Coordinate suppression of B cell lymphoma by PTEN and SHIP phosphatases

Although deletion of Pten in mouse T cells results in lethal T cell lymphomas (Suzuki et al., 2001), animals lacking Pten in B cells (bPTEN^−/−) show no evidence of malignancy (Anzelon et al., 2003). Similarly, there are no reports of malignancies in mice lacking SHIP in B cells. To determine if PTEN and SHIP coordinately suppress B cell transformation, we generated mice with a B cell–specific deletion of both Pten and Ship (bPTEN/SHIP^−/−) by mating bPTEN^−/− mice (Anzelon et al., 2003) with a novel strain of mice lacking SHIP only in B cells (bSHIP^−/−; see Materials and methods; Fig. S1 A). As early as 4 mo of age, bPTEN/SHIP^−/− (but not bPTEN^−/−, bSHIP^−/−, or bPTEN/SHIP^+/−) mice exhibited lethargy, weight loss (Fig. 1 A), hunched posture, ruffled fur, and splenomegaly (Fig. 1, A and B). Severe morbidity and death occurred in all bPTEN/SHIP^−/− animals by 1 yr of age (Fig. 1 C and not depicted). In contrast, WT, bPTEN^−/−, and bSHIP^−/− controls did not develop splenomegaly and failed to display any signs of disease noted in bPTEN/SHIP^−/− mice (Fig. 1, B and C). As compared with spleens of control animals, spleens of aged bPTEN/SHIP^−/− mice (>6 mo of age) displayed an expansion of CD19⁺ B cells, resulting in enlarged white pulp areas that often infiltrated and compressed the red pulp (Fig. 1, D and E; and not depicted). Notably, tissues containing the expanded B cell population also displayed an expansion of CD11b⁺ myeloid cells (Fig. 1 E and not depicted). Concurrent with disease onset, bPTEN/SHIP^−/− B cells and myeloid cells were found in liver and lung and, less frequently, kidney infiltrates and often formed visible masses in nonlymphoid tissues (Fig. 1 F). Thus, unlike loss of either PTEN or SHIP alone, loss of both PTEN and SHIP in B cells results in an aggressive and fatal disease associated with immune cell expansion.

To determine whether the disease observed in bPTEN/SHIP^−/− mice represents a lymphoproliferative disease or a neoplastic disorder, we examined the phenotypes of peripheral B cells in WT, bPTEN^−/−, bSHIP^−/−, and bPTEN/SHIP^−/− animals. Flow cytometric analyses of peripheral blood lymphocytes (PBL) showed a reduction in the frequency of B cells in asymptomatic bPTEN/SHIP^−/− mice (Fig. 2 A, pre, ≤6 mo) as compared with WT, bPTEN^−/−, or bSHIP^−/− mice (Fig. 2 A), perhaps as a result of retention of B cells in the lymphoid tissues. bPTEN/SHIP^−/− animals older than 6 mo had a threefold increase in the percentage of recirculating B cells in the blood, concurrent with the onset of disease (Fig. 2 A). Notably, the percentage of CD19⁺ B cells in bPTEN/SHIP^−/− animals was comparable to that of WT, bPTEN^−/−, and bSHIP^−/− mice, even in the blood of diseased bPTEN/SHIP^−/− mice (Fig. 2 B). However, in the latter group, the associated severe splenomegaly likely indicates a parallel expansion or recruitment of non–B cells into the spleen. Moreover, although diseased bPTEN/SHIP^−/− mice had hematocrits and hemoglobin values comparable to those of age-matched WT controls, peripheral blood smears revealed lymphocytes that resembled lymphoma cells based on cell size and nuclear features (unpublished data).

Flow cytometric analyses of recirculating CD19⁺ bPTEN/SHIP^−/− B cells showed that they were larger than B cells from WT, bPTEN^−/−, or bSHIP^−/− mice, as indicated by forward light scatter (Fig. 2 C). bPTEN/SHIP^−/− B cells were predominantly B220^low/−IgM^+/lowIgD⁻CD11b⁺CD5^+/lowCD43⁺CD21^−/lowCD23⁻ (Fig. 2, B and C) but did not express markers characteristic of T cells (CD3, CD4, and CD8) or markers of dendritic cells (CD11c), granulocytes (Gr1), or macrophages/myeloid cells (F4/80; not depicted). In asymptomatic prelymphoma mice, splenic B cells exhibited bimodal expression of some markers, indicating the presence of normal and altered B cell populations (Fig. 2 C). By comparison, splenic B cells and B cells present in liver masses of diseased bPTEN/SHIP^−/− mice shared an altered yet uniform phenotype, suggesting that bPTEN/SHIP^−/− B cells found in nonlymphoid organs had metastasized from secondary lymphoid tissues (Fig. 2 C). The phenotype of bPTEN/SHIP^−/− B cells was similar to that of B1a B cells that predominate in the peritoneal and pleural cavities of WT mice (Hardy et al., 2006) but was quite distinct from that of normal follicular, MZ, GC, or memory B cells. Importantly, however, most documented mouse B cell lymphomas, whether derived from MZ or GC B cells, are CD5^low, and a significant proportion also express CD11b (Fig. 2 C; Morse et al., 2002). Collectively, the surface phenotype and increased size of bPTEN/SHIP^−/− B cells is consistent with B cell neoplasia.

The majority of histologically defined mouse lymphomas are monoclonal or oligoclonal, as demonstrated by Southern blot analysis of Ig heavy chain (IgH) J gene organization or by PCR analysis of IgH V gene usage (Morse et al., 2002). Thus, to further distinguish between neoplastic and lymphoproliferative disease, genomic DNA from splenic B cells of age-matched WT and bPTEN/SHIP^−/− mice was analyzed by Southern blotting for clonal rearrangements at the IgH locus using the J_H-specific probe pJ11 (Stall et al., 1988). Numerous J_H rearrangements were detected in genomic DNA from B cells of WT and asymptomatic (prelymphoma) bPTEN/SHIP^−/− mice (Fig. 3 A), indicative of a polyclonal B cell pool. In striking contrast, distinct nongermline bands were identified in DNA from several diseased (lymphoma) bPTEN/SHIP^−/− mice, revealing the presence of clonal B cell outgrowths (Fig. 3 A). PCR analyses using V_H1- and V_H5-specific 5′ and shared 3′ J_H primers were used to determine if B cells present in nonlymphoid tissues and spleen were of common origin in individual mice (Fig. 3 B). In several cases, the same dominant rearrangement was observed in B cells isolated from several tissues within the same animal, indicating that the same clones had disseminated from lymphoid to nonlymphoid organs in diseased bPTEN/SHIP^−/− mice (Fig. 3 B).

Mouse B lymphoma cells differ from untransformed WT B cells in their capacity to expand and propagate disease after transfer into immunocompromised hosts. To determine if bPTEN/SHIP^−/− B cells would transfer disease, WT or bPTEN/SHIP^−/− B cells were transferred intravenously into T cell–deficient (TCR-βδ^−/−) recipients, and the presence of transferred cells was monitored in peripheral blood at regular intervals. Consistent with their designation as transformed cells, bPTEN/SHIP^−/− B cells expanded after adoptive transfer into TCR-βδ^−/− recipients, whereas WT B cells failed to persist (Fig. 4, A and B).

By 3 mo after transfer, CD19⁺B220^lowCD5⁺CD11b⁺ B cells represented 60–70% of peripheral blood B cells in recipients that received bPTEN/SHIP^−/− B cells (Fig. 4, B and C). In contrast, WT B cells neither survived nor expanded after transfer (Fig. 4, B and C). Examination of IgH usage in B cells transferred into TCR-βδ^−/− recipients showed that all WT B cells (before and after transfer) showed a diverse polyclonal IgH repertoire (Fig. S2). Several TCR-βδ^−/− recipient animals that received bPTEN/SHIP^−/− B cells showed expansion of the same clones as the original transferred B cells, whereas others showed an expansion of unique clones that may not have been well represented in the original transferred population but expanded after transfer (Fig. S2). Additionally, 2–5 mo after adoptive transfer, recipients of bPTEN/SHIP^−/− B cells exhibited signs of disease similar to those observed in unmanipulated bPTEN/SHIP^−/− mice (Fig. 4 D) and, at necropsy, were found to have extensive cellular infiltrations of lungs, livers, and other nonlymphoid tissues (not depicted). The combined evidence of cell surface phenotype, clonality, and ability to transfer disease indicates that bPTEN/SHIP^−/− mice develop B cell neoplasia rather than a B cell proliferative disease.

Similar to B cells from healthy mice, most explanted lymphoma cells generally do not replicate spontaneously ex vivo (despite the aggressive nature of most lymphomas in vivo), indicating that environmental stimuli are required for continued growth. In addition, however, mature B cells require tonic signaling via the BCR for maintenance and survival (Lam et al., 1997; Kraus et al., 2004). To understand which signals may contribute to the development of lymphomas in bPTEN/SHIP^−/− mice, we first examined the in vitro proliferative responses of B cells lacking PTEN and/or SHIP to BCR stimulation. Consistent with previous results, bPTEN^−/− and bSHIP^−/− B cells proliferated robustly upon ligation of the BCR with anti-IgM F(ab’)₂ fragments (Helgason et al., 2000; Anzelon et al., 2003; Suzuki et al., 2003). Intriguingly, however, bPTEN/SHIP^−/− B cells were less responsive than bPTEN^−/−, bSHIP^−/−, or WT B cells (Fig. 5 A). Notably, bPTEN/SHIP^−/− B cells were not generally refractory to proliferative stimuli, as they showed robust proliferative response to LPS or anti-CD40 that exceeded those of WT B cells and were comparable to those of bPTEN^−/− B cells (Fig. S3). Because the adoptive transfer studies indicated that T cells or T cell–derived factors were not required for lymphomagenesis, we considered a role for BAFF, as it is broadly available in vivo and critical for B cell survival (Mackay and Schneider, 2009). The addition of BAFF to anti-IgM–stimulated cells enhanced proliferation in B cells from WT, bPTEN^−/−, bSHIP^−/−, and bPTEN/SHIP^−/− mice (Fig. 5 A). Strikingly, treatment with BAFF alone induced proliferation of bPTEN/SHIP^−/− but not of WT, bPTEN^−/−, or bSHIP^−/− B cells (Fig. 5 A). Proliferation of bPTEN/SHIP^−/− B cells to BAFF was unique, as treatment with the related cytokine, a proliferation-inducing ligand (APRIL), failed to induce a proliferative response (Fig. 5 A).

To examine how BCR ligation and BAFF stimulation promote cell cycle entry in bPTEN/SHIP^−/− B cells, expression levels of p27^kip1 and cyclin D3 were examined. The inability of BAFF alone to induce proliferation of WT B cells results from a failure to down-regulate p27^kip1 (Huang et al., 2004), a cyclin-dependent kinase inhibitor which is negatively regulated by Akt either directly or via inhibition of the FOXO factors. As compared with WT and single-deficient B cells, bPTEN/SHIP^−/− B cells assayed directly ex vivo displayed diminished expression of p27^kip1 as well as enhanced expression of cyclin D3, an indicator of entry into the G1 phase of the cell cycle (Fig. 5 B). Consistent with these results, activated Akt, indicated by phosphorylation at serine 473, was readily detected in unstimulated bPTEN/SHIP^−/− B cells. This is in contrast to WT and single knockout controls which failed to show basal Akt activity (Fig. 5 C). These data suggest a mechanism by which bPTEN/SHIP^−/− B cells are biochemically poised to proliferate after BAFF stimulation.

WT, bPTEN^−/−, bSHIP^−/−, and bPTEN/SHIP^−/− B cells were stimulated in vitro with anti-IgM in the presence or absence of BAFF. Compared with controls, bPTEN/SHIP^−/− B cells consistently showed decreased expression of p27^kip1 and increased cyclin D3 expression under all stimulation conditions (Fig. 5 B). Furthermore, signaling through the BCR in bPTEN/SHIP^−/− B cells in the presence or absence of BAFF also resulted in hyperactivation of Akt in bPTEN/SHIP^−/− B cells (Fig. 5 C). Together, these data indicate that although bPTEN/SHIP^−/− B cells are biochemically hyperresponsive to signals through the BCR and BAFF receptors and up-regulate cyclin D3 in response to BCR stimulation, signals through the BCR and BAFF receptors alone are likely not sufficient to support their survival and expansion in vivo.

To directly assess the contribution of BAFF signaling to lymphoma progression, WT or bPTEN/SHIP^−/− B cells were transferred into sublethally irradiated BAFF^−/− animals and monitored over a 6-mo period for the appearance of lymphoma cells in the peripheral blood. As expected, WT B cells failed to survive and expand in the absence of BAFF in vivo (Fig. S4 A). However, we found that bPTEN/SHIP^−/− B cells were able to expand in some (4/11) BAFF^−/− recipients (Fig. S4 A). This result cannot be explained by autocrine production of BAFF, as BAFF message was not detectable by RT-PCR analysis of bPTEN/SHIP^−/− B cells (Fig. S4 B). Moreover, blockade of BAFF in in vitro cultures of bPTEN/SHIP^−/− B cells did not result in reduced viability of bPTEN/SHIP^−/− B cells as compared with untreated cells (Fig. S4 D). Thus, although BAFF provides a mitogenic signal to bPTEN/SHIP^−/− B cells, it is not an absolute requirement for lymphoma progression in situations in which PI3K signaling is deregulated.

Mouse B cell lymphoma cells not only survive and expand after transfer into immunocompromised recipient animals but they also persist in vitro. Given the presence of activated Akt in freshly isolated bPTEN/SHIP^−/− B cells (Fig. 5 C), we proceeded to examine the survival of B cells lacking expression of PTEN and/or SHIP. Splenic B cells were purified from WT, bPTEN^−/−, bSHIP^−/−, and bPTEN/SHIP^−/− mice and cultured in the absence of mitogenic stimuli or survival factors. In contrast to their rapid in vivo expansion after transfer into TCR-βδ^−/− hosts, bPTEN/SHIP^−/− lymphoma cells persisted but did not proliferate when cultured in the absence of stimulation ex vivo (Fig. 6 A). In contrast, B cells from WT and single mutant bPTEN^−/− and bSHIP^−/− mice did not persist and died more rapidly than bPTEN/SHIP^−/− B cells (Fig. 6 A). In this regard, bPTEN/SHIP^−/− B cell cultures contained an increased percentage of live cells as indicated by annexin V and propidium iodide staining compared with cultures of WT or single knockout B cells (Fig. 6 B). Consistent with increased survival and basally activated Akt (Fig. 6 C), levels of GSK3β(S9) phosphorylation, corresponding to the negative regulatory site for Akt, were also augmented in bPTEN/SHIP^−/− B cells (Fig. 6 C). Notably, phosphorylation of Akt and GSK3β in bPTEN/SHIP^−/− B cells was detected not only in freshly isolated B cells but also in cells from long-term culture. Consistent with inhibition of GSKβ by Akt and GSK3β-dependent degradation of the prosurvival factor MCL1 (Maurer et al., 2006), bPTEN/SHIP^−/− B cells expressed elevated amounts of MCL1 compared with WT, bPTEN^−/−, or bSHIP^−/− B cells at all time points examined (Fig. 6 C). Moreover, freshly isolated bPTEN/SHIP^−/− B cells expressed lower levels of the pro-apoptotic factor and FOXO-regulated gene product Bim (Dijkers et al., 2000) than WT or bSHIP^−/− controls (Fig. 6 D). Interestingly, bPTEN^−/− B cells also expressed lower levels of Bim than WT or bSHIP^−/− B cells but did not survive in vitro, suggesting that factors in addition to Bim may be responsible for the enhanced survival and expansion of bPTEN/SHIP^−/− B cells both in vivo and in vitro.

The histological and cytological features of lymphoid tissues of bPTEN/SHIP^−/− mice were consistent with the diagnosis of B cell lymphoma. To investigate the type of lymphomas that develop in bPTEN/SHIP^−/− animals, lymphomas were classified in accordance with the Bethesda proposals for mouse lymphoid neoplasms (Morse et al., 2002). Although the tumors of nearly all mice exhibited histological and cytological features most closely resembling those described for mouse splenic MZ lymphomas (MZL; Fig. 7, A and B; Morse et al., 2002), certain distinguishing characteristics suggested that bPTEN/SHIP^−/− lymphomas were distinct from previously observed cases of MZL (Fredrickson et al., 1999; Morse et al., 2002; Shin et al., 2004). For example, MZLs are initially low-grade tumors restricted to the splenic MZ, which is located just outside the follicles, but later invade both the white and red pulp when they progress to high-grade tumors (Morse et al., 2002; Shin et al., 2004). Histological examination of spleen sections revealed that bPTEN/SHIP^−/− lymphomas were localized to follicles in this organ, often completely replacing normal structures, leaving virtually no small lymphocytes in the periarteriolar lymphoid sheaths (Fig. 1 D and Fig. 7 A). Moreover, splenic red pulp involvement was less prominent compared with reported cases of MZL but exhibited similar degrees of white pulp compression (Fig. 1 D; Morse et al., 2002; Shin et al., 2004).

Low-grade MZLs are typically comprised of uniform cells of medium size that are larger than, but cytologically very similar to, normal MZ B cells (Fredrickson et al., 1999; Morse et al., 2002). In contrast, high-grade MZLs reportedly retain their physical relationship to the MZ but take on features indistinguishable from those of centroblastic DLBCLs (CBLs; Fig. 7 A). Some cases of bPTEN/SHIP^−/− lymphomas exhibited mixed cell populations of centrocytes and centroblasts, which is consistent with a diagnosis of follicular B cell lymphoma (FL), whereas others were dominated by centroblasts as seen in CBL. Still others were characterized by sheets of plasma cells, which is consistent with a diagnosis of plasmacytoma (Fig. 7, A and B). It is noteworthy that both low- and high-grade MZLs are almost always restricted to the spleen, whereas we consistently observed bPTEN/SHIP^−/− lymphoma infiltrates in nonlymphoid tissues, including liver, lung, heart, and, occasionally, kidney (Fredrickson et al., 1999; Morse et al., 2002). bPTEN/SHIP^−/− B cells did not predominate in the bone marrow, which occurs in B cell leukemia.

To corroborate these findings, we assessed expression of SHIP and PTEN transcripts in splenic samples from NFS.V congenic mice that spontaneously develop a spectrum of B lymphomas (Hartley et al., 2000; Shin et al., 2008). PTEN is down-regulated in all lymphoma types, and SHIP is down-regulated in two types of DLBCL (CBL and IBL) and MZL but is statistically unchanged for FL. Collectively, these findings indicate that bPTEN/SHIP^−/− mice develop lymphomas with overlapping similarities to several previously described subsets of mouse mature B cell lymphomas.

To examine the potential utility of the bPTEN/SHIP^−/− mouse as a model for human disease, levels of PTEN and/or SHIP transcripts in human B cell lymphomas were examined. The two most prevalent human B cell lymphomas, FL and DLBCL, represent >55% of all human mature B cell lymphomas (Evans and Hancock, 2003). Mutations that augment PI3K signaling and inactivating PTEN mutations have recently been reported in some human DLBCL cases (Uddin et al., 2006; Abubaker et al., 2007; Lenz et al., 2008). Accordingly, existing patient microarray gene expression profiling data revealed a reduction in mean PTEN and SHIP expression in DLBCL compared with normal B cells (including naive, memory, and GC B cells) and the less aggressive FL (Alizadeh et al., 2000; Fig. 8 A).

To determine if subsets of DLBCL (previously distinguished based on differences in BCL6 expression; Alizadeh et al., 2000) could be distinguished by levels of transcripts for PTEN and SHIP, data for 254 cases of DLBCL were examined and classified into the following three groups based on PTEN, SHIP, and BCL6 expression: (1) the 25% of cases with the lowest levels of PTEN transcripts; (2) the 25% of cases with the lowest levels of SHIP transcripts; and (3) the 25% of cases with the highest levels of BCL6 transcripts. These analyses revealed that patient samples with the lowest relative levels of PTEN (group 1) had SHIP transcripts significantly lower than the mean for all patients (Fig. 8 B, right, blue bar). Similarly, patient samples with low SHIP expression (group 2) had significantly lower levels of PTEN transcripts than the mean for all patients (Fig. 8 B, left, orange bar). Patients with the highest levels of BCL6 transcripts (group 3), most of which reportedly belong to the GC subset of DLBCL (Alizadeh et al., 2000), had levels of transcripts for both PTEN and SHIP that were significantly above the mean (Fig. 8 A), suggesting that the lymphomas of bPTEN/SHIP^−/− mice are probably not closely aligned to human GC-DLBCL. These results are consistent with coordinate tumor suppressor functions of these two phosphatases in B cells.

Cellular transformation in mammals is a complex and often multigenic process. Therefore, identifying and testing the genetic and epigenetic bases of particular malignancies has been challenging. In recent years, various studies have identified the tumor suppressor PTEN as a target in multiple tumor types and disease models. Studies of malignancies driven by loss-of-function PTEN mutations and gain-of-function mutations in PI3K subunit genes have established the relevance of the PI3K pathway in promoting cancer (Yuan and Cantley, 2008). Surprisingly, a tumor suppressor role for INPP5D, the gene encoding SHIP, which also regulates PI3K signaling, has not been demonstrated (Rohrschneider et al., 2000). In this paper, we provide the first evidence that SHIP acts as a tumor suppressor and that it acts in concert with PTEN to suppress B cell transformation.

Elevated PI3K signaling has recently been noted in selected B cell malignancies, namely MCL and DLBCL (Rudelius et al., 2006; Uddin et al., 2006). MCL patients are highly responsive to rapamycin treatment to inactivate mTOR activity (Witzig and Kaufmann, 2006), and Akt activity is enhanced in many MCL lines that are dependent on PI3K for cell cycle progression and survival (Rudelius et al., 2006). Similarly, bPTEN/SHIP^−/− lymphoma cells are responsive to PI3K inhibitors and rapamycin (unpublished data). Although the incidence of mutations in PTEN in human cancers is second only to that of TP53 (Stiles et al., 2004; Cully et al., 2006), we previously found that its loss in B cells does not cause transformation (Anzelon et al., 2003). In the case of DLBCL, reduced PTEN expression or predicted oncogenic PIK3CA mutations occurred in 37 and 8% of cases, respectively. Both alterations were associated with poor overall survival (Abubaker et al., 2007). Consistent with these findings, PTEN expression was reduced but still present in primary MCL cases exhibiting elevated Akt activity (Rudelius et al., 2006), raising the possibility that SHIP expression or function may also be affected.

Importantly, SHIP has recently been identified as a miR-155 target, causing reduced SHIP expression in a subset of DLBCL cases (Pedersen et al., 2009). This finding was confirmed in a miR-155 transgenic B lymphoma model (Costinean et al., 2009) and also reported in macrophages responding to inflammatory stimuli (O’Connell et al., 2009). Intriguingly, these findings suggest that epigenetic down-regulation of SHIP expression by inflammation-induced miR-155 expression may promote transformation of B cells with impaired PTEN function.

PTEN regulates PI(3,4,5)P₃ levels in all resting and stimulated cells. SHIP, in contrast, is expressed only in hematopoietic cells and depends on membrane recruitment via phosphotyrosine residues or SH3-based interactions for its activation (Leslie et al., 2008). In particular, upon recruitment to ITIM residues after coengagement with FcγRIIB, SHIP mediates an inhibitory effect on BCR signaling. Thus, in the absence of PTEN, BCR-induced PI(3,4,5)P₃ production likely remains dampened by SHIP. Accordingly, we postulate that disease in bPTEN/SHIP^−/− mice is promoted, at least in part, by tonic and/or antigen-mediated BCR signaling, which becomes lymphomagenic in the absence of both phosphatases. Indeed, BCR recognition of microbial or auto-antigens is thought to promote B-NHLs including MZL, CLL, and FL (Damle et al., 1999; Guyomard et al., 2003; Hervé et al., 2005). Clonal bPTEN/SHIP^−/− lymphoma populations support a role for antigen-mediated selection in lymphomagenesis. Moreover, heightened serum IgM and IgM deposits in bPTEN/SHIP^−/− heart and kidneys (unpublished data) are consistent with poly- or autoreactive antigen receptor signaling that may help drive lymphoma expansion in this animal model. Interestingly, however, bPTEN/SHIP^−/− lymphoma cells, which do not spontaneously proliferate, are also quite hypoproliferative in response to BCR stimulation in ex vivo cultures. It is possible that this reflects a refractory or tolerized state, perhaps caused by sustained antigen receptor ligation or tonic signaling. However, biochemistry studies indicate that Akt and other downstream effectors of the PI3K pathway are hyperactivated both basally and after various stimuli, including ligation of the antigen receptor. Thus, it is clear that B cells lacking PTEN and SHIP are not completely refractory to BCR ligation and that activation of PI3K effectors are enhanced after this stimulus. This observation, taken with the collective results presented here, suggests that despite the hypoproliferative phenotype observed ex vivo, bPTEN/SHIP^−/− lymphoma progression is driven, at least partly, by signals through the antigen receptor but likely requires the contribution of additional receptor-mediated signals.

In this regard, BAFF, which promotes B cell survival and supports BCR-dependent clonal expansion of normal B cells, is an attractive candidate. BAFF activates B cells to enter G₁ phase of the cell cycle marked by up-regulation of cyclin D, but cell cycle progression is halted before S phase without p27 down-regulation and cyclin E up-regulation (Huang et al., 2004). Notably, here we found that BAFF alone stimulated proliferation of bPTEN/SHIP^−/− B cells and that the combination of BAFF and anti-IgM stimuli synergistically enhanced proliferation. Indeed, in addition to its role in dampening BCR signaling, SHIP has been linked to suppression of BAFF-induced signals and acts in a BCR-dependent fashion to suppress signaling via other receptors such as CXCR4 (Brauweiler et al., 2007; Crowley et al., 2009). Thus, it is possible that in the absence of PTEN, SHIP prevents transformation by regulating PIP₃ levels induced by the BCR and other mitogenic signals and that in the absence of both phosphatases, the combination of unchecked signals via multiple receptors promotes disease.

A role for BAFF in B cell malignancies has been suggested. Transgenic expression of either BAFF (when combined with TNF deficiency) or the related molecule APRIL results in B cell neoplasia in mice (Mackay et al., 1999; Planelles et al., 2004). Studies in humans implicate BAFF and/or APRIL in mature B cell lymphomas, multiple myeloma, B cell chronic lymphocytic leukemia, and Waldenstrom’s macroglobulinemia (Shivakumar and Ansell, 2006; Tangye et al., 2006). We did not find evidence of autocrine BAFF production by bPTEN/SHIP^−/− cells. Moreover, serum BAFF levels were not reduced, even in mice bearing a large tumor load (unpublished data), confirming that BAFF is not limiting in vivo. bPTEN/SHIP^−/− B cells acquire the ability to proliferate to BAFF, which typically only promotes survival. Although BAFF is absolutely required for late maturation of WT B cells, transfer of bPTEN/SHIP^−/− B cells into BAFF^−/− recipients showed that its presence is not a strict requirement for B lymphoma progression. Nonetheless, lymphoma progression appeared to be less aggressive in BAFF^−/− than in TCR-β/δ^−/− recipients, suggesting that acute depletion of BAFF could deliver a therapeutic benefit in reducing tumor burden.

In summary, the bPTEN/SHIP^−/− model presented in this paper provides the first evidence that SHIP is a tumor suppressor. As such, this model will be useful for elucidating the specific roles of PTEN and SHIP in B cell neoplasia, as well as providing a novel platform for interrogating and therapeutically manipulating the PI3K pathway in cancer. In a broader sense, the bPTEN/SHIP^−/− mice also provide a system with which to investigate B cell–intrinsic (e.g., the BCR) and microenvironmental (e.g., BAFF/APRIL) factors that promote lymphoma induction and progression.

bPTEN^−/− mice were generated as previously described (Anzelon et al., 2003). bSHIP^−/− mice were generated by crossing SHIP^flox/flox (Karlsson et al., 2003) mice with Cd19Cre mice (Rickert et al., 1997) to generate B cell–specific deletion of SHIP. bPTEN/SHIP^−/− mice were generated by crossing PTEN^flox/flox (Lesche et al., 2002), SHIP^flox/flox, and Cd19Cre^+/− mice. All animals were heterozygous for Cd19Cre to achieve deletion of loxP-flanked exons while maintaining one CD19 allele. TCR-βδ^−/− mice were purchased from The Jackson Laboratory. All animals were maintained in the animal facility of the Sanford-Burnham Medical Research Institute (SBMRI). All protocols were approved by the Institutional Animal Care and Use Committee at the SBMRI and were performed in accordance with institutional guidelines and regulations.

To assess disease progression and animal viability/morbidity, cohorts of 15–20 animals per genotype were monitored over the course of 1 yr. Animals displaying signs of morbidity (lethargy, severe weight loss, hunched posture, ruffled fur, and slow movement) were considered morbid, euthanized, and scored as nonsurviving (note: if left alone, mice die within 2–3 d of development of the symptoms described). Hematology, including differentials and cytology, was performed by the University of California San Diego (UCSD) Animal Care Program Diagnostic Laboratory.

For paraffin-embedded sections, tissues were fixed immediately after harvest in 4% paraformaldehyde (Polysciences Inc.) for 24–48 h and stored in 95% ethanol. Paraffin embedding, sectioning, and H&E staining were performed by the UCSD Cancer Center and by the SBMRI Histology Core facilities according to standard protocols. Immunofluorescent staining of frozen tissue sections was performed as previously described (Anzelon et al., 2003) with indicated antibodies and reagents (see Flow cytometry and antibodies).

Single cell suspensions were prepared, counted, and stained with antibodies according to standard procedures. First, cells were stained for 20 min at 4°C with biotinylated antibodies in flow cytometry buffer (1% FBS and 0.1% azide in PBS), followed by incubation with a cocktail of antibodies conjugated to FITC, PE, peridinin chlorophyll protein complex–cyanine 5.5 (PerCP-Cy5.5.), PE-indotricarbocyanine (PE-Cy7), allophycocyanin (APC), or APC-Cy7. The following antibody conjugates were used (eBioscience): CD3 (145-2C11), IgM (II/41), IgD (11–26), CD5 (53–7.3), CD19 (ID3), B220 (RA3-6B2), CD11b (M1/70), CD43 (S7), CD21 (4E3), CD23 (B3B4), F4/80 (BM8), Gr1 (RB6-8C5), CD4 (GK1.5), CD8 (53–6.7), and CD11c (N418). Biotinylated reagents were detected with streptavidin (SAV) conjugated to FITC, PE, APC, PerCP-Cy5.5 (BD), and PE-Cy7 (eBioscience). All data were collected on a FACSCanto flow cytometer (BD) and analyzed using FlowJo software (Tree Star, Inc.). Data are displayed with logarithmic scale. Each plot represents analysis of 1–3 × 10⁴ events.

B cells were purified from spleen cell suspensions by magnetic removal of cells stained with biotinylated antibodies against CD4 (GK1.5), CD3 (145-2C11), CD11c (N418), and Gr-1 (RB6-8C5; eBioscience) followed by negative selection with anti-biotin–conjugated MACS beads (Miltenyi Biotec). Purity of B cells routinely exceeded 85% as determined by flow cytometry. For Southern blot analysis, genomic DNA was digested with EcoRI, fractionated on a 1% agarose gel, and transferred to Zeta probe membrane (Bio-Rad Laboratories) followed by hybridization with oligonucleotide pJ11 probe. The probe was labeled by random hexamer priming using α-[³²P]dCTP.

For PCR analysis, DNA was isolated using phenol-chloroform extraction and was subjected to PCR using one of two degenerate 5′ primers homologous to sequences from one of two different V_H families (J558, 5′-CGAGCTCTCC­ARCACAGCCTW­CATGCATCTCARC-3′; and 7183, 5′-­CGGTACCAAGA­ASAMCCTGTWC­CTGCAAATGASC-3′) and a common 3′ J_H4 primer (5′-TCCCTCAAATGAGCCTCCAAAGTCC-3′). Degenerate nucleotide positions are coded as: R = A or G; W = A or T; S = C or G; and M = A or C. PCR products generated represent V-DJ_H rearrangements having V_H segments from either the distal J558 (VH1) family or the proximal 7183 (VH5) family. Separation on a 1.5% agarose gel allowed for size-based separation of V-DJ_H products containing J_H1 (1.3 kb), J_H2 (0.95 kb), and J_H3 (0.4 kb) segments.

B cells were purified from spleen cell as described in Assessment of IgH rearrangements. For transfers into TCR-βδ^−/− recipients, 10⁷ WT or bPTEN/SHIP^−/− B cells were injected into unmanipulated recipient animals via the tail vein. For transfers into BAFF^−/− recipients, BAFF^−/− mice were sublethally irradiated (5 Gy), followed by injection of 10⁷ WT or bPTEN/SHIP^−/− B cells via the tail vein. Presence of WT and bPTEN/SHIP^−/− was monitored by reduced expression of CD19 (as a result of CD19 heterozygosity) and later by lymphoma-like phenotype (B220^−/lowCD5⁺CD11b⁺CD43⁺IgM^lo) of bPTEN/SHIP^−/− B cells.

B cells were purified from spleen cell suspensions according to standard procedures and cultured in DME medium supplemented with 10% FCS, pen/strep, l-glutamine, Na-pyruvate, nonessential amino acids, and 2-mercaptoethanol (10% media). For proliferation assays, purified B cells were cultured at a concentration of 5 × 10⁴ cells/100 µl in 96-well round-bottom tissue culture plates at 37°C with 10 µg/ml anti-IgM F(ab′)₂ fragments (Jackson ImmunoResearch Laboratories) in the presence or absence of 25 ng/ml BAFF (R&D Systems) or 100 ng/ml APRIL (PeproTech), as indicated. Cells were pulsed with 1 µCi ³H-thymidine at 48 h for an additional 12–16 h and then collected and scintillation counted. The data are displayed as raw cpm values. All assays were conducted in triplicate. For survival assays, 200 µl of purified splenic B cells was plated in flat-bottom 96-well plates in triplicate at a concentration of 5 × 10⁵ cells/ml in 10% media. For assays involving BAFF blockade, cells were either cultured in media alone or in the presence of 10 ng/ml BAFF. Anti-BAFF IgG and control IgG (R&D Systems) were used at a concentration of 500 ng/ml. Every 24 h, cells were harvested and counted in triplicate using a Coulter counter (Beckman Coulter). Data are presented as averaged total cell numbers of three wells counted three times. Survival was also determined by flow cytometry using the AnnV-FITC Apoptosis Detection kit (BioVision Inc.) according to manufacturer’s instructions.

Purified B cells were stimulated with 10 µg/ml anti-IgM F(ab’)₂ fragments in the presence or absence of 25 ng/ml BAFF for the indicated time points and lysed on ice with RIPA buffer (PBS, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, and 10 mM EDTA) supplemented with a protease inhibitor cocktail (Boehringer Mannheim), 10 mM NaF, 1 mM Na₃VO₄, and PMSF. Protein content of cleared lysates was measured using the BCA Protein Assay kit (Thermo Fisher Scientific), and equal protein amounts were resolved on 10% Bis-Tris Gels (Bio-Rad Laboratories or Invitrogen) followed by Western blotting for the indicated proteins. Antibodies raised against p27^kip1, cyclin D3, actin, phospho-Akt (S473), total Akt, phospho-GSK3β (S9), total ERK1/2, and Bim were obtained from Cell Signaling Technology. Anti-MCL1 was purchased from Rockland Immunochemicals. Total anti-GSK3β was purchased from Millipore. Primary antibodies were detected using horseradish peroxidase–labeled donkey anti–rabbit (Jackson ImmunoResearch Laboratories) or anti–mouse antibodies (GE Healthcare). Immune complexes were detected by SuperSignal West Pico Detection chemiluminescence (Thermo Fisher Scientific).

Relative SHIP and PTEN transcript abundance in human samples was acquired from data provided by Alizadeh et al. (2000) via the Oncomine Cancer Profiling Database (www.oncomine.org). To obtain reported values, Alizadeh et al. (2000) normalized all raw array signals to the median chip expression signal (median expression = 1), followed by Log₂ transformation (Log₂ values of 0 indicate mean expression across entire genome). For analysis of PTEN/SHIP co–down-regulation in human diseases, relative phosphatase expression among DLBCL patient subsets was determined using microarray data for 254 patients from Rosenwald et al. (2002; available at http://llmpp.nih.gov/DLBCL/). Mean expression values of PTEN (probe U92436), SHIP (probe U57650), and BCL6 (probe U00115) were calculated in patient subgroups. Low and high expression groups were defined as the lowest or highest 25% for gene expression among all patients. P-values were calculated using a two-tailed unpaired Student’s t test assuming equal variance.

The relative transcript levels of SHIP and PTEN in mouse B cell lymphoma were determined by oligonucleotide arrays (available at GEO Datasets under accession no. GSE12157). The origins and characteristics of primary mouse B cell lineage lymphomas from NFS.V⁺ congenic and the techniques used for transcriptional profiling of those lymphomas were detailed previously (Lee et al., 2006; Shin et al., 2008).

Fig. S1 shows that PTEN and SHIP are efficiently deleted in splenic bPTEN/SHIP^−/− B cells. Fig. S2 shows that bPTEN/SHIP^−/− B cells have clonal Ig repertoires before and after adoptive transfer into TCR-βδ^−/− recipients in contrast with WT B cells which are polyclonal. Fig. S3 shows poor proliferation of bPTEN/SHIP^−/− B cells in response to anti-IgM stimulation but robust proliferation of bPTEN/SHIP^−/− B cells after LPS or anti-CD40 stimulation. Fig. S4 shows that bPTEN/SHIP^−/− B cells can survive and expand after adoptive transfer into BAFF^−/− recipients but do not produce autocrine BAFF.

We thank Dr. Lou Staudt and colleagues for making publicly available the mRNA array data and prognostic information from their analyses of DLBCL patients; Nissi Varki at the University of California, San Diego Cancer Center Histology Core and Robin Newlin at the Sanford-Burnham Medical Research Institute (SBMRI) Histology Core for tissue processing, staining, and analysis; M. Scott (Biogen) for the BAFF^−/− animals; the SBMRI animal care staff; Dr. M. David; and members of the Rickert laboratory for discussions and critical evaluation of the manuscript.

This work was supported by National Institutes of Health Grants HL088686 and CA135531 (R.C. Rickert), F32CA132350 (A.V. Miletic), and K01CA122192 (I.M. Pedersen). D.M. Mills was supported by an Arthritis Foundation Postdoctoral Fellowship. A.V. Miletic was supported by an Irvington Institute for the Cancer Research Institute Postdoctoral Fellowship. This work was also supported, in part, by the Intramural Research Program of the National Institutes of Health, National Institute of Allergy and Infectious Diseases (H.C. Morse, D-M. Shin, and S. Bolland).

The authors have no conflicting financial interests.