|Home | About | Journals | Submit | Contact Us | Français|
Following primary infection, human herpesvirus 6 (HHV-6) establishes a persistent infection for life. HHV-6 reactivation has been associated with transplant rejection, delayed engraftment, encephalitis, muscular dystrophy, and drug-induced hypersensitivity syndrome. The poor understanding of the targets and outcome of the cellular immune response to HHV-6 makes it difficult to outline the role of HHV-6 in human disease. To fill in this gap, we characterized CD4 T cell responses to HHV-6 using peripheral blood mononuclear cell (PBMC) and T cell lines generated from healthy donors. CD4+ T cells responding to HHV-6 in peripheral blood were observed at frequencies below 0.1% of total T cells but could be expanded easily in vitro. Analysis of cytokines in supernatants of PBMC and T cell cultures challenged with HHV-6 preparations indicated that gamma interferon (IFN-γ) and interleukin-10 (IL-10) were appropriate markers of the HHV-6 cellular response. Eleven CD4+ T cell epitopes, all but one derived from abundant virion components, were identified. The response was highly cross-reactive between HHV-6A and HHV-6B variants. Seven of the CD4+ T cell epitopes do not share significant homologies with other known human pathogens, including the closely related human viruses human herpesvirus 7 (HHV-7) and human cytomegalovirus (HCMV). Major histocompatibility complex (MHC) tetramers generated with these epitopes were able to detect HHV-6-specific T cell populations. These findings provide a window into the immune response to HHV-6 and provide a basis for tracking HHV-6 cellular immune responses.
Human herpesvirus 6 (HHV-6) was identified in 1985 in lymphocyte cultures obtained from immunocompromised individuals and patients with lymphoproliferative diseases (59). HHV-6 is a lymphotropic virus (genus Roseolovirus) of the betaherpesvirus family, which also contains human cytomegalovirus (HCMV or HHV-5) and HHV-7. There are two major HHV-6 variants, HHV-6A and HHV-6B (1). The HHV-6 genome is highly conserved between the HHV-6 variants (>90%) (18, 38). Conversely, significant differences are observed in the biology, immunology, and epidemiology of the two variants of the virus (51). Primary infection with HHV-6B occurs in most individuals early in life and results in roseola infantum or exanthem subitum (5, 82). This primary infection generally induces a febrile episode that is considered one of the most important reasons for fever-associated hospital visits in children up to 2 years old (34). In contrast, primary infection with the HHV-6A variant seems to occur without clinical symptoms and later in life and has not been documented in HHV-6-naïve individuals (18).
Like other herpesviruses, HHV-6 upon primary infection establishes a lifelong persistence in the host and can be found in peripheral blood, salivary glands, kidneys, and the central nervous system (CNS) (reviewed in references 15 and 51). Active replication in vivo occurs in activated CD4+ T cells and salivary glands (28).
Although HHV-6 reactivation in healthy individuals usually occurs without significant morbidity, in immunocompromised persons such as solid-organ or hematopoietic cell transplant recipients, reactivation can result in CNS morbidity, pneumonitis, transplant rejection, or delayed engraftment (19, 63). Furthermore, reactivation of HHV-6, suggested by an increase in viral load and antibody titers, has been associated with CNS pathologies such as encephalitis, meningitis, and multiple sclerosis (MS) (64, 84). HHV-6 reactivation also has been associated with chronic fatigue syndrome (CFS) (41), febrile seizures (83), myocarditis (3, 14), pityriasis rosea (23), and drug-induced hypersensitivity syndrome (DIHS) (8). However, the role of HHV-6 in MS, DIHS, and CFS remains controversial (51, 64, 79).
To date, there have been limited studies of the specificity of the immune response to HHV-6 infection, despite the fact that cellular immune responses are fundamental in controlling primary infection and reactivation of the closely related virus HCMV (56, 76) and likely are involved in pathology associated with HHV-6 reactivation (27, 85). Early studies established T cell proliferative responses to HHV-6 in most healthy adults (69, 81, 87). Subsequent studies using peripheral blood mononuclear cells (PBMCs) have demonstrated T cell responses by both CD4+ and CD8+ T cells (24, 40, 77). However, HHV-6-specific T cell clones and T cell lines (TCLs) have been established only for CD4+ T cells (40, 42, 69, 80, 81, 87). The fine specificity of the T cell response to HHV-6 is largely unexplored. To date, HHV-6 T cell responses have been mapped only to 4 relatively large segments of the U11 tegument protein (69) and to seven HHV-6 peptide sequences (70). These seven peptide sequences, however, are cross-reactive with peptides from other pathogens (37, 48) or candidate autoantigens (7, 47, 70, 72).
Development of subunit vaccines that confer protection while avoiding pathologies associated with the immune response will rely on the identification of antigens, and more specifically T cell epitopes, associated with protective or deleterious immune responses. In this matter, the field of HHV-6 immunology lags well behind that of other herpesviruses like CMV and Epstein-Barr virus (EBV). T cell epitopes are identified classically using overlapping peptides spanning the full sequence of a protein. This type of approach is well suited for small pathogens with relatively few potential epitopes or for pathogens for which immunodominant proteins have already been defined. In contrast, for HHV-6, there are more than 40,000 individual potential 9-mer peptide epitopes. An alternative to comprehensive testing of overlapping peptides is a candidate epitope approach, in which T cell epitope prediction algorithms are used to narrow the number of peptides to screen (11, 60, 67, 74). This approach is particularly useful in the analysis of complex organisms with very large numbers of potential epitopes.
Here we combined proteomics, T cell epitope prediction, and synthetic peptide screening to identify major histocompatibility complex (MHC) class II-restricted CD4+ T cell epitopes from HHV-6. We evaluated the recognition of predicted epitopes by PBMCs obtained from healthy donors and by T cell lines (TCLs) generated by expansion in vitro. Overall, 11 HHV-6-derived CD4+ T cell epitopes were identified, predominantly derived from abundant virion proteins. Memory responses to these epitopes were observed by enzyme-linked immunosorbent spot (ELISpot) assay, intracellular cytokine staining (ICS), and MHC tetramer staining assays and included polyfunctional cytokine and potential cytotoxic responses.
For the generation of HHV-6-specific T cell lines, we used purified preparations of HHV-6 that were disrupted with Triton X-100 and also lysates of HSB-2 cells infected with HHV-6A (GS strain). Preparations of both viral strains (HHV-6A strain GS and HHV-6B strain Z29) were obtained from Advanced Biotechnologies Inc. (Columbia, MD). Seed culture of infectious HHV-6A (GS) was kindly provided by the HHV-6 Foundation and propagated in CCRF-HSB-2 cells (ATCC CCL-120.1).
HSB-2 cells were grown in Iscove modified Dulbecco medium (IMDM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were infected with HHV-6A (GS) in IMDM plus 5% FBS. After 4 to 5 days, cells were collected and washed with phosphate-buffered saline (PBS) and a cell lysate was prepared by repeated freeze-thaw cycles followed by centrifugation at 1,200 rpm for 10 min. The supernatant was collected and incubated at 57°C for 1 h to inactivate the virus, divided in small aliquots, and stored at −70°C. Lysate from noninfected HSB-2 cells was generated in a similar manner and used as a control.
Cytokines in PBMCs and T cell line (TCL) tissue culture supernatants were determined in supernatants at multiple time points by using a Th1/Th2/Th17 cytometric bead array (CBA) assay according to the instructions provided by the manufacturer (BD Biosciences, San Diego, CA). Using this system, we measured the concentrations of interleukin-2 (IL-2), IL-4, IL-10, IL-17, tumor necrosis factor alpha (TNF-α), and gamma interferon (IFN-γ). Independent triplicate cultures were set for each one of the times to analyze in order to avoid distortions of the culture conditions. Cell culture supernatants of PBMCs were generated by stimulation of 1 × 106 cells with pretitrated HHV-6 antigen preparations, medium, or control lysate in a 200-μl final volume in 96-well plates. Supernatants of TCLs were generated in a similar manner using 5 × 104 T cells and 1 × 105 irradiated PBMCs as antigen-presenting cells (APCs).
To predict HLA-DR1 (DRB1*0101) binding epitopes, we used the HHV-6B (Z29) sequence from the NCBI genome database (Refseq accession number NC 000898 and GenBank accession number AF157706). Each potential 9-mer binding frame was evaluated using two independent prediction algorithms, P9 (11, 35, 67) and SYFPEITHI based on peptide elution data (61), as previously described (11). Overall, 42,838 possible 9-mer epitopes in 104 predicted open reading frames were evaluated in the whole-genome analysis, and 4,253 possible 9-mer epitopes present in 6 abundant virion proteins were evaluated in the virion-only analysis. In both analyses, peptides scoring highly for both algorithms as described in the text were synthesized and tested for immunogenicity.
Peptides for initial screening and pool deconvolution experiments were synthesized by Sigma Genosys (Texas) using standard 9-fluorenylmethoxy carbonyl (Fmoc) chemistry and were characterized by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry after deprotection and desalting. All preparations exhibited molecular masses consistent with the expected sequences. Peptides showing positive responses in initial screening assays were resynthesized by 21st Century Biochemical (Marlborough, MA) using standard methods and were purified by high-performance liquid chromatography using a Vydac C18 reverse-phase column to >90% purity before use in subsequent T cell and MHC binding assays.
Blood from six healthy donors was collected under a protocol approved by the Medical School Institutional Review Board of the University of Massachusetts. The HLA class II haplotype was determined by the UMass HLA Typing Core Facility using PCR-based protocols. Participants were recruited after obtaining written informed consent.
To determine binding affinities of peptides to HLA-DR1 (DRB1*0101) molecules, a competition assay was performed in a 96-well setting. Serial dilutions of unlabeled peptide competitors (16 μM to 0.02 μM) were incubated with 25 nM Alexa 488–HA306–318–FRR peptide (acetyl-PRFVK*QNTLRLAT, where K* is lysine modified by Alexa Fluor 488 tetrafluorophenyl ester; Invitrogen) and with 100 nM peptide-free HLA-DR1, produced in Drosophila melanogaster S2 cells. Serial dilutions of unlabeled HA306-318 peptide were used as a control. The mixture was incubated for 3 days at 37°C in 100 mM sodium citrate (pH 5)–50 mM sodium chloride buffer (pH 5.5), containing 0.25% octylglucoside, 185 μg/ml iodoacetamide (IAA), 25 nM EDTA, 0.1% NaN3, and freshly added 0.5-μg/ml phenylmethylsulfonyl fluoride (PMSF). Measurements were performed in a POLARstar Optima plate reader (BMG Labtech), using the fluorescence polarization detection mode and 485-nm excitation and 520-nm emission filters. Polarization values for free and bound Alexa 488–HA306-318–FRR peptide were ~70 and ~350 mPa, respectively. Under the conditions of the assay, ~70% of the Alexa 488–HA306-318–FRR peptide becomes bound in the absence of competitor. Fifty percent inhibitory concentrations (IC50s) were evaluated using a competitive binding equation in Kaleidagraph (Synergy Software; v.4.0).
T cell lines were generated from the PBMC component of Ficoll-fractionated freshly drawn blood of healthy donors that had been obtained by venipuncture using heparin as an anticoagulant. One million PBMCs per well were incubated in T cell medium (RPMI 1640; Invitrogen Corporation, Carlsbad, CA) supplemented with 10% human AB-positive serum (Interstate Blood Bank Inc., Memphis, TN), 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, 2 mM l-glutamine, and 1 mM nonessential amino acids (all from Invitrogen) and 50 μM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). Antigen-specific populations were expanded by culture in the presence of one of the following antigens: (i) purified and disrupted HHV-6B (Z29) (10 μg/ml), (ii) purified and disrupted HHV-6A (GS) (10 μg/ml), (iii) HHV-6A (GS)-infected HSB-2 cell lysate (1:100), and (iv) noninfected HSB-2 cell lysate (1:100). After 3 to 4 days of incubation at 37°C, fresh T cell medium supplemented with 100 U/ml IL-2 (Proleukin; Chiron Corporation, Emeryville, CA) was added.
IFN-γ ELISpot assays were performed using Ready Set Go human IFN-γ ELISpot reagents (eBiosciences, San Diego, CA) in Immobilon-P membrane multiscreen HTS plates (Millipore, Billerica, MA), essentially as described previously (11). Responses were considered positive if a primary TCL showed an average response of >80 spots/106 T cells (5 to 10 spots/well over background) or PBMCs showed an average response of >10 spots/106 cells, respectively. Furthermore, a specific response had to be at least 2-fold greater than the background (medium controls) and at least 2-fold greater than the standard deviation of duplicate measurements.
Intracellular cytokine staining (ICS) was performed for the functional and phenotypic characterization of TCLs and PBMCs. Antigen-presenting cells (APCs) were prepared from autologous PBMCs that were irradiated, pulsed with peptide (5 μg/ml) overnight, and washed to remove unbound peptide. To evaluate responses by TCLs or PBMCs, 1 × 106 peptide-pulsed APCs were incubated with 3 × 105 T cells for 6 or 10 hours, respectively, with brefeldin A (Golgi-plug; BD Biosciences, San Jose, CA) and monensin (Golgi-stop; BD) added for the final 4 or 6 hours of incubation, respectively. If indicated, fluorescein isothiocyanate (FITC)-conjugated antibodies to CD107a and CD107b were added at the point of stimulation. TCL responses were considered positive if the frequency of functional cells was at least 2-fold over that of background (medium controls or irrelevant peptide stimulations). PBMC responses were considered positive if the frequency of functional cells was at least 2.3-fold over background. Cell responses were analyzed using a Cytofix/Cytoperm Plus kit and antibodies to IFN-γ–phycoerythrin (PE), CD4-allophycocyanin, and CD3-peridinin chlorophyll protein (PerCP) (all from BD Biosciences, San Jose, CA). After staining, cells were analyzed on a 4-color FACSCalibur cytometer (BD Biosciences, San Jose, CA).
Antibody titers to HHV-6 were determined by indirect immunofluorescence assay (IFA) using an HHV-6 IgG antibody kit according to the protocol described by the manufacturer (Advanced Biotechnologies, Columbia, MD). Antibody titers presented are the highest of a series of 2-fold serum dilutions giving a positive response. The antigen used in the kit consists of HSB-2 cells infected with HHV-6A (GS).
Biotinylated HLA-DR1-peptide complexes were prepared using insect cell-derived protein as described previously (12). For staining of peptide-specific cells, 1 × 106 to 3 × 106 PBMCs were incubated with a 2-μg/ml tetramer preparation for 2 h at room temperature. During the last 20 min of incubation, antibodies to CD4 (allophycocyanin), CD14 (PerCP), and CD19 (PerCP) (all from BD Biosciences, San Jose, CA) were added. Unbound tetramer and antibodies were removed by two washes with PBA (PBS supplemented with 1% bovine serum albumin [BSA] and 0.02% sodium azide). Ten minutes before loading into the cytometer, 20 μl of Via-Probe solution was added in order to identify dead cells during analysis. Measurements were performed on an LSR-II flow cytometer (BD Biosciences, San Jose, CA), and data sets were analyzed using FlowJo software, v.9.1, using gates on CD4+, CD14−, CD19−, and Via-Probe-negative cells (viable cells).
Samples were enriched in tetramer-positive cells using magnetic beads essentially as described previously (17). Briefly, PBMCs were labeled and washed as described above and then incubated with 20 μl of anti-PE microbeads (Miltenyi Biotec Inc., Auburn, CA) at 4°C. After 15 min, the cells were centrifuged, washed with PBE (PBS-1% BSA-2 mM EDTA), resuspended in PBE, and loaded into Minimacs columns (Miltenyi Biotec Inc., Auburn, CA). Unbound cells were removed by washing, and bound cells (tetramer-positive cells) were recovered, centrifuged, resuspended in 200 μl of PBE, and analyzed after adding the cell viability tracker as described above.
As a first step to determination of the frequency of T cell responses to HHV-6, we measured the kinetics of production of various cytokines. PBMCs from 2 donors were stimulated with heat-inactivated lysates from HHV-6A-infected HSB-2 cells or with a control lysate of noninfected HSB-2 cells. Cell culture supernatants were collected at several time points during the next 120 h and analyzed by a multiplex assay for the presence of multiple cytokines. Supernatants of PBMCs incubated with lysate of HHV-6A (GS)-infected cells showed increased levels of multiple cytokines compared to control lysate. Figure 1A presents the profile for cytokines IL-2, IL-4, IL-10, TNF-α, and IFN-γ in cell culture supernatants between 48 and 120 h after stimulation. IL-2 and IFN-γ showed a high ratio of cytokine levels in HHV-6 versus control lysates for both donors. IL-10 also showed high ratios but in only one donor, and IL-4 was observed at considerably lower levels in both donors. TNF-α levels were not appreciably higher in HHV-6-infected cells than in control lysates. IL-17 was not detected, and consequently, responses to this cytokine are not shown. Two main kinetic patterns of cytokine production were observed. The first pattern was observed only for IL-2, where peak responses were detected at 48 h and the kinetics and levels of cytokine production were similar for the two donors. The second pattern was observed for all the other cytokines, where peak levels were variable for the 2 donors but were detected between 72 and 96 h. These data suggest that IFN-γ and IL-2 are consistent markers of the CD4 T cell response to HHV-6A and that IFN-γ may be the most sensitive marker. In the work described below, we use IFN-γ as a signature cytokine of the HHV-6 T cell response.
In order to estimate the number of CD4+ T cells responding to HHV-6, ex vivo IFN-γ responses in PBMCs were studied by intracellular cytokine staining (ICS) combined with staining for surface markers CD3 and CD4. Cells were stimulated with heat-inactivated lysate of HHV-6A-infected cells, control noninfected-cell lysate, or medium-only control, as in Fig. 1A, or with a purified preparation of HHV-6A (GS) virus that had been disrupted with mild detergent treatment. IFN-γ responses of CD4-positive CD3-gated T cells were observed after stimulation with either of the HHV-6 antigen preparations. Responses to purified virus preparation and to infected-cell lysate were approximately four times and two times the response to the noninfected cell lysate, respectively (Fig. 1B). Similar studies with 5 additional healthy donors (not shown) suggest a frequency of 0.04 to 0.12 CD4+ T cells responding specifically to HHV-6 viral antigens (n = 6; mean, 0.06 ± 0.07).
The low frequency of HHV-6-specific T cells detectable in PBMCs of healthy donors limits the analysis of the specificity and targets of the T cell response to HHV-6. To overcome this limitation, TCLs were generated from PBMCs of multiple donors by in vitro expansion with an enriched and detergent-disrupted preparation of HHV-6A (GS) virus. Figure 2A shows the kinetics of the cytokine production profile for TCLs expanded in vitro from four donors, with the same antigen and controls as those used in Fig. 1A. IL-2, IL-4, IL-10, and IFN-γ levels in supernatants of cells stimulated with HHV-6 antigens were consistently greater than those for mock antigen or medium, suggesting antigen-specific responses. Cytokine levels for cells stimulated with HHV-6 antigens were significantly higher for TCLs than for PBMCs, with the notable exception of IL-2. Peak responses were reached in most cases at 12 h, regardless of the donors and cytokine. In general, IFN-γ is the cytokine produced at the highest levels. These findings suggest that IL-4, TNF-α, IL-10, and particularly IFN-γ are appropriate markers to follow HHV-6-specific responses in in vitro-expanded TCLs. Finally, as in the case of PBMCs, IL-17 was not detected.
To address the possibility that these cytokine release responses might be the result of non-antigen-specific polyclonal activation, we performed the experiment shown in Fig. 2B. We raised two TCLs from donor SL131: one generated by expansion with an enriched preparation of disrupted HHV-6B (Z29) and the second one by mock expansion with medium only. Both TCLs were analyzed for their IFN-γ responses using ELISpot assay. As shown in Fig. 2B, there is a dose-dependent response to the HHV-6 preparation by the TCL raised to the virus but not by the mock-expanded TCL of the same donor, indicating a specific T cell response to HHV-6 viral antigens. A CD4+ T cell clone specific for a different viral antigen (HA1.7) (44) was not activated by incubation with HHV-6B (Z29), although it was activated by its cognate influenza virus-derived peptide antigen (HA pep), indicating that the HHV-6 preparation, at the concentration employed in our assays, does not contain substantial mitogenic or nonspecific activating components and that the cytokine responses that we observed are not the result of polyclonal activation by viral components but rather represent antigen-specific T cell responses.
In summary, we observed polyfunctional responses to HHV-6 antigens in PBMCs and at higher levels in TCLs expanded with HHV-6 antigens. Although the responses included several cytokines in both cases, it was notable that IL-2 responses were not significantly expanded in TCLs compared to PBMCs.
To characterize the specificity of T cells responding to HHV-6 antigens, we raised new TCLs by expansion with purified preparations of HHV-6A (GS) and HHV-6B (Z29) and also with a heat-inactivated lysate of HHV-6A (GS)-infected HSB-2 cells. These strains are representative of the two variants of this virus. We analyzed the responses of the TCLs to various antigens by intracellular IFN-γ staining combined with antibodies to CD4. A substantial fraction of each of the cell lines secretes IFN-γ in response to the antigen used for expansion, with little response to control lysate or medium, with most of the IFN-γ production limited to CD4+ T cells (Fig. 3A to F). The skewing toward CD4+ cells is expected since TCLs were generated with a disrupted virus preparation that would not be expected to efficiently access the class I MHC compartment to expand CD8+ T cells (68). CD4+ responses to HHV-6A (GS) lysate (Fig. 3A) and virus preparation (Fig. 3B) were 50-fold to several-hundred-fold over background. This response is substantially amplified relative to ex vivo responses from the same donors (compare with Fig. 1B, 2-fold to 4-fold over background). Reponses to the HHV-6B (Z29) virus preparation also were observed for both donors, 26-fold and 9-fold over background for donors SL131 and SL139, respectively (Fig. 3C). TCLs expanded with infected-cell lysate recognized purified virus (Fig. 3D) and vice versa (Fig. 3E), suggesting that the observed responses are in fact due to recognition of HHV-6A (GS) antigens, and not contaminants that might be present in the virus preparation or nonviral components present in the cell lysate. Finally, TCLs expanded with HHV-6B (Z29) were able to recognize antigens from HHV-6A (GS) (Fig. 3F).
CD4+ T cells can function as effector cells capable of killing infected cells, in addition to their usual function in cytokine secretion (4, 6, 13, 49, 50, 66). We investigated the cytolytic potential of the HHV-6-specific TCLs by measuring the expression of CD107a and CD107b, components of the lytic granule membrane that become surface accessible after degranulation (10, 43), in combination with surface staining for CD4 and a conventional IFN-γ intracellular staining assay. As can be seen in Fig. 3G, over 70% of the cells expanded with HHV-6A (GS)-infected HSB-2 cell lysate or purified HHV-6A (GS) viral preparation show expression of degranulation markers CD107a and -b after exposure to HHV-6 antigen. To determine whether the TCL subpopulations responding to HHV-6 by degranulation overlapped those responding by IFN-γ secretion, we used the gating strategy shown in Fig. 3H, whereby IFN-γ-positive cells are colored red on the CD107a/b-versus-CD4 dot plot. Percentages of CD4+ T cells expressing CD107a/b and capable of producing IFN-γ are shown as insets to the dot plots. This analysis shows that most of the HHV-6A (GS)-specific T cell population that is functional in producing IFN-γ also is capable of degranulation.
To identify the epitopes targeted by the HHV-6-specific CD4+ T cell response, we used a “bottom-up” peptide screening strategy (Fig. 4A). The large size of the HHV-6 genome, which codes for approximately 100 open reading frames and more than 40,000 potential epitopes, precludes a conventional systematic search of overlapping peptide series covering each protein encoded by the genome. Instead, we used a computational algorithm to predict which peptide sequences were likely to bind to HLA-DR1 (DRB1*0101), a common human MHC class II protein, and tested high-scoring peptides using PBMCs from HLA-DR1+ donors. Because of the broadly specific “promiscuous” peptide binding motifs characteristic of most human MHC class II proteins (65), we expect that many of the epitopes identified in HLA-DR1+ donors will be recognized by donors of other haplotypes, but only HLA-DR1+ donors were used to characterize HHV-6 T cell epitopes. Peptides predicted to bind to HLA-DR1 initially were screened by IFN-γ ELISpot assay in small pools using PBMCs, with positive pools deconvoluted to individual peptides using the same assay. Individual peptides inducing IFN-γ responses in at least one donor were considered candidate T cell epitopes and were analyzed further by ICS assays using PBMCs, TCLs expanded with HHV-6 viral preparations, and TCLs expanded with candidate peptides and also by competition binding to purified HLA-DR1. Similar approaches have been used recently to identify CD4+ T cell epitopes in other large-genome pathogens, including Plasmodium falciparum, Mycobacterium tuberculosis, HIV, and vaccinia virus (11, 22, 26, 54).
To predict HLA-DR1 (DRB1*0101) binding epitopes, we used the HHV-6B (Z29) sequence from the NCBI genome database (20). Each potential 9-mer binding frame was evaluated using a combination of two prediction algorithms, P9, based on HLA-DR1 binding preferences (35, 36), and SYFPEITHI, based on sequences of naturally processed peptides found bound to HLA-DR1 in cells (61), using a consensus approach as previously described (11). Overall, 42,838 possible 9-mer epitopes were evaluated (Fig. 4B). One set of peptides includes the top-scoring peptides from the entire translated viral genome, regardless of the protein source. Potential epitopes were picked using strict criteria with cutoff scores of ≥1.5 for P9 and ≥29 for SYFPEITHI. Each of these criteria separately selects about the top 1% of the scores. Together, they select 113 peptides, corresponding to the top 0.26% of all predicted 9-mer binding frames in the translated HHV-6B (Z29) genome.
Since two recent studies in vaccinia virus (11, 39) and one in HCMV (68) suggest that proteins included in viral particles are frequent targets of antiviral CD4+ T cells, we also included another set of peptides, limited to six proteins present in high abundance in the HHV-6B (Z29) virion: three tegument proteins (U11, U14, and U54), glycoprotein H (U48), major capsid protein (U57), and capsid binding protein (U65) (Table 1). A more complete analysis of the proteins present in HHV-6B (Z29) virions will be reported elsewhere (A. Becerra, J. M. Calvo-Calle, and L. J. Stern, unpublished data). Peptides were selected from this set of proteins using relaxed selection criteria, with cutoff scores of ≥2.9 for P9 and >17 for SYFPEITHI. Each of these separately selects the top quarter of all sequences. Together, they select 220 peptides, corresponding to the top 5% of 9-mer binding frames present in the six identified virion proteins.
Potential epitopes with high SYFPEITHI and P9 scores from the whole translated genome and from the six identified virion proteins were used to design 333 17-mer peptides in which the 9-mer predicted sequence was extended at each end by 4 flanking amino acids present in the protein sequence. Eleven peptides were present in both sets, for a total of 322 unique peptide sequences. All cysteine residues were replaced by serine, to prevent complications from oxidation or chemical modification during peptide synthesis and cell culture assay. The final set of peptides together with P9 and SYFPEITHI scores is provided as Table S1 in the supplemental material. (Peptides carrying cysteine-to-serine alterations are indicated with a lowercase “s.”).
We arranged the 333 peptides into 64 pools consisting of four to six peptides each (see Table S1 in the supplemental material) and tested these for responding T cell frequencies using PBMCs from three HLA-DR1 donors (Table 2) using a direct ex vivo IFN-γ ELISpot assay. Representative experiments are shown in Fig. 5A. Donor SL131 recognized 14 pools (pools 1 to 3, 8, 11, 17 to 19, 31, 32, 41, 42, 53, and 59), whereas donor SL139 recognized six pools. To identify particular peptides recognized in each peptide pool, PBMCs from responding donors were tested individually with single peptides present in each positive pool. An example of this analysis is shown in Fig. 5B for donor SL131. Overall, 11 peptides were found to induce an IFN-γ response in PBMCs from donors 085, SL131, and SL139 (Table 3).
To verify that the responses to synthetic peptides were associated with the desired peptide sequence and not with contaminants or synthetic by-products present in the preparations, we resynthesized and repurified all peptides showing significant IFN-γ ELISpot responses in PBMC assays. Subsequent in vitro and ex vivo assays described below were performed using these peptides.
A limitation of the studies conducted in PBMCs by ELISpot assays is the overall low frequency of antigen-specific T cells and the lack of information about the phenotype of responding cells. Since we have shown in Fig. 3 that in vitro treatment of PBMCs with our HHV-6 preparation results in significant expansion of antigen-specific responses, we followed the same idea and generated TCLs from multiple donors by expansion with HHV-6A and used these to corroborate the recognition of the 11 candidate epitopes (Fig. 6). In addition to the three DRB1*0101 donors described above, donors expressing DRB1*0104 (donor 098), DRB1*08 (donor 037), and DRB1*0301 (donor 066) alleles (Table 2) were included in this analysis. Data from a representative ELISpot assay are shown in Fig. 6A, and results from all experiments with these TCLs are summarized in Table 3. In general, every donor recognizes several peptides, with TCLs from donors carrying the DRB1*0101 haplotype showing responses to more peptides (6 to 10) than do TCLs from donors with related non-0101 DRB1 subtypes (4 peptides). Two peptides, 1162 and 1211s, both from the major capsid protein U57, each were recognized by TCLs from five out of six donors. Four peptides (1051, 1106, 020, and 083) were recognized by four donors, and three peptides (1017, 1066, and 1105) were recognized by three donors. For one peptide (1094s), TCLs from only a single donor exhibited IFN-γ responses above background, although for this peptide PBMCs from a different donor previously had been shown to be responsive.
To demonstrate that the IFN-γ responses observed by ELISpot assay in the virus-expanded TCLs were in fact produced by CD4+ T cells, we performed IFN-γ ICS combined with phenotypic marker CD4 and cytolytic marker CD107a/b. Figure 6B shows responses severalfold over the background in CD4+-gated T cells for peptides 1096, 1105, 1106, 1162, 1211s, 020, and 083s. Furthermore, most of the responding IFN-γ+ CD4+ T cells also mobilize large amounts of CD107a/b to the cell surface, suggesting a cytotoxic phenotype of these cells (Fig. 6B).
Expansion of TCLs with whole-virus preparations might select against low-frequency and/or slowly growing clonotypes present in PBMCs, with potential loss of some antigen-specific responses. In an attempt to detect such responses, we generated TCLs by stimulation of PBMCs from donors SL131 and SL139 with purified peptides rather than with disrupted whole-virus preparations. Figure 6C shows analysis of TCLs from donor SL131 raised by expansion with each one of the 11 candidate peptides. In vitro expansion of PBMCs with peptides allowed the visualization of CD4+ T cell responses to peptides 1051 and 1066 and a weak response to 1094s. For donor SL139, a similar analysis revealed responses to peptides 1017, 1051, 1066, and 1094s that were not observed in virus-expanded TCLs (Fig. 7).
To ensure that TCLs expanded with synthetic peptides are in fact specific for the intended peptide sequence, and not for synthesis by-products or contaminants that might be present in the purified peptide preparations, we also tested peptide-expanded TCLs from donor SL131 for their ability to recognize HHV-6 viral antigens, using antigen-presenting cells (APCs) pulsed with either disrupted HHV-6A (GS) (Fig. 8A) or a lysate of HHV-6A (GS)-infected cells (Fig. 8B). Recognition of whole-antigen-pulsed APCs was observed to some degree for all peptide-expanded SL131 TCLs tested as evidenced by the production of IFN-γ. These results suggest that peptide-expanded T cells recognize in fact epitopes generated by processing of HHV-6 proteins.
In summary, each of the 11 peptides identified in the original PBMC screening was recognized by virus-expanded TCLs from at least three of six HLA-DR1-family donors. In TCLs from two donors for which epitope-specific responses were examined in detail, the majority of the responding IFN-γ cells were CD4+ and CD107a/b.
HHV-6A and HHV-6B variants share a high degree of homology, reaching up to 99% in central regions of the genome that contain the core genes, but with less conservation and even variant-specific genes in the flanking regions (73). Epidemiological studies suggest that the HHV-6B variant is widely distributed in the world, infecting over 90% of people (46). The HHV-6A variant seems to be more geographically dispersed, with lower prevalence rates in Japan and rates similar to those of HHV-6B in Africa (31). In the experiments described above, we had observed some degree of HHV-6A/HHV-6B subtype cross-reactivity for TCLs raised against viral preparations (Fig. 3F) or individual peptides (Fig. 6C), and we wanted to investigate in more detail the potential for subtype cross-reactive responses to the identified epitopes. Our epitope prediction analysis and peptide testing used sequences from the translated HHV-6B (Z29) genome (20). We compared these sequences to the corresponding sequences in the translated HHV-6A (U1102) genome (32) (Fig. 9A). Of the 11 HHV-6B peptides for which we were able to demonstrate CD4+ T cell responses in multiple donors, five peptides (083s, 1017, 1096, 1162, and 1211s) have identical sequences in the two strains or differ only by the cysteine-to-serine alteration that we designed (083s and 1211s). The remaining six peptides (20, 1051, 1066, 1105, 1106, and 1094s) have HHV-6A versus HHV-6B sequence differences as shown in Fig. 9A. For three peptides (20, 1066, and 1106), these differences are outside the predicted MHC binding frame or immediately flanking residues. Of the other three, peptide 1051 differs at the predicted P1 anchor residue (I versus V), whereas peptides 1094s and 1105 have differences in the predicted binding frame and also in flanking regions. In order to study the functional implications of the variations in HHV-6A versus HHV-6B, TCLs raised against a purified preparation of HHV-6A (GS) (Fig. 9B) or against HHV-6B peptides (Fig. 9C) were tested for reactivity with autologous PBMCs pulsed with HHV-6A (GS) or HHV-6B (Z29) versions of the six epitope peptides with sequence differences. In each case, reactivity with both HHV-6A and HHV-6B variants was observed, suggesting a substantial degree of cross-reactivity in these peptide-specific responses, as previously observed for TCLs raised and tested against viral preparations (Fig. 3F).
The detection of T cell responses directly ex vivo would be a useful tool in tracking HHV-6-specific immune responses, for example in transplant patients or patients with clinical pathologies associated with HHV-6. Since we had demonstrated specific CD4+ T cell responses to the set of 11 predicted peptides (Table 3), it was important to measure the frequencies of these peptide-specific responses in PBMCs in order to establish basal responses in healthy donors. Therefore, we performed intracellular IFN-γ staining assays using freshly isolated PBMCs as a source of effector cells and autologous irradiated PBMCs as APCs. Representative responses from 3 healthy donors (SL131, SL139, and 085) to peptides 1017, 1096, 1106, and 20 are shown in Fig. 10A, with a summary of the response to all epitope peptides shown in Fig. 10B. In some cases, peptide-specific responses could be observed above background levels (peptides 1017, 1096, and 020) but the responses are present at very low frequencies, with the largest responses being only around 0.02% over background. Although these low frequencies of peptide-specific responses are consistent with the low frequency of responses observed to the whole virus (Fig. 1B), the difficulty in establishing a true background level and the lack of seronegative donors to use as negative controls lessen the utility of this assay in following epitope-specific T cell responses to HHV-6 in unexpanded PBMC samples.
In order to generate reagents that could be used to track epitope-specific T cell responses to HHV-6 in PBMCs from clinical samples, MHC class II HLA-DR1 tetramers were produced for the set of 11 CD4 T cell epitopes (Table 3). Positive- and negative-control HLA-DR1 tetramers also were produced, carrying immunodominant epitopes from influenza virus HA306–318 (44) and HIV-1 Gag167–182 (52), respectively. Representative staining of freshly isolated PBMCs from a healthy donor by three HHV-6-specific tetramers and one control tetramer is shown in Fig. 11A, with a summary of all tetramers tested shown as a bar graph at the right of the plots. We performed the staining experiment both directly on PBMCs (Fig. 11B, gray bars) and on samples that had been enriched for tetramer-positive cells using an antistreptavidin magnetic bead-based procedure (17, 62) (Fig. 11B, black bars). Without enrichment, the highest staining was observed for tetramer DR1-020 with a frequency of 1%, slightly over 3 times the frequency of the negative-control Gag tetramer. Among the remaining eight HHV-6 tetramers, only staining by DR1-1106 is observed above background frequency. After enrichment, the frequency of tetramer-positive cells increased for all but the negative-control DR1-Gag tetramer: 3.2 times for the DR1-HA tetramer, 7.7 times for DR1-1066, 20.9 times for DR1-1096, 8.5 times for DR1-020, and 3.3 to 29.8 times for the other tetramers. In order to corroborate the specificities of the tetramers, new TCLs were generated from two DR1 donors and used for side-by-side evaluation of their ICS IFN-γ and MHC tetramer staining. As shown in Fig. 11D, there is a significant overall correlation between tetramer staining and IFN-γ ICS. These results validate the specificity of the tetramer staining and suggest that upon enrichment it is possible to track with confidence HHV-6 epitope-specific CD4 T cell responses in samples of peripheral blood using MHC tetramers.
In the present study, we characterized in healthy adults the cellular immune responses to a variety of HHV-6 preparations and defined 11 immunodominant CD4+ T cell epitopes. These represent the first HHV-6-specific immunodominant epitopes defined for this virus. Using samples of peripheral blood, we found that the CD4+ T cell response to HHV-6 is characterized by a low frequency of cytokine-producing cells (<0.12%). We expanded the HHV-6-specific population using purified disrupted HHV-6A and HHV-6B preparations to reveal HHV-6-specific responses in T cell lines many orders of magnitude over the background response. Using these in vitro-expanded TCLs, we confirmed the polyfunctional cytokine and potential cytolytic nature of the anti-HHV-6 CD4+ response, and we evaluated the specificity and variant cross-reactivity in the CD4+ T cell response to the 11 HLA-DR-restricted epitopes. MHC tetramers carrying these epitopes were able to detect HHV-6-specific CD4+ T cells in samples of peripheral blood.
Characterization of CD4 T cell epitopes in HHV-6 is complicated by poor knowledge of the immune response to this virus, the low frequency of responding cells, the large number of proteins encoded by the HHV-6 genome (20, 32), and the lack of a small animal model (30). We approached the characterization of the T cell response to HHV-6 using a candidate epitope approach. Initial screening of synthetic peptides was performed with fresh PBMCs rather than expanded cell populations, to assess the full diversity of the HHV-6 T cell response and avoid potential skewing toward cells capable of preferential expansion in vitro. In spite of using fresh PBMCs and a relatively low number of peptides in the pools (4 to 6 peptides), we observed that for some pools eliciting positive responses we were not able to identify individual peptides that met our criteria for a positive response (Fig. 5, pool 53). This potential limitation of a bottom-up, peptide-based approach has already been highlighted in studies of the T cell response to HIV (57) and hepatitis C virus (HCV) (78). PBMC responses to candidate epitopes identified in the peptide screening were validated using TCLs raised against viral preparations and individual peptides. In order to obtain a representative sample of T cell responses in healthy adults, TCLs were generated from 6 donors, including the two used in the PBMC studies. The responses of virus-expanded TCLs were found to be variable between individuals. Each candidate epitope was recognized by at least one TCL, but a TCL from any single donor did not recognize all 11 peptides identified. In spite of differences in haplotype, 5 peptides were recognized by at least 4 of the 6 donors (1106, 1162, 1211s, 020, and 083s). Peptide-expanded TCLs generated from PBMCs of two donors recognized each one of the 11 candidate HHV-6 CD4+ T cell epitope peptides. These antipeptide TCLs showed consistent IFN-γ CD4+ T cell responses to both purified viral antigen and heat-inactivated lysate of HHV-6-infected cells, demonstrating that the corresponding epitopes are generated upon processing of the source protein and so represent immunodominant and not cryptic epitopes.
MHC tetramers are very useful reagents in following antigen-specific T cell responses ex vivo, and we were able to identify epitopes that could be used for tetramer staining in PBMCs. However, the low frequency of responding cells resulted in low signals relative to background staining unless tetramer-positive cells were enriched using magnetic beads prior to flow cytometry. A previous study has indicated that one of the benefits of enrichment prior to staining is the reduction of background staining (62). We observed this reduction in background staining as well. Tetramer staining signals were significantly above background for the HA and all the 11 HHV-6 tetramers postenrichment compared to only two tetramers before enrichment. However, comparison of the frequencies of tetramer-positive cells before and after enrichment indicates that the enrichment procedure introduces some skewing of the tetramer-positive population. Considering the frequency of tetramer-positive cells after enrichment, HLA-DR1-peptide complexes could be sorted into two well-delineated groups. The first group includes HLA-DR1-peptide complexes with 1096, 083s, and 020 with MHC tetramer staining frequencies of over 8% after enrichment. The relative affinities of peptides in this group for HLA-DR1 are high, with IC50 values below 1,500 nM (range, 80 to 1,500 nM). The second group, with lower staining frequency, includes the remaining eight HHV-6 complexes. Relative affinities in this second group are more variable, with an IC50 range from 220 nM in peptide 1105 to 30 μM in peptide 1106. These findings suggest that the tetramer-positive frequencies observed postenrichment could be a function of not only the preenrichment frequency but also the relative affinity of a peptide for HLA-DR1. The tetramer enrichment procedure potentially could be improved by using covalently attached peptides (16, 75), particularly for weaker-affinity binders. Also, it is possible that the tetramer enrichment technique might not be appropriate for T cell populations with relatively low functional avidities (62). Despite these caveats, MHC tetramer staining experiments using peptide-expanded TCLs demonstrated significant correlation between tetramer staining and IFN-γ ICS responses. Thus, HLA-DR1 tetramers carrying the HHV-6 epitopes identified here should be useful in following HHV-6-specific CD4+ T cell responses.
Overall, peptides derived from HHV-6 virion proteins appear to be recognized by CD4+ T cell epitopes more frequently than are peptides from virally encoded proteins in general. In our studies, 10 out of 220 (~4.5%) high-scoring virion peptides were identified as containing CD4+ T cell epitopes, compared to two out of 113 (~1.8%) high-scoring peptides from the translated genome as a whole. In a previous study of CD4+ T cell responses to vaccinia virus, we had observed that T cell epitopes were not found frequently among peptides with the very highest MHC binding prediction scores (nor among the peptides experimentally shown to have the very highest affinity) (11). One possible factor contributing to the epitope skewing that we observed might have been higher overall MHC binding predictions for the whole-genome set; however, the skewing is observed even when sets in the same prediction score range were compared (normalized combined P9-SYFPEITHI score, 0.54 to 0.67): with 2 out of 9 virion peptides recognized but only 1 out of 84 whole-genome peptides (Fig. 4B). Lower expression levels for proteins selected from the whole translated genome than for virion components do not appear to explain the observed skewing, since the whole-genome set included 19 peptides from six proteins abundantly expressed in persistent infection (U18, U20, U27, U85, U90, and U94) (53), none of which were identified as T cell epitopes. Finally, the observed epitopes derive from proteins exhibiting early, intermediate-early, and late expression patterns (71, 86), suggesting that temporal regulation of gene expression is not a critical factor for CD4 T cell immunogenicity in HHV-6. Thus, several factors suggested previously to regulate CD4+ T cell immunogenicity in other scenarios do not appear to explain the skewing toward virion components that we observe for long-term responses to HHV-6. It is possible that HHV-6 may employ immunoevasion strategies to limit generation of CD4+ T cell epitopes in infected cells, as it does for CD8+ T cell epitopes (29), perhaps employing strategies related to those reported for other herpesviruses (9, 45, 58).
One of the major problems in assessing the involvement of HHV-6 in human diseases is the lack of reagents that can be used to track immune responses to this virus, especially reagents that could differentiate between HHV-6 and HHV-7. A previous study that analyzed a large number of CD4+ T cell clones demonstrated that 70% of the clones respond to both HHV-6 and HHV-7 (87). Sequence analysis of the 11 epitopes reported here reveals that seven of these peptide sequences do not have significant homologies with other human pathogens, including HHV-7 (see Table S2 in the supplemental material). Thus, these epitopes should be useful in tracking immune responses specific for HHV-6, although it should be noted that it remains to be directly experimentally verified that reactivity to these epitopes is elicited upon infection/reactivation of HHV-6. The four remaining peptides, 083s, 1017, 1106, and 1211s, have significant homology with HHV-7 (see Table S2). HHV-6A but not HHV-6B is characterized by neurotropism (2, 21, 33), and so it would be advantageous to have reagents to corroborate if relapses in diseases like MS are associated with immune responses to HHV-6A rather than HHV-6B, as suggested by Fogdell-Hahn et al. (25). Sequence analysis of the 11 epitope peptides and cross-reactivity studies (Fig. 9) suggest that they might not fulfill this requirement. An epitope screening strategy focused on variable regions together with a fine mapping study could in principle provide immunological reagents to differentiate between responses to HHV-6 variants, although it would be preferable to have sequencing of a larger number of variant strains to select variant-specific sequences.
Overall, we found that IFN-γ-secreting CD4+ T cells responding to HHV-6 represented less than 0.1% of the total CD4+ T cell population. This can be contrasted to the related betaherpesvirus HCMV, for which much higher responding-cell frequencies, typically 2 to 6% of total T cells, were observed (55, 68). Like HHV-6, HCMV infection usually is asymptomatic with establishment of a lifelong chronic latency, and the continued presence of virus has been suggested to be important in maintaining high responding-cell frequencies. In our experiments, we observed relatively low levels of IL-2 compared to other cytokines in supernatants of PBMCs or TCLs challenged with inactivated HHV-6 viral preparations. Moreover, IL-2 levels did not increase substantially after expansion of HHV-6-specific subpopulations using heat-inactivated or detergent-disrupted (not shown) virus preparations, although levels of other cytokines were significantly increased (Fig. 1 and and2).2). Previous publications have demonstrated impaired IL-2 production and cell proliferation of PBMCs and the induction of IL-10 in cells infected with HHV-6 (24, 77). Consequently, it is reasonable to suggest that the combined effects of impaired IL-2 production and the induction of IL-10 by HHV-6 could limit the expansion of HHV-6 T cell responses. Despite the differences in responding-cell frequencies, the targets of the response are similar, with most HCMV-specific CD4+ T cells targeting abundant virion proteins (68), as we observed also for CD4+ T cells recognizing HHV-6.
In summary, we report that IFN-γ can be used as a surrogate marker of the cellular response to HHV-6. Using HHV-6 viral preparations, lysates of infected cells, and peptides, we observed very low frequencies of responding CD4+ T cells. The response mapped preferentially to peptides derived from abundant virion proteins. Although the T cell response to HHV-6 was much lower than that to the related betaherpesvirus HCMV, homologous proteins were targeted. Finally, we present seven virus-specific peptide sequences that could be used to track immune responses to HHV-6.
We gratefully acknowledge the blood donors and also Jaclyn Longtine, Karen Longtine, and Melissa O'Neill for assistance in obtaining blood samples; Liying Lu for the production of MHC tetramers; Dharam Ablashi and Steven Jacobson for assistance with HHV-6 cultures and sharing cell lines and virus stocks; and Alan Rothman for helpful discussions. The assistance of the UMass Medical School Core Flow Cytometry Lab and Proteomics and Mass Spectrometry Facility is acknowledged.
L.J.S. is a member of the UMass DERC (DK32520). Core resources supported by the Diabetes Endocrinology Research Center grant DK32520 were used. This work was supported by a grant from the NIH AI-U19-57319 (J.M.C.-C. and L.J.S.).
Published ahead of print 22 February 2012
Supplemental material for this article may be found at http://jvi.asm.org/.