Effects of CXCR3 Signaling on Development of Fatal Encephalitis and Corneal and Periocular Skin Disease in HSV-Infected Mice Are Mouse-Strain Dependent

Abstract

purpose. The host inflammatory response to ocular infection with herpes simplex virus (HSV) can be either protective, with disease-free survival, or it can promote diseases such as HSV corneal disease (or herpes stromal keratitis [HSK] in humans) and encephalitis (HSE), depending on mouse strain. The role of CXCR3 chemokine signaling in HSV-induced central nervous system (CNS) inflammation and corneal disease was evaluated, and responses in genetically susceptible and resistant strains of mice were contrasted.

methods. Resistant C57BL/6J (B6) and susceptible 129S6 (129) mice were given monoclonal antibodies (mAbs) to neutralize the CXCR3 ligands monokine induced by interferon-γ (MIG, CXCL9) and interferon inducible protein-10 (IP-10, CXCL10) during HSV infection. In addition, the development of HSV disease was monitored in CXCR3-null mutant mice derived from resistant (B6) and susceptible (BALB/c) strains. Inflammatory cells infiltrating the cornea and brain stem were isolated and stained for flow cytometric analysis.

results. MIG and IP-10 were induced in nervous system tissue after HSV inoculation by the corneal route. HSV-infected 129 mice treated with MIG- or IP-10-neutralizing mAbs showed significantly enhanced survival compared with mice treated with control isotype antibody, whereas survival of the B6 mice was unaltered. Similarly, greater survival was observed for BALB.CXCR3^−/− mice compared with control BALB/c mice. Reduced CNS inflammation was documented that extended to the cornea, such that HSV corneal disease severity was reduced in susceptible BALB.CXCR3^−/−. In contrast, although survival of B6 and B6.CXCR3^−/− mice was indistinguishable, B6.CXCR3^−/− mice developed more severe corneal and periocular skin disease.

conclusions. The effects of CXCR3 signaling in HSV infection are strongly dependent on mouse strain.

Leukocyte trafficking through the central nervous system (CNS) and cornea is normally restricted, but during infection or inflammatory diseases, a large number of monocytes and lymphocytes infiltrate and accumulate in the brain. ¹ Although, the precise mechanisms regulating trafficking of distinct leukocyte populations to sites of viral replication in different tissues are yet to be elucidated, there is little doubt that inflammatory chemokines and chemokine receptors play a critical role. ^{2 3} Activated T cells, macrophages, natural killer (NK) cells and immature dendritic cells (DCs) that are functional for antigen capture rather than presentation express inflammatory chemokine receptors including CCR1, CCR2, CCR5, and CXCR3. ³ Activated Th1 T cells commonly induced by viral inflammatory responses express CCR2, CCR5, and CXCR3. Activated effector cells found in the brain during multiple sclerosis (MS) and experimental autoimmune encephalitis (EAE) express CXCR3 and CCR5, which indicates an important role for these receptors in neuroinflammatory responses. ^{1 4 5}  

In susceptible mouse strains, corneal HSV infection results in induction of potent inflammatory responses that cause bystander damage to the corneal stroma. This immunopathologic disease known as a herpes stromal keratitis (HSK) in humans, involves primarily activated CD4⁺ T cells infiltrating the cornea usually after HSV has been cleared from the eye at approximately day 7 postinfection (PI). ^{6 7} HSV corneal disease in the mouse is a reasonable approximation of the human condition that also involves T-cell and neutrophil infiltration. ⁸ HSV-infected mice show rapid upregulation and sustained expression of several chemokines in the cornea, trigeminal ganglion, and brain, including macrophage inflammatory protein (MIP)-1β, MIP-2, monocyte chemotactic protein (MCP)-1, regulated on activation normal T expressed (RANTES, CCL5), monokine induced by interferon-γ (MIG, CXCL9), and interferon inducible protein-10 (IP-10, CXCL10) ^{9 10 11 12 13 14} (for a review, see Ref. ¹⁵ ). The IFN-γ regulated CXCR3 ligands MIG and IP-10 are potent chemoattractants for activated T cells but the effect of CXCR3 signaling on chemotaxis of other CXCR3⁺ cells is less well defined. Expression of IP-10 and MIG in inflamed tissues is important for defense against various viruses, including MCVM, adenovirus, mouse hepatitis virus (MHV), hepatitis B virus (HBV), and West Nile virus (WNV). ^{16 17 18}  

Primary HSV infection of mucosal tissues is detected by specialized pattern recognition receptors (PRRs) on cells of the innate immune system that recognize virion components or virus-specific structures on the surface or within subcellular compartments of infected cells. ¹⁹ Signaling through such receptors—for example, Toll-like receptor protein family member 2 (TLR2) and -9—results in activation of innate inflammatory responses that are crucial for limiting HSV replication and spread to and within the CNS during acute infection. ^{20 21 22 23} We reported that HSV DNA is highly immunostimulatory, inducing potent proinflammatory responses including production of IFN-γ via TLR 9 activation, ²⁴ and Lund et al. ²⁰ reported recently that HSV-2 DNA triggers plasmacytoid dendritic cells (pDC) to secrete IFN-α via TLR9 activation. Based on observations that the IFN-γ-inducible chemokines MIG and IP-10 can be directly or indirectly induced by pathogen TLR activation, ^{15 25} we speculate that they may be involved in defense against HSV infection. 

In the present study, we examined the role of MIG and IP-10 signaling via CXCR3 in genetically resistant B6 mice or susceptible 129 and BALB/c mice infected on the cornea with HSV. The effect of CXCR3 signaling on the outcome of HSV infection was highly dependent on mouse genetic background. Counterintuitively, CXCR3 signaling facilitated development of fatal herpes simplex encephalitis (HSE) in genetically susceptible 129 and BALB/c strains. Though CXCR3 deficiency had no effect on survival of B6 mice, it resulted in exacerbated corneal eye disease and periocular skin disease. In contrast, BALB/c mice lacking CXCR3 developed less severe corneal and skin disease than did normal BALB/c mice. Thus, CXCR3 signaling contributes to immune pathology that results in severe disease in susceptible mouse strains, whereas in resistant B6 mice, it is associated with regulated responses that effectively curtail the infection without bystander immune pathology. This conclusion is fully consistent with results obtained in a separate study in which we showed that fatal encephalitis in genetically susceptible 129 mice is due to excessive inflammatory responses during acute HSV infection (Lundberg et al., unpublished observations, 2007). 

Materials and Methods

Mouse Strains

Wild-type C57BL/6J and BALB/cJ mice were obtained from Jackson Laboratories (Bar Harbor, ME), and wild-type 129S6 mice were from Taconic Farms (Germantown, NY). B6.CXCR3^−/− and BALB.CXCR3^−/− mice originally derived by Lu and Gerard ²⁶ were obtained from Sally Sarawar (Torrey Pines Institute for Molecular Studies, San Diego, CA) and Thomas Lane (University of California, Irvine), respectively. Mice were bred and housed under SPF conditions in the City of Hope vivarium. 

HSV-1 Inoculation of Mice

Master stocks of HSV strain 17+ or strain F comprising only cell-released virus were prepared in and titered on mycoplasma free CV-1 cell monolayers; single-use aliquots of 4 × 10⁷ PFU/mL virus in Hanks' balanced salt solution (HBSS) supplemented with 2% low endotoxin FBS (Omega Scientific, Tarzana, CA) were stored at −80°C. Unless otherwise stated, male mice at 6 to 8 weeks of age were inoculated with 3200 PFU HSV 17+ (diluted from frozen stocks into HBSS), which corresponds to 10× LD₅₀ for 129 mice. By comparison, 10× C57BL/6 mice exceeds 10⁶ PFU and this is the basis for characterizing 129 mice as susceptible and B6 mice as resistant. The right cornea of mice deeply anesthetized by intraperitoneal (IP) injection of ketamine and xylazine was gently scarified with a 27-gauge needle, with 10 vertical strokes, followed by application of HSV in a volume of 4 μL of HBSS, followed by another 10 horizontal strokes and gentle massaging of the eye with the eyelid to promote virus uptake. ²⁷ The same virus master stock was used for all experiments reported herein. Infected mice were observed daily for symptoms signifying development of HSE and euthanatized as necessary. All animal procedures were reviewed and approved by the City of Hope Institutional Animal Care and Use Committee (IACUC) and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

cDNA Filter Array Analysis

Total RNA isolated from pooled right trigeminal ganglia (Tg) from groups of 10 strain 129 mice mock infected or infected with 10⁶ PFU HSV, F strain, by corneal scarification on day 4 PI was used to prepare ³²P-labeled cDNA probes by reverse transcription. The cDNA probes were hybridized to cDNA filter arrays (Atlas; BD-Clontech, Mountain View, CA) that were washed according to the manufacturer's recommendations before images were captured on a phosphorescence imager screen. Signals on the filters were normalized to 12 housekeeping genes and the net ratio of HSV:mock signal was calculated. A complete list of the 588 genes on a mouse cDNA expression array (Atlas) is available from the BD-Clontech Web site (http://www.clontech.com/support/tools.asp?product_tool_id=157578&tool_id=157579). 

Semiquantitative RT-PCR

Left and right Tgs from three 129 mice bilaterally inoculated on the cornea with HSV were collected and pooled (six Tgs) at day 0, 2, 4, and 8 PI. Total RNA was isolated (TRIzol reagent; Invitrogen, Carlsbad, CA), according to the manufacturer's instructions, to prepare cDNA. Equivalent amounts of cDNA, normalized through PCR amplification for GADPH transcripts, were used for amplification of MIG and IP-10. PCR primers used were designed by us (Oligo 5.0; Molecular Biology Insights, Cascade, CO) and were muMIG, 5′-CCTTTTGGGCATCATCTTCCTGGAG-3′ (forward), 5′-GACGACGACTTTGGGGTGTTTTG-3′ (reverse); muIP-10, 5′-TCTGCCTCATCCTGCTGGGTCTGA-3′ (forward), 5′-TCCCTATGACCCTCATTCTCACTGG-3′ (reverse); and muGAPDH, 5′-GGTGAGGCCGGTGCTGAGTATGTC-3′ (forward), 5′-CATGAGCCCTTCCACAATGCCAAAG-3′ (reverse). Amplification products were analyzed by agarose gel electrophoresis and visualized by ethidium bromide staining. 

Quantitative SYBR Green RT-PCR

Total RNA was isolated from cells recovered from spleen or peritoneal lavage isolated (TRIzol; Invitrogen) reagent, and cDNA was produced using MMLV-RT according to the manufacturer's instructions. One microliter of a 20-μL RT reaction was then used for SYBR green RT-PCR with a 2× SYBR green master mix and gene-specific primers used according to the manufacturer's instructions (Bio-Rad, Hercules, CA) and amplified with a generic two-step (94°C/60°C) cycle program. Relative amplification (ΔC_t) was calculated by subtracting the replicate C_t average of GAPDH amplification from the corresponding C_t average of the gene of interest. The relative difference between strains (x-fold) was then calculated by subtracting the lower ΔC_t from the higher ΔC_t. PCR primers used were as described in Liu et al. ²⁸ or were designed by us (muGAPDH; Oligo 5.0; Molecular Biology Insights): muTLR2, 5′-CAGCTTAAAGGGCGGGTCAGAG-3′ (forward), 5′-TGGAGACGCCAGCTCTGGCTCA-3′ (reverse); muTLR9, 5′-CCAGACGCTCTTCGAGAACC-3′ (forward), 5′-GTTATAGAAGTGGCGGTTGT-3′ (reverse); and muGAPDH, as for semiquantitative RT-PCR. Correct amplification products were assessed by melting curve analysis and visually by agarose gel electrophoresis for all data used in the study. 

Histology

Corneas and skin biopsy samples isolated from infected mice were embedded in optimal cutting temperature (OCT) compound, sectioned at 6 and 10 μm, respectively, and hematoxylin and eosin (H&E) stained before being examined microscopically for infiltrating cells. 

In Vivo Antibody Depletion

HSV-infected mice were given 500 μL rabbit anti-mouse MIG (1 mg/mL total protein) or 200 μg affinity-purified rat anti-mouse IP-10 (hybridoma kindly provided by Thomas Lane) antibody intraperitoneally (IP) on days 0, 2, 4, 6, 8, and 10 PI and monitored. Mice displaying overt symptoms of encephalitis were euthanatized. Control mice received normal rabbit serum or whole rat IgG, respectively. 

Clinical Corneal and Skin Disease Scores

A blinded observer examined mice for periocular skin disease and corneal eye disease by using a hand-held slit lamp. HSK was graded on the following scale: 0, normal/disease free; 1, pupillary dilation and asymmetry; 2, corneal clouding; 3, corneal opaqueness; 4, opaqueness with ulcer (but not complete destruction of eyeball); and 5, eyeball collapsed and destroyed. Corneal perforations or neovascularization added an additional 0.5 score, so that an opaque cornea with neovascularization received a score of 3.5. Periocular skin disease was graded on the following scale: 0, normal/disease free; 1, only eyelid affected/blepharitis; 2, hair loss <3 mm from eye; 3, hair loss up to 50% of the side of the face; 4, hair loss on 50% to 75% of the side of the face; and 5, hair loss on greater than 75% of the side of the face. In addition, skin breakdown indicated by bleeding and/or scabbing or swelling of the eye lid to the point of closure added an additional 0.5 score, so that 60% hair loss with skin breakdown was given a score of 4.5. HSV-induced uveitis or retinitis was not monitored. 

Isolation and Staining of Mononuclear Cells Infiltrating the Cornea and Brain Stem

We adapted the method described by Ford et al. ²⁹ Briefly, two to three pooled brain stems (BSs) were minced and digested with collagenase (grade D; Roche Biochemicals, Nutley, NJ) and DNase I (grade II; Roche), after which the cell suspension was centrifuged through a two-step gradient (Percoll; Roche). The resultant enriched population of viable mononuclear cells included lymphocytes as well as microglia identified as CD45^intCD11b⁺ and CNS macrophages characterized as CD45^hiCD11b⁺. The latter two populations are morphologically and functionally distinct. ^{29 30} Mononuclear cell yields from a normal brain range from 0.8 × 10⁵ to 2 × 10⁵, with no T cells present, ³⁰ but higher cell yields were found for inflamed Tg and BSs of HSV-inoculated mice. Cell viability was usually greater than 95%; control digestions of spleen and draining lymph node (dLN) cells with collagenase indicated that the enzyme had no effect on the expression of cell surface markers or functionality of CD4⁺ and CD8⁺ T cells. ⁷ Corneas were processed directly by digestion with collagenase, and all cell preparations were washed twice with PBS supplemented with 2% calf serum, 2 mM EDTA, and 0.05% sodium azide before staining for flow cytometric analysis. Because T cells, neutrophils, and other monocytes have been implicated in HSE and corneal disease development, we used CD4, Gr-1, and CD11b as cell surface markers to identify these cell populations, respectively. 

Antibodies Used for Flow Cytometry Analysis

Unlabeled rat-anti-mouse CD16/32 (clone 93; eBioscience, San Diego, CA) was used to block nonspecific mAb binding to murine Fc receptor. Antibodies were titrated for optimal performance and purchased from commercial sources: APC- or biotin-conjugated Gr-1 (clone RB6-8C5; eBioscience); anti-MHCII (clone M5; eBioscience); FITC- and PerCP-conjugated CD45-PerCP (clone 30-F11; BD PharMingen, San Jose, CA); PE- or biotin-conjugated CD4 (clone GK1.5; BD PharMingen); PE-, FITC-, or APC-conjugated CD11b (clone M1/70; eBioscience); polyclonal goat-anti-mouse CXCR3 (clone Y-16; Santa Cruz Biotechnology, Santa Cruz, CA). Streptavidin conjugates purchased were: Cy2 (Jackson ImmunoResearch Laboratories, West Grove, PA), PE (eBioscience), and APC and PerCP (BD-PharMingen). For CXCR3 staining, RPE-F(ab′)2-donkey-anti-goat IgG (H+L) and cells were additionally blocked with normal donkey serum (Jackson ImmunoResearch Laboratories) before secondary antibody incubation. 

Results

Induction of MIG and IP-10 Expression in the Nervous System of HSV-Infected Mice

To determine what chemokines and cytokines are produced in the nervous system, 129 mice inoculated with HSV were killed on day 4 PI, and total trigeminal ganglionic RNA was used to probe gene filter arrays (Atlas; BD-Clontech), as detailed in the Materials and Methods section. Figure 1A , left, shows the overall data quality, in that the filters have reproducible patterns and lack background spots. The enlarged spots on Figure 1A , right, show more than 100-fold upregulation of the CXCR3 ligand MIG, a 12.5-fold upregulation of the interferon-γ signaling chemokine STAT1, an eightfold upregulation of IRF-1 and marginal upregulation of MIP-1β and GADDip. The CXCR3 ligand IP-10 was not included in the cDNA filter arrays. To determine whether IP-10 is upregulated similarly to MIG, a semiquantitative RT-PCR assay was used. Neither IP-10 nor MIG was detected in Tg of 129 mice before HSV inoculation, but IP-10 was strongly and MIG weakly induced 2 days PI, and both ligands were strongly induced at days 4 and 8 PI (Fig. 1B) . IFN-γ was not included on the expression array, and therefore its expression was not monitored in these experiments. 

Effect of Abrogated CXCR3 Signaling on HSV Mortality and Cells Infiltrating the Brain Stem

B6 mice are known to be resistant to HSV infection. ²⁷ Because activated T cells express CXCR3, in vivo neutralization of MIG or IP-10 was expected to cause mortality in HSV-infected B6 mice. However, survival was indistinguishable from that in B6 mice treated with isotype control antibody (data not shown). In contrast, Figure 1Cindicates a significant protective effect on mortality in MIG- and IP-10-depleted 129 mice. Compared with 10% survival in 129 control mice, survival was ∼50% in depleted mice (P = 0.009 for anti-MIG serum, P = 0.035 for anti-IP-10 mAb). A second trial of anti-MIG antibody (Figs. 1D 1E) , in which the HSV inoculum was varied, showed increased protection at 10⁴ and 10³ PFU (70%–80% survival compared with 25%–40% in mice receiving normal rabbit serum [NRS]), 100% survival compared to 90% in mice receiving NRS at 10² PFU, and 100% survival in both groups of mice inoculated with 10¹ PFU HSV. 

To determine whether genetic ablation of CXCR3 receptor repeated these results, B6.CXCR3^−/− mice were inoculated with HSV and monitored for survival. As shown, in Figure 1F , there was no increase in mortality in B6.CXCR3^−/− mice 3 weeks PI. Fourteen of 15 wild-type B6 mice survived and 9 of 10 B6.CXCR3^−/− mice survived, with the two deaths occurring at 10 and 12 days PI. With a 30-fold increase in the inoculum to 10⁵ PFU, mortality remained at ∼10% in both wild-type and B6.CXCR3^−/− mice (not shown). BALB/c mice are known to be susceptible to HSV infection. ^{27 31} In contrast to B6 mice, only 1 of 15 BALB/c mice survived corneal inoculation of 3.2 × 10³ PFU of HSV, and the deaths occurred at day 6 to 7 PI, earlier than B6 mice. As shown in Figure 1F , mortality was delayed in BALB.CXCR3^−/− mice, and survival was significantly greater than wild type: 8 of 30 BALB.CXCR3^−/− mice survived (P < 0.0001). This counterintuitive result suggests a detrimental effect of CXCR3 signaling on HSV encephalitis in susceptible mice only. 

Mortality from HSV corneal infection in the mouse model results from nervous system infection. To compare different mouse strains and the effect of CXCR3 signaling, infiltrating cells were isolated from the BS 7 days PI and stained for flow cytometric analysis. Figures 2A and 2Bshow results in BALB/c and B6 mice with respective histograms (Figs. 2C 2D) , giving the absolute number of cells and the proportion of activated cells as indicated by major histocompatibility complex (MHC) II expression. In general, resistant B6 mice had a lower absolute number but a higher proportion of activated infiltrating cells. There were fewer macrophages (CD45^hi CD11b^hi/^int) in B6 than in BALB/c BS (27-fold less) and also fewer T cells and macrophages (4.6–13-fold) and microglia (1.7-fold) recruited to B6 BSs. Infiltrating macrophages were the predominant MHC II⁺ population in B6 mice and though microglia (CD45^int CD11b^int) were less activated (lower MHCII⁺ expression) in B6 mice, they were more numerous than in BALB/c mice (Figs. 2C 2D) . Loss of CXCR3 signaling resulted in a marked reduction of cellular infiltration, with greater reduction in B6.CXCR3^−/− than in BALB.CXCR3^−/− mice. The BALB.CXCR3^−/− mice showed a two- to fourfold reduction of all infiltrating cells, demonstrating that the aggressive inflammatory response in the BS of susceptible BALB/c mice correlates with the observed mortality shown in Figure 1F . 

Expression of CXCR3 on CD11b⁺ splenocytes from B6 mice was determined since a simultaneous reduction in monocyte infiltration and CXCR3 expression has been observed in the cornea after anti-IP-10 antibody depletion. ³² CXCR3 was constitutively expressed (mean fluorescence intensity [MFI] of 100–400; shaded peak) and further inducible by Con A supernatant (shift to MFI, 50–70) on CD11b⁺ B6 splenocytes (Fig. 2E)compared with isotype control (MFI, 20–30). Furthermore, CXCR3 was absent and not inducible on CD19⁺ B6 splenocytes (Fig. 2F)and was normally absent but inducible on CD4⁺ B6 splenocytes (Fig. 2G) . Thus, constitutive expression of CXCR3 on CD11b⁺ cells suggest that monocytes are likely to respond to the early induction of IP-10 and MIG (Figs. 1A 1B 1C) , whereas prior activation would be necessary to recruit CXCR3⁺ CD4⁺ T cells. 

Effect of Abrogation of CXCR3 Signaling on HSV Clinical Eye Disease and Cells Infiltrating the Eye

To determine whether blocking CXCR3 signaling has a protective effect on eye disease similar to the protection of mortality, mice were inoculated on the right cornea and monitored weekly for clinical disease. Scores of 1 to 5 were assigned for the severity of corneal and periocular skin disease by a grader who was unaware of the mouse group tested. As shown in Figures 3A and 3B , the severity of corneal disease was reduced in BALB.CXCR3^−/− mice compared with wild-type BALB/c mice. This difference was relatively modest in the inoculated eye and only became statistically significant in the third week PI (P < 0.05 on day 21; P < 0.01 on day 35; Fig. 3A ). However, in the left eye, contralateral to HSV inoculation, corneal disease was consistently less severe in CXCR3^−/− mice with median scores <1 from 1 to 4 weeks PI (Fig. 3B) . In contrast, there was no difference in periocular skin disease in CXCR3^−/− compared with wild-type BALB/c mice (not shown). 

HSV corneal disease does not ordinarily develop in virus-inoculated B6 mice, ³³ and so it was unexpected when grossly evident corneal disease was noted in our initial mortality trials with B6.CXCR3^−/− mice. This observation led to studies assessing clinical eye disease in B6.CXCR3^−/− mice, and results are shown in Figures 3C 3D 3E 3F . Median scores for corneal disease in mice inoculated with 3.2 × 10³ PFU HSV was 0 to 1 at day 7 PI, similar to wild type; but, although scores remained unchanged in wild-type mice, CXCR3^−/− corneal disease scores increased to 3 on day 14 and only slightly began to decrease at day 35 PI (Fig. 3C) . With a 30-fold increase in HSV inoculum to 10⁵ PFU, corneal clinical disease developed in the inoculated eyes of wild-type mice (median score 2 at day 14 PI; Fig. 3E ); and although B6.CXCR3^−/− mice attained a higher median score, ³ this difference was no longer statistically significant. 

Grading results of periocular skin disease in B6 mice had a pattern similar to that of corneal disease: 0 to 1 for wild-type B6 mice inoculated with 3.2 × 10³ HSV, whereas the median score was 3 at day 14 PI for B6.CXCR3^−/− mice with scores slightly lower at days 21 and 35 PI (Fig. 3D) . With an increased inoculum of 10⁵ PFU HSV (Fig. 3F) , median scores increased for skin on the inoculated side compared with mice inoculated with 3.2 × 10³ PFU. There was an additional increase in median score to 2 in B6.CXCR3^−/− mice receiving the higher inoculum, with this increase being statistically significant. 

Histopathologic sections of cornea and periocular skin in B6 mice at day 14 PI were consistent with the clinical scoring: minimal inflammatory cells in wild-type mice (Fig. 4A) , modest inflammatory cell infiltration in B6.CXCR3^−/− mice with clinical scores of 0 to 1 (Fig. 4C) , and massive mononuclear and polynuclear cell infiltrates with tissue destruction in B6.CXCR3^−/− mice with scores of 3 to 5 (Fig. 4B) . Moderately affected B6.CXCR3^−/− mice (score, 1–2) had corneal edema but few infiltrating cells (Fig. 4C) . Neutrophils, identified morphologically, predominated infiltrates in the cornea (Fig. 4B , inset, arrows) and skin (Fig. 4E , inset, black arrows) in which a few lymphocytes were seen (Fig. 4E , inset, white arrows); no cellular infiltrate was seen in wild-type skin (Fig. 4D) . To characterize the early cellular infiltrate better, cells were isolated from corneas 7 days after HSV inoculation and stained for flow cytometric analysis. As summarized in Table 1 , the percentage of CXCR3-positive cells was low in wild-type mice (1.5% in B6, 2.8% in BALB/c). Despite the ultimate development of corneal disease in BALB/c and not B6 wild-type mice, the percentage of CD11b⁺, CD4⁺, and Gr-1⁺ cells were similar in the two strains at day 7 PI. The main cellular difference in the CXCR3^−/− mice was the marked increase in the percentage of Gr-1⁺ cells: from 5.1% in B6 wild type to 22.5% (4-fold increase) and 3.1% in BALB/c wild type to 51.8% (17-fold increase). Although CD4⁺ cells were present at a lower percentage than were Gr-1⁺ cells in the CXCR3^−/− mice, the increase compared with the count in wild type was greater (a 24-fold increase in B6.CXCR3^−/− and a 30-fold increase in BALB.CXCR3^−/− mice). 

To determine the effect of CD4⁺ and Gr-1⁺ cells on clinical eye disease, wild-type B6 mice were compared to B6.CD4^−/− mice or wild-type B6 mice treated with anti-Gr-1 mAb. In CD4^−/− mice, the median corneal score of the HSV-inoculated eye was 3 on day 21 PI compared with 0 to 1 in wild-type B6 mice (P = 0.025) and periocular disease was also greater in CD4^−/− mice (P = 0.0185; Fig. 5A 5B ). In contrast, B6 mice depleted of Gr-1⁺ cells did not develop significant corneal disease (Fig. 5C)although they did have higher periocular skin scores in the inoculated eye (Fig. 5D) . 

Discussion

We inferred from early induction of MIG and IP-10 in Tg of HSV-infected mice that interfering with CXCR3 signaling would be deleterious, causing increased mortality due to impaired T-cell responses. ¹⁵ Abolition of CXCR3 signaling by neutralization of MIG and IP-10 or genetic deletion of CXCR3 surprisingly had no effect on survival of resistant B6 mice, whereas survival of susceptible 129 and BALB/c mice was significantly enhanced. These results were not affected by HSV dose, since the same pattern was observed when susceptible and resistant strains were inoculated with much lower or higher doses of virus, respectively. 

Thus, CXCR3 signaling is associated with development of lethal HSV encephalitis in genetically susceptible 129 and BALB/c mice but not resistant B6 mice. This contention is consistent with results from ongoing studies of genetic resistance to HSV in 129 and B6 mice, which showed that fatal encephalitis in susceptible 129 mice results from excessive inflammatory responses causing severe tissue damage in the BS ²⁷ (Lundberg et al., manuscript submitted). Kurt-Jones et al. ²¹ reported analogous results implicating TLR2 signaling in induction of excessive inflammatory responses causing lethal HSV encephalitis in B6.129F2 mice and, as we have reported, ²⁷ mortality did not correlate with elevated HSV titers in the CNS. It is notable that the customary approach of focusing on changes in viral burden in mice treated with neutralizing antibodies to CXCR3 or its ligands MIG and IP-10 failed to reveal beneficial, or indeed (as shown in the present study) deleterious, effects of CXCR3 signaling during HSV infection. ^{32 34}  

Primary human microglia produced significant amounts of MIG, IP-10, and RANTES when infected with HSV, even though they are nonpermissive for replication. ³⁵ Chemokine induction occurred independent of TLR2 signaling, which was essential for induction of other proinflammatory cytokines and chemokines. ²³ Expression of TLR2 and -9 was respectively 8- and 20-fold higher in peritoneal exudate cells (PECs) from B6 compared with BALB/c mice (not shown). However, a role for TLR9 in protection against fatal HSE was not evident from the results of studies of HSV infection in TLR9-null mutants in either strain background. The critical PRR(s) on innate cells whose activation induces expression of CXCR3 ligands during HSV infection thus remains to be determined. The clinical relevance of studies of HSV pathogenesis in the mouse is highlighted by the observation that IP-10 and RANTES are detected at elevated levels in the CSF of patients with HSE. ³⁶  

Contrary to our results, a recent study reported significantly greater survival for B6.CXCR3^−/− compared with control B6 mice inoculated with HSV by the corneal route. ³⁷ A troubling aspect of this study is the uncharacteristic susceptibility of the control B6 mice, only 40% of which survived inoculation with 500 PFU of the HSV McKrae strain compared with 85% for B6.CXCR3^−/− mice. The LD₅₀ for HSV 17⁺ in B6 mice was greater than >10⁵ PFU, and in our hands McKrae was equally or perhaps slightly more virulent. With the same B6.CXCR3^−/− strain, mortality was approximately 10% for both B6 and B6.CXCR3^−/− mice inoculated with 3200 PFU HSV 17⁺ by the corneal route. The mortality rate for the B6.CXCR3^−/− mice was not significantly different in the two studies (both studies used the same B6.CXCR3^−/− strain ²⁶ ), whereas mortality of B6 was significantly different. Wickham et al. ³⁷ obtained B6 mice from The Jackson Laboratories (Bar Harbor, ME) and bred in house. Extensive in-house breeding of the B6 strain (>3–10 generations, depending on the sibling mix in the breeding scheme) compared with the B6.CXCR3^−/− would be compatible with the occurrence of genetic drift that may well have resulted in selection of a B6 substrain with decreased resistance to HSV. Indeed, more than 40 substrains of B6 are now recognized, most of which arose as a result of maintaining subcolonies separated from the originating colony for several generations. Detailed information on inbred mouse strain stability and an extensive bibliography on genetic drift are available from the Jackson Laboratories (http://jaxmice.jax.org/geneticquality/drift.html). The indistinguishable difference in mortality rates of the B6.CXCR3^−/− strain in the two studies suggests that genetic drift is a plausible explanation for the disparate mortality rates observed for the wild-type B6 strains. 

The overall outcome of loss of CXCR3 was a more moderate inflammatory response during acute infection, characterized by reduced cellular infiltration that differed in cellular subset responses in mice of different genetic backgrounds. However, the phenotypic result was more severe corneal and skin disease in resistant mice, contrasted with reduced mortality and skin and eye disease in susceptible mice. B6.CD4-deficient mice showed increased susceptibility to fatal HSE (not shown) and had higher corneal and skin disease scores. This result is consistent with those obtained with mice rendered CD4 deficient by antibody depletion or genetic knockout ³⁸ and the report that BALB.CD4^−/− have decreased corneal disease and reduced neurologic dysfunction. ³⁹ In a recent study, transient HSV corneal disease mediated by CD8⁺ T cells infiltrating BALB.CD4^−/− corneas was observed only after relatively high-dose corneal inoculation with HSV but not with lower doses that routinely induced HSV corneal disease in wild-type BALB/c mice. ⁴⁰ The increase in clinical disease scores noted herein for B6.CD4^−/− mice implies loss of a regulatory population that may be involved in regulating CD8⁺ T cell influx into the cornea as proposed for BALB.CD4^−/−. ⁴⁰  

Depletion of Gr-1⁺ cells in B6 mice had no effect on survival (compared to enhanced resistance observed in susceptible 129 mice; Lundberg et al., manuscript submitted) or corneal disease, and skin disease scores were only modestly increased. B6 mice inoculated on the cornea with HSV are resistant to development of corneal and periocular skin disease. Given this and the role of CXCR3 in trafficking of activated CD4⁺ Th1 T cells that are implicated in chronic HSV corneal disease, ⁴⁰ it was surprising that B6.CXCR3^−/− developed significant and sustained corneal and periocular skin disease. Examination of HSV-infected corneas from B6.CXCR3^−/− mice displaying clinical disease revealed massive cellular infiltration predominated by polymorphonuclear neutrophils (PMNs) with lymphocytes being relatively sparse, which suggests they may have a more limited role in HSV corneal disease in this model. 

Our studies of CXCR3 signaling in HSV-infected resistant and susceptible mouse strains reveal dramatically different phenotypes depending on the mouse strain. In susceptible strains, CXCR3 signaling promotes deleterious inflammatory responses associated with induction of severe disease and fatal HSE. Though CXCR3 signaling is dispensable for survival of HSV-infected resistant mice, this chemokine pathway nonetheless contributes to regulated immune responses that ensure virus elimination without bystander immune disease that results in eye and skin disease. Strain differences in immune response to HSV infection is a recurring theme that extends beyond differences in chemokine signaling and TLR expression. For example, we have shown recently that Gr-1⁺ neutrophils and macrophages contribute to destructive inflammatory responses in 129 mice that lead to HSE mortality (Lundberg et al., unpublished observations, 2007). In contrast, depletion of neutrophils in B6 mice had no effect on mortality, whereas depletion of macrophages in B6 mice increased their susceptibility to fatal HSE, ⁴¹ suggesting that macrophages are protective in the B6 genetic background. These results warrant caution in interpreting results obtained with hybrid 129xB6 strains harboring null mutations in immune responses genes that have been variably backcrossed, since dependent phenotypic effects may confound interpretation of results because obtaining an isogeneic control strain maybe difficult or impossible. ^{42 43 44}  

⁴  Present affiliation: Doheny Eye Institute, Department of Ophthalmology, University of Southern California Keck School of Medicine, Los Angeles, California.

 Supported by Grant EY013814 from the National Eye Institute.

 Submitted for publication March 1, 2007; revised April 11, 2007; accepted June 8, 2007.

 Disclosure: P. Lundberg, None; H. Openshaw, None; M. Wang, None; H.-J. Yang, None; E. Cantin, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

MIG and IP-10 expression in HSV-infected Tg and effects of CXCR3 signaling on development of HSE. cDNA filter analysis of RNA from 129 mouse Tg at day 4 PI. Right: genes showing greater than threefold induction between infected and mock mice (A). Semiquantitative PCR for detection of MIG and IP-10 in infected Tg at various times PI (B). Mortality in 129 mice (squares) after corneal infection was reduced by treatment with anti-serum to MIG (black circles) or antibody to IP-10 (gray circles) (C). Compared with normal rabbit serum treatment (D), the effect of MIG antiserum treatment (E) persisted at HSV-1 doses from 10¹ to 10⁴. Mortality of CXCR3^−/−mutant mice (gray) after corneal HSV infection was compared to their wild-type counterparts (black) for C57BL/6 (squares) and BALB/c (circles) and revealed an effect only in susceptible mice (F). Data combined from five experiments.

Figure 2.

 

Cellular infiltration of BS in wild-type and CXCR3^−/− mice and cell surface expression of CXCR3 in B6 splenocyte subsets. Overall cellular infiltration of BS on day 7 PI is reduced in both BALB.CXCR3^−/− (A) and B6.CXCR3^−/− mice (B). MHC II expression was used as a measure of activation of cell subsets and the histograms in (C) and (D) correspond to the CD45-CD11b gates in (A) and (B) and represent absolute numbers of infiltrating cells with the proportion MHC II+ in the black division of each bar. Representative of two experiments. Histograms showing CXCR3 expression on naive B6 splenocytes that were cultured overnight in medium (solid line, gray histogram) or in the presence of ConA supernatant (solid line, unfilled histogram), gated on CD11b⁺ (E), CD19⁺ (F) or CD4⁺ (G) cells. Isotype control stain (dashed line) was performed on stimulated cells. Results shown are representative of two experiments.

Figure 3.

 

Corneal and skin disease in wild-type C57BL/6 and BALB/c mice and their respective CXCR3^−/− mutants. Wild-type (square) and CXCR3^−/− mutant (circle) mice were infected with 320 (A, B), 3200 (C, D), or 10⁵ (E, F) PFU HSV-1 17+ applied to the right cornea. Infection of BALB/c mice show significantly decreased clinical scores for HSK in mice lacking CXCR3 on the infected (E) and contralateral (F) eye. B6.CXCR3^−/− show significantly increased clinical scores for HSK (A) and skin disease (B) when compared to B6 controls (peak scores at day 14 shown). High-dose infection of B6 mice show slightly increased clinical scores in CXCR3^−/− mice for HSK (C) and skin disease (D) when compared with B6 control mice. Data were combined from seven experiments.

Figure 4.

 

Cellular infiltration in corneas from B6.CXCR3^−/− mice. Hematoxylin-eosin stained 6-μm corneal sections from B6 WT (A) and B6.CXCR3^−/− (B, C) and 10 μm skin sections from B6.CXCR3^−/− (D) and B6 WT (E) from mice on day 14 PI are shown. (A) Absence of cellular infiltrates in B6 WT mice that did not develop HSK (clinical score, 0–1). (B) Massive PMN cellular infiltration apparent in B6.CXCR3^−/− corneas that developed severe HSK (clinical score, 3–5). (C) Modest cellular infiltration in B6.CXCR3^−/− mice that developed mild HSK (clinical score, 1–2). (D) Massive cellular infiltration was apparent in two affected skin sections (clinical score, 3–4) from B6.CXCR3^−/− mice, whereas (E) the B6 WT skin sections lacked such cellular infiltration (clinical score, 0–1). (B, E, insets) Almost exclusively PMN cells (black arrows) although a few cells with lymphocyte morphology (white arrows) were found in the skin sections (E). Magnification: (A–C) ×200; (D, E) ×100; (E) ×960.

Table 1.

 

Cellular Infiltration in HSV-Infected Cornea

Strain	CXCR3+	CD11b+	CD4+	Gr-1+
B6 wt	1.5	4.1	0.1	5.1
B6.CXCR3−/−	NA	ND	2.4	22.5
BALB wt	2.8	3.0	0.2	3.1
BALB.CXCR3−/−	NA	ND	6.1	51.8

 Corneas were collected on day 7 PI, digested, and stained for the cell surface markers CXCR3, CD11b, CD4, and Gr-1. Data are representative of three experiments and are expressed as percentages of total cell count. NA, not applicable; ND, not done.

Figure 5.

 

HSK and skin disease in B6.CD4^−/− and Gr-1-depleted C57BL/6 mice. When compared with wild-type mice (squares), B6 mice lacking CD4 (circles) show significantly increased HSK (A) and skin disease (B) at day 21 PI. Representative of two experiments. HSK did not develop in B6 mice depleted with 200 μg anti-Gr-1 mAb on days 0, 2, 4, 6, and 8 PI in either infected (circles) or contralateral (squares) eyes (C) but skin disease did develop on the infected side (D). Representative of two experiments.

The authors thank Paula V. Welander and Seung-Jae Jung for technical assistance and Chandran Ramakrishna and Massimo D'Apuzzo for helpful discussions.