Evolution of somatic mutations in mammary tumors in transgenic mice is influenced by the inherited genotype

Background MMTV-Wnt1 transgenic mice develop mammary hyperplasia early in development, followed by the appearance of solitary mammary tumors with a high proportion of cells expressing early lineage markers and many myoepithelial cells. The occurrence of tumors is accelerated in experiments that activate FGF proto-oncogenes or remove the tumor suppressor genes Pten or P53, implying that secondary oncogenic events are required for progression from mammary hyperplasia to carcinoma. It is not known, however, which oncogenic pathways contribute to Wnt1-induced tumorigenesis – further experimental manipulation of these mice is needed. Secondary events also appear to be required for mammary tumorigenesis in MMTV-Neu transgenic mice because the transgene in the tumors usually contains an acquired mutation that activates the Neu protein-tyrosine kinase. Methods cDNA or DNA from the mammary glands and mammary tumors from MMTV-Wnt1, MMTV-Wnt1/p53-/-, MMTV-Neu transgenic mice, and newly generated MMTV-Wnt1/MMTV-Neu bitransgenic mice, was sequenced to seek activating mutations in H-Ras, K-Ras, and N-Ras genes, or in the MMTV-Neu transgene. In addition, tumors from bitransgenic animals were examined to determine the cellular phenotype. Results We found activating mutations at codons 12, 13, and 61 of H-Ras in just over half of the mammary tumors in MMTV-Wnt1 transgenic mice, and we confirmed the high frequency of activating mutations of Neu in tumors in MMTV-Neu transgenic mice. Tumors appeared earlier in bitransgenic MMTV-Wnt1/MMTV-Neu mice, but no Ras or MMTV-Neu mutations were found in these tumors, which were phenotypically similar to those arising in MMTV-Wnt1 mice. In addition, no Ras mutations were found in the mammary tumors that arise in MMTV-Wnt1 transgenic mice lacking an intact P53 gene. Conclusions Tumorigenic properties of cells undergoing functionally significant secondary mutations in H-Ras or the MMTV-Neu transgene allow selection of those cells in MMTV-Wnt1 and MMTV-Neu transgenic mice, respectively. Alternative sources of oncogenic potential, such as a second transgenic oncogene or deficiency of a tumor suppressor gene, can obviate the selective power of those secondary mutations. These observations are consistent with the notion that somatic evolution of mouse mammary tumors is influenced by the specific nature of the inherited cancer-promoting genotype.


Background
Several transgenic mouse models have been generated in efforts to identify or confirm genes that can participate in breast carcinogenesis and to study the pathogenic process (for review, see [1]). In general, these models have revealed that a wide variety of proto-oncogenes can initiate mammary tumorigenesis when expressed under the control of an appropriate promoter (usually the mouse mammary tumor virus long terminal repeat (MMTV LTR) or the whey acidic protein (WAP) transcriptional control domain). They have also revealed that tumorigenesis is a stochastic process, resulting in the appearance of solitary tumors in one or a few of the 10 mammary glands several months after birth, and that secondary mutational events, inherited deficiencies in tumor suppressor genes, or a combination of transgenes can accelerate the onset of tumor growth. In addition, persistent tumor growth generally requires the continued expression of the initiating oncogene, unless certain secondary mutations have occurred to render the tumor independent of the transgenic oncogene [2][3][4].
Despite these generalizations, the characteristics of mammary tumors arising in transgenic mice differ in a fashion characteristic of the initiating oncogene. For example, the expression profiles of tumors reflect the transgenic oncogene [5]; moreover, those tumors induced by genes that activate the Ras signaling pathway (for example, polyoma middle T antigen, ErbB2/Neu, and H-Ras) are more similar to each other than to tumors induced by components of the Wnt signaling pathway (Wnt1, β-catenin, and c-Myc) with respect both to gene expression patterns ( [5], S. Huang et al., in preparation) and to cellular composition and histology [6][7][8][9]. In tumors induced by components of the Wnt signaling pathway, a high proportion of cells produce proteins (Sca-1, Keratin 6) associated with early stages of mammary development, whereas few such cells are observed in tumors induced by activators of Ras signaling [9]. In addition, the presence of many tumor cells with myoepithelial features (for example, expression of smooth muscle actin) and other observations suggest that in the Wnt pathway oncogenes transform cells positioned early in the developmental program for mammary tissue, whereas oncogenes like ErbB2/Neu that activate Ras proteins, either transform cells that are more advanced developmentally or drive cells to a more mature epithelial character during oncogenesis.
Efforts to understand these complex pathogenic events must take into consideration the secondary oncogenic events that are presumed to be required to convert one or a few of the many mammary cells expressing a transgenic oncogene into a frankly tumorigenic cell. Several findings offer clues to the nature of such secondary events. For example, the appearance of mammary tumors in MMTV-Wnt1 transgenic mice can be accelerated by a second transgene (MMTV-Fgf3, [10]), by proviral insertion mutations in Fgf genes [11][12][13], and by inherited deficiencies of P53 or Pten [14,15]. Similarly, tumors appear earlier in MMTV-c-Myc/MMTV-v-H-Ras bitransgenic mice than in monotransgenic mice [16], and, in c-Myc transgenic mice, tumors that do not require continued expression of c-Myc usually contain somatically mutated K-Ras2 or N-Ras genes [3]. When tumors occur in animals that inherit a transgene encoding normal ErbB2/Neu protein, the tumors usually exhibit a secondary somatic mutation of ErbB2/Neu that stimulates the protein-tyrosine kinase activity of the gene product [17].
In this report, we have attempted to identify additional somatic events that can promote progression to a tumorigenic phenotype in MMTV-Wnt1 transgenic mice by seeking acquired mutations in Ras. In addition, we have asked whether accelerating factors, such as inheritance of a second transgenic oncogene, MMTV-Neu, or deficiency in a tumor suppressor gene, p53, diminishes selection for cells that have acquired potentially oncogenic mutations. In the course of these studies, we were also able to determine that the phenotypic properties of tumors arising in MMTV-Wnt1/MMTV-Neu bitransgenic mice resemble those of tumors induced by Wnt rather than Ras signaling.

RNA, DNA, and protein extraction from tumors
Mouse mammary tumors were collected at the time of necropsy or tumor extraction surgery and snap frozen in liquid nitrogen. Frozen tumors were ground in liquid nitrogen and the resulting powder was placed in three tubes containing appropriate buffers. For RNA extraction, about 50 mg of tumor powder was dissolved in Trizol (#15596-018; Invitrogen, Carlsbad, CA, USA), followed by phenol-chloroform extraction and ethanol precipitation. For DNA extraction, about 50 mg of tumor powder was digested overnight with 2 mg/ml Proteinase K (#1 373 200; Roche, Indianapolis, IN, USA) in 20 mM Tris, 200 mM NaCl, 5 mM EDTA, 0.2% SDS, pH 8.5 buffer, followed by phenol-chloroform extraction and ethanol precipitation. For protein extraction, about 50 mg of tumor powder was dissolved in 20 mM Tris, 150 mM NaCl, 1% Triton X, 2 mM EDTA, 1x protease inhibitor cocktail (complete tablets, #1 697 498, Roche), 1 mM Na 3 VO 4 , 40 mM NaF, incubated on ice for up to 30 min and cleared by centrifugation.
Some DNA samples from MMTV-Wnt1 tumors with wild type, heterozygous or null p53 allele (generated by L. Donehower [18]) on a mixed genetic background were a gift from Larry Donehower (Baylor College of Medicine).

Ras and MMTV-Neu cDNA synthesis
Extracted RNA was copied with gene-specific primers to make cDNA, which was then amplified with the appropriate primers using the SuperScript™ One-Step RT-PCR Kit (#10928-034; Invitrogen) according to manufacturer's instructions. For mouse K-Ras and N-Ras amplification, primer sequences and cycling conditions were kindly provided by Lewis Chodosh and Robert Boxer (University of Pennsylvania School of Medicine). Mouse H-Ras codons 3-99 were amplified either with primers generated based on the sequences provided by Chodosh and Boxer, or with an additional reverse primer spanning exons 2 and 3: HARAS.B2: 5'GATCTGCTCCCTGTACTGATGG3'.
MMTV-Neu transgene cDNA was amplified with primers AB2913 and AB1310 [17], which amplify the region corresponding to nucleotides 1492-2117 of rat Neu cDNA.

Sequencing
All sequencing was carried out by the Memorial Sloan-Kettering Cancer Center (MSKCC) Sequencing Core Facility. Every cDNA was sequenced with both forward and reverse primers used for amplification, and some samples were also sequenced with the nested reverse primer. Sequencing peaks were inspected for overall integrity and legibility. The sample was scored as mutant when heterozygous peaks at positions 12,13, 59, and 61 were at least the same height as the wild type peaks in at least one sequencing direction. Sequences were also downloaded into Vector NTI 5.2.5 (InforMax, Inc., Frederick, MD, USA) program and aligned on the Baylor College of Medicine Search Launcher [19] against the sequences deposited in GenBank (mouse H-Ras: nm008284; mouse K-Ras: xm110615; mouse N-Ras: nm010937, rat Neu: X03362), to detect mutations (or lack of thereof) in the transcripts.

Activating mutations in H-Ras occur in about half of mammary tumors in MMTV-Wnt1 mice
Wnt1-induced mammary tumors contain elevated levels of c-Myc RNA and protein, as expected from stimulation of the Wnt signaling pathway [20]. Since an H-Ras transgene can hasten tumorigenesis when combined with a c-Myc transgene [16] and since K-Ras or N-Ras is frequently mutated in c-Myc-induced tumors [3], we decided to look for secondary somatic mutations in mammary tumors arising in MMTV-Wnt1 transgenic mice by sequencing cDNA copies of Ras mRNAs in the tumors. Mutations were scored only if approximately half of the resulting amplified DNA had the same mutation, implying growth selection of an oncogenic cell in which the observed mutation occurred.  Primary tumor cDNA was amplified and sequenced with H-Ras-specific primers at least once in each direction as described in Methods. Products containing mutant peaks equal to or greater in height than the wild type peaks in at least one sequencing direction were scored as mutationpositive. Mutations in codons 12,13, 59, and 61 were scored as activating mutations.

Mutations in the H-Ras gene
We initially sought activating Ras mutations by sequencing codons 3-99 of H-Ras, 1-98 of N-Ras, and 1-94 of K-Ras. All of the sequences contain codons 12, 13, 59, and 61 -sites of the mutations most commonly associated with oncogenic Ras proteins. We found mutations in codons 12 and 61 of H-Ras, but not in those of N-Ras or K-Ras in cDNA from the first 10 tumor samples (Figure 1). We subsequently focused on the H-Ras cDNA and found activating mutations in codons 12,13, or 61 in 24 of an additional 36 primary mammary carcinomas (Table 1). We then examined 20 tumors without activating mutations in H-Ras for activating point mutations in K-Ras and N-Ras, but failed to find mutations in these two genes. In addition, we were unable to detect mutant H-Ras cDNA made from RNA extracted from hyperplastic mammary glands from MMTV-Wnt1 transgenic mice (n = 6) or from virgin or lactating mammary glands from non-transgenic mice (n = 6), implying that, if H-Ras mutations occurred before the appearance of primary tumors, an insufficient number of cells in the glands harbored the mutations to allow detection, and that selection during tumorigenic growth would be required.
We made a preliminary effort to define the selectable trait conferred on mammary cells in MMTV-Wnt1 transgenic mice by an activating mutation in H-Ras. Tumors in these mice arise after a latency of similar length, have similar growth characteristics, and metastasize to the lungs with similar frequencies regardless of whether they do or do not contain a mutant H-Ras gene (data not shown).
These findings may simply mean that any requirement for activation of the Ras pathway in Wnt1-induced tumors may be satisfied by some other means, such as other oncogenic mutations; this issue is further addressed in a later section.
We have also used biochemical methods to look for the consequences of the H-Ras mutations. For example, the oncogenic activity of mutant Ras proteins in mouse cells is dependent on the Raf kinase pathway, which phosphorylates and activates the ERK1/p44 and ERK2/p42 mitogen-activated protein kinases (MAPKs) [21]. We have compared levels of phospho-ERK in Wnt1-induced tumors with mutant and wild type H-Ras genes by protein blotting and immunohistochemistry (

Mammary tumorigenesis is accelerated in MMTV-Wnt1/ MMTV-Neu bitransgenic mice
Her-2/Neu acts in part through the Ras signaling pathway [22][23][24], a feature that may account for the remarkable similarity between the phenotypes of mammary tumors found in MMTV-Neu and MMTV-v-H-Ras transgenic mice [25]. These observations suggest that an inherited Neu transgene might mimic the effects of somatic mutation of a Ras gene. If so, tumors might arise earlier in MMTV-Wnt1/MMTV-Neu bitransgenic mice than in mice carrying a single transgene, and any cells that acquired H-Ras mutations would not have a selective growth advantage; hence H-Ras mutations would not be detected in our tests of tumor genotypes. In addition, with bitransgenic mice, we could ask whether a co-existing MMTV-Wnt1 transgene eliminated the selective advantage conferred by an activating mutation in the coding domain of the MMTV-Neu transgene, and we could ask whether the phenotype of tumor cells in resulting tumors resembled the phenotype of tumors induced by components of the Wnt pathway, the Neu/Ras pathway, or both.
Therefore we crossed MMTV-Wnt1 transgenic mice to a MMTV-Neu transgenic line that carries a cDNA encoding normal rat ErbB2/Neu protein [26].

Tumors in the MMTV-Wnt1/MMTV-Neu bitransgenic mice are morphologically similar to Wnt1-induced tumors, despite expression of the Neu transgene
MMTV-Wnt1-induced tumors contain multiple mammary cell types, and many cells express early lineage markers, such as Sca-1 and keratin-6 [9]. In contrast, MMTV-Neuinduced mammary tumors are composed nearly exclusively of luminal epithelial cells and have a low proportion of Sca-1-and keratin-6-positive cells [9]. Remarkably, all tumors from bitransgenic mice were similar to the Wnt1-induced tumors, and not to the Neu-induced ones ( Figure 4A). We detected multiple cell types within these tumors, including smooth muscle actin-positive myoepithelial cells, keratin-8-positive luminal epithelial cells, and keratin 6-positive cells ( Figure 4B, panels a-c). Expression of the MMTV-Neu transgene was observed in the luminal epithelial cells from these tumors by immunohistochemistry ( Figure 4B, panel d) and by an RT-PCR assay with transgene-specific primers ( Figure 4C).

Activating H-Ras mutations are not found in Wnt1induced tumors arising in p53 null mice
We have previously shown that deficiency of p53 dramat-ically accelerates the appearance of mammary tumors in MMTV-Wnt1 transgenic mice [14]. acquire Wnt1 independence in a p53-deficient background [2]. We therefore asked whether we could detect H-Ras mutations in mammary tumors induced by an MMTV-Wnt1 transgene in a p53-null background; a failure to find such mutations would imply either that p53 is somehow required for such mutations to occur or, more likely, that an inherited deficiency of p53 conferred a growth advantage to Wnt1-expressing mammary cells that would outweigh any selective advantage provided by a secondary somatic mutation in H-Ras.
To explore this issue, we examined cDNA or DNA from 16 tumors from MMTV-Wnt1 transgenic, p53 -/mice and from 11 tumors from MMTV-Wnt1 transgenic, p53 +/mice, with or without loss of heterozygosity (LOH). Wnt1-induced tumors arising on a p53 heterozygous background, irrespective of the LOH status, contain H-Ras mutations at a frequency (5/11) similar to that described above for Wnt1-induced tumors on a wild type p53 background. Importantly, however, no H-Ras mutations were found in any of the sixteen Wnt1-induced tumors on a p53 -/background (P < 0.05). We conclude that an inherited deficiency of p53 is likely to reduce or eliminate, by an unknown mechanism, any selective growth advantage that might otherwise be contributed by secondary somatic mutations of H-Ras in Wnt1-producing mammary cells.
The finding of H-Ras mutations in the small number of tumors with loss of p53 heterozygosity implies that Ras mutation occurs before p53 LOH, since Ras mutations appear not to confer a selective advantage in p53 null cells. However, a larger cohort of tumors undergoing LOH at the p53 locus will be required to establish the conclusion more firmly.

Discussion
Most contemporary accounts of oncogenesis assume that multiple mutations affecting tumor suppressor genes and proto-oncogenes collaborate to convert a normal cell into a cancer cell. When these genetic events occur sequentially by somatic mutations during tumor development, it is assumed that the functionally significant mutations confer a selective advantage, such as an augmented rate of growth, protection from apoptosis, or promotion of further mutations, which explains why most if not all cells in a tumor exhibit multiple acquired mutations, consistent with repeated clonal selection.
In this report, we have used several features of recent studies of breast carcinogenesis in genetically manipulated mice to explore the contributions made to a multi-step oncogenic process by a variety of inherited and somatic mutations. We have identified mutations in the H-Ras proto-oncogene as a significant feature of Wnt1-induced tumorigenesis (Table 1), and we have shown that an inherited Neu transgene or an inherited deficiency of the p53 tumor suppressor gene can provide a selective advantage to Wnt1-expressing mammary cells that overrides the selective advantage conferred by somatic mutation of H-Ras. This conclusion is based on our failure to find H-Ras mutations in any tumors arising in bitransgenic (MMTV-Wnt1/MMTV-Neu) or p53-deficient (MMTV-Wnt1/p53 -/-) mice ( Table 2).
It is not difficult to reconcile the lack of H-Ras mutations in tumors from bitransgenic mice based on the overlapping signaling pathways in which Neu and Ras participate [22,23,27,28] and on the evidence for similarities in gene expression profiles and cell-type composition of mouse mammary tumors induced by these two genes [5,9,24,25]. However, the absence of H-Ras mutations in Wnt1-induced tumors from p53 null mice is surprising, since it is well known that an activated Ras gene can collaborate with p53 deficiency to promote both transformation of cultured cells [29] and tumor formation in vivo. For example, a mutant K-Ras gene induces lung adenocarcinomas more rapidly in p53-deficient than in p53-wild type mice [30,31], and K-Ras mutations are often accompanied by loss of p53 function in human cancers of the colon, pancreas, and other organs [32][33][34]. However, effects of inherited mutations on the detection of somatic mutations were previously reported in other mouse models. For example, mammary tumors from MMTV-TGF-α/ MMTV-Neu and from MMTV-Neu/ p53 R172H bitransgenic mice do not contain somatic mutations in the Neu transgene [35,36].
One curious aspect of our findings is the observation of Ras mutations affecting only H-Ras and never K-or N-Ras in Wnt1-induced mammary tumors. This contrasts sharply with the report by D'Cruz et al. [3] of mutations affecting mostly K-Ras and also N-Ras, but never H-Ras, in mouse mammary tumors induced by an MMTV-c-Myc transgene. However, three tumors from the small cohort of the MMTV-c-Myc mice maintained in our lab were found to carry mutations in H-Ras, and one had a mutation in K-Ras (KP, unpublished data). We do not have an explanation for these differences in mutation spectrum and do not know whether they reflect differences in mutational rates at Ras loci (for example, as a consequence of environmental exposures), differences in the selective advantage conferred by different mutant Ras proteins in cells expressing different oncogenes, or differences in the genetic backgrounds of the mouse lines used.
In the course of these experiments we have also found that tumors from MMTV-Wnt1/MMTV-Neu bitransgenic   Primary tumor cDNAs (11 MMTV-Wnt1/p53 -/samples) or DNAs (5 MMTV-Wnt1/p53 -/samples and all of the MMTV-Wnt1/p53 +/samples) were amplified and sequenced with H-Ras-specific primers at least once in each direction, as described in Methods. Mutations were scored by the same criteria described in Table 1.
animals were remarkably similar to those induced in MMTV-Wnt1 mice, although they appeared much earlier than tumors in monotransgenic animals, indicating cooperative effect of the two transgenes. The dominant effect of Wnt1 expression in these tumors closely resembles the dominant effect of the Myc transgene in tumors from the MMTV-c-Myc/MMTV-Neu and MMTV-c-Myc/MMTV-v-Ha-Ras bitransgenic mice [6]. This observation suggests that the Wnt signaling pathway has a dominant effect over the Ras signaling pathway in transformation of mammary epithelial cells. This might be related to the recently described effects of Wnts on stem cell maintenance during normal development [37][38][39][40]. Dominance of the Wnt phenotype in this cross might also be explained by a difference in time and level of transgene expression in the mammary tissue.