Virus-specific CD8⁺ T cells accumulate near sensory nerve endings in genital skin during subclinical HSV-2 reactivation

To study the spatiotemporal distribution of HSV-2–specific CD8⁺ T cells during the evolution of HSV-2 reactivation, serial 3-mm punch biopsies from six HSV-2 seropositive, HIV seronegative subjects with recurrent genital herpes were obtained. Patients were selected from a cohort of persons who participated in prior studies and who had defined CD8⁺ T cell responses to HSV-2 (8,14). Patients underwent from two to seven separate biopsies over the course of study (Table I). Qdot multimers recognizing peptide epitopes encoded by HSV-2 UL26, UL27, UL46, UL47, and UL49 genes were synthesized, and in situ staining for HSV-2–specific CD8⁺ T cells was conducted.

During the acute phase of genital recurrence, productive HSV-2 viral replication was associated with disruption of the epidermal layer (Fig. 1 a). Immune cell infiltration was massive at the site of infection and differed markedly from that observed in biopsies taken at the same time from distant skin sites (Fig. 1 a). CD8⁺ T cells were detected both below and within the HSV-2–infected epidermis (Video 1); three-dimensional modeling constructed from confocal images of the late ulcer/early crust stage (Fig. 1 b) revealed cell-to-cell interactions between CD8⁺ T lymphocytes and the HSV-2–infected epidermal keratinocytes. The presence of multiple junctions between CD8⁺ T cells and HSV-2–infected keratinocytes is consistent with the formation of immunological synapses between antigen-specific CD8⁺ T cells and their targets in vivo (Video 2). To demonstrate the functional competence of these CD8⁺ T cells, lymphocytes were recovered from recurrent genital lesions biopsied at the time of HSV-2 culture conversion from positive to negative. The lymphocytes were expanded and examined for cytolytic function by chromium release assays, as previously described (9). CD4⁺ CD8⁻ and CD4⁻ CD8⁺ T lymphocytes and CD3⁻ CD56⁺ NK cells were the three main subpopulations detected after expansion (Fig. 1 c). HLA class I–restricted, HSV-2 antigen–specific cytolytic activity of CD8⁺ T cells was clearly present both at bulk culture and selected T cell levels (Fig. 1 c), indicating that CD8⁺ cytolytic T lymphocytes specific for HSV-2 contribute to the resolution of lesions in patients with genital herpes.

Based on previous work defining the HSV-2–specific epitopes and immunodominant responses restricted by HLA A*0101, A*0201, and B*0702 alleles (8, 24), peptides encoded by the HSV-2 UL26, UL27, UL46, UL47, and UL49 genes were used to synthesize Qdot 655–conjugated multimers (see Materials and methods). Qdot 655 conjugates were constructed to average between seven and eight streptavidin molecules covalently attached to each Qdot particle, thus binding up to 32 peptide–MHC monomers under saturated condition (Fig. 2 a). The photostability of the Qdot multimer allowed high quality confocal laser scanning, including detailed imaging of TCR molecules on the cell surface (Fig. 2 b and Video 3). The Qdot 655 multimers exhibited similar specificity but had higher signal intensity and better sensitivity than PE- (Fig. 2 c) or allophycocyanin-labeled (not depicted) tetramers bearing the same peptide.

We first quantitated the number of total CD8⁺ cells in the lesional area during the acute stage of the herpetic lesion. Infiltrating CD8⁺ cells averaged 1,504 ± 337 cells/mm² in a 10-μm thick tissue (n = 6). Genital skin biopsies from 2 to 5 cm away (n = 4) revealed a dramatically decreased density of CD8⁺ cell infiltration (mean = 60 ± 48 cells/mm²), and skin biopsies of the arm region (n = 4) collected at the same time as the lesional biopsies averaged 12 ± 6 cells/mm² (Fig. 3 a). In addition, 30–50% of the CD8⁺ cells detected in both the distal arm area and neighboring genital skin were located within blood vessels, not in extravascular tissue. Thus, the CD8⁺ cell infiltration into dermal and epidermal skin tissue in persons with genital herpes lesions was limited spatially to the site of viral infection.

We then stained HSV-2–specific CD8⁺ T cells in situ by the Qdot 655–conjugated multimers during the acute lesional stage. In four out of the five patients tested, HSV-2–specific CD8⁺ T cells were detected in the epidermis, dermis, and dermal–epidermal junction of active lesions (Fig. 3 b) and were not visualized in either biopsies of distant skin or from areas 2–5 cm away from the active lesions. HSV-2–specific CD8⁺ T cells were not seen if Qdot multimers containing HSV-2 peptide combinations HLA-mismatched to the donor, or CMV pp65 peptide combinations HLA-matched to the donor, were used (Fig. 3 c). Additional specificity of the Qdot multimers for HSV-2 was demonstrated in subject P1, who developed a folliculitis in the genital skin during this study. Biopsy of the follicular lesion showed CD8⁺ T cell infiltration (mean = 434 ± 24 cells/mm²) but no evidence of HSV-2–specific CD8⁺ T cells (Fig. 3 c).

The frequency of CD8⁺ T cells specific for the HLA B*0702–restricted epitope UL49 (aa 49–57) comprised ∼10% of all CD8⁺ cells in lesional tissue, ∼100 times the frequency that these cells were found in blood (0.1% of CD8⁺ cells). The enrichment of HSV-2–specific CD8⁺ T cells in genital lesions was also seen in the two HLA A*0201–bearing subjects, where the frequency of detection was increased from undetectable in blood to 0.7 and 1.9% in tissues for subject P2 and P6, respectively (see Fig. 6 c). These data demonstrate that lesional CD8⁺ T cells recognizing HSV-2 epitopes can be identified with a high degree of specificity using the Qdot multimer technique, that HSV-2–specific CD8⁺ T cells home specifically to HSV-infected skin lesions, and that the CD8⁺ T cell response to HSV-2 is highly focused anatomically.

To study the potential persistence of HSV-2–specific CD8⁺ T cells after resolution of a herpetic lesion, subjects underwent repeated biopsies of the genital skin at the time of healing and sequential time periods from 2 to 8 wk after lesion resolution (Table I). The epidermis was intact and appeared clinically normal when these biopsies were taken. In contrast to acute infection, where HSV-2–specific CD8⁺ T cells could be detected in all skin layers (Fig. 3 b), CD8⁺ T cells in newly healed lesions were concentrated at the dermal–epidermal junction and in the upper dermis. HSV-2– specific CD8⁺ T cells were detected in small clusters in the upper dermis, in particular just beneath the basement membrane, and very few CD8⁺ cells infiltrated the epidermis (Video 4).

Resolution of the genital lesion was associated with skin reepithelialization and loss of HSV by culture or DNA detection in lesional biopsy tissue. However, HSV-2–specific CD8⁺ T cells were consistently detected in healed genital skin along the dermal–epidermal junction. Most of the CD8⁺ T cells, including HSV-2–specific CD8⁺ T cells, were in close contact to basal keratinocytes (Fig. 4, a–c; and Video 5). This persistence of HSV-specific T cells at the dermal–epidermal junction in the lesional biopsies was documented in all of the postresolution biopsies of all six subjects studied. HSV-2–specific CD8⁺ T cells persisted at similar frequencies to those in acute infection (see Fig. 6 c), suggesting that HSV-2–specific CD8⁺ T cells remain locally enriched at the dermal–epidermal junction even in the wake of a waning overall inflammatory response.

Because HSV reactivates from cutaneous sensory nerve endings, we postulated that HSV-2–specific CD8⁺ T cells persisting in healed genital herpes lesions would position themselves in close proximity to these nerves to optimally suppress reactivation. To test this hypothesis, we analyzed the spatial relationship between CD8⁺ T cells and peripheral nerve endings. To visualize the three-dimensional topography of the cutaneous nerve plexus, we prepared dermal sheets, stained them with a monoclonal antibody against neural cell adhesion molecule (NCAM), and performed confocal microscopy. Free-ending nerve fibers and enveloped terminal Schwann cells formed a distinct network throughout the entire superficial dermis (Fig. 5 a). Nearly all the terminal nerve endings spread out within a thin layer below the dermal–epidermal junction, whereas large nerve trunks were located deeper in the dermis. In normal unaffected skin, CD8⁺ cells were mainly detected in capillary blood vessels deep in the dermis and had no association with nerve-ending organs in the skin (Fig. 5 d and Fig. 3 c, normal skin). However, in posthealed lesional skin, many CD8⁺ T cells were found adjacent to basal cells of the epidermis that were contacted by free nerve endings and adjacent to terminal Schwann cells (Fig. 5, b and c; and Video 6). We also documented CD8⁺ T cells with HSV-2 specificity at these sites (Fig. 5 b). These findings showed that persisting CD8⁺ T cells associate with sensory cutaneous nerve endings after lesion healing, often in the vicinity of basal keratinocytes, the primary infected cell type during reactivation of HSV. This spatial relationship suggests that HSV-specific CD8⁺ T cells may control HSV reactivation peripherally by positioning themselves as sentinels at the interface between reactivating nerve endings and adjacent basal keratinocytes.

To investigate the temporal relationship between HSV-2 infection and CD8⁺ T cell infiltration during and after genital herpes episodes, we subjected serial biopsies to viral DNA detection and CD8⁺ T cell quantification. All 10 biopsies taken during acute lesions were both culture and PCR positive for HSV at the time of collection. For HSV-2 DNA, genome copies ranged from 2.2 × 10³ to 5.4 × 10⁶ per 50,000 cells (Fig. 6 a). HSV-2 DNA, as detected by PCR from tissue sections, rapidly decreased after healing (Fig. 6 a). Similarly, a rapid decrease in the number of CD8⁺ T cells occurred during the course of healing (Fig. 6 b). All 31 biopsies done after reepithelialization, including 24 taken from the original lesion site and 7 taken from unaffected normal areas (4 adjacent to and 3 distal from the lesion), were cultured negative for HSV. 20 out of these 24 posthealing biopsies taken from the original lesion site and all 7 biopsies taken from unaffected sites were also negative for HSV-2 DNA.

Four posthealing lesional biopsies from two separate subjects demonstrated reappearance of HSV-2 DNA. Subject P2 had HSV-2 DNA detected in the 2-wk and the 8-wk posthealing biopsies (21 and 19 copies of HSV-2 DNA per 50,000 cells, respectively). No HSV-2 DNA was detected in the biopsies that preceded and followed the 8-wk biopsy. A similar pattern was seen in subject P3, who had HSV DNA detected at the 4-wk and 8-wk posthealing tissues (2 and 15 copies of HSV-2 DNA per 50,000 cells, respectively). Histologic sections of these skin biopsies showed an intact epithelium in all four of these biopsies; with no clinical or histologic evidence of ulceration or epidermal break (Fig. 6, d and e), as such, the HSV-2 shed into tissue was “subclinical” by definition.

Increased numbers of both HSV-2–specific and total CD8⁺ T cells were detected in the posthealing biopsies in which HSV-2 DNA was present (Fig. 6, b–e). For example, the number of total CD8⁺ T cells detected in the 8-wk posthealing biopsies for subjects P2 and P3, respectively, was 452 cells/mm² and 355 cells/mm², an average of four times higher than the two 8-wk biopsies from subject P1, who exhibited no detectable viral DNA after healing (104 cells/mm²; Fig. 6 b). Subjects P2 and P3 possessed different HLA types, and hence, HSV-specific CD8⁺ T cells in the 8-wk biopsies were detected with different Qdot multimers, corroborating the specificity of the in situ detection of HSV-2–specific T cells during these episodes of subclinical HSV-2 reactivation (Fig. 6c–e). The association of CD8 infiltration with detection of HSV-2 DNA in the intact epidermis, in the absence of clinical symptoms, is most compatible with subclinical viral reactivation in which host containment of viral replication occurred before reactivated virus underwent extensive viral replication and disrupted the epithelial surface. These findings suggest that the HSV-2–specific CD8⁺ T cells at the site of genital herpes lesions contribute to local containment of viral replication.

One of the major questions in HSV-2 immunobiology has been the disparity between animal model data suggesting that functional host T cell responses in the ganglia have the ability to markedly limit viral reactivation at the ganglion level (25, 26) and human clinical studies of HSV-2 reactivation, which demonstrate that mucosal and genital skin shedding of HSV-2 in the immunocompetent person is extremely frequent (27–29). With once daily sampling, HSV-2 reactivation is seen in 20–60% of days in immunocompetent persons. More recent data suggest that high titer reactivations of HSV-2 lasting only 6–12 h are common (Mark, K., personal communication). These human clinical studies indicate that control of mucosal HSV-2 reactivation at the ganglion level in humans is incomplete at best. It is likely that immune control at both the ganglion and peripheral mucosal level is needed for containment of HSV reactivation. This study was designed to define whether local host control of HSV-2 played an important role in containing viral reactivation after lesional clearing. Our data indicate that much of the control of HSV-2 reactivation in humans appears to be consolidated in the peripheral mucosal immune compartment.

Our study indicates that the peripheral mucosal immune response to HSV-2 infection is highly concentrated both spatially and functionally. The cytolytic CD8⁺ T cell infiltration in genital herpes lesions is very specific to the site of infection. The dramatic reduction of CD8⁺ T cell density 2 cm outside of the affected skin-lesion area indicates that CD8⁺ T cell infiltration to an HSV infection is a very focused local event. The HSV-2–specific CD8⁺ T cells we detected were highly concentrated in the genital lesion. Quantitatively, HSV-2–specific T cells were present at 100 times higher concentrations in lesional skin than that in circulating blood. Even after lesion resolution, the enrichment of HSV-2–specific CD8⁺ T cells persisted in genital skin for an extended time period. Our study extended for only 2 mo, and local persistence was documented to last at least this long. Moreover, the spatial localization of HSV-2–persistent CD8⁺ T cells appears to be of physiological importance. Unlike active lesions, where HSV-2–specific CD8⁺ T cells were distributed in all layers of skin to eliminate viral infection and prevent spreading, HSV-2–specific CD8⁺ T cells that persisted after healing were detected in the upper dermis and the dermal–epidermal junction, coincident with sensory nerve endings. This physical closeness between persistent CD8⁺ T cells and sensory nerve endings would provide a potential mechanism in which the local immune system could quickly respond to and eliminate reactivated virus before it undergoes extensive replication. We demonstrated four such instances of such containment in two patients, as well as the apparent ability of host T cell responses in the periphery to contain viral reactivation before epithelial ulceration could occur.

Estimates of the frequency of HSV-specific CD8⁺ T cells in genital skin are likely to be underestimates of their true prevalence in our studies. The cocktail of Qdot multimers we used was based on only a limited mapping of the breadth of the T cell response. As such, it is likely we were assaying only a portion of the CD8⁺ T cell response to the virus. In more recent studies using a wider breadth of peptides, we have shown that many subjects possess CD8⁺ T cells specific to a wider multitude of HSV proteins than we had available here (14). As such, our quantitative assessments are likely to be underestimates of the number of HSV-specific CD8⁺ T cells in genital skin.

This study also provides evidence that HSV-specific CD8⁺ T cells play a role in immune protection against clinically symptomatic HSV-2 reactivation in genital skin. The biopsies in which we detected HSV DNA in skin tissue, without evidence of epithelial cell destruction and in association with increased localized infiltration with HSV-2–specific CD8⁺ T cells, suggest that resident HSV-2–specific T cells appear to be a critical component of cutaneous viral containment. The need to evolve a rapid response to viral replication in the desired region is emphasized by a clinical study, which showed that almost 40% of genital HSV-2 reactivation in humans were asymptomatic and lasted 6 h or less (Mark, K., personal communication). These short reactivations average from 10³ to 10⁴ copies of HSV-2 DNA per millimeter in swab specimens, indicating that prompt elimination of virus by the host mucosal immune responses occurred. Prospective studies investigating the relationship between in vivo persistence of HSV-2–specific T cells and the site and timing of clinical and subclinical reactivation are needed to define this association more closely. Although such studies require serial biopsies, patients with genital HSV-2 are affected greatly by this infection. The Qdot multimers and the in situ staining techniques we describe in this paper provide the tools for such studies to be undertaken.

In summary, HSV-2–specific CD8⁺ T cells appear to play a direct role in viral clearance in vivo. These CD8⁺ T cells migrate very specifically into genital lesions and persist preferentially in genital skin in the subepidermal layer contiguous to peripheral nerve endings, suggesting that they play a role in the earliest response to viral reactivation. Our studies indicate that persisting HSV-2–specific CD8⁺ T cells function in immune surveillance and that the peripheral T cell response influences the frequency and clinical course of HSV-2 reactivation in humans.

Six subjects (five female and one male) with recurrent genital HSV-2 infection and lesions were recruited into a study protocol approved by the University of Washington Institutional Review Board. All subjects were HIV-1/2 seronegative and had a history of culture-proven herpes disease. Serial 3-mm punch biopsies were obtained in subjects during an episode of symptomatic recurrence from the active lesion and at different stages thereafter (Table I). Lesional punch biopsies were taken from the erythematous skin immediately adjacent to a blister. After healing, punch biopsies were obtained from the site of the original lesion. In addition, 3-mm punch biopsies from an unaffected area 2–5 cm away from a lesion and/or the arm were also performed. Punch biopsies were divided into two halves: one half was snap frozen in optimum cutting temperature compound (Tissue-Tek; Sakura Finetek USA, Inc.) and stored at −80°C, whereas the other half was cut into three to four pieces and used immediately for immunofluorescence staining. Blood samples were drawn at the time of each biopsy.

In some biopsies, the epidermis and dermis were separated as described elsewhere (30). In brief, the subcutaneous tissue was removed with surgical scissors, and the remaining tissue was placed in cold PBS containing 5 mM EDTA and incubated at 4°C overnight. The epidermis was then dissected off the underlying dermis with a stereoscope (KL1500; Carl Zeiss MicroImaging, Inc.).

Peptide–MHC complexes (pMHCs) were formed in vitro as previously described (31). The HSV-2 peptides used were A*0101/ASD (UL46, aa 354–362), A*0201/FLI (UL27, aa 443–451), A*0201/FLV (UL47, aa 289–298), A*0201/GLA (UL47, aa 551–559), B*0702/RPR (UL49, aa 49–57), and B*0702/GPH (UL26, aa 475–483). The CMV peptide used was B*0702/NLV (pp65, aa 495–503). Streptavidin-coated Qdot 655 (Quantum Dot Corporation) was conjugated with a saturating amount of biotinylated pMHC on ice for 15 min and purified by gel filtration.

Fresh tissue sections were stained overnight on ice in a casein solution (Vector Laboratories) with Qdot-conjugated MHC–peptide multimers (1:100), mouse anti–human CD8 (1:100; Caltag Laboratories), and rabbit anti–HSV-2 (1:500; DakoCytomation). Tissues were washed, fixed in 4% paraformaldehyde for 20 min at room temperature, and incubated with donkey anti–mouse Alexa Fluor 488 and donkey anti–rabbit Alexa Fluor 594 (1:100; Invitrogen) for 1 h at room temperature. After washing and incubating with DAPI (Fluka), tissue sections were mounted in Mowiol 40-88 containing 2.5% wt/vol DABCO (Sigma-Aldrich). Infiltrating CD8⁺ T cells were enumerated in tissue fields of 460 × 460 μm in size, including the epidermis and 150–300 μm deep into the dermis. For calculating the median number of CD8 and HSV-specific CD8⁺ T cells, at least five fields were counted for each biopsy. To calculate the percentage of HSV-2–specific CD8⁺ T cells, Qdot multimer and CD8 double-positive cells were located, and total CD8⁺ cells along those with HSV specificity were enumerated in tissue with a volume of 460 × 460 × 20 μm. For costaining of CD8⁺ cells and nerve fibers, monoclonal CD8 Alexa Fluor 647 (1:100; Caltag Laboratories) and NCAM (1:50; Beckman Coulter) were used as primary antibodies, and donkey anti–mouse Alexa Fluor 488 was used as a secondary antibody.

Confocal microscopy of immunofluorescence-labeled tissue was performed on a confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.). For excitation of fluorescent dyes, the following laser lines were used: Alexa Fluor 488 and FITC, 488 nm; Alexa Fluor 594, 543 nm; Alexa Fluor 647, 633 nm; and Qdot 655, 458 nm. DAPI was excited in two-photon mode with a near-infrared laser (Chameleon; Coherent Inc.) tuned at 780 nm. The detection bands for the different dyes were as follows: Alexa Fluor 488 and FITC, 500–550 nm; Alexa Fluor 594, 565–615 nm; Alexa Fluor 647, 650–710 nm; Qdot 655, 650–710 nm; and DAPI, 435–485 nm. To minimize cross talk between dyes, imaging of multiple labeled samples was performed individually for each dye in a sequential mode. To reduce image noise, Kalman averaging (generally two to four frames) was used. For viewing of Qdot 655, a filter set with a 460-nm short-pass excitation filter, a 475-nm beam splitter, and a 635–675-nm band-pass emission filter was used (Chroma Technology Corp.). Three-dimensional rendering of confocal stacks was performed with Imaris (Bitplane AG) or Volocity (Improvision) software.

5 × 10⁶ PBMCs were incubated with Qdot 655–conjugated pMHC multimers for 15 min at room temperature. CD8-FITC antibody (Caltag Laboratories) was added, and cells were incubated for 30 min on ice, washed, and fixed. Acquisition was performed on a cytometer (LSR II; Becton Dickinson), and data were analyzed using CELLQuest software (Becton Dickinson). The detection limitation for HSV-2–specific CD8⁺ T cells in PBMCs by flow cytometry was 0.01%.

DNA from skin biopsy samples was extracted using tissue kits (EZ1; QIAGEN). At least two sets of 60–80-μm thick cross-section tissues from each biopsy specimen were treated and followed previously described procedures for extraction and testing of HSV DNA (32, 33). Subtype-specific typing was performed by real-time PCR (34). The HSV DNA copy numbers were normalized to 50,000 cells by measuring the β-globin gene copy numbers using 5′-TGAAGGCTCATGGCAAGAAA-3′ and 5′-GCTCACTCAGTGTGGCAAAGG-3′ as primers and 5′-TCCAGGTGAGCCAGGCCATCACTA-3′ as a probe. The detection limitation of the assay was one copy per 50,000 cells.

Video 1 shows detection of CD8⁺ cells in a genital herpes lesion. Video 2 provides a 180° view showing direct interaction between CD8⁺ T cells and HSV-2–infected keratinocytes. Video 3 depicts the surface distribution of CD8 and TCR molecules on an HSV-2–specific CD8⁺ T cell. Video 4 shows the distribution of CD8⁺ and HSV-specific T cells in a newly healed genital herpes lesion. Video 5 provides a 360° display of persistent HSV-specific CD8⁺ T cells in healed genital skin. Video 6 presents a 360° display of persistent HSV-specific CD8⁺ T cells contiguous to cutaneous sensory nerve endings in healed genital skin.

We would like to thank David McDonald and Adrian Quintanilla for support with imaging; Christopher McClurkan, Alexis Klock, Khamsone Phasouk, and Hong Xie for technical assistance; Michael Remington and Steve Kuntz for clinical assistance; and Drs. David Myerson and Christopher Wilson for helpful discussions and suggestions.

This work was supported by National Institutes of Health grants AI-30731, AI-42528, AI-50132, and HD-051455.

The authors declare that they have no competing financial interests.