Histopathology

After 6 days in culture, E15 SMGs typically progress to the Terminal Bud Stage: bilaminar, stratified epithelial ducts and single-layered epithelial terminal buds display distinct lumina, and are embedded in a loosely-packed stroma sparsely populated with fibroblasts (Fig. 1A, B). E15 + 6 SMGs infected with 100,000 PFU mCMV show a marked decline in branching epithelia; duct epithelia is pseudostratified and poorly organized; duct, and perhaps bud, lumina are greatly dilated (Fig. 1C, D). There is a several-fold increase in cellularity of the stroma, particularly at the periphery of the SMG (Fig. 1C). This zone of atypia contains clusters of large, basophilic, pleiomorphic cells, with high nuclear-to-cytoplasmic ratios, prominent nuclei and nucleoli, and frequently inclusion bodies pathognomonic of mCMV infection (Fig. 1D).

As the infection progresses, the stromal cells are composed of two distinct cell types: large basophilic round cells and smaller eosinophilic cells. In cross-section, the basophilic cells appear to be emigrating from the epithelia (Fig. 2A, B) and invaginating into ductal lumina (Fig. 2A, B). By 12 days in culture, there is a further decline in branching epithelia, and the further dilated lumen of the pseudostratified ducts are filled with eosinophilic cells, some living, some apoptotic (Fig. 2C, D). Throughout the gland the stroma is hypercellular and no longer resembles mesenchyme (Fig. 2C, D). The stromal cell type at E15 + 12 is unusual: large polygonal cells with large, darkly staining, nuclei containing prominent nucleoli, densely eosinophilic cytoplasm, and the frequent presence of inclusion bodies (Fig. 2D). This plump, eosinophilic cell type with mitochondrial hyperplasia (Fig. 2D, insert) is suggestive of oncocytic metaplasia [24, 25]. Other than in some intraductal epithelial cells, there is scant evidence of cell death in epithelial or stromal cells. This is consistent with the well-documented CMV suppression of cell death [26–33].

Dose response and time course

The severity of the morphologic changes in SMG development is mCMV dose-dependent (Fig. 3). At 10,000 PFU, SMGs are only moderately infected at the organ periphery, and there is an approximate 25% decrease in gland size (P < 0.01). At 50,000 PFU, SMGs are almost totally infected at the periphery, and about a 40% decrease in gland size (P < 0.001); this differs significantly from the smaller dose (P < 0.01). There are no significant differences between 50,000, 100,000, and 200,000 PFU with respect to SMG size. This is consistent with the fact that viral titers in SMGs exposed to any of these three doses, are not significantly different at E15 + 6 (F_2,9 = 1.48; P > 0.25).

At 100,000 PFU, the negative effect on branching morphogenesis is evident early. There is a significant 50% reduction in SMG size (P < 0.001) at E15 + 3, and this reduction significantly increases (P < 0.02) through E15 + 12 (Fig. 4A, C, E), as epithelial structures are increasingly replaced by metaplastic stromal cells (Fig 4B, D, F), and viral titers significantly increase with increasing time of exposure (P < 0.05). Immunodetection of mCMV immediate early protein 1 (IE1) reveals that through the first 6 days of culture, mCMV infection is localized to mesenchymal cells (fibroblasts)(Fig 4B, D); by 12 days, the infection is confined mostly to metaplastic stromal cells, but may also be found in a few ductal epithelial cells and in lumen-filling cells (Fig. 4F–J). These data are supported by β-galactosidase localization, an indicator gene product expressed from the virus we used (data not shown). It is important to note that at least through 6 days of culture, the infected fibroblasts are quite distant from much of the affected branching epithelia (Figs. 1C, 4B, 4D), suggesting paracrine factor diffusion from virus-infected areas. In this regard, it is also important to note that, although there is a progressive spread of infection over the 12-day time course, the majority of affected stromal cells remain uninfected throughout the observation period.

Cell proliferation

The clear decline in branching morphogenesis and increase in stromal cellularity is directly correlated with cell proliferation (Fig. 5). At E15 + 6, Terminal Bud Stage SMGs typically exhibit evidence of proliferation mostly in the branching epithelia, and very little in the mesenchyme (Fig. 5A). In mCMV infected SMGs, this pattern is reversed (Fig. 5B); mostly mesenchymal, not epithelial, cells are proliferating.

IL-6 signaling plays an important dual role in SMG development: branching morphogenesis through the SHP-2/Ras mitogenic pathway and ductal maturation through the STAT3 pathway [34]. In uninfected SMGs, IL-6 is expressed in proliferating epithelial cells surrounding the forming lumina (Fig. 5A, C), consistent with its mitogenic pathway. In mCMV infected SMGs, however, IL-6 is more intensely expressed in the non-proliferating epithelial cells surrounding dilated lumina (Fig. 5B, D). This finding suggests an upregulation of IL-6 expression and a premature shift to the STAT3 pathway for epithelial maturation. This proposition is supported by our gene expression studies below (Table 1) and by premature SMG terminal differentiation (Fig. 5E, F), as determined by the expression of mucin (Muc10), a SMG-specific marker of epithelial histodifferentiation [35, 36]. To wit, in mCMV-infected E15 + 6 SMGs, we see an increase in mucin protein translocated from cytoplasm to epithelial apical surfaces surrounding dilated lumina. In this regard, it is important to note that by E15 + 12, many of the metaplastic stromal cells are also expressing mucin (Fig. 5H–I), consistent with an epithelial origin.

Cell characterization

SMG epithelium is characterized by cytoplasmic cytokeratin and adherens junctions (Fig. 6). In uninfected SMGs, cytokeratin is exclusively seen in branching epithelia (Fig. 6A). In mCMV infected SMGs, cytokeratin is expressed in the abnormal, dilated ductal structures composed of pseudostratified squamous and lumina-filling epithelia, and more diffusely in the metaplastic stromal cells adjacent to this epithelia (Fig. 6B).

E-cadherin and p120 are important constituents of adherens junctions. In uninfected SMGs, E-cadherin is immunolocalized exclusively to epithelial plasma membranes (Fig. 6C). With mCMV infection, the pseudostratified lumina epithelia display a decline in expressed E-cadherin in cells facing the lumina and in cells facing the stromal space; there is also subsequent cytoplasmic localization of E-cadherin in metaplastic stromal cells adjacent to ductal epithelia (Fig. 6D). In uninfected SMGs, p120 is expressed adjacent to plasma membranes in epithelia (Fig. 6E); in mCMV infected SMGs, p120 is expressed in all epithelia, but markedly less in those cells which appear to be emigrating from ductal epithelia into the stroma (Fig. 6F).

Given these observations (Fig. 6B, D, F), as well as the recent study by Davis and Reynolds [37] of embryonic SMG epithelial dysplasia and E-cadherin deficiency in p120 null mice, it is reasonable to suggest that the lumen-filling cells are completely derived, and the metaplastic stormal cells are at least partially derived, from epithelium.

Recent studies indicate unexpected links between fibronectin expression, COX-2 expression, and β-catenin nuclear localization [38, 39], as well as between CMV infection and COX-2 expression [40–42]. Thus, we investigated the cellular distribution of fibronectin (FN), α5β1 integrin (FN receptor), COX-2 and β-catenin in Terminal Bud Stage SMGs, with and without mCMV infection (Fig. 7). Normally, FN is primarily immunolocalized in ductal and terminal bud epithelia basement membranes (Fig. 7A). With mCMV infection, there is a remarkable shift in FN distribution: FN surrounds individual metaplastic stromal cells, and there is a notable decline in basement membranes (Fig. 7B). A similar change in α5β1 integrin is seen with mCMV infection (Fig. 7B, insert). COX-2 is not expressed in uninfected stromal cells (not shown), but is localized in the cytoplasm of infected fibroblasts (Fig. 7C) and metaplastic stromal cells (Fig. 7D, E) in mCMV infected SMGs. This dramatic change is associated with a change in Cox2 transcript level (Table 1). Concurrently, β-catenin exhibits a most significant shift in location with mCMV infection of SMGs. In Terminal Bud Stage SMGs, β-catenin (an adherens junction constituent) is immunolocalized to the cytoplasm adjacent to epithelial plasma membranes (Fig. 7F). With mCMV infection, β-catenin is still expressed in intact epithelia, but, more importantly, there is de novo expression in the nuclei of metaplastic stromal cells adjacent to that epithelium (Fig. 7G, H). Nuclear localization of β-catenin is indicative of its alternate function as a transcription factor.

Transcription changes

Functional studies in our laboratory and elsewhere have demonstrated that embryonic SMG organogenesis is regulated through interconnected growth factor, cytokine, and transcription factor-mediated signaling pathways, including EGF, TGF-β, IGF, IL-6, Shh, and FGFs [22, 23, 34, 43–49]. The hub of this complex network of parallel and broadly-related pathways is NFκB [44]. In the present study, we employed real-time quantitative PCR to determine transcription changes with mCMV infection (Table 1). The 27 genes chosen include sentinel genes from the signaling network [44], as well as those that characterize cellular changes with mCMV infection (Figs. 5, 6, 7).

The results of these measurements are presented in Table 1. The relative expression ratio (R) is the mean increase or decrease in gene expression in mCMV infected glands compared to uninfected glands. The variation of R is calculated as gene expression noise (η); η is statistically equivalent to the coefficient of variation and ranges from 0 to 1 [50]. The value of η reflects fluctuations in the level of promoter-binding or the abundance of a particular transcription factor, variation in post-transcriptional modifications, and a host of other stochastic events that are characterized as intrinsic or extrinsic noise [50]. In the absence of prohibitively large sample sizes, as η approaches 1 it becomes extremely difficult to detect small, but important, true differences in gene expression levels, should they exist. Of the 27 genes measured, 6 exhibited statistically significant changes in expression with mCMV infection: Nfkb2, Relb, Il6, Stat3, Erk1, Cox2. These results are entirely consistent with previous studies [41, 44, 51, 52] and with our immunohistochemical studies (Figs. 5, 6, 7).

mCMV replication and pathology

There are two key questions about the relationship of viral replication to the subsequent SMG pathology. First, is mCMV replication necessary to initiate the pathogenesis? Second, is mCMV replication necessary to maintain the pathogenesis? To answer these questions, we utilized acyclovir, an antiherpesviral nucleoside active against mCMV [53].

In the first experiment, E15 SMG explants were infected for 24 hrs with mCMV and then cultured in the presence or absence of acyclovir for an additional 5 days (Fig. 8). Acyclovir treatment, as expected, surpresses mCMV replication (Fig. 8A, B); also, acyclovir treatment results in histologically normal SMGs (Fig. 8C–F) with normal patterns of cell proliferation (data not shown). This outcome is associated with a highly significant mean reduction in SMG mCMV titer from 2.1 × 10⁴ PFU to 20 PFU (P < 0.01). The replication cycle is not completed by 24 hrs of infection when antiviral treatment was initiated, so it appears that completion of the viral replication cycle beyond DNA replication is critical to the initiation of SMG pathogenesis.

In the second experiment, E15 SMG explants were infected with mCMV for 72 hrs, which allows for complete viral replication, followed by culture for an additional three (E15 + 6) or nine (E15 + 12) days in the presence or absence of acyclovir (Fig. 9). At E15 + 6, there is an evident inhibition of mCMV with acyclovir treatment (Fig. 9A); concomitantly, the SMGs are histologically near normal with only mildly dilated ducts, increased branching, and mostly normal stroma (Fig. 9C, D). By E15 + 12, there is only sparse evidence of detectable mCMV in acyclovir-treated SMGs (Fig. 9B), with histologically normal epithelial branching and the relatively rare appearance of metaplastic stromal cells (Fig. 9E, F). As is normally seen (Fig 5A), cell proliferation is almost exclusively seen in the branching epithelia (data not shown). Thus, eventhough SMGs were subjected to unimpeded mCMV replication for 72 hrs, the impact of reduced replication and spread was sufficient to dramatically reduce viral cytopathology and associated developmental changes. Embryonic cellular memory [54] appears insufficient to maintain progressive pathogenesis.

NFκB and mCMV-induced pathogenesis

Evidence indicates that the early CMV protein, IE1, activates canonical NFκB (p50/RelA) by inducing its nuclear localization, rather than transcriptional/translational upregulation [55, 56]. Activated NFκB binds to the NFκB recognition sites of viral ie genes and a large array of host cell genes. It is reasonable, then, to expect that inhibition of NFκB nuclear localization would moderate mCMV-induced pathogenesis, either through an impact on the virus or host cells. In fact, the results of just such an experiment are quite the reverse of expected (Fig. 10). mCMV infection of E15 SMGs in the presence of SN50, a cell permeable inhibitor of canonical NFκB nuclear translocation, exhibits an early accelerated viral replication (Fig. 10A, B) and SMG dysplasia. At E15 + 6 (Fig. 10E, F), the pathology is more characteristic of the typical mCMV-infected E15 + 12 SMGs (e.g. Fig. 9E).

These results are consistent with recent studies in human and mouse fibroblast cells grown in culture suggesting that canonical NFκB has paradoxical roles in infected cells [56]. Further, Sonenshein's group [51, 52] presents evidence that non-canonical RelB dimers may be more critical to de novo host tissue gene expression and pathology. Here we find an upregulation of Relb and Nfkb2 (Table 1), as well as the nuclear localization of RelB (Fig. 11). Indeed, compensatory upregulation of Relb and Nfkb2 in the presence of SN50 may ultimately prove to be the explanation for accelerated viral replication and pathology (Fig. 10).

Embryonic SMG culture system and mCMV infection

Female B10A/SnSg mice, obtained from Jackson Laboratories (Bar Harbor, ME), were maintained and mated as previously described [22]; plug day = day 0 of gestation. Timed-pregnant females were sacrificed on gestation day 15 (E15) and embryos were dissected in cold phosphate-buffered saline (PBS). All animal studies were conducted with the approval of the appropriate committees regulating animal research. An Animal Review Board and a Vivaria Advisory Committee review all applications to ensure ethical and humane treatment.

E15 SMG (mostly Canalicular Stage) primordia were cultured using a modified Trowell method as previously described [34, 44, 45]. The defined media consisted of BGJb (Invitrogen Corporation, Carlsbad, CA) supplemented with 0.5 mg ascorbic acid/ml and 50 units/ml penicillin/streptomycin (Invitrogen Corporation), pH 7.2; media was changed daily. Since notable differences in SMG branch number and size are usually seen among littermates, we employed a paired-design which compared right and left glands (treated and control) from each embryo.mCMV infection: on day 0, E15 SMGs were incubated with 10,000 to 200,000 plaque-forming units (PFU)/ml of lacZ-tagged mCMV RM427⁺ [86] for 24 hrs and then cultured in virus-free BGJb defined media for an additional 2–11 (E15 + 3 to E15 + 12) days. Explants were collected and processed for whole mount morphology, routine histology, immunohistochemistry or multigene expression.

For whole mount morphological and size analyses, SMGs were photographed using a Wilde dissecting microscope at 25× and the area of each gland was determined using Image-Pro Version 4.0 (Media Cybernetics, Silver Spring, Maryland). The following groups were analyzed: 10,000 PFU: E15 + 3 (n = 7), E15 + 6 (n = 18); 50,000 PFU: E15 + 3 (n= 11), E15 + 6 (n = 12); 100,000 PFU: E15 + 3 (n = 17), E15 + 6 (n = 29), E15 + 12 (n = 5); 200,000 PFU: E15 + 3 (n = 6); E15 + 6 (n = 8). The significance of area differences between viral-infected and control SMGs was determined by Student t-test using a matched-pairs design.

For histological analysis, SMGs were fixed for 4 hrs in Carnoy's fixative at 4°C or overnight in 10% neutral buffered formalin at room temperature, embedded in paraffin, serially-sectioned at 8 μm and stained with hematoxylin and eosin as previously described [22, 49]. For each experimental protocol, 3–20 SMGs/mCMV concentration/day were analyzed.

mCMV analysis

To obtain a measure of mCMV infection, we assayed for β-galactosidase (lacZ) activity, viral titer/gland and localization of viral immediate early (IE1) proteins. β-galactosidase (β-gal) staining: Paired E15 + 3, E15 + 6 and E15 + 12 SMGs were fixed for 20 min at room temperature in 0.2% gluteraldehyde in PBS, washed 3 times in rinse solution (0.005% Nonidet P-40 and 0.01% sodium deoxycholate in PBS), stained for 4–6 hrs at room temperature in standard staining solution (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl₂, 0.4% X-gal in PBS), rinsed twice in PBS and microphotographed at 25×. β-gal whole mounts were then dehydrated through graded alcohols, embedded in paraffin, serially-sectioned at 8 μm and counterstained with eosin. Viral titer: E15 + 3 and E15 + 6 SMGs were collected, stored at -80°C in DMEM shipping media [87] and titered by plaque assay as previously described [87]. IE1 distribution: SMGs were fixed in Carnoy's fixative, serially-sectioned at 8 μm, and incubated overnight with anti-IEI as previously described [88]. Controls consisted of sections incubated with mouse IgG alone. For each experimental protocol, 3–10 SMGs/mCMV concentration/day were analyzed.

Cell proliferation assay

Cell proliferation was determined by the localization of PCNA (proliferating cell nuclear antigen) as previously described [44]. Briefly, paired E15+ 3, E15 + 6 and E15 + 12 SMGs were fixed in Carnoy's fixative, serially-sectioned, incubated with anti-PCNA using the Zymed mouse PCNA kit (Invitrogen Corp.) and counterstained with hematoxylin. In this experiment, the cytoplasm appears blue and PCNA-positive nuclei appear dark brown. Three sections per slide and 3–5 SMGs per group were analyzed.

Antibodies and immunostaining

Immunolocalization was conducted essentially as previously described [34, 44, 45]. The following monoclonal (Mab) and polyclonal (Pab) antibodies were used: Mab ATP-synthetase (Mitochondria marker) (#MAB3494, Chemicon International, Temecula, CA); Mab α5-integrin (# 103801, Biolegend, San Diego, CA); Mab β1-integrin (# 102201, BioLegend); Pab β-catenin (# AB19022, Chemicon International); Pab cytokeratin (# ab9377, Abcam Inc., Cambridge, MA); Pab COX-2 (# 160106, Cayman Chemical Company, Ann Arbor, MI); Mab E-cadherin (# 610181, BD Biosciences, San Jose, CA); Pab FN (# F3648, Sigma-Aldrich Corp., St. Louis, MO); Pab IL-6 (# sc-1265, Santa Cruz Biotechnology, Inc., Santa Cruz, California); Pab p120 (# sc-1101, Santa Cruz Biotechnology, Inc); Pab mucin [35]. For immunofluorescent analyses, Pab's were incubated with biotin-labeled rabbit IgG or anti-goat IgG (MP Biomedical, Aurora, OH) and then with Alexa-Fluor-labeled streptavidin (Invitrogen Corporation). Mab's were incubated with biotin-labeled anti-mouse IgG or anti-rat IgG (Jackson Laboratories, West grove, PA) and then with Alexa-Fluor-labeled streptavidin. Hamster antibodies (β-1 integrin) were incubated in FITC-labeled hamster IgG (Biolegend). Nuclei were counterstained with DAPI (Invitrogen Corporation). Immunohistochemistry was conducted essentially as previously described, using the Chemicon Tissue Staining Kit. Sections were viewed on a Zeiss Axioplan Microscope and photographed using 10×, 20× and 40× objectives. Confocal images were obtained using a Zeiss LSM-510 laser scanning confocal/multiphoton microscope (Carl Zeiss Inc. Thornwood NY) and the accompanying LSM version 3.2 image acquisition and analysis software. Cells were imaged using a plan-neofluor 1.3 numerical aperture, 40× objective lens. Alexa-Fluor- and FITC-labeled images were captured using a 488 nm Argon laser for excitation and a 505–530 nm band-pass filter to detect emission. DAPI images were captured using a 800 nm Mira titanium sapphire laser for excitation and a 390–465 nm band-pass filter to detect emission.

CMV replication and pathology

Acyclovir, a synthetic purine nucleoside analgogue, is a highly selective agent for CMV with low toxicity to the host cell [53]. Acyclovir sodium (100 mg/20 ml) was purchased from American Pharmaceutical Partners, Inc (Schaumberg, Il). Since acyclovir has been shown at high doses to be teratogenic to rat embryos [89–91], we first determined the highest dose that is not teratogenic to E15 SMGs in vitro. Paired E15 SMGs were cultured in acyclovir (10 μg/ml, 20 μg/ml, 50 μg/ml or 100 μg/ml) or control medium for 3 (E15 + 3) or 6 days (E15 + 6) and acyclovir and control SMGs were compared by whole mount and histological analyses. Since doses ≥ 20 μg/ml acyclovir inhibited SMG branching and development, we employed 10 μg/ml acyclovir in all future experiments. Since no significant size differences were seen in SMGs infected with 50,000, 100,000 or 200,000 PFU mCMV, we infected SMGs with 50,000 PFU mCMV in this set of experiments. In the first experiment, we compared paired E15 SMGs infected with 50,000 PFU mCMV for 24 hrs and then cultured in control medium +/- 10 μg/ml acyclovir for a total of 6 days (CMV v. CMV + Acy) in culture; controls consisted of paired E15 SMGs cultured in control medium for 24 hrs and then in control medium +/- 10 μg/ml acyclovir for a total of 6 days (CONT v. Acy). E15 + 6 SMGs were collected and analyzed for whole mount morphology, mCMV infection (β-gal staining, titer determination), histopathology, and cell proliferation as described above. In the second experiment, E15 SMG explants were infected with mCMV for 72 hrs and cultured for an additional 3 (E15 + 6) or 9 (E15 + 12) days in the presence or absence of 10 μg/ml acyclovir (CMV v. CMV + Acy3). Controls consisted of E15 SMGs cultured in control medium (CONT) or control medium to which 10 μg/ml acyclovir was added after 72 hrs (Acy3). E15 + 6 and E15 + 12 SMGs were analyzed for whole mount morphology, mCMV infection (β-gal staining), histopathology, and cell proliferation. For each assay, 3–15 SMGs/group/day were analyzed.

SN50 inhibition of canonical NFκB nuclear translocation

The cell permeable peptide SN50 (Biomol Research, Plymouth Meeting, PA) has been shown to inhibit canonical NFκB translocation into the nucleus. Previous studies in our laboratory have demonstrated that 100 μg/ml SN50 results in significant inhibition of SMG development [44]. In this set of experiments, we compared E15 + 6 SMGs infected with 50, 000 PFU mCMV and cultured in the presence or absence of 100 μg/ml SN50 for the entire culture period; concurrent control and SN50-treated E15 + 6 SMGs were also compared. As an additional control experiment, we cultured mCMV-infected SMGs in the presence or absence of SN50M, the control peptide; no differences were seen between E15 + 6 mCMV-infected and mCMV-infected + SN50M explants. The SMGs were analyzed for mCMV infection (β-gal staining) and histopathology; 3–10 explants/group were analyzed.

RT-PCR

Paired E15 + 6 SMGs infected with 100,000 PFU CMV or cultured in BGJb control medium were pooled (10–12 SMGs/sample) and stored at -80°C in Trizol (Invitrogen Corporation). The glands were homogenized using lysing matrix D (FastRNA Pro Green kit, Q-Biogene, Morgan Irvine, CA) and RNA was extracted using the Trizol protocol according to manufacturer. One microgram RNA was reverse transcribed into first strand cDNA using ReactionReady™ First Strand cDNA Synthesis Kit: C-01 for reverse transcription (Superarray Biosciences, Frederick, MD). The kit uses random primers (hexamers) and MMLV reverse transcriptase to reverse transcribe the entire population of RNA in an unbiased manner. Real time quantitative PCR was conducted with a BioRad iCycler^® using primers and templates mixed with the PA011 master mix (RT² Real-Time™ SYBR Green/Fluorescein PCR master mix, Superarray Bioscences). The primer sets used were prevalidated to give single amplicons and purchased from Superarray Biosciences: Bcl2 (#PPM02918A-24), Caspase3 (#PPM02922A-22), c-myc (#PPM02924A-24), Cdh1 (#PPM03652A-24), Ctnnb1 (#PPM03384A-24), Cox2 (#PPM03647A-24), cyclinD1 (#PPM02903A-24), cyclinD2 (#PPM02900A-24), Egfr (#PPM03714A-24), Fn1 (#PPM03786A-24), Il6 (#PPM03015A-24), Jnk1 (#PPM03234A-24), Itga5 (#PPMO3609A-24), Itgb1 (#PPM03668A-24), Lef1 (#PPM05441A-24), Mapk3 (#PPM03585A-24), Mdm2 (#PPM02929A-24), Nfkb1 (#PPM02930A-24), Nfkb2 (#PPM03204A-24), p53 (#PPM02931A-24), Rela (#PPM04224A-24), Relb (#PPMO3202A-24), Stat3 (PPM04643A-24), TGFb1 (# PPM02991A-24), Tgfb2 (#PPM02992A-24), Tgfb3 (#PPM02993A-24), Tnf (#PPM03113A-24). Primers were used at concentration of 0.4 microM. The cycling parameters were 95°C, 15 min; 40 cycles of (95°C, 15 sec; 55°C, 30–40 sec and 72°C, 30 sec). Specificity of the reactions was determined by subsequent melting curve analysis. RT-PCRs of RNA (not reverse transcribed) were used as negative controls. GAPDH was used to control for equal cDNA imputs and the levels of PCR product were expressed as a function of GAPDH. The relative fold changes of gene expression between the CMV-infected and control glands were calculated by the 2^-ΔΔCT method. For each gene of interest, we analyzed 3–5 independent samples. Significant departures from the hypothesis that there are no differences between CMV-infected and control glands were determined by Student t-test.