Naturally variant autosomal and sex-linked loci determine the severity of iron overload in β₂-microglobulin-deficient mice

Abstract

Hereditary hemochromatosis (HH) is a common chronic human genetic disorder whose hallmark is systemic iron overload. Homozygosity for a mutation in the MHC class I heavy chain paralogue gene HFE has been found to be a primary cause of HH. However, many individuals homozygous for the defective allele of HFE do not develop iron overload, raising the possibility that genetic variation in modifier loci contributes to the HH phenotype. Mice deficient in the product of the β₂-microglobulin (β₂M) class I light chain fail to express HFE and other MHC class I family proteins, and they have been found to manifest many characteristics of the HH phenotype. To determine whether natural genetic variation plays a role in controlling iron overload, we performed classical genetic analysis of the iron-loading phenotype in β₂M-deficient mice in the context of different genetic backgrounds. Strain background was found to be a major determinant in iron loading. Sex played a role that was less than that of strain background but still significant. Resistance and susceptibility to iron overload segregated as complex genetic traits in F₁ and back-cross progeny. These results suggest the existence of naturally variant autosomal and Y chromosome-linked modifier loci that, in the context of mice genetically predisposed by virtue of a β₂M deficiency, can profoundly influence the severity of iron loading. These results thus provide a genetic explanation for some of the variability of the HH phenotype.

Hereditary hemochromatosis (HH) is a common recessively inherited human genetic disorder caused by a chronic imbalance in iron homeostasis that leads to systemic tissue damage, fibrosis, and, ultimately, organ failure (1). A conspicuous disease hallmark is the accumulation of storage (nonheme) iron in the liver parenchymal cells, which is a consequence of an inability to limit the intestinal uptake of dietary iron. A mutation in the HLA-linked class I gene HFE is a primary risk factor for HH. Depending on the populations analyzed, 75–90% of affected individuals with clinical HH are homozygous for a position Cys-282 → Tyr mutation (C282Y) (for review, see ref. 2). This mutation prevents the formation of a critical disulfide bond in the HFE α3 domain, impairing its ability to form its obligate heterodimer with the β₂-microglobulin (β₂M) light chain and thus its delivery to the plasma membrane (3). However, clinical expression of the disease is highly variable, and many individuals homozygous for the prevalent C282Y allele fail to develop iron overload (2, 4, 5). At present, there is little coherent information that explains the incomplete penetrance of this disease. Although environmental factors, including dietary iron and blood loss, are thought to play an important role (for reviews, see refs. 2 and 5), studies are also consistent with an additional genetic component to the HH phenotype (6–10). Understanding these genetic components could provide important diagnostic and clinical information along with clues to the genes that are crucial for iron homeostasis.

The mouse model has been pivotal in elucidating the genetics and pathophysiology of many human genetically determined diseases. Mice deficient in β₂M and mice deficient in Hfe or engineered to express the disease-associated C282Y HFE mutation demonstrate systemic iron overload with iron deposition in liver parenchymal cells (11–16). The B2m null (−) mutation has been highly back-crossed onto an array of inbred mouse strain backgrounds (17–19). These strains, analyzed in a standardized laboratory environment, provide an approach for investigating the extent to which modifier genes play a role in normal and pathological iron regulation in the β₂M-deficient mouse model.

To determine whether genetic variation plays a role in controlling iron overload, we performed classical genetic analysis of the iron-loading phenotype by using β₂M-deficient mice with different genetic backgrounds. We show that strain background plays a major role in determining the severity of iron overload and that sex plays a role that is less but still significant. Susceptibility to iron overload segregates as a complex genetic trait in F₁ and back-cross progeny. Our results suggest that naturally variant autosomal loci and sex-linked loci play a substantial role in determining the severity of iron overload in β₂M-deficient mice.

Materials and Methods

Mice.

All mouse stocks used in these studies carried a null (−) allele of B2m (nomenclature designation, B2m^tm1Unc) produced by gene targeting (20). These mouse stocks were bred from B2m^tm1Unc founder mice provided by B. Koller and O. Smithies (University of North Carolina). Mouse strains congenic for B2m−/− were generated after a minimum of 10 back-cross generations onto their respective strain backgrounds, as described (17–19). The resulting AKR/J-B2m^tm1Unc/Dcr, C3H/HeJ-B2m^tm1Unc/Dcr, C57BL/6J (B6)-B2m^tm1Unc/Dcr, and NOD/LtJ-B2m^tm1Unc/Dcr mice were maintained as homozygous B2m−/− mice under specific pathogen-free conditions at The Jackson Laboratory's research colony. All mice were provided with an unlimited supply of standard NIH31 chow (Purina 5K52, 345 parts iron per million) and acid-treated water. Note that unlike type I diabetes-susceptible NOD-B2m^wild-type mice, NOD/LtJ-B2m^tm1Unc mice are diabetes resistant (21).

Nonheme Iron Quantitation.

Nonheme iron quantitation of liver tissue was performed by a modification of the bathophenanthroline assay described by Torrance and Bothwell (22). Either 1-cm² sections or complete resected livers were dried overnight in a 65°C oven and then weighed. The dried liver was then placed in 10 ml of acid digestion mixture (3 M HCl/10% trichloroacetic acid) and heated in a 65°C oven overnight. After cooling the digested liver mixture to room temperature, 5 μl was placed in 995 μl of working BPS chromagen [1 vol of stock BPS chromagen (0.1% bathophenanthroline sulfonate/1% thioglycolic acid) and 10 vol of 2.5 M sodium acetate] and vortex mixed. Working BPS chromagen reagent was prepared fresh daily. For a standard curve, serial dilutions of a ferric iron standard (Sigma Diagnostics; 500 μg/dl) were used. Color was allowed to develop for 15 min and then measured as absorbance at 535 nm on a Beckman 560 spectrophotometer.

Statistics.

Student's t test, one-way ANOVA with Newman–Keuls posttest analysis, and two-way ANOVA with Bonferroni posttest analysis were performed as indicated (GRAPHPAD PRISM). P < 0.05 was considered statistically significant.

Histology.

For histological assessment of nonheme iron deposition, tissue sections were stained with Perls' Prussian blue, counterstained with nuclear fast red, and examined by light microscopy.

Results

To determine whether strain background influences the amount of iron overload, we measured hepatic nonheme iron concentrations of C3H-B2m−/−, B6-B2m−/−, AKR-B2m−/−, and NOD-B2m−/− mice at 3 months of age that had been fed a standard basal iron diet. Iron concentrations of livers from C3H-B2m−/− and B6-B2m−/− mice were considerably lower than iron concentrations of livers from AKR-B2m−/− and NOD-B2m−/− mice (Fig. 1). These results suggested that the severity of liver iron loading in β₂M-deficient mice is strongly influenced by strain background.

Figure 1.

Influence of strain background on hepatic nonheme iron concentration in β₂M-deficient mice. Iron concentration was determined from ≈1-cm² sections of liver from B2m−/− mice with the following backgrounds: C3H, four female and five male; B6, two female and eight male; AKR, five female and five male; NOD, five female and five male. Data are the mean ± SEM. C3H vs. B6, P < 0.05; C3H vs. AKR, P < 0.001; C3H vs. NOD, P < 0.001; B6 vs. AKR, P < 0.001; AKR vs. NOD, not significant.

To determine whether strain differences are apparent histologically, Perls' Prussian blue-stained liver sections from C3H-B2m−/−, B6-B2m−/−, AKR-B2m−/−, and NOD-B2m−/− mice of various ages were examined (Fig. 2). Evidence of iron staining of parenchymal cells, both diffuse and punctate, that is consistent with developing HH was present by 12 weeks of age in all β₂M-deficient strains analyzed (Fig. 2 A, D, G, and I), but iron staining was most evident in livers of mice of the AKR-B2m−/− and NOD-B2m−/− strains (Fig. 2 G, H, and I). C3H-B2m−/− and B6-B2m−/− mice, in contrast, showed much less iron deposition regardless of their age (Fig. 2 A–F). These results suggested that strain-specific differences in liver iron deposition persist regardless of the age of β₂M-deficient mice. For comparative purposes, β₂M-wild-type C3H, B6, AKR, and NOD mice that were fed a normal laboratory diet failed to accumulate liver iron, in agreement with earlier studies (23, 24).

Figure 2.

Influence of strain background on hepatic iron accumulation in β₂M-deficient mice determined by histopathological analysis. Photomicrographs of liver sections stained with Perls' Prussian blue from representative B2m−/− mice of the indicated ages and backgrounds. Photomicrographs of liver sections from B2m+/+ mice showed no stainable iron (data not shown).

Human males tend to have a higher incidence and severity of HH than females (25). To determine whether sex influences iron loading in β₂M-deficient mice, we compared groups of age-matched female and male C3H-B2m−/−, NOD-B2m−/−, and an F₁ cross of (C3H-B2m−/− × NOD-B2m−/−) mice. We observed in serial histological sections that iron deposition was not distributed equally throughout the liver (data not shown). Consequently, to guarantee the accuracy of iron quantitation, hereafter, we analyzed the whole liver rather than a portion of the liver as in the initial studies shown in Fig. 1. At 3 months of age, female mice had higher levels of hepatic iron loading than strain-matched males (Fig. 3A). A 52-week longitudinal study of NOD-B2m−/− mice confirmed this sexual dimorphism and showed that the amount of hepatic iron loading in female mice was higher than in male mice, at least to 52 weeks of age (Fig. 3B). Thus, in contrast to humans with HH, female β₂M-deficient mice are more susceptible to iron overload than are genotypically matched male β₂M-deficient mice.

Figure 3.

Sex differences in hepatic nonheme iron concentration in β₂M-deficient mice. Iron concentration was determined in whole livers of B2m−/− mice. Data are the mean ± SEM, and the number of animals is shown for each point. (A) Three-month-old female and male C3H, NOD, and (C3H × NOD)F₁ mice. P < 0.01 by two-way ANOVA with Bonferroni posttest analysis: C3H female vs. C3H male, NOD female vs. NOD male, F₁ female vs. F₁ male. (B) Longitudinal analysis of hepatic iron concentration of female and male NOD-B2m−/− mice. Comparisons between age-matched female and male by Student's t test: *, P < 0.05; **, P < 0.01; ***, P < 0.005; NS = not significant.

Interestingly, hepatic iron overloading in β₂M-deficient mice appeared to be nonprogressive in older β₂M-deficient mice, in contrast to the progression typical of human HH (2). The concentrations of hepatic iron in NOD-B2m−/− increased sharply until 20–30 weeks of age and then stabilized or decreased (Fig. 3B). Similarly, the loading of hepatic iron revealed histologically in B6-B2m−/− and C3H-B2m−/− mice did not appear higher after about 30 weeks of age compared with younger mice. In fact, in many cases, hepatic iron staining was virtually undetectable in B6-B2m−/− and C3H-B2m−/− mice approaching 2 years of age (Fig. 2 A–F). In this context, is interesting to note that total iron stores of (β₂M intact) Lewis rats decline with age, raising the possibility that, unlike humans, rodents have a natural means of decreasing iron loads as they age (26).

To investigate the genetic nature of susceptibility in β₂M-deficient mice, iron concentrations from whole livers of (C3H- B2m−/− × NOD-B2m−/−) × NOD-B2m−/− back-cross progeny were analyzed at 3 months of age. These results were compared with results from C3H-B2m−/−, NOD-B2m−/−, and F₁-B2m−/− mice. Results in Fig. 4, grouped by sex, show that both male and female F₁ progeny had intermediate levels of iron loading compared with sex-matched parental mice. The relative level of susceptibility observed in parental strain mice was thus inherited by their progeny as an additive trait. Moreover, back-cross mice had levels of iron loading that were intermediate between the F₁ and NOD parental mice, with a variance that encompassed both the F₁ and NOD mouse groups. These results indicate that susceptibility to iron overload in the β₂M-deficient mouse model is heritable and can be explained by polymorphic autosomal loci.

Figure 4.

Back-cross analysis of susceptibility to iron overload in β₂M-deficient mice. Distribution of hepatic iron concentrations among the indicated 3-month-old C3H and NOD parental, F₁, and back-cross B2m−/− mice. Total numbers of mice in each group are indicated. Histogram indicates distribution of numbers of mice at the indicated iron concentration. P < 0.0005 by two-way ANOVA with Bonferroni posttest analysis: C3H female ♀ vs. NOD female, C3H female vs. F₁ female, NOD female vs. F₁ female, C3H male ♂ vs. NOD male, C3H male vs. F₁ male, NOD male vs. F₁ male, back-cross female vs. back-cross male.

Discussion

The evidence suggesting a role for MHC class I family molecules in HH arose from the finding that β₂M-deficient mice have elevated levels of iron in serum and in liver parenchymal cells (11, 12, 27). The phenotype-based positional cloning of the HFE gene (28) and recapitulation of HH in mice with a defective Hfe gene (13, 14, 23) proved that HFE was the class I protein involved in regulation of dietary iron absorption. Evidence that homozygosity for the HFE C282Y mutation is the most substantial risk factor for developing HH is compelling (2). However, clinical expression is remarkably variable even in individuals proven genotypically to be homozygous for the defective C282Y allele. In a study of C282Y homozygous Australians, only half of C282Y homozygous individuals expressed clinical disease; the remainder had subclinical or asymptomatic disease (5, 29). Clinical expression was similarly variable in other studies, with between 6 and 81% of C282Y homozygous individuals showing normal serum iron parameters (4, 5, 30). Several possibilities have been proposed to account for this discordance, including concurrence of HH with other diseases leading to misdiagnosis, environment (including dietary iron variation and blood loss), and genetic variation in loci other than HFE (2, 4, 6–8, 10).

Our studies are consistent with an important role for natural genetic variation in determining the severity of iron overload in the β₂M-deficient mouse model. One component of the observed phenotypic variation can be explained by the presence of the Y chromosome, which is partially protective in β₂M-deficient mice regardless of genetic background (Figs. 3 and 4). This result contrasts with the human situation in which females have a lower risk for developing serious iron overload, possibly because they undergo blood loss during menstruation and childbirth (25). Female mice do not lose blood during estrus (Linda Washburn, The Jackson Laboratory, personal communication), which could explain why female humans with HH are protected from iron overload better than female mice. Whether the protective effect exerted by the mouse Y chromosome is conferred by a gene(s) directly controlling iron homeostasis or, more likely, is a secondary consequence of hormonal differences between female and male mice remains to be determined.

Genetic variation in the autosomal chromosomes confers a second and more potent component to the phenotypic variation observed in iron overloading of β₂M-deficient mice. C3H and B6 backgrounds were protective, and AKR and NOD backgrounds conferred susceptibility. Very similar results were observed recently in studies involving three of these strain backgrounds (C3H, B6, and AKR) in the context of an Hfe deficiency (31). Other recent studies have found that loss-of-function defects in genes known to be critical for iron transport influence the severity of iron overload in Hfe-deficient mice (32). Our results and those of Fleming et al. (31) suggest that, in addition to null mutations, naturally occurring alleles can also play a major role in determining susceptibility and resistance to iron overload in mice lacking HFE or β₂M. The intermediate iron storage phenotypes of our F₁ mice and the broad distribution of iron loading levels in the back-cross mice (Fig. 4) are consistent with susceptibility to iron overload being controlled by more than one quantitative trait locus. It should be straightforward to use genome scanning techniques to map the loci responsible.

These studies provide a blueprint for determining the loci whose naturally occurring polymorphisms control iron overload in mice. Multiple examples exist in which quantitative trait loci detected in mice map to regions syntenic to those associated with the same disease state in humans (33–35). It is thus possible that the loci responsible for susceptibility to iron loading in Hfe-deficient mice are the same as those that modify susceptibility to iron loading in HFE mutant humans. Results of multiple family and twin studies are consistent with the existence of polymorphic non-HFE modifier loci controlling susceptibility to iron loading in humans with HH (6–10). Given the high frequency of HH in humans and the morbidity in those most severely affected, identification of modifier genes that determine the severity of iron loading in HH patients could have great practical significance, in genetic counseling for prognosis and in guiding therapy.

Abbreviations

HH

hereditary hemochromatosis

β₂M

β₂-microglobulin