Skip to main content

Generation 3 Platform

Specifica’s Generation 3 Library Platform has quality built in by design. Developability, extremely high diversity, and the exclusion of sequence liabilities are intrinsic to the four sub-libraries making up the Platform, each of which is based on a therapeutic antibody scaffold chosen for its biophysical properties, lack of liabilities and germline gene variety.

Diversity is all derived from natural CDR sequences: HCDR3s are amplified directly from purified B cells, while replicated natural CDRs, derived from Specifica’s databases, are used for the rest. The same unique HCDR3 diversity can be repackaged into different formats, including VHH, scFv and Fab. In the case of scFv and Fab, libraries can also be made with common light chains as well as traditionally.

Visionary antibody technologies

Why Generation 3?

Generation 3

  • B cells purified from Leukopaks
  • Amplification of B cell HCDR3s
  • 5 of 6 CDRs replicated natural
  • Developable clinical scaffolds
  • All but HCDR3 sequence liabilities purged
  • NGS quality control
  • HCDR3 diversity >108

Generation 2

  • B cells purified from Leukopaks
  • Amplification of VH / VL
  • Individual primer pairs
  • Biased towards better V genes
  • Sequence liabilities at natural frequencies
  • NGS quality control
  • HCDR3 Diversity: 107-8

Generation 1

  • Non purified peripheral B cells
  • Amplification of VH / VL
  • Pooled primers
  • Unbiased V genes
  • Sequence liabilities at natural frequencies
  • No NGS quality control
  • HCDR3 Diversity: 3×106

Library design

The exclusive use of combinatorially reassembled natural CDR sequences, rather than intra-CDR synthetic diversity, ensures correct folding, avoids co-variance violations, and allows the complete elimination of sequence based liabilities for five of the six CDRs.

As a result of the high functionality of our Generation 3 Library Platform, you can expect a broad diversity of different antibodies from your selections, most of which will have affinities below 10 nM, combined with developability properties usually at least as good as the clinical candidate antibodies from which they were derived.

We construct each Generation 3 library de novo, ensuring your library comprises a unique combination of naturally replicated CDRs, combined with at least one hundred million unique HCDR3s derived from a donor pool used only once, guaranteeing each supplied Generation 3 Library Platform is one of a kind.

Customization

Specifica builds Generation 3 libraries using validated therapeutic frameworks and HCDR1-2 and LCDR1-3 diversity sets.

In addition to building libraries with these source elements, all library aspects can be customized, including scaffolds, mutations to revert frameworks to germline or eliminate sequences of concern, antibody format, CDR sequence liabilities retained or eliminated, HCDR3 sources, display platform (phage or yeast), and the peptide tags used for detection or purification.

We will work with you to design the optimal library for your purposes.

Affinity

One silver lining to the emergence of COVID-19 has been the intense focus of numerous groups to solving problems related to the pandemic, providing a unique global opportunity to directly compare different technologies applied to the same problem. In a recently published paper (Hastie, 2021, Science) the affinities of 181 antibodies from 56 different groups were assessed using a single technique (surface plasmon resonance) in a single lab, on a single platform (Carterra LSA).

173 antibodies generated directly from the Generation 3 Platform using AbXtract demonstrated far better affinities than this global dataset, with over 30 having affinities better than 100 pM (and the best at 13 pM).

This is consistent with our finding that 60% of antibodies selected directly from the Generation 3 scFv Platform have affinities below 10nM, with ~20% in the sub-nanomolar range.

Developability

Inherent within the design and construction of the Generation 3 Platform is the ability to select high affinity, drug-like, therapeutic antibodies directly from the naive libraries, without the need for downstream optimization.

Confirmation of desired developability characteristics of selected antibodies was assessed by the expression of full length IgGs from selected antibodies and testing for: HEK titer, Tm, salt-gradient affinity capture self-interaction nanoparticle spectroscopy, ELISA on a panel of commonly used targets, cross-interaction chromatography and size-exclusion chromatography within the context of accelerated stability.

Over 80% of full length IgG tested to date behave as well as, or better, than the therapeutic scaffolds from which they are derived.

Broad Diversity

Antibodies selected from our Generation 3 Library Platform show broad paratopic diversity by next generation sequencing of antibody populations fully positive for targets of interest.

We translate the extensive sequence diversity into clontoypes, or clusters, using proprietary machine learning algorithms embedded within AbXtract. Each cluster is made up of up to one hundred different but related sequences.

Our selection pipeline usually identifies one hundred to one thousand different clusters, with Levenshtein distances between different clusters of 9-14 for the HCDR3 and 30-40 for concatenated CDRs, as illustrated in the figures.

This extensive clonotypic diversity provides antibodies against a wide variety of target epitopes important for drug efficacy. Different antibodies within specific clusters show similar epitopic binding, but variations in affinity, allowing straightforward exploration of the role of affinity in biological activity.

Antibodies in the same cluster may also differ in the presence or absence of HCDR3 sequence liabilities, often allowing selection of antibodies with reduced liabilities, but similar biological activity.

Levenshtein distances between different antibody clonotypes selected against a therapeutic target for differing numbers of CDRs, from all six (top) to HCDR3 only (bottom).

Continue Exploring

Go Back

Continue Exploring

Read More