WO1996012009A2 - Methods for engineering antigen-presenting cells - Google Patents

Abstract

A protein transfer method for producing a cell having a defined MHC: nominal antigen peptide or costimulator on its membrane.

Description

DESCRIPTION

Methods For Engineering Antiσen-Presenting Cells

Field of the Invention

This invention relates to methods and reagents for engineering antigen-presenting cells ("APCs") for purposes of immunotherapy.

Background of the Invention

The following is a general discussion of relevant art, none of which is admitted to be prior art to the invention.

An objective in the field of immunotherapy is the development of strategies for activating or inhibiting T-cells that are therapeutic or pathogenic, respectively. APCs provide one efficient means for accessing antigen- specific T-cells.

In general, APCs are responsible for initiating most immune responses through their pivotal role in antigen presentation to T-cells. During antigen presentation, endogenously processed nominal antigen peptides associate intracellularly with either class I major histo- compatibility complex ("MHC") heterodimers or with class II MHC heterodimers, and the resulting heterotrimeric complexes are then translocated to APC surfaces (reviewed in Germain and Margulies, 11 Annu. Rev. Immunol. 403, 1993; Germain 76 Cell 287, 1994) . For a single immuno- genic protein that is pulsed with APCs, processing fre- quently yields a diverse MHC:nominal antigen peptide repertoire, the composition of which may differ between different APCs (Falk et al., 348 Nature 248,1990; Falk et al., 174 J. EXP. Med. 425, 1991; Henderson et al . , 255 Science 1264, 1992; Henderson et al. , 90 Proc. Natl. Acad. Sci. USA 10275, 1993; Hunt et al. , 255 Science 1261, 1992; Rammensee et al. , 11 Annu. Rev. Immunol. 213, 1993; Huczko et al., 151 J. Immunol. 2572, 1993) . In order to achieve a more defined MHC:nominal antigen peptide repertoire on a given APC surface, it is experimentally feasible to load exogenously synthesized oligopeptides onto class I or class II MHC heterodimers (Harding et al . , 86 Proc. Natl. Acad. Sci . USA 4230, 1989) . Such exogenously-loaded peptides do not require intracellular proteolytic processing.

For purposes of antigen presentation, it is common to employ one of two categories of cells, that is, cells that naturally express MHC heterodimers and cells transfected with an MHC gene expression cassette. In principle, another category of an MHC-bearing cell is one in which the MHC molecule has been exogenously attached to the cell surface. A method for delivering an MHC molecule to an APC surface has been described which entails the use of an H A-A2.1-streptavidin chemical conjugate which can be added to a pre-biotinylated cell (Elliott and Eisen, 87 Proc. Natl. Acad. Sci. USA 5213, 1990) . The chemical moiety was shown to be alloantigenic, but no data was presented that it can bind and present a nominal antigen peptide.

Functional interactions between APCs and T-cells are known to be mediated by both cell surface-associated and soluble molecules. Antigen-specificity in this inter- action is provided by the major histocompatibility com¬ plex:nominal antigen peptide complex of the APC inter¬ acting with the T-cell receptor (TCR) of the T-cell, constituting a trimolecular axis. However, effective interactions additionally require certain cell surface- associated and soluble costimulator molecules of APCs binding to receptors for these costimulators on T-cells.

Examples of known costimulators on APCs are B7 ( insley et al., 87 Proc. Natl. Acad. Sci. U.S.A. 5031, 1990) ; ICAM-I

(van Seventer et al. , 144 J. Immunol. 4579, 1990) ; VCAM-I (van Seventer et al. , 174 J. EXP. Med. 901, 1991) ; LFA-3 (van Seventer et al. , 21 Eur. J. Immunol. 1711, 1991; fibronectin (Shimizu et al. , 145 J. Immunol. 59, 1990; Nojima et al . , 172 J. Ex . Med. 1185, 1990; Davis et al . , 145 J. Immunol . 785, 1990) .

An objective in the field of tumor immunotherapy is the development of strategies for enhancing tumor immuno- genicity, with potential applications in both tumor prevention and cure. In general, tumors that arise de novo are poorly immunogenic, thereby escaping host anti- tumor responses (Hewitt et al., 33 Br. J. Cancer 241, 1976) . The basis for this low immunogenicity is unclear. Several experimental strategies have been described for enhancing tumor immunogenicity, for example, use of mutagen or drug treatment (Van Pel and Boon 79 Proc. Natl. Acad. Sci. USA 4718, 1982, Frost et al. , 159 J. EXP. Med. 1491, 1984) ; by transfection with a foreign gene encoding an exogenous antigen such as influenza hemagglutinin (Fearon et al. , 38 Cancer Res. 2975, 1988) ; by reducing the expression of certain molecules in a tumor that regulate its differentiation state (Tykocinski & Ilan, PCT/US93/03637) ; by transferring a gene expressing a lymphokine into a tumor, for example, interleukin-2 (Fearon et al. , 60 Cell 397, 1990) , interleukin-4 (Tepper et al., 57 Cell 503, 1989, Golumbek et al. , 254 Science 713, 1991) .

Summary of the Invention Applicant has determined a method for delivering a nominal antigen peptide to a cell surface that is not contingent upon prior expression of MHC by these cells. Both the MHC polypeptide and the nominal antigen peptide are externally delivered to the cell surface; intracel- lular protein processing pathways are not utilized. The method is predicated upon the use of a membrane reincor- porable variant of an MHC polypeptide. According to alternative embodiments of the present method, a nominal antigen peptide can be contacted with the MHC polypeptide variant prior to, concurrent with, or subsequent to the membrane reincorporation event. The MHC component is configured in a way that both facilitates its exogenous reincorporation into membranes and preserves its antigen presentation capacity. Hence, by such a strategy, APCs can in essence be "painted" with defined MHC:nominal antigen peptide complexes. This method allows for the tailoring of the antigen repertoire of an APC and enhance¬ ment of its therapeutic efficacy.

Applicant has determined that certain MHC polypeptide derivative can be genetically engineered to be suitable for use as a vehicle to deliver a nominal antigen peptide to a cell surface. Applicant believes that molecular topology substantially influences the MHC:nominal antigen peptide :T-cell receptor tri olecular axis and functional signaling through this axis. Hence, molecular topological aspects of an MHC moiety are critical determinants of its capacity to present antigen to T-cells. This indicates that a subset of engineered MHC polypeptide derivatives will support efficient antigen presentation to T- cells. Such a subset is determined by routine screening of different MHC classes using the techniques described below. The present invention discloses a class of MHC polypeptide derivatives which is particularly well-suited for effective antigen presentation. This class of MHC polypeptide derivatives has the added feature of reincor- porability into cell membranes. This feature permits it to be coated, or "painted", onto the surface of a cell.

The present invention is based upon the finding that lipid-modified MHC heterodimers, which acquire the property of membrane reincorporability by virtue of their lipid modification, retain an efficient antigen presenta¬ tion function. A preferred lipid modification for this purpose is a glycosyl-phosphatidylinositol ("GPI") modi¬ fication.

GPI-modified proteins are a class of native cell surface molecules that can be exogenously reincorporated back into cell membranes after purification (Medof et al . , 160 J. EXP. Med. 1558, 1984; Moran et al. , 149 J. Immunol. 1736, 1992; Zhang et al. , 89 Proc. Natl. Acad. Sci. USA 5231, 1992; Bulow et al. , 27 Biochemistry 2384, 1988; Hitsumoto et al., 5 Int. Immunol. 805, 1993) . This pro¬ perty stems from their amphiphilic properties and their solubility in exceedingly low detergent concentration, and as Applicant has found, in the complete absence of deter¬ gent. Protein transfer has been reported for a limited set of natural GPI-anchored proteins, including decay- accelerating factor ("DAF") (Medof et al., 160 J. Exp. Med. 1558, 1984; Moran et al. , 149 J. Immunol . 1736, 1992) , Thy-1 (Zhang et al . , 89 Proc. Natl. Acad. Sci. USA 5231, 1992) , T. brucei variant surface glycoprotein (Bulow et al . , 27 Biochemistry 2384, 1988) , and mouse heat-stable antigen (Hitsumoto et al., 5 Int . Immunol . 805, 1993) . For DAF and heat stable antigen, biological functions have been demonstrated for the exogenously reincorporated pro¬ teins (Moran et al. , 149 J. Immunol. 1736, 1992; Zhang et al., 89 Proc. Natl. Acad. Sci. USA 5231, 1992; Hitsumoto et al . , 5 Int . Immunol. 805, 1993)) . Polypeptide sequences can be artificially GPI- modified by linking their coding sequences to a GPI modi¬ fication signal sequence (Tykocinski et al . , 85 Proc. Natl. Acad. Sci. USA 3555, 1988; Straus et al . , 82 Proc. Natl. Acad. Sci. USA 6245, 1985; Caras et al. , 238 Science 1280, 1987; aneck et al. , 85 Proc. Natl. Acad. Sci. USA 577, 1988) . This finding has been substantiated by other more recent studies for a variety of target proteins (Lin et al . , 249 Science 677, 1990;- Sleckman et al. , 147 J. Immunol . 428, 1991; Zhang et al . , 115 J. Cell Biol . 75, 1991; Mann et al . , 142 J. Immunol. 318, 1989; Matsui et al., 254 Science 1788, 1991; ettstein et al . , 174 J. Exp. Med. 219, 1991; Kemble et al. , 122 J. Cell Biol. 1253, 1993; Weber et al. , 210 EXP. Cell Res. 107, 1994; Huang et al., 31 Mol. Immunol . 1017, 1994; Scheirle et al . , 149 3_. Immunol . 1994, 1992) . Gene transfer of GPI-modified MHC is disclosed by Mann et al. , 142 J. Immunol . 318, 1989; Matsui et al . , 254 Science 1788, 1991; Wettstein et al . , 174 . EXP. Med. 219, 1991; Scheirle et al. , 149 J. Immunol . 1994, 1992; and Huang et al . , 31 Mol . Immunol . 1017, 1994.

The methods of protein transfer of the present invention also allow for the delivery of one or more MHC allelic variants to an APC.

In a related invention, protein transfer can also be applied to an immunomodulatory polypeptide other than a MHC polypeptide for purposes of APC engineering. In this case, the objective is generally to modulate the immunogenic potential of the APC in order to enhance its therapeutic efficacy. An immunomodulatory polypeptide, such as a cell surface costimulator, is modified through genetic or other types of manipulation to confer to it the property of membrane reincorporability. This artificial polypeptide, which by virtue of its modification is amen¬ able to protein transfer, can be purified and then painted onto the surface of a cell, such as an APC.

Applicant has discovered that the immunogenicity of a tumor cell can be remarkably enhanced by expressing on the surface of that tumor cell an artificial costimulator polypeptide, such as a costimulator:GPI chimeric poly¬ peptide, that is amenable to protein transfer. One or more of such artificial costimulator polypeptides can be selected from known costimulators, for example, B7-1, B7- 2, B7-3, intracellular adhesion molecule ICAM-1, ICAM-2, ICAM-3, LFA-3, VCAM-I, and fibronectin, and the modified costimulator can be used for- tumor APC engineering. Methods that relate to tumor APC engineering can be readily generalized to other types of APCs.

The present invention discloses several membrane- binding polypeptide variants that are effective for coating tumor cells in order to enhance the immunogenicity of the cells. A preferred variant is a polypeptide including at least the extracellular domain of a costimu¬ lator polypeptide with a covalently linked glycosylphos- phatidylinositol (GPI) moiety. Such GPI-modified poly- peptides can be readily generated through chimeric gene transfer technology. Moreover, GPI-modified polypeptides, once purified from cells, can be readily reincorporated into cell membranes. In contrast to MHC protein transfer to APCs, costimulator delivery may allow for greater latitude in the applicable protein transfer techniques. Other chime¬ ric polypeptide variants can be used effectively for cell coating. An example of such a polypeptide is one in which the extracellular domain of a costimulator polypeptide is linked to streptavidin. This chimeric polypeptide can be used to coat tumor cells that have been pre-coated with a universal binding agent, such as a biotin:lipid conjugate. Yet another example of a polypeptide variant that can be used for coating cells is one in which the extracellular domain of a costimulator polypeptide is linked to a second polypeptide. The second polypeptide can be one which binds to a receptor on the tumor cell, to specifically target the costimulator. The present invention also provides an efficient method for identifying combinations of costimulator mole¬ cules that enable optimal enhancement of tumor cell immu¬ nogenicity. Protein transfer offers the significant ad¬ vantage of being able to deliver multiple molecules simul- taneously. Since substantial advantage is obtained by coating with multiple costimulators, protein transfer allows for the rapid deciphering of the optimal costimu¬ lator array for a given cell type.

According to yet another embodiment of the present invention, costimulator coating is combined with other known methods for enhancing the immunogenicity of a cell such as a tumor cell. It is anticipated that additive or synergistic effects may be obtained by such combinatorial approaches. According to one combinatorial approach, costimulator coating is applied to the same cell that has been otherwise manipulated to enhance its immunogenicity. For example, a transfected tumor cell expressing a soluble lymphokine, such as interleukin-4, can be coated by protein transfer with one or more membrane-binding GPI polypeptide derivatives of costimulators. Moreover, even three or more distinct approaches can be combined in the same cell, e.g. , costimulator coating, soluble lymphokine expression, and insulin-like growth factor inhibition. More optimal "universal" immunogenic tumor cells can be generated in this way.

According to a second combinatorial approach, cell populations are combined to constitute a therapeutic cellular mixture, wherein individual cell populations have been modified by only a single approach for enhancing tumor cell immunogenicity. Any number of such mixed cell populations, including cells coated with costimulators, can be coordinately administered to enhance therapeutic efficacy.

The present invention addresses the need for engineering tumor cell surfaces in complex ways to achieve selected cell surface molecular arrays. Methods are pro- vided for identifying for each tumor type the optimal molecular array for maximal immunogenic potential. In turn, methods are provided for engineering primary tumor cells in clinical settings to efficiently effect said tumor phenotypes. Thus, in a first aspect the present invention features a method for producing a cell having a defined MHC:nominal antigen peptide on its membrane. The method can be practiced by contacting the external portion of the cell with an externally applied nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind the nominal antigen peptide. Alternatively, the external portion of the cell can be first contacted with an externally applied artificial lipid-modified MHC polypeptide able to bind the nominal antigen peptide and then contacted with the nominal antigen peptide so that the artificial lipid-modified MHC polypeptide binds the peptide. In a third variation, the nominal antigen peptide is first contacted with the artificial lipid- modified MHC polypeptide able to bind the nominal antigen peptide and the external portion of the cell is then con¬ tacted with externally applied nominal antigen peptide bound to the lipid-modified MHC polypeptide.

By "cell" is meant an antigen presenting cell or any other cell which is capable of effectively interacting with a T cell, leading to activation or inhibition of the T cell. By "defined" is meant predetermined molecular species. By "nominal antigen peptide" is meant an oligopeptide that is capable of binding to the antigen-binding groove of an MHC molecule.

Many functional oligopeptides that are capable of binding to the antigen-binding pockets of different allelic (polymorphic) variants of class I and class II MHC molecules are known to those who practice the art. For some of the allelic MHC variants, nominal antigen peptide sequence motifs have been determined that allow one to predict whether any given oligopeptide will be capable of binding to a given MHC polypeptide. In practicing the present invention, one can draw upon this knowledge to either select a functional nominal antigen peptide from nominal antigen peptides known to bind a given MHC allelic variant or else configure nominal antigen peptide candi¬ dates based upon sequence motif requirements. Any oligo¬ peptide of interest can be readily synthesized using a commercial oligopeptide synthesizer and then contacted with an MHC:GPI molecule. By "contacting the external portion of the cell with an externally applied..." is meant the exterior surface of the cell membrane is contacted by applying externally the nominal antigen peptide and the lipid-modified MHC polypeptide either separately or together. This is in contrast with intracellular insertion of these molecules into the membrane, which occurs naturally or with genetic¬ ally engineered molecules expressed within the cell. By "artificial lipid-modified" is meant a covalent lipid modification that is not naturally present on the MHC molecule so modified; and functions so that the MHC molecule can be reincorporated into the cell membrane. That is, such a molecule does not occur naturally in nature. Artificial lipid-modified includes any other type of modification that can function to reincorporate an MHC molecule into a membrane. A preferred artificial lipid modification according to the present invention is a GPI modification. There is no known polymorphic MHC polypep¬ tide that is naturally GPI-modified. However, the non- polymorphic mouse MHC molecule called Qa-2 is naturally GPI-modified. In particular, applicant provides human MHC polypeptide with a GPI-type modification. Other embodi- ments of the present invention include the use of lipid- modified MHC molecules that comprise non-GPI lipid modifi¬ cations, for example, chemical coupling of a lipid moiety directly to the MHC polypeptide. Methods for covalently coupling lipids to polypeptides are described in the prior art. Of note, the advantage of GPI modification over non- GPI lipid coupling methods is the former's site-specifici¬ ty. Another advantage of the former is that the pep¬ tide:anchor link is natural and hence is less likely to be immunogenic. GPI-modified MHCs are made via chimeric gene transfer in cells in which the recombinant protein can be produced in large amounts.

By "MHC molecule" is meant any molecule that functions to bind nominal antigen peptides and can interact with T- cells. It need not be a complete MHC molecule as it exists in nature, rather it can be the active portion as determined by routine procedures. An MHC molecule capable of binding nominal antigen peptide and the T-cell receptor preferably consists of a heavy chain:j82m light chain heterodimer for class I MHC and an alpha:beta heterodimer for class II MHC. However, in the case of class I MHC, as opposed to class II MHC, the antigen-binding pocket is made up entirely of the heavy chain polypeptide subuni . In practicing the present invention for purposes of acti¬ vating or inhibiting T-cells with nominal antigen peptide- specificity, but not alloantigen-specificity, it is pre¬ ferred to select an MHC:GPI allelic variant that is syn- geneic with the T-cells undergoing modulation. For example, in the case of human class I MHC, there are usually six candidate MHC polypeptides to select from, that is, two allelic variants for each of the HLA-A, HLA- B, and HLA-C genetic loci. The choice among these candi- dates depends upon the therapeutic context. For example, in treating patients with chronic active hepatitis, there is information available that generating T cells with specificity for hepatitis B virus nominal antigen peptides restricted to the HLA-A2.1 allelic variant may be bene- ficial. Consequently, in treating a chronic active hepa¬ titis patient with HLA-A2.1 as one of the six class I MHC allelic variants, one would select HLA-A2.1:GPI in implementing the therapeutic method of the present inven¬ tion. The optimum MHC is readily determined by routine procedures.

Not all potential membrane-binding MHC molecular derivatives will be capable of effective antigen presenta¬ tion. Effective antigen presentation requires both capa¬ city to bind relevant nominal antigen peptide and capacity to bind and trigger a relevant MHC-restricted nominal antigen peptide-specific T-cell receptor. Even in the case of GPI-modified MHC derivatives, not all of them will be able to fulfill both requirements. Such MHC molecules can be determined by routine procedure. In a preferred embodiment of the present invention, a recombinant MHC heterodimer is produced in a cell that expresses empty MHC complexes at the cell surface, that is, MHC complexes free of nominal antigen peptide. This facilitates the loading of an exogenous antigen peptide onto purified recombinant MHC heterodimer, since there is no resident peptide to be displaced. The recombinant MHC molecule is purified by immunoaffinity chromatography or standard biochemical purification procedures. In the case of a GPI-modified MHC heterodimer, it is ready at this stage for exogenous nominal antigen peptide loading (prior to reincorporation into a cell) or for reincorporation into a cell surface with simultaneous or subsequent nominal antigen peptide loading. Another significant feature of the present invention is the ability to control the amount of MHC delivered to the cell surface. This is a consequence of the dose-dependence of the coating process.

Moreover, in the case of GPI-modified MHC, relatively small amounts (in the picomolar range) of recombinant protein are needed to achieve functional MHC molecular densities. Yet another significant feature of the present invention is the simplicity of the delivery process. Since empty MHC heterodimers are being used, there is no prerequisite for prolonged co-incubation of the MHC molecule and the nominal antigen peptide, although the degree of loading can be maximized by prolonging contact between the two. One simply combines a purified MHC:GPI and a nominal antigen peptide in a standard buffer, and then either co-incubates the two for several hours prior to adding the mixture to cells or immediately adds the mixture to cells in a medium that can support cell via- bility. The kinetics of MHC:GPI membrane reincorporation are rapid, with detectable reincorporation within minutes. Optimal reincorporation is achieved after one to two hours, and is best accomplished in the absence of serum or excess hydrophobic peptides. The coated cells are ready at this stage for immunotherapeutic applications.

In a second aspect the present invention features a method for producing a biological membrane having a defined MHC:nominal antigen peptide on its surface. The method can be practiced by contacting the biological mem- brane with a nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind the nominal antigen peptide. Alternatively, the biological membrane can be first contacted with an artificial lipid-modified MHC polypeptide able to bind the nominal antigen peptide and second contacted with the nominal antigen peptide so that the artificial lipid-modified MHC polypeptide binds the peptide. In a third variation, the nominal antigen peptide is first contacted with the artificial lipid- modified MHC polypeptide able to bind the nominal antigen peptide and the biological membrane is then contacted with the nominal antigen peptide bound to the lipid-modified MHC polypeptide.

A "biological membrane" can be either isolated from a cell or artificially produced which is capable of interacting with a T-cell, and activating or inhibiting the T-cell. Methods for preparing membrane extracts from cells are numerous and well-known to those familiar with the art. For example, sonication and differential ultra- centrifugation can be combined to prepare enriched mem¬ brane fractions. Similarly, methods for producing biolog¬ ical membranes de novo are widely known, for example, com- bining lipid and non-lipid components to generate lipo- somes. Lipid-modified MHC molecules, such as a GPI- modified MHC molecule, can be incorporated into membrane extracts or liposomal membranes by simply combining the two in a standard buffer. Alternatively, one can incor- porate a lipid-modified MHC molecule into a cell surface prior to preparing a membrane extract from that cell. Also, a lipid-modified MHC molecule can be combined with the other ingredients in the process of liposome forma¬ tion. Optimizing the liposome formation process in this context for any given MHC:GPI is a straightforward pro¬ cess.

By "membrane extract" is meant an extract of a cell enriched for membranes, but not necessarily containing only membranes. Such an extract is chosen because it will have the immunogenic properties necessary to induce an immune response in vivo or ex vivo. By "membrane" is meant a sheet, usually about 10 nm thick and normally com- posed of a bimolecular layer of lipid and protein, enclo¬ sing or partially enclosing a cell, organelle, or vacuole. In preferred embodiments of these two aspects of the invention the MHC polypeptide comprises a class I MHC polypeptide sequence; the MHC polypeptide comprises a class II MHC polypeptide sequence; the lipid-modified MHC polypeptide comprises a glycosyl-phosphatidylinositol- modified MHC polypeptide; the lipid-modified MHC polypep¬ tide comprises a poly-histidine tag, epitope tag or other appended amino acid sequence designed to simplify purifi¬ cation or detection of said polypeptide; the glycosyl- phosphatidylinositol-modified MHC polypeptide is produced by gene transfer of a chimeric gene expression construct comprising a GPI modification signal into a host cell and isolation of the glycosyl-phosphatidylinositol-modified MHC polypeptide from the host cell.

By "class I MHC polypeptide sequence" is meant an amino acid sequence corresponding to a portion of the extracellular domain of a class I major histocompatibility complex heavy chain, for example, an HLA-A, HLA-B, or HLA- C heavy chain.

By "class II MHC polypeptide sequence " is meant an amino acid sequence corresponding to a portion of the extracellular domain of either a class II major histocom- patibility complex alpha chain or a class II major histo¬ compatibility complex beta chain, for example, an HLA-DQ, HLA-DR, or HLA-DP alpha or beta chain.

By "glycosyl-phosphatidylinositol-modified MHC peptide is meant an MHC polypeptide which has a covalently attached glycosyl-phosphatidylinositol molecule, so as to allow membrane insertion.

By "poly-histidine tag" is meant two or more clustered histidines that are inserted into a polypeptide sequence in order to permit purification of said polypeptide by nickel-sepharose chromatography. In the case of a glycosyl-phosphatidylinositol-modified polypeptide, Applicant has discovered that an optimal site for inser- tion of the polyhistidine amino acid sequence is in be¬ tween the sequence for the polypeptide of interest, for example, an MHC polypeptide sequence, and the GPI modifi¬ cation signal sequence. In the case of GPI-modified poly- peptides, the N-terminal signal peptide and C-terminal GPI moiety preclude polyhistidine insertion into the conven¬ tional N-terminal and C-terminal sites. Insertion of the polyhistidine sequence is accomplished by inserting the coding sequence for this polyhistidine sequence into the desired site of an MHC expression construct.

By "chimeric gene expression construct" is meant an expression vector comprising a coding sequence comprising two coding sequences linked in-frame that are not natu¬ rally linked to each other. By "GPI modification signal" is meant an amino acid sequence, generally derived from a natural GPI-modified polypeptide, that upon artificial insertion into another polypeptide results in GPI modification of said second polypeptide. By "host cell" is meant a cell that is capable of expressing a transfected gene.

By "isolation" is meant purification of the glycosyl- phosphatidylinositol-modified MHC polypeptide from the host cell. Possible purification methods include affinity chromatography, ion-exchange chromatography, size- exclusion chromatography, hydrophobic chromatography, and salting-out procedures.

In a third aspect the invention features a method for activating an antigen-specific T-cell comprising the step of contacting a cell having a defined MHC: nominal antigen peptide complex on its membrane with a T-cell population comprising a T-cell with specificity for said MHC:nominal antigen peptide complex.

By "activating an antigen-specific T cell" is meant stimulating a T cell in a manner that results in T cell proliferation or triggering of T cell cytotoxicity or triggering T cell cytokine production, for example, interleukin-2 and interleukin-2 receptor production. Generally, it is believed that in order to activate a T cell, two signals comprising an antigen-specific signal, for example, a signal delivered by an MHC:nominal antigen peptide complex, and a non-specific signal, for example, a signal delivered by a cell surface costimulator, must be conveyed to the T cell .

By "contacting" is meant bringing one cell into proximity with a second cell in a way that permits the first cell to modulate the second cell. In the present invention, the contacting step can be effected ex vivo or in vivo. According to the present invention, an antigen- presenting cell exogenously coated with a lipid-modified MHC:nominal antigen peptide complex is contacted with a T- cell to induce activation of said T-cell. Methods for contacting antigen-presenting cells with T-cells in order to induce activation of said T-cells ex vivo or in vivo are described extensively in the literature. These stan¬ dard T-cell activation methods can be directly applied in the use of lipid-modified MHC:nominal antigen peptide com¬ plexes. Numbers of cells to be used, time of incubation, and other conditions are well known to those familiar with the art, as are methods for optimizing stimulation condi¬ tions and tailoring them for specific antigenic systems and clinical therapies.

By "T-cell with specificity for the MHC:nominal antigen peptide complex "is meant a T-cell that bears a T- cell receptor that can bind an MHC:nominal antigen peptide complex in a manner that leads to T-cell activation. In preferred embodiments contacting is performed ex vivo; contacting is performed in vivo; the MHC polypeptide is syngeneic with said antigen-specific T-cell; the MHC polypeptide is allogeneic with said antigen-specific T- cell; the cell is a dendritic cell; the cell is an acti- vated B-cell; the cell is another cell that bears a costim¬ ulator and is thereby capable of activating an antigen- specific T-cell; the cell is presenting a tumor antigen. By "ex vivo" is meant outside of the body of a patient to be treated. A preferred embodiment of the present invention is the use of an antigen-presenting cell coated with a lipid-modified MHC:nominal antigen peptide complex for purposes of producing large numbers of antigen- specific T-cells. Such T-cells can then be administered to patients as a cellular therapeutic. Patients in need of such therapeutic T-cells include ones who suffer from viral diseases, cancer, and immunodeficiency disorders. For instance, it is recognized that clinical progression to chronic active hepatitis in patients infected with hepatitis B virus may result from a deficiency in certain T-cells with antigen-specificity for defined hepatitis B virus nominal antigen peptides. Such patients are good candidates for ex vivo therapy with antigen-specific T- cells. To treat a patient in need of therapeutic T-cells, antigen-presenting cells, such as dendritic cells, are first recovered' from the patient's peripheral blood. In the case of dendritic cells, their numbers are amplified ex vivo by adding cytokines, for example, a combination of GM-CSF and interleukin-4 as described (Romani et al. , 180 J. Exp. Med. 83, 1994) . A preparation consisting of an MHC:GPI:nominal antigen peptide comprising MHC and nominal antigen peptide components that are relevant to the patient being treated, for example, in the case of a chronic active hepatitis patient bearing the common HLA- A2.1 allele, a combination of HLA-A2.1 and hepatitis B virus peptides as disclosed in the example below, are added to the cells in culture. T-cells from the patient are then co-cultured with the MHC:GPI: nominal antigen peptide-coated antigen-presenting cells in order to amplify the numbers of antigen-specific T-cells. The latter cells are then administered to the patient in need of such therapeutic T-cells. By "in vivo" is meant in the body of a patient to be treated. A preferred embodiment of the present invention is the use of an antigen-presenting cell coated with a lipid-modified MHC:nominal antigen peptide complex for purposes of inducing the proliferation of antigen-specific antigen-presenting cells in a patient. The modified antigen-presenting cells are administered directly to a patient as a cellular therapeutic. Patients in need of such therapeutic T-cells include ones who suffer from viral diseases, cancer, and immunodeficiency disorders. For instance, it is recognized that T-cell anti-tumor responses can be beneficial to a patient who suffers from a cancer. For certain cancer types, there is information available as to specific tumor antigen peptides, with well-defined MHC allelic restriction, that can elicit T- cell antitumor responses, for example, an HLA-A2.1-re¬ stricted melanoma nominal antigen peptide. Hence, to treat such a patient, a preferred method is to isolate and amplify dendritic antigen-presenting cells from said patient's peripheral blood, coat the cells with the rele¬ vant MHC:GPI :nominal antigen peptide complex, and admin¬ ister the modified antigen-presenting cells back to the patient. These modified antigen-presenting cells function as a cellular immunogen in the patient to generate anti¬ gen-specific T-cells with therapeutic benefit.

By "syngeneic" is meant that the antigen-specific T- cell bears at least one MHC allelic variant that is identical to that of the MHC molecule that is being transferred.

By "allogeneic" is meant that the antigen-specific T- cell does not bear any MHC allelic variants that are identical to that of the MHC molecule that is being transferred.

By "dendritic cell " is meant a well-described cell generally derived from peripheral blood that has potent antigen-presenting capacity and grows with a dendritic morphology in culture. By "activated B-cell" is meant a B-cell that has been triggered with a polyclonal stimulator, for example, lipopolysaccharide . By "presenting a tumor antigen" is meant bearing a nominal antigen peptide corresponding to a tumor-specific polypeptide.

In a fourth aspect the invention features a method for inhibiting an antigen-specific T-cell comprising the step of contacting a cell having a defined MHC: nominal antigen peptide complex on its membrane with a T-cell population comprising a T-cell with specificity for said MHC:nominal antigen peptide complex. By "inhibiting" is meant modulating a T-cell in a manner that results in the inability of said T-cell to proliferate or to undergo triggering of T-cell cytotoxi¬ city or to undergo triggering of T-cell cytokine produc¬ tion, for example, interleukin-2 and interleukin-2 recep- tor production. A modulated T-cell in such an inactive state can be either apoptotic (undergoing programmed cell death) or anergic (unresponsive) .

In preferred embodiments contacting is performed ex vivo; contacting is performed in vivo; the MHC polypeptide is syngeneic with said antigen-specific T-cell; the MHC polypeptide is allogeneic with said antigen-specific T- cell; the cell is expressing a coinhibitor; the cell is lacking a costimulator.

By "expressing a coinhibitor" is meant that a cell surface molecule capable of delivering an inhibitory signal to a T-cell, for example, the coinhibitor CD8, is present on the cell surface. Applicant has previously established that an antigen-presenting cell can be con¬ verted from a T-cell activator to a T-cell inhibitor by expressing on said antigen-presenting cell a cell surface coinhibitor such as CD8. This modified antigen-presenting cell is referred to by Applicant as an "artificial veto cell". Furthermore, it is believed that an artificial veto cell comprising the CD8 coinhibitor triggers apop- tosis in antigen-specific T-cells. Thus, the present invention provides a means for producing artificial veto cells with well-defined antigenic properties. By coordi- nately coating a cell with a lipid-modified MHC:nominal antigen peptide complex and a coinhibitor, an artificial veto cell can be produced which is capable of deleting T- cells with specificity for said MHC-restricted nominal antigen peptide. When a syngeneic MHC polypeptide is used, the method can be applied for the treatment of an autoimmune disorder. Through the use of an allogeneic MHC polypeptide, this method can be used for inhibiting an allogeneic response for the treatment of an alloimmune disorder.

By "lacking a costimulator" is meant that a cell surface molecule capable of delivering an activating signal to a T-cell, for example, the costimulators B7-1, B7-2, B7-3, ICAM-1, is absent from the cell surface. Generally, it is believed that in order to activate a T cell, two signals comprising an antigen-specific signal, for example, a signal delivered by an MHC:nominal antigen peptide complex, and a non-specific signal, for example, a signal delivered by a cell surface costimulator, must be conveyed to the T cell. Antigen presentation in the absence of requisite costimulation leads to the induction of T-cell anergy. Thus, the present invention provides a method for producing a new category of artificial veto cells that are capable of inducing anergy in antigen- specific T-cells. A critical aspect of the present inven¬ tion is that it bypasses the need to use a cell that is able to process antigens on its own. This implies that essentially any cell can be used as a vehicle to present antigen. In turn, for purposes of T-cell inhibition, it is therefore possible to avoid conventional antigen- presenting cells which generally express costimulators and instead select other cells which are naturally costimu¬ lator negative. By coating such a cell with a lipid- modified MHC:nominal antigen peptide complex, one can generate an artificial veto cell. Three candidate cell types for this purpose are hepatocytes, fibroblasts, and erythrocytes. In a fifth aspect the invention features a glycosyl- phosphatidylinositol-modified polypeptide which is separate from a cell membrane.

By "separate from a cell membrane" means not incorporated in a cell membrane and located external to a cell, as opposed to contained internally.

In a sixth aspect the invention features a cell having a membrane exogenously coated with lipid-modified MHC polypeptide. Exogenous coating of the cell may result in a MHC polypeptide with a cell surface topology (e.g. association with proteins) that is distinct from a cell that has been genetically engineered to contain a MHC polypeptide on its surface (insertion into the membrane via an intracellular route) .

In a seventh aspect the invention features a modified antigen-presenting cell comprising a cell with an artificial costimulator polypeptide exogenously reincor- porated into the surface of the cell. By "modified antigen-presenting cell" is meant an antigen-presenting cell that has a non-natural molecular moiety on its surface. For example, such a cell may not naturally have such a costimulator on its surface or may have additional artificial costimulator in addition to natural costimulator on its surface.

By "artificial costimulator polypeptide" is meant a molecule that functions as a costimulator and has the capacity to be exogenously reincorporated into a cell membrane. Such a molecule may include less than an entire naturally occurring costimulator molecule, or be a varia¬ tion on a naturally occurring costimulator molecule, but still retains the ability to act as a costimulator. Such polypeptides can be determined by routine procedures.

By "exogenously reincorporated" is meant the external portion of the cell membrane is contacted with the artificial costimulator molecule such that it becomes bound to that surface. This is in contrast to intracel- lular insertion of these molecules into the membrane, which occurs naturally or with genetically engineered molecules expressed within the cell.

In preferred embodiments the modified antigen- presenting cell is an immunogenic tumor cell; the arti¬ ficial costimulator polypeptide comprises a glycosylphos- phatidylinositol moiety; artificial costimulator polypep¬ tide comprises a streptavidin moiety; artificial costimu¬ lator polypeptide comprises a molecule with affinity for a molecule on the surface of the cell; the modified antigen-presenting cell is pre-coated with a molecule with affinity for an artificial costimulator polypeptide; the costimulator portion of the artificial costimulator poly¬ peptide is selected from the group consisting of B7-1, B7- 2, B7-3, intracellular adhesion molecule-1, intracellular adhesion molecule-2, intracellular adhesion molecule-3, LFA-3, VCAM-1, and fibronectin.

By "immunogenic tumor cell" is meant a modified tumor cell such that the tumor cell is now immunogenic. Most naturally occurring tumor cells are not immunogenic.

By "glycosyl-phosphatidylinositol moiety" is meant a GPI molecule attached to the costimulator polypeptide. The GPI moiety can be attached to the costimulator as described for an MHC molecule. By "strepavidin moiety" is meant any molecule or portion thereof able to bind avidin. Attachment of strep¬ avidin to a polypeptide is performed by standard proce¬ dures.

By "a molecule with affinity for a molecule on the cell surface" is meant a molecule which allows for the binding of the artificial costimulator to the cell sur¬ face. Such a molecule could be a second polypeptide that binds a receptor of the cell surface. Linkage of the second polypeptide to the costimulator could be by stan- dard procedures such as the use of a hybrid sequence encoding both molecules. In a eighth aspect the invention features a method for producing a modified antigen presenting cell comprising a costimulator on its membrane. The method is practiced by contacting the external portion of the cell with an externally applied artificial costimulator polypeptide.

In preferred embodiments the artificial costimulator polypeptide comprises a glycosyl-phosphatidylinositol moiety; the costimulator portion of said artificial costimulator polypeptide is selected from the group con- sisting of B7-1, B7-2, B7-3, intracellular adhesion molecule-1, intracellular adhesion molecule-2, intracel¬ lular adhesion molecule-3, LFA-3, VCAM-1, and fibronectin. In a ninth aspect the invention features a method for enhancing a cell's immunogenicity comprising contacting the cell with a membrane-binding artificial costimulator polypeptide.

By "membrane-binding" is meant the polypeptide has the capacity to noncovalently or covalently attach to a biological membrane. In an tenth aspect the invention features a method for inducing anti-tumor immunity against an autologous tumor cell in a patient by administering to the patient a tumor cell or tumor cell membrane extract with externally reincorporated costimulator polypeptide on its membrane. By "autologous tumor cell" is meant a tumor cell of the same tumor type that is derived from the patient undergoing treatment.

By "tumor cell membrane extract" is meant a membrane extract, as defined above, derived from a tumor cell. In preferred embodiments the tumor cell is an autologous tumor cell, the tumor cell is a heterologous tumor cell.

By "heterologous tumor cell" is meant a tumor cell of the same tumor type that is derived from an individual other than the patient.

In a eleventh aspect the invention features a method for identifying a molecule that enhances the immuno- genicity of an antigen presenting cell. The method encompasses coating the antigen presenting cell with a membrane binding polypeptide derivative of the molecule and determining the immunogenicity of the resulting cell compared to an uncoated cell.

By "coating" is meant exogenously incorporating a polypeptide into the external aspect of a cell membrane.

By "enhancing immunogenicity" is meant increasing the immunogenic potential of a cell. In preferred embodiments the antigen presenting cell is a tumor cell; determining the immunogenicity comprises determining the capacity of the coated tumor cell to induce an antitumor immune response.

By "antitumor immune response" is meant an immunological cell-mediated or antibody-mediated response directed against a tumor cell.

A significant aspect of this invention is the discovery that protein transfer can be used to deliver a functional MHC polypeptide or costimulator polypeptide to a cell surface. Previously only gene transfer was used for this purpose. The use of protein transfer for effec¬ ting said delivery process provides substantial advantages for engineering therapeutic APCs, since it bypasses the significant limitations imposed by those delivery methods based upon gene transfer. Advantages of protein transfer over gene transfer for APC engineering include, but are not limited to, the following.

First, most APC candidates, by virtue of being primary cells (that is, nontransformed) with low proliferative rates, are not amenable to efficient gene transfer. In contrast, primary cells can readily be coated with exogenously added artificial polypeptides designed for efficient membrane reincorporation and cell surface coating. Thus, the present invention can be practiced with a greater variety of cells, and most importantly with cells that are APC candidates . Second, nucleic acid, such as DNA, that is exogenously introduced into cells is often unstable, with concomitant loss of expression of the encoded RNA product. In contrast, protein that is exogenously introduced into cellular membranes has defined stability characteristics. Third, transfecting the same cell with multiple genetic constructs is a cumbersome process. In fact, from a practical standpoint, cotransfecting more than two genes into even a transfectable cell constitutes a major under- taking. Hence, this approach is not well-suited for the clinical setting, where many samples of often difficult to transfect cells are to be processed. In contrast, protein transfer allows for the simultaneous delivery of any number of membrane-binding polypeptides to the cell sur- face, with no additional burden imposed when additional molecules are to be included. The latter simply requires the addition of more molecules into the mixture. Hence, multiple MHC:nominal antigen peptide complexes or costimu¬ lators can be coordinately delivered to the same cell by protein transfer.

Fourth, certain gene transfer-based immunotherapeutic strategies require selection for stable trans ectants. This can be a time consuming process and complicates the clinical practice of such methods and imposes a delay period between biopsy and treatment. In contrast, protein transfer is a relatively rapid process and does not impose a burden of excessive cell culturing, shortening the interval between obtaining of cells from the patient and treatment. Fifth, there is considerable variability in the number of molecules that are expressed at the cell surface of transfected cells. In contrast, protein transfer provides for fine control of the degree of coating, with the possibility of either very low or high levels of protein delivery to the cell surface, and more uniformity with respect to the number of molecules on individual cells of the population. Sixth, gene transfer is dependent in most instances upon genetic vectors comprising viral components which carry with them some degree, albeit small, of biosafety hazard. In contrast, no biosafety hazards are associated with polypeptide derivatives to be used for practicing protein transfer onto APCs. The polypeptide derivatives of the present invention are designed to incorporate molecular components that are, in most instances, derived from native polypeptides, and hence, low toxicity and immunogenicity constitute a fundamental feature of such polypeptide derivatives.

The method of the present invention also overcomes several of the limitations of the conventional method of loading oligopeptides onto APCs. First, in order to replace pre-engaged, endogenously- loaded peptides from the antigen binding grooves of MHC molecules, in particular class I MHC molecules, by in situ loading at the cell surface, a large excess of exogenous peptide is required. Under such circumstances, there is a lack of fine control of the final density of any given MHC:nominal antigen peptide complex at the cell surface. This represents a serious limitation since there is evidence that MHC:nominal antigen peptide complex density can dictate the balance between T-cell activation and inhibition. Second, certain peptides cannot be efficiently loaded at all. Third, one is restricted to the use of cells already bearing relevant MHC molecules and to the levels of specific MHC products that naturally occur at the cell surface. The protein transfer method of the claimed invention allows for the coating of any cell with any relevant MHC molecule and any nominal antigen peptide that can bind to that MHC molecule. Also, the method allows for fine control of the MHC:nominal antigen peptide complex density on the surface of an APC as the levels of various reagents can be precisely controlled. Other features and advantages of the invention will be apparent from the following description of the preferred embodiments thereof, and from the claims.

Brief Description of the Drawings The drawings will first briefly be described.

Figure 1 is a schematic drawing of expression con¬ structs used for HLA-A2.l:GPI//3₂m and HLA-A2. l/jS₂m heterodimers in Schneider S2 cells.

Figure 2 is a silver-stained SDS-PAGE gel which documents the identity of purified HLA-A2.1 :GPI/ ₂m and HLA-A2.1//S₂m heterodimers. Designations to the right cor¬ respond with the HLA-A2.1 heavy chain (HLA-A2.1) , HLA- A2.1:GPI heavy chain (HLA-A2.1:GPI) , and human β₂ (β₂ ) . Figure 3 is a bar graph plotting data from a sandwich ELISA which documents heterodimerization in both purified

HLA-A2.l:GPI/|S₂m and HLA-A2.l/jβ₂m. Differing quantities of

HLA-A2.1//S₂m (cross-hatched boxes) and HLA-A2.1 :GPI//?₂m

(filled boxes) heterodimers bound to PA2.1 antibody pre-coated microtiter wells are shown on the x-axis. HRP activity from bound horseradish peroxidase-conjugated goat anti-human β₂m as measured by absorbance at 490 nm is shown on the y-axis.

Figure 4 are tracings from flow cytometric and immunofluorescence analyses that compare protein transfer of empty HLA-A2.1:GPI/jS₂m heterodimers under different experimental conditions. The data are plotted as number of cells (y-axis) versus log fluorescence intensity (x- axis) . Figure 4A shows the time course of HLA-A2.1 :GPI/S₂m protein transfer. Incubation time were 0 minutes (light solid line) , 1 minute (long broken line) , 10 minutes (medium broken line) , 30 minutes (short broken line) and 60 minutes (heavy solid line) . Figure 4B shows the dose- dependence of HLA-A2.1: GPl/3₂m protein transfer. Amounts of HLA-A2.1:GPI/S₂m are: 0 pmoles (light solid line) ; 0.1 pmoles (short broken line) ; 0.2 pmoles (medium broken line) ; and 1 pmole (heavy solid line) . Figure 4C docu- ments that PI-PLC treatment releases protein-transferred HLA-A2.1 :GPI/jS₂m from the cell surface. Coated cells were incubated at 4'C or 37^*C ± PI-PLC: CIR + PI-PLC (light solid line) ; CIR + HLA-A2.1 :GPI + PI-PLC, 37°C (medium broken line) ; CIR + HLA-A2.1:GPI, 37°C (short broken line) ; and CIR + HLA-A2.1 :GPI, 4°C (heavy solid line) . Figure 4D shows the effect of loaded peptide upon HLA-A2. - l:GPI//3₂m protein transfer: CIR (light solid line) ; CIR + HLA-A2.1.GPI (heavy solid line) ; CIR + H A-A2.1 :GPI + MA58-66 (short broken line) ; and CIR + HLA-A2.1.GPI + HBV env 335-343 (medium broken line) .

Figures 5 consists of bar graph plotting data that document CTL recognition of protein-transferred HLA-A2.1 :GPl/jS₂m/peptide heterotrimeric complexes. Percent specific lysis is plotted on the x-axis. Femtomo- les of HLA-A2.1:GPI/S₂m or HLA-A2. l/jS₂m, plus or minus peptide (MA58-66 or HBV env 335-343) are indicated on the y-axis. Error bars correspond to one standard deviation.

Figure 6 consists of a bar graph that compares different HLA-A2.1 :GPI/3₂m peptide loading regimens (either during a pre-incubation period or during protein transfer only) . Cytotoxicity was assessed without (left) or with

(right) 1.25 μg/Ml HBV env335-343 peptide added to CIR targets during the CTL assay. The x-axis indicates per- cent specific lysis. The y-axis indicates the loading regimens. Error bars correspond to one standard deviation.

Figure 7 consists of a graph plotting data from a cytotoxicity assay demonstrating that artificial GPI- modified B7-1 functions well as a costimulator. Various effector cells were used: 1 EL4 (solid square) ; 2 EL4

(solid circle) ; 3 MB7-DS/EL4 (solid triangle) ; and 4 MB7-

DS/EL4 (solid diamond) . Effector- to-target (E:T) ratios are shown on the x-axis. Percent specific lysis is plot- ted on the y-axis. Error bars correspond to one standard deviation. Description of the Preferred Embodiment

Protein transfer of MHC:GPI complexes to cell surfaces

The present invention addresses the need for delivering antigenic complexes in a controlled fashion to the surfaces of diverse cell types. Methods are provided for coating cells with lipid-modified MHC:nominal antigen peptide complexes.

In the experiments illustrated by the following examples, recombinant glycosyl-phosphatidylinositol (GPI)- modified HLA-A2.1 (HLA-A2.1:GPI/3₂m) was used as a protein transfer vehicle to deliver a hepatitis B virus (HBV) nominal antigen peptide to the surfaces of cytotoxic T- cell targets. The model system utilized an HLA-A2.1- restricted hepatitis B virus (HBV) peptide and well- defined human T-cell clones with specificity for this peptide. This peptide is one of a subset of HBV peptides that previous studies have defined as being dominant in T-cell responses to HBV. Clinical correlations have suggested that lack of CTL responses to these HBV peptide determinants may be an important factor in clinical pro¬ gression to chronic active hepatitis (Penna et al. , 174 J. Ex . Med. 1565, 1991; Nayersina et al. , 150 J. Immunol. 4659, 1993) .

Empty HLA-A2.1:GPI/_/β₂m was first produced in D. melanogaster co-transfectants and immunoaffinity purified. Cell coating with HLA-A2.1:GPI/β₂m was shown to occur rapidly, and to be protein concentration-dependent. Protein-transferred HLA-A2.1:GPI/_/S₂m effectively presented an HBV peptide to peptide-specific, HLA-A2.1-restricted T-cell clones in cytotoxicity assays. Protein transfer of GPI-modified class I MHC:nominal antigen peptide complexes represent a novel strategy for delivering functional anti¬ genic complexes to cell surfaces which bypasses limita¬ tions of gene transfer and permits control of nominal antigen peptide densities at cell surfaces. Example 1 : Production of GPI-modified Human Class I MHC In order to produce HLA-A2.1 polypeptides devoid of nominal antigen peptides, expression was performed in Schneider S2 D. melanogaster cells. Schneider S2 cells, obtained from J. Incardona (Case Western Reserve Universi¬ ty) , were maintained in 67.5% M3 medium (Sigma) supple¬ mented with 20% 5X BPYE (12.5 g/1 bacto-peptone, 5 g/1 TC yeastolate) , 12.5% fetal bovine serum, and 20 U/ml penicillin, and 20 mg/ml streptomycin. Such cells have been shown to produce empty (nominal antigen peptide- free) , native HLA-A2.1 when cotransfected with HLA-A2.1 and human /S₂m coding sequences (Jackson et al . , 89 Proc. Natl. Acad. Sci. USA 12117, 1992; Jackson et al. , 263 Science 384, 1994) . Moreover, these cells are known to support GPI-modification (Hortsch and Goodman, 265 J. Biol. Chem. 15104, 1990) . To produce HLA-A2.1 :GPI, Schneider S2 cells were cotransfected by lipofection with a combination of three expression constructs (Figure 1) , pHLA-A2.1:DAF-S /Pac (encoding HLA-A2.1 :GPI heavy chain) , ph3₂m/Pac (encoding human jβ₂m) and pHph/Pac (encoding the selectable marker hph which confers hygromycin B- resis¬ tance) . Alternatively, to produce native HLA-A2.1, stable cotransfectants were generated in which pHLA- A2.l/Pac (encoding native HLA-A2.1 heavy chain) was substituted for pHLA-A2.1:GPI/Pac in the above combination of expression constructs .

The HLA-A2.1:GPI heavy chain .and HLA-A2.1 heavy chain, and β₂m expression constructs were produced by ligating coding sequences for HLA-A2.1, HLA-A2.1:DAF-S, and human jβ₂-microglobulin into the D. melanogaster actin 5C promoter-based expression vector, pPac (Krasnow et al., 57 Cell 1031, 1989) . DNA inserts for HLA-A2.1, HLA-A2.1 : GPI, and β₂m were excised from pHLA-A2.1/REP7/3 with Xho I, from pHLA-A2.1 :DAF-S/REP7β (Huang et al. , 31 Mol . Immunol. 1017, 1994) with Xho I and Bam HI, and ph3₂m/REP10 (Huang et al., 31 Mol . Immunol . 1017, 1994) by BamHI and Hind III digestion, respectively. Insert DNAs were blunted by Klenow fill-in reactions and ligated into the Klenow filled-in BamHI site of pPac.

S2 cells were co-transfected as follows: 30 μg of lipofectin (BRL) , 10 μg of Qiagen column-purified expression construct (a 1:1 mixture of ph_/S₂m/Pac plus either pHLA-A2.1 :DAF-S/Pac or pHLA-A2. l/Pac) , and 1 μg of pHph/Pac, to allow for selection by growth in hygromycin B, were combined with 10⁶ washed S2 cells in 1 mL of 1.25X M3 medium (Sigma) in polystyrene tubes. After 3 hours at room temperature, cells were washed with 10 mL complete M3 medium (67.5 % 1.25X M3 media, 20% 5X BPYE (12.5 g/L bacto-peptone/5 g/L yeast extract) , and 12.5% fetal bovine serum and resuspended in 10 mL of complete medium. After three days, selection was initiated by adding hygromycin B (CalBiochem) to 0.1 mg/ml with fresh selective medium which was replaced every 3-4 days. Transfected cell colonies were evident at three weeks and were cloned from soft agar. Transfectant cell lines were screened for expression of cell surface HLA-A2.1 epitopes by immuno- staining with the HLA-A2.1, -Aw68-specific monoclonal antibody, PA2.1 (hybridoma cell line purchased from the American Type Culture Collection) , followed by FITC- conjugated goat anti-mouse immunoglobulin and analysis by flow cytometry (FACSCAN, Becton-Dickinson) . GPI- modifi¬ cation of HLA-A2.1:GPI was confirmed by treatment of transfected cells with 10 units of PI-PLC prior to PA2.1 immunostaining as described (Huang et al. , 31 Mol. Immuno- __. 1017, 1994) . For quantitative production of HLA-A2.1:GPI/_/S₂m and HLA-A2.1/jβ₂m, 2-5 X 10¹⁰ S2 transfectant cells from 8-16 L of culture were processed batchwise for immunoaffinity purification by PA2.1-conjugated Sepharose CL-4B chromato¬ graphy (Parham, 92 Meth. Enzvmol . 110, 1983) . Elution fractions containing heterodimer were identified using both SDS-PAGE analysis and antigen-capture ELISA specific for HLA-A2.1. Typical yields were approximately 1 to 10 μg recombinant HLA-A2.1 :GPI//3₂m or HLA-A2. l/_/S₂m heterodimer per batch with purity estimated at 60% by scanning densi- tometry. Neutralized immunoaffinity chromatography frac¬ tions were used directly in assays. Empty HLA-A2.1:GPI or HLA-A2.1, each complexed with β₂m, were immunoaffinity purified using the HLA-A2.1 heavy chain-specific monoclonal antibody PA2.1 conjugated to sepharose CL-4B (Parham, 92 Methods Enzymol . 110, 1983) . The association of β₂τ with either HLA-A2.1:GPI or HLA-A2.1 heavy chains in respective cotransfectants was demonstrated by subjecting immunoaffinity-purified materi¬ al to SDS-PAGE analysis (Figure 2) . Approximately 1 ng of HLA-A2.1 (lane 1) and HLA-A2.1:GPI (lane 2) heterodimers, immunoaffinity purified using the monoclonal antibody PA2.1, were separated on reducing, 15% acrylamide SDS-PAGE gels, according to standard conditions, and visualized by silver staining. As can be seen in both lanes 1 and 2, molecular specie's corresponding to both HLA-A2.1 or HLA- A2.1:GPI and β₂m are present in both lanes, indicating that hetorodimers were formed.

The association of β₂ with either HLA-A2.1:GPI or HLA- A2.1 heavy chains was also demonstrated using a sandwich ELISA (Parker et al. , 29 Mol. Immunol . 371, 1992) incorporating both anti-heavy chain and anti-3₂m antibodies (Figure 3) . Molecules detected by this particular ELISA bear both heavy chain and jβ₂m epitopes.

Sandwich ELISA assays were performed as follows: 1 μg of the HLA-A2.1, -Aw68 specific monoclonal antibody, PA2.1, was coated per well of 96 well Immulon 4 (Dynatec- h) plates by overnight incubation at 4^*C. After washing with ddH₂0, plates were blocked with 0.25% (w/v) bovine serum albumin plus 0.05% (w/v) tween-20 overnight at 4^*C. Wells were washed thrice with ddH₂0 prior to use. Heterod¬ imers were diluted into blocking buffer and captured for 90 minutes at room temperature. Wells were washed thrice with ddH₂0 and captured antigens detected with 1:1000 diluted rabbit anti-human β₂ra immunoglobulin conjugated to horseradish peroxidase (Accurate) and incubated at room temperature for 30 minutes. Wells were washed thrice and ELISA wells developed with 0.1 mL of 0.6 mg/mL ortho-phen- yl-diamine (Sigma) in 0.1 M sodium citrate/0.01% (w/v) H₂0₂ at room temperature for 15-20 minutes. Reactions were terminated with 2N H₂S0₄. Well absorbance was measured at 490 nanometers.

Example 2 : Protein Transfer of HLA-A2.1:GPI/β.,m Heterodimers Immunoaffinity-purified HLA-A2.1 :GPI/_/S₂m heterodimers were evaluated for their reincorporability into cell membranes. Co-incubation at 37°C of HLA-A2.1: GPI /β₂m with the HLA-A-negative, HLA-B-reduced B lymphoblastoid cell line, CIR, conferred HLA-A2.1 epitopes to CIR cell surfaces. CIR cells (Storkus et al. , 138 J. Immunol. 1657, 1987) were grown in RPMI-1640 supplemented with 10% fetal bovine serum, 2 mM glutamine, and 20 U/ml penicil¬ lin, and 20 mg/ml streptomycin. The transfer of HLA-A2.1 : GPI/3₂m molecules onto CIR cells was dependent upon co- incubation time, temperature, HLA-A2.1:GPI/jβ₂m concentra¬ tion, and fetal bovine serum concentration.

To examine transfer dependence on time, washed CIR cells were coated with HLA-A2.1 :GPI/_/S₂m by a protein transfer method (vide infra) . Approximately 4 pmoles of HLA-A2.1:GPI/ β₂m was combined with washed CIR cells at 37°C for different incubation times (0 minutes, 1 minute, 10 minutes, 30 minutes, 60 minutes) , and coated cells were then washed twice with lxPBS/0.1% BSA/0.05 % NaN₃ and immunostained with either 1 μl of PA2.1 ascites or 1 μl of BB7.1 ascites as negative control. BB7.1 staining was negative (not shown) . Reincorporation was detected within one minute of combining HLA-A2.1:GPI/3₂m with cells, and HLA-A2.1 epitope density, as monitored by immunofluores- cence and flow cytometry, increased steadily over a one hour co-incubation period at 37°C (Figure 4A) . In con- trast, less protein transfer was observed at either 4'C or 25°C (data not shown) .

To examine transfer dependence on HLA-A2.1 :GPI/ β₂m concentration washed CIR cells were incubated with varying amounts (0 pmoles, 0.1 pmoles, 0.2 pmoles, 1 pmole) of HLA-A2.1 :GPI//3₂m at 37^*C for one hour and processed for immunostaining with PA2.1 or BB7.1 antibody. BB7.1 staining was negative (not shown) . Increasing the HLA-A2.1 :GPI/3₂m concentration resulted in a dose-dependent increase in HLA-A2.1 epitope bound to CIR cell surfaces (Figure 4B) .

In parallel experiments, two additional HLA- A2.1- negative cell lines, the human chronic myeloid leukemia cell line K562 and the murine thymoma cell line EL-4 were comparably coated with equivalent amounts of HLA-A2.1:GPI//3₂m (data not shown) .

The influence of peptides on GPI-modified protein re¬ incorporation was also determined. HLA-A2.1:GPI/_/S₂m was pre-incubated overnight on ice with 5 μg/mL of either MA58-66 or HBV env335-343 peptide, as indicated in the legend inset. Reactants were added to washed CIR cells, and the treated cells were immunostained with PA2.1 or BB7.1 prior to flow cytometry. No immunostaining with BB7.1 was observed (not shown) . Short peptides at concentrations below 5 μg/mL generally had no effect on the efficiency of protein transfer (Figure 4D) . However, for one peptide, MA58-66, some inhibition was observed at higher (>5 μg/mL) peptide concentrations (data not shown) , perhaps an effect of its relative hydrophobicity compared to the other peptides examined and potential hydrophobic interactions with the GPI anchors.

GPI anchorage of reincorporated HLA-A2.1:GPI was substantiated by enzymatic cleavage analysis. Four pmoles of HLA-A2.1:GPI//3₂m was protein transferred onto washed CIR, cells and coated cells were then incubated at 4"C or 37"C with or without PI-PLC. As a negative control, uncoated CIR cells were also treated with PI-PLC at 37°C. Cells were processed for immunostaining with PA2.1 or BB7.1 and subsequent flow cytometry. No immunostaining with BB7.1 was detected (not shown) . As expected, PI-PLC treatment of HLA-A2.1 :GPI/jS₂m-coated CIR cells released the HLA-A2.1 epitopes from cell surfaces (Figure 4C) . In addition, protein transfer of HLA-A2.1: GPI was inhibited by inclusion of 10% fetal bovine serum in the co- incubation mixture (data not shown) , but once reincor¬ porated, HLA-A2.1:GPI//S₂m could not be stripped from the cell surface by washing with 10% fetal bovine serum. In contrast, insect cell-derived native HLA-A2. l//?₂m non- specifically adsorbed to cell surfaces was readily strip¬ ped off by this treatment (data not shown) . Taken to¬ gether, these findings are consistent with GPI membrane anchorage for HLA-A2.1 :GPI//S₂m.

Example 3 : Coating of Cells With

HLA- A2.1:GPI/β.,m/peptide Cells were coated with HLA-A2.1:GPI molecules as follows. Approximately 20 femtomoles (-1 ng) of HLA- A2.1:GPI heterodimer in 50 μl were pre-incubated with an equal volume of lx PBS +. 0.5 μg/ml of peptide and incubated on ice overnight in a silanized microfuge tubes. Peptides were resuspended in DMSO at 20 mg/ml and diluted to 1 mg/mL in RPMI-1640 immediately prior to use. Cells to be coated were washed twice with lx PBS, resuspended at 1.25 x 10^β/mL in lxPBS, and 0.25-0.5xl0⁶ cells added to the HLA-A2.1:GPI/ β2m mixtures. Final volumes were 0.25-0.5 L. Tubes were rotated at 37"C for one hour and washed twice with lxPBS/ 0.5% bovine serum albumin/0.1% NaN₃ prior to staining with monoclonal antibodies for immunostaining and subsequent flow cytometry (FACSCAN, Becton-Dickinson) . Example 4 : Cytolytic T Lymphocyte Recognition of

Protein Transferred HLA-A2.1:GPI/ β₇m/peptide

Complexes

The capacity of protein transferred HLA-A2.1 : GPI/jβ₂m and HLA-A2.l//3₂m to engage and present nominal antigen peptides was assessed in a series of functional studies.

The experimental system used HLA-A2.1:GPI//S₂m- or

HLA-A2. l/S₂m-coated CIR cells as cytolytic T lymphocyte

("CTL") targets. In a first experiment designed to test peptide binding, HLA-A2.1 :GPl/jβ₂m or HLA-A2. l/_/S₂m heterodimers were pre-incubated with (or without) HLA-A2.1- restricted peptides overnight at 4^*C prior to protein transfer. Under the coating conditions used in this experiment, a modest level (about two-fold log fluorescence units over background) of cell surface HLA-A2.1 epitope was attained

(Figure 4D) . HLA-A2.1 :GPI/ β₂m + HBV envelope peptide (HBV env335-343) -coated, ⁵¹Cr-labeled CIR cells were efficiently lysed by a CD8\ HBV env335-343- specific, HLA-A2.1- restricted human T-cell clone (Figure 5) . HBV env335-343- specific T cell clones, obtained from F. Chisari, were derived and maintained as described (Nayersina et al . , 150 J. Immunol. 4659, 1993) . In contrast, there was no CTL recognition of CIR cells combined with native HLA-A2.l//3₂m and the same HBV env335-343 peptide. Similar results were obtained using a second distinct human CD8⁺ T-cell clone with the same antigenic specificity (data not shown) . Neither CTL clone lysed cells coated with HLA-A2.1 :GPI/3₂m heterodimers alone or HLA-A2.1 :GPl/3₂m heterodimers loaded with an irrelevant HLA-A2.1-binding peptide, namely, the influenza matrix peptide MA58-66. When a larger quantity of HLA-A2. l:GPI/3₂m/HBV env335-343 heterotrimeric complexes were coated onto CIR target cells, the extent of CTL- mediated cytolysis increased (Figure 5) . The specificity of CTL recognition of protein- transferred, peptide-loaded HLA-A2.1:GPI/3₂m was probed in additional experiments . An excess of the irrelevant HLA-A2.1-binding peptide MA58-66, added prior to protein transfer, effectively blocked cytolysis by HBV env335-343- specific CTL of coated CIR targets (data not shown) . Cytolysis was also shown to be HLA-A2.1-specific in monoclonal antibody blocking studies. Both an HLA-A2.1- specific antibody (MA2.1, hybrido a cell line purchased from the American Type Culture Collection) and a pan-class I HLA-specific antibody (W6/32, hybridoma cell line purchased from the American Type Culture Collection) independently blocked killing of HLA-A2.1:GPI/jS₂m/HBV env335-343-coated target CIR cells. An irrelevant HLA-B7-specific monoclonal antibody (BB7.1) had no inhibitory effect (data not shown) . Of note, cytolysis was also partially blocked using a human CDΘαr-specific antibody (Leu2a, hybridoma cell line provided by R. Evans, Roswell Park Memorial Institute) , consistent with partial CD8-dependence for the CTLs.

In the above experiments, HLA-A2.1:GPI//3₂m/ peptide heterotrimers were formed in the course of an overnight co-incubation at 4^*C prior to protein transfer.

Additional experiments were performed employing alter¬ native heterotrimer co-incubation strategies. In these experiments, heterodimers were incubated at 4'C overnight in the absence of peptide. Peptides were subsequently added during the one-hour protein transfer procedure and/or the CTL assay. As shown in Figure 6, inclusion of peptide during the protein transfer phase only was suf¬ ficient to permit significant specific CTL lysis. This is consistent with either rapid formation of MHC:peptide complexes in solution during protein transfer and/or binding of peptide to reincorporated MHC heterodimers in si tu . The latter possibility is supported by the addi¬ tional finding that addition of peptide to (peptide-free) heterodi er-coated, but not uncoated, cells during the CTL assay phase only also yields specific cytolysis (Figure 6) . The extent of CTL-mediated cytolysis of target CIR cells coated with pre-formed HLA-A2.1:GPI/_/S₂m/peptide heterotrimeric complexes was consistently greater (by 1.25- to two-fold) than that for CIR cells coated with heterodimers and later exposed to free peptide during protein transfer (Figure 6) . This likely indicates that a higher proportion of loaded HLA-A2.1:GPI/_/8₂m/peptide heterotrimers can be achieved by prolonged overnight incu¬ bation. Moreover, it is consistent with the notion that heterotrimeric complexes are forming prior to coincubation with cells. Additional evidence for the formation of pre¬ formed heterotrimeric complexes prior to protein transfer comes from a sandwich ELISA in which exogenous peptide was shown to increase β₂va-dependent conformational epitopes on the HLA-A2.1:GPI heavy chain (data not shown) .

CTL assays were performed as follows. Twice washed CIR cells were labeled with ⁵¹Cr for 1 h at 37^*C, and washed twice with lx PBS. Labeled cells were resuspended in lx PBS and combined with the pre-incubated heterodimer + peptide in silanized microfuge tubes. Final concen¬ trations were IO^{6 51}Cr-labeled CIR cells/mL, and unless indicated otherwise, 1 ng heterodimer/mL. Coated cells were washed twice with RPMI-1640 supplemented with 10% heat inactivated fetal bovine serum, 2 mM glutamine and penicillin/streptomycin. 5xl0³ coated target cells and CTL were added per well in 0.1 L each into conical bottom microtiter wells (Linbro) . For assays done in the pre¬ sence of peptide, coated target cells and peptide were added in 50 μl each. CTL clone 0'R875.20-44 was used at an effector to target ratio of 1. Spontaneous release was measured from samples incubated with 0.1 mL of medium instead of CTL. Total release was obtained from wells treated with 0.1 mL of 4% Triton X-100. All determina¬ tions were performed in triplicate. Plates were spun for 2 minutes at 400 x g and incubated 4 hours at 37^*C in a humidified 5% C0₂ incubator. Plates were centrifuged again and 0.1 L of supernatant harvested for gamma counting. Percent specific release = 100* (experimental release - spontaneous release) / (Total release - spontaneous release) .

These experiments define a novel approach for delivering pre-formed class I MHC/3₂m/peptide heterotrimers to cell surfaces that is based upon the use of GPI- modified, membrane-reincorporable class I heavy chains. A significant feature of the experimental system is the use of S2 Schneider insect cells to produce empty (peptide-free) HLA-A2.1, which presumably facilitates exogenous peptide loading.

By design, relatively small quantities of HLA- A2.1:GPI//3₂m/peptide heterotrimers were used in these experiments for the cell coating step. The goal here was to deliver limited amounts of HLA heterotrimers to cell surfaces. The immunofluorescence data indicate that fine control of heterotrimer densities can be achieved using recombinant GPI-modified heterotrimers. In view of recent data indicating that surface HLA complex density can influence the balance between activation versus inhibition during antigen presentation, such fine control is likely to be useful. Substantially higher surface densities can be achieved if desired by increasing the amount of GPI- modified proteins added to the cells (data not shown) . The findings of the present study with HLA-A2.1 are applicable to other class I allelic products, as well as to class II MHC polypeptides. In the case of GPI-modified class II MHC polypeptides, both- the alpha and beta chains are produced as GPI-modified polypeptide variants. When co-expressed in the same cell via gene transfer, the two GPI-modified polypeptide chains associate in the proper manner to constitute an antigen-binding pocket at the amino-terminal ends. In order to achieve this end, it is imperative to use similar GPI modification signal sequences for both polypeptide chains so that the molecular topology permits dimerization. Example 5 : Production of an artificial veto cell by protein transfer of an HLA-A2.1 :GPI/β-,m peptide complex

A fundamental aspect of the present invention is the capacity to deliver an MHC:nominal antigen peptide complex to a cell that does not normally express that MHC molecule or to a cell that does not efficiently process antigen or to a cell that cannot be readily transfected with an MHC gene expression construct. Particular advantages that ensue from this are especially apparent in therapeutic applications directed towards the inhibition of pathogenic antigen-specific T-cells. In this case, the present invention permits the production of highly useful artificial veto cells that function to apoptose or aner- gize antigen-specific T-cells.

Patients suffering from autoimmune or alloimmune disorders are in need of the elimination of pathogenic T- cells. This can be accomplished through the use of an artificial veto cell that targets the culprit antigen- specific T-cells via the specific MHC:nominal antigen pep¬ tide complex on the artificial veto cell which engages the T-cell receptor on the T-cell. Applicant has previously discovered that an antigen-presenting cell can be con¬ verted into an artificial veto cell by expressing the coinhibitor CD8 on the antigen-presenting cell surface. However, in that case, antigen presentation was accom¬ plished by endogenous antigen processing by the artificial veto cell. The present invention provides a more flexible approach in that one is not dependent upon an endogenous antigen processing event, and furthermore, one can control the amount of a uniform population MHC:nominal antigen peptide complex at the surface of the artificial veto cell.

According to one preferred embodiment of the present invention, a costimulator-negative cell from a patient is coated simultaneously with an MHC:GPI :nominal antigen peptide complex and the artificial coinhibitor protein CD8;GPI, both of which are amenable to protein transfer. There are numerous candidate cell types, but one simple cell type to work with in this context is an autologous erythrocyte from a patient. In treating an HLA-A2.1- positive patient with multiple sclerosis, one proceeds according to the following steps. Peripheral blood is obtained from the patient by venipuncture, and erythrocytes are isolated by differential centrifugation. 10 ng of HLA-A2.1:GPI, pre-incubated overnight with a myelin basic protein peptide known to be pathogenic, is added together with 25 ug of CD8:GPI to the erythrocytes. The erythrocytes and recombinant lipid-modified poly¬ peptides are coincubated for two hours at 37C. The coated erythrocytes are then washed three times with suitable buffer and then transfused back into the patient. The simplicity of this procedure stems from the fact that mem¬ brane reincorporation is spontaneous, efficient, predict¬ able, and easily performed. Generally, 10-15 cc of patient red blood cells can be conveniently processed in this way, and the procedure can be repeated as needed.

Example 6 : Treatment of a chronic active hepatitis patient with hepatitis B virus-specific T- cells amplifiedusing HLA-A2.1:GPI:hepatitis B virus peptide-coated dendritic cells It is known that progression to chronic active hepatitis following hepatitis B virus infection is assoc¬ iated with the lack of T-cells reactive with certain well- defined hepatitis B virus-derived nominal antigen pep¬ tides. For patients with HLA-A2.1, the nominal antigen peptides have been well characterized. In examples 1-4 above, the feasibility of performing protein transfer of a functional lipid-modified HLA-A2.1:hepatitis B virus nominal antigen peptide complex was illustrated. In turn, cells coated in this way can be used clinically for immu- notherapeutic purposes. To treat a patient with chronic active hepatitis, peripheral blood is obtained from the patient by venipuncture. Dendritic cells are cultured from the peripheral blood sample using the cytokines GM-CSF and interleukin-4, as described (Romani et al. , 180 J. Exp. Med. 83, 1994) . Once sufficient numbers of dendritic cells have been grown up, the cells are coated with a functional lipid-modified HLA-A2.1:hepatitis B virus nominal antigen peptide complex by simply combining the two, using 10 picomolar of the recombinant protein and coincubating for two hours at 37°C in the absence of serum. Following coating, peripheral blood mononuclear cells are added to the coated dendritic cells, and routine procedures for amplifying T-cells are followed. Generally, once approximately 10⁸-10⁹ T-cells have been accumulated via ex vivo stimulation, the cells are harvested and administered intravenously to the patient. Such T-cell transfer is now done widely and standard procedures can be followed. In essence, the present invention provides a preferable method for carrying out the T-cell amplification step.

Whereas most artificial MHC:GPI chimeric polypeptides can be loaded with nominal antigen peptide at neutral pH, some molecular combinations may require that the loading step be performed at a more alkaline pH, for example, pH 5. In the case of class II MHC, in some instances efficient loading can be obtained at neutral pH (Scheirle et al., 149 J. Immunol. 1994, 1992; Stern et al. , 68 Cell 465, 1992; Sette et al . , 148 J. Immunol . 844, 1992) , whereas in other instances, alkaline pH seems preferable

(Jensen, 174 J. Exp. Med. 1111, 1991; Wettstein et al . , 174 J. EXP. Med. 219, 1991) .

Example 7: A functional artificial GPI-modified costimulator (B7-1:GPI) The following example is provided to show that a B7- 1:GPI molecule can function as a costimulator. Such a molecule can be delivered by protein transfer methods of the claimed invention.

Recombinant GPI-modified murine B7-1 was expressed on the surface of a tumorigenic murine thymoma cell line, EL- 4. This was accomplished using a chimeric coding sequence in which the sequence encoding the extracellular domain of B7-1 was linked in-frame with the GPI-modification signal sequence from the 3'-end of human decay-accelerating factor. C57BL/6 mice (two per group) were injected subcutaneously with either 2 X IO⁶ mB7-l:DAF-S/BSRalphaEN- transfected (3 MB7-DS/EL4 and 4 MB7-DS/EL4) or the identical number of nontransfected (1 EL4 and 2 EL4) EL-4 tumor cells. After fourteen days, splenocytes were recovered from each animal, re-stimulated in vitro for five days with irradiated nontransfected EL4 cells, and then used as effectors in a cytotoxicity assay against ⁵¹Cr-labeled EL-4 target cells at effector-to-target (E:T) ratios as shown. Splenocytes from mice immunized with B7- 1 :GPI-expressing EL-4 transfectants, but not from mice treated with nontransfectants, demonstrated enhanced antigen-specific EL-4-directed cytotoxicity in vitro (Figure 7) . Furthermore, B7-1 :GPI-expressing EL-4 transfectants exhibited loss of tumorigenicity when injected into mice. This study establishes that it is possible to engineer functional B7-1 derivatives with alternative modes of membrane anchorage. B7-1:GPI is a particularly interesting B7-1 derivative since it is amenable to protein transfer.

The B7-1 expression construct was produced as follows. The coding sequence for the extracellular domain of murine B7-1 (Genbank Accession #X60958) , from amino acids 1 to 247, was amplified by polymerase chain reaction from the murine B7-1-containing plasmid pBJ (obtained from L. Lanier) . The sequences of the 5'- and 3' -primers used for this PCR amplification were 5' -AAAAGCTTATGGCTTGCAATT GTGAG-3' and 5 ' -TTTAAGCTTGTGTTCTTGCTATCAGG-3 ' , respectively. These primers added Hindlll restriction endonuclease sites to both ends of the amplified B7-1 cassette. In parallel, 3'-end DAF-S sequence was mobilized from the plasmid pA2 :DAF-S/EE6 via Hindlll plus BamHI digestion, and it was ligated into the corresponding sites of the plasmid vector Bluescript (Stratagene) . The resulting plasmid, pDAF-S/BT, was digested with Hindlll (at the upstream end of the DAF-S sequence) , and the PCR- amplified, Hindlll-digested B7-1 fragment was ligated into this site, generating the plasmid pB7-l:DAF-S/BT. In turn, the chimeric B7-1:DAF-S coding sequence was mobilized using flanking Xhol and Xbal sites and was subcloned into the corresponding sites of the eukaryotic expression vector BSRalphaEN (obtained from D. Lublin) , generating the expression construct mB7-l:DAF- S/BSRalphaEN.

EL-4 cells, obtained from the American Type Culture Collection, were transfected with the mB7-l:DAF- S/BSRalphaEN gene construct by a standard lipofection pro¬ cedure according to the manufacturer's protocol (Bethesda Research Laboratories) . mB7-l epitope expression on the surface of stable EL-4 transfectants was verified by immunofluorescence and flow cytometry. The staining procedure entailed the addition of 0.25ug of CTLA4-Ig to approximately 5 X IO⁶ cells in a volume of lOOul for 1 hour on ice. After washing the cells three times, FITC- conjugated goat anti-human IgG was incubated with the cells for an additional hour. Cells were then washed, fixed and analyzed by flow cytometry.

The costimulator function of B7-1:GPI was assessed through a cellular immunization protocol. According to this protocol, C57BL/6 mice were injected subcutaneously with either mB7-l:DAF-S/BSRalphaEN-transfected or nontransfected EL-4 tumor cells. 2 X IO⁶ tumor cells were injected into each animal. After 14 days, splenocytes were recovered from each animal, washed twice, and plated out in 24-well cell culture plates (10⁷ cells/well) in RPMI 1640 supplemented with 10% fetal bovine serum (BioWittaker) , 1% glutamine, 1% penicillin/streptomycin, and lOum 2-mercaptoethanol . 2 X IO⁵ irradiated (10,000 rads) nontransfected EL-4 cells were added to each well (1:50 E:T ratio) . After five days of co-incubation, cyto- toxicity assays were performed using ⁵¹Cr-labeled EL-4 cells as targets. Treated splenocyte effectors were com¬ bined with the targets at different E:T ratios [1.3 X IO⁴ targets with either IO⁶ (E:T 76:1), 0.5 X IO⁶ (38:1) 0.25 X IO⁶ (19:1) , or .12 X IO⁶ (E:T 9.6:1) effectors] in indi- vidual wells of a 96-well plate and incubated for four hours at 37°C. Supernatants were harvested with Skatron harvest filters and counted in a gamma counter.

In an experiment comparing two mice in each category, there was a significant increase in CTL generation for mice pre-immunized with EL-4 cells bearing GPI-modified B7-1 (Figure 7) . At an effector-to-target cell ratio of 76:1, GPI-modified B7-1-primed animals showed approxi¬ mately 60% specific lysis, whereas controls displayed less than 30% specific lysis. At a lower effector-to-target cell ratio (9.6:1) , the % specific lysis values were 45% and 25%, respectively. These data demonstrate that the B7-1:DAF-S chimeric sequence encodes a functional B7-1 polypeptide derivative.

The costimulator function of B7-1:GPI was further evaluated via an assessment of its effects on EL-4 tumorigenic potential. Two syngeneic C57BL/6 mice were injected subcutaneously with 2 X IO⁶ B7-1:GPI-expressing EL-4 transfectants, and two control animals were injected with 2 X IO⁶ nontransfected EL-4 cells. Whereas large tumors formed in all control animals by seven days post- inoculation, there was no tumor formation in the animals receiving transfected cells. This result is identical to that obtained previously with EL-4 transfectants bearing native B7-1. These results indicate that B7-1:GPI retains the potential of native B7-1 to confer loss of tumorigeni- city to tumor cells. Taken together, these data indicate that membrane anchorage via its native transmembrane hydrophobic peptide anchor is not a prerequisite for B7-l's costimulator function. This flexibility in the choice of membrane anchor opens up unique protein engineering opportunities. B7-1:GPI is of particular interest in this regard since it can be immunoaffinity purified and then reincorporated back into cell membranes.

Other embodiments are within the following claims.

SEQUENCE LISTING

(1) GENERAL INFORMATION:

(i) APPLICANT: Mark L. Tykocinski

(ii) TITLE OF INVENTION: METHODS FOR ENGINEERING ANTIGEN-PRESENTING CELLS

(iii) NUMBER OF SEQUENCES:

(iv) CORRESPONDENCE ADDRESS

(A) ADDRESSEE Lyon & Lyon

(B) STREET: 633 West Fifth Street

Suite 4700

(C) CITY: Los Angeles

(D) STATE: California

(E) COUNTRY: U.S.A.

(F) ZIP: 90071-2066

(v) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: 3.5" Diskette, 1.44 Mb storage

(B) COMPUTER: IBM Compatible

(C) OPERATING SYSTEM: IBM P.C. DOS 5.0

(D) SOFTWARE: Word Perfect 5.1

(vi) CURRENT APPLICATION DATA:

(A) APPLICATION NUMBER: 08/324,125

(B) FILING DATE: October 14, 1994

(C) CLASSIFICATION: Not Yet Classified

(vii) PRIOR APPLICATION DATA:

Prior applications total, including application described below:

(A) APPLICATION NUMBER: 08/260,547

(B) FILING DATE: June 16, 1994 (viii) ATTORNEY/AGENT INFORMATION:

(A) NAME: Warburg, Richard J.

(B) REGISTRATION NUMBER: 32,327

(C) REFERENCE/DOCKET NUMBER: 209/244

(ix) TELECOMMUNICATION INFORMATION:

(A) TELEPHONE: (213) 489-1600

(B) TELEFAX: (213) 955-0440

(C) TELEX: 67-3510

(2) INFORMATION FOR SEQ ID NO: 1:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 26 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) SEQUENCE DESCRIPTION: SEQ ID NO: 1: AAAAGCTTAT GGCTTGCAAT TGTGAG 26

(2) INFORMATION FOR SEQ ID NO: 2:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 26 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) SEQUENCE DESCRIPTION: SEQ ID NO: 2: TTTAAGCTTG TGTTCTTGCT ATCAGG 26

Claims

Claims

1. A method for producing a cell having a defined MHC:nominal antigen peptide on its membrane comprising the step of: a) contacting the external portion of said cell with an externally applied nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind said nominal antigen peptide.

2. A method for producing a cell having a defined MHC:nominal antigen peptide on its membrane comprising the steps of: a) first contacting the external portion of said cell with an externally applied artificial lipid- modified MHC polypeptide able to bind a nominal antigen peptide; and b) second contacting said cell with said nominal antigen peptide so that said artificial lipid- modified MHC polypeptide binds said peptide.

3. A method for producing a cell having a defined MHC:nominal antigen peptide on its membrane comprising the steps of: a) first contacting an nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind a nominal antigen peptide; and b) second contacting the external portion of said cell with externally applied nominal antigen peptide bound to said artificial lipid-modified MHC polypeptide.

4. A method for producing a biological membrane having a defined MHC:nominal antigen peptide on its surface comprising the step of: a) contacting said membrane with a nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind said nominal antigen peptide.

5. A method for producing a biological membrane having a defined MHC:nominal antigen peptide on its surface comprising the steps of : a) first contacting said membrane with an artificial lipid-modified MHC polypeptide able to bind a nominal antigen peptide; and b) second contacting said membrane with said nominal antigen peptide so that said artificial lipid- modified MHC polypeptide binds said peptide.

6. A method for producing a biological membrane having a defined MHC:nominal antigen peptide on its surface comprising the steps of: a) first contacting a nominal antigen peptide and an artificial lipid-modified MHC polypeptide able to bind a nominal antigen peptide; and b) second contacting said membrane with said nominal antigen peptide bound to said artificial lipid- modified MHC polypeptide.

7. The method of claim 1, 2, 3, 4, 5, or 6 wherein said MHC polypeptide comprises a class I MHC polypeptide sequence.

8. The method of claim 1, 2, 3, 4, 5, or 6 wherein said MHC polypeptide comprises a class II MHC polypeptide sequence.

9. The method of claim 1, 2, 3, 4, 5, or 6 wherein said lipid-modified MHC polypeptide comprises a glycosyl- phosphatidylinositol- odified MHC polypeptide.

10. The method of claim 9 wherein said GPI-modified MHC polypeptide is produced by the steps of: a) gene transfer of a chimeric gene expression construct comprising a GPI modification signal sequence into a host cell;and b) isolation of said GPI-modified MHC polypeptide from said host cell.

11. A method for activating an antigen-specific T- cell comprising the step of contacting a cell having a defined MHC:nominal antigen peptide complex on its membrane with a T-cell population comprising a T-cell with specificity for said MHC:nominal antigen peptide complex.

12. A method for inhibiting an antigen-specific T- cell comprising the step of contacting cell having a defined MHC:nominal antigen peptide complex on its membrane with a T-cell population comprising a T-cell with specificity for said MHC:nominal antigen peptide complex.

13. The method of claim 11 or 12 wherein said contacting step is performed ex vivo.

14. The method of claim 11 or 12 wherein said contacting step is performed in vivo.

15. The method of claim 11 or 12 wherein said MHC polypeptide is syngeneic with said antigen-specific T- cell.

16. The method of claim 11 or 12 wherein said MHC polypeptide is allogeneic with said antigen-specific T- cell.

17. The method of claim 11 wherein said modified cell is a dendritic cell.

18. The method of claim 11 wherein said modified cell is an activated B-cell.

19. The method of claim 12 wherein said modified cell is a cell lacking a costimulator.

20. The method of claim 11 wherein said modified cell is presenting a tumor antigen.

21. The method of claim 12 wherein said modified cell is expressing a coinhibitor.

22. The method of claim 1, 2, 3, 4, 5, or 6 wherein said lipid-modified MHC polypeptide comprises a polyhistidine tag.

23. A glycosyl-phosphatidylinositol-modified MHC polypeptide which is separate from a cell membrane.

24. A cell having a membrane exogenously coated with a lipid-modified MHC polypeptide.

25. A modified antigen-presenting cell comprising a cell with an artificial costimulator polypeptide exogenously reincorporated into the surface of said cell.

26. The modified antigen-presenting cell of claim 25 comprising an immunogenic tumor cell.

27. The modified antigen-presenting cell of claim 25 wherein said artificial costimulator polypeptide comprises a glycosyl-phosphatidylinositol moiety.

28. The modified antigen-presenting cell of claim 25 wherein said artificial costimulator polypeptide comprises a streptavidin moiety.

29. The modified antigen-presenting cell of claim 25 wherein said artificial costimulator polypeptide comprises a molecule with affinity for a molecule on the surface of said cell.

30. The modified antigen-presenting cell of claim 25 wherein said cell is pre-coated with a molecule with affinity for an artificial costimulator polypeptide.

31. The modified antigen-presenting cell of claim 25 wherein the costimulator portion of said artificial costimulator polypeptide is selected from the group consisting of B7-1, B7-2, B7-3, intracellular adhesion molecule-1, intracellular adhesion molecule-2, intracellular adhesion molecule-3, LFA-3, VCAM-1, and fibronectin.

32. A method for producing a modified antigen presenting cell having a costimulator on its membrane comprising the step of: a) contacting the external portion of said cell with an externally applied artificial costimulator polypeptide.

33. The method of claim 32 wherein said costimulator polypeptide comprises an artificial lipid-modified costimulator polypeptide.

34. The method of claim 33 wherein said artificial lipid-modified costimulator polypeptide comprises a glycosyl-phosphatidylinositol moiety.

35. The method of claim 32 wherein said costimulator polypeptide is selected from the group consisting of B7-1, B7-2, B7-3, intracellular adhesion molecule-1, intracellular adhesion molecule-2, intracellular adhesion molecule-3, LFA-3, VCAM-1, and fibronectin.

36. A method for enhancing a cell's immunogenicity comprising contacting said cell with a membrane-binding artificial costimulator polypeptide.

37. A method for inducing anti-tumor immunity against an autologous tumor cell in a patient comprising the step of: a) increasing the amount of a costimulator molecule on the surface of said tumor cell by protein transfer.

38. A method for inducing anti-tumor immunity against an autologous tumor cell in a patient comprising the step of: a) administering to said patient an autologous tumor cell with externally reincorporated costimulator polypeptide on its membrane.

39. A method for inducing anti-tumor immunity against an autologous tumor cell in a patient comprising the step of: a) administering to said patient an heterologous • tumor cell with externally reincorporated costimulator polypeptide on its membrane.

40. A method for inducing anti-tumor immunity against an autologous tumor cell in a patient comprising the step of: a) administering to said patient a tumor cell membrane extract with externally reincorporated costimulator polypeptide.

41. A method for identifying a molecule that enhances the immunogenicity of an antigen presenting cell, comprising the step of: a) coating said cell with a membrane-binding polypeptide derivative of said molecule, and b) determining the immunogenicity of the resulting cell compared to an uncoated cell.

42. The method of claim 41 wherein said antigen- presenting cell is a tumor cell.

43. The method of claim 42 wherein determining the immunogenicity comprises determining the capacity of said coated tumor cell to induce an anti-tumor immune response.