Fingolimod reduces circulating tight-junction protein levels and in vitro peripheral blood mononuclear cells migration in multiple sclerosis patients

There are no data on the effects of fingolimod, an immunomodulatory drug used in treatment of multiple sclerosis (MS), on circulating tight-junction (TJ) protein levels as well as on peripheral blood mononuclear cells (PBMC) migration. Serum TJ protein [occludin (OCLN), claudin-5 (CLN-5) and zonula occludens-1 (ZO-1)] levels, sphingosine-1 phosphate 1 (S1P1) receptor expression on circulating leukocyte populations as well as in vitro PBMC migration were longitudinally assessed in 20 MS patients under 12-months fingolimod treatment and correlated with clinical and magnetic resonance imaging (MRI) parameters. After 12 months of treatment, a significant reduction of mean relapse rate as well as number of active lesions at MRI was found. TJ protein levels significantly decreased and were associated with reduction of S1P1 expression as well as of PBMC in vitro migratory activity. A significant correlation of CLN-5/OCLN ratio with new T2 MRI lesions and a significant inverse correlation of CLN-5/ZO-1 ratio with disability scores were found. These findings support possible in vivo effects of fingolimod on the blood-brain barrier (BBB) functional activity as well as on peripheral cell trafficking that could result in avoiding passage of circulating autoreactive cells into brain parenchyma. Circulating TJ protein levels and respective ratios could be further studied as a novel candidate biomarker of BBB functional status to be monitored in course of fingolimod as well as of other immunomodulatory treatments in MS.

Serum TJ proteins levels. All three proteins levels were found to be significantly reduced at 6 months as well as after 12-months treatment. This reduction was substantially similar for all three proteins. In particular, serum OCLN decreased by 47.4% and 70,5% after 6 and 12 months of therapy, respectively (p < 0.0001) (Fig. 1A). Serum CLN-5 significantly decreased by 54.8% and 73.6%, respectively (p < 0.0001) (Fig. 1B) and ZO-1 by 53.5% and 74.5%, respectively (p < 0.0001) (Fig. 1C). However, for all three proteins, the levels at 12 months treatment were significantly higher than those of NHS. There was no significant correlation between single TJ protein levels and clinical parameters such as relapse rate and disability score assessed by EDSS at baseline as well as at 12 months. No significant correlation between single TJ protein levels and neuroimaging parameters at MRI such as new T2 weighted lesions and T1 GAD + lesions was observed. (Supplementary Table S1). However, when calculating CLN-5/OCLN and CLN-5/ZO-1 ratios, we found a significant correlation of CLN-5/OCLN ratio at baseline with the number of new T2 weighted MRI lesions both at baseline (r = 0.56; p = 0.0095) ( Fig. 2A) and at 12-months treatment (r = 0.50; p = 0.023) (Fig. 2B). Conversely, there was a high statistical trend towards a negative correlation (r = −0.40; p = 0.056) and a significant inverse correlation (r = −0.46; p = 0.037), of CLN-5/ ZO-1 ratio at baseline with EDSS at baseline and at 12-months treatment, respectively (Fig. 2C,D).
Expression of S1P 1 in PBMC. Representative flow cytometry analysis of S1P 1 receptor in leukocyte populations from MS and NHS is shown in Fig. 3. In MS patients, a significantly reduced expression of S1P 1 by 30% and 61% was observed in CD4 + cells at 6 and 12 months, respectively (p = 0.0001) (Fig. 4A). A similar extent of reduction by 24% and 52% was observed in CD8 + cells (p = 0.0001) (Fig. 4B). In both subpopulations, S1P 1 expression extent after 12 months of therapy, reached that of NHS. In CD19 + cells, the reduction was 11% and 28%, respectively (p = 0.019) (Fig. 4C) whilst in CD14 + cells, the S1P1 expression was unchanged at 6 months but significantly decreased by 29% at 12 months (p = 0.0001) (Fig. 4D). However, in both CD19 + and CD14 + cells, treatment reduced the receptor expression to values similar to those of NHS.
Chemotaxis assay. A very significant reduction of PBMC migration rate by 39.5% and 68% was observed at 6 and 12 months, respectively (p < 0.0001) (Fig. 5A). To test whether the effect on migration was specifically due to FTY720 therapy, we assayed PBMCs from NHS (n = 12) or MS patients (n = 20) at baseline after a 24-hour in vitro exposure to FTY720 and observed a migration rate similar to that of MS patients at 12 months therapy (Fig. 5B). There was no correlation between the extent of S1P 1 expression in CD4 + (r = 0.03; p = 0.89) and CD8 + cells (r = −0.16; p = 0.48) and PBMC migration rate indicating no direct role of lymphocyte S1P 1 expression on the peripheral cell migration.

Discussion
In this study, we demonstrated that treatment with fingolimod is associated with significant reduction of circulating TJ proteins levels contributing to the architecture of BBB. BBB changes constitute one of the pathological hallmarks in MS 8 and are associated with TJ proteins alterations. In MS brains, changes in ZO-1 and OCLN were found in both active lesions and normal-appearing white matter (NAWM) 9 and ZO-1 was also found altered in inactive chronic lesions 10 . These findings suggest that damage to TJ proteins concurs to cause subtle alterations of the BBB architecture even in areas where the inflammatory process should be exhausted as well as in areas far from active or chronic lesions such as NAWM supporting the concept that BBB function is profoundly compromised in MS and with difficulty normalizes 11 . Moreover, there is experimental evidence that serum from active MS patients reduces the expression of OCLN and VE-cadherin in endothelial cultures suggesting that circulating proinflammatory factors such as cytokines, metalloproteinases as well as reactive oxygen radicals contribute to TJ protein changes and subsequent endothelium alterations 12 . These pathological findings parallel the MS real clinical life where often GAD + lesions at MRI, thought a surrogate marker of BBB alteration 13 , were also detected during clinical stability stages independently of occurrence of new clinical relapses 14,15 . Despite the lack of any correlation between single TJ protein levels and number of GAD + lesions, we, however, detected the highest TJ protein levels at baseline, in patients showing at least one GAD+ lesion or experiencing a clinical relapse at a time close to blood collection and thus being in active disease, consisting with the association of TJ proteins alterations with BBB leakage. However, considering that the average duration of contrast enhancement at a MRI lesion was found to be around 3 weeks 16 and the different time of blood collection respect to MRI in our cohort, we can't rule out that TJ protein levels could be lower than those detected and thus may be underpowered to allow to reach significant values at the correlation analysis. Of interest is the significant correlation of CLN-5/OCLN ratio with new T2 weighted lesions at MRI suggesting a common pathway underlying the release of both TJ proteins into the peripheral blood and related to brain inflammatory lesion formation as indicated by detection of new T2 lesions. It has been demonstrated that, at the cerebral microvessels, CLN-5 and OCLN are assembled into oligomers contributing to selective inhibition of the paracellular transport as well as to endothelial cell polarity fundamental for maintaining BBB integrity 17 . Moreover, the significant inverse correlation between CLN-5 /ZO-1 ratio and EDSS suggests that increased release of ZO-1 reduces this ratio and is associated with the complex process underlying the disability accrual. This hypothesis is supported by ZO-1 location at the brain endothelial cell cytoplasm binding both to the occludin-CLN-5 complex and to cytoskeleton proteins 18 . It is likely that the cascade of events causing TJ protein release leads to disassembly of cytoskeleton proteins such as actin thus resulting in BBB breakdown associated with diffuse inflammation and nervous tissue damage and causing MS disability 11,19 . In our study, the lowest TJ protein levels were detected after 12-months treatment showing a significant decrease of GAD + lesions and thus supporting an in vivo effect of fingolimod on BBB functional activity and likely on repair of BBB damage due to neuroinflammation. However, these levels, although reduced, tended to be higher than NHS supporting the persistence of subtle BBB alterations in clinically stable MS patients, even though we can't rule out that an extension of longitudinal measurement over 12-months treatment could show further level reduction.
Of interest are the findings on the S1P 1 expression in different leukocyte populations never, to the best of our knowledge, investigated in MS patients. It is demonstrated that reduction of S1P 1 receptor expression by fingolimod is due to its internalization 20 and subsequent degradation 21 . In our study, the extent of expression reduction differed between lymphocyte subsets and B cells as well as monocytes, respectively, likely reflecting possible differences in the effects of fingolimod on circulating immunocompetent cells rather than fluctuations of single cell populations counts 22 . Indeed, several distinct modulatory effects on B cell subsets and related cytokine phenotype 23 as well as on monocytes 24 were recently found in MS patients under fingolimod treatment. Unique are the findings on PBMC migration suggesting that circulating PBMC not entrapped in lymph nodes by fingolimod, are subjected to changes in the complex pathway of adhesion molecules and cell surface receptors involved in cell adhesion to and passage across BBB endothelium. Experimental evidence of an effect of fingolimod on these molecules has been achieved in EAE where reduction and downregulation of a number of endothelial adhesion molecule transcripts has been shown 5 . In addition, a significant reduction in CXCL13 levels, a chemokine involved in leukocyte cell trafficking, has recently been found in cerebrospinal fluid from MS patients under treatment with fingolimod supporting a modulatory role of this drug in the cell migration 25 . Recently, any insights into the possible underlying mechanisms of fingolimod effects on BBB cell migration have been provided in an animal experimental model by Zhao et al. demonstrating that FTY720 blocks in vivo leukocyte recruitment into the brain parenchyma by inhibition of brain endothelium activation through reduction of phosphorylation of a number of signaling molecules such as serine/threonine-specific protein kinases, STAT6 and NFkB 26 . In our study, the demonstration of a direct effect of fingolimod on cell migration was provided by the finding that PBMC from NHS as well as MS patients prior to starting treatment, display a significant reduction of migratory activity up to the extent detected in MS patients at 12-months treatment, after in vitro exposure to this molecule. Although we were not able to find any correlation between S1P 1 expression and extent of migration rate, the overlapping of migration index of fingolimod-treated cells with that of untreated cells, suggests however a link of this receptor subtype with migration pathway that remains to be elucidated. A possible limitation of these findings comes from the characteristics of the chemotaxis assay lacking in an endothelial cell monolayer more representative of the BBB architecture in vivo and thus not allowing to achieve direct data on BBB transmigration.
In conclusion, our study provides novel data on possible in vivo effects of fingolimod on BBB functional activity and repair during neuroinflammation and extends its effect on peripheral blood cell trafficking speculating on a possible reduction of migratory activity of circulating immunocompetent cells across BBB and likely avoiding passage of circulating autoreactive cells into brain parenchyma. Further studies are however needed to confirm this hypothesis. All biomarkers of BBB alteration in MS, investigated to date, resulted poorly sensitive when correlated with the number of GAD + lesions 27 . It is therefore reasonable in searching for new BBB investigating tools, to take into account other quantitative MRI parameters than number of GAD + lesions such as volume of gadolinium enhancement even though different patterns of in vivo BBB alteration should be considered in MS 28 . However, circulating TJ proteins levels and respective ratios could be further studied as a novel candidate biomarker of BBB functional status to be monitored in course of fingolimod as well as of other immunomodulatory treatments in MS patients.

Materials and Methods
Patients. This is an observational prospective study. Twenty RRMS patients (6 male and 14 female subjects) 29 were recruited. Twenty-five sex-and age-matched NHS, blood donors at the Blood Transfusion Centre of the Siena University Medical School and giving informed consent, were included as control group. All patients signed informed consent and the study was conducted according to the Declaration of Helsinki and approved by the Ethics Committee of the Medical School of University of Siena. All patients started second-line therapy with FTY720 (0.5 mg/die) according to the criteria approved by Agenzia Italiana del Farmaco (AIFA) including the failure of a first-line therapy or occurrence of highly active MS as demonstrated by at least two clinical relapses in the last year with at least one GAD + lesions or increase in number of T2 weighted lesions at MRI 30 . Number of relapses in the last year prior to enrollment in the study as well as sustained disability score by EDSS 31 at the enrollment time and after 12 months of treatment were assessed. Relapses were defined as occurrence of new clinical neurological signs or symptoms not associated with fever. Brain MRI at 1.5 Tesla with and without gadolinium infusion was performed according to the standard procedures for clinical follow-up at baseline time and after 12 months of treatment and number of new T2 weighted and T1 GAD + lesions was estimated. In addition, NEDA at 12 months defined as a composite consisting of absence of relapses, no sustained EDSS score progression, and no new or enlarging T2 or T1 GAD + lesions at MRI was assessed. In vitro migratory activity of PBMC from multiple sclerosis (MS) patients prior to start fingolimod treatment and from normal healthy subjects (NHS) prior to and after in vitro exposure to fingolimod (B). MI = migration index; T 0 = baseline time prior to treatment; T 6 = at 6 months of treatment; T 12 = at 12 months of treatment; S1P = sphingosine 1-phosphate; MS T 0 FTY720 + = PBMC from MS patients at baseline time exposed to fingolimod; NHS T 0 FTY720 + = PBMC from NHS after in vitro exposure to fingolimod. S1P+ FTY720 = assays performed with S1P at bottom of the chemotaxis chamber in the presence of FTY720. The results are shown as scatter plots with mean ± SD. Blood collection and PBMC separation. PBMC were isolated by standard density gradient centrifugation procedures on Ficoll-Hypaque from buffy coat preparations of NHS obtained on the same day by Transfusion Centre, as well as from MS patients. In MS subjects, blood sample was drawn at baseline time after a wash out of the first-line therapy ranging from three to four weeks and at 6 and 12 months of FTY720 therapy. In NHS, serum sample was obtained by Transfusion Centre on the same day as blood donation.
Chemotaxis assay. To test the effects of FTY720 binding to S1P 1 on migration of circulating immunocompetent cells, a chemotaxis assay was performed. PBMC were labelled with sulfo-NHS-biotin (Pierce, Rockford, IL) (0.2 mg/ml) for 30 minutes at room temperature according to a previously established method 33 . Biotin-labelled cells (25 × 10 3 ) in original culture medium (DMEM-10% FCS) were seeded in the upper compartment of a commercial chemotaxis chamber with 3 μm pore size membranes (NeuroProbe, MD, USA). As chemotactic agent, 100 nM sphingosine 1-phosphate (S1P) (Sigma) alone or, in some assays, in the presence of 100 nM FTY720 (Sigma), was placed in lower chamber. For each sample, as negative control, S1P was omitted in lower chamber. After 3 h at 37 °C, streptavidin-POD conjugate (1:5000) for 1 h at room temperature was added in lower chamber and after washes with PBS pH 7,4 and addition of the substrate 0,1% o-phenylendiamine (OPD) in citrate buffer pH 4,5, the product reaction was read at ELISA photometer at 490 nm and the absorbance of biotin-labelled cells after passage in the bottom compartment was measured. In some assays, PBMC (25 × 10 3 ) from NHS or from MS patients at baseline were exposed with 100 nM FTY720 for 24 h and then placed in chemotaxis chamber for assay as above described. Chemotactic activity was assessed as migration index calculated according to the following formula 34 : − . Absorbance of biotin labelled cells/absorbance of negative control Statistical analysis. Multiple comparisons were performed with ANOVA and Tukey post test for parametric data. Comparisons between paired groups was performed with Wilcoxon matched-pairs signed rank test for non-parametric data. Normality of data was estimated with D' Agostino & Pearson normality test. Correlation analysis was performed with Spearman's test and Pearson r coefficient for non-parametric and parametric data, respectively. P < 0.05 values were considered significant. All statistical tests were performed with GraphPad Prism (version 7.02), San Diego, CA, USA.