Compartmentalized cytotoxic immune response leads to distinct pathogenic roles of natural killer and senescent CD8⁺ T cells in human cutaneous leishmaniasis

Summary

Cytotoxic activity mediated by CD8⁺ T cells is the main signature of the immunopathogenesis of cutaneous leishmaniasis (CL). Here, we performed a broad evaluation of natural killer (NK) cell phenotypic and functional features during cutaneous leishmaniasis. We demonstrate for the first time that CL patients present the accumulation of circulating NK cells with multiple features of replicative senescence including low proliferative capacity and shorter telomeres, elevated expression of CD57, KLRG1 but diminished CD27 stimulatory receptor expression. Moreover, they exhibited higher cytotoxic and inflammatory potential than age‐matched controls. The accumulation of circulating senescent NK cells (CD56^dim CD57^bright) correlated positively with skin lesion size in the same patients, suggesting that they, like circulating senescent CD8⁺ T cells, may contribute to the immunopathology of CL. However, this senescent population had lower cutaneous lymphocyte antigen expression and so had diminished skin‐homing potential compared with total or senescent CD8⁺ T cells. This was confirmed in CL skin lesions where we found a predominance of CD8⁺ T cells (both senescent and non‐senescent) that correlated with the severity of the disease. Although there was also a correlation between the proportions of senescent NK cells (CD56⁺ CD57⁺) in the skin and lesion size, this was less evident. Collectively our results demonstrate first‐hand that senescent cytotoxic cells may mediate skin pathology during human cutaneous leishmaniasis. However, as senescent cytotoxic CD8⁺ T cells predominate in the skin lesions, they may have a greater role than NK cells in mediating the non‐specific skin damage in CL.

Senescent CD8 and natural killer cells accumulate in lesions during human cutaneous leishmaniasis and mediate the skin pathogenesis through cytotoxicity.

Abbreviations

CL

cutaneous leishmaniasis

CLA

cutaneous lymphocyte antigen

EMRA

effector memory T cells that re‐express CD45RA

KLRG1

killer cell lectin‐like receptor

MCL

mucocutaneous leishmaniasis

TCR

T‐cell receptor

Th1

T helper 1

Introduction

Leishmania braziliensis is the main causal agent of American tegumentary leishmaniasis, a serious neglected tropical disease where destructive cutaneous lesions develop.1 In cutaneous leishmaniasis (CL), a good prognosis is related to the predominance of a cellular T helper type 1 (Th1) response with production of interferon‐γ, tumour necrosis factor‐α (TNF‐α) and activation of parasite‐infected macrophages. On the other hand, the exacerbated Th1 immunity and cytotoxic response were identified as key elements in the immunopathogenesis of both murine and human CL caused by L. braziliensis.2, 3, 4, 5 The gene expression analysis in CL skin lesions revealed that cytotoxicity‐related genes are overexpressed compared with the cytokine pathway,6 representing the main signature of L. braziliensis infection. In patients with CL, the accumulation of cytotoxic CD8⁺ T cells or granzyme B‐producing cells was linked with the severity of CL.2 We showed recently that CD8⁺ T cells are driven towards senescence acquiring high cytotoxic potential and skin‐homing capacity, which may promote skin damage.7 Although the role of cytotoxic and senescent CD8⁺ T cells in the immunopathology in CL is established, it is not clear if senescent natural killer (NK) cells also have a role in this process.

Natural killer cells comprise 5–20% of peripheral blood mononuclear cells (PBMC) in humans and play a central role in immunosurveillance through their cytotoxic and pro‐inflammatory activities, without a requirement for prior sensitization.8 Similar to observations in the T‐cell pool, the differentiation state of NK cells modulates their functional capacity, which is still unknown in the context of Leishmania infection. NK cells can be divided into distinct phenotypic and functional subsets based on the relative expression of cell‐surface CD56 and CD16 (FcRIIIa).8 The CD56^bright NK subset has increased immunoregulatory and proliferative capacity after stimulation with cytokines, whereas the CD56^dim cells (the majority population ~90%) represents the most differentiated subset. The protective role of NK cells during CL is demonstrated by the increased proliferative activity in cured individuals compared with patients with active lesions.9 Furthermore, higher numbers of CD56⁺ cells are found in the peripheral blood of patients with CL before and after treatment,10 as well as in lesions of patients with diffuse CL who have a positive response to immunotherapy.11 Conversely, increased NK cell activity is linked to susceptibility and severity of human visceral leishmaniasis,12 CL13, 14 and mucocutaneous leishmaniasis.15

The pivotal balance that regulates either the functional activity of senescent CD8⁺ T cells or NK cytotoxic cells in blood and lesions of patients with CL is poorly understood. Here, we characterized the phenotypic and functional profiles of circulating NK cell subsets in these individuals. Similar to the CD8⁺ compartment, we found that L. braziliensis infection induces the terminal differentiation of NK cells with a high cytotoxic and inflammatory potential that is related to the pathology of CL. We also found that while senescent NK cells predominate in the blood compartment, senescent CD8⁺ cells are preferentially localized in the cutaneous lesions and their presence is significantly associated with tissue damage. Our results provide a broad understanding of the relationship between systemic and skin immunity and establish for the first time the relative roles of NK and CD8⁺ T cells in the pathogenesis of CL.

Materials & methods

Study subjects

Peripheral blood from 16 patients with untreated CL attending University Hospital (HUCAM) of Universidade Federal do Espirito Santo, Brazil were investigated in this study. They comprised nine males and seven females with illness duration ranging from 30 to 120 days and lesion sizes ranging from 200 to 600 mm². The diagnosis of CL was based on clinical and laboratory criteria and all patients in this study were positive for the polymerase chain reaction/restriction fragment length polymorphism of L. braziliensis and reported no previous infections or treatment. The control group consisted of 16 healthy age‐ and gender‐matched individuals (HC) living in a non‐endemic area without a history of leishmaniasis. All study participants (patients and healthy volunteers) were seronegative for HIV, hepatitis B virus and hepatitis C virus infections, had no history of chemotherapy, radiotherapy or treatment with immunosuppressive medications within the last 6 months. They provided written informed consent, and study procedures were performed in accordance with the principles of the Declaration of Helsinki. This study was registered with the HUCAM ethics committee under reference number 735.274.

PBMC isolation, cell sorting and culture

PBMC from CL and HC patients were isolated by centrifuging whole blood through a Ficoll‐Hypaque (GE Healthcare, Chalfont St Giles, UK) gradient followed by haemocytometry to determine absolute live cell number. Both NK and K562 cells were cultured in complete medium (RPMI‐1640 supplemented with 10% heat‐inactivated fetal calf serum, 100 U/ml penicillin, 100 mg/ml streptomycin and 2 mm l‐glutamine; Invitrogen, Carlsbad, CA). NK cells were negatively isolated from the PBMC fraction using an NK Cell Isolation Kit/VARIOMACS system (Miltenyi Biotec, Bisley, UK) according to the manufacturer's instructions.

Flow cytometric analysis

The lymphocytes were live gated using Live/Dead stain after exclusion of the doublet cells. The NK cell population was further identified and differentiated into immature (CD3⁻ CD7⁺ CD56^bright) and mature (CD3⁻ CD7⁺ CD56^dim) subsets on the CD56 (Fig. 1a). Moreover, the NK differentiation phenotype was also evaluated based on the expression of CD16, in which were identified immature (CD3⁻ CD7⁻ CD16⁻ CD56^bright), intermediate (CD3⁻ CD7⁻ CD16⁺ CD56^bright) and mature (CD3⁻ CD7⁻ CD16⁻ CD56^dim) subsets (Fig. 1d). The analyses were performed with the following antibodies: Live/Dead Blue Fixable Stain (L23105, Thermo Scientific, Waltham, MA), anti‐CD3 (UCHT1), anti‐CD16 (3G8), anti‐CD27 (M‐T271), anti‐CD57 (NK‐1), human cutaneous lymphocyte antigen (CLA) FITC (HECA‐452), granzima B (GB11), perforina (∂G9), anti‐CD107a and Ki67 (B56) from BD Biosciences (San Jose, CA). Anti‐CD7 (M‐T701), anti‐CD56 (HCD56) and anti‐KLRG1 (REA261) from Miltenyi Biotec. For surface markers, staining was performed at 4° for 30 min in the presence of saturating concentrations of antibodies. For intracellular analysis, cells were fixed and permeabilized with the Fix & Perm^® Kit (Invitrogen, Life Technologies, Carlsbad, CA), before incubation with the indicated antibodies. Intracellular staining for Ki67 was performed with a Foxp3 Staining Buffer Set (Miltenyi Biotec). Samples were acquired in a Fortessa X‐20 cytometer (BD Biosciences) and analysed using flowjo software (Treestar, Woodbury, OR).

Figure 1.

Patients with cutaneous leishmaniasis present increased frequency on circulating natural killer (NK) cells within a CD56^dim phenotype. (a) Representative gate strategy. (b) Cumulative frequencies of total NK cells and (c) CD56^dim and CD56^bright subsets from healthy control donors (HC n = 16) and patients with cutaneous leishmaniasis (CL n = 16). (d) Representative gate strategy and (e) cumulative frequencies of CD56^dim and CD56^bright subsets within CD16⁺ cells. The graphs or dot plots show median with 95% confidence. The P values were calculated using Mann–Whitney U‐test. *P < 0·05, **P < 0·01

Telomere hybridization

Isolated NK cells were fixed on poly‐l‐lysine‐coated glass slides and processed for fluorescence in situ hybridization, as described previously.16 Samples were hybridized with peptide nucleic acid probes for the telomere (TelC‐Cy3 probe, 5′‐CCCTAACCCTAACCCTAA‐3′; (Cambridge Research Biochemicals, Billingham, UK) and the nuclei were stained with 4′,6‐diamidino‐2‐phenylindole (DAPI; Molecular Probes, Eugene, OR). Images were acquired using a Leica SPE2 confocal microscope using LAS X software. Images corresponded to a full Z‐stack with 0·5‐µm step size and were projected at maximum intensity and analysed using imagej version 2.0.0‐rc‐43/1.51q.

Calcein‐release cytotoxicity assay

The cytotoxic activity was assessed using the K562 cells (human erythroleukaemia cell line) as target. Briefly, 20 000 K562 cells were labelled with Calcein‐AM (Sigma‐Aldrich, St Louis, MO) at 10 μm for 1 hr and plated in a 96‐well flat‐bottom plate for co‐culture with NK cells in complete medium containing 500 IU/ml recombinant human interleukin‐2 (rhIL‐2) (Miltenyi Biotech). Effector and target cells were combined at a ratio of 40 : 1 in triplicate. After 4 hr of co‐culture, fluorescence was measured in 75 μl of cell culture supernatant using a Spectramax Gemini spectrofluorimeter. Specific lysis was calculated as % killing = (test release–spontaneous release)/(max release–spontaneous release) × 100.

CD107a degranulation assay

Isolated NK cells were incubated at 37° for 4 hr with target K562 cells, at an effector to target (E : T) ratio of 2 : 1, in the presence of allophycocyanin‐conjugated CD107a antibody (BD Biosciences), as previously described.17 After incubation, cells were stained for surface markers and CD107a expression and analysed by flow cytometry.

Immunofluorescence analysis

For confocal microscopy, 8‐mm punch biopsies from the edge of lesional skin were embedded in OCT, frozen on liquid nitrogen and preserved at –80° until use. Healthy (uninfected) skin samples were taken from volunteers using a 4‐mm punch. The frozen skin samples were longitudinally sectioned at 8‐μm thickness and placed on glass microscopy slides. Immunofluorescence staining was performed and images were analysed on the SlidePath Digital Image Hub (Leica, Breckland, UK) with definiens tissue studio 3.6 (Definiens AG, Breckland, UK). Multiple primary staining of nuclei with DAPI (1 μg/ml), granzyme B (Abcam‐4059; Cambridge, MA), CD57 (187274‐Abcam), CD8 (MCA352‐BioRad) and CD56 (56C04‐Thermo) was performed. The secondary staining was carried out using the following antibodies: anti‐rat (A11006), anti‐mouse (A21124) or (A21046) and anti‐rabbit (A21244), all from Thermo Scientific. Analysis was performed by manually selecting regions of interest (ROI) using a Chromoplex Staining Detection system (Leica Biosystems). The cell and markers frequencies were defined according to the evaluated marker and cell nuclei of each ROI. Imaged with a 20 × objective (200 × magnification) on an upright confocal microscope with a 710 Metahead (Zeiss, Oberkochen, Germany) by z‐stacktile‐scans.

Granzyme B and cytokines determination

Granzyme B, IL‐2, interferon‐γ and TNF‐α were determined in cell culture supernatant of NK‐cytotoxicity assays using cytokine bead array (Flex Set from BD Biosciences) according to the manufacturer's instructions.

Proliferation assay

Sorted NK cells were cultured in the presence of rhIL‐2 (100 U/ml) for 48 hr, followed by intracellular staining for the cell‐cycle‐related nuclear antigen Ki67 Alexa Fluor 647 (B56; BD Bioscience) and analysed by flow cytometry. Supplemented culture medium without rIL‐2 was used as control.

tSNE analysis

Unbiased representations of multi‐parameter flow cytometry data were generated using the t‐distributed stochastic neighbour embedding (tSNE) algorithm. The R package ‘Rtsne’ available on CRAN (http://github.com/jkrijthe/Rtsne) was used to perform the Barnes Hut implementation of tSNE on flow cytometry data, where a similar number of events from each sample were analysed in parallel based on the fluorescence parameters and grouped near to each other based on the similarity expression levels.

Statistical analysis

graphpad prism (version 7) was used to perform statistical analysis (GraphPad, San Diego, CA). Statistical significance was evaluated using the paired Student's t‐test. Mann–Whitney U‐test was performed for all continuous, non‐parametric variables and correlations were calculated using Pearson's correlation coefficient. Differences were considered significant when P was < 0·05.

Results

Patients with localized CL have an increase in the number of circulating NK cells that present a mature/senescent phenotype

First, we investigated the heterogeneity of circulating NK cell populations in the HC and CL groups. The percentage of total NK cells identified as CD3⁻ CD7⁺ CD56⁺ (Fig. 1a) was similar in PBMC from the CL group compared with the HC group (Fig. 1b). However, compared with the HC group, NK cells from patients with CL have a decreased frequency of CD56^bright and increased frequencies of CD56^dim subsets (Fig. 1c) as well as higher frequencies of the mature CD16⁺ CD56^dim subset (Fig. 1e), both constituting the most mature NK populations.

The simultaneous expression of CD57 and KLRG1 and the loss of CD27 co‐stimulatory receptor have been used to define highly differentiated T and NK cells that exhibit characteristics of cellular senescence.8, 18 Patients with CL demonstrate increased frequencies of CD57 and KLRG1 in the total NK pool compared with the HC group that was mainly found to have the mature CD56^dim subset (Fig. 2a,b). The acquisition of CD57 expression by immature NK cells is gradual, supporting the concept of a continuous process of maturation. We identified an increased accumulation of CD57^bright NK subset in patients with CL, representing the late‐differentiated phenotype (Fig. 2c,d). This supports the idea that circulating NK cells are driven towards senescence during CL.

Figure 2.

Circulating natural killer (NK) cells from patients cutaneous leishmaniasis (CL) increase the expression of differentiation markers. (a) Representative histogram and (b) cumulative data of percentage of CD57, KLRG1 and CD27 receptors in total NK cells and CD56^dim and CD56^bright subsets. (c) Representative histogram and (d) cumulative data of NK subsets frequencies expressing CD57. The analyses were performed by flow cytometry using samples frm HC (n = 10 to n = 16) and CL (n = 10 to n = –16) groups. The graphs show median with 95% confidence. The P values were calculated using Mann–Whitney U‐test. *P < 0·05, **P < 0·01, ***P < 0·001

NK cells from patients with CL exhibit increased effector function but low proliferative capacity and telomere erosion

We, therefore, next assessed the pro‐inflammatory cytokine secretion, degranulation and the killing potential of circulating NK cells against MHC class I‐deficient K562 cells. Compared with controls, CL NK cells displayed both higher degranulation and greater killing capacity of the target cells (Fig. 3a). Moreover, they secreted significantly increased levels of interferon‐γ, TNF‐α and granzyme B (Fig. 3a) but showed less proliferation after IL‐2 activation (Fig. 3b), which was accentuated within the CD56^dim CD57^bright subset (Fig. 3c). Moreover, fluorescence in situ hybridization analysis in isolated NK cells from patients with CL showed shorter telomeres that were significantly lower (1·7‐ to 1·9‐ fold) than those observed in the HC group (Fig. 3d).

Figure 3.

Natural killer (NK) cells from patients with cutaneous leishmaniasis (CL) present pronounced inflammatory and functional profiles and display features of cellular senescence. Purified NK cells (over 95% purity) from healthy controls (HC) (n = 6) and CL (n = 6) groups were co‐cultured with K562 target cells (effector : target ratio of 10 : 1) for 4 hr. (a) The cytotoxic activity assessed by calcein‐release lysis assay; degranulation capacity or production of granzyme B, interferon‐γ (IFN‐γ) and tumour necrosis factor‐α (TNF‐α) were determined after co‐culture. (b) Representative histogram and cumulative data of the proliferative capacity assessed by Ki67 staining of purified NK cells or (c) their subsets stratified by CD57 expression. Cells were cultured with interleukin‐2 (IL‐2) (100 U/ml) for 48 hr. Results show the relative fold change normalized to the unstimulated control. (d) Telomere fluorescence in situ hybridization image in NK cells hybridized with telomere probe (TelC Cy3) and fold change of quantitative fluorescence intensity levels normalized with CL group. The graphs show median with 95% confidence. The P values were calculated using Mann–Whitney U‐test. *P < 0·05, **P < 0·01, ***P < 0·001, ****P < 0·0001

We next evaluated the ex vivo capacity of senescent NK cells to produce cytotoxic mediators. Hence, we found that CD57^dim and CD57^bright cells spontaneously produce a higher amount of granzyme B and perforin compared with the less mature phenotype (CD57^neg) (Fig. 4a,b). Senescent NK and T cells have increased cytotoxic and pro‐inflammatory capacity,19 which may potentially contribute to the pathogenesis of CL.

Figure 4.

Highly differentiated natural killer (NK) cell subset is cytotoxic and correlates with lesion size in patients. Cumulative data of basal production of (a) granzyme B and (b) perforin in NK cell subsets stratified by CD57 expression. The analyses were performed by flow cytometry using samples from healthy controls (HC; n = 7) and cutaneous leishmaniasis (CL) (n = 7) groups. Results show the relative fold change normalized to the respective early differentiated CD57^neg subset. (c) Spearman's correlation test between frequencies of CD57 subsets within CD56^dim NK cells and lesions size (mm²) of CL patients (n = 16) or (d) between frequencies of NK CD57^bright and CD8⁺ EMRA subsets

Accumulation of senescent cytotoxic NK cells correlates with lesion size observed in patients with CL

We found that patients with CL who had the highest proportion of senescent/cytotoxic CD57^bright NK cells (Fig. 2c) had the largest cutaneous lesions (Fig. 4c), suggesting that they have a role in the cutaneous immunopathology, as was shown previously for senescent (EMRA, or effector memory T cells that re‐express CD45RA) CD8⁺ T cells in these patients.7 Furthermore, we found a significant correlation between the increase in senescent NK and senescent CD8⁺ T cells in the same patients (Fig. 4d), suggesting that related mechanisms induce the expansion of both populations in the circulation of patients with CL.

Senescent CD8 T and NK cells in the circulation have different cytotoxic capacity and migratory potential in the skin of CL pathogenesis

We next compared the spontaneous skin‐homing capacity, and proliferative and cytotoxic potential of circulating senescent NK (CD57^bright) and senescent CD8⁺ (EMRA) T cells from the same donors. We found that senescent NK cells (CD57^bright) produced increased amounts of perforin, granzyme B and CD107 compared with senescent CD8⁺ T cells (EMRAs) in the same individuals. Furthermore, although the senescent NK subset exhibits a high proliferative potential (Ki67⁺), they showed a lower capacity for skin homing, as demonstrated by the CLA expression compared with CD8⁺ EMRA T cells (Fig. 5d,e). The total CD8⁺ T and NK cell populations from the patients showed higher levels of perforin, granzyme B and CD107 expression compared with healthy controls (see Supplementary material, Figure S1A–C). In contrast, the circulating CD8⁺ T cells exhibited increased skin homing but decreased proliferative potential compared with the NK cells from the same CL patients (Figure S1D,E). In addition, no differences in total frequencies of both the NK⁺ and CD8⁺ cell pools were observed in patients and controls (data not shown).

Figure 5.

Cytotoxic granules production, and proliferative and migratory capacity are highly diverse across senescent CD8⁺ T and natural killer (NK) cells from healthy controls (HC) and patients with cutaneous leishmaniasis (CL). (a–e) Peripheral blood mononuclear cells (PBMCs) from healthy donors (HC) (n = 7) and patients (CL) (n = 7) were directly stained for surface and intracellular markers and analysed by flow cytometry. (f) tSNE was performed gating on CD8⁺ EMRA and NK CD57^bright cells from HC and CL groups. The level of expression of Ki67, CLA, CD107a, perforin and granzyme B were evaluated separately on live cells generating the expression levels of the hierarchical clusters, represented in red for high expression, whereas blue represents low expression (cold‐to‐hot heat map)

We confirmed the functional profile of cytotoxic senescent cells by the tSNE algorithm.20 We arbitrarily identified two different clusters of CD8⁺ EMRA (Green) and CD57^bright (Red) subsets on the basis of population boundaries distinguishable on the tSNE density plots (Fig. 5f) and the expression intensities of markers in each cluster were displayed using a heat map. As expected, NK CD57^bright had increased Ki67, CD107a and perforin while expressing lower levels of CLA compared with CD8⁺ EMRA T cells. Granzyme B expression overlapped across the distinct clusters defined in the tSNE map and showed a similar distribution in both subsets (Fig. 5f).

Senescent cytotoxic cells accumulate in the skin and correlate with lesion size of patients with CL

We next investigated the presence and function of NK and CD8⁺ T cells and their senescent subsets in the skin lesions to determine their relative contribution to the observed immunopathology. Representative immunofluorescence micrographs demonstrate a significant accumulation of granzyme‐producing cells in the lesions (Fig. 6a,c,e) with a predominance of CD8⁺ T cells compared with NK cells (Fig. 6f). Interestingly, there was a significantly greater accumulation of senescent CD8⁺ T cells compared with senescent NK cells in the skin of patients with CL (Fig. 6g). This suggests that the increased potential for skin homing through the higher expression of CLA leads to a preferential accumulation of senescent CD8⁺ T cells in the skin lesions of CL patients, which extends previous observations.7

Figure 6.

Senescent cytotoxic cells accumulate in the skin and correlate with lesion size of patients with cutaneous leishmaniasis (CL). Histological sections of human skin punch‐biopsy from patients with CL (n = 6) stained for DAPI (blue); natural killer (NK) or CD8⁺ T cells (green); granzyme B (red) and CD57 (grey). The yellow arrows indicate senescent cytotoxic cells. Representative images (top panels a–d) and cumulative data of total cytotoxic cells and granzyme‐producing senescent cell frequencies (bottom graphs e–g). Scatterplot showing the Spearman's correlation test relationship between frequencies of NK and CD8⁺ T cells and lesion size (h) or NK^CD57 and CD8⁺ T EMRA and lesion size. The graphs show median with 95% confidence. The P values were calculated using Mann–Whitney U‐test. *P < 0·05, **P < 0·01, ***P < 0·001, ****P < 0·0001

We correlated the size of the skin lesions in patients with cutaneous leishmaniasis with the extent of CD8⁺ T cells or NK cell infiltration. We found that there was a positive correlation between the accumulation of total CD8⁺ T cells but not NK cells in the skin and lesion size (Fig. 6h). Interestingly, the accumulation of senescent CD8⁺ T cells strongly correlates with lesion size while a weak correlation is linked to the differentiated NK cells (Fig. 6i). This suggests that CD8⁺ T cells in the skin have a relatively greater impact than NK cells on the immunopathology in the skin of patients with CL.

Discussion

CD8⁺ T cells play a key role in CL immunopathology through cytotoxicity; however, the contribution of NK cells to this process is unresolved. NK cell function changes from immature CD56^bright and mature but naive CD56^dim CD57⁻ populations that produce high amounts of cytokines to a terminally differentiated/senescent CD56^dim CD57⁺ subset exhibiting strong cytotoxic activity.8, 21 Moreover, the acquisition of senescence characteristics in NK cells, acquired through persistent antigen exposure and/or chronic infection, is associated with the expression of CD16, CD57, KLRG1 and loss of CD27,22, 23 as demonstrated in our findings.

Given the pathogenic potential of both T and NK cell cytotoxic compartments, we tested their potential for skin‐homing. Although circulating NK cells exhibited potent cytotoxicity and inflammatory function compared with CD8⁺ T cells they showed reduced CLA expression and hence capacity to home to the skin. This raised the possibility that NK cells in CL may exert their function preferentially in the circulation of the patients whereas CD8⁺ T cells may exhibit their function in the skin of the patients.

CD8 T cells expressing higher levels of CLA are found in patients with cutaneous disorders24, 25 including cutaneous leishmaniasis,5, 7, 26 compared with healthy individuals. This raises the question of how CLA expression is induced on these cells. Although pro‐inflammatory cytokines have been shown to induce CLA expression on T cells,27 both T cells and NK cells in patients with CL are exposed to the same inflammatory environment in vivo yet there is a difference in expression of this molecule between both subsets. Alternatively, it is possible that specific CD8⁺ memory T lymphocytes are programmed to express CLA after antigenic stimulation. For example, the expression of CLA by memory CD8⁺ T lymphocytes increases during CL and is seen after Leishmania antigen recall in vitro.5, 7, 26 Furthermore, CLA‐negative precursor cells can be induced from CLA‐negative CD8⁺ T‐cell populations after TCR stimulation with viral and bacterial antigens.25, 28 Therefore, the increase in CLA‐expressing CD8⁺ T cells that harbour senescence characteristics during CL may arise from extensive TCR activation and proliferation resulting from stimulation with L. braziliensis antigenic epitopes. In contrast, as NK cells are not activated by the TCR, CLA expression is not induced by them.

Analysis of gene expression in CL skin lesions has shown that cytotoxicity‐related genes are overexpressed compared with the Th1 cytokine pathway and this represents the main signature associated with the promotion of the tissue damage and skin ulceration during CL caused by L. braziliensis.6 Despite this, not much is known of the overall effector mechanisms including cytotoxicity mediated by NK cells in human CL. It has been previously shown that the pathogenic role of CD8⁺ T cells in both human and murine CL caused by L. braziliensis infection occurs through cytotoxic and pro‐inflammatory activity that correlates with lesion size in patients.2, 4, 5, 29 In contrast to a recent report,14 our results show that cytotoxic (and senescent) CD8⁺ T cells preferentially accumulate in the skin lesions of patients with CL compared with NK cells, suggesting that a functional compartmentalization of these cells occurs in CL. Whereas senescent NK cells preferentially exert their function in the circulation, senescent CD8⁺ T cells preferentially exert their functional activity in the skin.

A key unanswered question is the mechanism by which senescent CD8⁺ T cells induce the immunopathology in the skin of patients with CL. It is well recognized that there are very few parasites in CL lesions, indicating that it is unlikely that there is extensive activation of these cells via the TCR. One possibility is that the CD8 T cells in the lesions are inducing non‐specific killing of healthy tissue. Previous studies have shown that senescent CD8⁺ T cells acquire the expression of functional NK receptors.30, 31 Furthermore, inflammation and also the development of senescence induces the expression of NK ligands in tissue cells.31 Therefore, skin‐homing CD8⁺ T cells in CL may mediate non‐specific pathology by killing NKR‐ligand‐expressing cells such as fibroblasts and keratinocytes in the skin. Supporting this possibility, both NKG2D and Rae1γ (an NKG2D ligand in the mouse) are expressed by a large number of leucocytes within skin lesions induced by Leishmania major infection in mice,32 regardless of the parasitic load of the lesion. In addition, the blockade of NKG2D in L. major‐infected mice prevents lysis of target cells and reduces lesion development in vivo.32 In preliminary studies we have also observed increased expression of both NKG2D on CD8⁺ T cells and MICA/MICAB on stromal cells within the lesions of patients with CL (Covre et al., unpublished observations). This suggests that Leishmania‐specific CD8⁺ T cells in the skin may participate in the development and severity of CL through non‐antigen‐driven bystander cytotoxicity mechanisms.

Collectively our results underpin a broad understanding of systemic and local changes in both blood and skin of patients with CL identifying the discrete compartmentalization of NK activity in the blood and CD8⁺ T‐cell activity in the skin of the patients. Furthermore, we suggest that CD8⁺ T cells and not NK cells are mainly responsible for the non‐specific skin lesional pathology. This is the first demonstration of an immunopathogenic role for senescent T cells in vivo.

Disclosures

The authors state no conflict of interest.

Supporting information

Figure S1. Circulating cytotoxic cells in patients with cutaneous leishmaniasis (CL). (A) Peripheral blood mononuclear cells (PBMC) from healthy donors (HC) (n = 7) and patients with CL (n = 7) were directly stained for surface and intracellular markers. Data show circulating total natural killer (NK) and T CD8⁺ cells expressing (A) perforin; (B) granzyme B; (C) CD107; (D) Ki67 and (E) CLA. The graphs show median with 95% confidence. The P values were calculated using Mann–Whitney U‐test. *P < 0.05, **P < 0.01, ***P < 0.001.