Varicella Zoster Specific CD4⁺Foxp3⁺ T Cells Accumulate after Cutaneous Antigen Challenge in Humans

Abstract

We investigated the relationship between varicella zoster virus (VZV) specific memory CD4⁺ T cells and CD4⁺Foxp3⁺ regulatory T cells (Tregs) that accumulate after intradermal challenge with a VZV skin test antigen. VZV-specific CD4⁺ T cells were identified with a MHC class II tetramer or by intracellular staining for either IFN-γ or IL-2 after antigen re-challenge in vitro. VZV-specific T cells, mainly of a central memory (CD45RA⁻CD27⁺) phenotype, accumulate at the site of skin challenge compared to the blood of the same individuals. This resulted in part from local proliferation since >50% of tetramer defined antigen-specific CD4⁺ T cells in the skin expressed the cell cycle marker Ki67. CD4⁺Foxp3⁺ T cells had the characteristic phenotype of Tregs, namely CD25^hiCD127^loCD39^hi in both unchallenged and VZV challenged skin and did not secrete IFN-γ or IL-2 after antigenic re-stimulation. The CD4⁺Foxp3⁺ T cells from unchallenged skin had suppressive activity, since their removal led to an increase in cytokine secretion after activation. After VZV antigen injection, Foxp3⁺CD25^hiCD127^loCD39^hi T cells were also found within the VZV tetramer population. Their suppressive activity could not be directly assessed by CD25 depletion since activated T cells in the skin were also CD25⁺. Nevertheless there was an inverse correlation between decreased VZV skin responses and proportion of CD4⁺Foxp3⁺ T cells present indicating indirectly, their inhibitory activity in vivo. These results suggest a linkage between the expansion of antigen-specific CD4⁺ T cells and CD4⁺ Tregs that may provide controlled responsiveness during antigen-specific stimulation in tissues.

Introduction

The crucial role of CD4⁺CD25^hiFoxp3⁺ regulatory T cells in the maintenance of peripheral tolerance, protection against autoimmunity and the regulation of immunity to infections and tumours is well recognized (1). However, CD4⁺CD25^hiFoxp3⁺ Tregs are heterogenous and consist of cells that are a distinct lineage of thymic origin (reviewed in (2,3)) and also cells that may be induced from either naïve (4–6) or memory T cells (7–10) in the periphery. Foxp3⁺ Tregs, irrespective of how they are generated, have a similar phenotype(2,3)) however subsets of these cells may require different levels of antigen for activation and may have a relatively distinct dependence on cytokines for their expansion and homeostasis(11).

It has been suggested that during the course of an immune response in humans, CD4⁺CD25^hiFoxp3⁺ Tregs may be induced from responding memory T cells (7) and that this may provide a mechanism that links the extent of activation to the level of regulatory activity that is induced during an immune response (12). There is indirect evidence for this possibility because circulating CD4⁺Foxp3⁺CD127⁻ cells specific for cytomegalovirus (CMV) and other foreign antigens can be identified by the upregulation of CD154 following activation in vitro (13). In addition, cloning experiments have demonstrated the existence of Ag-specific Tregs in vitro (14–17). Furthermore using MHC class II tetramers it has been shown that antigen specific Tregs are induced to proliferate by low dose antigenic stimulation in vitro (11). However, neither of these studies addresses the kinetics, proliferative activity or antigen specificity of Foxp3⁺ Tregs during an immune response to antigen in vivo. In order to demonstrate an obvious relationship between Tregs and memory T cells it is crucial to show that both populations with the same antigenic specificity arise together during a memory response to an antigen in vivo.

In a previous study, we found that after a recall challenge with tuberculin purified protein derivative in the skin, both memory CD4⁺ T cells and CD4⁺FoxP3⁺ T cells increase in parallel during the course of the reaction (7). However, it was not clear whether both populations shared the same antigen specificity. To extend these observations we investigated the antigen specificity of CD4⁺Foxp3⁺ Tregs during a cutaneous immune response to varicella zoster virus (VZV) antigens in vivo using the HLA-DRB1*1501 IE63 epitope specific tetrameric complexes (18). VZV induces an acute self-limiting vesicular eruption followed by a persisting latent infection in humans(19). Anti-viral immunity largely involving VZV-specific CD4⁺ T cells usually controls viral reactivation (19). However, comprised immunity in immunosuppressed subjects or in older humans can lead to viral re-activation and shingles (20).

We found that VZV specific CD4⁺ T cells in healthy humans accumulate at the site of VZV antigen challenge in the skin, due in part to proliferation of these cells in situ. Furthermore we demonstrate the presence of tetramer-defined CD4⁺Foxp3⁺ VZV specific T cells that also proliferate in the skin during this recall response. Since this relationship inextricably couples immune activation and immune regulation, the alteration of this balance may lead to immune dysregulation such as that which is found in the skin of older humans after antigenic challenge in vivo (21).

Materials and methods

Subjects

This work was approved by the Ethics Committee of the Royall Free Hospital. Healthy individuals who had a history of previous chickenpox infection (n=94, median age = 32.5 years, age range 20–92 years, 38 male, 56 female) were recruited for the study. All volunteers provided written informed consent and study procedures were performed in accordance with the principles of the declaration of Helsinki. Individuals with history of neoplasia, immunosuppressive disorders or inflammatory skin disorders and individuals on immunosuppressive medication were excluded.

Skin tests

Delayed type hypersensitivity responses (DTH) were induced by intradermal injection of antigen on non-sun exposed skin of the medial proximal volar forearm. Varicella Zoster Virus (VZV) skin test antigen from The Research Foundation for Microbial Diseases of Osaka University (BIKEN) was a kind gift of Professor Michiaki Takahashi, Osaka University. VZV skin-test antigen, was licensed in 1990 in Japan and contains viral glycoproteins prepared from the culture fluid of VZV (attenuated Oka parental strain) infected MRC-5 cells as described previously (22,23). 0.1ml dose was used as described previously(23) and no adverse effects have been observed with its use. Induration, palpability, and the change in erythema from baseline were measured to generate a clinical score (0–15) as described previously (24). The leukocytes at the injection site were investigated by immunohistochemical analysis of skin biopsies or by flow cytometric evaluation of leukocytes isolated from skin suction blisters that were induced at the site of injection at an indicated time point between 0 and 7 days after the skin test injection as described (21). Mantoux test reactions were induced in small number of healthy BCG (Bacille Calmette-Guerin) vaccinated volunteers by the intradermal injection of 2U of tuberculin purified protein derivative (PPD) (Statens Serum Institut, Copenhagen, Denmark).

Skin biopsies

Punch biopsies (5mm diameter) from the site of antigen injection were obtained from 15 young volunteers at various time-points (day 1, day 3 and day 7) post-VZV skin test injection. Control skin punch biopsies from normal non-injected forearm skin (n=6) were also obtained. Biopsies were frozen in OCT (optimal cutting temperature compound; Bright Instrument Company Ltd). 6µm sections were cut and left to dry overnight and then fixed in ethanol and acetone and stored at −80°C. For functional analysis of skin cells 5mm punch biopsies were digested overnight with 0.8mg/ml of collagenase IV (Sigma) as described (25).

Preparation of suction blister cells and PBMC preparation

Skin suction blisters were induced by the application of a negative pressure of 25–40 kPa (200–300 mmHg) below atmospheric pressure via a suction chamber for 2–4 h using a clinical suction pump (VP25; Eschmann) until a unilocular blister measuring 10–15 mm in diameter was formed. Suction blisters were raised over the sites of VZV skin test injection or normal skin 18–24 h before sampling to ensure maximum cell recovery. The blister fluid was microcentrifuged at 650g for 4 min to pellet the cells present. The pellet was resuspended in complete medium (RPMI GIBCO, BRL Life Technologies) containing 10% human AB serum, 100 U/ml penicillin, 100 µg/ml streptomycin, and 2 mM l-glutamine (all obtained from Sigma-Aldrich). Heparinized blood was collected at the time of blister aspiration or as specified. PBMCs were prepared by density centrifugation on Ficoll-Paque (Amersham Biosciences) and re-suspended in complete medium.

Immunofluorescence

6µm skin sections were blocked with Dako non-serum protein block for 20 minutes. Primary antibodies were incubated for 1 hour at room temperature and amplified with the appropriate secondary antibody: goat anti-mouse IgG1 conjugated with Alexa Fluor 546 or Alexa 488 (1hr at room temperature). For CD4 and Foxp3 staining, skin sections were incubated with primary antibodies (biotin anti-human Foxp3 and mouse anti- human CD4) overnight at 4°C, followed by incubation with strepCy3 and anti-mouse IgG1 Alexa Fluor 488. For Ki67 staining, skin sections were incubated overnight at 4°C with primary antibodies, (CD4 biotin, and Ki67 FITC, both Becton Dickinson) followed by strepCy3. Slides were then washed twice in PBS and mounted with Vectashield containing DAPI (Vector lab). Images were acquired using appropriate filters of a Leica DMLB microscope with Leica N PLAN 20× /0.40 objective and a Cool SNAP-Pro cf Monochrome Media Cybernetics camera, controlled by Image-Pro PLUS 6.2 software. When counting the numbers of cells in perivascular infiltrates, at least 5 largest perivascular infiltrates present in the upper and mid dermis of each section were counted (7). Cell numbers were expressed as the mean absolute cell number per perivascular infiltrate.

Flow Cytometric Analysis

Multi-parameter analysis of blister and blood T cell phenotype was performed on a FACSCalibur™ or LSR II cytometer (Becton Dickinson) as previously described. PBMCs or blister cells were stained with different combinations of antibodies including CD3, CD4, CD8, CD45RA, CD28, CD27 (all from Becton Dickinson) Ki67, Foxp3, CD25, CD127 and CD39. All surface staining was performed for 30 minutes on ice. Intracellular Ki67 staining (clone B56, Beckton Dickinson) was performed by intracellular staining to identify cells in all stages of cell cycle (23). Foxp3 staining was performed with either anti-human Foxp3-APC (clone 3G3, Miltenyi Biotec) or Pacific blue anti-human Foxp3 (clone 206D, Biolegend). Foxp3 Staining Buffer Set (Miltenyi Biotec) was used according to manufacturer’s instructions. In cases where Ki67 and Foxp3 were used together, the Foxp3 staining protocol was used.

Tetramer staining

DRB1*1501 iTAg MHCII tetramer was purchased from Beckman Coulter. DRB1*1501 tetramer was complexed to VZV IE63 peptide 24 (QRAIERYAGAETAEY). CLIP peptide (PVSKMRMATPLLMQA) was used as a control and tetramer staining was performed as previously described (18). Briefly, cells were first incubated with 2µg/ml HLA Class II tetramers for 1–2 hours at 37°C, in the dark, washed in PBS and then stained for different surface markers as needed. We analysed the tetramer expression within the CD4⁺ T cell subset by gating on the lymphocytes and excluding B cells, monocytes and dead cells (Via-Probe positive population). In Foxp3 and Tetramer co-staining experiments, standard Foxp3 protocol was used following tetramer staining. Cells were analyzed on LSR II or BD Fortessa using FACS Diva software (both Becton Dickinson) and further analyzed using FlowJo software (TreeStar, Inc). For analysis of mycobacteria specific cells DRB1*0301 iTAg MHC class II tetramers (Beckman Coulter), were complexed either to the mycobacterial Ag85A 20mer peptide, (VPSPSMGRDIKVQFQSGGAN (DR3-Ag85A)), or the human Apolipoprotein B-100 peptide (ISNQLTLDSNTKYFHKLN, (DR3-ApoB)) as a control tetramer(26).

Intracellular cytokine staining

Cells prepared from blisters or peripheral blood were stimulated with VZV lysate (Virusys corproration) or CMV lysate (1:10 dilution), and incubated for 15 hours at 37°C in a humidified 5% CO2 atmosphere. 5 µg/ml Brefeldin A (Sigma-Aldrich) was added after 2 hours. Unstimulated controls were also included. Following stimulations cells were first stained for surface markers (CD4, CD8 and a combination of additional differentiation markers (CD27, CD28, CD45RA, CCR7) for 30minutes at 4oC. After washing, cells were fixed and permeabilised (Fix & Perm Cell Permeabilisation Kit, Caltag Laboratories) before staining for IL-2 and IFN-γ. In experiments where Foxp3 and cytokine staining was performed in parallel, Foxp3 staining protocol was used according to manufacturer’s instructions.

Suppression assay

Cells prepared from collagenase digested skin were separated into CD25 enriched (CD25⁺ fraction) and CD25 depleted (CD25⁻ fraction) using CD25 microbeads and a standard MACS separation protocol (Miltenyi biotec). Non depleted skin cells, CD25⁻ skin cells and CD25⁻ cells mixed with CD25⁺ cells at 1:2 ratio were incubated with CD3/CD28 beads (GIBCO, BRL Life Technologies, 2:1 bead to cell ratio) for 18hrs in the presence of Brefeldin A as standard. Following stimulation cells were stained for surface markers, Foxp3 and cytokines. Foxp3 staining allowed us to gate on equivalent populations in all samples.

Statistics

Statistical analysis was performed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, California, USA). Non-parametric tests were predominantly utilised as the data was not normally distributed. The Kruskall-Wallis test was used to compare three or more unpaired groups and a 2-tailed Mann-Whitney test was used when comparing only two unpaired groups. The Wilcoxon matched pairs test or a paired t-test was used when comparing two groups of matched data. Correlation between clinical score and proportion of Tregs in the skin was assessed by linear regression analysis.

Results

CD4⁺ effector T cells and CD4⁺Foxp3⁺T cells accumulate following intradermal challenge with antigen

A secondary cutaneous immune response was induced by the intra-dermal injection of VZV antigen in healthy individuals who had a previous history of chickenpox. The clinical manifestation of this response is measured by the induction of erythema, induration and palpability at the injection site, which peaked at between 48–72 hours. By day 7 after induction, this reaction is significantly decreased compared to the peak response (p<0.036; data not shown). Skin punch biopsies taken at day 0, 1, 3 and 7 following VZV injection were stained for the expression of CD4 and Foxp3 (Fig. 1A). We found that CD4⁺ T cells accumulated around blood vessels after antigen injection (Fig. 1A, left panels, blood vessels indicated by asterisks). The five largest perivascular infiltrates per section were photographed and counted, and data expressed as mean absolute cell number per frame. The numbers of CD4⁺ T cells increased from day 1 onwards and peak between day 3 and day 7 (white squares). Therefore the peak of the clinical response (2–3 days) and the peak of CD4⁺ T cell infiltration (day 7) after VZV challenge in the skin do not coincide. However the number of infiltrating CD4⁺ T cells that were found at the peak of the cellular response after VZV challenge correlates positively with the clinical score at 48 hours (Fig. 1C; p<0.0001). This suggests that the early inflammatory response after VZV injection is essential to enable the accumulation of CD4⁺ T cells that occurs later.

Figure 1. CD4⁺ T cells and CD4⁺Foxp3⁺ T cells accumulate in parallel after VZV challenge in the skin.

Healthy young volunteers were injected with 0.02ml VZV skin antigen test and 5mm punch biopsies were performed on days 0, 1, 3 or 7 post injection (with 4–7 volunteers per time point). (A) 6µm frozen skin sections were stained for CD4 (green) and Foxp3 (red) using an indirect immunofluorescence method (original magnification: x400). Cell numbers were expressed as the mean absolute number of cells counted within the frame. Graph shows numbers of total CD4⁺ cells (white squares) and numbers of CD4⁺Foxp3⁺ cells (black squares) in perivascular infiltrates (PV) following VZV skin antigen injection. (B) Graph shows the proportion of total CD4⁺ cells expressing Foxp3 in perivascular infiltrates before and after VZV challenge. Each symbol represents a mean of n=4–7 individuals per time point; average of 5 perivascular infiltrates counted for each individual. (C) The number of infiltrating CD4⁺ T cells in the skin correlates with clinical response (linear regression analysis, p<0.0001).

We found a significant increase in the number of CD4⁺Foxp3⁺ T cells following VZV injection (p =0.0061, Kruskal Wallis test, Figure 1A, black squares) which coincides with the increase in total CD4⁺ T cell numbers. Therefore, when the data is expressed as the proportion of CD4⁺ T cells expressing Foxp3, this value remains fairly constant during the response at around 8–15% of CD4⁺ cells, Figure 1B). This suggests that memory and Foxp3⁺ (putative regulatory) T cells accumulate at the same rate after a recall response to VZV in vivo.

Accumulation of VZV-specific CD4⁺ T cells after cutaneous VZV antigen injection

We next investigated the specificity and phenotype of skin infiltrating T cells after VZV antigen challenge. To do this we induced skin suction blisters at the site of antigen injection, at the peak of cellular response at 7 days post VZV injection. CD4⁺ T cells harvested from skin suction blisters at this time were tested for their ability to synthesize IFN-γ and/or IL-2 after overnight re-stimulation with VZV lysate as previously described (21,27). We found a significantly higher proportion of IFN-γ⁺ VZV-specific CD4⁺ T cells in the skin compared to the blood of the same individuals after VZV injection (Figure 2A; Wilcoxon paired test, p=0.0002). Similar results were observed for IL-2 secreting cells (Figure 2B, Wilcoxon paired test, P=0.001). Furthermore, a large proportion of VZV specific CD4⁺ T cells synthesized both cytokines (mean 40±15%, (range 19–62%) of all cytokine producing cells, Figure 2C)). Minimal IFN-γ or IL-2 was produced by blood or skin CD4⁺ T cells in the control cultures without added VZV antigens or when the cells were stimulated with irrelevant antigens (Figure 2A-C, and data not shown), However CD4⁺ T cells isolated from the skin showed significantly increased background secretion of both cytokines compared to the blood population (pooled data for IFN-γ or IL-2 p<0.014), possibly reflecting endogenous cytokine production of cells that were activated in situ (Fig. 2 A,B). Very low levels of IL-17 and IL-10 producing cells were observed following VZV lysate stimulation of CD4⁺ T cells (data not shown), suggesting that the response to VZV injection is predominately driven by Th1 cells.

Figure 2. Cellular infiltration at the site of VZV injection is Ag-specific.

Skin suction blisters were induced on day 7 following VZV injection. (A) PBMCs and blister cells were stimulated with VZV lysate for 15hrs in the presence of Brefeldin A and stained for intracellular expression of IFN-γ and IL-2. Representative FACS plots are shown on the left. The percentage IFN-γ producing antigen specific CD4⁺ T cells was significantly increased compared to peripheral blood (n=15 for IFN-γ; Wilcoxon paired test ***P=0.0002, *P=0.015). (B) IL-2 producing antigen specific CD4⁺ T cells were also significantly increased compared to peripheral blood (n=8, Mann Whitney test **P=0.001, *P=0.016). (C) A Large proportion of VZV specific cells produce both cytokines. Representative dot plot shows day 7 blister cells (gated on CD4⁺) secreting IL-2 and IFN-γ. (D) The number of antigen specific CD4⁺ cells was also determined by class II tetramer staining. Blister cells and PBMC were stained with HLA-DRB1*1501 restricted IE63 tetramer (18). Left panel shows representative FACS plots for tetramer staining. Graph shows cumulative data, each data point represents one individual, horizontal bars represent the mean (n=11, Wilcoxon paired test **P=0.0046).

In addition, we used a VZV specific Class II tetramer to further investigate the characteristics of the T cells in the skin after antigen challenge. Leukocytes that were present in skin suction blisters and PBMC were collected from the same individuals on day 7 following VZV antigen injection and stained with HLA-DRB1*1501 restricted IE63 tetramer (18). The representative tetramer staining is shown in Figure 2D (left panels) and the data confirmed the presence of a large proportion of VZV-specific CD4⁺ T cells in the skin compared to the blood of healthy subjects (Fig. 2D, right panel, p<0.005, Wilcoxon paired test). We confirmed that the tetramer staining was specific by showing a lack of staining when a control tetramer (CLIP) was used and by demonstrating a lack of tetramer staining when HLA DRB1*1501-negative individuals were tested (Supplementary Fig. 1).

A higher proportion of VZV-specific CD4⁺ T cells (4.5 ± 1%, mean ± SEM) were identified using cytokine secretion compared to class II tetramer (2.7 ± 0.6%, mean ± SEM). While this confirms that the IE63 peptide contained within the tetramer is immunodominant, it also suggests that additional VZV specific CD4⁺ T cells of different specificities that are not recognized by the tetramer are also present within the skin after challenge. The exact proportion of VZV specific cells at the site of the immune response is difficult to ascertain. The proportion of cytokine secreting cells is likely to be an underestimate of the VZV specific cells, firstly because cells recovered from the site of the acute response may be refractory to further activation in vitro and secondly because it is possible that not all VZV specific cells have the capacity for IL-2 and/ or IFN-γ production. Another possibility is that these cells may be inhibited by Tregs that are present at the site of the response. Nevertheless it is clear that a significant proportion of infiltrating cells are specific for the antigen used in the cutaneous challange (21,27)

Are VZV specific CD4⁺ T cells in the skin highly differentiated?

We investigated the differentiation state of total and VZV-specific CD4⁺ T cells in the skin after VZV challenge using the relative expression of CD45RA, CD27 and CD28. In a previous study we have used both CD27 and CCR7 together with CD45RA to identify subpopulations of CD4⁺ T cells(28). We found that the subsets identified by CD27 and CCR7, were very similar but not totally overlapping, in agreement with many other groups (29). We chose to use CD27 instead of CCR7 together with CD45RA as it gives a better discrimination between all the 4 subsets identified, enabling more accurate gating. IFN-γ or IL-2 production was used to identify VZV-specific CD4⁺ T cells in blisters induced on day 7. We found that both total and VZV-specific CD4⁺ T cells found in the skin are predominantly CD45RA⁻CD27⁺ and CD45RA⁻CD27⁻ (central and effector memory cells respectively (Figure 3). Similar phenotype was observed when tetramer staining was used for identification of VZV-specific cells. Tetramer⁺ CD4⁺ cells in the skin were predominantly CD45RA⁻CD27⁺ central memory cells (data not shown).

Figure 3. VZV-specific CD4⁺ T cells in the skin predominantly have a central memory phenotype.

Skin suction blisters were induced on day 7 following VZV injection. PBMCs and blister cells were stimulated with VZV lysate for 15hrs in the presence of Brefeldin A and stained for intracellular expression of IFN-γ, IL-2, CD27 and CD45RA. (A) Representative FACS plots show expression of CD27 and CD45RA on total CD4 or IFN-γ secreting CD4⁺ T cells in the skin suction blisters and blood. Graph shows cumulative data (n=12) for total CD4 (black symbols) and IFN-γ⁺ CD4⁺ T cells (VZV- specific, white symbols) in the skin. (B) Representative FACS plots and cumulative data showing expression of CD27 and CD45RA on total and IL-2 secreting CD4 T cells (n=7).

On the basis of CD27 and CD28 expression, both total and VZV-specific skin infiltrating CD4⁺ T cells they are predominantly CD27⁺CD28⁺ and are therefore not very differentiated (data not shown). VZV-specific cells found in the blood also have similar phenotype (Figure 3). This is in agreement with other studies describing VZV-specific CD4⁺ T cells in the circulation, (18,28).

VZV specific CD4⁺ T cells proliferate in the skin following VZV injection

The accumulation of VZV-specific CD4⁺ T cells in the skin after antigen injection could be due to preferential infiltration of these cells from the blood and/or proliferation of these cells in the skin. To address this we investigated the proliferation of CD4⁺ T cells, using Ki67 expression as a marker of cells in cell cycle (7) in tissue biopsies that were taken after VZV injection (Fig. 4A). Very few Ki67⁺ CD4⁺ T cells were observed in normal skin (day 0, not shown), and the perivascular infiltrates were very small or absent before antigen injection (see Fig. 1). At day 7 post challenge, Ki67 expression was significantly increased and many CD4⁺ T cells in the skin were in cycle (Figure 4A, p=0.0051 Kruskal Wallis test).

Figure 4. Antigen specific memory CD4⁺T cells proliferate at the site of the VZV injection.

6µm frozen skin sections were stained for CD4 and Ki67 by double immunofluorescence method (original magnification: x400). (A) Photographs show representative staining on days 3 and 7 post-injection (Ki67 is green, CD4 is red). Graph shows percentage of CD4⁺ cells expressing Ki67 found per perivascular infiltrate in each donor. Each symbol represents an average of 5PVs counted for each individual (n=4–5 per time point, line indicates the mean, p=0.0051, Kruskal Wallis test). (B) Skin suction blisters were induced on day 7, in young individuals who were HLA-DRB1*1501 positive (n=4). PBMC and blister cells were stained with CD4, tetramer and Ki67 and proliferation of CD4⁺ tetramer⁻ and CD4⁺ tetramer⁺ T cells was compared. Dot plot shows representative tetramer staining of blister cells and the Ki67 staining of the two gated populations is shown in the histograms. Cumulative data is shown in the bar chart (n=4, p=0.02 paired t-test). (C) CD4⁺Foxp3⁺ T cells proliferate at the site of VZV injection. 6µm skin sections were stained with Ki67 (green) and Foxp3 (red). 5 largest PVs were counted for each individual (n=4–5 per time point). Photographs show representative day 3 and day 7 staining (original magnification x400, inset shows an area of interest at higher magnification). Bar chart shows cumulative data (white bars) in comparison to total CD4⁺ T cells (black bars).

We also investigated the proliferation of VZV-specific CD4⁺ T cells, identified by tetramer staining of cells that were isolated from skin suction blisters at day 7 after injection (Fig. 4B) Proliferation was observed in both the tetramer positive and tetramer negative CD4⁺ T cells in the skin. This suggests that in addition to T cells that are specific for the immunodominant IE63 peptide, other VZV specific cells that may be responding to additional VZV peptides are activated and proliferate in the skin after challenge. Nevertheless there was higher proliferative activity in tetramer population compared to the negative cells supporting the imunodominance of these cells (p=0.02, paired t-test). These data suggests that the increase in VZV specific CD4⁺ T cells in the skin after antigenic challenge occurs in part through their extensive local proliferation, however we do not rule out the possibility that some pre-activated VZV specific CD4⁺ T cells may also be recruited from the blood.

To examine whether CD4⁺Foxp3⁺, putative regulatory cells also proliferate at the site of cutaneous VZV challenge, biopsies obtained from the site of injection were co-stained for Ki67 and Foxp3 (Figure 4C, left panels). We observed that CD4⁺Foxp3⁺ cells proliferated to a similar extent as the CD4⁺ effector T cells after VZV challenge (Figure 4C, right panel). This is in agreement with previous observations when a different antigen, PPD, was injected in the skin (7).

CD4⁺Foxp3⁺T cells display a Treg phenotype

Human CD4⁺ T cells can transiently upregulate Foxp3 after activation; however, these recently activated T cells do not express a typical CD25^hiCD127^loCD39^hi Treg phenotype (7,30,31). Furthermore, typical Foxp3⁺ Tregs do not secrete cytokines after stimulation (7,32). We therefore examined the phenotypic and functional characteristics of the CD4⁺ T cells that were isolated from skin suction blisters 7 days after VZV injection (Fig. 5A). We found that CD4⁺Foxp3⁺ cells express very high levels of CD25, low levels of CD127 and high levels of CD39, which corresponding to the classical Treg phenotype (Figure 5A). Furthermore CD4⁺Foxp3⁺ T cells produced virtually no IL-2 (not shown) and considerably less IFN-γ and TNF-α when compared to CD4⁺Foxp3⁻ T cells cytokines after VZV activation in vitro (Fig. 5B). Interestingly the most cytokines were produced by CD4⁺Foxp3⁻CD127^lo cells which represent the memory population(33). In additional experiments, we also tested for the production of IL-17 and IL-10, but neither cytokine was produced at significant level by CD4⁺ T cells after VZV injection (not shown).

Figure 5. CD4⁺Foxp3⁺ T cells have a regulatory phenotype.

(A) Skin suction blisters were induced on day 7 following VZV injection. Blister cells and PBMC were stained in parallel and analysed by multi parameter flow cytometry. First, live cells were gated on either CD4⁺Foxp3⁺ or CD4⁺Foxp3⁻ (left dotplot). Representative expression of CD25, CD127 and CD39 is shown in the histograms (n=6). (B) Day 7 blister cells were stimulated o/n with VZV lysate and IFN-γ production was measured by intracellular cytokine staining. CD4⁺ population was gated based on Foxp3 and CD127 expression and the subsets (as indicated) compared in terms of cytokine production. Representative dot plots of n= 5 are shown. Quadrants were set on the isotype control and unstimulated cells (not shown). (C) The proportion of Foxp3⁺CD4⁺ T cells in the skin inversely correlates with the clinical score (each symbol represents one individual, linear regression analysis p=0.007). Representative immunofluorescence staining from individuals with low or high clinical score are shown. The clinical score on day 3 following injection is plotted against the % of Foxp3 expressing CD4⁺ cells found in the skin at the peak of the cellular response (mean of 5 largest PVs counted for each individual).

In order to determine whether CD4⁺Foxp3⁻ T cells that were found in the skin after VZV challenge were actually suppressive in vivo, we compared the proportion of CD4⁺ T cells expressing Foxp3 at the peak of the cellular response (day 7 after VZV injection) to the clinical score. There was a significant inverse correlation between the clinical score and the proportion of CD4⁺Foxp3⁺ T cells during the immune response (Fig. 5C, right panel, p=0.007). This provides indirect support for the possibility that the CD4⁺ Foxp3⁺ T cells that are found in the skin after VZV challenge are suppressive and regulate the extent of the response. This raises the question about the antigen-specificity of this regulatory population.

CD4⁺Foxp3⁺ T cells in the skin include VZV -specific Tregs

In order to determine whether the regulatory Foxp3⁺ T cells observed following VZV challenge are antigen specific we used the class II tetramer to identify VZV-specific CD4⁺ T cells in the skin of HLA-DRB1*DR1501⁺ individuals after VZV injection. In the majority of DRB1*DR1501⁺ individuals tested we could identify tetramer positive CD4⁺Foxp3⁺ T cells (Fig. 6A). The proportion of tetramer⁺ cells in the CD4⁺Foxp3⁺ population was very similar to that seen in the CD4⁺Foxp3⁻ T cell subset suggesting that both populations may arise in parallel (Fig. 6A). In a small number of individuals we investigated the presence of antigen specific Tregs following an intradermal challenge with tuberculin-purified protein derivative (PPD; Mantoux test, (MT)). Suction blisters were induced on day 7 following MT induction and using a DRB1*0301 restricted class II Ag85A tetramer (mycobacterial Ag85A 20mer peptide, from Mycobacterium tuberculosis) we could identify mycobacteria specific CD4⁺ T cells in the skin. In addition, a small proportion of Tetramer⁺Foxp3⁺ cells was also seen (Supplementary figure 2), and these cells had regulatory phenotype (CD25⁺CD127^low). This indicates that the accumulation of antigen specific Foxp3⁺ T cells in the skin is not unique to stimulation with VZV.

Figure 6. CD4⁺Foxp3⁺ T cells in the skin include VZV specific Tregs.

Skin suction blisters were induced in HLA-DRB1*1501 individuals on day 7 following VZV injection. (A) Representative dot-plot showing class II tetramer staining of blister cells gated on either total CD4⁺ cells (left panel) or CD4⁺Foxp3⁺ cells (right panel). Percentage of tetramer positive cells in each gate is indicated in the top quadrant. Graph shows cumulative data from n=13 showing percentage of total CD4⁺ T cells and CD4⁺Foxp3⁺ cells staining with the class II tetramer in the blister fluid. (B) Day 7 blister derived CD4⁺ T cells were subdivided into 4 subsets based on Foxp3 and tetramer staining. Representative dot plots show CD25 and CD127 staining on CD4⁺Foxp3⁻Tet⁺ and CD4⁺Foxp3⁺Tet⁺ populations. Cumulative data for the CD4⁺Foxp3⁺Tet⁺ subset is shown in the graph. Horizontal bar represents the mean (n=5).

We next investigated whether the VZV-specific Foxp3⁺ T cells had a typical Treg phenotype. In 5 individuals tested we found that a mean of 60.6±11.81% (mean±SEM) of the VZV-specific CD4⁺Foxp3⁺ T cells in the skin were CD25^hiCD39^hiCD127^lo. Therefore while the majority of VZV specific CD4⁺Foxp3⁺ T cells have a Treg phenotype, a proportion of these cells may represent a recently activated population of specific T cells that may be in the process of differentiation into a regulatory population.

Suppressive activity of CD4⁺Foxp3⁺ T cells in normal skin

We wanted to confirm that the Foxp3⁺ Tregs found in the skin following VZV challenge have suppressive function. However we found that their suppressive activity could not be directly assessed by CD25 depletion since there was a large proportion of activated T cells in the skin that were also CD25⁺ (Fig 5A). The possibility of FACS sorting based on CD25 and CD127 expression was also limited as the cell numbers recovered from skin blisters/ biopsies were very small (in the region of 50–300,000 cells). Therefore we investigated the function of CD4⁺Foxp3⁺ cells in normal skin where CD25 expression is confined to the Foxp3⁺ population and where larger skin samples could be used. Firstly, we demonstrate that Foxp3⁺ T cells in normal skin have the same phenotype as Foxp3⁺ T cell from the site of VZV induced immune response (Figure 7A). Secondly, we purified CD25⁺ cells from normal skin using MACS beads and performed stimulation in vitro with antiCD3/ anti CD28 beads to measure cytokine production. TNF-α production was compared from CD25 depleted skin cells with and without the addition of the CD25⁺ fraction. As shown in Figure 7B, TNF-α production was significantly reduced when CD25⁺ cells were added back, supporting the notion that Foxp3 expression in the skin identifies regulatory T cells. Taken together with the observation that there is a strong inverse correlation between the proportion of CD4⁺Foxp3⁺ T cells in the skin and clinical responses to antigen (Fig 5C and (21)) this provides indirect evidence that Foxp3⁺ T cells accumulating in the skin following VZV-challenge have suppressive activity in vivo.

Figure 7. CD4⁺Foxp3⁺ T cells in the skin have regulatory function.

Normal skin biopsies were collected from healthy individuals or form discard surgery skin and cells were isolated by enzymatic digestion. (A) Skin cells and PBMC were stained in parallel and analysed by multi parameter flow cytometry. Live cells were gated on either CD4⁺Foxp3⁺ or CD4⁺Foxp3⁻ (left dotplot) and representative expression of CD25, CD127 and CD39 is shown in the histograms. (B) Skin cells were fractionated into CD25⁺ and CD25⁻ fraction using MACS beads. CD25 depleted cells and mixed CD25⁻/CD25⁺ cells (1:2 ratio) were stimulated o/n with antiCD3/antiCD28 and TNF-α production was measured by intracellular cytokine staining. Dotplots are gated on CD4⁺Foxp3⁻ cells to allow a comparison between the two samples.

Discussion

The induction of an immune response has to be balanced with mechanisms that prevent excessive effector function and alterations of this balance may lead to immunopathology (1). Conversely, an excessive manifestation of inhibitory activity may hinder the generation of effective immunity. An important goal therefore is to clarify the relationship between effector and regulatory CD4⁺ T cells and thus to understand how a successful but controlled immune response can be achieved. In a previous study we suggested that the increased numbers of CD4⁺Foxp3⁺ T cells that are found in the skin of older humans before and after recall antigen challenge may lead to a defective immune response in this organ (21). This is in agreement with the observations in murine cancer model systems where the increase in Tregs around the tumour leads to inadequate anti-tumour responses that can be improved by removal of these cells (34). The present study was therefore designed to clarify the relationship between VZV-specific effector/memory and regulatory T cells, identified with a MHC class II restricted tetramer, in the skin of healthy humans during the response to a recall antigen.

We found that following intradermal challenge of healthy humans with VZV antigen, a significant proportion (1–8%) of all CD4 T cells at the peak of cellular infiltration may be specific for an immunodominant epitope (peptide 24) of the IE63 protein (18). This viral component is present during different phases of the viral life cycle and it is thought that IE63 specific CD4⁺ T cells are important for control of viral reactivation (35). Furthermore, the accumulation of the VZV-specific T cells is due in part to proliferation at the site of cutaneous antigen challenge. However we have not been able to define whether the VZV specific CD4⁺ T cells are recruited initially from the blood or whether they are derived from VZV-specific memory T cells that are resident in the skin.

A study of cutaneous infection with HSV-1 in mice showed that CD8⁺ HSV-specific cells that migrated to the site of antigen exposure, remain resident in the skin following antigen-clearance (36,37). These long-lived skin resident cells provide a rapid response to subsequent re-exposure to the same pathogen, presented by tissue resident DC and could contribute to the initial steps of activation and to the initiation of the inflammatory cascade(36,37). Similar observations were made in mice that were challenged with Vaccinia virus where skin resident CD8⁺ T cells were shown to be relatively long lived (6 months) and important for the initial response to viral re-activation(38). Interestingly different patterns of migration were noted by CD4⁺ compared to CD8⁺ T cells. Following the resolution of infection in the HSV skin infection model, virus specific CD8⁺ T cells were sequestered in the epidermis and slow moving while HSV-specific CD4 trafficked rapidly through the dermis (39). Furthermore it is also not clear how skin resident CD4⁺ regulatory T cells are involved in these responses. As VZV re-activation in humans has a strong skin manifestation (shingles), the existence of skin resident memory cells to this virus would be analogous to the murine models described above. However, since the activation of endothelium in the skin is essential for an optimal recall response to antigen to occur (21), it is likely that the recruitment of T cells from blood also has an important role in skin immunity. It is possible that skin resident antigen specific T cells are involved in the initiation of the cutaneous response that is subsequently amplified by the recruitment and proliferation of antigen specific T cells from the blood following antigenic challenge (40).

The Foxp3⁺ Tregs that are found during an antigen response in the skin may be recruited from the circulation as they express high levels of the skin homing receptors CLA and CCR4 (41,42). Alternatively, the accumulation of CD4⁺CD25^hiFoxp3⁺Tregs in the skin after antigen challenge may result from the proliferation of skin resident memory CD4⁺FoxP3⁺ T cells generated during primary infection with VZV as resident Tregs have been identified in normal human skin (7,43,44). A third possibility is that some CD4⁺CD25^hiFoxp3⁺Tregs may be derived from responsive memory CD4⁺ T cell populations during the immune response as suggested previously (7,12,45). However these possibilities are not mutually exclusive. Previous studies have shown that naïve CD4⁺Foxp3⁻ can differentiate into Foxp3⁺ regulatory T cells under the influence of TGF-beta(46,47), retinoic acid (48) and other cues (reviewed in (2)), there is only indirect evidence that effector or memory T cells can be induced to become Tregs(7–10). Many of the mechanisms that have been proposed in other studies including presence of TGF-β, DCs expressing B7H1 and B7DC(49), langerin⁺ dermal dendritic cells(6), RANK/ RANKL signalling (50) could be compatible with the skin microenvironment and this is under current investigation. Furthermore, other studies have suggested that extensive proliferation is involved in the induction of iTregs (6,10,11,51). When we examined Ki67 expression in Tetramer+ and Tetramer- Tregs it was the Tetramer⁺Foxp3⁺ subset which was the most proliferative (not shown).

The presence of VZV specific Tregs may be relevant to the maintenance of immunity to this virus during ageing. It is well recognized that older humans are susceptible to VZV re-activation that results in shingles(19). Furthermore, older humans(21,52,53) and mice (54) have increased numbers of CD4⁺Foxp3⁺ Tregs in many organs including the skin. If VZV specific CD4⁺Foxp3⁺ Tregs comprise the increased population of Tregs in the skin this may suppress the immune response to virus reactivation and shedding from nerve endings in the skin that leads to shingles. This requires further study.

In conclusion, our current study shows that VZV-specific memory T cells and Foxp3⁺ Tregs specific for the same antigen accumulate in parallel at the site of specific antigen challenge in vivo, suggesting that Tregs may be derived from memory T cells during a localized site of an immune response in vivo. A crucial question that remains is the specificity of the CD4⁺Foxp3⁺ Tregs that are increased in normal skin of older humans (21) and whether the influence of these inhibitory cells may directly hinder effective cutaneous immunity to pathogens such as VZV.

Abbreviations used

VZV

varicella zoster virus

DTH

delayed type hypersensitivity

Treg

regulatory T cells

MT

Mantoux test

PPD

purified protein derivative

CMV

cytomegalovirus

HSV

herpes simplex virus