Murine Gammaherpesvirus 68 Infection of IFNγ Unresponsive Mice: A Small Animal Model for Gammaherpesvirus-Associated B-Cell Lymphoproliferative Disease

Gammaherpesviruses are tightly controlled by the host immune response, with gammaherpesvirus-associated malignancies prevalent in immune-suppressed individuals. Previously, infection of IFNγ-unresponsive mice with gammaherpesvirus 68 (γHV68) showed that IFNγ controlled chronic infection, limiting chronic diseases including arteritis and pulmonary fibrosis. Here, we show that γHV68-infected IFNγ receptor–deficient (IFNγR^−/−) mice uniformly develop angiocentric inflammatory lesions in the lung. Prolonged infection revealed a range of outcomes, from spontaneous regression to pulmonary lymphoma. By 12 months of infection, 80% of mice had lymphoid hyperplasia or pulmonary lymphoma; 45% of infected mice developed frank tumors between 5 and 12 months postinfection, with some mice showing systemic involvement. Lymphomas were composed of B lymphocytes and contained latently infected cells. Although IFNγR^−/− mice control chronic γHV68 infection poorly, both early and late pathologies were indistinguishable between wild-type and reactivation-defective virus infection, indicating that, in contrast with other previously described γHV68-associated pathologies, these chronic diseases were not dependent on the reactivation of latent infection. This distinct combination of latent infection and defined host defect led to a specific and consistent lymphoproliferative disease. Significantly, this mouse model of virus-associated pulmonary B-cell lymphoma closely mimics the full spectrum of human lymphomatoid granulomatosis, an EBV-associated malignancy with no effective treatment. [Cancer Res 2009;69(13):5481–9]

The gammaherpesviruses are a group of viruses associated with lifelong infection. Although healthy individuals could typically control this infection, individuals with immune deficiencies are at an increased risk for disease associated with chronic infection. Murine gammaherpesvirus 68 (γHV68) is closely related to the human gammaherpesviruses EBV and Kaposi's sarcoma-associated herpesvirus in sequence and in genomic organization (1). Additionally, γHV68 establishes latency in B cells, macrophages, dendritic cells, and lung epithelial cells, analogous to the human gammaherpesviruses (2, 3). γHV68 infection of normal, healthy mice results in acute virus replication that is cleared by 10 days postinfection, followed by a self-limiting mononucleosis response and a latent infection which is maintained for a lifetime (4), similar to the clinical course in EBV-infected humans (5). However, γHV68 infection of immunodeficient mice results in persistent infection, leading to diseases including arteritis (6), fibrosis (7), lymphoproliferation, and malignancy (8–10). Thus, γHV68 infection of mice is a valuable animal model for gammaherpesvirus pathogenesis.

Although type I IFNs are required for the control of acute γHV68 infection (11), type II IFN (IFNγ) is not (12). Instead, IFNγ plays a role in the control of γHV68 reactivation and chronic infection (6, 7, 13–15). Consequently, mice deficient in the receptor for IFNγ(IFNγR^−/−) are unable to control chronic γHV68 infection efficiently and succumb to large-vessel arteritis (6) and pulmonary fibrosis (7). These pathologies are attenuated by cidofovir treatment (which inhibits γHV68 replication), consistent with productive virus infection in these pathogenic outcomes (16, 17). We have shown that the viral cyclin of γHV68 plays an important role in reactivation from latency (18), and also in the development of chronic disease (18–20). Thus, infection with a viral cyclin-deficient virus (cycKO) also results in reduced severity and degree of observed arteritis (18) and fibrosis (17).

Gammaherpesvirus infection results in a number of lymphoproliferative disorders and malignancies, some of which are specifically of B-cell origin. Lymphomatoid granulomatosis (LyG) is an EBV-associated angiocentric and angiodestructive lymphoproliferative disease that can result in B-cell lymphomas (21–25). The disease tends to be aggressive in nature, resulting in high mortality (24), but with evidence of spontaneous resolution in some patients (26). The lung is the most common site due to its large vascular bed (23, 24), although multiple organ involvement has also been reported (27). Initially, the angiocentric lesions contain a polymorphous mixture of lymphocytes and histiocytes, with little to no detectable virus (23, 24). The inflammation can progress to lymphomas at late stages, and viral RNA can be found within large atypical B cells of the tumor (22–25, 28). Although the mechanism of LyG is unknown and LyG is not an AIDS-defining malignancy (29, 30), there is some increased incidence in patients with immune deficiencies (23, 31–33).

Another poorly understood EBV-associated lymphoproliferative disorder is pyothorax-associated lymphoma, which primarily develops in patients who receive an artificial pneumothorax as treatment for pulmonary tuberculosis (34). Like LyG, virus-positive B cells are present in the malignancy, which is generally localized to the lungs but has also been reported in the central nervous system (35, 36). Although angiocentricity and vessel destruction have been reported in pyothorax-associated lymphoma (37), this is not a common finding (34). Most cases of pyothorax-associated lymphoma have been documented in the Japanese population after therapeutic pneumothorax (38), and patients with this disorder usually do not have notable immune deficiency (37, 39).

Here, we provide a detailed analysis of pulmonary inflammation and chronic disease following γHV68 infection of C57BL/6 and IFNγR^−/− mice. At 90 days postinfection, we observed multiple angiocentric infiltrates in the lungs of IFNγR^−/− mice. Longer infection times revealed a spectrum of disease outcomes, from spontaneous regression to progression to advanced disease. B-cell lymphoma was observed in 45% of IFNγR^−/− mice between 5 and 12 months postinfection. Furthermore, inflammation and tumorigenesis occurred independently of infectious virus or reactivation of latent infection. The development of pulmonary lesions with potential to progress to lymphoma shares remarkable similarity with some EBV-associated human lymphoproliferative diseases. The reproducibility of specific disease on a well-defined genetic background makes γHV68 infection of IFNγR^−/− mice an excellent small animal model to study gammaherpesvirus-associated B-cell lymphoproliferative diseases.

Viruses and tissue culture. γHV68 clone WUMS [ATCC VR1465; wild-type (WT)] and viral cyclin-deficient γHV68 (cycKO) were passaged, grown, and titered as previously described (1, 18). NIH 3T12 cells and mouse embryonic fibroblasts were cultured as previously described (40).

Mice, infections, and organ harvests. IFNγR^−/− mice were bred in-house at the University of Colorado Denver (originally from Jackson Laboratory, B6.129S7-Ifngr1tm1Agt/J; ref. 41). Eight-week-old C57BL/6 mice purchased from Jackson Laboratory and 8- to 17-week-old age-matched and sex-matched IFNγR^−/− mice were infected intranasally with 4 × 10⁵ plaque-forming units of WT or cycKO γHV68 in 40 μL of balanced salt solution for ≥90 d. Mock-infected mice were inoculated with 40 μL of balanced salt solution for ≥90 d. C57BL/6 mice were included as strain controls for 90-d infections. Heart, lung, and spleen were harvested from mice infected for 90 d at the time of sacrifice. Only IFNγR^−/− mice were included in long-term infections (>90 d). Heart, lung, and spleen (and in some cases, brain and mediastinal lymph nodes) were harvested from long-term–infected mice at necropsy or 1 year postinfection. All animal studies were conducted in accordance with the University of Colorado Denver Institutional Animal Use and Care Committee.

Measurement of viral replication in lungs. Infectious virus was measured from mechanically disrupted lung (lower left lobe from each animal) by plating serial dilutions of lung homogenates on mouse embryonic fibroblasts for the determination of cytopathic effect, as previously described (20). This method detects very small quantities of virus (five plaque-forming units per organ; mouse embryonic fibroblast assay sensitivity is 10-fold greater than conventional plaque assay; ref. 40).

Histology and immunofluorescence. For histologic examination, lungs were fixed in 10% formalin, paraffin embedded, sectioned (4–6 μm) and stained with H&E for analysis using a Zeiss Axiocam HR camera and KS 300 Imaging System 3.0 software (Zeiss). For identification of B cells, tissues were deparaffinized, transferred through progressive ethanol gradients, and stained with purified rat anti-mouse CD45R/B220 clone RA3-6B2, followed by biotin-conjugated goat anti-rat immunoglobulin (BD PharMingen) and streptavidin R-PE (Invitrogen). Staining with rabbit anti-γHV68 serum (6) was performed on acetone-fixed frozen lung tissue and visualized with donkey anti-rabbit AlexaFluor 488 (Invitrogen). Slides were mounted with ProLong Gold antifade reagent with 4′-6-diamidino-2-phenylindole (Invitrogen). Immunofluorescence analysis was performed using an Olympus IX81 inverted motorized scope with spinning disc (Olympus), a Hamamatsu ORCA IIER monochromatic CCD camera (Hamamatsu), and Intelligent Imaging Slidebook v.4.067 (Intelligent Imaging Innovations). Histologic analysis of inflammation and lymphoma determination was performed in blinded fashion by board-certified pathologists (S.D. Groshong, C.D. Cool, and B.K. Kleinschmidt-DeMasters).

Generation of single-stranded biotinylated RNA probes. Genomic DNA was harvested from 3T12s infected with WT γHV68 at a multiplicity of infection of 0.5 plaque-forming units per cell for 5 d as previously described (1). Nucleotides corresponding to genome coordinates 10 to 5,699 of the γHV68 genome were PCR-amplified and cloned into pCR-Blunt II-TOPO (Invitrogen) to create pCR-Blunt II-TOPO-γHV68 left end (γHV68 LE).⁵

A 70 to 80 bp region of each of the three most abundant polymerase III (polIII) transcripts (polIII-1, polIII-4, and polIII-5) expressed during latency and the viral M2 gene were PCR-amplified from γHV68 LE (Supplementary Fig. S1A). The amplimers were each cloned into pCR-Blunt II-TOPO (Invitrogen) and sequenced. pCR-Blunt II-TOPO containing each of the inserts and pGEM-3z (Promega) were digested with EcoRI and BamHI, ligated and transformed into One Shot TOP10 chemically competent E. coli (Invitrogen). The resulting plasmids were linearized with BamHI for in vitro transcription with T7 polymerase (Epicentre Biotechnologies) in the presence of biotin-labeled d-UTP (Roche). The transcription products were verified by Northern analysis using CDP-STAR (Roche; Supplementary Fig. S1B).

In situ hybridization. Virus-infected cells were detected using the following protocol (detailed schematic in Supplementary Fig. S1C). After tissue sections were deparaffinized and transferred through progressive ethanol gradients, they were treated with protease I (Ventana Medical Systems) for 20 min. A mixture of biotinylated probes directed against polIII-1, polIII-4, polIII-5, and M2 (500 pg/mL each) was added to the slides and heated at 65°C for 10 min, followed by a 2-h incubation at 42°C. The slides were then incubated in stringent wash solution (in situ hybridization kit for biotinylated probes; Dako) for 30 min at 48°C. Streptavidin-alkaline phosphatase (Dako) was incubated on slides for 30 min, followed by nitroblue tetrazolium chloride/5-bromo-4-chloro-3′-indolyphosphate p-toluidine salt (Dako) for 45 min to 1 h. Slides were counterstained with nuclear fast red (Vector Labs) and mounted with VectaMount AQ aqueous mounting medium (Vector Labs). Cells positive for viral transcripts appear blue-purple and localize to the nucleus, perinucleus, or cytoplasm, whereas nuclear fast red stains the nucleus a pink to light red shade. Positive and negative controls were included in each experiment (Supplementary Fig. S1D).

Statistical methods. Severity of pulmonary inflammation in the 90-d–infected tissues were independently and blindly assessed by S.D. Groshong. For all slides, at least 10 fields per section were analyzed. Pathologic processes (necrosis, fibrosis, and lymphoid hyperplasia) were assessed by C.D. Cool. Statistical significance was determined using the nonparametric Mann-Whitney test (GraphPad Software). The percentage of B-cell lymphoma–free mice (survival) was determined using the Kaplan-Meier survival curves and log rank test (GraphPad Software).

γHV68-infected IFNγR-deficient mice exhibit significant inflammation around pulmonary vessels. Previous studies have anecdotally noted pulmonary inflammation in γHV68-infected IFNγR^−/− mice (7, 10). To quantitate the incidence, severity, and outcome of pulmonary inflammation in chronic virus infection, healthy C57BL/6 and immunodeficient IFNγR^−/− mice were infected with γHV68 for 90 days. At this time, acute infection had been cleared, and only chronic, latent infection remained. H&E staining of lung tissue from infected IFNγR^−/− mice showed inflammation characterized by multiple angiocentric infiltrates surrounding both arteries and veins (Fig. 1A). In some instances, hemosiderin was present in the infiltrates, indicating previous hemorrhage (Fig. 1B). In contrast, mock infection of IFNγR^−/− mice, as well as γHV68 infection of C57BL/6 mice (data not shown), resulted in normal lung architecture characterized by “lacey” air spaces (Fig. 1C). The pathology resulting from infection of IFNγR^−/− mice (Fig. 1A) bore strong similarity to that seen during early-stage LyG (Fig. 1D), and prompted prolonged analysis of these infected mice, as described later.

Next, we conducted a thorough analysis of the observed pulmonary inflammation in 90-day–infected mice. We established a quantitative scoring system based on H&E staining to objectively measure the severity of inflammation present in infected mouse lungs (Supplementary Fig. S2). Each vessel was given a score (0–3) based on the thickness of inflammation present around the vessel. The sum of the scores was then divided by the total number of vessels to yield the mean vascular score. IFNγR^−/− mice infected with either WT or cycKO γHV68 uniformly showed angiocentric inflammation, and exhibited significantly greater mean vascular scores than infected C57BL/6 or mock-infected IFNγR^−/− mice (Fig. 1E).

Vascular infiltrates in 90-day–infected IFNγR^−/− lungs are primarily composed of B cells. Because γHV68 and the human gammaherpesviruses use B cells as the major latent reservoir, we performed immunofluorescent staining for the mouse B cell marker, B220, to determine whether the inflammatory lesions contained B cells. We found that B220⁺ B cells in mock-infected IFNγR^−/− mice and in mock and infected C57BL/6 mice were sparse and randomly dispersed throughout the lungs (Supplementary Fig. S3; data not shown). This is in distinct contrast to the concentrated B220⁺ cells in infected IFNγR^−/− lungs (Fig. 2A), which were far more numerous and were localized to inflammatory lesions around the vessels.

Infectious virus can be detected only in the lungs of wild-type–infected IFNγR^−/− mice. IFNγ has been shown to be important in controlling persistent γHV68 infection, as mice that have a defect in the IFNγ signaling pathway develop and succumb to disease (6, 7, 13–15). The use of cidofovir in these studies inhibited the development of disease (16, 17). Similarly, infection with the cycKO virus, which has a defect in reactivation, reduced the degree of observed fibrosis (17), implying that ongoing viral replication was necessary for pathogenesis. The demonstration of significant inflammation at 90 days postinfection (Fig. 1) suggests that ongoing viral replication could contribute to the pathology. Using an extremely sensitive assay, we analyzed infectious virus in the lungs of 90-day–infected mice, and detected low levels only in the lungs of WT-infected IFNγR^−/− mice (Fig. 2B). This amount of virus is not detectable by conventional plaque assay (40), nor by immunofluorescent analysis of lytic protein using anti-γHV68 serum (data not shown). This data is the first direct demonstration of infectious virus in the lungs of WT-infected IFNγR^−/− mice at late times after infection. No virus was detected in the lungs of infected C57BL/6 mice, nor was virus detected in cycKO-infected IFNγR^−/− mice (Fig. 2B), despite significant inflammation following cycKO infection of IFNγR^−/− mice (Fig. 1E).

The lungs of 90-day–infected IFNγR^−/− mice harbor latently infected cells. Although ongoing inflammation did not correlate with the detection of infectious virus, it remained possible that inflammation correlated with latent virus infection in the lungs. Latent gammaherpesvirus infection is typified by very limited gene expression and is virtually undetectable apart from the detection of viral nucleic acid (9, 42–45). Although in situ detection of EBV is commonly performed in clinical laboratories, the methods described thus far for in situ detection of γHV68 have been technically difficult and/or required days to complete (3, 9, 10).

We designed an optimal, enzyme-based in situ hybridization (ISH) protocol for the detection of latent γHV68, with high specificity that can be completed in 1 day (Supplementary Fig. S1). The γHV68 ISH protocol described here was developed by modification of a protocol for detection of EBV RNA as described by D. Davis (University of Colorado Denver Department of Pathology).⁶

Using this protocol, we were able to detect cells containing latent viral transcripts in lung tissue from 90-day–infected IFNγR^−/− mice (Fig. 2C; Supplementary Fig. S4). Virus-positive cells were specifically localized to areas of inflammation around the vessels. In healthy, normal mice infected with γHV68, the peak number of latently infected cells is found in the spleen at 16 days postinfection (46); therefore, this tissue served as a positive control in our ISH protocol (Supplementary Fig. S4). No virus-infected cells were detected in mock-infected IFNγR^−/− mice, or γHV68-infected C57BL/6 mice (Supplementary Fig. S4), which also do not show any evidence of ongoing inflammation.

We also showed viral gene expression of the polIII-1 transcript in the lungs of 90-day–infected IFNγR^−/− mice by reverse transcriptase-PCR but did not detect expression of Rta, despite using a sensitive nested PCR analysis (Supplementary Fig. S4). PolIII-1 is constitutively expressed in both latent and lytic infection (42), whereas Rta is the immediate early activator expressed in lytic replication and reactivation (47). This finding, along with the ISH data, shows similar latent gene expression in IFNγR^−/− mice infected with either WT or cycKO virus.

The severity of inflammation is variable in long-term–infected IFNγR^−/− mice. Because the inflammatory infiltrates observed in 90-day–infected IFNγR^−/− mice closely resembled those seen during the early stages of LyG (Fig. 1A and D), IFNγR^−/− mice were infected for up to 1 year to determine the consequences of prolonged inflammation (Table 1). We found a modest decrease in the average level of inflammation in infected mice (Figs. 1E and 3A), which was evidenced by H&E staining (Fig. 3B–D). However, the mean vascular scores were widely variable, in that a few of the infected mice exhibited more severe inflammation (Fig. 3C) and some maintained the same level of inflammation. In some mice, the inflammation resolved dramatically such that the lungs resembled those of mock-infected mice (Fig. 3D). These data show variable resolution of the inflammation noted at 90 days, both after infection with WT and cycKO γHV68.

Some of these mice also showed signs of other disease. For example, two of the WT-infected IFNγR^−/− mice exhibited fibrosis. These mice, along with seven other mice, also showed lymphoid hyperplasia or mitoses, indicative of lymphoproliferation (Table 1).

Long-term infection of IFNγR^−/− mice leads to the development of pulmonary lymphoma with evidence of systemic involvement. Interestingly, in the remainder of the long-term–infected mice, we observed that the pulmonary inflammation seen at 90 days postinfection had progressed to frank lymphoma (Table 1). The incidence of B-cell lymphoma in mice infected for >1 year was 45% (9 of 20; Fig. 4A), with nine additional mice showing signs of lymphoproliferation characterized by lymphoid hyperplasia or mitoses (Table 1). Three mock-infected mice developed T-cell lymphoma over this time (Supplementary Fig. S5; Table 1), consistent with previous reports that IFNγR^−/− mice develop tumors more rapidly and with greater frequency than normal mice (8). H&E staining of lymphomas from IFNγR^−/− mice showed a number of features commonly observed in cancers (48). First, the tumor cells in the lung overran anatomic boundaries and crossed the blood vessel walls (Fig. 4B and C). Second, higher magnification revealed various stages of mitoses (arrows) among numerous large, atypical cells (Fig. 4C). Third, areas of necrosis, due to hypoxia, invasion, or disruption of surrounding tissue were detected (Fig. 4C,, inset). Finally, there was evidence of systemic involvement in a number of mice (Table 1), in which lymphoma was also present in the heart, brain, and/or lymph nodes (Fig. 4D; data not shown).

Lymphomas in IFNγR^−/− mice are of B-cell origin and contain virus-infected cells. Most of the lymphomatous cells from γHV68-infected IFNγR^−/− mice expressed the B cell markers B220 and CD79α (Fig. 5A and B; Table 1; Supplementary Fig. S6). This is in contrast with the lymphomatous cells found in mock-treated IFNγR^−/− mice, which were negative for B-cell markers but positive for the T-cell marker CD3 (Supplementary Fig. S5; Table 1). The lymphoma in the mock mice exhibited a starry sky pattern accompanied by massive splenomegaly, neither of which was observed in γHV68-infected mice. Furthermore, virus-infected cells were detected by ISH for polIII-1, polIII-4, polIII-5, and M2 transcripts in all of the pulmonary lymphomas analyzed from both WT-infected and cycKO-infected IFNγR^−/− mice (Fig. 5C; Table 1), and virus-infected cells were more numerous than those detected in 90-day inflammatory lesions (Fig. 2C). Virus-positive cells were also detected in brain tissue from a cycKO-infected mouse that showed signs of systemic involvement (Fig. 5D). To corroborate our ISH findings of virus being present in the tumor samples, DNA isolation was attempted from several tumors for PCR detection of viral Rta (see Supplementary Materials and Methods). Although formalin-fixed, paraffin-embedded tissue provides a suboptimal source of DNA, we confirmed the presence of viral Rta DNA by nested PCR analysis of tumors from both WT-infected and cycKO-infected tissues, including the brain tissue represented in Fig. 5D (data not shown).

Prior studies of γHV68-infected IFNγR^−/− mice documented the development of arteritis and fibrosis (6, 7). In this report, we fully characterized pulmonary inflammation and lymphoproliferation of chronic γHV68 infection in IFNγR^−/− mice. In doing so, we noted a striking disease outcome that was not previously recognized. We identified multiple angiocentric infiltrates in the lungs of infected IFNγR^−/− mice at 90 days postinfection, but not in those of control mice (Fig. 1). Pulmonary inflammation is a common result of virus infection, but is usually coincident with acute infection (4), and is not apparent at late times after infection. Consequently, we developed an objective scoring system (Supplementary Fig. S2), which allows quantitative comparison of pulmonary inflammation following virus infection. Our application of this scoring system established that significant inflammation was present only in infected IFNγR^−/− mice, and not in any of the control mice (Fig. 1E). We also showed that virus-infected lungs contained angiocentric lesions abundant in B cells, whereas lungs of control mice displayed only scarce and randomly dispersed B cells (Fig. 2A).

Previous studies in IFNγR^−/− mice also showed that treatment with cidofovir inhibited the development of disease (16, 17). The degree of fibrosis could be reduced by infecting mice with cycKO virus (17), implying that ongoing viral replication was necessary for the observed pathogenic outcomes. Using a very sensitive assay, we directly detected infectious virus, but only in the lungs of WT-infected IFNγR^−/− mice (Fig. 2B). We detected cells positive for latent viral transcripts in lung tissue from IFNγR^−/− mice infected with either WT or cycKO virus (Fig. 2C) using ISH and reverse transcriptase-PCR, but not in control mice (Supplementary Fig. S4).

Due to the similarity of the histologic findings in 90-day–infected IFNγR^−/− mice to that characteristically seen in early stage human LyG, we analyzed longer infection times in IFNγR^−/− mice to address the potential outcomes of virus-associated disease. We noted variability in the level of pulmonary inflammation observed at later times, with some mice exhibiting more severe inflammation and others showing spontaneous resolution (Fig. 3). Chronic infection of IFNγR^−/− mice resulted in B-cell lymphoma development (Table 1), with a surprisingly robust incidence of 45% in mice infected for up to 1 year (Fig. 4A). Each lymphoma-positive mouse displayed characteristic features of cancer (Fig. 4B–D) and contained abundant virus-infected B cells (Fig. 5). Seven of the nine lymphomas presented in the lung, and it is likely that the remaining two mice had pulmonary lymphomas which were not captured in the sections of lung tissue analyzed. At this point, it is unclear whether the involvement of other vascular organs is a result of initial seeding at multiple sites, or a metastatic process.

Three mock-infected mice also developed lymphoma, not surprising given that IFNγR^−/− mice are known to develop tumors more rapidly than normal mice (8). However, lymphoma present in the mock-infected mice was not of B-cell origin (Supplementary Fig. S5), was not virus-associated, and was not preceded by any evidence of earlier inflammation (Fig. 1E), suggesting a different mechanism involved in its development. Among the long-term infection mice that did not develop lymphoma, most exhibited lymphoid hyperplasia or mitoses (Table 1), consistent with early lymphomagenesis. The mean vascular score at late times did not seem to correlate with other disease signs. The variability in the outcome of long-term infection closely resembles the course of disease in patients with LyG, with some showing complete resolution without treatment, whereas others progress to lymphoma (26). We also showed that disease developed in mice infected with cycKO γHV68. This is notable because, prior to this study, various chronic diseases associated with γHV68 infection of immunodeficient mice have shown dependence on the viral cyclin (17, 18, 20). This is the first report of chronic disease comparable to that of WT infection in the absence of the viral cyclin, suggesting that the pathology we describe here does not depend on virus reactivation but rather is associated with latent infection.

It was previously shown that γHV68 infection of IFNγR^−/− mice resulted in advanced arteritis (6). We did not note this disease in our analyses for a number of possible reasons: (a) the incidence of arteritis is greatest in mice infected at 6 weeks or younger, (b) arteritis was shown after intraperitoneal inoculation, and (c) arteritis was described in IFNγR^−/− mice with a 129Ev/Sv background. Furthermore, the angiocentric lesions we observed at 90 days postinfection were noted around both arteries and veins. This is in contrast to the described arteritis, which specifically involves the infection of smooth muscle. We observed two cases of fibrosis in WT-infected IFNγR^−/− mice, similar to previous reports (7, 17). These mice exhibited low mean vascular scores (Table 1), suggesting that there was little inflammation coincident with fibrosis. It is possible that the greater level of reactivation in these mice preferentially leads to tissue damage rather than inflammation. The key features that distinguish this model from previous studies of γHV68-associated malignancy (9, 10) include the use of C57BL/6 mice (a non–tumor-prone strain) with a defined immune defect, which upon infection, develops B-cell lymphoma in the lung; that is, a single reproducible disease outcome with high incidence over a relatively short infection period.

Our findings to date indicate that the infection model described here bears striking and reproducible similarity at early and late stages to LyG, a rare human disease with no animal model. We have identified two factors, chronic gammaherpesvirus infection combined with a defined primary immunodeficiency, which are strongly associated with the development of pulmonary inflammation and lymphomagenesis. Although the mechanism of lymphomagenesis in LyG is not known, immunosuppressed patients are more prone to developing LyG (23, 31–33). It has also been shown that an IFNγ gene polymorphism correlated with a greater risk of lymphoproliferative disease in an EBV animal model (49), and that the gene expression and circulating levels of the IFNγ-associated chemokines Mig and IP-10 are elevated in patients with LyG (50).

This study raises the intriguing possibility that unrecognized immune defects, such as an impairment of the IFNγ signaling pathway, in combination with common infections such as EBV, may lead to chronic disease. Further application of this model will facilitate our understanding of the mechanism of gammaherpesvirus-associated B-cell lymphoproliferative diseases and may provide a platform for therapeutic development and testing.

No potential conflicts of interest were disclosed.

Grant support: NIH T32 AI07537-09 and T32 AI007405-18 (K.S. Lee), and NIH R01 CA103632 and Burroughs Wellcome Foundation Career Award (L. van Dyk).

We thank Dr. H.W. Virgin (Washington University in St. Louis) for anti-γHV68 serum, Dr. Laurel Lenz (National Jewish Health) for IFNγR^−/− mice (originally from Jackson Laboratory), Dr. John Ryder and David Davis (University of Colorado Denver) for expert ISH assistance, and Dr. Han Myint (University of Colorado Denver) for expert review of the data. We also thank Kevin Diebel (University of Colorado Denver) for expert advice in RNA probe design and Dr. Eric Clambey (National Jewish Health) for critical insights.