The present invention relates to the field of biotechnology. In particular, the present invention relates to a large antibody phage display library, a process for arriving at the library and a method of producing antigen-specific Fabs through the said library.

BACKGROUND OF THE INVENTION

Antibody discovery was enabled by immunizing a non-human host species to generate polyclonal antibodies from serum or generate monoclonal antibodies by the hybridoma technology (Kohler G and Milstein C, 1975). The process of antibody discovery by immunization is however dependent on the uncertainties in the biology of immunization and results in full length non-human IgGs only. Further, such non-human IgGs are recognized as foreign by our immune system resulting in anti-species antibodies. The advent of technologies of grafting the antigen-recognizing rodent variable domains on human constant domains (-ximAbs) or grafting the complementarity-determining regions from rodents on human variable region frameworks (-zumabs) as well as the technology of mice transgenic for human Ig loci have reduced the problem of generating anti-species antibodies. Nonetheless, this method of human antibody generation, particularly when it comes to toxic antigens or protein targets that are highly conserved in amino acid sequence between humans and rodents, is not preferred.

Another alternative to this technology is Antibody Phage Display Library (APDL). Antibody phage display is a technique that can be used for the production of antigen-specific antibodies. One of the early references of APDL pertains to George P Smith (1985), who first described the phage display concept. APDL uses in vitro recombinant antibody synthesis techniques that rely on cloning of immunoglobulin gene segments to create libraries of antibodies. In this technique, antibody genes or gene fragments are fused to phage genes, thus allowing the antibody genes to be expressed and displayed as fusions to coat proteins on phage surfaces.

The ability to generate large repertoires of human antibodies combined with total display of the expressed repertoire allows the selection of individual antibodies with a desired specificity after biopanning against a target molecule. With this technique, tailor-made antibodies can be synthesized and selected to acquire the desired affinity of binding and specificity for in vitro and in vivo diagnosis, or for immunotherapy of human disease. This technique is particularly useful for production of antibodies for difficult antigens such as proteins homologous to human proteins, an area where conventional antibody producing methods of immunization often fail.

The antibody format commonly used for phage display include single -chain variable fragment (scFv) and Fragment antibody binding (Fab) fragments. scFv are monovalent structures that comprises the V region of the heavy and light chains of antibody molecules fused into a single polypeptide chain via a short flexible linker. Each Fab displays a single antigen binding site and consists of an Fd chain (VH -CHI -hinge fragment) and a light chain (VL -CL) bonded to each other at the C-terminus by a disulfide bond.

Prior art discloses certain scFv libraries and synthetic libraries. For instance, WO 1992/001047

Al discloses an antibody library in scFv format. Libraries with sizes of 10 7' and 2x107' cfu constructed from a naive human IgM source, and 5x10 cfu from naive human IgG source cloned in phagemid vector pHENl in scFv format are disclosed. Secretion of scFv and Fab antibody fragments from a phagemid vector transformed in amber suppressor host HB2151 and their detection in Westerns using polyclonal or tag-specific antibodies is disclosed. However, WO'047 does not disclose as to whether such secreted scFvs or Fabs from bacterial cultures can bind to an antigen of interest or not.

US6794128 B2 discloses a scFv phage antibody library of 7.0xl09 members. It further discloses selection of library on antigens and confirmation of the binders by phage ELISA. However, it appears that there are significant differences between the number of samples screened for such antigen-specific ELISA and the number of corresponding hits. Furthermore, the proportion of conversion of hits to water-soluble scFvs capable of recognizing the cognate antigen is very low.

US6696248 Bl discloses a fully synthetic human combinatorial antibody library (HuCAL) based on consensus sequences of the human immunoglobulin repertoire. The format used is the scFv

7 8

format and the size of the libraries varied between 10 and greater than 10 members. US'248 discloses yields of 5-10mg/L of purified scFvs but affinities of such scFvs are not disclosed.

US2005/0119455 Al discloses (i) single chain Fv (scFv); (ii) single chain Fv with zipper domain (scFvzip); (iii) Fab fragment (Fab); and (iv) Fab fragment with zipper domain (Fabzip) fused to the C-terminal domain of the minor coat protein (gill). Further, the polypeptide is expressed as a fusion protein on the surface of viral particle where the heavy chain variable domain is fused to a portion of a viral coat protein. US'455 further discloses display of 1011 polypeptide sequences or antibody variable sequence in the scFv format. Construction of F(ab')2 libraries with L3/H3 diversity and both Fab and F(ab' )2 libraries with H1/H2/H3 diversities are also disclosed but it does not provide any method to arrive at such libraries and size of the libraries are also not set out in this document.

US2005/0164180 Al discloses a method for generating a dAB (single heavy domain antibody fragment) phage library derived from human monoclonal antibody BT32/46 having VH portion of the said monoclonal antibody mutated in the CDR regions for enhanced expression. The description discloses a library size of 2.4x10° cfu.

KR2009/100961392 discloses a method for producing an antibody phage surface display library using the VH3-23/VLlg gene as a framework, and the phage display vector pFDV. The scFv gene library consisting of sublibraries of AE (2.1xl09 cfu), BE(2.7xl09 cfu), CF(l. lxl09 cfu) and DF( 1.7xl09 cfu) was obtained by sequential polymerisation between the cleaved sublibraries and pFDV vector at Sfi\ sites. Borate extracts of periplasm from single colonies of strain ER2537 from panning rounds for performing ELISA and declaring hits are disclosed, however score ratio and affinity estimates for such hits are not disclosed.

US2009/0054254 Al provides a method for generating immunoglobulin libraries by isolating RNA from a subset of B cells. It discloses a method for generating an immunoglobulin library, comprising the steps of: a) isolating a subset of B cells essentially of IgM memory B origin; b) isolating RNA from this subset of B cells; c) converting the isolated RNA into cDNA; d) amplifying immunoglobulin sequences of cDNA; e) inserting the amplified immunoglobulin

sequences into vector, and f) transforming a host cell with the vector containing the amplified sequences to obtain an immunoglobulin library. The format disclosed is both scFv and Fab, although the size of the library disclosed is for the scFv format only (10 cfu).

US201 1/236372 Al discloses synthetic antibody libraries in scFv format. The description discloses total library size to be higher than 109 individual clones, of which -80% were determined to be capable of secreting scFv in bacterial supernatant.

Thus, the prior arts are predominantly drawn to methods of producing antibody phage display library in scFv format. However, these methods do not disclose an ultra-large naive library in the Fab format. Using Fab format for producing antibody phage display library offer certain advantages when compared to the other formats of phage display such as single chain Fv (scFv). The advantage of Fab phage display library is that antigen-selected Fabs out of such libraries have high stability as water-dissolved proteins. In contrast, antibodies in single chain fragment variable (scFv) format have the tendency to form aggregates and are relatively unstable over longer periods of time (Weidner KM et al, 1992; Holliger P et al, 1993; Kortt et al, 1997; Quintero-Hernandez V et al, 2007). Furthermore, the scFvs may show a reduced affinity of up to one order of magnitude compared to the corresponding Fab fragments.

Certain prior art disclose naive human Fab libraries. For instance, EP2067788 A2 discloses a naive human Fab library with 4.3xl010 individual clones. EP'788 further discloses scoring antigen-specific hits using periplasmic extracts, although the method cited for hit screening from periplasmic extracts is not designed to allow periplasmic extraction of Fabs but scFvs.

From the available citations, it is evident that while it is easy to get libraries of a small size (10 -109 clones), it is not at all easy to obtain a library of a size that is the theoretical limit of mammalian immune diversity (1014 permutations). Furthermore, the chances of generating false positives or no binders at all remain very high due to the inherent flaws of the recombinant antibody assembly and discovery process, and also due to the biology of phages that tend to propagate non -recombinant (wild type) or partially recombinant phages (parasite phages) over

the recombinant ones. Therefore, the present invention is drawn so as to overcome this disadvantage.

Regardless of the library format, biopanning may be employed to select binders from antibody libraries. Biopanning may be conducted in vitro by immobilizing pure antigens on solid surfaces such as polystyrene, or by biotinylating the pure antigens and immobilizing them on streptavidin-coated polystyrene surfaces, followed by exposure to phages displaying the Fv domains in various formats. Biopanning may also be conducted in vitro by capturing the biotinylated antigens on streptavidin coated magnetic microbeads, followed by exposure to phages displaying the Fv domains in various formats. The latter approach has the advantage of being able to carry out panning in liquid phase, where the laws governing reaction equilibrium and kinetics may be applied far more confidently to pull out binders with desirable affinity and thermodynamic characteristics. Biopanning may also be carried out against target antigens present on the surface or inside a living cell, or against antigens such as cell surface receptors stabilized in lipid bilayers.

During the process of biopanning, the number of specific binders (binders) to a given antigen is a minuscule proportion of the entire gamut of non-specific binders (background) present in the phage population. Several rounds of panning are therefore required to enrich the specific binding subpopulation over the background. Furthermore, the small proportion of specific binders captured at each round of panning does require amplification of these binders by transduction in F+ hosts to be able to conduct the next round of panning. These amplification cycles however have the potential to propagate any genome (with phage replication ori) that has a growth advantage - phages containing a shorter length of genomes or translational stop codons within open reading frames appear to possess such an advantage. This biological fact can complicate the retrieval and analysis of genuine binders, particularly in case of phages derived from phagemid libraries where most phages are bald to begin with.

The various methods of biopanning discussed in prior art involves incubation of phage clones with target antigen followed by recovery of the bound phage to the target antigen by various elution strategies. Further, methods for affinity assessment by ELISA, Westerns and SPR has

been disclosed in the prior art, but the exact method for obtaining optimal yield of the antibodies against the target antigen with desired manufacturability characteristics is not disclosed.

Thus, the various existing antibody phage display libraries and the method of their production are described with diversity capture as their major goal, with little or no attention paid to the problems inherent in monoclonal lead identification or manufacturability aspects. Therefore, large antibody phages display libraries and the methods to produce antibodies reproducibly, confidently, and speedily out of them needs to be developed. The present invention provides methods to produce large and high diverse antibody phage display libraries which are commercially viable and could be produced in a short span of time period. The present invention provides novel methods of making large antibody phage display libraries as well as antibody retrieval from such libraries that are suitable for manufacture using established tools of biotechnology.

OBJECT OF THE INVENTION

The object of the invention is to create large antibody phage display library for its use as therapeutic and diagnostic purposes.

SUMMARY OF THE INVENTION

The present invention discloses a naive antibody phage display library (APDL) having a size ranging between 8.86 x 1010 to 9.13 x 1011 (3.06 x 1011) cfu, that includes 5.38 x 1010 to 2.55 x 1011 (1.26 x 1011) cfu kappa library and 7.33 x 1010 to 3.59 x 1011 (1.79 x 1011) cfu lambda library.

The present invention discloses a process for producing the APDL, wherein the immune repertoire capture comprises the steps of:

i) RNA isolation and cDNA synthesis;

ii) amplification of VL (lambda and kappa) and VH domains using primers comprising the SEQ ID 1-23 and 42-54;

iii) amplification of C domains using SEQ ID 24-26 and using primers comprising the

SEQ ID 27-31;

iv) overlap PCR of light chains by fusion of V and C domains and V and C domains obtained from step (ii) and (iii), respectively, using primers comprising the SEQ ID 30, 32, 35-37 and 55;

v) overlap PCR of heavy chains obtained from fusion of VH and CHI obtained from step (ii) and (iii) using primers comprising the SEQ ID 28 & 33;

vi) overlap PCR of light chains and heavy chains obtained from steps (iv) and (v) respectively to obtain Fabs using primers comprising the SEQ ID 32, 34, 35-37 and 55;

vii) purifying the amplicons at each step.

The present invention also discloses a method of obtaining manufacturable antibodies as soluble Fabs from the antibody phage display library in a defined order comprising the steps of: i) target specific panning;

ii) periplasmic quantitative ELISA (qELISA);

iii) kinetic ranking;

iv) bioassay;

v) manufacturability assessment;

resulting in a phenotype to genotype correlation of >90% so obtained after kinetic ranking.

BRIEF DESCRIPTION OF FIGURES

Figure 1 depicts quality check of total RNA extracted from isolated PBMC's.

Figure 2 depicts variable and constant gene amplifications in Sso Fast Evagreen. Panels A & B show amplification and melt curve peaks, respectively. Lanes 2, 3 and 4 of panel C contain PCR products of lambda, kappa and heavy constant synthetic genes as a template (20ng/25μl reaction) with respective constant gene primers viz. SEQ ID 29/SEQ ID 30, SEQ ID 31/SEQ ID 30 and SEQ ID 27/SEQ ID 28. Lanes 5, 6 and 7 contain PCR products of cDNA (20ng/25μl reaction) as a template with respective lambda, kappa and heavy variable primers viz. SEQ ID 14/SEQ ID 23, SEQ ID 9/SEQ ID 13 and SEQ ID 1/SEQ ID 7. Lane 1 contains lkb plus DNA marker from Fermentas. PCR products were run on 1.2% gels cast in lxTBE containing 0.0^g/ml ethidium bromide, and run at 5V/cm for 90 min.

Figure 3 depicts testing of amplifiability of human antibody variable genes using an optimized PCR buffer and DNA polymerase combination. After pre -heating at 94°C for 5min, reactions were cycled 30 times with a denaturation step at 94°C for 15s, simultaneous annealing and extension step at 72°C for 45s, followed by an extension and nick-sealing step at 72°C for lOmin. Panel A: Heavy V-genes; Lanes 3, 5, 7, 9, 11 and 13 contain products from six forward primers (SEQ ID 1-6) with reverse primer SEQ ID 7, while lanes 2, 4, 6, 8, 10 and 12 contain their respective Negative (No Template) controls. Lanes 15, 17, 19, 21, 23 and 25 contain products from the same forward primer set with reverse primer SEQ ID 8. Lanes 14, 16, 18, 20, 22 and 24 contain their respective Negative (No Template) controls. Panel B: Kappa V- genes; Lanes 3, 5, 7 and 9 contain products from four kappa forward primers (SEQ ID 9-12) with reverse primer SEQ ID 13, while lanes 2, 4, 6, and 8 contain their respective Negative (No Template) controls. Lane 10 was empty. Panel C: Lambda V-genes; Lanes 3, 5, 7, 9, 11, 13, 15, 17 and 19 contain products from nine forward primers (SEQ ID 14-22) with reverse primer SEQ ID 23, while lanes 2, 4, 6, 8, 10, 12, 14, 16, and 18 contain their respective Negative (No Template) controls. Lane(s) 1 contains lkb plus DNA marker (Fermentas). All PCR products were run on 1.2% gels cast in lxTBE containing 0.0^g/ml ethidium bromide, and run at 5V/cm for 90 min.

Figure 4 depicts re-testing of optimized amplification conditions of all V primer pairs. All the amplifications were carried out using Pfu Ultra II HS and PCR Extender buffer. Primer pairs are SEQ ID 14-22 with reverse primer SEQ ID 23. Input cDNA was 50ng per 50μ1 reaction. After pre-heating at 94°C for 5min, reactions were cycled 30 times with a denaturation step at 94°C for 15s, simultaneous annealing and extension step at 72°C for 45s, followed by an extension and nick-sealing step at 72°C for lOmin. Even numbered lanes contain products derived from respective V forward primers mentioned above the lanes with reverse primer SEQ ID 23. Odd numbered lanes contain their respective negative (No template) controls. M is the Generuler lkb plus DNA marker from Fermentas. All PCR products were run on 1.2% gels cast in lxTBE containing 0.0^g/ml ethidium bromide, and run at 5V/cm for 90 min.

Figure 5 depicts amplification by all V primers. Reactions in Panel A were amplified using Pfu Ultra II HS polymerase and Pfu buffer. Lanes 2 and 4 contain products derived from primers

SEQ ID 9 and 10 respectively with reverse primer SEQ ID 13, while lanes 1 and 3 contain their respective negative (No template) controls. Panel B and C show enzyme -buffer matrix to get better amplification by SEQ ID 11 and 12. Reactions in both panels were amplified using PCR Extender polymerase blend. Panel B shows amplification by SEQ ID 11 while Panel C shows amplification by SEQ ID 12, both paired with reverse primer SEQ ID 13. After pre -heating at 95°C for 5 min, reactions were cycled 30 times with a denaturation step at 94°C for 15s, annealing at 60°C for 30s, extension at 72°C for 30s, followed by an extension and nick-sealing step at 72°C for lOmin. Lanes 2, 4, 6, and 8 contain PCR products with Advantage 2 buffer, Advantage 2 SA buffer, PCR Extender buffer and Tuning buffer respectively. Lanes 1, 3, 5, and 7 contains their respective negative (No template) controls. Input cDNA was 50ng per 50μ1 reaction for all panels. M is the Generuler lkb plus DNA marker from Fermentas. All PCR products were run on 1.2% gels cast in lxTBE containing 0.0^g/ml ethidium bromide, and run at 5V/cm for 90 min.

Figure 6 depicts testing of amplification of VH genes using an optimized PCR buffer and DNA polymerase combination with all Scripps VH primer pairs. Top panel exhibits reactions carried out in Advantage 2 SA buffer, while the bottom panel exhibits reactions carried out in Tuning buffer. Enzyme used was Pfu Ultra II HS polymerase for all reactions. Lanes 1 to 12 contain all 6 VH sense primers (SEQ ID 1-6) with reverse primer SEQ ID 7. Lanes 13 to 24 contain all 6 VH sense primers with reverse primer SEQ ID 8. Input cDNA amount per 50μ1 reaction was 50ng. Even numbered wells contain products from respective primer pairs mentioned above them, while odd numbered wells contain their respective negative (No template) controls. M is the Generuler lkb plus DNA ladder from Fermentas.

Figure 7 depicts application of optimized enzyme -buffer matrix to check amplification of all constant domain primers (CK, Cx, CH)- All reactions were amplified using Pfu Ultra II HS polymerase and PCR Extender buffer. Input DNA concentration was 50ng per 50μ1 reaction. Lane 1 contains product derived from SEQ ID 26 as template that was amplified using primer pair SEQ ID 29/SEQ ID 30, lane 2 contains product derived from SEQ ID 25 as template that was amplified using primer pair SEQ ID 31/SEQ ID 30, while lane 3 contains product derived from SEQ ID 24 as template that was amplified using primer pair SEQ ID 27/SEQ ID 28. M is

the Generuler lkb plus DNA marker from Fermentas. All PCR products were run on a 1.2% gel cast in lxTBE containing 0.0^g/ml ethidium bromide, and run at 5V/cm for 90 min.

Figure 8 depicts input DNA concentration versus varying polymerase concentration matrix to improve 1st overlap PCR of V C . All the reactions were carried out using Advantage 2 polymerase and Advantage 2 SA buffer. Input DNA was equimolar for purified V and C product per 50μ1 reaction. The primer pair was SEQ ID 32/SEQ ID 30. Panels A, B, and C show results for 10, 25, and 50ng input DNA, respectively. Even numbered lanes contain products from respective enzyme concentrations (0.25x, 0.5x, 0.75x and l.Ox) mentioned above the lanes whereas odd numbered lanes contain their respective negative (No template) controls. M is the Generuler 1 kb plus DNA marker (Fermentas).

Figure 9 depicts enzyme buffer matrix for the optimization of 1st overlap product of VHCHI . Panels A, B, C, D, E, and F show PCR products in Advantage 2 buffer, Advantage 2 SA buffer, Exact polymerase buffer, PCR Extender buffer, Tuning buffer and Vent buffer, respectively. Primer pair for VHCHI overlap is SEQ ID 33/SEQ ID 28. Input DNA concentration for VH was 50ng, and equimolar amount of Cjjl was used for the overlap reactions. Lanes 2, 4, 6, 8, 10, 12 and 14 contain products derived from Advantage 2 polymerase, Exact polymerase, Pfu Ultra II HS enzyme, AmpliTaq polymerase, PCR Extender, Vent polymerase and Deep Vent polymerase, respectively. Lanes 1, 3, 5, 7, 9, 11, and 13 contain their respective negative (No template) controls. M is the Generuler 1 kb plus DNA marker from Fermentas.

Figure 10 depicts SOE PCR megaprimer strategy with enzyme buffer matrix for amplification of final Fab product. Initial 15 cycles were carried out without addition of primers with 50ng each of the 1st overlap products. After 15 cycles, 30 more cycles were carried out with addition of primer pair SEQ ID 32 and SEQ ID 34. Panels A, B, C, and D show amplification in Advantage 2 SA buffer, Expand LT buffer, PCR Extender buffer and Thermopol buffer respectively. Lanes 2, 4, 6 and 8 show amplification products derived by using Advantage 2 Polymerase mix, Expand LT Polymerase, PCR Extender Enzyme and Deep Vent Polymerase, respectively, while lanes 1, 3, 5 and 7 contain their respective negative (No template) controls. M is the Generuler 1 kb plus DNA marker from Fermentas.

Figure 11 depicts 2-step PCR strategy with enzyme buffer matrix for amplification of final Fab product. Primer pair was SEQ ID 32 and SEQ ID 34 with 50ng each of the 1st overlap products. Panels A, B, C, and D show amplification in Advantage 2 SA buffer, Expand LT buffer, PCR Extender buffer and Thermopol buffers, respectively. Lanes 2, 4, 6 and 8 show amplification products derived from Advantage 2 polymerase mix, Expand LT polymerase, PCR Extender Enzyme and Deep Vent polymerase respectively, while lanes 1, 3, 5 and 7 contain their respective negative (No template) controls. M is the Generuler 1 kb plus DNA marker from Fermentas.

Figure 12 depicts circular plasmid map of pCOMB3XSS.

Figure 13 depicts Sfil digestion of pCOMB3XSS vector. Panel A shows S zI-digested pCOMB3XSS vector and stuffer fragments before band-cutting while Panel B shows the same gel after the desired bands were cut out. Lane 2 contains uncut pCOMB3XSS while lanes 4 to 15 contain S zI-digested pCOMB3XSS. The upper band is the 3.3kb vector backbone while lower band is the 1.6 kb stuffer fragment. lOU^g of Sfil enzyme was used for overnight digestion at 50°C. M is the 1 kb plus DNA marker from Fermentas.

Figure 14A depicts effect of heat inactivation of ligation mix. Panel A shows the number of transformants from the 1 :0.35 ligation mix without any heat inactivation, while panel B shows the number of transformants for the same 1 :0.35 ligation mix with heat inactivation treatment. For transformation, neat (undiluted) Ιμΐ from either the untreated or the heat inactivated ligation mixes were electroporated per 25μ1 of TGI cells, and 1 , 10 and ΙΟΟμΙ of the culture was plated after electroporation. Data are shown only for the 1 :0.35 ligation ratio for convenience. Other two ligation ratios (1 : 1 and 1 :3.5) also follow the same trend but with much more matted growth. Figure 14B depicts calculation of vector background. Panel A shows vector control ligation in which only 140ng vector was added without any insert, while panel B shows the plate of 1 : 1 heat inactivated ligation mix. Data are shown for the Ιμΐ plating volume only.

Figure 15 depicts TOPO-Fab clone confirmation by Sfil digestion. All reactions contain ^g of miniprep plasmid from respective TOPO-Fab clones. Digestions were carried out in NEB buffer

4 with 5U of Sfil per μg DNA in 20μ1 reactions. Numbers above the lanes indicate the respective clone numbers. Samples were run in 1% analytical grade agarose gel in lxTBE with 0Λμg/m ethidium bromide at 5V/cm for 1.5h. M is the lkb plus DNA marker from Fermentas. Arrow indicates the 1.5kb Fab band released after Sfil digestion.

Figure 16 depicts schematic illustrating the concept of self-circularization to release Fabs with ligatable Sfil ends.

Figure 17 depicts Sfil digestion of linear Fabs by self-ligation strategy. The Fab used was of single type amplified from plasmid isolated from a TOPO-Fab clone. Lanes 2, 3 and 4 contain ^g each of PCR amplified Fab product, Fab after self-ligation and Fab after Sfil digestion, respectively. Products were analyzed on a 1% agarose gel prepared in lxTBE containing O.^g/ml ethidium bromide and run at 5V/cm for 1.5h. M is the lkb plus DNA marker from Fermentas. Numbers at the left show marker size in base pairs.

Figure 18 depicts results of the first self-ligation test on a single population of Fab. Panel A shows Vector control plate while Panel B shows that for the self-ligated Fab. Panel C is the stuff er control. Plates from Ιμΐ plating volume are shown.

Figure 19 depicts results of the self-ligation test on a diverse population of Fabs. Panel A shows vector control plate whereas Panel B shows the lambda Fab pool prepared by the self-ligation strategy. Panel C shows the stuffer control plate.

Figure 20 depicts characterization of self-ligation library clones. Panel A shows PCR amplification for the kappa self-ligation clones while Panel B shows the same for lambda self-ligation clones. ΙΟμΙ of PCR products were loaded in each lane. Panel C shows Sfil digestion pattern for the kappa self-ligation clones while Panel D shows Sfil digestion pattern for the lambda self-ligation clones. All lanes in Panels C & D contain ^g of Sfil digested plasmid DNA isolated from respective clones. Numbers above the lanes indicate clone numbers. Analysis was done in 1% agarose gels prepared in lxTBE containing 0Λμg/m ethidium bromide and run at

5V/cm for 1.5h. M is the lkb plus DNA marker from Fermentas. Numbers on the left show marker size in base pairs.

Figure 21 depicts BstNl analysis of kappa and lambda Fab clones. All lanes contain 40μ1 of the BstNl digestion reaction. Top panel shows the fingerprint of kappa Fabs while the lower panel shows lambda Fab fingerprint. Images represent ethidium bromide stained 3% agarose gels cast in lxTBE, and run at 4V/cm for 3h. Numbers above the lanes indicate the respective clone numbers. M is the lkb plus DNA marker (Fermentas). Numbers on the left indicate marker size in base pairs.

Figure 22 depicts ClustalW report of all sequenced lambda Fabs. Nucleotide sequences of lambda Fabs were compared against pCOMB3XSS to verify presence and intactness of Sfil on both 5' and 3' ends.

Figure 23 depicts the process of improving the self-ligation method by increasing DNA concentration per unit volume and use of PEG 8000. Lane 2 contains the phosphorylated Fab pool alone, while lanes 4, 5 and 6 contain samples of ligation mixtures after 16h of incubation at 16°C. Equal amounts (2^g) of all were loaded for visual comparison. Lane 4 contains sample from the standard 83ng/ 1 ligation reaction, while lane 5 contains the standard reaction supplemented with 6% PEG. Lane 6 samples a ligation reaction that contained 200ng/ 1 of DNA, supplemented with 6% PEG. Lanes 1 and 3 are empty. M is the lkb plus DNA marker from Fermentas. Numbers on the left show marker size in base pairs. Panel on the left shows the epifluorescent image of a 1% agarose gel prepared in lxTBE containing 0.^g/ml ethidium bromide and run at 5V/cm for 1.5h. Image on the right is the photographic negative of the same.

Figure 24A depicts photographic evidence of the Cell to DNA ratio titration experiment. Plates from the 1 :25000 dilutions are shown. Figure 24B depicts graphical representation of the cell to DNA ratio titration experiment listed in Table 37.

Figure 25 depicts self-ligation of linear Fabs for final large Fab library making. Panels A and B show the epifluorescent image for kappa and lambda Fabs, respectively, while panel C is the

photographic negative of the lambda Fab gel (panel B). Lanes 2 contain the 500ng of Fab pool alone, while lanes 4 contain ^g of sample of Fab self-ligation mixtures after 16h of incubation at 16°C. Lanes 6 contain Sfil digested salt purified self-ligated ligation mixture. Lanes 3 and 5 are empty. M is the lkb plus DNA marker from Fermentas. Numbers on the left show marker size in base pairs. Samples were run in a 1% agarose gel prepared in lxTBE containing O.^g/ml ethidium bromide and run at 5V/cm for 1.5h.

Figure 26 depicts sequences and paired alignment of the two final overlap primers SEQ ID 32 and SEQ ID 34. The upper panel shows the primer sequences and orientation along with SEQ ID. The portion aligning to the VL-CL or VH-CH1 template is in bold. The overhangs are in regular font. The Sfil and Sacl sites in SEQ ID 32 are italicized. The bottom panel shows a paired alignment of the two sequences using the Martinez-Needleman-Wunsch algorithm with a similarity index of 66.7%.

Figure 27 depicts conventional Sfil digestion of PCR assembled kappa Fab pool. All PCR overlap products digested with Sfil were run on 1.2% gels cast in IxTAE containing IxSYBR safe™, and run at 5V/cm for 90 min. Top panels show images of these gels after the run, while bottom panels show the same gels after cutting out the 1.5kb band of the Sfil digested Fab pool. The primers used for final Fab assembly, along with the length of the overhangs 5' of the annealing site, are indicated on the top of each column.

Figure 28 depicts circular plasmid map of pSSYl (SEQ ID 38).

Figure 29 depicts verification of the quality of prepared cDNA by testing variable gene amplifications. This figure shows test amplification by a V -specific reverse primer (SEQ ID 13) paired with all V forward primers (SEQ ID 42-45) and test amplifcation by a V reverse primer (SEQ ID 23) paired with all V forward primers (SEQ ID 46-54). Numbers on the top of the gel shows SEQ IDs of respective forward primers. Products were analyzed in 1.2% agarose gels prepared in lxTBE containing O. ^g/ml ethidium bromide and run at 5V/cm for 1.5h. M is the lkb plus DNA marker from Fermentas.

Figure 30 depicts V-family coverage of the ultra-large library.

Figure 31 depicts quality check of phage library by colony PCR from phage transductant clones. The Fab inserts were PCR-amplified from diluted culture of randomly selected clones using vector backbone primers. Products were analyzed on a 1% agarose gel prepared in lxTBE containing 0.1 g/mL ethidium bromide and run at 5V/cm for 1.5h.

Figure 32 depicts BstNl fingerprinting of P04 clones (from 3 round panned pools against target antigen). The digests were run on 3% agarose gel to analyze the restriction pattern. Agarose gel in invert mode is shown for better contrast between banding patterns. Products were run in lxTBE buffer containing O. ^g/mL ethidium bromide in 3% agarose gel prepared in same buffer for 2.5h at 6V/cm. Numbers on top of lanes indicate respective repeat pattern clones. Number at the left shows respective DNA marker positions in bp.

Figure 33 depicts periplasmic Western of Fabs. Western blots of periplasmic extracts were obtained from a panning campaign such as described in Examples 26 and 27 and probed with a HRP-conjugated rabbit polyclonal IgG (Jackson ImmunoResearch#309-036-003). Arrow indicates the presumed ~50kDa Fab heterodimer, while the * indicates the presumed 23 or 27 kDa monomer (light or heavy chain). Box on top indicates the clone number along with the L VH families to which these clones belong. Numbers along the left edge of each image represent molecular weight of the markers in kDa (Pre-stained All-Blue SDS-PAGE marker; BioRad).

Figure 34 depicts phage pool ELISA. Panel A shows simultaneous comparison amongst phage pools derived from the naive library, round 1 and round 2 solution panning eluates. Phages were incubated with 500nM biotinylated antigen in round 1 for lh at RT, followed by various concentrations of the same antigen in round 2 under the same incubation conditions. Black bars indicate reactivity against the target antigen immobilized on Polysorp wells while grey bars indicate reactivity of the same pool against human serum albumin (HsSA) immobilized in parallel wells. Higher reactivity in antigen-specific wells suggests enrichment of antigen-specific binders by the 2nd round, although bait dose dependent enrichment is not seen. Panel B shows the same simultaneous comparison amongst phage pools derived from the naive library, round 1 and round 2 solution panning eluates as in panel A, with the exception that phages were allowed to incubate with the antigen for 16h at RT in round 2. Bait dose dependence is apparent under longer incubation conditions from round 2. Panel C shows the same simultaneous comparison amongst phage pools derived from the naive library, round 1 and round 2 solution panning eluates as in panels A and B, with the exception that phages were incubated with ΙΟηΜ biotinylated antigen for lh at RT before addition of ΙΟΟηΜ non-biotinylated antigen for various lengths of time in round 2. The persistence of enrichment over 2h suggests presence of high affinity (slow dissociation) binders in such pools.

Figure 35 depicts antigen-specific ELISA. Four 96-well plates are shown in this screenshot. Odd-numbered wells were coated with 2μg/ml human serum albumin while even-numbered wells were coated with the same amount of target antigen. Whole cell extracts prepared as described in the text from monoclonal recombinants were incubated at optimized dilutions in these pre-coated wells in duplicates, such that one aliquot of the dilution is pipetted into the odd-numbered well while the other aliquot is pipetted into the even-numbered well of the pair. Bound Fabs were detected with a human Fab-specific polyclonal serum (Jackson ImmunoResearch 309-036-003). Clones that show antigen-specific reactivity at least 2-folds over the paired nonspecific antigen well are highlighted and bolded.

Figure 36 depicts the periplasmic gate. The image is a composite of 30 Western blots of periplasmic extracts from a solution panning campaign. The blots were probed with a HRP-conjugated rabbit polyclonal IgG (Jackson ImmunoResearch 309-036-003). Arrows indicate the presumed ~50kDa Fab heterodimer and the presumed 23 or 27 kDa monomer (light or heavy chain). Labels on top of each lane indicate the clone number. Numbers along the left edge of each image represent molecular weight of the markers in kDa (Pre-stained All-Blue SDS-PAGE marker; BioRad).

Figure 37 depicts the confounding nature of a polyclonal detection antibody for Fab hit identification.

Figure 38 depicts the revised periplasmic gate. The image is a composite of 33 Western blots of periplasmic extracts from a solution panning campaign as in Example 30. The blots were probed with a HRP-conjugated mouse monoclonal anti-HA IgG (clone 3F10; Roche). Arrows indicate the presumed ~50kDa Fab heterodimer and the presumed 23 or 27 kDa monomer (light or heavy chain). Labels on top of each lane indicate the clone number. Numbers along the left edge of each image represent molecular weight of the markers in kDa (Pre-stained All-Blue SDS-PAGE marker; BioRad).

Figure 39 depicts periplasmic subtyping by Western. The image is a composite of 16 Western blots of periplasmic extracts from a solution panning campaign as in Example 30. The blots were probed with a HRP-conjugated anti-kappa or anti-lambda monoclonal (Sigma). Arrows indicate the presumed ~50kDa Fab heterodimer and the presumed 23 or 27 kDa monomer (light or heavy chain). Labels on top of each lane indicate the clone number. Numbers along the left edge of each image represent molecular weight of the markers in kDa (Pre-stained All-Blue SDS-PAGE marker; BioRad).

Figure 40 depicts the in-frame versus off-frame clone experiment. Western analysis of periplasmic extracts from 3 deliberately tandem in-frame clones versus 3 deliberately off-frame (in the HC) clones in non-reduced conditions probed with anti-lambda (panel a), anti-kappa (panel b), anti-Cnl (panel c) and HRP-conjugated anti-Hu (H+L) F(ab')2 fragment antibodies (panel d).

Figure 41 depicts the chain-switch phenotyping concept to weed out false Fab protein hits away from true hits.

Figure 42 depicts example of a fitted curve and back prediction from the Fab chain-switch quantitation ELISA. Panel A shows the 4-parameter fitted curve for mass of input standards on the X-axis versus output A450 values on the Y-axis while Panel B shows the estimates of the fit parameters as well as the goodness of the fit as indicated by the probability score from an F-test. Panel C shows the input Fab concentrations, the raw A450 values and the back-predicted concentrations from the fitted curve.

Figure 43 depicts discrimination of in-frame clones and off-frame clones by qELISA. Panel A shows A450 values, while Panel B shows the same data as a bar graph for easy visualization of the output. Panel C shows the fitted standard curve for this experiment, while Panel D shows the 4-PL fit parameters and the goodness-of-fit.

Figure 44 depicts summary of maximum response units possible from various SPR Fab capture surfaces. Experiments were performed using ProteOn XPR36 instrument (BioRad) on GLC, GLM or NLC chips using recombinant standard human Fab or PPE-Fab. Running buffer was 20mM PBS, pH 7.4 and 0.05% Tween-20 or lOmM HEPES with 0.05% Tween-20. Capture antibodies viz. polyclonal anti-Fab, anti-His, anti-HA or 1 : 1 mixture of bivalent anti-CHl/anti-or anti-CHl/anti- were immobilized vertically on respective chip. 1:2 to 1: 10 dilution of test Fabs and 5μg/ml of standard Fabs were captured on respective horizontal channels for 180s-300s at 25μ1/ηΉη flow rate. If required, two consecutive captures were performed to increase the capture level and, surfaces were stabilized by H3PO4 injection for 18s at ΙΟΟμΙ/min. The sensorgrams were referenced appropriately and the Fab capture levels were noted. The bivalent anti-Fab antibodies resulted into maximum Fab capture level (800-2000RU) compared to rest of the antibodies (70-600RU).

Figure 45 depicts SPR based kinetic analysis of VEGF165 interaction with purified Fab' fragment of Bevacizumab. Fab capture method was validated using the purified Fab' fragment of Bevacizumab as the ligand and VEGF165 as the analyte. Running buffer was PBS, pH 7.4 containing either physiological salt/0.005% Tween-20 OR 0.5M NaCl/0.05% Tween-20. Panel A shows the SPR profile for troubleshooting of non-specific binding (NSB) of analyte (VEGF165) to the capture surface. Increasing the Tween-20 and salt concentrations in running and sample buffers drastically reduced the NSB. Panel B shows the binding curves at different capture levels of the Fab' fragment along with their respective residuals plots. Panel C shows derived kinetic values (ka, kd and KD) and other relevant parameters at each capture level.

Figure 46 depicts SPR based kinetic analysis of VEGF165 interaction with expressed BevacizuFab present in crude periplasmic extracts. Fab capture method was validated using BevacizuFab as the ligand in periplasmically expressed Fab format and VEGF165 as the analyte.

Optimized conditions as described in Figure 45 were used. Panel A shows the binding curves of BevacizuFab whereas Panel B shows its respective residual plot. Panel C shows corresponding kinetic values (ka, k and ¾) and other relevant parameters at each capture level (compare with panel C of Figure 45). PPE-Fab LB and PPE-Fab MM refer to periplasmic extracts of BevacizuFab grown in different culture media.

Figure 47 depicts the funnel of antibody discovery from a Fab library.

Figure 48 depicts b-TNF purity and quality check. Samples were heated at 90°C for 5min and 5μg each of reduced and non-reduced proteins were electrophoresed in a 4-15% polyacrylamide gel at 150V for 40min. The image on right shows the gel stained in Coomassie Brilliant Blue R-250 for 2h and de-stained using water: methanol: acetic acid (50:40: 10) for 2h. M is Precision plus All Blue SDS-PAGE marker from Biorad - NR is for non-reduced, R is for reduced. For Western blotting, ΙΟΟηΜ (50ng) and 30nM (~16ng) reduced protein samples were prepared and electrophoresed as previous and transferred to nitrocellulose membrane at 100V for 1.5h. Blot was blocked for lh in 3% BSA in TBST (0.05% Tween-20) and probed with streptavidin-HRP (Dako# P0397) at 1:40000 dilution in 3% BSA for lh. Blot was developed using Clarity Western ECL substrate (Biorad#170-5060). Image of the resultant Western blot (left) was recorded using the ChemiDoc XRS system (Biorad).

Figure 49 depicts representative data for anti-TNF soluble Fab screening using chain switch qELIS A. Upper panel shows layout of plate 6 of the TNF campaign with clone numbers 481 to 576. Middle panel shows standard curve and A450 values using lambda detection antibody whereas bottom panel shows standard curve and A450 values using kappa detection antibody. Black color indicates high expressing clones, dark grey color indicates moderate expressing clones while light grey color indicates low expressing clones.

Figure 50 depicts kinetic screening profiles of SPR positive clones of anti-TNF at 500nM analyte concentration. Experiments were performed using ProteOn XPR36 instrument (BioRad) on neutravidin-coated (NLC) chips. Running buffer was 20mM PBS, pH 7.4 with 0.5M salt and 0.05% Tween-20. Capture antibody was 1: 1 mixture of biotinylated bivalent anti-CHl/anti- and anti-CHl/anti- antibodies. Three different concentrations (10, 3 and ^g/ml) of this mixture were immobilized vertically in duplicate; one as a test surface while other as reference surface for respective capture concentration on the NLC chip. 1: 10 dilutions of test Fabs were captured on respective horizontal channels for 300s at 25μ1/ηΉη flow rate. Two to three consecutive captures were performed to saturate the capture surface. One horizontal channel was dedicated as reference channel where non-specific (non-TNF binder) commercial human Fab was used to saturate the surface such that the reference surface exactly mimics the test surface. Before the analyte injection, the baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for 5min after the 1st buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. 500nM of analyte (sTNF ) was injected horizontally at 25μ1/ηιήι for 120s (2min) followed by dissociation for 300s (5min). Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. The sensorgrams were referenced appropriately and analyzed using Langmuir 1: 1 fitting models. Resultant affinity constant values (ka, kd, ¾), and other relevant parameters like Rmax and were noted.

Figure 51 depicts schematic of design and results of epitope binning of 10 anti-TNF SPR positive clones. In the first experiment, five Fabs viz. bTl, bT16, bT38, bT59 and bT75 were immobilized on horizontal channels 1 to 5 of a NLC chip, respectively. Surfaces were saturated using three consecutive injections of test Fabs for 300s at 25μ1/ηΉη. Analyte (sTNF ) was injected next in vertical direction to interact with these Fabs and to block the respective target epitopes. Lastly, same five Fabs were flown again over these surfaces, but this time vertically to see the interaction pattern. Second experiment was performed similarly using set of next five clones viz. bT76, bT77, bT84, bT86 and bT88. In the third and final experiment, previous two sets of five clones were tested similarly with each other. (V) mark indicates positive SPR response while (X) mark indicates negative SPR response. Tables below each experimental schematic show bins generated for respective combination of clones.

Figure 52 depicts summarized view of SPR profiles and parameters of anti-TNF monoclonal Fabs bTl, bT59 and bT88. Biotinylated bivalent anti-CHl/anti-κ and anti-CHl/anti-λ capture antibody was immobilized at three different concentrations (10, 3 and ^g/ml) vertically in

duplicate; one as a test surface while other as the reference surface for respective capture concentration on NLC chip. 1 : 10 dilutions of test Fabs were captured on three vertical channels (LI, L3 and L5) for 300s at 25μ1/Γηίη flow rate. Two to three consecutive captures were performed to saturate the capture surface. Reference surfaces (L2, L4 and L6) were saturated using non-specific (non-TNF binder) commercial human Fab to exactly mimic the test surfaces. Before the analyte injection, baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for lOmin after first buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. Five concentrations (reciprocal dilution) of sTNF ranging between 10nM-0.625nM for bTl and ΙΟΟΟρΜ to 62.5pM for bT59 and bT88 were injected horizontally at 25μ1/ηιήι for 900s (15min) followed by dissociation for 900s (15min). Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. For data analysis, last three concentrations were considered i.e. 2.5nM-0.625nM for bTl whereas for bT59 and bT88, the range used was 250pM to 62.5pM. The sensorgrams were referenced appropriately and analyzed using Langmuir 1: 1 fitting models. Resultant affinity constant values (ka, kd, ¾), and other relevant parameters like Rmax and were noted as required.

Figure 53 depicts /Rh5 purity and quality check. Samples were heated at 90°C for 5min and 2μg each of reduced and non-reduced proteins were electrophoresed in a 4-15% polyacrylamide gel at 150V for 40min; gel was stained in Coomassie Brilliant Blue R-250 for 2h and de-stained using water: methanol: acetic acid (50:40: 10) for 2h. M is Precision plus All Blue SDS-PAGE marker from Biorad - NR = non-reduced, R = reduced.

Figure 54 depicts representative data for anti- Rh5 soluble Fab screening using chain switch qELISA. Upper panel shows layout of plate 11 of the Rh5 campaign with clone numbers 193 to 288. Middle panel shows standard curve and A450 values using lambda detection antibody whereas bottom panel shows standard curve and A450 values using kappa detection antibody. Black color indicates high expressing clones, dark grey color indicates moderate expressing clones while light grey color indicates low expressing clones.

Figure 55 depicts kinetic screening profiles of SPR positive clones of anti- /Rh5 at 500nM analyte concentration. Biotinylated bivalent anti-CHl/anti- capture antibody was immobilized at 3μg/ml on LI to L3 channels and similarly anti-CHi/anti- was immobilized on L4 to L6 channels of a NLC chip in the vertical direction. 1 :5 dilutions of a set of five test Fabs at a time were captured in horizontal direction for 300s at 25μ1/ηΉη flow rate. Two consecutive captures were performed to saturate the capture surface. Reference surfaces (sixth horizontal channel) was saturated using non-specific (non-P/Rh5 binder) commercial human Fab to exactly mimic the test surfaces. Before the analyte injection, baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for 5min after first buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. Single concentration of 500nM of Rh5 was injected horizontally on all six horizontal channels at 25μ1/Γηίη for 120s (2min) followed by dissociation for 300s (5min). Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. The sensorgrams were referenced appropriately and analyzed using Langmuir 1: 1 fitting models.

Figure 56 depicts summarized view of SPR profiles and parameters of anti-P/Rh5 monoclonal Fabs. Biotinylated bivalent anti-CHl/anti-λ capture antibody was immobilized at 3μg/ml on LI to L3 and similarly anti-CHl/anti-κ was immobilized on L4 to L6 channels of a NLC chip in the vertical direction. 1:5 dilutions of test Fab were captured on four vertical channels; LI and L2 for lambda clones while channels L3 and L4 for kappa clones for 300s at 25μ1/ηΉη flow rate. Two consecutive captures were performed to saturate the capture surface. The reference surfaces (L3 and L6) were saturated using non-specific (non-Rh5 binder) commercial human Fab to exactly mimic the test surfaces. Before the analyte injection, baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for lOmin after first buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. Five concentrations (reciprocal dilution) of /Rh5 ranging between 500nM-31.25nM were injected in the horizontal direction for 600s (lOmin). Dissociation of bound Fabs to target antigens was carried out for 900s ( 15min) with running buffer. Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. For data analysis, the sensorgrams were referenced appropriately and analyzed using Langmuir 1: 1 fitting models. Resultant affinity constant values (ka, k&, KD) were noted as required.

Figure 57 depicts CSP purity and quality check. Samples were heated at 90°C for 5min and 5μg each of reduced and non-reduced proteins were electrophoresed in a 4-15% polyacrylamide gelat 150V for 40min; gel was stained in Coomassie Brilliant Blue R-250 for 2h and de-stained using water: methanol: acetic acid (50:40: 10) for 2h. M is Precision plus All Blue SDS-PAGE marker from Biorad - NR is for non-reduced, R is for reduced.

Figure 58 depicts representative data for anti-P CSP soluble Fab screening using chain switch qELISA. Upper panel shows layout of plate 8 of the CSP campaign with clone numbers 289 to 384. Middle panel shows standard curve and A450 values using lambda detection antibody whereas bottom panel shows standard curve and A450 values using kappa detection antibody. Black color indicates high expressing clones; dark grey color indicates moderate expressing clones while light grey color indicates low expressing clones.

Figure 59 depicts kinetic screening profiles of SPR positive clones of anti-P CSP at 500nM analyte concentration (depicted as two parts 59A and 59B). 1: 1 mixture of biotinylated bivalent anti-CHl/anti-K and anti-CHl/anti-λ capture antibodies were immobilized at three different concentrations (10, 3 and ^g/ml) vertically in duplicate. 1:5 dilutions of a set of five test Fabs at a time were captured in horizontal direction for 300s at 25μ1/ηιιη flow rate. Two consecutive captures were performed to saturate the capture surface. Reference surface (sixth horizontal channel) was saturated using non-specific (non-P CSP binder) commercial human Fab to exactly mimic the test surfaces. Before the analyte injection, baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for 5min after first buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. Single concentration of 500nM was injected horizontally on all six horizontal channels at 25μ1/ηιιη for 120s (2min) followed by dissociation for 300s (5min). Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. The sensorgrams were referenced appropriately and analyzed using Langmuir 1: 1 fitting models.

Figure 60 depicts summarized view of SPR profiles and parameters of anti-P CSP monoclonal Fabs. The 1: 1 mixture of biotinylated bivalent anti-CHl/anti-κ and anti-CHl/anti-λ capture antibodies were immobilized at three different concentrations (10, 3 and ^g/ml) in duplicate on a NLC chip in the vertical direction. 1 :5 dilutions of test Fabs were captured on five vertical channels (LI to L5) for 300s at 25μ1/ιηίη flow rate. Two to three consecutive captures were performed to saturate the capture surface. Reference surface (L6) was saturated using nonspecific (non-P CSP binder) commercial human Fab to exactly mimic the test surfaces. Before the analyte injection, baseline was stabilized using three consecutive injections of running buffer at ΙΟΟμΙ/min for 60s. The system was paused for 5min after first buffer injection followed by the remaining two - this helps to stabilize the signal rapidly. Five concentrations (reciprocal dilution) of CSP ranging between 500nM-31.25nM were injected in the horizontal direction for 600s (lOmin). Dissociation of bound Fabs to target antigens was carried out for 900s (15min) with running buffer. Surfaces were regenerated using glycine pH 2.0 for 60s followed by second injection for 30s. For data analysis, the sensorgrams were referenced appropriately and analyzed using Langmuir 1 : 1 fitting models. Resultant affinity constant values (ka, k&, KD) were noted as required.

DETAILED DESCRIPTION

The present invention discloses a large antibody phage display library, a method for producing the antibody phage display library and a method of screening various antigens to obtain manufacturable antibodies against the said antigens.

The present invention discloses a large and diverse library in the Fab format which significantly increases the chances of identifying binding compounds with high affinity and high specificity for the target. Further, the invention offers easy isolation of target-specific soluble Fabs with minimal sequence engineering. The Fabs in the present invention are expressed as self -folded proteins in E. coli periplasm as output which is easier to detect and manufacture and further use for clinical or diagnostic purposes. The present invention offers the advantage of speed and cost-effectiveness as the process of constructing the library to panning and identification of lead monoclonal can be done in few weeks. The present invention is thus a high fidelity productive process for obtaining manufacturable antibodies and establishes a method of screening of various antigens to obtain such antibodies in a controlled manner.

The present invention discloses a process of antibody discovery that matches pre -set critical quality attributes for manufacturing arranged as a series of staged assessments starting from a large repertoire of antibody fragments displayed on phages - this process is exemplified and discussed in the Examples that follow. The benefits of this invention include phage displayed antibody fragments and are possible to extend to other in vivo display systems such as yeast display or bacterial display that also use antibody fragments for display and that are subject to the same stringent requirements of staged assessments, especially in high throughput format as water soluble proteins.

CQAs in Antibody Discovery: Industrial antibody discovery and manufacture necessitates monitoring of the protein product throughout the process (Alt N et al., 2016; Kepert JF et al., 2016). In other words, the binding, activating, agonistic or conjugating phenotypes that are predefined for a therapeutic target (reviewed in Labrijn AF et al., 2008) should be assignable to the protein moiety, so that the Critical Quality Attributes (CQA's) responsible for such phenotypes can be defined as early as possible. Such definitions usually include assessment of affinity, specificity against the target, and biological functionality as common aims, and may include additional assessment of productivity, tendency to aggregate, thermodynamic stability (Thiagarajan G et al., 2016) and potential immunogenicity (Hai S-H et al., 2009. Immunogenicity screening using in silico methods: Correlation between T-Cell epitope content and clinical immunogenicity of monoclonal antibodies. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic), all of which are properties directly assignable to the behaviour of an antibody as a protein dissolved in water, or are inherent in its structural features. The intent of defining such early stage CQAs is therefore to mitigate the risks innate in the mismatch between targeted and observed phenotype as well as the observed phenotype and the underlying genotype as early as possible. This invention demonstrates how to achieve this aim for antibodies discovered from naive human phage display library platforms.

Therapeutic or diagnostic antibody usage has very much depended upon the format available for discovering antibodies. Historically, antibody discovery was enabled by immunizing a non-human host species to generate polyclonal antibodies from serum (von Behring EA and Kitasato S, 1890), or generate monoclonal antibodies by the hybridoma technology (Kohler G and

Milstein C, 1975). The process of antibody discovery by immunization is therefore dependent on the uncertainties in the biology of immunization and results in full length non-human IgGs only. Such non-human IgGs are usually not suitable for human use as they are recognized as foreign by our immune system, resulting in anti-species antibodies that neutralize the therapeutic benefit after repeated use (Chester KA and Hawkins RE, 1995; Glennie MJ and Johnson PWM, 2000). Antibodies generated by the non-human immunization path therefore remained useful only as research or diagnostic reagents for many years until the technologies of grafting the antigen -recognizing variable domains on human constant domains (-ximabs; Liu AY et al., 1987) or grafting the complementarity-determining regions on human variable region frameworks (-zumabs; Jones PT et al., 1986; Carter P et al., 1992) were developed by industry pioneers. Use of mice transgenic for human Ig loci (Briiggemann M and Taussig MJ, 1997; Green LL, 1999; Lonberg N, 2008; Dechiara TM et al., 2009) has today largely bypassed the non-human Ig origin problem for generating therapeutic monoclonal antibodies from immunization and several antibodies have now been marketed from this technology (Panitumumab aka Vectibix®, Golimumab aka Simponi®, Canakinumab aka Ilaris®, Ustekinumab aka Stelara®, Ofatumumab aka Arzerra®, and Denosumab aka Xgeva/Prolia®). Nonetheless, this method of human antibody generation remains subject to the uncertainties of immunization, particularly when it comes to toxic antigens or protein targets that are highly conserved in amino acid sequence between humans and rodents (Frenzel A et al., 2016).

Full length IgGs have the in-built benefit of long half-life of circulation in blood or capability to engage immune effector cells or both by virtue of features encoded with the Fc domain, and therefore have a structural CQA advantage by design in many therapeutic scenarios. An additional advantage is embedded in the fact that as such IgGs are discovered as secreted proteins, attribution of the phenotypic qualities to the protein per se and assignment of CQAs to such proteins is straightforward. Linking the phenotypes to the underlying genotypes, which is a critical knowledge for manufacturing constancy, is also technically feasible for antibodies secreted by hybridomas on a routine basis (Bradbury A, 2010. Cloning hybridoma cDNA by RACE. In: Antibody Engineering; Vol. 1), although it remains a formidable technical challenge for polyclonal species secreted by a multitude of B -cells in vivo for routine use (Meijer PJ et al., 2006; Tiller T et al, 2008).

With the advent of systems capable of displaying antibody fragments in vivo or in vitro, bypassing the vagaries of immunization (antigen toxicity, antigen homology, lack of response) has become possible and the format for discovery is no longer confined to full length IgG alone. These systems depend upon the power of recombinant DNA technology, which in essence utilizes the modularity inherent in Ig protein structure as well as their genomic origin to recombine the antigen recognizing elements of immunoglobulin genes in vitro in a variety of forms. The source of such genes may be natural or synthetic. Upon appropriate capture of these diverse V-domain permutations in host cells of prokaryotic or eukaryotic origin and subsequent expression and display, these recombinant formats can bind to antigens in vitro, and such binders can then be isolated and sequenced easily to assign the binding phenotype to a finite genotype. The possibility of easy phenotype-to-genotype linkage therefore allows possibility of assignment of gross CQA's to the binders as proteins. Examples cited herein document that when it comes to obtaining antibodies from phage display, such assignment is not obvious and needed to be invented.

The present invention discloses a process in which immunoglobulin fragments may be discovered as secreted proteins in E. coli periplasm, and allows attribution of the phenotypic qualities to the protein per se. The present invention links the phenotypes to the underlying genotypes, which is a critical knowledge for manufacturing constancy.

The present invention also discloses a method of obtaining manufacturable antibodies as soluble Fabs from the antibody phage display library in a defined order comprising the steps of: i) target specific panning;

ii) periplasmic qELISA;

iii) kinetic ranking;

iv) bioassay;

v) manufacturability assessment;

resulting in a phenotype to genotype correlation of >90% in the antibodies so obtained after kinetic ranking.

Phage display technology rests on five key ideas: (a) bacteriophages can express heterologous peptides or polypeptides fused to their coat proteins when transduced in host bacteria; (b) given a multitude of such peptides or polypeptides, a library of recombinant phages can be created that display all these variants on a coat protein of choice; (c) such a library of phages can be screened as a population for ability to bind (recognize) a target molecule in vitro; (d) the binders can be separated from the non-binders by washing them away in a process akin to washing sand away from gold dust (panning), and (e) the isolated binders can be analyzed for the sequence of the variant encoded within its genome. Since its conceptualization (Smith GP, 1985), this technology has proven invaluable for a variety of investigations that includes antibody discovery, epitope mapping, protein interaction site mapping, enzyme substrate discovery and molecular evolution (reviewed in Burton DR, 1995; Azzazy HM and Highsmith WE, 2002).

The present invention discloses a method of obtaining antibodies from the antibody phage display library, wherein the panning is conducted in solid or solution phase at various temperatures ranging between 4 and 37°C and for various lengths of time ranging between lh and 16h. The solid phase panning may comprise the steps of:

i) optimizing the maximal coating concentration for a given antigen on a solid surface such as charged polystyrene;

ii) conversion of the phagemid library to phage format;

iii) coating the selected surface with the optimal concentration of the antigen as determined at step(i) followed by blocking with protein or non-protein molecules to block non-specific sites;

iv) pre-adsorption of phage pool as obtained at step(ii) on unblocked polystyrene surface to eliminate plastic binders;

v) incubation of pre-adsorbed phages from step(iv) with immobilized target antigen (step iii) for defined periods of time;

vi) multiple rounds of washings to eliminate unbound phages from step(v);

vii) elution of bound phages from step(v) by trypsin digestion and concurrent transduction in amber suppressor as well as non-amber suppressor hosts to obtain phage titers; viii) amplification of eluted phages from step (vii) by transducing in amber suppressor host for next round of panning;

ix) performing the next round of panning by using reduced antigen concentration and repeating steps (iii) to (viii) to enrich the target specific antibody population;

x) repetition of steps (vii) to (ix);

xi) evaluation of eluted phages from step (vii) and (x) for enrichment of binding over rounds of panning using target specific ELISA.

The solution phase panning may comprise the steps of:

xii) optimizing the reaction conditions for optimal biotinylation of a given antigen to achieve a biotin to protein molar ratio of <10, preferable 1-5;

xiii) conversion of the phagemid library to phage format;

xiv) blocking the phages obtained at step(ii) with protein or non-protein molecules to block non-specific sites for defined periods of time simultaneous with streptavidin bead washing followed by blocking the beads with protein or non-protein molecules to block non-specific sites;

xv) incubation of blocked phages from step(xiii) with soluble target biotinylated antigen (step xii) for defined periods of time;

xvi) incubation of phage-antigen complex obtained at step(xiv) with pre-blocked streptavidin beads;

xvii) multiple rounds of washings of the beads bound to antigen -phage conjugates at step(xv) to eliminate unbound phages;

xviii) elution of bound phages at step(xvi) by DTT or trypsin digestion and concurrent transduction in amber suppressor as well as non-amber suppressor hosts to obtain phage titers;

xix) amplification of eluted phages from step (xviii) by transducing in amber suppressor host for next round of panning;

xx) performing the next round of panning by using reduced antigen concentration and repeating steps (xiv) to (xviii) to enrich the target specific antibody population;

xxi) repetition of step (xix) to step (xx);

xxii) evaluation of eluted phages from step (xviii) and (xxi) for enrichment of binding over rounds of panning using target specific ELISA.

A very important concept embedded within the phage display technology is the physical linkage of the binding phenotype to the encoded genotype within the recombinant phage. In contrast, cDNA expression libraries can also encode a multitude of polypeptides, but the phenotype of a particular clone after a population-based screening can only be linked to its encoded genotype after a separate investigative step conducted on a parallel master set of clones (hybridization with radiolabeled probe and colony picking, for example; Sambrook J and Russell DW, 2001a. Preparation of cDNA libraries and gene identification. In: Molecular Cloning: A Laboratory Manual; Vol. 2). The speed and throughput that can be obtained from phage display libraries are therefore incomparably faster and higher compared to screening cDNA libraries.

When it comes to assessing antibodies as a protein however, the display technology loses its power, for in principle; the industrial proposition of must-assess-discovered-entity-as-protein puts it at a same advantage or disadvantage level as a cDNA library. Furthermore, the high throughput that is possible today for assessment of individuals of a variant population of proteins and that are applied to secreted IgGs from hybridomas on a routine basis for antibody discovery (Hay FC and Westwood OMR, 2002. Preparation of human B-cell hybridoma. In: Practical Immunology), are not possible to apply for phage displayed antibodies. The primary obstacle to high throughput with phage display technology is the necessity for transducing the growing bacterial cultures with infective phages and then harvesting the amplified phages, which involves centrifugation to separate the bacteria from the phages, a second step of polyethylene glycol (PEG) mediated precipitation to concentrate the phages, and then repeated washes to eliminate any bacterial contamination before beginning to assess any binding phenotype. To put this process in contrast, hybridomas can be cultured in 96-well plates and centrifuged to collect the supernatant, which can be directly assessed for binding phenotype (Green LL, 1999).

A secondary but critical obstacle rests on the fact that by design, phage displayed antibodies are linked N- or C-terminally to a much larger phage particle. The presence of such a large "tag" to the discovered antibody fragment can most certainly be predicted to influence the basic phenotype of binding, as has been documented in several reports (Lou J et al., 2001; Chowdhury PS, 2002. Targeting random mutations to hotspots in antibody variable domains for affinity improvement. In: Methods in Molecular Biology, Vol. 178: Antibody Phage Display: Methods and Protocols; Pavoni E et al., 2007). The inbuilt limitations of phage biology result in propagation of a majority of clones that (a) do not display the antibody fragments at all (Winter G et al., 1994; Azzazy HM and Highsmith WE, 2002; Hust M et al., 2009. Antibody phage

display. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic), (b) that are shorter in size compared to the expected size of the recombinant genomes (de Bruin R et al., 1999; Lowe D and Vaughan TJ, 2009. Human antibody repertoire libraries; In: Therapeutic Monoclonal Antibodies: From Bench to Clinic), and that (c) have the greatest growth advantage as opposed to the experimental objective of amplifying the antibody displaying recombinants only (de Bruin R et al., 1999; L0set GA et al., 2005; Lowe D and Vaughan TJ, 2009. Human antibody repertoire libraries; In: Therapeutic Monoclonal Antibodies: From Bench to Clinic). These inbuilt errors can result in situations where binding ability of a monoclonal phage to a target antigen is not mirrored by the binding ability of the protein antibody produced from the same clone (Vaughan TJ et al., 1996; Chowdhury PS, 2002. Targeting random mutations to hotspots in antibody variable domains for affinity improvement. In: Methods in Molecular Biology, Vol. 178: Antibody Phage Display: Methods and Protocols; US 6,794,128; Pavoni E et al., 2007), as would be expected from the principle of genotype -phenotype linkage. Examples cited herein document such phenomenon in great detail, and necessitate inventing methods that recognize the pitfalls of the process and therefore, can bypass them.

A common method of mitigating the risk that the binding phenotype shown by a phage -antibody fusion may not be mirrored by the antibody as a protein when expressed minus the phage tag is by inquiring whether the genome of the binder represents an expressible open reading frame (ORF) - usually achieved by sequencing (Buckler DR et al., 2008). Despite major advances in automation, such an approach, although invaluable in clone assessment, remains labor-intensive and cannot be construed as a high throughput method. Furthermore, sequencing per se does not guarantee whether an antibody ORF will be expressible. Rules for what kinds of antibody ORFs might be expressible in bacteria and mammalian cells have been suggested (Ewert S et al., 2004; Rothlisberger D et al., 2005) and synthetic antibodies libraries today depend upon such rules for their success (Knappik A et al., 2000; Rothe C et al., 2008; Prassler J et al., 2011; Tiller T et al., 2013). No such deterministic rules are obviously possible for binders derived out of a random combination of V-domains that represent a naive or immune phage display library. Therefore the present invention discloses a method that would allow assessment of antibody fragments as proteins secreted out from bacteria akin to the high throughput assessment of IgGs secreted out of hybridoma cultures. Such method as disclosed herein is one of the factors in enabling high

throughput. Certain prior art discloses such methods (Winter G et al, 1994., Kirsch M et al., 2005; L0set GA et al., 2005; Petropoulos K, 2012. Phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols; Rader C, 2012a. Selection of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), but these methods do not suggest further resolution of the genotype-phenotype dissociation expected from this approach. Examples cited herein therefore document a novel staged assessment system that continues to harnesses the essential power of the phage display technology (phenotype-linked-to-genotype) when screening en masse, but employs biochemical principles of analyzing antibodies as proteins when screening the phenotype of each clone (phenotype delinked from genotype), while maintaining the capability to assign the observed phenotype to the underlying genotype.

The process of the present invention achieves the objectives by examining and establishing the theory and process pertaining to the format of the antibody that can be displayed by the phages but that is also stable as an isolated protein in water, and the ability of the host bacteria to secrete such forms as protein, amongst other factors and processes. scFvs as proteins are known to be generally less stable and more prone to aggregation compared to Fabs when in an aqueous environment (Weidner KM et al, 1992; Holliger P et al, 1993; Kortt AA et al, 1997; Quintero-Hernandez V et al, 2007), although careful studies suggest that this an epiphenomenon directed by the given combination of VL- and VH-domains (Rothlisberger D et al, 2005) as well as the fact that the unfolding kinetics of Fabs is slower rather than any intrinsically enhanced thermodynamic stability (Honegger A, 2008). Since Fabs have higher intrinsic stability then scFvs, the present invention discloses the use of Fabs for construction of the said library. The use of Fabs is further advantageous since Fabs or pegylated Fabs can directly serve as therapeutic structures, which is preferred when considering the gain in reducing the time between discovery and manufacture. In contrast, scFvs can only serve as diagnostic or conjugated therapeutics due to their short life in blood circulation (Chames P et al, 2009) and therefore, generally need to be re-formatted for alternative therapeutic use. The loss of speed between discovery and manufacture can therefore be considerable. Furthermore, when considering the structural elements of Fab, the paratope (binding surface) created by the VL-VH interface in a Fab is a natural protein interaction domain while that created by the VL-VH interface in a scFv is

constrained by linker length and therefore, artificial (Kortt AA et al., 1997). Therefore, the affinity shown by a Fab towards a particular antigen would be driven by a natural VL-VH interface while that shown by a scFv by an unnatural one, which may have unpredictable consequences during reformatting of a candidate scFv to Fab or IgG formats. Fab reformatting to a therapeutic IgG format, on the other hand, would actually benefit from a stability gain (Casadevall A and Janda A, 2012).

In order to enable the host bacteria to secrete these stable forms as proteins, the present invention advantageously utilizes the design inherent in phage display technology that uses periplasmic space to direct phage production (Webster R, 2001. Filamentous phage biology. In: Phage Display: A Laboratory Manual). The present invention discloses a hybridoma-like screen in which periplasmic extracts from growing bacteria recombinant for antibody fragments may be examined for the presence, absence and relative yield of Fabs - enabling high throughput without the loss in fidelity of the present process. The application of a periplasmic Western for qualitative analysis of expressed scFvs has been reported in prior art (L0set GA et al., 2005; Eisenhardt SU and Peter K, 2010. Phage display and subtractive selection on cells. In: Antibody Engineering; Vol. 1), but the present application for the first time discloses such a process for screening Fabs as proteins out of a panned Fab display library. This approach also allows the crucial validation of whether the standard Fab design of two different periplasmic leaders for two different cistrons (light and heavy chains) have actually resulted in a heterodimeric Fab protein expressed in the periplasm. The advantage with this approach therefore is that it focuses attention and efforts to obtain functional Fab proteins that exist at a reasonable level of detection, and rejects clones that are at a low detection level, at an early discovery stage. However, this approach of expressing human Fabs as soluble proteins in E. coli periplasm can cause low yields (Better M et al., 1993; Humphreys DP, 2003). The proximate cause for low yields include misfolding in the periplasm (Skerra A and Pliickthun A, 1991; Humphreys DP, 2003) and presence of periplasmic proteases that are able to digest such misfolded polypeptides, light chains in particular (Chen C et al., 2004). The distal causes of misfolding may include the inbuilt limitation of differential codon usage between the host and guest species, as well as the family-specific VL-VH interface stability properties (Ewert S et al., 2004, Tiller T et al., 2013).

Certain methods are available for increasing Fab or Fab' yields in E. coli periplasm for synthetic library building or making grams of proteins for downstream purposes, however most of the methods in prior art have certain limitations (Humphreys DP and Bowering L, 2009. Production of antibody Fab' fragments in E. coli. In: therapeutic Monoclonal Antibodies: from Bench to Clinic; US 8,062,865; Tiller T et al., 2013), and none exist that can be applied for screening hundreds of Fabs as proteins from a naive library at an early discovery stage. The present invention discloses for the first time such assessment as a primary phenotypic screen to weed out clones that do not produce heterodimeric Fabs in the periplasm at the limit of detection for enhanced chemiluminescence based Westerns (1-3 pg/band). Hence the present invention advantageously benefits in terms of time and cost and overcomes the disadvantages of the need to handle poorly expressed clones, even though there is a possibility of losing some antigen specific binders. Another advantage of this gate is that one can type these periplasmic hits by simultaneous immunoblotting with anti-human kappa and lambda-specific antibodies, thus avoiding the need to sequence these clones when panning is done with mixed kappa and lambda libraries.

The present invention discloses methods/process to arrive at optimum set of methods and protocols to overcome the limitations in getting antigen-specific Fab protein binders with high fidelity. A limitation of the initial approach described above was that antibodies discovered as present in the periplasm and assumed to be heterodimeric were actually homodimeric for the light or heavy chains in many cases. The said limitation was overcome by utilizing the 2-site concept for ELISA development (Harlow E and Lane D, 1988. Immunoassays. In: Antibodies: A Laboratory Manual; Lawson ADG et al., 1997) to identify clones that are likely to produce heterodimeric Fabs. Examples included herein demonstrate that the present application has extended the basic 2-site concept to develop a novel chain-switch ELISA system that is not only able to distinguish clones producing heterodimeric Fabs away from clones producing homodimeric Fabs, but also allows to achieve a very crucial aim of estimating yields quantitatively in mass Fab/volume terms. This breakthrough improvement is a major step forward from the initial system of qualitative assessment by Westerns and demonstrates one of the inventive merits of the present application.

The present invention discloses a method of obtaining antibodies from the panned antibody phage display library, wherein the periplasmic qELISA comprises the steps of:

i) obtaining soluble Fabs from single bacterial colonies from eluate titer plates;

ii) coating the surface of 96-well charged polystyrene plates with a capture antibody against heavy chain;

iii) capturing the soluble Fab from step(i) on the coated surface of step(ii)

iv) detection of light chain by utilization of light chain specific antibody to identify full length, tandem in-frame, heterodimeric, soluble Fabs.

Furthermore, development of Fab protein assessment in an ELISA format immediately allows high throughput as potential binder clones out of a pool of binders from a panning campaign can be grown in 96-well plates, such cultures induced to produce Fabs in the periplasm (Kontermann RE, 2010. Immunotube selections. In: Antibody Engineering; Vol. 1; Hust M and Mersmann M, 2010. Phage display and selection in microtitre plates. In: Antibody Engineering; Vol. 1; Petropoulos K, 2012. Phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), the periplasm harvested by simple lysis in situ (W01/94585; Humphreys DP and Bowering L, 2009. Production of antibody Fab' fragments in E. coli. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic), and the supernatants harvested for assessment of Fabs as proteins in the same 96-well format akin to hybridoma culture supernatants. Replicas of these master cultures can be stored frozen as glycerol stocks (Petropoulos K, 2012. Phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols; Protocol: Use of Glycerol Stocks and Preparation of Transfection-Quality Plasmid DNA; Broad Institute, Boston, MA, 2015.), and clones with desirable protein properties regrown and interrogated most easily. None of these steps require further involvement of phage display.

The present invention discloses a method of obtaining soluble Fabs from the panned antibody phage display library, wherein obtaining the soluble Fabs comprises the steps of:

i) picking single clones from titer plates of non-amber suppressor hosts and liquid culture in 96-well deepwell plates for overnight growth at 37°C and 250 rpm;

ii) diluting the overnight cultures 10-folds and allowing growth to log phase under conditions identical to (i);

iii) inducing the log phase cultures at step (ii) with ImM IPTG and allowing overnight growth at 30°C and 250 rpm;

iv) centrifuging the cultures at step (iii) in 96-well plates to pellet down the induced cells; v) periplasmic extraction of the pelleted cells at step (iv) by using high concentrations of EDTA in a buffered solution while slowly shaking the buffer-suspended cells in the same 96-well plate overnight at 30°C;

vi) centrifugation to isolate the diffused periplasmic fraction at step (v) away from the spheroplast and cell debris.

Prior art discloses certain high throughput screening of monoclonal Fabs as soluble proteins after obtaining binder pools from a solution panning campaign (Petropoulos K, 2012. Phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols; Rader C, 2012a. Selection of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols). However, a major disadvantage with the protocol described by Petropoulos K, 2012, is that it requires re-cloning of the phagemid DNA of potential monoclonal binders as a pool into an expression vector before screening as soluble Fabs. Such a requirement will surely result in significant slowdown in speed, plus the potential of losing binders during the subcloning process. Furthermore, the protocol makes no effort to isolate the periplasmic fraction away from the cytoplasmic fraction, which is a disadvantage insofar as the periplasmic translocation property of the Fab cannot be validated. Lastly, the prior art protocol uses an anti-Fab polyclonal antibody for detecting but not quantifying the Fab yield. In contrast, the invention of the present application (a) does not require re -cloning of binder pools by taking advantage of the amber stop codon in between the heavy chain and the gill and use of non-amber suppressor hosts (also see the section on pSSYl in this context), (b) utilizes a gentle periplasmic isolation method that allows separation of the periplasmic fraction away from the remainder of the E. coli fractions, and (c) uses the chain switch concept to unambiguously qualify and quantify heterodimeric Fabs only.

The protocol set out in the Rader C, 2012a, uses the concept of 2-site ELISA by immobilizing light chain polyclonal antibodies on polystyrene plates, capturing crude Fab preparations on such antibodies, and detecting with a heavy chain C-terminal tag. However, the process is not developed into a quantitative assay. As described, the process is low throughput (14ml tubes used for sampling 32 clones), and actually samples Fab-pIII fusions isolated from a simple centrifugation of induced cultures that is unlikely to isolate a fraction enriched in periplasmic proteins. In contrast, the present application exemplified herein (a) uses a high throughput culture, induction and storage method very similar to that described (Petropoulos K, 2012. Phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), (b) takes advantage of the amber stop codon in between the heavy chain and the gill and use of non-amber suppressor hosts to produce Fabs without the pill (also see the section on pSSYl in this context), (c) takes care to use a gentle periplasmic isolation method that allows separation of the periplasmic fraction away from the remainder of the E. coli fractions and (d) uses the chain switch concept to unambiguously qualify and quantify heterodimeric Fabs only without any pill fusion.

The present application further discloses a method of obtaining antibodies from the antibody phage display library, wherein the surface may be charged polystyrene surface such as MaxiSorp™ or PolySorp™, or may be coated with avidin or streptavidin or neutravidin, preferably the Maxisorp™ surface is coated with streptavidin at a concentration ranging between 20 and 100μg/ml, most preferably 100μg/ml.

The capture antibody is selected from the group comprising goat anti-human IgG (goat anti-Human IgG (H+L); F(ab')2 fragment) or Capture Select Biotin Anti-IgG-CHl Conjugate, preferably the biotinylated anti-CHl antibody at a concentration of 1000-lOOng/ml, most preferably 250ng/ml).

The light chain specific antibody is selected from the group comprising goat anti-human lambda LC specific peroxidase conjugate, goat anti- human kappa LC specific peroxidase conjugate, goat anti-human F(ab')2-HRP, mouse anti-human kappa light chain peroxidase conjugate, mouse anti-human kappa light chain monoclonal and rabbit anti -human kappa chain monoclonal, preferably at a dilution ranging between 1-20000, most preferably 1 : 10000 for anti-lambda and 1 :2000 for anti-kappa.

Development of the quantitative chain switch ELISA as set out herein permits direct linking in series to an industry-standard assessment method - kinetic screening or affinity ranking for high throughput antibody discovery (Schraml M and Biehl M, 2012. Kinetic screening in the antibody development process. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols; Drake AW and Papalia GA, 2012. Biophysical considerations for development of antibody-based therapeutics. In: Development of Antibody -based Therapeutics). The practical reasons for affinity ranking are set out in literature (Tabrizi MA. 2012. Considerations in establishing affinity design goals for development of antibody-based therapeutics. In: Development of Antibody -based Therapeutics). It is one of the best predictor of antibody dose required for the maximum therapeutic benefit (potency) at a manageable cost-of-goods. Higher affinity antibodies will generally result in higher potency primarily determined by antigen concentration and turnover in vivo. Ability to rank antibodies in terms of their kinetic parameters early in the discovery phase enables the process of the present invention to match the capabilities of the phage display system to that of the hybridoma system in this aspect. The other benefits of kinetic ranking are high throughputs and possibility of assessment of thermodynamic stability (Schraml M and von Proff L, 2012. Temperature -dependent antibody kinetics as a tool in antibody lead selection. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), which in turn is often a predictor of protein aggregation (Thiagarajan G et al., 2016) - a very important concern for manufacturing antibodies.

Affinity ranking can be done using competition ELISA or Surface Plasmon Resonance. Competition ELISA requires a labeled antigen that can be used as a detection handle. Due to this additional and often difficult-to-achieve requirement, such assays are used more for validation rather than a primary method of data generation today. SPR, in contrast, is a label-free method, and along with continuous improvements in software and hardware, has emerged to become the method-of-choice today for generating antibody affinity data. An SPR surface that is desirable for kinetic analysis is where the Fabs are oriented with their Fv surfaces (paratopes) facing the flowing water phase (with or without antigen) in the SPR flow cell. This requires a surface that would allow quantitative capture of crude Fab candidates in an oriented manner and allow kinetic ranking. Such oriented surfaces for full length IgGs are well described in literature (Canziani GA et al., 2004; Schraml M and Biehl M, 2012. Kinetic screening in the antibody development process. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), but for Fabs (that lack the Fc domain by definition) such literature is indeed sparse (Leonard P et al., 2007). Pioneers therefore had no option but to immobilize the antigen itself on the surface of the SPR chip to determine kinetic parameters (de Haard HJ et al., 1999; Steukers M et al., 2006). Such an approach mimics a direct ELISA with the distinct possibility of masking interesting epitopes. New reagents have now become commercially available that allow creation of such oriented surfaces even for Fabs, although detailed protocols are not available in the prior art. The present invention investigates several such surfaces, and examples presented herein.

The present invention discloses a method of obtaining antibodies from the antibody phage display library, wherein the kinetic ranking comprises the steps of:

i) obtaining soluble Fabs from qELISA positive clones in 50ml individual cultures;

ii) dialyzing the obtained Fabs from step(i) against lxPBS;

iii) use of running buffer of physiological strength and pH for kinetic analysis - the buffer could be phosphate or HEPES, more preferably phosphate, containing NaCl or KC1 concentration of 0.1 to 1.0M, preferably 0.25 to 0.75M, more preferably 0.4 to 0.6M, and Tween-20 concentration of 0.005 to 0.05%;

iv) selecting the SPR (surface plasmon resonance) chip immobilization surface - such surface could be charged dextran, charged alginate, nickel nitrilotetraacetic acid coated on charged dextran or alginate surface, or streptavidin or neutravidin coated on charged dextran or alginate surface;

v) selecting the immobilization chemistry for the SPR surface at step(iv) - such chemistry could be amine coupling using EDAC(l-Ethyl-3-(3-dimethylaminopropyl)carbodiimide) and sulfo-NHS (N-hydroxysuccinimide), Ni2+ charging using lOmM nickel sulfate, or streptavidin-biotin recognition chemistry;

vi) immobilizing the anti-Fab capture antibody on the chip surface from step(v) - capture antibodies may include anti-Fab IgG, anti-tag antibodies such as anti-His, anti-HA or biotinylated anti-CHl or biotinylated bivalent anti-CHl/anti-C or biotinylated anti-

CHl anti-C or a 50:50 mixture of both biotinylated bivalent anti-Cnl/anti-C and biotinylated anti-CHl/anti-C ;

vii) capturing the crude periplasmic Fabs obtained from step(ii) on the capture antibody- coated surface of the chip from step(vi);

viii) signal stabilization by 1-3 rounds of running buffer injection over the chip surface with intermediate pause of 2-15 min;

ix) testing the analyte response on captured Fabs of step(vii) at an optimal concentration of analyte to distinguish between target antigen binders and non- binders;

x) removing the Fab-analyte complex using regenerating reagent for the surface to be reused for the next round of screening - regenerating agent could include 2M MgCb, 0.85% H3PO4, 50mM NaOH or lOmM glycine, pH 2.0.

The combination of methods for Fab discovery as described in the preceding paragraphs achieves two aims. The first is that extension of the concept of manufacturabilty to antibody discovery from phage libraries is enabled as demonstrated herein from the initial step - Fabs discovered from such a system can be confidently assessed as quantifiable heterodimeric molecules expressed in the periplasm by design, with a range of yields that is not by design but on which a gating system can be applied. The second is that establishment of a consistent method for crude Fab capture on SPR chips allows us to interrogate binding of a target antigen to the captured Fab on the chip itself - a function conventionally carried out by direct/indirect ELISA where the antigen is immobilized on polystyrene surface. The latter approach can mask interesting and relevant epitopes as discussed earlier. In other words, the novel crude Fab capture system on SPR chips of the present invention is doubly advantageous in that it not only bypasses antigen-specific ELISA that is error prone, but also allows obtaining kinetic parameters directly for Fabs that do recognize the antigen, thus shortening the time required for discovery. Examples included herein demonstrate that Fabs assessed in this manner are also genotype true - that is, the combination of chain switch ELISA on periplasmic extracts and on-chip kinetic ranking (phenotyping) picks up only those clones that have intact ORFs in their tandem in-frame light chain-heavy chain Fab structure as determined by post hoc sequence analysis (genotyping).

Creating an ultra-large naive phage display library as a source of antibodies

The staged assessment system described in the previous sections suggests that Fabs are the preferred format for antibody discovery for stability reasons, and can be assessed successfully as proteins in a high throughput manner. This design decision, added to the reality that a large number of potential binders will be lost due to the constraints of phage amplification, poor expression in E. coli, as well as the stringencies imposed by the staged assessment process itself, therefore necessitates that an ultra-large library of Fabs be created to compensate for the losses, and be made available for antigen recognition by a display method.

The primary rationale for creating large libraries is capturing as many diverse combinations of V-domains as possible to retrieve antibodies in the therapeutic range (KD in sub-nM to pM range). Library size therefore represents this diversity - the underlying assumption being that each recombinant clone represents a different VL-VH combination (Hoogenboom HR et al., 1991 ; Waterhouse P et al., 1993). This assumption is difficult to verify without sequencing a very large number of clones. Digesting plasmid DNA with frequent cutting enzymes (BstNI, BstOl, Alu\ etc.) and running the digests on agarose gels to study the resultant fingerprints is conventional for assessing diversity but most certainly a weak substitute of sequence information. Sanger sequencing, because of its power to read long fragment lengths, is highly suitable to read the VL-VH pairings embedded in scFv or Fab formats, but the throughput of sample preparation, sequencing and analysis does not allow it to be executed for more than a few hundred clones. Next-Generation Sequencing (NGS) has been applied to resolve this problem (Geyer CR et al., 2012. Recombinant antibodies and in vitro selection technologies. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols; Glanville J et al., 2015), but limitations in read lengths do not allow it to read VL-VH pairs of a scFv library unambiguously for more than 106 clones. The procedure is therefore more successful for enriching binders after the diversity of the library has been reduced after at least one round of panning (Ravn U et al., 2013; Glanville J et al., 2015). The limitation in read lengths implies that Fab libraries, where other than VL-VH pairings, one also needs to confirm tandem in-frame nature of the light chain and heavy chain cassettes, are not currently amenable to verification by NGS. Recent developments in de novo assembly of short reads to construct maps of complex libraries (Cho N et al., 2015) may solve this problem. Regardless of this verification problem, the core principle of maximal paratope coverage is the key to the phage display libraries' ability to find antibodies against almost any antigen including self-antigens that are nearly impossible to obtain from immunizations. Furthermore, there exists an almost linear correlation between library size and affinities of binders - larger the library, greater the chances of finding therapeutic range antibodies (Hust M et al., 2009. Antibody phage display. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic; Lowe D and Vaughan TJ, 2009. Human antibody repertoire libraries; In: Therapeutic Monoclonal Antibodies: From Bench to Clinic). Making a large library is therefore advantageous from this respect as well.

While building an ultra large naive library, the present invention and the process utilized herein balance the diversity in the library, while ensuring the size, rapid throughput and the economic advantage.

The present invention discloses a naive antibody phage display library (APDL) having a size ranging between 8.86 x 1010 to 9.13 x 1011 (3.06 x 1011) cfu, that includes 5.38 x 1010 to 2.55 x 1011 (1.26 x 1011) cfu kappa library and 7.33 x 1010 to 3.59 x 1011 (1.79 x 1011) cfu lambda library.

Several attempts have been made earlier to overcome the disadvantage of the prior art by building alternative display systems. Alternative display systems may be categorized as in vivo systems that include yeast display (Boder ET and Wittrup KD, 1997; Weaver-Feldhaus JM et al., 2004), bacterial display (van Blarcom TJ and Harvey BR, 2009. Bacterial display of antibodies. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic) and mammalian display (Tomimatsu K et al., 2013; Horlick RA et al., 2013), and in vitro systems that include ribosome display (Hanes J and Pluckthun A, 1997), DNA display (Sumida T et al, 2009), mRNA display (Roberts RW and Szostak JW, 1997) and bead display (Diamante L et al., 2013). In vivo systems usually rely on anchoring the expressed antibody format (scFv, Fab) on a cell surface protein to maintain the phenotype-genotype linkage. Similarly, in vitro systems anchor the transcription unit through a linker to the anchor (ribosome, puromycin or polystyrene beads) to maintain the phenotype-genotype linkage. However, similar to the phage display system, in vivo systems are limited by transformation efficiencies that do not allow library sizes to usually exceed 109

variants, thus limiting their utility for capturing a naive immune repertoire. In contrast, in vitro systems can display as many as 1012-1014 variants, but are limited both by the difficulties in manipulation as well as inability to produce heterodimeric proteins by design (the only exception being the difficult-to-optimize DNA display system). These systems are therefore most useful in scenarios where molecular evolution of a candidate protein, such as affinity maturation of a parent Fab template, is desired. Phage display libraries, for their simplicity, robustness and track record in discovering multiple therapeutic or diagnostic antibodies, therefore remain as the method of choice for first-pass retrieval of a VL-VH combination that has ability to recognize a given antigen (Hust M et al, 2009. Antibody phage display. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic; Lowe D and Vaughan TJ, 2009. Human antibody repertoire libraries; In: Therapeutic Monoclonal Antibodies: From Bench to Clinic).

Phage display libraries can be created either on phage or phagemid vector backbones with consequent advantages and disadvantages (Scott JK and CF Barbas III. 2001. Phage-display vectors. In: Phage Display: A Laboratory Manual). Phagemids are preferred both for monovalent display of antibody fragments that allow selection by true affinity as well for the fact that transformation efficiencies in E. coli are far superior compared to the phage vectors. Nonetheless, even this superior efficiency is limited by the maximum transformation efficiencies usually achievable for restriction digested plasmid vectors ligated to insert fragments, which is about 109cfu^g. In contrast, the maximum efficiencies achievable for supercoiled plasmid preparations are at least a log higher (1010cfu^g; Hoogenboom HR et al., 1991 ; Sambrook J and Russell DW. 2001b. The Hanahan method for preparation and transformation of competent E. coli: High efficiency transformation. In: Molecular Cloning: A Laboratory Manual; Vol. 1). However, as set out herein above, these limits of transformation efficiency limit the number of random VL-VH permutations that can be captured in E. coli to create libraries, and have therefore forced the antibody community to create such large libraries in two steps or two independent cassettes transformed simultaneously (Waterhouse P et al., 1993; Griffiths AD et al., 1994; de Haard HJ et al, 1999; Ostermeier M and Benkovic SJ, 2000; Hoet RM et al, 2005) to ensure capture of both VL and VH diversity. It is therefore advantageous for building large Fab-phage display libraries if these limits can be pushed higher. Examples presented herein demonstrate that exceeding these limits is indeed possible, and can be applied to build large Fab-phage display

libraries in one step wherein the only limitation is the maximum number of E. coli that can reside in one liter of bacterial culture (10 12 /L; Hoogenboom HR et al., 1991).

The present invention discloses process for producing the APDL, wherein transformation is carried out at a DNA to cell volume ratio of 25 to 400, preferably 100 to 350, more preferably, 200 to 300 ng per 50μ1 of ultracompetent cells.

The present invention further discloses a process for producing the APDL, wherein transformation is carried out at a voltage in the range of 1500 to 3500 volts, preferably 2500 to 3200 volts, capacitance in the range of 10 to 30μΡ, preferably 20 to 28μΡ and resistance of 100 to 400 ohms, preferably 250 to 350 ohms in a cuvette of 0.1, 0.2, 0.4 cm inter-electrode distance, preferably 0.2 cm. The host is an amber suppressor t-RNA encoding host selected from the group comprising TGI, XL-1 Blue and ER2537, preferably TGI of ultrahigh competence (4 x 1010 cfu^g).

The present invention discloses the use of higher rated transformation efficient cell (procured from Lucigen). Hence the present invention reports higher transformation efficiencies of restriction digested phagemid vectors ligated to Fab insert fragments. This advantageous utilization enables the present invention to arrive at a large naive library as set out herein utilizing lesser number of transformations.

The present invention also discloses that the transformation efficiency of vector and insert ligations depends upon DNA quality and adequacy of ligation. While the problem of DNA quality has been focused upon and improved earlier (Martineau P, 2010. Synthetic antibody libraries. In: Antibody Engineering; Rader C, 2012b. Generation of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols), the problem of adequacy of ligation has not. To elaborate, most antibody cloning formats use a cassette cloning approach in which the VL, VH, CL or Cjjl fragments are treated modularly for PCR-based amplification, fusion, restriction digestion with enzymes that cut rarely within V-domains (Persic L et al., 1997), and ligation to a vector. The present invention discloses that highest transformation efficiencies may be obtained when the sticky ends generated after

restriction digestion of a population of PCR amplified/fused fragments have the minimum number of mismatch with the compatible sticky ends in the vector. Other improvements such as the importance of salt removal from ligation mixes for achieving higher transformation efficiencies by electroporation have been described (Chowdhury PS, 2002. Targeting random mutations to hotspots in antibody variable domains for affinity improvement. In: Methods in Molecular Biology, vol. 178: Antibody Phage Display: Methods and Protocols), and the present invention demonstrates a similar beneficial effect using ultrafiltration through microconcentrators (de Haard HJW, 2002. Construction of large naive Fab libraries. In: Methods in Molecular Biology, Vol. 178: Antibody Phage Display: Methods and Protocols; Green MR and Sambrook J, 2012a. Concentrating and desalting nucleic acids with microconcentrators. In: Molecular Cloning: A Laboratory Manual; Vol. 1) in the examples. The beneficial effect of high fidelity amplification and seamless fusion of these fragments for increasing potential translatability of the captured Fab that advantageously improves on the utility is also disclosed in the present application.

The present invention discloses a process for producing the APDL, wherein displaying the captured immune repertoire in a vector comprises the steps of:

i) ligating the Fabs in a phagemid vector;

ii) transforming the ligated mixture into a suitable host.

The present invention discloses a process for producing the APDL, wherein the ligation of Fab repertoires obtained using the SEQ ID 32 and 34 is conducted by

i) blunting the Fabs using the 3 '-5' exonuclease property of T4 DNA polymerase at 11- 37°C, preferably 11°C, and phosphorylation of the 5' ends of blunted Fabs using T4 polynucleotide kinase at 37°C for l-1.5h, preferably 1.5h;

ii) by self-ligation of Fabs obtained at step(i) at a temperature range of 4-16°C, preferably 16°C for 16h followed by 25°C for lh, with a concentration range of 50- 400ng^l of total DNA, preferably 200ng^l, in the presence of additives selected from the group comprising polyethylene glycol of molecular weight 6000-32000 Daltons, preferably 8000 Daltons, in a final percentage ranging between 1.5-9% w/v, preferably 4-7% w/v, more preferably 6% w/v;

iii) restriction digestion of the self-ligated Fab population from (ii) with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification;

iv) by sticky end ligation of linear Fabs to pCOMB3XSS obtained at step(iii) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

The present invention further discloses a process for producing the APDL, wherein the ligation of Fab repertoires obtained using SEQ ID 34 and 35-37 to the pCOMB3XSS vector is conducted by i) restriction digestion of the linear Fab population with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification; ii) by sticky end ligation of linear Fabs obtained at step(i) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

The present invention discloses a process for producing the APDL, wherein the ligation of Fab repertoires obtained using SEQ ID 34 and 55 to the pSSYl vector is conducted by

i) restriction digestion of the linear Fab population with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification; ii) by sticky end ligation of linear Fabs obtained at step(i) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

The present invention also discloses vide examples a maximum amount of DNA that can be transformed into a fixed volume of high transformation efficiency TGI cells without any decrease in efficiency. Although it seems obvious that building a library is a compromise between size and efficiency of transformation, a library having optimum size, diversity and balance of cost has not been found in prior art as of now. The present application discloses such a number by utilizing the principle of titration. Examples included herein demonstrate that this determination allowed the process of the present invention as set out herein in reducing the number of transformations drastically. Such reduction resulted in a very large library size, with a comparatively small amount of ligated DNA as well with much shorter turnaround times. The gain in library-making speed can be advantageous when immune libraries may be required to be

created to discover pathogen-specific antibodies rapidly in biodefense-like scenarios. Similarly, significant cost savings in reagent use may be achieved by using smaller amounts of PCR-amplified and fused DNA. The present invention discloses a process to reduce the number of transformations thereby increasing the size, with benefits of time and cost and is one of the several advantages of the library and the process of arriving at the library as set out herein.

In the present invention, the APDL is obtained from 15-160 μg of ligated DNA, preferably 20-100μg, more preferably 40 to 50μg of ligated DNA in a single step of transformation. The present invention discloses a process for producing the kappa subtype APDL, wherein the APDL is obtained from 10 to 70μg of ligated DNA, preferably 20 to 50μg, more preferably 25 to 30μg of ligated DNA as obtained in a single step of transformation. The present invention discloses a process for producing the lambda subtype APDL, wherein the APDL is obtained from 5 to 60μg of ligated DNA, preferably 8 to 50μg, more preferably 10 to 20μg of ligated DNA as obtained in steps 15-20 in a single step of transformation.

The present invention discloses a process for producing the APDL, wherein the kappa APDL is obtained with an efficiency of 1.92 x 109 to 1.98 x 1010 cfu^g and the lambda APDL is obtained with an efficiency of 1.92 x 109 to 9.1 x 109 cfu^g.

pSSYl - A new phagemid display vector: The present invention discloses a new phagemid vector for making ultra-large libraries (pSSYl; SEQ ID 38 and Figure 28). As illustrated in the Examples, creation of this vector was necessitated by the inherent defects in the parental vector pCOMB3XSS (Figure 12; Barbas CF III et al, 1991; Andris-Widhopf J et al, 2001. Generation of antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences. In: Phage Display: A Laboratory Manual) that did not allow creation of ultra-large libraries due to inefficient ligations. The present invention discloses the re-design of the vector based on the design intent of the parental vector, but this effort also introduced major changes in the overall sequence and provided significant advantages that are illustrated in the Examples herein. Further features of this plasmid, and similarities as well as dissimilarities with pCOMB3XSS are illustrated in Example 24.

Producing recombinant antibodies by PCR amplification and fusion of V- and C-genes: This invention designs and utilizes an optimized set of PCR conditions for high fidelity amplification of human V-genes and their fusion with human C-genes for creating a combinatorial human immunoglobulin repertoire for subsequent cloning and display with the phagemid display vector pSSYl. For this purpose, it utilizes a set of 35 primers set out in the prior art (Andris-Widhopf J et al., 2001. Generation of antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences. In: Phage Display: A Laboratory Manual) that were subsequently modified to allow efficient ligation to the pSSYl vector. The present invention discloses a novel process for amplification of the said 35 primers and their subsequent ligation in the pCOMB3XSS vector. The present invention also discloses modified primers and their amplification and subsequent ligation into the novel vector pSSYl.

Recombinant antibody generation depends on the modularity inherent in how an immunoglobulin is assembled in vivo. Briefly, both the light and heavy chains of an immunoglobulin molecule are composed of variable (V), hinge (H) and constant (C) regions in a N-terminal - C-terminal direction that originate from their respective gene clusters (loci) within the chromosomes in a mirrored 5' - 3' exon-intron fashion. The chromosomal location of these gene clusters do differ - while the human heavy chain locus is on chromosome 14 (IGH locus; 14q32.33), the human light chain loci are on chromosomes 2 (kappa or IGK locus; 2pl l.2) and 22 (lambda or IGL locus; 22ql l.2). Each locus can contain multiple genes - the overall number of all human Ig genes is estimated to range between 371 and 422. Diversity in the repertoire of heavy chains is primarily generated by recombination of the germline V, Diversity (D) and Junctional (J) exons, while diversity in the repertoire of light chains is primarily generated by recombination of the germline V and J exons. Three additional mechanisms contribute to the immunoglobulin chain diversity. The first is called the N diversity (N, for Nucleotides) which results from the deletion and/or addition of nucleotides at random by terminal deoxynucleotidyl transferase (TdT) at the V-D-J junction, resulting in a region that is not encoded in the germline DNA. The second mechanism is called somatic hypermutation (SHM), which specifically affects the V-D-J rearranged genes, and is believed to be controlled either by activation-induced cytidine deaminase-uracil DNA glycosylase-DNA polymerase eta enzyme complex, or an error prone RNA-directed DNA polymerase. Regardless of mechanism, the end result is change of

nucleotides at "hot-spots" that therefore differ from the germline code, and generally result in improvements of affinity to a target antigen. The third mechanism is called class switch, which joins the rearranged V-D-J clusters of heavy chains to various hinge and constant exons. Class switch does not impact on the antigen recognition ability of the immunoglobulin, but allows differential interaction of immune effector cells to the constant regions that provides final functionality to the molecule.

The immunoglobulin genes are expressed only in the cells from B-lineage, first as membrane-bound receptors followed by secretion as immunoglobulin proteins. The type of recombination, SHM or class switch that a B-cell light and heavy chain combination might contain depends on the stage in its life cycle. The source of immune tissues for recombinant antibody generation therefore depends on the project goal - building a naive library requires harvesting of B -cells that have undergone minimum re-arrangement of the germline, preferably without N-addition or SHM. In contrast, building an immune library requires harvesting of B -cells that have preferentially undergone SHM to retrieve the affinity matured V-D-J or V-J combinations. Different compartments of the human body encase B -cells at different stages of their life cycle, and therefore care must be exercised for appropriate tissue harvest (Dobson CL et al., 2012. Naive antibody libraries from natural repertoires. In: Phage Display in Biotechnology and Drug Discovery). As exemplified herein, the ultra-large naive Fab library described in this invention use B -cells harvested from human peripheral blood mononuclear cell population, tonsil and bone marrow.

The present invention discloses a process for producing the APDL comprising the steps of:

i) immune repertoire capture to obtain a Fab;

ii) displaying the captured immune repertoire of as above in a suitable vector.

The present invention further discloses a process for producing the APDL, wherein the immune repertoire capture comprises the steps of:

i) RNA isolation and cDNA synthesis;

ii) amplification of VL (lambda and kappa) and VH domains using primers comprising the SEQ ID 1-23 and 42-54;

iii) amplification of C domains using SEQ ID 24-26 and using primers comprising the SEQ ID 27-31;

iv) overlap PCR of light chains by fusion of V and C domains and V and C domains obtained from step(ii) and (iii), respectively, using primers comprising the SEQ ID 30, 32, 35-37 and 55;

v) overlap PCR of heavy chains obtained from fusion of VH and CHI obtained from step(ii) and (iii) using primers comprising the SEQ ID 28 & 33;

vi) overlap PCR of light chains and heavy chains obtained from steps(iv) and (v) respectively to obtain Fabs using primers comprising the SEQ ID 32, 34, 35-37 and 55.

vii) purifying the amplicons at each step.

In contrast, synthetic or semi-synthetic libraries either depend upon a fixed natural Ig template or fixed synthetic framework regions of the variable domains on which the core antigen-recognizing amino acids of the variable domains (the complementarity determining region or CDR) are subcloned after in vitro synthesis.

Recombinant antibody generation from natural sources begins after harvesting the transcriptome (mRNA) of the target B-cell population and then reverse transcribing it to create complementary DNA (cDNA). Domain specific primers are then used to amplify the subtype-specific (kappa or lambda) V-gene repertoire from the cDNA templates by PCR. The amplified V-domains of one subtype are then allowed to pair randomly with the opposite subtype in vitro to allow creation of a multitude of paratopes. Further in vitro manipulations that are recombinase or DNA polymerase based allow joining of H- or C-regions to create Ig fragments or full length Igs that can be displayed. As should be obvious, V-domain specific amplification and random pairing destroys the inherent paratope information. To some extent, this is intentional for one of the design goals of recombinant antibody generation is to generate new paratopes that can recognize self-antigens or have very high affinities for a target antigen. Such paratopes would normally be deselected in vivo and therefore, would be not found in a B-cell immune repertoire (Foote J and Eisen HN, 1995; Hai S-H et al., 2009. Immunogenicity screening using in silico methods: Correlation between T-Cell epitope content and clinical immunogenicity of monoclonal antibodies. In: Therapeutic Monoclonal Antibodies: From Bench to Clinic). However, for some

applications, it is important to capture the paratope information. A system of linked PCR has been described (Meijer PJ et al., 2006) for such purposes.

Regardless whether the design intent is to capture the paratopes as they are or recombine the V-domains randomly, PCR primers are required to specifically amplify the V-domains. Both due to the multiplicity of V-gene families that further encode allelic variations, as well as the fact that a particular class of V-D-J or V-J combination needs to be captured depending on whether one wishes to create a naive or immune library, a variety of thoughtful primer sets have been designed and successfully used to create antibody phage display libraries (Marks JD et al., 1991; de Boer M et al., 1994; Sblattero D and Bradbury A, 1998). All such primers incorporate degenerate nucleotides at specific positions to accommodate the variations in amino acids at those positions as revealed by sequences available in the antibody databases ( t^J/wvvw.bknnf.org.uk abs/sirnkab.html; http;//www.vbase2.org/; http://www.imgt.org/). The present invention discloses a process utilizing the primer set described by a Scripps group (Andris-Widhopf J et al., 2001. Generation of antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences. In: Phage Display: A Laboratory Manual). The design intent and method used to design these primers is available (Burton DR, 2001. Overview: Amplification of antibody genes. In: Phage Display: A Laboratory Manual).

The present invention also discloses a set of novel primers as the earlier primers set out in the prior art when progressed to the recipient phagemid display vector pCOMB3XSS did not allow for efficient cassette cloning of the V-domains amplified by these primers. This re-design involved the restriction ends of the primers. In particular, the re-design involved changing the core pentanucleotide sequence of the Sfil sites that were built in all the V and V forward primers, as well as the final overlap forward primer. Further re-design involved reducing the homology between the final overlap forward and reverse primers. As exemplified herein, these design changes were crucial for high efficiency ligation of the amplified Fabs to the new phagemid display vector pSSYl. As discussed earlier and as exemplified herein, highly efficient ligation combined with parameter-controlled high efficiency transformation as disclosed herein enables for the first time a single step transformation to create a >10n cfu Fab-phage display library.

PCR primers alone are not sufficient in themselves to ensure high fidelity amplification of templates, particularly templates encoding multiple variations such as those encoded by the V-domains. Prior art primarily discloses use of Taq polymerase, partly because that was the only thermostable DNA polymerase available for a long time. Taq polymerase however has a fairly high error rate (Tindall KR and Kunkel TA, 1988; Gelfand DH and White TJ, 1990) that results in introduction of stop codons and frameshifts in V -domains or V-C fusions (reviewed in Lowe D and Vaughan TJ, 2009. Human antibody repertoire libraries; In: Therapeutic Monoclonal Antibodies: From Bench to Clinic; Azzazy HM and Highsmith WE, 2002). In recognition of this problem, more recent practitioners of this art have used thermostable polymerase blends (Andris-Widhopf J et al., 2001. Generation of antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences. In: Phage Display: A Laboratory Manual; Rader C, 2012b. Generation of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols) based on the LA-PCR principle (Barnes WM, 1994), although this still does not ensure amplification of all V-domains (L0set GA et al., 2005; Rader C, 2012b. Generation of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols). Regardless whether Taq or Taq blends are used, almost all practitioners have used conservative annealing temperatures (~56°C) in order to allow the degenerate primers to amplify maximally (Marks JD et al., 1991 ; de Boer M et al., 1994; Sblattero D and Bradbury A, 1998; Andris-Widhopf J et al, 2001. Generation of antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences. In: Phage Display: A Laboratory Manual; Rader C, 2012b. Generation of human Fab libraries by phage display. In: Methods in Molecular Biology Vol. 901: Antibody Methods and Protocols). As exemplified herein, our detailed analysis of the templates suggests that most of human VH and V family genes are GC-rich for most of their amplicon length, particularly in the 1st framework (FR1) regions where all forward V-domain primers are designed to anneal, including the Scripps primers described in prior art. The V family genes, in contrast, have average GC content (-50%) for most of the length of their amplicons, including the FR1 region, although GC-rich stretches also exist within these gene families. The present invention discloses through extensive experimentation that neither the DNA polymerase combinations nor the low annealing temperature suggested in the prior art are appropriate for efficient amplification from such templates regardless of the accuracy of primer design. As exemplified herein, specific buffer- polymerase combinations were designed that follow the current standard of amplifying through such GC-rich regions (Green MR and Sambrook J. 2012b. PCR amplification of GC-Rich Templates. In: Molecular Cloning: A Laboratory Manual; Vol. 1). Furthermore, analysis of a limited number of clones from the resultant ultra-large library suggest that no loss of diversity has resulted from this inventive stratagem and the library was able to produce different binder sequences against each of the three tested antigens as determined by detailed analysis. In contrast to most published data, all of these binders were also kinetically stable with off -rates in the 10"4 to 10~5.s_1 range. This suggests that the inventive amplification system has captured the desired diversity of a naive B-cell immune repertoire with high fidelity and efficiency. Examples illustrating these principles are presented herein.

We claim

1. A naive antibody phage display library (APDL) having a size ranging between 8.86 x 1010 to 9.13 x 1011 (3.06 x 1011) cfu, that includes 5.38 x 1010 to 2.55 x 1011 (1.26 x 1011) cfu kappa library and 7.33 x 1010 to 3.59 x 1011 (1.79 x 1011) cfu lambda library.

2. A process for producing the APDL as claimed in claim 1 comprising the steps of:

i) immune repertoire capture to obtain a Fab;

ii) displaying the captured immune repertoire of step 2(i) in a suitable vector.

3. A process for producing the APDL as claimed in claim 2(i), wherein the immune repertoire capture comprises the steps of:

i) RNA isolation and cDNA synthesis;

ii) amplification of VL (lambda and kappa) and VH domains using primers comprising the SEQ ID 1-23 and 42-54;

iii) amplification of C domains using SEQ ID 24-26 and using primers comprising the SEQ ID 27-31;

iv) overlap PCR of light chains by fusion of V and C domains and V and C domains obtained from steps 3(ii) and 3(iii), respectively, using primers comprising the SEQ ID 30, 32, 35-37 and 55;

v) overlap PCR of heavy chains obtained from fusion of VH and CRI obtained from steps 3(H) and 3(iii) using primers comprising the SEQ ID 28 & 33;

vi) overlap PCR of light chains and heavy chains obtained from steps 3(iv) and 3(v) respectively to obtain Fabs using primers comprising the SEQ ID 32, 34, 35-37 and 55;

vii) purifying the amplicons at each step.

4. A process for producing the APDL as claimed in claim 3(ii), wherein the amplification of variable lambda domains are conducted in a two-step PCR using primers comprising the SEQ ID 14-23 and 46-54 and comprising the steps of:

i) obtaining a mixture of cDNA template, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 4(i) to a temperature range of 90 to 96°C to denature the templates;

iii) simultaneous annealing and extension of the denatured templates from 4(ii) at a temperature of 65 to 72°C to obtain variable lambda domains such that it results in a diverse V repertoire capture.

5. A process for producing the APDL as claimed in claim 3(ii), wherein the amplification of variable kappa domains are conducted in a three- step PCR using primers comprising the SEQ ID 9-13 and 42-45 and comprising the steps of:

i) obtaining a mixture of cDNA template, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 5(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) annealing the primers to the denatured templates from step 5(ii) at a temperature range of 55 to 70°C;

iv) extension of the primers on annealed templates from step 5(iii) at temperature of 65 to 72°C to obtain variable kappa domains such that it results in a diverse V repertoire capture.

6. A process for producing the APDL as claimed in claim 3(ii), wherein the amplification of variable heavy domains are conducted in a three- step PCR using primers comprising the SEQ ID 1-8 and comprising the steps of:

i) obtaining a mixture of cDNA template, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 6(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) annealing the primers to the denatured templates from step 6(ii) at a temperature range of 55 to 70°C;

iv) extension of the primers on annealed templates from step 6(iii) at a temperature of 65 to 72°C to obtain variable heavy domains such that it results in a diverse VH repertoire capture.

7. A process for producing the APDL as claimed in claim 3(iii), wherein the amplification of CHI domains is conducted as a three-step PCR using primer comprising the SEQ ID 27-28 and templates comprising the SEQ ID 24 and 39, and comprising the steps of:

i) obtaining a mixture of synthetic Cnl-domain templates, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 7(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) annealing the primers to the denatured templates from step 7(ii) at a temperature range of 55 to 70°C;

iv) extension of the primers on annealed templates from step 7(iii) at a temperature of 65 to 72°C to obtain the constant heavy domain.

8. A process for producing the APDL as claimed in claim 3 (iii), wherein the amplification of C and C domains are conducted in a two-step PCR using primers comprising the SEQ ID 29-31 and templates comprising the SEQ ID 25-26 and 40-41, and comprising the steps of:

i) obtaining a mixture of synthetic C and C domains, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 8(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) simultaneous annealing and extension of the denatured templates from step 8(ii) at a temperature of 65 to 72°C to obtain the constant kappa and lambda domains.

9. A process for producing the APDL as claimed in claim 3(iv), wherein the fusion of V and C domains and V and C domains are conducted in a two-step PCR using primers comprising the SEQ ID 30, 32, 35-37 and 55 and comprising the steps of:

i) obtaining a mixture of light chain variable and constant gene templates from claims 4-5 and 8, respectively, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 9(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) simultaneous annealing and extension of the denatured templates from step 9(ii) at a temperature of 65 to 72°C to obtain lambda and kappa light chain repertoires.

10. A process for producing the APDL as claimed in claim 3(v), wherein the fusion of VH and CHI domains is conducted in a three-step PCR using primers comprising the SEQ ID 28 and 33, and comprising the steps of:

i) obtaining a mixture of heavy chain variable and constant gene templates from claims 6 and 7, respectively, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 10(i) to a temperature range of 90 to 96°C to denature the templates;

iii) annealing the primers to the denatured templates from step 10(ii) at a temperature of 55 to 70°C;

iv) extension of the primers on annealed templates from step 10(iii) at a temperature of 68 to 72°C to obtain heavy chain repertoires.

11. A process for producing the APDL as claimed in claim 3(vi), wherein the fusion PCR of light and heavy chains are conducted in a two-step PCR using primers comprising the SEQ ID 32, 34, 35-37 and 55, and comprising the steps of:

i) obtaining a mixture of light and heavy chain repertoires from claims 9-10, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step l l(i) to a temperature range of 90 to 96 °C to denature the templates;

iii) simultaneous annealing and extension of the denatured templates from step 11 (ii) at a temperature of 65 to 72°C to obtain lambda and kappa Fab repertoires.

12. A process for producing the APDL as claimed in claim 3(vi), wherein the fusion PCR of light and heavy chains are conducted in a three-step PCR using primers comprising the SEQ ID 32, 34, 35-37 and 55, and comprising the steps of:

i) obtaining a mixture of light and heavy chain repertoires as obtained in steps 9-10, polymerase enzyme, primers, buffer and dNTP mix in an aqueous solution;

ii) subjecting the mixture of step 12(i) to a temperature range of 90 to 96°C to denature the templates;

iii) annealing the primers to the denatured templates from step 12(ii) at a temperature of 55 to 70°C;

iv) extension of the primers on annealed templates from step 12(iii) at a temperature of 65 to 72°C to obtain lambda and kappa Fab repertoires.

13. A process for producing the APDL as claimed in claim 5-12, wherein the buffer is selected from the group comprising AmpliTaq® Gold buffer, AmpliTaq® PCR buffer, AmpliTaq® PCR buffer II, Expand™ buffer 2, Expand™ buffer 3, Expand™ buffer 4, Thermopol® buffer, Pfu Ultra II buffer, Exact™ polymerase buffer, PCR Extender buffer, Tuning buffer, Vent® Buffer, Advantage®2 buffer, Advantage®2 SA buffer and the thermostable DNA polymerase enzyme is selected from the group comprising AmpliTaq® Gold DNA polymerase, Expand™ LT Taq DNA polymerase blend, Phusion® High Fidelity DNA polymerase, PfuUltra™ II HS DNA polymerase, PCR Extender™ DNA polymerase blend, Exact™ DNA polymerase, Vent® DNA polymerase, Deep Vent® DNA polymerase, and Advantage®2 DNA polymerase Mix.

14. A process for producing the APDL as claimed in claim 2(ii), wherein displaying the captured immune repertoire of claims 5-13 in a vector comprises the steps of:

i) ligating the Fabs in a phagemid vector;

ii) transforming the ligated mixture into a suitable host.

15. A process for producing the APDL as claimed in claim 14(i), wherein the ligation of Fab repertoires obtained using the SEQ ID 32 and 34 is conducted by:

i) blunting the Fabs using the 3'-5'exonuclease property of T4 DNA polymerase at 11-37°C, preferably 11°C, and phosphorylation of the 5' -ends of blunted Fabs using T4 polynucleotide kinase at 37°C for 1-1.5h, preferably 1.5h;

ii) by self-ligation of Fabs obtained at step 15(i) at a temperature range of 4-16°C, preferably 16°C for 16h followed by 25°C for lh, with a concentration range of 50-400ng^l of total DNA, preferably 200ng^l, in the presence of additives selected from the group comprising polyethylene glycol of molecular weight 6000-32000 Daltons, preferably 8000 Daltons, in a final percentage ranging between 1.5-9% w/v, preferably 4-7% w/v, more preferably 6% w/v;

iii) restriction digestion of the self-ligated Fab population from 15(ii) with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification;

iv) by sticky end ligation of linear Fabs obtained at step 15(iii) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

16. A process for producing the APDL as claimed in claim 15, wherein the vector is pCOMB3XSS (Figure 12).

17. A process for producing the APDL as claimed in claim 14(i), wherein the ligation of Fab repertoires obtained using the SEQ ID 34 and 35-37 is conducted by:

i) restriction digestion of the linear Fab population with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification;

ii) by sticky end ligation of linear Fabs obtained at step 17(i) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

18. A process for producing the APDL as claimed in claim 17, wherein the vector is pCOMB3XSS (Figure 12).

19. A process for producing the APDL as claimed in claim 14(i), wherein the ligation of Fab repertoires obtained using the SEQ ID 34 and 55 is conducted by:

i) restriction digestion of the linear Fab population with 32U^g Sfil at 50°C for 16h to release linear Fabs with sticky ends followed by agarose gel purification;

ii) by sticky end ligation of linear Fabs obtained at step 19(i) at a temperature of 16°C for 16h followed by 37°C for lh and heat-inactivation at 70°C for 15min.

20. A process for producing the APDL as claimed in claim 19, wherein the vector is pSSYl (SEQ ID 38; Figure 28).

21. A process for producing the APDL as claimed in claim 14(H), wherein transformation is carried out at a DNA to cell volume ratio of 25 to 400, preferably 100 to 350, more preferably, 200 to 300ng per 50μ1 of ultracompetent cells.

22. A process for producing the APDL as claimed in claim 14(ii), wherein transformation is carried out at a voltage in the range of 1500 to 3500 volts, preferably 2500 to 3200 volts, capacitance in the range of 10 to 30μΡ, preferably 20 to 28μΡ and resistance of 100 to 400 ohms, preferably 250 to 350 ohms in a cuvette of 0.1, 0.2, 0.4 cm inter- electrode distance, preferably 0.2 cm.

23. A process for producing the APDL as claimed in claim 14(H), wherein the host is an amber suppressor t-RNA encoding host selected from the group comprising TGI, XL- 1 Blue and ER2537, preferably TGI of ultrahigh competence (4 x 1010cfu^g).

24. A process for producing the APDL as claimed in claim 1, wherein the APDL is obtained from 15-160μg of ligated DNA, preferably 20-100μg, more preferably 40 to 50μg of ligated DNA as obtained in claims 15 to 23 in a single step of transformation.

25. A process for producing the kappa subtype APDL as claimed in claim 1, wherein the APDL is obtained from 10 to 70μg of ligated DNA, preferably 20 to 50μg, more preferably 25 to 30μg of ligated DNA as obtained in claims 15 to 23 in a single step of transformation.

26. A process for producing the lambda subtype APDL as claimed in claim 1, wherein the APDL is obtained from 5 to 60μg of ligated DNA, preferably 8 to 50μg, more preferably 10 to 20μg of ligated DNA as obtained in claims 15 to 23 in a single step of transformation.

27. A process for producing the APDL as claimed in claim 1, wherein the kappa APDL is obtained with an efficiency of 1.92 x 109 to 1.98 x 1010 cfu^g and the lambda APDL is obtained with an efficiency of 1.92 x 109 to 9.1 x 109 cfu^g.

28. A method of obtaining manufacturable antibodies as soluble Fabs from the antibody phage display library as claimed in claim 1 in a defined order comprising the steps of:

i) target specific panning;

ii) periplasmic quantitative ELISA (qELISA);

iii) kinetic ranking;

iv) bioassay;

v) manufacturability assessment;

resulting in a phenotype to expected genotype correlation of >90% in the antibodies so obtained at step 28(iii).

29. A method of obtaining antibodies from the antibody phage display library as claimed in claim 28(i), wherein the panning is conducted in solid or solution phase at various temperatures ranging between 4 and 37 °C and for various lengths of time ranging between lh and 16h. The solid phase panning may comprise the steps of:

i) optimizing the maximal coating concentration for a given antigen on a solid surface such as charged polystyrene;

ii) conversion of the phagemid library (as obtained in claims 24 to 27) to phage format;

iii) coating the selected surface with the optimal concentration of the antigen as determined at step 29(i) followed by blocking with protein or non-protein molecules to block non-specific sites;

iv) pre-adsorption of phage pool as obtained at step 29(ii) on unblocked polystyrene surface to eliminate plastic binders;

v) incubation of pre-adsorbed phages from step 29(iv) with immobilized target antigen from step 29(iii) for defined periods of time;

vi) multiple rounds of washings to eliminate unbound phages from step 29(v);

vii) elution of bound phages from step 29(v) by trypsin digestion and concurrent transduction in amber suppressor as well as non-amber suppressor hosts to obtain phage titers;

viii) amplification of eluted phages from step 29(vii) by transducing in amber suppressor host for next round of panning;

ix) performing the next round of panning by using reduced antigen concentration and repeating steps 29(iii) to 29(viii) to enrich the target specific antibody population;

x) repetition of steps 29(vii) to 29(ix);

xi) evaluation of eluted phages from step 29(vii) and 29(x) for enrichment of binding over rounds of panning using target specific ELISA.

The solution phase panning may comprise the steps of:

xii) optimizing the reaction conditions for optimal biotinylation of a given antigen to achieve a biotin to protein molar ratio of <10, preferable 1-5;

xiii) conversion of the phagemid library (as obtained in claims 21-27) to phage format;

xiv) blocking the phages obtained at step 29(xiii) with protein or non-protein molecules to block non-specific sites for defined periods of time simultaneous with streptavidin bead washing followed by blocking the beads with protein or non-protein molecules to block non-specific sites;

xv) incubation of blocked phages from step 29(xiv) with soluble target biotinylated antigen from step 29(xii) for defined periods of time;

xvi) incubation of phage-antigen complex obtained at step 29(xv) with pre- blocked streptavidin beads;

xvii) multiple rounds of washings of the beads bound to antigen-phage conjugates at step 29(xvi) to eliminate unbound phages;

xviii) elution of bound phages at step 29(xvi) by DTT or trypsin digestion and concurrent transduction in amber suppressor as well as non-amber suppressor hosts to obtain phage titers;

xix) amplification of eluted phages from step 29(xviii) by transducing in amber suppressor host for next round of panning;

xx) performing the next round of panning by using reduced antigen concentration and repeating steps 29(xiv) to 29(xviii) to enrich the target specific antibody population;

xxi) repetition of step 29(xix) to step 29(xx);

xxii) evaluation of eluted phages from step 29(xviii) and 29(xxi) for enrichment of binding over rounds of panning using target specific ELISA.

30. A method of obtaining antibodies from the panned antibody phage display library as claimed in claim 28(ii), wherein the periplasmic quantitative ELISA comprises the steps of:

i) obtaining soluble Fabs from single bacterial colonies from eluate titer plates;

ii) coating the surface of 96-well charged polystyrene plates with a capture antibody against heavy chain;

iii) capturing the soluble Fab from step 30(i) on the coated surface of step 30(ii)

iv) detection of light chain by utilization of light chain specific antibody to identify full length, tandem in-frame, heterodimeric, soluble Fabs.

31. A method of obtaining soluble Fabs from the panned antibody phage display library as claimed in claim 30(i), wherein obtaining the soluble Fabs comprises the steps of:

i) picking single clones from titer plates of non-amber suppressor hosts (claims 29(vii) and 29(xviii) and liquid culture in 96-well deepwell plates for overnight growth at 37°C and 250 rpm;

ii) diluting the overnight cultures 10-folds and allowing growth to log phase under conditions identical to 31(i);

iii) inducing the log phase cultures at step 31(ii) with lmM IPTG and allowing overnight growth at 30°C and 250 rpm;

iv) centrifuging the cultures at step 31(iii) in 96-well plates to pellet down the induced cells;

v) periplasmic extraction of the pelleted cells at step 31(iv) by using high concentrations of EDTA in a buffered solution while slowly shaking the buffer-suspended cells in the same 96-well plate overnight at 30°C;

vi) centrifugation to isolate the diffused periplasmic fraction at step 31(v) away from the spheroplast and cell debris.

32. A method of obtaining antibodies from the antibody phage display library as claimed in claim 30(ii), wherein the surface may be charged polystyrene surface such as MaxiSorp™ or PolySorp™, or may be coated with avidin or streptavidin or neutravidin, preferably the Maxisorp™ surface is coated with streptavidin at a concentration ranging between 20 and 100μg/ml, most preferably 100μg/ml.

33. A method of obtaining antibodies from the antibody phage display library as claimed in claim 30(iii), wherein the capture antibody is selected from the group comprising goat anti-human IgG (goat anti-Human IgG(H+L); F(ab')2 fragment) or Capture Select Biotin Anti-IgG-CHl Conjugate, preferably the biotinylated anti-CHl antibody at a concentration of 1000-lOOng/ml, most preferably 250ng/ml.

34. A method of obtaining antibodies from the antibody phage display library as claimed in claim 30(iv), wherein the light chain specific antibody is selected from the group comprising goat anti-human lambda LC specific peroxidase conjugate, goat anti- human kappa LC specific peroxidase conjugate, goat anti-human F(ab')2-HRP, mouse anti-human kappa light chain peroxidase conjugate, mouse anti-human kappa light chain monoclonal and rabbit anti-human kappa chain monoclonal, preferably at a dilution ranging between 1-20000, most preferably 1: 10000 for anti-lambda and 1:2000 for anti-kappa.

35. A method of obtaining antibodies from the antibody phage display library as claimed in claim 28(iii), wherein the kinetic ranking comprises the steps of:

i) obtaining soluble Fabs from quantitative ELISA positive clones (as obtained in claims 30-34) in 50ml individual cultures;

ii) dialyzing the obtained Fabs from step 35(i) against lxPBS;

iii) use of running buffer of physiological strength and pH for kinetic analysis - the buffer could be phosphate or HEPES, more preferably phosphate, containing NaCl or KC1 concentration of 0.1 to 1.0M, preferably 0.25 to 0.75M, more preferably 0.4 to 0.6M, and Tween-20 concentration of 0.005 to 0.05%;

iv) selecting the SPR (surface plasmon resonance) chip immobilization surface - such surface could be charged dextran, charged alginate, nickel nitrilotetraacetic acid coated on charged dextran or alginate surface, or streptavidin or neutravidin coated on charged dextran or alginate surface;

v) selecting the immobilization chemistry for the SPR surface at step 35(iv) - such chemistry could be amine coupling using EDAC(l-Ethyl-3-(3- dimethylaminopropyl) carbodiimide) and sulfo-NHS (N-hydroxysuccinimide), Ni2+ charging using lOmM nickel sulfate, or streptavidin-biotin recognition chemistry;

vi) immobilizing the anti-Fab capture antibody on the chip surface from step 35(v) - capture antibodies may include anti-Fab IgG, anti-tag antibodies such as anti-His, anti- HA or biotinylated anti-Cnl or biotinylated bivalent anti- CHl nti-C or biotinylated anti-CHl/anti-C or a 50:50 mixture of both biotinylated bivalent anti-CHl/anti-C and biotinylated anti-CHl/anti-C ;

vii) capturing the crude periplasmic Fabs obtained from step 35(ii) on the capture antibody-coated surface of the chip from step 35(vi);

viii) signal stabilization by 1-3 rounds of running buffer injection over the chip surface with intermediate pause of 2-15 min;

ix) testing the analyte response on captured Fabs of step 35(vii) at an optimal concentration of analyte to distinguish between target antigen binders and non- binders;

x) removing the Fab-analyte complex using regenerating reagent for the surface to be re-used for the next round of screening - regenerating agent could include 2M MgCl2, 0.85% H3P04, 50mM NaOH or lOmM glycine, pH 2.0.

36. Use of the APDL as claimed in claim 1 for obtaining antibodies for diagnostic and therapeutic purposes and in vitro assays.