Method for the generation of large, multispecific antibody libraries and applications for ultra-high-throughput functional screening

Abstract

Embodiments are described that relate to methods and compositions for generating large combinatorial libraries of multispecific antibodies, in aspects asymmetric multispecific antibodies, including nucleic acids, collections of nucleic acids, expression cassettes, vectors for delivery comprising the expression cassettes, cells comprising the vectors, libraries of cells, and methods for screening including methods involving autocrine screening for binding or function. The methods for screening large combinatorial libraries of asymmetric or symmetric multispecific antibodies are useful for antibody-based drug discovery.

Description

[0001] METHOD FOR THE GENERATION OF LARGE, MULTISPECIFIC ANTIBODY LIBRARIES AND APPLICATIONS FOR ULTRA-HIGH-THROUGHPUT FUNCTIONAL SCREENING

[0002] FIELD OF INVENTION

[0003] [1] The invention relates to a new method for generating large combinatorial libraries of asymmetric multispecific antibodies and screening them for antibody-based drug discovery.

[0004] BACKGROUND

[0005] [2] Existing antibody generation platforms can conduct drug discovery programs on repertoires containing thousands to millions of monoclonal antibodies. However, these approaches rely on established processes that are not well suited for accessing the biological and functional space available to multispecific antibodies. As such, these platforms have not proven equally successful at discovering and developing functional multispecific antibodies able to manipulate unique biological mechanisms and improve patient outcomes.

[0006] [3] Multispecific antibodies are increasingly of interest as therapeutics. While natural antibodies are monospecific and symmetric homodimers, multispecific antibodies recognize at least two different epitopes, on either the same or different antigens. Multispecific antibodies can also include antibodyfusion proteins, a class of antibody -like molecules in which an effector molecule, such as a cytokine or receptor ligand, is fused to antibody constant domains or replaces at least one of the antigenbinding paratopes. These abilities allow multispecific antibodies to uniquely access intricate biological mechanisms that can be harnessed to treat complex diseases. However, the generation of functional multispecific antibodies is non-trivial. There is no universal format that can be used to optimally access diverse biological mechanisms, meaning that many formats, which may vary in size, arrangement, valency, flexibility, and geometry, must be tested to find a combination of features that leads to the desired behavior and biological functionality. This challenge is further exacerbated by the need to screen large libraries of epitope / paratope space within any given format, as well as the need to address development challenges inherent to most multispecific antibody formats. Overall, these combinatorial factors greatly increase the format-epitope / paratope space that must be searched through, and as a result, a multispecific antibody that produces the desired functional effect and clinical outcome, may be an extremely rare occurrence.

[0007] [4] Current platforms for generating multispecific antibodies are limited as universal platforms, despite having generated clinical candidates and market-approved molecules. Mature multispecific platforms, using formats such as BiTE, XmAb, DART and CrossMAb, were developed primarily as single-format platforms to address specific biological mechanisms using, in part, known targets and epitope s / known paratopes. As such, the performance of these platforms has not always translated when applied to novel targets, epitopes, or biological mechanisms. Indeed, it is becoming apparent that even small differences in binding affinities or geometries can greatly impact the functional effects of multispecific antibodies. Furthermore, a reliance on a single format has hampered the broad applicability of these platforms. Many single-format legacy platforms tend to rely on symmetric formats that have a fixed valency, such as 1 : 1 or 2:2. Symmetric formats bring advantages for both antibody discovery and developability but sacrifice the ability to address targets using asymmetric geometries that can make it easier to access mixed binding valency, trispecific epitopes, or substitute a paratope with an effector molecule, any of which may be required to elicit a given biological function and therapeutic effect. Other platforms screen multiple paratopes that are covalently linked as part of the same polypeptide. The pursuit of many valuable therapeutic targets with functional antibodies will require a departure from single format-driven platforms due to the complexities of screening through differences in molecular geometry, epitope / paratope / affinity space, pharmacokinetics, and developability.

[0008] [5] New methods are needed to overcome the inability of current antibody discovery platforms to efficiently search through biological and functional space in order to identify functional-multispecific antibodies with the greatest potential to achieve therapeutic outcome.

[0009] BRIEF DESCRIPTION OF DRAWINGS

[0010] [6] Figure 1 depicts a non-limiting example of different methods for assembling asymmetric multispecific antibody libraries that may be used to identify functional antibodies.

[0011] [7] Figure 2 depicts non-limiting examples of symmetric and asymmetric multispecific antibodies (dashed box) containing one, two or three paratopes. In some embodiments, multispecific antibodies can contain effector molecules in addition to, or in place of, paratopes.

[0012] [8] Figure 3 depicts a non-limiting example of a workflow for assembling and screening an asymmetric multispecific antibody library for functional antibodies.

[0013] [9] Figure 4 depicts the use of flow cytometry to detect GFP expression in Jurkat-NFAT.GFP cells. Figure 4A depicts results for cells with no stimulation. Figure 4B depicts results for cells that have been activated using plate-bound anti-CD3 stimulation to drive the expression of GFP via NFAT signalling. These results validate that the Jurkat-NFAT.GFP cells are able to produce GFP in response to NFAT signalling pathways, in this case driven by anti-CD3 stimulation

[0014]

[0010] Figure 5 depicts the use of flow cytometry to confirm the overexpression of the hybrid PD1.CD3 target antigen on Jurkat-NFAT.GFP-PD1.CD3 cells using an antibody-based anti-PDl detection reagent. Figure 5 A shows the absence of PD1 overexpression on Jurkat-NFAT.GFP cells which have not been transduced with vector encoding the hybrid PD1.CD3 target antigen. Figure 5B shows the presence of PD1 on the surface of Jurkat-NFAT.GFP-PD1.CD3 reporter cells that have been transduced with vector encoding the hybrid PD1-CD3 target antigen.

[0015]

[0011] Figure 6 depicts the use of flow cytometry to assess GFP expression by Jurkat-NFAT.GFP- PD1.CD3 reporter cells (Figs. 6D, 6E, 6F) or control Jurkat-NFAT.GFP cells (Figures 6A, 6B, 6C) that have been stimulated using U2OS-PDL1 cells to agonize PD1 and drive the expression of GFP. Cells received (a) no stimulation, (Figures 6A, 6D), or (b) stimulation by co-culture with U2OS-PDL1 cells that express PDL1 at a 1:1 ratio (Figures 6B, 6E) or (c) stimulation by co-culture with U2OS- PDL1 cells in combination with atezolizumab, a monoclonal antibody that disrupts the interaction of PD1 and PDL1 (Figures 6C, 6F). Blockade of the PD1 / PDL1 interaction by the addition of Atezolizumab abrogates GFP expression, demonstrating that PDL1 -derived agonism of PD1 is driving the expression of GFP. Jurkat-NFAT.GFP cells lacking the hybrid target antigen did not express GFP under any of the tested conditions.

[0016]

[0012] Figure 7 depicts the use of flow cytometry to analyze the expression of fluorescent markers showing that Jurkat-NFAT.GFP-PD1.CD3 were transduced with one or both halves of a biparatopic antibody, during a sequential transduction protocol that generated a large, high-quality, combinatorial cell library of asymmetric biparatopic anti-PDl antibodies. In the first round of transductions, reporter cells were transduced with lentiviral particles encoding for the scFv half of the biparatopic antibody, operatively linked to the expression of the LSSmOrange fluorescent protein, and were then purified by FACS on positivity for scFv, through the expression of LSSmOrange. Figure 7A depicts flow cytometry results for Jurkat-NFAT.GFP-PD1.CD3 cells before transduction, which shows that they express neither half of the biparatopic antibody. Figure 7B depicts flow cytometry results for Jurkat- NFAT.GFP-PD1.CD3 cells after transduction, but prior to FACS purification, with lentiviral vector encoding the scFv half of the biparatopic antibody, which shows that they express the scFv half. In the second round of transductions, scFv-containing cells were transduced with lentiviral particles encoding for the Fab half of the biparatopic antibody, operatively linked to the expression of the mTagBFP fluorescent protein, and then purified by FACS on positivity for both scFv and Fab expression, through LSSmOrange and mTagBFP expression, respectively. Figure 7C depicts flow cytometry results for Jurkat-NFAT.GFP-PDl.CD3-scFv cells after purification by FACS, but before the second round of transductions, which shows that they express only the scFv half of the biparatopic antibody. Figure 7D depicts flow cytometry results for Jurkat-NFAT.GFP-PDl.CD3-scFv after transduction with lentiviral vector encoding the Fab half of the biparatopic antibody, but prior to purification by FACS. Figure 7E depicts flow cytometry results for Jurkat-NFAT.GFP-PD1.CD3- scFv.Fab cells after purification by FACS for scFv and Fab positivity, through the expression of LSSmOrange and mTagBFP.

[0013] Figure 8 depicts flow cytometry results demonstrating that Jurkat-NFAT.GFP-PD1.CD3- scFv.Fab library cells secrete antibodies able to bind to Jurkat-PDl target cells. Figure 8A shows flow cytometry results quantifying the PD1 binding ability of supernatant from control Jurkat-NFAT.GFP- PD1.CD3 reporter cells that express no anti-PDl antibody. Figure 8B shows flow cytometry results quantifying the PD1 binding ability of supernatant from Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library cells. Figure 8C shows flow cytometry results quantifying the PD1 binding ability of 2 pg / mL Palivizumab, an isotype control antibody. These results establish that cells within the Jurkat- NFAT.GFP-PDl.CD3-scFv.Fab library are secreting antibodies able to bind PD1.

[0017]

[0014] Figure 9 depicts flow cytometry results demonstrating that Jurkat-NFAT.GFP-PD1.CD3- scFv.Fab library cells secrete antibodies able to functionally activate PD1, as determined by GFP expression that results from PD1 agonism in Jurkat-NFAT.GFP-PD1.CD3 reporter cells. Figure 9A shows flow cytometry results quantifying the ability of supernatant from control Jurkat-NFAT.GFP- PD1.CD3 reporter cells, which do not express anti-PDl antibody, to activate reporter cells. Figure 9B shows flow cytometry results quantifying the ability of supernatant from Jurkat-NFAT.GFP- PDl.CD3-scFv.Fab library cells, to activate reporter cells. These results establish that cells within the Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library are secreting antibodies able to functionally agonize PD1, and as such, can be mined for rare functional antibodies.

[0018]

[0015] Figure 10 depicts the use of FACS to select and isolate a population of library cells enriched for reporter activity. Figure 10A shows flow cytometry results quantifying the percentage of active library cells, based on positivity for LSSmOrange and mTagBFP expression as well as GFP expression, for the unenriched Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library cells. Figure 10B shows flow cytometry results quantifying the percentage of active library cells, based on positivity for LSSmOrange and mTagBFP expression as well as GFP expression, in the enriched population 1 week after isolation and expansion. These results demonstrate that enriched populations of active library cells can be generated through an iterative sorting process to mine a combinatorial cell library for rare functional activity.

[0019]

[0016] Figure 11 depicts the use of flow cytometry to analyze various clonal populations of autocrine reporter cells that were isolated from the cell library, each expressing a representative single biparatopic anti-PDl antibody. Figure 11A shows flow cytometry results of various clonal populations as analyzed for scFv and Fab positivity, through the expression of LSSmOrange and mTagBFP. Figure 11B shows flow cytometry results of various clonal populations as analyzed for the ability to express GFP through activation of their autocrine reporter system.

[0020]

[0017] Figure 12 depicts the fold-activation of a series of clonal populations of autocrine reporter cells, isolated from the cell library and each expressing a representative single biparatopic anti-PDl antibody, over control Jurkat-NFAT.GFP-PD1.CD3 cells that express no anti-PDl antibody, and a comparative result for autocrine reporter cells treated with GS-0151, a clinical PD1 agonist. Many isolated antibodies elicited a signal at least as good as, or better than, the clinical PD1 agonist.

[0021]

[0018] Figure 13 depicts the use of flow cytometry to analyze the ability of various antibody containing supernatants from isolated clones to bind and functionally agonize PD1. Figure 13 A shows flow cytometry results of various antibody containing supernatants binding to Jurkat-PDl target cells. Figure 13B shows flow cytometry results of various antibody containing supernatants agonizing PD1 on Jurkat-NFAT.GFP-PD1.CD3 reporter cells, leading to GFP expression in the reporter cells.

[0022]

[0019] Figure 14 depicts assay results showing that antibody containing supernatants from a series of isolated clones bind and functionally agonize PD1. Figure 14A shows the results of a PD1 binding assay that compared antibody containing supernatants from isolated clones to control supernatant, expressed as relative-fold binding of the clonal supernatants compared to supernatant from control Jurkat-NFAT.GFP-PD1.CD3 reporter cells that do not express anti-PDl antibody. Figure 14B shows the results of a PD1 reporter assay, detecting PD1 agonism using Jurkat-NFAT.GFP-PD1.CD3 reporter cells, that compared antibody containing supernatants from isolated clones to control supernatant, expressed as relative-fold agonism of the clonal supernatants compared to supernatant from control Jurkat-NFAT.GFP-PD1.CD3 reporter cells that do not express anti-PDl antibody.

[0023] SUMMARY OF INVENTION

[0024]

[0020] The embodiments described herein relate to methods and compositions for generating large combinatorial libraries of multispecific antibodies, preferably asymmetric multispecific antibodies, including nucleic acids, collections of nucleic acids, expression cassettes, vectors for delivery comprising the expression cassettes, cells comprising the vectors, libraries of cells, and methods for screening including methods involving autocrine screening for binding or function. The methods for screening large combinatorial libraries of asymmetric or symmetric multispecific antibodies are useful for antibody -based drug discovery.

[0025]

[0021] According to a first aspect, the disclosure provides a library of cells expressing multispecific antibodies comprising at least about 104cells that each comprise: (a) one or more nucleic acids encoding one or more antigen-binding domains of a first paratope that specifically bind a first epitope of a first target antigen, (b) one or more nucleic acids encoding one or more antigen-binding domains of a second paratope that specifically bind a second epitope, wherein the second epitope is on the same first target antigen or on a second target antigen, and (c) optionally, one or more nucleic acids encoding a reporter system; optionally wherein the antigen-binding domain(s) of the first paratope are not covalently linked as part of the same polypeptide with the antigen-binding domain(s) of the second paratope.

[0022] In some embodiments, the library of cells express an asymmetric multispecific antibody comprising the antigen-binding domain(s) of the first paratope and the antigen-binding domain(s) of the second paratope. In some embodiments, the library of cells expressing the asymmetric multispecific antibody comprises one or more non-paratopic regions, optionally including a heterologous effector molecule.

[0026]

[0023] In some embodiments, the library of cells expressing the asymmetric multispecific antibody further comprises one or more nucleic acids encoding one or more antigen-binding domains of a third paratope that specifically binds a third epitope, wherein the third epitope is on the first target antigen, or on the second target antigen, or on a different third target antigen. In some embodiments, the asymmetric multispecific antibody comprises an Fc region with a first heterodimerization domain and an Fc region with a second heterodimerization domain, such that the two Fc regions bind each other. In some embodiments, the first or second heterodimerization domain comprises a knob, a hole, a leucine zipper, a coiled coil, or a polar amino acid residue capable of forming an electrostatic interaction. For example, the first heterodimerization domain is a knob structure and the second heterodimerization domain is a hole structure.

[0027]

[0024] In some embodiments, the library of cells expressing the asymmetric multispecific antibody comprises one or more light chain constant regions (CL). In some embodiments, the asymmetric multispecific antibody comprises a first CL with a first heterodimerization domain and a second CL with a second heterodimerization domain.

[0028]

[0025] In some embodiments, the library of cells expressing the asymmetric multispecific antibody comprises a first collection of nucleic acids encoding antigen-binding domains of at least 100 different paratopes and a second collection of nucleic acids encoding antigen-binding domains of at least 100 different paratopes, wherein the first and second collections comprise the same paratopes or the first and second collections comprise different paratopes. In any or all of the embodiments, the library of cells expressing the asymmetric multispecific antibody expresses at least 104, 105or 106different asymmetric multispecific antibodies. In some embodiments, the library of cells expressing the asymmetric multispecific antibody are mammalian cells. In some embodiments, the library of cells expressing the asymmetric multispecific antibody are bacterial cells or yeast or fungal cells or insect cells.

[0029]

[0026] In a second aspect, the disclosure provides a method of producing a library of cells expressing an asymmetric multispecific antibody comprising the steps of: (a) transfecting a library of cells with a first collection of delivery vectors comprising expression cassettes encoding antigen-binding domains of at least 100 different paratopes; and (b) transfecting the library resulting from step (a) with a second collection of delivery vectors comprising expression cassettes encoding antigen-binding domains of at least 100 different paratopes. In some embodiments, the cells in step (a) express an autocrine reporter system, optionally wherein the cells in step (a) stably express an autocrine reporter system. In some embodiments the method of producing a library of cells expressing an asymmetric multispecific antibody comprises transfecting the library of cells with one or more nucleic acids encoding the autocrine reporter system prior to step (a), or after step (b). In some embodiments, the expression cassettes of step (a) and / or step (b) are in a viral vector, optionally a lentiviral vector. In some embodiments, transfection results in at least 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85% or more %, preferably at least 70%, of the cells in the library expressing a multispecific antibody and an autocrine reporter system.

[0030]

[0027] In a third aspect, the disclosure provides a method of using the library of cells expressing an asymmetric multispecific antibody to screen for asymmetric multispecific antibodies comprising: (a) detecting when binding of an asymmetric multispecific antibody produced by a cell within the library to a first target antigen results in autocrine reporter system activation of said cell, wherein the autocrine reporter system activation identifies said cell as producing a functional multispecific antibody, and optionally (b) isolating said antibody -producing cell from the library.

[0031]

[0028] In a fourth aspect, the disclosure provides a method of using the library of cells expressing an asymmetric multispecific antibody to screen for functional multispecific antibodies comprising: (a) detecting when binding of an asymmetric multispecific antibody produced by a cell within the library to the first paratope and second paratope, on the same or different target antigens, results in reporter system activation, wherein the reporter system activation identifies said cell as producing a functional multispecific antibody, and optionally (b) isolating said antibody -producing cell from the library.

[0032]

[0029] In any or all of the embodiments, the method is an ultra-high throughput antibody screening method that permits screening of at least about 104, 105, 106, 107, 108, 109to up to about 1010, or more cells, preferably at least about 105cells individual cells per day for producing functional asymmetric multispecific antibody. In some embodiments, the autocrine system comprises a reporter protein that is a fluorescent or membrane -bound protein detectable by fluorescence activity cell sorting (FACS) or by bead-based positive selection. In some embodiments, step (b) comprises FACS or bead-based selection. In a further embodiment, the method of using the library of cells expressing an asymmetric multispecific antibody comprises reducing secreted free antibody in solution by adding decoy beads or decoy cells displaying target antigen, optionally comprising the step of removing the decoy beads or decoy cells.

[0033]

[0030] In some embodiments, a cell expressing an autocrine reporter system and an asymmetric multispecific antibody are individually encapsulated or separated. In some embodiments, a cell expressing a reporter system and a cell expressing an asymmetric multispecific antibody are encapsulated together. In any or all of the libraries or methods the library comprises at least about 104, 105, 106, 107, 108, 109to up to about 1010, or more cells, preferably at least about 105cells, that express asymmetric multispecific antibody. In any or all of the libraries or methods the reporter system comprises a reporter protein selected from the group consisting of: a fluorescent protein, a bioluminescent protein, an enzyme, a split enzyme, an engineered cell surface protein, a unique affinity tag, or combination thereof. In any or all of the libraries or methods the target antigen is a soluble antigen attached to the cell with an anchor or a cell-surface receptor, optionally an enzyme- linked receptor, G-protein-coupled receptor (GPCRs), or ion channel-linked receptor. In some embodiments, the reporter system detects agonism, antagonism, internalization, or inverse agonism of the target antigen.

[0034]

[0031] In a fifth aspect, the disclosure provides a linear lentiviral transfer vector comprising (a) one or more nucleic acids encoding antigen-binding domain(s) of a multispecific antibody, operatively linked to (b) a regulatory element suitable for expressing the multispecific antibody, (c) flanking 5 ’ and 3’ Long Terminal Repeats (LTRs), (d) a Rev responsive element (RRE), and (e) a lentiviral domain, (f) optionally, a central polypurine tract (cPPT), and (g) optionally, a woodchuck post- transcriptional regulatory element (WPRE). In some embodiments, the vector comprises nucleic acids encoding a light chain variable region and nucleic acids encoding a heavy chain variable region. In some embodiments, the vector further comprises nucleic acids encoding one or more non-paratopic regions, optionally including a heterologous effector molecule. In some embodiments, the vector comprises a woodchuck post-transcriptional regulatory element (WPRE). In some embodiments, the vector comprises a central polypurine tract (cPPT).

[0035]

[0032] The present disclosure provides novel methods and nucleic acid systems useful for generating and screening large combinatorial libraries of asymmetric multispecific antibodies to identify functional antibodies. In some embodiments, the disclosure provides methods for the identification of a functional asymmetric multispecific antibody, by establishing a large combinatorial antibody library via (a) generating paratope diversity, (b) assembling paratopes, and / or other effector molecules, into asymmetric multispecific antibody formats, (c) modifying one or more cells to express a single asymmetric multispecific antibody and optionally, one or more reporter systems, (d) contacting the one or more cells with a reporter that detects functional multispecific antibodies and / or effector molecules, and (e) detecting when binding of the antibody to the target antigen results in reporter system activation, wherein reporter system activation identifies the cell as producing a functional multispecific antibody.

[0036]

[0033] In some embodiments, paratope diversity is generated from primary B-cells isolated from an immune animal. In further embodiments, paratopes are generated by single cell rescue of primary B- cells exhibiting antigen specificity. This method may include steps to generate nucleotide sequences that encode paratopes as paired heavy and light chains; or as reformatted-paratopes including, but not limited to, scFv or scFab formats; or as a combination thereof. In any of the embodiments of the foregoing embodiments, nucleotide sequences encoding the paratopes are generated with flanking nucleotide sequences that allow for the assembly of asymmetric multispecific antibodies, including, but not limited to, sequences recognized by restriction enzymes, sequences recognized by recombinases, homology regions, or a combination thereof.

[0037]

[0034] In further embodiments, the disclosure provides expression cassettes and delivery vectors for assembling generated paratopes, and effector molecules of interest, into combinatorial libraries of asymmetric multispecific antibodies. Expression cassettes contain one or more gene regulatory elements operatively linked to said assembled asymmetric multispecific antibodies, and may include a promoter, terminator, enhancer, silencer, or insulator. Expression cassettes may further contain one or more elements required for delivering said assembled asymmetric multispecific antibodies to cells, and may include, but are not limited to, elements facilitating the production of viral delivery vectors; elements encoding integrases, recombinases, transposases, or nucleases; elements facilitating integration, recombination, or transposition; or combinations thereof. In some embodiments, nucleotide sequences encoding non-paratopic regions, including, but not limited to, signal peptides, antibody constant domains, hinges, linkers, and variants or combinations thereof, are included within the expression cassette. In other embodiments, nucleotide sequences that encode non-paratopic regions may be provided separately from the expression cassette during assembly. In some embodiments of the foregoing embodiments, the nucleotide sequences encoding for antibody constant domains may contain mutations that modify their function or promote the heterodimeric assembly of asymmetric multispecific antibodies. In any of the embodiments, sequences encoding expression cassettes and non-paratopic regions may contain flanking nucleotide sequences complementary to those on the generated paratopes or effector molecules, to allow for assembly of multispecific antibodies, including, but not limited to, sequences recognized by restriction enzymes, sequences recognized by recombinases, homology regions, or a combination thereof.

[0038]

[0035] The disclosure provides methods for assembling nucleotide sequences encoding expression cassettes, non-paratopic regions, and generated paratopes or effector molecules into a library of asymmetric multispecific antibody formats. In a further embodiment, this disclosure provides methods to concurrently create multiformat libraries of asymmetric multispecific antibodies.

[0039]

[0036] The disclosure provides further methods for delivering a library of asymmetric multispecific antibodies to cells such that each cell produces a single asymmetric multispecific antibody. In some embodiments, cells are delivered, either sequentially or simultaneously, two expression cassettes such that each are operatively linked to opposing halves of an asymmetric multispecific antibody, such that each cell produces a single asymmetric multispecific antibody. In other embodiments, each cell is delivered a single expression cassette operatively linked to an asymmetric multispecific antibody. In other embodiments, the compositions and methods can be applied similarly to symmetric antibodies, and each cell is delivered a single expression cassete that is operatively linked to a symmetric multispecific antibody. In embodiments of any of the foregoing embodiments, said asymmetric antibody halves or multispecific antibody, are delivered via a viral vector or a non-viral vector delivery system. In some embodiments, delivered expression cassetes operatively linked to asymmetric multispecific antibodies are stably inserted into the genome of the receiving cells in a targeted, pseudorandom, or random manner, mediated by mechanisms including, but not limited to, integration, recombination, transposition, or a combination thereof.

[0040]

[0037] In further embodiments, the disclosure provides methods for contacting one or more cells, each producing an asymmetric multispecific antibody, with a reporter system. In some embodiments, reporter systems may be delivered to cells prior, during, or after delivery of the libraries of asymmetric multispecific antibodies. In additional embodiments, the screening system is an autocrine reporter system that comprises (i) one or more nucleic acid sequence(s) encoding a target antigen, and / or (ii) one or more nucleic acid sequence(s) encoding an reporter module that detects antibody binding to the target antigen and results in activation of said autocrine reporter system, wherein the autocrine reporter system activation identifies the cell as producing a functional multispecific antibody through a detectable signal, marker, or selectable trait from said autocrine reporter module. In any of these embodiments, the reporter system may detect agonism, antagonism, inverse agonism, or internalization of the target antigen. In any aspects of the foregoing embodiments, said reporter system may be operatively linked to one or more gene regulatory elements, and may include promoters, enhancers, silencers, or insulators.

[0041]

[0038] In some embodiments the screening method is applied to pools of cells producing asymmetric multispecific antibodies. These cell pools may comprise at least 104and up to 1010or more asymmetric multispecific antibody -producing cells. The methods may also include one or more steps to prevent false positives arising when antibody secreted by one host cell binds to an adjacent host cell in a paracrine manner.

[0042]

[0039] Other embodiments of the present invention will be evident to one skilled in the art upon reading the present patent specification.

[0043] DETAILED DESCRIPTION OF THE INVENTION

[0044]

[0040] The present disclosure provides compositions and methods for generating large combinatorial libraries of asymmetric multispecific antibodies and collections of nucleic acids encoding such antibodies or portions of the antibodies, e.g. paratopes, non-paratopic regions, and / or effector molecules. In some embodiments, these libraries are interrogated for target antigen binding or functional activity as single multispecific antibody -producing cells. In some embodiments, the engineered cells also comprise an autocrine reporter system that allows the multispecific antibodyproducing cell to assay its own functional activity or binding activity.

[0045]

[0041] The methods described herein greatly increase the format-epitope / paratope space that can be screened by a function-first approach, allowing the interrogation of large combinatorial libraries to identify asymmetric multispecific antibody -drug candidates in an empirical manner, rather than a priori. The methods described herein involve one or more of the following steps: (a) optionally, generating libraries of antigen-specific paratopes and optionally effector molecules; (b) optionally, creating libraries of non-paratopic regions and expression cassettes; (c) optionally, assembling these libraries into asymmetric multispecific antibody formats; (d) combinatorically delivering nucleic acids encoding different polypeptide chains of asymmetric multispecific antibodies to cells, such that each cell produces an asymmetric multispecific antibody; (e) optionally delivering nucleic acids encoding an autocrine reporter system before or after step (d) or alternatively contacting the resulting asymmetric multispecific antibody -producing cells with one or more reporter systems; (f) identifying or selecting cells that produce the desired asymmetric multispecific antibodies; and (g) optionally, producing recombinant asymmetric multispecific antibody encoded by the identified or selected cells, characterizing said antibody, as well as further engineering and / or humanizing said antibody. These one or more steps are further described below.

[0046] DEFINITIONS

[0047]

[0042] As used herein, an “antibody” includes natural formats such as IgG, including IgGl, IgG2, IgG3 and IgG4, IgA, IgM and IgE that include an Fc domain, as well as engineered formats that may contain antibody fragments including but not limited to Fab or Fab’, F(ab’)2, single-chain fragment variable regions (scFv), or single domain antibodies (e.g., VHH). Antibodies may include one or more mutations, e.g. outside of the CDRs, as long as they retain target antigen-binding activity.

[0048]

[0043] As used herein, an “asymmetric antibody” comprises two different, non-identical halves that form a heterodimer through the interaction of regions, e.g. Fc regions, that have been engineered to promote heterodimeric pairing. In contrast, a conventional “symmetric antibody” comprises two identical halves, each typically composed of a paired light and heavy chain, joined through the interactions of their Fc regions such that they form a homodimer. In some embodiments, the compositions and methods described herein exclude BiTEs and BiTE-like formats, which are typically two or more scFv antigen-binding domains joined by a polypeptide linker on a single polypeptide, typically lacking an Fc region.

[0049]

[0044] As used herein, an “antigen-binding domain” refers to a portion of an antibody that binds antigen, e.g. one or more CDRs, or a light chain variable region, or a heavy chain variable region, or a VHH region.

[0045] As used herein, a “multispecific antibody” recognizes two or more epitopes located on the same or different target antigen. An antibody may be multispecific if it comprises two or more different paratopes. Alternatively, an antibody may be multispecific if it comprises at least one paratope and an effector molecule.

[0050]

[0046] As used herein, a “paratope” or “paratopic region” is a part of an antibody that binds to an epitope of a target antigen. A paratope often includes multiple antigen-binding domains that together bind the epitope, e.g. multiple light chain CDRs; multiple heavy chain CDRs; one or more light chain CDRs with one or more heavy chain CDRs; or a light chain variable region with a heavy chain variable region. A paratope may be a single antigen-binding domain, e.g. VHH.

[0051]

[0047] As used herein, a “non-paratopic region” or “non-paratope” is a part of an antibody that does not bind target antigen. Examples include an Fc domain, a linker, or an effector molecule.

[0052]

[0048] As used herein, an “effector molecule” is a non-paratopic region that has an independent functional activity, e.g. a cytokine or receptor ligand or ligand-binding domain. Although the Fc domain has effector functions, “effector molecule” herein refers to a heterologous effector molecule that is not present in natural antibodies.

[0053]

[0049] As used herein, a “heterodimerization domain” is a portion of an antibody, typically within the Fc region, that is modified to promote polypeptide chain pairing. Examples include knob-in-hole mutations, charge-pair mutations or a leucine zipper.

[0054]

[0050] As used herein, a “reporter system” comprises one or more polypeptides that together produce a detectable signal or marker when an antibody binds, activates, agonizes or antagonizes a target antigen. A reporter system includes the target antigen or a fragment thereof that retains one or more epitopes, optionally one or more functional binding partners of the target antigen, and optionally one or more proteins responsible for the signal or marker. The target antigen may be heterologous or native to the cell. Where the multispecific antibody binds two target antigens, the reporter system may include the second target antigen or a fragment thereof that retains one or more epitopes, and optionally include further functional binding partners of the second target antigen, or optionally a further reporter protein.

[0055]

[0051] “Nucleic acid” and “nucleotide” sequences are used herein interchangeably and can refer to a nucleobase sequence encoding a polypeptide or to gene regulatory or translational regulatory elements. ASSEMBLING ASYMMETRIC MULTISPECIFIC ANTIBODIES

[0056] Generating libraries of paratopes

[0057]

[0052] The disclosure provides collections of nucleic acids encoding different paratopes that can be used to produce asymmetric multispecific antibodies, libraries of such paratopes (paratope libraries), and methods for creating such collections and libraries. In some embodiments, the different paratopes in the paratope library (encoded by the collection of nucleic acids) have specificity for two or more antigens. In other embodiments, the paratope library is specific to one antigen, but contains a range of paratopes that are specific for one or more epitopes on said antigen, such that the resulting library can be used to create multiparatopic antibodies.

[0058]

[0053] In some embodiments, antibody paratopes used for library creation are retrieved from primary B-cells isolated from an immune animal of any species, which optionally have been engineered to contain a human immune repertoire or components thereof. The immune animal may be immunized with a target antigen or be naturally immune to a target antigen. Isolated B-cells, or hybridomas, may produce antibodies that bind to or have functional activity against said antigen. Functional antibodies may neutralize pathogens, have anti-tumor activity, modulate the immune system, including inflammation or autoimmunity, or have other biological functional activity. Methods of isolating, synthesizing or mutating antibody cDNA or variants thereof are known in the art, e.g., Babcook et al, 1996 [PMID: 8755564], McClean et al, 2005 [PMID: 15814702], Kosuri et al, 2010 [PMID: 21113165], Kohler & Milstein, 1975 [PMID: 1172191],

[0059]

[0054] In other embodiments, antigen-specific paratopes are selected from an existing library using an antibody display technology. Display technologies may include, but are not limited to, phage display, mammalian display, yeast display, ribosome display, bacterial display, or a combination thereof. For example, scFv libraries expressed and displayed on the surface of a bacteria phage can be used to select antibody paratopes that bind an antigen (Smith, 1985) [PMID: 4001944], Phage libraries can be obtained from a commercial vendor; constructed from B-cells isolated from naive or immune animals, including humans; or generated using DNA synthesis in a random or semi-random fashion. Once an antibody paratope is selected by a display library, as needed, standard molecule biology techniques can be used to isolate the nucleic acids encoding the paratopes and to reformat the expressed antibody paratopes into single-chain formats, paired heavy (VH) and light (VL) chain formats, or a combination thereof.

[0060]

[0055] In some embodiments, paired heavy (VH) and light (VL) chain nucleic acid sequences retrieved from individually isolated B-cells are reformatted into nucleic acids encoding single-chain paratope formats that include, but are not limited to, scFv or scFab formats. For example, scFv formats are created by bridging paired heavy (VH) and light (VL) chain sequences, in either orientation, with a nucleotide sequence encoding for a polypeptide linker 12 to 24 amino acids in length, optionally composed of a repeating pattern of glycine and serine residues (GnS)n. As a further example, scFab formats are created by bridging paired heavy (VH) and light (VL) chain sequences, in either orientation, with a nucleotide sequence encoding for the constant region that corresponds to the first variable region, either the heavy (CHI) or light (CL) chain constant domain, followed by a polypeptide linker 32 to 104 amino acids in length, optionally composed of repeating patterns of glycine and serine residues (GnS)n. To complete the scFab, a further constant region that corresponds to the second variable region, either the heavy (CHI) or light (CL) chain constant domain, is appended to the second variable region. Paratope reformatting can be accomplished using standard molecular biology techniques such as, but not limited to, those described by Andris -Widhopf et al, 2011 [PMID: 21880816],

[0061]

[0056] In further embodiments, paratopes may be a single domain antibody (e.g., VHH). VHH are found in nature, e.g. camelids and sharks naturally produce heavy -chain-only antibodies where the antigen-binding region consists of a single variable domain of the heavy chain (VHH). Artificial humanized VHH may also be produced by transgenic mice or phage display. VHH are also called nanobodies or single domain antibodies (sdAbs).

[0062]

[0057] In some embodiments, parental antibody paratopes retrieved from immune animals or from pre-existing antibody display libraries may be further mutated (i.e. introduction of amino acid insertions, deletions or substitutions) in a random or semi-random fashion, inside or outside of the CDRs, by methods known in the art. The mutations can be done, for example, as part of affinity maturation approaches or for humanization, or for other approaches known in the art.

[0063]

[0058] In any of the foregoing embodiments, the nucleotide sequences encoding antibody paratopes, as paired heavy (VH) and light (VL) chains or single-chain formats, are appended with flanking sequences that allow paratopes to be assembled into asymmetric multispecific antibodies. Flanking assembly sequences are added to paratope sequences via standard molecular biology techniques, and can contain, but are not limited to, nucleotide sequences recognized by restriction enzymes, nucleotide sequences recognized by recombinases, homology regions, or a combination thereof. Assembly sequences are designed such that the nucleic acids which encode paratopes can be combined with nucleic acids encoding non-paratopic regions, and optionally nucleic acids encoding one or more effector molecules, and placed into expression cassettes that allow for delivery to host cells and production of antibody, as described in detail below.

[0064]

[0059] Collections or libraries of the present disclosure containing at least 10, 50, 75, 100, 250, 500, 1000, or up to 104different paratope sequences may be produced in this manner, resulting in combinatorial libraries that contain, for example, 104, 105, 106, 107, 108, 109or 1010, and / or optionally up to 1010unique asymmetric multispecific antibodies. Generating libraries of effector molecules

[0065]

[0060] The disclosure provides collections of nucleic acids encoding dilferent elfector molecules that can be used to produce asymmetric multispecific antibodies, libraries of such elfector molecules, and methods for creating such collections and libraries. This provides a means to create multispecific antibody -fusion proteins that can access novel biology that would be unavailable to either an antibody or an effector molecule alone (Kiefer and Neri, 2016) [PMID: 26864112], Elfector molecules are typically fused to antibodies via a polypeptide linker. In some embodiments, antibody -fusion proteins can be created by fusing effector molecules to the C-terminus of one or more antibody constant domains, on either the heavy and / or light chain. In other embodiments, antibody-fusion proteins can be created by replacing part or all of one or more antibody paratopes with an elfector molecule via an N-terminal fusion to the Fc region of a heavy chain or an N-terminal fusion to a truncated paratope. In yet other embodiments, antibody -fusion proteins can be created by fusing effector molecules to the N- terminus of an antibody chain, e.g. light chain or heavy chain, optionally with an intervening linker. Such antibody -fusion proteins are encompassed by the multispecific antibodies described herein.

[0066]

[0061] Effector molecules could be, as non-limiting examples: a cytokine or chemokine such as, IL- 2, IL-4, IL-6, IL-10, IL-12, IL-15, TNF, INF y, GM-CSF, CXCL8, CCL2, CCL3, CCL4, CCL5, CCL11, or CXCL10; a growth factor such as VEGF, IGF, or FGF, a receptor ligand such as GLP-1; immune modulating receptor or ligand such as PD-1 or PD-L1; or a ligand-binding domain such as, the extracellular domain of TNF receptor II. In other embodiments, one or more mutations is introduced into effector molecules (for example, by introducing mutations into a collection of nucleic acids encoding different effector molecules) to create a library of effector molecules with varied functional properties, such as, but not limited to, differing affinity to their binding partner; or varied physical properties such as, but not limited to, differing linker lengths or linker flexibilities.

[0067]

[0062] In any of the foregoing embodiments, the nucleotide sequences that encode effector molecules contain flanking sequences that allow for the assembly into asymmetric multispecific antibodies. Flanking assembly sequences are added to effector molecule sequences via standard molecular biology techniques, and can contain, but are not limited to, sequences recognized by restriction enzymes, sequences recognized by recombinases, homology regions, or a combination thereof. Assembly sequences are designed such that the nucleic acids encoding effector molecules can be combined with nucleic acids encoding non-paratopic regions and one or more paratopes and placed into expression cassettes for the delivery to host cells and the production of antibody -fusion protein, as described in detail below. Generating libraries of non-paratopic regions

[0068]

[0063] The disclosure also provides collections of nucleic acids encoding different non-paratopic regions that can be used to produce asymmetric multispecific antibodies, libraries of such nonparatopic regions, and methods for creating such collections and libraries. The nucleotide sequences encoding non-paratopic regions include, but are not limited to, antibody constant domains, Fc domains, linkers and hinges, signal peptides, and variants or combinations thereof.

[0069]

[0064] Antibody constant domains are the scaffold to which binding paratopes are connected, facilitating pairing of light and heavy chains into an intact and functional antibody. Examples of antibody constant domains may include, but are not limited to, those found in natural antibody sequences such as human lambda light chains, human kappa light chain, human heavy chain IgGl, human heavy chain IgG2, human heavy chain IgG3, or human heavy chain IgG4, as well as engineered variants of such chains, such as those carrying mutated or chimeric hinge regions, truncations for creating antibody -fusion proteins, mutations to modify binding to Fc receptors, mutations to promote heterodimeric chain pairing (e.g. introduce heterodimerization domains), or combinations thereof. As a non-limiting example, mutations to promote heterodimeric chain pairing of an Fc region can include polar amino acid residues capable of forming electrostatic interactions, Knob-in-Hole (KiH) mutations, or a combination thereof, such as Knob-T366W / K409A and Hole- T366S / L368A / Y407V / F405K (Wei et al, 2017) [PMID: 28881627], alternative KiH-like mutations such as Chain A-T350V / L351 Y / F405A / Y407V and Chain B-T350V / T366L / K392L / T394W (Von Kreudenstein et al, 2013) [PMID: 23924797], or domain swapping mutations such as a strandexchange engineered domain (SEED) approach which comprises alternating IgG- and IgA-derived CH3 domains (Davis et al, 2010) [PMID: 20299542], As a further non-limiting example, mutations to promote heterodimeric chain pairing of an LC region can include polar amino acid residues capable of forming electrostatic interactions, Knob-in-Hole (KiH) mutations, domain swapping mutations, or a combination thereof, such as CrossMab approaches, which comprises various permutations on swapping the CL and CHI domains (Schaefer et al, 2011) [PMID: 21690412], EFab domain substitutions that replaces the native CL and CHI domains with IgE -derived CH2 domains (Cooke et al, 2018) [PMID: 30215570], DuetMab approaches that modify the formation of disulfide bonds between CL and CHI domains (Mazor et al, 2015) [PMID: 25621507], electrostatic mutations such as CH1-K26D and CL-T18R, or CH1-A20L and CL-F7S / A / V (Bbnisch et al, 2017) [PMID: 28981885], or mutations such as CH1-V37W and CL-F98A, CH1-Q39E and CL-Q38R, or CH1-A139G / V190A and CL-L135W / N137A (Corper et al, 20180 [US11078296B2], KiH mutations introduce a sterically bulky amino acid, like tryptophan or tyrosine on one chain and small amino acids like glycine or alanine on the other, to create a protruding “knob” and “hole”, respectively. In a further embodiment, secondary protein structures such as, but not limited to, a coiled coil or a leucine zipper, can be used to promote heterodimeric chain pairing, either used in addition to or in place of, antibody constant domains.

[0070]

[0065] Linkers and hinges are designed to connect effector molecules or reformatted paratopes, such as an scFv, a binding protein, or a cytokine, with the non-paratopic regions required to create an antibody -fusion protein. Examples of linkers and hinges may be derived from natural sequences, such as, but not limited to, the hinge regions of an immunoglobulin, an antibody, CD8, or CD28. Linkers and hinges may also be derived from artificial or synthetic sequences, such as, but not limited to, various permutations on repeating patterns of glycine and serine residues.

[0071]

[0066] Signal peptides allow encoded polypeptides to be trafficked to the host cell secretory system. Non-limited examples of signal peptides include those naturally found in antibody coding sequences, such as mouse Ig kappa, mouse Ig lambda, mouse Ig heavy, human Ig kappa, human Ig lambda, or human Ig heavy, as well as those found in other secreted or transmembrane anchored protein, such as human IL-2, human IL-12, human trypsinogen-2, human insulin, VSV-G, influenza haemagglutinin, gaussia luciferase, or silkworm fibroin light chain.

[0072]

[0067] Libraries of non-paratopic regions (or the collection of nucleic acids encoding the different non-paratopic regions) can be used to assemble asymmetric multispecific antibodies with increased diversity; thereby improving the chances of finding a functional multispecific antibody. Examples of non-paratopic region libraries that can be used to modify the biophysical characteristic of an antibody include, but are not limited to, libraries of constant domains in which modifications to the hinge region is used to create diversity in paratope flexibility, libraries of linkers or hinges that change the flexibility or distance of the connection between an effector molecule, or single -chain paratope, and the constant domains, or libraries of modified constant domains having different stabilities. Examples of non-paratopic region libraries that can be used to modify the biological features of an antibody include, but are not limited to, libraries of constant domains that bind to Fc receptors with varying affinities, or libraries of constant domains that have altered glycosylation patterns.

[0073]

[0068] As a further embodiment, libraries of non-paratopic regions can be used to simultaneously assemble asymmetric multispecific antibodies in multiple formats. This allows different formats to be interrogated alongside variations in epitope / paratope space, allowing the interrogation of large diversity libraries to empirically identify antibody -drug candidates. As a non-limiting example, and further described below, a library of constant domains that allows for heterodimeric chain pairing can be used to simultaneously assemble bispecific paratopes into an antibody library in which each combination of paratopes is configured in multiple different formats, such as, but not limited to, Fab- Fab, Fab-scFv, and scFv-scFv. These three formats all have different geometries and biophysical properties, which allows a large amount of diversity to be searched through for an ultra-rare functional antibody.

[0069] In any of these foregoing embodiments, nucleotide sequences encoding non-paratopic regions are appended with flanking sequences, complementary to those on the nucleic acids encoding paratopes and / or effector molecules, to allow for the assembly of asymmetric multispecific antibodies. Flanking assembly sequences are added via standard molecular biology techniques, and can contain, but are not limited to, nucleotide sequences recognized by restriction enzymes, nucleotide sequences recognized by recombinases, homology regions, or a combination thereof. In some embodiments, the nucleotide sequences of one or more non-paratopic regions are included within the expression cassette prior to the assembly of asymmetric multispecific antibodies.

[0074] Generating expression cassettes for expressing antibodies in host cells

[0075]

[0070] This disclosure also provides expression cassettes that can be used to express libraries of asymmetric multispecific antibodies (including asymmetric multispecific antibody -fusion proteins). The disclosure further provides methods for producing the expression cassettes. Expression cassettes may comprise (a) nucleic acids encoding paratopes (optionally with nucleic acids encoding for one or more non-paratopic regions and optionally nucleic acids encoding effector molecules), operatively linked to (b) the gene regulatory and optionally translational regulatory elements, required to express and produce an antibody within a host cell, and (c) the assembly sequences required to assemble collections of nucleic acids into the expression cassette,

[0076]

[0071] To produce a polypeptide from the nucleotide sequence encoding for an asymmetric multispecific antibody, expression cassettes are designed to contain one or more gene regulatory elements that are operatively linked to said assembled asymmetric multispecific antibodies, and may include a promoter, enhancer, terminator, silencer, or insulator. In some embodiments, the gene regulatory element is optionally, a cytomegalovirus promoter (CMV), a CMV enhancer, an SV40 promoter, a SFFV promoter, a MNDU3 promoter, an HIV-1 LTR promoter, an EF- la promoter, a U6 promoter, a Hspa5 promoter, a A2UCOE promoter, a Ubiquitous Chromatin Opening Element (UCOE), a Tetracycline Regulatory Element (TRE), a Woodchuck Hepatitis Virus Post- Transcriptional Regulation Element (WPRE), a polyadenylation signal, a (3-Globin Intron Element, or a combination thereof.

[0077]

[0072] In some embodiments, expression cassettes are designed to contain translational regulatory elements, operatively linked to an assembled multispecific antibody, and can include, but are not limited to internal ribosome entry site (IRES) or ribosome skipping motifs such as 2A or 2A-like sequences.

[0078]

[0073] In further embodiments, nucleotide sequences that encode gene regulatory elements or translational regulatory elements may be added to an expression cassette during assembly.

[0074] In any of these foregoing embodiments, the nucleotide sequences encoding expression cassette contain assembly sequences that are complementary to the flanking assembly sequences of the nucleic acid collections encoding paratopes, non-paratopic regions, and / or effector molecules, to allow asymmetric multispecific antibodies to be assembled within, and operatively linked to, the gene regulatory element(s) in the expression cassette.

[0079]

[0075] In some embodiments, the nucleotide sequences encoding the expression cassette may further comprise one or more non-paratopic regions prior to the assembly of asymmetric multispecific antibodies.

[0080] Generating vectors for stably delivering antibody-containing expression cassette to host cells

[0081]

[0076] The disclosure also provides delivery vectors comprising (a) an expression cassette, (b) one or more delivery elements for stably delivering assembled asymmetric multispecific antibodies to host cells, such that each host cell produces one type of asymmetric multispecific antibody, and (c) a selectable marker. The disclosure further provides methods for creating these delivery vectors. In some embodiments, nucleotide sequences that encode delivery elements and selection markers may be appended to an expression cassette, and optionally, placed into a vector during assembly.

[0082]

[0077] In some instances, delivery elements may facilitate the packaging and production of viral delivery vectors such as, but not limited to, DNA-based viral vectors, such as an adenovirus, an adeno-associated virus, a herpes virus, a polyomavirus, a poxvirus, or a reoviruses, as well as RNA- based viral and retroviral vectors, such as a lentivirus, a pseudotyped lentivirus, an alphavirus, a flavivirus, a rhabdovirus, a measles virus, an avulavirus, or a picomavirus.

[0083]

[0078] A non-limiting example is lentiviral or pseudotyped lentiviral delivery vectors that can be used as a tool to stably integrate nucleic acid sequences encoding asymmetric multispecific antibodies into a host cell. Pseudotyped lentiviral vectors are lentivirus in which the native viral envelope protein is complemented or replaced by a heterologous envelope protein to alter the viral particle cell tropism and to facilitate entry into a host cell (Gutierrez-Guerrero et al, 2020) [PMID: 32933033], Nonlimiting examples of envelope proteins used to create pseudotyped lentiviral vectors include vesicular stomatitis virus G glycoprotein (VSV-G), MLV envelope glycoprotein, BaEV envelope glycoprotein, gibbon ape leukemia virus glycoprotein (GALV), or cat endogenous retrovirus envelope glycoprotein (RD 114).

[0084]

[0079] Typically, lentiviral particles are produced by co-transfecting a production cell line, such as HEK293T, with a series of circular plasmids that contain the regulatory elements and gene products required for lentiviral production and subsequently, harvesting and purifying the resulting viral particles (Merten et al, 2016) [PMID: 27110581], This includes the transfer vector, which contains a transgene and the cis-acting regulatory elements required to package said transgene into a viral particle and deliver it into the host cell genome, such as the domain, the central polypurine tract (cPPT), the Rev responsive element (RRE), and the 5’ and 3’ Long Terminal Repeats (LTRs) (Johnson et al, 2021) [PMID: 33981779]; and the packaging vectors, typically, two (second-generation systems) or three (third-generation systems) vectors, one encoding a viral envelope protein such as VSV-G that facilitates host cell entry, and the others encoding the Gag, Pol, Rev, and Tat genes required to create an viral particle capable of integrating into a host genome (Dull et al, 1998) [PMID: 9765382], To create an expression cassette compatible with lentiviral delivery, required cis-acting regulatory elements must flank the expression cassette. In some embodiments of this disclosure, these elements are already present in a circular vector that contains the expression cassette, such that after assembly of an asymmetric multispecific antibody into the expression cassette, lentivirus particles containing antibody-encoding nucleotide sequences can be created using a transfer vector that is a circular plasmid. Circular transfer vectors can be produced by methods such as, but not limited to, passage through a host cell or host bacteria, or through enzymatic methods, such as PCR, restriction-ligation, rolling-circle, or isothermal amplification. Viral particles can be produced as described above, and then used to transduce host cells.

[0085]

[0080] The disclosure provides as a separate and independent aspect a linear lentiviral transfer vector comprising (a) one or more nucleic acids encoding the antigen-binding domain(s) and optionally nonparatopic regions of an asymmetric multispecific antibody, operatively linked to (b) a regulatory element suitable for expressing said antibody, (c) flanking 5’ and 3’ Long Terminal Repeats (LTRs), (d) a Rev responsive element (RRE), and (e) a lentiviral domain, (f) optionally, a central polypurine tract (cPPT), and (g) optionally, a woodchuck post-transcriptional regulatory element (WPRE).

[0086]

[0081] LECs are linear DNA molecules that typically contain gene regulatory elements (for example, a promoter and transcriptional terminator) operatively linked to antibody coding sequences. When transfected into a host cell, LECs are transcribed and translated, resulting in the production of antibodies. LECs accelerate the process of antibody discovery and characterization, as they can be directly created through high-throughput DNA manipulation and amplification, without the need to passage through bacterial hosts (Zhang et al., 2021). However, unlike lentiviral vectors, LECs are not able to stably integrate DNA sequences into a host cell. Stable integration of a transgene offers advantages over transient transfection, including long-term expression and the ability to manipulate the number of integration events in a host cell by precisely controlling the multiplicity of infection (MOI) during transduction.

[0087]

[0082] Antibody linear expression cassette (LEC) were first disclosed by Liao et al, 2009 [PMID: 19428587], but have not included lentiviral cis-acting regulatory elements and have not previously been used in this manner. Lentiviral-competent LECs (LTR-LECs) can be produced by appending flanking lentiviral cis-acting regulatory elements to a LEC using standard molecule biology techniques. Specifically, the 5’ end of the LEC is modified to contain a 5’ lentiviral LTR, a domain, a cPPT, and an RRE upstream of the LEC promoter, while the LEC is modified to contain a selectable marker and a 3’ LTR downstream of the antibody coding sequence. Importantly, the termination and polyadenylation signals typically used for the creation of LECs are incompatible with the production of lentiviral particles and must be replaced by a viral-derived termination signals compatible with retroviral vectors, such as WPRE. In practice, LTR-LECs are assembled using standard molecular biology techniques in a similar manner to that used for LECs. After assembly, DNA for viral particle production, comprising LTR-LECs, can be generated through enzymatic methods, such as PCR, rolling-circle, or isothermal amplification. In further embodiments, nucleic acids comprising LTR- LECs may be modified to increase transfection efficiencies or cellular half-life. Non-limiting examples of modifications include, covalently closed ends, phosphorothioate bonds, 2’-O-methoxy- ethyl bases, 2' fluorine bases, or a combination thereof. As described above, to produce LTR-LEC viral particles, linear DNA molecules comprising the LTR-LEC are co-transfected with circular packaging plasmids into a production cell line such as HEK293T. Subsequently, the resulting viral particles are harvested, purified, allowing them to be used to transduce host cells.

[0088]

[0083] In a further embodiment, delivery elements are integrase attachment site(s) flanking the expression cassette and gene regulatory elements operatively linked to a nucleotide sequence(s) encoding for the corresponding integrase(s), such as, but not limited to, Bxbl and / or cpC31 sitespecific serine integrases. In further embodiments, delivery elements are recombinase attachment sites that flank the expression cassette and gene regulatory elements operatively linked to a nucleotide sequence(s) encoding for the corresponding recombinase(s), such as, but not limited to Flp or Cre sitespecific tyrosine recombinase. Site-specific integrases or recombinases allow DNA to be inserted into specific locations within a host cell genome. In some examples, this may use a single integration site to insert an entire circular plasmid or use two integration sites to facilitate a recombinase-mediated cassette exchange strategy.

[0089]

[0084] Integration- or recombination-competent delivery vectors can be delivered to host cells using approaches such as, but not limited to, mechanical, chemical, electroporation techniques, as well as non-integrative viral vectors. Once in the cell, the integrase or recombinase is expressed, and the vector is integrated into the host cell genome. In some embodiments, nucleotides encoding for an integrase or recombinase may be delivered separately from the vector containing the multispecific antibody expression cassette. In other embodiments, the host cell contains a landing pad, containing attachment sites corresponding to those on the antibody expression cassette, that increase the efficiency of DNA integration and allow for specific targeting within the host cell. Landing pads may contain gene regulatory elements and selectable markers used to identify host cells that have undergone successful integration events.

[0085] In some embodiments of this disclosure, the expression cassette is flanked by the appropriate delivery elements required to allow for transposon-mediate integration into a host cell genome via a DNA transposase or retrotransposase. In some embodiments, nucleotides encoding for the transposase or retrotransposase flank the antibody expression cassette, in other embodiments, these nucleic acids may be delivered separately. Non-limiting examples of transposons that could be used to deliver an expression cassette into a host cell genome include Sleeping Beauty, piggy Bac, Mage, and Tol2. Transposon-competent delivery vectors can be delivered to host cells using approaches such as, but not limited to, mechanical, chemical, electroporation techniques, as well as non-integrative viral vectors. Once in the cell, the transposase is produced, leading to the vector being integrated into the host cell genome. In some embodiments, nucleotides encoding for the transposase may be delivered separately from the vector containing the expression cassette.

[0090]

[0086] In some embodiments, the delivery elements are flanking homology arms that allow the expression cassette to be integrated into a host cell genome via the action of a site-specific nuclease such as, but not limited to, a transcription activator-like effector nuclease (TALEN), a zinc finger nuclease, or a CRISPR-based nuclease. In some embodiments, delivery vectors contain gene regulatory elements that are operatively linked to nucleotides encoding for a nuclease and any required accessory sequences, as a non-limiting example, one or more guide RNAs. In other embodiments, these gene regulatory elements and nucleic acids may be delivered separately.

[0091]

[0087] In any of these foregoing embodiments, the delivery vector may contain gene or translational regulatory elements operatively linked to an encoded protein which allows for the identification and / or selection of cells to which the delivery vector has been successfully integrated into the host cell DNA. As a non-limiting example, this could include resistance genes such as puromycin or hygromycin resistance markers; a fluorescent protein such as blue, green, red, or infrared fluorescent proteins; a surfaced expressed marker that can be used for detection via an antibodies or staining reagent, such as truncated CD34, truncated CD19, or truncated CD4; a surfaced expressed marker that can be used for affinity selection, such an extracellular strep-tag fused to a transmembrane anchor; or a combination thereof.

[0092] Methods of assembling libraries of vectors containing asymmetric multispecific antibodies

[0093]

[0088] As depicted in Figure 1, the disclosure provides methods for assembling collections of nucleic acids, comprising sequences that encode paratopes, effector molecules, and non-paratopic regions, in such a way as to place them into an expression cassette that can subsequently be delivered to a host cell, resulting in the expression and production of one or more polypeptide chains that represents one half of an asymmetric multispecific antibody (e.g. a light chain polypeptide and heavy chain polypeptide, or a single polypeptide encoding a scFv-Fc fusion). Assembly of these sequences in an ordered manner is facilitated by flanking nucleotide sequences that can contain, but are not limited to, sequences recognized by restriction enzymes, sequences recognized by recombinases or integrases, homology regions, or a combination thereof. To create a library, appropriate collections nucleic acid encoding portions of the asymmetric multispecific antibody (e.g. encoding paratopes, non-paratopic regions, effector molecules) are combined, and then assembled using standard molecular biology techniques, into a collection of nucleic acids in which each nucleic acid encodes for a different half of an asymmetric multispecific antibody.

[0094]

[0089] Standard molecular biology techniques, including but not limited to polymerase chain reaction (PCR), fusion-PCR, adaptor-ligation, homology -mediated cloning, or ligation-mediated cloning are used to create recombinant DNA sequences.

[0095]

[0090] In some embodiments of the present disclosure, a method for creating libraries of asymmetric multispecific antibody halves is described. Using antibody constant domains that contain mutations promoting heterodimeric chain pairing, two vector libraries are assembled. Each vector within these libraries comprises an expression cassette that contains one of the two asymmetric constant domains and encodes a specific half of an asymmetric multispecific antibody (e.g. a light chain polypeptide and heavy chain polypeptide, or a single polypeptide encoding a scFv-Fc fusion). As each asymmetric multispecific antibody half contains delivery elements that allow a single copy of the expression cassette to be stably integrated into the host cell genome, large amounts of combinatorial diversity can be created during the cell creation process. In some embodiments, each library is assembled in such a way that asymmetric constant domains are combined with paratopes configured in multiple formats such as, but not limited to, Fab or scFv. During the cell creation process, this allows a cell population carrying the library to simultaneously contain different formats, such as, but not limited to, Fab-Fab, Fab-scFv, and scFv-scFv paratope configurations.

[0096]

[0091] Fab paratopes are generally composed of the VH and CHI domains of the HC and the whole LC (comprising VL and CL domains). ScFv paratopes are molecules generated by connecting the variable regions of heavy (VH) and light (VL) chains, often via a flexible linker. VHH paratopes are heavy-chain-only antibodies where the antigen-binding region consists of a single variable domain of the heavy chain (VHH).

[0097]

[0092] As depicted in Figure 2, the methods described above can be used to assemble asymmetric multispecific antibodies in different formats. Non-limiting examples of depicted formats include: (1) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that one is a Fab and the other is an scFv, depicted paratopes are specific for different epitopes on one or more antigens; (2) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that both are in an scFv format, depicted paratopes are specific for different epitopes on one or more antigens; (3) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that both are in a Fab format and light and heavy chain pairing is facilitated by modification of the light chain constant regions (CL) and heavy chain constant regions (CHI), depicted paratopes are specific for different epitopes on one or more antigens; (4) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that one is a Fab and the other is an scFv, specific for different epitopes on one or more antigens, and the polypeptide chain encoding the scFv paratope contains an additional scFv binding-domain fused to the C-terminal of its Fc domain that is specific for a third epitope; (5) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that both are in an scFv format and are specific for different epitopes on one or more antigens, in addition, both heterodimeric Fc regions contain an additional scFv binding-domains fused to the C-terminal of their Fc domain; (6) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that it contains two fused Fab domains on one half and one Fab domain on the other, where light and heavy chain pairing is facilitated by modification the light chain constant regions (CL) and heavy chain constant regions (CHI), depicted paratopes are specific for different epitopes on one or more antigens; (7) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which one paratope is configured such that it contains one scFv with an N-terminally fused Fab domains on one half and an scFv domain on the other, depicted paratopes are specific for different epitopes on one or more antigens; (8) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that it contains one scFv with an N-terminally fused Fab domains on one half and another Fab domain on the other half, where light and heavy chain pairing is facilitated by modification the light chain constant regions (CL) and heavy chain constant regions (CHI), depicted paratopes are specific for different epitopes on one or more antigens; (9) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that one is a Fab and the other is an scFv, specific for different epitopes on one or more antigens, and the polypeptide chain encoding the scFv paratope contains an effector molecule fused to the C-terminus of its Fc domain; (10) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, which is configured such that one half is a Fab paratope and the other is an effector molecule; (11) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that both are in a Fab format and light and heavy chain pairing is facilitated by modification the light chain constant regions (CL) and heavy chain constant regions (CHI), in addition, the polypeptide chain encoding one of the Fab domains contains an effector molecule fused to the C-terminus of its Fc domain, depicted paratopes are specific for different epitopes on one or more antigens; (12) an asymmetric multispecific antibody, comprising two halves joined for example at the Fc region, in which paratopes are configured such that one is a Fab and the other is an scFv, specific for different epitopes on one or more antigens, the polypeptide chain encoding the Fab domains contains an effector molecule fused to the C-terminus of its Fc domain, the polypeptide chain encoding the scFv domains contains an additional scFv binding-domain, specific for a third epitope, fused to the C-terminus of its Fc domain.

[0098]

[0093] In some embodiments of this disclosure, the paratope libraries have specificity for two or more antigens and can be used to create multispecific antibodies. In other embodiments, the paratope library is specific to one antigen target, but contains a range of paratopes that are specific for different epitopes on said antigen, such that the resulting library can be used to create multiparatopic antibodies. Multiparatopics are a form of multispecific antibodies that contain multiple paratopes that typically bind to distinct epitopes on a target antigen. Biparatopics, targeting two epitopes on a target antigen, and triparatopics, targeting three epitopes on a target antigen are multiparatopic antibodies known to the art, and are all forms of multispecific antibodies. In some embodiments, multispecific antibodies can be antibody -fusion proteins. These fusion proteins contain effector molecules, such as, but not limited to, a cytokine, a receptor ligand, or a ligand-binding domain, that are typically fused to an antibody constant domain or replace at least one antigen-binding paratopes. These antibody -fusion proteins are a form of multispecific antibody because the effector molecule provides specificity to one target, while the remaining paratopes provide specificity for either another target (multispecific -like) or an epitope unique from the effector molecule binding site on said target (multiparatopic -like). A multispecific antibody that contain a paratope that has been reformatted into a single-chain, such as a scFv or scFab, could also be considered as a form of antibody -fusion protein. In a further embodiment, various paratope libraries and effector molecules are combined, such that the resulting multispecific antibodies combine the characteristics of multispecific antibodies, multiparatopic antibodies, antibody -fusion proteins, or a combination thereof.

[0099]

[0094] In another embodiment, nucleic acids encoding asymmetric multispecific antibody halves are pre-assembled, as linear nucleic acids or as circular plasmids, and then pooled such that each collection of nucleic acids represents one half of an asymmetric multispecific antibody. These two libraries are then used in a combinatorial assembly step that results in a single circular vector, containing an expression cassette which encodes two halves and produces an intact asymmetric multispecific antibody. This combinatorial assembly step is used to generate library diversity and can be accomplished using standard molecular biology techniques. Assembly is accomplished in such a way as to allow the asymmetric multispecific antibody library to simultaneously contain different formats, such as, but not limited to, VHH-VHH, Fab-Fab, Fab-scFv, and scFv-scFv paratope configurations. The method described above can be used to assemble asymmetric multispecific antibodies in different formats, as depicted in Figure 2. In a further embodiment, nucleic acids encoding multispecific antibodies are assembled into a circular vector in a symmetric format where the encoded multiple paratopes and / or effector molecules are expressed as one or more polypeptide chains. Diversity is generated during assembly, and the resulting expression cassettes produce intact antibodies from one symmetric antibody half. In any of the forgoing embodiments, assembled vectors contains delivery elements that allow a single copy of the expression cassette to be stably integrated into a host cell genome, such that after the cell creation process, each cell in the population produces a single multispecific antibody.

[0100]

[0095] In some embodiments, one or more parts may be pre-assembled, as either a linear nucleic acids or as circular plasmids, prior to assembly into a collection of nucleic acids encoding multispecific antibodies. As a non-limiting example, a nucleic acid encoding antibody light chain variable (VL) and constant regions (CL) could be pre-assembled with their corresponding nucleic acid encoding heavy chain variable region (VH) to create a single nucleic acid encoding for a Fab paratope. These combined parts, which preserve heavy (VH) and light chain (VL) pairing, could then be pooled, and used as a collection to assemble a library of asymmetric multispecific antibodies. As a further non-limiting example, nucleic acids encoding one or more single-chain paratopes or effector molecules could be pre-assembled in such a way as to fuse them with a linkers sequence to a nucleic acid encoding the C-terminus of a heterodimeric constant domain or the N-terminus of a given paratope. These nucleic acids could then be pooled into a collection and used to assemble libraries of trispecific or tetraspecific antibodies.

[0101] Methods for creating reporter systems

[0102]

[0096] The disclosure also provides methods for creating reporter systems capable of identifying asymmetric multispecific antibodies, preferably functional asymmetric multispecific antibodies. Reporter systems described herein apply to any type of target antigen, soluble, secreted, or cell- surface-expressed. Cell-surface antigens include cell-surface receptors. Classes of cell-surface receptors include enzyme-linked receptors, G-protein-coupled receptors (GPCRs), and ion channel- linked receptors. The enzyme-linked receptor class includes receptor tyrosine kinases (RTKs), serine / threonine kinases, receptor guanylyl cyclase’s, tyrosine-kinases associated receptors and receptor tyrosine phosphatases. The G-protein-coupled receptors include rhodopsin-like, the secretin family, metabotropic glutamate receptors, fungal mating pheromone receptors, cAMP receptors and frizzled and smoothened receptors. The ion channel-linked receptors include voltage-gated, ligandgated, ionotropic glutamate receptors, P2X receptors and mechano-sensitive receptors (Purves et al., 2001). Upon antibody engagement of a cell-surface receptor several structural and physiological cellular events may occur, including but are not limited to, receptor dimerization, receptor shape change, receptor internalization and degradation, recruitment of additional interacting protein(s), the production of second messengers (e.g., cAMP), the activation of protein kinases and subsequent phosphorylation of pathway protein(s), transcription of pathway -specific genes (e.g., IFNy), cytoskeletal re-arrangements, cell survival or cell death.

[0097] There are many assays known in the art that enable the functional interrogation of most, if not all, receptor types and associated cellular events. These previously described assays can be adapted into reporter systems which enable the efficient discovery of functional antibodies against all receptor types and against cognate ligands and binding partners. When the reporter system is expressed in the antibody -producing cell so that it can be activated by antibody produced by that cell, it is an autocrine reporter system.

[0103]

[0098] Examples of target antigens include, but are not limited to 4-1BB, 5HT3, APRIL, av(38, B7H3, BCMA, BTLA, CB1, CCL2, CD3, CD19, CD22, CD28, CD33, CD40, CD47, CD79b, CD30, Claudinl8.2, cMet, CTLA-4, EGFR, FGFR3, Folate receptor-a, Frizzled-1, GIPR, GLP1R, Her2, Her3, Integrin |36, IL-4, IL-17, IL-21, LAG-3, LRP5, LRP6, Navi.7, Nectin-4, NPR1, P2X3, PD-1, PD-L1, R0R1, TGF(3R, TIGIT, TIM-3, Tissue Factor, TNFa, TNFR2, TROP2, VEGFR, and their ligands, decoy proteins, and / or co-receptor etc.

[0104]

[0099] Assays suitable for detecting various functions, such as agonism, inverse agonism, antagonism or internalization of a target antigen, are known in the art and may be adapted for use as a reporter system. In some embodiments, the reporter system comprises (i) nucleic acid sequence(s), e.g. DNA and / or RNA, encoding a target antigen, and optionally, (ii) nucleic acid sequence(s), e.g. DNA and / or RNA, encoding a reporter protein or marker that is detectable when binding to target antigen occurs or (iii) a detection reagent that indicated binding to the target.

[0105]

[0100] In some embodiments, the reporter system comprises (i) nucleic acid sequence(s), e.g. DNA and / or RNA, encoding one, two or more target antigens and (ii) nucleic acid sequence(s), e.g. DNA and / or RNA, encoding a reporter module (one or more proteins that interact functionally with the one, two or more target antigens, including a reporter protein or marker that is detectable when the desired function occurs). The target antigen(s) is (are) “functionally linked” to the reporter module when the desired function (for example, agonism, inverse agonism, antagonism, or internalization), causes the autocrine reporter module to be activated (as in, produce a detectable signal or selectable trait). In this manner, a cell secreting a functional asymmetric multispecific antibody can be identified and / or isolated from a pool of other antibody -producing cells by the detectable signal or selectable trait that occurs when the desired function (for example, agonism, inverse agonism, antagonism or internalization) of the target antigen(s) causes activation of the autocrine reporter module.

[0106]

[0101] The nucleic acids encoding the reporter systems may be on the same vector or different vectors. In some embodiments, one or more vectors of the reporter system can be delivered transiently or by stable integration into the host cell. In further embodiments, viral and non-viral vector strategies are used to deliver the autocrine reporter system. Viral vectors suitable for use herein include but not limited to, DNA-based viral vectors, such as an adenovirus, an adeno-associated virus, a herpes virus, a polyomavirus, a poxvirus, or a reoviruses, as well as RNA-based viral and retroviral vectors, such as a lentivirus, a pseudotyped lentivirus, an alphavirus, a flavivirus, a rhabdovirus, a measles virus, an avulavirus, or a picornavirus. Non-viral systems suitable for use herein include, but not limited to, virus-like particles, nanoparticles, liposomes, lipids, cationic lipids, polycationic polymers, dendrimers, biolistics, microinjection, magneto-transfection, chemical transfection, hydrodynamic gene transfer, cell-penetrating peptides, electroporation, mechanical delivery systems and / or sonoporation. Libraries may be interrogated in a reporter system in which the antibody -producing cell is incubated with another cell containing a reporter system, allowing evaluation of multispecific antibodies that bind two different target antigens as well as multiparatopic antibodies that bind a single target antigen.

[0107] Methods to create combinatorial cell libraries producing asymmetric multispecific antibodies

[0108]

[0102] The disclosure provides methods for creating cell populations in such a way as to create combinatorial libraries of cells where each member of the population expresses a single species of asymmetric multispecific antibody. In some embodiments, two vector libraries, each comprising an expression cassette that contains one of the two asymmetric constant domains and encoding a specific half of an asymmetric multispecific antibody, are sequentially delivered to a population of cells such that each cells receives one copy of each half of an asymmetrical multispecific antibody. In doing so, the resulting library of intact antibodies contains a large amount of combinatorial diversity.

[0109] Generating diversity in this manner: (1) reduces the complexity, time, burden of creating large combinatorial libraries compared to strategies purely reliant on molecular biology techniques such as high-throughput cloning, and (2), allows for the each of the two antibody halves to be carried on separate delivery vectors, which avoids the packaging-limits associated with many viral vectors, and the loss of delivery efficiency due to increased nucleic acid size that is associated with many non-viral methods.

[0110]

[0103] In some embodiments, the reporter system used to identify antibodies is expressed by the antibody-producing cell (i.e. autocrine, allowing for the antibody-producing cell to identify that the antibody it produces binds a target antigen or is functional). In some embodiments, the autocrine reporter system is delivered to the host cells prior to creation of the asymmetric multispecific antibody library (e.g. the introduction of one or more collections of nucleic acids encoding paratopes or antibodies). In other embodiments, the autocrine reporter system is delivered to the host cells after creation of the asymmetric multispecific antibody library (e.g. the introduction of one or more collections of nucleic acids encoding paratopes or antibodies). In a further embodiment, the autocrine reporter system is delivered to the host cells during the creation of the asymmetric multispecific antibody library.

[0111]

[0104] One example embodiment is depicted in Figure 3. In this workflow, a library of asymmetric biparatopic antibodies is created and then screened to identify functional antibody-drug candidates. In the first step, single B-cells producing antibodies specific to a target antigen are isolated from an immune animal immunized against said antigen. Antibody cDNA is then generated from each individually isolated B-cells, using standard molecular biology techniques such as RT-PCR, to create a template for amplification of the nucleotide sequences encoding the variable regions of antibody heavy (VH) and light (VL) chains in a manner that preserves chain pairing. Optionally, if paratopes are already known, DNA synthesis can be used to create paired antibody heavy and light chains. The nucleotide sequences encoding the paired antibody chains are then divided into two or more groups, and standard molecular biology techniques are used to modify each group to either: (1) add flanking assembly sequences to both the heavy and light chain sequences, or (2) reformat paired antibody chains into single-chain formats and append flanking assembly sequences. In the next step, the generated paratopes are assembled with appropriate non-paratopic regions, using standard molecular biology techniques, such that multispecific antibody halves are placed in expression cassettes within vectors capable of forming lentiviral particles. Overall, this process yields two collections of vectors, which each correspond to one of the halves of an asymmetric multispecific antibody. These vector collections are then individually co-transfected with packaging plasmids into production cell lines, such as HEK293T, to create two collections of lentiviral particles. Each collection encodes for one of the two specific halves of an asymmetric multispecific antibody. The first lentiviral collection is then transduced at a low multiplicity of infection (MOI) into a cell population that stably expresses an autocrine reporter system. Transduction with a low MOI ensures that most cells receive a single viral integration event. Cells are then recovered and purified for successfully transduced cells using methods such as fluorescence-activated cell sorting, antibiotic selection, magnetic-bead selection, or a combination thereof. The resulting purified cells express a single copy of one half on an asymmetric antibody library. These cells are then expanded, transduced at a low MOI with the second lentiviral pool, such that transduced cells receives a single copy of the second half of the asymmetric antibody library, and purified to select for successfully transduced cells. The cells resulting from this two-step transduction method now each contain a single copy of both halves of an asymmetric multispecific antibody, and as such, produce an intact antibody that can be screened for functional activity. The resulting cell population is then subjected to screening via the autocrine reporter system, allowing cells expressing functional asymmetric multispecific antibodies to be isolated and the antibody sequences recovered by sequencing, as detailed below.

[0112]

[0105] According to this aspect, collections of linear lentiviral transfer vectors containing nucleic acids encoding at least 10, 50, 75, 100, 250, 500, 1000, or up to 104different antibody polypeptide chains may be produced in this manner, resulting in combinatorial libraries that contain, for example, 104, 105, 106, 107, 108, 109or 1010, and / or optionally up to 1010unique asymmetric multispecific antibodies.

[0106] In the non-limiting example described above, to ensure that each cell in the population expresses a single asymmetric multispecific antibody, the cell population is subjected to two stepwise lentiviral transductions, each at a low MOI, to ensure that each cell receives one copy of nucleic acids from each collection, such that it receives nucleic acids encoding both specific halves of an asymmetric multispecific antibody, and as such produces a single asymmetric multispecific antibody species. In a further embodiment, combinatorial libraries of cells, each expressing a single asymmetric multispecific antibody, can be produced by alternative delivery strategies. In one non-limiting example, two vector collections are simultaneously delivered to a cell that contains an autocrine reporter system, where each vector comprises an expression cassette encoding a specific half of an asymmetric multispecific antibody, in a site-specific manner such that each cell receives one copy of each half of an asymmetrical multispecific antibody and produces an intact asymmetric multispecific antibody. In some embodiments, site-specific integration into host cell genomes is facilitated by a sitespecific nuclease such as, but not limited to, a transcription activator-like effector nuclease, a zinc finger nuclease, or a CRISPR-based nuclease. As a non-limiting example, a CRISPR-based nuclease, such as Cas9, is expressed in the host cell with two or more guide RNAs specific to two locations within the host cell genome. These locations may be natural safe-harbor sites, known to allow for efficient integration and expression of recombinant proteins, or may be located on a landing-pad that has been previously engineered into the host cell genome. Each vector collection may contain flanking homology arms specific to one of the two sites, such that when co-delivered, the nuclease facilitates the integration of a vector from each collection, at each integration site. The resulting cell expresses two halves of an asymmetric multispecific antibody and produces an intact asymmetric multispecific antibody. In another embodiment, site-specific integration into host cell genomes is facilitated by site-specific integrases and / or recombinases. As a non-limiting example, each vector collection contains one or more unique integrase / recombinase attachment sites, so that when the vectors are co-delivered into a cell producing the corresponding integrase / recombinase, a single copy of each vector is integrated into the host cell genome and the cell produces an intact asymmetric multispecific antibody composed of two different halves. These integrations events may occur at a two different, natural, sites within the host cell genome, or at a landing-pad that has been previously engineered into the host,

[0113]

[0107] In another embodiment, a single vector, containing an expression cassette which encodes two asymmetric antibody halves and produces an intact asymmetric multispecific antibody is delivered to cells to create a library of cells that each express a single multispecific antibody. In other embodiments, the compositions and methods can be applied similarly to symmetric antibodies, and a single vector, containing an expression cassette encoding a multispecific antibody in a symmetric format is delivered to cells to create a library of cells that each express a single symmetric multispecific antibody. Methods for screening a cell library for functional multispecific antibodies

[0114]

[0108] The disclosure provides methods for screening libraries of cells, expressing multispecific antibodies, using a reporter system.

[0115]

[0109] The disclosure also provides methods using an autocrine reporter system to identify functional multispecific antibodies. In some embodiments, when screening cell libraries to identify functional antibodies, various methods are used to prevent false positives arising from antibodies secreted by one reporter cell binding to an adjacent reporter cell in a paracrine manner. These methods can include, but are not limited to, mixing the cell library with decoy cells, in suspension, that express the target antigen, or similarly with antigen coated beads, such that they absorb excess antibody within the assay, or physically separating individual cells using microfluidic or microencapsulation, including stationary and flowed droplet-based systems and microwell systems as known to the art (Broketa and Bruhns, 2022) [PMID: 35173713], such that the antibodies secreted by said cells do not interact in a paracrine manner.

[0116] [HO] In some embodiments, cells identified to be expressing a functional asymmetric multispecific antibody, by activation of the autocrine reporter system, are isolated using methods such as, but not limited to, fluorescence-activated cell sorting, antibiotic selection, magnetic-bead selection, or a combination thereof. Activated cells could be isolated as single cells, or optionally, as a bulk population. In another embodiment, the asymmetric multispecific antibodies responsible for activating the autocrine reporter system will be identified. As a non-limiting example, in cases in which reporter cells are isolated as single cells, the resulting isolated cells, can optionally be clonally expanded, and can be sequenced in such a manner as to preserve pairing of the asymmetric halves using technique such as, but not limited to, DNA barcoding during the preparation of next-generation sequencing libraries, or sanger sequencing. As a further non-limiting example, in cases where the activated reporter cells have been isolated as a bulk population, the resulting cells can be subjected to singlecell sequencing to preserve and identify pairing of the asymmetric halves.

[0117] [Hl] The disclosure further provides methods for using a paracrine reporter system to identify functional asymmetric multispecific antibodies. In some embodiments, a library of cells producing asymmetric multispecific antibodies are physically separated as individual cells, using a microfluidic or microencapsulation approach, and contacted with one or more reporter cells containing a reporter system. Activation of the reporter system produces a detectable signal, allowing for the identification of library cells that produce a functional antibody.

[0118]

[0112] In a further embodiment, identified asymmetric multispecific antibodies are recombinantly produced and characterized. In another embodiment, identified asymmetric multispecific antibodies are further modified by engineering and / or humanization. EXAMPLES

[0119] EXAMPLE 1 : Establishment of linear transfer vectors (LTR-LECs)

[0120]

[0113] In some embodiments of the present disclosure, we will demonstrate that linear expression cassettes (LECs) designed for antibody production can be modified to create linear transfer vectors (LTR-LECs) that can be used for rapidly producing LEC-containing lentiviral particles that stably integrate LECs in a host cell genome.

[0121]

[0114] Linear transfer vectors (LTR-LECs) that combine the advantages of antibody producing LECs with the ability to create lentiviral particles are produced as follows. Standard techniques known in the art are used to append flanking lentiviral cis-acting regulatory elements (such as domain, the central polypurine tract (cPPT), the Rev responsive element (RRE), and the 5’ and 3’ Long Terminal Repeats (LTRs)) to a LEC that contains suitable regulatory element(s) operatively linked to antibody coding sequences and optionally a reporter protein sequence. LTR-LECs are co-transfected with lentiviral packaging plasmids into a standard lentiviral production cell line such as HEK293T. Resulting lentiviral particles are harvested, quantified by viral titration, and used to infect a cell line such as ExpiCHO or Expi293. The transduced cells will express the antibody and optionally the reporter, which is detected using flow cytometry, ELISA, high-content imaging, or other standard techniques. LEC lacking lentiviral cis-acting regulatory elements are shown to be unable to produce functional viral particles and are unable to transduce cells.

[0122] EXAMPLE 2A: Creating a lentiviral library of asymmetric multispecific antibodies

[0123]

[0115] Creating large combinatorial lentiviral libraries of asymmetric multispecific antibodies.

[0124]

[0116] The DNA sequences of paratopes (paired heavy and light chain antibody variable regions) from at least 10, 50, 75, 100, 250, 500, 1000, or up to 104antibodies, specific for a given antigen or eliciting a functional response, are generated through either PCR amplification from an isolated B-cell or through DNA synthesis of known antigen binders. Standard molecular biology techniques are used to append flanking assembly sequences onto the paratopes sequences. Optionally, paratopes may be synthesized with the appropriate assembly sequences. Standard techniques known in the art are used to individually assemble paratopes into lentiviral expression cassettes, such that a given paratope will be formatted as a Fab and as a scFv, in both light-heavy and heavy-light conformations. Expression cassettes are designed to each encode for one of the two halves of an asymmetric multispecific antibody and are operatively linked to different reporter proteins, such as a green or red fluorescent protein, and / or different selection makers, such as a puromycin or hygromycin resistance marker. The coding sequences of the antibody constant domains contain mutations that promote the heterodimeric assembly of an asymmetric multispecific antibody, such as electrostatic steering or Knob-in-Hole, to enable proper heterodimeric Fc chain pairing.

[0125]

[0117] The lentiviral transfer vector containing the expression cassettes is an LTR-LEC or a circular transfer plasmid. The individually assembled transfer vectors are pooled into two populations with each containing expression cassettes encoding for one of the two halves of an asymmetric multispecific antibody. For LTR-LECs, pooled transfer vectors are optionally further amplified by PCR. For expression cassettes contained within circular transfer plasmids, pooled transfer vectors are transformed into E. coli and amplified on solid agar. The use of circular plasmids exemplifies: (1) Colony counts post-transformation enumerates the size of the libraries and demonstrates the libraries are of sufficient size (i.e. at least 10-, 50-, 75-, 100-, 250-, 500- or 1000-times coverage of the theoretical library; and (2), sequencing of select colonies is used to sample the composition and quality of the vector library.

[0126]

[0118] Transfer vector collections, each corresponding to one of the two halves of an asymmetric multispecific antibody, are separately co-transfected with lentiviral packaging plasmids into a standard lentiviral production cell lines known in the art (e.g. HEK293T). The resulting collections of lentiviral particles are individually harvested and quantified by viral titration to demonstrate maintenance of at least 10-, 100-, or 1000-times coverage of the theoretical library.

[0127] EXAMPLE 2B: Creating a large combinatorial cell library of asymmetric multispecific antibodies

[0128]

[0119] The lentiviral collections prepared in Example 2A are used to create a large combinatorial cell library in which each cell has been delivered a single copy of each arm of an asymmetric multispecific antibody.

[0129]

[0120] A cell population numbering at least 106, 107, 108, 109, IO10, or more cells, which alternatively, stably expresses an autocrine reporter system, is transduced with the first lentiviral collection, encoding one of the two halves of an asymmetric multispecific antibody, at an MOI of 0.05, 0.1, 0.2, 0.3, 0.4, or 0.5, such that most cells are subjected to a single viral integration event. Cells are allowed to recover from transduction and are then purified by fluorescence-activated cell sorting or antibiotic selection. Enough cells are recovered to ensure that at least 10-, 100-, or 1000- times coverage of a single-arm library is maintained.

[0130]

[0121] Recovered cells are expanded, and a cell population numbering at least 106, 107, 108, 109, 1010, or more cells are transduced with a second lentiviral collection, encoding for the second corresponding half of an asymmetric multispecific antibody. Cells are transduced at an MOI of 0.05, 0.1, 0.2, 0.3, 0.4, or 0.5, such that most cells receive a single viral integration event. Enough cells are transduced to maintain 10-, 100-, or 1000-times coverage of the theoretical size of the combinatorial library created by stepwise transduction of two arms of an asymmetric multispecific antibodies. Cells are allowed to recover from transduction and are then purified by fluorescence-activated cell sorting or antibiotic selection. Enough cells are recovered to maintain at least 10-, 100-, or 1000-times coverage of the combinatorial library.

[0131]

[0122] The creation of a large combinatorial cell library is exemplified by interrogating the cell library via sequencing. Next- generation sequencing of bulk amplicons is used to show the cell library has captured 60%, 80%, 90%, 95%, or more of the possible asymmetric arms from the designed library. Optionally, a single-cell sequencing approach is used to generate paired-reads for both asymmetric arms to demonstrate the combinatorial library creation approach can generate 60%, 80%, 90%, 95%, or more of the possible combinations of asymmetric multispecific antibodies in the designed library.

[0132] EXAMPLE 3: Screening a cell library to identify functional asymmetric multispecific antibodies using an autocrine reporter assay

[0133]

[0123] A library of cells producing combinatorial libraries of asymmetric multispecific antibodies is screened for functional activity using an ultra-high-throughput autocrine assay as follows.

[0134]

[0124] The cells used to prepare the combinatorial cell library in Example 2B are engineered prior to library creation to stably express an autocrine reporter system. Alternatively, the autocrine reporter system can be added to host cells following library creation. In either case, an autocrine reporter system is assembled in a host cell, to create a reporter cell, by introducing (i) one, two, three, or more nucleic acid sequence(s) that are operatively linked to gene regulatory elements, such that they encode a target antigen(s), and / or (ii) one, two, three or more nucleic acid sequence(s), operatively linked to gene regulatory elements, encoding an autocrine reporter module. The autocrine reporter system is designed such that activation identifies the reporter cell as producing an asymmetric multispecific antibody with functional activity. Alternatively, the autocrine reporter system is inducible. Nucleic acids for creating the autocrine reporter system are delivered either through plasmids or lentiviruses.

[0135]

[0125] Reporter cells containing the asymmetric multispecific antibody library are interrogated in an autocrine assay. Library cells are expanded, or thawed and expanded, to obtain a cell population of at least 107, 108, 109, 1010, or more cells to guarantee 10-, 100-, 1000-times, or more coverage of the theoretical size of the combinatorial library. To prevent false positives arising from antibodies secreted by one reporter cell bind to an adjacent reporter cell in a paracrine manner, expanded reporter cells are transferred to fresh media and optionally mixed with decoy cells expressing target antigen. Decoy cells are optionally stained with various reagents or engineered to allow for identification separate from the reporter cells. If an autocrine reporter system is operatively linked to an inducible promoter, the induction condition is applied to allow the reporter system to be expressed. Alternatively, reporter cells are microencapsulated as individual cells such that the antibody secreted by one cell stays within the microencapsulated environment. Additional reagents or components needed in the autocrine assay are optionally added at this time. Reporter cells are incubated under assay conditions for up to 1,2, 4, 8, 12, 16, 24, 48, or more than 48 hours. After incubation the decoy cells are removed. Reporter cells are then further processed as needed to allow for assay read-out, which may include, but is not limited to, staining with various reagents, the addition of enzymatic substrates, and incubating cells as needed. Once processed, reporter cells are subjected to fluorescence-activated cell sorting to isolate cells in which the autocrine reporter system is activated. Activated reporter cells are isolated as single cells, or alternatively, a bulk population.

[0136]

[0126] The asymmetric multispecific antibodies responsible for activating the autocrine reporter system are identified. In cases in which reporter cells are isolated as single cells, the resulting isolated cells may be clonally expanded and sequenced in such a manner as to preserve pairing of the asymmetric halves. Isolated clones are lysed and subjected to PCR to amplify the nucleic acids that encode each half of the asymmetric multispecific antibody. The resulting amplicons are then sequenced using sanger or next-generation sequencing techniques. Alternatively, activated reporter cells isolated as a bulk population are subjected to single-cell sequencing to preserve and identify pairing of the asymmetric halves.

[0137] EXAMPLE 4: Biochemical and functional characterization of asymmetric multispecific antibodies identified using an autocrine reporter assay

[0138]

[0127] The asymmetric multispecific antibodies identified in Example 3 are recombinantly produced and characterized to demonstrate that the antibodies identified through an autocrine reporter assay are functionally active.

[0139]

[0128] Reporter cells isolated as single cells and / or then clonally expanded may produce, recombinant asymmetric multispecific antibody for further characterization. Alternatively, nucleic acids encoding for identified asymmetric multispecific antibodies are synthesized, assembled into expression cassettes, and transfected or transduced into an antibody production cell line, such as ExpiCHO or Expi293. Antibody producing cells are grown for at least 1, 2, 3, 5, or 7 days, and the resulting culture supernatant is collected and clarified. Antibodies present in the supernatant are purified using techniques such as Protein A affinity capture. Purified antibodies are quantified and then analyzed by standard techniques known in the art (e.g. non-reducing microchip capillary electrophoresis, analytical size-exclusion chromatography, liquid chromatography-mass spectrometry, or analytical cation exchange chromatography).

[0129] Candidate asymmetric multispecific antibodies are validated using a series of assays. Functional activity is initially characterized using a functional assay analogous to that used in the autocrine reporter system. Functional activity is further validated using a series of orthogonal biological assays. Alternatively, the binding affinity and / or avidity of the purified asymmetric multispecific antibodies to one or more target antigens is characterized.

[0140] EXAMPLE 5: Creation of reporter cells containing an autocrine reporter system for detecting the agonism of PD1

[0141]

[0130] A reporter cell line containing an autocrine reporter system for detecting the agonism of PD1 was produced as follows.

[0142]

[0131] Jurkat E6.1 cells were transduced with a lentiviral vector containing a green fluorescent protein (GFP) operatively linked to a minimal promoter containing three NFAT response elements, such that activation of NFAT signaling results in the production of GFP, and a puromycin selection marker. Three days post-transduction, transduced Jurkat E6.1 cells were subjected to six days of drug selection using 0.5 pg / mL of puromycin. After selection, the resulting cells were treated for 16 hours with 10 pg / mL of plate-bound anti-CD3 (OKT3, BioLegend, Cat. No. 317326) after which FACS was used to isolate transduced Jurkat E6.1 cells that produced GFP in response to NFAT signaling, as driven by anti-CD3 stimulation. Isolated cells were allowed to recover for nine days before being subjected to a second round of FACS to isolate transduced Jurkat E6.1 cells that did not produce GFP in the absence of stimulation. The isolated cells (Jurkat-NFAT.GFP hereafter) were allowed to recover and expand, before being validated for the ability to produce GFP in response to NFAT signaling. In brief, Jurkat-NFAT.GFP cells were incubated for 16 hours with either 10 pg / mL of plate-bound anti- CD3 or without any stimulation. As depicted in Figure 4, analysis by flow cytometry determined that 36.10% percent of Jurkat-NFAT.GFP cells produced GFP upon stimulation with anti-CD3, while in the absence of stimulation, only 1.08% of cells produced any GFP.

[0143]

[0132] To create an autocrine reporter system for detecting PD1 agonism, Jurkat-NFAT.GFP cells were transduced with a lentiviral vector containing a hybrid PD1.CD3 target antigen operatively linked to a constitutive promoter, and a hygromycin selection marker. The hybrid PD1.CD3 target antigen is designed such that the extracellular domain (ECD) and the transmembrane domain (TMD) of PD1 is fused to the intracellular signaling domain (ICD) of CD3 zeta in such a way that agonism of PD1 will result in activation of immunostimulatory pathways that signal through NFAT, rather than the recruitment of immunosuppressive phosphatases that are normally associated with the wild-type ICD of PD1. Converting an immunosuppressive signal to an immunostimulatory signal using this hybrid target antigen allows PD1 agonism to be detected using Jurkat-NFAT.GFP cells. Three days post-transduction, transduced Jurkat-NFAT.GFP cells were subjected to six days of drug selection using 800 pg / mL of hygromycin. After selection, the cells were stained with a conjugated anti-PDl monoclonal antibody (EH12.2H7, BioLegend, Cat. No. 329908) and subjected to FACS to isolate cells that strongly expressed the hybrid PD1.CD3 target antigen. Isolated cells (Jurkat-NFAT.GFP- PD1.CD3 hereafter) were allowed to recover and were then analyzed by flow cytometry for the expression of the target antigen. As depicted in Figure 5, the Jurkat-NFAT.GFP-PD1.CD3 cells strongly express the hybrid PD1.CD3 target antigen (99.6% of the population), compared to the Jurkat-NFAT.GFP cells which express only low levels of wild-type PD1 (6.33% of the population).

[0144]

[0133] Jurkat-NFAT.GFP-PD1.CD3 cells were validated for the ability to report on PD1 agonism by co-culture with U2OS cells overexpressing PDL1, the ligand for PD1 (U2OS-PDL1 hereafter).

[0145]

[0134] To agonize PD1, Jurkat-NFAT.GFP-PD1.CD3 cells or Jurkat-NFAT.GFP cells were cultured for 16 hours under three different conditions: 1) Cultured alone; 2) Co-cultured in the presence of U2OS-PDL1 at a 1:1 ratio of Jurkat to U2OS cells; and 3) Co-cultured in the presence of U2OS-PDL1 at a 1 : 1 ratio of Jurkat to U2OS cells with the addition of 5 pg / mL of Atezolizumab, a monoclonal antibody that disrupts the interaction of PD1 and PDL1. After 16 hours, Jurkat cells were analyzed for GFP expression using flow cytometry. As depicted in Figure 6, Jurkat-NFAT.GFP-PD1.CD3 cells strongly expressed GFP when co-cultured with U2OS-PDL1 cells, which could be abrogated by the addition of Atezolizumab, demonstrating that the PDL1 -derived agonism of PD1 is driving the expression of GFP. In contrast, Jurkat-NFAT.GFP did not express GFP under any of the tested conditions, further demonstrating that the ability of Jurkat-NFAT.GFP-PD1.CD3 to report on PD1 agonism is a direct result of the hybrid target antigen.

[0146] EXAMPLE 6: Construction of a large combinatorial cell library of biparatopic anti-PDl antibodies

[0147]

[0135] A large combinatorial cell library of biparatopic anti-PDl antibodies was produced as follows.

[0148]

[0136] Antibodies with specificity for PD1 were identified from single B cells using linear expression cassettes as described by Kurosawa et al (2012) [PMID: 23017270] and Zang et al (2021) [PMID: 33899674], In brief, FACS was used to isolate B-cells with specificity for PD1 from the splenocytes of PD1 -immunized mice. Isolated B-cells were then sorted as single cells and lysed to release mRNA. The resulting mRNA was converted to cDNA using a reverse transcriptase reaction and then used as a template for two-step nested PCR to amplify antibody variable gene regions. For amplification of the heavy chain variable region, the first round of PCR used forward primers specific to IGH leader sequences and reverse primers specific to IGHG CHI domains. For amplification of the kappa chain variable region, the first round of PCR used forward primers specific to IGK leader sequences and reverse primers specific to IGK constant domain. DNA resulting from the first round of PCR was used as template for their respective second round of PCR. For both the heavy and kappa variable regions, the forward and reverse primers in the second round PCR were specific to the 5’ and 3 ’ ends of the variable genes and contained overhang sequences homologous to the upstream and downstream DNA fragments used to construct their respective linear expression cassettes. Amplified variable genes were purified, quantified, and then used in a fusion-PCR reaction to construct linear expression cassettes. Fusion-PCR reactions were assembled using three fragments of DNA, an upstream fragment containing a promoter and a signal peptide, a variable gene, and a downstream fragment containing a termination signal and the respective constant regions of either the heavy or kappa chain, as well as a forward primer specific to the 5 ’ end of the upstream fragment and a reverse primer specific to the 3’ end of the downstream fragment. Amplified linear expression cassettes for matching heavy and kappa chains were purified, quantified, and transfected into ExpiCHO cells for antibody expression.

[0149]

[0137] Antibody containing supernatants were harvested after 7 days and used to test antibody specificity via on-cell binding to PD1. Expi293 target cells were transfected with vectors expressing either human PD1 or an irrelevant fluorescent protein and allowed to recover for 24 hours. Transfected target cells were then incubated with antibody containing supernatants for 30 minutes, washed, and subsequently stained with 2 pg / mL of a polyclonal fluorophore-conjugated secondary reagent specific to human Fc-domain (Jackson ImmunoResearch Laboratories, Cat. No. 109-605- 170). Flow cytometry was then used to assess antibody binding by comparing the ability of an antibody to bind to target cells expressing PD1 versus binding to target cells expressing the irrelevant fluorescent protein. Binders, as selected based on >2 -fold binding-over-background, were subjected to targeted NGS amplicon sequencing, as described by Chen et al, 2018 [PMID: 29357282], to identify the nucleic acid and amino acid sequences of the antibody variable regions.

[0150]

[0138] The paratopes of a diverse panel of 144 unique PD1 binders were synthesized and assembled into lentiviral transfer vectors such that each vector encodes for half of an anti-PDl antibody operatively linked to a reporter protein, as exemplified in Figure 1 and Figure 3, that can be assembled into an asymmetric biparatopic antibody. Specifically, each paratope was synthesized in two different formats: As a Fab, composed of paired VH and VL nucleic acid sequences; and as an scFv, a single nucleic acid sequence in which the paratope variable regions are joined in VL-to-VH orientation by a glycine-serine linker. Formatted paratopes were synthesized with appropriate assembly sequences to allow for ordered Golden Gate assembly into lentiviral transfer vectors. For the Fab format, the assembled lentiviral transfer vectors contained an antibody expression cassette designed to encode for one of the two halves of an anti-PDl antibody that can be assembled into an asymmetric biparatopic antibody; comprising a nucleic acid sequence encoding a Fab light chain and a Fab heavy chain, the CH3 domain of which contains mutations that promote the heterodimeric assembly of an asymmetric multispecific antibody, all operatively linked to mTagBFP, a blue fluorescent protein. For the scFv format, the assembled lentiviral transfer vectors contained an antibody expression cassette designed to encode for the other of the two halves of an anti-PDl antibody that can be assembled into an asymmetric biparatopic antibody; comprising a nucleic acid sequence encoding an scFv-Fc fusion protein, the CH3 domain of which contains mutations that promote the heterodimeric assembly of an asymmetric multispecific antibody, operatively linked to LSSmOrange, an orange fluorescent protein. All expression cassettes contained gene regulatory elements for expressing polypeptides, as well as signal peptides that allowed encoded antibodies to be trafficked to a host cell’s secretory system. Using standard molecular biology techniques, each paratope, in each format, was individually assembled into their respective lentiviral transfer vector.

[0151]

[0139] The individually assembled transfer vectors were pooled into two populations, each containing expression cassettes encoding for one of the two halves of an asymmetric biparatopic anti- PD1 antibody. One population (the first population) contains mutations that promote heterodimeric chain pairing, and the other population (the second population) contains matching mutations that promote heterodimeric chain pairing, such that any of the antibody halves from the first population can pair with any of the antibody halves from the second population. For example, the first population are the scFv-encoding vectors, and the second population are the Fab-encoding vectors. Pooled transfer vectors were then transformed into E. coli and amplified overnight on solid agar. As shown in Table 1, enumeration by colony counts demonstrated that the libraries were sufficiently sized to achieve at least 1000-times coverage of the theoretical library size, while Sanger sequencing demonstrated that 23 out of 24 colonies were correctly assembled. The resulting E. coli biomass was harvested, and plasmid DNA was purified. The resulting plasmid DNA was subjected to targeted NGS amplicon sequencing to interrogate the paratope coverage of each plasmid pool, demonstrating that 143 out of 144 paratopes were present in the scFv pool, while in the Fab pool 144 out of 144 paratopes were present.

[0152]

[0140] Pooled transfer vectors, each corresponding to one of the two halves of an asymmetric multispecific antibody, were separately co-transfected with lentiviral packaging plasmids into standard lentiviral production cell lines. The resulting lentiviral particles were harvested and quantified by functional viral titering. As shown in Table 1, viral particle production maintained greater than >100,000-times coverage of the theoretical library.

[0153] Table 1. Determination of library coverage during construction

[0154] Construction step Theoretical Recorded Fold-coverage of theoretical

[0155] Molecular cloning of library scFv arm viral vectors 144 228,000 ± 63,000 Colonies 1580 ± 440

[0156] Fab arm viral vectors 144 195,000 ± 77,000 Colonies 1350 ± 530

[0157] Generation of lentiviral particles scFv arm viral particles 144 3.1 ± 0.5 x 107Infectious units >100,000

[0158] Fab arm viral particles 144 8.7 ± 0.3 x 107Infectious units >100,000

[0159] Creation of reporter cell library

[0160] Transduction of scFv arm 144 4.5 x 106Cells recovered >20,000

[0161] Transduction of Fab arm 20,736 4.2 x 106Cells recovered >200

[0162] Final library 20,736 >200

[0163]

[0141] A sequential transduction protocol was used to create a large combinatorial cell library of asymmetric biparatopic anti-PDl antibodies (Figure 7). Reporter cells, generated in Example 5, were first transduced with lentiviral particles encoding for the scFv half of the biparatopic antibody. Specifically, 10 million Jurkat-NFAT.GFP-PD1.CD3 cells were transduced with diluted lentiviral particles at an MOI of 0.3 to ensure that most cells were subjected to a single viral integration event while allowing for enough transduction events to provide coverage of >20,000-times the theoretical library size. After overnight incubation, transduced cells were washed to remove any residual viral particles and suspended in fresh media. Transduced cells were then allowed to expand over 6 days, before being subjected to FACS to isolate cells positive for the expression of LSSmOrange. After isolation, 4.5 million live cells (Jurkat-NFAT.GFP-PDl.CD3-scFv hereafter) were recovered, maintaining >20,000-times the theoretical library size (Table 1). The Jurkat-NFAT.GFP-PD1.CD3- scFv cells were then expanded for 7 days. Prior to the second round of transduction, expanded Jurkat- NFAT.GFP-PDl.CD3-scFv cells were analyzed by flow cytometry, demonstrating that 85.3% of the cells were positive for LSSmOrange expression, and therefore expressed the scFv half of a biparatopic antibody. Jurkat-NFAT.GFP-PDl.CD3-scFv cells were then transduced with lentiviral particles encoding for the Fab half of the biparatopic antibody. As described above, 20 million Jurkat- NFAT.GFP-PDl.CD3-scFv cells were transduced with diluted lentiviral particles at an MOI of 0.3 to ensure that most cells were subjected to a single viral integration event while allowing for enough transduction events to provide coverage of >200-times the theoretical library size. Cells were transduced, washed, and allowed to expand as described. The transduced cells were then subjected to FACS to isolate cells positive for both LSSmOrange and mTagBFP expression. After isolation, 4.2 million live cells (Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab hereafter) were recovered, maintaining >200-times the theoretical library size (Table 1). The Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab cells were expanded, and the resulting cell library was analyzed by flow cytometry, demonstrating that 92.7% of the library cells were positive for both LSSmOrange and mTagBFP expression, and therefore expressed both the scFv and Fab halves of a biparatopic antibody. To assess the quality of the library, genomic DNA was extracted from 5 million cells and subjected to targeted NGS amplicon sequencing. Sequencing detected 143 out of 144 of the scFv paratopes and 140 out of 144 of the Fab paratopes within the combinatorial cell library, demonstrating that this method can generate large high-quality combinatorial cell libraries containing at least 98.3% of expected paratopes.

[0164] EXAMPLE 7: Screening a large combinatorial cell library of biparatopic anti-PDl antibodies for functional activity

[0165]

[0142] A large combinatorial cell library of biparatopic anti-PDl antibodies was screened for functional activity using ultra-high-throughput autocrine assays.

[0166]

[0143] The biparatopic anti-PDl antibody library contained within the Jurkat-NFAT.GFP-PD1.CD3- scFv.Fab cells secrete antibodies able to bind and functionally agonize PD1. Specifically, 3 million Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library cells were seeded in cell media at a density of 0.15 million cells / mL and allowed to expand for 24 to 48 hours, at which point the culture supernatant was harvested. To assess binding to PD1, supernatant was incubated with Jurkat target cells stably overexpressing wild-type PD1 (Jurkat-PDl hereafter) for 30 minutes, washed, and subsequently stained with 2 pg / mL of a polyclonal fluorophore-conjugated secondary reagent specific to human Fc- domain (Jackson ImmunoResearch Laboratories, Cat. No. 109-605-170). Flow cytometry was then used to assess the PD1 binding ability of antibodies contained within the supernatant. As shown in Figure 8, supernatant from the Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab cell library bound Jurkat-PDl target cells to a 1980-fold greater extent than supernatant from control Jurkat-NFAT.GFP-PD1.CD3 reporter cells that were not transduced with vector encoding any anti-PDl antibody, and 800-fold greater than 2 pg / mL of Palivizumab, an isotype control antibody. To assess functional agonism of PD1, freshly prepared Jurkat-NFAT.GFP-PD1.CD3 reporter cells were incubated for 16 hours with supernatants, then analyzed for GFP positivity using flow cytometry. As shown in Figure 9, supernatant from the Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab cell library demonstrated the ability to agonize PD1 and stimulate NFAT signaling through CD3 ICD, increasing GFP positivity by 4.5-fold compared to supernatant from control Jurkat-NFAT.GFP-PD1.CD3 reporter cells that were not transduced with vector encoding any anti-PDl antibody. Together, these results establish that cells within the Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library are secreting antibodies able to both bind and functionally activate PD1. This proves that the methods established, and provided herein, can produce combinatorial cell libraries of multispecific antibodies that can be mined for rare functional antibodies using ultra-high-throughput autocrine assays.

[0167]

[0144] An ultra-high-throughput autocrine assay was used to isolate clones secreting functional biparatopic anti-PDl antibodies. Jurkat-NFAT.GFP-PDl.CD3-scFv.Fab library cells were expanded, 60 million cells were then collected and washed twice with phosphate-buffered saline to remove excess secreted antibodies that could lead to false positives due to paracrine binding, washed cells were transferred to fresh media and incubated for 16 hours. FACS was then used to isolate individual library cells in which the autocrine reporter system, designed to detect agonism of PD1, was activated and therefore identified the reporter cell as putatively producing an asymmetric biparatopic antibody with functional activity towards PD1. Activated cells expressing both scFv and Fab halves of a biparatopic antibody were selected, e.g., through positivity for both LSSmOrange and mTagBFP expression, as well as GFP expression 90 standard deviations above the mean GFP expression of Jurkat-NFAT.GFP-PD1.CD3 reporter cells. From this activated cell population, comprising approximately 0.03% of the overall population of the cell library, single cells were sorted into 96-well plates and allowed to expand for 2 weeks to create a series of clonal cell populations each derived from a different activated cell.

[0168]

[0145] Prior to sorting individual single cells, an enriched population of activated cells can optionally be generated through sequential rounds of isolation using an ultra-high-throughput autocrine assay. Library cells, prepared as described above, were subjected to FACS to identify and isolate a pool of library cells in which the autocrine reporter system was activated. Activated cells expressing both scFv and Fab halves of a biparatopic antibody were selected, e.g., through positivity for both LSSmOrange and mTagBFP expression, as well as GFP expression 16 standard deviations above the mean GFP expression of Jurkat-NFAT.GFP-PD1.CD3 reporter cells. In Figure 10, activated cells were isolated and allowed to expand for 1 week before being analyzed by flow cytometry. As illustrated, the method described herein resulted in a population of cells enriched 5.7-fold for reporter activity when compared to the unenriched population, increasing the percentage of activated cells from 0.2% to 1.13%. Enriched populations of cells can be subsequently sorted as single cells to isolate clones secreting functional biparatopic anti-PDl antibodies. As needed, multiple rounds of enrichment could be used to mine a combinatorial cell library for rare functional activity.

[0169]

[0146] Clonal cell populations producing putative functional biparatopic anti-PDl antibodies were characterized using a series of assays. Flow cytometry was used to characterize clonal populations and assess their ability to activate the autocrine reporter system, exemplified in Figure 11. This analysis demonstrated that clonal cell populations, isolated and recovered using the methods described herein, can express both scFv and Fab halves of biparatopic antibodies, indicated by LSSmOrange and mTagBFP expression, and vary in their ability to activate the autocrine reporter system. Illustrated in Figure 12, isolated clones exhibited GFP positivity 70- to 450-fold higher than Jurkat-NFAT.GFP- PD1.CD3 reporter cells, a level of activity that was greatly enriched when compared to the source library. Moreover, many of the clonal populations displayed greater GFP positivity than Jurkat- NFAT.GFP-PD1.CD3 cells treated for 16 hours with 2.5 pg / mL of GS-0151, a clinical PD1 agonist. This suggests that many of the antibodies produced by the isolated cells elicited a signal at least as good as, or even better than, the clinical PD1 agonist. Some were multiple-fold better, up to almost 3- fold better. Overall, these results demonstrate that a combinatorial cell library of multispecific antibodies combined with an ultra-high-throughput autocrine assay, as described herein, can effectively isolate cells enriched for functional activity and even identify cells that may produce antibodies demonstrating rare and useful activities.

[0170]

[0147] Isolated clonal populations secreted antibodies able to bind and functionally agonize PD1. Antibody containing supernatants were generated from isolated clones as described above. To assess binding to PD1, supernatants from clonal populations were incubated with Jurkat-PDl target cells for 30 minutes, washed, and subsequently stained with 2 pg / mL of a polyclonal fluorophore-conjugated secondary reagent specific to human Fc-domain (Jackson ImmunoResearch Laboratories, Cat. No. 109-605-170) and analyzed by flow cytometry to assess antibody binding. To assess functional agonism of PD1, freshly prepared Jurkat-NFAT.GFP-PD1.CD3 reporter cells were incubated for 16 hours with supernatants from clonal populations, then analyzed for GFP positivity using flow cytometry. As exemplified in Figure 13, Supernatants from isolated clones contained antibodies able to bind and functionally activate PD1 to varying degrees. Supernatant from Clone 1 showed 4600-fold higher binding and a 27-fold greater activation of PD1 compared to supernatant from control Jurkat- NFAT.GFP-PD1.CD3 reporter cells not transduced with vector encoding any anti-PDl antibody. In contrast, supernatant from Clone 2 bound PD1 strongly but only moderately activated PD1, while Clone 3 supernatant displayed both moderate binding and activation. As illustrated in Figure 14, supernatants from clonal populations demonstrated 436- to 4600-times greater binding and 2- to 27- times more functional activation of PD1 than supernatant produced by control Jurkat-NFAT.GFP- PD1.CD3 reporter cells not transduced with vector encoding any anti-PDl antibody. These results confirm that the cells isolated from the combinatorial cell library by an ultra-high-throughput autocrine assay are secreting antibodies able to bind and functionally activate PD1.

[0171]

[0148] The functional asymmetric biparatopic anti-PDl antibodies responsible for activating the autocrine reporter system are identified by sequencing. Genomic DNA can be extracted from expanded clonal populations. PCR, using a forward primer specific to the 5’ end of the lentiviral expression cassette and a reverse primer specific to the nucleic acid sequences encoding either LSSmOrange or mTagBFP, can then be used to separately amplify the lentiviral cassettes that encoded each half of the biparatopic antibody. Amplified DNA can then be sequenced using Sanger or NGS amplicon sequencing, which will identify the nucleic acid sequences of the antibody, while preserving pairing information of the asymmetric halves. Antibodies identified by these methods can then be recombinantly produced and further characterized.

[0172]

[0149] The example embodiments described herein and in the figures are intended to be non-limiting examples of embodiments, and those skilled in the art will recognize additional embodiments within the scope of the disclosure.

Claims

CLAIMS1. Alibrary of cells expressing antibodies comprising at least about 104cells that each comprise:(a) one or more nucleic acids encoding one or more antigen-binding domains of a first paratope that specifically bind a first epitope of a first target antigen,(b) one or more nucleic acids encoding one or more antigen-binding domains of a second paratope that specifically bind a second epitope, wherein the second epitope is on the same first target antigen or on a second target antigen, and(c) optionally, one or more nucleic acids encoding a reporter system; optionally wherein the antigen-binding domain(s) of the first paratope are not covalently linked as part of the same polypeptide with the antigen-binding domain(s) of the second paratope.

2. The library of claim 1 wherein the cells express an asymmetric multispecific antibody comprising the antigen-binding domain(s) of the first paratope and the antigen-binding domain(s) of the second paratope.

3. The library of any of claims 1-2 wherein the asymmetric multispecific antibody comprises one or more non-paratopic regions, optionally including a heterologous effector molecule.

4. The library of any of claims 1-3 further comprising one or more nucleic acids encoding one or more antigen-binding domains of a third paratope that specifically binds a third epitope, wherein the third epitope is on the first target antigen, or on the second target antigen, or on a different third target antigen.

5. The library of any of claims 1-4 wherein the asymmetric multispecific antibody comprises an Fc region with a first heterodimerization domain and an Fc region with a second heterodimerization domain.

6. The library of claim 5 wherein the first or second heterodimerization domain comprises a knob, a hole, a leucine zipper, a coiled coil, or a polar amino acid residue capable of forming an electrostatic interaction.

7. The library of claim 5 wherein the first heterodimerization domain is a knob structure and the second heterodimerization domain is a hole structure.

8. The library of any of claims 1-7 wherein the asymmetric multispecific antibody comprises one or more light chain constant regions (CL).

9. The library of claim 8 wherein the asymmetric multispecific antibody comprises a first CL with a first heterodimerization domain and optionally, a second CL with a second heterodimerization domain.

10. The library of any of claims 1-9 wherein the library comprises a first collection of nucleic acids encoding antigen-binding domains of at least 100 different paratopes and a second collection of nucleic acids encoding antigen-binding domains of at least 100 different paratopes, wherein the first and second collections comprise the same paratopes or the first and second collections comprise different paratopes.

11. The library of any of claims 1-10 that expresses at least 104, 105or 106different asymmetric multispecific antibodies.

12. The library of any of claims 1-11 wherein the cells are mammalian cells.

13. The library of any of claims 1-11 wherein the cells are bacterial cells or yeast or fungal cells or insect cells.

14. A method of producing the library of claim 1 comprising the steps of:(a) transfecting a library of cells with a first collection of delivery vectors comprising expression cassettes encoding antigen-binding domains of at least 100 different paratopes; and(b) transfecting the library resulting from step (a) with a second collection of delivery vectors comprising expression cassettes encoding antigen-binding domains of at least 100 different paratopes.

15. The method of claim 14 wherein the cells in step (a) express an autocrine reporter system, optionally wherein the cells in step (a) stably express an autocrine reporter system.

16. The method of claim 15 comprising transfecting the library of cells with one or more nucleic acids encoding the autocrine reporter system prior to step (a), or after step (b).

17. The method of any of claims 14-16 wherein the expression cassettes of step (a) and / or step (b) are in a viral vector, optionally a lentiviral vector.

18. The method of any of claims 14-17 wherein transfection results in at least 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85% or more %, preferably at least 70%, of the cells in the library expressing a multispecific antibody and an autocrine reporter system.

19. A method of using the library of any of claims 1-14 to screen for asymmetric multispecific antibodies comprising:(a) detecting when binding of an asymmetric multispecific antibody produced by a cell within the library to a first target antigen results in autocrine reporter system activation of said cell, wherein the autocrine reporter system activation identifies said cell as producing a functional multispecific antibody, and optionally(b) isolating said antibody -producing cell from the library.

20. Amethod of using the library of any of claims 1-14 to screen for functional multispecific antibodies comprising:(a) detecting when binding of an asymmetric multispecific antibody produced by a cell within the library to the first paratope and second paratope, on the same or different target antigens, results in reporter system activation, wherein the reporter system activation identifies said cell as producing a functional multispecific antibody, and optionally(b) isolating said antibody -producing cell from the library.

21. The method of claim 19 or 20, wherein the method is an ultra-high throughput antibody screening method that permits screening of at least about 104, 105, 106, 107, 108, 109to up to about IO10, or more cells, preferably at least about 105cells individual cells per day for producing functional asymmetric multispecific antibody.

22. The method of any of claims 19-21, wherein the autocrine system comprises a reporter protein that is a fluorescent or membrane-bound protein detectable by fluorescence activity cell sorting (FACS) or by bead-based positive selection.

23. The method of claim 19 or 20 wherein step (b) comprises FACS or bead-based selection.

24. The method of any of claims 19-23 further comprising reducing secreted free antibody in solution by adding decoy beads or decoy cells displaying target antigen, optionally comprising the step of removing the decoy beads or decoy cells.

25. The method of any of claims 19-23 wherein a cell expressing an autocrine reporter system and an asymmetric multispecific antibody are individually encapsulated or separated.

26. The method of any of claims 1-25 wherein a cell expressing a reporter system and a cell expressing an asymmetric multispecific antibody are encapsulated together.

27. The library of any of claims 1-13, the method of any of claims 14-18 or the method of any of claims 19-25, wherein the library comprises at least about 104, 105, 106, 107, 108, 109to up to about 1010, or more cells, preferably at least about 105cells, that express asymmetric multispecific antibody.

28. The library of any of claims 1-13, the method of any of claims 14-18 or the method of any of claims 19-25, wherein the reporter system comprises a reporter protein selected from the group consisting of: a fluorescent protein, a bioluminescent protein, an enzyme, a split enzyme, an engineered cell surface protein, a unique affinity tag, or combination thereof.

29. The library of any of claims 1-13, the method of any of claims 14-18 or the method of any of claims 19-25, wherein the target antigen is a soluble antigen attached to the cell with an anchor ora cell-surface receptor, optionally an enzyme-linked receptor, G-protein-coupled receptor (GPCRs), or ion channel-linked receptor.

30. The library of any of claims 1-13, the method of any of claims 14-18 or the method of any of claims 19-25, wherein the reporter system detects agonism, antagonism, internalization, or inverse agonism of the target antigen.

31. A linear lentiviral transfer vector comprising (a) one or more nucleic acids encoding antigenbinding domain(s) of a multispecific antibody, operatively linked to (b) a regulatory element suitable for expressing the multispecific antibody, (c) flanking 5’ and 3’ Long Terminal Repeats (LTRs), (d) a Rev responsive element (RRE), and (e) a lentiviral domain, (f) optionally, a central polypurine tract (cPPT), and (g) optionally, a woodchuck post-transcriptional regulatory element (WPRE).

32. The linear lentiviral transfer vector of claim 31, wherein the vector comprises nucleic acids encoding a light chain variable region and nucleic acids encoding a heavy chain variable region.

33. The linear lentiviral transfer vector of claim 31, wherein the vector further comprises nucleic acids encoding one or more non-paratopic regions, optionally including a heterologous effector molecule.

34. The linear lentiviral transfer vector of claim 31, wherein the vector comprises a woodchuck post- transcriptional regulatory element (WPRE).

35. The linear lentiviral transfer vector of claim, 31, wherein the vector comprises a central polypmine tract (cPPT).

36. The method of claim 19 comprising:(a) isolating from said library a first subset of antibody -producing cells that exhibit autocrine reporter system activation,(b) expanding said first subset of cells for about 1 week, followed by(c) isolating from said first subset of antibody -producing cells a second subset of antibody producing cells detected as exhibiting autocrine reporter system activation, wherein the percentage of antibody -producing cells that exhibit autocrine reporter system activation is enriched at least 5 -fold, followed by(d) isolating one or more single cells exhibiting autocrine reporter system activation, optionally repeated steps (a)-(d).

Images