Potential conflict of interest: Nothing to report.
This work was supported by The Hong Kong Research Grants Council (HKU763509 and HKU7/CRF/09), The University of Hong Kong (Small Project Funding 201007176079; to W-L.L. and Y-P.C.), and The S.K. Yee Medical Research Fund 2011 (to D-Y.J. and Y-P.C.). I.O-L.N. is a Loke Yew Professor in Pathology.
Deregulation of cellular-signaling pathways by the inactivation of tumor-suppressor genes is one of the major causes of hepatocellular carcinoma (HCC). In this study, we identified Tax1 binding protein 2 (TAX1BP2) as a novel tumor-suppressor gene in HCC. TAX1BP2 transcript was frequently underexpressed (42.2% with T/NT <0.5; P < 0.03) in HCCs, and underexpression of TAX1BP2 was associated with poorer overall survival rates in patients after surgical resection. An effector domain (ED) for TAX1BP2 tumor-suppressor activity was mapped to the amino-acid residues 267-756. Transient or stable expression of either full-length or ED of TAX1BP2 significantly suppressed HCC cell tumorigenicity through the activation of the p38/p53/p21 pathway. In contrast, silencing of TAX1BP2 by short interfering RNA remarkably suppressed the activation of the p38/p53/p21 pathway. Finally, phosphorylation of TAX1BP2 at serine-763 by cyclin-dependent kinase (CDK)2 abolished the TAX1BP2-mediated p38 activation and tumor-suppressive activity, indicating that TAX1BP2 can adapt CDK2 signaling to the p38/p53/p21 pathway. Conclusion: Taken together, our data provide the first evidence that TAX1BP2 is a CDK2-regulated tumor-suppressor gene in HCC and is a novel activator of the p38/p53/p21 pathway. (HEPATOLOGY 2012;56:1770–1781)
Hepatocellular carcinoma (HCC) is the most-common form of primary hepatic tumor and one of the most-common malignancies worldwide.1 Activation of oncogenes or inactivation of tumor-suppressor genes is a common cause for hepatocarcinogenesis.2
The centrosome is the microtubule-organizing center, coordinating the microtubule network critical for cell shape, polarity, adhesion, motility, intracellular transport, organelle positioning, chromosome segregation, and cytokinesis.3 Deregulation of centrosome duplication leads to centrosome supernumerary, which is highly associated with aneuploidy, genomic instability, and tumorigenesis.4 Our previous study has demonstrated that a centrosomal protein, Tax1 binding protein 2 (TAX1BP2 or TAXBP121), binds with Tax, which is an oncogenic transcriptional activator of human T-cell leukemia virus type I (HTLV-1).5 TAX1BP2 acts as an intrinsic block to centrosome overduplication, and inactivation of TAX1BP2 by Tax may be critical for HTLV-1-mediated aneuploidy and carcinogenesis. TAX1BP2 is a coiled-coil protein and has a significant sequence homology to another centrosomal protein, centrosomal Nek-2 associated protein 1, which is involved in centrosome cohesion.6 However, other functions of TAX1BP2 are still poorly understood. In this study, the role of TAX1BP2 in hepatocarcinogenesis was explored. Our data demonstrated that TAX1BP2 was significantly down-regulated in human HCCs. Stable overexpression of TAX1BP2 or its effector domain (ED) exerted the growth/tumor-suppressive effect by activation of the p38/p53/p21-signaling pathway, whereas silencing of TAX1BP2 expression reversed the tumor suppression. In addition, we demonstrated that TAX1BP2 tumor-suppressor activity could be modulated by cyclin-dependent kinase (CDK)2 phosphorylation. Taken together, our data provide the first evidence that TAX1BP2 is a putative tumor suppressor in HCC, and that loss of TAX1BP2 may preclude the activation of the p38/p53/p21 pathway.
ATP, adenosine triphosphate; BrdU, bromodeoxyuridine; CDK, cyclin-dependent kinase; CFA, colony formation assay; ED, effector domain; FL, full length; GFP, green fluorescent protein; HA, hemagglutination; HCC, hepatocellular carcinoma; HTLV-1, human T cell leukemia virus type-1; HU, hydroxyurea; IF, immunofluorescence; LOH, loss of heterozygosity; MAPK, mitogen-activated protein kinase; MKK, MAPK kinase; p-TAX1BP2 (S763), phospho-TAX1BP2 (Serine-763); qRT-PCR, quantitative reverse-transcription polymerase chain reaction; siRNA, short interfering RNA; TAX1BP2, Tax1 binding protein 2.
Materials and Methods
Antibodies and Chemicals.
Rabbit anti-phospho-TAX1BP2 (S763) antibody was raised against a synthetic peptide, PRPVPGpSPARDAP (aa757-769) (Applied Biological Materials Inc., Richmond, British Columbia, Canada). Rabbit polyclonal anti-TAX1BP2b antiserum was raised using a TAX1BP2 (1-815) polypeptide (Genscript USA Inc., Piscataway, NJ). Other antibodies are listed in Supporting Table 2. Cisplatin and etoposide were from Merck KGaA (Darmstadt, Germany). All chemicals were from Sigma-Aldrich (St. Louis, MO), unless otherwise stated.
Centrosome Duplication Assay.
HeLa cells were seeded on coverslips and treated with 2 mM of hydroxyurea (HU) for 24 hours. Cells were then transfected with constructs of green fluorescent protein (GFP)-tagged full length (FL) or ED. After 6 hours, cells were further treated with 2 mM of HU for 42 hours. The number of centrosomes in at least 30 transfected cells was counted. Three independent experiments were performed for each transfection.
In Vitro Kinase Assay.
Purification of bacterially expressed His-tagged proteins and in vitro CDK2 assay were performed as described previously.5 Briefly, 2 μg of purified TAX1BP2 protein was incubated with reconstituted CDK2/cyclin A complex in the presence of 1 μCi/mL of γ-[32P]ATP (adenosine triphosphate) at 30°C for 30 minutes. After electrophoresis, the gel was stained by Coomassie blue and autoradiography was performed.
Human HCC Samples.
Paired samples of primary HCCs and the corresponding nontumorous liver tissues from 45 Chinese patients were collected at the time of surgical resection at The University of Hong Kong, Queen Mary Hospital (Pokfulam, Hong Kong) from 1991 to 2000. All specimens were snap-frozen in liquid nitrogen and kept at −80°C immediately after surgical resection. After resection, all patients were followed up monthly in the first year and quarterly thereafter. Actuarial survival was measured from the date of hepatic resection to the date of death or last follow-up.
Clinicopathological Correlation, Survival, and Statistical Analyses.
Clinicopathologic features of patients, including gender, tumor size, number of tumor nodules, cellular differentiation by Edmondson grading, presence of tumor encapsulation, tumor microsatellite formation, venous invasion without differentiation into portal or hepatic venules, direct liver invasion, background liver disease in the nontumorous liver tissues, serum hepatitis B surface antigen status, and tumor staging, were analyzed using SPSS 18 software (SPSS, Inc., Chicago, IL). Continuous data between groups were compared using the Student t test, and categorical variables were compared using the chi-square or Fisher's exact test, as appropriate. Overall survival was analyzed by Kaplan-Meier's method, and results were compared by log-rank test. Tests were considered significant when P < 0.05.
Underexpression of TAX1BP2 Was Frequent in HCC Patient Samples and HCC Cell Lines.
To examine the role of TAX1BP2 in HCC, TAX1BP2 transcript in paired samples of HCC and the corresponding nontumorous liver tissues were quantified by quantitative reverse-transcription polymerase chain reaction (qRT-PCR). After normalization with β-actin, 42.2% (19 of 45 cases) of the HCC samples had a lower expression of TAX1BP2 (T/NT <0.5), compared to the corresponding nontumorous tissues. The TAX1BP2 expression in tumor was significantly lower than that in the nontumorous tissues (P < 0.03; Fig. 1A). Protein expression level of TAX1BP2 correlated well with the transcript level in 12 representative HCC sample pairs (Fig. 1B). To elucidate whether the underexpression of TAX1BP2 was the result of allelic loss, we performed loss of heterozygosity (LOH) analysis. Using three microsatellite markers (D1S2697, D1S170, and D1S2644), flanking different regions of the Tax1bp2 gene, LOH in HCC samples was determined to be 28.5% (12 of 42 detectable cases; 3 noninformative cases). Upon clinicopathological correlation, underexpression of TAX1BP2 in HCC had no significant association with the parameters analyzed (Supporting Table 1). As for survival analysis, overall survival rates of TAX1BP2-underexpression patients were 100.0%, 84.8% and 63.6% months at 1, 3 and 5 years, respectively. Patients whose tumors had TAX1BP2 underexpression had significantly shorter overall survival rates, as compared with those whose tumors had no TAX1BP2 underexpression (median, 70.2 and 116.1 months, respectively; P = 0.040, log-rank test) (Fig. 1C).
To examine the expression of TAX1BP2 in HCC cell lines, endogenous protein expression of TAX1BP2 was examined in the nontumorigenic, immortalized healthy liver cell line, MIHA, and seven HCC cell lines (Fig. 1D). Compared with MIHA, five of the seven HCC cell lines had a lower TAX1BP2 protein level (<70% of MIHA). None of the HCC cell lines had a higher TAX1BP2 expression than MIHA, indicating that TAX1BP2 is frequently down-regulated in HCC cells. Because of high transfection efficiency, SMMC-7721, HepG2, Hep3B, and PLC/PRF/5 cells were chosen for the subsequent study.
TAX1BP2 Suppressed Colony Formation of HCC Cells and Its Growth/Tumor-Suppressive ED Was Mapped to the Central Region.
To examine whether TAX1BP2 possesses growth- or tumor-suppressive activity, plasmid containing FL complementary DNA of TAX1BP2 was transfected into HCC cells in colony-formation assay (CFA). Ectopic expression of TAX1BP2 FL significantly suppressed the colony-formation ability of SMMC-7721 (Fig. 2B) and HepG2 cells (Supporting Fig. 1A). To map the tumor-suppressive ED of TAX1BP2, a series of truncation mutants were generated (Fig. 2A). Expression of 1-941 and 1-817, but not 1-539, 1-266, 757-1039, and 1040-1313 mutants, significantly suppressed colony formation in SMMC-7721 cells (Fig. 2B). Surprisingly, the 757-1039 mutant enhanced colony formation. Similar results were obtained using HepG2 cells (Supporting Fig. 1A). Hence, the ED was mapped to the region of 267-756 (Fig. 2A). To further confirm the tumor-suppressive activity of ED, it was subcloned into an expression vector. Ectopic expression of ED significantly suppressed the colony formation of SMMC-7721 (Fig. 2C) and HepG2 cells (Supporting Fig. 1A).
Similarly to FL, ED also significantly inhibited the centrosome overduplication assayed by centrosome duplication assay (Fig. 2D). Whereas more than 65% of cells with vector expression had supernumerary centrosomes, remarkably, less cells with FL and ED expression (only 31.6% and 48.2%, respectively) had more than two centrosomes, indicating that the expression of FL and ED inhibited centrosome overduplication. Therefore, the central region of TAX1BP2 carries both tumor and centrosome overduplication suppressive activities.
Centrosome Targeting Was Essential for Tumor-Suppressive Activity of TAX1BP2.
To determine the importance of centrosome-targeting property for the tumor-suppressive activity of TAX1BP2, centrosome localization of FL, ED, and all truncation mutants was examined by immunofluorescence (IF) staining with reference to the centrosomal marker, γ-tubulin. FL, ED (Fig. 2E), and mutants of 1-941, 1-817, and 1040-1313 (Supporting Fig. 1B, 2, 3, and 7) showed an intense colocalization signal with γ-tubulin, whereas the 757-1039 mutant only showed a weak colocalization signal (Supporting Fig. 1B, 6). However, cells transfected with GFP or mutants of 1-539 and 1-266 were uniformly fluorescent (Supporting Fig. 1B, 4 and 5), and no colocalization with γ-tubulin was observed. Hence, we confirmed a second centrosome-targeting domain at 540-817 (Supporting Fig. 1B, 8), overlapping with ED, in addition to the C-terminal centrosome-targeting domain found in our previous study.5 This result suggested that the centrosome localization of TAX1BP2 may be important for its tumor-suppressive activity.
TAX1BP2 Suppressed HCC Cell Proliferation and Tumorigenicity.
To evaluate whether TAX1BP2 can suppress HCC cell proliferation, stable FL and ED of TAX1BP2 expression clones in SMMC-7721 cells were established (Fig. 3A). In the cell-proliferation assay, all stable FL and ED expression clones had a significantly slower growth rate than the stable vector control clones (Fig. 3B). The doubling time increased from 25.1 ± 0.1 (GFP-C1)/25.2 ± 0.1 (GFP-C2) hours in the stable vector control clones to 29.2 ± 0.2 (FL-C1)/29.2 ± 0.2 (FL-C2) hours in the FL clones and to 38.5 ± 0.4 (ED-C1)/43.8 ± 0.1 (ED-C2) hours in the ED clones. To test whether the proliferation suppression was the result of a blockage in G1/S progression, a bromodeoxyuridine (BrdU)-labeling assay was performed. All the stable FL and ED clones had significantly slower rates of BrdU incorporation than the stable vector clones (Fig. 3C). To further assess the tumorigenicity of stable TAX1BP2-expressing clones in vivo, a nude mouse xenograft assay was performed. Tumors generated from the stable FL clone had a significant delay in the onset of tumor formation, slower growth rates, and lower weights than that from the vector control cells (Fig. 3D), suggesting that expression of TAX1BP2 suppressed tumor growth in vivo.
TAX1BP2 Up-regulated p53 and p21CIP.
Because p53 is a well-known tumor suppressor that inhibits G1/S progression,7 we next investigated whether the delay in S-phase entry was the result of an up-regulation of p53. In the stable vector clones, protein expression of p53 and p21CIP was at a low basal level (Fig. 4A, left panel). However, significantly higher protein expression of p53 and p21CIP was detected in the stable FL and ED clones. Interestingly, when the transcript levels of p53 and p21CIP in the stable clones was examined by RT-PCR, the p21CIP, but not p53, transcript was found to be significantly increased (Fig. 4A, right panel), suggesting that TAX1BP2 may stabilize the p53 protein post-transcriptionally, which, in turn, enhanced the transcription of p21CIP.
To further confirm the up-regulation of p53-transactivating activity by TAX1BP2, a luciferase reporter assay was performed by cotransfecting constructs of FL or ED and a p53 luciferase reporter into SMMC-7721 and HepG2 cells. Transcriptional activity of p53 increased dose dependently with an increasing amount of the FL and ED constructs used in transfection (Fig. 4B and Supporting Fig. 2A). To investigate the dependence of p53 on the tumor-suppressor activity of TAX1BP2, two p53-defective cell lines (Hep3B and PLC/PRF/5) were used in the CFA. Ectopic expression of FL and ED did not suppress colony formation in these cell lines, suggesting that p53 may be essential for the tumor-suppressor activity of TAX1BP2 (Fig. 4C). Consistently, the phenomenon that TAX1BP2 up-regulates p21 transcript level was also observed in human HCCs, because the expression of TAX1BP2 was significantly correlated with p21CIP at the transcript level in HCC samples (P = 0.0001, Pearson's correlation; Fig. 1E). Because up-regulation of p53 plays an essential role in response to genotoxic stress and chemoresistance, the expression level of TAX1BP2 in HCC cells treated with two common chemotherapeutic agents (etoposide and cisplatin) separately was examined. Endogenous protein expression of TAX1BP2 and p53 was induced by both agents in HepG2 and SMMC-7721 cells, respectively, demonstrating that both TAX1BP2 and p53 are involved in the response to genotoxic stress (Fig. 4D and Supporting Fig. 2B).
TAX1BP2 Up-regulated p53 and p21CIP Through p38 MAPK Activation.
Several studies reported that loss of centrosome integrity by the depletion of centrosome assembly proteins induces p38/p53-dependent G1/S arrest.8, 9 Thus, we queried whether TAX1BP2 would activate the p38 MAPK–signaling pathway. A high level of p38 and its upstream regulator, MAPK kinase (MKK)3/6, activating phosphorylation was detected in stable FL and ED clones, but not in vector clones (Fig. 5, left panel). Incubation of the stable FL and ED clones with a specific p38 inhibitor (SB203580) remarkably suppressed the up-regulation of p38-phosphorylated p53 (S33), p53, and p21CIP (Fig. 5A, right panel). Furthermore, silencing TAX1BP2 expression by specific short interfering RNAs (siRNAs) in stable FL and ED clones reversed the effects of TAX1BP2 on p38 phosphorylation, p53 and p21CIP induction (Fig. 5B), and BrdU incorporation (Fig. 5C and Supporting Fig. 2C).
Consistently, silencing endogenous TAX1BP2 expression by siRNA in HepG2 cells also significantly inhibited the expression of p53 and p21CIP and promoted cell proliferation, as demonstrated by an increase in BrdU incorporation (Fig. 5D) and colony formation (Supporting Fig. 2D). In line with this, knockdown of TAX1BP2 abolished the effect of etoposide and cisplatin on the induction of p53 and p21CIP (Fig. 5E) and their growth-suppressive effect measured by CFA (Fig. 5F). These results suggested that TAX1BP2 is an upstream regulator of the MKK/p38/p53/p21 pathway, and, upon genotoxic stress, TAX1BP2 may induce cell-cycle arrest by activation of the p38/p53/p21 pathway.
CDK2 Phosphorylated TAX1BP2.
Up-regulation of CDK2 activity was reported in HCC, but the underlying pathogenic mechanism is not completely clear.10 In our previous study, TAX1BP2 was shown to be a substrate of CDK2.5 Based on the CDK2 phosphorylation-site consensus sequence, S/T-P-X-K/R (S/T, serine/threonine; P, proline; X, any; K/R, lysine/arginine), serine residues at the positions of 763, 779, 786, 799, 802, and 1305 of TAX1BP2 were identified as potential CDK2 phosphorylation sites. Because the expression level of the TAX1BP2 FL protein in bacteria was low, truncation mutants of TAX1BP2 corresponding to amino acids 1-815 and 1040-1313, which contained all the potential CDK2 consensus sites, were generated. His-tagged proteins were purified and used for the in vitro kinase assay. The result showed that the 1-815 and 1040-1313 truncation fragments of TAX1BP2 protein were good substrates of CDK2 (Fig. 6A, lanes 3 and 9). Next, serine residues at the positions of 763, 779, 786, 799, and 802 on the TAX1BP2 (1-815) and that at the position of 1305 on the TAX1BP2 (1040-1313) were substituted by an alanine residue to generate a panel of nonphosphorylatable mutants. Though S786A and S799A mutants of TAX1BP2 (1-815) were phosphorylated by CDK2 to a similar extent as the wild type (Fig. 6A, lanes 6 and 7), the S763A, S779A, S802A, and S1305A mutants were phosphorylated at a lower extent (Fig. 6A, lanes 4, 5, 8, and 10), suggesting that S763, S779, S802, and S1305 are putative CDK2-targeted sites.
Phosphorylation of TAX1BP2 By CDK2 at S763 Abolished Its Tumor-Suppressive Activity.
To investigate the effect of CDK2 phosphorylation on the tumor-suppressive activity of TAX1BP2, nonphosphorylable mutants (S763A, S779A, S802A, and S1305A) and phosphorylation mimetic mutants (S763D, S779D, S802D, and S1305D) of TAX1BP2 were generated for the CFA. Similarly to FL, ectopic expression of all the mutants, except S763D, significantly suppressed the colony formation of SMMC-7721 cells (Fig. 6B), suggesting that phosphorylation at S763 can abrogate the growth-suppressive effect of TAX1BP2. To exclude the possibility that the mutations have an effect on the subcellular localization of TAX1BP2, we performed confocal IF staining and found that both S763A and S763D mutants were still targeted to the centrosomes (Supporting Fig. 3B). To further characterize S763 phosphorylation on TAX1BP2, we raised a phosphorylation-specific antibody for the S763 site. To characterize the specificity of the antibody, we showed that the anti-p-TAX1BP2 (S763) (phospho-TAX1BP2 [Serine-763]) antibody detected a strong signal on TAX1BP2 FL protein in the presence of the CDK2/cyclin A complex (Supporting Fig. 3A, lane 3), but not in the absence of the CDK2/cyclin A complex or ATP (Supporting Fig. 3A, lanes 2 and 4), suggesting that the anti-p-TAX1BP2 antibody recognizes CDK2 phosphorylation on TAX1BP2. Furthermore, the anti-p-TAX1BP2 antibody, detecting a strong signal on the TAX1BP2 (1-815) (Fig. 6C, lane 2), but not on the TAX1BP2 (1-815) S763A, mutant (Fig. 6C, lane 4) when the CDK2/cyclin A complex was coincubated in the kinase assay, strongly indicated that the anti-p-TAX1BP2 antibody recognizes the p-TAX1BP2 S763 signal specifically. Next, we tested whether CDK2 phosphorylation of TAX1BP2 at S763 occurs in vivo by transfecting SMMC-7721 cells with constructs of FL or S763A mutant. The anti-p-TAX1BP2 antibody only detected the p-TAX1BP2 (S763) signal in cells expressing FL (Fig. 6D, lane 1), which was the result of endogenous CDK2 activity, but not in cells expressing the S763A mutant (Fig. 6D, lane 3). Incubation of transfected cells with a pharmacological inhibitor of CDK2 (roscovitine) significantly reduced the phospho-TAX1BP2 (S763) signal (Fig. 6D, lane 2), strongly suggesting that the phosphorylation at S763 is specific for CDK2 and that the phospho-specific anti-p-TAX1BP2 antibody can specifically recognize phosphorylation at S763.
To address the effect of S763 phosphorylation on the activation of the p38/p53 axis, FL, S763A or S763D mutants of TAX1BP2 were co-transfected with hemagglutination (HA)-tagged p38 expression constructs into SMMC-7721 cells. Both FL and the S763A mutant, but not the S763D mutant, significantly activated p38, as detected by the phospho-specific anti-p38 antibody (Fig. 6E). To further examine the effect of S763 phosphorylation on TAX1BP2-mediated p53 transcription activity, the S763A and S763D mutants were used for the p53 luciferase reporter assay. Luciferase activity was significantly enhanced in cells with an expression of FL or S763A mutant, but not the S763D mutant, suggesting that phosphorylation of S763 on TAX1BP2 abolished the p53-activating activity (Fig. 6F). Taken together, these results showed that CDK2 phosphorylation of TAX1BP2 at S763 abolishes its tumor-suppressive and p38/p53-activating activity.
Our previous study demonstrated that TAX1BP2, which localizes at the centrosome, was a novel interacting partner of the HTLV-1 oncoprotein, Tax.5 TAX1BP2 acts as an intrinsic block to centrosome overduplication, and inactivation of TAX1BP2 by Tax may be one of the major mechanisms for HTLV-1-mediated aneuploidy and carcinogenesis. However, the functions of TAX1BP2 in carcinogenesis are still far from clear. Our LOH and others' studies had shown that the chromosomal region, 1p36.1-36.3, which coincides with the TAX1BP2 locus, is a region frequently deleted in HCC.11 In this study, we further confirmed that TAX1BP2 suppressed hepatocarcinogenesis. First, we showed that TAX1BP2 transcript and protein were frequently underexpressed (∼42.2%) in human HCC samples and cell lines (Fig. 1A,D). Clinicopathological analysis indicated that TAX1BP2 underexpression was significantly associated with poorer overall survival rates in patients after surgical resection (Fig. 1C). Second, we demonstrated that ectopic expression of TAX1BP2 significantly inhibited HCC cell growth using proliferation, BrdU labeling, and CFA assays (Figs. 2B,C and 3B,C and Supporting Fig. 1A). Third, TAX1BP2 stable expressing HCC cells showed a significant reduction in the in vivo tumor growth (Fig. 3D). Taken together, these data provide the first evidence that TAX1BP2 is a putative tumor suppressor in HCC.
Using a panel of truncation mutants, we mapped an ED that suppresses tumor growth at the central coiled-coil region (aa267-756) of TAX1BP2 (Fig. 2A,C). Similarly to FL, ED also localized at the centrosome (Fig. 2E) and inhibited centrosome overduplication, but to a lesser extent (Fig. 2D). This result indicated that inhibition of centrosome overduplication is related to the tumor-suppressor function of TAX1BP2. Thus, TAX1BP2 plays a dual role in suppressing both tumorigenicity and aberrant centrosome duplication. Though the possibility that additional domains are involved in the suppression of centrosome overduplication is currently under investigation, our data, however, established that TAX1BP2 contains an ED that is essential for both tumor and centrosome overduplication suppressive activities.
Previous studies reported that depletion of centrosome assembly proteins induces p38/p53-dependent cell-cycle arrest at the G1/S phase.8 However, our data revealed that instead of depletion, ectopic expression of TAX1BP2 remarkably activated the p38/p53/p21 axis (Figs. 4A,B and 5A), which, in turn, suppressed cell proliferation and caused a delay in S-phase entry (Fig. 3B,C). In support of this, we observed a significant correlation of TAX1BP2 and p21CIP at the transcript level in human HCCs (Fig. 1E). Thus, the possibility that TAX1BP2 inhibits centrosome overduplication through the up-regulation of p53/p21 warrants further investigation. Further investigation is also required to understand how TAX1BP2 activates MKK. It may be interesting to test whether TAX1BP2 also plays a role in transforming growth factor beta signaling, which activates MKK3/6. In view of this, we believe that restoration of TAX1BP2 expression and function may be a promising approach to suppress centrosome supernumerary and growth of aneuploidy cells.
The p38 pathway is activated by a wide range of cellular stresses, such as genotoxic, osmotic, thermal, hypoxic, and oxidative stresses as well as cytokine treatment.12 Silencing TAX1BP2 expression or treatment with a specific p38 inhibitor in stable TAX1BP2-expressing clones precluded the activation of the p38/p53/p21 pathway and reversed the inhibition of S-phase entry (Fig. 5A-C and Supporting Fig. 2C). In addition, depletion of endogenous TAX1BP2 in HepG2 cells inhibited p53/p21 expression with an enhancement in cell proliferation (Fig. 5D and Supporting Fig. 2D). These results indicated that p38/p53/p21 plays a critical role in TAX1BP2′s tumor-suppressor activity. Furthermore, we confirmed that the TAX1BP2-mediated growth-suppressive effect is p53 dependent by performing the CFAs using p53-defective cells (Fig. 4C).
So far, we have not observed that TAX1BP2 has any effect on cell-survival pathways, namely protein kinase B, p42/44 extracellular signal-related kinase, and nuclear factor light-chain enhancer of activated B cells (Lai WL, unpublished data). Because p53 plays an important role in chemoresistance,13 we showed that genotoxic stress exerted by chemotherapeutic agents could significantly induce the expression of TAX1BP2 (Fig. 4D and Supporting Fig. 2B). Consistent with this, silencing TAX1BP2 expression by siRNA significantly repressed the induction of p53 and suppression of colony formation upon genotoxic stress (Fig. 5E,F), suggesting that restoration of TAX1BP2 can potentially be used to enhance the chemosensitivity of cancer cells, particularly those carrying functional p53. Currently, it remains unclear how genotoxic stress induces the accumulation of TAX1BP2, but it is noteworthy that TAX1BP2 contains a potential ataxia telangiectasia mutated (a DNA damage kinase) phosphorylation site at residue 922 that has been reported on previously.14 Thus, TAX1BP2 may function to inhibit cell growth in response to genotoxic stress, thereby preventing cancer formation. Taken together, our results provided strong evidence that TAX1BP2 is a novel activator of the MKK/p38/p53/p21 pathway, which may play a central role in TAX1BP2 tumor-suppressor function, particularly in response to genotoxic stress.
CDK2 is a key regulator in initiating centrosome duplication, and up-regulation of CDK2 activity was reported in HCC.10 Six putative CDK2-targeted sites were found on TAX1BP2 (Fig. 6A), but only the mutation of S763 to a phosphorylation mimetic mutant (S763D) abolished the growth-suppressive effect of TAX1BP2 (Fig. 6B). Furthermore, S763D mutant was the only mutant that could not activate p38 and p53 (Fig. 6E,F), suggesting that activation of p38/p53 is coupled with the tumor-suppressive function of TAX1BP2. Using a phospho-specific antibody, we demonstrated that S763 was phosphorylated by CDK2 in vivo (Fig. 6D). The possibility that S763 is the target of other CDK kinases apart from CDK2 needs further investigation. However, our current data provide the first evidence that CDK2 can abolish the tumor-suppressive and p38/p53-activating activities of TAX1BP2 by phosphorylation of S763. Also, it suggests the possibility that TAX1BP2 can couple CDK2 and p53/p21 signaling to coordinate proliferation or cell-cycle progression and apoptosis of cells. To this end, we have obtained data to show that TAX1BP2 S763 phosphorylation was significantly enhanced in S and G2 phases during cell-cycle progression (Lai WL, unpublished data).
In conclusion, we demonstrated that TAX1BP2 was frequently underexpressed in HCC and that loss of TAX1BP2 promoted the growth of HCC cells by suppressing activation of the p38/p53/p21 pathway, particularly under genotoxic stress. In CDK2-overexpressing HCC, phosphorylation of TAX1BP2 by CDK2 can override its tumor-suppressor activity, thereby promoting hepatocarcinogenesis. Thus, TAX1BP2 is a novel tumor suppressor in HCC and restoration of TAX1BP2 may present an attractive therapeutic approach for cancer therapy.