BTG1 mutation yields supercompetitive B cells primed for malignant transformation

Abstract

Multicellular life requires altruistic cooperation between cells. The adaptive immune system is a notable exception, wherein germinal center B cells compete vigorously for limiting positive selection signals. Studying primary human lymphomas and developing new mouse models, we found that mutations affecting BTG1 disrupt a critical immune gatekeeper mechanism that strictly limits B cell fitness during antibody affinity maturation. This mechanism converted germinal center B cells into supercompetitors that rapidly outstrip their normal counterparts. This effect was conferred by a small shift in MYC protein induction kinetics but resulted in aggressive invasive lymphomas, which in humans are linked to dire clinical outcomes. Our findings reveal a delicate evolutionary trade-off between natural selection of B cells to provide immunity and potentially dangerous features that recall the more competitive nature of unicellular organisms.

Fig. 1.. Characterization of BTG1 missense mutations in DLBCL.

(A) Quantile-quantile plot of observed and expected P values from a gamma-Poisson distribution for single nucleotide variants across 101 germline-matched whole genome–sequenced DLBCLs. Mutated genes with FDR < 0.001 are depicted with a red dot. (B) BTG1 protein and frequency, type, and location of mutations in n = 272 unique cases out of n = 2407 DLBCLs. LxxLL motifs mediate interaction with nuclear receptors. (C) BTG1 homology model (residues 1 to 129) based on a structural reference of BTG2 (Protein Data Bank ID 3DJU) showing mutation frequency. N- and C-terminal ends are denoted. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Fig. 2.. Btg1^Q36H generates supercompetitor GC B cells.

(A) In vivo competitive assay schematic and representative flow cytometry plots for gating Q36H and CREneg from transferred germinal center (GC) B and non-GC B cells. Graph represents pooled mice from three independent experiments, each with n = 2 or 3 mice per time point. Mean ± SD, paired t test (two-tailed). (B) Competitive assay between mature B cells expressing Btg1^Q36H versus Btg1^WT from Rosa26 (R26) knock-in alleles that differ by a single nucleotide. Representative flow cytometry plots for gating antigen-specific (Ag-spe, lambda⁺NP⁺) naïve B (NB, CD138⁻B220⁺IgD⁺) and GC B (FAS⁺CD38⁻) WTKI and Q36H cells at day 14 after immunization. Graph represents pooled mice from two independent experiments, representative of three [day 7 (d7) and d14] or two (d10) independent experiments, each with n = 3 to 5 mice per time point. Mean ± SD, paired t test (two-tailed). (C) Experimental schematic for the adoptive transfer of B cells with an endogenous B cell receptor repertoire to compete with wild-type recipient cells. Representative flow cytometry plots on live B cells (DAPI-B220⁺) at day 21 after NP-KLH immunization, for gating transferred cells (CD45.2) of the indicated genotypes within GC B cells. Graph represents mice combined from three independent experiments, with 9 to 14 mice total per genotype and time point. Mean ± SD, unpaired t test (equal SD, two-tailed). s.c., subcutaneously.

Fig. 3.. Mutant BTG1 induces MYC-related biosynthetic programs.

(A) Gene set enrichment analysis (GSEA) network showing canonical and hallmark gene sets positively enriched in Q36H versus CRE up-regulated genes [normalized enrichment score (NES) ≥ 1.25; FDR < 0.05]. Distance between two enriched gene sets was calculated as the Jaccard index (size intersection: size union) using leading-edge genes. The circle size represents the degree of connectivity to which each gene set is attached to others. (B) GSEAs showing LZ-to-DZ recycling GC B, MYC, and mTORC1 activation signatures enrichment in Q36H versus CRE mouse GC B cells (left) and in BTG1 mutant versus WT DLBCL cases in the BC Cancer Agency (BCCA) cohort (right) (23). LZ-to-DZ recycling GC B signature is DECP_UPREG [n = 221 mouse genes (left); n = 201 human orthologs (right)]; MYC signature is SCHUHMACHER_MYC_TARGETS_UP [n = 80 mouse orthologs (left); n = 75 genes represented in the BCCA dataset (right)]; mTORC1 signature is PENG_RAPAMYCIN_RESPONSE_DN [n =242 mouse orthologs (left); n = 230 genes represented in the BCCA dataset (right)]. Enrich., enrichment score. (C) Graph showing quantification of RNA extracted from 200,000 mouse GC B cells. One experiment, n = 4 and 5 mice per genotype. Mean ± SD, unpaired t test (equal SD, two-tailed). (D) Representative flow cytometry histograms showing cell size measurement as forward scatter area (FSC-A) in GC B (FAS⁺CD38⁻) and non-GCB (FAS⁻CD38⁺) cells (left). Graph showing geometric mean (gMean) of FSC-A from one experiment, n = 5 and 6 mice per genotype, representative of at least three independent experiments (right). Mean ± SD, unpaired t tests (two-tailed). (E) Experimental schematic and representative flow cytometry plots showing Myc^GFP+ cells (green) in centroblasts (CB) and centrocytes (CC). (F) Bar plot showing proportion of Myc^GFP+ cells in CC. One experiment with n = 3 and 4 mice per genotype, representative of at least three independent experiments. Mean ± SD, unpaired t test (two-tailed). (G) Genes (n = 201) depicted with red dots present higher expression (FC > 1.2) in GFP^Myc+ versus GFP^Myc− GC B cells comparing Q36H mice (y axis) to CRE mice (x axis). (H) Hypergeometric mean analysis for the 201 genes identified in (G). The Myc immediate early signature consists of Myc-dependent genes induced by 8 hours in Myc WT/flox B cells (39). Genes in this and other gene signatures are included in supplementary table S4.

Fig. 4.. Q36H mutation disrupts BTG1-MYC mRNA association and enhances MYC protein synthesis kinetics.

(A) Heatmap showing n = 732 BTG1^WT-associated (BTG1^WT RIP/input log2FC > 2, q < 0.05) and BTG1^Q36H-lost (BTG1^Q36H/BTG1^WT log2FC > 2, q < 0.05) transcripts by RIP-seq, using V5 antibody in SU-DHL4 DLBCL cells overexpressing V5-tagged BTG1^WT or BTG1^Q36H. MYC and leading edge genes from positively enriched LZ-to-DZ recycling GC B, MYC, and mTORC1 signatures in BTG1^Q36H versus BTG1^WT are labeled. (B) Signatures enriched for up-regulation of the n = 732 BTG1^WT-associated and BTG1^Q36H-lost transcripts as determined by hypergeometric mean analysis. (C) RIP-qPCR for MYC mRNA in indicated SU-DHL4 V5-tagged DLBCL cells. Mean ± SD of enrichment in RIP over input samples, normalized to BTG1^WT-V5. Three independent experiments, each with n = 2 or 3 independently generated lines per genotype. Unpaired t test (equal SD, two-tailed, versus EGFP or as indicated). (D) Representative MYC flow cytometry histograms of SU-DHL4 DLBCL cells expressing V5-tagged BTG1^WT (WT-V5) or BTG1^Q36H (Q36H-V5) and treated with 5 μM MG132 for 0 or 2 hours. Proportions of MYC-positive cells (MYC⁺) are indicated. Staining with isotype control is included. Graphs show proportion of MYC⁺ cells (left) or MYC protein levels within MYC⁺ cells (right) as measured by flow cytometry in three independent experiments, each with n = 2 or 3 independently generated lines per genotype, shown as fold change relative to WT-V5 0 hour mean value per experiment. Mean ± SD, two-way analysis of variance (ANOVA) for Q36H-V5 versus WT-V5. gMeanFI, geometric mean fluorescence intensity. (E) Polysome profile of BTG1^WT-V5 and BTG1^Q36H-V5 expressing SU-DHL4 cells. Mean (line) ± SD (shade) from n = 3 independently generated lines per genotype except fractions 4 to 9: n = 2 Q36H lines owing to technical loss of one line. One experiment representative of two independent experiments. (F) Bar plot shows the area under the curve (AUC) for indicated portions of the polysomal traces from (E), as percent of the total AUC average per genotype. Mean ± SD, unpaired t test (equal SD, two-tailed). (G) Absolute MYC mRNA levels from fractions in (E) and (F), pooled as indicated. n = 2 independently generated lines per genotype. Mean ± SD, unpaired t test (equal SD, two-tailed).

Fig. 5.. Faster S phase completion and DZ program commitment in Btg1^Q36H LZ-to-DZ recycling cells.

(A) Experimental design for targeted single-cell RNA sequencing for n = 496 genes in competing Q36H and CREneg GC B cells from n = 4 mice pooled into three biological replicates. (B) Dark zone (DZ), light zone (LZ), and LZ-to-DZ recycling GC B cells were defined on the basis of signatures enrichment and of Mki67 and Cd86 marker expression levels, projected onto the uniform manifold approximation and projection (UMAP) distribution of cells (n = 2982). Centroblast (CB) and centrocyte (CC) signatures correspond to genes down-regulated or up-regulated in CC versus CB, respectively. Myc^GFP+ signature represent genes up-regulated in Myc^GFP+ versus Myc^GFP− CRE control GC B cells from Fig. 3E; mTORC1 signature is PENG_RAPAMYCIN_RESPONSE_DN; DECP signature is LZ_DECP_upreg and Myc immediate early signature consists of Myc-dependent genes induced by 8 hours in Myc WT/flox B cells. Genes in these signatures are included in table S4. (C) Distribution of cells enriched for G1, S, and G2/M signatures (top) and of Q36H and CREneg GC B cells differential density across a Slingshot cell cycle pseudotime axis. Difference in distribution of Q36H and CREneg cells by Wilcoxon test. (D to F) Distribution of Q36H and CREneg GC B cells based on their enrichment for (D) LZ-to-DZ recycling GC B signature, (E) DZ signature, or (F) G2/M signature, across the Slingshot cell cycle pseudotime axis. (D) Pseudotime units 9 to 14 correspond to LZ-to-DZ recycling GC cells (score > 0). [(E) and (F)] Pie charts represent Q36H or CREneg GC B cells with (E) a DZ score > 0 or (F) a G2/M score > 0, as a proportion of LZ-to-DZ recycling cells (pseudotime units 9 to 14) from (D). Chi-square P value. (G) Experimental design for in vivo cell cycle profiling by EdU/BrdU dual staining of competing Q36H and CTRL GC B cells (top). Representative flow cytometry plots for CC and CB in S phase, mid-late S phase, and post S phase (middle). Graphs show data pooled from three independent experiments, with n = 3 or 4 mice each (bottom). Mean ± SD, paired t test (two-tailed). (H) Experimental design to measure S phase-experienced Q36H and CRE Myc^GFP+ cells in vivo by EdU staining coupled with immunofluorescence (IF). Representative IF images are shown. Bar plot shows quantification from two independent experiments, each with n = 2 or 5 mice per genotype. Each dot represents one individual field of view with a minimum five cells per field of view. Mean ± SD, unpaired t test (two-tailed).

Fig. 6.. Accelerated LZ-DZ recycling kinetics in Btg1^Q36H GC B cells.

(A) In silico readout for the number of LZ-DZ cycles undergone by competing Btg1 mutant versus WT GC B cells at day 10 after immunization, using the GC mathematical model in which mutant GC B cells are provided with 1.12 times faster T cell help response (faster up-regulation of Myc and mTORC1) and 21% shorter S phase. Each dot is a readout from a single simulation (n = 300). Difference by Wilcoxon test considering the mutant and WT readout from the same simulation as paired samples. (B) Schematic of the system used to deliver targeted T_FH cell help to competing GC B cells in vivo. (C) Experimental design for the targeted delivery of T_FH cell help to competing DEC205^+/+ WT and Q36H GC B cells in vivo and tracking their centroblast (CB) versus centrocyte (CC) identity over time. (D) Representative gating flow cytometry plots for C. (E) Graph shows pooled data for (C) and (D), from two experiments, each with n = 3 to 5 mice per time point. Paired t test, two-tailed. (F) Schematic model. More-rapid GC LZ-DZ cycles in Btg1 mutant (MUT) GC B cells, because of faster response to T_FH cell help via accelerated Myc protein induction, more rapid S phase completion and earlier commitment to DZ transcriptional program, explain the progressive competitive fitness gain of Btg1^Q36H GC B cells over the course of the GC reaction. Rec, LZ-to-DZ recycling.

Fig. 7.. Mutant BTG1 drives formation of highly aggressive B cell lymphomas in mice and humans.

(A) Experimental design for testing contribution of the R26^{lsl.Btg1Q36H/+} allele to Bcl2-driven lymphomagenesis. (B) Kaplan-Meier curves depicting overall survival of groups described in (A) in days after transplantation, assessed by either time of death or euthanasia upon sickness development. Log-rank (Mantel-Cox) test performed as indicated. (C) Image of n = 5 spleens per group at 8 months after transplantation. (D) Consecutive spleen sections from Bcl2 and Bcl2+Q36H mice at 8 months after transplantation stained with H&E or by immunohistochemistry (IHC) for B cells (B220 antibody) or GC B cells (PNA). Representative of n = 5 mice per genotype. (E) Consecutive lymph node sections from Bcl2 and Bcl2+Q36H mice at 8 months after transplantation stained with H&E or by IHC for proliferating cells (Ki67). Representative of n = 5 mice per genotype. (F) Consecutive liver sections from Bcl2 and Bcl2+Q36H mice at 8 months after transplantation stained with H&E or by IHC for B cells (B220). Representative of n = 4 mice per genotype. (G) B cell lineage trees from BCR sequences analysis (ImmunoSeq) in Bcl2 and Bcl2+Q36H mice at 8 months after transplantation. The B cell clone with the most sequence reads for each mouse (mouse number indicated in the bottom) is shown. Branch lengths represent the estimated number of somatic hypermutations between nodes. Scale bar denotes 10 mutational events. Trees from n = 2 mice, representative of n = 5 mice per genotype (fig. S16F), are shown. (H) H&E staining on indicated tissue sections from Bcl2 and Bcl2+Q36H moribund mice. Representative of n = 6 mice per genotype. Low-grade follicular lymphoma–like centrocytes shown in Bcl2 mice. In Bcl2+Q36H mice, white arrowheads depict large immunoblastic cells; yellow, plasmacytoid cells; and orange, a large-size DLBCL-like cell. (I) Kaplan-Meier curves depicting overall survival of ABC-DLBCL patients combined from publicly available cohorts (9, 23, 24), based on their BTG1 status. Log-rank test was performed. Censored events are indicated below the graph. (J) Univariate and multivariate analyses in ABC-DLBCL cases from the same combined cohorts as in (I). HR, hazard ratio with 95% confidence interval. P value from Wald chi-square test for H0:HR = 1.