Ectopic Doublecortin Gene Expression Suppresses the Malignant Phenotype in Glioblastoma Cells

Glioblastomas are the most malignant brain cancers, with a median survival of 10 to 12 months (1). The morbidity and mortality of these brain tumors are severe (2). Based on Affymetrix gene chip analysis in relation to glioma patient survival, the three dominant genes that are associated with patient survival are osteonectin, semaphorin 3B, and doublecortex (doublecortin), all of which regulate cellular motility (3). Osteonectin is a secreted cysteine-rich acidic protein, a component of bone, and plays an important role in development, tissue healing and remodeling, and angiogenesis (4). Osteonectin, also known as secreted protein acidic and rich in cysteine (SPARC), has a suppressive effect on U87 glioma cell proliferation in vitro and on tumor formation in vivo (5). Semaphorin 3B regulates neuronal migration and acts as a tumor a suppressor gene and is found in the 630-kb lung cancer homozygous deletion region on chromosome 3p21.3 (6). Among the three genes, SPARC and semaphorin 3B both act as tumor suppressors, but there is no report on the other gene, doublecortin (DCX), in cancer. We sought to investigate the effect of DCX expression in glioblastoma cell lines. These three genes (SPARC, semaphorin 3B, and DCX) together provide explanatory markers of survival (3). Poorer survival is associated with higher levels of each of these three genes individually. As noted by Rich et al. (3), none of them serves as a useful predictive marker alone, but the concordance of higher values together seems to associate with poorer survival.

Doublecortex is also known as doublecortin (DCX), a microtubule binding protein with two actin-binding domains that regulate neuronal migration and located on chromosome Xq22.3-q23 (7–11). Mutations in the X-linked DCX gene result in lissencephaly, which is a severe brain disorder in newborns and is a result of abnormal neuronal migration (12). The role of DCX in cancer is not known. Deletions of all or part of chromosome 10 are the most common genetic alterations in high-grade gliomas. A tumor suppressor gene, phosphatase and tensin homologue deleted on chromosome 10 (PTEN), maps to chromosome region 10q23 and is mutated by the most common genetic alteration of loss of heterozygosity. PTEN up-regulates DCX at the protein level in subventricular zone precursor cells (13). Coinjection of glioblastoma cells with neural precursors from the subventricular zone, which express nestin and DCX, improves the survival time of old mice to a level similar to that in young mice (14). DCX- and nestin-expressing subventricular zone neural precursors migrate toward the glioma, surround it, decrease the tumor size after glioma cells inoculation, and prolong the survival of the animal (14, 15). More recently, we have shown that DCX expression in DCX-deficient human glioma cell line U87 induces E-cadherin, VE-cadherin, N-cadherin, microtubule-associated protein 2, and nestin, and promotes cell survival from severe serum oxygen glucose deprivation (16). In this report, we provide data showing the direct effect of DCX on suppression of glioma. When DCX is ectopically expressed in U87 glioma cells, there is a marked suppression of the transformed phenotype; the cells manifest a reduced rate of growth in vitro, are arrested in the G₂ phase of the cell cycle, and do not generate tumors in nude rats; and the tumors can be restored by exposure to DCX small interfering RNA (siRNA).

All experimental protocols have been approved by the Henry Ford Hospital Institutional Animal Care and Use Committee. Male nude rats (1–2 months of age) were used. All efforts were made to minimize the number of animals used and their suffering.

Cell cultures. Human glioblastoma U87, A172, and U251N, rat glioblastoma RG2 and 9L, mouse NIH 3T3, and human embryonic kidney 293T cells, obtained from American Type Culture Collection (ATCC; Manassas, VA), were maintained in DMEM supplemented with 5% fetal bovine serum, 2 mmol/L glutamine, 100 units/mL penicillin, and 50 μg/mL streptomycin.

Construction of cytomegalovirus expression vector and stable transfection. Full-length mouse DCX cDNA was subcloned into pcDNA3.1Myc-His(−)A (Invitrogen Corp., San Diego, CA) as previously described (16). Briefly, full-length mouse DCX cDNA from pYX-Asc library (ATCC) was digested with restriction endonucleases EcoRI and KpnI and subcloned into pcDNA3.1Myc-His(−)A (Invitrogen). The orientation of the insert was verified by restriction endonuclease digestion and DNA sequencing by Applied Genomics Technology Center, Wayne State University (Detroit, MI). U87 cells (10⁶) and mouse NIH 3T3 cells (10⁶) were transfected with 10 μg of purified DNA by Lipofectamine 2000 Reagent (Invitrogen) according to the manufacturer's protocol and incubated at 37°C in a humidified incubator for 12 to 16 hours. The medium was removed and incubated for 48 hours with fresh medium for transient transfection. For stable transfection, cells were isolated and cultured routinely as previously described (17). Five highly DCX-expressing U87 clones were pooled on the basis of real-time PCR data (data not shown). The pooled clones were used as DCX-synthesizing U87 and NIH 3T3 cells.

Construction of lentivirus vector, DCX siRNA, and virus infection. DCX-expressing lentivirus vector driven by human cytomegalovirus early gene promoter/enhancer was constructed as previously described (16). DCX siRNA was generated as previously described (16). All recombinant lentiviruses were produced according to the method of Wiznerowicz and Trono (18, 19). Infection of virus was monitored by observing green fluorescent protein expression under a fluorescent illumination microscope (Olympus IX71/IX51, Tokyo, Japan) by infecting the cells with green fluorescent protein–containing lentivirus. DCX expression was confirmed by real-time PCR and Western blot analysis (Figs. 1 and 2).

Real-time semiquantitative PCR and cell proliferation assays. The cDNA synthesis and real-time PCR were done as previously described (18). DNA primers CTTTTGGTTCAGCAGAAGGG-sense, CAAATGTTCTGGGAGGCACT-antisense; TTTCCAGATTCAATCAGCCC-sense, AAAGATCTGCTGAGGGGGAT-antisense; and ACCGCCAAATTTAATTGCAG-sense, TTCGTCCCTTTCCAGCTTTA-antisense for rat/mouse DCX, human DCX, and human PTEN, respectively, were used for real-time PCR. Dissociation curves and agarose gel electrophoresis were used to verify the quality of the PCR products. Each sample was tested in triplicate using quantitative real-time PCR. All values were normalized to β-actin. Values obtained from five independent experiments were analyzed relative to gene expression data using the 2^−ΔΔCT method (20). The CellTiter 96 Aqueous Non-Radioactive Cell Proliferation Assay (Promega Corp., Madison, WI) was used to determine the number of viable cells in the proliferative phase as previously described (17).

Fluorescence-activated cell sorting analysis, soft agar colony formation assay, and implantation of tumor cells in nude rats. Fluorescence-activated cell sorting (FACS) analysis was done by dual labeling with bromodeoxyuridine (BrdUrd) and propidium iodide according to the method of Agami and Bernards (21). In vitro tumorigenicity was tested by growth in a soft agar colony assay as previously described (17). Tumor cells were implanted into nude rats (5–6 weeks old) according to the method of Zhang et al. (15).

Immunoprecipitation, Western blot analysis, and alkaline phosphatase treatment. Immunoprecipitation and Western blot analysis were done as previously described (22). For alkaline phosphatase treatment, cells were lysed with ice-cold cell lysis buffer for Western blot, without phosphatase inhibitors such as 1 mmol/L NaF, 1 mmol/L Na₃VO₄, at 4°C. The cell lysates were incubated at 37°C for 5 to 30 minutes with calf intestinal alkaline phosphatase (1 units/μg of protein; Promega). For treatment with specific inhibitors for c-jun NH₂-terminal kinase (JNK), mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase kinase (MEK), and p44MAPK, the cells were incubated for 3 hours with JNK inhibitor II, AG126, and PD 98059 (Calbiochem, San Diego, CA). Then, the cells were lysed with lysis buffer and analyzed by Western blot and immunoprecipitation. Proteins were transferred onto a nitrocellulose membrane and incubated with goat antiserum for DCX (1:1,000; Santa Cruz Biotechnology, Santa Cruz, CA) as primary antibody and with donkey anti-goat horseradish peroxidase antibody (Jackson ImmunoResearch Labs, West Grove, PA) as secondary antibody. Rabbit anti–spinophilin/neurabin-II polyclonal antibody (1:1,000; Upstate, Charlottesville, VA), mouse monoclonal PTEN (1:1,000; Sigma, St. Louis, MO), mouse monoclonal anti-phospho-tyrosine (1:1,000; Sigma), and mouse monoclonal β-actin (1:5,000; Santa Cruz Biotechnology) were used as primary antibodies, and antirabbit and antimouse horseradish peroxidase (1:10,000; Jackson ImmunoResearch Labs) as secondary antibodies. Mouse monoclonal epidermal growth factor receptor (EGFR) antibody (1:1,000; Santa Cruz Biotechnology) was used for immunoprecipitation.

Statistical analysis. One-way ANOVA followed by Student-Newman-Keuls test was used. The values were the mean of 5 to 10 independent experiments for real-time PCR data and three independent experiments for Western blot analysis. The data are presented as mean ± SD. P < 0.05 is considered as significant.

DCX expression in glioma and noncancer cells. DCX is expressed in migrating neurons throughout the central and peripheral nervous system during embryonic and postnatal development (23). DCX is up-regulated in subventricular zone cells after severe ischemic insult, in vivo and in vitro (16). Here we examined DCX expression in different glioma cell lines, such as human A172, U87, and U251N and rat RG2 and 9L, and noncancer primary culture, subventricular zone cells, and different cell lines, NIH 3T3 and 293T. There were no detectable DCX mRNA and protein found in the cancer and noncancer cell lines (Fig. 1A and B). In contrast, subventricular zone cells expressed DCX at the mRNA and protein levels (Fig. 1A and B). DCX-transfected U87 cells and subventricular zone cells expressed both phosphorylated DCX and nonphosphorylated DCX (Fig. 1B). In contrast, DCX-transfected NIH 3T3 expressed only nonphosphorylated DCX (Fig. 1B). These data indicate that U87 and subventricular zone cells have MAPKs that phosphorylated DCX, and these MAPKs were not activated in NIH 3T3 cells.

Phosphorylation of DCX in glioma and noncancer cells. To determine cell specificity of phosphorylation of DCX, we infected additional glioma cells such as A172, U251N, 9L, and RG2 and noncancer cells 293T with lentivirus carrying DCX (Fig. 1C and D). Alkaline phosphatase treatment for 30 minutes of cell lysates from DCX-infected cells and subventricular zone cells altered the migration pattern of DCX protein bands in Western blot analysis (Fig. 1C). Alkaline phosphatase digestion of upper band of DCX represented the phosphorylated band of DCX (Fig. 1C). All glioma cells infected with DCX showed double bands of DCX as phosphorylated DCX (upper) and nonphosphorylated DCX (lower) in Western blot analysis (Fig. 1C). In contrast, noncancer cells infected with DCX showed a major protein band as nonphosphorylated DCX (lower) and very faint phosphorylated DCX (upper; Fig. 1D,, bottom). The magnitude of phosphorylation of DCX was remarkably high in all glioma cells compared with noncancer cells based on quantitative histogram analysis (Fig. 1D , top). DCX is a substrate of JNK and interacts with both JNK and JNK-interacting protein (24). Eighty-six percent of primary human glial tumors show activation of almost exclusively the M_r 55,000 isoforms of JNK (25). Aside from EGFR, this is the only other kinase that has been shown to be basally active in glioma (25). These data suggest that activation of JNK in glioma is involved in DCX phosphorylation.

Effect of DCX on cell proliferation. We overexpressed DCX in DCX-deficient glioma cell lines such as A172, U87, U251N, RG2, and 9L and noncancer 293T with DCX lentivirus, and in U87 and NIH 3T3 cells with DCX stable transfection (Fig. 1A–D). The tumor suppressor gene PTEN, which is missing in these glioma cells due to loss of heterozygosity of chromosome region 10q23, up-regulates DCX at the protein level in brain (13). We studied the effect of DCX on proliferation of glioma cells. To knock down DCX expression in DCX-synthesizing cells, the most reduced DCX knockout cells were selected for proliferation assay from these cells infected with DCX siRNA for several times and analyzed by quantitative real-time PCR and Western blot and monitored during the time course of proliferation assay (data not shown). Proliferation assay was done everyday for 5 days. After proliferation assay, at day 6 the cells were analyzed again by Western blot to examine DCX expression in all samples including DCX siRNA sample (Fig. 2). All DCX-synthesizing glioma cells such as U87, U251N, A172, 9L, and RG2 exhibited significantly suppressed growth, which was reversed on DCX siRNA treatment (Fig. 2). In contrast, DCX had no effect on proliferation of noncancer cells NIH 3T3 (Fig. 2).

DCX regulates cell cycle progression. To examine the mechanism of growth suppression of glioma cells by DCX, we analyzed DNA synthesis. The growth state of the cells was tested with FACS by dual labeling with BrdUrd and propidium iodide. The control cells actively synthesized DNA, whereas DCX-synthesizing glioma cells such as U87, A172, U251N, 9L, and RG2 were blocked in the G₂-M phase of the cell cycle (Fig. 3A). Abrogating DCX gene expression with DCX siRNA treatment specifically reversed this block (Fig. 3A). These data show the specific and reversible effect of DCX in suppressing the growth of glioma cells.

In vitro anchorage-independent growth of human glioma. Anchorage-independent growth in semisolid medium is a useful variable for evaluating the malignant potential of a given cell population (reviewed in ref. 26). We tested control and DCX-transfected U87 cells for their ability to grow in soft agar. The DCX stably expressing U87 cells generated only small colonies (Fig. 3B) when compared with mock-transfected control cells (Fig. 3B). The maximal size of the DCX-transfected colonies was on the order of ∼10-fold less than the control colonies based on quantitative histogram analysis of areas occupied by the colonies (Fig. 3B). When the DCX-transfected U87 cells were infected with DCX siRNA, these cells produced almost the same size of colonies as controls in soft agar (Fig. 3B). Quantitative histogram data showed that the inhibitory effect of DCX on colony formation in soft agar was reversed by DCX siRNA up to ∼9-fold. These results were corroborated by growth suppression of DCX-synthesizing glioma cells such as A172, U87, U251N, RG2, and 9L (Fig. 2).

In vivo tumorigenicity of DCX-transfected U87 glioma cells. To evaluate the malignant potential of glioma cells, we measure the formation of tumor xenograft in immunocompromised nude rats. When DCX-transfected U87 cells were implanted into the brain of nude rat, no tumors were detected after 3 weeks (Fig. 3). In contrast, the mock-transfected U87 cells generated tumor xenograft (10 of 10 rats) at 3 weeks after U87 cell implantation (Fig. 3). These results corroborate the anchorage-dependent and anchorage-independent growth kinetics. The inhibitory effect of DCX on in vivo tumorigenicity was reversed by DCX siRNA treatment. The tumor xenograft (1 of 5 rats) was found at 3 weeks after inoculation of DCX-synthesizing U87 cells infected with DCX siRNA (Fig. 3).

Interaction of phosphorylated DCX with spinophilin/neurabin II. Phosphorylation on serine/threonine or tyrosine residues of proteins plays a major role to control protein function in eukaryotic cells and cancer cells. However, the relative activities of both kinases and phosphatases regulate the steady-state level of protein phosphorylation. Dephosphorylation of the proteins by phosphatases can be as effective as phosphorylation. As an example, phosphorylated retinoblastoma susceptibility gene pRb is inactive with respect to growth suppression, but dephosphorylated pRb is a very active growth suppressor and regulates cell cycle progression after removal of phosphates (reviewed in ref. 27). Here, we examine the interaction of phosphorylated DCX with spinophilin/neurabin II, a type 1 protein-phosphatase-binding protein (28). Immunoprecipitation of the total protein extracts from DCX-transfected U87 cells with neurabin II and DCX antibodies and Western blot analysis with DCX and neurabin II antibodies showed that the phosphorylated DCX bands had significantly increased binding ability with neurabin II (Fig. 3D). In contrast, cell lists from DCX-transfected NIH 3T3 contained the majority of dephosphorylated DCX; immunoprecipitation showed no bands in five independent Western blot analyses (Fig. 3D). Western blot analysis of immunoprecipitation of alkaline phosphatase–treated cell lysates from DCX-transfected U87 with both neurabin II- and DCX-specific antibodies showed no signal in five independent experiments (Fig. 3D). These data show that phosphorylated DCX binds specifically to neurabin II in DCX-transfected U87 cells.

Involvement of JNK in DCX phosphorylation. In Western blot analysis, alkaline phosphatase treatment of cell lysates from DCX-transfected U87 cells altered the migration pattern of DCX protein bands by digesting the upper band several times, indicating that DCX was phosphorylated (Fig. 4A). The total cell lysates from DCX-synthesizing U87 cells were immunoprecipitated with EGFR antibodies as a positive control (22) and with DCX as an experimental sample, and subjected to Western blotting with anti-phospho-tyrosine antibodies. The presence of very strong bands for positive control EGFR, heavy and light chains of immunoglobulin G, but no bands in the range of molecular size of DCX (∼49 kDa) in the blot (Fig. 4B) shows that DCX was not phosphorylated on tyrosine.

The specific inhibitors for JNK, MEK, and p44MAPK were used to investigate the involvement of JNK in the phosphorylation of DCX. The specific inhibitor for JNK (JNK inhibitor II) reduces the phosphorylation on DCX in a dose-dependent manner (Fig. 4C). In contrast, the specific inhibitors for MEK (AG126) and p44MAPK (PD 98059) have no effect on the phosphorylation on DCX (Fig. 4C). These data indicate that DCX is phosphorylated by JNK, but not by MEK and p44MAPK. Neurabin II interacted with phosphorylated DCX in a dose-dependent manner when these were subjected to JNK inhibitors (Fig. 4C). To augment these data, reciprocal experiments were also done with constitutively active JNK (a fusion of MKK7 and JNK1; ref. 29). The expression vectors of the fusion of MKK7 and JNK1 (α and β) provided by Dr. L.E. Heasley (University of Colorado Health Sciences Center, Denver, CO) were transiently transfected into DCX stably transfected U87 cells and into U87 cells. Analysis of phosphorylation of DCX in Western blot shows that JNK up-regulates phosphorylation of DCX (Fig. 4D). No DCX bands were visible in MKK7-JNK1 (α and β)–transfected U87 cells. Immunoprecipitation of total cell lysates with DCX and neurabin II antibodies and blotting with neurabin II and DCX antibodies showed that only phosphorylated DCX were associated with neurabin II. These data confirm the involvement of JNK in DCX phosphorylation.

Ectopic expression of PTEN induces DCX synthesis in U87 glioma cells. Overexpression of PTEN in U87 cells transfected with PTEN expression vector (30), provided by Dr. A.H. Ross (Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, MA), induced DCX expression at mRNA and protein levels in PTEN-synthesizing U87 cells (Fig. 5A). Treatment with DCX siRNA significantly reduced the growth suppression and antitumorigenic effect of DCX in PTEN-expressing U87 cells (Fig. 5C and D). These data suggest that DCX is also involved in the mechanism of tumor suppression mediated by PTEN. Thus, DCX may play an important role in antitumorigenesis.

Association of phosphorylated DCX with neurabin II influences tumor suppression. The site of association of DCX with neurabin II and protein phosphatase 1 (PP1) was recently identified by Shmueli et al. (31) within the conserved KVRF motif of DCX by mutating one amino acid (F57L), and in JNK sites (T331, S334) in DCX, identified by Gdalyahu et al. (24). To study the importance of interaction and phosphorylation of DCX in glioma suppression, the mutated DCX expression vector (DCX-F57L) and three site-directed DCX mutants (24) such as T331A, S334E, and T331A + S334A for both amino acid substitution, provided by Dr. O. Reiner (Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel), were transiently transfected into U87 cells. These DCX mutants failed to interact with neurabin II, as analyzed by immunoprecipitation and Western blot with both DCX and neurabin II antibodies (Fig. 6A), and showed significantly reduced ability for growth suppression of U87 cells in proliferation assay and anchorage-independent soft agar culture assay (Fig. 6B–D). However, the mutant DCX-F57L remained phosphorylated (Fig. 6A). These data are consistent with those of Shmueli et al. (31). These data show that the conserved KVRF motif and JNK sites in DCX are responsible for association of neurabin II and PP1 with DCX. Thus, both phosphorylation and dephosphorylation of DCX in JNK sites may be important in DCX-mediated suppression of glioma. These data also show that interaction between DCX and neurabin II is dependent on DCX phosphorylation in either site of JNK for tumor suppression.

Proliferation assay and anchorage-independent soft agar culture assay of both the DCX-overexpressing and nonexpressing glioma cells treated with siRNA to neurabin II (Santa Cruz Biotechnology) and the DCX interacting coiled coil domain of neurabin II (ref. 24; provided by Dr. O. Reiner) were also done (Fig. 6A). Treatment with siRNA to neurabin II and the DCX interacting coiled coil neurabin II significantly reversed the DCX-mediated tumor suppression effect on cell proliferation assay and anchorage-independent colony formation assay in soft agar culture in U87 cells (Fig. 6B–D). Human embryonic kidney cells HEK 293 are a neurabin II null cell line (31). Deficiency of neurabin II expression in HEK 293 cells is also confirmed in Western blot analysis (Fig. 6A). Overexpression of DCX in HEK 293 cells has no effect on the cell proliferation and on the anchorage-independent colony formation in soft agar culture (Fig. 6B–D). These data suggest that neurabin II is important in the suppression of anchorage-independent growth proliferation of glioma, and loss of neurabin II function causes anchorage-independent growth, and DCX overexpression can no longer suppress this effect.

The results of this study imply that DCX regulates the growth and maintenance of the malignant phenotype in glioma. De novo expression of DCX, a gene that is totally suppressed in wild-type cells, leads to an arrest in the G₂ phase of the cell cycle. During this process, the neoplastic cells cease to proliferate and lose their ability to form large colonies in semisolid medium and to induce tumor xenograft in immunocompromised hosts. DCX siRNA treatment reverses the antitumorigenic effect of DCX. These data point toward a specific and direct effect of DCX as a tumor suppressor gene. This growth suppression can be reversibly abrogated by blocking DCX transcription via DCX siRNA, a process that leads to a complete restoration of the overall growth abilities of the transfected clones. Fundamental questions are how this microtubule-associated protein modifies the growth rate and tumorigenic potential of transformed cells, and whether this growth suppression is also operational in normal cells. Interestingly, Affymetrix gene chip analysis in relation to glioma patient survival detects three separate genes (3). Two of them, osteonectin (SPARC) and semaphorin 3B, are antitumorigenic and tumor suppressor genes, respectively (6), but there is no report of involvement of DCX in glioma. SPARC is a collagen and fibronectin binding protein (32). Another collagen and fibronectin binding leucine-rich protein, decorin, a chondroitin sulfate proteoglycan, shows similar properties to SPARC. Decorin is an antioncogenic and antitumorigenic gene (17, 22, 33, 34). Decorin promotes glioma survival in vitro via the phosphatidylinositol 3-kinase/AKT pathway, as previously reported (18), and glioma survival in vivo in rats as reported by others (35). DCX also enhances glioma survival under conditions of oxygen and glucose deprivation in vitro (18). Here, we show that DCX shows a pronounced antitumorigenic effect on glioma in vitro as well as in vivo. DCX is highly phosphorylated by JNK in glioma cells. The phosphorylated DCX interacts with spinophilin/neurabin II. Neurabin II inhibits anchorage-independent growth of human and mouse cancer cell lines, regardless of p53 and ARF status (28). Ectopic expression of p14ARF (ARF) is ineffective for suppression of colony formation in Saos-2 cells whereas neurabin II and ARF coexpression and interaction synergistically inhibits anchorage-independent growth (28). Our data also show that DCX and neurabin II interaction inhibits proliferation and anchorage-independent growth in glioma cells. In contrast, DCX-mediated growth suppression is lost in neurabin II null HEK 293 cells and is reversed by knocking down neurabin II with siRNA. DCX and neurabin II interaction may contribute to strong antitumorigenicity to glioma.

PTEN gene, also known as mutated in multiple advanced cancers (MMAC1) and transforming growth factor-β-regulated and epithelial cell-enriched phosphatase (TEP-1), is a tumor suppressor gene located at chromosome 10q23.3 (36, 37). Loss of heterozygosity on the long arm of chromosome 10 is found in 75% to 90% of high-grade gliomas (38, 39). Here we show that DCX-deficient glioma cells expressed DCX when they were forced to overexpress PTEN by DNA transfection. These data are also consistent with up-regulation of DCX by PTEN in subventricular zone precursor cells (13). These data suggest that DCX deficiency in glioma cells may be because of lack of PTEN. Overexpression of wild-type PTEN into PTEN-deficient glioma cells causes in vitro and in vivo growth suppression (40, 41). Here we show that DCX siRNA treatment significantly reduced the growth suppression effect in PTEN-overexpressing glioma cells. These data indicate that involvement of DCX in PTEN-mediated tumor suppression is a novel mechanism.

Regulation of many cellular functions depends on protein phosphorylation and dephosphorylation of multiple substrates by protein kinases and phosphatases (reviewed in ref. 42). Phosphorylation and dephosphorylation lead to association and dissociation of many proteins such as a phosphoprotein encoded by mouse α4 and the catalytic subunit of protein phosphatase 2A that regulates their activation (43). Neurabin II belongs to this class of regulators because it negatively regulates the PP1 catalytic subunit activity. Native neurabin II associates with PP1 (44). DCX, neurabin II, and PP1 are found in the same protein complex when PP1 is pulled down with PP1-specific microcystin-agarose beads and also when DCX is immunoprecipitated from mouse brain extracts (31). Our data show that JNK inhibitors and JNK site-directed mutagenesis (T331, S334) in DCX reduced the interaction between DCX and neurabin II in glioma cells and JNK activation by MKK7-JNK1, and increased DCX phosphorylation and its interaction with neurabin II in glioma cells. Phosphatase inhibitors reduce the interaction between DCX and neurabin II and also inhibit DCX-neurabin II-PP1 complex formation (31). Thus, phosphorylation and dephosphorylation both may be required for interaction between DCX and neurabin II and DCX-neurabin II-PP1 complex formation. Interaction between phosphorylated DCX and neurabin II induces association of DCX, neurabin II, and PP1 in vivo (31). PP1, one of the key eukaryotic serine/threonine protein phosphatases, is involved in the mitotic dephosphorylation of pRb (reviewed in ref. 27), as well as in the dephosphorylation of specific residues of p53 (reviewed in ref. 45), and regulates the control of cell cycle progression (27, 45). PP1 dephosphorylates DCX specifically on amino acid residues S331 and T334 both in vitro and in vivo (31). Microinjection of PP1-neutralizing antibodies (46) and PP1 inhibitors such as okadaic acid (46, 47) blocks mitosis and alters the progression of the cell cycle by accumulating at the nucleus to associate with chromatin during G₂ and M phases (46). Our data showed that DCX overexpression blocked the G₂-M phase of the cell cycle in glioma cells. DCX-mediated growth arrest in the G₂-M phase of the cell cycle in glioma cells may be through inactivation of PP1 by neurabin II-DCX interaction. The expression of DCX and neurabin II is dynamic, and they are coexpressed in migrating neurons (31). Overexpressing the coiled-coil domain of neurabin II leads to interaction with DCX, recruits the endogenous neurabin II with PP1, and induces dephosphorylation of DCX on one of the JNK phosphorylated sites (31). In vitro, DCX is site-specifically dephosphorylated by PP1 without the presence of neurabin II (31). Overexpression of phosphorylated DCX, therefore, itself may competitively inhibit PP1 and block G₂-M phase of cell cycle progression.

In summary, we showed that DCX is a growth suppressor of glioma. Our studies also suggest that the absence of the DCX gene may increase propensity to develop tumors; thus, boosting endogenous DCX gene expression may be therapeutically beneficial and, alternatively, pharmacologic delivery of recombinant DCX may have beneficial therapeutic effects in the treatment of glioma.

Grant support: NIH National Institute of Neurological Disorders and Stroke grant PO1 NS23393.

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 Drs. D. Trono, A. Ross, L.E. Heasley, and O. Reiner for providing us with lentivirus vectors, PTEN expression vector, MKK7-JNK1 expression vectors, and DCX mutants.