Skeletal myofiber VEGF deficiency leads to mitochondrial, structural, and contractile alterations in mouse diaphragm

Abstract

Diaphragm dysfunction accompanies cardiopulmonary disease and impaired oxygen delivery. Vascular endothelial growth factor (VEGF) regulates oxygen delivery through angiogenesis, capillary maintenance, and contraction-induced perfusion. We hypothesized that myofiber-specific VEGF deficiency contributes to diaphragm weakness and fatigability. Diaphragm protein expression, capillarity and fiber morphology, mitochondrial respiration and hydrogen peroxide (H₂O₂) generation, and contractile function were compared between adult mice with conditional gene ablation of skeletal myofiber VEGF (SkmVEGF^−/−; n = 12) and littermate controls (n = 13). Diaphragm VEGF protein was ~50% lower in SkmVEGF^−/− than littermate controls (1.45 ± 0.65 vs. 3.04 ± 1.41 pg/total protein; P = 0.001). This was accompanied by an ~15% impairment in maximal isometric specific force (F[1,23] = 15.01, P = 0.001) and a trend for improved fatigue resistance (P = 0.053). Mean fiber cross-sectional area and type I fiber cross-sectional area were lower in SkmVEGF^−/− by ~40% and ~25% (P < 0.05). Capillary-to-fiber ratio was also lower in SkmVEGF^−/− by ~40% (P < 0.05), and thus capillary density was not different. Sarcomeric actin expression was ~30% lower in SkmVEGF^−/− (P < 0.05), whereas myosin heavy chain and MAFbx were similar (measured via immunoblot). Mitochondrial respiration, citrate synthase activity, PGC-1α, and hypoxia-inducible factor 1α were not different in SkmVEGF^−/− (P > 0.05). However, mitochondrial-derived reactive oxygen species (ROS) flux was lower in SkmVEGF^−/− (P = 0.0003). In conclusion, myofiber-specific VEGF gene deletion resulted in a lower capillary-to-fiber ratio, type I fiber atrophy, actin loss, and contractile dysfunction in the diaphragm. In contrast, mitochondrial respiratory function was preserved alongside lower ROS generation, which may play a compensatory role to preserve fatigue resistance in the diaphragm.

NEW & NOTEWORTHY Diaphragm weakness is a hallmark of diseases in which oxygen delivery is compromised. Vascular endothelial growth factor (VEGF) modulates muscle perfusion; however, it remains unclear whether VEGF deficiency contributes to the onset of diaphragm dysfunction. Conditional skeletal myofiber VEGF gene ablation impaired diaphragm contractile function and resulted in type I fiber atrophy, a lower number of capillaries per fiber, and contractile protein content. Mitochondrial function was similar and reactive oxygen species flux was lower. Diaphragm VEGF deficiency may contribute to the onset of respiratory muscle weakness.

INTRODUCTION

Respiratory muscle weakness develops in many clinical conditions, such as acute critical illness and chronic cardiopulmonary disorders, and in aging (23). In particular, impairments to the main muscle of respiration, the diaphragm, contribute substantially to pulmonary complications and poor clinical outcomes in patients (18, 33). However, the mechanisms that induce diaphragm weakness and effective rescue treatments remain poorly resolved. Most clinical disorders associated with diaphragm dysfunction are characterized by abnormal microvasculature and O₂ delivery (49) [e.g., critical illness, chronic heart failure, chronic obstructive pulmonary disease (COPD)]. Abnormalities in the O₂ transport system may be a key mechanism for triggering the onset of diaphragm weakness (23). Vascular endothelial growth factor (VEGF) is a transmembrane glycoprotein that is requisite for blood vessel development and maintenance in all mammalian organs. VEGF is a family of 5 growth factors (VEGF-A, VEGF-B, VEGF-C, VEGF-D, placental growth factor) that have various roles during embryonic/adult tissue development and maintenance. However, VEGF-A (referred to hereafter as VEGF) is the most predominant form in the majority of tissues/organs in adults. In addition to signal transduction for angiogenesis, VEGF is critical for stem cell recruitment, maintenance of vulnerable barriers (i.e., lung, nephron, and kidney cells), and protection of neural tissue (10, 51).

VEGF plays a critical role in locomotor skeletal muscle structure and function, and these variables impact whole-organism functions, such as exercise tolerance (26, 50). Furthermore, skeletal muscle remodeling following exercise training is highly dependent on VEGF being properly expressed in a variety of tissues, including skeletal myofibers (16, 25). VEGF is downregulated in leg muscle samples taken from patients with diseases characterized by abnormalities in O₂ supply, such as COPD (5). In contrast, the role of VEGF expression in the diaphragm remains poorly characterized, and whether low VEGF contributes to diaphragm contractile dysfunction is unknown. Interestingly, in rats undergoing mechanical ventilation in which the diaphragm undergoes arrest to induce severe fiber weakness, VEGF is reduced by around twofold (11, 17). However, in conditions with high mechanical respiratory loading, such as COPD (2), or under hypoxic conditions (46), VEGF mRNA expression in the diaphragm is elevated. Thus, it seems that in conditions characterized by diaphragm disuse, such as critical illness, a reduction in VEGF expression may contribute to respiratory muscle weakness. The mechanisms that underpin diaphragm dysfunction have been suggested to include fiber atrophy (because of an imbalance in protein synthesis/degradation), elevated reactive oxygen species (ROS), an oxidative to glycolytic fiber type transition, and mitochondrial dysfunction (23, 41).

Based on the importance of VEGF in locomotor skeletal muscle during adaptation to exercise training and various pathologies, we aimed to measure whether VEGF deficiency modulates diaphragm contractile function, capillarity, fiber structure, and mitochondrial bioenergetics. Given that diaphragm tissue demonstrates a high degree of cellular abnormalities in a variety of diseases (6, 8, 9, 31, 32), we hypothesized that low myofiber VEGF impairs diaphragm contractile function and is accompanied by fiber remodeling and mitochondrial functional deficits, including capillary regression, fiber atrophy, shifts from oxidative to glycolytic fiber types, and reduced maximal mitochondrial oxygen flux with increased ROS generation.

MATERIALS AND METHODS

Animals and study design.

We measured diaphragm biochemistry, structure, and contractile function in adult mice with conditional deletion of the VEGF-A gene in skeletal myofibers (SkmVEGF^−/−). To achieve this, we maintained and bred mice homozygous for the VEGFLoxP transgene (22) with mice heterogeneous for HSA-Cre-ER^T2 (44). Animals were housed in a pathogen-free vivarium in plastic cages with a 12:12 h light/dark cycle and fed a standard chow diet (Harlan Teklad 8604, Madison, WI) with water ad libitum. Conditional myofiber-specific deletion of VEGF was initiated at 10 wk of age (~20 g body wt) using a tamoxifen-inducible HSA-Cre-ER^T2 system in the VEGFLoxP mice on a C57BL/6J background (16). Male SkmVEGF^−/− mice (n = 12) were compared with floxed wild-type (WT) controls (n = 13) that did not express Cre recombinase (HSA-Cre-ER^T2−/−). All mice received tamoxifen (1 mg/day ip) for 5 consecutive days (D0–D4). Following completion of the tamoxifen treatment period and sufficient time for VEGF expression to be decreased, mice were anesthetized on day 21 (D21) with inhaled isoflurane vaporized in 100% O₂ and killed through removing the major organs. Our protocol was approved by the University of California, San Diego, Animal Care and Use Committee and was conducted in accordance to guidelines outlined by the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Genotyping.

The presence of HSA-Cre-ER^T2 transgene was measured by PCR from tail DNA and forward 5′-CTAGAGCCTGTTTTGCACGTTC-3′ and reverse 5′-TGCAAGTTGAATAACCGGAAA-3′ primers. The conditions during cycling were a 2-min polymerase activation incubation at 95°C, 35 cycles of 30 s denaturation at 94°C, 30 s annealing at 52.1°C, and 60 s elongation at 72°C, followed by one 8-min elongation at 72°C.

Diaphragm dissection and contractile function.

The diaphragm was excised through a laparotomy and thoracotomy. The right costal diaphragm muscle was divided for measurement of mitochondrial function and the remainder immediately snap frozen in liquid N₂ for later biochemical analyses. The left costal diaphragm muscle was prepared in a Krebs–Henseleit buffer solution (120.5 mM NaCl, 4.8 mM KCl, 1.2 mM MgSO₄, 1.2 mM NaH₂PO₄, 20.4 mM NaHCO₃, 1.6 mM CaCl₂, 10 mM dextrose, 1 mM pyruvate at a pH of 7.4), and a muscle bundle (~3 mg) connected from rib to central tendon was dissected, sutured (size 4.0), and mounted horizontally in a buffer-filled organ bath at room temperature equilibrated with 95% O₂-5% CO₂. The suture connected to the rib was secured to a hook in the organ bath, and the tendon was tied to an adjustable-length force transducer (model 920CS, DMT, Aarhus, Denmark). The muscle was stimulated via platinum electrodes with a supramaximal current (500-ms train duration; 0.25-ms monophasic pulses) via a high-power stimulator (model S88, Grass Medical Instruments, Quincy, MA). The muscle bundle was set at an optimal length equivalent to the maximal twitch force produced and a 15-min equilibration period followed. A force-frequency protocol was then performed at 1, 15, 30, 50, 80, 120, 150, and 300 Hz, separated with 1-min rest intervals. After a 5-min period in which muscle length was measured using a digital micrometer, the muscle underwent a fatigue protocol over 5 min (40 Hz every 2 s with a 500-ms train duration). The muscle was subsequently detached, trimmed free from rib and tendon, blotted dry on filter paper, and weighed. Muscle force (N) was normalized to muscle cross-sectional area (cm²) by dividing muscle mass (g) by the product of optimal length (cm) and estimated muscle density (1.06 g/cm³) (13), which allowed specific force (in N/cm²) to be calculated.

Mitochondrial function.

Dissected diaphragm tissue (~5–10 mg) was placed immediately in preservation solution at 4°C. Preservation medium (BIOPS) contained 10 mM Ca²⁺EGTA buffer, 20 mM imidazole, 50 mM K⁺-MES, 0.5 mM dithiothreitol, 6.56 mM MgCl₂, 5.77 mM ATP, 15 mM phosphocreatine, and a pH of 7.1. Fiber bundles (each ~0.5–1 mg) were transferred to a plastic culture dish and kept in BIOPS at 4°C. The bundles were mechanically separated using sharp forceps and a dissection microscope (Leica, Germany). The bundles were permeabilized with a 30-min incubation in saponin (50 µg/mL) dissolved in BIOPS at 4°C. Tissues were washed of saponin for 10 min in a respiration medium at 4°C. Fiber bundles were weighed to the nearest microgram using an ultrabalance (UMX2, Mettler-Toledo, Greifensee, Switzerland) and transferred into a calibrated respirometer (Oxygraph 2k, Oroboros Instruments, Innsbruck, Austria) containing 2 mL of media in each chamber. Respirometry and fluorometry was performed in duplicate at 37°C in stirred media (MiR05+Cr) containing 0.5 mM EGTA, 3 mM MgCl₂, 60 mM K-lactobionate, 20 mM taurine, 10 mM KH₂PO₄, 20 mM HEPES, 110 mM sucrose, 20 mM creatine, and 1 g/L BSA essentially fatty acid free, adjusted to pH 7.1. [O₂] in the media was kept between 300 and 500 µM.

A substrate-uncoupler-inhibitor-titration protocol (37) included 10 mM glutamate and 2 mM malate to support electron entry through complex I (GM; “LEAK” state), 5 mM ADP to stimulate oxidative phosphorylation (“OXPHOS_CI” state), 10 mM succinate to maximize convergent electron flux at the Q-junction (ADP+S; OXPHOS_CI+II), CCCP titrated in 0.5 µM steps to achieve maximal uncoupled respiration for measurement of electron transport system capacity (“ETS” state), 0.5 μM rotenone to inhibit complex I (Rot; ETS_CII). The flux control ratio for OXPHOS was calculated as (OXPHOS_CI/ETS) and (OXPHOS_CI+II/ETS). OXPHOS coupling efficiency was calculated as (1-LEAK/OXPHOS_CI+II). The substrate control ratio for succinate was calculated as (OXPHOS_CI+II/OXPHOS_CI).

In parallel to respirometry, the green fluorescence sensor of the O2k-Fluo LED2-Module (Oroboros) and the Amplex UltraRed (AmR) assay were used to measure hydrogen peroxide (H₂O₂) production during of the respiratory states, an index of total mitochondrial ROS production (27). AmR (10 µM), horseradish peroxidase (HRP; 1 U/mL) and SOD (5 U/mL) were injected before addition substrates. AmR is oxidized by H₂O₂ in the presence of HRP and allows for an excitation/emission at 563 and 587 nm. [H₂O₂] was determined by calibrating to a series of injections of 0.1 µM H₂O₂ using a 40 μM stock solution made fresh daily.

Muscle biochemistry.

VEGF protein in diaphragm tissue was quantified by ELISA and normalized to total protein, in accordance with the manufacturer’s instructions (VEGF Mouse ELISA, R&D Systems, La Jolla, CA). For Western blot analysis, frozen muscle samples were homogenized in relaxing buffer (90 HEPES, 126 KCl, 36 NaCl, 1 MgCl₂, 50 EGTA, 8 ATP, and 10 creatine phosphate in mmol/L at pH 7.4) containing a protease inhibitor mix (Inhibitor Mix M, Serva, Heidelberg, Germany), and sonicated for 10 cycles (Sonopuls GM70, Bandelin Electronics, Berlin, Germany), with protein content of the homogenate subsequently determined (bicinchoninic acid assay, Pierce, Bonn, Germany). Citrate synthase activity (a marker of mitochondrial content) was assessed at room temperature, as previously described (6, 8, 9, 31, 32). Diaphragm homogenates (5–20 µg) mixed with loading buffer (126 mM Tris·HCl, 20% glycerol, 4% SDS, 1.0% 2-mercaptoethanol, 0.005% bromophenol blue; pH 6.8) were separated by SDS-polyacrylamide gel electrophoresis. Proteins were transferred to a PVDF and incubated overnight at 4°C with the following primary antibodies for the contractile proteins of myosin heavy chain (MyHC; 1/1,000, Sigma-Aldrich, Taufkirchen, Germany) and sarcomeric actin (1/500; Sigma-Aldrich): the signaling proteins of PGC1-α (1/200, Santa Cruz, Heidelberg, Germany) and hypoxia inducible factor (HIF)-1α (1/200, Santa Cruz) and the stress-related proteins muscle atrophy F-box (MAFbx; 1/2,000; Eurogentec, Seraing, Belgium) and NADPH oxidase subunit gp91^phox (1/1,000; Abcam, Cambridge, UK). Membranes were subsequently incubated with a HRP-conjugated secondary antibody and specific bands visualized by enzymatic chemiluminescence (SuperSignal West Pico, Thermo Fisher Scientific, Bonn, Germany) and densitometry quantified using a one-dimensional scan software package (Scanalytics, Rockville, MD). Blots were then normalized to the loading control GAPDH (1/30,000; HyTest, Turku, Finland), which we confirmed was not different between experimental groups. All data are presented as fold change relative to control group.

Histology.

Liquid-nitrogen isopentane frozen diaphragm sections prepared for cryosectioning were cut at 10 µm, mounted on glass coverslips, and incubated overnight at 4°C in antibody diluent (Dako, Hamburg, Germany) with primary antibodies against MyHC type I fibers (M8421, 1/400; Sigma) or laminin (1/200; Sigma). After being washed with Tris-buffered saline-Tween 20, sections were incubated with fluorescently conjugated (Alexa 488) secondary antibodies for 1 h, further washed, and then visualized under a fluorescent microscope. Images were captured at ×5 magnification and merged to allow fiber morphology to be determined using imaging software (Analysis Five, Olympus, Münster, Germany), which included ~300–500 fibers in a muscle area of ~500,000 μm² (i.e., ~40%–50% of the total muscle section). Stained fibers (bright green) were taken as type I and unstained fibers (black) as type II, with fiber boundaries demarcated in red. In addition, capillaries were stained with lectin specific to endothelial cells using rhodamine-conjugated Griffonia simplicifolia lectin-1 (1:250 dilution; Vector Laboratories, Peterborough, UK), with the capillary-to-fiber (C/F) ratio (number of capillaries to number of fibers) and capillary density (number of capillaries per mm² of muscle tissue) calculated alongside the mean fiber cross-sectional area (FCSA). Images were captured at ×40 magnification, which included ~100 fibers in a muscle area of ~120,000 µm² (i.e., ~15% of the total muscle section).

Statistical analyses.

Means were compared, when appropriate, with t tests. Diaphragm muscle contractile function data were analyzed using a two-factor repeated-measure ANOVA (contraction time × genotype and contraction frequency × genotype). Mitochondrial function data were analyzed using a two-factor repeated-measure ANOVA (respiratory state × genotype). Data are presented as means ± SD, and, when appropriate, the 95% confidence interval (CI₉₅) is included.

RESULTS

In this conditional VEGF deficiency mouse model, we confirmed that VEGF protein levels were lower in the diaphragm of SkmVEGF^−/− mice compared with WT mice by ~50% (1.45 ± 0.65 vs. 3.04 ± 1.41 pg/total protein; P = 0.001).

Contractile function.

Specific force generated by diaphragm fiber bundles was impaired following VEGF deletion by ~15% in SkmVEGF^−/− compared with WT mice (F[1,23] = 15.01, P = 0.001; main effect of genotype), which occurred at both low and high stimulation frequencies ranging from 15 to 300 Hz (Fig. 1A). Similarly, there was a rightward shift in the normalized force-frequency relationship (F[1,23] = 2.98, P = 0.045; interaction), with relative forces lower between 30 and 120 Hz in SkmVEGF^−/− diaphragm fiber bundles (Fig. 1B). In contrast, fatigue resistance tended to be higher in SkmVEGF^−/− compared with WT mice during the fatigue protocol (F[1,23] = 4.15, P = 0.053; main effect of genotype), such that relative force tended to be ~10% higher throughout the repeated contractions (Fig. 1C). Twitch kinetics remained unaltered following VEGF deletion, with no differences (P > 0.05) between WT and SkmVEGF^−/− mice in terms of time to peak tension (43 ± 7 vs. 42 ± 3 ms, respectively) and half-relaxation time (60 ± 17 vs. 63 ± 16 ms, respectively).

Fig. 1.

Diaphragm contractile function in skeletal muscle-specific vascular endothelial growth factor (VEGF)-deficient (SkmVEGF^−/−) mice compared with wild-type (WT) controls. Isolated diaphragm fiber bundles were stimulated at increasing stimulation frequencies to measure isometric specific force (A) and normalized relative to maximum force (B), with fatigability assessed during repeated stimulations (C). Data are means ± SD; *P < 0.01; §P < 0.05 SkmVEGF^−/− vs. WT.

Muscle structure and protein expression.

The C/F ratio was ~40% lower in SkmVEGF^−/− compared with WT mice (P = 0.008; Fig. 2A). In contrast, the capillary density remained unchanged between groups (P > 0.05; Fig. 2B), which can be explained by the scale-dependent consequence of VEGF deletion lowering mean FCSA in parallel (Fig. 2C; P = 0.023). Representative muscle sections from WT and SkmVEGF^−/− are presented in Fig. 2, D–E. Fiber proportion for type I slow and type II fast MyHC isoforms were not different between SkmVEGF^−/− and WT mice (P = 0.455; Fig. 2F), with type I fibers representing ~10% of the overall fiber population. In contrast, however, type I FCSA was lower by ~25% in SkmVEGF^−/− compared with WT mice (P = 0.027; Fig. 2G). No difference was detected in type II FCSA between the groups (P = 0.786; Fig. 2G). Although no difference was observed between groups for MyHC expression (P = 0.429; Fig. 3A), the smaller type I fiber size corresponded to a ~30% lower level of the key contractile protein actin (P = 0.011; Fig. 3B). We subsequently measured key proteins known to be involved in fiber atrophy signaling but found no differences between groups in relation to the key E3 ligase MAFbx (P = 0.189; Fig. 4A) and major ROS source NADPH oxidase (P = 0.091; Fig. 4B). We also probed for protein expression of key upstream regulatory proteins in the VEGF signaling cascade but found no differences in protein expression of HIF-1α (P = 0.417; Fig. 4C) and PGC-1α (P = 0.176; Fig. 4D).

Fig. 2.

Diaphragm structural properties in skeletal muscle-specific vascular endothelial growth factor (VEGF)-deficient (SkmVEGF^−/−) mice compared with wild-type (WT) controls, which included assessment of capillary-to-fiber (C/F) ratio (A), capillary density (CD) (B), and mean fiber cross-sectional area (FCSA) (C) (n = 5 per group). Representative immunofluorescent diaphragm sections from WT (D) and SkmVEGF^−/− (E) mice (fiber boundaries stained red, type I fibers green, and type II fibers unstained), revealed no change in fiber type proportions (F) but a type I specific fiber atrophy (G) in SkmVEGF^−/− mice compared with WT controls (n = 8–11 per group). Data are presented as means ± SD; *P < 0.05 SkmVEGF^−/− vs. WT.

Fig. 3.

Protein expression in the diaphragm of the key contractile proteins myosin heavy chain (MyHC) (A) and sarcomeric actin (B) as well as respective representative blots (C) in skeletal muscle-specific vascular endothelial growth factor (VEGF)-deficient (SkmVEGF^−/−) mice compared with wild-type (WT) controls. Data were normalized to the loading control GAPDH and calculated as fold change (∆) vs. WT. Data are presented as means ± SD; *P = 0.01 SkmVEGF^−/− vs. WT.

Fig. 4.

Diaphragm protein expression along with representative blots of regulatory muscle signaling proteins measured in skeletal muscle-specific vascular endothelial growth factor (VEGF)-deficient (SkmVEGF^−/−) mice compared with wild-type (WT) controls, which included the key atrogene muscle atrophy F-box (MAFbx) (A) and reactive oxygen species (ROS) source NADPH oxidase (subunit gp91^phox) (B), as well as the upstream VEGF activators hypoxia inducible factor (HIF)-1α (C) and PGC-1α (D). Data were normalized to the loading control GAPDH and calculated as fold change (∆) vs. WT. Data are presented as means ± SD.

Mitochondrial function.

Diaphragm mitochondrial O₂ consumption was not different between SkmVEGF^−/− and WT (F[1,80] = 0.85, P = 0.36; Fig. 5A), nor was an interaction present (respiratory state × genotype: F[4,80] = 0.1, P = 0.98; Fig. 5A). However, H₂O₂ production was lower in SkmVEGF^−/− versus WT (F[1,75] = 14.57, P = 0.0003; Fig. 5B), with particular differences observed at OXPHOS_CI+II and ETS_CII respiratory states. OXPHOS coupling efficiency, the ratio of free to total OXPHOS capacity, was not different between SkmVEGF^−/− and WT (Fig. 5C). There were no differences (P > 0.05) in the flux control ratios for ADP-stimulated respiration (Fig. 5, D–E) or in the substrate control ratio for succinate (Fig. 5F). Citrate synthase activity was also not different (P > 0.05) between WT and SkmVEGF^−/− mice (8.38 ± 0.73 vs. 8.20 ± 0.96 µmol/min/mg protein, respectively).

Fig. 5.

Diaphragm mitochondrial respiratory function and reactive oxygen species (ROS) production. Rate of oxygen consumption (JO₂) measured during a high-resolution respirometry substrate-uncoupler-inhibitor-titration (SUIT) protocol. A: LEAK: glutamate+malate for LEAK state respiration. OXPHOS_CI: ADP for the phosphorylating state with substrates provided to complex I. OXPHOS_CI+II: ADP+succinate. ETS: Carbonyl cyanide m-chlorophenyl hydrazine for uncoupled respiration and electron transport system capacity. ETS_CII: Rotenone added to inhibit complex I. B: H₂O₂ flux measured simultaneous to JO₂. Main effect of genotype present (F[1,75] = 14.57, P = 0.0003); *different to wild-type (WT). C: OXPHOS coupling efficiency was calculated as (1-LEAK/OXPHOS_CI+II). D: Flux control ratio (FCR) for OXPHOS_CI was calculated as (OXPHOS_CI/ETS). E: FCR for OXPHOS_CI+II was calculated as (OXPHOS_CI+II/ETS). F: Substrate control ratio (SCR) for succinate was calculated as (OXPHOS_CI+II/OXPHOS_CI). Error bars are SD; n = 8–10. SkmVEGF, skeletal myofiber vascular endothelial growth factor.

DISCUSSION

We found that myofiber-specific VEGF-deficient mice produce reduced specific tension in the diaphragm accompanied by a lower C/F ratio, fiber atrophy, and loss of sarcomeric actin content. Although the mechanisms linking VEGF signaling to diaphragm weakness remain unclear, we suggest it may be related to the lower number of capillaries supplying each fiber. This could limit oxygen and nutrient availability and affect contractile function via tissue hypoxia, inflammation, and/or protein homeostasis (23). Interestingly, diaphragmatic mitochondrial function and content were maintained in VEGF-deficient mice, and mitochondrial-derived ROS generation was lower. These findings suggest that vascular dysfunction may lead to metabolic compensation in myofibers that manifests as improved fatigue resistance measured in the absence of neural or vascular systems.

Myofiber VEGF deletion leads to diaphragm weakness.

Conditional gene deletion of skeletal myofiber-specific VEGF in adult mice reduced VEGF protein levels by ~50%. Although other cells, including endothelial cells, macrophage stem cells, and fibroblast express VEGF, myofibers are the major source of VEGF in skeletal muscle (50). Rather than lifelong VEGF gene ablation, conditional deletion provides a more clinically relevant model to investigate whether diaphragm VEGF deficiency is linked to respiratory muscle weakness in adult patients. The reduction in diaphragm VEGF levels was similar to that reported previously in lifelong skeletal myofiber-targeted VEGF gene ablated (~60% reduction) 20-wk-old mice (50). Lifelong VEGF deficiency reduces diaphragm capillarity by around 40%, and this reduction in capillaries was almost twofold greater than that observed in locomotor muscles in this same mouse line (50). We observed a similar trend in the present study. Conditional VEGF gene deletion induced in adult mice resulted in a ~40% lower C/F ratio in the diaphragm. However, in contrast to mice with lifelong loss of the VEGF gene, locomotor skeletal muscle capillaries are stable when VEGF gene deletion is initiated in adult mice (16, 26).

Although low VEGF levels are associated with conditions in which diaphragm dysfunction is developed, such as mechanical ventilation (11, 17), a direct link had not been previously explored to our knowledge. Given the diaphragm’s high degree of sensitivity when homeostasis is challenged (6, 8, 9, 31, 32), we anticipated that low VEGF expression would impair diaphragm contractile function. A major finding of our current study, therefore, was confirming that VEGF deficiency impairs isometric specific forces by ~15% in the diaphragm. These functional differences are similar to that observed in patients and experimental animal models in which diaphragm fiber function is impaired, such as mechanical ventilation, critical illness, and heart failure (24, 31, 40). The contractile dysfunction was present at both low (15–30 Hz) and high (120–300 Hz) stimulation frequencies, which indicates that low VEGF-dependent signaling could impact a range of respiratory movements, such as resting and exercising ventilation in addition to acute respiratory exacerbations involving airway clearance.

Our data add important knowledge to indicate that respiratory muscle shows a substantial degree of sensitivity to VEGF ablation. This is not surprising given that the diaphragm shows a high sensitivity to dysfunction compared with the limbs when challenged by disease states, such as heart failure (53), systemic hypertension (8), and pulmonary hypertension (14). Thus, the onset of diaphragm weakness may represent an early marker of disease. The mechanism(s) is/are likely underpinned by the diaphragm’s persistent contraction pattern, which renders it highly susceptible to early changes in mechanical loading (possibly related to pulmonary/acid-base disturbance) and/or delivery of oxygen, nutrients, inflammatory cytokines, and ROS (23). Deficient skeletal myofiber VEGF levels are also linked to whole body reductions in exercise tolerance (26, 50), and our data provide support that respiratory muscle weakness may be a contributing factor.

Mechanisms of VEGF-induced diaphragm weakness.

The mechanism(s) by which VEGF deficiency induced diaphragm contractile dysfunction remain(s) unclear. Our data confirmed impaired force generation at both low and high stimulation frequencies in isolated fiber bundles concomitant with a rightward shift in the normalized force-frequency relationship. Impaired forces at the low stimulation frequencies and the rightward shift in the normalized force-frequency relationship in VEGF-deficient diaphragm fibers could be interpreted as a result of a slow-to-fast fiber type switch and/or alterations in calcium handling (i.e., more rapid Ca²⁺ release/reuptake dynamics). However, we found no changes between groups in terms of fiber type or twitch kinetics, which would argue against such mechanisms acting. However, we did find that protein expression of sarcomeric actin was ~25% lower in the diaphragm of SkmVEGF^−/− mice, which could directly limit cross-bridge cycling, thus limiting force-generating capacity. It remains unclear why we observed a preferential loss of actin compared with myosin. A disproportionate loss between the major contractile proteins is not uncommon, and this has been reported in various conditions, including critical illness myopathy, immobilization, and microgravity (21, 35). Under conditions of VEGF deficiency, therefore, various mechanisms may be induced in the diaphragm, including the preferential degradation of actin (following specific targeting of numerous activated E3 ligases), impaired transcription and synthesis of actin, and/or an imbalance in protein turnover rates between actin and myosin (21, 35).

The lower actin content was also a likely factor underlying atrophy preferentially observed in type I fibers in the diaphragm of SkmVEGF^−/− mice. The reason why we observed a preferential fiber type I atrophy in VEGF-deficient skeletal muscle remains unclear, but it may be consequent to type I fibers having a richer capillary supply than type II fibers. Therefore, type I fibers may be more susceptible to contractile protein loss (and thus atrophy) under conditions associated with VEGF deficiency that severely impair capillary maintenance [e.g., cigarette smoke exposure (34, 52)]. Interestingly, loss of thin filaments may increase the maximal shortening velocity (V_max) even without a change in MyHC (42). Without a measure of V_max, however, we can only speculate at this point.

It is important to note that it remains unclear how much this loss of actin actually contributed to the observed contractile weakness/type I fiber atrophy given the small proportion of type I fibers present in mouse diaphragm. Nonetheless, in other diseases in which O₂ delivery is impaired, such as heart failure and COPD, reduced actin content in combination with muscle dysfunction and atrophy are present (7, 30, 47). In muscle wasting following dexamethasone treatment or cigarette exposure in mice, VEGF protein expression is also reduced (4). In combination, these findings suggest VEGF may play a critical role in modulating protein homeostasis. However, we wish to note a few limitations in our methodology, which include not distinguishing between the type II fiber subtypes (a, x, b) and use of a low salt buffer to assess levels of MyHC and actin in which total abundance may not be accurately reflected if incompletely solubilized (43).

The signaling pathways responsible for protein loss and fiber atrophy in VEGF deficiency remain unclear but likely include an upregulation in protein degradation and/or a downregulation in protein synthesis in muscle with an insufficient supply of O₂. Given that conditional VEGF ablation led to around a ~40% lower C/F ratio and, as previous data have shown, it can also impair muscle perfusion (26, 50), it is possible that VEGF deficiency in muscle induces tissue hypoxia to activate proteolytic pathways (15). It is important to note that this is speculative, as we do not have a measure of diaphragm tissue hypoxia in vivo from the current study. Nonetheless, the fact that protein levels of the key E3 ligase atrogene MAFbx were not different between WT and SkmVEGF^−/− mice suggests that diaphragm VEGF deficiency may upregulate other key atrophic pathways associated with contractile protein degradation [e.g., calpain, caspase-3, MuRF-1, or increased ubiquitin proteasome activity (7, 30, 48)] or via inhibition of protein synthesis [e.g., Akt, mTOR, p70S6K (45)]. Future investigation into the role VEGF exerts over various protein synthesis/degradation signaling molecules may therefore prove worthwhile.

Elevated ROS in the diaphragm in disease can also stimulate protein degradation (45) and induce post-translational oxidative contractile protein modifications (9). However, as our ROS measures of mitochondrial as well as transmembrane NADPH oxidase were not elevated in SkmVEGF^−/−, with both known to be strong mediators of diaphragm weakness (1), this argues against VEGF deficiency inducing diaphragm weakness via a ROS-related mechanism. However, an activity-based assay of NADPH oxidase should be considered for future experiments. We also probed upstream regulators involved in the VEGF signaling cascade for potential compensatory adaptations to the VEGF deficiency. Both HIF-1α and PGC-1α independently stimulate VEGF (3), with VEGF generally secreted by the transcription factor HIF-1α in response to hypoxic conditions (19) and by the transcriptional coactivator PGC-1α in response to exercise (29). Both HIF-1α and PGC-1α signaling in diaphragm tissue were not impacted in the present study, indicating they do not form part of a compensatory response to VEGF deficiency.

VEGF deletion does not impact diaphragm mitochondrial respiration.

Mitochondrial impairments are developed in diseases associated with diaphragm dysfunction (38, 40). However, in our study, in situ mitochondrial O₂ consumption was not impacted. Neither mitochondrial content (as assessed by citrate synthase) nor the coupling efficiency or respiratory capacity were different under VEGF deficiency. Whether the fiber atrophy that occurred following VEGF deficiency acted as a compensatory mechanism to maintain mitochondrial respiratory function remains unclear but seems a reasonable suggestion. For example, in response to derangements in muscle perfusion, a lower muscle size may help maintain adequate O₂ diffusion between capillary and myocyte (26).

In contrast, mitochondrial-derived ROS flux was lower following VEGF ablation. The mechanisms underlying lower mitochondrial ROS flux in SkmVEGF^−/− mice remain uncertain, but it could be related to a higher antioxidant enzyme capacity. For example, a compensatory increase in oxidative enzyme activity (i.e., citrate synthase, β-hydroxyacyl-CoA dehydrogenase) occurs in both locomotor and diaphragm muscle in mice with VEGF deficiency from birth (36) and in the gastrocnemius following conditional deletion (16). As a shift toward greater muscle oxidative capacity is commonly associated with increased antioxidant enzyme activities (e.g., superoxide dismutase, glutathione peroxidase, catalase) (39), this could be a plausible explanation for why mitochondrial ROS flux may be lower. However, our current data show citrate synthase activity remained unchanged.

High mitochondrial ROS production is particularly common in conditions associated with diaphragm weakness, such as heart failure (28) and mechanical ventilation (40). Surprisingly, we found that mitochondrial-derived ROS flux was lower in VEGF-deficient diaphragm fibers. As NADPH oxidase, a cytosolic ROS marker, was also not different between groups, our data indicate that diaphragm weakness induced by VEGF deficiency is unlikely to be mediated by elevated oxidative stress. SkmVEGF^−/− mice may have been better protected against ROS-induced muscle fatigue, as increased ROS production during repeated contractions inhibits force generation (12). Interestingly, improvements in relative fatigue of the diaphragm in vitro are also present in disease (31), which suggests the SkmVEGF^−/− mouse model may closely reflect pathological states and thus mirror adaptations to that often observed after long-term endurance training (39). However, the observation of this improved fatigue resistance may be dependent on the in vitro muscle fatigue protocol employed. Using a “matched stimulus” rather than “matched initial specific force” protocol, healthy (or control) diaphragm fibers generate higher absolute forces initially and throughout to induce an apparent more rapid fatigue (20, 31). Therefore, the fatigue protocol employed in our study may explain why relative fatigue resistance was higher in mice with VEGF deficiency and extrapolation of these findings should be interpreted with caution.

CONCLUSIONS

Skeletal myofiber-specific VEGF deletion resulted in diaphragm contractile dysfunction that was accompanied by lower C/F ratio, smaller type I fibers, lower sarcomeric actin protein, and lower mitochondrial ROS generation. Whether differences in type I fiber size, capillarity, and ROS generation compensated to protect mitochondrial respiratory function remains unclear. Deficient diaphragm VEGF levels may be a contributing factor to the onset of diaphragm dysfunction and provide a viable treatment target for patients afflicted with respiratory muscle weakness.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grant P01-HL-091830-01A1 and Tobacco-Related Disease Research Program 26IP-0033. T. S. Bowen acknowledges support from the Fondation Leducq (TNE 13CVD04) and the Medical Research Council UK (MR/S025472/1).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.T.C., E.C.B., and T.S.B. conceived and designed research; D.T.C., L.R., V.A., E.C.B., and T.S.B. performed experiments; D.T.C., L.R., V.A., and T.S.B. analyzed data; D.T.C., L.R., V.A., E.C.B., and T.S.B. interpreted results of experiments; D.T.C. and T.S.B. prepared figures; D.T.C. and T.S.B. drafted manuscript; D.T.C., L.R., V.A., E.C.B., and T.S.B. edited and revised manuscript; D.T.C., L.R., V.A., E.C.B., and T.S.B. approved final version of manuscript.