Method for isolating an anti-ligand

Abstract

The present invention relates to an improved method for isolating an anti-ligand having a desired binding specificity, and more particularly, to a method for isolating an anti-ligand from an anti-ligand library that is specific for a ligand that is differentially and / or rarely expressed.

Description

Technical Field

[0001] The present invention relates to an improved method for isolating antiligands having a desired binding specificity, and more particularly to a method for isolating antiligands from an antiligand library that is specific for ligands that are differentially and / or rarely expressed.

[0002] Protein- or peptide-based libraries are often used for the selection of antiligand molecules having specificity for a particular ligand. Such libraries are constructed such that the protein molecules are physically linked in a manner to the genetic information encoding the particular protein molecule. Thus, the protein molecules are displayed together with their genes.

[0003] Commonly used display formats rely on cells or viral host particles to present the protein molecules, including bacterial display (Francisco et al., 1993) and phage display (Smith, 1985, Smith and Scott, 1993, Winter et al., 1994). Such systems display potential antiligand molecules on the surface of the host particles, while the genetic information about the presented molecules is retained inside the particles, and the method has been successfully used for the selection of specific protein-based antiligands.

[0004] There are other display formats that rely on in vitro translation, including various forms of ribosome display (Mattheakis et al., 1994; Hanes and Pluckthun, 1997; He and Taussig, 1997) that rely on non-covalent attachment of genetic information to the protein molecule, and other display formats that also rely on in vitro translation where a covalent bond exists between the genetic information and the potential antiligand protein molecule, such as Profusion (Weng et al., 2002) or Covalent Display Technology (Gao et al., 1997).

[0005] The displayed peptide or proteinaceous anti-ligand library can be completely randomized (e.g., if a peptide library is used), or they can be based on a constant region scaffold structure that incorporates additional structures conferring variability.

[0006] Scaffold structures that are often used are based on antibody heavy and light chain variable domains (McCafferty et al., 1990), but can also be based on fibronectin (Jacobsson and Frykberg, 1995, Koide et al., 1998), protein A domains (Stahl et al., 1989), or other scaffolds such as small stable protein domains, e.g., BPTI (Markland et al., 1991).

[0007] The selection of anti-ligands exhibiting a particular binding specificity from a display library is often performed using so-called "biopanning" methods.

[0008] The target ligand can be immobilized on a solid surface, and specific anti-ligand members of the library are exposed to the immobilized target ligand to allow the anti-ligand of interest to bind to the target ligand. Subsequently, unbound library members are washed away, and the anti-ligand of interest is recovered and amplified.

[0009] In another selection method, the target ligand binds to specific anti-ligand library members while in solution. The bound anti-ligand is then isolated, e.g., using a recoverable tag that binds to the target ligand. The most commonly used tag is biotin, which allows the complex between the target molecule and a specific library member displayed to be recovered using avidin bound to a solid support, e.g., magnetic beads (Parmley and Smith, 1988).

[0010] These methods are used when the target ligand is well-known and available in a purified form. Selection for a single target ligand at a time is routine. Selection for several defined target ligands can be done simultaneously. The target ligand can be one or more of a small hapten, protein, carbohydrate, DNA, and lipid.

[0011] Selection can also be done against ligands expressed on the surface of cells. This can be done using endogenous cell lines, primary cells, or cells overexpressing the ligand of interest.

[0012] Proteinaceous particles other than members of the anti-ligand library, e.g., phage expressing antibody fragments, can be "sticky" and result in the binding and isolation of some non-target-specific molecules. Nonspecific binding can be minimized by adding certain compounds, e.g., milk, bovine serum albumin, serum (human / fetal bovine), gelatin, and for certain (non-cellular) applications, surfactants, to the anti-ligand display construct / ligand mixture, as they act as blocking agents to reduce this background binding of non-specific anti-ligands.

[0013] Several washing procedures have been devised to reduce non-specific binding of library members to cells and to aid in the separation of cells from contaminating and / or non-specifically bound library members. Such methods include washing cells magnetically immobilized in a column (Siegel et al., 1997) to minimize shear forces and to allow for the re-binding of dissociated phage. Another method of washing cells is by centrifugation in a higher density medium such as Ficoll or Percoll (Carlsson et al., 1988, Williams and Sharon, 2002) to selectively remove non-specific and low affinity anti-ligands and to spatially separate cells and cell-bound anti-ligands from free anti-ligand and non-specifically bound anti-ligand.

[0014] Depending on the efficiency of the selection process, several rounds of panning may be required to eliminate non-specific anti-ligands or reduce them to a sufficiently low level, at least to the desired level (Dower et al., 1991).

[0015] For many applications, specific anti-ligands to differentially expressed ligands are of interest. For example, proteins can be differentially expressed on cells when compared to those from healthy controls, or can be present in different amounts in tissues and samples from patients with diseases. Such diseases include microbial, viral, or parasitic infections, asthma, chronic inflammation and autoimmune disorders, cancer, neurological diseases, cardiovascular diseases, or gastrointestinal diseases. Similarly, the protein composition of body fluids (e.g., plasma, cerebrospinal fluid, urine, semen, saliva, and mucus) can vary among patients with diseases compared to healthy controls.

[0016] Thus, in addition to their general applicability as research tools for identifying differentially expressed ligands, anti-ligands specific to differentially expressed ligands can be used as tools for use in the diagnosis, prevention, and / or treatment of diseases.

[0017] Advances in the fields of genomics and proteomics have indicated the existence of numerous yet undefined differentially expressed molecules, highlighting the importance of methods for generating specific anti-ligands to these potential target ligands. Many of these differentially expressed molecules are present on the cell surface and thereby are expected to constitute potential targets for targeted chemotherapy using, for example, specific antibodies that can be conjugated to bioactive (e.g., cytotoxic) agents.

[0018] Large, highly diversified anti-ligand display libraries provide a way to isolate anti-ligands that have specificity for unknown cell ligands that act as carbohydrates, proteins, lipids, or combinations thereof.

[0019] Currently available biopanning processes include whole-cell, cell-part, and cell-membrane-based methods, which in principle enable the isolation of display constructs presenting anti-ligands specific for cell-membrane ligands in their native conformation.

[0020] Human and humanized therapeutic antibodies are increasingly being used to treat a variety of diseases, including acute and chronic inflammatory disorders, immune and central nervous system disorders, and cancer. Human therapeutic antibodies are considered the most attractive modality for treating human diseases because of their fully human nature and lack of associated immunogenicity, their optimal ability to engage antibody Fc-dependent host immune effector mechanisms, and their superior in vivo half-life compared to their murine, chimeric, and humanized counterparts. Human antibodies are routinely generated today by different technologies, including humanized mice and highly diversified phage antibody libraries.

[0021] Human antibody libraries are also thought to offer advantages compared to transgenic mice carrying human immunoglobulin genes when selecting antibodies that bind to receptor epitopes that are structurally conserved between humans and mice. This is because antibodies in this category are negatively selected in vivo by the mechanisms of self-tolerance. Conserved regions are often functionally related (e.g., ligand-binding domains required for the binding and imparting of ligand / receptor-induced cell responses), and such conserved epitopes are of particular therapeutic interest because antibodies targeting such conserved regions can be screened for in vivo therapeutic activity in syngeneic experimental disease model systems.

[0022] Large binder libraries (>10 10Members typically include antibodies that bind with high (nM) affinity and selectivity to many clinically relevant target biomolecules (Hanes et al, 2000, Soderlind et al, 2000, Rothe et al, 2008). Thus, antibodies to diverse biomolecules can be isolated from such libraries by applying positive selection pressure for binding to a defined target biomolecule (e.g., a cell surface receptor) and negative selection pressure for binding to highly homologous non-target biomolecules (e.g., related receptors of the same superfamily) (Winter et al, 1994). Similarly, antibodies to empirically unknown biomolecules that are differentially expressed between a target sample and a non-target sample can be isolated from an antibody library by applying positive and negative selection pressures in the form of a complex biomolecule population (e.g., diseased cells versus normal cells, blood, or tissue).

[0023] When combined with clinically predicted high-throughput functional screening (Frendeus, 2013; Ljungars et al., 2018) and target deconvolution (Mattsson et al., 2021), this enables phenotypic drug discovery (PDD) of medically relevant antibodies, including first-in-class antibodies specific to new targets (Ljungars et al., 2018; Waldmann et al., 1984; Veitonmaki et al., 2013; Roghanian et al., 2015; Williams et al., 2016) and best-in-class antibodies specific to intrinsically functional epitopes on validated targets (Semmrich et al., 2022). The broad applicability of PDD to biologics is further demonstrated by multiple selection strategies that incorporate cells, tissues, and fluids to generate antibodies for PDD (Veitonmaki et al., 2013; Roghanian et al., 2015; Dyer et al., 1989). However, realizing the full potential of biological PDD (i.e., functional screening of antibodies against all disease-related biomolecules) will require significant improvement and output compared to current target-independent methods that have generated a small number (101 - 103) of antibodies specific to a limited number of highly expressed biomolecules (Ljungars et al., 2018; Williams et al., 2016; de Kruif et al., 1995; Ridgway et al., 1999; Qin et al., 2014; Egloff et al., 2019; Sandercock et al., 2015; Nixon et al., 2019).

[0024] Furthermore, therapeutic efficacy cannot be easily predicted from antibody receptor specificity, and antibodies against the same target receptor can vary widely in therapeutic efficacy regardless of their binding affinity (Beers et al., 2008; Cragg and Glennie, 2004), and antibodies against alternative molecular targets can exhibit promising and sometimes unexpected therapeutic potential (Beck et al., 2010; Cheson and Leonard, 2008). For example, different CD20-specific antibody clones that bind with similar affinity to the CD20 antigen and possess the same murine IgG2a constant region differ substantially in their ability to deplete B cells in vivo (Beers et al., 2008; Cragg and Glennie, 2004), and antibodies against other tumor-associated cell surface receptors rather than CD20 can have significant anti-cancer activity against B cell cancer (see Cheson and Leonard, 2008 for a review). Thus, in a highly diversified antibody library, the most therapeutically effective, potent, and resistant antibodies for any given type of cancer can be specific for any of several different receptors, and identifying the therapeutically optimal antibody clones in a highly diversified library requires functional screening of multiple, and ideally all, library members specific for different disease cell-associated receptors.

[0025] The present applicant has previously developed two different screening techniques (biopanning methods) that enable the activation of antibody clones that bind to different surface receptors differentially expressed on one cell population (target cells) compared to another cell population (non-target cells) from a human phage antibody library (hereinafter known as differential biopanning). The first of these methods was described in International Publication No. WO 2004 / 023140 (and also in Fransson et al., 2006, Frendeus 2006). The second of these methods was described in International Publication No. WO 2013 / 041643. The disclosures of International Publication No. WO 2004 / 023140 and WO 2013 / 041643 (and all domestic applications derived therefrom) are hereby incorporated by reference in their entirety.

[0026] The process of International Publication No. WO 2004 / 023140 included the following steps in the following order: 1) Differential biopanning, followed by 2) Screening for target vs. non-target specificity, followed by 3) Conventional sequencing of a smaller number of clones by the Sanger technique. [[ID=*]]

[0027] Using this technique, it was possible to generate a pool of antibodies that showed high specificity for differentially expressed surface receptors on target cells vs. non-target cells.

[0028] The process of International Publication No. WO 2013 / 041643 included the following steps in the following order: 1) Differential biopanning, followed by 2) High-throughput sequencing, followed by 3) Confirmatory screening for anti-ligand specificity against differentially expressed ligands.

[0029] Using the process of International Publication No. 2004 / 023140, Sanger sequencing is an example of a technique currently used to identify unique binders in a "low throughput" format. Other examples include electrophoresing antibody gene DNA on a gel before and after restriction enzyme digestion to reveal unique sizes and indirectly reveal different sequences through different sensitivities to different restriction enzymes.

[0030] When applied to the isolation of antibodies targeting differentially expressed surface receptors of cancer B cells (target) versus T cells (non-target) ("BnonT" differential biopanning), this process identified antibodies specific for differentially expressed surface receptors of different target cells, including HLA-DR, surface Ig, and ICAM-1 (Table 1).

[0031] Table 1. Frequency and specificity of antibodies isolated by existing screening methodologies, e.g., sequential differential biopanning, screening for binding, and Sanger sequencing, targeting of cancer B cells to differentially expressed T cell surface receptors ("BnonT" differential biopanning).

Table 1

[0032] However, all of the targeted receptors were relatively well-expressed, and the number of unique antibody sequences identified by this process (8 out of 81 screened) was limited.

[0033] Only a limited number of clones specific for differentially expressed surface receptors of the target cells were sequenced, but the high frequency of one antibody clone indicated limited antibody diversity in the activated "BnonT" antibody pool. Thus, this technique provided a significant improvement compared to previous cell-based panning techniques in that antibodies with therapeutic potential against several different differentially expressed receptors were identified by a limited screening effort (Fransson et al., 2006), but this observation indicates that, according to the widely held general view, further improvement is needed because limited screening for binding can only find the most common antibodies, which consist of antibodies against surface receptors with limited diversity, relatively high expression, and strongly differential expression (Hoogenboom, 2002)(Liu et al., 2004, Mutuberria et al., 1999, Osbourn et al., 1998).

[0034] In silico calculations performed as taught in previous biopanning methods (International Publication No. 2004 / 023140 and Frendeus 2006) indicated that the differentially selected "BnonT" antibody pool should contain far more antibodies against each of several differentially expressed surface receptors.

[0035] At that point, the sequencing capacity made it extremely difficult (practically impossible) to sequence a significantly greater number of antibody clones in the pool, and thus, the hypothesis that the differentially selected antibody pool should be far more diverse than is apparent from the first screening was tested using an indirect approach. Thus, immunobeads conjugated with recombinant ICAM-1 protein (ICAM-1 is a cell surface receptor targeted by a single antibody clone among the first 81 clones sequenced in the differentially selected antibody pool of Table 1) were used to pan the differentially selected "BnonT" antibody pool for the presence of additional ICAM-1-specific antibody clones. Screening of the 1260 antibody clones activated after panning of the differentially selected antibody pool against recombinant ICAM-1 identified 21 additional ICAM-1-specific antibody sequences / clones.

[0036] These observations showed that the original differential biopanning method of WO 2004 / 023140 was able to identify antibody clones against differentially expressed antigens, but that the differentially selected antibody pool was far more diverse than was apparent from these first screenings and significantly more diverse than determined by conventional screening approaches.

[0037] The process of WO 2013 / 041643 improved the accuracy of the original differential biopanning method for detecting multiple different anti-ligands against a ligand of interest. This improved biopanning method enabled the activation of a pool of high-affinity anti-ligands, such as human antibodies, specific for different ligands (e.g., receptors) differentially expressed at low to high levels in their native cell surface conformation in a target cell population compared to another cell population, from a human antibody library (and other molecular libraries). This method combined the differential biopanning method with next-generation deep sequencing and used a confirmation screening, virtually a "reverse screening", for anti-ligand specificity against differentially expressed surface receptors of target cells.

[0038] Importantly, all anti-ligands such as antibody clones identified by this approach are, firstly, differentially expressed on target cells versus non-target cells and, secondly, expressed on the receptors in their native cell surface conformation on target cells, and secondly, based on the documented ability of these antibodies to mediate a therapeutic effect in relevant in vitro and in vivo experimental disease model systems, they may have therapeutic potential (Beck et al., 2010, Fransson et al., 2006).

[0039] Despite the progress made in the applicant's previous methods, there remains a need to further improve methods for identifying antibodies specific for ligands that are differentially expressed between two or more samples, cell types, tissues, body fluids, etc. Depending on the relative and absolute expression levels in target and non-target complex biological molecule samples (as described above), such anti-ligands may have distinct therapeutic and / or diagnostic value.

[0040] This application addresses this problem by the aspects of the invention described below. Specifically, this application relates to a method that enables means for the preferential identification of anti-ligands to a ligand based on ligand expression levels in target and non-target complex biomolecule / ligand samples by generation of predictive and experimental antibody enrichment signatures. This enables the identification, expression, and focused analysis of the most promising clones based on their targeted ligand expression in target and non-target complex biomolecule / ligand samples in the most appropriate functional assays, which are often limited in throughput due to the lack of the biological sample of interest (e.g., primary patient-derived cells). For example, diagnostic and therapeutic antibodies that rely purely on blocking ligand-receptor signaling (e.g., anti-IL-6R (31)) can be specific for receptors expressed over a wide dynamic range. In contrast, Fc-dependent and armed therapeutic antibodies that mediate ADCC and CAR-T cell specificity require low (Fc-dependent) receptor expression or no (armed) receptor expression on important normal cells and tissues due to their potent cytolytic properties. By enabling the identification of anti-ligands specific for ligands expressed over a therapeutically and diagnostically relevant range, and by generating distinct predicted, in silico calculated, and / or experimentally derived anti-ligand display selection enrichment profiles, the present invention enables both the identification of orders of magnitude more anti-ligands compared to existing target-independent methods, and the preferential and focused functional screening and target deconvolution in lower throughput downstream assays for distinct therapeutic and / or diagnostic potential anti-ligands. Thus, the method of the present invention enables the activation of the diversity of therapeutically useful antibodies enriched and identified in differential biopanning studies, and the efficient activation of rare antibody clones found in such studies.

[0041] In a first aspect of the present invention, there is provided a method for isolating at least one anti-ligand from a library of anti-ligands against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest, the method comprising: (a) providing one or more reference enrichment signatures for the library of anti-ligands used; (b) performing one or more rounds of differential biopanning on the library of anti-ligands to generate an anti-ligand pool; (c) performing high-throughput sequencing on the anti-ligand pool generated during step (b) to generate a discovery enrichment signature for each anti-ligand in the anti-ligand pool.

[0042] In a preferred embodiment of the first aspect, the method further comprises: (d) comparing one or more reference enrichment signatures provided in step (a) with the discovery enrichment signatures for the anti-ligands generated in step (c) to isolate at least one anti-ligand against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest.

[0043] By "enrichment signature", the inventors mean a pattern of changes in the frequency of a given anti-ligand in the anti-ligand population during successive rounds of biopanning. The change in the frequency of a given anti-ligand from one round to another is determined based on the complete anti-ligand population from the preceding round. The enrichment signature provides a means by which the frequency of anti-ligands in a given anti-ligand population can be tracked and / or predicted during successive biopanning rounds.

[0044] "Reference enrichment signature" includes the meaning of known enrichment signatures for known anti-ligands. A reference enrichment signature can be for an anti-ligand having a known sequence, for an anti-ligand against a known target ligand, and / or for an anti-ligand against a target ligand having a known expression level in a cell, tissue, or sample of interest. A reference enrichment signature can be derived from one or more previously performed anti-ligand library selection experiments (e.g., using one or more differential biopanning steps), or can be generated using in silico calculations and modeling.

[0045] By "discovery enrichment signature", the inventors include the meaning of an unknown enrichment signature for a previously unknown anti-ligand. A discovery enrichment signature can be for an anti-ligand having a previously unknown sequence, for an anti-ligand against a previously unknown target ligand, and / or for an anti-ligand against a target ligand having a previously unknown expression level in a cell, tissue, or sample of interest. A discovery enrichment signature can be derived by performing an anti-ligand library selection experiment (e.g., using one or more differential biopanning steps) and subsequently sequencing the resulting anti-ligand pool.

[0046] Using enrichment signature information, anti-ligands specific for differentially expressed ligands can be identified by comparing and matching a discovery enrichment signature against a reference enrichment signature for an anti-ligand against a known ligand target and / or an anti-ligand against a ligand target having a known expression level in a cell, tissue, or sample of interest. Enrichment signature information can also be used to optimize the parameters of a biopanning experiment to obtain anti-ligands against specific different classes of ligands based on the absolute expression level of a ligand in a target cell, tissue, or sample, or the relative expression level of a ligand in a target cell, tissue, or sample versus a non-target cell, tissue, or sample.

[0047] "Aligning" means aligning a reference enrichment signature with a discovery enrichment signature having a similar profile of changes in anti-ligand frequency in the anti-ligand population during successive rounds of biopanning.

[0048] "Tissue" includes any of the different types of materials from which animals are made, including cells, extracellular matrix and their components, and cell secretory substances.

[0049] "Sample" includes the meaning of a biological sample or specimen. For example, a biological sample / specimen may be a sample / specimen taken from a larger entity (e.g., a sample / specimen of blood, urine, tissue, or saliva), a cell lysate, a microbial culture (e.g., a bacterial or fungal culture), or a population of virus particles.

[0050] In some embodiments of the method of the first aspect, the method further includes a step of screening for anti-ligand specificity for differentially expressed ligands, and the screening is (i) before step (c), performing a screening step to identify target-to-non-target binding, and / or (ii) after step (c), performing a confirmation screening step for anti-ligand specificity for differentially expressed ligands, and / or (iii) after step (d), by performing a confirmation screening step for anti-ligand specificity for differentially expressed ligands.

[0051] In some embodiments of the method of the first aspect, the confirmation screening step is performed by binding analysis using methods such as flow cytometry, FMAT, ELISA, MSD, or CBA.

[0052] In some embodiments of the method of the first aspect, the method of isolating at least one anti-ligand does not include a screening step.

[0053] The statement that "a method for isolating at least one anti-ligand does not include a screening step" means that the method for isolating the anti-ligand does not include, either before step (b), a screening step for identifying target-to-non-target binding, or after step (b), a confirmation step for screening the anti-ligand specificity for differentially expressed ligands, or after step (c), a confirmation screening step for screening the anti-ligand specificity for differentially expressed ligands, or after step (d), a confirmation screening step for screening the anti-ligand specificity for differentially expressed ligands.

[0054] In some embodiments of the method of the first aspect, one or more reference enrichment signatures provided in step (a) are reference enrichment signatures derived in silico.

[0055] "Reference enrichment signature derived in silico" includes the meaning of a reference enrichment signature generated using a computational method such as in silico modeling of anti-ligand enrichment during successive biopanning rounds.

[0056] In some embodiments of the method of the first aspect, the enrichment signature derived in silico is generated using an equation derived from the universal law of mass action. In some embodiments of the method of the first aspect, the enrichment signature derived in silico is the following equation:

[0057]

Number

[0058]

Number

[0059] In some embodiments of the method of the first aspect, the in-silico-derived reference enrichment signature is for an anti-ligand against a highly expressed ligand of the target cell, tissue, or sample of interest, for an anti-ligand against an intermediate-expressed ligand of the target cell, tissue, or sample of interest, or for an anti-ligand against a low-expressed ligand of the target cell, tissue, or sample of interest.

[0060] In some embodiments of the method of the first aspect, one or more reference enrichment signatures provided in step (a) are experimentally derived reference enrichment signatures. In some embodiments of the method of the first aspect, the experimentally derived reference enrichment signature is generated from at least one biopanning experiment comprising at least one reference anti-ligand.

[0061] In some embodiments of the method of the first aspect, the experimentally derived reference enrichment signature is for an anti-ligand against a highly expressed ligand of the target cell, tissue, or sample of interest, for an anti-ligand against an intermediate-expressed ligand of the target cell, tissue, or sample of interest, or for an anti-ligand against a low-expressed ligand of the target cell, tissue, or sample of interest.

[0062] Experimentally derived reference enrichment signatures can be generated from one or more previously performed anti-ligand library selection experiments (e.g., using one or more differential biopanning steps). Experimentally derived reference enrichment signatures can be related to a reference anti-ligand, and such reference anti-ligands can be identified in an anti-ligand library by several approaches and isolated from the anti-ligand library. For example, a reference anti-ligand can be identified in an anti-ligand library and isolated from the anti-ligand library using an isolated reference ligand (e.g., a purified recombinant protein), cells, tissues, or samples enriched for a ligand of interest (e.g., a cell line engineered to overexpress a protein of interest, or a cell line or tissue transformed / transfected with a construct expressing a protein of interest), or using a differential biopanning protocol without prior knowledge of a specific ligand target. Thus, it is not always necessary to know the identity of the ligand targeted by a reference anti-ligand; instead, it is often sufficient to know only the expression levels of the target ligand in target and non-target cells / tissues / samples.

[0063] "Reference anti-ligand" includes the meaning of an anti-ligand having a known sequence and / or specificity whose frequency can be tracked in an anti-ligand pool during successive biopanning rounds. The presence of such a reference anti-ligand in an anti-ligand pool can be confirmed, for example, using nucleic acid sequencing (e.g., next-generation sequencing).

[0064] "Isolated reference ligand" includes the meaning of an isolated ligand having a known expression level (i.e., with respect to the copy number per cell / tissue / sample) in a cell, tissue, or sample of interest. An isolated reference ligand is substantially pure and separated from its normal cellular and / or physiological context. For example, an isolated reference ligand can be a purified recombinant protein of interest immobilized on affinity beads or a solid surface.

[0065] In some embodiments of the method of the first aspect, two or more reference enrichment signatures are provided in step (a), and one or more of the reference enrichment signatures are in-silico-derived reference enrichment signatures, and one or more of the reference enrichment signatures are experimentally-derived enrichment signatures.

[0066] In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), one or more in-silico-derived enrichment signatures are generated using equations derived from the universal law of mass action. In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), one or more in-silico-derived enrichment signatures are the following formula:

[0067]

Number

[0068]

Number

[0069] In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), one or more in-silico-derived reference enrichment signatures are for an anti-ligand against a highly expressed ligand of the target cell, tissue, or sample of interest, an anti-ligand against an intermediate expressed ligand of the target cell, tissue, or sample of interest, or an anti-ligand against a low expressed ligand of the target cell, tissue, or sample of interest.

[0070] In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), the reference enrichment signatures derived from experiments are generated from at least one biopanning experiment comprising at least one reference anti-ligand.

[0071] In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), one or more reference enrichment signatures derived from experiments are for an anti-ligand against a highly expressed ligand of a target cell, tissue, or sample, an anti-ligand against an intermediate expressed ligand of the target cell, tissue, or sample, or an anti-ligand against a low expressed ligand of the target cell, tissue, or sample.

[0072] In some embodiments of the method of the first aspect, the biopanning step (a) comprises (i) a sub-step of providing a library of anti-ligands, and (ii) a sub-step of providing a first population of ligands comprising ligands immobilized or incorporated into a subtractor ligand construct, and (iii) a sub-step of providing a second population of ligands comprising ligands immobilized or incorporated into a target ligand construct, and (iv) determining the amounts of the subtractor ligand construct and the target ligand construct in the population using one or more equations

[0073]

Number

[0074] The steps of the present invention are not necessarily intended to be performed in any particular order.

[0075] "Providing the determined amount" means providing an amount of a ligand already known such that the equations of the present invention are used to verify that the provided known amount is suitable for isolating the desired anti-ligand.

[0076] The reaction parameters utilized for a given selection process can be optimized according to the present invention by calculations that apply the law of mass action and the equations derived therefrom, taking into account parameters such as molecular library diversity, anti-ligand copy number, desired detection limit of upregulation, desired anti-ligand affinity, and ligand concentration.

[0077] In some embodiments of the method of the first aspect, the equation in sub-step (iv) is

[0078]

Number

[0079]

Number

[0080] In some embodiments of the method of the first aspect, step (b) of the method includes performing biopanning for 2 or more rounds, 3 or more rounds, or 4 or more rounds of biopanning. Preferably, step (b) includes performing biopanning for 2 or more rounds, more preferably 3 or more rounds, and most preferably 4 or more rounds of biopanning.

[0081] In some embodiments of the method of the first aspect, at least one differentially expressed ligand is A high-expression ligand of a target cell, tissue, or sample of interest, an intermediate-expression ligand of a target cell, tissue, or sample of interest, or a low-expression ligand of a target cell, tissue, or sample of interest.

[0082] In some embodiments of the method of the first aspect, the high-expression ligand is expressed at more than 1,000,000 copies per target cell of interest, the intermediate-expression ligand is expressed at 100,000 to 1,000,000 copies per target cell of interest, and the low-expression ligand is expressed at less than 100,000 copies per target cell of interest.

[0083] In some embodiments of the method of the first aspect, the ligand is not expressed by either the target construct or the subtracter construct.

[0084] In some embodiments of the method of the first aspect, the method may further include the step of releasing an anti-ligand from the ligand.

[0085] In some embodiments of the method of the first aspect, substeps (ii) to (ix) are performed in parallel to isolate a plurality of anti-ligands against a plurality of different ligands. In some embodiments, substeps (ii) to (ix) are repeated one or more times.

[0086] In some embodiments of the method of the first aspect, the amount of one of the subtracter construct or the target construct is provided in excess of the amount of the other of the subtracter construct or the target construct. In certain embodiments, the excess of the ligand is 10 to 1000-fold, or 2 to 10-fold, or 1000 to 1,000,000-fold.

[0087] The size of the excess subtractor ligand population determines the highest possible "resolution" that can be detected (i.e., how well it can distinguish between anti-ligands that are specifically low, moderately, highly upregulated, or constitutively expressed ligands), and how well it can distinguish between differentially expressed ligands. For example, when using a library with 100 target ligand-specific anti-ligands and adding a sufficiently high concentration of positive ligand such that all anti-ligands bind to the ligand at equilibrium, a 10-fold excess subtractor ligand population generally enables a 90% reduction in the frequency of anti-ligands with specificity for commonly expressed ligands, and a 200-fold excess (twice the number of anti-ligand specific binding agents) enables the removal of common binders (WO 2004 / 023140: See FIGS. 5 and the last paragraph of Example 4 of that document for data confirming this).

[0088] In some embodiments of the method of the first aspect, the high-throughput sequencing step (c) is performed using 454 sequencing, Illumina, SOLiD method or Helicos system, or from Complete Genomics and Pacific Biosciences.

[0089] The advent of next-generation sequencing has enabled the sequencing of large numbers (1,000 to 1,000,000) of candidate genes in a high-throughput manner (hereinafter referred to as "deep sequencing").

[0090] 454 sequencing is described by Margulies et al. (2005) (incorporated herein by reference). In the 454 method, the DNA to be sequenced can either be fractionated and supplied with adapters or segments of the DNA can be PCR amplified using primers that contain adapters. The adapters are 25-mer nucleotides necessary for binding to DNA capture beads and for annealing of emulsion PCR amplification primers and sequencing primers. The DNA fragments are made single-stranded and are bound to DNA capture beads in a manner that allows only one DNA fragment to bind to one bead. Next, the DNA-containing beads are emulsified in a water-in-oil mixture to obtain microreactors that each contain only one bead.

[0091] Within the microreactors, the fragments are PCR amplified, resulting in millions of copies per bead. After PCR, the emulsion is broken and the beads are loaded onto a picotiter plate. Each well of the picotiter plate can contain only one bead. Sequencing enzymes are added to the wells and nucleotides are flowed over the wells in a fixed order. Incorporation of the nucleotides results in the release of pyrophosphate, which catalyzes a reaction that results in a chemiluminescent signal. This signal is recorded by a CCD camera and software is used to translate the signal into a DNA sequence.

[0092] In the Illumina method (Bentley (2008)), single-stranded adapter-supplied fragments are bound to an optically transparent surface and subjected to "bridge amplification". This procedure results in millions of clusters, each containing copies of a unique DNA fragment. DNA polymerase, primers, and four labeled reversible terminator nucleotides are added and the surface is imaged by laser fluorescence to determine the position and nature of the labels. The protecting groups are then removed and this process is repeated for several cycles.

[0093] The SOLiD process (Shendure (2005)) is similar to 454 sequencing, where DNA fragments are amplified on the surface of beads. Sequencing involves cycles of ligation and detection of labeled probes.

[0094] Several other technologies for high-throughput sequencing are currently being developed. Such examples are the Helicos system (Harris (2008)), Complete Genomics (Drmanac (2010)), and Pacific Biosciences (Lundquist (2008)). Since this is a very rapidly evolving technical field, the applicability of high-throughput sequencing methods to the present invention will be apparent to those skilled in the art.

[0095] In some embodiments of the method of the first aspect, the separating means is selected from at least one of a solid support, a cell membrane and / or a portion thereof, a synthetic membrane, beads, chemical tags and free ligands, or fluorescence-activated cell sorting.

[0096] In some embodiments of the method of the first aspect, sub-step (ix) can be performed by at least one of density centrifugation (Williams and Sharon, 2002), solid support isolation, magnetic bead isolation using beads specific for receptors expressed on a single cell population or biotin-specific beads after biotinylation of the single cell population (Siegel et al., 1997), chemical tag binding and aqueous partitioning, fluorescence-activated cell sorting.

[0097] In some embodiments of the method of the first aspect, the library of anti-ligands is a display library comprising a plurality of library members that display anti-ligands. In some embodiments of the method of the first aspect, the anti-ligand display library comprising a plurality of library members that display anti-ligands is a phage display library, an RNA display library, a ribosome display library, a yeast display library, or a mammalian display library. Preferably, the library of anti-ligands is a phage display library.

[0098] In some embodiments of the method of the first aspect, the library of anti-ligands can be constructed from at least one of an antibody, its antigen-binding variant, derivative, and / or fragment; a scaffold molecule having an engineered variable surface; a receptor; and an enzyme. Preferably, the library of anti-ligands is constructed from an antibody, its antigen-binding variant, derivative, and / or fragment.

[0099] In some embodiments of the method of the first aspect, the library of anti-ligands comprises at least one anti-ligand having a known copy number. In some embodiments of the method of the first aspect, at least one anti-ligand having a known copy number is added to the library of anti-ligands at a known copy number. For example, the library of anti-ligands may contain a sequence encoding a commercially available anti-ligand (e.g., a commercially available antibody) added to the library at a known copy number and can be used to generate an enrichment signature from experiments.

[0100] In some embodiments of the method of the first aspect, the library of anti-ligands is a pool of anti-ligands generated by a previously conducted anti-ligand library selection experiment.

[0101] The display of proteins and polypeptides on the surface of a bacteriophage (phage) fused to one of the phage coat proteins provides a powerful tool for the selection of specific ligands. This "phage display" technology was originally used in 1985 by Smith to generate large libraries of antibodies for the purpose of selecting antibodies with high affinity for a particular antigen. More recently, this method has been used to display peptides, protein domains, and intact proteins on the surface of phage to identify ligands with desired properties.

[0102] The principle behind phage display technology is as follows. (i) Clone the nucleic acid encoding the protein or polypeptide for display into the phage. (ii) The cloned nucleic acid is expressed as a fusion to the coat anchoring portion of one of the phage coat proteins (typically, p3 or p8 coat protein in the case of filamentous phage) such that the foreign protein or polypeptide is presented on the surface of the phage. (iii) Phage presenting the protein or polypeptide with the desired property are then selected (e.g., by affinity chromatography), thereby providing the genotype (linked to the phenotype) that can be sequenced, propagated, and transferred to other expression systems.

[0103] Alternatively, the foreign protein or polypeptide can be expressed using a phagemid vector (i.e., a vector containing origins of replication derived from both a phage and a plasmid) that can be packaged as single-stranded nucleic acid in a bacteriophage coat. When a phagemid vector is used, a "helper phage" is used to provide the functions of replication and packaging of the phagemid nucleic acid. The resulting phage expresses both wild-type coat protein (encoded by the helper phage) and modified coat protein (encoded by the phagemid), whereas when a phage vector is used, only the modified coat protein is expressed.

[0104] The use of phage display for isolating ligands that bind to biologically relevant molecules is reviewed in Felici et al. (1995), Katz (1997), and Hoogenboom et al. (1998). Several randomized combinatorial peptide libraries have been constructed to select polypeptides that bind to different targets, e.g., cell surface receptors or DNA (Kay and Paul, (1996)).

[0105] Proteins and multimeric proteins have been successfully phage-displayed as functional molecules (see Chiswell and McCafferty, (1992)). Furthermore, functional antibody fragments (e.g., Fab, single-chain Fv [scFv]) have been expressed (McCafferty et al. (1990), Barbas et al. (1991), Clackson et al. (1991)), and some of the drawbacks of human monoclonal antibody technology have been replaced since the isolation of human high-affinity antibody fragments (Marks et al. (1991) and Hoogenboom and Winter (1992)).

[0106] Further information regarding the principles and implementation of phage display is provided in Phage display of peptides and proteins: a laboratory manual Ed Kay, Winter and McCafferty (1996), the disclosure of which is incorporated herein by reference.

[0107] The library of anti-ligands can be constructed from at least one selected from antibodies and their antigen-binding variants, derivatives, or fragments; scaffold molecules having engineered variable surfaces; receptors; and enzymes.

[0108] In some embodiments of the method of the first aspect, the ligand is at least one selected from antigens; receptor ligands; and enzyme targets comprising at least one of carbohydrates, proteins, peptides, lipids, polynucleotides, inorganic molecules, and conjugated molecules. Preferably, the ligand is a cell surface receptor. More preferably, the cell surface receptor is in its native form.

[0109] In some embodiments of the method of the first aspect, the method further comprises exposing the ligand and its separation means to a stimulus that affects the expression of the target ligand on the ligand construct.

[0110] In a second aspect of the present invention, there is provided a method of isolating at least one anti-ligand from a library of anti-ligands against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest, the method comprising (a) providing one or more reference enrichment signatures for anti-ligands present in the library of anti-ligands used; and (b) providing one or more discovery enrichment signatures for anti-ligands present in the library of anti-ligands used.

[0111] In a preferred embodiment of the second aspect, the method comprises (c) To isolate at least one anti-ligand against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest, further comprising comparing one or more reference enrichment signatures provided in step (a) with one or more discovery enrichment signatures provided in step (b).

[0112] In some embodiments of the method of the second aspect, the method further comprises screening for anti-ligand specificity against the differentially expressed ligand, and the screening is (i) After step (b), performing a confirmation screening step for anti-ligand specificity against the differentially expressed ligand, and / or (ii) After step (c), performing a confirmation screening step for anti-ligand specificity against the differentially expressed ligand, which is implemented by

[0113] In some embodiments of the method of the second aspect, the method of isolating at least one anti-ligand does not include a screening step.

[0114] "The method of isolating at least one anti-ligand does not include a screening step" means that the method of isolating the anti-ligand Before step (b), a screening step for identifying target-non-target binding is also After step (b), a confirmation step for screening anti-ligand specificity against the differentially expressed ligand is also After step (c), a confirmation screening step for anti-ligand specificity against the differentially expressed ligand is also After step (d), a confirmation screening step for anti-ligand specificity against the differentially expressed ligand is also not included, which means.

[0115] In some embodiments of the method of the second aspect, one or more reference enrichment signatures provided in step (a) are in-silico derived reference enrichment signatures.

[0116] In some embodiments of the method of the second aspect, in-silico derived enrichment signatures are generated using equations derived from the universal law of mass action. In some embodiments of the method of the second aspect, in-silico derived enrichment signatures are the following equations:

[0117] [Number] (where FA T = frequency of the recovered antibody rA T = number of the recovered antibody

[0118] [Number] HR = hit rate, the proportion of target cell-specific antibodies (determined experimentally)) are used to generate.

[0119] In some embodiments of the method of the second aspect, in-silico derived reference enrichment signatures are for anti-ligands against high-expressing ligands of the target cell, tissue, or sample of interest, for anti-ligands against medium-expressing ligands of the target cell, tissue, or sample of interest, or for anti-ligands against low-expressing ligands of the target cell, tissue, or sample of interest.

[0120] In some embodiments of the method of the second aspect, one or more reference enrichment signatures provided in step (a) are experimentally derived reference enrichment signatures. In some embodiments of the method of the second aspect, experimentally derived enrichment signatures are generated from at least one biopanning experiment containing at least one reference ligand.

[0121] In some embodiments of the method of the second aspect, the experimentally-derived reference enrichment signature is for an anti-ligand against a highly-expressed ligand of the target cell, tissue, or sample of interest, for an anti-ligand against an intermediate-expressed ligand of the target cell, tissue, or sample of interest, or for an anti-ligand against a low-expressed ligand of the target cell, tissue, or sample of interest.

[0122] The experimentally-derived reference enrichment signature can be generated from one or more previously performed anti-ligand library selection experiments (e.g., using one or more differential biopanning steps). The experimentally-derived reference enrichment signature can be related to a reference anti-ligand, and such reference anti-ligands can be identified in an anti-ligand library by several approaches and isolated from the anti-ligand library. For example, a reference anti-ligand can be identified in an anti-ligand library and isolated from the anti-ligand library using an isolated reference ligand (e.g., a purified recombinant protein), a cell, tissue, or sample enriched for a ligand of interest (e.g., a cell line engineered to overexpress a protein of interest, or a cell line or tissue transformed / transfected with a construct expressing a protein of interest), or a differential biopanning protocol without prior knowledge of a specific ligand target. Thus, it is not always necessary to know the identity of the ligand targeted by a reference anti-ligand; instead, it is often sufficient to know only the expression levels of the target ligand in target and non-target cells / tissues / samples.

[0123] In some embodiments of the method of the second aspect, two or more reference enrichment signatures are provided in step (a), and one or more of the reference enrichment signatures are in-silico-derived reference enrichment signatures and one or more of the reference enrichment signatures are experimentally-derived enrichment signatures.

[0124] In some embodiments of the method of the second aspect, when two or more reference enrichment signatures are provided in step (a), one or more in-silico-derived enrichment signatures are generated using equations derived from the universal law of mass action. In some embodiments of the method of the first aspect, when two or more reference enrichment signatures are provided in step (a), the in-silico-derived enrichment signature is the following formula:

[0125] [Number] (wherein, FA T = frequency of the recovered antibody rA T = number of the recovered antibody

[0126] [Number] HR = hit rate, the proportion of target cell-specific antibodies (determined experimentally)) is used to generate.

[0127] In some embodiments of the method of the second aspect, when two or more reference enrichment signatures are provided in step (a), one or more in-silico-derived reference enrichment signatures are for an anti-ligand against a highly expressed ligand of the target cell, tissue, or sample of interest, an anti-ligand against an intermediate expressed ligand of the target cell, tissue, or sample of interest, or an anti-ligand against a low expressed ligand of the target cell, tissue, or sample of interest.

[0128] In some embodiments of the method of the second aspect, when two or more reference enrichment signatures are provided in step (a), the experimentally-derived enrichment signature is generated from at least one biopanning experiment including at least one reference anti-ligand.

[0129] In some embodiments of the method of the second aspect, when two or more reference enrichment signatures are provided in step (a), the reference enrichment signatures derived from experiments are for antiligands against highly expressed ligands of the target cell, tissue, or sample of interest, antiligands against moderately expressed ligands of the target cell, tissue, or sample of interest, or antiligands against lowly expressed ligands of the target cell, tissue, or sample of interest.

[0130] In some embodiments of the method of the second aspect, one or more discovery enrichment signatures provided in step (b) are derived from a previously conducted antiligand library selection experiment.

[0131] In some embodiments of the method of the second aspect, at least one differentially expressed ligand is a highly expressed ligand of the target cell, tissue, or sample of interest, a moderately expressed ligand of the target cell, tissue, or sample of interest, or a lowly expressed ligand of the target cell, tissue, or sample of interest.

[0132] In some embodiments of the method of the second aspect, the highly expressed ligand is expressed at more than 1,000,000 copies per target cell of interest, the moderately expressed ligand is expressed at 100,000 to 1,000,000 copies per target cell of interest, and the lowly expressed ligand is expressed at less than 100,000 copies per target cell of interest.

[0133] In some embodiments of the method of the second aspect, the library of anti-ligands is a display library comprising a plurality of library members that display anti-ligands. In some embodiments of the method of the first aspect, the anti-ligand display library comprising a plurality of library members that display anti-ligands is a phage display library, an RNA display library, a ribosome display library, a yeast display library, or a mammalian display library. Preferably, the library of anti-ligands is a phage display library.

[0134] In some embodiments of the method of the second aspect, the library of anti-ligands is constructed from at least one of an antibody, an antigen-binding variant, derivative, and / or fragment thereof; a scaffold molecule having an engineered variable surface; a receptor; and an enzyme. Preferably, the library of anti-ligands is constructed from an antibody, an antigen-binding variant, derivative, and / or fragment thereof.

[0135] In some embodiments of the method of the first aspect, the library of anti-ligands comprises at least one anti-ligand having a known copy number. In some embodiments of the method of the second aspect, at least one anti-ligand having a known copy number is added to the library of anti-ligands at a known copy number. For example, the library of anti-ligands may comprise a sequence encoding a commercially available anti-ligand (e.g., a commercially available antibody) added to the library at a known copy number and can be used to generate an enrichment signature from experiments.

[0136] In some embodiments of the method of the second aspect, the library of anti-ligands is a pool of anti-ligands generated by a previously performed anti-ligand library selection experiment.

[0137] In some embodiments of the method of the second aspect, the ligand is at least one selected from an antigen; a receptor ligand; and an enzyme target comprising at least one of a carbohydrate, a protein, a peptide, a lipid, a polynucleotide, an inorganic molecule, and a conjugated molecule. Preferably, the ligand is a cell surface receptor. More preferably, the cell surface receptor is in its native form.

[0138] In a third aspect of the present invention, an enriched signature of the anti-ligand is provided, and the enriched signature is generated by step (c) of the method according to the first aspect.

[0139] The selected anti-ligands identified by the method of the first aspect of the present invention can then be used in the manufacture of a pharmaceutical composition for use in the treatment, imaging, diagnosis, or prognosis of a disease. Anti-ligands based on antibodies and most importantly human antibodies have great therapeutic potential.

[0140] Accordingly, in a fourth aspect of the present invention, there is provided a method for preparing a pharmaceutical composition, which comprises adding an anti-ligand having desired properties identified by the method according to the first or second aspect to a pharmaceutically acceptable carrier after identifying the anti-ligand.

[0141] In a fifth aspect of the present invention, there is provided a pharmaceutical composition prepared by the method according to the fourth aspect.

[0142] In a sixth aspect of the present invention, there is provided a pharmaceutical composition prepared by the method according to the fourth aspect for use in a medicament. The pharmaceutical composition can also be used in the manufacture of a medicament for the prevention, treatment, imaging, diagnosis, or prognosis of a disease.

[0143] Other Definitions "Biopanning" includes the meaning of a method of selecting one member from a desired anti-ligand-ligand binding pair based on its ability to bind to other members with high affinity.

[0144] "Differential biopanning" includes the meaning of a biopanning method of selecting one member from a desired anti-ligand-ligand binding pair that is expressed in different amounts in or on two different sources (e.g., subtractor / control and target) based on its ability to bind with high affinity to other members.

[0145] "High-throughput sequencing" and "deep sequencing" include the meaning of a sequencing process in which a large number of sequences are sequenced in parallel (up to millions), as a result, the speed of sequencing a large number of molecules is actually achievable, and it is made significantly rapid and inexpensive.

[0146] "Confirmation screening" includes the meaning of detecting specific ligand binding of an isolated anti-ligand pool and / or individual anti-ligand clones to a target construct versus a subtractor construct using any assay dealing with ligand / anti-ligand binding (e.g., flow cytometry, FMAT, ELISA, MSD, and CBA). This term further includes the meaning that once an anti-ligand is identified to bind to a differentially expressed ligand, the nature and identity of the ligand and the binding interaction between the anti-ligand and the ligand are studied.

[0147] "Ligand" includes the meaning of one member of a ligand / anti-ligand binding pair. A ligand can be, for example, one strand of a complementary hybridized nucleic acid double-strand binding pair; an effector molecule in an effector / receptor binding pair; or an antigen in an antigen / antibody or antigen / antibody fragment binding pair.

[0148] "Anti-ligand" includes the meaning of the opposite member of a ligand / anti-ligand binding pair. An anti-ligand can be the other strand of a complementary hybridized nucleic acid double-strand binding pair; a receptor molecule in an effector / receptor binding pair; or an antibody or antibody fragment molecule in an antigen / antibody or antigen / antibody fragment binding pair, respectively.

[0149] "Antigen" includes a molecule or chemical compound that can interact with an antibody but does not necessarily elicit an immune response. Such antigens include, but are not limited to, proteins, peptides, nucleotides, carbohydrates, lipids, or conjugate molecules thereof.

[0150] "Differentially expressed ligand" includes a ligand that is expressed only under certain specific conditions / places and not under other conditions / places, and is expressed at different levels between a target and a subtractor source, or either the target ligand or the subtractor ligand is a modified version of the other of the target ligand and the subtractor ligand. For example, some antigens are highly expressed on the cell surface of diseased cells (e.g., cancer cells) and are expressed at low levels or not at all on equivalent healthy cells (e.g., non-cancerous cells).

[0151] "Low-expressed ligand" means a ligand that is expressed at a relatively low level (i.e., less than 100,000 copies per cell / tissue / sample), or a ligand that occurs at a frequency of less than 1% of any other more highly expressed ligand in a positive ligand population sample.

[0152] "Intermediate-expressed ligand" means a ligand that is expressed at a relatively intermediate level (i.e., 100,000 copies per cell / tissue / sample to 1,000,000 copies per cell / tissue / sample).

[0153] "High-expressed ligand" means a ligand that is expressed at a relatively high level (i.e., more than 1,000,000 copies per cell / tissue / sample).

[0154] "Ligand construct" includes a system that includes a target and / or subtractor ligand associated with a separation means.

[0155] The term "antibody variant" is taken to refer to any synthetic antibody, recombinant antibody, or antibody hybrid, including but not limited to single-chain antibody molecules produced by phage display of immunoglobulin light and / or heavy chain variable and / or constant regions, or other immune interaction molecules capable of binding to an antigen in an immunoassay format known to those skilled in the art.

[0156] The term "antibody derivative" refers to any modified antibody molecule capable of binding to an antigen in an immunoassay format known to those skilled in the art, such as fragments of an antibody (e.g., Fab or Fv fragments), or an antibody molecule modified by the addition of one or more amino acids or other molecules to facilitate coupling of the antibody to another peptide or polypeptide, to a large carrier protein, or to a solid support (e.g., among others, the amino acids tyrosine, lysine, glutamic acid, aspartic acid, cysteine, and their derivatives, an NH2-acetyl group or a COOH-terminal amide group). The term "antibody derivative" also refers to bispecific antibodies and cells expressing an antibody on their surface (such as chimeric antigen receptor T cells).

[0157] "Density centrifugation" includes the meaning of the separation of species (e.g., cells, cell organelles, and macromolecules) according to their density differences. This separation is achieved by centrifugation using a density gradient of an appropriate solution, and the species being separated move based on their density.

[0158] The "law of mass action" is a universal natural law applicable under any circumstances. This law states the following for a reaction, aA + bB → cC + dD When this system is in equilibrium at a given temperature, the following ratio is constant,

[0159] [Number] (wherein, A, B, C, and D = the participants (reactants and products) in the reaction a, b, c, and d = coefficients necessary for the balanced chemical equation) The constant is calculated with respect to concentration (indicated by []), and K has the unit M c+d-(a+b) has. Examples embodying certain aspects of the present invention will now be described with reference to the following drawings.

Brief Description of the Drawings

[0160]

Figure 1

Figure 2

Figure 3-1

Figure 3-2

Figure 4-1

Figure 4-2

Figure 5-1

Figure 5-2

Figure 6

Figure 7

Figure 8-1

Figure 8-2

Figure 8-3

Figure 9

Example

[0161] Example 1 - Derivation of the equations used in the experiments of the examples According to the law of mass action, the interaction between an anti - ligand (A), its target receptor (B), and their complex (AB) is given by an equilibrium interaction,

[0162]

Number

[0163]

Number

[0164] The equilibrium interaction between (A) and (B) can be described as follows,

[0165]

Number

[0166]

Number

[0167] Total A or B ([A] or [B]) is the sum of free and bound A or B, i.e., [A]=[fA]+[bA], and [B]=[fB]+[bA].

[0168] Therefore, in (1), substituting [fA] with [A]-[bA] and [fB] with [B]-[bA], the following is obtained:

[0169]

Number

[0170] This equation has the following solution:

[0171]

Number

[0172]

Number

[0173] Substituting the concentration into the number of antibodies / number of particles per mole (N A ) / volume (V) gives the following, or it is simplified.

[0174]

Number

[0175]

Number

[0176] When target cells and non-target cells are mixed, the total number of receptors B is as follows. B = (B T C T + B N C N ) (In the formula, C T = number of target cells C N = number of non-target cells B T = number of receptors on C T above B N = number of receptors on C N above)

[0177] The number of anti-ligand A bound to receptor B on target cells at equilibrium is equal to the total number of anti-ligands bound on target cells and non-target cells, multiplied by the ratio of the number of receptors on target cells to the total number of receptors (both receptors on target cells and non-target cells).

[0178]

Number

[0179] Furthermore, combining equations (3) and (4) gives

[0180]

Number

[0181] Since not all antibodies are recovered after selection, the number of recovered antibodies (rA T ) is as follows:

[0182]

Number

[0183] In the n-CODeR (registered trademark) library, the average copy number of each anti-ligand is 2,000, and the display level is 10%. Therefore, A in Selection 1 is set to 200. In subsequent selections, A is calculated as the number of recovered anti-ligands (rA T ) in the previous selection multiplied by the amplification factor. The amplification factor is determined experimentally (or set to 10,000, 100,000, and 10,000 respectively between Selections 1-2, 2-3, and 3-4 during selection optimization). E and Y are determined experimentally (or set to 0.5 during selection optimization).

[0184] The frequency of recovered anti-ligands (FA T ) in the selected phage pool is calculated as follows,

[0185]

Number

[0186]

Number

[0187]

Number

[0188] Example 2 - Overview of the study using materials and methods Overview of the study The results of this study are presented in the following Examples 3 to 6. This study used a novel prediction-based methodology that combined iterative experimental antibody display selection and large-scale parallel sequencing analysis with in-silico modeling of antibody enrichment to rationally identify hundreds of thousands of sequences encoding antibodies against a priori unknown differentially expressed biomolecules present at unknown levels in complex biological samples. When applied to a naive phage display human scFv antibody library to select antibodies against surface receptors that are differentially expressed on cancer compared to immune cells, 10 5 enriched antibody sequences were predicted to bind to the differentially expressed biomolecules. After subset sampling and expression in the full-length human IgG format, 90% of these antibodies were confirmed to bind to cancer cell-associated receptors, which were expressed over a therapeutically and diagnostically relevant range (from thousands to millions of receptors per cell). The method of the present invention is widely applicable to large molecular libraries that link genotype to phenotype during the screening of complex biological populations, such as body fluids, immune cells or cancer cells, or pandemic microorganisms.

[0189] The present invention integrates computer modeling involving experimental (phage, yeast, or ribosome display) antibody selection and large-scale parallel sequencing to rationally enrich and identify antibodies against a priori unknown biomolecules that are differentially expressed between two samples from a large antibody library that links genotype to phenotype (Figure 1).

[0190] First, the categories of differential target biomolecule specificities for the models are defined by the absolute and relative expression levels of virtual biomolecules in ranges of approximately 10-fold continuous increments from lowest to highest estimated in target and non-target samples. The affinity (K D ) of the antibody for the target biomolecule, selection reaction parameters that enable enrichment of the antibody by drive binding, and, in the case of competitive selection, depletion of minor antibodies specific for biomolecules expressed at similar levels in target and non-target samples are identified by in silico modeling of the selection (Figure 3).

[0191] Second, perform experimental selection of the antibody library and parallel sequence analysis of the recovered antibody pool to provide (1) experimentally enriched antibody sequences, (2) their associated enrichment signatures, and (3) the percentage of displayed antibodies enriched in an antibody- and biomolecule-dependent manner (hit rate).

[0192] Third, incorporate the hit rate into in silico modeling to generate a predicted / reference enrichment signature for the antibody for the required category of differentially expressed target biomolecules. The predicted enrichment signature can be experimentally verified using reference binder sequences for molecules with known expression in target and non-target antigen populations (optional).

[0193] Finally, compare the experimentally obtained enrichment signatures of individual clones to the predicted enrichment signature to identify clone sequences (genotypes) that encode specificity (phenotype) for the required category of differentially expressed target biomolecules (Figure 1).

[0194] Materials and Methods for Examples 3 and 6 Cells Cell lines DU145 (prostate cancer) and Jurkat (clone E6-1, acute T cell leukemia) were obtained from ATCC and cultured according to the supplier's instructions. DU145 cells were stimulated with INF-γ (R&D Systems), 27 ng / mL, for 16 - 24 hours prior to use.

[0195] In-silico Optimization of Selection Criteria Using Equation 6 described in Example 1, the number of target and non-target cells required to recover and enrich a 10 nM antibody against (1) a receptor that is ≧5000 copies per target cell and not expressed on non-target cells, and (2) a receptor that is at least 5-fold upregulated compared to non-target cells and is ≧200000 copies on target cells was calculated. Since the average copy number of each antibody in the n-CoDeR® library is 2,000 at a display level of 10%, the total number of antibody A in Selection 1 was set to 200. In subsequent selections, A was calculated as the number of recovered antibodies (rAT) in the previous selection multiplied by the amplification factor. Amplification factors of 10,000, 100,000, and 10,000 were used in the calculations between Selections 1-2, 2-3, and 3-4, respectively, and it was shown to promote the required specificity enrichment and minimal specificity depletion. The proportion of antibodies eluted from target cell E and the proportion of recovered target cell Y were set to 0.5.

[0196] Predictive Guide Cell Selection For selection with non-target cell competition, DU145 target cells were harvested, washed, biotinylated with EZ-LinkTM Sulfo-NHS-SS-Biotin (Thermo Fisher Scientific), and labeled with anti-biotin microbeads (Miltenyi Biotec) according to the manufacturer's instructions. The labeled target cells (10, 2.5, 5, or 5,000,000 cells in Selections 1, 2, 3, and 4, respectively) were mixed with approximately 1,000-fold excess Jurkat non-target cells and incubated overnight at +4 °C on a rocking platform with n-CoDeR® scFv phage display library 1 (BioInvent International). The cell-phage mixture was loaded onto a MACS column and subsequently washed thoroughly. After eluting the target cells from the column, the phage bound to the target cells was recovered and amplified for use in successive selections. As previously described (Ljungars et al, 2019), phagemid DNA was purified, the gene encoding the scFv was used for soluble scFv production, and Illumina sequencing was performed.

[0197] For selection without non-target cell competition, phage was incubated with 10, 2.5, 5, or 5,000,000 DU145 cells in Selections 1, 2, 3, and 4, respectively, at +4 °C for 4 hours on a rocking platform. After washing the cells four times with phosphate-buffered saline (PBS), phage was recovered and proceeded as described above.

[0198] Hit rate determination The supernatants containing 300 - 1,100 ScFv from each selection were filtered (0.45 μm Millipore), and 25 μL / well was added to a 1:1 mixture of DU145 cells and CellTraceTM CSFE-labeled (Thermo Fisher Scientific) Jurkat cells (50,000 cells in 25 μL of PBS + 0.5% BSA / well) and allowed to bind at +4°C for 1 hour. After washing, scFv binding to live cells (eBioscience™ Fixable Viability Dye eFluor™ 780, Thermo Fisher Scientific) was detected using anti-His-AF647 (R&D Systems) and analyzed by flow cytometry (iQue, Intellicyt Sartorius FortCyt v8.0). The hit rate was determined as the percentage of analyzed clones having a mean fluorescence intensity on target cells at least 3-fold higher than the isotype control.

[0199] In silico calculation of predicted enrichment signatures The in silico predicted enrichment signature, defined as the predicted antibody frequency observed over four consecutive selection rounds of 10 nM antibodies targeting receptors with various expression profiles, was calculated using equations 6 and 7 described in Example 1. Since the average copy number of each antibody in the n-CoDeR® library is 2,000 at a display level of 10%, the total number of antibody A in selection 1 was set to 200. In subsequent selections, A was calculated as the number of recovered antibodies (rAT) in the previous selection multiplied by an amplification factor (determined experimentally). The proportion E of antibodies eluted from target cells and the proportion Y of recovered target cells were determined experimentally. Phage-antibody binding to biomolecules expressed over the experimentally determined expression range (5 × 103 - 4 × 106 copies / target cell) was modeled, assuming the same median affinity (KD = 10 nM) and the same number of antibodies specific to different categories of biomolecules present in the non-selected naive antibody library.

[0200] Array library preparation and Illumina sequencing Phagemid DNA was purified from the enriched phage pool using the QIAprep Spin Miniprep Kit (Qiagen). One-step PCR was performed using PfuUltraII Fusion HS DNA polymerase (Agilent) to amplify the scFv-encoding gene from the phagemid DNA and bind Illumina adapters and indexes to the sample. The reaction volume was 50 μl / sample and contained 50 ng of template and 0.2 μM of each primer. Samples for MiSeq sequencing were amplified using two different primer pairs, with the first pair covering CDR-H1, CDR-H2, and CDR-H3, and the second pair covering CDR-L1, CDR-L2, CDR-L3, and CDR-H3. Samples for NextSeq sequencing were amplified using a primer pair covering CDR-H3. The reverse primer contained a 10-bp index sequence to facilitate multiplexing. The primer sequences used are provided in Table 2. PCR amplification was carried out under the following conditions: 95 °C / 2 min; 12 cycles of 95 °C / 20 s, 62 °C / 30 s, 72 °C / 30 s; followed by 72 °C / 3 min. The PCR products were purified from a 2% agarose gel (MinElute Gel Extraction Kit, Qiagen), quantified (Qubit™ dsDNA HS Assay Kit, Thermo Fisher Scientific), and analyzed for purity and size on an Agilent 2100 Bioanalyzer using the Agilent 1000 DNA kit and Agilent 2100 Expert software (version B.02.08.SI648 (SRI)). The concentration of the pooled sequence library was measured using the KAPA Library Quant Kit Universal qPCR Mix (Roche, KK4824).

[0201] Table 2. Primer Sequences

Table 2

[0202] Samples for MiSeq were combined on a flow cell (Illumina MiSeq Reagent Kit v3 (600 cycles)), 10% PhiX was added, and sequencing was performed on the MiSeq to a median depth of 8,500,000 usable reads / sample using paired-end reads. Samples for NextSeq were combined on four flow cells (Illumina NextSeq 500 / 550 High Output Kit v2.5 (300 cycles)), 25% PhiX was added, and sequencing was performed on the Illumina NextSeq 500 to a median depth of 25,000,000 usable reads / sample using single reads. Base calling and demultiplexing were performed using bcl2fastq version 2.20.0.422, and then the quality of the resulting fastq files was inspected using fastQC version 0.11.8 with default settings. For each sequence, the number of reads was normalized to the total number of reads in the corresponding pooled library. Sequences with fewer than two reads were excluded from further analysis.

[0203] Identification of reference receptors covering the expression range of targets related to diagnosis and treatment Ten cell surface receptors with various gene expression profiles on target cells (DU145) and non-target cells (Jurkat) were identified through searches in the Cancer Cell Line Encyclopedia (Broad Institute, Cancer cell line encyclopedia, 2019) and literature studies (Liu et al, 2000) (see Table 3). These cell surface expressions were measured by flow cytometry using antibodies labeled according to the manufacturer's instructions (see Table 4) and quantification beads (Bang Laboratories, 815B).

[0204] Table 3. Number of receptors expressed on target and non-target cells experimentally determined by flow cytometry.

Table 3

[0205] Table 4. Labeled Antibody / Recombinant Reference Receptor

Table 4

[0206] Production of Antibodies Targeting Reference Receptors As previously described (Ljungars et al, 2019), polystyrene beads (Polysciences, 17175) coated with 25 pmole / bead and 4 beads / receptor were used to select for recombinant reference receptors using either the n-CoDeR® library or phage amplified from cell selection 2 using non-target cell competition. For selections starting with the n-CoDeR® library, a second selection against the recombinant protein was performed, followed by a third selection against DU145 cells. Phage-binding antibodies were converted to soluble scFvs, expressed, and analyzed by flow cytometry as described above for hit rate determination. The scFvs that bound to DU145 cells were analyzed for binding to each receptor in ELISA. Reference receptors (Table 4) were coated onto plates overnight at +4°C. The next day, scFv supernatants diluted 1:4 in PBS containing 0.05% Tween 20 and 0.45% fish gelatin (both from Sigma-Aldrich) were allowed to bind to the washed ELISA plates for 1 hour at room temperature. Bound scFvs were detected using an AP-conjugated anti-FLAG M2 antibody (Sigma-Aldrich), followed by addition of a luminescent substrate (CDP Star Emerald II, Thermo Fisher Scientific), and the plates were read on a plate reader (Tecan Ultra using Tecan Magellan v.3.0). All receptor-binding clones were cherry-picked, grown overnight in 96-well microtiter plates, and subjected to Sanger sequencing.

[0207] Predicted guide antibody classification Frequency through sequential selection (F S1 , F S2 , F S3 , F S4 ) was obtained from NextSeq data. Antibodies were classified into three steps.

[0208] First, using the signature from selection by non-target cell competition, F S2 >F S1 and while F S2 and F S3 >0, the antibody was classified as a binder. Antibodies that did not meet these criteria were classified as non-enriched.

[0209] Next, the antibodies in the binder group were classified into three binder types based on selection by non-target cell competition again. To guide the classification, F S4 was compared with the predicted signature for binders having a 10 nM affinity for receptors expressed at 100,000 and 1,000,000 copies / target cell without non-target cell expression (1,3 ppm and 750 ppm respectively). 1) >1,000,000 (antibodies having a frequency signature higher than the predicted signature of 1,000,000) - inclusion criterion: F S4 >750 ppm 2) 100,000 - 1,000,000 (antibodies having a frequency signature of the predicted signature of 100,000 - 1,000,000) - inclusion criterion: 1,32 ppm < F S4 <750 ppm 3) <100,000 (antibodies having a frequency signature lower than the predicted signature of 100,000) - inclusion criterion: F S4 <1,32 ppm

[0210] Thirdly, the antibodies classified as <100,000 were further classified by comparing the signatures from selection with and without non-target cell competition. 1) Upregulation - inclusion criterion

[0211] [Number] or competing F S4 If = 0, non-competing F S4 > 0 and

[0212] [Number] 2) Low expression, target cell-limited - inclusion criterion F S1 ≠ 0 and F S2 , F S3 , F S4 = 0 (no non-target cell competition) 3) Unclassified - does not meet any of the above criteria.

[0213] Production of IgG The complete scFv sequences required for clone synthesis were obtained from the Miseq data. Briefly, the VH and VL+CDR-H3 sequences were combined based on the CDR-H3 sequence. When one CDR-H3 was associated with two or more sets of VH and / or VL sequences, the frequencies in the two libraries were used to join the correct VH / VL pairs. The antibody genes were synthesized (Twist Bioscience), ligated into a vector containing genes encoding the heavy chain constant region and light chain constant region of human IgG1, produced in HEK293 cells, and purified as previously described (Ljungars et al, 2018).

[0214] Confirmation binding analysis of predicted guide antibodies Purified IgG was diluted to 100 μg / mL, titrated in 25 μL of PBS + 0.5% BSA, added to 50,000 DU145 (target) and Jurkat (non-target) cells / well, and allowed to bind at +4 °C for 1 hour. After washing, IgG bound to live cells was detected using an APC-conjugated anti-human Fc antibody (Jackson Immunoresearch, 109-136-098) together with a live / dead cell marker (SYTOX green, Thermo Fisher Scientific) and analyzed by flow cytometry (iQue, Intellicyt Sartorius using FortCyt v8.0). To generate a calibration curve for converting the MFI at saturation binding to the number of receptors, a subset of IgG with different signal intensities at the saturation cell binding concentration was selected for receptor number determination. Purified IgG was labeled with AF647 using Alexa Fluor 647 carboxylic acid succinimidyl ester (ThermoFisher Scientific) according to the manufacturer's instructions. The labeled antibody was used for receptor number determination using calibration beads (Bang Laboratories, 816) according to the manufacturer's instructions (Figure 2). The assay detection limit (receptor number of the isotype control) was 1,000 receptors / cell.

[0215] Quantification and statistical analysis Statistical analysis was performed using GraphPad Prism 9 and carried out by the Mann–Whitney test with *P ≤ 0.05 (Figure 8C).

[0216] Example 3 - The predicted enrichment signatures accurately model the enrichment of binders according to the absolute and relative expression levels of the target receptor. Using the method of the present invention, a large (>1010 member) naive human antibody library (Soderlind et al, 2000) against cell surface receptors differentially expressed between two cell types of DU145 prostate cancer (target) cells compared to Jurkat T (non-target) cells was screened.

[0217] In the first step, a category of differentially expressed surface receptors was defined that covered a broad dynamic expression range spanning from thousands (1,000) to millions (1,000,000) of receptors per cell. Any receptor that was upregulated by five-fold or more on the target compared to non-target cells was potentially considered therapeutically or diagnostically interesting. Therefore, selection reaction parameters were determined that would allow for the enrichment of relevant antibody binding agents to receptors that were >5-fold upregulated on target cells relative to non-target cells and the removal or selective release of minor binding agents to receptors that were <5-fold upregulated. Through in silico modeling, a selection protocol was identified that would result in the enrichment of high-affinity (K 3 ~10 6 ≦10 nM) antibodies to receptors expressed across this range (10 D receptors / cell) and at least 5-fold upregulated on the target compared to non-target cells (Figure 3A). The resulting protocol involved the application of positive and negative selection pressures in the form of target cells and excess non-target cells, respectively. Modeling indicated that in the absence of negative selection pressure, selection was unable to reduce minor antibodies specific to receptors that were equally expressed or <5-fold upregulated on the target relative to non-target cells, and distorted enrichment towards highly expressed rather than differentially expressed receptors (Figure 3B).

[0218] Four rounds of experimental selection were performed according to the in silico optimization protocol described above (i.e., selection with or without competition). Enrichment signatures based on four (compared to fewer) data points generated using selection with or without competition were considered useful for discriminating antibodies against various types of differentially expressed receptors, such as receptors whose expression was restricted to target cells but varied by expression level (restricted receptors), and receptors that were upregulated on target cells, i.e., also expressed on non-target cells (upregulated receptors). The percentage of phage antibodies enriched in an antibody- and target cell receptor-dependent manner in each selection round (i.e., the hit rate) was determined by flow cytometry after conversion of the phage antibodies to the scFv antibody format (Figure 4).

[0219] The experimentally determined hit rate and a fixed K equal to 10 nM were used to model in silico the cell receptor-specific phage antibody enrichment according to driven equilibrium binding. To experimentally verify the predicted enrichment signature, candidate reference cell surface receptors with expression profiles that match the classified receptors were identified by probing the Broad Institute Cancer Cell Line Encyclopedia (CCLE) for cell surface receptor-encoding genes that were differentially expressed in the target as compared to non-target cells. Based on flow cytometry analysis using commercially available antibodies and ABC quantification beads, ICAM-1, CD44, EGFR, HER2, ROR1, CD40, CD130, and CD55 were differentially expressed on the target as compared to non-targets, while CD59 and CD71 represented receptors that were <5-fold upregulated on target cells as compared to non-target cells (Figure 5A, Table 3). The number of molecules per target cell of the restricted receptors was 10 D (CD130), 10 D (CD40, ROR1, HER2), 10 3 (CD44, EGFR), up to 10 4 (ICAM-1), and there was no detectable expression on non-target Jurkat cells. The upregulated receptors CD55, CD59, and CD71 were similarly expressed on target cells (2 - 4×10 5 receptors / cell), but were >5-fold (CD55, 10-fold) and <5-fold (CD59, 4-fold and CD71, 2-fold) upregulated on the target as compared to non-target cells (Figure 5A, Table 3). Antibodies and their coding sequences (n = 4 - 69) for each of the reference receptors were identified by selection using each of the recombinant extracellular domains of each of the reference receptors (see Example 2), enabling the evaluation of the enrichment profiles of antibodies against biomolecules with established expression differences between target and non-target cells. 6 5

[0220] ​​The experimental enrichment of antibodies specific for a reference receptor was monitored by massively parallel sequencing of an antibody pool obtained by experimental cell-based selection and compared to that predicted by in silico modeling (Figure 5B). When the antibody-receptor equilibrium was modeled according to K D = 10 nM, a strong correlation was observed between the median in silico predicted frequency and the experimentally determined frequency of antibodies for differentially expressed reference receptors (Figure 5B). This value corresponded well to the median affinity of antibodies for diverse cell surface receptors of different types isolated from the n-CoDeR® antibody library used herein (Roghanian et al, 2015, Fransson et al., 2006, Schiopu et al, 2007). A similar strong fit between the in silico predicted frequency and the experimentally determined frequency was observed for selection responses with or without non-target cell competition (Figure 5B). Notably, the signatures of antibodies for low-expressing target cell-restricted receptors (CD40 and CD130), and upregulated receptors (CD55) were very similar in selection with non-target cell competition, but were very different in selection without competition.

[0221] Selection of antibodies from a highly diversified library was further modeled for antibody enrichment according to 10-fold higher (1 nM) or 10-fold lower (100 nM) affinities to generate receptor-specific antibodies of various affinities. The enrichment signatures were overall similar, but the modeled antibody frequencies increased with higher affinity and decreased with lower affinity (Figure 6A). The enrichment signatures were more sensitive to antibody affinity the lower the expression of the targeted receptor. This correlated with a greater variation in the frequencies observed for antibodies against low-expression (CD130, CD40, ROR1, HER2) receptors compared to high-expression (ICAM-1, CD44, EGFR) receptors (Figure 5B). To evaluate how possible errors in receptor quantification could affect the antibody enrichment signature, enrichment was modeled with a number 2-fold higher or 2-fold lower than the experimentally determined number. This mainly affected the signature of antibodies against upregulated receptors (CD71, CD59, CD55; Figure 6B).

[0222] In summary, in silico modeled enrichment of target receptor-specific antibodies closely reflected experimental antibody enrichment, which was driven by the absolute and relative expression levels of the target receptor in target and non-target cells (Figure 5B).

[0223] Example 4 - Enrichment profiles show hundreds of thousands of antibodies against receptors differentially expressed over a wide dynamic range. The fact that the method accurately modeled the enrichment of reference antibodies against receptors expressed over a broad dynamic range supports its potential utility as a discovery tool for identifying antibodies against targets relevant to distinct therapeutic and diagnostic possibilities (Figure 5B). The value of such a discovery tool is primarily determined by the number of unique binders to differentially expressed targets that this technology generates. To evaluate this, in silico optimization and selected antibody pools were interrogated for antibody sequences with enrichment profiles that match in silico predicted signatures of antibodies specific for different categories of therapeutically or diagnostically relevant receptors (Figure 5B). Binders developed for diagnostic purposes are independent of functional regulatory properties and can thus be specific for any differentially expressed receptor, regardless of expression level. Similarly, antibodies that rely purely on blocking ligand-receptor signaling (e.g., anti-IL-6R) (Sebba, 2008) can be specific for receptors expressed over a broad dynamic range. In contrast, for example, Fc-dependent and armed therapeutic antibodies that mediate ADCC and CAR-T specificity require low (Fc-dependent) receptor expression on important normal cells and tissues or no receptor expression at all (armed) due to their potent cytolytic nature. Thus, antibodies with enrichment signatures that match those predicted to represent specificity for restricted receptors expressed at (1) >10 6 molecules / target cell, (2) 10 5 ~10 6 molecules / target cell, or (3) <10 5 molecules / target cell, or are >5-fold upregulated on target compared to non-target cells were quantified by analyzing the frequency of individual antibody sequences in the eluted phage-antibody pools at each of 4 rounds of experimental selection.

[0224] Analysis of the complete sequence dataset showed the presence of hundreds of thousands of antibodies against any receptor category (Figure 7A). Signature-guided analysis showed that after 2 or more rounds of panning, the binder pool (10 2 ~10 5showed that antibodies specific for the most highly expressed receptors dominate (ppm / sequence), and conversely, antibodies specific for lower-expressed target cell-restricted or up-regulated receptors are rare (10 -2 ∼10 ppm / sequence) (Figures 7A-7B). Interestingly, in contrast to their low frequency, the number of antibody clones specific for upregulated or target-restricted receptors with low expression was 10 4 It was shown that the α-glucan concentration exceeded 1000kJ / μg (Figure 7C).

[0225] The data presented above support the idea that the methods of the present invention enable the discovery of antibodies against receptors spanning a range of medically relevant expression, including low- and up-regulated receptors (Figure 7). Importantly, the observation that antibodies against low- and up-regulated receptors were present at very low frequencies in the selected pools (less than one clone per million) is consistent with the observed shortcomings of existing methods for generating such antibodies. In the absence of information-based enrichment signatures identified in data generated by integrated computer modeling and massively parallel sequencing, robust identification of these specificities requires the production of millions of antibody clones and labor-intensive cell-based screening.

[0226] Example 5 - Prediction allows for the preferential discovery of binding agents with different therapeutic and diagnostic potential. Detailed analysis of the antibodies belonging to the least frequent but largest group (Fig. 7A, bottom group) revealed low expression (10 3 ~10 5 Receptor / cell) restricted receptors, such as HER2, ROR1, and CD130, as well as antibodies against medium- to high-expressing (10 5 ~10 6It has been shown to contain clones specific for a receptor / cell) upper control receptor, e.g., CD55 (Figure 5B). As discussed above, these two target expression profiles are each associated with the therapeutic potential of agonistic and function-blocking antibodies. Therefore, it stands to reason that depending on the intended therapeutic modality, one should focus on either type of antibody specificity.

[0227] The observations in Example 4 above showed that the enrichment signatures for antibodies against low-expressing, restricted receptors and upregulated receptors were similar in selection with non-target cell competition, but significantly different in selection without competition (Figure 5B). This suggests that it is possible to identify sequences encoding antibodies with therapeutic potential as agonistic (against restricted receptors) or function-blocking (against upregulated receptors) antibodies by potentially using comparative analysis of enrichment signatures generated from selection with and without competition. Antibodies. In support of this concept, in silico modeled enrichment signatures showed a decrease in the frequency of antibodies against upregulated receptors and, depending on the expression level, an increase or no change in the frequency of antibodies against restricted receptors in selection with non-target cell competition compared to selection without competition (Figure 8A). These observations are consistent with the negative selection pressure imposed in the form of non-target cells and promote the activation of binders while increasing selectivity for the target compared to non-target cells, as modeled in silico (Figure 3) and experimentally verified by flow cytometry analysis (Figure 4), reducing binders specific for generally expressed receptors and weakly upregulated receptors of target cells.

[0228] The comparative analysis was extended to include a total of 94,429 sequences represented by enrichment signatures encoding antibodies against low-expression restricted or up-regulated receptors (lower panel, Figure 7A), enabling the classification of the proportion of these antibodies into any category. Thus, 13,709 antibodies specific for low-expression restricted receptors (increased frequency) and 753 antibodies against up-regulated receptors (decreased frequency) were shown (Figure 8B). The remaining antibodies were not significantly affected by competitive selection (Figure 8B, n = 79,967).

[0229] Example 6 - Prediction discovers an unprecedented number of antibodies with different therapeutic and diagnostic potentialities. Thus, the collected data show that predictive-guided selection generated a highly diversified pool of antibodies against empirically unknown differentially expressed surface receptors expressed over a wide dynamic range, and that the enrichment signature was used to identify an unprecedented number (about 10 3 compared to about 10 5 ) of antibodies against therapeutically and diagnostically relevant receptors. To address its relevance to such antibody discovery, (Group 1) > 10 6 molecules / cells (n = 8), (Group 2) 10 5 ~10 6 molecules / cells (n = 16), (Group 3) < 10 5Antibodies specific to receptors encoded by enriched signatures (n = 102) that are expressed by molecules / cells or upregulated receptors on target cells (n = 67), or (group 4) antibodies to receptors that are <5-fold upregulated on target cells or lack specificity for target cell receptors (n = 11) were selected for production in the full-length IgG format (Figure 9, upper panel). After expression and purification, the number of antibody epitopes (i.e., expression of the targeted receptor on target and non-target cells) was quantified by flow cytometry using ABC beads (Figure 9, lower panel). Consistent with the ability of the method of the invention to generate a unique number of antibodies to differentially expressed surface receptors, 93% (85 / 91) of the antibodies in groups 1-3 were confirmed to be specific for the differentially expressed receptor on target cells. Equally important and demonstrating the value of the predictive guidance approach, 9 / 11 (82%) of the antibodies that showed clear enrichment between selection rounds 1 and 2 but lacked specificity for medically relevant receptors as indicated by the enrichment signature were confirmed to be negative for binding to target cells.

[0230] Furthermore, consistent with the enrichment signature informing antibody specificity, the experimentally determined number of antibody-targeted receptors on target cells correlated well with the predicted number. Antibodies in group 1 were predicted to bind the receptor at >10 6 copies / cell, and the experimentally determined target cell expression was 1.4×10 6 (7.3×10 5 ~2.6×10 6 )(geometric mean (95% confidence interval)). Group 2 was predicted to bind the receptor at expression levels of 10 5 ~10 6 , and the experimentally determined number was 3.6×10 5 (1.0×10 5 ~1.3×10 6 ). Finally, group 3 was predicted to bind receptors with restricted expression levels <10 5 or upregulated receptors.

[0231] To assist in separating these different categories of antibodies, a comparative analysis of the Group 3 enrichment signatures from selection with or without non-target cell competition was performed. This dataset included sequences that showed either an increase (28 / 67) or a decrease (7 / 67) in frequency in the presence compared to the absence of competitive selection (Figure 9, inset). Consistent with these respective enrichment signatures informing antibody specificity for upregulated receptors and restricted low-expression receptors, the experimentally determined numbers of epitopes on target and non-target cells for antibodies with predicted specificity for upregulated receptors were 5.4×105 (1.4×105~2.2×106) and 1.3×104 (2.3×103~7.9×104), and for antibodies with predicted specificity for restricted low-expression receptors were 3.0×104 (1.3×104~7.1×104) and not determinable, the latter because 22 / 28 of the antibodies in this group bound to receptors with non-target cell expression below the limit of detection (limit of detection = 1×103 receptors / cell) (Figure 9, inset).

[0232] Overall, antibody specificity could be predicted by the enrichment signature, although the specificity of some individual antibodies deviated from the prediction (Figure 9). For example, only Group 1 (left panel) was predicted to contain antibodies specific for receptors expressed at >1,000,000 copies / target cell and not expressed on non-target cells. However, all groups included a proportion of antibodies with this expression profile. Modeling data showed differential enrichment of antibodies with different affinities for the same receptor (or receptors with the same expression level), and since the naive antibody library included antibodies of various affinities, we next analyzed how antibody affinity affects the antibody enrichment signature. Antibodies with similar determined epitopes (i.e., similar receptor expression levels on target cells) were analyzed for their >3 nM or <3 nM EC for binding to endogenously expressed receptors 50They were divided into two groups according to the value, and their enrichment during selection was compared. Consistent with the modeling shown in Figure 6A, for both monitored target expression levels (1,000,000 - 4,000,000 copies per cell, n = 22 and 300,000 - 1,000,000 copies per cell, n = 10), EC 50 the frequency of antibodies with an EC 50 value < 3 nM was found to be higher than the frequency of antibodies with an EC

[0233] value > 3 nM. For antibodies against receptors with 1,000,000 - 4,000,000 copies per cell, this difference was statistically significant after selections 2, 3, and 4 (p < 0.05) (Figure 8C).

[0234] Discussion of the Results in Examples 3 - 6 This study describes the use of a novel selection and screening methodology that enables the rational identification of antibodies against a priori unknown differentially expressed cell surface receptors. Compared to previously described techniques, this method has identified several orders of magnitude more antibodies that have been shown to bind receptors expressed over a broad dynamic range, including low-expressing, restricted receptors. Thus, prediction-based discovery overcomes some of the limitations of selection and screening methodologies in the art.

[0235] The method of the present invention is particularly useful for phenotypic discovery (PD) of biologics (see Moffat et al, 2017 for an overview of PD). In PDD, small molecules or antibodies from a macromolecular library are screened for functional activity (e.g., inhibition of pro-inflammatory cytokine release or induction of tumor cell death) without prior knowledge of their molecular targets. As a result, PDD enables the discovery of the most functional molecules and antibodies across multiple receptors and epitopes for specific disease-related pathways. PDD is a well-validated strategy for first-in-class small molecule drug discovery (Swinney, 2013, Swinney and Lee, 2020) and has been used to identify several first-in-class antibodies and their associated targets, such as CD52 (Waldmann et al, 1984), ICAM-1 (Veitonmaki et al, 2013), CD32b (Roghanian et al, 2015, Ljungars et al, 2018), TNFR2 (Williams et al, 2016), which are currently in clinical development. By combining the ability of prediction-based discovery to identify a large number of antibodies against a wide range of surface receptors with appropriate functional screening (Ljungars et al, 2018, Chandrasekaran et al, 2021) and recently described high-throughput CRISPR-based methods for target deconvolution (Mattsson et al, 2021), biologics PDD can be moved to the next level, directly equivalent to small molecule PDD.

[0236] An important feature of the methods of the invention related to both diagnostic and therapeutic uses is its ability to identify antibodies against low-expressing disease-related molecules. Antibodies against low-expressing tumor-restricted antigens and rare disease-related conformational epitopes can have significant therapeutic potential when developed as empowered biologics. With respect to diagnosis, liquid biopsies can be easily sampled but typically contain very low concentrations of biomarkers. Thus, technologies that assist in the identification and quantification of rare disease-related biomarkers serve the pursuit of early diagnosis and personalized medicine. The observation in this study that antibodies against low-expressing and upregulated receptors are present at a very low frequency (less than 1 clone per million) in a selected pool is consistent with the observed drawbacks of existing methods for generating such antibodies. The robust identification of these specificities in the absence of enrichment signatures identified in data generated by integrated computer modeling and massively parallel sequencing requires the production of millions of antibody clones and labor-intensive cell-based screening. Antibodies against low-expressing receptors are also valuable in diagnosis.

[0237] In this study, prediction-based discoveries were applied to identify antibodies that discriminate one cell type from another, which is widely applicable to other complex antigen systems associated with diverse inflammatory, immunological, neurological, infectious diseases and cancer, such as blood, urine, cerebrospinal fluid (Cortese et al, 1996), tissue (Larsen et al, 2015), bacteria (DiGiandomenico et al, 2012), or virus (van der Brink et al, 2005). Another highly relevant use of the invention is to identify antibodies against important pathogenic factors, such as the adhesin glycoproteins of pandemic microorganisms (Bertoglio et al, 2021) and their receptors on host cells. The identification of such antibodies and their related molecular targets has the potential to generate both passive (antibody-based) and active (vaccine) immunotherapies to assist in the treatment and prevention of drug-resistant microbial infections.

[0238] A further advantage of prediction-based discovery is the parallel discovery of target molecules and candidate therapeutic or diagnostic antibodies to these targets, compared to other discovery methodologies (e.g., gene expression-based approaches or proteomics-based approaches), which can be of complex or processed nature, such as oxidized low-density lipoprotein (Schiopu et al, 2007, Lehrer-Graiwer et al, 2015, Li et al, 2013), and can include disease-related epitopes and conformations.

[0239] An important application underlying the ability of prediction-based discovery to generate comprehensive panels of binders to differentially expressed molecules relates to the integration of in silico modeling and experimental selection. First, in silico modeling is utilized to establish the selection reaction parameters necessary to achieve robust enrichment of relevant binders and de-selection of minor binders without knowing the actual binder specificity or abundance of the targeted molecules. This is accomplished by simulating the enrichment of virtual binders defined by their specificity for molecules with different absolute and relative expression levels in diseased compared to two complex antigen populations, e.g., healthy blood, serum, or cells. As modeled and experimentally verified in this study, insufficient positive selection pressure results in a risk of depleting fewer retained binders and binders to lower expressed molecules (not shown in the data). On the other hand, insufficient subtractive selection pressure generates binders to the most highly expressed molecules rather than the most strongly differentially expressed molecules (i.e., disease-related molecules) (Figure 2). The retroactive application of in silico modeling to previously described differential selection protocols is consistent with both insufficient positive selection pressure and subtractive selection pressure underlying the poor quality and quantity of the observed binders (not shown in the data).

[0240] Second, the method of the invention enables the discovery of an unparalleled class of binders with different therapeutic and diagnostic potential by integrating in silico modeling with experimental differential selection. Specifically, the proportion of binder enriched in an antigen-specific manner (i.e., binder hit rate) is determined experimentally. The enrichment signature of the classified binders, as defined above by their specificity for molecules having different absolute and relative expression levels in two antigen populations, is then generated by in silico modeling of the proportion of binder enriched antigen-specifically according to the abundance of the targeted molecule and the presence or absence of competitive selection. By collating the enrichment signature of individual experimentally identified binder sequences against a reference enrichment signature modeled in silico, binders to several differentially expressed molecules can be identified in an unparalleled number (about 100,000) with high precision (90%). This same methodology enables prioritization based on binder information according to the indicated therapeutic and diagnostic uses (e.g., naked or armed IgG).

[0241] In this study, the presented equation was used to optimize the selection reaction in silico and generate a predicted reference enrichment signature for the identification of (experimentally enriched) antibody sequences encoding specificity for therapeutically and diagnostically relevant cell surface receptors. However, the modeling can also be used to model the antibody enrichment of selections already made. The retroactive application of in silico modeling to the inventors' and other previously described differential selection protocols is consistent with both insufficient positive and subtractive selection pressures underlying the observed antibody quality and quantity limitations. Thus, predictions can help "rescue" existing antibody pools generated using precious patient-derived materials by reanalysis incorporating the enrichment signature and / or by demonstrating the benefit of performing additional (optimized) selection rounds in the presence or absence of competition.

[0242] In conclusion, particularly with respect to PDD, prediction-based discoveries have shifted the current bottleneck from the identification of numerous antibodies for summing differentially expressed disease-related biomolecules to efficient antibody production and the development of high-throughput clinically predicted functional assays, which enables the screening of thousands (tens of thousands) of clones.

[0243] References · Barbas CF, Kang AS, Lerner RA and Benkovic SJ, (1991), Assembly of combinatorial antibody libraries on phage surfaces: the gene III site. Proc. Natl. Acad. Sci. USA 88, 7978-7982. · Barreto, K. et al. (2019). Next-generation sequencing-guided identification and reconstruction of antibody CDR combinations from phage selection outputs. Nucleic acids research, 47(9):e50. · Beck et al. (2010). Strategies and challenges for the next generation of therapeutic antibodies. Nat Rev Immunol 10, 345-352. · Beers et al. (2008). Type II (tositumomab) anti-CD20 monoclonal antibody out performs type I (rituximab-like) reagents in B-cell depletion regardless of complement activation. Blood 112, 4170-4177. · Bentley DR, et al. Accurate whole human genome sequencing using reversible terminator chemistry. Nature. 2008;456:53 - 59。 · Bertoglio, F., et al. (2021). SARS-CoV-2 neutralizing human recombinant antibodies selected from pre-pandemic healthy donors binding at RBD-ACE2 interface. Nat Commun 12, 1577。 · Boder and Wittrup (1997). Yeast surface display for screening combinatorial polypeptide libraries. Nat Biotechnol 15, 553 - 557。 · Carlsson, R., et al. (1988) Binding of staphylococcal enterotoxin A to accessory cells is a requirement for its ability to activate human T cells. J Immunol 140, 2484。 · Chandrasekaran, S.N., Ceulemans, H., Boyd, J.D. & Carpenter, A.E. (2021). Image-based profiling for drug discovery: due for a machine-learning upgrade? Nat Rev Drug Discov, 20, 145 - 159。 · Chiswell DJ and McCafferty J, (1992), Phage antibodies: will new ‘coliclonal’ antibodies replace monoclonal antibodies? Trends Biotechnol. 10, 80 - 84。 · Clackson T, Hoogenboom HR, Griffiths AD and Winter G, (1991), Making antibody fragments using phage display libraries. Nature 352, 624-628。 · Cortese, I. et al. (1996). Identification of peptides specific for cerebrospinal fluid antibodies in multiple sclerosis by using phage libraries. Proc Natl Acad Sci USA 93, 11063-11067。 · Cragg, M.S., and Glennie, M.J. (2004). Antibody specificity controls in vivo effector mechanisms of anti-CD20 reagents. Blood 103, 2738-2743。 · de Kruif, J., Terstappen, L., Boel, E. & Logtenberg, T. (1995). Rapid selection of cell subpopulation-specific human monoclonal antibodies from a synthetic phage antibody library. Proc Natl Acad Sci USA 92, 3938-3942。 · DiGiandomenico, A. et al. (2012). Identification of broadly protective human antibodies to Pseudomonas aeruginosa exopolysaccharide Psl by phenotypic screening. J Exp Med 209, 1273-1287 · Dower, W.J., CWIRLA, S.E. and N.V., A.T. (1991) Recombinant Library Screening Methods. In: World Intellectual Property Organization. SMITH, W.M., United States。 · Drmanac R, et al. Human Genome Sequencing Using Unchained Base Reads on Self-assembing DNA Nanoarrays. Science 2010;327:78-81。 · Dyer, M.J., Hale, G., Hayhoe, F.G. & Waldmann, H. (1989). Effects of CAMPATH-1 antibodies in vivo in patients with lymphoid malignancies: influence of antibody isotype. Blood 73, 1431-1439。 · Egloff et al(2019). Engineered peptide barcodes for in-depth analyses of binding protein libraries. Nat Methods 16, 421-428 · Ellington and Szostak(1990). In vitro selection of RNA molecules that bind specific ligands. Nature 346, 818-822。 · Felici F, Luzzago A, Manaci P, Nicosia A, Sollazzo M and Traboni C, (1995), Peptide and protein display on the surface of filamentous bacteriophage. Biotechnol.Annual Rev.1, 149-183。 ·Francisco, J.A., Stathopoulos, C., Warren, R.A., Kilburn, D.G. and Georgiou, G. (1993) Specific adhesion and hydrolysis of cellulose by intact Escherichia coli expressing surface anchored cellulase or cellulose binding domains. Biotechnology (NY) 11, 491。 ·Fransson, J., Tornberg, U.C., Borrebaeck, C.A., Carlsson, R., and Frendeus, B. (2006). Rapid induction of apoptosis in B-cell lymphoma by functionally isolated human antibodies. Int J Cancer 119, 349 - 358。 ·Frendeus, B. (2013). Function-first antibody discovery: Embracing the unpredictable biology of antibodies. Oncoimmunology 2, e25047。 ·Gao, C., Lin, C.H., Lo, C.H., Mao, S., Wirsching, P., Lerner, R.A. and Janda, K.D. (1997) Making chemistry selectable by linking it to infectivity. Proc Natl Acad Sci USA 94, 11777。 ·Hanes, J. and Pluckthun, A. (1997) In vitro selection and evolution of functional proteins by using ribosome display. Proc Natl Acad Sci USA 94, 4937。 · Hanes, J., Schaffitzel, C., Knappik, A. & Pluckthun (2000). A. Picomolar affinity antibodies from a fully synthetic naive library selected and evolved by ribosome display. Nat Biotechnol 18, 1287 - 1292。 · Harris TD, et al. (2008). Single - molecule DNA sequencing of a viral genome. Science; 320: 106 - 109。 · He, M. and Taussig, M. J. (1997) Antibody - ribosome - mRNA (ARM) complexes as efficient selection particles for in vitro display and evolution of antibody combining sites. Nucleic Acids Res 25, 5132。 · Hoogenboom HR and Winter G, (1992), By - passing immunisation. Human antibodies from synthetic repertoires of germline VH gene segments rearranged in vitro. J.Mol.Biol.227, 381 - 388)。 · Hoogenboom HR, deBruine AP, Hutton SE, Hoet RM, Arends JW and Rooves RC, (1998), Antibody phage display technology and its application. Immunotechnology 4(1), 1 - 20。 · Hoogenboom, H.R. (2002). Overview of antibody phage display technology and its applications. In Methods in Molecular Biology P.M. O’Brien, and R.Aitken, eds. (Totowa, NJ: Humana Press Inc.), pp. 1-37。 · Jacobsson, K. and Frykberg, L. (1995) Cloning of ligand-binding domains of bacterial receptors by phage display. Biotechniques 18, 878。 · Katz BA, (1997), Structural and mechanistic determinants of affinity and specificity of ligands discovered or engineered by peptide display. Annual Rev. Biophys. Biomol. Struct. 26, 27-45 · Kay BK and Paul JI, (1996), High-throughput screening strategies to identify inhibitors of protein-protein interactions. Mol. Divers. 1, 139-140)。 · Koide, A., Bailey, C.W., Huang, X. and Koide, S. (1998) The fibronectin type III domain as a scaffold for novel binding proteins. J Mol Biol 284, 1141。 ·Larsen, S. A., Meldgaard, T., Lykkemark, S., Mandrup, O. A. & Kristensen, P. (2015). Selection of cell-type specific antibodies on tissue-sections using phage display. J Cell Mol Med 19, 1939-1948。 ·Lehrer-Graiwer, J. et al. (2015). FDG-PET imaging for oxidized LDL in stable atherosclerotic disease: a phase II study of safety, tolerability, and anti-inflammatory activity. JACC Cardiovasc Imaging 8, 493-494。 ·Li, S. et al. (2013). Targeting oxidized LDL improves insulin sensitivity and immune cell function in obese Rhesus macaques. Mol Metab 2, 256-269。 ·Liu et al. (2000). Differential expression of cell surface molecules in prostate cancer cells. Cancer research; 60, 3429-3434。 ·Liu et al. (2004). Mapping tumor epitope space by direct selection of single-chain Fv antibody libraries on prostate cancer cells. Cancer Res 64, 704-710。 ·Ljungars et al. (2018). A platform for phenotypic discovery of therapeutic antibodies and targets applied on Chronic Lymphocytic Leukemia. NPJ Precision Oncology; 2, 18。 ·Ljungars et al. (2019). Deep Mining of Complex Antibody Phage Pools Generated by Cell Panning Enables Discovery of Rare Antibodies Binding New Targets and Epitopes. Frontiers in Pharmacology; 10, 847。 ·Lopez, T., Nam, D. H., Kaihara, E., Mustafa, Z. & Ge, X. (2017). Identification of highly selective MMP-14 inhibitory Fabs by deep sequencing. Biotechnology and bioengineering 114, 1140 - 1150。 ·Lundquist PM, et al. (2008). Parallel confocal detection of single molecules in real time. Optics Letters 33:1026 - 1028。 ·Margulies M, et al. (2005). Genome sequencing in microfabricated high - density picolitre reactors. Nature; 437:376 - 380。 ·Markland, W., Roberts, B. L., Saxena, M. J., Guterman, S. K. and Ladner, R. C. (1991) Design, construction and function of a multicopy display vector using fusions to the major coat protein of bacteriophage M13. Gene 109, 13。 ·Marks JD, Hoogenboom HR, Bonnert TP, McCafferty J, Griffiths AD and Winter G, (1991), By-passing immunisation. Human antibodies from V-gene libraries displayed on phage. J. Mol. Biol. 222, 581 - 597 ·Mattheakis, L. C., Bhatt, R. R. and Dower, W. J. (1994) An in vitro polysome display system for identifying ligands from very large peptide libraries. Proc Natl Acad Sci USA 91, 9022。 ·Mattsson, J. et al. (2021). Accelerating target deconvolution for therapeutic antibody candidates using highly parallelized genome editing. Nat Commun 12, 1277。 ·McCafferty, J., Griffiths, A. D., Winter, G. and Chiswell, D. J. (1990) Phage antibodies: filamentous phage displaying antibody variable domains. Nature 348, 552。 ·Moffat, J.G., Vincent, F., Lee, J.A., Eder, J. & Prunotto, M. (2017). Opportunities and challenges in phenotypic drug discovery: an industry perspective. Nat Rev Drug Discov 16, 531-543。 ·Mutuberria et al. (1999). Model systems to study the parameters determining the success of phage antibody selections on complex antigens. J Immunol Methods 231, 65-81。 ·Nixon et al(2019). A rapid in vitro methodology for simultaneous target discovery and antibody generation against functional cell subpopulations. Sci Rep 9, 842 ·Osbourn et al. (1998). Pathfinder selection: in situ isolation of novel antibodies. Immunotechnology 3, 293-302。 ·Parmley and Smith(1988). Antibody-selectable filamentous fd phage vectors: affinity purification of target genes. Gene, 73(2): 305-318。 ·Qin, H. et al. (2014). Generation of a new therapeutic peptide that depletes myeloid-derived suppressor cells in tumor-bearing mice. Nat Med 20, 676-681。 ·Ridgway, J.B. et al. (1999). Identification of a human anti-CD55 single-chain Fv by subtractive panning of a phage library using tumor and nontumor cell lines. Cancer Res 59, 2718-2723。 ·Roghanian, A. et al. (2015). Antagonistic human FcgammaRIIB(CD32B) antibodies have anti-tumor activity and overcome resistance to antibody therapy in vivo. Cancer Cell 27, 473-488。 ·Rothe, C. et al. (2008). The human combinatorial antibody library HuCAL GOLD combines diversification of all six CDRs according to the natural immune system with a novel display method for efficient selection of high-affinity antibodies. J Mol Biol 376, 1182-1200。 ·Sandercock et al(2015). Identification of anti-tumour biologics using primary tumour models, 3-D phenotypic screening and image-based multi-parametric profiling. Mol Cancer 14, 147 · Schiopu, A. et al. (2007). Recombinant antibodies to an oxidized low-density lipoprotein epitope induce rapid regression of atherosclerosis in apobec-1(- / -) / low-density lipoprotein receptor(- / -) mice. J Am Coll Cardiol 50, 2313 - 2318。 · Sebba, A. (2008). Tocilizumab: the first interleukin-6-receptor inhibitor. Am J Health Syst Pharm 65, 1413 - 1418。 · Semmrich et al(2022). Vectorized Treg-depleting aCTLA-4 elicits antigen cross-presentation and CD8(+) T cell immunity to reject ’cold’ tumors. J Immunother Cancer 10。 · Shendure J, et al. (2005). Accurate multiplex polony sequencing of an evolved bacterial genome. Science; 309: 1728 - 1732。 · Siegel et al. (1997). Isolation of cell surface-specific human monoclonal antibodies using phage display and magnetically-activated cell sorting: applications in immunohematology. J Immunol Methods 206, 73 - 85 · Smith, G.P. (1985) Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 228, 1315。 · Smith, G.P. and Scott, J.K. (1993) Libraries of peptides and proteins displayed on filamentous phage. Methods Enzymol 217, 228。 · Soderlind et al. (2000). Recombining germline-derived CDR sequences for creating diverse single-framework antibody libraries. Nat Biotechnol 18, 852 - 856。 · Stahl et al. (1989) A dual expression system for the generation, analysis and purification of antibodies to a repeated sequence of the Plasmodium falciparum antigen Pf155 / RESA. J Immunol Methods 124, 43。 · Swinney, D.C. (2013). Phenotypic vs. target-based drug discovery for first-in-class medicines. Clin Pharmacol Ther 93, 299 - 301。 · Swinney, D.C. & Lee, J.A. (2020). Recent advances in phenotypic drug discovery. F1000Res 9。 ·van den Brink, E.N. et al. (2005). Molecular and biological characterization of human monoclonal antibodies binding to the spike and nucleocapsid proteins of severe acute respiratory syndrome coronavirus. J Virol 79, 1635 - 1644。 ·Veitonmaki, N. et al. (2013). A human ICAM-1 antibody isolated by a function-first approach has potent macrophage-dependent antimyeloma activity in vivo. Cancer Cell 23, 502 - 515。 ·Waldmann, H. et al. (1984). Elimination of graft-versus-host disease by in-vitro depletion of alloreactive lymphocytes with a monoclonal rat anti-human lymphocyte antibody (CAMPATH-1). Lancet 2, 483 - 486。 ·Weng, S., Gu, K., Hammond, P.W., Lohse, P., Rise, C., Wagner, R.W., Wright, M.C. and Kuimelis, R.G. (2002) Generating addressable protein microarrays with PROfusion covalent mRNA-protein fusion technology. Proteomics 2, 48。 ·Williams, B. R. and Sharon, J. (2002) Polyclonal anti - colorectal cancer Fab phage display library selected in one round using density gradient centrifugation to separate antigen - bound and free phage. Immunol Lett 81, 141。 ·Williams, G. S. et al. (2016). Phenotypic screening reveals TNFR2 as a promising target for cancer immunotherapy. Oncotarget 7, 68278 - 68291。 ·Winter and McCafferty(1996)Phage display of peptides and proteins: a laboratory manual Ed Kay, Academic Press, Inc ISBN 0 - 12 - 402380 - 0 ·Winter, G., Griffiths, A. D., Hawkins, R. E. and Hoogenboom, H. R. (1994) Making antibodies by phage display technology. Annu Rev Immunol 12, 433。

[0244] Next, embodiments of the present invention will be described in the following numbered paragraphs. 1. A method for isolating at least one anti - ligand against at least one differentially expressed target ligand in a target cell, tissue, or sample from a library of anti - ligands, comprising: (a) providing one or more reference enrichment signatures for the library of anti - ligands used; (b) performing one or more rounds of differential biopanning on the library of anti - ligands to generate an anti - ligand pool; (c) A step of performing high-throughput sequencing on the anti-ligand pool generated during step (b) to generate a discovery enrichment signature for each anti-ligand in the anti-ligand pool, and a method comprising the same. 2. The method is (d) Further comprising comparing one or more reference enrichment signatures provided in step (a) with the discovery enrichment signature for the anti-ligands generated in step (c) to isolate at least one anti-ligand for at least one differentially expressed target ligand in the target cell, tissue, or sample of interest, the method according to paragraph 1. 3. The method according to paragraph 1 or paragraph 2, wherein one or more reference enrichment signatures provided in step (a) are reference enrichments derived in silico. 4. The reference enrichment signature derived in silico is generated using an equation derived from the universal law of mass action, and optionally, the reference signature derived in silico is the following equation:

[0245]

Number

[0246]

Number

[0247]

Number

[0248]

Number

[0249]

Number

[0250]

Number

[0251]

Number

[0252]

Number

[0253]

Number

[0254]

Number

[0255]

Number

Claims

Claim 1 A method for isolating at least one anti-ligand from a library of anti-ligands against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest, comprising: (a) providing one or more reference enrichment signatures for the library of anti-ligands to be used; (b) performing one or more rounds of differential biopanning on the library of anti-ligands to generate an anti-ligand pool; (c) performing high-throughput sequencing on the anti-ligand pool generated during step (b) to generate a discovery enrichment signature for each anti-ligand in the anti-ligand pool; and (d) comparing the one or more reference enrichment signatures provided in step (a) with the discovery enrichment signatures for the anti-ligands generated in step (c) to isolate at least one anti-ligand against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest. Claim 2 The one or more reference enrichment signatures provided in step (a) are in-silico-derived reference enrichments, optionally, the in-silico-derived reference enrichment signatures are generated using equations derived from the universal law of mass action, and optionally, the in-silico-derived signatures are the following equations: 【Number 1】 (wherein FA Ｔ = Frequency of the recovered antibody rA Ｔ = Number of antibodies recovered 【Number 2】 HR = hit rate, the proportion of antibodies specific to the target cell, tissue, or sample (determined experimentally)) are used to generate, Optionally, the in-silico-derived reference enrichment signatures are for anti-ligands against highly expressed ligands of the target cell, tissue, or sample of interest, for anti-ligands against moderately expressed ligands of the target cell, tissue, or sample of interest, or for anti-ligands against lowly expressed ligands of the target cell, tissue, or sample of interest. The method according to claim 1. Claim 3 The one or more reference enrichment signatures provided in step (a) are reference enrichment signatures from experiments, and optionally, the reference enrichment signatures from experiments are generated from at least one biopanning experiment comprising at least one reference anti-ligand, and optionally, the reference enrichment signatures from experiments are for anti-ligands against highly expressed ligands of a target cell, tissue, or sample, for anti-ligands against moderately expressed ligands of a target cell, tissue, or sample, or for anti-ligands against lowly expressed ligands of a target cell, tissue, or sample. The method according to claim 1.

4. In step (a), two or more reference enrichment signatures are provided, one or more of the reference enrichment signatures are reference enrichment signatures from in silico, one or more of the reference enrichment signatures are enrichment signatures from experiments, and optionally, the reference enrichment signatures from in silico are generated using equations derived from the universal law of mass action, and optionally, the enrichment signatures from in silico are generated using the following equation: [Number 3] (where FA Ｔ = Frequency of the recovered antibody rA Ｔ = Number of antibodies recovered 【Number 4】 HR = hit rate, the proportion of antibodies specific to a target cell, tissue, or sample (determined experimentally)) and / or optionally, the reference enrichment signatures from experiments are generated from at least one biopanning experiment comprising at least one reference anti-ligand. The method according to claim 1.

5. The reference enrichment signatures from in silico are for anti-ligands against highly expressed ligands of a target cell, tissue, or sample, for anti-ligands against moderately expressed ligands of a target cell, tissue, or sample, or for anti-ligands against lowly expressed ligands of a target cell, tissue, or sample, and / or the reference enrichment signatures from experiments are for anti-ligands against highly expressed ligands of a target cell, tissue, or sample, for anti-ligands against moderately expressed ligands of a target cell, tissue, or sample, or for anti-ligands against lowly expressed ligands of a target cell, tissue, or sample. The method according to claim 4.

6. The biopanning step (a) comprises (i) a sub-step of providing a library of anti-ligands, (ii)providing a first population of ligands comprising a ligand immobilized or incorporated in a subtractor ligand construct; (iii)providing a second population of ligands comprising a ligand immobilized or incorporated in a target ligand construct; (iv)determining the amounts of the subtractor ligand construct and the target ligand construct in said populations using one or more equations 【Number 5】 derived from the universal law of mass action; wherein A, B, C, and D = participants (reactants and products) in the reaction a, b, c, and d = coefficients necessary for the balanced chemical reaction equation (to enable isolation of the anti-ligand against the differentially expressed target ligand) (v)providing the amount of the subtractor ligand construct determined in step (iv); (vi)providing the amount of the target ligand construct determined in step (iv); (vii)providing separating means for isolating the anti-ligand bound to the target ligand construct from the anti-ligand bound to the subtractor ligand construct; (viii)exposing the library of (i) to the ligand constructs provided by (v) and (vi) to enable binding of the anti-ligand to the ligand; (ix)isolating the anti-ligand bound to the ligand immobilized or incorporated in the target ligand construct using said separating means, comprising optionally, the equation in step (iv) being 【Number 6】 wherein bA = number of receptor-bound anti-ligand A = total number of anti-ligand A B = total number of target receptor B K ｄ = anti-ligand affinity (M) V = reaction volume (dm ３ ) N Ａ = Avogadro's constant (6.022 × 10 ２３ particle moles －１ )) or 【Number 7】 wherein bA = number of receptor-bound anti-ligand A = total number of anti-ligand A B = total number of target receptor B K ｄ = anti-ligand affinity (M) V = reaction volume (dm ３ ) N Ａ = Avogadro's constant (6.022 × 10 ２３ particles per mole －１ ) C Ｔ = number of target cells C Ｎ = number of non-target cells B Ｔ = C Ｔ The number of receptors above B Ｎ = C Ｎ The method according to any one of claims 1 to 5, which is any of the above receptor numbers).

7. The method according to any one of claims 1 to 6, wherein step (b) comprises performing two or more rounds of biopanning, three or more rounds of biopanning, or four or more rounds of biopanning.

8. The differentially expressed ligand is a highly expressed ligand of the target cell, tissue, or sample of interest, an intermediate expressed ligand of the target cell, tissue, or sample of interest, or a lowly expressed ligand of the target cell, tissue, or sample of interest, the method according to any one of claims 1 to 7.

9. The method according to any one of claims 6 to 8, wherein the ligand is not expressed in either the target construct or the subtractor construct.

10. The method according to any one of claims 1 to 9, wherein the method further comprises the step of releasing the anti-ligand from the ligand.

11. The method according to any one of claims 6 to 10, wherein steps (ii) to (ix) are performed in parallel to isolate a plurality of anti-ligands for a plurality of different ligands and / or steps (ii) to (ix) are repeated one or more times.

12. The method according to any one of claims 6 to 11, wherein the amount of one of the subtractor construct or the target construct is provided in excess of the amount of the other of the subtractor construct or the target construct and / or the excess of the ligand is 10 to 1000 times, or 2 to 10 times, or 1000 to 1,000,000 times.

13. The method according to any one of claims 1 to 12, wherein the high-throughput sequencing in step (c) is performed using 454 sequencing, Illumina, SOLiD method or Helicos system.

14. The method according to any one of claims 6 to 13, wherein the separating means is selected from at least one of a solid support, a cell membrane and / or a part thereof, a synthetic membrane, beads, a chemical tag and a free ligand, or fluorescence-activated cell sorting.

15. The method according to any one of claims 6 to 14, wherein step (ix) is performed by at least one of density centrifugation, solid support isolation, magnetic bead isolation, chemical tag binding and aqueous phase partitioning, fluorescence-activated cell sorting.

16. The library of anti-ligands is a display library comprising a plurality of library members displaying anti-ligands, optionally, the library is a phage display library and / or the library of anti-ligands is constructed from at least one of an antibody, an antigen-binding variant, derivative, and / or fragment thereof; a scaffold molecule having an engineered variable surface; a receptor; and an enzyme, optionally, the library of anti-ligands is a pool of anti-ligands generated by a previously performed anti-ligand library selection experiment. The method according to any one of claims 1 to 15.

17. The method according to any one of claims 1 to 16, wherein the library of anti-ligands comprises at least one anti-ligand having a known copy number, and optionally, at least one anti-ligand having a known copy number is added to the library of anti-ligands at the known copy number.

18. The method according to any one of claims 1 to 17, wherein the ligand is at least one selected from an antigen; a receptor ligand; and an enzyme target comprising at least one of a carbohydrate, a protein, a peptide, a lipid, a polynucleotide, an inorganic molecule, and a conjugated molecule, optionally, the ligand is a cell surface receptor, and optionally, the cell surface receptor is in its native form.

19. A method for isolating at least one anti-ligand from a library of anti-ligands against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest, comprising: (a) providing one or more reference enrichment signatures for anti-ligands present in the library of anti-ligands used; (b) providing one or more discovery enrichment signatures for anti-ligands present in the library of anti-ligands used; (c) comparing the one or more reference enrichment signatures provided in step (a) with the one or more discovery enrichment signatures provided in step (b) to isolate at least one anti-ligand against at least one differentially expressed target ligand in a target cell, tissue, or sample of interest.

20. The one or more reference enrichment signatures provided in step (a) are in silico-derived reference enrichments, optionally, the in silico-derived reference enrichment signatures are generated using an equation derived from the universal law of mass action, and optionally, the in silico-derived enrichment signatures are generated using the following equation, 【Number 8】 wherein FA Ｔ = Frequency of recovered antibody rA Ｔ = Number of antibodies recovered 【Number 9】 HR = hit rate, the proportion of antibodies specific for the target cell, tissue, or sample (determined experimentally) Optionally, the in-silico-derived reference enrichment signature is for an anti-ligand against a highly expressed ligand of a target cell, tissue, or sample of interest, for an anti-ligand against an intermediate expressed ligand of a target cell, tissue, or sample of interest, or for an anti-ligand against a lowly expressed ligand of a target cell, tissue, or sample of interest, the method according to claim 19. **Claim 21** The one or more reference enrichment signatures provided in step (a) are in-silico-derived reference enrichment signatures, and optionally, the in-silico-derived reference enrichment signatures are generated from at least one biopanning experiment comprising at least one reference anti-ligand, and optionally, the in-silico-derived reference enrichment signatures are for an anti-ligand against a highly expressed ligand of a target cell, tissue, or sample of interest, for an anti-ligand against an intermediate expressed ligand of a target cell, tissue, or sample of interest, or for an anti-ligand against a lowly expressed ligand of a target cell, tissue, or sample of interest, the method according to claim 19. **Claim 22** In step (a), two or more reference enrichment signatures are provided, one or more of the reference enrichment signatures are in-silico-derived reference enrichment signatures, one or more of the reference enrichment signatures are in-silico-derived enrichment signatures, and optionally, the method according to claim 35, the in-silico-derived reference enrichment signature is generated using an equation derived from the universal law of mass action, and optionally, the in-silico-derived enrichment signature is generated using the following equation, 【Number 10】 (wherein, FA Ｔ = Frequency of the antibody recovered rA Ｔ = number of antibodies recovered 【Number 11】 HR = hit rate, the proportion of antibodies specific to the target cell, tissue, or sample (determined experimentally)) and / or optionally, the in-silico-derived reference enrichment signature is generated from at least one biopanning experiment comprising at least one reference anti-ligand, the method according to claim 19. **Claim 23** The in-silico-derived reference enrichment signature is for an anti-ligand against a highly expressed ligand of a target cell, tissue, or sample of interest, for an anti-ligand against an intermediate expressed ligand of a target cell, tissue, or sample of interest, or for an anti-ligand against a lowly expressed ligand of a target cell, tissue, or sample of interest, and / or The method according to claim 22, wherein the reference enrichment signature from the experiment is for an anti-ligand against a highly expressed ligand of a target cell, tissue, or sample of interest, for an anti-ligand against an intermediate expressed ligand of the target cell, tissue, or sample of interest, or for an anti-ligand against a low expressed ligand of the target cell, tissue, or sample of interest.

24. The method according to any one of claims 19 to 23, wherein the one or more discovery enrichment signatures provided in step (b) are derived from a previously conducted anti-ligand library selection experiment.

25. The at least one differentially expressed ligand is a highly expressed ligand of a target cell, tissue, or sample of interest, an intermediate expressed ligand of the target cell, tissue, or sample of interest, or a low expressed ligand of the target cell, tissue, or sample of interest, the method according to any one of claims 19 to 24.

26. The library of anti-ligands is a display library comprising a plurality of library members displaying anti-ligands, optionally, the library is a phage display library, and / or the library is constructed from at least one of an antibody, an antigen-binding variant, derivative, and / or fragment thereof; a scaffold molecule having an engineered variable surface; a receptor; and an enzyme, optionally, the library of anti-ligands is a pool of anti-ligands generated by a previously conducted anti-ligand library selection experiment, the method according to any one of claims 19 to 25.

27. The library of anti-ligands comprises at least one anti-ligand having a known copy number, optionally, the at least one anti-ligand having a known copy number is added to the library of anti-ligands at the known copy number, the method according to any one of claims 19 to 26.

28. The ligand is at least one selected from enzyme targets including an antigen; a receptor ligand; and at least one of a carbohydrate, protein, peptide, lipid, polynucleotide, inorganic molecule, and conjugated molecule, optionally, the ligand is a cell surface receptor, optionally, the cell surface receptor is in its native form, the method according to any one of claims 19 to 27.

29. An anti-ligand enrichment signature generated by step (c) of the method according to any one of claims 1 to 18. **Claim 30** A method for preparing a pharmaceutical composition, the method comprising the step of adding an anti-ligand isolated by the method according to any one of claims 1 to 28 to a pharmaceutically acceptable carrier. **Claim 31** A pharmaceutical composition prepared by the method according to claim 30. **Claim 32** A pharmaceutical composition prepared by the method according to claim 31 for use in medicine.

Images