Bone marrow stromal cell antigen 2 (BST-2) restricts mouse mammary tumor virus (MMTV) replication in vivo

Background Bone marrow stromal cell antigen 2 (BST-2) is a cellular factor that restricts the egress of viruses such as human immunodeficiency virus (HIV-1) from the surface of infected cells, preventing infection of new cells. BST-2 is variably expressed in most cell types, and its expression is enhanced by cytokines such as type I interferon alpha (IFN-α). In this present study, we used the beta-retrovirus, mouse mammary tumor virus (MMTV) as a model to examine the role of mouse BST-2 in host infection in vivo. Results By using RNA interference, we show that loss of BST-2 enhances MMTV replication in cultured mammary tumor cells and in vivo. In cultured cells, BST-2 inhibits virus accumulation in the culture medium, and co-localizes at the cell surface with virus structural proteins. Furthermore, both scanning electron micrograph (SEM) and transmission electron micrograph (TEM) show that MMTV accumulates on the surface of IFNα-stimulated cells. Conclusions Our data provide evidence that BST-2 restricts MMTV release from naturally infected cells and that BST-2 is an antiviral factor in vivo.

In vivo, MMTV first infects antigen presenting cells (APCs) such as B cells and dendritic cells (DCs) at the site of infection [15][16][17][18][19]. MMTV infected APCs present virus-encoded superantigen (Sag) to T cells expressing Sag-specific T-cell receptor (TCR) Vβ chains. This immunological synapse causes stimulation of Sag-reactive T cells and proliferation of lymphocytes thereby promoting virus replication. Both lymphoid and myeloid cells infected with MMTV are capable of producing infectious virus [19] and infected lymphoid cells are necessary for virus spread and mammary carcinogenesis [20,21].
Although MMTV infects and causes mammary cancer in infected mice; nonetheless, infected cells do not produce high virus titer, and time to MMTV-induced cancer is rather long, suggesting that virus replication and spread may be restricted by host factors like BST-2.
Indeed, we demonstrate that BST-2 co-localizes with MMTV Gag and Env, and inhibits MMTV particle release in tissue culture cells. Importantly, MMTV infection of mice was significantly inhibited by BST-2.

Results
IFNa induces BST-2 expression and restricts MMTV release BST-2 expression results in retention of a wide range of virus-like particles (VLPs) assembled in tissue culture [1][2][3][4], and IFNα or IFNγ treatment induces the transcription of the gene encoding BST-2 [1,22]. We hypothesized that IFNα induction of BST-2 in an MMTV producing cell line would enhance BST-2 expression and suppress MMTV release to the culture medium. We treated MMTV-producing GR cells with IFNα or vehicle and then measured BST-2 mRNA and surface protein levels, as well as extra-and intra-cellular viral particles. We observed a significant increase in BST-2 mRNA ( Figure  1A) and protein ( Figure 1B) levels upon IFNα treatment. This increase in BST-2 levels was accompanied by a reduction in the accumulation of MMTV particles in the culture supernatant, with no change in intracellular viral load as assessed by viral RNA (Figures 1C and 1D) and viral protein ( Figure 1E). Additionally, the reduction in particle accumulation in culture supernatant results in lower viral load when such supernatants were used to infect MMTV susceptible cells ( Figure 1F). This data demonstrate that IFNα induces endogenous BST-2 in MMTV infected cells and suggest that BST-2 may be the critical factor limiting the accumulation of MMTV particles in the culture supernatant.
Next, we examined whether BST-2 is responsible for IFNα-dependent inhibition of MMTV release. GR cells stably silenced for BST-2 expression with a lentiviral construct carrying BST-2 specific shRNA or control shRNA were stimulated with IFNα or vehicle. Culture supernatants were collected 24 hours later and subjected to Western blot analysis for MMTV capsid protein p27. The corresponding cells were examined for BST-2 protein (Figure 1G). Results show that while shRNA was able to silence BST-2 and enhance accumulation of viral particles in culture supernatant ( Figure 1H), IFNα stimulation of shRNA cells results in increased cell surface BST-2 protein expression ( Figure 1G) and a reduction in the amount of viral particles accumulated in the culture supernatant (Figure 1H). This result shows that the reduction in the amount of extracellular viral particles observed in IFNα treated cells is linked to BST-2 induction.

IFNa causes accumulation of MMTV structures at the cell surface
In IFNα treated cells, less MMTV particles were detected in the culture supernatant without apparent defect in accumulation of intracellular viral components ( Figures 1E, 1H). As BST-2 was shown to block the release of various enveloped viruses/VLPs from the cells [1,23], we hypothesize that exposure of MMTV-producing GR cells to IFNα will result in accumulation of MMTV virions on the cell surface. To test this, GR cells were stimulated with IFNα or vehicle and 24 hours later, cells were processed for scanning electron microscopy (SEM). In vehicle treated GR cells, 100 nm sized buds were readily observed (Figure 2A). The buds were commonly found on the tip of filamentous structures projecting from the cells surface. IFNα induction drastically increased the number of bud-containing filaments ( Figure 2B). To confirm that the buds were indeed virions, thin sections of GR cells prepared under the same treatment conditions were analyzed with transmission electron microscope (TEM). Electron dense virion structures, about 100 nm in diameter (black arrowheads), were detected at the cells surface ( Figure 2C and 2D). As expected, in IFNα treated cells, particles that are linked to each other were also observed ( Figure 2D). The TEM thus validated that the filamentous structures observed with SEM contain virion(s). Together the electron microscopy data strongly suggest that exposure of MMTV infected cells to IFNα results in retention of MMTV virions on the cell surface.
BST-2 co-localizes with MMTV structural proteins at the cell surface Consistent with its role in tethering virus particles to the surface of the infected cell, BST-2 has been shown to colocalize with structural proteins of other retroviruses along the plasma membrane of cells [1,3]. To examine whether BST-2 co-localizes with MMTV structural proteins, BST-2, rat glucocorticoid receptor (RSVGR), and MMTV proviral genome (HP) were co-transfected into 293T cells. Two days later, cells were processed and immuno fluorescently stained to detect BST-2 and MMTV structural proteins. Results show partial co-localization of MMTV Env ( Figure 3A) or Gag ( Figure 3B) with BST-2 at the cell surface. Our data support previous demonstrations of co-localization of BST-2 with HIV-1 Env in infected cells [24] and co-localization of SIV Env and BST-2 in activated primary rhesus macaque lymphocytes infected with SIV [25]. In addition, our data are in line with previous report of BST-2 co-localization with the Gag proteins of other retro-and filo-viruses [3].
Depletion of endogenous BST-2 enhances MMTV particle release and replication in cell culture To test the hypothesis that BST-2 is responsible for inhibition of MMTV release, we used an RNAi assay to determine whether silencing of endogenous BST-2 expression would lead to higher MMTV particle release and higher

Overexpression of BST-2 inhibits MMTV replication
To confirm our finding that BST-2 restricts MMTV replication in cultured cells, we transiently expressed varying amounts of BST-2 or empty vector in TRH3 (293T cells that stably express MMTV entry receptor transferrin receptor 1). Twenty four hours later, a portion of cells were examined for BST-2 expression  Figure 5B), indicating that BST-2 restricts MMTV replication in a dose dependent manner. Whether lower infection observed in BST-2 transfected cells is a result of cells being infected at low multiplicity or lower number of cells being infected is unclear. To address this question, we first generated TRH3 cells stably expressing BST-2 or empty vector ( Figure 5C) and infected cells with MMTV.gfp purified from MMTV-producer cell line, CGRES6. These cells produce virions derived from a molecular clone of MMTV (pGR102ES) with a gfp gene inserted into its long terminal repeat (LTR) [26]. Infected cells were collected 24 and 96 hours later and examined for level and intensity of gfp expression. Results show that both at 24 and 96 hours after infection, level of gfp expression was significantly lower in BST-2 expressing cells compared to cells expressing empty vector control ( Figure 5D). In addition, gfp mean fluorescence intensity (MFI) was also lower in cells expressing BST-2 at both time points ( Figure 5E). These results suggest that BST-2 potently restricts MMTV replication as evidenced by the larger proportion of cells infected in the absence of BST-2 ( Figure 5D, solid blue line and solid red line) compared to lower number of cells infected in the presence of BST-2 ( Figure 5D, broken blue line and broken red line).

BST-2 specific siRNA depleted endogenous BST-2 in murine lymph nodes
We have demonstrated that loss of BST-2 in tissue culture cells enhances MMTV replication and spread (Figures 4H and 4I lower panels). To examine the role of BST-2 in virus infection in vivo, we first determined whether MMTV target cells and tissues such as dendritic cells (DCs), macrophages (MΦs), lymph nodes and spleens from different mouse strains (C3H/HeN and C57BL/6) that express BST-2 mRNA. We used the mammary epithelial cell line, NMuMG as a reference and show that BST-2 transcript is present in all MMTV targets ( Figure 6A). BST-2 mRNA is higher in bone marrow derived DC (BMDC) compared to bone marrow derived macrophages (BMMΦ), popliteal lymph nodes and the spleen, with a modest difference in BST-2 levels between C3H/HeN and C57BL/6 mice. Since lymph nodes express BST-2, we examined whether BST-2 specific siRNA is capable of silencing BST-2 in vivo. Following inoculation of mice with siRNA or siControl subcutaneously in the hind footpad, we evaluated the levels of BST-2 mRNA and protein and show that in 48 hours, siRNA significantly depleted lymph node BST-2 mRNA to about 7 fold ( Figure 6B) with a modest reduction in protein level ( Figure 6C: upper and lower), and no effect on BST-2 levels of mice inoculated with siControl or no siRNA. We also examined the expression of IFNα, IFNβ, and 2'-5'-oligoadenylate synthetase (OAS) transcripts, and observed that mice treated with siRNA sequences have similar levels of IFNα, IFNβ, and OAS transcripts despite reduced BST-2 mRNA level (Additional file 1). These data suggest that BST-2 specific siRNA reduced BST-2 expression in vivo without eliciting IFN response.

siRNA mediated silencing of endogenous BST-2 enhances MMTV replication and spread in vivo
Local administration of BST-2 siRNA mediates BST-2 depletion in vivo (Figures 6B and 6C). To determine the role of endogenous BST-2 in virus replication in vivo, we administered either a control siRNA, or siRNA that targets BST-2 mRNA to mice, followed by infection with MMTV. Virus replication was significantly higher in siRNA treated mice compared to their siControl or no siRNA counterparts ( Figure 6D). In addition, we observed significantly higher virus load in the spleens of mice that received siRNA ( Figure 6E). These data show for the first time that siRNA is capable of mediating BST-2 knock down in vivo (Figures 6B and 6C); and that loss of BST-2 in the draining lymph node for the site of initial infection results in higher rate of virus replication and dissemination in vivo (Figures 6D and  6E).

Discussion
Here, we describe an important anti-viral function for BST-2 in the mouse. Many in vitro and ex vivo studies have shown that BST-2 inhibits the replication of a number of viruses [1][2][3][4][5][6][7][8][9]. By using MMTV, we directly demonstrate that BST-2 plays a role in virus replication in vivo. Administration of siRNA sequences into mouse footpad results in knock-down of endogenous lymph node BST-2. This assay allowed us to demonstrate that BST-2 functions as a virus restriction factor in a natural host organism. We chose to silence lymph node BST-2 because lymphocytes play a critical role in the in vivo infection with MMTV and other retroviruses such as HIV-1; and in the absence of a knockout model, siRNA provides the best alternative. The significantly higher virus load observed in BST-2 silenced lymph nodes implies that loss of BST-2 enhances virus replication.
BST-2 expression inhibits the egress of HIV-1, HIV-2, Ebola, Marburg and MLV; and we found that it also inhibits the accumulation of MMTV particles in the culture medium. The inhibitory effect of BST-2 on accumulation of extracellular MMTV particles in culture supernatant is consistent with the effect of BST-2 on virus release from the cell surface as has been observed for other viruses [1][2][3][4][5]. The observation that BST-2 colocalizes with MMTV Env and Gag further suggests that BST-2 prevents the accumulation of MMTV particles in culture supernatants. Our EM data support previously documented studies that MMTV buds off from the plasma membrane and microvilli [27]. The cell surface visualization by SEM provides an opportunity to demonstrate that induction of BST-2 with IFNα facilitates accumulation of MMTV on the cell surface. This finding was backed by our protein analyses showing that stimulation of MMTV producer cells with IFNα results in less accumulation of MMTV particles in the culture supernatant.

Conclusion
We have directly demonstrated in this study that loss of BST-2 enhances replication of MMTV in vivo. However, it remains to be determined why mice become infected with MMTV in the presence of the endogenous BST-2. It is worth noting that MMTV has co-existed with mice for over 20 million years [28] and perhaps must have evolved to avoid most hosts' anti-viral defense mechanisms. The findings of this study are important as it has shown that BST-2 function as anti-viral in the host. Based on this report, we are now focused on using MMTV as a model to understand the role of BST-2 in viral pathogenesis.

Ethics statement
All experiments involving mice were performed in accordance to NIH guidelines, the Animal Welfare Act, and US federal law. The experiments approved by the University of Iowa Animal Care and Use Committee (IACUC). Mice were housed according to the policies of the Institutional Animal Care and Use Committee of the University of Iowa.

Molecular constructs
BST-2 plasmids used in the study were previously described [4] and kindly provided by Dr. Paul Bates, University of Pennsylvania, Philadelphia, PA. The MMTV env (Q61) construct [20], MMTV molecular clone HYB PRO (HP) [29], and the rat glucocorticoid receptor (RSVGR) [30] were kindly provided by Dr. Susan Ross, University of Pennsylvania, Philadelphia, PA.
Animals C57BL/6 and C3H/HeN mice were purchased from the Jackson Laboratory and National Cancer Institute (NCI) respectively.

Cell lines
NMuMG (derived from normal mammary tissue of a Namru mouse strain), GR (a mammary carcinoma cell line), Mm5MT (a mammary carcinoma cell line from C3H/HeN mouse strain), 3T3 MEF and 293T cell lines were purchased from American Type Culture Collection (ATCC). TRH3 cells are 293T-mTfR1 (293T cells that stably express MMTV entry receptor transferrin receptor 1 [mTfr1]) and have been previously described [31]. TRH3-BST-2 and TRH3-Vector cells (TRH3 cells stably expressing BST-2 or empty vector). CGRES6-GFP cells (feline kidney stably transfected with a GFP-tagged MMTV proviral construct [26] was obtained from Dr. Susan Ross of the university of Pennsylvania. All cells were maintained according to the suppliers' instructions. Both GR and Mm5MT cells are from MMTV-induced mammary tumor and both produce infectious MMTV particles.
The cells were left in culture with three changes of medium to remove dead cells. By day 10 after transfection, several individual clones of stably transfected cells emerged and clones were picked and expanded. Level of BST-2 expression was examined by FACs analysis. After examination, clones with the highest BST-2 expression were expanded and some cells frozen or used for experiments.
Antibodies, siRNAs, shRNAs, and other reagents Antibodies used in this study includes goat anti-gp52 (Env) and anti-p27 (capsid) previously described [32,33]. Others are unconjugated rat anti-mouse BST-2 clone 129c, Alexa flour 647 conjugated anti-mouse BST-2 clone 129c [34], and FITC conjugated anti-mouse BST-2 clone 927 purchased from eBioscience. Rabbit anti-BST-2 and mouse anti-tubulin were from Abnova and Li-core respectively. The following isotype control antibodies were used: Rat IgG2a and Rat IgG2b, both from eBiosciences. The following secondary antibodies from Li-core were used: RDye 680 donkey anti-mouse, IRDye 680 LT conjugated donkey anti-rabbit, IRDye 800CW donkey anti-goat, and IRDye 800CW conjugated goat anti-rabbit. Rabbit anti-rat Alexa Fluor 594 and chicken anti-goat Alexa Fluor 488 were from Invitrogen. The in vivo ready silencer Select target specific siRNA, control siRNA, and SiPort NeoFX transfection reagent were from Ambion while plasmids for target-specific Mission shRNA, control shRNA, and polybrene were obtained from Sigma. BST-2 shRNA and control shRNA lentiviral particles were produced from plasmids of target-specific Mission shRNA and control shRNA by the University of Iowa Vector Core Facility. Recombinant mouse interferon alpha (IFN-α) was from Miltenyi Biotech. Vector-shield was from Vector Laboratories and Lipofectamine 2000 was from Invitrogen.

Infection of mice
In vivo infection of mice was performed with 50 μl of MMTV.MM5MT inoculated subcutaneously on the hind footpad. At different times after infection, mice were sacrificed and the popliteal lymph node draining the site of infection from each mouse was harvested. Lymphocytes were obtained and used for fluorescenceactivated cell-sorting (FACS) analysis or used for DNA and RNA extraction.
BST-2 knockdown by shRNA and siRNA in cell culture BST-2 shRNA lentiviral particles containing 3 targetspecific constructs and shRNA constructs encoding a scrambled sequence (Santa Cruz Biotechnology, Santa Cruz, CA) or target specific Mission shRNA and a corresponding control shRNA (Sigma) were used to knock down BST-2 gene expression. Briefly, mouse mammary epithelial tumor cell line GR were transduced with control or BST-2 shRNA lentiviral particles at a ratio of 5 infectious units of virus per cell in the presence of 8 μg/mL polybrene. The next day, fresh media was added to cells and incubated for another 24 hours. In some experiments, puromycin at 2 μg/ml was added to the culture medium for the selection of cells that had stably incorporated shRNA. BST-2 gene silencing was confirmed at the mRNA and protein levels by qPCR and FACS. Knockdown of BST-2 by siRNA was achieved using Ambion's in vivo ready silencer Select target specific or siRNA in vivo ready scrambled siRNA. Briefly, 250 pico moles of siRNA sequences were mixed with Lipofectamine 2000 following manufacturer's recommendation and used to transfect NMuMG, GR or MEFs. The next day, fresh medium was added to cells and incubated until assayed for BST-2 and MMTV mRNA. NMuMG and MEFs were passaged and used for detection of BST-2 and virus replication experiments. Experiments were repeated at least 3 times with similar results.

Western blots
Western blots of virus preparations, cell lysates from MMTV infected cells, cells transiently transfected with env and capsid plasmids, or immuno-precipitates of protein complexes were probed with anti-total MMTV, anti-gp52, and anti-p27. The species-appropriate IRDye secondary antibodies were used, followed by detection with the Odyssey Infrared Imaging System (LI-COR Biosciences).
DNA, RNA isolation and real-time quantitative PCR (qPCR) DNA or total RNA was isolated from cells, tissues, and mice lymphocytes using the QIAGEN-QIAamp DNA Mini Kit or RNeasy Mini Kit (Qiagen, Inc.) respectively according to manufacturer's instructions. For cDNA synthesis, equivalent amounts of RNA treated with DNase I (Qiagen, Inc.) were reverse-transcribed with high capacity cDNA reverse transcription Kit (Applied Biosystems, ABI), and the cDNA was amplified with primers specific to BST-2, MMTV.Mm5MT, IFNα, IFNβ, 2'-5'-OAS, and GAPDH. DNA was used to quantitatively detect integrated proviruses [32,33], using primers specific to the MMTV long terminal repeats and puromycin. Semi-quantitative PCR was performed using Veriti 96-Well Thermal Cycler, and real time qPCR was carried out using ABI 7500 FAST thermal cycler (ABI). All amplifications were normalized to GAPDH and data are presented as fold change relative WT or naïve sample depending on experiment. Sequence of primer pairs used is available (Table 1).

Virus release assay
GR cells were induced for virus production with 1 μM of dexamethasone (DEX) [35] and treated with IFNα or vehicle for 24 hours. Culture supernatants from cells were clarified by low speed centrifugation, passed through a 0.45 μm filter, and virions were pelleted through a 20% sucrose layer at 32,000g as previously described (30)(31). Pellets of virus particles and corresponding cell extracts were denatured and analyzed by SDS-PAGE and Western blot assays. In some experiments, virus particles were used for qPCR analysis of viral RNA or to infect TRH3 cells.

Statistical analysis
Statistical analysis of significant differences between experimental groups was tested using paired two-tailed Student's t test, and a p value of 0.05 was considered significant. Error bars represent standard deviations.

Additional material
Additional file 1: BST-2 siRNA did not elicit interferon response in mice. Age-matched C3H/HeN mice were inoculated with siRNA (n = 3), siControl (n = 3), or PBS (WT, (n = 3) subcutaneously on the hind foot pad. Forty-eight hours after inoculation, mice were sacrificed and cells of the draining popliteal lymph node used for total RNA extraction. Quantitative PCR was used to examine mRNA levels of (A) BST-2 (B) IFNα, (C) IFNβ, and (D) 2'-5'-oligoadenylate synthetase (OAS). Data is presented as fold change relative to WT mice. Error bars are standard deviation, and p is significance level. Experiments were performed with 3 mice per group and repeated at least three times with similar results.