Mössbauer Spectra of Mouse Hearts Reveal Age-dependent Changes in Mitochondrial and Ferritin Iron Levels

Abstract

Cardiac function requires continuous high levels of energy, and so iron, a critical player in mitochondrial respiration, is an important component of the heart. Hearts from ⁵⁷Fe-enriched mice were evaluated by Mössbauer spectroscopy. Spectra consisted of a sextet and two quadrupole doublets. One doublet was due to residual blood, whereas the other was due to [Fe₄S₄]²⁺ clusters and low-spin Fe^II hemes, most of which were associated with mitochondrial respiration. The sextet was due to ferritin; there was no evidence of hemosiderin, a ferritin decomposition product. Iron from ferritin was nearly absent in young hearts, but increased steadily with age. EPR spectra exhibited signals similar to those of brain, liver, and human cells. No age-dependent EPR trends were apparent. Hearts from HFE^−/− mice with hemochromatosis contained slightly more iron overall than controls, including more ferritin and less mitochondrial iron; these differences typify slightly older hearts, perhaps reflecting the burden due to this disease. HFE^−/− livers were overloaded with ferritin but had low mitochondrial iron levels. IRP2^−/− hearts contained less ferritin than controls but normal levels of mitochondrial iron. Hearts of young mice born to an iron-deficient mother contained normal levels of mitochondrial iron and no ferritin; the heart from the mother contained low ferritin and normal levels of mitochondrial iron. High-spin Fe^II ions were nearly undetectable in heart samples; these were evident in brains, livers, and human cells. Previous Mössbauer spectra of unenriched diseased human hearts lacked mitochondrial and blood doublets and included hemosiderin features. This suggests degradation of iron-containing species during sample preparation.

Introduction

From an early stage of fetal development until the end of life, the heart functions unceasingly to pump blood throughout the body. This requires a high and sustained level of metabolic energy unrivaled by any other organ except perhaps the brain. Mitochondria, the organelle responsible for generating most of the chemical energy in cells, serve a critical role in cardiac function. Improving our understanding of cardiac physiology is important because heart disease is the most common cause of death in the Western world (1). The preponderance of iron-containing centers in respiration-related mitochondrial proteins makes this metal a major player in cardiac physiology.

Much of the iron that enters cardiomyocytes is used to build iron-sulfur clusters (ISCs)³ and heme centers many of which are installed into mitochondrial respiratory complexes and respiration-related proteins. Excess cellular iron is generally stored within the core of cytosolic ferritin, a spherically shaped protein complex composed of variable ratios of H- and L-subunits (2). H-ferritin is especially important in cardiac function (3, 4). Besides storing iron, H-ferritin helps cells manage the metal (5). H-ferritin levels are significantly lower in mice hearts after a myocardial infarct (6). Such hearts contain spotty iron deposits and suffer from excessive oxidative stress. Deleting H-ferritin reduces the viability of cardiomyocytes and increases these deposits (6). Excessive iron and increased oxidative stress damage cardiomyocytes and contribute to heart failure (6). Mitochondrial ferritin (mt-ferritin) is highly expressed in heart mitochondria (7). It functions to sequester Fe^II in mitochondria and thus protect the organelle from iron-dependent oxidative damage (8).

Cardiomyopathy is common in iron-overload diseases such as hereditary hemochromatosis (HH) (9). HH initially impacts the liver, but eventually affects the heart, causing ROS damage especially to mitochondria (10–12). The most common cause of HH is a mutation in the HFE gene. HFE^−/− mutant livers contain 4–6 times more iron than controls, whereas the iron content of mutant hearts is slightly elevated (13, 14). Hemosiderin, a poorly defined iron-loaded breakdown product of ferritin (15), accumulates along with ferritin in mutant livers but not in mutant hearts (13, 14). Hemosiderin is found in some iron-overloaded organs (15–17).

Mössbauer (MB) spectroscopy has been used to investigate the iron content of hearts from patients with β-thalassemia, another iron-overload blood disease (18–20). The average iron concentration in the β-thalassemia heart is nearly 3 times higher than normal (11). MB spectra of β-thalassemia heart tissue exhibit features of ferritin, a high-spin Fe^III species, and perhaps hemosiderin (11, 18). The control MB spectrum was too noisy to evaluate because the heart was neither overloaded nor enriched in ⁵⁷Fe.

Individuals with Friedreich's ataxia (FA) develop cardiomyopathy in conjunction with mitochondrial iron accumulation and ROS damage (21, 22). FA arises from a deficiency in frataxin, a mitochondrial protein involved in ISC assembly (23). Eighty percent of FA patients die of cardiac disease (24).

Mice in which frataxin was deleted in cardiac muscle exhibited the classical progression of FA (20, 24–26). At 7 weeks, mutant mice became deficient in ISCs, and at 9 weeks, the iron concentration in mutant hearts increased and respiratory activity declined (27). At 10 weeks, the concentration of iron in mutant hearts contained nearly 10 times more iron than WT hearts, which contained about 1.3 mm iron (26, 28). The excess iron in the mutant cells flowed into mitochondria where it aggregated and generated ROS (23, 25, 27, 29).

MB spectroscopy was used to characterize the iron aggregates in hearts of 9-week frataxin KO mice (23). The spectrum was noisy because mice were not enriched in ⁵⁷Fe. It exhibited a single quadrupole doublet due to Fe^III oxyhydroxide nanoparticles (23). The low-temperature MB spectrum of the liver from the same frataxin KO mice was dominated by a sextet due to ferritin (11, 23).

Cardiac failure is the most prevalent cause of death in the elderly and is commonly associated with impaired energy homeostasis (30, 31). Aged mitochondria often have respiratory defects (32). Ischemia is associated with a decline in the EPR signal due to the Rieske ISC protein associated with respiratory complex III (32).

Iron-related proteins in mammals are primarily regulated by the IRP1/IRP2 system (33). Under iron-deficient cellular conditions, IRP1/2 bind mRNA transcripts of ferritin and TfR1. Doing so increases TfR1 expression and decrease ferritin levels. IRP2^−/− mice accumulate large amounts of iron in the liver (33–36), whereas IRP2^−/− brains contain WT levels of iron and WT MB features (34).

MB spectroscopy is the most powerful spectroscopic tool for probing the iron content of biological materials, but it has not been applied extensively to vertebrate animals (37). MB is relatively insensitive and can only detect ⁵⁷Fe, which accounts for just 2% of naturally occurring iron. As a result, spectra of unenriched mammalian organs are noisy, even when iron overloaded. Enriching mammals in ⁵⁷Fe is inconvenient and costly. Nevertheless, we enriched mice with ⁵⁷Fe for this study. In earlier studies, we used MB spectroscopy to characterize the iron content of brain and liver (34, 38, 39). Here we use MB to probe the iron content of healthy mouse hearts during development. We also investigate hearts from HFE^−/− mice with hemochromatosis, hearts from IRP2^−/− mice, and hearts from an iron-deficient mouse and her offspring.

Results

We wanted to characterize the iron content of healthy mammalian hearts at different developmental stages, and used MB spectroscopy as our primary tool. We enriched mice with ⁵⁷Fe to improve spectral quality. Analyses were augmented by EPR spectroscopy and ICP-MS analysis. Mice were euthanized at different ages and immediately imported into a refrigerated anaerobic glove box where blood was flushed extensively with Ringer's buffer. Hearts and other organs were removed by dissection, and frozen in MB cups for later analysis. Low-field MB spectra were collected on whole intact hearts from elderly (Fig. 1), young (Fig. 2), and adult (Fig. 3) mice. For each group, multiple spectra are shown to help distinguish sample-to-sample variations from age-dependent changes. Distinguishing this is a challenge for studies like this in which only small numbers of animals can be investigated. After collecting spectra, samples were analyzed for metal concentrations. These results, and the MB parameters used in simulations are compiled in Table 1. EPR spectra were collected (Fig. 4) on different heart samples that had been homogenized and packed into EPR tubes by centrifugation.

FIGURE 1.

Mössbauer spectra of hearts from elderly mice. All spectra presented in the text were collected at 5 K except for in C, which was collected at 100 K (and in Fig. 5D, which was collected at 70 K). In all MB spectra presented in the text, a 0.05 tesla magnetic field was applied parallel to the gamma rays. The solid red lines overlaying spectra in all MB figures are simulations using parameters listed in Table 1. A, C60, raw spectrum (all other spectra have the blood doublet removed). The solid green, blue, and gold lines above the data are simulations of the blood, CD, and ferritin components, respectively. B, C60 with contribution from blood removed; C, same as B but at 100 K; D, C52; E, I104.

FIGURE 2.

Mössbauer spectra (5 K) of hearts from young mice. A, C00; B, C01; C, C02a; D, C02b; E, C03; F, C03_D; G, C04a; H, C04b.

FIGURE 3.

Mössbauer spectra (5 K) of hearts from adult mice. A, C06; B, I08; C, C16; D, C24; E, C28; F, C28_D.

TABLE 1.

Metal concentrations in ⁵⁷Fe-enriched mouse hearts (and livers), and associated Mössbauer parameters

Metal concentrations are in μm, calculated by assuming a tissue density of 1.06 g/ml. Masses of hearts are in mg; uncertainties are ±1 mg. Values in parentheses refer to the number of hearts in the sample. If more than 1 heart was used, the mass refers to the average. Entry indicated with bold and italics is the collective mass of 4–7 hearts that had been combined. This mass, on a per-heart basis, is an outlier relative to the other samples. Sample designations C, H, and I refer to C57BL/6, HFE^−/−, and IRP2^−/− strains, respectively. The number that follows indicates the age of the mouse in weeks. Samples of the same age are distinguished with a and b. Samples from iron-deficient mice are distinguished by a subscript D. For ICP-MS analysis, each tissue was analyzed in triplicate; reported concentrations are averages; uncertainties are standard deviations. Cobalt and molybdenum concentrations were also determined but no age-dependent trends were apparent; average [Co] and [Mo] (n = 18) were 0.16 ± 0.11 μm and 0.05 ± 0.02 μm, respectively. Isomer shifts δ, quadrupole splittings ΔE_Q, and line widths Γ (all in mm/s) used in simulations were: central doublet (0.45 ± 0.01, 1.15 ± 0.02, and 0.44 ± 0.07); low-temperature ferritin (−0.10 ± 0.05, 0.44 ± 0.07, and 0.77 ± 0.14) with H_eff = 490 ± 6 kG; blood (0.91 ± 0.06, 2.30 ± 0.04, and 0.43 ± 0.11); high-temperature ferritin (0.44, 0.70, and 0.55). Mössbauer spectra were calibrated against α-iron foil at room temperature. For convenience, the sum of the percent relative intensities for the three major MB components were forced to 100%; however, 10–15% of the actual spectral intensity cannot be accounted for by these three components.

Sample	Mass of heart	[57Fe]	[Fetot]	[Cu]	[Mn]	[Zn]	Central doublet	Ferritin sextet	Blood doublet
	mg						%; μm	%; μm	%; μm
C00 (1 day)	65 (1)	240 ± 3	300 ± 8	10 ± 1	1.6 ± 0.2	31 ± 7	20; 60	20; 60	60; 180
C01	100	350 ± 7	470 ± 10	33 ± 1	4.9 ± 0.1	43 ± 1	38; 180	17; 80	45; 210
C02a	180 (1)	240 ± 9	290 ± 10	12 ± 0.4	2.3 ± 0.1	26 ± 1	53; 150	00; 00	47; 140
C02b	59 (2)	170 ± 3	240 ± 9	19 ± 2	3.3 ± 0.1	26 ± 1	46; 110	13; 30	41; 100
C03	NDa	ND	ND	ND	ND	ND	50; ND	0; ND	50; ND
C04a	98 (4)	600 ± 8	710 ± 9	29 ± 0.3	8.9 ± 0.2	46 ± 1	40; 280	10; 70	50; 360
C04b	61 (2)	570 ± 10	610 ± 10	48 ± 1	11 ± 0.2	45 ± 1	44; 270	12; 70	44; 270
C06	140 (1)	620 ± 1	670 ± 10	42 ± 1	8.1 ± 0.1	43 ± 1	36; 240	17; 110	47; 320
C12	150 (3)	63 ± 1	610 ± 10	38 ± 1	5.8 ± 0.1	37 ± 1	31; 190	32; 190	37; 230
C16	ND	ND	ND	ND	ND	ND	33; ND	28; ND	39; ND
C24	240 (2)	950 ± 8	1050 ± 8	45 ± 1	7.7 ± 0.1	47 ± 1	30; 320	28; 290	42; 440
C28	140 (2)	740 ± 10	780 ± 10	36 ± 0.4	5.8 ± 0.1	34 ± 1	28; 220	36; 280	36; 280
C52	250 (2)	700 ± 10	1100 ± 20	39 ± 1	8.1 ± 0.1	48 ± 1	22; 240	48; 530	30; 330
C60	130 (1)	650 ± 30	960 ± 40	40 ± 10	3.4 ± 0.1	34 ± 1	16; 150	41; 390	43; 420
C03D (pups)	94 (4)	260 ± 10	310 ± 20	40 ± 2	24 ± 1	56 ± 2	54; 170	00; 000	46; 140
C28D (mom)	390 (1)	740 ± 5	850 ± 5	48 ± 0.3	8.8 ± 0.1	49 ± 1	35; 300	13; 110	52; 440
H12	160 (3)	58 ± 2	640 ± 20	58 ± 1	9.9 ± 0.2	60 ± 1	27; 170	37; 240	36; 230
H12 (liver)	1600 (1)	1200 ± 200	5600 ± 700	29 ± 4	14 ± 2	90 ± 2	0.5; 30	90; 5040	9.5; 530
C12 (liver)	1300 (1)	20 ± 90	960 ± 500	60 ± 30	14 ± 6	120 ± 70	18; 170	60; 580	22; 210
I08	95 (2)	520 ± 7	560 ± 7	35 ± 2	7.0 ± 0.1	32 ± 1	38; 210	10; 60	52; 290
I104	200 (1)	600 ± 20	800 ± 30	36 ± 1	7.7 ± 0.3	32 ± 1	30; 240	34; 270	36; 290

^a ND, not determined.

FIGURE 4.

X-band EPR spectra of heart homogenates from mice of different ages. A, 4 weeks; B, 12 weeks; C, 32 weeks; D, 52 weeks; E, 60 weeks. EPR conditions: temperature, 4 K; microwave frequency, 9.37 GHz; microwave power, 0.2 milliwatt (30 dB); modulation amplitude, 10 G. Dashed vertical lines indicate the fields corresponding to g values of 2.16, 2.00, 1.93, and 1.86.

MB spectra were composed of three major features, including a sextet and two quadrupole doublets (Fig. 1A). One doublet had parameters (Table 1) of high-spin Fe^II hemoglobin. This doublet, simulated in green in Fig. 1, was assigned to residual blood. This so-called “blood doublet” (BD) was present in all raw spectra. The average relative intensity of the BD was 43%, with a standard deviation of just ±5% (n = 18). To view other spectral features better, we have removed the BD contribution from all spectra (except for that in Fig. 1A). The consistent high percentage of iron due to blood in our samples raises doubts that flushing by cardiac puncture is effective in removing blood from the heart itself. Nevertheless, knowing the percent contribution of the BD to the overall intensity of each spectrum and the absolute concentration of iron in each sample allowed us to calculate the concentration of iron within heart cells themselves for the first time. We have already done this for brain and liver; another group has done the same for liver and spleen (40). As such, the iron concentrations listed in Table 1 are probably the most accurate determinations for hearts from iron-sufficient mammals. The average iron concentration in the healthy heart samples (with blood excluded) was 400 ± 200 μm (n = 11), depending on age. The approximate concentration of iron in young, adult, and elderly hearts was 300, 450, and 700 μm, respectively.

We assigned the sextet in heart MB spectra to Fe^III aggregates in ferritin cores. The gold line above the data of Fig. 1A simulates the ferritin sextet using parameters in Table 1. To assess whether hemosiderin contributed to this absorption, we collected a 100 K spectrum of the C60 heart. The baseline was devoid of sextet features (Fig. 1C), suggesting the absence of this degradation product.

The other doublet in MB spectra of hearts, called the “central doublet” or CD, was simulated as the blue line in Fig. 1A using parameters in Table 1. The CD is due collectively to [Fe₄S₄]²⁺ clusters and low-spin Fe^II hemes; the two types of centers cannot be resolved by MB spectroscopy because they are diamagnetic and have similar δ and ΔE_Q values. The CD dominates the spectrum of mitochondria isolated from brain, liver, and human cells (34, 38, 41), which allows us to assign this spectral feature to respiration-related complexes in that organelle. [Fe₄S₄]²⁺ clusters and LS Fe^II hemes are undoubtedly present in non-mitochondrial regions of the heart, but they are minor contributors to spectral intensity. We have quantified the intensity of the CD in the spectra of elderly hearts (Fig. 1) and find that it corresponds to 110–200 μm iron.

The CD dominated the MB spectra of young hearts (defined as newborn to 4 weeks) with an intensity that corresponded to 60–220 μm (Fig. 2 and Table 1). The intensity of the ferritin sextet was dramatically lower in spectra of young hearts than in those of elderly hearts; indeed the feature was absent in some spectra of young hearts. Adult hearts (Fig. 3, 6–28 weeks) exhibited even higher concentrations of the CD (220–350 μm iron) and intermediate levels of ferritin (55–320 μm iron).

EPR spectra of packed mouse heart homogenates of different ages are shown in Fig. 4. The g = 2 region consisted of numerous overlapping signals including a g_ave = 1.94 signal due to [Fe₂S₂]¹⁺ and/or [Fe₄S₄]¹⁺ clusters, a g_ave = 1.90 signal due to the [Fe₂S₂]¹⁺ cluster in the Rieske ISC protein, a radical signal at g = 2.00, and an unassigned resonance at g = 2.16. The same signals were observed in mouse brain and liver homogenates and in mitochondria isolated from these tissues (34, 38). No age-dependent EPR trends were apparent.

MB spectra of HFE^−/− hearts and matching controls are shown in Fig. 5, A and B, respectively. Spectra were noisy because the mice were enriched in ⁵⁷Fe for only 4 weeks. Relative to controls, the HFE^−/− heart contained slightly more total iron, including more ferritin and less CD. In a qualitative sense, the HFE^−/− heart appeared slightly older. The spectra of the HFE^−/− liver and control (Fig. 5, C and E) were strikingly different, with only the HFE^−/− liver overloaded with ferritin iron. The 70 K spectrum of the HFE^−/− liver (Fig. 5D) was devoid of any sextet features in the baseline, confirming the absence of hemosiderin in this diseased liver. The HFE^−/− liver contained significantly less CD than the control liver (notice the absence of spectral intensity between the two inner lines of the sextet of Fig. 5C), implying that the HFE^−/− liver contains less mitochondria than WT livers. The MB spectrum of another diseased liver (Fig. 4F of Ref. 34) also showed a deficiency of mitochondrial iron.

FIGURE 5.

Mössbauer spectra of 12 weeks HFE^−/− hearts and livers versus controls. A, HFE^−/− hearts; B, control hearts; C and D, HFE^−/− livers; E, control livers. All spectra were collected at 5 K except for D, which was collected at 70 K.

Spectra of hearts from IRP2^−/− mice (I08 in Fig. 3B and I104 in Fig. 1E) were similar to controls. The intensities of the ferritin sextets were slightly less in mutant spectra but the effect was modest. The intensities of the CD in the mutant spectra were comparable with (if not slightly stronger than) the same feature in the controls. The concentration of iron in I08 and I104 (Table 1) were slightly lower than in the controls. Like HFE^−/− livers, IRP2^−/− livers accumulated large amounts of ferritin iron; however, they contained more mitochondrial iron (see Fig. 4E of Ref. 34) than did HFE^−/− livers.

We fed a pregnant ⁵⁷Fe-enriched mouse an iron-deficient diet starting 1 week before she gave birth to 4 offspring, and then continued to feed her that diet until the pups were 3 weeks old. Mother and pups were then euthanized and their hearts were examined by MB spectroscopy. The MB spectrum of the pup hearts (Fig. 2F) exhibited standard features of young hearts, i.e. an intense CD and no ferritin sextet. The MB spectrum of the 28-week-old heart from the mother (Fig. 3F) exhibited less than half as much ferritin as a control heart of the same age (110 versus 280 μm). Surprisingly, the CD of the iron-deficient heart was more intense than the control (300 versus 220 μm).

Discussion

This is the first Mössbauer study of the iron content of healthy ⁵⁷Fe-enriched mammalian hearts. All previous MB studies have investigated diseased human hearts that were overloaded in iron. In those earlier studies, spectra from healthy controls could not be obtained because samples were not ⁵⁷Fe enriched and so spectral intensities would have been exceedingly weak.

One advantage of obtaining MB spectra of whole hearts is the ability to distinguish iron due to blood in vessels permeating the heart ([Fe]_B) from iron in the heart cells themselves ([Fe]_H). The total measured iron in whole hearts was [Fe]_T = [Fe]_B + [Fe]_H. Previous determinations of heart iron concentration probably overestimated [Fe]_H by nearly a factor of 2 due to the large proportion of blood in the tissue. An exception is an earlier study from our group in which 40% of the iron in newborn to 6-week-old mice hearts was assumed to arise from blood; [Fe]_H values between 450 and 720 μm iron were reported (39). Another study reported that hearts from 12-week-old mice contained 1.4–1.6 mm iron (42) (millimolar units calculated from those reported). In another study, heart iron concentrations of 1.1–1.5 mm (28) were reported. If 43% of those concentrations were presumed to be due to blood, [Fe]_H would be between 630 and 900 μm, reasonably similar to the concentrations we obtained. Comparing samples more precisely would require matching ages and the concentration of iron in the diet. In other studies, hearts from WT mice reportedly contained 4 (14, 43), 6 (14), 7 (23), and ∼100 mm (36) iron. These values overestimated the actual heart iron concentrations by factors of 10–200.

Apart from blood, the two major forms of iron in heart tissues were ferritin and mitochondrial iron. Iron in hearts of young mice was mostly found in respiration-related ISCs and heme centers within mitochondria. The level of mitochondrial iron increased as young mice matured into adults, and then it stabilized or declined slightly in old age. As mice aged, their hearts accumulated increasing amounts of ferritin iron such that the iron content of elderly mice was dominated by this storage form of iron.

We used the overall concentration of iron in ⁵⁷Fe-enriched hearts and the percentage of CD and ferritin as determined by MB spectroscopy to calculate the absolute concentration of these two features (Table 1 and Fig. 6). Age-dependent mitochondrial iron concentrations ([CD]_age) were assessed by assuming the function in Equation 1.

$${\left[\text{CD}\right]}_{\text{age}}={\left[\text{CD}\right]}_{\text{newborn}}+\frac{{\left[\text{CD}\right]}_{\text{g}}\text{age}}{K_{CD}+\text{age}}$$ (Eq. 1)

FIGURE 6.

Heart iron concentrations (excluding blood) as a function of age. Black circle, total iron; blue circles, ferritin; red circles, central doublet. Values were obtained from Table 1. The solid red line is the best-fit simulation generated assuming Equation 1. The solid blue line is the equivalent for ferritin. The solid black line is the sum of the red and blue lines. Simulations were optimized against the data by minimizing root mean square deviation. Best-fit parameters (concentrations in μm) were: [CD]_newborn, 31; [CD]_g, 250; K_CD, 2.8 weeks; [FN]_newborn, 0; [FN]_g, 450; K_FN, 14 weeks. Log₂ numbers 0, 2, 4, and 6 refer to 1, 4, 16, and 64 weeks, respectively.

A similar function was assumed to assess age-dependent ferritin iron concentrations, [FN]_age. In this case, parameters [FN]_newborn, [FN]_g, and K_FN were used. The optimized functions are given as the solid lines in Fig. 6. Total heart iron at any age (ignoring the blood contribution) was calculated as the sum of these two components, i.e. [Heart Fe]_age = [CD]_age + [FN]_age. The best-fit parameters suggest that the heart from a newborn should contain about 60 μm mitochondrial iron and little ferritin, as observed. Mitochondrial iron should increase during the first 6 weeks, maximizing at 280 μm. Ferritin iron levels should increase more slowly over the first 24 weeks, maximizing at 450 μm in adults and elderly.

These trends, illustrated qualitatively in Fig. 7, are easily rationalized given the requirement of the heart to function unceasingly from before birth until death. This high-energy requirement is satisfied by the high levels of mitochondria (and perhaps by not storing much iron). As hearts age, they store increasing amounts of ferritin iron.

FIGURE 7.

Illustration of how iron content changes with age in healthy mouse hearts. The opacity of the background reflects the overall iron concentration in the heart. The size of the mitochondria reflects the concentration of iron present in that organelle. The number of yellow hexagons represents the concentration of iron present as ferritin.

These age-dependent changes differ from those occurring in the developing brain and liver (38, 34). The iron content of the newborn brain is dominated by ferritin iron, with modest levels of mitochondrial iron. This makes sense because the newborn does not use its brain extensively. Then, during the first few weeks of life, the animal opens its eyes, and begins to sense and interact with its environment, increasing its need for energy to drive brain function. To accommodate these needs, ferritin iron in the brain is transformed into mitochondrial iron. In fact, the overall iron concentration in the brain declines during the first few weeks of life, as the combined rate of mitochondriogenesis and brain volume growth outpaces the import rate of new iron (39). Gradually, as the animal ages, the brain, like the heart, accumulates iron in the form of ferritin.

The changes in the iron content of the liver differ from those of brain and heart. The newborn liver contains very high concentrations of ferritin iron, much of which is exported from the liver during the first week of life (34). This burst of exported iron is most likely used to help other organs develop (including the heart but perhaps not the brain). In healthy mice raised under iron-sufficient conditions, the liver accumulates a modest amount of iron as it ages, but not an excessive amount (this might depend on the iron concentration in the diet).

Another result that highlights an important difference between heart, brain, and liver is the absolute concentration of mitochondrial iron in these organs. The concentrations of mitochondrial iron in the developing brain and liver maximize at about 110 and 180 μm, respectively, whereas it maximizes at about 280 μm in the heart. These differences may again reflect the high and sustained level of chemical energy required by the heart relative to these other organs.

Another difference between mouse hearts, brains, livers, and human Jurkat cells is the concentration of nonheme high-spin (NHHS) Fe^II in these organs/cells. Jurkat cells are used for this comparison because they have been characterized by Mössbauer spectroscopy (41, 44) and so the concentration of NHHS Fe^II can be quantified. The concentration of such ions in the brain is 10 μm of 170 μm total iron (38). Their concentration in the liver is 20–40 μm of 300–650 μm total iron (34). In Jurkat cells, the concentration of NHHS Fe^II is 40 μm of 400 μm total iron (41). By contrast, we had difficulty detecting the NHHS Fe^II doublet in most heart spectra. The heart contains high levels of mt-ferritin, which is thought to sequester Fe^II. It is interesting to consider that the concentration of NHHS Fe^II ions is so low in heart tissue because mt-ferritin sequesters these ions.

It is also interesting to compare the total iron concentrations in heart, brain, liver, and Jurkat cells (given above). Heart, liver, and Jurkat cells contain about 400 μm iron, whereas the brain contains less than half of that concentration. Perhaps the danger associated with Fenton chemistry has minimized the use of this metal in the most complicated and delicate of all organs.

Relative to age-matched controls, HFE^−/− hearts contained slightly more total iron, including more ferritin and less mitochondrial iron; these differences were typical of somewhat older hearts, perhaps reflecting an increased burden on the heart due to hemochromatosis. The HFE^−/− liver was iron-overloaded, as expected, but it also contained lower than normal concentrations of mitochondrial iron, which was not expected. Given the literature suggesting that hemosiderin ought to be present in diseased livers, we were surprised not to detect it in our samples.

The lack of IRP2 in IRP2^−/− mutant mice should allow ferritin transcripts to be translated and TfR1 transcripts to be degraded quickly under all nutrient iron conditions. Under iron-deficient conditions, the absence of IRP2 should lead to insufficient cellular iron, whereas under iron-excess conditions, the same absence should lead to excess cellular iron. In our study, mice were fed iron-sufficient chow (but just barely so) and so cellular iron would be expected to be near normal levels. Consistent with this, mutant hearts contained about 85% of the iron found in WT hearts, with an iron content similar to that of WT hearts. Similarly, spectra of the brains of IRP2^−/− mice were virtually indistinguishable from controls (34, 38). By contrast, IRP2^−/− livers were iron-overloaded with ferritin (34). Perhaps the IRP1 system is sufficient to retain iron homeostasis in the mutant heart and brain, whereas the liver relies more on the IRP2 system for regulation, and so dysregulation was more severe. Interestingly, the level of mitochondrial iron in IRP2^−/− livers was substantially higher than in the HFE^−/− livers, even though both livers were overloaded with iron.

We also examined the effect of iron deficiency on a pregnant mouse and her offspring. As expected, the iron-deficient mother contained subnormal levels of ferritin iron, but normal levels of mitochondrial iron. A similar phenomenon was observed in spectra of iron-deficient brains (38), in which the CD was actually more intense than it was in spectra of iron-sufficient brains. The hearts of the 3-week-old pups from the iron-deficient mother contained normal levels of mitochondrial iron. This suggests that the iron-deficient mother does not limit the transfer of iron to her offspring during pregnancy or lactation. Interestingly, the iron-deficient mother also had a higher concentration of mitochondrial iron in the heart than did the iron-sufficient control.

Our study provides “control” spectra for the published MB spectrum of a frataxin KO mouse heart (23). That spectrum contained only a doublet species due to Fe^III oxyhydroxide nanoparticles (no ferritin or CD was observed). It should also be compared with the MB spectrum of human Jurkat cells in which frataxin expression was knocked-down by RNAi (44). That spectrum also consisted of a nanoparticle doublet with parameters (δ = 0.48 mm/s, ΔE_Q = 0.57 mm/s) very near to those reported for the frataxin KO heart doublet. We conclude that the type of nanoparticles in both mouse hearts and human Jurkat cells must be similar.

Finally, we have pondered as to why previously published MB spectra of human hearts lacked the mitochondrial doublet and the blood doublet that we have observed with such strong intensities. Published spectra only exhibit features due to ferritin and hemosiderin. Our samples certainly contained ferritin, but none appeared to contain hemosiderin. Perhaps the iron associated with mitochondria and blood in published samples of human organs was converted into ferritin and/or hemosiderin iron during sample preparation. Few details of sample preparation were provided in those studies, but we suspect that the period between death and freezing samples for MB analysis was longer for human (days?) than mouse tissues (30 min). Also, embalming solutions and/or exposure to O₂ might alter the iron content of human tissues. In contrast, our mouse tissues were dissected immediately in an anaerobic and refrigerated glove box. They were flushed with an isotonic aqueous buffer, and were frozen within 30 min of death. Clearly there are advantages to using non-human mammals in MB studies besides the ability to enrich them with ⁵⁷Fe. These advantages, along with the availability of many iron-related transgenic strains of mice, make these small mammals ideal for future MB investigations of iron-related diseases.

Experimental Procedures

All procedures involving mice were approved by the Animal Care and Use committee at Texas A&M University. The original C57BL/6 mice used in the study were a gift from Louise Abbott (Texas A&M University). IRP2^−/− mice were a gift from Tracey Rouault (National Institutes of Health). C57BL/6 and IRP2^−/− mice were bred as described (38). They were fed iron-deficient chow (Harlan Teklad, Madison, WI; number 80396) to which 50 mg of ⁵⁷Fe^III (Isoflex USA, San Francisco, CA) citrate per kg of chow was added. An iron-deficient pregnant mouse was raised similarly except that she was fed iron-deficient chow unenriched in ⁵⁷Fe for 1 month, starting a week before giving birth. Water was distilled and deionized. Three 8-week-old male HFE^−/− (B6.129S6-Hfe<tm2Nca>/J; 017784) mice and matching C57BL/6 controls were purchased from The Jackson Laboratory. These mice were fed iron-deficient chow to which 200 mg of ⁵⁷Fe/kg of chow and 1.2 g of ascorbic acid/kg of chow were added. They were also given triple-distilled water to which 100 μm ⁵⁷Fe^III citrate and 1 mm ascorbic acid had been added. These 6 mice were euthanized 1 month after arrival, at 12 weeks of age. Mice were raised in iron-deficient cages, and euthanized with ketamine and xylazine, as described (38). Immediately after death, carcasses were imported into a refrigerated (8 °C) nitrogen-atmosphere glove box (Mbraun Labmaster) containing <10 ppm O₂. Preparing samples anaerobically prevented any oxidation or degradation of iron centers in these samples. Animals older than 1 week were flushed by puncturing the heart with a needle, cutting the caudal vena cava, and flowing Ringer's buffer into the heart at 700 μl/min for 5 min per 10 g of animal mass. Organs were removed, weighed, transferred to MB cups, frozen, and removed from the box. EPR samples were prepared by thawing frozen samples anaerobically, adding ∼5 ml of buffer, homogenizing using a tissue grinder, transferring to an EPR tube, capping the tube, removing it from the glove box, and spinning the tubes to pack the material. MB spectroscopy was performed as described (38). Continuous-wave X-band EPR spectra were obtained using a Bruker Elexsys E500A spectrometer with a cryogen-free cooling system. After spectra were collected, samples were thawed and transferred quantitatively to 15-ml plastic tubes with screw-tops. An equal volume of concentrated trace metal-grade nitric acid was added to each sample, and sealed samples were heated to 90 °C overnight. Samples were diluted with 6.0 ml of distilled deionized water and analyzed by ICP-MS (Agilent 7700x).

Author Contributions

M. C. led in raising the mice, dissections, sample preparation, and early spectroscopy characterization. J. D. W. collected the remaining MB spectra, the EPR spectra, performed the metal analysis, prepared figures, and helped write the paper. P. A. L. designed the study, helped with all aspects of it, and wrote most of the paper. All authors approved the final version of the manuscript.