CA2196679A1 - Target specific screens and their use for discovering small organic molecular pharmacophores - Google Patents

Abstract

The invention relates to a general process by which recombinantly derived variable domains of antibodies encompassing either or both light and heavy variable regions with or without respective constant regions are engineered and selected for identification of unique surface domains of pharmaceutical targets or parts thereof which regulate target function. The recombinant antibodies are useful as reagents for high volume, rapid screening of occupation of the active surface domains by natural or synthetic entities.
This invention is also directed to elucidating the three-dimensional conformation of the antibodies, or parts thereof, which bind to the pharmaceutical targets and confers activity. Methods for creating high resolution molecular models which can direct the synthesis of biologically active small organic molecules useful as viable discovery drug leads are also provided.

Description

W096l04s57 ~g& 6 7 g PCTrUS95rlO182 o -- 1 ~
TARGET SPECIFIC SCREENS AND THEIR USE FOR
DISCOVERING SMA~L ORGANIC ~n~RcTTT,~R pHARMACOPHORES

This is a cnnt;ml~t1nn-in-part of pending United States application 08/286,084 filed August 3, 1994 and which i9 incorporated in its entirety herein.

FTRTn OF T~TR INVR~TT¦ON

The inYention relates to a general process by which recombinantly derived antibodies (rVab) are engineered and selected to identify unique active surfaces of pharmaceutical targets. These recombinant antibodies are useful as reagents to identify natural or synthetic entities which occupy active surfaces of pharmaceutical targets and which therefore may be useful as therapeutics.
This invention also relates to el~rl~t;ng the three dimensional conformations of the various rVabs which bind to the ph~r~plltical target and confers target regulation and the use of high resolution molecular models to identify or synthesize biologically active small organic molecules useful as viable discovery drug leads.

R~rT~RnTJN~ OF TT-TR IN~ENTION

Today there are many approaches to identifying c_emical entities which have a desired effect on a ph~rr-nPl1t;cal target and therefore potential as drugs.
Common to all o~ these processes is the sequential use of multiple assays to identify a test compound's composite activity profile. This activity profile usually consists of information on four basic attributes: potency, .activity, selectivity and specificity. Selectivity indicates the ability to distinguish among closely related members of a particular target family. Specificity is the ability to distinguish between unrelated targets. Only two types of assays are used to develop the activity ~6~7 9 profile of a potential drug: one, a binding assay to measure affinity (i.e. potency) of the compound; and a second, an activity assay, to measure the compounds effect (i.e. agonistic or antagonistic) on the target. Binding assays measure the formation of the complex between target (T)and ligand (L). Targets include receptors, enzymes or structural components. Ligands include signals such as hn~mnn~n, neurotransmitters, growth factors::or test compounds. Until recently, L was labelled in some fashion (Lf) for identification and ~uantitation of the L:T
complex. Recently, binding assays have been developed which use a tagged R (R~) to asse8s L afflnity (see below). All these processes of labelling and R:L complex isolation and o~uantitation are known to those skilled in the art and have been reviewed.
In the process of searching for small organic molecules with appropriate potency, activity, selectivity and specificity for a particular pharmaceutical target, the order of testing is most often affinity, activity, selectivity and then specificity. In addition, some form of binding and/or activity assay, is interspersed with synthetic chemistry efforts at improving the compounds attributes. This generates an iterative cyclic discovery processes in which variou9 assays and synthesis are repeated over until a nnmpoiln~ possessing all of the desirable properties is obtained.
The present iterative process, although successful, is extremely time consuming~and has a high probability of failure for several reasons. Although binding and activity assays have now been automated, screening takes significant time as it is done on individual entities within chemical files cnnt~;n~n~ over lO0,000 entities.
In addition, the properties:of potency, activity, specificity and selectivity are separable, such that the presence in.a compound of any one property is not predictive of attaining another. For example, bindlng W096/04ss7 2 1 9 ~ ~7 ~ PCT~S95/10182 ~ .

assays give no cQnclusive data on the activity (i.e., a compound with high affinity may be an ahtagonist), and - activity assays do not predict selectivity or affinity.
As a result, modifying a compound so as to change one o~
its attributes (i.e., agonist activity) without modifying another (i.e., target affinity or selectivity) is unpredictable and considerable time is added to the discovery program when high affinity compounds identified early in the discovery process turn out to have ~ LU~' late activity or select vity.
10The relatively large number of biologically active small organic ligands having different general structures and which are capable of binding to a particular p~rm~utical target suggest that the binding surface of the target is not singularly uni~ue. Furthermore, binding assays using an endogenous ligand or close analog thereof are inherently biased to campounds which bind to only a fraction of the available surface of the target. Even where the labelled ligand ls not an endogenous one, this c~n~;n~mDrt means that the vast majority of active compounds identified by this process will be greatly restricted to the surface domain of the target which is used for lnteraction with the endogenous ligand.
This limitation is often viewed as desirable because the recognition damain for the endogenous ligand are those known via previous studies to have the ability to modify target activity. However, investigation of only ane target area severely restricts the ability to identify useful li~ands. As endogenous ligands in most instances are agonists peptides as in the case of opiate receptors, antagonist discovery can become a rare event. In addition, because e~dogenous binding domains often exhibit limited diversity among receptor members of a single target family, it becomes difficult for active compounds to discriminate among target family members. This often occurs when the ~nduy~uus signal for the family is a W096104SS7 PCT~S9S/10182

2~9&67~ --o single entity and not a group of closely related entities.
Acetylcholine (ACh) receptors are an example of a target family with only one signal=entity. The caterhnl~min~
receptors are an example of a target family with a few but highly related endogenous catechol signals.
In many cases, target diversity is found in target domains other than the specific binding site of the endogenou~ ligand. Some of these domains may be associated with the target's other functions, i.e., signal transmission while others are quiescent domains not being used by any endogenous signals recognition or tr~nqmi~ion. An example of a dilemma in discriminating among target family members is that found for the muscarinic receptor family lAChRm) where the binding domain for acetylcholine is-used to monitor a test compound's potency, yet finding AChRm agoni~ts which distinguish among the five ACHRm subtypes has proven illusive to date.
The task for drug discovery is to devise a screening approach~which provides detectable ligands to be used to screen compounds which bind=to the target and provide info~mation regardiny potency, activity, specificity and selectivity, as well as the three dimension~13D) conformation of compounds active at that particular slte on the pharmaceutical target.
AB part of any solution of these problems it is also necessary to establish binding assays which report the interaction of test .- ~Jullds with allosteric modulatory sites on targets. An allosteric site is one which modifies the endogenous ligand binding site yet is disrnnt;nnr~us and non-overlapping with that site. Such target sites have important physiological and pharmaceutical consequences and have been reported. For example, the allosteric site on the Gaba A receptor bind~
b~n70fli~7epines (BDZ) and thereby modulates the binding of the endogenous n~uLuLL~n~ tter Gaba. Occupation of the allosteric BDZ site, which can be done by chemicals from many unrelated structural groups, has a significant and recognized thrr~r~lt;c infl~lrnre on physiological processes tnrlnfl;ng anxiety and sedation.
It is also known that active allosteric sites exist which are modulatory for endogenous ligand binding and have observable effects of their own on the target Such an allosteric site is present on the Gaba receptor [Garrett, Blume and Abel 1986; Garrett, Abel and Blume 1986].
l~ Present screening technir~ues which monitor direct binding of test compounds to allosteric target sites are not routinely done because a) high affinity tagged ligands which bind to these sites are u,qually unavailable at the start of a discovery program; and b) the necessary monitoring of detectable endogenous ligand dissociation or bioassays are too time consuming in initial screening protocols. Without a simple, rapid and comprehensive way to observe all potential target sites, investigation of the surface of a ph~r~r-~llt;r~l target for potential modulation remains limited to a small part of the target surface :New methods are r~r~qS~ry to survey the entire target surface in early screening for discovery leads.
Recently methods of identifying various entities which recognize target surfacec have been reported which do not depend upon the av~;l Ah; 1; ty of tagged ligands with high affinity for the target. [Delvin, J.J., P~nJ~n;h~n, L.C., and Devlin, P.~., l990]. These assays detect a compounds surface recognition activity directly via formation of an ~ flPnt; f; ~hl e tagged target (T*):Ligand complex. In one version, test compound is coupled in r ; flPnt; f; ~ compartments to a solid matrix of varied composition at rnnr~ntr~t;r,nq which allow sufficient amounts of labelled target to bind and form stable ligand-labelled target complexes for subsequent detection via rh~m; r~l, radioactive, or biological methods known to W096/04557 PCT~S95/10182 ~ ~ ~ . L .
21966~

o .
those skilled in the art. Subser~uent isolation (or identification) of test compound from the compartments rrnr~ining labelled target provide active c~emical structures. In one such version where test compounds are free oligonucleotides, the oligonucleotides are isolated in complexes with the target, and are amplified and sequenced by PCR technology. [Delvin, J.J., Panganiban, L.C , and Devlin, P.E., 1990].
Phage display is a particularly sensitive method of presenting peptide test compounds to a target Phage may be engineered to express the gene rnrrfl1nr~ the test peptide as a fusion protein with one of its surface proteins. Methods involving phage display are referred to in Winter et al. PCT application W0 92/20791; Xuse, W092/06204; and Ladner et al. W090/02809.
Although these newer approaches have now been incorporated into random drug screenlny~protocols, they do not resolve the following problems: the assays of the critical attributes of potency, activity, selectivity and specificity are still unconnected; active target surfaces including the endogenous ligand site and allosteric sites have not been identified; and 3D information on conformation of the active agent is not provided. More importantly, most of the agents available for screening, i.e., peptides, nucleotides, lipids, and carbohydrates which are available in large libra~ies, are not totally satisfying as discovery leads because none are expected to be orally active, or pass ' dne barriers to get at intracellular or central nervous system targets. In addition, these classes of~compounds are 8c flexible as to obscure their active 3D-configuration to such a degree as to prevent or severely limit their use as models for ~ organic synthetic efforts. An ill~Luv~-~nt in screening would then encompass a regolution of thege defiri~nri~r 80 that these broad surface recognition libraries could attain their full usefulness.

W096104ss7 PCT~S95/10182 219~67~ ~ .
....
0 - 7 - =
In covering the prior art for high throughput binding screens ior target modifiers, it also is necessary to review what is known of the endogenous ligand signals as well as their targets. Both shed significant light on additional problems and limitations encountered in the binding assays available today for discovery approaches.
Endogenous ligand signals are those ligands which directly modi~y target activity. The size of endogenous ligands varies greatly, ranging from lO0 Daltons (e.g., as for glycine in its regulatory role as an excitatory amino acid neurotrangmitter) to over lOOkD (e.g., as for some P~trp~1lnl~r1y active growth factors (GF) with a proportionea increase on surface area. The com~osition of endogenous ligand is eo~ually varied ~n~lnrT1ng organics such as neurotransmitters; peptides e.g. somatostatin, DX, BHRH and TRH; proteins eg., growth factors; and lipids;
carbohydrates; and inorganics such as ions.
For discovery purposes, common to all is the desire to replace the endogenous ligand with a small organic molecule. The problem of screening for rPplp~m~nt~
appears to be very different for most small endogenous ligands, i.e, neurotransmitters and neuropeptide m~rTlTl~t~r5 compared to large endogenous ligands i.e., ~rm~n~, growth and differentiation factors. Although small organic molecules have been found which can be active a~ targets for small endogenous ligands, few, if any have been found for the larger molecules such as proteins.
Corresponding to the diversity in endogenous ligands is the equally extensive diversity in target domains which are responsible for recognT~ing (i e., binding) and ~ responding to endogenous ligand signals. It is generally accepted that both signal and target have specific domains involved in forming the actual contact points found within the endogenous ligand:target complex ~EnL:T). Recent data on crystallized growth hormone (GH) and its receptor W096/04557 2 ~ 7 ~ PCT~S9~/10182 , .; ,'' complex provides detailed molecular information on the amino acid~ within the GH hormone ligand and its target GX
receptor interactive domains.
Recent data on the crystal structure of G~ and its receptor has shown a single GH molecule to contact the same set of amino acids in each of two i~Pnt;rAl G~
receptor units complexed with one GH molecule.
[Cunningham and Wells 1989; ~nnnlnrJhAm et al. 1991; DeVos, et al 1992]. Each of the receptor units therefore has only one target site which i8 the same on both units.
Each receptor uses the same 7 amino acids to define the binding site which participate in GE binding and receptor dimerization necessary for activity. [Cunningham and Wells 1989; Cunningham et al. 1991; DeVos, et al. 1992].
Dimerization of at least two receptor subunits by monomeric or multimeric hnrmnn~ is required for receptor activation for the majority of hnrmnn~F studied to date, such as growth factors, including ~erve growth factor (NGF), rrl~rrr-l growth factor (EGF), fibroblast growth factor (FGF), interleukins (IL2, 4 and 6), interferDns and insulin. [DeFronzo, BnnA~nnnA, and Ferrannini, 1992;
Bamborough, Hedgecock and Richards 1994; Kishimoto, et al.
1994; Claesson-Welsh, 1995]. In some cases, the two units of the hormone, as well as receptor are not genetically related. In such cases one subunit provides high affinity hormone binding and the other intrAr~llnlAr signalling (e.g., tyrosine kinase activity). [Ullrich, et al., 1986;
Kaplan, Martin-Zanco, and Patrada, 1991; Kaplan, et al.
1991; Klein, et al. 1991; Argetsinger, et al. 1993;
Obermeier, et al. 1993; Weiss 1993]. In some cases, the lower affinity receptor when dimerized can be activated.
[Ullrich and Schlessinger 1990; Stahl and Yancopoulos 1993; Claesson-Nelsh 1995].
Among many hnrrnn~s and hormone receptors, it is now apparent that an unexpected and un~anticipated degree of structural homology exists with subgroups of:these signals 21~g7~ , W096/04557 PCT~S95/10182 ~ .. .. .
. . .
~ ,. :

O -- g and receptors forming homologous families which sometimes follow along different genetic evolutionary lines. Other functional simiIarities may be brought about as a result of c~11v~~ t evolution. In either case, the active 3D
co~formations of ligands and receptors appear to follow some general prinr;p~lr. Xowever, fbr drug discovery, the principals gleaned from these studies have not yet been detailed enough to bypass crystallography of particular hormone/receptor complexes in order to gain sufficient specific information as to deduce the molecular shape o_ active small organic molecules.
Deciphering the elements necessary in a signal to activate a hormone/growth factor receptor has inrl1~
crystal formation and analysis at <3A of receptor and endogenous ligand complexes; (2) the ;nrlllpnre on function (i.e, ligand bindi~g and receptor:activation) caused by molecular biological mutagenesis of single amino acids or short peptide del~etion/replacement, or chimera formation of both the hormone and receptor units In addition, monoclonal antibody binding to surface domains available when ligand and receptor are either uncomplexed or in the R:~ complex, along with the ability of Fab2 versus Fabl to activate or block receptor activation ~ vitro, n situ or 'n ViVQ has been studied_ ~
The above studies when taken toyether, provide information rnnrP~n~ng (l) the contact points between hormone and receptor; (2) the amount of energy of binding involved in these contact points; (3) amino acids outside of the receptor:ligand contact points essential for global ~ receptor/ligand stability or dimer stability, or receptor ~lrJn~ll;ng activity (i.e. tyrosine kinase, binding of ~ other intracellular regulatory factors, internalization, ~ uncoupling for effector system).
Critical for identifying small organic molecules which are active at hormone receptors are the data from the above indicating (l) number of units/active complex;

W096/04557 PCT~S95/10182 21 ~679 (2) amino acids of the target specifically involved in the binding domain with the endoyenous ligand; and 3) amlno acids of the ligand specifically involved in binding and/or activating the target. Of all of the above~
information, clearly the rate limiting event=today is obtaining sufficiently resolved crystallography data of hormone/receptor complexes. ~owever, complexes of receptor--and ligand are often difficult to identify and crystalize thus preventing on~from~obtaining the structural information. It is also recognized that the various molecular, biological, immunological~studies, biochemical and pharmacological studies noted above, also take considerable time and effort. Accordingly, prior art approaches to identifying active small organic molecules are long and arduous with unpredictable results.
In the approach outlined above, it is important that both structural and biological data be obtained as each has its own limitations and artifacts. Also, contact points could reflect specific aspects of crystal formation which do not re~lect the structure at the protein in ~iL~, or the crystal may contain an ind~ Liate number of subunits. On the other hand, the biological data generates both false positives and negatives.
Furthermore, if antibodies are used to probe the binding site of the target, not all receptor or~ligand surfaces may be immunogenetic accessi~le tQ Fab2 or Fabl antibody.
Another problem is the difficulty studying allosteric sites which do not interact directly with the signal ligand.
Despite c~nci~rable e~fort, a major problem in drug discovery has been the identification of small organic molecules capable of activating peptide hormone/growth factor receptors. This is likely the result of the multivalent nature of ~u~uy~uuus ligands for these receptors and the requirement to dimerize or simultaneously activate multiple attcrl t sites on a W096/045~ 219 6 6 7 ~ PCT~S95/10182 ~, single receptor (receptor subunits) for receptor activation. Even for receptors which are h~m~;m~rs~ such as GH receptor (G~R~, a single small organic molecular monovalent att~hmpnt to the G~R site I is not sufficient to cause activation, nor ~i ~rl ~ t of growth hormone from its active divalent dimer receptor complex.
Failure to find single small organic molecules in conv~nt;~n~l binding assays stems from the fact that the labelled hormone is bivalent, and its disrl~ nt from two receptor units by a single monovalent small organic molecule (i.e. compounds which attach to only one receptor target at a time) is thermodynamically unfavorable in the present day binding assay. Furth~ ~, in the large ma~ority of cases the receptor for a given hormone is a heterodimer. Thus, for a given hormone/growth factor-receptor binding pair, there may exist at least two different binding sites on the target which may be due to the multimeric rature of the target or a target consisting of allosteric sites on a monomeric unit. In all o~ these cases, the endogenous ligand must therefore comprise at least a sufficient number cf binding sites which are properly spaced to bind to the multiple sites on the target n~c~c5~ry for activation. Obviously, one would then require a multimeric or a multivalent small organic molecule for displacemènt of these hl ~ from their targets.
Given the c~mpl~;ty required of each small organic molecule to bind the receptor at the multiple sites n~ ry for activity, or to displace the endogenous ligand, one could expect that the occurrence of a single small organic molecule with two unrelated yet active binding domains would be equal to the chance of finding one multiplied by the chance of finding the other independently. As active small organic molecules are iound by random robotic assays at a frequency of between l/l000 to l/l0,000 on most screens for ligands requiring W096/04557 PCT~S95110182 21n~7~

only one binding site on the ligand, and which have correspondingly a single binding slte on the receptor, one would expect to screen an organic chemical libraries c~nt~n~ng from 106 to 10~ compounds in order to identify an active molecule. Such libraries exceed those which could be screened in some reasonable assay format and actually exceed most made by even the largest pharmaceutical companies.
Therefore, a different approach to screening for small organic molecules which can activate hormone receptors is needed.
A number of libraries now exist for screening such large numbers Two have been noted already, the oligonucleotide and peptide library. Another such file contains natural products.
Classical chemical libraries consisting of synthetically derived small organic molecules are routinely available from commercial sources (e.g.
Alldrich, and Kodak) and consist of upwards of a 1-200,000 entities Recently a survey of the chemical entities within such libraries uncovered 100,000 or so chemical structures as being the cores upon which most of the individual entities were crafted. The average molecular weight of the entity within such files ranges between 200-400 Daltons which would account for no more than one such contact site per target.
Screening of small chemical compound libraries is limited only by their availability, which most often is <100,000.
With the advent of molecular biology and gene cloning and se~uencing, it has been discovered that most pharmaceutical targets are~ot uni~ue entities unto themselves, but in fact belong to families of sometimes rather large size and close relatedne8s Recognition of~
this fact has mandated a much more serlous look at all of the members of the family to which the target under W096/04s57 2 l ~7~ 7 9 PCT~S95~1~182 investigation belongs so as to identify lead compounds which can distinguish among its family members. If one used only binding assays as a primary screen for potency, activity, selectivity and specificity, one would require affinity labelled standards for each~Qf the family members Although this is potentially possible when the endogenous ligand signal are proteins due to their native affinity and ease of labelling, it is not presently feasible where small organics are the only known signals This approach is also unsuitable _or targets with l~nlr7~7lt;f;r~ gignal ligands. ~ny discovery of how to include such widespread specificity ~esting into primary binding screen assays would greatly increase the probability of drug discovery success.

S7J~ARY OP T~R Ih7VE~TION

This invention provides compositions and methods for identifying active surfaces of biologically active sites of pharmar~--r;r~l targets. Identification of these sites is useful for preparing reagents suitable for use in screening assays of small organic molecules to identify those as r~rr7;r7~te lead compounds possessing desired attributes of biological activity, specificity, selectivity and affinity.
Reagents are provided by this invention which are suitable for identifying active sites on ph~r~r~l7tlca targets. The reagents comprise libraries of variable regions of antibodies obtained and modified by molecular biology techniques which are used to prepare recombinant Pab fragments (rVab) useful for scanning the surface of a - target in a manner so as to identify those rVab's having desired potency, activity~ specificity and selectivity.
The attributes of potency, activity, specificity and selectivity are rrllpr~;vely referred to as a "composite activity profile'7 (CAP) The rVab's which are made and W096/04557 PCT~595/l0l8~
2 1 ~

identified by this invention as possesslng the deslred CAP
attributes speclflcally blnd the target (i.e. are T~), are selective for the target (S+) and activate the target or are capable of activating the target when combined with another ligand (A~).
By c~mhining structural features of va~ious members of the recombinant antibody llbrary which possess activity at a defined pharmaceutical target, this lnventlon provldes a method of det~rm; nl ng a composlte structure possesslng the deslred composlte activity proflle. Thls composite structure may then be used to identify small organlc molecules capable of actlng at the target surface wlth elther agonlst or antagonist actlvlty wlth the sufflclent speclficity and electivity.
The method according to thls invention of identifylng llgands capable of blndlng to actlve sites and possessing a composite activity proflle ~or a glven pharmaceutical target comprises combining members of a recnmh; n~nt antibody library wlth a pharmaceutlcal target coupled to a reporter whlch reporter is çapable of signaling activation ~ or lnhibition~of the pharmaceutlcal target. Reporters of pharmaceutlcal activity may include but are not limlted to, for example, receptor coupllng to ~~~~ tors such as the G protein; ollgomerlzatlon of receptor subunlts;
changes in enzymatic activity such as klnase actlvity; or changes ln lon flux Accordlng to this method, lndlvldual members of the library possesslng-deslred actlvlty as demonstrated by the reporter, are useful indivldually or collectlvely ln subse~uent assays to ldentlfy small organlc molecules capable of possesslng the desired actlvlty at the pharmaceutical target. By cnmhinl~g structural features ln common between multlple members of the llbrary possesslng the deslred actlvlty, a composlte structure for açtivity may be derived whlch may then be used to create a model for a=compdund~possesslng the desired activity attributes.

WO 96/04!i~i7 . PCT/lJS9!illO182 2~g6~i79 ~ ,", . , o ~ 1~ -This invention also provides a method of identifying small organic molecules which are active at the target sites comprising screening potential drug ~nA;~tP~ in a binding assay for their ability to displace 1 Ah~ A, rVab members possessing a desired composite activity profile consisting of potency activity, selectivity and specificity for the pharmaceutical target.
Small organic molecules as ~nA;~tes for drugs may also be iaentified by analyzing the structure o~ the model derlvea ~ro~he structure of at ~east two active members of the rVab library and det~rm;n;ng common characteristics including, but not limited to charge and spacial orientations which participate in binding to the active sites of the pharmaceutical target. Using the model, small organic molecules may be obtained by synthesizing compounds possessing the common structural features identified in the model, or screening a chemical file data base ~or memoers possessing features in common with the model.
This inve~tion also provides means oi identifying structural re~uirements of ligands capable of binding to pharmacological targets comprising multiple binding sites existing on one or more molecular entities which when bound by a single ligand are capable of activating the pharmacological target. Similarly, this invention provides a means of identifying structural re~uirements of multivalent ligands capable o~ activating pharmacological targets comprising binding sites-too large to be occupied by a monovalent small organic molecule or reo~uiring - concurrent binding of a multivalent ligand to ef~ect oligomerization o~ separate molecular entities to form an - active pharmacological target.
This invention also provides reagents comprising recombinant antibody libraries (rVab's) which have been constructed to encode CSR and ~DR regions with specific , . . ~ .

W096/04ss7 PCT~S95/10182 variations and ln which the CDR and CSR regions are expressed on a specific ~pntlfl~h~e fL~..~.r~Lh structures.
The recombinant libraries of the invention may be packaged in various forms including bacterial phage which express the recombinant antibodies on their surface.
It is therefore an object of the present invention to provide a process for the identification of~small organic molecular replacements capable of modifying a pharmaceutical target with a desired composite activity profile comprising sufficient potency, activity, specificity and selectivity to be considered as an--initial drug discovery lead.
It is a particular object of this proce8s to identify surfaces of a pharmaceutical targets capable of discriminating among members of a family of related targets which are activatèd by the same or similar endogenous ligand or utilize similar signal transduction merh~n~l It is a particular object of this process to identify active or regulatory surfaces of a ph~rm~rPlltjcal target which may or may not be usea by an endogenous ligand for the target of interest, and which is nevertheless capable of modifying the pharmaceutical target in some ph~rm~r~ntically useful manner. ~ =
It is a particular object of this process to identify allosteric sites on the pharmaceutical target which are not used by endogenous siynals nor have activity on their own, as well as active allosteric sites which are used by endogenous signals other than the pharmaceutical target activating signal and which~have some type of activity on their own.
It is a particular obiect of this process to provide a repertoire of surface recognition libraries which together recognlze diverse pharmaceutical target surfaces by constructing a small number of combinatorial antibody 1;hr~riP~-~VO 96104S57 PCI/US95/10182 ~ 2 1 9 ~ ~ 7 ~ . ! . s It is a particular object of~this process to convert ~ by a single simple and rapid procegs any llnl ~h~
recombinant variable antibody fld~ ' (rVab) isolated r from a library to a labelled one to act as a reagent capable of identifying small organic molecules which possess any one, or combination thereof, of the attributes of potency, activity, speci~icity or selectivity simultaneously when screening random chemical libraries It is an object of this process to identify the specific binding regions of ph~rm~c~utlcal targets requiring binding to 8ites in at least two different regions to cause a response of the target. Such regions may be present on monomeric or oligomeric pharmaceutical targets. The endogenous ligands for such sites generally are multivalent monomeric or oligomeric proteins which bind to the multiple regions which define the active surface of the pharmaceutical target.
This invention provides a method for identifying the structural requirements for ligands to bind at the separate regions and identifying such ligands. By combining the ligands capable of individ~ally binding to the separate regions into a single molecule, fully active ligands are provided.
It is another object of this invention to identify the monovalent det~rm;n~ntc making up the active surfaces on the targets for large protein signals guch as h~rm~n~
and growth and ~;ff~r~nt;~t;nn ~actors consisting of oligomeric receptors. Such receptors may contain homologous or heterologous components with one or more o~
these units c~nt~;n;ng a part of the signal recognition det~rm1n~nt.
It is a particular object of this process to use chemical oligomerization of small organic molecules for each of multiple binding sites to derive an active oligomer for~large proteins such as growth factors and :

W096/04557 PCT~S95/10182 ., --21~667~

~rmnn~ which contain multiple binding sites within their active binding domains.
Accordingly, another object of this invention is to identify small organic molecule repl~c~m~nt~ for large protein signals such as growth factors and protein h~rm~n~ be they allosteric or competitive modifiers and whether they be monovalent or multivalent.
It is a particular object of this invention to identify small organic molecule replacements for pharmaceutical targets which have no bioorganic endogenous ligand signals, such as certain ion ~h~nn~.i, pump9, and exchangers. .~ ~
It is a particular object of this lnvention to provide high volume binding assays which discriminate agonist from antagonist small organic molecule replacements.
It is a particular object of this invention to be able to identify from large antibody variable region libraries, lndividual variable regions which distinguish from one another binding sites which confer selectivity of pharmaceutical targets for specific members of a gene family.
It is a particular object of this process to provide labelled antibody variable regions which lnt~r~rt with and modify the activity of targets which have no 1~n~
endogenous ligand, nor exogenous natural signals, and which labelled ligands have sufficient affinity for the pharmaceutical target to be used in competing binding assays in which small organic molecule8 may compete for binding with the labelled ligands.
It is another object of this invention to provide a plurality of different recombinant antibody variable regions which recognize at least one common binding 8ite of a pharmaceutical target and which collectively provide structural information useful for designing active small WO 96104S57 ~ = PCT/IT595/10182 2i~667~ :

o organic molecules which are active at the pharmaceutical target.
It is another object of this process to provide a general method to rapidly obtain peptide structures which are useful as 3D models comprising the minimum characteristics of small organic molecule replacements which have sufficient po~ency, activity, selectivity and gp~r; ~; ~; ty to clagsify ag viable discovery leads.
It is a particular object of this process to provide molecular models for ac~ive ligands wherein the pharmaceutical target necessary to be occupied by active ligand comprises one or more binding sites on one or more molecular entities.
It is a particular object of this process to be able to solve the canonical structures of the CDR VH3 of rec~mh;n~nt antibodies which have b~en identified as possessing the desired properties of potency, activity, selectivity and specificity.
It is a particular object of this process to be able to use composite structural characteristics to direct a synthetic effort capable of directly synthesizing active small organic molecules.

RRTT~T~ DFSCRIPTION OF TT-TT~ FIGTTRT~.C

FIG.l. Stages of the Topogr=aphic System Assay (TSA).
Fig. l shows the activities and products of the three main stages of the TSA. When combined together, Stage I and II, or Stage I a~d III, allow the identification of small ~ organic molecules (SOMERS) which are active at pharmacological targets(T). A MUBTIMER is at least two SO~ERs covalently linked togethe~ to produce an active molecule. A BEEP is a biologically enhanced ensembled pharmacophore, and Tn is subunit n of pharmacological target.

W096/04S~7 21 9 6 6 7~ PCT~S9~1101~2 FIG 2 Related Antibody Structures and Varlable Region Domains. A. Shows various forms of antibody structures including the variable(V) and constant (C) regions of immunoglobulin (Ig) heavy (H) ana light(L) chains. Antibodies constructed in this invention by molecular biology technology have a r prefix. ~. Shows details of the antigen recngn~tlnn Variable=region (V) domains of the VL and VH. FW is the 'constant' framework regions; "CDR" refers to the complementary detPrmining regions as defined by Kabat (Kabat l99l); CSR refers to canonical structures found in CDRs as originally defined by Chothia (Chothia and Lesk, 1987); V (with leader sequence), D(diversity) and J(V/C junction) are the genes which are n~ h;npd to create the mature VH and VL genes. V
Regions are attached via genetic recnmh-n~t~on for V~ to either a kappa or lambda Constant region. VH are recombined with three Constant regions in sequence with CHl being attached to V~ The V regions of the invention can used either without C regions, or with kappa or lambda for CL, and up to three C regions for CH
FIG.3. Potential Planar, Cavity and Grove Antibodies of ~nown Crystalline Structure for rVab Library Construction. Fig.3 lists a n~mber of ~ntlho~es for which there data is known cnnnPrn~ng their crystalline structure and which are potential parental antibody structures for construction of the rVab library as described in this invention. The antibodies are grouped according to their type of antibody combining site : i.e., planar, grove or cavity-type structure.
FIG.4. Comparison of:Natural Fab and rVab Library Diversification. A Nature's Immune Repertoire: V, D and J
are the genes recombined to make the mature V gene; rf~
are the reading frames of the D gene which can be used to make sense protein seq~ences upon recombination with V and J. CDR~ are there are no CSRs for the VH3 region. ~The number of known CSR for each CDR is given in parentheses.

w096l04557 2 1 9 ~ 6 7 g PCT~S95/10182 B. The rVab Repertoire: Div~rc;f;r~t;on arises by using all p~ ~ct;on~ of the known CSRs, 3 d~fferent length CDRH3 and r~n~ 7~t;~n of amino acids at two positions within each CSR ~or CDRH3) within a single VH and VL
parental fL~r._..J-k structure. Primary r~n~rm;7~t;~nc are made duri~g construction o~ the rVH and rVL (see Figs.
7,8) and allows all 20 essential amino acids to appear at given positions within V regions among members of the rVH
and rVL libraries. CDRH3 are three known CRDH3s of different=sequence and=three different lengths covering V~
amino acid positions 95-102 (see text for details). rVab is encoded by one rV~CHl and one rVLCL gene on the same piece of DNA. Totals of CSR=include:CSR and CDRB3 . combinations.
FIG.5. Type and Diversification of Amino Acids at various positions within V region. Numbering of the amino acid (A~) positions as per Kabat (Kabat,1991). Library Diversification identifies the high priority r~n~ te amino acid position for primary library diversification during construction of rVH.lib and rVL.lib as described in this invention.
FIG.6. CDR and Canonlcal Structures (CSR) of V
Regions.
Particular amino acid (single letter code) at V gene positions rr;t;rcl for particular CSRs are given as defined by Chothia (Chothia and Lesk, 1987). ~ represents amino acids not within CSR or CDR which participate in ~rf;n;ny the CSR. The diversity position is the amino acid position used for primary library r~n~rm;7~tion as ~ described ii this invention.
FIG.7. Construction of the rVLCL Lib. of Diversified rcnr~n;r~l CSRs: rVLCL.Lib. A-F ~re sequential steps of the process constructing rVL.lib. G is the final step of recombination of rVL.Lib with a rCL to form rVBCL.Lib.
Amino acid positions occur in brackets; nucleotide poaitions are given left to right as 5' -3' in W096/04557 PCT~S95/10182 ,, ' 219~67~ 22 -parenthesis; restriction sites (rs) also appear in brackets and have a "p" prefix to denote when they are located within the plasmid and not the V region.
Restriction sltes are denoted by combinations of letters and numbers. Prlmer direction is denoted by arrows (left ls forward (FWD)), and right is a reverse (BCK) prlmer).
* denotes more than one amino acid at a CSR position which is critical for a particular CSR; ~ denotes that diversification by rAn~ 7Ation of amino acids with CSR
or CDR has occurred. ~ib suffixes indicate a library of many individual members. Heavy Ilne lndicates that the product(single entity or library) has been cloned in to plasmid pCLONALL:~pC).
FIG.8. Construction of the CSR and CDRH3 Diversified rVHCXl.Lib. Construction of CSRHl and U2 and three CDR~3 of different lengths ~i.e., 5.7.or 10 amino acid insertions); diversification by amino acid rAn~m~7Ation=and combination of CSR and CDRH3 in all possible permutations is as illustrated in a manner analogous to that descrlbed for rVLCL.Llb (see legend to Fig.7).
FIG.9. General Usage Plasmids..A. Illustrates the se~uence of restriction sites (rs) which occurs in the cloning site of pCLONALL. Use of each in rVH.Lib and rVL_Lib construction is noted wherein "---" denotes a restriction site used and defined by parental AB se~uence;
wherein X denotes a re8triction site not used in that particular rV lib construction. General positions of restriction sites within the rV and rC regions under construction are shown. JCH and JJCL are the natural J/C
gene recombination region with included amino acid positions given in brackets. JCLINK is the position of the J/C recombination restriction site, also referred to as rs3. B. Events used in constructing the plAo~q carrying rC regions and in the final step of rVab.Lib construction wherein rV regions are appended to rC regions. The two W096/04sS7 ~9~ 7 PCT~595/l0l82 I' '~ ' . .

plAq~ needed for this are listed as pVx~CCEPTORs. C
Plasmids used in creating expression vectors for the rVHCH1 and rV~Ch chains of the rVab when not attached to phage coat protein gpIII EK is the enterokinase cleavage site ISOTAG is the additional amino acid seriuence useful in isolation and labelling rVab as rVab~ O~
constructs.
FIG 10. General Primer Table. Primers are written as 5'-3' Numbers and single letters designate individual amino acid positions which in the primers would be corresponding triplet codon seriuences for the amino acid at these positions The letter N within parpnthpc;~
denotes the random ~pFPAr~nre of the nucleotides A,T,U,C
used to rAn~n~; 7P the amino acid at this position.
~etters, without parpnt~p~;r/ are used ~or sequence5 nPrPrgAry for a desired CSR or CDRH3 structure; numbers are used for sequences which are not critical to CSR or CDRH3 structure. rs is a restriction site sequence.
Sequences for all FWD primers are , l~ t~ry to the sense sequerce. Approximate primer sizes in nucleotides are listed as #mer. The right hand column signifies general use of primer with amino acid rAn~t '7At;on; and SEQ. is sequencing FIG ll. Constructs for CRE-~OX rer 'inntnrial formation of''rVab lib: PartI Expression of rVab with or without one attached random octamer peptide (Pep 8) library. Figure 11 illustrates the steps generating the necessary phagmid and plasmid constructs to allow in vivo rPrn~h;n~tinn of individual rVHCHl.lib and rV~C~.lib ~ members, by the Cre rPrr~mh;nA~e, and the construction of a single phagmid cnntA;n;nr an rVHCH1 and rV~C~ member on one piece of DNA (i.e , an rVab). This procedure i8 used for rVab.~ib construction where~there is no need in the TSA discovery process'for subse~uent A~;t;r,n to rVab of more than one random octamer peptide (Pep~.~ib). Wild type (wt) and mutant (511) loxP sites are as defined in legend WO96/0455? PCT~S95/10182 , 2 1 9 ~ = 2 to Fig.12 LpelB and LgpIII are leader sequences for pel3 and gpIII.
FIG.12. CRE-LOX Plasmid and Phagmid Se~uences used for rVab.lib Construction. For use in rVab.lib construction ~y in vivo Cre-recombinase directed recombination of rVXCX1 and rVhCL onto singIe phagmids where there i9 a subsequent neea ln the TSA process for attachment to rVab of no more than one random peptide library.
FIG.13. Constructs for CRE-LOX Recombinatorial Formation of rVab.Lib: Part II. Expression~of rVab with or without one or two attached random octamer (Pep 8) peptide l;hr~r~oct. Steps involved in adding Pep8.Lib; i) illustrates expressing one peptide (PEPI'') at the amino terminus of VX (Pap8~A'); ii) illustrates expressing one peptide at the carboxyt~rmi nllct of CL ~Pepl"); and iii) illustrates expressing one peptide at the aminoterminus of VX (Pep8~"' and one peptide at the carboxy~rm-nnq of V~
(Pep82"). Step E illustrates use of two primers required to append Pep8.Lib to either VX or CL
FIG.14. In vivo, Generation and Expression of rVab.Lib members. The generation of rVL and rVX gene pairs (rVab) as one DNA molecule, as well as the expression and phage display of rVab attached to coat proteins of fd is illustrated. Synthesis of rVXCX1- and rVLCL proteins and their complexation to form gpIII
attached rVab for phage display is illustrated showing cells, such as bacteria, infection of bacteria with phage carrying rVLCL and transformation with DNA plasmids carrying the rVXCH1- construct; and in vivo recombination of rVXCX1 and rVLCL onto~a single ~d via the LOX seque~ces and the P1 provided CRE-recombinase. Following recombination and replication, a combined single expressible pair of rVab genes is packaged per phage.
When induced, rVLCL is made~and introduced via its leader into the periplasmic space were its complexes under WO 96/04!i~i7 PC'r/lJS95/10182 ~ 2196g79 . ~

reduced rnn~lt;nnr with synthesized rVECEl-gpIII coat protein to create the desired rVab complex attached to the gpIII phage coat protein (see text fo~ details).
FIG.15 Flow Diagram of Diversification and Simplification Paths of the TsA~ Steps are~outlined for optimizing TSA+ attributes of rVabs for a given pharmacological target. The library attributes are potency of binding to Target (T), specificity and selectivity for Target ~S) a~d regulation of target Activity (A). "+"
denotes that the attribute is present in the rVab member.
FIG.16. Isolation of Target (T+) Specific/
Selective(S+) rVab. A. Isolation by panning for Target recognition(binding) (T+). B. Isolation by panning for Target Specificity and/or Selectivity (S+). Isolation of T+ and S+ rVab can be done in any order, and when used together isolate rVabTS+ members. T denotes the pharmacological target; ~ phage displayed rVab; com-T-pep represents the entity, holotarget, subunit or peptide fragment, which is to be distinguished from the Target.
Binding to the com-T-pep prevents rVabS~ binding to matrix attached T.
FIG.17. Selection of rVab Scanners ~or Active Target Surfaces Used by Signals with Single Atr~r~m~nt Sites. Fig 17 presents a flow diagram of the TSA procegg ;~n1~t;nr;
rVabTSA+ members from a rVab library previously ;fl~nt;f;ea as T+S+. T, S and A are defined in legend to Fig.15.
Native signal is the endogenous or previously ;flPnt;~ied agonist entity (e.g., protein, peptide, neurotransmitter) which activates Target by interaction at a single att~rhm~nt site (see Text for detàils). Allosteric Fffector is an endogenous or previously identified entity whlch binds to a single att~rl t site on Target which modifies agonist activity but has no activity on its own.
rVabA+ are isolated by competition by native signal or allorteric c_ange in T by allosteric ~ff~rtnr which prevents normal rVabTS+ binding to T. The binding of ., , W096/04557 PCT~S95110182 21~7~

rVabTS+A- members is unaffected by the presence of the allosteric effector or native signal and therefore is not isolated free ln the supernatant during this process.
FIG 18. Discovery of SOMERs for a Target with a Single Univalent Actlve Site. Fig.18 illustrates the steps of the TSA process of rVab- Scanner to Reporter conversion and Reporter use in competitive binding assays to identify active SOMERs for the pharmacological Target. Both competitive and allosteric active SOMERs are identified in this process.
FIG.19. Identification and Isolation of Active rVabTSA+ for the Muscarinic Acetylcholine Receptor subtype ml (AChR~). (s)-denotes matrix attached Wheat Germ Agglutinin; T,S and A are as defined in the legend to Fig.15; R denotes receptor target; G denotes guanine nucleotide binding protein; RG denotes RG noncovalent complex; ~ denotes phage displaying rVab. The TSA process isolating rVab based on specificity/selectivity (see Fig.
16) is illustrated for the isolation of AChR~ rVabS~+
using Agonist-~ike rVabT+A+(type 1 rVabTA+). The same process is used for isolating Partial-Agoni8t-Bike, Allosteric-Agonist-Bike and~Competitive Antagonist-Like S+
rVabTA+ (i.e., respectively type 2,3 and 4 rVabTA+) FIG.20. Isolation of Active rVabTSA~ for Complex Active Sites on Dimeric Receptor Targets ~T~2). Fig. 20 illustrates the TSA process by which the rVab pair~for each part of the active site on each of two receptor target subunits (Tl or T2) is isolated. The process is shown in full for one member of the pair; that for the active site region on T1, and is duplicated for the active site region on T2. m-T denotes matrix attached Target;
comp-T-receptor denotes comp-T-pep as described in Fig.16.
~ denotes phage displayed. Pep8.Lib is the random octapeptide library displayed as a fusion protein with phage coat protein gpIII. Pep~T2+ is the library of peptides which bind to T2. rVabT1-Pep8T2.Lib is the W096l04557 2 1 ~ 6 ~ 7 9 PCT~S95/10182 ~ . ~ ~ . ,, rVabTlS+.Bib to which the Pep8T2+ Bib has been appended ~ ~see Fig.12 and 13 for details of rVab-Pep.Bib construction). Preselection of the T2+ Pep8.lib is not ~ re~uired and a random Pep8 ~ib can be used in this process Testing for rVabS+ is optional and can be done at any step along the process The related rVabT2m+S+A+-Pep8Tl+ member of the active domain pair is obtained in parallel analogous manner.
FIG.21. ~sing Active Bivalent rVabT1-Pep8T2 to Screen for Disomer Replacements of a Multivalent Signal Fig.21 presents a flow diagram of the steps of the TSA in which each rVab member of the active pair of rVab-Pep8 for both domains of the active site, which occur on separate T
subunits are used to find a DISOMER r~pl~c t for the native signal and which rerulates Target activity. [A+]
denotes that the rVab-Pep entity is active in regulating the T1-2 dimeric Target. A* denotes that the rVabTS+
member is derived from a rVab-Pep entity which is [A+].
DISOMERmn denotes covalent linked SOMERs for the pair of active site domains identified by the paired rVabTSA*
members.
FIG.22 Summary of the Discovery of DISOMERs for a Bivalent ~ormone.
FIG.23. Elow Chart of TSA Steps Creating and Using a Biologically Rn~nr~ Ensembled Pharmacophore (BEEP).
FIG.24. The TSA Process of Finding and Relating Sets of Surface AEtributes of rVabTSA+ to Create a BEEP.
FIG.25. The TSA Process of Finding the Surface Common to All Active rVabTSA+ Scanners for an Active Site of a - Target.

~rrr~n DESCRIPTION OF T~r~ TNVE~IION

This inventio~ provides methoas and compositions for identifying ligands capable o~ identifying active sites on pharmacological targets. This invention utilizes .

-- . " ~

W096/04~7 ; PCT~S9~10182 2~6679 recombinant antibodies whic~ possess the combined attributes of potency (affinity), selectivity, specificity, and activity as reagents useful for modelling active ligands and identifying smalI~~organic molecules which also possess these attributes and therefore utility as drug leads or therapeutic compounds~
I. ph~rm~r~loqir~l T~rqets Iden~ified ~y Th;C Tnv~ntio~.
ph~rm~r~l ogical targetg may be receptors for~
endogenous or other ligands which evoke a physiological response by the cells on which the receptors are present.
Besides receptors, the pharmacological target may be ion rh~nn~l q, transport proteins, 7~h~q;nn proteins such as N-CAM, or any other physiological regulatory surface which i9 excessible to being identified by the rer~ l n~nt antibodies and which is activated by a specific ligand. A
non-limiting list of ~mplAry physiological ligands for which active surfaces may be identified by using the methods and compositions of this invention are listed in Example 4.
Receptors may include those for neurotransmitters, hormones, growth or trophic factors, modulatory peptides, ions or other moieties which act as signal ligands for the ph~rr-c~logical target. Preferred nonlimiting examples of neurotransmitter and peptide receptors for which active ~urfaces may be identified include those for acetylcholine, i.e., nicotinic, and the various forms of the muscarinic ml-5 receptor subtypes; adrenergic receptors including ~ 2, ~ 2,; ~rp~m;nrrgic receptors including D~, D2" D2b, D3 and D~, and D5; serotonin receptors including 5-~T~, 5-HT1AD' 5-HT2, 5-HT3, and 5-HT~;
benzodiazepine receptors; opiod receptors including O, K, and ~; and others. Also preferred are receptors for h,- ~ and growth factors~which may, for example, include those for insulin; growth hormone; erythropoetin;
neurotrophic factors, including but not limited to nerve growth factor, ciliary neurotrophic factor, brain derived W096~04557 2 1 ~ ~ 5 7 g PCT~S95/10182 neurotrophic factor, NT-3 and NT-4. ~eceptors for cytokines such as interferons, and the interleukins are also preferred as are receptors for nonpeptide hormones such as thyroid hormone, and glucocorticoids The methods and compositions of this invention described herein may be adapted by methods known in the art and applied generally to identifying the specific binding surfaces of other pharmacological targets as~well_ ~
Other target surfaces for which active ligands may be ;~Pntified include extracellular, ;ntracrllular~ nuclear or mitochondrial located soluble or membrane associated proteins, carbohydrates, lipids nucleic acids or complexes thereof which play a role in a physiological or pathophysiological process involving a predictable indication for which one would like to have a drug based therapy.
The ph~rm~rological targets according to this invention, are physiological molecules, or combinations of molecules associated through covalent or non-covalent _orces, which alone or in combination with other molecules, evoke a physiological or therapeutic response when activated by a ligand which binds the "active surface'~ of the pharmacological target. By "active surface" is meant the region of the pharmacological target which can bind a ligand, whether or not there are native endogenous ligands for these sites, and translate that binding into a physiological meaningful response rh~r~ctrristic of the target. Where the response re~uires oligomerization of at least two separate molecular ~ entities by a ligand, binding to the active surface on only one o_ the molecular entities is insufficient to evoke the physiological response.
The active surface is comprised of specific atoms or other chemical moieties which participate in the binding of the ligand to the ph~rm~r~l ogical target, for example by contributing to changes in enthalpy or entropy. The . .

W096104ss7 ~ PCT~S95/10182 7 ~ ~

active surface of the pharmacologicaI target may be small, capable of being bound by a single monovalent ligand having a molecular weight of less than about lO00 aaltons;
or large, re~uiring a multivalent ligand for binding to a plurality of binding sites which contribute=to the active surface. Multiple binding sites may be present in=a larger binding domain in a single region of-the pharmacological target. Alternatively, multiple binding sites may be present as separate non-contiguous regions which may be bound by a ligand capable of spanning the pharmacological target to simultaneously bind the different binding sites of the target. In addition, binding sites may be present on two or more molecular entities, which may be the same or different, and which re~uire oligomerization by binding to a multivalent ligand.
Growth Factors (GF), including ~GF, EGF, FGF, interleukin ~e.g. IL2, 4, 6) interferons, insulins and many other extracellular biosignals along with their respective receptor targets~a~d~ tly contain multiple target binding sites. Such protein signals are in the order of 20-lO00 X Daltons and exist as monomers or homo-or heterodimers or more complex multimers, which ~nr~mp~cc surface areas of tens of th~nc~n~c of A2.

Estimates of the surface area of such endogenous ligands and receptors which are occluded by their association ranges from 500-l600~2 By the above definition, each ligand has >2 bindin=g sites and each receptor has ,2 corresponding binding site which are discontinuous and non-overlapping with each other.

II. Use of Recombinant Ant;ho~C rVab's As Scanners To Identifv Active Snr~c~c This invention identifies and characterizes active surfaces by constructing and using a sufficiently large repertoire of diverse liganas capable of rscanning" the surface of pharmacological targets and binding to their W096/04sS7 219 6 ~ 7 ~ PCT~595/10182 active surfaces. Confirmation of binding to active surfaces is accomplished according to t~is invention by monitoring a change in function of the phArm~rnlogica target or by monitoring a biochemical or hiophysical change which reports binding and~or~activation of the pharmacolo~ical target or receptor on the target.
Ant;hofl;Pr have most of ~he ahove required attributes and can be recomhinantly engineered 80 as to acriuire uniriue attributes reruired for use in this invention It is well known that antibodies occur which are neutralizing and therefore by definition antagonistic in that they prevent, competitively or allosterically, the binding of signal to receptor, or receptor activity.
Antibody epitopes in protein targets range from a few amino acids to about 20 amino acids and cover from hundreds to thmlq~nflq A20f target surface In addition, epitopes can comprise se~uential or nnnrnnt;guouc groups of amino acids. However, it is e~ually clear that antibodies can recognize organic epitopes which are relegated to much smaller volumes, (i.e., <50-200A2) as are those associated most frequently with small organic haptins (i e., dinitrophenol or morphine). As antibody affinity and selectivity can be equal with both large and small epitopes, it is assumed that anti-target rVab antibodie~ will have landscape recognition surfaces which range over all of these A' ; nnr .
A. Use of rV~h T~; hrarieS
The repertoire of A; ffPrPnt ligands for scanning the pharmacological target ~ccnrfl;nJ to this invention is ~ provided by an antibody library comprising recombinant Fab fr~Jr~ntr, or portions thereof, constructed to present a sufficiently large repertoire of:different identifiable structures, some of which will be expected to bind and, depending on whether CU~U~LL~ binding to multiple sites is reriuired, activate the ph~rm~rnlogical target. These active antihodies are identified as specific memhers of a W096/04ss7 PCT~S95/10182 21~7~ --~ - 32 -library which may be considered to scan t~ ~nt;re surface of the pharmacological targe~ and possess the desired composite activity profile for the binding site.
According to this invention, the rec~mhin~nt antibodies used with this invention are referred to as "rVab" to indicate that they are constructed using recombinant techniques and are-made as libraries which incorporate diversified amino acid sequences in one or more regions of the antibody associated with target recognition or binding. ~ ~ ~
Where the pharmacological target comprises multiple binding sites on one molecular entity, or requires oligomerization of at least two molecules to form a single binding site with contributions from the individual subunits, or requires oligomerization of two or more molecular entities which each bind to the ligand at a different site, activity will only be observed using antibodies modified according to this invention to contain at least one additional separate binding e~tlty. In the preferred embodiment of this invention, the separate binding entity comprises at least one random sequence of amino acids having a structure a~ Liate to bind a binding site not bound by the antibody's variable region.
In some cases two such random sequences of amino acids would be reguired although it is contemplated that additional sequences may also be required. Additional binding sites on rVab can also be provided by more or less complicated prctein based structures including smailer peptides, larger proteins in~ ing intact enzymes or even another antibody, in st~lctures described in the literature such as diabodies (Winter et al. 1994).
Additional peptide sequences-which may be used to add additional binding sites preferably are between about 5 and 30 amino acids in length. More preferably such sequences are between about 6 and 12 amino acids in W096/045s7 PCT~59~/10~82 21~6679 , - 33 - ~
length. Most preferably, such sequences contain 8 amino acids.
An antibody identified as recognizing the binding - site is simultaneously or se~l~nt;Ally further characterized by determining its selectively and activity for the phArmArological target. To streamline an rVab selection process for=more than one target attribute, target specificity (T) and some of the activity (A) testing may be simultaneously characterized.
The order of isolating Vab for A+ and S+ can be varied, most often depending upon~which is the more difficult attribute to find among entities which modify the target of interest. For example, if selectivity among highly homogeneous target members of a single family is the critical missing attribute of existing agents, S+
could be determined first, or after isolation of the population which is A+.
Although antibodies which recognize (i.e. bind) the target's landscape in such a way as to modify its function make up a small percentage of those capable of passively rerognl7;ng the target (i.e. not modifying its activity), their presence is likely because of the size and diversity of the rVab library of the invention. In addition, active antibodies would also be expected to be present which have the additional desired attribute of specificity for that target. Furth~ ~ an o~;m~nt of this invention includes that the biological suitcases (i.e, phage or bacteria) used to individuaIly package each rVab library member allows their recoverability after a biological replicative cycle even if present in the original library in rare copies.
This invention utilizes recent advances in molecular biology which allow the generation and manipulation of sufficiently large and diverse V (VX and/or V~) region libraries, along with both m;nim~Ation and directed secondary diversification of their CDRs and CSRs to allow r ;:

W096/04SS7 2 1 ~ ~ 6 7 ~ PCT~S95/10182 . ;:, . .

selected rVabs, when l~h~lle~l to act as reporters and affinity selectors i~ assays which identify potential active ligands. Such active ligands are preferably small organic molecules which are useful as drug leads or as therapeutics themselves.
B. Use Of rVAB Members To Iaentify Small Organic Molecule Re~lacements (SOMERS) At least two methods are provided fpr identifying SOMERS based on the identification of the recombinant antibodies (rVab's) possessing the attributes of T
(specificity/potency) S (seiectively) and A (activity).
According to one method of the invention, these [rVab T+
S+ A+] scanners are converted to reagents for reporting the presence of other ligands capable of binding to the active site on the pharmacological target. Conversion to reporters is accomplished by labelling the active scanners with a detectable label. The reporter rVab fr5~mPnt~ may then be used in classical competitive binding assays to idectify SOMERS. For simple active surfaces, single SOMERS represent active small organic molecules, while for complex active surfaces c~nt~in1ng more than one ligand binding site, corresponding numoers of SOMERS, found in the fashion disclosed by this invention, are covalently coupled together to represent the active small organic molecules In another embodiment o~ this i~vention, SOME~S are identified based on the col~ective attributes of an ensemble of active rVab scanners which have been characterized as T+, S+ and A+. By providing a sufficiently large repertoire of antibodies, multiple antibodies possessing these de9ired binding attributes are expected to be identified. Common structural features of this ensemble of scanners possessing the desired CAP
attributes are then used to con9truct a model ligand for binding to the active surface of pharmasolo~isal target.

w096/04ss7 6 ~ ~ PCT~S9~10182 ~ . , , By n~ h;n;ng structural features of multiple antibodies ~ identified as being active for a specific ph~rm~nl ogical target, biologically enhanced ensem~ble pharmacophores (nl3EEPSn), i.e., drug models, may be derived which may then be used to identify small organic molecules as drug leads or therapeutics This molecular model, ~EEP, then serves to provide a basis for screening chemical databases to identify SOMERS either by electronic screening of available chemical data bases, or as a basis for rational drug design to synthesize SOMERS expected to possess the combined attributes of specificity/potency, selectively and activity. This solves the prior art problem of access to all compounds within=a chemical data base, decreases the time needed for screening and amount of manpower necessary, and could ~l ;m; n~te screening if used to direct a synthetic chemistry effort to create SOMERS.

III. Recombinant Antibody Libraries Provide Sufficiently Large~ Re-pertoires Of Different Ligands To Identify Active Snrfaces A. Eunction Of Resnmhin~nt J,.; hr5ry The objects of the invention are provided by a process which makes and then uses separately and/or in batch mode, combinatorial repertoire libraries of variable regions (V~ and/or VL) of r~rnmh;n~nt antibodies (rVab) to scan the surface of a pharmacological target so as to identify and select those which have a desired potency, specificity, selectivity and activity profile. These four attributes are collectively defined herein as the compound activity profile (CAP). Members of the library possessing the desired attributes are then grouped according to the local surface domain recognized. ~y using a sufflciently large and diverse library as described herein it is expected that essentially most if not alI relevant active surface of pharmacological targets should be identifiable using the=method of this inventiQn_ In addition, because the library is r~nl ';n~ntly made in a random fashion and 4 . . ~

W096/04557 PCT~595/10182 21~7~ --selected in vitro, recognition of sites which would not otherwise be detected as non-self, or antigenic, or immunogenic, should occur using the rVab library described in this invention.
In addition, the objects of the invention relating to discovery of three dimensional shapes of surface areas are provided by use of the active rVab's of this invention as reporters of target structure. As described below, these rVab reporters are constructed using VX and V~ domains wherein the CDR regions which may be diversified are cnntA1n~a in a fLom_..JLh of an Ig (or Fab) having a three dimensional structure which has been=determined by crystallography has CDRs which contain the known canonical structures:~(CSRs) Such structural information about rVabs for a given delimited active target surface domain allows for the molecular resolution ana deduction of the essential elements of the rigid organic structure of the constellation of critical amino acids constituting the active target surface recognition~portions of the ensemble of active rVabs and thereby provide the essential elements of the rigid organic structure of active SOMERS which can bind with specificity to and modify that target.
Construction of the ~EEP requires PCR det~rminAtinn of the amino acid sequences of rVab CDR, CSR and some fL~I_..JLh residues in these active surface scanners through a process which uses computers and genetic~
algorithms. It is also possible that with sufficiently large enough active rVabs, the information obtained in the above manner will enable resolution o~ the active surface of the targets. This process provides the objects of the invention related to electronic screening for SOMERS by cnmhin;ng common structural elements in~ ~ Ational packages called biologically enhanced ensemble pharmacophores (i.e., ~EEPS).
The r~cnmhlnAnt antibodies used in accordance with this irvention also provide an ilLI~L~v~ll~llL over the prior WO 96/04557 219 ~ ~ 7 9 PCT/US9~i~10182 art of typical l~hPllP~ target-reporter binding assay screens. One l~ uv ~~~ comprises obtaining via recl ' tn~nt molecular biology technology, antibody ~ variable regions (V) in sufficient numbers, with sufficieit affinity and desired activity so as to identify those members of the library which function as surface reporters capable o~:recognizing active target surfaces, modulating the target through these recognition sites and distinguishing its target from amQng closely related targets Iselectivity).
B. Size Of ~ecombinant BibrarY
In order to have a sufficien~ l;kPl;hnod of identifying the active surface of a pharmacological target, the recombinant library preferably rnnt~tnc at least between about 109 and 1014 entities. Preferably the IS library rnnt~;nr between about 10l~ and 1013 e~tities. Most preferably, the library cnnt~;n~ about 10l2 entities. The specific size of the library required to provide a reasonable l;kPl;~nod of identifying the active site will depend on the overall surface area of the target surface ~ and the surface area of the binding domain to be ~Pnt;f;ed The surface of most targets is of the order of 50,000-100,000 A2, with each ligand binding domain encompassing from about 100-200 A' to about 1,000-2,000 A'.
As each rVab covers only about 20-40 ~' of surface area, one reriuires about 2,000 rVab's to cover the target landscape, ana at least 10 times that (2x104) allowing for o~erlapping recognition~domains. Another increase of two orders of magnitude (2x106) allows for d~Lu~Liate surface interactions which produce specifIc agonist or antagQnist action. Another 100 fold Increase allows for such rVabs to be recoverable from the library upon batch analysis.
~ An additional I04-104 fold increase allows n~nn~nl~r affinities and agonistic activity. Accordingly, the ~ preferred useful surface scanning libraries have on the 3S order of about 10l2 entities.

W096/04sS7 PCT~S95/10182 , 2~9~7~

It is recognized that antibodies have the ability to distinguish among closely related targets. Accordlngly, recombinant libraries possessing sufficient numbers of entities are reached according to this invention by constructing rec-ombinant libraries comprising variable re~ions=of either, or both light (~) or heavy ~H) chains which are modified or unmodified and which may or may not be expressed ln combination with a constant region. These libraries may be selectively varied not only during their===
original construction, but also after the initial round of selection for any one or all of ~he three composite profile activities of target binding, selectivity and activity. Such secondary additional divers~fication as well as secondary simplification may be carried out by combinations of primer based PCR or oligonucleotide insertion at convenient restriction sites. Furthermore, the secondary variations may be localized to each of the 6 CDRs (i.e. the three in VB and the three in V~) or any particular combiration or singular location. Variability is introduced in-the CDR'9 by modifying the CDRs to contain random amino acid substitutions of positions involved in contact with the target. The positions of variation, including further diversification or simplification, are preferentially tkose within the CDR
which do not alter the CSR structure of that region and are known to those skilled in the art. The number of amino acid positions to be diversifled is dependent on tke number of active rVab members desired to be obtained.
Thus, if an insufficient number of members are identified, the library diversification can be increased by diversifying additional amino acid positions in a CDR as described below.
Given that there are twenty naturally occurring amino acids, diversification at a single amino acid position results in about 20 different potential antigen binding (touch) sites. By diversifying at two amino acid W0961045s7 2 1 9 ~ 6 ~ 9 SgS/10182 o positions in each of the 3 V~ CSR, 2 VH CSR, and the one V~ CDRH3 which are randomly combined into VH:V~ pairs by the invention, one obtains a diversity in the rVab library ~ of 210l~ members (see Fig. 4). Since a given phage library can package about lX10l4 members, several libraries are preferably constructed and packaged in phage to contain the entire population of diversified members. Although, it is preferable to diversify two amino acids in each CDR
as shown in Figure 6, other combinations are possible.
Randomization in only some of the CSRs and one CDR allows for libr~y sizes approximating 1012 such that one phage rVab library could con~ain multiple copies of each divPr~ f ~ P~ member. In addition, three or more non-essential amino acids in a given CDR may be diversified ~see, Figure 5 for non-essential amino acids) preferably with a corresponding decrease in divers; f; r~t;on of amino acids in other CDRs so as to ~-;nt~;n the total size of the library within an attainable number. Resultant libraries of 4X10l2 members can be approached using, for example, bacteriophage as vectors. A single rVab library of this invention of at least about 10l2 members, ln~PpPn~Pntly=of how diversity is obtained, provides enough surface probes with the minimum CAP at the target to allow ;~Pnt;fi~tion of most active surfaces of interest.
An adva~tage of this invention over prior art screening methods is that it scans the entire available surface of the target for active surfaces and provides active sur~ace reporters. This allows for identification of active sites and SOMERS for targets without endogenous signals (endogenous ligand) and:at~target surfaces not used by natural endogenous ligands but which result in modulatlon of that target. The latter surfaces, referred to as allosteric surfaces, are of two types: those without activity in the absence of endogenous ligand binding to target (i.e., cryptic allosteric sites); and those having W096/04ss7 PCT~S95/1~182 2 1 ~ ~ ~ 7 ~i - 40 - ~ .
activity on their own and yet are still able to modify the action of an endogenous ligand (i.e., active allosteric sites). Obviously, the larger the target surface under scrutiny, the greater the opportunity of finding appropriate active surfaces. As endogenous ligand contact surfaces probably represent some lO~ of total target surface area, including allosteric surfaces greatly increases the surface area under investigation.
The use of re~ombinant libraries also provides a means of reduclng or increasing the number of complementary det~rm;n;ng regions (CDR~ within the variable domain of the rVab n~rr,qq~ry to confer desired CAP attributes to the rVab. Thus, one can attain a minimal active CDR complement. Alternatively, large scale r~n~rm;7~tion of up to most of the amino acids within the rVab CDRH3 domain may be used to increase the population of active rVab from which to identify the best rVab reporter. For example, if the initial library screened does not possess members with the sufficient constellation of CAP attributes, secondary diversification of the best r~n~ t~q1 by a number of procedures including PCR and various in vivo and Yi~Q mutagenesis systems known to those skilled in the art, and then recycling through the oriyinal identification and selection procedures described below, may be used to recover an antitarget rVab with a full complement of the aesired CAP which might have been too rare to be found among the original:antitarget rVab library. In addition, by identifying and sequencing active rVab CDR complements one may also obtain accurate and detailed structural information useful for modeling the essential elements of active SOMERS, i.e., as in BEEPS.
C. Aff;n;ty Of Recrmh~n~nt Ant;hrS;e5 ~ The rVab of the invention are used to detect and characterize active sites by providing information related to their structure, and/or to function as reporters in 2196~
W096l04ss7 PCT~S9SJl0182 competition assays to identify SOMERS. Accordingly, the affinity of the rVab's useful in this invention should allow for one:or both of t~ese functions. I~ the rVab is used only to detect and characterize the active binding site or to rnntr;h~lte in developing a BEEP, its affinity may be high and a slow dissociatiQn rate (i.e., half time of dissociation, preferably between about 5 and 30) would be suitable ~owever~ the affinity of the rVab's useful to identify SOMERS for a ph~rr-rnlogical target should not be so high as to prevent dissosiation and competition for use in sompetition assays. Preferably, this affinity will be in the range of from about 0.0~ to about 100 nM. More preferable the affinity will be between about 0.1 to about 30 nM Even more preferably the affinity will be about 0.
~-10 nM. Most preferably, the affinity will be between about 1-5 nM.

D. Characterization of Ligand And Target Binding Sites ~inding domains on a signal are referred to as ligand att~rhm~nt sites (LIGATTS) and those on the target as target att~rh~nt sites (TARGATTS). Where each is protein in nature, both can be defined as the eurface area of the entity made up of contiguous (e.g., amino acids n and n+1) or discontiguous (e.g., amino acids n and i where i is not n+1) ~l~mentq~so confined in space as to be accessible and in contact at the same time with the surface of the other partner in the complex so as to contribute to the binding energy of that interaction. Where there are multiple binding sites, by our definition, each TARGATT domain forms contact points with amino acids on the signal and one SOMER would not be expected to rnrnmr~rq two LIGATTS.
Where endogenous ligands are nonproteinaceous, other compound bn;l~;nr blocks would replace the amino acid as the unit entity.

W096/04557 2 ~ 9 PCT~S95110182 The sizes of ~IGATTS and TARGATTS are ~uite variable.
We have arbitrarily confined TARGATTS to the volume which can be ~n~mp~q5ed by a syn~hetic small organic molecule r~plsc~m~nt (i.e., SOMER) of less than about lkD. This TARGATT size, is practical and modeled by the opiate receptors' attachment site for its 30 amino acid endogenous ligand, endorphin, which easily binds morphine (<600D) and all of the pharmaceutically known opiate analgesics, with nM affinity and is fully activated by their attachment. Identification and characterization of larger TARGATTS is considered within the scope of=this invention as such sites should also be recognized by members of rVab libraries~o~ this invention.

~ sociation of Activity A+ With 3indinq of rVab An important feature of this invention is that the rVab's which are ;~nt;f;~ as possessing the desired CAP
attributes and in particular, activity at a target, function to create a linkage between binding to a target and activity at that target. Accordingly, once an rVab is identified which is both T+ and A+, that rVab may then be used to identify other ligands which are also T+ and A+
based on competition binding assays alone.
Several methods are available to initially provide a connection between binding and activity Of- a rVab. In a preferred method, an active surface for~a target is associated with a secondary biochemical response which may be detected upon binding of~an active ligand at the active surface. Such biochemical responses~may include changes in affinity of the ligand or allosteric ligands, oligomerization with other subunits, phosphorylatiQn state, ion flux, etc. For example, ana as discuss'e'd more fully below, the changes in agonist affinity of a receptor coupled to G protein based on the presence of a guanine nucleotide can provide the necessary linkage between binding and activity.

W09~04557 PCT~S95110182 ~ 21~679 o Also, as discussed in U.S. Patent 4,859,609, which is ~ incorporated herein by reference, receptors may be expressed as ~usion proteins comprising the ligand binding domain of the receptor fused to a "reporter" polypeptide which undergoes an assayable change in conformation or function when the active ligand binding domain of the receptor binds to an agonist or antagonist.

IV. Method of Identifyinq SOMFRg The method of obtaining small organic molecules (SOMERS) which are active at pharmacologic targets is summarized as comprising the following (See FIG l):
Staqe I (a): Construction of the scanning rVab library.
Staqe ~ (b,c): T~ntlflcation of rVab's which bind and activate target. If target is a multivalent site requiring att~r' at two sites, pairs of rVab's are identified using rVab-peptide scanners to detect activity.
Staqe II: Use labelled rVab's as reporters to detect SOMERS or MULTIMERS (i.e., DISOMERS).
Stage ~lI: Create 3EEPS from composite of structural information derived from rVabTSA+ for screening or synthesizing SOMERS or MULTIMERS.
A. Construction of Scanninq rVab Library (Staqe la) Molecular biology te~nnlogy is used to construct a limited number of large combinatorial libraries of recombinant antibodies (rVab libraries) wherein the VL and VH CSRs and CDRH3 occur within each library within a single Ig VH and V~ framework, respectively, and optionally attached to their respective constant region (CHl and CL). An antibody whose structure has been ~ determined by crystallography is preferably used to provide the f .IJLh for construction for these rVab libraries. Antibodies of undetPrmln~fl structure can also be used for library construction and identification of active rVab8 (i.e., Stage l abc, Fig. I) useful as W096l04557 PCT~S95/10182 reporter rVabs to detect SOMERS and other MUBTIMERS (Stage II Fig. 1) according to the process of the invention, but only antibodies of determined structure can be used in creation of BEEPS (Stage III, Fig. 1).
In the preferred '~imPn~ of the invention, antibodies of solved structure are~used to create the original rVab library. In another embodiment, one or two of the isolated active rVabs for a given target are subse~uently crystalli~ed and the structure ~ptprm~np~ to allow their use in Stage III. The later is useful as it allows use of the newly published sequences of the human VH and V~ genes [Tomlinson et al. 1992; Williams and Winter~1993; Cox, Tomlinson and Winter 1994; Nissim et al.
1994; Tomlinson et al. 1994] for Stage III work.
In all cases, the rVab lipraries constructed Py the process of the invention have a sufficient numper of diverse memPers to encompass an immunological antigenic repertoire approaching man's natural one or are made from human VH and V~ genes [Roitt, 1991; Nossal 1993; Griffiths et al., 1994] which are capable of reAo~n~7;ng an enormous diversity of surfaces including but not restricted to proteins, nucleic acids, carbohydrates, lipids and organic haptens.
There are basically three sources of genes to Pe used as the starting material for construction the rVab libraries.
a) the pnhliqhP~ data on cloned and sequenced antibodies;
b) the antibody clone~s themselves, carried ln various cell types, including hybridomas, spleen cells, bacterial plant cells, yeast and viruses, on various DNAs ;n~ ing p1~qmi~, phagmids and chromosomes; and most recently c) the published sequences of a human repertoire of V~ and VL genes [Roitt, 1991; Tomlinson et al. 1992;
Nossal 1993; Williams and Winter 1993; Cox, Tomlinson and W096/045S7 2 1 9 6 6 7 ~ ' PCT~59Sl101~2 Winter 1994; Griffiths et al., 1994; Nissim et al. 1994;
~omlinson et al. 1994].
Most of the se~uence information is available in at least two data bases, i.e., the Brookhaven Protein data base and that of Kabat at NIH (which is also available in text form) [~abat et al. 1991]. The structure of the ma~ority of the crystallized antibodies is also available from the Brookhaven Protein data base. histings of such crystallized antibodies are presented in Example 1. An example of an antibody which has been crystallized to determine its structure is described in (Tulip et al. ~, Mol. Bio., (1992) 227:149-150).
In the preferred embodiment, the antibody sequence is obtained first and is the starting point of rVab library construction using the following steps to construct the rVab library. The order of steps may be varied to suit particular circumstances.
I. Selection of Parental Fabs of known crystalline strUctllre sq rV~h l;hr~ry rL~ .7J~h t~-Cl~t,es II. ~rPs t; ng the Nucleic ~cids Encoding the Heavy ' and Light Chains (rVHCH1 and rVLCL) for ABXXX
rVsh.lib.
Stev l~a): Cr~nctructiQn of 5'VL Section Step 2 r Diversificati~n By PCR

Stev l(b): ~ nctru~tlQn of th~~ MIDVh section Step l(c): C~nctruction Or th~ 3'Vh section Q;~
~L~L~l higation , III Construction of the Constant reqi~nC o~ ~R~
IV. Construction Of rVHCHl.lib (Fiq. 8) Conqtruction of 5' hAl~ of the VH Region C~cnctruction of the 3' H~~f of th~ VH Resion V. V~ and Vh library sizes:

W096/045~7 PCT~S9~/10182 ~lg667q f VI Construction of the rVab.lib ~the V~CHllib x VLCTfl;h Cl ;n~tor;~l l;h,~ (F~q.ll.l2.l4) SteT~ 4: In vivo recnmh;n~tion of V~T~TTl snfl VLCL
qenes Details of the Individual Steps for Expressing the rVLCL l.6 and rVHCHl.L.b by CRE-LOX
yT~C~MRTN~TORT~T, FORMA~ION
VI Step ~ - Generating Phage and~Displaying the rV~h.lib on Phaqe Surfaces (Fiq. I4) The critical steps are~shown in Figs. 7, ~, ll and 14 which describe respectively the cQnstruction of rVLCL and rVHCHl libraries, their pairing in the rVab library, and finally their expression attached to the surface of~phage as functional complexes.
Both construction of the rVLCL and rVXCHl libraries follow a similar outline wherein:
a. a limited number of oligonucleotides are synthesized rnn~A;n~ng convenient restriction sites and which cover both ends, and in one case the middle domain, of the V region, b. the oligonucleotides are ligated together, c. PCR is used to append missing and ~unctional regions as well as provide the means of r~nfl~ 7stion of amino acids at defined positions, d. the completed rVX and rVL libraries are ligated to ~Lu~Llate constant domains wherein one-library is placed within a plasmid and the other phagmid, and e the rVH and rVL libraries-are rrmhin~fl in vivo by the CRE-LOX recombinase provided by coinfection by Pl.
Following this outline, rVab libraries of about l012 ~ members are constructed.
In other embodiments, ~ a the VH and VB gene~s, wiEhout constant regions, encoding an antibody of known structure are cloned via PCR
to obtain the se~uences encoding the VHCHl and VLCL
sections of the lgs using methods known to those in skilled in the art, and W096/04557 ~ PCT~S9~110182 -- 21~667~
. . 1 .

b. the Vs may then be altered via PCR to remove ~ unwanted restriction sites, and develop convenient restriction sites bording the CSR and CDR domalns.
c. selectively r~n~1 'zPfl oligonucleotides with appropriate end pnc;tinn~l restriction sites may be used 5 to replace each of the 6 CDR regions having appropriate matching restriction sites in the basic V f~ h to allow directional cloning. These oligonucleotides vary in length (i.e., n, n+l and n+2) to match the known CSR and some length changes in CDRX3 and contain all of the amino acids at one or two positions within each CDR most often involved in antigen contact.
In the preferred and other embodiment, with 2 amino acid r~nflnm~7~t;ons within each CSR and CDRH3 and three rPnt lengths of CDRX3 used, the numbers of diverse 15 members in the final rVab BIB ~i.e., rVHCHl x rVLC~) reach lO18 (see Fig. 4 for details).

l. Sources of FL~ "JL~
FL ..~rh~ in which the optimally diversified CSRs 20 and CDRH3 are cloned into may be derived from antibodies of hnown structure.
FL~ Lh~ may be chosen from antibodies which present the canonical regions in different orientations with respect to the C region. Thus, it may be desirable 25 to prepare multiple rVab libraries on different f ~ Lha to maximize different special orientations of the CDR's.
FL~.~..~Lha may be chosen which will favor binding over small to large surface areas. As discussed above, a small surface area would cover an area of about 200 ~2, a medium surface area about 750 ~2 and a large surface area about 1500 A2 Examples of ~ntihnfl~es which can provide ~ fL~,~.~JLha for these three different size targets are found among the planar, cavity and grooved type antigen . ;recognition domain present in various antibodies of hnown structure (Fig. 3 respectively). Fr~r ..~Lhs may be chosen . ~

W096/045s7 PCT~'iS95/10182 219667d -~48 -slmply based on the shapes of the antigen recognition domain or in combination with other structural ~actors.
2. The Ey~ressihle V~h Region Cnnctxuct Preferentially, construction may be done in one of two general type vectors, a. fd and M13 (Pharmacia, USA [Smith, 1985; Scott and Smith, 1988; Parmley and Smith 1988; Cwirla, et al., 1990, McCafferty, et al.~1990; Winter and Milstein 1991;
Waterhouse, et al. 1993, Recombinant Phage Antibody System Instruction Manual, Pharmacia P-L Biochemicals, USA].
i. the inserted V(H and h) with CH1 at the carboxy terminus preceded by the lac promoter and a ribocomal binding site [RBS], an export leader sequence in front of gpIII phage coa~ protein or EielB, a cloning site followed by either an in frame'linker and then gpIII, or a double set of suppressible termination codons.
ii. the VH or VL without C~1 or CL or with partial NH2 terminal constant region amino acids may be preceded by the lac promoter -RBS-PelB-with internal cloning sites allowing in frame ligation of VH at both 5 and 3' ends and followed by -C(H or L) and either an in frame linkage to gpIII or two guppressible t~r~;n~t;nn codons .
b. , n~pII (lambda) Stratacycte, CA, [Skerra, and Pluckthun 1988; Mulinax, et al. 1990, T~m-lno~p Cloring ~it, Instruction Manual Stratacyte Corp. CA USA;
~ang, et al_ 1991; and Barbas, et al., 1991]
i. as above for V region, with and without intact CH1.
ii as above for V regio~, with and without intact CH1 Expression of Single V(H, L) -C~H or L).
Expression of single V(H or L)-C peptide9 may be used to confirm proper con8truction of the V regions, or rVHCH
and rVLCL libraries, before either expression a9 mature VC

W096/04557 PCT~S95110182 21g667~9 c~ ,., O - ~9 (rVHCH or rVBCB) or CRE-BOX rFrnmh;nAt;n~ and phage ~ expression. ~d (~13) or Lambda expressiqn is lnduced with glucose a~ des~ribed in phArm-~;A ~USA) Kit~ or the Stratacyte (CA) ~y~tem Berner. The product may be ;fl~nt~f;~ wi~h CHl antibody (stanaard Elisa technology known to tho~e ~killed in the art) either with fd a~ phage di~played molecule~, or wIth l~mbda after expres~ion induction, and generation of peripla~mic located molecule~. When using phage,1the induction of the lytic cycle may also be u~ed to determine the ratio of lam.~bda to intact rV a~ an indication of ~ize of library. With fd, one can assay antibiotic (e.g. ampicillin) re~i~tance colony forming unit~ (cpu) tran~fer from within fd genome v~. the number of phage with rV display attached to the viral ~urface. Di~he~ coated with viral or rV antigen may be u~ed to provide information on the ~ize of the rV
library.
In another e-m-bodiment~ only the rVH and rVB domains are expre~~ed and connected through a flexible linker to form a ~ingle=chain V region antibody (termed ~cFv by Winter [Huston, et al. 1988; Bird, et al. 1988;
McCafferty, et al. l990; Hoggenboom, et al. l99l; Barba~, et al. l99l; Garrard, et al. l99l; Breitling, et al. l99l]
which may be expressed using phage di~play. The expres~ed V antibodie~ are fu~ed to gIII on Ml3 u~ing a Recombinant Phage Antibody Sy~tem Kit~ (Pharacia, USA), according to in~truction~ provided the m~nnfActllrerer for con~truction, expression and~detection.
c. General information on ~rimer u~e and PCR.
To allow the library con~truction of variou~ domain~
of rVH and rVB, and CH and CB as well, each primer include~ a ~equence encoding a re~triction ~n~nml~lea~e recognition ~ite. The ~eguence of the primer which cnntA;n~ the restriction ~ite may be located within, partially within, and ~ometimes precede~ the ~ection of the primer AnnPAl ;ng to the target Vab ~eguence. When it W096/04s57 ~ PCT~S95/10182 219667~ ' ' ~

is present as an extension to the sequence homologous to the rU section under construction, it will not participate in ~nn~l;nr duriny first strand forward and second strand reverse synthesis but will participate in annealing subsequent PCR amplification cycles. Although not essential, the restriction sites (at either or both ends) are such as to generate 3' or 5' overhangs to aid in subsequent ligation ut;~;7;ng restriction enzymes which m~;nt~in the appropriate reading frames. Products of PCR
may be isolated from the reaction mixtures by a variety of techniques known to those skilled in the art. A number o~
restriction sites which have been successfuIly encoded within rVH and rV~ gene constructs for insertion in the available expression vectors are known to those skilled in the art and are available from manufacturers of IG
expression systems and Ig primers such as ph~rm~rl~ (USA), Stratacyte (CA), and ~'-3' Prime (USA).
Insertion in frame can be into vectors r~nt~;ning sequences encoding other proteins to produce fusion proteins not only containing one or more C constant regions, but also the coat protein gpIII and VIII of fd filamentous phage, or t,~,~ ...',r~n~ proteins to provide rVHCH or rC~V~ anchoring for d~L~Liate extracellular or phage displays.

3. Preparation Of rVabs With Multiple Att~rh~-nt Sites The grouping of active rVabs based on recognition of different target surface domains is simplified by using small peptides which cover in an overlapping fashion, the liner amino acid sequence of the target. Such grouping simplifies the pairing o~ açtive rVab for a M~TIMER (e.g.
DISOMER or TRISOMER) obtained from multlvalent rVab-PEP
libraries (example 3 and 4) as well as forms the basis o~
selection of active rVab for conversion to reporters for simple SOMER identification.

W096/045s7 2 PCT~S95/10182 ~ I ~ 6 6 7 ~

O
Glven that many antigenic sites are less than 12 ~ amino aclds, peptides of 10-20 amino acids, made in overlapping fashion (i.e amino acids 1-15, 5-20 10-25 ~ etc.) would provide most of the se~uential target epitopes. This would mean that=for an average protein of 50,000 Rd, i.e., some 90 would be~needed to cover the entire surface. For many pharmaceutical targets, mutagenesis and alanine scanning has provided information, known to those skilled in the art, of particular amino acids, and small groups of~amino acids which are involved in signal binding and receptor activation. Such information is used here to reduce to a much smaller number the peptides needed to provide most of the desired surface epitope information. Another possibility for target fra~ t~t;on is the use of synthetic polypeptides, bought commercially or produced by biotechnology means, using commercially available expression vectors harboring specific sites for cloning and expression of peptides in fusion with easiIy and t~uantitatively recoverable proteins.

4. CSR ~n~ rnR Diversificatit~n ~nt1 Rt~t1nt~tion CSR and CDR R~ntlt 7~tion: A preferred embodiment will be to use synthetic oligonucleotides which vary at increasing number of amino acid positions within each CSR
and CDREB but which do not alter the CSR Minimal r~n~tm;7~t;tn of amino acids would be to have only 1 position within each CDR filled with all 20 amino acids.
One could include up to about 24 amino acid posltions within the CDR ~3. As the number of positions randomized increases, the total possible different rVH and rVL
rapidly exceeds the practical limitation of 10l~-l4 on phage library size, and one has to limit the number to fit within the library size that is attainable. Increased r~nt~t~i7~tion at larger number of positions can be acc~ tht~d by putting amino aclds lnto classes, i.e., W096/04557 PCT~S95/10182 21~6679 basic, acid, hydrophobic, hydrophilic, etc., and then usiny only one or two amino aclds of each group at each 'randomized' position. Secondly, since not every amino acid within a CDR is involved in contact, one can identify those which are most often involved in contact and focus amino acid r~nflnm17~tion at those positions. Lastly, one does not need to use the same type or degree of r~nflnmi7~tion for all CSR and CDRH3s. In one e~mbodiment, one could~use only CSRHl, H2 and H3 for randomization as VHs alone have been pnhl;ch~fl to have nM antigen affinity [Ward, E.S. et al. 1989].
In the preferred embodiment, r~nflnmi7~tinn ~may be accl pl;~h~d during construction of the rVab library. In addition, secondary r~nfl, '7~tion after isolation of the initial active rVabs may also be utilized if desirable.
Secondary r~nfll 7~t;on can be used to obtain a single, or pairs of missing attributes of the desired TSA CAP, or to increase or decrease one or more present CAP attributes.
CDR Reduction: To determine the smallest target binding domain it may be desirable to reduce the size of the potential rVab target binding domain For CSR and CDR
reduction there i9 the posslbility of using only one VH or one V~, making PCR copies, cloning with primers which include only the first, first two, or last one or two CSR
and CDRs within rVH and rV~, and subse~uently ligating the constructs into parental f ' ~.~ Lk~ wherein the missing CSR or CDR has been replaced with a string of glycines (Winter EP 0 368 684 A1). After alteration each library may be retested for its new CAP. In another approach, one can start with a preferred rVH:rVD pair and~delete (again r~pls~;ng each with a glycine heximer) a) one CSR or CDRH3 at a time (there being 5 such possibilities); b) two at a time (there being 14 such possibilities); c) three at time (there being 9 of these); and d) four at a time (there~
being 6 of these). With reduction in CSR and or CDRs, the potency of the altered rVab can be tolerated up to 30nM, -WO 9611)4557 ,~ J..,J~ 6~
~ 21' 9 b~ 6r ~

(that re~uired for use of the rVab in subsequent binding screens for organic r~plsc tR). ~owever, an affinity of lOD~nM is tolerable in the minimal CSRjrCDR cl '~n~t;on if it is put through mutagenesis for potency ; ~ ~vr~ Ls later on as such processes have been shown to produce increases in binding affinity of up to two oraers of magnitude r~ass, Greene and Wells, 1990; Marks, et al.
1991] The reduction in number places all of the critical contact atoms within the smallest number of semifixed domains making 3D mn~r~1; ng of critical atomic spacial 0 rr~l ~t; nnRh;ps easier by means known to those skilled in the art.

5 Ry~ression of rV~hR
a. ~Ynression of rV~h ~R a rh~ge l;hrAry, In one ~mhn~ of this invention, the rVab library is displayed on phage This process i8 best described recently by Griffiths et al. 1994~. Methods for using phage display of antibodies have previously been published (see, Ladner et al., Int~rn~t;nn~l patent application W090/02809; Winter et al. W092/20791 and Huse et al.
W092/06204 which are incorporated herein by reference) and some reagents are commercially available in kits.
In another n~; ~, only the rVLCL is placed in a library for expression as a bacterial plasmid construct (VLCL.bact) with a leader which allows product release to the periplasmic space. This library is then expressed and product is combined yith either one of a rVHCEL or rVLCL
phage displayed library to derive the two phage and one soluble protein libraries. An anti-CL antibody attached to solid matrix may be used to harvest their VLCL protein ~ library.
To identify members of the rVHCH1 phage library with one or more CAP attributes, the soluble rVLCL protein library is added to the above phage library and panned for target surface recognition with target protein attached to W096l045~7 ~ ~ t , PCT~S9~10182 "~

a matrix (plastic, chromatographic or magnetized beads) in the absence or presence of competing proteins ~see example 2) to derive rv~r~l rvBcL protein T+(S+) members. The phage ~nt~lnlng the rVXC~1 gene is harvested afte~
allowing the phage to multiply with helper ~using commercial kits). Isolation and enrichment steps may be repeated as re~uired. This~library may be referred to as T+S+rVHCHhalfLIB. Assays fOr A+ may be~then be done to obtain TsA+rvHrH~AlfT~TT~.
The T+(S+A+optional)rVHCHhalfLIB may then be cloned, in, for example, lambda and expressed as periplasm soluble entities. The library may then be mixed with the phage display rVLCL.~IB, and the above isolation steps repeated to obtain a T+(S+A+)rVLCLhalfLIB. The specific methodology for this procedure has been published by Lerner and group [Cabilly, et al. 1984; Burton, et al.
1988; Huse, et al. 1989; Mullinax et al. 1990; Zebedee, et al. 1992; Tmmllffn~r Cloning Kit Stratacyte Corp. CA., and SurfZap Cloning ~it (instruction manual) Stratagene Corp, CA] and is herein included in entirety by reference. See below section on Functional VH and V~ combinations for details. The active rVHCHhalf library is cloned into pVHACCEPTOR (see Fig. 11) and the active rVLCHhalf library into pVLACCEPTOR. The CRE-LOX recombination system may then be used to derive a rVab LIB combinatorial library which may be tested for the.TSA+CAP.

b. Isolation Of rVab Library 0~ Target Binders And Ph~qe DisplaY
This step isolates all rVab existing within the original library which recognize some part of the target's surface and form a complex with sufficient stability for isolation (i.e, target affinity ~30nM). Those with this recognition ability are termed T+ In the preferred method, the rVab genes are mixed and packaged in and displayed functionally on phage surfaces. Accordingly, WO 96104557 2 1 9 6 6 7 ~ PCl~/US95/10182 ~ . ~

rVab are displayed on the surface of phage and the phage are incubated with target surfaces. In other embn~;- t~, the library number can be reduced by prior selection of - active rv~TrT~hAlfT~TR and rvTlrR~AlfRTR which allows pac~aging and expression in bacteria, either as soluble or membrane anchored rVabs by methods known to those skilled in the art using commercially available kits (e.g.
Stratacyte USA) following manufactures directions.
As discussed above, the target can be any surface one desires to scan _or recognition by members of the rVab phage library. Permissible incubation conditions, of which there are many known to those skilled in the art, would include those which do not disrupt the vehicle packaging the~rVab, or inactivate rVab recognition of the target, nor prevent display of its target epitopes. In lS addition, in all cases the rVab:target complex preferably is one which is ~uantitatively separable from free Tr-rVab phage packages.
After ;nrnhAt;nn of target and the rVab phage, there are many pnh1;~hPd methods for separation of complexes known to those skilled in the art which are all based on the principle of having the target tagged (denoted Tr~) in such a form as to allow its convenient ~uantitative separation from all reactio~ soluteE. Preferably such tags are lnseparable or act as labels to follow the target:rVab complexes through separation procedures.
Among such preferred tags are matrixes such as agarose, magnetic beads and the surface of culture dishes. In these cases attArhmrrt of the target to the tag would have been made prior to incubation with the rVab library.
There are also non-matrix target-tags which allow ~ target:rVab complex separation from solute and unbound rVab. Among such tags are fluorescent compounds, (for use in fluorescent activated sorting), biotin, (for avidin directed sorting) and polyhistine cnnt~A;n;ng 6 residues [his 6] (for metal chelate column~chromatography) and very ~.

W096/04557 PCT~S9~10182 219~679 small antibody epitopes which are known to those skilled in the art.
Incubation conditions can be varied extensively.
Variation~ in temperature, time, pH, buffer and media additives are all to be considered as those attributes which influence target:rVab complex formation and stability in manners known to those:skilled in the art.
The preferred conditions here are phosphate, MOPS, Hepes or Tris buffer at about neutral pH (6.8-7.2) with 1% BSA
at room temp. for about 4-6, up to about 6-12 hrs.
After formation of target:rVab complexes, any matrix bound rVab is separated from unbound free rVab. In the preferred embodiment, the target is attached to plastic culture dish surfaces, and one of any number of rapid procedure~, such as panning, i8 used for separating free and target complexed rVab. The general approach of pan ing at different temperatures, pH and the presence of the antigen have been shown to allow isolation of rVab with controlled affinity.
After ~Pt~ t from matrix or affinity associated tag, by procedures such as low pH or others known to those skilled in the art, the recovered rVabT+ can be recycled through the selection procedure or any variant thereof any number of times. pllhllRhPd panning and affinity chromatographic procedures have shown single step enrichments of 5 x 102-103 per cycle. Although, the number of cycles can be varled, fl~pPn~;ng upon the enrlchment found per cycle, the abundance of a particular rVabT+, the total size and diversity of rVabT+ recovered, 3 cycles i8 preferred. Other number of cycle8 may be 30 chosen based on recovered rVabT+ characteristics such as S
or A.
Isolation of rVabT+ members can be done with different types of packaged rVab expres8ing functional rVab, including phage packages or soluble entities as discussed earlier. ~ = :

WO 96104Xi7 PCT/US95/10182 2lgli67g . ~1 .

-=57 -In this preferred isolation step, the rVab can have ~ one o~ the following functional forms Fv(rVCvh or vl only), Fab, or scFv as described below:
- a. Single functional VH or V~ without (Fv) or with associated constant regions (Fvc) for the V heavy (CHn) S and Vlight (C~ or k) genes or parts thereof. Both types of F can recognize targets'using only three of the 8ix V
region CDRs present in a natural Fab. In the preferred case, these Fvc genes are first packaged in fd phage and expressed with the C region (or some part thereof) attached, and in frame, being respectively a CHn, C~ kappa or CL lambda) in which in all cases the constant regions, are devoid o~ their C t~rm1 n~l cysteine. There are a number of CX regions available ;nr1n~;ng CH gamma, or delta, selected based on the required solubility and known to those skilled in the art. These V(or VC) genes can be expressed as soluble entities with or without tags or, as i~ the preferred case, ~used, in frame, to one of the phage's coat proteins ~i.e., gpIII) for functional display. These libraries comprising only V regions are termed rFv and may be expressed packaged in phage for phage display. rFv phage libraries may be screened for members possessing CAP attributes of T, S and A and may be further diversified as described above. Such libraries with ~nt1t1~r cnnt~;n1ng a reduced number of CDR or CSRs may be derived as part of the secondary simplification process when there are a very large number o~ active rVabs or when simplification is desired to foster the development of a more accurate BEEP.

c. Fnnrtirn~1 VH ~n~ V~ rl h;n~ti~nr (rV~h) These combinations have two V genes with, or without, partial or intact constant genes. Although they may contain like members, the pre~errea combination is one VH
(or VHCX) and one V~, (or VBCB). In the preferred method, rVabT+ with the particular V~ and VB couple are co-W096/04557 2 ~ 9 ~ 6 7 9 pcTNs9sr~o~82 ~' ~ 8 -o packaged in a single phage, on a single piece of DNA, as two individual gene products. For each phage, either VX
and VL, may be expressed as-soluble protein~and the other attached to gpIII to cause surface phage display of the Fab. This coupling and expression of VX and VL can be made with or without identiiying separately the VH phage libra~y and VL phage library which can recognize the target when in the presence of a library of;~oluble VLCL
protein or VXCX1 protein respectively (see, supra).
In one embodiment, the sequential procedure to obtain l~ ~unctional rUabs is as follows: Three individual libraries are made. Two of ~107 phage packages each expressing and cr~nt~;nlng only one V gene (VXCH or VLCL) attached to phage for surface display. The other i8 of the same size but is made of VL genes expressed in lambda as soluble VL
proteins which can be harvested from periplasm of bacteria expressing the VL soluble library. First rVabs are then made by mixing the soluble protein library with the VX
phage library in solution prior to testing for target recognition. This mixin~ allows all VL proteins present to complex with any one VX expressed on a single phage surface package to form a phage attached noncovalent (disulfide b ~ds excluded) functional rVab. This allows the formation of all possible rVab combinations. To this mixture is then added the matrix associated target under study and after lnrnh~t;on~ and complex formation, all phage carrying a matrix associated target displayed as part of rVab displayed on the surface of phage are isolated, preferably by one of the above noted panning procedures. Subsequent isolation of phage DNA gives an expre8sible library of functional rVXCXl phage which can be T+ (i e., the T+rV~h~l fT,TT~) .
All the VH phage library inserts, before or after phage amplification as needed, are next excised via simple endonuclease restriction digestion, and directionally cloned into a lambda able to express inserts as soluble W09~104557 21 ~ 66 79 ~US95/~0182 ~ . .,;

periplasm proteins. After~induction as noted above, the - protein from the T+rvTTrT~h~lfT~TR is harvested from the periplasmic ~ace to give a protein library with the - potential to form complexes with the entities within the other original phage library, i.e., the rV~CL library.
A~ter mixing the soluble protein library and this phage library, as above, the T+rVLCLhalfLI3 is isolated, as noted above.
In the final step of this embodiment, a combinatorial library of packaged pairs of T+rVab is produced in which individual packages contain one VH and one VL pair of genes co-expressed as separate entities but associated together in functional rVab complexes. In the preferred embodiment of this procedure, these two genes are combined via the C~E-LOX recombinase system reported originally by lS Xoess [Hoess, et al. 1982, Hoess and Abremski, 1985;
Hoess, et al. 1986] and recently by Griffiths et al. 1994, which are included herein by reference. In another : ~ -'im~nT,, the package ig also a phage, and expression is similar to the preferred embodiment but in this procedure, the , 1n~t;cns of rVHCH and rVLCL are made by ~r;r;ng and ligating in vitro the DNA in a fashion which allows r~n~l 7~t'rn of VH and VL pairs but only one pair per DNA
construct. These constructs can be phagmid or phage to allow either bacterial or phage expression of the rVab.
In bacteria the rVab are isolated and tested by protein lifts, whereas in phage, the rVab is attached to a surface protein for display and assay. Both methods have been published [Hoogenboom, et al. l991; Xang, et al. 1991;
Waterhouse, et al. 1993; Figini, et al. 1994; Jespers, et al. 1994] and are commercially available in kit form (e.g.
Stratacyte, CA). The preferred method is phage display of rVab.
The advantage of the 'o~;m~nt in which active rV~h~lfT,TT~ are identified before rnmh1nini them into rVabs, is that where combinations of VH and VL are made . ~ ~ . . , W096/04ss7 ~ PCT~S95/10182 .
2196~

randomly from a preselected T+ active rVhalf library, the independent preselection of~active VHCH T+ and VLT+ genes i8 likely to have reduced the number of active rVh~lfT.
members to less than lO~-~. This reduction in number greatly increases the chances of deriving within a single phage library of lO12 members, which is attainable with the methodology disclosed herein, all possible active rVabs.
The procedure used to isolate single VH and single VL
and pairs of VH~VL which recognize the target has the added benefit of being rapid, and controllable as to the strength and nature of Vab target binding that is desired.
3y the procedures outlined,~a paired rVabT+ (~nnt~in;ng 5 about 101~ entities) can be generated The procedures discussed above result in the .
isolation of a) rVH or rVL, which alone do not need the other to recognize the target, and b) the recombinantly derived co-mb;n~t;nn~ of rVH and rVL termed rVabs and scFv which, in the later case, have rVH and rVL linked together by a short peptide chain and expressed as gpIII phage protein fusion products or even as soluble entities.
Additionally, rVab in which both V domains are of one type, i.e., either VH2 or VL2 are possible by this invention~ VHVH Fab have been reported with increased solubility. Altering CXl for CH delta regions or changing speci~fic and identifiable C amino acids, could also f~r~l;t5te expression of novel rVabs.
The basic and preferred t~rhnnlngy for cloning individual heavy and light chain variable regions either alone, or attached at their N terminus to leader sequences, or parts thereof, or at their C terminus attached to a constant region, or parts thereo', and pl~r~m~nt into suitable expression vectors, transformation and P~pr~;nn in a compatible host cell in active form by re~nmhin~nt DNA technology are described in the art_ See, Huse W092/06204; Ladner WO90/02809; Winter WO92/20791, which are incorporated herein by reference.

W096/04ss7 21~g~67g PCT~S95/10182 To achieve high yield and faithful cloning of each ~ active IgG, secretion of protein either as soluble extracellular protein or in the periplasmic space is - suitable. In addition, protein may be expressed as an extrar~ r (or on the surface of phage) facing tL~ ne or me-m-brane-anchored functional protein which allows sprnt~n~o7lr dimerization of heavy and light chain intact IgG or V domains.
Methods of cloning from naive or ~mnn;7~ animals, entire spleen repertoires of Vab heavy (Vabh) and Vab light (V~l) in their natural or random pairings to derive enormously diverse crlmhin~trnial repertoire libraries are known in the art. [Huse, et al. 1989; Sastry, et al.
1989; Milstein 1990; ~lArkr~n, et al. 1991; Marks et al.
1991; Winter and Milstein 1991; ~awkins and Winter 1992;
~oogenboom, et al. 19~2; Lerner, et al. 1992; Marks et al.
1992; Winter, et al. 1994].

B Identification of rVab's Which Bind ~n~ Activate Tarqets (Staqe lh) In a preferred e-mbodiment of the invention, pairs of V~ and V~ antibody domains (rVab) are selected both as biological scanners of specific target surfaces and information reporters of activity related to the molecular 3D structure o~ the antibody site involved in surface interactions as well as the molecular 3D structure of the active P1Pm~ntr of the binding site. This structural i~iormation is relevant to identifying the minimum structure of the ~IGATT, which would need to be inuuL~oL~Led into a SOMER or DISOMER, to reconstitute the CAP of t~e~acFive rVab and regul~ate the target in the desired fashion. This invention identifies the unique ability of rVab when used as libraries crnt~in~ng at least about 101~ members to identify those portions of a target~s surface connected to function in such a manner as to ~ tely provide the tools necessary and sufficient for W096104s57 ~I~6~ PCT~S95/10182 ;-:
~, - 62 -screening ~or organ ~ replcc~m~ntc at the target with a desired CAP. In addition, an embodiment of the process uses genetic algorithms to construct 3D high resolution molecular models of the shapes of organic molecules which can fit into the active target and regulate activity so as to electronically screen for~or synthesize via computer ~L~ r ~ SOMERS or DiSOMERS.
Active target landgcapeg are thoce gurfaces ~nn~t~
to target function as defined as those able, when occupied by a ligand, of influencing target activity. It is known that antibodies, in a wide variety of forms, e.g. Ig, Fab2, Fab, or sFv (i.e., VH or V~ alone~, have exceptional selectivity as well as high affinity for their targetC.
This invention uses rVab which are identified as possessing the desired CAP attributes in two ways.
Structural characteristics of multiple rVab's iden~ified as possessing the desired CAP attributes are rnmh; n~ to produce a composite structural map which is used to define a ~3EEP. In addition, individual rVab's which are ;~nt;f;~ ag poggegging the degired CAP may be labelled 80 that they may be used as reports in competitive binding assays to identify SOMERS, DISOMERS or other ligands active at the pharmacological target.

l. Identification of rVab's with TSA+
For T~rqets ~Av;ng Endoqen3ug ~ie~n~c The approaches to isolation and ;~nt;fication of Vab for targets having endogenous ligands and rVab processing all TSA+ attributes, are divided based on two flln~
issues: first whether the rVab induced target modification is allosteric (alloA) or competitive (compA) with the native signal (endogenous ligand) and second, whether the active surface is a simple or complex landscape found one or more:different submits of the target. Target modification is considered anything which alters target activation by any means ;n~ ;ng native W096l045s~ 2~ 9 66 7~ pcr~sss/lol82 signal recognltion (i.e., signal binding) and/or the - signal tn~nsfl~ tion process directed by the active target.
For example, the binding of ACh to~ the muscarinic subtype - 1 receptor ana the interaction and activation of the Gi protein, respectively. In both cases, the process uses libraries already selected for, preferably by batch mode selection, target recognition i.e., rVabT+, Batch mode selection is preferably than used to identify and separate rVabT+A+ from those which are inactive under specified ~nn~;t;nn~. T,;hrAr;P~ of 10~ to 1ol~ individuals are used and the process is therefore applicable to rVab libraries which have both VH and VB chains, noncovalently (as Fab) or covalently attached (as scFv [Hoston, et al. 1988;
Bird, et al. 1988; McCafferty, et al. 1990; Hoggenboom, et al. 1991; Barbas, et al. 1991; Garrard, et al. 1991;
Breitling, et al. 1991] or fl;~ho~;P~ [Holliger, Prespero, and Winter, 1993] as well as those with only one V chain.
By methods known to those skilled in the art, individual rVabTSA+ within an active rVab A+ library (BIB) can be simply and rapidly isolated, assayed, tagged and used to screen various chemical libraries for SOMERS which compete with rVabA+ for binding to the target.
For allosteric Vab-modulators, the presence of allosteric activity within a rVabT+ library is indicated by the occurrence o_ an alteration in the association between rVabT+ and the target induced by the binding to the target of another entity. This entity could be the native signal or any known target ~ffectQr entity.
Examples of allosteric entities include such nucleotides ~ as ATP for receptor ~nnt~;n;ng ki~ases, or GTP for G-protein associated targets, or a protein which couples - to the target during signal tr~n~nct; nn such as G-proteins, or even other receptor subunits.

W096l04557 6 6 ? ? ~ PCT~S95110182 ~ ~-64 -a. Identification of rVabTSA+ from rV~hT+ n~in~ sllosteric ~ifiers The isolation of rVabTA+ from rVabT+ is tied directly to the action of the signal at the target. In the preferred process, matrix-linked target (m-Tr)is mixed with t~e rVabT+ and incubatad so as to allow m-Tr:rVabT+
complexes to form. In general these are the same conditions used to isolate rVabT+ in Step I (b). After sufficient time to allow appreciable complex formation, which=may or may not be sufficient to allow the 0 interaction to come to e~uilibrium, the temperature is lowered to about 4~C so as to trap bound rVab in the m-Tr:rVab complex by slowing its dissociation rate. With the temperature at 4~C, free rVab is rapidly washed away and the complex is resuspended in original buffer. This proeess is done ~uickly and uses a matrix such as, for example beads or plastic surfaces, and takes <1 min. For this process, preferentially one first determines or estimates the normal dissociation rate of rVabT+ from the target. This may be determined by methods known to those skilled in the art. For example, in~parallel reactions, the dissociation constant (k-l) for target (Tr) and signal are determined using either a labeled target (T*) and monitoring the dissociation of T*-rVabT+-matrix complexes, or unlabeled target and following its release from the rVabT+-matrix complexes using anti-rVab constant reyion ~ntiho~ (or anti-phage ~nt;h~;es) or by simply assaying phage in the supernatant if a rVab phage library is used. The half time (t~) for k-~ at 4~C for rVabT+
library from the target, for the entire population, is then determined.
With the t~ for k-, known, a new pop~ t;on of washed rVabT+-matrix complexes of the entire rVab library are formed at 4~C and allosteric effectors are added in saturating concentrations. Xalf the population is centrifuged to isolate the free rVabT+ members from the W096/04557 PCT~S95/l0l82 ~ 21~667~

- 65 - _ _ library which remain in the supernatant within about the ~ first minute (or ~1/30th) of the population's dissociation t,~. The remaining half is allowed to dissociate for about ~ 10 x t~, centrifuged and the pelle~ resuspended and allowed to dissociate for about another 10 x t~ to isolate the second population of free rVabT+. In both cases, centrifugation is used to rapidly isolate the free rVaoTf.
In the first instance the free rVabT+ library is enriched for those rVab members induced to rapidly dissociate, referr~ tn ~r rVabT+A+ allofast, while the second is enriched for those which have been induced to dissociate slowly, referred to as rVabT+A+ alloslow. Each is thoroughly washed and then recycled through the above isolation procedure a second time. Such enrichment cycles are rr,nt;nn~fl until a clear change in entire populations t~ for dissociation is seen at which time the population is termed rVabT+A+ (fast or slow). Their numbers are then flotrrm;nPfl/ if neèd be after amplification. If these populations are small, individual rVabT+A+ (fast or slow) can be isolated at this time and assayed directly in ~ 20 subsequent procedures. If large populations are obtained, they can be analy~ed in 9ubsequent steps to isolate subpopulation9 which have other desirable target attributes, e.g. speci_icity (S+) among one of a large number of target family members.
b. Td~nt; f;r~t;on of rVabT+A+ from rVsh r;hrary Usinq Competition Assays The second approach to isolating rVab capable of ~ target modification is used for the isolation of rVabT+, whether or~not the S properties have~et been determined, which are target regulators which bind to targets at the same domain or at a domain overlapping with that used by the target's natural signal (nS) endogenous ligand. These are considered as competitors with nS for binding to the nS binding domain, and therefore are co~petitive ~ . , , I

W096/045~7 PCT~S9~/10182 ''';-'L'~ _ : - ~
219667~ - 66 -modulators, not allosteri-c modurators. Both agonists and antagonist replacements for endogenous ligand will be found within this population.
This proce8s requires the use of a high affinity nS
which is labelled (nS*) and capable of rapid and ~uantitative isolation. There are many such labels possible, one is biotin, another, for example, is the small antibody epitopes for which high affinity sera (or monoclonal antibodies) exists commercially. Methods of making such a labelled nS and the available epitope/antibody combination for protein signals and organic molecules are known to those skilled in the art.
Labelling is a relatively easy procedure for protein nS.
~or organic molecules it i9 much more difficult but in the preferred cases where labelling has not yet been done, non-neutralizing monoclonal antibodies or biotin will be used by methods known to those skilled in the art.
The preferred process of identification and isolation o~ competitive rVabT+ (S det~rmln~ or unde~rmin~d) which is outlined here u~es biotin as the nS label ('~tag"). The process works similarly using other labelling tags such as iodination with I~I, or [32p] ATP phosphorylation.
The biotinylated high af~inity signal, nS~, and the rVabT+ library to be tested (previously isolated and identified as T+) are combined with a soluble active form of the target ~Tr) and incubated so as to allow fnrr-tinn of significant numbers of nS~:Tr as well as rVab:Tr complexes. The incubation conditions used here are those previously used to allow binding of the rVab library to m-Tr as long as these conditions also allow nS~ binding to Tr. The temperature is then lowered to 4~C and all nS~ and nS~:Tr complexes are removed from solution with strepavidin (or another tag recognizer coupled to some matrix). The supernatant, containing T:rVabT+ complexes and free rVabT+ is~affinity separated to isolate only Tr:rVabT+ by either panning over anti-Tr antibody coated W0961045s7 T~~ J~IOI6~
21g6~79 7, ~

dishes or passed through anti-Tr ~ntlho~;P~ coupled to ~ agarose. The anti-Tr ~nt;hn~ies used in this step do not alter rVabT+ b; n~; ng to Tr Such antibodies are known to o~ten be those which have epitopes at either the amino or carboxy termini of the Tr under study or some other noi-modulatory (i.e., non-active) target domain. The population of rVabT+ bound to Tr in solution and obtained by association with anti-Tr antibody on their own matrix can be isolated and recycled~through the above procedure any number o~ times ~or enrichment and ampli~ication This population contains all rVabT+ library members which bind to Tr at the binding site used by the target's nS
This population is there~ore made up o~ rVab which bind to the nS binding site and will be labeled rVabT+'~~P. Even though at thig point these active rVabs are uncharacterized as to agonist or antagonist activity, their classification as active rVab is appropriate based on the ~pf;n;t;nnc and disclosure=of this invention.
Indivi~ual Pnt;t;pn within these poplllstinnq may be isolated, testea for agonist or antagonist activity using standard ~ vitro, cellular or i YLY~ assays known to those skilled in the art, and/or labeled by procedures known to those skilled in the art and used ~or screening for agonist and or antagonist SOMERS. Furt~P -~L~, where a l~hP1lP~ nS~ exists for Tr, individual rVabT+A+compt will be tested for competitive ~;f;n~t;on of nS~
binding to T by methods known to those skilled in the art c. Isolation of rVabT+ Which Are A+ By ~llostericallv Mo~ifyinq Tarqets The next process outlines the isolation of rVabT+
which allosterically modify Tr (i.e., are A~) by binding to sites which do not alter nS binding but do alter the ability of the target to be active even ~or targets devoid of native signals. In these cases, active rVab will be isolated by virtue of their ability to alter the W0~6l04sS7 PCT~S95/10182 21~6~79 association of T and some component of the signal transduction system used by the target. For G coupled receptors, that would be the GTP-G protein complex; for targets with catalytic or stoichiometric enzymatic activity that would be nonhydrolyzable substrate analogs;
and for channels or:transporters it would be ions, molecules transported, electrochemical gradients or other channel subunits. In these cases the i801ation of~this type of rVabT+A+ would occur either by a) testing in batch mode limited sized libraries i e., rVabT+A+ for agoni8t or antagonist action in vitro; or b) isolating in batch mode those which altered Tr activation, i.e.
phosphorylation, binding of ATP or GTP, or binding of other proteins involved in signal transduction as outlined above. hibrary members which are T+A+ may be diluted and retested until single entities are i~ntif;~d.

d. Identification of rVabT+A+ Pairs ~h~n Sinole ~v~hT+~+ ~re Not Identi~l If no single allosteric or competitive rVab is found in cases where an nS exists by one o_ the above approaches, the following procedures are capable of identifying pairs of entities which, are both required simultaneous as the n~c~s~ny condition for modification of the target. In these prQcedures, the pairs of entities tested will be provided by two differentially identifiable rVab libraries or preferably one rVab library and another large and highly diverse library of ;~nt;c;~hle molecules. For targets with large protein 8ignals, such as growth factors cytokines, etc (i.e., ~lO,OOOD~ which may be expected to have more than one hIGATT this dual modifier assay will be the preferred approach in one of two general alternative forms.
The basic procedure will be described first using two differentially labelled rVab library as 80urce8 of the two paired modulatory entities. In addition to t e rVab G7~
WO 9610455~ PCI/US9!i/~0182 ~ i, l;hr~r;Pc there are both a labelled Tr (Tr*) and a - labelled high affinity signal (haS) which are also recoynizable ;n~PpPn~Pntly and separably from each other as well as from the rVab by high af f inity probes. In each case, recognition of target occurs whether or not these entities are part of any type of Tr complex but does not perturb the target's ability to bind baS* or rVab. For example, the labeiling epitope c~ntclnp~ within the Tr*
could be one which is recognized by a high affinity Ig at sites commonly known to those skilled in the art as non-neutraiizing epitopes ~arge protein targets are known to Pnc~mpAcg such sites within internal peptide seguences, N- or C-terminus or unmodified or modified amino acids. These epitopes need only be exposed during complex formation and non-active, i.e. unable to modulate target binding of nS when occupied by recognition antibody which can be easily est~hl; ChP~ in eacb case.
For signal labels, either biotin or an integral Ig epitope, are the preferred label, allowing avidin- or Ig-agarose respectively, to be the guantitative recovery probe as long as the labels do not significantly reduce affinity for the target. Other possible labels include i~ont;f;~hle peptides or protein sequences, such as substance P, partial XSV viral coat protein sequences, and ~nkPph~l;n ~he ~nt;ho~;Pc for such small epitopes or peptides could be either polyclonal or monoclonal Ig, commercially available or rVab as procured by the recombinant methods referred to for targets disclosed herein. ~iotinylation of various signals and testing for non-interference with native target signal binding to Tr is available by many methods known to those skilled in the art.
~ sing an Ig epitope l~hPllP~ or tagged Tr-(Tr*) and a biotin-lch~llP~ high affinity signal (haS), the identification and isolation of a pair of modulatory entities (ir this example both are rVab) is initiated by W096/04557 PCT~S951101~2 .

219~6~!~

combining sufficient numbers of two previously isolated t large rVabT+ populations, each with a specific Ig epitope (epitope 1 for rVabl and epitope 2 for rVab2~), with the haSbi'~' and the epitope tagged Target (Tr*) to allow formation of the trimeric rVabl:T:rVab2 complex which does not bind haSb'~'i=.
rVabl and rVab2 may be added initially at a variety of about equal concentrations from 10 x 1I down to 101 M.
The lowest rnnn~nt~ation at which target activation occurs will be used for subsequent manipulations. ~The upper 0 number is arbitrary but should theoretically exceed by about 30 fold ~he concentration needed for rYabl or rVab2 to bind to Tr 90 as to saturate the site and prevent binding of haS*. The mixture is then allowed to incubate at room temp for at least approximately 6 hr, or overnight and then saturating amounts of avidin-agarose is added and the mixture centrifuged and the supernatant, devoid of any free haSbi~' or Tr:haSbiot complexes, is removed for subsequent use. The supernatant, ~nnt~;n~ng dimers of Tr:VAB1 and Tr-VAB2 and th~ desired trimers of VAbl:Tr:V,b2 are then panned over anti-T Ig attached to a solid matrix or support such as for example, plastic culture dishes or agarose column matrixes.
T~nt1f;cation and isolation of Tr complexes having both rVabl and rVab2 concurrently bound can be made by panning successively over matrixe8 coated with anti-rVabl and then anti-rVab2 Igs. Phage displayed rVabs isolated by this procedure can be separated, amplified and then used for secondary cycling through the above isolation procedure. Finally, individually purified phage are tested in identified combinations for competition of haStag binding.
In the above case, the~two rVabT libraries (i.e., rVabT1, 2+) can be easily distinguished for example by nt;l;7;ng the CHl domain of humans on one and the C~l domain of mice on the other_ Ig specific for human and WO 9610455~ ~ 2 t 9 6 6 7,~ PCT/US95~10182 ' ~ .
- 71 ~
mouse C~l are available commercially. Use of other constant regions from one specie i8 also possible.

e. Use Of rVab-Peptide ~ibraries And Other Probes To Identify Multiple T,Tr~T Tarqet8 i. Identification Of First ~igand For A Multiple ~IGATT Tarqet There are a number of variant8 to the above procedures in which the second entity of the pair needed to compete for haS~ binding would not be another rVab but instead would be a member of another library cnnt~in;n~
~ diverse small organic molecules, peptides, nucleic acids, carbohydrates or even natural products. Excluding the possibility of stearic htn~Pr~nre, the frequency in the rVab library of entities which bind to a target in a modifying manner (given their paired entity is also present) should be no different than that for rVab which are able on their own to bind to Tr surfaces and modify signal binding. Accordingly, rVab libraries of the size generated by this invention may be used to identify both rVab members of the sought after pair. All of the libraries stated above having in excess of 101l members~ml should be suitable for use with this invention provided the frequency for each binding event is not less than 10-3. A useful library or pair of libraries should contain sufficient members so that two binding events will occur simultaneously on the same Tr, the condition necessary for inhibition of haS~ binding, at less than about 10-ll and therefore be present at least once per rP~rti nn, If the frequency of each event is greater, i.e., 10' or 10-3 then these modulatory complexes will nrrnr ~ frequently as 10 to 100 times per assay. Ac the purification of an active phage displayed rVab per cycle is 10-3 Eo 10-3 then up to 4 cycles may be needed to purify the active entity. To obtain one member of the pair, one only has to purify from the final step, one of the two rVab entities. When other W096/04S57 ~ PCT~S9~/10182 21~67~ - 72 -libraries are ln use as the source of the second pair member, they need not be isolated at all.
ii Identification Of Second Or Subsequent ~igands For Secondary ~IGATTS Of A
Ml I 1 tiple T~T~TT T~get Once one member (primary member) of the pair is 1fl~nti~;rfl, which in the above cage would be a rVab the isolation of the second 18 made straightforward by using the first member, at saturating r~nr~ntration in all reactions. This simplifie~ to ~ search for a single entity, which for a rVab, would be done as outlined above.
However, when one rVab of a pair is in hand, one can search through a chemical as well as a rVab library for the second member of the pair of Tr binders which regulate Tr activity when simultaneously bound to the target. Each member of the pair, particularly those which are identified as members of a chemical library, are potential r~nfl;~t~c ag one half of a pair of smaIl organic molecules, one for each active surface domain required for target regulation, which when covalently linked together would provide a single active organic molecule referred to as a DISOMER. Such DISOMERs would be valid interesting drug discovery leads.
Another protocol for identifying an active pair, i e., a pair which is necessary and sufficient to bind to Tr in such a manner a8 to displace haS~g, is to perform the orlginal inr~lh~t;on of tagged target (Tr*), high affinity target siynal (haS) and target binding rVab (rVabTrl or Tr2+) in the presence of excess labelled Tr~
to reduce to a minimum the pre8ence of unbound rVabTrl or Tr2+ If these incubations are done ln the presence of haS at about a 100 fold exce8s of the Tr-saturating dose, the only rVab in solution will be tho~e which has been competed from binding by haS Accordingly, those rVab - prevented from binding to Tr by haS, ~hould, with high probability, be those which can preve~t haS binding to Tr WO 96/04557 2~ 9 6 6 7g PCI~/US95/10182 S~,r, ~ ' f .

and are expected to possess the deslred activity. As bound rVab can be separated from free rVab via panning ~ver_anti-Tr Ig (or avidin with a biotinylated Tr), upon such removal of all rVab:Tr* complexeE, the only rVab Ll ;n;ng in solution will be those pairs which when bound S together, and possibly individually, prevent haS binding.
Recycling of the supernatant A~it;nnAl times through such a paradigm will eventually result in identifying the rVab pair or at least one of its members if another type of ligand is used as the source of the other half of the active pair.

iii. Use Of rVab-Peptiae As Surface SrAnn~rb For signals such as protein hrrmnn~ and growth factors, where dimerization or timerization of identical (i.e., homoligomeric) or different (i.e., heteroligomeric) receptor units is required for receptor activation. This invention solves the problem in one embodiment by creating bivalent rVabs which allow for the isolation of bivalent active rVab surface reporters capable of identifying each receptor subunit endogenous ligand TARGATT attAr1 site. In this process, identification of bifunctional active surface reporters, proceeds by taking a plurality of rVabs which have previously been ;~nt;fied as z5 recogn; 7; ng either a particular limited surface of one of the target's subunits (i.e. are T+) L or a larger number of one or two selected groups of amino acids which are known to be involved with endogenous ligand binding. The genes ~nrn~;ng these rVabT+ ligands are modified to encode for a flexible amino acid which attaehes in frame to one end of either the heavy or light chain construct, a library of small random peptides to create a bifunctional scanner (rVabPEP). In one embodiment, the peptide is encoded by DNA used to that encoding the heavy or light constant domains. In another embodiment an rVab is expressed with W096/04557 ~,~ PCT~S95/10182 ~t~
21g~6~
~ - 74 -o at least two peptides for ldentification of~trimeric receptors.
In a preferred embodiment, a bifunctional scanner library consisting of rVLCL~and one rVXCHl iB constructed to identify rVab-PEPs which recognize an active surface S consisting of two TARGATTS on the surface of the target.
rVab-PEP are then isolated in batch mode and individual member are subsequently identified as active competitors for endogenous ligand binding. Such rVab-PEPs do not significantly bind the target in the presence of excess endogenous ligand. These bivalent rVab-PEPs will then prebound to target will prevent binding of the target endogenous ligand which has been immobilized on a solid matrix_ - ~
For h~m~ ric receptors where each target subunit has a TARGATT which binds to the ligand (as per Growth Xormone Receptor, GXR), rVab-PEP would be isolated. The rVab portion of a first active rVab-PEP is then labelled for use as a reporter to identify SOMER replacements for the ~IGATT which resides within the rVab portion of the active rVab-PEP entity and recognizes one TM GATT on the surface of the receptor. To identify a second SOMER
r~p~ t for the second ~IGATT of the rVab-PEP entity, which resides in the PEP portion of the rVab-PEP entity, a second rVab without peptide is identified from the library of active rVab-PEP which competes for binding with the peptide portion of the first rVab-PEP. The process of finding the two rVab which correspond to the two LIGATT
residing within an active rVab-PEP entity is referred to as rVab Pairing. The second rVab is then labelled for conversion to a reporter for i-dentification of SOMERS for the second LIGATT site.
Where the targets are~heterodimers, the preferred approach is as follows. The rVabT+ for receptor subunit surface I, are grouped based upon recognition of common domains and/or surfaces cnnt~;n;ng amino acid known to ~ 21g667~ ,, affect bindir,g of endogenous ligand. These rVab's are then expressed as rVab-PEP as described above to generate a series of bivalent ligands. Members of this rVab-PEP
library which are displaced from target by endogenous ligand and which also displace endogenous ligand from the target are selected as above for ~ r receptors. A
limited number (< about 10) of rVab-PEPs with endogenous ligand dlsplacing activity,at the target are then selected for identifying a ligand for,the,s=econd ~II) binding site An alternative selection method _or identifying site I
ligands is to select rVab-PEPs based on their ability to activate target. Activation may be detected as described above based on modification of an allosteric effector or on some other detectable change associated with receptor activation. Por example, activation may be associated with self phosphorylation or dimerization rVabs for the second TARGATT site on the second receptor subunit of the heterodimeric are identlfled ln one embodlment, by expressing rVabs as a rVab-PEP library using rVabs previously identlfied as being competit,ive for the endogenous ligand at slte II The resultlng,rVab-PEP
llbrary for slte II i9 then tested for actlvlty as descrlbed above and actlve members are lsolated.
V. T~nti fication Of rVab Whlch Are Selective (S+) In order to lsolate those rVab whlch are selective for and dlstingulsh among closely related members of a target family or any target of co~cern (i.e. selective), the following batch mode selectlo~ procedure may be used.
The rVabT+ under lnvestigatlon ls mixed wlth matrlx lmmobilized target (m-T) and allowed to form complexes ln the presences of soluble peptldes, recomblnantly obtained protein fragments or intact targets whose ldentlcal (or related) sequences or conformations are found in targets for which the investigator does not wish the rVab to bind.
These sequences are typically between about 6 to 12 amino W096/04557 PCT~595/10182 ' 21966~

acids in length and~are present ir. the targets for~other endogenous ligands of the same gene family. After=
sufficient time for complex formation the rVabT+ still bound to matrix are isolated by panning and preferably recycled 2-3 times for enrichment as noted above to derive S rVabT+S+. This procedure can be done before or after any of the above procedures related to isolating Active(A+ ? or Target recognition positive(T+) library members. =~
If all screens for ~, S, and A are acco-mplished, the final library would be rVabT+A+S+ given that there_was only one LIGATT and one TARGATT required for~ regulation of the target and thereby represent individual~entities which describe target sites suitable for screening for SOMERS
with all three attributes of a CAP. Where there are more than one BIGATT and one TARGATT required for target regulation, i.e., when the target is multimeric or even monomeric but r~nt~;nr multiple TARGATT domains, the full CAP, includlng activity (A+), can only be observed with a bivalent rVab, such as would be found in an active rVab-PEP. In such cases, the rVab portion of the active bivalent rVab would not be active on its own.
Nevertheless, since it still can identify SOMERS we referto it as A~.
Clearly, high affinity~(less than or e~ual to about 30 nM) and selective target recognition do not require the antigen pocket of the Vab be made up of two V domains as found in native Ig molecules but can exlst in single V~
domains r~nt~;n;ng only 3 CDRs. Based on the information in the art, i ~ ~v~~ 8 in making useful single chain (rVvx; i.e., vh or vl) with T+, S+ and A+ properties are expected by utilizing constant domains other than CHl, i.e., using gamma 2 or 3 or delta. ~This~invention al80 rerogn; 7Pq the need for solubility of the recombinant proteins used to construct the members of the rVab, rVvx and rVab-PEP libraries. To be acceptable, changes in W096/04s57 2 1 9 6 6 7 9 PCT~S95/10182 solubility would not adversely effect V~; VL rtructure in ~ ar rVab When using single chain libraries, select the rVvx entities which modify pharmacological Earget activity via binding to its surface. Refer to these as the active rVvxT+A+ libraries (LI3). Isolate acEives based on:
i. those whose binding is modified by the presence of the end'ogenous ligand;
ii. those whose binding is modified by any allosteric regulator of the target iii. those whose binding alters target ~i.e.
target phosphorylation or association with G
proteins).
In the case of i and ii, actives are isolated as soluble entities and in iii precipitated by anti POI-protein or G-protein antibodies. In i endogenous ligand is used ad 300x Kd. In all cases harvest positives, amplify, and reisolate.
Group as to common surface domain recogn;7~fl by rescreening active rVvxT+A+. LI~ against target in presence of small peptides (10-12 amino acids) or large peptides made rer~ ';nAntly ~20-50 amino acids) which define the target domain. In this assay, tho8e soluble in presence of peptide are grouped together, and all data are used to construct an antibody surface map.
The members of the rVab library which are particularly useful in automated binding assays and screens ~or SOMERS at pr~fl~ntified target sites possess preferably the following characteristlcs.
a. <30nM affinity for target;
b. recognized target sites are smaller than ~. those used by endogenous ligand signals;
c. possesses agonist or antagonist activity when bound to an active landscape whether . it be those used by endogenous ligand or allosteric sites;

. . - .

W096/04557 ~ ; :. PCT~S95110182 219~6~'t', d. specificity for binding to oLly one~among many related members of a target family;
e. little nonspecific binding to unrelated targets and ~ubstances related to the assay it~elf;
f. easy and homogeneous and single tagging with a label ;
g. labelling which allows both rapid and sensitive ~ua~titation of target binding and;
h. a fL .iJlh of known structure which delineates the location in space of the contact points of the reporter with its target.
The latter attribute is critical to the solution of the 3D structure of active SOMERS as it allows the problem of deducing the 3D-shape of the LIGATT on the target surface scanners which are active and in contact with the target to be solved after obtaining the one dimensional linear amino acid sequence of the reporter with the use of genetic algorithms. The 3P landscape of the LIGATT on the active rVab is directly transformable into a 3D landscape of the sought after SOMERS.

VI. Identification of biologically ~nh~nrPd ensemble ~ht~rm~ro~hores (PEEP) A Combine structural information from identified members of library possessing desired attributes of potency, activity, selectivity, and s~ecificity In trying to identify useful ~Vabs and to deduce the structure of the ~3EEP, the ability to genetically simplify (e.g., reduction in number or.size) or further diversify (e.g., increased number of randomized amino acid positions, or increased size) of CDRs and CSRs within active rVab libraries or within one rVab is of critical significance. This is because not all contact amino acids W096104557 21 9 6 6 7 ~ PCT~895/10182 }. ~ " ~-contribute the same energy to antibody binding and ~ sometimes one amino acid can account for >99~ of binding ene~yies. Just the 3 CDRs of one VH can provide lO-lOO nM
of Ig target affinity. rVab phage libraries of about lO12 members with secondary diversifications in any number of reyions can be derived from a small number o~ active rVabs found initially by processes of the invention previously described, by PCR as used to congtruct the rVab library (see below) and or oligonucleotide insertion, known to those bkilled in the art to provide an acceptably large enough sQurce of target surface scanners and reporters as envisioned by this embodiment of the invention. In addition, it is clear that active surface scanner rVab will be needed which recognize different local surfaces on the target in order to generate sufficiently large amounts lS of one dimensional amino acid ser~uence information so as to accurately deduce a BEEP which is not only accurate for predicting the structure of one SOMER but is capable of predicting the ensemble of active SOMERS which can attach to that site.
A particularly novel aspect of this invention is that it estAhl;qh~ a way for the CDR regions of a VH or a V~
alone or complexed together as rVab to be reduced to a minimum structure which occupies the target sites recoynized by the rVab and have a deslrable CAP. An advantage of identifying such a minimum structure is the potential reduction of taryet affinity to a level which is competable in standard binding assays by endogenous ligand and potential $0MERS and of the number of crltical atoms participating in target contact. The smaller the number of contact points the simpler the resolution of the BEEP.

B. Create .3eeps For Earh Ar~ive rVAh Subset According to this invention, BEEPS are created which contain the coordinates and attributes of the active elements of the 3D surface of active SOMERS for a .~ , W096/04557 PCT~S95/10182 219667~
~ . - 80 -particular surfacè domain on particular pharmacological target~. The starting point for this i9 grouping together of rVabT+S+A+ members of the rVab library according to common target surface domain recognized which in the first instance will be that which is overlapping, or ;fl~nt;r~l to endogenous ligand.
In a preferred embodiment:
a. Each surface group is partitioned and one rVabT+S+A+ for that group is isolated. ~The VHCH gene is then cloned out and used to derive a new combinatorial library. To derive this new combinatorial library the cloned rVHCXn is paired with all rVBCD for rVab members which bind to Ehe common surface.
b. Isolate via panning (as done for the original DIB) all new combinational rVab members (i.e., rVXCHn:
rV~C~' rVab) which are T+S+A+ for the original common target surface domain. This library is called rVabv~
Repeat for each VHCH in the original rVab thereby deriving a rVabv~+a+2n+ set which ;~nt;f;es all related VH and V~
for a particular surface domaln. These libraries will provide multiple combinations of defined VH genes with all V~s for a given surface. :Alternatively, these various libraries may be made by identifying specific VD genes and cloning them into libraries c~rt~;n;ng all VX genes identified ~or a given surface target.
c. Determine via PCR the amino acid sequence of all V~ in the set which can bind to all VHs in the library.
d. repeat a-c for all active V~using [VL]nn~n=~+
e. The spacial coorfl;n~t~ for the fL~.._..Jlh of the parert antibody in which all randomized CDRs were placed, along with the coordinates of the various CSR and CDRX3 for the active ~ and VL for tho9e entities found in the particular local target surface domain rVab library under study along with the amino acids ;fl~nt;+;~fl in these CSRs and CDRs are solved in a genetic algorithm to determine the 3D conformation of the p~rr ~logical~target -W096/04ss7 2I 9 6 6 79 PCT~S9~10182 landscape occupied by all active rVab members which - recognize the same surface domain~ This solution is a biological ~nh~nr~fl ensembled ph~rm~phore (i.e., a BEEP) f Repeat for rVab library for other local active target surface domains.
S g. If any data base is not sufficient, take the relative set of VH genes and excise their CDR~3 domain and replace with a random oligonucleotide ~n~ofl~ny a peptide library o~ preferably 8 to 10 amino acids. The potential size of this library is between about 82~-lOZ~ me~mbers.
Repeat selections to obtain new diversity ~nh~n~efl BIP.

C. U9e of qenetic alqorithmc to Create BEEPS
Creation of the BEEP begins after isolation of a set of active rVabs {Vi}i=N, which contain members lvi) which have been verified as having the desired attributes of affinity, selectivity and activity at the target, where N = the ~umber of such members within the set. In the preferred instance, each active rVab will have all three of the above attributes, but it is also possible that only two, or only one, of the attributes will be desired and therefore will be present. For this description, TSA+
will refer to the active rVab irrespective.of which attributes are present. Each TSA+ rVab member is then isolated and its amino acid sequence det~rmi n~fl using procedures known and available to those skilled in the art. For example, commercially supplied kits and an automated sequencer (ABI, ~SA).
According to this model, it is assumed that an active target surface binds different rVabs, through the same site of the target surface, and accordingly, at least a subset of those rVab are expected to possess similar surfaces Thus, finding a recurriny, i.e., common, surface motif twhich we refer to as the BEEP) in different rVabs indicates either: a) the common rVab surface plays a role in target:rVab interactions; and b) that this W096/04sS7 ' PCT~S95/10182 ..
21~67~ 82 -interaction could be duplicated by other molecules with similar surfaces. Therein,~it is a common surface which is responsible for the common phenotype of at least a subset of the L, members of~the originai set of TSA+
rVabs. There may be one or more common surfaces within S the original set of TSA+ rVabs. This duplication takes the form of the BEEP first, and subsequently small organic molecules.
Given such a c~ll prt ~ ~n of TSA+ rVabs and their amino acid sequences, a prPl;m;nAry set of_surface scanners {Li}j=N, where each L; is a model of an antibody molecule, i5 constructed a~rfl;ng to the invention using the canonical structural pr;n~lpAl R of Chothia (Chothia and Lesk 1987, Chothia 1989, and Chothia 1992) and the information on the crystalline form of the parental antibody used as f /JLk for construction of the rVab library as described by this invention, N is the number of such TSA+ rVab surface scanners which define the fnn~;~mPnt~l geometry which is the position of surface atoms within acceptable diRtances from each within a generally known structure. Shape descriptors rely on known CSR and CDRX3 shapes, and the amino acid sequence within these domains. Subsequently, chemistry characteristics, such as charge, hydrophobic interactions, exposed/buried surface area, hYdLO~11 bond formation etc., known to those skilled in the art will be considered.
In the preferred case, each TSA+ rVab contains one VX
and one VL chain, with 6 complementary detPrm;n;ng regions (CDR) wherein three (CDRVLl,2,3) are within VL and three (CDRX1,2,3) are within VX Furt~ ~, in the preferred ca8e, there are the 5, 1 and 6 different canonical structures consisting of a different kuown canonical loop structure possible for every CDRVL1,2 and 3 respectively, and 3, and 4 different canonical structures consisting of known canonical loop structures possible _or every CDRH1 and 2 according to the invention. The QR for X3r W096l04557 PCT~S95/10182 21 9~G79 ~ .t9 although not ri9nnn; r~l, in the parental library will have - one of three defined structures in its parental mode before the amino acids posltions within each are ; 7Pd. Furthermore, the prior knowledge of rVab fL ..JLh and rPl~t;rnrh;p of the 6 CDR domains within the S fL~h.~..JLk provides additional structural information for constructing an ~j and eventually a BEFP. In addition, as the number of known antibody,structures increases, new canonical structures become known and may be incorporated into the rVab libraries to allow isolation of TSA+ rVabs containing such structural loops.
Each Lj can be represented, for the purposes here, by the atomic coordinates of the constituent atoms of the rVab which i8 a member of TSA+ set. The surface (Si) of the prpl;mln~ry model ~i can be parsed by its CSRs and CDRs wherein ~
S~ ~ [(CSRl)i,(CSR2)1,(CSR3)l,(CSR4)i,(CSR~)l,(CDR6)~]
wherein 1 through 5 denote CSRV~1, 2, and 3 and CSR~1, 2, and 6 denotes CDRH3, respectively, and wherein with each (CSR)i, for Lj there is a particular serluence~
The surface (Sij) can be repositioned and reoriented in space by transforming the atomic coordinates of the Li ~cror~n~r to SU=Ga'Lj, where Lj, is a model of surface scanner i defined by the coordinates of its constituent atoms and Gu is a matrix that transforms Lj. Furth~ ~, Gjj is paramaterized by the translation and rotational parameters (~i, Xi, ~i, xj, Yi, zi)j. Thus, as scanner i is rotated and movéd into a new position j, and the CDR are carried along with it.
The genetic algorithm of this invention, referred to here as DIOGAM, takes the initial set of {~i~}, where the superscript (~) means 'prPl;m;ni~ry model', as input data to produce from that data as output the theoretical common surface (i.e., the ;3EEP) which represents the best overlap W O 96/04557 ~7 ~ P~rAUS95/10182 21~6~9 - 84 - ~
in terms of chemistry and geometry for members of the set.

In general, a genetic algorithm (Holland, ~.~., 1992 and Goldberg, D.B. 1989, which are herein i~corporated by reference~ operates on 'genes~ to produce vAr;At;~n which through selection yields 'survivors~. The genes of survivors (as judged by 'fitness~) are~then mutated to produce newer progeny for further fitness selection.
Thus, mutated genes, according to the genetic algorithm of the invention DIOGAM, are produced a~d encode altered surfaces, which in turn=are:altered phenotypes.
The definition of a "gene" for use in the model of this invention is a specific sets of values for the parameters of Gj: (~i, Xi, ~i, ~, Yi, Zi)i- Varying these parameters changes the position of the surface Sij which we define here as the phenotype of the given gene.
Herein, [{Gi~}]j=l,M is a population of _ variations of the model Li, which encompass all possible ways to vary the surface of the model, on each member of the TSA+ rVab set which gives rise to subsequent models (lst progeny generation, 2nd progeny generation, nth progeny generation models [l-n]) wherein n = the number of=the generation.
The initial creation of pr~l;minAry models follows in one embodiment the Computer Vision algorithm for structural and surface comparison of proteins (Fisher et al.; 1994) using a small number of points, rotational and translational in nature for unique definition This method is based on the previous method of the Geometric Hashing Paradigm (Lamdan ar~ Wolfson 1988 and Lamdon Schwartz and Wolfson, 1990). This method finds 3D motifs within different segments or by isolated single amino acids, independently of any~linear sequence of amino acids. The later provides for incorporation of all important amino acids or groups thereof located within the ~ CSRs and ~ CHDH3 and which by themselves aO not occur in a singularly linear sequence within any rVab.

W096/04ss7 ~ 6 7~ Pcr/US95/10182 ., .~ }"

Using only distance lnvariants, this program obtains data from surface superpositioning which is then used to solve for portions of the rVab which represent analogous portions of surfaces of ligands directly involved in ligand-target bi~ding requirements, i.e., the 'docking 5 problem.' Various types of surface superpositiQning can be used, and includes docking of rVabs, one rVab and one target, and one rVab and on~e target related ligand.
DIOGAM uses an efficient automated computer vision based technique for detection of three dimensional gtructural motifs (Fisher, D., et al., 1992; and 3achar, O., et al.
1993). In this process, seed matches are found first;
based on the Geometric Hashing Paradigm, the clusters of seed matches are found using rotational and translation parameters to fix 3D motion. Here the seed matches will 15 be done within specific sized balls, using different pairs of balls, the subsequent clustering added by known CSR
structure~and CSR and CDR relationships within each rVab.
Eixtensions will be extensive, eventually including all amino acids within each CSR and CDR, using reiterate ever 20 growing cycles.
Such clustering and extension (referred to here as additional level ~t;nr~C (see below)) can be used for both chemistry and energy analyses. ~odeling will initially be done individually, then in an aggregate 25 manner.
Therein for each progeny generatiQn, the sum of {SuD}, wherein j=jth member of the ith scanner as appearing in the nth generation gives us a Target Fitness Landscape (Ti): which is a set of numbers representing chemical and 30 gecmetric properties of the maximally overlapped set of SD. For the purposes of this invention, Tn is a vector whose components, tj, include but are not limited to scaled electrostatic energy, buried surface area, hYdLU~
bon~Ling, and local curvature.

W096/045~ PCT~S95/10182 219667'~

As the algorithms proceeds, it r~lrnli~te5 at each stage, the target fitness landscape (T) and ascertains a mutational strategy for the next stage. Thus, ~p~n~;ng upon the strategy, all ~ genes are mutated, producing new phenotypes for which a new value of T is calculated. The process is complete when T can be m~lm- ~d no further.
Thus DIOGAM alters the set of {~i, Xi, ~i, X, Yi~ Z;} in order to achieve the best overlaps in the general sense (geometry, energy and chemistry) and the result is new Target Fitness T,~n~ri~p~r (i.e., T~ de_ined to be a minimum when maximum generalized overlap has been achieved. __ _ _ _ The next or intervening phases o$ DIOGAM allow variation (i.e., mutation) in the ~i themselves thus the genetic algorithm include s genetic varation of CSRs and CDRs. Eor DIOGAM, the mutated gene (i.e., the augmented or varied gene) is the collection of rotamer angles of the side chains themselves within the CSRs and CDRs. Such changes would include, as example, changing the rotation around a C~-C~ bond (C=carbon), which for a valine put it in result in 3 different positions). For an arginine, there are up to 27 rotomers of the guanidium group. In the preferred mode, structural variations will be carried out early on. ~nnc;~rr;n~ mutational events, another level of variation could be rocking of the models.
Further mutation (i~e~, variation) would be changes in the angle between VH and VL from O-l~ degrees, which has the e~fect of shifting the target residues within the genes over a longer distance which can be rr,nq1~r~ shifting C
~ positions~ These mutations will include 'catastrophic events' having global implications for the position of the amino acid within the CSR or CDR~3~ These mutations enable local minima trapping to be avoided~_ Although the above mutational events are the first two preferred, the order of changes will ~e mn~ 1 f; r~ during the overall DIOGAM program~

W 0 96104557 ~ 1 g6 ~ 7~ PC~r/US95/10182 No~e that VH CDRH3 is a special case. This is so because first there are no canonical 8tructures for CDRH3, second, it is by far the largest CDR region with insertion sizes of up to close to 24 amino acids; and third, because it can influence the angle between VH and VL. Therefore, this region is the one of most variations with the least structural restrictions.
~ rrnr~n~ to the prefer~ed mode of the invention, there are two positions within each CSR gene, which do not alter its canonical structure, and which are r~nr~ ;~ed in the rVab lib. as to amino acid. This translates to the po8sibility of any one of 20 amino acids being present at these two po8ition8 within each CSR and CDHR3 within any one of the r; m~mhers selected TSA+ rVab set under analysis. Therein, in the first level variation phase of DIOGAM, there is an arbitrary 'mutation', herein meaning rotation, of the gene allowing prP~Gnt=t;nn of the various possible rotamers for these two particular amino acids found within one TSA+ rVab at each of the two r~nr~ ;~P~
po8itions within the gene. Such mutation events will also be used later with VH CDRH3 at its two randomized amino acid position.
These mutants will then be analyzed by DIOGAM to derive other sets of Tl~ in the manner described above.
Additional mutational events may also be utilized to produce further diversity to more fully describe the miniml1m gtructural re~uirements to define the common overlap (i.e., BEEP) which has the best TSA+ phenotype for the active ~ite of the Target. Mutational event8 which ~ effect fitness, will involve, but not be restricted to hydrophobic, electrostatic and confnrm~;nn~l entropy effects, 8urface roughness, surface curvature, avoidance of unpaired charges, favorable and unfavorable steric interaction of functional groups and will be characterized by available ~L~L~,I~ like COGEN (Bruccoleri, R.E., and Kar,olus, M., 1937; Novotny, J., Bruccolerir and R.E.

L ~ . . .

W096l04557 ' ~i PCT~S95/10182 2i96~7~

Saul, F.A., 1989; and Tulip, W.R., et al. 1994) and the multiple copy simultaneous search method of CHARMM
(~;r~nk~r, A., and ~arplus, M., 1991; Patai, S. 1989 and Brooks, B.R., et al , 1993) using functionality descriptors with fewer atoms (Andrews, P.R., Craik, D J., and Martin, J ~., 1984) or a spherical apprn~;mi~t;nn to a multi-atom group (Goodford, P.J., 1985 and Goodsell, D.S., and Olson, A.J. 1990) based on time dependent Tartree approximation or m;n;mi~iqtion (Flber, R., and ~arplus, M
1 9 9 0 ) .
Once these mutational levels ~1~-n~ level mutations) have been=gone through one~time, for each ~j~, there will be new children ~perhaps hundreds to th~nqi~n~c) of the original parental rVabs. Structural parameters of the second are then put through the 'Nussinov-Computer ~ision' algorithm (Fisher, et. al ~1994), which is included herein by reference, to obtain the best alignment. Details of this method and some ;~rl;~ist;nnq o~ the program (Fisher, D., et al., 1992 and Bachar, O. et al. 1993) are included herein by reference. The lowest values of the target functions for each Tn, will be dif~erent. The values will include, but not be restricted to, rms (for geometric overlap), ~G (Gibbs free energy) and chemistry. The mutational events will produce progeny which will be selected as having ~rms, ~energy and cnegative chemistry 2~ values than those of the pi~r~nt~l targets. Together the sum of these values define an overaIl Target Fitness Bandscape for=each Tn.
At this stage, DIOGAM will use commerclally available algorithms, as described (see Goldberg 1998~ by providers, and known to those skilled in the art, to score and register the results of each fitness test. At this stage then, there will be a list of ~i~x~ xilyi~zi for each Ljn and a running fitness score (Tij=). DIOGAM then goes back to next cycle of genetic variations, doing these iterations for th~ncAn~ and th~n~;~n~ of generations, W096/04557 21 9 66 79 PCT~S9~10182 simultaneous, or in an ordered fashion, which at its ternination will provide a list of best minima, which will be the 1st level PEEP, i.e., the best overlap of the surfaces r~ntA; n~ within the set of active TSA+rVab We have done this manually in the case of two antibodies (N 10 and NC41) to the same site (epitope) on the surface of n~nrAm1n;dase (Tulip, W.L., et al., 1994) and Malby, R.L., et al., l994? which have been defined crystallographically and which provides us with a population, here only rrntA;n;ng two member~, which apprr~1mAt~r t~he TSA+ rVab population isolated by this invention Analysis of this population has shown overlap of antiboay CSR and CDR surfaces w~ich are bound to the same epitope. Therefore, a Sij surface as envisioned by this inVentiQn can be made.
At thi8 stage, DIOGAM now goes back to the mutation stage and iterates, i.e., arbitrary changes rotamer positio~, overlapping the ~et, yet in so doing producing a slightly different set of ~i, X~ , X, y" z, but more importantly, finding Ts which are different (higher or lower) frQm its predecessors. Thus every character of every gene will be updated to reflect the fact that it increm~ntAlly (differently) contributed to a more robust phenotype (target ~itness landscape).
DIOGAM directs the algorithm to enter into its next stage, initiated after many such mutational iterations, its crossover or recombination stage, wherein it creates new co-mb;nAt;r~n~ of yenes, even without knowing what is good (better fitness) about an existing gene mutations.
These rl ~;nAtions~ i.e., mating, of genotypes (or isogenotypes) are based on T scores, equal phenotype selection of better fitness, wherein fitnes~ is defined as contributing to maximal overall overlap.
- It is noted here that overlap is not restricted to physical occupation of identical space, but includes overlap defined, for example, as charge n~ntrAl;7At;on ., ~
W096/04557 ~ PCT~S95/10182 , 219~7 o - 90 wherein, for example, two negative charged residues may be scored as 'overlapping~ i~ they each could be within some distance o$ a positive charge.
In this entire process, it is important that the test tube selection of TSA+ rVab from the large rVab libraries, selects the right cnmhin~t;on of genes which presently in no way can be guessed in advance. By ~f;n;t;nn, the combination existing in th~ ~t;ve TSA+ r~ab is 'correct' as it rnnt~;nC the surface n~pqs~ry for desired activity profile, i e., consisting of one or more of the desired attributes of affinity, selectivity and or activity on the target.
To summarize, in our genetic algorithm, DIOGAM, the gene is the object, the mutation is the change and the early selection is the testing by iteration to get a better number O~f individual genes. This is then followed by crossover using genetic logic of=pieces of genes which are responsible for the fitness. This crosslng over and ro~nmh;n~t;on in the preferred instance ;n~ln~ deletions and additions Of single amino acids or groups (referred to a seed clustering, or extension or simplification~. With regard to additions, this ; n~l n~ those amino acids within the CS~s, CDR and fL~..~.~J~h domains of the rVab which have not been r~n~ '7~, and ;n~ln~ those within the CSRs which are critical to the canonical loop structure itself. The importance of deletions and additions to genes as later mutatio~al events is important as pnhl; ~h~ data (Malby et al. 1994) shows that for two antibodies binding to the same antigen epitope, one of the CSR in the pair does ~ot make contact with the target surface and that large target recognition domains may themselves contain much smaller domains which are ~:
responsible for the most of the energy of target interaction (Clackson and Wells, 1995). For the purpose ~ of this invention, the Ti o~ the best common overlap, i.e., the BEEP, is related ~o the existence of a small WO 96/04557 2 ~ PCT/[rS95/10182 196679 :
~ ~ ,. i ,. . .
,i ~ .

subset of high energy density points in the atoms target surface (Clackson, T. and Wells, J.A 1995; and Tulip, W.R., et al., 1994), which i9 considerable less than all contact residues. This is expected to simplify the alignment (i.e., overlapping) of the B; for example if the target domain which is responsible for the TSA+ phenotype of the set selected rVabs is assumed to have iust two hot spots then tfhere is a very restricted number of ways a given antibody, known to interact with the site so as to have a TSA+ phenotype, can bind to that site.

D. Identify small organic molecules active at t~rr~et sitrC
l Use of BEEP as high volume screening rea~ent The BEEP provided by this inv~ention may be used as follows to identify SO~ERS or drug leads.
a. Use BEEP to electronically screen C~EMEIBE to identify SOMERS as discovery leads uslng computer structural pLO~l~L~ commercially available and known to those skilled the art.
b Use the coordinates of the BEEP to 8creen via existing computer technology entire chemical data bases for matching SOMERS.
c. Select a few SOMERS and test in vitro and in vivo to confirm discovery lead.
d. Use BEEP to direct synthesis of active SOMERS
via techni~ues known to those skilled in the art of medlcinal synthetic chemistry.
2. Identification of SOMERS using rVAB-Report~r8 a. Select 1-2 representatives of each surface domain group within the active-selective rVabTSA+ library and enzymatically label with, for example a r~1O~r~1~P

W096/04557 PCT~S951~0182 219~

b. Establish competition binding assays using endogenous ligand and known allosterlc target regulators as displacer labelled rVab~SA reporter.
c. Screen chemical l;hr~r;~ via standard automated binding assays for SOMERs which displace labelled rVab from its target. Identify all close analogs of active SOMERS and perform SAR for target binding.

3. In a preferred Pmhr~;m~nt, DISOMERS are identified as follows (See Figs. 21-and 22~: s 0 a. Start with all rVab which recognize a surface on pharmacological targets These can be selected following steps described above.
b. Modify the phage rVab, rVvx library to contain one or two large random peptide libraries sufficient to occupy the other one or two TARGATTS which together make up the active surface of the target. After identifying a scanner rVab to identify one TARGATT ;~rnt;~;r~t;o~ of the others is ~rcrm~l;shed which may also be done in the presence of the first discovered SOMER. Do limited SAR on each SOMER to identify the inactive Pl~m~nt~, covalently oligomerize the two or three SOMERS via linkage through their inactive surfaces to make a DISOMER or TRISOMER.
Test in vivo and in vitro to identify best Discovery Lead.
c. Test most potent SOMERs for activity using an in vitro target assay. _ d Test in vitro active SOMERS with best CAP in vivo ~via I.P. route to identify Discovery ~eads.
If no aLalogs exi8ts of originally discovered SOMERS, carry out limited synthetic effort, use A*rVabs or rVvx to do a limited SAR binding study and then select best and test in vitro and in vivo for entire CAP.
If label reporter A*rVab or rVvx for a particular target domain does not uncover SOMERS or none are displayed by ~ndu~ us ligand, perform secondary W096l04ss7 PCT~S95/10182 ~I ~6~7g - 93 - = : -simplication or diversification of CSRs and CDRs, reselectfor the TSA+ and carry out 3A again.
Screening for small organic molecular replacements (SOMERS) will be done by me~hods known to those skilled in the art using robotic assay employing labelled n[~]rVab with specific CAP and searching for compounds which displace [~]rVabT+ binding to targets e. Excise all rVXCHrdomains from rVXCXT+.BIB, move into the plasmid for bacterial periplasmic expression and create a library of soluble VSCXT+. Mix this library of soluble rVXCB~+ entities and a phage library of rVBC~
displayed attached to the phage coat protein through its CB region (rVBCB.BIB) to make a combinatorial library wherein only one member is packaged in the isolated phage and pan against target protein as in 2Aa. After enrichment (2-4 cycles of selection for one or more of the three desired properties) the genes for the active rV~C~
entities are obtained. The genes for the active rVXCXT+
entities may then be obtained in a manner similar to that used to obtain the rV~VB genes. After exision of both the rVBCX and rVXCX genes, the Cre-Box rec~mh;nAt1~n system ~see below) may be used to construct a single phage ~n~A1n;ng both chaing and for expression of the rVab.BIB
as a phage displayéd functional complex. In another embodiment, the libraries may expressed as single chain versions with VX and VB coupled through a linker using commercially available kits, such as those from Cambridge according to the manufacturer. Finally, enrichment and selection of VX:V~ ~mh1nA~10ns which possess the desired target attributes may be obtained by, for example, panning-W096/04557 PCT~S95/10182 21 ~66 7~ - 94 ~

Rl~MPT.R 1 Construction of a Recnmh;n~nt Surface Scanner rVab Lihr~ry (rv~h.lib)~

VII. Selection of Parental Fabs of known crystalline structure ~c rVab librarv fr~m~work ~emnlates The amino acid se~uences and crystalline structure of the light and heavy chains of the antibody ABXXX which is used as the parental Fab for construction of the rVab.library are obtained from the Brookhaven Data Base, the Kabat Data Base, ~NRR~N~ (email: NCBI.NIH.GOV_) or Kabat, F.A., et. al. (Rabat, T.T.Wu et al. 1991). The V
regions of the light and heavy chains are subdivided in domains as follows: the highly variable complementary ~rtrrm;n~nr~ regions (CDR), the r~nnn~r~1 structure region (CSR) within each CDR, and the intervening rL JLh regions (FWR)(Fig.2.5.6). Individual amino acids not within a CSR or CDR, but nevertheless essential to the canonical structure (Chothia and ~esk 1987; Chothia, ~esk et al. 1939; Kabat, T.T.Wu et al. 1991; Chothia, Besk et al. 1992) are also listed (Fig. 5, 6) ABxxx is selected as the parental fL ..~Lh template for the construction of the ABxxx rVab.lib for recognition of target surfaces by an antibody with a planer type antigen r, ;n;ng site. This selection is based on the following: 1) av~ h;l;ty of the crystal structure of the antibody (bound or free of corresponding binding partner, i.e. antigen); 2) the antibody is a member of the planer type r~ ;n;n~ site group of Fabs (Webster, Henry et al. 1994) which have been found to recognize protein ~ surfaces; 3) the antibody has canonical structures for CSR
H1-2 and ~1=3; 4) the antibody's CDRH3 size is in the mid-range of sizes of CRDH3 (so as to favor er~ual usage of all 6 ~DRs of the rVab in target recognition=(Wu, Johnson et al. 1993); and 5) the antibody's antigen is a protein (Fig.3). Parental antibody fL~~ Lh~ found in antibodies WO 9610455'7 219 6 6 7 ~ rC~/~J595~10182 g ~
with a cavity and a grove yroup type combining site ~classification as reviewed by Webster [Webster, Henry et al. 1994]) will al80 be used to make two additional rVab libraries ln a fashion 8imilar to that described below for the rVab lib based on ABXXX. To~ether these three libraries generate a 5l-ff;r;Pntly large number of probes for surface recognition~of relevant binding sites.
In the A~xxx rVab.lib the natural diversification of antibodies is provided by placing within the library varied comb;n~t; ~n~ of VH and VL domains which themselves have varied co-m-bination8 of the known canonical CSRs, variable length CDRH3s, and randomized amino acid8 (one of 20 essential amino acidg) at one or more amino acid positions within the CSR or CDRs of each V region=within each rVab (Fig.4).
VIII. ~-~ ~reating t~e Nuclelc Aclds Bncoding the Heavy and Light Chains (rVHCH1 and rVLCL) for ABXXX
rVAh.~ ;hr The nucleotide sequence of A~xxx i8 obtained from Se~Pnres of p~otP;n~ of-Jmmlln~loq~r~ TntPrest~ 5th ed.
(Kabat, E.A., T.T.Wu et al. 1991); the Kabat Data Base (NCBI.NI~. GOV); or GENBANK T~Pntlf;catlon and analysis of all restriction sites present within these sequences may be accomplished using a commercially available program (GCG [Univ. Wisconsin,USA], MacVector [IBT,Kodak,New Haven, CT], DNAStrider (C. Ma~ck, Gif-Sur-Yvette Cedex, France, Service de Biochemie, Inst. Reg. Fnn~m~nt~1, Aloric Energy Commission of France) and SeqBd,[Applied Bio8ystem]~.
Restriction sites endogenous to ABxxx and conflicting with construction of the rVab.lib as outlined below are removed and replaced with other nucleotides not encoding the r~nrl;rt;n3 restriction site. This is done using sequences which keep unchanged the identlty of the parental amino acid(s).
.

..

W096/0455~ 6 7 '~ PCT~895/10182 v The sequences are then analyzed again for the changes necessary to place the convenient and unique restriction sites throughout the V and C genes needed for library construction as outlined below.
The PEXXX rVab.lib is built according to this invention from separate rVLCL ~Fig. 7) and rVHCXl (Fig.8) chains which are combined randomly in an 'n vivo process (Fig.14). The construction of the rVLCL and rVHCH nucleic .
acid libraries encoding the rVLCL and rVXCHl chains, is accomplished in steps outlined as follows: ~tep l) 10 oligonucleotide synthesis constru~iDn Df a) amino terminus end (5'V), b) a midregion ~MIDV) for V~ only, and c) a carboxy-t~rm;mlr end (3'V) o~ the V region; step 2) diversification via PCR Df some CSRs; step 3) ligation of the sections; step 4) diversification of the L~ ;n;nr 15 CSRs; and step 5) ligation of the appropriate constant (CXl or CL) region derived by PCR or:oligonucleotide construction to generate the complete recombinant heavy and light chain libraries ~rVXCXl.lib and rVLCD.lib).
Ste~ l: Construction of rV~Cr!.l;h (Fi~. 7) In the oligonucleotide phase (step A, Fig. 7), construction of a) the 5' (5'VL) end; b) the Mid section (MIDVL) and c) 3' (3'VL) end oi the VL region uses eight synthetic oligonucleotides comprising four complementary pairs. Each oligonucleotide (x) has a 25 complementary mate labelled x'. T~o oligonucleotide pairs, a/a' and b/b' are used to make the 5' end. The MIDVL (c/c'), and the 3'VL~(d/d') sections are each synthesized from one oligonucleotide pair. The amino acid and nucleic acid positions encoded by the specific 30 oligonucleotides are shown in Fig. 7.
The variance in amino acids at position 2 (within a/a') and 71 (appended to c/c') n~r~qr~ry to allow for construction of all the desired VLl CSRs is added during later steps as described below. All oligonucleotides are 35 synthesized so as to have at lea~t one overlapping W096l04ss7 ~ Ol~
~ 21966~
f complementary sticky end, an absence of hairpln forming ends, and to be nnncnmplementary to se~uences other than that of-the desired oligonucleotide joining partner based on analysis by a commercially available oligonucleotide primer analysis software program.
Step l~a): ~nnrtr~ction of ~VL Sert;~n For construction of the 5~V~ end section in step l(a), the oligonucleotides are first phosphorylated, then mixed together in one reaction mixture, heated, annealed and ligated together using generally known molecular biology technology (Sambrook, FritsCh et al. l990). The product is then isolated and ligated in 60 ~l reactions with 1200U T4~DNA ligase (New England BioBabs) to ~ ~g pCBONABB (see Eig.9 which lists all general use plasmids) digested at restriction site (rsj prsO and rs4 ("p"
signifies that the location of the restriction site is within the plasmid and outside of the rVab se~uence) (Sambrook, Fritsch et al. 199D).
DNA is purified from the ligation mixture using Gleneclean II (siolol)~ resuspended in water and used for transfection by electroporation (Dower, Miller et al.
1988) of ~. coli TGl (Gibson 1984) grown in broth rnnt~;n;ng l~ glucose for lh an~ then plated on dishes in antibiotic rnnt~1n1ng media. After overright (o.n.) ;nrllh~t;nn at 37'C, individual colonies are picked.
Colonies are identified as rV~3-24.bact first by diagnostic PCR using primers pCFWD and pCBCR (see Primer Table, Fig.10) and subseriuently confirmed by se~uence analysis via automated a~ A}3I ser~uencer and commercially c available related kits as olltl;nr~ by manufacturer (ABI,USA). Storage of positive clones at -70 C is done in broth (Miller, 1972) rnnt~;n1nr~ 15~ (v/v) glycerol.
Step 2: Diversif~cation BY p~R
Toothpicked froze~ glycerol stocks of rV~3-24 are used in PCR reactions to append primers corferri~g diversi~ication to the rVB section One of the five W096l04557 ~ PCT~S951l0l82 ~ ,, ~
2~g~

O - 9~
different CSRLl diversified with random amino acids at two positions is used as the FWD primer~at the 3' end of the parental ab~a'b' 5'VL section_ The BCE primer for the 5' end comprises nucleic acids encoding one of the tb~ee different amino acids I,V or S at position VL2, and the amino acid of:the parental ABXXX at position VLl.~ These appendings are done i~ 5 primary PCR reactions, each rnnt~;n;ng one FWD prlmer (i.e., Ll.lFWD, Ll.2FWD, ~1 3FWD, Ll 4FWD or Ll_5FWD) and one of three different BCK primers i~ the following combinations: Ll.1-3BCK
primer mixed with the 3 reactions rnnt~;n;ng Ll.l, L.12 and ~.13 FWD primers, and Ll.4BCk and Ll 5BCk mixed correspondingly with one of the two Ll in;nr Ll-~FWD
primers. Subsequently, amino acids VL34-44 are appended to the primary PCR products in secondary PCR reactions by taking an aliquot of the primary reaction and carrying out secondary PCR with primers LlALLFWD and LlALLBCK. The products of=~he secondary reactions are kept separate and are labelled rVLl-44CSRl.1-5.1ib.pcr. These constructs allow subsequent generation of all 5 known r~nnn; r~l CSR
~1 in the rVL.lib after cloning when these products are ~oined with the appropriate MIDVL section having one of three different amino acids in position VL71. Each of the primary PCR uses Taq polymerase, FWD and BCK primers as noted above, in 50 ~1 reaction mixtures and is cycled 25 times (94 C for 1 min, 60'C for 1 min and 72 C for 1 min).
The secondary PCR reactions (25 ~1) use fresh Taq polymerase and 1 ~1 of amplified appended diversified primary PCR reaction mixture product, FWD and BCK primer pairs as noted, and the reaction is cycled 30 times (94 C
for 1 min, 55 C for 1 min and 72 C for 2 min). A list of the sequences of all primers appears in Primer Tab~le (Fig.10).
In step C, the five products of the secondary amplification reaction of correct size, are designated rVLl-44CSRl.1-5, ana are isolated on ~ow percenta~ge W096/04557 ~ 9 6 ~ 7 9 PCT~595/10182 ~ .

_ 99 acrylamide gels, recovered, restricted and ligated to pCLOXALL precut with prs4 and rs2 and cloned via electroporation (Dower, Miller et al. 1988) into E. coli as described ~step B, Fig. 7). These five 5'VL section products are designated rVLl-44CSRl.1-5.1ib.bact. Twenty clones of'eac~'library are checked first by diagnostic PCR
and 8ubsequently five (5) clones are analyzed for diversification of CSRl by automated sequencing as described above using pCFWD and pCBC~ sequencing primers and commercially available kits (ABI,USA). This procedure generates greater than 10~ transformants per each of the five VLl CSRs.
Ste~ ~(b): Crnctruction of the ~T~VL section In parallel fashion, a second set of reaction steps A-C constructs the MIDVL section of rVLlib. The MIDVL
section originally rr,nt~;nc amino acids rVL53-68. The oligonucleotides for this reaction are rrnt~;n~fl in the one pair c/c'.
In step A, each oligonucleotide is phosphorylated, the pair hybridized together under annealing conditions, and the c/c' double stranded DNA complex is purified and ligated in a 60 ~l volume with 1200U of T4 DNA ligase (New England BioLabl to apprr~;~-trly 5 ~g rs2 and prs5 cut pCLONALL (Sambrook, Fritsch et al. 1990). Ligated product is isolated from the mixture using Genecleanb II (BiolOl), resuspended in water and used to transform E, coli via electroporation (Dower, Miller et al 1988). After 1 hr in broth rrnt~;n;ng 1~ glucose, the cells are placed on dishes in antibiotic rrlnt~;n;ng media After overnight - ~nrnhst;nn at 37 C, individual colonies are picked and the MIDVL section transformants are ;flrnt;f;~fl from among 30 transformants generated by diagnostic~PCR. ~r~nf;r~t;on=
of ser~uences is by automated sequencing using an ABI
automated sequencer using pCFWD and pCBC~ primers (ABI,~SA). Positives are labelled rVL53-68.bact. and frozen glycerol stocks are produced.

W096/04557 ~ PCT~S95/10182 2 ~ 7 g In step B diversification, PCR is used to append diversified CSRB2 to the 5'~end of MIDVL. Three different amino acids at V~71 (i.e., Y, F and A) followed by restriction site rsC between V~72 and V~76 followed by a rs4 restriction site are appended with primers to the 3' end of MIDVB These additions are done in three separate reaction mixtures, one each cnnts;n;ng FWD primer L2.71YFWD, ~2.71FWD and B2.71FWD All three ~WD primers contain the rsC site which will allow joining of MIDVB to 5'V~ sections. For each of these reactions, the BCK
primer is ~2A~LBCK which cnnt~;nq an rsB site as well as DCSRB2 diversified at amino acid VB5~ and 51. Each mixture cnnt~;n~ a toothpicked frozed glycerol stock of rV~53-68 (see Primer Table,Fig.10), Taq polymerase, in 50 ~l mixtures, and is cycled 25 times (94 C 1 min, 60'C 1 min 72'C 2 min).
In the following step C, approximately 1 ~g of the amplified diversified appended MIDVL products are isolated using Magic PÇR Preps (Promega), cut with prsl and rs4, reisolated and ligated to 5 ~g pC~ONA$L precut with prsl and rs4 in 60 ~l volume with 1200U T4 DNA ligase (New England Biolabs) (Sambrook, Fritsch et al 1990). The ligated plasmid DNA products are isolated using Geneclean II (BiolO1), resuspended in water and used to electroporate E. cQli to generate, as noted~above, a library of transformants ~Dower, Miller et al. 1988).
The three separate groups of successful transformants (one for each type of VL71) are identified by diagnostic PCR
and confirmed regarding diversification of VLCSR2 by automated sequencing of 10 clones of each group. These transformants are designated rVL38-73CSR2:71(Y,F,A)lib.bact. This procedure gives >104 transformations for each group.
Ste~ l(c): Construction of the 3'V~ section of rVL
In the third set of parallel steps~ A-C~ the 3'V~
section of rVB lib is constructed. This section is W096l04sS7 2 1 9 6 6 7 9 PCT~59s/l0l82 originally buiit to contain amino aclds VL72-90 and uses the one oligonucleDtldes palr dJd'. In step A, this pair is phosphorylated and the two oligonucleotides annealed.
The double stranded complex is then isolated and is ligated to pCLONALL precut with prsO and rs4' Ligated product is isolated and used to transform E. coli via electroporation ~Dower, Miller et al 1988~ as above.
3'VL section transformants are isolated from among the transformants generated, and diagnostic PCR is preformed on twenty of them, the positives being confirmed by automated se~uencing and labelled rVL76-90.bact. Frozen glycerol stoc~s are prepared.
In the next phase, diversification (step B), the six diversified CSRL3s, followed by a new prs5 site, as well as amino acids VL72-75 which contain the convenient restriction site (rsC), are appended to VL76-90 to make the followiig 5'VL PCR product: rVL72-lOOCSR3.1-6.pcr.
Diversification of CSR3.1-6 occurs at positions VL92 and 93. These proce8ses are done in six ~6) separate 50 ~l PCR reactions each cnnt~inJng one L3.1-6FWD primer, all nnnt~ln1ng L3ALLBCK (see Primer Table, Fig.10), and Ta~
polymerase in 50 ~l mixtures. The reactions are cycled 25 times (94~C 1 min, 60~C 1 min and 72~C 2 min).
In step C, the amplified diversifled appended products are isolated using Magic PCR Preps (Promega), cut with prs2 and rs5, reisolated and ligated into pCLONALL
precut with prsl and prs5. The ligated plasmid DNA
products are isolated and used to electroporate E cQli to generate a library of transformants as noted above and designated rVL72-lOOCSR3.1-6.1ib.bact. This procedure gives greater~than 10~ transformations which are identified by diagnostic PCR and sequencing to contain a~ LIately randomized amino acids at the diversified positions within VLCD3 for each of the six (6) VLCSR3s.

=_ - : - - . . - - .~

W096/04557 . ~ ; '';A PCT~S95/10182 -~l02 -Ste~ 3: Li~atiQn E~ _ In step 3, the 5'VL and MIDVL sections are joined (see Fig. 7). Five ~g of DNA of each of the:five ~
rVL1-44.1ibs (i.e., CSR1 1-5) is digested with rsB~and rs5 and ligated to 1 ~g of insert isolated from the three rVL38-70CDRL2:71* using 1200U T4 DNA ligasez((New England BioLabs) (Sambrook, Fritsch et al. 1990). In these reactions, ligation pairiny of 5'VL~rVLi-44CSRs~ to MIDVL[rVL38-76CSR2:71*] is ~lnt~lnP~ as: 5'VL1.1-3 x MIDVL2:71Y; 5'VL1.4 x MIDVL2:7lF and 5'VL1.5 x MIDVL2:7lA
to create the five rVL1-76CSRD1&2.DN~s. Each of tXese is used to electroporate E. coli (Dower, Miller et al.
1988).
The bacteria are then grown in broth cnnt~ln1ng 1 glucose for 1 h and are plated on dishes in antibiotic cnnt~1n1ng media. After overnight incubation at 37~C, individual colonies are picked and are characterized first by diagnostic PCR and then by automated sequencing. Some 100 colonies are analyzed by diagnostic PCR and 20-30 by sequencing to confirm the random presence of different CSR
pairing and diversified amino aclds within the various CSRs. Frozen stocks of the five groups are then prepared and are designated rVL1-76CSR12.1ib.bact.
In step F, the o~tPn~P~ 5'VL halves, consisting of the five rVL1-76CSR1&2.1ibs., are joined in 30 separate PCR reactions in combinatorial fashion with the six 3'VL
halve sections, consisting of the six (6) rVL72-lOOCSR3.1-6.1ib. This process generates 30 ~ull length rVL1-lOOCSR1~2&3 1ib. (as diagramed in Fig. 7). In each of these library constn~ctions, about 5 ~g of DNA of each of the five rVL1-71CSR1&2.libs (i.e., CSR1.1-5) is digested with rsC and prs5 and ligated to 1 ~g of each of the inserts isolated from the six rVL72-lOOCSR3.1-6 digested with rsC and prs5 using 1200~ T4 DNA ligase (New England BioLabs) (Sambrook, Fritsch et al. 1990) to create the 30 rVL1-lOOCSRD1&2&3.dna preparations. Equal aliqouts Wo96104557 2 1 9 6 6 7 9 PCT~Sg5110182 '.~ ?.~

from each ~igation mixture are:pooled and the pooled DNA
i-s-puri~ied using Geneclean II (BiolOl) and resu~pended in 30 ~1 water~to create the completed rVLCL.lib.dna. PCR is then used to append to the 3' end of this DNA library, the nucleotides encoding the r~m~tning amino acids of V~ (i.e.
rVL101-107),~ amino acids at the 5 'end of CL (i.e., amino acids C~ 108-110), and within this sequence the convenient rs3 site. The rs3 site, also designated the rsCLLNK site tFig.9~, subser~uently allows the ~oining of rVD.lib with its cloned rCD section.
These appending reactions are done by carrying out a PCR reaction with an aliquot of the purified rVLl-lOOCSR1&2&3.1ib.dna, the primers W CB~N~FWX and ~lALLBCX, and the Taq polymerase in 50 ~1 volume mixtures cycles. The PCR reaction is cycled 25 times (94~C for 1 min, 60~C 1 min and 72~C for 2 min).
Amplified DNA is then purified using Magic PCR Preps (Promega~. After suspension in water, i ~g of the purified DNA is digested with rs2 and prs5 and ligated to 5 ~g of pCBONADL DNA precut with rs2 and prs5 using 1200U
T4 ligase (Sambrook, Fr1tsch et al. 1990) and used to electroporate E. coli (Dower, Miller et al. 1988). The bacteria ~rown in broth cnnt~;n;ng 1~ glucose for 1 h are then plated on dishes in antibiotic cnnt~inlnr~ media.
After overnig~t inrnh~tlrn at 37~C, individual colonies are plcked and characterized first by diagnostic PCR and then by automated sequencing. So_e 100 colonies are ~min~ by diagnostic PCR and some (about 5-10) by sequencing to confirm the presence of amino acids Vhl-110 - and the random presence of different CSR pairings and diversification of amino acids within the various CSRs.
More than 10~ transformants are gen-eratea in this process and a frozen stock of the library:is then prepared and designated rVD.lib.bact.
In the last step (step G) of~rVB.lib construction, DNA from rVLlib is digested with prsl and rsJCLNK, and 1 W0 96/04557 ~ t~ PCT~S95110182 21~6~9 ~g is ligated to 5 ~g of pVBACCEPTOR (Fig. 9~, precut with prsl and rsJCLLN~, using 1200U T4 ligaee (Sambrook, Fritsch et al.~l990). The product is then purified from the ligation mixture uslng Gleneclean II (BiolOl) and resuspended in water This material ie used to electroporate E. coli (Dower, Miller et al. 1988), and the bacteria are grown, after 1 hr in broth supplemented with 1~ glucose, gvernight at 37~C on dishes in antibiotic cont~1n~ng media. Individual colonies are picked and characterized by diagnostic PCR and automated se~uencing l~ to confirm the presence of c~ in the library. Frozen glycerol stocks of rVBl-110~CSRl-31ib are made and designated rVBCL.lib.bact (Fig 7).
The above detailed reactions where double amino acid randomization occurs within each CSR theoretically allows lS the construction of 2000, 400 and 2400 different CSR
~1,2,3 respectively, and a rVHlib size of 1.92 x109. This exceeds the largest published recombinant VB library made by similar (Griffiths,Williams et. al. 1994) technology by about 2 fold.
IX. t~nnqtructis~ of the Conq~n~ re~itnq of ~
The constant region (C) of the light (C~) and heavy chain (C~l) region for the selected parental Fab A~xxx (Fig. 9) is obtained either by annealing and ligating a series of eynthetic overlapping oligonucleotides, as done 2S for the V regisns, or via standard PCR of the C regions of ABxxx or any other antibody mRNA or DNA with identical C
regions. Nucleic acids encoding speciflc antibodies may be obtained from hybridmas from various sources including the ATCC. In sither case, the constructions includes the removal of endogenous restriction sites that interfere with library co~structisn and the creation of a number of convenient restriction sltes at and aro~nd the 5' and 3' ends of the C reglons so as to allow simple cloning into pCLONAL, pEXPRESSION and pV(H or ~)ACCEPTOR ~(Fig. 9).
3S For both CHl and C~ reglons, the C genes have inserted w096l045s7 21 g 6 6 7 9 PCT~S95/10182 ~

within them an rs3 site for specific ~oining of V and C
sections of rVL at or about the natural V/J gene junction for heavy and light chains. These sites are referred to ~ as either rsJCHLNK and rsJCLL~K respectively. In constructing the C sections, these two junctional rs are appended by standard PCR using BCK primers CLBCK and CHBCK
and FWD prlmers CLFWD and CHEWD (see Primer Table for sequence details ~Fig. 10). ~
The parental C nucleic acid sequence of ABXXX is amplified by PCR with Taq polymerase using primers CLFWD
and CLBCK which places the rs3 restriction site within the JC segme~t of the parental Fab at the 5' end of the C
sequence and two stop codons (TAA) and the rs4' site (AscI) just outside the 3~-end of the C region. The reaction mixture ~50 ~1) is cycled 25 times (94~C for 1 min, 60~C for 1 minr and 72~C for 1 min.) and the amplified appended C sequence is purified using Magic PCR
Preps (Promega) and resuspended in 50 ~1 water.
The reaction amplifying the parental Fab CH1 gene of ABXXX is identical, except for the following: the primers for the PCR reaction are different, being CHFWD and JCXBCK, and the CXEWD primer ~n~ i n~ a Notl site at the 3~ t~rminn~ of the CH1 region.
To complete construction of the VLCL, the amplified and J appended recombinant VL diversified CSR1 and 2 and 3 (rVLCSR1&2&3j genes are jolned to the amplified CL gene int he standard ligation fashion used above, or using PCR
(Horton, Hunt et al. 1989). Assembly PCR reactions (25~1) use Taq polymerase, 1 ~1 amplified parental JC, and 0.8 ~1 of the rVL.lib gene from above. The d~L~Llate VLBCK
primer is used together with the CLFWD and the reaction cycled l0 times (94~C for 1 min, 60~C for 1 min. and 72~C
for 2 min.).
X. (~Nsl~u~llQN OF rV~ .lih (Eiq. ~) ~n the oligonucleotide phase (step A), construction of a 5' and 3' half of the VH region is accomplished using W096/W557 ~ t ~ PCT~S95/10182 2~6~

16 synthetic oligonucleotides, comprising 8 complementary pairs. Six oligonucleotides are for the 5' half and are=
labelIed VE a-c with their complementary partners~labelled VH a'-c'. Within the 5'.VH half, the oligonucleotide b/b' pair has the rsB restriction site between amino acias rVH
22-26; Ten oligonucleotides are for the 3' half and are labelled VH d-f and d'-f'. Construction of the 3' half of the VH region is done in a similar fashion but uses three forms of the "e" complementary pair, designated as follows VH e/e', VH e2/e2' and VH e3/e3'. These correspond to the "e" oligonucleotides with elther a valine ~V), alanine (A) or arginine ~R~ at amino acid posltion VE71, respectively.
In the ~nnP~l;ng step, three types of the 3'VH half are constructed: 3'VHdef/d'e'f', 3iVHde2f/d'e2'f' and 3'VHde3f~d'e3'f. The variance ln "e" oligonucleotides within the 3'VH half is necessary to allow for subsequent construction in the rVHlib of all four of the known CSRX2 as outllned below. All oligonucleotides are synthesized so as to have a least one overlapping complementary sticky end, an absence of hairpin formlng ends, and an absence of complementary sequences other than those of the desired oligonucleotide joining partner based on analysis by a commercially available oligonucleotide primer analysis software.
Cnnq~ruction of 5' ~lf of th~ VX Reqion For constructing the 5' half of the VH region,~the appropriate oligonucleotides~are phosphorylated and are mixed together in one reaction mixture, after which they are heated and are annealed ana ligated together using generally known molecular biology technology (Sambrook, Fritsch et al. 1990). As outlined, the first phase annealing and ligation (step A, Fig. 8~ allows the formation of the 5' VH abc~a'b'c' pair. In the next step (step B~, the correct construct of 5' VX, cnnt~;n;ng a convenient rsB within its b/b' segment, is amplified with primers 5'VHFWD and 5'VHBCK (a list of names and sequences w096l04ss7 ,~ , for primers~used in VHCXl.lib co~struction appears in the Primer Table, Fig. 10j by carrying out PCR on an ali~uot of the liyated and isolated abc DNA duplex product of step A. In this step, an ali~uot from the step A reaction is amplified using the above noted primers and Ta~ polymerase S in 50~1 reactions and is cycled 25 times (94~C l min, 60~C
for 1 min, 72~C for 2 min.). The amplified DNA is purified using Magic PCR Preps (Promega) and is suspended in 5 ~l water.
Next, the product of the amplification reaction having the correct size and designated rVX1-51, is cut at rs4 (Notl) and prsl. The cut fragment is purified by Magic PCR Preps (Promega) and 1 ~g is ligated in a 60 ~l volume with 1200U of T4 DNA ligase (New England BioBabs) to 5 ~g of rs4 and prsl digested pCBONALB (Sambrook, Fritsch et al. 1990). DNA is purified from the ligation mixture using Geneclean II (BiolOl) resuspended in 30 ~l water and electroporated (Dower, Miller et al. 1988) into E. coli whic~ is then grown in broth nnntA~n;ng l~ glucose for 1 h and plated into antibiotic cnntA;n;ng media.
After overnight ;n~nhAt jnn at 37~C~ individual colonies are picked and ;~nt;f;~_ Transformants nnntA;n;ng the r~n~ h;nAnt parental 5'VH half, rVXl-51, are identified by diagnostic PCR for d~Lu~Liate size (with plasmid primers pCFWD and pCBCR). Those transformants suspected of cnntA;n;ng the rV~1-51 are P~pAn~o~ The nucleic acid amplified with PCR using PCFWD and_pCBCK are se~uenced via automated ABI se~uencing with commercially available kits as outlined by the manufacturer (ABI,USA) to confirm the ' identity of the rVH1-51 fragment. Cultures are then grown and stored as frozen glycer'ol (15~v/v) stocks and designated rV~1-51bact.
In the next step, step C (Fig.8), a diversified version of each of the four known CSR~2 is appended to rV~1-51_ This process is done in four separate standard PCR reaction mixtures (see above). Each reaction mixture , ~.

W096l04ss7 , PCTNS95110182 - 108 - = -comprises the rVH1-51 fragment (obtained from toothpicked frozen glycerol bacterial stocks), one of four FWD primers (H2.1FWD, H2.2FWD, H2.3FWD and H2.4FWD~ and the BCK primer H2A~;3CK. The four FWD prlmers are'constructed to span from amino acid 47 through 59 of CSR2~ and contain amino acid diversiflcation at posltion ~3. The four library products, are isolated, and are cut at rsB and prs5, and then 1 ~g of each purified DNA product is ligated using T4DNA ligase to 5 ~g pC~ONA~H precu~ at rsB and prs5.
As described abover the ligated,DNA is purified and l~ used in step D to trans_orm ~. coli Yia electroporation.
Transformants are isolated and characterized first by diagnostic PCR and then by automated sequencing to contain appropriate examples of~the randomized diversified versions of all four CSRH1. Frozen stocks of each, designated rVHrsB-59CSR2.1-4 lib. bact. are made.
Constructi~n of th~ 3~ ~l f Qf th~ ~H ~i9~
In a parallel fashion, another set of reaction steps ~-C are conducted to construct the 3' half of the VH
region which incorporates nucleic acid encoding amino acids 57-95 of the variable heavy (VH) chain (Fig. 8).
The oligonucleotides for this reaction contain the three sets of pairs of VH oligonucleotides, e/e'and e2/e2', and e3/e3' in which amino acid VH71 is valine, alanine or arginine respectively. Appropriate mixing (as outlined above) allows for annealing and ligation of the three different rVH57-95 ~ouble stranded cbmpIementary oligonucleotides 3'VHdef/def (i.e., VH5i-95[71V]) and 3~VHde2f/d'e2'f (i.e.' VH57-95[71A]) and 3'VHde3f/d'e3'f (i.e., VH57-95[71R]). Aliquots of these three reactions are then amplified and appended with rsD and prs5 sites in step B by PCR using 3'VHFWD and 3'VH~CK. These reactions contain Taq polymerase, as described above, and are cycled 25 times (94~C for 1 min, 60~C for 1 min, 72~C for 2 min).
The correct products are purified using Magic PCR Preps (Promega), suspended in 50 ~1 water and are then cut at W~ 96104557 PCT/US95~10182 21g667g - 109 - '~' '' " ' o prs2 and prs5 and reisolated. Approximately 1 ~g of the reisolated rVH56-95 gene fragment is ligated into 5 ~g pCLONALL precut with prsl and prs5. Plasmid pCLONALL with the rVH56-95 insert is isolated and purified using Geneclean II (BiolOl), and i8 used in:step C to transform E. coli by electroporation (Dower, Miller.et al. 1988).
Transformants are selected, and the correct three products, rVH56-95:71V;A;R, are identified by diagnostic PCR and r~n~ rrP~ by automated ~BI sequencing. Frozen stocks of each, designated rVH56-95[71V;A; R].bact. are made.
Completion of construction of the nucleic acids encoding the four known CSRH2 regions genes is accomplished in steps D and E. The three rVHrsD-56-71~-95-prs5 inserts, freed by digestion of plasmid DNA are ligated to the four rVHrsB-59CSR2.1-4.1ib which have been precut at rsD and prs5. The resultant rVHrsB-95CSR2.1-4 library is cloned into E. coli using the standard purification, ligation and electroporation processes m~tlinP~ above. Transformants are isolated and about 5Q are characterized by diagnostic PCR and 20 by automated sequenciny to confirm that they contain the expected diversified versions of the four known CSRH. The ligation combinations of rVHrsB-59 CSR2 and rVH56-71~-95 necessary to construct the fully diversified rVHCSR2 library are rVHrsB-59CSR2.11ib. with rVH56-95:71V;
rVHrsB-59CSR2.21ib. with rVH56-95:71A; and rVHrsB-59CSR2.3 and 2.41ib. with rVH56-95:71R in steps D and E.
Step E, c ~prises sequential PCR reactions to append - to the 3' end of the four diversified CSRH2~constructs rVHRSB-95CSR2.1-4 diversified CDRH3s of different lengths and the convenient JCHlLNX restriction sites (i.e., rs3), and at their 5' ends diversify their parental CSRHl and to append nucleic acids encoding VH17-rsB-24. The ~inal PCR
products of these reactions are designated rVH17-lI8CSRl&2&3.1ib and contain8all combinations of the W096/0~557 PCTNS95/10182 ., _ 2196~3 ' - llo -diversified known CSRHl & 2's and diversified CDR~ of ~hree different lergths.
These steps are carried out in the following 36 PCR
r~rtirnr. Nine aliquots of each of the four different toothpicked frozen ylycerol stocks of rVHrsB-95~CSR2.1-41ib.bact. are added to separate 50 ~1 primary PCR reaction mixtures cr,nt~;ninr Ta~ polymerase.
The forward primers H3.5FWD, X3.7FWD and H3.-IOFWD are added to 3 of the 9 tubes containing each of the four CSR2s.bact. To each triplicate set of unique forw~ard primers is added one of the following: the BCK primers Hl.lPCK, Hl.2BCK, or H1.3BCK. Tfiese primary PCR reactions are cycled 25 times (94~C for 1 min., 6~0~C for 1 min. and 72~C for 2 min.). Following completion of the primary PCR, aliquots of each of the 36 reactions are taken for a r~c~n~ry PCR reaction with new Taq polymerase, and primers X31FWD and H31BCK. The secondary reactions append VX100-rs3-118-rs4 and VH17-rsB-24 to the 3' and 5' ends respectively. The products are designated rVH17-118CSR123.1ib. followed by a combination number (e.g., l.lx2.2x3.5) which denotes the cDmbinatorial arrangement of the three CDRHs in these products. Each of the 36 library products are~characterized by diagnostic PCR and sequence analysis. Aliquots of the 36 libraries-are pooled to ge~erate the rVH17-118CSRl&2&3.1ib.
In step G, DNA from the rVH17-118CSR1&2&3 library is digested with rsB and rs4. The digested DNA is purified using Magic PCR Prep (Promega) ligated i~to pCLONAL cut with rsB and rs4, purified and used to transform E. coli as detailed above. The transformants are isolated, characterized and designated rVHrsB-118CSR1&2&3.1ib.bact.
In step H the rVXrsB-rs3 inserts are removed from the DNA of the rVHrsB-118&2&3.1ib using restriction enzymes specific for rsB and rs3 to form fragments designated rVHrsB-114CDR1&2&31.3. These fragments are ligated using T4DNA ligase (New England BioLabs) to 5~g rsB and rs3 ,, : _, . _ : ___ ,, .. , ,,,, , ,_ _ _ _ _ _ w096l04ss7 2 1 9 6 ~ 7 9 PCT~S95/10182 ~ ~. . ~
t.

digested rVH1-S1-rs3.kact. DNA. The product is then c isolated, purlfied and used to transform E. roli to generate rVHl-JCHLNK-~CSR1&2&31ib.bact. Individual clones from the library are then isolated and their ser~uence is confirmed by diagnostic ~CR and se~uenclng. The library is then stored as frozen glycerol stocks. The bacterial transformants cnntA~ning this library contain the canonical CSRH1 and H2 regio~s d'iverslfied in greater than one amino acid position, and CDRH3 of three different lengths and diversified in greater than one amino acid position. This procedure gives at~ least 105 transformations which are identified by diagnostic PCR and seguencing tQ contain appropriately randomized amino aclds at the dlversifled positions within the CSRH2 and H3 regions for the rVH1-114CDR2-3.1ibrary.
In step I, 5 ~g of the rs2 and rs3 precut pVLACCEPTOR
DNA (also referred to as pVH-CH, Fig. 9) is ligated to the rs2 and rs3 released insert rVH1-JCHLN~DCSR1&2&3.1ib DNA
(also referred to as rVHlib, Fig. ~), and the recovered purified product is designated rVHCHl.lib. This rVHCHl.lib product is used to transform E. cQli to generate a frozen stock of bacteria cnnt~;n;ng the rVHCHl.lib. Greater than 106 total members are obtained.
XI. VH ~n~ V~ l;hr~ry sizes.
The above detalled reactions where two amino acld randomlzations occur withln each CSR theoretlcally allows the constrlrt;rn of 1200, 1600 and_1200 different CSR
H1,2,3 respec~ively, and a rVH library size of 2.3 x109.
This exceeds the largest pnhl; ch~ recombinant VHCH1 library made by similar tech~ology (Griffiths,Williams et.
al., 1994) by only about 2 fold. A smaller rVH library can be made using only 2 randomizations within the CSRH1 and H2 and on~~r~nfl~'~;7~tion within each of the three differently sized CDRH3. This procedure theoretically allows the construction of 1200, 1600 and 60 different CSR
H1,2,3 respectively, and a rVH library size of 1.152 x -- , ~, :

21~657~ i --- l12 - ~:
lO8. This is similar to the largeEt rVHCHl library reported. The procedure ou~tlined below allows subsequent pairing of individual members of such sized rVXCHl libraries with individual members of ecually sized rVLCL
libraries (i.e., of 109 as noted above and Fig. 4) on one piece of DNA in single bacteria. Based on the sizes of the rVXCXl library and rVLCL library that are generated above, the potential size of the combinatorial rVab.lib (i.e , VHCHllib x VLCL lib) is greater than lO1~ members (Fig. 4).
XII. Construction of the rVab.lib (the VHCHllib x VLCT,l;h c~mh' n~ tor'~l l;h.)(Fiq.ll,12,14) In this section the phagmid (fd~) which carry the rVHCXllib, designated Lox Receiver (LoxREC) (fd~RECEIVER, Fig. ll) and the plasmid (p) which carries the rVLCL
library, designated Lox Provider (LoxPro) (p~TCl9PROVIDER, 15 Fig. ll) are constructed and then are randomly recombined in vivo within individual bacteria onto a single phage vector (fd~CARRIER) which expresse:s the rVab rCXCHl and rVLCL genes and produces on~the surface of the phage functional versions of:the rVab rVHCLl:rVLCL proteins.
The rVab library construction phase is outlined in Figs.
ll, 12.
Construction ls begun by reamplification of the rVXCHl library m~nt~;nP~ in the pVLACCEPTOR.lib.bact.
using PCR, as described above, with primers pCFI~D and 25 pCBCK. The DNA product is fsolated and cut with VXrs2 (Ncol) and VHrs4 (Notl) and is ligated using T4 ligase and standard methodology into LoxPRO precut with Ncol and Notl. The LoxPRO used in this example is fashioned after fdDOGl-2lcxVkdel as described by Griffiths, A.D. et al.
lg94) and contains an endog~ous VHCHl, bounded by a Sfil and Notl rs, preceded by a ribosome binding site (rbs), an in frame LpelB leader seque~ce (LpelB), followed by an inframe wild type loxP sequence~(XOess et al. 1982) and then an inframe gpIII qequence. In ~oxPRO, upstream fro~
35 the endogenous VXCH gene, and to be replaced by the w096/04ss7 219 6 6 7 9 rcT~s9sllol82 S
- lI3 -o lncrJmins rVLCL.lib. there is an endoyenous CL gene which is preceded leader sequence which ends in a Apa~l in frame sequence5which is followed by two terminator triplet ~ codons. The endogenous CL gene~is followed by two torm;n~tnr triplet codons, an Ascl and HindIII restriction site, and a mutant 511 loxP site (Hoess et al. 1986~. DNA
from the ligation mixture is purified and electroporated (Dower, Miller et al. 1988) into E. cQli TGl tGibson 1984) to create the pUC based library LoxPRO.rV~CHllib. (i.e , pUCLoxPROVIDER-rV~CHllib). ~More~than 105 clones are obtained and the diversity is confirmed by sequencing ;n~PpPn~Pnt clones.
In parallel, DNA is purified from the rVLChlib.bact.
(Fig. 8) and~amplified by PCR with primers pCFWD and pCBC~. The PCR product is isolatedl cut with V~rs2 (ApaLl) and VLrs4' (Ascl) and ligated using standard methodology into fd based LoxREC (i.e., fdfDOGRECEIVER).
DNA amplified by PCR is purified using Magic PCR Prep.
The DNA is then cut with ApaLI an~ AscI and the digested DNA (about 6 ~g), is purified on a 1.5~ low melting-point agarose gel using Magic PCR Prep ~Promega). Approximately 1 ~g of the purified and cut rVLC~.lib DNA (Fig.7) is ligated to about 5 ~g of digested fdDOG-21OxVkdel (Sambrook, Fritsch et al. 1990) in a 60 ~1 volume with 1200U of T4 DNA ligase (New England Biolabs) (Fig. 11).
Ligated DNA is purified from the ligation mixture using nPrlP~n II tBiolOl), resuspended in 30 ~l water and electroporated (Dower, Miller et al. 1988) into four 5Q ~l aliquots of E. coli TGl cells grown in 1 ml 2 x TY broth - rnnt~;n;ng 1~ glucose for lh. Cells are then plated in dishes (Nunc) in TYE ~Miller, 1972) medium with 12 5 ~g/ml tetracycline (TYE-TET). After overnight ;nrnh~t;on at 37~C, colonies are scraped o~f tEe plates into 7 ml 2 x TY
broth (Miller, 1972) rnnt~;n;ng 15~ (v/v) glycerol for storage at -70~C.

W096/0~557 PCT~S95/10182 ' 1'; ' 219~7~ ~
~ 114 -The freguency of inserts is determined by PCR for each of the pools. Sequence diversity is confirmed by seguencing a clones of each pool. The pools are then ~ ;nP~ to create the rV~CL.lib-fdDOG-210x rVdLlib.
outlined above. DNA from the ligation mixture is purified and electroporated (Dower, Miller et al. 1988) into E.
sgli ~G1 (Gibson, 1984) to create the library LoxRE~rVHCHllib. (i.e., pucl9-loxrvHcHIlib) having greater than ~ x 10~ clones. Diversity i8 confirmed by seo,uencing 3~ ;n~p~n~nt clones.
Ste~ 4: In ViVD recnmh-n~tion of v~r~l an~ V~cr q~n~
In this step, summarized in Fig.14, VHCH1 and VLCL
genes-are recombined in pairs, onto single pieces of DNA
to make the rVab library. Individual member8 of the VLCL
and rVHCH1 library are placed within a single bacteria via sequential incorporation within that bacteria of the rVLCL
member via phage mediated infection_and of the rVHCH1 member via DNA-meaiated plasmid transformation. Once inside the bacteria, the two chains are combined onto the same piece of replicating DNA (fd~CARRIER'~ within the 20 bacterium by the P1 CRE recombinase, provided by P1 phage in~ection, which catalyzes recombination at loxP site in a process termed 'recombinatorial infection'(Waterhouse, Griffiths et al. 1993). The process of recombinatorial infection for expressing recl ';n~nt proteins was 25 originally described by St~rnh~rg and Hamilton (Sternberg and Hamilton 1981); and Hoess et al. (Hoess, Ziese et al~
1982; Hoess, Wierzbicki et al. 1986) which are incorporated herein by reference and depicted in Fig. 14.
In the process according to the invention, only those bacteria transformed with a rVHCHl~rVLCL combination (i.e., an rVab member) survive. Given the size of the rVHCH1 library (greater than 10~, see above) and the rVLCL
library (greater than 107, see above), this type of combination, given unlimited bacteria, could yield a rVab.lib of greater than 1ol7 members.

wos6lo4ss7 PCT~59~l0l82 ~t~6~7~
~'~ 3~ f~ ' ~

According to the invention, the diversified rVLCL lib is cloned into a tetracyclineR fd phage (lst antibiotic resistance~ rnnt~;n~ng any VH chain which is easily recog~ized and which will be replaced later in the process by rVH lib chains. The diversified rVHCHl chains are cloned into provider:ampicillin resistant p~nm;~c (2nd antibiotic resistance). The two libraries are then joined in E. coli via phage infection with fd phage rnnt~;n;ng the receiver VLCL chains (the rVLCL.lib) of bacteria previously transformed with plasmid D~A cnnt~;n;ng provider VHCH1 chains. A 1 liter~culture of these bacteria is then co-infected with fP1 which is chorampenicol resistant (3rd anti~iotic resistance) carrying the Cre recombinase. fd phage recovered from P~p~n~ colonies resistant to the antibiotics are used to infect E, coli. The percent of receptor phage with acquired rYHCH1 genes from the provider vector is expected to be greater than 5% based on the assumption that each bacteria generates 60 phage after overnight culture (Griffiths, Williams et al. 1994). It is also estimated that as long as this percent of the original triantibiotic resistant recovered cells ac~uires a rVHCH1 chain from the provider vector, the number of different phage within the rVab library will be close to the number of surviving bacteria.~5 Details of the Individual Steps for Expressing the rVLCL 1 6 and rVHCHl.L.b by CRE=LOX RE~M~TN~TORIAL
FO~ TION
Phage P1 lysates are made by thermal induction (Rosner, 1972). E. coli C600 Su- (Appleyard, 1954) rnnt~;n;ng phage PlCm cl.lO0r-m- (Yarmolinsky, Hansen et al. I98g) are grown in a 2 l baff1ed fIasks rr,nt~;n;ng 1 l of 2 X TY,:25 ~g/ml chloramphenicol, 10 mM MgS04 with vigorous shaking at 30 C to an optical density of 0.6 at 600 nffi. The temperature is then raised quickly to 42 C by shaking in a 70 C water bath. Shaking is continued for W096/04ss~ PCT~S95/10182 2~ ~667'~ 116 -another 35 min and then at 37 C untll lysis is visible.
Cultures are centrifugea to remove debris ana intact cells. Chloroform ~100 ~Ir is added to the ~upernatant and P1 phage after 30 min. 30 C infection o~ midlog E.
~Qli TG1 (Gibson, 1984) grown in 2 x TY broth with 5 mM
CaCl2. Phage infected E. ~Qli are tittered by plating E.
coli on TYE medium ~Miller, 1972) ~nnt~;n1ng 30 ~g~ml chloramphenicol. Resistant colonies are counted after 24h n~nh~t;nn at 30 C and when expressed ~~ transducing units (t.u.) are greater than 109/ml.
I0 One liter:of 2 x TY broth cnnt~ln1ng 12.5 ~g/ml tetracycline (2 x Ty-TET) is inoculated with 109 E. coli carrying the rVBC~.lib cloned in ~oxREC (i.e , fdDOG-21Ox Vkldel Griffiths, A.D.,et.al. 1994). The culture is in~llh~t~ for 12h at 30 C in two 500 ml aliquots in 2 l baffled Erlenmeyer flasks. Polyethylene glycol is added to precipitate the phage (McCafferty, Griffiths et al.
1990), which are then suspended in PBS (phosphate buffered saline: 25 mM NaX2PO4, 125 mM NaCl, p~ 7.0) and filtered through a 0.45 ~m sterile filter (M;nl~rt, Sartorius).
The resulting phage, are tittered on mid-log ~. coli TG1 (30 min, 37 C) and plated on:TYE-TET, (Griffiths,A.D., et.al.,1994) reaches -101~ t.u./ml.
The recombination process is monitored by withdrawing aliquots of the phage infected bacteria and serially diluting the bacteria onto TYE plates supplemented with l~
glucose and a variety of the three antibiotics, ampicillin (100 ~g/ml), tetracycline (15 ~g/ml) and chlnr~mrh~n;col (30 ~g/ml) and calculating the library size. The rV~CX1 library cloned into ~oxPRO (i.e., pUC19-211OxV~del in Griffiths, A.D., et al. 1994, see above) and cnnt~;n~ in about 109 E. ~Qli, is inoculated in 100 ml 2 X TY broth cont~;n~ng 100 ~g/ml ampicillin and 1~ (w/v) glucose (2 x TY:AMP:G~U). An aliquot is taken for c f u titering and the r~m~;n~r of the culture ls grown overnight at 30 C.
A second aliquot is then taken for.c f u. titering and one ;

wos6/~4ss7 PCT~S9~/10182 ~ ~f ~ f ~..' if f ~

5 ml aliquot i5 used to inoculate 500 ml of 2 x TY:AMP:G~U
s in a 21 Erl, y~r flask and the culture is grown at 37 C
to an OD of 0.5 t600 nm~. To this culture, 2 x 1012 t.u. of ~ rV~C~.lib in LoxREC is added and the culture is then divided into 5 x 100 ml aliquots.~Each aliquot is mixed with 1 l of 2 x TY:AMP:GBU, prewarmed to 37 C, and incubated at 37'C without shaking for 30 min, and then with shaking until they reach an pD600 of 0.4 (about 30 min). Aliquots are then taken for c.f.u. titering. Two hundred ml of phage PlCmcl.lOOr-m- lysate ~about 6 x 10 t.u.) are added to each flask (at an m.o.i. of about 1) after the addition of CaCl2 to obtain a final concentration of 5 mM in CaCl~. This incubation is e~ntln~lo~, with short durations of shaking every 15 min.
for lh at 30 C, followea by the centri~ugation at 5,000 x g for 15 min. The resultant pellets are suspended in 5 l 2 X TYB (the original volume) with 100 ~g/ml ampicillin (lOOA), 12.5 ~g/ml tetracycline (12.5T) and 25 ~g/ml chloramphenicol (25C) and 1~ glucose (lG). An aliquot is taken for c.f.u. titering and the library size (number of ATC resistant c.f.u.) is confirmed to be greater th~n 101~.
An aliquot is centrifuged at 12,000 x g for 5 min. the supernatant filtered through a 0.45 ~m sterile filter, and the fd phage titer is determined by infecting log phase E.
s~li TG1 (30 min. 37 C) and plating on TYE-TET.
The culture, in 5 x 1 liter aliquots, is incubated over~ight at 30 C (all culturing is with shaking unless specified) for 24h in 2 l baffled flasks Aliqout~ are taken for bacterial c.f.u and fd phage (using log phase - E. ccli TG1~ titering with the total yield of fd phage being confirmed to be greater than 10l3 t.u. The culture is centrifuged at 5,000 x g for 15 min. at 4 C and the fd phage are precipitated uslng PEP (McCa~ferty et al. 1990) and resuspended in a final volume of 10 ml PBS.
Five 2 l flasks, each with 1 l 2 x TYB, are inoculated with E. cQli TG~ and grown at 37-C until W096/04557 PCT~S95/10182 2~6~9 reaching an OD600;of 0~4' ~a~out 4 x 10l2 bacteria). About 1-2 x 10~2 t u rVab are then added to the 5 1 of E~ coli and the cultures are ;nrllh~t~ without shakIng at 37'C for 30 min. The number of E. cQli infected~with fd phage is confirmed by plating bacteria on TYE-TET plates to be greater than 10l2. Tetracycline (12.5 ~g/ml) is then added and the culture is grown for 16h at 30 C. The culture is then centrifuged at 5,000 x g for 10 min. and the pellet comprlsing the library is suspended in 250 ml 2 x TYB
r~nt~in~ng 15~ glycerol and is stored in 15 ml aliquots at -70 C.
The efficiency of replacement of the endogenous VH to be exchanged in the phagemid receiver vector ~oxREC with rVHCHl chains from the provider vector BoxPRO (i.e., pUl9-210xVHlib)~Griffiths A_D.,et.al.,1994), is flrt~rm;n~
to be less than about 20~ by analyzing 200-300 in~ividual colonies from the rVablib. Colonies are transferred onto TYE-TBT plates and grown overnight at 30 C.
Identification of color,ies possessing the recombinant VH
genes is accomplished using colony hybridization (Tomlinson et al. 1992) with a primer complementary with the CDR3 region of the exchangeable VH of the BoxREC
Between 40-50 clones lacking the endogenous VH gene (i.e., the antiTNF VH as used in fdDOG-210x Vdel by Griffiths, A.D. et al., 1994) are screened by PCR ~Gussow and Clackson, 1989) for the presence of heavy chains with the primers similar to PE~BBCK (5'GAA ATA CCT ATT GCC TAC GG) and CHl BIBSEQFWD (i.e., 5'GGT GCT CTT GGA GGA GGG TGC) and for the presence of light chains with the primers like fdBCK (5'GCG ATG GTT GTT GTC ATT GTC GGC) and C~ (or C~)BIBSBQFWD (respectively,~5'CAA CTRG CTC ATC AGA TGG CG
OR 5'GTG GCC TTG TTG GCTTGA AGC) ~Gri~fiths,A.D., et al.
1994). Both chains are expected to appear among the clones at frequency of about 20-30~.
Aliquots are then spread on TYE-TET in dishes (Nunc), and are incubated overnight at 30 C as well as being wos6/04ss7 ~ 6 7 PCT~595/10l82 tittered by serial dilution on small TYE-TET plates to e allow ~Pt~rr;n~tion of the number of colonies on the large plates. The~plates r~nf~lnlng the nccessary bacteria to generate 107 ~lones are ~rrnm~ ted, and the bacteria are scraped into 10 ml 2 x TYB c~nt~.;nlng 15~ glycerol to make stocks correspbnainy to r~ab libraries of greater than Io7 clones XII. S~ep 5 - Generating Phage and Displaying t~P ry~h,lih O~ P~ge Sllrfarp~ ~Fiq.14) As constructed above, each phagemid carries and expresses an individual member of the rVab.lib. AB shown in Fig. 14, VHCH1 protein is expressed as a fusion protein coupled in open reaainy frame to~Fhe N~2-tPrm;nn~ of the fd gpIII coat protein gene and is~there~ore displayed on the mature phage surface as an attached surface protein.
The V~CD protein, expressed via d~ Llate leader and double tP~m1n~tor codons as a soluble protein, is released into the bacterial periplasmic space wherein under reducing conditions it spont~npol~cly forms active disulfide linked dimmers with VHCH to produce the desired functional recombinant rVab on the surface of the mature phage. Phage lysates expressiny th.e entire co-mbinatorial rVab library ione rVHCH and one rVLCL gene per phage) are made with the aid of helper phage.
Phage, helper phage, ~lasmid construction, and titering are as generally descri~ed in the literature and phage and helper phage are available from commercial sources (Stratacyte CA, or Cambridge Antibodies Technologies, UK). The lysates are in general made as followsH five l of 2 x TY-TET is inoculated with a 15 (5-20) ml aliguot of the rYab phage library (greater than 2 x 10'~ c.~.u.), the cultures are grown overnight at 30 C
in baffled flasks (1 l medium/fl), centrifuged at 5,000 x g for 15 min at.4 C and the fd phage are precipitated with W096/04557 ~ ~ PCT~595/10182 21~67g PEP (McCafferty et al. l990). Phage is then resuspended in a final volume of l0 ml PBS.
These~lysates are designated rVab.lib.F and have total yields of rVab expressing nature phage of from l013 to 1014 t.u.
R~MP~E 2 Preparation of SOMERs For The Human Type l Mllrcarin1c Acetylcholize ReceotQr In this example, following Stages I and II of~the TSA
process (Fig.l), rVabs from the rVab.lib are identified, isolated and used to establish an assay for small organic molecules (SOMER) which bind to and regulate the activity of only one subtype of human m~uscarinic cholinergic receptor (huAChRm). Such SOMFRS are useful new discovery leads for such diseases as Alzheimer's and other memory Y and learning deficits. The steps outlined below constitute Stages I~ ee Fig. l) of the process of the invention and are those necessary to isolate from the rVab.lib those rVab members which bind (T+) to type l of the AChRm subtypes, regulate its activity (A+), and are specific and selective (S+) for subtype l of the human muscarinic receptor (hl1A~hRm1) Stage III of the invention, usiny these TSA+ rVabs to generate 3D models of ACHRml-specific ph~rr-r~phore5 (PEEPS, see below) and obtain SOMERs is briefly outlined at the end.
Stages I-II detail the steps necessary to obtain and use the specific AChRml rVab to establish simple rapid radioreceptor assays for small organic molecules (SOMFRs) which specifically bind and regulate ~nA~h~m1 As disclosed herein, and illustrated in~Fig. l~ and l9, these rVabs are used to aiscover active:surfaces on the h11~hRm1 which are not present on the other h1~OhRm~-s subtypes. In addition, the rYabs may be agonists or antagonists at selective huAChRm subtypes (i.e., m15) and may exhibit specificity(S+) of action between one m subtype and the other four.

W096l04557 2 1 ~ ~ ~ 7 9 PCT~Sg5110182 ~1 , .

O
Phase I of this process reconstitutes functional huAChRm which are the target of these assays. Phase II
first identifies the rVabs contained within the rVab.lib which bind to hn~rhRml (i.e., are Ti), and are selective among the five huACh~m subtypes (Andre, Marullo et al.
l987~ ag well as specific for huAChRm over non-cholinergic neurotransmitter receptors In this example these two attributes are referred to~ether as S+. Subsequently, Phase II identifies and isoIates the subpopuIation of TS+
huAChRm rVab which regulate the activity of the hll~rhRm1(A+) with similar TS+ attributes. The rVabs with all these attributes are referred to as TSA+ rVabs. Phase III conver~s~ e TSA+ rVabs to reporters (i.e., rVab.reporters) and est~hl;~ho~ validated automated rapid receptor binding screens for small organic molecules (SOMERS~ which competitively displace active rVab reporters ~rom active surfaces on ~ rhRm1 Among these SOMERS are those having the desired acti~ity profile of a pharmaceutical discovery lead, i.e., selective specific regulation of AChR~l ~~ - ph~Re I-~ : Obt~;r~ng~rhRm Cortical mPmhr~nP~ enriched in huAChmR are prepared from brains (fresh or frozen, human, porcine or bovine) as outlined by Haga O'C Haga (Haga and Haga l983). Membranes are prepared by homogenization In standard fashion (i.e., with protease inhibitors) and AChRm is solubilized by treatment with l~ digitonin, 0.l~ NaCholate in ~0 mM
~aCl/buffer. The soluble receptor is purified over an 3-(2~-amino benzhydryloxy)tropane-~ABT) affinity colum~
- and is eluted from the A~T column by atropine. Soluble receptor is subse~uently applied onto a hydroxyapatite column to remove the free atropine. The receptor is then eluted with high potassium phosphate and 0.l~ digitonin a~d is further purified through a second round of A~T
purification as noted above. Two rounds of HP~C
purification over tandem linked TSK4000SW and TSK3000SW

W096~04~7 PCT~S9~110182 '. ' I

21~6679 columns provides the final=purification and the receptor is su8pended in 0.I M potassium phosphate with 0.1~ -digitonin. = ~ ==
As a secondary source, the five huAChRmI-5, expressed RS recombinant proteins (rhn~h~ml-5) in Sf9 cells cont~nlng an expression vector baculovirus construct carrying one of the huAChRm as originally described by Vasudeva (Vasudevan, R~ n~r et al. 1991) are obtained from commercial sources (BioSignal, Inc., Montreal, Canada). Other alternative sources o~ huAChRm are various tissue culture cell lines transfected and expressing cloned huAChRm =(~ubo, Fukuda et al. 1986;
Shapiro, Scherer et al. 1988; Buckley, Bonner et al. 1989;
Buckley, Hulme et al. 1990; Tietje, Goldman et al. 1990;
van Koppen and Nathanson 1990; K~h1hAra, Varga et al.
1992; Beth 1993; Bazareno, Farries et al. 1993; van ~oppen, and ~enz et al. 1993).
ph~ce I-B Obt~;nin~ thP-G prot~1nc (GP) Go, Gi and Gn (referred to as G protein [GP] in text and G in Figures) are purified as described (Sternweis, 1984; Haga, 1986, and Xaga, Uchiyama, et.al.,1989). Brains (150g), porcine, bovine or human (obtained from commercial or non-profit sources) are homogenized, the membranes pelleted and then solubilized with 1~ NaCholate in~2D m~M
TrisXCl (pX 8.0) 1 mM EDTA/ 1 mM DTT (l~Cho-TED) with 0.1 mM b~n7~m1~1n~ (2~ vol.). After centrifugation, the supernatant is applied to DAE Sephacel and the fractions binding [35S] GTPS are eluted with linear NaCl, in l~Cho-TED, concentrated, and app:lied and elu~ted from Ultrogel AcA 34 in 0.lM NaCl in Cho-TED. The fractions with [35S] GTPS binding activity are pDoled with TED + 0.lM
NaCl (450ml) and applied to heptylamine-Sepharose, washed and finally are eluted with a linear gradient of 0.25 NaCho-TED + 0.2M NaCl vs. 1.3~ NaCho-T~D + 0.05M NaCl.
This material (a mixture of Gi and Go) is applied to DEAE-Toyopearl, prewashed with TED + ~.6~ ~ubrol PX

wo 96104s~7 2 1 g 6 ~ 7 9 PCTlUSg~/10182 O 1 ~
(0.6~BPX-TED~ and eluted wlth a linear gradient of NaCl in 0.6~PX-TED. The Gi fractions elute first, then the Go fractions. Each is collected separately and is stored at -80 C until use. Before use, the Bubrol is changed to 0.8~ NaCholate, in TED+0.5M ~ phosphate buffer S pH7,0.1MNaCl) on a small column of hydroxyapatite.
Ph~e I-~ : Reconsti~ution of an active A~h~m:~p com~lex Reco~stitution is accomplished as per Florio and Sternweis (Florio, 1985). Porcine [or human brain total lipids: as per Folch, J., ~ees, M., and Stanley, G.H.S.
(Folch/ Bees et al. 1957). The lipid mixture is prepared (Xaga, 1986 ) from brain extract (Folch fraction I) (1.5 mg each) and total lipids ~1.5mg each) suspended in 1 ml HEN (20 mM Hepe~s-KOH buffer pH 8.0, 1 mM EDTA and 160 mM
NaCl) wlth 0.18~ deoxycholate and 0_04~ sodium cholate.
rhn~h~m (0.2-0.4 nmol/ml [3H]QNB binding sites in PD (.5M
potassium phosphate buffer pH 7.0 and 0.1% digitonin (10-40~1j) are mixed with 0.1 mM oxotremorine in XEN, and then with 100 ~l of lipid mixture (final vol. 200 ~l) to give QNB:R complex. The complex is then run through a Sephadex G50 column and the void volume (1-8 pmol [3H]QNB
binding sites:, 400 ~l) is collected. The huAChRm:QNB
complex is mixed with G protein (mlxtures or separate G-proteins, 0-200 pmol of [35S]GTPgS binding sites in 40 ~l cholate solution) CN-TED and HEN (50 ~1) containing MgCl2 and DTT (final cnnr~rtr~tion 10 and 5 ~M respectively) and 1nr1~hat~ at 0 C for 1 hr. This hui~h~ml:GP mixture is diluted before use with 3-5 vol of HEN.
Phase I-D : Attachment of ac~ive huAChRm to matrices (Fiq. 19~
huAChRm (abbreviated AR in text and R or T in Figures), alone or complexed with GP, is attached to a Sepharose (or agarose)-type matrix by taking 5 ml of matrix (WGA-Sepharose, mmolWGA/ml Sepharose, 50~ v/v, prewashed and suspended in ~uffer A (25 mM Potassium phosphate buffer,[pH7.0],0.8 mM EDTA, 10 mM MgCl2, 230 mM

~ 79 ~ - 124 -NaCl, 0.06~ BSA, and 4mM HEPES KOH buffer [pH 8.0]) and mixing it with less than 1 ml reconstituted AR:GP
complexes ~100 pmol AR/ml). The mixture is then ;nrnh~t~
at room temp ~r.t.) for 30 min, diluted with buffer A to 20 ml and the Sepharose is allowed to settle (or centrifuge at low speed [5,000 x g, 1-2 minl). The Sepharose is then resuspended in 20 ml buffer~A ana the washes are repeated twice to provide purified AR
complexed-Sepharose WGA [sWGA:ARGP] material.
Recombinantly derived or native AR:GP complexe-s with i0 appropriate sugar residues bound to WGA in this process remain active as matrix-attached receptor in agreement with published data showing glycosylation is not required for AChRm activity (Habecker, Tietje et al. 1993).
Quantitation of bound AR:GP to sWGA is ~verified by [3H]QNB
+ 10 ~M atropine and [3~SIGPTS or [3H]GppNHp + 0.1mM GTPS
or GppNHp binding using standard binding assays (Berrie, Birdsall et al. 1985; Haga, Haga et al. 1986; Wheatley, Hulme et al. 1986; Poyner, Birdsall et al. 1989).
In parallel reactions,-~AChRm (or GP), natural or recombinantly expressed preparations, are attached by standard techniques to plastic, directly or~secondarily, through matrix attached antibodies, naturally derived or rVab-type, which recognize epitopes on the receptor, glycoprotein, G-protein or small peptide tags (i.e., the c-myc and other amino or carboxy terminal in frame tagging peptides, available in various spaced commercial -expression vectors). After att~rhmrnt of AR, the unoccupied reactive matrix surfaces are~blocked by application of~various standard blocking agents (i.e., BSA, milk etc.).
ph~Ce II p~nn;nr for TSA+rV~h In this stage, rVabs which possesses TSA+ attributes are identified as those which bind to AChR directly or indirectly attached to the matrix~ with or without G, in buffer conditions similar to those used for AChRm 21~679 w09~104ss7 ~ ~. PcT~sgs/1~182 - 125 ~
radioreceptor binding studies. These conaitions m~int~ln receptor activity. In all cases plastic and not glass is used for direct attachment matrix surfaces and reaction vesicles to minimize rVab nonspecific absorption to glass.
The buffer for these reactions is a lO m,M potassium S phosphate (pX 7.0), 0.8 mM EDTA, lO mM MgCl2, 0.230 m,M
NaCl, 0.'06~ ;3SA, 4 m,M Xepes-ROH (pX 3.0) buffer, and optionally further comprising guanine nucleotide (GTP) and/or muscarinic agonist or ant~agonist as detailed below.
This stage isolates four types~of A+ TSA+ rVab,antibodies:
agonist like~Ago+~, partial agonist-like (partAgo+l, allosterically agonist (Alloago+) and antagonist-like (Antago+) (outlined in Fig l9).
Ph~e II-P- p~nninq for receptor ~Tarqet (T+)l recoq~ition The general proce8s is summarized in Fig.16 and the specific application in Fig. l9: Five ml of the rVab.lib (lOIl12PFU/5 ml, and suspended in buffer) is mixed with l.O
ml settled s-WGA:GAR in buffer'A, and ;nrnh~tP~ at 30 C
with gentle shaking for 60 min. The mixture is then centrifuged at low speed (~SS) of 500 x g for 15 min. The supernatant is decanted and diluted with buffer A to lO
ml. These washes are repeated 3 times rapidly and the rVab in the ~inal pellet resuspended in buffer A and designated as the T+rVab.lib. (Fig. l9). Phage are released by elution with lOOm,M triethylamine (Marks, Hoogenboom et al.
l99l). Alir~uots are withdrawn and tittered for phage.
The population of,isolated phage are then amplified by infection and induction of new lysates and panned again 2-4 more times to generate the final T+rVab population of ~ phage for subse~u=ent isolation of the four types of A+
rVabs.
Phase II~ p~nn~n~ for Active r~hs ~+rV~h)(ph~qe TT~) In this process (general outline=in Fig. 17, specific application in Fig. l9), the subset of rVab from the amplified T+rVab population which are potPnti~11y agonistic are=induced by the addition of guanine W096/04557~ PCT~S9~/10182 ~1 ~667'1 ~

- 12~ -nucleotides to dissoc~at'e~ from the matrix attached R:G
complex and be isolated as free TA+rVab in:the supernatant. In this proce~s, the rVab which bind and act as antagonists, or bind to nonactive surfaces, remain matrix-receptor associated after the addition of guanine nucleotide. The negative influence of GTE on:T+rVab binding is taken as indicative of potentlal agonist action of the bound rVab based on the observation that in functionally coupled AR:GP complexes there is a negative reciprocal interaction between the binding of GTP or GDP
l~ to the G protein, and agonist to the receptor, which can be observed aS an immediate~dissociation of either from the complex (Smith, Perry et al. 1987; Poyner, Birdsall et al. 1989; Bazareno, Farries~et al. 1993). No such reciprocal interactions occur between antagonist and guanine nucleotide binding (Buckley, Bonner et al. 1989).

The TA+ rVab released into the supernatant are further separated and isolated as one of three types of agonists in separate panning steps (see below Phase IIB-i,ii,iii). The specific muscarinic activity of the rVab i9 confirmed at the end of all isolations using AChRml activity assays in which potential TSA+rVabs (a) compete with radiolabelled antagoni~t (or agonist), (b) dissoclate prebound [35S] GPTS or [3~] GppNHp from matrix bound ~ChRm:GP complexes, ~c) stimulate GTPase and or GTP
exchange, and d) regulate the activity of other effector systems coupled to the AChRml (i.e., adenylate cyclase, phospholipase, K rh~nn~l r) in various published ~n Vit~Q, cellular or animal assay systems (Yatani, Mattera et al.
1988; ~raser, Wang et al i989; Shapiro and Nathanson 1989; Kobayashi~ .Ch;h~k; et al. 1990; van Koppen and Nathanson lg90; Weiss, Bonner et al.~l990; Yatani, Okabe et al. 1990).
In Phase IIB7 addition to bound T+rVab of ACh itself can also be used, via the same type of induction of rVab w~s6/04ss7 ~ ~ 6 ~ 7 ~ PCT~S95/l0l82 dlssociation from AChR, to isolate those rVab which bind not to the ACh binding pocket but to GP at active nucleotide binding surfaces or to other surfaces on AR or GP which are active and allosterically connected with the rhrlin~rgic binding surfaces of the AChRml Specifically, at the start of Phase IIB, the amplified T+rVab.lib isolated in Phase IIA is mixed with matrix-bound AChRml:G compl,ex, in 10 volumes buffer A as noted above for 30 mi~=at 37 C The pellet is centrifuged at low speed, resuspend in 10 vol cold buffer A and ~ tely recentrifuged The washed pellet is resuspended in lO vol cold buffer A r~nt~in~ng 100 uM GTP
After less than or equal to about~l min the matrix:AR:GP
complex is centrifuged at low speed, and the supernatant i9 separated from the pellet to be used to isolate three different types of agonistic rVab in Phase IIS-i,ii-iii.
The pellets are washed in similar fashion with buffer A
three (3~ tines and analyzed in phase II~3-iv for muscarinic antagonist (Antago+) activity as detailed below. Throughout these phases, aliquots of supernatant are taken to titer the phage, and lf less than lO~ml, the phage are amplified and recycled as above 2-3 additional times To the final Pnr~rn~t~ntl cnnt~inlng rVab induced to dissoclated via GTP addition, GTPase and GDPase are added and the supernatant lnrnh~t~ 30 min at 30 C. The solution is then chilled and passed over a Sephadex G50 fine column using buffer A and the_void volume, free of any remaining-nucleotides, is taken and labeled TA+~G~+)rVab lib nt;ficatio~ of ~nta~nlst ~rtivity The T+rVab lib, for which binding is not modified by addition of GTP, and which is recovered bound to matrix in the presence of GTP, i8 released from matrix a~d the phage harvested by PEP precipitation in Phase II~-iv. The phage are~then resuspended and mixed with s-WGA:RG in 2ml buffer A rnnt~;ninJ saturating amounts of antagonist .. . .
.

W096/04557 , ., PCT~S9S/10182 21~6~7~
-~128 -(atropine l0 ~M, perenzepine l ~M, scopolamine, l ~M).
Following incubation for 60' at 30 C phage a~nd s-WGA:RG
are ~ntri fnged at low speed and the supernatant is collected. The free phage are isolated, and amplified (as noted above) aud the popuIation recycled an additional 2 S to 3 times by combining with s-WGA:RG to remove from the supernatant phage which, in the presence o~:artagonist, aO
not bind to s-WGA:AR. The phage in the final supernatant contain the expressed A+rVab me-mbers which are muscarinic antagonist-like (Antago+) are deslgnated at the end of Phase II.3-iv, as TAntago+ hnA~hRm1 rVab.lib. [see-Fig.
l9, rVab-4].
The pellet from incllhist;on with muscarin~c antagonlst in the above Phase IIB-iv ~nntA1nc a T+rVab sublibrary which has members which interact directly with surfaces on the G protein~of the AR:GP complex and are guanine nucleotide like regulators of the AChRml:G complex. Phage are freed from the matrix, amplified and incubated with matrix bound G-protein in b~f~er A. :The matrix, and attached rVab, are then oentri~uged, washed and attached phage isolated_ Confirmation of G-like activity among these isolated rVabs is done in standard radioreceptor binding assays estAh~;Ah;ng competition with radiolabelled GppNHp or GTP~S for binding to GP.
PhA~e IT ~i.ii,iii: Separating GTP Sensitive A+rVab into Aqo+(c~+ .~hive) An~ i~ 1 loAqo+(~ +Cch-i~e~ ve) A~hRml - rVAh . (Fiq.l9) One to l0 ml of the TAffG~+)rVab.lib is mixed with 1 ml sWGA-GR, incubated 60' min, 30 C in buffer A with 300 ~M stable muscarinic agonist carbachol (CCh) and is then centri+7uged at low speed. ~In Phase IIt3-~iii, the pellet is wa~hed with buffer A three (3) times, and resuspended in buffer A and the phage isolated in standard fashion This phage population, labelled TA+~G~+CCH~~rVab.lib, contains the allosterically acting muscarinic like agonist (alloAgo+) rVab members (Fig. I~, rVab-3).

w096104557 21 9 6 B 7 n pCT~S9~110182 The supernatant from the above Phase IIB inmlh~t;nn with CCh is passed over Sephadex G50 (fine~ in Phase IIB-i,ii and ~he phage are~coIlected i~ the void volume o~
the column (as outlined above ) to obtain CCh free rVab which are blocked from binding to AChRm by CCh. These S phage are labeled as the TA+(~P+C~+~rVab lib and contain the Phase IIB-i and ii rVab.lib members which are competitive-ACh muscarinic full(i) or partial(ii) agonist-like (Ago+) antibodies (i.e., rVab-l and 2 in Fig l9).
Phase II-C : Separating Selective(S+) from non Selective(S-) T~+ rV~h~
All four types of AChRm A+T+rVab phage isolated in Phase II~ (labelled rVab-1,2,3 & 4 Fig. l9), are taken separately, and mixed with l ml sWGA:GR ml in buffer A
cnnt~ln1ng soluble complexes of GP and AChR of subtypes 2-5 (i.e., G:AChRm2-5 complexe~). These complexes are added a8 the competing target peptide (analogous to comp-T-pep in Fig. 16) which contain greater than lO ~old exce8s of surface epitopes which are not to be recognized by the ml specific A+rVabs, incubated 30'C, 60 min and then centrifuged at low speed( Fig. l9). The pellets contain the S+rVabs.lib members and these are resuspended in lO vol buffer A and washed ; ~ tely~ The phage are recovered in 8tandard ~ashion, amplified and cycled through Phase IIC two ~o four additional times. Frozen stock bacterial cultures and phage lysates are prepared for each of the four A+ types of AChRml specific(S+) and are designated TS(Ago; partAgo; alloAgo; or antAgo)+rVab.lib. In an alternative embodiment, isolation of the AChRml specific rVab library is done on the T+rVab.lib before selecting for the A+rVab.lib (Fig. 16) and the population is amplified for subsequent A+
selection as defined above.

, W096/04557 PCT~S95110182 219~n ',, Stage II-E: Confirmation of A+ activity among i~divi~n~l mPmher5 o~ thP T~A+ rv~h A~hml l;h Individual members (10-20) of each of the four A+
type TSA+ rVab ACh~ml library identified above are obtained and phage lysates are gene~ated for each by standard technology. The A+ profile for individual phage members of each of the above four A+ library is cnnf;
and quantitated by a nM ED5D value in one or more-Df the following standard radioreceptor and receptor-coupled activity assays. The radioreceptor assays use 1) active soluble targets (i.e., AChRm, AchRm:G and G-protein complexes); 2) radiolabelIed AChRm [3H]agonist or antagonist, or [3H, or 32P]GTP, or GMPPNP or [35S]GTPS in buffers used for rVab isolation; and 3) various dilutions of individual rVab members to be tested. The reaction mixture cnntpntq are incubated at 30~C for 30 min and the targets are recovered free of soluble radioligand by standard filtration or PEG precipltation. The reduction in specifically bound radiolabel is then ~uantitated. ~ ~
The degree of agonist activity forsAgo+~ partAgo+ and alloAgo+ rVab members is demonstrated by dose respo~se alteration of any one of a number of AChRml coupled~
effector systems. Individual antagonism ~Antago+) is demonstrated by dose response blockage of the ACh agonist 25 effect on the particular receptor coupled system.
Phase III. Convergion o~ Selected A+rV~h to rV~h Reporterg A Preparation of Reporters and Competitive ~in~;nq Aq~ys to Ident,ify SOMRR~ (Fiq.la,l9) DNA is isolated from phage lysates prepared from bacteria grown from two to ~ive individual TSA+rVab.bact stocks from each of the four classes of A+ libraries characterized above to have A+ activities with ED~0 values of 1-30 nM. The DNA is digested with Apa~1 and Notl to release from the fd~Carrier the rV~CL-rVHC~l rVab w096io4ss7 21~ ~ ~ 7 3 PCT~595/l0l82 O
construct. One ~g of the insert is isolated and mixed with 5 ~g DNA from pEXPRESSORrVab (pEXPRESSORrVab-1, see Fig. 9), precut with ApaL1 and Notlv and 1200 U T4 ligase - (Sambrook, Fritsch et al. 1990). The ligated products are purified and electroporated into ~. CQli (Dower, Miller et al. 198~). Transformants are grown and characterized ~y diagnostic~PC~R and_then se~uenced. Correct constructs of each are then grown, the recombinant rVab (i.e., VHCHl:V~CL dimmer chains) induced and the rVab products are recovered in the supernatant by precipitation with Sepharose coupled VH or VL chain antibodies or antibodie~
to peptide se~uences~(ISOTAGS) in~luded in pEXPRE~SORrVab-I (Fig. 9C) a~d fused in frame to~the carboxyterminus of CH1. The rVab are then released from the precipitating antibody. The VHCH1 chain of the rVab is then phosphorylated in a constant region C term;n~l domain attached in ~rame (Li, et al. 19~9) when rVab is ligated to pEXPRESSrVab. The phosphorylation reaction uses protein kirase and [32p] ATP following published methodology and the radiolabelled product i9 isolated in the vold volume of a G50 column. The radiolabelled rVab is mixed with '3SA and stored at -4~C until use To establish a 5~nr~t;on isotherm and ED50 for the labelled rVab with its active target (soluble or membrane bound; GP, ~ChrRml, or AChrRml:GP complexes), the binding of rVab is determined from reactio~ mixtures (50 ~1) comprising from 1000-1,000,000 cpm of radiolabelled rVab with and without 1000 folded excess of unlabelled rVab in bu~fer ~ T~ent; ~1 cortrol assays are done with O AChRm2-5, AChRnicotinic, or other-non- chnl; n~rgic G-protein linked neurotransmitter receptors (e.g., ! beta-and alpha adrenergic, and opiate receptor). These assays are incubated for 30 min at 30~C. The [32p] rVab_target complex is PEG precipitated ~or ~iltered with -- n~ bound target) and co~nted for radioactivity.
The induced dissociation of rVab from its target by , , ......... - . . -W096/04557 , PCT~S95/10182 2~9~
- 132 -~
an allosteric effector li.e., the Ago+rVabs with GTP) defines~the class of allosteric rVab agonists. A series of competition binding assays-is the~ performed using lese than, or e~ual to, the ED50 amount of [32p] rVab with increasing cor,centrations of the nonlabelled form of the same rVab, other rVab, standard muscarinic specific ligands (agonists and antagonists), and a number of noncholinergic ligands as controls to further characterize these rVabs.
These assays establish a saturation binding isotherm, an apparent Kd for rVab and target association, and IC50 values for various ligands and other=rVabs. The reactions carried out in the presence of increasing co~centrations of other members of the same TSA+ rVab group define the rVab with the lowest IC50 value. This rVab is then converted to a radiolabelled form for use in obtaining saturation isotherms and various competition curves. In addition to the radiolabelled rVab, these assays further may contain l) target agonist; 2) antagonist; 3) GTP;
and 4) combinations of all three. Standards such as nicotine, muscarine, ATP, GMP, and the various small organic molecules previously reported in the literature to have affinity for regulation of AChRm receptor of t~e ml-5 type regardless of affinity or selectivity may also be included. Saturation isotherms are yenerally conducted over a conc~ntr~ti~n ranye of four to six orders of magnitude.
rVab's with affinity for AChRml of less than about l0 nM, selectivity for AchRml over AchR types m2-5 of ~l00 fold, and speclficity regarding non-cholinergic soluble receptors of l000 fold are appropriate as rVab-REPORTERs for A+ activity for use in ~5tages II~and III of this invention wherein SOMERs are identified in CHEMFIEES or synthesized based on BEEP models (see below).

W096l04ss7 21 g 6 6 7 9 PCT~595/10182 0 133 '~
ph~e~ IV-VI
In the last three phases of the invention, which are part of TSA Stage III , the TSA+ rVabs are grouped ~ according to common epitopes and attributes (Phase IV), 3D-models of~active pharmacophores ~BEEPS) are derived (Phase V) and the pharmacophores used to find SOMBRs in existing CHBMFILES or by synthesis (Phase VI). The grouping of TSA+rhuAChRml in Phase IV is accomplished according to a) the common surfaces recognized by the rVab (defined by competition by peptide fragments of the AChR; b) the type of activity exhibited by the rVab (partial or full agonist, antagonist, competitive or allosteric with ACh or GTP) and; cr the diversified amino acids of the V regions found in the rYab.
The Stage III analysis of the TSA+rVabs which creates a 3D model ph~rr-cophore (Fig.=23-25) i8 performed based on a genetic algorithm directed comparison of the array and positions of the amino acids in_the V regions of the active rVab's, including CSR, CDR~and fL~~ Lh residues.
The 3D atomic model formulated by t~is process is designated a nbiologically ~nh~nc~d ensembled pharmacophore" (BEEP). The BEEP cnnt~;n~ suffici~ent infnrmiqt;nn to describe the ~l~m~nt~ of a SOMBR necessary for the activity profile of the active rVabs within that particular group.
In Phase VI, the BEBP is used in a variety of available programs (HOOK, BOOK, and DOCK) for computational screening (Phase VIa) of available CHBMFIBES
for hn~rhRml SOMBRs and, in a rational drug design effort, to direct the actual synthesize of hn~h~ml SOMBRs (Phase VIb). 50MBRs obtained by either approach are then cnn~;rm~1 as TSA+ AChRml agonists~or antagonists in n Yi~Q, cellular and animaI as8ays, known to those versed in nhn~;nnm;m~tics.
Additional diversification of TSA+ rVabs within CSRs and CDRH3 is carried out by PCR (as detailed in the : .,, ~ i.

W096l04557 PCT~S95/10182 ~ 134 -construction of the original rVab.lib) in Phase IVb whenever the number of rVab within a group is less:than 10 or when sufficient information is not obtainable from the number of A+ rVab's identified to develop PEEPS with the desired usefulness for idertifying SOMERs and simplification:of the TSA+ population is done when the number of rVab within a group is >lQ0 (Fig. 15) EXAMPD~ 3 - -This example outlines the TSA process establishingsimple competitive binding assays for multimeric small organic molecules, which in this example are DISOMERs, capable of regulating the activity of growth hormone receptor. ~ere, DISOMER discovery is based on the discovery of pairs of rVab which identify active surfaces on Growth ~ormone Receptor and their conversion to rVab.REPORTERs according to the method of the invertion. ~:
This methodology establishes a generic approach for discovery of drugs active at oligomeric receptor targets, or targets requiring activation at multiple sites of a monomeric unit. In 9uch systems the "receptor" is defined by multiple surfaces which must be iu contact with the signal to cause activation. - ~
The process of this invention provldes a means of identifying active ligands Lor multiple site receptors a) which have more than one active surface; b) more than one subunit per active receptor complex; or c) different subunits and active surfaces. This method is also suitable where more than one subunit contains a portion of an active surface, the surface reguired for activation is too large to be occupied=by a single small organic molecule present within a ~ MFILE; and activation of oligomeric rec-eptors is intimately associated with the hormone lnduced formation ~of complexes of at least two receptor subunits (~nnn~ngh~m, 1991; Kelly, 1991; DeVos, 1992; and Wells, 1993).

219~679 ~096l04ss7 PCT~S95/lOlX2 f) . ; ., f ~ ~
- 13~ -U~like standard screens to identify a single chemical entity to replace a large multi-site binding hormone, the approach described according to this invention, identifies pairs of active 8urfaces, finds SOMERs for each individual active~surface, and then links the SOMERS together to 5 create multlmeric units (e.g.,DISOMER) large enough to replace the multivalent hormone, c.g., growth hormone (GH). In the example provided, the target oligomeric receptor is the homo-~iimeric growth hormone receptor (GHR) and the active surfaces identified are the two surfaces l~ u8ed by GH for active GHR dimerization. For GHR there i8 only one type of receptor subunit, referred to here as T1.
Activation of the receptor requires GH to dimerize two receptor subunits (T12) by maintaininy binding of active surfaces on two T1.
1. Identification ana Isolation of rVabs Specific fDr ~R
Step la: Identification of GHRT+rVab.lib for t~T1 GFfR Subl~nits Isolate from the rVab.lib the subpopulation which binds to the surfaces of the T1 GHR subunit. These rVabs are designated GHR.T+rVab.lib.
Library surface scanners arel~provided by the rVab.lib constructed as outlined in Example 1 of this invention.
This rVab.lib, i.e., rVHCH:VLCL complexes, is expressed on phage surfaces attached to the phage gpIII coat pro~ein.
A one ml aliquot of phage lysate ~10~t.u.) is mixed with GHR receptor~subunits ~Tl) which are prebound to an immolized solid support i.e., agarose beaa-type isolation matrix=(mat-T1~. In this example, the basic GHR subunit (T1) used is that which ~n~mr~cses only the ~ llnl~r domain of the hGHR, including hGHR amino acids 1 to 238 (~eung, 1987; Fuh, 1990) with an unpaired penultimate cysteine (!3ass, Greene et al. 1990). This form is referred to as sGHR and is expressed in E. coli as an W096/04557 ~ PCT~S95/10182 .

21g~79 ~ 136 -extrArPllnlArly released soIuble protein ~Fuh, Mulkerrin et al 1~90). This soluble protein IS then purifiea (Fuh, Mulkerrin et al. 1990) and bound to ~eaas or:plastlc through its unpaired cysteine (Bass, Greene et al. 1990), or to plastic through an antibody which recognizes the sGHR but does not interfere wlth GH binding or active GHR
dimerlzation (Fuh, Mulkerrin et al. 1990; Cunningham, Ultsch et al. 1991). All forms of sGHR~bind GH as does the endogenous membrane~assoclated entact GHR (~eung, 1987; Fuh, 1990). An excess of soluble prolactin receptor (PRLR) as competing peptide (comp-T-peptide) (see=Fig. 16 or various mutant hGHR, or PRBR mlssing either H binding~
site I or II (~llnn1nr;~Am, 1991; DeVos, 1992;~and Rozakis-Adcock, 1992) to compete binding of non-specific rVab binders which have no selectivity for GHR binding is routinely added to the mixture to define rVab specificity.
With sGHR attached to 0.2 mg of oxivane polyacrylamide beads (Sigma) the reaction mixtures can be as small as 50 ul beads. The excess of soluble prolactin receptor competes for binding of non-specific rVab binders which have no selectivity for GHR binding. The mixture is incubated for at least 3 hr=at 30~C in buffer A which supports normal GHg and GHR association with one entity displayed as an attached phage coat protein (Bass, Greene et al. 1990) and consists of c50 mM Tris, pH 7.4, 1 mM
EDTA 50 mM NaCl, 1 mg/ml BSA and 0.02~ Tween 20 and washed three (3) times in 30C buffer A The rVab bound to the matrix associated GHR, in the presenc~ of the excess competing soluble non-GHR related peptide (i.e., the comp-T-pep) is deslgnated the GHRTS+ rVab.lib. The phage are recovered by washing (2x~ either in Buffer A with 20 nM hGH or 0.2M glycine (pH2.1) (Bass, Greene et al. 1990) and tittered.
The phage libraries are mixed with E. coli (at a multiplicity of infection) of approximately one (1), ~nrnhAt~ without shaking fo~r 30 min and then plated in w096l04ss7 2 1 9 6 6 7 9 PCT~Sg5/10182 antibiotic media and grown overnight and tittered. The survivors are pooled and yrown Qvernight and frozen as bacteria~ stocks, in 15~ glycerol. An aliquot of the - stock i5 grown up and new phage lysates are made and tittered. This phage population,=GH~.TS+rVab recognize~
all surface~ on the T1 subunit of GHR. Definition of S+
in this populatiQn at this time is not mandatory, and can be omitted, i.e., by not adding prolactin receptor (or any other comp-T-pep) to the original reaction mixture above, if the number of GHR.TS+rVab members obtained in Step 1 which are competed by GH (see below) is less than 100.
An additional phase of V region amino acid diver~ification within CSRs and/or CDRH3, as per outlined in the Example 1 and summarized in Fig. 15, is performed if greater numbers of GXR.T+ or TS+rVab are desired.
Step lb. Subdivision of TS+rVab based ~n ~.RR 5l~rfac-e epi~ope recognized lb) Group library members according to common receptor surfaces recognized. Designate groups as GXRlx-y).T+rVab.lib, where x-y is the amino acid domain of the Tl unit nnnt~n~ng the common group epitope ~Fig. 16).
S~p~r~t;~n according to the receptor surface recognlzed is accomplished by adding aliquots of TS+rVab to plastic dishes to which have been preabsorbed peptides ~obtained commercially) of 10-20 amino acid overlapping amino acid seq~ences of GHR and those domains cnnt~ining amino acid sequences known to influence GH binding li.e., hGXR amino acids 54-68, 171-185, 9 [GHR siteI]); and 116-119 and 8-14~GHRsiteII) as described ~cnnn;ngh~m~
Henner et al. 1990). TS+rVab are ir~nhsto~ with preadsorbed peptides in buffer A ~20 mM TrisXCl buffer pH
7.5, 1 mM EDTA, 0.1 ~ bovine serum albumen) for 3 hr at 30~C. The dishes are washed to remove unbound rVabs.
~30und rVabs are released from the matrix, tittered and amplified again via lnfection in coli. Binding to -- , . - .
-~ .. .. i ~ . . .: , .

W096/04ss7 , PCT~595/10l82

6 ~

these overlapping GHR peptides prod~ces a grouping according to primary receptor amino~acid sequence a~d Y
hormone binding Each of the separate groups are then mixed with soluble matrix-GHR_ ~see s~ep=la) in buffer A
with greater than 100 fold excess GH and incubated 3 hr at 30~C and centrifuged. The phage in the supernatant are tittered, amplified and further enric~ed by panning 2-3 additional times for TS+rVabs which aO not bind to GHR in the presence o~ GH_ This recycling produces a pop~lation of GHR.TS+rVab which bind to a surface of the GHR which is normally occupied by bound GH. Although these steps do not identify and/or subdivide all GHR hormone related epitopes, they divide the original GHRTS+rVab.lib into workable sized subgroups based on binding to various amino acid sequences and domains involved ln GHR recognition.
Each group i8 tittered, amplified, in~ected into _. coli and bacterial stocks and subse~uent new phage lysates are prepared Each group is designated by its amino acid receptor sequence or domain recognized (e.g.' amino acid x-y) as follows: GHR.T(x-y)S+rVab.lib. Competition by these rVab for~I125hGH binding to sGHR is done in standard binding assays (Spencer, ~r ~ et al. 1988) in buffer A
with terminated by precipitation by polyethylene glycol 8000, at 4~C in phosphate buffered saline as described (Leung, 1987) Competition binding to membrane associated GHR is per~ormed under identical conditions and reactions are terminated by filtration and washing.

2. Formation and Identi~ication of Bifunctional Active rVabs Possessing ,R~n~m Se~l~nce~ of ~m~n~ Acid Step 2a: Prep~ration And E~sression Qf r~sh-peo rl;~r~ry 2a) Attach a random 8 amino acld peptide library (Pep8) in frame to the light chain (VLCL) of all members of a rVab library recognizing a coLmmon GHR surface=~Fig.

wos6l04ss7 ~CT~S95110182 ~, 2~9~7~,.,.,.,.~,, ~ , o 11 and 20). Designate these bifunctional surface binder librarie B G~R(xy).T+rVab-pep.lib.
Each of t~e group Iibraries is genetically engineered - to be expressed, in a coupled manner, with a short random peptide of-8 amino acids (pep 8) attached through a short linker (LNKR) to one chain of the rVab (Fig. 11).
Att~rhm~nt can be at different positions on different chains s~r~n~;ng upon which Cre-Lox recombination system is used to combine the rVXCXl.lib ana rV~CL.lib onto the same piece of DNA when the rVab.lib is made (see Fig. 11 vs. 13). In~this example, the rVab.lib is made according to Example 1 (Fig. 11) and attachment of the pep8 is to the amino terminus of the V~ region of the rVLCL.lib (Fig.
11 and 20). In Example 4 below, the construction of a dif~erent rVab.lib where addition to a single pep8 could be made to either the carboxyterminus of the constant domain (CL) of the rV~CL or to the aminoterminus domain of the VX of the rVXCX1 is described ialso see Fig. 13).
In this example, att~rhm~nt is accomplished by using PCR to append the pep8 library to the 5' end of the VL
region within the rY~Ch mem~ers of the GXR.TS+rVab.lib.
This reaction uses forward primer CH2Q9-216-NotlFWD and back primer APAPEP8LNK~3CK (i.e., leader seq.Apal-(NNN)8(GGGGS)~VL1-7) (see Primer Table, Fig. 10).
These reactions contain an aliquot of bacteria from each GHRT(x-y)S+rVab.lib., Taq polymerase and forward and back prlmers and are cycled 25 times (94~C 1 min, 60~C for 1 min and 72~C for 2 min). The amplified, appended DNA is purified using Magic PCR PREPS (Promega) and after suspension in:water, 1 ~g of the purified DNA is digested with Notl and Apal and ligated using 1200U T4 ligase (Sambrook, Fritsch et al. 1990) to fdrVabpCARRIER (see Fig. 11) precut with Notl and Apal. The ligated product, designated forVabPEPpCARRIER is isolated with GeneClean and electroFor~ted into E. coli. Transformants are grown, tittered and ~rozen stocks are made. A sufficient nu-mber W096l04S57 PCT~S95/10182 .
2~6~

o of colonies are picked ana seo~uenced to~confirm the presence of the random pepB library. The bacteria, designated GHRT(x-y).rVab-PEP.lib bact are then grown and phage are induced for expresslon with helper phage so that the GHRT(x-y)rVab-pep constructs are dlsplayed on the phage surface~attached to gpIII (see Fig. 20~.
With the amino acids of the octapeptide being random at each position, there are~greater than 101~ peptide combinations for each library. Accordingly, with less than about lO0 GHR.TS+rVab in each group the combinatorial rVab-pep.lib number is less~than lOI~and is therefore accommodated in a normal:phage lysate. If the number of GXR.TS+ rVab is greater than about lO0, the random octapeptide library is expEessed alone as a fusion protein fused to the gpIII on the surface of fd phage via the same ls linker ([GGGGS]2) and the octapeptides which recognize GHR
surfaces are isolated first by panning over matrix attached GHR complexes. Those phage which stick to the matrix, are isolated, amplified and the oligonucleotide sublibrary encoding the pep8 octa~eptides which bind to GXR are excised and amplified with primers cnnt~-n1ng a leader restriction site (in the 5CK primer) and ApaLl (in the FWD primer). This smaller pep8 oligonucleotide sublibrary, which is T+ (pepT+), is then ligated into the grouped GXR.TS+EVab.lib precut at the rsl site in the VL
Lgplll=leader se~uence and~at Apa l (See Fig. llD) to produce a GHR.TS+rVab-pep(T+) library. In such cases the members of this combinatorial library, less than lol4, are gro~wn, the phage induced and the library of surface:
attached GHR.TS+rVab-pep(T+) harvested and tittered.
Ste~ 2b: Identificati~n Of ~tive ~iV~l~n~ rVsh-Pep Members ~ -2b) Isolate GHR(x-y)T+.rVab-pep members which actively (A+) dimerize the receptor as does GH. Label these GHR(x-y)TA+rVab-pep.

WO 961045S7 2 1 9 ~ 6 7 9 PCT/rJ595110182 r I

The bivalent rVab-P~P, are expressed as a phage di-splayed library and are panned ~or combinatorial members which actively dimerize GHR. The positives are labeled G~R.rVabT(x-y)SA+-pepTl+.lib. In this step, activation is recognized b~ the occurrence of one or more of the following o~servabIe events: 1) dimerization of two GHR
T1 subunits; 2) dimerization of two T1 subunits which allow fluorescence transfer between the same or different modified amino acids in the two subunits as described by Cunnlngham (Cunningham, ~ltsch et al. 1991); 3) dimmer formation which generates~an antibody recognized epitope which ~nnt~;nq amino acids from two T1 subunits which occur only in activated dimeric Tl2 structures (Taga, Narazaki et al. 1992); 3) GHR-GH-GHR-matrix complexes which are dissociated by wild type hGH, or only a mutant hGH with only site I or site II binding capability (C~lnn;ng~?m, ~ltsch et al. 1991); or 4) antibody recognizable phosphorylation of one of the receptor subunits associated with active receptor dimerization. In the later case, incubation of GHR.rVabT(x-y)S+pep.lib with ATP and PKC is carried out before panning and the ATP and P~C is present during the panning procedures. It is also possible to monitor for in vitro active dimerization by the co-presence of some third G~R associated protein in the active complex (Taga, Narazaki et al. 1992).
2c) Confirm activity by testing for activation of a cell associated GHR. Those GHR_TSA+rVab-pepT which appear active ;n vitro, are tested in an intact cell assay system such as GH induced growth of myeloid leukemia cell line FDCPl expresslng ~ybrid extracellular domain GHR-intracellular granulocyte colony-stimulating factor receptor (GCSFR) (Fuh, Cunningham et al. 1992) or IM-g cells (Silva, ~eber et al. 1993) to confirm the agonist nature of the rVab-pep complex.

L: ,~' - . -.

W096/04557 ~ PCT~S95110182

7 g ' 3. Identification of Active GH-rVah Pairs For Use As ReDorterg Step 3a, Ex~ression of Soluble rv~hs - 3a) Identify from among ~he membe~rs of di~ferent A+
rVabA+-pep groups, those which have a rVab which by itself ~
competes with the peptide member of the same or dlfferent _ rVabA+-pep group. This is accomplished by~carrying out competition binding assays designed:to identify tkosç
rVabs and peptides which compete with each other for binding to the GHR. The peptide portion of an active rVab-pep is separately expressed without the corresponding rVab to perform these binding assays. 3y this process rVabs which can mimic and replace the pep8 portio~=of an active rVab-pep member are identified. The rVab of a first A+rVab-pep member and the rVab of a second A+rVab-pep member which competes with the peptide portion of the first member, are designated an active pair of G~-rVabs.
Specifically, after confirmation of activation is obtained, the active rVab-pep are modified by appropriate digestion of the construct to allow expression of soluble rVab without any linkage to phage coat protein gpIII and to the octapeptide as well.= Such simplified~entitiçs are labeled rVabTS+A~. To prepare the modified constructs allowing for expre8sion of free soluble rVab, DNA from rVab-pep is obtained, digested with Apal and Notl and isolated. One ~g of the isolated DNA is then ligated with 5 ~g pEXPRESSIONrVab DNA precut with ApaL1 and Notl by ;nmlh~tinn with T4 ligagse. The ligated products are isolated by GeneClean II and electroporated into E. coli and transformants obtained and confirmed by diagnostic PCR ~
and sequencing. Fro2en stocks are prepared. These stocks are denoted GHR.rVabTS+A~ and not A+ because by themselves they cannot activate the G~R but are members of active pairs ~i.e., rVabs and pep8s) which do activate thç
receptor. Expression of the octapeptide member of the W096/04s57 21~ 6 ~ 7 9 PCT~895/l0182 r 1,~

-~143 --active rVab-pep is carried out by excision and ligation of the olio~nucleotide portions encoding the pep8 and transfer to expression vectors in which the pep8 is - expressed as a soluole extracellular entity fused with a easlly purifiable tagged carrier protein (using a variety of commercially available expression vectors) or attached via GGGGS Iinker to gpIII coat protein and displayed as a phage surface entity. These Pn~;t;p~ are labelled pep8A~
and are u-sed as descrlbed below to identify rVab for the other portion of the GHR active sur~ace utilized by the active rVab-pep entity.
3b~ rVab and pep8 members of active pairs are grouped-accordlng to common GHR surfaces recognized (as described above).
4. PrPn~ration of GH-rVsh-RP~Qrters Convert a rVab representative of at least one active pair of GH-rVabs into a GH.rVab-Reporter The CH domain of the heavy chain of the rVab is labelled (as described in Example 2) and the labelled entity, designated GH.:rVab-REPORTER, is used to establish saturation and competition binding assays as described in Example 2.
The isolated and expressed separated pep8 members from active rVabA+-pep constructs are used in standard binding competitio~ assays to identify (see Fig. 11) those GHRrVabT+ which bind to the same GHR domain as the pep8 entities. Those which compete are-designated as the second member of the active pair of rVab for the two active G~R
surfaces re~uire~ for receptor activation. This second member ls then converted to a rVab-Reporter (see above).
The rVab member of the rVabA+-pep construct from which the pep8 was obtained is the second member of the active pair.
St~"o 5 SnMR~ St'RRRNTNG
Establish binding assays with each member of an active pair of GH.rVab-REPORT_Rs for a pair of SO~ERS, each capable of binding to at least one of the two domains .. . , . , .. . _ _ _ . _ _ W096/04557 PCT~S95/10182 2~66~9 - 1~4 -of an active pair of recept:or surfaces invoIved in~active GHR dimerization.
The GH.rVab-REPORTER is used u~der standardi2e'~ and automated binding assay conditions to identify SOMERs within a chemical data base (i.e., CHEMFI~E) which will campete at an active* (A*) surface on the Tl subunit of the GH receptor. These SOMERs are designated SOMER~-Tl. In a parallel:fashion, us~ing t~e other rVab-Reporter member of the active rVab pair (as defined above~ SOMERs are isolated for the second active surface on GHR requ~red for its activation (Figs. 21 and 22). The SOMERs which recognize the second site are designated SOMER-Tl.:~
Identification of specific interaction with site I
(i.e., Tl) or site II (i.e_, Tl') of huGHR iB made in binding assays measuring the ability of~these entities to compete with mutant 125I-GH which can only bind to:~site I
or II as de~cribed (Cunningham, Ultsch et al l99l).
Ste~ ~: DISOM~ Pr~nAration ~n~ T~ntification of Druq I~ = ~: :
In the last step of this process, SOMER-Tl and SOMER-Tl' are covalently combined to create a bivalent SOMER (i.e., a DISOMER) which can recognize the two sites of the active surface pair, i.e., the Tl and Tl' receptor dimmer subunit active surfaces. This DISOMER can actively dimerize the GH receptor subunits as does the native hormone. Confirmation of DISOME~ GH activity is obtained in standard radioreceptor binding assays (competitive with intact labelled GH) for GHR-binding and standard activity assays (in vitro and/or GHR~cellular activation systems).
Additional assay systems for active hormone receptor subunit oliqomerizations~in which a free e~cellular receptor:hormone complex associates with other membrane proteins in intact cells to.form active oligomeric complexes which direct auto-, and substrate phosphorylation, and other down stream activation responses (Taga, Narazaki et al. 1992).

Wos6/04ss7 PCT~S95/10182 ~ 2~6~79 ., . , ' I ~ !; .; ~

Steps 1-4 of the process, which find active surface landscapes involved in active dimerization of two TI
subunits of G~R are outlined in Figs. 20, 21 and 22. Fig.
- 20 is a flow diagram for creation of rVab-pep. lioraries and isolation of rVab-peptides fQr the two active GHR
surfaces. In the example presented here of oligomeric receptor targets, there is only one type of subunit (Tl) in the active GXR dimer compl~ex, and therefore subunit T2 - Tl. Fig. 21,~2 illustrate GHTl- and GHTl'-SOMER and GX-DISOMER ti.e., GXTl-GHTl') identification.
0 R~ MPT,R 4 Example 4 is a variation of Example 3 which recognizes the fact that many hormonal receptors are comprised of different receptor subunits. Often at least two or three subunits which may all be different from each other are required for activity. In these cases, hormone induced receptor oligomerization a~sociated with receptor activation, requires interaction of the hormone with at least three active surfaces, each being on a different receptor 8ubunit. Examples of heterodimeric (alpha/beta, or alpha/gamma) receptors include the yroup of interleukin (IB) I~3;IL4,I~ 7,I~9 receptors and the GMCSF receptor, and the group-of growth factor PGF, PDGF, CSF and NGF
receptors, while an example of a heterotrimeric receptors ( alpha, beta and gamma) is the IB2 receptor (see reviews Pierce, 1989; Boulay,1993; Cosman, 1993; ~ishimoto, 1994;
~cnchcncky, 1993; Xondo, 1994; Noguchi, 1993; Russell, 1993 and Bamborough, 1994).
The use of rVab to identify active surfaces involving ~ two or more:sites distributed on multiple subunits involves certain adaption8 from the process used when activation requires only one site.~ First, with the heterooligomeric receptors, a different rVabT(x)S+ lib is i~nti ~ for each subunit (x) using the soluble receptor subunits as initial targets (e.g. Tavernier, 1991), as Second, that for trimeric receptors two random peptide 8 . ~ = . . , .. . ~ . - .

W096/04557 PCT~S95/10182 2 ~ f ~ --libraries are attached to each rVabT(x)S+ library. Third, where the rVab is T+ for the alpha receptor subunit (i.e., rVabT+S), the other two members of the active trio=(i.e., those binding to each of the other two subunit surfaces n~r~cc~ry for active receptor trimeriZation~, designated rVabT+ and rVabT+, are identified as those which compete for-bindiny with one of the_two octapeptide members o~ an active rVabTSA+-pepl. For such trimeric receptors,~the indivldual rVHCX.lib and rU~CL.lib made in Example 1 are combined into different fdRECEIVERs and pUC19PROVIDERs as l~ rl~t~ in Fig. 13.
In this application~ rVHC~.lib is placed into a fdRECEIVER which allows expression of rVHC~ fused to gpIII :
coat protein and with, or without, peptide (preferably 8 amino acids) attached to its aminoterminus. The rVLCL.lib 5 is placed into a pUCPROVIDER which allows for expression of rVLCL as soluble entitieq with, or without, peptide, preferably 8 amino acids, attached to its CL domain.
After in vivo Cre-~ox -reco~bination of these two libraries, as detailed in Example l, (see also. Fig. 13) the product rVab.lib is cloned as a single fdDNA
designated fdrVabPEPCARRIER rVab members which bind to each of the receptor subunits (i.e., Tx+rVab) are then isolated and grouped as described in Example 3.
Subser~uent addition of one or two random octapeptide 25 libraries (Pep8n), which in some cases have been prescreened and selected for hinding to an identified receptor subunit is accomplished via PCR. As described above and in Fig. 13, oligonucleotides ~nrn~n~ the peptides are added to the DNA encoding the rVab library 30 using FWD primer CLLNKs~w~ ~~ ~ ~~~~~~ ~ ~ ~~~
(Ascl-(NNN)8(GGGGS)3CLL208-216) and VHLNKPEPBCK
(rsPE~B-(~NN)8~GGGGS)VH1-8)-together or in combination with primers having no Pep8 or linker- appending ser~uences. Use of one of~these primers with a primer 35 devoid of a Pep8 library could be used to generate a rVab w096l04ss7 PCT~S95/10182 2~9~79 .'.'?

with ~ne attached pep8 (i.e., rVab-Pep8') as described above in Example 2 with the single Pep8 library appended through linker to the either the aminotPrm;nllq of the - rVHCH1 member or the carboxyterminus of the rVLCB member (~ig. 13). ~ ~
According to this process, each attached peptide and the rVab portion of the rVab-PEP2 each bind to a specific target site. Binding to all three sites is re~uired for activity of the receptor. Therefore, the trimeric rVab-PEP2-unit deflnes three binding domains: one defined by the rVab portion~((T(x)), and bne each by each of the pep8 (i.e., pep3l and pep82) present in the construct.
Isolation of active rVab-Pep2members utilizes enrichment cycles in which all three receptor units are complexed together in active trimeric structures. Such structures, complexed with their phage expressed rVab-Pep2 entities, are enriched by use of matrix-bound active subunits, antibodies to each of the three units, antibodies to modifications of receptor units which occur upon active oligomerization, such as phosphorylation or association with additional non-receptor membrane components (Argetsinger, Campbell et al. 1993;
Silv~nn~;n~n, Witthuhn et al. 1993; and Witthuhn, 1993).
Confirmation of agonist and antagonist activity is done using standard hormone : receptor binding assays to 2S establish competitive binding of hormone to its receptor (~itamura, Sato et al. 1991; Imler and Zurawski l9S2;
Pietzho, Zohlnhofer et al 1993) and cellular receptor dependent activity assays measuring growth, DNA synthesis, ~ protein phosphorylation etc. (Yokota, Otsuki et al. 1986;
Pierce, Ruggiero et al. 1988; Solari et al. 1989;
AnXlesaria, Teixido et al. 1990; ~eidaran, Pierce et al.
1990; Pierce, Di et al. 1990; Heidaran, Pierce et al.
1991; Xeegan, Pierce et al. 1991; Murakami, Narazaki et al. 1991; Xruse, Tony, et al. 1992; Otani, Siegel et al.

,~ c W096/04ss7 PCT~S95/10182 7~
- l~B -1992; Taga, Narazaki et aL.~992; and Wang, Ogorochi et al 1992) ~ ~ -For a given rVab-Pep82,~we identify rVabA*s which are cnnt~;no~ in other rVab-Pep82A+ which bind to each o~ the target sites bound by the peptides o~ the original active rVab-Pep82 (trimer rVabA* unit), following the same process outlined in Example~3. ~sing this process, ~embers of the trimeric unit are identified as ~) any rVabT+ from another active construct (i.e., rVabTSA+-Pep82) which competes with one of the two PEP lib on the original active rVabTSA+-PEPI, b) with any rVabT+ from a third active construct (i.e , rVabTSA+-Pep82) which competes with the other PEP on the D~iginal active rVabTSA+-PEP2 and c) the rVabTS+ of the origlnal active rVabTS+-PEP2.
Competition for binding to GHR is de~rm;~ by assaying for competition of P~P8 units expressed either attached to gpIII coat protein and presented as phage~dlsplayed entities or as soluble fusion proteins with labelled rVabTx+-Reporters which are made as described above in example 2 and 3. After ideLtification of all three of the active trimer members, each rVab member of the active timeric unit is then cloned minug its Pep8 library member(s), expressed, isolated and converted to a rVab-REPORTER r~as detailed ln Example 1, and used to establish competitive binding assays which then find competing SOMERs (i.e., Somer-T" T2-or T3) In the final stage covalent linking of the three-Somer-Ts) is done so as to construct the active multimer, in this case a TRISOMER (i e., T,-T2-T3), substitute for the native -hormone In these systems, a~ additional receptor activation assay system is available for heterooligomeric receptor activation which monitors the induction ofi~n~; f; ~hl e holoreceptor induced cellular responses by preformed soluble complexes of hormone and one of the receptor subunits in response to the binding of these complexes to intact cells expressing the other subunit(s) WO 96/04557 . 21 9 6 ~ 7 9 pcT~r89sllol82 ~ -~ T
, - 149 - - -of the active receptor complex and formation of active holoreceptor complexes (Taga, Hibi et al. 1989).
In these systems, an additional receptor activation - assay ~ystem.may be used to,confirm heterooligomeric receptor activation. Such systems monitor the induction of i~nt; f; Ahle cellular responses induced by the ~ ~;n~tion of preformed 601uble complexes comprising hor~one and one of the receptor:6ubunits and intact cells expressing th,e,other subunit(s) of the active receptor complex and the subsequent formation o_ active complete holorece~tor complexes (Taga, Hibi et al. 1989).
The following Table lists exemplary ligands and heterooligomeric receptor systems for which this invention provides a means for identifying their pharmacologic target sites as well as SOMERS or DISOMERS.
ntrr~ ;n1 Immune System Supres3ion/S~ t;An Agonist/Antag.

IL2-7, 9-Il Immune System Supression/Sf; 1~;An Agonist/Antag.

Insulin Like Growth Factors: N~AP1 AA; A U ~nfArrnn;Ut Erythropoiesis Agonist(synergis.

w Epo) ~rAnn1npn;rA;~ Agonist(synergis.

w GMSCF) - ~GFbetas Wound Healing (Matrix proteins) Agonist . Tnfl -;nn ~n~5rnn;ct Carr;, - n Antagonist ~nfAT Disease Antagonist GCSF Chemotherapy Agonist Bone Marrow TrAnAr1At~nn Agonist CSF Bone Marrow Failure Syndromes (re: rAAiA~1nn/rh~mn~h~rAry) Agonist Tnf1: -nry ~n~gnn;ut Neoplasms (acute myeloid leukemia) Antagonist Erythropoietin T~ APA; rU; A (anemias) Agonist GMCSF Immune Suppression/St; lA~;An AgoniAt/Antag.

PDGF Wound Repair Ayonist T~Jr;l_ ;u P~t~-~n ut V~u,.,,,,.~l,,r~;An ~r~A ~n ut D~hrrnur1rrOsis T~nt~Tnn ~t Neoplasms T~nr,rnn ut Pulmonary Fibrosis Dnt~rnn Ut Tnf1~ nry Joint Diseases Pnf~rnn ~t EGF Wound Repair Agon.st Neoplasms Antagonist FGF Neoplasms Antagoni_t Wound Repair Agonist Angiogenesis (Capillary Blood Vessels) Antagonist NGF Antil._ll~d~ ~tive Diseases Agonist (Acute/Chronic); (Peripheral/Central) Small Organic Molecules r_~ tturS

i.e. OhA1;nnm; ;ru (ACh C

W096/04557 ~ PCT~S95110182 7 ~ --mReceptor 1-~) Agonist/~ntag.
Trans~orter/Channel ~.l At~rA Agonist/Antag.

References ~ ~ ;
Andre, C., et al. (1987). ~Immunochemical studies of the muscarinic ac~etylcholine receptor.
~ecept ~es 7(1-4): 89-103.
Andrews, P.R., Craik, D.J., Martin, J.~., (1984). Functional group contributions to drug receptor interactions. J. Med. Chem. 27:1648-1657.
Anklesaria, P., et al. (1990). "Cell-cell adhesion mediated by binding of membrane-anchored transforming growth factor alpha to epidermal growth factor receptors promotes cell proliferation.~ Proc Natl Acad sci u s A 87(9): 3289-93.
Appleyard, R. K. (1954). Genetic~7 39: 440-452.
Argetsinger, L. S., et al. (1993).
~Identification of JAK2 as a growth hormone receptor-associated tyrosine kinase.~ Cell 74(2): 237-44.
Bachar, O., Fisher, D., Nussinov, R., Wolfso~, H.J. (1993). A computer vision based technique for 3-D
se~uence ln~PpPn~Pnt structural comparison of proteins.
Protein Eng. in press.
Bamborough, P., et al. (1994). ~The interleukin-2 and interleukin-4 receptors studied by molecular modelling.r Curr.Biol. 2: 839-851.
Barbos, C.F., Kang, A.S., Lerner, R.A. and Benkovic, S.J. (1991). Assembly of cn~h;n~torial antibody libraries on phage surfaces: the gene III site."
Proc. Natl . Aca~. Sci. USA 88:7978-7982.
Bass, S., et al. (1990). ~Hormone phaye: an enrichment method for variant proteins with altered binding properties. r Proteins 8(4): 309-14.
Berrie, C. P., et al. (1985). BioC~em Soc.
~rans. 13: 1101-1103.
Beth, A. (1993). ~i~e Sci. 52: 429-432.

w096l04557 2 ~ 7 ~ PCT~S95110182 .;
. ,, Bird, R.E., Hardman, K.D., Jacobson, J.W., Johnson, S., Kauifman, B.M., Lee, S.M. Lee, T., Pope, S.H., Riordan, G.S. and Whitlow, M. (1988J. "Single-chain antigen-binding Froteins." Sci . 242:423-426.
Boulay, J. L. and W. E. Paul 11993).
"Hematopoietin sub-family classification based on size, gene organization and sequenc~ homology.[Review] " Curr.
Biol. 3: 573-581.
Breitling, S.D., Seehaus, T., Klewinghaus, I.
and Little, M. (1991). "A surface expression vector for lO antibody screening." Gene 104:147-153.
Brooks, B.R., Bruccoleri, R.E., Olafson, B.D., 9tates, D.J., Swaminathan, S., Karplus, M., (1993).
C~ARMM: A program for macromolecular energy, m;n;m; 7~tion and dynamics calculation. J. Comp. Chem. 4:187-217.
I5 Bruccoleri, R.E., Karplus, M., (1987).
Prediction of the folding of short polypeptide segments by uniform conform-t;-~n~l sampling. Biopolymers, 26:137-168.
Buckley, N. J., et al. (1989). Antagonist binding properties of five cloned muscarinic receptors 20 expressed in CH0-K1 cells." Mol Pharmacol 35(4~: 469-76.
Buckley, N. J.! et al. (1990). "Use of clonal cell line~ in the analysis of neurotransmitter receptor me~h~n;r-- and function [Review] " Siochim i9iophys Acta 1055(1): 43-53.
Burton, D.R., Barbas, C.R., Persson, M.A.A., Liening, S., Chanock, R.M. and Lerner, R.A. (1988). ~A
large array of human monoclonal antibodies to type 1 human ; ~t~t~ficieLcy virus from combinatorial libraries of asymptomatic seropositive individuals." Proc. Natl . Acad.
30 sci. USA 88:10134-10137.
Cabilly, S., Rigg~, A.D., Pande, H., Shively, J.E., Bolmes, W.E., Rey, M., Perry, L.J., Wetzel, R., Heyneker, H.L. (1984). "Generation of antibody activity from immunoglobulin polypeptide chains produced in 35 Escherichia coli." Proc. Natl. Acad. Sci. US 81:3273-3277A.

W096/04557 PCT~S95/10182 6 7 ~

Chothia, C. and A M. Lesk (1987). "Canonical structures for:the hypervariable regions of immunoglobulins.~ Journal of Molecular Biology 196(4):
901-17.
Chothia, C., et al. (1992). "Structural repertoire of the human VH segments.- J Mol Biol 227(3):
799-817.
Chothia, C., et al. (1989). "Conformations of immunoglobulin hypervariable regions [see c~m~nt~].
[Review~.~ Nature 342(6252): 877-83.
l~ Clackson, T., Hoogenboom, H.R., Griffiths, A.D.
and Winter, G. (1991). Nature 352:624-628.
Clackson, T. and Wells, J.A. (1995). A Xot Spot of Binding Energy in a Hormone Receptor Interface Science 267:383-386.
Claesson-Welsh, L. (1995). Platelet-derived Growth Factor receptor Signals.~ ~. Biol. Chem.
269:32023-32026.
Cosman, D. (1993). "The hematopoietin receptor superfamily. [Review].~ Cytokine 5(2): 95-106.
Cox, J.P.L., Tomlinson, I.M. and Winter, G., (1994). "A directory of human germ-line V kappa segments reveals a strong bias in their usage.~ Eur. J. Immunol .
24:827-836.
~nnn;ngh~m, B. C. and J. A. Wells :(1989).
~xigh-resolution epitope mapping of_hGH-receptor interactions by alanine-scanning mutagenesis.r Science 244(4908): 1081-5.
~nnningh~m, B. C., et al. (1990). ~Zinc mediation of the binding of human growth hormone to the human prolactin receptor.~ Science: 25D(4988): 1709-12.
Onnningh~m, B. C., et al. (1990). ~Engineering human prolactin to bind to the human growth hormone receptor.~ Sci ence :
rnnningh~m~ B. C., et al. (1991). "Dimerization of the extracellular domain of the human growth hormone =

-~O 96/045~7 21 g ~ 6 7 !~ PCT/US9~lO182 receptor by a 6ingle hormone molecule. n Science 254(5033): 821-5.
Cwirla, S.~., Peters, E.A., Barrett, R.W., and - Dower, W.J. (1990). ~Peptides on phage: a vast library of peptides for identifyi~g ligands." Proc. Natl . Acad .
5ci. USA, 87:6378-6382.
DeFronzo, R.A. Bnn~nnn~ R.C. and Ferrannini, E. ~1992). ~Diabetes Care 15:318-368.
Delvin, J.J., Panganiban, D.C., and Devlin, P.E., (1990). "Random Peptide Libraries: A Source of Specific Protein Binding Molecules." Science 249:404-406.
DeVos, A. M., et al. (1992). ~Human growth hormone and extracellular domain of its receptor: crystal structure of the complex.~ Science 255(5042): 306-12.
Dower, W. J., et al. (1988). High efficiency transformation of E. coli by high voltage electroporation." Nucleic Acid~ Research 16(13): 6127-45.
Elber, R., Xarplus, M. (1990). ~nh~nr~
Sampling in molecular dynamicg: Use of the time-d~r~n~nt Hartree approximation for a simulation of carbon r- n~; ~G
diffusion through my myoglobin. J. Am. Chem. Soc.
112:gl61-9175.
Figini, M., Marks, J.D., Winter, G. and Griffiths, A.D. (1994). "In Vitro Assembly of Repertoires of Antibody Chains on the Surface of Phage by Renaturation." ~. Mol. ~iol. 239:68-78.
Fisher, D., Bachar, O., Nussinov, R., Wolfson, H.J. (1992). An efficient automated computer vision based techni~ue for:detection of three dimensional structural - motifs in proteins. J. Biomolec. Struct. Dyn. 9(4):769-789.
Fisher, D.R., Norel, R.H., Wolfson, and Nussinov, R., (1994). Surface motifs by a Computer Vision Techni~ue: Sçarches, Detection and Implications for Protei~ Ligand Recognition, Protein: Structure, Function and Genetics, 16:278-292.

W096/04557 , ~ . PCTN595/l0l82 2 1 ~ ~ 6 r ~

Florio, V. A. and P. C. Sternweis (1985).
"Recon~titution of resolved muscarinic cholinergic receptors with purified GTP-binding proteins." J. Biol.
Chem. 260(6): 3477-83.
Folch, J., et al. (1957). J. Bio. Chem.
Fraser, C. M., et al. (1989). ~Site-directed mutagenesis of ml muscarinic acetylcholine receptors:
conserved aspartic acids play important role~ in receptor function.~ ~ol Pharmacol 36~6): 840-7.
Fuh, G., et al. (1992). "Rational design of potent antagonists to the human growth hormone receptor.
Science 256(5064): 1677-80.
Fuh, G., et al. (1990). "The human growth hormone receptor. Secretion from Escherichia coli and disulfide bonding pattern of the extracellular binaing domain.~ J. Biol. Chem. 265(6): 3111-5~
Garrard, L.J., Yang, M., O'Connell, M.P., Kelly, R.F. and ~enner, D.J. (1991). ~Fab assembly and enrichment in a monovalent phage display system.
Bio/~echnology 9:1373-1377.
Garrett, K.M., Blume, A.J. and M.S. Abel (1989).
~Effect of halide ions on [35S] TBPS binding to Gaba-A
Receptors.~ J. Neurochemistry 53:935-939.
Garrett, K.M., Abel, M.S. and A.J. Blume (1989).
~Effects of various GABA-A Receptor Mn~ nrs on [35S] TBPS
~5 binding in the pre~ence of various halindes.~ J.
Neurochemi~try 53:940-945.
Gibson, T. J. (1984). University of Cambaridge, UK.
Goldberg, D.E. (1989). Genetic Algorithms in Search, Optimization and Machine ~earning Addison-Welsey pnhl; f~h 1 n g CO .
Goodford, P.J., (1985). A computational procedure for ~t~rm;n;ng energetically favorable binding sites on biologically important macromolecules. J. Med.
Chem. 28:849-857.

WO 96~04557 219 6 6 7 ~ pcr/lTs95/lol82 - - f ~

Goodsell, D.S., Olson, A.J. (1990). Automated docking of substrates to proteins by simulated a~nealing Proteins:3:195-202.
~ Griffiths, A. D., et al. (1994). ~Ieolation of high affinity human antibodies directly from synthetic S repertoires." EMB~ 13 (14): 3245-3260.
Habecker, B. A., et al (1993). ~Regulation of Expression and Function of Muscarinic Receptors.~ Life Science~ 52: g29-432.
Xaga, K. and T. Haga (1983). "Affinity chromatography of the muscarinic acetylcholine receptor. n . Biol. Chem. 258(22): 13575-9.
Haga, K., et al. (1986). "Reconstitution of the muscarinic acetylcholine receptor. Guanine nucleotide-sensitive high affinity binding of agonists to purified muscarinic receptors reconstituted with GTP-binding proteins (Gi and Go).~ ~. Biol . Chem. 261(22):
10133-40.
~akwins, R.E. and Winter, G. (1992). ~Cell selection strategies for making ~rt;ho~ from variable gene libraries: trapping the memory pool." Eur. ~.
Immunol . 22:86~7-870.
Heidaran, M. A., et al. (1990). ~Chimeric alpha-and beta-platelet-derived growth factor (PDGF) receptors define three immunoglobulin-like domains of the alpha-PDGF
receptor that ~t~rm;nP PDGF-AA binding specificity.~ ~.
Biol. Chem. . 265(31): 18741-4.
Heidaran, M. A., et al. (1991). ~Role of alpha beta receptor heterodimer formation in beta platelet-derived growth factor (PDGF) receptor activation by PDGF-AB.r ~. Biol. Chem. 266(30): 20232-7.
~oess, R. H., et al. (1986). "The role of the loxP spacer region in P1 site-specific re ~;n~tion.
Nucleic Acis Re~. 14: 2287-2300.
~oess, R. H , et al. (1982). ~P1 site-specific 35 recombination: nucleotide gequence of the r~ n;n~

W096/0~557 ~ t; ~ ~'. PCT~S95/10182 2~6~7~
~ 156 - ~
sites. n PrPceedings of the National Academy of Sciences of the United States of America 79111): 3398-402.
Holland, J.H., (1992~. Adaption i~ Natural and Artificial Systems. The MIT Press, Cambridge, Ma~s.
Holliger, P., Prespero, T. and Winter, G.
(1993). "Diabodies": Small bivalent and bispecific antibody fragments . " Proc. Natl . Acad. sci . USA 90:6444-6448.
Hoggenboom, J.R., Griffiths, A.D., Johnson, K.S., Chi~well, D.J., Hudson, P., and Winter, G. (1991).
"Multi-subunit proteins on the gurface of f;l; tpllC
phage: methodologies for displaying antibody (Fab) heavy and light chains." Nucleic Acid. Res. 19:4133-4137.
Hoogenboom, H.R. et al. (1991), Nucleic Acid Res. 4133-4137; Jespers, L.S., Roberts, A., Mahler, S.M., Winter, G. and Hoogenboom, J.R. (1994). "Guiding the selection of human antibodies from phage display repertoires to a single epitope of an antigen." ~=
Bio/Technology 2:899-903.
Hoogenboom, H.R , Marks, J.D., Griffiths, A.D., and Winter, G. (1992). "Building antibodies from their genes." Immunol. Rev. 130:41-68.
Horton, R. M., et al. (1989). "Engineering hybrid genes without the use of restrictiPn enzyme~: gene splicing by overlap extension." ~ene 77(1). 61-8.
Hulme, E. C., et al. (199D). "Muscarinic receptor subtypes [Review].~ Annu Rev ph~rm-co7 Toxicol 30: 633-73.
Huse, W.D., Sastry, L., Iverson, S.A. ~ang, A.S., Alting-Mees, M. Burton, D.R_, Benkovic, S.J. and Lerner, R.A. (1989). "GeneratiPn of a Large Combinatorial ~ibrary of the Immunoglobulin Repertoire in Phage Lambda."
Sci. 346:1275-1281.
Huston, J.S., Levinson, D., Mudgett, H.M., Tai, M.S. Novotny, J., Margolies, M.N., Ridge, R.J., Bruccoleri, R.E., Haber, E., Crea, R., and Opperman, H.
(1988). "Protein ~ng;n~ring of antibody binding sites;

W096l04ss7 21~ ~ 6 7 9 PCT~ss5/l0l82 recovery of specific activity in a anti-digoxin single-chain Fv analogue produced in Escherichia coli." Proc.
Natl . Acad. Sci . USA 85:5879-5883~
ImIer~ J. L. and G. Zurawski (1992). "Receptor binding and intPrn~l;zi~tion of mouse interleukin-2 derivatives that are partial agonists.~ J. Biol . Chem.
267(19): 13185-90.
Kabat, E. A., et al. (1991). Se~uences of Prote;nq of T ~locical Interest. 5th ~n US Dept Health And Human Services, Bethesda, MD. USA.
I0 Kang, A.S., Barbas, C.F., Janda, K.D., Benkovic, S.J. and Lerner, R.A. (1991). "Linkage of recognition and replication functions by assembling combinatorial antibody Fab libraries along phage surfaces." ~roc. Natl. Acad.
sci. USA 88:4363-4366.
I5 Kaplan, D.R., Martin-Zanco, D. and Patrada, L.F.
(1991). ~Tyrosine ph~rh~rylation and tyrosine kinase activity of the trk proto oncogene~product induced by NGF." Nature, 350:156-160.
Kaplan, D.R., Martin-Zanco, D. and Patrada, L.F.
(1991). ~The trk proto-oncogene product: a signal transducing receptor for nerve growth factor.~ Sci.
252_554-558.
~hihi~ra, K., et al. (1992). Cloning of the rat M3, M4 and M5 muscarinic acetylcholine receptor genes by the polymerase chain reaction (PCR~ and the pharmacological characterization of the expressed genes.
Life Sci. 51(12): 955-71.
~nll~h~n~ky, K. and P. A. Karplus (1993).
"Hematopoietic growth factors: understanding functional diversity in structural terms. [Review].~ Blood 82(11):
3229-40.
Keegan, A. D., et al. (1991). "Ligand stimulation of transfected and endogenous growth factor receptors ~nhnnr~ cytokine production by mast cells.
Embo J. 10(12): 3675-82.

_ _ .. _ _ _ . .. . _ . _ _ _ _ _ _ Wo9~/04ss7 PCT~S95110182 6 7 9; . ' ~ ~

Kelly, P. A., et al. (1991). ~The prolactin/growth hormone receptor family." Endocr Rev 12(3): 235-51 Kishimoto, T., et al. (1994). "Cytokine signal transduction. [Review] " Cell 76(2): 253-62.
Kitamura, T., et al. (1991). "Expression cloning of the human IL-3 receptor cDNA reveals a shared beta subunit for the human IL-3 and GM-CSF receptors." Cell 66(6):-1165-~4~ ~
Klein, R., Jing, ~.~., Nanduri, V., O'Rourke, E., and B~rh~ M. (1991). "The trk proto-oncogene encodes a receptor for nerve growth factor.~ Cell 65:189-197.
Kobayashi, I., et al. (1990). ~Purification and characterization of five different alpha subunits of guanine-nucleotide-binding proteins in bovine brain membranes. Their physiological properties ~n~rn;n~ the activities of adenylate cyclase and atrial muscarinic K+
~h~nn~lc." Eur ~. Biochem. 191(2): 499-506.
Kruse, N., et al. (1992). ~Conversion of human interleukin-4 into a high affinity ~nt~n; ct by a single amino acid repl~- ." Embo J 11(9): 3237-44.
Kubo, T., et al. (1986). ~Cloning, se~uencing and expression of complementary DNA encoding the muscarinic acetylcholine receptor." Nature 323(6087):
411-6.
Lamden, Y., Schwartz, J.T., Wolfson, H.~.
(1990). Affine invariant model-based object recognition IEEE Trans. Robotics Ant~ t10n 6(5)578-589.
Lamden, Y., Wolfson, H.J. ~1988). Geometric Hashing: A general and efficient model-based recognition scheme. Proc. IEEE Int. Conf. Computer Visions, Tampa, FL
Dec. pp. 238-249. : ~
Lazareno, 9., et al. ~1993). "Pharmacological characterization of guanine nucleo~ide exchange reactions in membranes from CX0 cells stably transfected with human muscarinic receptors ml-m4." Life Sci. 52~5-6): 449-56.

w096~04ss7 . pcT~sss/lol82 21~6~7~ ' ~ - 159 -O
Lerner, R.A., Kang, A.S.-Bain, J.D., Burton, D.R. and~Barbas C.F. (1992). ~Antibodies without ; i7ation~ Sci. 258:1313-1314.
- Leung, D. W., et al. (1987). "Growth hormone receptor and serum binding protein purlfication, cloning and expression.~ Nature 330(6148): 537-43.
Li, B. L., et al. (1989). Prox. Natl. Acad. Sci.
USA 86: 558-562.
Malby, R.L., Tulip, W.R., Harley, V.R., McKimm-Beschkin, J.L., Laver, W.G., Webster, R.G. and Colman, P.M. (1994) The structure of a complex between the NC10 antibody and influenza virus neuri~m;ni~A~e and comparison with the overlapping binding site of the NC41 antibody.
Structure, 2(8)733-746.
Marks, J. D., et al. (1991). ~By-passing ;r-nni7~ti~n~ ~uman antibodies from V-gene libraries displayed on phage.~ ~ournal of Molecular Biology 222(3):

Mark6, J.D., Hoogernboom, H.J.R., Griffiths, A.~., and Winter, G. (1992). Molecular evolution of protein on f;li -tous phage: mimicking the strategy of the immune system.~ J. Piol. Chem. 267:1-4.
McCafferty, J., et al. (1990). ~Phage antibodies: ~ilamentous phage displaying antibody variable domains.~ Nature 348(6301): 552-4.
McCafferty, J., Griffiths, A.D., Winter, G. and Chiswell, D.J. tl990). "Phage antibodies: filamentous phage displaying antibody variable domains.~ Nature 348:552-554.
- Miller, J. H. (1972). ~eriments ;n Mclecl]li~r Gen~tics~ Cold Spring Harbor, NY, USA, Cold spring Harbor Laboratory Press.
Milstein, C. (1990); The Croonian Lecture (1989). "~ntiho~;es: a paradigm for the biology of molecular recognition.~ Prox. R. Socl . ~on~. Piol. 239:1-16.

W096/0~557 ~ PCT~S95/1018z '6~i~9 Miranker, and A., Karplus, M. (1991).
Fnnrt;nn~l;ty Maps of Binding Sites: A Multiple Copy Simultaneous Search Method ~roteins, Structure, Function, and Genetics, 11:29-34.
Mullinax, R.L., Gross, E.A., Amberg, J.R., Hay, B.N., Xogrefe, H.H., Kubitz, M.M., Greener, A., Alting-Mees, M., Ardourel, D., Short, J.M., Sorge, J.A., and Shopes, B. ~1990). "Identification of human antibody fragment clones specific for tetanus toxid in a bacteriophage ~ immunoexpression library." Proc. Natl.
Acad. Sci. USA 87:8095-8099.
Murakami, M., et al_ ~1991).
Proc.Natl.Acad.Sci.USA 88: 11349-11353.
Nissim, A., Hoogenboom, H.R., Tomlinson, I.M., Flynn, G., Midgley, C., Lane, D. and Winter G. ~1994).
"Antibody fragments from a 'single pot' phage di~play library as immunochemical reagents.~ E~BO J. 13:692-698.
Nossal, G.J. ~1993). ~Tolerance and ways to break it." [Review]. Ann_ N.Y. Acad. Sci. 690:34-41.
Novotny, J. Bruccoleri, R.B., Saul, F.A., ~1989). On the attributionlof binding energy in antigen-antibody complexes McPC603,D1.3 and HyHBL-5.~Biochem.
28:4735-4749.
Obermeier, A., Halfter, H., Wiesmuller, K., Jung, G., Schlessinger, J. and Ullrich A. ~1993).
"Trosine 785 is a major determinant of Trk-substrate interaction.~ E~9O J. 12:933-941. _ ~
Otani, H., et al. ~1992). ~Interleukin ~IL)-2 and IL-3 lnduce distinct but overlapping responses in murine IL-3-dependent 32D ce~ls transduced wi~h human I~-2 receptor beta chain: involvement of tyrosine kinase~s) other than p561ck.~ Proc. Natl. Acad. Sci. USA 89~7):
2789-93.
Parmely, S.F., and Smith, G.P. ~1988).
~Antibody-selectable filamentous fd phage vectors:
affinity purification of target genes." Gene 73:3Q5-318.

WO 961045~7 ~ 79 PCT~S95~10182 Patai, S. (1989). Patai's Guide to the Chemistry of Functional Groups, New York, John Wiley and Sons, and Brooks - Pierce,-J. H., et al. (1988). "Signal transduction through the EGF receptor transfected in IL-3-dependent hematopoietic cells.~ Science 239(4840):
628-31.
Pierce, ~. H. (1989)~. "Oncogenes, growth factors and hematopoietic cell transformation [Review].~ Biochim Biophys Acta 989(2): 179-208.
~ Pierce, J. ~., et al. (1990).
"Macrophage-colony-stimulating factor (CSF-1) induces proliferation, chemotaxis, and reversible monocytic differ~nt;At;on in myeloid progenitor cells transfected with the human c-fms/CSF-1 receptor cD~7A.~ Proc. Natl.
Acad. Sci. USA 87(15): 5613-7.
Pietzho, D., et al. (1993). "The Hepatic Interleukin-6 Receptor.~ ~. Biol. Chem. 268: 4250-4258.
Poyner, D. R., et al. (1989). ~Binding and hydrodynamic properties of ~uscarinic receptor subtypes solubilized in 3-(3-cholamidopropyl)dimethylammonio-2-hydroxy-1-propanesu lfonate.~ ~ol Ph~rr-~o7 36(3): 420-9.
Roitt, I.M. (1991). ~ssential immunology 7th edition, pgs 35-64. Blackwell Scientific Publications.
Boston.
Rosner, J. L. (1972). "Formation, induction, and curing of basteriophage P1 lysogens.r Virology 48(3):
679-80.
- Rozakis-Adcock, M. and P. A. Kelly (1992).
" T~ent; f; cation of Ligand Binding Determinants of the Prolactin Receptor.~ ~. Biol. Chem. 267(11~: 7428-7433.
Russell, S. M., et al. (1993). "Interleukin-2 receptor gamma~chain: a functional component of the interleukin-7 receptor.~ Science 262: 1880-1883.

W096l04sS7 PCT~S951l0l82 9, Sambrook, J., et al. (1990~. Molecular Cloninq:
A LabQratorY Manual. Cold Spring Harbor, NY. USA, Cold Spring Harbor Laboratory Press.
Sastry, L., Alting-Mees, M., Huse, W.D., Short, J.M., Sorge, J.A., Hay, B.N., Handa, K.D., Benkovic, S.J.
and Lerner, R.A. (1989). "Cloning of the immunological repertoire in Escherichia coli for generation of moniclonal catalytic antibodies: Construction of a:heavy chain variable region-specific cDNA library." Proc Natl.
Acad. Sci. USA 86:5728-5732.
I0 Scott, K.M., and Smith, G.P. (1990). ~Searching for peptide ligands with an epitope library.~ Sci.
249:386-390. ~
Shapiro, R. A. and N. M. Nathanson (1989).
"Deletion analysis of the mouse ml muscarinic acetylcholine receptor: effects on phosphoinositide metabolism and down-regulation." Biochemistry 28(22):
8946-50.
Shapiro, R. A., et al. (1988). ~Isolation, eequence, and functional expression of the mouse M1 muscarinic acetylcholine receptor gene [pnhli-~h~d erratum appears in J Biol Chem 1989 Apr 15j264(11):6596]." ~.
Biol. Chem. 263(34): 18397-403.
Short, J.M., et al. ~1988). Nucleic Acid Res.
16:7583-75.
Silva, C. M., et al. (1993). ~Stimulation of tyrosine phosphorylation in human cells by activation of the growth hormone receptor." E~ndocrinology 132(1): 101-8.
Silv~nnn;n~n~ O., et al. (1993). Structure of the murine Jak2 protein-tyrosine kinase and its role in interleukin 3 signal transduction." Proc. Natl. Acad. Sci.
USA 90 (18): 8429-33.
Skerra, A., and Pluckthun, A. (1988). ~Assembly of a Functional immunoglobuli~ Fv Fragment in Escherichia coli." Sci. 240:1038-1041.
Smith, C. J., et al. (1987). "Guanine nucleotide modulation of muscarinic cholinergic receptor binding in ~, .. :. . .: .... ... ... .. ... _ wos6l04ss7 , PCT~S95/10182 219~67~

postmortem human brain--a prr-;mincry study in ~l7hP;m~r~8 disease.~ Neurosci Lett 82(2): 227-32.
Smith, G.P. (1985). ~Filamentous fusion phage:
- novel expression vectors that display cloned antigens on the virions surface.~ Sci. 228:1315-1317.
Solari, R. and e. al. (1989). Biochem J. 262:
897-908.
Spencer, S. A., et al. (1988). Rabbit liver growth hormone receptor and serum binding protein.
Purification, characterization, and sequence.~ J. Biol.
C~em. 263(16): 7862-7.
Stahl, N. and Yancopoulos, G.D., (1993). ~The Alpha, Betas and Kinases of Cytokine Receptor Complexes.
Cell . 74:587-590.
Stern'oerg, N. and D. ~amilton (1981).
~Bact~r1Oph~ge Pl site-specific rrl h;nction~ I.
~r ';n~tion between loxP sites.~ Journal of ~olecular Biology 15P(4): 467-86.
Sternweis, P. C. and J. D~ Rr~h;Rh~w (1984).
~Isolation of two proteins with high affinity for guanine nucleotides from membranes of bovine brain.~ Journal o~
Biological Chemistry 259(22): 13806-13.
Taga,;T., et al. (1989). ~Interleukin-6 triggers the ~cs~ri~tion of its receptor with a possible signal tr~nC~n~r, gpl30.~ Cell 58(3): 573-8I.
Taga, T., et al. (1992). ~Functional inhibition of hematopoietic and neurotrophic cytokines by blocking the interleukin 6 signal transducer pgl30.
Proc.Natl.Acad. Sci. USA 89: 10998-11001.
Tavernier, J., et al. (1991). ~A human high affinity interleukin-5 receptor (IL5R) is composed of an IL5-specific alpha chain and a beta chain shared with the receptor for GM-CSF.r Cell 66(6): 1175-84.
Tietje, K. M., et al. (1990). "Cloning and functional analysis of a gene encoding a novel muscarinic acetylcholine receptor expressed in chick heart and brain.~ J Biol Chem 265(5): 2828-34.

.: ., . , :, .

WO96/04s57 PCT~S95/10182 ~gl~5 -. .~ . .
~ 164 -Tomlinson, I.M., Watler, G., Marks, J.D., Llewelyn, M.B. and Winter, G. (1992). The repertoire of human germline VH sequences reveal~s about fifty groups of VH segments with different hypervariable loop." J. Mol.
Biol. 227:776-798.
T~ l;nqnn, I.M., Cook, G.P., Carter, N.P., Elaswarapu, R., Smith, S., Walter, G., Buluwela, L., Rabbits, T.H. and Winter, G. (1994). ~The human immunoglobulin VH and D segments on chro~n~ s 15qll.2 and 16pll.2. Human Mol. ~en. 3:853-860.
I~ Tulip, W.R., Harley, V.R., Webster, R.G., Novotny, J. (1994). N9 Neuraminidase Complexes with ~nt;ho~ies NC41 and NC10: Empirical Free-Energy Calculations Capture Specificity Trends Observed with Mutant Binding Data. ~. T3iochemistry, 33:7986-97.
Ullrich, A., Riedel, HJ., Yarden, Y., Coussens, L., Gray, A., Dull, T., Schlessinger, J., Waterfield, M.D., Parker, P.J. (1986). "Protein kinases in cellular signal transduction: tyrosine kinase growth $actor_=
receptors and protein kinase C. Cold Spring Harb. Symp.
Quant }3iol. 2:713-724.
Ullrich, A. and Schlessinger, J. (1990) Cell 61:203-212.
van Koppen, C. J., et al. (1993). ~Isolation, sequence and $linnt;on~l expression of the mouse m4 muscarinic acetylcholine receptor g~ne.~ Biochim }3iophys Acta 1173(3): 342-4.
van Koppen, C. J. and N M. Nathanson (1990).
"Site-directed mutagenesis of the m2 muscarinic acetylcholine receptor. Analysis of the role of N-glycosylation in receptor~expression and function.~ ~.
Piol. Chem. 265(34): 20887-92.
Vasudevan, S., et al. (1991). "Expression and cell membrane localization of rat M3 muscarinic acetylcholine receptor produced in Sf9 insect cells using the baculovirus system.~ ~ebs Lett 283(1): 52-6.

2l9~67~
W096l04~7 PCT~595/10182 ' 3 ~

Ward, E.S., Gussow, D., Griffiths, A.D., Jones, P.T. and Winter, G. (1989). ~Binding activities of a repertoire of :single immunoglobulin variable domains - secreted from Escherichia coli.~ Nature 351:554-546.
Waterhouse, P., et al. ~1993). ~Combinatorial infection and in vivo recombination: a strategy for making large phage antibody repertoires.n Nucleic Acids Research 21(9): 2265-6.
Wang, H. M., et al. (1992). Structure of mouse ~n~rl~nk;n 3 (IL-3) binding protein (AIC2A). Amino acid residues critical for Ih-3 binding.~ J. Biol. Chem.
267(2): 979-83.
Webster, D. M., et al. (1994). Antibody-antigen interactiQns [Review].~ Current Biology 4: 123-129.
Weiss, A. (1993) Cell 73:209-212.
Weis3, J., et al. (l990). "Delineation of Muscarinic Receptor Domains ConferrirLg Selectivity of Coupling to Guanine Nucleotide-Binding Proteins and Second Messengers. r MQ1 . Pharmacol . 38: 517-523.
Wheatley, M., et al. (1986). Proc. 6th International Symp. Cell and Molec. Bab7iL7 Cholinergic Function:
Williams, S.C., and Winter, G. (1993). ~Cloning and sequencing of human i Inngl nhlll i n V lambda gene segments.~ Eur J. Immunol . 23:1456-1461.
Winter, G., Griffiths, A.D., Xawkings, R.~., and Xoogenboom, H.R. (1994). ~Making Antibodies by phage display technology.~ Annu. Rev. IDunol . 12:433-455.
Winter, G. and Milstein C. (1991). "Man-made antibodies." ~ature 349:293-299.
Witthuhn, B. A., et al. (1993). ~JAK2 associates with the erythropoietin receptor and is tyrosine phosphorylated and activated following stimulation with erythropoietin.~ Cell 74(2): 227-36.
Wu, T. T., et al. (1993). rLength Distribution of CRDX3 in Antibodies . ~ Protein~: Structure, Function and Genetics 16: 1-7.

.

W096/04ss7 PCT~S95/10182 7 g - 166 - = -Yarmolinsky, M. B., et al. (1989).
"Participation of the lytic~replicon in bacteriophage P1 plasmid maintenance ~ ~our~a1 of Bacteriology 171(9):
4785-91.
Yatani, A., et al. (1988). "The G protein-gated atrial K+ channel is stimulated by three distinct Gi alpha-subunits.~ Nature 336(6200): 680-Z.
Yatani, A., et al. (1990). "ras p21 and GAP
inhibit coupling of muscarinic receptors to atrial X+
rh~nn~l~,~ Cell 61(5): 769-76.
Yokota, T., et al. (1986). Proc.Natl.Acad.Sci.
USA 83: 5894-5898.
Zebedee, S.L., Barbas, C.F. 3d, Hom, Y.~., Caothien, R.H., Graff, R., DeGraw, J., Pyati, J., ~aPolla, R., Burton, D.R. and ~erner, R.A. (1992). "Human combinatorial antibody libraries to hepatitis B surface antigen." Proc. Natl. Acad. Sci. USA 89:3175-3179.

While we have hereinbefore described a number of embodiments of this invention, it is apparent that the basic constructions can be altered to provide other embodiments which utilize the methods and compositions of this invention. Therefore, it will be appreciated that the scope of this invention is defined by the claims appended hereto rather than by the specific ~mho~im~nts which have been presented hereinbefore by way of example.

Claims (75)

I CLAIM:

1. A method of identifying a ligand capable of binding to at least one determinant of a biologically active site on a target, which determinant participates in conferring biological activity of said target, the method comprising:
a) providing at least one reporter antibody to be used as a reporter of binding of said ligand to the biologically active site, and wherein said antibody is selected from an antibody library of sufficient diversity to possess at least one antibody member capable of binding to at least one determinant in the biologically active site as determined by the ability of said antibody member, either alone or in combination with at least one other ligand, to possess agonist or antagonist activity;
b) identifying as potential ligands for activity at the target, those ligands which are capable of competing with the reporter antibody for binding to the target.

2. The method according to claim 1 wherein the reporter antibodies are members of a recombinant library wherein each antibody member (rVab) of the recombinant library comprises at least one variable region selected from the group consisting of VH and VL regions, and optionally comprising a constant domain attached by its amino terminus to the variable region.

3. The method according to claim 2 wherein the rVab unit is displayed on the surface of a carrier.

4. The method according to claim 2 wherein the rVab unit is soluble.

5. The method according to claim 3 wherein the carrier is a bacteria.

6. The method according to claim 3 wherein the carrier is a bacteriophage.

7. The method according to claim 2 wherein a parental VL region comprising at least one CDR is used to derive the VL region of the rVab by deleting, inserting or substituting at least one amino acid within at least one CDR.

8. The method according to claim 2 wherein a parental VH region comprising at least one CDR is used to derive the VH region of the rVab by deleting, inserting or substituting at least one amino acid within at least one CDR.

9. The method according to claim 2 wherein parental VL and VH regions comprising at least one CDR, are used to derive a pair of VL and VH regions of a rVab by deleting, inserting or substituting at least one amino acid within at least one CDR of each variable region.

10. The method according to any one of claims 7, 8 or 9 wherein the crystal structure of the parental V
regions used to derive rVab are known.

11. The method according to claim 9 wherein the crystal structure of the parental VH and VL pair used to derive the rVab is known.

12. The method according to claim 2 wherein at least one of the parental V regions used to derive rVab is unmodified.

13. The method according to claim 2 wherein the crystal structure of the rVab is determined after isolation as a rVab which binds to a biologically active site on the target.

14. The method according to claim 2 wherein at least two V regions are modified by deleting, inserting or substituting at least one amino acid in at least one CDR
after isolation as rVab which binds to a biologically active site on the target.

15. The method according to claim 1 wherein the target is a polypeptide, protein, nucleic acid, oligosaccharide, carbohydrate or lipid.

16. The method according to claim 1 wherein activity of the target is coupled to an assayable biochemical response at the target which biochemical response functions as a signal of target activation.

17. The method according to claim 16 wherein the biochemical response is detectable as a change in a protein or polypeptide characteristic.

18. The method according to claim 16 wherein the biochemical response is associated with an organometallic moiety, a metal or other nonprotein.

19. The method according to claim 16 wherein the biochemical response is associated with a portion of the bioactive structure.

20. The method according to claim 16 wherein the biochemical response comprises a detectable free radical, fluorescent or chemiluminsecent group, radioactive isotope or involves oligomerization.

21. The method according to claim 16 wherein the biochemical response is phosphorylation and the signal is a change in the phosphorylation state of the target.

22. The method according to claim 17 wherein the signal protein is a G protein and the signal is a change in either the prepense of a G protein regulatory agent or the binding of rVab due to the presence of a G
protein regulatory agent.

23. The method according to claim 16 wherein the signal is a change in the binding of rVab to its binding site.

24. The method according to claim 2 wherein the recombinant antibody comprises a single polypeptide chain comprising a VH functionally coupled to a VL to produce a binding site.

25. A method of identifying ligands capable of binding to at least two determinants which together are required for biological activity of a pharmacological target, the method comprising:
a) screening and isolating from an rVab library, rVab members comprising at least one VH and VL
regions, and optionally comprising a constant domain attached by its amino terminus to the V region, and capable of binding to at least one of the determinants of the pharmacological target;
b) making and expressing an rVab-peptide (rVab-PEP) library comprising the isolated rVab members coupled to at least one peptide comprised of a random sequence of amino acids;
c) screening the rVab-PEP library for first rVab-Pep members which bind and activate the pharmacological target wherein the rVab component binds to a first determinant of the pharmacological target and the peptide component binds to a second determinant of the pharmacological target;
d) screening the rVab-Pep library and identifying a second rVab-pep member capable of actively binding to the pharmacological target, and wherein the rVab component binds to a third determinant of the pharmacological target and the peptide component binds to fourth determinant of the pharmacological target.

26. The method according to claim 25 wherein the rVab component of the second rVab-Pep member competes with the peptide component of the first rVab-Pep member for binding to a determinant on the pharmacological target.

27. The method according to claim 25 wherein the rVab component of the first rVab-Pep member competes with the peptide component of the second rVab-Pep member for binding to a determinant on the pharmacological target.

28. The method according to claim 25 wherein the first determinant of the pharmacological target is the same as the fourth determinant, and wherein the second determinant of the pharmacological target is the same as the third determinant.

29. The method according to claim 25 wherein the rVab component used to construct the rVab-Pep has at least one other attribute of an active ligand, besides affinity for the target, and wherein the attribute is selected from selectivity and biological activity.

30. The method according to claim 29 wherein rVabs which bind to determinants of active sites are identified by their ability to competitively or allosterically alter the binding on an endogenous ligand.

31. The method according to claim 25 wherein the active rVab-Pep possess agonist or antagonist activity.

32. The method according to claim 31 wherein activity of the target is coupled to an assayable biochemical response at the target which biochemical response functions as a signal of target activation.

33. The method according to claim 32 wherein the biochemical response is detectable as a change in a protein or polypeptide characteristic.

34. The method according to claim 32 wherein the biochemical response is associated with an organometallic moiety, a metal or other nonprotein.

35. The method according to claim 32 wherein the biochemical response is associated with a portion of the bioactive structure.

36. The method according to claim 32 wherein the biochemical response comprises a detectable free radical, fluorescent or chemiluminsecent group, radioactive isotope or involves oligomerization.

37. The method according to claim 32 wherein the biochemical response is phosphorylation and the signal is a change in the phosphorylation state of the target.

38. The method according to claim 33 wherein the signal protein is a G protein and the signal is a change in either the prepense of a G protein regulatory agent or the binding of rVab due to the presence of a G
protein regulatory agent.

39. The method according to claim 32 wherein the signal is a change in the binding of rVab to its binding site.

40. The method according to claims 25 wherein the peptide component of the rVab-Pep members comprising VH and CL regions are expressed attached to either or both of the amino terminus of VH and the carboxy terminus of CL.

41. The method according claim 40 wherein the peptide component is attached to the amino terminus of the VH region.

42. The method according to claim 40 wherein the peptide component is attached to the carboxy terminus of the CL region.

43. The method according to claim 40 wherein two peptides are attached to the rVab component to form rVab-Pep2.

44. The method according to claim 40 wherein the peptide comprises between about 5 and 50 amino acids.

45. The method according to claim 44 wherein the peptide comprises between about 7 and 25 amino acids.

46. The method according to claim 45 wherein the peptide comprises about 8 amino acids.

47. A reporter of binding of a ligand to a determinant of a pharmacological target, which target requires binding of ligand to at least two determinants of said target to produce a biological response, said reporter comprising an rVab portion of an active rVab-Pep, and wherein said rVab component of said rVab-Pep binds to a first determinant of said target, and the peptide component binds to a second determinant of said target.

48. The reporter of claim 47 wherein the rVab comprises VH and CL regions and the peptide is expressed bound to either or both of the amino terminus of the VH
and the carboxy terminus of the CL.

49. The reporter according claim 48 wherein the peptide component is attached to the amino terminus of the VH region.

50. The reporter according to claim 47 wherein the peptide component is attached to the carboxy terminus of the CL region.

51. The method according to claim 47 wherein two peptides are attached to the rVab component to form rVab-Pep2.

52. The method according to claim 47 wherein the peptide comprises between about 5 and 50 amino acids.

53. The method according to claim 52 wherein the peptide comprises between about 7 and 25 amino acids.

54. The method according to claim 53 wherein the peptide comprises about 8 amino acids.

55. A method of identifying a ligand capable of binding to at least one determinant of a biologically active site on a target, which target requires activation of at least two determinants to express biological activity of said target, the method comprising:
a) providing at least one rVab reporter antibody according to claim 47 to be used as a reporter of binding of said ligand to the biologically active site, and wherein said antibody is selected from an antibody library of one antibody member capable of binding to at least one determinant in the biologically active site as determined by the ability of said antibody member, either alone or in combination with at least one other ligand, to possess agonist or antagonist activity;
b) identifying as potential ligands for activity at the target, those ligands which are capable of competing with the reporter antibody for binding to the target.

56. The method according to claim 55 wherein multiple ligands are identified which when bound together covalently, are capable of binding to the determinants necessary to cause a bioligical response of the target, the method comprising:

a) providing reporter rVab antibodies for each of the determinants for which ligands are to be identified;
b) for each of the rVab reporter antibodies, identifying as potential ligands for activity at each of the determinants of the target, those ligands which are capable of competing with each of the rVab reporter antibodies for binding to the target;
c) covalently linking the identified ligands so as to form active multivalent ligands capable of activating the pharmacological target

57. The method according to claim 56 wherein the identified ligands are non-protein organic molecules.

50. The method according to claim 56 wherein the two rVab reporter antibodies are used to identify two ligands which are combined to form the multivalent active ligand.

59. The method according to claim 56 wherein the pharmacological target is a polypeptide receptor.

60. A recombinant rVab antibody library comprising rVab members possessing at least one VL or VH
region derived from a parental variable region with at least one CDR which is diversified to form different rVab members by deleting, inserting or substituting at least one amino acid within at least one CDR.

61. The recombinant antibody library according to claim 60 wherein a parental VH region comprising at least one CDR is used to derive the VH region of the rVab members by deleting, inserting or substituting at least one amino acid within at least one CDR.

62. The recombinant antibody library according to claim 60 wherein parental VL and VH regions comprising at least one CDR, are used to derive a pair of VL and VH
regions of rVab members by deleting, inserting or substituting at least one amino acid within at least one CDR of each variable region.

63. The recombinant antibody library according to any one of claims 60, 61, or 62 wherein the crystal structure of the parental V regions used to derive rVab members are known.

64. The recombinant antibody library according to claim 60 wherein the crystal structure of the parental VH and VL pair used to derive the rVab members is known.

65. The recombinant antibody library according to claim 60 wherein at least one of the parental V regions used to derive rVab is unmodified.

66. The recombinant antibody library according to claim 60 wherein the CDR regions of a specific antibody are expressed on a plurality of frameworks which provides for variable geometric orientation of the CDR regions.

67. The recombinant antibody library according to claim 60 wherein the rVab members further comprise a peptide sequence covalently bound to the rVab members to form rVab-Pep members.

68 The recombinant antibody library according to claim 67 wherein the peptide component of the rVab-Pep members comprising VH and CL regions are expressed attached to either or both of the amino terminus of VH and the carboxy terminus of CL.

69. The recombinant antibody library according claim 68 wherein the peptide component is attached to the amino terminus of the VH region.

70. The recombinant antibody library according to claim 68 wherein the peptide component is attached to the carboxy terminus of the CL region.

71 The recombinant antibody library according to claim 68 wherein two peptides are attached to the rVab component to form rVab-Pep2.

72. The recombinant antibody library according to claim 68 wherein the peptide comprises between about 5 and 50 amino acids.

73. The recombinant antibody library according to claim 72 wherein the peptide comprises between about 7 and 25 amino acids.

74. The recombinant antibody library according to claim 73 wherein the peptide comprises about 8 amino acids.

75. A method of providing a model for a ligand capable of binding to a determinant of an active site of a pharmacological target, the method comprising:
a) providing at least two rVab identified as binding to an active surface of a pharmacological target;
b) identifying the regions of the rVabs that bind the biologically active site or individual inactive surface determinants of the bioactive structure, c) grouping the rVabs by overlapping structures which bind to common epitopes;
d) determining the relative spatial orientation, charge and energetics of the identified binding sites e) determining the molecular structure necessary to bind the target and confer activity.