By-passing in vitro screening--next generation sequencing technologies applied to antibody display and in silico candidate selection

Abstract

In recent years, unprecedented DNA sequencing capacity provided by next generation sequencing (NGS) has revolutionized genomic research. Combining the Illumina sequencing platform and a scFv library designed to confine diversity to both CDR3, >1.9 × 10(7) sequences have been generated. This approach allowed for in depth analysis of the library's diversity, provided sequence information on virtually all scFv during selection for binding to two targets and a global view of these enrichment processes. Using the most frequent heavy chain CDR3 sequences, primers were designed to rescue scFv from the third selection round. Identification, based on sequence frequency, retrieved the most potent scFv and valuable candidates that were missed using classical in vitro screening. Thus, by combining NGS with display technologies, laborious and time consuming upfront screening can be by-passed or complemented and valuable insights into the selection process can be obtained to improve library design and understanding of antibody repertoires.

Figure 1.

Schematic representation of immunoglobulin heavy and light chain variable regions and CDR3 diversification strategy. (A) Framework regions (FR1 to FR4) and CDR regions (H1 to H3 and L1 to L3) are indicated. Stars indicate the location of Type IIS restriction sites in the stuffer fragment located between FR3 and FR4 and in the flanking regions of the diversified VH_CDR3 and L3. The sizes of the designed VH_CDR3 and L3 are 9–15 and 8–11 amino acids, respectively. The arrows indicate the location of the primers used for next generation sequencing and the dashed arrows indicate the region covered by the 76 bp reads. The heavy chain primer, which is located in the FR4 region, permits partial sequencing of the FR3 region (VH_FR4-CDR3-FR3). In contrast, as the light chain primer is located downstream of the FR4 region, the sequencing read does not cover the full CDR3 sequence (VL_CDR3-FR4). (B) ScFv rescue strategy based on VH_CDR3 sequences. Arrows indicate complementary primers corresponding to a VH_CDR3 sequence as well as primers flanking the scFv coding region. These primers were used to amplify and assemble the scFv sequence from the pool of clones following the third selection round.

Figure 2.

Germline gene family analysis. (A) Frequency of heavy chain variable gene families identified and (B) proportion of V κ and V λ light chains in the AE1 library and after each selection round (R1–R3) against the target, 5E3. (C) Frequency of heavy chain variable gene families identified in the AE1 library and after each selection round (R1–R3) against the target, hIFNγ. Sequences were considered as undetermined if they did not match exactly the signature sequence used for family assignment.

Figure 3.

Frequency of VH_CDR3 lengths and distribution within the three VH families included in the AE1 library and after each selection round against different targets. VH_CDR3 lengths are expressed as amino acids. VH_CDR3 lengths of 9–15 amino acids were included in the library design. Bars appearing between these defined lengths correspond to non-functional VH_CDR3 that are out of frame due to errors in oligonuclotide synthesis or cloning artifacts. (A) library, (C–D) selection rounds against 5E3, (E–G) selection rounds against hIFNγ.

Figure 4.

Frequency and evolution of top 10 sequences. Frequency of the 10 VH_CDR3 sequences that were the most abundant after the third selection round against 5E3 (A) and against hIFNγ. (B) Amino acid sequence, length of the VH_CDR3 according to the IMGT nomenclature and VH family are shown. Sequences that were also identified during screening by ELISA as well as clones that were rescued based on their VH_CDR3 sequence and frequency are indicated.

Figure 5.

Frequency and evolution of CDR3 sequences identified by classical binding screening against the target 5E3. Frequency evolution of VH_FR4-CDR3-FR3 (A) and VL_CDR3-FR4 (B) sequences corresponding to six scFv binding specifically to the monoclonal 5E3. The EC₅₀ for binding of the soluble scFv to the target in ELISA and the frequency ranking after the third round of selection are indicated below each clone.