Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Transcriptional Analysis of T Cells Resident in Human Skin

  • Jane Li ,

    jane.melb@gmail.com (JL); jzma@unimelb.edu.au (JZM)

    Affiliations Department of Microbiology and Immunology, The University of Melbourne at The Peter Doherty Institute for Infection and Immunity, Parkville, Victoria, Australia, Department of Medicine (St Vincent’s Hospital), The University of Melbourne, Fitzroy, Victoria, Australia

  • Moshe Olshansky,

    Affiliation Department of Microbiology and Immunology, The University of Melbourne at The Peter Doherty Institute for Infection and Immunity, Parkville, Victoria, Australia

  • Francis R. Carbone,

    Affiliation Department of Microbiology and Immunology, The University of Melbourne at The Peter Doherty Institute for Infection and Immunity, Parkville, Victoria, Australia

  • Joel Z. Ma

    jane.melb@gmail.com (JL); jzma@unimelb.edu.au (JZM)

    Affiliation Department of Microbiology and Immunology, The University of Melbourne at The Peter Doherty Institute for Infection and Immunity, Parkville, Victoria, Australia

Transcriptional Analysis of T Cells Resident in Human Skin

  • Jane Li, 
  • Moshe Olshansky, 
  • Francis R. Carbone, 
  • Joel Z. Ma
PLOS
x

Abstract

Human skin contains various populations of memory T cells in permanent residence and in transit. Arguably, the best characterized of the skin subsets are the CD8+ permanently resident memory T cells (TRM) expressing the integrin subunit, CD103. In order to investigate the remaining skin T cells, we isolated skin-tropic (CLA+) helper T cells, regulatory T cells, and CD8+ CD103- T cells from skin and blood for RNA microarray analysis to compare the transcriptional profiles of these groups. We found that despite their common tropism, the T cells isolated from skin were transcriptionally distinct from blood-derived CLA+ T cells. A shared pool of genes contributed to the skin/blood discrepancy, with substantial overlap in differentially expressed genes between each T cell subset. Gene set enrichment analysis further showed that the differential gene profiles of each human skin T cell subset were significantly enriched for previously identified TRM core signature genes. Our results support the hypothesis that human skin may contain additional TRM or TRM-like populations.

Introduction

Human skin at steady state contains a vast number of memory T cells [1]. Traditionally, memory T cells have been divided into two populations: central memory T cells (TCM) that circulate mainly between the lymphoid tissues and effector memory T cells (TEM) that migrate to extralymphoid peripheral tissues [2]. TCM and TEM are distinguished by the expression of CCR7 and CD62L, or lack thereof (TCM−CCR7+CD62L+, TEM−CCR7-CD62L-), and both may be found in normal human skin [1]. Recently, a subset of CD8+ T cells has been discovered that resides permanently in peripheral tissues post-infection, without returning to the circulation [35]. These T cells provide accelerated long-lived site-specific immunity and have been termed resident memory T cells (TRM) [3,5,6]. TRM are generally defined by surface expression of CD103 (αE integrin) and CD69 but lack of CCR7 and CD62L, and have been described in both mice and humans in many non-lymphoid tissues such as gut, brain, lung, skin and genital mucosa [3,711]. Since their discovery, CD8+CD103+ TRM have been studied extensively. Microarray analyses in mouse models have identified the transcriptomes of these CD8+CD103+ TRM in several tissues, including skin [7,12], demonstrating that these TRM are a separate subset distinct from TCM and TEM.

Apart from CD8+CD103+ TRM, skin contains other TRM, as well as a heterogeneous population of recirculating memory T cells (TRCM) comprising TEM, TCM, and other subsets yet to be described in detail [13,14]. TRCM presumably recirculate between blood and skin through the expression of skin addressins such as cutaneous lymphocyte antigen (CLA), CCR4 and CCR10 [15,16]. Studies in murine skin have found CD4+CCR7+ TRCM with effector functions more akin to TCM than TEM [14], and CD4+ regulatory T cells (Treg) which reversibly traffic between skin and blood [17]. Interestingly, these experiments also identified a subset of CD4+CD103+CCR7- T cells that did not reenter the circulation, suggesting that the skin may also harbour CD4+ TRM [14]. A comparable complexity appears to exist in human skin. In a study of patients with cutaneous T cell lymphoma treated with the monoclonal antibody alemtuzumab, which depletes circulating T cells but spares TRM, both CD8+ and CD4+ T cells, including Treg, persisted in the skin [13]. Thus, the present literature indicates that skin contains multiple T cell subsets, some of which have yet to be fully defined.

We sought to further characterize human skin TRM and TRCM by undertaking a gene expression microarray analysis of skin-tropic memory T cells in blood compared to non-CD8+CD103+ T cells in the skin. We reasoned that T cells in skin would comprise both TRM and TRCM, while the skin-tropic memory T cells in blood would comprise only TRCM. Our aim was to identify a gene expression “signature” that distinguished cutaneous CD8+ T cells, CD4+ T cells and Treg from their blood equivalents. A secondary aim was to compare the transcriptional profile of these skin T cell groups with the currently known core signature of CD8+CD103+ TRM in mouse models. We showed that skin-tropic T cells derived from skin and blood had distinct patterns of gene expression, with a shared pool of genes contributing to the skin/blood discrepancy. We also found that the human skin T cells were significantly enriched for established TRM core signature genes compared to human blood T cells.

Materials and Methods

Tissue sample collection and pooled cell suspension preparation

The IMMGEN protocol (http://www.immgen.org) was consulted in the design of this microarray experiment. Peripheral blood mononuclear cells (PBMC) were obtained from 15 healthy donors (age range 17–72) and human skin samples were obtained as surgical discard from 15 healthy volunteers (age range 18–64). All donors were female to avoid gender-based disparities. The University of Melbourne human ethics committee approved this study and patients provided written informed consent. PBMC were acquired as cryopreserved samples from the Victorian Cancer Biobank or Australian Red Cross Blood Service. PBMC samples from donors aged 17 were from the Australian Red Cross Blood Service. Written consent was obtained directly from these donors and not from the next of kin as the minimum age for blood donation (parental consent not required) in Australia is 16 years.

Fresh skin samples were incubated in collagenase type 3 (3mg/ml; Worthington) with 5μg/ml DNAse in RPMI medium containing 2% (v/v) FCS for 1.5 hours at 37°C. Collagenase-digested tissue samples were then dissociated using the Becton Dickinson (BD) Medimachine into cell suspensions. To reduce the effect of inter-individual variability, pooled cell suspensions of PBMC or skin cells were obtained by combining samples from 3 donors. Cells were pooled such that the cumulative age of the donors was similar across all pools. The pooled cell suspensions were subjected to negative magnetic selection using the Dynabeads Untouched Human T cells Kit (Thermo Fisher Scientific) to deplete contaminating B cells, NK cells, monocytes, platelets, dendritic cells, granulocytes and erythrocytes.

Fluorescence activated cell sorting and antibodies

Pooled cell suspensions were stained with a cocktail of fluorescence-conjugated antibodies for 30 minutes at 4°C. BD FACSAria III was used to collect live sorted cells at the Melbourne Cytometry ImmunoID Flow Cytometry facility (The University of Melbourne). The live cell populations collected are shown in S1 Table and representative gating is shown in Fig 1. Antibodies used from BD were: anti-CD8a (RPA-T8), anti-CD3 (UCHT1), anti-CLA (HECA-452), anti-CD25 (M-A251). Antibodies from eBioscience were: anti-CD4 (OKT4), anti-CD127 (eBioRDR5), anti-CD103 (BerACT8; used for skin samples only). Anti-CD45RO (UCHL1) antibody was obtained from Biolegend. To further insure against non-T cells contaminating the sort, a dump channel was created to exclude cells binding PerCP-Cyanine5.5-conjugated antibodies (all obtained from Biolegend) targeting CD138 (epithelial cells; used for skin samples only), CD235a (erythrocytes), CD14 (monocytes), CD19 (B cells), CD335 (NKp46; NK cells). Propidium iodide was used as a viability marker.

thumbnail
Fig 1. Multiple subsets of skin-tropic T cells are present in human skin and blood.

Representative images are shown to demonstrate gating strategy for fluorescence activated cell sorting of T cells from blood and skin. In each case, gates were used to exclude debris, doublets and non-viable cells, and a dump channel was used to exclude irrelevant cell types. (a) For blood samples, T cells were identified based on CD3 expression, then divided into CD4+ and CD8+ populations. Skin-tropic CD8+ and CD4+ memory T cells were isolated based on their expression of memory marker CD45RO and skin addressin CLA. Skin-tropic regulatory T cells (Treg) were identified based on their CD25hiCD127lo surface profile. (b) For skin samples, CD45 was used to eliminate cells of non-haematopoietic origin. CLA+ T cells were identified based on CD3 expression. CD8+CD103- T cells were captured; CD4+ T cells (CD127hi) and Treg (CD25hiCD127lo) were identified on the basis of CD25 and CD127 expression. FSC = forward scatter; PI = propidium iodide; SSC = side scatter.

http://dx.doi.org/10.1371/journal.pone.0148351.g001

RNA extraction, whole transcriptome amplification and microarray hybridization

After sorting, RNA was extracted using the Qiagen RNEasy Micro kit according to manufacturer’s instructions, then stored immediately at -80°C. RNA quality analysis, amplification, microarray hybridization and data acquisition were performed at the Ramaciotti Centre for Genomics at the University of New South Wales. In brief, RNA quality was assessed using a Bioanalyzer 2100 (Agilent). RNA was then used to generate complementary DNA (cDNA) through whole-transcriptome amplification using the Ovation PicoSL WTA V2 linear RNA amplification system (NuGEN). The cDNA was then labeled with the Encore Biotin module (NuGEN) and hybridized to HumanHT 12v4 BeadArrays (Illumina) prior to data acquisition.

Microarray data analysis

Data was extracted from the raw intensity files from the Illumina HumanHT 12v4 BeadArrays using Bioconductor R limma package [18]. Bioconductor illuminaHumanv4.db package [19,20] (reannotation) was used to exclude bad probes and to accurately assign probes to transcripts. Due to high sample variability a standard Differential Expression analysis (like in limma package) did not produce satisfactory results, therefore RUVinv method (from ruv R package [21]) was used to generate log2transformed fold-change and corresponding p-values for pairwise comparisons [22,23]. A log2fold-change cut-off of ≥ 1.5 and a p-value cut-off of ≤ 0.05 after Benjamini-Hochberg false discovery rate multiple testing correction were used to identify significantly differentially expressed transcripts. Microarray data was submitted to the Gene Expression Omnibus (accession code GSE74158).

The PANTHER (protein analysis through evolutionary relationship) Classification System [24] version 10.0, 2015 (http://pantherdb.org) was used for gene ontology (GO) analysis. The PANTHER Statistical overrepresentation test with GO-Slim Biological Process annotation data was used; GO terms with p≤0.05 after Bonferroni correction were deemed significant. Pathway analysis was conducted with PathVisio version 3.2.1 software [25] using curated biological pathways from WikiPathways [26]. Gene sets for Gene Set Enrichment Analysis (GSEA) were obtained of genes upregulated and downregulated in murine skin, lung and gut TRM [7] (for complete lists see S2 Table). GSEA was performed on log2transformed and quantile normalized data using GSEA v2.2.0 software from the Broad Institute [27]. Principal component analysis was performed on the normalized data and visualized using XLSTAT (Addinsoft SARL, US).

Quantitative PCR

Quantitative (q)PCR was performed on the StepOne Plus system (Thermo Fisher Scientific) using 1–3 cDNA samples from each category of sorted cells (blood CD4+ = 3 samples; skin CD4+ = 1 sample; rest = 2 samples). Taqman Gene Expression Assays (Thermo Fisher Scientific) were obtained for NR4A2 (Hs00428691_m1), S1PR1 (Hs00173499_m1), CCR7 (Hs01013469_m1), CD8A (Hs00233520_m1), CD4 (Hs01058407_m1) and CTLA4 (Hs03044418_m1), as well as UBC (Hs00824723_m1) and B2M (Hs00984230_m1), with the latter two used as housekeeping genes. Raw gene expression data was normalized to the geometric mean of housekeeping gene expression using the 2-ΔCт comparative CT method [28].

Results

Memory T cells in the skin share a common gene expression signature

We identified and isolated skin-homing memory T cells from human blood and skin based on the expression of CLA (Fig 1). The proportion of CLA+ memory T cells (CD8+, Treg and CD4+) present in our samples was consistent with previous studies, constituting 80–90% of CD3+ cells in the skin, but only ~15% of CD3+ cells in the blood [1]. Gene expression profiles of these CD8+, CD4+ and Treg CLA+ memory T cells in the skin and blood were examined through the use of microarray gene chips. Principal component analysis (PCA) was used to delineate variations in the transcriptional data. The PCA showed that the transcriptomes of the memory T cells segregated according to the tissue of origin on PC1 and 2 (73.4%; Fig 2A), whereas the variations due to T cell lineage were less prominent, as represented by PC18 and PC21 in blood (1.15%; Fig 2B). These results suggested that the magnitude of transcriptional dissimilarity between skin and blood T cells outweighed the gene expression differences between CD8+, Treg and CD4+ T cells.

thumbnail
Fig 2. T cells in human skin are transcriptionally distinct from skin-tropic T cells in the blood.

Principal component analysis of gene expression of sorted skin and blood CD4+, CD8+ and regulatory T cells. Graphs plotted showing (a) principal components 1,2 and 3 and (b) principal components 18 and 21. Each symbol represents one array; 5 arrays/cell type. (c) Venn diagram showing overlap of the significantly differentially expressed genes (DEGs, listed in S3 Table; P≤0.05) in pairwise comparison between skin and blood CD4+, CD8+ and regulatory T cells using RUVinv analysis. (d) Venn diagram of DEGs between T cell lineages in blood and skin (listed in S4 Table; P≤0.05) showing overlaps between categories. Numbers in brackets indicate total number of DEGs from each pairwise comparison. Treg, regulatory T cells.

http://dx.doi.org/10.1371/journal.pone.0148351.g002

To determine which genes contributed to the transcriptional dissimilarities, we first performed pairwise comparisons between each skin Treg, CD4+ and CD8+ group and their blood equivalent. We found 80–100 significantly differentially expressed genes (DEGs) in each case (for complete lists see S3 Table). There was substantial overlap in DEGs, with a shared group of 24 genes that were differentially expressed in all three T cell types (Fig 2C and Table 1). As blood CLA+ T cells are known to be CD69- [1], the significant upregulation of CD69 (Table 1) provided some reassurance that cross-contamination between skin and blood T cells was minimal.

We then compared CD8+ to CD4+ T cells and Treg to CD4+ T cells in skin and, separately, in blood to determine if there were also shared DEGs between T cell lineages independent of tissue origin. In contrast to the previous analysis between skin and blood T cells, there were ~40 DEGs identified in these pairwise comparisons (S4 Table), and these DEGs rarely overlapped between categories (Fig 2D). The difference in the number of DEGs identified between comparing cells from different sources and lineages reaffirmed the PCA findings, in that the environment the T cells are in greatly influences their transcriptional profiles, with some contributions from their developmental programs.

Validation of microarray data using quantitative real-time PCR

We sought to validate the DEGs identified in the microarray data analysis using qPCR. Due to limited specimen availability, we could only test a few DEGs in the remaining samples from each T cell subset. In concordance with the microarray data, T cells derived from skin showed higher expression of NR4A2 compared to those from blood, whereas S1PR1 expression was higher in T cells from blood than skin (Fig 3A). In the limited number of samples tested, CCR7 showed a trend of increased expression in CD8+ T cells from blood compared to skin (Graph A in S1 Fig). Different T cell subsets from the skin and blood also showed elevated expression of their respective markers, CTLA4, CD8A and CD4 (Fig 3B and Graph B in S1 Fig).

thumbnail
Fig 3. Skin and blood T cell subsets demonstrate differential gene expression by quantitative PCR.

(a) qPCR was performed for selected genes differentially expressed between skin- and blood-derived T cells (combined CD8+, CD4+ and regulatory T cells) as identified on microarray analysis (Table 1 and S3 Table). Expression values were normalized to the geometric mean of two housekeeping genes. (b) Expression of lineage-related genes in the CD8+, CD4+ and regulatory T cells (combined blood and skin samples). *P≤0.05; **P≤0.01 using two-tailed Mann-Whitney U-test.

http://dx.doi.org/10.1371/journal.pone.0148351.g003

Functional analysis suggests that skin T cells are active and responding to stress signals

Next, gene ontology (GO) terms were assigned to the genes that were upregulated and downregulated in skin T cells compared to blood counterparts, so as to identify their functions. GO terms were successfully matched to 75 of the 81 genes that were upregulated in skin T cells compared to blood T cells. Subsequent statistical analysis of GO terms enrichment showed that several biological processes were significantly overrepresented: protein folding, response to stress, apoptotic process, cell death, death and immune system process (Fig 4; for detailed results see S5 Table). Conversely, amongst the 115 genes that were upregulated in the blood T cells compared to skin, there were no significantly overrepresented biological processes found (data not shown). These results suggested that the skin T cells were poised to respond to danger signals and were in a more activated state than their blood equivalents.

thumbnail
Fig 4. Skin T cells upregulate genes in immune system and stress response processes.

Gene ontology analysis of transcripts upregulated in skin compared to blood T cells. Significantly overrepresented biological processes (PANTHER GO-Slim annotation; P≤0.05) in skin T cells shown plotted against Enrichment Score (calculated as -log10p-value).

http://dx.doi.org/10.1371/journal.pone.0148351.g004

To investigate how the products of these gene transcripts might act within the cells, we performed pathway analysis on the T cell subsets by mapping our microarray results to relevant biological pathways found in the WikiPathways repository [26]. This pathway analysis was intended to provide a broad visual overview of the transcripts involved in a biological process, even though not all genes were necessarily significantly differentially expressed by our earlier stringent criteria. We found that the expression of genes involved in the Cytokines and Inflammatory Response pathway [29] (WP530) was skewed toward skin T cells (Fig 5). Similarly, many of the genes in the Oxidative Stress pathway [30] (WP408) were upregulated in skin T cells (Fig 6). Combined, the increased transcription of inflammatory cytokines and stress response-related genes provides additional evidence that skin T cells are indeed in a highly activated state.

thumbnail
Fig 5. CD4+ T cells in the skin are transcriptionally active for cytokine and inflammatory response genes.

Graphic representation of relative gene expression between skin CD4+ T cells and skin-tropic T cells derived from blood. Cytokines and inflammatory response pathway (WP530) depicted using PathVisio v3.2.1 and WikiPathways. Red-green colour bar denotes magnitude of log2fold-change (green = upregulated in skin, red = upregulated in blood). Similar results were found in CD8+ T cells and for regulatory T cells (data not shown).

http://dx.doi.org/10.1371/journal.pone.0148351.g005

thumbnail
Fig 6. CD8+ T cells in the skin overexpress genes in the oxidative stress pathway.

Graphic representation of relative gene expression between skin and blood CLA+ CD8+ T cells in the Oxidative Stress pathway (WP408) using PathVisio 3.2.1 and WikiPathways. Similar results were found for CD4+ T cells and for regulatory T cells (data not shown). Red-green colour bar denotes magnitude of log2fold-change (green = upregulated in skin, red = upregulated in blood). TF = transcription factor.

http://dx.doi.org/10.1371/journal.pone.0148351.g006

Transcriptional profiles of skin T cells are significantly enriched for TRM core signature genes

Finally, we sought to compare the human skin T cell transcriptome to the core signature of CD8+ TRM as established in the mouse [7]. The various skin T cell types loosely conformed to the gene expression pattern of the murine TRM core signature, with all significant DEGs corresponding exactly to the TRM profile (Fig 7A). Further analysis using GSEA was performed on the genes typically up- and downregulated in murine lung, skin and gut TRM compared to circulating TEM and TCM (S2 Table). Surprisingly, we found that the transcriptional profiles of human skin CD4+ T cells, Treg and CD8+CD103- T cells were all individually significantly enriched for genes upregulated in murine TRM compared to the human skin-tropic T cells in blood (Fig 7B). In keeping with this result, the T cells derived from human blood were significantly enriched for genes downregulated in murine TRM (S2 Fig).

thumbnail
Fig 7. Human skin T cell transcription profiles are enriched for signature resident memory T cell genes defined in the mouse.

(a) The fold change of various genes in the murine skin TRM core signature in human blood versus skin CD4+, CD8+ and Treg cells on microarray analysis. Green (upregulated in TRM) and red arrows (downregulated in TRM) below indicate the expected direction of expression in TRM. Asterisks indicate significantly differentially expressed genes (P≤0.05; n = 5 arrays per cell type). (b) Enrichment scores for the various skin T cell types following Gene Set Enrichment Analysis using gene set lists containing lung, gut and skin TRM gene signatures. All gene sets shown are significantly enriched at false discovery rate <25%.

http://dx.doi.org/10.1371/journal.pone.0148351.g007

We were interested in ascertaining which genes in our T cell samples were most important in accounting for their similarity to the murine TRM gene signatures. To identify the subset of genes that had the greatest contribution to the enrichment scores [27], we performed a leading edge analysis on the GSEA datasets. We found that there were many shared genes between the leading edge subsets for CD8+CD103-, Treg and CD4+ T cells (Table 2; for complete leading edge subset lists see S6 Table). Among these genes, we identified aryl hydrocarbon receptor (AHR), which is a critical determinant in enabling T cells to persist in the skin [31]. Taken together, our results suggest that while in the skin, CD4+, Treg and CD8+CD103- T cells acquire a similar gene expression profile to that of TRM, thereby potentially allowing them to reside long-term within skin.

thumbnail
Table 2. Shared genes of human skin T cells and murine TRM signature with greatest contribution to the gene set enrichment.

http://dx.doi.org/10.1371/journal.pone.0148351.t002

Discussion

Normal human skin contains both TRM and TRCM. The best-characterized TRM are intraepithelial CD8+CD103+ T cells, although recently CD4+ TRM have also been described, and the necessity of CD103 expression has been disputed [13,14,32,33]. In this paper, we analyzed the transcriptomes of CD8+, Treg and non-Treg CD4+ T cells bearing the skin-homing marker CLA in the blood to the same populations in the skin. In essence, we compared groups of T cells containing both TRM and TRCM (skin) to groups composed exclusively of TRCM (blood). We showed that skin-derived T cells possess a gene signature different from blood-derived T cells and this signature is largely conserved in both humans and mice.

Several key genes that were differentially expressed between skin and blood T cells are associated with tissue persistence and cell activation status. For example, S1PR1, which mediates T cell egress [34], was significantly downregulated in CD4+ and CD8+CD103- cutaneous T cells, although fell short of significance in cutaneous Treg. In addition, skin T cells displayed higher expression of the early activation marker CD69 and upregulated genes involved in the stress response and immune process pathways. Further comparison of human skin T cells to the common transcription profile of murine CD8+CD103+ TRM found in various murine tissues revealed that the expression pattern of murine TRM signature genes was largely conserved in human skin CD8+CD103-, Treg and CD4+ cells, and that a shared group of genes accounted for their likeness to murine TRM, including potential lineage transcription factors associated with murine TRM.

There are a number of ways to interpret our findings. The first is that the differences we observed between skin and blood T cells could be, wholly or partially, due to changes induced by the different microenvironments. It may be that gene expression is altered so that circulating T cells can enter and take up residence in the skin, with the initial changes then triggering downstream effects on cell behavior and effector function [34]. Moreover, since healthy skin contains numerous commensal microbes [35] extending even into the dermis [36], there would conceivably be a constant supply of antigens to stimulate T cell activation. These stimuli are not usually present in the circulation, which could result in a difference in activation status. The implication of this model is that T cells leaving the skin would eventually lose the skin-specific gene expression signature. Interestingly, in vivo tracking of skin TRCM in the Kaede transgenic murine model demonstrated that TRCM displayed the same surface phenotype for days after leaving the skin [14]. However, as surface markers may not accurately reflect changes at the gene expression level, further investigation matching the gene expression in TRCM derived from blood and skin is required to definitively exclude this possibility.

Another possible interpretation is that each of the CD8+CD103-, Treg and CD4+ T cell groups in human skin contain substantial amounts of TRM, or cells that are developing into TRM, thereby accounting for the differences in the gene expression profiles. This interpretation would certainly be compatible with the situation found in patients with cutaneous T cell lymphoma (CTCL) treated with alemtuzumab, where a similarly diverse range of T cells were spared from depletion and remained in the skin post-treatment [13]. If this is the case, the DEGs identified between the skin and blood T cells would be representative of the transcription profile of the TRM population within skin. Taken together with the results from the pathway, gene ontology and principal component analyses, our findings imply that significant functional distinctions exist between TRM and TRCM within a T cell lineage [4,37], which are potentially as important as the differences between the various T cell lineages themselves.

Finally, an intriguing hypothesis that can also be drawn from our studies, is that TRM and TRCM are two ends of a spectrum of extralymphoid memory T cells, with a number of intermediate phenotypes that could reside in skin for varying periods of time. Their time in residency may even depend on their strength of expression of the TRM core signature. The existence of intermediate phenotypes would explain the observed expression of TRM-associated genes even in cutaneous T cells that lack defining features of TRM, such as CD103. Furthermore, this theory is supported by the recent discovery that TRCM in CTCL patients could be further delineated into two major subgroups with differing rates of depletion from the skin by alemtuzumab [32]. As the duration of observation in previous studies has been limited by logistical and technical factors (up to 6 months for mouse experiments and ~6 weeks in alemtuzumab patients [32,38]), it is unknown whether some of the T cells that persisted early on in the skin might eventually migrate out into the circulation. Importantly, the abovementioned hypotheses need not be mutually exclusive, as features unique to the skin microenvironment may both recruit T cells as well as encourage their development along the TRCM-TRM continuum [39].

In summary, our report provides evidence that CD8+CD103-, Treg and CD4+ TRM may exist in healthy human skin, sharing a common skin-associated gene expression signature. The distinction between TRCM and TRM will provide new insights into T cell biology, with potential transcriptional and functional consequences on par with the CD4+ and CD8+ T cell lineages.

Supporting Information

S1 Fig. Validation of microarray results using quantitative real-time PCR.

(a) PCR was performed to examine CCR7 expression in sorted skin- and blood-derived CD8+ T cells. (b) Expression of CD4 in skin-derived CD8+ and CD4+ T cells as determined by PCR.

doi:10.1371/journal.pone.0148351.s001

(PDF)

S2 Fig. Skin-tropic T cells in the blood are enriched for genes downregulated in resident memory T cells.

Results of Gene Set Enrichment Analysis using gene set lists of genes downregulated in lung, gut and skin resident memory T cells (TRM). Negative enrichment scores indicate that gene sets are enriched in blood compared to skin samples. All gene sets shown are significantly enriched at False Discovery Rate <25%. N = 5 arrays per cell type.

doi:10.1371/journal.pone.0148351.s002

(PDF)

S1 Table. T cell populations sorted from blood and skin for microarray.

Surface markers were used to identify and sort live T cell populations from skin and blood for RNA extraction. For each of the 6 cell types, 5 biological replicates were obtained.

doi:10.1371/journal.pone.0148351.s003

(PDF)

S2 Table. Gene sets used for Gene Set Enrichment Analysis.

Gene sets contain lists of genes, compiled in Illumina probe ID format, that are typically up- or downregulated in resident memory T cells (TRM) from lung, skin and gut.

doi:10.1371/journal.pone.0148351.s004

(PDF)

S3 Table. Significantly differentially expressed genes between blood and skin T cells.

Significantly differentially expressed genes (DEGs) identified after pairwise comparison of microarray results with the RUVinv statistical method. Log2Fold-Change (log2FC) cutoff of 1.5 used. P<0.05 after multiple testing correction for all genes shown. Bold = differentially expressed genes shared between all 3 groups.

doi:10.1371/journal.pone.0148351.s005

(PDF)

S4 Table. Significantly differentially expressed genes between T cell lineages in blood and in skin.

Significantly differentially expressed genes identified after pairwise comparison of microarray results with the RUVinv statistical method. Log2Fold-Change (log2FC) cutoff of 1.5 used. P<0.05 after multiple testing correction for all genes shown. Bold = common between blood and skin CD8 versus CD4 T cells. Bold italicized = common between blood and skin Treg versus CD4 T cells.

doi:10.1371/journal.pone.0148351.s006

(PDF)

S5 Table. Gene ontology (GO) analysis of differentially expressed genes upregulated in skin T cells compared to blood T cells.

Data obtained from PANTHER version 10.0 Overrepresentation Test (release 20150430) using PANTHER GO-Slim Biological Process annotation data set. P-values are adjusted for multiple testing with the Bonferroni method.

doi:10.1371/journal.pone.0148351.s007

(PDF)

S6 Table. Results of leading edge analysis of Gene Set Enrichment Analysis.

Leading edge analysis was performed to determine which genes in the various skin T cell types contributed most to the enrichment score for the gene sets pertaining to skin resident memory T cells (TRM), i.e. gene sets containing the genes upregulated in skin TRM and downregulated in skin TRM. Treg = regulatory T cells. Bold = shared leading edge subset genes between the 3 groups.

doi:10.1371/journal.pone.0148351.s008

(PDF)

Acknowledgments

We thank all patients for participating in this study, and the team at the Royal Women’s Hospital for assistance in obtaining tissue samples. We are grateful to Dr. D Fernandez-Ruiz for help in compiling the TRM gene set lists. Biospecimens and data used in this research were obtained from the Victorian Cancer Biobank, Victoria, Australia, with appropriate ethics approval. The Victorian Cancer Biobank is supported by the Victorian Government. We thank Dr. K Shaw for assistance with proofreading.

Author Contributions

Conceived and designed the experiments: FRC. Performed the experiments: JL. Analyzed the data: JL MO JZM. Contributed reagents/materials/analysis tools: JL MO FRC JZM. Wrote the paper: JL MO FRC JZM.

References

  1. 1. Clark RA, Chong B, Mirchandani N, Brinster NK, Yamanaka K, Dowgiert RK, et al. (2006) The vast majority of CLA+ T cells are resident in normal skin. Journal of Immunology 176: 4431–4439.
  2. 2. Sheridan BS, Lefrancois L (2011) Regional and mucosal memory T cells. Nat Immunol 12: 485–491. pmid:21739671
  3. 3. Gebhardt T, Wakim LM, Eidsmo L, Reading PC, Heath WR, Carbone FR (2009) Memory T cells in nonlymphoid tissue that provide enhanced local immunity during infection with herpes simplex virus. Nat Immunol 10: 524–530. doi: 10.1038/ni.1718. pmid:19305395
  4. 4. Zhu J, Peng T, Johnston C, Phasouk K, Kask AS, Klock A, et al. (2013) Immune surveillance by CD8alphaalpha+ skin-resident T cells in human herpes virus infection. Nature 497: 494–497. doi: 10.1038/nature12110. pmid:23657257
  5. 5. Gebhardt T, Mueller SN, Heath WR, Carbone FR (2013) Peripheral tissue surveillance and residency by memory T cells. Trends Immunol 34: 27–32. doi: 10.1016/j.it.2012.08.008. pmid:23036434
  6. 6. Schenkel JM, Fraser KA, Beura LK, Pauken KE, Vezys V, Masopust D (2014) T cell memory. Resident memory CD8 T cells trigger protective innate and adaptive immune responses. Science 346: 98–101. doi: 10.1126/science.1254536. pmid:25170049
  7. 7. Mackay LK, Rahimpour A, Ma JZ, Collins N, Stock AT, Hafon ML, et al. (2013) The developmental pathway for CD103(+)CD8+ tissue-resident memory T cells of skin. Nat Immunol 14: 1294–1301. doi: 10.1038/ni.2744. pmid:24162776
  8. 8. Wakim LM, Woodward-Davis A, Bevan MJ (2010) Memory T cells persisting within the brain after local infection show functional adaptations to their tissue of residence. Proceedings of the National Academy of Sciences of the United States of America 107: 17872–17879. doi: 10.1073/pnas.1010201107. pmid:20923878
  9. 9. Tang VA, Rosenthal KL (2010) Intravaginal infection with herpes simplex virus type-2 (HSV-2) generates a functional effector memory T cell population that persists in the murine genital tract. Journal of Reproductive Immunology 87: 39–44. doi: 10.1016/j.jri.2010.06.155. pmid:20688399
  10. 10. Steinert EM, Schenkel JM, Fraser KA, Beura LK, Manlove LS, Igyarto BZ, et al. (2015) Quantifying Memory CD8 T Cells Reveals Regionalization of Immunosurveillance. Cell 161: 737–749. doi: 10.1016/j.cell.2015.03.031. pmid:25957682
  11. 11. Sathaliyawala T, Kubota M, Yudanin N, Turner D, Camp P, Thome JJ, et al. (2013) Distribution and compartmentalization of human circulating and tissue-resident memory T cell subsets. Immunity 38: 187–197. doi: 10.1016/j.immuni.2012.09.020. pmid:23260195
  12. 12. Wakim LM, Woodward-Davis A, Liu R, Hu Y, Villadangos J, Smyth G, et al. (2012) The molecular signature of tissue resident memory CD8 T cells isolated from the brain. Journal of Immunology 189: 3462–3471.
  13. 13. Clark RA, Watanabe R, Teague JE, Schlapbach C, Tawa MC, Adams N, et al. (2012) Skin effector memory T cells do not recirculate and provide immune protection in alemtuzumab-treated CTCL patients. Sci Transl Med 4: 117ra117.
  14. 14. Bromley SK, Yan S, Tomura M, Kanagawa O, Luster AD (2013) Recirculating memory T cells are a unique subset of CD4+ T cells with a distinct phenotype and migratory pattern. Journal of Immunology 190: 970–976.
  15. 15. Agace WW (2006) Tissue-tropic effector T cells: generation and targeting opportunities. Nat Rev Immunol 6: 682–692. pmid:16932753
  16. 16. Fuhlbrigge RC, Kieffer JD, Armerding D, Kupper TS (1997) Cutaneous lymphocyte antigen is a specialized form of PSGL-1 expressed on skin-homing T cells. Nature 389: 978–981. pmid:9353122
  17. 17. Tomura M, Honda T, Tanizaki H, Otsuka A, Egawa G, Tokura Y, et al. (2010) Activated regulatory T cells are the major T cell type emigrating from the skin during a cutaneous immune response in mice. Journal of Clinical Investigation 120: 883–893. doi: 10.1172/JCI40926. pmid:20179354
  18. 18. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43: e47. doi: 10.1093/nar/gkv007. pmid:25605792
  19. 19. Dunning M, Lynch A, Eldridge M (2015) illuminaHumanv4.db: Illumina HumanHT12v4 annotation data (chip illuminaHumanv4). R package version 1.26.0.
  20. 20. Barbosa-Morais NL, Dunning MJ, Samarajiwa SA, Darot JF, Ritchie ME, Lynch AG, et al. (2010) A re-annotation pipeline for Illumina BeadArrays: improving the interpretation of gene expression data. Nucleic Acids Res 38: e17. doi: 10.1093/nar/gkp942. pmid:19923232
  21. 21. Gagnon-Bartsch JA (2014) ruv: Detect and Remove Unwanted Variation using Negative Controls. R package version 0.9.4.
  22. 22. Gagnon-Bartsch JA, Jacob L, Speed TP (2013) Removing unwanted variation from high dimensional data with negative controls. Technical Report 820: Department of Statistics, University of California, Berkeley.
  23. 23. Gagnon-Bartsch JA, Speed TP (2012) Using control genes to correct for unwanted variation in microarray data. Biostatistics 13: 539–552. doi: 10.1093/biostatistics/kxr034. pmid:22101192
  24. 24. Mi H, Muruganujan A, Thomas PD (2013) PANTHER in 2013: modeling the evolution of gene function, and other gene attributes, in the context of phylogenetic trees. Nucleic Acids Res 41: D377–386. doi: 10.1093/nar/gks1118. pmid:23193289
  25. 25. Kutmon M, van Iersel MP, Bohler A, Kelder T, Nunes N, Pico AR, et al. (2015) PathVisio 3: an extendable pathway analysis toolbox. PLoS Comput Biol 11: e1004085. doi: 10.1371/journal.pcbi.1004085. pmid:25706687
  26. 26. Kelder T, van Iersel MP, Hanspers K, Kutmon M, Conklin BR, Evelo CT, et al. (2012) WikiPathways: building research communities on biological pathways. Nucleic Acids Res 40: D1301–1307. doi: 10.1093/nar/gkr1074. pmid:22096230
  27. 27. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences of the United States of America 102: 15545–15550. pmid:16199517
  28. 28. Schmittgen TD, Livak KJ (2008) Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc 3: 1101–1108. pmid:18546601
  29. 29. Kataoka T (2009) Chemical biology of inflammatory cytokine signaling. Journal of Antibiotics 62: 655–667. doi: 10.1038/ja.2009.98. pmid:19834513
  30. 30. Morel Y, Barouki R (1999) Repression of gene expression by oxidative stress. Biochemical Journal 342 Pt 3: 481–496. pmid:10477257
  31. 31. Zaid A, Mackay LK, Rahimpour A, Braun A, Veldhoen M, Carbone FR, et al. (2014) Persistence of skin-resident memory T cells within an epidermal niche. Proceedings of the National Academy of Sciences of the United States of America 111: 5307–5312. doi: 10.1073/pnas.1322292111. pmid:24706879
  32. 32. Watanabe R, Gehad A, Yang C, Scott LL, Teague JE, Schlapbach C, et al. (2015) Human skin is protected by four functionally and phenotypically discrete populations of resident and recirculating memory T cells. Sci Transl Med 7: 279ra239.
  33. 33. Park CO, Kupper TS (2015) The emerging role of resident memory T cells in protective immunity and inflammatory disease. Nature Medicine 21: 688–697. doi: 10.1038/nm.3883. pmid:26121195
  34. 34. Skon CN, Lee JY, Anderson KG, Masopust D, Hogquist KA, Jameson SC (2013) Transcriptional downregulation of S1pr1 is required for the establishment of resident memory CD8+ T cells. Nat Immunol 14: 1285–1293. doi: 10.1038/ni.2745. pmid:24162775
  35. 35. Grice EA, Segre JA (2011) The skin microbiome. Nat Rev Microbiol 9: 244–253. doi: 10.1038/nrmicro2537. pmid:21407241
  36. 36. Nakatsuji T, Chiang HI, Jiang SB, Nagarajan H, Zengler K, Gallo RL (2013) The microbiome extends to subepidermal compartments of normal skin. Nat Commun 4: 1431. doi: 10.1038/ncomms2441. pmid:23385576
  37. 37. Schenkel JM, Fraser KA, Vezys V, Masopust D (2013) Sensing and alarm function of resident memory CD8(+) T cells. Nat Immunol 14: 509–513. doi: 10.1038/ni.2568. pmid:23542740
  38. 38. Jiang X, Clark RA, Liu L, Wagers AJ, Fuhlbrigge RC, Kupper TS (2012) Skin infection generates non-migratory memory CD8+ T(RM) cells providing global skin immunity. Nature 483: 227–231. doi: 10.1038/nature10851. pmid:22388819
  39. 39. Turner DL, Farber DL (2014) Mucosal resident memory CD4 T cells in protection and immunopathology. Front Immunol 5: 331. doi: 10.3389/fimmu.2014.00331. pmid:25071787