Affinity gaps among B cells in germinal centers drive the selection of MPER precursors

Current prophylactic human immunodeficiency virus 1 (HIV-1) vaccine research aims to elicit broadly neutralizing antibodies (bnAbs). Membrane-proximal external region (MPER)-targeting bnAbs, such as 10E8, provide exceptionally broad neutralization, but some are autoreactive. Here, we generated humanized B cell antigen receptor knock-in mouse models to test whether a series of germline-targeting immunogens could drive MPER-specific precursors toward bnAbs. We found that recruitment of 10E8 precursors to germinal centers (GCs) required a minimum affinity for germline-targeting immunogens, but the GC residency of MPER precursors was brief due to displacement by higher-affinity endogenous B cell competitors. Higher-affinity germline-targeting immunogens extended the GC residency of MPER precursors, but robust long-term GC residency and maturation were only observed for MPER-HuGL18, an MPER precursor clonotype able to close the affinity gap with endogenous B cell competitors in the GC. Thus, germline-targeting immunogens could induce MPER-targeting antibodies, and B cell residency in the GC may be regulated by a precursor–competitor affinity gap.

The frequency of precursor B cells for modest-and low-affinity immunogens is reported to affect B cell recruitment to GCs 29 .To investigate whether precursor frequency influenced 10E8-UCA H B cell recruitment to GCs, we adoptively transferred 10E8-UCA H CD45.2 + B cells into CD45.1 wild-type mice to establish 10E8-UCA H B cell frequencies of 2:10 6 , 8:10 6 , 4:10 5 and 4:10 4 (Extended Data Fig. 3b) 1 day before immunization with 10E8-GT9.212mer or 10E8-GT9-KO 12mer.On day 7 p.i., some 10E8-UCA H CD45.2 + GC B cells were detected in mice with initial 4:10 5 and 4:10 4 10E8-UCA H B cell frequencies, but not mice with 2:10 6 and 8:10 6 initial frequencies (Extended Data Fig. 3c,d).In the 10E8-GT9.212mer-immunized cohort, we also detected a high proportion (~20%) autoantigens and/or polyreactivity and to their blockade by immune checkpoints in bnAb knock-in mouse models [15][16][17][18] .The bnAb 10E8, which binds to the distal MPER, has less poly-and autoreactivity 2,12,19 and appreciable neutralization breadth 12,20 .An unmutated common ancestor (UCA) of 10E8 proposed based on next-generation sequencing (NGS) data combined with structure-function studies lacks affinity for MPER and does not neutralize even the most sensitive HIV-1 isolates 21 .To gain recognition of Env in its membrane context, extensive SHM, particularly in HCDR2 and HCDR3, would be required 21,22 .One strategy to overcome the lack of affinity of 10E8-like precursor antibodies for native Env, referred to as germline targeting (GT), relies on priming with immunogens designed for increased affinity to diverse precursors, followed by sequential boosting with increasingly native-like immunogens to shepherd SHM toward the mature bnAb sequence.The hydrophobic MPER is excluded from most soluble Env immunogens to prevent aggregation 14,23,24 .A series of scaffolds displaying engineered MPER epitopes, named 10E8-GT and displayed on multivalent nanoparticles for improved immunogenicity, can bind 10E8-like precursors in vitro 25 .
Here, we generated two mouse models expressing the heavy chain sequence 10E8-UCA 21 or 10E8-NGS-04 (a 10E8-like heavy chain identified by sequencing naive B cells from healthy human donors) 25,26 in functional BCRs with mouse light chains.We also used a pre-existing line expressing a heavy chain sequenced from genuine human B cell precursors binding to 10E8-GT9 immunogens, referred to as MPER-HuGL18 ref. 25.We found that B cells in the knock-in mice developed normally and underwent affinity maturation in GCs after immunization with 10E8-GT immunogens 25 and identified a dynamic affinity gap between precursor B cells and endogenous competitors driving GC kinetics.Furthermore, variations among B cell lineages in BCR affinity maturation capacity were of substantial importance for maintenance in GCs.
IGHV sequences from 10E8-UCA H GT10 ++ KO − GC B cells sorted at days 7, 14 and 21 p.i. with 10E8-GT10.312mer from CD45.1 wild-type mice transferred with 1:10 4 CD45.210E8-UCA H B cells showed ongoing SHM and diversification (Extended Data Fig. 9e,f).BCRs from epitope-specific CD45.2 + GT10 ++ KO − GC B cells acquired a median of less than two amino acid (maximum of six amino acid) mutations by day 7 p.i., which increased to a median of two amino acid mutations (maximum of six) at day 14 p.i. and a median of two amino acids (seven maximum) at day 21 p.i. (Extended Data Fig. 9e).However, there was no correlation (R 2 = 0.01, P > 0.05) between affinity and the number of amino acid mutations in the heavy chain of CD45.2 10E8-UCA H B cells on day 21 p.i. with 10E8-GT10.3(Extended Data Fig. 9g).Furthermore, phylogenetic analysis of sequences from day 21 p.i. with 10E8-GT10.3suggested that lower-affinity CD45.2 10E8-UCA H  Article https://doi.org/10.1038/s41590-024-01844-7cells (Fig. 6g), which might have lost monovalent affinity through detrimental mutation(s) 41 .Several mouse light chains were commonly found paired with the heavy chains of 10E8-UCA H CD45.2 + GT10 ++ KO − GC B cells isolated at day 21 p.i., and there was substantial affinity variation within specific heavy chain-light chain sequence pairings (Extended Data Fig. 9h,i), indicating that SHM-generated diversity in BCRs might have caused occasional affinity loss or, alternatively, that low affinities at later time points might have resulted from ongoing naive B cell entry to GCs 42 .These observations indicated that immunization with 10E8-GT10.312mer enabled some 10E8-UCA H B cells to compete and mature in GCs up to day 21 p.i.
To follow the affinity maturation of MPER-HuGL18 H B cells, we sequenced BCRs from epitope-specific MPER-HuGL18 H CD45.2 + GT10 ++ KO − GC B cells and wild-type CD45.1 + GT10 ++ KO − GC B cells at days 14 and 21 p.i. with 10E8-GT10.312mer, expressed them as GT10.3-MPER-HuGL18H and GT10.3-WT mAbs and used SPR to determine their affinities to 10E8-GT10.2monomers.GT10.3-MPER-HuGL18H mAbs reached a median affinity of 20 nM on day 21 p.i., whereas the naive MPER-HuGL18 H mAbs, which consisted of both iGL and pretransfer (day −1) BCR sequences, had a median affinity of 161 nM (Fig. 7g).The median affinity of GT10.3-WT mAbs from day 21 p.i. (30 nM) also increased relative to naive wild-type mAbs (iGLs, ~8 µM); however, GT10.3-MPER-HuGL18H mAbs showed a better median affinity than GT10.3-WT at days 14 and 21 p.i. (Fig. 7g).MPER-HuGL18 H B cells using the prevalent mouse light chain IGKV1-117 had a high median affinity (~1.71 nM) but also encompassed a broad affinity range when examined on day 21 p.i. (Extended Data Fig. 10g), whereas heavy chain SHM within this specific pairing did generally increase with improved affinities, but the relationship was not significant (R 2 = 0.01, P > 0.05; Extended Data Fig. 10h).When we plotted affinity on a phylogenetic tree of 23 heavy chains paired with IGKV1-11 and other light chains, we observed that, while some highly mutated lineages were high in affinity, other highly mutated branches lost affinity (Extended Data Fig. 10i).These observations suggested that MPER-HuGL18 H B cells might differ from 10E8-UCA H B cells in their capacity to gain sufficient affinity to overcome endogenous competition and to remain in GCs.

MPER-HuGL18 H B cells are more competitive in GCs
To assess the importance of initial antigen affinity of the BCR for entry to the GC, we adoptively transferred CD45.2 10E8-NGS-04 H , CD45.2 10E8-UCA H

Discussion
Here, we showed that knock-in mice expressing heavy chain sequences derived from 10E8 precursor B cells isolated from individuals infected with HIV-1 (10E8-UCA H ) or healthy individuals (10E8-NGS-04 H ) and mice expressing a heavy chain sequenced from genuine human B cell precursors binding to 10E8-GT9 immunogens (MPER-HuGL18 H ) 25 had normal B cell development, in contrast to previously reported 2F5 H and 4E10 H knock-in mice [16][17][18] , and that B cells from these mice underwent activation in response to 10E8-GT immunogens.We furthermore identified off-target sites of immunogenicity through structural analysis, and we also found that the relative affinities of 10E8 precursor B cells and endogenous precursor B cells over time drive intra-GC competition.Endogenous competition, while observed in other models 26,[29][30][31][32]43,44 , was particularly critical to 10E8 immune responses in mice. We fond that the 10E8-GT9 immunogen variants with affinities ranging from ~1 to 3 µM provided limited activation of 10E8 B cell precursors, although immunogens of similar or lower affinities induce sustained GC responses in models for other classes of bnAbs 29,30,32 .Crystal structures of 10E8-GT10.2complexes showed that a wild-type mouse antibody (GT10.2-WT)isolated at day 14 p.i. bound a nearby epitope, inhibited the binding of GT10.2-10E8-UCAH mAbs in vitro and the activation of 10E8-UCA H B cells in vivo.Recruitment and maturation of B cells in GCs are controlled by competition and not absolute affinity 44,45 , and low-affinity B cells can remain in GCs in the absence of high-affinity competitors 46 .We suspect, therefore, that the increase in baseline affinity to 10E8-UCA between 10E8-GT10.2and 10E8-GT10.3 allowed ome maintenance of 10E8-UCA B cell precursors in the GC by narrowing the affinity gap with epitope-specific endogenous B cells during competition for antigen in the GC.However, reasonable starting affinities alone were insufficient to overcome endogenous competition and ensure retention in GCs.
A major goal of GT design is the activation of diverse bnAb precursors 10,26,47,48 .The responses of MPER-HuGL18 H B cells, which use a different IGH-V gene than 10E8-UCA H B cells, demonstrated that GC retention of B cells bearing 10E8-like HCDR3s after immunization with 10E8-GT10.312mers is possible.Higher initial recruitment of precursors to the GC did not sufficiently explain the difference in residence time between 10E8-UCA H and MPER-HuGL18 H B cells, as 10E8-UCA H B cells transferred at 25-fold higher frequencies and thus present in day 7 GCs at higher frequencies were still lost at day 14.The fraction of p.i. MPER-HuGL18 H Fabs achieving affinity gains compared to 10E8-UCA H B cell-derived Fabs suggests interclonal variation in the likelihood that SHM will produce affinity maturation, perhaps due to probabilistic variation in some types of mutations or the favorability of certain codons 49,50 , which may underpin the distinctive GC dynamics of 10E8-UCA H and MPER-HuGL18 H B cells.Whatever the underlying cause, the fact that MPER-HuGL18 H B cells were substantially favored in the GCs against 10E8-UCA H and 10E8-NGS-04 H after 10E8-GT10.2immunization highlighted the importance of GT immunogens engaging multiple B cell precursor lineages.
Affinity is only one factor in GC recruitment and maintenance.Delivery with alhydrogel enhanced epitope-specific 10E8-UCA H B cell recruitment, in line with reports that adjuvants strongly affect Env trimer immunogenicity 51 .Adjuvant activity is dependent on many factors, some of which (for example, Toll-like receptor activation [52][53][54] ) vary across species; nonetheless, this model confirmed the integrity of 10E8-GT epitopes in formulations with alhydrogel.High valency, effective for the eOD-GT8 60mer, which induces VRC01-bnAb-class responses in mice 55 and in clinical trials in humans 37 , did not improve 10E8-GT10.2.10E8-GT10.260mers induced higher serum binding titers than 10E8-GT10.212mers but lower epitope-specific 10E8-UCA H B cell recruitment to GCs, potentially because the increased differentiation into short-lived antibody-secreting cells that is associated with multimerization 55 limited GC residency.
.Further details on mouse strains as well as all reagents and tools described below can be found in Supplementary Table 3.

Generation of 10E8-UCA H -knock-in mice
10E8-UCA H -and 10E8-NGS-04 H -knock-in mice were generated following published protocols 27,32 .In brief, the 4E10 targeting vector 56 was modified by the incorporation of human rearranged 10E8-UCA or 10E8-NGS-04 V(D)J (heavy chain construct) sequences downstream of the promoter region and by elongation of the 5′ and 3′ homology regions using the Gibson assembly method (New England Biolabs).The targeting vector DNA was confirmed by Sanger sequencing (Eton Bioscience).Next, fertilized mouse oocytes were microinjected with a donor plasmid containing the prerearranged 10E8-UCA or 10E8-NGS-04 Igh with the mouse VHJ558 promoter, two pairs of single guide RNAs (25 ng ml -1 ) targeting the Igh locus and AltR-Cas9 protein (50 ng ml -1 ) and injection buffer 32 .Following culture, the resulting zygotes were implanted into the uteri of pseudopregnant surrogate C57BL/6J mothers.

Immunizations
Both male and female donor mice were used for adoptive transfers.All transfers and immunizations were performed in male CD45.1 +/+ mice due to risk of rejection of cells from male donor mice to female recipient mice.Preparations of immunogens (10E8-GT9 12mer and 10E8-GT10 12mer nanoparticles at 5 µg per mouse (or as specified in the text)) were diluted in PBS at a volume of 100 µl per mouse for i.p. injection and then mixed at a 1:1 ratio with 2% alhydrogel (Invitrogen) or Sigma adjuvant system (Ribi) for at least 20 min.The final formulation was injected i.p. (total volume of 200 µl per mouse).

Immunogen and flow cytometry probe production
All proteins were expressed in Freestyle 293F cells as previously described 25,48 .In brief, nanoparticles were purified by Galanthus Nivalis lectin affinity purification, followed by size-exclusion chromatography.Antibodies and Fabs were purified by protein A affinity chromatography and CaptureSelect CH1-XL affinity chromatography, respectively, and buffer exchanged into TBS.Monomeric proteins contained C-terminal His tags and were purified by immobilized metal affinity chromatography, followed by size-exclusion chromatography.
For flow cytometric probe binding, Avi-tagged 10E8-GT9 or 10E8-GT10 was biotinylated by BirA enzymatic reaction (Avidity) according to the manufacturer's protocol.Biotinylated 10E8-GT9 or 10E8-GT10 probes and respective knockouts were prereacted in independent tubes for at least 30 min in a 4:1 molar ratio with fluorescently labeled streptavidin (streptavidin-AF488 and/or streptavidin-AF647).Reagents were then combined with fluorescently labeled antibodies for fluorescence-activated cell sorting (FACS) staining.

Enzyme-linked immunosorbent assay
Antigen-specific antibody titers were detected by enzyme-linked immunosorbent assay using anti-His (2 mg ml -1 ) to capture GT10 or GT10-KO antigen (2 mg ml

BCR probe binding and dissociation measurement by flow cytometry
Peripheral blood mononuclear cells containing naive B cells from 10E8-UCA H and 10E8-NGS-04 H mice were surface stained using 10E8-GT10.2probe conjugated with streptavidin-BV421.After final washing, stained cells were incubated at room temperature for the indicated lengths of time to allow dissociation before being subjected to FACS analysis.

10x gene expression analysis
The Seurat R package (v4.3.0) was used to perform 10x gene expression analysis 57 .Droplets containing more than 500 unique molecular identifiers, more than 300 unique genes and a mitochondrial ratio of less than 20% were considered of poor quality and were filtered out before subsequent analysis (log 10 GenesPerUMI > 0.80 and housekeeping genes > 50 were also applied).The top 2,000 highly variable genes were selected using the 'FindVariableFeatures' function on the normalized dataset before scaling and principal-component analysis.

Fig. 1 |
Fig. 1 | 10E8 knock-in mice express heavy chains from human 10E8 precursor B cells.a, Representative flow cytometry plots (left) and quantification (right) of spleen cells from 8-to 10-week-old wild-type, 10E8-NGS-04 H and 10E8-UCA H mice and gating strategy for the quantification of splenic B220 + TCRβ − B cells and TCRβ + B220 − T cells.Data are pooled from two independent experiments (n = 3-4 mice per independent group).b, Representative flow cytometry plots (left) and quantification (right) of spleen cells from 8-to 10-week-old wild-type, 10E8-NGS-04 H and 10E8-UCA H mice and gating strategy for the quantification of CD21 hi CD24 lo follicular B cells, transitional CD21 lo CD24 hi T0/ T1 B cells, CD21 hi CD24 hi CD23 + T2 B cells and CD21 hi CD24 hi CD23 − marginal zone B cells.Data are pooled from two independent experiments (n = 3-4 mice per independent group).Error bars indicate mean ± s.d.from mice in pooled groups.c, Representative flow cytometry (left) and quantification (right) of B220 + CD4 − CD8 − F4/80 − Gr-1 − naive B cells from peripheral blood binding biotinylated 10E8-GT9 in wild-type or 10E8-NGS-04 H mice. Data are pooled from two independent experiments.Error bars indicate mean ± s.d.from mice in each group (n = 3-6).d, Representative flow cytometry (left) and quantification (right) of naive B cells binding to biotinylated 10E8-GT9 in wild-type or 10E8-UCA H mice. Error bars indicate mean ± s.d.from mice in each group (n = 3).Data from a single experiment are presented and are representative of three independent experiments.e, Top, 10x Genomics scBCR-seq data from 10E8-GT10.2-specificnaive B cells from 10E8-NGS-04 H mice (n = 1,026 pairs amplified) showing human 10E8-NGS-04 IGH-V gene frequency (left) and mouse IGK-V genes (right).Cells from three mice were sequenced, and representative data from one mouse are presented.Bottom, 10x Genomics scBCR-seq data from 10E8-GT10.2-specificnaive B cells from 10E8-UCA H mice (n = 1,418 pairs amplified) showing the frequency of the human 10E8-UCA IGH-V gene (left) and mouse IGK-V genes (right).Cells from three mice were sequenced, and representative data from one mouse are presented; HC, heavy chain; LC, light chain.

Fig. 8 |
Fig. 8 | Survival of 10E8 B cells in GCs after immunization with a clinically relevant candidate immunogen varies by precursor clonotype.a, Quantifications of the percentages of GC B cells (left) and epitope-specific CD45.2 + GT10 ++ KO − GC B cells (right) in spleens at day 7 p.i. in CD45.1 wild-type mice adoptively transferred with 10E8-NGS-04 H , 10E8-UCA H and MPER-HuGL18 H CD45.2 B cells to reach a frequency of 1:10 4 for each precursor lineage 1 day before immunization with 5 µg of 10E8-GT10.212mer or 10E8-GT9-KO 12mer with alhydrogel.Data are pooled from two independent experiments (n = 5-7 per treatment).Error bars indicate mean ± s.d.from mice in the pooled group.Significance was calculated by using a two-tailed unpaired t-test.b, Pie plots

2 | 10E8-UCA H B cells are primed and recruited to GCs. a, Affinities
5//doi.org/10.1038/s41590-024-01844-755.Kato, Y. et al.Multifaceted effects of antigen valency on B cell response composition and differentiation in vivo.Immunity 53, 548-563 (2020).Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons.The Ragon Institute of Mass General, MIT and Harvard, Cambridge, MA, USA.2Department of Immunology and Microbiology, The Scripps Research Institute, La Jolla, CA, USA.3IAVI Neutralizing Antibody Center, The Scripps Research Institute, La Jolla, CA, USA.4Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery, The Scripps Research Institute, La Jolla, CA, USA.5Institute for Drug Discovery, Leipzig University Medical Faculty, Leipzig, Germany.6Department of Integrative Structural and Computational Biology, The Scripps Research Institute, La Jolla, CA, USA.All animal experiments were performed under the approval of the Institutional Animal Care and Use Committee of Harvard University and the Massachusetts General Hospital (animal study protocols 2016N000022 and 2016N000286) and were conducted in accordance with the regulations of the Association for Assessment and Accreditation of Laboratory Animal Care International. https