Carriers of the Complex Allele HFE c.[187C>G;340+4T>C] Have Increased Risk of Iron Overload in São Miguel Island Population (Azores, Portugal)

Iron overload is associated with acquired and genetic conditions, the most common being hereditary hemochromatosis (HH) type-I, caused by HFE mutations. Here, we conducted a hospital-based case-control study of 41 patients from the São Miguel Island (Azores, Portugal), six belonging to a family with HH type-I pseudodominant inheritance, and 35 unrelated individuals fulfilling the biochemical criteria of iron overload compatible with HH type-I. For this purpose, we analyzed the most common HFE mutations– c.845G>A [p.Cys282Tyr], c.187C>G [p.His63Asp], and c.193A>T [p.Ser65Cys]. Results revealed that the family’s HH pseudodominant pattern is due to consanguineous marriage of HFE-c.845G>A carriers, and to marriage with a genetically unrelated spouse that is a -c.187G carrier. Regarding unrelated patients, six were homozygous for c.845A, and three were c.845A/c.187G compound heterozygous. We then performed sequencing of HFE exons 2, 4, 5 and their intron-flanking regions. No other mutations were observed, but we identified the -c.340+4C [IVS2+4C] splice variant in 26 (74.3%) patients. Functionally, the c.340+4C may generate alternative splicing by HFE exon 2 skipping and consequently, a protein missing the α1-domain essential for HFE/ transferrin receptor-1 interactions. Finally, we investigated HFE mutations configuration with iron overload by determining haplotypes and genotypic profiles. Results evidenced that carriers of HFE-c.187G allele also carry -c.340+4C, suggesting in-cis configuration. This data is corroborated by the association analysis where carriers of the complex allele HFE-c.[187C>G;340+4T>C] have an increased iron overload risk (RR = 2.08, 95% CI = 1.40−2.94, p<0.001). Therefore, homozygous for this complex allele are at risk of having iron overload because they will produce two altered proteins—the p.63Asp [c.187G], and the protein lacking 88 amino acids encoded by exon 2. In summary, we provide evidence that the complex allele HFE-c.[187C>G;340+4T>C] has a role, as genetic predisposition factor, on iron overload in the São Miguel population. Independent replication studies in other populations are needed to confirm this association.


Introduction
Iron overload in humans is associated with a variety of acquired and genetic conditions, the most common being the hereditary hemochromatosis type-I (HH, OMIM #235200), an autosomal recessive disorder caused by mutations in the HFE (High Iron Fe, OMIM Ã 613609) gene [1]. HFE encodes an HLA-A class 1-like protein and is located on 6p21.3, 4 megabases (Mb) telomeric to the human leukocyte antigen region (HLA).
Two HFE mutations-c.845G>A [p.Cys282Tyr] and c.187C>G [p.His63Asp]-were originally described in association with HH [2]. The majority (60% to 90%) of clinically diagnosed probands were homozygous for c.845A [p.282Tyr], and 5% were c.845A/c.187G compound heterozygous. In terms of molecular pathology, the c.845A is the most severe mutation. Its frequency decreases from the north to the south of Europe and is very low in non-European derived populations, such as Asians [3], Africans [4] and Ashkenazi Jewish [5]. The second originally described mutation-c.187G [p.63Asp]-is found as a highly frequent polymorphism in general populations [2]. Nonetheless, it was observed at an increased frequency in HH patients' chromosomes that do not carry the c.845A mutation [2,6], suggesting a possible role as a modifier of iron overload. A third sequence variant in the HFE gene − c.193A>T [p. Ser65Cys] − was increased in some HH patients' groups in comparison to healthy controls [7,8]. Despite the fact that individuals with the c.845A/c.193T genotype may have an increased risk to express a milder form of HH, the penetrance of this genotype is low and other genetic and environmental factors may influence the expression of iron overloading [9]. Other HFE single nucleotide polymorphisms and rare variants have been implicated in hemochromatosis [10,11], including intronic splicing mutations, such as c.340+4T>C [IVS2+4T>C] [12] and c.1008+1G>A [IVS5+1G>A] [13]. For example, Floreani and colleagues [11] showed that two cis-variantsc.193T and c.340+4C -generate alternative splicing by HFE exon 2 skipping. The corresponding protein misses the α1-domain, which is essential for the interaction of HFE with TfR1 (Transferrin Receptor-1) [14].
Despite recent advances, a better understanding of the molecular basis of iron overload is needed in order to improve patients' disease outcome through early diagnosis and treatment. Here, we conducted a hospital-based case-control study of individuals living in the Azorean island of São Miguel in order to genetically characterise a family with pseudodominant inheritance of hereditary hemochromatosis (HH) type-I, and an additional group of unrelated patients that fulfilled the biochemical criteria of iron overload compatible with HH type-I. To that end, we genotyped four HFE mutations/variants-c.845G>A [p.Cys282Tyr], c.187C>G [p.His63Asp], c.193A>T [p.Ser65Cys], and c.340+4T>C [IVS2+4T>C]-and performed an association analysis of HFE haplotypes and genotypic profiles with this condition. Since several HLA haplotypes have been associated with HFE-c.845A and -c.187G, we also studied the HLA-A and -B group alleles and haplotypes linked to these mutations in the São Miguel Island.

Ethics statement
The present investigation follows the international ethical guidelines and was approved by the Health Ethics Committee of the Hospital of Divino Espirito Santo of Ponta Delgada, EPE (HDES). The study design includes, from all participants, written informed consent, confidentiality, and an abandonment option in the case of expressed will.
The general population consists of 469 DNA samples of unrelated healthy blood donors from São Miguel Island selected from the anonymized Azorean DNA bank, located at HDES. This DNA bank was established after approval by the Health Ethics Committee and follows, as well, the international ethical guidelines for sample collection, processing, and storage.  4.6%) and Nordeste (NOR, 4937 inhabitants; 3.6%). Around half of the São Miguel population lives in small rural localities. The rural area is characterized by agriculture and cattle-breeding economy, and its inhabitants show great similarity in life style, as well as in eating habits. Regarding iron metabolism, the literature review did not reveal any previous study on iron overload in this population.
The general population cohort is composed of 469 DNA samples of unrelated healthy blood donors from São Miguel Island (Azores, Portugal). This sample was geographically representative of the six municipalities of the island: PDL (n = 176), VFC (n = 87), RG (n = 76), POV (n = 51), NOR (n = 41) and LAG (n = 38). Ninety-seven percent of the subjects studied have parents born in the same locality.

Iron overload patients
A total of 41 patients (Table 1) were clinically characterized by six physicians (Internal Medicine, Gastroenterology, Hematology, and Pneumology departments) at the HDES. Six patients belong to one family studied in the context of a family screening of hemochromatosis. They live in a small rural locality with less than 600 inhabitants. The remaining 35 unrelated patients were referred for HFE genotyping, since they fulfilled the biochemical criteria of iron overload compatible with the classical form of hereditary hemochromatosis (type-I)-1. serum ferritin > 400 ng/mL (males) or > 300 ng/mL (females), and/or 2. transferrin saturation (TS) > 50% (males) or > 45% (females), and/or 3. serum iron > 160 μg/dL (males) or > 145 μg/dL (females). Patients with evidence of secondary iron overload, namely exogenous iron intake, hepatitis B or C infection, and daily alcohol consumption higher than 60g, were excluded from the study. All patients and their parents were born in São Miguel Island, being the majority with both parents born in the same municipality (66%; Table 1). Blood samples were collected by venipuncture into dry and EDTA-K 3 tubes for biochemical and mutation analysis, respectively. Genomic DNA was extracted using the PUREGENE 1 DNA Purification (Gentra systems Inc.) or Citogene 1 DNA Purification (Citomed) kits. Serum transferrin, iron, and transaminases (aspartate transaminase, AST, and alanine transaminase, ALT) were measured   193A>T, and c.845G>A was carried out by two methods: i) polymerase chain reaction followed by specific restriction enzyme (PCR-RFLP) [15] for 35 patients and 469 general population individuals, or ii) real-time PCR using TaqMan 1 genotyping assays for 6 patients.
In RFLP analysis, PCR products from exons 2 (208 base pairs, bp, for c.187C>G and c.193A>T) and 4 (390 bp for c.845G>A) were digested for 2 hours at 37°C with RsaI for c.845G>A, MboI for c.187C>G and HinfI for c.193A>T (New England Biolabs). The digestion products were size resolved by electrophoresis on 4% agarose gel and visualized by SYBR 1 Green I nucleic acid gel stain (Molecular Probes). For the c.845G>A, the RsaI produced two fragments of 250 and 140 bp in the wild type DNA and three fragments of 250, 111 and 29 bp in the mutated DNA. In the case of wild type DNA for c.187C>G, the MboI generated two fragments of 138 and 70 bp, whereas for HinfI the two fragments have 147 and 61 bp; both c.187C>G and c.193A>T mutated DNA were not cut.
Considering patients' clinical manifestations and their HFE status (the majority of patients are not c.845A homozygous), we also performed direct sequencing of HFE exons 2, 4 and 5, including exon-intron boundaries. For this, we used the primers described above for RFLP analysis; however, for exon 5 the following primers were used: 5'-GATGAGAGCCAGGAGCTGAG-3' and 5'-CCCTGGGGCAGAGGTACT-3'. The exon 2 and its intron-flanking sequences analysis allows for the identification of the c.340+4T>C [IVS2+4T>C] splice site variation, which has been reported as associated with iron overload [11], whereas sequencing of exon 5 and its intron-flanking regions allows for the evaluation of the splice site mutation c.1008+1G>A [IVS5+1G>A] [13]. Each 20 μl PCR amplification reaction contained 100 ng of genomic DNA, 10 μM primers, 200 μM dNTPs (Promega), 25 nM MgCl 2 (Qiagen), 1x Q-Solution (Qiagen), 1x buffer (Qiagen), 5 U of HotStart Taq (Qiagen), and sterile water. The PCR started with an enzyme activation step at 95°C for 15 min, then 40 cycles at 94°C for 30s, 56.5°C for 30s and 72°C for 1 min, followed by a final extension step at 72°C for 10 min. Amplification products were purified using the QIAquick PCR Purification kit (Qiagen), according to the manufacturer's instructions. Purified products were sequenced, using the same primers of the PCR amplification, with the BigDye Terminator v1.1 cycle sequencing kit (Applied Biosystems) under the following conditions: 1 μl ready reaction mix, 5 μl BigDye sequencing buffer, 3.2 pmol of forward or reverse primer, 7 ng DNA, and sterile water to a final reaction volume of 20 μl. Cycle sequencing was performed using a initial denaturation step at 96°C for 1 min followed by 25 cycles at 96°C for 10s, 50°C for 5s, and 60°C for 4 min in a GeneAmp 1 PCR System 2700 (Applied Biosystems). The sequencing products were purified with a BigDye XTerminator 1 Purification kit, and separated by capillary electrophoresis in an automated sequencer (ABI 3130 Genetic Analizer, Applied Biosystems) with a 36 cm length capillary and POP-7 TM polymer, according to the manufacturer's instructions. Data were analyzed with the Sequencing Analysis software version 5.3.1 (Applied Biosystems). The alignment and edition of sequences were carried out using the Bioedit™ software version 7.0.0.
Sequencing results revealed the presence, in some patients, of the splice site variation c.340 +4T>C (rs2071303). Consequently, this variant was genotyped in the general population (469 individuals) by TaqMan 1 Pre-Designed SNP Genotyping Assays (Applied Biosystems) on an ABI 7500 Fast Real-Time PCR System, according to manufacturer's instructions.
HLA genotyping HLA-A and-B genotyping was performed in the 41 patients by PCR amplification with sequence-specific primers (PCR-SSP), according to the manufacturers' instructions (Olerup SSP™, GenoVision Inc.). After electrophoresis on a 4% agarose gel stained with SYBR 1 Green, the PCR products were visualized, followed by HLA group allele identification using the Helmberg-SCORE TM -Sequence Compilation and Rearrangment Evaluation, for research onlysoftware (Olerup SSP AB). This methodology only allows a low resolution genotyping. Consequently, HLA was characterized in a group level resolution, and the specific alleles present in each group were not discriminated.

Statistical analysis
Allele and genotype frequencies were estimated for the HFE-c.187C>G, -c.193A>T, -c.340 +4T>C, and -c.845G>A mutations in the general population (n = 469). Hardy-Weinberg equilibrium (HWE) for the general population was determined by the Arlequin software v.3.5.1.2. No departure from HWE was observed. Then, contingency tables were constructed to calculate the statistical differences between the allele frequencies in the six municipalities of the island. Fisher's exact test was used instead of a chi-square test, since the last may be inapplicable for very small expected frequencies (number of subjects in a cell should be 5 or more). Data analysis was carried out using the statistical package SPSS software, version 10.0 (SPSS, Inc.). The results were considered statistically significant when the p values were less than 0.05.
The HLA-A and-B group alleles' frequencies were calculated by direct counting. The HFE-c.845A-HLA-A-B haplotypes were directly obtained by segregation analysis through five consecutive generations in the family affected with pseudodominant inheritance of HH, and by-A and-B group alleles' homozygosity in two unrelated patients (patients 3 and 5). The same strategy was applied to determine the HFE-c.187G-HLA-A-B (patient 14). For a random subset (93 individuals) of the general population, the HLA haplotypes were estimated with the expectation-maximization (EM) algorithm provided in the Arlequin package, as previously described [17].
Association analysis of iron overload with HFE and HLA haplotypes, as well as with HFE genotypic profiles, was performed by calculating relative risks (RRs) with 95% confidence intervals (CI) using the 2-way Contingency

Study of a family affected with pseudodominant inheritance of hereditary hemochromatosis type-I
The study of hereditary hemochromatosis in São Miguel Island began in 1988 by the identification of two brothers-the proband (patient IV.7) and one sibling (patient IV.8)-with a classical clinical picture of HH and one paternal uncle affected with a minor form of the disease (III.3, Fig 1). In 2000, after the introduction of the HFE gene testing, this family, which is from a small village with less than 600 inhabitants, was reinvestigated in order to understand the HH segregation and to offer a better follow up of their relatives [18]. Their family pedigree, which spans five generations, is shown on Fig 1. The HH phenotype was identified in nine family members, from II to V consecutive generations: three patients (II.2, III.2 and III.7) by clinical and family history, and six patients (III.3, IV.7, IV.8, IV.12, V.10 and V.11) by clinical, biochemical, and HFE genotyping. A summary of the demographic, clinical, biochemical and genetic data of these patients is shown in Table 1. The proband (42y, IV.7) and one brother (40y, IV.8) presented the classical clinical picture of hereditary hemochromatosis type-I, including a transferrin saturation higher than 80% and a serum ferritin higher than 1000 ng/ mL. The paternal uncle (74y, III.3) had only a slightly elevated serum ferritin. They were all homozygous for the mutated allele HFE-c.845A [p.282Tyr]. Following the characterization of these first three patients, a biochemical and genetic screening was given to 11 additional family members. First, we investigated three sisters (IV.5, IV.10 and IV.12) of the two affected brothers and the daughter (IV.2) of the affected paternal uncle. The mutation analysis revealed that one clinically asymptomatic sister (37y, IV.12) was HFE-c.845A homozygous, being the other three heterozygotes for this mutation (c.845GA). We performed a clinical evaluation and genetic testing to the other five family members living in the São Miguel Island, namely the proband spouse (IV.6), three husbands of proband sisters (IV.4, IV.9 and IV.11), and the husband of one first-cousin (IV.3). The HFE analysis showed that the spouse of the proband (IV.6), which is non-related, is heterozygous for the c.187G [p.63Asp] mutation. Therefore, the molecular screening was also offered to their three children, aged 20 (V.10, female), 18 (V.11, male) and 15 (V.12, male) years-old, since they were suspected to be HFE-c.845A/c.187G compound heterozygous. This hypothesis was confirmed in the daughter (V.10) and in one son (V.11). Both were clinically asymptomatic, but presented high transferrin saturation levels.
HFE mutation analysis of unrelated patients with biochemical evidence of iron overload HFE genotyping for the three most common mutations associated with iron overload was performed in 35 unrelated patients from the São Miguel Island (Azores) who were referred by their physicians due to suspicion of the classical form of hereditary hemochromatosis (type-I), based on biochemical criteria of iron overload (see M&M). A summary of the demographic, clinical, biochemical, and genetic data of these patients is also presented in Table 1. Six of the 35 patients (17%) were homozygous for HFE-c.845A [p.282Tyr], and three were c.845A/ c.187G compound heterozygous. In these nine patients, the predominant clinical manifestations were skin pigmentation and arthropathy; three patients presented liver cirrhosis, one in association with classical hemochromatosis and bronze diabetes (patient 5 in Table 1). Although the highest TS values were found in patients homozygous for c.845A mutation, there was not, in general, a clear-cut correlation between patients' genotype and biochemical data.

Association analysis of HFE haplotypes and genotypic profiles with iron overload
In order to assess the association of HFE haplotypes (H) and genotypic profiles (GP) with iron overload we calculated relative risk (RR) by comparing the 41 patients against the general population. General population's HLA-A-B haplotypes were determined, by indirect inference, without HFE-c.845A and -c.187G carrier information. In this subset, a total of 84 haplotypes were identified, being A Ã 01-B Ã 08 (0.086), A Ã 02-B Ã 44 (0.066), and A Ã 24-B Ã 08 (0.043) the most frequent (S1 Table). The patients' direct haplotype inference revealed that, with the exception of A Ã 03-B Ã 50 (H 6 ) and A Ã 03-B Ã 49 (H 7 ), all other patients' haplotypes were present in the general population (Table 2).
To investigate the association of HLA group alleles and haplotypes with iron overload, we calculated relative risk by comparing the 41 patients against the general population. The data demonstrated a significant positive association of B Ã 35 (RR = 2.65, 95% CI = 1.20−5.87,

The HFE mutations in São Miguel population and its geographical distribution
Since the São Miguel population lives on an island, we studied the prevalence of the four HFE mutations.  c.193T mutation was found in 19 subjects: two (0.004) were compound heterozygous for c.845A, and five (0.011) were compound heterozygous for c.187G (Table 4). In order to assess the relationship between the four HFE mutations (c.845A, c.187G, c.193T, and c.340+4C) and their geographical distribution in the São Miguel Island, we compared allele frequencies between the six municipalities (Fig 2). For c.845A, the highest value is found in Nordeste (9.8%) followed by Povoação (5.9%), and the lowest in Lagoa (2.6%). We observed a significant difference (p = 0.048) between Nordeste and the other five municipalities (PDL, RG, LAG, VFC, and POV), indicating a relatively non-uniform island distribution for the c.845A mutation. On the other hand, the other three variants-c.187G, c.193T, and c.340+4C -showed a uniform pattern with no significant differences among municipalities. Considering that the São Miguel population has an admixed genetic background composed mainly of Europeans and less by Jews and Africans [19][20][21][22][23][24][25], we compared the HFE mutations frequencies with other populations. The analysis revealed that islanders have a c.845A [p.282Tyr] frequency similar to that found in several countries in northern Europe (%5.7%) [26], but significantly different from the reported frequencies for the Azorean island of Terceira (2.1%) [27]. Regarding c.187G [p.63Asp], the highest frequency observed in Portugal has been detected in the Madeira archipelago (20.5%), followed by São Miguel Island (20.4%), north and center mainland (19.7%) [28] and Terceira Island (18.3%) [27]. The frequencies detected for these two HFE mutations (c.845A and c.187G) in São Miguel are significantly higher than those found in Jewish (1.3 and 9.7%) [5] and African (0.9% and 13.2%) [4] populations. As expected, the c.340+4C variation presented similar allelic frequency with phase III CEU Hap-Map population (34.1%).

Discussion
Iron overload disorders represent a heterogeneous group of conditions resulting from inherited and acquired means. Despite recent advances, a better understanding of the molecular basis of iron overload is needed in order to improve patient's disease outcome through early diagnosis and treatment. Here, we carried out, for the first time, a clinical evaluation of 41 iron overload patients from the Azorean island of São Miguel (Portugal), which were referred for HFE genotyping. Of these, six patients belong to a family with a pseudodominant inheritance of HH. In the present research, the patients' clinical evaluation demonstrate that around 43.9% (18 out 41) present serum ferritin higher than 1000 ng/mL, and 29.3% (12 out 41) show HH's clinical manifestations, the most frequent being hyperpigmentation and hepatic cirrhosis. Moreover, there is a higher frequency of HFE-c.845A homozygous individuals in the patient cohort (24.4%) compared to the general population (0.2%). The majority of patients presenting an iron overload did not fulfill the criteria of HH, suggesting that additional environmental or genetic factors could contribute to iron overload. Sequencing of HFE exons 2, 4 and 5, and intron-flanking regions did not reveal any other mutation, but allowed the identification of c.340+4C [IVS2+4C] splice site variant in 63.4% patients (26 out of 41, Table 1). This variant alone does not explain iron overload (RR = 0.28, 95% CI 0.10-0.66, p = 0.001; Table 2 HFE-H 4 ); however, around 7% (3 out 41) of patients (#32-34, Table 1) were heterozygous for c.340 +4T>C and did not show any other HFE mutation. Bioinformatic analysis using the Human Splicing Finder 2.4 (http://www.umd.be/HSF/) [29] revealed that c.340+4T>C has a splicing Δ consensus of +9.71%, a value higher than the expected 7% for a +4 position of 5' splice site. In fact, this significant impact on splicing is consistent with the biological evidence of exon 2 skipping observed in a patient with histologically-demonstrated iron overload [11]. According to functional studies performed by Martins et al. [14], the protein produced by the c.340+4T>C alternative splicing is retained in the endoplasmic reticulum and do not efficiently reach the plasma membrane with the β 2 -microglobulin chaperone. A similar situation is observed with the p.282Tyr mutated HFE protein.
Although the pattern of inheritance of HH is usually horizontal, i.e. all patients belong to the same generation, as expected for an autosomal recessive disease, here we describe a vertical (pseudodominant) pattern due to the segregation within the studied family of at least three HFE mutant alleles in each generation. Common causes for a pseudodominant inheritance pattern are: i) birth of an affected child from an affected individual and a genetically related (consanguineous) reproductive partner, who is an unsuspecting carrier, and ii) high carrier frequency, enhancing the risk that the spouse of a patient is a carrier of a mutation in the same gene [30]. Interestingly, these two conditions are observed in their family pedigree: the proband's (IV.7) parents are consanguineous and responsible for the transmission of the c.845A allele and the proband's spouse (IV.6), who is genetically unrelated, carries the second most frequent mutation-c.187G.
The knowledge of mutation origin improves the comprehension of population genetic background and evolution. The analysis of co-segregation of HFE mutations and HLA-A and-B with HH in the family pedigree revealed three non-ancestral HLA-A-B haplotypes associated with the HFE-c.845A mutation: A Ã 01-B Ã 35, A Ã 02-B Ã 44 and A Ã 02-B Ã 55. The first two were also observed in HFE-c.845A homozygous HH patients from the north of Portugal [31]. However, the third haplotype (A Ã 02-B Ã 55) and the A Ã 24-B Ã 15 are, to our knowledge, two new nonancestral haplotypes associated with this mutation. These two haplotypes reinforce the association of HFE-c.845A mutation with A Ã 02 (linked to, for example, B Ã 07, B Ã 14 and B Ã 35) or A Ã 24 (linked to B Ã 18, B Ã 35 and B Ã 57) group alleles, both observed in northern Portuguese HFEc.845A homozygous patients [31]. Regarding HFE-c.187G, a previous study [32] reported a significant association of this mutation with A Ã 29 or B Ã 44 group alleles, as well as with A Ã 29-B Ã 44 haplotype in the mainland Portuguese population. The obtained results only show association of the HFE-c.187G mutation alleles with A Ã 03-B Ã 49 or A Ã 32-B Ã 49. Nevertheless, patient 9, who is c.845A/c.187G compound heterozygous, also presents the A Ã 29-B Ã 44 haplotype.
In order to investigate the geographical distribution of the HFE-c.187C>G, -c.193A>T, -c.340+4T>C, and -c.845G>A mutations in the São Miguel Island, we compared the allele frequencies between the six municipalities (Fig 2). The data demonstrate a higher frequency of c.845A mutation in Nordeste compared to the other five municipalities (p = 0.048), suggesting a geographic cline from east to west. This observation may lead to theorize the presence of a founder effect; however, the high diversity of HLA haplotypes associated with this mutation does not corroborate this hypothesis. Furthermore, this trend was not observed for the other three HFE variants-c.187G, c.193T and c.340+4C -, which show no significant differences among the municipalities. Additionally, we cannot rule out the hypothesis that this result may be due to the small sample size. Nevertheless, the observed pattern validates the importance of carrying out screening studies of recessive mutations in relatively small populations like the Azorean island of São Miguel.
In general, the frequency of c.845A [p.282Tyr] in Europe shows a decreasing north-south cline, with values ranging from 5 to 10% in north Europe, and from 1 to 5% in central and south Europe [33]. The frequency of the c.845A mutation in São Miguel was similar to the one reported in north/central mainland Portugal [28], but significantly higher than in the Azores Terceira Island [27], south mainland Portugal [28], and Madeira Island [34]. This data validates previous results where mainland Portuguese, especially from north/center, were the main contributors to the Azorean settlement. Moreover, these results may be suggestive of different population dynamics between the Azorean islands. Comparison with other populations shows that c.845A frequency is similar to central European countries, reflecting our previous results, where Flemish, French, and Germans also contributed to the settlement of the Azores [21][22][23][24].
Concerning the c.187G [p.63Asp] mutation, the highest frequency observed in Portugal has been detected in the Madeira archipelago followed by São Miguel Island. This high frequency is similar to that found in southern Europe. Although the relationship between this mutation and HH is unclear, it constitutes a genetic predisposing factor causing iron overload when present with another genetic (HFE or other gene mutation) or an environmental factor [35]. The results indicated that alone c.187G is not associated with iron overload, but together with c.340+4C [IVS2+4C] splice variant is responsible for an increase in the risk of developing the disease. This result is corroborated by simulation studies, where adding just one patient homozygous for the mutated allele of both variants (GP 4 GG AA CC GG , Table 3) had a significantly increase in risk (RR = 3.10, 95% CI 1.03-8.24, p = 0.016). We hypothesize that double homozygous individuals will produce two altered HFE proteins-the p.63Asp [c.187G] mutated protein, and the protein lacking 88 amino acids encoded by exon 2. Overall, these data point to the complex allele HFE c. [187C>G;340+4T>C] being an iron overload genetic predisposition factor.
The c.193T [p.65Cys] mutation, also considered a polymorphism, may, alone or combined with other mutations, be associated with mild iron accumulation [36]. In the present study, the small number of patients carrying the c.193T polymorphism makes it difficult to establish conclusions about the relation of iron overload with this variant. However, genotypic profile analysis revealed that carriers of GP 3 CG AT CC GG present a significant risk (RR = 11.44, 95% CI 1.17-112.26, p = 0.002; Table 3). In terms of gene expression, these individuals will most likely produce three types of HFE altered proteins: the p.63Asp [c.187G] protein, the p.65Cys [c.193T] protein, and the protein missing exon 2 encoded amino acids. In summary, we provide evidence that at least the complex allele HFE-c.[187C>G;340 +4T>C] has a role, as a genetic predisposition factor, on iron overload in the São Miguel population (Azores, Portugal). Independent replication studies in other populations are needed to confirm this association. Additionally, the HFE-c.845A [p.282Tyr] mutation has a diverse HLA genetic background, since it is associated with six haplotypes, three of them described here for the first time-A Ã 02-B Ã 55, A Ã 24-B Ã 15 and A Ã 03-B Ã 50. These data validate the importance of carrying out epidemiological studies of recessive mutations in relatively small populations like São Miguel Island and are a valuable contribution to a perspective study on iron overload in this population.

Limitations
The present work has some limitations, the major one being the small number of patients; however, it includes practically all island's diagnosed cases of iron overload that meet the inclusion criteria and is the first report on HFE mutation distribution in the healthy population of São Miguel Island. Nonetheless, replication with a larger sample size is needed in order to validate the results observed here. Another constraint is the lack of functional analysis of the complex allele HFE c.[187C>G;340+4T>C].
Supporting Information S1 Table. HLA-A-B haplotypes observed in the São Miguel Island general population. (DOCX)