Although γherpesvirus infections are associated with enhanced lung fibrosis in both clinical and animal studies, there is limited understanding about fibrotic effects of γherpesviruses on cell types present in the lung, particularly during latent infection. Wild-type mice were intranasally infected with a murine γherpesvirus (γHV-68) or mock-infected with saline. Twenty-eight days postinfection (dpi), ∼14 days following clearance of the lytic infection, alveolar macrophages (AMs), mesenchymal cells, and CD19-enriched cell populations from the lung and spleen express M3 and/or glycoprotein B (gB) viral mRNA and harbor viral genome. AMs from infected mice express more transforming growth factor (TGF)-β1, CCL2, CCL12, TNF-α, and IFN-γ than AMs from mock-infected mice. Mesenchymal cells express more total TGF-β1, CCL12, and TNF-α than mesenchymal cells from mock-infected mice. Lung and spleen CD19-enriched cells express more total TGF-β1 28 dpi compared with controls. The CD19-negative fraction of the spleen overexpresses TGF-β1 and harbors viral genome, but this likely represents infection of monocytes. Purified T cells from the lung harbor almost no viral genome. Purified T cells overexpress IL-10 but not TGF-β1. Intracellular cytokine staining demonstrated that lung T cells at 28 dpi produce IFN-γ but not IL-4. Thus infection with a murine γherpesvirus is sufficient to upregulate profibrotic and proinflammatory factors in a variety of lung resident and circulating cell types 28 dpi. Our results provide new information about possible contributions of these cells to fibrogenesis in the lungs of individuals harboring a γherpesvirus infection and may help explain why γHV-68 infection can augment or exacerbate fibrotic responses in mice.
- B cell
- mesenchymal cell
- interstitial lung disease
idiopathic pulmonary fibrosis (IPF) is an interstitial pneumonia with no known cause characterized by variable inflammation and progressive fibrosis leading to impaired lung function. Patients experience progressive loss of lung function with an average time of survival of 3–5 yr postdiagnosis (1). Although the pathogenesis remains unclear, spatial and temporal variance of fibrotic lesions suggests a repeated stimulus (perhaps by a virus) causes lung injury over the course of the disease (10).
The Epstein-Barr virus (EBV) was positively identified by PCR and immunohistochemistry in 41% of IPF patients compared with 10% of control subjects (40). In a 2003 study, 97% of diagnosed IPF patients tested positive for one or more of four common human herpesviruses, EBV, cytomegalovirus (CMV), human herpesvirus 7 (HHV-7), and human herpesvirus 8 (HHV-8), compared with only 36% of control patients (43). It should be noted that other studies did not find an association between herpesviral infection and IPF, and it is not clear whether these discrepancies represent geographical or technical differences (53, 54). Coincidence of IPF with herpesvirus infection could be secondary to immunosuppressive therapy and does not establish a mechanism for herpesviruses in IPF pathogenesis.
Murine γherpesvirus-68 (γHV-68) is a good model for human pulmonary herpesvirus infection due to similarities to the primate herpesviruses in sequence and pathogenicity (51). In IFN-γR−/− mice, a model used to mimic the T helper type 2 (Th2) environment seen in the lungs of IPF patients, γHV-68 causes chronic inflammation and progressive interstitial fibrosis (27). The virus alone is not sufficient to induce fibrosis in wild-type mice, but its presence in the lung pre- or postfibrotic stimulus increases inflammation and collagen deposition in murine models (16, 20, 49). Lytic γHV-68 in the lungs at the time a fibrotic stimulus (bleomycin) is administered significantly increased inflammation and the amount of collagen measured in BALB/c mice (16). Additionally, our laboratory (20) has shown that infection with γHV-68 14 days after administration of a different fibrotic stimulus, FITC, exacerbates fibrosis measured 7 days postinfection (dpi). Furthermore, long-term latent γHV-68 can also augment the response to FITC or bleomycin; we (48) have observed significant increases in lung collagen when infection occurred 70 days before a fibrotic stimulus. Interestingly, an increase in the number of inflammatory cells accompanies increased fibrosis in each of these models, but how these inflammatory cells contribute to the augmentation is unclear.
Intranasal infection with γHV-68 results in lytic infection of alveolar epithelial cells (AECs) followed by infection of a variety of cell types involved in the antiviral immune response including macrophages, dendritic cells, and B cells (29). After the active lytic infection has been cleared, AECs, macrophages, dendritic cells, and B cells have been shown to harbor latent γHV-68 in the lungs (9, 41, 48). It has also been demonstrated that herpesviruses can infect mesenchymal cells, and these cells are often used for plaque assays for viral quantification (20).
Recent studies suggest potential γherpesvirus-mediated mechanisms for increased fibrosis involving AECs and macrophages. AECs latently infected with γHV-68 express increased levels of CCL2 and CCL12 mRNA. Latently infected AECs also produce higher levels of cysteinyl leukotrienes and the potent profibrotic cytokine transforming growth factor (TGF)-β1 (48). These results are consistent with human studies demonstrating that EBV upregulates TGF-β1 in primary human AEC lines and can influence epithelial-to-mesenchymal transition in human AEC lines through alteration of CUX1/Wnt signaling (18). Additionally, increased levels of ER stress proteins BiP, EDEM, and XBP-1 were noted in AECs from familial and sporadic IPF patients as well as colocalization of herpesviral peptides with XBP-1 (14). Regarding macrophages, IFN-γR−/− mice infected with γHV-68 showed increased levels of alternatively activated macrophage proteins Ym1/2 and Fizz-1 in bronchoalveolar lavage (BAL) as well as arginase-1 in whole lung lysates 180 dpi, suggesting that induction of an M2 macrophage phenotype may contribute to the virally induced fibrosis in these Th2-biased mice (26).
Although our (48) work has demonstrated the development of a profibrotic phenotype in latently infected AECs in wild-type mice, no studies to date have addressed the potential contribution of other inflammatory and resident cells to the profibrotic alterations induced by γHV-68 in wild-type mice. Thus we chose to study changes 28 dpi during viral latency following the clearance of acute lytic infection (7, 42). We found that a variety of circulating and resident cell types harbor latent virus and express higher levels of profibrotic or proinflammatory mediators.
MATERIALS AND METHODS
Male C57BL/6 mice (2–4 mo) were purchased from The Jackson Laboratories (Bar Harbor, ME). All procedures were approved by the University of Michigan Committee on the Use and Care of Animals.
After mice were anesthetized with a low dose of ketamine and xylazine, 5 × 104 plaque-forming units of γHV-68 (American Type Culture Collection, Manassas, VA) diluted in 20 μl of sterilized PBS were administered intranasally.
FITC-induced fibrosis and fibrosis measurements.
FITC (28 mg/ml) was dissolved in sterile saline and sonicated for 30 s, and 50 μl was administered intratracheally in anesthetized mice as previously described (24). Fibrosis was determined at indicated time points via hydroxyproline assay, which is a surrogate for collagen deposition (24).
Cytokines and chemokines were measured in cell culture supernatants using DuoSet ELISA Development System kits (R&D Systems, Minneapolis, MN) following the manufacturer's instructions. For analysis of TGF-β1, supernatants were first acidified to allow for measurements of total TGF-β1.
Semiquantitative real-time RT-PCR.
Semiquantitative real-time RT-PCR was performed with ABI PRISM 7000 thermocycler (Applied Biosystems, Foster City, CA) using a previously described protocol (4). Primers and probes, listed in Table 1, were made using Primer Express Software (Applied Biosystems). All samples were run in triplicate.
Analysis of viral genome loads.
DNA was prepared from cells isolated from mock-infected or γHV-68-infected mice using the QIAGEN DNeasy Blood & Tissue Kit (Valencia, CA), and PCR was performed to detect the glycoprotein B (gB) viral coding sequence as previously described (30). Values were compared with a standard curve consisting of gB plasmid DNA diluted at known copy numbers. Reported values were normalized to 100 ng of input DNA for each reaction and represent the copy number in mock-infected mice (background) subtracted from the virus-infected samples. For gB DNA analysis, the forward primer was 5′-GGCCCAAATTCAATTTGCCT-3′, the reverse primer was 5′-CCCTGGACAACTCCTCAAGC-3′, and the probe was 5′-6FAM-ACAAGCTGACCACCAGCGTCAACAAC-TAMRA-3′.
Alveolar macrophage isolation and culture.
Alveolar macrophages (AMs) were extracted from lungs by BAL as previously described (4) and cultured in serum-free media at 5 × 105 cells per well in 24-well plates for 24 h before collection of cell supernatants and isolation of RNA or DNA.
Mesenchymal cell isolation.
Whole lung was minced and cultured for 14 days to enrich for mesenchymal cells in complete media supplemented with 10 μM cidofovir (Gilead Sciences, Foster City, CA) to prevent viral reactivation and cell lysis. Then, mesenchymal cells were cultured in serum-free media on 6-well plates at 4 × 105 cells per well for 24 h before collection of cell supernatants and isolation of RNA or DNA.
Splenic cell enrichment.
To enrich for B cells, splenocytes were incubated with biotinylated CD19 antibody (BD Biosciences, San Jose, CA) followed by MACS Streptavidin MicroBeads. The cells were then separated on a magnetic column (Miltenyi Biotec, Auburn, CA). CD19-enriched cells were considered B cells. The CD19-negative cells were enriched for T cells, monocytes, and dendritic cells. Both populations were plated on 6-well plates at 5 × 106 cells per well in serum-free media for 24 h before cell supernatants were collected and RNA or DNA was isolated.
Lung B and T cell enrichment.
Cells were extracted from lung tissue by collagenase and DNase digest as previously described (11). B cells were purified using biotinylated CD19 antibody (BD Biosciences). T cells were isolated using the Pan T Cell Isolation Kit from Miltenyi Biotec. The cells were sorted using the MACS Streptavidin MicroBead system (Miltenyi Biotec) and cultured on 96-well plates at 1 × 106 cells per 200 μl in serum-free media for 24 h before cell supernatants were collected and RNA or DNA was isolated.
Intracellular staining for lung T cell cytokines.
Whole lungs were prepared for flow cytometry by collagenase digestion, as described previously (11). For intracellular cytokine staining, cells were first stimulated with PMA (0.05 μg/ml; Sigma, St. Louis, MO) and ionomycin (0.75 μg/ml; Sigma) for 6 h in the presence of GolgiStop protein transport inhibitor (BD Pharmingen). Cells (2.5 × 106) were then stained using fluorochrome-conjugated antibodies against the cell surface markers CD45, CD4, and CD8 (BD Pharmingen). Next, cells were fixed, permeabilized, and stained with anti-IFN-γ and anti-IL-4 (BD Pharmingen). To enumerate lymphocyte subsets, gates were first set on CD45-expressing cells followed by gating on the lymphocyte-sized subset.
Complete media were DMEM (Lonza, Walkersville, MD) with 10% fetal bovine serum (Fisher Scientific, Pittsburgh, PA), 1% penicillin-streptomycin (GIBCO/Invitrogen, Carlsbad, CA), 1% l-glutamine (Fisher Scientific), and 0.1% amphotericin B (Lonza). Serum-free media were DMEM with 1% bovine serum albumin (Sigma), 1% penicillin-streptomycin, 1% l-glutamine, and 0.1% amphotericin B.
Methods of statistical analysis.
Statistical significance was measured by Student's t-test using GraphPad Prism 5 software. Data represent means ± SE, and P < 0.05 was considered significant. Error bars represent the SE between cell culture wells.
γHV-68 infection can augment fibrotic responses.
To demonstrate that γHV-68 infection can augment fibrotic responses, mice were first infected with γHV-68 or mock-infected on day 0. On day 15, mice received an intratracheal injection of FITC, and fibrosis was measured on day 36. This time point represents 36 dpi and 21 days postfibrotic insult. Figure 1A demonstrates that a prior infection with γHV-68 significantly augments the fibrotic response to FITC. Similarly, in an exacerbation model, in mice given FITC on day 0 that were subsequently infected with γHV-68 on day 14 and harvested for fibrosis measurements at day 28 or 35, the viral infection significantly augments the fibrotic response (Fig. 1B). These data are consistent with our (20, 48) earlier reports of the ability of γHV-68 infection to augment or exacerbate fibrotic responses at various time points. Because these fibrotic alterations were noted at time points beyond the lytic replication of γHV-68 in the lung (20), we were curious to determine whether latent infection was associated with profibrotic changes in various cell types. Thus we chose to analyze a variety of resident and circulating cell types in the lung or spleen 28 dpi.
Twenty-eight dpi, AMs are infected and produce more TGF-β1, TNF-α, CCL2, CCL12, and IFN-γ.
Mice were infected with γHV-68 or mock-infected on day 0. On day 28, AMs were harvested by BAL and cultured for 24 h before supernatants and RNA or DNA were collected. AMs isolated from mice 28 dpi expressed significantly higher levels of total TGF-β1 (P < 0.0001), TNF-α (P < 0.001), CCL2 (P < 0.0001), CCL12, and IFN-γ (P < 0.0001) compared with AMs from mock-infected mice (Fig. 2A). Expression of IL-12, IL-4, and IL-13 were not detectable from AMs of either group. Herpesvirus latency is often defined by the expression of latency-associated genes at higher levels than lytic-associated genes when preformed, lytic virus is not present (35, 50). We (48) have previously demonstrated that preformed lytic virus is not present in the lungs by 14 dpi. To ascertain latent infection, expression of the viral genes gB (a lytic gene) and M3 (a gene expressed during both lytic and latent infection) (35) was examined in AMs by real-time RT-PCR. M3 gene expression was higher than gB gene expression in 15 of 21 cell culture wells tested, indicating persistent and predominantly latent infection of AMs 28 days after the initial γHV-68 infection (Fig. 2B). To formally prove that γHV-68 genome was present in AMs purified from latently infected mice, AMs were harvested on day 28 postinfection, and DNA was prepared for real-time PCR analysis for the gB gene. Figure 2C demonstrates that ∼2,000 copies of gB DNA were detectable in 100 ng of DNA isolated from AMs on day 28 postinfection.
Twenty-eight dpi, AMs are classically activated.
It was previously shown that alternative activation of macrophages occurs in the lungs of IFN-γR−/− mice in the presence of a long-term γHV-68 infection (26). However, the upregulation of IFN-γ expression noted above suggested a classic activation of the AMs. To test this, we measured markers of classic [inducible nitric oxide synthase (iNOS)] and alternative (arginase-1, Ym1/2, and Fizz-1) activation by real-time RT-PCR in AMs isolated 28 dpi. iNOS transcript levels were significantly higher in AMs from infected mice compared with controls (P < 0.001), whereas expression of alternative activation markers were not different (Fig. 3). Interestingly, expression of heparin-binding EGF-like growth factor (HB-EGF), a cytokine associated with fibrosis in some systems (6, 31, 46), was also elevated in AMs from infected mice.
Forty-two dpi, latently infected mesenchymal cells produce more TGF-β1, CCL12, and TNF-α.
To isolate mesenchymal cells, lungs were harvested 28 dpi, and lung minces were grown for 14 days in complete media containing cidofovir before protein expression and mRNA expression were measured at a time point that corresponds to 42 dpi. We could not isolate mesenchymal cells in the absence of cidofovir due to viral reactivation in vitro, which led to cell destruction. TGF-β1 (P < 0.0001), CCL12 (P < 0.0001), and TNF-α (P < 0.05) were expressed at significantly higher levels by mesenchymal cells harvested from γHV-68-infected mice compared with mesenchymal cells from mock-infected mice (Fig. 4A). CCL2 was expressed by mesenchymal cells from both groups, but viral infection did not change the degree of expression. Expression of IFN-γ, IL-12, IL-4, and IL-13 was not detected from mesenchymal cells of either group. Gene expression of M3, but not gB (Fig. 4B), was detected in all cell culture wells of mesenchymal cells from γHV-68-infected mice. Figure 4C demonstrates that mesenchymal cells harbor ∼2.8 × 108 copies of viral genome per 100 ng of input DNA, confirming that mesenchymal cells harvested from γHV-68-infected mice are latently infected 42 dpi following in vitro treatment with cidofovir. The complete lack of expression of the lytic viral protein gB is likely a reflection of inhibition of viral replication by cidofovir.
Twenty-eight dpi, latently infected CD19-enriched splenic and lung cell populations make more TGF-β1.
Spleens were harvested from mice 28 dpi, and single cell suspensions were magnetically enriched for B cell fractions using anti-CD19-conjugated beads. Flow cytometry analysis of isolated cells confirmed that ∼93.1% of cells isolated from the mock-infected mice and 87.3% of cells isolated from the virus-infected mice expressed CD19. CD19-enriched cells from the spleens of infected mice expressed significantly increased levels of TGF-β1 and significantly decreased levels of IL-6 compared with uninfected controls (Fig. 5A). IL-10 was expressed by splenic CD19-positive cells, but viral infection did not significantly change the degree of expression. Expression of CCL2, CCL12, TNF-α, IFN-γ, IL-4, and IL-13 was not detected from CD19-positive cells of either group. Both M3 and gB transcripts were detectable in the CD19-positive cells harvested from infected mice, but M3 transcript levels were higher than gB, suggesting predominantly latent infection (Fig. 5B). When viral genome loads were analyzed, splenic B cells contained ∼187 copies of gB DNA per 100 ng of input DNA (Fig. 5C). Because B cells are a circulating cell type, we wanted to confirm the upregulation of TGF-β1 in B cells isolated from the lungs 28 dpi. Lungs were removed and subjected to collagenase and DNase digestion. Isolated leukocytes were then further enriched following CD19 staining and magnetic sorting. Flow cytometry analysis confirmed that cells isolated from mock-infected mice were ∼84.3% CD19+, whereas cells isolated from virus-infected mouse lungs were 70.4% pure. CD19-positive cells isolated from the lungs of infected mice expressed a less pronounced but significant increase in TGF-β1 production as well (Fig. 5D). Real-time RT-PCR demonstrated latent viral gene expression predominated in CD19-enriched cells from the lung similar to what was seen in the spleen (Fig. 5E). Latent infection of the lung B cells was confirmed by measuring viral genome loads (Fig. 5F). Interestingly, viral genome loads in lung B cells were higher (∼781 copies/100 ng DNA) than in the spleen at day 28 postinfection.
Twenty-eight dpi, splenic CD19-negative cells show increased levels of TGF-β1 expression.
In the spleen, we collected the CD19-negative fraction of cells as a source enriched for T cells, myeloid cells, and dendritic cells. These CD19-negative cells from spleens of infected mice produce more TGF-β1 and CCL2 than CD19-negative cells from uninfected controls (Fig. 6A). CCL12, IL-6, and IL-10 were expressed by CD19-negative cells from both groups, but viral infection did not significantly change the degree of expression. Expression of TNF-α, IFN-γ, and IL-13 were not detected from CD19-negative cells of either group. Viral gene expression analysis by real-time RT-PCR indicates a modest increase in the levels of M3 over gB mRNA transcripts in the CD19-negative spleen fraction (Fig. 6B). The levels of viral genome in this population of cells averaged 206 copies per 100 ng of input DNA.
T cells from lungs of latently infected mice produce IL-10 more than TGF-β1.
We next wanted to purify T cells from the lung to more carefully measure viral load and cytokine production. For these experiments, lungs were digested on day 28 following mock or γHV-68 infection, and lung T cells were enriched using the Pan T Cell Isolation Kit (Miltenyi Biotec) in which T cells are isolated by depletion of nontarget cells. Flow cytometry analysis verified that T cells isolated from the lungs of mock-infected mice were 84.3% CD3+, whereas those isolated from virus-infected mice were more pure (98.5%). Contrary to what was seen in CD19-negative splenic cells, TGF-β1 production was not increased. In fact, the trend in combined experiments demonstrated a decrease in TGF-β1 production (Fig. 6D). This may reflect the fact that IL-10 levels were significantly elevated in lung T cells from latently infected mice (Fig. 6D). Genome load in these highly purified T cells from the lung was extremely low and even undetectable in half of the samples (Fig. 6E). Consequently, mRNA analysis was not performed.
Lung T cells produce IFN-γ but not IL-4 at day 28 postinfection.
Previous studies have demonstrated that γHV-68 infection can cause fibrosis in Th2-biased IFN-γR−/− mice (27). Thus we wanted to determine whether the T cells in the lungs of latently infected mice produced IL-4. Collagenase digestions were performed on the lungs of mock- or virus-infected mice at day 28 postinfection, and cells were analyzed by flow cytometry for surface expression of CD4 or CD8 and intracellular staining of IL-4 or IFN-γ. Figure 7A shows the representative flow cytometry plots, and Fig. 7B shows the results of five total mice that were analyzed in each group. Viral infection strongly induced production of IFN-γ in both the CD4 and CD8 T cell subsets even during latency, but there was no induction of IL-4 expression. To verify that our intracellular cytokine staining was adequate to detect IL-4 expression, we stained lung T cells from allergen-challenged mice and could detect increased intracellular IL-4 in CD4 T cells (Supplemental Fig. S1, available in the data supplement online at the AJP-Lung Cellular and Molecular Physiology web site). Thus the profibrotic effects of γHV-68 infection can occur in the presence of a mixed cytokine response in lung T cells. T cells from lungs of latently infected mice overexpress IL-10 and IFN-γ but not IL-4 or TGF-β1.
There is limited understanding about fibrotic effects of γherpesviruses on cell types resident or circulating in the lung, particularly following the clearance of the acute lytic infection. Because of its ability to persist in the lung, herpesvirus infection can be a potential cause of chronic injury. It is possible that persistent reactivation of the virus to a lytic state may lead to chronic AEC destruction and persistent immune activation. Recently, however, we (48) have demonstrated that latent infection in the absence of reactivation can also promote fibrogenesis in response to a second insult. For this reason, we became interested in better understanding the potential profibrotic consequences that a latent infection may have on both resident and recruited cells in the lung. AMs, mesenchymal cells, and B cells expressed a latent viral gene expression signature 28 dpi, 2 wk after the clearance of initial lytic infection. We observed novel cell-specific increases in profibrotic and proinflammatory mediators in all cell types other than lung T cells. Lung T cells from latently infected mice strongly expressed IL-10 and IFN-γ, and AMs were classically activated. These data provide new information about possible contributions of these circulating and resident cell types to fibrogenesis in the lungs of individuals harboring a γherpesvirus infection and suggest that viral-induced augmentation of fibrosis does not require a strongly biased Th2 environment.
Using an animal model of γHV-68 infection before or following fibrotic stimulation with FITC, we demonstrate that γHV-68 infection on day 0 can augment the fibrotic response to FITC injected on day 15 when measured at day 36 (Fig. 1A). Additionally, infection with γHV-68 on day 14 post-FITC can exacerbate the fibrotic response when measured on day 28 or 35 (Fig. 1B). Thus γHV-68 infection can enhance or exacerbate fibrogenesis measured at time points beyond the clearance of the initial lytic infection (20). These results are consistent with our (20, 48) earlier observations made at different time points and prompted our investigations of the profibrotic effects of latent infection on various resident and circulating cell types in the lung.
Our results demonstrate that in wild-type mice, AMs remain infected for at least 28 dpi and produce higher total protein levels of the potent profibrotic factor TGF-β1 as well as CCL2, CCL12, TNF-α, and IFN-γ than AMs from uninfected controls. Because the majority of the AMs analyzed by real-time PCR express higher levels of the latent viral gene M3 than the lytic viral gene gB in the absence of preformed virus 28 dpi, the AMs are predominantly latently infected. This was confirmed by demonstrating the presence of viral genome in DNA isolated from AMs (Fig. 2). Some AM samples did express higher levels of gB than M3, however, and some gB gene expression is present in all of the AMs tested, suggesting that γHV-68 may maintain infection of various cell populations by periodically reactivating to the lytic phase. Our data agree with previous studies that demonstrated that AMs harbor viral infection for 14–90 dpi (9). Previous studies in Th2-biased mice lacking the IFN-γR have demonstrated that γHV-68 infection is associated with persistent reactivation and the accumulation of TGF-β1-producing AMs 180 dpi in areas of moderate and severe fibrosis (27). The TGF-β1-producing macrophages in IFN-γR−/− mice were shown to be alternatively activated. These findings were intriguing given that the IPF lung has been described as a Th2-skewed environment (52) and suggested that the Th2 environment was necessary for the fibrotic effects of γHV-68. However, these results could not explain the observation that γHV-68 latent infection was able to augment fibrotic responses in wild-type mice, particularly in mice with strong antiviral IFN-γ responses (Ref. 48 and Figs. 1 and 7). Interestingly, our studies demonstrate that AMs isolated from γHV-68-infected wild-type mice 28 dpi produce more total TGF-β1 than controls despite being classically activated (Fig. 3) and producing IFN-γ with no detectable IL-4 or IL-13 (Fig. 2). AM overproduction of TGF-β1 as well as TNF-α, CCL2, CCL12, and HB-EGF are also consistent with our (20) previous observation that IL-4 and IL-13 are not required for γHV-68 infection to exacerbate established fibrosis in wild-type mice.
AMs isolated from γHV-68-infected wild-type mice express factors that are known to be profibrotic. For example, infusion of recombinant TNF-α promotes whereas neutralization limits silica-induced lung fibrosis (33). Similarly, transgenic expression of TNF-α in murine type II cells induces a profound fibrotic phenotype (21). HB-EGF has been shown to be profibrotic in models of kidney and cardiac fibrosis (6, 31, 46). Additionally, HB-EGF expressed in lung macrophages (13) and epithelial cells (55) has been shown to act as a fibroblast mitogen. Finally, γHV-68 infection of AMs also upregulates profibrotic CCL2 and CCL12, potent chemotactic ligands for recruitment of inflammatory cells and fibrocytes during pulmonary fibrosis. CCL2 mRNA expression is increased in the lungs of patients with IPF (2), and increased expression of CCL2 protein in BAL fluid of patients with IPF predicts that they die within 5 yr of diagnosis, often due to acute exacerbation (37). Furthermore, CCL12, a murine homolog of human CCL2, is responsible for fibrocyte recruitment in mice (25), and CCL2 has been shown to promote collagen synthesis by fibrocytes (24).
One possible explanation for the increased production of TGF-β1 by AMs in latently infected mice is that AMs produce TGF-β1 in response to ingestion of apoptotic AECs secondary to viral-mediated AEC injury. To determine whether AMs from mice with another form of AEC injury would display a similar phenotype, we utilized a transgenic mouse expressing the diphtheria toxin (DT) receptor under the control of the SpC promoter (SpC-DTR mice) (39). Delivery of DT to these mice results in specific damage to the type II AECs and the development of fibrosis. To mimic the timing of viral infection, SpC-DTR mice were treated with DT for 14 days before AMs were isolated on day 28, and total TGF-β1 production was measured (Supplemental Fig. S2). Interestingly, AMs from SpC-DTR mice did not show significant upregulation of TGF-β1. Thus we believe that TGF-β1 secretion by latently infected AMs is directly attributable to viral infection and not simply the induction of AEC damage.
We next wanted to compare the fibrotic phenotype of AMs from latently infected mice with that of AMs isolated from mice treated with bleomycin, a known inducer of fibrosis within the lung (23). Notably, AMs isolated from mice 28 days after intratracheal administration of bleomycin upregulate TGF-β1, CCL2, and TNF-α to the same degree that γHV-68 does 28 dpi (Supplemental Fig. S3). This suggests that in mice, the presence of γherpesvirus prompts AMs in the lung to have an expression profile of fibrotic mediators very similar to the profile AMs have during known fibrotic insults. The emergence of this profibrotic profile in lung resident AMs could help explain how γherpesvirus has served as a cofactor and exacerbating agent in mouse models of pulmonary fibrosis (16, 20, 27, 43, 48).
In addition to the pivotal role of mesenchymal cells as producers of extracellular matrix proteins during fibrosis, mesenchymal cells also contribute to inflammation and fibrosis through mediator secretion. We demonstrate that latently infected mesenchymal cells isolated from lung digests 28 dpi make significantly more total TGF-β1, CCL12, and TNF-α after an additional 14-day in vitro isolation period compared with mesenchymal cells isolated from uninfected mice (Fig. 4). Mesenchymal cell populations from infected and uninfected mice were treated with cidofovir during the isolation period to prevent viral reactivation and cell destruction in the cultures from infected mice. Real-time RT-PCR analysis of mesenchymal cells 42 dpi confirmed that the mesenchymal cells were latently infected after the cidofovir treatment because M3 was expressed at significantly higher levels than gB and viral genome loads were extremely high. The cidofovir treatment itself slowed the growth of mesenchymal cells sufficiently so that it was not possible to separate fibrocytes from fibroblasts at the end of the culture period. Therefore, we cannot say for sure whether these profibrotic alterations would be seen equally in both fibroblasts and fibrocytes. It is especially noteworthy that the profibrotic mediator signature is seen in these cells despite a pharmacological blockade of viral replication. These data provide strong support for the concept that long-term latent infection (even in the total absence of viral reactivation) can result in durable profibrotic mediator secretion in infected cells.
Generally considered the major reservoir of long-term latent γHV-68 (8), CD19-positive populations isolated from the spleen and lungs had significantly higher levels of M3 and gB gene expression 28 dpi than AM, mesenchymal cell, or CD19-negative populations (compare gB and M3 gene expression relative to β-actin in Figs. 2 and 4⇑–6). The CD19-positive cells may serve as a source of chronic reinfection of lung resident cell types after clearance of the original lytic infection (41). It is interesting to note that the lung B cells seem to harbor more viral genome than do splenic B cells (Fig. 5), although we cannot entirely rule out that this difference is due to other cell types contaminating each isolation. CD19-positive cell populations from the lungs and spleen express more total TGF-β1 28 dpi than the same cell populations from mock-infected mice. CD19-positive cells from the spleen also produce IL-10 at the same levels as controls and less IL-6 than controls. We were somewhat surprised that splenic B cells from γHV-68-infected mice did not overproduce IL-10 since two previous studies have suggested that γHV-68 infection can induce IL-10 production in B cells (32, 38). In the study by Peacock and Bost (32), IL-10 mRNA was shown to be strongly induced in T cells and macrophages from the spleen at 15 dpi but was much weaker in B cells. We confirmed the upregulation of IL-10 production in T cells from lungs of latently infected mice (Fig. 6D). Additionally, Siegel et al. (38) have demonstrated that IL-10 is elevated in the serum of γHV-68-infected mice 14 dpi and that the γHV-68 M2 protein, when transduced into B cells, can induce IL-10 secretion. Both of these studies suggest that latent γHV-68 can augment IL-10 production and that B cells may contribute. In our studies performed 28 dpi, we did not see statistically elevated expression of IL-10, thus we believe this discrepancy is best explained by the difference in timing of when our experiments were performed.
A case has been made for the existence of regulatory B cells that produce TGF-β1 and/or IL-10 (22, 44), but little is known about B cell production of TGF-β1, particularly in the lung. Regulatory B cells may dampen the antiviral immune response by balancing Th1/Th2 skewing or inducing T cell apoptosis. It is not yet clear whether viral infection induces a regulatory phenotype in the B cells or whether regulatory B cells may be particularly susceptible to harboring latent infection. B cell activity is increased in the lungs of IPF patients. IgG and IgA levels in lavage fluids from IPF patients are significantly higher than healthy controls, indicating an increase in the amount of activated B cells present in the lung (34, 36). Mature, nonproliferating B cells, the primary reservoir of γHV-68 in the blood (9), have also been found in neolymphoid structures in the lungs of IPF patients (19). In addition, TGF-β1-mediated growth inhibition in human B cells is much less pronounced in the setting of EBV infection (3). The presence of B cells in the lung carrying a high viral load could provide a reservoir of consistent reinfection that accounts for the persistence of the virus, temporal heterogeneity, and enhanced TGF-β1 production all noted in the IPF lung.
The role of T cells in IPF is not fully defined. T cell depletion experiments followed by a fibrotic stimulus have produced contradictory results as reviewed in Ref. 17. A CD19-negative spleen cell population we isolated from latently infected mice 28 dpi expresses more total TGF-β1 and CCL2 than controls. This population expresses similar levels of gB and M3, indicating that γHV-68 is readily reactivated in this population. The CD19-negative spleen populations likely include T cells, monocytes, and dendritic cells. However, it is likely that the latently infected cells in this population and the cells that express TGF-β1 are in fact monocytes. Enriched T cells isolated from lung collagenase digests 28 dpi did not produce more total TGF-β1, and measurements of viral genome load in these highly purified T cells (98.5% CD3+) were undetectable or extremely low (Fig. 6). These findings are consistent with previous studies that have demonstrated viral genome in the non-T cell fraction but not in the T cell fraction of the thymus (12). Our ability to isolate T-enriched cells from lung digests was extremely limited, only allowing assessment of TGF-β1 and IL-10 levels; thus we do not have ELISA data regarding IL-4, IL-13, or IFN-γ production by T cells in the lung. Rather, we assessed production of IL-4 and IFN-γ in lung T cells by performing intracellular cytokine staining and analyzing the cells via flow cytometry. Figure 7 demonstrates that both CD4 and CD8 lung T cells showed a preference for IFN-γ production over IL-4 production in mock-infected animals. Not surprisingly, postinfection, the percentage of CD4 and CD8 cells producing IFN-γ was strikingly enhanced, but there was no evidence of induction of IL-4 expression. Taken together, the T cell phenotype in the lungs of latently infected mice is mixed. These cells produce IL-10 (Fig. 6D), a cytokine often associated with Th2 responses but not IL-4 (Fig. 7). Production of TGF-β1 by T cells is not increased. In fact, in one experiment, TGF-β1 production was significantly decreased, but this was not noted in all experiments (compiled data are shown in Fig. 6D). It is interesting to speculate that the enhanced production of IL-10 in lung T cells may inhibit the production of TGF-β1, as this has been previously reported (28, 47). In some circumstances, IL-10 has also been shown to be profibrotic (5, 15). Overall, we conclude that the ability of γHV-68 to augment fibrotic responses in wild-type mice can occur in the presence of a strong antiviral IFN-γ response.
An overarching question regarding the ability of viruses to augment fibrotic responses is whether these effects are specific to particular viruses or perhaps could be generalized. We were particularly interested to note the studies by Turpin et al. (45) using respiratory syncytial virus infection pre- or postexposure to vanadium pentoxide actually showed reduced levels of inflammation and fibrosis, and this was associated with decreased expression of profibrotic growth factors. Additionally, we (20) have previously noted that murine adenovirus infection does not exacerbate established fibrosis. Thus there may be some specificity for viruses that can establish latency or for herpesviruses in general in promoting fibrotic responses. Understanding the differences in how the host responds to the different viruses may offer additional insights for therapeutic intervention.
In summary, 28 days after intranasal infection with γHV-68, AMs, mesenchymal cells, and CD19-positive cells from the spleen and the lung all demonstrate latent viral infection. Collectively, these cell populations express increased levels of TGF-β1, CCL2, CCL12, TNF-α, and IFN-γ compared with the same cell populations from mock-infected mice. The upregulation of TGF-β1, TNF-α, CCL2, and CCL12 during viral infection has clear profibrotic implications in terms of the recruitment of inflammatory cells and the activation of matrix secretion. Our studies suggest novel roles that these circulating or resident cell types may have in herpesvirus-mediated fibrogenesis in the lungs. Additionally, although clinical studies have found evidence of γherpesviral gene expression in IPF tissues (40, 43), the fact that infectious virus is not always present has made it difficult to ascertain whether the infection is causally related to the fibrotic process. Our results suggest that presence of latent virus alters the secretory phenotype of both resident and recruited lung cells to create a profibrotic lung environment even in the situation where the latent viral load may be low.
This work was supported by National Institutes of Health Grants HL-087846 and AI-065543.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We thank Linda van Dyk for helpful discussions. We also thank Jason Weinberg for provision of the gB plasmid and protocol for DNA quantification. Additionally, Bryan Petersen and Nick Lukacs provided lungs from allergen-exposed mice for intracellular cytokine staining controls.
- Copyright © 2011 the American Physiological Society