Herpesvirus-Associated Lymphadenitis Distorts Fibroblastic Reticular Cell Microarchitecture and Attenuates CD8 T Cell Responses to Neurotropic Infection in Mice Lacking the STING-IFNα/β Defense Pathways

Type I IFN (IFN-α/β)–driven immune responses to acute viral infection are critical to counter replication and prevent dissemination. However, the mechanisms underlying host resistance to HSV type 1 (HSV-1) are incompletely understood. In this study, we show that mice with deficiencies in IFN-α/β signaling or stimulator of IFN genes (STING) exhibit exacerbated neurovirulence and atypical lymphotropic dissemination of HSV-1 following ocular infection. Synergy between IFN-α/β signaling and efficacy of early adaptive immune responses to HSV-1 were dissected using bone marrow chimeras and adoptive cell transfer approaches to profile clonal expansion, effector function, and recruitment of HSV-specific CD8+ T cells. Lymphotropic viral dissemination was commensurate with abrogated CD8+ T cell responses and pathological alterations of fibroblastic reticular cell networks in the draining lymph nodes. Our results show that resistance to HSV-1 in the trigeminal ganglia during acute infection is conferred in part by STING and IFN-α/β signaling in both bone marrow–derived and –resident cells, which coalesce to support a robust HSV-1–specific CD8+ T cell response.

T ype I IFNs (IFN-a/b) are pleiotropic cytokines with diverse functional roles ranging from innate host defense to immunoregulation (1,2). Acute viral infections stimulate rapid expression of IFN-a/b through various pathogen recognition receptor pathways to induce an antiviral state and prime adaptive immune responses (3). HSV type 1 (HSV-1) is a prototypical, neurotropic member of the herpesvirus family, which includes eight human pathogens (e.g., HSV-2, varicella-zoster virus, EBV, CMV, and so on) that establish chronic infections with varying tissue tropisms and clinical consequences. Clinical manifestations of HSV-1 typically result from viral recrudescence in orofacial mucosal sites innervated by infected neurons within the trigeminal ganglia-the reservoir for HSV-1 latency.
Ocular morbidities arising from herpesvirus infections represent a particularly significant clinical concern because diagnosis can be challenging (4)(5)(6). Herpesviruses are ubiquitous in the human population and often a danger for immunocompromised patients (7), thus identifying the molecular and cellular determinants of host resistance during acute infection could aid in the development of targeted therapies or vaccines.
The cytosolic DNA sensor signaling adaptor protein stimulator of IFN genes (STING) is paramount for host resistance to HSV-1 infection (8)(9)(10)(11)(12). However, the function of STING beyond immune surveillance and induction of IFN-a/b during HSV-1 infection is incompletely understood. In the current investigation, we sought to identify the contributions of STING and IFN-a/b signaling relative to innate and early adaptive immune responses to HSV-1 in the trigeminal ganglia (TG) to further define the host-pathogen interactions that arbitrate the severity of viral pathogenesis.
Animal models of ocular HSV-1 infection have long been used to investigate immunity and pathogenesis. We have recently identified the antiviral effector tetherin to be of particular importance in controlling HSV-1 neuroinvasion from the mucosal epithelium to the cornea-innervating TG in a STING-dependent manner (12). Along these lines, we hypothesized that STING counters HSV-1 replication within the TG by upregulating IFN-a/b-dependent innate responses. Collectively, our results show that STING is dispensable for IFN-a/b-dependent immunosurveillance in the peripheral nervous system but that IFN-a/b-induced innate responses are insufficient to control HSV-1 replication within the TG in the absence of STING. Furthermore, our data corroborate the longstanding principle that CD8 + T cell recruitment is essential for immunologic control of HSV-1 in the peripheral nervous system (13)(14)(15)(16). This investigation also dissects the requirements of IFN-a/b as a link between innate and adaptive immune responses to HSV-1 because of the established role of IFN-a/b in driving optimal CD8 + T cell activation and clonal expansion in other models (2,(17)(18)(19). Finally, our investigation explores deficiencies in adaptive immunity resulting from atypical lymphotrophic dissemination of HSV-1 in

Infection and plaque assays
Six-to twelve-week-old male and female mice were anesthetized for partial debridement of corneal epithelium with a 25-gauge needle prior to application of virus. Eyes were then blotted to remove the tear film, and HSV-1 McKrae was topically applied at an inoculum of 1000 PFU in 3 ml of saline. All infections were bilateral. Standard plaque assays on confluent CCL-81 Vero cell (American Type Culture Collection, Manassas, VA) monolayers were used to quantify HSV-1 titers in homogenized tissue supernatants as described previously (21).

PCR techniques and quantitation
RNA was isolated from TG and converted into cDNA as described previously (9). Relative gene expression was calculated by the standard 2 2DDthreshold cycle method, standardized to reference genes, and normalized to WT uninfected (UI) controls following semiquantitative real-time PCR using a CFX Connect thermocycler (Bio-Rad, Hercules, CA). Viral transcripts were amplified using iTaq supermix (Bio-Rad) with real-time primers from Integrated DNA Technologies (Coralville, IA). Primer sequences are listed in Supplemental Table I. PrimePCR technology (Bio-Rad) was used for transcript expression studies on IFN-stimulated genes (ISG) according to the manufacturer's instructions. Profiles of transcript expression represented by the cluster image map (or heat map) data supplement were generated using the National Cancer Institute's CIMminer tool freely accessible online.

Immunoassays
For evaluation of cytokines and growth factors, tissue was harvested at the indicated times postinfection (p.i.) and homogenized using a Tissue-Tearor homogenizer (Biospec Products, Bartlesville, OK) in 500 ml 13 PBS containing 13 Calbiochem protease inhibitor set 1 (EMD Millipore, Billerica, MA), and supernatants clarified by centrifugation as described previously (21). ELISAs for IFN-a and vascular endothelial growth factor (VEGF) were obtained from eBioscience (San Diego, CA) and R&D Systems (Minneapolis, MN), respectively. All other analytes were assayed using Luminex-based multiplex platforms from Bio-Rad (Bio-Plex) or EMD Millipore (Milliplex MAP). Data were transformed to reflect total analyte per milligram tissue. For quantification of intracellular signaling proteins, tissues were mechanically dissociated in a Bullet Blender homogenizer (Next Advance, Averill Park, NY) using 1.7-ml bead lysis tubes containing radioimmunoprecipitation assay lysis buffer supplied with protease inhibitor (Santa Cruz Biotechnology, Dallas, TX) and supplemented with PhosSTOP phosphatase inhibitor mixture tablets (Roche, Indianapolis, IN) at twice the suggested concentration. Lysates were incubated in a water-bath sonicator for 10 min to dissociate aggregates, supernatants were collected following centrifugation at 15,000 3 g for 10 min, and protein concentrations were determined using a Pierce bicinchoninic acid assay kit (Thermo Fisher Scientific, Pittsburgh, PA). Total and phosphorylated proteins were quantified using Luminex-based Bio-Plex Pro magnetic cell signaling assays (Bio-Rad); data reflect measured fluorescence obtained from 15 mg of sample protein input. All immunoassays were performed according to the manufacturers' specifications.

Bone marrow chimeras
Chimeric mice were produced as described previously (22). Briefly CD45 congenic WT and CD118 2/2 mice subjected twice to 600 Gy of gamma irradiation at a 4-h interval. Irradiated mice were subsequently treated with 3 3 10 6 CD45 congenic bone marrow cells (BMC) i.v. to reconstitute the hematopoietic compartment. Ten weeks later, BMC grafts were verified by analysis of leukocytes in the blood, which showed a .90% donor BMC composition relative to the CD45 congenic recipient allele. See Fig. 3A for a schematic.

Cell isolation and adoptive transfer
For adoptive transfer experiments, CD8 + T cells were obtained from singlecell suspensions of secondary lymphoid organs of naive or infected TCRtransgenic gBT-I.1 mice by MACS immunomagnetic isolation (Miltenyi Biotec, Auburn, CA). Isolated cells were incubated in 1 mM CFSE (eBioscience) and washed, and 3 3 10 6 cells were injected i.v. into recipients retro-orbitally without irradiation. Purities of immunomagnetically enriched cells were evaluated by flow cytometry and found to be .80% CD3 + CD8 + double-positive cells relative to the total CD45 + population.

Flow cytometry
All tissues were dissociated in RPMI 1640 medium supplemented with 10% heat-inactivated FBS, 13 antibiotic/antimycotic, and 10 mg/ml gentamicin (Life Technologies) (i.e., complete media). Lymph nodes were macerated into single-cell suspensions over 40-mm mesh. TG specimens were mechanically dissociated using a Dounce homogenizer (Thermo Fisher Scientific) and filtered through 40-mm mesh. For analysis of circulating leukocytes, 100 ml of blood was collected from the superficial temporal facial vein, mixed with 5 ml of 0.5 M EDTA to prevent coagulation, and treated twice for 2 min with 1.0 ml of lysing buffer (150 mM ammonium chloride, 10 mM postassium bicarbonate, and 0.1 mM EDTA) and resuspended in complete medium. Abs for flow cytometry were purchased from eBioscience, BD Biosciences (San Jose, CA), and Tonbo Biosciences (San Diego, CA). Mouse MHC Class I Tetramer [K(b)-SSIEFARL, HSV-1 gB 498-505 ] was provided by the NIH Tetramer Core Facility (Atlanta, GA) and incubated with samples prior to the addition of Ab. Cells were incubated with anti-CD16/32 Fc block for 10 min prior to the addition of other Abs. For surface Ag labeling and wash steps, cells were incubated with Ab in wash buffer (1% BSA in 13 PBS) for 30 min, washed twice, fixed in phosphate-buffered 1% paraformaldehyde, and resuspended in wash buffer for analysis. Intracellular immunolabeling was conducted using Cytofix/Cytoperm fixation permeabeabilization solution and Perm/ Wash buffer (BD Biosciences) following surface labeling for functional assays. Lymphocyte viability was evaluated using the Miltenyi Biotec Annexin V-FITC kit, according to the manufacturer's directions. Samples were analyzed on a MACSQuant-10 flow cytometer with MacsQuantify software (Miltenyi Biotec) unless indicated otherwise. Flow plots substantiating gating strategies are shown in Supplemental Fig. 1.

T cell functional assays
Spleens were teased and filtered into single cell suspensions, subjected to 0.84% ammonium chloride for 2 min twice to lyse red cells, washed, and resuspended in complete medium. For functional assays, 1 3 10 6 splenocytes were cultured in 1.0 ml complete medium for 3 h at 37˚C, 5.0% CO 2 in the presence of DMSO as a vehicle control, 2.0 mg of gB 498-505 peptide, or 50.0 ng of PMA and 800 ng of ionomycin as shown previously (21). After 1 h in culture, 0.67 ml of GolgiStop protein transport inhibitor (BD Biosciences) was added to facilitate intracellular IFN-g or IL-17 immunolabeling. Cells were fixed and labeled after a total of 4 h in culture.

Lymphocyte depletion
Cell-specific or isotype control Abs (BioXCell, West Lebanon, NH) were injected i.p. for systemic delivery. For NK cell depletion, 300 mg of anti-NK.1.1 or IgG control Ab was injected 1 d before and after HSV-1 infection as reported by Crouse et al. (23). For CD8 + T cell depletion, 200 mg of anti-CD8a or IgG control Ab was injected on days 3 and 5 p.i. Depletion efficiency was confirmed by flow cytometry in tissues of interest.

Histology and microscopy
Lymph nodes harvested from HSV-1-infected mice and UI controls were fixed in Lymph-ID (BioSafe Supplies, Orlando, FL) prior to paraffin embedding and sectioning. Five-micrometer mandibular lymph node (MLN) sections were cut and mounted on glass slides. H&E staining and a Modified Gomori's Reticulum stain (StatLab, Baltimore, MD) were performed on sequential serial sections. Prepared slides were imaged on a Nikon E400 microscope outfitted with a 340 objective lens using MetaVue imaging software (Nikon, Melville, NY).

Statistical analysis
Normal distribution was assumed a priori for all data sets. Graphpad Prism 5 was used for statistical analysis. All data reflect mean 6 SEM. One-way ANOVAs with Student-Newman-Keuls multiple comparisons tests or two-way ANOVAs with Bonferroni posttests were used to assess data with one or two independent variables, respectively. Significance thresholds for each comparison are denoted as follows: *p , 0.05, **p , 0.01, and ***p , 0.001.

Loss of STING enhances HSV-1 neuropathogenesis independent of canonical IFN-a/b responses
The neurovirulence of HSV-1 is determined in part by the efficiency of neuroinvasion and subsequent replicative proficiency within neuronal ganglia. Therefore, to investigate our hypothesis that STING is essential for host resistance to HSV-1 in the TG, we assessed viral titers and lytic gene expression following ocular HSV-1 infection of WT, STING 2/2 , and type I IFN receptor a-chain-deficient (CD118 2/2 ) mice. By day 5 p.i., titers of infectious virus in the TG of STING 2/2 and CD118 2/2 mice were substantially higher than their WT counterparts (Fig. 1A). We have previously shown that STING 2/2 and CD118 2/2 mice harbor more virus in the TG at day 3 p.i. than WT (12). Real-time PCR was performed on viral transcripts representing all three temporally regulated HSV-1 lytic gene classes to distinguish whether the increased viral burden was merely because of enhanced neuroinvasion from the infected cornea or compounded by efficient replication within the TG. Excised ganglia from each experimental group were analyzed for expression of infected cell protein 0 (ICP0), thymidine kinase (TK), and glycoprotein B (gB) as representatives of immediate-early, early, and late viral gene products, respectively (Supplemental Table I). Viral gene expression was .10 times greater in TG from STING 2/2 and CD118 2/2 mice relative to WT (Fig. 1B). Thus, the elevated viral burden observed in STING 2/2 mice resulted not only from enhanced neuroinvasion (11,12) but was augmented by efficient viral replication within the TG.
Host resistance to acute HSV-1 infection is predominantly thought to be dependent upon induction of IFN-a/b followed by transcriptional activation of ISG and generation of cell-mediated immunity. A PCR-based ISG array including antiviral effectors, positive regulators, and feedback inhibitors was used to substantiate whether enhanced viral replication in TG from STING 2/2 mice corresponded to diminished ISG responses. TG specimens from STING 2/2 mice exhibited an ISG profile remarkably similar to WT (Fig. 1C, Supplemental Fig. 2) despite the amplified HSV-1 neurovirulence. Furthermore, STING-dependent inflections in suppressor of cytokine signaling (SOCS)-1 and SOCS-3 expression previously reported in BM-derived macrophages (24) were not observed in the TG during  Fig. 2. * and^, differences from WT and STING 2/2 , respectively. *p , 0.05, ***p , 0.001 differences from WT. ∧ p , 0.05, ∧∧ p , 0.01 differences from STING 2/2 . HSV-1 infection (Fig. 1C). Collectively, our data show that STINGindependent innate sensing pathways compensate for IFN-a/b responses in the TG during HSV-1 infection but that ISG induction alone is insufficient to suppress viral replication.
Multiple studies have confirmed the importance of canonical IFN signaling through STAT1 with respect to ISG-mediated resistance to HSV-1 in the TG (25)(26)(27). Ancillary IFN signaling also propagates physiological changes through STAT1-independent mediators such as the MAPK, ERK, and protein kinase B (AKT) pathways to elicit antiviral defenses (28)(29)(30)(31). Although these signaling pathways are also active in healthy nerve ganglia, modulation of such responses is likely important for control of HSV-1 in the peripheral nervous system. For example, IFN-a/b/g-dependent STAT1 phosphorylation is reportedly subdued in neurons relative to mitotically active cells as a measure of protection against cytotoxicity (31,32).
To further analyze IFN-related signal transduction in TG from STING 2/2 mice during HSV-1 infection, concentrations of IFN-a and IFN-g were evaluated along with phosphorylation of ancillary downstream signaling mediators at day 5 p.i. in TG homogenates ( Fig. 2). Briefly, IFN-a protein levels were higher in WT than STING 2/2 samples, whereas the concentration of IFN-g was elevated in STING 2/2 mice relative to WT ( Fig. 2A, 2B). Although IFN-g-induced neuroprotective signaling through Erk1/2 has been previously described (31), Erk1/2 phosphorylation was increased in TG from all experimental groups pursuant to HSV-1 infection irrespective of measured IFN ligand availability (Fig. 2C). No changes in AKT phosphorylation were detected in parallel (data not shown). However, efficient phosphorylation of p38 MAPK was only observed in WT samples in terms of phosphorylated-tototal protein ratios (Fig. 2D). Multiple studies have reported protective mechanisms for IFN-and stress-associated p38 MAPK signal transduction including regulation of the balance between apoptosis and autophagy (33)(34)(35). Autophagy was recently inferred to be an important STING-dependent protective countermeasure against HSV-1 in the nervous system (11). In this study, susceptibility to HSV-1 neuropathogenesis in the TG of STING 2/2 mice was essentially independent of local ISG induction through canonical IFN signaling pathways. Taken together, ancillary STAT1-independent IFN-signal transduction involving p38 MAPK may confer protection against HSV-1 in the TG through an unknown mechanism possibly involving autophagy-as reflected in other models (11,34,36). Furthermore, the early contributions of CD8 + T cells with respect to control of HSV-1 remained to be explored in STING 2/2 mice.

IFN-a/b signaling in resident and BM-derived cells contributes to resistance against HSV-1 neuropathogenesis via CD8 + T cell recruitment
BM chimeras were generated (Fig. 3A) in an effort to dissect the contributions of IFN-a/b signaling in resident and BM-derived cells with respect to host resistance to HSV-1 at day 5 p.i. Intact IFN-a/b signaling was required in both resident and BM-derived cells for optimal resistance to HSV-1 in terms of viral burden in the TG (Fig. 3B). However, loss of IFN-a/b signaling in the resident cell population (i.e., CD118 2/2 recipients) compromised recruitment of gB 498-505 -specific CD8 + T cells to the TG (Fig. 3C, Supplemental Fig. 1), reflecting the immunodominant CD8 + T cell epitope of HSV-1 in C57BL/6 mice based on MHC class I tetramer labeling (37,38). Neurotropism is not the exclusive route of HSV-1 dissemination because infectious virus is also detectable in the draining MLN (Fig. 3D) through a mechanism involving hematogenous and/or lymphatic spread where it can disrupt lymph node integrity and generation of effective T cell responses (22,39). Consistent with this notion, total CD8 + T cell infiltration into the TG of BM-chimeric mice was inversely proportional to the amount of virus recovered from the MLN (Fig. 3C, 3D). Upward of 80% of the CD8 + T cell infiltrate in the TG during the early stages of acute HSV-1 infection responds to known viral epitopes, and 50% are specific for the gB 498-505 epitope (38). However, 10-20% of the CD8 + T cells reflect bystander-activated cells that do not target HSV-1 based on analysis of TCR specificity for known epitopes and observations of OVAspecific transgenic T cells within HSV-1-infected TG in cotransfer experiments (38,40). The phenomenon of cytokine-driven bystander CD8 + T cell activation is well documented in other contexts (41)(42)(43).
Concentrations of IFN-driven chemokines associated with recruitment of effector CD8 + T cells (16,(44)(45)(46), specifically CXCL10 and CCL2, were produced in the TG by resident cells and/or BM-derived cells (Fig. 3E) in a compensatory manner largely . Total and phosphorylated Erk1/2 (C) and p38 MAPK (D) protein in TG measured by suspension array on tissue homogenates (n = 8-11 per group; three independent experiments). Values reflect median fluorescence intensity (MFI). The phosphorylated to total ratios are also shown. Statistical differences were determined by one-way ANOVA with Student-Newman-Keuls multiple comparisons tests (A and B) or with Dunnetts multiple comparisons tests comprising experimental groups to UI controls (C and D). Bars represent mean 6 SEM. For (A) and (B), * and^reflect differences from WT and STING 2/2 , respectively. *Differences from UI for (C) and (D). *p , 0.05, **p , 0.01, ***p , 0.001 differences from WT, ∧∧∧ p , 0.001 differences from STING 2/2 . dependent on IFN-a/b signaling in one component or the other. However, no apparent differences in CXCL1 or CCL5 concentrations were observed among experimental groups (data not shown). To further substantiate the importance of CXCL10, viral titers in the TG were compared in mice deficient in CXCL10, its receptor CXCR3, or both. Global loss of CXCR3 had a modest effect on viral burden in the TG; however, CXCL10 2/2 mice exhibited a magnified susceptibility to HSV-1 in the TG that was paralleled in DKO mice (Fig. 3F). Taken together, these data support a paradigm in which IFN-a/b signaling in tissue-resident and BM-derived cell populations is important for countering viral dissemination, preserving early adaptive CD8 + T cell responses, and facilitating lymphocyte recruitment to infected tissues.

IFN-a/b signaling is dispensable in CD8 + T cells without a concomitant loss in effector function against HSV-1
TCR-transgenic mice (gBT-I.1) in which nearly all CD8 + T cells recognize the gB 498-505 epitope of HSV-1 (47) were backcrossed with CD118 2/2 mice (gBT-I.1 3 CD118 2/2 ) to further identify the impact of IFN-a/b signaling in CD8 + T cells relative to early immunologic control of HSV-1 in the TG. Adoptive transfer of naive HSV-1-specific CD8 + T cells isolated from gBT-I.1 or gBT-I.1 3 CD118 2/2 mice into virally infected CD118 2/2 mice corroborated the importance of CD8 + T cell responses with respect to IFN-a/b-independent mechanisms of protection against acute HSV-1 infection (Fig. 4). Specifically, addition of HSV-specific CD8 + T cells into highly immunocompromised CD118 2/2 mice lowered virus titers in the TG 10-fold (Fig. 4A) correlative with a restoration of T cell infiltration into the TG (Fig. 4B, 4C). This treatment also lowered viral titers in the MLN (Fig. 4D) and restored T cell populations therein (Fig. 4E-H) independent of IFN-a/b signaling in the exogenous CD8 + T cells. Furthermore, recovery of T cell responses in recipient mice (Fig. 4H) was principally attributed to restoration of the endogenous host immune response because the majority of T cells contained in the MLN or infiltrating into the TG of recipient mice were not exogenous CFSE-labeled TCR-transgenic cells that colabel with the gB 498-505 tetramer (Fig. 4B, 4C, 4F, 4G; quantification of CFSE + cells not shown).
Direct infection of TCR-transgenic mice reflected a delayed yet important contribution of CD8 + T cells with respect to HSV-1 titers in the TG among CD118-sufficient and -deficient mice ( Fig. 5A-C). Virus was not detected in the MLN of WT or gBT-I.1 mice at day 5 p.i. However, HSV-1 disseminated to the MLN in gBT-I.1 3 CD118 2/2 and CD118 2/2 mice by day 5 p.i. with 10to 100-fold higher titers in the CD118 2/2 mice (Fig. 4D, data not shown). As previously noted (47), TCR transgenic mice exhibit deficiencies in the total numbers of CD4 + T cells (Fig. 5D). Consistent with the observed reduction in viral burden, CD8 + T cell responses were restored in the MLN of gBT-I.1 3 CD118 2/2 mice compared with CD118 2/2 mice (Fig. 5E, 5F). Analysis of IFN-g production in TCR-transgenic CD8 + T cells following in vitro stimulation with PMA and ionomycin or SSIEFARL peptide corresponding to the gB 498-505 epitope indicated no functional abnormalities (Fig. 5G-I).
Elevated NK cell counts were observed in the MLN of the gBT-I.1 3 CD118 2/2 animals relative to their CD118 2/2 counterparts (data not shown). NK cell-mediated fratricide of CD8 + T cells in secondary lymphoid organs is observed in models of LCMV infection in the absence of IFN-a/b signaling, such that NK cell depletion enhances the number of LCMV-specific CD8 + T cells (23,48). We sought to identify the impact of NK cells on CD8 + T cell fratricide in CD118 2/2 and gB 3 CD118 2/2 mice following HSV-1 dissemination to the MLN. In contrast to LCMV models, NK cell depletion did not enhance the total number or affect the functional capacity of CD8 + T cells in CD118 2/2 or gBT 3 CD118 2/2 mice infected with HSV-1 (Fig. 5J-L). Rather, depletion of NK cells reduced the number of HSV-specific CD8 + T cells. These data are consistent with established findings showing that NK cells facilitate HSV-1 containment to reduce viral dissemination (49,50). These results also support another report suggesting that NK cell depletion does not enhance CD8 + T cell responses to adenovirus in the absence of IFN-a/b signaling (51).
CD8 + T cell responses are distorted in HSV-1-infected STING 2/2 mice Because host resistance in STING 2/2 mice was lost despite preservation of IFN-a/b signaling within the TG, we investigated early adaptive and cellular innate immune responses in these animals. Infiltration of CD4 + , CD8 + , and gB 498-505 -tetramer + CD8 + T cells into the TG was reduced in STING 2/2 mice relative to WT such that the numbers of TG-infiltrating cells in STING 2/2 mice were indistinguishable from that observed in the highly immunocompromised CD118 2/2 mice (Fig. 6A-C). However, NK cell infiltration was not impacted by deficiencies in STING or CD118 mice (Fig. 6D). Patterns in TG-infiltrating CD11b + myeloid cells produced several noteworthy observations but were not associated with the susceptibility of STING 2/2 mice. Consistent with the reported repressive role of IFN-a/b signaling (52), neutrophil infiltration was limited in TG from STING 2/2 mice (Fig. 6E) despite a high viral burden (Fig. 1A). In addition, the phenotypes of TG-infiltrating monocytes diverged from what is seen in mucosal tissues in CD118 2/2 mice. Specifically, Ly6G 2 Ly6C + inflammatory monocytes are virtually absent from the corneas and lungs of HSV-1-and influenza-infected CD118 2/2 mice, respectively (12,53,54). In contrast, inflammatory monocytes were abundant in the TG of CD118 2/2 mice following HSV-1 infection (Fig. 6F). However, there were no significant differences in total numbers of infiltrating monocyte populations among TG specimens from WT, STING 2/2 , or CD118 2/2 mice (Fig. 6F, 6G). Distinct anatomical host niches differ in profiles of neutrophil recruitment and monocyte differentiation during inflammation and infection, as indicated above comparing cellular innate responses in nervous and mucosal tissue and validated by others (52,55,56).
Because differences in viral burden in the TG of STING 2/2 mice could not be readily interpreted as a consequence of overt deficits in IFN-a/b signaling or cellular innate responses, the observed defect in CD8 + T cell infiltration into the TG of STING 2/2 mice was explored further. Chemokines associated with CD8 + T cell recruitment were analyzed in the TG at day 5 p.i. (Fig. 6H-K). Concentrations of CCL2, CCL5, and CXCL10 were significantly elevated in TG from STING 2/2 mice relative to WT (Fig. 6H, 6J, 6K); thus, the availability of chemotactic factors in the TG microenvironment did not substantiate the dearth of CD8 + T cell infiltration in these animals.
T cell responses to HSV-1 infection have not been described in STING 2/2 mice despite their elevated susceptibility to multiple routes of infection (8,11). Immune responses were consequently evaluated to compare profiles of T cell generation, circulation, and function in WT, STING 2/2 , and CD118 2/2 mice. Flow cytometrybased analysis of T cell populations in the MLN revealed that CD4 + , CD8 + , and gB 498-505 -tetramer + CD8 + T cells in STING 2/2 mice expand identically to WT following infection (Fig. 7A-C). In contrast, T cells precipitously decline in the MLN of HSV-1infected CD118 2/2 mice concomitant with viral dissemination to the MLN (39) as was observed in the present investigation. Further phenotypic characterization of STING 2/2 CD8 + T cells in the MLN revealed proficient activation as determined by upregulation of the IL-2 receptor CD25 (Fig. 7D) yet a 2-fold increase in proliferating CD69 + CD8 + T cells relative to WT (Fig. 7E). No differences were noted in sphingosine-1-phosphate receptor 1 (S1P1) + CD8 + T cells as a transient marker of lymph node egress (Fig. 7F). In addition, a 2-to 3-fold increase in CD44 high CD62L low effector memory CD8 + T cells was observed in MLN of STING 2/2 relative to WT with no observable differences in the CD44 high CD62L high central memory population (data not shown). However, the number of circulating CD8 + T cells in the blood were substantially lower in STING 2/2 and CD118 2/2 mice compared with WT (Fig. 7G), with a similar trend in activated CXCR3 + CD8 + T cells (Fig. 7H). Approximately 50% of the circulating CXCR3 + CD8 + T cells coexpressed high levels of CD44 in WT and STING 2/2 mice, although only 30% of the CXCR3 + CD8 + T cells in CD118 2/2 exhibited the same phenotype (data not shown).
In contrast to T cell responses observed in DNA plasmid-vaccinated and tumor-bearing STING 2/2 mice (8,57), in vitro stimulation revealed no intrinsic defect in the ability of CD4 + or CD8 + T cells recovered from HSV-1-infected STING 2/2 mice to produce IFN-g in mixed splenocyte cultures containing WT-equivalent T cell frequencies (Fig. 7I) following PMA and ionomycin stimulation (Fig. 7J,  7K). However, it was noted that CD4 + T cells obtained from HSV-1infected CD118 2/2 mice were hyperresponsive with respect to IFN-g and IL-17 production (Fig. 7K, 7L). Furthermore, TCR-dependent IFN-g production by CD8 + T cells did not differ between WT and STING 2/2 splenocyte cultures pulsed with SSIEFARL peptide corresponding to the gB 498-505 epitope of HSV-1 (data not shown).
Because CD8 + T cells were functional in STING 2/2 mice, systemic depletion of CD8 + T cells was pursued to determine whether the reduced number of responding CD8 + T cells in the TG-impacted HSV-1 titers. Ab-mediated CD8 + T cell depletion was effective based on cell counts in the MLN reaching levels comparable to the CD118 2/2 mice, yet the CD4 + T cell population was unaffected (Fig. 8A, 8B). Relative to IgG isotype control-treated STING 2/2 mice, CD8 + T cell-depleted STING 2/2 mice displayed an increase in viral titer in the TG comparable to CD118 2/2 mice (Fig. 8C).
T cell complementation studies were subsequently explored in STING 2/2 mice. Adoptive transfer of CD8 + T cells from naive gBT-I.1 mice into STING 2/2 mice had no effect on HSV-1 titers in the TG by day 5 p.i. (Fig. 8D) despite seeding of exogenous T cells in the MLN (data not shown). However, adoptive transfer of in vivo-activated (i.e., primed) CD8 + T cells from HSV-1-infected gBT-I.1 mice into STING 2/2 mice led to a 10-fold reduction of HSV-1 titers in the TG relative to STING 2/2 controls (Fig. 8D). Consistent with this result, adoptive transfer of primed, but not naive, gBT-I.1 CD8 + T cells also reduced viral titers in the MLN (Fig. 8E). Collectively, our results substantiate the importance of early responding CD8 + T cells in the peripheral nervous system to control primary acute HSV-1 infection and suggest that the MLN microenvironment is affected by viral dissemination such that mobilization of recently activated CD8 + T cells is compromised.

HSV-1 affects T cell responses in the MLN by modulating the supportive network of FRCs
MLN were surveyed for HSV-1 titers and cytokines or growth factors associated with T cell survival to identify whether the  Fig. 1 for flow cytometry gating strategies. Statistical differences were determined by one-way ANOVA with Student-Newman-Keuls multiple comparisons tests. Bars represent mean 6 SEM. * and^Differences from WT and STING 2/2 , respectively. *p , 0.05, **p , 0.01, ***p , 0.001 differences from WT. ∧ p , 0.05, ∧∧ p , 0.01, ∧∧∧ p , 0.001 differences from STING 2/2 . discrepancy between the numbers of CD8 + T cells in the MLN and in circulation in STING 2/2 mice were elicited by pathological alterations within the MLN following viral dissemination, as has been reported in CD118 2/2 mice (22,39). Specifically, we hypothesized that FRC networks are lost in HSV-1-infected lymph nodes, leading to compromised leukocyte trafficking and survival. Viral dissemination to the MLN was not observed in WT mice by day 5 p.i., but HSV-1 was readily detectable in MLN from STING 2/2 mice albeit at lower titers than was observed in CD118 2/2 mice (Fig. 9A). Variances in IFN-a 2,4 and IFN-g were noted comparing STING 2/2 and WT MLN (Fig. 9B, 9C) in addition to a trend in increased levels of VEGF in STING 2/2 MLN (Fig. 9D). The total quantity of IL-2 was consistent in the MLN of all groups (Fig. 9E). However, a conspicuous elevation of IL-6, , and sphingosine-1-phosphate receptor 1 (S1P1) (F) to evaluate activation, proliferation, and egress, respectively, at day 5p.i. (n = 6-16 mice per group; three to four independent experiments). Total number of CD8 + T cells (G) and activated CXCR3 + CD8 + T cells (H) in the peripheral circulation at day 5 p.i. (n = 9-16 mice per group; four to five independent experiments). Functional analysis of T cells in mixed splenocyte cultures at day 5 p.i. showing (I) T cell ratios, IFN-g production in CD8 + (J) and CD4 + (K) T cells over background in PMA and ionomycin-stimulated cultures, and (L) IL-17 production in stimulated CD4 + T cells over background (n = 6-11 mice per group; three to four independent experiments). See Supplemental Fig. 1 for flow cytometry gating strategies. Statistical differences were determined by one-way ANOVA with Student-Newman-Keuls multiple comparisons tests. Bars represent mean 6 SEM. * and^Differences from WT and STING 2/2 , respectively. *p , 0.05, **p , 0.01, ***p , 0.001 differences from WT. ∧∧ p , 0.01, ∧∧∧ p , 0.001 differences from STING 2/2 . IL-7, IL-12p70, and IL-15 was noted in CD118 2/2 mice despite the synchronous loss of T cells in these animals (Fig. 9F-I). Moreover, the abundance of these cytokines did not differ between STING 2/2 and WT mice (Fig. 9C-F). Propidium iodide and annexin V dyes were used to assess survival of CD8 + T cells in the MLN. Although STING 2/2 mice had a total CD8 + T cell count in the MLN similar to WT (Fig. 7B), the number of dead CD8 + T cells in STING 2/2 mice was disproportionately high with a parallel reduction in the viable cell population (Fig. 9J). Thus, CD8 + T cell viability in STING 2/2 mice is compromised following viral dissemination to the MLN, as was previously reported for CD118 2/2 mice (39) and corroborated in this study.
Gross visual examination revealed that MLN excised from STING 2/2 and CD118 2/2 mice were morphologically hypertrophic compared with WT specimens (data not shown). Pathologic edema is prevalent in MLN of CD118 2/2 mice following HSV-1 dissemination to the MLN, as previously shown by T-2 weighted magnetic resonance imaging highlighting gross lymphadenopathy (22). Diffuse hypocellularity and edema were substantiated histologically in MLN sections from CD118 2/2 mice within the central paracortical-medullary zone (Fig. 10A) and peripheral subcapsular-perifollicular zone (Fig. 10B) following HSV-1 dissemination. In contrast, MLN from STING 2/2 mice exhibited WT-equivalent densities of mononuclear cells without extensive edema at day 5 p.i. (Fig. 10A, 10B). No hypoplasia or other aberrant pathology were observed in UI MLN from any experimental group (Fig. 10A). Although edema in the corneas and lymph nodes of HSV-1-infected CD118 2/2 mice is associated with the loss of lymphatic vessels (22), we have recently shown that lymphatic vessels are preserved in the mucosae of HSV-1-infected STING 2/2 mice (12).
Lymph node architecture and adaptive immune responses are also modulated by changes within the FRC network during infection (20,58). Histologic examination of FRC networks within MLN sections from WT, STING 2/2 , and CD118 2/2 mice was achieved using a classical silver stain technique to identify reticulin fibers (Fig. 10A, 10B). Although no differences were observed in the FRC network microarchitecture in UI MLN samples (Fig. 10A), distinct alterations following infection were observed in STING 2/2 and CD118 2/2 mice (Fig. 10A, 10B). Extended tracks of fine branching fibers were observed in WT MLN before and postinfection (Fig. 10A, 10B) with clear compartmentalization of germinal centers (Fig. 10B). In contrast, the FRC network within the MLN of STING 2/2 mice was predominately composed of discontinuous bone spicule-like formations indicative of FRC damage at day 5 p.i. (Fig. 10A). Contrary to our hypothesis, the FRC network in CD118 2/2 MLN exhibited an exceptionally dense, continuous labyrinthine phenotype following infection consistent with FRC proliferation and fibrosis (Fig. 10A, 10B). However, changes in FRC networks were not observed along MLN trabeculae (Fig. 10B).
Collectively, these observations support disparate mechanisms of FRC-associated T cell dysfunction in STING 2/2 and CD118 2/2 mice resulting in impaired immunity to HSV-1. High concentrations of cytokines associated with lymphocyte survival (Fig. 9F-I) despite concurrent loss of both CD4 + and CD8 + T cells (Fig. 7A-C) in the MLN of CD118 2/2 mice is consistent with reticulin fibrosis (Fig. 10) and loss of lymphatic vessels within the MLN (22). Soluble factors in fibrotic tissues are prone to entrapment in the extracellular matrix, and T cells require efferent lymphatic vessels to enter the bloodstream from lymph nodes. FRCs respond to IFN-g and IFN-b by increasing MHC class I expression, are able to activate naive CD8 + T cells, but restrict effector memory differentiation via upregulation of the coinhibitory molecule programmed death-ligand 1 (PD-L1) (59). On the basis of the high concentration of IFN-g (Fig. 9C) and observations of FRC damage (Fig. 10A) in the MLN of STING 2/2 mice, we attribute the lack of effector CD8 + T cell mobilization (Fig. 7H) to discontinuity of the FRC network within the MLN. Another possible mechanism may involve FRCmediated inhibition of proliferating or recently activated effector CD8 + T cells responding to Ag within the lymph node, as substantiated by the high proportion of dying/dead CD8 + T cells therein (Fig. 9J) in STING 2/2 mice relative to WT. Consistent with this, adoptive transfer of primed but not naive HSV-1-specific CD8 + T cells were able to reduce HSV-1 titers in STING 2/2 mice (Fig. 8D, 8E). Thus, virus-associated pathological alteration of the FRC network is a likely (but not necessary a causal) mechanism responsible for the T cell immunodeficiency. However, it is T cells in the MLN at day 6 p.i., (C) HSV-1 titers in the TG following CD8 + T cell depletion (n = 5-7 mice per group; three independent experiments). Adoptive transfer of 3 3 10 6 naive CD8 + T cells isolated from UI gBT-I.1 mice or primed CD8 + T cells isolated from HSV-1-infected gBT-I.1 mice and injected i.v. into STING 2/2 mice 24 h postinfection showing impact on viral titers in the TG (D) and MLN (E) at day 5 p.i. relative to WT controls (n = 6-12 mice per group; three to five independent experiments; ND, not detected). See Supplemental Fig. 1 for flow cytometry gating strategies. Statistical differences were determined by one-way ANOVA with Student-Newman-Keuls multiple comparisons tests. Bars represent mean 6 SEM. For (A)-(C) * and^differences from STING 2/2 IgG and STING 2/2 anti-CD8a, respectively. In panels (D) and (E), * and^reflect differences from WT and STING 2/2 controls, respectively, unless indicated otherwise. *p , 0.05, **p , 0.01, ***p , 0.001 differences from WT. ∧ p , 0.05 differences from STING 2/2 . conceivable that T cell priming is also impaired in STING 2/2 mice as deficiencies in innate immunity involving APCs have been observed (8,12,57,60).

Discussion
The current investigation provides insight into host countermeasures that facilitate immunologic resistance to acute HSV-1 infection in divergent susceptible cell populations and anatomic compartments. Notably, our data suggest that IFN-a/b-mediated resistance to HSV-1 replication in the peripheral nervous system involves STING and noncanonical IFN-a/b signaling that we speculate includes the p38 MAPK pathway. Modulation of MAPK/ERK and AKT signaling pathways by HSV has been reported with varied results primarily in mitotically active cells, yet the contributions of IFN to these pathways has been neglected (61)(62)(63)(64)(65). Downstream signaling pathways mediated by IFN-a/b signaling and STING are highly regulated (66). Accumulating evidence suggests that neurons exhibit a predilection for alternative IFN signaling cascades relative to mitotically active cells-a strategy of protection against cytotoxicity (31,32); this differs from the orthodox supposition that STAT1-dependent ISG responses arbitrate antiviral defenses ubiquitously.
The protective effects of IFN-a/b signaling and CD8 + T cell responses during the acute phase of HSV-1 infection have been well documented individually; however, the mechanistic link between these countermeasures is incompletely understood. In this study, we show that a deficiency in immunosurveillance against HSV-1 via loss of STING or CD118 leads to atypical viral dissemination, lymphadenitis, and defective adaptive immune responses. Our experiments detail the impact of IFN-a/b and STING with respect to activation, function, mobilization, and recruitment of effector CD8 + T cells during acute infection. We and others have demonstrated previously that CD118 2/2 and STING 2/2 mice succumb to encephalitis by day 5-7 p.i. when ocularly infected with neurovirulent HSV-1 strains such as McKrae or strain 17 (11,67). Thus, the experiments conducted in this study reflect early stages of the adaptive CD8 + T cell responses to infection rather than the peak of T cell infiltration into the TG, which occurs around days 12-14 p.i. in WT mice (14,15). Our data show that early T cell responses help control viral replication in the TG; however, CD8 + T cell responses do not restrict viral dissemination to the CNS or prevent lethality because of encephalitis in CD118 2/2 or STING 2/2 mice (data not shown).
Naive CD8 + T cells require Ag presentation, costimulation, and a cytokine-dependent signal for activation and optimal effector function. Moreover, IFN-a/b serves as one cytokine that supports CD8 + T cell activation (17). Our data contribute two seminal observations that challenge current understandings of IFN-a/b signaling on T cell responses to viral infection (23,48). First, our data show that IFN-a/b signaling enhances but is not requisite for CD8 + T cell effector function during HSV-1 infection. Second, we show that immunodeficiency can manifest as a secondary outcome of pathological alterations to FRC networks pursuant to HSV-1 dissemination into secondary lymphoid organs. This is consistent with other models showing abrogated T cell responses resulting from virus-associated damage to FRCs (59,(68)(69)(70). One caveat to our studies is that adoptive transfer of HSV-specific TCR-transgenic CD8 + T cells creates a high precursor frequency that may circumvent physiological T cell priming.
Lymph nodes enlarge during immune responses to provide for optimal engagement between foreign Ags and lymphocytes. This process involves the coordinated activities of many constituents ranging from hematopoietic cells, including macrophages, dendritic cells, mast cells, and lymphocytes, to nonhematopoietic structures, including lymphatic vessels and FRC conduits for Ag transport and ; the immunomodulatory growth factor VEGF A (D); T cell differentiation or survival including IL-2 (E), IL-6 (F), IL-7 (G), IL-12p70 (H), and IL-15 (I) were evaluated at day 5 p.i. (n = 9 mice per group; three independent experiments). Statistical differences for (A)-(I) were determined by one-way ANOVA with Student-Newman-Keuls multiple comparisons tests. * and^, differences from WT and STING 2/2 , respectively. (J) Viability of CD8 + T cells recovered from MLN on day 5 p.i. assessed by propidium iodide and Annexin V staining (n = 5-9 mice per group; two to three independent experiments). Parameters for determination of cell viability are shown in inset dot plot; differences were determined by two-way ANOVA with Bonferroni posttests. Bars represent mean 6 SEM. *p , 0.05, **p , 0.01, ***p , 0.001 differences from WT. ∧ p , 0.05, ∧∧ p , 0.01, ∧∧∧ p , 0.001 differences from STING 2/2 . leukocyte migration (71)(72)(73)(74). Within the lymph node microenvironment, FRCs provide structural support (75), are a prominent source of the T cell survival factor IL-7 (76), promote interactions between dendritic cells and T cells (77), and are critical for antiviral immunity (78). As noted, FRCs are targets of multiple viral infections and can adversely affect T cell responses in the virally infected lymph node (20). Mechanisms of FRC-mediated T cell attenuation involve PD-L1, inducible NO synthase, and arguably IDO (59,68,(79)(80)(81)(82). Notably, experimental neutralization of PD-L1 ameliorates the neurovirulence of HSV-1 and abrogates HSV-1-specific CD8 + T cell responses (83). Consistent with this, anti-PD-L1 Ab treatment contributes to immunopathology and disruption of FRC networks in secondary lymphoid organs during LCMV infection (68).
In this study, we show that adoptive transfer of HSV-1-specific CD8 + T cells into CD118 2/2 mice can reduce viral burden and partially restore endogenous T cell responses (Fig. 4). Although the mechanism driving this phenomenon remains to be elucidated, we speculate that reducing the number of infected FRCs and/or limiting virus-associated disruption of the lymph node microarchitecture is responsible for restoration and mobilization of the endogenous T cell responses. Titrations of different numbers of adoptively transferred TCR-transgenic T cells are known to modulate the gBprecursor frequency during acute HSV-1 infection (84); however, regulatory mechanisms are thought to control the size of the CD8 + T cell effector population based on Ag availability (85). One caveat to our adoptive transfer studies using 3 3 10 6 gB-specific CD8 + T cells is that it reportedly leads to a 3-fold higher effector population by day 5 p.i. relative to the physiological response observed in immunologically naive WT mice (84). Although the increase in effector cell frequency is much greater than 3-fold in CD118 2/2 mice following adoptive transfer (Fig. 4H), it is important to note that this is in comparison with a substantial pathological loss of endogenous lymphocytes because of virus-associated disruption of the lymph node integrity in CD118 2/2 mice as characterized previously (39). However, IFN-a/b signaling enhances but is not requisite for CD8 + T cell effector responses and viral clearance as demonstrated by reductions in HSV-1 titers in the MLN and TG of CD118 2/2 mice receiving gBT-I.1 or gBT-I.1 3 CD118 2/2 T cells. Although the contributions of STING or IFN-a/b to FRC physiology have not been explored in depth, the effects of STING and CD118 2/2 on T cell responses are of considerable interest. Rational vaccine design for HSV-1/2 has primarily yet unsuccessfully sought the development of a strong T cell response to confer protection (6,86). Furthermore, cyclic dinucleotide STING agonists and STING ligand-inducing adjuvants have been used to improve the efficacy of vaccine-elicited Th1 responses to viral, bacterial, and tumor Ags (87)(88)(89)(90). Furthermore, various models have produced mixed evidence for STING and IFN-a/b-dependent T cell responses. CD8 + T cell responses are reportedly muted in STING 2/2 mice following DNA plasmid vaccination (8) and in response to solid tumor challenge (57) or baculovirus infection (60). Consistent with these reports, STING 2/2 macrophages fail to upregulate MHC class II following stimulation (8,12). In contrast, cell-mediated immunity induced upon Listeria monocytogenes infection is enhanced in STING 2/2 mice (91). Moreover, CD8 + T cell responses are largely unaffected in CD118 2/2 and STING 2/2 mice inoculated with recombinant attenuated adenoviral vectors (51). In this study, we show efficient activation and proliferation of HSVspecific CD8 + T cells in the MLN of STING 2/2 mice with defective mobilization during HSV-1 infection. These data correlate with pathologic alteration of the MLN FRC network. Moreover, adoptive transfer of primed but not naive HSV-1-specific CD8 + T cells into STING 2/2 mice leads to a reduction in viral burden (Fig. 8D, 8E), suggesting a defect in appropriate T cell priming in this model. As such, we suggest the lymph node serves as a trap for which T cells exposed to Ag are unable to egress because of disorganization of the FRC network. Collectively, our data definitively highlight the importance of IFN-a/b and STING-mediated immune responses relative to control and susceptibility to acute HSV-1 infection. Although not necessarily unexpected, we show that early responding CD8 + T cells are necessary to repress viral replication within neuronal ganglia. Finally, we posit that future HSV vaccine development should entertain specific targeting of the STING-IFN-a/b pathway to elicit protective immunity because this signaling pathway is essential for the generation of protective countermeasures against HSV-1 in the naive host.