Human cytomegalovirus induces TGF-β1 activation in renal tubular epithelial cells after epithelial-to-mesenchymal transition

Abstract

Human cytomegalovirus (HCMV) infection is associated epidemiologically with poor outcome of renal allografts due to mechanisms which remain largely undefined. Transforming growth factor-β1 (TGF-β1), a potent fibrogenic cytokine, is more abundant in rejecting renal allografts that are infected with either HCMV or rat CMV as compared to uninfected, rejecting grafts. TGF-β1 induces renal fibrosis via epithelial-to-mesenchymal transition (EMT) of renal epithelial cells, a process by which epithelial cells acquire mesenchymal characteristics and a migratory phenotype, and secrete molecules associated with extracellular matrix deposition and remodeling. We report that human renal tubular epithelial cells infected in vitro with HCMV and exposed to TGF-β1 underwent morphologic and transcriptional changes of EMT, similar to uninfected cells. HCMV infected cells after EMT also activated extracellular latent TGF-β1 via induction of MMP-2. Renal epithelial cells transiently transfected with only the HCMV IE1 or IE2 open reading frames and stimulated to undergo EMT also induced TGF-β1 activation associated with MMP-2 production, suggesting a role for these viral gene products in MMP-2 production. Consistent with the function of these immediate early gene products, the antiviral agents ganciclovir and foscarnet did not inhibit TGF-β1 production after EMT by HCMV infected cells. These results indicate that HCMV infected renal tubular epithelial cells can undergo EMT after exposure to TGF-β1, similar to uninfected renal epithelial cells, but that HCMV infection by inducing active TGF-β1 may potentiate renal fibrosis. Our findings provide in vitro evidence for a pathogenic mechanism that could explain the clinical association between HCMV infection, TGF-β1, and adverse renal allograft outcome.

Figure 1. HCMV replication in HFFs and HK-2 cells.

HFF (A) and HK-2 cells (B) were infected with HCMV strain TR at MOI of 1, cells and media harvested daily, and viral titers determined by DEAFF assay. A parallel set of HK-2 cells were infected with HCMV and exposed to recombinant active TGF-β1 (C), and cells and media analyzed as for HK-2 cells. A separate set of cells were uninfected or infected with HCMV strain TR at MOI of 1, cell pellets harvested at day 5 post-infection, and western blotting for viral proteins performed using HCMV hyperimmune globulin (insets). HFFs supported logarithmic viral replication with viral progeny in both cells and media, whereas HK-2 supported linear productive infection in cells only. Upright triangles, cells; inverted triangles, media.

Figure 2. HCMV infected renal tubular epithelial cells undergo epithelial-to-mesenchymal transition (EMT) after TGF-β1 exposure.

(A) Primary human renal tubular epithelial cells were uninfected (top row) or infected with HCMV strain TR (bottom row), without (left column) or with recombinant human active TGF-β1 (raTGF-β1) to induce EMT (right column). Cells were stained using a monoclonal antibody against HCMV IE1 (mab63-27) and an isotype-specific AlexaFluor 488-conjugated secondary antibody (green nuclei), and co-stained with AlexaFluor 594-conjugated phalloidin (red) and Topro3 (blue nuclei) nuclear stain. Images were collected by confocal microscopy using similar exposure time and identical gain. Cells at baseline had epithelioid morphology with concentric structural actin cytoskeleton, both without (top left) and with (bottom left) HCMV infection, whereas cells after raTGF-β1 stimulation showed elongated mesenchymal morphology indicative of EMT in both HCMV uninfected (top right) and infected (bottom right) cells. (B) Primary human renal tubular epithelial cells were untreated, or were infected with HCMV strain TR at MOI of 1 and/or treated with raTGF-β1, lysed, and subjected to western blotting using antibodies against HCMV IE1, e-cadherin, vimentin, or actin. Nuclear extracts from cellular lysates were also subjected to western blotting for phospho-SMAD2. Both uninfected (IE1 negative) and HCMV infected (IE1 positive) cells expressed E-cadherin but not vimentin at baseline, but after raTGF-β1 stimulation, both uninfected and infected cells lost E-cadherin expression and demonstrated both vimentin expression and SMAD2 phosphorylation, indicative of EMT in both. (C) HK-2 cells were untreated, or were infected with HCMV strain TR at MOI of 1 and/or stimulated with raTGF-β1 to induce EMT, lysed, total RNA reverse transcribed to cDNA, and cDNA analyzed for presence of extracellular matrix associated mRNAs using the SuperArray extracellular matrix PCR array. Results were normalized to GAPDH expression and quantitated as fold-change compared to mRNA levels in uninfected, unstimulated HK-2 cells. HCMV infected cells (grey bars) induced some mRNA transcripts of fibrogenic molecules, but only at less than 10-fold induction. Both uninfected cells stimulated with raTGF-β1 (hatched bars) and HCMV infected cells stimulated with raTGF-β1 (black bars) demonstrated induction of many fibrogenic molecules represented in this array, consistent with induction of EMT in both uninfected and HCMV infected cells. Transcripts upregulated in HCMV infected HK-2 cells after raTGF-β1 stimulation (denoted with asterisks) were confirmed by individual RT-PCR assays using both HK-2 cells and primary renal tubular epithelial cells (Figure S2).

Figure 3. HCMV infected renal tubular epithelial cells induce active TGF-β1 production after EMT.

Immortalized renal tubular cells, HK-2 (A, B), or primary human renal tubular epithelial cells (C, D) were untreated, or were infected with HCMV strain TR (HCMV TR −/+) and/or treated with raTGF-β1 (raTGF-β1 −/+) at 15 ng/ml (0.6 nM) to induce EMT. Cells were washed 3 times and re-incubated in fresh media not containing raTGF-β1. Supernatants were assayed for de novo active and total TGF-β1 production using a TGF-β1 responsive luciferase bioassay. Only the HCMV infected cells stimulated with raTGF-β1 induced production of active TGF-β1 in both HK-2 and primary cells. (E) HK-2 cells were infected with HCMV strain TR at MOI of 1 and/or stimulated with raTGF-β1 at 15 ng/ml as described, and supernatants were assayed using the Quantikine ELISA. Only the HCMV infected, raTGF-β1 stimulated cells produced detectable active TGF-β1 in this assay. (F) HK-2 cells were infected with HCMV strain TR at MOI of 1 and/or stimulated with raTGF-β1 at 15 ng/ml as described, and a blocking antibody against TGF-β1 was added to cell cultures simultaneously with raTGF-β1. After 48 hours, cells were washed and re-incubated in fresh media, and luciferase assay for active TGF-β1 performed. The blocking antibody reduced TGF-β1 activation in a dose-dependent manner. Legend: (*) p>0.05, ns; (***) p<0.01.

Figure 4. Effect of TGF-β1 and viral input upon TGF-β1 activation.

(A) HK-2 cells were infected with HCMV strain TR at MOI of 1 and stimulated with increasing doses of raTGF-β1 from 1–20 ng/ml (0.04 nM–0.8 nM), washed, and assayed for de novo TGF-β1 production as described. HCMV infected HK-2 cells induced active TGF-β1 production in proportion to the amount of stimulating raTGF-β1. (B) HK-2 cells were infected with HCMV strain TR at increasing MOI (2-8) with raTGF-β1 stimulation at 15 ng/ml (0.6 nM), and luciferase bioassay performed to quantitate active TGF-β1 production. Active TGF-β1 production increased with increasing HCMV MOI. (C) HCMV AD169 strain BADrUL131 at MOI of 1 was used to infect HK-2 cells prior to raTGF-β1 stimulation, and TGF-β1 luciferase bioassay performed. Similar to results for HCMV strain TR, HCMV AD169 strain BADrUL131 induced active TGF-β1 production after EMT. (D) HCMV strain TR at MOI of 1 was inactivated by UV irradiation (HCMV TR UV+) and used to infect HK-2 cells prior to raTGF-β1 stimulation, and active TGF-β1 measured by luciferase bioassay. Irradiated virus failed to induce active TGF-β1 production after EMT. Legend: (*) p>0.05, ns; (***) p<0.01.

Figure 5. MMPs are expressed and a MMP complex forms in HCMV infected cells after EMT.

(A) HK-2 cells were infected with HCMV strain TR at MOI of 1 and incubated with inhibitors, GM6001, aprotinin, anti-thrombospondin 1(α-TSP), or anti-α_vβ₆ integrin (α- αvβ6), prior to stimulation with raTGF-β1 at 15 ng/ml (0.6 nM), washed, and TGF-β1 luciferase bioassay performed for active TGF-β1. Results were compared to those from uninfected, unstimulated HK-2 cells (HCMV TR-/raTGF-β1-) as well as HK-2 cells infected with HCMV and stimulated with raTGF-β1 (HCMV TR+/raTGF-β1+). Both GM6001 and aprotinin significantly inhibited active TGF-β1 production. Legend: (**) p<0.05; (***) p<0.01. (B) HK-2 cells were untreated, or were infected with HCMV at MOI of 1 and/or treated with raTGF-β1. Cell lysates were subjected to gelatin zymography (zymogram) and western blotting using anti-MMP-2 (anti-MMP-2). Pro- and active MMP-2 could be detected only in HCMV infected, raTGF-β1 stimulated cells. (C, D) HK-2 cells were treated as in (A), but lysates were either subjected directly to western blotting for TIMP-2, MT3-MMP, MT1-MMP, or actin (C) or incubated with mouse anti-MMP-2 followed by protein A-agarose, and immunoprecipitated proteins subjected to western blotting using rabbit anti-MMP-2, anti-TIMP-2, anti-MT3-MMP, and anti-MT1-MMP. TIMP-2 and MT3-MMP immunoprecipitated with MMP-2 only in HCMV infected, raTGF-β1 stimulated cells. (E) HK-2 cells were transfected with MMP-2 shRNA plasmid (MMP-2), or a control scrambled plasmid (Ctrl). Cells were infected with HCMV strain TR at MOI of 1 and/or stimulated with raTGF-β1 at 15 ng/ml. Supernatants were subjected to luciferase assay for active TGF-β1 (top panel). A portion of the cell pellets were subjected to western blotting for MMP-2, GFP, and actin (middle panel). RNA was extracted from the remainder of the cell pellets and RT-PCR performed for MMP-2 (bottom panel), with results depicted as fold change between raTGF-β1 exposed and non-exposed transfections. These assays showed that MMP-2 shRNA transfection reduced active TGF-β1, MMP-2 protein and mRNA; the control transfections stimulated with raTGF-β1 did induce active TGF-β1, MMP-2 protein and mRNA.

Figure 6. Effects of viral inhibitors and HCMV IE1 or IE2 ORFs upon TGF-β1 production.

(A) HK-2 cells were infected with HCMV strain TR in the presence of increasing concentrations of ganciclovir or foscarnet prior to raTGF-β1 stimulation, and active TGF-β1 measured by luciferase bioassay (left panel). These inhibitors did not affect active TGF-β1 production by HCMV infected cells after EMT. DNA was extracted from cell pellets and quantitative DNA PCR for HCMV gB was performed (right panel), and confirmed efficacy of the viral polymerase inhibitors. (B) HK-2 cells were transiently transfected with expression plasmids containing lacZ either alone or co-transfected with plasmids containing either HCMV IE1, IE2, or UL55 (gB), followed by raTGF-β1 stimulation and TGF-β1 luciferase bioassay. Results from the TGF-β1 luciferase bioassay were normalized to transfection efficiency as measured by β-galactosidase activity in cell lysates (dns). Both the IE1 and IE2 constructs, but not the gB construct, induced active TGF-β1 production after EMT. Legend: (***) p<0.01. (C, D) Cells were transfected with plasmids containing either HCMV IE1, IE2, or UL55, and cell pellets were subjected to western blotting for MMP-2 and actin (C), or RT-PCR for MMP-2 mRNA (depicted as fold change between raTGF-β1 exposed and non-exposed transfections). Only the IE1 or IE2 transfected cells exposed to raTGF-β1 induced MMP-2 protein and mRNA.