Drosophila melanogaster discs large (dlg) is an essential tumor suppressor gene (TSG) controlling epithelial cell growth and polarity of the fly imaginal discs in pupal development. A mammalian ortholog, Dlg1, is involved in embryonic urogenital morphogenesis, postsynaptic densities in neurons, and immune synapses in lymphocytes. However, a potential role for Dlg1 as a mammalian TSG is unknown. Here, we present evidence that loss of Dlg1 confers strong predisposition to the development of malignancies in a murine model of pediatric B-cell acute lymphoblastic leukemia (B-ALL). Using mice with conditionally deleted Dlg1 alleles, we identify a novel “pre-leukemic” stage of developmentally arrested early B-lineage cells marked by preeminent c-Myc expression. Mechanistically, we show that in B-lineage progenitors Dlg1 interacts with and stabilizes the PTEN protein, regulating its half-life and steady-state abundance. The loss of Dlg1 does not affect the level of PTEN mRNAs but results in a dramatic decrease in PTEN protein, leading to excessive phosphoinositide 3-kinase signaling and proliferation. Our data suggest a novel model of tumor suppression by a PDZ domain-containing polarity gene in hematopoietic cancers. Cancer Immunol Res; 1(6); 426–37. ©2013 AACR.
Drosophila melanogaster discs large (Dlg) was one of the first genes isolated in the Drosophila mutational screens causing neoplastic mutations (1, 2). In the fly, dlg loss results in neoplastic transformation of epithelial tissues forming imaginal discs characterized by a loss in apico-basal polarity and an increase in cell proliferation, suggesting that dlg functions in the regulation of cell polarity and cell-cycle progression (2, 3). Mammalian ortholog of dlg, Dlg1, has been implicated in the development of mammalian tumors based on the observations that it can be targeted for degradation by E6 proteins of high-grade human papillomavirus (HPV) strains (4). Human Dlg1 has been shown to interact with various proteins in the cell cycle, such as mitotic Ser/Thr kinase PBK (5), and classic tumor suppressors APC and PTEN (2, 3, 6–8). Dlg1 may function in cell division as it shows dramatic changes in cellular localization during the cell cycle: Dlg is localized at the plasma membrane during G1, in the cytoplasm during S-phase, bound to the mitotic spindle during mitosis, and then localizes to the spindle body and the midbody during cytokinesis (9, 10).
Several recent studies indicate that Dlg1 may be involved in lymphocyte development and activation (3, 11–14). Although the precise mechanism by which Dlg functions remains to be elucidated, Dlg1 has been implicated in the regulation of c-Myc, a proto-oncogene mutated or constitutively expressed in approximately 30% of human cancers and a critical transcription regulator during proliferative expansion of early-stage B-lineage cells (14, 15). In this context, early B-cell precursors undergo vigorous proliferative expansion driven by interleukin (IL)-7 and pre–B-cell receptor (BCR) signaling, and must exit the cell cycle to initiate first the heavy- and then the light-chain immunoglobulin (Ig) genes recombination (16, 17); however, the molecular mechanism that controls these transitions remains incompletely understood. Thus, despite requiring IL-7 signaling to proliferate and survive, B-cell precursors must terminate IL-7 signaling and proliferation to develop past the large pre-B stage (18). Evidence from past studies indicates that signals emanating from the pre-BCR play a role in terminating IL-7 signaling to permit rearrangement of the Ig light-chain gene loci (19, 20). Moreover, these pre-BCR signals have been shown to suppress cell proliferation as well as c-Myc expression (16). Thus, paradoxically, loss of key signaling components of the pre-BCR complex including BLNK and Btk results in increased proliferation of pre-B cells in response to IL-7 and increased rates of development of spontaneous B-cell leukemias and lymphomas (21–23).
In pre-B cells, IL-7 initiates signaling events by heterodimerization of the IL-7Rα chain and γc, leading to trans-phosphorylation of JAK3 and JAK1, phosphorylation of the IL-7Rα chain, and recruitment of SH2-containing STAT proteins, STAT5A and STAT5B (15, 24). This permits STATs to dimerize and translocate into the nucleus, where they act as transcription factors for a number of target genes. The IL-7Rα chain also participates in direct recruitment and activation of the p85 subunit of phosphoinositide 3-kinase (PI3K) that binds to phosphorylated-Tyr449 on the IL-7Rα via its SH2 domain. Mutations of this binding site preclude PI3K-dependent proliferation of B-lineage cells and lead to impairment in B-cell development. The main target of PI3K signaling is the serine/threonine kinase Akt/PKB, which is responsible for many downstream survival- and proliferation-related events (25).
Alterations in the function of the PTEN protein are of major relevance for a wide variety of human cancers (26). The PI3K/Akt pathway has been implicated as a target of PTEN in cancer cells as phosphatidylinositol lipids phosphorylated at the 3′ position by PI3K are the most physiologically relevant PTEN substrates (27). Thus, PTEN is capable of regulating both proliferation and survival by opposing the effects of PI3K activation. Although many mutations of PTEN found in human cancers target the phosphatase domain, a subset of mutations creates a truncated protein lacking the C-terminal PDZ-binding motif thought to have decreased stability (27). In this context, reduction in PTEN levels has been found in approximately 75% of human acute myelogenous leukemia (AML), and PTEN has been functionally linked with proliferation rates of various malignant leukemias and lymphomas.
On the other hand, mutations of p53 are found in approximately 60% of nonhematologic malignancies and inactivation of p53 is frequently observed in B-lineage lymphomas including chronic lymphocytic leukemia (CLL), AML, and acute lymphoblastic leukemia (ALL; refs. 28, 29). Recently, conditional deletion of the p53 gene in B-lineage cells was shown to result in the development of spontaneous B-cell lymphomas (30), indicating that p53 knockout “sensitizes” B cells to develop a gamut of malignancies at various stages of differentiation.
In this report, we show that the interaction of PTEN with a cell polarity protein Dlg1 in early-stage B-lineage cells is required for the regulation of PTEN stability. Thus, Dlg1-loss disrupts PTEN function and leads to an expansion of large pre-B cells and a pool of leukemogenic progenitors resulting in increased morbidity and mortality in two different murine models of precursor B-cell ALL (B-ALL).
Materials and Methods
The generation of the null allele of Dlg1 and of mice in which Dlg1 expression is conditionally terminated by the Cre recombinase–mediated deletion of loxP-flanked Dlg1 (Dlg1flox/flox) has been described previously (11, 14). Dlg1 conditional knockout mice were bred with CD19-Cre, MX1-Cre, and VAV1-Cre mice. c-MyceGFP mice were a gift from Dr. Barry Sleckman (Washington University, St. Louis, MO). c-MyceGFP mice were crossed with both CD19-Cre and MX1-Cre Dlg1 conditional knockout mice. p53flox/flox mice were crossed with CD19-Cre Dlg1 conditional knockout mice. Rag2−/− mice were a gift from M. White (Washington University, St. Louis, MO). Mice were maintained in the specific pathogen-free facility of Washington University School of Medicine (St. Louis, MO) in accordance with institutional policies for animal care and usage.
Single-cell suspensions were prepared from the bone marrow, spleen, and peritoneal cavity and stained with antibodies according to standard protocols. Antibody conjugates specific for the following markers were used: B220, CD2, CD5, CD21/35, CD23, CD43, CD127, CD132, C-Kit, IgD, IgL Lambda, IgL Kappa, and Pre-BCR (BD Pharmingen); CD24, CD25, Tdt (eBioscience); BP.1 (BioLegend); and IgM (Southern Biotech). Cell sorts were performed on FACS Aria II (Becton Dickinson). Intracellular stains were performed by fixing the cells in 2% paraformaldehyde for 15 minutes followed by washing with permeabilization buffer (PBS+2% FBS and 0.1% Saponin). Antibodies were then added for 30 minutes, washed with permeabilization buffer, and analyzed.
Quantitative RT-PCR analysis
IL-7 B-cell cultures, or purified B-cell subsets, were harvested in TRizol (Invitrogen), and RNA was extracted according to the manufacturer's instructions. cDNA was generated using SuperScript First-Strand Reverse Transcriptase Synthesis System (Invitrogen) according to the manufacturer's instructions. For quantitative real-time PCR (qRT-PCR) analysis, primers are listed in Supplementary Table S1. qRT-PCR was performed with the Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) according to standard PCR conditions in a Stratagene Mx3000P RT-PCR system.
A total of 5 × 105 B cells from culture were mixed with 500 μL of p210 BCR/ABL retroviral supernatant and transduced by centrifugation at 7,000 × g for 50 minutes. The transduced cells were placed in Dulbecco's modified Eagle medium (DMEM)-10 until more than 95% of cells became enhanced GFP (eGFP) positive, indicating the presence of BCR/Abl. Cells were then seeded at 105 cells per well in 96-well plates and counted every 2 days to assess proliferation. Alternatively, 2 million BCR/ABL–transduced cells were injected retro-orbitally into Rag2−/− mice. Injected mice were checked daily for signs of ataxia, paralysis, palpable masses, posturing, and lethargy and were scored accordingly. Mice were sacrificed if tumor masses became larger than 2 cm or if they scored 5/5 signs in daily observation. Spleen, blood, liver, lymph nodes, and tumor masses were processed into single-cell suspensions and analyzed by flow cytometry.
CD19-Cre p53−/−–induced B-cell tumorigenesis
CD19-Cre+ p53flox/flox Dlg1flox/flox or flox/+ mice were monitored daily. For survival analysis, the endpoint was determined either by spontaneous death or by sacrifice of the animal because of signs of pain and suffering, rapid weight loss, or the presence of palpable tumor mass in the spleen or lymph nodes.
Histology and immunohistochemisty
Spleen and lymph node samples were placed in formalin and processed in the Washington University Anatomic and Molecular Pathology Core facility.
Data are expressed as mean ± SD, datasets derived from the indicated genotypes were compared using the two-tailed unpaired Student t test or log-rank test for Kaplan–Meier survival studies. Differences were considered statistically significant (*) when P < 0.05 and (**) when P < 0.005.
Additional methods are described in the Supplementary Data.
Loss of polarity gene Dlg1 leads to a partial developmental arrest at the C'-1 stage pre-B cells
We have shown previously that Dlg1-deficient hematopoietic progenitors are capable of generating all major populations of mature lymphoid cells including T and B lineages using RAG-deficient complementation approaches (11, 14). However, the requirement for Dlg1 in early B-cell development has not been investigated. Given recent evidence for the involvement of the polarity gene Dlg1 in the regulation of c-Myc expression in lymphoid cells and its ability to regulate cell-cycle progression in a variety of mammalian cell lines (14, 31), we analyzed the requirement of Dlg1 in the control of cell proliferation during early stages of B-cell differentiation. Recently, we found c-Myc protein expression in a novel stage of B-cell development termed C'-1 (20), in which the expression of c-Myc was associated with proliferative burst upon Ig heavy-chain gene rearrangement and expression in pre-BCR complex. Subsequently, the pre-BCR signals render these cells unresponsive to IL-7 and differentiate into CD25-expressing large-pre B cells (C'-2 subset) that rearrange the Ig light-chain gene loci (20).
Because the deletion of Dlg1 in murine germline is lethal (11), we generated mice with Cre recombinase–mediated deletion of loxP-flanked Dlg1 gene (Dlg1flox/flox; ref. 14). These mice were bred with CD19-Cre gene knockin mice, in which the expression of Cre is restricted to B-lineage cells (32). As an alternative strategy, we generated Mx1-Cre+ Dlg1flox/flox mice in which the Mx1 type I IFN-inducible promoter can be activated by intraperitoneal injection of polyinosinic-polycytidylic acid [poly(I:C)] resulting in inducible expression of Cre followed by deletion of loxP-flanked Dlg1 gene in early lymphoid progenitors, including all pro/pre-B cells (33). As a third approach, we generated Vav1-Cre+ Dlg1flox/flox mice, in which transgenic Cre expression is driven by the Vav1 promoter that is active in all hematopoietic cells (34). To confirm efficient deletion of Dlg1 in pre-B cells, we used CD19-Cre+ Dlg1flox/flox, Mx1-Cre+ Dlg1flox/flox, and Vav1-Cre+ Dlg1flox/flox bone marrow and control (Cre+ Dlgh1flox/+ or Cre− Dlgh1flox/flox) cells cultured in IL-7–supplemented media for 4 days, at which time nearly all cells from both cultures are B220+ with surface traits marking populations of pro/pre-B (or fractions B/C) cells. In all cases, cells derived from Cre+ Dlg1flox/flox, but not Cre+ Dlg1flox/+ or Cre− Dlg1flox/flox mice, show complete deletion of loxP-flanked Dlg1 gene segments and Dlg1 protein (Supplementary Fig. S1A–S1D).
To determine the effects of Dlg1-loss on B-lymphopoiesis in vivo, first we carried out competitive repopulation experiments in which control (wild-type, WT) CD45.1 hematopoietic stem cells (HSC) were mixed at a 1:1 ratio with HSCs from CD19-Cre+ Dlg1flox/flox mice and injected i.v. into irradiated RAG2-deficient recipients. The contribution of CD45.1 (WT) and Dlg1-deficient (knockout) HSCs to generate mature B and T cells in the peripheral lymphoid organs was analyzed after 4 weeks. Strikingly, while the contribution to non–B-lineage cells, including T cells, was similar, analysis of B-lineage cells shows a dramatically increased contribution of WT cells as compared with knockout cells (Fig. 1A–C).
To determine the developmental stage at which Dlg1-loss affects B-lymphopoiesis in vivo, we crossed CD19-Cre+ Dlg1flox/flox mice with a reporter strain harboring c-Myc gene knockin (c-MyceGFP/eGFP mice) that permits direct analyses of c-Myc protein expression in live cells (20, 35). Strikingly, our analyses of bone marrow cells from CD19-Cre+ Dlg1flox/flox c-MyceGFP/eGFP (knockout) and control mice (WT) revealed increased percentages and total numbers of a novel stage C'-1 large pre-B cells marked by the expression of c-MyceGFP in the knockout mice, as compared with WT control mice (Fig. 1D–F).
In this context, we have previously identified and characterized the two novel subsets of pre-B cells (C'-1 and C'-2; described in ref. 20), of which only the C'-1 cells and not the C'-2 cells, respond to IL-7 stimulation (20). These data indicate that Dlg1 is required for the regulation of c-Myc protein expression in large pre-B cells and their proliferative expansion in vivo. Our analyses also show what seems to be a compensatory decrease in percentages and total number of cells at the C'-2 stage in the knockout c-MyceGFP/eGFP mice (Fig. 1E and F). We found similar alterations in the development of C'-1 and C'-2 pre-B cells in Mx1-Cre+ Dlg1flox/flox c-MyceGFP/eGFP mice (data not shown). Taken together, these results show that despite the expansion of the C'-1 stage of Dlg1 conditionally deficient pre-B cells, differentiation into later-stage B cells is delayed indicating a “bottleneck” in B-cell development at the C'-1 stage upon Dlg1-loss. This observation is supported by results from competitive repopulation analyses of the bone marrow. A 50/50 mix of WT and knockout HSC cells (distinguishable by congenic markers CD45.1/2) was adoptively transferred into sublethally irradiated RAG-deficient hosts. As predicted by our hypothesis, the skewing in ratios of WT versus knockout cells in the periphery, but not in the bone marrow where contributions of WT and knockout B-lineage progenitors are equal up to the C/C' stage of development, indicates at least a partial developmental block in the knockout pre-B cell differentiation. These results are consistent with a role for Dlg1 in C'-1 pre-B-cell differentiation (Supplementary Fig. S2), and they are also consistent with our previous reports that immature and mature B-cell subsets in the periphery (Supplementary Fig. S1F), and B1 B cells (Supplementary Fig. S1G) are intact in young adult Dlg-1 knockout mice. Taken together, these data indicate a developmentally regulated, stage-specific role for Dlg1 in the regulation of c-Myc induction and proliferative burst during early stages of B-lymphopoiesis. These observations indicate that the expression of the c-Myc protein in C' large pre-B cells is subject to Dlg1-dependent regulation, thus raising an intriguing possibility that Dlg1 may function to suppress signals emanating from the IL-7R. Furthermore, it is conceivable that Dlg-1 may be involved in the recruitment or expression of components of the pre-BCR complex necessary for the suppression of c-Myc expression and the induction of C'-1 cell differentiation into the nonproliferative C'-2 stage.
Dlg1 regulates IL-7R signaling output and c-Myc induction in large pre-B cells
Given the requirement for Dlg1 in the regulation of C'-1 pre-B-cell proliferation, and the amplification of c-Myc in Dlg1-deficient mice, we assessed the requirement for Dlg1 in regulating IL-7R signaling. Because STAT5 plays a key role in the survival and proliferation of pro-/pre-B cells downstream of IL-7 (15), we analyzed the kinetics of STAT5 activation in response to IL-7 using STAT5 phospho-specific antibodies. We found that in Dlg1-deficient cells STAT5 activation is significantly increased and prolonged as compared with control cells (Fig. 2A), indicating that Dlg1-deficiency leads to amplification of signals mediated by STAT5. Given the effects of Dlg1-loss on c-Myc induction (Fig. 1D–F; ref. 20), alterations of c-Myc expression may account for an abnormal proliferative response in Dlg1-deficient cells in vitro. To test this hypothesis, we purified B220+CD43+ c-MyceGFP-negative cells from WT and knockout c-MyceGFP/eGFP mice and stimulated them with IL-7. Within 24 hours of stimulation, knockout cultures contain significantly higher numbers of c-MyceGFP–expressing cells than WT cultures (Fig. 2B), indicating that c-Myc induction in response to IL-7R stimulation is enhanced in Dlg1-deficient cells.
To determine whether the hyper-responsiveness to IL-7 leads to increased proliferation of Dlg1-deficient pro-B-cell blasts, we cultured knockout and WT cells in vitro in the presence of IL-7 (Fig. 2C). We found a dramatic increase in the proliferative expansion of knockout cells, as compared with WT cells; these findings were corroborated by an increase in the fraction of cells undergoing cell-cycle progression, as evidenced by direct analyses of DNA content (Fig. 2D), and S-phase transition (Fig. 2E). Importantly, staining with anti-IL-7Rα antibody showed similar levels of surface expression of IL-7R complex in both knockout and WT cells, indicating that the increased growth of Dlg1-deficient cells in IL-7 is not due to alterations of receptor expression (Fig. 2F). In this context, analyses of cell viability showed no significant differences between Dlg1-deficient and control cells (Fig. 2G), indicating that the relative increase in the number of knockout cells is not due to improved survival. Cultures in media supplemented with various doses of IL-7 showed similar dose response curves for both knockout and WT cells (Fig. 2H), indicating that loss of Dlg1 does not impart increased sensitivity to low doses of IL-7. Together, these results suggest that while Dlg1 somehow controls IL-7R signaling output, it does regulate its activation threshold.
Consistent with the view that the increased number of B cells in Dlg1-deficient mice is due to increased proliferation in response to IL-7, analyses of the expression of Mcl1, a critical survival factor in early-stage B cells induced by IL-7 (36), showed similar levels of Mcl1 in knockout and WT cells (Fig. 2I). Moreover, we analyzed the expression of Ccnd3, which encodes cyclin D3, a critical cell-cycle regulator and a major transcriptional target of the IL-7R–STAT5 pathway. We found the levels of Ccnd3 transcripts drastically increased in Dlg1-deficient cells, as compared with control cells (Fig. 2J). These results show that Dlg1 is required in the regulation of several key IL-7–inducible genes involved in early B-lymphocyte development, and the loss of Dlg1 results in enhanced c-Myc protein induction in response to IL-7.
Dlg1 is required for stabilization of PTEN in pre-B cells
Alterations of growth in response to IL-7 caused by Dlg1-loss suggested that Dlg1 may be involved in the regulation of PI3K/Akt signaling, a key proproliferative pathway induced by IL-7. First we analyzed Akt activity in lysates of Dlg1-deficient and -sufficient cells stimulated with IL-7. Using surrogate assays with phospho-specific antibodies, we found that phosphorylation of both Akt and one of its major targets, GSK3β, was significantly enhanced in cells lacking Dlg1 (Fig. 3A). A key regulator of this signaling pathway is PTEN; previous studies in neuronal cells (6, 7) suggest that PTEN-Dlg1 interacts through binding mediated by PDZ domains. We tested this potential interaction by overexpressing Myc-tagged Dlg1 and/or Flag-tagged PTEN proteins in HEK 293 cells. Although Dlg1 and PTEN could be readily coimmunoprecipitated (co-IP) from lysates of HEK 293 cells expressing WT Dlg1 and PTEN constructs, mutants of Dlg1 lacking PDZ domains did not co-IP with PTEN; likewise, PTEN mutants lacking the c-terminal domain were unable to co-IP with Dlg1. These results indicate that the PDZ domains of Dlg1 may interact directly with the C-terminus of PTEN (Fig. 3B).
Next, we analyzed the localization of endogenous PTEN and Dlg1 proteins in primary pre-B cells. Previous studies indicated a propensity of Dlg1 to localize to uropod structures in activated lymphocytes (20, 37, 38), as PTEN regulates the PI3K/Akt pathway through lipid phosphatase activity that requires membrane localization to gain access to its substrates (39). We used pre-B cells activated with the chemokine ligand CXCL12, which can induce cell migration and uropod formation (40). We found that both PTEN and Dlg1 colocalized in uropod areas of pre-B cells stimulated with CXCL12 (Fig. 3C). Next, we tested whether endogenous Dlg1 and PTEN proteins interact in B-lineage cells. Endogenous Dlg1 was immunoprecipitated with specific or control antibodies from lysates of Ableson-transformed pre-B cells and immunoblotted with PTEN-specific antibodies (Fig. 3D). These experiments show that Dlg1 can co-IP with PTEN, indicating that endogenous Dlg1 and PTEN can interact in early-stage B-lineage cells.
To determine whether the Dlg1–PTEN interaction is involved in the regulation of PTEN function, we analyzed the expression of PTEN in Dlg1-deficient pre-B cells. Strikingly, we found that levels of PTEN protein are decreased dramatically in Dlg1-deficient cells as compared with control cells (Fig. 3E). In contrast, the levels of PTEN mRNA appeared similar between Dlg1-deficient and control cells (Fig. 3F). These observations suggest a possible role for Dlg1 in stabilization of PTEN protein. To address this, we performed pulse-chase experiments in which cells were pulsed with [35S]methionine for 30 minutes, and then chased for 3 and 6 hours. Cells were lysed and PTEN was immunoprecipitated with an anti-PTEN mAb, and the [35S]methionine-labeled PTEN protein bands were quantified in relation with the time zero chase time (Fig. 3G). These experiments showed that PTEN displays a significantly faster rate of degradation in Dlg1-deficient cells as compared with control cells (Fig. 3G). These data indicate that endogenous Dlg1 is required for stabilization of PTEN protein and predict that, in the absence of Dlg1, diminished levels of PTEN protein in pre-B cells may lead to enhanced PI3K/Akt signaling in response to IL-7. Notably, consistent with the notion that a reduction of PTEN level may be causatively linked to increased growth of pre-B cells in IL-7 cultures, experiments with enforced PTEN expression using lentiviruses encoding eGFP alone, or PTEN and eGFP within a bicistronic vector, showed that a restoration of PTEN expression in knockout pro-B cells reduced their growth in IL-7 cultures to levels similar to control cells (Fig. 3H).
Taken together, these data indicate that the loss of Dlg1 leads to decreased levels of PTEN protein, followed by hyper-activation of Akt, and inactivation of GSK3β, which has been shown to lead to the induction of c-Myc and increased cell cycling.
Dlg1-loss predisposes pre-B cells to oncogenic transformation by BCR/Abl1 in a model of Ph+ pre-B-ALL
Given that Dlg1-deficiency leads to decreased PTEN, enhanced STAT5 activation and c-Myc induction in response to IL-7 stimulation, we hypothesized that it could predispose early-stage B-lineage cells to oncogenic transformation. To test this hypothesis, we first assessed the susceptibility of Dlg1-deficient B lymphoid cells to transformation by BCR/Abl1 in vitro. Bone marrow cells from knockout and WT mice were transduced with BCR/Abl1 retroviruses and cultured under conditions that favor outgrowth of transformed B lymphoid cells (Whitlock–Witte cultures; ref. 41). BCR/Abl1-transformed knockout B lymphoid cells have the same surface phenotype (B220+BP.1+) as BCR/Abl1-transformed and control B lymphoblasts (data not shown). Our analyses of pools of BCR/Abl1-transformed knockout and WT B lymphoblasts showed a substantial increase in proliferation of knockout cells (Fig. 4A). Transduction efficiency of knockout and WT bone marrow cells was equivalent as comparable numbers of pre-B cells from WT and mutant bone marrow were eGFP-positive (data not shown). Thus, Dlg1-deficient B-lymphoid progenitor cells show increased susceptibility to the establishment and/or maintenance of lymphoid transformation by BCR/Abl1 in vitro.
To test the role of Dlg1 in B-cell malignant transformation in vivo, we used a previously established protocol (8, 42) in which B lymphoblasts from knockout and WT mice were transfected with the BCR/Abl1 retroviral expression cassette and subsequently injected i.v. into RAG-2–deficient mice. All injected mice developed leukemia with an initial expansion of immature B-cell blasts evidenced by flow cytometry analyses of peripheral blood and lymphoid organs (Supplementary Fig. S3A and S3B). Flow cytometry analyses of bone marrow, blood, lymph nodes, and spleen from mice transplanted with BCR/Abl1 cells revealed the presence of a distinct population of cells that express the retrovirally encoded eGFP. These cells showed high eGFP intensity and expressed exclusively B-cell markers (B220/CD19/BP.1) confirming that BCR/Abl1 in this in vivo model caused leukemia with a primary lymphoid component (immature B lymphoblasts). Leukemia induced by BCR/Abl1 caused marked enlargement of the spleen and infiltration of peripheral organs in mice transplanted with both knockout and WT B lymphoblasts (Supplementary Fig. S3A–S3C). However, animals reconstituted with BCR/Abl1-transduced knockout B lymphoblasts showed a dramatic increase in disease morbidity rate (Fig. 4B). Thus, morbidity rate was accelerated in mice that received knockout BCR/Abl1 transformants compared with those that received control transformants, even though both cohorts eventually succumbed to B-ALL with the dissemination of eGFP+B220+ lymphocytes in primary and secondary lymphoid organs (Supplementary Fig. S3A–S3C). Our analyses of BCR/ABL–transformed knockout cells showed significantly reduced levels of PTEN protein compared with that in WT cells (Supplementary Fig. S4A and S4B). These results clearly show that loss of Dlg1 in a model of BCR/Abl1-induced precursor B-ALL results in drastic acceleration of disease morbidity.
Dlg1 functions as a tumor suppressor in a p53-loss “sensitized” B-cell model
Previous studies have shown that conditional deletion of p53 in B-lineage cells leads to development of tumors in various stages of B-cell development ranging from the pro-B to mature B-cell stages (30). Thus, deleting p53 in the B-lineage could unmask the oncogenic potential of putative tumor suppressor genes (TSG). We hypothesized that if Dlg1 serves as a TSG in developing B cells, loss of Dlg1 in tumor-prone (i.e., p53-deficient) mice may hasten malignancy. To test this hypothesis, we generated mice harboring p53flox/flox, Dlg1flox/flox, and CD19-Cre+ alleles (CD19-Cre: p53f/f Dlg1f/f or DKO) to permit a simultaneous deletion of both Dlg1 and p53 genes in early B-lineage precursors; cohorts of CD19-Cre: p53f/f Dlg1f/+ (p53 knockout) mice were generated and used as controls (Fig. 5A). As expected on the basis of previously published studies, all mice in both cohorts succumbed to B-cell leukemia (30). Strikingly, however, we found that the cohort of CD19-Cre: p53f/f Dlg1f/+ mice (p53 knockout) had a median age of mortality at 428.5 days, whereas the cohort of DKO mice had a median age of mortality at 345 days (Fig. 5B). These mice were moribund from high tumor burden and had associated leukemia characterized by large B220+ B-cell blasts (Supplementary Fig. S5). All mice from both cohorts developed malignancies that were disseminated in the spleen and lymph nodes (Supplementary Fig. S5 and not shown), which showed effacement and infiltrate of lymphoid cells (Supplementary Fig. S5). No tumors were evident in mice that lacked the expression of Cre recombinase. Importantly, SKY analysis of tumor samples revealed clonal chromosomal rearrangements (Supplementary Fig. S6), whereas i.v. transfer of both P53 knockout and DKO tumor cells into RAG−/− mice resulted in malignancy and death in all of the recipients (data not shown). Taken together, these data reveal that the loss of Dlg1 confers strong predisposition to development of malignancy in a murine model of Ph+ pre-B-ALL.
Detailed morphologic, histologic, and immunophenotyping analyses of tumors derived from p53 knockout and DKO cells are shown in Table 1. Remarkably, these data suggest that DKO tumors may, on average, arise at earlier developmental stages as compared with p53 knockout tumors, as all four of the DKO tumors sampled showed evidence of pre-BCR expression and two thirds of cells lacked mature B-cell markers, as opposed to p53 knockout tumors, in which only one tumor sampled displayed pre-BCR expression and majority of which expressed IgD (Table 1). Further studies are needed to confirm that Dlg1-loss is linked to pre-BCR tumors; however, our results are consistent with the view that a developmental arrest of Dlg1-deficient pre-B cells at the novel C'-1 stage predisposes DKO mice to develop pre-B cell leukemia.
Developing B cells have the delicate task of balancing responses to IL-7. Following IL-7–dependent proliferative burst, successful rearrangement of Ig heavy chain must result in the termination of IL-7 signaling, allowing the proliferating cells to rest and begin rearrangement of the Ig light chain genes. Although this checkpoint is regulated through the pre-BCR complex (16, 17, 43, 44), the loss of pre-BCR signaling components can lead to developmental arrest and pre-B cell leukemia (21–23, 45). In this study, we report a novel requirement for Dlg1 during B-cell development. Using a competitive repopulation assay, we demonstrate that Dlg1-loss leads to an increase in the amount of c-Myc–positive large pre-B cells that seems to act as a “bottleneck” at the C'-1 stage. This result indicates that Dlg1 regulates this newly discovered stage of large pre-B cells. C'-1 cells lack the expression of CD25, a marker of cells that pass the checkpoint at the C/C' (pre-BI/large pre-BII) transition; C'-1 cells rapidly proliferate in response to IL-7 and have high levels of c-Myc, cyclin D3, and pSTAT5 (20).
A recent study revealed that the ability of large pre-B cells to respond to IL-7R stimulation is controlled in a cell-intrinsic manner, likely due to the strength of pre-BCR signaling (20). Blocking signals downstream of the pre-BCR resulted in a substantial increase in proliferation in response to IL-7 while simultaneously blocking differentiation of C'-1 into C'-2 cells. Several recent reports showed that deletion of genes integral in pre-BCR signaling results in hyper-responsiveness to IL-7 (21, 46, 47); deletion of pre-BCR signaling components has been associated with the development of pre-B cell malignancies (45). Thus, it is possible that Dlg1 may interact with or regulate proteins involved in IL-7 or pre-BCR signaling. The guanylate kinase domain of Dlg1 has been shown to bind phosphorylated serine residues on the LGN protein that is involved in cell division (48). Moreover, the pre-BCR signaling protein Btk contains a proline-rich region, which may interact with the SH3 domain of Dlg1. Although the PDZ domains of Dlg1 have the ability to bind the tumor suppressor PTEN in fibroblasts and Schwann cells (6, 7), we demonstrate that in pre-B cells Dlg1 binds and stabilizes PTEN protein, resulting in the inhibition of the Akt/PKB signaling pathway, which has been shown to regulate pre-B cell differentiation (49).
Our data reveal the critical role of Dlg1 in the regulation of STAT5 activation. The importance of the JAK/STAT5 pathway has been demonstrated in Abelson murine leukemia virus (A-Mu-LV)–transformed B-lymphoid cells, as infected cells form malignant tumors that grow independently of growth factors. In this context, the virally encoded v-Abl oncogene directly phosphorylates JAK3 and JAK1, and leads to the activation of STATs, PI3K, and Akt, thus mimicking IL-7R signals that can transform B lineage cells to become growth factor independent (50, 51). In addition to the regulation of proliferation and survival of early-stage B lymphocytes, the JAK–STAT pathway has been implicated in the pathogenesis of human B-ALL, the most common childhood malignancy (50). B-ALL comprises a heterogeneous group of disorders characterized by distinct chromosomal alterations including the t(9;22)(q34;q11) translocation (also known as Philadelphia chromosome, or Ph), which encodes the constitutively active BCR/Abl1 tyrosine kinase that is associated with poor prognosis. The BCR/Abl1 fusion protein leads to constitutive activation of STAT5 and contributes to leukemogenesis by signaling via the JAK–STAT pathway (50, 51). STAT5-deficient mice are completely protected from BCR/Abl1-induced leukemia, illustrating an important role for STAT5 in the initiation of B-ALL in this model (8). In contrast, transfection of a constitutively active form of STAT5 transforms hematopoietic cells in vitro to grow independently of growth factors, and in vivo, results in a large increase in the number of pro-B cells, suggesting that this is a key pathway downstream of the IL-7R (52).
Recent studies have highlighted the importance of leukemia-initiating cells (LIC), which are often early-stage cells. Although these cells are thought to drive the disease, they may not respond to cancer treatments resulting in relapses. A recent study investigating the high-risk B-precursor ALL in humans revealed a significant correlation between the loss of genes involved in early B cell differentiation and disease prognosis (53). We have shown that several of these genes are critical for the regulation of the novel developmental checkpoint at the C'-1 to C'-2 transition (20). In addition, changes in the Wnt/β-Catenin pathway (which regulates c-Myc expression) have been implicated in ALL and chronic myelogenous leukemia (54). In this context, we propose that an abnormal expansion of c-Myc+ C'-1 precursors upon Dlg1-loss creates a reservoir of “pre-leukemic” target cells promoting the development of pre-B cell leukemia. Such cells seem to maintain high levels of c-Myc expression and proliferate at a high rate as they become arrested in transition to the quiescent, C'-2, cell stage.
Drosophila dlg has been shown to function as a tumor suppressor, as its loss results in dramatic neoplastic overgrowth of the imaginal disc epithelia and disruption of cell junctions (3). Here, we used two different murine models of pre-B cell leukemia, a BCR/Abl1-driven model of Ph+ precursor B-ALL, and a p53-loss “sensitized” B-ALL tumor model to demonstrate that the mammalian Dlg1 functions as a TSG in B-lineage cells. Results of our analyses in both models indicate that Dlg1-loss leads to poor outcomes. This is likely due to reduced PTEN levels and increased STAT5 signaling in Dlg1-deficient cells. These results are consistent with the notion that STAT5 is critical for the development of precursor B-ALL. Remarkably, conditionally Dlg1-deficient mice display increased mortality in the p53-sensitized tumor model. We propose that this results directly from the accumulation of preleukemic C'-1 cells upon Dlg1-loss. Consistent with this view, our analyses show clear skewing in the developmental stage of DKO tumors that display markers of pre-B cells, as compared with the p53 knockout tumors that display markers of mature B cells at a higher rate. Thus, during normal development, Dlg1 is required for the regulation of IL-7R signaling outputs, whereas Dlg1-loss results in developmental arrest at a novel C'-1 stage characterized by c-Myc expression, a potential reservoir of cellular targets for oncogenic transformation in precursor B-cell leukemia.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: G.J. Sandoval, D.B. Graham, R.J. Xavier, W. Swat
Development of methodology: G.J. Sandoval, G.B. Gmyrek, K. Fujikawa, S. Srivatsan, W. Swat
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.J. Sandoval, D.B. Graham, G.B. Gmyrek, D. Bhattacharya, S. Srivatsan, A. Kim, K. Yang-Iott, E. Duncavage
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G.J. Sandoval, D.B. Graham, G.B. Gmyrek, K. Fujikawa, A. Kim, A.S. Shaw, E. Duncavage, R.J. Xavier, W. Swat
Writing, review, and/or revision of the manuscript: G.J. Sandoval, D.B. Graham, S. Srivatsan, R.J. Xavier, W. Swat
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.M. Akilesh, K. Fujikawa, A. Kim, K. Yang-Iott, W. Swat
Study supervision: D.B. Graham, C.H. Bassing, W. Swat
Cloned and developed DLG1 knockout mice in W. Swat laboratory: B. Sammut
This work was supported by NIH grants R01AI061077 and R01AI073718 (to W. Swat), DK083756 and DK043351 (to R.J. Xavier), RO1DK058366 and R37-AI57966 (to A.S. Shaw), CA 125195 and CA 136470 (to C.H. Bassing), the Leukemia & Lymphoma Society Scholar Award (to W. Swat), and Leukemia & Lymphoma Society Special Fellowship Award (to D.B. Graham), and the Howard Hughes Medical Institute (to A.S. Shaw).
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.
The authors thank Drs. Barry Sleckman for c-MyceGFP/eGFP mice, and Rick Van Etten for providing p210BCR/Abl expression constructs. The authors also thank Drs. Chyi Hsieh, Jerry Lio, Jeff Bednarski, and Beth Helmink for discussion and experimental help, and Drs. Gene Oltz and Barry Sleckman for discussion and critical reading of the article.
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).
- Received May 30, 2013.
- Revision received September 3, 2013.
- Accepted September 16, 2013.
- ©2013 American Association for Cancer Research.