S100A4/Mts-overexpressing mice have thick elastic laminae and mild pulmonary arterial hypertension (PAH), and the occasional older mouse develops occlusive neointimal lesions and perivascular inflammation. We hypothesized that a vasculotropic virus could induce neointimal lesions in the S100A4/Mts1 mouse by facilitating breakdown of elastin and migration and proliferation of smooth muscle cells. To test this hypothesis, we infected S100A4/Mts1 mice with gammaherpesvirus 68 (γHV68). We observed, 6 mo after γHV68 [4 × 103 plaque-forming units (PFU)], perivascular inflammation in 10/15 S100A4/Mts1 mice and occlusive neointimal formation in 3/10 mice, accompanied by striking degradation of elastin. We then compared the early response after high-dose γHV68 (4 × 106 PFU) in C57Bl/6 and S100A4/Mts1 mice. In S100A4/Mts1 mice only, significant PAH, muscularization of distal vessels, and elastase activity were observed 6 wk after γHV68. These features resolved by 3 mo without neointimal formation. We therefore infected mice with the M1-γHV68 strain that reactivates from latency with higher efficiency and observed neointimal lesions at 3 mo in 2/5 C57Bl/6 (5–9% of vessels) and in 5/5 S100A4/Mts1 mice (13–40% of vessels) accompanied by mild PAH, heightened lung elastase activity, and intravascular viral expression. This suggested that enhanced generation of elastin peptides in S100A4/Mts1 mice may promote increased viral entry in the vessel wall. Using S100A4/Mts1 PA organ culture, we showed, in response to elastase activity, heightened production of elastin peptides associated with invasion of inflammatory cells and intravascular viral antigen. We therefore propose that early viral access to the vessel wall may be a critical determinant of the extent of vascular pathology following reactivation.
- pulmonary hypertension
- γ herpes virus
- pulmonary vascular occlusive disease
pulmonary arterial hypertension (PAH) is characterized by progressive neointimal formation causing obstruction of pulmonary arteries. On the basis of experimental systems, our group has related the development of PAH to increased elastolytic activity reviewed in (49), release of growth factors (61), and activation of growth factor receptors (34). Elastin peptides, breakdown products of elastin, can upregulate the production of fibronectin (30), a glycoprotein critical to smooth muscle cell migration related to neointimal formation (41). Elastin peptides also possess chemotactic activity for monocytes and lymphocytes and can, in this way, facilitate perivascular or intravascular inflammatory processes (17, 19, 25, 31). Perivascular infiltrates composed of macrophages and lymphocytes have been described in lung biopsies of patients with PAH (62), but the role of inflammation and the immune system in the pathogenesis of PAH is poorly understood. PAH is a common complication of infectious and autoimmune diseases such as human immunodeficiency virus (HIV), mixed connective tissue disease, and scleroderma (67). Consistent with this, highly active antiretroviral viral therapy (HAART), including proteinase inhibitors, has been shown to improve the prognosis in HIV-associated PAH (72), whereas immunosuppression has proven beneficial in PAH associated with mixed connective tissue disease (55). Furthermore, a large portion of patients with idiopathic (I)-PAH show signs of autoimmunity and active inflammation, including the detection of circulating autoantibodies, cytokines, and chemokines such as IL-1, IL-6, macrophage inflammatory protein (MIP)-1α, fractalkine, and regulated on activation, normal T cell expressed, and presumably secreted (RANTES) (3, 14, 15, 32, 33). In addition, the involvement of a latent infection with the human herpesvirus (HHV)-8, also called Kaposi sarcoma-associated herpesvirus, has been shown in one series of I-PAH patients (8), although not in others (6, 9, 35, 39, 45, 51).
The goal of our study, therefore, was to determine whether we could use an animal model to establish a direct link between a viral infection and the development of PAH. We chose to carry out our experiments using the transgenic mouse that overexpresses S100A4/Mts1, a member of the S100 family of EF hand calcium-binding proteins implicated in inflammation and cancer (60). Although this mouse was originally designed to study propensity to cancer and metastatic disease (2), we made the observation quite serendipitously that the occasional older mouse (>1.5 yr old) can spontaneously develop obstructive neointimal formation in the pulmonary arteries (pulmonary vascular disease; PVD) in association with perivascular inflammation (23). We then went on to show that S100A4/Mts is increased in the neointimal lesions of patients with PAH (23) and that circulating levels of S100A4/Mts1 are also elevated in these patients (unpublished data). Clinical studies also show that S100A4/Mts1 can generate an antibody that suppresses production of interferon (IFN)-γ by γδ-T cells, potentially perpetuating a viral infection (36, 54).
In response to chronic hypoxia, S100A4/Mts1 mice have greater PAH than C57Bl/6 controls, in association with increased fibulin-5 and thickened elastic laminae, and S100A4/Mts1 mice also have elevated levels of stromal cell-derived factor-1 (SDF-1) (43). We therefore hypothesized that elastase activity produced by viruses such as those of the herpes family (48) could promote PVD in S100A4/Mts1 mice if this would render the elastic laminae susceptible to degradation.
In this study we used, the murine γ-herpesvirus (γHV68) (also referred to as MHV-68 or γHV), a vasculotropic virus, to develop a murine model of PVD. After an acute lytic infection, γHV68, a virus closely related to the human HHV-8 and Epstein-Barr virus, establishes life-long latency in B cells, macrophages, epithelial cells, and dendritic cells. It expresses a serine protease late in the lytic cycle (48) and is known to cause large-vessel arteritis when an underlying immune deficiency is present [i.e., in IFN-γ receptor (IFNγR)−/−, IFN-γ−/−, and B cell-deficient mice] (65). Immunohistochemical analyses demonstrate γHV68 antigen in arteritic lesions and reveal striking tropism of γHV68 for smooth muscle cells (10). Ongoing viral replication seems to be required for γMV68-induced vascular damage (11). In the following studies we describe the development of PVD in S100A4/Mts1 mice infected with γHV68 when over 1 yr of age and relate these observations to viral reactivation from latency and degradation of elastin.
MATERIALS AND METHODS
Infection of mice.
S100A4/Mts1 mice generated on a C57Bl/6 background and C57Bl/6 mice as controls were inoculated once intraperitoneally with γHV68 [4 × 102 to 4 × 106 plaque-forming units (PFU); a gift from Dr. M. Blackman, Trudeau Institute, New York, NY]. Hanks' balanced salt solution (HBSS; GIBCO/Invitrogen, Carlsbad, CA) was used as vehicle. The mutant M1-γHV68, characterized by a higher efficiency to reactivate (a gift from Dr. H. W. Virgin IV, Washington University, St. Louis, MO) was inoculated intraperitoneally at 4 × 106 PFU. Evaluation of a pathological response was carried out between 6 wk and 6 mo following viral inoculation. Animals were assessed for signs of PAH by measurement of right ventricular systolic pressure (RVSP), right ventricular hypertrophy (RVH), muscularization of distal arteries, and loss of vessels. The development of lesions similar to those associated with human pulmonary vascular disease (PVD) was assessed by the formation of a neointima in small intra-acinar arteries (50–300 μM), leading in some cases to complete occlusion of the vessel lumen, and was documented as a percentage of vessels with neointima relative to total vessels visible on a section of the whole lung. All animal studies were performed under a protocol approved by the Animal Care Committee at Stanford University under the guidelines of the American Physiological Society.
Assessment of RVSP.
43). Mice were killed by an overdose of isoflurane (5%), the lungs were flushed with 1× PBS, and the heart and lungs were removed en bloc. The right lung was immediately frozen in liquid nitrogen, and the left lung was inflated with 10% formalin (Fisher Scientific, Hampton, NH) and embedded in paraffin.
Assessment of RVH.
RVH was assessed by the ratio of right ventricular weight divided by the weight of the left ventricle plus septum (RV/LV+S) as previously described (20).
Transverse lung sections were stained to view elastic laminae with the Hart's stain (4). Muscularization was assessed as the percentage of fully and partially muscularized alveolar wall and alveolar duct arteries. The total number of peripheral arteries at alveolar duct and wall level was calculated as the number of arteries per 100 alveoli (43). Calculations were based on assessments of five fields per two lung sections per mouse, at magnification ×400 for assessments of muscularization and at ×200 for arterial number. All morphometric analyses were performed by one observer, blinded to the treatment or genotype.
This assay represents a slight modification from a method previously described (4, 43). Mouse lung tissue (∼30 mg), which was fresh frozen in liquid nitrogen and stored at −80°C, was homogenized in 1 ml 0.9% NaCl plus 2 mM N-methylamine (NMA) at 4°C using a Polytron homogenizer (Brinkmann Instruments, Westbury, NY) and centrifuged at 3,300 g for 30 min at 4°C. The pellet was kept as the source of elastase. To extract elastase, we added 1 ml of extraction buffer [0.5 M NaAc, pH 4, 0.02% (wt/vol) Na azide, and 10% glycerol] and 2 mM NMA. Samples were mixed, rotated overnight (O/N) at 4°C, and centrifuged at 8,100 g for 30 min at 4°C. Protein was concentrated with Centricon filters (cut off 3 kDa, YM-3; Millipore, Billerica, MA) at 6,500 g for 6 h at 4°C. The retentate was recovered by centrifugation at 300–1,000 g at 4°C for 2 min. The protein was precipitated by adding 1 ml 75% (NH4)2SO4 extraction buffer with 2 mM NMA and BSA to a final concentration of 70 μg/ml. The samples were mixed, shaken on a rotator for 1 h at room temperature, and then incubated at 4°C O/N without shaking to help protein to precipitate. The next day, samples were centrifuged at 12,000 g for 30 min at 4°C to pellet the protein. To reactivate the elastase, the pellet was resuspended in 150 μl of 50 mM Tris·HCl assay buffer (pH 8). Elastase activity was measured by a fluorescent assay, using the EnzChek elastase assay kit (E-12056; Invitrogen) and bovine DQ-Elastin as fluorogenic substrate. In addition, a serine elastase inhibitor (N-methoxysuccinyl-Ala-Ala-Pro-Val-chloromethyl ketone) was added to each sample in duplicate wells. After incubation for 1 h, elastase activity was calculated in fluorescent units by subtracting elastase activity that was not inhibited from the elastase activity observed. According to our standard curve, 500 fluorescent units equal 6.4 μU/ml porcine pancreatic elastase.
Hart's stain for elastin fibers.
According to a method previously described (4), slides were deparaffinized (twice for 10 min in xylene; Fisher Scientific, Hampton, NH) followed by graded alcohol: 100% (twice for 5 min), 95%, 70%, and 50% alcohol and distilled water (each 5 min). Slides were dipped in 0.25% potassium permanganate solution (5 min) and kept in running tap water until the color residue was gone. Slides were then dipped in 5% oxalic acid until they turned from yellow to clear and were washed again in running tap water (5 min), dipped in distilled water, and kept in resorcin-fuchsin working solution overnight. The next day, slides were rinsed in running tap water (10 min), dipped in distilled water, and counterstained in tartrazine solution (1–2 min). Slides were dehydrated in alcohol and xylene and then coverslipped (4).
Slides were deparaffinized in xylene (twice for 5 min) and hydrated in 100% (twice), 95%, 90%, and 70% ethanol (3 min each). Antigen retrieval was performed using the heat-mediated citrate buffer epitope retrieval method (10 mM sodium citrate, 0.05% Tween 20, pH 6.0). Staining was performed using the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA) according to the manufacturer's protocol. Primary antisera were used in the following concentration and incubation times: rabbit polyclonal CD-3 antibody (Ab; 1:100; Abcam, Cambridge, MA), 30 min at 37°C; the following Abs were all incubated overnight at 4°C: rat anti-mouse Mac-3 monoclonal Ab (1:100; BD Pharmingen, Bedford, MA); rat anti-mouse neutrophil monoclonal Ab, clone 7/4 (Ly 7/4) (1:200; Serotec, Raleigh, NC); affinity-purified anti-mouse macrophage aminopeptidase Ab (CD13) (1:100; Cedarlane Laboratories, Burlington, NC); polyclonal goat anti-mouse SDF-1 Ab (1:100; Santa Cruz Biotechnology, Santa Cruz, CA); polyclonal rabbit anti-mouse CXCR4 Ab (1:50; Abcam); monoclonal rat anti-CD-45 Ab (1:50; Novus Biologicals, Littleton, CO); rabbit anti-mouse collagen-1 Ab (1:100; AbD Serotec, Raleigh, NC); and rat anti-mouse IL-12 monoclonal Ab (1:100; Biolegend, San Diego, CA). Staining for viral antigen (γHV68) was performed using polyclonal rabbit anti-mouse γHV68 Ab (a gift from Dr. H. W. Virgin IV, Washington University, St. Louis, MO)at 1:2,500 and 1:5,000 dilutions O/N at 4°C. For each set of experiments, a secondary antibody-only control was run. 3,3-Diaminobenzidine (DAB) was used as a substrate for peroxidase (Peroxidase substrate kit, DAB; Vector Laboratories). Slides were rinsed, counterstained with hematoxylin for 20 s, rehydrated in graded ethanol and xylene, and coverslipped. All the histological and immunohistochemical assessments described below were carried out by an observer blinded to the experimental group being studied. The intensity of immunoreactivity was judged semiquantitatively as minimal (+), moderate (++), or intense (+++). The severity of inflammation was evaluated as mild, moderate, or severe, representing <10, 10–50, and >50 perivascular inflammatory cells, respectively. The presence of neointima formation was assessed and documented as a percentage of vessels with neointima relative to total vessels visible on a section of the whole lung.
Extraction and quantification of viral DNA.
Total DNA including endogenous and viral DNA was extracted from the lung using QIAamp DNA mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Viral load was quantified in the lungs of infected mice by real-time PCR using the ABI Prism 7700 sequence detector (PE Biosystems, Foster City, CA). We used primers and probes to detect a 70-bp region of the γHV68 glycoprotein B gene: forward, 5′-GGCCCAAATTCAATTTGCCT-3′; reverse, 5′-CCCTGGACAACTCCTCAAGC-3′; probe (5′) 6-(FAM)-ACAAGCTGACCACCAGCGTCAACAAC-(TAMRA) (3′), where FAM is a reporter dye and TAMRA is a quencher dye. Ten nanograms of DNA extracted from each mouse lung were used per reaction (66).
Organ culture of pulmonary arteries.
One day after intraperitoneal infection with 4 × 106 PFU γHV68, pulmonary arteries from C57Bl/6 and S100A4/Mts1 mice were isolated and placed in organ culture by using a slight modification of the method previously described by our group (42). While still attached to the heart, the main pulmonary artery (PA) and right and left branch PAs were dissected and cleared of fatty tissue. The main PA was then slit open, leaving the branches intact, and then the Y-shaped structure was removed en bloc and embedded in a collagen gel (Purcol; Inamed, Foster City, CA) prepared according to the manufacturer's protocol. The isolated PA subsequently settled slightly into the gel during collagen fibrillogenesis, which was initiated by warming at 37°C for 1 h. DMEM with 10% fetal bovine serum (GIBCO), 1% penicillin, streptomycin, and amphotericin B (Cascade Biologics, Portland, OR) and 1 μg/ml human leukocyte elastase (HLE) solution (1 μg/μl; Elastin Products, Owensville, MO) was then added to the tissue in the gel. The organ culture was incubated for 1 h at 37°C before addition of the homologous lymphocytes and monocytes.
Isolation of lymphocytes and monocytes.
Blood (500 μl) was aspirated with a heparin-coated syringe from the right ventricle of the mouse before death and removal of the PA. The blood was diluted with HBSS (1:1), 750 μl of Ficoll-Hypaque (Sigma-Aldrich, St. Louis, MO) was layered below the sample, and lymphocytes/monocytes were isolated using a density gradient centrifugation technique as described by the manufacturer. The cells were added to the elastase-treated PA organ culture and incubated at 37°C. On days 1 and 3, PAs were harvested, fixed in 10% formalin, and embedded in paraffin. Hart's stain was used to assess whether fragmentation of elastin was evident, and immunohistochemistry as described above was performed in addition to hematoxylin and eosin staining to assess the perivascular and intravascular accumulation of inflammatory cells and the presence of viral antigen.
Blockade of elastin peptides.
Two elastin-blocking agents were used. VGVAPG (Elastin Products), a hexapeptide ligand for the elastin-binding protein on lymphocytes/monocytes, was incubated at a concentration of 10−8 M with the lymphocytes/monocytes for 30 min before they were added to the organ culture (26). In other experiments, BA-4 (Sigma-Aldrich), a monoclonal anti-elastin antibody that binds to peptide fragments generated by proteolytic digestion of insoluble elastin, was added to the organ culture at a concentration of 1:1,000 for 1 h (69). Anti-mouse IgG (Southern Biotech, Birmingham, AL) was used as a control. After 3 days in organ culture, the extent of perivascular inflammation was judged as the number of perivascular inflammatory cells in the adventitia (mean of 3 separate fields at ×400 magnification). The presence of intravascular viral antigen was identified by immunoreactivity and assessed qualitatively.
The number of animals or samples in each group is given in legends. All quantitative results are means ± SE. Statistical significance was determined by one-way ANOVA followed by a Bonferroni post hoc test when comparisons involved three or more groups. A P value <0.05 was considered significant.
S100A4/Mts1 mice infected at 3 mo of age do not develop PAH or PVD.
Three-month-old S100A4/Mts1 and C57Bl/6 control mice (n = 5–6 per group) were infected either intraperitoneally with the vasculotropic virus γHV68 (4 × 102 and 4 × 103 PFU) to induce a latent infection in the lung or intranasally with the murine influenza virus x31 [4 × 102 egg infectious dose (EID)] to produce a more chronic pulmonary infection. After 2 and 6 mo, mice were killed and RVSP, RVH, and PVD were assessed. Neither virus produced an inflammatory response in the lung or elevation in RVSP, RVH, or neointimal lesions (PVD) (data not shown).
Since PVD occurred spontaneously only in S100A4/Mts1 mice >1.5 yr old, subsequent experiments were carried out in older animals. We therefore inoculated S100A4/Mts1 mice >1 yr old with either γHV68 (4 × 103 PFU) or influenza x31 (4 × 103 EID) and assessed the development of inflammation and/or PVD 6 mo later. We observed PVD in 20% (3/15) of mice infected with γHV68 but in none (0/12) infected with influenza x31 (Fig. 1). Infection with γHV68 was associated with a severe perivascular inflammatory infiltrate (>50 inflammatory cells) in the 3/15 mice with PVD (Fig. 1A) and a milder inflammatory response (<50 inflammatory cells) in an additional 7/15 mice without PVD. Compared with γHV68, influenza x31 led to a more severe peribronchial and parenchymal inflammation in 92% (11/12) of S100A4/Mts1 mice, yet the perivascular inflammation was similar without evidence of PVD (Fig. 1B). This suggests that a specific vasculotropic virus rather than the induction of perivascular inflammation may be necessary to induce PVD. None of the animals in the vehicle-treated group (n = 15) spontaneously developed PVD. Although RVSP was not measured in this group, we observed a significant difference (P < 0.05) in RVH in the γHV68 compared with the vehicle- or influenza-treated group (Fig. 1C), which was largely due to values of RV/LV+S in the 3/15 animals who developed severe PVD (0.46 ± 0.035) compared with values in the other S100A4/Mts1 mice without PVD (0.32 ± 0.02).
We then characterized the nature of the PVD lesions as well as the perivascular inflammation (Fig. 2). One of the most striking observations in the PVD lesion was the extensive fragmentation of elastin as shown by Hart's stain (Fig. 2A), and the development of a neointima composed largely of α-smooth muscle actin (α-SMA)-positive cells (Fig. 2B). Occasionally, cells staining positive for the macrophage marker Mac-3 (Fig. 2C) or the lymphocyte marker CD3 (Fig. 2D) were seen within the media or neointima, whereas these cells were abundant in the perivascular region. The absence of neutrophils (Ly 7/4; data not shown) around PVD lesions is in keeping with the chronic nature of the inflammation. We then assessed expression of the chemokine SDF-1, because it was induced selectively in S100A4/Mts1 mice exposed to hypoxia (43), and we also evaluated expression of its cognate receptor, CXCR4. We observed abundant intravascular expression of SDF-1 (Fig. 2E) codistributing with CXCR4 (Fig. 2F).
Viral antigen was found in the vessel wall, in the neointima in mild (Fig. 3, A and B) as well as in more occlusive lesions (Fig. 3C), and in surrounding inflammatory cells and epithelial cells, which are known to carry latent virus. In vessels with perivascular inflammation but without PVD, we did not observe inflammatory cells or viral antigen in the vessel wall.
We investigated whether fibrocytes were present in association with the PVD lesions, since they are associated with experimental chronic hypoxia-induced PAH (16, 59). We saw very few of these cells, as characterized by CD45, Col1, and Cd13 positivity (46), in association with the inflammatory response and none within the vessel wall (data not shown). We then determined whether activated dendritic cells, characterized by expression of IL-12 (50), were associated with PVD, since monocyte-derived dendritic cells are attracted to SDF-1-expressing cells (44). Furthermore, dendritic cells have been described as a latency reservoir for γHV68 (68), and the antimicrobial/elastase inhibitor elafin has been shown to regulate lung dendritic cells and adaptive immunity (52). Dendritic cells marked by IL-12 expression were observed in association with perivascular inflammation but were not selectively increased in S100A4/Mts1 mice with PVD (data not shown).
A high viral load transiently induces PAH but not PVD.
We investigated whether a higher initial viral load, potentially leading to a more severe inflammatory response, could initiate PVD lesions after shorter incubation periods of 6 wk and 3 mo. To this end, 1-yr-old C57Bl/6 and S100A4/Mts1 mice were inoculated intraperitoneally with a 1,000-fold higher concentration of γHV68 (4 × 106 PFU). Six weeks after infection, a transient increase in RVSP and RVH was detected in S100A4/Mts1 but not in C57Bl/6 mice (Fig. 4, A and B). Immunohistochemistry revealed the development of neither lung inflammation nor PVD, yet we observed an increase in muscularization of distal vessels, a morphometric indicator of pulmonary arterial changes associated with PAH, but no loss of arteries (Fig. 4C), a feature also associated with PAH (43). Lung elastase activity was more than twice as high in infected S100A4/Mts1 mice compared with C57Bl/6 mice (P < 0.01), but in uninfected mice the difference was not statistically significant (Fig. 4D). Three months after infection, the elevation in RVSP and RVH had regressed to normal in the S100A4/Mts1-infected mice (Fig. 4, E and F). Perivascular inflammation was detected in only 1/7 γHV68-infected S100A4/Mts1 mice and in 1/3 infected C57Bl/6 mice, but in none of the animals was PVD (neointimal lesions) or increased muscularization of distal vessels observed (Fig. 4G). Four of eleven S100A4/Mts1 mice showed a persistent twofold elevation in elastase activity compared with C57Bl/6 mice, but the difference between the groups was no longer statistically significant (Fig. 4H).
A mutant virus with higher reactivation efficiency produces PVD within 3 mo.
Since a greater viral load was not sufficient to induce PVD within a 3-mo time frame, we next investigated whether a mutant form of γHV68 with a higher efficiency to reactivate from latency could do so. One-year-old C57Bl/6 and S100A4/Mts1 mice were infected intraperitoneally with 4 × 106 PFU of mutant M1-γHV68 virus. This mutant strain contains a deletion at the 5′ end of the M1 open reading frame (ORF) (M1Δ511), leading to enhanced reactivation efficiency from latency (7). The mice were examined 3 mo after viral infection. Although the difference between vehicle- and virus-infected mice was not statistically significant, S100A4/Mts1 mice infected with M1-γHV68 tended to have the highest RVSP values (Fig. 5A), but significant RVH was not evident (Fig. 5B). However, elastase activity was elevated in the M1-γHV68-infected mice compared with both vehicle-treated S100A4/Mts1 animals (P < 0.01) and C57Bl/6 mice after M1-γHV68 infection (P < 0.05; Fig. 5C). When we compared viral load by quantitative RT-PCR in whole lungs from C57Bl/6 compared with S100A4/Mts1 M1-γHV68-infected mice, there were no significant differences (Fig. 5D).
Compared with vehicle-treated S100A4/Mts1 mice (Fig. 6A), all five infected with M1-γHV68 developed PVD (Fig. 6B) associated with fragmentation of elastin (Fig. 6, C and D). However, the degree of neointimal formation associated with α-SMA-positive cells was considerably milder and nonocclusive (Fig. 6, E and F) compared with the lesions observed in the 20% of S100A4/Mts1 mice followed 6 mo after inoculation with the wild-type virus. Quantifying the presence of neointima in all four groups by assessing the percentage of vessels per lung with neointima formation, we found neointima formation after M1-γHV68 infection in 5/5 S100A4/Mts1 mice with 25.2 ± 7.5% of vessels affected, compared with 2.8 ± 1.8% of vessels affected in 2/5 C57Bl/6 mice after infection (P < 0.05). It is interesting that we also observed severe PVD (neointima in 40% of vessels) in one S100A4/Mts1 mouse treated with a vehicle. γHV68 viral antigen could be detected in this animal, suggesting that it might have been infected secondarily by contamination from being housed in the same room and weighed once a week using the same device. Viral antigen was not detected in any of the other noninfected mice. We also found PVD in one C57Bl/6 vehicle-treated mouse, which at this point we speculate may be due to the age of our study mice population. Mild to severe perivascular inflammation was observed in some animals with PVD, yet mild neointima formation also seemed to occur without an extensive inflammatory reaction.
To further characterize differences in S100A4/Mts1 compared with C57Bl/6 mice that might be associated with susceptibility to PVD, we evaluated the composition of the perivascular and intravascular inflammatory infiltrate and examined the PAs for evidence of expression of viral antigen and the chemokine ligand/receptor pair SDF-1/CXCR4. The composition of the perivascular and intravascular inflammatory infiltrate, judged by macrophage and T cell staining, did not differ between C57Bl/6 and S100A4/Mts1 mice (data not shown). Viral antigen expression in the vessel wall was a consistent feature of PVD and hence was evident in all the S100A4/Mts1 and in the two infected C57Bl/6 mice with PVD. γHV68 infection increased expression of SDF-1 and CXCR4, but no difference in intensity of immunoreactivity was detected when C57Bl/6 and S100A4/Mts1 mice were compared. Thus it appears that with reactivation from latency, virus-mediated PVD in S100A4/Mts1 mice is associated with intravascular expression of viral antigen and reinduction of elastase activity.
S100A4/Mts1 PA elastin produces abundant elastin peptides associated with viral entry into the vessel wall.
Because we observed an initial elevation in elastase activity that accompanied viral infection at 6 wk in the S100A4/Mts1 mice that did not result in neointima formation at that time point, we hypothesized that an early increase in elastase activity or elastin susceptibility to degradation, or both, might have facilitated the later development of PVD in S100A4/Mts1 mice by facilitating seeding of virus in the vessel wall. We set out to determine whether in S100A4/Mts1 mice the abnormally thick elastic laminae may be more susceptible to degradation following viral inoculation and whether more functional elastin peptides, as judged by their chemoattractant properties, were produced that could attract inflammatory cells depositing virus in the vessel wall.
Explanted PAs from 1-yr-old S100A4/Mts1 and C57Bl/6 mice that were infected with γHV68 1 day before death to mobilize inflammatory cells were assessed for susceptibility to elastin degradation (Fig. 7). Western immunoblotting for tropoelastin on four isolated PAs from C57Bl/6 and S100A4/Mts1 mice treated with HLE for 1 h (10 μg/ml) showed a heavy smear of degraded elastin in S100A4/Mts1 mouse PAs compared with the few bands of elastin peptides noted in the C57Bl/6 PAs (Fig. 7A). The vessels were then placed on collagen in an organ culture system (Fig. 7B), and we assessed whether the elastin peptides produced were functional in mobilizing an inflammatory response by adding lymphocytes/monocytes from the same mouse to the organ culture. Compared with the C57Bl/6 PA (Fig. 7C), note the abundant elastin associated with elastic lamellae and in the interlamellar space in the S100A4/Mts1 PA (Fig. 7D).
After 3 days in organ culture, inflammatory cells were not seen surrounding any of the PAs from C57Bl/6 mice, and no viral antigen was present in the vessel wall (Fig. 8, A–D). In contrast, 4/4 PAs from S100A4/Mts1 mice showed an extensive accumulation of perivascular inflammatory cells in the adventitia, and in 2/4, there was invasion of the outer portion of the vessel wall (Fig. 8, E–H). These cells were largely macrophages (Mac-3) and T cells (CD3). Viral antigen for γHV68 was found in association with the periadventitial inflammation. The quantification of inflammatory cells in the adventitia in C57Bl/6 and S100/A4 PAs (Fig. 8I) showed 14 ± 1.5 cells in C57Bl/6 PAs compared with 72 ± 11 cells (n = 4, P < 0.01) in S100A4/Mts1 PAs per ×400 field.
To test whether the generation of elastin peptides was responsible for the accumulation of inflammatory cells in PAs of S100A4/Mts1 animals compared with C57Bl/6 PAs (Fig. 9, B vs. A), we preincubated lymphocytes/monocytes with the elastin peptide VGVAPG to inhibit binding to endogenous elastin peptides. In another experiment, the PAs were treated with the specific elastin antibody BA-4 to block inflammatory cell interaction with the endogenous elastin peptides or with mouse IgG as a control. Although accumulation of inflammatory cells was detected in 2/3 IgG-treated PAs (Fig. 9B), this was completely abrogated in 3/3 VGVAPG (Fig. 9C) and in 3/3 BA-4-treated organ cultures (Fig. 9D). The quantification of inflammatory cells in the adventitia (Fig. 9E) showed a significant decrease after treatment with VGVAPG (P < 0.05) and BA-4 (P < -0.01) compared with IgG-treated S100A4/Mts1 PAs. This indicates that the elastin peptides produced in the S100A4/Mts1 PAs were biologically active in inducing inflammatory cell chemotaxis.
On the basis of our findings, we propose the following model of γHV68-induced PVD in S100A4/Mts1 mice (Fig. 10). During the “acute” lytic phase, the virus induces mild elevation in pulmonary arterial pressure that may be related to vasoconstriction and some vascular remodeling associated with elevated elastase activity. These features reverse when the virus goes into a latent phase, in which only the viral genome is present in smooth muscle cells, but no inflammatory reaction is evident. The reduction in elastase activity correlates with the induction of latency of the virus. Reactivation of virus in the vessel wall from latency again induces elastase, but rather than the level of elastase activity per se, it is also the susceptibility of the elastin to degradation in the S100A4/Mts1 mice with this “second hit” that results in an elastin peptide-mediated smooth muscle cell chemotactic and proliferative response. In addition, degradation of elastin leads to activation of growth factors that are normally bound to the ECM (61) and that stimulate smooth muscle cell proliferation.
We describe for the first time how a vasculotropic virus mediates neointimal lesions seen in PVD. Intravascular viral expression and reactivation from latency coupled to host susceptibility, in this case related to overexpression of S100A4/Mts1, elastase activity, and elastin susceptibility to degradation, appear to be determinants of lesion formation.
Several epidemiological reports indicate a possible association of various forms of systemic vascular disease with the presence and titer of viral antibodies (1, 12, 22, 53). Some studies clearly show the presence of virus, viral antigens, or DNA in atherosclerotic lesions (29, 64), whereas other studies report the absence of viral antigen in these lesions (24). That it may be difficult to detect viral antigen or DNA in tissue does not preclude the possibility that a prior viral infection initiated an injurious process in the vessel wall. In transplant vasculopathy, the effect of the cytomegalovirus seems to be more an upregulation of an immune response to alloantigens and less of a direct intravascular effect (12). Although γHV68 could be detected by quantitative RT-PCR, there was no increase in the amount detected in whole lungs from S100A4/Mts1 compared with C57Bl/6 mice at 3 mo. Further studies are necessary to refine ways to detect whether there might be a difference in viral antigen within the PAs.
Vasculitis of the main PAs following γHV68 infection has been described in IFNγR−/−, IFN-γ−/−, and B cell-deficient mice (65). Mild pulmonary hypertension has been described in a murine model of retrovirus-induced immunodeficiency (21) with modest muscularization of distal vessels, but neointima formation was not documented.
It is intriguing that the S100A4/Mts1 mice inoculated with virus do not develop PVD unless they are over 1 yr of age. This could be a function of higher lung elastase activity in older S100A4/Mts1 (>1.5 yr) mice (13) or the presence of an impaired immune system that results in a smoldering viral infection, inducing elastase activity. In addition to age, overexpression of S100A4/Mts1 can result in suppression of IFN-γ by γδ-T cells, potentially perpetuating a viral infection (36, 54). The resistance to clearing of a viral load, causing increased seeding and infection of smooth muscle cells, might be critical in inducing PVD when the virus gets reactivated from latency. However, at least in the whole lung, there was no evidence of impaired viral clearance in S100A4/Mts1 mice. It is also possible that the propensity for virus-mediated subversion of endogenous microRNAs (5, 56) is greater in the older animal. Because viruses, particularly the herpesvirus family, can produce their own microRNAs (63), they may interfere with host cell function in a way that leads to impaired differentiation, resistance to apoptosis, and proliferation (47).
Inoculation of influenza did not result in PVD, despite a perivascular and intraparenchymal inflammatory response as great as or greater than that observed with γHV68. This is in keeping with our unpublished observations that bleomycin-induced inflammation in the S100A4/Mts1 mice did not induce PAH or PVD. Thus it is not the inflammation but perhaps the early invasion of the virus in the vessel wall that is required for PVD to occur following viral reactivation. The development of PVD was strongly associated with the expression of viral antigen in the vessel wall. This is in keeping with previous observations in IFNγR−/− mice (11) that persistent virus replication was a prerequisite for chronic arteritis. In those studies, antiviral therapy in mice with established disease resulted in increased survival, clearance of viral antigen from the media of the affected vessel, and substantial amelioration of arteritic lesions. Furthermore, it was documented that persistence of disease in the great elastic arteries was due to inefficient clearance of viral infection from the vessel (10). This was associated with failure of T cells and macrophages to enter the virus-infected elastic media, thereby demonstrating immunoprivilege of this site.
In both the S100A4/Mts1 and C57Bl/6 mice, γHV68 infection induced heightened expression of SDF-1 and its receptor, CXCR4. In Kaposi's sarcoma, the presence of a soluble SDF-1 gradient in the adventitia promoted transendothelial migration of virus-infected cells (70). This goes along our observation of CXCR4 expression mainly on endothelial and inflammatory cells and SDF-1 expression in the adventitia close to the outer elastic lamina. This ligand/receptor pair has been extensively studied in injured lung tissue, and a major focus of investigation has been its role in the recruitment of extrapulmonary progenitor cells and fibrocytes that leads to pulmonary fibrosis (38). Fibrocytes are abundant in the adventitia of hypoxia-induced pulmonary vascular lesions of calves with PAH and are thought to be key players in hypoxia-induced pulmonary vascular remodeling (16, 28, 59). Although SDF-1/CXCR4 expression was induced by the virus, the level of expression was similar in both genotypes, and fibrocytes were not evident in the lesions.
A higher initial inoculation of 4 × 106 PFU γHV68 transiently induced PAH in S100A4/Mts1 compared with C57Bl/6 mice, yet did not result in PVD 3 mo after infection. We interpret the increase in RVSP and RVH 6 wk after infection as a virus-mediated vasoconstriction; however, other processes such as endothelial edema could also contribute to the elevation in RVSP and RVH. Immune system-mediated endothelial damage and vasoconstriction induced by viruses have been previously proposed in systemic arteries (27), and big endothelin-1 has been shown to be elevated in human microvascular endothelial cells after HHV-8 infection (57). In addition, our group (43) has shown that vasoactive stimuli, such as acute hypoxia, elicit a greater response in S100A4/Mts1 compared with C57Bl/6 mice.
We can only speculate as to whether the early increase in elastase activity is caused by the initial viral replication. Because we did not observe an influx of inflammatory cells at the 6-wk postinoculation time point, the source of the heightened elastase activity may be the replicating virus (48) or the infected smooth muscle cells (71). It is interesting that the pulmonary hypertension (elevated RVSP), RVH, and muscularization of distal arteries observed in the S100A4/Mts1 mice regressed 3 mo after infection, during the time that the virus becomes latent. However, this initial insult may render the elastin of the S100A4/Mts1 mice more susceptible to degradation.
Interestingly, a higher initial viral load did not lead to development of PVD, whereas inoculation with the mutant M1-γHV68 with a fivefold higher efficiency to reactivate led to PVD and neointimal formation in all of the S100A4/Mts1 and in 20% of C57Bl/6 mice. This finding emphasizes that establishing a latent infection and reactivation from latency are critical aspects of herpesvirus infection mediating target tissue pathology. Processes that result in productive acute infection are fundamentally different from those giving rise to reactivation from latency and persistent infection, since they require different viral genes (58). Three months after M1-γHV68 infection, at a time coinciding with viral reactivation, we again observed heightened elastase activity in infected S100A4/Mts versus C57Bl/6 mice. At this time point, fragmentation of the elastic laminae and deposition of ectopic elastin were clearly evident. RVSP and RVH showed only a trend toward elevation in S100A4/Mts1 versus C57Bl/6 mice, which we attributed to a neointima that was nonocclusive. This is also of great interest because the model dissociates neointimal lesions from the other structural changes that have been related to pulmonary hypertension in rodent models and human disease, i.e., muscularization of distal vessels and loss of small vessels. It could be that the early stage of reactivation and nonocclusive lesions is associated with relatively low pulmonary vascular resistance that progressively increases as the lesions become more obstructive. The small proportion of the mice inoculated with the wild-type strain of γHV-68 studied 6 mo after inoculation, which had more occlusive neointimal lesions, had pulmonary hypertension as judged by RVH.
Our organ culture experiments of explanted PAs documented greater production of elastin peptides in S100A4/Mts1 PAs after exogenous application of elastase. We attribute this effect to the degradation of considerably thicker but less well-assembled elastin laminae in S100A4/Mts1 versus C57Bl/6 mice (43). The heightened production of biologically active, chemoattractant elastin peptides (18, 19, 31, 37) in S100A4/Mts versus C57Bl/6 mice was substantiated by the accumulation of inflammatory cells in the adventitia and the vessel wall. Our interventional studies with the elastin-blocking antibody BA-4 and the elastin receptor-binding peptide VGVAPG reinforce the involvement of elastin peptides in the chemoattractant process and in the seeding of virus that may produce its own elastolytic and proteolytic enzymes in association with PVD.
Our study refocuses attention on the way in which a viral infection can mediate neointimal lesions in association with pulmonary arterial hypertension in the susceptible host. Viruses can produce serine pretenses or induce endogenous elastase activity, but the propensity of the elastin to degrade may be a key determinant of the evolution of the pathology. How the pathology of neointimal formation evolves to produce occlusive lesions or plexogenic lesions seen in the clinical setting in some patients with HIV (40) or other forms of PAH not previously linked to viral infection will be of great interest to pursue in future studies now that we have a model of virus-mediated disease.
This work was supported by National Heart, Lung, and Blood Institute Grant 1-R01-HL-074186–01, the National Scleroderma Foundation, and the Scleroderma Research Foundation and by the Dwight and Vera Dunlevie Endowed Professorship (M. Rabinovitch), postdoctoral fellowships from the Pulmonary Hypertension Association (E. Spiekerkoetter), and the American Heart Association (Y.-M. Kim), and a studentship from Institut Jacques Monod (A. Bruneau).
We thank Dr. Marcy Blackman (Trudeau Institute, New York, NY) for the gift of γHV68, Dr. Herbert W. Virgin IV, (Washington University, St. Louis, MO) for the γHV68 antibody and M1-γHV68, and Dr. Edward S. Mocarski (Emory University School of Medicine, Atlanta, GA) for supplying the IFNγR−/− mouse, and we also appreciate the very helpful discussions with all three investigators during the course of our studies. We are indebted to Dr. Michal Bental Roof for help with the figures and with editorial review of our manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society