Dynamic mitochondrial transcription and translation in B cells control germinal center entry and lymphomagenesis

Germinal center (GC) B cells undergo proliferation at very high rates in a hypoxic microenvironment but the cellular processes driving this are incompletely understood. Here we show that the mitochondria of GC B cells are highly dynamic, with significantly upregulated transcription and translation rates associated with the activity of transcription factor A, mitochondrial (TFAM). TFAM, while also necessary for normal B cell development, is required for entry of activated GC precursor B cells into the germinal center reaction; deletion of Tfam significantly impairs GC formation, function and output. Loss of TFAM in B cells compromises the actin cytoskeleton and impairs cellular motility of GC B cells in response to chemokine signaling, leading to their spatial disorganization. We show that B cell lymphoma substantially increases mitochondrial translation and that deletion of Tfam in B cells is protective against the development of lymphoma in a c-Myc transgenic mouse model. Finally, we show that pharmacological inhibition of mitochondrial transcription and translation inhibits growth of GC-derived human lymphoma cells and induces similar defects in the actin cytoskeleton.

Germinal center (GC) B cells undergo proliferation at very high rates in a hypoxic microenvironment but the cellular processes driving this are incompletely understood. Here we show that the mitochondria of GC B cells are highly dynamic, with significantly upregulated transcription and translation rates associated with the activity of transcription factor A, mitochondrial (TFAM). TFAM, while also necessary for normal B cell development, is required for entry of activated GC precursor B cells into the germinal center reaction; deletion of Tfam significantly impairs GC formation, function and output. Loss of TFAM in B cells compromises the actin cytoskeleton and impairs cellular motility of GC B cells in response to chemokine signaling, leading to their spatial disorganization. We show that B cell lymphoma substantially increases mitochondrial translation and that deletion of Tfam in B cells is protective against the development of lymphoma in a c-Myc transgenic mouse model. Finally, we show t ha t p ha rmac ol ogical inhibition of mitochondrial transcription and translation inhibits growth of GC-derived human lymphoma cells and induces similar defects in the actin cytoskeleton.
The germinal center (GC) reaction is a highly spatially organized process in secondary lymphoid tissue essential for humoral immunity 1 . B cells responding to antigen captured by follicular dendritic cells introduce random mutations in their immunoglobulin genes in a process known as somatic hypermutation (SHM), which occurs in the anatomically defined dark zone, and are then selected through competitive interaction with follicular helper T cells (T FH ) in the light zone. GC B cells cycle between these zones, leading to antibody affinity maturation and eventually formation of memory B or plasma cells. During SHM, mutations are introduced in immunoglobulin gene loci through the action of activation-induced cytidine deaminase (AID encoded by AICDA). This can give rise to oncogenic mutations, for example, the translocation of MYC with the IGH or IGL loci 2 . GC B cells are the origin of most diffuse large B cell lymphomas (DLBCL), the most common non-Hodgkin lymphoma.
The metabolic processes that support GC B cell homeostasis are incompletely understood. GC B cells are highly proliferative, with division times as short as 4-6 h, and reside within a hypoxic Article https://doi.org/10.1038/s41590-023-01484-3 which we confirmed by flow cytometry (Fig. 1c). We found that the mitochondrial GFP signal was highest in the gray zone, followed by the light zone and then the dark zone (Fig. 1c). We identified highly distinct mitochondrial morphology between naive and GC B cells, with small, fragmented mitochondria in naive B cells and large fused mitochondria in GC B cells ( Fig. 1d and Extended Data Fig. 1a). The mito-QC reporter mouse also allows detection of autophagy of mitochondria (mitophagy), as mCherry is resistant to quenching in the acid environment of the autolysosome. Using multispectral imaging flow cytometry, we screened GC B cells for the presence of GFP − mCherry + punctae. Although autophagy has been reported in GC B cells 24 , we did not identify evidence of active mitophagy (Extended Data Fig. 1b-d). We hypothesized that for B cells entering the GC reaction to acquire the high mitochondrial mass we observed, activation of mitochondrial transcription and translation would be required. Using Aicda-Cre × Rosa26 STOP tdTomato reporter mice to label GC B cells, we found markedly increased levels of cytochrome c oxidase subunit 1 (COXI), part of the electron transport chain (ETC) complex IV, in GC B cells (Fig. 1e,f).
To examine mitochondrial transcription, we next labeled B cells in vivo or ex vivo with 5-ethynyl uridine , which is incorporated into actively synthesized RNA. We detected colocalization between 5-EU and COXI, with 5-EU incorporation highly prominent in GC B cells and most often seen in areas of high COXI signal (Fig. 1g).
To understand the regulation of this high rate of mitochondrial protein transcription and translation, we examined the expression and distribution of TFAM. We found TFAM protein organized into punctate structures representing mitochondrial nucleoids, which colocalized with 5-EU (Fig. 1g). These were more numerous and larger in GC B cells (Fig. 1h) and had a more elliptical morphology, which is associated with elevated transcriptional activity (Extended Data Fig. 1e) 25 . We confirmed the increase in nucleoid number using stimulated emission depletion (STED) super-resolution microscopy (Fig. 1i). Mitochondria are therefore highly dynamic in GC B cells, with prominent transcriptional activity associated with increased TFAM-nucleoid content.

Tfam is essential during B cell development and differentiation
TFAM has a role in the prevention of long-term inflammatory immunosenescence in T cells but is dispensable for their development 20,26 . Whether TFAM is important for B cell development or function is unknown, as are the dynamics of ETC component expression in B cell developmental trajectories. To address these questions, we first conditionally deleted Tfam in B cells using Cd79a-Cre (hereafter B-Tfam), microenvironment [3][4][5] . Typically, rapidly proliferating immune cells mainly use aerobic glycolysis, but GC B cells rely on fatty acid oxidation (FAO) and oxidative phosphorylation, although to what extent is somewhat controversial [4][5][6][7][8][9][10] . This metabolic phenotype is carried over into a substantial proportion of DLBCLs, which are also oxidative phosphorylation-dependent 11 . In vitro, mitochondria have been shown to be regulators of B cell signaling through redox-related mechanisms, although whether these studies can be generalized to the hypoxic microenvironment of the GC found in vivo is uncertain [12][13][14] .
Regulation of mitochondrial transcription and translation is important in other immune cell types, including cytotoxic CD8 + T cells, where it controls cell killing independent of the effects on metabolism, and cytokine production in CD4 + T cells [15][16][17] . A key regulator of mitochondrial transcription and translation is transcription factor A, mitochondrial (TFAM), a DNA-binding high mobility box group protein that aids in the packaging of the mitochondrial genome into nucleoids, analogous to the role of histone proteins, and also controls transcription and translation of mitochondrial DNA, serving as a regulator of mitochondrial biogenesis 18,19 . In T cells, TFAM controls immunosenescence by restraining the production of inflammatory cytokines 20 ; in fibroblasts and myeloid cells, it acts to regulate an antiviral immune state 21 . Recent work has shown that the mitochondrial DNA helicase TWINKLE is required for GC formation and plasma cell differentiation 22 . However, the dynamics, function and regulation of mitochondria in GC B cell biology remain incompletely understood.
We show that GC B cell mitochondria are highly dynamic organelles, undergoing profound structural changes as they transition through the GC reaction. We find that TFAM is dynamically regulated in B cells and is required for their development and transcriptional and spatial entry into the GC reaction by modulating cellular motility. We also demonstrate, using genetically modified mouse models, that TFAM is essential for the development of lymphoma and that pharmacological inhibition of mitochondrial transcription and translation in human lymphoma cells is a potential treatment target for human disease.

GC B cells undergo active mitochondrial remodeling
To determine mitochondrial density and structure in GC B cells, we first immunized mitochondrial reporter mice expressing green fluorescent protein (GFP) and mCherry (mito-QC) 23 with sheep red blood cells (SRBCs) and examined spleen by confocal microscopy and flow cytometry. The mitochondrial GFP signal was strongly localized to GCs and significantly higher than the surrounding B cell follicle ( Fig. 1a,b),

Fig. 1 | GC B cells undergo active mitochondrial remodeling. a, Spleen sections
from mito-QC mice immunized with SRBC and analyzed on day 12. Scale bar, 50 μm. Representative of three independent experiments. b, Quantification of mitochondrial GFP signal intensity and area (μm 2 ) in GCs. Each point represents an individual GC (n = 9) and surrounding B cell follicle (n = 9), pooled from three individual mito-QC mice immunized as in a. Representative of two independent experiments. c, Gating strategy for GC B cells and their subsets in the spleen. GC and naive B cells were gated as CD19 + CD38 − GL-7 + and CD19 + IgD + GL-7 − , respectively. GC B cells from the dark zone, gray zone and light zone were identified based on CXCR4 and CD86 expression signatures as depicted. Quantification and comparison of gMFI of mitochondrial GFP in the indicated subsets from immunized mito-QC mice (n = 7), pooled from two independent experiments. d, Maximum intensity projection of three-dimensional (3D) Airyscan ICC images depicting the mitochondrial network (GFP) in MACSenriched and flow-sorted GC B cells (light zone, gray zone and dark zone) and magnetic bead-sorted naive B cells collected from immunized (enhanced SRBC immunization protocol at day 12) and unimmunized mito-QC spleens, respectively. Scale bar, 3 μm. Three-dimensional mitochondrial volume (μm 3 ) quantification in naive (n = 10), light zone (n = 49), gray zone (n = 29) and dark zone (n = 28) cells isolated from four mice. Representative of two independent experiments. e, Spleen sections from Aicda-Cre × Rosa26 STOP tdTomato mice immunized with SRBC and analyzed on day 12, with IHC for COXI and mitochondrial import receptor subunit TOM20 homolog (TOMM20). Scale bar, 50 μm. Representative of two independent experiments. f, Flow cytometry histogram plots and quantification of COXI protein levels in GC B cells normalized to IgD + naive B cells from the same mice (n = 4). Representative of three independent experiments. g, Representative Airyscan confocal images of in vivo 5-EU incorporation (indicating RNA synthesis), with COXI and TFAM in ex vivo-isolated naive and GC B cells. Quantification of mitochondrial 5-EUintegrated signal density in ex vivo-labeled naive (n = 23) and GC B (n = 32) cells isolated from two mice. Scale bar, 3 μm. Representative of two independent experiments. h, Two-dimensional Airyscan ICC images of TFAM in naive and GC B cells, with 3D reconstruction indicating individual mitochondrial nucleoids. Quantification of nucleoid number and volume (μm 3 ) in naive (n = 22) and GC B (n = 24) cells isolated from two mice. Scale bar, 1 μm. Representative of two independent experiments. i, Deconvoluted confocal and STED super-resolution ICC images of TFAM in naive and GC B cells. Scale bar, 1 μm. Representative of two imaging experiments. Statistical significance was calculated using an unpaired two-tailed t-test (b,g,h), a two-tailed Mann-Whitney U-test (f) or an ordinary one-way analysis of variance (ANOVA) with Tukey's multiple comparisons test (c,d). Data are presented as the mean ± s.e.m. DZ, dark zone; LZ, light zone; GZ, gray zone. Nature Immunology | Volume 24 | June 2023 | 991-1006 993 Article https://doi.org /10.1038/s41590-023-01484-3 which is active from the early pro-B cell stage 27 . TFAM was efficiently deleted in mature splenic B cells (Extended Data Fig. 2a,b). B-Tfam mice appeared healthy, with no clinical signs of immunosenescence or overt autoimmunity. However, B-Tfam mice had a profound reduction in the peripheral B cell compartment . Analysis of B cell development in the bone marrow indicated a failure of progression from the pro-to the pre-B cell stage (Fig. 2b,c and Extended Data Fig. 2f). We noted that heterozygous B-Tfam mice had normal B cell development and an intact peripheral B cell compartment (Extended Data Fig. 2g,h).
To understand the expression of ETC proteins during B cell development, and the effect of Tfam deletion, we quantified a panel of ETC proteins by high-dimensional spectral flow cytometry in B wild-type (B-WT) and B-Tfam bone marrow and spleen (Fig. 2d,e). We found that in B-WT bone marrow, there was a progressive increase in the expression of most ETC proteins from the pre-pro-B stage (Hardy fraction A 28 ), peaking at the earliest pre-B subset (fraction C′) and then substantially falling in the later stages (fractions D-F). Interestingly, TFAM expression peaked early (fraction A), suggesting that it might initiate a mitochondrial transcription program leading to upregulation of ETC proteins. The maximum expression of ETC proteins corresponded to the developmental block seen in B-Tfam mice (pro-to pre-B cell stage). B-Tfam mice deleted TFAM from fraction A onwards in the bone marrow and TFAM remained deleted in splenic B cell subsets; accordingly, mitochondrially encoded COXI was downregulated. The nuclear-encoded ETC components COXIV, cytochrome C, ATP5A1 and succinate dehydrogenase A (SDHA) were increased.
In the periphery, ETC component expression was similar across transitional and follicular B cells but increased in marginal zone B cells (Fig. 2e). TFAM deletion led to a much more marked upregulation of

HSP60
Stage Condition nuclear-encoded ETC proteins. This was evident in a mismatch in the ratio of proteins encoded by mitochondrial or nuclear genomes in the same ETC complex, that is, COXI and COXIV of complex IV. Extracellular flux analysis of unstimulated peripheral B cells showed no difference in oxygen consumption rate (OCR) but a significant increase in extracellular acidification rate in B-Tfam mice (Extended Data Fig. 2i).
Also increased in B-Tfam B cells were the mitochondrial unfolded protein response (UPR mt )-associated proteins heat shock protein 60 (HSP60) and mitochondrial 70-kDa heat shock protein (mtHSP70) (Fig. 2d,e). These results demonstrate that ETC components are dynamically regulated across B cell developmental trajectories and that TFAM is required for the expression of mitochondrially encoded proteins.
We next examined the ability of B-Tfam mice to generate GCs. Examination of immunized B-Tfam spleens revealed smaller and fewer B cell follicles and an almost complete absence of GL-7 + GCs compared to B-WT mice (Fig. 2f,g). Tfam heterozygosity in B cells also led to a significantly reduced GC response after SRBC immunization (Extended Data Fig. 2j). These results collectively suggest that TFAM is essential for normal B cell differentiation and that in situations of low-energy demand, partial respiratory compensation is possible despite substantially reduced translation of mitochondrial proteins.

GC B cells require TFAM
Having established a role for TFAM in B cell development and ETC complex balance, we next focused on its function in the GC reaction. To specifically delete Tfam in GC B cells, we generated Aicda-Cre × Tfam loxP × Rosa26 STOP tdTomato × Blimp1-mVenus mice (hereafter Aicda-Tfam), which allow simultaneous identification of cells that have expressed Aicda (activated, GC, memory B and plasma cells) and/or currently express Blimp1 (encoded by Prdm1) (by detecting the fluorescent reporter proteins tdTomato and mVenus, respectively) ( Fig. 3a). Aicda-Tfam mice were immunized with SRBCs and analyzed on day 12. Immunofluorescence staining of Aicda-Tfam spleen sections demonstrated small GCs in reduced numbers compared with Aicda-WT controls (Fig. 3b). Clusters of plasma cells in the splenic cords were much smaller in Aicda-Tfam mice and Blimp1-mVenus + tdTomato + cells were poorly represented, suggesting impairment of GC output (Fig. 3c). The proportion of CD138 + tdTomato + long-lived plasma cells in bone marrow was also significantly lower (Extended Data Fig. 3a). We found that the generation of CD19 + IgD lo CD38 − GL-7 + (and correspondingly CD19 + IgD lo CD38 − tdTomato + ) GC B cells was substantially reduced in Aicda-Tfam mice as measured by flow cytometry (Fig. 3d,e). We also confirmed a reduction in Blimp1-mVenus + tdTomato + post-GC plasma cells (Fig. 3f).
Next, we examined antigen-specific GC formation by immunizing mice with (4-hydroxy-3-nitrophenyl) acetyl-chicken gamma globulin (NP-CGG). The proportion and absolute numbers of NP-binding GC B cells were significantly reduced in Aicda-Tfam mice, as were those of plasma cells measured in situ; the plasma cell to GC B cell ratio was correspondingly decreased . Affinity maturation was also compromised, with decreased binding of IgG1 antibodies to NP 1-4 compared with NP >20 (Fig. 3h and Extended Fig. 2 | Tfam is essential during B cell development and differentiation. a, Representative flow cytometry plots of B220 + B cells from the spleens of unimmunized 6-week-old B-Tfam and B-WT mice, with quantification of total splenic B cell counts and absolute counts of follicular (CD23 + CD21 int ) and marginal zone (CD23 − CD21 + ) B cell subsets (n = 5 mice). Representative of two independent experiments. FSC, forward scatter. b, Representative flow cytometry plots and quantification of bone marrow IgM − IgD − B220 + pre-pro-B, pro-B and pre-B cells from B-Tfam and B-WT mice (n = 5 mice). Results are representative of two independent experiments. Equivalent Hardy stages are shown in parentheses. c, Representative flow cytometry plots and quantification of bone marrow-resident B220 + IgM + IgD − immature and B220 + IgD + mature B cell subsets from B-WT and B-Tfam mice (n = 5 mice). Results are representative of two independent experiments. Equivalent Hardy stages are shown in parentheses. d, Heatmap of row z-scores for the gMFI of indicated mitochondrial proteins, measured by spectral flow cytometry in bone marrow B cell subsets (Hardy stages) of unimmunized B-WT and B-Tfam mice (mean of n = 2). Results are representative of two independent experiments with n = 4 mice per group in total. Fr A, pre-pro-B; Fr B-C, pro-B; Fr C′-D, pre-B; Fr E, immature-B; Fr F, mature-B. e, Heatmap of row z-scores for the gMFI of indicated mitochondrial proteins, measured by spectral flow cytometry in splenic B cell subsets of unimmunized B-WT and B-Tfam mice (mean of n = 2). Results are representative of two independent experiments with n = 3 mice per group in total. f,g, B-WT and B-Tfam mice were immunized with SRBCs (days 0 and 5) intraperitoneally; on day 12 spleens were analyzed by IHC and flow cytometry. f, Representative tile-scan images of spleen sections from B-Tfam and B-WT mice depicting GL-7 + GCs and CD21/35 + B cell follicles. Scale bar, 500 μm. Results are representative of two independent experiments. g, Flow cytometry gating strategy and quantification of CD38 − GL-7 + GC B cells in the spleens of immunized B-Tfam (n = 6) and B-WT (n = 3) mice. Results are representative of two independent experiments. Statistical significance was calculated using an unpaired two-tailed t-test (a,g) or a two-way ANOVA with Šidák's correction (b,c) or Tukey's (d,e) multiple comparisons tests, with experimental batch as a covariate (d,e). Data are presented as the mean ± s.e.m. . b-f, Aicda-WT and Aicda-Tfam mice were immunized with SRBCs intraperitoneally; on day 12, spleens were analyzed using flow cytometry or IHC. b,c, Representative Airyscan IHC confocal images of spleen sections (b) and tdTomato + Blimp1-mVenus + plasma cell clusters in red pulp (c) from Aicda-Tfam and Aicda-WT mice. Scale bars, 500 μm and 75 μm, respectively. Representative of three independent experiments. d,e, Representative flow cytometry plots for the identification of GC B cells from Aicda-WT (n = 6) and Aicda-Tfam (n = 4) mice. Quantification of CD38 − GL-7 + GC B cell (d) and CD38 − tdTomato + GC B cell (e) percentages within the CD19 + B cell compartment. Representative of four independent experiments. f, Representative flow cytometry plots and quantification of tdTomato + post-GC plasma cells within the Dump¯(IgD‾, Gr1‾, CD3‾) Blimp1-mVenus + cell compartment (n = 3 mice per group). Data are representative of four independent experiments. g,h, Aicda-WT and Aicda-Tfam mice were immunized with 50 μg NP-CGG in alum (at a 1:1 ratio) intraperitoneally; on day 14, spleens and sera were collected for flow cytometry, IHC and enzyme-linked immunosorbent assay analyses. g, Representative flow cytometry plots depicting CD19 + CD38 − GL-7 + GC B cells binding NP-PE or allophycocyanine (NP-APC) from Aicda-Tfam (n = 10) or Aicda-WT mice (n = 4). Quantification of NP-specific GC B cell absolute counts is shown. Data are representative of three independent experiments. h, Enzyme-linked immunosorbent assay quantification and comparison of the ratio of IgG1 NP-specific high-affinity antibodies to lowaffinity antibodies detected by binding to NP [1][2][3][4] and NP >20 antigens, respectively, from Aicda-WT (n = 6) and Aicda-Tfam mice (n = 5). Data were pooled from two independent experiments. i-k, Aicda-WT and Aicda-Tfam mice were immunized with 50 μg NP-CGG in alum (at a 1:1 ratio) intraperitoneally; on day 30, they were boosted with 50 μg NP-CGG in PBS. Spleens and bone marrow were collected for flow cytometry on day 70. i, Representative flow cytometry plots depicting tdTomato + IgG1 + CD38 + memory B cells binding NP-APC from Aicda-WT and Aicda-Tfam mice. j, Proportional comparison and absolute count quantification of NPbinding within CD38 + IgG1 + tdTomato + memory B cells from Aicda-WT and Aicda-Tfam mice (n = 4 per group). Representative of two independent experiments. k, Absolute plasma cell counts in the bone marrow of Aicda-WT and Aicda-Tfam mice (n = 6 per group). Data were pooled from two independent experiments. Statistical significance was calculated using a two-tailed Mann-Whitney U-test (d,e) and an unpaired two-tailed t-test (f-h,j-k). Data are presented as the mean ± s.e.m. Article https://doi.org/10.1038/s41590-023-01484-3 Data Fig. 3e,f). However, there was no difference in IgM anti-NP antibodies, which is in keeping with the extrafollicular origin of this response, produced by plasmablasts that have not expressed Aicda (Extended Data Fig. 3c,g).
To understand if loss of Tfam compromised B cell memory, we immunized Aicda-Tfam and Aicda-WT mice with NP-CGG in alum and then boosted them on day 30 with NP-CGG in PBS. On day 70 there were substantially fewer NP-binding tdTomato + CD38 + IgG1 + memory B cells (Fig. 3i,j) and on day 49 a reduction in IgG1 anti-NP antibodies reactive against NP [1][2][3][4] and NP >20 (Extended Data Fig. 3h). The number of plasma cells in the bone marrow was also significantly reduced on day 70 (Fig. 3k).
Mitochondria play a central role in the regulation of apoptosis, but surprisingly the apoptosis rate detected by means of active caspase 3 staining in Aicda-Tfam GC B cells was comparable with that of Aicda-WT controls (Extended Data Fig. 3i). In situ TUNEL also demonstrated unaltered apoptosis (Extended Data Fig. 3j). There were no significant differences in cell cycle dynamics between Aicda-Tfam and Aicda-WT mice (Extended Data Fig. 3k,l). We also confirmed that Tfam deletion did not affect proliferation or cell viability in vitro (Extended Data Fig. 3m-p).
These data demonstrate that loss of Tfam markedly compromises GC B cell differentiation and output but without detectable effects on proliferation and survival in those cells already committed to the GC fate.

TFAM controls transcriptional entry into the GC program
To examine the effects of Tfam deletion on the transcriptional program of GC B cells, we performed combined single-cell gene expression profiling and V(D)J sequencing. We immunized Aicda-Tfam and Aicda-WT mice with NP-CGG and sorted tdTomato + cells on day 14 (Fig. 4a). This population includes any B cells that have expressed Aicda and so will include pre-GC, GC, memory and plasma cells. We identified nine shared clusters, broadly separated into B cells from GC, CD38 + non-GC and plasma cell populations (Fig. 4b,c). Cluster 0 (C0), which expressed markers of immaturity suggestive of an activated precursor (AP) state (Ighd, Ccr6, Gpr183, Cd38, Sell and low levels of Bcl6), was significantly expanded in Aicda-Tfam compared with Aicda-WT mice (Fig. 4d). Examination of differential gene expression in C0 revealed, as expected, broad dysregulation of mitochondrial gene expression in Aicda-Tfam mice, in keeping with the function of TFAM as a regulator of mitochondrial transcription (Fig. 4e). Pathway analysis 29 demonstrated substantial downregulation of translation initiation and elongation gene sets (Fig. 4f). Gene components of the activator protein 1 (AP-1) signaling pathway Jun, Junb, Jund and Fos were all significantly upregulated (Fig. 4g). The AP-1 pathway is broadly upregulated by cellular stress signaling, reactive oxygen species and in response to environmental cues 30,31 .
Analysis of the main GC cluster (C1) demonstrated as before dysregulation of mitochondrial and ribosomal gene transcription, with upregulation of Jun, Fos and Junb (Fig. 4h,i). Notably increased in Aicda-Tfam cells was Rgs1 (regulator of G-protein signaling-1), a GTPase-activating protein that has an important role in the negative regulation of cell movement in response to chemokines (for example, CXCL12) 32,33 (Fig. 4j). Downregulated were Coro1a (coronin-1), which encodes an actin-binding protein required for cell migration 34 , and Arpc3 (actin-related protein 2/3 complex subunit 3) (Fig. 4h), which mediates branched actin polymerization and actin foci formation 35 .
We next examined B cell clonality by evaluating V(D)J sequences. Because we sorted all tdTomato + B cells, the overall clonal diversity of all samples was high (Fig. 4k). As expected in the anti-NP immune response, V H 1-72 usage was dominant (Fig. 4l). However, there was more diversity in Aicda-Tfam mice, with fewer larger clones, suggesting that there was less ability for the evolution of dominant clones.
There was significantly less SHM in Aicda-Tfam B cells (Extended Data Fig. 4a), including in the Ighv1-72 gene in GC B cells (Extended Data Fig. 4b,c). The W33L mutation in CDR1 and substitution of K59 of the V H 1-72 heavy chain confer increased affinity for NP 36,37 . There were significantly lower W33L and K59 substitution rates in the Aicda-Tfam GC B cell cluster, in keeping with our observation that high-affinity NP binding was reduced (Fig. 4m). There was no W33L mutation and negligible K59 substitution in the AP cluster (C0) of either Aicda-WT or Aicda-Tfam mice, reflecting their pre-GC state (Extended Data Fig. 4d).
Therefore, transcriptional analysis revealed evidence of failure of Aicda-Tfam B cells to transition from an AP to a committed GC B cell phenotype.

TFAM is required for GC B cell commitment
Given the relative accumulation in immunized Aicda-Tfam mice of AP B cells with high expression of Sell (selectin L), Ccr6 and a GC transcriptional profile suggestive of altered cell trafficking and cytoskeleton dynamics, we hypothesized that Tfam is required for activated B cells to enter the GC and remain appropriately spatially positioned.
We first confirmed the proportional expansion of APs, defined as tdTomato + CD38 + IgD + , in Aicda-Tfam mice after NP-CGG immunization (Fig. 5a). Despite a significant numerical reduction in GC B cells, the AP population was maintained (Extended Data Fig. 5a). This was also seen to a more striking extent in B-Tfam mice immunized with SRBC, with APs defined as IgD + GL-7 int (Extended Data Fig. 5b). We found that in Aicda-WT mice, a lower proportion of AP B cells bound NP compared to GC B cells, and did so with lower affinity, in keeping with their lack of a W33L mutation and low levels of SHM (Extended Data Fig. 5c).
We then examined the splenic sections of immunized Aicda-Tfam mice under high magnification. We found a highly disorganized GC architecture, with poor GC B cell compartmentalization; within the follicle, there were relatively many more tdTomato + IgD + B cells in Aicda-Tfam mice (Fig. 5b). The level of B cell lymphoma 6 (BCL6) protein, a master transcriptional regulator of GC commitment and entry, was lower in Aicda-Tfam GC B cells (Extended Data Fig. 5d).
We next asked whether it was possible to overcome the failure of AP B cells to enter the GC by adoptively transferring preactivated Tfam −/− AP-like B cells into primed congenically marked mice. We used the induced GC B cell (iGB) culture system 38 and TAT-Cre to delete Tfam in Tfam loxP × Rosa26 STOP tdTomato B cells, or with WT congenically marked CD45.1/2 control B cells. This experimental design allows competitive transfer of activated B cells to take place (Fig. 5c). The resulting Tfam −/− iGB cells effectively deleted TFAM on day 4 of culture and this resulted in decreased expression of mitochondrially encoded ETC proteins and upregulation of nuclear-encoded proteins, as seen in splenic B cells from B-Tfam mice (Extended Data Fig. 5e). Loss of Tfam did not affect cell expansion over 4 days of culture (Extended Data Fig. 5f). However, after 5 days, Tfam −/− iGB cells were at a substantial competitive disadvantage in GC participation after adoptive transfer (Extended Data Fig. 5g and Fig. 5d-f).
We reasoned that the defects we observed on deletion of Tfam might be due to abnormalities in T FH -B cell interaction, or a defective T FH pool. Although there was no numerical defect in the T FH compartment (Fig. 5g,h), and antigen presentation was intact (Extended Data Fig. 5h), we noted impairment of T FH -induced in vitro GC B cell differentiation (Fig. 5i,j and Extended Data Fig. 5i).
To determine whether the defect we saw in Aicda-Tfam mice was B cell-intrinsic, we generated mixed competitive bone marrow chimeras, in which T FH generation would be intact. Aicda-Tfam GC B cells were outcompeted by WT cells compared with Aicda-WT controls in the spleen and Peyer's patches (Extended Data Fig. 5j). Overall these data suggest that TFAM promotes the entry or maintenance of activated B cells into the GC and that this function is principally cell-intrinsic.

TFAM regulates mitochondrial translation in activated B cells
We next examined the expression of TFAM and mitochondrial ETC components after immunization. There was dynamic expression of TFAM, which peaked at the AP stage and then subsided after GC were immunized with 50 μg NP-CGG with alum (at a 1:1 ratio) intraperitoneally; live Dump − (CD3 − , Gr1 − , CD11c − ) tdTomato + cells were flow-sorted before bead capture and 10X library preparation and sequencing. A total of 9,948 cells from Aicda-Tfam and 9,667 cells from Aicda-WT mice were sequenced. b, UMAP and clustering of integrated Aicda-WT and Aicda-Tfam (n = 3 mice per group) datasets. IFN, interferon. c, Heatmap of selected differentially expressed marker genes by cluster for all samples. d, Cluster proportions between groups. e, Volcano plot of differentially expressed genes in C0 (APs). f, SCPA of Gene Ontogeny biological processes (GO BPs) in the AP cluster. g, Jund gene expression projected onto clustered data. h, Volcano plot of differentially expressed genes in cluster 1 (C1) (GC). i, SCPA of GO BPs in the GC cluster (C1). j, Rgs1 gene expression projected onto clustered data. k, Plot of clonotype abundance distribution for each sample. l, Plot of proportions of top clonotypes for each sample. Clonotypes with IGHV1-72 usage are highlighted. m, Quantification of substitution at Ighv1-72 amino acid coding positions 33 and 59 for the GC B cell cluster. The number of GC B cells with Ighv1-72 usage analyzed are indicated on the plot. Statistical significance was calculated with a two-tailed t-test with correction for multiple comparisons using the Bonferroni method (d,e,h), a multisample Mahalanobis crossmatch test, with correction for multiple comparisons using the Bonferroni method (f,i) or a twotailed chi-squared test (m).
Article https://doi.org/10.1038/s41590-023-01484-3 entry, and a progressive increase in COXI, which was maximal in GC B cells from Aicda-WT mice (Fig. 6a,b). Other ETC proteins were also highly expressed in GC B cells, including those encoded in nuclear and mitochondrial DNA. Deletion of Tfam led to a program of alteration in mitochondrial ETC protein expression in AP and GC B cells in B-Tfam mice (Fig. 6c), much like that observed during B cell development.
To directly confirm the activity of mitochondrial translation, we then measured incorporation of the amino acid analog Zfp36 Rpl36

Rsrp1
Jchain Tsc22d3 Protein targeting Protein localization Ribosomal biogenesis Complement activation Adaptive immune resp. Adaptive immune resp. SHM Activation of immune resp. Immune resp. signaling Electron transport chain Generation precursor metabolites O-propargyl-puromycin (OPP) in isolated ex vivo mitochondria from Aicda-WT and Aicda-Tfam B cells by flow cytometry (Fig. 6d,e). The proportion of red fluorescent protein (RFP) + mitochondria (originating from tdTomato + AP/GC B cells) (Extended Data Fig. 6a) was substantially lower in Aicda-Tfam mice, reflecting their defective GC formation (Fig. 6e). We detected significantly more OPP incorporation in RFP + mitochondria

37.6% CD38-PE-Cy7
Aicda-WT tdTomato + Aicda-Tfam tdTomato + than those of non-AP or GC B cell origin (RFP − ) (Fig. 6f). Translation was decreased in RFP + mitochondria from Aicda-Tfam B cells compared to Aicda-WT cells (Fig. 6g); using Tfam −/− iGB cells, treatment with the mitochondrial translation inhibitor and oxazolidinone antibiotic chloramphenicol (CHL) 39 did not lead to additional suppression, confirming the defective translation in mitochondria (Extended Data Fig. 6b). We then examined the metabolic consequences of Tfam deletion on APs and GC B cells, quantifying cytoplasmic protein translation rate by the incorporation of OPP as a proxy for metabolic capacity, as recently reported 40 . GC B cells had significantly higher basal OPP incorporation than AP B cells, reflecting their high levels of metabolic activity (Extended Data Fig. 6c). Loss of Tfam led to an increase in glucose dependence and a marked reduction in FAO/amino acid oxidation (AAO) capacity in AP B cells, seen to a lesser extent in GC B cells ( Fig. 6h and Extended Data Fig. 6c,d). These results were mirrored by Seahorse extracellular flux analysis in activated B cells and iGBs ( Fig. 6i and Extended Data Fig. 6e,f).

GC B cell 62%
Therefore, loss of TFAM compromised GC B cell mitochondrial translation and impaired metabolic homeostasis.

Tfam deletion disrupts B cell mobility
We found that Aicda-Tfam GCs were poorly compartmentalized, with smaller dark zones (Fig. 7a) and a disrupted dark zone to light zone ratio measured using flow cytometry (Fig. 7b).
Positioning of B cells in the GC is controlled by the chemokines CXCL12 and CXCL13, which promote migration to the dark zone and light zone, respectively 41 . As our preceding data indicated that AP B cells need TFAM to enter the GC, and given that the transcriptional profile of Aicda-Tfam GC B cells was suggestive of cytoskeletal and mobility defects, we hypothesized that TFAM was required for proper cellular positioning in GCs.
We examined the cellular actin network of TFAM-deficient B cells and found a significant increase in filamentous actin (F-actin) in Aicda-Tfam GC B cells (Fig. 7c,d), which was also evident in B-Tfam B cells (Extended Data Fig. 7a,b). Rearrangement of the actin cytoskeleton is critical for B cell migration 42 ; to understand if Tfam deletion compromised GC B cell motility, we performed a transwell migration assay to determine chemotaxis in response to the chemokines CXCL12 and CXCL13. We found that Aicda-Tfam GC B cells migrated poorly, with significantly reduced chemotaxis compared to Aicda-WT cells (Fig. 7e).
B cell receptor (BCR) and chemokine-driven cytoplasmic calcium mobilization critically regulates F-actin organization in B cells 43 . As mitochondria are an important reservoir of intracellular calcium 44 , we asked whether Tfam deletion led to dysregulated intracellular calcium levels. After CXCL12 stimulation, B cells from B-Tfam mice failed to sustain peak cytoplasmic calcium levels (Fig. 7f), which we also observed after BCR stimulation with anti-IgM (Fig. 7g). Interestingly, we also detected a significant upregulation in the levels of the mitochondrial calcium uniporter (MCU), suggesting elevated mitochondrial calcium uptake capacity in B-Tfam B cells, whereas CD3 + T cells showed comparable MCU expression ( Fig. 7h and Extended Data Fig. 7c).
Mitochondrial reactive oxygen species (mtROS) were also substantially increased in B-Tfam naive and APs ( Fig. 7i and Extended Data Fig. 7d). ROS activate the AP-1 signaling pathway 31 and this observation was therefore consistent with the transcriptional profile of Aicda-Tfam AP B cells. and COXI (b) (n = 5 mice per group) gMFI in tdTomato + CD38 − GC B, tdTomato + IgD + AP and tdTomato − IgD + naive B cell compartments from SRBC-immunized Aicda-WT mice. Data are representative of three independent experiments. c, Heatmap of row z-scores for the gMFI of the indicated mitochondrial proteins, measured using spectral flow cytometry in the indicated B cell subsets of NP-CGG-immunized (day 14) B-WT and B-Tfam mice (mean of n = 3 mice per group). Results are representative of two independent experiments with n = 4 mice per group in total. d, Schematic of singlemitochondrion translation assay. e, Gating strategy for isolated mitochondria based on side scatter area (SSC-A), sizing beads and COXIV-AF488 signal, and for tdTomato + (RFP + ) AP/GC-derived mitochondria. f, Flow cytometry histogram plots depicting OPP-Alexa Fluor (AF) 647 signal in AP/GC-derived RFP + and naive B cellderived RFP − mitochondria from immunized Aicda-Tfam and Aicda-WT mice, and no OPP (DMSO) vehicle control from pooled samples. Quantification of OPP gMFI in RFP + versus RFP − mitochondria from immunized Aicda-WT mice (n = 3 mice) is shown. Data were pooled from three independent experiments. g, Normalized OPP in RFP + AP/GC-derived mitochondria from Aicda-Tfam and Aicda-WT mice (n = 3 per group). The experiment was performed in three batches and values were pooled after batch correction. h, Quantification of glucose dependence and FAO or AAO capacity of cells based on OPP incorporation into splenic IgD + GL-7 int APs (n = 3 mice per group) and CD38 − IgD − GL-7 + GC B cells from B-WT (n = 7) and B-Tfam mice (n = 5) treated ex vivo with metabolic inhibitors (oligomycin or 2-deoxy-d-glucose). Data were pooled from two independent experiments. See Supplementary Methods for details of the SCENITH calculations. i, Real-time OCR measurements (from the Mito Stress test) of 3 × 10 5 B cells stimulated overnight (anti-CD40 and IL-4) from B-Tfam (n = 4) and B-WT (n = 3) mice, with quantification of basal OCR and spare respiratory capacity. Results were pooled from two independent experiments. FCCP, carbonyl cyanide-p-trifluoromethoxyphenylh ydrazone; Rot/AA, rotenone/antimycin A. Statistical significance was calculated using a one-way ANOVA with Tukey's multiple comparisons test (a,b), an unpaired two-tailed t-test (f,i), an unpaired two-tailed t-test after batch correction (g) or a two-way ANOVA with Tukey's multiple comparisons test (c) or Šidák's correction (h). Data are presented as the mean ± s.e.m. Given these observations, we next tested whether either suppression of mtROS with the scavenger mitoTEMPO 45 or inhibition of MCU function with the ruthenium compound Ru265 (ref. 46) improved cell motility in B-Tfam B cells in response to CXCL12. MCU inhibition did not improve transwell migration but strikingly mitoTEMPO largely rescued the defect in B-Tfam cells (Fig. 7j).
Our data therefore collectively suggest that TFAM is required for proper cellular motility to enable entry into and movement within the GC, and that this effect is associated with regulation of ETC function and mtROS generation.

Tfam deletion in B cells prevents lymphoma
Having demonstrated an essential role for TFAM in B cell development and entry into the GC, we next asked whether it was required for the development of lymphoma. One of the most common mutations giving rise to DLBCL is translocation of MYC to immunoglobulin gene loci, leading to its unregulated expression. c-Myc is a key transcription factor regulating cell cycle and growth, cellular metabolism and mitochondrial biogenesis 47 . We reasoned that deletion of Tfam would counter the oncogenic effects of c-Myc overexpression. We employed the well-established Eμ-Myc transgenic mouse model of lymphoma, in which Myc is expressed under the control of the Igh enhancer 48 . Eμ-Myc mice develop lymphoma with high penetrance from a median age of 11 weeks. To understand what effects overexpression of Myc had on mitochondrial translation, we transferred established lymphoma cells from Eμ-Myc mice into WT CD45.1 congenic hosts to allow us to compare B cells from the same environment (Fig. 8a). We found very markedly higher expression of COXI in transferred lymphoma cells compared with WT B cells, with significantly upregulated TFAM (Fig. 8b,c). We next generated Cd79a-Cre × Tfam loxP × Eμ-Myc mice (B-Tfam-Myc). B-Tfam-Myc mice were completely protected from the development of lymphoma during the observation period of 30 weeks, compared with Eμ-Myc controls, which had a median survival of 12.5 weeks (Fig. 8d).
Finally, to establish whether inhibition of mitochondrial transcription and translation might be a therapeutic target in human lymphoma, we treated Daudi B cell lymphoma cells (originally arising from Burkitt lymphoma) with IMT1, a specific inhibitor of mitochondrial RNA polymerase (POLRMT), which functions along with TFAM to initiate mitochondrial RNA transcription 49 . We found that treatment with IMT1 led to a progressive reduction of COXI levels, with increasing COXI:succinate dehydrogenase B (SDHB) mismatch (Fig. 8e,f). Imaging of Daudi cells treated with IMT1 revealed mitochondrial enlargement and confirmed the loss of COXI (Fig. 8g). IMT1 reduced cell growth, inhibited cell cycle progression, increased mtROS levels and recapitulated the F-actin dysregulation we observed with Tfam deletion (Fig. 8h-k and Extended Data Fig. 8a). To inhibit mitochondrial translation, we used CHL. CHL reduced COXI expression and increased expression of F-actin, and the UPR mt protein LON peptidase 1 (LONP1) in Daudi cells (Fig. 8l,m and Extended Data Fig. 8b). Cell growth was substantially reduced (Fig. 8n).
These results define high rates of mitochondrial translation enabled by Tfam expression as an essential requirement for the development of B cell lymphoma and show the therapeutic potential of POLRMT and mitochondrial translation inhibition in human disease.

Discussion
In this study, we show that on entry to the GC, B cells dramatically remodel their mitochondria, increasing their mass and radically altering their morphology. As part of this transition, mitochondrial translation is highly active and we demonstrate that the nuclear-encoded mitochondrial transcriptional and translational regulator TFAM is not only required for B cell development but also for their entry into the GC program, proper spatial anchoring and for the subsequent development of lymphoma.
GC B cells have a highly distinct metabolism, predominantly relying on oxidative phosphorylation despite their very rapid rate of proliferation in a hypoxic microenvironment 6,8 . For rapidly dividing cells to maintain high mitochondrial mass in the face of dilution to their daughter cells, a high rate of mitochondrial protein translation and division must be maintained. We were surprised to find that TFAM is required for entry into the GC program itself; when deleted, there is a proportional accumulation of B cells with an AP phenotype, which have expressed Aicda but maintain markers of immaturity. Recently, the TCA metabolite ɑ-ketoglutarate was shown to be upregulated by interleukin-4 (IL-4) in B cells; this leads to epigenetic alteration of the Bcl6 locus 9 . It is therefore possible that TFAM is required for an initial burst of mitochondrial biogenesis to facilitate GC program entry.
Deletion of TFAM substantially altered the balance of ETC protein expression, with loss of mitochondrially encoded subunits and compensatory upregulation of nuclear-encoded proteins. This was reflected in metabolic disturbance, with impaired oxidative phosphorylation after activation, upregulation of glycolysis and, importantly, mtROS generation, which was seen even in unstimulated cells. Temporal control of Tfam deletion either very early in B cell development using Cd79a-Cre, or after activation with Aicda-Cre or in vitro with TAT-Cre did not lead to unexpected or major phenotypic differences but it is possible that adaptations might occur in a dynamic manner.
An increase in F-actin was consistent across genetic or pharmacological interference with mitochondrial transcription and translation. Dynamic actin cytoskeletal modification is essential for normal cell movement and its probable disruption through ROS accumulation, reversible with the addition of a ROS scavenger, probably contributes to the positioning and motility defects we observed 50 . This adds weight to the idea that mitochondria are intimately linked to cytoskeletal function and that this role may operate independently of ATP generation 51 . These defects collectively have the potential to compromise cellular interaction, in particular that with T FH cells required to enter the GC program, and for normal dynamics once within. We observed that Tfam −/− B cells failed to differentiate normally into GC B cells in an ex vivo T FH coculture system and yet were unaffected when CD40 ligation was artificially provided in the iGB culture system. The capacity of Tfam −/− iGB cells to present antigen to cognate T cells was normal, however, as were T FH cell numbers within GCs after immunization. How Tfam deletion in B cells may affect T cell interaction is therefore deserving of future study. We have used the iGB system developed by Nojima et al. 38 to precisely control deletion of Tfam before adoptive transfer; while this is an important and widely used tool to generate GC B cell precursors in vitro, which can then participate in the GC reaction in vivo, uncertainties remain about their fidelity to true APs and this would also benefit from future study.

Eµ
Although we have directly established the importance of TFAM as a regulator of B cell development and activation, the factors driving the counterintuitive switch to oxidative phosphorylation in GC B cells is to be uncovered, as do the signaling mechanisms controlling the differences in mitochondria we observed between GC microenvironments. Disruption of mitochondrial integrity also induces a phenotype associated with immune aging; to what extent the mechanisms we describe might hold true in the diminished humoral immune response seen with age is another area deserving of further exploration 20 .
We show that deletion of Tfam is sufficient to completely prevent the development of Myc-driven lymphoma. Although loss of Tfam at an early developmental stage leads to B cell lymphopenia, and thus the pool of B cells that may become malignant is reduced, the high penetrance of the model contrasted with the complete protection against lymphoma suggests that this is insufficient to explain the phenotype we observed. How TFAM acts to support lymphomagenesis requires further study but may be associated with its promotion of mitochondrial translation 52 . Our observation that inhibition of mitochondrial transcription and translation reduces growth of lymphoma cells suggest that this should be prioritized as a therapeutic target.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41590-023-01484-3.  Gt(ROSA)26Sor tm1(CAG-mCherry/GFP)Ganl (mito-QC) was a kind gift from I. Ganley (University of Dundee). B6.SJL.CD45.1 mice were provided by the central breeding facility of the University of Oxford. Male and female mice between the ages of 6 and 15 weeks were used. Mice were bred and maintained under specific pathogen-free conditions at the Kennedy Institute of Rheumatology, University of Oxford. Mice underwent regular checks to ensure they did not have any pathogenic microorganisms. They were housed in cages that had individual ventilation and were provided with items to stimulate their environment. The temperature was kept between 20 and 24 °C, with a humidity level of 45-65%. They were exposed to a 12-h cycle of light and darkness (7:00 to 19:00), with a 30-min period of dawn and dusk. All procedures and experiments were performed in accordance with the UK Scientific Procedures Act (1986) under a project license authorized by the UK Home Office (PPL no. PP1971784).

Flow cytometry and imaging flow cytometry
Flow cytometry was performed as described previously 54 v.7.12). Flow cytometry data was analyzed using FlowJo (FlowJo LLC).

Cell sorting
Naive B cells were isolated using the Pan B cell Isolation II Kit, anti-CD43, and/or anti-CD23 Microbeads (all Miltenyi Biotec) according to the manufacturer's instructions. Purity validated by flow cytometry was more than 90%. Isolation of the dark zone, light zone and gray zone subsets of GC B cells and T FH cells (CD19 − CD4 + CXCR5 + ICOS + GITR − ) was performed via fluorescence-activated cell sorting (FACS).
For some experiments, untouched GC B cells were isolated using a magnetic bead-based protocol as described by Cato et al. 55 . Briefly, spleens were collected from SRBC-immunized mice and single-cell suspensions were prepared in ice-cold magnetic-activated cell sorting (MACS) isolation buffer (PBS with 0.5% BSA + 2 mM EDTA) followed by ACK RBC lysis (Gibco) for 4 min at 20 °C with occasional mixing every 30 s. After washing, cells were labeled with anti-CD43 microbeads (Miltenyi Biotec) and biotinylated antibodies against CD38 and CD11c (clones 90 and N418, respectively; both eBioscience), followed by incubation with anti-biotin microbeads, and were subsequently run through a MACS LS column (Miltenyi Biotec). Purity was confirmed by flow cytometry and immunocytochemistry (ICC) and exceeded 95%.
When sorting for single-cell RNA sequencing (scRNA-seq), spleens were crushed using the rough ends of microscope slides to maximize cell yield. Subsequently, non-B cells were depleted using the Pan B cell Isolation II Kit (Miltenyi Biotec). Enriched cells were then incubated with viability dye, anti-CD16/32 (Fc Block) and surface flow antibodies, including markers for an exclusion (dump) channel (anti-CD3, anti-Gr1 and anti-CD11c); live Dump − tdTomato + cells were sorted by BD FACSAria.

Detection of ETC and UPR mt proteins with spectral flow cytometry
A complete list of antibodies is shown in the antibody table (see the spectral flow cytometry antibody panel section). All antibodies targeting intracellular mitochondrial proteins were directly conjugated and of mouse origin, with the exception of rabbit anti-Tfam monoclonal antibody (Abcam). Goat anti-rabbit AF405 Plus secondary antibody (Thermo Fisher Scientific) was used for detection. All antibodies used in the panel had been either validated for flow cytometry by the vendor or in the literature. Mouse anti-mitochondrial complex 1 antibody was conjugated in house using a Lightning-link PE-Cy7 Antibody Labeling kit (catalog no. 762-0005, Novus Biologicals). Cells were labeled with Zombie NIR viability dye (BioLegend) and Fc Block in PBS for 30 min on ice in 96-well V-bottom plates (Sarstedt). After washing, surface staining was performed in Brilliant stain buffer (catalog no. 563794, BD Biosciences) and FACS buffer (at a 1:1 ratio) for 30 min. Cells were then fixed in 4% freshly-made PFA at 20 °C for 15 min and permeabilized in freezer-cold methanol for 10 min with occasional vortexing. Anti-Tfam primary staining was performed in 50 μl FACS buffer supplemented with 2% goat serum for 45 min at 20 °C. After two washes, goat anti-rabbit secondary antibody and the remaining directly conjugated antibodies for ETC and UPR mt proteins were added for 30 min at 20 °C. The cells were subsequently washed in FACS buffer and acquired on a Nature Immunology Article https://doi.org/10.1038/s41590-023-01484-3 Cytek Aurora (5-laser) spectral flow cytometer. Exploratory pilot experiments were performed to determine the most suitable single-stained references (beads or cells) for individual marker-fluorochrome combinations. If cells were used as reference controls, they were obtained from matching organs (spleen or bone marrow). Reference controls were processed similarly to fully stained samples, with parallel fixation, permeabilization and washing steps. Acquired samples were unmixed using SpectroFlo and analyzed with FlowJo software. The 'Autofluorescence (AF) as a fluorescent tag' option was enabled during unmixing to minimize AF interference. In rare cases, minor adjustments were applied to unmixing on the SpectroFlo software. Geometric mean fluorescence intensity (gMFI) values for ETC and UPR mt proteins were calculated using FlowJo.

Single-mitochondrion translation assay
The technique was performed as described previously but with modifications 56 . Total B cells were isolated from the spleens of Aicda-WT and Aicda-Tfam mice immunized with NP-CGG using magnetic beads (Pan B cell Isolation Kit II). Experiments were performed in batches so that one Aicda-WT and one Aicda-Tfam mouse spleen were processed in each replicate. Cells were pulsed with OPP (20 μM,Jena Bioscience) in complete RPMI 1640 supplemented with the cytoplasmic translation inhibitor harringtonine (1 μg ml −1 , catalog no. ab141941-5mg, Abcam) for 45 min at 37 °C. After a wash with complete medium, cells were resuspended in ice-cold mitochondria isolation buffer (1×, 320 mM sucrose, 2 mM EGTA,10 mM Tris-HCl, at pH 7.2 in water, prepared as 2× stock and stored in aliquots at −20 °C). Cells were then homogenized with a Dounce homogenizer with a 2-ml reservoir capacity (catalog no. ab110171, Abcam), using ten strokes with a type B pretzel. The homogenizer was rinsed with distilled water before each sample was processed to avoid cross-contamination. Homogenates underwent differential centrifugation at 4 °C, with intact cells and isolated nuclei first pelleted at 1,000g for 8 min. The supernatant containing the mitochondria was then transferred into new microcentrifuge tubes and centrifuged at 17,000g for 15 min. Enriched mitochondria, which appeared as brown-colored pellets, were fixed in 1% PFA in 0.5 ml PBS on ice for 15 min, followed by a wash with PBS. Permeabilization and subsequent antibody staining was performed in 1× BD Perm/Wash buffer (diluted in MilliQ-filtered water) at 20 °C. Preadsorbed primary rabbit anti-RFP (1:250,Rockland) antibody labeling was performed at 20 °C for 30 min. After a single wash, the Click reaction was performed using the Click-iT Plus OPP AF647 Protein Synthesis Assay Kit (catalog no. C10458, Thermo Fisher Scientific) at 20 °C for 30 min. After a wash, mitochondria were resuspended in 1× BD Perm/ Wash buffer containing AF488-conjugated anti-COXIV (1:100, catalog no. 66110-1-Ig, Proteintech) and PE-conjugated donkey anti-rabbit IgG (1:200,catalog no. 406421,BioLegend). After washing with Perm/Wash buffer, mitochondria were resuspended in 250 μl filtered (0.2 μm) PBS and acquired using a BD Fortessa X-20 flow cytometer. The threshold for SSC-A (log-scale) was set to the minimum value (20,000) to allow acquisition of subcellular particles. Submicron Particle Size Reference Beads (catalog no. F13839, Thermo Fisher Scientific) were also used to identify mitochondria. For analysis, mitochondria were identified based on the COXIV and SSC-A properties; RFP + (mitochondria with AP-GC origin) and RFP − (mitochondria from naive B cells) were gated based on the anti-RFP antibody signal. The gMFI for the OPP-AF647 fluorescence was calculated for the RFP + and RFP − mitochondrial subsets and their ratio was used as a translation index and pooled after batch correction. Mitochondria from reporter-free cells, and those untreated with OPP, served as negative controls. In the validation experiments, 300 μg ml −1 CHL or ethanol vehicle was used to inhibit mitochondrial translation along with the cytoplasmic translation inhibitor harringtonine; mitochondria were detected by COXIV-AF488 (1:100, catalog no. 66110-1-Ig, Proteintech) and SDHA-AF594 antibodies (1:250, catalog no. ab170172, Abcam) as described above.

Immunohistochemistry (IHC)
Collected spleens were immediately transferred to Antigenfix (Diapath) solution in 1.5 ml and fixed overnight at 4 °C. The next day, spleens were washed in PBS followed by overnight incubation in 30% sucrose (in PBS) at 4 °C for cryoprotection. On the following day, spleens were snap-frozen in 100% methanol on dry ice and stored at −80 °C until cryosectioning at 8-12-μm thickness. Slides were then rehydrated in PBS at 20 °C, then permeabilized and blocked in PBS containing 0.1% Tween-20, 10% goat serum and 10% rat serum at 20 °C for 2 h. For panels requiring intracellular detection, Tween-20 was replaced by 0.5% Triton X-100. All primary antibody staining was performed overnight at 4 °C in PBS supplemented with 2% goat serum and 0.1% Triton X-100 (intracellular) or Tween-20 (surface); secondary staining was performed at 20 °C for 2 h the next day in PBS 0.05% Tween-20. TUNEL imaging was performed using the Click-iT Plus TUNEL In Situ Imaging Far Red kit according to the manufacturer's protocol (catalog no. C10619, Thermo Fisher Scientific). Slides were mounted with Fluoromount G (catalog no. 0100-01, Southern Biotech) and imaged with a ZEISS LSM 980 equipped with an Airyscan 2 module. See the relevant image analysis section in Supplementary Methods.

ICC
Isolated cells were transferred onto 18-mm coverslips coated with 0.01% (weight/volume) poly-l-lysine (catalog no. P8920-100ML, 10× stock, Merck) and incubated for 10 min at 37 °C to ensure cell attachment. Cells were then fixed in prewarmed 1× PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA and 4 mM MgSO 4 ·7H 2 0, pH 6.8) with 4% PFA for 10 min at 37 °C, followed by permeabilization and blocking in 0.2% Triton X-100 with 10% goat serum for 30 min. Primary antibody labeling was performed overnight at 4 °C; secondary antibody staining was performed for 45 min at 20 °C (see antibody table). For the mitochondrial transcription assay based on 5-EU incorporation, isolated untouched naive B cells and GC B cells were resuspended in complete RPMI 1640 supplemented with 1 mM 5-EU (catalog no. C10330, Thermo Fisher Scientific) and transferred to 18-mm coverslips coated with poly-l-lysine. After incubation for 45 min, cells were briefly washed and fixed in warm 4% PFA diluted in PHEM buffer. After permeabilization and blocking for 30 min, incorporated 5-EU was detected by the Click-iT RNA AF594 Imaging Kit (catalog no. C10330, Thermo Fisher Scientific). Intracellular antibody labeling was performed after the Click reaction to minimize the interference of Click reagents with fluorochromes. For in vivo measurement of mitochondrial RNA synthesis, 2 mg 5-EU (Thermo Fisher Scientific) was injected intraperitoneally on day 12 after SRBC immunization and similar preparation and labeling steps described for the ex vivo 5-EU assay were followed. DAPI (catalog no. D8417-1MG, Sigma-Aldrich) staining was performed at 1 μM at 20 °C for 5 min, followed by a brief wash and mounting in Fluoromount G. For MitoTracker staining, cells were labeled with MitoTracker Red CMX ROS (150 nM, catalog no. M7512, Thermo Fisher Scientific). Slides were imaged with a ZEISS LSM 980 equipped with an Airyscan 2 module. See the relevant image analysis section in Supplementary Methods.
iGB adoptive transfer iGB cells were generated as above. On day 4, fibroblasts and in vitro-differentiated plasmablasts (generally less than 10% frequency) were removed using biotinylated anti-H-2Kd (catalog no. 116604, BioLegend) and anti-CD138 (catalog no. 142511, BioLegend) followed by anti-biotin microbeads (Miltenyi Biotec); negative selection was performed using LS columns (Miltenyi Biotec). In some experiments, FACS was used to purify tdTomato + iGB cells. For competitive experiments, purified WT iGBs (CD45.1/2 + ) were mixed 1:1 (ratio confirmed by flow cytometry before transfer) with CD45.2 + tdTomato + Tfam −/− iGBs and injected intravenously (6 × 10 6 total cells in competitive setting or 3 × 10 6 cells in noncompetitive setting) into CD45.1 + or CD45.2 + congenic hosts that were immunized with SRBC according to the enhanced protocol to maximize the recruitment of transferred iGB cells into the GC. On day 6 after iGB adoptive transfer, spleens were collected and analyzed by flow cytometry and confocal imaging to assess GC entry.

Transwell migration assay
A total of 5 × 10 5 enriched total B cells isolated from SRBC-immunized Aicda-Tfam and Aicda-WT mice were placed in a 6.5-mm transwell chamber with 5-μm pore size (catalog no. 3421, Corning) and incubated at 37 °C for 2 h in the presence of murine CXCL12 (200 ng ml −1 , BioLegend) or CXCL13 (1 μg ml −1 , BioLegend) in complete RPMI 1640. The relative chemotaxis/migration index was calculated as follows: percentage of GC B cells (CD38 − GL-7 + tdTomato + ) in migrated live total cells divided by the percentage of GC B cells in total input cells . A total of 2 × 10 5 purified total B cells from B-Tfam and B-WT mice were placed in a 6.5-mm transwell chamber with 5-μm pore size and incubated for 5 h in the presence or absence of murine CXCL12 (100 ng ml −1 , BioLegend) with or without mitoTempo (100 μM, Merck) and Ru265 (30 μM,Merck). After 5 h, cells were incubated with LIVE/DEAD and anti-B220 AF488 antibody and resuspended in 100 μl in 96-well V-bottom plates and acquired on a Cytek Aurora flow cytometer at high-flow setting with a stopping volume of 60 μl. Technical duplicates were also included. Specific migration (%) was calculated according to this formula: 100 × (number of B220 + cells migrated in response to CXCL12 − number of B220 + cells migrated in the absence of CXCL12)/number of input B cells.

scRNA-seq analysis
Approximately 17,000 cells per sample were loaded onto the 10X Genomics Chromium Controller (Chip K). Gene expression and BCR sequencing libraries were prepared using the 10X Genomics Single Cell 5′ Reagent Kits v2 (Dual Index) according to the manufacturer's user guide (CG000330 Rev B). The final libraries were diluted to approximately 10 nM for storage. The 10-nM library was denatured and further diluted before loading on the NovaSeq 6000 sequencing platform (v.1.5 chemistry, 28/98 bp paired-end, Illumina) at the Oxford Genomics Centre.
Filtered output matrices from Cellranger v.6.0.1 were loaded in Seurat v.4.1.0. Cells with more than 5% mitochondrial reads and fewer than 200 genes were removed from the analysis. Data were normalized and transformed with SCTransform, with regression of cell cycle phase and mitochondrial reads, and integrated with the FindIntegrationAnchors and IntegrateData functions. The RunUMAP, FindNeighbors and FindClusters functions were used to cluster cells. Marker genes between clusters were identified using the FindAllMarkers function. Two small contaminant clusters (less than 1% of cells) were identified based on the expression of non-B cell genes and were removed from subsequent analyses. Clusters were identified by expression of canonical markers. Cluster proportions were calculated using DittoSeq. For visualization of uniform manifold approximation and projection (UMAP), equal number of cells from each experimental condition were displayed by random downsampling. Differential gene expression between conditions was calculated using the FindMarkers function with the 't-test' parameter. Pathway analysis was performed using the R package single-cell pathway analysis (SCPA). Pseudobulk differential gene expression between individual biological replicates was performed using EdgeR after count aggregation across cells using Scuttle.
Filtered contig outputs generated by Cellranger v.6.0.1 from cells processed in the Seurat workflow above were combined, filtered and visualized using scRepertoire v.1.4. For quantification of mutational load, the Immcantation pipeline was used. Germline segment assignment was performed with Change-O; the SHM count was calculated using SHazaM.

Statistical analysis
The statistical tests used are indicated in the respective figure legends, with error bars indicating the mean ± s.e.m. P ≤ 0.05 was deemed to indicate significance. Analyses were performed with Prism 9 (GraphPad Software) or R v.4.1. No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications 57 . The distribution of data was determined using normality testing to determine appropriate statistical methodology; otherwise, it was assumed to be normally distributed. For the in vivo experiments, we matched the sex and age of the mice in the experimental batches; however, other modes of randomization were not performed. Data collection and analysis were not performed blind to the conditions of the experiments in most of the experiments. Mice with complete absence of GCs and lacking alum spots after immunization were considered as failed intraperitoneal immunization and therefore excluded from the analysis. In selected experiments, batch correction was performed with the R package Batchma.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Nature Immunology
Corresponding author(s): Alex Clarke NI-A34325B Last updated by author(s): Mar 5, 2023 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.
nature portfolio | reporting summary

March 2021
Flow Cytometry Plots Confirm that: The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.

Methodology
Sample preparation Flow cytometry and imaging flow cytometry Briefly, harvested spleens were injected with ice cold PBS and mashed through a 70μm strainer (Falcon) or crushed between the frosted ends of microscope slides. For Peyer's patch dissociation, a 40μm strainer (VWR) was used. RBCs were depleted by incubating splenocytes with ACK Lysis Buffer (Gibco) for 3-4 mins at 20°C. Single cell suspensions were incubated with Fixable Viability Dye eFluor™ 780 (eBioscience) in PBS, followed by FcBlock (5 mins) and surface antibodies (30 mins on ice) in FACS buffer (PBS supplemented with 0.5% BSA and 2mM EDTA). For intracellular staining, cells were fixed at 20°C with 4% freshly prepared PFA (Cell Signaling) for 15 mins and permeabilized with methanol (90% ice cold for 10 mins on ice with frequent vortexing) unless specified otherwise. Phalloidin-based F-Actin (Thermofisher, cat no: A30107) staining was performed using BD Perm/Wash reagent (BD, Cat. No. 554723) following 4% PFA fixation. For in vivo cell cycle analysis, 5ethynyl-2'-deoxyuridine (EdU) (1 mg, Thermofisher, cat no: A10044) was injected intraperitoneally and mice were sacrificed after 2.5 h. Cells were stained for surface markers, fixed and permeabilized then labelled using Click chemistry according to manufacturer's instructions (Click iT Plus EdU Flow cytometry kit, Thermo Fisher, cat: C10634). FxCycle Violet (Thermo Fisher, cat: F10347) reagent was used for cell cycle characterization. For mitochondrial superoxide deep red (mtSOX, Dojindo) uptake, following viability dye and surface staining, cells were resuspended in warm complete RPMI supplemented with mtSOX (10μM) and incubated for 30 mins at 37°C. Cell were washed twice before flow cytometry acquisition. Flow cytometry was performed on BD Fortessa X-20 or LSR II instruments (both BD), or using a Cytek Aurora (5 laser) spectral flow cytometer. For imaging cytometry, single cell suspensions were prepared from spleens of MitoQC mice, incubated with live-dead dye, stained for surface markers, and then fixed with 4% PFA. Washed cells were then resuspended in 50μl FACS Buffer and run on an Amnis ImageStreamX Mark II Imaging Flow Cytometer and analyzed with IDEAS (EMD Millipore) and FCS Express software (v7.12). Flow cytometry data was analyzed using FlowJo (BD).

Detection of ETC/UPRmt proteins with spectral flow cytometry
A complete list of antibodies is shown in the antibody table. All antibodies targeting intracellular mitochondrial proteins were directly conjugated and of mouse origin, with the exception of rabbit anti-Tfam monoclonal antibody (Abcam). Goat antimouse AF405 Plus secondary antibody (ThermoFisher) was used for detection. All antibodies used in the panel had been either validated for flow cytometry by the vendor or in the literature. Mouse anti-mitochondrial complex 1 antibody was conjugated in house using a PE-Cy7 Lightning-link kit (Novus Bio, cat: 762-0005). Cells were labelled with Zombie NIR viability dye (Biolegend) and Fcblock in PBS for 30 min on ice in 96-well V-bottom plates. Following washing, surface staining was performed in Brilliant staining buffer (BD, cat: 563794) for 30 min. Cells were then fixed in 4% freshly-made PFA at 20°C for 15 mins and permeabilized in freezer-cold methanol for 10 mins with occasional vortexing. Anti-Tfam primary staining was performed in 50μl FACS Buffer supplemented with 2% goat serum for 45 mins at 20°C. Following two washes, goat antirabbit secondary antibody and the remaining directly conjugated antibodies for ETC/UPRmt proteins were added for 30 mins at 20°C. The cells were subsequently washed in FACS buffer and acquired on a Cytek Aurora (5 laser) spectral flow cytometer. Exploratory pilot experiments were performed to determine the most suitable single-stained references (beads or cells) for individual marker-fluorochrome combinations. If cells were used as reference controls, they were obtained from matching organs (spleen or bone marrow). Reference controls were processed similarly to fully stained samples, with parallel fixation, permeabilization, and washing steps. Acquired samples were unmixed using SpectroFlow and analyzed with FlowJo. The 'Autofluorescence (AF) as a fluorescent tag' option was enabled during unmixing to minimize AF interference. In some rare cases minor adjustments were applied to unmixing on the SpectroFlow software. gMFI values for ETC/UPRmt proteins were calculated using FlowJo. For some experiments, untouched GC B cells were isolated using a magnetic bead-based protocol as described.55. Briefly, spleens were harvested from SRBC-immunized mice, and single cell suspensions were prepared in ice cold MACS isolation buffer (PBS with 0.5% BSA + 2mM EDTA) followed by ACK RBC lysis (Gibco) for 4 mins at 20°C with occasional mixing every