Selective retention of herpes simplex virus-specific T cells in latently infected human trigeminal ganglia

High Prevalence of Latent HSV-1 and VZV Infection of Human TG.

We first determined the prevalence and load of latent α-herpesvirus DNA in paired left and right TG of 17 donors by real-time PCR. Eleven donors had DNA of both viruses; 1 donor had only HSV-1; 4 donors had only VZV; and 1 donor had no detectable DNA of either virus. No HSV-2 DNA was detected in any TG sample (Fig. 1A and data not shown). DNA of the same virus was always detected in both TG of each individual, and the donors' antibody serostatus correlated with the presence of the respective viral DNAs (data not shown). In line with a previous report, the median number of HSV-1 genome copies per 10⁵ TG cells (3,321 ± 879) was significantly higher than that for VZV genomes (565 ± 639) and correlated significantly between the paired TG (Fig. 1B) (20). Transcription of HSV-1 LAT, but not the genes encoding the late viral structural proteins glycoprotein D and G, was detected in the HSV-1 DNA-positive TG samples, showing true HSV-1 latency (data not shown) (1).

Human TG-Resident CD8⁺ T Cells Are Activated and Express a Late Effector Memory Phenotype.

Whereas extravasated T cells have been identified in human latently infected TG, their phenotypic and functional characteristics are largely unknown (7, 8). The activation and differentiation status of the human TG-resident T cells was determined by flow cytometry on single cell suspensions of the left and right TG and was compared with paired peripheral blood (PB) T cells of 10 donors. Overall, the total TG cell pool contained 0.47 ± 0.15% CD45⁺ cells, with, on average, one-third of the CD45⁺ cells coexpressing CD3 and CD8. The number of CD8⁺ T cells expressing the activation marker CD69 (21) was significantly higher in the TG compared with PB, indicating that the TG-derived CD8⁺ T cells had been recently activated in situ (Fig. 1C).

Naïve CD8⁺ T cells express CD45RA and high levels of CD27 and CD28. After primary activation, the effector T cells differentiate into memory CD8⁺ T cells, generally characterized by a loss of CD45RA and a gain of CD45RO (22). After repeated antigenic stimulation, the costimulatory molecules CD28 and, subsequently, CD27 are down-regulated. These T cells, characterized as CD45RA⁻RO⁺CD28⁻CD27⁻, are considered late memory T cells (23). Studies of the human TG CD8⁺ T cell pool indicated that they contained significantly fewer naïve cells and displayed lower expression of CD27 and CD28 compared with paired PB CD8⁺ T cells (Fig. 1C). Notably, the numbers of CD8⁺ T cells expressing CD69, CD45RA, CD27, or CD28 correlated significantly between the paired TG (Fig. 1D).

Human memory T cells can be assigned to two broad classes on the basis of lymph node homing potential and function. Central memory T cells express the lymph node homing receptors CCR7 and CD62L, entailing their ability to migrate to secondary lymphoid organs. Effector memory T cells, which lack these receptors, are enriched in peripheral tissues and exert immediate effector functions (24). Analyses on three paired TG samples containing both viruses demonstrated that, contrary to the paired PB samples, the majority of the CD8⁺ T cells lacked expression of CCR7 and CD62L, implicating that other homing receptors are used by the CD8⁺ T cells to enter the TG (Fig. 1E). Collectively, the flow cytometric analyses indicate that human TG-resident CD8⁺ T cells are activated and express a late effector memory phenotype.

The T Cell Infiltrate in Human TG Is Oligoclonal.

To investigate the clonal diversity of the TG-resident T cell infiltrates, we determined the rearrangement of the T cell receptor gamma (TCRG) locus by multiplex PCR on DNA isolated from TG samples of 14 donors (25). The TCRG locus is rearranged early during T cell development in both TCRαβ and -γδ lineage precursors and is considered the prototypic TCR locus for the detection of T cell clonality in clinical specimens (26). Typical polyclonal Gaussian curves, signifying polyclonality, or one or two peaks, illustrative for the rearrangement of the TCRG locus on one or both chromosomes, were detected when human PB mononuclear cells (PBMCs) or a monoclonal T cell leukemic cell line were assayed, respectively (Fig. 2 A–D) (25). TCRG GeneScan analyses demonstrated a reproducible, markedly restricted pattern of TCRG rearrangements in the majority of the analyzed TG samples irrespective of the latent virus type present. Moreover, donor-specific dominant TCRG rearrangements strongly correlated between the left and right TG (Table 1). These findings suggest that the TG were infiltrated with a limited number of T cell clones, which were similar between the left and right TG of the same donor.

Human TG-Resident T Cells Are Located in the Proximity of Neurons Latently Infected with HSV-1 but Not with VZV.

The spatial relation between T cells and the latently infected neurons in human TG was determined by in situ hybridization and immunohistochemistry on TG samples from 12 additional donors. The TG of nine donors contained both viruses, and three contained only VZV. VZV latency is characterized by expression of a restricted set of VZV proteins, including those from ORFs 4 (ORF4), 29, 62, 63, and 66 (3, 4). Expression of these proteins is restricted to the cytoplasm of latently infected neurons (3, 4, 7). As reported previously, positive in situ hybridization signals with the HSV-1 LAT-specific oligonucleotide probe (Fig. 3A) and VZV ORF62 protein expression (Fig. 3C) were confined to the nuclei or cytoplasmic vesicles of the neuronal cell bodies, respectively, delineating latency of each virus within neurons (1, 3, 4, 7, 8).

T cells were found in all TG analyzed and were generally dispersed between the neurons. T cell numbers varied between donors. The median number of CD3⁺ cells per sensory neuron cell body (0.32 ± 0.08) was significantly lower in TG sections of HSV-1 seronegative donors compared with HSV-1 seropositive donors (2.11 ± 0.41). Occasionally, T cells clusters were detected in juxtaposition to the sensory neuron cell bodies. The T cell clusters were particularly associated with LAT⁺ neurons (Fig. 3 A–B), but not with VZV ORF62⁺ neurons (Fig. 3 C–D). However, not all LAT⁺ neurons were surrounded by T cells, suggesting that the cognate antigen was expressed only by a subset of the HSV-1 latently infected neurons (Fig. 3 A–B). The T cell clusters composed of both CD8⁺ and CD4⁺ cells and, in line with the flow cytometric analyses (Fig. 1C), both T cell subtypes expressed CD69 (Fig. 3 E–H). GrB and CD8 immunohistochemistry on consecutive tissue sections demonstrated that the TG-infiltrating CD8⁺ T cells, particularly those within the neuron-interacting T cell clusters, expressed grB, implicating their cytolytic potential (Fig. 3 I–J). Conversely, histological examination of the TG samples revealed only limited neuron damage, which was not different between neurons with or without T cell clusters (data not shown).

Human TG-Derived T Cells Are Reactive to HSV-1 but Not to Latency-Associated VZV Proteins.

The data suggest that the TG-resident T cells may be differentially involved to control neuronal HSV-1 and VZV latency. To evaluate whether these T cells are directed to the latent virus, TG-derived T cells were expanded by mitogenic stimulation, and the resultant T cell lines (TCLs) were assayed for the presence of virus-specific T cells by using autologous EBV-transformed B cell lines (BLCLs) as antigen-presenting cells (APCs). Because BLCLs are not permissive to VZV infection (2), we used an alternative strategy in which BLCLs were infected with recombinant vaccinia viruses (rVV) expressing prototypic latency-associated VZV proteins ORF4, -29, -62, and -63 (2, 3). In parallel, BLCLs were HSV-1 or mock infected or infected with a vaccinia virus with no insert as a control. Antigen-specific T cell reactivity was determined by IFN-γ enzyme-linked immunosorbent spot-forming (ELISPOT) on TCLs derived from the left TG of six donors. Four of five TCLs, all generated from HSV-1 and VZV double-infected TG, demonstrated HSV-1 T cell reactivity (Fig. 4). One of two HSV nonresponding TCLs was obtained from a TG devoid of both viruses (TG donor 54004). No significant T cell responsiveness toward the rVV–VZV ORF-infected BLCL was detected compared with the BLCL infected with vaccinia virus control (Fig. 4). The T cell response to the rVV–VZV ORF-infected BLCL, which is higher compared with mock infected BLCL, is likely due to the intrinsic superantigen activity of vaccinia virus, as has been described for the analogous virus parapox ovis (27).

The HSV-specific T cell subset was determined by intracellular IFN-γ flow cytometry. Compared with TCLs recovered from TG devoid of HSV-1, higher numbers of HSV-reactive T cells were identified in the TCL obtained from HSV-1-infected TG (Table 2). The HSV-1-specific T cell reactivity was concordant with the IFN-γ ELISPOT assays performed on the same TCL (Fig. 4 and Table 2). The HSV-specific T cells identified were mainly CD8⁺ and HSV serotype-specific (Table 2). These results suggest the selective retention of HSV-1- but not VZV-specific T cells in latently infected human TG.

Donors.

The TG biopsies (left and right TG) and heparinized PB samples were obtained from 29 subjects (median age 81, ranging from 49 to 98 yr) at autopsy with a median postmortem interval of 5.5 h (ranging from 4 to 10 h). The clinical specimens were generously provided by the Netherlands Brain Bank (Amsterdam). This unique, large TG tissue panel consisted of nine controls and 20 donors with a neurologic disease history affecting the CNS (mainly Alzheimer's disease and Parkinson's disease). The cause of death was not related to an α-herpesvirus infection. No significant differences in the virologic and immunologic parameters analyzed were detected between the control donors and those with a history of CNS disease (data not shown). Written, informed consent from either the donor or the next of kin was given, and the protocol was approved by the local ethical committees. The study was conducted according to the tenets of the Declaration of Helsinki.

Generation of Single-Cell Suspensions and T Cell Lines from TG.

Human TG were removed aseptically and either immediately snap-frozen in liquid nitrogen for in situ analysis or transferred to tubes containing RPMI medium 1640 containing 10% FBS and antibiotics for nucleic acid extraction and cell isolation (GIBCO/BRL, Carlsbad, CA). In case of the latter procedure, TG were fragmented in PBS containing 1% BSA (referred to as P1B medium) and subsequently treated with Liberase blendzyme 3 (0.2 units/ml) (Roche, Basel, Switzerland) at 37°C for 1 h. Dispersed cells were filtered through a 100 μM pore-size mesh, and the flow-through was collected and resuspended in P1B medium. TG-derived TCLs were generated by phytohemagglutinin (PHA) stimulation of one-fifth of the TG single-cell suspension in the presence of γ-irradiated (3,000 rad) allogeneic PBMCs in RPMI medium 1640 supplemented with 10% human pooled serum, antibiotics, and recombinant human IL-2 (50 units/ml) essentially as described in ref. 46. The limited number of T cells necessitated two rounds of PHA stimulation to obtain sufficient T cells for cryopreservation and subsequent functional assays. PHA stimulation was chosen as the best alternative to limit selective outgrowth of TG T cell subpopulations. Nonetheless, we cannot rule out the option that the culture conditions may have influenced the repertoire of the TG-infiltrating T cells. Limiting dilution assays on freshly isolated TG T cells may overcome this potential problem and should be considered in future studies.

IFN-γ ELISPOT.

The BLCLs were infected overnight with HSV-1 or the rVV-expressing VZV ORF4, -29, -62, or -63 at a multiplicity of infection of 10, as described in refs. 46 and 47. An rVV poly 186 without insert was used as control (46). The antigen-specificity of TCLs was determined in an IFN-γ ELISPOT assay, as described in ref. 48, using 10⁴ T cells and 2 × 10⁴ APCs per well. Spots were counted with an automated ELISPOT reader (Sanquin, Amsterdam, The Netherlands).

In Situ Analyses.

Primary mAbs directed to the following antigens were used at predefined concentrations: CD3 (DAKO), CD4 (DAKO), CD8 (Monosan, Canton, MA), CD69 (Biolegend, San Diego, CA), IL-15 (R&D Systems, Minneapolis, MN), grB (DAKO), HLA class II (DAKO), VZV ORF62 (Millipore, Billerica, MA), and the appropriate isotype controls. Stainings were made on acetone-fixed cryosections, or paraformaldehyde-fixed paraffin sections, in case of the combined analysis of grB and CD8, with the avidin–biotin complex system (DAKO) (49). HSV-1 LAT in situ hybridization was performed on cryosections by using a digoxigenin-conjugated LAT oligonucleotide probe, as described (7).

For additional information on isolation of nucleic acids and PCR analyses, isolation of PBMC and generation of BCLs, flowcytometry, and statistical analyses, see supporting information (SI) Text.