1
i l f i i f h i Precise Control of Diversity for Synthetic Precise Control of Diversity for Synthetic Precise Control of Diversity for Synthetic A tib d Lib D i &C t ti Antibody Library Design & Construction Antibody Library Design & Construction Chris Hebel 1 Christoph Eicken 1 Qi Zhu 1 Xiaochuan Zhou 1,2 Xiaolian Gao 3 Chris Hebel 1 , Christoph Eicken 1 Qi Zhu 1 , Xiaochuan Zhou 1,2 , Xiaolian Gao 3 , 1 LC Sciences and 2 Atactic Technologies, Houston, TX 77054 , 3 Dept of Biology and Biochemistry, University of Houston, Houston, TX 77004 LC Sciences and Atactic Technologies, Houston, TX 77054 , Dept of Biology and Biochemistry, University of Houston, Houston, TX 77004 Electron microscope Synthetic antibody libraries have proven to be effective tools for drug discovery and development through the generation of functional, high-affinity antibodies against a wide variety photo of the Paraflo ® Synthetic antibody libraries have proven to be effective tools for drug discovery and development through the generation of functional, high affinity antibodies against a wide variety of antigens They are an expanding alternative to standard hybridoma technology especially for application to particularly difficult therapeutic challenges that cannot be addressed chip. of antigens. They are an expanding alternative to standard hybridoma technology especially for application to particularly difficult therapeutic challenges that cannot be addressed with antibodies from the natural repertoire The performance of a synthetic antibody library depends in large part on the diversity of the library which must be designed based on It features: with antibodies from the natural repertoire. The performance of a synthetic antibody library depends in large part on the diversity of the library which must be designed based on th h d t di f th tib d t t df ti F d di it id ffi i t th t tib d didt d i df ti l f i • 10 μl total volume thorough understanding of the antibody structure and function. Focused diversity can provide an efficient path to antibody candidates designed for exceptional performance in • 10 μl total volume 4000 rxn chambers specialized applications if precise control over design and construction is exercised. The use of degenerate oligos and other standard methods of diversity introduction lack this precise • 4000 rxn chambers control and can introduce unwanted or useless codons into the library, thus limiting its performance. Fully designed library diversity is enabled through parallel in-situ (on-chip) • 270 pl/rxn chamber control and can introduce unwanted or useless codons into the library, thus limiting its performance. Fully designed library diversity is enabled through parallel in situ (on chip) synthesis of tens of thousands of specific (non-degenerate) oligonucleotide sequences We demonstrate the bioinformatics-based design and high-throughput synthesis of a mutant • uniform flow rate synthesis of tens of thousands of specific (non-degenerate) oligonucleotide sequences. We demonstrate the bioinformatics-based design and high-throughput synthesis of a mutant phage display library to improve affinity of anti ErbB2 single chain monoclonal antibody A21 across the chip phage display library to improve affinity of anti-ErbB2 single chain monoclonal antibody A21. bl h lh h fl d fl ® fl d l h l f Programmable DLP Photolithography Microfluidic Reaction Devices μParaflo ® Microfluidics Oligo Synthesis Platform Digital Light Projection drives light directed chemical reactions at specific sites in an array Three components are at the core of the μParaflo ® Technology: Mi fl idi hi (4K 30K format and eliminates the need for expensive, inconvenient microfabricated photomasks. A Fluid Reaction Chamber Microfluidic chip (4K -30K ti h b ) id computer generates the digital mask and a Digital Light Projector (DLP) projects the light beam 1. Photogenerated reagent chemistry (PGR -chemistry). Cover Glass Fluid Channel Chamber reaction chambers) provides l d t th t very accurately into the micro reaction chambers where a photogenerated reagent is produced. 2 Digital photolithography directed (programmable) synthesis Channel an enclosed system that f ilit t th f th 2. Digital photolithography directed (programmable) synthesis. Distribution facilitates the use of the l ti ht h it 3. Microfluidic devices containing high density 3D-reaction chambers of pico-liter volumes. Distribution Channel solution photochemistry, t t if it 3. Microfluidic devices containing high density 3D reaction chambers of pico liter volumes. Channel Ph t G t d A id (PGA) D t ti Ch it promotes spot uniformity d d ibilit d Photo Generated Acid (PGA) Deprotection Chemistry and reproducibility and h ki ti f Light Beam enhances kinetics for h i l th i ll Light Beam A novel solution photochemical approach employing chemical synthesis as well i bi di / t Photogenerated R light directed parallel synthesis and deprotection as various binding/capture Reagent l k with a photogenerated reagent enables high-yield assays. Uniform Flow Sequence input & Digital mask Programmable light parallel synthesis using standard DMT protected Si Substrate Distribution Across all R ti Ch b Sequence input & mask generation projection Programmable light driven reactions phosphoramidites. Si Substrate Reaction Chambers mask generation driven reactions Other Methods Have Limitations Too Limits of Conventional Codon Randomization Methods Designed Library Diversity with Specific Oligo Sequences Other Methods Have Limitations Too Limits of Conventional Codon Randomization Methods Designed Library Diversity with Specific Oligo Sequences Library Design & Diversity Amino Acid Codon Representations Using Different Randemization Synthesis Methods • TRIM – Trinucleotide Mutagenesis - The codon-based Library Design & Diversity Synthesis Units (3) NNG/T (4) NNC/G/T (1) NNN (2) NNC/T trinucleotide synthesis for building oligonucleotides provides Diversity - the variability carried by the amino acid # of Codon Amino Acid 32 48 64 32 specific codon for each amino acid, but it is expensive to be of sequences of a synthetic antibody library Amino Acid Codon Presence# Presence% Presence# Presence% Presence# Presence% Presence# Presence% 32 48 64 32 routine use and suffers from biased or low efficiency coupling Alanine(A) GCX 4 6.25 2 6.25 2 6.25 3 6.25 Arginine(R) CGX, AGA/G 6 9.38 2 6.25 3 9.38 4 8.33 issues. • Synthetic libraries are diversified by design – design is in-silico using one of many Asparagine(N) AAC/T 2 3.13 2 6.25 1 3.13 2 4.17 Aspartic Acid(D) GAC/T 2 3 13 2 6 25 1 3 13 2 4 17 • S lit Mi S litM th d Off b tt t l f i id published algorithms. Naturally occurring antibodies provide guidance on how to limit Aspartic Acid(D) GAC/T 2 3.13 2 6.25 1 3.13 2 4.17 Cysteine(C) UGC/T 2 3.13 2 6.25 1 3.13 2 4.17 • Split -Mix-Split Method – Offers better control of amino acid di t ib ti d iti f d i d h synthetic diversity. Glutamine(Q) CAA/G 2 3.13 0 0.00 1 3.13 1 2.08 Glutamic acid(E) GAA/G 2 3.13 0 0.00 1 3.13 1 2.08 distribution and composition of randomized sequences, however, th i i ifi t l it d t ti l f h • Diversity designs can be created in one or more of the light and/or heavy chain Glutamic acid(E) GAA/G 2 3.13 0 0.00 1 3.13 1 2.08 Glycine(G) GGX 4 6.25 2 6.25 2 6.25 3 6.25 Hi tidi (H) CAC/T 2 3 13 2 6 25 1 3 13 2 4 17 there is significant complexity and potential for human error as th i i i d lti l ti Diversity designs can be created in one or more of the light and/or heavy chain variable domains (V H and V L ) and by varying the length of the hypervariable loops Histidine(H) CAC/T 2 3.13 2 6.25 1 3.13 2 4.17 Isoleucine(I) ATA/C/T 3 4.69 2 6.25 1 3.13 2 4.17 synthesis resins are mixed multiple times. variable domains (V H and V L ) and by varying the length of the hypervariable loops present in the variable domain (length diversity) to allow diverse structural Leucine(L) CTX,TTA/G 6 9.38 2 6.25 3 9.38 4 8.33 Lysine(K) AAA/G 2 3 13 0 0 00 1 3 13 1 2 08 • XYZ or Custom Mixtures of building blocks can combat codon present in the variable domain (length diversity) to allow diverse structural conformations to form in this region and further increase the diversity of the library Lysine(K) AAA/G 2 3.13 0 0.00 1 3.13 1 2.08 Methionine(M) ATG 1 1.56 0 0.00 1 3.13 1 2.08 Ph lli (F) TTC/T 2 3 13 2 6 25 1 3 13 2 4 17 biases but also are limited by the potential for imprecise conformations to form in this region and further increase the diversity of the library. • Often randomization or tailored randomization of oligonucleotide sequences is Phenylalnine(F) TTC/T 2 3.13 2 6.25 1 3.13 2 4.17 Proline(P) CCX 4 6.25 2 6.25 2 6.25 3 6.25 mixing/handling. • Often randomization or tailored randomization of oligonucleotide sequences is employed to introduce diversity Serine(S) UCX,AGC/T 6 9.38 4 12.50 3 9.38 5 10.42 Threonine(T) ACX 4 6 25 2 6 25 2 6 25 3 6 25 employed to introduce diversity. h bili i l d fi h fi l di i f lib f ili h Threonine(T) ACX 4 6.25 2 6.25 2 6.25 3 6.25 Tryptophan(W) UGG 1 1.56 0 0.00 1 3.13 1 2.08 • The ability to precisely define the final diversity of a library facilitates the fi l i h ii d i ii ib d l d Tyrosine(Y) UAC/T 2 3.13 2 6.25 1 3.13 2 4.17 Valine(V) GTX 4 6.25 2 6.25 2 6.25 3 6.25 Synthetic Antibody Library Construction Process process of isolating, characterizing, and optimizing an antibody lead. Stop codon UAA/G,UGA 3 4.69 0 0.00 1 3.13 1 2.08 Synthetic Antibody Library Construction Process Max. 9.38 12.50 9.38 10.42 Min 1 56 0 00 3 13 2 08 Min. 1.56 0.00 3.13 2.08 in excess 6.0 3.0 5.0 Recombinant Monoclonal Antibody Library Recombinant Monoclonal Antibody Library Li k L1 L2 L3 N • NNN d Linker L1 L2 L3 N • NNN codon • has six fold difference in amino acid representation. H3 H2 H1 C • the amount of the expressed full length sequences drops quickly due to stop codon H3 H2 H1 C • the amount of the expressed full-length sequences drops quickly due to stop codon termination and codon-dependent low expression Antigenic ligand binding termination and codon-dependent low expression. CDRH1 Sequence ligand binding to surfaces of CDRs • generates an impractically large library where almost all variants are inactive, reducing th lik lih d f fi di i t ith d i bl ti Sequence Variations CDRs the likelihood of finding a variant with desirable properties. • NNC/T codon is incomplete in representing amino acids (missing 5 of the 20 amino acids). • NNG/T and NNC/G/T codons overcome some limitations but still suffer from the difficulties • NNG/T and NNC/G/T codons overcome some limitations but still suffer from the difficulties of being inflexible in codon choices and sequence bias of being inflexible in codon choices and sequence bias. • None of the above methods would allow deletion or insertions. • None offer the flexibility in choosing specific codons for protein expression optimization. None offer the flexibility in choosing specific codons for protein expression optimization. Example: Recombinant l l tib d lib Designed (Non-Random) Sequence Variations monoclonal antibody library Designed (Non Random) Sequence Variations Application Ex Directed Evolution for Increased Affinity Bioinformatics Based Library Design Results Application Ex – Directed Evolution for Increased Affinity Bioinformatics Based Library Design Results Bi i f ti b dd i dhi h th h t i fl idi th i f Nt ll i tib di id id h t li it th ti di it A Bioinformatics-based design and high-throughput microfluidic synthesis of Naturally occurring antibodies provide guidance on how to limit synthetic diversity. An l ith hl t dt i d t d P d d it d d h Several rounds of screening process mutant phage display library to improve affinity of single chain antibody A21 algorithm helps us to determine advantageous codons. Pseudo-codons are introduced such th t th t t lib ti i bl bi ti f t ti (d t i d Scanning Liquid biotin antigen screening more than 10 3 clones that the mutant library contains every conceivable combination of mutations (determined t b d t ) t t h ti ll f f hil ii ii th Scanning - Liquid biotin antigen screening, more than 10 3 clones Anti ERBB2 Antibody A21 to be advantageous) at a stochastically even frequency of occurrence while minimizing the it d ti f d i d id d th if tl b di t l tib di Greater than 200 clones Anti-ERBB2 Antibody A21 introduction of undesired residues and those infrequently observed in natural antibodies. The next (l ll d / d 8) 8 b d l h A21 CDR region Kabat classification Screening - Screening to clone the soluble antibody and phage antibodies by ELISA round of screening ERBB2 (also called HER2/neu and p185) is a 185 KDa trans-membrane epidermal growth f ( )f l h k A21 CDR region Kabat classification CDRs Sequence Class Single Mutation Double Mutation Total phage antibodies by ELISA screening factor receptor (EGFR) family protein with intrinsic tyrosine kinase activity. ERBB2 contains l l bl h h l d bl f h h Positions Peptide Seq Positions Peptide Seq Positions Peptide Seq ~30-50 clones SPR determination of wild-type and mutant single chain antibody affinity C2C10B several plausible tyrosine phosphorylation sites and is capable of associating with other h d fk l d h L1 KSSQTLLYSNNQKNYLA Class 3 13 272 84 2372 97 2644 A21scFv Western-Blotting or SPR Preliminary single-chain antibody affinity C2C10B EGFR proteins to enhance a cascade of kinase activities in signal transduction pathways k b ll lf d/ ll h h L2 WAFTRKS Class 1 7 42 24 5461 31 5503 comparison of affinity known to be active in cell proliferation and/or cell apoptosis. ERBB2 is shown to have up- l d l l h h b d L3 QQYSNYPWT Class 1 8 72 28 6176 36 6248 ~20 clones regulated expression in several malignant human tumors such as breast and ovarian d l ll k h l d H1 GYSFTGYFIN Class 1 9 90 36 3219 45 3309 cancer compared to normal cells, making the protein an intensively pursued target as h d b k f ll d H2 HISSSYATSTYNQKFKN Class 2 17 272 136 19362 153 19634 CDR, L3 Sequence analysis cancer therapeutics and as a biomarker for tumor cell detection. H3 SGNYEEYAMDY Length specified 11 110 55 20425 66 20535 5 10 clones bd d h hb f ll h Q Q Y S N Y P W T specified Total 65 858 363 57015 428 57873 ~5-10 clones Antibody A21 is a predominant growth inhibitor of cells over expressing the ERBB2 gene h f kd d d hh ff Determination of the ka (1/Ms) kd (1/s) KD with significant in vivo work done toward producing higher affinity mutants. However, bd ff f fl d bd d fh d Lib C t ti relative affinity of eukaryotic expression SKN3 1.72e5 8.99e-5 5.21e-10 antibody-affinity maturation in vivo often fails to produce antibody drugs of the targeted k f h ff b ld dd d Library Construction eukaryotic expression C2C10B 2.56e5 3.73e-6 1.45e-11 potency making further in vitro affinity maturation by computational design and directed l It k i i ! evolution necessary. Oligonucleotide Synthesis It works in vivo! It works in vitro! Oligonucleotide Synthesis It works in vitro! Non-random C l Non random mutagenic oligos are synthesized on Complex tl are synthesized on a microfluidic chip crystal Synthesized oligos are cleaved from the Eukaryotic expression vector: chip and recovered in solution mutant antibodies with human IgG Fc fusion protein chA21 chA21 Epitopes chA21 chA21 Epitopes LC Sciences LLC LC Sciences, LLC 2575 W Bellfort Suite 270 2575 W. Bellfort, Suite 270 Houston Texas 77054 Houston, Texas 77054 T l 713 664 7087 Tel. 713 664-7087 Tel. 713 664 7087 F 713 664 8181 Fax 713 664-8181 www lcsciences com www.lcsciences.com
