Learn more: PMC Disclaimer | PMC Copyright Notice
Isoform-Specific Ras Activation and Oncogene Dependence during MYC- and Wnt-Induced Mammary Tumorigenesis▿
Abstract
We have previously shown that c-MYC-induced mammary tumorigenesis in mice proceeds via a preferred secondary pathway involving spontaneous activating mutations in Kras2 (C. M. D'Cruz, E. J. Gunther, R. B. Boxer, J. L. Hartman, L. Sintasath, S. E. Moody, J. D. Cox, S. I. Ha, G. K. Belka, A. Golant, R. D. Cardiff, and L. A. Chodosh, Nat. Med. 7:235-239, 2001). In contrast, we now demonstrate that Wnt1-induced mammary tumorigenesis proceeds via a pathway that preferentially activates Hras1. In addition, we find that expression of oncogenic forms of Kras2 and Hras1 from their endogenous promoters has markedly different consequences for the progression of tumors to oncogene independence. Spontaneous activating Kras2 mutations occurring in either MYC- or Wnt1-induced tumors were strongly associated with oncogene-independent tumor growth following MYC or Wnt1 downregulation. In contrast, Hras1-mutant Wnt1-induced tumors consistently remained oncogene dependent. Additionally, Kras2-mutant tumors exhibited substantially higher levels of ras-GTP, phospho-Erk1/2, and phospho-Mek1/2 compared to Hras1-mutant tumors, suggesting the involvement of the ras/mitogen-activated protein kinase (MAPK) pathway in the acquisition of oncogene independence. Consistent with this, by use of carcinogen-induced ras mutations as well as knock-in mice harboring a latent activated Kras2 allele, we demonstrate that Kras2 activation strongly synergizes with both c-MYC and Wnt1 in mammary tumorigenesis and promotes the progression of tumors to oncogene independence. Together, our findings support a model for tumorigenesis in which c-MYC and Wnt1 select for the outgrowth of cells harboring mutations in specific ras isoforms and that these secondary mutations, in turn, determine the extent of ras/MAPK pathway activation and the potential for oncogene-independent growth.
A wealth of experimental evidence indicates that the development and progression of human cancers require multiple genetic events that occur in a stepwise fashion. Although few genetic alterations have been definitively placed in the sequence of events that contribute to breast cancer in humans, several genetic pathways have been implicated in determining breast cancer susceptibility, response to therapy, and prognosis.
c-MYC amplification occurs in 10 to 30% of human breast cancer (3, 20) and is correlated with aggressive tumor behavior and poor prognosis (9, 30). Wnt1, an upstream activator of c-MYC, was initially identified on the basis of its insertional activation by the mouse mammary tumor virus (MMTV) during the process of murine mammary tumorigenesis (27). Although mutations in Wnt family members are rare in human breast cancer, overexpression of Wnt family members and alterations in components of the Wnt signaling pathway are common (5). For example, β-catenin, which translocates to the nucleus when activated by the Wnt pathway, is localized in the nucleus in up to 60% of human breast cancers (21).
A third oncogene that has been extensively studied in human breast cancer is ras. Unlike the relatively high frequency of genomic alterations in c-MYC, ras family members are activated by mutation in less than 5% of human breast cancers (24, 29, 31). Nevertheless, like that of Wnt, ras pathway activation is extremely common in human breast cancer (7, 38). This likely results from the fact that regulatory molecules that signal through ras—such as HER-2/neu (1, 33), insulin-like growth factor, insulin-like growth factor receptor (17, 34), and c-Src (2)—are frequently amplified, overexpressed, or otherwise activated in human breast cancer.
While it is clear that multiple genetic alterations are required for the formation of breast cancers, relatively few such secondary alterations have been identified. In this regard, mouse models bearing defined initiating genetic alterations relevant to human cancer have been a valuable means to identify collaborating oncogenic events. For example, forced overexpression of c-MYC and v-Ha-ras in the mammary glands of bitransgenic mice results in strongly synergistic tumor formation (19, 32). Consistent with this, MYC-induced mammary tumors frequently harbor spontaneous activating point mutations in Kras2 (8). Similarly, over 50% of MMTV-Wnt1 tumors contain activating point mutations in Hras1 (28). These observations suggest that activation of the ras family pathway may contribute to MYC- and Wnt1-induced tumorigenesis.
To study mammary tumor progression, our laboratory has developed conditional bitransgenic systems for the doxycycline-inducible expression of c-MYC or Wnt1 in the murine mammary gland (8, 14). In these systems, the reverse tetracycline transcriptional activator (rtTA) is specifically expressed in the mammary epithelium of transgenic mice under the control of the MMTV promoter and, in the presence of doxycycline, induces expression of c-MYC or Wnt1 from a tetracycline-dependent promoter. Using these models, we have previously reported that expression of c-MYC in bitransgenic MMTV-rtTA/TetO-MYC (MTB/TOM) female mice leads to the development of mammary adenocarcinomas with an average latency of 22 weeks (8). Nearly half of these tumors harbor spontaneous activating point mutations in Kras2 or Nras, with the majority found in Kras2. Upon downregulation of c-MYC in tumors by doxycycline withdrawal, 50% of MYC-induced tumors regress to a nonpalpable state whereas the remaining tumors resume growth in the absence of c-MYC overexpression or MYC pathway activation (4, 8). Notably, in MYC-induced mammary tumors, activating mutations in Kras2—but not Nras—are strongly associated with the acquisition of oncogene independence (8). This suggests that Kras2 may contribute both to the development of MYC-induced mammary tumors and to their progression to oncogene independence.
Similar to c-MYC, conditional expression of Wnt1 in bitransgenic MMTV-rtTA/TetO-Wnt1 (MTB/TWNT) mice leads to mammary tumor formation with an average latency of 20 weeks (14). In contrast to MYC-induced tumors, however, abrogation of Wnt1 expression by doxycycline withdrawal results in the regression of virtually all tumors to a nonpalpable state (14). Of note, it has recently been reported that more than half of MMTV-Wnt1 tumors harbor activating mutations in Hras1, but not in Kras2 or Nras (28), raising the question of whether MYC and Wnt1 preferentially synergize with different ras family members.
We now report that the majority of Wnt1-induced tumors in MTB/TWNT mice harbor activating point mutations in Hras1, whereas a far smaller percentage of tumors harbor activating point mutations in Kras2. Nevertheless, treatment of MYC and Wnt1 transgenic mice with the carcinogen MNU (N-methyl-N-nitrosourea) demonstrated that activation of either Hras1 or Kras2 can strongly synergize with c-MYC or Wnt1 in mammary tumor formation. However, tumors bearing Hras1 mutations remain oncogene dependent, whereas tumors bearing Kras2 mutations uniformly progress to oncogene independence. Oncogene independence, in turn, is strongly associated with high levels of ras and mitogen-activated protein kinase (MAPK) pathway activity, suggesting a biochemical basis for the differential oncogene dependence exhibited by tumors bearing mutations in different ras isoforms. In aggregate, our findings suggest a model for tumorigenesis in which c-MYC and Wnt1 select for the outgrowth of cells harboring mutations in specific ras isoforms, which in turn determines the extent of MAPK pathway activation and the potential for oncogene-independent growth.
MATERIALS AND METHODS
Animals and tissues.
MTB, TOM, and TWNT transgenic mice were maintained on an FVB background and have previously been described (8, 13, 14). K-rasLA2 mice (18) were obtained from Jax. MTB/TOM and MTB/TWNT mice were bred to K-rasLA2 mice to generate MTB/TOM/K-rasLA2 and MTB/TWNT/K-rasLA2 tritransgenic mice on a mixed FVB/C57Bl6/129sv background. Expression of the c-MYC or Wnt1 transgenes was induced by the administration of 2.0 mg/ml doxycycline with 5% sucrose in drinking water.
For MNU treatment, 1.0 g of MNU (Sigma-Aldrich) was dissolved in 100 ml of 0.85% NaCl adjusted to pH 7.0 with acetic acid. Animals were treated with 50 mg of MNU/kg of body weight by intraperitoneal injection.
For all studies, mice were monitored at least once per week for tumor formation. Tumor size was measured using calipers. Mice were withdrawn from doxycycline when tumors reached ∼10 by 10 mm in size and were monitored for regression. Tissue for biochemical analysis was obtained by biopsy prior to doxycycline withdrawal.
ras mutation analysis.
Total RNA was harvested as previously described (23). cDNA was synthesized using a First Strand cDNA synthesis kit (Amersham Biosciences Corp.) and PCR amplified using primers specific for Hras1, Nras, or Kras2. PCR products were purified with a GeneClean III kit (Qbiogene Inc.) and sequenced to detect point mutations at codons 12, 13, and 61 as previously described (4, 8).
Western analysis and ras activity assays.
Snap-frozen mammary and tumor tissues were homogenized using EBC buffer (0.5% NP-40, 50 mM Tris [pH 8.0], 0.12 M NaCl, 25 mM NaF, 1 mM β-glycerophosphate, 1 mM sodium orthovanadate, and 1 tablet protease inhibitors/50 ml EBC buffer) and a Dounce homogenizer. Equal amounts of protein from each extract were electrophoresed on 10% sodium dodecyl sulfate-polyacrylamide electrophoresis gels and transferred to nitrocellulose membranes. Membranes were stained with Ponceau S to confirm equal protein loading and efficient transfer. Membranes were blocked in phosphate-buffered saline-0.1% Tween 20-5% nonfat dry milk (PBST-MILK) for 1 h and incubated with primary antibody diluted in PBST-MILK at room temperature (RT) for 1 h or at 4°C overnight. Membranes were washed three times with PBST at RT for 5 min each and then incubated with secondary antibody diluted in PBST-MILK at RT for 1 h. Blots were washed three times with PBST and twice with phosphate-buffered saline, developed using the ECL Plus system (Amersham Biosciences), and exposed to film (Kodak XAR-5).
Ras activity assays were carried out using a ras activation assay kit (Upstate Biotechnology) according to the manufacturer's specifications. Briefly, the ras binding domain of human RAF1 fused to glutathione S-transferase was used to precipitate GTP-bound ras from protein lysates. Precipitated ras-GTP was visualized by Western blotting.
The primary antibodies used were anti-ras clone RAS10 (Upstate Biotechnology) (1:1,000 dilution); anti-phospho-p44/42 MAPK (Thr202/Tyr 204) (anti-phospho-Erk1/2) (Cell Signaling) (1:1,000 dilution); anti-phospho-MEK1/2 (Ser217/221) (Cell Signaling) (1:1,000 dilution); anti-β-tubulin (Biogenex) (1:10,000 dilution); and anti-TFIIH p89, clone S-19 (Santa Cruz Biotechnology) (1:1,000 dilution). The secondary antibodies used were horseradish peroxidase-conjugated goat anti-mouse and horseradish peroxidase-conjugated goat anti-rat antibodies (Jackson ImmunoResearch) (1:10,000 dilution).
Quantitative PCR for Kras2 amplification.
For genomic DNA preparation, tissue was incubated with 500 μl digestion buffer (100 mM EDTA, 50 mM Tris [pH 8.0], 0.5% sodium dodecyl sulfate, 1.7 μl/ml proteinase K) overnight at 55°C. Following addition of 500 μl PCIA (50% phenol, 48% chloroform, 2% isoamyl alcohol), samples were vortexed and spun at maximum speed in a microcentrifuge for 10 min and 300 μl of the top layer was transferred to a fresh tube. A 600-μl volume of 100% EtOH was added to precipitate DNA, which was then washed with 70% EtOH and resuspended in 100 μl of water.
The Kras2 probe (ACCCTACGATAGAGGTAACG) for quantitative PCR anneals to the intron 1-exon 1 junction of genomic Kras2, and the primers (K2651-4100F [GCCTTGACGATACAGCTAATTCAGAAT] and K2651-4100R [GCGCACGCAGACTGTAGA]) amplify a region spanning the intron 1-exon 1 junction. 18S RNA (assay identification number 4319413E-0412015) was used as an endogenous loading control. Quantitative PCRs were performed in a mixture consisting of 10 μl TaqMan Universal MasterMix, 1 μl TaqMan primer-probe mix, and 5 μl of DNA in a total volume of 20 μl. Reactions were performed in duplicate in 384-well microtiter plates in an ABI Prism 7900HT real-time PCR system (Applied Biosystems). Reaction conditions were as follows: 95°C denaturation for 10 min followed by 40 cycles consisting of 95°C denaturation for 15 s and 60°C annealing-extension for 1 min. Kras2 amplification levels were normalized to 18S RNA amplification levels to control for differences in DNA concentration.
RESULTS
Isoform-specific ras mutations develop during MYC- and Wnt1-induced mammary tumorigenesis.
We have previously reported that MYC-induced mammary tumorigenesis is commonly accompanied by spontaneous activating point mutations in Kras2 and Nras (8). As c-MYC is a downstream target of the Wnt signaling pathway, we reasoned that Wnt1-induced tumors might also harbor activating ras mutations. In support of this hypothesis, Podsypanina et al. have shown that MMTV-Wnt1 mammary tumors frequently harbor activating Hras1 mutations (28). Therefore, we examined mammary adenocarcinomas arising in MTB/TWNT mice for activating mutations in Hras1, Kras2, and Nras (Table (Table1).1). This analysis revealed that 54% (51/95) of Wnt1-induced tumors harbored activating point mutations in Hras1 and 3% (3/95) harbored activating point mutations in Kras2, whereas none harbored activating point mutations in Nras (Kras2 versus Hras1, P < 0.0001; Nras versus Hras1, P < 0.0001). No tumor had detectable mutations in more than one ras isoform. Consistent with these and previous findings, we further determined that 20% (2/10) of MMTV-Wnt1 tumors harbored activating Hras1 mutations (28). These results demonstrate that, in contrast to the preferential association of MYC-induced mammary tumors with Kras2 mutations, Wnt1-induced mammary tumors are preferentially associated with mutations in Hras1.
TABLE 1.
Frequency of ras mutations in tumors from MTB/TWNT and MTB/TOM mice
| Transgenic line | % ras mutations
| ||
|---|---|---|---|
| Hras1 | Kras2a | Nrasb | |
| MTB/TWNT (n = 95) | 54 | 3 | 0 |
| MMTV-Wnt1 (n = 10) | 20 | 0 | 0 |
MNU synergizes with c-MYC and Wnt1 in mammary tumorigenesis and alters ras mutation frequency and distribution.
The preferential occurrence of Kras2 mutations in MYC-induced tumors and Hras1 mutations in Wnt1-induced tumors suggested the possibility that c-MYC and Wnt1 may preferentially synergize with different ras isoforms during mammary tumorigenesis. Therefore, we hypothesized that c-MYC overexpression in the mammary gland might select for the outgrowth of Kras2-mutant cells, whereas Wnt1 might select for the outgrowth of Hras1-mutant cells. To address this hypothesis, we treated female MTB/TOM and MTB/TWNT mice with MNU, an agent with a 10-min biological half-life that induces mammary tumors harboring Hras1 and Kras2 mutations in mice and rats (15, 16, 25, 35). One week following MNU treatment, expression of either c-MYC or Wnt1 was induced in the mammary epithelium by doxycycline administration.
MNU treatment prior to oncogene induction decreased the median latency of MYC-induced tumor formation from 22 weeks to 8.3 weeks (Fig. (Fig.1A;1A; P < 0.0001) and increased tumor multiplicity from 1.2 to 2.1 tumors per animal (P = 0.0004). Similarly, MNU decreased the median latency of Wnt1 tumorigenesis from 20 weeks to 3.5 weeks (Fig. (Fig.1B;1B; P < 0.0001) and increased tumor multiplicity from 1.2 to 5.9 tumors per animal (P < 0.0001). These data suggest that mutations induced by MNU treatment strongly synergize with either c-MYC or Wnt1 in mammary tumor formation.
MNU synergizes with Wnt1 and MYC in mammary tumorigenesis. (A) Mammary tumor-free survival curves for MTB/TOM mice (n = 7) injected intraperitoneally with 50 mg of MNU/kg of body weight at 6 weeks of age and induced with doxycycline at 7 weeks of age compared to untreated, doxycycline-induced MTB/TOM animals (n = 148). (B) Mammary tumor-free survival curves for MTB/TWNT animals (n = 11) subjected to the same MNU treatment as the animals represented in panel A compared to untreated, doxycycline-induced MTB/TWNT animals (n = 58). (C) Mammary tumor-free survival curves for MTB/TOM animals (n = 11) induced with doxycycline at 6 weeks of age and injected intraperitoneally with 50 mg/kg of MNU at 7 weeks of age compared to untreated, doxycycline-induced MTB/TOM animals (n = 148). (D) Mammary tumor-free survival curves for MTB/TWNT animals (n = 9) subjected to the same MNU treatment as the animals represented in panel C compared to untreated, doxycycline-induced MTB/TWNT animals (n = 58). Animals were palpated weekly for mammary tumors.
Analysis of mammary tumors harvested from MNU-treated c-MYC-induced mice revealed an increased frequency of activating mutations in Hras1 (25% versus 0%; P = 0.0021) compared to tumors arising in non-MNU-treated animals (Table (Table2).2). In addition, in MYC-induced tumors from MNU-treated mice, there was a modest increase in the frequency of activating mutations in Kras2 (42% versus 34%; P = not significant [NS]) and a modest decrease in the frequency of Nras mutations (0% versus 11%; P = NS) compared to that seen with tumors arising in animals not treated with MNU (Table (Table2).2). Analysis of Wnt1-induced tumors arising in MNU-treated animals revealed a modest decrease in the frequency of activating Hras1 mutations (21% versus 34%; P = NS) without change in the frequency of Kras2 or Nras mutations compared to that seen with tumors in untreated animals (Table (Table22).
TABLE 2.
Frequency of ras mutations in tumors from MNU-treated and untreated MTB/TWNT and MTB/TOM mice
| Transgenic line and treatment | No. of mice | % ras mutations
| |||
|---|---|---|---|---|---|
| Hras1a | Kras2b | Nrasc | Totald | ||
| MTB/TOM | |||||
| None | 85 | 0 | 34 | 11 | 45 |
| MNU before Doxe | 12 | 25 | 42 | 0 | 67 |
| MNU after Dox | 12 | 0 | 83 | 0 | 83 |
| MTB/TWNT | |||||
| None | 32 | 34 | 3 | 0 | 37 |
| MNU before Dox | 24 | 21 | 4 | 0 | 25 |
| MNU after Dox | 12 | 8 | 33 | 0 | 41 |
As the majority of ras mutations identified in Wnt1-induced tumors in MNU-treated mice were in Hras1, and the majority of ras mutations identified in MYC-induced tumors in MNU-treated mice were in Kras2, these findings were consistent with a model in which Wnt1 expression selects for the outgrowth of cells harboring Hras1 mutations whereas c-MYC expression selects for the outgrowth of cells harboring Kras2 mutations. However, our observation that up to 25% of MYC-induced tumors in MNU-treated mice harbored Hras1 mutations suggested the possibility that Hras1 is able to synergize with c-MYC as well as Wnt1. Of note, we did not observe a significant increase in the overall frequency of ras mutations in either MYC- or Wnt1-induced tumors in MNU-treated mice. Given the striking synergy observed between MNU treatment and each of these oncogenes, these findings raise the possibility that MNU may induce genetic alterations other than ras mutation that synergize with c-MYC or Wnt1.
As an alternate approach to this question, oncogene expression in MTB/TOM and MTB/TWNT mice was induced with doxycycline 1 week prior to MNU treatment. Similar to the effects of MNU treatment prior to oncogene induction, MNU treatment following oncogene induction greatly accelerated mammary tumorigenesis and increased tumor multiplicity (Fig. 1C and D). MYC-induced tumor latency decreased from 22 to 9.0 weeks (P < 0.0001), and multiplicity increased from 1.2 to 4.1 tumors per animal (P < 0.0001). Similarly, Wnt1-induced tumor latency decreased from 20 to 5.0 weeks (P < 0.0001) and multiplicity increased from 1.2 to 6.9 tumors per animal (P < 0.0001). Thus, MNU treatment either before or after oncogene induction strongly synergizes with MYC- or Wnt1-induced tumorigenesis.
Analysis of ras mutation frequencies in mammary tumors revealed a marked increase in the percentage of MYC tumors bearing activating ras mutations in mice treated with MNU after oncogene induction compared to the results seen with untreated animals (83% versus 45%; P = 0.012) (Table (Table2).2). The distribution of ras mutations in MYC tumors was also altered, with 83% of MYC-induced tumors from mice treated with MNU after oncogene induction harboring Kras2 mutations compared to 34% of MYC tumors in untreated mice (P = 0.003).
For Wnt1-induced tumors, while the overall frequency of ras mutations remained similar, the distribution of ras mutations between Wnt1 tumors in MNU-treated and untreated mice changed significantly (Table (Table2).2). Whereas 3% of Wnt1-induced tumors in untreated mice harbored activating Kras2 mutations, 33% of tumors in MTB/TWNT mice treated with MNU after oncogene induction harbored Kras2 mutations (P = 0.019). This increased frequency of Kras2 mutation was accompanied by a corresponding decrease in Hras1 mutation frequency (8% versus 34%; P = NS) in MNU-treated versus untreated animals. In aggregate, these findings provide additional support for the model that ras mutation is a rate-limiting step in both MYC- and Wnt1-induced tumorigenesis and further suggest that activated Kras2 and Hras1 are each able to synergize with Wnt1 or c-MYC in mammary tumorigenesis.
Loss of p53 in Wnt1-induced tumors partially abrogates selection pressure for Hras1 mutation during mammary tumorigenesis.
It has previously been shown that half of tumors arising in MMTV-Wnt1 transgenic mice bear activating point mutations in Hras1 (28). Podyspanina and colleagues further demonstrated that mammary tumors arising in p53 null MMTV-Wnt1 mice lacked activating Hras1 mutations. Surprisingly, mammary tumors arising in MMTV-Wnt1 mice heterozygous for a null allele of p53 exhibited an Hras1 mutation frequency indistinguishable from that observed in p53 wild-type mice, regardless of whether tumors displayed loss of heterozygosity (LOH) (28). These findings suggest that loss of p53 function early in tumor development may obviate the selective pressure for Hras1 mutation.
To determine whether similar changes in selective pressure due to loss of p53 occurred in our conditional system, we analyzed ras mutation frequencies in 36 MTB/TWNT tumors that arose in mice heterozygous for a null allele of p53. Of 30 MTB/TWNT/p53+/− tumors that had not undergone LOH at the p53 locus as evaluated by Southern hybridization, 11 (37%) harbored activating Hras1 mutations. In contrast, of six MTB/TWNT/p53+/− tumors that had undergone p53 LOH, none harbored Hras1 mutations. Consistent with this difference and previous reports, no Hras1 mutations were detected in four MTB/TWNT tumors arising in p53 null mice. These data suggest that reduced p53 pathway activity decreases the selective pressure for the outgrowth of cells bearing Hras1 mutations during Wnt1-induced mammary tumorigenesis.
Activating Kras2 mutations are associated with Wnt1 oncogene-independent tumor growth.
We have previously demonstrated that the presence of spontaneous activating point mutations in Kras2 in MYC-induced mammary tumors is tightly correlated with their ability to resume growth following doxycycline withdrawal and MYC downregulation (4, 8). To determine whether activating ras mutations in Wnt1-induced mammary tumors were also associated with their acquisition of oncogene-independent growth, tumors were subjected to biopsy while they were still expressing Wnt1 to establish their ras mutation status. Mice were then withdrawn from doxycycline and tumor sizes monitored to determine their regression behavior in response to Wnt1 downregulation (Table (Table33).
TABLE 3.
Kras2 mutations are associated with the progression of Wnt1-induced mammary tumors to oncogene independence
| MTB/TWNT tumor ras status | % Tumors with indicated status
| |
|---|---|---|
| Full regression | Incomplete regression | |
| Wild type (n = 30)a | 93 | 7 |
| Hras1 mutant (n = 33)b | 94 | 6 |
| Kras2 mutant (n = 3) | 33 | 67 |
Analysis of tumor biopsies for ras mutations revealed that 93% (28/30) of Wnt1-induced tumors wild type for Hras1, Kras2, and Nras regressed to a nonpalpable state following doxycycline withdrawal. Similarly, 94% (31/33) of tumors with activating point mutations in Hras1 also regressed to a nonpalpable state. In contrast, of three Wnt1-induced tumors identified with activating point mutations in Kras2, two grew in a Wnt1-independent manner following doxycycline withdrawal (Hras1 versus Kras2; P = 0.0014). These findings are consistent with a model in which activating Kras2 mutations in Wnt1-induced tumors correlate with the acquisition of oncogene-independent growth in a manner analogous to that observed for Kras2-mutant MYC-induced tumors. In contrast, Hras1 mutation during Wnt1-induced mammary tumorigenesis was not associated with oncogene-independent growth. These data suggest that—in the context of Wnt1-induced tumorigenesis—activating Kras2 mutations promote the progression of tumors to oncogene independence, whereas Hras1 mutations do not.
Wnt1 tumors do not progress to oncogene independence despite repeated cycles of Wnt1 activation.
We previously demonstrated that nearly all MYC-induced mammary tumors progress to oncogene independence following repeated cycles of MYC induction and deinduction (4) in part attributable to the emergence of novel Kras2 mutations in recurring tumors. The fact that neither MYC- nor Wnt1-induced tumors have been identified with activating point mutations in more than one ras isoform, despite the fact that ras mutation is common in these tumors, implies that activation of one ras isoform relieves the selective pressure for activation of other ras isoforms. This, in turn, suggests that Hras1, Kras2, and Nras are to some extent functionally redundant in their ability to synergize with c-MYC and Wnt1 during mammary tumorigenesis. As a consequence, since Wnt1-induced mammary tumors are most commonly associated with activating mutations in Hras1 rather than Kras2, and since Hras1 mutations are not associated with oncogene-independent growth, we hypothesized that Wnt1 tumors would be more likely than MYC tumors to remain dependent upon oncogene expression for tumor growth.
To determine whether Wnt1-induced tumors could be induced to progress to oncogene independence in a manner similar to that observed for MYC-induced tumors, we subjected Wnt1-induced tumors to multiple cycles of oncogene induction and deinduction (Fig. (Fig.2A).2A). Wnt1 was initially induced in the mammary gland until tumors developed and reached 1 cm3 in size. Wnt1 expression was then abrogated by doxycycline withdrawal and tumor regression monitored. Tumors that regressed to a nonpalpable state were then induced to regrow by readministration of doxycycline. When tumors again reached 1 cm3 in size, Wnt1 expression was once again terminated. This cycle of Wnt1 induction and deinduction was repeated three times.
Wnt1-induced mammary tumors remain stably dependent upon Wnt1 for tumor maintenance and growth. (A) Representative growth curves of three mammary tumors that underwent repeated cycles of Wnt1 induction and deinduction, showing progressing to Wnt1 oncogene independence during the first, second, or third cycle of oncogene induction. (B) Chart displaying the percentages of tumors that remained dependent upon Wnt1 for tumor maintenance and growth at each cycle of deinduction. Blue bars represent the percentages of tumors that remained Wnt1 dependent in each round of deinduction. Maroon bars represent the cumulative percentages of Wnt1-dependent tumors determined for by each deinduction round.
Consistent with our prior observations, nearly all Wnt1-induced tumors (27/30) regressed to a nonpalpable state following the first round of doxycycline withdrawal (Fig. (Fig.2B).2B). When mice harboring oncogene-dependent tumors were reinduced with doxycycline, in all cases tumors rapidly regrew at the site of the initial tumor. This indicates that residual tumor cells survive following oncogene deinduction and remain susceptible to the oncogenic effects of Wnt1. Similar to their behavior during the first round of Wnt1 downregulation, 100% of tumors (23/23) fully regressed during the second round of deinduction (Fig. (Fig.2B).2B). This oncogene-dependent behavior persisted through a third round of deinduction, in which 100% (19/19) of the remaining tumors underwent regression to a nonpalpable state. Thus, in marked contrast to our findings in MYC-induced mammary tumors, Wnt1-induced mammary adenocarcinomas remain stably dependent upon Wnt1 for tumor maintenance and growth.
Differential ras pathway activation in Kras2 and Hras1-mutant tumors.
Given the differential oncogene-dependent behavior of tumors bearing Kras2 versus Hras1 mutations, we wished to investigate potential differences in the biochemical consequences of Kras2 and Hras1 activation in vivo. Specifically, we considered the possibility that Kras2 mutation activates the ras pathway to a different extent than Hras1 mutation. To explore this possibility, we analyzed the extent of ras and MAPK pathway activation by determining levels of ras-GTP, phosphorylated Erk1/2, and phosphorylated Mek1/2 in protein lysates prepared from MYC- and Wnt1-induced mammary tumors harboring mutations in different ras isoforms (Fig. (Fig.33 and and44).
Ras and MAPK pathway activities in Wnt1- and MYC-induced mammary tumors differ according to ras mutation status. (A) Immunoblots showing levels of ras-GTP, ras, phospho-Mek1/2, and phospho-Erk1/2 in Kras2 mutant MTB/TOM tumors compared to the levels seen with MTB/TOM tumors without detectable ras mutations. The results for mammary glands from doxycycline-induced and uninduced MTB/TRAS mice that permit expression of v-Ha-ras in the mammary gland are shown as negative and positive controls. β-Tubulin is shown as a loading control. (B) Immunoblots showing levels of ras-GTP, ras, phospho-Mek1/2, and phospho-Erk1/2 in Hras1 mutant MTB/TWNT tumors compared to the levels obtained with MTB/TWNT tumors without detectable ras mutations. Uninduced and induced mammary glands from MTB/TRAS mice were used as negative and positive controls. β-Tubulin is shown as a loading control.
Ras and MAP kinase pathway activity in Wnt1- and MYC-induced mammary tumors bearing mutations in different ras family members. (A) Immunoblots showing levels of ras-GTP, ras, phospho-Mek1/2, and phospho-Erk1/2 in Hras1-mutant MTB/TWNT tumors compared to the levels seen with Kras2-mutant MTB/TOM tumors. Uninduced and induced mammary glands from MTB/TRAS mice were used as negative and positive controls. β-Tubulin is shown as a loading control. (B) Immunoblots showing levels of ras-GTP and ras for MTB/TOM and MTB/TWNT tumors without detectable ras mutations or with Kras2 mutations and for MTB/TWNT tumors with Hras1 mutations. Uninduced and induced mammary glands from MTB/TRAS mice were used as negative and positive controls. β-Tubulin is shown as a loading control.
Compared to MYC-induced mammary tumors without detectable ras mutations, MYC-induced tumors harboring activating Kras2 mutations exhibited significantly higher levels of ras-GTP as well as higher levels of phospho-Erk1/2 and phospho-Mek1/2 (Fig. (Fig.3A).3A). Notably, Wnt1 tumors with activating Hras1 mutations had significantly, but only modestly, elevated levels of ras-GTP compared to Wnt1-induced tumors without ras mutations (Fig. (Fig.3B).3B). Moreover, no elevations in phospho-Erk1/2 or phospho-Mek1/2 were observed in Hras1-mutant Wnt1 tumors. Direct comparison of Hras1-mutant Wnt1 tumors to Kras2-mutant MYC tumors revealed markedly higher levels of ras-GTP, phospho-Erk1/2, and phospho-Mek1/2 in Kras2-mutant MYC tumors (Fig. (Fig.4A).4A). This comparison also demonstrated that phosphorylation of the MAPK pathway targets p90RSK and Elk1 was elevated in MYC tumors harboring Kras2 mutations compared to Hras1-mutant Wnt1 tumors (Fig. (Fig.4A).4A). Moreover, consistent with the inference that Kras2 mutation is a more potent activator of the MAPK pathway than Hras1 mutation in mammary tumors, Wnt1-induced tumors with activating Kras2 mutations displayed high levels of ras-GTP that were comparable to those observed in MYC tumors harboring Kras2 mutations (Fig. (Fig.4B4B).
In addition to MAPK pathway activation, we also addressed the possibility that other downstream ras effector pathways, such as the RalGDS or PI3K pathways, might be differentially activated in MYC- and Wnt1-induced mammary tumor bearing mutations in different ras family members. To study the RalGDS pathway, we examined the levels of Ral-GTP in Hras1-mutant Wnt1 tumors, Kras2-mutant MYC tumors, and MYC- and Wnt1-induced tumors wild type for ras (data not shown). Analogously, to study the PI3K pathway, we compared the levels of phospho-Akt in a series of Western blot analyses that included Hras1-mutant Wnt1 tumor samples, Kras2-mutant MYC tumor samples, and Wnt1- and MYC-induced tumors samples without detectable ras mutations (data not shown). These experiments failed to reveal any consistent differences in the levels of Ral-GTP or in the levels of phospho-Akt between MYC- and Wnt1-induced tumors harboring different ras mutations or MYC- and Wnt1-induced tumors wild type for ras.
In aggregate, our findings demonstrate that tumors bearing spontaneous activating Kras2 mutations in which Kras2 is expressed from its endogenous promoter exhibit higher levels of ras and MAPK pathway activation than tumors without ras mutations or tumors bearing Hras1 mutations. This, in turn, suggests that the oncogene-independent growth observed in tumors bearing activating Kras2—but not Hras1—mutations could be due to increased activation of the ras and MAPK pathways selectively conferred by Kras2 mutation.
Activated Kras2 synergizes with c-MYC and Wnt1 in mammary tumorigenesis.
The high frequency of spontaneous activating ras mutations that we observed in MYC- and Wnt1-induced mammary tumors, as well as the increased Kras2 mutation frequency and synergy observed between MNU treatment and MYC and Wnt pathway activation, suggested that expression of activated Kras2 from its endogenous promoter can synergize with MYC and Wnt1 in mammary tumor development. However, in each of the above-described experimental contexts, multiple genetic alterations in addition to Kras2 mutation likely occur during the process of tumorigenesis, thereby complicating assessment of the contribution of Kras2 activation to tumor growth and the acquisition of oncogene independence. Moreover, while v-Ha-ras has been shown to synergize with MYC in carcinogenesis in several model systems, including the mammary gland (19, 32), to date synergy between c-MYC and Kras2—or between Wnt1 and any ras isoform—has not been demonstrated in vivo.
In light of these observations, we analyzed mice bearing a latent activated Kras2 allele knocked into the endogenous locus (K-rasLA2) (18) to determine directly whether expression of activated Kras2 from its endogenous promoter could synergize with c-MYC or Wnt1 in mammary tumorigenesis and confer oncogene-independent growth. In this strain, spontaneous recombination of the transcriptionally inactive K-rasLA2 allele results in the expression of an activated K-rasLA2 allele from the endogenous Kras2 locus. We reasoned that if Kras2 activation was selected for during the process of MYC- or Wnt1-induced tumorigenesis, the presence of cells bearing a spontaneously rearranged—and activated—K-rasLA2 allele would accelerate the tumorigenic process. Therefore, K-rasLA2 mice were bred to MYC- and Wnt1-inducible mice to generate tritransgenic MTB/TOM/K-rasLA2 and MTB/TWNT/K-rasLA2 mice. Oncogene expression was induced in these tritransgenic mice, along with littermate controls, by doxycycline administration.
MTB/TOM/K-rasLA2 tritransgenic mice developed mammary tumors far faster (median latency, 8.6 weeks versus 25.9 weeks; P < 0.0001) (Fig. (Fig.5A5A and Table Table4)4) and had higher tumor multiplicities (2.9 versus 1.2; P = 0.0002) (Table (Table4)4) than control doxycycline-induced MTB/TOM bitransgenic mice. Similarly, MTB/TWNT/K-rasLA2 mice exhibited markedly shorter tumor latencies (5.3 weeks versus 18.3 weeks; P < 0.0001) and higher tumor multiplicities (2.8 versus 1.2; P = 0.0004) than MTB/TWNT controls (Fig. (Fig.5B5B and Table Table4).4). MTB/K-rasLA2 mice treated with doxycycline only rarely developed mammary tumors (Fig. (Fig.5).5). Finally, analysis of genomic DNA from tumors arising in MTB/TOM/K-rasLA2 and MTB/TOM/K-rasLA2 mice indicated that all tumors had rearranged and activated the latent Kras2 allele (data not shown). These data demonstrate that activated Kras2 expressed from its endogenous locus can strongly synergize with either the MYC or Wnt1 pathway in mammary tumorigenesis.
A latent K-rasLA2 allele synergizes with MYC- and Wnt1-induced mammary tumorigenesis. (A) Mammary tumor-free survival curves for doxycycline-induced MTB/TOM/K-rasLA2 mice (n = 30). Results for doxycycline-induced MTB/TOM (n = 43) and MTB/K-rasLA2 (n = 33) littermates are shown as controls. (B) Mammary tumor-free survival curves for doxycycline-induced MTB/TWNT/K-rasLA2 mice (n = 15). Results for doxycycline-induced MTB/TWNT (n = 19) and MTB/K-rasLA2 (n = 33) littermates are shown as controls.
TABLE 4.
A latent K-rasLA2 allele synergizes with c-MYC and Wnt1 in mammary tumorigenesis
| Transgenic line | Median latency (wks) | Average multiplicity (tumors/mouse) |
|---|---|---|
| MTB/TWNT/K-rasLA2 (n = 30)a | 5.3 | 2.8 |
| MTB/TWNT (n = 18) | 18.3 | 1.2 |
| MTB/TOM/K-rasLA2 (n = 41)b | 8.6 | 2.9 |
| MTB/TOM (n = 28) | 25.9 | 1.2 |
Activated Kras2 promotes the progression of Wnt1- and c-MYC-induced mammary tumors to oncogene independence.
If activation of an endogenous Kras2 allele by point mutation were sufficient to confer MYC- or Wnt1-independent growth in mammary tumors, we reasoned that tumors arising in MTB/TOM/K-rasLA2 and MTB/TWNT/K-rasLA2 tritransgenic mice would continue to grow following abrogation of MYC or Wnt1 transgene expression. To test this hypothesis, we removed doxycycline from the drinking water of tumor-bearing tritransgenic MTB/TOM/K-rasLA2 and MTB/TWNT/K-rasLA2 mice and bitransgenic MTB/TOM and MTB/TWNT mice and monitored tumors for regression (Table (Table55).
TABLE 5.
A latent K-rasLA2 allele promotes the progression of MTB/TWNT and MTB/TOM tumors to oncogene independence
| Transgenic line | No. of tumors with indicated status
| |
|---|---|---|
| Full regression | Incomplete regression | |
| MTB/TWNT/K-rasLA2 (n = 30)a | 22 | 8 |
| MTB/TWNT (n = 18) | 18 | 0 |
| MTB/TOM/K-rasLA2 (n = 41)b | 17 | 24 |
| MTB/TOM, Kras2 wild type (n = 10) | 8 | 2 |
Consistent with our prior observations, 0 of 18 tumors arising in MTB/TWNT mice were able to grow in an oncogene-independent manner. In contrast, 27% (8/30) of tumors in MTB/TWNT/K-rasLA2 mice grew in an oncogene-independent manner following downregulation of Wnt1 overexpression (P = 0.016). Thus, Wnt1-induced tumors bearing an activated K-rasLA2 allele were more than sixfold more likely to progress to Wnt1 oncogene independence than those that did not. Nevertheless, the majority of Wnt1-induced tumors bearing activated Kras2 alleles regressed to a nonpalpable state following Wnt1 downregulation. These findings demonstrate that while an activated Kras2 allele is not sufficient for Wnt1-independent growth, it is strongly associated with the progression of tumors to oncogene independence.
Similarly, of MYC-induced tumors in MTB/TOM mice that were wild type for Kras2, 20% (2/10) were able to grow following the downregulation of MYC expression, whereas 59% of tumors (24/41) arising in MTB/TOM/K-rasLA2 mice grew in an oncogene-independent manner following doxycycline withdrawal (P = 0.029). Thus, MYC-induced tumors with an activated K-rasLA2 allele were 5.6-fold more likely to progress to oncogene independence than MYC-induced tumors wild type for Kras2. This indicates that, as observed for Wnt1, expression of an activated Kras2 allele from its endogenous promoter is strongly associated with MYC-independent growth. Notably, however, activation of the K-rasLA2 latent allele in tumors arising in either MTB/TWNT/K-rasLA2 or MTB/TOM/K-rasLA2 mice was not sufficient in and of itself to confer oncogene-independent growth.
Activation of the latent K-rasLA2 allele is not sufficient for high levels of ras or MAPK pathway activation.
A potential explanation for the continued oncogene-dependence of some MTB/TWNT/K-rasLA2 and MTB/TOM/K-rasLA2 tumors, despite having activated the K-rasLA2 allele, is that these tumors do not activate ras or the MAPK pathway to the same extent as MTB/TWNT and MTB/TOM tumors bearing spontaneous activating Kras2 mutations.
Biochemical analysis of biopsies from mammary tumors in MTB/TWNT/K-rasLA2 mice that were subsequently shown to be oncogene dependent following doxycycline withdrawal demonstrated that, despite expressing the activated K-rasLA2 allele from the endogenous locus, these tumors did not universally display the high levels of ras-GTP, phosphorylated Mek1/2, and phosphorylated Erk1/2 seen in MTB/TWNT tumors harboring spontaneous Kras2 mutations (Fig. (Fig.6A).6A). Ras-GTP was undetectable in two of nine tumors from MTB/TWNT/K-rasLA2 mice, and in six of the remaining seven tumors, ras-GTP levels—though elevated—were somewhat lower than those observed in Kras2-mutant MTB/TWNT tumors. Consistent with this, in many tumors the levels of phospho-Mek1/2 and phospho-Erk1/2 were lower than those observed in Kras2-mutant MTB/TWNT tumors.
Ras and MAPK activity in tumors arising in MTB/TWNT/K-rasLA2 mice. MTB/TWNT/K-rasLA2 mice were induced with doxycycline until tumors reached 1 cm3. Tumors were subjected to biopsy, deinduced, and monitored weekly for regression behavior. (A) Immunoblots showing the levels of ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 in biopsy samples from nine tumors from MTB/TWNT/K-rasLA2 mice that were later demonstrated to be Wnt1 dependent following doxycycline withdrawal. An Hras1-mutant MTB/TWNT tumor, a Kras2-mutant MTB/TWNT tumor, and uninduced and induced mammary glands from MTB/TRAS mice were used as controls. β-Tubulin is shown as a loading control. (B) Immunoblots showing the levels of ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 in biopsy samples from seven tumors from MTB/TWNT/K-rasLA2 mice that were later demonstrated to be oncogene independent following doxycycline withdrawal. An Hras1-mutant MTB/TWNT tumor, a Kras2-mutant MTB/TWNT tumor, and uninduced and induced mammary glands from MTB/TRAS mice were used as controls. β-Tubulin is shown as a loading control.
In contrast to the biochemical characteristics of Wnt1-dependent tumors in MTB/TWNT/K-rasLA2 mice, nearly all Wnt1-independent tumors from such mice were found to have ras-GTP levels at least as high as those observed in Kras2-mutant MTB/TWNT tumors (Fig. (Fig.6B).6B). Analogously, biochemical analysis demonstrated that nearly all tumors arising in MTB/TOM/K-rasLA2 mice that were subsequently demonstrated to be oncogene independent following doxycycline withdrawal had high levels of ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 comparable to those found in Kras2-mutant MTB/TOM tumors (Fig. (Fig.77).
Ras and MAPK activity in tumors arising in MTB/TOM/K-rasLA2 mice. Immunoblots showing the levels of ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 in biopsy samples from seven tumors from MTB/TOM/K-rasLA2 mice that were later demonstrated to be oncogene independent following doxycycline withdrawal are presented. An MTB/TOM tumor lacking detectable ras mutations, a Kras2-mutant MTB/TOM tumor, and uninduced and induced mammary glands from MTB/TRAS mice were used as controls. β-Tubulin is shown as a loading control.
These findings indicate that rearrangement and activation of a latent K-rasLA2 allele in MYC- and Wnt1-induced mammary tumors does not uniformly result in the high levels of ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 typically found in MTB/TWNT or MTB/TOM tumors bearing spontaneous Kras2 mutations. This, in turn, implies that genetic alterations in addition to Kras2 mutation may be required for high levels of ras pathway activation and progression of MYC- and Wnt1-induced tumors to oncogene independence. Nevertheless, our observation that virtually all oncogene-independent MTB/TWNT/K-rasLA2 and MTB/TOM/K-rasLA2 tumors displayed levels of activated ras, Mek1/2, and Erk1/2 comparable to those found in oncogene-independent MTB/TWNT and MTB/TOM tumors harboring spontaneous Kras2 mutations supports a model in which high levels of ras pathway activation promote the progression of MYC- and Wnt1-induced tumors to an oncogene-independent state.
High levels of ras activation and oncogene independence are not due to Kras2 amplification in MYC- or Wnt1-induced tumors.
It has previously been reported that amplification of the Kras2 locus can lead to activation of the MAPK pathway (22). We considered this as a potential explanation for the high levels of MAPK pathway activation observed in Kras2-mutant MTB/TOM tumors and for the observation that some tumors with rearranged latent K-rasLA2 alleles exhibited high levels of MAPK pathway activation whereas others did not. Therefore, we assessed Kras2 genomic copy numbers by quantitative PCR to determine whether the Kras2 locus was amplified in MTB/TOM, MTB/TOM/K-rasLA2, or MTB/TWNT/K-rasLA2 tumors.
Of the tumors arising in MTB/TOM mice, all Kras2-mutant tumors examined exhibited high levels of ras-GTP and MAPK pathway activity and were able to grow in an oncogene-independent manner. Nevertheless, Kras2/18S genomic ratios for Kras2-mutant, oncogene-independent MYC-induced tumors did not differ significantly from those observed for MYC-induced tumors that were wild type for ras and oncogene dependent (1.13 versus 1.01; P = 0.447). Similarly, Kras2/18S genomic ratios for MTB/TOM/K-rasLA2 or MTB/TWNT/K-rasLA2 tumors that were oncogene independent or that had high levels of ras-GTP were not significantly different from those of tumors that were oncogene dependent or that had lower levels of ras-GTP (data not shown). These results indicate that in MYC- and Wnt1-induced tumors, either with or without a latent K-rasLA2 allele, amplification of the Kras2 locus does not appear to account for high levels of ras pathway activity or the acquisition of oncogene independence.
DISCUSSION
Our findings demonstrate that MYC-induced mammary tumorigenesis proceeds via a preferred pathway involving Kras2 mutation, whereas Wnt1-induced tumorigenesis proceeds via a preferred pathway involving Hras1 mutation. Furthermore, our observation that MNU treatment strongly synergizes with c-MYC and Wnt1 in mammary tumorigenesis, and significantly alters the frequency and distribution of ras mutations, suggests that ras mutations directly contribute to—and represent a rate-limiting step in—MYC- and Wnt1-induced mammary tumorigenesis. Consistent with this, expression of a latent activated K-rasLA2 allele from the endogenous Kras2 locus strongly synergized with both c-MYC and Wnt in mammary tumor formation. Additionally, we have shown that alterations in the p53 tumor suppressor pathway can alter secondary pathways in tumorigenesis, as Wnt1-induced tumors lacking functional p53 exhibited significantly fewer Hras1 mutations than tumors arising in p53 wild-type mice. In aggregate, our data suggest a model in which c-MYC induction preferentially selects for the outgrowth of cells harboring Kras2 mutations, whereas Wnt1 induction preferentially selects for the outgrowth of cells harboring Hras1 mutations.
The experiments described above suggest that a specific synergistic interaction exists between c-MYC and Kras2, and between Wnt1 and Hras1, during mammary tumorigenesis. However, the finding that Kras2 mutations do arise in Wnt1-induced tumors, and that Wnt1 induction in K-rasLA2 mice dramatically accelerates mammary tumorigenesis and increases tumor multiplicity, suggests that it is possible for Wnt1 to synergize with Kras2 as well as with Hras1 in mammary tumorigenesis. Similarly, Hras1 mutations were found in a substantial fraction of MYC-induced mammary tumors in the setting of carcinogen treatment, providing evidence that—like Wnt1—c-MYC can synergize with Hras1 as well as with Kras2 in mammary tumorigenesis.
Notably, while ras mutation frequently accompanies both MYC- and Wnt1-induced mammary tumorigenesis, the behaviors of these two tumor types following oncogene downregulation were found to be markedly different. Our data show that Kras2 mutation—whether in the context of MYC- or of Wnt1-induced tumorigenesis—is strongly associated with the acquisition of oncogene-independent tumor growth. In contrast, Hras1 mutation was not associated with oncogene-independent growth. Our further observation that ras-GTP, phosphorylated Erk1/2, and phosphorylated Mek1/2 levels were all significantly elevated in tumors harboring Kras2 mutations compared to the levels seen in tumors with Hras1 mutations suggests a biochemical basis for the differential oncogene independence exhibited by Kras2-mutant MYC and Hras1-mutant Wnt1 mammary tumors.
Several in vitro studies have demonstrated differential activation of ras effector pathways by different ras family members. In a variety of cell types, activated Kras2 stimulates the Raf-1/MAPK pathway more strongly than does activated Nras or Hras1 (37, 40). Conversely, activated Hras1 stimulates the PI3K/Akt pathway more strongly than activated Kras2 (6, 40). Overexpression of specific ras isoforms has also been shown to result in differential cellular behavior, in part through activation of different downstream ras effector pathways (6, 26). Consistent with these in vitro experiments, we observed that spontaneous activating Kras2 mutations in MYC- and Wnt1-induced mammary tumors are associated with a level of activation of the Raf-1/MAPK pathway greater than that seen with spontaneous activating mutations in Hras1. Moreover, tumors harboring spontaneous point mutations in Kras2 behave differently than tumors bearing Hras1 mutations, as demonstrated by the ability of Kras2-mutant MYC- and Wnt1-induced mammary tumors to progress to oncogene independence. As such, these models for MYC- and Wnt1-induced mammary tumorigenesis provide a novel in vivo demonstration of differential activation of ras effector pathways depending on the ras isoform mutated.
Notably, a subset of MYC-induced tumors exhibit high levels of Ras-GTP, phospho-Mek1/2, and phospho-Erk1/2 despite harboring wild-type ras alleles (Fig. (Fig.3A3A and data not shown). Similarly, levels of phospho-Mek1/2 and phospho-Erk1/2 were higher in some ras wild-type Wnt1-induced tumors than in Hras1-mutant Wnt1-induced tumors (Fig. (Fig.3B).3B). This raises the possibility that in both MYC- and Wnt1-induced tumors, mechanisms other than activating point mutation in ras likely impinge on ras and the MAPK pathway. These alternative mechanisms could include an increase in Ras-GEF function or a decrease in Ras-GAP function.
To more directly address the contribution of Kras2 activation to oncogene-independent growth, MYC- and Wnt1-inducible mice were bred to mice bearing a latent activated K-rasLA2 allele at the endogenous Kras2 locus. Surprisingly, in contrast to tumors arising in MTB/TOM and MTB/TWNT mice bearing spontaneous activating point mutations in Kras2, a subset of K-rasLA2 tumors remained oncogene dependent despite having recombined and activated the endogenous latent K-rasLA2 allele. Nevertheless, Wnt1- and MYC-induced tumors bearing rearranged and activated latent K-rasLA2 alleles were five- to sixfold more likely to be able to grow in a transgene-independent growth. This suggests that expression of an activated Kras2 allele from its endogenous promoter contributes to, but is not be sufficient for, progression of Wnt1- and MYC-induced tumors to an oncogene-independent state.
Remarkably, analysis of over 600 tumors indicated that once a ras allele has been activated by mutation in a tumor, mutations in other ras family members either do not occur or are not selected for during tumorigenesis. This implies that the specific ras mutation selected for during tumor formation ultimately determines the extent of oncogene independence that will be observed in the resulting tumor. In this model, c-MYC induction in the mammary gland preferentially selects for cells harboring activating Kras2 mutations during tumorigenesis. Kras2 activation, in turn, leads to high levels of MAPK pathway activation that ultimately enable tumors to grow in an oncogene-independent manner following MYC downregulation. In contrast, Wnt1 induction preferentially selects for cells harboring activating Hras1 mutations, which do not result in high levels of ras or MAPK pathway activity. As a consequence, the majority of Wnt1-induced tumors regress to a nonpalpable state following Wnt1 downregulation. This model further predicts that Wnt1 tumors do not become oncogene independent following repeated cycles of Wnt1 induction and deinduction as a consequence of the apparent ability of Hras1 mutation to preclude the subsequent acquisition of mutations in other ras isoforms.
As predicted by this model, a significantly higher fraction of tumors arising in MTB/TOM/K-rasLA2 and MTB/TWNT/K-rasLA2 tritransgenic mice grew in an oncogene-independent manner compared to tumors arising in MTB/TOM and MTB/TWNT littermates. However, a substantial number of MYC and Wnt1 tumors remained oncogene dependent despite having rearranged and activated the latent K-rasLA2 allele. This was unexpected based upon our previous finding that virtually all Kras2-mutant tumors arising in MTB/TOM mice grow in an oncogene-independent manner following MYC downregulation. A potential explanation for this differential behavior may be found in our observation that although each of the tumors in tritransgenic animals had rearranged and activated the latent K-rasLA2 allele, tumors did not uniformly exhibit high levels of activated ras, Mek, and Erk. Potentially, subtle differences in expression may exist between the rearranged latent K-rasLA2 allele and endogenous activated Kras2 alleles that contribute to their differential biochemical activation of ras. An alternate explanation, however, is that mutation and activation of endogenous Kras2 may not be sufficient for high levels of ras or MAPK pathway activation. This, in turn, would suggest that alterations that arise nearly universally in MTB/TOM and MTB/TWNT tumors that spontaneously activate endogenous Kras2 do not occur universally in MYC- and Wnt1-induced mammary tumors bearing a latent K-rasLA2 allele.
Consistent with this inference, it has recently been demonstrated that activation of Kras2 at its endogenous locus is not sufficient to confer high levels of MAPK pathway activation (12, 36). In this regard, reports that wild-type Kras2 may inhibit Kras2 alleles activated by mutation raise the possibility that loss of wild-type Kras2 may contribute to tumor progression and MAPK pathway activation in Kras2-mutant tumors (10, 41). Conversely, other studies have implicated amplification of the Kras2 locus in the upregulation of MAPK pathway activity (22). While our results suggest that amplification of the Kras2 locus does not commonly play a role in ras pathway activation in MYC-induced mammary tumors, they also suggest that alterations in addition to Kras2 mutation are required for tumor cells to exhibit high levels of ras and MAPK pathway activation.
Notably, the origins of the Hras1 and Kras2 mutations found in MYC- and Wnt1-induced mammary tumors remain obscure. One possibility is that MYC or Wnt1 induction results in the outgrowth of rare mammary epithelial cells that harbor spontaneous activating ras mutations. With the expression of an additional oncogene, these cells may have a proliferation or survival advantage over cells lacking such mutations. Alternatively, MYC or Wnt1 induction could lead to molecular changes within mammary cells that predispose them to the acquisition of additional oncogenic mutations. For example, MYC dysregulation has been shown to cause genetic instability in several systems (11, 39). Genetically unstable cells that sustain activating mutations in ras oncogenes may possess a growth advantage over other cells, eventually producing tumors that harbor ras mutations.
Taken in concert, our data demonstrate that the nature of a particular initiating oncogenic event plays a major role in determining the specific secondary genetic alterations that are subsequently selected for during the process of tumorigenesis. These secondary events, in turn, play a major role in determining the likelihood that tumors will progress to oncogene independence. Whether such mechanisms of selection are relevant to human tumorigenesis is as yet unclear. However, the fact that preferentially associated sets of mutations have been identified in human cancers and the fact that a significant number of specific genetic alterations in human breast cancer—including c-MYC amplification—have been correlated with aggressive tumor behavior and poor prognosis suggest that these mechanisms of selection are operative during mammary tumorigenesis in both mice and humans. As such, elucidating the genetic events that contribute to the progression of tumors to oncogene independence may provide insights into the processes by which human breast cancers develop and progress to more aggressive states.
Acknowledgments
This research was supported in part by grants CA92910, CA93719, CA98371, and CA105490 from the National Cancer Institute, grants W81XWH-05-1-0405, DAMD17-03-1-0327 (J.W.J.), and DAMD17-00-1-0397 (R.B.B.) from the U.S. Army Breast Cancer Research Program, and a grant from the Susan G. Komen Breast Cancer Foundation.
We thank C. Sterner and J. Gutnick for assistance with animal husbandry and E. Gunther for contributing mouse tissue and initial observations in synergy between MNU treatment and Wnt1 pathway activation.
Footnotes
▿Published ahead of print on 14 August 2006.