Anti–HIV-1 B cell responses are dependent on B cell precursor frequency and antigen-binding affinity

Significance

An essential requirement for an HIV vaccine is to elicit antibodies to conserved regions of the spike protein (i.e., Env) because these antibodies can protect against infection. Although broadly neutralizing antibodies develop naturally in rare individuals after prolonged HIV infection, eliciting them by vaccination has only been possible in artificial knock-in mouse models in which the number of B cells expressing the antibody precursor is superphysiologic. To understand the relationship among precursor frequency, antigen affinity, and germinal center recruitment, we have performed adoptive transfer experiments in which fixed numbers of precursor cells are engrafted in WT mice. Our results provide a framework for an understanding of how precursor frequency and antigen affinity shape humoral immunity to HIV.

Abstract

The discovery that humans can produce potent broadly neutralizing antibodies (bNAbs) to several different epitopes on the HIV-1 spike has reinvigorated efforts to develop an antibody-based HIV-1 vaccine. Antibody cloning from single cells revealed that nearly all bNAbs show unusual features that could help explain why it has not been possible to elicit them by traditional vaccination and instead would require a sequence of different immunogens. This idea is supported by experiments with genetically modified immunoglobulin (Ig) knock-in mice. Sequential immunization with a series of specifically designed immunogens was required to shepherd the development of bNAbs. However, knock-in mice contain superphysiologic numbers of bNAb precursor-expressing B cells, and therefore how these results can be translated to a more physiologic setting remains to be determined. Here we make use of adoptive transfer experiments using knock-in B cells that carry a synthetic intermediate in the pathway to anti–HIV-1 bNAb development to examine how the relationship between B cell receptor affinity and precursor frequency affects germinal center (GC) B cell recruitment and clonal expansion. Immunization with soluble HIV-1 antigens can recruit bNAb precursor B cells to the GC when there are as few as 10 such cells per mouse. However, at low precursor frequencies, the extent of clonal expansion is directly proportional to the affinity of the antigen for the B cell receptor, and recruitment to GCs is variable and dependent on recirculation.

Most vaccine responses are elicited by immunization with attenuated, heat-killed, or genetically inactivated pathogens or pathogen-derived proteins (1, 2). High-affinity antibodies are the key effectors of protection in the great majority of these vaccines (3, 4). These antibodies develop in germinal centers (GCs), which are the microanatomic sites of B cell clonal expansion and antibody affinity maturation (5).

Although there is no effective vaccine against HIV-1, potent broadly neutralizing antibodies (bNAbs) that bind to HIV-1 envelope glycoprotein (Env) with high affinities can prevent infection in animal models even when present at low concentrations (6–9). However, elicitation of bNAbs to HIV-1 Env presents a series of what may be unique problems, including a requirement for one or more of the following unusual features: high levels of somatic hypermutation, incorporation of self-glycans into a complex peptide–glycan epitope, long CDRH3s, and self- or polyreactivity (10–13). Among these, the requirement for higher than normal levels of somatic mutation, a reaction that occurs in GCs, is nearly universal (14).

B cell entry into and persistence in GCs is dependent on a large number of variables including, but not limited to, the presence or absence of T cell help, antigen concentration, B cell receptor affinity for the antigen, responding B cell precursor frequency, valency of the antigen, and physiological state of the B cell (5, 15, 16). The role of B cell receptor affinity and precursor frequency in B cell entry into GC responses has been studied in WT mice (17–20). For example, GCs in WT mice immunized with the hapten (4-hydroxy-3-nitrophenyl)-acetyl (NP) are colonized by B cells with highly varied affinities, some as low as 500 μM (K_a) (18). Similar results have also been found in WT mice immunized with Bacillus anthracis protective antigen, influenza hemagglutinin, and chicken γ-globulin (19, 20).

In contrast to WT mice, results using Ig transgenic or knock-in B cells, with defined affinities for their cognate antigen, display far more variable results. Transgenic or knock-in B cells that carry anti-NP antibodies with affinities as low as 300 μM enter GCs and undergo affinity maturation unless outcompeted by higher-affinity antibody-expressing B cells (21–24). However, knock-in B cells that carry an anti-hen egg lysozyme (HEL)–specific receptor require high-affinity interactions (K_a < 23 μM) for efficient expansion (25, 26). Similarly, B cells that carry anti–HIV-1 Ig knock-in genes also appear to require high-affinity interactions with polymerized antigen to gain access to and participate in germinal-center reactions (27–30).

Whether the high-affinity requirements found in the anti–HIV-1 Ig knock-in mice represent a general rule for HIV-1–specific B cells or, alternatively, whether lower-affinity interactions characteristic of B cells participating in germinal-center reactions under physiologic conditions can induce HIV-1–specific B cell responses has not been determined. This question is particularly important for vaccines aimed at eliciting bNAbs to HIV-1 because they appear to require the recruitment of rare B cell precursors into GCs and sequential exposure to different immunogens (31–35).

Here we examine the relationship between precursor frequency and antigen binding affinity for B cells that carry a knock-in B cell receptor specific for the HIV-1 CD4-binding site.

Results

B Cell Development in 3BNC60^SI Knock-In Mice.

To study the relationship among B cell precursor frequency, affinity, and epitope-specific B cell responses to HIV-1 Env, we sought to use B cells that develop normally, display nearly complete allelic exclusion, and show normal levels of cell-surface IgM and IgD. This is particularly important for anti–HIV-1 heavy and light chain knock-in mice because a number of these mice, including 2F5 (29, 36, 37), 4E10 (38), 3BNC60 germline (30), and VRC01 germline (28), display combinations of abnormal B cell development, anergy, and/or absence of allelic exclusion.

B cell development in the bone marrow of 3BNC60^SI knock-in mice, expressing a synthetic intermediate antibody composed of the mature 3BNC60 heavy chain and the germline light chain (30, 31), showed the expected decreases in pre-B cells and immature B cells and increases in mature follicular B cells that are associated with expression of a non–self-reactive prerearranged knock-in B cell receptor (Fig. 1 A and B) (15, 39–41). Similarly, the spleen contained decreased levels of transitional T1 and T3 B cells, characteristics found in non–self-reactive Ig transgenic mice, whereas mice that carry self-reactive Igs show increased frequencies of T3 B cells (42–44). 3BNC60^SI knock-in mice also displayed increased frequencies of splenic follicular and marginal-zone B cells (Fig. 1 C and D). Thus, B cells in the bone marrow and spleen of 3BNC60^SI knock-in mice display no signs of self-reactivity or anergy, as measured by the size of B cell subpopulations and IgM/IgD expression levels, respectively.

Fig. 1.

B cell development in 3BNC60^SI knock-in mice. Representative flow-cytometric analysis (A) and quantification (B) of B cell development in bone marrow of WT C57BL/6 and 3BNC60^SI knock-in mice. Pre-B cells (IgM⁻IgD⁻), immature B cells (IgM⁺IgD⁻), and mature B cells (IgM^+/loIgD⁺) are shown (n = 5 mice per genotype; mean ± SEM). Plots are pregated on B220⁺CD19⁺CD2⁺ cells. Representative flow-cytometric analysis (C) and quantification (D) of B cell populations in spleen of WT and 3BNC60^SI knock-in mice. Transitional (T) 1 (CD93⁺CD23⁻IgM⁺), T2 (CD93⁺CD23⁺IgM^hiIgD⁺), T3 (CD93⁺CD23⁺IgM^loIgD^hi), follicular (FoB; CD93⁻CD23⁺IgM⁺IgD^hi), and marginal-zone B cells (MZB; CD93⁻CD23^loIgM⁺CD21/35⁺) are shown (n = 5 mice per genotype; mean ± SEM). Plots are pregated on B220⁺CD19⁺ cells. (E) Frequency of IgH-a⁺ and IgH-b⁺ B220⁺ cells (Upper) and frequency of hIgκ constant and mouse (m) Igλ among IgH-b⁺B220⁺ cells (Lower) in heterozygous WT (Left) and 3BNC60^SI knock-in (Right) mice crossed to IgH-a, hIgκ mice (gl, germline; mt, mature). Representative flow-cytometric graphs of WT (n = 3) and 3BNC60^SI knock-in (n = 6) mice are shown. (F) Frequencies of B cells that express the knock-in (KI) B cell receptor (BCR; IgH-b⁺hIgκ⁻mIgλ⁻) from the heterozygous offspring of anti–HIV-1 Ig knock-in mice (indicated on x-axis) crossed to IgH-a, hIgκ mice (n = 2–6 mice per genotype; mean ± SEM; HC, heavy chain; LC, light chain). (G) Representative flow-cytometric analysis of TM4 core (high-affinity antigen) or N276D (intermediate-affinity antigen) binding B220⁺IgM⁺ cells from WT (n = 3) and 3BNC60^SI knock-in (n = 11) mice. Numbers in plots indicate the mean frequency ± SEM of antigen-binding B cells (**P< 0.01 and ****P < 0.0001, two-tailed unpaired Student’s t test).

Absence of allelic exclusion is associated with self-reactivity or otherwise abnormal B cell receptors (45). To evaluate allelic exclusion and receptor editing, we combined IgH^a/b allotypes with a heterozygous human Igκ constant region knock-in (hIgκ) allele (46). As a result of allelic exclusion, ∼50% of WT B cells express IgH^a or IgH^b. Of the IgH^b- or IgH^a-expressing cells, ∼45% express hIgκ, ∼45% express mouse (m)Igκ, whereas the remaining 5–10% express mIgλ light chains (Fig. 1E, Left). In contrast, and as expected from non–self-reactive prerecombined knock-in Igs, 90% of heterozygous 3BNC60^SI knock-in B cells express the knock-in IgH^b allele in combination with the 3BNC60 Igκ light chain (Fig. 1E, Right). This is in contrast to numerous other strains of anti–HIV-1 Ig knock-in mice that we and others have generated that show far lower levels of knock-in BCR expression (Fig. 1F) (28, 36, 38). In addition, 99.6% and 99.1% of circulating B cells from homozygous 3BNC60^SI knock-in mice bound high-affinity (TM4 core) and intermediate-affinity (N276D) HIV-1 Env proteins, respectively, compared with 1.0% and 2.6% in WT controls. Furthermore, the magnitude of IgM expression by Env-binding 3BNC60^SI knock-in B cells is comparable to WT, indicating that these cells are not anergic (39) (Fig. 1G). In summary, 3BNC60^SI knock-in B cells show efficient allelic exclusion of WT alleles and are therefore relatively homogenous in terms of their antigen-binding properties. Moreover, 3BNC60^SI knock-in B cells show no appreciable correlates of self-reactivity.

Relationship Between Precursor Frequency and Clonal Expansion.

To track the fate of antigen-stimulated epitope-specific precursor B cells, allotype-marked 3BNC60^SI knock-in B cells (CD45.2) were transferred into congenic (CD45.1) WT mice. As reported by others (17), the efficiency of transfer was ∼5% as determined by dilution experiments (Table S1). To determine whether 3BNC60^SI knock-in B cells are responsive to soluble HIV-1 Env, we immunized WT mice engrafted with 100,000 3BNC60^SI knock-in B cells with 10 μg of soluble N276D Env in alum, which binds to the knock-in BCR with ∼40 μM affinity (Fig. S1 and Table S2). Three days after immunization, an average of 0.3% of all B cells in naive and immunized recipient mice were knock-in cells (Fig. 2A). GCs in draining lymph nodes peaked as early as 7 d after immunization (Fig. 2B), when the knock-in cells had expanded to represent 4–10% of all B cells (Fig. 2A, closed circles), and, on average, more than 60% of all B cells in GCs (Fig. 2C) compared with nonimmunized recipients in which there was no such expansion (Fig. 2 A, open circles, and D). We conclude that, when 100,000 3BNC60^SI knock-in B cells are engrafted into WT mice, they can develop into GC B cells and proliferate extensively after stimulation with soluble antigen that binds to the B cell receptor with ∼40 μM affinity.

Fig. 2.

3BNC60^SI knock-in B cells transferred into WT recipient mice expand and form GCs following soluble HIV-1 Env immunization. A total of 10⁵ 3BNC60^SI knock-in (KI) B cells (CD45.2) were engrafted into recipient B6.SJL (CD45.1) mice that were subsequently immunized with soluble N276D Env trimers (see also Table S2). (A) Frequency of 3BNC60^SI knock-in B cells in immunized (closed circles) and nonimmunized (open circles) recipient mice. (B) Total GC B cells of immunized recipient mice. (C) Frequency of 3BNC60^SI knock-in B cells among GC B cells of immunized recipient mice. (D) Representative flow-cytometric analysis of GC B cells (CD95⁺CD38⁻) among transferred 3BNC60^SI knock-in B cells (red) and endogenous B cells in immunized and nonimmunized animals. Data in A–D are pooled from two independent experiments (n = 2–8 mice per time point).

To examine the relationship between precursor frequency and B cell expansion in response to a soluble HIV-1 Env antigen, we engrafted varying numbers of 3BNC60^SI knock-in B cells into WT recipients and analyzed B cell responses 14 d after immunization with 10 μg of soluble N276D Env (∼40 μM) in alum. Mice engrafted with 1,200–100,000 B cells showed 10–50-fold increases in the fraction of 3BNC60^SI knock-in B cells after immunization (Fig. 3A). In mice engrafted with a high number of knock-in cells, these cells contributed to, on average, 30% of all B cells in GCs (Fig. 3B). Moreover, engraftment with as few as 100 3BNC60^SI knock-in B cells was sufficient for their recruitment into the GC after immunization with a soluble ∼40-μM affinity antigen (Fig. 3C and Fig. S2). We conclude that immunization with a soluble antigen of modest affinity expands epitope-specific precursor B cells in a manner that is directly correlated to the number of epitope-specific precursors present at the time of immunization.

Fig. 3.

3BNC60^SI knock-in B cell expansion in response to immunization correlates with precursor frequency. Frequency of 3BNC60^SI knock-in (KI) B cells of total B cells (A) and per GC B cell (B) in recipient mice engrafted with the indicated numbers of 3BNC60^SI knock-in B cells (below) 14 d following soluble N276D Env immunization (closed circles). Mice receiving the same number of cells with no immunization are shown for comparison (open circles). (C) Representative flow-cytometric analysis of mice in A and B. Percentage of GC B cells among 3BNC60^SI knock-in B cells are shown in plots (see also Fig. S2). Data in A–C are representative of two independent experiments (n = 2–3 mice per group per time point). Bars in all graphs indicate mean values.

Relationship Among Affinity, Precursor Frequency, and Clonal Expansion.

To examine the role of HIV-1 Env antigen affinity in 3BNC60^SI knock-in B cell responses to immunization, we engrafted WT mice with 10,000, 100, or 10 cells and immunized with 10 μg of one of the three soluble Env trimer antigens of varying affinities (∼7 nM, ∼40 μM, and ∼200 μM; Fig. S1 and Table S2) in alum. The frequency of knock-in B cells in the draining lymph nodes and their phenotype was determined 14 d after immunization. In the absence of immunization, transferred B cells were detected in only mice engrafted with 10,000 cells, and they remained naïve (CD95⁻CD38⁺; Fig. 4A).

Fig. 4.

Relationship between antigen affinity and precursor frequency. (A) Representative flow-cytometric analysis of GC B cells in B6.SJL (CD45.1) recipient mice engrafted with 10,000 (n = 6 mice per antigen), 100 (n = 8 mice per antigen), or 10 (n = 20 mice per antigen) 3BNC60^SI knock-in (KI) B cells (CD45.2) 14 d following immunization with the indicated soluble Env trimer protein (Upper; see also Table S2). 3BNC60^SI knock-in (red) and endogenous (gray) GC B cell responses are shown. (B) Frequency of 3BNC60^SI knock-in B cells from mice in A. Animals showing undetectable frequencies of knock-in B cells were set to 0.001 to indicate the number of nonresponders. (C) Frequency of GC B cells among 3BNC60^SI knock-in B cells from mice in A. Data are pooled from two independent experiments. Bars in all graphs indicate mean values.

All antigens produced similar-sized GCs in the control mice that did not receive 3BNC60^SI knock-in B cells, indicating similar overall immunogenicity by the different antigens (Fig. S3). Mice immunized with high-affinity soluble antigen (∼7 nM) showed high levels of 3BNC60^SI knock-in B cell expansion irrespective of the number of precursors present at the time of immunization (Fig. 4 A and B). In addition, nearly all of the transferred B cells in the draining lymph nodes of these mice displayed a GC B cell phenotype (CD95⁺CD38⁻; Fig. 4C). However, the number of mice with precursor cells that responded and the extent of their expansion was variable in mice engrafted with smaller numbers of precursors. Knock-in B cells also responded to immunization with intermediate-affinity soluble antigen (∼40 μM) even when present at levels as low as 10 precursor B cells per mouse. However, in the presence of limited numbers of knock-in cells (100 and 10 cells per mouse), the degree of B cell expansion and their relative contribution to the GC reaction was decreased and variable in comparison to high-affinity antigen. Finally, immunization with low-affinity soluble antigen (∼200 μM) produced GC responses by the engrafted cells, but this response was highly variable and mainly found in mice engrafted with the highest number of precursor cells (Fig. 4).

The difference in the relative amount of B cell expansion in response to antigens of differing affinities could be seen as early as 3–4 d after immunization in mice engrafted with high numbers of cells (10,000 cells/mouse). CellTrace Violet (CTV)-labeled 3BNC60^SI knock-in B cells diluted the dye to a greater extent in mice immunized with high-affinity than with intermediate-affinity antigen, in which no cell division was detected at this time point (Fig. S4). However, the amount of expansion in response to high- and intermediate-affinity antigens was similar by day 14 after immunization (Fig. 4 A and B). The discrepancy between days 3 and 14 suggests a kinetic effect that is antigen affinity-dependent.

In summary, in the context of a polyclonal immune system in which a given precursor is present in only small numbers, a higher-affinity antigen is essential for reproducible induction of high levels of epitope-specific B cell expansion. Nevertheless, even lower-affinity soluble antigens can recruit B cell precursors present in limiting numbers to the GC reaction, but this response is more variable than when large numbers of precursors are present.

Rare Precursor B Cell Expansion Is Dependent on Lymph Node Homing.

Our analysis is limited to the draining lymph nodes, and the probability that a rare engrafted cell is present in those nodes at the time of immunization is small. Thus, it is possible that under conditions of limiting precursor frequency, GC participation requires that engrafted 3BNC60^SI knock-in B cells survey the animal and enter lymph nodes containing ongoing GC reactions even after the initiation of the reaction (24, 47). To examine this possibility, we blocked lymphocyte homing and egress from lymph nodes with a combination of a neutralizing antibody to CD62L and FTY-720 (an S1PR1 antagonist). Compared with controls engrafted with 10 3BNC60^SI knock-in B cells, anti-CD62L and FTY-720–treated mice immunized with high-affinity (7 nM) soluble antigen showed significantly decreased levels of precursor B cell expansion (Fig. 5A) and GC B cell entry (Fig. 5 B and C). We conclude that the efficient activation of rare naïve precursor cells with a high-affinity antigen is dependent on the recirculation of naïve B cells.

Fig. 5.

Recruitment of rare precursors to GCs is dependent on B cell recirculation. A total of 10 3BNC60^SI knock-in (KI) B cells (CD45.2) were engrafted into recipient B6.SJL (CD45.1) mice and immunized with soluble TM4 core (high-affinity Env, ∼7 nM). To block homing and egress of lymphocytes, recipient mice were treated with neutralizing antibody to CD62L and FTY-720 (treated; n = 20 mice) on the day of immunization and every 48 h thereafter until the day of analysis (day 14). Control mice were treated with Rat IgG2a isotype antibody and vehicle (control; n = 20 mice). Frequency of 3BNC60^SI knock-in B cells of total B cells (A), frequency of GC B cells among 3BNC60^SI knock-in B cells (B), and frequency of 3BNC60^SI knock-in B cells among GC B cells (C). Data in A–C are pooled from two independent experiments (*P < 0.05 and ****P < 0.0001, two-tailed unpaired nonparametric Mann–Whitney U test). Bars in all graphs indicate mean values.

Discussion

Potent bNAbs to HIV-1 Env are protective in animal models even at low concentrations, and it is generally accepted that a vaccine that elicits such antibodies would be efficacious (6–9, 12, 48). However, these antibodies have been elicited only by immunization in mice that carry human Ig knock-in genes using a sequence of immunogens specifically designed to target the knock-in B cell receptor (33–35). Although these experiments establish that sequential immunization can produce bNAbs from specific germline precursors, there are several important issues that must be resolved before these results can be translated to genetically unmanipulated animals and humans. Among these are the requirements for antigen affinity in recruiting anti–HIV-1 bNAb precursor cells to GCs and how affinity requirements relate to precursor frequency and the clonal expansion of the target cell.

To date, experiments with knock-in mice expressing the 2F5 or VRC01 germline anti–HIV-1 antibodies suggest a requirement for high-affinity polymerized antigen for B cell clonal expansion (27–29). 2F5 knock-in mice carry a self-reactive receptor that fails to support B cell development, and only small numbers of cells escape central tolerance and enter the periphery with an anergic phenotype (27, 29, 36, 37). Nevertheless, these self-reactive B cells can expand when the knock-in mice are immunized with a high-affinity polymerized antigen (27, 29). Similar results have also been obtained with the model antigen HEL using B cells from mice that express HEL as a neo–self-antigen and carry a self-reactive anti-HEL antibody gene (49).

A requirement for high-affinity (0.5 μM) polymerized antigens was also found in adoptive transfer experiments using germline VRC01 knock-in B cells (28). Moreover, high levels of B cell expansion were found only when specific precursors were present at concentrations of 1:10⁵ or 200–400 cells per mouse. One potential explanation for the difference between VRC01 knock-in B cells and the results obtained in studies of GCs in WT mice (18–20) is that VRC01 knock-in B cells display some features of abnormal B cell development (28). Heterozygous VRC01 knock-in B cells fail to display high levels of allelic exclusion. These cells were shown to coexpress knock-in Igκ and endogenous Igλ light chains (28). Under physiologic circumstances, only 2–5% of B cells express dual light chain receptors (46, 50). Ig transgenic or knock-in mice like 3BNC60^SI that express non–self-reactive receptors also show the same high levels of allelic exclusion (46, 50–52). In contrast, dual light chain-expressing B cells, like VRC01 germline knock-in, in which an unusually large proportion of B cells displayed double expression of κ and λ-light chains (28), are typically found in autoimmune B cells of mice and humans (45, 53–56). These cells arise by receptor editing, whereby B cells expressing self-reactive receptors or incompatible Ig heavy and light chains have their receptors replaced or diluted by continuing V(D)J recombination (39, 46, 57–60). Transgenic or knock-in B cell receptors that are designed to be resistant to this process can nevertheless be edited by coexpressing a second light chain, as seen in VRC01 knock-in B cells (28). Coexpression of a second light chain produces chimeric receptors, which dilute the effects of the receptor that fails to support normal B cell development (39, 46, 57–60). In such cases, each developing B cell expresses a unique and random combination of receptors composed of the knock-in Ig heavy and/or the knock-in light chain and endogenous light chains. The result is that many of the developing B cells fail to bind to their cognate antigen and those that do so may display variable levels of affinity and avidity.

Our experiments were performed using 3BNC60^SI knock-in B cells that display normal levels of allelic exclusion and the expected uniform levels of antigen binding. In addition, 3BNC60^SI knock-in B cells show no measurable signs of anergy or self-reactivity. The data indicate that, even when present in limiting numbers (10 precursor B cells per mouse), such cells can be recruited to GCs by soluble HIV-1 Env antigens. Soluble high-affinity antigen (∼7 nM) induced high levels of clonal expansion in draining lymph nodes when as few as 10 specific precursor cells were present in the mouse. Although lower-affinity antigens (∼40–200 μM) were able to recruit 3BNC60^SI knock-in B cells into the GC, B cell expansion was limited and their recruitment was more variable. Our results are in keeping with observations made in anti-NP knock-in and WT mice in that B cells found in GCs of mice immunized with the hapten NP or with protein antigens can display micromolar affinities for the immunogen (18–20). The differing levels of expansion induced by antigens of varying affinity are likely related to B cell competition for T cell help at the T–B border and within GCs (19, 23, 61–63).

3BNC60^SI differs from 2F5 and VRC01 in that it is not a germline precursor but a synthetic intermediate in HIV-1 bNAb development composed of a mutated heavy chain and a germline light chain. Moreover, B cells in 3BNC60^SI appear to develop normally, whereas several different lines of anti–HIV-1 Ig knock-in mice show abnormalities associated with self- or polyreactivity, including germline 2F5, VRC01, 3BNC60, CH103, and, to a lesser extent, PGT121 (Fig. 1F) (28, 30, 33, 36, 37). The abnormalities in development found in these knock-in mice are entirely consistent with the observation that anti–HIV-1 antibodies are frequently self- and/or polyreactive (10, 64, 65). Whether B cells expressing authentic bNAb precursors in humans show the same developmental defects and differ from B cells that develop normally in their requirements for GC recruitment and clonal expansion remains to be determined.

Materials and Methods

Mice.

3BNC60^SI (synthetic intermediate) knock-in mice (CD45.2) carry the prerearranged Ig V(D)J genes encoding the mature heavy chain (somatically mutated) and predicted germline light chain of human bNAb 3BNC60 (30, 31). Mice used in experiments were homozygous for both knock-in alleles (CD45.2). IgH^a/aIgκ^hu/hu, PGT121 germline (glHCLC) and PGT121^SI were described previously (33, 46). CH103 germline (glHCLC) knock-in mice were generated as described previously (31, 33) C57BL/6 WT and B6.SJL (CD45.1) mice were obtained from Jackson Laboratories. All experiments were conducted with approval from the institutional review board and the institutional animal care and use committee at The Rockefeller University.

HIV-1 Env Antigens.

Soluble trimeric gp140 Env antigens were expressed in 293-F cells and purified as previously described (30). All antigens are derived from the Clade C 426c gp140 Env (66) or 426c TM4ΔV1-3 Env (herein called TM4 Core) (31). A more detailed description is given in SI Materials and Methods.

Bio-Layer Interferometry Assay.

Bio-Layer Interferometry (BLI) assays were performed on the Octet Red instrument (ForteBio) as previously described (30). Biotinylated trimeric recombinant gp140 proteins were immobilized on streptavidin biosensors (ForteBio) immersed into wells containing dilutions of purified recombinant Fabs. Details of this process and determination of K_a values are given in SI Materials and Methods.

Lymphocyte Preparation.

Lymphocytes were harvested as described in SI Materials and Methods. B cells were enriched by negative selection by using anti-CD43 MicroBeads (Militenyi Biotec) with magnetized LS columns according to manufacturer instructions. In some experiments, B cells were labeled with CTV (Thermo Scientific) according to manufacturer instructions.

Animal Experiments.

3BNC60^SI knock-in B cells were transferred i.v. into B6.SJL recipient mice. Ten micrograms of indicated soluble Env protein in Imject Alum (Thermo Scientific) was injected s.c. into hind footpads. To block lymph node homing and egress of lymphocytes, mice were administered neutralizing antibody to CD62L and FTY-720 (SI Materials and Methods).

Flow Cytometry.

Single-cell suspensions of lymphocytes were stained for flow cytometry as described in SI Materials and Methods. Antigen-specific B cells were identified by using biotinylated TM4 core monomers or N276D-gp140 trimers.

Determination of B Cell Transfer Efficiency.

A defined number of knock-in B cells (CD45.2) were transferred into B6.SJL recipient mice (CD45.1). The following day, peripheral lymphoid organs were collected and absolute numbers of lymphocytes were calculated by using a hemocytometer. Samples were next analyzed by flow cytometry to determine the frequency of knock-in B cells present in the sample. This information was used to determine the absolute number of knock-in B cells engrafted in recipient mice at the time of immunization.

Statistics.

Data were analyzed with Prism 6 (GraphPad) by using two-tailed unpaired Student’s t tests (normally distributed data sets comparing the means of two samples) or two-tailed nonparametric Mann–Whitney U tests (data sets that were determined by an F test not to have a normal distribution and in which there was a comparison of the mean of two samples) as indicated in the text. For all analyses, P ≤ 0.05 was considered statistically significant.