Associations of iron metabolism genes with blood manganese levels: a population-based study with validation data from animal models

Manganese (Mn) is an essential nutrient but also a known neurotoxicant at high chronic exposure levels. In particular, occupational studies have documented adverse effects on motor, psychological, and neurological function following long term exposures in adults, with progressively worse symptoms persisting long after cessation of exposure [1–6]. While most epidemiologic literature on manganese toxicity has focused on occupational exposure, there is emerging evidence that elevated environmental levels of manganese may cause cognitive and motor deficits, as well as behavioral problems, among both adults and children [7–15]. High maternal blood manganese levels have also been associated with lower birth weights, although this effect appears nonlinear [16].

Given the emerging evidence for adverse effects from excess manganese exposure, it is critical to understand host factors that affect manganese metabolism. Such factors may provide mechanistic clues for future interventions or may identify susceptible subpopulations. For example, there is a well-known inverse association between iron stores and manganese absorption. In laboratory animals, iron deficiency results in increased absorption and retention of manganese in various organs [17–21]. Iron overload, in contrast, reduces manganese accumulation in the brain and in other organs [20]. In humans, similar inverse associations of blood manganese with ferritin and dietary intake of nonheme iron have been reported [22–24]. It is therefore likely that iron and manganese share the same transport and regulatory proteins, and at least one factor, Divalent Metal Transporter-1 (DMT1), is known to mediate uptake of both cations [25].

Manganese and iron are both transition metals and, given their similar chemical properties, likely share genetic factors that regulate their metabolism. Such genetic factors may alter manganese absorption, excretion, or tissue/cellular distribution, thereby altering blood manganese levels. Variants of iron metabolism genes with known functional properties, such as the HFE gene, may be good candidates to regulate manganese metabolism. The HFE gene product regulates levels of the regulatory hormone hepcidin, thereby controlling iron absorption in response to body iron status. Two well-characterized, functional missense variants in the HFE gene, C282Y and H63D, have been associated with increased dietary iron uptake [26, 27]. The C282Y variant is the primary cause of the common form of hereditary hemochromatosis, a genetic disease of excess iron absorption. Transferrin is an iron transport protein that delivers iron from the gastrointestinal tract to tissues. A common missense variant in the TF (transferrin) gene, P570S (or TFC2), has been associated with differential binding of iron to transferrin relative to the wildtype protein [28]. Finally, ALAD (δ-aminolevulinic acid dehydratase) is part of the heme synthesis pathway [29]. Heme is a necessary component of hemoglobin and all cytochrome enzymes, and requires iron for enzyme activity. This iron metabolism gene has been linked to metal toxicity [30–32] as well as metal binding [33]. The K59N (or ALAD2) variant of this gene has also been associated with increased hemoglobin [34]. We hypothesized that, because each of these iron metabolism genes has common functional variants and because iron metabolism is linked to manganese metabolism, these variants may be associated with altered blood manganese levels. We also explored whether any observed effects of genotype are mediated through iron status.

We examined these questions using blood manganese levels in a cohort of women at time of delivery. Pregnancy is a time at which blood manganese levels increase [35, 36] as well as a time of relative iron deficiency. We believed the increase in variability of both iron stores and blood manganese would increase our ability to detect differences in blood manganese related to genetic variants in iron metabolism genes. This time point is also relevant because maternal blood manganese is a measure of prenatal exposure to the fetus, and there is evidence that environmental manganese exposure in utero is associated with later life behavioral disinhibition [9] and lower psychomotor development test scores among offspring [12]. Cost constraints prevented a large scale or full interrogation of the iron metabolism gene pathway. Instead, we focused specifically on three candidate iron metabolism genes, HFE, TF, and ALAD, which have either known or putatively functional variants. In order to validate that our findings were not due to chance or linkage disequilibrium (LD) with a nearby functional variant, we used a knockout mouse model to determine if we could replicate our results in an experimental setting. This is a novel method of validating observed associations involving genetics and observational epidemiologic data. It is not subject to population substructure and is more cost effective than replicating findings in an independent population. To our knowledge, this is also the first study to examine blood manganese levels in HFE-deficient mice.

Blood manganese levels in our study are similar to levels observed among women around the time of delivery in other studies. Most similar are concentrations among a cohort residing in Quebec, Canada (GM = 16 μg/L; 5^th, 95^th percentiles: 10 μg/L, 26 μg/L)[45]. At 34-weeks gestation, the mean blood manganese level among a group of Australian women was 13 μg/L (SD = 4 μg/L) (i.e., mean (SD) = 230 (68) nmol/L)[35]. Higher manganese concentrations were reported in a mother-infant cohort living near a Superfund site in northeast Oklahoma (median = 22 μg/L; 5^th, 95^th percentiles: 13 μg/L, 41 μg/L)[16], and among groups of women in Montreal, Canada (mean = 23 μg/L; 5^th, 95^th percentiles: 6 μg/L, 52 μg/L)[46], and Paris, France (mean = 23 μg/L; 5^th, 95^th percentiles: 12 μg/L, 40 μg/L [46]; and GM = 20 μg/L; 5th, 95th percentiles: 11 μg/L, 40 μg/L [12]). The differences in blood manganese levels between studies may be due to differences in analysis method (e.g., Takser et al. 2003 and 2004, Smargiassi et al. 2002, and Spencer 1999 used atomic absorption spectrometry, while Zota et al. 2009 used ICP-MS), but may also be due to differences in environmental exposure levels or to differences in a factor related to manganese absorption/metabolism (e.g., iron status).

We found that carriers of the HFE variant genotype had 12% lower blood manganese levels than wildtype subjects. TF and ALAD genotypes were not associated with manganese levels in our data. In our experimental animal study, our findings of reduced blood manganese among Hfe ^-/- mice compared to "wildtype" control mice recapitulated the findings of our human population study and affirm the relationship observed between deficiency of HFE function and reduced blood manganese. The results from the animal study support the hypothesis that HFE genotype is the cause of the lower blood manganese levels in the epidemiologic study. This step is critical as observational epidemiologic studies may be confounded by ethnicity (population stratification) or by linkage disequilibrium with a nearby functional variant in a different gene (i.e. the measured genotype is not truly causative). The animal study cannot be subject to population stratification. While a knockout model cannot confirm that the results are not due to a variant within the HFE gene in LD with the H63D/C282Y variants, it does confirm that LD with a variant outside the HFE gene is not driving the results.

Few studies have examined associations between iron metabolism genes and biomarkers of manganese, and results are inconsistent. Contrary to our findings, Nichols and Bacon reported that hereditary hemochromatosis patients (i.e., patients with clinical disease) accumulate manganese [47]. However, given that this study was conducted in patients with known hemochromatosis, it is not analogous to our report. In a pilot study of 141 Ohio residents, Haynes et al. observed no significant associations of HFE and TF with hair or blood manganese [48]. The relationship between ambient air manganese and hair manganese, however, was significant after adjusting for HFE and TF genotypes, suggesting that manganese absorption varies by genotype. In a separate analysis of the same cohort as the present study, we observed lower hair manganese levels among one-month-old infants whose mothers carried the HFE variant genotype (mean (SD) = 1.8 (1.4) μg/g) compared to wildtype mothers (mean (SD) = 3.2 (4.4) μg/g) (natural log-transformed hair manganese: β = -0.5, 95% CI: -0.9 to -0.1), which is again consistent with results presented here for maternal blood manganese.

As in previous animal and human studies [17, 20, 23], ferritin was inversely associated with blood manganese, although an association between iron metabolism genes and hemoglobin or ferritin was not observed. One explanation for the lack of an observed association might be the relative insensitivity of hemoglobin and ferritin to iron status. In the clinical setting, measuring body iron status accurately is notoriously difficult, and the limitations of serum ferritin and whole blood hemoglobin are well known. Bone marrow iron concentration is considered the gold standard, but is seldom used due to the invasive nature of this measure. In HFE-associated hemochromatosis patients, increased transferrin saturation typically precedes elevated ferritin levels and may therefore provide a more accurate representation of iron stores in these individuals than ferritin [49]. If the mechanism by which HFE genotype is associated with reduced blood manganese does reflect a change in iron status, transport mechanisms affecting both metals could be involved and reduced manganese levels may reflect altered uptake and distribution due to excess iron stores competing for manganese binding and transport. Loss of HFE function is associated with abnormally low levels of the iron regulatory hormone hepcidin, which result in up-regulation of the basolateral iron exporter ferroportin and net dietary absorption of iron [50]. It remains uncertain whether ferroportin is involved in absorption of dietary manganese across the intestine. While increased ferroportin levels may enhance transfer of iron from the enterocyte to circulation, manganese export out of the enterocyte may be blocked. Alternatively, increased iron loading upon loss of HFE function may affect distribution of manganese between the blood and soft tissue compartments and/or may facilitate excretion of manganese via biliary secretion.

It is also conceivable that the HFE variant genotype might alter blood manganese levels independent of iron status. In Figure 1, a directed acyclic graph depicts two possible mechanisms by which iron metabolism genes may affect blood manganese concentrations: either via iron levels or independent of iron levels. The lack of association between genotype and iron variables in our human data suggests that HFE could impact manganese levels in a pathway independent of iron levels (i.e., via dashed arrow in Figure 1). Finally, we cannot exclude the possibility that our findings are due to another polymorphism in the HFE gene or in a proximal gene with which HFE variants are in LD. However, we note that H63D and C282Y are not in high LD and no single variant would be expected to be in LD with both HFE variants. The animal study results would mitigate the possible role of a variant with high LD outside the HFE gene driving the results. The physiological role of HFE in manganese metabolism can be explored in future animal experiments, and more closely examined in patient studies of individuals with HFE-associated hemochromatosis and other iron overload disorders.

Adverse health effects from manganese exposure are variable, and have been described as a "continuum of dysfunction" [11]. Some of this variability may be due to differences in how individuals metabolize manganese. A given dose may be preferentially absorbed or excreted conferring different levels of susceptibility to an individual. Iron status is one factor that may affect manganese absorption. Likewise, genetic factors may influence susceptibility to manganese health effects. Because manganese is also a nutrient, both low and high levels of exposure may produce toxicity, further highlighting the importance of genes that regulate manganese metabolism. Our study suggests that genetics, possibly independent of iron stores, may contribute to variability in manganese exposure and perhaps also to manganese effects by affecting manganese absorption, distribution, or excretion.

In summary, we found that carriers of HFE variant alleles have lower blood manganese levels, which may be beneficial among individuals exposed to high levels of environmental manganese, or could have negative implications if dietary manganese intake is insufficiently low. Data from our complementary animal experiments strongly support the conclusion that loss of HFE function is associated with reduced blood manganese. This is among the first studies to observe an effect of genetics on human manganese biomarkers and the first, to our knowledge, that affirmed results in an animal model. If our findings are confirmed in other populations, genotype will be critical to consider in studies that rely on environmental manganese measurements.