Cross-Reactivity with Self-Antigen Tunes the Functional Potential of Naive B Cells Specific for Foreign Antigens

During hematopoietic development, B cell progenitors undergo VDJ recombination to generate Ig expressed on the cell surface as a BCR (1). The random nature of this process gives way to the broad spectrum of receptor specificities needed to recognize diverse pathogens but can also yield cells expressing receptors that bind self-antigens. To prevent Ab-mediated autoimmunity, self-reactive B cells are subject to tolerance mechanisms including receptor editing, clonal deletion, and the restriction of functional responsiveness, termed anergy (2). The clones that survive these checkpoints comprise the repertoire of naive B cells that may respond to vaccination or infection (3). There is growing evidence suggesting that Ab responses are hindered by peripheral tolerance mechanisms due to cross-reactivity of Ag-specific BCR with host-derived self-antigens. Studies of individual Ab lineages have identified isolated cases where such cross-reactivity may influence activation of particular clones (4–10). Additionally, several studies have found increased protective Ab responses in animals and patients prone to autoimmunity (11–13). However, it is not clear from these examples how universal this feature may be for naive B cell populations targeting different foreign Ags.

Such studies pose a number of technical difficulties, as they require analysis of rare Ag-specific B cells prior to somatic hypermutation. Previous work has shown that as many as 20% of the BCRs in the naive human repertoire exhibit detectable reactivity with self-antigens (14). In this work however, the specificity of these B cells was not fully defined. More recently, Zikherman and colleagues (15–18) used a reporter mouse in which BCR signaling in response to self-antigen binding was assessed by the expression of eGFP as a surrogate for the expression of the orphan nuclear receptor Nur77. These studies revealed that mature follicular B cells expressed a range of Nur77, and high levels were associated with self-antigen reactivity and diminished function (15). Importantly, whereas other factors such as CD40 and TLR signaling could result in modest Nur77^eGFP upregulation in vitro, Nur77^eGFP expression can be detected in naive B cells at steady-state in the absence of these signals (16). These studies also revealed that Nur77^eGFP expression was low in naive MD4 B cells, which express a transgenic BCR specific for the foreign Ag hen egg lysozyme (15, 16). These data may indicate that naive B cells specific for foreign Ags exhibit low cross-reactivity with self-antigens. However, this conclusion is drawn upon the examination of a single BCR, while there are potentially hundreds of thousands of unique BCRs able to bind a given Ag. Furthermore, the MD4 BCR was originally isolated from an animal postimmunization following germinal center selection and extensive somatic hypermutation, which may have reduced cross-reactivity from the unmutated BCR from which the MD4 Ab was derived. A more recent study analyzed B cells with a fixed VH186.2 transgenic H chain specific for the foreign Ag 4-hydroxy-3-nitrophenyl (NP) paired to different nontransgenic L chains. In this context, positive selection was more evident in cells expressing BCR that exhibited higher self-antigen reactivity (18). These data examining oligoclonal populations of B cells indicates that mild self-reactivity is a key part of B cell development and suggests that nontransgenic polyclonal populations of foreign Ag-specific naive B cells would have similar features.

A recent study examining the reactivity of foreign and self-reactivity of Abs produced by cultures of thousands of single the transitional and mature human B cells may have shed light on this question (19). These data revealed that on average, half of the B cells binding both foreign and self-antigens were deleted between the transitional and mature stage (19). Importantly, self-antigen cross-reactivity was never completely removed from the foreign Ag-specific populations, and the extent of deletion appeared to vary from Ag to Ag (19).

In this study, we used Ag-specific enrichment to analyze populations of naive murine B cells specific for both model Ags as well as pathogen-derived Ags and found that Nur77^eGFP was expressed similarly among most populations. Unexpectedly, these levels were indistinguishable from the larger population of naive B cells whose specificities were unknown. Similarly, naive human B cells specific for influenza virus, respiratory syncytial virus (RSV), or HIV-1 proteins contained B cells with the IgD⁺ IgM^LOW phenotype that has been previously reported to contain human self-antigen-reactive B cells (20, 21). Importantly, the level of self-antigen cross-reactivity appears to influence the likelihood that a naive B cell will respond to Ag immunization. Naive Ag-specific B cells with high Nur77^eGFP levels were ∼2-fold less likely to proliferate in response to Ag injection compared with their counterparts that expressed moderate levels of Nur77^eGFP. However the cells with low Nur77^eGFP levels were also less likely to proliferate compared with the population expressing moderate levels of Nur77^eGFP, indicating that some level of cross-reactivity with self-antigen may support a cell state optimally poised to proliferate during an immune response. Together, our data suggest that all polyclonal populations of B cells specific for foreign Ags contain BCRs with a range of self-antigen reactivities that tunes cells for optimal function.

Six- to fourteen-week-old male and female mice were used for experiments. CD45.2 (C57BL/6), CD45.1 congenic (B6.SJL-Ptprca Pep3b/BoyJ), IgH^a congenic (B6.Cg-Gpi1^a Thy1^a Igh^a/J), Nur77^eGFP (C57BL/6-Tg(Nr4a1-EGFP/cre)820Khog/J) (22), MD4 (C57BL/6-Tg(IghelMD4)4Ccg/J) (4) and Rag1^−/− (B6.129S7-Rag1tm1Mom/J) (23) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). IgH^a and Nur77^eGFP were backcrossed to generate IgH^a/a Nur77^eGFP mice before use. MD4 and Rag1^−/− were backcrossed to generate MD4 Rag1^−/−. Animals were maintained in a specific pathogen-free facility in accordance with Fred Hutchinson Cancer Research Center Institutional Animal Care and Use Committee approval and National Institutes of Health guidelines.

Purified R-PE and allophycocyanin were purchased from ProZyme, and OVA was purchased from MilliporeSigma. Biotinylated peptides (GGGEQKLISEEDLGGG) conjugated to NP (NP-biotin) or DNP (DNP-biotin) were purchased from GenScript.

NP650 was created by first conjugating streptavidin–allophycocyanin (PJ27S; ProZyme) to DyLight 650 NHS ester following the manufacturer’s instructions (Thermo Fisher Scientific). The molar concentration of streptavidin–allophycocyanin–DyLight 650 was calculated by measuring the absorbance of streptavidin–allophycocyanin at 280 nm using the extinction coefficient of 0.268 μM⁻¹cm⁻¹. The resulting streptavidin–allophycocyanin–DyLight 650 was mixed at a molar ratio of 1:6 with NP-biotin and incubated for 30 min on ice. The resulting conjugate was purified from excess NP-biotin using a 100-kDa Amicon Ultra size exclusion column (MilliporeSigma), and the concentration was determined as described for streptavidin–allophycocyanin–DyLight 650.

RSV prefusion and postfusion F Ags, HIV-1 envelope (Env) 426C^TM4ΔV1–3, and influenza hemagglutinin (HA) stem were produced as described previously (24–26).

Purified Ags were biotinylated using an EZ-Link Sulfo-NHS-LC-Biotinylation kit (Thermo Fisher Scientific) using a 1:1.3 M ratio of biotin to Ag. Unconjugated biotin was removed by centrifugation using an Amicon Ultra size exclusion column (MilliporeSigma) with a smaller m.w. cutoff than the Ag. To determine the average number of biotin molecules bound to each Ag, streptavidin–PE (PJRS25; ProZyme) was mixed with a fixed amount of biotinylated Ag in increasing concentrations and incubated at room temperature for 30 min. Samples were run on an SDS-PAGE gel (Bio-Rad Laboratories) and transferred to nitrocellulose prior to incubation with streptavidin–Alexa Fluor 680 (Thermo Fisher Scientific) diluted 1:10,000 in 1× DPBS containing 0.2% Tween 20 (MilliporeSigma) to determine the ratio at which there was excess biotinylated Ag available for streptavidin–Alexa Fluor 680 to bind. To make tetramers, biotinylated Ags were mixed with streptavidin–PE or streptavidin–allophycocyanin (PJ27S; ProZyme) at the ratio determined above or at a 5 to 1 ratio using the biotin concentration provided by the manufacturer. Following a 30-min incubation on ice, unconjugated biotinylated Ag was often removed by several rounds of dilution and concentration using a 100 kDa Amicon Ultra (MilliporeSigma) or 300 kDa Nanosep centrifugal devices (Pall). Tetramers were stored at 1 μM in 1× DPBS at 4°C or 1× DPBS containing 50% glycerol at −20°C prior to use. Control PE594, PE650, and allophycocyanin–DyLight 755 (Allophycocyanin755) tetramers were created by mixing a biotinylated control Ag with streptavidin–PE preconjugated with DyLight 594 (PE594) or DyLight 650 (PE650), and streptavidin–Allophycocyanin755 following the manufacturer’s instructions for DyLight conjugation (Thermo Fisher Scientific). On average, PE594/PE650 and Allophycocyanin755 contained four to eight DyLight molecules per PE or allophycocyanin. The concentration of each tetramer was calculated by measuring the absorbance of PE at 565 nm combined with an extinction coefficient of 1.96 μM⁻¹cm⁻¹ or allophycocyanin at 650 nm using an extinction coefficient of 0.7 μM⁻¹cm⁻¹.

The spleen and inguinal, axillary, brachial, cervical, mesenteric, and periaortic lymph nodes from individual mice were pooled and dissociated with forceps prior to incubation with 0.8 mg/ml Dispase (Thermo Fisher Scientific), 0.2 mg/ml Collagenase P (Roche), and 0.1 mg/ml of DNase I (Roche) in RPMI 1640 (Thermo Fisher Scientific) for 20 min at 37°C. Tissue fragments were next forced through a 100-micron mesh to create single-cell suspensions, which were filtered using another 100-μm mesh to remove tissue debris and centrifuged at 300 × g for 5 min at 4°C.

Venipuncture was used to obtain blood from healthy, HIV-seronegative adult volunteers enrolled in the General Quality Control study in Seattle, WA. This protocol was approved by the Fred Hutch institutional review board, and informed consent was obtained from all study participants before enrollment into the parent protocols. PBMCs were isolated from whole blood using Accuspin System Histopaque-1077 (MilliporeSigma), and the resulting cell fraction was washed in 1× DPBS and centrifuged at 300 × g for 5 min at 4°C. The supernatant was aspirated and the cell pellet resuspended in 10% DMSO in heat-inactivated FBS and cryopreserved in liquid nitrogen before use.

Ag-specific B cells were enriched using previously described protocols (27–30). For labeling with fluorescent model Ags, 0.2-ml samples containing 1–2 × 10⁸ cells were incubated with 1 pmole of allophycocyanin, NP650, or PE for 25 min on ice in FACS buffer (1× DPBS containing 1% heat-inactivated newborn calf serum) containing 2 μg of anti-Fc receptor Ab 2.4G2 (Bio X Cell) for murine cells or 2.5% heat-inactivated mouse serum (Thermo Fisher Scientific) and 2.5% heat-inactivated rat serum (Thermo Fisher Scientific) for human cells. For Ag tetramer labeling, cells were first incubated with 1–3 pmole of tetramer containing a control Ag conjugated to PE650, PE594, and/or Allophycocyanin755 for 10 min on ice prior to incubation with 1 pmole of allophycocyanin- or PE-conjugated Ag tetramer for 25 min on ice.

After the incubation, 15 ml of FACS buffer was added and the sample was centrifuged at 300 × g for 5 min at 4°C. The supernatant was discarded and the pellet was resuspended prior to the addition of 25 μl of anti-allophycocyanin and/or 25 μl of anti-PE microbeads (Miltenyi Biotec). Following a 15–30 min incubation on ice, 5 ml of FACS buffer was added, and the sample was passed over a magnetized LS column (Miltenyi Biotec). The tube and column were washed once with FACS buffer and then removed from the magnetic field. Five milliliters of FACS buffer was pushed through the column with a plunger twice to elute column-bound cells.

Cells from the column-bound fraction and 1/40th of the column flow through fractions were incubated in 50 μl of FACS buffer containing a mixture of Abs and a 0.05 μl/ml fixable viability dye (FVD) eFluor 506 (eBioscience), FVD eFluor 780 (eBioscience), or Ghost Dye Violet 510 (Tonbo Biosciences) for 30 min on ice. Abs for murine experiments included various combinations of 2 μg/ml GL7 Fitc (GL7; BD Biosciences), 2 μg/ml anti-IgD PerCP-Cy5.5 (11-26c.2a; BD Biosciences), 4 μg/ml anti-IgM^b BV650 (AF6-78; BD Biosciences), 4 μg/ml anti-Igλ BV421 or BUV395 (R26-46; BD Biosciences), 2 μg/ml anti-Igκ BV605 (187.1; BD Biosciences), 2 μg/ml anti-CD9 PE (eBioKMC8; eBioscience), 2 μg/ml anti-CD19 BUV395 or BUV737 (1D3; BD Biosciences), 2 μg/ml anti-CD21 PE-Cy7 (eBio8D9; eBioscience), 2 μg/ml anti-CD23 BV786 (B3B4; BD Biosciences), 2 μg/ml anti-CD38 Pacific Blue (90; BioLegend), 2 μg/ml anti-B220 BV711 or BV786 (RA3-6B2; BD Biosciences), 2 μg/ml anti-CD45.1 PE-eFluor 610 (A20; eBioscience), 2 μg/ml anti-CD45.2 PerCP-Cy5.5 (104; eBioscience), 2 μg/ml anti-CD45.2 PE-Cy7 (104; BioLegend), 4 μg/ml anti-CD79b BUV661 (HM79b; BD Biosciences), 2 μg/ml anti-CD93 BV421 or BV650 (AA4.1; BD Biosciences), 2 μg/ml anti-CD3 BV510 (145-2C11; BD Biosciences), 2 μg/ml anti-F4/80 BV510 (BM8; BioLegend), and 2 μg/ml anti-Gr-1 BV510 (RB6-8C5; BD Biosciences). Following surface staining, samples were centrifuged for 5 min at 300 × g and supernatant was discarded. For intracellular Ig assessment, samples were resuspended and incubated with 250 μl of Cytofix/Cytoperm (BD Biosciences) for 30 min on ice. Four milliliters of permeabilization buffer (BD Biosciences) was added, samples were centrifuged for 5 min at 400 × g, and supernatant was discarded. Samples were resuspended with 40 μg/ml F(ab′)₂ goat anti-IgG (H+L) Alexa Fluor 350 and incubated for 30 min on ice. Four milliliters of permeabilization buffer (BD Biosciences) was added, samples were centrifuged for 5 min at 400 × g, and supernatant was discarded.

For human PBMC experiments, cells were labeled with various combinations of 6.24 μg/ml anti-IgM FITC (G20-127; BD Biosciences), 0.3 μg/ml anti-IgD PerCP-Cy5.5 (IA6-2; BD Biosciences), 25 μg/ml anti-Igλ Pacific Blue (MHL-38; BioLegend), 2.5 μg/ml anti-Igκ BV786 (G20-193; BD Biosciences), 50 μl/ml of anti-CD10 BUV395 (HI10a; BD Biosciences), 20 μl/ml of anti-CD19 PE-Cy7 (HIB19; BD Biosciences), 100 μl/ml of anti-CD19 BUV496 (SJ25C1; BD Biosciences), 20 μg/ml anti-CD20 eFluor 450 (2H7; eBioscience), 25 μl/ml anti-CD20 BUV395 (2H7; BD Biosciences), 50 μl/ml of anti-CD27 BV650 or BV480 (L128; BD Biosciences), 20 μg/ml anti-CD79b allophycocyanin/Fire750 (MHL-38; BioLegend), 60 μl/ml of anti-CD3 BV711 (UCT1; BD Biosciences), 60 μl/ml of anti-CD14 BV711 (MφP9; BD Biosciences), and 2.8 μg/ml of anti-CD16 BV711 (3G8; BioLegend). Following surface staining, samples were centrifuged for 5 min at 300 × g and supernatant was discarded. Human samples were resuspended with 250 μl of Cytofix/Cytoperm (BD Biosciences) or 2% paraformaldehyde and incubated for 15 min on ice prior to the addition of 4 ml of FACS buffer. Human samples were centrifuged for 5 min at 400 × g, and supernatant was discarded.

Prior to flow cytometry, both human and murine samples were resuspended in FACS buffer containing 20,000 Fluorescent AccuCheck counting beads (Thermo Fisher Scientific) to calculate cell numbers. Flow cytometry was performed on a five-laser (355, 405, 488, 561, 640 nm) LSR II, LSRFortessa, or FACSymphony (BD Biosciences) and analyzed with FlowJo software (Tree Star, Ashland, OR).

Naive follicular B cells were enriched from pooled spleen and lymph nodes from 18–20 mice using a B cell negative selection kit (Miltenyi Biotec, Auburn, CA) supplemented with Biotin-conjugated anti-CD93 (BD Biosciences) to eliminate transitional B cells (31). The B cells were washed in warm RPMI medium, adjusted to a final concentration of 2 × 10⁷ cells/ml in prewarmed RPMI medium, and incubated with 5 μM CellTrace Violet (CTV) (Thermo Fisher Scientific) for 12 min at 37°C prior to the addition of media containing 10% newborn calf serum. Following washing, cell surface marker staining was performed as described above. B cells expressing different levels of Nur77^eGFP were sorted by a FACSAria II cell sorter (BD Biosciences) using side scatter height/side scatter width–based duplet discrimination, followed by gating on CD9⁻ CD3⁻ F4/80⁻ Gr-1⁻ FVD eFluor 780⁻. CD9 was used to exclude marginal zone cells (32).

Purified CTV-labeled B cells were adjusted to a concentration of 2 × 10⁶ cells/ml in RPMI (Thermo Fisher Scientific, Logan, UT) containing 10% FBS (Thermo Fisher Scientific), 100 U/ml penicillin (Thermo Fisher Scientific), 100 μg/ml streptomycin (Thermo Fisher Scientific), 2 mM l-glutamine (Thermo Fisher Scientific), and 27.5 μM 2-ME (MilliporeSigma). Five milliliters of cells was added per well of a six-well flat bottomed plate and cultured in the presence or absence of 1–10 μg/ml F(ab′)₂ goat anti-mouse IgG (H+L) (Jackson ImmunoResearch) for 72 h at 37°C. Following incubation, cell surface marker staining was performed as described above prior to analysis of CTV dilution in gated CD19⁺ CD3⁻ Gr-1⁻ F4/80⁻ FVD eFluor 780⁻ cells using flow cytometry.

A total of 1–3 × 10⁶ of FACS-purified B cells in 0.1 ml of 1× DPBS was injected retro-orbitally into CD45.1⁺ recipient mice. The following day, animals were injected s.c. in the base of the tail with 50 μl of CFA (MilliporeSigma) mixed 1 to 1 with 1× DPBS with or without 62.5 pmoles of NP650. Seven days following Ag injection, NP650⁺ cells were enriched and CD45.2⁺ CD45.1⁻ donor B cells were assessed using flow cytometry as described above.

Single human B cells were isolated using a FACSAria II cell sorter (BD Biosciences) following tetramer enrichment and cell surface marker staining using human PBMCs as described above. Specifically, Env-specific B cells were sorted using side scatter height/side scatter width–based duplet discrimination, followed by gating on CD19⁺ CD20⁺ CD3⁻ CD14⁻ CD16⁻ FVD⁻ B cells that bound HIV-1 Env–PE (or allophycocyanin) tetramers but not PE650 (or PE594 or Allophycocyanin755) tetramers containing control Ags. Single cells were sorted into individual wells of 96-well PCR plates (Eppendorf) containing 10 μl/well of ice-cold lysis buffer containing 0.25 μl 12.5 U RNase out (Thermo Fisher Scientific), 2.5 μl 5× SuperScript IV First Strand Buffer (Thermo Fisher Scientific), 0.625 μl 0.1 M DTT (Thermo Fisher Scientific), 0.3125 μl 10% Igepal detergent (MilliporeSigma), and 6.625 μl DEPC-treated water. Plates were sealed with adhesive PCR plate seals (Thermo Fisher Scientific), centrifuged briefly, and immediately frozen on dry ice before storage at −80°C.

Reverse transcription was performed using SuperScript IV (Thermo Fisher Scientific) as previously described (33, 34). Briefly, 3 μl of reverse transcription reaction mix consisting of 1.5 μl of 50 μM random hexamers (Thermo Fisher Scientific), 0.4 μl 25 mM dNTPs (Thermo Fisher Scientific), 0.5 μl 10 U SuperScript IV RT, and 0.6 μl water was added to each well containing a single-sorted B cell in 10 μl lysis buffer and incubated at 50°C for 1 h. Following reverse transcription, 2 μl of cDNA was added to 19 μl of PCR mix containing 0.2 μl 0.5 U HotStarTaq Polymerase (Qiagen), 0.075 μl 50 μM 3′ reverse primers, 0.115 μl 50 μM 5′ forward primers, 0.24 μl 25 mM dNTPs, 1.9 μl 10× buffer (Qiagen), and 16.5 μl water. The PCR program for IgM/IgG and Igκ was 50 cycles of 94°C for 30 s, 57°C for 30 s, and 72°C for 55 s, followed by 72°C for 10 min. The PCR program for Igλ was 50 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 55 s, followed by 72°C for 10 min. After the first round of PCR, 2 μl of the PCR product was added to 19 μl of the second round PCR so that the final reaction contained 0.2 μl 0.5 U HotStarTaq Polymerase, 0.075 μl 50 μM 3′ reverse primers, 0.075 μl 50 μM 5′ forward primers, 0.24 μl 25 mM dNTPs, 1.9 μl 10× Buffer, and 16.5 μl water. PCR programs were the same as the first round of PCR. Five microliters of the PCR product was run on an agarose gel to confirm the presence of an ∼500-bp H chain band or 450-bp L chain band. Five microliters from PCR reactions showing the presence of H chain or L chain amplicons was mixed with 2 μl of ExoSAP-IT (Thermo Fisher Scientific) and incubated at 37°C 15 min, followed by 80°C for 15 min to hydrolyze excess primers and nucleotides. Hydrolyzed second-round PCR products were sequenced by Genewiz with the respective reverse primer, and sequences were analyzed using IMGT/V-Quest to identify V, D, and J gene segments. Sequences with more than five mutations or <85% V gene similarity were excluded from Supplemental Table I.

Paired H chain VDJ and L chain VJ sequences that lacked signs of somatic hypermutation from Env-specific or control B cells of unknown specificity were cloned into pTT3-derived expression vectors containing the human IgG1, Igκ, or Igλ constant regions (25) using In-Fusion cloning (Clontech), and sequences were confirmed using Sanger sequencing (Genewiz). Expression vectors were transfected into 293F cells and Abs purified from culture supernatant using protein A columns (Thermo Fisher Scientific) followed by media exchange into 1× DPBS in house or by the Fred Hutch Ab Development Shared Resource.

Bio-layer interferometry (BLI) using the Octet Red instrument (Forte Bio) was performed at room temperature with shaking at 1000 RPM following the manufacturer’s instructions. Briefly, 100 μg/ml of cloned Abs were loaded on anti-human IgG Fc capture biosensors (Forte Bio) for 240 s, followed by a wash step in kinetics buffer (1× DPBS, 0.01% BSA, 0.02% Tween 20, and 0.005% NaN₃ [pH 7.4]) for 60 s. After washing, 1 μM HIV-1 Env was associated for 300 s, followed by dissociation into kinetics buffer for an additional 300 s. The HIV-1 Env–specific Ab VRC01 was used as a positive control in all experiments, and the RSV-specific Ab palivizumab was used as a negative control.

The ZEUS IFA ANA HEp-2 Test System was used following the protocol recommended by the manufacturer (Zeus Scientific) with minor changes. Briefly, HEp-2–coated slides from the kit were incubated with 25 μl of Ab at 0.1 mg/ml for 30 min at room temperature. Slides were then washed three times in 1× DPBS, followed by incubation of 25 μl of 0.01 mg/ml goat anti-human IgG Alexa Fluor 594 (Thermo Fisher Scientific) in 1× DPBS for 30 min at room temperature in the dark. After washing three times with 1× DPBS, slides were coated in 50% glycerol, and coverslips were applied. The EVOS Cell Imaging System (Thermo Fisher Scientific) was used to acquire images, which were analyzed using ImageJ to determine the average Alexa Fluor 594 fluorescence per HEp-2 cell.

Prism software (GraphPad) was used to calculate p values using an unpaired two-tailed t test, a ratio paired two-tailed t test, or one-way ANOVA followed by Dunnett multiple comparison test.

To assess the degree to which cross-reactivity with self-antigen could exist within polyclonal naive B cells specific for foreign Ags, we measured Nur77^eGFP expression in rare naive follicular Ag-specific B cells using previously described Ag-specific enrichment approaches (27–30). In our initial experiments, we analyzed B cells specific for a complex model Ag we called NP650, which contains four NP molecules bound to streptavidin–allophycocyanin and ∼6 molecules of DyLight 650. In Nur77^eGFP transgenic mice, a relatively large population, ∼0.4%, of naive follicular B cells bound NP650, which was increased to ∼18% following enrichment using anti-allophycocyanin microbeads (Fig. 1A, 1B). The binding of NP650 was confirmed to be BCR mediated, because MD4 Rag1^−/− B cells expressing a transgenic BCR specific for hen egg lysozyme did not bind this Ag above the limit of detection even with enrichment (Fig. 1B). Additionally, we confirmed that immunization with NP650 in CFA resulted in activation and differentiation of B cells specific for NP, allophycocyanin and DL650 when these populations were assessed individually (Supplemental Fig. 1).

We next compared the level of Nur77^eGFP in naive follicular B cells that bound NP650 compared with cells that were NP650^NEGATIVE (Fig. 1C). This analysis revealed a roughly 1:1 ratio of Nur77^eGFP geometric mean fluorescence intensities (gMFI) between the two populations (Fig. 1D). These results raised the concern that the enrichment methodology resulted in an upregulation of Nur77^eGFP expression. For the enrichment protocol, fluorescent Ags are bound to BCRs expressed by Ag-specific B cells, followed by magnetic enrichment using microbeads specific for the fluorochrome. Although the labeling and enrichment procedure is carried out largely on ice, it is possible that BCR signaling and subsequent Nur77^eGFP upregulation could occur during short portions of this 2–3 h procedure when the cells are not on ice. Against this notion, Nur77^eGFP was not upregulated in enriched NP650-specific B cells compared with a portion of the sample that was analyzed immediately following NP650 labeling without enrichment (Fig. 1E). In contrast, Nur77^eGFP upregulation was present in NP650-specific B cells if the samples were incubated with NP650 at 37°C for 3 h (Fig. 1D). Nur77^eGFP expression was correlated with the level of NP650 bound by the cells when the samples were incubated at 37°C for 3 h, but not when samples were kept on ice (Fig. 1F, 1G). Together, these data indicate that Nur77^eGFP expression is not altered by Ag-specific enrichment and instead is reflective of in vivo BCR signaling.

We next assessed whether B cells specific for individual components of NP650 expressed either high or low levels of Nur77^eGFP. This did not appear to be the case because the NP-specific and allophycocyanin-specific naive follicular B cells expressed indistinguishable Nur77^eGFP gMFI and robust coefficient of variation (rCV) compared with their counterparts that did not bind these Ags (Fig. 2A–D). These results appear to conflict with the results of a recent study that found low Nur77^eGFP levels in NP-specific B cells using a transgenic H chain paired to endogenously produced λ (Igλ) L chains (18). In agreement with this report, we found that Nur77^eGFP levels were slightly lower in NP-specific naive follicular B cells using Igλ (Fig. 2E, 2F). This did not appear to be unique to NP-specific cells because Nur77^eGFP expression was slightly lower in all naive follicular B cells using Igλ, including those specific for the model Ag DNP (Fig. 2E, 2F). Lower Nur77^eGFP expression in B cells using Igλ likely resulted from slightly lower surface BCR expression compared with their counterparts using Igκ, which was detected when surface levels of CD79b (Igβ) were assessed (Fig. 2G, 2H). Similarly, human Igλ⁺ naive follicular B cells expressed slightly lower CD79b levels compared with their Igκ⁺ counterparts (Supplemental Fig. 2). Differences in Nur77^eGFP expression between B cells expressing Igλ and Igκ were eliminated when normalized to the level of surface BCR (Fig. 2I, 2J).

Overall, the expression of Nur77^eGFP in DNP-specific naive follicular B cells was also similar to the DNP^NEGATIVE population (Fig. 2B–D). Unexpectedly, increased expression of Nur77^eGFP was detected in naive follicular B cells specific for the model Ag R-PE (Fig. 2A, 2C). Previous work has demonstrated that the PE-specific population is large and dominated by a large fraction of cells using V_H1-81 in IgH^b mouse strains such as C57BL/6 (35). The increased expression of Nur77^eGFP appeared linked to the usage of V_H1-81 because the PE-specific population from IgH^a mice expressed similar levels of Nur77^eGFP compared with the allophycocyanin-specific and Ag^NEGATIVE populations (Fig. 2C). Together, these data suggest that a range of cross-reactivity with self-antigens is common within populations of B cells specific for foreign Ags.

We next considered whether these results were applicable to Ags derived from human pathogens. For this, we examined the populations of B cells specific for an influenza HA stem Ag (26), an HIV-1 Env Ag called 426C^TM4ΔV1–V3 (25), and Ags representing the prefusion and postfusion forms of the RSV F protein (24, 36) using tetramer enrichment. We found that the follicular B cells specific for influenza HA stem, HIV-1 Env, RSV prefusion F, and postfusion F expressed a similar level of Nur77^eGFP (Fig. 3). These data suggested that a range of self-antigen cross-reactivity is a common feature among populations of naive B cells specific for pathogen-derived Ags.

B cells expressing BCRs able to bind self-antigens often exhibit reduced responsiveness to BCR stimulation (4–10). Previous work demonstrated that calcium signaling in response to BCR stimulation was reduced in B cells expressing high levels of Nur77^eGFP compared with cells expressing low levels (15). In this study, we assessed whether this reduced calcium flux was mirrored by reduced proliferation of FACS-purified mature follicular B cells expressing high levels of Nur77^eGFP following BCR stimulation with anti-Ig in vitro. For this, we incubated FACS-purified naive B cells expressing high, mid, and low levels of Nur77^eGFP with anti-Ig for 3 d in vitro and measured CTV dilution. In these experiments, we found that 39% of cells expressing high levels of Nur77^eGFP had diluted CTV following stimulation with 2 μg/ml anti-Ig for 3 d in vitro (Fig. 4A–C). In contrast, 61% of B cells expressing low levels of Nur77^eGFP had low CTV levels in response to the same amount of anti-Ig (Fig. 4B, 4C). In contrast, 44% of B cells expressing moderate levels of Nur77^eGFP had low CTV levels in response to the same amount of anti-Ig (Fig. 4B, 4C). These differences in CTV dilution were diminished as the concentration of anti-Ig was increased (Fig. 4C), indicating that strong BCR stimulation could overcome proliferative defects. Similar differences in proliferation were detected when B cells binding NP650 were examined within populations expressing high or low levels of Nur77^eGFP (Fig. 4D). These data suggest that Nur77^eGFP expression is a surrogate for reduced function in B cells specific for foreign Ags.

The modest in vitro response of B cells expressing high levels of Nur77^eGFP suggested that these cells may respond poorly to Ag immunization in vivo. To test this, 1–3 × 10⁶ FACS-purified mature CD45.2⁺ follicular B cells that expressed high, moderate, or low levels of Nur77^eGFP were adoptively transferred into CD45.1⁺ recipients 1 d prior to injection of NP650 in CFA or CFA alone. We used NP650 for these experiments because the frequency of B cells able to bind this combination Ag is large enough to reliably detect hundreds of these cells within the CD45.2⁺ donor population after transfer (Fig. 5A). Seven days after the injection of CFA alone, the differences in Nur77^eGFP expression were maintained within the three groups of NP650-specific cells (Fig. 5B). Following the injection of NP650 in CFA, the majority of donor NP650-specific B cells in all three groups had fully diluted levels of CTV compared with their counterparts in control mice injected with CFA alone (Fig. 5C, 5D), indicating proliferation. However, significantly fewer, 66%, NP650-specific cells that expressed high levels of Nur77^eGFP had low CTV compared with the 79% of donor cells that expressed low levels of Nur77^eGFP (Fig. 5C, 5D). Unexpectedly, NP650-specific B cells that expressed moderate levels of Nur77^eGFP contained the highest frequency of CTV^LOW cells, 88% (Fig. 5D). These data suggest that the expression of moderate levels of Nur77^eGFP corresponds with optimal function.

An increase in the percentage of CTV^LOW cells could be a reflection of a higher number of divisions made by the naive cells that were activated in response to Ag. In these experiments, the responding cells in each group were essentially fully CTV diluted (Fig. 5C), indicating seven or more cell divisions. These data could reflect differences in migration of activated cells out of the spleen and lymph nodes, but the differences in calcium signaling reported previously (15) led us to hypothesize that differences in the frequency of CTV^LOW cells resulted from differences in the number of naive B cells that participated in the response and divided at least once. The frequency of naive B cells that failed to divide can be determined by comparing the frequency of Ag-specific donor B cells that are CTV^HIGH in mice injected with and without Ag (30). Notably, naive CTV^HIGH B cells binding low levels of NP650 were excluded from these analyses because previous experiments revealed that the frequency of these cells was unaltered by NP650 injection (Supplemental Fig. 3). Within recipient animals receiving Nur77^eGFP-high cells, ∼0.19% of donor B cells remained undivided and CTV^HIGH in animals injected with NP650 in CFA, compared with ∼0.26% in animals injected with CFA alone (Fig. 5E). Although this decrease was not significant, it suggested that ∼three-fourths (19/26%) of Nur77^eGFP-high NP650-specific naive B cells fail to proliferate in response to immunization (Fig. 5F). In contrast, only 45% of NP650-specific naive B cells that expressed moderate levels of Nur77^eGFP failed to proliferate (Fig. 5E, 5F). An intermediate frequency, 61%, of NP650-specific naive B cells that expressed low levels of Nur77^eGFP failed to proliferate (Fig. 5E, 5F). Importantly, the average frequencies of naive B cells that failed to proliferate inversely correlated with the average frequencies of CTV diluted cells (Fig. 5G). Together, these data indicate that naive foreign Ag-specific cells expressing moderate levels of Nur77^eGFP are more than twice as likely to respond to Ag immunization compared with their counterparts expressing high levels of Nur77^eGFP.

Having observed functional impairment in vaccine-relevant populations in mice, we sought to gauge the level of self-antigen cross-reactivity within analogous populations of naive human B cells specific for pathogen-derived Ags. Unlike Nur77^eGFP expression in mice, high levels of intracellular Nur77 in human B cells do not appear to correlate with self-antigen reactivity (37). Instead, human naive B cells expressing low levels of IgM have been shown to be enriched for self-antigen–reactive B cells (20, 21), a phenotype commonly exhibited by transgenic B cells specific for self-antigens (38–42). Additionally, both human and murine B cells expressing low levels of IgM have reduced calcium signaling in response to BCR stimulation similar to cells expressing high levels of Nur77^eGFP (15, 20, 21, 43). Overall, naive follicular murine B cells specific for NP650 or HIV-1 Env expressed indistinguishable levels of IgM compared with cells that did not bind these Ags (Fig. 6A, 6B). In agreement with published reports (15), naive B cells of unknown specificity that expressed higher levels of Nur77^eGFP expressed reduced levels of surface IgM (Fig. 6C–E). Similar results were found within murine B cells specific for NP650 or HIV-1 Env (Fig. 6E). Downregulated IgM levels in cells expressing high levels of Nur77^eGFP was mirrored by an overall decrease in surface BCR levels when assessed based upon CD79b staining (Fig. 6F, 6G). Together, these data suggest that self-antigen binding results in Nur77^eGFP upregulation and a corresponding downregulation of IgM BCRs.

Using Ag tetramer-based enrichment, we examined IgM expression in IgD⁺ naive human B cells specific for HIV-1 Env and RSV F prefusion or postfusion F proteins (Fig. 7A–C). Because most people are naturally infected with RSV early in their lives, we excluded memory B cells and focused upon naive B cells by gating upon cells that expressed IgD but not CD27 (Fig. 7A). Cells expressing CD10 were also excluded because transitional B cells circulating in the blood express this protein (44). Using this gating strategy, we found that each Ag-specific population expressed a similar level of IgM compared with their counterparts that did not bind Ag (Fig. 7D, 7E). These features suggest that populations of B cells specific for foreign Ags exhibit cross-reactivity with self-antigens.

To directly analyze BCR cross-reactivity with self-antigens, we next cloned 27 BCRs from HIV-1 Env–specific B cells and assessed binding to the HEp-2 cell line, a common test of self-antigen reactivity (45). The cloned BCRs were selected from a data set of 103 HIV-1 tetramer-binding B cells from human PBMCs in which paired H and L chain sequences were recovered using single-cell RT-PCR (Supplemental Table I). The 27 cloned Abs exhibited no signs of somatic hypermutation, and 24 bound HIV Env above the limit of detection when assessed by BLI (Fig. 8A, 8B). We next assessed self-antigen reactivity of the 24 Env-specific cloned Abs compared with a set of 18 Abs cloned from naive B cells of unknown specificities (Supplemental Table I) along with an Env-specific positive control, 4E10 (46), and the RSV-specific negative control, palivizumab (47). Using this approach, 8.3% of the Env-specific Abs bound to HEp-2 cells above background, which was similar to the 11% of Abs derived from B cells of unknown specificity that also bound (Fig. 8D, 8E). Together, these data indicate that self-antigen cross-reactivity is a common feature among B cells and is composed of a range of affinity within a given population.

Overall, our results demonstrate that B cells specific for foreign Ag display a breadth of self-reactivity and diminished function that may limit vaccine responses. We initially hypothesized that different populations of B cells would have different levels of cross-reactivity with self-antigens based upon the unique properties of the epitopes displayed by each Ag. Although this could still be true, our data suggest that cross-reactivity with self-antigen is generally present among polyclonal BCRs specific for foreign Ags including small haptens, large complex Ags, and even rationally designed immunogens. This suggests that cross-reactivity with self-antigens may have more to do with intrinsic biochemical features of the BCRs themselves rather than similarity between the self- and foreign Ags. There are a number of ways this could occur, for example Abs using V_H4–34 are known to be more self-reactive compared with other alleles (48–50), whereas the unique CDRH3 regions expressed by BCRs using these alleles can mediate binding to diverse sets of foreign Ags. Ab cross-reactivity to structurally distinct self- and foreign Ags could also occur if different amino acids within the H and L chain CDR3s mediated binding to the two Ags.

Uniquely high cross-reactivity with self-antigen within the PE-specific B cell population was surprising. Cross-reactivity with self-antigens was previously reported within the PE-specific memory B cell population, but this was attributed to somatic hypermutation in the germinal center (51). Our data suggest that this self-reactivity was present in the naive population and either not selected against or increased in the germinal center. Interestingly, increased levels of Nur77^eGFP were not detected in PE-specific B cells in IgH^a mice, which lack the V_H1–81 allele used by the majority of the large PE-specific B cell population in IgH^b C57/BL6 mice (35). This could suggest that V_H1–81 exhibits a mild level of unrecognized self-reactivity with self-antigens that maintains a higher frequency of PE-specific B cells through increased positive selection similar to results obtained from studies of cells expressing a transgenic H chain specific for NP (18).

In our analyses we found slightly lower expression of Nur77^eGFP on B cells using Igλ compared with their Igκ counterparts. Although we initially hypothesized that this was the result of Igλ having inherently lower levels of self-antigen reactivity because overall, B cells that express low levels of Nur77^eGFP expressed higher levels of CD79b and IgM. This is likely because binding to self-antigens results in Nur77^eGFP upregulation and subsequent downregulation of surface BCR. However, our data instead suggests that the slightly lower Nur77^eGFP levels in B cells using Igλ are instead a result of slightly lower BCR expression rather than slightly lower levels of self-antigen reactivity.

Our results contrast with studies of CD4⁺ and CD8⁺ T cells showing that increased cross-reactivity with self-peptide bound to MHC class I was associated with higher CD5 expression and enhanced responsiveness (52, 53). Instead, we find that there is a “sweet spot” in that B cells with too little or too much cross-reactivity with self-antigen function poorly. Moderate signaling in response to self-antigen may provide the level of tonic BCR signaling known to be essential for B cell survival (54). We hypothesize that the differences between B and T cells are the result of differences in positive and negative selection. If T cells with higher affinity for self-antigens are more effectively deleted from the repertoire, the remaining CD5^HIGH T cells may represent the sweet spot of reactivity necessary for optimal function. In contrast, mild reactivity with self-antigens appears to increase positive selection of developing B cells (18). The ability of B cells to undergo somatic hypermutation may reduce the need for stringent deletion of cells expressing BCRs with higher affinity for self-antigen. Indeed, naive B cells with high affinity for both foreign and self-antigens are likely the source of “redeemed” Abs that appear to lose self-reactivity as a consequence of somatic hypermutation and affinity in the germinal center (55, 56). Together these data suggest that the benefit of redeeming potentially pathogenic Abs to become protective may outweigh the risks of maintaining B cells cross-reactive with self-antigens. Because TCRs cannot be somatically mutated and redeemed, the benefits of keeping T cells with cross-reactivity to self-peptides/MHC is lower.

During the preparation of this article, it was reported that a population of Nur77^eGFP-high CD4⁺ exhibits reduced function (57) similar to what we and others have observed for Nur77^eGFP-high B cells. These Nur77^eGFP-high T cells appear to compose only a portion of the CD5^HIGH CD4⁺ T cell populations analyzed previously (52), which may have concealed their reduced function. In light of these new data, B and T cells expressing moderate levels of Nur77^eGFP may represent the sweet spot for responsiveness to foreign Ags.

One limitation to our analyses of in vivo responsiveness is that we only measured this for B cells specific for our combination Ag NP650. The reason we did not assess multiple Ag-specific populations is because frequency of individual populations is typically below 0.01% of the total B cell population. Below this level we could not obtain enough donor cells sorted upon Nur77^eGFP expression level to reliably detect naive Ag-specific B cells that failed to respond to immunization. For this reason, we chose to conjugate NP, streptavidin, allophycocyanin, and Dylight 650 to form NP650. Using this combination Ag, we essentially simultaneously probed the response of multiple naive populations together. Therefore, our data showing that superior response of the naive B cells expressing moderate levels of Nur77^eGFP is not confided to a single Ag-specific population but reflective of the larger response targeting all of these Ags. Although it is possible that one Ag-specific population is dominating this response and skewing our results, we think it is more likely that an unusual response from any one population of Ag-specific B cells in this combination would be masked by the response of the other populations.

In our experiments, we also found that naive B cells binding lower levels of foreign Ag were less likely to proliferate in response to immunization. In fact, we could find no evidence that the third of naive B cells binding lowest levels of Ag proliferated using these immunization conditions. These results were not surprising and do not mean that these B cells bind Ag with an affinity below the threshold for activation and are therefore irrelevant. However, it is possible that with increased T cell help, increased Ag availability, Ag multimerization, or different adjuvants could induce these lower binding B cells to participate in a relevant immune response. It will also be interesting to determine whether more low-affinity B cells participate in the response to Ags provided by infection rather than Ag immunization. In fact, some infections are able to stimulate the activation of B cells expressing BCRs with affinities below the limit of detection for most assays (58).

Our results indicate that two key affinities help control which cells respond to Ag exposure. First, the BCR affinity for the injected Ag must be high enough to allow activation. However, overlaid on top of this is the BCR affinity for self-antigens, in which a moderate amount primes cells for optimal responsiveness. Future work is focused upon the elucidation of other factors that tune the functional potential of naive B cells.

We thank P. Culver, R. Putnam, L. Yates, and R. Eberts for administrative support, D. Alwan, E. Naibert, A. MacCamy, and E. Hayes for technical expertise, M. D. Grey, A. T. McGuire, and L. Stamatatos for providing HIV-1 Env, B. Graham for plasmids encoding RSV prefusion and post-fusion F Ags, P. Kwong for plasmids encoding influenza HA stem, and M. J. McElrath for PBMCs from Seattle Area Control cohort.

The authors have no financial conflicts of interest.