Genes commonly affected by mouse mammary tumor virus proviral insertions are found in primary mammary lesions in mice and humans. I chose to expand the study of MMTV CIS to include additional strains of MMTV and another mouse strain in a secondary panel of host-viral junction nucleotide sequences from MMTV induced mammary tumors in Balb/cfMMTVHP-REC, Balb/cf MMTVY1Y2-REC and C3H/HeN mice provided by Dr. Susan Ross (Ross et al., 2006). The genomes of MMTVHP-REC and MMTVY1Y2-REC are comprised of the 5half the endogenous Mtv1 viral genome from C3H/He mice and the 3 half from the milk borne MMTVC3H. In addition, MMTVY1Y2-REC contains a point mutation in the immunoreceptor tyrosine-based activation motif (ITAM) in the Env region of the viral genome. Nucleotide sequences from host-viral junction fragments of Balb/cfMMTVC3H induced mammary tumors were provided by Dr. John Hilkens (Theodorou et al., 2007). Thus in the current study the host-viral junction sequences of five different strains of MMTV and three mouse strains were examined using the same iteration of the mouse genome nucleotide sequence and the same methodology to define retroviral insertion sites (RIS) and CIS. A novel aspect of our study was to use microarray analysis to examine the effect MMTV integration on the expression genes on either side of the RIS and at varying distances from the integration site. Using this approach we identified 55 genes that were CIS for MMTV. Among the CIS genes, members of the Wnt, Fgf and Rspo gene families were the most frequently activated without preference for feral, inbred mouse or virus strain and are referred to as Core CIS genes. Likewise, but at a lower frequency, MMTV integration at Sfmbt2 and Pdgfra was independent mouse or virus strain. The majority of MMTV CIS (n=36) occurred at a low frequency (n=2-5) and are specific to a particular mouse or virus strains. There were also 17 """"""""predicted"""""""" or """"""""hypothetical"""""""" MMTV CIS genes that I have not studied further. The different stages of mammary tumor progression were analyzed to determine whether there is a linkage between particular CIS genes and stage of tumor development. Rspo2, but not Rspo3, has been found as a CIS in 10 out of 15 independent CzechII HOGs. In this series of HOGs one had a viral insertion at Wnt1. These results confirm and extend an earlier smaller study in which the Core CIS genes;Rspo2, Wnt1 and Fgf3/4 were each found to be activated in CzechII HOGs (Lowther et al.). Surprisingly out of 90 independent HOG derived mammary tumors only 10 had a CIS in addition to the one observed in the precursor HOG. These CIS genes include: Vash2, Cryba4, Pdgfra, Fgf3, Rspo2 and Col4a5. Similarly, none of the 15 independent lung metastasis had a CIS in addition to those in the precursor HOG or mammary tumor. Accepting the caveat that RIS and CIS may have been missed due to host-viral junction fragment size, these results suggest that MMTV induced mutations selectively contribute to the early events required for initiation of preneoplastic lesions and primary mammary tumors. However, subsequent new MMTV proviral insertions do not necessarily contribute to the transition through the different stages of malignant progression. To determine whether the genes identified by MMTV tagging in mouse mammary tumors are deregulated in human breast tumors, the Oncomine Database was interrogated for expression of the 55 human homologues of the MMTV CIS genes. The genes studied are ranked top to bottom based on their significance as seen by the differential expression analysis (Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R, et al. ONCOMINE: a cancer microarray database and integrated data-mining platform. Neoplasia. 2004;6:16). Thirty of the CIS genes (9 from feral MMTV, 10 from MMTVC3H, and 11 core CIS genes) are deregulated in tumor tissue relative to normal tissue. The genes could be further grouped on the basis of their ranking in the top 5-10% of genes in the particular study and in which type of breast carcinomas expression was detected (Table 1). Thus NCKAP5 expression, with a 5-10% ranking, was detected in invasive ductal carcinomas (IDC), mixed IDC with ductal carcinoma in situ (DCIS) and invasive lobular carcinoma (ILC) expression patterns. Likewise, PHF19 was expressed in IDC, ILC and DCIS;ASTN2, FOXL1 and FGFR2 were expressed in DCIS;FGF10, FGF19 and KIAA1267 were expressed in mixed IDC/DCIS and ILC;and PCSKIN was expressed in ILC. One of the challenges in modern cancer genetic research is to identify which of the many mutated genes found in cancers are most likely to be drivers of malignant progression. The human homologues corresponding to the list of 55 CIS genes found in MMTV induced tumors was compared with lists of """"""""high risk"""""""" genes in The Cancer Gene Atlas (TCGA), Broad, Cosmic, Sanger and Vogelstein databases of human tumors. Twenty of the MMTV CIS genes were ranked as high risk (Table 2). Of these high-risk genes, 10 were also found in the Broad, Cosmic, Sanger and Vogelstein databases. In addition, Dr. Bert Vogelstein (unpublished data) has found that 14 of the 55 CIS genes have been shown to have non-synonymous mutations in at least one human tumor, which approaches statistical significance (p =0.05, Chi2*) (Personal communication). Eight of the genes found on the Vogelstein list were not ranked in the TCGA gene list. Interestingly, for 5 of these human genes (FGFR2, PHF19, ASTN2, NXN and NOTCH4) the mutations occurred in breast tumors and in the case of ASTN2, NXN and NOTCH4 In the literature there are several claims that MMTV Env related sequences have been detected in cellular DNA from primary human breast tumors by PCR assays. The nucleotide sequence of the amplified DNA fragments was 95-100% identical to MMTVC3H Env. However, no evidence has been present that the proviral genome was actually integrated into the host cellular DNA. Having developed the inverse PCR methodology in my laboratory to detect MMTV host-viral junction fragments, it seemed reasonable to address the question of whether MMTV is an etiological agent in human breast cancer. Using this approach we analyzed 76 primary human breast tumor DNAs from Finland and found no evidence of MMTV host-viral junction fragments. This result is consistent with the conclusion MMTV is not an etiologic agent of breast cancer at least in this population.
