Received: 29 October 2018 | Revised: 1 April 2019 | Accepted: 7 April 2019 DOI: 10.1111/cns.13142 ORIGINAL ARTICLE Optimal response to dimethyl fumarate is mediated by a reduction of Th1‐like Th17 cells after 3 months of treatment María José Mansilla1,2 | Juan Navarro‐Barriuso1,2 | Silvia Presas‐Rodríguez3,4 | Aina Teniente‐Serra1,2 | Bibiana Quirant‐Sánchez1,2 | Cristina Ramo‐Tello3,4 | Eva María Martínez‐Cáceres1,2 1 Immunology Division, LCMN, Germans Trias i Pujol University Hospital and Abstract Research Institute, Campus Can Ruti, Aim: Dimethyl fumarate (DMF) is one of the most promising therapies for relaps‐ Badalona, Spain 2 ing‐remitting multiple sclerosis (RRMS) patients since it has shown immunomodula‐ Department of Cellular Biology, Physiology and Immunology, Universitat Autònoma de tory and neuroprotective effects. However, a percentage of RRMS patients do not Barcelona, Cerdanyola del Vallès, Spain exhibit an optimal response to DMF. The objective of this study was to identify early 3 Multiple Sclerosis Unit, Department of Neurosciences, Germans Trias i Pujol biomarkers of treatment response by analyzing changes in peripheral leukocyte sub‐ University Hospital, Badalona, Spain populations directly in whole blood samples. 4 Department of Medicine, Universitat Methods: A longitudinal and prospective study analyzing peripheral blood leukocyte Autònoma de Barcelona, Cerdanyola del Vallès, Spain subpopulations in 22 RRMS patients before initiating DMF treatment (baseline) and at 1, 3, 6, and 12 months of follow‐up was performed. Differences between no evi‐ Correspondence Eva María Martínez‐Cáceres, Immunology dence of disease activity (NEDA) and ongoing disease activity (ODA) patients were Division, LCMN, Edifici IGTP‐MAR, planta analyzed. 2ª, Camí de les Escoles s/n, 08916 Badalona, Barcelona, Spain. Results: The beneficial effect of DMF was associated with a specific depletion of Email: emmartinez.germanstrias@gencat.cat memory CD4+ and CD8+ T lymphocytes and B cells. Importantly, only NEDA patients Cristina Ramo‐Tello, Germans Trias i Pujol Universitari Hospital, Multiple Sclerosis Unit, showed (a) a shift from a pro‐ to an antiinflammatory profile, with an increase of Th2 Carretera del Canyet s/n, 08916 Badalona, cells and a decrease of Th1‐like Th17 lymphocytes; and (b) an increase of regulatory Barcelona, Spain. Email: cramot.germanstrias@gencat.cat CD56bright NK cells. Conclusion: The optimal response to DMF is mediated by a shift to antiinflammatory Funding information Plan Nacional de I+D+I, Grant/Award and immunoregulatory profile, which puts forward Th1‐like Th17 lymphocytes as a Number: PI14/01175; ISCIII‐Subdirección potential early biomarker of treatment response. General de Evaluación; Fondo Europeo de Desarrollo Regional; Agència de Gestió d'Ajuts Universitaris i de Recerca, KEYWORDS Grant/Award Number: 2017 SGR 103; biomarkers, dimethyl fumarate, immunomonitoring, multiple sclerosis, Th1‐like Th17 Government of Catalonia, Grant/Award lymphocytes Number: 02/2015-01/2018; EU Framework Programme Horizon 2020 Cristina Ramo‐Tello and Eva María Martínez‐Cáceres: shared senior coauthorship. This is an open access article under the terms of the Creat​ive Commo​ns Attri​bution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2019 The Authors. CNS Neuroscience & Therapeutics Published by John Wiley & Sons Ltd. CNS Neurosci Ther. 2019;25:995–1005.  wileyonlinelibrary.com/journal/cns | 995 996 | MANSILLA et al. 1 | I NTRO D U C TI O N 2 | PATI E NT S A N D M E TH O DS Multiple sclerosis (MS) is a chronic immune‐mediated disease 2.1 | Patients and study design characterized by inflammation, demyelination, and axonal damage A total of 22 RRMS patients from the MS unit of the Germans Trias in the central nervous system (CNS). During the last years, numer‐ i Pujol Hospital (Spain), fulfilling McDonald's 2010 criteria, were ous new immunomodulatory drugs have been approved for the included in the study before initiating DMF treatment. Clinical, ra‐ treatment of MS patients. From them, delayed‐release dimethyl diological, and immunological monitoring of patients was conducted fumarate (DMF) is one of the most promising new treatments from baseline (before treatment) until 12 months of treatment. for relapsing‐remitting MS (RRMS) patients, since it has showed Clinical and demographic characteristics of the patients are summa‐ both antiinflammatory and neuroprotective properties. Clinically, rized in Table 1. DMF treatment reduces the annualized relapse rate (ARR) and the Clinical data included disability progression (defined as wors‐ number of new lesions measured by magnetic resonance imaging ening of 1 point or more on the Expanded Disability Status Scale (MRI).1,2 [EDSS] score) and disease relapses (defined as new or worsening Nowadays, due to the numerous disease‐modifying therapies neurological deficits lasting 24 hour or more in the absence of fever (DMT) available for MS patients, the selection of the proper treat‐ or infection) after 12 months of treatment. ment for each patient is getting more difficult. Currently, the deci‐ Radiological activity was measured by determining the presence sion is based on the balance between the potential clinical benefits of new or enlarging T2 lesions and/or gadolinium‐enhancing T1 le‐ and the adverse effects of each treatment. However, it is known sions (Gd+) in brain MRI. Either 1.5 or 3.0 Tesla MRI scans were used that a percentage of patients exhibit a suboptimal clinical response to evaluate the number of Gd+ lesions before and after 12 months to some DMT. Research groups around the world have been inves‐ of follow‐up. The same equipment was applied for each patient. The tigating the mechanism of action of each DMT, trying to identify number of Gd+ lesions was visually determined, comparing the first molecules or biomarkers of treatment response. Dimethyl fumarate and second MRI scans. mediates immunomodulatory as well as antioxidative and cytopro‐ Patients were classified as NEDA (no evidence of disease activ‐ tective effects through the inhibition of the transcription factor ity)—defined as the absence of relapses, disability progression, and nuclear factor kappa b (NF‐κB) and promoting the activation of radiological activity during the follow‐up—or ODA—when an in‐ the nuclear factor‐erythroid 2‐related factor (Nfr2) transcriptional crease in clinical and/or radiological activity was reported. pathway.3-6 Blood samples were obtained from patients at baseline and after Studies focused on the mechanism of action of DMF demon‐ 1, 3, 6, and 12 months of treatment, to determine changes in lympho‐ strated that the treatment induces lymphopenia, selectively cyte subpopulations related to an optimal clinical response to DMF. depleting memory T and B cells and promoting a shift toward a The study protocol was approved by the ethics committee of tolerogenic immune profile.7-12 Specifically, Medina and colleagues the Germans Trias i Pujol Hospital, and all the patients gave their reported that DMF induces a change in the ratio between reg‐ informed consent to participate in the study. ulatory CD56 bright NK cells and effector TNF‐α‐producing CD8+ T cells, which is required for the efficacy of the treatment. The authors also suggested that monitorization of leukocyte subpopu‐ 2.2 | Flow cytometry analysis lations could be a useful tool for early identification of responder and nonresponder patients.12 2.2.1 | Analysis of peripheral leukocyte In this study, we performed a longitudinal and extensive subpopulations characterization of whole blood immunophenotype of RRMS For the analysis of the different peripheral blood leukocyte sub‐ patients before and after 1, 3, 6, and 12 months of DMF treat‐ populations, whole blood samples, obtained by standard venipunc‐ ment by flow cytometry. Changes in lymphocyte subpopula‐ ture in EDTA tubes, were kept at room temperature and processed tions in patients with no evidence of disease activity (NEDA) in within the next 24 hour. Briefly, 100µl of peripheral blood was incu‐ comparison with patients with ongoing disease activity (ODA) bated for 20 minutes at room temperature and protected from light, were analyzed in order to identify early biomarkers of treat‐ with the appropriate combination of antibodies (see below) to ana‐ ment response. This technique is an easy and fast way to di‐ lyze T‐cell, Treg, B‐cell, and DC/monocyte/NK‐cell subpopulations. rectly analyze (without the limitations of leukocyte isolation or After erythrocyte lysis and cell fixation, samples were washed twice cryopreservation) the ex vivo effect of DMF or other treatments and acquired on a LSR II Fortessa cytometer and analyzed using in leukocyte subpopulations. Using this protocol, we identified FACSDiva software (BD Biosciences). Cell markers used to define that a reduction of effector memory Th1‐like Th17 cells is asso‐ each leukocyte subpopulations are specified in Table S1. The gating ciated with increased DMF efficacy, which occurs already after strategy used is based on an international consensus and was previ‐ 3 months of treatment in NEDA patients and it is maintained ously described by Teniente‐Serra and Quirant‐Sánchez.13,14 until 12 months of treatment. MANSILLA et al. | 997 TA B L E 1 Clinical and demographic Total MS patients NEDA ODA characteristics of multiple sclerosis patients n 22 15 6 Age at baseline 36.7 ± 7.42 38.0 ± 7.53 33.8 ± 7.47 Gender (% female) 15/22 (68.2%) 11/15 (73.3%) 3/6 (50.0%) Disease duration (y) 4.18 ± 4.72 4.60 ± 5.12 3.67 ± 4.13 Prior treatment Naïve 15 11 3 IFN‐β 4 3 1 Fingolimod 1 – 1 Aubagio 1 – 1 Othersa 1 1 – Baseline ARR 1.14 ± 0.62 0.99 ± 0.48 1.39 ± 0.83 Relapses during the previ‐ 1.50 ± 0.91 1.33 ± 0.90 1.83 ± 0.98 ous year Relapses after 12 mo of 0.15 ± 0.37 – 0.50 ± 0.55 DMF treatment Baseline EDSS 2.34 ± 1.47 2.27 ± 1.58 2.75 ± 1.21 EDSS after 12 mo of DMF 2.26 ± 1.93 2.04 ± 1.84 3.17 ± 1.97 treatment Baseline T2 lesions <9 T2 lesions 5/22 (22.7%) 4/15 (26.7%) 1/5 (20.0%) >9 T2 lesions 17/22 (77.3%) 11/15 (73.3%) 5/6 (80.0%) New T2 lesions after 12 mo 4/21 (19.04%) – 4/6 (66.7%) of treatment (%) Presence of gadolinium‐en‐ 10/21 (47.6%) 6/15 (40.0%) 4/6 (66.7%) hancing lesions at baseline New gadolinium‐enhancing 1/21 (0.05%) – 1/6 (16.7%) lesions after 12 mo of DMF treatment Abbreviations: ARR: annual relapse rate; EDSS: Expanded Disability Status Scale; NEDA: no evi‐ dence of disease activity; ODA: ongoing disease activity. a Others: cell therapy clinical trial. Leukocyte subpopulations were analyzed by using the follow‐ samples was incubated with anti‐CD45‐PerCP (BD Biosciences) ing four panels of monoclonal antibodies, previously reported by for 20 min at room temperature and protected from the light. 14 Quirant‐Sánchez et al : (a) T‐cell panel: CD3‐V450, CD4‐PerCP‐ Erythrocytes were removed using 450 µL lysis buffer (BD FACS Cy5.5, CD45RA‐PE‐Cy7, CCR7‐PE, CD38‐APC, CD8‐APC‐H7, Lysing Solution; BD Biosciences). In order to quantify the ab‐ HLA‐DR‐V500 (BD Biosciences), CXCR3‐AF488, CCR6‐BV605, and solute number of cells, Perfect‐Count Microspheres (Cytognos CD45‐AF700 (BioLegend); (b) Treg panel: CD4‐PerCP‐Cy5.5, CD25‐ SL) were added to each sample in the same proportion of PE, CCR4‐PE‐Cy7, CD127‐AF647, CD45RO‐APC‐H7, CD3‐V450, blood, 25 µL (ratio 1:1), and acquired on a LSR II Fortessa cy‐ HLA‐DR‐V500 (BD Biosciences), and CD45‐AF700 (BioLegend); (c) tometer and analyzed using FACSDiva software (both from BD B‐cell panel: CD24‐FITC, CD19‐PerCP‐Cy5.5, CD38‐APC, CD20‐ Biosciences). The number of total lymphocytes was expressed APC‐H7, CD3‐V500 (BD Biosciences), IgD‐PE‐Cy7, CD27‐BV421, in cells/µL. and CD45‐AF700 (BioLegend); and (d) DC/monocyte/NK panel: CD3 + CD19‐APC‐H7, CD56‐PE, CD16‐APC, CD14‐V450, CD123‐ 2.3 | Statistical analysis PerCP‐Cy5.5, CD11c‐PE‐Cy7, HLA‐DR‐V500 (BD Biosciences), and slan‐FITC (Miltenyi Biotec). Statistical analyses were performed using GraphPad Prism version 6.00 for Windows. Parametric and nonparametric tests were used Quantification of absolute cell numbers depending on the normality of the distribution of the variables. For quantification of absolute lymphocyte and peripheral blood Differences between samples obtained at different time points mononuclear cell (PBMC) counts, 25 µL of peripheral blood were compared to baseline measurements using the repeated 998 | MANSILLA et al. measures 1‐way ANOVA test with Dunnett's or Dunn's correction 3 | R E S U LT S for multiple comparisons. To compare data from NEDA and ODA groups, Mann‐Whitney U or t tests were applied. Fisher's exact 3.1 | Patients test was used to compare qualitative variables. Results were ex‐ A total of 22 MS patients were enrolled in the study. To determine pressed as mean ± standard deviation (SD) values, unless other‐ the effects of DMF related to an optimal response, 21 out of 22 pa‐ wise stated. Differences were considered statistically significant tients were classified as NEDA (n = 15) or ODA (n = 6) patients. One when P < 0.05. F I G U R E 1 Changes induced by DMF treatment on T‐ and B‐lymphocyte subpopulations. Representation of percentages (%) of central memory (CCR7+ CD45RA−), effector memory (CCR7− CD45RA−), and naïve (CCR7+ CD45RA+) subpopulations from CD4+ and CD8+ T lymphocytes (A and B, respectively); and % of preswitched memory (CD27+ IgD+), class‐switched (CD27+ IgD−) and (CD27− IgD+) naïve CD19+ B lymphocytes (C), before the treatment (baseline) and after 1, 3, 6 and, 12 months of follow‐up. Data are expressed as mean SD. Each dot represents the value of an individual (n = 22). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 MANSILLA et al. | 999 patient was excluded from this stratification due to the lack of MRI number of lymphocytes (P = 0.0005). This reduction was statisti‐ at 12 months. No differences were found on baseline characteristics cally significant after 6 months of treatment and remained after between groups, and, as expected, the ODA group showed a higher 12 months of follow‐up (Figure S1A and Table S2). number of relapses (P = 0.0016) and an increased presence of new When the effect of DMF on T cells was investigated, a decrease in T2 lesions (P = 0.0025) after 12 months of follow‐up (Table 1). the percentage of CD3+ T cells was found (P < 0.0001) following just 3 months of treatment (Figure S1B). Further analyses revealed that the number of both CD8+ and CD4+ T cells was reduced although CD4+ lym‐ 3.2 | Optimal response to DMF is related to a phocytes were decreased in a lesser extent (Table S2). Consequently, reduction of memory CD4+ T cells the proportion of CD4+ T cells and the ratio of CD4+/CD8+ T lympho‐ As has been widely described, the analysis of whole blood samples cytes were significantly increased at 6 and 12 months of DMF treat‐ revealed that DMF induced a progressive decrease in the absolute ment (P < 0.0001 and P = 0.0001, respectively) (Figure S1C). F I G U R E 2 DMF specifically depleted memory T‐ and B‐lymphocyte subpopulations in  NEDA patients. Representation of percentages (%) of central memory (CM), effector memory (EM), naïve, and terminally differentiated RA (EMRA) CD4+ and CD8+ T lymphocytes from patients with no evidence of disease activity (NEDA) (A, C) and patients with ongoing disease activity (ODA) (B, D), before the treatment (Baseline) and after 1, 3, 6, and 12 months of follow‐up. In addition, % of preswitched memory, class‐switched, naïve, and double negative B lymphocytes from NEDA and ODA patients (E, F) was analyzed before and after treatment. Data are expressed as mean of each lymphocyte subpopulation of NEDA (n = 15) and ODA (n = 6) patients. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 1000 | MANSILLA et al. F I G U R E 3 Transitional B cells increased in  NEDA patients following DMF treatment. Representation of percentages (%) of transitional B lymphocytes (CD27− IgD+ CD24hi CD38hi) from the total of B cells (CD3− CD19+ cells), before the treatment (Baseline) and after 1, 3, 6, and 12 months of follow‐up in total RRMS patients (n = 22) (A), and after been classified in NEDA (n = 15) (B) and ODA (n = 6) (C) groups. Data are expressed as mean SD. Each dot represents the value of an individual. *P < 0.05, **P < 0.01 and ***P < 0.001 A deeper analysis on T‐lymphocyte subpopulations showed a 3.3 | DMF reduces Th1‐like Th17 lymphocytes in progressive depletion of T central memory (CD4: P = 0.0106; CD8: NEDA patients P < 0.0001) and effector memory cells (CD4 and CD8: P < 0.0001), resulting in an increase of the relative percentage of naïve CD4+ When the pro‐ and antiinflammatory effects of DMF were evalu‐ and CD8+ lymphocytes (CD4 and CD8: P < 0.0001) (Figure 1A,B ated (Figure S2) by analyzing whole blood samples, a reduction of and Table S3). These changes were statistically significant at 6 and the Th1‐like Th17 population was found after 6 and/or 12 months of 12 months of treatment for effector memory CD4+ and CD8+ T cells. treatment in central memory and effector memory cells, respectively + + In the case of central memory CD8 and CD4 T lymphocytes, dif‐ (P < 0.0001 in both subsets, Figure S3). Importantly, an increase in ferences were found after 3 or 12 months of follow‐up, respectively. the percentage of central and effector memory Th2 lymphocytes was Interestingly, when patients were classified as NEDA or ODA, the also observed (P < 0.0001 and P = 0.0065, respectively). This antiin‐ effects caused by the DMF treatment on CD4+ and CD8+ T‐lympho‐ flammatory effect was significantly relevant after 3 months of treat‐ cyte subpopulations were only observed in NEDA patients, but not ment for central memory Th2 cells and after 12 months in the case of in ODA patients, with the exception of CD8+ central memory cells, effector memory Th2 cells (Table S3). Surprisingly, the analysis exhib‐ which were reduced in both groups of patients (Figure 2A‐D and ited an increase of effector memory Th17 cells after 6 and 12 months Table S3). In addition, NEDA patients exhibited a significant reduc‐ of follow‐up (P < 0.0001). However, when the global pro‐ and antiin‐ tion of both CD4+ and CD8+ central memory T lymphocytes after flammatory balance was compared in central and effector memory 3 months of treatment (P < 0.0001 and P = 0.0021, respectively) cells, a progressive decrease of the pro‐inflammatory cells was evi‐ (Figure 2A,C and Table S3). denced (P < 0.0001 and P = 0.0065, respectively) after 3 months of The analysis of B‐cell subsets revealed comparable results to DMF treatment on central memory cells and after 12 months on ef‐ those found in T‐cell subpopulations (Tables S4 and S5). Although fector memory cells (Figure 4A,B, respectively) (Tables S2 and S3). the total percentage of B cells remained stable (Figure S1C), DMF The stratification of patients according to their clinical and ra‐ induced a specific depletion of memory B cells, both switched and diological activity revealed similar effects of DMF on the different preswitched class cells (P < 0.0001), thus increasing the percentage pro‐ and antiinflammatory lymphocytes subpopulations, being only of naïve B cells (P < 0.0001) (Figure 1C). Moreover, transitional B statistically significant in NEDA patients (Figure 4C,D). It is worth cells were already found increased in both, percentage and absolute noting that a reduction in the percentage of effector memory Th1‐ cell counts, after 3 months of treatment (P < 0.0001) (Figure 3A). No like Th17 cells was detected in NEDA patients from month 3 to changes in double negative (CD27− IgD−) B cells, and plasma cells month 12 of follow‐up, but no differences were found in the ODA were observed on DMF‐treated patients (Tables S4 and S5). group (Figure 4E,F, and Table S3). Interestingly, differences in the The analysis of NEDA and ODA patients showed that B cells from percentage of Th1‐like Th17 effector memory cells between NEDA NEDA and ODA patients were affected in a similar way, although the and ODA patients were statistically significant after 3 and 6 months group of ODA patients did not reach statistical differences in most of treatment (P = 0.0202 and P = 0.0373, respectively). of the time points compared to baseline (probably due to the low On the other hand, the analysis of memory Treg in the whole group number of samples of the ODA group) (Figure 2E,F): (a) decrease of of MS patients showed a progressive reduction of this population, being preswitched and switched class memory B cells (NEDA: P < 0.0001 statistically significant after 3 months of DMF treatment (P = 0.0001; and P = 0.0005 vs ODA: P = 0.2148 and P = 0.0049, respectively); Figure 4G). However, the percentage of activated memory Treg re‐ (b) increase of naïve B cells (NEDA: P < 0.0001 vs ODA: P = 0.0415); mained stable during the whole observation period (Figure 4H). Results and (c) increase of transitional B cells (NEDA: P = 0.0006 vs ODA obtained considering NEDA and ODA patients were similar for both P = 0.0791) (Figure 3B,C and Table S5). memory Treg and activated Treg subpopulations (Tables S2 and S3). MANSILLA et al. | 1001 F I G U R E 4 Selective depletion of the pro‐inflammatory Th1‐like Th17 cells in  NEDA patients after 3 months of DMF treatment. Representation of the pro‐/ antiinflammatory ratio in central memory (CM) and effector memory (EM) CD4+ T cells for total RRMS patients (n = 22) (A, B) as well as for NEDA and ODA patients (n = 15 and n = 6, respectively) (C, D), before the treatment (Baseline) and after 1, 3, 6, and 12 months of follow‐up. For ratio calculations, the % of Th2 cells was considered as antiinflammatory and the sum of the % of Th1, Th17, and Th1‐like Th17 was used as pro‐inflammatory values. The % of Th1‐like Th17 (CXCR3+ CCR6+) T lymphocytes was analyzed on NEDA (E) and ODA (F) patients. In addition, the % of regulatory T cells (Treg, CD127− CD25 +CCR4+ CD45RO+) (G) and activated Treg (HLA‐DR+ Treg) (H) were determined. Data are expressed as mean SD. Each dot represents the value of an individual. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 did not show any alteration of these subpopulations caused by 3.4 | DMF shifts the balance between cytotoxic and DMF treatment. However, after 6 months of follow‐up, the per‐ regulatory NK cells centage of circulating cytotoxic CD56 dim CD16 + NK cells was The evaluation of percentages and absolute cell counts of reduced (P = 0.0032), while the proportion of immunoregula‐ monocytes, natural killer (NK) cells, and dendritic cells (DC) tory CD56 bright NK cells increased (P = 0.0003) (Figure 5A,B). 1002 | MANSILLA et al. F I G U R E 5 DMF shifted the balance from cytotoxic to regulatory NK cells. Representation of percentage (%) of cytotoxic (CD56dim CD16+) and regulatory (CD56bright CD16‐) natural killer (NK) cells (A and B, respectively), before the treatment (Baseline) and after 1, 3, 6, and 12 months of follow‐up in total RRMS patients (n = 22); and regulatory NK cells in NEDA (n = 15) (C) and ODA (n = 6) (D) patients. Data are expressed as mean SD. Each dot represents the value of an individual. *P < 0.05 and **P < 0.01 The analysis in NEDA and ODA patients revealed that the ODA Although etiology of MS still remains unknown, it is considered an group did not reach a statistical increase of regulatory NK cells immune‐mediated disease in which myelin‐reactive CD4+ T lympho‐ (NEDA, P = 0.0008; ODA, P = 0.4152) (Figure 5C,D; and Tables cytes trigger a complex immune response that leads to inflammation, S4 and S5). myelin destruction, and axonal loss.17-19 The main players initiating the autoimmune attack are autoreactive CD4+ Th1 and Th17 cells, which produce pro‐inflammatory cytokines, such as IFN‐γ and IL‐17, 4 | D I S CU S S I O N respectively. These cells are able to cross the blood‐brain barrier and reach the CNS and recognize myelin antigens, secreting pro‐in‐ During the last years, the multiple treatment options available for flammatory mediators and activating other immune cells, such as B MS patients have driven the attention of physicians and researchers cells and cytotoxic CD8+ T cells. 20 Consequently, different pharma‐ on identifying the most suitable DMT depending not only on the cological strategies have been developed to abrogate or block these clinical and immunological characteristics of the patient, but also on immune cascades. the specific mechanism of action of each therapy and its potential In this study, we prospectively evaluated changes induced by risks. To this end, cross‐sectional studies have contributed notably to DMF treatment performing a direct whole blood cytometric anal‐ elucidate the most relevant effects of different DMT such as DMF or ysis. This approach provided a deep analysis of the adaptive and in‐ fingolimod. However, due to the interindividual heterogenicity, some nate immune response, being able to reproduce most of the DMF less obvious changes could remain undetectable. In contrast, longi‐ effects reported during last years. tudinal studies are limited by the final number of patients completing As described, DMF reduced the absolute number of circulating the follow‐up period, but results obtained avoid interindividual vari‐ lymphocytes, affecting more extensively CD8+ cytotoxic T cells8,21 ability and are useful to monitor changes induced by the treatment and therefore increasing the CD4+/CD8+ T‐cell ratio.9,22 Further over time. With this purpose, our group has been collaborating in an analyses evidenced the characteristic phenotypic alterations pro‐ international consortium to establish standardized multiparametric duced by DMF treatment on CD4+ and CD8+ T‐cell subpopulations, cytometric protocols to evaluate changes in leukocyte subpopula‐ a selective reduction of central and effector memory T cells, and tions of patients with autoimmune diseases such as type I diabetes, an expansion of naïve T cells. This effect of DMF is considered as rheumatoid arthritis, or MS15,16 Importantly, these cytometric panels one of the most important effects reducing MS activity on patients, are performed on fresh whole blood samples, providing an easy, fast, since memory cells are responsible to induce a rapid and efficient and direct approach to obtain a readout of circulating leukocytes of response against a previously recognized antigen. While effec‐ patients able to be translated into clinical practice. tor memory T cells rapidly migrate to the foci of inflammation to MANSILLA et al. | 1003 orchestrate a rapid immune response, central memory T cells home an increase of transitional B cells following 12 months of treat‐ to lymph nodes, proliferate following antigen restimulation, and then ment, 22,31,32 although our work is the first reporting this effect differentiate into effector memory T cells. Therefore, the depletion after only 3 months of DMF therapy. of memory cells could be beneficial to reduce disease activity. In this Among monocytes, DC, and NK cells, only NK‐cell subpopula‐ regard, we observed that only NEDA patients were exhibiting a sig‐ tions were affected by DMF treatment. Specifically, this treatment nificant decrease of CD4+ central and effector memory subpopula‐ reduced cytotoxic CD56 dim CD16 + NK cells and increased the per‐ + tions, as well as CD8 effector memory T cells. centage of circulating CD56 bright CD16− NK cells after 6 months Interestingly, the expression analysis of CXCR3 and CCR6, surro‐ of follow‐up. NK cells are an important component of innate im‐ gate markers to define Th1 (CXCR3+ CCR6−), Th17 (CXCR3− CCR6+), munity that can be divided in 2 main subsets, CD56 dim CD16+ and Th1‐like Th17 (CXCR3+ CCR6+), and Th2 (CXCR3− CCR6−) CD4+ lym‐ CD56 bright CD16− NK cells. The most frequent NK‐cell population phocyte subpopulations revealed an expansion of antiinflammatory in peripheral blood are the CD56 dim CD16+ cells, which have per‐ Th2 cells (both central memory and effector memory), together with forin‐mediated cytolytic activity. In contrast, CD56 bright CD16− NK a marked decrease of central memory and effector memory Th1‐ cells are predominant in secondary lymphoid tissues and have high like Th17 cells. These Th1‐like Th17 lymphocytes, also described proliferative activity and cytokine production, both pro‐ and an‐ as Th1/Th17, are Th17 cells with Th1‐like features, producing both tiinflammatory, depending on the context. It is worth noting that IFN‐γ and Th17 pro‐inflammatory cytokines. 23,24 In a recent pub‐ clinical efficacy of daclizumab in MS patients has been correlated lication, Langelaar et al25 described that effector memory Th1‐like with an expansion of this CD56 bright NK cells. 33 Consequently, an‐ Th17 express the α4β1 integrin, very late antigen‐4 (VLA‐4), and other mechanism mediating DMF efficacy could be related to the therefore, their activation promotes their migration into the CNS, reduction of cytotoxic NK cells as well as the expansion of regula‐ inducing neuroinflammation. In our study, we showed for the first tory CD56 bright NK cells induced by the treatment. In this regard, time that DMF exerts an antiinflammatory effect by increasing Th2 our data indicate that only the expansion of CD56 bright CD16− NK cells and decreasing Th1‐like Th17 cells. Surprisingly, a reduction cells was related to the optimal response of DMF, since a statisti‐ in the proportion of circulating memory Treg was found. Although cally significant change was not reached in ODA patients. These not addressed in this work, it would be interesting to analyze the results are in accordance with the study of Medina et al,12 which expression of CXCR3 and CCR6 in the Treg population to know found a negative correlation between CD56 bright NK cells and var‐ whether this reduction is related to a decrease of pro‐inflammatory ious cytokine‐producing T cells, postulating that this CD56 bright 26,27 Th1 and/or Th17 Treg. Regarding discrepancies with other stud‐ NK cells could inhibit pathogenic effector lymphocytes. In this re‐ ies reporting a reduction on Th1 and Th17 cells and no alteration gard, our study has revealed a reduction of Th1‐like Th17 effector or even expansion of Treg,8,9,12 they might be caused by the use memory lymphocytes, a population of IFN‐ϒ and IL‐17 secreting in those studies of isolated, cryopreserved, and stimulated PBMC cells that might be downregulated by the action of CD56bright NK (since it has been reported that these procedures can modify the cells. Moreover, among Th1‐like Th17 cells, Th17.1 lymphocytes expression of chemokine receptors). 28-30 We consider that the direct (CCR6+CXCR3+CCR4−) are a subpopulation of lymphocytes with analysis of whole blood leukocytes subpopulations can reflect better superior capacity to reach the CNS. It was found selectively ac‐ the effects caused by DMF treatment, avoiding variability caused cumulated in peripheral blood of natalizumab‐treated MS pa‐ by cell manipulation. Strikingly, our results showed a reduction of tients who did not experience clinical relapses. 25Further studies effector memory Th1‐like Th17 cells only in NEDA patients, already are needed to investigate whether the effect of DMF increasing after 3 months of treatment, meaning that this population could be CD56 bright NK cells mediates the reduction of Th17.1 cells, promot‐ a candidate biomarker for the early identification of optimal DMF ing, consequently, the reduction of neuroinflammation in NEDA responders. patients. On the other hand, we also observed a selective decrease on In summary, we have shown that whole blood multiparametric circulating memory B cells (both switched and preswitched), in flow cytometry analysis is a useful tool to perform an exhaustive parallel with an increase of naïve B lymphocytes, in accordance immune monitoring of MS patients under DMF treatment. Our with previous publications.10,22,31 B cells can produce pro‐inflam‐ longitudinal study demonstrated that DMF mediates its beneficial matory cytokines, and treatments depleting the B‐cell compart‐ effect by depleting memory T and B cells, as well as cytotoxic NK ment such as rituximab (anti‐CD20 monoclonal antibody) have cells, while increasing the prevalence of transitional B cells and demonstrated beneficial effects on MS patients. However, DMF CD56 bright NK cells, shifting the balance from a pro‐ to an antiin‐ not only mediated a depletion of mature or differentiated B cells, flammatory profile. but also an increase of transitional B cells, a population of cells en‐ Furthermore, we have identified effector memory Th1‐like Th17 riched in regulatory IL‐10 producing B cells. It was not addressed lymphocytes as a potential biomarker for the early identification of whether this increase was due to an increase of transitional B cells DMF responder patients. However, further studies with larger num‐ or to a suppression of their differentiation to naïve mature B cells. bers and independent cohorts of MS patients are necessary to con‐ This finding is in accordance with previous studies also showing firm these results. 1004 | MANSILLA et al. AC K N OW L E D G E M E N T S the balance between Th1/Th17 and Th2 in multiple sclerosis pa‐ tients. J Immunol. 2017;198(8):3069‐3080. We thank Mr. Marco A. Fernández of the Cytometry Facility of the 10. Li R, Rezk A, Ghadiri M, et al. Dimethyl fumarate treatment medi‐ Germans Trias i Pujol Research Institute for his continuous help and ates an anti‐inflammatory shift in B cell subsets of patients with suggestions, and Mr. Armando Ruiz Bazorla for his technical sup‐ multiple sclerosis. J Immunol. 2017;198(2):691‐698. 11. Longbrake EE, Cantoni C, Chahin S, Cignarella F, Cross AH, Piccio L. port. This study was partially supported by project PI14/01175, Dimethyl fumarate induces changes in B‐ and T‐lymphocyte func‐ integrated in the Plan Nacional de I+D+I, and cosupported by the tion independent of the effects on absolute lymphocyte count. ISCIII‐Subdirección General de Evaluación and the Fondo Europeo Mult Scler. 2018;24(6):728‐738. de Desarrollo Regional (FEDER). JNB was beneficiary of a FI‐DGR 12. Medina S, Villarrubia N, Sainz de la Maza S, et al. Optimal re‐ sponse to dimethyl fumarate associates in MS with a shift from PhD contract from AGAUR (Agència de Gestió d'Ajuts Universitaris an inflammatory to a tolerogenic blood cell profile. Mult Scler. i de Recerca) and supported by the Government of Catalonia 2018;24(10):1317‐1327. (02/2015‐01/2018). The authors are members of a consolidated 13. Teniente‐Serra A, Hervás JV, Quirant‐Sánchez B, et al. Baseline group recognized by AGAUR (2017 SGR 103) and supported by the differences in minor lymphocyte subpopulations may predict re‐ sponse to fingolimod in relapsing‐remitting multiple sclerosis pa‐ Government of Catalonia. This work has been supported by positive tients. CNS Neurosci Ther. 2016;22(7):584‐592. discussion through both ENTIRE and A FACTT Europeans networks 14. Quirant‐Sanchez B, Hervas‐Garcia JV, Teniente‐Serra A, et al. (COST Actions BM0907 and BM1305, respectively). COST is sup‐ Predicting therapeutic response to fingolimod treatment in multi‐ ported by the EU Framework Programme Horizon 2020. ple sclerosis patients. CNS Neurosci Ther. 2018;24(12):1175‐1184. 15. Popadic D, Anegon I, Baeten D, et al. Predictive immunomonitoring – the cost entire initiative. Clin Immunol. 2013;147(1):23‐26. C O N FL I C T O F I N T E R E S T 16. Maecker HT, McCoy JP, Nussenblatt R. Standardizing immuno‐ phenotyping for the human immunology project. Nat Rev Immunol. The authors declare no conflict of interest. 2012;12(3):191‐200. 17. Noseworthy JH, Lucchinetti C, Rodriguez M, Weinshenker BG. Multiple sclerosis. N Engl J Med. 2000;343(13):938‐952. ORCID 18. Sospedra M, Martin R. Immunology of multiple sclerosis. Annu Rev Immunol. 2005;23:683‐747. María José Mansilla https://orcid.org/0000-0002-5044-6313 19. Comabella M, Khoury SJ. Immunopathogenesis of multiple sclero‐ sis. Clin Immunol. 2012;142(1):2‐8. Bibiana Quirant‐Sánchez https://orcid. 20. Hemmer B, Nessler S, Zhou D, Kieseier B, Hartung HP. org/0000-0002-3875-8074 Immunopathogenesis and immunotherapy of multiple sclerosis. Nat Clin Pract Neurol. 2006;2(4):201‐211. 21. Spencer CM, Crabtree‐Hartman EC, Lehmann‐Horn K, Cree BA, REFERENCES Zamvil SS. Reduction of cd8(+) T lymphocytes in multiple sclero‐ sis patients treated with dimethyl fumarate. Neurol Neuroimmunol 1. Fox RJ, Miller DH, Phillips JT, et al. Placebo‐controlled phase 3 Neuroinflamm. 2015;2(3):e76. study of oral BG‐12 or glatiramer in multiple sclerosis. N Engl J Med 22. Montes Diaz G, Fraussen J, Van Wijmeersch B, Hupperts R, Somers 367(12), 1087‐1097. V. Dimethyl fumarate induces a persistent change in the composi‐ 2. Gold R, Kappos L, Arnold DL, et al. Placebo‐controlled phase 3 tion of the innate and adaptive immune system in multiple sclerosis study of oral BG‐12 for relapsing multiple sclerosis. N Engl J Med patients. Sci Rep. 2018;8(1):8194. 367(12), 1098–1107. 23. Acosta‐Rodriguez EV, Rivino L, Geginat J, et al. Surface pheno‐ 3. Gold R, Linker RA, Stangel M. Fumaric acid and its esters: an emerg‐ type and antigenic specificity of human interleukin 17‐producing T ing treatment for multiple sclerosis with antioxidative mechanism helper memory cells. Nat Immunol. 2007;8(6):639‐646. of action. Clin Immunol. 2012;142(1):44‐48. 24. Cohen CJ, Crome SQ, MacDonald KG, Dai EL, Mager DL, Levings 4. Linker RA, Lee DH, Ryan S, et al. Fumaric acid esters exert neuro‐ MK. Human Th1 and Th17 cells exhibit epigenetic stability at protective effects in neuroinflammation via activation of the Nrf2 signature cytokine and transcription factor loci. J Immunol. antioxidant pathway. Brain. 2011;134(Pt 3):678‐692. 2011;187(11):5615‐5626. 5. Scannevin RH, Chollate S, Jung M‐Y, et al. Fumarates promote 25. van Langelaar J, van der Vuurst de Vries RM, Janssen M, et al. 17.1 cytoprotection of central nervous system cells against oxidative cells associate with multiple sclerosis disease activity: Perspectives stress via the nuclear factor (erythroid‐derived 2)‐like 2 pathway. J for early intervention. Brain. 2018;141(5):1334‐1349. Pharmacol Exp Ther. 2012;341(1):274‐284. 26. Duhen T, Duhen R, Lanzavecchia A, Sallusto F, Campbell DJ. 6. Gillard GO, Collette B, Anderson J, et al. DMF, but not other fuma‐ Functionally distinct subsets of human FOXP3+ Treg cells that phe‐ rates, inhibits NF‐kappaB activity in vitro in an Nrf2‐independent notypically mirror effector Th cells. Blood. 2012;119(19):4430‐4440. manner. J Neuroimmunol. 2015;283:74‐85. 27. Halim L, Romano M, McGregor R, et al. An atlas of human regula‐ 7. Longbrake EE, Ramsbottom MJ, Cantoni C, Ghezzi L, Cross AH, tory t helper‐like cells reveals features of Th2‐like Tregs that sup‐ Piccio L. Dimethyl fumarate selectively reduces memory T cells in port a tumorigenic environment. Cell Rep. 2017;20(3):757‐770. multiple sclerosis patients. Mult Scler. 2016;22(8):1061‐1070. 28. Cossarizza A, Chang H‐D, Radbruch A, et al. Guidelines for the use 8. Gross CC, Schulte‐Mecklenbeck A, Klinsing S, et al. Dimethyl of flow cytometry and cell sorting in immunological studies. Eur J fumarate treatment alters circulating T helper cell subsets Immunol. 2017;47(10):1584‐1797. in multiple sclerosis. Neurol Neuroimmunol Neuroinflamm. 29. Faint JM, Tuncer C, Garg A, Adams DH, Lalor PF. Functional conse‐ 2016;3(1):e183. quences of human lymphocyte cryopreservation: implications for 9. Wu QI, Wang Q, Mao G, Dowling CA, Lundy SK, Mao‐Draayer Y. subsequent interactions of cells with endothelium. J Immunother. Dimethyl fumarate selectively reduces memory T cells and shifts 2011;34(8):588‐596. MANSILLA et al. | 1005 30. Nieto JC, Canto E, Zamora C, et al. Selective loss of chemokine S U P P O R T I N G I N FO R M AT I O N receptor expression on leukocytes after cell isolation. PLoS ONE. 2012;7(3):e31297. Additional supporting information may be found online in the 31. Holm Hansen R. Hojsgaard Chow H, Sellebjerg F, Rode von Essen Supporting Information section at the end of the article. M. Dimethyl fumarate therapy suppresses B cell responses and fol‐ licular helper T cells in relapsing‐remitting multiple sclerosis. Mult Scler. 2018. [Epub ahead of print]. https​://doi.org/10.1177/13524​ How to cite this article: Mansilla MJ, Navarro‐Barriuso J, 58518​790417 32. Lundy SK, Wu Q, Wang Q, et al. Dimethyl fumarate treatment of re‐ Presas‐Rodríguez S, et al. Optimal response to dimethyl lapsing‐remitting multiple sclerosis influences b‐cell subsets. Neurol fumarate is mediated by a reduction of Th1‐like Th17 cells Neuroimmunol Neuroinflamm. 2016;3(2):e211. after 3 months of treatment. CNS Neurosci Ther. 2019;25: 33. Bielekova B, Catalfamo M, Reichert‐Scrivner S, et al. Regulatory 995–1005. https​://doi.org/10.1111/cns.13142​ CD56(bright) natural killer cells mediate immunomodulatory ef‐ fects of IL‐2Ralpha‐targeted therapy (daclizumab) in multiple scle‐ rosis. Proc Natl Acad Sci USA. 2006;103(15):5941‐5946.