key: cord-0817580-g5odwj7g
authors: Murji, Amyn A.; Raju, Nagarajan; Qin, Juliana S.; Kaldine, Haajira; Janowska, Katarzyna; Fechter, Emilee Friedman; Mapengo, Rutendo; Scheepers, Cathrine; Setliff, Ian; Acharya, Priyamvada; Morris, Lynn; Georgiev, Ivelin S.
title: Sequence and functional characterization of a public HIV-specific antibody clonotype
date: 2021-12-03
journal: iScience
DOI: 10.1016/j.isci.2021.103564
sha: 2eef2a560a0844edf074e51c4d3977317a5a1a73
doc_id: 817580
cord_uid: g5odwj7g

Public antibody clonotypes shared among multiple individuals have been identified for several pathogens. However, little is known about the determinants of antibody “publicness”. Here, we characterize the sequence and functional properties of antibodies from a public clonotype targeting the CD4 binding site on HIV-1 Env. Our results showed that HIV-1 specificity for the public antibodies studied here, comprising sequences from three individuals, was modulated by the V(H), but not V(L), germline gene. Non-native pairing of public heavy and light chains from different individuals suggested functional complementation of sequences within this public antibody clonotype. The strength of antigen recognition appeared to be dependent on the specific antibody light chain used, but not on other sequence features such as native-antibody or germline sequence identity. Understanding the determinants of antibody clonotype “publicness” can provide insights into the fundamental rules of host-pathogen interactions at the population level, with implications for clonotype-specific vaccine development.

Antibody discovery from HIV-infected individuals is a hallmark of HIV-1 research, paving the way toward the development of effective therapeutic and vaccine candidates (Bar et al., 2016; Lynch et al., 2015) . These discovery efforts have identified broadly neutralizing antibodies (bNAbs) as potential therapeutic candidates and antibodies as templates for engineering antigens to elicit epitope-specific antibody responses to vaccination (Bricault et al., 2019; Jardine et al., 2015; Xu et al., 2018) . Large-scale profiling of human antibody repertoires has shown that the antibody response to infection is vast and complex and, therefore, may contain unexplored avenues for vaccine design (Briney et al., 2019; Galson et al., 2015) . One currently under-explored area is vaccine design informed by population-level antibody responses (Davis et al., 2019; Kreer et al., 2020) . Although the majority of a person's antibody repertoire is unique because of the vast potential diversity generated in part by V (variable), D (diversity), J (joining) recombination, light chain selection, and somatic hypermutation (SHM) (Briney et al., 2019) , individuals can nevertheless possess identical or similar antibodies. Such ''public'' antibodies have been identified not only for various disease states including HIV-1 infection, SARS-CoV-2 infection, dengue infection, influenza vaccination, and others, but also in healthy individuals (Arentz et al., 2012; Ehrhardt et al., 2019; Jackson et al., 2014; Parameswaran et al., 2013; Setliff et al., 2018; Soto et al., 2019; Voss et al., 2021; Yuan et al., 2020) , though each study provides their own criterion for what will be defined as ''public.'' As there is no currently accepted consensus definition for a public clonotype, there exist opportunities to examine the variables that contribute to what may be considered ''public.''

To gain a better understanding of the properties of public antibodies, we focused on a CD4 receptor binding site (CD4bs)-targeting clonotype that had been previously identified in samples from multiple HIV-infected individuals from the Centre for the AIDS Programme of Research in South Africa (CAPRISA) cohort (Setliff et al., 2018) . In particular, this public clonotype included antibodies from three CAPRISA donors, with publicness defined by the same V H -gene, J H gene, and junction length, and CDRH3 amino acid sequences of high identity among donors (Setliff et al., 2018) . In that study, two antibody clonotype members with natively paired heavy and light chains from the public clonotype were produced experimentally and confirmed to be HIV-specific. Subsequent analysis of the antibody sequencing data revealed the existence of additional antibody sequences with high CDRH3 identity to the antibodies from the public clonotype but paired with different V H and/or V L genes. Therefore, here we sought to build on our previous work (Setliff et al., 2018) by investigating the genetic and phenotypic characteristics that define the members of this public antibody clonotype to include analyses on the importance of V-gene usage, CDR3 identity, and the relationship of sequence identity to native and germline sequences. The resulting analysis offers insight into the public antibody response in the context of chronic HIV-1 infection and explores the boundaries of antibody ''publicness.'' More broadly, an understanding of the role of shared elements may shed some light on the immunological role of public antibodies and their potential as templates for population-level vaccine design.

Antigen-specific sorting was performed to obtain sequences from three CAPRISA donors. Bulk sequencing was performed on donor CAP351, whereas paired heavy and light chain sequencing was performed on donors CAP314 and CAP248. The antibody sequences from the three CAPRISA donors were combined and complete linkage clustering was performed to assign clonotype membership for each sequence (Gupta et al., 2015) . In contrast to our previously published work (Setliff et al., 2018) , we expanded our parameters such that sequences were clustered using the following criteria: CDRH3 amino acid sequence identity of at least 70% with the same CDRH3 and junction length and no consideration for V H -and J H -gene usage. This allowed for a more inclusive definition of potential public antibodies that would enable systematic exploration of the boundaries of antibody publicness. Among the 24,218 clonotypes encompassing sequences from one or more of the three donors, clonotype #13905 contained the previously reported public heavy chain sequences (Setliff et al., 2018) and was selected for further analysis. Clonotype #13905 includes 171 nucleotide sequences from all three donors (donors CAP314, CAP248, and CAP351), spanning four different V H -gene assignments (IGHV1-69, IGHV5-51, IGHV1-18, and IGHV3-23). Because paired heavylight chain sequences were available for the datasets for donors CAP314 and CAP248, the corresponding light chain sequences were also retrieved. Two different light chain genes were identified in sequences from clonotype #13905: IGKV1-27 in both donors CAP314 (20% of light chain sequences from that donor) and CAP248 (100%), and IGKV3-20 in donor CAP314 (80% of light chain sequences from that donor).

To determine whether the CDR3 sequences from the same donor exhibited greater levels of similarity compared to sequences from other donors, we constructed Hamming distance matrices among all the unique CDRH3 and CDRL3 sequences, respectively, from clonotype #13905. The 171 sequences in the public clonotype comprised 25 unique CDRH3s, including 16 from donor CAP351, five from donor CAP314, three from donor CAP248, and one shared between donors CAP314 and CAP351. It also included six unique CDRL3s, with four from donor CAP314 and two from donor CAP248 ( Figure 1A ). The CDRH3 sequence distance values ranged from zero to four, with lower values corresponding to greater identity between two CDRH3 sequences ( Figure 1A ). High CDR3 similarity was observed both within donors (mean: 2.57, SEM: 0.08) as well as among donors (mean: 2.46, SEM: 0.06), with some CDR3s exhibiting greater similarity among, as opposed to within, donors ( Figure 1B ).

To investigate the patterns of pairing between CDR3 sequences and germline V genes, we constructed a Hamming distance-based network graph for the antibody sequences from clonotype #13905 (Figure 2 ). The majority (97%) of heavy chain sequences utilized IGHV1-69, but antibodies using three other heavy chain germline genes were also observed: IGHV1-18 (1.2%), IGHV3-23 (1.2%), and IGHV5-51 (0.6%). Of note, the antibodies from donors CAP314 and CAP248 only utilized IGHV1-69, whereas donor CAP351 utilized all four of the V H -genes. We observed two unique CDRH3 sequences from donor CAP351 that each utilized two different V H -genes: ARGADGDYYYYMAV (IGHV1-69 and IGHV5-51) and ARGADGDYRYYMDV (IGHV1-69 and IGHV1-18). These results suggest that the same CDRH3 sequence can be associated with multiple diverse V H -genes. With the exception of a single node in the heavy chain portion of the graph, all other antibody sequences from clonotype #13905 were within a Hamming distance of one from at least one other sequence in that clonotype, revealing a tight network of sequence similarity among members of 12 These authors contributed equally 13 

The observation that multiple diverse V H and V L germline genes can be associated with highly similar CDR3 sequences led us to address the limits of what constitutes a public antibody clonotype. To assess this question, we selected a diverse set of antibody heavy and light chain sequences for experimental validation, making sure sequences from all three donors and all observed V H and V L genes were selected. The set included nine heavy chain sequences (six from CAP351, two from CAP314, and one from CAP248) and three light chain sequences (two from CAP314 and one from CAP248). Of these, five, two, one and one sequences were from IGHV1-69, IGHV3-23, IGHV1-18, and IGHV5-51, respectively; and two and one were from IGKV1-27 and IGKV3-20, respectively. Among the nine heavy chain sequences, three pairs utilized identical CDRH3s, and sequences in two of these pairs (CDRH3 sequences ARGADGDYRYYMDV and ARGADGDYYYYMAV) utilized different V H -genes (IGHV1-18 and IGHV1-69; IGHV5-51 and IGHV1-69, respectively), whereas the third pair (CDRH3 sequence ARGADGDYYYYMDV) used the same V H -gene but was found in two different donors (Table S1 ). Thus, the selected set was representative of the diversity of sequences found in clonotype #13905. Each node represents a unique CDR3 and the node diameter is proportional to the corresponding number of sequences, which varied from 1 to 72 for heavy chain and 1 to 7 for light chain. Nodes are connected by an edge if their Hamming distance is 1, irrespective of edge length. Node colors correspond to the respective donor for the given CDR3 sequence, with multiple colors in a single node representing a CDR3 that is shared by multiple donors. The node fill pattern corresponds to the V-genes used for the given set of sequences, with multiple patterns in a single node representing different V-genes associated with the same CDR3 sequence.

4 iScience 25, 103564, January 21, 2022 iScience Article All natively paired heavy-light chain sequences, as well as all other non-native heavy-light chain pairs from all three donors (a total of 27 unique antibody heavy-light chain pairs) were successfully expressed as recombinant IgG proteins. Thus, we sought to determine whether and to what extent these antibodies could recognize HIV-1 Env-derived antigens. Two of these antibodies, CAP248_30 and CAP314_30 (native heavylight chain pairs from donors CAP248 and CAP314), had been previously validated (Setliff et al., 2018) and were also confirmed to be HIV-specific in our experiments (Figures 3 and S1 ). We tested all 27 heavy-light chain pairs against two different HIV Env-derived antigens, clade CRF01_AE 93TH975 gp120 monomer (Figures 3A and S1A) and clade A BG505.SOSIP.664 prefusion-stabilized gp140 trimer (Sanders et al., 2013) ( Figures 3A and S1B ). Only antibodies with heavy chains utilizing IGHV1-69 were able to bind the gp120 monomer ( Figures 3A and S1A ), suggesting the importance of V H -gene usage for antigen recognition. Interestingly, the CAP351_04 (IGHV1-69) and CAP351_01 (IGHV1-18) heavy chains had the same CDRH3 sequence, but different V H -genes; however, only CAP351_04 (IGHV1-69) bound 93TH975 gp120.

In contrast, all IGHV1-69 heavy chains paired with light chains utilizing any of the three V L genes were able to the bind gp120 monomer, albeit to different extents ( Figures 3A and S1A ). Together, these results suggest potential V H , but not V L , germline gene-mediated antigen specificity for this public clonotype.

The strength of binding to the monomer appeared to be associated with the choice of light chain pairing with a given IGHV1-69 heavy chain (p < 0.0001, two-way ANOVA with p value corrected for multiple comparisons using Tukey's multiple comparisons test), suggesting that even though there appears to be greater promiscuity for the choice of light chain compared to heavy chain, strength of antigen reactivity can be modulated by optimizing heavy chain-light chain pairings. We next determined if the antibodies from this public clonotype recognized gp140 trimer, which is designed to mimic a neutralization-sensitive perfusion conformation of Env (Sanders et al., 2013) (Figures 3A and S1B). For most antibodies, there was markedly increased binding to BG505.T332N.SOSIP.664 single-chain trimer , though the antibody-antigen binding patterns between monomer and trimer showed a significant correlation (p < 0.0001, Spearman correlation) ( Figure 3B ). Binding to different forms of the BG505 trimer for the native CAP248_30 and the non-native CAP248_30 H /CAP314_30 L antibody pairs was further validated by surface plasmon resonance ( Figure S2 ).

Overall, the finding that only antibodies using IGHV1-69 heavy chains were capable of recognizing HIV-1 antigens, despite also testing sequences with highly similar or even identical CDR3 sequences but different V H gene usage, indicated that this specific public antibody clonotype may be restricted to only sequences with IGHV1-69 but with either IGKV1-27 or IGKV3-20. This conclusion was further reinforced by the fact that non-native pairs of heavy and light chains from either the same or different donors could successfully recognize HIV-1 Env, indicating functional complementation of antibody sequences from the public clonotype.

We previously determined that binding to HIV-1 Env by CAP248_30 and CAP314_30 was affected by the CD4 receptor binding site (CD4bs) epitope knockout, D368R (Setliff et al., 2018) . To confirm whether other members of the public clonotype also mapped to CD4bs, we generated epitope knockouts in the context of the BG505.T332N.SOSIP.664 trimer and confirmed that all tested heavy-light chain pairs were indeed affected by CD4bs knockout mutations (D279K and D368R) ( Figure S3 ). These results suggest that nonnative pairings of heavy and light chains from the same or different donors did not affect epitope targeting, further corroborating the conclusion that these antibody sequences which were shared among all three donors, are indeed public. Further, in pseudovirus neutralization assays, both native and non-native heavylight chain pairs exhibited generally consistent ability to neutralize tier 1 HIV-1 strains ( Figure S4 ), in agreement with our previously published work (Setliff et al., 2018) for the native CAP248_30 and CAP314_30 antibodies.

To visualize the overall sequence similarity among antibodies from different donors, we generated a phylogenetic tree with the sequences from this public antibody clonotype ( Figure 4A ). Varied levels of somatic hypermutation were observed both in the heavy chains (2.64%-11.16%) and the corresponding light chains (1.4%-4.9% for IGKV1-27 and 1.04%-5.58% for IGKV3-20 iScience Article particularly in the heavy chain tree, suggesting a diversity of antibody evolution within this clonotype (Figure 4A ). Of note, however, sequences from different donors were interspersed in the trees. This suggests that in some cases, greater similarity was observed among sequences from different donors, as opposed to sequences from the same donor. This observation of among versus within donor similarities was further supported by the V-gene hamming distances for these sequences ( Figure 4B ), with distances ranging between 12 and 22 among donors and 1-19, 2-17, and 6-24 within donors CAP351, CAP248, and CAP314, respectively. Together, these results suggest a diversity of pathways in which antibodies from this clonotype evolved within each donor, although inter-donor similarities were also observed.

A B Figure 3 . Public antibody recognition of HIV-1 protein (A) Binding data for each antibody (both native and non-native heavy-light chain pairs) against 93TH975 gp120 or BG505.T332N.SOSIP.664 are displayed as a heatmap of AUC analysis calculated from the ELISA curves in Figure S1A and S1B, respectively. Statistical significance was determined via two-way ANOVA p value (p < 0.0001) corrected for multiple comparisons using Tukey's multiple comparisons test for both monomer and trimer. (B) Spearman correlation between antibody binding to HIV-1 monomer (x axis) and trimer (y axis). To better understand the types of somatic hypermutation changes that are characteristic of these specific public antibodies, we analyzed the per-residue frequency of mutations from germline ( Figure S5A ). Although in each donor, a large number of residue positions retained their germline identity in the majority of sequences, a number of residue positions had high frequency of mutations compared to germline ( Figure S5A ). Of note, several of these residue positions overlapped in multiple donors, with the levels of somatic hypermutation and residue entropy at each position having a significant correlation between all three donors (Spearman correlation test with p value correction for multiple comparisons using 

Benjamini-Hochberg method) (Benjamini and Hochberg, 1995) . This finding suggests the existence of somatic hypermutation hotspots that are common to all three donors ( Figures S5A and S5B ).

To interrogate whether the somatic hypermutation hotspots in the IGHV1-69 antibodies are characteristic of this public antibody clonotype, we compared them to a set of unrelated representative antibody sequences that also utilized the IGHV1-69 germline gene, retrieved from cAb-Rep (Curated Antibody Repertories) (Guo et al., 2019) ( Figures 5A and S6A) . Although for the majority of residue positions, no major iScience Article differences among these public antibody sequences and the reference dataset were observed, there were also residue positions with notably higher frequencies of non-germline amino acid identities in the public antibody sequences that were also identified when comparing to non-public IGHV1-69 sequences from the three CAPRISA donors ( Figures 5B and S6A) , suggesting the existence of unique IGHV1-69 mutations that are specific to this public antibody clonotype.

Heavy chain identity to either germline or native antibody sequence does not modulate Env trimer recognition

We next examined additional factors that could affect antigen recognition by the different antibody variants from the public clonotype ( Figure 3 ). We first explored the potential role of heavy chain identity to germline sequence. Notably, antibodies at both ends of the V H germline identity scale showed lower levels of HIV-1 protein binding, whereas some of the strongest binding was observed for antibodies with intermediate V H germline identity ( Figure 6A ). We then explored whether antigen binding could be dependent on the sequence identity of the heavy chain in a given antibody pairing to the heavy chain present in the native pair for the given light chain ( Figure 6B ). For CAP248_30 L , the native heavy-light chain pairing was optimal in terms of binding to both gp120 monomer and to the stabilized trimer; however, high identity to native for the non-native heavy chains was not necessarily associated with improved antigen recognition ( Figure 6B ). Further, the other two light chains, CAP314_30 L and CAP314_21 L , exhibited better antigen recognition when paired with non-native, compared to their native, heavy chains ( Figure 6B ). Together, these data indicate that heavy chain identity to germline or native sequence may not be strong determinants of antigen recognition by this public antibody clonotype. ). Yet, our understanding of what constitutes antibody publicness has been limited to date. At one extreme, stringent definitions that restrict public clonotypes to only identical antibody sequences have been employed (Soto et al., 2019) . Such a stringent approach aims to guarantee complete confidence in the identification of truly public antibodies. However, this approach fails to account for the diversification potential of antibodies undergoing evolution in response to antigen exposure. For example, antibodies from an individual clone from a single donor have been identified with up to 50% or more divergence in CDRH3 sequence (Wu et al., 2015) . It is therefore important to understand what levels of sequence-based similarity can be reasonably used for defining public clonotypes among antibodies from multiple individuals.

To that end, here we characterized a previously identified public clonotype found in a cohort of HIV-infected donors by expanding selection criteria and performing phenotypic and genotypic analyses to more carefully define the limits of antibody publicness for this clonotype. To achieve this, we explored the choice of V H genes, the choice of V L genes, and a range of CDR3 identities for sequences in the putative clonotype. We also explored the potential of heavy and light chains from different antibodies and different donors to produce functional antibodies with unaltered antigen specificity to show functional complementation between non-native heavy-light chain pairs and further support the phenotypic similarities among different antibodies in this public clonotype. Of course, as expected, the specific choice of heavy and light chain pair appeared to be important for strength of antigen recognition.

In this article, we report on sequence and functional features of antibodies shared between three donors in the CAPRISA cohort. We observed that the heavy chain gene is responsible for antigen recognition and that recognition is modulated by the light chain but not in other sequence features including heavy chain or native sequence identity. However, there are a number of limitations to this study. The results reported here are specific to the public antibody sequences identified in clonotype #13905. Structural details about the atomic-level interactions of this public antibody clonotype with HIV-1 Env would be useful in putting into context the variation observed in this clonotype. A high-resolution antibody-antigen complex structure can reveal clonotype-specific mutations within the heavy and light chains that participate in antigen recognition and would therefore be a good future direction of this study. Furthermore, the antibody sequence criteria that we initially used to select for members of this public clonotype were very lenient. As such, it would be worthwhile to compare these selection criteria on known antibody-antigen complexes in the Protein DataBank which can reveal the biological relevance of using ll OPEN ACCESS iScience 25, 103564, January 21, 2022 9

iScience Article more relaxed constraints to identify antibodies shared among individuals. In addition, our conclusions on the determinants of antigen recognition were predicated on a limited number of natively paired sequences. Although our analyses on the available native-paired sequences yielded no discernible reliance on native-pair identity, these conclusions would be better informed with greater numbers of paired sequences. Furthermore, because this study is limited in the number of paired light chain sequences, our analyses of rates of somatic hypermutation were made solely based on the heavy chain. A clearer understanding of rates of somatic hypermutation on antigen recognition would benefit from an analysis of SHM that also included the light chain.

Although public antibody clonotypes have been identified in a variety of disease-state contexts, it remains to be seen how common this public clonotype is outside of the CAPRISA cohort from which clonotype iScience Article #13905 was discovered. Interestingly, a small number of antibodies from the cAb-Rep dataset were found to have CDRH3 sequence identity of greater than 70% to at least one sequence from the public clonotype studied here ( Figure S6B ). The overall prevalence and HIV-1 antigen recognition specificity of these public antibody sequences will be an interesting area of future research.

Whether our findings for the specific clonotype studied here will be generalizable for other public antibodies and in contexts beyond HIV-1 infection will be of significant interest in the antibody field. Additional investigation of public clonotypes may shed light on specific, population-level responses to infection and vaccination. A comprehensive assessment of the publicness of antibody repertoires has the potential to make significant contributions to vaccine development for difficult targets such as HIV-1, influenza, and other diseases.

In this article, we report on sequence and functional features of antibodies with high sequence similarity from three donors in the CAPRISA cohort. Although the results reported here are based on one specific public clonotype and may not be necessarily generalizable to all public clonotypes, these findings represent an important step toward understanding the determinants of antibody publicness, and highlight the significance of assessing multiple variables for defining antibody publicness. We note that the conclusions from this study related to the determinants of antigen recognition are based on a small number of natively paired antibody sequences and light chain sequences in the public clonotype, which represent a limitation of our study. Addressing these limitations may play an important role for gaining a better understanding of the role of factors such as somatic hypermutation and heavy/light chain sequence diversity play in defining the functional phenotypes of public antibody clonotypes like the one studied here. Although public antibody clonotypes have been identified in a variety of disease-state contexts, it remains to be seen how prevalent public clonotypes are in the context of HIV-1, including outside of the CAPRISA cohort from which clonotype #13905 was discovered. Structural details about the atomic-level interactions of this public antibody clonotype with HIV-1 Env can help further our understanding of how the public antibody sequence features are associated with antigen recognition, including a better understanding of the light chain variation observed in this clonotype. Furthermore, in future studies, it will be interesting to explore the potential for systematic selection of the antibody sequence criteria for defining public clonotypes. For example, it would be worthwhile to apply such selection criteria on known antibody-antigen complexes in the Protein DataBank or other functional antibody databases, with the goal of assessing the biological relevance of using different levels of sequence identity cutoffs to identify antibodies shared among individuals.

Detailed methods are provided in the online version of this paper and include the following: (Shingai et al., 2013) ; Anti-HIV-1 gp120 Monoclonal (VRC01), from Dr. John Mascola (cat# 12,033) (Wu et al., 2010) .

We used preprocessed B-cell repertoire data of donors CAP351, CAP314 and CAP248 from our previously published sequencing data which is available for public access under BioProject PRJNA415492 (Setliff et al., 2018) . For donor CAP351, heavy chain variable gene sequences were available from three time points (pre-infection, six months post-infection and three years post-infection) whereas for donors CAP314 and CAP248, paired heavy-light chain sequences were available for a single time point. Preprocessed sequences were annotated using IgBLAST (Ye et al., 2013) to assign gene information compared to the germline repertoire obtained from IMGT (Lefranc et al., 2015) . To find the public clonotype, irrespective of V-gene and J-gene identity, we combined data from all three donors and performed clonal clustering using Change-O (Gupta et al., 2015) with the following criteria: complete linkage, same CDRH3 length and 70% CDRH3 amino acid sequence identity.

For each of the heavy chain and light chain sequence sets, sequences were grouped based on the respective V-genes. Hamming distances were computed between all the pairs of unique CDR3s. Further, sequences were grouped based on CDR3, and a network graph was generated using each unique CDR3 as a node; an edge between two nodes was shown only for Hamming distance value of one. A multiple sequence alignment with phylogenetic analysis was performed using Clustal Omega (Madeira et al., 2019) by also including the respective germline V-gene sequences. Phylogenetic trees were annotated and visualized using iTol (Letunic and Bork, 2019); for better visualization, sequences were clustered based on 98% sequence identity and one (original and/or consensus) or more sequences (if they came from different donors) from each cluster was included.

Residue-wise analysis was performed by computing somatic hypermutation (SHM) and entropy values. SHM was computed using in-house scripts that compared each sequence in the public clonotype to the respective germline sequence. This revealed the frequency of mutation at each residue position, which can vary from 0 (non-mutated) to 1 (always mutated). Entropy values were calculated by using a log-based formula via Bioedit (Hall, 1999) . The entropy of a residue position represents the frequency of different amino acids at that position. Entropy values at each residue position varied from 0 (only one type of amino acid at that position) to 4.322 (different frequencies of multiple amino acids at that position).

Sequences in the public clonotype were also compared to sequences in the cAb-Rep database (Guo et al., 2019) . cAb-Rep is a curated database containing, at the time of analysis, antibody sequences from 306 B cell repertoires from 121 donors. Among the 267.9 million heavy chain sequences from cAb-Rep, sequences with gene IGHV1-69 were extracted, and sequences with less than 98% identity and SHM >10% were included in the ''cAb dataset'' (30422 sequences). Donor-wise amino acid frequencies were computed for the sequences in the public clonotype the cAb dataset. The ratio of the per-residue amino acid frequencies was calculated with respect to the cAb dataset. The sequence logo plot to visualize residuewise mutations and amino acid diversity was generated by using Weblogo (Crooks et al., 2004) .

BG505 gp140 SOSIP variants were expressed as recombinant soluble antigens. The single-chain variants iScience Article membrane Nalgene Rapid Flow Disposable Filter Units and then run slowly over an affinity column of agarose bound Galanthus nivalis lectin (Vector Laboratories cat no. AL-1243-5) at 4 C. The column was washed with PBS, and proteins were eluted with 30 mL of 1 M methyl-a-D-mannopyranoside. The protein elution was buffer exchanged three times into PBS and concentrated using 30 kDa Amicon Ultra centrifugal filter units. Concentrated protein was run on a Superose 6 Increase 10/300 GL or Superdex 200 Increase 10/300 GL sizing column on the AKTA FPLC system, and fractions were collected on an F9-R fraction collector. Fractions corresponding to correctly folded antigen were selected, and antigenicity by ELISA was characterized with known monoclonal antibodies specific for that antigen. Proteins were stored at À80 C until use.

For producing BG505 DS-SOSIP Env (Do Kwon et al., 2015) expressing high mannose glycans, we produced the Env in GnT1-cells. Env was affinity purified using a PGT145 IgG affinity column (De Taeye et al., 2015) , followed by a Superdex 6 Increase 10/300 column.

For each antibody, variable genes were inserted into plasmids encoding the constant region for the heavy chain (pFUSEss-CHIg-hG1, Invivogen) and light chain (pFUSE2ss-CLIg-hl2, Invivogen and pFUSE2ss-CLIghk, Invivogen) and synthesized from GenScript. mAbs were expressed in FreeStyle 293F or Expi293F mammalian cells (ThermoFisher) by co-transfecting heavy chain and light chain expressing plasmids using polyethylenimine (PEI) transfection reagent and cultured for 5-7 days. FreeStyle 293F (ThermoFisher) and Expi293F (ThermoFisher) cells were maintained in FreeStyle 293F medium or FreeStyle F17 expression medium supplemented with 1% of 10% pluronic F-68 and 20% of 200 mM L-Glutamine. These cells were cultured at 37 C with 8% CO 2 saturation and shaking. After transfection and 5-7 days of culture, cell cultures were centrifuged at 6000 rpm for 20 minutes. Supernatant was 0.45 mm filtered with PES membrane Nalgene Rapid Flow Disposable Filter Units. Filtered supernatant was run over a column containing Protein A agarose resin that had been equilibrated with PBS. The column was washed with PBS, and then antibodies were eluted with 100 mM Glycine HCl at pH 2.7 directly into a 1:10 volume of 1 M Tris-HCl pH 8. Eluted antibodies were buffer exchanged into PBS three times using 10 kDa or 30 kDa Amicon Ultra centrifugal filter units.

For gp120 ELISAs, soluble 93TH975 (Aids Reagent Program) protein was plated at 2 mg/mL overnight at 4 C. The next day, plates were washed three times with PBS supplemented with 0.05% Tween20 (PBS-T) and coated with 5% milk powder in PBS-T. Plates were incubated for one hour at room temperature and then washed three times with PBS-T. Primary antibodies were diluted in 1% milk in PBS-T, starting at 10 mg/mL with a serial 1:5 dilution and then added to the plate. The plates were incubated at room temperature for one hour and then washed three times in PBS-T. The secondary antibody, goat anti-human IgG conjugated to peroxidase, was added at 1:10,000 dilution in 1% milk in PBS-T to the plates, which were incubated for one hour at room temperature. Plates were washed three times with PBS-T and then developed by adding TMB substrate to each well. The plates were incubated at room temperature for 10 minutes, and then 1 N sulfuric acid was added to stop the reaction. Plates were read at 450 nm.

For recombinant single-chain SOSIP trimer ELISAs, 2 mg/mL of recombinant trimer proteins diluted in PBS-T were added to the plate and incubated overnight at 4 C. Primary and secondary antibodies, along with substrate and sulfuric acid, were added as described above.

Areas under the ELISA binding curves (AUC) were determined with GraphPad Prism 8.0.0.

The binding of native CAP248_30 to different concentrations of BG505.DS.SOSIP/GnTI-was assessed by surface plasmon resonance on Biacore T-200 (Cytiva) at 25 C with HBS-EP+ (10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA, and 0.05% surfactant P-20) as the running buffer. 200 nM of CAP248_30 IgG was captured on flow cell of immobilized human Anti-Fc chip (9,000 RU) and it was assayed by flowing over 50, 100, 200, 400 nM of BG505.DS.SOSIP/GnTI-in running buffer. The surface was regenerated between injections by flowing over 3M MgCl2 solution for 10 s with flow rate of 100 mL/min. Blank sensorgrams were obtained by injection of the same volume of HBS-EP+ buffer in place of BG505.DS.SOSIP/GnTI-solutions. Sensorgrams of the concentration series were corrected with corresponding blank curves.

iScience 25, 103564, January 21, 2022

Secreted human Ro52 autoantibody proteomes express a restricted set of public clonotypes

Effect of HIV antibody VRC01 on viral rebound after treatment interruption

Controlling the false discovery rate -a practical and powerful approach to multiple testing

HIV-1 neutralizing antibody signatures and application to epitope-targeted vaccine design

Commonality despite exceptional diversity in the baseline human antibody repertoire

WebLogo: a sequence logo generator

Recent progress in the analysis of ab T cell and B cell receptor repertoires

Polyclonal and convergent antibody response to Ebola virus vaccine rVSV-ZEBOV

In-depth assessment of withinindividual and inter-individual variation in the B cell receptor repertoire

Single-chain soluble BG505.SOSIP gp140 trimers as structural and antigenic mimics of mature closed HIV-1 Env

A database of curated antibody repertoires for exploring antibody diversity and predicting antibody prevalence

Change-O: a toolkit for analyzing largescale B cell immunoglobulin repertoire sequencing data

BioEdit: A User-Friendly Biological Sequence Alignment Editor and Analysis Program for Windows 95/98/NT

Human responses to influenza vaccination show seroconversion signatures and convergent antibody rearrangements

HIV-1 VACCINES. Priming a broadly neutralizing antibody response to HIV-1 using a germlinetargeting immunogen

Exploiting B cell receptor analyses to inform on HIV-1 vaccination strategies

Crystal structure, conformational fixation and entry-related interactions of mature ligand-free HIV-1 Env

IMGT, the international ImMunoGeneTics information system 25 years on

A nextgeneration cleaved, soluble HIV-1 Env trimer, BG505 SOSIP.664 gp140, expresses multiple epitopes for broadly neutralizing but not nonneutralizing antibodies

Optimization and validation of the TZM-bl assay for standardized assessments of neutralizing antibodies against HIV-1

Multi-donor longitudinal antibody repertoire sequencing reveals the existence of public antibody clonotypes in HIV-1 infection

Antibody-mediated immunotherapy of macaques chronically infected with SHIV suppresses viraemia

High frequency of shared clonotypes in human B cell receptor repertoires

Immunogenicity of stabilized HIV-1 envelope trimers with reduced exposure of nonneutralizing epitopes

Prevalent, protective, and convergent IgG recognition of SARS-CoV-2 non-RBD spike epitopes

Rational design of envelope identifies broadly neutralizing human monoclonal antibodies to HIV-1

et al.; NISC Comparative Sequencing Pro-gram. Maturation and diversity of the VRC01-antibody lineage over 15years of chronic HIV-1 infection

Epitope-based vaccine design yields fusion peptide-directed antibodies that neutralize diverse strains of HIV-1

IgBLAST: an immunoglobulin variable domain sequence analysis tool

Structural basis of a shared antibody response to SARS-CoV-2

Resource and reagent requests should be directed to the corresponding author.

Custom scripts used to analyze data in this manuscript are available upon request to the corresponding author. 

Antibody neutralization was assessed using the TZM-bl assay as described (Montefiori, 2004; Sarzotti-Kelsoe et al., 2014) . This standardized assay measures antibody-mediated inhibition of infection of JC53BL-13 cells (also known as TZM-bl cells) as a reduction in luciferase gene expression after a single round of infection by molecularly cloned Env-pseudoviruses. Murine leukemia virus (MLV) was included as an HIV-specificity control and VRC01 was used as a positive control. 

Spearman correlation tests were performed using cor.test function in R to obtain p and r values to determine the statistical significance of the differences. For the comparison of multiple independent tests, the p values were adjusted using the Benjamini-Hochberg method (Benjamini and Hochberg, 1995) using the p.adjust function in R. GraphPad Prism 8.0.0 was used to calculate 2 way ANOVAs as well as p-values adjusted for multiple comparisons via Tukey's multiple comparison test.