Cerebrospinal fluid growth-associated protein 43 in multiple sclerosis

Neurodegeneration in multiple sclerosis (MS) correlates with disease progression and reparative processes may be triggered. Growth-associated protein 43 (GAP-43) exhibits induced expression during axonal growth and reduced expression during MS progression. We aimed to evaluate if GAP-43 can serve as a biomarker of regeneration in relapsing-remitting MS (RRMS) and whether disease-modifying therapies (DMTs) influence GAP-43 concentration in cerebrospinal fluid (CSF). GAP-43 was measured using an enzyme-linked immunosorbent assay in 105 MS patients (73 RRMS, 12 primary progressive MS, 20 secondary progressive MS) and 23 healthy controls (HCs). In 35 of the patients, lumbar puncture, clinical assessment, and magnetic resonance imaging was performed before initiation of therapeutic intervention, and at follow-up. CSF GAP-43 concentration was significantly lower in progressive MS compared with HCs (p = 0.004) and RRMS (p =  < 0.001) and correlated negatively with disability (p = 0.026). However, DMTs did not alter CSF GAP-43. Interestingly, in RRMS CSF GAP-43 levels were higher in patients with signs of active inflammatory disease than in patients in remission (p = 0.042). According to CSF GAP-43 concentrations, regeneration seems reduced in progressive MS, increased during disease activity in RRMS but is unaffected by treatment of highly active DMTs.

The influence of clinical and demographic factors and blood-brain barrier function on CSF GAP-43 concentrations. While no correlation was found between CSF GAP-43 concentrations and age in HCs, it correlated negatively with age (r s = −0.339, p < 0.001), disease duration (r s = −0.303, p = 0.002), and EDSS (r s = −0.369, p < 0.001, Fig. 1B) in the MS population. Multiple regression analysis showed that only EDSS independently correlated with CSF GAP-43 concentration (p = 0.026). When dividing the MS population into RRMS, PPMS and SPMS patients, EDSS correlation was significant only in PPMS (PPMS: r s = −0.651, p = 0.03, RRMS: r s = −0.123, p = 0.301, SPMS: r s = −0.242, p = 0.304). After adjustment for age in the PPMS group the correlation was still significant (p = 0.009). No significant differences were found between CSF GAP-43 concentrations in females and males, and baseline GAP-43 concentration did not correlate with the albumin ratio.

Discussion
The data in this study are based on a heterogeneous group of MS patients in different stages of disease. We confirmed that CSF GAP-43 was significantly lower in progressive MS compared with HCs and RRMS patients 21,22 , with the lowest levels in PPMS. CSF GAP-43 concentrations correlated negatively with age, disease duration, and EDSS, but only independently with EDSS. However, we found that the CSF GAP-43 concentration was significantly higher in RRMS patients with signs of active inflammatory disease compared with RRMS patients in remission, whereas fingolimod and alemtuzumab treatment did not alter CSF GAP-43 concentrations.
We confirmed no difference between CSF GAP-43 concentrations of RRMS and HCs 22 . In contrast, the GAP-43 was reduced in progressive MS, suggesting lost or reduced regenerative potential in late MS. This ability seems to become more marked with increased disability and may be the result of atrophy development and progressive neuro-axonal loss. This finding is in contrast to our findings in Alzheimer's disease, in which CSF GAP-43 was increased 23 . Thus, the nature of neurodegeneration seems to be more important than the degree of neurodegeneration, as the level of CSF GAP-43 was not increased in other neurodegenerative diseases 23 and not related to the extent of atrophy 24 . Although the pathogenesis of MS progression is unknown, and clearly different from that of  www.nature.com/scientificreports www.nature.com/scientificreports/ Alzheimer's disease, our results suggest that regenerative processes such as synaptogenesis and axonal outgrowth, is reduced in progressive MS. This interpretation is supported by the decreased expression of GAP-43 in the vicinity of white matter lesions 20 .
Our results suggest that the CSF GAP-43 concentration increases in association with new inflammatory activity in MS patients. Increased CSF GAP-43 concentration was found in RRMS and patients with clinically isolated syndrome with >10 T2 lesions compared with those with fewer T2 lesions 22 . However, increase of CSF GAP-43 during relapses or in the presence of contrast enhancing lesions has not been reported before. This elevation seemed independent of blood-brain barrier function, since no correlation was found between CSF GAP-43 and the albumin ratio. Previous studies of axotomy and experimental models of ischemia, traumatic brain injury, and MS show that GAP-43 protein expression is induced temporarily and adjacent to neuro-axonal damage and the formation of new lesions [12][13][14][15][16][17][18] . Thus, immune-mediated damage of the CNS may explain the transient release of GAP-43 that we found in the CSF of MS patients with ongoing disease activity. Another explanation could be that the CSF GAP-43 concentration increases during MS relapse in an attempt to regenerate injured axons.
We could not show any significant impact of DMTs on CSF GAP-43 concentration. Similar CSF GAP-43 concentrations were observed at baseline in patients without prior treatment and those on first-or second-line treatment. MS treatments primarily reduce CNS inflammation in MS, and not the regenerative capacity. The lack of change in CSF GAP-43 across different therapies suggests that reduced inflammation does not influence regeneration involving GAP-43.
HCs were of younger age than the MS population. However, we found no association between CSF GAP-43 concentrations and age in HCs. While multiple regression analysis revealed an independent relationship between disability and the CSF GAP-43 concentration, this were not the case for disease duration and age. Thus, our study confirmed previous findings 22,23,25 , and thus, differences in age between HCs and patients should not have influenced the results. Similar differences existed between the gender distributions of the study groups. However, neither did gender seem to influence the CSF GAP-43 concentrations. Moreover, we could only report an association between ongoing inflammatory activity and increased CSF GAP-43 levels, but we lacked MRI data on lesion load or cerebral and spinal cord atrophy. Relationships between CSF GAP-43 concentrations and such MRI measures should be further explored to better characterize the possible role of GAP-43 in the pathogenesis of MS.
In conclusion, studies of GAP-43 in MS concordantly show that, this protein is decreased in CSF in progressive MS, and we found an association with disability and also with disease activity. However, effective DMTs had no effect on the CSF GAP-43 concentration. Previous studies have not shown correlation between GAP-43 and NFL in CSF 22 , indicating that axonal damage does not influence the release of GAP-43 in CSF. Although the clinical potential of GAP-43 as a biomarker in MS seems limited at this stage, it contributes to further understand the pathogenesis behind progression, and that of degeneration and regeneration in MS.

Methods
Patients and healthy control subjects. We included 23 HCs and 105 MS patients, including 73 with RRMS, 20 with SPMS, and 12 PPMS, fulfilling the revised McDonald criteria from 2010 26 . Ninety of these patients had previously participated in studies of CSF biomarkers in MS 6,27 , including one investigating the influence of fingolimod treatment 28 . The remaining patients (n = 15) were recently recruited from consecutive patients at the MS Centre, Sahlgrenska University Hospital, Gothenburg, Sweden, to explore the influence of alemtuzumab therapy on CSF biomarker concentrations. At baseline, 24 patients received first-line treatment (19 interferon beta, 4 glatiramer acetate, 1 dimethyl fumarate), 20 received second-line treatment (6 fingolimod, 1 rituximab, 13 natalizumab) and 61 were treatment naïve. Descriptive clinical and demographic characteristics are presented in Table 1.
Clinical evaluation, sampling of CSF, and magnetic resonance imaging. All patients were assessed clinically at baseline by MS-specialized neurologists. The EDSS 29 was used to score neurological deficits and impairment. A relapse was defined as an episode of neurological disturbance lasting for at least 24 h that could not be better explained by another cause 30 . Lumbar puncture was performed at baseline (n = 127), and in the fingolimod (n = 20) and alemtuzumab (n = 15) treatment groups, CSF was obtained again after a median of 7 (range 3-13) and 24 (range 24-26) months, respectively. One patient had only a follow-up lumbar puncture sample. The CSF samples were handled according to the consensus protocol of the BioMS-EU network for CSF biomarker research in MS 31 . MRI of the brain was performed on 66 patients at baseline and in close association with the clinical neurological examinations and lumbar puncture (median 1 month, range 0-7 months). A standard MRI protocol for MS including intravenous gadolinium contrast was performed on a 1.5 or 3 Tesla MRI scanner and included T1, T2, and fluid attenuation inversion recovery (FLAIR) sequences, according to the Swedish guidelines 32 .
CSF GAP-43 analysis. The GAP-43 protein concentration in CSF was determined by an in-house ELISA as described previously 22 , with minor modifications. Briefly, plates were coated with NM4 monoclonal antibody (1.35 μg/mL, Fujirebio, Tokyo, Japan) in carbonate buffer (pH 9.6), and incubated over night at 4 °C. After three washes with phosphate-buffered saline with tween (PBST), wells were blocked with a solution of PBST-milk (2% non-fat dry milk, Biorad, Hercules, CA, US) for 1 hour on a shaker at room temperature and placed at −20 °C for at least 12 hours to enable higher throughput during sample runs. After thawing and three more washes, the detection antibody (polyclonal ABB-135, Nordic Biosite, Täby, Sweden), 50 μL of twofold prediluted samples, and calibrators (recombinant GAP-43) in PBST-milk were co-incubated overnight at 4 °C. Plates were washed three times and secondary antibody (anti-rabbit IgG HRP, Promega, Wisconsin, US) diluted in 1% bovine serum albumin (BSA)/PBST at 1:20000 added and incubated on the bench for 2 hours at room temperature. Wells were washed and 100 μL of 3,3′,5,5′-tetramethyl-benzidine (TMB One, KemEnTech Diagnostics, Taastrup, Denmark) added. Plates were incubated in the dark for 30 min before adding 100 μL of 0.