Iron Deficiency May Slow Muscle Stem Cell Growth and Muscle Repair by Increasing HIF-2α Levels

All animal studies were approved by the University of Georgia IACUC and performed following the guidelines. Animals were housed in a temperature‐controlled (22°C) environment with 12:12‐h light–dark cycles. The following mouse strains were used in this study, all obtained from the Jackson Laboratory: C57BL6/J (#000664), Pax7^creERT2 (#017763) and R26R^{CAG‐Sun1/sfGFP} (#021039). Equal numbers of males and females were used in all experiments. To induce Cre activity in mice carrying Pax7^creERT2 allele, tamoxifen (20 mg/mL in corn oil, Sigma #C8267) was administered intraperitoneally for three consecutive days at a dosage of 100 mg/kg/day.

For iron sufficient (IS, 35 ppm, Dytes #115127) and iron deficient (ID, 3 ppm, Dytes #115072) chow feeding, 4‐week‐old C57BL6/J mice were housed in plastic microisolator cages with raised floor beds. Haemoglobin levels were measured using tail‐vein blood after 8‐week feeding to confirm iron‐deficient states by HemoCue Hb201 + Haemoglobin analyser (HemoCue).

To induce muscle injury, cardiotoxin (CTX; Sigma #C3987) was first reconstituted with sterile PBS (no Ca²⁺/Mg²⁺) and then injected into the tibialis anterior (TA) muscle of adult mice (0.5 nmol in 50 μL PBS). All injections were performed under general anaesthesia induced by 2–3% isoflurane delivered via a precision vaporizer, with mice maintained on a heated pad to prevent hypothermia. For pharmacological inhibition, PT2385 (1 μg in 50 μL 0.1% DMSO/PBS solution) or vehicle alone was injected into the TA muscle at 2, 3 and 4 days after injury (dpi), also under isoflurane anaesthesia.

MuSC sorting was performed on a MoFlo XDP cell sorter (UGA CTEGD Cytometry Laboratory). GFP^+pos MuSC (~5 × 10⁴) were FACS‐sorted from 3‐month‐old Pax7^creERT2;R26R^{CAG‐Sun1/sfGFP} mice (at 1 week after tamoxifen induction). For cell cycle analysis, cells were stained with Hoechst 33342 (10 μg/mL) and analyzed on a HyperCyAn flow cytometer (UGA CTEGD Cytometry Laboratory). Cells of different cell cycle phases were calculated by FlowJo v10.

C2C12 myoblasts (CRL‐1772, ATCC) were cultured in DMEM (4.5 g/L glucose) medium supplemented with 10% FBS or 0.1% FBS (serum starvation) and 1% penicillin–streptomycin‐neomycin under 21% O₂ and 5% CO₂. For cell cycle manipulation, C2C12 myoblasts were serum‐starved in DMEM with 0.1% FBS for 48 h to induce quiescence (G0/G1 arrest), followed by refeeding with DMEM with 10% FBS for 72 h to promote cell cycle re‐entry. For Fe²⁺ depletion, 2,2′‐bipyridyl (BPD; 20 μM; Sigma Aldrich #D216305) were added to culture media.

Cell culture medium was removed, and cells were washed with PBS and then incubated with 0.25 μM Calcein AM (Life Technologies, C3100MP) in reaction buffer (4.5 g/L glucose, 2 mM glutamine in PBS) for 30 min at 37°C with 5% CO₂. Then cells were washed twice and incubated in PBS for fluorescence intensity measurement at Ex/Em = 495/530 nm using CytoFlex Cytometer (Beckman Coulter) or fluorescence microscopy.

Cells were stained with ReadyProbe Cell viability imaging kit (Green and Blue, ThermoFisherScientific) under the guidance of manual instruction. Briefly, add two drops of NucGreen Dead reagent and two drops of NucBlue Live reagent per mL cell culture medium and incubate for 20 min at 37°C with 5% CO₂. Fluorescence images were captured by EVOS fluorescence microscope (FITC/GFP filter for dead cells and DAPI filter for live cells).

For EdU staining, 10 μM EdU was added into the culture medium 4 h before assay. EdU Click‐It kit (ThermoFisherScientific #C10340) was utilized for the detection of the EdU signal.

For cell number counting, Hoechst 33342‐stained or EdU‐stained C2C12 myoblasts were counted by BioTek Lionheart FX automated microscope.

For immunostaining, TA muscle sections were fixed with paraformaldehyde (1% in PBS, 10 mins), permeabilized with Triton X‐100 (0.5%, 10 mins), blocked with Mouse on Mouse Blocking Reagent (Vector Lab) and 5% BSA/5% normal goat serum/PBS and incubated with primary antibodies: anti‐Pax7 (1:5; DSHB #Pax7), anti‐HIF‐2α (1:250; ThermoFisherScientific #PA‐16510), anti‐Ki67 (1:1000; Abcam #ab15580), anti‐Myogenin (1:50; DSHB #F5D), anti‐Laminin B2 (1:1000; Millipore #05–206), anti‐embryonic MyHC (1:50, DSHB #F1.652) overnight at 4°C. Sections were washed in PBS/0.1% Tween‐20, incubated with AlexaFluor‐labelled secondary antibodies (1:200, 1 h), washed and mounted with DAPI‐containing mounting medium (Life Technologies). Mounted slides were imaged on a Zeiss LSM 710 confocal microscope. eMyHC,

For quantification of quiescent MuSC, Pax7⁺ nuclei were counted manually per cross‐section of uninjured muscle.

H/E staining was performed on muscle sections following an online protocol from the Treat‐NMD consortium (SOP: MDC1A_M.1.2.004).

The automated measurements of myofibre cross‐sectional areas on muscle IM images were performed as described in a previous publication (Mayeuf‐Louchart A. et al. Skeletal Muscle 2018).

Three non adjacent cryosections (~200 μm apart) were stained with anti‐eMyHC (DSHB F1.652, 1:50) together with anti‐Laminin B2 (1:1000) and DAPI. Images were acquired with identical exposure/gain across all groups. Quantification was performed in Fiji/ImageJ by an investigator blinded to group. Laminin B2 staining was used to generate a myofibre mask, defining total myofibre area. eMyHC signal was background‐subtracted and thresholded using an a priori fixed threshold determined from negative control images. Percents of eMyHC⁺ area were calculated as eMyHC‐positive pixels within the laminin mask divided by the total myofibre areas.

Total RNA was extracted from primary mouse myoblasts in three conditions: WT control, HIF‐2α overexpression (HIF‐2α OE) and HIF‐2α OE with HIF‐2α inhibition (100 nM PT2385). For each condition, three biological replicates were generated from separate time points using different founding myoblast isolates and clones. In each replicate, total RNA was isolated from two independently cultured wells and pooled to minimize intra‐replicate variability. To reduce sequencing costs while maintaining representative transcriptomic profiles, total RNA from the three biological replicates per condition was combined into a single composite sample for RNA‐seq library preparation. RNA was purified using RNA Extraction and Clean‐up kits (Zymo Research) and sequenced by BGI Genomics on the DNBSEQ platform.

RNA‐seq raw reads were first quality‐checked using FastQC. Cleaned reads were aligned to the mouse reference genome (GRCm38/mm10) using TopHat2. Following alignment, HTSeq‐count was used to generate the gene‐level count matrix based on RefSeq gene annotations. Normalization and differential expression analysis were performed using DESeq2 in R.

GSEA was performed using GSEA 4.0.3 (UC San Diego, Broad Institute) application with gene set databases: h.all.v2023.Hs.symbols.gmt [Hallmarks]. Gene sets were permutated 2000 times. Gene sets with normalized enrichment score (NES) ≥ 1 and a normalized p value ≤ 0.05 were collected.

Unsupervised hierarchical clustering was performed using the online GenePattern platform (genepattern.org↗).

Rb1 promoter sequences were scanned for consensus hypoxia response elements (RCGTG) using the JASPAR database.

Statistical analyses were conducted using Prism GraphPad 6.0/9.0 software. When specific pairs of groups are compared, the Shapiro–Wilk tests for normality (α = 0.05) were performed. If both groups follow normality, Two‐tailed Welch's t tests were conducted to test the equality of the means. When three or more than three groups are compared, one‐way ANOVA with Bonferroni's post hoc tests were employed to assess the significance of differences among multiple groups. The significant level (α) was set at 0.05. Data presented in figures represent mean ± SEM.

Iron importers transferrin receptor 1 (TfR1) and divalent metal transporter 1 (DMT1) are key regulators of intracellular iron homeostasis. We compared their expression levels in FACS‐isolated MuSC (FI‐MuSC) from uninjured TA muscle and activated MuSC from cardiotoxin (CTX)‐injured muscle at 5 days after injury (5 dpi). Activated MuSC exhibited elevated expression of cell cycle RNAs (Ccne2, Cdk2) and Myod1 RNA>, and downregulation of quiescence‐associated genes (Epas1, Spry1), confirming their activation states (Figure 1A; Figure S1A). Notably, Tfrc (encoding TfR1) and Slc11a2 (encoding DMT1) mRNA levels were significantly increased in activated MuSC compared to FI‐MuSC (Figure 1A).

We further examined their expression in C2C12 myoblasts under cell cycle arrest and serum refeeding‐induced proliferation states. Serum refeeding increased the proportion of proliferating cells from ~27% to ~97%, as indicated by fluorescence of cells with abundant Ki67 + proteins (Figure 1B). Importantly, this proliferative response occurred without induction of differentiation (Myogenin⁻) or cell death (PI⁻; Figure S1B–C). Compared to quiescent myoblasts, Tfrc mRNA was significantly elevated, whereas Slc11a2 was unchanged in proliferating myoblasts (Figure 1C). Western blot confirmed increased TfR1 protein following serum refeeding, and flow cytometry of live cells stained for CD71 (TfR1) showed elevated surface TfR1 in proliferating cells (Figure 1D–E). These findings indicate that Tfrc is dynamically regulated during myoblast activation, resulting in increased surface expression of TfR1 in proliferating myoblasts.

To assess whether increased TfR1 expression correlates with changes in intracellular chelatable iron, we quantified intracellular chelatable Fe²⁺ using calcein‐AM staining (Supplemental Material—Ref. S1). Because Fe²⁺ quenches calcein fluorescence, a reduced signal indicates a higher free Fe²⁺ level. FACS analysis revealed significantly lower calcein‐AM fluorescence in proliferating myoblasts, consistent with elevated chelatable Fe²⁺ (Figure 1F).

To assess the role of chelatable Fe²⁺ iron in myoblast proliferation, we treated C2C12 myoblasts with the Fe²⁺ chelator 2,2′‐bipyridyl (BPD; Supplemental Material—Ref. S2) to model partial iron deprivation. BPD treatment for 5 days did not induce cell death (Figure S2A). However, proliferation assays showed that BPD‐treated myoblasts proliferate more slowly than controls, and EdU incorporation was significantly reduced (Figure 2A–B). These results suggest that Fe²⁺ depletion attenuates the proliferative capacity of myoblasts in vitro.

To examine the effect of iron depletion in vivo, we compared mice maintained on an iron‐deficient diet with those on an iron‐sufficient diet (Figure 2C). The number of quiescent Pax7⁺ MuSC per cross‐section was comparable between IS and ID in TA and soleus (Figure 2D, Figure S2B).

To assess regenerative proliferation, we induced muscle injury by CTX injection and analyzed TA muscles at 10 dpi. ID mice exhibited a significant reduction in total Pax7⁺ MuSC (Figure 2E). Moreover, co‐staining for Pax7 and Ki67 revealed a marked decrease in the proportion of proliferating (Pax7⁺/Ki67⁺) MuSC among total MuSC in ID mice compared to controls, indicating impaired MuSC activation and proliferation under iron‐deficient conditions (Figure 2F).

To evaluate whether iron deficiency affects skeletal muscle regeneration, we collected TA muscles from IS and ID mice at 0, 10 and 30 dpi for histological analysis. After 8 weeks on the respective diets (0 dpi), TA muscle weights were significantly reduced in ID male mice while female mice showed a similar trend that did not reach statistical significance (Figure S3A). Body weights remained comparable between IS and ID groups (Figure S3B). At baseline (0 dpi), both groups showed normal muscle morphology, with well‐organized myofibres and peripheral myonuclei, and no notable differences in myofibre size (Figure 3A). A moderate reduction in haemoglobin levels, with unchanged haematocrit, confirmed a state of iron deficiency without anaemia in the ID mice (Figure S3C, S3D).

At 10 dpi, IS mice displayed robust regeneration with numerous newly formed myofibres containing centrally located nuclei—a hallmark of regeneration. In contrast, ID mice showed impaired regeneration, characterized by smaller myofibres, increased fibrotic areas and disorganized architecture (Figure 3B). By 30 dpi, IS muscles had largely returned to normal morphology, with mature, enlarged fibres and resolved inflammation. However, ID muscles still exhibited reduced fibre size, a higher proportion of small fibres (CSA ≤ 1200 μm²) and fewer large fibres (CSA ≥ 1800 μm²), along with persistent fibrotic areas, indicating delayed or incomplete regeneration (Figure 3C).

To further assess myofibre maturation, we stained for embryonic myosin heavy chain (eMyHC), a marker of nascent myofibres that is typically downregulated upon maturation. At 10 dpi, eMyHC⁺ fibres were abundant in IS muscles but relatively sparse in ID muscles, suggesting delayed differentiation. Quantification at 10 dpi confirmed a lower percent eMyHC⁺ area in ID versus IS (Figure 3D). By 30 dpi, eMyHC expression was nearly absent in IS muscles but remained in ID muscles, consistent with prolonged regeneration and delayed fibre maturation in the ID condition (Figure 3D).

To determine whether iron deficiency impacts MuSC self‐renewal, we stained for Pax7⁺ MuSC at 30 dpi. Pax7⁺ MuSC numbers were comparable between IS and ID groups, indicating that mild iron deficiency does not compromise MuSC maintenance after one round of regeneration (Figure 3E).

Free Fe²⁺ is a cofactor for Prolyl Hydroxylases that are responsible for HIF‐α degradation under normoxia. Our previous work showed that HIF‐2α stabilization impairs muscle regeneration by reducing MuSC proliferation [15, 16], prompting the hypothesis that HIF‐2α mediates the anti‐proliferative effects of iron deficiency. To test this, we first examined whether HIF‐2α is stabilized in MuSC under iron deficiency. At 10 dpi, we co‐stained TA muscle sections from IS and ID mice for Pax7 and HIF‐2α. HIF‐2α⁺ nuclei were observed in three cell types: mononuclear immune cells, myocytes and Pax7⁺ MuSC (Figure 4A). Both IS and ID muscles showed HIF‐2α expression in immune cells (white arrowheads), but only ID muscles showed prominent nuclear HIF‐2α accumulation in myonuclei (white stars). In MuSC, most Pax7⁺ cells in IS muscle were HIF‐2α⁻, consistent with our prior findings that HIF‐2α is downregulated during MuSC activation [15]. In contrast, ~84% of Pax7⁺ MuSC in ID muscle were HIF‐2α⁺, compared to ~22% in IS muscle (white circles), indicating strong stabilization of HIF‐2α in the MuSC population under iron deficiency.

Consistent results were also observed in vitro. In C2C12 myoblasts, BPD treatment induced nuclear accumulation of HIF‐2α, whereas vehicle‐treated cells showed minimal peri‐nuclear staining (Figure 4B). Immunoblot analysis confirmed that both total and nuclear HIF‐2α levels were markedly increased following BPD treatment (Figure 4C).

To determine whether HIF‐2α mediates the proliferation block caused by iron depletion, we co‐treated BPD‐exposed myoblasts with the HIF‐2α–selective inhibitor PT2385. PT2385 significantly rescued the proliferation defect in BPD‐treated cells, as demonstrated by cell counts and restored EdU incorporation (Figure 4D–E). To validate this in vivo, we administered PT2385 intramuscularly in ID mice at 2–4 dpi following injury (Figure 4F). Treatment with PT2385 significantly increased the number of Pax7⁺ MuSC and the proportion of Ki67⁺ proliferating MuSC compared to vehicle‐treated ID controls, indicating that HIF‐2α is a critical mediator of the proliferation block caused by iron deficiency (Figure 4F).

To determine whether HIF‐2α is sufficient to inhibit myoblast proliferation, we overexpressed a stabilized triple‐mutant form of HIF‐2α (P405A/P530V/N851A, which is resistant to hydroxyproline‐induced degradation) in C2C12 myoblasts. HIF‐2α overexpression (OE) resulted in robust induction of nuclear HIF‐2α, as confirmed by immunofluorescence and immunoblotting (Figure 5A, Figure S4A). HIF‐2α OE myoblasts exhibited reduced proliferation and EdU incorporation compared to Dox‐treated control cells (infected with empty vector, EV), similar to BPD‐treated cells (Figure 5B, Figure S4B). Flow cytometry revealed a G0/G1 arrest in HIF‐2α OE cells, with an increased proportion of cells in G0/G1 (70.2%) compared to EV control cells (57.7%), suggesting that HIF‐2α inhibits proliferation by blocking G1/S transition (Figure 5C).

We next performed RNA‐seq comparing WT and HIF‐2α OE cells. As expected, canonical HIF‐2α targets such as Atp7a, Cav1 and Vegfa were upregulated (Figure S4C). Gene Set Enrichment Analysis (GSEA) showed enrichment of hypoxia‐ and inflammation‐related pathways among transcripts with increased expression. Strikingly, the top downregulated gene set was 'E2F TARGETS,' which include transcripts encoding G1/S cell cycle regulators such as Ccna2, Ccnb1 and Cdk2 (Figure 5D–E). Hierarchical clustering and RT‐qPCR confirmed repression of E2F targets upon HIF‐2α OE, which was fully rescued by PT2385 co‐treatment (Figure 5F, 6A). Similarly, BPD treatment in normal C2C12 myoblasts reduced the expression of E2F target genes, which was partially rescued by PT2385, as evidenced by an increase of Ccna2 mRNA to ~45% and Ccnb1 mRNA to ~33% of control levels—both significantly higher than BPD alone (Figure S4D). This supports that HIF‐2α mediates the repression of E2F targets under iron deficiency.

The Rb pathway is a well‐established regulator of E2F activity. We hypothesized that HIF‐2α represses E2F targets through Rb induction. Indeed, Rb1 (encoding Rb) mRNA was significantly upregulated in HIF‐2α OE cells, and this induction was suppressed by PT2385 (Figure 6B). BPD‐treated myoblasts also showed elevated Rb1 expression (Figure S5A). A conserved hypoxia response element (HRE) locates in the Rb1 promoter (Figure S5B). Similar to known HIF‐2α targets Vegfa and Cav1, ChIP‐qPCR confirmed HIF‐2α binding to the Rb1 promoter, supporting direct transcriptional regulation (Figure 6C, Figure S5C).

To investigate whether HIF‐2α represses E2F targets through Rb‐mediated chromatin remodelling [19, 20], we performed ChIP‐qPCR for the transcriptionally active histone mark H3K27ac at E2F target promoters. HIF‐2α OE significantly reduced H3K27ac levels at Ccna2, Ccnb1, Cdk2 and Cdc25a, consistent with transcriptional repression (Figure 6D, Figure S5D). Finally, Rb1 knockdown via siRNA reversed the proliferation defect in HIF‐2α OE cells and restored RNA expression of E2F targets, confirming that Rb is functionally required for HIF‐2α–mediated repression (Figure 6E–F).

Together, these results identify Rb as a direct transcriptional target of HIF‐2α and demonstrate that HIF‐2α suppresses myoblast proliferation by activating Rb and inhibiting E2F target RNA expression.

Given the central role of HIF‐2α in mediating proliferation arrest under iron‐deficient conditions, we tested whether transient pharmacological inhibition of HIF‐2α could improve muscle regeneration under iron deficiency. Following the same dosing schedule as described in Figure 4, we administered PT2385 directly into the TA muscle of ID mice after injury and collected tissue at 10 and 30 dpi for analysis. At both 10 dpi and 30 dpi, PT2385‐treated ID muscles exhibited larger regenerated myofibres compared to vehicle‐treated ID controls, indicating improved myofibre growth (Figure 7A–B). Immunostaining for eMyHC revealed enhanced expression at 10 dpi and near‐complete resolution by 30 dpi in PT2385‐treated muscles, suggesting that HIF‐2α inhibition accelerates myofibre maturation (Figure 7C). At 30 dpi, the weights of regenerated TA muscle were comparable with those from 0 dpi/uninjured states in the IS group; where a significant reduction was observed in the ID group (Figure 7D). Notably, PT2385 treatment at 2–4 dpi rescued the injury‐associated muscle mass loss in the ID group (Figure 7E). The number of Pax7⁺ MuSC at 30 dpi was comparable between PT2385‐treated and control ID groups, indicating that short‐term HIF‐2α inhibition does not affect MuSC self‐renewal (Figure S5A).