Distinct metabolic requirements regulate B cell activation and germinal center responses

Abstract

Following infection or vaccination, activated B cells at extrafollicular sites or within germinal centers (GCs) undergo vigorous clonal proliferation. Proliferating lymphocytes have been shown to undertake lactate dehydrogenase A (LDHA)-dependent aerobic glycolysis; however, the specific role of this metabolic pathway in a B cell transitioning from a naïve to a highly proliferative, activated state remains poorly defined. Here, we deleted LDHA in a stage-specific and cell-specific manner. We find that ablation of LDHA in a naïve B cell did not profoundly affect its ability to undergo a bacterial lipopolysaccharide-induced extrafollicular B cell response. On the other hand, LDHA-deleted naïve B cells had a severe defect in their capacities to form GCs and mount GC-dependent antibody responses. In addition, loss of LDHA in T cells severely compromised B cell-dependent immune responses. Strikingly, when LDHA was deleted in activated, as opposed to naïve, B cells, there were only minimal effects on the GC reaction and in the generation of high-affinity antibodies. These findings strongly suggest that naïve and activated B cells have distinct metabolic requirements that are further regulated by niche and cellular interactions.

Extended Data Fig. 1 |. Confirmation of Ldha deletion in naïve B cells.

a, PCR analysis of genomic DNA from naïve splenic B cells. Two Ldha^fl/flCd23^Cre and two Ldha^+/+Cd23^Cre control mice were analyzed for the presence of the deleted (null), floxed and wild type Ldha alleles. Data is representative of 3 independent experiments. b, Naïve splenic B cells harvested from mice of the indicated genotypes (n = 2 for Ldha^+/+Cd23^Cre, n = 3 for Ldha^fl/flCd23^Cre) or splenic B cells activated with LPS + IL-4 for 48 h (n = 3 for each group) were examined for expression of Ldha by qPCR. Heart tissue harvested from Ldha^+/+Cd23^Cre mice (n = 1) was used as a control. c, Naïve splenic B cells were activated ex vivo with LPS + IL-4 for 72 h and whole cell protein extracts were analyzed by immunoblotting using antibodies against LDHA or α-tubulin (control). Data is representative of three independent experiments. d, Naïve and activated B cells were analyzed for expression of Ldhb by qPCR. n = 3 for each group, except n = 2 for naive Ldha^+/+Cd23^Cre B cells and n = 1 for heart tissue. For panels b and d, data represents mean ± s.d.

Extended Data Fig. 2 |. LDHA governs aerobic glycolysis in B cells.

a, Schematic representation of mito stress assays performed in a Seahorse analyzer. The mito stress assay measures the oxygen flux of cells cultured in glucose-sufficient media supplemented with glutamine and sodium pyruvate. During this assay, the oxygen consumption rate (OCR) is first measured at baseline and after the sequential addition of oligomycin (which inhibits ATP synthase and reduces OCR), FCCP (an uncoupling agent that disrupts the mitochondrial membrane potential without inhibiting the electron transport chain and allows oxygen consumption to reach a maximum), and a mixture of rotenone and antimycin A (AA) (which shuts down mitochondrial respiration). b, Representative graph of OCR of naïve B cells at baseline and following the indicated perturbations. c, Quantification of OCR and of spare respiratory capacity (SRC) of naïve splenic B cells from Ldha^fl/flCd23^Cre and Ldha^+/+Cd23^Cre (control) mice. SRC was calculated as the difference of FCCP-stimulated maximum OCR and basal OCR. d, e, Naïve splenic B cells were activated with LPS + IL4, anti-CD40 + IL4 or anti-CD40 + IL4+anti-IgM for 48 h and OCR and SRC was quantified. p values were calculated using unpaired, two-tailed t-tests. For panels b–e, n = 6 for each indicated genotype and representative of three independent experiments. Data points and bars represent mean ± s.e.m.

Extended Data Fig. 3 |. Cellular homeostasis in LDHA-deficient activated B cells.

a, Uptake of the fluorescent glucose analog 2-NBDG. Naïve B cells (n = 5 for each genotype) were activated ex vivo with LPS + IL4 for 48 h, incubated with 11 mM of 2-NBDG for 15 min, and 2-NBDG uptake was quantified by flow cytometry. Data is representative of two independent experiments. b, Splenic B cells from Ldha^fl/flCd23^Cre (n = 3) and Ldha^+/+Cd23^Cre (control, n = 3) mice were activated ex vivo as indicated for 48 h and intracellular ATP level was determined. Data is representative of two independent experiments. c, d, Splenic B cells from Ldha^fl/flCd23^Cre and Ldha^+/+Cd23^Cre (control) mice were labeled with Cell Trace Violet (CTV) dye, activated ex vivo for 48 h with LPS + IL4, and CTV dilution was determined by flow cytometry as a measure of cell proliferation. Data shown is representative of three independent experiments. d, Splenic B cells (1 × 10⁶) were harvested from mice of the indicated genotypes (n = 7 for each group), stimulated with LPS + IL-4 and cell numbers were quantified at the indicated time points. e, Splenic B cells (1 × 10⁶) harvested from mice of the indicated genotypes (n = 4 for each group) were stimulated with LPS + TGFβ + anti-IgD ex vivo and cell numbers were determined at 96 h post-activation. Data is representative of two independent experiments. Bars represent mean ± s.e.m. *p ≤ 0.05, ****p ≤ 0.0001 by unpaired, two-tailed t-test.

Extended Data Fig. 4 |. B cell subsets at homeostasis.

a, Representative flow plot of splenic transitional T1 (B220⁺CD93⁺CD23⁻IgM^hi), T2 (B220⁺CD93⁺CD23⁺IgM^hi) and T3 (B220⁺CD93⁺CD23⁺IgM^low) B cells. Live singlets from the spleen were first gated for B220⁺CD93⁺ cells and then gated for surface expression of IgM and CD23 as indicated. b, Quantification of T1, T2, and T3 B cells in the spleens of mice of indicated genotypes. c, Representative flow plots of B1a (B220^lowCD19⁺CD5⁺) B cells in the peritoneum of mice. Live singlets from the peritoneal cavity were first gated for B220^lowCD19⁺cells and then gated for surface expression of CD5. d, Quantification of B1a cells in the peritoneum of mice of the indicated genotypes. e, Representative flow plots of marginal zone (MZ) B cells and follicular zone (FO) B cell subsets in the spleen of mice of the indicated genotypes. Live singlets from the spleen were first gated for mature B220⁺CD93⁻B cells and then analyzed for surface expression of CD21 and CD23. f,g, Frequency and absolute number of FO B cells. h, i, Frequency and absolute number of MZ B cells. j, Frequency of CD4⁺ and CD8⁺ T cells (among live singlets) in peripheral blood and k, in the spleen of mice of the indicated genotypes. Each datapoint represents a single mouse. Bars represent mean. *p ≤ 0.05, **p ≤ 0.01, p-values were calculated using unpaired, two-tailed t-test. For panels b, d, f–k, n = 5 of each genotype and data is representative of two independent experiments.

Extended Data Fig. 5 |. Confirmation of Ldha deletion in CD138+ plasmablasts, marginal zone (MZ) and follicular (FO) B cells; IgM ELISA of culture supernatants of ex vivo FO and MZ B cells.

a, Gating strategy for the analysis of B220^lo CD138⁺ cells in the spleens of mice challenged with LPS. b, Representative PCR showing the genomic deletion of the floxed Ldha exon in CD138⁺ plasmablasts sorted from the spleens of Ldha^fl/flCd23^Cre and control mice at d5 after LPS challenge. Tail DNA was used as a control to detect both the null and floxed alleles. Image of ethidium bromide-stained gel shown is representative of three independent experiments. c, Purified FO or MZ B cells of the indicated genotypes (n = 3) were cultured ex vivo for 72 h with LPS + IL4 or LPS respectively and IgM concentration in the culture supernatants was determined by ELISA. Bars represent mean ± s.e.m. d, Representative PCR depicting genomic deletion of Ldha from sorted FO and MZ B cells cultured with LPS. Image of ethidium bromide-stained gel shown is representative of three independent experiments.

Extended Data Fig. 6 |. Gating strategy for analysis of GC B cells following NP-CGG immunization, quantification of GC density and gating strategy for sorting of GL7⁺ and GL7⁻ B cells.

a, Gating strategy for analysis of GC B cells and associated populations in the spleens of NP-CGG immunized mice. b, Mice were immunized intraperitoneally with NP-CGG, boosted at d10 and the GC density in the spleen sections from Ldha^+/+ Cd23^Cre (grey bar) and Ldha^fl/fl Cd23^Cre (red bar) mice was quantified at d14 post-immunization. Each data point represents one independent experiment in which data from the spleen sections of 3 mice per genotype were pooled. Data presented is mean ± s.e.m. c, Detection of pre-GC B cells. Wild type mice were immunized intraperitoneally with NP-CGG or with PBS and the frequency of GL7⁺ cells quantified at d4 post-immunization. d, Gating strategy for the sorting of GL7⁺ and GL7⁻ cells from the spleens of NP-CGG immunized mice at d4 post-immunization. Sorted GL7⁺ and GL7⁻ cells were mixed 3:1 and subjected to scRNA sequencing analysis. Flow cytometry of the post-sort cells is shown.

Extended Data Fig. 7 |. LDHA in T cells regulates GC B cell responses.

Ldha^fl/fl and Ldha^fl/flCd4^Cre mice were immunized intraperitoneally with NP-CGG, boosted at d10, and analyzed at d14. a, Immunofluorescence staining of frozen spleen sections harvested at d14. Image is representative of n = 3 mice per genotype. Scale bar represents 50 μm. b–e, Splenic B cells (5 × 10⁵) from wild type mice were co-cultured with T_FH cells (3 × 10⁵) purified from Ldha^fl/flCd4^Cre (n = 2 pooled samples) or Ldha^fl/fl (n = 2 pooled samples) control mice. The cells were analyzed at d6 post initiation of culture. b, c, Number of B and T_FH cells at d6. d, e, Antibody titers of IgM and IgG1 in culture supernatant. *p ≤ 0.05, by unpaired, two-tailed t-test. Data represents two biological samples, each pooled from 4 mice per genotype. Bars represent either mean ± s.e.m (panels b, c, and e) or mean ± s.d. (panel d).

Extended Data Fig. 8 |. LDHA in T cells is dispensable for LPS-dependent extrafollicular responses.

The spleen of Ldha^fl/flCd4^Cre (n = 8) and Ldha^fl/fl (control, n = 7) mice were analyzed at d5 after LPS administration. a, Staining of Ki67 in activated (GL7⁺) and unactivated (GL7⁻) B cells in mice of indicated genotypes. b, mTORC1 activity was quantified from the frequency of phosphorylated S6 (p-S6) positive cells in activated (GL7⁺) B cells. c, Frequency of CD4⁺ and CD8⁺ T cells among live singlets. d, Ki67 staining of CD4⁺ and CD8⁺ T cells. e, Quantification of naïve (CD62L^hiCD44^lo), central memory (CD62L^hiCD44^hi) and effector (CD62L^loCD44^hi) CD4⁺ and CD8⁺ T cells at d5 post-LPS challenge. Data is representative of three independent experiments. Bars represent mean ± s.e.m.

Extended Data Fig. 9 |. Ldha^fl/flAicda^Cre/+ mice can mount a robust GC response.

Ldha^fl/flAicda^Cre/+ and Ldha^+/+Aicda^Cre/+ (control) mice were immunized intraperitoneally with NP-CGG, boosted at d10, and analyzed at d14. a, Viability of GC B cells as measured by cells negative for Zombie Red viability dye. b, Cell size of viable GC B cells as measured by quantifying the MFI of forward scatter (FSA). c, Mitochondrial mass was quantified by measuring the MFI of MitoTracker^™ Green FM Dye among the GC and non-GC B cells. d, Frequency of dysfunctional mitochondria was measured by quantifying the fraction of live GC and non-GC B cells negative for MitoTracker^™ Red CMXRos. e, PCR of genomic DNA of GC and non-GC B cells sorted from spleens of mice of the indicated genotypes. f, Immunoblotting of whole cell extracts prepared from GC and non-GC B cells sorted from the Peyer’s patches and spleens of mice of the indicated genotypes to detect LDHA and α-tubulin (control) proteins. g, Expression of Ldha and Ldhb in sorted GC B cells. Bars represent mean. Each datapoint represents a single mouse (n = 3). The box plot depicts the first quartile, third quartile, and the median of the dataset. Values that fall within 1.5 times the interquartile range above the third quartile and below the first quartile are represented by the whiskers. The gating strategy for the sorting of GC B cells is the same as that in Extended Data Fig. 6a. For panels a and b, n = 8 for each genotype; data is representative of two independent experiments. For panels c and d, n = 5 for each genotype and data is representative of two independent experiments. For panels e, and f, data is representative of 2 independent experiments with 3 to 5 mice per experiment.

Extended Data Fig. 10 |. Working model depicting metabolic requirements in naïve and in activated B cells.

Proposed model suggesting different metabolic constraints in B cells committed to either a T-dependent or a T-independent pathway for an optimal humoral response. LDHA-mediated glycolysis is essential for the early activation of B cells during a T-dependent GC response but is largely dispensable once the activated B cells are fully engaged in a GC reaction (‘>‘ sign symbolizes ‘greater than’). On the other hand, a T-independent extrafollicular response is not reliant on LDHA-mediated glycolysis.

Fig. 1 |. LDHA governs aerobic glycolysis in B cells.

a, Schematic representation of a glycolysis stress assay performed in a Seahorse analyzer. During this assay, ECAR is first measured when cells are grown in media without glucose, and then measured after the sequential addition of glucose (which is converted into lactate leading to acidification of the media), oligomycin (which inhibits mitochondrial respiration) and 2-deoxyglucose (2-DG; a glucose analog that inhibits glycolysis). b–d, Representative graphs showing ECAR of naïve B cells activated with LPS + IL-4 (b), anti-CD40 + IL-4 (c) or anti-CD40 + IL-4 + anti-IgM (d) for 48 h. Cells were cultured in glucose-deprived Seahorse minimal media for 30–40 min before glycolytic flux analysis. The time point at which glucose, oligomycin and 2-DG are added to the cells is indicated by the dashed green line. e, Quantification of glucose-induced glycolysis in B cells activated ex vivo as indicated. Glycolysis was measured as the difference of maximum ECAR after glucose injection and minimum ECAR after 2-DG treatment. f, Lactate produced by B cells after 48 h of culture with LPS + IL-4. n = 5 for each genotype except for f wherein n = 3 for each group; data were pooled from three independent experiments. Bars represent the mean ± s.e.m.; **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by unpaired, two-tailed t-test.

Fig. 2 |. LDHA is dispensable for lipopolysaccharide-dependent extrafollicular responses.

a, Schematic of experimental strategy. Mice intravenously (i.v.) injected with LPS or PBS were analyzed at d5. b,c, Representative flow cytometry plots (b) and quantification (c) of class-switched IgG3⁺ B cells gated among live singlets in the spleen of LPS-injected Ldha^+/+Cd23^Cre (control, n = 8), Ldha^fl/fl Cd23^Cre (n = 8), Aicda^Cre/Cre (used as an AID knockout control, n = 4) and PBS-injected (n = 4) mice. Data are representative of three independent experiments. d,e, Representative flow cytometry plots (d) and quantification (e) of plasmablasts (B220^loCD138^hi) among live singlets in Ldha^+/+Cd23^Cre (n = 8), Ldha^fl/fl Cd23^Cre (n = 8) and PBS-injected (n = 4) mice. Data are representative of three independent experiments. f, Quantification of cell proliferation in activated (GL7⁺) and unactivated (GL7⁻) B cells as assessed by Ki67 staining (n = 8 for each group). g, mTORC1 activity quantified as the frequency of phosphorylated S6 (p-S6) in activated (GL7⁺) and unactivated (GL7⁻) B cells. Data are representative of three independent experiments, with n = 8 for each group. h, Concentrations of IgM and IgG3 antibodies at d5 after LPS administration. For LPS-injected animals, n = 8 for each group, except for Ldha^fl/flCd23^Cre mice that were examined for IgG3 antibodies, n = 7; PBS-injected mice, n = 4. Bars represent the mean in h and the mean ± s.e.m. for all other graphs. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by unpaired, two-tailed t-test.

Fig. 3 |. Loss of LDHA in naïve B cells impairs germinal center responses.

a–c, GC response at homeostasis. a, Representative flow plots (a) and quantification (b) of GC B cell frequencies in the PPs of Ldha^+/+Cd23^Cre (n = 7), Ldha^+/flCd23^Cre (n = 6) and Ldha^fl/flCd23^Cre (n = 6) mice. c, Serum antibody titers of IgM, IgG1 and IgA isotypes in Ldha^+/+Cd23^Cre (n = 6) and Ldha^fl/flCd23^Cre (n = 6, except for serum IgG1 measurement in which n = 7) mice at homeostasis. d,e, GC response following NP-CGG immunization. d, Schematic of NP-CGG immunization. Mice were immunized intraperitoneally (i.p.) with NP-CGG, boosted at d10 and analyzed at d14. e, The frequency (n = 7 for Ldha^+/+Cd23^Cre and n = 8 for Ldha^fl/flCd23^Cre) and absolute number (n = 6 for Ldha^+/+Cd23^Cre and n = 8 for Ldha^fl/flCd23^Cre) of GC B cells were quantified in the spleens of immunized mice of the indicated genotypes at d14. f, Representative immunofluorescence staining and quantification of GC area in spleen sections from mice of the indicated genotypes at d14 after immunization. Data are representative of three mice of each genotype. Scale bar, 50 μm. g, Relative titers of high-affinity (NP₈) and all-affinity (NP₃₀) anti-NP IgM antibodies at d14 after immunization (n = 7 for Ldha^+/+Cd23^Cre and n = 8 for Ldha^fl/flCd23^Cre mice). h, Relative titers of high-affinity (anti-NP₈) and all-affinity (anti-NP₃₀) anti-NP IgG1 antibodies at d14 after immunization (n = 8 for mice of each genotype). i, Bone marrow from wild-type (WT; CD45.1) or Ldha^fl/flCd23^Cre mice (CD45.2) were mixed at a 1:1 ratio, transplanted into γ-irradiated RAG2-deficient recipient mice. Following 8 weeks of reconstitution, mice were immunized with NP-CGG as described above and analyzed. Reconstitution experiments with CD45.1 and CD45.2 wild-type bone marrow cells mixed at a 1:1 ratio served as controls. Representative flow plots and quantification of congenically marked GC B cells (GL7⁺Fas⁺ of live B220⁺ B cells) in the spleens of immunized mice are shown. n = 3 for CD45.1 and n = 5 for CD45.2 mice. Bars represent the mean ± s.e.m. for all the graphs except in c, g and h where bars indicate the mean; **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by unpaired, two-tailed t-test and by two-way analysis of variance (ANOVA) for i. a.u., arbitrary units.

Fig. 4 |. LDHA is required for pre-germinal center B cell proliferation.

a–d, Analysis of pre-GC B cells antibody responses by single-dose immunization of NP-CGG measuring serum titers at d0 (pre-immunization), d4, d7 and d10. The green arrow indicates the time point for immunization. n = 4 animals for each indicated genotype. e–j, Single-cell transcriptome and BCR mutation analysis of pre-GC B cells. Mice (Ldha^+/+Cd23^Cre or Ldha^fl/flCd23^Cre) were immunized with NP-CGG and at d4 after immunization. GL7⁺ (activated) and GL⁻ (naïve) B cells from the spleen of indicated genotypes were enriched, mixed at a 3:1 ratio and subjected to single-cell BCR and transcriptome analysis. e, Uniform manifold approximation and projection (UMAP) of B cells from mice of the indicated genotypes. Of the four clusters, only the proliferating cluster, defined by the expression of Mki67 and Top2a, was markedly reduced in Ldha^fl/flCd23^Cre mice. RP, ribosomal protein. f,g, UMAP analysis of Bcl6 expression in pre-GC B cells of Ldha^fl/flCd23^Cre and control mice. g, Violin plots for Bcl6 expression in proliferating and non-proliferating pre-GC B cells. h,i, UMAP analysis of Ccr6 expression in pre-GC B cells of Ldha^fl/flCd23^Cre and control mice. i, Violin plots for Ccr6 expression in proliferating and non-proliferating pre-GC B cells. j, Total number of mutations in the immunoglobulin heavy chain (IgH) of pre-GC B cells. For a–d, data points represent the mean ± s.d. **P ≤ 0.01, ***P ≤ 0.001, by two-way ANOVA. NS, not significant.

Fig. 5 |. LDHA in T cells regulates germinal B cell responses.

a–g, Ldha^fl/fl and Ldha^fl/flCd4^Cre mice were immunized intraperitoneally with NP-CGG, boosted at d10, and analyzed at d14. a,b, Frequency (a) and absolute number (b) of T_FH cells (PD1⁺CXCR5⁺ gated among live singlet B220⁻CD3⁺CD4⁺CD44⁺ cells) in the spleens of immunized mice. c,d, Frequency (c) and absolute number (d) of GC B cells in the spleens of immunized mice. e, Frequency of NP⁺ B cells within the GC population. f, Relative titers of high-affinity (NP₈) and all-affinity (NP₃₀) anti-NP IgM antibodies at d14 after immunization. g, Relative titers of high-affinity (anti-NP₈) and all-affinity (anti-NP₃₀) anti-NP IgG1 antibodies at d14 after immunization. h–j, Ldha^fl/fl, Ldha^fl/flCd4^Cre and Aicda^Cre/Cre (AID knockout) mice were intravenously injected with LPS or PBS (control) and analyzed at d5. h, Representative flow plots and quantification showing IgG3⁺ B cells. i, Representative flow plots and quantification of plasmablasts (B220^loCD138^hi) in the spleens of LPS-challenged mice. j, Absolute serum titers of IgM and IgG3 at d5 after LPS administration. Bars represent the mean for f, g and j and the mean ± s.e.m. for all other graphs. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by unpaired, two-tailed t-test. In a–e, n = 7 for Ldha^fl/fl and n = 9 for Ldha^fl/flCd4^Cre mice; in f and g, n = 8 for Ldha^fl/fl and n = 10 for Ldha^fl/flCd4^Cre mice, although some of these datapoints fell below the detection limit of the assay and thus were not plotted (4 for anti-NP₈ IgM and IgG1 titer measurements, and 2 for anti-NP₃₀ IgG1 titer measurement); in h and i, n = 7 Ldha^fl/fl and n = 8 for Ldha^fl/flCd4^Cre mice; in j, n = 5 for each group (IgM antibody titer) and n = 7 for each group (IgG3 antibody titer). All data are representative of three independent experiments.

Fig. 6 |. LDHA is largely dispensable for germinal center responses after B cell activation.

a,b, PPs of Ldha^+/+Aicda^Cre/+ and Ldha^fl/flAicda^Cre/+ mice. a, Representative flow plots of GC B cells (GL7⁺Fas⁺ of live B220⁺ B cells). b, Frequency of GC B cells and frequency of IgA⁺ B cells within the GC population (n = 8 for each group). Data are representative of two independent experiments. c, Concentration of IgM, IgG1 and IgA antibodies in the sera of Ldha^+/+Aicda^Cre/+ (n = 5) and Ldha^fl/flAicda^Cre/+ (n = 6) mice at homeostasis. d–h, Mice of the indicated genotypes (n = 8 animals per group) were immunized with NP-CGG, boosted at d10 and analyzed at d14. d,e, Frequency (d) and absolute number (e) of GC B cells in the spleens of immunized mice. f, Frequency of NP⁺ B cells among the GC B cell population in the spleens of immunized mice. g, Frequencies of DZ (CXCR4^hi CD86^lo of GC B cells) and LZ (CXCR4^lo CD86^hi of GC B cells) GC B cells. h, Frequency of IgG1⁺ class-switched B cells in the splenic GCs of immunized mice. Data are representative of two independent experiments.

Fig. 7 |. Activated B cells do not rely on LDHA for affinity maturation.

Mice of the indicated phenotypes were immunized with NP-CGG, boosted at d10 and analyzed at d14. a,b, Cell-cycle analysis of GC B cells. a, Representative flow plots of BrdU incorporation versus total DNA stained with 7-AAD, gated on GC B cells. b, Quantification of the fractions of GC B cells at the different stages of the cell cycle (n = 5). Data are representative of two independent experiments. c,d, Measurement of all-affinity (anti-NP₃₀) (c) and high-affinity (anti-NP₈) (d) anti-NP IgM antibodies at indicated time points. e,f, Measurement of all-affinity (anti-NP₃₀) (e) and high-affinity (anti-NP₈) (f) anti-NP IgG1 antibodies at the indicated time points. Data for c–f are representative of two independent experiments (n = 5 for Ldha^+/+Aicda^Cre/+, n = 6 for Ldha^fl/flAicda^Cre/+ mice). g, Mutations detected by JH4 intron sequencing in GC B cells sorted from Ldha^fl/fl Aicda^Cre/+ (n = 6) and Ldha^+/+Aicda^Cre/+ (n = 5) mice. Data are representative of two independent experiments. Bars represent the mean and mean ± s.e.m, wherever applicable. **P ≤ 0.01, by unpaired, two-tailed t-test. ND, not detectable. In the box plot (g), the top and bottom edges of the box correspond to the first and third quartiles, the middle line denotes the median and the whiskers represent the largest and smallest values no greater than 1.5 times the interquartile range. Outliers are plotted as individual points outside the boundary of the whiskers.