HK1218144B - Methods for high throughput receptor:ligand identification - Google Patents

Description

High-throughput receptor:ligand identification methods

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of U.S. Provisional Application No. 61/735,791, filed December 11, 2012, and U.S. Provisional Application No. 61/833,588, filed June 11, 2013, the contents of each of which are hereby incorporated by reference.

Government support statement

This invention was made with government support under Grant Nos. 3U54GM094662-02, 5U01GM094665-02, and AI057158 awarded by the National Institutes of Health. The government has certain rights in this invention.

Background of the Invention

Throughout this application, various publications are referenced within parentheses. Full citations for these references can be found at the end of the specification. The disclosures of these publications and all patents, patent application publications, and books mentioned herein are hereby incorporated by reference in their entirety into this application to more fully describe the field to which this invention pertains.

Cell surface receptors and adhesion molecules are the defenders of cellular function, and are important for the development, morphogenesis and environmental processes of normal physiology and pathology. These molecules are major therapeutic targets. The high-resolution structural characterization of these complexes is limited to the chemical and physical determinants of the basis of receptor: ligand specificity, affinity, oligomeric state, potency and the overall structural characteristics that are important for integrating these interactions and their related signal transduction pathways into the overall cell physiology. All of these features are crucial for understanding the basic mechanisms that drive complex cellular processes and provide rare opportunities for therapeutic intervention. Unfortunately, the systematic structural characterization of these key complexes (i.e., the structural genomics of the secretome) is currently an unrealistic goal because many receptor: ligand pairs, if not most, are still unclear and therefore cannot be structurally characterized.

The present invention addresses this need by providing a technology for efficiently and systematically identifying all components of receptor:ligand interactions relevant to human physiology, disease, and pharmaceuticals.

SUMMARY OF THE INVENTION

A cell microarray is provided, comprising:

(i) a first plurality of cells transformed to express a first predetermined heterologous secreted protein, a heterologous membrane protein, or a heterologous cell surface protein and a first fluorescent protein and (ii) at least a second plurality of cells transformed to express a second predetermined heterologous secreted protein, or a second heterologous protein and a second fluorescent protein,

wherein the first and second pluralities of cells are attached to a solid surface of the microarray, and wherein the first and second pluralities of cells are in spatially distinct locations on the solid surface.

A cell microarray is provided, comprising:

(i) a first plurality of cells transformed to express (a) a first predetermined heterologous protein and (b) a first fluorescent protein and (ii) at least a second plurality of cells transformed to express (a) a second predetermined heterologous protein and (b) a second fluorescent protein,

wherein the first and second pluralities of cells are attached to a solid surface of the microarray, and wherein the first and second pluralities of cells are in spatially distinct locations on the solid surface.

A method for preparing a cell microarray as described herein is provided, comprising: immobilizing a first plurality of expression constructs encoding a first heterologous protein and a first fluorescent protein on a solid surface of a microarray at locations spatially distinct from the immobilized first plurality of expression constructs, immobilizing at least a second plurality of expression constructs encoding a second heterologous protein and a second fluorescent protein on the solid surface of the microarray, and contacting the expression constructs with a plurality of cells under conditions including the presence of a transfection agent so as to allow the cells to attach to the solid surface and transfection of at least a portion of the cells occurring at each spatially distinct location with the corresponding expression construct.

Also provided is a method for determining whether a candidate protein or peptide binds to a second protein or peptide, the method comprising expressing the second protein as a heterologous protein on a cell microarray as described herein, and contacting the cell microarray with the candidate protein or peptide, wherein the candidate protein or peptide has a third fluorescent protein or peptide immobilized thereon, washing the cell microarray in contact with the candidate protein or peptide to remove unbound candidate protein or peptide, and determining whether any candidate protein or peptide is bound to the cell microarray after washing, wherein the presence of the candidate protein or peptide bound to the cell microarray at a first spatial location corresponding to cells transformed with the first heterologous protein after washing indicates that the candidate protein or peptide is bound to the first heterologous protein, and wherein the absence of the candidate protein or peptide bound to the cell microarray at the first spatial location after washing indicates that the candidate protein or peptide is not bound to the heterologous protein.

Also provided is a system comprising (i) a microarray solid surface and a suspension-adapted cell line transformed to express a candidate ligand protein or peptide and a first C-terminal cytoplasmically expressed fluorescent protein on its cell surface, and (ii) at least a) a second plurality of cells transformed to express a predetermined heterologous protein and a second fluorescent protein on its cell surface, or b) a plurality of microbeads to which the heterologous protein has been immobilized and to which the second fluorescent protein has been immobilized, wherein a) or b) is immobilized to the microarray solid surface. Also provided is a system as above, mutatis mutandis, wherein the candidate ligand protein or peptide is expressed on the second plurality of transformed cells or the plurality of microbeads, and the heterologous protein is expressed on the cell surface of the transformed suspension-adapted cell line.

A method for determining whether a candidate ligand protein or peptide binds to a second protein or peptide, the method comprising expressing the candidate ligand protein or peptide and a first fluorescent protein in a plurality of cells of the suspension-adapted cell line of the current system, contacting the plurality of cells with a) a second plurality of cells transformed to express the heterologous protein and the second fluorescent protein, or b) a plurality of microbeads having the heterologous protein and the second fluorescent protein immobilized on their surfaces, washing to remove unbound candidate ligand protein or peptide, and identifying cells showing colocalization of the first and second fluorescent proteins by FACS analysis, wherein cells showing colocalization of the first and second fluorescent proteins at spatially different locations indicate that the first protein or peptide binds to the heterologous protein corresponding to the spatially different locations.

A system is provided that includes a first plurality of suspension-adapted cell line cells transformed with a vector so as to express a first heterologous protein on their cell surface and a first cytoplasmically expressed fluorescent protein, wherein the vector comprises a single predetermined sequence of 15-35 nucleotides of the first heterologous protein, the single sequence being capable of being primed by one or more universal primers; and a second plurality of suspension-adapted cell line cells transformed with a second vector so as to express a second heterologous protein on their cell surface and a first cytoplasmically expressed fluorescent protein, wherein the second vector comprises a different single predetermined sequence of 15-35 nucleotides of the second heterologous protein; and (i) one or more additional pluralities of suspension-adapted cell line cells transformed with a vector so as to express a candidate ligand protein or peptide on their cell surface and a second fluorescent protein, wherein the second suspension-adapted cell line comprises a stably expressed peptide cell surface epitope, or (ii) a plurality of magnetic microbeads to which the candidate ligand protein or peptide has been immobilized on its surface and to which the second fluorescent protein has been immobilized.

Also provided is a method for determining whether a candidate ligand protein or peptide binds to a second predetermined protein, the method comprising expressing the second predetermined protein as a heterologous protein of the current system and contacting the candidate ligand protein or peptide with (i) one or more additional plurality of suspension-adapted cell line cells transformed so as to express the candidate ligand protein or peptide on their cell surfaces and express a second fluorescent protein, the second suspension-adapted cell line comprising a stably expressed peptide cell surface epitope, or with (ii) a plurality of magnetic microbeads to which the candidate ligand protein or peptide has been immobilized and to which the second fluorescent protein has been immobilized;

separating any one of the first plurality of suspended adaptive cell line cells that is bound to one or more of the second plurality of cells or to the plurality of magnetic microbeads by magnetic attraction;

obtaining DNA from such isolated cell-cell or cell-bead conjugates and amplifying a unique sequence, if present in the DNA, using said universal primers;

sequencing copies of the unique sequence to confirm their presence;

The single sequence thus identified is compared to a database associating the single predetermined sequence of 15-35 nucleotides with specific heterologous proteins or peptides, and any heterologous proteins or peptides thus associated are identified as binding, thereby identifying the specific heterologous protein or peptide as binding to the candidate protein or peptide.

A system comprising (i) a first plurality of suspension-adapted cell line cells, wherein the plurality of cells are transformed with a vector so as to (a) express a heterologous protein on their cell surface and (b) express a first cytoplasmically expressed fluorescent protein, and wherein the vector comprises a single predetermined sequence of 15-35 nucleotides of the expressed heterologous protein, such that the first plurality of suspension-adapted cell line cells express at least two different types of the first heterologous protein; and (ii) a second plurality of suspension-adapted cell line cells, wherein the cells are transformed with a second vector so as to express a second heterologous protein on their cell surface and a second cytoplasmically expressed fluorescent protein, wherein the second plurality of suspension-adapted cells express a single type of the second heterologous protein. In one embodiment, any single cell in the first plurality of cells expresses only one heterologous protein on its cell surface. In one embodiment, no different type of the first heterologous protein of the first plurality of cells has the same sequence as the second heterologous protein of the second plurality of cells.

Also provided is a method for determining whether a candidate ligand protein or peptide binds to a second protein or peptide, the method comprising expressing the candidate ligand protein or peptide as a first heterologous protein in a first plurality of cells in a system as described herein and expressing the second protein or peptide as a second heterologous protein in a system as described herein under conditions permissive for binding of the first heterologous protein to the second heterologous protein, and optionally, washing to remove any unbound first heterologous protein, then recovering cells in which the first and second heterologous proteins co-localize, obtaining nucleic acid from the recovered cells and sequencing the nucleic acid to identify a single sequence of 15-35 nucleotides contained therein to identify the candidate ligand protein or peptide that has bound the second protein or peptide corresponding to the single 15-35 nucleotides.

Also provided is a method for determining the effect of a predetermined amino acid residue of a first protein on the binding of the first protein to a second protein, the method comprising expressing a protein mutated by one or more point mutations relative to the first protein as a plurality of different types of heterologous proteins in a plurality of cells of a first suspension-adapted cell line of a system as described herein, contacting the plurality of proteins with a second protein in the form of a second heterologous protein in a second plurality of cells transformed to express the second protein and a second fluorescent protein in the system as described herein, recovering cells that exhibit colocalization of the first and second fluorescent proteins, obtaining nucleic acid from the recovered cells and sequencing the nucleic acid to identify a unique sequence of 15-35 nucleotides contained therein to identify the first protein that has bound to a second protein or peptide, and comparing the level of the protein that has bound to the second protein or peptide to a predetermined reference level.

Wherein the level of the protein to which the second protein or peptide has been bound exceeds a predetermined reference level, it indicates that the one or more residues mutated in the protein enhance the binding of the first protein to the second protein, and where the level of the protein to which the second protein has been bound is below a predetermined reference level, it indicates that the one or more residues mutated in the protein inhibit the binding of the first protein to the second protein.

Also provided is a system comprising (i) a first plurality of suspension adaptive cell line cells transformed with a vector so as to express a first heterologous candidate ligand protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein, and a second plurality of suspension adaptive cell line cells transformed with a second vector so as to express a second heterologous candidate ligand protein or peptide on their cell surface and a second cytoplasmically expressed fluorescent protein, and (ii) a plurality of magnetic microbeads to which target proteins, peptides or antibodies have been immobilized.

Also provided is a method for determining whether one or more of two candidate ligand proteins or peptides binds to a target protein, peptide, or antibody, the method comprising expressing a first candidate ligand protein or peptide as a first heterologous protein in a first plurality of cells in the current system and expressing a second candidate ligand protein or peptide as a second heterologous protein in the current system under conditions that allow binding of the first heterologous protein and the second heterologous protein to the target protein, peptide, or antibody, and recovering any microbeads complexed with first fluorescent protein-expressing cells and/or complexed with second fluorescent protein-expressing cells, and identifying the candidate ligand proteins in the complexes, wherein recovering microbeads in complex with the first fluorescent protein-expressing cells indicates that the first candidate ligand protein or peptide binds to the target protein or peptide, and wherein recovering microbeads in complex with the second fluorescent protein-expressing cells indicates that the second candidate ligand protein or peptide binds to the target protein or peptide, and wherein not recovering microbeads in complex with the first fluorescent protein-expressing cells or the second fluorescent protein-expressing cells indicates that the first candidate ligand protein does not bind to the target protein or peptide, and the second candidate ligand protein does not bind to the target protein or peptide, respectively.

Also provided is a system comprising (i) a first plurality of suspension adaptive cell line cells transformed with a vector so as to express a first heterologous target protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein and one or more second plurality of suspension adaptive cell line cells transformed with a second vector so as to express a second heterologous candidate ligand protein or peptide on their cell surface and a second cytoplasmically expressed fluorescent protein, and (ii) a plurality of magnetic microbeads to the surface of which antibodies against the candidate ligand protein or peptide or against the target protein or peptide have been immobilized. Also provided is a method for determining whether a candidate ligand protein or peptide binds to a target protein or peptide, the method comprising expressing the candidate ligand protein or peptide as a second heterologous protein of the second plurality of cells in the current system and expressing the target protein or peptide as a first heterologous protein in the current system under conditions permitting binding of the candidate ligand protein or peptide to the target protein or peptide, and recovering any microbeads complexed with a first fluorescent protein-expressing cell and a second fluorescent protein-expressing cell, wherein recovering microbeads attached to a complex of the first fluorescent protein-expressing cell and the second fluorescent protein-expressing cell indicates that the candidate ligand protein or peptide binds to the target protein or peptide, and wherein not recovering microbeads attached to a complex of the first fluorescent protein-expressing cell and the second fluorescent protein-expressing cell indicates that the candidate ligand protein does not bind to the target protein or peptide.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1. Crystallographic view of the immune synapse formed between a T cell and an antigen-presenting cell. Composite model of the TCR:MHC and co-stimulatory receptor:ligand complexes within the central region of the immune synapse. The TCR:MHC (PDB code 1G6R), PD-1:PD-L1 (3BIK), PD-1:PD-L2 (3BP5), and CTLA-4:B7-1 (1I8L) complexes are representative of existing crystal structures; the model of the monovalent CD28:B7-1 complex is based on the CD28 monomer (1YJD) and CTLA-4:B7-1 structures. The approximate size (i.e., length) of the complex is shown, as well as the number of residues connecting the structured Ig domain to the membrane. The distance of approximately 140 Å between the plasma membranes in the immune synapse is also indicated. The cytoplasmic signaling and scaffolding proteins, which are positioned and organized by interactions between the receptor:ligand extracellular domains, are schematically shown (i.e., geometric symbols). This figure highlights the importance of receptor:ligand structure in defining fundamental characteristics (e.g., oligomeric state, valence, ligand specificity) and overall organizational principles that are key determinants of function for cell surface receptors and ligands [45].

Figure 2: Receptor:ligand structure studies. Left: Structure of the mouse PD-1:PD-L2 complex. Right: Locations of point mutations in the PD-1 receptor that result in increased affinity and altered specificity.

Figure 3: Cell microarray platform. Left: Schematic diagram of cell microarray generation. Right: For illustration, a GFP expression construct (i.e., plasmid) was "pinned" onto a glass surface to create an expression array. HEK293 cells were seeded with the spotted cDNA and subsequently transfected to generate a live cell array. This 2000-spot grid was constructed using a custom microarray printer. Each spot is a cluster of 50-80 cells.

Figures 4A-4C: Detection of T cell ligand and receptor binding using a cell microarray platform. (A) High-density cell microarray containing alternating rows of HEK cells expressing cytoplasmic GFP or plasma membrane-embedded PD-L1-GFP (schematically shown). (B) The array was treated with purified PD-1 IgG premixed with Alexa 594 secondary antibody (RED) and imaged on an Axon 4000B microarray scanner. Cells presenting only the PD-L1 extracellular domain showed staining; cells expressing cytoplasmic GFP did not show staining and served as a negative control. (C) Merged images showing colocalization of labeled PD-1-Ig fusion protein and PD-L1 GFP, but not cytoplasmic GFP constructs. These results demonstrate the specificity of the microarray. The difference in GFP fluorescence intensity between PD-L1 and the control (GFP alone) is shown by the expression method.

Figure 5: Three cell microarrays showing specific binding between different members of the Ig superfamily. Slides were spotted with alternating rows of plasmid DNA encoding GFP fusion constructs of PD-1, B7-1, and CD200R, or GFP alone. Three spotted slides were placed in a 10 cm square culture dish, treated with transfection agent, and then seeded with HEK cells. Three days after transfection, the slides were washed and subsequently treated with Ig-fusions of PD-L2 (slide 1), CTLA-4 (slide 2), or CD200 (slide 3); the Ig-fusions were pre-incubated with Cy7 secondary antibody for detection. All spotted rows were successfully transfected and GFP-positive (data not shown). The images show only the fluorescent signal from the Cy7 channel. For each array, significant binding of the Ig-fusion was detected only for those rows where its cognate receptor or ligand was present. For example: PD-L2, the ligand of the receptor PD-1, only binds in those rows where PD-1 GFP was spotted. Individual slides were pseudocolored for clarity and overlays reveal specific binding patterns.

Figure 6: Generation of an Ig superfamily mammalian expression library. Schematic (top) shows a ligation-independent vector specifically engineered for rapid and efficient cloning of type I secretory proteins. 283 members of the Ig superfamily have been cloned and tested for expression by transient transfection of HEK293 cells. Using fluorescence microscopy, 240 (approximately 85% successful) clones expressed above background and demonstrated correct membrane localization. GFP fluorescence images are shown for a panel of 24 representatives of the 240 expression library members. Preliminary validation of each expression construct highlighted proteins that required additional "rescue work" and were critical for a fully characterized, robust platform.

Figure 7: Co-transfection in a cell microarray format. Left) HEK293 cells transfected with cytoplasmic GFP. Center) HEK293 cells transfected with cytoplasmic mCherry. Right) HEK293 cells co-transfected with cytoplasmic GFP and mCherry. Co-transfection with green and red fluorophores produces yellow fluorescence. This provides proof of principle for expressing multiple polypeptides, as required to achieve functional multicomponent receptors in a cell microarray format.

Figures 8A-8C: Strategies for receptor:ligand discovery using high-affinity microbeads. A) Suspension-adapted HEK293 cells were transiently transfected with selected cell surface markers (e.g., PD-L1; red cells) linked to cytoplasmic mCHERRY. B) 50 nm protein A-coated GFP-labeled microbeads were modified with PD-1Ig or B7-1Ig-fusion proteins (green beads). C) Incubation of red cells and green microbeads produces red:green conjugates. D) Flow cytometry allows detection and quantification of receptor:ligand interactions. Along the Y-axis are unbound red transfected cells; along the X-axis are unbound green microbeads; along the diagonal are red:green cell:microbead conjugates reporting receptor:ligand interactions.

Figure 9: Microbead-based demonstration of PD-L1:PD-1 and PD-L1:B7-1 interactions. Protein A microbeads were saturated with IgG premixed with FITC-IgG and Ig-fusion (i.e., PD-1 or B7-1) at a 1:4 ratio that was precipitated and then resuspended in PBS. The conjugated microbeads were incubated with HEK cells expressing either mCherry alone or PD-L1mCherry fusions, and the extent of binding was determined by flow cytometry. These data clearly demonstrate that both PD-1-conjugated microbeads and B7-1-conjugated microbeads bind to HEK cells expressing only PD-L1 and support the strategy of increasing the affinity of the ligand to increase the dynamic range of potential receptor-ligand interactions that can be measured experimentally.

Figure 10: Detection of specific engineered cell-cell interactions by flow cytometry. HEK 293 cells were transfected with GFP, mCherry, PD-L1mCherry, or PD-1GFP. The cells were resuspended in ice-cold DMEM containing 2% BSA and the cell lines were mixed together in a 1:1 stoichiometric ratio. Individual populations and mixed binary pairs were incubated at 4°C for 2 hours. A significant (approximately 60-fold) increase in the number of red and green fluorescent events (i.e., conjugates) was only observed in the presence of ligand and receptor.

Figure 11: Detection of CD200:CD200-receptor interactions by flow cytometry. HEK293 cells were transfected with GFP, mCherry, CD200mCherry, or CD200-receptor GFP. Cells were processed and analyzed as in Figure 10. A significant increase in the number of red:green fluorescence events (i.e., conjugates) was only observed when cells expressing the ligand and receptor alone were present.

Figure 12: Identification of highly selective functional PD-L1 mutants using microbeads and cell-cell FACS assays. A) A panel of >100 PD-L1mCherry mutant constructs were transiently transfected into HEK293 cells. These lines were then stimulated with microbeads pre-saturated with PD-1 Ig-fusions (microbead data in orange) or, in a separate set of experiments, with cells pre-saturated with transiently expressing PD-1 GFP (cell-cell data in blue). In each experiment, the percent binding determined by FACS analysis was normalized to wild-type binding. The data show a direct comparison of the average binding observed using microbeads or cell-cell methods. For clarity, only a subset of the data is shown. A similar set of experiments was also performed using B7-1 (another known ligand for PD-L1) to stimulate PD-L1 mutants (data not shown). B) Screening of PD-L1 mutants yielded several mutants that showed impaired binding to PD-1, B7-1, or both. To further validate these results, seven PD-L1 mutants were expressed as Ig-fusions. Purified PD-L1 mutants were pre-incubated with Alexa 594 secondary antibodies (red) for detection and added to HEK293 cells expressing PD-1 GFP or B7-1 GFP (green). FACS data show GFP fluorescence (X-axis) versus Alexa 594 fluorescence (Y-axis). Although the data appear to correlate more closely with the results obtained using the cell-cell binding method, the extent of binding observed using purified proteins reflects the extent of binding observed in both screening methods.

Figures 13A-13E: Strategy for parallel identification of interactions between cell surface proteins. A) A barcoded library is generated by LIC cloning into a barcode library vector (BC-L vector, which generates a GFP reporter gene and a unique core barcode of 28 nucleotides flanked by universal T7 forward and reverse primer sites). The libraries are pooled and transfected together into suspension-adapted HEK293 cells. B) In separate reactions, a single query receptor is transfected into HEK293 cells that are also mCherry-tagged and present the FLAG epitope on the cell surface to allow magnetic capture/isolation. C) The pooled library is mixed with query-expressing cells (1:1) to allow productive interactions between cognate receptors. D) Positive interactions (i.e., conjugates) are recovered from the expression pool by magnetic capture ("sorting") of the query receptor cells. E) Plasmid DNA is extracted and the barcodes are PCR amplified using universal primers. F) The sequence of each PCR product is obtained using next-generation massively parallel sequencing. A unique barcode signature uniquely identifies the ligand complexed with the query receptor. Notably, this approach can be multiplexed by performing the above steps for a large number of query proteins, individually capturing the resulting conjugates in a multi-well plate format (e.g., 24-well plates), and adding a "good specificity" identifier (a unique 8-nucleotide barcode) to the primers in step E. These amplicons are then pooled into a single next-generation sequencing run and deconvoluted after sequencing to reduce the cost per reaction.

Figures 14A-14B: Specific cell-cell binding events can be captured and isolated using magnetic microbeads. A) HEK293 cells transiently expressing a PD-L1 mCherry fusion were mixed with cells transiently expressing either GFP or PD-1 GFP fusions at a 1:1 ratio (4 × 10 ⁶ cells total). Magnetic protein A-coated microbeads (Miltenyi) modified with PD-L1 antibodies were added and a sample (total amount); this represents the composition of the starting mixture). The mixture was applied to a magnetic cell separation column, and bound cells were washed and eluted. The "total" and "bound" samples were analyzed by FACS to determine the percentage of GFP-positive cells. Of the PD-L1-expressing cells captured by the magnetic beads, significantly more cells expressing the PD-1 GFP fusion were isolated than cells expressing GFP alone, demonstrating clear and specific enrichment of the cognate receptor:ligand conjugate. B) To demonstrate the ability to isolate "rare" events, as would be encountered in a high-throughput screen, 10 ⁶ PD-L1 mCherry cells were mixed with 0.1 × 10 ⁶ cells expressing either PD-1 GFP-fusion or GFP and 5 × 10 ⁶ untransfected cells (to simulate a pseudo-pool). In this experiment, GFP-positive cells comprised approximately 1.5% of the total cells. Two hours later, anti-PD-L1-coated microbeads were added, a "total" sample was taken, and the bound fraction was isolated as described in A. The data show a significant enrichment of PD-1 GFP-fusion-expressing cells over magnetically captured PD-L1-expressing cells.

Figure 15: Simulated signal-to-noise ratio: Enrichment of "rare" receptors from the collection by cell-cell binding coupled with FACS sorting. A) For the mock expression library, 10 ⁷ HEK293 cells transiently expressing GFP were mixed with 0.02×10 ⁶ cells expressing PD-1 GFP fusion (0.2% GFP-positive cells). The "library" was then stimulated with 10 ⁶ HEK293 cells transiently expressing mCherry (negative control) or PD-L1 mCherry-fusion. The data shows flow cytometric analysis of a total of 3×10 ⁶ events. Gates were set based on 10,000 event reads of GFP "library", mCherry, or PD-L1 mCherry-fusion alone. B) Schematic showing the location of primers used to verify PD-1 GFP-fusion enrichment. C) For PD-L1 mCherry stimulation (right panel in A), positive binding events (Q2) were sorted and 10,000 events were collected. For comparison, 10,000 cells were also collected from the cell mixture before sorting (pre-sort sample). PD-1 GFP-fusion PCR control is in the left lane. Comparison of pre- and post-sort PCR products confirmed that the sorted positive binding events were enriched for PD-1 GFP.

Figure 16: Multiplexing of cell-cell FACS assays: A secretory protein mammalian expression library is generated with each expression construct carrying a unique barcode (a variable sequence approximately 20 base pairs downstream of the 3' end on the expression plasmid). These constructs are expressed with a GFP fluorescent protein tag. For example, a non-barcoded "stimulator" secretory protein expression construct is also generated with an mCherry fluorescent protein tag. The library constructs and the "stimulator" construct are expressed separately in mammalian cells. The "library" and "stimulator" cell populations are mixed together. Cell-cell interactions are then sorted from non-interacting cells. Sorting can be performed by fluorescence (i.e., sorting all double-positive events, mCherry+GFP), or magnetically using magnetic beads to capture all "stimulator" cells along with any bound library partners. Cell samples are taken before and after sorting, lysed, and the supernatant used for PCR amplification of the expression plasmid. The pooled PCR sample is sent for deep sequencing to provide a quantitative assessment of how many copies of a specific barcode are present. In our experiments, enrichment of a specific barcode over background would indicate a specific protein-protein interaction. We can then use the specific barcode sequence to identify which protein came from the library.

Figure 17: Cell-cell "rare event" assays: Multiplexing allows for the extraction of specific binding interactions from libraries where the target protein will be relatively rare compared to the entire library. For example, if one has a library of 100 genes and they are individually expressed in mammalian cells and those 100 expression populations are mixed together in equal amounts, any one gene expressed will represent 1/100 of the total library.

This experiment demonstrates that even when the target gene (in this case, PD-1) is present in 1/100 of the total GFP-positive cell population, it can still be enriched for the PD-1/PD-L1 interaction using the techniques described herein. As an initial test of the deep sequencing approach, a panel of previously characterized PD-L1 mutants was utilized. The idea was to use the mutant sequence as a "barcode" because each PD-L1 mutant sequence is inherently different. Several of these mutants showed reduced binding to PD-1, and it was possible to identify those same mutants using a multiplexed deep sequencing approach.

Figure 18: PD-L1 mutants: Test case - The sequence shown is for a portion of the mouse PD-L1 gene. The highlighted amino acids indicate the location of unique point mutations. Green represents residues that showed no effect on PD-1 binding when mutated. Red residues represent residues that showed a strong loss of PD-1 binding when mutated, and yellow residues show a more moderate loss of PD-1 binding. The DNA sequences highlighted in blue show the locations of the primers used to PCR amplify the pooled PD-L1 library before and after sorting. FACS plots show a negative control mock sort and an experimental PD-1-stimulated sort using GFP to stimulate the PD-L1 mutant library. For the experiment, 20,000 cells were collected before sorting (pre-sort) and from sorted population Q2 (post-sort) and PCR amplified using the primers highlighted above. The agarose gel image in the lower right shows the PCR products obtained from samples before and after sorting. These samples were then sent for deep sequencing analysis.

Figure 19: Deep sequencing results were analyzed to determine the number of times each unique PD-L1 single sequence was identified (total number of occurrences). The enrichment ratio between the samples before and after sorting was then calculated when it occurred. For example, if 10 wild-type PD-L1 sequences were counted in the pre-sort sample and 100 were counted in the post-sort sample, a 10-fold enrichment of wild-type PD-L1 was observed (100 divided by 10). If a specific PD-L1 mutant does not bind to PD-1, such as D122A, we can still count 10 D122A sequences in the pre-sort sample, but only 5 sequences in the post-sort sample. This results in an enrichment ratio of 0.5 (5 divided by 10). In order to compare the data with the data obtained using the traditional FACS binding method (light blue column), all data were normalized to wild-type PD-L1 binding. The data clearly demonstrate that the use of multiplexed deep sequencing methods similarly identified those mutants previously identified as weak PD-1 binding factors.

Figure 20: Detection of T cell ligand-receptor binding using the cell microarray platform described herein was used to examine several IgSF receptor/ligand pairs. Binding was specific.

Figure 21: Identification of selectively functional PD-L1 mutants using a bead-based platform.

Figure 22: Slides were spotted with plasmid DNA encoding wild-type PD-L1, mCherry alone, or a series of PD-L1 mutants. The position of each spotted construct is shown on the far right of the figure. The fluorescent signal from the mCherry (Cy3) laser reports the expression level of each spotted construct. The Alexa 647 (Cy5) laser shows the binding of PD-1 (top panel) or B7-1 (bottom panel) Fc fusion protein to PD-L1-expressing cells. Therefore, spots that do not fluoresce in the Cy5 channel do not show binding. The grid on the right is color-coded to indicate the binding observed using the microbead binding experiment. Those green constructs showed binding similar to wild-type. Yellow mutants showed reduced binding and red mutants showed little to no binding. The same binding pattern was observed on the microarray.

Figure 23: Extending the cell-cell binding assay to a 96-well plate format suitable for high-throughput protein-protein interaction screening. A) Two replicates of 16 control cell samples were set to three different total cell concentrations and incubated for 2 hours at 4°C in a 96-well deep-well plate. Cell-cell mixtures containing cognate ligand: receptor pairs and therefore should exhibit significant binding are highlighted in green (the lower two rows of Table A). After incubation, an aliquot of cells was transferred to a 96-well U-bottom plate and analyzed using an Intellicyt HTFC continuous flow system connected to a BD Accuri cell counter. B) A well detector view from a representative 96-well plate run. Each peak represents an event collected from one well. Identifiers mark the end of each row of the 96-well plate. Note that the size of the peak reflects the concentration of cells in the well, making the three different cell concentrations clearly distinguishable. C) Gate GFP fluorescence, mCherry fluorescence, or both (double-positive "hits") from all singlet cells in the entire 96-well plate. D) Heat map showing double positive "hits" as a percentage of all singlet cell events. E) Graph showing double positives (hits) as a percentage of total singlet cell events for all 16 control cell samples at each cell concentration tested. Note that the percentage of cell-cell binding decreases as the total cell concentration decreases, but the fold difference between negative and positive pairs increases slightly from approximately 6-fold at 1× ¹⁰ cells/mL to approximately 10-fold at 0.2× ¹⁰ cells/mL. Data represent the mean of two independent experiments, and error bars represent standard deviations.

Figure 24: Cell microarray expressing members of the IgG superfamily. A) Poly-l-lysine-coated slides were spotted with expression constructs for 144 human genes in the IgG superfamily. Each construct was spotted, and each row was repeated 4 times, resulting in an array of 4×144 spots in total. After transfection, the expression of each spotted construct could be directly observed via mCherry fluorescence (pseudo green, single channel not shown). The cell array was then treated with recombinant PD-L1-Fc pre-incubated with Alexa 647-labeled anti-human IgG, washed and fixed with 4% formaldehyde (pseudo red, single channel not shown). The data show overlapping green and red pseudo-color images, in which binding was observed as yellow/orange produced by the merging of green and red fluorescence signals. The rows labeled A and B contain two known binding targets of PD-L1, PD-1 (A) and B7-1 (B). B) 10X magnification of the highlighted row in A, showing the positive signal observed for PD-L1 binding to PD-1 and B7-1 compared to the signal observed from surrounding spots.

Detailed Description of the Invention

A cell microarray is provided, comprising:

(i) a first plurality of cells transformed to express a first predetermined heterologous secreted protein, a heterologous membrane protein, or a heterologous cell surface protein and a first fluorescent protein and (ii) at least a second plurality of cells transformed to express a second predetermined heterologous secreted protein, or a second heterologous protein and a second fluorescent protein,

wherein the first and second pluralities of cells are attached to a solid surface of the microarray, and wherein the first and second pluralities of cells are in spatially distinct locations on the solid surface.

A cell microarray is provided, comprising:

(i) a first plurality of cells transformed to express (a) a first predetermined heterologous protein and (b) a first fluorescent protein and (ii) at least a second plurality of cells transformed to express (a) a second predetermined protein and (b) a second fluorescent protein,

wherein the first and second pluralities of cells are attached to a solid surface of the microarray, and wherein the first and second pluralities of cells are in spatially distinct locations on the solid surface.

In one embodiment, the first or second predetermined protein is a classical secreted protein. In one embodiment, the first or second predetermined protein is a non-classical secreted protein. Non-classical secretion includes proteins such as FGF2 that have a well-defined non-classical secretory pathway and cytoplasmic proteins released due to cell lysis/death.

In one embodiment, the cell microarray further comprises a fusion protein comprising (i) a candidate protein or peptide ligand of one of the heterologous proteins and (ii) a third fluorescent protein bound to one of the heterologous proteins, or further comprises a compound comprising a peptide or protein ligand of one of the heterologous proteins, the compound having a third fluorescent protein bound thereto via a non-peptide bond, wherein the compound is bound to one of the heterologous proteins of the cell microarray.

In one embodiment, the cell microarray further comprises a third plurality of cells as a control, the third plurality of cells optionally transformed to express the first fluorescent protein but not transformed with the first or second predetermined heterologous protein.

In one embodiment, each plurality of cells is a plurality of mammalian cells.

In one embodiment, the mammalian cell is an isolated human cell.

In one embodiment, the mammalian cell is a cell of the human embryonic kidney (HEK) cell line.

In one embodiment, the mammalian cell is a HEK293 cell line cell.

In one embodiment, the microarray comprises at least 10 different pluralities of cells, each plurality of cells transformed to express a predetermined heterologous protein and a first fluorescent protein, the heterologous protein being different from the heterologous protein expressed by each other plurality of transformed cells in the microarray.

In one embodiment, the microarray comprises at least 100 different pluralities of cells, each plurality of cells transformed to express a predetermined heterologous protein and a first fluorescent protein, the heterologous protein being different from the heterologous protein expressed by each other plurality of transformed cells in the microarray.

In one embodiment, the first and/or fluorescent protein is green fluorescent protein or yellow fluorescent protein.

In one embodiment, the third fluorescent protein is red fluorescent protein.

In one embodiment, each plurality of cells is transformed only to express the first predetermined heterologous protein and the first fluorescent protein, and is not transformed to express any other heterologous protein.

In one embodiment, the first predetermined heterologous protein is a subunit of a multi-subunit heterologous protein, and the plurality of cells are further transformed to express one or more remaining members of the multi-subunit heterologous protein.

In one embodiment, the first predetermined heterologous protein is linked to the first fluorescent protein via its C-terminus when expressed.

In one embodiment, the first predetermined heterologous protein is linked to a transmembrane anchor peptide when expressed.

In one embodiment, the cell microarray is made by immobilizing a first plurality of expression constructs encoding a first heterologous protein and a fluorescent protein on a solid surface of a microarray at locations spatially distinct from the immobilized first plurality of expression constructs, immobilizing at least a second plurality of expression constructs encoding a second heterologous protein and a fluorescent protein on the solid surface of the microarray, and contacting the expression constructs with a plurality of cells under conditions including the presence of a transfection agent to allow at least a portion of the cells at each spatially distinct location to be transfected with the corresponding expression construct.

In one embodiment, the expression construct comprises a pEGFP-N1 expression construct. In one embodiment, the expression construct comprises a CMV promoter.

In one embodiment, the cell is an insect cell. In one embodiment, the cell is a Drosophila S2 cell.

In one embodiment, the first or second predetermined heterologous protein is an immunoglobulin superfamily protein, a TNF receptor protein, a cytokine, a chemokine, a type 1 transmembrane receptor protein, a type 2 transmembrane receptor protein, an ion channel protein, or a membrane transporter protein.

In one embodiment, the first or second predetermined heterologous protein as described herein is 1) a protein in the entire secretome of humans (i.e., approximately 8,000 secretory proteins and integral membrane proteins, including GPCRC); 2) a non-classical secretory protein of humans/mouse; 3) a cytoplasmic protein or a secreted protein that exhibits an extracellular function via binding to the cell surface; or 4) a pathogen secreted protein or an integral membrane protein.

In one embodiment, the first or second predetermined heterologous protein is a toll-like receptor, a TNF receptor, a GPCR, a growth factor receptor, a nectin, an interleukin, or an interleukin receptor.

In one embodiment, the first or second predetermined heterologous protein is a mammalian protein.

In one embodiment, the first or second predetermined heterologous protein is expressed at a localized location on the plasma membrane. In one embodiment, the first and/or second heterologous protein is a secretory protein, a transmembrane protein, or a cell surface protein. In one embodiment, the cell microarray comprises one of 100, 200, 300, 400, or 500 or more different cells transformed to express a heterologous protein, wherein each of the multiple cells expresses a heterologous protein different from each other heterologous protein expressed by multiple transformed cells of other species. In one embodiment, the cell microarray comprises 750 or more different cells transformed to express a heterologous protein, wherein each of the multiple cells expresses a heterologous protein different from each other heterologous protein expressed by multiple transformed cells of other species. In one embodiment, the cell microarray comprises 1000 or more different cells transformed to express a heterologous protein, wherein each of the multiple cells expresses a heterologous protein different from each other heterologous protein expressed by multiple transformed cells of other species.

In one embodiment, the heterologous protein is a secreted protein and is expressed as a fusion to a transmembrane helix.

In one embodiment, the first fluorescent protein and the second fluorescent protein are of the same type, and the third fluorescent protein is of a different type.

In one embodiment, each plurality of cells is divided into spots of a plurality of cells, each multiple of cells being less than the total number of cells in the plurality of cells, and wherein each spot is arranged so as to be closer to another spot of the same plurality of cells than to a spot of another plurality of cells.

A method for preparing a cell microarray as described herein is provided, comprising: immobilizing a first plurality of expression constructs encoding a first heterologous protein and a first fluorescent protein on a solid surface of a microarray at locations spatially distinct from the immobilized first plurality of expression constructs, immobilizing at least a second plurality of expression constructs encoding a second heterologous protein and a second fluorescent protein on the solid surface of the microarray, and contacting the expression constructs with a plurality of cells under conditions including the presence of a transfection agent so as to allow the cells to attach to the solid surface and transfection of at least a portion of the cells occurring at each spatially distinct location with a corresponding expression construct.

In one embodiment, the expression construct can encode a single transcript of a fusion protein comprising a heterologous protein and a fluorescent protein as a single covalent fusion polypeptide. In one embodiment, the expression construct can encode the heterologous protein and the fluorescent protein as two different polypeptides (e.g., an IRES construct). In one embodiment, the expression construct is prepared using ligation-independent cloning (LIC). In one embodiment, conventional restriction site cloning is used.

Also provided is a method for determining whether a candidate protein or peptide binds to a second protein or peptide, the method comprising expressing the second protein as a heterologous protein on a cell microarray as described herein, and contacting the cell microarray with the candidate protein or peptide, wherein the candidate protein or peptide has a third fluorescent protein or peptide immobilized thereon, washing the cell microarray contacted with the candidate protein or peptide to remove unbound candidate protein or peptide, and determining whether any candidate protein or peptide is bound to the cell microarray after washing, wherein the presence of the candidate protein or peptide bound to the cell microarray at a first spatial location corresponding to cells transformed with the first heterologous protein after washing indicates that the candidate protein or peptide is bound to the first heterologous protein, and wherein the absence of the candidate protein or peptide bound to the cell microarray at the first spatial location after washing indicates that the candidate protein or peptide is not bound to the heterologous protein.

In one embodiment, determining whether any candidate protein or peptide is present that is bound to the cell microarray after washing is accomplished by measuring the fluorescence of a third fluorescent protein and determining its location on the cell microarray, wherein colocalization of the third fluorescent protein with the first or second fluorescent protein at a spatially distinct location indicates that the first protein or peptide is bound to a heterologous protein corresponding to the spatially distinct location.

Also provided is a system comprising (i) a microarray solid surface and a suspension-adapted cell line transformed to express a candidate ligand protein or peptide and a first C-terminal cytoplasmically expressed fluorescent protein on its cell surface; and (ii) at least a) a second plurality of cells transformed to express a predetermined heterologous protein and a second fluorescent protein on its cell surface, or b) a plurality of microbeads to which the heterologous protein has been immobilized and to which the second fluorescent protein has been immobilized, wherein a) or b) is immobilized to the microarray solid surface. As above, mutatis mutandis, a system is also provided, wherein the candidate ligand protein or peptide is expressed on the second plurality of transformed cells or the plurality of microbeads, and the heterologous protein is expressed on the cell surface of the transformed suspension-adapted cell line.

Cells on the microarray can be probed using 1) fluorescently labeled probe proteins; 2) probe proteins presented on fluorescent microbeads; and/or 3) cells expressing probe molecules on their surface.

In one embodiment, the system further comprises c) one or more additional pluralities of cells transformed so as to express a different predetermined heterologous protein and a second fluorescent protein on their cell surface, or d) one or more additional pluralities of microbeads to which the different predetermined heterologous protein has been immobilized on its surface and to which a second fluorescent protein has been immobilized, wherein c) or d) is immobilized on the microarray solid surface at a spatially different location from the plurality of a) and/or b).

In one embodiment, the heterologous protein is immobilized to the beads via Protein A molecules.

In one embodiment, a suspension-adapted cell line transformed to express a candidate ligand protein or peptide on its cell surface has been transiently transfected with a nucleic acid construct encoding the candidate ligand protein or peptide. In one embodiment, the heterologous protein is immobilized on microbeads by binding to an antibody attached to the microbeads. In one embodiment, the first and second fluorescent proteins are different colors. In one embodiment, one fluorescent protein is green and the other fluorescent protein is red. Non-limiting examples include green fluorescent protein and mCherry ^™ .

In one embodiment, the plurality of cells are a plurality of mammalian cells. In one embodiment, the mammalian cells are isolated human cells. In one embodiment, the mammalian cells are human embryonic kidney (HEK) cell line cells. In one embodiment, the mammalian cells are HEK293 cell line cells.

In one embodiment, the predetermined heterologous protein is a subunit of a multi-subunit heterologous protein, and the plurality of cells are also transformed to express one or more remaining members of the multi-subunit heterologous protein. In one embodiment, the predetermined heterologous protein is a secreted protein, a membrane protein, or a cell surface protein. In one embodiment, the predetermined heterologous protein is linked to a fluorescent protein via its C-terminus during expression. In one embodiment, the predetermined heterologous protein is a secreted protein and, during expression, is linked to a transmembrane anchor peptide or protein. In one embodiment, the expression construct comprises a pEGFP-N1 expression construct and/or a CMV promoter. In one embodiment, the heterologous protein is an immunoglobulin superfamily protein, a TNF receptor protein, a cytokine, a chemokine, a type 1 transmembrane receptor protein, a type 2 transmembrane receptor protein, an ion channel protein, or a membrane transporter protein. In one embodiment, the heterologous protein is a toll-like receptor, a TNF receptor, a GPCR, a growth factor receptor, an adhesion protein, an interleukin, or an interleukin receptor. In one embodiment, the heterologous protein is a mammalian protein. In one embodiment, the heterologous protein is expressed at a localized location on the plasma membrane.

A method for determining whether a candidate ligand protein or peptide binds to a second protein or peptide, the method comprising expressing the candidate ligand protein or peptide and a first fluorescent protein in a plurality of cells of a suspension-adapted cell line of the present system, contacting the plurality of cells with a) a second plurality of cells transformed to express the heterologous protein and the second fluorescent protein, or b) a plurality of microbeads having the heterologous protein and the second fluorescent protein immobilized on their surfaces, washing to remove unbound candidate ligand protein or peptide, and identifying cells showing colocalization of the first and second fluorescent proteins by FACS analysis, wherein cells showing colocalization of the first and second fluorescent proteins at spatially different locations indicate that the first protein or peptide binds to the heterologous protein corresponding to the spatially different locations.

In one embodiment, colocalization of the first and second fluorescent proteins is determined by FACS analysis.

In certain embodiments of hemophilic interactions, the candidate ligand protein or peptide and the second protein or peptide have the same sequence.

A system is provided that includes a first plurality of suspension-adapted cell line cells transformed with a vector so as to express a first heterologous candidate ligand protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein, wherein the vector comprises a single predetermined sequence of 15-35 nucleotides of the first heterologous candidate ligand protein or peptide, the single sequence being capable of being primed by one or more universal primers; and a second plurality of suspension-adapted cell line cells transformed with a second vector so as to express a second heterologous candidate ligand protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein, wherein the second vector comprises a different single predetermined sequence of 15-35 nucleotides of the second heterologous candidate ligand protein or peptide; and (i) one or more additional pluralities of suspension-adapted cell line cells transformed with a vector so as to express a receptor protein or peptide on their cell surface and a second fluorescent protein, the second suspension-adapted cell line comprising a stably expressed peptide cell surface epitope, or (ii) a plurality of magnetic microbeads to which a receptor protein has been immobilized and to which a second fluorescent protein has been immobilized.

In one embodiment, the receptor protein may be a classical recognition receptor. In one embodiment, the receptor protein may not be a classical recognition receptor, but merely a receptor protein for the ligand.

A system is provided that includes a first plurality of suspension-adapted cell line cells transformed with a vector so as to express a first heterologous protein on their cell surface and to express a first cytoplasmically expressed fluorescent protein, wherein the vector comprises a single predetermined sequence of 15-35 nucleotides of the first heterologous protein, the single sequence being capable of being primed by one or more universal primers; and a second plurality of suspension-adapted cell line cells transformed with a second vector so as to express a second heterologous protein on their cell surface and to express the first cytoplasmically expressed fluorescent protein, wherein the second vector comprises a different single predetermined sequence of 15-35 nucleotides of the second heterologous protein; and (i) one or more additional pluralities of suspension-adapted cell line cells transformed with a vector so as to express a candidate ligand protein or peptide on their cell surface and to express a second fluorescent protein, wherein the second suspension-adapted cell line comprises a stably expressed peptide cell surface epitope, or (ii) a plurality of magnetic microbeads to which the candidate ligand protein or peptide has been immobilized on its surface and to which the second fluorescent protein has been immobilized.

In one embodiment, the universal primers include T7 forward and reverse universal primers.

In one embodiment, the peptide cell surface epitope is a FLAG epitope (DYKDDDDK) (SEQ ID NO: 1). In one embodiment, the system further comprises an anti-FLAG epitope antibody comprising a magnetic molecular entity, wherein the antibody binds to the FLAG epitope.

In one embodiment, the magnetic molecular entity is a superparamagnetic impregnated bead.In one embodiment, the single predetermined sequence of 20-35 nucleotides is 28 nucleotides in length.

Also provided is a method for determining whether a candidate ligand protein or peptide binds to a second predetermined protein, the method comprising expressing the second predetermined protein as a heterologous protein of the current system and contacting the candidate ligand protein or peptide with (i) one or more additional plurality of suspension-adapted cell line cells transformed so as to express the candidate ligand protein or peptide on their cell surfaces and express a second fluorescent protein, the second suspension-adapted cell line comprising a stably expressed peptide cell surface epitope, or with (ii) the candidate ligand protein or peptide of the plurality of magnetic microbeads to which the candidate ligand protein or peptide has been immobilized and to which the second fluorescent protein has been immobilized;

separating any one of the first plurality of suspended adaptive cell line cells that is bound to one or more of the second plurality of cells or to the plurality of magnetic microbeads by magnetic attraction;

obtaining DNA from such isolated cell-cell or cell-bead conjugates and amplifying the unique sequence, if present in the DNA, using said universal primers;

sequencing copies of the unique sequence to confirm their presence;

The unique sequence thus identified is compared to a database associating the unique predetermined sequence of 15-35 nucleotides with specific heterologous proteins or peptides, and any heterologous proteins or peptides thus associated are identified as binding, thereby identifying the specific heterologous protein or peptide as binding to the candidate protein or peptide.

In one embodiment, the candidate ligand protein or peptide is fixed to the microbeads via protein A molecules. In one embodiment, the candidate ligand protein or peptide is fixed to the microbeads by being bound by an antibody attached to the microbeads. In one embodiment, the first and second fluorescent proteins are different colors. In one embodiment, the one fluorescent protein is green and the other fluorescent protein is red. Non-limiting examples of such fluorescent proteins are provided above. In one embodiment, the multiple cells are plural. In one embodiment, the mammalian cell is a human embryonic kidney (HEK) cell line cell. In one embodiment, the mammalian cell is a HEK293 cell line cell. In one embodiment, the predetermined heterologous protein is a subunit of a multi-subunit heterologous protein, and the multiple cells are also transformed to express one or more remaining members of the multi-subunit heterologous protein. In one embodiment, the predetermined heterologous protein is connected to a fluorescent protein via its C-terminus when expressed. In one embodiment, the predetermined heterologous secretory protein is connected to a transmembrane anchor peptide when expressed. In one embodiment, the heterologous protein is an immunoglobulin superfamily protein, a TNF receptor protein, a cytokine, a chemokine, a type 1 transmembrane receptor protein, a type 2 transmembrane receptor protein, an ion channel protein, or a membrane transporter. In one embodiment, the heterologous protein is a toll-like receptor, a TNF receptor, a GPCR, a growth factor receptor, an adhesion protein, an interleukin, or an interleukin receptor. In one embodiment, the heterologous protein is a mammalian protein. In one embodiment, the heterologous protein is expressed at a localized location on the plasma membrane.

A system comprising (i) a first plurality of suspension-adapted cell line cells, wherein the plurality of cells are transformed with a vector so as to (a) express a heterologous protein on their cell surface and (b) express a first cytoplasmically expressed fluorescent protein, and wherein the vector comprises a single predetermined sequence of 15-35 nucleotides of the expressed heterologous protein, such that the first plurality of suspension-adapted cell line cells express at least two different types of the first heterologous protein, and (ii) a second plurality of suspension-adapted cell line cells, wherein the cells are transformed with a second vector so as to express a second heterologous protein on their cell surface and a second cytoplasmically expressed fluorescent protein, wherein the second plurality of suspension-adapted cells express a single type of the second heterologous protein. In one embodiment, any single cell in the first plurality of cells expresses only one heterologous protein on its cell surface. In one embodiment, no different type of the first heterologous protein of the first plurality of cells has the same sequence as the second heterologous protein of the second plurality of cells.

In one embodiment, the second heterologous protein is a membrane receptor. In one embodiment, each heterologous protein expressed in the first plurality of suspension-adapted cell line cells is a secreted peptide, polypeptide, or protein. In one embodiment, the plurality of different types of first heterologous proteins are each mutant of a predetermined wild-type protein. In one embodiment, the second heterologous protein is a wild-type protein. In one embodiment, each type of heterologous protein in the first plurality of different proteins differs from each other type of heterologous protein in the plurality by 1, 2, 3, 4, or 5 amino acid residue point mutations. In one embodiment, each type of protein in the plurality of different proteins differs from each other type of heterologous protein in the plurality by 1 amino acid residue point mutation.

In one embodiment, the unique sequence can be primed by one or more universal primers. In one embodiment, the unique sequence is 15-35 nucleotides. In one embodiment, the first or second fluorescent protein is green. In one embodiment, the other fluorescent protein is red.

Also provided is a method for determining whether a candidate ligand protein or peptide binds to a second protein or peptide, the method comprising expressing the candidate ligand protein or peptide as a first heterologous protein in a first plurality of cells of a system as described herein and expressing the second protein or peptide as a second heterologous protein in a system as described herein under conditions permissive for binding of the first heterologous protein to the second heterologous protein, and optionally, washing to remove any unbound first heterologous protein, then recovering cells in which the first and second heterologous proteins co-localize, obtaining nucleic acid from the recovered cells and sequencing the nucleic acid to identify a single sequence of 15-35 nucleotides contained therein to identify the candidate ligand protein or peptide that has bound the second protein or peptide corresponding to the single 15-35 nucleotides.

Also provided is a method for determining the effect of a predetermined amino acid residue of a first protein on the binding of the first protein to a second protein, the method comprising expressing a protein mutated by one or more point mutations relative to the first protein as a plurality of different types of heterologous proteins in a plurality of cells of a first suspension-adapted cell line of a system as described herein, and contacting the plurality of proteins with a second protein in the form of a second heterologous protein in a second plurality of cells transformed to express the second protein and a second fluorescent protein in the system as described herein, and recovering cells that exhibit colocalization of the first and second fluorescent proteins, obtaining nucleic acid from the recovered cells and sequencing the nucleic acid to identify a unique sequence of 15-35 nucleotides contained therein to identify the first protein that has bound the second protein or peptide, and comparing the level of the protein that has bound the second protein or peptide to a predetermined reference level.

Wherein the level of the protein to which the second protein or peptide has been bound exceeds a predetermined reference level, it indicates that the one or more residues mutated in the protein enhance the binding of the first protein to the second protein, and where the level of the protein to which the second protein has been bound is below a predetermined reference level, it indicates that the one or more residues mutated in the protein inhibit the binding of the first protein to the second protein.

In one embodiment, the predetermined level is a control. In one embodiment, the predetermined level is obtained by measuring the level of the first protein that does not mutate and the second protein combination. In one embodiment of the method, the cells showing the colocalization of the first and second fluorescent proteins are recovered by FACS analysis.

Also provided is a system comprising (i) a first plurality of suspension adaptive cell line cells transformed with a vector so as to express a first heterologous candidate ligand protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein, and a second plurality of suspension adaptive cell line cells transformed with a second vector so as to express a second heterologous candidate ligand protein or peptide on their cell surface and a second cytoplasmically expressed fluorescent protein, and (ii) a plurality of magnetic microbeads to which target proteins, peptides or antibodies have been immobilized.

Also provided is a method for determining whether one or more of two candidate ligand proteins or peptides binds to a target protein, peptide, or antibody, the method comprising expressing a first candidate ligand protein or peptide as a first heterologous protein in a first plurality of cells in a current system and expressing a second candidate ligand protein or peptide as a second heterologous protein in the current system under conditions permissive for binding of the first heterologous protein and the second heterologous protein to the target protein, peptide, or antibody, and recovering any microbeads complexed with the first fluorescent protein-expressing cells and/or complexed with the second fluorescent protein-expressing cells, and identifying the candidate ligand proteins in the complexes, wherein recovering microbeads in complex with the first fluorescent protein-expressing cells indicates that the first candidate ligand protein or peptide binds to the target protein or peptide, and wherein recovering microbeads in complex with the second fluorescent protein-expressing cells indicates that the second candidate ligand protein or peptide binds to the target protein or peptide, and wherein not recovering microbeads in complex with the first fluorescent protein-expressing cells or the second fluorescent protein-expressing cells indicates that the first candidate ligand protein does not bind to the target protein or peptide, and the second candidate ligand protein does not bind to the target protein or peptide, respectively.

Also provided is a system comprising (i) a first plurality of suspension adaptive cell line cells transformed with a vector so as to express a first heterologous target protein or peptide on their cell surface and a first cytoplasmically expressed fluorescent protein and one or more second plurality of suspension adaptive cell line cells transformed with a second vector so as to express a second heterologous candidate ligand protein or peptide on their cell surface and a second cytoplasmically expressed fluorescent protein, and (ii) a plurality of magnetic microbeads to the surface of which antibodies against the candidate ligand protein or peptide or against the target protein or peptide have been immobilized. Also provided is a method for determining whether a candidate ligand protein or peptide binds to a target protein or peptide, the method comprising expressing the candidate ligand protein or peptide as a second heterologous protein in a second plurality of cells in the current system and expressing the target protein or peptide as a first heterologous protein in the current system under conditions that allow binding of the candidate ligand protein or peptide to the target protein or peptide, and recovering any microbeads complexed with a first fluorescent protein-expressing cell and a second fluorescent protein-expressing cell, wherein recovering microbeads attached to a complex of the first fluorescent protein-expressing cell and the second fluorescent protein-expressing cell indicates that the candidate ligand protein or peptide binds to the target protein or peptide, and wherein not recovering microbeads attached to a complex of the first fluorescent protein-expressing cell and the second fluorescent protein-expressing cell indicates that the candidate ligand protein does not bind to the target protein or peptide.

In one embodiment of the method, cells that exhibit colocalization of the first and second fluorescent proteins are magnetically sorted. Magnetic entities, such as beads, that can be attached to the second plurality of cells are magnetically separated when identifying cells that exhibit colocalization of the first and second fluorescent proteins. Thus, the methods and systems described herein can include magnetic entities, such as magnetic microbeads, attached to the second plurality of cells and can include attaching magnetic entities, such as magnetic microbeads, attached to cells in the second plurality of cells.

In one embodiment, the heterologous protein or peptide is heterologous to the cell in which it is expressed with respect to the source of the protein (e.g., another cell type or another species). In one embodiment, the heterologous protein or peptide is heterologous to the cell in which it is expressed with respect to its location, e.g., the protein is not expressed at a location (e.g., a cell surface) under normal physiological conditions (e.g., in vivo).

In one embodiment of the method, PCR is performed on a single sequence of 15-35 nucleotides. In one embodiment, deep sequencing is performed on the merged PCR products to identify the single sequence of 15-35 nucleotides. In one embodiment of the method, the method includes determining whether the single sequence of 15-35 nucleotides is enriched after sorting (or after recovery) and before sorting (or before recovery).

In one embodiment of the methods and systems described herein, the single sequence is 20-35 nucleotides. In one embodiment of the methods and systems described herein, the single sequence is 20-30 nucleotides. In one embodiment of the methods and systems described herein, the single sequence is 25-30 nucleotides. In one embodiment of the methods and systems described herein, the single sequence is 20 nucleotides in length. In one embodiment of the methods and systems described herein, the single sequence is 28 nucleotides in length.

In one embodiment of the methods described herein, the co-localized cells are lysed or the cells are recovered and the contents of their supernatant are sequenced.

In another embodiment of the methods described herein, the method is performed in each well of a multiwell dish using a different amplicon than in the remaining wells. In one embodiment, different wells of the multiwell dish contain different receptor proteins.

All combinations of the elements described herein are within the scope of the invention unless otherwise indicated herein or otherwise clearly contradicted by context.

The present invention will be better understood from the experimental details that follow. However, one skilled in the art will readily recognize that the specific methods and results discussed are merely illustrative of the present invention, which is more fully described in the claims that follow.

Experimental details

The need for continuous structural characterization of receptor:ligand complexes: A wide range of biomolecules, including members of the immunoglobulin (Ig), TNF/TNFR, GPCR, chemokine, and receptor kinase superfamilies, are crucial for the goal of systematic structural characterization of the secretome. The CD28 receptor family (i.e., CD28, CTLA-4, ICOS, and PD-1) is described below, a subclass of the immunoglobulin superfamily (IgSF) that provides the primary signal for optimal T cell function [41–43]. These signaling receptors share structural features and recognize related cell surface ligands (e.g., B7-1, B7-2, ICOS-L, PD-L1, and PD-L2) with similar interaction patterns (Figure 1) [44,45]. For example, engagement of CD28 by B7-1 and B7-2 leads to T cell activation, whereas engagement of the same B7 ligand by CTLA-4 provides an inhibitory signal. The inducible co-stimulatory receptor (ICOS) provides an additional important positive signal (i.e., co-stimulation) upon binding ICOS-L, and engagement of PD-1 with either of its two B7-like ligands, PD-L1 and PD-L2, initiates an additional inhibitory pathway (i.e., co-inhibition). The structures of these molecules and complexes, including several from this laboratory [28,35,46-48], have helped define the basic mechanistic features (e.g., oligomeric state, valence, ligand specificity, etc.) and overall organizational principles required for co-stimulatory and co-inhibitory signaling as summarized in Figure 1 [45].

In addition to defining fundamental biophysical and histological features, these structures also provide the basis for the generation of novel biochemical reagents and unique mechanistic information model systems. For example, guided by the structure of the PD-1:PD-L2 complex disclosed herein [35], a mutant mouse PD-1 receptor was generated that binds to mouse PD-L2 with wild-type affinity but does not exhibit interaction with PD-L1. Based on these findings, the generation of Ig-fusion proteins and knock-in mouse models is disclosed herein, which provide unprecedented opportunities to dissect the mechanistic role of two ligand-related signaling pathways in normal physiology and disease (Figure 2). In addition, based on the structure of this complex, a single point mutant of the human PD-1 receptor was generated that has 50-fold and 30-fold higher affinity for human PD-L1 and PD-L2, respectively (unpublished data). This reagent represents a new class of high-affinity agents that may provide unique therapeutic opportunities (see below). These examples highlight the value of such receptor:ligand structures and the ways in which they can be influenced for biological insights and new therapeutic opportunities.

New key receptor:ligand interactions remain to be defined. Of particular relevance to this application, even within the much-studied CD28 family of receptors, additional important interactions have only recently been discovered. B7-1 has been shown to bind to PD-L1, generating a bidirectional inhibitory signal, while ICOS-L has been shown to bind to both CD28 and CTLA4, and the CD28:ICOS-L interaction is required for human T cell activation [49, 50]. These cross-competitive interactions generate a highly complex network of signaling pathways. These examples highlight the value of systematically defining the entire component of receptor:ligand interactions, as the discovery of even a single new receptor:ligand pair can significantly impact the mechanistic understanding of signaling pathways relevant to T cell function, human physiology, and disease.

The therapeutic relevance of receptor:ligand complexes. Importantly, many cell surface molecules and their associated binding partners are excellent targets for deliberate modulation of signaling pathways to treat a variety of human diseases. Function-blocking antibodies targeting cell surface immune receptors and ligands are an important class of protein therapeutics used to manipulate immune responses to treat infectious diseases, autoimmune diseases, and malignancies. A leading example includes Yervoy ^™ (Bristol Myers Squibb), a function-blocking mAb against the CTLA-4 inhibitory receptor that produces comprehensive immune stimulation and was approved by the FDA in March 2011 for the treatment of advanced melanoma [51]. These immune receptors are not only targets but also effective therapeutic agents in their own right. For example, the soluble form of CTLA-4 marketed as Orencia ^™ (BMS) competes with CD28 for binding to the B7 ligand, resulting in inhibition of CD28-related stimulation pathways. Blockade of CD28 stimulation results in comprehensive immune suppression, making Orencia a leading treatment for autoimmune diseases including rheumatoid arthritis [52]. Of particular note is Belatacept ^™ (BMS), a soluble CTLA-4 variant of Orencia with two point mutations. Belatacept was approved by the FDA in November 2011 for the prevention of acute renal transplant rejection, showing efficacy comparable to existing therapies and, due to the mutations, greatly reduced side effects and toxicity. Remarkably, Belatacept only increased its affinity for the B7 ligand by two-fold, but its biological potency was enhanced by 10-fold [26, 53]. This finding strongly supports the continued focus on structural and biochemical analysis of major co-stimulatory molecules and their cognate complexes to gain molecular insights that will aid in the development of new therapeutics. These principles can be extended to the entire secretome.

Implementation of the proposed high-throughput technology will provide powerful tools for, for example, the study of interactions associated with the human secretome, defining the scope of extracellular host:pathogen interactions associated with viral, bacterial, fungal, and parasitic diseases (e.g., [54]), and identifying host:pathogen interactions. In addition, recent evidence suggests that many "ostensibly" cytoplasmic proteins also have extracellular functions [55-59]. Non-classical secretion mechanisms (i.e., secretion that is independent of signal sequences) are increasingly being described and are the subject of considerable research [60,61]. Notably, the cell surface receptors for many of these non-classical secretory proteins have not yet been identified.

The considerable value of defining interactions associated with the mammalian secretome has long been recognized and has attracted considerable attention from small biotechnology companies, large pharmaceutical companies, and academic groups. Efforts originating from academic laboratories have been conducted on a very modest scale [62, 63]; the most prominent/extensive examples include contributions from Genentech and Five Prime Therapeutics, Inc. Genentech has used its considerable resources to generate a library of >1000 Ig-fusion proteins for direct binding analysis using surface plasmon resonance technology. These efforts have led to the discovery of new ligands for the Ig superfamily members BTLA and TIGIT [64, 65]. Given that each individual target (e.g., TIGIT) requires individual screening against each member of the library, this approach lacks the features required to achieve true high throughput. Genentech recently described a protein array in which approximately 700 secreted proteins were individually pinned to a solid support; this array was subsequently screened with a multivalent reagent that individually presents approximately 90 human Ig-fusion proteins [66]. This platform facilitated the discovery of new and surprising receptor:ligand interactions, including an unexpected interaction between B7-1 and NGFR.

Compared to these "array" approaches, Five Prime Therapeutics, Inc. took a more "brute force" approach in which approximately 3400 constructs of secretory proteins and extracellular domains of transmembrane proteins were individually expressed in 293T cells [67]. These proteins were examined in 30 different HTP assays that probed metabolic, transcriptional, and growth responses related to immune and cardiovascular function and cancer proliferation in a wide variety of cell lines. These efforts demonstrated that the previously uncharacterized protein IL-34 is a ligand for the (ostensibly) well-characterized colony stimulating factor 1 receptor. This study is a prime example of the need to de-orphanize molecules (e.g., IL-34) and highlights the fact that even well-characterized cell surface molecules can have unknown interactions.

These high-throughput methods for identifying receptor:ligand interactions are among the most exciting recent developments in the biological sciences, as they hold the potential to uncover new fundamental biological mechanisms and generate novel therapeutic strategies. However, for several reasons, these studies may not achieve the broad impact they might otherwise require. First, the ability to generate the large quantities of secreted proteins/Ig-fusions required for these assays is beyond the capabilities of even the most ambitious academic laboratories, including those supported by the Protein Structure Initiative. Furthermore, all of these approaches fail when proteins cannot be purified, exhibit instability during storage and subsequent manipulation, or exhibit adverse solution behavior (e.g., aggregation, which typically affects Ig-fusion proteins). In the case of the Five Prime Therapeutics screen, biologically functional proteins (or cognate binding proteins) not covered by the selected cell-based screen did not interact. Of particular note, all of these methods are unsuitable for some of the most important classes of integral membrane proteins, including GPCRs, transporters, and channels, as these proteins are generally incompatible with high-throughput purification of functionally active species and cannot tolerate the physical procedures of the assays. Finally, and perhaps most importantly, the results reported by these commercial efforts represent only those interactions deemed acceptable for release to the public; many "non-scientific" factors influence these decisions and it is almost certain that a large percentage of these important data will never enter the public domain.

Disclosed herein are three affordable, efficient, and high-throughput technologies for identifying interactions involving, for example, mammalian secretomes for high-resolution structural discovery, biochemical analysis, and therapeutic agent development. The disclosed technologies offer numerous advantages over existing methods: 1) expression in a cell microarray format allows for the systematic expression of all classes of proteins (including multi-spanning membrane-spanning integral proteins such as GPCRs and transporters, as well as multicomponent receptors such as integrins); 2) cell microarray expression is very tractable because only DNA (i.e., expression vectors) is required, not the purification of the protein itself; 3) the technologies are all based on the detection of direct physical interactions and therefore do not require any understanding of biological function; 4) flow cytometry-based methods allow both decoys and captures to be expressed on the surface of separate and distinguishable cells, thereby eliminating all need for purified proteins; and 5) magnetic separation coupled with deep sequencing and barcoded libraries of secretory protein-expressing cells provides massively parallel interrogation of many (all) ligands relative to the entire set of potential receptors.

Development of cell microarrays for high-throughput characterization of cell surface protein-protein interactions: Cell microarray technology has been modified to systematically screen a pan-genomic library of cell surface receptors (i.e., the secretome) for a single query ligand. This approach presents many receptors in the presence of live host cells in a previously arrayed format. Cell microarray technology has been successfully modified to efficiently screen libraries of potential receptor constructs [68,69]. Each expression construct (e.g., a plasmid based on the pEGFP-N1 backbone and other fluorescent variants driven by the CMV promoter) is individually "pinned" to a glass surface to produce an expression array of library molecules. Mammalian cells, when seeded on the spotted cDNA in the presence of a transfection reagent (e.g., a lipid reagent), become transfected, producing a living cell array with each individual cluster expressing a different member of the library (Figure 3). (Cells growing between the spotted cDNAs are not transfected and remain "black"). These expression arrays are then stimulated with purified fluorescently labeled query proteins. After washing to remove unbound ligand, positive interactions are scored based on fluorescence. Because each construct is "pinned" to a known position in the microarray, positive "hits" can be immediately associated with their interacting partners.

The cell microarray platform was validated using PD-1:PD-L1 interaction (Kd of approximately 5.5 μM). Live cell microarrays consisting of cells expressing GFP fusion proteins of PD-L1 or GFP alone in alternating rows were generated ( FIG. 4A ). As shown in the diagram ( FIG. 4 , left panel), the GFP control was expressed in the cytoplasm (green solid circles), while the GFP fused to PD-L1 (green ring) was localized to the membrane (the lower absolute GFP fluorescence observed for PD-L1-GFP was due to the higher expression level of the cytoplasmic GFP construct). By exciting these arrays with red Alexa 594-bound bivalent Ig-fusion (i.e., Fc-fusion) constructs of the PD-1 extracellular domain, those spots that specifically expressed the homologous PD-L1 ligand were correctly identified ( FIG. 4B / C). These experiments clearly demonstrate the cell microarray technology for identifying receptor:ligand complexes. Figure 5 shows the cell microarray technology for a variety of different protein interactions (PD-1:PD-L2, CTLA-4:B7-1, and CD200R:CD200); these results further validate the broad applicability of the cell microarray platform and highlight the signal-to-noise ratio and specificity of this approach.

Similarly, 14 members of the fibronectin/fibronectin-like family, which belongs to the Ig superfamily (IgSF), were studied. At least 10 of these proteins exhibit homophilic interactions, and at least 20 heterophilic interactions exist between fibronectin family members [70-72]. In addition, approximately 500 extracellular domains and secreted proteins comprising the entire human IgSF were passed through this system. In these experiments, expression vectors for each member of the IgSF were spotted to generate microarrays probed with Ig fusion constructs of specific IgSF members (i.e., each spot represents a member of the IgSF). Because most IgSF members bind to other members of the IgSF, this provides an exciting opportunity to define new receptor:ligand interactions in the IgSF. Given their considerable mechanistic and therapeutic importance in cancer biology and autoimmune diseases, it is also necessary to identify ligands for the IgSF members B7-H4[73-80], VISTA[81], B7-H3[74,79,82-85], LAG-3[86-90], and 10 members of the butyrophilin family[91-94]. Other members of the IgSF family, including extended B7, carcinoembryonic antigen-related cell adhesion molecule (CEACAM)[95], and other members of the leukocyte receptor complex[96] family, are candidate targets. Expression reagents for a large number of IgSFs have been successfully generated and validated (Figure 6).

All members of the TNF and TNFR superfamily can be part of the platform; all of these proteins are important mechanistic and therapeutic targets and are type II membrane proteins (i.e., members of the TNF superfamily). The technology can be adapted to the entire secretome, including GPCRs, Toll-like receptors, growth factor receptors, interleukins, interleukin receptors, ion channels, etc.

Cloning: Access to large numbers of desired cloning templates is available. For example, the NYSGRC hosts the entire Mammalian Human Genome Collection (MGC) cDNA set from OpenBiosystems and these cloning templates are freely available. In a preferred embodiment, highly efficient ligation-independent cloning (LIC) [97] is used to generate expression libraries; the target insert gene will be followed by a transmembrane anchor and will be covalently fused to a cytoplasmically localized GFP expression reporter gene at its C-terminus (type I membrane protein) (Figures 4 and 6). Recently, implementation of LIC cloning has supported the generation of >15,000 sequence-verified constructs for large-scale structural (nysgrc.org) and functional genomic (enzymefunction.org) programs over the past 18 months. For the high-throughput molecular biology required to rapidly generate such libraries, the method is facilitated by automated liquid handling robots (e.g., the Biomek FxP liquid handler and the Perkin-Elmer EP3 robot). All required expression vectors are readily generated (Figure 6).

Quality generation of Ig fusion constructs of query proteins: High-throughput transient transfection and lentiviral driven platforms have been established to generate secretory proteins and in particular Ig fusion proteins. Figure 12 shows a number of mutant PD-L1 Ig fusion constructs that have been recently generated. This platform is based on the Daedelus system [98] and has the capacity to generate 48 lentiviruses per week. In addition, numerous secretory proteins and ectodomains have been efficiently generated as soluble Ig fusion proteins for mechanistic studies [21-23] and therapeutic applications [24-27]. This vast literature demonstrates that covalent fusion with non-native domains has no deleterious effects and is compatible with a variety of secretory proteins and ectodomains.

Expression of functional plasma membrane-localized GFP fusion proteins in cell microarrays: In the context of cell microarray screening, native integral membrane proteins will utilize native transmembrane elements to avoid the problem of distinguishing between type I and type II integral membrane proteins. Importantly, many examples of biologically relevant fluorescent protein fusion proteins (e.g., GFP fusion proteins) have been reported, including members of the Ig, TNF/TNFR[1-3], GPCR[4-6], integrin[7], and transporter[8] superfamilies. For cell microarrays, secreted proteins can be effectively engineered into integral membrane proteins by adding transmembrane helices that anchor them to the cell surface for subsequent detection. Tethering is not an issue given the existence of many proteins from many families that have biologically important membrane-anchored and secreted forms due to alternative splicing and/or shedding (i.e., proteolysis)[9-19]. In addition, a variety of secreted proteins (e.g., IL-2 and GM-CSF) have been intentionally engineered into single-spanning integral membrane proteins to provide novel therapeutic strategies (e.g., vaccine design)[20].

Some receptors require multiple components in order to exhibit binding activity to their cognate ligands (e.g., T cell receptors, integrins). As appropriate, these more complex receptors were addressed by co-expressing multiple components at a single location on the cell microarray (Figure 7).

Selection of cell lines for microarray presentation: Cell microarray technology has been firmly established using HEK293 cells. To distinguish those query proteins that bind to cell surface proteins endogenously expressed by HEK293 cells, binding to untransfected control cells (i.e., those cells that have not received an expression vector encoding a plasma membrane-localized protein) present in all microarrays is used as a convenient control. However, in most cases, saturating levels of overexpression driven by the strong CMV promoter will outperform low levels of endogenous cell surface expression. Moreover, appropriate statistical methods can statistically identify signal binding events. To aid these statistical analyses, all expression vectors can be spotted in duplicate in the cell microarray. Importantly, various alternative cell lines can also be used as "rescue host lines" in microarrays. For example, Drosophila S2 cells have been used by Sabatini for genome-wide loss-of-function studies in microarray format [99,100].

Avidity and dynamic range: Bivalent Ig-fusions have been effectively used to identify interactions with moderate affinities (i.e., PD-1:PD-L1; Kd = 5.5 μM). Activation of PD-L1-expressing cells with higher titer B7-1-modified microbeads allowed robust recruitment and specific identification of receptor:ligand binding in flow cytometry-based experiments ( FIG9 ). This comparison supports the view that increasing the avidity of the ligand expands the dynamic range of potential receptor:ligand interactions that can be measured experimentally, improving the ability to detect binding to dot-blot cell microarrays. For lower affinities, it is useful to probe the microarray with, for example, high-affinity multivalent microbeads modified with Ig-fusion proteins.

Higher affinity using transiently transfected cells: The experiments described above involve probing live cell arrays with purified query ligands, ultimately moving the focus of the experiments towards query protein generation and labeling. To increase the simplicity, practicality, and throughput of the described platform, a suspension-adapted mammalian cell line with reduced adhesion (i.e., HEK293 Freestyle (Invitrogen) [101]) expressing the query protein immobilized on its surface by a single transmembrane helix fused to a cytoplasmic C-terminal mCherry reporter protein (or other suitable fluorescent protein) can be used. The mCherry (red) suspension cells can then be used to excite the immobilized green "receptor" cells on the array. Co-localized cell spots containing GFP (e.g., microarray-localized receptor) and mCherry (e.g., suspension query ligand) will generate a positive score. Expressing the query ligand in the cellular environment removes the burden of query ligand purification and labeling; it has the added advantage of maintaining the query protein ligand in an environment closer to its native state, which is critical for proteins such as GPCRs. To address nonspecific binding and background, the following is proposed. Each “spot” in a cell microarray represents a cluster of cells overexpressing a defined gene product within a monolayer of untransfected cells. The observed background is due to nonspecific interactions with the monolayer of untransfected cells on the microarray, combined with the general inability to wash the microarray vigorously before fluorescence detection. These problems can be alleviated by enhancing the adhesion of the monolayer to withstand rigorous washing or by spatially restricting the deposition of cells to well-defined, well-defined areas. To improve the localized adhesion of spotted cells in the microarray environment, a HEK293 cell line stably expressing a functional cell surface-intrinsic single-chain avidin (scAvidin) was successfully engineered by using a non-classical secretion system to direct and anchor scAvidin to the outer leaflet of the plasma membrane (data not shown) [102]. This stable cell line specifically binds to non-cell permeable Alexa 594-labeled biotin and this strategy can be used to globally anchor cells to the array, or to specifically tether cells to defined areas via site-directed dotting of biotin conjugates if more rigorous washing steps are required. Both approaches will reduce background signal.

Statistical analysis of cell microarrays: Although very obvious interactions are easily discernible by eye (eg, Figures 4 and 6), it is preferable to apply appropriate statistical criteria to identify weaker interactions that are statistically significant.

An automated flow cytometric technique for high-throughput characterization of cell surface protein-protein interactions. Also disclosed is an effective alternative to using flow cytometry to determine specific receptor:ligand interactions. This platform allows for easy examination of affinity probes with a variety of avidities (i.e., bivalent Ig-fusions, high-affinity microbeads, and very high-affinity transiently transfected cells). The use of Ig-fusion proteins is conceptually similar to the experiments described above. The utility of microbeads and transiently transfected cells for discovering new receptor:ligand interactions is described more fully below.

The microbead-based approach was demonstrated using the same PD-1:PD-L1 interaction described above and expanded by including the PD-L1:B7-1 interaction. Figure 8 shows the overall strategy for the microbead experiment. Figure 9 shows a highly specific interaction between microbeads loaded with PD-1 or B7-1Ig-fusion proteins and HEK293 cells expressing plasma membrane-localized PD-L1:mCherry fusion proteins. These experiments show that multiple proteins (e.g., PD-1 and B7-1) are affected by microbead presentation and exhibit a predictable high signal-to-noise ratio (i.e., very low background binding was detected using cells expressing only plasma membrane GFP). These proof-of-principle experiments highlight the practicality and utility of flow cytometric analysis based on microbead presentation plus defined receptor:ligand interactions. The cell-cell flow cytometric approach is further illustrated using the same PD-L1 interaction described for the cell-microbead approach. While microbead presentation offers higher avidity relative to Ig-fusion proteins, expression of the query protein on the eukaryotic plasma membrane provides higher receptor density, significantly higher avidity, and an expanded dynamic range for detecting weaker receptor:ligand interactions.

The following were expressed individually in suspension-adapted HEK293 cells: 1) full-length PD-L1 as an mCherry fusion protein, 2) full-length PD-1 as a GFP fusion protein, 3) cytoplasmic mCherry, and 4) cytoplasmic GFP. Flow cytometric analysis of individual and mixed populations clearly demonstrated a significant increase (approximately 60-fold) in the signal representing specific cell-cell interactions only in the presence of both PD-1-expressing cells and PD-L1-expressing cells (Figure 10). Because the additional negative control PD-1 was also expressed as an mCherry fusion protein and, as expected, PD-1 itself did not interact (data not shown). Like the microbead analysis, the cell-cell approach also demonstrated an interaction between PD-L1-mCherry and B7-1-GFP from transiently transfected HEK293 cells (data not shown). This approach appears to have general utility because it also clearly revealed the expected interaction between CD200 and the CD200 receptor (CD200R) (Figure 11), which is independent of the PD-1, PD-L1, and B7-1 protein families.

These flow cytometric methods can be adapted to other known T cell co-stimulatory receptor:ligand pairs, including homophilic and heterophilic interactions within the aforementioned cohesin family (Figure 5). Both microbeads and transiently transfected HEK293 cells can be used to challenge methods using known immune receptors to probe binding across the entire Ig superfamily, the TNF/TNFR superfamily, and ultimately extend these experiments to probe the entire secretome.

Biochemical Functional Profiling: Bead-cell and cell-cell interactions can be used to dissect complex biochemical functions by screening a large number of mutant molecules. These capabilities have been demonstrated by generating PD-L1 point mutants that exhibit different affinities for PD-1 and B7-1, and more importantly, PD-L1 point mutants that bind specifically to PD-1 or B7-1. These studies used a first generation HEK293 cell line that was transiently transfected with a large number (i.e., >100) of PD-L1 mutant-mCherry fusions alone. The ability of these cells to bind to GFP-loaded microbeads modified with wild-type PD-1Ig-fusions or wild-type B7-1Ig-fusion proteins or to HEK293 cells transiently transfected with plasma membrane-localized wild-type PD-1-GFP or B7-1-GFP fusions was probed by flow cytometry. Of particular note, several mutants that were observed to be unbound in the microbead array showed significant binding (e.g., K124A and K125A) in the cell-cell format (Figure 12). This difference can be directly attributed to the increased potency/avidity associated with cell surface expression and emphasizes the value of multiple platforms with a range of affinities. These studies generated a generation of mutant PD-L1 Ig-fusion proteins that specifically bind to PD-1 (G119A, G120A) or B7-1 (D122A, Y123A) (Figure 12) and provided a mapping of the different but overlapping PD-L1 surfaces responsible for PD-1 and B7-1 recognition (data not shown); these unique reagents allow for the distinct contributions of PD-L1:PD-1 and PD-L1:B7-1 interactions to the definition of mammalian immunity. These results highlight the selectivity, utility, and complementarity of microbead-cell and cell-cell interaction platforms. In addition, with respect to binding affinity, ongoing quantitative Kd determination (e.g., surface plasmon resonance) of the interactions of these PD-L1 mutants will help determine the sensitivity of these microbead-cell and cell-cell platforms, as well as the cell microarray platform.

Improve throughput. The above studies were conducted using the BD FacsAria III, which supports a moderate throughput of approximately 1 sample/minute in a disposable format, requiring continued user attention. These methods can be used in other systems, such as 96/384-well plate formats to support high-throughput screening using, for example, the Intellicyt™ HTFC system. Intellicyt™ supports a throughput of 3 minutes per 96-well plate/12 minutes per 384-well plate in hands-free mode. This flow cytometry-based method, when performed using a multi-well cell counter and a fully automated tissue culture robot, provides the high throughput required for large-scale receptor de-orphanization experiments. Other examples of systems that can be used with the method include the Perkin Elmer Cell::Probe; a fully automated tissue culture-based liquid handler, a Janus workstation, a Liconic shaking incubator, and an Envision microplate reader—all accessible via a six-axis robotic arm housed in a BSL-2 biosafety hood, for example to ensure sterility. Fully implemented automated tissue culture capabilities, including cell growth, culture medium replacement, transfection, etc., contribute to efficiency. The platform in a multi-well format can optionally be based on proven interaction pairs (PD-1:PD-L1, PD-1:PD-L2, PD-L1:B7-1, CTLA-4:B7, CD200R:CD200; Figures 4, 6, 10, 11) as well as the entire panel of PD-L1 mutants (Figure 12). This can be generalized to all members of the Ig superfamily as described above and ultimately to the entire secretome. It is important to note that although many laboratories have demonstrated that cell-cell based interactions can be readily examined by FACS analysis (Figures 10 and 11), these efforts have all been low-throughput and not easily transferable to a complete screen that confers the advantages described herein.

Modification of magnetic capture technology and next-generation sequencing for highly multiplexed identification of cell surface protein-protein interactions: Another platform described herein uses magnetic capture technology to rapidly enrich cell-bead (or cell-cell) conjugates formed due to specific receptor:ligand interactions [103] and uses next-generation massively parallel sequencing (e.g., Illumina/454 [104-106]) to deconvolute the resulting pool (e.g., [107-110]). This platform leverages the tagging of expression vectors for each member of an expression library with a unique nucleotide barcode (in the example described, 28 nucleotides, but other ranges can be used) that can be a barcode amplified with "universal primers" and easily identified by deep sequencing (Figure 13) [107-110]. Libraries of barcoded vectors can be pooled and transfected together into suspension-adapted HEK293 cells. The expression library that converges is mixed with query protein (when microbeads or cell surface presentation) to form conjugate, and conjugate is reclaimed by porous magnetic separation (for example, in less than 30 minutes, carry out 24 parallel separations).Although there is query protein in the converging library, magnetic query protein (when microbeads or cell surface presentation) is greatly excessive, thereby eliminating the competition from converging library components.Amplification comes from the barcode of enrichment set member and carries out next generation deep sequencing (for example, 75 nucleotides each up to 10,000,000 readings) to identify the barcode enriched by capture process.These enrichment barcodes directly identify the potential receptor for subsequent verification by in vitro biochemical method (SPR, ITC, SEC, FACS): ligand interaction.Described strategy allows to rapidly identify the binding partner of single query protein, but can be easy to multiplex to greatly increase throughput and reduce cost. For example, using tissue culture automation, each of the 500 IgSF members can be used individually as a query protein, and the capture conjugates from each query protein are collected in separate wells of a multiwell plate. This physical separation of candidate interactors for each query protein allows the use of "composite" primers in the amplification step (i.e., each well receives a unique primer set in which two "universal T7 priming sequences" are flanked by additional 8 unique, highly specific nucleotides). The use of composite primers allows the specific identification of those library members (28-nucleotide core barcode) that are enriched due to interaction with the query protein (additional, highly specific nucleotide barcode, e.g., 8 nucleotides) corresponding to a particular well (Figure 13). After amplifying each well individually with these composite primers, the amplicons are pooled and deconvoluted using a single deep sequencing run. This identifies the interacting members of the library based on the unique nucleotide barcodes assigned to each member of the IgSF. These interactors are identified as binding partners of the specific query protein by their unique (e.g., 8 nucleotide) barcodes specific to each well.

Magnetic capture/enrichment: Cell enrichment using the Miltenyi system is straightforward in the context of cell-bead conjugates [103]. Cell enrichment is preferably performed using, but not limited to, 50 nm magnetic beads. Figure 14 illustrates the use of magnetic beads to separate/enrich specific cell-cell conjugates formed due to cognate receptor:ligand interactions. To generalize this approach for high-throughput capture of cell-cell conjugates by magnetic separation, the query protein is expressed in a cell line stably expressing a transmembrane anchor tag (e.g., FLAG) on its surface to allow capture by anti-FLAG-carrying magnetic beads. This strategy prevents bead-bound capture reagents (e.g., anti-FLAG) from interfering with the specific receptor:ligand complex. Furthermore, it is possible to examine a range of library pool:query cell stoichiometries and vary the amount of magnetic beads utilized, as these variables will affect the yield and stringency of the selection. Importantly, magnetic bead capture allows for extensive washing steps and results in reduced background. The extremely high strength of multivalent interactions associated with cell-cell conjugate formation is fully compatible with the relatively mild magnetic separation technique. This approach does not rely on the measurement of a fluorescent signal to determine "bound" and "unbound" events and removes the complications of gating, laser settings, etc.

Signal-to-noise ratio: Nonspecific binding can occur between query protein-expressing cell lines and "non-targets" (i.e., cells that do not express the cognate ligand). Figure 14B demonstrates the ability of the magnetic capture technology to specifically enrich for "rare events" (i.e., 1.5% of total possible interactions). Figure 15 provides evidence that the barcode method can detect even rarer events. For a typical binary cell-cell assay, in which the population consists of a 1:1 mixture of cells expressing a cognate receptor: ligand pair (e.g., PD-1:PD-L1 and CD200:CD200R), 10-30% of events are typically scored as positive binding interactions (see Figures 10 and 11). For negative controls (e.g., GFP and mCherry), typically about 0.2-0.5% of events are scored as binding, where "binding" is defined as the number of events that are GFP and mCherry positive (i.e., Figure 15A, quadrant 2).

To specifically assess the challenges associated with identifying cognate interactions in the context of expression libraries, background was simulated by mixing 107 HEK293 cells transiently expressing GFP with 0.02 × 106 cells expressing PD-1 GFP-fusion protein (0.2% GFP-positive cells, which would represent a single member of the IgSF if all were transfected with equal efficiency). The library was stimulated with 106 mCherry (negative control) or HEK293 cells transiently expressing PD-L1 mCherry-fusion. Figure 15 shows that the GFP:mCherry conjugate is clearly enriched (quadrant 2) due to the specific PD-1:PD-L1 interaction. Importantly, PCR-based validation of the enrichment of PD-1 expressing cells was completely similar to the barcoding strategy performed to deconvolute the confluent amplicons (i.e., the PD-1 coding sequence acts as an intrinsic barcode). Importantly, the enrichment levels achieved in Figures 14 and 15 are well within the detection limit of the next generation deep sequencing method employed [113].

PD-L1:PD-1 and PD-L1:B7-1 interactions can be examined using the unique barcoding approach described in Figure 13, as can 500 genes belonging to the human IgSF that are incorporated into each of two expression vectors (i.e., 500 receptors, 500 ligands) and subjected to interaction screening. This system enables simultaneous interrogation of many ligands for the entire set of potential receptors, allowing for simultaneous, efficient, and cost-effective interrogation of large query lists. This approach can cover the entire secretome.

Additional aspects of the invention and their verification are illustrated in Figures 16-24.

References

1.Cheung, T.C., et al., T cell intrinsic heterodimeric complexes between HVEM and BTLA determine receptivity to the surrounding microenvironment. JImmunol, 2009.183(11): p.7286-96.

2.Cheung, T.C., et al., Unconventional ligand activation of herpesvirusentry mediator signals cell survival. Proc Natl Acad Sci U S A, 2009.106(15): p.6244-9.

3. Olszewski, M.B., et al., TNF trafficking to human mast cell granules: mature chain-dependent endocytosis. J Immunol, 2007.178(9): p.5701-9.

4. Giebing, G., et al., Arrestin-independent internalization and recycling of the urotensin receptor contribute to long-lasting urotensin II-mediated vasoconstriction. Circ Res, 2005.97(7): p.707-15.

5. Tarasova, N.l., et al., Visualization of G protein-coupled receptor trafficking with the aid of the green fluorescent protein. Endocytosis and recycling of cholecystokinin receptor type A.J Biol Chem, 1997.272(23): p.14817-24.

6.Bohme, I.and A.G.Beck-Sickinger, Illuminating the life of GPCRs. CellCommun Signal, 2009.7: p.16.

7.Carman, C.V. and T.A. Springer, Integrin avidity regulation: are changes in affinity and conformation underemphasized? Curr Opin Cell Biol, 2003.15(5): p.547-56.

8. Chiu, C.S., et al., Number, density, and surfatce/cytoplasmic distribution of GABA transportets at presynaptic structures of knock-in micecarrying GABA transporter subtype I-green fluorescent protein fusions. JNeurosci, 2002.22(23): p.10251-66.

9. Geng, H., et al., Soluble form of T cell Ig mucin 3is an inhibitorymolecule in T cell-mediated immune response. J Immunol, 2006.176(3): p.1411-20.

10. Sakimoto, T., A. Yamada, and M. Sawa, Release of soluble tumor necrosisfactor receptor I from corneal epithelium by TNF-alpha-converting enzyme-dependent ectodomain shedding. Invest Ophthalmol Vis Sci, 2009.50(10): p.4618-21.

11. Sanderson, M.P., et al., Generation of novel, secreted epidermalgrowth factor receptor (EGFR/ErbB1) isoforms via metalloprotease-dependentectodomain shedding and exosome secretion. J Cell Biochem, 2008.103(6): p.1783-97.

12. Chalaris, A., et al., The soluble Interleukin 6 receptor: generation and role in inflammation and cancer. Eur J Cell Biol, 2011.90(6-7): p.484-94.

13. Shibata, M.A., el al., The endogenous soluble VEGF receptor-2 isoformsuppresses lymph node metastasis in a mouse immunocompetent mammary cancer model. BMC Med, 2010.8: p.69.

14. Pavlakovic, H., et al., Soluble VEGFR-2: an antilymphangiogenic variant of VEGF receptors. Ann N Y Acad Sci, 2010.1207 Suppl 1: p.E7-15.

15. Khankin, E.V., et al., Soluble erythropoietin receptor contributes toerythropoietin resistance in end-stage renal disease. PLoS One, 2010.5(2): p.e9246.

16.Jones, D.C., et al., Alternative mRNA splicing creates transcriptsencoding soluble proteins from most LILR genes. Eur J Immunol, 2009.39(11): p.3195-206.

17.Chen, Z., et al., Identification of an expressed truncated form of CD200, CD200tr, which is a physiologic antagonist of CD200-inducedsuppression. Transplantation, 2008.86(8): p.1116-24.

18. Eshel, D., et al., Characterization of natural human antagonistic soluble CD40isoforms produced through alternative splicing. Mol Immunol, 2008.46(2): p.250-7.

19. Levine, S.J., Molecular mechanisms of soluble cytokine receptor generation. J Biol Chem, 2008.283(21): p.14177-81.

20. Yang, Y., et al., A novel method to incorporate bioactive cytokines as adjuvants on the surface of virus particles. J Interferon Cytokine Res, 2009.29(1): p.9-22.

21.Cao, E., et al., T cell immunoglobulin mucin-3 crystal structure reveals a galectin-9-independent ligand-binding surface. Immunity, 2007.26(3): p.311-21.

22.Bitonti, A.J., et al., Pulmonary delivery of an erythropoietin Fcfusion protein in non-human primates through an immunoglobulin transportpathway. Proc Natl Acad Sci U S A, 2004.101(26): p.9763-8.

23. Ozkaynak, E., et al., Programmed death-l targeting can promote allograft survival. J Immunol, 2002.169(11): p.6546-53.

24. Yang, T., et al., A variant of TNFR2-Fc fusion protein exhibits improved efficacy in treating experimental rheumatoid arthritis. PLoS ComputBiol, 2010.6(2): p.e1000669.

25. Moreland, L., G. Bate, and P. Kirkpatrick, Abatacept. Nat Rev Drug Discov, 2006.5(3): p.185-6.

26. Vincenti, F., A. Dritselis, and P. Kirkpatrick, Belatacept. Nat Rev Drug Discov, 2011.10(9): p.655-6.

27. Moreland, L.W., Soluble tumor necrosis factor receptor (p75) fusion protein (ENBREL) as a therapy for rheumatoid arthritis. Rheum Dis Clin North Am, 1998.24(3): p.579-91.

28. Zhang, X., et al., Structural and functional analysis of the costimulatory receptor programmed death-l. Immunity, 2004.20(3): p.337-47.

29.Bhatia, S., et al., Different cell surface oligomeric states of B7-1and B7-2: implications for signaling. Proc Natl Acad Sci U S A, 2005.102(43): p.15569-74.

30.Cao, E., et al., NTB-A receptor crystal structure: insights into homophilic interactions in the signaling lymphocytic activation moleculereceptor family. Immunity, 2006.25(4): p.559-70.

31. Chattopadhyay, K., et al., Structural basis of inducible costimulatorligand costimulatory function: determination of the cell surface oligomericstate and functional mapping of the receptor binding site of the protein. JImmunol, 2006.177(6): p.3920-9.

32. Chattopadhyay, K., et al., Assembly and structural properties of glucocorticoid-induced TNF receptor ligand: Implications for function. ProcNatl Acad Sci U S A, 2007.104(49): p.19452-7.

33.Yan, Q., et al., Structure of CD84 provides insight into SLAM family function. Proc Natl Acad Sci U S A, 2007.104(25): p.10583-8.

34. Chattopadhyay, K., et al., Evolution of GITRL immune function: murineGITRL exhibits unique structural and biochemical properties within the TNFsuperfamily. Proc Natl Acad Sci U S A, 2008.105(2): p.635-40.

35. Lazar-Molnar, E., et al., Crystal structure of the complex between programmed death-I(PD-I) and its ligand PD-L2. Proc Natl Acad Sci U S A, 2008.105(30): p.10483-8.

36. Zhan, C., et al., Biochemical and structural characterization of the human TL lA ectodomain. Biochemistry, 2009.48(32): p.7636-45.

37.Bhatia, S., et al., Dynamic equilibrium of B7-1 dimers and monomers differentially affects immunological synapse formation and T cell activationin response to TCR/CD28 stimulation. J Immunol, 2010.184(4): p.1821-8.

38.Zhan, C., et al., Decoy strategies: the structure of TLIA: DcR3complex.Structure, 2011.19(2): p.162-71.

39.Samanta, D., et al., Structure of Nectin-2 reveals determinants ofhomophilic and heterophilic interactions that control cell-cell adhesion. ProcNatl Acad Sci U S A, 2012.109(37): p.14836-40.

40.Eads, J.C., et al., Structure determination and characterization of Saccharomycescerevisiae profilin. Biochemistry, 1998.37(32): p.11171-81.

41.Bout-Jordan, H., et al., Intrinsic and extrinsic control of peripheral T-cell tolerance by costimulatory molecules of the CD28/B7family. Immunol Rev, 2011.241(1): p.180-205.

42. Nurieva, R.I., X. Liu, and C. Dong, Yin-Yang of costimulation: critical controls of immune tolerance and function, Immunol Rev, 2009.229(1): p.88-100.

43. Wang, S. and L. Chen, T lymphocyte co-signaling pathways of the B7-CD28 family. Cell Mol Immunol, 2004.1(1): p.37-42.

44.Zang,

45. Chattopadhyay, K., et al., Sequence, structure, function, immunity: structural genomics of costimulation. Immunol Rev, 2009.229(1): p.356-86.

46. Ostrov, D.A., et al., Structure of murine CTLA-4 and its role inmodulating T cell responsiveness. Science, 2000.290(5492): p.816-9.

47. Schwartz, J.C., et al., Structural basis for co-stimulation by the human CTLA-4/B7-2complex. Nature, 2001.410(6828): p.604-8.

48. Zhang, X., et al., Crystal structure of the receptor-binding domain of human B7-2: insights into organization and signaling. Proc Natl Acad Sci U SA, 2003.100(5): p.2586-91.

49.Butte, M.J., et al., Programmed death-I ligand I interacts specifically with the B7-1 costimulatory molecule to inhibit T cellresponses. Immunity, 2007.27(1): p.111-22.

50.Yao, S., et al., B7-h2 is a costimulatory ligand for CD28 inhuman. Immunity, 2011.34(5): p.729-40.

51. Mansh, M., Ipilimumab and cancer immunotherapy: a new hope for advanced stage melanoma. Yale J Biol Med, 2011.84(4): p.381-9.

52. Bluestone, J.A., E.W. St Clair, and L.A. Turka, CTLA4lg: bridging the basic immunology with clinical application. Immunity, 2006.24(3): p.233-8.

53.Larsen, C.P., et al., Rational development of LEA29Y(belatacept), ahigh-affinity variant of CTLA4-Ig with potent immunosuppressive properties. AmJ Transplant, 2005.5(3): p.443-53.

54. Lum, G. and X.J.Min, FunSecKB: the Fungal SecretomeKnowledgeBase.Database (Oxford), 2011.2011: p.bar001.

55. Mishra, S. K., H. R. Siddique, and M. Saleem, S100A4 calcium-binding protein is key player in tumor progression and metastasis: preclinical and clinical evidence. Cancer Metastasis Rev, 2011.

56.Nickel, W., The unconventional secretory machinery of fibroblastgrowth factor 2.Traffic, 2011.12(7): p.799-805.

57. Kadono, N., et al., The impact of extracellular syntaxin4 on HaCaTkeratinocyte behavior. Biochem Biophys Res Commun, 2012.417(4): p.1200-5.

58. Rawat, P., et al., The multifunctional glycolytic proteinglyceraldehyde-3-phosphate dehydrogenase (GAPDH) is a novel macrophagelactoferrin receptor. Biochem Cell Biol, 2012.90(3): p.329-38.

59. Kumar, S., et al., Characterization of glyceraldehyde-3-phosphatedehydrogenase as a novel transferrin receptor. Int J Biochem Cell Biol, 2012.44(1): p.189-99.

60. Nickel, W., Pathways of unconventional protein secretion. Curr OpinBiotechnol, 2010.21(5): p.621-6.

61. Prudovsky, I., et al., Secretion without Golgi.J Cell Biochem, 2008.103(5): p.1327-43.

62. Bushell, K.M., et al., Large-scale screening for novel low-affinity extracellular protein interactions. Genome Res, 2008.18(4): p.622-30.

63. Jiang, L. and A.N. Barclay, identification of leucocyte surfaceprotein interactions by high-throughput screening with multivalent reagents. Immunology, 2010.129(1): p.55-61.

64.Gonzalez, L.C., et al., A coreceptor interaction between the CD28 andTNF receptor family members B and T lymphocyte attenuator and herpesvirusentry mediator. Proc Natl Acad Sci U S A, 2005.102(4): p.1116-21.

65.Yu,

66. Ramani, S.R., et al., A secreted protein microarray platform for extracellular protein interaction discovery. Anal Biochem, 2012.420(2): p.127-38.

67. Lin, H., et al., Discovery of a cytokine and its receptor by functional screening of the extracellular proteome. Science, 2008.320(5877): p.807-11.

68.Palmer, E., Cell-based microarrays: overview. Methods Mol Biol, 2011.706: p.1-12.

69. Ziauddin, J. and D. M. Sabatini, Microarrays of cells expressing defined cDNAs. Nature, 2001.411(6833): p.107-10.

70. Takai, Y., et al., Nectins and nectin-like molecules: roles in contactinhibition of cell movement and proliferation. Nat Rev Mol Cell Biol, 2008.9(8): p.603-15.

71. Takai, Y., et al., The immunoglobulin-like cell adhesion moleculenectin and its associated protein afadin. Annu Rev Cell Dev Biol, 2008.24: p.309-42.

72.Chan, C.J., D.M.Andrews, and M.J.Smyth, Receptors that interact with nectin and nectin-like proteins in the immunosurveillance and immunotherapy of cancer. Curr Opin Immunol, 2012.24(2): p.246-51.

73.Wang, X., et al., Blockade of both B7-H4 and CTLA-4 co-signalingpathways enhances mouse islet allograft survival, Islets, 2012.4(4).

74.Yi, K.H.and L.Chen, Fine tuning the immune response through B7-H3and B7-H4. Immunol Rev, 2009.229(1): p.145-51.

75. Mirza, N. and D. Gabrilovich, Comment on "Cutting edge: induction of B7-H4 on APCs through IL-10: novel suppressive mode for regulatory T cells". JImmunol, 2007.178(8): p.4705-6; author reply 4706.

76.Krambeck, A.E., et al., B7-H4 expression in renal cell carcinoma and tumor vasculature: associations with cancer progression and survival. Proc NatlAcad Sci U S A, 2006.103(27): p.10391-6.

77.Sica, G.L., et al., B7-H4, a molecule of the B7 family, negatively regulates T cell immunity. Immunity, 2003.18(6): p.849-61.

78. Wei, J., et al., Tissue-specific expression of B7x protects from CD4T cell-mediated autoimmunity. J Exp Med, 2011.208(8): p.1683-94.

79. Zang,

80. Zang,

81.Wang, L., et al., VISTA, a novel mouse Ig superfamily ligand that negatively regulates T cell responses. J Exp Med, 2011.208(3): p.577-92.

82.Sun, Y., et al., B7-H3 and B7-H4 expression in non-small-cell lungcancer. Lung Cancer, 2006.53(2): p.143-51.

83.Fauci, J.M., et al., A review of B7-H3 and B7-H4 immune molecules and their role in ovarian cancer. Gynecol Oncol, 2012.127(2): p.420-5.

84.Guo, G., et al., The characteristic expression of B7-H3 and B7-H4 inliver biopsies from patients with HBV-related acute-on-chronic liverfailure. Pathol Int, 2012.62(10): p.665-74.

85.Guo, G., et al., The expression and distribution of immunomodulatoryproteins B7-H1, B7-DC, B7-H3, and B7-H4 in rheumatoid synovium. Clin Rheumatol, 2012.31(2): p.271-81.

86. Triebel, F., et al., LAG-3, a novel lymphocyte activation gene closely related to CD4. J Exp Med, 1990.171(5): p.1393-405.

87. Triebel, F., LAG-3: a regulator of T-cell and DC responses and its use in therapeutic vaccination. Trends Immunol, 2003.24(12): p.619-22.

88.Chun, T., et al., The effect of soluble LAG-3(CD223)treatment infetal thymic organ culture. Biotechnol Lett, 2004.26(17): p.1371-7.

89. Okazaki, T., et al., PD-1and LAG-3 inhibitory co-receptors actynergistically to prevent autoimmunity in mice. J Exp Med, 2011.208(2): p.395-407.

90. Sierro, S., P. Romero, and D.E. Speiser, The CD4-like molecule LAG-3, biology and therapeutic applications. Expert Opin Ther Targets, 2011.15(1): p.91-101.

91.Abeler-Dorner, L., et al., Butyrophilins: an emerging family of immuneregulators. Trends Immunol, 2012.33(1): p.34-41.

92. Cubillos-Ruiz, J.R. and J.R. Conejo-Garcia, It never rains but itpours: potential role of butyrophilins in inhibiting anti-tumor immune responses. Cell Cycle, 2011.10(3): p.368-9.

93.Arnett, H.A., S.S.Escobar, and J.L.Viney, Regulation of costimulation in the era of butyrophilins. Cytokine, 2009.46(3): p.370-5.

94. Elleder, D., et al., The receptor for the subgroup C avian sarcoma and leukosis viruses, Tvc, is related to mammalian butyrophilins, members of the immunoglobulin superfamily. J Virol, 2005.79(16): p.10408-19.

95. Kuespert, K., S. Pils, and C. R. Hauck, CEACAMs: their role in physiology and pathophysiology. Curr Opin Cell Biol, 2006.18(5): p.565-71.

96. Barrow, A.D. and J. Trowsdale, The extended human leukocyte receptor complex: diverse ways of modulating immune responses. Immunol Rev, 2008.224: p.98-123.

97. Aslanidis, C. and P.J. de Jong, Ligation-independent cloning of PCRproducts (LIC-PCR). Nucleic Acids Res, 1990.18(20): p.6069-74.

98. Bandaranayake, A.D., et al., Daedalus: a robust, turnkey platform for rapid production of decigram quantities of active recombinant proteins in human cell lines using novel lentiviral vectors. Nucleic Acids Res, 2011.39(21): p.e143.

99. Lindquist, R.A., et al., Genome-scale RNAi on living-cell microarrays identifies novel regulators of Drosophila melanogaster TORCI-S6K pathwaysignaling. Genome Res, 2011.21(3): p.433-46.

100.Wheeler, D.B., et al., RNAi living-cell microarrays for loss-of-function screens in Drosophila melanogaster cells. Nat Methods, 2004.1(2): p.127-32.

101.Graham, F.L., Growth of 293 cells in suspension culture.J GenVirol, 1987.68(Pt 3): p.937-40.

102.Stegmayer, C., et al., Direct transport across the plasma membrane of mammalian cells of Leishmania HASPB as revealed by a CHO export mutant. JCell Sci, 2005.118(Pt 3): p.517-27.

103. Grutzkau, A. and A. Radbruch, Small but mighty: how the MACS-technology based on nanosized superparamagnetic particles has helped toanalyze the immune system within the last 20 years. Cytometry A, 2010.77(7): p.643-7.

104.Thudi, M., et al., Current state-of-art of sequencing technologies for plant genomics research. Brief Funct Genomics, 2012.11(1): p.3-11.

105.Koboldt, D.C., et al., Massively parallel sequencing approaches for characterization of structural variation. Methods Mol Biol, 2012.838: p.369-84.

106. Morozova, O. and M.A. Marra, Applications of next-generationsequencing technologies in functional genomics. Genomics, 2008.92(5): p.255-64.

107.Whitehead, T.A., et al., Optimization of affinity, specificity and function of designed influenza inhibitors using deep sequencing. NatBiotechnol, 2012.30(6): p.543-8.

108. Smith, A.M., et al., Competitive genomic screens of barcoded yeastlibraries. J Vis Exp, 2011(54).

109.Lee, W., et al., Genome-wide requirements for resistance to functionally distinct DNA-damaging agents. PLoS Genet, 2005.1(2): p.e24.

110.Pierce, S.E., et al., Genome-wide analysis of barcoded Saccharomycescerevisiae gene-deletion mutants in pooled cultures. Nat Protoc, 2007.2(11): p.2958-74.

111.McLellan, A.S., et al., The Wasp System: An open source environment for managing and analyzing genomic data. Genomics, 2012.

112.Golden, A., et al., The Einstein Genome Gateway using WASP-a highthroughput multi-layered life sciences portal for XSEDE.Stud Health TechnolInform, 2012.175: p.182-91.

113. Jiang, L., et al., Synthetic spike-in standards for RNA-seqexperiments. Genome Res, 2011.21(9): p.1543-51.

Claims (17)

1. A system comprising: (i) A first plurality of suspension-adapted mammalian or insect cells, wherein the cells are transformed by a carrier so as to: (a) It expresses a first heterologous protein on its cell surface; and (b) expressing a first cytoplasmic fluorescent protein, such that the first plurality of suspension-adapted mammalian or insect cells express at least two different types of the first heterologous protein; and (ii) A second plurality of suspension-adapted mammalian or insect cells, said cells being transformed by a second carrier to express a second heterologous protein on their cell surface and to express a second cytoplasmic fluorescent protein, wherein said second plurality of suspension-adapted mammalian or insect cells express a single type of the second heterologous protein; in: i) Each vector expressed by the first plurality of suspension-adapted mammalian or insect cells contains a predetermined sequence of 15-35 nucleotides of the expressed heterologous protein; and ii) The colocalization of the first and second cytoplasmic fluorescent proteins was determined by fluorescence activated cell sorting (FACS) analysis. 2. The system of claim 1, wherein any single cell of the first plurality of cells expresses only one first heterologous protein on its cell surface. 3. The system according to claim 1 or 2, wherein the plurality of different types of first heterologous proteins are each mutant of a predetermined wild-type protein. 4. The system according to any one of claims 1-3, wherein the second heterologous protein is a wild-type protein. 5. The system according to any one of claims 1-4, wherein each type of heterologous protein in the first plurality of cells differs from each other type of heterologous protein in the plurality of cells in that there is a point mutation of 1, 2, 3, 4 or 5 amino acid residues. 6. The system according to any one of claims 1-5, wherein the first or second heterologous protein is an immunoglobulin superfamily protein, TNF receptor protein, cytokine, chemokine, type 1 transmembrane receptor protein, type 2 transmembrane receptor protein, ion channel protein, membrane transport protein, Toll-like receptor, G protein-coupled receptor, growth factor receptor, adhesion protein, interleukin, or interleukin receptor. 7. The system according to any one of claims 1-6, wherein the first or second heterologous protein is a PD-1 peptide, PD-L1 peptide, PD-L2 peptide, B7-1 peptide, B7-2 peptide, CD28 peptide, CTLA-4 peptide, ICOS peptide, or ICOS-L peptide. 8. The system according to any one of claims 1-7, wherein the single sequence is capable of being initiated by one or more universal primers. 9. The system according to any one of claims 1-8, wherein the first and second cytoplasmic fluorescent proteins are different colors. 10. The system according to any one of claims 1-9, wherein each plurality of cells is a plurality of mammalian cells. 11. The system according to any one of claims 1-10, wherein the second heteroprotein is a ligand of at least two different types of the first heteroprotein. 12. A method for determining whether a candidate ligand protein or peptide binds to a second protein or peptide, the method comprising: a) Expressing the candidate ligand protein or peptide as a first heterologous protein of the first plurality of cells in the system according to any one of claims 1-11; b) Expressing the second protein or peptide as the second heteroprotein in the system according to any one of claims 1-11, provided that the first heteroprotein is allowed to bind to the second heteroprotein; and c) Recover cells co-localized with the first and second heterologous proteins. 13. The method of claim 12, comprising a washing step between step b) and step c) to remove any unbound first heterologous protein. 14. The method of claim 13, further comprising obtaining nucleic acid from the recovered cells and sequencing the nucleic acid to identify a single sequence of 15-35 nucleotides contained therein in order to identify the candidate ligand protein or peptide corresponding to the single 15-35 nucleotides that has been bound to the second protein or peptide. 15. A method for determining the effect of predetermined amino acid residues of a first protein on the binding of the first protein to a second protein, the method comprising: a) Expressing proteins mutated relative to the first protein by one or more point mutations as the first two different types of first heterologous proteins in the first plurality of suspension-adaptive mammalian or insect cells of the system according to any one of claims 1-11; b) Contact the first plurality of suspension-adapted mammalian cells or insect cells with the second protein in the form of a second heterologous protein of the second plurality of suspension-adapted mammalian cells or insect cells transformed in any one of claims 1-11 to express the second protein and the second cytoplasmic fluorescent protein; c) Recover cells showing co-localization of the first and second cytoplasmic fluorescent proteins; d) Obtain nucleic acids from the recovered cells and sequence the nucleic acids to identify a single sequence of the 15-35 nucleotides contained therein in order to identify the first protein that has bound the second protein or peptide, and compare the level of the protein that has bound the second protein or peptide with a predetermined reference level. The presence of a protein that has already bound the second protein or peptide at a level exceeding the predetermined reference level indicates that one or more mutated residues in the protein enhance the binding of the first protein to the second protein, and the presence of a protein that has already bound the second protein or peptide at a level below the predetermined reference level indicates that one or more mutated residues in the protein inhibit the binding of the first protein to the second protein. 16. The method of claim 15, wherein the predetermined level is a control, optionally wherein the predetermined level is obtained by measuring the level of binding between the unmutated first protein and the second protein. 17. The method according to any one of claims 15-16, wherein cells showing co-localization of the first and second cytoplasmic fluorescent proteins are recovered by fluorescence activated cell sorting analysis.