Deficiency of mitochondrial calcium uniporter abrogates iron overload-induced cardiac dysfunction by reducing ferroptosis

Abstract

Iron overload associated cardiac dysfunction remains a significant clinical challenge whose underlying mechanism(s) have yet to be defined. We aim to evaluate the involvement of the mitochondrial Ca²⁺ uniporter (MCU) in cardiac dysfunction and determine its role in the occurrence of ferroptosis. Iron overload was established in control (MCU^fl/fl) and conditional MCU knockout (MCU^fl/fl-MCM) mice. LV function was reduced by chronic iron loading in MCU^fl/fl mice, but not in MCU^fl/fl-MCM mice. The level of mitochondrial iron and reactive oxygen species were increased and mitochondrial membrane potential and spare respiratory capacity (SRC) were reduced in MCU^fl/fl cardiomyocytes, but not in MCU^fl/fl-MCM cardiomyocytes. After iron loading, lipid oxidation levels were increased in MCU^fl/fl, but not in MCU^fl/fl-MCM hearts. Ferrostatin-1, a selective ferroptosis inhibitor, reduced lipid peroxidation and maintained LV function in vivo after chronic iron treatment in MCU^fl/fl hearts. Isolated cardiomyocytes from MCU^fl/fl mice demonstrated ferroptosis after acute iron treatment. Moreover, Ca²⁺ transient amplitude and cell contractility were both significantly reduced in isolated cardiomyocytes from chronically Fe treated MCU^fl/fl hearts. However, ferroptosis was not induced in cardiomyocytes from MCU^fl/fl-MCM hearts nor was there a reduction in Ca²⁺ transient amplitude or cardiomyocyte contractility. We conclude that mitochondrial iron uptake is dependent on MCU, which plays an essential role in causing mitochondrial dysfunction and ferroptosis under iron overload conditions in the heart. Cardiac-specific deficiency of MCU prevents the development of ferroptosis and iron overload-induced cardiac dysfunction.

Introduction

Iron is an essential biological element involved in normal cellular homeostasis and plays an important role in a wide range of physiological processes including red blood cell production, oxygen transport, DNA synthesis, and electron transport [60]. However, iron can become toxic when its homeostasis is perturbed resulting in the formation of free radicals. Iron overload occurs in patients with hereditary forms of systemic iron dysregulation (primary hemochromatosis) or as a result of multiple blood transfusions in attempts to correct anemia such as in patients with sickle cell disease or severe β-thalassemia (secondary hemochromatosis).

The heart is one of the major target organs for iron deposition which can be manifested as iron overload cardiac dysfunction [28]. Cardiac dysfunction and/or arrhythmias are the leading cause of morbidity and mortality in patients with hemochromatosis and resulting cardiac iron overload. Nevertheless, the underlying mechanisms have not been well defined [27, 42, 52].

Iron overload may occur at the cellular as well as mitochondrial level. Recent studies have suggested that excess iron in mitochondria plays a critical role in causing cardiomyocyte dysfunction not only under known iron overload conditions such as those seen in patients with hereditary hemochromatosis, thalassemia or sickle cell disease [25, 27, 43], but also during cardiac ischemia–reperfusion injury. Furthermore, iron related cardiotoxicity has been associated with some anticancer drugs (particularly anthracyclines, e.g. doxorubicin) [5, 37].

The mitochondrial Ca²⁺ uniporter (MCU) is known to be involved in the uptake of Ca²⁺ into mitochondria [11]. Very recently it has been proposed that the MCU may also transport ferrous ion (Fe²⁺) into mitochondria with mitoferrin-2 (Mfrn2) being a regulator of MCU [53]. The role of MCU as a transporter for both Ca²⁺ and Fe²⁺ has been suggested in liver cells [35, 66]. In addition, MCU-dependent accumulation of both Ca²⁺ and Fe²⁺ has also been demonstrated in neutrophils [51]. Supporting this notion, it has been shown that MCU is involved in brain and heart cell mitochondrial dysfunction under iron loaded conditions [43]. However, it is not clear how Mfrn2 and MCU might be involved in mitochondrial iron uptake. Moreover, there is currently no mechanistic evidence that links MCU-dependent mitochondrial iron uptake to mitochondrial or cardiac dysfunction.

Ferroptosis has been recently identified as a new form of non-apoptotic cell death caused by the accumulation of iron-dependent lipid peroxidation [14, 38]. Ferroptosis can be inhibited by iron chelation, antioxidants, and ferroptosis-specific inhibitors such as ferrostatin-1 (Fer-1), a lipid radical scavenger. Lipid peroxidation and subsequent induction of ferroptosis have been linked to two parallel pathways which include (1) the glutathione peroxidase 4 (GPX4) pathway and (2) the ferroptosis suppressor protein 1-coenzyme Q10 (FSP1-CoQ10) pathway [3, 21, 31]. Despite recent intensive studies by oncology and neuroscience groups, the study of ferroptosis in the heart is just emerging [1, 17–19, 69]. Although ferroptosis has been suggested to play a crucial role in the cardiotoxicity associated with some anti-cancer drugs, iron overload states as well as ischemia–reperfusion injury, it remains less clear on how iron mechanistically links the above proposed ferroptosis pathways in the heart. As a result, the role of MCU in iron overload-induced ferroptosis and cardiac dysfunction remains unknown.

In the present study, we have used inducible conditional MCU knockout mice [44] (MCU^fl/fl-MCM) and determined the pivotal role of MCU and mitochondrial iron uptake in the occurrence of ferroptosis and cardiac dysfunction under iron loading conditions. Our results suggest that cardiac-specific deficiency of MCU is protective against the development of iron associated cardiac dysfunction and ferroptosis. A preliminary report of our findings has appeared in abstract form [20].

Methods

All animal experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at Rutgers-New Jersey Medical School.

Control (MCU^fl/fl) and conditional MCU knockout (MCU^fl/fl-MCM) mice were used in this study [44, 49]. Western blot analysis showed that MCU expression levels were efficiently reduced in heart tissue and cardiomyocytes of MCU^fl/fl-MCM mice after tamoxifen treatment (online Fig. S1A–D).

An expanded “Methods” section is available in the Online Data Supplement.

Results

MCU deficiency abrogates cardiac dysfunction in chronically iron loaded mice

We have modified a previously described protocol [10, 58] to establish a chronic iron overload mouse model. Iron dextran was injected intraperitoneally (i.p.) for 6 weeks in both MCU^fl/fl (WT) and MCU^fl/fl-MCM (KO) mice. Iron administration did not alter the MCU expression in WT mice (online Fig. S1E, F). ECHO was performed to evaluate LV function at both baseline and after chronic iron injection. As shown in the M-mode ECHO images, no differences were observed in LVEF or LVFS before iron or vehicle (Ctl) injection at baseline for any of the experimental groups (online Fig. S2). However, in MCU^fl/fl mice after chronic iron injection both LVEF and LVFS were significantly reduced compared to control MCU^fl/fl group that received vehicle injections (Fig. 1A–C) indicating the development of cardiac dysfunction in MCU^fl/fl mice that received iron. In contrast, no reduction in LVEF or LVFS was observed in MCU^fl/fl-MCM mice after chronic iron administration (Fig. 1A–C). Furthermore, we found an increase in the susceptibility to arrhythmias in iron treated MCU^fl/fl mice that have been subjected in vivo to sympathetic stress by administration of 1 mg/kg body weight of isoproterenol. In contrast, in iron treated MCU^fl/fl-MCM mice the susceptibility to arrhythmias was significantly attenuated (online Fig. S3). Taken together, these results suggest that MCU deficiency both abrogates the development of cardiac dysfunction and reduces susceptibility to arrhythmogenesis.

Fig. 1.

Left ventricular function and histological properties in the hearts from WT (MCU^fl/fl) and MCU KO (MCU^fl/fl-MCM) mice after chronic iron (Fe) vs. vehicle control (Ctl) injection. A Representative M-mode echo images. B, C Summary data (14 randomly selected mice from each group) of left ventricular ejection fraction (LVEF) and left ventricular fractional shortening (LVFS). D Representative Prussian blue staining images showing iron deposition within the left ventricles. E Quantification of Prussian blue staining levels (%). F Representative images of Picrosirius red staining to visualize fibrosis within left ventricles. G Quantified data of Picrosirius red staining levels (%). Number of tissue fields analyzed (from 3–5 animals in each group) is listed in columns enclosed in brackets. **p < 0.01, NS not significant, two-way ANOVA followed by Tukey test

Although a reduction in heart function was observed in MCU^fl/fl mice after chronic iron administration (Fig. 1), iron administered MCU^fl/fl and MCU^fl/fl-MCM mice showed similar amounts of iron deposition in LV tissue (Fig. 1D, E). Furthermore, similar levels of fibrosis (Fig. 1F, G) in the LV of MCU^fl/fl vs. MCU^fl/fl-MCM mice were observed after chronic iron treatment.

Effect of iron overload on mitochondrial function in ventricular cardiomyocytes from MCU^fl/fl and MCU^fl/fl-MCM mice

To further understand the cellular mechanism(s) that might contribute to the observed reduction in heart function in MCU^fl/fl mice after chronic iron loading, we performed experiments at the level of the isolated cardiomyocyte. LV cardiomyocytes were isolated from mice that had been chronically injected with iron or vehicle. We postulated that the protection from iron loading in MCU^fl/fl-MCM mouse hearts might be mediated by a lower mitochondrial iron uptake and lower mitochondrial/cellular oxidative stress. Consistent with our hypothesis that MCU is involved in mitochondrial iron uptake, a higher mitochondrial Fe²⁺ level as indicated by a lower RPA fluorescence intensity was detected in intact cardiomyocytes from MCU^fl/fl mice (Fig. 2A, B). In contrast, the mitochondrial Fe²⁺ level was not changed in cardiomyocytes from MCU^fl/fl-MCM mice (Fig. 2A, B). MCU-dependent iron uptake by mitochondria was further supported by studies using saponin-permeabilized cardiomyocytes exposed to acute ferrous Fe²⁺ treatment (online Fig. S4). Acute Fe²⁺ treatment caused significant Fe²⁺ loading into mitochondria in MCU^fl/fl as indicated by the apparent decline in the intensity of RPA fluorescence (online Fig. S4A, D). Consistent with our hypothesis that MCU is involved in mitochondrial Fe²⁺ uptake, Fe²⁺ loading was attenuated in MCU^fl/fl-MCM cardiomyocytes (online Fig. S4B, D) and in the presence of the MCU inhibitor Ru360 (online Fig. S4C, D).

Fig. 2.

Intramitochondrial iron (Fe²⁺) accumulation, mitochondrial membrane potential $\Delta {\psi }_m$, ROS generation, mitochondrial Ca²⁺, and mitochondrial respiration in ventricular cardiomyocytes isolated from MCU^fl/fl and MCU^fl/fl-MCM mice after chronic iron (Fe) vs. vehicle (Ctl) injection. A Representative images of RPA-loaded ventricular cardiomyocytes. B Quantified data of Fe²⁺ levels detected by RPA fluorescence. A decreased level (quenching) in RPA fluorescence indicates an increased level of mitochondrial Fe²⁺. C Representative images of TMRM-loaded ventricular cardiomyocytes. D Quantified data of $\Delta {\psi }_m$ detected by TMRM fluorescence. E Representative images of DCF-loaded cardiomyocytes. F Quantified data of intracellular ROS levels detected by DCF fluorescence. G, H Same as E and F, except that the cells were loaded with MitoSOX Red that detects mitochondrial ROS levels. I Representative Ca²⁺ traces evaluating mitochondrial Ca²⁺ content in cardiomyocytes. The cells were perfused in a Ca²⁺- and Na⁺-free Tyrode’s solution containing L i⁺ (LiT) and inhibitors of RyR (tetracaine, TTC) and SERCA (thapsigargin, THG). J Quantified data of basal mitochondrial Ca²⁺. **p < 0.01, *p < 0.05, NS not significant, two-way ANOVA followed by Tukey test. Number of cardiomyocytes from 3 to 4 randomly selected mice from each group is indicated in each bar graph (B, D, F, H, J). K, L Evaluation of oxygen consumption rate (OCR) in MCU^fl/fl and MCU^fl/fl-MCM using the Seahorse analyzer. M Quantified data of maximum OCR. N Quantified data of spare respiratory capacity (SRC). The number of cell culture wells from 3 to 4 randomly selected mice is indicated in each bar graph. *p < 0.05, two-way ANOVA followed by Tukey test

Previous studies have suggested that iron overload may cause mitochondrial dysfunction [25, 63]. Consistent with this notion, we found depolarization of the $\Delta {\psi }_m$ in cardiomyocytes isolated from MCU^fl/fl after chronic iron overload, as indicated by a reduction in the intensity of TMRM fluorescence. Meanwhile, the $\Delta {\psi }_m$ remained unchanged in cardiomyocytes from MCU^fl/fl-MCM with chronic iron overload (Fig. 2C, D). Furthermore, iron loading increased cytosolic ROS levels (measured by DCF) (Fig. 2E, F) as well as mitochondrial superoxide levels (measured by MitoSOX red) (Fig. 2G, H) in MCU^fl/fl cardiomyocytes. However, no significant increase in ROS levels (measured by DCF or MitoSOX red) was detected in MCU^fl/fl-MCM cardiomyocytes (Fig. 2E–H). These results indicate that MCU deficiency may prevent mitochondrial dysfunction by reducing iron load and ROS production as well as by maintaining the $\Delta {\psi }_m$ under iron overload conditions.

Since MCU is thought to be primarily a uniporter for Ca²⁺, we also evaluated mitochondrial Ca²⁺ levels in isolated intact cardiomyocytes as previously reported [45, 72]. As shown in Fig. 2I, application of 2 μM FCCP induced a significant increase in the intracellular Ca²⁺ level. This increase in intracellular Ca²⁺ upon FCCP application was considered to be proportional to basal mitochondrial Ca²⁺ load. In control groups, the basal Ca²⁺ level in mitochondria was not different in cells from MCU^fl/fl (1.52 ± 0.03) vs. MCU^fl/fl-MCM (1.57 ± 0.05). Furthermore, chronic iron treatment resulted in a similar reduction in mitochondrial Ca²⁺ level in cells from both MCU^fl/fl (1.34 ± 0.02) and MCU^fl/fl-MCM mice (1.35 ± 0.03) (Fig. 2I, J).

To evaluate mitochondrial respiratory function, we conducted a mitochondrial stress test in intact cardiomyocytes using the Seahorse XF Analyzer. We found that chronic iron treatment reduced the maximum respiration oxygen consumption rate (OCR in pmol/min) and spare respiratory capacity (SRC) in cardiomyocytes from MCU^fl/fl (Fig. 2K, M, N), while the expression levels of OXPHOS complexes (I–V) were not significantly altered (online Fig. S5). However, the maximum OCR and SRC were not reduced in cardiomyocytes from MCU^fl/fl-MCM (Fig. 2L–N), suggesting that MCU^fl/fl-MCM cardiomyocytes were not susceptible to iron overload-induced mitochondrial energetic perturbation.

Ferroptosis induced by iron overload and the involvement of MCU

It is well accepted that increased cardiac cell death contributes to the development of cardiac dysfunction in a wide range of cardiac diseases [41, 67]. However, as shown in online Fig. S6A–C, we did not detect the presence of apoptosis in either the MCU^fl/fl or the MCU^fl/fl-MCM group after chronic iron administration. Activated (cleaved) caspase 3 has been used as one of the markers for the presence of apoptosis. As shown in online Fig S6, although there was a tendency towards an increase in caspase 3 (35 kDa) level after iron administration in MCUfl/fl (p > 0.05) (online Fig. S6D, E), the active cleaved form (17 kDa) of caspase 3 was not detected in any of the experimental groups (p > 0.05) (online Fig S6D, F). These results indicate that apoptosis is not likely to be involved in the observed cardiac dysfunction.

Since ferroptosis has been related to mitochondrial dysfunction as well as myocardial contractile dysfunction under various pathological conditions [1, 17, 64, 69], we next determined the incidence of ferroptosis under iron overload conditions. We performed the live/dead cell viability assay on cardiomyocytes isolated from MCU^fl/fl (without prior chronic iron injection) to determine the incidence of ferroptosis after acute iron treatment (overnight) and in the presence or absence of other regulators. As shown in Fig. 3A, B, treatment with iron (ferric ammonium citrate, FAC) [16] for ~ 18 h induced cell death in a dose-dependent manner. The value of effective concentration of iron causing 50% ferroptosis (${\mathrm{E}\mathrm{C}}_{50}$) was 2.36 mM with a Hill coefficient of n = 2.0 (Fig. 3B).

Fig. 3.

Ferroptosis induced by Fe overload in mouse ventricular cardiomyocytes. A Representative images of MCU^fl/fl cell death induced by vehicle (Ctl), 1 mM, 2 mM, 5 mM FAC (Fe), and 2 mM Fe + 10 μM Fer-1. Live cells were stained green with calcein AM and nuclei of dead cells were stained red with ethidium homodimer-1 (EthD-1). B Quantitation of cell death showing Fe induced cell death in a dose-dependent manner. The dose-death rate relation was plotted and fit to the Hill $\mathrm{D}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}=\mathrm{B}+(100-\mathrm{B})/\left(1+{\left({\mathrm{E}\mathrm{C}}_{50}/\left[\mathrm{F}\mathrm{e}\right]\right)}^{\mathrm{n}}\right)$, where ${\mathrm{E}\mathrm{C}}_{50}$ (2.36 mM) is the FAC concentration ($\left[\mathrm{F}\mathrm{e}\right]$) that caused 50% cell death, n is the Hill coefficient. C Summary data. D, E Representative images and quantified data of MCU^fl/fl cell death in 2 mM Fe, 2 mM Fe + 20 μM emricasan (EMR), and 2 mM Fe + 30 μM Necrostatin-1 (Nec-1) groups. F Representative images of MCU^fl/fl cell death in 2 mM Fe, 2 mM Fe + 2 mM BPD, 2 mM Fe + 2 mM DFO, 2 mM Fe + 5 μM MitoTEMPO and 2 mM Fe + 20 μM Trolox groups. G, H Quantitation of cell death data. **p < 0.01, compared to the control (Ctl) group. ^##p < 0.01 compared to 2 mM Fe group, one-way ANOVA followed by Tukey test. Cardiomyocytes were isolated from 3 to 4 randomly selected mice in each group and cultured. Numbers on bar graphs indicate the number of fields tested

Fer-1 is a lipid peroxide scavenger and a selective ferroptosis inhibitor. As shown in Fig. 3A, C, 2 mM iron-induced cell death (52.9 ± 4.8%) was prevented by co-treatment with 10 μM Fer-1 (21.1 ± 3.2%). We also tested the effect of the apoptosis inhibitor emricasan (EMR, a pan-caspase inhibitor) as well as the necroptosis inhibitor, necrostatin-1 (Nec-1). As shown in Fig. 3D, E, neither EMR (20 μM) nor Nec-1 (30 μM) prevented iron-induced cell death further supporting the concept that neither apoptosis nor necroptosis play a significant role in the observed cardiac dysfunction. These data were consistent with our findings using the TUNEL assay and cleaved caspase 3 data regarding apoptosis (online Fig. S6).

It has been reported that 2,2′-bipyridyl (BPD) is a powerful iron chelator that can also reduce mitochondrial Fe²⁺ content [5, 25]. We found that 2 mM BPD significantly reduced the incidence of ferroptosis caused by iron treatment (Fig. 3F, G). In addition, the gold standard iron chelator deferoxamine (DFO, 2 mM) also significantly prevented iron overload-induced ferroptosis (Fig. 3F, G). Furthermore, the involvement of mitochondrial ROS was confirmed by the mitochondria-targeted antioxidant MitoTEMPO (5 μM) which effectively reduced the occurrence of iron overload-induced ferroptosis (Fig. 3F, H). The same protective effect was observed when cardiomyocytes were pretreated with another broadly used antioxidant trolox (20 μM) (Fig. 3F, H).

To further determine whether MCU plays an essential role in chronic iron overload-induced cardiac ferroptosis, we conducted additional live/dead cell assays in ventricular cardiomyocytes. As shown in Fig. 4A, B, iron overload-induced ferroptosis was prevented by MCU deficiency. The ferroptotic death rate in cardiomyocytes was 52.9 ± 4.8% in MCU^fl/fl compared to 9.5 ± 1.1% in MCU^fl/fl-MCM after treatment with 2 mM iron (p < 0.01). It is also notable that the baseline death rate (control groups without iron treatment) was also reduced by MCU deficiency (control MCU^fl/fl group: 12.2 ± 1.2% vs. control MCU^fl/fl-MCM group: 5.7 ± 0.8%, p < 0.01). Taken together, these results demonstrate that MCU deficiency is highly protective from iron overload-induced cardiac ferroptosis.

Fig. 4.

Prevention of Fe overload-induced ferroptosis by a deficiency of MCU in mouse ventricular cardiomyocytes. A Representative images of MCU^fl/fl and MCU^fl/fl-MCM cell death induced by vehicle (Ctl) and 2 mM FAC. B Quantitation of cell death data (Note the MCU^fl/fl data are the same as in Fig. 3E). Cardiomyocytes were isolated from 3 to 4 randomly selected mice in each group and cultured. Numbers on bar graphs indicate the number of fields tested. C Representative photos of MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes showing lipid peroxide levels using specific Liperfluo dye after treatment with vehicle (Ctl), 2 mM Fe, and 2 mM Fe + 10 μM Fer-1. D Quantified data of lipid peroxide levels detected by Liperfluo fluorescence. *p < 0.05; **p < 0.01, NS not significant, two-way ANOVA followed by Tukey test. Number of cells from 3 to 4 randomly selected mice studied is indicated in each bar graph

Lipid ROS accumulation is a hallmark of ferroptosis. We therefore determined the level of lipid peroxidation in cultured cardiomyocytes using Liperfluo which interacts directly with lipid hydroperoxides [70]. As shown in Fig. 4C, D, Liperfluo fluorescence was significantly elevated in the iron-treated MCU^fl/fl group, but not in the iron + Fer-1 group. Consistent with the live/dead cell assay result, the overall lipid peroxidation level in MCU^fl/fl-MCM cardiomyocytes was significantly lower compared to MCU^fl/fl cardiomyocytes after iron treatment.

RAS-selective lethal 3 (RSL3), a selective GPX4 inhibitor, has been widely used as a tool to induce ferroptosis in multiple cell types [1, 22]. Our results show that RSL3 efficiently induced ferroptosis in ventricular cardiomyocytes (online Fig. S7). RSL3 at 0.4 μM induced ferroptosis at a death rate of 83.3 ± 4.8% in MCU^fl/fl cardiomyocytes. The ferroptotic death rate was not attenuated in MCU^fl/fl-MCM cardiomyocytes (76.6 ± 5.0%) compared to MCU^fl/fl cardiomyocytes. Therefore, deficiency of MCU rendered no resistance to RSL3-induced ferroptosis contradictory to iron overload-induced ferroptosis. Still, the RSL3-induced death rate was significantly reduced by co-treatment with either Fer-1 (10 μM), DFO (2 mM) or MitoTEMPO (5 μM) in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes (online Fig. S7A–D). These combined findings suggest a novel pathway in cardiomyocytes for the induction and inhibition of iron overload-induced ferroptosis (online Fig. S7E) (see “Discussion”).

The ferroptosis inhibitor Fer-1 abrogates iron overload-induced cardiac dysfunction

To further determine whether ferroptosis is involved in the occurrence of cardiac dysfunction after chronic iron overload in vivo, we determined the effect of a specific ferroptosis inhibitor (i.e. Fer-1) on iron overload-induced cardiac dysfunction. WT mice (C57BL/6J) were randomly divided into three groups and injected (i.p.) either with: (1) vehicle (dextrose), (2) iron dextran, or (3) iron dextran + Fer-1. ECHO was performed to evaluate LV function at both baseline and after 6 weeks of iron or iron + Fer-1 (10 μM) treatment. No difference was observed in cardiac function (LVEF and LVFS) at baseline for any of the experimental groups (online Fig. S8A–C). While cardiac function was significantly reduced in the iron-injected WT group compared to the vehicle group, heart function was maintained in the iron + Fer-1 group (Fig. 5Aa–c) suggesting that the inhibition of ferroptosis may serve as an effective approach to prevent the development of iron overload-induced cardiac dysfunction. In addition, to further connect the observed cardiac dysfunction to lipid peroxidation and ferroptosis, we conducted 4-hydroxynonenal (4-HNE, a cytotoxic aldehyde product of lipid peroxidation) staining. Consistent with live/dead cell assay results, the level of lipid peroxidation (4-HNE intensity) in the left ventricle was significantly increased in the iron-treated group which was reversed in the iron + Fer-1 group (Fig. 5Ba, b). These results confirm that Fer-1 prevents lipid peroxidation and thereby attenuates ferroptosis in the heart (Fig. 5C).

Fig. 5.

Protective effect of Fer-1 on heart dysfunction in chronic in vivo Fe overloaded mice. Aa Representative M-mode ECHO images measured in wild-type (C57BL/6) control (vehicle) group, Fe overload group, and in Fe + Fer-1 group. Ab, c Quantified data of LVEF and LVFS. Number of randomly selected animals studied is indicated in bar graphs. Ba Representative 4-HNE staining images assessing lipid oxidation within left ventricles of MCU^fl/fl hearts after chronic Fe overload and Fe + Fer-1 groups. Bb Quantified data of lipid oxidation (4-HNE staining). Number of left ventricles from randomly selected mice is indicated *p < 0.05; **p < 0.01, NS not significant, one-way ANOVA followed by Tukey test. C Schematic demonstrating that Fer-1 prevents lipid peroxidation and thereby attenuates ferroptosis in the heart

Cytosolic Ca²⁺ and contractile function in ventricular cardiomyocytes from MCU^fl/fl and MCU^fl/fl-MCM after chronic iron overload

Reduced heart function may be related to a loss of cardiomyocytes (possibly induced by ferroptosis) and/or result from a negative inotropic effect due to reduced SR intracellular Ca²⁺ content and Ca²⁺ release [29] and /or changes in myofilament Ca²⁺ sensitivity and responsiveness [30, 61]. Therefore, we determined whether iron overload causes intracellular Ca²⁺ mishandling and thereby contractile dysfunction. We did not observe significant differences in either resting cell lengths (online Fig. S9A) or heart size (heart weight/tibia length) (online Fig. S9B) between the experimental groups. However, consistent with the observed reduction in vivo of contractile function (Fig. 1B, C), we detected a significant reduction in contractility at the level of the cardiomyocyte in MCU^fl/fl mice after chronic iron injection (percentile cell shortening in iron group: 4.06 ± 0.45% vs. vehicle group 7.82 ± 0.50%, p < 0.05). However, ventricular cardiomyocyte contractility from MCU^fl/fl-MCM mice after chronic iron administration was not different from that observed in the vehicle treated group (percentile cell shortening in iron group: 8.80 ± 0.37% vs. vehicle group: 8.65 ± 0.31%, NS) (online Fig. S10A, B). To further evaluate if Ca²⁺ handling might be involved at the level of the cardiomyocyte, we measured Ca²⁺ transients and SR Ca²⁺ content as shown in online Fig. S10C–G. After chronic iron treatment the Ca²⁺ transient amplitude and SR Ca²⁺ content were both significantly reduced in MCU^fl/fl cardiomyocytes (online Fig. S10D, E). In contrast, neither the Ca²⁺ transient amplitude nor SR Ca²⁺ content was changed in MCU^fl/fl-MCM cardiomyocytes after chronic iron administration (online Fig. S10D, E). There was no change in fractional Ca²⁺ release by the SR (online Fig. S10F) and there was no change in the time course of Ca²⁺ release or reuptake (T₅₀) (online Fig. S10G) with chronic iron administration in either the MCU^fl/fl vs. MCU^fl/fl-MCM groups.

We performed Western blot experiments to further delineate the molecular mechanism(s) that might be involved in the observed changes in cytosolic Ca²⁺ handling induced by chronic iron injection in MCU^fl/fl vs. MCU^fl/fl-MCM cardiomyocytes. As shown in online Fig. S11A–D, while no significant changes were observed in the expression level of sarcoplasmic reticulum Ca²⁺ ATPase (SERCA2a) and phospholamban (PLN) in either group, the ratio SERCA2a/PLN was significantly reduced by chronic iron injection in the MCU^fl/fl group, but not in the MCU^fl/fl-MCM group. The expression level of total RyR2 (online Fig. S11I, J) was reduced by chronic iron injection in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes, whereas the ratio of phosphorylated RyR2 at S2814 (CaMKII site) and S2808 (PKA site) were unchanged (online Fig. S11I, K, L). In addition, sarcolemmal DHPR (L-type Ca channel) expression level was decreased in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes after chronic iron injection while sarcolemmal NCX1 levels remained unchanged (online Fig. S11E–G). It is notable that the expression level of Mfrn2, another putative mitochondrial Fe²⁺ transporter, was lower in MCU^fl/fl-MCM than in MCU^fl/fl mice and was further reduced by chronic iron injection in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes (online Fig. S11E, H).

Whole-mitoplast patch clamp current recording with Ca²⁺ or Fe²⁺ as a charge carrier

To test whether MCU is able to directly transport Fe²⁺ into mitochondria as it does with Ca²⁺, we carried out whole-mitoplast patch clamp experiments (online Fig. S12A, B) [24]. Although we observed a significant MCU-mediated inward Ca²⁺ current, we did not detect any inward current carried by Fe²⁺ when Ca²⁺ (1 mM) was replaced with Fe²⁺ (2 mM FeCl₂ + 5 mM ascorbate) as the charge carrier. Importantly, the Ca²⁺ current was reduced to half of its prior level even when Fe²⁺ was washed-out from the bath suggesting that Fe²⁺ binds tightly to the MCU pore and dissociates very slowly. Perhaps under our experimental conditions, Fe²⁺ permeation via MCU is too slow to give detectable currents (coulombs per second) via the direct patch clamp methodology or MCU may not be a direct Fe²⁺ transporter.

Discussion

It has previously been demonstrated that MCU is involved in mitochondrial Ca²⁺ uptake and provides needed energetic support for the heart during acute fight-or-flight responses. Furthermore, it has been shown that MCU mediates mitochondrial Ca²⁺ overload and causes cell death during ischemic cardiac injury. However, the role of MCU in mitochondrial iron load and cardiac toxicity is not well understood. To determine the potential pivotal role of MCU in mitochondrial iron load under iron loading conditions to ferroptosis and cardiac dysfunction, we carried out both single cell (in vitro) and whole animal (in vivo) experiments on MCU^fl/fl and MCU^fl/fl-MCM mice [44]. We demonstrate for the first time a close link between MCU-dependent mitochondrial iron load, ferroptosis, and iron associated cardiac dysfunction. We show that MCU deficiency prevents the development of iron overload-induced cardiac dysfunction by reducing the incidence of ferroptosis, thereby maintaining cell contractility and heart function.

The essential role of MCU in iron overload-induced ferroptosis

Ferroptosis is a recently proposed form of regulated cell death which is distinct from apoptosis, necrosis, and other modes of cell death. It is iron-dependent and characterized by the accumulation of oxidized membrane lipids [14, 38, 69]. Although the term “ferroptosis” was coined by Dixon et al. in 2012 [14], ferroptosis-like cell death had been observed in earlier studies. For example, Dolma et al. found that erastin, a small molecular antitumor agent, initiated a non-apoptotic form of cell death in 2003 [15]. Later, RSL3 was identified and was shown to activate the same pattern of cell death, which was shown to be dependent on iron [71], glutathione (GSH) synthesis and redox state [2], as well as GPX4 and membrane lipid metabolism [59]. Based on these findings, ferroptosis has been defined as a new form of cell death that depends on iron-dependent lipid peroxidation [14]. More details regarding the development of the concept and molecular mechanisms of ferroptosis can be found in recent comprehensive review articles [34, 38, 64, 69].

Since it was reported by Dixon et al. in 2012, ferroptosis has been intensely studied in the field of oncology and neuro-degenerative diseases (e.g. Alzheimer’s and Parkinson’s disease). In recent emerging studies, ferroptosis has also been implicated in iron and doxorubicin-induced cardiotoxicity as well as in ischemia–reperfusion injury [1, 17, 19, 69]. However, the detailed underlying mechanisms for ferroptosis in the heart, specifically how iron is involved in the ferroptosis process, is still not well known. Answers to this question will provide insights into the mechanism(s) of ferroptosis and associated human diseases as well as potentially direct new therapeutic strategies for such diseases. Our study points to MCU playing an important mechanistic role in the ferroptosis process in the heart. Our studies also indicate that there are possibly yet to be identified pathways other than those reported to be important in cancer and other cell types that might be involved in diseases impacting the heart.

Currently there is no available assay to directly evaluate ferroptosis at the tissue level. The presence of ferroptosis at the cellular level can be confirmed by determining whether cell death is prevented by the selective ferroptosis inhibitor Fer-1 (a lipid peroxide or ROS scavenger) and by evaluating levels of lipid peroxides. Our studies using the live/dead cell viability assay have clearly demonstrated that ferroptosis is induced by iron in isolated cardiomyocytes. We confirmed that iron induced ferroptosis in a dose-dependent manner. An increased level of lipid oxidation, which is a hallmark of ferroptosis, was confirmed by both 4-HNE and Liperfluo staining. Iron overload-induced cell death was prevented by the selective ferroptosis inhibitor Fer-1 in vitro and in vivo indicating that ferroptosis rather than other types of cell death (e.g. apoptosis or necroptosis) are involved as previously reported by Das et al. [9]. Nevertheless, some studies have reported that the level of apoptosis markers is increased in iron overloaded hearts [65].

We found that administration of Fer-1 maintained heart function when administered with iron further indicating that iron overload-induced cardiac dysfunction is likely to be mediated at least in part by ferroptosis. The iron dependency of ferroptosis was tested by evaluating the impact of DFO on ferroptosis, as it remains the gold standard for clinical iron chelation [33, 56]. In addition, we also tested the effect of another iron chelator BPD, which has a high membrane permeability and thus is able to access mitochondria and effectively chelate labile iron [5, 12, 25, 55]. Both BPD and DFO efficiently reduced ferroptosis cell death further indicating that mitochondrial iron overload contributes to ferroptosis.

Most importantly, we have demonstrated for the first time that a deficiency of MCU dramatically ameliorates mitochondrial dysfunction indicating that MCU plays an essential role in mitochondrial iron uptake and iron overload-induced ferroptosis. Furthermore, our results indicate that mitochondrial iron loading most likely results in ferroptosis via ROS and lipid oxidation, which is consistent with our observation that ferroptosis was prevented by various antioxidants such as Trolox and MitoTEMPO (mitochondria targeting) as well as Fer-1 (lipophilic radical trapping).

Ferroptosis can be promoted by various inducers that act through different pathways [39, 65, 69]. The predominant pathways leading to ferroptosis in cardiomyocytes, especially under pathophysiological conditions, have not been determined. It is well accepted that GPX4 reduces the level of lipid peroxidation thereby protecting cell membranes from ferroptosis. RSL3, a selective GPX4 inhibitor, induced ferroptosis in cardiomyocytes from WT hearts confirming previous findings by Baba et al. [1]. However, we found that RSL3-induced ferroptosis showed the same properties in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes (online Fig. S7A–D). RSL3-induced ferroptosis was not attenuated in MCU^fl/fl-MCM cardiomyocytes, while it was still prevented by the iron chelator DFO in both MCU^fl/fl and MCU^fl/fl-MCM cardiomyocytes (online Fig. S7A–D). These results can be explained by the schematic in online Fig. S7E that demonstrates that GPX4 inhibition by RSL3 increases the lipid peroxidation pathway bypassing the involvement of mitochondria or MCU. Regulation of ferroptosis in a GPX4-independent manner has also been demonstrated in recent studies [7, 57]. The iron-dependence of RSL3-induced ferroptosis is likely mediated by lipid peroxidation reactions catalyzed by lipoxygenases (LOX, an iron-dependent enzyme) and Fe²⁺ (lipid Fenton reaction) (as indicated in online Fig. S7E by circles numbered 2 and 3), while direct iron administration may trigger all iron-dependent steps including mCU dependent mitochondrial iron load (circled 1 in online Fig. S7E). Therefore, under iron overload conditions, MCU and mitochondrial iron likely play a more essential role in ferroptosis, which is not the case in ferroptosis induced by direct inhibition of the GPX4 pathway. Our results suggest that MCU activity, mitochondrial vs. cytosolic iron, ROS, and lipid oxidation may play different roles in cardiac ferroptosis depending on the different triggers and pathways involved (online Fig. S7E).

How is MCU involved in mitochondrial Fe²⁺ transport?

A remaining question is how iron is transported into mitochondria. Previous studies have suggested that Mfrn2 may function as a mitochondrial Fe²⁺ transporter [6]. However, a very recent report has suggested that it is MCU that is responsible for mitochondrial Fe²⁺ uptake, which is regulated by Mfrn2 [53]. Our present results (Fig. 2A, B and online Fig. S4) also demonstrated the essential MCU dependency of mitochondrial Fe²⁺ uptake as the mitochondrial Fe²⁺ content is significantly higher in MCU^fl/fl compared to MCU^fl/fl-MCM cardiomyocytes in both chronic and acute iron overload models. However, whole-mitoplast patch clamp recordings did not show an inward current carried by Fe²⁺ (online Fig. S12) suggesting that MCU is not likely a direct Fe²⁺ transporter or Fe²⁺ permeation via MCU is too slow to give detectable currents. The potent block of Ca²⁺ current even after Fe²⁺ washout indicates that Fe²⁺ does not enter, but instead appears to bind tightly to the MCU pore and in turn inhibits Ca²⁺ entry and that the release of Fe²⁺ to the matrix-side is minimal or is too slow to be detected using the patch clamp method. Alternatively, Fe²⁺/Fe³⁺ may be transported into mitochondria in the form of a neutralized complex. The transporting process may be regulated by the interaction between MCU and Mfrn2 [13, 53]. In this regard, we observed that the expression level of Mfrn2 was lower in MCU^fl/fl-MCM than in MCU^fl/fl cardiomyocytes and that it was further reduced by chronic iron injection in both cardiomyocytes isolated from MCU^fl/fl and MCU^fl/fl-MCM hearts.

Mitochondrial dysfunction after chronic iron injection: mediated by iron effects or altered Ca²⁺ homeostasis in mitochondria?

In our previous study [25], we have demonstrated the role of the mitochondrial permeability transition pore (mPTP) in Ca²⁺ dysfunction and arrhythmogenesis in cardiomyocytes after acute iron overload. We have found that acute iron overload induces mitochondrial ROS generation and $\mathrm{\Delta }{\psi }_m$ depolarization, thus opening the mPTP and promoting calcium waves and cardiac arrhythmias. Conversely, the inhibition of mPTP ameliorates the proarrhythmic effects of iron overload.

In the present study, we have found mitochondrial dysfunction such as depolarized $\Delta {\psi }_m$, increased ROS, and reduced SRC in MCU^fl/fl cardiomyocytes from mice after receiving chronic iron injection, while mitochondrial function remained normal and protected from ferroptosis in MCU^fl/fl-MCM cardiomyocytes. Since MCU is primarily a uniporter for Ca²⁺, a question that should be considered is whether cardioprotection results from direct Fe²⁺ transport effects or is due to alterations in mitochondrial Ca²⁺ homeostasis. In this regard, we found the same reduction in basal mitochondrial Ca²⁺ levels in MCU^fl/fl vs. MCU^fl/fl-MCM cardiomyocytes with or without chronic iron injection, suggesting that mitochondrial Fe²⁺ rather than Ca²⁺ is critical for iron overload-induced mitochondrial dysfunction and ferroptosis.

Histological alterations, single cardiomyocyte contractility, and cardiac dysfunction in mice with iron overload

Chronic iron loading resulted in a decrease in cardiac contractility in MCU^fl/fl mice, despite both groups (i.e. MCU^fl/fl and MCU^fl/fl-MCM) having the same level of cardiac iron deposition and fibrosis suggesting that iron tissue load and/or fibrosis are not sufficient to cause cardiac dysfunction. However, a significantly reduced SR Ca²⁺ release as reflected by a reduced amplitude of the Ca²⁺ transient associated with reduced SR Ca²⁺ content could contribute to the observed reduced contractility seen at the level of the cardiomyocyte as well as the whole heart. These functional data were consistent with the Western blot data showing a lower SERCA2a/PLN ratio, which is considered to be an index of SR Ca²⁺ uptake and content. The reduced expression of total RyR2, DHPR, and Mfrn2 may be due to compensatory effects.

The mitochondria-specific changes, i.e. depolarized $\Delta \psi$, increased ROS, and reduced respiratory function may also contribute to reduced single cell contractility (e.g. changes in myofilament Ca²⁺ responsiveness [32] and reduced energy reserves [46]) in MCU^fl/fl cardiomyocytes, but not in cardiomyocytes from MCU^fl/fl-MCM hearts. The observed iron overload-induced cardiac dysfunction in conjunction with oxidative stress, mitochondrial dysfunction, and the susceptibility to arrhythmogenesis are consistent with our recent studies in the setting of acute iron treatment [25, 26, 62]. The concerted effects of iron overload on cardiac dysfunction via induction of ferroptosis and negative inotropy are illustrated in Fig. 6.

Fig. 6.

A schematic presenting the concerted effects of iron overload in inducing cardiac dysfunction via ferroptosis, negative inotropy, and the involvement of MCU. MCU mitochondrial Ca²⁺ uniporter, ROS reactive oxygen species, $\Delta {\psi }_m$ mitochondrial membrane potential, OCR oxygen consumption rate, SRC spare respiratory capacity, CaT Ca²⁺ transient

Conclusions and perspectives

Our results indicate for the first time that MCU activity is involved in mitochondrial Fe²⁺ loading and leads to ferroptosis and cardiac dysfunction. It appears that ferroptosis sits at the nexus of linking a toxic and deadly partnership between Fe²⁺ and Ca²⁺ resulting in contractile dysfunction. The inhibition of ferroptosis may serve as an effective way to prevent the development of cardiac dysfunction.

It has recently been suggested that mitochondrial iron accumulation and probably ferroptosis mediate cardiac damage during ischemia–reperfusion (I/R) injury [5, 19, 47] making iron an important pathophysiological factor that warrants further study. It should be noted that the role of MCU in ischemia–reperfusion (I/R) injury is still under debate [23, 48]. It has been shown that conditional cardiomyocyte-specific MCU KO mouse hearts (used in our present study) are protected from acute ischemia–reperfusion (I/R) injury [44] suggesting that MCU may be linked to acute Ca²⁺ and Fe²⁺ uptake and mitochondrial permeability transition (mPT)-driven cell damage. However, a lack of cardiac protection has been reported during global I/R injury in global germline MCU KO mice [54]. One explanation for this discrepancy might be that the gene deletion in germline knockout mice is chronic and may induce cellular adaptations in the chronic absence of MCU. Another explanation might be that during global ischemia there is a significant change in inorganic phosphate concentration and pH [32]. Both of these factors are known to impact myofilament calcium responsiveness [29, 30, 32] resulting in changes in myocardial contractility. The effects of changes in pH and P_i during hypoxia on myocardial contractility have been previously demonstrated [32, 50].

Another important finding of our present study is that iron treatment significantly disrupted mitochondrial respiratory function by reducing maximum OCR and SRC in MCU^fl/fl cardiomyocytes. However, cardiomyocytes from MCU^fl/flMCM hearts showed resistance to iron overload-induced dysfunction in mitochondrial energetics. Recently ferroptosis has also been implicated in doxorubicin-induced cardiotoxicity pointing to the important role played by iron homeostasis in pathophysiological states in the heart. The protective effect of MCU deficiency against cardiac iron dysregulation may be attributed to its anti-ferroptosis feature and maintenance of myocardial energetics and energy reserves.

Study limitations

We realize that our current mouse model of iron overload does not completely mimic the clinical manifestations of hemochromatosis-associated cardiomyopathy e.g. an EF less than 40% and hepatic cirrhosis-related right sided heart failure. However, a significant reduction in LVEF was apparent as shown in Fig. 1B making the model useful for both in vivo and in vitro studies. Further studies in human cardiomyocytes from patients with iron overload cardiomyopathy are warranted to demonstrate potential clinical relevance of the current study.

Ru360 has been proven to be useful in isolated mitochondria, mitoplasts, and permeabilized myocyte studies (e.g. online Fig. S4C, D), however, its application in vivo and in intact cardiomyocytes is limited due to poor membrane permeability and off-target effects [4, 8, 40, 68]. For these reasons, the effect of Ru360 on iron overload-induced ferroptosis could not be reliably evaluated using the live/dead cell viability assay with an intact plasma membrane. However, cardiomyocytes isolated from MCU KO mice provided an ideal and selective approach for assessing the role of MCU in ferroptosis. It has also been reported that Ru360 may also reduce Mfrn2-dependent mitochondrial iron uptake [36]. This is consistent with our postulation that mitochondrial iron load is MCU-dependent and may also involve other potential iron transporting mechanisms such as MCU regulation by Mfrn-2 resulting in the formation of neutralized ionic complexes.

Abbreviations

ECG

Electrocardiography

ECHO

Echocardiography

Fer-1

Ferrostatin-1

GPX4

Glutathione peroxidase 4

LVEF

Left ventricular ejection fraction

LVFS

Left ventricular fractional shortening

LV

Left ventricle

MCU

Mitochondrial calcium uniporter

OCR

Oxygen consumption rate

SRC

Spare respiratory capacity

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and supplementary materials. Raw data can be made available on reasonable request.