Microfibrillar-associated protein 4 as a potential marker of acute relapse in inflammatory demyelinating diseases of the central nervous system: Pathological and clinical aspects

Background: Microfibrillar-associated protein 4 (MFAP4) is an extracellular matrix protein not previously described in the human central nervous system (CNS). Objectives: We determined MFAP4 CNS expression and measured cerebrospinal fluid (CSF) and serum levels. Methods: Tissue was sampled at autopsy from patients with acute multiple sclerosis (MS) (n = 3), progressive MS (n = 3), neuromyelitis optica spectrum disorder (NMOSD) (n = 2), and controls (n = 9), including 6 healthy controls (HC). MFAP4 levels were measured in 152 patients: 49 MS, 62 NMOSD, 22 myelin oligodendrocyte glycoprotein-associated disease (MOGAD), and 19 isolated optic neuritis (ION). Results: MFAP4 localized to meninges and vascular/perivascular spaces, intense in the optic nerve. At sites of active inflammation, MFAP4 reactivity was reduced in NMOSD and acute MS and less in progressive MS. CSF MFAP4 levels were reduced during relapse and at the onset of diseases (mean U/mL: MS 14.3, MOGAD 9.7, and ION 14.6 relative to HC 17.9. (p = 0.013, p = 0.000, and p = 0.019, respectively). Patients with acute ON (n = 68) had reduced CSF MFAP4 (mean U/mL: 14.5, p = 0.006). CSF MFAP4 levels correlated negatively with relapse severity (rho = −0.41, p = 0.017). Conclusion: MFAP4 immunoreactivity was reduced at sites of active inflammation. CSF levels of MFAP4 were reduced following relapse and may reflect disease activity.


Introduction
Inflammation during disease activity in inflammatory demyelinating diseases (IDDs) of the central nervous system (CNS) frequently associates with disruption of the blood-brain barrier (BBB) and clinical exacerbation. 1The three best-defined IDDs are multiple sclerosis (MS), aquaporin-4-IgG positive neuromyelitis Microfibrillar-associated protein 4 (MFAP4) is an extracellular matrix (ECM) protein and a member of the fibrinogen-related domain family. 4MFAP4 is expressed in the perivascular ECM of microvessels, for example, in alveolar septa. 5,6MFAP4 binds to arginylglycyl-aspartic acid (RGD)-dependent cellular integrin receptors, including integrin α V β 3 and α V β 5 . 6,7The ECM is a dynamic component of the neurovascular unit of BBB and ECM and its receptors and plays an important regulatory role in maintaining homeostasis and BBB integrity. 4,8,9Soluble MFAP4 is recognized as a marker of the matrix remodeling processes in peripheral vascular diseases. 4,5,10In a pilot study, we provided preliminary data that MFAP4 levels in cerebrospinal fluid (CSF) are reduced in patients with acute optic neuritis (ON) in population-based samples, and that MFAP4 is expressed in the optic nerve. 11Therefore, in two parallel and independent studies, we aimed to (1) determine the localization of MFAP4 in the human CNS (pathology part); and (2) measure CSF and serum MFAP4 levels in patients with IDDs (clinical part).

Pathology study
Tissue was sampled at brain and spinal cord autopsy from patients with different stages of MS and NMOSD and from controls (Table 1).The tissues originated from Austria.For validation purposes, tissues sampled from NMOSD (1 optic nerve) from Japan, and HC (1 brain and 3 optic nerves) from Denmark, were investigated.
A total of seven control subjects without infection, neurological disease, brain, or optic nerve lesions were included (Table 1).

Immunohistochemistry
The monoclonal anti-MFAP4 antibody (HG-HYB 7-14) stained for MFAP4, and the isotype control anti-OVA (HYB099-1) antibody was used as a negative control (State Serum Institute, Copenhagen, Denmark).C57BL/6/N Mfap4-deficient mice were immunized to produce monoclonal antibodies (HG HYB 7-14) against rMFAP4 as previously described. 12he specificity of the anti-MFAP4 antibody has previously been demonstrated. 10In brief, we have produced Mfap4-deficient mice and performed staining of MFAP4.Positive staining of MFAP4 was evident in wildtype and complete absence of staining was seen in Mfap4-deficient mice. 134-µm-thick sections were cut from formalin-fixed, paraffin-embedded tissue blocks.The sections were mounted on FLEX IHC microscope slides (Dako/Agilent, Glostrup, Denmark), and they were dried at room temperature and baked at 60°C for 60 minutes before immunostaining.Staining was automated at the Discovery Ultra immunostainer (Ventana Medical Systems, Tucson, AZ) using the OmniMap anti-goat-HRP detection system (Ventana) and Discovery Purple chromogen kit (Ventana).Incubation with goat antihuman IgG Fc (Abcam, Cambridge, UK), diluted at 1:2000, was done for 32 minutes at 36°C.Epitope retrieval was performed in Cell Conditioning Solution 1 (CC1) for 32 minutes at 100°C.Nuclear counterstaining was performed using hematoxylin II (Ventana).Finally, the slides were washed, dehydrated, and coverslipped using an automated Dako coverslipper (Dako/Agilent).
In parallel sections, staining was performed as described for AQP4, 14,15 glial fibrillary acidic protein (GFAP), 14,15 and C9neo (a marker of membrane attack complex), 14,15 hematoxylin and eosin.(HE), 15 Iba-1 13,16 and Klüver-Barrera (KLb). 13,16 situ hybridization.The RNA chromogenic in situ hybridization was performed using a well-established enhanced RNAscope 2.5 procedure.Laboratory methods.Blood and CSF were collected according to international research standards. 24antification of MFAP4 was done with an AlphaLISA assay (Perkin Elmer, MA, USA) at the University of Southern Denmark.Details of the procedure are described elsewhere. 5Samples were run in duplicates, and occasional samples with CV > 10% (12/161 samples 7.45%) were reanalyzed to obtain a valid measurement for every sample.Patients were tested for AQP4-IgG and MOG-IgG using live cell-based assays, as previously described.

Pathology
MFAP4 is expressed within the healthy CNS.MFAP4 was expressed in the connective tissue spaces of CNS from all controls, including meninges and the vascular/perivascular spaces (Figure 1).MFAP4 immunoreactivity was present in the ECM in the intima and adventitia of larger vessels (Figure 1

Widespread MFAP4 immunoreactivity throughout the optic nerve.
There was much more connective tissue around the vessels of the optic nerve than in the brain parenchyma, occasionally forming septae between nerve fiber bundles.MFAP4 reactivity predominated in these connective tissue bundles using immunostaining (purple) for MFAP4 (Figure 2(a)).In situ hybridization analysis demonstrated that MFAP4 was expressed in small vessels and the capillaries in healthy optic nerve (Figure 2 In progressive MS, MFAP4 immunoreactivity was seen in vascular/perivascular spaces and meninges and cortical sulcus (Supplemental Figure 2(a)).In addition, MFAP4 immunoreactivity was diminished in lesions with demyelination in progressive MS to a lesser degree than in acute MS (Supplemental Figure 2(b)).In patients with stroke, MFAP4 immunoreactivity was detected to a similar degree as in the other controls in the connective tissue spaces, including meninges and ECM in the intima and adventitia of larger vessels (Supplemental Figure 2(c) and (d)).

Clinical study
Demographics and disease characteristics.Serum and CSF were collected from 161 patients with IDDs.Of the 161 patients, 9 were excluded because either AQP4-IgG or MOG-IgG were not tested.Of the resulting 152 patients, 62 were AQP4-IgG positive NMOSD, 22 MOG-IgG positive (MOGAD), and 68 were double-seronegative.In all, 19 patients out of the 68 had isolated optic neuritis (ION) and 49 patients had MS.Samples from MS patients were obtained before treatment with corticosteroids or disease-modifying drugs (Table 2 and Supplemental Figure 3).

No correlation between MFAP4 serum and CSF levels
Serum MFAP4 (sMFAP4) levels in IDD patients did not differ (mean U/mL: 27.4) compared with HC (19.9; p = 0.149).When stratified by IDD type, the mean sMFAP4 level was elevated only in MS patients (mean U/mL: 32.0 MS vs. 19.9HC; p = 0.025).No correlation was observed between sMFAP4 and CSF MFAP4 levels in IDD patients (rho = 0.14 p = 0.25), and in the fraction of patients with MS (rho 0.15 = p = 0.265).
In MOGAD CSF MFAP4 levels were decreased significantly in the acute stage relative to chronic stage (mean U/mL: 13.1 vs. 21.1;p = 0.001), independent of age.

CSF MFAP4 levels are affected by number of relapses
Non-Caucasian patients (n = 26, predominantly with NMOSD (20), consisting of Afro-Americans, 2 Asians, 24 and Latin Americans 1 ) experienced more attacks (median 2) before sample collection than did Caucasians (n = 87), (median 1) (p = 0.000).Comparison between age, sex, and severity of attacks in non-Caucasian patients versus Caucasians did not reveal any differences.

CSF MFAP4 levels and time from onset of recent relapse to sample collection
Univariate analysis indicated in patients with MOGAD in the acute stage a correlation between CSF MAP4 levels and time from onset of recent relapse to sample collection (rho = 0.61, p = 0.007) (95% CI = 0.08: 0.36), (Figure 6(e)), and strongly associated with time to sample collection within ⩽15 days (rho = 1), (Figure 6(f)).Based on a linear regression model, average MFAP4 levels increased to 0.22 U/mL per day following attack onset.From these data we have estimated a time limit on sampling CSF as 6.2 weeks.

CSF MFAP4 levels as a biomarker for acute optic neuritis
Patients with acute ON relapse (n = 68) had reduced CSF MFAP4 levels (mean U/mL: 14.5 vs. 17.9;Correlation between CSF MFAP4 levels and relapse severity MFAP4 levels were not associated with EDSS in IDD patients.In total, a relapse severity score at the most recent relapse prior to lumbar puncture was available for 44 IDD patients.Relapse severity score was associated with lower CSF MFAP4 levels independent of number of relapses in acute IDDs (rho = −0.41,p = 0.017) (95% CI = −0.13:−0.69), (Figure 8).

Discussion
Our data demonstrate MFAP4 protein immunoreactivity in the vascular/perivascular spaces and meninges, particularly of the optic nerve.At sites of active inflammation in tissue from NMOSD and acute MS, MFAP4 immunoreactivity was reduced, suggesting that inflammation altered the composition of extracellular matrix.Clinically, CSF MFAP4 levels were reduced in the acute stage during relapses of IDDs and more pronounced in patients with MOGAD and acute ON relapse.In MOGAD we observed an association between CSF MFAP4 levels and time from onset of recent relapse.A positive correlation was observed between number of relapses and CSF MFAP4 levels, suggesting a dynamic reorganization of MFAP4 in CNS as a reaction to inflammatory relapses.Relapse severity was negatively associated with CSF MFAP4 levels, suggesting that MFAP4 may serve as a potential marker of disease activity and relapse severity.Overall, the clinical findings are consistent with the pathological data since a reduction of CSF MFAP4 occurred in acute disease stages.The recognized cellular receptor for MFAP4 is integrin αVβ3. 10Two main receptors/adhesion proteins are involved in the cell-cell and cell-matrix interactions of the BBB: dystroglycan and integrins. 9,10In our study, MFAP4 was absent in CNS during relapse as documented by the presence of inflammatory infiltrates in parallel sections, suggesting that MFAP4 remodeling may occur as a consequence of BBB damage and vascular reorganization.Notably, in situ hybridization of the optic nerve demonstrated mRNA MFAP4 expression in the capillaries.
Reduction of MFAP4 reactivity at sites of active inflammation in NMOSD and active MS suggests that MFAP4 reflect CNS tissue damage and possibly relate to astrocytes and oligodendrocytes being the target as well as the source of MFAP4.In line with pathology data, substantial CSF MFAP4 levels reduction was seen in the acute stage, and the reduction was more pronounced in patients with MOGAD and acute ON.Notably, the reduction of CSF MFAP4 correlated with the severity of all types of relapses.
This study raises a potential role of MFAP4 in severity of relapses.The pathology data were collected without knowledge of clinical data, including CSF   A negative correlation between CSF MFAP4 levels and relapse severity was observed.
and serum levels and vice versa, diminishing bias in the study.Patients originated from a large cohort of samples from multiple centers, which increases the generalizability of the findings.The study conclusions are limited by the cross-sectional design of our study; longitudinal design with consecutive samples will be necessary for further validation, including treatment effects.A large prospective study designed to validate these observations in our study would be an obvious continuation.
MFAP4 immune reactivity is increased in liver fibrosis, cardiovascular disorders, and asthma. 27,28In these diseases the MFAP4 levels are increased in serum, whereas we in this study observed that CSF levels in acute IDDs were decreased.
In the present study, the mean serum MFAP4 level was elevated only in MS patients.However, no correlation was observed between sMFAP4 and CSF MFAP4 levels including the MS patients.CSF MFAP4 levels were decreased in patients with MS, in line with the pathology data.All controls in our study were without morbidity whereas limited information was obtained on comorbidity in the IDD patients.Future studies will illuminate the putative effects of comorbidity.
In conclusion, MFAP4 is expressed in the human CNS and diminished at sites of active inflammation.CSF MFAP4 levels were lower in the acute stage, especially in patients with MOGAD and acute ON, where it correlated negatively with the severity of the relapse.Our findings underscore the timepoint of MFAP4 determinations relative to an attack and suggest a role for MFAP4 of disease activity and relapse severity.

Figure 1 .
Figure 1.MFAP4 immunoreactivity in the control brain.Micrographs show tissue sampled from control brain.(a-f) Sections were stained for MFAP4 using monoclonal antibodies against MFAP4.(a-c, e, and f) MFAP4 was expressed to a variable extent in the connective tissue spaces of meninges and around vessels, (e and f) including pial vessels and (d) choroid plexus, denoted by black arrows.Scale bars = 200 µm (a), 100 µm (c), 50 µm, (b, d, f), and 20 µm (e).
(b)).Negative controls showed lack of staining (Figure 2(c) and (d)).Reduction of MFAP4 reactivity at sites of active inflammation.At sites of inflammation, documented by the presence of inflammatory infiltrates in parallel sections, MFAP4 expression was diminished in NMOSD (Figure 3(a)) in areas that colocalized with loss of AQP4 (Figure 3(b)), GFAP staining (Figure 3(f)), and increased microglia activation with Iba-1 reactivity (Figure 3(e)).MFAP4 positive staining was preserved and co-localized (Figure 3(c)) with AQP4 (Figure 3(d)) and GFAP (top of Figure 3(f)) immunoreactivities in the perilesional area.This observation was confirmed in optic nerve tissue from an NMOSD patient showing a reduction of MFAP4 reactivity co-incident with loss of AQP4 expression and deposition of C9neo and infiltration of neutrophils and macrophages (Supplemental Figure 1).In acute MS, demyelination, and cellular infiltration (Figure 4(a)) coincided with diminished MFAP4 immunoreactivity (Figure 4(b) and (c)).Weak MFAP4 reactivity was seen at the sites of preserved myelin (Figure 4(b) (insert), and in the perilesional demyelination area.

Figure 2 .
Figure 2. MFAP4 immunoreactivity in a control human optic nerve.Micrographs show longitudinal sections of the optic nerve.(a) Immunostaining for MFAP4 indicated in purple.Insert in (a) shows the whole tissue cross section of the eye and the red box indicates the magnified area.(b) Punctuate in situ hybridization (indicated in red) of MFAP4 mRNA.(c) Corresponding negative controls, respectively, with isotype antibody control staining or (d) in situ hybridization without probe pair.Widespread MFAP4 immunoreactivity was observed throughout in the optic nerve, including the perivascular areas and in small vessels and capillaries (a and b).Areas of MFAP4 immunoreactivity are marked by black arrows.Lack of MFAP4 detection in negative controls (c and d).Scale bars = 250 μm (a and c); 100 μm (b and d).

Figure 3 .
Figure 3. MFAP4-negative staining co-localizes with loss of AQP4 in NMOSD.Micrographs show sections of the mesencephalon from an NMOSD patient.Sections were stained for MFAP4 (a and c), AQP4 (b and d), Iba1 (e), and GFAP (f) (indicated by brown staining).(b) MFAP4-negative staining co-localizes with loss of AQP4; (c) MFAP4 positive staining area adjacent (left) to a lesion; (d) AQP4 positive area adjacent (right) to a lesion; (e) microglia activation, Iba-1 reactivity in corresponding areas to (a) and (b); and (f) GFAP staining: reduction of GFAP in areas corresponding to areas in (a), (b), and (e).Areas of histopathological changes are marked with arrows.Original magnification 4× (a-f).

Figure 4 .
Figure 4. MFAP4 immunoreactivity is diminished in acute MS.Micrographs show sections of an active parenchymal brain lesion in an MS patient.Sections were stained for hematoxylin and eosin.(HE), Klüver-Barrera (KLb) (HE-KLb) (a) and MFAP4 (b and c, brown staining).(a) The image shows demyelination and massive inflammation, with myelin staining in bottom-left corner.(b) MFAP4 staining reveals a similar pattern to the myelin staining (a) (insert b), with additional staining at vessel in the center.(c) The bottom of the same areas as shown in (a) and (b) is illustrated by MFAP4stained sections with MFAP4-positive staining area adjacent (below) to a lesion.In (c) the bottom of the same areas as shown in (a) and (b) has MFAP4 positive staining in the area adjacent to a lesion.Original magnification 25× (a-c).

Figure 5 .
Figure 5. Reduction of CSF MFAP4 in the acute stage.Comparing IDD patients in the acute stage and HC showed a trend toward (a) decreased CSF MFAP4 levels (p = 0.05), (b) significantly in MOGAD, ION, and MS.(c) Comparison between CSF MFAP4 levels in Caucasians and non-Caucasians showed lower CSF MFAP4 levels in Caucasians.(d) Comparing Caucasian IDD patients in the acute stage and HC showed significantly decreased CSF MFAP4 levels.

Figure 6 .
Figure 6.A positive correlation between number of relapses and CSF MFAP4 levels.(a) Number of relapses was associated with higher CSF MFAP4 levels.(b) CSF MFAP4 decreased relative to HC in patients at presenting attack, (c) specifically for MOGAD patients and ION patients.(d) In MOGAD the CSF MFAP4 levels decreased more significantly relative to HC in patients with less relapses: >2 relapses (mean U/mL: 13.1; p = 0.02), 2 relapses (mean U/mL: 12.3; p = 0.006), one relapse (mean U/mL: 9.7; p = 0.000).(e) CSF MFAP4 levels in the acute stage within ⩽60 days was positively associated with time from recent relapse to sample collection (rho = 0.61), and (f) associated strongly with time to sample collection within ⩽15 days (rho = 1) in MOGAD.

Figure 7 .
Figure 7. CSF MFAP4 as a marker for acute optic neuritis relapse.

(
a and b) Patients with acute ON relapse had significantly reduced CSF MFAP4 levels.

Table 1 .
Pathology study: clinical information of patients with inflammatory demyelinating diseases of CNS and controls.
25,26Statistics Statistical analyses were performed using Stata 16 and 17 SE (StataCorp, USA).Multiple linear regression was used to compare MFAP4 across groups.
Model validation was performed by graphical inspection of quantile plots of residuals and plots of residuals against fitted values.When model assumptions appeared violated, concentrations were analyzed following logarithmic transformation.Consequently, MFAP4 in serum and CSF were analyzed on a log 10scale.If necessary, the comparison was supplemented by an overall nonparametric Kruskal-Wallis test.Test for trend across ordered groups was performed equidistantly coding the groups and entering the grouping as continuous variable into the regression.Correlations were measured using Spearman's rank correlation.This was a hypothesis-generating exploratory study so no adjustment for multiple comparisons was made.p-valuesbelow0.05 were considered statistically significant.Standard protocol approvals, registrations, and patient consentsThe Research Ethical Committees for the Region of Southern Denmark approved the study protocol (ref.no.S-20130137 and ref. no.S-20080142) as did the Danish Data Protection Agency (ref.no.14/26345).Approval by local ethical Institutional Review Boards from all centers were obtained following informed consent from the patients.

Table 2 .
Clinical and demographic data of the study population.