Hyperactivation of Ha-ras oncogene, but not Ink4a/Arf deficiency, triggers bladder tumorigenesis (original) (raw)
Ink4a/Arf is not a cooperative partner of Ha-ras in urothelial tumor initiation. A surprising finding of the present study is the apparent lack of synergism between activated Ha-ras and Ink4a/Arf deficiency in urothelial tumorigenesis. The germline loss of one, or even both, alleles of the Ink4a/Arf gene failed to accelerate urothelial tumor formation in mice expressing a constitutively active Ha-ras in the urothelium (Figure 2). This result stands in stark contrast to the demonstrated synergism between these 2 genetic events in other cell types, including melanocytes, astrocytes, and lung and pancreatic epithelial cells (31–33, 52). Our result does not support the idea that since deletion of the Ink4a/Arf locus occurs frequently in the precancerous lesions of the bladder, it must be invariably involved in urothelial tumor initiation (53). However, our result does support the important concept that not only the tumorigenicity of an oncogene but also its cooperative tumor suppressor partner can be cell type specific (3). Precisely why activated Ha-ras and Ink4a/Arf deficiency are not synergistic in the urothelium is presently unclear, but the following scenarios could be involved. First, the activated Ha-ras may require a cooperative partner other than Ink4a/Arf. For example, it cannot be ruled out that the deletion of the 9p21 locus, where Ink4a/Arf resides, actually reflects more of a loss of the adjacent Ink4b gene rather than of the Ink4a/Arf gene per se. Ink4b encodes a CDK inhibitor that is critical in negatively regulating G1/S-phase transition. Although it has not been as well studied as the Ink4a gene, there was reduced expression of the p15Ink4b product in one cohort, occurring in as many as 66% of early-stage bladder tumors (54). In our study, we observed a robust increase in expression of the Ink4b gene in response to Ha-ras even in the hyperplastic phase of urothelial proliferation (Table 1), suggesting that the loss of Ink4b might cooperate with Ha-ras activation in accelerating urothelial tumorigenesis. It would be interesting to test this hypothesis by crossing the Ink4b knockouts with the ras transgenics. Other collaborative partners of Ha-ras may also play a urothelium-specific role. For instance, we recently demonstrated that the loss of p53 significantly shortened the latency of ras-induced urothelial tumors, converting urothelial hyperplasia to full-fledged bladder tumors (55). The fact that p19Arf acts in the p53 pathway but failed to exert the same effect during Ink4a/Arf deficiency (Figure 2 and Table 2) as does p53 raises the possibility that the biological effects resulting from the deficiency of these 2 tumor suppressor genes are not equivalent.
Second, as far as the “partnership” is concerned, it is also possible that, in urothelium, Ink4a/Arf deficiency requires the cooperation of an oncogene other than Ha-ras for tumor initiation. However, even if this were the case, it would only account for a minority of the early-stage bladder tumors, because an overwhelming majority of the early-stage tumors in humans have ras pathway activation in one form or another (see below). Third, while Ink4a/Arf deficiency does not appear to be involved in bladder tumor initiation, it remains possible that it could cooperate with activated Ha-ras or another genetic event at later stages to promote bladder tumor progression. We attempted to test this hypothesis by generating compound mice homozygous for activated Ha-ras and nullizygous for Ink4a/Arf; this approach failed, however, due to the infertility of the ras-homozygous mice (data not shown). Finally, Ink4a/Arf deficiency could be involved in a tumorigenic pathway that is distinct from the one involving the activated Ha-ras. It has been well established that human urothelial tumors are a mixture of biologically, phenotypically, and genetically different entities (23, 56–58). It is possible then that Ink4a/Arf and Ha-ras can act separately in divergent pathways of urothelial tumorigenesis. Definitive answers to these unresolved issues will come from further experimental analyses using both animal models and human tumor specimens.
Tumorigenicity of Ha-ras: critical role of quantitative differences. We were also surprised by the finding that doubling the gene dosage of the activated Ha-ras in the ras-homozygous background, even in the presence of the wild-type Ink4a gene, triggered early-onset, rapidly growing, and fully penetrant urothelial tumors throughout the urinary tract (Figures 3 and 4). This finding strongly suggests that while the low-level expression of a constitutively active Ha-ras is insufficient, even in collaboration with Ink4a/Arf deficiency, to initiate urothelial tumors, overexpression of the constitutively active Ha-ras is both necessary and sufficient to initiate urothelial tumor formation. These results highlight an important mechanism of Ha-ras activation that has not been fully appreciated to date and underscore the importance of evaluating not only structural but also quantitative aspects of Ha-ras activation when assessing the tumorigenic potential of activated Ha-ras oncogene.
Evidence is accumulating that overactivation of Ha-ras occurs in human urothelial tumors. Thus, findings reported here are relevant to human urothelial tumorigenesis as well. Czerniak et al. showed that mutation and overexpression of Ha-ras occurred concurrently in about 10% of the bladder tumors by means of alterative splicing of the last intron of the mutated Ha-ras gene (11, 59). Another mechanism may involve transcriptional upregulation of the mutated Ha-ras gene, as suggested by several independent studies that show that more than half of all human bladder tumors overexpress the ras gene (16, 17). Based on a conservative estimate that 30%–40% of the bladder tumors harbor Ha-ras mutations (59), there is reason to believe that ras mutation and overexpression overlap in a significant number of cases. Furthermore, activation of the ras pathway, via upstream-acting RTKs, such as FGFR3, EGFR, and Erb family proteins, are extremely prevalent in human bladder tumors (19–22). It is highly likely that these RTKs can functionally overactivate the ras pathway in the presence or absence of ras mutations.
Whether the enhanced tumorigenicity from overexpression of an activated Ha-ras applies to other oncogenes is currently unknown, but it has been documented that amplification and/or overexpression of oncogenes are common features of many human tumors (4). Hence, it seems likely that what we have learned here about the gene-dosage dependence of oncogene activation is not limited to Ha-ras but can be extended to other oncogenes. Further experimental studies are needed to investigate this idea.
The mechanism whereby overactivated Ha-ras transforms the urothelium may lie in the imbalance of positive and negative pressures exerted on cell-cycle progression. Under this paradigm, either the increase in positive forces or the decrease in negative forces could tilt the balance toward cell-cycle progression. This concept has been proven valid with tumor suppressors, in which reduced gene dosage per se, via the loss of 1 tumor suppressor gene allele (haploinsufficiency), enhances tumorigenesis (60). As for the Ha-ras oncogene, at low expression levels, its mitogenic effects may be counterbalanced by the antiproliferative forces, such as prosenescence and tumor-suppressing molecules (Figure 1 and Table 1). Consequently, cell cycle regains control, with no tumor formation. In many cases, a cooperative effect, such as the inactivation of a tumor-suppressor gene, is required for Ha-ras to be fully transforming. However, as we showed here, when Ha-ras is overactivated, it may be able to override the antagonistic effects of induced tumor-suppressor genes, resulting in tumorigenesis without the need for a cooperative event. This explanation is completely in line with the paradigm of cell-cycle control and tumorigenesis.
Ras activation causes urothelial tumorigenesis along the low-grade, noninvasive papillary pathway. Unlike most epithelial tumors that follow a single path of evolution from benign to malignant stages, urothelial tumors manifest themselves as at least 2 drastically different diseases (22, 56–59, 61). The low-grade, noninvasive papillary tumors, accounting for 70%–80% of all clinical cases, are often multifocal in the bladder and tend to recur, but they rarely advance to the muscle-invasive and metastatic stages. On the other hand, the muscle-invasive tumors almost invariably follow an aggressive clinical course, with more than 50% of tumors eventually metastasizing to distant organs, despite the complete surgical removal of the bladder. There is compelling evidence from clinicopathological and longitudinal studies that most muscle-invasive tumors are not derived from the low-grade, noninvasive papillary tumors. Rather, they seem to arise de novo or are derived from high-grade, flat, carcinoma-in-situ lesions. Genetic studies of muscle-invasive bladder tumors from humans suggest that more than half of these lesions harbor structural and functional defects in the components of p53 and/or Rb pathways (62, 63). Therefore, the current consensus is that these p53/Rb defects may underlie urothelial tumor development along the invasive pathway. By contrast, the genetic defect(s) that drive urothelial tumor formation along the low-grade, noninvasive pathway have been long sought after, with little success, due partly to the controversy surrounding the role of Ha-ras activation. In this study, we demonstrated that hyperactivation of Ha-ras induced urothelial tumors that were consistently of low pathological grade, papillary, and noninvasive. These data provide direct and strong experimental evidence indicating that hyperactivation of the ras signaling pathway is responsible for the low-grade, noninvasive papillary bladder tumors (Figure 7).
Schematic diagram of signaling effectors underlying the genesis of low-grade, noninvasive papillary urothelial tumors. Low-level expression of a constitutively active ras-GTPase (ras-GTP) activates the MAPK pathway and converts quiescent normal urothelial cells into simple urothelial hyperplasia, which is persistent for an extended period (8–10 months) without progressing to urothelial tumors. In contrast, overactivation of the ras-GTPase profoundly activates survival and angiogenesis signals along PI3K/AKT and STAT pathways, leading to the formation of nodular hyperplasia and low-grade, noninvasive urothelial tumors. The activation of the STAT pathway that is unique to nodular hyperplasia and urothelial tumors, both of which contain mesenchymal components, may result from signaling from growth factors (GFs) and cytokines (CTKs) produced by the tumor mesenchyme. The activation of AKT and STAT pathways may also be facilitated by the functional inactivation of their upstream inhibitor, PTEN, via its C-terminal hyperphosphorylation. FGFR3 mutations, which occur in up to 80% of low-grade, noninvasive human bladder tumors, are likely to transmit the signals along pathways similar to those described above. C, carboxyl terminus.
In direct support of our transgenic data is the recent finding that humans with Costello syndrome, which is caused by germline mutations in the Ha-ras gene, are highly predisposed to developing early-onset bladder tumors (64). Interestingly, all 3 cases of bladder tumors reported so far are of low pathological grade, papillary, and recurrent (65–67). These data strongly and independently support the role of ras activation in this distinct phenotypic pathway of bladder tumorigenesis.
Coactivation of AKT and STAT pathways is required for urothelial tumorigenesis. We observed that during Ha-ras hyperactivation, the preferential activation of certain signaling pathways is essential for urothelial tumor initiation. While activation of the MAPK pathway was sufficient to induce simple urothelial hyperplasia, as observed in heterozygous mice, this pathway alone did not lead to urothelial tumor initiation (Figures 3–6 and Figure 7). In contrast, the AKT pathway underwent profound changes only in the homozygous mice that exhibited severe nodular hyperplasia and urothelial tumors (Figure 5). These changes in AKT included its phosphorylation at both Thr308 and Ser473, nuclear translocation, and activation of its downstream targets (GSK3B and FKHR). These findings suggest that activation of the AKT pathway is crucial for urothelial tumor initiation (Figure 7). Interestingly, there was a strong association between AKT overactivation and the phosphorylation of the C terminus of PTEN (Figure 5). PTEN is a membrane-associated lipid phosphatase that suppresses PI3K by lowering the level of phosphatidylinositol-3-phosphate, thereby inhibiting AKT activity (47). Phosphorylation of PTEN at the C terminus in vitro prevents the recruitment of the protein from the cytosol to the plasma membrane, thus reducing PTEN’s phosphatase activity (50, 68). What we observed here, however, provides new evidence that such a phosphorylation-mediated inactivation of PTEN occurs naturally in tumor cells and in the context of Ha-ras hyperactivation. Although ras itself is capable of activating the AKT pathway by directly interacting with the regulatory subunit of PI3K (69), inactivating a potent upstream inhibitor, PTEN, within the signaling pathway could be regarded as an additional safeguard employed by tumor cells to free themselves from the “check and balance” and to allow for signal amplification to go forward. Similarly, the cooperativity between ras activation and loss of PTEN allele(s) has recently been shown from transgenic models of skin and ovarian tumorigenesis (70, 71), suggesting that this mechanism is not confined to the urothelium. In human bladder tumors, while somatic mutations of PTEN are rare, loss of heterozygosity of the PTEN locus is fairly common, occurring in about 25% of the cases (72–74). Our study suggests yet another mechanism whereby PTEN can be inactivated during urothelial tumor formation. This underscores the importance of studying the functional status of PTEN when assessing the involvement of this protein in the tumorigenesis of bladder and other tissues (Figure 7).
In addition to AKT overactivation, we also observed a profound activation of STAT3 and STAT5, as evidenced by their elevated expression and phosphorylation, almost exclusively in the homozygous Ha-ras mice (Figure 6). As members of the latent STAT gene family, STAT3 and STAT5 are considered oncogenes that are persistently activated in a variety of tumors in response to growth factors and cytokines produced by the extracellular matrix (40). Phosphorylation of STAT3 and STAT5 promotes homotetramerization and nuclear translocation of the proteins and increases transcription of targets critical for cell growth, survival, and angiogenesis (40). The fact that these 2 proteins were not activated at all during simple hyperplasia of the heterozygous ras mice but were markedly activated in nodular hyperplasia and tumors where a substantial amount of extracellular matrix comes in close contact with the tumor cells reflects the importance of signaling crosstalk between transformed urothelial cells and their microenvironment in establishing and maintaining tumor growth (Figure 7). Another possible mechanism of STAT activation may be related to the functional inactivation of PTEN, which normally inhibits not only the PI3K/AKT pathway, but also the STAT pathway (48, 75). Given the remarkable degree of AKT and STAT activation in low-grade, noninvasive urothelial tumors, selectively inhibiting these molecules and/or restoring the function of their upstream antagonist such as PTEN will likely have an important therapeutic effect in treating and preventing the recurrence of this type of bladder tumor.