S1P2 sphingosine 1-phosphate receptor signaling can regulate proliferation, survival, morphology, and migration in many cell types in vitro. Here, we report that S1P2−/− mice develop clonal B-cell lymphomas with age, such that approximately half of the animals display this neoplasm by 1.5 to 2 years of age. Histologic, immunophenotypic, and molecular analyses revealed a uniform tumor phenotype with features of germinal center (GC)–derived diffuse large B-cell lymphoma (DLBCL). Tumor formation was preceded by increases in GC B cells and CD69+ T cells, as well as an increased formation of spontaneous GCs, suggesting that S1P2 loss may promote lymphomagenesis in part by disrupting GC B-cells homeostasis. With the sole exception of rare lung tumors, the effect of S1P2 gene disruption is remarkably restricted to DLBCL. In humans, 28 of 106 (26%) DLBCL samples were found to harbor multiple somatic mutations in the 5′ sequences of the S1P2 gene. Mutations displayed features resembling those generated by the IgV-associated somatic hypermutation mechanism, but were not detected at significant levels in normal GC B cells, indicating a tumor-associated aberrant function. Collectively, our data suggest that S1P2 signaling may play a critical role in suppressing DLBCL formation in vivo. The high incidence of DLBCL in S1P2−/− mice, its onset at old age, and the relative lack of other neoplasms identify these mice as a novel, and potentially valuable, model for this highly prevalent and aggressive human malignancy. [Cancer Res 2009;69(22):8686–92]

Sphingosine 1-phosphate (S1P) can potently regulate the proliferation, survival, migration, and morphology of many cell types in vitro (1, 2). In vivo data implicate S1P in the physiology of the cardiovascular, immune, reproductive, and nervous systems (13). S1P elicits most of its effects via one or more members of a family of G protein–coupled receptors currently called S1P1-S1P5 (15). The S1P2 receptor (previously named AGR16, H218, Edg-5, and LPB2) binds S1P with nanomolar affinity and is ubiquitously expressed throughout development and adulthood (14).

S1P2 couples to Gi, Gq, and G13, leading to signaling through multiple pathways involving extracellular signal-regulated kinase/mitogen-activated protein kinase, cyclic AMP, c-Jun, c-Fos, Ca2+, Rho, and Rac (15). S1P2-dependent signaling has also been shown to mediate S1P-induced proliferation, survival, morphology, and migration in a wide variety of cell types in culture (1, 5).

S1P2 knockout studies indicate essential in vivo roles in neuronal excitability during development (6), auditory and vestibular maintenance (79), and vascular tone (10). The signaling pathways and the cellular effects of S1P2 both raise the possibility of S1P2 involvement in cancer. In vitro manipulation studies have specifically addressed this issue, revealing that S1P2 activation can inhibit cancer cell migration, invasion, and metastasis (1113). However, the widespread in vivo expression of S1P2, and its many cell type–dependent actions revealed in culture, make it difficult to predict what in vivo roles it may play in cancer.

We report here that targeted disruption of the S1P2 gene in mice leads to a high incidence of lymphoma formation with aging. These tumors display a remarkably uniform diffuse large B-cell lymphoma (DLBCL) phenotype of germinal center (GC) origin preceded by an increase in GC B cells and spontaneous GC formation. Moreover, the S1P2 5′ sequences were found to be mutated in 26% of human DLBCL, but not in normal GC B cells, suggesting that S1P2 is aberrantly targeted by somatic hypermutation (SHM) in this lymphoma (14). Overall, these data suggest that S1P2 may play a nonredundant in vivo role in suppressing DLBCL formation.

S1P2 knockout mice

The mouse S1P2 gene was targeted through homologous recombination as previously described, leading to the deletion of the complete protein-coding region of this gene (6). S1P2−/− mice and S1P2+/+ controls were generated on a 50:50 129SvEvBrd/C57Bl/6 background via S1P2+/− intercrosses and were genotyped by Southern blot or PCR of tail DNA. An additional cohort was generated on a pure 129SvEvBrd background. S1P2−/− mice were born at the predicted Mendelian frequency (6). They were maintained under standard housing conditions and monitored for abnormalities up to 18 to 24 mo of age.

In our previous study, some S1P2−/− mice in the mixed genetic background were observed to have rare (mean, <4), brief (mean, <5 s) seizures during development and an ∼15% overall death rate from the seizures (6). On a pure 129SvEvBrd background, the overall S1P2−/− death rate drops below 5%.5

5A.J. MacLennan et al., unpublished data.

All experimental data were collected and analyzed without prior knowledge of genotype. All mice were cared for in accordance with the University of Cincinnati and Columbia University Institutional Animal Care and Use Committees.

SRBC immunization

For analysis of T cell–dependent immune responses, age- and sex-matched mice were immunized i.p. with 0.5 mL of a 2% SRBC suspension in PBS (Cocalico Biologicals), and sacrificed 10 d later.

Flow cytometry

The following antibody combinations were used on single-cell suspensions from normal organs or tumors: A: B220, IgM, IgD, and Ig κ light chain; B: B220, CD23, CD21, and CD11b; C: IgM, CD43, CD5, and CD19; D: B220, CD138, IgM, IgD, and Ig κ light chain; E: B220, CD69, and CD80; F: B220, A44.1, IgM, and CD43; G: CD3, CD4, CD8, and Vβ8 TCR. The enumeration of GC cells was performed by four-color staining with B220, Peanut Agglutinin (PNA), CD38, and GL7. All antibodies were from BD Pharmingen, except for the IgD-Pe (Southern Biotechnology, Inc.). Data were acquired on a FACSCalibur (Becton Dickinson) and analyzed with the FlowJo software (Tree Star, Inc.).

For cell proliferation measurements, 2 mg BrdUrd (Roche Biochemical) was injected 2 h before sacrifice. Four-color BrdUrd staining (Pharmingen) was performed on surface-labeled cells according to the manufacturer's instructions. Membrane exposure of phosphatidylserine in apoptotic cells was measured with recombinant, phycoerythryn-labeled AnnexinV (Pharmingen), and 7-AminoActinomycin D (Sigma) staining identified dead cells.

Histology, immunohistochemistry, and immunofluorescence

Four-micrometer-thick formalin-fixed, paraffin-embedded sections were stained for H&E or immunostained as published previously (15) using the following primary antibodies: rabbit anti-BCL6 (N3), goat anti-IRF4 (sc-6059), goat anti-Pax5, goat anti-CD21 (sc-7028; Santa Cruz Biotechnology), donkey anti-IgM (Jackson Immunoresearch Laboratories), rat anti-B220, CD138 (BD Pharmingen), and PNA lectin (Vector). Double immunohistochemical stainings were performed as published previously (16) with noncrossreacting combinations of primary and secondary antibodies.

Morphometric quantitation of GCs was performed with ImageJ (RSB, NIH). Grayscale or color images were acquired at room temperature using a Nikon E600 microscope (Nikon USA), fitted with Planachromat 4×/0.10/30.0, 10×/0.25/10.5, 20×/0.40/1.3, PlanApo 40×/0.095/0.12 to 0.16 (light microscopy) objectives, with a SPOT-2 CCD camera and software (Diagnostic Instruments). All images were edited for optimal color contrast with Adobe Photoshop 7 whole-image autocontrast feature, cropped, then mounted and labeled with Adobe Illustrator 10 (Adobe Systems), on an Apple G4 computer (Apple Computers).

Southern blot analysis and SHM analysis of Ig genes

Genomic DNA was isolated from tumor specimens by the salting out method, and screened for the presence of clonal immunglobulin heavy chain (IgH) gene rearrangements by Southern blot analysis according to standard procedures. Briefly, 5 μg of DNA were digested with EcoRI, separated by electrophoresis through a 0.8% agarose gel, transferred to nylon membranes, and hybridized with a radiolabeled fragment specific for the murine JH4 region.

For mutational analysis, the rearranged Ig variable heavy chain genes were amplified from genomic DNA as described (17); in a subset of samples, cDNA synthesis and reverse transcription-PCR (RT-PCR) amplification were also performed using primers specific for all the major murine IGVH families, and the protocols are available upon request. Amplicons indicative of clonal rearrangements (i.e., appearing as a discrete band on gel electrophoresis) were gel purified using the QIAQUICK purification method (QIAGEN) and sequenced directly to confirm the clonal nature of the rearrangement and to identify the presence of mutations. Since direct sequencing can only detect changes present in a significant fraction of the template DNA (i.e., clonally represented mutations), Taq errors are negligible and only mutations present in the clonal, tumor population are identified. Sequences were compared with the National Center for Biotechnology Information mouse genome databases and Ig sequence database, as well as to our own database to rule out polymorphisms or contamination with previously amplified rearrangements. All mutations were confirmed on independent PCR products.

RNA extraction and Northern blot analysis of S1P2 expression

Total RNA was extracted from 20 DLBCL cell lines using TRIZOL (Invitrogen), and Northern blot analysis was performed according to standard procedures, with radiolabeled probes corresponding to a fragment of the human S1P2 cDNA, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control for loading. The primers used for generation of the S1P2 probe are S1P2RT-F1 (5′-tgccaaggtcaagctgtatg-3′) and S1P2RT-R1 (5′-agtccagaaggaggatgctg-3′). S1P2 expression levels were then quantitated by Phosphorimager analysis and normalized with the GAPDH levels.

S1P2−/− mice develop lymphomas with aging

To examine the function of S1P2 in tumorigenesis, a cohort of age-matched S1P2−/− and S1P2+/+ mice was monitored for the occurrence of pathologic events over a period of 2 years. Complete necropsies, performed on eight S1P2+/+ and nine S1P2−/− mice found dead between 14 and 23 months of age, revealed the presence of extensive lymphoma involvement in four of the S1P2−/− mice, whereas only one of the S1P2+/+ mice had a pathologic mass on its spleen that could not be diagnosed due to severe tissue autolysis. Following these observations, we sacrificed and analyzed 13 randomly selected 19- to 24-month-old S1P2−/− mice and 11 age-matched S1P2+/+ controls deriving from the same breeding pairs (in many cases, littermates). Complete necropsies of these mice revealed lymphomas in 7 of 13 (54%) of the S1P2−/− mice (total, 11 of 22) but none of the S1P2+/+ controls (total, 1 possible of 19). Tumors were found in mesenteric, submandibular, and mediastinal lymph nodes; in the spleen; and rarely in liver and retroperitoneal soft tissue, with frequent involvement of multiple sites within individual animals. Tumors with the same characteristics were subsequently seen on a pure 129SvEvBrd background, with 6 of 9 (66%) of the 18- to 24-month-old S1P2−/− mice examined displaying lymphomas, in contrast to 0 of 9 of the age-matched S1P2+/+ controls. Therefore, in total, 17 of 31 (54.8%) S1P2−/− mice were found affected compared with 1 possible of 28 (3.6%) S1P2+/+ controls. In addition to the lymphoid malignancies, 3 of 31 (10%) S1P2−/− mice showed evidence of lung tumors, including two adenocarcinomas and one very large lung adenoma that could be classified as a carcinoma based on its size. No such abnormalities were present in the 28 S1P2+/+ controls. With the exception of the comparatively low incidence of lung lesions, no other neoplasms were associated with S1P2 gene disruption despite the high incidence of lymphomas. The few S1P2+/− mice that underwent full histopathologic analysis (data not shown) displayed neither lymphomas nor lung lesions, consistent with previous reports documenting the lack of other S1P2−/−–associated phenotypes in heterozygous animals (e.g., refs. 6, 7).

S1P2−/− lymphomas uniformly display the features of GC-derived DLBCLs

Histologic characterization of the lymphoma tissue from the S1P2−/− mice revealed infiltration by large atypical blast cells admixed with variable amounts of reactive inflammatory and lymphoid cells, consistent with the diagnosis of DLBCL, according to the Bethesda classification for murine lymphoid neoplasms (Fig. 1, left; ref. 18). Benign splenic marginal zone hyperplasia was diagnosed in three mice, two of which concomitantly displayed a nodal DLBCL. Immunohistochemical and flow cytometric analysis of the lymphomas indicated a GC origin (B220+, PNA+, BCL6+, Pax5+, IgMk dim/negative, IgD−, mostly CD5−, CD23−) without plasmacytic (CD138−, IRF4−) or monocytoid/marginal zone differentiation (CD21−, CD11b−, CD80−; data not shown; Figs. 1 and 2A). One of five tumors also displayed evidence of isotype switching (i.e., surface IgK positive, negative for preswitch isotypes IgD and IgM). RT-PCR analysis consistently documented expression of AID, the molecule required for SHM and class switch recombination in GC B cells (19), in all samples studied (Fig. 2B). The lymphomas did not express activation markers (CD69−, IRF4−).

Figure 1.

Histopathology of DLBCL in S1P2−/− mice. Representative DLBCL, featuring H&E staining (note the large tumor cells in the Wright-Giemsa–stained touch preparation in the inset). These tumor cells display positive BCL6 (dark blue), B220 (purple gray), and PNA (reddish brown) staining, and negative IRF4 (dark blue) and CD138 (dark blue) staining. The lymph node is totally replaced by diffuse large cell lymphoma proliferation, containing lymphocytes and inflammatory cells (BCL6, B220, and PNA negative, and IRF4 and CD138 positive). The PNA-stained section also received a light blue nuclear counterstain. Scale bars, 10 μm (bottom) and 1 μm (top).

Figure 1.

Histopathology of DLBCL in S1P2−/− mice. Representative DLBCL, featuring H&E staining (note the large tumor cells in the Wright-Giemsa–stained touch preparation in the inset). These tumor cells display positive BCL6 (dark blue), B220 (purple gray), and PNA (reddish brown) staining, and negative IRF4 (dark blue) and CD138 (dark blue) staining. The lymph node is totally replaced by diffuse large cell lymphoma proliferation, containing lymphocytes and inflammatory cells (BCL6, B220, and PNA negative, and IRF4 and CD138 positive). The PNA-stained section also received a light blue nuclear counterstain. Scale bars, 10 μm (bottom) and 1 μm (top).

Close modal
Figure 2.

DLBCL immunophenotype in S1P2−/− mice. A, surface immunophenotype by FCM of a representative DLBCL from a S1P2−/− mouse and a histologically normal spleen from a littermate control S1P2+/+ mouse. Note the low/absent surface Ig, the loss of CD23, and the absence of CD138 in the DLBCL sample. B, AID expression in representative DLBCL from S1P2−/− mice, compared with purified normal GC B cells from two S1P2−/− and one S1P2+/+ mice. Mouse lung tissue was used as a negative control for the AID RT-PCR.

Figure 2.

DLBCL immunophenotype in S1P2−/− mice. A, surface immunophenotype by FCM of a representative DLBCL from a S1P2−/− mouse and a histologically normal spleen from a littermate control S1P2+/+ mouse. Note the low/absent surface Ig, the loss of CD23, and the absence of CD138 in the DLBCL sample. B, AID expression in representative DLBCL from S1P2−/− mice, compared with purified normal GC B cells from two S1P2−/− and one S1P2+/+ mice. Mouse lung tissue was used as a negative control for the AID RT-PCR.

Close modal

Southern blot analysis of the IgH locus by using a JH probe confirmed the clonal origin of the tumors in all samples studied (n = 4/4), as documented by the presence of distinct bands representing clonally rearranged Ig genes (Fig. 3A). In cases 297 and 246, the detection of two and three bands, respectively, may reflect the derivation from a clone where both Ig alleles had been rearranged, as it occurs in ∼30% of B cells when the first rearrangement is not productive (20), or it may indicate a biclonal tumor, or both. The relative intensity of the residual germline band observed in all tumors is consistent with the presence of nonneoplastic cells in the biopsy, especially evident in the splenic tumors (case #240). Direct sequencing analysis of the rearranged IgV genes confirmed the clonal origin of the lymphomas and showed evidence of SHM in all samples analyzed (Fig. 3B and C). Overall, these data indicate that the DLBCL in S1P2−/− mice derive from the clonal expansion of a cell that transited the GC.

Figure 3.

Clonal rearrangements of the IgH locus and evidence of SHM in DLBCL from S1P2−/− mice. A, Southern blot analysis of EcoRI-digested DNA from representative S1P2−/− DLBCLs, and control tail DNA, hybridized with a murine JH region probe. The germline immunoglobulin heavy chain EcoRI fragment is ∼6.5 Kb (arrow). *, clonal rearrangements of the IgH gene in the tumors analyzed. The predominant germline band observed in some tumors is most likely due to the presence of infiltrating normal cells (see text). B, mutation analysis of clonally rearranged IgV genes in lymphomas from representative S1P2−/− animals. Note that although Southern blot analysis in tumors #246 and #297 suggests the presence of two distinct clones, only one rearranged allele could be successfully amplified, possibly due to the usage of a less common V gene family member not recognized by the primers used, or to the presence of mutations in the primer binding site. C, nucleotide sequence alignment of the rearranged IgV gene from S1P2−/− DLBCL #245 (as obtained by direct sequencing) with the germline VH7183 and JH region. Dashes, sequence identities; the indicated nucleotides correspond to mutations. Note that since direct sequencing only detects clonally represented events, the observed mutations can be unequivocally attributed to the malignant clone.

Figure 3.

Clonal rearrangements of the IgH locus and evidence of SHM in DLBCL from S1P2−/− mice. A, Southern blot analysis of EcoRI-digested DNA from representative S1P2−/− DLBCLs, and control tail DNA, hybridized with a murine JH region probe. The germline immunoglobulin heavy chain EcoRI fragment is ∼6.5 Kb (arrow). *, clonal rearrangements of the IgH gene in the tumors analyzed. The predominant germline band observed in some tumors is most likely due to the presence of infiltrating normal cells (see text). B, mutation analysis of clonally rearranged IgV genes in lymphomas from representative S1P2−/− animals. Note that although Southern blot analysis in tumors #246 and #297 suggests the presence of two distinct clones, only one rearranged allele could be successfully amplified, possibly due to the usage of a less common V gene family member not recognized by the primers used, or to the presence of mutations in the primer binding site. C, nucleotide sequence alignment of the rearranged IgV gene from S1P2−/− DLBCL #245 (as obtained by direct sequencing) with the germline VH7183 and JH region. Dashes, sequence identities; the indicated nucleotides correspond to mutations. Note that since direct sequencing only detects clonally represented events, the observed mutations can be unequivocally attributed to the malignant clone.

Close modal

S1P2 gene disruption leads to alterations in the immune system before tumor formation

To test whether lymphoma development in S1P2−/− mice is preceded by polyclonal abnormalities in B-cell differentiation and maturation, we performed an extensive characterization of the B-cell and T-cell compartments in young (7- to 8-week-old) and 9- to 12-month-old animals, by using flow cytometric and immunohistochemical analysis. Mature, immature, apoptotic, and necrotic T- and B-cell subsets were examined in bone marrow, spleen, thymus, peritoneal cavity, and peripheral blood. These experiments showed no significant abnormalities at the relatively early age of 7 to 8 weeks, except for a small increase in CD4+ T cells and a marginal, statistically insignificant increase in CD23−/CD21− B cells (Supplementary Fig. S1). In contrast, flow cytometry analysis of nonimmunized, 9- to 12-month-old animals revealed an increase in the number of B220+PNA+ GC B cells in S1P2−/− mice (Fig. 4A,, inset), compared with their littermate S1P2+/− and S1P2+/+ controls. Immunohistochemical and morphometric analysis of GC formation in spleen tissue from the same animals confirmed that the S1P2−/− spleens contain an increase in both number and size of spontaneous GCs (Supplementary Fig. S2; Fig. 4A and B), in the absence of overt tumor formation. Given the BCL6+, PNA+ GC phenotype of the DLBCL that develop later in life, these data support the possibility that S1P2 gene disruption leads to DLBCL by altering GC B-cell regulation. These differences were not detected when the animals were analyzed after immunization with SRBC as a potent stimulus to evaluate T cell–dependent GC responses (Fig. 4C), possibly because the maximal response evoked by strong antigen delivery masked the effects of chronic low level GC stimulation.

Figure 4.

Morphometric analysis of splenic GC formation in S1P2−/− mice. A, unimmunized mice; total area of BCL6+ aggregates in each size range [measured in pixels (1,000 pixels = 1 Kpx)] normalized for spleen section area. Inset, percentage of splenic B220+ PNA+ GC cells in old, unimmunized control and knockout mice measured by flow cytometry. The difference is significant (P < 0.0015). B, unimmunized mice; number of BCL6+ aggregates per spleen section area, calculated as described in A. The numbers presented are per average spleen section. *, ANOVA analyses indicated that both the size and number of >2 Kpx aggregates are significantly (P < 0.05) increased in S1P2−/− spleens. C, number and area of the GC aggregates in SRBC-immunized mice. Note the expected increase in larger aggregates for immunized control and knockout mice compared with values for unimmunized mice in A and B. n = 7 young control, 3 young S1P2−/−, 4 old control, 3 old S1P2−/−. Columns, mean; bars, SEM.

Figure 4.

Morphometric analysis of splenic GC formation in S1P2−/− mice. A, unimmunized mice; total area of BCL6+ aggregates in each size range [measured in pixels (1,000 pixels = 1 Kpx)] normalized for spleen section area. Inset, percentage of splenic B220+ PNA+ GC cells in old, unimmunized control and knockout mice measured by flow cytometry. The difference is significant (P < 0.0015). B, unimmunized mice; number of BCL6+ aggregates per spleen section area, calculated as described in A. The numbers presented are per average spleen section. *, ANOVA analyses indicated that both the size and number of >2 Kpx aggregates are significantly (P < 0.05) increased in S1P2−/− spleens. C, number and area of the GC aggregates in SRBC-immunized mice. Note the expected increase in larger aggregates for immunized control and knockout mice compared with values for unimmunized mice in A and B. n = 7 young control, 3 young S1P2−/−, 4 old control, 3 old S1P2−/−. Columns, mean; bars, SEM.

Close modal

S1P2−/− mice were indistinguishable from the wild-type controls with respect to apoptosis (measured by AnnexinV binding and 7-AminoActinomycin D uptake) and proliferation (measured by BrdUrd incorporation) of GC cells (data not shown), as well as spleen size (absolute weight or relative to body weight), spleen follicle immunoarchitecture, in situ distribution of marginal zone B cells, defined by CD21 staining (Supplementary Fig. S3), and other B-cell subpopulations including CD69+ B cells (data not shown; Supplementary Fig. S4A). Therefore, the immunologic defects resulting from S1P2 gene disruption are clearly distinct from the marginal zone B-cell abnormalities recently reported for S1P3−/− mice (21).

In the T-cell compartment, analysis of 9- to 12-month-old mice revealed a significant increase in both the percentage (S1P2−/−, 14.9%; S1P2+/+, 8.0%; P < 0.025) and intensity of CD69-labeled cells (Supplementary Fig. S4C). CD69 is an activation-associated molecule that has a pleotropic immunoregulatory role (22) including fine tuning of the inflammatory and tumor-suppressor response through transforming growth factor β. It remains to be established whether the increase in CD69 expression results from retention of CD69+ cells or increased activation of T cells. No significant differences were observed in other parameters relative to T cell subsets from the spleen, thymus, lymph nodes and peripheral blood, including the percentage of Vβ8 TCR+ cells (Supplementary Fig. S4B).

The S1P2 5′ sequences are targeted by multiple somatic mutations in human DLBCL

To determine whether the S1P2 gene is altered in human DLBCL, we used PCR amplification and direct sequencing to analyze 106 samples (20 cell lines and 86 primary biopsies) for mutations in S1P2 5′ sequences, and 20 samples (10 cell lines and 10 primary biopsies) for mutations in the S1P2 single protein-coding exon. Missense mutations of this exon were found in 2 of 20 cases (10%), including one cell line and one primary biopsy (Supplementary Table S1). In addition, 28 of 106 (26%) DLBCL samples carried one or more mutations distributed within the first ∼1.3 Kb from the transcription start site, i.e., the hypermutable region in IgV and BCL6 genes (Table 1; Fig. 5; refs. 2325). In total, 114 sequence variants were detected in the 28 mutated cases, with an average mutation frequency of 0.17 × 10−2/bp (range, 1–24 events/case; Supplementary Table S2; Table 1). Mutations were somatic in origin, as confirmed by analysis of paired normal DNA in cases with available material, and consisted in the vast majority of cases of single bp substitutions (n = 105), with occasional deletions and duplications of short DNA stretches (n = 6 and 3, respectively; Supplementary Table S2; Table 1). Of the 105 single bp substitutions observed, 61 were transitions and 44 were transversions, with a transition/transversion ratio of 1.38 (expected, 0.5; P < 0.0001; χ2 test) and a preference for G+C versus A+T substitutions (Table 1). Furthermore, although the relatively low number of events prevented the assessment of statistical significance, the frequency of mutations targeting G residues within an RGYW motif was considerably higher than expected (14% versus 3.4%). Taken together, these features strongly resemble those described for the IgV-associated SHM mechanism (26), suggesting that the same machinery targets the S1P2 gene in a sizable fraction of DLBCLs. Notably, sequencing analysis of this same region in 150 clones obtained from purified tonsillar centroblasts did not show significant levels of mutations, when compared with control pre-GC naïve B cells (n = 75) and fibroblasts (n = 82), both of which are known to lack SHM activity (mutation frequencies in the three populations: 0.013%, 0.006%, 0.011%; P = not significant; Supplementary Table S3). Thus, SHM of S1P2 does not seem to represent a measurable phenomenon during the physiologic GC reaction, suggesting that the high mutation load suffered by the S1P2 5′ sequences in DLBCL reflects a tumor-associated aberration (14).

Table 1.

Mutational analysis of the S1P2 5′ sequences in DLBCL

TotalCell linesPrimary cases
n of cases mutated/tested (%)  28/106 (26%) 5/20 (25%) 23/86 (27%) 
Mutation frequency* 0.17% 0.18% 0.17% 
n of single bp substitutions 105 21 84 
n of deletions+duplications 
Transitions/transversions (ratio) 61/44 (1.38) 16/5 (3.2) 45/39 (1.15) 
G+C/A+T mutations 81/24 17/4 64/20 
TotalCell linesPrimary cases
n of cases mutated/tested (%)  28/106 (26%) 5/20 (25%) 23/86 (27%) 
Mutation frequency* 0.17% 0.18% 0.17% 
n of single bp substitutions 105 21 84 
n of deletions+duplications 
Transitions/transversions (ratio) 61/44 (1.38) 16/5 (3.2) 45/39 (1.15) 
G+C/A+T mutations 81/24 17/4 64/20 

*Calculated in the mutated cases only, considering two alleles/sample (1,200 bp × 2).

Figure 5.

Mutational analysis of the S1P2 gene in human DLBCL. Top, S1P2 genomic locus with untranslated (empty boxes) and translated (filled boxes) exons; arrow, the transcriptional start site per GeneBank accession no. NM_004230. The region amplified for analysis is expanded in the bottom and aligned to sequences of mutated DLBCL cases (one line = two alleles), where each small segment represents a 25-bp interval and position +1 corresponds to the first nucleotide of the reference mRNA. Ovals, single basepair substitutions; brackets, deletions; triangles, insertions.

Figure 5.

Mutational analysis of the S1P2 gene in human DLBCL. Top, S1P2 genomic locus with untranslated (empty boxes) and translated (filled boxes) exons; arrow, the transcriptional start site per GeneBank accession no. NM_004230. The region amplified for analysis is expanded in the bottom and aligned to sequences of mutated DLBCL cases (one line = two alleles), where each small segment represents a 25-bp interval and position +1 corresponds to the first nucleotide of the reference mRNA. Ovals, single basepair substitutions; brackets, deletions; triangles, insertions.

Close modal

Northern blot analysis in the 20 DLBCL cell lines revealed heterogeneous S1P2 expression, ranging from undetectable levels to levels comparable with those of GC B cells or type I Burkitt lymphoma lines, representing transformed GC B cells (Supplementary Fig. S5). No correlation was observed between steady-state expression levels and the presence or absence of aberrant SHM, nor with the mutation load.

The present data show that S1P2 gene disruption in mice leads to tumor formation, indicating a role for this gene as a nonredundant tumor suppressor in vivo. Although a low percentage of the tumors observed were lung adenocarcinomas—warranting further studies with substantially larger groups of animals—the overwhelming majority of the tumors were mature B-cell lymphomas displaying histologic and molecular features of GC-derived DLBCL, suggesting that S1P2 plays a particularly critical role in suppressing such lymphoma formation.

The mechanism by which loss of S1P2 promotes DLBCL development remains to be elucidated. One of the main effects of S1P2 deletion in this model is the expansion of the GC compartment, as documented by an increase in the number and size of spontaneous GCs before any anatomic indication of tumor formation. Moreover, the histologic, immunophenotypic, and molecular features of the lymphomas developing later in life indicate their derivation from a cell that has experienced the GC reaction. Taken together, these observations suggest that loss of S1P2 promotes lymphomagenesis in part by disrupting GC B-cell homeostasis, which in turn may lead to the accumulation of target cells for transformation, as seen in the Bcl-2-Ig model of lymphoma (27).

Future cell type–specific gene disruption studies will be required to definitively determine which cell types are involved in the S1P2-dependent DLBCL suppression. One possibility is that the signaling involves S1P2 receptors in GC B cells, given that: (a) purified normal GC centroblasts from mouse and human tonsils express S1P2 (Supplementary Fig. S6), consistent with the presence of S1P2 cDNA clones in an EST library constructed from purified GC B cells (NCI_CGAP_GCB1); (b) the S1P molecule is present in lymph and plasma at high enough concentrations to constitutively saturate such S1P2 receptors (28); (c) S1P2 receptors can regulate cellular differentiation, proliferation, survival, and migration in an apparently cell-autonomous manner, in vitro (15); and (d) the S1P2 gene is aberrantly targeted by SHM in DLBCL cells, although the functional consequences of these mutations are still unknown (this study). Alternatively, S1P2 receptors may regulate GC B-cell behavior, and DLBCL formation, by acting through other cell types involved in the GC reaction. These include dendritic cells, where expression of S1P2 receptors may regulate their cytokine release (29), thereby indirectly influencing B cells. Furthermore, S1P2 receptor expression has been detected in T cells (30, 31) and the subtle dysregulation of T-cell homeostasis in S1P2−/− mice may mediate enhanced GC precursor activation and increase the risk of lymphomagenesis. Notably, a recent report indicates that T cells can promote the production of large B-cell lymphomas, albeit with different characteristics than those of the S1P2−/− lymphomas (32).

Our study also identifies the S1P2 locus as a novel target of aberrant SHM in DLBCL. Mutations affecting putative S1P2 regulatory sequences were found in >25% of human DLBCL and were comparable in frequencies to other genes that are targeted by aberrant SHM in this disease, including PIM1, PAX5, RhoH/TTF, and MYC (Supplementary Fig. S7; ref. 14). Mutations also display the characteristic features of the IgV-associated SHM mechanism, but were not found at significant levels in purified normal GC B cells, indicating that they represent a tumor-specific phenomenon, presumably resulting from an aberrant function of the SHM machinery in DLBCL (14), and raising the possibility that some of these events may have been selected by the neoplastic clone for a role in deregulating S1P2 expression. In the cell lines analyzed, the expression levels of S1P2 are variable, with some cases displaying very low to absent S1P2 mRNA. This scenario may in part reflect the known phenotypic and molecular heterogeneity of DLBCL (see the lack of significant S1P2 expression in both Ly3 and Ly10, which represents activated B-cell type DLBCL; ref. 33). However, S1P2 mRNA was also essentially absent in cell lines belonging to the GC B cell–derived subgroup, including the one (Ly1) that carries the highest number of mutations (Supplementary Fig. S5). Because S1P2 is normally expressed in GC B cells, its absence in some DLBCL may represent a pathogenetic event. Given the extreme heterogeneity of the mutations, and their variable combination in individual cases, additional experiments will be required to determine the direct functional effect of each mutated allele on the regulation of S1P2 expression and on DLBCL pathogenesis in vivo.

Notably, in a number of human DLBCL cases, the protein-coding sequence and presumed regulatory regions of the S1P2 gene are apparently intact. Therefore, although S1P2 mutations may contribute to DLBCL in some cases (as discussed above), the intact S1P2 receptors in the remaining cases raise the possibility that the receptor could serve as a potential therapeutic target in which S1P2 agonist treatment may decrease DLBCL formation by enhancing S1P2-dependent tumor suppression mechanisms. Clearly, the cellular and molecular aspects of the tumor suppression mechanisms need to first be elucidated.

The immunosuppressive drug FTY720 (FTY720-P) acts on S1P receptors to reduce T-cell and B-cell egress from lymphoid organs (31). Some data suggest it downregulates S1P2 in addition to its effects on other S1P receptors (34). However, the S1P2−/− effects described here are likely unrelated given that: (a) the lymphocyte egress effects of FTY720 are mediated by the S1P1 receptor (31, 35); and (b) the S1P2−/− mice display neither the reduced blood lymphocyte populations (36, 37) nor the marginal zone B-cell displacement (35) seen with FTY720 administration.

The histology of the S1P2−/− lymphomas, along with their expression of specific GC markers such as BCL6 and PNA, and the presence of IgV mutations, all strongly resemble the features of human DLBCL (38). In this respect, the S1P2−/− tumors share significant properties with the tumors resulting from constitutive expression of BCL6 in B cells (16). In both models, GC-derived DLBCL develops with age following an increase in GC number, although the BCL6 mice display a modest splenomegaly and evidence of lymphoproliferative disease not seen with S1P2−/− mice. TCL1 overexpression in B and T cells and disruption of the Bad gene have also been reported to promote GC-derived DLBCL, although via different mechanisms that seem to be dependent on modification of B-cell proliferation and/or apoptosis (39, 40). Given the relatively higher incidence and the homogeneous phenotype of the S1P2−/− lymphomas, this model may serve as a valuable new tool in the effort to determine the mechanisms regulating DLBCL formation.

A.J. MacLennan holds a minority interest in S1P2-related patents. The other authors declared no potential conflicts of interest.

Grant support: NIH grant CA37295 and a Leukemia & Lymphoma Society SCOR grant (R. Dalla-Favera).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Danae Washton, Adina Grunn, and Glenn Doerman for excellent technical help, Lin Yang, Greg Boivin, the Molecular Pathology Facility of the Herbert Irving Cancer Center of Columbia University Medical Center, and the University of Cincinnati College of Medicine Pathology Core Facility for the superb histology service. L.P. is a Julie Gould Scholar.

1
Takabe
K
,
Paugh
SW
,
Milstien
S
,
Spiegel
S
. 
“Inside-out” signaling of sphingosine-1-phosphate: therapeutic targets
.
Pharmacol Rev
2008
;
60
:
181
95
.
2
Rivera
R
,
Chun
J
. 
Biological effects of lysophospholipids
.
Rev Physiol Biochem Pharmacol
2006
;
160
:
25
46
.
3
Choi
JW
,
Lee
C-W
,
Chun
J
. 
Biological roles of lysophospholipid receptors revealed by genetic null mice: an update
.
Biochim Biophys Acta
2008
;
1781
:
531
9
.
4
Anliker
B
,
Chun
J
. 
Cell surface receptors in lysophospholipid signaling
.
Cell Dev Biol
2004
;
15
:
457
65
.
5
Toman
RE
,
Spiegel
S
. 
Lysophospholipid receptors in the nervous system
.
Neurochem Res
2002
;
27
:
619
27
.
6
MacLennan
AJ
,
Carney
PR
,
Zhu
WJ
, et al
. 
An essential role for the H218/AGR16/Edg-5/LP(B2) sphingosine 1-phosphate receptor in neuronal excitability
.
Eur J Neurosci
2001
;
14
:
203
9
.
7
MacLennan
AJ
,
Benner
SJ
,
Andringa
A
, et al
. 
The S1P2-sphingosine 1-phosphate receptor is essential for auditory and vestibular function
.
Hear Res
2006
;
220
:
38
48
.
8
Herr
DR
,
Grillet
N
,
Schwander
M
,
Rivera
R
,
Muller
U
,
Chun
J
. 
Sphingosine 1-phosphate (S1P) signaling is required for maintenance of hair cells mainly via activation of S1P2
.
J Neurosci
2007
;
27
:
1474
8
.
9
Kono
M
,
Belyantseva
IA
,
Skoura
A
, et al
. 
Deafness and stria vascularis defects in S1P2 receptor-null mice
.
J Biol Chem
2007
;
282
:
10690
6
.
10
Lorenz
JN
,
Arend
LJ
,
Robitz
R
,
Paul
RJ
,
MacLennan
AJ
. 
Vascular dysfunction in S1P2 sphingosine 1-phosphate receptor knockout mice
.
Am J Physiol Regul Integr Comp Physiol
2007
;
292
:
R440
6
.
11
Arikawa
K
,
Takuwa
N
,
Yamaguchi
H
, et al
. 
Ligand-dependent inhibition of B16 melanoma cell migration and invasion via endogenous S1P2 G protein-coupled receptor. Requirement of inhibition of cellular RAC activity
.
J Biol Chem
2003
;
278
:
32841
51
.
12
Yamaguchi
H
,
Kitayama
J
,
Takuwa
N
, et al
. 
Sphingosine-1-phosphate receptor subtype-specific positive and negative regulation of Rac and haematogenous metastasis of melanoma cells
.
Biochem J
2003
;
374
:
715
22
.
13
Lepley
D
,
Paik
JH
,
Hla
T
,
Ferrer
F
. 
The G protein-coupled receptor S1P2 regulates Rho/Rho kinase pathway to inhibit tumor cell migration
.
Cancer Res
2005
;
65
:
3788
95
.
14
Pasqualucci
L
,
Neumeister
P
,
Goossens
T
, et al
. 
Hypermutation of multiple proto-oncogenes in B-cell diffuse large cell lymphomas
.
Nature
2001
;
412
:
341
6
.
15
Ye
BH
,
Cattoretti
G
,
Shen
Q
, et al
. 
The BCL-6 proto-oncogene controls germinal-centre formation and Th2-type inflammation
.
Nat Genet
1997
;
16
:
161
70
.
16
Cattoretti
G
,
Pasqualucci
L
,
Ballon
G
, et al
. 
Deregulated BCL6 expression recapitulates the pathogenesis of human diffuse large B cell lymphomas in mice
.
Cancer Cell
2005
;
7
:
445
55
.
17
Jolly
CJ
,
Klix
N
,
Neuberger
MS
. 
Rapid methods for the analysis of immunoglobulin gene hypermutation: application to transgenic and gene targeted mice
.
Nucleic Acids Res
1997
;
25
:
1913
9
.
18
Morse
HC
 III
,
Anver
MR
,
Fredrickson
TN
, et al
. 
Hematopathology subcommittee of the mouse models of human cancers consortium: Bethesda proposals for classification of lymphoid neoplasms in mice
.
Blood
2002
;
100
:
246
58
.
19
Muramatsu
M
,
Kinoshita
K
,
Fagarasan
S
,
Yamada
S
,
Shinkai
Y
,
Honjo
T
. 
Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme
.
Cell
2000
;
102
:
553
63
.
20
Yamada
M
,
Wasserman
R
,
Reichard
BA
,
Shane
S
,
Caton
AJ
,
Rovera
G
. 
Preferential utilization of specific immunoglobulin heavy chain diversity and joining segments in adult human peripheral blood B lymphocytes
.
J Exp Med
1991
;
173
:
395
407
.
21
Girkontaite
I
,
Sakk
V
,
Wagner
M
, et al
. 
The sphingosine-1-phosphate (S1P) lysophospholipid receptor S1P3 regulates MAdCAM-1+ endothelial cells in splenic marginal sinus organization
.
J Exp Med
2004
;
200
:
1491
501
.
22
Sancho
D
,
Gomez
M
,
Sanchez-Madrid
F
. 
CD69 is an immunoregulatory molecule induced following activation
.
Trends Immunol
2005
;
26
:
136
40
.
23
Storb
U
,
Peters
A
,
Klotz
E
, et al
. 
Somatic hypermutation of immunoglobulin genes is linked to transcription
.
Curr Top Microbiol Immunol
1998
;
229
:
11
9
.
24
Shen
HM
,
Peters
A
,
Baron
B
,
Zhu
X
,
Storb
U
. 
Mutation of BCL-6 gene in normal B cells by the process of somatic hypermutation of Ig genes
.
Science
1998
;
280
:
1750
2
.
25
Pasqualucci
L
,
Migliazza
A
,
Fracchiolla
N
, et al
. 
BCL-6 mutations in normal germinal center B cells: evidence of somatic hypermutation acting outside Ig loci
.
Proc Natl Acad Sci U S A
1998
;
95
:
11816
21
.
26
Neuberger
MS
,
Ehrenstein
MR
,
Klix
N
, et al
. 
Monitoring and interpreting the intrinsic features of somatic hypermutation
.
Immunol Rev
1998
;
162
:
107
16
.
27
McDonnell
TJ
,
Korsmeyer
SJ
. 
Progression from lymphoid hyperplasia to high-grade malignant lymphoma in mice transgenic for the t(14; 18)
.
Nature
1991
;
349
:
254
6
.
28
Goetzl
EJ
,
Wang
W
,
McGiffert
C
,
Huang
M-C
,
Graler
MH
. 
Sphingosine 1-phosphate and its G protein-coupled receptors constitute a multifunctional immunoregulatory system
.
J Cell Biochem
2004
;
92
:
1104
14
.
29
Idzko
M
,
Panther
E
,
Corinti
S
, et al
. 
Sphingosine 1-phosphate induces chemotaxis of immature and modulates cytokine-release in mature human dendritic cells for emergence of Th2 immune responses
.
FASEB J
2002
;
16
:
625
7
.
30
Graeler
M
,
Goetzl
EJ
. 
Activation-regulated expression and chemotactic function of sphingosine 1-phosphate receptors in mouse splenic T cells
.
FASEB J
2002
;
16
:
1874
8
.
31
Matloubian
M
,
Lo
CG
,
Cinamon
G
, et al
. 
Lymphocyte egress from thymus and peripheral lymphoid organs is dependent on S1P receptor 1
.
Nature
2004
;
427
:
355
60
.
32
Zangani
MM
,
Froyland
M
,
Qiu
GY
, et al
. 
Lymphomas can develop from B cells chronically helped by idiotype-specific T cells
.
J Exp Med
2007
;
204
:
1181
91
.
33
Alizadeh
AA
,
Eisen
MB
,
Davis
RE
, et al
. 
Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling
.
Nature
2000
;
403
:
503
11
.
34
Graler
MH
,
Goetzl
EJ
. 
The immunosuppressant FTY720 down-regulates sphingosine 1-phosphate G-protein-coupled receptors
.
FASEB J
2004
;
18
:
551
3
.
35
Cinamon
G
,
Matloubian
M
,
Lesneski
MJ
, et al
. 
Sphingosine 1-phosphate receptor 1 promotes B cell localization in the splenic marginal zone
.
Nat Immunol
2004
;
5
:
713
20
.
36
Mandala
S
,
Hajdu
R
,
Bergstrom
J
, et al
. 
Alteration of lymphocyte trafficking by sphingosine-1-phosphate receptor agonists
.
Science
2002
;
296
:
346
9
.
37
Chiba
K
,
Yanagawa
Y
,
Masubuchi
Y
, et al
. 
FTY720, a novel immunosuppressant, induces sequestration of circulating mature lymphocytes by acceleration of lymphocyte homing in rats. I. FTY720 selectively decreases the number of circulating mature lymphocytes by acceleration of lymphocyte homing
.
J Immunol
1998
;
160
:
5037
44
.
38
Küppers
R
,
Klein
U
,
Hansmann
ML
,
Rajewsky
K
. 
Cellular origin of human B-cell lymphomas
.
N Engl J Med
1999
;
341
:
1520
9
.
39
Hoyer
KK
,
French
SW
,
Turner
DE
, et al
. 
Dysregulated TCL1 promotes multiple classes of mature B cell lymphoma
.
Proc Natl Acad Sci U S A
2002
;
99
:
14392
7
.
40
Ranger
AM
,
Zha
J
,
Harada
H
, et al
. 
Bad-deficient mice develop diffuse large B cell lymphoma
.
Proc Natl Acad Sci U S A
2003
;
100
:
9324
9
.