CA2197232A1 - Method of producing single-chain fv molecules - Google Patents

Abstract

The invention relates to a method of producing single-chain Fv molecules in eukaryotic cells, and to secretable sFv proteins having at least one nonnaturally occurring glycosylation site. The single-chain Fv molecules produced by this method are biologically active and capable of being secreted from eukaryotic cells into the cell culture medium.

Description

~ W096/0s228 r~ 'e ~ MFT~nn OF 8K~U~lN~ SINGLE-CHAIN Fv MOLE~.F

Bach~vu-ld of the Invention Single-chain Fv (sFv) proteins are genetically engineered molecules that consist of the two variable domains of an antibody or T cell receptor cnnnected by a polypeptide linker and that contain the antigen binding function of the parental protein in a single 30 hD
polypeptide chain. (Huston, J.S., et al., Proc. Natl.
Acad. Sci. U.S.A. 85:5879-5883 ~1988); Bird, R.E., et al.
Science 242:423-426 (1988); Huston, J.S., et al., Meth.
Enzvmol. 203:46-88 (1991)).
The Fv portion of an antibody i5 the smallest f L _ _ - to bear the complete antigen-binding site. It is a 25 kD heterodimer consisting of the N-trrm;nAl variable (V) domains of the heavy (H) and light (L) chain.
(Inbar, D., et al., Proc. Natl. Acad. Sci. U.S.A.
69:2659-2662 (1972); Hochman, J., Biochemistrv 15:2706-2710 (1976); Hochman, J., et al., Bioch~m;~trv 12:1130-1135 (1973)). More recently, a g~n~tirAlly engineered single-chain Fv (sFv) with antigen binding activity has been produced by cnnn~ct;ns the C-t~nm;nl~R
of one V domain to the N-t~rm;nll~ of the other with a peptide linker. Huston, J.S., et al., Proc. Natl. Acad.
Sci U~s~A~ 85:5879-5883 (1988); and Bird, R.E., et al., Science 242:423-426 (1988) Since then, sFv proteins have been produced from a large number of different Ant;ho~;es (Huston, J.S., et al., Intern. Rev. Immunol. 10:195-217 (1993); Winter, G. and Milstein, C. Nature 349:293-299 (l991)) and initial studies (Kurucz, I., et al., Proc.
Natl. Acad. Sci. U.S.A. 90:3830-3834 (1993); Novotny, J., et al., Proc. Natl. Acad. Sci. USA 88:8646-8650 (1991);
Soo Hoo, W.F., et al., Proc. Natl. Acad. Sci. USA
89:4759-4763 (1992); Ward, E.S., J. Mol. Biol.

W09~05228 ~ 9 7232 224.885-890 (1992~ have ~k~ribed the production of sFv analogues of T cell receptors ~TcR), cell sur~ace molecules that are highly ~ 10gQU9 to immunoglobulins (Hedrlck, S.M., et al., ~ L~ 308:153-158 (1984); Davis, M.M. and Bjorkman, P.J., ~~~E_ 334, 395-402 (1988)).
Most sFv proteins have been generated in bacteria, often as insoluble, cytoplAr~ic inclusion bodies. Protein from inclusion bodies is not active and must be solubilized, renatured n vitro and n~; ~; 7Pd to form parent ~;cnlf;~P bonds, (~uston, J.S., et al. Methods Enzvmol. 203:46-78 (1991)) Alternatively the intro~nrt;rn of an ~-terminal leader se~uence can direct sFv into the periplasmic space of bacteria by a secretion process wherein the leader sequence is removed (Holland, I.B., et al., Methods Enzvmol. 182:132-143 (1990)) and protein folding is ac, ~l;chpd~ aided by enzymes that catalyze disulfide bond formation (Bardwell, J.C.A., et al., Cell 67:581-589-(1991~ and cis-trans isomerization of proline residues (~ayanD, T., et al., Biochemistrv 30:3041-3048 (l991)). ~owever, even with these enzymes, secreted sFv proteins 8~ t; - exist as insoluble aggregates in the periplasmic space, which must be solubilized and refolded i vitro (Johnson, S. and Bird, R.E., Methods FnzYmol. 203:88-98 (1991); George, A.J.T., et al., J. Immunol. 152, in press (1994)). Xnappik, A., et al. (Bio/Technoloqy 11:77-83 (1993)) have recently attempted to uV~l~ this problem by overexpressing protein disulfide isomerase and prolyl cis-trans isomerase in the piroplasm of bacteria. ~owever, neither enzyme induced a significant change in folding rff;r;Pnry of sFv proteins when expressed either alone or together with the other enzyme.
A different approach to produce active sFv would be to use the more sophisticated rpfrl~;ng r-rh;nPry that is 35 located in the endoplasmic reticulum (ER) of l; ~n ~ w096/o5~ 2 1 9 7 2 3 2 = ~ ,J~ ~

cells. The potential benefit of this ~L~-dl could be ~ substantial, since the ER not only co~;nq enzymes that catalyze specific isomerization steps but it also rnn~;nc a number of proteins (e.g., chaperones) that aid in the folding process and prevent the secretion of inc~L,e~Lly folded proteins (Gething, M.J. and Sambrook, J~ E~ 355 33-45 (1992); Pelham, ~.R., Annu. Rev. Cell ~içl- 5:1-23 (1989); Hurtley, S.M. and Helenius, A., Annu Rev. Cell Biol. 5:277-307 (1989)). A number of sFv fusion proteins have been expressed in or on the surface of l; ~n cells. Examples include an anti-HIV-KDEL
fusion protein, or anti-HIV sFv alone, that remains bound in the ER (Marasco, W.A., et al. Proc. Natl. Acad. Sci.
g~A 90:7889-7893 (1993)) and anti-tumor sFv proteins fused to TcR-~ or Fc(RI-~ that trigger cell mediated cytolysis (Eshhar, Z., et al. Proc. Natl. Acad. Sci. USA
90:720-724 (1993); Hwu, P., et al. J. EXD. Med.
178:361-366 (1993); Stancovski, I., et al. J. Immunol.
151:6577-6582 (1993)). However, production-of this class of proteins by , li~n cells is generally very low, varying from a few mi~yL~ a to a few milligrams, if production is possible at all. (Davis, S.J., et al., J.
Biol.Chem, 265:10410-10418 (1990); Tr~nn~rk~r, A. et a]., EMBO J., 10:3655-3659 (1991)).
To date, it is not certain what the rate-limiting step, or steps, in the efficient expression and secretion of sFv proteins in ~r~ n cells may be, nor has it been apparent how to induce, or increase existing production levels.
Summary of the Invention The present invention relates to a method of producing single-chain Fv molecules in l; ~n cells and single-chain Fv molecules produced by this method.
The single-chain Fv molecules produced by this method are W096/0s228 capable of being secreted from , 1; An cellg into the cell culture medium, thus greatly facilitating their ;~olAt;r,n and pnr;~;rpt;on. Importantly, the single-chain Fv molecules produced by this method are secreted as correctly folded, biologically active binding molecules capable of reacting with their respective ligands, thus eliminating the need of further n vitro manipulation to remove bacterial endotoxin and to further purify or to refold these molecules to recover biological activity.
The method described herein is based on the finding that, in the secretion of single-chain Fv molecules from r l; An cells, their exit from the endoplasmic reticulum can be rate-limiting and that glycosylation of the single-chain Fv molecules can enhance the rate of secretion.
More specifically, the parent nucleic acid sequence ~nC~;ng a gingle-chain Fv molecule is modified by oligon-lclrr~tide-directed mutagenesis of the coding sequence to include one, or more, non-naturally occurring glycosylation site, or sites. As used herein, the term single-chain Fv r-lecnlP includes novel analogs of the single-chain Fv. These sFv analogs include, for example, the sFv' and the (sFv')2 molecules wherein a cystine-r~ntA;n;ng peptide is fused to the sFv carboxy terminus.Another example ; nrl ~l~r~ a BiBABS molecule (V~-VL-V~-VL) wherein two sFv ~lerll1~ are linked together.
Descriptions of additional single-chain Fv molecules rnl -~sed by this invention are found in Huston, J.S., et Al. Cell Bio~hy8ic8.. 24 in press (1994), the tr~rh;ngR of which are incorporated herein by reference.
Also ~n~ cqed ~y this invention are chimeric multivalent protein analogs described in W0 93/23537, the t~Pr~;nr,~ of which are also incorporated herein by reference, and antibody fl _ ~ such as Fv, Fab and os6lo5228 ~ JI

Fab' fragments (Better, M. and Howitz A.H., Enzvmol.
~ 178:476-796 (1989)). As used herein, the term non-naturally occurring glycosylation Eite means a glycosylation site not encoded by the parent single-chain Fv nucleic acid sequence. The novel glycosylation site, or sites, are incorporated into the parent single-chain Fv nucleic acid sequence at an appropriate amino acid residue, or residues cnnt~;nPd within the sequence.
Preferably, the novel glycosylation site, or sites, are either N-linked (asparagine-linked) or 0-linked (serine-or threonine-linked) glycosylation sites. The parent sFv coding sequence is ~ ';f;ed in such a manner (e.g., by insertion, deletion, or substitution of nucleotides) so as to result in the consensus amino acid sequence Asn-X-Ser/Thr, which leads to N-linked glycosylation, or Ser/Thr, which leads to 0-linked glycosylation. In a preferred ~ '-lir ~, the novel glycosylation site(s) is(are) located in a region of the sFv protein product that is not buried within the folded protein (e.g., exposed on the protein surface at a ~-turn, in a loop, or within a linker sequence).
The modified sFv nucleic acid construct (also referred to herein as the sFv construct) is introduced into a vector capable of expressing the glycosylated sFv protein construct in a eukaryotic cell. In a preferred embodiment, the eukaryotic cell is a l; ~n cell. The term vector, as used herein means any nucleic acid sequence comprising a nucleic acid sequence of interest, competent to be incorporated into a eukaryotic host cell resulting in the expression of the nucleic acid sequence of interest. Vectors can include, for example, linear nucleic acid sequences, plasmids, cosmids, phagemids, and extrachromosomal DNA. Spp~;f;c~llyl the vector can be a recombinant DNA vector. Also as used herein, the term expression, or gene expression, is meant to refer to the W096105228 2~ ~ 7~ r~ J~1~,~8 --,;
production of the protein $Toduct-of the n~cleic acid sequence of interest, ;nclll~;ng transcription of the DNA
and translation of the RNA transcript. The eukaryotic host cell can be any , liAr cell capable of expressing protein, including for example, immortalized 1; ~n cells such as COS-7 cells, 293 cells, myeloma, Chinese hamster ovary (CHO) cells. The host cells can also be cultured yeast cells.
The vector is transfected into the eukaryotic cell, for example, by calcium rh~srhAte precipitation, and the transfected cell is ~-intAin~ under conditions sufficient for propAgAtinn of the cells, expression of the sFv construct within the cell and secretion of the sFv protein product into the cell culture medium. For lS example, if a , liAn cell is the transfected host cell, the cell is cultured in suitable culture medium and under an ai _,~~~e conducive for growth of the cell. As the host cell grows~ ~he ~e~tcr i~tegrates into the host cell genome ana e~Leases ~he 8P~ construct within the host cell resulting in the sFv protein product.
Importantly, this sFv protein product ~ntA;r~ at least one novel, engineered glycosylation site that was not present in the parent sFv molecule. This new glycosylation site signals the AttA ' of 2S oligosaccharide (~ ohydl~te) chains to the sFv protein, which takes place within the endoplasmic reticulum (ER) of the eukaryotic cell. Under the conditions described herein, glycosylated sFv proteins are secreted from the ER at an increased rate relative to the rate of Eecretion of parent sFv proteins. Importantly, the secretable, glycosylated sFv molecules ~e~r;h~ herein, typically exhibit hi~l~gir~l acti~ity and properties which have not been previously accessible through bacterial (prokaryotic) expressian or secretion. Additionally, glycosylation of CDRs in some ~nt;ho~;es can be important _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~ W096~5228 ~ P~

to improve binding, especially to~carbohydr~te antigens ~ ~Wright, A. and Morrison, W., in SPRING SEMINARS IN
IMMUNOLOGY 15:259-273 (1993)),. Thus specific or random ~ intro~ tinn of glycosylation sites into CDRs of sFv proteins, as described by the present invention can be of value.
The present invention further relates to modified secretable sFv proteins having one, or more, non-naturally occurring glycosylation site(s), and to the DNA
sequences encodins these proteins. More specifically, these proteins have at least one N-linked or O-linked glycosylation site that is not encoded by the parent sFv protein. These modifications are also referred to herein as post-translational modifications. Although these sFv proteins are modified to contain non-naturally occurring glycosylation sites, they retain the same specificity of binding as exhibited by the parent, unglycosylated sFv protein.
Thus, as a result of the method described herein, post-translationally modified sFv molecules capable of specifically binding ligand can be successfully produced in and secreted from eukaryotic cells.

Brief Descri~tion of the Drawinqs Figure 1 is a diayL ~;c representation of the sFv gene constructs.
Figure 2A shows the nucleic acid sequence (SEQ ID
NO: 14) an encoded amino sequence (SEQ ID NO: 15) of the V~ region of U7.6 sFv.
Figure 2s shows the nucleic acid sequence (SEQ ID
NO: 16) and the encoded amino acid sequence (SEQ ID NO:
17) of the V~ region of U7.6 sFv.

W096/05228 2 1 ~ ~ 2 ~ 2 . ~u~

Figure 3A is an electrophoretic gel showing the distribution of sFv in COS-7 cells and culture gnr~rnAtAnt at different times in a pulse chase experiment.
Figure 33 is a graphic ~ se~Lation showing the rate of secretion of the different sFv molecules.
Figure 4 is an electrophoretic gel showing the processing of N-linked ca,bGhyd,dte during 8Fv secretion.
Figure 5A is an electrophoretic gel showing the presence of ~7.6 sFv mutants in cells and supernatant in a pulse chase experiment.
Figure 5B is a graphic L~es~,ltation showing the densitometric analysis of the data shown in Figure 5A.
sFv present in the sup~rn~tAnt is plotted as percent of total i ~L~cipitation material.
Figure 6 is an electrophoretic gel showing the preferential secretion of glycosylated forms of sFv proteins from tunicamycin-treated cells.
Figure 7 18 an electrophoretic gel showing that secreted sFv proteins specifically bind their antigen.
Figure 8 is a graphic repres~ntAtinn showing the results of experiments demonstrating the inhibition of binding of ~7.6 sFv mutant proteins by DNP-hapten.
Figure 9 is a graphic repr~ntAt;~n showing the results of experiments demonstrating that OKT9-sFvs produced in 1i An cells and bacteria dissociate from K562 cells with similar rates.

Detailed De8cription of the Invention The invention described herein relates to a method of producing single-chain Fv ~olec~ in eukaryotic cells and single-chain Fv molecules produced by this method. ~perif;~Ally, the invention relates to a method of producing single-chain Fv proteins that are capable of 35 being secreted from 1; An cells directly into the 096/05Z~ e cell culture medium. Importantly, these single-chain Fv ~ proteins are secreted as correctly folded, biolog;r~l1y active binding molecules, capable of binding ligand with ~ sper;f;city.
The method described herein is based on the finding that, in the secretion of single-chain Fv proteins from ~ n cellg, their exit from the endoplasmic reticulum can be rate-limiting and that glycosylation of the single-chain Fv protein can enhance rates of secretion.
A wide variety of proteins are secreted by vertebrate cells. In fact, some cells are highly sper~l;7ed to secrete specific proteins, such as B-lymphocytes that secrete ; ~ hl1l ;n~. R;hos ~ that synthesize secreted proteins are bound to the endoplasmic reticulum (ER). After synthesis these proteins are translocated into the lumen of the ER and then move in small transport vesicles through the Golgi complex and eventually exit the cell. This entire process is often termed protein maturation.
Specific maturation steps known to occur in the ER
include proteolytic cleavage of leader sequences, addition and modification of carbohydrate residues, formation of disulfide bonds, and the folding of the nascent polypeptide chain into its correct three-~; ~;rn~l structure. The first two processes occur very rapidly with all proteins, and for example, with antibodies, the formation of disulfide bonds occurs as the peptide passes into the lumen of the ER.
(sergman, ~.W. and Kuehl, W.M., J. Biol. Chem.
254:8869-8876 (1979)).
ln the case of antibodies, nascent heavy (H) and light (B) chains are known to bind to chaperones, proteins resident in the ER that facilitate the folding and assembly of H and L chains into functional W096/05228 ~3 ~72 -io-~n~iho~;~q Both chai~ns~aggociate with heavy chain binding protein ~BiP or GRP78) (Knittler, M.R. and Haas, I.G. MBO J. 11:1573-1581 (1992)) and GRP94 (Melnick, J., et a]. J. Biol. Chem. 267:21303-21306 (1992)) during folding and assembly, and IP90, another putative chaperone, interacts with partial complexes of membrane immunoglobulin in the ER of B-cells (Hochstenbach, F., et al. Proc. Natl~ Acad. Sci. U.S.A. 89-4734-4738 (1992)).
Once ~n~;h~;es have assumed their proper configuration they dissociate from the chaperones and proceed through the Golgi on~their way to being secreted. In the case of 8FY proteins, little is known about protein folding. One recent report discussed a non-secretable sFv (i.e., an sFv which was produced but not secreted by , l; ~n cells) that interacted with BiP. (Marasco, W.A., Haseltine, et al . Proc. Natl. Acad. Sci. ~SA 90:7889-7893 (1393)). Another report discussed that the V~ domain contributes to the-hindln3 of ; ~ h~ n heavy chains to BiP. (Pollak, B_a., et al. Proc. Natl. ~cad. Sci.
~ - 84:9199-9203 (1987)).
r.oL~v~r, many antibodies, other secreted proteins and most cell surface proteins are glycosylated. That is, they have oligsaccharides covalently linked to amino acid residues. (Machamer, C.E. and Rose, J.K., J. Biol.
Chem. 263:5948-5954 (1988)). Many potential functions have been suggested for these oligsaccharides, including assistance with polypeptide folding, prevention of intracellular aggregation, protection from proteolytic breakdown and signals for intracellular targeting and cellular recognition. (Olden, X. et al., Biochem Bio~hYs. Acta 650:209-232 (1982)). As described herein, it is now demonstrated that l; ~n cells transfected with different sFv genes secrete the corr~qpAn~;ng active sFv molecules at different rates and that glycosylation can affect these secretion rates.

2 ~ 9 773.~
~ W096~s228 .~,/~

A nucleic acid sequence ~nro~;ng a 6ing1e-chain Fv (sFv) protein can be modified to include one, or more non-naturally occurring N-linked, or O-linked ~ glycosylation site(s). The I -';f1ed nucleic acid sequence Pnro~;nS the sFv protein is referred to herein as the parent sFv nucleic acid sequence. As used herein, the term non-naturally occurring glycosylation site means a glycosylation site that is not encoded for in the parent nucleic acid sequence. Thus, the modified sFv protein rnnt~;nr at least one glycosylation site that is not encoded by the parent nucleic acid sequence.
sFv proteins suitable for modification by the method described herein, include single-chain antibody proteins, such as U7.6, other Ig superfamily analogues, such as the T-cell receptor protein and chimeric derivatives of these aFv proteins. A ~t~iled description of sFv molecules is found in U.S. Patents Nos. 5,091,513, issued February 25, 1992, and 5,132,405, issued July 21, 1392, the t~rh;nga of which are incorporated herein by reference. A
description of chimeric single-chain protein analogues is found in International Patent Application, W0 93/23537, the t~rh; ng~ of which are incuL~uLwted by reference.
References to nucleic acid sequences and constructions of sFv molecules are described, for example, in Huston, J.S., et al., Intern. Rev. Immunol. 10:195-217 (1993).
The sFv proteins ~nr ,~r~ by this invention also include sFv fusion proteins where effector domains are fused to either chain terminus of the sFv, as described, for example, in Huston, J.S., et al., Meth. Enzvmol.
203:46-88 (1991).
Nucleic acid sequences encoding single-chain Fv proteins, Ig superfamily analogues, chimeric proteins, sFv fusion proteins and other sFv protein species can be modified, as described in Example I, by oligonucleotide-directed mutagenesis of the coding sequence to include W096/052~ 2~ 3~ e ~

one, or more, glycosylati~n site(s) that are not present in the parent secIuence. The ~ nucleic acid ser~ nre encoding an sFv protein r~nt~;n;ng at least one novel gylcosylation site that was not encoded by the parent nucleic acid sequence is referred to herein aa the modified sPv nucleic acid seriuencel or sFv construct.
N-linked (asparagine-linked) sites are the preferred sites of glycosylation contemplated by this invention.
The parent sFv coding secIuence is altered in such a manner (e.g., by addition, deletion, or substitution of nucleotides) 80 as to ~esult in a nucleic acid ser~uence that encodes the consen8us amino acid secIuence Asn-X-Ser/Thr. This consensus secluence signals an N-linked glycosylation site ~owever, O-linked glycosylation sites are also , ~ cced by this invention. If an O-linked glyco8ylation site is inserted, the parent sFv nucleic acid secIuence is al~ered sb as to result in a modified nucleic acid ser1uence t~at encodes the amino acid se~uence Ser/Thr. This consensu~ sequence signals an o-linked glycosylation site.
The selection of the novel glycosylation site, or sites, is based on the c~ ' ln~t;~n of the primary nucleic acid ser~uence onrr,~ ng the sFv molecule and the local tertiary structure in the parent protein, which can, for example, represent a ~-turn or loop structure. (Aubert et al, Arch. Biochem. Bio~hvs. 175:~10 (1976)). The novel glycosylation site(s) is(are) located in a region of the protein that is not expected to be buried within the fol~ed protein, nor ser~uestered at the VH_V~ interface.
That is, the novel glycosylation site i6 attached to an amino acid residue that is exposed on the protein surface. For example, the presence of an N-linked glycosylation site in the first f~ c k region of VH cf two sFvs (OKT9 Ab-sPv, and ~7.6Ab-sFv) increases their ~ w096/05x~ ~ 9y~3~

rate of secretion without sir~nif;c~nt1y altering their - antigen binding affinities. If more than one glycosylation site is added, the sites can be constructed ~ to be adjacent to each other, (e.g., attached to adjacent amino acid residues located within the amino acid sequence) or they can be interspersed at various/random positions within the sP~lPrre (e.g., attached to non-adjacent amino acid residues of the sequence).
The modified sFv nucleic acid construct is then introduced into a vector capable of expressing the modified construct in a host cell as described in Example 2. Such vectors, especially rec ';n~nt DNA vectors, are well known to those skilled in the art. These vectors supply a promoter and other Pl c necessary to express the construct in eukaryotic cells. Specifically, plasmid vectors are rrnt 15ted for use in the present invention, which include both prokaryotic sequences and r 1;~n transcription units, such as described in ~OT.T.'.~TJT.~R CLONING: A L~30RATORY M~UAL, 2d Ed. Sambrook, J., et al., eds., Cold Spring Harbor Laboratory Press, NY, (1989).
The vector is then introduced into a host cell by methods known to those of skill in the art. Intro~llrt;~n of the vector into the host cell can be ~r l;chPd by any method that introduces the construct into the cell, ;nr1ll~;nrJ, for example, calcium phr~rh~te precipitation, microinjection, or ele~8lu~u~tion. See, e.g., Cockett et al., sio/technoloqv~ 8:662-667 (1990); CURRENT PROTOCOLS
IN M~T.T'~TTT.~R BIOLOGY, Ausubel, F.M., ed., John Wiley &
Sons, NY (1989)). For example, as described in Example 2, a , 1;~n cell is used as the host cell. The 1icn cell can be transfected using the calcium phosphate method. The transfected host cell is r~;nt~;nP~ under conditions sufficient for propagat;~n of the cell and expression of the sFv construct within the W09~052~ ~ ~7~32 r~

cell. Host cells can in~lude, for example, immortalized liAn cells 8uch as COS-7 cells, 293 cells, myeloma or Chinese hamster ovary (CH0) cell line8, and cultured yeast cells. Conditions sufficient for p~ n of the cell and expression of the sFv construct include any type of culture system and media known to those of skill in the art suitable for the propag~ti~n of the host cell.
Such culture systems include microcarrier or hollow fiber culture systems, as well as suspension systems, cultured under optimal conditions of media, A' ~~ph~re and temperature. As the host cell grows, the vector, for example, integrates into the host cell genome and expresses the sFv construct within the host cell resulting in the 8Fv protein product.
Importantly, the expressed sFv construct protein product ~ntA;nR at least one novel glycosylation site that is not encoded by the parent sFv sequence. Within the endoplasmic reticul~m (ER) of the l; An cell olig~ac~h~ride ~carbohydrate) chains are attached to the sFv protein at these glycosylation sites and the sFv protein is folded into its biologically active conformation. These correctly folded, glycosylated sFv proteins are then tran8ported from the ER to the cell surface and secreted from the cell. Key to this method is the fact that, under the conditions described herein, sFv molecules with one, or more, non-naturally occurring glycosylation~sites are secreted from the cell at a faster rate relative to the secretion rate of unglycosylated sFv molecules, and that these glycosylated sFv molecules retain their biological activity.
The secretion of sFv molecules into cell culture medium as biologically active, soluble proteins greatly facilitates their isolation and purification. Most current strategies used to produce sPv molecules have relied on bacterial expression of fusion proteins.

~ W096~s228 (Xuston, ~.S., et al., Meth. Enzymol. 203:46-88 (1991)).
sFv proteins produced in bacterial culture often contain bacterial endotoxin which contAm;nAt~ the protein and requires further purification procedures to produce clinically acceptable proteins. MJLe~y~rl if periplasmic secretion methods do not produce active sFv, fusion proteins may be secreted or expressed in bacterial cells as insoluble inclusion bodies which require further golnhi1;7~t;nn and refolding to obtain biologically active proteins. In most cases, denaturants such as SDS, urea, or gnAn;~;n~ HCl have to be used to extract the inclusion body sFv protein. Subsequently, the protein has to be renatured (i.e., correctly folded) to achieve biological activity and this may prove difficult, if not impossible. (PRINCIPLES OF GENE EXPRESSION, 4th Ed., Old, R.W. and Primrose, S.B., eds., Blackwell Scientific Publications, Cambridgel MA (1989)). In the case of binding proteins, e.g., a protein such as an ~Fv protein, folding to achieve the c~rrect three-~ c;nnAl conformation is a prerequisite of its biological activity.
In addition to correct folding to achieve biological activity, many different post-tranglational modifi~Atlnnc have been described as n~n~ccAry for active proteins.
(PRINCIPLES OF GENE EXPRESSION, 4th Ed., Old, R.W. and Primrose, S.B., eds., Blackwell Scientific Publications, Cambridge, MA (1989)). Besides N- and O-linked glycosylation, these modifications include phnsrhnrylationl acetylation, Am;~At;nn, gl~lphllrAt;nn, attachment of fatty acids and the formation of unu3ual amino acids. ~hese post-translational modifications are not known to occur in bacteria. Some glycosylation has been reported in yeast and in baculovirus cells, however, the composition and sequences of the resulting oligosaccharide chains can differ significantly from W096/05228 ~1 q 7 ~ 3 2 r~ ,6, :E -those found in ~ l;An cells. Thus, expression and secretion of sFv molecules in 1; ~n cells, as described herein, provides a rapid and efficient method of producing hirloq;c~lly active sFv molecules. The sFv protein ~duced by the described method can be ;~rl~tr~, purified and tested for biological activity using known laboratory terhn;~l~, such ag degcribed in r 1P~ 3 and 4. For example, if the sFv protein is an antibody sFv, the biological activity can be tested by ; ~precipation, ELISA, or r~;o; r~say.
Furthermore, testing biological activity of the sFv protein produced by the method described herein is greatly facilitated because the sFv protein is secreted in soluble form, directly into the cell culture medium.
It is simply a matter of obtaining an ali~uot of culture medium to test the activity of the sFv protein directly without further ~-n;p7l1~t;rn.
The present invention also relates to modified, secretable sFv proteins having one, or more, non-naturally occurring glycosylation site(s) and to the DNAsequences ~nco~ing these modified sFv proteins.
Additional advantages are achieved by producing glycosylated sFv molecules. Non-glycosylated proteins can be subject to aberrant aggregation such as that mediated by non-covalent ~or;~t;rn (McCoutney et ~1., Prote;n Enq., ~1994) or intermolecular disulfide bonding.
(Machamer, C.F. and Rose, J.K., J.Biol.Chem. 263:5955-5960 (1988)). For example, as protein ~ccllmnl~tes in the FR, cysteine residues in neighboring proteins can cross-link to form multimers. This results in large aggregatesof protein which can be ~;f~iclllt to disassociate in order to further isolate and purify the protein.
Alternate types of aggregation can involve non-covalent self-association, as for the first example, m~ ted by the bottom surface of an sFv, sFv analog or an antibody ~ 096/05228 2 1 9 7 2 3 2 r~ o~8 fragment, which is diametrically opposite to the antigen - , ' ;ninr~ site. Thig can result in an equilibrium mixture of monomers, dimers and even higher aggregates.
- Oligos~rrh~ride chains positioned at specific amino acid residues in the proteins can play an important role in the prevention of aggregation and non-covalent a880r; zlt i r,n In clinical applications, the presence of oligos~nh~ride chains can protect the protein from proteolytic degradation, resulting in increased circnlat;nr, half-life of the protein. (Goto, M, et al., 3io/Technoloqv, 6:67-71 (1988)). The presence of an oligosaccharide chain, term; n~t; ng in sialic acid, can also modify the n vivo biodistribution, phaLI~k~n~tics~
~l;m;n~t;nn and/or renal uptake of a single-chain Fv molecule. For example, if the terminal sialic acid is removed, the L~ ;n;nnJ carbohydrate chain residue can be a residue recogni~ed by hepatic receptors, thus signaling elimination. If a sialic acid residue is present, the signal for elimination can be blocked, again resulting in longer cirrnl~t;nr, half-life of the protein. Thus, the present invention provides a method of modifying the phaL k~npt; cs of sFv proteins in the body.
The att~' t of oligos~rrh~ride chains can also decrease the antigenicity of the sFv molecule. For example, the linker sequence used in an sFv molecule to connect V~ and VL could elicit an antigenic response to the sFv molecule. The addition of a glycosylation site within the linker can prevent the unwanted response, or decrease the severity of the response by masking the antigenic amino acid residues responsible for the response.

W096~5~ ~ 1 9 7 ~ 3 2 sFv Constrnrt~
To evaluate sFv pro~nrt;nn in ~ n cells, COS-7 cells were transfected with vectors ~nro~;nr~ four different sFv molecules; ~7.6-sFv directed against the hapten DNP, OKT9-sFv against the human TfR, 2Cll-sFv against the murine CD3-~ chain, and 2B4, a TcR-sFv that recognizes a cytochrome C peptide bound to I-E~. These sFv molecules are described in detail in Example 1.
Light chain leaders were used to direct newly-synthesized Ab-sFv proteins to the endoplasmic reticulum, and a c-myc peptide ser~ueuce was introduced at the 3~ end of each Ab-sFv construct to facilitate ~t~rt;rn of the proteirLs.
The three Ab-sFv proteins cnnt~;n~ a ~G~-S)Ilinker connecting the 3' end of Vrwith the 5' end of VE~ as shown in Figure 1. OKT9-sFv is exceptional in that it has an N-linked glycosylation site located at position 19 in FRl of VE. The U7.6 antibody also differs from the other Ab-sFv ~r~tein in that it has an extra cysteine in v~, located i~ C3~3, 3 amino acid residues C-t~m;n~l to the second domain-forming cys, as shown in Figure 2A and 2s. The TcR-sFv construct rnrt~;n~d a Va leader sequence, Va, and a linker consisting of 14 amino acid residues of CO followed by two additional residues resulting from a Pst I site. This joined Vp and 7 CO
2~ residues. The.~2B4 sFv has three glycosylation sites, one in Va, one in V~ and one in C~, and two additional cysteine residues in the third L~ .;J~ k portion of Vs.

Production ~n~ 8ecretion of sFv molecule8 Transfected COS-7 cells were labeled as described in detail in the Example 2, with [35S]-met for 60 min and then chased with unfolded methionine for either 0, 2, or 6 h. The sFv secretion products were then precipitated from both the cell lysates and the negative supernatants u~ing mouse antibody to the C-terminal myc peptide tag ~, ~1q7~32 = -and rabbit anti-mouse Ig bound to Protein A-Sepharose.~
Figure 3A and 3B shows the results of the production and secretion of sFv in COS-7 cells. Figure 3A shows the distribution of sFv in cells and culture supernatant at different times during a pulse chase experiment. COS-7 cells were trans~ected with plasmid DNA encoding the~
indicated sFv and 48 h later pu-lsed for 1 h with [35S]-methionine- Cells were subsequently chased for 0 ("P" in the figure), 2 or 6 h as indicated. The sFv proteins were immunoprecipita~ed, subjected to SDS-PAGE
under reducing conditions, and visualized by autoradiography. Figure 3B shows the rate of secretion of the different sFv molecules. Autoradiograms were ~uantified by densitometry. Total amounts of sFv in cell lysate and supernatant were determined and the percentage of total sFv present In the supernatant at each time point was calculated and plotted. Both OKT9 and 2Cll sFv proteins were secreted from the cells at similar rates, culminating in most sFv being present in the medium after a 6 h chase.
By contrast, the U7.6-eFv was secreted much more slo~ly, with only a minor amount being present in the medium a~ter the 6 h chase. The 2B~ TcR-sFv was handled .
differently by the COS cells than the Ab-sFv proteins, as most of this sFv could not be found in the supernatant, even after a 6 h chase. Data from three comparable independent experiments were guantified by densitometry and the percentages of total sFv in the supernatant were calculated and plotted (Figure 3B). The percentages of 30 sFv present in the sup~rnAtAnt at various times are --comparable for the 2C11 and OKT9-sFv proteins, resulting in a~out 70~ secretion after a 6 h chase. At this time only 25~ of the U7.6 sFv and 5~ of the 2B4-sFv were .
secreted in corresponding experiments.

AMEND~ S}!EET

~ W096r~s228 ~l'9 ~ 2 3 2 ~ ~"~ ~ t~ -E~;t from the ER is the rate limitin~ steD in OKT9-sFv 8ecretion The transfer of OKT9 and 234-sFv proteins from the ER to the Golgi complex was followed by measuring the acquisition of endo H resistance that is conferred on N-linked oligosaccharides by additional changes in glycosylation that occur in the Golgi. Transfected COS-7 cells were pulse labeled and chased. sFv and immunoprecipitates were either treated or not treated with endo H prior to SDS-PAGE analysis, as described in Example 3. Figure 4 shows that the major +raction of cell-associated OKT9 sFv L~ ; n~d sensitive to endo H at all time points, while OKT9 sFv present in the medium was always endo H resistant. Endo H resistance was used to determine the lo~Ali7~t;on of sFv in COS-7 cells at different times following pulge-lAhPl;ng.
T ~ ~cipitate8 were either not treated (-) or treated (+) with endo H prior to analysis by SDS-PAGE. Samples indicated by "Cells" represent immunoprecipitates from cell-associated material at the designated times; samples indicated "Supe" are ; nprecipitates from culture 8llr~nAtAnts 6 h after the pulse lAhPl;n~.
The fact that most cell-associated OKT9-sFv was endo H sensitive indicated that it had not yet passed through 2s the Golgi complex, and was therefore located in the ER.
As a control, OKT9-KDEL sFv, which r~ntA;n~ an ER
retention signal, mostly L~ ;n~ cell associated and endo H-sensitive at all time points (Figure 4). The small amount of sFv reaching the medium from these cells was 3~ endo H-sensitive, suggesting that it was released directly into the medium from the ER, most likely as a result of cell death. The 2B4 TcR-sFv behaved quite differently from the OKT9. This protein remained cell-associated and endo ~-sensitive at all time points (Figure 4), and was never secreted. Unlike the OKT9-KDEL

~ 096~S2~ 2 1 ~ 7 2 3 2 ~ ,3'.. If sFv, the glycosylation pattern of the 2B4-sFv changed with time.

Introduction of a 31ycosvlation-site in the U7.6 sFv enhances its secretion rate The U7.6 sFv was secreted much more slowly than the others antibody sFv constructs. In order to find the source of this defect, two mutations were introduced into the U7.6-sFv, as described in Example 1. First, the extra cysteine present in CDR 3 of VL (-CYS in Figure 5A and 5B) was removed, and secondly an N-linked glycosylation site was introduced in the same position as the glycosylation site in OKT9 (+Asn). Figure 5~ shows SDS-PAGE analysis of ; nprecipitates from cell lysates and culture supernatants from COS cells transfected with mutated U7.6 sFv proteins at different times during a pulse chase experiment. Figure 5B shows a densitometric analysis of the data shown in Figure 5A plotted as percent of immunoprecipitated material present in the culture supernatant.
A third construct incorporated both mutations (-Cys+Asn). COS-7 cells were transfected with the U7.6 mutants, and pulse chase experiments were performed. The data from three inde~endenL experiments were quantified and the percentages of total sFv in the sup~rnAtAnt were calculated and plotted in Figure 5B. The removal of the extra cysteine residue did not result in a change in secretion level. However, the introduction of a glycosylation site both in the presence (+Asn) and absence (-Cys+Asn) of the additional cysteine markedly ~nhAnre~ the rate of sFv secretion; after 6 hr, 20-25~ of the non-glycosylated sFv proteins were present in the medium, as compared with 50-60~ of the glycosylated proteins. Endo ~ experiments indicated that U7.6 +Asn ::~
and U7.6 -Cys+Asn were indeed glycosylated and that the w096/0~8 r~ 48 -rate limiting step in their secretion was the exit of the proteins from the ER, similar to OKT9-sFv.
To confirm the implication that glycosylation of the sEv proteins Pnh~n~e~ their rates of secretion, COS-7 cells transfected with the OKT9, U7.6+Asn and U7.6-Cys+Asn sFv constructs were treated with suboptimal c~n~Pntrationg of tunicamycin, as described in Example 2.
Matched plates of transfected COS-7 cells were pretreated for 2 h, pulsed for 1 h, and chased for 2 h, all in the presence of 3 pg/ml tunicamycin. The sFv proteins were immunoprP~;p;tated from cell lysates and analyzed by SDS-PAGE ; ~ rely after the pulse step (Cells/Pulse), and secreted sFv proteins were analyzed after a 2 h chase (Supe/Chase). On the right of Figure 6, the percentages of ; ~Lecipitated sFv proteins that were glycosylated are noted, as determined by densitometry, and based on the quotient of density of the hiqher Mr band divided by the sum of both bands~ densities.
As shown in Figure 6, after pulsing with 35S-methionine, the ; ~pnecipitated sFv proteins migrated as two bands corrPRp~n~;ng to glycosylated and non-glycosylated forms. After a 2 hr cha8e, there was an enrichment of the glycosylated, relative to the non-glycosylated band in supernatants from all three transfectants. Thus, glycosylation accelerates the rate of secretion of both ~7.6 and OKT9-sFv proteins.

Secreted antibodv sFv Rroteins sPecificallv bind antiqen To determine if secreted sFv proteins were active, the ability of radiolabeled material from the medium of transfected COS-7 cells to spe~;~;r~l1y bind antigen was tested as described in Example 4. Each radiolabeled sFv present in 6 h chase media was tested for antigen binding activity. For~OKT9 and 2C11-sFv proteins, cells bearing the relevant antigen were ;n~llh~tPd for 4 h at 4~C with ~ 096~5228 7 1 S 7 2 ~ 2 r ~

sFv in the presence or absence of an inhibiting antibody.
The cells were subsequently washed, lysed, and bound sFv ; nprecipitated and analyzed by SDS-PAGE under ~ reducing conditions. The reactivity of U7.6 sFv (wild type) was assessed by ;rrllh~t;ng DNP-Sepharose beads with the radiolabeled chase media for 16 h at 4~C in the absence or presence of 1 mM DNP-~-aminocaproate. The beads were subsequently washed, boiled in SDS-loading buffer, and eluted material was analyzed by SDS-PAGE
under reducing conditions and the bands were quantified by densitometry.
Figure 7 shows that all three sFv proteins were indeed active; OKT9-sFv bound to K562 cells, which express high levels of human TfR, binding was inhibited by intact OKT9 antibody, but not by W6/32, an antibody that recognizes MHC Class I molecules on these cells.
Similarly, 2C11 sFv bound to 2B4 cells, which are CD3', and the binding was totally inhibited by 2~11 whole antibody but only slightly inhibited by H57, an antibody that recognizes the ~ chain of the TcR. Finally, U7.6 and the three U7.6 sFv mutants bound to DNP-Sepharose beads in the absence, but not in the presence of inhibiting hapten.
Sequential depletion experiments were done to determine the percentages of secreted sFv that had antigen binding activity. After two sequential incubations with DNP beads, 87-96~ of the secreted U7.6-sFv constructs were specifically absorbed to the beads.
Similar experiments using multiple sequential ;n~nhat;ons of secreted OKT9-sFv with K562 cells indicated that the vast majority of this sFv was also functional, that is, specifically bound antigen.
To determine if the introduced mutations have any effect on the binding affinities of the U7.6-sFvs, the binding of 35S-meth;nn;n~ labeled U7.6-sFv constructs to w096~s228 ~ ;F -DNP-beads was inhibited with increasing amounts of DNP
hapten as described in Example 4. A shown in Figure 8, the greatest changes in affinity resulted from removal of CY8 91L: ;nh;h;t;nn of binding re~uired about lO fold less hapten for the -Cys mutants than for the +Cys, suggesting that the affinity for hapten increased about lO fold on removal of cys 91L. Introduction of a glycosylation site at position 19 in FR' of V~ in the +Cys mutants had no significant effect on binding activity, while it caused an approximately 4-fold decrease in affinity in the -Cys mutants. Thus, introduction of an N-linked glycosylation site had only a small effect on the binding activity of ~7.6-sFv.
The effect of glycosylation on the binding affinity of ORT9-sFv was assessed by comparing ~ o~; ~t; nn rates of bpnt~r;~lly produced (and therefore non-glycosylated) and refolded sFv with that of OKT9-sFv secreted from COS-7 cells. Figure 9 shows that the two products dissociate from K562 cells with very similar rates, suggesting that they have similar affinities for the TfR, as described in Example 5.
The amount of OKT9-sFv secreted by the tranafected CoS-7 cells was estimated by inhibiting its binding to the TfR with unlabeled ORT9 Fab, which binds with essentially the same affinity as the bacterially produced ORT9-sFv. By this assay, it i8 estimated that a COS cell sup~rn~t~nt cnnt~in~ about 0.17 ~g/ml of 35S-methionine labeled OKT9-sFv.
Thus, as described herein, transfected l;~n cells can secrete active antibody sFv at different rates, and the rate of secretion can be governed by glycosylation. In pulse chase experiments, the bulk of cell-associated Ab-sFv l ;nP~ in an endo H-sensitive form during a period when substantial amounts of sFv were 35 ~ ting in the medium in an endo H-resistant form.

~ ~7~
096/05228 P~ e~e This indicates that most of the cell-associated sFv had not yet passed through the medial Golgi, where carbohydrate modifications conferring endo ~ resistance occur (Dunphy, W.G. and Rothman, J.E. Cell 42:13-21 (1985); Tarentino, A.L., et al. J. Biol. Chem.
249:818-824 (1974)). Thus the exit of Ab-sFv from the ER
appears to be the rate limiting step in their secretion.
As a result of the work presented herein, soluble, secretable sFv molecules, including sFv fusion molecules, can be produced in 1; An cellg. A11 secreted antibody sFv specifically bound the ~ Liate antigen, and where tested, at least 90~ of the secreted sFv was fnn~tion~l. Therefore, , l;~n cellg can properly fold and secrete antibody sFv and the presence of oligosaccharide can enhance their rates of secretion.
These sFv molecules are biologically active and readily isolated and purified from cell culture medium.
The sFv molecules produced by the method described herein are useful in any procedure where intact immunoglobulins (IgG), frA~ IgG, analogous Ig superfamily sFv analogues, or chimeric derivatives are used. An sFv antibody produced by the method described herein can be used in a diagnostic i ~RR~y procedure to detect the presence of a specific protein which is ;n~;rnt;ve of a disease condition. For example, an sFv antibody specific for a tumor marker found in blood or urine can be used in an ELISA to screen patients for a particular type of cancer. As another example, an sFv receptor protein can be used in an assay to screen peptides for biological activity which enhance or inhibit receptor activity. In particular, due to their smaller size, these sFv molecules are useful as i vivo targeting agents. For example, the sFv molecule can be used to deliver effector molecules such as cellular toxins to targeted cells or to deliver radioisotope to tumor8 W096/05~ 2 1 9 7~3~ r~

(~uston, J.S., et al., Intern Rev. Immurol.. 10:195-219 (1993)).
The expression of sFv molecules in m l; ~n cells, as de3cribed herein, allows a rapid determination o~ the biological activity of a particular construct. This could be especially useful in the pro~ tion and testing of fusion proteins, where both the sFv and its fusion partner must be correctly folded. In many such cases, it may prove impossible to fold such constructs i vitro, while the folding m-ch;npry in the ER of l;~n cells can produce active sPv fusion proteins i vivo.
Additionally, the method described herein provides a rapid means of screening numerous sPv constructs, or provides the means to determine if a particular sFv construct encodes an active sFv protein (e.g., on a pilot scale). Once it has been determined that an active sFv protein is produced, other systems, such as stably transfected cell lines or transgenic animals or plants, can be used for large-scale prn~nt;nn of the protein.
The present invention will now be illustrated by the following , ~lPR, which will further and more spPc;f;c~lly illustrate the invention.

F le 1: aFv Constructs Constructs and primers used are shown in Figure 1.
A11 antibody sFv constructs contain a VL-linker-V~, L-chain leader sequences, and a C-tprT;n~l peptide tag (from the c-myc proto-oncogene). The TcR EFv construct rnnt~;nP~ a leader sequence to direct protein to the ER.
The following primers were used to produce the constructs:
Primer 1, tgttaactgctcact TCT ~r.~ ATG AGG ACC CCT
GCT CAG TTT CTT GGA ATC TTG TTG CTC TGG TTT CCA GGT ATC
A~A TGT GAC ATC AAG ATG ACC CAG TCT. (SEQ ID NO: 1) =

~ Wo ~/05~8 2 1 9 ~ 2 3 2 rcT~sgs/l0~8 Primer 2, atata GAA TTC CTC GAG ~r CTC TTA TTA ATT
CAG ATC CTC TTC TGA GAT GAG TTT TTG TTC TGA TAA AGC TTT
TGA GGA GAC TGT. (SEQ ID NO: 2) Primer 3, atata GAA TTC CTC GAG ~ CTC TTA TTA GAG
TTC GTC CTT TTC GCT ATT CAG ATC CTC TTC TGA GAT GAG TTT
TTG TTC TGA TAA AGC TTT TGA GGA GAC TGT. (SEQ ID NO: 3) Primer 4, tgttaactgctcact TCT AGA ATG AGG ACC CCT
GCT CAG TTT CTT GGA ATC TTG TTG CTC TGG TTT CCA GGT ATC
A~A TGT GAC GTC GTC ATG ACC CAG TCT CCA GCA.
(SEQ ID NO: 4) Primer 5, atata GGA TCC ATG AGG GCC CCT ACT GTC.
(SEQ ID NO: 5) Primer 6, atata GCG GCC GCC ACT CCC ACC TCC GCC AGA
ACC TCC GCC TCC TGA TCC GCC ACC TCC TTT GAT TTC CAG CTT
GGT GCC. (SEQ ID NO: 6) Primer 7, atata GGC GGC CGC GAG GTG QG CTG GTG GAG.
(SEQ ID NO: 7) Primer 8, atata CTC GAG TTA TTA ATT CAG ATC CTC TTC
TGA GAT GAG TTT TTG TTC TGA TGA GGA GAC GGT GAC CAT GGT.
(SEQ ID NO: 8) Primer 9, atata TCT AGA GAG AAG ACA ACC AGC GAT TGG
ACA GGG GCC ATG CAG AGG AAC CTG GGA GCT GTG CTG GGG ATT
CTG TGG GTG C~G ATT TGC TGG GTG AGA GGA GAT CAG GTG GAG
CAG AGT CCT TCA GCC. (SEQ ID NO: 9) Primer 10, atata ~ TCC TCA CTA AGT CAC ATT TCT CAG
ATC CTC. (SEQ ID NO: 10) In all primers, underlined sequences indicate restriction sites, lower case letters designate nucleotides added to facilitate cutting with restriction enzymes, and bold letters designate added sequences encoding for amino acid residues not present in the template.
U7.6 and OKT9 AbsFv proteins and the 2B4 TcR-sFv constructs have been described in (Kurucz, I., et al.
PrQc. Natl. Acad. Sci. U.S.A. 90:3830-3834 (1993);

W09~05z~ 21~7232 P~ s~

Nicholls, P.J., et al. J. Immunol. Method~ 165:81-91 (1993)). The nucleic acid sequence of U7.6 (SEQ ID NOs:
14 and 16) are shown in Figure 2A and 2B (V~ region and VL
region, respectively). Plasmids rnnt~;n;ng these constructs were used as polymerase chain reaction (PCR) templates. The OKT9 sFv construct was amplified using sense primer 1 that introduced an Xba I site and a light chain leader sequence and anti-sense primer 2 that introduced a c-myc peptide sp~tpnre~ two stop codons, and a Sac I site. To construct the OKT9+KDEL sFv sense primer 1 and an anti-sense primer 3 that is identical to primer 2 but adds an additional SEKDEL sequence at the C
terminal end were used. The PCR products were directly ligated into PCRIOOO (Invitrogen, San Diego, CA), according to r-mlf~cturers instructions, and snhcloned into the Xba I and Sac I site of pSVL (Pharmacia, Piscataway, NJ). The U7.6 sFv construct was amplified u3ing sense PCR-primer 4 that introduced an Xba I site and a light chain leadel sP~l~nre followed by an Aat II
site, and anti-sense primer 2.
Following the digestion of the purified PCR product with Xba I and Sac I the con8truct wa8 directly ligated into the Xba I and Sac I site of pSVL. The 2C11 sFv was constructed from cDNA using primers based on the 2C11 V
region sequences, kindly supplied by Dr. J. Yun Tso (Protein Design Labs, Mountain View, CA). Specifically primed first strand cDNA was synth~ized from mRNA
isolated from 2C11 cells (FastTrack Invitrogen, San Diego, CA) by primer extension using reverse transcriptase (Superscript, Gibco/BRL, Grand Island, NY).
The light chain cDNA was synthesized using anti-sense primer 6 that introduces a (G4S)3linker and a Not I site.
This cDNA was used for amplification of VL using anti-sense primer 6 and sense primer 5, that introduces a BamH I site. The heavy chain cDNA was synthesized using _ ~ W096~s228 2 1 ~ 7 2 3 2 ~ 0~48 anti-sense primer 8 that introduces a myc-peptide sequence and a Xho I site. This cDNA was used for amplification of the light chain using anti-sense primer 8 and sense primer 7 that introduces a Not I site at the 5' end. The PCR products were purified, blunted using T4 DNA polymerase (Boehringer M-nnh~;m Tn~; ~nApol; ~, IN) rhngphnrylated with T4 polynucleotide kinase (Gibco/BRL, Grand Island, NY) and sub-cloned into pcDNA/AMP
(Invitrogen, San Diego, CA) that had been cut with EcoR V
and treated with calf nlk~l ;n~ phosphatase (Boehringer M~nnhP;m) The light chain was excised with BamH I and Not I and the heavy chain with Not I and Xho I and both chains were simultaneously ligated into pcDNA/AMP
digested with BamH I and Xho I.
Cloned 2B4 TcR-sFv (Kurucz, I., et al. Proc. Natl.
Acad. Sci. U.S.A. 90:3830-3834 (1993)) was used as the PCR template for the TcR-sFv construct. The 2B4 sFv construct was amplified using sense primer 9 introducing an Xba I site (underlined), 30 bp of the ~-chain 5' non-coding region and the ~-chain leader sequence and anti-sense primer 10 that introduced a BamH I site. The PCR products were directly ligated into PCRIOO0 and El-h~lnnPd into the Xba I and BamH I sites of pSVL.

Site directed mutaqenesis of U7.6 Mutations in the U7.6 sFv constructs were introduced using a Transformer Mutagenesis Kit (Clonetech Laboratories Inc, Palo Alto, CA) according to the manufacturer's instructions except that ten times more plasmid was used than rP ~Pd. Colonies were screened using restriction enzyme digestion and plasmids from mutant clones were sequenced with a Sequenase Version 2.0 Kit (United States Biochemical, Cleveland, OH). Constructs were recloned to ~l;m;n~tP possible changes introduced into the vector during mutagenesis.

W096~5228 ? 1~3~ r~

The primers used were. U7.6 -CYS, C TGC CAG CAG TaC AGT
GGT TAC CCG, (SEQ ID NO: 11) introduced a tyrosine in place of cysteine 91 of VL (the original U7.6 VL clone r~nt~inP~ a tyrosine regidue at position 91; cysteine 91 was inadvertently introduced at a subsequent recloning step, probably as a PCR-induced ~t; on); U7.6+ASN, GGC
GCT TCA GTG AAt ATA TCC TGC AAG GC, (SEQ ID NO: 12) introduced an asparagine for lysine 19 ofV~. As a selection primer we used CCC TTT CGT CTT CAA Gtt TTC TCA
TGT TTG ACA GC (SEQ ID NO: 13) which removed an EcoR I
site from the vector. In the above primers, lower case letters designate the mutated nucleotides. The double mutant, U7.6 -CYS+ASN, was produced by using all three primers simult~nPo-lq1y ~le 2: Tr~n~fection and Met~holic T.~hPlina Cell ~ines and ~nt;hn~ies The following cell lines and monoclonal ~ntihoflies were used: COS-7 monkey kidney fibroblasts and K562 human erythroleukemia cells (American Type Culture Collection, Rockville, MD), 2B4 murine T hybridoma cells (Hedrick, S.M., et al., Cell, 30:141-152 (1982)), 145-2C11 hybridoma cells and mAb against murine CD3 ~ chain (Leo, 0., et al., Proc. Natl. Acad. Sci. ~SA 84:1374-1378 (1987)), OKT9 mAb against human transferrin receptor (TfR) (Goding, J.W. and Burns, G.F. J. Immunol.
127:1256-1258 (1981); S~nPi~pr, C., et al., J. Biol.
Chem. 257:8516-8522 (1982)), W6/32 mAb against human MHC
Class I molecules (Barnstable, C.J., et al., Cell, 3C 14:9-20 (1978)), H57 mAb against the C~ domain of murine TcRs (Kubo, R.T., et al., J. Immunol., 142:2736-2742 (1989)), 9EIO mAb against a c-myc peptide (Evan, G.I., et al., Mol. Cell ~3iol., 5:3610-3616 (1985)), and A2B4 mAb, which is specific for the 2B4 TcR (c l~n, L.E., et al., Proc. Natl. Acad. Sci. U.S.A. 80:6972-6976 (1983)).

21 q7232 096l05228 r~ 9 Polyclonal rabbit anti-~ouse IgG was from Cappel (Organon Technika, Durham, NC). COS-7 cells cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with penicillin (100 units/ml), streptomycin (100 units/ml), 2 mM L-glutamine, (all from Biofluids, Rockville, MD), and 10~ fetal calf serum (~7~ltnn R;o~h~m;c~lq, Lenexia, KS) were plated in 10-cm culture dishes. The next day the medium was changed and two hours later the cells were transfected with 20 ~g of plasmid DNA [produced in E.
coli strain B IO1 (Gibco/BRL) and purified with a Qiagen (Chatsworth, CA) Plasmid Kit] using the calcium p~nsphAt~
precipitation method (Davis, L.G., et al., Basic methods i~ molecular bioloqY, Elsevier, New York (1986)).
For pulse chase experiments COS-7 cells were trypsinized 16 h after transfection and replated in three 10-cm culture dishes to generate a uniform population of cells and m;n;m;7e dish to dish variation. The cells were then allowed to grow foi an additional 24 h. Cells in a 10-cm dish that were 90~ confluent were pr~;n~nhat~d for 20 min at 37~C in I th;nn;n~ free-DMEM rnnt~;n;n~ 10 fetal calf serum (dialyzed against PBS) after which [35S]methionine (Trans35S-label [ICN Radiochemicals, Irvine, CA~) was added to a ~nnr~nt~ation of O. 15 mCi/ml. The cells were pulsed for l h at 37~C and then washed and chased in DMEM cnnt~;n;ng 15 mg/ml of L-met~;nn;nP (chase medium)(Sigma Chemical Co., St.
Louis, MO). Cells and culture media were collected separately. In some pulse-chase experiments cells were pretreated for 2 h, pulsed for 1 h, and chased for 2 h in medium that ~nnt~;n~d 3/~g/ml of tunicamycin (Sigma) throughout the experiment.

21 q72~2 W096l05228 PCT~S9~10~8 -~2-F le 3: Isolation ~n~ ~Iri~; rnt;~n of Proteins ImmunoDreci~itation and endoqlvcosidase treatment Culture media were used immediately for ~ cipitation. Cells were washed 3 times with cold PBS, scraped from the dishes and lysed overnight in 250 ~1 lysis buffer (50 mM Tris, 5 mM EDTA, 150 mM NaCI, 0.5 NP-40 (Calbiochem, San Diego, CA), lmM phenyl methyl sulfonyl fluoride (Sigma), pH 3.0). Nuclei were pelleted by centrifugation for 30 min at 12,000 g. Culture media (3 or 6 ml) or cell lysates (250 ~1) were ;n~nhated for 3 h at 4~C with gentle tnr~l ;n~ with 20 ~l of packed Protein A-Sepharose beads (Pharmacia) precoated with rabbit IgG. The precleared samples were then centrifuged and incubated with 3 ~g oi either 9EI0 or A2B4 m~b for 3 h, followed by 3-16 h ;n~nhat;on at 4~C with 20-30 ~l of packed Protein A-Sepharose precoated with rabbit anti-mouse Ig. The ; , ecipitates were washed three times with 5% sucrose, 1~ NP40, 0 5 M NaCI, 50 mM Tris, 5 mM EDTA, pH 7.2 and once with-5~ ~M Tris, 1~0 mM NaCI, 1 Triton X-loo (Siqma), 0.15~ SDS, 1~ sodium deoxycholate, and solubilized in 40 ~1 reducing SDS polyacrylamide gel electrophoresi6 (SDS-PAGE) sample buffer. Sequential immunoprecipitationS using the same anti~oiy were performed until less than 1~ of the original protein L~ ; n~ in the supernatant. In some exp~; a immunoprecipitates were digested with endoglycosidase H
(endo H) (Genzyme, Cambridge, MA), as described in (Kearse, ~.P. and Singer, A., ~. Immunol. Meth~a (1994)). Following endo H digestion, an equal volume of reducing SDS-PAGE sample buffer was added.

SDS-PAGE, autoradioqra~hv and densitometrv SDS-PAGE was performed using a Pharmacia PhastSystem with 12.5~ homogeneous Phastgels. Samples were heated to 95=C for 5 min, beads were removed by centrifugation and ~ 096/os~ 2 1 9 7 2 3 2 . 11~ ~

3.5 ~l aliquot8 of snrPrn~t~nt were applied to the gel.
After electrophoresis, gels were ;n~~lh~tPd with shaking twice for 20 min in dimethyl sulfoxide, once for 2 h in 20~ w/v 2,5-diphenyloxazole in dimethyl sulfoxide, and once for 30 min in ~0. Gels were dried in a microwave oven for 20 min at the lowest power level and exposed to film (Kodak X-OMAT AR, Rochester, NY) at -70~C for 1-3 days. Autoradiograms were quantified with a Molecular Dynamics C~ i ng Dengitometer using Imagequant soilware. E~U~UL~ times were chosen to ensure linearity between radioactivity and band density. Multiple loadings of identical samples gave standard deviations in band densities of less than 10~.

Exam~le 4: sFv Bindinq Studies Binding of sFv proteins was tested using the radiolabeled material present in the media of transfected COS-7 cells after either a 2 h or a 6 h chase. Binding of OKT9 and 2C11 sFv media was tested on K562 (TfR') and 2B4 ~CD3') cells, respectively. Chase media c~nt~;n;ng (4 ml) 10 mM HEPES (Biofluids) were in~nhatPd for 4 hr at with 5-10 x 106cells in 6 well plates, in the presence or absence of excess inhibiting or control ~nt;ho~;PR
Inhibiting antibodies were the parental mAbs from which the sFv proteins were derived, and control ~nt;ho~;es were W6/32 and H57 for OKT9 and 2C11 respectively, both of which bind to the same cells as the sFv proteins, but to different antigens. After ;n~nhat;~n and washing, cells were lysed in 250 ~1 lysis buffer, and sFv proteins were ; ~precipitated and analyzed by SDS-PAGE as described above. U7.6 sFv was tested for binding by ; n~nh~t; on with gentle tumbling of 1.5 or 4ml of chase medium with 50 ~1 of packed DNP-Sepharose beads for 16 h in the absence or presence of 1 mM DNP-e-~m;n~r~p~oate.
The beads were washed and the bound sFv wag 8olllh; 1 i 7 Wo96/OS~h in 40 ~1 reducing sample buffer in preparation for SDS-PAGE. In eome studies, U7.6 sFv sup~n~tAntR were incubated a second time with 50 ~1 of DNP-beads. The doubly-depleted Snr~rnAt~ntq were then assayed for residual sFv by i nprecipitating with 9E 10 and rabbit anti-mouse protein A beads. FractionE were then analyzed using SDS-PAGE and the bands were quantified by densitometry.
In order to measure relative affinities of the U7.6 mutant sFv constructs, 1 ml aliquotg of 35S-. th;nn;n~
labeled chase media nnntA;n;nr .01~ sodium azide, 10 mM
HEPES, and graded amounts of DNp-~-Am;nncAr~oate were incubated overnight at 4~C with 40 ~1 of packed DNP-Sepharose beads. The beads were washed, and bound sFv removed by addition of 1 mM DNP-hapten for 6 hr at 4~C.
The eluted sFv was ; ~,ecipitated with 9E10 mAb and protein A-Sepharose, and the beads were washed, divided into 4 equal portions, and~heated for 5 min at 94~C in 100 ~1 of elutior. bLiier (20mM Tris, lmM EDTA, 2~ SDS and 5~ 2ME p~ 7.8). The eluate (8~1) was transferred directly to 2.5ml 03f Ecolume (TCN Cleveland Ohio) ,qr; nt;llAtion 801ution ad counted in a liquid 8r;rt;11~tion counter.

Exam~le 5: Dissociation of OKT9-sFv from K562 Cells To measure the ~;ccor;~t;nn rate o~ COS-7 cell-produced OKT9-sFv from K562 cells, tubes cnnt~;n;nr 6X106 K562 or 2B4 (control) cells and 1 ml of COS-7 medium rnnt~;n;ng 3s9-meth;on;n~ labeled OKT9-sFv and supplemented with .01~ sodium azide, 10 mM ~EPES were ;ncnhAt~d 1 h at 4~C. Cells were centrifuged for 5 min at 425xG. To determine the maximal amount of sFv bound (O time), cells were resuspended in ~ank's hAlcnr~ salt solution without phenol red nnnt~;n;nr 0.1~ BSA and .01 sodium azide (wash buffer), immediately spun for 10 sec ~o ~/05228 2 ~ 9 7 2 3 2 P~ ~48 in an ~ppPn~nrf mi~LocellLLifuge, and the pellet frozen.
Other samples were sl~pPn~pd in 1 ml wash buffer cnnt~;n;ng 50 ~g OKT9 mhb, incubated for various times at ~ 4~C, apun for 10 sec, and pellets frozen. Cell pellets were lysed and sFv ; nprecipitated with 9E10 mAb and protein A-Sepharose beads, and taken for ~r;nt;ll~t;n~
rollnt; ng as described above. Quadruplicate samples were averaged, and the amount of radioactivity associated with 2B4 cells was rnn~i~pred background and was subtracted from the amount binding to K562 cells.
The dissociation rate was also measured using bacterially-produced OKT9-sFv and flow cytometry. OKT9-sFv was produced and refolded from bacterial inclusion bodies, and labeled with fluorescein isothiocyante (FITC) as described in Segal, D.M. et al., Meth. EnzYmol.
150:478-492 (1987) K562 cells (107) were incubated for 1 h at 4~C with 2 ~g OKT9-sPv-FITC, with or without SO ~g OKT9 mAb in a total volume of 1 ml. Cells were centrifuged for 5 min at 425xG, resuspended in 1 ml wash buffer rnnt~;n;nS 50 ~g OKT9 mAb, and incubated at 4~C.
At various times, samples were analyzed by flow cytometry. Mean fluorescence values were determined and control values (mean fluorescence of samples inrllhat with FITC-sFv in the presence of:excess unlabeled mAb) were subtracted at each time point.

Ecuivalents Those skilled in the art will recognize, or be able to ascertain, using no more than routine experimentation, many equivalents to the specific Pmhr~;- t~ of the invention described herein. Such equivalents are ;ntPn~pd to be ~n~ sed by the following claims:

PCTnUS95/10348 21 9723?

(1) GENBRAL L~rl ~TON
li) APPLICANT.
(A) NAUB: The United States of America, as reprenented by the Secretary of the Health and ~uman Service~
.B: STRBBT: 6011 Executive Boulevard, Susite 325 :C: CITY: Rockville D: STATE/PROVINCE: Maryland E COUNTRY: U.S.A.
:F POSTAL CODE/ZIP: 20852-3804 (i) APPLIC~NT:
A NAMB: Creative ~nMrl~r~ Inc.
B STRBET: 35 South Street ~C CITY: ~opkinton D STATE/PROVINCB~ - A.
E COUNTRY: U.S.A.
P POSTAL COD3/ZIP: 01748 (ii) TITLE OF INVENTION: Method of ProduciLg Single-Chain Fv Molecules (iii) NOMBER OP SEQUENCE8: 17 (iv) uu~ U~N~ ADDRE9S:
A AlD-BSSEE: ~amilton, Brook, Smith & Reynolds, P.C.
.B:~ ~ ET: mwO Militia Drive .Ol r : Lexington n, . I.~mE: ~
: CO~lTRY: U.S.
F ZI': 02173 (v) C0M~UTBR RB~DA3LB PûRM:
.A.I ME~IT~M mLYPE. Floppy dink B~ CûYPUTBR: I~M PC I ihl~
C opeRATING SYSTEM: PC-DOS/MS-DOS
D SO TWARE: PatentIn Release #1.0, Version #1.25 (vi) CURRBNT APPLICATION DATA:
(A) APPLICATION NUMBER:
(B) FILING DATE:
(C) CLASSIFICATION
(vii) PRIOR APPLICATION DATA:
(A) AppLIcaTIoN NOMBER: U8 08/292,124 (B) FILrNG DATE: 17-AUG-1994 (viii) ATTORNBY/AGENT INFORMATION:
(A) NAME: Brook, David E.
(B) REGISTRATION NOMBER: 22,592 (C) RBFERENCB/DOCXBT NGMBER: CBM94-01 (iX) mRT. --Tnr~ T~W:
(A) TELEPXONE: 617-861-6240 (B) TBLEFAX: 617-861-9540 ~ 096/05228 PCT/US95/10348 21 ~J~3~

.

(2) lN~'U~_.Il~ FOR SEQ ID NO:1:
(i) SBQUENCE ~D~Dr ~ r.~T.~ll~b:
(A) LENGT~: 102 base pairs (B) TYPE: nucleic acid (C) b : ~ingle (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE DESCRIPTION: SEQ ID NO:1:
TGTTAACTGC TCACTTCTAG AATGAGGACC CCTGCTCAGT TTCTTGGAAT ~ll~l~lC 60 TGGTTTCCAG GTATCAP,ATG TGACATCAAG ATGACCCAGT CT 102 (2) lN~I - T~N FOR SEQ ID NO:2:
(i) SEQUENCE ~ O D ~ I h l l~b:
(A) LENGTE: 86 base pairs (B) TYPE: nucleic acid (C) b ~: ~ingle (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE ~b~l~l~N: SEQ ID NO:2:

(2) lN~ T~ FOR SEQ ID NO:3:
(i) SEQUENCE f~nD~ hll~b:
(A) LENGT~: 104 base pairs (B) TYPE: nucleic acid (C) b ~ ~h: ~ gle (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE DESCRIPTION: SEQ ID NO:3:
ATATAGAATT CCTCGAGGAG ~Lc~ ~G AGTTCGTCCT ll'~l~il CAGATCCTCT 60 TCTGAGATGA ~lllll~llC TGATAAAGCT TTTGAGGAGA CTGT 104 (2) INFORMATION FOR SEQ ID NO:4:
(i) SEQUENCE frD~D~ ~ _ ~ l hl l~'b:
(A) LENGT~: 108 base pairs (13) TYPE: nucleic acid (C) ~ h: ~ingle (ii) MOLECULE TYPE: synthetic DNA

W 096~S228 .~1~ ,10~48 ~
2~ 97~3~

(xi) SEQUENCE ~LKl~llUN: SEQ }D NO:4:
TGTTAACTGC TCACTTCTAG AATGAGGACC CCTGCTCAGT TTCTTGGAAT LLL~L'~LLU 60 TGGTTTcrAG GTATCAAATG TGACGTCGTC ATGACCrAGT CTCCAGCA 108 t2) INFORMATION FOR SEQ ID NO:S:
(i) SEQUENCE r~ T~I lLS:
(A) BENGT~: 29 base pairs (B) TYPE: nucleic acid (C) ST~r ~: ringle (ii) MOLEC~EE TYPB: ~ynthetic DNA
(Xi) SEQUENrE n~S~l~llUN: SEQ ID NO:5:
ATATAGGATC r~Tr~-Grcr CCTACTGTC 29 (2) lN~ lVN FOR SEQ ID NO:6:
(i) SEQuENcE rFr n~
(A) LENGTH: 80 base pairs (B~ TYPE nucleic acid (C) ST~Nn~N~Cq: 3ingle (ii) MOLEC~BE TYPE: ~ynthetic DNA
(xi) SEQUENrE ~U~l~LluN: SEQ ID NO:6:

TTrATTTCCA ~ L ' ~L~.~ 80 (2) lN~U~ .11UN FOR SEQ ID NO:7:
(i) SEQUENCE r~D~ ~ l LS ' (A) BENGTE: 32 base pair~
(B) TYPE: nucleic acid (C) ST~Rr~-cc: single (ii) MOLECULE TYPE: ~ynthetic DNA
(xi) SEQUENCE ~KSL~l~ I lUN: SEQ ID NO:7:
~rr,rrr. ccGcr~Ar~GTr~ CAGCTGGTGG AG 32 (2) INFORMATION FOR SEQ ID NO:8:

(i) SEQUENCE r~ lCS:
(A) BENGT~: 74 ba8e pairs (B) TYPE: nucleic acid (c) ~ l ~r~ c: ~ingle ~0 96/05228 r~ U ;P
2 1 ~232 (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE L~LK~ uN: SEQ ID NO:8:
ATATACTCGA GTTATTAATT CAGATCCTCT TCTGAGATGA ~lll~l~llC TGATGAGGAG 60 (2) lNr~l E_I1UN FOR SEQ ID NO:9:
(i) SE9UENCE r~D~ U~
(A) LENGT~: 131 base pair~
(B) TYPE: nucleic acid (C) S~ N~CC: si~gle (ii) MOLECULE TYPE: syLthetic DNA
(xi) SEQUENCE DESCRIPTION: SEQ ID NO:9:
A~l~T~rl~l~G r~ r~D ACCAGCGATT r~Dn~,aar CATGCAGAGG ~ r~ 60 ~ ~hll~l~l~ GTGCAGATTT GCTGGGTGAG AGGAGATCAG GTGGAGCAGA 120 (2) lN~ Tn~ FOR SEQ ID NO:10:
(i) SEQUENCE r~D~DI ~ ~ ~ x~
(A) LENGT~: 38 base pairs (3) TYPE: nucleic acid (C) ST.~.~ : si~gle (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE ~Kl~llUN: SEQ ID NO:10:
ATATAGGATC CTCACTAP~T CACATTTCTC AGATCCTC 38 (2) INFO~MATION FOR SEQ ID NO:11:
(i) SEQUENCE a~roD~ I xll~
(A) LENGT~: 2s ba~e pairs (B) TYPE: nucleic acid (C) ST.~. r : single (ii) MO~ECULE TYPE: synthetic DNA
(xi) SEQUENCE DESCRIPTION: SE~ ID NO:11:

W 09~05228 r~
21 97~3~

(2) INFORMATION FOR SEQ ID NO:12:
(i) SEO~ENCE ~~~
(A) LENGTH: 29 base p2irs (B) TYPE: nucleic acid (C) ~ : Yingle (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENOE ~UKl~ ~ lUN: SEQ ID NO:12:
GGCGCTTCAG TGAATATATC CTGcAaGGC 29 (2) lN~I - TON FOR SEQ ID NO.13:
(i) SEQ~EN OE ~FD~D~
(A) LENGTH: 35 base pairs (B) TYPE: nucleic acid (C) ST~Dr-l~T ~ RR: single (ii) MOLECULE TYPE: synthetic DNA
(xi) SEQUENCE ~UKl8llUN: SEQ ID NO:13:
UUU~ll~4l~ TTC~AGTTTT ul~I~ ACAGC 35 ~2) INFORMATION FOR SEQ ID NO:14 (i) SEQ~EN OE ~r~~D~
A LENGTH: 360 base pairs 3. TYPE: nucleic acid C " . . ~ N ~ i: double D, TûPO~OGY: linear (ii) MOLEC~DE TYPE: DNA (ge~omic) (ix) FEATURE:
(A) NAME/KEY: CDS
(B) LOCATION: 1..36C
(xi) SEQUENCE DESCRIPTION: SEQ ID NO:14:

Gln Val Gln Leu Gln Gln Ser Gly Pro Glu Leu Glu Lys Pro Gly Ala l 5 10 15 TCA GTG AAG ATA TCC TGC AaG GCT TCT GGT TAC TCA TTC ACT GGC TAC 96 Ser Val Lys Ile Ser Cys Lys Ala Ser Gly Tyr Ser Phe Thr Gly Tyr ATC ATG AAC TGG GTA Aa~ C~G AAC AAT GGA AAG AGC CTT GAG TGG ATT 144 Ile Met Asn Trp Val Lys Gln Asn Asn Gly Lys Ser Leu Glu Trp Ile Gly Asn Ile Ala Pro Tyr Tyr Gly Gly Thr Ser Tyr Asn Gln Lys Phe ~ 096~5228 21 q7232 A-a-G GGC A~G GCC ACA TTG ACT GTA GAC AaA TCC TCC AGC ACA GCC TAC 240 Lys Gly Lys Ala Thr Leu Thr Val Asp Lys Ser 8er 8er Thr Ala Tyr Met Gln Leu 8er 8er Leu Thr 8er Glu Asp 8er Ala Val Tyr Phe CYG

Ala Arg Trp Gly Gly Thr Met Ile Thr Gly Leu Asp Tyr Trp Gly Gln GGC ACC A~T CTC ACA GTC TCC TCA 360 Gly Thr Thr Leu Thr Val 8er Ser (2) lN~ - FOR SBQ ID XO:15:
~i) 8EQUENCE r~va~
~A) LENGTE: 120 amino acids (B) TYPE: amino acid (D) TOPOLOGY: linear (ii) MOLECOLE TYP3: protein (xi) SEQ~EN OE DESCRIPTION: SEQ ID XO:15:
Gln Val Gln Leu Gln Gln Ser Gly Pro Glu Leu Glu Ly~ Pro Gly Ala l 5 10 15 ser Val Ly8 Ile Ser Cy8 LYB Ala Ser Gly Tyr Ser Phe Thr Gly Tyr Ile Met A~n Trp Val Ly8 Gln Asn A8n Gly Lys Ser Leu Glu Trp Ile Gly Acn Ile Ala Pro Tyr Tyr Gly Gly Thr Ser Tyr Asn Gln Ly~ Phe Lys Gly Ly~ Ala Thr Leu Thr Val Asp Lys Ser Ser Ser Thr Ala Tyr Met Gln Leu Ser Ser Leu Thr Ser Glu A~p Ser Ala Val Tyr Phe Cy~

Ala Arg Trp Gly Gly Thr Met Ile Thr Gly Leu Asp Tyr Trp Gly Gln Gly Thr Thr Leu Thr Val Ser Ser (2) 1Nr~ ~TON FOR SEQ ID NO:16:
(i) SEQ-~ENCE ~V~~~rT~RTcTICs:
A LENGT~: 327 ba~e pair~
B TYPE: nucleic acid C ~ rr ~ : double D TOPOLOGY: linear WO 96/05228 PCTIIJS95/10348 ~
2~9i'~3~ _ (ii) MOLECULE TYPB: DNA (ge~omic) (ix) FEATU~E:
(A) NAME/~EY: CDS
(3) LOCATION: 1..327 (xi) 8EQUEN OE ~bb~l~-loN: SEQ ID NO:16:

ABP Ile Val Met Thr Gln Ser Pro Ala Ile Met Ser Ala Ser Pro Gly Glu Lys Val Thr Met Thr Cys Arg Ala Ser Ser Ser Val Ser Ser Thr TAC TTC CAC TGG TAC CAG CAG AAG TCA GGT GCC TCC CCC AaA CTC TGG 144 Tyr Phe ~is Trp Tyr Gln Gln Lys Ser Gly Ala Ser Pro Lys Leu Trp Ile Tyr Ser Thr Ser Thr Leu Ala Ser Gly Val Pro Ala Arg Phe Ser Gly Ser Gly 8er Gly Thr Ser Tyr Ser Leu Thr Ile Ser Ser Val Glu GCT GAA GAT GCT GCC ACT TAT TAC TGC caG CAG TAC AGT GGT TAC CCG 288 Ala Glu Asp Ala Ala Thr Tyr Tyr Cy8 Gln Gln Tyr Ser Gly Tyr Pro 85 90 gs CTC ACG TTC GGT GCT GGG ACC AaG CTG GAG CTG ~AA CGC 8Z7 Leu Thr Phe Gly Ala Gly Thr LYB Leu Glu Leu Lya Arg lOO 105 (2) IN'FORM~TION FOR SEQ ID NO:17:
(i) SEQ~EN OE r~nP,.~
(A) LENGT~: 109 amino acids (3) TYPE: amino acid (D) TOPOLOGY: linear (ii) MOLECULE TYPE: protein (xi) SEQUENCE DES OEIPTION: SEQ ID NO:17:
~BP Ile Val Met Thr Gl~ Ser Pro Ala Ile Met Ser Ala Ser Pro Gly ~lu Lys Val Thr Met Thr CYB Arg Ala Ser Ser Ser Val Ser Ser Thr Tyr Phe ~is Trp Tyr Gln Gln Lys Ser Gly Ala Ser Pro LYB Leu Trp Ile Tyr Ser Thr 8er Thr Leu Ala Ser Gly Val Pro Ala Ar~ Phe Ser Gly Ser Gly Ser Gly Thr Ser Tyr Ser Leu Thr Ile Ser Ser Val Glu ~ 0 96t05228 .~~ 0348 ~ ~72~

Ala Glu Aap Ala Ala Thr Tyr Tyr Cys Gln Gl~ Tyr Ser Gly Tyr Pro Leu Thr Phe Gly Ala Gly Thr Lys Leu Glu Leu Lya Arg

Claims (31)

The invention claimed is:

1. A method of producing single-chain Fv molecules in eukaryotic cells comprising:
(a) modifying a nucleic acid sequence which encodes a single-chain Fv molecule to include at least one non-naturally occurring glycosylation site in the nucleic acid sequence, thereby producing a single-chain Fv construct;
(b) introducing the single-chain Fv construct of step (a) into a vector capable of expressing said construct in a eukaryotic cell and transfecting said vector containing the single-chain Fv construct into a eukaryotic cell; and (c) maintaining said cell transfected with the vector of step (b) in cell culture medium under conditions sufficient for expression of the single-chain Fv construct within the cell and secretion of the expressed single-chain Fv protein product from the cell into the cell culture medium, thereby producing a single-chain Fv molecule.

2. A method of producing single-chain Fv molecules in mammalian cells comprising:
(a) modifying a nucleic acid sequence which encodes a single-chain Fv molecule to include at least one non-naturally occurring glycosylation site in the nucleic acid sequence thereby providing a single-chain Fv construct;
(b) introducing the single-chain Fv construct of step (a) into a vector capable of expressing said construct in a mammalian cell and transfecting said vector into a mammalian cell; and (C) maintaining said cell transfected with the vector of step (b) in cell culture medium under conditions sufficient for expression of the single-chain Fv construct within the cell and secretion of the expressed single-chain Fv protein product from the cell into the cell culture medium, thereby producing a single-chain Fv molecule.

3. The method of Claim 2 wherein the mammalian cell is selected from the group consisting of: COS-7 monkey kidney fibroblast cells; K562 human erythroleukemia cells; 293 cells; myeloma cells; and Chinese hamster ovary cells.

4. The method of Claim 2 wherein the glycosylation site of step (a) is an N-linked glycosylation site.

5. The method of Claim 4 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

6. The method of Claim 2 wherein the nucleic acid sequence encodes single-chain Fv U7.6 (SEQ ID NOs: 14 and 16).

7. The method of Claim 6 wherein the modification to the U7.6 nucleic acid sequence to include at least one glycosylation site consists of an asparagine amino acid residue substituted for lysine 19 of VH.

8. A single-chain Fv molecule produced by the method of Claim 2.

9. A method of increasing the rate of secretion of single-chain Fv molecules from mammalian cells comprising introducing into a single-chain Fv molecular at least one non-naturally occurring glycosylation site.

10. The method of Claim 9 wherein the glycosylation site is an N-linked glycosylation site.

11. The method of Claim 10 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

12 A method of decreasing intermolecular aggregation of single-chain Fv molecules in solution comprising introducing into a single-chain Fv molecule at least one non-naturally occurring glycosylation site.

13. The method of Claim 12 wherein the glycosylation site is an N-linked glycosylation site.

14. The method of Claim 13 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

15. A method of protecting a single-chain Fv molecule from proteolytic degradation comprising introducing into a single-chain Fv molecule at least one non-naturally occurring glycosylation site.

16. The method of Claim 15 wherein the glycosylation site is an N-linked glycosylation site.

17. The method of Claim 16 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

18. A method of decreasing the antigenicity of a single-chain Fv molecule comprising introducing into a single-chain Fv molecule at least one non-naturally occurring glycosylation site.

19. The method of Claim 18 wherein the glycosylation site is introduced into the linker sequence.

20. The method of Claim 18 wherein the glycosylation site is an N-linked glycosylation site.

21. The method of Claim 20 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

22. A method of modifying the ligand binding affinity of a single-chain Fv molecule comprising introducing into a single-chain Fv molecule at least one non-naturally occurring glycosylation site.

23. The method of Claim 22 wherein the glycosylation site is an N-linked glycosylation site.

24. The method of Claim 23 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

25. A method of modifying the pharmokenetics of a single-chain Fv molecule comprising introducing into a single-chain Fv at least one non-naturally occurring glycosylation site.

26. The method of Claim 25 wherein the glycosylation site is an N-linked glycosylation site.

27. The method of Claim 26 wherein the glycosylation site is encoded by the nucleic acid sequence Asn-X-Ser/Thr.

28. A method of screening single-chain Fv molecules for biological activity comprising:
(a) modifying a nucleic acid sequence which encodes a single-chain Fv molecule to include at least one non-naturally occurring glycosylation site in the nucleic acid sequence, thereby producing a single-chain Fv construct;

(b) introducing the single-chain Fv construct of step (a) into a vector capable of expressing said construct in a mammalian cell and transfecting said vector containing the single-chain Fv construct into a mammalian cell;
(c) maintaining said mammalian cell transfected with the vector of step (b) in cell culture medium under conditions sufficient for expression of the single-chain Fv construct within the cell and secretion of the expressed single-chain Fv protein product from the cell into the cell culture medium; and (d) obtaining an aliquot of cell culture medium containing the single-chain Fv molecule and testing the aliquot for biological activity.

29. A secretable single-chain Fv molecule having one, or more, non-naturally occurring glycosylation site, or sites.

30. The single-chain Fv molecule of Claim 29 wherein the glycosylation site is an N-linked glycosylation site.

31. The single-chain molecule of Claim 29 wherein the single-chain Fv molecule is U7.6 and the glycosylation site of the U7.6 Fv molecule consists of an asparagine amino acid residue substituted for lysine 19 of VH.