Association between EBV serological patterns and lymphocytic profile of SjS patients support a virally triggered autoimmune epithelitis

Sjögren's syndrome (SjS) is characterized by lymphocytic infiltration of exocrine glands, i.e. autoimmune epithelitis. Lymphocytes are central in SjS pathogenesis, with B-cell hyperactivity mediated by T-cells. B-cells are main targets of Epstein-Barr virus (EBV) infection, a frequently-suggested trigger for SjS. We aimed to evaluate how the EBV infection modulates B and T-cell subsets in SjS, including as controls Rheumatoid arthritis patients (RA) and healthy participants (HC). SjS patients presented decreased CXCR5+T-cells, although IL21-secreting Tfh and Tfc cells were increased. Tfc were positively correlated with ESSDAI scores, suggesting their relevant role in SjS pathogenesis. As previously described, SjS patients showed expanded circulating naïve B-cell compartments. SjS patients had a higher incidence of EBV-EA-D-IgG+ antibodies, characteristic of recent EBV-infection/reactivation. SjS patients with past infection or recent infection/reactivation showed increased CXCR3+Th1 and CXCR3+Tfh1 cells compared to those without active infection. SjS patients with a recent infection/reactivation profile presented increased transitional B-cells compared to patients with past infection and increased plasmablasts, compared to those without infection. Our results suggest EBV-infection contributes to B and T-cell differentiation towards the effector phenotypes typical of SjS. Local lymphocyte activation at ectopic germinal centres, mediated by Tfh and Tfc, can be EBV-driven, perpetuating autoimmune epithelitis, which leads to gland destruction in SjS.

B-cell subsets. Considering the IgD/CD27 classification, the percentages of IgD + CD27 − B-cells (naïve) were higher in SjS patients when compared to HC (p = 0.028) and RA patients (p = 0.043), and SjS patients also presented higher absolute counts of this subset compared to RA (p = 0.015). Total memory B-cells (CD27 + IgD +/− ) and unswitched memory B-cells (CD27 + IgD + ) were lower in SjS patients compared to HC (p = 0.001). However, only absolute counts of switched memory B-cells (CD27 + IgD − ) were lower in SjS compared to HC (p < 0.001), and no differences were observed towards RA patients.
Using the Bm1-5 classification, B-cells were classified as Bm1, Bm2, Bm2′, Bm3 + 4, eBm5, and Bm5 subsets 21 . The percentages of Bm1 cells were significantly lower in SjS compared to RA (p = 0.005) and HC (p = 0.008), and the absolute counts were also significantly lower in SjS (p = 0.002) compared to HC. Bm2 (naïve) and Bm2′ (transitional) percentages were significantly higher in SjS compared to RA patients (p = 0.015 for Bm2 and p = 0.041 for Bm2′), and Bm2′ absolute counts followed the same trend (SjS vs RA; p = 0.003). Lower percentages (p = 0.037) and absolute values (p < 0.001) of eBm5 cells were found in SjS when compared to HC. As for Bm5 cells, SjS patients presented lower percentages than RA (p = 0.011), and lower absolute counts than HC (p = 0.001). The results are summarized in Table 1 and Supplementary Table 2. EBV serological markers. All patients and controls were negative for anti-VCA IgM and anti-EA IgA, except for 1 HC that presented borderline levels for anti-EA IgA. All samples were positive for anti-VCA IgG, except for 2 SjS patients, who showed negative values for these antibodies. Most patients and HC showed positive values for anti-EBNA IgG (76.5% of SjS; 80.0% of RA and 85.0% of HC), and negative values for anti-VCA IgA (79.4% of SjS; 75.0% of RA and 80.0% of HC), without significant differences between groups. Interestingly, for anti-EA IgG, significant differences were observed between SjS patients and HC (32.4% in SjS; 20.0% in RA; 5.0% in HC). The results are presented in Table 2.

EBV serological patterns in SjS patients.
Recognizing that SjS patients presented an increased prevalence of Anti-EBV EA-D IgG and also an important presence of anti-EBNA IgG, we further divided these patients into 3 subgroups according to the serological EBV profile observed: G1 (n = 18), previous infection (EA-IgG-, EBNA IgG +); G2 (n = 11), recent infection/reactivation (EA IgG + , EBNA IgG + /-), and G3 (n = 5), no  20 . Demographic and clinical data are presented in Table 3. SjS patients with recent infection/reactivation markers (G2) had earlier disease manifestations and shorter disease duration.
These data are presented in Table 4 and Supplementary Table 3.

Discussion
Our study aimed to explore the relation between the EBV serological profile of SjS patients and the distribution of circulating B and T-lymphocyte subsets. First, we report interesting differences in follicular T-cell subsets between SjS patients and both HC and RA patients. Despite circulating CXCR5 + T cell subsets were decreased in SjS patients, functionally IL21-secreting CD4 + (Tfh) and CD8 + (Tfc) T cells seem to be more pronounced in these patients. IL21-secreting CD8 + T cells (Tfc) were even positively correlated with ESSDAI scores, suggesting their relevant role in SjS pathogenesis. Moreover, we confirmed the enriched circulating naïve B-cell compartment of SjS patients (compared to both control groups, healthy and autoimmune), previously reported in the literature 5 .
The major observation of our study, however, comes from the EBV profile, with SjS patients presenting a greater incidence of EBV-EA-D-IgG positivity, a profile characteristic of recent infection/reactivation of EBV infection. Furthermore, SjS patients with either serological evidence of past EBV infection or recent infection/ reactivation presented higher values of CXCR3 + CD4 + T cells (Th1) and CXCR3 + CXCR5 + CD4 + T cells (Tfh1) compared to those without serological evidence of active infection. Also, the B-cell compartment was distinctive in SjS patients with signs of recent EBV infection/reactivation: showing higher levels of transitional Bm2′ cells compared to patients with past infection and increased plasmablasts, compared to patients without serological evidence of infection.
The factors underlying the onset and development of SjS are still uncertain. Nevertheless, typical immune profiles have been characterized in these patients, which can be relevant to unveil important links to other triggering players in this autoimmune disease. Despite B-cells are the main target for EBV latent infection, T-cells have also a role in this play, and have been studied in autoimmune diseases for which EBV is considered a potential trigger. For instance, EBV-specific CD8 + T-cells are increased during B-cell transformation and in the productive viral replication phases of EBV in infected RA 22 and SLE patients 23 . Additionally, the EBV-specific CD8 + T-cell pool is reduced by immunosuppressive therapy 24 .
Our study supports the presence of a promoted follicular T-cell environment in SjS patients, traduced by the increased secretion of IL21 by both CD4 + and CD8 + T-cells. We found no differences in the percentages of circulating CXCR5 + follicular T-cells between groups, and absolute counts for this subset were even decreased www.nature.com/scientificreports/ in SjS patients, possibly due to the decreased absolute counts of CD4 + T-cells observed in these patients. Similar data had been described in the work of Brokstad 25 , which reported no differences for total CXCR5 + CD4 T-cell percentages, but only changes in particular subsets of these cells in SjS patients, such as the increase in Tfh-like ICOS + PD-1 + cells. Interestingly, in our study, another Tfh-like subset was increased in SjS patients, the IL21 + Tfh cells. Indeed, both CD4 + and CD8 + T-cells were more prone to produce IL21, the Tfh modulating cytokine. If the lower absolute numbers may indicate retention of CXCR5 + follicular T-cells at the exocrine glands, as supported by previous studies showing a T-cell predominance in lymphocytic infiltrates of these organs 26 , SjS patients seem to be predisposed to promote Tfh differentiation. Also, when naïve T-cells and salivary gland epithelial cells are co-cultured, Tfh differentiation is observed, i.e. T-cells acquire a classical Tfh phenotype and are able to secrete IL21 27 . Thus, this systemic follicular function may be overexpressed in SjS patients, also as an effect of the local altered interplay. We have previously reported that the ESSDAI score, which is a measure of disease activity in SjS, seemed to be correlated with IL21 + CD8 + T cells (Tfc) levels 28 . In fact, patients with more active disease present increased circulating Tfc cells, though the causative link between these observations is still to be clarified (i.e. whether higher levels of Tfc cells lead to increased disease severity or, on the contrary, happen in response to disease aggravation).
In addition, the increase in IL21-expressing T-cells resembles the profile of a chronic active viral infection, as proposed by Fahey and collaborators, who showed that viral persistence redirects T-cell differentiation towards the Tfh profile in animal models 19 . Moreover, patients with infectious mononucleosis show an increase in a particular subset of Tfh cells in peripheral blood 29 , which supports our hypothesis that viral triggers may take part in the modulation of the immune responses also in SjS patients 29 .
Interestingly, Fahey and colleagues 19 proposed that viral-induced Tfh cells deviate from an original Th1 profile. In line with this, we observed that SjS patients with serological evidence for recent infection/reactivation presented increased Th1 and Tfh1 subsets. Thus, the autoimmune background of SjS patients could provide T-cells with alternate activation signals leading them to assume both Th1 and Tfh1 profiles under viral persistence, since it is accepted that the pathogenesis of SjS is mediated by Th1-derived responses 27 .
Strikingly, the implication of Tfc cells in SjS pathogenesis is supported by their increase in patients with higher disease activity. In line with our results, serum levels of IL21 had already been associated with systemic disease activity in SjS 30 , but our results seem to highlight a role for CD8 T-cells in this scenario. Initially, CXCR5 + CD8 T-cells were described as early effector memory CD8 T-cells present in B-cell follicles of human tonsils 13 . Recently these cells were implicated in the control of chronic viral infections 29,31,32 . Also, associations between humoral responses and CXCR5 + CD8 T-cells 32,33 were identified, as these cells express co-stimulatory molecules. In fact, increased immunoglobulin production by B-cells occurs when they are co-cultured with CXCR5 + CD8 T-cells, suggesting these cells have other immune functions besides cytotoxic activities 33 . Considering that dysregulated humoral responses are present in SjS, Tfc cells, along with Tfh cells, may induce the atypical antibody production of SjS patients. Nevertheless, the major function of the follicular CD8 + T-cells may still be limiting the replication of viral agents in B-cell follicles, as these cells show increased cytotoxic capacities 34 .
As for B-cells, it is accepted they are EBV's main target 35 . Several studies tried to relate SjS pathogenesis with a specific clonality of B-cells. One of the hallmarks of Sjögren's syndrome is, in fact, the formation of ectopic lymphoid structures (ELS) in the SG. ELS are composed of B-cell/T-cell follicles, supported by networks of stromal follicular dendritic cells, which support ectopic GC reactions 36 . Active EBV infection has been associated with ELS in the SG of SjS patients and appears to contribute to local growth and differentiation of diseasespecific autoreactive B-cells 37 . Despite the possibility of an EBV-triggered B-cell proliferation in SjS, EBV-infected memory B-cells were found to express lower levels of self-and poly-reactive antibodies than their uninfected counterparts 38 .
As corroborated by our data, SjS patients present a typical circulating B-cell compartment, enriched in transitional/naïve subsets, in opposition to memory subsets. If we consider the observations from Coleman and colleagues on the effect of EBV in B-cells 39 , we may also suggest a possible role for EBV in the alterations observed in the B-cell compartment of SjS patients. In fact, these authors have recognized that murine transitional B-cells from the spleen can be reservoirs for gammaherpesvirus like EBV, which can remain latent in these cells, prolonging their life span indefinitely. Our results are in line with this hypothesis, as SjS patients with serological evidence of recent infection/reinfection presented higher percentages of transitional Bm2′. Whether this is an effect of EBV or other concomitant viral infection, remains to be elucidated. However, the viral input for this feature of SjS patients can be also supported by our observation that transitional B cells were particularly increased in patients with recent infection/reactivation. We acknowledge that the assessment of the viral genome in different B-cell subsets could clarify this idea.
Regarding the serological EBV markers, we found an increased prevalence of Anti-EBV EA-D IgG in SjS patients, compared to both RA and HC, as described in previous works 37,40 . The anti-EBV EA-D IgG prevalence in our SjS patients (about 33% vs 5% in HC) was very close to the one reported by Pasoto and colleagues 41 (36% vs 4.5% in HC), which strengths our data, and led us to further assess the immune compartments according to the EBV serological profile of SjS patients. Interestingly, patients with evidence of EBV infection, and particularly those with recent infection/reactivation (EA-D IgG positive) had earlier disease manifestations, but also a distinct immune profile, with a shift towards pro-inflammatory Th1/Tfh1 subsets in the T-cell compartment. Furthermore, SjS patients with evidence of recent infection/reactivation exhibited higher levels of transitional B-cells and plasmablasts, which may traduce the importance of EBV in the modulation of the immune responses in SjS patients, with possible clinical impact, as suggested by an earlier onset of clinical manifestations. The effect of cytotoxic T-cell (CTL) responses, with both CD8 + and CD4 + T cells, or even other unconventional T cell subsets, may restrict the expansion of latently infected B-cells in long-term carriers or patients with past infection 42,43 . This later immune balance may be the cause for the differences observed in transitional B-cells between SjS patients with recent infection/reinfection and patients with previous infection (G1). Usually increased in SjS, these cells www.nature.com/scientificreports/ represent a potential EBV reservoir, and are markedly increased in G2 patients, showing however a relapse to lower values in past infections as an effect of an effective immune control happening in these patients (G1).
To better comprehend how EBV profiles modulate immune populations, or whether the changes are SjSdriven, a comparison between SjS patients and HC with similar EBV serological patterns would be very helpful. However, considering the reduced number of HC that could be included in the EBV subgroups G1 (3 HC) and G2 (1 HC), such analysis was not possible. In the future, we aim to better address this question, extending the study to a larger cohort, with more SjS patients, but also healthy controls.
Clinical differences between groups were also difficult to assess due to the small size of our patients' groups. The ESSDAI score in both groups of patients with positive EBV serology (G1 and G2) was higher than in EBVnegative patients (G3). G1 patients had a non-significantly higher ESSDAI compared to G2 patients, in line with a recent report by Sanosian 44 , who didn't find distinct ESSDAI scores in anti-EBV EA-positive compared to anti-EBV EA-negative patients. No differences were found in specific clinical ESSDAI domains between G1 and G2 groups, contrasting to the report of Pasoto 41 , who found higher articular activity in anti-EBV EA-positive patients. Nevertheless, the greater frequency of active disease in our G2 patients may suggest an influence of recent EBV infection/reactivation in disease activity status, which may be further supported by the higher levels of IgG observed in this group, despite no statistical significance was achieved. Furthermore, our observations on the B-cell compartment are in line with this, as the increased transitional Bm2′ and plasmablasts observed in the EBV infection/reactivation group also suggests a higher B-production and differentiation on these patients, possibly contributing in parallel for the increased gammaglobulin levels and greater disease activity.
Interestingly, in our study, IgA and IgM EBV-antibodies behaved similarly in SjS, RA and HC, with a predominance of negative samples for these biomarkers in all groups. As for VCA IgM, an acute-phase marker, despite it may be present in different viral scenarios, no differences were expected in SjS according to previous reports 45 .
In other systemic autoimmune diseases, particularly SLE, increased levels of IgA antibodies against the two lytic antigens studied (VCA and EA) have been reported 45,46 . Literature is scarce for this evaluation in SjS patients, nonetheless, both VCA-IgA and EA-IgA seem to be less present in SjS that in other autoimmune conditions such as SLE. Along with a few previous studies 45,46 , our data suggest a lower mucosal immune response against EBV in SjS, compared to other autoimmune conditions, with anti-EBV IgA antibodies prevalence similar to HC. Whether this corresponds to a better or worst control of latent infections in SjS remains to be elucidated.
To our knowledge, our study is the first to report an association between EBV serological patterns and the immune profile of SjS patients, despite EBV EA (early antigen) had already been correlated with autoantibodies production 40 . Although we observed no further differences in the clinical manifestations of SjS patients according to their EBV serology, we were able to identify distinct immune profiles according to the EBV serological pattern. Still, we must recognize the low number of patients considered and the absence of other confirmatory methodologies for the effective viral infection. Nonetheless, our data support the idea that different EBV serological profiles affect circulating B and T-cells in SjS patients. In fact, a more active serological background, as the ones observed in groups G1 and G2, may suggest a viral influence in the immune system driving it to the more pro-inflammatory scenario observed in SjS patients when compared to both other autoimmune conditions or healthy controls.
Other authors have supported the hypothesis that reactivation in the lytic phase of EBV infection promotes immunological dysfunction in SjS 37 . Considering our results, we also believe special attention should be given to the group of SjS patients with serological evidence for recent infection/reactivation, which present a proinflammatory profile, with increased Th1/Tfh1 ratio cells along with elevated transitional B-cells and increased plasmablast differentiation.
We acknowledge limitations in our study, such as the absence of standard molecular biology assays to confirm EBV infection. In future studies, it would be relevant to assess not only serology, but also EBV viral load, and eventually other viruses with potential impact in SjS development, such as CMV. Also, our study was performed exclusively in peripheral blood, and we realize it may not properly reflect the numbers and interactions of immune cells at exocrine glands. For instance, SG biopsies would not only clarify the hypotheses on cell traffic between affected organs and the circulating lymphocyte pool but would also allow us to prove the presence of EBV in such organs.
Nevertheless, from our results, it is possible to suggest that EBV plays a role in inducing B and T-cells towards an effector phenotype. EBV enters the replicating phase in the exocrine glands, where this facilitated interaction between EBV antigens and effector T-cells might lead to a breakdown of tolerance. The ensuing autoimmune response mediated by effector B and T-cells might lead to a localized lymphocyte activation with the formation of ectopic GC or GC-like structures. This process, mediated by Tfh and Tfc, can thus perpetuate the autoimmune epithelitis and result in gland destruction.
Our work provides a new perspective on how EBV might be involved in lymphocytic alterations known to be a feature in SjS. Clarifying the role of follicular CD4 and CD8 T-cells in the context of viral infection can be of great value in confirming a viral-triggered autoimmune response in SjS, but a specific strategy for the characterization of these cells in peripheral blood and target organs is still needed.
Our study can also constitute a starting point for approaching the role of CXCR5 + and IL21 + CD8 T-cells (Tfc) in the context of autoimmunity. The association between CXCR5 + CD8 T-cells and disease activity in SjS observed in our study may be an indicator of their involvement in the pathophysiology of autoimmune epithelitis. In the current scenario where Tfc cells involvement in autoimmune pathologies is yet to be elucidated, our study pioneers the association of Tfc cells with human autoimmunity and paves the way for further studies regarding Tfc cells in autoimmune diseases. Indeed, considering the possible pathogenic role of EBV in the pathogenesis of SjS, therapies directed towards the interaction between EBV and activated effector T-cells and B-cells could halt the EBV-triggered lymphocytic activation, and have a relevant clinical applicability.  49 . Clinically active disease was defined as activity in any ESSDAI domain, except the hematologic and biologic. The healthy control group (HC) consisted of women without symptoms or signs of xerostomia or xerophthalmia, or any history of autoimmune rheumatic diseases, selected from the Ophthalmology outpatient clinic of Hospital CUF Descobertas.
Informed consent was obtained from all participants. The study was approved by the Ethics committees of both recruiting institutions, and NOVA Medical School Ethics Committee (no. 17/2016/CEFCM). CellQuest Pro (BD Biosciences) software was used for acquisition and analysis purposes and Infinicyt 2.0 (Cytognos S.L., Salamanca, Spain) software was also used for more differentiated subset analysis.
Whenever appropriate, fluorescence-minus-one control tubes were prepared to assess the positivity of dimer expressions. The subsets analyzed, and the respective gating strategies, are displayed in Fig. 1. Within T-cells, we characterized CD4 + and CD8 + (CD4-) subsets, including CXCR5 + Tfh and Tfc cells, and the Tfh1 and Tfh17 profiles, according to the expression of CXCR3 and CCR6, respectively. B-cells' subsets were addressed according to the classical IgD/CD27 classification, and the Bm1-5 classification, often used in autoimmunity settings 21 . Functional assays for the evaluation of IL21 production by T-cells. Heparinized peripheral blood samples were used to assess IL21 and IL-17 production by CD4 + and CD8 + T-cells.
In brief, cells were stimulated with PMA and ionomycin, for 5 h at 37ºC in a 5% CO 2 atmosphere in the presence of brefeldin-A. After stimulation, cells were lysed, washed and incubated with anti-CD3 and anti-CD8 mAbs for surface staining. For intracellular stain, cells were treated according to the protocol defined by the manufacturer for the BD Fixation/Permeabilization Solution Kit with BD GolgiPlug™ (BD Biosciences) and then marked with anti-IL21 and anti-IL-17 mAbs, after cell fixation and permeabilization. For each patient, stimulated and unstimulated tubes were run in parallel to assure proper stimulation and staining controls. Gating strategy is presented in Fig. 2, with IL21 + (IL17 − ) cells being identified within CD8 + and CD8 − T cells, respectively considered as Tfc and Tfh cells.
EBV serological markers. Enzyme-linked immunosorbent assays (ELISA) were used for the assessment of IgG, IgA and IgM antibodies (Abs) against EBV antigens (Ags). All ELISA kits were obtained from Euroimmun (Euroimmun, Luebeck, Germany) and used according to the manufacturers' instructions. The following Abs for EBV Ags were determined: IgG for diffuse early Ag (EA-D), IgG for viral capsid Ag (VCA), IgG for nuclear Ag-1 (EBNA1), IgA for EA-D, IgA for VCA and IgM for VCA. All tests for IgG Abs were quantitative, while IgA and IgM were semiquantitative. In quantitative assays, sample concentration was determined using 3-point calibration curves constructed with ELISA-Logit software, available at https ://ednie uw.home.xs4al l.nl/Calib ratio n/ Logit /Logit .htm (V24May2017). The cut-off level for all IgG antibodies assayed was 20 RU/ml. For semiquantitative assessments, a single calibrator was determined in triplicate per assay. The ratio sample/calibrator was used www.nature.com/scientificreports/ to assess positivity levels (Negative: ratio < 0.8; Borderline: ratio ≥ 0.8 to < 1.1; Positive: ratio ≥ 1.1). Patients were randomly assigned to undergo EBV serology evaluation.
Statistics. Graph Pad Prism™ 6.0 (Graph Pad Software, San Diego, CA, USA) was used for statistical analysis. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.