|Home | About | Journals | Submit | Contact Us | Français|
Mouse mammary tumor virus (MMTV), which was discovered as a milk-transmitted, infectious, cancer-inducing agent in the 1930s, has been used as an animal model for the study of retroviral infection and transmission, antiviral immune responses, and breast cancer and lymphoma biology. The main target cells for MMTV infection in vivo are cells of the immune system and mammary epithelial cells. Although the host mounts an immune response to the virus, MMTV has evolved multiple means of evading this response. MMTV causes mammary tumors when the provirus integrates into the mammary epithelial and lymphoid cell genome during viral replication and thereby activates cellular oncogene expression. Thus, tumor induction is a by-product of the infection cycle. A number of important oncogenes have been discovered by carrying out MMTV integration site analysis, some of which may play a role in human breast cancer.
The existence of mammary tumors in mice was first recognized in the early 19th century by “mouse fanciers” who bred mice for specific physical characteristics (for an excellent review on the history of mice in breast cancer research, see Cardiff and Kenney 2007). In the 1930s, J. J. Bittner at the Jackson Laboratory demonstrated that a milk-borne agent caused mammary tumors in female mice and suggested that this agent might be a virus (Bittner 1936); Bittner's hypothesis was based on the groundbreaking studies of Peyton Rous that showed that sarcomas could be induced in chickens after injection with tumor filtrates (Rous 1911). From the 1930s to the 1960s, many studies on the genetics of mammary tumor induction were performed, including identifying a genetically transmitted form of the virus in some inbred mouse strains (Bentvelzen and Daams 1969). Subsequently, mouse mammary tumor virus (MMTV) was shown to have an RNA genome like Rous sarcoma virus (RSV) and avian leukosis virus (ALV) and thus was definitively classified as a virus (Duesberg and Blair 1966). Following the discovery of reverse transcriptase in both ALVs and murine leukemia viruses (MLVs) (Baltimore 1970; Temin and Mitzutani 1970), MMTV was quickly found to contain this enzymatic activity as well (Spiegelman et al. 1970). Also following upon the observation that ALV sequences could be detected in the chicken genome, endogenous copies of the MMTV provirus were found in the mouse genome (Varmus et al. 1972).
Like all retroviruses, MMTV infects host cells by binding its surface envelope glycoprotein (Env) to a cellular receptor, in this case transferrin receptor 1 (TfR1) (Ross et al. 2002). After endocytosis to a low pH compartment, the viral membrane fuses with that of the host and capsids enter the cytoplasm. MMTV replicates by reverse-transcribing its RNA genome, and the double-stranded DNA that results from this replication integrates into cellular DNA (Figure 1). Since its discovery, MMTV has been used as a model for studying oncogenesis and virus transmission, as well as the genetics of susceptibility to infection. Here we briefly review the biology of MMTV and its impact on both the host response to virus infection and breast cancer research.
Endogenous retroviruses arise when exogenous viruses infect germ or early embryonic cells, thereby establishing themselves as germline copies. It has been estimated that endogenous retroviruses account for approximately 8% of the human genome (Hohn et al. 2013). The endogenous copies of the MMTV provirus found in the house mouse (Mus) genome are called Mtv loci (Kozak 1987). Most commonly used laboratory mouse strains have from one to six germline copies of Mtvs, the majority of which have one or more mutations/deletions of viral genes and thus do not produce infectious virions. Inbred mice that inherit functional, fully infectious endogenous Mtvs, likely representing recent germline integrations that have not yet been silenced by mutation, have very high mammary tumor incidence, which occurs in the absence of milk-borne virus (Michalides et al. 1978; Nandi and McGrath 1973). At least 10 different exogenous and more than 30 endogenous Mtvs have been identified (Callahan and Smith 2000; Kozak 1987). Most of these viruses retain functional viral superantigen (sag) genes, which can have profound effects on the mouse immune system (see MMTV Transmission). Although the rat (Rattus) genome contains MMTV-like sequences, not all wild Mus species have endogenous Mtv proviruses, and based on this, it has been estimated that MMTV first infected Mus approximately 10 million years ago, after their speciation from rats (Baillie et al. 2004). Thus, it is likely that the germline MMTV-like sequences in rat represent a de novo zoonotic transmission, occurring between 1.6 and 3.5 million years ago; transmission into the North American deer mouse Peromyscus also seems to have occurred within a similar time frame (Demogines et al. 2013).
The different exogenous strains of MMTV are generally named for the inbred mouse strains from which they were derived (e.g., MMTV-C3H, MMTV-GR, MMTV-BR6). There is greater than 95% identity both at the nucleic acid and amino acid level among the different MMTVs, with the greatest amount of divergence localized to the sag genes encoded within the long terminal repeats (LTRs) (Brandt-Carlson et al. 1993). Exogenous MMTV strains infect and cause mammary tumors in different strains of mice to varying degrees. Some of this variability can be attributed to the ability of the different MMTVs to interact with the immune systems of the different strains (Ross 2010) (see MMTV Transmission) but may also result from the susceptibility of the mammary cells to transformation (Callahan and Smith 2000).
In the 1970s, the identification of a retrovirus that causes breast cancer in mice created great interest in determining whether similar viruses exist in other species, particularly humans. A number of early studies showed that human breast cancer cells and tissues had proteins that cross-reacted with anti-MMTV antisera; these studies also showed the presence of MMTV-like RNA and particles in tumors and human milk (Axel et al. 1972; Keydar et al. 1984; Mesa-Tejada et al. 1982; Spiegelman et al. 1970). More recently, several groups, largely using nested polymerase chain reaction–based techniques, have identified MMTV-like sequences in human breast cancers but not normal mammary tissue (Lawson et al. 2010; Liu et al. 2001; Wang et al. 1995). However, these findings have not been replicated in all studies (Bindra et al. 2007; Mant et al. 2004; Morales-Sanchez et al. 2013; Park et al. 2011). The MMTV sequences found in human samples are very closely related, if not identical, to those found in mice, suggesting either zoonosis from mice to humans or contamination of the human samples with mouse DNA, as was shown for the putative zoonotic retrovirus xenotropic-related murine retrovirus associated with human prostate cancer (Delviks-Frankenberry et al. 2012). Whether MMTV can infect human cells is controversial, although one group has reported that this occurs in some but not all human mammary tumor cell lines (Indik et al. 2007; Konstantoulas et al. 2015). However, only the Mus or Rattus—not human—TfR1 can efficiently support virus entry (Demogines et al. 2013; Ross et al. 2002; Wang et al. 2008). Moreover, it is likely that the antiviral restriction factors found in human cells efficiently prevent MMTV infection and replication (Okeoma et al. 2007; Stavrou et al. 2014).
Although the existence of an exogenous MMTV-like virus in human breast cancer is controversial, many species, including humans, harbor endogenous retrovirus sequences related to MMTV. As discussed above, the genomes of rats and deer mice contain proviruses highly similar to MMTV, likely the result of zoonosis sometime in their ancestry. Sequences similar to MMTV have also been found in dwarf hamsters, certain species of bats, horses, and most recently in American pikas and polar bears, although these proviruses are more highly divergent from MMTV than those found in rats and deer mice (Escalera-Zamudio et al. 2015; Hayward et al. 2013; Lemos de Matos et al. 2015; Mayer et al. 2013; Tikhonenko et al. 1990; van der Kuyl 2011). The first family of human endogenous retroviruses (HERVs) were identified because of their homology to MMTV (Ono et al. 1986). The most recently integrated HERV proviruses, HERV-K, belong to a subgroup that are the most highly related to MMTV (Hohn et al. 2013). However, no fully functional copies of HERV-K have yet been identified in the human genome, although they have been resurrected by molecular cloning and correction of the mutations in the viral genes (Dewannieux et al. 2006; Lee and Bieniasz 2007). Interestingly, these resurrected HERV-Ks are highly inhibited by host antiviral restriction factors, suggesting that host controls preventing reemergence of infectious retroviruses is in place (Esnault et al. 2005; Lee et al. 2008; Lemaitre et al. 2014).
Exogenous MMTV is spread by the milk of infected females and is acquired by suckling pups (Nandi and McGrath 1973) (Figure 2). The primary targets for exogenous MMTVs are dendritic cells (DCs), T cells, and B cells located in Peyer's patches of the gastrointestinal tract of neonatally infected pups (Beutner et al. 1994; Held, Shakhov, et al. 1993). MMTV gains access to these cells by passing through M cells located in the follicular-associated epithelium of the intestinal Peyer's patches (Golovkina et al. 1999). The spread from DCs to T cells requires activation of T and B lymphocytes by the viral superantigen (Sag) encoded by sag mapped within the 3′ LTR (Choi et al. 1991). Conventional antigens are presented by major histocompatibility (MHC) class II molecules expressed on antigen-presenting cells and are recognized by both the Vβ and the Vα chains of the T cell receptor (TCR), resulting in stimulation of a relatively small percentage of T cells. Sags are also presented by MHC class II molecules. However, in contrast with conventional antigens, Sag recognition relies predominately on the Vβ domain of the TCR, resulting in stimulation of approximately 10% to 30% of all T cells (Acha-Orbea and MacDonald 1995; MacDonald et al. 1988; MacDonald et al. 1989). CD40L/CD40-mediated interaction between activated T cells and B cells activates expression of costimulatory molecules on B cells and promotes further activation and proliferation of T cells and B cells (Chervonsky et al. 1995).
Overall, the LTR sequences of different MMTVs are highly conserved (Brandt-Carlson et al. 1993). However, the region encoding the C-terminal segment of Sag is more diverse. The amino acid sequence of this region, also known as the hypervariable region, contacts the Vβ chain of the TCR and thus determines which T cells are affected (Yazdanbakhsh et al. 1993). Sag-activated T cells are gradually lost, and their deletion is used as a read-out for virus infection (Marrack et al. 1991). Sags encoded by endogenous Mtvs stimulate deletion of the Vβ+ T cell subsets during formation of the immune repertoire in utero. Mice, which inherit endogenous Mtvs, are born without subsets of T cells affected by their Sags.
Sag function is crucial to the MMTV life cycle because mice that lack Sag-cognate T cells due to the expression of sag transgenes or endogenous Mtvs cannot be infected with exogenous viruses bearing Sags of the same Vβ specificity (Golovkina et al. 1992; Held, Waanders, et al. 1993). In addition, MMTVs without functional Sags cannot activate cognate T cells or be transmitted to the offspring of infected mothers (Golovkina et al. 1998). Furthermore, mice with MHC haplotypes that do not express MHC class II I-E molecules, essential for presentation of the vast majority of MMTV Sags (Held et al. 1994), are relatively resistant to MMTV (Pucillo et al. 1993; Wrona and Dudley 1996).
Sag presentation by infected cells results in a pool of proliferating cells, which are essential for efficient MMTV infection (Held, Waanders, et al. 1993) because MMTV requires dividing cells for efficient delivery of the preintegration complex into nuclei of infected cells (Varmus et al. 1977). Sag-mediated activation also allows the virus to amplify in lymphocytes prior to transmission of the virus to mammary epithelial cells, which likely occurs during puberty (Nandi and McGrath 1973). Ultimately the virus is transported by infected lymphocytes to the mammary glands (Finke and Acha-Orbea 2001; Golovkina et al. 1998). The virus is shed into milk during lactation, therefore completing the replication cycle (Figure 2).
Innate and adaptive immunity evolved as host defense mechanisms against different pathogens. The major targets of innate immune recognition are pathogen-associated molecular patterns (PAMPs) (Janeway 1989). PAMPs represent conserved molecular structures that are produced by pathogens and are essential for the survival of the pathogen. PAMPs are recognized by the germ line–encoded innate receptors also known as pattern recognition receptors (PRRs), which are expressed either on the cell surface or intracellularly (Medzhitov et al. 1997). PRRs, including Toll-like receptors (TLRs), cyclic GMP-AMP synthase (cGAMP), and Aim2-like receptors (ALRs), bind to PAMPs, such as retroviral RNA and reverse-transcribed DNA (see below), and then signal through several different pathways, ultimately leading to the activation of various transcription factors, including NF-κB and interferon-response factors (IRFs), and the transcription and production of type I interferons, proinflammatory cytokines, and costimulatory molecules (Bhat and Fitzgerald 2014; Medzhitov 2001; Medzhitov et al. 1997). Costimulatory molecules are necessary for antigen presentation by MHC class I and II molecules to naive CD4+ T cells and their subsequent differentiation to TH1 and TH2 subsets, leading to production of pathogen-specific humoral (antibody) and cytotoxic CD8 T cell responses, respectively.
Successful pathogens have evolved to specifically evade both innate and adaptive immune responses, and retroviruses are notoriously known for evasion of antiviral immune response. Studying inbred mice with different susceptibilities to the virus led to discovery of the host protective responses controlling MMTV and the mechanisms evolved by the virus to counteract these responses.
Apolipoprotein editing complex 3 (APOBEC3) proteins are cytidine deaminases that play major roles in restricting retrovirus infection (Malim and Bieniasz 2012). Mouse APOBEC3 blocks MMTV early reverse transcription, thereby inhibiting viral replication (MacMillan et al. 2013) and MMTV infection, and virus-induced mammary tumor incidence is much higher in APOBEC3-knockout mice compared with wild-type mice (Okeoma et al. 2007). Certain mouse strains, including those derived from the C57BL/6 line, which are relatively resistant to MMTV infection, inherit a unique allele of Apobec3 that more effectively blocks reverse transcription than the allele found in MMTV-susceptible strains such as BALB/c or C3H (Okeoma et al. 2009). Recent data also suggest that retroviral reverse transcripts serve as PAMPs, which are sensed by host DNA sensor PRRs, leading to type I interferon production that contributes to the antiviral response (Dempsey and Bowie 2015). Although this has not yet been formally demonstrated for MMTV, it is clear that genetic differences in the ALR genes in different inbred mouse strains alter their ability to respond to MLV reverse transcripts and thus to inhibit infection (Stavrou et al. 2015).
I/LnJ mice, which are susceptible to MMTV infection, do not transmit the virus to subsequent generations (Purdy et al. 2003). The I/LnJ mechanism of resistance is different than that discussed in the preceding section, in which endogenous Mtv loci prevent infection by milk-borne virus (Golovkina 2000). After one passage through I/LnJ mice, infectious milk-borne MMTV becomes avirulent because it is coated with virus-neutralizing antibodies produced in infected mothers, thus preventing the virus from reinfecting new cells in nursing pups (Purdy et al. 2003). The virus-neutralizing antibody response is triggered by virion RNA, a PAMP recognized by the endosomal PRR TLR7, as I/LnJ mice deficient in TLR7 or in MyD88 (an adaptor TLR7 signals through) failed to make antiviral antibodies (Kane, Case, Wang, et al. 2011). A single recessive gene, virus infectivity controller (vic1), governs the production of antiviral neutralizing antibodies in I/LnJ mice (Purdy et al. 2003). Unlike I/LnJ mice, MMTV-susceptible mice such as C3H/HeN carry a vic1 allele that negatively regulates the immune response and thereby prevents the formation of antiviral antibodies by an as-of-yet unknown mechanism (Case et al. 2008; Golovkina 2000; Kane, Case, Wang, et al. 2011).
Another mechanism evolved by the virus to evade host protective responses resulted from studies of two closely related mouse lines (C3H/HeN and C3H/HeJ). C3H/HeJ mice carrying a mutation in Tlr4 exhibited a significant delay in mammary tumor development (Outzen et al. 1985). Later it was found that mice deficient in TLR4 decrease milk-borne transmission, thereby eliminating MMTV when passed through successive generations (Jude et al. 2003). Subsequent studies using mice reared in germ-free (GF) conditions and antibiotic-treated mice suggested that the virus relies on the gut microbiota for its transmission. Specifically, MMTV interacts with TLR4 and triggers production of interleukin 10 (IL-10) (Jude et al. 2003). Because TLR4 recognizes Lipopolysaccharide (LPS) produced by Gram-negative bacteria and IL-10 is an immunosuppressive cytokine, it was proposed that the virus exploits the commensal bacteria to counteract the antiviral immune response (Kane, Case, Kopaskie, et al. 2011). In fact, LPS-free MMTV stocks failed to trigger TLR4 signaling (Kane, Case, Kopaskie, et al. 2011). Furthermore, specific pathogen-free (SPF) mice treated with high doses of antibiotics or deficient in either TLR4 or IL-10, as well as GF wild-type mice, were all unable to efficiently transmit infectious virus through successive generations (Kane, Case, Kopaskie, et al. 2011). However, when MMTV-infected GF wild-type mice were reconstituted with a defined bacterial community (altered Schaedler's flora), their ability to transmit the virus was completely restored (Kane, Case, Kopaskie, et al. 2011). Furthermore, infected GF MyD88- as well as SPF MyD88/TLR4-double deficient mice reversed their virus-resistant phenotype and transmitted MMTV, suggesting that signaling through MyD88 was required for the microbiota-induced antiviral response (Kane, Case, Kopaskie, et al. 2011). Taken together, these results suggest that the virus exploits the microbiota, specifically LPS, to counteract the antiviral immune response.
A series of genetic and biochemical experiments established that, after infiltrating the host, the virus cloaks itself in LPS. However, the nature of the virus–LPS interactions remained unknown. Mammals have multiple LPS-binding proteins, such as the membrane-anchored protein CD14, which binds LPS for transfer to the MD-2-TLR4 complex (Viriyakosol et al. 2001). MD-2 directly binds LPS, and LPS-bound MD-2 triggers TLR4 dimerization and activation of downstream signaling (Park et al. 2009). MMTV, an enveloped retrovirus, targets cells that express and display these LPS-binding proteins (Lee et al. 2009; Miyake 2006; Wilks et al. 2015). Thus, the virus could potentially acquire these proteins from the host during budding and utilize them to bind LPS. Subsequently, mammalian LPS receptors were shown to be incorporated into the viral envelope and bind LPS, enabling the virus to activate the immune evasion pathway (Figure 2) (Wilks et al. 2015).
MMTV induces mammary tumors typically within 6 months to 1 year, a longer time frame than that observed for retroviruses that encode an oncogene (Rosenberg and Jolicoeur 1997). A variety of factors dictate the emergence of MMTV-induced mammary tumors, including the viral strain, the presence of endogenous Mtvs, and multiple host genes that influence the immune response and expression of viral proteins (reviewed in Bentvelzen 1974; Cardiff and Kenney 2007; Hilgers and Bentvelzen 1978; Holt et al. 2013; Ross 2010). Moreover, MMTV does not carry a classical oncogene, such as v-myc or v-src (Bishop 1981), which gives viral transformation in cell culture; retroviruses that express such proteins produce rapidly appearing, polyclonal tumors (Rosenberg and Jolicoeur 1997). The long latency of mammary tumors resulting from MMTV infection is consistent with insertional mutagenesis, and the resulting tumors are often clonal (Nusse and Varmus 1982; Peters et al. 1983). Nevertheless, the MMTV Env protein has been implicated in mammary tumorigenesis because Env overexpression has been shown to result in morphological changes to normal 3-dimensional mammary cultures (Katz et al. 2005; Ross et al. 2006). Env-induced phenotypic changes required an immunoreceptor tyrosine-based activation motif (ITAM) (Katz et al. 2005), and mutation of this motif decreased signaling through SRC and increased susceptibility of mammary cells to apoptosis (Kim et al. 2012). Furthermore, an infectious MMTV provirus lacking a functional Env ITAM motif developed mammary tumors with decreased incidence and higher latency (Ross et al. 2006). One interpretation of these results is that MMTV Env is not a typical oncogene but synergizes with gene products activated by proviral insertions in mammary cells.
Retroviruses are retrotransposons, which are known to act as insertional mutagens on their hosts (Carreira et al. 2014). Early studies of tumors induced by ALVs and MLVs revealed insertions near and deregulation of particular cellular proto-oncogenes (Corcoran et al. 1984; Hayward et al. 1981; Neel et al. 1981; Payne et al. 1982; Steffen 1984). Like human cancers, MMTV-induced mammary tumors are believed to be the result of multiple mutagenic events (Nusse and Varmus 1982, 1992). The initiation of tumors is dependent on hormonal cycles that stimulate division of mammary epithelial cells and multiple integrations of the virus (Bentvelzen 1974; Kordon 2008). As a result, the incidence increases with additional pregnancies and lactations (Bentvelzen 1974; Hilgers and Bentvelzen 1978).
Forerunners of MMTV-induced tumors are preneoplastic lesions known as hyperplastic alveolar nodules (HANs), which then undergo further genetic changes and progress to tumors (Schwartz et al. 1992; Squartini et al. 1983). Both human and MMTV-induced breast tumors can metastasize to distant sites (Callahan 1996). Studies of HANs and mammary tumors derived from MMTV-infected animals indicate that both have acquired integrated copies of viral DNA (Callahan and Smith 2008; Gama-Sosa et al. 1987). In most cases, multiple proviruses are present in each tumor, presumably resulting from lack of superinfection resistance by MMTV (Dzuris et al. 1999) and high levels of viral replication in the differentiated mammary gland (Zhu et al. 2000), as well as selection for insertion mutations. Nevertheless, rare tumors have been shown to contain one or two MMTV integrations (Nusse and Varmus 1982; Peters et al. 1983). The identification of tumors with single proviral insertions indicates that many MMTV-induced breast cancers are clonal outgrowths of virus-infected cells (Nusse and Varmus 1982; Varmus et al. 1981). Because MMTV does not have preferred insertion sites (Cohen et al. 1979; Faschinger et al. 2008), this observation strongly suggests that the clonality of these tumors results from growth selection of a small subset of these cells (Peters et al. 1983). Characterization of proviruses and their insertion sites has been the primary focus for studies of the molecular basis of MMTV-induced cancers.
The cloning of MMTV proviruses and their cellular flanking sequences from tumors with low numbers of integrated copies led to the identification of the Int1 gene (Nusse and Varmus 1982), later renamed as Wnt1 due to its sequence homology to the wingless gene of Drosophila (Rijsewijk et al. 1987). MMTV insertions in the Wnt1 gene of different mammary tumors were verified, typical of common insertion sites (CISs) and likely a target for selection of the outgrowth of tumor cells (Nusse and Varmus 1992) (Figure 3). Shortly after the identification of Wnt1, another CIS, Fgf3, a member of the fibroblast growth factor family, was identified in other MMTV-induced mammary tumors (Dickson et al. 1990; Peters et al. 1983). Examination the insertion sites indicated that proviruses were integrated either upstream in the opposite transcriptional orientation or downstream in the same transcriptional orientation as Wnt1 and Fgf3 and that MMTV integration activated their transcription. Neither gene is normally expressed in the mammary gland, although both are expressed during embryonic development or in other adult tissues (Jakobovits et al. 1986; McMahon and Moon 1989; Nusse et al. 1984). MMTV insertions also were found at large distances from Wnt1 (approximately 25 kb) (Callahan and Smith 2000), and more recent work suggests that viral integration may occur more than 200 kb from the target gene (Callahan et al. 2012; Theodorou et al. 2007). The infrequency of integration within the Wnt1 and Fgf3 coding regions and the absence of hybrid transcripts containing viral sequences indicated that MMTV primarily induces breast cancers by enhancer activation of cellular genes rather than promoter insertion (Clausse et al. 1993; Nusse 1991).
Transgenic mice expressing either WNT1 or FGF3 under the control of the MMTV LTR showed multiple mammary hyperplasias (Muller et al. 1990; Shackleford et al. 1993; Tsukamoto et al. 1988). These observations suggest that the inappropriate expression of these growth factors in mammary tissue leads to division of differentiated cells (Muller et al. 1990; Tsukamoto et al. 1988). The clonal nature of the derivative tumor cells indicates that expression of these growth factors alone is insufficient for mammary cancer formation. Furthermore, mating of Wnt1- and Fgf3-transgenic mice leads to acceleration of mammary tumors, indicating that these growth factors act cooperatively (Kwan et al. 1992). Nonetheless, the resulting tumors remain clonal, consistent with the requirement for additional mutations (Callahan and Smith 2000). Interestingly, infection of mice with an MMTV provirus that lacks the ITAM motif in the envelope gene had a reduced frequency of Wnt1 and Fgf3 insertions relative to wild-type MMTVs, suggesting that other proto-oncogenes may synergize with MMTV Env (Ross et al. 2006).
In addition to Wnt1 and Fgf3, a number of other MMTV-specific CISs have been identified, both by traditional and high-throughput screening methods (Callahan et al. 2012; Gallahan et al. 1987; Kim et al. 2011; Klauzinska et al. 2012; Marchetti et al. 1995; Morris et al. 1991; Roelink et al. 1990; Theodorou et al. 2007). Other members of the Wnt and Fgf family, including Wnt3a, Fgf6, Fgf8, and Fgf10, as well as the FGF receptor Fgfr2, are CISs, providing supporting evidence for the involvement of these growth factors in mammary tumors (Katoh 2002). Because some family members are clustered on mouse chromosomes, a single MMTV insertion may activate multiple genes, only one of which may participate in tumorigenesis (Callahan et al. 2012; Kim et al. 2011; Klijn et al. 2013; Koudijs et al. 2011; Theodorou et al. 2007). The majority of CISs also appear to be activated by MMTV enhancer insertion outside of coding regions, but exceptions can be found (Callahan et al. 2012; Klijn et al. 2013; Theodorou et al. 2007). In some cases, viral insertions truncate the target genes, allowing expression of the intracellular portion of the NOTCH4 receptor (Gallahan and Callahan 1997), or inactivate expression (Kcnj6) (Callahan et al. 2012).
Other CISs include gene families encoding thrombospondin (Rspo2 and Rspo3) or NOTCH (Notch4), which do not function as growth factors (Theodorou et al. 2007); instead, these proteins are part of signaling pathways important for mammary cell proliferation—for example, enhancement of WNT signaling by RSPO proteins (de Lau et al. 2014; Klauzinska et al. 2012). Both Rspo and Wnt genes can be activated in the same breast cancers (Theodorou et al. 2007). Of note, despite the large numbers of WNT receptors and coreceptors, only Antxr1/Tem8 has been identified as an MMTV CIS by high-throughput screening (Kim et al. 2011), possibly acting as a stabilizer of the LRP6 coreceptor needed for WNT signaling (Chen et al. 2013). Therefore, multiple signaling events related to proliferation and embryonic development are targeted by MMTV insertions selected during tumorigenesis.
It appears likely that MMTV CISs identify genes important for signaling and constitutive proliferation of mammary stem cells, many of which are dependent on the WNT pathway (Jang et al. 2015). These initiation events then may sustain additional genomic damage to acquire tumorigenic properties (Callahan et al. 2012; Klijn et al. 2013). However, only a few of the described MMTV CISs have been functionally shown to influence the growth behavior of mammary cells (Callahan et al. 2012; Kim et al. 2011). Finally, many MMTV CISs are overexpressed in human mammary carcinomas, validating the use of MMTV as a molecular tag for identification of novel therapeutic targets (Callahan et al. 2012; Kim et al. 2011; Klijn et al. 2013; Theodorou et al. 2007).
An MMTV variant (type B leukemogenic virus [TBLV]) induces T cell lymphomas rather than mammary cancers (Ball et al. 1983; Ball et al. 1985). The TBLV genome, as well as other MMTV variants isolated from T cell lymphomas, have alterations in the U3 region of the LTR, including deletion of a negative regulatory element, whereas other viruses, including TBLV, also had duplications of sequences flanking the deletions (Ball et al. 1988; Dudley et al. 1986; Hsu et al. 1988; Lee et al. 1987; Michalides et al. 1985; Yanagawa et al. 1990). In TBLV, triplication of sequences flanking the negative regulatory element deletion acted as an enhancer in T cells but not mammary or B cells (Mertz et al. 2001). Substitution of the mammotropic provirus with LTRs of the lymphomagenic provirus enabled the resulting virus to generate T cell lymphomas (Bhadra et al. 2005; Yanagawa et al. 1993).
Unlike MMTV-induced mammary tumors, T cell lymphomas resulting from TBLV infection appear to be polyclonal (Broussard et al. 2002). Although a high-throughput approach has not been used, standard cloning and sequencing of flanking regions from tumors induced by lymphotropic MMTVs identified at least four CISs (Myc, Rorc, Notch1, and Tblvi1), each of which is overexpressed in lymphomas induced by T cell–tropic MMTVs compared with normal thymus (Broussard et al. 2004; Mueller et al. 1992; Rajan et al. 2000; Yanagawa et al. 2000). TBLV integrations were upstream or downstream of Rorc and Myc, consistent with the enhancer insertion model proposed for most MMTV-induced mammary tumors (Broussard et al. 2002; Broussard et al. 2004; Rajan et al. 2000).
High-throughput identification of CISs from MLV-induced lymphomas or MMTV-induced mammary tumors has been used to suggest that target genes are not virus specific but dependent on the tissue of origin (Theodorou et al. 2007). Another interpretation of these data is that selection for tumor cell growth occurs due to the ability of particular retroviral enhancers to promote expression of key proto-oncogenes. In support of this, three of the lymphotropic MMTV CISs (Myc, Rorc, and Notch1) are identical with those identified by screens of MLV-induced lymphomas (Akagi et al. 2004). Moreover, the TBLV LTR enhancer binds the transcription factors MYB and RUNX1 (Mertz et al. 2007), similar to the T cell–tropic MLVs (Zaiman et al. 1998). Thus, the tissue specificity of retroviral CISs is the result of recruitment of particular transcription factors to retroviral enhancers that allow high-level replication of the virus, as well as the overexpression of discrete linked proto-oncogenes that promote tumor growth and survival.
The use of MMTV as a model for studying in vivo virus infection and tumorigenesis has clearly provided important advances in our understanding of both of these processes. The strength of this experimental system is the ability to use a natural pathogen of mice coupled with the study of infection in inbred and genetically altered mice. Because MMTV has been in Mus species for millions of years, it has clearly evolved to use host factors for its replication while avoiding the host antiviral response. At the same time, the long latency for tumor induction in mice indicates that the host has evolved as well to control MMTV pathogenesis. This includes intrinsic and innate restriction factors as well as the adaptive immune response.
MMTV has also served as an important means for studying cancer in mice. In addition to the identification of novel CISs, including genes in the WNT pathway now known to be important in many different human cancers, the MMTV LTR has been one of the major enhancer/promoter elements used to direct gene expression, particularly of oncogenes, to mammary tissue in transgenic mice. It is likely that the study of MMTV will continue to serve as an important model to further our understanding about multiple aspects of viral infection and oncogenesis.