Viral targeting of fibroblastic reticular cells contributes to immunosuppression and persistence during chronic infection

LCMV CL-13 Infects FRC.

LCMV Armstrong and CL-13 are responsible for very different outcomes of infection in adult mice: acute and chronic infections, respectively. Clearance of Armstrong occurs within 1 week of infection, coinciding with a large specific T cell response. Conversely, CL-13 persists at high titers in blood and tissues for several months, resulting in functional exhaustion of virus-specific T cells and a generalized immune suppression. Although CL-13 can infect DC (10), other additional mechanisms may also play a role in the functional impairment of the responding T cells (14). Assessment of viral antigen distribution in the spleen after CL-13 infection revealed a distinctly different localization in comparison with LCMV Armstrong infection. Armstrong largely localized to the marginal zones (Fig. 1A), with minor staining within white pulp (WP) and red pulp (RP). In contrast, CL-13 was found in the marginal zones and extensively throughout the WP and also in the RP. Moreover, a distinctly reticular pattern of LCMV antigen was apparent throughout the WP after CL-13 infection, suggesting potential infection of the lymphoid stroma.

To visualize cells of the reticular network, we used the monoclonal antibody ER-TR7 (15), which stains the network of FRC in the WP and RP [supporting information (SI) Fig. 6]. Detailed examination of WP FRC was also performed by staining for laminin, smooth muscle actin, and desmin, as well as the collagen III⁺ reticular fibers ensheathed by the FRC (16) (SI Fig. 6).

To determine whether LCMV CL-13 was infecting stromal cells we performed a kinetic analysis of viral infection in the spleen. Spleens were harvested between 1 and 30 days after infection and costained for LCMV and ER-TR7 antigens. Initially (day 1), both in mice infected with LCMV Armstrong and in mice infected with CL-13, virus predominantly localized to the marginal zone (Fig. 1B). At this point, the degree of cellular infection appeared to be very similar between the two strains. Within 3 days of infection, however, CL-13 could be observed throughout the WP and more extensively in the RP. Infection of both the WP and RP by CL-13 further increased, peaking at day 8 and remaining high for at least 30 days after infection. In stark contrast, Armstrong infection remained mostly localized to the marginal zone and was controlled by day 8. Importantly, LCMV antigen appeared to colocalize with ER-TR7⁺ FRC after CL-13, but not Armstrong, infection. Indeed, a substantial proportion of the infected cells after CL-13 infection appeared to correspond to cells of the reticular network.

A closer look at infection of splenocytes after Armstrong infection revealed large cells of distinct morphology staining positive for LCMV (Fig. 2A). Macrophages were infected by both Armstrong and CL-13, yet neither strain was observed to infect T or B lymphocytes (data not shown). Whereas very few FRC were positive for viral antigen in Armstrong-infected spleens, many were heavily infected within just 3 days of CL-13 infection. Furthermore, colocalization of LCMV antigen with collagen III, laminin, smooth muscle actin, and desmin was used to clearly determine infection of the FRC by LCMV CL-13 (Fig. 2B). Laminin⁺ ER-TR7⁻ cells in the B cell zones were also found to colocalize with LCMV, suggesting that these stromal cells may also be a target of LCMV CL-13 infection (data not shown).

Given the systemic nature of LCMV infection, we also examined peripheral lymph nodes (LN) for the presence of LCMV antigen-positive FRC. Similar to the pattern of infection in the spleen, LCMV antigens were observed in LN FRC after CL-13 infection (Figs. 1B and 2A). Minimal infection of FRC was observed in the LN during Armstrong infection.

Infection of the FRC by CL-13 was determined by transmission electron microscopy. Staining for LCMV antigen revealed highly infected reticular cells and cell processes ensheathing bundles of collagen fibers (Fig. 2C). Lymphocytes observed in close association with the FRC were not infected. Infected reticular cells did not express MHC class II or Fc receptors (SI Fig. 7). Lastly, infection of FRC was quantified by flow cytometry after dissociation of the spleen by collagenase treatment (Fig. 2D). Almost two-thirds of ER-TR7⁺ cells isolated from the spleen stained positive for LCMV antigen after CL-13 infection. This was in contrast to only 6.6 ± 2.8% of CD11c⁺ DC that were found to be infected with LCMV at this time. Together, these data demonstrate that LCMV CL-13 targets reticular stromal cells in the lymphoid tissues.

We examined the genetic basis of CL-13 FRC tropism using reassortants of the Armstrong and CL-13 viruses. The LCMV genome consists of two single-stranded RNA segments designated large (L) and small (S), which encode the viral polymerase protein and the glycoprotein, respectively. LCMV CL-13 differs by only two amino acids from the parent strain, LCMV Armstrong (17). These have been mapped to a single amino acid change in the viral polymerase that may increase viral yield within infected cells and a single change in the glycoprotein that is important in determining the degree of infectivity (18, 19). We infected mice with viruses containing the L segment from CL-13 and the S segment from Armstrong (13/Arm) or vice versa (Arm/13). Mice infected with either reassortant virus displayed large numbers of infected cells and infected FRC (SI Fig. 8). However, neither reassortant demonstrated a pattern of infection equivalent to that of CL-13, which appeared to infect greater numbers of cells in the spleen. Thus, mutations in both the polymerase and glycoprotein genes of LCMV CL-13 appeared to contribute to the tropism exhibited by the virus.

Disruption of the Conduit Network During CL-13 Infection.

The stromal network is thought to play numerous roles in lymphoid organs, influencing cell recruitment, migration, activation, and survival through organization of the lymphoid cellular and molecular microarchitecture. The FRC form a contiguous network of conduits that allow transport of small molecules and antigens directly into the T zones without perfusing through the parenchyma (20). CL-13 infection is characterized by disruption of the lymphoid architecture, including smaller, disorganized WP regions and reduced cellularity within the WP. Infected FRC may die, lose cellular integrity, or experience disruption of cell–cell contacts that keep the conduit lumen separate from the splenic parenchyma. To examine the integrity of the conduit network after LCMV infection, mice were injected intravenously with fluorescently labeled tracer molecules (Fig. 3A). Small molecules accumulate within the conduit network in the spleen after injection (21), making it possible to trace the functional framework of the conduit network. In uninfected (control) mice, Alexa Fluor 568-conjugated dextran (Fig. 3B) and OVA-FITC (data not shown) were clearly observed in the conduits of the spleen 10 min after injection. In mice infected with LCMV Armstrong, greater amounts of tracer accumulated within the marginal sinuses (data not shown); however, the tracers were found to accumulate within the conduits, similar to that in uninfected mice. In contrast, during CL-13 infection there was marked disruption in the ability of the tracers to accumulate within the conduit network in the WP of the spleen. We observed an ≈86% and ≈74.5% reduction in Dex-568 or OVA-FITC tracer accumulation, respectively, in WP areas after CL-13 infection when compared with uninfected mice (Fig. 3E and data not shown). Tracer could be found in association with some cells in the RP, possibly because of leakage into the splenic parenchyma and association with resident cells. These data suggest that LCMV CL-13 infection is detrimental to the function of the FRC conduit network in the spleen.

CD8⁺ T Cell Involvement in Disruption of the Conduit Network.

The altered drainage of antigen in the conduit networks in mice infected with LCMV CL-13 may have resulted from damage due to direct infection of the FRC or immune-mediated damage. Although LCMV is a noncytolytic virus, infection of cells is known to have detrimental effects on the health, or the “luxury functions,” of the cells it infects (22). Such an alteration in function may account for the disrupted FRC function. We examined carrier mice infected with LCMV CL-13 in utero, which are immunologically tolerant to the virus because of deletion of virus-specific T cells and display no generalized virus-induced immunosuppression. Carrier mice were found to have entirely normal splenic conduit drainage despite large numbers of infected FRC (Fig. 3 C and E). This suggested that aspects of the immune response were responsible for disruption of tracer drainage in CL-13-infected mice.

To examine the role of CD8⁺ T cells in damage to the conduit network during CL-13 infection, mice were depleted of CD8⁺ T cells before infection. In these mice, accumulation of tracer molecules was similar to that in uninfected mice (Fig. 3 D and E). This corresponded with the reduced disruption of the lymphoid architecture and increased cellularity observed during chronic LCMV infection in mice depleted of CD8⁺ T cells (data not shown) (9). Mice deficient in perforin and IFN-γ showed improved tracer accumulation in the WP after CL-13 infection when compared with WT mice (SI Fig. 9). However, neither knockout strain demonstrated levels of tracer accumulation comparable to uninfected mice, indicating that both cytotoxicity and cytokine production play a role in disruption of the conduits during CL-13 infection. Together, these data suggest that the integrity of the conduit network may be disrupted by responding CD8⁺ T cells during CL-13 infection, either preventing molecules from entering the network or allowing them to escape from the conduit lumen and perfuse into the splenic parenchyma.

FRC Express High Levels of PD-L1 During LCMV Infection.

Despite infection of significant numbers of FRC by LCMV CL-13 and immune-mediated disruption of conduit function, the complex network of FRC was not obviously killed or depleted from the spleen during infection. This suggested that the FRC may use mechanisms that protect them from T cell-mediated death. We recently demonstrated that one of the major mechanisms involved in the impairment of T cell function during LCMV CL-13 infection involves up-regulation of the molecule PD-1 on T cells (23). Interaction of this inhibitory receptor with its ligand PD-L1 directly influences T cell responses. PD-L1 is widely expressed on cells such as macrophages, DC, T and B cells, Kupffer cells in the liver, and β cells in the pancreas, among others (24). We examined expression of PD-L1 on FRC isolated from the spleen. The ER-TR7⁺ FRC were found to constitutively express PD-L1 and up-regulated the inhibitory ligand considerably after LCMV CL-13 infection (Fig. 4A). Up-regulation of PD-L1 expression of FRC was confirmed by staining the cells for gp38, a surface marker found on the lymphoid reticular cells (25). When spleen sections were examined via microscopy, high levels of PD-L1 were observed on the FRC in CL-13-infected mice (Fig. 4B). This suggested that PD-L1-expressing FRC in the lymphoid tissues may play a role in regulating the function of PD-1-expressing virus-specific T cells during chronic LCMV infection and that this inhibitory interaction (PD-1/PD-L1) may protect the FRC from being killed.

PD-L1 Blockade Induces Severe Immunopathology in Lymphoid FRC.

We recently demonstrated that blockade of the PD-1/PD-L1 inhibitory pathway can restore function in exhausted CD8⁺ T cells during chronic CL-13 infection (23). Given that PD-L1 was highly expressed on FRC during the early stage of CL-13 infection, we sought to examine the role for this expression by treating mice with anti-PD-L1 antibody (Fig. 5A). Two days after treatment, mice exhibited severe immunopathology in the spleen. In particular, mice treated with anti-PD-L1 antibody displayed noticeably increased disruption of the splenic stromal architecture (Fig. 5B). This included smaller WP regions and considerable alteration to the ER-TR7⁺ stromal network in the RP and WP. We found that ER-TR7 antigen was reduced by ≈54% (Fig. 5C) and appeared less uniform, indicative of loss or alteration of the stromal cells. Furthermore, PD-L1 blockade induced redistribution of viral antigen. Many large deposits of LCMV antigen were observed after treatment (Fig. 5B, arrows), suggesting possible uptake by phagocytic cells. As described above, no significant destruction or loss of the stromal network was detected in untreated CL-13-infected mice. Moreover, in mice lacking CD8⁺ T cells, FRC immunopathology was not observed after anti-PD-L1 treatment: ER-TR7 antigen expression was similar to untreated mice, and the spleen architecture appeared relatively intact (Fig. 5 B and C). Together, these data clearly demonstrate that PD-L1 expression prevents severe immunopathology and architectural disruption of the FRC network in the spleen, promoting persistence of LCMV CL-13 in infected lymphoid stromal cells.

Infections.

C57BL/6 mice were infected with 2 × 10⁵ pfu of Armstrong or 2 × 10⁶ pfu of CL-13 or the reassortant viruses Arm/13 and 13/Arm (18) intravenously. For a description of mouse strains used and antibody treatment see SI Materials and Methods.

Microscopy.

For a description of standard immunohistochemistry and transmission electron microscopy methods used, see SI Materials and Methods. For immunofluorescence, spleens were removed from mice and frozen in TissueTek OCT (Sakura FineTek, Torrance, CA). Twenty-micrometer cryostat sections were fixed in ice-cold acetone for 10 min. Sections were stained with ER-TR7 (Biogenesis, Bournemouth, U.K.), anti-type III collagen (Southern Biotechnology Associates, Brimingham, AL), anti-laminin (AbD Serotec, Raleigh, NC), anti-α-smooth muscle actin (Sigma, St. Louis, MO), polyclonal rabbit anti-desmin (Monosan, Uden, The Netherlands), polyclonal anti-LCMV guinea pig serum, or anti-PD-L1 biotin (eBioscience, San Diego, CA). Stains were visualized with Alexa Fluor 488-, 568-, or 647-conjugated goat anti-rat, anti-hamster, anti-rabbit, or anti-guinea pig Ig (Invitrogen, Carlsbad, CA) and analyzed on an LSM510 confocal microscope (Zeiss Microimaging). Images were prepared in NIH ImageJ, with a colocalization plug-in, and compiled in Photoshop (Adobe Systems, San Jose, CA). Analysis of ER-TR7⁺ antigen distribution after anti-PD-L1 treatment was performed by using ImageJ software and measurement functions to count ER-TR7⁺ pixels as a proportion of tissue area (n = 5–10 mice per group).

Fluorescent Tracers.

Ten-kilodalton lysine fixable dextran 568, OVA, and OVA-FITC (5 mg/ml) were purchased from Invitrogen. Tracers were injected intravenously in 200 μl of PBS, and mice were killed after 10 min. Spleens were immersion-fixed overnight in 4% paraformaldehyde, washed twice in PBS, and frozen in OCT. For quantitation of tracer-positive pixels per area of WP, spleens were analyzed by using NIH ImageJ software and built-in measurement functions (n = 14–22). Statistical analysis was performed by a two-tailed t test with 95% confidence intervals using Prism software (GraphPad).

Flow Cytometry.

Spleens were prepared as described previously (33). Cells were fixed by using the Cytofix/Cytoperm kit (BD Biosciences, San Jose, CA) and stained intracellularly with ER-TR7 antibody and Alexa Fluor 488 goat anti-rat IgG (Invitrogen) and costained with anti-LCMV nucleoprotein antibody conjugated to Alexa Fluor 647. Alternately, cells were stained with phycoerythrin-conjugated anti-PD-L1 (eBioscience) before fixation, or with anti-gp38 antibody (25) and anti-PD-L1-biotin, followed by goat anti-Syrian hamster FITC and streptavidin–allophycocyanin (Invitrogen). Appropriate rat or Syrian hamster IgG isotype antibodies were used as controls. PD-L1^−/− mice were used to control for PD-L1 staining. Samples were acquired by using a FACSCalibur flow cytometer (BD Biosciences).