Mutually exclusive mutations of the Pten and ras pathways in skin tumor progression

Abstract

Pten heterozygous (Pten^+/-) mice develop increased papilloma numbers and show decreased carcinoma latency time in comparison with controls after skin treatment with dimethyl benzanthracene (DMBA) and tetradecanoyl-phorbol acetate (TPA). H-ras mutation is normally a hallmark of DMBA-TPA-induced skin tumors, but 70% of carcinomas from Pten^+/- mice do not exhibit this mutation, and in all cases have lost the wild-type Pten allele. Tumors that retain the Pten wild-type allele also have H-ras mutations, indicating that activation of H-ras and complete loss of Pten are mutually exclusive events in skin carcinomas. Mitogen-activated protein kinase (MAPK) is consistently activated in the tumors with H-ras mutations, but is strongly down-regulated in Pten^-/- tumors, suggesting that this pathway is dispensable for skin carcinoma formation. These data have important implications in designing individual therapeutic strategies for the treatment of cancer.

The Pten tumor suppressor gene encodes a dual protein and lipid phosphatase that antagonizes PI3K/Akt signaling, a pathway that triggers a cascade of responses important for tumor development (Vivanco and Sawyers 2002). Somatic alterations that negate Pten function are commonly detected in a number of human tumor types (Li et al. 1997; Steck et al. 1997). In addition, germ-line mutations in Pten have been described in heritable cancer risk syndromes, Cowden disease (CD) and Bannayan-Zonana syndrome (Liaw et al. 1997; Marsh et al. 1997). Homozygous deletion of Pten results in embryonic lethality (Di Cristofano et al. 1998; Stambolic et al. 1998; Podsypanina et al. 1999), indicating that it is a developmentally essential gene. Mice that are heterozygous for Pten are viable, but they show abnormal pathological features and are susceptible to developing tumors in a number of tissues (Di Cristofano et al. 1998; Podsypanina et al. 1999). The skin of these mice shows focally increased epidermal thickness due to hyperkeratosis (Di Cristofano et al. 1998), a feature commonly observed with CD patients. The association of squamous cell carcinomas with CD has also been reported (Camisa et al. 1984; Hildenbrand et al. 2001), but these lesions were not observed in Pten^+/- animals. On the other hand, 100% of mice with complete Pten deficiency, and a proportion with Pten heterozygosity, in their keratinocytes spontaneously develop papillomas and squamous cell carcinomas (Suzuki et al. 2003), indicating that Pten is an important suppressor of skin tumorigenesis.

We have used dimethyl benzanthracene (DMBA)-tetradecanoyl-phorbol acetate (TPA)-induced skin carcinogenesis to study the molecular mechanisms that underlie the development of this malignancy. A signature of DMBA treatment is an activating mutation in H-ras, which is the most common initiating event for tumor formation (Balmain et al. 1984; Quintanilla et al. 1986). We demonstrate here that most skin carcinomas from Pten^+/- mice do not have this signature mutation, but instead show complete loss of Pten by deletion of the wild-type allele. These genetic changes are correlated with major differences in signaling through the mitogen-activated protein kinase (MAPK) and PI3K/Akt pathways.

Results and Discussion

To investigate the role of Pten in skin carcinogenesis, we treated an equal number of Pten^+/- mice and wild-type littermates with DMBA and TPA. We used germ-line Pten^+/- mice (Di Cristofano et al. 1998), as opposed to tissue-specific knockout (KO) models (Suzuki et al. 2003) to better recapitulate the condition of CD. The basal level of Pten protein is clearly reduced in the skin of these heterozygous animals (Fig. 1A), and this is relevant given that Pten expression is rapidly up-regulated, both at the RNA and protein levels, in response to TPA treatment (Fig. 1B). The increase in levels of a tumor suppressor protein such as Pten is intriguing in view of the fact that TPA treatment strongly induces proliferation in the epidermis. A similar situation is, however, seen with other negative growth regulators such as TGFβ (Akhurst et al. 1988), which are also induced in response to TPA treatment, possibly as part of a mechanism that ensures a regulated return to homeostatic equilibrium within the epidermis. In light of these observations, we predicted Pten^+/- mice to be more susceptible to DMBA-TPA-induced skin carcinogenesis, although the risk of spontaneously developing skin cancer was not noticeably altered (Di Cristofano et al. 1998; Podsypanina et al. 1999). Both Pten^+/- and wild-type mice developed papillomas after 7–8 wk of DMBA-TPA treatment. However, the papilloma number was significantly higher in Pten^+/- animals in comparison with wild type in both males and females (Fig. 1C). By 20 wk posttreatment, the mean number of papillomas per mouse was more than three-fold higher in Pten^+/- mice (p < 0.0001), indicating that loss of one Pten allele renders animals more susceptible to DMBA-TPA-induced skin carcinogenesis.

Figure 1.

Pten heterozygosity increases skin cancer susceptibility. (A) Pten protein expression in the skin of wild-type and Pten^+/- mice. (B) Expression of Pten is induced in wild-type mice at RNA (top) and protein (bottom) levels at indicated time points after TPA treatment. Pten protein levels in skin of two different mice are shown at each time point. (C) Average papilloma number per mouse at the indicated time points after DMBA-TPA treatment. (D) Latency of carcinoma onset.

Papillomas are generally benign, but a small number eventually progresses to the carcinoma stage. Mice were monitored for carcinoma development for up to 60 wk post-DMBA treatment. We observed carcinomas on Pten^+/- mice as early as 20 wk post-DMBA treatment (Fig. 1D). By 44 wk posttreatment, carcinomas were found on all Pten^+/- animals, whereas >70% of the wild-type littermates remained carcinoma free. The decrease in carcinoma latency period in Pten^+/- versus wild-type mice was highly significant (p < 0.005), indicating that an impartial loss of Pten function not only increased susceptibility to tumor initiation, but also accelerated tumor progression.

A two-hit mechanism to inactivate tumor suppressor genes has been well documented, and Pten appears to follow this paradigm in many tumor types. We examined the status of Pten in skin tumors using microsatellite markers at, and also proximal and distal to, the Pten genomic locus. In 15 papillomas and 12 carcinomas from wild-type mice, we did not detect any deletions at the Pten locus. In contrast, loss of heterozygosity (LOH) was observed in papillomas and a high number of carcinomas from Pten^+/- mice (Fig. 2A). In many instances, LOH extends to markers as far as 20 cM from the Pten locus, and may reflect a loss of the entire chromosome 19. However, LOH at adjacent markers did not occur independently of LOH at the Pten locus, indicating that Pten is the primary target for LOH in these tumors.

Figure 2.

H-ras and Pten alterations are mutually exclusive. (A) LOH analysis of tumors from Pten^+/- mice using the indicated microsatellite markers on chromosome 19, as obtained from the Celera database. Papilloma #16 (fifth from the right) shows loss of distal markers on chromosome 19, but is wild type at Pten, and has a mutation in H-ras (Table 1). (B) Papillomas and carcinomas from wild type (top panel) and Pten^+/- (bottom panel) were analyzed for H-ras mutation by digestion of PCR products as described in text, and for Pten expression by RT–PCR. Asterisk indicates samples with TTA alteration, as opposed to the more common CTA, at codon 61 of H-ras, and hence was not digested by XbaI. (C) Chromatogram showing the CAA → TTA alteration at codon 61 of H-ras that occurred in a small number of papillomas.

Loss of expression due to regulatory mutations or epigenetic events is one mechanism of inactivating gene activity. However, RT–PCR analysis indicated that the Pten transcript is present in all papillomas and carcinomas from wild-type animals (Fig. 2B). Pten protein is also clearly detectable in these papillomas and carcinomas (Fig. 3A). We sequenced the entire coding region of Pten in these tumors and did not detect any mutation. In papillomas and carcinomas from Pten^+/- animals, the presence or absence of the Pten wild-type transcript correlated with Pten protein status, and these were in agreement with the microsatellite results. In the majority of carcinoma samples, the wild-type Pten transcript is clearly absent (Fig. 2B). On the other hand, the transcript corresponding to the KO allele is retained in all papillomas and carcinomas, indicating that LOH selectively targets the chromosome that carries the wild-type Pten allele.

Figure 3.

H-ras activation and Pten inactivation elicit differential biochemical responses. The status of pten for each tumor, as determined by microsatellite and RT–PCR analysis, is indicated. (A) Pten and H-Ras protein levels in papillomas and carcinomas. (B) Separation of wild-type and mutant (Gln61Leu) H-Ras protein in carcinomas by high-percentage SDS-PAGE. (C) Analysis of Erk and Akt activation in papillomas and carcinomas using antibodies to detect phosphorylated and total protein, as indicated.

The incidence of Pten LOH in papillomas was 25% (5/20), which is significantly lower than the 66% (6/9) we observed for carcinomas (p < 0.05), indicating that a complete loss of Pten efficiently drives the conversion of benign papillomas to malignant carcinomas. The concept of subtypes of papillomas with high or low malignant potential has been previously described (Hennings et al. 1985), with those having the higher potential for malignant conversion originating from hair follicle stem cells (Perez-Losada and Balmain 2003). One interpretation of our data is that Pten may be important in regulating epidermal stem cell proliferation, similar to that described for neural stem cells (Groszer et al. 2001). In this respect, a complete loss of Pten could result in an increase in the number of hair follicle stem cells and thus the number of potential targets for malignant skin tumor development. In support of this idea, animals with keratinocyte-specific Pten deficiency do indeed have increased hair follicles and decreased interfollicular epidermis (Suzuki et al. 2003). Alternatively, complete loss of Pten may be able to drive progression regardless of the cell of origin.

A lack of Pten alterations in tumors from wild-type mice was a surprising finding because homozygous inactivation of Pten in keratinocytes can clearly cause skin cancer (Suzuki et al. 2003). On the other hand, our data are consistent with findings in human skin tumors, where a lack of somatic Pten alterations has also been reported (Kubo et al. 1999). It is possible that inactivation of both Pten alleles is required during skin carcinogenesis, a situation that is more favorable when the first event has already occurred through germ-line transmission, as is the case with CD and Pten^+/- mice. In this respect, genetic alterations at other loci may substitute for Pten inactivation in tumors from wild-type mice and may even preclude the need for Pten LOH in some tumors from Pten^+/- animals.

Activating mutations in H-ras, usually a CAA-to-CTA change at codon 61, occur at a high frequency in DMBA-initiated skin tumors (Balmain et al. 1984; Quintanilla et al. 1986). H-Ras has a stimulatory effect on PI3K/Akt signaling; thus its activation could occur in place of Pten inactivation. We analyzed skin tumor DNA for H-ras-activating mutations, taking advantage of the XbaI restriction site created by the CTA mutation (Nagase et al. 2003). Activating H-ras mutations were detected in all 10 carcinomas from wild-type animals (Fig. 2B). In two papillomas where the mutant product was not detected, DNA sequencing revealed a TTA alteration at codon 61 in place of the more common CTA (Fig. 2C). Interestingly, both CTA and TTA alterations resulted in the same Gln61Leu amino acid change, pointing to importance and selectivity of this residue in activated H-Ras protein. Taken together, our data showed that the codon 61-activating H-ras mutation occurs in every DMBA-TPA-induced papilloma and carcinoma from wild-type mice.

In contrast, the majority of skin carcinomas derived from Pten^+/- mice did not carry this H-ras mutation, and all of the tumors in this category had undergone Pten LOH. We did not detect any mutations in H-ras despite sequencing the entire coding region. Because mutations in K-ras have been found in skin tumors from DMBA-TPA treated H-ras KO mice (Ise et al. 2000), we also sequenced the entire coding region of K-ras in these carcinomas and did not detect any mutation. There is therefore a highly significant correlation between the absence of ras mutations in carcinomas and the loss of the wild-type Pten allele. Table 1 summarizes results of Pten and H-ras status in papillomas and carcinomas from wild-type and Pten^+/- mice. It can be seen that although the majority of the benign papillomas in Pten^+/- mice still had activated H-ras in the usual manner, ∼30% had lost the Pten allele. Of these, two had both activated H-ras and loss of Pten, in contrast to the situation seen in the carcinomas. This may indicate that papillomas carrying both mutations are less likely to progress to malignancy than those showing complete loss of Pten.

Table 1.

Summary of H-ras and pten alterations

Pten(+/-)	H-ras	Pten	Pten(+/+)	H-ras	Pten
Papilloma			Papilloma
1	MUTa	WT	1	MUT	WT
2	MUT	WT	2	MUTa	WT
3	MUT	WT	3	MUT	WT
4	MUT	WT	4	MUT	WT
5	MUT	LOSS	5	MUT	WT
6	MUT	WT	6	MUT	WT
7	WT	LOSS	7	MUT	WT
8	MUT	LOSS	8	MUT	WT
9	MUT	WT	9	MUT	WT
10	MUT	WT	10	MUT	WT
11	MUT	WT	11	MUT	WT
12	WT	LOSS	12	MUT	WT
13	MUT	WT	13	MUT	WT
14	MUT	WT	14	MUTa	WT
15	MUT	WT	15	MUT	WT
16	MUT	WT
17	MUT	WT	Carcinoma
18	WT	LOSS	1	MUT	WT
19	MUT	WT	2	MUT	WT
20	MUT	WT	3	MUT	WT
			4	MUT	WT
Carcinoma			5	MUT	WT
1	MUT	WT	6	MUT	WT
2	WT	LOSS	7	MUT	WT
3	WT	LOSS	8	MUT	WT
4	WT	LOSS	9	MUT	WT
5	MUT	WT	10	MUT	WT
6	WT	LOSS
7	WT	LOSS
8	MUT	WT
9	WT	LOSS

(WT) Wild type; (MUT) mutant.

^aCodon 61 mutation CAA → TTA

The strong up-regulation in H-ras protein level in carcinoma demonstrates the importance of this gene in papilloma-to-carcinoma conversion (Fig. 3A). The mechanism underlying this increase in expression appears to be at the level of gene copy number (Nagase et al. 2003). By using high-percentage polyacrylamide SDS gels to discriminate between mutant and wild-type H-Ras protein by mobility, we could show that it is the oncogenic form of H-Ras protein that is selectively up-regulated in carcinomas (Fig. 3B). In addition, the extent of up-regulation is stronger in tumors with intact Pten than in those that are Pten heterozygous. In carcinomas with complete Pten inactivation, not only is there a lack of H-ras-activating mutation, but there is also no up-regulation at the expression level. On the basis of these results, it appears that Pten inactivation confers similar advantages to activated H-ras and its subsequent overexpression, at least at the stage of carcinomas. This could account for the decreased carcinoma latency time in Pten^+/- mice because LOH involving the wild-type Pten allele would be the single limiting event, whereas with H-ras there is a requirement for mutagenesis as well as allele-specific duplication or amplification (Nagase et al. 2003).

We analyzed the status of downstream targets of H-Ras and Pten in order to better understand the biochemical consequences of their alterations. Several lines of evidence suggest that the activation of Erk-1 and Erk-2 MAPK proteins, which are important components of the H-Ras network, can also be dependent on the status of Pten (Shan et al. 2000; Suzuki et al. 2001, 2003). Despite relatively similar levels of total Erk proteins, the phosphorylated or active forms of both Erk proteins (P-Erk) were only detectable in papillomas or carcinomas that do not have complete Pten inactivation (Fig. 3C) and are most likely a direct consequence of the oncogenic H-Ras protein present in these tumors (see earlier). Even in these cases, the level of P-Erk proteins was barely detectable in papillomas. The Erk proteins are more strongly activated in carcinomas, which are consistent with the up-regulated mutant H-Ras expression in these malignant tumors, but the extent of activation can be variable. The almost complete lack of P-Erk in papillomas and carcinomas with complete Pten inactivation may be a signature of these tumors, and suggests that Erk activation is dispensable in skin carcinogenesis or at least during papilloma-to-carcinoma conversion.

Central to the tumor-suppressing function of Pten is the inhibition of Akt signaling (Stambolic et al. 1998). In contrast, H-Ras activates PI3K activity, which in turn stimulates the Akt pathway (Cox and Der 2003). We observed an increasing trend in levels of phosphorylated Akt (P-Akt) from papillomas to carcinomas as previously reported (Fig. 3C; Segrelles et al. 2002), suggesting that Akt activation is important for carcinoma development and consistent with the role of Akt signaling in promoting cell proliferation and survival (Vivanco and Sawyers 2002). Interestingly, whereas many studies with cultured cells have shown regulation of Akt signaling in response to Pten alterations to be largely at the level of phosphorylation (Stambolic et al. 1998; Suzuki et al. 2003), we found that in papilloma and particularly carcinoma specimens the regulation appears to occur at the level of total Akt protein as well. Tumors with retention of the wild-type Pten allele and carrying H-ras mutations in general had higher levels of total Akt protein. It is possible that these higher levels may be a consequence of the activity of Ras/MAP in inducing cell death (Deng et al. 2004), causing a feedback response leading to up-regulation of Akt-mediated survival signals. A dynamic balance between these opposing pathways may explain why no tumors carrying both genetic alterations have progressed to malignancy, whereas those with only Pten loss, and no MAPK activity, have high malignant potential. These results may be relevant to prognosis of human cancer carrying both kinds of genetic changes, which may be expected to have a more indolent growth pattern.

A lack of concurrent Pten alteration and N-ras or K-ras mutation has been reported for melanoma cell lines (Tsao et al. 2000) and endometrial cancer (Ikeda et al. 2000), respectively. However, the relationship between these mutational events is still unclear, and some melanomas or uncultured cells have been reported to have both mutations in B-raf and loss of Pten (Tsao et al. 2004). In this study, we have excluded the possibility that the common sites of activation of K-ras or B-raf (Davies et al. 2002) are mutated in the tumors that show complete loss of Pten (data not shown). We conclude that Pten inactivation and H-ras activation are mutually exclusive in malignant skin carcinomas, suggesting that these events are genetically redundant, and that alteration in both genes does not confer a further advantage. It is also possible that H-ras activation and Pten loss is in fact disadvantageous when these events occur together in the same cell. We detected two papillomas with both H-ras mutation and Pten inactivation, but no carcinomas with both changes. Although these numbers are small, papillomas with both genetic alterations are possibly at a relative disadvantage for tumor progression to carcinomas, possibly due to reported effects of activated ras in induction of apoptosis (Chang et al. 2003) or growth arrest (Serrano et al. 1997).

These observations of discrete pathways leading to malignancy in this model system have important implications for the design of therapeutic strategies for the treatment of skin cancer as well as other cancers with frequent Pten and/or ras alterations. Tumors with complete loss of Pten and low levels of signaling through the Ras/Raf/Erk pathway will obviously not respond to small molecule drugs that inhibit these targets. Such treatments may even be detrimental, because they may alter the balance between cell death and survival pathways in cancers carrying both mutations.

Materials and methods

Mice and chemical carcinogenesis

Pten^+/- mice (C57BL/6), provided by Dr. P.P. Pandolfi (Memorial Sloan-Kettering Cancer Center, New York; Di Cristofano et al. 1998), were crossed with FVB/N to generate F1 wild-type and Pten^+/- mice. To induce skin carcinogenesis, we used a single dose of DMBA treatment followed by twice-weekly application of TPA for 20 wk as described (Balmain et al. 1984; Quintanilla et al. 1986). Papilloma number was recorded up to 20 wk and carcinoma development monitored up to 60 wk post-DMBA treatment. Mice were killed if moribund, if the tumor load was excessive, if any individual tumor exceeded 1.5 cm in diameter, or at the termination of the experiment. Surgically removed tumors were immediately snap-frozen in liquid nitrogen, and stored at -70°C. We scored tumors as carcinomas or papillomas on the basis of their morphological appearance and histological features.

DNA/RNA and protein preparation

Tumors were ground into a fine powder in liquid nitrogen and aliquotted for DNA and RNA extraction and protein lysate preparation. DNA was isolated using standard phenol/chloroform extraction following an overnight 55°C incubation with Proteinase K. Total RNA was purified using TRIzol Reagent (Invitrogen). For total protein lysates, powdered tumors were incubated in ice-cold lysis buffer (50 mM Tris-Cl at pH 7.4, 2 mM EDTA, 150 mM NaCl, 1% NP-40, 0.5% Triton X-100, 0.1% SDS) supplemented with Complete Mini Protease Inhibitor Cocktail (Roche). Soluble protein lysates were quantified using BCA Protein Assay kit (Pierce).

LOH analysis

Tumor and corresponding normal DNA was used for LOH studies with microsatellite markers at the Pten locus and flanking regions chosen from the Celera database. PCR amplification was performed in 20-μL volume containing 1× PCR buffer (Bioline), 200 μM of each dNTP (Pharmacia), 6.6 μM of each primer (Qiagen Operon), 1.5 mM MgCl₂, 2 units of Taq polymerase (Bioline), and 80 ng of DNA. Amplifications were initially denatured for 3 min at 94°C, followed by 35 cycles of 30 sec at 94°C, 30 sec at 55°C or 52°C, and 30 sec at 72°C. PCR products were electrophoresed in 4% (3% NuSieve/1% normal) agarose gel and visualized by ethidium bromide staining.

Pten expression analysis

RT–PCR was performed using Thermoscript RT–PCR (GIBCO-BRL). Mutation analysis of the Pten coding sequence was performed by direct sequencing of gel-purified PCR products of tumor cDNA. Primer pairs used in the analysis were Pten1F (5′-CATCTCTCTCCTCCTTTTTCT TCA-3′) and Pten1R (5′-CAAACATCATCTTGTGAAACAGC-3′), Pten2F (5′-GGCAAATTTTTAAAGGCACAAG-3′) and Pten2R (5′-AAGTTGA ACTGCTAGCCTCTGG-3′), and Pten3F (5′-ACTTTGAGTTCCCTCA GCCA-3′) and Pten3R (5′-TTCATGGTATTTTATCCCTCTTGA-3′).

Immunoblotting

Equal amounts of proteins were resolved by SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membrane using standard protocol. We used 18% polyacrylamide gel in order to distinguish between wild-type and codon 61 H-Ras mutant protein. Rabbit polyclonal antibody to phospho-Ser 473–PKB/AKT was a gift from Dr. David Stokoe (University of California at San Francisco). We purchased Pten (A2B1), H-Ras (C-20), and ERK1 (C-16) antibodies from Santa Cruz Biotechnology, and P-Erk (#9101) and Akt (#9272) antibodies from Cell Signaling Technology.

Statistical analysis

SPSS statistical package was used for all statistical analysis. We used the nonparametric test to compare papilloma number and the Kaplan-Meire method to compare latency in progression to carcinoma.