Iron deficiency is related to lower muscle mass in community‐dwelling individuals and impairs myoblast proliferation

Abstract Background Loss of muscle mass is linked with impaired quality of life and an increased risk of morbidity and premature mortality. Iron is essential for cellular processes such as energy metabolism, nucleotide synthesis and numerous enzymatic reactions. As the effects of iron deficiency (ID) on muscle mass and function are largely unknown, we aimed to assess the relation between ID and muscle mass in a large population‐based cohort, and subsequently studied effects of ID on cultured skeletal myoblasts and differentiated myocytes. Methods In a population‐based cohort of 8592 adults, iron status was assessed by plasma ferritin and transferrin saturation, and muscle mass was estimated using 24‐h urinary creatinine excretion rate (CER). The relationships of ferritin and transferrin saturation with CER were assessed by multivariable logistic regression. Furthermore, mouse C2C12 skeletal myoblasts and differentiated myocytes were subjected to deferoxamine with or without ferric citrate. Myoblast proliferation was measured with a colorimetric 5‐bromo‐2′‐deoxy‐uridine ELISA assay. Myocyte differentiation was assessed using Myh7‐stainings. Myocyte energy metabolism, oxygen consumption rate and extracellular acidification rate were assessed using Seahorse mitochondrial flux analysis, and apoptosis rate with fluorescence‐activated cell sorting. RNA sequencing (RNAseq) was used to identify ID‐related gene and pathway enrichment in myoblasts and myocytes. Results Participants in the lowest age‐ and sex‐specific quintile of plasma ferritin (OR vs middle quintile 1.62, 95% CI 1.25–2.10, P < 0.001) or transferrin saturation (OR 1.34, 95% CI 1.03–1.75, P = 0.03) had a significantly higher risk of being in the lowest age‐ and sex‐specific quintile of CER, independent of body mass index, estimated GFR, haemoglobin, hs‐CRP, urinary urea excretion, alcohol consumption and smoking status. In C2C12 myoblasts, deferoxamine‐induced ID reduced myoblast proliferation rate (P‐trend <0.001) but did not affect differentiation. In myocytes, deferoxamine reduced myoglobin protein expression (−52%, P < 0.001) and tended to reduce mitochondrial oxygen consumption capacity (−28%, P = 0.10). Deferoxamine induced gene expression of cellular atrophy markers Trim63 (+20%, P = 0.002) and Fbxo32 (+27%, P = 0.048), which was reversed by ferric citrate (−31%, P = 0.04 and −26%, P = 0.004, respectively). RNAseq indicated that both in myoblasts and myocytes, ID predominantly affected genes involved in glycolytic energy metabolism, cell cycle regulation and apoptosis; co‐treatment with ferric citrate reversed these effects. Conclusions In population‐dwelling individuals, ID is related to lower muscle mass, independent of haemoglobin levels and potential confounders. ID impaired myoblast proliferation and aerobic glycolytic capacity, and induced markers of myocyte atrophy and apoptosis. These findings suggest that ID contributes to loss of muscle mass.


Introduction
Skeletal muscle mass is a key determinant of physical condition and mobility. 1 Reduced muscle mass, or sarcopenia, is strongly linked with impaired quality of life and premature mortality, fuelling the need for modifiable risk factors. 1,2 Several factors may promote sarcopenia, including physical inactivity, reduced mobility and poor physical endurance. 1,2 Iron deficiency (ID) may adversely influence muscle strength, exercise capacity and endurance in chronically ill individuals and the elderly, but so far, it is unknown whether there is a relationship between ID and muscle mass in the general population.
ID is among the most common mineral deficiencies. The largest amount of iron in the body is used for haemoglobin production in the bone marrow and ID is a common cause of anaemia. In recent years, there has been increasing interest in implications of ID beyond erythropoiesis. Iron plays a role in deoxyribonucleic acid (DNA) synthesis and is essential to the function of numerous enzymes. Moreover, iron is a key component of myoglobin, responsible for oxygen storage in muscle cells. Finally, iron is crucial for mitochondrial function and aerobic metabolism, being involved in the citric acid cycle as well as the electron transport chain. 3,4 The relationship between ID and muscle cell function has been studied in cardiomyocytes and skeletal muscle cells. In cardiomyocytes, ID impairs oxidative phosphorylation and contractile function. 5,6 ID reduced left ventricular function in rodents. 7,8 In heart failure patients, ID has been associated with worse skeletal muscle strength and increased lactate production after exercise, while iron supplementation improves aerobic glycolysis. [9][10][11] Also, in skeletal muscle cells, ID may influence energy metabolism. [12][13][14] In humans, anaemia has been linked with reduced muscle mass, 15 but data about the association between ID and muscle mass are scarce. Moreover, little is known about potential mechanisms by which ID might be related to reduced muscle mass.
Therefore, in the current study, we addressed whether ID is related to reduced muscle mass in the general population. Moreover, we investigated whether ID influences proliferation and differentiation rate, aerobic metabolism and cell viability in cultured mouse C2C12 myoblasts and differentiated myocytes.

Iron deficiency and muscle mass in community-dwelling individuals
We used cross-sectional data from the Prevention of Renal and Vascular End-stage Disease (PREVEND) study, a prospective, population-based cohort of Dutch community-dwelling individuals aged 25 to 75 years. The study was approved by the local ethics committee and complied with the principles of the Declaration of Helsinki. A full description of the methods, including extensive description of the statistical methods, is provided in the supplemental material. Written informed consent was obtained from all participants before enrolment. In this study, 8592 participants were enrolled at baseline. For the current analysis, we excluded participants with missing data on urinary creatinine excretion rate (CER) or ferritin ( Figure S1). All participants were asked to collect two consecutive 24-h urine samples before their visit to the clinic, of which the average was calculated. To limit potential 24-h urine collection errors, we excluded participants at the lowest or highest 2.5% of difference between expected and measured 24-h urine volume, 16 leaving 5571 participants available for analysis. Fasting blood samples and 24-h urine samples were collected at the same study visit. Transferrin saturation (TSAT) was calculated as 100 × plasma iron (μmol/L)/(plasma transferrin (g/L) × 25). The average CER of the two 24-h urine samples was used as a parameter for muscle mass. CER was divided into quintiles stratified for age and sex, two major factors influencing muscle mass. The odds ratio of being in the lowest quintile of age-and sex-stratified CER was compared between participants across quintiles of age-and sex-stratified ferritin and TSAT levels, using logistic regression (model 1). Upon multivariable analysis (model 2), we adjusted for estimated glomerular filtration rate (eGFR), body mass index (BMI), high-sensitive C-reactive protein (hs-CRP), 24-h urinary urea excretion (reflecting protein intake), alcohol consumption and smoking status. In model 3, we additionally adjusted for plasma haemoglobin (Hb). In sensitivity analyses, we substituted CER for length-indexed CER 17 and length^2-indexed CER, 18 given that muscle mass depends on body size. Furthermore, we repeated the analyses after exclusion of the 5% most extreme ferritin or TSAT values by calculating absolute differences from the median.

Cell culture
Mouse C2C12 skeletal myoblasts (ATCC, Cat. no. CRL-1772) were cultured in Dulbecco's Modified Eagle Medium (DMEM, Gibco Cat. No. 41966-029) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich Cat. No. F7524) and 1% penicillin-streptomycin (ps; 5000 U/mL penicillin, 5000 μg/mL streptomycin, Gibco) using T25 flasks (Thermo Scientific Cat. No. 156367). Cells were incubated at 37°C, 5% CO 2 (Thermo Scientific Forma Series II Water Jacket CO2 incubator) and 100% humidity. Sub-culturing and refreshing medium was performed three times per week. For sub-culturing, cells were washed with phosphate buffered saline (PBS; Biowhittaker Cat. No. 15-512Q) before trypsinization with 0.05% Trypsin-EDTA (Thermo Scientific, Cat. No. 15400-054). Collected cells were then diluted for sub-culturing or counted for seeding on plate. Counting of living cells was done using a trypan blue staining and a Bürker cell counting chamber (Marienfeld-superior Cat. No. 0640211). For the differentiation from myoblasts to myocytes, cell culture was grown to a confluence of >70% in cell culture plates before switching from culture medium to differentiation medium (DMEM, 2.5%; horse serum (HS, Sigma-Aldrich; Merck KgaA Cat. No. H1138), 1% ps). After seven days, this resulted in the gene expression of differentiation markers (Myh7, Myod, Myog, Figure S2) and formation of skeletal myotubes across the plate and experimental intervention was initiated for up to three days.

Induction of iron deficiency and iron repletion
To induce ID during cell culture, deferoxamine (DFO; Sigma-Aldrich Cat. No. D9533) was added to the medium. Unless stated otherwise, a DFO concentration of 7.5 μM was used. Repletion of iron to the cells was achieved by supplementing medium containing DFO with ferric citrate (FC; Sigma-Aldrich Cat. No. F3388). Unless stated otherwise, a FC concentration of 10 μM was used. The cells were typically incubated with or without DFO and FC for three days, except for production of quantitative polymerase chain reaction (qPCR) samples for which they were incubated one day.

Assays
Experimental methods are described in brief below. A more extensive description is provided in the Supplemental material.
Differentiation rate of C2C12 myoblasts was determined using staining techniques. The fusion index, or the percentage of nuclei being in a differentiated cell, was counted.
Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of skeletal myocytes were measured using a Seahorse Mito Stress test. During the assay, three baseline measurements were taken, after which ATP synthase inhibitor Oligomycin A was injected into the wells to inhibit complex I of the oxidative phosphorylation process. Then, carbonyl cyanide 4-(trifluoromethoxy)-phenylhydrazone (FCCP) was injected to uncouple the oxidative phosphorylation and induce rapid oxidation of energy substrates. All data were acquired using a Seahorse XF24 or XF96 analyser and normalized for total protein in the well. To assess ATP-linked respiration, the decrease in oxygen consumption rate after the injection of oligomycin A was calculated (OCR basal À OCRoligomycin ). To assess respiratory reserve of the cells, the increase in oxygen consumption after injection of FCCP compared with baseline was calculated (OCR FCCP À OCR basal ).
For western blotting, C2C12 cell lysate was collected and total protein concentration measured using the Bicinchoninic acid (BCA) protein assay (ThermoScientific, Cat. No. 23227). Using normalized protein amounts, ferritin heavy chain I (Fth) and myoglobin bands were visualized and analysed after normalization for β-actin.
For quantitative real-time polymerase chain reaction (PCR) and for ribonucleic acid (RNA) sequencing, total RNA was isolated from C2C12 myoblasts and myocytes. Quantitative PCR (qPCR) technology was then performed by mixing collected cDNA with primers for target genes Fbxo32, Trim63, Becn1, TfR, Slc39a14, Slc40a1, Myh7, Myod, Myog, Bax, Bcl2, Casp9 and Rplp0 (Table S1). For RNA sequencing, RNA samples were sent to Single Cell Discoveries (Utrecht, the Netherlands) for transcriptome analysis. Samples were multiplexed into a sequencing library and sequenced. Total RNA concentration was measured and normalized to 20 ng/μL. Normalized total RNA (with RNA integrity number (RIN) scores >7) was used for library preparation and sequencing.
To measure apoptosis rate in differentiated C2C12 myocytes, fluorescence-activated cell sorting (FACS) was used. Cells were stained with 5 μL FITC Annexin-V and 5 μL propidium ionide staining solution to distinguish early apoptotic cells and late apoptotic cells from living cells.
Data analyses were performed with Prism 8.4.2 (GraphPad Software). For all assessments, experimental groups consisted of at least three biological replicates and all analyses were performed in multiplo. Data are presented as mean ± standard error of the mean. To test for normality of the distribution, a Shapiro-Wilk test was used. Differences between two groups were analysed by an independent samples t-test in case of a normal distribution or a two-sided Mann-Whitney U-test in case of a skewed distribution. Trends in concentration series were analysed with the ANOVA test followed by a Tukey's multiple comparisons test in case of a normal distribution or a Kruskal-Wallis test followed by Dunn's multiple comparisons test in case of a non-normal distribution. A P-value of <0.05 was considered statistically significant. In column scatter plots, unless stated otherwise, the median and individual values are presented.

Iron deficiency is associated with lower muscle mass in community-dwelling individuals
We first studied the association between ID and lower muscle mass in a cohort of 5571 individuals (51% men, median age 52 (44-63) years, eGFR 92 ± 17 mL/min/1.73 m 2 , plasma Hb 13.7 ± 1.2 g/dL and plasma ferritin 97 (48-173) μg/L). Further Iron deficiency, myoblast proliferation and muscle mass participant characteristics are provided in Table 1. Individuals in the lowest age-and sex-stratified quintile of plasma ferritin level had a significantly higher risk of having low muscle mass, defined as the lowest age-and sex-specific quintile of CER, compared with the middle quintile ( Figure 1). The association remained significant upon adjustment for BMI, eGFR, hs-CRP, urinary urea excretion, alcohol consumption and smoking status (Model 2, OR 1.46, 95% CI 1.13-1.88, P = 0.004), and upon further adjustment for plasma haemoglobin (Model 3, OR 1.62, 95% CI 1.25-2.10, P < 0.001). Results were similar in secondary analyses where CER was indexed for length (fully adjusted OR 1.48, 95% CI 1.15-1.92, P = 0.003) or length 2 (fully adjusted OR 1.76, 95% CI 1.36-2.28, P < 0.001, Figure S3). Excluding outliers of ferritin also did not materially change the results ( Figure S4). In addition, participants in the highest age-and sex-stratified quintile of plasma ferritin level had a higher risk of being in the lowest age-and sex-specific quintile of CER (fully adjusted OR 1.34, 95% CI 1.03-1.75, P = 0.03, Figure 1).
Participants in the lowest versus the median age-and sexstratified quintile of plasma TSAT also had a higher risk of being in the lowest age-and sex-specific quintile of CER after multivariable adjustment (Model 3: OR 1.34, 95% CI 1.03-1.75, P = 0.03, Figure 1). Results were similar if CER was indexed for length (fully adjusted OR 1.34, 95% CI 1.02-1.74, P = 0.03), and slightly weaker, but with a similar trend for CER indexed for length 2 ( Figure S3). The risk of lower CER for those in the lowest age-and sex-specific quintile of TSAT was attenuated after exclusion of outliers ( Figure S4). There was a non-significant trend suggesting a higher risk of being in the lowest age-and sex-specific quintile of CER in participants with a higher TSAT (Figure 1), which was not consistent in secondary analyses ( Figures S3 and S4).

Deferoxamine induces iron deficiency and reduces myoblast proliferation but does not affect differentiation to myocytes
Incubation of C2C12 skeletal myoblasts and myocytes with DFO for three days resulted in dose-dependently reduced concentrations of Fth ( Figure 2A, Figure S5A). Co-incubation of myoblasts and myocytes with increasing concentrations of FC in addition to 7.5 μM DFO dose-dependently restored concentrations of Fth ( Figure 2B, Figure S5B). Gene expression of Transferrin receptor (TfR), involved in cellular uptake of transferrin-bound iron, increased with 28% by 7.5 μM DFO treatment of myocytes for one day (P = 0.002) and decreased by 78% upon co-treatment with 10 μM FC (P = 0.002, Figure 2C). Expression of Slc39114, encoding the iron uptake marker ZIP14, was not affected by treatment with DFO or FC ( Figure 2D). Expression of Slc40a1, encoding ferroportin, a cellular iron export protein, was reduced after DFO treatment (P = 0.04, Figure 2E), which was reversed by co-treatment with FC ( Figure 2E). DFO or FC did not affect cell viability ( Figure 2F). In C2C12 myoblasts, DFO dose-dependently reduced the proliferation rate (P-trend <0.001, Figure S6). Co-incubation with 10 μM FC fully restored proliferation rate (P < 0.001, Figure 3A). DFO treatment with or without FC during differentiation of myocytes did not affect expression of differentiation markers Myh7 (encoding myosin heavy chain), Myod (encoding myoblast determination protein 1) and Myog (encoding myogenin) ( Figures 3B-D) or the fusion index ( Figure 3E).

Iron deficiency leads to mitochondrial dysfunction and reduces myoglobin content in skeletal muscle cells
Basal OCR was slightly and non-significantly lower upon incubation with 7.5 μM DFO ( Figure 4A). During a mitochondrial stress test, DFO treatment resulted in a small and borderline significant decline (À9%, P = 0.07) in ATP-synthase linked respiration reflected by the difference between basal respiration rate and remaining respiration rate after oligomycin injection (OCR basal À OCR oligomycin ), which was not restored by co-treatment with 10 μM FC ( Figure 4B). Subsequently, FCCP was injected to assess respiratory reserve, reflected by the increase in oxygen consumption rate relative to baseline (OCR-FCCP À OCR basal ). Myocytes treated with DFO tended to have a lower maximal mitochondrial capacity, compared with untreated myocytes (À28%, P = 0.10, Figure 4C). Additional treatment with FC restored respiratory reserve. ECAR during basal respiration, ATP-linked respiration and maximal respiration, reflecting lactate metabolism, increased after addition of DFO and decreased after additional treatment with FC, although these differences did not reach statistical significance ( Figure 4D). We next investigated the effects of ID on oxygen storage and mitochondrial function in differentiated C2C12 myocytes. ID induction reduced myoglobin concentration in C2C12 myocytes (À52%, P < 0.001); iron repletion with FC did not restore myoglobin levels ( Figure 4E).

Figure 1
Association between iron status as reflected by age-and sex-specific quintiles of ferritin levels (A) or TSAT (B) and creatinine excretion rate (CER) reflecting muscle mass in community-dwelling individuals. Odds ratios and corresponding 95% confidence intervals are provided for the risk of being in the lowest age-and sex-specific quintile of 24-h CER in a crude model (model 1), a multivariable model, adjusted for BMI, eGFR, hs-CRP, urinary urea excretion, alcohol consumption and smoking status (model 2) and with additional adjustment for haemoglobin (model 3). BMI, body mass index; CER, creatinine excretion rate; eGFR, estimated glomerular filtration rate; hs-CRP, high sensitive C-reactive protein; OR, odds ratio; TSAT, transferrin saturation.
Iron deficiency, myoblast proliferation and muscle mass

Iron deficiency affects key pathways of energy and nucleotide metabolism
To explore potential mechanisms explaining impaired proliferation, mitochondrial respiration and increased apoptosis, we profiled transcriptomic changes in C2C12 myoblasts and myocytes after treatment with DFO for one day with or without 10 μM FC. RNA sequencing revealed 110 differentially-expressed genes (DEGs) in myoblasts compared with 270 DEGs in myocytes; 65 DEGs overlapped between the cell types. In myoblasts, expression of 96 genes was significantly up-regulated after treatment with DFO and restored after co-treatment with FC, while 14 genes showed the opposite pattern. In myocytes, 104 genes were up-regulated after treatment with DFO and restored after additional treatment with FC, and 145 genes showed the opposite pattern ( Figure 6A-D). Table 2 shows genes with most significant differences upon DFO treatment with or without FC.
Next, we assessed whether alterations in genes related to energy metabolism could explain reduced mitochondrial function in iron-deficient myocytes. Expression of genes encoding all major enzymes involved in the anaerobic steps of glucose metabolism, glycolysis and lactate production was up-regulated in myocytes after treatment with DFO ( Figure 6E). On the contrary, apart from two subunits of  Iron deficiency, myoblast proliferation and muscle mass complex I of the electron transport chain, none of the genes encoding enzymes involved in the mitochondrial steps of energy metabolism and in fatty acid oxidation (both oxygen-dependent) were affected ( Figure 6E).
Although RNA sequencing analysis did not confirm significantly altered expression of Fbxo32, Trim63, Becn1 or Bax, we did observe up-regulated expression of pro-apoptotic genes Ddit4, Bnip3, Bnip3L, Fam162a and Ero1L in iron-deficient myoblasts and myocytes. Pathway enrichment analysis showed that in myoblasts, ID most strongly affected energy metabolism, nucleotide metabolism and hypoxic responses ( Figure S8a). In differentiated myocytes, nucleotide metabolism, cell cycle regulation and hypoxic responses were most affected ( Figure S8b). Computational analysis of literature-based putative transcription factors showed that transcription factors that best predicted clusters of identified DEGs after ID induction were Hif1A, encoding hypoxia inducible factor subunit 1-alpha, in myoblasts and Foxm1, a gene involved in cell proliferation, in myocytes (Table S2).

Discussion
To our knowledge, this is the first study to reveal a relation between ID and reduced muscle mass in the general population, independent of haemoglobin levels and potential confounders. In cultured skeletal muscle cells, we found that ID strongly reduces proliferation in myoblasts and adversely impacts pathways related to energy and nucleotide metabolism in myocytes, potentially explaining our observations in humans. Our findings in humans are in line with previous studies, mostly in disease populations. In patients with chronic heart failure, ID was associated with impaired exercise tolerance and muscle strength, 9,10,[19][20][21] while ID correction improved exercise capacity [22][23][24][25] and cardiac function. 11,26 Also, in the elderly, lower iron status has been related with lower muscle strength. 27 Intravenous iron treatment of iron-deficient cancer patients enhanced muscle strength. 28 Finally, ID was associated with less recovery of muscle strength after stroke. 29 Other studies report a detrimental effect of ID on endurance. In women, a lower TSAT was correlated with worse aerobic fitness, but not with anaerobic capacity 30 while correction of non-anaemic ID improved peak VO 2 during exercise. 31 Chronic heart failure patients with ID displayed more pronounced phosphocreatine depletion in calf muscle after exercise and a higher muscular acidification rate, or a rapid switch to anaerobic glycolysis, 10 while intravenous iron reduced phosphocreatine depletion. 11 Not only low, but also higher ferritin levels were linked to lower muscle mass in our study. A negative correlation between ferritin and muscle mass has been previously observed in the NHANES cohort. 32 Iron overload triggers the Fenton reaction, inducing the formation of oxygen radicals, which may have detrimental effects on skeletal muscle. 33,34 Furthermore, ferritin is an acute-phase protein and high levels may occur in the context of inflammation, although the relation between ferritin and muscle mass remained significant after adjustment for CRP in our study.
Iron is a main component of haemoglobin, and thereby plays an important role in transport of oxygen to all tissues, including skeletal muscle. 35 Furthermore, iron is involved in nucleotide synthesis and cellular energy metabolism, and may therefore be crucial for cell proliferation and function. 35 Iron deficiency, myoblast proliferation and muscle mass Finally, iron is important for biosynthesis of testosterone 36,37 which promotes muscle mass.
To further study the effects of ID on skeletal muscle function, we used C2C12 cells, a well-documented model of immortalized mouse myoblasts that can differentiate into skeletal myocytes. ID was induced using the iron-chelator DFO. A reduction in intracellular Fth content confirmed that DFO captured iron from C2C12 myoblasts and differentiated myocytes. Addition of 7.5 μM DFO caused a > 50% reduction in Fth content, and therefore this concentration was used for further experiments. Co-treatment with the iron compound FC prevented the increase in intracellular Fth in myoblasts and myocytes. The expression of TfR, a marker of cellular ID, was increased by DFO, while expression of Slc40a1, a marker of iron abundance, was decreased ( Figure 2C-E). Co-treatment with FC prevented these changes, and therefore we considered these conditions to be representative of iron depletion and repletion, respectively, in skeletal myoblasts and myocytes.
DFO significantly impaired myoblast proliferation with a clear dose-effect response. To confirm that the reduction in proliferation rate was attributed to ID, rather than the chelation of other metal ions or direct toxic effects of DFO, iron availability was restored by co-treatment with FC, which reversed the effects of DFO ( Figure 3A). There are several possible explanations for this observation. First, proliferation could be reduced due to ID-induced adverse effects on energy metabolism. Furthermore, the observation from RNA sequencing and gene ontology that nucleotide synthesis is one of the main processes affected by ID also suggests direct effects on DNA replication. In myoblasts, two of the five genes that were most significantly suppressed were involved in the folic acid cycle, which is crucial for DNA stabilization and replication. DFO did not affect myocyte differentiation.
Subsequently, we assessed the effects of ID on energy metabolism. Although oxygen consumption during basal cellular respiration was not affected by ID, there was a non-significant trend suggesting a lower maximal mitochondrial reserve. Also, the acidification rate was higher after ID induction, although this was also non-significant. Previous studies showed a significant reduction in basal respiratory consumption rate as well as in maximal oxygen consumption in human cardiomyocytes treated with DFO 5 or in C2C12 skeletal muscle cells treated with deferiprone, another iron chelator. 13 Using RNA sequencing analysis, we found that expression of most genes encoding proteins involved in glycolysis and lactate production, the main processes of anaerobic energy metabolism, was up-regulated after ID induction. In contrast, expression of genes involved in aerobic processes such as fatty acid oxidation, the citric acid cycle and oxidative phosphorylation, was not affected, apart from two subunits of complex I. This finding is in line with a previous study in cardiomyocytes 5 and suggests that during ID, there is a malfunction of the oxidative energy metabolism and a compensatory shift towards the glycolytic anaerobic type of chemical energy production. Activity of the citric acid cycle may be hampered by a lower function of aconitase, which is known to be iron-dependent, 38 while oxidative phosphorylation may be impaired because the five complexes driving oxidative phosphorylation contain iron in the form of iron-sulfur clusters. 39 Moreover, their function requires oxygen, which may be insufficiently available because of a lower content of myoglobin, as also observed in the current study and, previously, in mice. 14 Our findings are in line with previous observations in iron-deficient rodents in whom complex I-IV activity was reduced in muscle. 14,40 Although glycolysis is usually sufficient for fast and powerful muscle contractions, oxidative metabolism of fatty acids and carbohydrates is required for sustained skeletal muscle use. A shift from aerobic towards anaerobic energy metabolism is a common hallmark of chronic diseases such as diabetes, heart failure or chronic obstructive pulmonary disease. 41 In patients with chronic heart failure, correction of ID improved skeletal muscle energy metabolism. 11 We also found that ID induced expression of two genes that are important for cellular atrophy and protein degradation, Fbxo32 and Trim63, in line with a previous study. 42 Our results regarding the effects of ID on apoptosis were The key enzymes involved in these processes are depicted as circles. Enzymes of which the gene expression is not affected by DFO are depicted in blue. Enzymes of which the gene expression is increased after treatment with DFO are depicted in green. None of the enzymes were down-regulated after treatment with DFO. ACAD, acyl-coenzyme a dehydrogenase; ACAT, acetyl-coenzyme a acetyltransferase; ACO, aconitase; ADP, adenosine diphosphate; ATP, adenosine triphosphate; CACT, carnitine-acyl carnitine translocase; CPT, carnitine palmitoyltransferase; CS, citrate synthase; DFO, deferoxamine; ECH, enoyl-coenzyme a hydratase; FAD, flavin adenine dinucleotide; HADH, hydroxyacyl-coenzyme a dehydrogenase; FC, ferric citrate; FH, fumarate hydratase; GAPDH, glycerylaldehyde 3-phosphate dehydrogenase; GPI, glucose-6-phosphate isomerase; HADH, hydroxyacyl-coenzyme a dehydrogenase; IDH, isocitrate dehydrogenase; α-KGDH, α-ketoglutarate dehydrogenase; LDH, lactate dehydrogenase; MDH, malate dehydrogenase; MPC, mitochondrial pyruvate carrier; NAD+, nicotinamide adenine dinucleotide; NADPH, nicotinamide adenine dinucleotide phosphate; PDH, pyruvate dehydrogenase; PFK-1, phosphofructokinase 1; PGAM, phosphoglycerate mutase; PGK-1, phosphoglycerate kinase; PKM, pyruvate kinase; SDH, succinate dehydrogenase; SUCL, succinyl-coenzyme A-synthetase; TPI, triose phosphate isomerase.
Iron deficiency, myoblast proliferation and muscle mass not equivocal: expression of both Bax and Bcl2 was slightly but non-significantly increased and the expression of caspase-9 was not altered. However, the expression of Bnip3, a pro-apoptotic gene encoding Bcl2 interacting protein 3, the highly similar Bnip3L, and Ddit4, encoding a marker of hypoxia and DNA damage, was increased in iron-deficient myoblasts. Also, expression of Fam162a and Ero1L, involved in hypoxia-induced apoptosis, was increased in iron-deficient myoblasts and myocytes. We did not find a change in percentages of dead, early apoptotic or total apoptotic cells, although there was a minor but significant increase in the percentage of late apoptotic cells. In addition, we found IDinduced up-regulation of Becn1, encoding beclin-1 which is the most important inducer of autophagy, suggesting that myocytes promote autophagy to recycle iron when its availability is scarce. However, protein expression of beclin-1 was reduced, not increased. This may be explained by posttranscriptional suppression of beclin-1 by apoptotic processes, known to regulate autophagy. 43 Potentially, induction of apoptosis could contribute to loss of muscle mass in patients with ID. Computational prediction of putative transcription factors analysis revealed that hypoxia inducible factor subunit 1-alpha (HIF1α) is the transcription factor involved in the expression of the largest cluster of DEGs related to ID. HIF1 is stabilized in response to hypoxia, promoting erythropoiesis and iron uptake from the gut, but may also be up-regulated by ID, independent of oxygen availability. 44 HIF1 may drive cellular compensatory mechanisms, such as inhibition of fatty acid oxidation and promotion of glycolysis and mitochondrial autophagy in response to skeletal muscle hypoxia. 45 Possibly, up-regulation or stabilization of HIF1α is the central effect of ID in skeletal muscle, explaining our other observations of altered energy metabolism and induction of cell-degradation. This hypothesis is partially supported by the observations of Leermakers et al, who reported that the increased expression of pro-apoptotic Bnip3 was HIF1Adependent. 13 Our study adds important new insights in the effects of ID on skeletal muscle. Its main strength is the availability of data from a large and extensively phenotyped cohort of community-dwelling individuals, combined with results of in vitro studies to investigate underlying mechanisms. Our study also has a number of limitations. First, we were not able to assess the relationship of ID with skeletal muscle strength in humans or with contractile strength in myocytes. Second, we did not assess morphological effects of ID on myocytes and therefore we could only provide circumstantial evidence that ID may induce atrophy and programmed cell death. Third, because we used a cell model, we could not assess any effects of ID on tissue vascularization, which might be increased by the HIF1a signalling cascade. Finally, the in vitro setting in C2C12 cells and the use of DFO to induce ID might not accurately reflect the impact of ID in vivo, and therefore the observations in C2C12 cells cannot be directly translated to the human in vivo setting. Nevertheless, our findings provide mechanistic insights that are supportive of the observations in humans.
In conclusion, ID was consistently associated with lower muscle mass in community-dwelling individuals, independent of haemoglobin and other factors. In skeletal muscle cells, ID reduced proliferation and resulted in increased expression of genes related to atrophy, apoptosis and autophagy, while inducing a switch from aerobic to anaerobic energy metabolism. Future studies should address whether correction of ID improves muscle mass.