Evaluation of Antigen-Conjugated Fluorescent Beads to Identify Antigen-Specific B Cells

Selection of single antigen-specific B cells to identify their expressed antibodies is of considerable interest for evaluating human immune responses. Here, we present a method to identify single antibody-expressing cells using antigen-conjugated fluorescent beads. To establish this, we selected Folate Receptor alpha (FRα) as a model antigen and a mouse B cell line, expressing both the soluble and the membrane-bound forms of a human/mouse chimeric antibody (MOv18 IgG1) specific for FRα, as test antibody-expressing cells. Beads were conjugated to FRα using streptavidin/avidin-biotin bridges and used to select single cells expressing the membrane-bound form of anti-FRα. Bead-bound cells were single cell-sorted and processed for single cell RNA retrotranscription and PCR to isolate antibody heavy and light chain variable regions. Variable regions were then cloned and expressed as human IgG1/k antibodies. Like the original clone, engineered antibodies from single cells recognized native FRα. To evaluate whether antigen-coated beads could identify specific antibody-expressing cells in mixed immune cell populations, human peripheral blood mononuclear cells (PBMCs) were spiked with test antibody-expressing cells. Antigen-specific cells could comprise up to 75% of cells selected with antigen-conjugated beads when the frequency of the antigen-positive cells was 1:100 or higher. In PBMC pools, beads conjugated to recombinant antigens FRα and HER2 bound antigen-specific anti-FRα MOv18 and anti-HER2 Trastuzumab antibody-expressing cells, respectively. From melanoma patient-derived B cells selected with melanoma cell line-derived protein-coated fluorescent beads, we generated a monoclonal antibody that recognized melanoma antigen-coated beads. This approach may be further developed to facilitate analysis of B cells and their antibody profiles at the single cell level and to help unravel humoral immune repertoires.


releVance TO The research TOPic: single cell aPPrOaches TO sTUDY The iMMUne sYsTeM
The selection and study of single antigen-specific B cells and the identification of their expressed antibodies can help shed light on the nature and functions of the human humoral immune response. In our study, we present an antigen-conjugated fluorescent bead methodology designed to identify and isolate single antibody-expressing cells and to derive and clone matched heavy and light chain antibody variable regions into full length antibodies. We believe that our single cell-to-functional antibody strategy falls within the scope of this research topic, as it opens the way for opportunities to analyze B cells and their antibody profiles at the single cell level and may be potentially applied to help unravel diverse humoral immune repertoires in different health and disease states.

inTrODUcTiOn
The study of humoral immunity may be key to understand human health, aging, and disease, as well as to predict or monitor host immune responses to pathogen challenge and disease progression (1). Moreover, dissecting B cell responses could help inform medical interventions such as preventative vaccines or therapeutic treatments (2,3). Currently, therapeutic monoclonal antibodies are part of the standard care of treatment for various conditions including diabetes, autoimmune and allergic diseases, and various types of cancer. A big proportion of these antibodies are fully mouse or mouse-human chimeras (4). Therefore, identification of fully human, heavy (H), and light (L) variable regionmatched antibodies may be of interest for clinical applications, as administration of fully human antibodies is considered less likely to induce antidrug antibody responses compared with chimeric or humanized antibodies (5). Immune monitoring of human B cells recognizing specific antigens, alongside identification, cloning, and production of their cognate monoclonal antibodies, may, therefore, be of substantial value to help evaluate reactivity and immunological response to these antigens.
Analyses of the antibody repertoires of B cells have been reported in the human setting and in animal models, with the most common applications in the study of autoimmune diseases (6), viral infections, and B cell malignancies (7)(8)(9). Some studies have employed high-throughput sequencing of heavy chain variable regions or of paired H and L variable region sequences (10,11). Ultra-high-throughput sequencing techniques have also been developed, capable of analyzing repertoires of thousands of B cells from small human blood samples, with some allowing the mapping of paired H and L variable regions, in a few hours (12)(13)(14).
B cells may also be a source of antigen-specific antibodies for clinical diagnostics or for therapeutic applications. Existing methods to identify antigen-specific B cells and generate monoclonal antibodies include ex vivo B cell culture approaches, which could promote the survival and expansion of certain B cell subsets, screening of the culture supernatants to identify B cell reactivity and fluorescent-activated cell sorting (15)(16)(17)(18)(19)(20). An essential element in the process of selecting antigen-specific B cells is detection of antibodies with a certain degree of specificity. This could be achieved by screening cell culture supernatants through ELISPOT assays or ELISA-based methods using immobilized recombinant antigens or cells (16,20). Screening cell culture supernatants by ELISA, although highly sensitive, represents only a surrogate parameter and antigen reactivity should ultimately be confirmed only after sequencing and expression of the selected clone. For all these applications, the gold standard of identifying antigen-specific antibodies remains the expression of the recombinant antibody and further evaluation of its antigen recognition properties. Workflows to facilitate selection of single human B cells without ex vivo growth, stimulation, and clone expansion, and which do not require sampling of cell culture supernatants could offer additional tools for the study of human B cell immunity. Novel approaches to address these requirements involve the use of modified fluorescent tetramers for direct B cell screening by fluorescent-activated cell sorting (21,22).
In this study, we describe the design of a bead-based methodology to identify single antibody-expressing B cells, and to clone and produce antigen-specific antibodies. The workflow features bead-based identification and isolation of specific B cells using direct fluorescent-activated cell sorting, sequencing, and cloning of matched heavy and light chain variable regions in a single full sequence antibody expression vector system, and expression and testing the antigenic reactivity of the antibody clone. The workflow is designed to avoid ex vivo B cell expansion and secondary clone selection and to facilitate antibody generation and downstream evaluation.

MaTerials anD MeThODs human samples
Human immune cells were isolated from venous blood of healthy volunteers and patients with malignant melanoma. Specimens were collected with informed written consent in accordance with the Declaration of Helsinki. The study was conducted at King's College London, King's College London, Guy's and St Thomas' NHS Foundation Trust (08/H0804/139 approved by London Bridge NRES committee; 16/LO/0366 approved by London-Central NRES Committee). Human peripheral blood mononuclear cells (PBMC) were isolated from 40 ml blood using Ficoll ® Paque Plus density centrifugation (GE Healthcare).

cell culture
Cell culture was performed using aseptic technique at 37°C in a humidified atmosphere in 5% CO2, unless otherwise specified. The human ovarian carcinoma cell line IGROV1 naturally over-expressing folate receptor alpha (FRα) was grown in RPMI 1640 GlutaMAX™ medium (Thermo Scientific) supplemented with 10% fetal calf serum (FCS). The human breast cancer cell line MDA-MB-231 was grown in DMEM GlutaMAX™ medium (Thermo Scientific) supplemented with 10% FCS. The permanently transfected murine myeloma cell lines SP2/0-MOv18 specific for FRα and SP2/0-SF25, recognizing a colon carcinoma antigen (23), were cultured in Dulbecco's Modified Eagle's Medium plus 10% FCS as previously described (24). The human embryonic kidney cell lines, Expi293F cells, were cultured in serum-free Expi293 expression medium (Thermo Scientific) on a Stuart orbital shaker at 125 rpm at 8% CO2.

Transient expression of human Monoclonal antibodies in expi293F cells
Expi293F cells were transfected with pVitro1-hygro-mcs antibody constructs using the ExpiFectamine293 Transfection kit (Thermo Scientific) as per manufacturer's instructions. The anti-human epidermal growth factor receptor 2 (HER2) and the melanomaassociated antigen-specific chondroitin sulfate proteoglycan (CSPG4) antibody constructs were previously described (25,26).

Fluorescent Beads
Different avidin-or streptavidin-coated fluorescent beads of different sizes were used (Table S1 in

generation of Melanoma cell line Protein extracts
To biotinylate proteins of melanoma cell lines, confluent cell monolayers were washed with PBS and borate buffer pH 9. Then, sulfo-NHS-LC-biotin at 1 mg/ml in PBS was added and incubated for 20 min on ice. The reaction was quenched with 200 mM Tris, 120 mM NaCl pH 7.4, and the cells were subsequently extensively washed with PBS. Cell lysis buffer (Cell Signaling Technology) containing 0.1% Triton and supplemented with 2 mM PMSF was added to the monolayer and the cells were harvested by scrapping. Lysates were briefly sonicated and incubated on an orbital rocker for 30 min at 4 C. The lysates were cleared by centrifugation at 900 × g for 15 min, followed by a second centrifugation at 12,000 × g for 30 min. Supernatants were either used immediately or stored at −80°C.

coupling of recombinant Proteins to Fluorescent Beads
Recombinant Folate Receptor α (FRα) (R&D Systems, Cat no 5646-FR) and human epidermal growth factor receptor 2 (ErbB2/HER2) Fc chimera (R&D Systems, Cat no 1129-ER) were reconstituted in PBS at 0.1 mg/ml as recommended by the manufacturer. Biotins with three different arms were used: (i) sulfo-NHS-LC-biotin with a spacer arm of 22.4 Å; (ii) sulfo-NHS-LC-LC-biotin with a spacer arm of 30.5 Å, and (iii) NHS-PEG12-biotin with a spacer arm of 56 Å (all from Pierce-Thermo Fisher Scientific). The three forms of biotin were reconstituted following the manufacturer's instructions and mixed with the recombinant proteins in PBS at a 20:1 (biotin:recombinant protein) molar ratio and incubated for 90 min at room temperature (RT). Free biotin was removed by dialysis in PBS.
For binding of biotin-conjugated FRα to all Sphero fluorescent particles, 100 µl of beads were washed with 1 ml of 1% BSA-PBS-0.05% sodium azide, spun at 13,000 × g for 20 min at RT, and the pellet was resuspended in the original volume of beads with 1% BSA-PBS-0.05% sodium azide. Then, 25 µg of biotin-conjugated recombinant protein was added and mixed, by rotation, for 2-16 h at RT. After binding, beads were washed three times with 1 ml 1% BSA-PBS-0.05% sodium azide and resuspended in 100 µl of 1% BSA-PBS-0.05% sodium azide.
For binding of biotin-conjugated cell extracts to LumAvidin 5.6 µm microspheres, 200 µl of microsphere solution were washed with 1 ml of 1% BSA-PBS-0.05% sodium azide, spun at 8,000 × g for 2 min, and pellets were resuspended in 200 µl of 1% BSA-PBS-0.05% sodium azide. Then, 300 µl of biotinconjugated cell extracts were added to the beads and mixed, by rotation, for up to 16 h at RT. After binding, beads were washed three times with 1 ml of 1% BSA-PBS-0.05% sodium azide and resuspended in 200 µl of 1% BSA-PBS-0.05% sodium azide. For each test, 15 µl beads were used.

Fluorescent labeling of conjugated Beads
To check that FRα/HER2 proteins were attached on the bead surface, conjugated beads were stained with specific antibodies for FRα/HER2. Conjugated beads (10 µl) were incubated in 200 µl of FACS buffer (5% FCS-PBS) for 20 min at RT, then, 1 µg of the monoclonal antibody MOv18 (anti-FRα), anti-HER2 [previously described in Ref. (27)] or isotype control antibodies were added. In all described experiments, the hapten-specific in-house produced monoclonal antibody anti-NIP IgG was used as an isotype control. The beads were incubated for 30 min at 4°C, washed with 2 ml of FACS buffer, and incubated with anti-human immunoglobulin (Ig) antibody conjugated to FITC (Vector Laboratories) for another 30 min at 4°C. After a final wash with FACS buffer, stained beads were analyzed on a FACSCanto™ flow cytometer (BD Biosciences). Beads coated with melanoma cell line extracts were tested by staining with a monoclonal IgG antibody recognizing the melanoma-associated antigen CSPG4.

spiking of PBMc with antibodyexpressing cells
Sp2/0 cells expressing MOv18 IgG or the control antibody SF25 were prestained with anti-mouse MHC Class I-PE antibody to be used with Streptavidin SA-Blue 1.1 µm or LumAvidin 5.6 µm beads (Table S1 in  ig Variable region cloning PCR tubes containing sorted cells in lysis buffer were thawed and reverse transcription was performed in a thermal cycler using the SuperScript VILO cDNA Synthesis kit (Thermo Fis her Scientific) following the manufacturer's recommendations. Variable regions of Ig heavy and light chains were amplified by two rounds of PCR amplification in 20 µl mix, using 10 µl of Phusion Flash polymerase mix per reaction (Finnzymes), each primer at 0.5 µM, 2 µl of the reverse transcriptase reaction as template for the first  Table 1), and the nested PCR protocol were previously described (28). PCR products were separated on agarose gels, purified using the QIAquick gel extraction kit (Qiagen), and cloned using the ZeroBlunt PCR cloning kit (Life Technologies). Six colonies were grown for each PCR band, plasmid DNA purified using QIAprep spin miniprep kit (Qiagen) and sequenced by Source Bioscience using the M13 forward primer. Sequences were analyzed using IMGT/V-QUEST (29).

recombinant antibody cloning and expression
Heavy and light chain variable regions were cloned in pVitrohygro-1 (Invivogen) using polymerase incomplete primer extension (PIPE) cloning as previously described (25). In brief, pVitro-hygro-1, containing pre-cloned human antibody heavy and light constant region cassettes (gamma 1/kappa) was used as a template in two separate PIPE PCR reactions to amplify two linear plasmid fragments with partially single-stranded 5″ ends. Similarly, the Ig heavy and light chain variable regions were PIPE PCR-amplified, using the pCR-Blunt constructs, generated previously, as PCR templates. Next, the PCR products were diluted four times with ddH2O and mixed in a ratio of 1

Workflow for identification of antibodyexpressing single cells Using antigenconjugated Fluorescent Beads
We designed a process to allow the identification of antigenspecific antibody-expressing B cells, which can be performed without prior ex vivo growth or secondary screening of B cells. To establish this system, we selected Folate Receptor alpha (FRα) as a model antigen. As the test antibody-expressing cells, we selected a B cell line expressing both the soluble and the membranebound [B cell receptor (BCR)] form of a human/mouse chimeric antibody (MOv18 IgG1 clone) specific for FRα. The workflow (Figure 1) entails coupling of fluorescent polystyrene beads with the nominal antigen, Folate Receptor alpha (FRα), followed by binding of FRα-coated beads to anti-FRα antibody-expressing B cells. Single bead-conjugated cells were identified and isolated by FACS sorting directly onto microplates containing lysis buffer. RNA released from single cells was converted to cDNA using reverse transcription followed by a semi-nested RT-PCR with Ig specific primers. Matched variable H and L chains were sequenced and cloned into a single expression vector containing the H and L constant IgG1 region sequences as previously reported (25,26). The full antibody was then expressed in a human expression host. The antibody was purified and antibody specificity for the antigen was tested on target antigen-expressing cells.

antigen-coupled Fluorescent Bead recognition by Monoclonal antibodies and by specific antibody-expressing B cells
We first established the workflow using APC fluorochrome LumAvidin Microspheres of 5.6 µm diameter (LumAvidin 5.6 µm; Table S1 in Supplementary Material). Prior to interrogating the ability of these beads to bind antibody-expressing cells, we used flow cytometric evaluations to confirm that fluorescent beads could be successfully coupled to antigens. We tested the antigens FRα and HER2 conjugated to fluorescent beads. The anti-FRα antibody MOv18 IgG bound to FRα-conjugated beads, and the HER2-specific antibody trastuzumab bound to HER-2conjugated beads. On the other hand, the hapten-specific monoclonal antibody NIP IgG showed only background binding to FRα-conjugated or to HER-2-conjugated beads, and no binding was detected on beads incubated with secondary antibody alone (Figure 2A). We also interrogated different coupling agents with varying biotin lengths (LC-biotin, 22.4 Å; LC-LC-biotin, 30.5 Å; PEG12-biotin, 56 Å) to assess whether the distance between bead and antigen from the bead surface could affect antigen recognition by anti-FRα-expressing B cells. We found no significant differences in MOv18 IgG antibody binding to bead-conjugated FRα with respect to any of these coupling agents ( Figure 2B).
We then established a model system using microspherecoupled FRα and anti-FRα antibody-expressing B cells. We confirmed that a small proportion of FRα+ LumAvidin 5.6 µm microspheres recognized single anti-FRα antibody-expressing B cells (Sp2/0-MOv18) ( Figure 3A). When single bead-bound cells were isolated into lysis buffer-containing microplate wells by flow cytometric sorting, single cell PCR yielded matched H and L chain variable region DNA products were detected to be of the expected size ( Figure 3B). Antibody variable region sequences were cloned into a pVITRO1 vector containing human IgG1/k constant region cassettes and the antibody was expressed in Expi293F™ cells. Subsequently, the cloned IgG1 antibody was used to assess binding to IGROV1 cells, which express native cell surface human FRα. Flow cytometric evaluations confirmed that the anti-FRα antibody cloned from a sorted single antibody-expressing B cell was able to specifically recognize antigen-expressing IGROV1 cells. The cloned anti-FRα antibody did not recognize MDA-MB-231 breast carcinoma cells that do not express FRα (Figure 3C).
These findings confirmed that it is possible to extract matched variable region sequences from single antibody-expressing cells selected by recognition of specific antigen-coated microspheres. Selection of single B cells allowed the identification of the expected matched variable regions and permitted the cloning and production of the corresponding monoclonal antibody from a single cell. This antibody was able to recognize natively expressed cell surface FRα.
influence of Microsphere Diameter and Biotin length on specific recognition of antigen-expressing B cells Since we observed that FRα+ microspheres were able to bind only a small proportion of single anti-FRα antibody-expressing B cells FigUre 1 | Workflow for identification of antigen-specific monoclonal antibodies derived from single cell cloning. Biotinylated antigen (Folate Receptor alpha, FRα) was conjugated to fluorescent beads conjugated to streptavidin or avidin. Antigen-coated beads could bind cells expressing an antigen-specific immunoglobulin (Ig) and bead-bound cells were purified using cell sorting. RNA from sorted cells was reverse transcribed and Ig variable regions were amplified by nested PCR, sequenced, and cloned using the dual expression plasmid pVITRO1 containing human heavy and light chain constant region cassettes (gamma1/kappa). Antibodies were expressed in a human cell expression system, purified, and tested for antigen binding.
(Sp2/0-MOv18), we investigated whether microsphere diameter or the FRα-biotin on the beads could influence specific recognition of antibody-expressing cells.
We confirmed that Sp2/0-MOv18 IgG cells expressing MOv18 IgG on the cell surface could be recognized by human recom binant FRα ( Figure 4A). As observed with recombinant antibody binding to FRα-coated microspheres (Figure 2), the distance between bead and antigen from the bead surface did not affect antigen recognition by anti-FRα antibody-expressing Sp2/0-MOv18 cells ( Figure 4B). However, the proportion of the Sp2/0-MOv18 cells recognized by fluorescent beads coated with FRα differed with bead size or type: microspheres of smaller bead diameters (SA-Red 0.5 µm and SA-Blue 1.1 µm) appeared to bind a higher proportion of B cells (70.3 and 54.9%, respectively) compared to the LumAvidin 5.6 µm beads (1.4%) (Figures 4C,D).
These findings suggest that bead size or type could influence the proportion of possible antibody-expressing B cells recognized by antigen-conjugated fluorescent beads.

evaluation of specific Detection of antibody-expressing B cells in PBMc samples by antigen-conjugated Fluorescent Beads
Since alongside specific binding of beads to antibody-expressing B cells, we also detected non-specific binding of FRα-coated beads to control Sp2/0 cells (Figure 4E), we further investigated the background binding of fluorescent beads to freshly isolated human PBMCs. A proportion (~1%) of human PBMCs (top panel) and also of human B cells (CD19+ cells, lower panel) could bind to FRα-coated beads in most likely a non-specific manner ( Figure 5A). Different blocking agents did not appear to reduce the levels of background binding (Table S2 in Supplementary Material). Furthermore, A-Red 0.8 µm, SA-Blue 1.1 µm, and LumAvidin 5.6 µm FRα-coupled (PEG12-biotin) beads were incubated with PBMCs and binding was directly compared to recognition of Sp2/0-MOv18 IgG cells (used as positive controls). FRα-coupled A-Red 0.8 µm microspheres showed 54.3% specific and 0.56% non-specific B cell recognition, while SA-Blue 1.1 µm microspheres showed 53.4% specific and 1.36% non-specific recognition of B cells. On the other hand, the larger sized LumAvidin 5.6 µm microspheres showed 8.7% specific and 0.07% non-specific binding ( Figure 5B). This suggested that the smaller diameter beads were more likely to select antibody-expressing cells.   In order to analyze whether, and at what frequency, anti-FRα antibody-expressing B cells in PBMC samples may be detected by FRα-coupled beads, we spiked human PBMC with Sp2/0-MOv18 cells at different frequencies and we detected double-positive (FRα+/Sp2/0+) events by flow cytometry (Figure 6A). FRα-coupled A-Red 0.8 µm fluorescent beads were incubated with PBMC (1 µl beads/10 6 PBMC, Table S1 in Supplementary Material) in the presence of serially diluted (1:50, 1:500, 1:5,000) Sp2/0-MOv18 cells pre-labeled with an anti-CD45 antibody. The actual dilution was calculated post acquisition ( Figure 6B). Actual positive events were defined as anti-FRα antibody-expressing B cells among all bead-bound cells and used to evaluate potential specific selection of real antigen-specific cells. The Actual true events were defined as antigen-specific B cells bound to beads among all possible antibody-specific Sp2/0+ events and used to estimate the selection of antigen-binding cells compared to all specific cells in a sample. We observed that FRα+ beads were able to identify 49% actual positive events (B cells) when the specific B cell frequency was ~1:50. The proportion of actual positive events (anti-FRα antibody-expressing B cells) was lower when the specific B cell frequencies in the PBMC pool were reduced (Figures 6B,C). The actual true events (B cells bound to beads among all possible antibody-expressing Sp2/0+ events) ranged between 9.4 and 18% (Figures 6B,C). Beads of different types and diameter sizes varied in ability to detect Actual positive events, and the proportion of Actual positive events (among all bead-bound cells) decreased when the frequencies of FRα+/Sp2/0+ B cells were lower among PBMCs. Between 5 and 75% of antibodyexpressing B cells recognized by beads were antigen-specific when B cells were found in high frequencies (≥1:100) in human blood. SA-Blue 1.1 µm LC-LC biotin beads detected the highest proportion (75%) when B cells were found at a frequency of 1:43 ( Figure 6D; Table S3 in Supplementary Material).
To confirm selection of antibody-expressing cells by specific antibody, PBMC spiking showed that FRα-coated beads were more likely than HER2-coated beads to be recognized by FRα-specific Sp2/0-MOv18 cells (Figure 7A; Figure S1 in Supplementary Material). Anti-SF25 antibody-expressing Sp2/0 cells were also less likely to be detected by FRα-coated beads in PBMC samples (Figure 7A). Similarly, HER2-coated beads were more likely to bind anti-HER2 (trastuzumab)expressing Expi293F cells compared with anti-CSPG4 control antibody-expressing Expi293F cells. HER2-coated beads were also more likely to bind anti-HER2 (trastuzumab)-expressing Expi293F cells compared with FRα-coated beads, which showed low binding to anti-HER2 (trastuzumab)-expressing Expi293F cells (Figure 7B). Human Ig expression on the surface of Expi293F cells is shown in Figure S1 in Supplementary Material. Together, these data suggest that antibody-expressing cells in PBMC preparations are able to recognize beads conjugated to their specific antigens.
To interrogate this approach in the human setting, we evaluated whether fluorescent beads conjugated to melanoma cell surface antigens were able to identify antigen-reactive B cells. We prepared fluorescent beads coated with antigens extracted from human melanoma SK-MEL-28 cells, which natively express the tumor-associated antigen chondroitin sulfate proteoglycan 4 (CSPG4). Melanoma antigen-coated beads were recognized by an anti-CSPG4 antibody but not by MOv18 antibody specific for the antigen FRα not expressed by SK-MEL-28 cells ( Figure  S2 in Supplementary Material), suggesting that antigen-reactive antibodies can specifically recognize antigen bound on these beads. We then screened for antigen-reactive antibodies from human B cells. PBMCs from patients with melanoma incubated with melanoma antigen-coated beads were screened to identify mature CD19/CD22+ B cells recognizing antigen-coated beads. Single bead-bound B cells were isolated by single cell sorting Ig heavy and light chain sequences were amplified, sequenced, and antibodies were cloned (Figures 8A,B). An antibody derived from a single bead-bound B cell could recognize melanoma antigen-coated fluorescent beads compared with secondary only or a non-specific antibody control. The B cell-derived clone showed binding comparable to that of a monoclonal antibody specific for the melanoma-associated antigen CSPG4 (Figure 8C).
Taken together, these findings suggest that antigen-reactive B cells in human blood could be recognized by antigen-conjugated fluorescent beads and could be sorted by flow cytometry for the expression of monoclonal antibodies.

DiscUssiOn
In this report, we present the development of a fluorescent bead method for the selection of single antigen-reactive B cell clones and the production of the cloned antigen-specific antibodies for downstream testing. The workflow comprises: (a) conjugation of fluorescent microspheres with a recombinant antigen of interest; (b) identification of fluorescently labeled single antibodyexpressing B cells by flow cytometry using antigen-conjugated beads; (c) flow cytometric sorting of bead-bound single cells directly into lysis buffer; (d) single cell retrotranscription and sequencing of matched heavy and light chain antibody variable regions; (e) cloning and expression using a vector containing human constant region cassettes; (f) confirmation of antigen reactivity of the produced full-length antibody. We designed this process to allow the identification of antigen-specific antibodyexpressing B cells without the requirement of prior ex vivo growth or secondary screening of B cells, and we aimed to conduct the workflow from clone identification to antibody production and characterization in a timeframe of approximately 23 days.
To design this protocol, we employed a model system featuring human recombinant Folate Receptor alpha (FRα) as the target antigen, a B cell line expressing both the soluble and the membranebound forms of MOv18 IgG1, a human/mouse chimeric antibody specific for FRα. We showed that fluorescent microspheres can be coupled with the recombinant antigen via biotin-streptavidin/ avidin bridging, and that immobilized antigens could be readily detected by antigen-specific monoclonal antibodies. Recognition of the epitope on FRα by the test antibody MOv18 is thought to be dependent on the native folding of the target antigen. Here, we demonstrated that it is possible to use antigen-coupled fluorescent beads of different sizes to identify and isolate single B cells expressing cell surface-expressed antibodies that recognize this conformational epitope. Antigen-antibody recognition could be improved by evaluating beads with different characteristics and diameters, while varying the lengths of the avidin coupling agents did not significantly influence antibody-expressing cell recognition by the beads. The latter observation also suggested that engagement of the antigen to the bead surface via different   biotins did not mask epitope recognition by specific antibody either in solution or when the antibody was expressed on the surface of a B cell.
A key feature of this streptavidin-biotin bead-based approach is that, in principle, it may enable the coupling of virtually any known native or recombinant antigen to fluorescent beads and facilitate the detection of B cells reactive to this antigen. Our strategy may offer different features compared with those of other detection methods. For instance, antibodies used for detection of BCR on B cells may also interact not only with BCRs but also with Fc receptors on human B cells (30). Tetramer technologies may be applicable to the selection of antigen-specific B cells, but fluorophores have been known to be released from antigen complexes and to bind non-specifically to immune cells, yielding false positive events (31). Our protocol is also specifically designed to avoid the requirement for B cell culture and ex vivo expansion or for antibody selection from culture supernatants prior to sub-cloning, limiting dilution and isolation of antigen-specific B cell clones. As well as being laborious and lengthy, these steps may also be limited by specific expansion of distinct B cell populations not always representative of the original B cell repertoire, and which could suffer from potential loss of the antigen-specific clone during the cell ex vivo culture processes (17).
Following flow sorting of single cells directly into lysis buffer, we confirmed that it is possible to extract matched H and L chain variable region sequences from single antibody-expressing cells selected by specific antigen-coated beads. Employing the FRα-MOv18 Sp2/0 B cell model, we confirmed that B cell selection by FRα+ fluorescent beads allowed the identification of the matched variable regions and permitted the cloning and production of the corresponding monoclonal antibody from a single cell. This was greatly expedited with the use of a single vector Expi293F™ cell line-based expression system, which permitted antibody production within a few days (25,26). While the vector used in this study contained the constant regions of human IgG1, in principle, this platform could be used for engineering of B cell-derived antibodies with constant regions of any isotype or species desired, potentially facilitating a wide range of downstream applications for this technology. Importantly, we showed that cloned and expressed full-length antibody with variable regions extracted from a single B cell could recognize native cell surface-expressed human FRα. The specificity of an identified antibody could ultimately be confirmed only by the expression of the antibody and subsequent testing against the natively expressed antigen. Previous published studies report the number of detected B cells and the antigenic reactivity of antibodies secreted in supernatants of ex vivo B cell cultures without cloning, production, or testing of the derived antibody clones (16,32). A similar B cell identification process to the one reported in our study described the frequency of antigen-specific B cells detected using a modified bead-based method (31). With this tool, it was possible to monitor the frequencies of anti-HLA, anti-tetanus toxin-, and anti-EBNA1-committed B cells in different individuals. However, antibodies from selected B cells are often not cloned and expressed subsequently in order to analyze their ability to recognize natively expressed cognate antigens. Our data confirming antibody sequencing, cloning, expression, and antigen recognition provide an early proof of principle that functional Ig sequences could be recovered with this methodology.
We ascertained that binding of FRα-coated fluorescent beads can specifically single out antibody-expressing B cells. We found that FRα-coated beads bound up to 71% of possible Sp2/0-MOv18 IgG cells. However, alongside enhanced recognition of possible antigen-reactive B cells, FRα-coated beads also bound to 6-9% of non-specific Sp2/0 B cells. Furthermore, a proportion (~1%) of human circulating B cells could bind to FRα-coated beads in most likely a non-specific manner.
Together, these suggest that background non-specific binding of beads to cells may be a limitation of this methodology. We employed two approaches to evaluate whether antigen-coupled beads could detect antigen-reactive antibody-expressing cells in PBMC samples.
Our first approach was to "spike" human PBMC with Sp2/0-MOv18 B cells at different frequencies. We demonstrated that FRα + fluorescent beads could identify antibody-expressing cells among PBMC populations when these antigen-reactive cells were present at higher frequencies in human blood. We also found that different bead types had different specific recognition and background recognition profiles. The proportion of actual positive B cells among all bead-bound cells decreased with lower frequencies of FRα+/Sp2/0+ B cells among PBMCs. When B cells were found in high frequencies (≥1:100) in human blood, the proportion of Actual positive events, the specific B cells recognized by the beads, ranged from 5 to 75% of the total bound cells, with the SA-Blue 1.1 µm LC-LC biotin beads able to detect the highest proportion (75%). Although fluorescent activated cell sorting could enable separating a single cell from a heterogeneous population, the detection of cell populations with a low frequency remains challenging and presents a technical limitation with different protocols (33,34). Consistent with previous studies, here, we observed great variability in the acquired numbers of cells when analyzing samples with very low antigen-specific B cell frequencies. This implies that only active and high frequency humoral responses may be readily studied in the present context, while the detection of low frequency B cells represents an inherent limitation of this methodology and requires further optimization. As a next step, we asked whether using this protocol, fluorescent beads coated with melanoma cell line antigens may be able to identify antigen-reactive B cells in individuals suffering of malignant melanoma. Using melanoma cell line surface antigen-coated fluorescent beads, we expressed a monoclonal antibody that bound to melanoma cell line protein-coated beads. This suggested that antigen-reactive B cells in human blood may be singled out using fluorescent beads.
The selection of single antigen-specific B cells and the identification of their expressed antibodies is critical to gaining a deeper understanding of the nature and functions of active human humoral immune responses. Identification of B cells, as well as cloning and production of their heavy and light chain matched antigen-specific monoclonal antibodies thus remain highly desirable in immunology research and antibody discovery. Consequently, discovery of antigen-specific B cell clones forms the focus of numerous high-and low-throughput approaches (12)(13)(14). Our bead-based protocol may provide an alternative, readily applicable means for exploring human B cells by facilitating the study of single B cell-derived antibodies and their functional profiles. Since the frequency of antigen-specific B cells in the human circulation remains ≤1% of the total B cell populations (35), increasing the probability for more specific selection of such low-frequency B cells remains a major challenge with this and many other available technologies. Future efforts may help improve specific selection by incorporating additional selection markers such as beads of multiple fluorophores (31), or B cell activation markers to single out BCR-activated cells more likely to be matured antibody-expressing clones (36,37). An alternative strategy may entail an additional imaging tool to verify selection of single antigen-reactive B cells subsequent to cell sorting (38). Furthermore, adjusting the cloning process to identify antibodies of different subclasses or specificities or from specific B cell subsets could improve the chance of detecting clones and clonal families in certain diseases, in which immunological conditions may promote specific antibody profiles (39)(40)(41)(42). Individually or combined, such approaches may help increase the chances of clonal selection.
In summary, we describe the establishment of a methodology to identify single antibody-expressing cells and to produce and test their sequenced recombinant antibodies in a workflow that may be readily applicable in any basic and translational immunology laboratory setting. This single cell-to-functional antibody strategy may open the way for new opportunities to analyze B cells and their antibody profiles at the single cell level and may be potentially applied to help unravel diverse humoral immune repertoires in blood and tissues and in different health and disease conditions.

eThics sTaTeMenT
Human immune cells were isolated from the venous blood of human volunteers. Specimens were collected with informed written consent in accordance with the Declaration of Helsinki.
aUThOr cOnTriBUTiOns SK, PK, KL, and FN conceived the study, and IC, KI, SC, PK, KL, FN, AT, and SK designed the methodology. IC, KI, PK, SC, SL, MF, and AC acquired data, generated materials, or helped with the data analysis and interpretation. SK, PK, KI, IC, and SC wrote the manuscript. IC, KI, SC, SL, MF, AC, JS, AT, FN, PK, KL, and SK discussed and interpreted the data and edited the manuscript. SK supervised the study, led and coordinated the project.

acKnOWleDgMenTs
We thank all volunteers who participated in this study. We acknowledge the Biomedical Research Centre Immune Monitoring Core Facility team at Guy's and St. Thomas' NHS Foundation Trust and the Nikon Imaging Centre at Kings College London for assistance.