After infection of epithelial surfaces, HSV-1 elicits a multifaceted antiviral response that controls the virus and limits it to latency in sensory ganglia. That response encompasses the CD8+ T cells, whose precise role(s) is still being defined; immune surveillance in the ganglia and control of viral spread to the brain were proposed as the key roles. We tracked the kinetics of the CD8+ T cell response across lymphoid and extralymphoid tissues after ocular infection. HSV-1-specific CD8+ T cells first appeared in the draining (submandibular) lymph node on day 5 and were detectable in both nondraining lymphoid and extralymphoid tissues starting on day 6. However, although lymphoid organs contained both resting (CD43lowCFSEhigh) and virus-specific cells at different stages of proliferation and activation, extralymphoid sites (eye, trigeminal ganglion, and brain) contained only activated cells that underwent more than eight proliferations (CD43highCFSEneg) and promptly secreted IFN-γ upon contact with viral Ags. Regardless of the state of activation, these cells appeared too late to prevent HSV-1 spread, which was seen in the eye (from day 1), trigeminal ganglia (from day 2), and brain (from day 3) well before the onset of a detectable CD8+ T cell response. However, CD8+ T cells were critical in reducing viral replication starting on day 6 and for its abrogation between days 8 and 10; CD8-deficient animals failed to control the virus, exhibited persisting high viral titers in the brain after day 6, and died of viral encephalitis between days 7 and 12. Thus, CD8+ T cells do not control HSV-1 spread from primary to tertiary tissues, but, rather, attack the virus in infected organs and control its replication in situ.
Herpes simplex virus type 1 infection of scarified murine cornea results in retrograde transport of the virus via sensory neurons to the trigeminal ganglion (TG).3 After multiplication in the TG, HSV-1 migrates back to the area of sensory innervation covered by the ganglion and causes typical vesicular rash. HSV-1 often spreads from TG into the brain, also along the sensory nerves (1, 2, 3, 4, 5, 6, 7). The incidence of viral spread to the brain is believed to be enhanced by immunosuppression or stress (8, 9), and HSV-1 encephalitis has emerged as one of the serious complications in immunosuppressed individuals (10, 11). In the course of this infection, HSV-1 is detected by the immune system, resulting in a multifaceted antiviral response that controls the acute infection and limits the virus to latency in ganglia of the peripheral nervous system. In fact, after the initial viral replication and the outbreak of vesicular rash, it is the antiviral T cells that usually limit and control viral replication (6, 12, 13, 14, 15, 16). By contrast, in the absence of T cell immunity, uncontrolled viral replication typically leads to lethal encephalitis (13, 15, 17, 18). Nevertheless, the exact role, basic biology, and mechanisms of antiviral action of various T cell subsets remain incompletely understood, including the kinetics of priming, kinetics and parameters of migration into infected organs, and interplay with the replicating virus.
CD8+ T cells were implicated as potential mediators of acute viral control in the nervous system and of neuropathogenesis (12, 13, 15, 17, 18, 19, 20, 21) and also as sentinels that potentially control viral reactivation from latency in the TG in situ (22, 23, 24, 25). Such evidence was obtained using infection of animals treated with depleting Abs (13, 15, 17) or studies in knockout mice lacking CD8α (26, 27), IFN-γ (26, 28, 29, 30), or molecules that mediate cytotoxicity (27, 31, 32, 33, 34). Direct proof that the missing effector molecules act via CD8 T cells, however, was often not obtained in these studies, and the results of some studies questioned an obligatory role of CD8+ T cells in fighting HSV (22, 35).
Perhaps more importantly, the kinetics of priming, proliferation, and release of activated CD8+ T cells into lymphoid and nonlymphoid organs in relation to the progression of viral replication after ocular infection have not been defined in this model, and such knowledge is critically important to dissect the precise role(s) of CD8+ T cells in controlling the virus during primary infection as well as its inducing potential immunopathogenesis in the CNS (9). To address the above questions, we studied the kinetics and biology of priming and dissemination of antiviral CTL, linking them to the kinetic distribution and the level of viral replication at different sites. We exploited the advantage of techniques allowing precise and definitive assessment of the specificity of antiviral CD8+ T cells.
Materials and Methods
Female C57BL/6-NCr (B6) and B6-Ly5.2/Cr (Ly-5.2) mice were purchased from the National Cancer Institute colony. Female CD8α−/− mice were purchased from The Jackson Laboratory and propagated in the Vaccine and Gene Therapy Institute vivarium. Female B6.gBT-I (mice transgenic for TCRαβ specific for the HSV-1 glycoprotein B498–505 peptide SSIEFARL (gBT-I in the text)) TCR transgenic mice (36), carrying rearranged tcr genes encoding TCR that recognizes the immunodominant HSV-1 epitope gB498–505+H-2Kb, were provided by Dr. F. R. Carbone (University of Melbourne, Melbourne, Australia) and were propagated at the Vaccine and Gene Therapy Institute vivarium by backcrossing to B6 mice. All animals were housed under specific-pathogen free conditions and were used at 6–12 wk of age.
Reagents, Abs, and flow cytofluorometric (FCM) analysis
The gB498–505 peptide (SSIEFARL) was purchased from Research Genetics. The gB498–505:Kb tetramers were prepared in our laboratory according to standard published protocol (37).
The following mAb were used: anti-CD8α (clone 53-6.7), anti-CD45.2 (clone 104), anti-CD43 (clone 1B11), anti-CD62L (clone MEL-14), anti-IFN-γ (all from BD PharMingen), and anti-human granzyme B (GzmB; clone GB12; Caltag Laboratories), the last two for intracellular staining.
For FCM analysis, cells isolated from various organs were dispersed into single-cell suspensions, stained on ice for 20 min with a predetermined optimal concentration of various fluorophore-conjugated mAb, washed, and fixed in 2% paraformaldehyde. For the IFN-γ production assay, 4 × 106 splenocytes were stimulated with 10−6 M gB498–505 peptide for 6 h in RPMI 1640 medium supplemented with 10% FBS and 1.5 μg/ml brefeldin A. After surface staining, the cells were fixed in 2% paraformaldehyde and permeabilized with PermWash (1 mg/ml saponin, 1% FBS, and 0.1% NaN3 in PBS). Intracellular staining for IFN-γ and GzmB was then performed in PermWash on ice, exactly as described above. Analysis was performed on a FACSCalibur instrument using CellQuest 3.3 software (BD Biosciences). At least 104 cells were analyzed per sample, with dead cells excluded by selective gating based on orthogonal and side light scatter characteristics.
Virus and ocular infection
HSV-1 strain 17 (passage 2, uncloned swarm), obtained from Dr. D. J. McGeoch (University of Glasgow, Scotland, U.K.), was grown and titrated on Vero cells, and was used in all experiments. Mice were anesthetized by i.p. injection of ketamine mixture (100 mg/ml ketamine, 20 mg/ml xylazine, and 10 mg/ml acepromazine). The corneas were gently scarified in a crisscross fashion with a sterile 26-gauge needle, and 3 μl of HBSS (Cellgro; Mediatech) containing 5 × 105 PFU of HSV-1 was applied to the surface of each eye.
Adoptive transfer of transgenic T cells
Spleens of 6- to 12-wk-old female gBT-I transgenic mice were enriched for CD8+ T cells by depletion of CD4+ and B220+ cells by MACS (Miltenyi Biotec) according to the manufacturer’s instructions. Magnetic depletion yielded a cell population that was >80% CD8+, and 80–90% of CD8+ cells stained with the gB498–505:Kb tetramer. These cells had a naive phenotype: >92% CD62Lhigh, >95% CD44low, and >98% CD43low. Tetramer+ cells (1 × 106) were adoptively transferred into congenic B6-Ly5.2/Cr female recipients by i.v. injection 24 h before infection. In some experiments the cells were labeled with 2 μM CFSE (Molecular Probes) before adoptive transfer (10-min incubation at 37°C). The transferred cells were distinguished from the recipient cells by FCM detection of CD45.2 (mAb clone 104; BD Pharmingen), Vα2 (BD Pharmingen), and/or fluorescein introduced by CFSE labeling.
Isolation of replicating virus and titer determination
At various times postinfection, the eyeballs, TG, brains, spleens, lungs, livers, and draining lymph nodes were collected and placed in DMEM (Invitrogen Life Technologies) supplemented with antibiotics on ice. The organs were then homogenized using a sterile glass homogenizer. The homogenate was centrifuged (800 × g, 4°C, 10 min), and the supernatant was frozen at −80°C and subsequently used to determine the titer of replicating virus in each organ in a standard plaque assay on confluent Vero cell monolayers. Briefly, serial 10-fold dilutions of the virus-containing supernatant were made in serum-free DMEM and added to confluent layers of Vero cells in six-well plates (1 ml/well). After 1-h incubation at 37°C (infection), the wells were overlaid with 2 ml of 1.5% methylcellulose in DMEM supplemented with 1% FBS and incubated for 2 days at 37°C. The methylcellulose was then washed off with PBS, and the cell monolayers were fixed with 2% formaldehyde and stained with 0.1% crystal violet/PBS solution. The plaques were counted, and the results were expressed as total PFU per organ. The limit of detection was 10 PFU/organ.
Isolation of lymphocytes from nonlymphoid organs
Mice were killed and perfused with PBS (Cellgro; Mediatech), and their eyeballs, TG, and brains were excised. The eyeballs and TG were incubated in RPMI 1640 (Invitrogen Life Technologies) containing 3 mg/ml collagenase type 1 (Sigma-Aldrich) for 1–2 h, dispersed into single-cell suspension by pipetting, and filtered through a 40-μm pore size filter before use in assays. The brains were first dispersed into single-cell suspensions by grinding over a metal screen. The cell suspensions were then layered over a 40/60% Percoll gradient (Amersham Biosciences) and centrifuged at 1200 × g for 30 min at room temperature, and the lymphocytes were isolated from the 40/60% interphase.
Initiation of HSV-1-specific CD8+ T cell response
To follow the early events in the development of the HSV-1-specific CD8+ T cell response, including the proliferation and expression of activation-associated Ags, which are difficult to measure in the endogenous population due to low precursor frequency, we used an adoptive transfer approach. The gBT-I TCR transgenic CD8+ T cells (38), specific for the immunodominant HSV-1 glycoprotein B epitope, gB498–505 (39, 40, 41), overwhelmingly of the naive phenotype (>92% CD62Lhigh, >95% CD44low, and >98% CD43low) were labeled with CFSE and transferred into syngeneic C57BL/6 (B6) mice. After ocular HSV-1 infection, their proliferation, frequency, and surface phenotype were analyzed in lymphoid (spleen, submandibular lymph node, and mediastinal lymph node) and relevant nonlymphoid organs (including eye, TG, and brain) on days 1–12 postinfection.
Clearly detectable proliferation of virus-specific T cells was seen in all animals on day 5 in submandibular lymph node (Fig. 1,A), albeit with some individual variation (15–40% of all gBT-I CD8+ T cells were found to proliferate at that time; Fig. 1 B). On that day, no other organs contained proliferating gBT-I CD8+ T cells, indicating that this lymph node probably served as the draining lymph node for the infected cornea.
Dissemination of activated HSV-1-specific CD8+ T cells into lymphoid and nonlymphoid tissue
The frequency of gBT-I CD8+ T cells increased from the background level on day 6 in both lymphoid and nonlymphoid organs 1 day after proliferation was first detected in the draining lymph node (Fig. 2). Within the limits of our sampling schedule (daily intervals up to day 6, every other day thereafter), the increase in frequency appeared to occur simultaneously in secondary lymphoid organs (spleen and nondraining mediastinal lymph node) and the extralymphoid organs (eye, TG, and brain). Overall, the activation, expansion, and dissemination of virus-specific CD8+ T cells took place in three stages: 1) up to day 5 postinfection, initial proliferation of small number of precursors in the draining lymph node; 2) on day 6, initial dissemination of antiviral CD8+ T cells into both lymphoid and nonlymphoid organs, correlating with an increase in frequency of virus-specific CD8+ T cells in those organs; and 3) days 8–12, further increase in frequency of virus-specific CD8 T cells in lymphoid organs and infected peripheral organs, and maintenance of steady, low frequency of these cells in the lymph nodes. Consistent with that pattern, on day 6 the relative frequency of gB498–505-specific CD8+ T cells was highest in the draining lymph node, but at subsequent time points this was no longer the case, because frequencies rose rapidly in other lymphoid and nonlymphoid organs.
To test whether the above kinetics mirror the kinetics of activation of endogenous CD8+ T cells, a parallel experiment was performed in which the kinetics of the endogenous anti-HSV-1 CD8+ T cell response were determined in B6 mice by gB498–505-Kb tetramer staining (not shown). The endogenous CD8+ T cells followed the same kinetic pattern of dissemination into peripheral lymphoid and nonlymphoid organs as the transgenic cells, but the detection of these cells was delayed by 24–36 h, as expected, due to lower precursor numbers of the latter. As in the transfer experiment, the frequency of virus-specific CD8+ T cells continued to increase in the peripheral organs before reaching a peak on days 8–12, although the frequency remained stable and low in the draining lymph nodes.
Incompletely differentiated CTL are excluded from nonlymphoid organs
We next determined the extent of proliferation (CFSE dilution), activation, and functional status of the gBT-I CD8+ T cells across the tested organs. Differences were observed in the proliferation status and phenotype of gBT-I cells isolated from lymphoid and nonlymphoid organs on days 6–12 postinfection. In the lymphoid organs, the gBT-I CD8+ T cells were CFSEhigh (have not started to proliferate), CFSEint (undergoing their first through seventh divisions), or CFSEneg (have divided more than seven or eight times; Fig. 3,A). Of interest, resting cells and cells in the early rounds of proliferation could be detected in the draining lymph node as well as in the spleen and nondraining lymph node until day 12 postinfection. By contrast, the nonlymphoid organs harbored only CFSEneg gBT-I CD8+ T cells at all time points tested (Fig. 3 B). Note that the draining lymph node had more gBT-I CD8+ cells in the early rounds of division (starting with 39% on day 6 and declining to 25% by day 12) than spleen (16% on day 6 declining to 10% on day 12).
These results were confirmed and extended by correlating cell proliferation to the expression of surface CD43, an activation and adhesion marker (isoform detected by mAb clone 1B11) known to be elevated on activated CD8+ T cells (42). We analyzed not only lymphoid and infected nonlymphoid tissues, but also blood as an intermediate compartment between the two. This analysis revealed that CD43 expression increased over consecutive rounds of proliferation. More than 95% of all CFSEhigh gBT-I CD8 T cells were CD43low, whereas both CFSEint and CFSEneg cells up-regulated the expression of CD43 (Fig. 4,A). Fewer CFSEint cells up-regulated CD43 on days 6 and 12 compared with CFSEneg cells, but both cell populations had similarly high expression of CD43 on day 8 (Fig. 4, B and C). This together with the finding that the nonlymphoid organs contained only CFSEneg gBT-I cells indicate that CD8+ T cells proliferate and become activated in lymphoid organs, and only the fully differentiated CTL (CTL that not only up-regulated CD43, but also went through extensive proliferation) are recruited to nonlymphoid organs. Indeed, the phenotype of gBT-I CD8+ T cells isolated from TG and brain was consistent with this idea (Fig. 5); these cells were CD43high and largely capable of IFN-γ production upon short stimulation. Of interest, the percentages of IFN-γ+ gBT-I cells were similar in the spleen, TG, and brain, suggesting that peripheral localization was not essential for the acquisition of this capacity. The situation was less clear with regard to the expression of GzmB; there, many fewer splenic gBT-I cells expressed this molecule compared with those in TG and brain. At present we are investigating whether this reflects selective recruitment of GzmB+ cells to the neural tissues or the existence of an in situ maturation step. Regardless of this issue, the finding that the nonlymphoid organs contained mostly, if not exclusively, CFSEneg gBT-I cells indicates that CD8+ T cells are activated and become functional in the lymphoid organs, but that only cells that underwent extensive proliferation (and, at least in part, arming) in the lymphoid organs were allowed access to peripheral nonlymphoid tissues. This is consistent with findings of Masopust et al. (43) on the expression of CD43 in lung-residing CTL.
Presence of antiviral CD8+ T cells in the infected peripheral organs correlates with increased survival and lower viral titers in nervous tissues
We next examined how the kinetics of the antiviral response correlated with the kinetics of viral spread and replication. First, we determined the kinetics of viral spread by measuring titers of replicating virus in the eye, TG, and brain on days 1–12 postinfection (Fig. 6,A). Replicating virus was initially detected only in the eye (day 1); however, it spread rapidly to the nervous tissues, reaching the TG on day 2 and the brain on day 3. We reproducibly detected the virus in the brains of all mice, albeit the titers, on the average, were lower compared with TG in this particular experiment. Between experiments and animals, the variation in viral titers was within 1–2 orders of magnitude (see Fig. 6 C for an example). Viral titers begun to decline on day 5 (eye) and day 6 (TG and brain), and viral replication ceased between days 8 and 10 in all organs tested. In parallel, nonlymphoid organs (spleen, submandibular lymph node, and mediastinal lymph node) were also tested for the presence of replicating virus; no viral replication was detected in these sites at any time. Of note, we also did not detect either the virus or virus-specific CD8+ T cells in the other nonlymphoid organs tested (lung, liver, and fat pad; not shown). Therefore, the progress of infection from the eye was restricted to the nervous tissue and did not cause systemic infection.
When superimposed upon the kinetics of CD8+ T cell response, it is clear that the early spread of the virus occurs much before CD8+ T cells are available to contain it in nonlymphoid organs. If so, one could argue that either CD8+ T cells are not necessary for viral control or that they exercise their effect rather late after dissemination. We thus tested whether CD8+ T cells are responsible for control of viral replication in situ after viral dissemination from the eye into the peripheral nervous system and CNS. If that were true, one would expect that starting on day 6 postinfection, CD8α-deficient mice would maintain high virus titers, perhaps for longer periods of time than the wild-type controls. To test this hypothesis, we infected CD8α knockout mice (CD8α−/−) and control B6 mice with the same dose of HSV-1 and compared their survivals (Fig. 6,B). We detected a significant difference in survival between the two groups of mice: 83% survival in B6 mice (n = 12) and only 14% survival inCD8α−/− animals (n = 14). The difference was statistically significant (p = 0.0011, by two-sided Fisher’s exact test). Mortality of CD8-deficient animals occurred after day 8, at the time when wild-type animals exhibited peak frequencies of activated virus-specific CD8+ T cells in the infected organs (Figs. 2 and 4). This correlated with the persistence of high viral titers in the TG and CNS of CD8α−/− mice at these late time points, in contrast to wild-type mice, in which HSV was no longer detectable (Fig. 6 C). These finding indicated that the virus, rather than CD8+ T cells, is responsible for brain pathology. Moreover, this finding strongly suggests that antiviral CD8+ T cells are critical in controlling the outcome of ocular HSV-1 infection and survival of the host.
The ideas that in the course of HSV infection, CD4+ T cells operate to clear local skin/mucosal infection, and that CD8+ T cells chiefly operate in containing the virus in neural tissues were introduced a long time ago (13, 27). Yet in some studies, the role of CD8+ T cells in containing the primary, acute infection was deduced to be either not essential or not obligatory (22, 35) and was proposed to be mostly at the level of limiting the viral spread into the CNS. The results reported in this study strongly affirm that naive, resting CD8+ T cells cannot prevent viral spread to different organs if activated at the time of viral infection. These cells are activated and proliferate with kinetics incompatible with the idea that they can contain early viral spread into the brain. Of interest, activated HSV-1-specific CD8+ T cells, injected within the first 24 h of infection in a cutaneous, zosteriform model of HSV, reduced or abrogated viral infection, suggesting that these cells could limit viral spread if present at the right time and in sufficient numbers (44). However, in that model (44) in the course of natural progression of primary infection, freshly primed CD8+ T cells were also reported to arise too late to limit initial viral spread. Our results are entirely consistent with that study.
Our study systematically analyzed the development of the anti-HSV CD8 T cell response after corneal infection using definitive and conclusive tools: adoptive transfer of naive, HSV-specific CD8+ T cells. Proliferation of antiviral CD8 T cells was initiated in the submandibular lymph node, which was previously shown to be the draining lymph node in corneal transplantation experiments (45). This result confirms and extends earlier results obtained in the flank scarification and footpad infection models (46, 47) and formally demonstrates the kinetics of CTL priming in the ocular model. The novel aspects of these studies are that 1) activated cells were released shortly after initial activation in the draining lymph node; 2) both draining lymph node and spleen contained a mixture of resting, Ag-specific cells and cells undergoing a range of cycles of proliferation as late as day 12 postinfection; and 3) the cells that infiltrated extralymphoid organs had invariably undergone extensive proliferation, full CD43 up-regulation, and arming for IFN-γ secretion, and CD43 expression on these cells began to recede shortly after termination of viral replication. We also found that in addition to the TG and eye, the brain is a site of a vigorous antiviral CTL response, exhibiting the same infiltration kinetics and functional characteristics as the response in TG.
Analysis of gB-specific CD8+ T cell proliferation revealed that proliferating cells in early rounds of proliferation as well as undivided, resting gB-specific CTL could be found in the draining lymph node as late as day 12 postinfection. Thus, the lymph node remains the site of continuous recruitment and proliferation of naive CD8 T cells well after the initial wave of priming and expansion. Except for day 5, when proliferation was detectable only in the draining lymph node, proliferating cells were also detected in the spleen, indicating that the spleen may also be a site of CD8 T cell proliferation in this model. Whether the spleen contains any viral Ags at this point (disseminated by local APCs from the site of infection, because we could not detect replicating virus in the spleen at any time point), or whether CD8 T cells merely proliferate as a consequence of the initial activation program (48, 49) is currently under investigation.
Coordinated analysis of proliferation and activation marker acquisition demonstrated that antiviral CD8 T cells up-regulated the expression of CD43 over several rounds of proliferation. Undivided CFSEhigh gBT-I cells were uniformly CD43low. Cells in the early rounds of proliferation (CFSEint) begun to up-regulate CD43, but the frequency of CD43high cells was greatest in CFSEneg cells, which had undergone extensive proliferation. This is consistent with the idea of a CD8+ T cell differentiation program in which proliferation, gradual changes in activation phenotype, and differentiation into functional CTL occur across a coordinated continuum. Because only CFSEnegCTL were detected outside the lymphoid organs, this implies that only differentiated CD8+ T cells can be recruited to the sites of infection. Indeed, we show that the extralymphoid CD8+ T cell from both brain and TG were functional, as shown by IFN-γ production after 6-h ex vivo peptide stimulation. This is consistent with the results reported by Khanna et al. (25). In addition, the CTL in the infected organs had elevated GzmB expression (compared with naive controls). GzmB is a component of the lytic granule; its expression is elevated in effector CD8 T cells and correlates with the CD43high and CD62Llow phenotype as well as with ex vivo CTL cytotoxic function. Whether CTL located in nervous tissues are capable of and engaged in target cell lysis is currently under investigation.
Finally, our current results suggest that the lethal outcome of HSV-1 infection can be, and often is, prevented by CD8+ T cells. This result is in agreement with the findings of a recent study by Anglen et al. (9) of HSV-1-mediated stress-induced encephalitis after intranasal infection, showing that a timely presence of gB498–505-specific CD8+ T cells protects against encephalitis. In our experiments using CD8α−/− animals, it is possible that the lack of CD8α affected the otherwise CD8α+ Ag-presenting dendritic cells (DCs), which, in turn, may have affected development of arms of the immune response other than CD8+ T cells. Indeed, in the HSV-1 epidermal infection model (50, 51), the CD8α+ DC subset is critical to priming CTL responses. However, it is unlikely that this defect abolished priming of CD4 T cells, which seem to rely upon CD11b+ DCs, but not CD8α+ DCs (52, 53). Therefore, our experiments affirm that in addition to the proposed role in controlling and surveying viral latency in sensory ganglia (25), the CD8+ T cell response to HSV in the brain is one of the mainstays of antiviral defense during acute infection.
The authors have no financial conflict of interest.
We thank Dr. D. J. McGeoch (University of Glasgow, Glasgow, Scotland) for supplying the virus; Dr. F. R. Carbone (University of Melbourne, Melbourne, Australia) for generously providing the gBT-I transgenic mice and for sharing unpublished results; Drs. J. Rosenbaum and S. Planck (Oregon Health & Science University’s Casey Eye Institute) for helpful discussions and encouragement; D. Nikolich-Žugich for flow cytometry, Drs. A. Hill, M. Slifka, S. Wong, and I. Messaoudi and J. Brien for careful reading of the manuscript; and the members of the Nikolich laboratory for support and discussion.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by U.S. Public Health Service Grant CA86803 (to J.N.-Ž.) and Core National Primate Center Support Grant P51RR00163 from the National Institutes of Health.
Abbreviations used in this paper: TG, trigeminal ganglion; DC, dendritic cell; FCM, flow cytofluorometric; gBT-I, mice transgenic for TCRαβ specific for the HSV-1 glycoprotein B498–505 peptide SSIEFARL; GzmB, granzyme B.