Loss of signalling via G[agr]13 in germinal centre B-cell-derived lymphoma

We sequenced the S1PR2 coding region in 117 GCB-DLBCL, 31 Burkitt’s lymphoma and 68 activated B-cell-like (ABC)-DLBCL samples. Twelve S1PR2 coding mutations were identified in the GCB-DLBCL samples versus one in each of the Burkitt’s lymphoma and ABC-DLBCL cohorts (Supplementary Tables 1 and 2). The majority of GCB-DLBCL mutations were in conserved transmembrane residues (Fig. 1a) and all were predicted to be structurally damaging. Cell-line transduction experiments showed that five of eight tested mutations disrupted S1PR2 protein expression (Fig. 1b and Extended Data Fig. 1a–c). These same mutations disrupted S1P-mediated inhibition of CXCL12-induced pAkt and migration (Fig. 1c, d). One additional mutant, R147C, which was expressed at levels similar to wild type (WT) (Fig. 1b and Extended Data Fig. 1), showed a strongly reduced ability to support S1P-mediated inhibition of pAkt and migration (Fig. 1c, d and Extended Data Fig. 1d, e). These observations suggested that tumours harbouring single mutant S1PR2 alleles (Extended Data Fig. 2) are often likely to be functionally heterozygous for S1PR2. Using a mixed bone-marrow chimaera system in mice3, S1pr2 heterozygous B cells showed marked expansion in the germinal centre (GC) relative to the follicular compartment in mesenteric lymph nodes and Peyer’s patches of unimmunized mice (Fig. 1e and Extended Data Fig. 3a, b). Overexpression of WT S1PR2 repressed the outgrowth of S1pr2+/− GC B cells and this was also seen for mutant R329C, whereas the R147C mutation caused the receptor to lose GC growth suppressive activity (Fig. 1f and Extended Data Fig. 3c, d). On the basis of molecular simulation analysis (Supplementary Information and Extended Data Fig. 3e–g) we propose that the R147C S1PR2 mutant cannot attain the active conformation necessary for G-protein recruitment and signalling. Gα12 and Gα13 often function redundantly in transmitting G-protein-coupled receptor signals8. Transcripts for both G-proteins are upregulated in GC B cells, with Gna13 transcripts appearing more abundant (Extended Data Fig. 4a). In accord with recent whole-exome sequencing studies that reported mutations in GNA13 but not GNA12 (refs 5, 6 and 9, 10, 11), we found frequent GNA13 coding mutations in GCB-DLBCL and Burkitt’s lymphoma biopsy samples, with a number of biallelic cases (Supplementary Table 2 and Extended Data Fig. 2). Analysis of mixed bone-marrow chimaeras revealed that Gα13 deficiency was sufficient to confer a GC B-cell growth advantage in mesenteric lymph nodes and to a lesser extent in Peyer’s patches (Fig. 1g and Extended Data Fig. 4b). Gα13-deficient mesenteric lymph node GC B cells showed increased pAkt relative to WT when incubated ex vivo with CXCL12 and S1P (Fig. 1h). Deficiency in the Gα13 effector, Arhgef1 (p115 RhoGEF or Lsc), led to a similar defect in the ability of S1P to repress chemokine induced pAkt (Fig. 1i). To determine whether loss of Gα13 in B cells could promote lymphomagenesis, we allowed a cohort of Gna13-deficient mice to age. At 1 year, 10 out of 18 Gna13-deficient mice showed a greater than tenfold expansion of GC B cells compared with littermate controls (Fig. 1j, k), and at least five of the outgrowths appeared clonal (Extended Data Fig. 4c). Three of the Gna13-deficient animals showed massive mesenteric lymphadenopathy (Fig. 1l and data not shown), with evidence in one case (number 307) of spleen and Peyer’s patch involvement (Fig. 1l and Extended Data Fig. 4c–e). Immunophenotyping of the Gα13-deficient tumours confirmed they were of GC origin (Extended Data Fig. 4f). To test the conservation of the Gα13-signalling pathway in human GC B cells, we performed gene rescue experiments in GCB-DLBCL cell lines. Sequencing of S1PR2, GNA13 and ARHGEF1 in a panel of GCB-DLBCL cell lines identified several with deleterious mutations in these genes (Supplementary Table 3 and Extended Data Fig. 5a). The mutations in GNA13 matched those previously described and were associated with reduced protein levels6. ARHGEF1 mutations have not previously been reported, probably because the large size (~24 kilobases) of this 27-exon gene and its multiple splice variants and low transcript abundance make sequence analysis difficult. Remarkably, 10 out of 20 cell lines with analysable ARHGEF1 sequence showed mutations in this gene, several of which resulted in premature stop codons (Supplementary Table 3 and Extended Data Fig. 5a). Using retroviral transduction to restore gene expression, we established that loss of S1PR2, Gα13 and ARHGEF1 were each sufficient to disrupt S1P-mediated suppression of pAKT and, in the case of cell lines that were migratory, to disrupt S1P-mediated inhibition of migration (see Supplementary Information and Extended Data Fig. 5). The mechanisms by which malignant GC B cells can exit the GC niche and lymphoid organ to spread among multiple lymph nodes or to systemic sites such as bone marrow have not been defined. Consistent with a lack of migration inhibition by S1P (Fig. 2a), mice lacking Gα13 in B cells showed marked disruption of GC architecture in mesenteric lymph nodes (Fig. 2b and Extended Data Fig. 6a). In a mixed transfer system, Gα13-deficient GC B cells were excluded from the interior of otherwise WT GCs (Extended Data Fig. 6b). Remarkably, Gα13-deficient GC B cells were readily detected in lymph and to a lesser extent in blood while WT GC B cells were absent from circulation (Fig. 2c). In mixed bone-marrow chimaeras, Gα13-deficient GC B cells were again detectable in the lymph, indicating that Gα13 was needed intrinsically in GC B cells to inhibit egress (Fig. 2d). Analysis of Arhgef1-deficient mice and chimaeras revealed a similar disruption of mesenteric lymph node GC architecture (Fig. 2b and Extended Data Fig. 6c) and GC B-cell appearance in lymph and blood (Fig. 2e, f). In contrast, S1PR2-deficient GC B cells were not significantly higher in lymph relative to littermate controls (Fig. 2g). Analysis of mice expressing constitutively active myristoylated Akt (myrAkt) or overexpressing BCL2 in B cells established that increased GC B-cell survival was not sufficient to lead to dissemination (Supplementary Information and Extended Data Fig. 7). GNA13 mutations and BCL2 rearrangements and potentially activating mutations frequently occur together in GCB-DLBCL6, 12. GC B cells in mice with combined Gα13 deficiency and BCL2 overexpression showed enhanced ex vivo survival (Fig. 3a), increased numbers (Fig. 3b), wider dispersal throughout the follicle and interfollicular regions in mesenteric lymph nodes (Extended Data Fig. 7f) and twofold increased frequencies in lymph and blood (compare Figs 3c and 2c), compared with cells in Gα13-deficient mice. To examine requirements for GC B-cell persistence after arriving at a distant site, we bypassed the egress step and intravenously transferred mesenteric lymph node cells to congenically distinct recipients. Transferred WT GC B cells were essentially undetectable in recipient spleen and bone marrow after 6 hours (Fig. 3d, e) and Gα13 deficiency alone was insufficient to cause a significant increase in their number (Fig. 3e). BCL2-overexpression alone caused an elevation in GC B-cell frequency in recipient spleens but not bone marrow (Fig. 3e). Loss of Gα13 combined with BCL2-overexpression led to greater accumulation of transferred GC B cells in spleen and now led to an increase in their frequency in bone marrow (Fig. 3e). This combinatorial effect probably reflects an ability of Gα13 deficiency and BCL2-overexpression to cooperate in promoting survival of GC B cells outside the GC niche (Fig. 3a). To determine whether GC B cells could seed distant lymph nodes after entry into lymphatics, we transferred mesenteric lymph node cells intraperitoneally. Small numbers of Gα13-deficient, but not WT, GC B cells were detectable in the draining parathymic lymph nodes after 6 hours (Fig. 3f). In this case, recovery of Gα13-deficient GC B cells was not enhanced by the BCL2 transgene. Bone marrow involvement occurs in a fraction of GCB-DLBCL patients and is a predictor of worse disease13. In some year-old Gα13-deficient mice showing mesenteric lymph node tumours, GC B cells could be detected in the bone marrow (Fig. 3g, h). Moreover, in aged BCL2-tg Gna13 knockout (KO) but not BCL2-tg Gna13 WT mice, GC B cells were frequently found in the bone marrow (Fig. 3i). The more frequent mutations of GNA13 than of S1PR2 in both GCB-DLBCL and Burkitt’s lymphoma, despite the similar size of their open reading frames, together with our finding of Gna13-deficient but not S1pr2-deficient mouse GC B cells in circulation (Fig. 2c, g), led us to hypothesize that additional Gα13-coupled G-protein-coupled receptors may be involved in GC B-cell regulation. In this regard, P2YR8, a gene situated in the pseudoautosomal region of the X chromosome, was a target of mutations in published whole-exome sequencing data of GCB-DLBCL and Burkitt’s lymphoma5, 7, 14 and was frequently mutated in our GCB-DLBCL and Burkitt’s lymphoma samples, with several of each lymphoma type carrying biallelic mutations (Fig. 4a, Supplementary Table 2 and Extended Data Fig. 2). P2RY8 is an orphan receptor and has orthologues in many vertebrates, but unexpectedly it lacks an orthologue in mouse (Fig. 4b). Like S1PR2, P2RY8 was abundant in human GC B cells (Fig. 4c). Five out of six tested mutations prevented surface P2RY8 expression (Extended Data Fig. 8a, b). Despite the lack of a mouse P2RY8 orthologue, we considered the possibility that if the ligand were a small molecule it may be conserved, and we therefore asked whether P2RY8 overexpression influenced GC B-cell growth. Remarkably, human P2RY8 led to a suppressive effect on GC B-cell growth in mouse Peyer’s patches and mesenteric lymph nodes, similar to the effect of S1PR2 overexpression (Fig. 4d and Extended Data Fig. 8c). This suppression required P2RY8 coupling to Gα13 as it was not seen if the cells lacked Gna13 (Fig. 4e and Extended Data Fig. 8d). In short-term transfers, P2RY8-transduced B cells localized in the centre of the follicle immediately around and often within GCs while vector transduced cells were dispersed throughout the follicle (Fig. 4f, Extended Data Fig. 8e, f and Supplementary Information). In the absence of Gα13, P2RY8 was unable to direct B cells to the follicle centre (Fig. 4g and Extended Data Fig. 8g). Importantly, a control Gα13-coupled G-protein-coupled receptor, Tbxa2r, could not suppress GC B-cell growth or confine cells to the GC niche (Extended Data Fig. 9 and Supplementary Information). These observations lead us to suggest that P2RY8 in humans acts to suppress GC B-cell growth and promote B-cell positioning in a GC location via Gα13-dependent pathways. GC B cells are normally tightly regulated in their growth and strictly confined to the GC, and they lack the ability to exit into circulation or to survive outside the GC niche. Each of these processes breaks down in the GC B-cell-derived malignancies, GCB-DLBCL and Burkitt’s lymphoma. We provide evidence that disruption of Gα13 signalling, via mutations in GNA13, ARHGEF1, S1PR2 or P2RY8, contributes to this breakdown. GNA13 is mutated in 15–33% of GCB-DLBCL and ~15% of Burkitt’s lymphoma6, 7, 9, 10, 11 (Supplementary Table 2 and Extended Data Fig. 2). This is similar to the frequency of mutations in the histone methyltransferases EZH2 and MLL2, deletions of PTEN and amplifications of miR17-92, genetic alterations that have been highlighted for their role in oncogenesis in GCB-DLBCL15, 16, 17, 18, 19, 20, 21. Our data support a model (Extended Data Fig. 10 and Supplementary Information) where deleterious mutations in Gα13 and its effector, ARHGEF1, are sufficient to deregulate AKT signalling and to cause loss of confinement, allowing egress of GC B cells into circulation; survival of the disseminating cells at distant sites such as bone marrow depends on co-operating mutations affecting additional genes, such as BCL212, 22. S1PR2 and P2RY8 mutations are also suggested to deregulate AKT signalling and growth but may lead to less dissemination due to overlapping roles in promoting confinement. Potentially inactivating mutations of RHOA, a direct target of ARHGEF123, have been reported in Burkitt’s lymphoma24. The mechanism by which RHOA inhibits AKT activation is not yet defined but might involve activation of PTEN or inhibition of RAC25, 26, 27. We suggest that small molecules that inhibit AKT may replace the missing repressive effects of RHO on growth or survival in cells that harbour defects in the S1PR2/P2RY8–Gα13–ARHGEF1–RHO pathway. Development of active RHO-mimetics may represent a novel therapeutic approach that addresses both lymphoma cell survival and disease dissemination. All clinical samples were studied with informed consent according to an institutional review board protocol approved by the National Cancer Institute. Genomic DNA for the single exon coding region of S1PR2 and complementary DNA (cDNA) for GNA13 or ARHGEF1 was amplified by PCR. PCR products were bidirectionally sequenced using an ABI 3730 Genetic Analyzer (Applied Biosystems). Sequence electropherograms were manually reviewed. ARHGEF1 encodes multiple splice variants with up to 28 coding exons per splice variant. We were unable to sequence the open reading frame of ARHGEF1 from cDNA in some cell lines in our panel, probably because of splice variation or insufficient transcript. In some cell lines, regions containing coding exons for ARHGEF1 were amplified from genomic DNA. Primers used for amplification and sequencing are shown in Supplementary Table 4. The following NCBI (RefSeq) accession numbers were used to report mutations: ARHGEF1, NM_004706 and NP_004697; GNA13, NM_006572 and NP_006563; S1PR2, NM_004230 and NP_004221. Adult C57BL6 Ly5.2 (CD45.1+) mice at least 7 weeks of age were from the National Cancer Institute. S1pr2−/− mice28 were backcrossed for at least six generations to C57BL6/J (B6/J). Arhgef1−/− mice29 were backcrossed to B6/J for at least six generations. Gna13 f/f mice were on a mixed background30. Mb1-cre mice (provided by M. Reth) express Cre in all B-lineage cells31. BCL2-tg mice were of the EµBcl2-22 line32 that overexpresses BCL2 selectively in B cells. MD4 Ig-tg mice were from an internal colony. Mice lacking Gna13 in B cells and littermate controls were generated by crossing mb1-cre + Gna13 f/+ mice to Gna13 f/f. In most experiments, bred mice of both sexes were used and were between 7 and 12 weeks of age except in the ageing cohort of Gna13 animals as indicated. Bone marrow chimaeras were made using Ly5.2 (CD45.1+) from National Cancer Institute as hosts as previously described33 and analysed at least 8 weeks after reconstitution. For one experiment using S1pr2 heterozygous and WT littermate donors, mice were also heterozygous for β-2-microglobulin. CD21-cre (Cr2-cre) mice expressing Cre in mature B cells were from Jackson Laboratory. The mouse genotype was not blinded from the investigator and mice were not randomized. Mice were housed in a specific pathogen-free environment in the Laboratory Animal Research Center at the University of California, San Francisco, and all animal procedures were approved by the Institutional Animal Care and Use Committee. S1PR2, P2RY8, GNA13, ARHGEF1 retroviral constructs were made by inserting the human open reading frame into the MSCV2.2 retroviral vector followed by an internal ribosome entry site (IRES) and Thy1.1 or green fluorescent protein (GFP) as an expression marker. The mouse Tbxa2r open reading frame was inserted into the Thy1.1 MSCV2.2 retroviral vector. S1PR2, P2RY8 and Tbxa2r were inserted in frame with a preprolactin leader and Flag-epitope encoding sequence. Lymphoma-associated mutations were introduced into S1PR2 or P2RY8 by quick-change PCR. WEHI231 or human lymphoma cell lines engineered to express an ecotropic retroviral receptor34 were spin-infected with retrovirus containing vector, WT or mutant S1PR2, P2RY8, Tbxa2r, GNA13 or ARHGEF1. For transduction of bone marrow, S1pr2 heterozygous or deficient, CD21-cre or Gna13 f/f mb1-cre donor mice were injected intravenously with 3 mg 5-fluorouracil (Sigma). Bone marrow was collected after 4 days and cultured in DMEM containing 15% (v/v) FBS, antibiotics (penicillin (50 IU ml−1) and streptomycin (50 μg ml−1); Cellgro) and 10 mM HEPES, pH 7.2 (Cellgro), supplemented with IL-3, IL-6 and stem cell factor (at concentrations of 20, 50 or 100 ng ml−1, respectively; Peprotech). Cells were ‘spin-infected’ twice at days 1 and 2 and transferred into irradiated recipients on day 3. Bone marrow chimaeras in which constitutively active myristoylated Akt (myr-Akt) was selectively expressed in B cells were generated by transducing CD21-cre bone marrow with retrovirus in which myr-Akt was downstream of a loxP–stop–loxP cassette3. To generate activated B cells that could be efficiently retrovirally transduced, MD4 Ig-transgenic mice (MGI 2384500) containing lysozyme-specific B cells were injected with 5 mg hen egg lysozyme, splenocytes were harvested 4 h later and the B cells further activated by culturing with 20 μg ml−1 anti-CD40 (FGK4.5; BioXcell) for 24 h as in past studies35. Alternatively, Gpr183+/− or Gna13 WT or KO spleen cells were harvested in media containing 1 μg ml−1 lipopolysaccharide or 0.25 μg ml−1 anti-CD180 (RP-105; clone RP14, BD Biosciences) and cultured for 24 h. Later experiments were performed using anti-CD180 activation as we found it much more effective in achieving high levels of transduction than lipopolysaccharide. The activated B cells were spin-infected for 2 h with retroviral supernatant, and cultured overnight before transfer into SRBC-immunized WT mice. Transferred cells were analysed after 24 h by flow cytometry and immunohistochemistry. B cells from spleen, mesenteric lymph nodes, Peyer’s patches and blood were isolated and stained as previously described3. Lymph was collected from the cisterna chyli via fine glass micropipette as previously described36. Assessment of clonality by PCR of J558 heavy chain, and κ and λ light chains, from genomic DNA from bulk mesenteric lymph node cells from year-old mice was performed as previously described37. For adoptive transfer experiments, mesenteric lymph nodes were harvested, washed once and transferred intravenously or intraperitoneally into CD45.1 recipient mice. Spleen and bone marrow were harvested 6 h after intravenous transfer; parathymic lymph nodes were harvested 6 h after intraperitoneal transfer. Harvested organs were analysed by FACS for the presence of donor GC B cells. For GC B-cell positioning experiments in a mixed setting, Gna13 WT or KO CD45.2+ B cells were transferred with WT CD45.1+ B cells into MD4 Ig-tg CD45.1+ recipients. Recipients were then immunized with SRBCs and analysed after 8 days. For pAkt analysis of mesenteric lymph node GC B cells, mesenteric lymph nodes were harvested in RPM-I1640 medium containing 0.5% (w/v) fatty-acid-free BSA (migration media; EMD Biosciences). Cells were RBC lysed twice and re-suspended in migration media. Cells were incubated for 10 min at 37 °C and then stimulated for 10 min with CXCL12 (300 ng ml−1) or S1P (10 nM). Cells were fixed at a final concentration of 1.5% PFA for 10 min at room temperature of 21–23 °C and then permeabilized in ice-cold methanol. Cells were washed twice in staining buffer, blocked with Fc-block (2.4G2; BioXcell) and 5% normal goat serum for 20 min at room temperature of 21–23 °C, stained for 45 min at room temperature of 21–23 °C for Akt phosphorylated at Ser 473 (D9E, number 4060; Cell Signaling Technology) followed by goat antibody to rabbit IgG conjugated to allophycocyanin (sc-3846; Santa Cruz Biotechnology) as well as antibodies to GC markers. For pAkt analysis by flow cytometry in transduced WEHI231 or human GCB-DLBCL lines, cells were stimulated for 5 minutes with or without CXCL12 (100 ng ml−1) with or without S1P (1 nM for WEHI-231 or 10 nM for human GCB-DLBCL lines) and fixed and stained as above for pAkt as well as anti-Thy1.1 conjugated to phycoerythrin (clone ox-7; Biolegend). Human cell lines used in this paper were tested for mycoplasma contamination. Mycoplasma-positive lines were treated with MycoZap (Lonza) and Plasmocin (InvivoGen). All human cell lines were tested for a unique profile of polymorphic DNA copy number variants (CNV fingerprint; unpublished protocol from L. Bergsagel). In some experiments, cells were treated with the PI3K inhibitors wortmannin (Sigma) or GS-1101 (Selleck Chemicals) as negative pAkt staining controls. For active caspase-3 staining, total mesenteric lymph node cells were harvested, washed once and incubated in RPMI-I1640 containing 10% FCS for 3 h at 37 °C; cells were stained for surface markers, fixed and permeabilized with BD Cytofix/Cytoperm and stained with anti-active caspase-3 conjugated to biotin (clone: C92-605; BD Biosciences) according to the manufacturer’s instructions. Chemotaxis assays of GC B cells were performed using total mesenteric lymph node cells that were RBC lysed twice or transduced WEHI231 or human GCB-DLBCL lines as described3. U-46619 was from Cayman Chemicals. Flow cytometry was performed on a FACSCalibur or LSRii (BD Biosciences). For quantitative PCR analysis of gene expression in GC B cells, Ptprc (encoding CD45) was used as a control since its expression was unchanged between follicular and GC B cells by microarray (http://www.immgen.org/ and unpublished data), RNA sequencing analysis (unpublished data) and by surface staining. In contrast, Gapdh and Hprt were both upregulated in GC B cells (http://www.Immgen.org and unpublished data). WEHI231 cells transduced with vector, WT or mutant human S1PR2 were washed twice in migration media and incubated at 37 °C for 30 min, washed once in cold PBS and lysed in 0.5% Brij 35, 0.5% NP40, 150mM NaCl, 10 mM Tris-HCl, pH 7.4 with protease inhibitor cocktail (Roche) for 1 h on ice. Lysates were centrifuged and supernatants were mixed with loading buffer and reducing agent and incubated at room temperature of 21–23 °C for 30 min. Samples were resolved by SDS–polyacrylamide gel electrophoresis (SDS–PAGE), and Flag expression was detected with rabbit polyclonal anti-Flag (Sigma). For pAkt western blot experiments, Ly7, Ly8 or WEHI cells that were sorted based on Thy1.1 expression and expanded were stimulated as above and lysed in 2× sample buffer, resolved by SDS–PAGE and probed with rabbit anti-pAkt S473 (D9E, number 4060; Cell Signaling Technology). Cryosections 7 μm in thickness from mesenteric lymph node and spleen were cut and prepared as described3. Tumour immunophenotyping was performed using goat polyclonal IRF4 antibody (Santa Cruz, sc-6059) or biotinylated anti-mouse CD138 (clone 281-2; BD Biosciences). For Bcl-6 staining, cryosections were fixed with 4% PFA for 10 min and stained with rabbit polyclonal Bcl6 antibody (Santa Cruz, sc-368). Images were captured with a Zeiss AxioOberver Z1 inverted microscope. Prism software (GraphPad) was used for all statistical analysis. Data were analysed with a two-sample unpaired (or paired, where indicated) Student’s t-test. P values were considered significant when less than 0.05. Download references We thank S. Coughlin for Gna13f/f and Arhgef1−/− mice and R. Proia for S1pr2−/− mice. We thank X. Geng and G. Doitsch for assistance with processing of human tonsil, A. Reboldi for discussion, and T. Arnon and O. Bannard for reading the manuscript. J.R.M. is supported by a Fellow Award from the Leukemia & Lymphoma Society and by National Institutes of Health (NIH) institutional training grants (T32 DK007636 and T32 CA1285835); R.S. is supported by the Dr Mildred Scheel Stiftung für Krebsforschung (Deutsche Krebshilfe). N.V. was supported by NIH grant GM097261 for the modelling work. J.G.C. is an Investigator of the Howard Hughes Medical Institute. The human lymphoma samples were studied under the auspices of the Lymphoma/Leukemia Molecular Profiling Project. The work was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research, and NIH grant AI45073.