B and T Cells Are Required for Mouse Mammary Tumor Virus Spread Within the Mammary Gland¹

It is well established that mouse mammary tumor virus (MMTV)³ utilizes cells of the immune system in its infection pathway. Exogenous MMTV is produced by the mammary glands of infected females and is acquired via milk by suckling pups (1). However, the ultimate targets of MMTV are mammary gland cells, which begin dividing during puberty (2). Thus, this virus enters the host at a time when its main targets are not yet capable of being infected. To overcome this temporal problem, MMTV initiates a local infection of first B and then T cells in the Peyer’s patches of the gastrointestinal tract (3). The infected lymphoid cells then bring virus to the cells of the developing mammary gland, thereby also allowing the virus to overcome its spatial problem.

MMTV is able to initiate lymphoid cell infection because it encodes a superantigen (SAg) in its long terminal repeat (LTR) (4). This SAg is presented by APC and causes the stimulation of cognate T cells bearing particular Vβ chains of the TCR (5). As a result, bystander B and other cells are also stimulated to divide, setting up a reservoir of infection-competent cells (5, 6). The major evidence for this pathway comes from studies done with mice lacking SAg-reactive T cells (5, 7) or B cell-deficient mice (8); both are resistant to milk-borne transmission of MMTV. Because both B and T cells become MMTV infected in this process (3) and are capable of producing infectious virus (9), it is hard to determine whether one or both types of lymphocytes are required to bring virus to the mammary gland.

Once MMTV infects mammary gland cells, virus amplification within this tissue is required both to maximize virion production and to induce mammary tumors. MMTV is a nonacute transforming retrovirus and does not encode an oncogene (1). Mammary tumorigenesis takes place after the insertion of proviral DNA near cellular proto-oncogenes and activation of their transcription (1, 10, 11). Because retroviral integration is not site specific (1), the more virus produced, the more likely it is that proviral DNA will integrate near such a proto-oncogene. Since MMTV transcription is induced by lactogenic hormones, including progesterone and glucocorticoids (12), virus production is increased dramatically during pregnancy (13). For example, virgin mice have much lower mammary tumor incidence than do multiparous females because their mammary glands are less infected with MMTV (13, 14). However, it is not known how virus spread occurs within the mammary gland.

Here, we present evidence that SAg function is required both for milk-borne viral transmission and for virus spread within the mammary gland. A transgenic mouse strain was created carrying a molecularly cloned MMTV provirus with a frameshift mutation in the sag gene that resulted in its premature termination (15). Infection in offspring that nursed on this milk was aborted at the level of viral amplification in lymphoid cells, due to the lack of SAg-induced T cell stimulation. In addition, we demonstrate here that SAg activity is required for efficient viral infection of the mammary epithelial cells and consequent tumorigenesis, since mice lacking SAg-cognate T cells as well as B cell-deficient mice could not be infected by direct injection of MMTV into mammary glands. Thus, SAg-mediated stimulation of lymphoid cells is needed for their infection and for virus spread between mammary gland cells.

All mice used in this study were bred and maintained at the animal facility of the University of Pennsylvania, Philadelphia, PA. C3H/HeN HYB PRO/Cla transgenic mice containing the genetically engineered MMTV provirus with the mutated sag gene were described elsewhere (15). To avoid viral recombination between the Mtv1 endogenous provirus in C3H/HeN mice and the transgene, the HYB PRO/Cla transgene was bred onto the BALB/c background (Mtv6, -8, and -9) for two generations. Transgenic females null for Mtv1 were identified by Southern blot analysis and showed no deletion of Vβ14⁺ T cells (not shown), in contrast to the HYB PRO/Cla females in the C3H/HeN (Mtv1⁺) background (15). The MTV-ORF no. 16 and LEL no. 1 transgenic mice were previously described (7, 16).

Homozygous IGH6/BL6 (H-2^b) females (μ-chain knockout mice) (22) purchased from The Jackson Laboratory (Bar Harbor, ME) were crossed with C3H/HeN MMTV⁻ (H-2^k) males (National Cancer Institute, Frederick Cancer Research Facility, Frederick, MD) to generate F₁ progeny. The F₁ progeny were intercrossed, and F₂ generation female mice were typed for surface IgM expression and the MHC class II H-2^k haplotype. C3H/HeN MMTV⁺ mice were purchased from National Cancer Institute, Frederick Cancer Research Facility.

Surface expression of IgM and MHC class II was determined by staining PBLs with FITC-labeled anti-mouse IgM and IgG Abs (Jackson ImmunoResearch Laboratories, West Grove, PA) or anti-mouse H-2^k Abs (PharMingen, San Diego, CA), respectively. FITC-coupled mAbs against the Vβ14 TCR chain were purchased from PharMingen. Anti-CD4 Abs (GK 1.5) coupled to phycoerythrin were purchased from Life Technologies (Grand Island, NY). A FACScan (Becton Dickinson, Mountain View, CA) flow cytometer and CellQuest software program were used for FACS analysis.

Virus was isolated from the milk-filled stomachs of 2- to 3-day old pups nursed on C3H/HeN MMTV⁺ mice (17) and injected into the mammary glands of the indicated 3-wk-old mice. In the first experiment (see Table II), 300 μl of the virus preparation (1 mg/ml of the protein level) was injected per mouse. In all other experiments, 300 μl of a different virus preparation (0.15 mg/ml of the protein level) was injected per mouse.

RNase T₁ protection assays were performed as previously described (16), using a probe specific for MMTV (C3H) viral transcripts (17). Forty micrograms of total RNA isolated from the lactating mammary glands and 5 μg of RNA isolated from the milk of the mice described in the text were used.

Mammary gland tumor incidence in the MMTV(C3H)-injected LEL transgenic mice and their nontransgenic littermates was monitored by weekly palpation of the animals. Tumor-bearing mice were sacrificed and DNA isolated from a portion of each tumor was subjected to Southern blot analysis as described previously (17). All of the tumors contained new MMTV integrants, indicating that the tumors were caused by the virus (not shown).

High m.w. DNA (0.25 μg) isolated from spleens, thymis, or Peyer’s patches was amplified by PCR. Amplification of newly integrated copies of exogenous MMTV(C3H) and HYB PRO/Cla viruses in nontransgenic mice was accomplished using the following LTR-specific primers: 5′-AATTCGGAGAACTCGACCTTCC-3′ (nt 268–289) and 5′-TAATGTTCTATTAGTCCAGCCACTGT-3′ (nt 923–898) (19). Semiquantitative PCR was carried out as previously described (18). Briefly, 31 cycles of 1 min at 55°C, 1 min at 72°C, and 1 min at 94°C gave linear DNA amplification, whereas after 35 cycles, the amplification had plateaued (nonquantitative conditions). After PCR amplification, 1 μl (1/40 of the volume, semi-quantitative conditions) or 20 μl (1/2 of the volume, nonquantitative conditions) of each reaction mixture were incubated with MfeI restriction enzyme (New England Biolabs, Beverly, MA), as indicated in the figure legend, and the resulting products were analyzed on 1.5% agarose gels. These primers also amplify some endogenous MMTVs under nonquantitative PCR conditions. To amplify the MMTV(C3H)-integrated proviruses in the MTV-ORF and LEL transgenic mice, different primers were used, since the primers described above amplified both the transgenes. In this case, a forward primer specific for the MMTV(C3H) LTR (5′-GACAGTGGCTGGACTAATAGAACATT-3′, nt 898–923) and a gag-specific MMTV BR6 reverse primer (5′-CCTACCTCTTCTCTGTAGGCGAGAC-3′, nt 1613–1589) were used. Amplification with this primer set was carried out as follows: 35 cycles of 1 min at 49°C, 1 min at 72°C, and 1 min at 94°C. After PCR amplification, 1 μl of each reaction (1/40 part of the volume) was run on 1.5% agarose gel. Southern blots of the PCR products were hybridized with LTR- (20) or gag-specific probes. The gag-specific probe was cloned as a 2-kb PstI-XhoI fragment from the HYB MMTV plasmid (21).

MMTV can infect only the mammary glands of mice with both functional B cells and SAg-cognate T cells (5, 7, 8). Although these immune system cells are important for MMTV transmission, an absolute requirement for SAg in the MMTV life cycle has not yet been shown. We introduced a transgene, HYB PRO/Cla, with a frameshift mutation at amino acid 113 in the sag gene into C3H/HeN mice. This frameshift caused premature termination of this protein (15). Because this termination occurred upstream of the SAg hypervariable region that determines TCR Vβ specificity, no deletion of Vβ14⁺ T cells should occur in these mice. However, this mutation was corrected in transgenic females by retroviral recombination with the endogenous Mtv1 virus present in C3H/HeN mouse strain, leading to the generation of a Vβ14-deleting, milk-transmitted virus (15). To prevent this recombination, the transgene was crossed into BALB/c mice, which lack Mtv1. The Mtv1⁻ HYB PRO/Cla transgenic animals generated from these crosses expressed the transgene, shed virus in their milk (Fig. 1 A, lanes 4 and 3, respectively) (15), and showed no deletion of their SAg-cognate Vβ14⁺ T cells (data not shown).

To determine whether the mutated MMTV produced by the Mtv1⁻, HYB PRO/Cla transgenic females was infectious, these mice were bred and offspring generated. None of the nontransgenic Mtv1⁻, HYB PRO/Cla-nursed offspring acquired virus, since no transgene-specific RNA was detected in either their mammary glands or milk (Fig. 1,A, lanes 2 and 1, respectively). In contrast, mice nursed on a Mtv1⁺, HYB PRO/Cla transgenic mice acquired infectious virus (Fig. 1 A, lanes 5 and 6), as previously shown (15). These results indicated that a SAg-minus MMTV could not infect mammary gland cells.

The primary targets for MMTV during milk-borne infection are lymphocytes. To determine whether SAg was also required for infection of lymphoid cells, we examined the relative levels of viral DNA present in the lymphoid organs of nontransgenic mice nursed on the Mtv1⁻, HYB PRO/Cla-transgenic mothers using PCR analysis (see Materials and Methods). After 31 cycles of PCR amplification, no exogenous MMTV DNA was detected in DNA isolated from the spleens, thymi, or Peyer’s patches of four different 3-mo-old Mtv1⁻, HYB PRO/Cla-infected mice (mice 1–4, Fig. 1,B, top panel), in contrast to mice nursed on Mtv1⁺, HYB PRO/Cla-infected mice (E.C., Fig. 1,B). However, after 35 cycles, viral DNA was detectable in these organs (Fig. 2 B, bottom panel), indicating that there was low level infection of lymphoid cells. Thus, the SAg-mutated virus did not amplify in the lymphoid cells and this was responsible for our failure to detect MMTV infection in the mammary gland.

Infected lymphoid cells are required for the primary infection of mammary gland cells during milk-borne transmission, but it is not known whether they play a role in MMTV spread within the mammary gland. Although mice naturally acquire virus through the milk, MMTV can also be introduced by injection into the mammary gland of pubescent mice. To determine whether MMTV injection into the mammary gland resulted in systemic infection of lymphoid cells, we performed PCR analysis on DNA isolated from spleen and Peyer’s patches of mice infected in this manner, using MMTV(C3H) LTR-specific primers. We found that both tissues from the infected mice acquired new proviral copies of MMTV (Fig. 2), indicating that lymphoid cells were infected after virus injection into the mammary gland.

B cells are believed to be the first cells infected during milk-borne MMTV transmission to offspring (3, 8). To determine whether they were needed for MMTV spread within the mammary gland, we injected virus into mice that lacked B cells due to the targeted mutagenesis of their Ig μ-chain gene (strain IGH6/BL6 22)). This gene disruption was available only in C57BL/6 mice, which are H-2^b and do not express the MHC class II I-E molecule. Because of this, most MMTV SAgs, including that encoded by MMTV(C3H), are presented inefficiently and these mice are less susceptible to MMTV(C3H) infection (23). IGH6/BL6 mice were crossed with C3H/HeN mice, which are H-2^k, an MHC haplotype that efficiently presents the MMTV(C3H) SAg (24, 25). F₁-(IgM⁺/H-2^k+) and F₂-generation female mice (IgM⁺H-2^k+, IgM⁺H-2^k−, IgM⁻H-2^k+, IgM⁻H-2^k−; see Fig. 3) derived from these crosses received mammary gland injections of MMTV at 3 wk of age. Four weeks after injection, immune response to the virus was assessed by determination of the percentage of peripheral CD4⁺/Vβ14⁺ T cells. Both the F₁- and F₂-generation IgM⁺H-2^k+ mice were infected, since they deleted about 30% of their Vβ14⁺ SAg-reactive T cells (Table I). In contrast, none of the H-2^k− or IgM⁻ mice showed deletion (Table I), indicating that these mice were not MMTV infected. However, the absence of cognate T cell deletion in the B cell-deficient and H-2^k− (I-E⁻) mice could also be due to inefficient SAg presentation, because of the lack of B cells or appropriate MHC haplotype. To determine whether the mice were infected, we isolated DNA from the spleens of the injected mice and performed PCR specific for MMTV(C3H) proviral DNA. As shown in Figure 4, only the IgM⁺H-2^k+ (lanes A–J) but not the IgM⁺H-2^k− (lanes K–N), IgM⁻H-2^k− (lane O), or IgM⁻H-2^k+ (lanes P–R) mice contained MMTV(C3H) proviruses in their splenocytes. Therefore, lack of deletion of SAg-cognate T cells in IgM⁻ or H-2^k− mice directly correlated with the lack of infection.

To determine whether mammary gland cells of these animals were infected, we isolated RNA from their lactating mammary glands. This RNA was subjected to RNase T₁ protection analysis with the probe specific for the LTR of the MMTV(C3H) exogenous virus. All the F₁ and F₂ generation IgM⁺H-2^k+ (shown only for the F₂ generation) were infected, since they expressed high levels of MMTV-specific RNA in their mammary glands (Fig. 5, lanes A–J). In contrast, no MMTV-specific transcripts were detected in either the IgM⁻H-2^k+ (lanes P–R), IgM⁻H-2^k− (lane O), or IgM⁺H-2^k− mice (lanes K–N). After an overnight exposure of the gel in Figure 5, a very faint signal was seen for two of the four IgM⁺ H-2k⁻ mice (not shown). As was previously reported, H-2^b mice can be infected at a low level (26). One IgM⁺H-2^k+ mouse also showed no mammary gland expression of MMTV(C3H) (Fig. 5, mouse J); however, this mouse did show lymphocyte infection (Fig. 4, bottom panel, IgM⁺H-2^k+, mouse J). Thus, in the absence of B cells, there was no virus spread within the immune system or mammary gland.

One consequence of B cell infection by MMTV is that these cells could function as SAg-presenting APCs that activate cognate T cells. This activation would lead to amplification of virus within the lymphoid compartment. That mice of the wrong MHC haplotype failed to be efficiently infected by mammary gland injection of MMTV indicated that SAg presentation might be a requisite step in the infection of mammary cells. To test directly whether SAg activity and infected lymphoid cells were required for virus spread within the mammary gland, we injected exogenous MMTV(C3H) into the mammary glands of two types of transgenic mice, MTV-ORF and LEL (7, 16). Because the MTV-ORF and LEL transgenic mice both expressed the MMTV(C3H) sag as an endogenous gene, they lack Vβ14⁺ T cells and thus were resistant to milk-borne MMTV(C3H) infection (17, 27).

Two independent experiments were performed. In the first, LEL transgenic females (n = 7) and their nontransgenic littermates (n = 6) were injected with a high virus dose, whereas in the second experiment, a smaller MMTV dose was injected into both LEL (n = 4) and MTV-ORF (n = 6) transgenic mice and nontransgenic (n = 10) littermates (see Materials and Methods) (Fig. 6). After MMTV(C3H) injection directly into the mammary glands of the pubescent mice, we assessed whether the nontransgenic mice were infected by determining their percentage of peripheral CD4⁺/Vβ14⁺ T cells; because the transgenic mice all express the MMTV(C3H) SAg as an endogenous protein, they already lack these T cells (Table II). All nontransgenic mice in both experiments were infected as judged by their deletion of SAg-reactive Vβ14⁺ T cells (Table II).

To determine whether MMTV injection into the mammary gland did result in systemic infection of lymphoid cells, we performed PCR analysis on DNA isolated from spleen and Peyer’s patches of the infected transgenic and nontransgenic mice, using MMTV(C3H) LTR, gag-specific primers. We found that lymphoid tissues from all of the infected nontransgenic mice acquired new proviral copies of MMTV (Fig. 7, LEL n/t and MTV-ORF n/t). In contrast, the lymphoid tissue isolated from the injected transgenic mice showed no evidence of viral infection, except for one LEL transgenic mouse (lane marked LEL p.p.*, Fig. 7). This mouse also developed a mammary tumor.

Two methods were used to determine whether the mammary gland cells of the injected transgenic mice were infected. First, we bred these mice and then examined their nontransgenic offspring for the deletion of Vβ14⁺ T cells. In the first experiment, where a high virus dose was injected, the offspring of five of seven transgenic females showed deletion of their Vβ14⁺ T cells (Table II). These results suggested that MMTV transmission was variable due to low level infection. In support of this, when a lower dose of virus was injected, there was no deletion of this subset of T cells in the nontransgenic offspring of either the LEL or the MTV-ORF transgenic mice (Table II). All offspring of nontransgenic mice that received mammary gland injections of virus showed Vβ14⁺ T cell deletion (Table II).

We also tested directly whether the mammary glands of the injected mice were MMTV infected by isolating RNA from their milk and subjecting it to RNase T₁ protection analysis using a probe specific for MMTV(C3H) transcripts. To ensure that there was ample time for virus spread, all of the mice were analyzed after their fourth pregnancy; we have previously shown, using this assay, that milk-borne infection of mammary gland tissue increases with parity (28). RNA isolated from the milk of the LEL transgenic mice injected with the high virus dose contained little or no detectable MMTV(C3H)-specific RNA, in contrast to the MMTV-injected nontransgenic mice (Fig. 8,A), whereas no viral RNA was detected in either the LEL or MTV-ORF mice that received the lower amount of virus (Fig. 8 B).

This lower viral load had a dramatic affect on MMTV-induced mammary tumorigenesis. Because MMTV integration next to cellular oncogenes is a stochastic event, the higher the viral load, the more rapidly mice develop mammary tumors (13). Both the transgenic and nontransgenic mice from experiment 1 were continuously bred and monitored for mammary gland tumor incidence. The multiparous nontransgenic mice had a 100% mammary gland tumor incidence by the age of 260 days (Fig. 9). At that time, none of the transgenic mice developed mammary tumors and only two of seven transgenic mice developed tumors by 300 days (Fig. 9). All the tumors had acquired exogenous MMTV proviruses as detected by Southern blot analysis (not shown).

Both T and B cells play an essential role in the transmission of milk-borne MMTV, since the mammary glands of mice lacking either of these lymphoid subsets are not efficiently infected (5, 7, 8, 29). B cells are necessary, as initial targets for MMTV infection at minimum, because they are effective SAg-presenting cells for cognate T lymphocytes (8). Similarly, SAg-reactive T cells are required for the amplification and establishment of MMTV infection, at least within the lymphoid compartment (5, 7). Here, we provide proof that a mutant MMTV lacking a functional SAg gene will infect lymphoid cells, but that insufficient amplification of this lymphoid infection blocks viral transmission to the mammary gland.

Although it is now clear how MMTV initiates infection in vivo, how this virus infects mammary gland cells and spreads within this tissue is poorly understood. Expression of integrated MMTV proviruses in mammary cells is regulated by lactogenic hormones. As a result, pregnancy has a dramatic effect on virus load, which also increases with parity (2). The number of infected cells also increases in multiparous animals, showing that there is virus spread within the mammary gland after the initial infection. Without this spread, tumor induction by MMTV is very inefficient. Because retroviral integration is a stochastic event, increased numbers of integration events will improve the chance of insertion near a given cellular oncogene. For example, MMTV-infected virgin mice have a significantly lower virus load that leads to a decreased mammary tumor incidence and an increased tumor latency (14). In addition, MMTV-induced mammary tumors usually contain large numbers of newly integrated copies of exogenous MMTV, indicating multiple reinfection events (10).

Here we addressed the requirement for lymphoid cells in MMTV infection of and spread within the mammary gland. Not surprisingly, based on experiments by others in which MMTV was injected into peripheral tissue (3), we found that lymphoid as well as mammary cells were efficiently infected after the injection of virus directly into the mammary glands of wild-type, pubescent mice. However, when the same mode was used to introduce MMTV into B cell-deficient mice, we found no infection of either lymphoid cells or mammary tissue. Since the B cell-deficient mice had the MHC H-2^b haplotype and thus did not express the I-E molecule required for efficient presentation of the MMTV(C3H) SAg, we crossed them to C3H/HeN mice (H-2^k), an MHC haplotype that presents this SAg. Both I-E positive and I-E negative F₂ mice were generated. This allowed us also to examine the role of MHC haplotype in virus spread within the mammary gland. Neither I-E⁺ nor I-E⁻ B cell-deficient animals showed evidence of mammary gland infection. Moreover, I-E⁻, B cell⁺ mice also showed no, or very limited, mammary gland infection. This led us to speculate that even in the mammary gland, infected B cells were required to present SAg to cognate T cells for efficient infection.

We therefore tested whether mice lacking such cognate T cells could be infected by mammary gland injection of virus. We found that transgenic mice lacking SAg-reactive T cells showed little or no virus spread and, as a result, did not develop mammary tumors even after MMTV was introduced directly into the mammary glands of pubescent mice. These data suggest that the SAg stimulation of T cells is required for efficient infection of mammary glands and consequent tumorigenesis. Even when infection of the mammary gland cells was achieved, as in the case of the LEL transgenic mice injected with a high virus dose, the virus spread within this tissue was less efficient than that which occurred in mice that had SAg-cognate T cells.

These results demonstrate that infected lymphoid cells play a critical role in infection of the mammary gland. One possibility is that virus spread between mammary gland cells cannot be achieved in the absence of infected lymphoid cells. This may be because cell-cell contact between lymphocytes and mammary cells is the most efficient way to deliver virus. The architecture of the murine mammary gland may also affect how virus spreads. Mammary gland ducts consist of two main cell types, an inner region of ductal cells and an outer monolayer of myoepithelial cells surrounded by a basement membrane (2). At puberty, the mammary epithelial cells proliferate, resulting in a lengthening and branching of the ductal tree until the whole mammary fat pad is filled with cells. The terminal end bud drives ductal morphogenesis and most DNA synthetic activity is associated with these buds, while the ducts are relatively quiescent. The cells that occupy the terminal buds are most likely the targets for MMTV during puberty, since as with most other retroviruses, it is thought to require cell division for its propagation (30). If infected lymphocytes are present near these terminal buds, they could increase infection level of these dividing cells.

Similarly, lymphocytes may be involved in the virus spread that occurs in multiparous mice. Once puberty is over, the mammary cells do not divide in the mouse until pregnancy. MMTV preferentially buds from the apical surface of the epithelial cells lining the alveolar lumen (31) so that the virus is released into the milk during lactation. Thus, for MMTV to cause a systemic infection of the mammary gland tissue, it would have to be produced by an infected cell that directed its expression toward subepithelial tissues. B or T cells could fulfill this requirement, since both produce MMTV (9), persist within the tissue, and can be activated by cytokines produced by SAg-activated T cells (32). An additional factor that may come into play is that lymphoid cells can circulate within the tissue, thereby possibly coming into contact with mammary cells at multiple locations.

It is also possible that cytokine production by SAg-stimulated T cells affects the mammary gland tissue directly. For example, cytokines produced by T cells could induce expression of gene products necessary for MMTV infection of mammary gland cells or for mammary tumorigenesis. Since MMTV-infected mice gradually delete SAg-cognate T cells after acquisition of milk-borne virus, this mechanism would require the persistence of such T cells in the mammary gland and not in other peripheral sites. Whether this occurs can be tested by examining the Vβ-repertoire of T cells isolated from the mammary gland and their ability to be activated by SAg.

Taken together, we have established that SAg activation of lymphoid cells is needed not only for the initial stages of MMTV infection, but also for virus spread within the mammary gland. These results demonstrate the importance of different cell types during in vivo virus infections.

We thank Kevin Norman for maintaining the mouse colony in excellent condition, members of our laboratories for helpful discussions, and T. Wrona and Dr. Alexander Chervonsky for critically reading the manuscript.