Multiple sclerosis (MS) is the leading cause of disability in young and middle-aged people in the developed world. It is an inflammatory, demyelinating disease of the central nervous system (CNS), and is widely considered to have an autoimmune etiology. The multistep mechanism of the disease involves inflammation, demyelination, and neurodegeneration [1].

A growing body of evidence, supported by both post-mortem studies and advanced MRI techniques, shows increased CNS iron stores in MS patients, particularly in the sub-cortical gray matter [2–5]. The hypotheses of iron excess as a cause of oxidative stress [6, 7], with possible contribution to neuronal injury and death, has been commonly proposed in other neurodegenerative disorders [5, 8, 9]. Indeed, iron level manipulation has been reported as being neuroprotective and neurorestorative in neurodegenerative diseases [10]. Moreover, it was demonstrated that iron deficiency provides protection from the development of experimental autoimmune encephalomyelitis, the animal model of MS [11].

A contribution to the development of iron-driven oxidative stress in several degenerative disorders is linked to the presence of one or more genetic variants leading to suboptimal iron balance in the tissue [5, 9, 12–14]. Some of the main genes and single nucleotide polymorphisms (SNPs) involved in iron management with possible effects on tissue injury are described below.

The HFE gene, locus 6p21.3, codes for a membrane protein similar to MHC class I-type proteins. This protein modulates iron absorption by regulating the interaction of the transferrin receptor with transferrin, and defects in this gene cause hereditary hemochromatosis [15]. C282Y and H63D are the two commonest disease-associated variants in the HFE gene, and iron-dependent inflammation seems to be influenced by both polymorphisms [12–15]. Apart from hemochromatosis, C282Y increases the risk of iron-dependent skin lesions and affects wound healing in patients with leg iron overload due to chronic venous diseases [14, 16]. Among neurodegenerative disorders, the HFE gene has been investigated as a modulator of the different clinical phenotypes. In the field of MS, controversial data have been published; C282Y was found to be overrepresented among MS cases of North-Western European origin [17], and it was considered a predictor for early onset, as well as the H63D homozygotes or the H63D/C282Y compound heterozygotes [18]. Other groups did not find any significant association, when comparing MS cases with low versus high disability scores [19]. Nevertheless, the C282Y variant has recently been considered a marker of poorer MS prognosis and it has been associated with MS aggressiveness [20].

The FPN1 gene, locus 2q32.2, codes for a multiple transmembrane domain protein. Its official name is Solute Carrier Family 40 (iron-regulated transporter), member 1 (SLC40A1). Differently from other iron transporters, it is the only identified mammalian molecule that exports iron outside the cell [21]. FPN1 expression is finely tuned by the iron responsive element (IRE) in the 5’untranslated region (5’UTR) of mRNA, which, under cell iron overloading, increases protein expression leading to iron exports. Four SNPs and one CGG microsatellite repeat in the FPN1 gene have been studied in relation to HFE[22]. Two of these, -8CG and -98GC, are close to the IRE element and are in complete linkage disequilibrium. To date, no data are available about the role of FPN1 gene variants in MS susceptibility or in other neurodegenerative disorders, and very few data have been reported on their potential role on other iron overload diseases [16].

The HEPC gene, locus 19q13.1, codes for a 25-amino-acid peptide, derived from cleavage of an 84-amino-acid long pro-peptide, which is mainly synthesized by hepatocytes [23]. Its official name is Hepcidin Anti-Microbial Peptide (HAMP), and it is a major regulator of iron balance acting by binding to the FPN1 protein on cell membrane, suppressing it. A polymorphism in the promoter region (−582AG) has recently been described as possibly associated with iron metabolism [24–26], but no data on the HEPC gene variants and neurodegenerative diseases are reported so far.

The TF gene, locus 3q22.1, codes for a molecule that forms a stable complex with the HFE protein facilitating iron transfer via transferrin receptor [27]. The effect of HFE on iron absorption depends on its relationship with the transferrin receptor: HFE variants affect TF binding, determining a loss of HFE-repressor function for TF uptake, thereby increasing iron transport within the cells. A common variant in the TF gene is the P570S (TF, C1C2) [28]. The role of the C2 allele in iron balancing [28, 29] and in neurodegenerative diseases [30, 31] has been debated; nonetheless, a joint effect of the HFE and TF genes, responsible for a greater synergic effect, suggested possible gene-gene and gene-environment interactions [32].

Considering that little is known and that there are controversial data about the role of iron trafficking genes in the natural history of MS, we decided to investigate whether common functional SNPs within the main iron genes might contribute to MS susceptibility, onset, disability/severity, and progression.

Several issues surround iron and neurodegenerative disease, due to the fact that iron is essential in neuronal cell life, yet brain iron accumulation can be toxic [5, 7–9].

Iron imbalance is strongly suspected in MS pathogenesis, even though there is no evidence that systemic iron overload occurs more frequently in MS patients than in general population [36, 37].

In contrast, at the brain level, susceptibility weighted imaging MRI techniques permit to reliably measure iron in the brain and to follow the natural history of iron accumulation. Interestingly, a correlation exists between iron storages and disability, manifested either by cognitive or motor symptoms, suggesting a role in the complex mosaic of MS pathogenesis [38–41].

The exact underlying mechanism by which brain iron accumulates in CNS of MS patients is not fully understood. Iron enters into the brain through the blood–brain-barrier, due to iron transport proteins expressed locally [42] and it is stored according to the efficiency of the transferring receptors. This can be controlled at the post-transcriptional level by iron regulatory proteins (IRPs) that interact with IRE motifs on mRNA to alter the expression on brain endothelial cells, neurons, glia, oligodendrocytes, and macrophages [43, 44]. When there is not enough iron in the milieu, IRPs bind IRE motifs to contextually decrease the expression of ferritin and ferroportin and increase that of the transferrin receptor, favouring mRNA stability. Basically, this allows the cell to uptake more iron to efficiently use it before it bounds to the storage protein ferritin [43]. In the literature there are two main hypotheses on the mechanism leading to iron accumulation in brain parenchyma in course of MS. The first is linked with microglia and astrocyte iron accumulation in course of unknown steps linked with neurodegeneration [5, 9, 44]. The second is linked with a vascular condition, known as chronic cerebrospinal venous insufficiency (CCSVI) related to reduced brain perfusion [45]. It has been hypothesized that CCSVI might favourite erythrocytes diapedesis, and subsequent iron deposition [12, 13, 38, 46]. Even though this is an intriguing and interesting hypothesis and a genetic dependence of CCSVI has recently been described by our group [47], other authors do not directly link CCSVI with increased iron and MS [48–50].

Therefore, in spite of the lack of concordance between blood and brain iron levels, whatever the mechanism causing brain iron deposition is, the same group of proteins regulate iron influx, efflux and storage [42, 51]. We hence looked at the commonest SNPs in the main iron-protein genes.

The main finding of our study was an increased MS susceptibility risk, of more than 4-fold, associated with the FPN1 -8GG homozygous genotype. In addition, stratifying disease progression and severity by FPN1 genotypes, PI and MSSS gradually increased as the number of the G alleles increased, ascribing to the GG-genotype the highest value. This suggests that MS patients carrying the -8G-allele might be at increased risk for disease worsening. These results can really be considered novel and peculiar findings in the field of MS since, to date, FPN1 SNPs have been only associated with particular diseases, such as venous leg ulcers [16], or reinvestigated as genetic modifiers of HFE[22]. FPN1 expression is regulated at different levels: by the IRE sequence in the 5’-UTR that, interacting with the IRPs, finely tunes how many FPN1 molecules can be expressed [43]; and post-translationally by the hepatic hormone hepcidin [23]. The IRE region, results in increased/reduced FPN1 expression respectively under high/low cellular iron, leading to personalized iron export. Hepcidin interacts and blocks FPN1 in the presence of high iron levels. Generally, FPN1 mutations return a molecule that cannot reach the cell surface or block FPN1 internalization and degradation affecting both hepcidin interaction and iron export. The strong closeness of FPN1 -8CG to the crucial IRE region, prompted us to investigate its role in MS. The significant associations we found can be speculatively interpreted as a direct role on the IRE-IRP interactions, or as an indirect role of still unknown molecular defects in linkage with the SNP. In CNS cells, or in macrophages, these situations may potentially affect iron-balancing, similarly as described for the HFE C282Y [52]. Micro-deletions in the IRE region lead to expected increased in FPN1 levels despite low cellular iron levels, and to date no mutations specifically affecting IRE have been identified in the FPN1 gene [53].

Our second relevant finding was related to the HEPC gene. Homozygous -582GG cases had an increased MS susceptibility of about 2.5-fold among progressive patients and the risk was kept when progressive cases were compared to the RR course. In addition, EDSS progressively increased among the three different HEPC genotypes, with homozygotes about 1.5-higher than the rest of cases. Noteworthy, the rate of -582GG homozygotes was higher among progressive cases (13.5%) when compared to RR group (5.5%), who retained the same rate observed among healthy controls (5.8%). This could suggest that those patients might have rapid disease progression and/or higher chance for progression. Though confounding, due to the unavoidable presence of a great proportion of RR who will develop secondarily the progressive clinical course, this result could even be underestimated, because of the few homozygotes found among RR could even decrease after progression, and improve the statistical comparison. To verify the hypothesis, we split the group of SP cases (n=103) in those with/without the -582G-allele in order to calculate how long these two subgroups stayed in the previous less severe clinical phenotype before becoming progressive. Indeed, during a ten-year retrospective analysis, those carriers had a 3-fold higher chance to progress in the SP-phenotype if compared to the counterpart -582AA genotype. Similarly, including in the same survival analyses also the RR patients, those carriers had a 4-fold higher chance to progress. If this was true, the complementary analysis, that is computing together the mean duration time of the RR-patients (n=273) and that of the previous RR status of the 103 SP patients (total, n=376), could indirectly confirm this hypothesis by yielding opposite results (i.e. -582AA-carriers show a longer disease duration). That is exactly what we observed (GG, 4.52y±3.6 < AG, 6.2y±5.6 < AA, 7.8y±6.8); a possible explanation is that SP G-carriers could have faster left the RR condition to switch in the progressive form. Therefore, the RR G-carriers could have a potential shorter mean duration time within the less severe condition (GG, 4.21y±3.9 < AG, 7.45y±5.9 < AA, 9.12y±7.7). We recognize the intrinsic limit of these partial and indirect results, but all are in favour of an earlier progression-switch role ascribable to the HEPC polymorphism. Conflicting and scanty results exist on the -582AG HEPC variant [24, 25]. The G-allele decreases the transcriptional activity by 20% respect to the A-allele in HepG2 cells in the presence of upstream stimulatory factor 1 (USF1) and by 12-14% with USF2 [26]. The Authors concluded that the promoter variant is not associated with serum iron parameters and that the in-vitro studies resulted in little reductions of the G-allele mediated trans-activation. Although they ascribed to the HEPC variant negligible in-vivo effects, we state that, regardless the small change in the promoter activity between the two alleles, this could be enough to have significant detrimental effects on long-staying iron overload as is the case in MS patients. Accordingly, also subtle chronic lower HEPC expressions in subjects with -582G-allele may be responsible for significant local iron dysregulation mostly in homozygous GG-patients. We previously reported that even minor SNP effects (i.e. those found in MMP12 -82AG) had significant results in another degenerative disease under chronic iron-overload conditions [16].

As far as the HFE gene is concerned, H63D and C282Y did not reveal in our population associations with MS. One exception was the 3-fold higher PI found among the 63DD-homozygotes. However, also in other studies the role of the HFE gene in MS, seems not to be particularly decisive, being often controversial [17–20]. HLA-DR15 is associated with younger age of onset in MS [54], though we found an appreciable delayed onset among HFE63 DD-Homozygotes. This could be explained by speculating that iron greedy-cells (i.e. those with the polymorphism) could even be protective, paradoxically helping myelin synthesis in the early phases of the disease [55]. After iron moves on insoluble-hemosiderin, iron-starved cells cannot use it, this favours energy crisis and cell apoptosis [5, 9, 56]. Similarly, this could also be speculated for the HEPC variant, in which G-carriers show delayed onset.

Controversial results exist in the association between TF P570S and Alzheimer disease (AD) [30, 31], hypothesizing a not definitively demonstrated defect in total iron binding capacity [28, 29] and a suggestive synergism between TF and HFE gene variants and AD [32]. We did not find such a synergism in MS, except a non-significant higher MSSS among the TF 570S-carriers.

Gender appears to play critical role in development, progression and treatment of MS. In addition, higher brain iron level was found associated with male gender in presence of common iron gene SNPs [57, 58]. For this reason, we performed a gender-related sub-analysis, and we found different risk associations related to the different SNPs considered, but definite results cannot be drawn due to the low number of patients obtained after subanalyses. Clarifying a possible differential gender-associated risk to develop neurodegenerative diseases, combining genetic and MRI biomarkers, may help clinicians to design primary intervention programs to select high-risk sub-groups.

We conclude that, all the SNPs investigated work in the same direction: potential iron dysregulation, oxidative tissue damage, and possible actions on MS [51]. This was the reason we looked at the combined effect that the coexistence of several at risk-alleles might have on MS. The fact that among multi-carriers the risk increased, as well as disability, progression, and severity did, strongly implies the multi-gene nature of iron unbalancing in MS.

We recognize that the main limitation of our study is linked to the low number of investigated SNPs. A relevant number of SNPs exist in other candidate genes related to tissue inflammation and degeneration. A further shortcoming in the interpretation of our results is linked with the lack of knowledge still present in MS pathogenesis as well as in the steps leading to iron accumulation.

Whatever the mechanism causing brain iron deposition is, our study shows strong influence of gene variants in MS onset and disease course in terms of expectation of disability and severity. Although, in our survey the homozygous prevalence of the investigated SNPs is low, ranging from 3% to 8%, we have to take into account that more than 80% of our patients carry at least one of these variants, and that about 50% are double carrier. On the basis that, combined carriers can have phenotypic effects greater than or comparable to single homozygotes, and that iron homeostasis is multi-genetically tuned, this opens new clinical concrete perspectives in monitoring iron accumulation as an underlying mechanism connected to the natural history of MS together with the prognostic value of iron trafficking genes. People carrying at risk alleles could be selected in advance for therapeutic trials aimed to iron chelation and dietary modification in the view that MS course could be in part genetically targeted. So, further larger investigations on iron genes should become mandatory in MS. Understanding the exact mechanism by which iron acts in the brain causing MS and how the brain would be impacted by iron chelation/supplementation could potentially furnish precious prognostic information and novel insights for alternative personalized treatments (pharmacogenetics) aimed in preventing or counteracting neuron loss and degeneration.

All this is in line with a recently published review, on the importance of individualised therapy in MS, based on genetic and biochemical determinations [59].