Phenotypic Analysis of Disease-Relevant T Cells in Dermatitis Herpetiformis

TO THE EDITOR Gluten-specific CD4þ T cells are central players in the pathogenesis of celiac disease (CeD), an inflammatory disease driven by exposure to dietary gluten proteins. Patients with CeD are treated with a lifelong gluten-free diet. When gluten is reintroduced to the diet of patients in remission, there is a wave of activated (CD38þ) gluten-specific CD4þ T cells in the blood that peaks on days 6‒8 after the first gluten exposure. This increase in the frequency of activated gluten-specific CD4þ effector-memory T cells (TEM) in the blood can be detected by IFN-g enzyme-linked immunospot assay (Anderson et al., 2000; Tye-Din et al., 2010) or by HLA-DQ:gluten tetramers (Ráki et al., 2007; Sarna et al., 2018b; Zühlke et al., 2019). The gluten-specific CD4þ T cells in the blood that increase in frequency also alter their phenotype to that of gluten-specific CD4þ T cells in the gut mucosa of untreated CeD (Christophersen et al., 2021, 2019). Parallel with the increase of glutenspecific CD4þ T cells in the blood, there is an increase in the frequency of gut-homing and activated CD8þ and gd T cells (Christophersen et al., 2021; Han et al., 2013; Risnes et al., 2021). Although dermatitis herpetiformis (DH) is considered an extraintestinal manifestation of CeD, little is known about the gluten-induced T-cell responses in patients with DH. Recently, Kalliokoski et al. (2020) performed IFNg enzyme-linked immunospot assay on PBMCs from patients with DH and CeD on day 6 after a 3-day gluten challenge as part of an oral challenge for a period of up to 1 year. Only 47% of the patients with DH displayed reactivity to deamidated gluten, leading to the conclusion that early IFN-g response to

TO THE EDITOR Gluten-specific CD4þ T cells are central players in the pathogenesis of celiac disease (CeD), an inflammatory disease driven by exposure to dietary gluten proteins. Patients with CeD are treated with a lifelong gluten-free diet. When gluten is reintroduced to the diet of patients in remission, there is a wave of activated (CD38þ) gluten-specific CD4þ T cells in the blood that peaks on days 6-8 after the first gluten exposure. This increase in the frequency of activated gluten-specific CD4þ effector-memory T cells (T EM ) in the blood can be detected by IFN-g enzyme-linked immunospot assay (Anderson et al., 2000;Tye-Din et al., 2010) or by HLA-DQ:gluten tetramers (Ráki et al., 2007;Sarna et al., 2018b;Zü hlke et al., 2019). The gluten-specific CD4þ T cells in the blood that increase in frequency also alter their phenotype to that of gluten-specific CD4þ T cells in the gut mucosa of untreated CeD (Christophersen et al., 2021(Christophersen et al., , 2019. Parallel with the increase of glutenspecific CD4þ T cells in the blood, there is an increase in the frequency of gut-homing and activated CD8þ and gd T cells (Christophersen et al., 2021;Han et al., 2013;Risnes et al., 2021).
Although dermatitis herpetiformis (DH) is considered an extraintestinal manifestation of CeD, little is known about the gluten-induced T-cell responses in patients with DH. Recently, Kalliokoski et al. (2020) performed IFNg enzyme-linked immunospot assay on PBMCs from patients with DH and CeD on day 6 after a 3-day gluten challenge as part of an oral challenge for a period of up to 1 year. Only 47% of the patients with DH displayed reactivity to deamidated gluten, leading to the conclusion that early IFN-g response to selected gluten peptides does not predict clinical relapse in patients with DH on long-term gluten provocation. The study was approved by the Finnish Regional Ethics Committee of Tampere University Hospital (Tampere, Finland), and all patients gave written informed consent.
Using cryopreserved PBMCs that were available from 7 of the 19 patients from the original challenge study (Mansikka et al., 2019), in this study, we aimed to identify the phenotypic markers of disease-relevant T cells in patients with DH by performing multiparametric flow cytometric analysis of CD4þ, CD8þ, and gd T cells. We sought to identify gluten-reactive T cells by use of HLA-DQ2.5:gluten tetramers (termed HLA-tetramers in the remaining part of this paper) representing a mixture of the immunodominant DQ2.5-glia-a1a, DQ2.5-glia-a2, DQ2.5-glia-u1, DQ2.5-glia-u2, and DQ2.5-hor3 epitopes (Tye-Din et al., 2010). We stained the T cells for CD45RA and CD62L as well for guthoming marker integrin b7, cutaneous lymphocyte antigen (CLA), and CCR4 (for further details, see Supplementary Materials and Methods).
We detected CD4þ T EM (CD45RA-CD62L-) cells that bound HLA tetramers in five of the seven patients investigated, but only two patients displayed a clear increase in the frequency of such cells on day 6 ( Figure 1a and b). The number of cells retrieved from the cryopreserved PBMC samples ranged from 0.35 to 3.9 million cells (median ¼ 1.1) with a viability of 78-97% (median ¼ 93%). Ideally, more cells should have been analyzed. Although the tetramer enrichment method is highly sensitive and distinguishes patients with CeD from HLA-matched controls, precise cell number estimates require higher numbers (>10 million) of PBMCs, especially for the baseline samples because the method detects low-frequency cells (typically one cell per million CD4þ T cells) (Sarna et al., 2018a).
The majority of the HLA-tetramerpositive T EM cells expressed integrin b7, suggesting that these are gut-homing cells. Curiously, 31% of the HLAtetramer-positive T EM cells in one of the patients (DH4) expressed the skinhoming marker CLA (Figure 1a and c). The CLAþ HLA-tetramer-positive T cells of this one patient also expressed high levels of CCR4. Whether additional patients with DH also have skin-homing gluten-specific CD4þ T cells in addition to gut-homing gluten-specific CD4þ T cells cannot be concluded from this single observation. More studies are warranted. Despite variation in the increase of numbers of HLA-tetramerpositive T EM cells, in all the five patients, we observed a uniform increase in CD38 expression in HLA-tetramerpositive T EM cells after gluten challenge ( Figure 1d). We further showed that all the HLA-tetramer-positive T EM cells clustered distinctly from the HLAtetramer-negative cells ( Figure 1e). In line with previous observations of patients with CeD (Christophersen et al., 2021(Christophersen et al., , 2019, the HLA-tetramer-positive T EM cells on gluten challenge upregulated CD38 and PD-1 and downregulated CD127 (Figure 1f). To identify which epitope the glutenspecific T cells in patients with DH were reactive to, we sorted CD4þ HLA-tetramer-positive T EM cells to generate T-cell clones and successfully established 10 T-cell clones from four patients. All the 10 T-cell clones displayed proliferative response against deamidated gluten and to at least one of the four immunodominant epitopes of wheat gluten that were represented in the HLA-tetramer cocktail used during sorting (Figure 1g). We also verified the HLA-tetramer binding of these T-    and an increase of CD103þCD38þ gd T cells in four of seven patients (Figure 2a and b). Moreover, the increase in the frequency of CD8þ T cells as fold change ranged from 2.2 to 104. Interestingly, patient DH5 had the lowest frequency of CD103þCD38þ CD8þ T cells on day 6, and this was the only patient who continued for the entire 12-months challenge period. CLA was expressed by a substantial proportion of total CD8þ (9-76%, median ¼ 32%) and gd T cells (13-63%, median ¼ 36%). Also among CD103þCD38þ CD8þ T cells, some expressed CLA (range ¼ 0-51%, median ¼ 13%) (Figure 2c and d). None of the CD103þCD38þ gd T cells expressed the CLA skin-homing marker. Whether any of the activated CD8þ T cells found in the blood on day 6 after gluten challenge home to skin and exert effector functions there remains to be proven. Although further studies with a higher number of patients and cells are warranted, our study suggests that patients with DH have gluten-induced Tcell responses with similar characteristics to regular patients with CeD. Furthermore, CD38 expression in gluten-specific integrin b7þ CD4þ T cells and CD103þCD38þ CD8þ T cells are promising markers to predict clinical relapse on a short gluten challenge in patients with DH.

Data availability statement
The data are not publicly available owing to Finnish legislation concerning patient-related data.

CONFLICT OF INTEREST
The authors state no conflict of interest.
For tetramer restaining of T-cell clones (TCCs), we stained with three different mixtures of HLA-DQ2.5:gluten tetramers consisting of one PE-conjugated and one allophycocyanin (APC)-conjugated tetramer to determine the epitope specificity of each TCC (indicated in Figure 1g).
The flow analyses were performed on Aria III cell sorter (BD Biosciences) for the tetramer-stained samples and BD Fortessa for the analyses of CD8þ and gd T cells, at the Flow Cytometry Core Facility at Oslo University Hospital (Oslo, Norway). Data analysis was done with FlowJo software (FlowJo LLC, Ashland, OR). The number of HLA-DQ2.5:gluten tetramer-positive effector-memory T cells was normalized to the number of million CD4þ T cells. Total PBMC was counted before enrichment, and the frequency of CD4þ T cells was analyzed in a preenriched sample.

Generation of TCCs
TCCs were generated by limiting dilution and antigen-free expansion. HLA-tetramer binding CD4þ T cells were sorted in a tube containing feeder mixture, which contains irradiated PBMCs (1 million/ml, 60 Gy) of three healthy donors in 10% human serum/RPMI medium with penicillin/ streptomycin, phytohemagglutinin (1 mg/ml), IL-2 (20 IU/ml), and IL-15 (1 ng/ml). This T cell/feeder mixture suspension was cultured on Terasaki plates for 10 days. Subsequently, growing TCCs were transferred to 48well plates containing 500 ml of feeder mixture and cultured for additional 10 days.