A 21‐bp deletion in the complement regulator CD55 promotor region is associated with multifocal motor neuropathy and its disease course

To further substantiate the role of antibody‐mediated complement activation in multifocal motor neuropathy (MMN) immunopathology, we investigated the distribution of promotor polymorphisms of genes encoding the membrane‐bound complement regulators CD46, CD55, and CD59 in patients with MMN and controls, and evaluated their association with disease course.


| INTRODUCTION
Multifocal motor neuropathy (MMN) is a rare, asymmetric, immunemediated motor neuropathy that mostly affects the relatively young to middle-aged.2][3][4] Although MMN responds to immunoglobulin treatment, approximately 20% of patients develop debilitating loss of hand and arm function. 1Patient characteristics associated with the heterogeneity in MMN severity are largely unknown. 1,4,50][11] Higher anti-GM1 IgM antibody titers, increased complement deposition triggered by anti-GM1 IgM in vitro and higher innate activity of the classical pathway of complement are all associated with more severe muscle weakness. 8,9,11These findings support the hypothesis that activation of complement is central in the inflammatory processes underlying MMN.
The contribution of variation in innate complement regulation in MMN has not been studied.Complement activation is tightly regulated at the fluid and tissue levels, among others by membrane-bound complement regulatory proteins (mCRPs), including CD46, CD55, and CD59. 12,135][16] Common polymorphisms in the promotor region of these mCRPs lead to variation between individuals in transcriptional activity and cellular mCRP expression.In general, higher mCRP expression will lead to higher thresholds for complement activity, and vice versa.[19][20][21] To further our understanding of the role of complement regulation in MMN immunopathology, we analyzed common polymorphisms in the promotor regions of CD46, CD55, and CD59 in MMN patients and controls and determined their association with MMN susceptibility and its disease course.

| Study population
All patients with MMN were diagnosed and enrolled at the outpatient clinic of the University Medical Center Utrecht (UMCU), a tertiary neuromuscular referral center and national center for MMN.All patients fulfilled the most recent diagnostic criteria for definite, probable, or possible MMN. 22These criteria rely on the combination of a typical clinical phenotype combined with conduction block found on nerve conduction studies or, in the absence of conduction block, on abnormal ancillary investigations and/or a response to treatment with immunoglobulins.
Dutch population-based controls were enrolled through the Prospective ALS study The Netherlands (PAN), a population-based casecontrol study performed in the UMCU. 23Control subjects did not have a motor neuron disorder, and those enrolled between January 2012 and August 2018 were included in this study.

| Clinical data
2][3][4] When necessary, we supplemented these data with the most recent data from the UMCU patient files.
Recorded baseline characteristics included sex, age at onset, diagnostic delay, treatment with IVIg, the presence of anti-GM1 IgM antibodies, and muscle strength testing on the first and last visit to the UMCU outpatient clinic.
We defined the onset of disease as the time of the first complaint of muscle weakness and diagnostic delay as the time between onset and diagnosis.We documented the presence or absence of anti-GM1 IgM antibodies as described previously. 24At the first visit and at the last follow-up visit to the UMCU, we quantified muscle strength using the 6-point Medical Research Council (MRC) scale.The MRC scale ranges between 0 (no contraction) and 5 (normal muscle strength against resistance).MRC scores were documented for left and right shoulder abduction, elbow flexion and extension, wrist flexion and extension, finger flexion and extension, finger spreading, hip flexion, knee flexion and extension, and foot dorsal and plantar flexion.We calculated an MRC sum score (MRCss) by summation of the MRC scores of all tested muscle groups (range 0-130) for the first and last visit to our hospital.

| DNA samples
We used standard DNA isolation methods to extract genomic DNA from whole blood obtained during two national studies on MMN performed in 2007 and 2015. 1 We used control DNA samples that were obtained upon participation in the PAN study.

| Genotyping
We used two techniques to determine the presence of five CD46, CD55, and CD59 promotor region variants, which were chosen because of both their common presence in healthy individuals (i.e., with a minor allele frequency [MAF] of at least 20%) and their previously described association with disease.
First, in patients with MMN only, we used the previously described Sanger sequencing methodology to genotype five common polymorphisms that have previously been found associated with disease or disease outcome. 17,18Next, we designed a second primer set to specifically target the polymorphisms identified in patients with MMN to genotype the control cohort.These primers and PCR programs are summarized in Table S1.We determined optimal annealing temperatures using a temperature gradient PCR.For CD46 rs2796267 and rs2796268, CD55 rs28371583, and CD59 rs141385724, we amplified genomic DNA by PCR, purified the PCR using Sephadex and subsequently performed Sanger sequencing. 25We validated the sequencing result using eight MMN samples of which we had obtained the full promotor region sequence of CD46, CD55, and CD59.Results matched in all samples.
We genotyped CD55 rs28371582 by gel electrophoresis of the PCR product.Since rs28371582 is a 21-base pair deletion, we designed primers predicted to provide products of 159 out of 180 base pairs, which could subsequently be resolved by gel electrophoresis to determine homozygous (Ins/Ins or Del/Del) and heterozygous (Ins/Del) genotypes.We included controls for each genotype (previously determined by Sanger sequencing) on each gel.We validated the genotyping results of 20 MMN samples by comparing results with the whole CD55 promotor sequence.These results correlated 100%.MMN samples that could not be genotyped by amplification of the whole promotor region of either CD46, CD55, or CD59, were retested once using the polymorphism-specific PCR approach.

| Statistical analysis
We used R (version 3.5.1) to perform statistical analyses.We used a Χ 2 test to determine the association between the polymorphisms and MMN susceptibility.When a p value <.05 was found in the overall genotype comparison, post-hoc Χ 2 tests with Bonferroni p value adjustment were performed per genotype, and a p value <.017 was considered statistically significant.Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated.We used the continuous variables age at onset, first (i.e., baseline) visit, and last visit MRC sum score (FV MRCss and LV MRCss, respectively) as clinical parameters.We compared age at onset in all patients.
The MRCss variables were compared in patients who were treatmentnaïve at baseline only.In all comparisons, patients were grouped as either carriers (i.e., having a heterozygous or homozygous positive genotype) or non-carriers (i.e., having a homozygous negative genotype).
We compared age at onset using a Student's t test.Next, we determined the association between either of the polymorphisms and muscle strength.First, we used a linear regression analysis to determine the association between a polymorphism and the FV MRCss, including the FV MRCss as an outcome variable, disease duration at the first visit, and polymorphism status as independent variables.To avoid outlier-driven effects, patients with a disease duration longer than 20 years at their first visit were excluded from MRCss analyses.The mean MRCss deterioration over time and its mean difference between the polymorphism groups were obtained.Thus, modeling the pre-diagnostic disease course, the model's interaction term indicated whether the change in muscle strength over time was different for patients carrying any of the polymorphisms in question.Second, we determined the effect of a polymorphism on the MRCss change between the first and last visit to our hospital using a linear mixed effects model (R lme4 package), including the FV and LV MRCss as separate values.The fixed effect part contained polymorphism status, time since baseline in months and the interaction between polymorphism and time and included a random intercept per patient.In this analysis too, the interaction term indicated whether the change in muscle strength over time differed according to polymorphism status.
Third, since it is known that anti-GM1 IgM antibody status is associated with MMN disease course, both MRCss analyses were performed in an anti-GM1 IgM antibody status independent and dependent way. 4,8,11In the anti-GM1 IgM antibody status dependent way, antibody status was introduced as extra independent variable/ fixed effect.In all three analyses, we obtained the mean MRCss deterioration over time for both groups, and calculated the mean difference between groups with its 95% CI and p value.Since we have selected the polymorphisms based on their association with the disease instead of studying all known polymorphisms in the CD46, CD55, and CD59 promotor region, and since the clinical correlation analysis was exploratory, we did not perform a p value correction for these analyses.
p Values <.05 were considered statistically significant.

| Clinical correlation
Next, we investigated whether specific genotypes of the complement regulatory proteins were associated with clinical features of MMN (Figure 1).We found no association between any polymorphism and age at onset.

| DISCUSSION
In this study, we show that MMN susceptibility is associated with a polymorphism in the promotor region of the complement regulatory protein CD55, and its disease course with promotor polymorphisms of both CD55 and CD59.More specifically, we found that MMN was significantly associated with rs28371582, a 21-bp deletion in the promotor region of CD55 ( p = .009;Del/Del genotype 16.8 vs. 7.7%, OR: 2.43 [1.27-4.58],p = .005),while CD59 rs141385724 was associated with less severe muscle weakness at baseline in treatmentnaïve patients.Patients carrying CD55 rs28371582 (55% of all patients) also showed a more favorable disease course after starting IVIg treatment.
7][28] CD55 encodes for decay-accelerating factor (DAF), a GPI-linked protein that regulates complement by enhancing the decay of the C3 convertases C4bC2a and C3bBb.CD55 thus inhibits the formation and amplification of C3b, the central complement factor. 13,26,27e CD55 promotor polymorphism rs28371582 has functional consequences.0][31] Interestingly, the CD55 rs28371582 deletion contains three such possible TC n elements.CD55 rs28371582 is associated with a lower CD55 transcriptional activity, suggesting lower cellular expression levels. 32,33Reduced CD55 expression has been associated with multiple autoimmune diseases, including systemic lupus erythematosus, autoimmune hemocytopenia, and myasthenia gravis, the latter also being associated with a CD55 promotor polymorphism. 20,21,34Lower CD55 expression or reduced CD55 upregulation may promote the formation of C3 convertase and thus the deposition of C3 fragments on peripheral nerves. 35This may explain the observed association of this CD55 promotor polymorphism with MMN susceptibility.
We found two associations of genetic variability of CRPs and MMN disease course.First, treatment-naïve patients carrying CD59 rs141385724 showed a trend of less severe weakness at the first visit.CD59 rs141385724 has previously been believed to have a tissue-and inflammation-specific functional effect and we hypothesize that it is associated with increased CD59 expression in motor neurons. 17,18After initiation of IVIg therapy, patients carrying this polymorphism had a similar disease course as compared to noncarriers.Combined with the negative result of an open-label trial where IVIg-treated patients with MMN were treated with eculizumab, an anti-C5 monoclonal antibody that exerts a CD59-like effect on the terminal complement cascade, the results of our study may indicate that muscle strength deterioration in patients with MMN treated with IVIg is mediated through complement factors of the proximal (C2, C4, and C3), rather than the terminal pathway (C5-C9, i.e., MAC). 36We hypothesize that CD59 expression on peripheral nerves is sufficiently C1q, thereby inhibiting C1q-initiated complement activation. 37IVIg, like CD55, also modulates complement at the C3 convertase level by scavenging activated C3 through binding to C3b.9][40][41] If complement factors such as C3 and C4 play an important role in MMN pathogenesis as suggested by experimental models, carriership of CD55 rs28371582 may provide a context in which IVIg may exert more pronounced beneficial effects. 10,35r study is the first study on genetic susceptibility factors involved in innate complement regulation in MMN.As MMN is a rare disease, the size of our cohort is significant and includes a long follow-up time with a median duration of over 8 years.Additionally, our control cohort is large and population-based.Since the MRC sum scores were collected retrospectively, relatively small groups were compared in the analyses and any change in MRCss was reduced to a linear correlation; the results of these correlations should be interpreted with care.Future in vitro studies in motor neurons could aid in finding biological explanations for the associations found in this study. 10Moreover, such studies could help in forming a stronger theoretical basis for treatment strategies targeting the pre-C5 level of the complement cascade in patients with MMN. In , and CD59 rs141385724, each assigned a separate row, correlated to clinical parameters in patients with MMN.The left column shows the age at onset in years.The middle column depicts the MRC sum score (MRCss) at the first visit in untreated patients, plotted against the disease duration at the first visit in years on the x-axis.The right column shows the course of the MRCss between patients' first and last visit in patients untreated at the first visit, plotted against MMN disease duration in years on the x-axis.Patients were grouped by either carrying or not carrying the polymorphism shown in each row (red: carriers; blue: non-carriers).Patients with MMN who were treatment-naïve at the time of their first visit and who carried CD59 rs141385724 showed less severe muscle weakness at their first visit (p = .032).CD55 rs28371582 was associated with less severe clinical deterioration during the follow-up period in patients with MMN who were treatment-naïve at baseline ( p = .019).MMN, multifocal motor neuropathy; MRC, Medical Research Council.high to-in combination with IVIg-hold MAC formation at bay after binding of anti-ganglioside antibodies to myelin or (para)nodes.The CD55 promotor polymorphism rs28371582 not only associated with susceptibility, but also showed a second association with disease course.Treatment-naïve patients carrying CD55 rs28371582 actually showed a trend to less severe muscle strength deterioration after IVIg therapy, despite the implication of lower CD55 expression and hence increased susceptibility to complement-mediated nerve damage.We believe that this association is best explained by the complement modulatory effects of IVIg.Monomeric IgG can compete directly with T A B L E 1 Baseline characteristics of patients with MMN and control subjects.We included 133 patients with MMN and 380 control subjects.Baseline characteristics are shown in Table 1.Age at onset (MMN) or age at inclusion in the PAN study (controls) is shown.The first visit MRC sum score is depicted for patients with MMN who were treatmentnaïve at their first visit (n = 103).The last visit MRC sum score is calculated only for patients with MMN who were treatment-naïve at baseline and for whom follow-up data were available.
None of the polymorphisms in CD46 and CD55 were 95% CI: 0.04-0.49,p .019), suggesting a more favorable long-term disease course after initiation of IVIg treatment.All other polymorphisms were not associated with the MMN disease course.Neither the FV MRCss, nor the longitudinal MRCss data were different among patients with or without anti-GM1 IgM antibodies ( p values >.05 for all analyses).
conclusion, our study reports novel findings that point towards the relevance of complement regulation at the C3 level in multifocal motor neuropathy.We show that a 21-bp deletion in the promotor region of CD55 is associated with MMN susceptibility and that patients carrying this polymorphism showed a trend to a better long-term clinical response upon IVIg treatment.With new therapies that target the early stages of the complement cascade emerging, future studies should aim to further our understanding of the importance of the proximal complement cascade in MMN immunopathology.