Dynamic changes of MMP-9 plasma levels correlate with JCV reactivation and immune activation in natalizumab-treated multiple sclerosis patients

The aim of the study was to investigate the changes of matrix metalloproteinase (MMP)-2 and MMP-9 plasma levels during natalizumab treatment and their correlation with JC virus (JCV) reactivation and T-lymphocyte phenotypic modifications in peripheral blood samples from 34 relapsing-remitting multiple sclerosis (RRMS) patients. MMP-9 levels were assessed by zymography in plasma samples. JCV-DNA was detected through quantitative real time PCR in plasma samples. T-lymphocyte phenotype was assessed with flow cytometry. MMP-9 plasma levels resulted increased from 12 to 24 natalizumab infusions. Stratifying plasma samples according to JCV-DNA detection, MMP-9 plasma levels were significantly increased in JCV-DNA positive than JCV-DNA negative samples. MMP-9 plasma levels resulted positively correlated with JCV viral load. CD4 immune senescence, CD8 immune activation and CD8 effector percentages were positively correlated to MMP-9 plasma levels, whereas a negative correlation between CD8 naïve percentages and MMP-9 plasma levels was found. Our data indicate an increase of MMP-9 plasma levels between 12 and 24 natalizumab infusions and a correlation with JCV-DNA detection in plasma, T-lymphocyte immune activation and senescence. These findings could contribute to understand PML pathogenesis under natalizumab treatment, suggesting a potential role of MMP-9 as a predictive marker of PML in RRMS patients.


Results
Detection of MMP plasma levels by zymography. MMP-2 and MMP-9 plasma levels were measured by zymography on gelatin-copolymerized gels in 116 plasma samples from 34 RRMS patients stratified according to natalizumab infusion number.
A representative gel from an individual patient is shown in Fig. 1a. No variations were observed for plasma levels of MMP-2, which is constitutively expressed in body fluids and was used as an internal control of sample processing. By contrast, MMP-9 levels increased in plasma accordingly with natalizumab infusion number. The longitudinal analysis, performed in 26 RRMS patients for whom two samples of plasma were available, collected within 12 (T ≤ 12) and at 24 natalizumab infusions (T24), showed a statistically significant increase in MMP-9 plasma levels from T ≤ 12 to T24 (Wilcoxon, p = 0.036) (Fig. 1b).
A bimodal trend in MMP-9 plasma levels was observed from the analysis of the 116 plasma samples collected from all the RRMS patients stratified according to natalizumab infusion number. Specifically, no differences were found in patients after 3, 6 and 12 natalizumab infusions. Conversely, MMP-9 plasma levels increased after 15, 18 and 24 natalizumab infusions (Spearman, ρ = 0.264 and p = 0.024) ( Supplementary Fig. S1a). Moreover, the same bimodal trend was observed for MMP-9/MMP-2 ratio (Table 1).
MMP-9 plasma levels were higher in untreated RRMS patients than HD ( Supplementary Fig. S1b), even though the difference was not statistically significant (Mann-Whitney, p > 0.05). No differences were found between male and females or considering previous therapies.

Evaluation of MMP
In order to avoid sampling biases due to inclusion of multiple samples from the same patient at different timepoints in the same analysis, samples of our cohort were stratified into three groups: T0, before the first

Evaluation of T-lymphocyte subsets, immune activation and senescence. T-lymphocyte immune
activation was evaluated considering CD38 and HLA-DR co-expression, whereas T-lymphocyte immune senescence was evaluated considering the percentages of CD28 − CD57 + cells by flow cytometry. CD4 and CD8 subsets were characterized by flow cytometry considering surface CD45RO and CD27 expression. Briefly, four CD4 and five CD8 subsets were identified as follows: naïve (N): CD27 + CD45RO − , central memory (CM): The evaluation of the different CD4 and CD8 subsets confirmed previous results 25 , showing that CD4 immune activation levels were unchanged during natalizumab treatment, whereas CD8 immune activation were increased in RRMS patients with a longer exposition to natalizumab (Table 1). Moreover, a longitudinal analysis was performed, considering those patients with two whole blood samples available (T ≤ 12 and T24 natalizumab infusions, respectively). An increment in CD8 + HLA-DR + CD38 + percentages was found from T ≤ 12 to T24 (Wilcoxon, p = 0.004) ( Supplementary Fig. S1c). Under natalizumab treatment, no differences in CD4 and CD8 immune senescence levels were observed. CD4 and CD8 subsets were unchanged comparing the groups at different natalizumab infusion numbers.

Discussion
The role of MMPs in MS pathogenesis and progression has been widely investigated. Specifically, it has been shown that the gelatinase subfamily, including MMP-2 and MMP-9, can facilitate the influx of inflammatory cells into the CNS, and contribute to the breakdown of the BBB 26 . Furthermore, gelatinases are able to cleave human myelin basic protein in vitro, thus playing a role in the immune pathogenesis of MS, leading to demyelination 27,28 . There is a large agreement on the pro-inflammatory role of MMP-9 in MS, considering that serum levels were found increased in patients with the RR form of MS compared to the progressive form of the disease 14 . Furthermore, MMP-9 serum levels were found increased in patients with MRI evidence of active lesions 13 and was predictive for the appearance of gadolinium positive MRI lesions in MS subjects with secondary progressive forms 29 . It has been demonstrated that MMP-9 protein concentrations or enzymatic activity were increased in the CSF of patients with MS, especially during the evidence of either clinical or MRI active disease 30 . Moreover, some authors demonstrated that higher MMP-9 plasma levels were associated to the severity of the disease 31 .
As reported by other authors 10,11 , the current study showed higher MMP-9 plasma levels in untreated RRMS patients (T0) compared to HD, strengthening the pathogenic role of MMP-9 in MS.
In this paper, we considered MMP-9 and MMP-2 plasma levels in RRMS patients treated with natalizumab, at different number of infusions, showing that MMP-2 plasma levels remained unchanged in all RRMS patients, independently from the exposition to natalizumab treatment. Conversely, despite an initial stability in MMP-9 plasma levels up to 12 natalizumab infusions, a linear increase was observed from 12 to 24 infusions. The longitudinal analysis further confirmed the increase in MMP-9 plasma levels at 24 natalizumab infusions in comparison to samples collected from the same patients within the first 12 month of treatment.
In our cohort, the increased MMP-9 plasma levels were not associated to clinical or MRI relapses. Recently, other authors have investigated serum levels of MMP-2 and MMP-9 active forms together with serum concentrations of tissue inhibitors of metalloproteinases (TIMP)-1 and TIMP-2 in a cohort of natalizumab treated RRMS patients. Although they did not find any increase in MMP-9 and MMP-2 active forms during natalizumab treatment and demonstrated that MMP-9/TIMP-1 and MMP-2/TIMP-2 ratios remained unchanged, an imbalance in MMP-9/MMP-2 ratio was found for RRMS patients with 15, 18 and 21 natalizumab infusions compared to RRMS patients with shorter exposition to the drug 32 . Interestingly, these data seem to be aligned to the increased MMP-9 levels observed in our cohort after the first year of natalizumab treatment.
In the present work, in order to study the influence of JCV reactivation on MMP-9 plasma levels, we stratified our samples in JCV+ and JCV−, according to JCV-DNA detection. A preliminary analysis, in which we included all the collected samples, evidenced a significant increase in MMP-9 plasma levels in JCV+ compared to JCV− samples, suggesting that JCV circulation in peripheral blood could be implicated in the increase of MMP-9 levels. This hypothesis is reinforced by the positive correlation found in our experiments between MMP-9 plasma levels and JCV viral load and by several studies that showed as MMPs are up-regulated by different infectious agents, including viruses 33 . Notably, a more detailed analysis after stratification of plasma samples into three groups (T0, T12 and T24, according to natalizumab infusion number), in which each patient contributed only with one sample, further confirmed the increase of MMP-9 plasma levels in JCV+ samples in T24 group. Accordingly, we can speculate that the elevated MMP-9 plasma levels observed in RRMS patients beyond 12 months of natalizumab treatment and in patients with detectable JCV-DNA in plasma, contribute to the increased risk of developing PML. It has been demonstrated that natalizumab can mobilize CD34 + precursors latently infected by JCV from the bone marrow to peripheral blood 34 . Therefore, the increased MMP-9 plasma levels can disrupt the integrity of the BBB and facilitate JCV entry into the CNS. Moreover, the impairment of the immune surveillance mechanisms in the CNS, secondary to natalizumab treatment 25 , represents another risk factor implicated in the pathogenesis of PML. In this respect, Fissolo et al. have recently reported decreased MMP-9 mRNA levels in PBMC and protein plasma concentrations at baseline in RRMS patients who developed PML, compared to patients who did not experience PML over 5 years of natalizumab treatment. In their study the differences of MMP-9 levels were inconstantly detected after 12 and 24 months of natalizumab treatment in pre-PML and not-PML patients 23 .
In that paper, they speculate that reduced levels of molecules involved in BBB disruption, such as MMP-9, could interfere with CD8 + T-lymphocyte immune surveillance mechanisms in the CNS, predisposing MS patients to JCV reactivation and PML development 23 . Our results cannot be directly compared with those obtained by Fissolo et al., considering that in our cohort we did not observe any case of natalizumab induced PML. Moreover, Fissolo et al. did not evaluate plasma JCV-DNA levels in RRMS patients of their cohort.
In previous papers, we evaluated T-lymphocyte immune activation, immune senescence and maturation subsets and demonstrated that CD8 + T-lymphocytes immune activation percentages were increased during natalizumab treatment. Furthermore, CD4 and CD8 naïve percentages were unchanged, while CD4 and CD8 effectors increased during natalizumab treatment 25,35 . Although in this study, the analyses of all 116 samples evidenced a positive correlation between JCV viral load and CD8 immuno activation percentages, surprisingly no differences were found in the CD8 immune activation percentages between JCV+ and JCV− samples. By contrast, the longitudinal analysis in samples from 26 RRMS patients, stratified according to natalizumab infusion number, showed a significant increase of CD8 immuno activation percentages in JCV+ in comparison to JCV− samples in the T12 group. The lack of increase in CD8 immune activation percentages in JCV+ samples of T0 and T24 groups might be explain considering the low number of samples analysed and the high variability of CD8 immune activation percentages in both JCV+ and JCV− stratified samples.
To the best of our knowledge, this is the first study that analyzes the relationship between MMP-9 plasma levels and T-lymphocyte subsets in a cohort of RRMS patients under natalizumab treatment. Here we studied the correlation between MMP-9 plasma levels and T-lymphocyte phenotype in RRMS patients under natalizumab treatment and found that MMP-9 plasma levels resulted positively correlated with CD4 immune senescence, CD8 immune activation and CD8 E percentages. By contrast, a negative correlation between MMP-9 plasma levels and CD8 N percentages was found. The correlation between MMP-9 plasma levels and CD4 immune senescence underlines that these cells are not only involved in the MS pathogenesis, but may contribute to pro-inflammatory mechanisms due to their cytotoxic function and resistance to apoptosis, as previously shown by other authors 4 .
All together, these data show that T-lymphocyte immune activation is directly correlated to MMP-9 plasma levels. We could not establish whether the increased MMP-9 level is a cause or a consequence of T-lymphocyte immune activation, considering that MMP-9 is produced by and modulate the activation state of T-lymphocytes. Interestingly, in a mouse model MMP-9 seemed to be essential for the induction of T-lymphocyte proliferation and cytokine production. Specifically, the inhibition of MMP-9 and MMP-2 activity was able to abrogate anti-CD3-induced T-cell proliferation and in MMP-9 −/− knock-out mice, T-lymphocytes showed impairment in cytokine transcription and protein expression 36 . In agreement with these evidences, it has been shown that CD4 + and CD8 + T-lymphocytes, among other cells, have the ability to produce MMP-2 and MMP-9 upon stimulation 37 .
In conclusion, we postulate that in RRMS patients, the combined effect of natalizumab treatment and JCV reactivation could enhance MMP-9 enzymatic activity, thus contributing to CD4 and CD8 immune activation and BBB impairment. Taken together all these findings could contribute to a better understanding of PML pathogenesis in RRMS patients under natalizumab treatment, suggesting the critical role of MMP-9. More studies involving larger cohorts of patients and evaluating MMP inhibitors together with MMPs levels are advisable to explore the potential use of MMP-9 as a predictive tool for PML onset during natalizumab treatment.

Materials and Methods
Ethic statement. This study was approved by the Ethics Committee of Policlinico Umberto I of Rome All patients were under a natalizumab-based treatment. A "wash out" period of at least 1 month for immunomodulatory drugs and 6 months for immunosuppressive drugs was mandatory before initial natalizumab administration. According to the therapeutic protocol, an intravenous dose of 300 mg of natalizumab was administered every 4 weeks. As shown in Table 2 all samples were stratified according to natalizumab infusion number (N0: no infusion, N3: three infusions, N6: six infusions, N12: twelve infusions, N15: fifteen infusions, N18: eighteen infusions and N24: twenty-four infusions).
All patients regularly underwent a complete physical and neurological examination and neurological disability was assessed with the Expanded Disability Status Scale (EDSS) score 38 . As a control group, 10 healthy donors (HD) age and sex matched with RRMS patients were enrolled. Sample collection. Blood samples were collected before natalizumab administration in heparin and ethylenediamine tetra-acetic acid (EDTA) tubes and plasma was obtained from whole blood after centrifugation and stored at −80 °C. Heparin plasma was used for zymography while EDTA plasma was used for viral genome detection. Heparin whole blood was used for multi-color flow cytometry immunophenotyping.
MMP-9 plasma activity detection. As described by Liuzzi et al. 10 , MMP-2 and MMP-9 plasma levels were detected by zymography as white bands of digestion on the blue background of the gel and were identified by co-localization on the zymogram with human MMP-2 or MMP-9 standards (ALEXIS Biochemicals, San Diego, CA, USA). Quantitation of MMP-2 and MMP-9 levels were performed using computerized image analysis (Image Master 1D, Pharmacia Biotech, Buckinghamshire, UK) through one-dimensional scanning densitometry (Ultroscan XL, Pharmacia Biotech). MMP levels were expressed as optical density (OD) × mm 2 , representing the scanning area under the curves, which considers both brightness and width of the substrate lysis zone. An example of zymogram gelatin gel is represented in Fig. 1a. Viral genome detection. Viral DNA was extracted from plasma obtained from EDTA blood with the DNeasy Blood & Tissue Kit (QIAGEN, S.p.A, Milan, Italy), according to the manufacturer's instructions. The extraction product was amplified with a real time PCR (qPCR) system using a 7300 Real-Time PCR System (Applied Biosystems, USA) and specific primers and probes, as previously described 39 .