Hybridization-based antibody cDNA recovery for the production of recombinant antibodies identified by repertoire sequencing

Abstract

High-throughput sequencing of the antibody repertoire is enabling a thorough analysis of B cell diversity and clonal selection, which may improve the novel antibody discovery process. Theoretically, an adequate bioinformatic analysis could allow identification of candidate antigen-specific antibodies, requiring their recombinant production for experimental validation of their specificity. Gene synthesis is commonly used for the generation of recombinant antibodies identified in silico. Novel strategies that bypass gene synthesis could offer more accessible antibody identification and validation alternatives. We developed a hybridization-based recovery strategy that targets the complementarity-determining region 3 (CDRH3) for the enrichment of cDNA of candidate antigen-specific antibody sequences. Ten clonal groups of interest were identified through bioinformatic analysis of the heavy chain antibody repertoire of mice immunized with hen egg white lysozyme (HEL). cDNA from eight of the targeted clonal groups was recovered efficiently, leading to the generation of recombinant antibodies. One representative heavy chain sequence from each clonal group recovered was paired with previously reported anti-HEL light chains to generate full antibodies, later tested for HEL-binding capacity. The recovery process proposed represents a simple and scalable molecular strategy that could enhance antibody identification and specificity assessment, enabling a more cost-efficient generation of recombinant antibodies.

Introduction

The adaptive immune system is capable of generating a vast antibody response against a great variety of possible pathogens through clonal selection. The antibody repertoire is dynamically shaped in response to external stimuli. Natural infection or immunization can evoke a strong polarization of the repertoire by selecting high affinity antigen-specific clonal groups. The high-throughput sequencing of the antibody transcripts has enabled a more thorough analysis of the immunoglobulin diversity^1,2 as a means to describe the complexity and dynamics of the antibody-mediated defense process, which has led to a better understanding of the adaptive immune response^3-5 and an improved antibody discovery process.^6,7

Used as tools for diagnosis, therapy and research, recombinant antibodies are indispensable in the biomedical field. Antibodies can be generated against different pathogens and many other molecules of interest, but the production of new recombinant antibodies against specific antigens is directly limited by the discovery strategy used to identify such antibodies. Traditionally, approaches such as hybridomas⁸ and phage display have been implemented to address this issue.⁹ More recently, techniques such as single B cell expression cloning, B cell immortalization^10-13 and deep sequencing of the antibody repertoire are improving the analysis of the human antibody response. The analysis of the antibody repertoire through high-throughput sequencing has been used to study the diversity of the immune response, and for the identification of antibodies of interest.^6,14-16 With an adequate bioinformatic analysis, it is possible to identify putative antigen-specific antibodies from the repertoire sequences based on certain parameters such as clonal frequency⁶ or structural relations with highly mutated antibodies of known specificity, such as broadly neutralizing antibodies against HIV.¹⁷ The production of recombinant antibodies based on the bioinformatic analysis of the repertoire relies mainly on the synthesis of DNA of specific light and heavy variable regions; however, the cost of the synthesis of DNA is a direct constraint affecting the evaluation throughput.⁶

The CDRH3 is the region with the highest diversity within the heavy chain variable region (VH) and a major determinant of antigen specificity.^15,16 Selecting the CDRH3 is an effective way to molecularly recover single clonal groups. In this study, we provide a proof-of-principle of targeting the CDRH3 by using restrictive PCR along with a hybridization-based recovery, allowing clonal groups with different relative abundances to be enriched and produced recombinantly without the need of chemical synthesis of the whole gene.

Results

Evidence of selection through enrichment

An immunoglobulin sequencing experiment was performed by immunizing two BALB-c mice with HEL at days 0 and 10. Spleens were collected at day 15, which corresponds to the peak of germinal center formation, somatic hypermutation and class switching. Spleen total RNA was used to generate IGHV amplicons of the IgG compartment followed by high throughput sequencing with the 454-Roche platform. A total of 45,996 and 36,512 reads were obtained from library A (LA) and B (LB), respectively. By clustering sequences according to VDJ rearrangement, CDRH3 length and identity, 783 (LA) and 902 (LB) clonal groups were identified (see Materials and Methods: Bioinformatics analysis and candidate selection).

To simplify the bioinformatic analysis, only the frequency of each clonal group was considered in the selection; the five most frequent from each library were selected for recovery (2G1–2G5, 3G1–3G5). The same amplicon libraries used for pyrosequencing were also used as templates for the amplification of the rearrangements found in the clonal groups of interest. Structural features of the five most frequent clonal groups in each mouse and the CDRH3 used for primers design are described in Table 1.

Table 1. Primer sequences and V/J usage.

Clonal group	Probe sequence (5′→3′)	IGHV and IGHJ usage‡
2G1	/B/CAAAGTA CCCCGTGTTA CCAACATT	IGHV14–4*02 IGHJ2*01
2G2	/B/GTAAGTCG TGGTAAGCAA TCCCACAT	IGHV14–4*02 IGHJ2*01
2G3	/B/CCAGGAG TAATGAACAT TACAG	IGHV14–4*02 IGHJ3*01
2G4	/B/AAACCCTC CGTACCCCGC TCTT	IGHV1–18*01 IGHJ3*01
2G5	/B/TAACGGG CCGTAGTACC CTCTC	IGHV1–63*02 IGHJ1*01
3G1	/B/TACCCGT AGTTACCCTC TCCTCTTGC	IGHV1–9*01 IGHJ2*01
3G2	/B/CCTTCGT AATCATACCC ATCCCCTTCT C	IGHV4–1*02 IGHJ3*01
3G3	/B/GTAAGGA GTCGTAATCA TAGCTGC	IGHV1–85*01 IGHJ3*01
3G4	/B/AAAGGGC TTCCCCTGCA TAAC	IGHV14–4*02 IGHJ3*01
3G5	/B/AACTACC GTAGTAATCT CCCCTCTTG	IGHV1–76*01 IGHJ2*01

/B/: Biotin molecule at 5′ end; bold sequence: sequence within the IGHJ segment; underlined sequence: sequence within the IGHV segment. ‡ Determined by IMGT/V-QUEST.

The whole enrichment process comprises two main selection points (Fig. 1). The first selection consisted of the PCR amplification of the desired rearrangements using IGHV and IGJH family-specific primers, for which stringent PCR conditions were used to minimize the amplification of undesired but structurally similar rearrangements. Ideally, only sequences with the same IGHV-IGHJ usage to that of the target group would reach this step. The PCR products generated at this point served as the input material for the second selection in which a biotinylated CDRH3-specific oligonucleotide was used for hybridization-based enrichment. After this step, most of the recovered sequences should have the desired rearrangement and CDRH3.

Figure 1. Heavy-chain variable region sequence recovery strategy. Clonal VH sequences are recovered from the whole VH repertoire through two main molecular selection steps and one following checkpoint. PCR products representing the VH repertoire are the starting material for the enrichment and also used for massive sequencing. The first selection step amplifies through PCR the desired IGHV-IGHJ rearrangement, while the second step recovers all molecules with the desired CDRH3 sequence through a hybridization-based strategy. These two selection steps are determined by the in silico selection that had been previously performed. A final checkpoint is performed to ensure the presence and overrepresentation of the desired sequences, which can then be used for cloning and recombinant expression.

As a preliminary assessment of the enrichment process, a PCR checkpoint was performed immediately after the second selection (Fig. 2). The enrichment products were used as templates in PCR with the IGHV- IGHJ and IGHV-CDRH3 primers pairs, comparing the enriched samples and the negative controls (mock enrichments). As this system enables the enrichment of the desired sequences instead of a complete isolation of them, undesired rearrangements were expected to inevitably yield an amplification product for the mock enrichments. Therefore, it was important to determine the number of amplification cycles at which the unwanted sequences (background seen in the mock enrichments) reach a significant amount to minimize the abundance of irrelevant sequences (Fig. 2A). Additionally, the amplification pattern observed when using IGHV-IGHJ and IGHV-CDRH3 primers pairs should be similar, since it is expected that the majority of the sequences amplified by the IGHV-IGHJ pair are the ones amplified by the IGHV-CDRH3 pair. Nevertheless, a notable difference in the abundance of PCR product between the enriched sample and mock enrichment was observed for all clonal groups (Fig. 2B). This amplification difference is the evidence of enrichment of the desired molecules.

Figure 2. PCR checkpoint after clonal enrichment. (A) An agarose gel stained with ethidium bromide with inverted grayscale showing the PCR products of the clonal enrichment of group 2G3 and a negative control (mock enrichment). Two main products were amplified, one covering the whole VH region (VH-JH) and the other up to the CDRH3 (IGHV-CDRH3), of approximately 350 and 300 base pair, respectively. Three different amplification cycles were used (13X, 15X, and 18X) to determine the amplification threshold required for further processing. In the enriched sample, both products (IGHV-IGHJ and IGHV-CDRH3) are present, while they are slightly visible in 18X of the mock control. This difference demonstrates a selection for VH sequences containing the desired CDRH3. (B) PCR products of the ten enriched samples and their respective mock controls shown in separate agarose gels stained with ethidium bromide with an inverted grayscale. For each group, the enriched sample (left) and mock control (right) amplifications were performed with the indicated amplification cycles and annealing temperatures. The PCR products from the enriched samples shown here were used for cloning and sequencing.

Validation of the enrichment of desired clonal groups by sequence analysis

Once the proper number of amplification cycles was determined for each group, the IGHV-IGHJ amplicon generated under such conditions was cloned into the expression vector. Only those clones with a CDRH3-specific insert detected through colony PCR were Sanger-sequenced. Assuming that those sequences amplified with their CDRH3 primer were also subject to enrichment in the recovery process, the positive to negative CDRH3-specific colony PCR ratio was used as an indicator of the enrichment efficiency (Fig. 3A).

Figure 3. Enrichment assessment through single sequence analysis. (A) Representation of the concentrations of each of the eight clonal groups that were successfully recovered, before (gray) and after (yellow) the enrichment. Concentrations before enrichment were determined in silico by the analysis of the massive sequencing results; the number of sequences considered for 2G1–2G5 was 46 062, and 36 393 for 3G1–3G4. Concentrations after enrichment were determined through PCR amplification of the desired IGHV-CDRH3 fragment from cloned enrichment products; N indicates the total number of clones evaluated in each case. The circle representing the total (blue) is used as a reference of 100% of the area. (B) CDRH3 conservation after enrichment determined through Sanger sequencing. A few clones for each clonal group were sequenced (N) and compared with their respective target sequence, showing that most of the sequences obtained had no or little variation in the CDRH3. (C) HEL-binding ELISA for the three positive combinations (VH/VL) of the 32 tested. 2G3/HH10, 3G1/HH5 and 3G1/F10.6.6 showed HEL-binding capacity, corresponding to two VH of the eight tested. HH10/HH10 served as positive control, and no-DNA transfectant was used as negative control. The y-axis shows the ratio between the OD₄₉₀ from the HEL-binding ELISA and the OD₄₉₀ from the IgG detection ELISA of the same sample.

This assumption, however, could consider CDRH3 primer mismatch to highly similar but not identical sequences. Therefore, 6 to 15 colonies from each of the eight clonal groups were sequenced and aligned against their corresponding target sequence. In eight of the ten groups, sequences with a correct rearrangement and identical CDRH3 were recovered (Fig. 3B), with an overall sequence identity of ≥98.6% ± 1.1. Partially complementary sequences with one or two mismatches in the CDRH3 were also recovered for these eight groups. Sequences with three or more mismatches in the CDRH3 were observed only in four of the groups. Nevertheless, some of these mismatches were synonymous, coding for the correct amino acid (Fig. S1). Recovery of groups 3G2 and 3G5 was unsuccessful; only partially complementary sequences or wrong rearrangements were obtained.

Expression of recovered antigen-specific recombinant antibodies

One sequence for each of the eight clonal groups recovered was selected to be recombinantly expressed in HEK 293T cells. Although the recovery process considered only VH sequences, interchangeable murine light-chain variable regions (VL) were used to generate complete recombinant chimeric IgG.^18,19 Because we did not have VL massive sequencing results readily available, the VH recovered were paired with four different VL (HH10, HH5,¹⁸ D1.3²⁰ and F10.6.6²¹) that had been previously reported to belong to HEL-binding antibodies. A total of 32 monoclonal antibodies (plus controls) were generated and evaluated by ELISA to assess expression and antigen-binding capacity. One irrelevant VH (from an unimmunized mouse) was paired with the four VL as negative controls to discard any evident specificity contribution by the VL chosen (data not shown).

The VH/VL combinations 2G3/HH10, 3G1/HH5 and 3G1/F10.6.6 were positive to the HEL-binding ELISA (Fig. 3C). In this case, the 3G1 VH showed binding capacity when paired separately with two different VL. Therefore, two of the eight VH evaluated showed binding capacity to HEL when paired to some of the VL chosen. The antigen-binding capacity of the antibodies obtained, however, does not rely on the enrichment process, but depends directly on an effective bioinformatic analysis and immunization protocol.

Discussion

Massive sequencing provides a broad approach to study the repertoire dynamics in an immune response to an antigen, as well as the identification of antigen-specific sequences. The in silico candidate selection of the target sequences is commonly followed by synthetic production of the desired genes. The cost of gene synthesis, however, limits their experimental specificity validation and, in some cases, may restrict the selection by relying on consensus sequences. In this study, we propose a strategy for the recovery of heavy chain variable region cDNA to enrich the sequences of clonal groups of interest and produce them recombinantly, aiding experimental validation for the in silico analysis without the use of synthetic genes.

Two main alternative strategies, one mechanical and the other molecular, have been reported that bypass DNA synthesis when producing recombinant antibodies from digital repertoires. In the first case, a micromanipulator was used to retrieve specific sequencing beads coated with the desired DNA²² used in the 454 sequencing platform. This strategy requires a sophisticated manipulating device, and can only be used for the 454 sequencing system.

In the second strategy, a molecular recovery of the specific heavy chain antibody sequences was implemented to target the CDRH3 through an overlap PCR.¹⁶ However, the use of a CDRH3 primer in PCR can artificially add the desired sequence to a slightly different CDRH3 with which the primer can still hybridize and extend. Because the IGHV-CDRH3 and CDRH3-IGHJ fractions are first amplified and later joined, it is possible for two fractions that do not belong to each other to be joined, generating artificial sequences. This particular detail did not represent a problem because this strategy was tested with samples in which diversity was restricted only to the CDRH3. The recovery strategy proposed in our study is not solely based in PCR and is capable of retrieving complete biological variable region sequences. Theoretically, its application could be extended to other massive sequencing platforms.

The recovery process proposed in this study consists of two main selection points, one for the selection of the desired rearrangement through PCR, and the other for the selection of CDRH3 through a hybridization-based enrichment. These two filters proved to be sufficient to recover eight of ten expanded clonal groups from two diverse samples. The success in the first selection point relies on the design of the primers and the amplification conditions. Although the set of primers used in this study was intended to select specific murine IGHV and IGHJ families, amplification was not entirely restricted to them because the primers sequence could also be completely or partially shared with other families. There are several primers set proposed for the selective amplification of murine IGHV and IGHJ subfamilies, however most do not cover the whole variable region, from the first to the last codon.^23-29 Despite the fact that the primers set used in this study lacked strict subfamily selectivity, such primer restriction was proven not to be essential. Nevertheless, the design of an extended and more selective set of primers would improve the overall recovery outcome. A similar approach can be done to amplify human VH sequences to broaden the extent of this strategy trough the design of a new set of primers.

The hybridization-based CDRH3 selection represents the main clonal selection step at which a CDRH3-specific probe was used to recover each group. Biotinylated oligonucleotides have been previously used for the enrichment of specific transcripts or other target molecules within a mixture,^30,31 and for capturing CDRs to exploit sequence diversity for combinatorial biology approaches.³² However, as far as we know, there are no reports of a targeted recovery of cDNA of specific antibody variable regions with this strategy, possibly due to the challenge it represents.

Since the CDRH3 sequence serves as a tag mark to identify clonal groups, a unique probe had to be designed for each group. The coverage of the probe was limited mainly to the IGHD and junctions (N and P nucleotides) because this region is the most diverse and distinctive for each clonal group within the repertoire. Inclusion of a high number of positions into the IGHV and IGHJ segments, resulting in the use of large probes, yields partially complementary sequences and hinders the recovery process (data not shown). Therefore, clonal groups with short and low diversity CDRH3 may be more difficult to recover.

The CDRH3-specific PCR after the VH cloning served as an indicator of the enrichment efficiency. The initial representation of the clonal groups before the enrichment started at 2.1% and up to 9.5% in the filtered digital repertoire. Thus, the actual representation is even lower, considering the sequences that were taken out in the in silico quality filters. At the CDRH3-specific colony PCR, group 2G5 was enriched up to a 100% of the colonies evaluated, and the lowest was 31% for 2G4. The abundance achieved after the enrichment varied among the eight clonal groups recovered. However, sequences with an identical CDRH3 were recovered in each case by testing a few colonies.

The presence of an amplification product at the CDRH3-specific colony PCR (checkpoint) did not ensure an identical CDRH3. Therefore, the clones were sequenced, showing that identical and nearly identical CDRH3 were recovered. The sequence of group 2G3 that was used in the transfections contained one non-synonymous mismatch (W→L) at the CDRH3 and still showed antigen-binding capacity. This indicates that sequences with one or two mismatches at the CDRH3 should still be considered for further evaluation, unless an amino acid known to be essential is lost.

We managed to recover sequences with a correct rearrangement and identical CDRH3 to that of the target molecule, whereas further variability was found along the rest of the VH region. This particular variety, presumably within the same clonal group, could be especially convenient when comparing different idiotypes with a common clonal origin. Through this approach one could identify specific positions in the VH of matured antibodies that confer them higher affinity when compared with their precursors.

While the discovery of HEL-binding antibodies from the enriched VH sequences does not provide any information regarding the recovery success, it does indicate if an appropriate bioinformatic selection was performed. The combinations 2G3/HH10, 3G1/HH5 and 3G1/F10.6.6 were positive to the HEL-binding ELISA. The fact that the same VH paired with two different VL retained antigen-binding capacity reaffirmed that generic VL may be used for screening, as it has been previously reported.^18,19,33 However, a more diverse and better selected pool of VL could allow high-throughput screening. Possibly, some of the negative VH recovered sequences could show antigen-binding capacity if paired with other VL sequences. To overcome VH-VL mispairing, massive sequencing of native VH-VL pairs could also be incorporated.³⁴ Considering that the bioinformatic analysis was scarce, that the identification was solely based on frequency and no other parameters (e.g., hypermutation rate, CDRH3 composition and framework/CDR mutation ratio), and that non-native light chains were used, the observed outcome was better than expected.

The antibody repertoire is a very diverse and complex mixture of closely related sequences. Hence, an efficient isolation of a single sequence from such a complex mixture is not straight forward. The method proposed in this study allowed the recovery of specific clonal groups with simple molecular techniques. This strategy can yet be further improved and expanded, even to similar immune diverse samples, such as the T cell receptor repertoire.³⁵ Coupled with improved bioinformatics methods to identify putative antigen-specific antibodies, this recovery method may also provide the proof of concept for scalable and automated devices for high-throughput antibody production.

In summary, ten presumably antigen-specific clonal groups were identified in silico from a VH antibody repertoire. Eight of these clonal groups were recovered through a PCR and hybridization-based strategy that enabled their recombinant expression. Sequences with the correct rearrangement and identical CDRH3 were recovered in each case. This alternative recovery strategy allows the enrichment of specific clonal groups without the need of synthetic DNA through the use of everyday molecular biology techniques.

Materials and Methods

Mice immunization

Adult (6–8 wk old) male BALB-c mice were immunized twice intraperitoneally with HEL (Cat. L6876, Sigma). The first immunization was performed with 30 μg of HEL with incomplete Freud’s adjuvant. Ten days later, a second immunization was performed with 5 μg of HEL with the same adjuvant. The spleen was removed 15 d after the first immunization and frozen in liquid nitrogen until further processing.

Libraries preparation and massive sequencing

Total RNA was extracted with Trizol from the immunized mice spleens. Splenic RNA (1.5 μg) was used to generate VH amplicons by a 5′ RACE-PCR approach. Briefly, total RNA was used to generate cDNA with a d(T)30 primer. In the 3′ end of the cDNA, a universal primer containing the adaptor A sequence (454-Roche) was incorporated using the protocol described by Matz et al.³⁶ IgG VH amplicons were generated by PCR with IgG-specific barcoded primers fused with adaptor B sequence (454-Roche) to allow discrimination of sequencing reads according to each mouse (5′-AdaptorB/4mer-454key/6mer-barcode/ CAGGGG CCAGTGGATA GACHGATGG-3′). Amplicons were quantified with the Agilent Bioanalyzer and pooled in a 1:1 molar ratio. Amplicons were sequenced with the GS FLX Titanium platform (454-Roche) in a 2-region gasket pooled with other barcoded libraries following the standard amplicon sequencing protocol. To have good quality coverage and depth in the CDRH3 region, sequencing was performed from IGHC to IGHV direction using primer B.

Bioinformatic analysis and candidate selection

The default signal processing pipeline for amplicon was used. The resulting SFF files were filtered using PRINTSEQ³⁷ to filter out reads > 200 bp or with average Q values > 30, and were split according library barcode. We have developed a pipeline for automated analysis of clonal diversity in Ig Rep-Seq data (data not shown). To define clonal groups, IGHV and IGHJ segments for each read are mapped with IgBLAST,³⁸ and the CDRH3 sequence is extracted with HMMER (http://hmmer.janelia.org/). For all the reads sharing the same VDJ rearrangement, the algorithm recursively clusters CDRH3 having ≥ 98% identical length and ≥ 98% sequence identity with USEARCH.³⁹ The result is the consensus of each clonal group in FASTA format, complemented with the number of reads belonging to each clonal group that is used as proxy of clonal frequency. Since the libraries are derived from mRNA, clonal groups having several reads could be derived from clonal expansion of few plasma cells. Higher frequency clonal groups from each mouse were selected as putative anti-HEL specific clonal groups as described previously.⁶

CDRH3 probes design

One CDRH3 probe was designed for each of the ten clonal groups intended to be recovered. They were designed to cover mainly the CDRH3 region and a few nucleotides from the end of IGHV and the beginning of IGHJ segments, according to the IMGT/V-QUEST annotation.⁴⁰ Ideally, no more than five positions outside the CDRH3 were included in each side, unless necessary to avoid significant secondary structures. These probes contained the complementary sequence of the CDRH3 so they could also be used in PCR as reverse primers along with a forward primer at an upstream position. Each probe was synthesized with a biotin molecule at its 5′ end (IDT).

Rearrangement-specific PCR

The reactions were performed with the following final concentrations: 1.6 fM VH PCR product as template, 0.5 mM IGHV forward primer, 0.5 mM IGHJ reverse primer, 1X PCR buffer, 0.2 mM dNTP’s, 2.5 mM MgSO₄ and 0.025 U/μL Platinum Taq DNA polymerase High Fidelity (Invitrogen). Three forward 5′-phosphorylated primers for the amplification of murine IGHV families IGHV1 (5′-AATTCAGAGG TCCAGCTGCA GCAGTCTGGA-3′), IGHV4 (5′-AATTCAGAGG TGAAGCTTCT CGAG-3′) and IGHV14 (5′-AATTCAGAGG TTCAGCTGCA GCAGTCTGGG-3′) were used separately. Three reverse primers were used for the amplification of the IGHJ families: IGHJ1 (5′-CTAGCTGAGG AGACGGTGAC CG-3′), IGHJ2 (5′-CTAGCTGAGG AGACTGTGAG AGTGGTG-3′) and IGHJ3 (5′-CTAGCTGCAG AGACAGTGAC CAGA-3′). These primers contained the partial recognition sites for EcoRI (forward) or NheI (reverse), and a phosphate at the 5′ end to allow cohesive-end cloning through Exonuclease III digestion (Sigma). Stringent amplification conditions were used through a touchdown PCR scheme. Amplification started with 95 °C for 5 min; 35 cycles of melting at 95 °C for 30 s, annealing at X °C for 30 s, and extension at 72 °C for 30 s; and a final extension of 72 °C for 5 min. The annealing temperature (X) decreased 2 °C (X-2 °C) at cycle six, 1 °C (X-3 °C) at cycle 11 and 2 °C (X-5 °C) at cycle 16. The initial annealing temperature (X) for the primers pairs IGHV1-IGHJ1, IGHV1-IGHJ2, IGHV1-IGHJ3, IGHV14-IGHJ2 was 81 °C; for IGHV4-IGHJ3 was 75 °C; for IGHV1-IGHJ3 was 71 °C; and for IGHV13-IGHJ3 was 64 °C. The PCR products were gel-purified before any further processing.

Clonal group cDNA enrichment

Streptavidin-coated magnetic beads were used to recover the hybridized target-molecule/biotinylated-probe complexes. Five μL of magnetic beads (Enrichment beads, 454-Roche sequencing system) per sample were pre-washed twice with 200 μL of enhancing buffer (454-Roche sequencing system)⁴¹ every time, retrieving the beads with a magnetic base (Cat. S1506S, New England Biolabs), and finally resuspended in 10 μL of enhancing buffer per sample. For every clonal group intended to recover, 7.5 ng of the corresponding rearrangement-specific PCR product were mixed with annealing buffer (454-Roche sequencing system) up to 26 μL and heated at 95 °C for 5 min, followed by immediate cooling in ice for 2 min. Four μL of the CDRH3-specific biotinylated probe at 5 pmol/μL were added, proceeding to an incubation 5 °C above the corresponding Tm for 8 min and cooling in ice for 2 min. Ten μL of pre-washed magnetic beads were added to every sample and incubated in gentle rotation at room temperature for 7 min. The buffer was removed from the magnetic beads in a magnetic base once the solution had completely cleared (≥90 s). The remaining beads were washed ten times with 500 μL of enhancing buffer. Thirty μL of previously warmed water at 95 °C were added to the beads and were incubated at 95 °C for 5 min. The beads were immediately transferred to a magnetic base and the enrichment product solution was recovered once it had completely cleared (90 s). Along with every enriched sample, a mock enrichment (negative control) was performed, for which the biotinylated probe was replaced with annealing buffer.

Enrichment testing and VH cloning

The enriched VH sequences were cloned into the expression vector pVAJO-CHG1, derived from pFUSEss-CHIg-hG1 (InvivoGen), which contains the human IgG1 constant region and an omega-cassette⁴² that confers resistance to chloramphenicol. To generate the inserts, the VH sequences were PCR-amplified from the enrichment products using the corresponding IGHV and IGHJ primers. The amount of template used represented 10% of the final volume of the reaction; however, various separate reactions were done to cover increasing amplification cycles, starting at 13 and up to 20. The amplification started at 95 °C for 5 min; 13 to 20 cycles of 95 °C for 30 s, X-5 °C for 30 s, 72 °C for 30 s; and a final extension of 72 °C for 5 min. The value of X and chemical composition of the reaction are as mentioned in the rearrangement-specific PCR methods. Along with every sample, the negative controls of the enrichment were also tested by PCR and analyzed by electrophoresis in a 1% agarose gel. The amplification products generated from an enriched sample were visually compared with the ones generated from the corresponding mock enrichments. The final number of cycles elected was the highest one at which a notable product was generated from the enriched sample, but none or very scarce was obtained from the mock enrichment. Once the final number of amplification cycles was defined for every sample, the reactions were repeated only for the chosen cycles and using the Direct Universal Cloning System (Sigma). The new PCR products were digested with Exonuclease III, and the pVAJO-CHG1 with EcoRI and NheI (Fermentas). Digestion with Exonuclease III and the cloning procedures were as indicated by the manufacturer. The introduction of recombinant plasmids in Escherichia coli strain TOP10 was performed by electroporation. The TOP10 strain was grown at 37 °C in Luria-Bertani medium and, when required, chloramphenicol was added to a final concentration of 30 μg/mL.

Transformants were analyzed by colony PCR for inserts by using in-vector primers, and for the specific CDRH3 by using the IGHV and CDRH3 (biotinylated) primers. Only those colonies positive for both PCRs were Sanger-sequenced. The clone with the closest sequence to its target was used for transfections. The recovered VH sequences were paired with four VL sequences: HH10 [GenBank: M35667.1], D1.3 [PDB: 1FDL_L], HH5 [PDB: 1BQL_L] and F10.6.6 [PDB: 2Q76_A], previously reported as positive for HEL binding.^18,20,21 The VL sequences were synthesized (gBlocks, IDT) and cloned into the expression vector pVAJO-CLhk, derived from pFUSE2ss-CLIg-hk (InvivoGen), which contains the human kappa constant region and an omega-cassette that confers resistance to chloramphenicol.

Recombinant antibody expression

HEK 293T cells (Thermo Scientific) were plated at a density of 125 000 cells per well in 24-well plates with DMEM complete medium (supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin (Life Technologies) and 10% FBS) and incubated at 37 °C and 5% CO₂. After 24 h, the cells reached 70–80% confluence and the growth medium was replaced with low-FBS DMEM (2% FBS). For each well, 2 μg of total DNA were used for transfections. Endotoxin-free plasmids were mixed in a 1:3 mass ratio of the heavy and light chains, respectively. The DNA was mixed with 2.5 μL of 2 M CaCl₂ and water to a 20 μL volume. Finally, 20 μL of 2X HBS (180 mM NaCl, 50 mM acid-free HEPES, 2 mM Na₂HPO₄, pH 7.1) were added and the transfectant was applied to the cells. The growth medium was replaced with low-glucose DMEM (12–725F, Lonza) without FBS, 24 h after transfection. The supernatants were harvested and passed through 0.2 μm filters (Advantec) 72 h after transfection. Sodium azide was added to a final concentration of 0.01% to each sample, which were then stored at 4 °C.

IgG and HEL-binding ELISA

ELISAs were performed in 96-well plates (Thermo Scientific) coated overnight (4 °C) with a goat anti-human IgG primary antibody (109–005–098, Jackson ImmunoResearch) at 10 μg/mL for the detection of IgG; or HEL (L6876, Sigma) at 50 μg/mL for the detection of HEL-binding antibodies. The plates were washed three times with PBS/0.1% Tween-20 and blocked with PBS/3% BSA (IgG ELISA) or PBS/5% skimmed milk (HEL ELISA) for 1 h at room temperature. After washing the plates three more times, 50 μL of the supernatants containing IgG were transferred to the wells and incubated for 2 h at room temperature with gentle shaking. Bound IgG were detected with a goat anti-human IgG antibody coupled with HRP (ab98567, Abcam). The assay was developed with OPD and H₂O₂, and 15 min later the absorbance was read at 490 nm.

Glossary

Abbreviations:

HEL

hen egg lysozyme

CDRH3

heavy chain complementarity determining region 3

FBS

fetal bovine serum

10.4161/mabs.27435