Initial B Cell Activation Induces Metabolic Reprogramming and Mitochondrial Remodeling

Summary

B lymphocytes provide adaptive immunity by generating antigen-specific antibodies and supporting the activation of T cells. Little is known about how global metabolism supports naive B cell activation to enable an effective immune response. By coupling RNA sequencing (RNA-seq) data with glucose isotopomer tracing, we show that stimulated B cells increase programs for oxidative phosphorylation (OXPHOS), the tricarboxylic acid (TCA) cycle, and nucleotide biosynthesis, but not glycolysis. Isotopomer tracing uncovered increases in TCA cycle intermediates with almost no contribution from glucose. Instead, glucose mainly supported the biosynthesis of ribonucleotides. Glucose restriction did not affect B cell functions, yet the inhibition of OXPHOS or glutamine restriction markedly impaired B cell growth and differentiation. Increased OXPHOS prompted studies of mitochondrial dynamics, which revealed extensive mitochondria remodeling during activation. Our results show how B cell metabolism adapts with stimulation and reveals unexpected details for carbon utilization and mitochondrial dynamics at the start of a humoral immune response.

Graphical Abstract

Highlights

• •Glucose is dispensable for B cell activation; OXPHOS is fueled by other nutrients
• •Few, elongated mitochondria remodel to many punctate mitochondria upon activation
• •mtDNA and nucleoid numbers remain similar for naive and 24 hr stimulated B cells
• •Combined RNA-seq and metabolomics used to study metabolism during B cell activation

Immunology; Immune Response; Components of the Immune System; Metabolomics

Introduction

B and T lymphocytes are key effector cells in an adaptive immune response. Both cell types express unique surface receptors that recognize an immense repertoire of antigens. Upon antigen recognition, B cells upregulate surface molecules that support the activation of T cells, including those for antigen presentation through major histocompatibility complex class II (MHC class II) and those for co-stimulation by CD86/B7-2. T cell-dependent B cell activation leads to selective recruitment into transient anatomical sites of differentiation known as germinal centers (GCs). In GCs, B cells proliferate and edit their antigen receptors to refine and augment a humoral immune response. This gene editing includes somatic hypermutation, which increases antibody specificity, and isotype class switch recombination (CSR), which exchanges the immunoglobulin heavy chain constant region to support specific antibody functions. Ultimately, GC B cells differentiate into long-lived plasma cells that secrete antibodies or into memory B cells that re-activate with antigen re-engagement, providing a basis for vaccination.

The metabolic requirements for initially activated B cell proliferation and differentiation are relatively unknown (reviewed in Table S1). Naive B cells are maintained in a quiescent state, and activation initializes cellular reprogramming to drive re-entry into the cell cycle. Rapid cell expansion requires production of biomolecules including proteins, lipids, and nucleotides at an extraordinary rate. The details for how activated B cells support a rapidly increasing biomass are unknown. Two recent reviews expressed a critical need for studies of metabolic pathways supporting B cell activation (Boothby and Rickert, 2017, Jellusova and Rickert, 2017). Stimulation of B cells with lipopolysaccharide (LPS), a Toll-like receptor (TLR) 4 receptor agonist, or an anti-IgM antibody, to cross-link the B cell antigen receptor, increases glucose import into activated B cells and leads to an increased oxygen consumption rate (OCR) (Caro-Maldonado et al., 2014, Cho et al., 2011, Doughty et al., 2006, Dufort et al., 2007, Jellusova et al., 2017). How imported glucose is utilized, and whether T cell-dependent B cell activation utilizes similar pathways remain unknown. More than a decade ago, Doughty and colleagues utilized isotopomer-labeled glucose in IgM-stimulated B cells to show that glucose shuttled into the pentose phosphate pathway (PPP), but reported results for only two metabolites, lactate and glutamate (Doughty et al., 2006). To date there are no studies that globally examine metabolism or glucose fate upon initial B cell activation beyond these individual measures.

Despite reports of increased oxygen consumption in multiple B cell stimulation conditions, little is known about how quiescent B cells alter their mitochondria to support increased oxidative phosphorylation (OXPHOS). Both T cell-dependent and T cell-independent stimulation have been observed to increase staining with membrane potential-dependent dyes, by immunofluorescence or flow cytometry. However, the resolution achieved fails to discriminate between fluorescence increases due to increases in mitochondrial mass versus an increase in membrane potential, or both. Recent studies in T lymphocytes have demonstrated major remodeling upon activation (Ron-Harel et al., 2016), although it remains unclear whether mitochondrial remodeling also occurs with or without similar dynamics for B lymphocytes to support a shift from quiescence to activation. Therefore, we examined primary mouse B cell metabolism using a well-established in vitro model of B cell activation that replicates a B cell interaction with T cells through co-stimulation of surface CD40 and addition of the T cell cytokine, interleukin (IL)-4 (Rush and Hodgkin, 2001). By combining publicly available RNA sequencing (RNA-seq) data with mass spectrometry (MS) tracing of glucose and intracellular metabolites, we report on key metabolic networks and requirements for B cell metabolism during early B cell activation.

Results

Activated B Cells Increase Glucose Uptake but Do Not Accumulate Glycolytic Metabolites

To isolate B cells, we sorted CD43⁻ cells from total mouse splenocytes (Figure S1A), resulting in a 97% pure CD19⁺ B220⁺ CD4⁻CD8⁻ naive B cell population (Figure S1B). After 24 hr of stimulation with CD40L and IL-4 (Rush and Hodgkin, 2001), flow cytometry analysis confirmed that B cells had undergone an increase in cell size as measured by forward scatter (FSC-A) and induction of activation markers including MHC class II, required for antigen presentation to T cells, and CD86/B7-2, a costimulatory molecule required for T cell activation (Figure S1C).

Previous studies have shown that B cells increase glucose import with activation (Caro-Maldonado et al., 2014, Cho et al., 2011, Doughty et al., 2006, Dufort et al., 2007). In agreement, we also measure an increase in import of the fluorescent glucose analog, 2-Deoxy-2-[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]-D-glucose (2-NBDG), in CD40/IL4 activated B cells (Figure 1A). To investigate carbon utilization from glucose, we performed metabolite tracing in naive and stimulated B cells. Growing cells in media with ¹³C₆-glucose enables tracing of carbons by analyzing the shifts in mass peaks of metabolites through MS (Table S2). We observe that 90% of glucose was fully m+6 labeled in both conditions, confirming import of the glucose label (Figure 1B). Multiple published reports suggest or assume that glycolysis is upregulated upon B cell activation (Caro-Maldonado et al., 2014, Doughty et al., 2006, Garcia-Manteiga et al., 2011, Jellusova et al., 2017). Unexpectedly, however, the total levels of glycolytic metabolites decrease upon activation, with the exception of 3-phosphoglycerate (3PG) (Figure 1C). Of note, lactate levels do not increase at 24 hr as expected with upregulation of glycolysis. We also investigated the isotopologue distribution in glycolytic metabolites. Despite decreases in the total amounts of glycolytic metabolites, we measured increased m+6 label in glucose-6-phosphate/fructose-6-phosphate and fructose-1,6-bisphosphate, and increased m+3 label in G3P and 3PG for activated versus naive B cells (Figure 1D). These results suggest that glucose is fluxing through the glycolytic pathway, although not accumulating, and is likely routed into alternative metabolic pathways in activated B cells.

Figure 1.

B Cell Activation Induces Glucose Import without Accumulation of Glycolytic Metabolites; Glucose Restriction Has Only Minor Impacts on B Cell Function

(A) Representative flow cytometry plot and quantification of 2-NBDG glucose import into naive and stimulated B cells with unstained control (n = 3).

(B) Isotopomer distribution for labeled glucose in naive and stimulated B cells (n = 3).

(C) UHPLC-MS quantification of relative levels of glycolytic metabolites between stimulated and naive B cells, data shown are log₂ converted average values of 3 technical replicates of 3 biological replicates (n = 3).

(D) Isotopomer distribution for labeled glucose in glycolytic metabolites in naive and stimulated B cells (n = 3).

(E) Flow cytometry analysis of cell size (FSC-A), activation markers (MHCII, CD86) at 24 hr, germinal center differentiation (GC, %B220+ Fas+ GL7+), and class switch recombination (CSR, %B220+ IgG1+) at day 3, and plasmablast differentiation (PB, %B220^lo CD138+) at day 5 in cells stimulated with or without glucose (n = 3). Data represent mean ± SD. p Values were determined by a paired two-tailed Student's t test. **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001.

G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; F16BP, fructose-1,6-bisphosphate; G3P, glycerol-3-phosphate; 3PG, 3-phosphoglycerate; Pyr, pyruvate; Lac, lactate.

Since multiple studies have found that glucose uptake is increased upon B cell activation (Caro-Maldonado et al., 2014, Cho et al., 2011, Doughty et al., 2006, Dufort et al., 2007), we sought to determine the functional outcome of glucose limitation by culturing B cells in media lacking glucose. For these studies low-level, >10-fold reduced, residual glucose (1.5 mM, data not shown) was unavoidably present from the media fetal bovine serum (FBS). Surprisingly, there was a small to absent impact of limiting glucose on B cell activation, differentiation, or proliferation (Figure 1E). B cells cultured in residual FBS glucose showed a defect in class switching to IgG1; however, glucose appeared dispensable in culture for other B cell functions (Figure 1E).

OXPHOS and TCA Cycle Elevation

Prior studies of metabolism during B cell activation provide an incomplete evaluation of metabolic reprogramming in B cells. To determine which metabolic pathways are upregulated, and thus likely active, we performed gene set enrichment analysis (GSEA) on a previously published RNA-seq dataset containing naive and 24 hr activated B cells stimulated by CD40L and IL-4 (GEO: GSE77744) (Wöhner et al., 2016). We identified 56 metabolic Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways comprising between 15 and 500 genes each, and 12 enriched metabolic pathways with a false discovery rate < 0.25 (Tables 1 and S3). Aminoacyl tRNA synthesis (KEGG: MMU00970) was the most enriched pathway and includes transcripts for all tRNA synthetase subunits. This result is concordant with a general increase in protein translation during the transition from a quiescent to a rapidly dividing cell (Vander Heiden et al., 2011).

Table 1.

Gene Set Enrichment Analysis (GSEA) for Induced Metabolic Transcripts during B Cell Activation

KEGG Pathway	ES	NES	Pval	FDR
Aminoacyl tRNA synthesis	0.556	2.215	<0.001	<0.001
Oxidative phosphorylation	0.447	2.183	<0.001	0.002
Terpenoid backbone synthesis	0.645	2.127	<0.001	0.003
Citrate cycle	0.542	2.012	<0.001	0.005
Pentose phosphate pathway	0.486	1.710	0.008	0.049
Biosynthesis of unsaturated fatty acids	0.500	1.728	0.004	0.054
Pyrimidine metabolism	0.333	1.552	<0.001	0.107
One carbon pool by folate	0.453	1.434	0.073	0.190
Cysteine and methionine metabolism	0.372	1.372	0.113	0.236
Glutathione metabolism	0.338	1.331	0.100	0.238
Glyoxylate and dicarboxylate metabolism	0.378	1.286	0.139	0.247
Purine metabolism	0.260	1.301	<0.001	0.249

KEGG metabolism gene set enrichment analysis (GSEA) based on genes differentially expressed between naïve (n = 2) and 24 hr stimulated (n = 2) B cells (GEO: GSE77744). Gene sets were derived from the KEGG database (n = 84, 56 after filtering). Elevated set signatures below a q value of 0.25 are listed (ES, enrichment score; NES, normalized enrichment score; Pval, nominal p value; FDR, FDR adjusted q value).

Two additional top KEGG pathways were OXPHOS (KEGG: MMU00190) and the tricarboxylic acid (TCA) cycle (KEGG: MMU00020), signifying key roles during B cell activation (Figure 2A). This finding is consistent with prior reports of increased oxygen consumption with B cell activation by LPS or IgM (Caro-Maldonado et al., 2014). To determine whether OXPHOS was also increased with CD40/IL-4 modeled T cell-induced activation, we analyzed the OCR of unstimulated and stimulated B cells. Basal OCR was higher in stimulated B cells than in naive cells, as was maximal respiratory capacity (Figures 2B and S2A). Interestingly, basal OCR of stimulated B cells was almost identical to the maximal OCR of unstimulated B cells (0.717 and 0.684 fmoles O₂/min/cell, respectively). Both subsets of B cells had significant spare respiratory capacity (SRC), which sharply contrasts with activated T cells and naive pre- and pro-B cells that exhibit minimal SRC (Stein et al., 2017, van der Windt et al., 2012). Our data suggest distinct mechanisms of metabolic adaptation for T, immature B, and mature activated B cells.

Figure 2.

B Cell Activation Induces Increases in OXPHOS and the TCA Cycle

(A) GSEA on RNA-seq transcriptome data from naive and 24 hr stimulated B cells shows enrichment for OXPHOS and TCA cycle metabolic transcripts.

(B) Seahorse extracellular flux analysis measurement of oxygen consumption rate (OCR) in naive and stimulated B cells (n = 4).

(C) UHPLC-MS quantification of log₂ converted average levels of TCA cycle metabolites between stimulated and naive B cells (n = 3).

(D) Isotopomer distribution of labeled glucose for TCA cycle metabolites in naive and stimulated B cells (n = 3). Data represent mean ± SD. p values determined by two-way ANOVA with Bonferroni correction for multiple comparisons. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001. Pyr, pyruvate; Cit, citrate; Glu, glutamate; Gln, glutamine; αKG, α-ketoglutarate; Succ, succinate; Fum, fumarate; Mal, malate.

See also Figure S2, Tables S2 and S3.

To further interrogate our transcriptome data mining results that the TCA cycle was upregulated, we measured TCA metabolite levels in stimulated versus naive B cells. We observed an increase in fumarate, malate, and α-ketoglutarate, with little to no change in pyruvate, citrate, or succinate (Figure 2C). We next quantified the isotopomer distribution of TCA cycle metabolites resulting from co-incubation with ¹³C₆-glucose. Fully labeled m+3 pyruvate comprised 96% and 97% of the total pyruvate pool in both naive and stimulated B cells, respectively. Inclusion of ¹³C₆-glucose and ¹³C₃-pyruvate into the TCA cycle would result in accumulations of m+2 peaks for all TCA cycle metabolites (Figure S2B). Surprisingly, examination of five TCA cycle intermediates showed minimal or no labeling in both naive and stimulated conditions (Figure 2D). For example, although there is an increase in the total fumarate level with B cell activation, indicating augmented synthesis of fumarate at 24 hr, only 9% was m+2 labeled in stimulated B cells. This result is despite an increase in the expression of the mitochondrial pyruvate carrier genes, Mpc1/2 (Figure S2C), which encode for proteins that import pyruvate into mitochondria to supply the TCA cycle, suggesting (not surprisingly) a post-transcriptional regulatory mechanism for pyruvate entry into naive and activated B cells. Increases in OCR and total TCA metabolite levels (Figures 2B and 2C) indicate increased TCA flux; however, labeled glucose and pyruvate are not contributing carbon atoms to these intermediates. Instead, it is likely that activated B cells utilize alternative nutrients to fuel the TCA cycle, which could include lipids, ketones, lactate, or glutamine. Indeed, measurement of glutamine in the culture media shows that B cells import glutamine at a rate of 12.4 μM glutamine/hour/10⁶ cells (Figure S2D).

We stimulated naive B cells in the presence of ATP synthase inhibitor, Oligomycin A, to determine the importance of OXPHOS upregulation in activated B cells that are actively increasing biomass. Oligomycin A inhibited B cell size increase and prevented upregulation of B cell activation markers (Figure 3A), suggesting that early activated B cells have coupled mitochondria, and a central role for the TCA cycle and OXPHOS in these processes. At day 3 of modeled T cell stimulation, 10 nM Oligomycin A also completely blocked naive B cell differentiation into GC B cells and immunoglobulin class switching to IgG1. By day 5, cells treated with Oligomycin A were unable to differentiate into plasmablasts (Figure 3A), further supporting the essential role of OXPHOS in B cell activation and differentiation. Since glutamine is robustly imported into activated B cells (Figure S2D) and could fuel the TCA cycle, we investigated the requirement for glutamine during B cell activation by culturing B cells in glutamine-free media. Loss of glutamine causes reductions in activated B cell size, differentiation, class switching, and proliferation (Figure 3B). This result is consistent with findings from Heyse et al. (2015) that show the importance of glutamine for biomass accumulation and DNA replication but not early activation, and is similar to our findings upon inhibition of OXPHOS with Oligomycin A. This, in conjunction with our findings that OXPHOS is increased but not fueled by glucose, suggests that glutamine import fuels the TCA cycle, and is required for B cell activation.

Figure 3.

Perturbation of OXPHOS Negatively Affects B Cell Differentiation

(A) Flow cytometry analysis of cell size (FSC-A), activation markers (MHCII, CD86) at 24 hr, germinal center differentiation (GC, %B220+ Fas+ GL7+) and class switch recombination (CSR, %B220+ IgG1+) at day 3, and plasmablast differentiation (PB, %B220^lo CD138+) at day 5 in cells stimulated with DMSO or 10 nM Oligomycin A (n = 6 DMSO, n = 3 Oligo).

(B) Flow cytometry analysis of cell size (FSC-A), activation markers (MHCII, CD86) at 24 hr, germinal center differentiation (GC, %B220+ Fas+ GL7+) and class switch recombination (CSR, %B220 + IgG1+) at day 3, and plasmablast differentiation (PB, %B220^lo CD138+) at day 5 in cells stimulated with or without glutamine (n = 3).

Data represent mean ± SD. p Values determined by paired two-tailed Student's t test. *p ≤ 0.05; **p ≤ 0.01; ****p ≤ 0.0001.

Glucose Fate(s)

Glucose import increases with naive B cell activation but does not appreciably supply the TCA cycle nor coincides with enrichment of glycolytic genes as measured by GSEA (KEGG: MMU00010) (Figure S3A). Therefore, we analyzed the distribution of glucose carbons through additional metabolic pathways (Figure 4A). Pyruvate is converted to lactate as an end product of glycolysis. However, whereas 97% of pyruvate is m+3 labeled in both naive and activated B cells (Figure 2D), only 31% of total lactate is m+3 labeled (Figure 4B). Interestingly, this is despite increased expression of the lactate to pyruvate conversion enzyme Ldha (Figure S3B). Although 31% m+3 label does represent an increase in labeling from naive B cells, it does not fully explain the final fates of carbon during B cell activation. Pyruvate and additional glycolytic intermediates are likely increasing their flux through alternative metabolic pathways. One metabolic pathway that branches from glycolysis is the PPP, which generates ribose-5-phosphate (R5P) for incorporation into nucleotides for DNA replication and RNA transcription. Among the enriched metabolic pathways, GSEA shows statistically significant enrichments of PPP, pyrimidine metabolism, and purine metabolism (KEGG: MMU00030, MMU00240, MMU00230) (Table 1, Figure 4C) in stimulated B cells. Consistent with these enriched gene set pathways, total and m+5 labeled R5P is massively increased (Figure 4D), providing a precursor for purine and pyrimidine biosynthesis. This m+5 label also tracks to downstream ribonucleotides, with less incorporation into deoxyribonucleotides (Figure S3C), despite markedly increased expression of Rrm1/2 genes, which convert ribonucleotides to deoxyribonucleotides (Figure S3D). It is likely that at 24 hr, increases in transcription favor de novo biosynthesis of ribonucleotides over deoxyribonucleotides because DNA replication does not occur until after 24 hr of activation of B cells (Figure S3E).

Figure 4.

Additional Metabolic Pathways Utilize Carbon with B Cell Activation

(A) Schematic of potential main branch points for glucose-derived carbon utilization.

(B) Isotopomer distribution in naive and stimulated B cells for lactate (n = 3).

(C) GSEA on RNA-seq transcriptome data from naive and 24 hr stimulated B cells shows enrichment for PPP, pyrimidine, and purine metabolism transcripts.

(D) Isotopomer distribution of labeled glucose in R5P in naive and stimulated B cells (n = 3).

(E) GSEA for terpenoid biosynthesis and biosynthesis of unsaturated fatty acids.

(F) Isotopomer distribution of labeled glucose in CDP-ethanolamine and phosphatidylcholine in naive and stimulated B cells (n = 3).

Data represent mean ± SD. p Values determined by two-way ANOVA with Bonferroni correction for multiple comparisons. *p ≤ 0.05; ****p ≤ 0.0001.See also Figure S3, Tables S2 and S3.

Glucose carbons can also route toward the synthesis of lipids. GSEA metabolic signatures enriched for both terpenoid backbone biosynthesis and biosynthesis of unsaturated fatty acids (KEGG: MMU00900, MMU01040, respectively) (Table 1, Figure 4E), including transcript elevations for Hmgcr, the rate-limiting enzyme in cholesterol biosynthesis, and Elovl6, an enzyme involved in fatty acid elongation (Figure S3F). This suggests increased utilization of an elevated glucose flux toward the generation of lipids for stimulated B cells, which is compatible with a rapid replication rate. Although many nonpolar lipids are incompatible with ultra-high-performance liquid chromatography-MS on polar columns, we were able to detect isotopomer incorporation of glucose into cytidine disphosphate ethanolamine (CDP-ethanoloamine), a precursor for phosphatidylethanolamine, suggesting that glucose is indeed fluxing toward lipid generation. Also, despite previous findings showing incorporation of radiolabeled glucose into the predominant lipid species of cellular membranes, phosphatidylcholine (PC), upon stimulation with LPS (Dufort et al., 2014), we did not observe incorporation of glucose into PC by CD40/IL-4 stimulation (Figure 4F), suggesting that T cell stimulation mechanisms may program differential metabolic programs to support different biological demands. Overall, these data suggest that increased glucose uptake contributes mainly to the synthesis of biomass, including ribonucleotides and lipids, as stimulated B cells grow in preparation for rapid cell division.

Mitochondrial Remodeling in Response to B Cell Activation

Augmented TCA cycle and OXPHOS activation in stimulated B cells suggests the possibility for coordinate changes in mitochondrial dynamics and morphology. To evaluate mitochondria in living mouse B cells, we employed super-resolution Airyscan microscopy with tetramethylrhodamine, ethyl ester (TMRE), a mitochondrial membrane-potential dependent dye, and PicoGreen, a double-stranded DNA intercalation dye that marks both genomic and mtDNA. In naive B cells, we detected few, long mitochondria with multiple nucleoids per mitochondrion (Figures 5A and S4A). On average, each naive B cell had 2.6 mitochondria, which increased to an average of 5.0 mitochondria per B cell upon activation (Figure 5B). The average area of a single mitochondrion remained relatively stable, whereas the total mitochondrial area increased from 2.3 to 4.2 μm² (Figure 5B). The aspect ratio (long axis_object/short axis_object) of mitochondria showed higher ratios in naive cells, supporting our observations that naive B cells have longer mitochondria than stimulated B cells (Figure S4B). Circularity (4πA/perimeter²) assessments also reflect a greater roundness of mitochondria in activated B cells (Figure S4B). In agreement with OCR increases in activated B cells, we also detected an increase in mean TMRE fluorescence intensity at 24 hr (Figure S4B), reflecting an increase in proton gradients within the mitochondria.

Figure 5.

B Cell Activation Induces Mitochondrial Remodeling without Changes in mtDNA Levels

(A) Live-cell super-resolution imaging of primary naive and stimulated B cells stained with TMRE and PicoGreen (n = 120 naive cells, 90 stimulated from at least 3 biological replicates). Arrowheads indicate representative mitochondrial nucleoids. Scale bar represents 2 μm.

(B) Quantification of the number of mitochondria per cell, single mitochondrial area, and total mitochondrial area (n = 313 naive mitochondria from 120 cells, 447 stimulated mitochondria from 90 cells, all from at least 3 biological replicates).

(C) Ratio of mtDNA to genome DNA measured by qPCR (n = 3).

(D) GSEA for genome DNA and mtDNA replication shows no enrichment for mtDNA replication genes.

(E) Quantification of the number of nucleoids per cell, number of nucleoids per mitochondria, and total nucleoid area per cell (n = 120 naive cells, 90 stimulated from at least 3 biological replicates, n = 120).

Data represent mean ± SEM (B and D) or mean ± SD (C). p Values determined by unpaired two-tailed Student's t test. *p ≤ 0.05; ****p ≤ 0.0001. See also Figure S4.

A striking feature of activated B cells is the lack of an increase in mtDNA measured by qPCR despite increased mitochondrial numbers after stimulation (Figure 5C). Analysis of RNA-seq data for enrichment of KEGG pathways for genomic DNA and mtDNA replication (KEGG: MMU03030, MMU03029) are consistent with this result, as there is no significant change in the expression of mtDNA replication genes upon activation (Figure 5D). The total nucleoid number also does not increase with B cell activation; however, there is a decrease in the number of nucleoids per mitochondrion (Figure 5E). This is consistent with our observation that naive B cells have fewer, elongated mitochondria with multiple nucleoids, and that stimulated B cells have more numerous and rounded mitochondria with a single nucleoid. Our data suggest that B cell mitochondria with multiple nucleoids undergo fission to generate more mitochondria without replicating their mtDNA, which agrees with minimal deoxyribonucleotide production at 24 hr activation (Figure S3B). Also, we detect little change in nuclear-encoded mitochondrial fission transcripts at 24 hr activation, although we do identify increases in the expression of MICOS complex genes, which are responsible for inner membrane remodeling (Cogliati et al., 2016) (Figure S4C). In addition, the total nucleoid area per cell increases with stimulation (Figure 5E), with increased nucleoid area previously suggested to represent increased mtDNA transcription (Alan et al., 2016, Lee and Han, 2017), consistent with increased emphasis on ribonucleotide production and augmented OXPHOS activity during the first 24 hr of B cell activation (Figure S3B). Combined, our data suggest a potential model in which B cells meet increased demand for mitochondrial transcripts and OXPHOS during initial T cell-assisted activation by maintaining an expanded nucleoid pool in naive B cells, rather than by replicating mtDNA upon activation.

Discussion

GC B cells are the fastest replicating non-cancerous cells in the human body, with little known about how they alter their metabolism from quiescence to support a massive numeric expansion to enable an effective humoral immune response. Using glucose isotopomer tracing we identified an unexpected decrease in total glycolytic metabolites with modeled T cell activation of naive B cells, with the exception of 3PG (Figure 1C). Increased 3PG may shunt toward synthesis of amino acids, and indeed, we detect increased total amounts of phosphoserine, serine, glycine, and threonine at 24 hr of activation (Figure S3G). Accumulation of 3PG may also suggest a block in pyruvate kinase (Pkm1), which catalyzes the conversion of phosphoenol pyruvate to pyruvate, or a switch to expression of the pyruvate kinase 2 (Pkm2) isoform. This switch has previously been observed in homocysteine-stimulated B cells (Deng et al., 2017) and occurs commonly in proliferating cells with a high level of nucleotide biosynthesis (Dayton et al., 2016).

A key unexpected finding was that glucose restriction had a relatively minor impact on T cell-dependent B cell activation (Figure 1D). Multiple studies have shown that B cells increase glucose uptake upon activation with multiple stimuli (Caro-Maldonado et al., 2014, Doughty et al., 2006, Garcia-Manteiga et al., 2011, Jellusova et al., 2017), yet B cells grown in only residual, 10-fold reduced glucose are able to activate and differentiate. The primary defects in glucose-restricted conditions were a minor decrease in cell size, likely from reduced contributions of glucose to the biosynthesis of lipids (Figures 4E and 4F) (Dufort et al., 2014), a minor increase in activation marker CD86, and a major defect in CSR of IgM to IgG1 (Figure 1E). It is unclear how glucose limitation may lead to reduced CSR; however, many of the signaling pathways required for B cell activation overlap with regulators of metabolic functions, including Pax5, Myc, and Hif (Boothby and Rickert, 2017). Further study is needed to tease apart the specific signaling pathways that link glucose metabolism to CSR.

Our surprising finding that glucose was largely dispensable for initial B cell activation prompted a more detailed study of metabolic pathways that support the energetic and biosynthetic requirements of B cell activation (Table 1). Analysis of an available RNA-seq dataset (Wöhner et al., 2016) enabled investigations of metabolic pathways utilized by CD40/IL-4-activated B cells. A caveat of this approach is that many metabolic processes are regulated allosterically rather than transcriptionally, including glycolysis, so we supplemented our studies with high-throughput metabolic tracing to evaluate consistency in our results. Specifically, we found that activated B cells upregulate OXPHOS and the TCA cycle (Figure 2A). Scant glucose is utilized in the TCA cycle (Figure 2D), suggesting that alternative carbon sources, such as glutamine, are anapleurotic fuels. Indeed, we and others have shown that B cell biomass accumulation, protein synthesis, and DNA replication, but not early activation, depend on glutamine (Figure 3B) (Heyse et al., 2015). Inhibition of OXPHOS with Oligomycin A substantially impaired B cell activation (Figure 3A). We cannot rule out, however, that Oligomycin A treatment may cause accumulation of reactive oxygen species (ROS), or that loss of mitochondrial function may cause other cellular phenotypes, such as lysosomal defects (Demers-Lamarche et al., 2016), that prevent usual B cell differentiation. Glutamine deprivation, however, phenocopies inhibition of OXPHOS by Oligomycin A (Figure 3A), which is at least consistent with glutamine utilization by the TCA cycle for supporting B cell functions. We also identified a flux increase in the PPP at 24 hr (Figure 4C), which, in addition to generating mainly ribonucleotides, provides reducing equivalents NADH and NADPH during this process. Activation of B cells induces intracellular ROS within the first 24 hr (Wheeler and Defranco, 2012), so PPP induction may also help maintain the B cell redox state.

Isotopomer tracing revealed 97% labeling of pyruvate in both naive and stimulated B cells; however, labeled pyruvate was not fully incorporated into lactate (Figure 4B) despite increases in Ldha expression. Previous studies have observed increased extracellular acidification rate (ECAR) upon B cell activation and have attributed this to increased lactate production (Caro-Maldonado et al., 2014, Jellusova et al., 2017); however, studies that specifically measure extracellular lactate (Garcia-Manteiga et al., 2011, Jellusova et al., 2017) did not find increases in actual lactate levels (Table S1). We propose that an ECAR increase originates from increased CO₂ and carbonic acid production from a highly active TCA cycle, and that activated B cells utilize pyruvate primarily for biosynthesis.

Pyruvate was not incorporated into lactate, but incorporation of labeled glucose carbons into the TCA cycle also did not appreciably occur (Figures 2D and 4B), leaving open the question of where pyruvate is utilized. We hypothesize that pyruvate converts into acetyl-CoA and then incorporates into fatty acids, which is at least consistent with significant gene induction in this KEGG pathway (Table 1). In addition Dufort et al. (2014) have shown that LPS stimulation and IgM with IL-4 induce incorporation of radiolabeled glucose into polar and nonpolar lipids, although the ratios of the labeled lipids differ according to the stimulation condition, likely due to different bioenergetic needs for the final fates of each condition. Further studies of targeted metabolites will be required to determine the final fate(s) of pyruvate during the various stimuli of naive B cell activation.

Studies of mitochondria in primary lymphocytes have been limited, with prior work using transgenic T cells expressing fluorescent-tagged mitochondria (Ron-Harel et al., 2016) or low-resolution immunofluorescence imaging that lacked structural insights (Table S1) (Jellusova et al., 2017). Here, we imaged live B cells on a wild-type background by utilizing cell-permeable stains to study mitochondrial dynamics upon activation. This initial imaging study of mature B cell mitochondria provides some unexpected results. Similar to T cells, B cell activation causes changes in mitochondrial architecture. In direct contrast to T cells, however, there are ∼50% fewer initial mitochondria in naive B cells (2.6/cell) than in naive T cells (∼5.0/cell) (Ron-Harel et al., 2016) (Figures 5A and 5B). In addition, in T cells, increases in mitochondrial number accompany increases in mtDNA copy number, which does not occur in B cells. Instead, mtDNA and nucleoid numbers remain similar between naive and 24 hr stimulated B cells (Figure 5C). This may be due to naive B cells maintaining a larger mtDNA pool than thymocytes (Milasta et al., 2016). Further work is required to determine the benefits of this strategy for robust proliferation, activation, and differentiation, which aligns well with the delayed timing of genomic DNA replication and the initial preference for ribonucleotide synthesis during early B cell activation (Figures S3B and S3C).

The increased mitochondrial fragmentation upon stimulation, concordant with increases in OXPHOS, has only been reported for lymphocytes thus far. Canonically, longer, fused mitochondria are associated with increased OXPHOS, and smaller, fragmented mitochondria have less OXPHOS capacity and are associated with nutrient deprivation or OXPHOS inhibition (Galloway et al., 2012). This is an apparently distinct morphological strategy employed by lymphocytes. Perhaps in this highly proliferative system, B cells sacrifice mitochondrial morphology associated with efficient OXPHOS in favor of morphology that may promote additional mitochondrial pathways, such as one-carbon metabolism or calcium signaling.

In summary, we present a detailed investigation of the metabolic adaptations utilized by modeling T cell-dependent early B cell activation. By combining RNA-seq data with metabolomics, we uncovered metabolic pathways, fluxes, and strategies that support rapid B cell proliferation and expansion to initiate a humoral immune response. These findings broaden and deepen our understanding of the usual metabolic pathways that support B cell activation and the transition from quiescence to proliferation and differentiation, which may enable targeting of metabolic regulatory nodes to block autoimmunity and B cell leukemia and lymphoma.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: July 27, 2018

Supplemental Information

Table S2. List of Relative Metabolite Amounts and Isotopomer Distribution Results from Naive and Stimulated B Cells with ¹³C₆-Glucose Labeling, Related to Figures 1–3 and S3

Relative amounts and isotopomer contributions of metabolites collected from B cells cultured for 2 hr in media containing ¹³C₆-glucose without stimulation (naive, n = 3) or 24 hr with CD40/IL4 stimulation (24 hr stim, n = 3).