Glutamine and glutaminolysis promote antibody response to ovalbumin.

(a) Reduced glutamine concentration in LN relative to the bloodstream. Shown are the results from measuring concentrations glutamine, leucine, and pyruvate in the centrifugal eluates (67, 68) of lymph nodes or plasma of mice (n=3), as described in the Methods. (b, c) B cells from WT B6 mice were activated and cultured together anti-IgM, BAFF, IL4, and IL5 in the presence of the indicated concentrations of Gln. Shown are the mean (±SEM) frequencies of CD138+ B220lo PC (b) and IgG1+ B cells (c) after culture for 5 days. (d, e) Effects of cell-permeable α-ketoglutarate precursor, dimethyl-α-ketoglutarate (DMK), on the impaired PC differentiation and IgG class switching caused by Gln restriction. CellTrace Violet-labeled B cells were activated and cultured as in (b), with addition of supplemental DMK as indicated. Shown are mean (±SEM) frequencies of % CD138+ (d) and IgG1+ B cells (e) within each division-counted peak. (b-e) Results aggregated from three biological replicate experiments. P values were calculated by Mann-Whitney U test. *p<0.05. Related data are presented in supplement 1 to this Figure. (f-k) An anti-ovalbumin Ab response is promoted by B cell expression of GLS. (f) Schematic of the immunization with priming and sensitization of tamoxifen-treated mice of the indicated genotypes [huCD20-CreERT2+ and either Gls +/+ (WT) or Gls f/f], followed at week three by challenges with intranasal instillations of sterile ovalbumin solution. (g) Mediastinal LN were collected from harvested mice and the frequencies of IgG1+ events in the GL7+ CD95+ B cell gate were measured by flow cytometry. Each dot represents an individual mouse, with bars denoting the mean values. (h-k) Single-cell suspensions of lung and bone marrow were analyzed by ELISpot assays with ovalbumin-coated filters to capture secreted Ag-specific Ab detected using anti-mouse IgG1. Shown are (h) representative wells from the indicated sources (organ; genotype of B cells) and aggregated frequencies of anti-ova IgG1+ ASCs in lung (i) and marrow (j). (k) Anti-ova IgG1 in sera of the mice with B cell-restricted depletion of GLS (Gls iB-Δ/Δ) or controls, as indicated. Shown are data from ELISA with serial 4-fold dilutions using individual samples from each mouse. The likelihood of each null hypothesis (no true difference between two samples) was calculated as noted in the Methods (two-tailed, and non-parametric testing where conditions not met for Student’s t-test). *p<0.05, **p<0.01, *** p,0.001, ****p<0.0001.

Glutaminolysis only conditionally supports the anti-NP Ab response.

(a) Schematic of the immunization with priming sensitization and inhaled challenge of tamoxifen-treated mice of the indicated genotypes [huCD20-CreERT2+ and the indicated combinations of either Gls +/+ (WT) or Gls f/f and either Slc2a1 +/+ (WT) or Slc2a1 f/f], followed at week three by challenges with intranasal instillations of sterile ovalbumin solution. (b-f) ELISpot analyses of cells secreting IgG1 anti-ovalbumin (b, c) and anti-NP (d-f) Ab in mice of the indicated genotypes. Shown are counts from the lung suspensions (b), bone marrow (c, d, f) and spleens (d, e) from each individual mouse, with means (±SEMs) for the aggregate data denoted as bar graphs. (d) Representative ELISpot wells scoring the frequencies of IgG1 anti-NP Ab-secreting cells in spleen and marrow of mice whose B cells were converted to the indicated genotypes [wildtype (), GlsΔ/Δ (), Slc2a1Δ/Δ (), and GlsΔ/ΔSlc2a1Δ/Δ ()] after immunization and rechallenge. (e, f) Aggregrate data with ASC counts for IgG1 anti-NP-secreting cells in spleen (e) and marrow (f). Data were pooled from two independent experiments (seven wildtype, six GlsΔ/Δ, four Slc2a1Δ/Δ, and six GlsΔ/ΔSlc2a1Δ/Δ mice. * p < 0.05, ** p < 0.01, *** p < 0.001 (Mann-Whitney U test).(g-i). Shown are the mean (±SEM) absorbances of aggregated results of ELISA analyses of the IgM (g) and IgG1 (h, i) responses in the immunized mice of the two independent experiments, comparing all-affinity (h) to high-affinity (i) anti-NP, using serial 3-fold or 4-fold dilutions when detecting IgM and IgG1, respectively.

Synthetic auxotrophy - glutaminase support of anti-NP response is dependent on mitochondrial pyruvate channel subunit 2.

(a) Schematic of the immunization with priming and a recall boost of mice of the indicated genotypes [huCD20-CreERT2+ and the indicated combinations of either Gls +/+ (WT) or Gls f/f and either Mpc2 +/+ (WT) or Mpc2 f/f], treated with tamoxifen prior to the initial immunization. (b, c) Results of serological analyses performed on sera from the time of harvest. Shown are the mean (±SEM) absorbances measured in ELISA for detection of all- and high-affinity Abs (α-NP20 and α-NP2, respectively) of the IgG1 (b) and the relative strengths of affinity maturation (c). Impacts of altered metabolic pathways in B cells on relative extents of affinity maturation in the recall responses are shown as mean ratios of absorbances for high- (NP2) / all-affinity (NP20) ELISA at two serum dilutions for the IgG1 isotype. The analogous serologic measurements of IgM and IgG2c anti-NP Ab are shown in Fig 3 - supplement 1. (d, e) Altered glutaminolysis and pyruvate import collaborate in support of the response to primary immunization as well as the Ab after a secondary immunization (’boost’). All- and high-affinity anti-NP IgG1 was measured in the sera of mice whose B cells were converted to the indicated genotypes by tamoxifen injections and then immunized as in (a). Shown are mean (±SEM) absorbances for (d) all- (αNP20) and high- (αNP2) IgG1 at a single dilution of sera before (’Pre-Boost’) and a week after (’Boost’) the second immunization. Data represent three temporally distinct cohorts, each including two mice of each of the four genotypes. Probabilities of the null hypothesis applying (P values) for differences in pairwise comparisons are indicated. (f) Shown are mean (±SEM) frequencies of splenocytes producing IgG1 anti-NP Ab of the indicated affinities at harvest after the booster immunization as in (a), with genotypes as shown and each dot representing an individual mouse. Data on IgG2c and IgM isotypes are presented in Fig 3 - supplement 2. (b-f) Results are aggregated from four temporally separate immunization experiments with mice of each genotype, totaling eight individual controls [(WT, i.e., only CreERT2+) )] and eight mice with each type of induced B cell type-specific deletion [GlsΔ/Δ (), Mpc2Δ/Δ (), and GlsΔ/Δ, Mpc2Δ/Δ () genotypes]. Related and additional results - affinity maturation (NP2/NP20 ratios), frequencies of GC B cells and the NP-binding population therein, MBCs, BrdU uptake rates in GC B cells, as well as Gls and Mpc2 mRNA - are presented in Fig 3 - supplements 1-3. (g, h) Functional requirements for GLS and MPC2 in B cells in a recall phase after secondary immunization. (g) Schematic of the priming immunization without deletion of the conditional alleles, followed by tamoxifen injections ∼ 3 wk later, and only then a recall boost of mice of the indicated genotypes [huCD20-CreERT2+ and the indicated combinations of either Gls +/+ (WT) or Gls f/f and either Mpc2 +/+ (WT) or Mpc2 f/f]. (h) Shown are serologies of the all- and high-affinity IgG1 anti-NP Ab elicited by a boost when Cre activation was initiated only after the primary response. Results are aggregated from three temporally separate immunization experiments with mice of each genotype, totaling nine individual controls [(WT, i.e., only CreERT2+) ()] and nine mice with each type of induced B cell-specific deletion [GlsΔ/Δ (), Mpc2Δ/Δ (), and GlsΔ/Δ, Mpc2Δ/Δ () genotypes]. Fig 3 - supplements 4 and 5 show additional results, with IgM, IgG2c and IgA isotypes; frequencies of ASCs, total and NP-binding GC B cells as well as Tfh / GC-Tfh cells.

GLS and MPC2 collaborate in supporting progression to plasma cell development.

(a) B cells were activated and cultured under conditions promoting plasma cell differentiation in the presence of the indicated combinations of vehicle (DMSO) or inhibitors of GLS (CD839) and the MPC (UK5099). Shown are representative histograms from flow cytometric analyses of CD138 within the live cell gate, with inset numbers denoting the %CD138+. The bar graph shows the mean (±SEM) results for generation of CD138+ cells under each treatment condition, pooling data from three temporally independent experiments, each with 3-5 independent B cell pools purified from separate mice (each dot represents a distinct sample). (b) Shown are the mean (±SEM) calculated numbers of PC generated in temporally independent replica experiments with a total of eight independent B cell pools cultured in vitro in (a). (c) Representative ELISpot results measuring the frequencies of IgM- and IgG1-secreting PC, as indicated, produced in the differentiation cultures under each treatment condition. (d) Bar graphs with mean (±SEM) ELISpot results pooled from the replicate experiments illustrated in (c), with each dot representing an individual sample. Shown are data normalized to the vehicle (DMSO) control for each set of cultures using an individual B cell pool. (e) Bar graphs show the mean (±SEM) absorbance values from ELISA measurements of IgM and IgG1 secreted into the media during the cultures as in (b). Additional data quantifying ASCs are presented in Supplemental Fig. 4. (f) Prdm1 gene expression promoted by GLS and MPC2. B cells were activated and cultured as above, but harvested after 3.5 d culture in BAFF, IL-4, IL-5, and the indicated inhibitor(s) or vehicle followed by qRT2-PCR to quantitate Prdm1 RNA encoding Blimp1. Shown are the results from four biologically independent mouse pools, B cell purifications and cultures, with the Prdm1-encoded RNA then normalized in each experiment to the level in the vehicle (DMSO) control (in each sample, relative to the averaged CT values of cyclophilin A and GAPDH). (g, h) Global gene expression identifies plasma cell identity as a main target of synthetic auxotrophy. Using three biologically independent replicate pools for each condition, RNA-seq was performed with the B cells cultured as in (f). Enriched genes identified by DESeq2 comparison were analyzed using the MyGeneset tool from ImmGen. (g) Genes enriched in vehicle treated cultures compared to cultures treated with both CB839 and UK5099 are shown as a W-plot with defined stages for mature B cells and PC indicated. (h) Genes enriched in CB839-treated cultures compared to cultures treated with both CB839 and UK5099 are shown as a heatmap of z-scored relative expression, with specific gene identities and defined stages for mature B cells and PC indicated. (i) Metabolic mitigation of the block imposed by synthetic auxotrophy. Graphs of aggregate results from six biologically independent B cell preparations (two biological replicates in each of three independent experiments), presented as in (a), are shown for differentiation assays performed with B cells purified, activated, and cultured as in (a), except that the cell permeable αKG analogue DMK was added as indicated. (j-l) Gene set enrichment analyses (GSEAs) were performed on RNA-seq data generated using RNA from flow-purified GC B cells and hallmark gene sets of the Mouse Molecular Signatures Database. CreERT2 -transgenic mice of each of the four genotypes (Gls +/+ or f/f; Mpc2 +/+ or f/f) were immunized with SRBC after being treated sequentially with tamoxifen to activate huCD20-CreERT2 as diagrammed in Fig 3a. One week after immunization, RNA was isolated from viable GC-phenotype B cells purified by flow sorting, as well as from IgD+ B cells. RNA-seq data were quality-controlled, processed, and organized into GSEAs as described in the Methods. Shown is a selected subset of analyses with high normalized enrichment scores (NES) for the indicated gene sets (j) oxidative phosphorylation (WT vs Gls Δ/Δ, Mpc2 Δ/Δ GC B cells); (k) regulated by c-Myc (Mpc2 Δ/Δ vs Gls Δ/Δ, Mpc2 Δ/Δ GC B cells); (l) oxidative phosphorylation (Mpc2 Δ/Δ vs Gls Δ/Δ, Mpc2 Δ/Δ GC B cells). Additional GSEA and other data are in Fig 4 - supplement 5).

Synthetic auxotrophy of B cell metabolism that supports a progressive post-activation increase in mitochondrial respiration.

(a) Pools of purified B cells from WT mice or those with the indicated gene-targeted loss(es) of function were activated and cultured (1 and 2 d) with αCD40, BAFF, IL-4, and IL5. Metabolic functions were then assayed by a metabolic flux analyzer. (b-f) As for (a), except WT cells were treated with inhibitors (CB839; UK5099) alone or in combination, as indicated by the color coding, and then subjected to mitochondrial stress-tested measurements of respiration (b), biochemical assays of [ATP] (g-i), flow cytometry (j, l) or qPCR (k). (a) Oxygen consumption rates (OCR) during mitochondrial stress testing of B cells, comparing loss-of-function B cells of the indicated genotypes, color-coded as in Fig. 3. (b-d) Changes in basal respiration and maximal respiration of B cells from day 1 to day 2 after activation (b), calculated from assays in (c) and (d) OCR values at day 2 were used for statistical analysis. (c, d) As in (a) except that purified WT B cells were used, with additions of vehicle (DMSO) or the indicated inhibitor(s) (CB839, 1 µM; UK5099, 10 µm), with each B cell pool assayed on both days 1 (c) and 2 (d) after purification and activation. (e) Extra-cellular acidification rates (ECAR) during glycolytic stress tests of WT B cells activated and cultured in the presence of the indicated agents after 2 d cultures as in (d). (f) ECAR during glycolytic stress test of B cells inducibly rendered Gls Δ/Δ and/or Mpc2 Δ/Δ, then activated and cultured as in (a). (g, h, i) Intracellular [ATP] in lysates of B cells activated and cultured (2 d) as in (c). (g) Metabolic inhibitor(s) were present throughout the period of culture (2 d) and assay. (h) Cells were activated and initially cultured in the presence of the indicated metabolic inhibitor(s), then washed, and assayed (90 min) in medium without inhibitors. (i) After activation and 2 d culture with no inhibitor present, the indicated agent(s) were added to block glutaminolysis and/or mitochondrial pyruvate import during the 90-minute assay. (j) Mitochondrial membrane potential determined by tetramethylrhodamine (TMRE) staining analyzed by flow cytometry. Shown are mean fluorescence intensity (MFI) values from each independent experiment after activation and culture (2 d) as in (c), then normalized to DMSO-treated condition in each experiment. (k) Flow cytometry results comparing inhibitor-treated cells vs controls after staining for ROS with DCFDA in three independent replication experiments, with each dot representing one experiment, normalizing as in (i).

Metabolism in B cells promotes B cell proliferation via IL-21 signaling to STAT3.

(a) Enrichment of the IL6-stimulated Jak-STAT3-induced gene set. (b) Immunoblot analyses of IL-21-induced tyrosine phosphorylation of STAT3 in B cell blasts, showing representative results from one individual experiment representative of three independent replications. Purified B cells were activated with anti-CD40 and BAFF, cultured for 16 hr in the presence of vehicle (DMSO) or inhibitors (CB839 and UK5099), as indicated, then stimulated (15 min) with IL-21. (c, d) CB839 and UK4099 attenuates IL-21-induced B cell proliferation. B cells were stained with CTV, stimulated with anti-CD40 and BAFF with and without CB839 and UK5099 for 3 days in the presence or absence of IL-21. (x = 4; n = 8). (c) Shown are the representative flow plot of CTV partitioning in viable B cell gates. (d) Aggregated mean (±SEM) frequencies of IL-21-induced proliferation of B cells calculated by subtracting CTVlow frequencies of without IL-21 from with IL-21. P values were calculated by Mann-Whitney U test. (e, f) Expression of a STAT3-gain of function mutant reverses the impact of GLS and MPC2 deficiency on B cell proliferation. WT and GlsΔ/Δ; Mpc2 Δ/Δ (dKO) B cells purified from tamoxifen-injected huCD20-CreERT2 mice were transduced with replication-deficient virions of either the MIT retrovector or MIT-STAT3-Y640F (99), as indicated, stained with CTV, stimulated with anti-CD40, anti-IgM, BAFF, IL-4, and IL-5, and cultured 5 days. Shown are the flow-cytometry results of a representative analysis of CTV partitioning in Thy1.2-positive transduced cell gates (e), and mean (±SEM) frequencies of B cells that divided ≥ 3 times, aggregating data from three biologically independent replicate experiments (f), comprising five separate pools each of WT and dKO samples. P values were calculated by Mann-Whitney U test.

Metabolic programming in B cells modulates interferon signaling to STAT1.

(a) Results from an over-representation analysis of differentially expressed protein-coding RNA in WT versus induced double-deficient (GLS; MPC2; “diKOB”) B cells, as counted in the RNA-seq analyses with GC B cells of SRBC-immunized mice (as in Fig 4, j-l). Additional GSEA and an overview of gene set comparisons for pairs of genotypes are presented in Fig 4 – supplement 5. (b) Selected analyses of gene sets enriched in WT GCBs compared to their counterparts subjected to disruption of both Gls and Mpc2 (diKOiB GCBs) using the Hallmark Pathway database. Shown are enrichment of IFN-α- and IFN-γ-associated pathways in WT GCBs compared to the Gls Δ/Δ, Mpc2 Δ/Δ samples. (c-h) Immunoblot analyses of IFN-induced STAT1 phosphorylation in activated B cells, showing representative results from individual experiments, each representative of three independent replications. (c, d) Purified B cells were activated with anti-CD40 and BAFF and cultured for 64 or 16 hr in the presence of vehicle (DMSO) or inhibitors (CB839 and UK5099), as indicated, then stimulated (15 min) with IFN-β (c) or IFN-γ (d) followed by immunoblotting using Ab specific for p-STAT1(Y701) or p-STAT1α(S727) along with anti-cyclophilin B Ab as a loading control. (e, f) Inhibition of mitochondrial ETC attenuates STAT1 activation. B cells were activated and cultured as in e and f but for 16 hr, in the presence or absence of metformin (2 mM; 16 hr) or rotenone (0.5 µM; final 2 h of the cultures) as indicated, then stimulated (15 min) with IFN-β (g) or IFN-γ (h) and analyzed as for panels c, d. (g, h) ROS inhibit STAT1 activation. B cells were activated and cultured for 16 hr, in the presence or absence of menadione (2 µM; all 16 h) or H2O2 (100 µM; the final 2 h), and then stimulated with IFN-β (g) or IFN-γ (h) for 15 min.

Glutamine and glutaminolysis promote antibody response to ovalbumin.

Additional data from or relating to experiments in Figure 1. (a) Mouse B cells were activated and cultured together BAFF, LPS, IL4, and IL5 in the presence of the indicated concentrations of Gln. Shown are the mean (±SEM) frequencies of PC (CD138+ B220lo; left panel) and IgG1+ B cells (right panel) after culture (4 d) and flow cytometry, averaging ≥ 3 independent replications. (b) Results of measuring IgM and IgG1 by ELISA. Using a dilution in the linear range for the samples, relative concentrations of IgM (left panel) and IgG1 (right panel) in the supernatants of cultures of Fig. 1b were measured by ELISA, using titrations of re-added glutamine concentrations as indicated. Shown are the mean (±SEM) values from analysis of supernatants in three biologically independent experiments, with indications of P values as in the main figure.(c) Reduction of PC differentiation and class switching caused by Gln restriction manifested at equal division numbers, and mitigated by a cell-permeable analogue of α-ketoglutarate. After labeling with CellTrace Violet, B cells were activated and cultured as in (a, b), except that dimethyl-ketoglutarate (DMK) was added to one of two cultures at 0.1 mM glutamine, as indicated. Shown are mean (±SEM) frequencies of % CD138+ (upper panel) and IgG1+ B cells (lower panel) within each division-counted peak. (e) Distributions of division counts under conditions of lower extracellular glutamine in experiments of panel c. Shown are the percentages of live cells in the gate for each peak of 2-fold CTV partitioning when grown under the indicated conditions. (e) Distributions of division counts under conditions of lower extracellular glutamine, with or without addition of dimethylketoglutarate (DMK) as indicated, in experiments of panel c. (e, f) Flow cytometric resolution of seven divisions. (f) A representative flow cytometry result from CTV partitioning experiments to measure conversion of B cells to CD138+ plasmablast / plasma cells as a function of division number. Shown are events after gating for viable cells (FSC x SSC, 7AAD exclusion). (g) A representative panel of results gated on CD19+ B lineage cells, generated after CTV labeling followed by mitogenic activation and culture as in panel c.

Glutamine and glutaminolysis promote antibody response to ovalbumin.

(a) Reduced Gls (alternatively labeled as Gls1) gene expression in B cells purified after tamoxifen treatments of huCD20-CreERT2, Glsf/f mice and activated ex vivo and cultured 2 d. Shown are representative data from imaging immunoblots probed with anti-GLS and anti-actin Ab. (b) Efficient depletion of GLS from B cells of tamoxifen-treated huCD20-CreERT2 mice that were either Gls+/+, Mpc2+/+ or Glsf/f, Mpc2f/f mice, as indicated. Using two separate mice, proteins resolved by SDS-PAGE with whole cell lysates of purified B cells were probed with Ab specific for GLS and cyclophilin B as indicated. (c) Relative levels of ovalbumin-specific IgM in experiments of Fig. 1f-k, showing mean (±SEM) absorbances in ELISA data across serial dilutions, as in Fig. 1k.

Synthetic auxotrophy - glutaminase support of anti-NP response is dependent on mitochondrial pyruvate channel subunit 2

(a-c) Using sera from the immunized mice analyzed in Fig 3a-c, relative concentrations of all- (α-NP20) and high- (α-NP2) affinity IgG2c (a) and IgM (b) were measured by ELISA, from which affinity maturation as a ratio of NP2/NP20 was determined for each subject (c). Shown are the means (±SEM) of aggregated data from the eight mice of each genotype, as well as indications of results from statistical testing (*, p<0.05; **, p<0.01). (d, e) Using results from experiments of Fig 3d, e, affinity maturation of IgG1 anti-NP Ab (d) and the capacity to increase their concentration upon secondary immunization (“boost effect”, i.e., for each subject, the difference between the pre-and post-boost absorbance at 1:100,000 dilution.

Synthetic auxotrophy - glutaminase support of anti-NP response is dependent on mitochondrial pyruvate channel subunit 2

(a, b) Using the immunized and boosted mice of Fig 3a-c, splenic ACSs were quantitated by ELISpot assays capturing all- and high-affinity anti-NP IgG2c (a) and all-affinity IgM. Shown are the means (±SEM) of aggregated data from the eight mice of each genotype, as well as indications of results from statistical testing (*, p<0.05). (b) Metabolic regulation of GC B cell cycling. Mice whose B cells were converted to the indicated genotypes by tamoxifen injections were immunized with SRBC, received intravenous BrdU (16 h and 4 h before harvest) one week after immunization with SRBC, followed by harvest and flow cytometric analyses. Shown are the mean (±SEM) frequencies of BrdU+ GC-phenotype (GL7+ CD38neg) B cells, aggregating results from three temporally independent experiments, each involving two mice of each the four genotypes as indicated. (d) Shown are the aggregated data scoring the frequencies of splenic GC-phenotype (GL7+ CD95+ IgDneg CD19+) cells in the viable cell gate), using mice from Fig 3a-c, with B cell genotypes color-coded as for panels a, b. (e, f) Quantitation of Gls and Mpc2 gene inactivation efficiency. Normalized read counts of inducible gene knockout targets in GCB cells purified by flow sorting one week after SRBC immunization of mice of the indicated genotypes.

Synthetic auxotrophy - glutaminase support of anti-NP response is dependent on mitochondrial pyruvate channel subunit 2

(a) Representative illustration of the flow cytometric identification of GC B cells used in the analyses. Shown is a series of 2-parameter flow plots in which CD19+ IgDneg events were selected from dumpneg B220+ cells in the live cell FSC x SSC gate, followed by identification of the GL7+ CD38neg (or, in some experiments, CD95+) population designated as GC-phenotype B cells. (b) Shown are the aggregated data on the frequencies of GC-phenotype B cells measured at 1 week after SRBC immunizations (six biological replicate samples of two temporally separate experiments, each with three mice of each genotype) initiated after B cells were converted to the indicated genotypes by serial treatments with tamoxifen. (c) Combined metabolic support preferential support to the NP-binding repertoire among GC B cells. Splenocytes from the mice harvested after primary and secondary (boost) immunizations of the experiments shown in Fig 3a were analyzed by flow cytometry that included NP-APC in addition to the direct immunofluorescent staining to identify viable GC B cells by their phenotype as shown in panel a. Shown on the two rows are (upper row of five plots) the CD38 vs GL7 profiles for B220+ CD19+ dumpneg IgDneg in the FSC x SSC gate of viable cells and, for events in the GL7++ CD38neg gate denoted by the polygon, the CD19 vs NP-APC fluorescence emission profile (lower row of five plots). Inset numbers show the percentage of positive events within the gate indicated to the left. Representative data are shown from one mouse of each of the indicated B cell genotypes in one experiment out of three performed with two mice of each genotype and a non-immunized control. (d) With each dot representing one mouse, the quantitative data on frequencies of GC-phenotype (GL7+ CD38neg) B cells among total B cells in the six subjects of these three new experiments were aggregated along with earlier data from experiments where the NP-APC staining was of insufficient quality. (e) Shown are the aggregated data on frequencies of NP-APC+ events among GC-phenotype (GL7+ CD38neg) B cells in the six mice whose B cells were of the indicated genotype. (c and e) each represent three temporally independent replicate experiments, in each of which two mice whose B cells were converted to the indicated genotypes by tamoxifen injections as diagrammed in Fig 3a. Statistical testing to calculate the indicated P values was performed using the Mann-Whitney test.

Glutamine and pyruvate metabolism in B cells affects their support to follicular helper T cells.

(a) Representative illustration of the flow cytometric identification of Tfh- and GC-Tfh T cells. Shown is a series of 2-parameter flow plots in which CD4+ TCR+ events were selected from viable (Ghostneg B220+ events in the live cell FSC x SSC gate) cells, followed by determination of the levels of CXCR5 and PD1. As demarcated by the indicated polygonal gates, PD1int CXCR5int CD4 T cells were scored as Tfh while GC-Tfh were PD1hi CXCR5hi. (b) Tfh and GC-Tfh cell frequencies in spleens of the immunized subjects analyzed in Fig 3-supplement 3b Shown are the aggregated data on the frequencies of Tfh- and GC-Tfh-phenotype CD4 T cells measured one week after SRBC immunization followed by harvest of mice whose B cells were converted to the indicated genotypes by serial treatments with tamoxifen. (c-e) Glutaminase and mitochondrial pyruvate channel promote response of reactivated memory B cells. Mice of indicated genotypes were immunized with NP-OVA, injected with tamoxifen, and boosted with NP-OVA as in Fig 3h. Shown are serologies of the all-and high-affinity NP-specific IgM (c), IgG2c (d), and IgA (e) antibodies from three independent replicate experiments. Shown are the mean (±SEM) of aggregated data from the three independent experiments with nine mice of each genotype, as well as indications of results from statistical testing (*, p<0.05).

Synthetic auxotrophy - glutaminase support of splenic production of anti-NP ASCs is enhanced by mitochondrial pyruvate channel subunit 2.

Panels (a) - (g) show data derived from the experiments of Fig 3g, h, in which tamoxifen injections to convert the conditional alleles (Glsf/f or Mpc2f/f) to loss-of-function were delayed until after 3 wk after primary immunization and response. (a-f) Aggregated results of ELISpot assays were used to measure the frequencies of ASCs producing all- [binding high valency NP20 albumin (a-c)] or high- [binding and retained on low valency NP2 (d-f)] -affinity anti-NP Ab of the indicated isotypes. (g) NP-specific GC B cells were analyzed in the spleen from immunized mice with indicated genotypes. Shown are the mean (±SEM) frequencies of NP+ cells in GL7+ CD95+-gated B cells from three independent replicate experiments.

Glutaminolysis, especially in concert with mitochondrial pyruvate import, promotes proliferation, viability, and differentiation of anti-IgM-stimulated B cells.

(a) Shown are the representative flow plots of CTV partitioning (upper) and CD138+ cells (lower) among 7AAD- viable lymphocytes gates. B cells were stained with CTV, stimulated with anti-IgM, anti-CD40, BAFF, IL-4, and IL-5, and cultured for 5 d with and without CB839 and UK5099 as indicated. (b) Aggregated mean (±SEM) frequencies of CD138+ cells among 7AAD- events in the gate viable lymphocytes. (c) Aggregated mean (±SEM) frequencies of the cells that divided ≥ 3 times. (d) Aggregated total numbers of viable cells obtained at the end of the 5 d cultures. (e) Aggregated mean (±SEM) frequencies of 7AADneg viable cells in the lymphocyte gate. Data derive from six biologically independent cell pools and samples, derived from three temporally independent experiments in each of which two separate cell pools from distinct mice were subdivided to test in each of the four indicated conditions. The indicated P values were calculated by Mann-Whitney U testing.

Metabolite evidence of the impact of CB839 on glutaminolysis and mitochondrial metabolism in activated B cells.

(a) A schematic illustration of glutaminolysis, its inhibition by CB-839, and GPT2-(mitochondrial alanine aminotransferase, also termed glutamate pyruvate transaminase) catalyzed conversion of glutamate and pyruvate for generation of alanine. PDH, pyruvate dehydrogenase; MPC, mitochondrial pyruvate channel. (b-j) Selected metabolic perturbations identified by metabolomic analysis of activated B cells. For each of three biologically and temporally independent replicate experiments, B cell pools were purified from several mouse spleens, then activated and cultured 2 d as in Fig 5, or under the indicated conditions [or with Fab2’ anti-IgM (1 µg/mL) added, yielding similar results (not shown)]. Shown are results for (b) glutamine, (c) glutamate, (d) alanine, along with results of calculating inferred glutaminase activities (e). (f) Schematic illustrating mitochondrial import of fatty acids via L-carnitine and acyl-carnitine intermediaries, along with use of the acetyl (Ac)-CoA for either entry into the Krebs (TCA) cycle or generation of acetyl (Ac)-carnitine. LC, long-chain; coA, coenzyme A; CPT, carnitine palmitoyl transferase, FAO, fatty acid oxidation; acetyl, Ac; TCA, tricarboxylic acid (Krebs cycle); CACT, carnitine-acylcarnitine transferase; CrAT, carnitine O-acetyltransferase. (g) Computationally inferred activity of fatty acid oxidation (FAO) derived from the metabolomic data. (h) L-carnitine and (i) acetyl-carnitine concentrations in the activated B cells, and the ratio calculated for each sample (j).

Impacts of glutaminolysis, alone or in concert with mitochondrial pyruvate import, on differentiation vs secretory function of B lymphoblast-derived plasma cells.

Using conditions akin to Fig. 4a-e, purified B cells were activated and cultured 4 d in the presence or absence of the indicated inhibitors. Equal number of viable cells were replated for ELISpot assays after rinsing and counting, with cultures in the wells performed in the absence (panels a, c) or presence (b, d) of the indicated inhibitors. (a, b) Photographs of spots in wells of the indicated cultures, scoring the frequencies as well as spot sizes of cells secreting IgM or IgG1 as indicated. Shown are single wells from one representative experiment, representative of the technical duplicates and of the five biological replicate experiments. (a) B cells were cultured 4 d in the presence of CB839 or UK5099 prior to plating culturing in inhibitor-free medium for the ELISpots. (b) B cells were cultured 4 d in inhibitor-free medium, followed by addition of the indicated compounds after plating in the ELISpot wells and overnight cultures. (c, d) Mean (±SEM) data from all five biologically independent replicate experiments quantitating the relative frequencies of ASCs and sizes of the spots (surrogates for amount of Ab secreted during the overnight culture) are shown. Left, middle, and right panels show relative frequencies of ASCs secreting IgG1, mean spot sizes after detection of IgM, and mean IgG1 spot sizes, respectively. For each experiment (a common pool of purified B cells activated and cultured in parallel with inhibitor(s) or vehicle alone), the ASC numbers and average spot sizes for inhibitor-treated cultures were normalized to those measured for the vehicle (DMSO) control. (c) Inhibitors were present during 4 d cultures, as in (a). (d) Inhibitors were added only after plating in ELISpot wells for overnight assays of secretion after a pool of activated B cells was aliquoted after culture 4 d without inhibitor present,

On the relationship between proliferation effects and differentiation efficiency of B cells treated with CB839, alone or in concert with hydroxychloroquine.

(a) Plasma cell differentiation cultures of purified B cells, labelled with CTV and activated in the presence of the indicated combinations of drugs (or DMSO vehicle), were performed as in Fig. 4 and analyzed by flow cytometry. (a) Mean (±SEM) %CD138+ cells (day 5) at levels of CTV fluorescence representing divisions 3 through 7 are plotted separately for each condition shown in the key. Open symbols, HCQ added; filled symbols - no HCQ. Line colors are coded as in Fig. 4, 5. Mean results derived from three biologically independent experiments. (b) Quantitative data on frequencies of plasma cells after independent cultures of purified B cells were performed and analyzed as in Fig. 4 (no CTV labeling), in the indicated combinations of drugs. Dots denote individual values for four independent B cell pools and cultures, with bars representing the mean % CD138+. (c, d) HCQ effect on PC development contingent on inhibition of glutaminolysis. Using data from (a), the % CD138+ for each indicated condition in each independent experiment was measured in the CTV peaks representing viable cells that divided four (c) and five (d) times. In each case, a ratio of control to HCQ-treated value was calculated for the condition (DMSO or CB839 present) (left graph) and the actual % CD138+ with and without HCQ (right graph).

Altered gene expression of metabolically reprogrammed B cells

Additional data relating to the analyses of RNA-seq results with flow-purified GC B cells (controls versus those with disruption of Gls, Mpc2, or both; Fig. 4j-l; Fig. 6a-c). (a) In the bubble plot summarizing the results of GSEA using the RNA-seq data, the heat-mapped color coding of each circle denoted the normalized enrichment score on the scale to the right, while the size of each circle indicates the adjusted P value. (b-d) Selected GSEA plots illustrative of the changes in transcriptional programs of GC B cells with altered metabolism due to post-maturation disruption of the genes Gls, Mpc2, or both, as summarized in (a). (b) Enrichment of Myc- and E2F-upregulated mRNA in WT samples as compared to Gls Δ/Δ, Mpc2 Δ/Δ. (c) GLS-dependent increases in expression of RNA of the oxidative phosphorylation and E2F pathways, with GSEA comparing Mpc2 Δ/Δ to Gls Δ/Δ, Mpc2 Δ/Δ GC B cells shown. (d) MPC-dependent increases in expression of RNA of the IFN-γ response and apoptosis program gene sets compared for Gls Δ/Δ versus Gls Δ/Δ, Mpc2 Δ/Δ GC B cells shown.

Synthetic auxotrophy of B cell metabolism supports ROS homeostasis and survival in GC B cells.

(a, b) mitochondrial (mt)DNA content (a) and MitoTracker Green (MTG) labeling (b) of B cells generated as in Fig 5 were measured by qPCR or flow cytometry, respectively. Mice whose B cells were of the indicated genotypes were generated by tamoxifen injections into huCD20-CreERT2-expressing subjects (Gls +/+; Mpc2 +/+ or Gls f/f; Mpc2 f/f), immunized with SRBC, and harvested 1 wk post-immunization. (a) qPCR measurement of mtDNA after activation and culture (2 d) as in Fig. 5c. (b) Cellular mitochondrial content determined by MitoTracker Green (MTG) labelling quantified by flow cytometry. Shown are MFI values from each independent experiment after activation and culture (2 d) as in Fig. 5c, then normalized to DMSO-treated condition in each experiment. (c) Increased ROS in GC-phenotype B cells with inactivated Gls and Mpc2 genes. In the analyses of each experiment DCFDA levels in GL7+ CD38- IgD- GCB cell gated cells were measured at 1 week after SRBC immunization. The geometric MFI (gMFI) of DCFDA in WT GCB cells was averaged and normalized to 1. Relative DCFCA fluorescence intensities in WT and GlsΔ/Δ; Mpc2Δ/Δ GCB cells were then calculated accordingly. Shown are mean (±SEM) relative gMFI of DCFDA in GCB cells from two independent replicate experiments. P values were calculated by Mann-Whitney U test.. (d) Impairment of Gls and Mpc2 genes leads to increased mitochondrial ROS in GC B cells. MitoSOX levels in GL7+ CD38- IgD- GCB-gated cells were measured at 1 week after NP-OVA boost. Shown are mean (±SEM) gMFI of MitoSOX in GCB cells from two independent replicate experiments. (e, f) Non-viable GC B cells increased in the loss-of-function B mice. (e) Representative flow plots of annexin V vs 7-AAD for identification of early-apoptotic cells in GL7+ CD38- IgD- GCB cell gated cells. (f) Shown are aggregated mean (±SEM) frequencies of annexin V+ 7-AAD+ cells. (c-f) Samples were from mice immunized after tamoxifen injections to activate the recombinase in B cells of HuCD20-CreERT2-transgenic mice that were either f/f or +/+ at the Gls and Mpc2 loci.

Gating to analyze cell division history in B cells transduced with the bi-cistronic MiT retrovector.

Relating to Fig 6e, f and using one representative sample from it, an illustration of the flow cytometric measurement of CTV fluorescence emission and designation of CTVlo cells in the CD90 (Thy1.1)+ and CD90 (Thy1.1)neg gates, corresponding to transduced and non-transduced cells.

Normal IFN-R expression yet decreased P-STAT1 in metabolically reprogrammed B cells

(a, b) Deletion efficiency of Gls1 and Mpc2 in vivo. CD19+ IgD+ naïve B cells were flow-purified from spleens of tamoxifen-injected and SRBC-immunized hu-CD20-CreERT2 mice (Glsf/f, Mpc2f/f, or wild-type). Shown are the levels of Gls- (a) and Mpc2-encoded RNA (b) in the cells with the indicated genotypes relative to WT control after normalized to β-actin as an internal control. (c) Immunoblot analysis of whole cell extracts of B cells of the indicated genotypes, purified from tamoxifen-treated mice (two of each genotype probed with anti-GLS and anti-cyclophilin B. (e, f) Active metabolism via GLS1 and MPC2 promotes interferon activation of STAT1. WT or GlsΔ/Δ; Mpc2Δ/Δ B cells were activated with anti-CD40 and BAFF for 2 days followed by IFN-β (c) or IFN-γ (d) treatment for 15 min. Shown are the representative western blot images from more than three independent experiments. (f) Cell surface IFNAR signal on B cells activated and cultured as in Fig 6 was determined by flow cytometry. Shown is a representative result from replicate experiments (n = 3) analyzing IFNAR expression on viable B cells.