Abstract Objective Metabolic flexibility refers to the ability of tissues to adjust cellular fuel choice in response to conditional changes in metabolic demand and activity. A loss of metabolic flexibility is a defining feature of various diseases and cellular dysfunction. This study investigated the role of microRNA-1 (miR-1), the most abundant microRNA in skeletal muscle, in maintaining whole-body metabolic flexibility. Methods We used an inducible, skeletal muscle-specific knockout (KO) mouse model to examine miR-1 function. Argonaute 2 enhanced crosslinking and immunoprecipitation sequencing (AGO2 eCLIP-seq) and RNA-seq analyses identified miR-1 target genes. Metabolism was investigated using metabolomics, proteomics, and comprehensive bioenergetic and activity phenotyping. Corroborating information was provided from cell culture, C. elegans, and exercised human muscle tissue. Results miR-1 KO mice demonstrated loss of diurnal oscillations in whole-body respiratory exchange ratio and higher fasting blood glucose. For the first time, we identified bona fide miR-1 target genes in adult skeletal muscle that regulated pyruvate metabolism through mechanisms including the alternative splicing of pyruvate kinase (Pkm). The maintenance of metabolic flexibility by miR-1 was necessary for sustained endurance activity in mice and in C. elegans. Loss of metabolic flexibility in the miR-1 KO mouse was rescued by pharmacological inhibition of the miR-1 target, monocarboxylate transporter 4 (MCT4), which redirects glycolytic carbon flux toward oxidation. The physiological down-regulation of miR-1 in response to hypertrophic stimuli caused a similar metabolic reprogramming necessary for muscle cell growth. Conclusions These data identify a novel post-transcriptional mechanism of whole-body metabolism regulation mediated by a tissue-specific miRNA. Keywords: eCLIP-seq, Aerobic glycolysis, PKM, MCT4, VB124, Resistance training Graphical abstract Image 1 [63]Open in a new tab Highlights * • miR-1 is necessary to maintain whole-body metabolic flexibility. * • miR-1 targets several genes in the pyruvate metabolism pathway. * • miR-1 is necessary for sustained endurance activity in mice and C. elegans. * • Downregulation of miR-1 results in a pro-growth metabolic reprogramming. 1. Introduction Metabolic flexibility is the ability to adapt to changes in energy demand by adjusting fuel selection between glucose and fatty acids [[64]1]. In a metabolically flexible state, the whole-body respiratory quotient (RQ) exhibits fluctuations characteristic of the ability of mitochondria to shift between carbohydrate and fat substrate oxidation depending on changes in activity level and nutrient availability [[65]1]. Metabolic inflexibility is now recognized as a hallmark of a number of clinical conditions including obesity, diabetes and insulin resistance, heart disease, polycystic ovary syndrome, non-alcoholic fatty liver disease, and chronic physical inactivity [[66]2]. Skeletal muscle is a versatile tissue that is normally characterized by a robust ability to switch freely between fuels and has a central role in whole-body metabolism [[67]2]. Another factor that supports a causal role for skeletal muscle in metabolic flexibility regulation is the fact that skeletal muscle is responsible for 70–80% of postprandial or insulin-stimulated glucose uptake from circulation and more than 95% of energy requirements during exercise [[68]2,[69]3]. Thus, acquiring a better understanding of the mechanisms regulating metabolic flexibility in skeletal muscle is critical for treating conditions associated with metabolic inflexibility. Adult skeletal muscle expresses a small set of tissue-specific microRNAs (miRNAs) referred to as myomiRs [[70]4]. Of all miRNAs expressed in striated muscle, including other myomiRs, microRNA-1 (miR-1) is by far the most abundant miRNA in adult mammalian (human, rodent, equine, ovine, caprine, bovine, and porcine) skeletal muscle, typically accounting for 80–90% of all miRNA reads [[71]5,[72]6]. While previous studies have investigated the function of miR-1 in the heart and/or skeletal muscle, it was during embryonic or post-natal development and in conjunction with other myomiRs such as miR-133a or miR-206 [[73][7], [74][8], [75][9]]. Thus, the specific function of miR-1 in adult skeletal muscle remains to be rigorously investigated. To address this fundamental gap in our understanding of miR-1, we generated an inducible mouse model to specifically inactivate only miR-1 in adult skeletal muscle fibers without affecting the expression of other myomiRs. Using RNA-seq and AGO2 eCLIP-seq, we identified bona fide miR-1 target genes that regulate pyruvate metabolism. As the nexus of carbon metabolism within the cell, pyruvate metabolism affects the entire cellular metabolic network and, as a consequence, metabolic flexibility. Thus, in the miR-1 KO, the muscle undergoes metabolic reprogramming towards aerobic glycolysis which impairs metabolic flexibility by directing the fate of pyruvate away from mitochondrial oxidation towards lactate. This loss in metabolic flexibility of skeletal muscle in the miR-1 KO causes a significant reduction in running and swimming exercise performance in mice and C. elegans, respectively. In addition, we provide evidence that the physiological down-regulation of miR-1 in response to a hypertrophic stimulus in both humans and mice results in a pro-growth metabolic reprogramming reminiscent of cancer cells, i.e., aerobic glycolysis, that supports biosynthesis for muscle cell growth. 2. Methods 2.1. Generation of inducible miR-1 KO mouse All animal procedures were conducted in accordance with institutional guidelines for the care and use of laboratory animals and approved by the Institutional Animal Care and Use Committee of the University of Kentucky. Mice were housed in a temperature- and humidity-controlled room and maintained on a 14:10-hr light–dark cycle with food and water ad libitum. To knockout miR-1 specifically in adult skeletal muscle, the skeletal muscle-specific inducible Cre mouse (HSA-MCM) [[76]10] was crossed with a floxed miR-1-1 and miR-1-2 (miR-1-1^f/f; miR-1-2^f/f) mouse [[77]7] to generate the HSA-MCM; miR-1-1^f/f; miR-1-2^f/f mouse, designated HSA-miR-1. Adult (4 months of age) female HSA-miR-1 mice were administered tamoxifen (2 mg/day) by intraperitoneal injection for five consecutive days to induce KO of miR-1. Tamoxifen-treated littermate miR-1-1^f/f; miR-1-2^f/f mice (HSA-MCM negative) pr vehicle-treated (15% ethanol in sunflower seed oil) HSA-miR-1 mice served as WT controls. Unless otherwise noted, after an 8 week-washout, (6 months of age), WT and KO mice were subjected to downstream experimentation or humanely euthanized via carbon dioxide asphyxiation followed by cervical dislocation. Hindlimb muscles were then carefully excised and prepared for storage or immediate downstream analyses, as described in sections below. 2.2. Genotyping For miR-1-1^f/f genotyping, the following forward (F) and reverse (R) primers were used: F: 5′-GGG AGC GGT TCC TTA CGA CCA TC-3’; R: 5′-TCC ATC GGT CCA TTG CCT TTC-3’. DNA samples for genotyping were ran with a negative control and positive control. The PCR product for the miR-1-1 WT allele was 1000 bp, while the miR-1-1^f/f allele generated a larger PCR product (∼1100 bp). For miR-1-2^f/f genotyping, genotyping primers were as follows: F: 5′-CGC AGG AGT GCC TAC TCA G-3′ and R: 5′-AGT GCT GAA GAT AGC ACT TGC C-3’. The PCR product for the miR-1-2 WT allele was <1500 bp (∼1400 bp), while the miR-1-2^f/f allele generated a larger PCR product 1500 bp in size. 2.3. Gene expression For RNA-seq and qPCR, total RNA was isolated from muscle previously snap-frozen in liquid nitrogen. Samples were homogenized in TRIzol using a THb Handheld Tissue Homogenizer and Hard Tissue Omni Tip Plastic Homogenizer Probes (Omni International, Kennesaw, GA, USA). RNA was isolated using a Direct-zol RNA Miniprep Kit (R2053, Zymo Research, Irvine, CA, USA) according to manufacturer's instructions. For RNA-seq, RNA samples were shipped to Novogene Corporation Inc. (Sacramento, CA, USA) for library preparation (polyA, non-strand specific) and sequencing (20 million PE150 reads, NovaSeq, Illumina, San Diego, CA). RNA-seq data were analyzed in R (Ver. 4.4) using the DESeq2 package [[78]11]. Pathway analyses were performed using the Enrichr web-based tool [[79][12], [80][13], [81][14]]. For RNA-seq and pathway analyses, significance was set using an FDR alpha of 0.05. RNA-seq data have been deposited in the Gene Expression Omnibus (GEO) database ([82]GSE274378). For comparison of miR-1 KO RNA-seq data with RNA-seq data during muscle hypertrophy, the list of differentially expressed genes from a previous study of 72 h of synergist ablation-induced MOV was downloaded [[83]15]. Raw MOV data are available in [84]GSE213406. For gene expression of mRNAs, cDNA was synthesized using a High-Capacity RNA-to-cDNA kit (4387406, Thermo Fisher, Waltham, MA, USA). Prepared cDNA templates were analyzed by qPCR using TaqMan probes (4331182, Thermo Fisher, Mib1: Mm00523008_m1, Pkm: Mm00834102_gH; 4351372, Thermo Fisher, PTBP1: Hs00914687_g1) and TaqMan Fast Advanced Master Mix (4444556, Thermo Fisher) run on a QuantStudio 3 PCR system (Thermo Fisher). Threshold cycle (Ct) values between these genes and the housekeeping gene beta-2-microglobulin (B2m, 4331182, Thermo Fisher, Mm00437762_m1) was used to normalize data, and the 2ˆ(-delta delta Ct) method was used to calculate fold change. For miRNA expression, cDNA was synthesized using a TaqMan MicroRNA Reverse Transcription Kit (4366596, Thermo Fisher) and TaqMan MicroRNA RT primers (4427975, Thermo Fisher). Gene expression of miRNAs was analyzed by qPCR using TaqMan MicroRNA Assays (4427975, Thermo Fisher) as follows: miR-1 Assay ID 002222, miR-133a Assay ID 002246, miR-206 Assay ID 000510. The endogenous control U6 snRNA (Assay ID 001973) was used for normalization, and the 2ˆ (-delta delta Ct) method was used to calculate fold change. 2.4. Single fiber isolation Isolation of single fibers for MuSC removal was based on the paper from Rosenblatt et al. [[85]16]. Extensor digitorum longus (EDL) muscles were carefully dissected and handled via the tendons. Muscles were immediately placed into warmed collagenase solution (2 mg/ml Type I collagenase [MB-122-0050, Rockland Immunochemicals, Limerick, PA, USA], 1% penicillin-streptomycin [Pen-Strep, 97063-708, VWR, Radnor, PA, USA] in DMEM [30–2002, ATCC, Manassas, VA, USA]), and incubated at 35 °C for 45 min, inverting the tube every 5–10 min. To stop digestion, muscles were transferred using a large bore pipette coated with heat inactivated horse serum (26050088, Thermo Fisher) to 60 mm culture plates (430166, Corning, Corning, NY, USA) pre-coated with heat inactivated horse serum to prevent fibers from sticking to the plates, and then filled with DMEM. Muscles were triturated with glass Pasteur pipettes until 20–30 fibers were released, then transferred to a new plate and placed in a 5% CO[2] incubator at 37 °C. 10% Matrigel Matrix- (354234, Corning) coated 24-well plates (3524, Corning) were prepared, then isolated fibers were retrieved from the incubator, and individual fibers were removed from suspension with normal Pasteur pipettes and placed in the center of Matrigel-coated wells. Fibers were allowed to settle and attach for 3 min to the Matrigel, and then 500 μl of plating medium consisting of 10% horse serum and 10 ng/ml basic fibroblast growth factor (bFGF) (354060, Corning) in DMEM was slowly added to wells, care being taken not to agitate the droplet in which the fiber was plated. Three days after plating, MuSCs were stuck to the Matrigel, and fibers were removed from the Matrigel without disturbing the surrounding cells by means of a Pasteur pipette pulled to a fine tip in flame. After fiber removal, fibers were immediately placed in TRIzol for RNA isolation. 2.5. Immunohistochemistry Muscle fiber size and fiber-type distribution were assessed using methodology previously described by our lab [[86]17]. After harvesting, soleus or plantaris muscles were embedded in Tissue TEK Optimal Cutting Temperature (4583, Sakura Finetek, Torrance, CA, USA) and frozen using liquid nitrogen chilled isopentane. Unfixed 7 μm thick cryosections placed onto SuperFrost Plus slides (12-550-15, Thermo Fisher) were incubated in primary antibodies against myosin heavy chain (MHC) Types I (BA.D5), IIA (SC.71), IIB (BF.F3) (1:100, Developmental Studies Hybridoma Bank, Iowa City, IA, USA) as well as anti-laminin (1:100, L9393, Sigma–Aldrich, St. Louis, MO, USA) for 90 min at room temperature. In order to visualize MHC and laminin, fluorescence-conjugated secondary antibodies were applied to different mouse and rabbit immunoglobulin subtypes for 1 h at room temperature (1:250, A-21242, A-21121, A-21426, A-11046, Thermo Fisher). Type IIX fibers were inferred from unstained fibers. Muscle sections were imaged using a Zeiss upright microscope (AxioImager M1, Zeiss, Oberkochen, Germany) at 20x magnification and analyzed by MyoVision software [[87]18,[88]19]. 2.6. Indirect calorimetry A Sable Promethion Core phenotyping system (Sable Systems International, North Las Vegas, NV, USA) was used for indirect calorimetry at 3-min intervals. Metabolic data quantified included O[2], CO[2], and H[2]O vapor; energy expenditure was then calculated from the Weir equation, and the respiratory quotient (RQ) and RER was calculated as the ratio of the volume of CO[2] produced to the volume of oxygen O[2] used, or VCO[2]/VO[2]. The web-based tool CalR (Ver. 2) was used for indirect calorimetry analysis [[89]20]. Mice were individually caged and acclimated for 24 h before data collection (48 h). Data were analyzed using a two-way ANOVA for light and dark cycles separately. 2.7. Glucose tolerance test Intraperitoneal glucose tolerance tests were performed by injecting 1 mg glucose/gram body mass. Mice were fasted for 4 h prior to glucose injection. Blood glucose was measured before glucose injection and 15-, 30-, 60-, and 120-min following glucose injection by tail bleed with a handheld glucometer (TD-4116, Metene, Shenzhen City, China). 2.8. Voluntary wheel running Voluntary wheel cages for mice were from Actimetrics (Wilmette, IL, USA). Cage wheels (model PT2-MCR1) are 11-cm inside diameter, 5.4 cm wide, with 1.2-mm wide bars placed 7.5 mm apart. A wireless node was used to sense wheel revolutions from the rotation of a magnet mounted on the wheel axle, recorded via the ClockLab software (Actimetrics). For WT vs. miR-1 KO wheel running experiments, mice underwent a 4-week washout after tamoxifen treatment and then were single-housed in voluntary wheel cages for an additional 5 weeks. ClockLab analyses included number of wheel revolutions/day and bout length, as previously described [[90]21]. The first week was used for acclimation, then data over the subsequent 4 weeks were averaged. Daily running volume was defined as the average distance covered per day, and maximum bout was defined as the longest continuous running period without stopping. 2.9. C. elegans swim exercise C. elegans were cultured on NGM plates seeded with E. Coli (OP50) at 20 °C. N2 (Bristol strain, wild type) and mir-1 mutant MT17810 [mir-1(n4102) I] strains were obtained from Caenorhabditis Genetics Center (CGC), supported by the National Institutes of Health - Office of Research Infrastructure Programs (P40 OD010440). Swimming exercise was performed according to previous descriptions [[91]22]. Synchronized worms were bleached by sodium hydroxide bleaching buffer, then incubated overnight in M9 buffer on a rocker at 20 °C. Larval stage L1 worm populations were transferred to 60 mm NGM plates seeded with OP50 for 2 days to obtain adult D1 worms, which were then washed off plates with 3 ml M9 buffer and allowed to settle under gravity, with the supernatant including OP50 and larvae removed, for a total of 3 times. Worms were then transferred to unseeded NGM plates with M9 buffer using a glass Pasteur pipette, and all plates were then moved to a 20 °C incubator for 90 min. After swimming exercise, worms were washed off with M9 buffer, gravity settled, and transferred to new 60 mm NGM plates seeded with OP50 in a 20 °C incubator. Swimming exercise was performed 2 × 90 min daily for 5 days. CeleST (C. elegans Swim Test) was conducted as previously described [[92]23]. One day following the 5-day swimming exercise regimen, a 10 mm pre-printed ring on the surface of a glass microscope slide was used and covered with 50 μl M9 buffer. 7–8 worms per group and 30 worms total were picked up and placed into the swimming area, and 60-sec movies with ∼16 frames per sec of the worms were taken using a Nikon LV-TC microscope (Nikon Instruments, Melville, NY, USA) at 1x magnification with an OPTIKA C–P20CM camera (Optika Microscopes, Ponteranica, Italy). 2.10. AGO2 eCLIP and data processing The AGO2 eCLIP (enhanced crosslinking and immunoprecipitation) protocol was executed with minor modifications to previously described methods [[93]24,[94]25]. Skeletal muscle tissue was obtained from the University of Kentucky Center for Muscle Biology (CMB) biobank. Human samples were collected with written informed consent from each participant and the approval of the Institutional Review Board of the University of Kentucky. Participants’ written consent forms and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. Muscle biopsies were derived from the vastus lateralis of participants (n = 3 males, n = 3 females, 24.43 ± 5.35 y). Skeletal muscle tissues were pulverized and subjected to UV-irradiation to crosslink protein-nucleic acid interactions. RNA associated with AGO2 was immunoprecipitated using a custom human anti-rabbit AGO2 antibody [[95]26], followed by adapter ligation and RT-PCR to create libraries for high-throughput sequencing. Single-end 100 bp reads were performed on an Illumina Hi-Seq (Illumina, San Diego, CA, USA). Sequencing reads were pre-processed and mapped with minor changes to established protocols. Adapter sequences were trimmed, and the resulting reads were mapped to the human genome (hg38). Identical alignments were collapsed to eliminate PCR duplicates removed using UMI-Tools dedup, and strand-specific read coverage was calculated from the alignments of the six samples. To identify significant AGO2 clusters, we applied a zero-truncated negative binomial (ZTNB) model to calculate the significance of read coverage at each mapped genomic position [[96]27]. The assumption was that read heights for each gene are sampled from an underlying ZTNB distribution. Parameters for gene-specific ZTNB probability density functions were estimated using read heights measured at all positions within an annotated gene region for each sample. P-values were calculated based on the probability of observing a read height as large as the observed height and were assigned to each position. Fisher's method was used to summarize p-values at each genomic position across the six samples. Positions with a false discovery rate (FDR) of less than 5% were considered significant. Significant positions within 60 nucleotides of each other were merged into single contiguous intervals, and regions shorter than 50 nucleotides were extended symmetrically to 50 nucleotides to account for the AGO2 binding footprint. Intragenic cluster positions were then annotated according to their overlapping gene structures using Ensembl v75 annotations. HOMER software was used to discover motifs enriched in AGO2 eCLIP cluster sequences. Additionally, motif enrichment was performed using a sliding window approach, as done previously [[97]27]. Briefly, a single base offset sliding window of 7 nts was used to determine heptamer frequencies in cluster and background sequences. Enrichment data were calculated for all possible heptamers. Enrichment data for miR-1 seed was calculated individually for enrichment of 7M8 and 7A1 sequences. Enrichment significance was calculated by Fisher's exact test, and p-values were transformed to Benjamini-Hochberg FDRs. TargetScan perl package was used to scan for miR-1 target genome sequences within significantly enriched clusters. AGO2 eCLIP-seq data sets are available as custom built web browser ([98]www.mir-engagemint.io). For the CDF generated, miR-1 targets were tested against non-miR-1 targets using a Kolmogorov–Smirnov goodness-of-fit test. 2.11. Western blotting For Western blotting, muscle samples were homogenized in RIPA lysis buffer (97063-270, VWR) with protease and phosphatase inhibitor cocktails (97063-010, VWR and 786-450, G-Biosciences, St, Louis, MO, USA) using a THb Handheld Tissue Homogenizer. Protein concentrations were determined using a Pierce BCA Protein Assay Kit (23252, Thermo Fisher). Samples were then prepared in Laemmli sample buffer (1610737, Bio-Rad Laboratories, Hercules, CA, USA), boiled, and 20 μg of proteins were separated by SDS-PAGE using 4–20% Criterion TGX Stain-Free Protein Gels (5678094, Bio-Rad). Total protein stains were activated by UV excitation using a ChemiDoc MP system (Bio-Rad), and then proteins were transferred onto LF PVDF membranes (1620263, Bio-Rad). Membranes were blocked in 5% nonfat dry milk (1706404, Bio-Rad) for 1 h at room temperature, followed by primary antibody incubation overnight at 4 °C. The primary antibodies and respective concentrations used were as follows: anti-PKM1 (1:1000, 7067, Cell Signaling, Danvers, MA) anti-PKM2 (1:1000, 4053, Cell Signaling), anti-LDHB (1:1000, ab52488, Abcam, Cambridge, United Kingdom), anti-MCT4 (1:1000, 81569, Cell Signaling) anti-phospho-PDHE1a (Ser293) (1:1000, NB110-93479, Novus Biologicals, Centennial, CO, USA) and anti PDHE1a (1:1000, NBP2-33922, Novus Biologicals). Following primary antibody incubation, membranes were washed in TBS- 0.1% Tween-20 and then incubated in secondary antibodies (1:10,000, goat anti-rabbit IgG H + L secondary antibody HRP conjugate, 31460, Thermo Fisher) at room temperature for 1 h. Blots were developed with enhanced chemiluminescence using Clarity Western ECL Substrate (1705061, Bio-Rad), and imaged with a ChemiDoc MP system. Blots were quantified using Image Lab Software (Bio-Rad). Restore Western Blot Stripping Buffer (21059, Thermo Fisher) was used to strip membranes that were reprobed with a second primary antibody. 2.12. Bioenergetic phenotyping For assessment of mitochondrial respiration in permeabilized fibers, muscles were removed, cut lengthwise into small samples, and immediately placed in ice-cold biopsy relaxing and preservation solution (BIOPS: 10 mM Ca-EGTA buffer, 0.1 μM free calcium, 20 mM imidazole, 20 mM taurine, 50 mM K-MES, 0.5 mM DTT, 6.56 mM MgCl[2], 5.77 mM ATP [A9062, Sigma–Aldrich], 15 mM phosphocreatine [PCr, P1937, Sigma–Aldrich], pH 7.1). Fiber bundles were mechanically separated using forceps in a petri dish on ice, then transferred into BIOPS solution containing saponin (S2149, Sigma–Aldrich) at a final concentration of 50 μg/ml. Fiber bundles were agitated in saponin solution on ice for 30 min, transferred to mitochondrial respiration medium (5 mM ATP [A9062, Sigma–Aldrich], 105 mM K-MES, 30 mM KCl, 10 mM KH[2]PO[4], 5 mM MgCl[2], 1 mM EGTA, 2.5 g/L BSA) for 10 min, and then fiber bundles of ∼1–2 mg wet mass were added to Oroboros O2k Oxygraph (Oroboros Instruments, Innsbruck, Austria) chambers. All respiration experiments were carried out at 37 °C in a 2 ml reaction volume. For permeabilized soleus fiber experiments, mitochondrial respiration medium was supplemented with 20 mM creatine monohydrate (C3630, Sigma–Aldrich), and substrates conditions tested were Pyr/Mal (10/2 mM, P2256/M1000, Sigma–Aldrich) or octanoyl-carnitine (B6371, APExBIO, Houston, TX, USA)/Mal (0.5/0.1 mM) to assess complex I leak, and 4 mM ADP (A5285, Sigma–Aldrich) to initiate OXPHOS. For the remainder of the respirometry experiments, buffer was supplemented with 5 mM creatine monohydrate, 1 mM PCr, and 20 U/ml creatine kinase (CK) (10736988001, Sigma–Aldrich) to determine steady-state oxygen consumption rates (JO[2]) ranging from near resting up to ∼95% of maximal using a modified version of the CK energetic clamp technique [[99]28]. In these assays, the free energy of ATP hydrolysis (ΔG[ATP]) was calculated based on known amounts of creatine, PCr, and ATP in combination with excess CK and the equilibrium constant for the CK reaction [[100]28]. Substrate conditions tested were as follows: Pyr/Mal (5/2.5 mM), succinate/rotenone (10/0.005 mM), glutamate (G1501, Sigma–Aldrich)/Mal (10/2.5 mM), and octanoyl-carnitine (50892, Sigma–Aldrich)/Mal (0.2/2.5 mM). Following substrate addition, sequential additions of PCr to 6, 15, and 21 mM were performed to gradually slow JO[2] back toward baseline. For functional assays involving isolated mitochondria, differential centrifugation was employed to prepare the mitochondria. Gastrocnemius complexes (gastrocnemius, soleus, and plantaris) were excised and immediately placed in ice-cold Buffer A (10 mM EDTA in PBS, pH 7.4). Tissue was minced and then incubated on ice for 5 min in Buffer A, supplemented with 0.025% trypsin. Skeletal muscle suspensions were then centrifuged at 200×g for 5 min at 4 °C to remove trypsin. Tissue pellets were then resuspended in Buffer B (50 mM MOPS, 100 mM KCl, 1 mM EGTA, 5 mM MgSO[4]), supplemented with 2 g/L BSA and homogenized with a Teflon pestle and borosilicate glass vessel. Homogenates were then centrifuged at 500×g for 10 min at 4 °C, and supernatant from each tissue was filtered through thin layers of gauze and subjected to an additional centrifugation at 10,000×g for 10 min at 4 °C. Mitochondrial pellets were washed in 1.4 ml of Buffer B, then transferred to microcentrifuge tubes and centrifuged at 10,000×g for 10 min at 4 °C. Buffer B was aspirated from each tube, and the final mitochondrial pellets were suspended in 100–200 μl of Buffer B. Yield was assessed by protein concentration using the BCA method. Cytochrome c (10 μM) was included in all assays to check the integrity of the outer mitochondrial membrane. Fluorescent determination of ΔΨ was carried out using a QuantaMaster Spectrofluorometer (QM-400, Horiba Scientific, Kyoto, Japan) via tetramethyl rhodamine-methylester (TMRM, T668, Thermo-Fisher), at 37 °C as previously described [[101]28]. The fluorescence ratio of the following excitation/emission parameters: Ex/Em (576/590)/(552/590) was recorded, and a KCl standard curve was used to convert the ratios to millivolts. Buffer C was supplemented with 10 μM cytochrome C and 0.2 μM TMRM, and isolated mitochondria (0.05 mg/ml) were added to the assay buffer. Respiratory substrates, CK clamp components were then added, followed by sequential PCr additions as in the respirometry experiments. Following the final PCr addition, 15 nM oligomycin (75351, Sigma–Aldrich) was added for hyper-polarization of the ΔΨ, and finally, 10 mM cyanide (60178, Sigma–Aldrich) was added to depolarize the ΔΨ. Mitochondrial H[2]O[2] emission (JH[2]O[2]) was measured fluorometrically via the Amplex Ultra Red (AUR)/horseradish peroxidase (HRP) system, as previously described [[102]28]. Fluorescence was monitored (Ex:Em 565:600) using a QuantaMaster Spectrofluorometer, and resorufin fluorescence was converted to pmoles H[2]O[2] via an H[2]O[2] standard curve. Assay buffer was buffer C, supplemented with 2.5 g/L BSA, 5 mM creatine, 1 mM PCr, 20 U/ml CK, 10 μM AUR (A36006, Thermo-Fisher), 1 U/ml HRP (P8375, Sigma–Aldrich), and 20 U/ml superoxide dismutase (S9697, Sigma–Aldrich). Isolated mitochondria (0.1 mg/ml) were added to assay buffer, followed by addition of respiratory substrates (Pyr/Mal), 0.1 μM auranofin (A6733, Sigma–Aldrich), 5 mM ATP, and sequential PCr additions of 6, 15, and 21 mM. 2.13. Mitochondrial content For measurement of CS activity in muscle lysates, a colorimetric plate-based assay was used, whereby CoA-SH, a byproduct formed by the CS-mediated reaction of oxaloacetate (OAA) and acetyl-CoA, interacts with 5′, 5′-dithiobis 2-nitrobenzoic acid (DTNB) to form TNB (OD: 412 nm). A 96-well round bottom plate was loaded with assay buffer C (105 mM K-MES, 30 mM KCl, 10 mM KH[2]PO[4], 5 mM MgCl[2], and 1 mM EGTA, pH 7.2) supplemented with 0.2 mM DTNB and 0.5 mM acetyl-CoA, and tissue lysates (10 μg/well), and then incubated at 37 °C for 5 min to deplete endogenous substrates. The assay was then initiated by the addition of 1 mM OAA to sample wells, and absorbance was measured at 412 nm every 30 s for 20 min. For mitochondrial DNA (mtDNA) analysis, total DNA was isolated from gastrocnemius muscles using a QiAmp DNA Mini kit (51304, Qiagen, Hilden, Germany), and 10 μg per reaction was used as a template for SYBR Green (Bio-Rad)-based qPCR, as previously described [[103]29]. Mitochondrial target genes were NADH Dehydrogenase 1 (Nd1) (F: GGCTATATACAACTACGCAAAGGC, R: GGTAGATGTGGCGGGTTTTAGG) and Mito1 (F: ACATAGCACATTACAGTCAAATCCCTTCTCGTCCC, R: TGAGATTGTTTGGGCTACTGCTCGCAGTGC). H19 (F: CACTGGCCTCCAGAGCCCGT, R: CGTCTTGGCCTTCGGCAGCTG) was used as nuclear DNA control. 2.14. Mitochondrial-targeted proteomics Bioenergetic phenotyping was combined with subcellular mitochondrial-targeted nLC-MS/MS [[104]30]. Isolated mitochondria were lysed in buffer consisting of 8 M urea in 40 mM Tris, 30 mM NaCl, 1 mM CaCl[2], 1 x cOmplete ULTRA mini EDTA-free protease inhibitor tablet (05892953001, Roche Holding AG, Basel, Switzerland), pH 8.0. Samples were subjected to three freeze–thaw cycles, then sonicated with a probe sonicator in three 5 s bursts at an amplitude of 30 (Q Sonica CL188, Qsonica, Newton, CT, USA). Samples were then centrifuged at 10,000×g for 10 min at 4 °C. Equal amounts of protein (determined by BCA protein assay) were reduced with 5 mM DTT at 37 °C for 30 min, alkylated with 15 mM iodoacetamide at room temperature for 30 min in the dark, and unreacted iodoacetamide was quenched with DTT up to 15 mM. Digestion was initially performed with Lys-C Protease (90307, Thermo-Fisher, 1:100 w:w, 2 μg enzyme per 200 μg protein) for 4 h at 37 °C. Samples were diluted to 1.5 M urea with buffer containing 40 mM Tris (pH 8.0), 30 mM NaCl, 1 mM CaCl[2], then digested overnight with 50:1 w/w protein:enzyme of trypsin (V5113, Promega, Madison, WI, USA) at 37 °C. Samples were subsequently acidified to 0.5% TFA and centrifuged at 4000×g for 10 min at 4 °C. Supernatant containing soluble peptides was desalted, and the eluate was frozen and lyophilized. Final peptides were resuspended in 0.1% formic acid, then quantified using a Pierce Quantitative Colorimetric Peptide Assay (23275, Thermo Fisher). Samples were diluted to a final concentration of 0.25 μg/μl, then subjected to nLC-MS/MS analysis using an Ultimate 3000 RSLCnano system (Thermo Fisher) coupled to a Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Fisher) via a nanoelectrospray ionization source. For each injection, 1 μg of sample was initially trapped on a trapping column (Acclaim Pep Map 100, 200 mm × 0.075 mm, 164535, Thermo Fisher) 5 μl/min at 98/2 v/v water/acetonitrile with 0.1% formic acid. Analytical separation was then performed at a flow rate of 250 nl/min over a 95 min gradient of 4–25% acetonitrile using a 2 μm EASY-Spray PepMap RSLC C18 75 μm × 250 mm column (ES802A, Thermo Fisher) with a column temperature of 45 °C. MS1 was performed at 70,000 resolution, with an AGC target of 3 × 10^6 ions and a maximum injection time of 100 ms. MS2 spectra were collected by data-dependent acquisition of the top 15 most abundant precursor ions with a change greater than 1 per MS1 scan and dynamic exclusion enabled for 20 s. MS2 scans were performed at 17,500 resolution, with an AGC target of 1 × 10^5 ions and a maximum injection time of 60 ms. Precursor ions isolation window was 1.5 m/z, and normalized collision energy was 27. Proteome Discoverer 2.2 (Thermo Fisher) was used for data analysis, with default search parameters, including oxidation and carbamidomethyl (15.995 Da on M and 57.021 Da on C, respectively) as variable and fixed modifications. Data were searched against the Uniprot Mus musculus reference proteome (UP 000000589) and the mouse Mito Carta 3.0 database [[105]31]. Peptide spectrum matches (PSMs) were filtered to a 1% FDR and grouped to unique peptides while maintaining a 1% FDR at the peptide level. Peptides were grouped to proteins using rules of strict parsimony, proteins were filtered to 1% FDR. Peptide quantification was done using the MS1 precursor intensity, and imputation was performed via low abundance resampling. Mitochondrial enrichment factor (MEF) was determined using only high confidence master proteins by comparing mitochondrial protein abundance (proteins identified as mitochondrial by cross-reference with the MitoCarta 3.0 database) to total protein abundance. Quantification of OXPHOS protein complexes was generated by the summed abundance of all subunits within a complex. To assess electron transport chain (ETC) co-regulation, we calculated all pairwise Pearson's correlations (r) separately for WT and miR-1 KO mitochondria. Correlation analysis was performed with rcorr() using Pearson's correlation coefficient method from Hmisc (version 4.5.0) library, as previously described [[106]32]. All mass spectrometry samples were normalized to total protein abundance, which was converted to the Log2 space. For pairwise comparisons between WT and miR-1 KO, tissue mean, standard deviation, and p-value (two-tailed Student's t-test, assuming equal variance) were calculated. 2.15. Metabolomics Frozen gastrocnemius samples, each weighing 25 mg were cryo-fractured in muscle sample extraction solution (50% HPLC-grade methanol and 100% formic acid in a 100:1 ratio) in tissueTUBE TT1 Extra Thick tubes (520007, Covaris, Woburn, MA, USA) using a Covaris CryoPREP CP02 Cryogenic Dry Pulverization system (Covaris). The pulverized samples were further homogenized with zirconium silicate beads (Next Advance, Inc., Troy, NY, USA) using a Bullet Blender Gold (Next Advance, Inc.). Samples were centrifuged at 20,000×g for 10 min, then filtered using Captiva EMR-Lipid cartridges (Agilent Technologies). Filtered samples were collected in LC/MS autosampler vials (Agilent Technologies) by centrifugation at 500×g for 20 min, followed by drying and concentration using a Savant SpeedVac Integrated Vacuum Concentrator System and Kit (SPD2030, Thermo Fisher). The sample pellets were resuspended in 5% acetonitrile, 0.1% formic acid, and 100% HPLC-grade water and then centrifuged at 6,000 × g for 30 s. Samples were subsequently transferred to a centrifuge tube filter with a 0.22 μm pore (Costar, Glendale, AZ, USA) and centrifuged at 20,000 × g for 10 min at 4 °C. The filtered samples were transferred into polymer feet autosampler inserts (Agilent Technologies) in LC/MS autosampler vials and stored at −20 °C until LC-MS analysis. For untargeted metabolomics, metabolites were separated using SeQuant ZIC-pHILIC columns (150 mm × 2.1 mm; 5 μm particle size, 150454, Sigma–Aldrich). Mobile phase A consisted of 10 mM ammonium acetate in water, pH 9.8, and mobile phase B consisted of 100% methanol. The column flow rate was set to 0.15 ml/min, and column temperature was set to 25 °C. Metabolites were separated over a 19-min gradient from 90% B to 30% B; the column was then washed for 5 min at 30% phase B, followed by re-equilibration at 90% B for 8 min. Metabolites were analyzed on an Orbitrap Exploris 240 mass spectrometer coupled to a Vanquish Neo UHPLC system (Thermo Fisher). For each polarity set, polarity-switching MS1 only acquisition was acquired at 120,000 FWHM resolution with a scan range of 80–1200 Da, an automatic gain control (AGC) target set to ‘Custom’ at 1E6 absolute AGC value, and maximum inject time of 50 ms. Compound Discoverer 3.3 SP (Thermo Fisher) was used for data analysis using the Untargeted Metabolite Processing Workflow. Data were searched against the following ChemSpider databases: BioCyc, Human Metabolome, and KEGG, to identify compound names. Metabolomic output was analyzed and visualized using the MetaboAnalystR R package [[107]33]. To test significance of clusters on the Principal Component Analysis (PCA), a permutational multivariate ANOVA (PERMANOVA) was used. For differential expression of metabolites, t-tests were used. The MetaboAnalyst 6.0 web-based platform [[108]34] was also used for joint pathway analysis using the gene list of differentially expressed genes from the RNA-seq together with the differentially expressed metabolite list. For all metabolomics statistical analyses, FDR <0.05 was used as the threshold for significance. 2.16. Seahorse XF analysis Hindlimb muscles from 6-month-old untreated miR-1-1^f/f; miR-1-2^f/f mice (WT) or HSA-miR-1 mice (miR-1 KO) were digested using a GentleMACS Octo Dissociator (130-096-427, Miltenyi Biotec, North Rhine-Westphalia, Germany) with a Skeletal Muscle Dissociation Kit, mouse and rat (130-098-305, Miltenyi Biotec), according to manufacturer's instructions. Following digestion, MPCs were isolated using an autoMACS Pro Separator (130-092-545, Miltenyi Biotec) with a Satellite Cell Isolation Kit, mouse (130-104-268, Miltenyi Biotec). Primary MPCs were expanded on 10% Matrigel-coated (Corning) Primaria culture plates (353846, Corning) in growth media consisting of Hams F-10 (10-070-CV, Corning), 20% FBS (35-070-CV, Corning), 1% Pen-Strep (VWR), and 10 ng/ml bFGF (Corning). MPCs were split by mild trypsinization (L0154-01000, VWR) at around 40% confluence until time of differentiation. For differentiation to myotubes, MPCs were allowed to reach 85–95% confluency before switching from growth medium to differentiation medium, consisting of DMEM (30–2006, ATCC) supplemented with 2% horse serum (35-030-CV, Corning). MPCs were allowed to differentiate for 5 days before adding differentiation medium supplemented with 50 nM 4-OH TAM (89152-604, VWR) for 48 h to induce recombination in HSA-miR-1-derived myotubes (or for TAM control in WT-derived myotubes). Following 48 h of 4-OH TAM treatment, myotubes were subjected to either ECAR assessment or RNA isolation and qPCR as described above. For ECAR experiments, culture and differentiation of myotubes were performed in Matrigel-coated Seahorse XF24 cell culture microplates (100882-004, Agilent Technologies, Santa Clara, CA). Before ECAR analysis using a Seahorse XF24 Analyzer (Agilent Technologies), media was replaced with Seahorse XF DMEM assay media (103575-100, Agilent Technologies). ECAR was assessed using a Seahorse XF Glycolysis Stress Test Kit (103020-100, Agilent Technologies). For assessment of metabolic flexibility in vitro, following 48 h of 4-OH TAM treatment, WT and miR-1 KO myotubes were exposed to glucose starvation for 24 h (“fasting-mimicking”, glucose-free DMEM incubation). Glucose (10 mM) and pyruvate (1 mM) (103680-100, Agilent Technologies) were then added to myotubes for 30 min (“refeeding”), and oxygen consumption rate (OCR) was immediately measured using a Seahorse XF Cell Mito Stress Test Kit (103015-100, Agilent Technologies). Immediately after ECAR and OCR measurement, protein concentrations were determined using a BCA assay, for protein normalization. All data were generated from four technical replicates (averaged) for each biological replicate (n = 3 WT and n = 4 KO). 2.17. Ex vivo [^3H]-2-deoxy-D-glucose uptake and lactate secretion Ex vivo glucose uptake and lactate secretion were measured in soleus and EDL muscle, as previously described [[109]35]. For glucose uptake, muscles were incubated in continuously gassed (95% O[2], 5% CO[2], Automated Multi-Myograph System 630 MA, Danish Myo Technology, Hinnerup, Denmark) 37 °C Krebs-Ringer-Bicarbonate (KRB) buffer (J67591.K2, Thermo Fisher) containing 1 mM [^3H]-2-deoxy-d-glucose (NET54900, Perkin Elmer, Waltham, MA, USA), 9 mM [^14C]-mannitol (NEC314, Perkin Elmer), for 15 min, in the presence or absence of insulin (200 μU/mL, 11376497001, Sigma–Aldrich). Contralateral muscles were used for basal and insulin-stimulated measurements. After incubation, muscles were blotted dry, snap-frozen, and stored at −80 °C until analyzed with liquid scintillation counting of [3H] and [^14C] labels for calculation of [^3H]-2-deoxyglucose uptake. For lactate secretion, muscles were incubated in continuously gassed KRB buffer supplemented with 2 mM pyruvate and 5.5 mM d-glucose for 60 min, and buffer lactate levels were assessed using a lactate colorimetric assay kit (10186-852, BioVision, Milpitas, CA, USA). 2.18. MCT4 inhibition VB124, a potent and selective small-molecule MCT4 inhibitor [[110]36,[111]37], was purchased from a commercial source (HY-139665, MedChemExpress, Monmouth Junction, NJ, USA). VB124 was dissolved in 0.5% methylcellulose and 0.1% Tween-20. Vehicle (VEH) control was 0.5% methylcellulose, 0.1% Tween-20 solution without VB124. For the MCT4 inhibition experiments, mice were given VB124 (30 mg/kg/day in 200 μl) or VEH (200 μl) by oral gavage once per day. For miR-1 KO + VEH vs miR-1 KO + VB124 voluntary wheel running experiments, mice underwent an 8-week washout after tamoxifen treatment and then were single-housed in voluntary wheel cages for 10 days. The first week was used for acclimation (7 days), then data over the subsequent 3 days were averaged. Mice were gavaged with vehicle or VB124 daily during this 3-day running period. 2.19. Synergist ablation MOV of the plantaris muscle was induced by bilateral synergist ablation surgery [[112]38,[113]39]. Mice were anesthetized (isoflurane 1–2%) and placed in prone position, with both hindlimbs immobilized. Hair was removed, then an incision was made on the skin from the mid-belly of the gastrocnemius down to the ankle to expose the Achille's tendon. The plantaris tendon was separated from the synergist muscle tendons, and a rounded probe was used to separate the fascia between the plantaris muscle and the gastrocnemius and soleus muscles. The soleus and roughly one-third of the gastrocnemius were sequentially excised, and the excision was closed using Silk 5 sutures. Sham surgery involved the same steps of synergist ablation without removal of muscle. Following surgery, mice resumed ambulatory cage behavior, and the plantaris was allowed to grow for 7 days. 2.20. Human resistance exercise training samples Human resistance exercise training vastus lateralis biopsy samples were collected in a previous study investigating skeletal muscle responses over the time course of 6 weeks of resistance exercise [[114]40]. Participants provided written informed consent prior to participation in the study. The local ethics committee of the German Sports University Cologne approved the study, which was conducted following the guidelines of the declaration of Helsinki. The lower limb exercise training regimen is described in detail in the original publication [[115]40]. Briefly, 14 young males (24 ± 3 y) conducted thrice weekly resistance exercise for a total of 14 resistance exercise training sessions. Muscle biopsies were obtained at baseline (T0), after the first (T1), 13th (T13), and after 10 days of rest, the 14th session (T14). The same samples were also used in another previously published study, which demonstrated that the protein abundance of PKM1 and PKM2 change in response to resistance exercise [[116]41]. In the current study, we aimed to assess miR-1 levels in the muscle biopsy samples at the different time points. Biopsy samples were shipped on dry ice, and RNA isolation, miR cDNA synthesis, and qPCR were performed as described above. 2.21. Statistical analyses Unless otherwise specified above, Student's t-tests (2-tailed) or two-way ANOVAs were used to determine the significance between WT and KO groups, as appropriate. For two-way ANOVAs, Tukey's multiple comparisons tests were used for post-hoc analyses. In the case of multiple t-tests, a two-stage step-up procedure (Benjamini, Krieger, and Yekutieli) was used for controlling the FDR. For human resistance exercise training samples, a repeated measures one-way ANOVA with Tukey's multiple comparisons test was used. For correlation analyses, a simple linear regression was used. Unless otherwise stated, statistical analyses were performed using GraphPad Prism software (Ver. 10.2.3), and the level of significance was set at p < 0.05. 3. Results 3.1. Generation of inducible skeletal muscle-specific miR-1 KO mouse Mature miR-1 is derived from two distinct genes that each encode a unique bicistronic primary miRNA transcript containing either miR-1-1/miR-133a2 or miR-1-2/133a1. The miRNA-1-1 cluster is derived from the lncRNA [117]NR_045331.1, while the miRNA-1-2 cluster resides within intron 12 of the Mib1 gene [[118]7]. For the first time, Wei and colleagues generated mice in which only the miR-1-1 and miR-1-2 genes were floxed without disrupting the miR-133a1/a2 loci of the respective bicistronic cluster [[119]7]. Using these floxed miR-1 mice, the authors reported that knocking out both miR-1 genes resulted in pups being born at the expected frequency but then all died before post-natal day 17 from dilated cardiomyopathy. To avoid any such developmental issues [[120]8], we generated a mouse model allowing for the specific, inducible deletion of miR-1 in adult skeletal muscle, without affecting nearby miR-133a1/2 or Mib1 expression. To this end, we crossed the skeletal muscle-specific inducible Cre mouse (HSA-MCM) [[121]10] with the floxed miR-1-1 and floxed miR-1-2 mice (miR-1-1^f/f; miR-1-2^f/f) [[122]7] to generate the HSA-MCM; miR-1-1^f/f; miR-1-2^f/f mouse, designated HSA-miR-1 ([123]Supplemental Fig. S1A). Tamoxifen-treated littermate miR-1-1^f/f; miR-1-2^f/f mice (negative for HSA-MCM) or vehicle-treated HSA-miR-1 mice served as wild-type (WT) controls. To avoid any skeletal muscle maturation issues that might arise upon miR-1 deletion, HSA-miR-1 mice were treated with tamoxifen to induce Cre-mediated inactivation of the miR-1 genes at 16 weeks of age. Following an 8-week washout period, whole-muscle miR-1 expression was ∼60 and ∼70% lower in the slow-twitch soleus and fast-twitch plantaris muscles of miR-1 KO mice, respectively, compared to WT mice ([124]Supplemental Fig. S1B–C). To assess the effectiveness of miR-1 knockdown in just myofibers, we next determined miR-1 expression of isolated single myofibers. The expression of miR-1 was reduced by >90% in miR-1 KO myofibers compared to WT myofibers, thus confirming our approach was highly effective in knocking down miR-1 expression ([125]Supplemental Fig. S1D). The discrepancy in the magnitude of miR-1 knockdown between whole-muscle and myofiber analyses likely reflects miR-1 expression in skeletal muscle stem cells. Importantly, the loss of miR-1 expression did not alter the expression of miR-133a1/2, Mib1, or miR-206 in skeletal muscle or miR-1 expression in cardiac muscle ([126]Supplemental Fig. S1E–H). There was also no difference in body weight between WT and miR-1 KO mice ([127]Supplemental Fig. S1I). Finally, we assessed the expression of Dlk-Dio3 imprinted genes which were previously shown to be significantly increased in the germline miR-1/miR-133 double-KO mouse [[128]8]. However, in our inducible, miR-1-specific KO model in adult mice, we observed a minor difference in the expression of only one gene in the Dlk1-Dio3 cluster ([129]Supplemental Fig. S2). Thus, the phenotype of miR-1 KO mice can be attributed to the specific loss of skeletal muscle miR-1 expression, which is distinct from the phenotype described in the miR-1/miR-133 double-KO mouse. 3.2. miR-1 is required for skeletal muscle metabolic flexibility A key feature of metabolic flexibility is the ability to switch from fatty acid oxidation to carbohydrate oxidation in response to an increase in physical activity. This switch in substrate utilization is most commonly measured using the respiratory exchange ratio (RER). The RER is the ratio between CO[2] production via metabolism and O[2] uptake, such that values closer to 0.7 reflect lipid oxidation, while a RER value closer to 1.0 indicates carbohydrate oxidation. In mice, the RER increases during the dark period as mice become more active and oxidize carbohydrates as a preferred energy source [[130]42]. To determine RER over a 48-hr period, miR-1 KO (n = 5) and WT (n = 5) mice underwent indirect calorimetry to measure whole-body oxygen consumption (VO[2]) and CO[2] production. In comparison to WT mice, the typical increase in RER during the dark cycle was 69% lower (p = 0.003) in miR-1 KO mice indicating an impaired ability to shift to oxidative carbohydrate metabolism ([131]Figure 1A–B). In addition, there was no difference in either oxygen consumption (VO[2]) or locomotor activity throughout the 48-hr testing period between WT and miR-1 KO mice ([132]Figure 1C–D). Thus, the difference in nighttime RER levels between WT and miR-1 KO were not caused by differences in activity levels but rather reflect a state of metabolic inflexibility. Furthermore, energy expenditure and food consumption were not different between groups ([133]Supplemental Fig. S3A–B), although water consumption tended to be lower in miR-1 KO mice compared to WT mice during the dark phase (p = 0.062) ([134]Supplemental Fig. S3C). Interestingly, water intake has previously been shown to be lower in mice fed a high-fat diet, which was thought to be related to the generation of metabolic water as a result of fat oxidation [[135]43]. Figure 1. [136]Figure 1 [137]Open in a new tab Effect of miR-1 loss on metabolic flexibility and exercise performance. (A) Respiratory exchange ratio (RER) via indirect calorimetry over a 48-hr period. (B) Box plots (min-to-max) showing median and interquartile range (IQR) for RER data. (C) Whole-body oxygen consumption (VO2) and (D) locomotor activity during indirect calorimetry period. Data in A-D are mean ± SEM, plotted using CalR (Ver. 2), and analyzed using a two-way ANOVA for light and dark cycles separately. (E) Fasting basal blood glucose levels, data analyzed using independent t-tests (2-tailed). (F) Glucose tolerance test (GTT), data are mean ± SD, data analyzed by two-way ANOVA, main effect of miR-1 knockout (KO) and interaction (genotype x time) shown. Data in A-F from n = 5 female wild-type (WT) mice and n = 5 female miR-1 KO mice. (G–H) ClockLab analyses of 4 weeks of voluntary wheel running. Average (G) daily running volume (km/day) and (H) maximum running bout (sec). Data in G-H from n = 5 female mice per group. (I–J) CeLeST analysis of (I) wave initiation rate and (J) activity index in Untrained N2 (WT) (n = 32) and mir-1 mutant worms (n = 35). (K–L) CeLeST analysis of (K) wave initiation rate and (L) activity index in Untrained (dark circle) or Exercised (open circle) N2 (n = 32 and n = 36, respectively), and Untrained (dark square) and Exercised (open square) mir-1 mutant worms (n = 35 and n = 36, respectively). Data in I-J analyzed using independent t-tests (2-tailed). Data in K-L analyzed using a two-way ANOVA, main effect of miR-1 KO, interaction (genotype x exercise), and results of post-hoc Tukey's multiple comparisons tests shown. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001, ns: not significant. Metabolic inflexibility is linked to metabolic dysfunction including insulin resistance [[138]44], although it is unknown whether metabolic inflexibility leads to insulin resistance or vice versa [[139]1]. To assess insulin sensitivity, we performed intraperitoneal glucose tolerance tests in miR-1 KO (n = 5) and WT (n = 5) mice. Although fasting basal blood glucose levels were higher in miR-1 KO mice ([140]Figure 1E), glucose tolerance was not different between groups ([141]Figure 1F), which indicates at this stage (post- 8-week washout), metabolic inflexibility precedes systemic insulin resistance. 3.3. miR-1 is necessary for endurance exercise performance A loss of metabolic flexibility is often associated with poor endurance exercise performance because of an inability to utilize the proper substrate for energy production. Thus, given the metabolic inflexibility observed in the miR-1 KO mouse, we hypothesized that voluntary wheel running, a murine model of endurance exercise, would be impaired because of an inability to switch substrate utilization to carbohydrate oxidation. To test this hypothesis, we quantified voluntary wheel running activity in WT and miR-1 KO mice. At 4 months of age, mice were treated with tamoxifen, and following a 4-week washout, mice were singly housed with access to a running wheel. After a week of acclimation, daily running volume and maximum running bout (longest continuous running period without stopping) were measured for 4 weeks. WT mice ran an average of 10.9 km/day, while miR-1 KO mice ran 4.7 km/day, a 57% reduction in wheel running distance ([142]Figure 1G). Similarly, the maximum bout length was lower in the miR-1 KO mice; the maximum bout length in WT mice averaged 134 s, while it was 43 s in miR-1 KO mice ([143]Figure 1H). The significant loss in endurance exercise performance of miR-1 KO mice suggested the loss of miR-1 impaired the ability of muscle to readily switch to carbohydrate oxidation during exercise, consistent with the RER finding. miR-1 is one of only 32 miRs found to be conserved among all Bilaterian animals [[144]45]. Given this high-degree of conservation, we next investigated the swimming performance of C. elegans of a mir-1 mutant. Similar to mammals, miR-1 is predominantly expressed in muscle of C. elegans [[145]46]. Importantly, muscle development of the mir-1 mutant appeared to be normal as adult worms had the same number of muscle fibers with the correct morphology [[146]47]. Adult N2 (WT) or mir-1 mutant worms were transferred to an unseeded NGM plate (Untrained) or an unseeded NGM plate flooded with M9 buffer (Exercise) for two 90-min sessions per day for 5 days, as previously described [[147]22]. Locomotion analysis of worm movement showed that Untrained mir-1 mutant worms had poorer swim performance parameters, including significantly lower wave initiation rate ([148]Figure 1I) and significantly lower activity index compared to Untrained N2 worms ([149]Figure 1J). Following the 5-day exercise protocol, wave initiation rate and activity index were significantly higher in N2 worms compared to Untrained N2 worms ([150]Figures 1K-L). The mir-1 mutant worms did not show a significant change in these exercise performance parameters in response to training compared to Untrained mir-1 mutant worms. While the reason for the failure of mir-1 mutant worms to show improvements in swim performance following training remains to be determined, a likely cause might be the training stimulus was not sufficient to induce muscle adaptations as a result of their low activity level. As observed with mice, miR-1 appears to be necessary in worms for endurance exercise performance which we hypothesize is required for the metabolic flexibility needed to perform high-levels of endurance exercise. 3.4. Identification of miR-1 target genes To determine how miR-1 is regulating metabolic flexibility of adult skeletal muscle, we performed RNA-seq and AGO2 eCLIP-seq. First, we used RNA-seq of total RNA from gastrocnemius muscles of WT (n = 4) and miR-1 KO (n = 4) mice to identify up-regulated differentially expressed genes (DEGs) in the miR-1 KO, with the understanding that many of these genes will be miR-1 target genes ([151]Supplemental Table S1). Next, as part of a parallel initiative to generate an atlas of empirically defined miRNA target binding sites across human tissues ([152]www.mir-engagemint.io), we performed AGO2 eCLIP on six adult (24.43 ± 5.35 y) human skeletal muscle tissue samples to generate a high-quality map of miRNA binding sites across the muscle transcriptome. AGO2 eCLIP-seq is currently the gold standard for identifying bona fide miRNA target genes because it allows for the direct biochemical determination of AGO2:mRNA binding [[153]27,[154]48,[155]49]. MicroRNA-1 and miR-206 represent a miRNA family because they share an identical seed sequence [[156]50]; however, because they share the same seed sequence it was not possible to distinguish miR-1 sites from miR-206 sites. Given that miR-1 is significantly (∼250-fold) more abundant in adult skeletal muscle than miR-206, it is reasonable to assume that miR-1/206 target genes are most likely being regulated by miR-1 [[157]6]. For clarity, we hereafter refer to miR-1/206 identified target genes as miR-1 target genes. We leveraged empirical binding events (defined as significantly enriched peaks known as clusters, FDR <0.05) with target site prediction simulation tools (TargetScan Pel Package) to identify miR-1 target sites across 1796 clusters and 1577 genes ([158]Supplemental Table S2). Integrating the RNA-seq and AGO2 eCLIP-seq data sets, we generated cumulative distribution function (CDF) plots which demonstrated that 3ʹ-UTR eCLIP-supported miR-1 target genes were globally up-regulated in skeletal muscle of the miR-1 KO (as indicated by a rightward shift, p = 0.0034, KS = 1.787) relative to non-miR-1 targets ([159]Figure 2A). Similar to findings in the liver when combining AGO2 eCLIP-seq-defined targets and liver-specific miR-122 KO mice [[160]51], ∼21% of the up-regulated DEGs in the miR-1 KO muscles were miR-1 target genes, thus indicating the loss of miR-1 regulation also had a major influence on the expression of non-target genes ([161]Supplemental Table S3). Figure 2. [162]Figure 2 [163]Open in a new tab Transcriptomic profiling of miR-1 KO skeletal muscle and identification of bona fide miR-1 target genes. (A) Cumulative density function (CDF) of AGO2 eCLIP-seq-defined miR-1 targets (green) Log2FC compared to non-miR-1 targets (blue). Zoomed in figure emphasizes rightward shift of miR-1 targets, indicated by arrow. RNA-seq data for integration with AGO2 eCLIP-seq are from gastrocnemius muscles (n = 4 female WT, n = 4 female miR-1 KO mice). (B) Pathway enrichment analysis of significantly up-regulated genes (false discovery rate, FDR <0.05, Log2 fold change, FC > 0) in miR-1 KO/WT that are AGO2 eCLIP-seq-defined miR-1 target genes. (C–D) miR-1 binding peaks in (C) polypyrimidine tract-binding protein 1 (PTBP1) and (D) pyruvate kinase M1/2 (PKM) mRNA outlined with red dotted line, miR-1 alignment on mouse target sequence shown below. (E) Representative Western blot of PTBP1 and corresponding total protein levels in WT and miR-1 KO. (F) Quantification of PTBP1 protein levels after densitometric analysis of the levels of each sample normalized to corresponding total protein levels, expressed as fold-change (FC). (G) Representative Western blot of PKM1 and PKM2, (H–I) Quantification of (H) PKM1 and (I) PKM2 protein levels. (J–K) miR-1 binding peaks in (J) lactate dehydrogenase A (LDHA) and (K) solute carrier family 16 member 3 (SLC16A3) mRNA. (L) Western blot of LDH, (M) quantification of LDH protein levels. (N) Western blot of monocarboxylate transporter 4 (MCT4), (O) quantification of MCT4 protein levels. For Western blotting experiments, gastrocnemius muscle lysates from n = 4–8 WT and n = 4–8 miR-1 KO females were used, and differences between WT and miR-1 KO were tested using an independent t-test (2-tailed). ∗p < 0.05, ∗∗∗p < 0.01, ∗∗∗∗p < 0.0001, ns: not significant. 3.5. miR-1 target genes involved in skeletal muscle pyruvate metabolism Pathway analysis of the 251 AGO2 eCLIP-seq-defined miR-1 target genes that were up-regulated in miR-1 KO muscle showed the most significantly enriched pathways were associated with glucose metabolism, specifically aerobic glycolysis ([164]Figure 2B, [165]Supplemental Table S4). The up-regulated miR-1 target gene pyruvate kinase M1/2 (Pkm) was of particular interest to us because of its known role in metabolic flexibility through the regulation of pyruvate metabolism. PKM has two isoforms that are generated through alterative splicing, with PKM1 promoting pyruvate flux to the mitochondrial TCA cycle. PKM2, on the other hand, is expressed primarily in proliferating cells and tumor cells and promotes aerobic glycolysis and lactate generation via lactate dehydrogenase (LDH) [[166]52]. The polypyrimidine tract binding protein 1 (PTBP1) is a splicing factor that is known to promote PKM2 expression through alternative splicing that leads to exon 9 replacing exon 10 in the PKM mRNA. Interestingly, previous studies in cancer found that elevated expression of PTBP1 promotes the Warburg effect, (i.e., aerobic glycolysis) by generating the PKM2 isoform [[167]53]. Our AGO2 eCLIP-seq analyses identified miR-1 binding peaks in both PTBP1 and PKM mRNAs, with ScanMiR [[168]54] confirming miR-1 binding sites in the orthologous mouse genes ([169]Figure 2C–D). RNA-seq analysis showed the expression of the Ptbp1 and Pkm transcripts were significantly higher in gastrocnemius of miR-1 KO compared to WT ([170]Supplemental Fig. S4A–B). We next performed western blot analysis, metabolomics, and mitochondrial bioenergetic phenotyping to determine if modulation of miR-1 target genes involved in pyruvate metabolism (i.e., Ptbp1 and Pkm) regulates the fate of pyruvate. Western blot analysis of whole-muscle lysates showed PTBP1 protein levels were >10-fold higher in miR-1 KO compared to WT ([171]Figure 2E–F). Furthermore, PKM1 protein expression was not different, while PKM2 protein expression was >2-fold higher in the miR-1 KO compared to WT ([172]Figure 2G–I). AGO2 eCLIP-seq also demonstrated miR-1 binding peaks in lactate dehydrogenase A (LDHA), the enzyme that catalyzes the conversion of pyruvate to lactate ([173]Figure 2J) as well as solute carrier family 16 member 3 (SLC16A3) ([174]Figure 2K), which encodes the lactate transporter monocarboxylic acid transporter 4 (MCT4). Ldha and Slc16a3 expression were significantly higher in RNA-seq analysis of miR-1 KO compared to WT mice ([175]Supplemental Fig. S4C–D). Protein levels of both LDH and MCT4 were significantly higher in miR-1 KO compared to WT (2.4-fold and 5-fold, respectively) ([176]Figure 2L–O). The remainder of the glycolytic genes up-regulated in miR-1 KO muscle (both miR-1 targets and non-miR-1 targets) are presented in [177]Supplemental Figures S4E-L. Notably, several of the significantly up-regulated glycolytic genes are highly expressed in skeletal muscle ([178]Supplemental Fig. S4M). Together, these findings demonstrate that upon the loss of miR-1, skeletal muscle underwent metabolic reprogramming to aerobic glycolysis as a consequence of the up-regulation of miR-1 targets including Ptbp1, Pkm2, Ldha, and Slc16a3. 3.6. Loss of miR-1 alters the skeletal muscle metabolome To determine if the change in pyruvate metabolism suggested at the transcript and protein levels in the miR-1 KO was observed at the metabolic level, we performed untargeted metabolomics (ultra-high performance liquid chromatography-mass spectrometry [UHPLC-MS]) on gastrocnemius muscle from WT (n = 4) and miR-1 KO (n = 4) mice. The metabolomic profiles between WT and miR-1 KO mice displayed clear and significant differences as observed by principal component analysis (PCA) (p = 0.024, [179]Figure 3A). In complete agreement with the gene expression data, pathway analysis of differentially abundant metabolites identified pyruvate metabolism as the most significantly enriched pathway that was different between WT and miR-1 KO ([180]Figure 3B, [181]Supplemental Table S5). Other metabolomic pathways enriched in miR-1 KO muscle compared to WT included amino sugar and nucleotide sugar metabolism. This observation confirms our previous study proposing miR-1 regulation of the pentose phosphate pathway, as sugar backbones for all nucleotides are derived from ribose-5-phosphate (Ru5P) in the pentose phosphate pathway [[182]55]. In addition to Ru5P, the abundance of other glycolytic metabolites such as glyceraldehyde 3-phosphate (GADP), 3-phosphoglycerate (3 PG), phosphoenolpyruvate (PEP), and lactic acid were significantly higher in miR-1 KO muscles ([183]Supplemental Fig. S5A–E). Since amino acids play a role in the synthesis of nucleotides, we assessed the expression levels of several critical amino acid metabolism genes in our RNA-seq dataset. A number of these genes were differentially expressed, including glutamic--pyruvic transaminase 2 (Gpt2), glutamate-ammonia ligase (Glul), and malate dehydrogenase 1 (Mdh1), which were identified by our AGO2 eCLIP-seq as miR-1 targets ([184]Supplemental Fig. S5F–H), and the non-miR-1 target genes, glutamic--pyruvic transaminase 1 (Gpt), glutamine phosphoribosylpyrophosphate (Ppat), solute carrier family 25 member 22 (Slc25a22), and aspartate aminotransferase (Got1) ([185]Supplemental Fig. S5I-L). Figure 3. [186]Figure 3 [187]Open in a new tab Skeletal muscle metabolomic profiling of miR-1 KO skeletal muscle. (A) Principal component analysis (PCA) scores plot generated using MetaboAnalyst based on gastrocnemius metabolites in n = 4 WT and n = 4 miR-1 KO female mice. Predictive component (PC) 1 and PC2 can differentiate the WT and miR-1 KO muscle. (B) Summary of altered metabolic pathways analysis with MetaboAnalyst reflecting the impact on the pathway and the level of significance. The colors of dots (varying from yellow to red) indicate the significance of the metabolites in the data, and the size of the dot is positively corelated with the impact of the metabolic pathway. Top 6 pathways labeled. (C) miR-1 expression of 4-Hydroxytamoxifen (4-OH TAM)-treated myotubes from WT or miR-1 KO mice (n = 3 untreated female mice per group). (D) Extracellular acidification rate (ECAR) trace over time after injection of indicated glycolytic modulators in WT (n = 3 female) and miR-1 KO (n = 4 female)-derived myotubes. Oligo: oligomycin, 2-DG: 2-deoxy-d-glucose. (E)Ptbp1 and (F)Pkm2 mRNA expression of WT and miR-1 KO myotubes. (G) Oxygen consumption rate (OCR) trace over time after injection of mitochondrial respiration modulators in WT (n = 3 female) and miR-1 KO (n = 4 female)-derived myotubes that were exposed to glucose starvation followed by re-addition. FCCP: carbonyl cyanide-p-trifluoromethoxyphenylhydrazone, AmA/ROT: antimycin A/rotenone. Data in C, E-F analyzed using independent t-tests (2-tailed). Data in D, G-H analyzed by two-way ANOVA, main effect of miR-1 KO, interaction (genotype x substrates), and results of post-hoc Tukey's multiple comparisons tests shown. ∗p < 0.05, ∗∗p < 0.01. 3.7. Loss of miR-1 leads to increased glycolysis in myotubes in vitro To more rigorously investigate miR-1 regulation of pyruvate metabolism in muscle cells we isolated myogenic progenitor cells (MPCs) from untreated WT or HSA-miR-1 mice at 6 months of age. MPCs differentiated for 5 days, and then myotubes were treated for 48 h with 50 nM 4-hydroxytamoxifen (4-OH TAM) to knockout miR-1 in HSA-miR-1-derived myotubes. 4-OH TAM induced Cre-mediated recombination effectively knocked down miR-1 by >90% in HSA-miR-1 myotubes compared to WT myotubes ([188]Figure 3C). Next, WT and miR-1 KO myotubes were analyzed using a Seahorse Bioanalyzer to determine the extracellular acidification rate (ECAR), a proxy measure of the lactate produced by glycolysis ([189]Figure 3D). Basal and maximal ECAR were significantly higher in miR-1 KO myotubes compared to WT, indicating greater lactate production as a result of enhanced glycolysis ([190]Figure 3D). Consistent with the higher rate of glycolysis in the miR-1 KO myotubes, Ptbp1 and Pkm2 mRNA expression were significantly higher in miR-1 KO myotubes relative to WT ([191]Figure 3E–F). Together, these data demonstrate that the loss of miR-1 directly promotes aerobic glycolysis in muscle cells as a result of higher expression of Ptbp1, a target of miR-1, which, in turn, causes greater Pkm2 expression via alternative splicing of the Pkm gene. To assess metabolic flexibility in vitro, we measured oxygen consumption rate (OCR) in myotubes that were re-fed glucose after 24 h of glucose starvation. The OCR in response to glucose re-addition was significantly lower in miR-1 KO myotubes compared to WT ([192]Figure 3G). In contrast, ECAR was significantly higher in miR-1 KO myotubes after glucose re-addition following fasting ([193]Figure 3H). This further suggests that miR-1 is necessary for the increase in glucose oxidation after fasting/refeeding. 3.8. Bioenergetic phenotyping of miR-1 KO mitochondria confirms miR-1 is necessary for mitochondrial pyruvate oxidation To investigate if miR-1 regulation of pyruvate metabolism altered skeletal muscle bioenergetics, we performed high-resolution respirometry (HRR) to assess mitochondrial respiration of permeabilized soleus fibers. Pyruvate/malate (Pyr/Mal)-supported LEAK and ADP-stimulated respiration were significantly lower in the miR-1 KO compared to WT ([194]Figure 4A). However, there was no difference in fatty acid-supported (octanoyl-carnitine/malate, Oct/Mal) LEAK or ADP-stimulated respiration between miR-1 KO and WT soleus muscles ([195]Figure 4A). Figure 4. [196]Figure 4 [197]Open in a new tab Bioenergetic phenotyping of miR-1 KO skeletal muscle. (A) Assessment of mitochondrial respiration in permeabilized soleus fibers (n = 8 female WT or miR-1 KO mice per group). Oxygen flux (JO[2]) normalized to mg tissue dry weight. Pyruvate/Malate-supported complex I leak (Pyr/Mal) and ADP-stimulated OXPHOS (Pyr/Mal/ADP), Octanoyl-carnitine/Malate-supported complex I leak (Oct/Mal) and ADP-stimulated OXPHOS (Oct/Mal/ADP). (B) Citrate synthase (CS) activity in gastrocnemius complex lysates, normalized to protein. (C) Quantification of mitochondrial DNA (mtDNA) by levels of NADH Dehydrogenase 1 (Nd1), and Mito1 relative to nuclear DNA (nDNA) in gastrocnemius muscle. Data in B–C from n = 6–9 female mice per group. (D) Assessment of OXPHOS kinetics using the creatine kinase (CK) clamp technique (increasing free energies [i.e., more negative ΔG[ATP] values correspond to an increased ATP/ADP ratio)] with Pyr/Mal and succinate as substrates in permeabilized fibers from the gastrocnemius, data normalized to mg dry weight. (E) Assessment of OXPHOS kinetics using the CK clamp technique with Pyr/Mal as substrates in isolated mitochondria, data normalized to total protein. (F) H[2]O[2] emission rate (JH[2]O[2]) assessed in isolated mitochondria in response to Pyr/Mal using a CK clamp, normalized to total protein. (G) Mitochondrial membrane potential (ΔΨ), expressed in mV, in response to Pyr/Mal using a CK clamp. Oligo: oligomycin, CN: cyanide. (H) OXPHOS kinetics using the CK clamp technique with glutamate/malate (G/M) as substrates in isolated mitochondria, data normalized to total protein. (I) OXPHOS kinetics using the CK clamp technique with octanoyl-carnitine/malate (Oct/Mal) as substrates in isolated mitochondria, data normalized to total protein. (J) [^3H]-2-Deoxyglucose uptake in EDL muscles with or without 200 μU/mL of insulin. Data in D-J from n = 3–4 WT and n = 3–4 miR-1 KO female mice, and data are mean ± SEM. Data in A, D-I analyzed by two-way ANOVA, main effect of miR-1 KO, interaction (genotype x ADP for A, genotype x ΔGATP for D-I), and results of post-hoc Tukey's multiple comparisons tests shown. Data in B–C analyzed by t-tests (2-tailed). Data in J analyzed by a two-way ANOVA, main effect of miR-1 KO and interaction effect (genotype x insulin), and results of post-hoc Tukey's multiple comparisons tests shown. ns: not significant, ∗p < 0.05, ∗∗p < 0.01. The lower mitochondrial respiration of permeabilized miR-1 KO myofibers could be due to reduced mitochondrial content or metabolic suppression independent of mitochondrial content/density. Therefore, we measured citrate synthase (CS) activity and the ratio of mitochondrial DNA (mtDNA) to nuclear DNA (nDNA) in gastrocnemius complex (gastrocnemius, soleus, and plantaris) samples to assess mitochondrial content. We found no difference between WT and miR-1 KO in either measure of mitochondrial content ([198]Figure 4B–C). Thus, in agreement with reduced pyruvate oxidation observed in sedentary mice [[199]42], our data demonstrate glucose metabolism can be significantly altered without a change in mitochondrial content of the muscle. To measure mitochondrial respiration under conditions that are more physiological, we applied a modified version of the creatine kinase (CK) energetic clamp to assess mitochondrial energy transduction across a range of energy demands [[200]28]. In permeabilized gastrocnemius fibers, miR-1 KO mice displayed significantly lower Pyr/Mal-stimulated mitochondrial respiration across a range of ATP free-energy states (23.41% lower on average, p = 0.013, [201]Figure 4D). We next tested different respiratory substrates in isolated mitochondria prepared from the gastrocnemius complex. In contrast to the findings from permeabilized fibers, Pyr/Mal-stimulated respiration was significantly higher in miR-1 KO mitochondria compared to WT across increasing ATP free energies (26.66% higher on average, p < 0.0001, [202]Figure 4E). The discrepancy in mitochondrial respiration between myofibers and isolated mitochondria is likely due to the fact that the miR-1 targets involved in pyruvate metabolism are all cytosolic, which are absent in isolated mitochondrial preparations [[203]56]. Pyr/Mal-stimulated respiration was also associated with elevated hydrogen peroxide emission (JH[2]O[2]) and hyper-polarization of the mitochondrial membrane potential (ΔΨ) across a range of ATP free energy in miR-1 KO mitochondria (39.03% and 49.31% higher on average, respectively, p < 0.0001 for both, [204]Figure 4F–G). These differences in mitochondrial membrane potential and hydrogen peroxide emission suggest that altered mitochondrial coupling efficiency and proton leak in response to pyruvate may also contribute to the oxygen flux response in isolated mitochondria from miR-1 KO muscle. There was no difference between WT and miR-1 KO mitochondrial oxygen consumption with glutamate/malate ([205]Figure 4H) or fatty acid substrates (Oct/Mal) ([206]Figure 4I). The higher pyruvate-stimulated respiration in miR-1 KO mitochondria, without a change in fatty acid-stimulated respiration, is indicative of pyruvate oxidation resistance and metabolic inflexibility [[207]42]. To examine whether the lower pyruvate oxidation was due to lower glucose uptake, EDL muscles were excised from WT and miR-1 KO mice for the measurement of ex vivo [^3H]-2-deoxyglucose uptake. There was no difference in basal glucose uptake between WT and miR-1 KO muscles, and in fact, insulin-stimulated glucose uptake was significantly higher in miR-1 KO muscles ([208]Figure 4J). Consistent with these findings, our RNA-seq analyses showed that genes encoding glucose transporter 3 (GLUT3) and glucose transporter 4 (GLUT4), (Slc2a3 and Slc2a4) were significantly higher in miR-1 KO muscle compared to WT ([209]Supplemental Figures S5M−N). Taken together, these data further suggest that pyruvate oxidation resistance in miR-1 KO mitochondria is downstream of glucose uptake. We combined our bioenergetic assessments with mitochondrial-targeted nano-liquid chromatography mass spectrometry (nLC-MS/MS) proteomics for further evaluation of mitochondrial enrichment [[210]30]. Proteomic data analysis showed that the mitochondrial enrichment factor (MEF), calculated as the proportion of all quantified proteins that could be identified as mitochondrial using the MitoCarta 3.0 database [[211]31], was not different between WT and miR-1 KO mitochondrial preparations ([212]Figure 5A). Moreover, the percentage of the mitochondrial proteome devoted to each OXPHOS complex was not different between WT and miR-1 KO mitochondrial preparations when normalized to total protein ([213]Figure 5B). These findings further demonstrate the altered pyruvate oxidation in miR-1 KO muscle occurred without changes in mitochondrial content. Figure 5. [214]Figure 5 [215]Open in a new tab Mitochondrial-enriched proteomics of miR-1 KO skeletal muscle. (A) Ratio of mitochondrial protein to total protein abundance across samples, referred to as Mitochondrial Enrichment Factor (MEF). (B) Quantification of the OXPHOS protein complexes generated by the summed abundance of all subunits within a given complex. Data are presented as a percentage of the max for each complex. (C–D) Hierarchically clustered heatmap of Pearson's correlation matrix for electron transport chain (ETC) proteins in (C) WT and (D) miR-1 KO mitochondria. (E) Pearson's correlation measurements comparing WT and miR-1 KO for each ETC complex. (F) Volcano plot depicting changes in the skeletal muscle mitochondrial proteome. Red color indicates significance (p < 0.05), and differentially expressed proteins involved in pyruvate metabolism are labeled. Data in A-C from n = 4 WT and n = 4 miR-1 KO female mice. (G) Representative Western blot of phosphorylation of Ser 293 on the PDHE1a subunit [p-PDHE1a(Ser293)], total PDHE1a, and corresponding total protein levels in WT and miR-1 KO gastrocnemius muscle lysates. (H) Quantification of p-PDHE1a/total PDHE1a protein levels after densitometric analysis of the levels of each sample (n = 8 WT and n = 8 miR-1 KO) normalized to corresponding total protein levels, expressed as a ratio. (I) Quantification of total PDHE1a protein levels after densitometric analysis of the levels of each sample normalized to corresponding total protein levels, expressed as FC. (J–K) Buffer lactate levels measured as an indicator of lactate secretion of (J) soleus muscles and (K) EDL muscles, n = 8 female mice per group. Data in A, H–K analyzed using independent t-tests (2-tailed), data in B analyzed using a two-way ANOVA, main effect of miR-1 KO, interaction (genotype x ETC complex), and results of post-hoc Tukey's multiple comparisons tests shown, ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. Furthermore, correlation matrices showed no difference in the associations between respiratory proteins in WT and miR-1 KO mitochondria ([216]Figure 5C–D). Statistical comparisons of pairwise correlations for each protein complex demonstrated no difference in the stoichiometric proportions of each protein complex between WT and miR-1 KO ([217]Figure 5E). We next focused on pyruvate metabolism-related proteins that were differentially expressed in the mitochondrial-targeted proteomics. As shown in [218]Figure 5F, lactate dehydrogenase D (LDHD) and pyruvate dehydrogenase kinase 4 (PDK4) were significantly up-regulated in miR-1 KO mitochondria compared to WT, while pyruvate carboxylase (PCX) was significantly down-regulated ([219]Supplemental Table S6). To determine if higher PDK4 expression was associated with greater phosphorylation of pyruvate dehydrogenase (PDH), phosphorylation levels of the E1α (Ser293) subunit of PDH were assessed relative to total PDH levels ([220]Figure 5G). Strikingly, we observed a significant up-regulation of the phosphorylation status of PDH ([221]Figure 5H), as well as higher total PDHE1a levels ([222]Figure 5I) in gastrocnemius muscles of miR-1 KO mice. The hyper-phosphorylation of PDH leads to enzyme inactivation, which causes reduced conversion of pyruvate to acetyl-CoA, and instead, increased conversion to lactate [[223]57]. This type of metabolic reprogramming is now recognized as causing a state of metabolic inflexibility [[224]1]. To this end, we measured ex vivo muscle lactate secretion, which demonstrated a significant increase in buffer lactate levels in glycolytic EDL muscles, but not in soleus muscles of miR-1 KO mice compared to WT ([225]Figure 5J–K). Taken together, these data suggest that high miR-1 levels are required for skeletal muscle to maintain normal pyruvate metabolism and metabolic flexibility. 3.9. Pharmacological inhibition of MCT4 rescues metabolic inflexibility in miR-1 KO mice We hypothesized that redirecting glucose carbon flux away from lactate secretion and toward oxidation would improve metabolic flexibility in the miR-1 KO mouse. To test this, we used a selective MCT4 inhibitor (VB124), which has been shown to cause LDH to produce more pyruvate that is shunted toward mitochondrial oxidation [[226]36,[227]37]. First, we assessed RER over a period of 48 h in miR-1 KO mice, which were administered by gavage either vehicle (VEH) or 30 mg/kg/day of VB124. Mice administered VB124 had significantly higher RER during the dark cycle (p = 0.034), with no difference in RER during the light cycle (p = 0.220) ([228]Figure 6A–B), and no difference in either VO[2] ([229]Figure 6C) or activity levels ([230]Figure 6D). There were also no significant differences in energy expenditure, food consumption, or water consumption between mice administered with VEH or VB124 ([231]Supplemental Fig. S6A–C). Consistent with the rescue in metabolic flexibility (increase in RER), mice administered with VB124 during a 3-day period of voluntary wheel running showed a significantly higher daily running volume ([232]Figure 6E). Thus, by promoting pyruvate oxidation through inhibition of MCT4, we were able to restore metabolic flexibility and improve endurance exercise performance despite the loss of miR-1. Figure 6. [233]Figure 6 [234]Open in a new tab Effect of MCT4 inhibition on metabolic flexibility and exercise performance in miR-1 KO mice. (A) RER via indirect calorimetry over a 48-hr period in miR-1 KO mice administered vehicle (VEH) or VB124 (30 mg/kg) by gavage once daily (n = 5 female mice per group). WT trace from data in [235]Figure 1 shown for comparison. (B) Box plots (min-to-max) showing median and IQR for RER data in A. (C) Whole-body oxygen consumption (VO2) and (D) locomotor activity during indirect calorimetry period. Data in A, C-D are mean ± SEM, data plotted using CalR (Ver. 2). Data analyzed using a two-way ANOVA for light and dark cycles separately. (E) Average daily running volume (km/day) over a period of 3 days in miR-1 KO mice administered VEH or VB124 by gavage once daily (n = 5 female mice per group). Data in E analyzed using an independent t-test (2-tailed), ∗∗p < 0.01. 3.10. Down-regulation of miR-1 may play a role in rewired glucose metabolism during skeletal muscle hypertrophy in mice and humans In response a hypertrophic stimulus induced by mechanical overload (MOV), we reported that miR-1 expression is rapidly down-regulated by ∼70% in the mouse plantaris muscle [[236]58,[237]59]. Thus, we hypothesized the loss of miR-1 would be sufficient to induce muscle hypertrophy in the miR-1 KO. In contrast to our hypothesis, however, the loss of miR-1 for 8 weeks in the miR-1 KO did not affect the muscle fiber size of fast-twitch, glycolytic plantaris or slow-twitch, oxidative soleus muscles ([238]Supplemental Fig. S7A–B). There was also no difference in fiber-type-specific cross-sectional area (CSA) in plantaris or soleus muscles ([239]Supplemental Fig. S7C–J). Moreover, there were only nominal differences in fiber-type composition between miR-1 KO and WT plantaris and soleus muscles which pointed towards a minor transition towards more fast-twitch, glycolytic fibers in the KO mice ([240]Supplemental Figures S7K-L). Thus, the down-regulation of miR-1 is not sufficient for hypertrophic growth nor necessary to maintain fiber-type composition of adult skeletal muscle. We next wanted to know if the genetic (miR-1 KO) and physiological (MOV-induced) knockdown of miR-1 regulated a common pathway as reflected in a shared set of up-regulated genes. We compared genes up-regulated in miR-1 KO relative to WT, genes up-regulated in MOV relative to sham control [[241]15], and our eCLIP-seq-identified miR-1 target genes ([242]Supplemental Table S7). This integration identified 12 common miR-1 target genes that were up-regulated in both the miR-1 KO and MOV-induced hypertrophy, which included Ptbp1 ([243]Figure 7A). To determine if the up-regulation of Ptbp1 with MOV was also associated with higher PKM2 expression, we performed western blot analysis. In agreement with the miR-1 KO, PKM2 expression was elevated in response to MOV compared to sham control muscle, while PKM1 expression was not different ([244]Figure 7B–D). Figure 7. [245]Figure 7 [246]Open in a new tab miR-1 down-regulation during MOV-induced muscle hypertrophy. (A) Venn diagram comparing significantly up-regulated genes in miR-1 KO compared to WT (“miR-1 KO up”), significantly up-regulated genes following 3 days of synergist ablation-induced MOV compared to sham (“MOV up”), and eCLIP-seq-defined miR-1 targets (“CLIP target”). Consensus genes listed. Venn diagram generated using [247]https://bioinformatics.psb.ugent.be/webtools/Venn/. (B) Western blot of PKM1 and PKM2 in mice subjected to sham surgery (SH, n = 13 female mice) or synergist ablation-induced mechanical overload (MOV, n = 9 female mice), (C–D) Quantification of (C) PKM1 and (D) PKM2 protein levels. (E) Outline of human resistance exercise training program and times of biopsy collection (T0: baseline, T1: after the 1st training session, T13: after the 13th training session, T14: after the 14th training session). (F) miR-1 expression in human skeletal muscle biopsies (from n = 14 males) at the different time points of the resistance exercise training program. Differences in miR-1 expression at the different time points tested using a repeated-measures ANOVA with Tukey's multiple comparisons. (G–H) Association between the change (Δ) in miR-1 expression and the Δ PKM1 protein levels at (G) T13 and (H) T14. (I–J) Association between miR-1 expression and PTBP1 mRNA expression at (I) T0 and (J) T13. Associations tested using simple linear regressions, p-values shown. (K) Type II fiber hypertrophy, demonstrated as percent change in Type II fiber CSA from T0 to T14. Participants divided into Low (n = 7) and High (n = 7) groups based on magnitude of CSA increases. (L) Percent change in miR-1 expression from T0 to T14 in participants in the Low and High fiber hypertrophy groups. Data in J-K analyzed using independent t-tests (2-tailed), ns: not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.0001. Verbrugge and colleagues reported that six weeks of resistance exercise (RE) caused an increase in skeletal muscle PKM2 expression, while PKM1 expression was down-regulated [[248]41]. To determine if the up-regulation of PKM2 following RE was associated with the down-regulation of miR-1, as observed in both the miR-1 KO and in response to MOV, we obtained human skeletal muscle biopsies from the participants of the aforementioned study ([249]Figure 7E) [[250]41]. As shown in [251]Figure 7F, miR-1 expression was significantly decreased after the first resistance training session (T1) compared to baseline (T0). The expression of miR-1 remained significantly lower after the 13th training session (5 weeks, T13) but increased after 10 days of detraining. The reduction in miR-1 expression at T13 and T14 tended to correlate with the reduction in PKM1 at the corresponding time-point (p = 0.07 and p = 0.06, respectively, [252]Figure 7G–H). In addition, higher miR-1 levels tended to correlate with lower levels of PTBP1 both at baseline and at T13 (p = 0.09 and p = 0.06, respectively, [253]Figure 7I–J). Finally, stratifying participants into “Low” (<10%, n = 7) and “High” (≥10%, n = 7) magnitude of muscle hypertrophy based on Type II fiber CSA changes from T0 to T14 ([254]Figure 7K) found the reduction in miR-1 was significantly greater in participants in the High fiber hypertrophy group ([255]Figure 7L). These findings indicate that the miR-1>PTBP1>PKM2 axis was conserved in humans following RE and further suggest the metabolic reprogramming of skeletal muscle to a cancer-like metabolic state i.e., aerobic glycolysis is a shared feature of muscle hypertrophy in response to mechanical loading. 4. Discussion Healthy metabolism is characterized by the cellular flexibility to freely switch between oxidative substrates to maintain energy status and glucose homeostasis. Metabolic inflexibility refers to the inability to appropriately shift substrate use in response to metabolic challenges, which often leads to cellular dysfunction [[256]1]. Metabolic inflexibility is now recognized as a key feature of the pathophysiology of various conditions, including obesity, type 2 diabetes, cardiovascular disease, and aging [[257]2]. Skeletal muscle is the largest organ in the body, comprising about 35–40% of the human body by mass [[258]60]. Skeletal muscle is also a highly metabolically active tissue that is critical to maintaining whole-body metabolic homeostasis [[259]61]. One of the unique features of skeletal muscle tissue is its capacity for remarkable flexibility in the usage of fuel in response to physiological cues [[260]62]. In the current study, we provide compelling evidence that the presence of miR-1 is necessary for maintaining this metabolic flexibility of adult skeletal muscle. The loss of miR-1 in skeletal muscle was sufficient to reduce whole-body metabolic flexibility (defined as the increase in RER from light to dark cycles) by ∼70%. The metabolic inflexibility induced by the loss of skeletal muscle miR-1 caused a significant reduction in endurance exercise capacity in both C. elegans and in mice. Importantly, although reduced wheel running activity in miR-1 KO mice is consistent with impaired skeletal muscle metabolism, we cannot fully exclude the contribution of behavioral changes (e.g., motivation, stress response) to the observed phenotype. Moreover, the impaired training response observed in C. elegans lacking miR-1 raises the possibility that similar adaptive pathways may be disrupted in miR-1 KO mice, which warrants future investigation. miR-1 was one of the first identified miRNAs [[261]63] and among the first miRNAs found to be highly enriched in particular organs (heart and skeletal muscle for miR-1) [[262]64]. Notably, cardiac and skeletal muscle are two organs that display unique metabolic adaptations to support their energy demands [[263]65]. Although miR-1 has been shown to be critical for proper muscle development in both invertebrates and vertebrates using in vitro techniques, technical barriers have prevented further interrogation of how miR-1 controls myogenesis in vivo [[264]65]. Furthermore, the specific function of miR-1 in adult skeletal muscle had not been rigorously investigated since prior mechanistic studies of miR-1 all used germline models or developing mice [[265]8,[266]9,[267]66]. Additionally, another major challenge in determining the function of miR-1 in adult skeletal muscle is the rigorous identification of bona fide target genes. The typical in silico approaches overpredict miRNA binding sites, inaccurately model target site accessibility, and miss biologically significant miRNA binding [[268]27,[269]48,[270]49]. Popular in vitro methods such as luciferase reporter assays also do not account for in vivo stochiometric relationships between miRNAs and accessible mRNA targets in the cell type of interest, and therefore do not accurately reflect true biological interactions in vivo [[271]27,[272]48,[273]49]. To address these prior limitations, we generated an inducible, skeletal muscle-specific miR-1 KO mouse, and utilized AGO2 eCLIP-seq analysis to biochemically define, for the first time, the miR-1:target gene interactions in adult skeletal muscle. By combining the AGO2 eCLIP-seq miR-1 interactome and RNA-seq data, we identified post-transcriptional regulation of metabolic genes involved in pyruvate oxidation. Pyruvate metabolism has been described as the central hub of the entire cellular metabolic network [[274]67]. Positioned at the critical juncture between mitochondrial carbohydrate oxidation and aerobic glycolysis (the ‘Warburg effect’), pyruvate synthesis and consumption are tightly regulated. Alterations in pyruvate metabolism have been shown to have a major influence on the dynamic nature of the cellular metabolic network and, as a result, modulating metabolic flexibility [[275]42,[276]67]. We identified miR-1 target genes that are involved in glycolysis and pyruvate metabolism which include PTBP1, PKM, LDHA, and SLC16A3 (which encodes MCT4). In addition, comprehensive bioenergetic phenotyping demonstrated that miR-1 is necessary for pyruvate oxidation. The altered mitochondrial bioenergetics in miR-1 KO skeletal muscle was also accompanied by a distinct metabolomic profile, characterized by an unbalanced pyruvate–lactate axis. We identified MCT4, the primary cellular lactate exporter, as a key element in this metabolic reprogramming. The inhibition of MCT4 through administration of VB124, a potent and selective inhibitor, redirects glycolytic carbon flux toward oxidation [[277]36,[278]37], which rescued metabolic flexibility and exercise performance in miR-1 KO mice. These data further confirm that miR-1 is necessary for skeletal muscle metabolic flexibility by modulating gene expression to ensure pyruvate oxidation. It is important to note that VB124 treatment was tested only in miR-1 KO mice; thus, we cannot determine whether the observed effects are specific to the miR-1 KO phenotype or reflect a general response to altered pyruvate handling. Additional studies in WT animals will be necessary to clarify this point. [279]Figure 8 summarizes the effects of miR-1 KO on glycolytic enzyme expression and the abundance of glycolytic intermediates. Figure 8. [280]Figure 8 [281]Open in a new tab Summary of the effect of miR-1 loss on glycolytic and pentose phosphate enzymes and intermediates. Glycolytic and pentose phosphate pathway enzymes (ovals) and intermediates (text) depicted. Enzymes up-regulated in miR-1 KO RNA-seq or mitochondrial proteomics indicated by green color, and inactivated enzymes indicated by red color. Up-regulated metabolites in miR-1 KO metabolomics indicated by bolded green text. Green plus sign indicates positive regulation. Target symbol indicates eCLIP-seq-defined miR-1 target. Created using Biorender.com. Although the loss of miR-1 has also been implicated in cardiac hypertrophy [[282]7,[283][68], [284][69], [285][70], [286][71]], we found no difference in the muscle size of miR-1 KO compared to WT mice. Therefore, the growth-promoting metabolic reprogramming induced by the loss of miR-1 was not sufficient to induce hypertrophy. Still, recent studies increasingly suggest that hypertrophying skeletal muscle cells display a Warburg-like rewiring of metabolism [[287][72], [288][73], [289][74], [290][75], [291][76], [292][77]]. Notably, miR-1 expression is dramatically down-regulated in response to a hypertrophic stimulus in mice [[293]58,[294]59,[295]78]. In the current study, we found that miR-1 is decreased during resistance exercise training-induced muscle hypertrophy in humans as well. Thus, while down-regulation of miR-1 may not be sufficient for muscle hypertrophy, a reduction in miR-1 and the ensuing metabolic reprogramming may be necessary for myofiber growth. While it may appear counterintuitive that mitochondrial OXPHOS is downregulated during muscle growth, it is important to note that while mitochondria provide cellular energy by coupling substrate oxidation to ATP synthesis, this neglects other pathways of energy transduction. For example, NADPH, which can be reduced by the pentose phosphate pathway, is also necessary as a reducing equivalent for nucleic acid and fatty acid synthesis [[296]79]. Hypertrophying muscle cells may thus bias against complete pyruvate oxidation and the ensuing loss of carbon as carbon dioxide in favor of maximizing biosynthetic precursors necessary for cell growth, similar to cancer cells [[297]67]. Physiological down-regulation of miR-1 may be a mechanism to temporarily reprogram the metabolism of skeletal muscle cells to aerobic glycolysis during MOV-induced hypertrophy. Few studies have identified factors which, when modulated in skeletal muscle, can influence whole-body metabolic flexibility. For example, skeletal muscle-specific alteration of the key metabolic enzymes and myokines, phosphatidylserine decarboxylase (PSD) [[298]42], PDK [[299]80], and brain-derived neurotrophic factor (BDNF) [[300]81], were shown to affect whole-body metabolic flexibility. More recently, Wang and colleagues demonstrated that muscle-specific KO of lysine (K)-specific demethylase 2A (Kdm2a) could also improve whole-body metabolic flexibility, which was achieved by alternative splicing of the key metabolic nuclear receptor, estrogen-related receptor gamma (ESRRG) [[301]82]. Our study demonstrates that the muscle-specific miRNA, miR-1, regulates whole-body metabolic flexibility to a similar extent as these key metabolic factors. It is important to note that one limitation of our multi-omics approaches was maintaining equal depth of coverage across the transcriptome for each methodology. While the advent of eCLIP-seq has dramatically reduced deduplication rates and input requirements from its predecessor, high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP), eCLIP in tissues remains technically challenging. Further, it is likely a percentage of differentially expressed genes identified by RNA-seq may possess miR-1 target-sites but were below our detection limits for peak calling. Additionally, our AGO2 eCLIP-seq peak-calling method uses neighboring genome read distribution to assess enrichment over background, which may mask weaker miR-1 bound states in favor of other non-miR-1 binding events. Nonetheless, our studies identified, with high confidence, several bona fide miR-1 target genes. We show that miR-1 represents a single miRNA that can regulate the entire pyruvate metabolic pathway. Thus, we have identified a novel target which may lead to effective strategies for managing metabolic disorders in various clinical contexts by modulating several functionally related genes. Another limitation of the current study is the lack of isotope tracing. While our current indirect measurements support impaired glucose oxidation, we acknowledge that they do not directly quantify flux. Direct measurement of glucose fluxes through isotope tracing represents an important future direction to fully elucidate the metabolic rewiring associated with miR-1 loss. Moreover, while our study focused on pyruvate metabolism, future in-depth exploration of fatty acid oxidation and amino acid handling (particularly glutamine) will demonstrate the possible effects of miR-1 on other metabolic pathways in skeletal muscle. Finally, mitochondrial ultrastructure in skeletal muscle fibers was not examined in this study; such analysis could reveal morphological correlates of the observed functional deficits and will be an important focus of future investigations. CRediT authorship contribution statement Ahmed Ismaeel: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation. Bailey D. Peck: Writing – review & editing, Visualization, Methodology, Investigation, Formal analysis, Data curation. McLane M. Montgomery: Investigation, Data curation. Benjamin I. Burke: Writing – review & editing, Investigation, Data curation. Jensen Goh: Investigation, Data curation. Abigail B. Franco: Methodology, Investigation, Data curation. Qin Xia: Investigation, Data curation. Katarzyna Goljanek-Whysall: Writing – review & editing, Supervision, Project administration, Methodology. Brian McDonagh: Writing – review & editing, Project administration, Methodology, Formal analysis. Jared M. McLendon: Investigation, Data curation. Pieter J. Koopmans: Investigation, Data curation. Daniel Jacko: Investigation, Data curation. Kirill Schaaf: Investigation, Data curation. Wilhelm Bloch: Investigation, Data curation. Sebastian Gehlert: Writing – review & editing, Resources, Project administration, Methodology. Kevin A. Murach: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Kelsey H. Fisher–Wellman: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Formal analysis. Ryan L. Boudreau: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis. Yuan Wen: Writing – review & editing, Software, Resources, Project administration, Conceptualization. John J. McCarthy: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Conceptualization. Funding This work was supported by National Institutes of Health grants from the National Institute on Aging (R01AG069909 to J.J.M. and R00AG063994 to K.A.M.) and the National Heart, Lung and Blood Institute (HL144717 and HL150557 to R.L.B.). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements