Advertisement

The tumor suppressive TGF-β/SMAD1/S1PR2 signaling axis is recurrently inactivated in diffuse large B-cell lymphoma

Anna Stelling, Hind Hashwah, Katrin Bertram, Markus G. Manz, Alexandar Tzankov and Anne Müller

Data supplements

Article Figures & Data

Figures

  • Figure 1.

    Genomic editing of the S1PR2 locus provides a growth advantage to DLBCL cell lines in vitro and in vivo. The DLBCL cell lines (A) RC-K8 and (B) SU-DHL-6 were subjected to S1PR2 inactivation using CRISPR/Cas9 editing. Absolute cell counts of 2 to 3 independent clones derived from FACS single cells of the indicated genotypes were compared under standard cell culture conditions over 10 days without medium change. Pooled results from (A) 2 of a total of 4 independent experiments and (B) of 4 independent experiments are shown. P values were calculated using the Student t test on the average value for each genotype pooling S1PR2+/− and S1PR2−/− clones. (C-E) Ten million cells each of 5 S1PR2+/+ clones (red), 4 S1PR2+/− clones (light blue), and 2 S1PR2−/− clones (all in the RC-K8 cell line; blue) were injected subcutaneously into the flanks of NSG mice. (C) Tumors were excised, representative macroscopic images were taken, and (D) tumor weights and (E) volumes were determined at the study end point 40 days post injection. Every dot represents 1 tumor, and plots show pooled data from 2 independent experiments. (F-H) Ten million cells per mouse of 2 to 3 replicates each of 3 independent S1PR2+/+ clones (red) and 3 S1PR2−/− clones (all in the SU-DHL-6 cell line; blue) were injected IV into MISTRG mice. Mice were euthanized 35 days after tumor cell injection, (F) their spleens were weighed, and the frequencies of hCD45+ cells in the (G) spleen and (H) bone marrow was determined by flow cytometry. Every dot represents 1 mouse; graphs represent pooled data from 2 independent experiments. (D-H) Horizontal lines indicate medians; P values were calculated using the Mann-Whitney U test. *P < .05; **P < .01; ***P < .001.

  • Figure 2.

    The monoallelic loss of S1pr2 promotes hyperproliferation of the GC B-cell compartment and increases the lymphoma burden in a spontaneous and a serial transplantation model of MYC-driven lymphomagenesis. (A-F) S1pr2+/+ and S1pr2+/− mice on the BL/6 background were immunized twice intraperitoneally with 200 μL 10% SRBC, with a 10-day interval between the 2 immunizations. (A-C) Mice were euthanized 10 days after the last immunization and GC B cells were flow cytometrically identified as CD95hi CD38lo in the CD19+ B-cell compartment. GC B-cell frequencies in % of all CD19+ B cells as well as absolute numbers per spleen are shown alongside representative FACS plots. Nonimmunized littermates are shown as control. (D-F) The GC area (arbitrary units) of immunized mice was determined by quantifying 3 Ki-67-stained spleen sections per mouse (D) using ImageJ. (E-F) Representative pictures of spleens of immunized S1pr2+/+ and S1pr2+/− mice are shown. Size bar represents 1000 μm; arrows point to GCs. (A-B,D) Every dot represents 1 mouse and data from 5 pooled experiments are shown. (G-H) One million lymph node cells per mouse, harvested from 3 S1pr2+/+ and 3 S1pr2+/− MYCtg donor mice of the cohorts shown in supplemental Figure 2F were injected IV into wild-type BL/6 recipients. Mice were palpated every other day for enlarged lymph nodes and euthanized after 20 days (ie, when the first mice showed disease symptoms). (G) Spleen and (H) lymph node weights were determined. Lymph node weights represent the average of 2 inguinal and 2 axillary lymph nodes. Control mice were not injected with tumor cells. (A-B,D,G-H) Horizontal lines indicate medians; P values were calculated using the Mann-Whitney U test. *P < .05; **P < .01.

  • Figure 3.

    S1PR2 expression is regulated by the TGF-β/SMAD signaling pathway. (A) S1PR2 expression after 24 hours of treatment with the indicated increasing doses of TGF-β, as assessed in the SU-DHL-6, Oci-Ly10, RC-K8, and Oci-Ly3 DLBCL cell lines by qRT-PCR. (B) The DLBCL cell line SU-DHL-6 was treated with FOXP1 targeting siRNA for 48 hours and subjected to treatment with 2 ng/mL TGF-β for an additional 24 hours. (A-B) Data are pooled from 3 or more independent experiments. Graphs show mean ± SEM; P values were calculated using the Student t test. (C) The indicated DLBCL cell lines were treated with 2 ng/mL TGF-β for 1 hour and subjected to immunoblotting with antibodies against the indicated SMAD proteins, p-SMAD1/5/9 and tubulin. Representative immunoblots of at least 2 independent experiments are shown. (D) TGF-βR2 surface expression of the indicated DLBCL cell lines, as assessed by flow cytometry. The plots are representative for 2 independent experiments. (E) pSMAD1/5/9 ChIP of cells treated or not with 5 ng/mL TGF-β for 4 hours; an unspecific rbIgG antibody was used as control. Eluted DNA was subjected to PCR using primers amplifying 2 regions 2.5 and 5 kb upstream of the S1PR2 TSS. MyoD was amplified as negative control; CDKN1A and ID1 were used as positive controls for canonical TGF-β and BMP signaling. Graphs represent the fold change of the yield relative to 1% input of the pSMAD1/5/9 sample vs rbIgG; means ± SD of 2 independent experiments are shown. *P < .05; **P < .01; ***P < .001; ****P < .0001.

  • Figure 4.

    TGF-β induces S1PR2-dependent apoptosis in DLBCL cell lines in vitro and in vivo. (A) Cell viability and apoptosis, as determined by Cell Titer Blue assay and Annexin V staining, of the indicated cell lines after 24 hours of exposure to increasing concentrations of TGF-β; values are normalized to the untreated control sample. (B) The DLBCL cell line SU-DHL-6 was treated with FOXP1 targeting siRNA for 48 hours, subjected to 2 ng/mL TGF-β for additional 24 hours, and analyzed as shown in panel A. (A-B) Data are pooled from 3 or more independent experiments. Graphs show means ± SEM; P values were calculated using the Student t test. (C) Three S1PR2+/+, 1 S1PR2+/−, and 3 S1PR2−/− clones generated in the SU-DHL-6 cell line were treated with 2 ng/mL TGF-β and analyzed for apoptosis by Annexin V staining. Bars represent pooled data for each genotype relative to the untreated control of each clone. Each clone was analyzed 3 to 6 times. Graphs represent means ± SEM; P values were calculated using the Student t test. (D-F) Ten million SU-DHL-6 cells were injected subcutaneously into both flanks of NSG mice. (D) One tumor per mouse was injected intratumorally with TGF-β at the depicted intervals; the other received vehicle only. Tumor volumes were measured (E) after excision and (F) RNA was extracted and qRT-PCR for S1PR2 was performed on excised tumor tissue. (F) Each dot represents 1 tumor and results are pooled from 2 independent experiments. S1PR2 expression analysis was performed in only 1 of the 2 studies with n = 10 per group. Two control and 1 TGF-β–treated tumor had to be excluded because of insufficient RNA quality. TGF-β–treated and control tumors are compared for each mouse. P values were calculated using the Mann-Whitney U test. *P < .05; **P < .01; ***P < .001; ****P < .0001.

  • Figure 5.

    Loss of TGF-β signaling in the GC compartment induces GC B-cell hyperproliferation. (A-E) Tgfbr2fl/fl mice were crossed with AID-Cre mice; Tgfbr2wt/wt, Tgfbr2fl/wt, and Tgfbr2fl/fl x AID-Cre mice were immunized IV with 200 μL 10% SRBCs, euthanized 10 days after immunization, and GC cells were analyzed by flow cytometry as described in Figure 2. (A-C) GC B-cell frequencies in % of all CD19+ B cells as well as absolute numbers per spleen are shown alongside representative FACS plots. Nonimmunized littermates are shown as control. (D-E) The GC area (arbitrary units) of immunized mice was determined by quantifying 3 Ki-67-stained spleen sections per mouse (D) using ImageJ. (E) Representative pictures of spleens of immunized mice of the indicated genotypes. Size bar represents 1000 μm; arrows point to GCs. (A-B,D) Every dot represents 1 mouse, and data from 3 pooled experiments are shown. Graphs show medians. *P < .05; **P < .01.

  • Figure 6.

    TGF-β signaling via TGF-βR2 and SMAD1 activates S1PR2 expression and induces apoptosis of DLBCL cells and SMAD1 expression is downregulated in DLBCL patients. (A) Three SMAD1+/+ and 2 SMAD1−/− as well as 3 TGFβR2+/+ and 3 TGFβR2−/− clones (all generated in the SU-DHL-6 cell line) were treated with 2 ng/mL TGF-β for 24 hours and analyzed for apoptosis with Annexin V staining. Bars represent pooled data for each genotype relative to the untreated control of each clone. (B) The same clones as in panel A were subjected to RNA extraction and S1PR2-specific qRT-PCR. (A-B) Each clone was analyzed twice; graphs represent means ± SEM; P values were calculated using the Student t test. (C) The DLBCL cell line SU-DHL-6 was treated with SMAD1-targeting siRNA for 48 hours and subjected to 2 ng/mL TGF-β for additional 24 hours. Cells were analyzed for apoptosis by Annexin V staining. Graphs show pooled results from 6 independent experiments. Means ± SEM are represented. P values were calculated using the Student t test. (D) Absolute cell counts of 2 to 3 independent clones derived from FACS single cells of the indicated genotypes were compared under standard cell culture conditions over 10 days without medium change. Two experimental replicates are shown. P values were calculated using the Student t test on the average value for each genotype. (E) Ten million cells each of 3 SMAD1+/+ clones (gray) and 2 SMAD1−/− clones (orange, in SU-DHL-6) were injected subcutaneously into the flanks of NSG mice. Tumors were excised and tumor weights and tumor volumes were determined at the study end point 24 days after injection. Every dot represents 1 tumor; plots show pooled data from 2 independent experiments. (F) Ten million cells of 6 SMAD1/TGFβR2+/+ clones (gray) and 2 SMAD1−/− (orange) or 3 TGFβR2−/− (blue, all in SU-DHL-6) clones were injected IV into MISTRG mice. Mice were euthanized 35 days postinjection, their spleens were weighed, and the frequencies of hCD45+ cells in the spleens and bone marrow was determined by flow cytometry. Every dot represents 1 mouse; graphs represent data from 3 experiments. (E-F) Horizontal lines indicate medians; P values were calculated using the Mann-Whitney U test. (G) Negative (left) and positive (right) SMAD1 immunohistochemical staining of DLBCL patient samples. Size bar represents 20 μm. *P < .05; **P < .01; ***P < .001; ****P < .0001. n.s., not significant.

  • Figure 7.

    Schematic summarizing the tumor-suppressive properties of the TGF-β/SMAD1/S1PR2 axis in DLBCL. Under physiological conditions, centrocytes and centroblasts express large amounts of S1PR2, which promotes GC confinement because of a gradient of S1P that increases in concentration toward the borders of the GC and leads to apoptosis in GC cells that attempt to exit the GC. In DLBCL, S1PR2 is either mutated (in the GCB subtype) or transcriptionally downregulated by FOXP1 (in the ABC subtype). Loss of S1PR2 thus is an early initiating event in both major subtypes of DLBCL. The expression of S1PR2 is further regulated by TGF-β, which binds to its receptor TGF-βR2 and activates SMAD1 phosphorylation and nuclear translocation. p-SMAD1 binds directly to regulatory elements in the S1PR2 promoter and activates S1PR2 expression; most cases of DLBCL exhibit aberrantly low or absent expression of SMAD1.