Myofibre injury induces capillary disruption and regeneration of disorganized microvascular networks

Abstract

Injury to skeletal muscle disrupts myofibres and their microvascular supply. While the regeneration of myofibres is well described, little is known of how the microcirculation is affected by skeletal muscle injury or its recovery during regeneration. Nevertheless, the microvasculature must also recover to restore skeletal muscle function. We aimed to define the nature of microvascular damage and time course of repair during muscle injury and regeneration induced by the myotoxin BaCl₂. To test the hypothesis that microvascular disruption occurred secondary to myofibre injury, isolated microvessels were exposed to BaCl₂ or the myotoxin was injected into the gluteus maximus (GM) muscle of mice. In isolated microvessels, BaCl₂ depolarized smooth muscle cells (SMCs) and endothelial cells while increasing intracellular calcium in SMCs but did not elicit death of either cell type. At 1 day post-injury (dpi) of the GM, capillary fragmentation coincided with myofibre degeneration while arteriolar and venular networks remained intact; neutrophil depletion before injury did not prevent capillary damage. Perfused capillary networks reformed by 5 dpi in association with more terminal arterioles and were dilated through 10 dpi. With no change in microvascular area or branch point number in regenerating capillary networks, fewer capillaries aligned with myofibres and were no longer organized into microvascular units. By 21 dpi, capillary orientation and microvascular unit organization were no longer different from uninjured GM. We conclude that following their disruption secondary to myofibre damage, capillaries regenerate as disorganized networks that remodel into microvascular units as regenerated myofibres mature.

Graphical Abstract

Model of microvascular injury and regeneration after acute trauma to skeletal muscle.

Introduction

Skeletal muscle regeneration is an intricate process that requires the activation, proliferation and differentiation of resident stem cells called satellite cells (Yin et al. 2013). However, the restoration of intact, functional muscle requires the coordinated recovery of additional cell types and tissue components during myogenesis, particularly the microcirculation. When compared to the well-defined molecular and cellular events of myofibre degeneration and regeneration (for review, see Karalaki et al. 2009), little is known of how skeletal muscle injury and regeneration affect its microvascular supply.

The microvasculature of skeletal muscle consists of arterioles, capillaries and venules comprising networks of branches arranged in series and in parallel (Segal, 2005). The microcirculation delivers oxygen and nutrients to myofibres while removing cellular debris and products of metabolism. To meet these physiological demands, the microcirculation responds acutely by regulating local blood flow (e.g. functional hyperemia in response to muscle contraction; Segal & Duling, 1986) and adapts to chronic use by modifying network morphology (e.g. increased capillarization and arteriogenesis; Olfert et al. 2001; Peirce & Skalak, 2003).

Acute muscle injury can result from lacerations, contusions, eccentric contraction or exposure to myotoxins. Regardless of the mechanism of injury, myofibres follow a similar pattern of degeneration and regeneration (Hardy et al. 2016). To study muscle injury and regeneration in mice, a standard approach is intramuscular injection of the myotoxic agent BaCl₂, which induces reproducible myofibre damage while sparing satellite cells to support myofibre regeneration (Caldwell et al. 1990; Casar et al. 2004; Cornelison et al. 2004). As shown in the mouse gluteus maximus (GM) muscle, BaCl₂ also damages capillaries (Fernando et al. 2019; Morton et al. 2019), which undergo fragmentation within 1 day post-injury (dpi), thereby eliminating local perfusion and solute transport. BaCl₂-induced myofibre death occurs through depolarization of the sarcolemma leading to Ca²⁺ overload, membrane disruption and proteolysis (Morton et al. 2019). However, it is unknown how BaCl₂ results in capillary fragmentation in vivo. While freeze injury or injection of snake venom toxins also disrupts capillaries along with myofibre damage (Hardy et al. 2016), it is unknown whether capillary fragmentation (i.e. damage to endothelial cells (ECs)) is a direct effect of the initial insult by BaCl₂ or is a consequence of myofibre degeneration. Nor has it been determined whether arteriolar and venular networks, which supply and drain capillary networks, are disrupted in the manner shown for capillaries.

Ischaemic injury to skeletal muscle is followed by robust angiogenesis with an increase in capillary-to-myofibre ratio (Hardy et al. 2016; Arpino et al. 2017). Following injury with BaCl₂, endothelial sprouts appear within 2–3 dpi, with the ensuing regeneration of capillary networks restoring local perfusion by 5 dpi (Fernando et al. 2019; Morton et al. 2019). At this time, arteriolar networks are abnormally dilated with recovery of blood flow control (vasomotor tone, dilatation and constriction) occurring by 21 dpi in the mouse GM when regenerating myofibres mature (Fernando et al. 2019). Nevertheless, the cellular dynamics of revascularization and microvascular remodelling during myofibre regeneration are poorly understood. In the present study, we tested the hypotheses that (1) BaCl₂ induces death of microvascular ECs and smooth muscle cells (SMCs) by triggering Ca²⁺ overload; and (2) capillaries proliferate during early regeneration with networks remodelling as nascent myofibres mature.

Methods

Ethical approval

All procedures were approved by the Animal Care and Use Committee at the University of Missouri (protocol no. 17720) and were performed in accordance with the National Research Council’s Guide for the Care and Use of Laboratory Animals and the animal ethics checklist of The Journal of Physiology.

Animal care and use

Male C57Bl/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA; RRID: IMSR_JAX:000664) at ~14 weeks of age, acclimated at the University of Missouri animal care facilities at least 1 week prior to study, and used in all experiments other than our analyses of capillary network morphology. For the latter experiments, male Cdh5-mTmG mice (cross of VE-cadherin-CreERT2 mice (RRID: MGI:3848984, gifted from Dr Luisa Iruela-Arispe; Wang et al. 2010) and Rosa26-mTmG mice (RRID: IMSR_JAX:007676; The Jackson Laboratory; both on C57BL/6 background), were bred and housed in animal care facilities of the University of Missouri. Mice (weight, ~30 g) were studied at ~4 months of age. Cre recombination for membrane-bound eGFP expression in ECs was induced through intraperitoneal injection of 100 μl tamoxifen solution (10 mg/ml + 5% ethanol in peanut oil; no. T5648, Sigma-Aldrich, St Louis, MO, USA) on three consecutive days with at least 1 week allowed after the first injection prior to study. All mice were maintained under a 12:12 h light–dark cycle at 22–24°C with fresh food and water ad libitum. To control for an order effect, criterion time points and treatment status were randomized. Prior to tissue collection, muscle injury, and/or intravital microscopy, mice were anaesthetized(ketamine (100 mg/kg) + xylazine (10 mg/kg) in sterile saline; intraperitoneal injection) and euthanized at the end of an experiment by anaesthetic overdose and cervical dislocation.

Preparation of isolated microvessels and endothelial tubes

On the morning of an experiment, a male C57Bl/6J mouse was anaesthetized with ketamine + xylazine, abdominal fur was shaved and a midline incision through the skin was made from the sternum to the pubis. The abdominal muscles were exposed, removed bilaterally and placed in a dissection chamber containing chilled, nominally Ca²⁺ free physiological salt solution (PSS, pH 7.4) containing (in mM): 140 NaCl (Fisher Scientific, Pittsburgh, PA, USA), 5 KCl (Fisher), 1 MgCl₂ (Sigma), 10 HEPES (Sigma) and 10 glucose (Fisher); standard PSS (sPSS) also contained 2 mM CaCl₂ (Fisher). Muscles were pinned as a flat sheet onto transparent silicone rubber (Sylgard 184; Dow Corning, Midland, MI, USA). While viewing through a stereomicroscope, an unbranched segment of the superior epigastric artery (SEA; length, ~2 mm; diameter, ~100 μm; comprising a single SMC layer surrounding the EC monolayer) was dissected from the surrounding tissue. Following isolation, a SEA was transferred to a tissue chamber (no. RC27-N; Warner Instruments, Hamden, CT, USA) for cannulation. The tissue chamber was secured in a platform with micromanipulators (MT-XYZ; Siskiyou Corp., Grants Pass, OR, USA) positioned at each end that held heat-polished cannulation micropipettes (external diameter, ~100 μm). The SEA was cannulated at each end and secured with silk suture. The vessel preparation was transferred to the stage of a Nikon E600FN microscope (Tokyo, Japan) mounted on a vibration isolation table (TMC Vibration Control, Peabody, MA, USA). The vessel was pressurized to 100 cmH₂O (~75 mmHg), maintained at 37°C, superfused at 3 ml/min with sPSS, and allowed to equilibrate for 15 min before experimentation (Norton et al. 2019).

To isolate intact microvascular endothelial tubes, a SEA segment (length, ~1 mm; diameter, ~60 μm) was placed into a round bottom test tube containing 0.62 mg/ml papain (no. P4762; Sigma), 1 mg/ml dithioerythritol (no. D8255; Sigma), and 1.5 mg/ml collagenase (no. C8051; Sigma) in sPSS and incubated for 30 min at 34°C (Behringer et al. 2012). The microvessel segment was transferred to the tissue chamber and gently triturated to remove the SMCs by aspirating and ejecting the segment through a borosilicate glass capillary tube that was heat polished at one end (tip internal diameter, ~80 μm). Following dissociation of SMCs (confirmed by visual inspection at ×200 magnification), the endothelial tube was secured against the bottom of the chamber with blunt fire-polished micropipettes held in the micromanipulators. The preparation was then moved to an inverted microscope (no. TS100; Nikon) mounted on a vibration isolation table (TMC Vibration Control). The endothelial tube was maintained at 33°C, superfused at 3 ml/min with sPSS and equilibrated for 15 min before experimentation.

Intracellular recording

Membrane potential (V_m) of SMCs (intact pressurized SEA) or ECs (endothelial tube) was recorded with an Axoclamp amplifier (2B; Molecular Devices, San Jose, CA, USA) using microelectrodes pulled (P-97; Sutter) from glass capillary tubes (no. GC100F-10; Warner) and backfilled with 2 M KCl (tip resistance, ~150 MΩ). A Ag/AgCl pellet was placed in effluent sPSS for the reference electrode. The output of the amplifier was connected to a data acquisition system (Digidata 1322A, RRID:SCR_021041; Molecular Devices) and an audible baseline monitor (ABM-3; World Precision Instruments, Sarasota, FL, USA). Data were recorded at 1000 Hz using pCLAMP 10.0 software (RRID: SCR_011323; Molecular Devices) on a personal computer. Successful impalements were indicated by sharp negative deflection of V_m, stable V_m >1 min and prompt return to 0 mV upon withdrawal of the microelectrode. Once a cell was impaled, V_m was recorded for at least 5 min to establish a stable baseline. The superfusion solution was then changed to sPSS containing 1.2% BaCl₂ until the V_m response had stabilized (~10 min). Each experiment represents paired data under resting baseline conditions and when stabilized during BaCl₂ exposure for an intact vessel or endothelial tube each obtained from a separate mouse.

Calcium photometry

A cannulated, pressurized SEA was secured in a tissue chamber and placed on an inverted microscope. The vessel was superfused (3 ml/min) at 37°C for 20 min with sPSS, then incubated in a static bath containing Fura-2-AM dye (no. F14185; Fisher). The dye was dissolved in DMSO, diluted to 1 μM in sPSS, and added to the tissue chamber for 40 min. Superfusion with sPSS was then resumed for 20 min to wash out excess dye. Fura-2 fluorescence was used to evaluate [Ca²⁺]_i by alternately exciting the preparation at 340 and 380 nm while recording emissions at 510 nm through a ×20 objective (Nikon Fluor20, numerical aperture (NA) = 0.45) using IonWizard 6.3 software (RRID: SCR_021764; IonOptix, Milford, MA, USA). After Fura-2 fluorescence and vessel diameter were recorded under baseline conditions, the SEA was exposed to 1.2% BaCl₂ in sPSS. Intracellular Ca²⁺ signals and vessel diameter were measured over 30 min of BaCl₂ exposure. Under these conditions (dye loaded from the bath), the [Ca²⁺]_i signal primarily originates from SMCs (Norton & Segal, 2018).

To measure [Ca²⁺]_i in ECs, an endothelial tube was incubated for 30 min with Fura-2 AM dye in a static bath and then washed for 20 min with sPSS. To maintain their integrity, these preparations were studied at 33°C (Socha et al. 2011; Norton et al. 2020). After baseline fluorescence was recorded, intracellular Ca²⁺ signals (F₃₄₀/F₃₈₀) were measured during the 30 min exposure to 1.2% BaCl₂ in sPSS; responses typically stabilized within ~10 min.

Each experiment represents paired data under resting baseline conditions and during the stable response to BaCl₂ for an intact vessel or endothelial tube, each obtained from a separate mouse.

Cell death

Following equilibration in sPSS, the superfusion solution was changed to sPSS containing 1.2% BaCl₂. After preparations were exposed to BaCl₂ for 1 h (which kills >90% myofibres; Morton et al. 2019), superfusion with sPSS was restored. The membrane-permeant nuclear dye Hoechst 33342 (1 μM; no. H1399; Fisher) was used to identify nuclei of all cells and propidium iodide (2 μM, no. P4170; Sigma) to identify nuclei in dead and dying cells (Norton et al. 2019; Norton et al. 2020). Following BaCl₂ exposure, respective nuclear dyes (in sPSS) were perfused through the lumen of a cannulated SEA (0.1 ml/min) or superfused over the surface of an endothelial tube for 10 min followed by 10 min wash in sPSS.

To evaluate cell death, fluorescence images of nuclear staining were acquired with appropriate filters using a ×40 water immersion objective (NA = 0.8) coupled to a DS-Qi2 camera with NIS-Elements Basic Research software (version 4.51, RRID: SCR_002776; Nikon) on an E800 microscope (Nikon). Z-stacks were acquired from the top half of the vessel segment and analysed using ImageJ software (RRID: SCR_003070; NIH, Bethesda, MD, USA). Stained nuclei were counted manually within a defined region of interest (150 μm × 500 μm in intact microvessels, 50 μm × 300 μm in endothelial tubes); nuclei of ECs are oval shaped and oriented parallel to the vessel axis while SMC nuclei are thin and oriented perpendicular to the vessel axis (Norton et al. 2019). For each cell type, cell death is expressed as a percentage as follows: (number propidium iodide⁺ nuclei/number Hoechst 33342⁺ nuclei) × 100.

Muscle injury by local injection of BaCl₂

A mouse was anaesthetized with ketamine + xylazine and placed on an aluminium warming plate to maintain body temperature. Depth of anaesthesia was assessed by lack of tail or toe pinch reflex. The skin over the injury site was shaved and sterilized by wiping 3× with Betadine Solution (Purdue Products LP, Stamford, CT, USA) followed by 70% alcohol. An incision (~5 mm) was made through the skin to expose the GM near the lumbar fascia. Using a Hamilton (Reno, NV, USA) syringe and 32 gauge needle, 75 μl of 1.2% BaCl₂ in water was injected under the GM to injure the muscle (Fernando et al. 2019). The incision was closed with surgical glue or two discontinuous sutures. Surgery required ~15 min. The mouse was placed on a heated platform, monitored until consciousness and ambulation were restored, then returned to its original cage. Mice routinely recovered normal activity and behaviour within 24 h and were studied up to 21 dpi with uninjured mice (0 dpi) serving as controls. The 21 dpi time point coincides with recovery of vasomotor control in arteriolar networks and restoration of myofibre number during regeneration of the mouse GM following BaCl₂ injury (Fernando et al. 2019).

Neutrophil depletion

In some experiments, neutrophils were depleted prior to BaCl₂ injection using a neutralizing Ly6G antibody. Mice were injected intraperitoneally with either 500 μg of anti-Ly6G 1A8 antibody (RRID: AB_10312146, no. BE0075; BioxCell; Lebanon, NH, USA; Bamboat et al. 2010) or vehicle (sterile saline) on −2, −1 and 0 dpi. Neutrophil depletion was confirmed by a differential blood count of a sample collected by cardiac puncture in mice anaesthetized with ketamine + xylazine. A drop of whole blood was spread on a glass slide and stained with a Wright Giemsa stain. Leukocytes were counted in sets of 100 cells, differentiating between lymphocytes, neutrophils and monocytes. Two sets of 100 counts were averaged per slide. Blood samples were obtained from mice without muscle injury (0 dpi) and at 1 dpi for comparison to mice injected with the vehicle at the same time points.

Dissection of gluteus maximus muscle

A mouse was anaesthetized with ketamine + xylazine. Supplemental doses (~20% of initial dose) were given throughout the experimental protocol (typically lasting 2–3 h) to maintain a stable plane of anaesthesia (checked every 15 min by lack of withdrawal to tail or toe pinch). Hair was removed by shaving and the mouse was placed on a warming plate. Skin and overlying connective tissue were removed using microdissection to expose the GM as previously described (Bearden et al. 2004). Once exposed, the GM was continuously superfused with a bicarbonate-buffered physiological salt solution (bbPSS; pH 7.4, 34–35°C) containing (in mM) 131.9 NaCl₂ (Fisher), 4.7 KCl (Fisher), 2 CaCl₂ (Sigma), 1.17 MgSO₄ (Sigma) and 18 NaHCO₃ (Sigma) equilibrated with 5% CO₂–95% N₂. While viewing through a stereomicroscope, the GM was cut along its origin from the lumbar fascia, sacrum and iliac crest to reflect the muscle away from the body and reveal its vascular supply on the ventral side. The GM was then either removed for immunostaining (and the mouse euthanized by cervical dislocation) or prepared for intravital microscopy as described below.

Whole mount immunostaining for confocal imaging

The GM from male C57BL/6J mice was excised, placed in ice-cold phosphate-buffered saline (PBS; pH 7.4; no. P3813; Sigma), and transferred to a 12-well plate coated with transparent rubber (Sylgard 184; Dow Corning). The GM was spread to approximate its dimensions in situ with the ventral surface facing up and secured at the edges with insect pins. Excessive connective tissue and fat were removed using microdissection and then the GM was fixed overnight at 4°C in 4% paraformaldehyde in PBS. After washing in PBS (3 × 5 min), the muscle was immersed in blocking buffer containing 0.5% Triton X-100 (no. T8787; Sigma), 2% bovine serum albumin (no. BP671; Fisher), and 4% normal goat serum (no. 50197Z; Sigma) in PBS. To identify SMCs, the GM was stained with monoclonal rabbit anti-Myh11 (1:500; RRID:AB_10975311, no. ab124679; Abcam, Cambridge, UK; Duivenvoorden et al. 2018) in blocking buffer overnight at 4°C, then washed with blocking buffer (3 × 5 min), incubated with goat anti-rabbit Alexa 633 (1:200; RRID: AB_2535732, no. A21071; Fisher) in blocking buffer for 2 h at room temperature and washed in PBS (3 × 10 min).

An immunostained GM was placed in a custom imaging chamber with the ventral surface facing the objective to optimize resolution of the microvasculature. A small volume of PBS (<10 μl) was added to the chamber and the GM was flattened by placing a glass block (2 cm × 2.5 cm ×1 cm; mass, 7.8 g) on the dorsal surface. Images of the microvasculature were acquired with a ×10 objective (NA = 0.4; image size, 1162.5 μm × 1162.5 μm) or ×25 water immersion objective (NA = 0.95; image size, 465 μm × 465 μm) on an inverted laser scanning confocal microscope (TCS SP8, Leica Microsystems, Buffalo Grove, IL, USA) using Leica LASX software (RRID: SCR_013673). Maximum projection z-stacks (~150 μm thick; 2 μm z-steps) were used to resolve SMC coverage and capillary network morphology (n = 5–12 images from 4 mice).

Analysis of capillary diameter and network morphology

Male Cdh5-mTmG mice expressing green fluorescent protein (eGFP) in ECs were used to analyse capillary diameters and network morphology at defined time points during regeneration. The GM was excised, cleaned, secured and imaged as described above, but without prior fixation; one to three images were analysed per muscle across four to seven mice. To measure the average diameter of capillaries from confocal z-stacks (image size, 465 μm × 465 μm), a calibrated 5 × 5 grid was centred over the image, creating 25 equal ROIs (93 μm × 93 μm). The diameter of five capillaries was measured manually from 10 ROIs selected at random; these 50 measurements were averaged for the image. To analyse microvascular networks using ImageJ, a threshold was applied to maximum projections of z-stacks to generate a binary image. The percentage vascular area (area fraction) was calculated by determining the area occupied by eGFP fluorescence within the entire image (465 μm × 465 μm).

To assess capillary network morphology from a larger field of view (1162.5 μm × 1162.5 μm), the binary image was skeletonized and analysed with the Analyse Skeleton plugin (ImageJ). Outputs included number of segments, length of segments, number of connected segments and number of branch points. Segments included in this analysis contained at least three continuous pixels, which corresponded to ~2.5 μm in length.

To calculate the connected capillary length as a measure of fragmentation at 1 dpi, the length of connected segments was summed for each network (defined as a group of continuous capillaries) and the maximum connected capillary length determined for a given image (465 μm × 465 μm). Branch point frequency was calculated as the number per total microvascular length (mm) for that image. For analysis of capillary orientation, images were digitally rotated during acquisition to align myofibres vertically. The OrientationJ plugin generated distribution histograms for capillary orientation with −90° and +90° defining the horizontal axis. All measurements in ImageJ were validated using reference networks created with known area, length, number of branch points and orientation.

Intravital microscopy

The GM of anaesthetized male C57BL/6J mice was dissected as described above, then spread onto a transparent rubber pedestal (Sylgard 184; Dow Corning) and pinned at its edges to approximate in situ dimensions (Bearden et al. 2004; Fernando et al. 2019). The preparation was transferred to the stage of a Nikon E600FN microscope and equilibrated for 30 min while continuously superfused with bbPSS at 3 ml/min. To assess vascular perfusion and permeability during maximum capillary damage (at 1 dpi), a fluorescein isothiocyanate (FITC) conjugated dextran (70 kDa, to approximate the mass of albumin) was injected into the retroorbital sinus to access the systemic circulation and allowed to circulate for ~10 min (Fernando et al. 2019). The preparation was illuminated by a mercury lamp for fluorescence imaging using an appropriate filter cube. Images were acquired through a ×4 objective (Nikon Fluor4, NA = 0.1; image size, 2.7 mm × 3.4 mm) coupled to a low light CMOS FP-Lucy camera (Stanford Photonics, Inc. (SPI)), Palo Alto, CA, USA) and displayed on a digital monitor. Time lapse images were recorded at 40 frames/s using Piper Control software (SPI). At the end of the experiment, the mouse was given an overdose of anaesthetic and killed by cervical dislocation.

Mapping arteriolar networks

In anaesthetized male C57Bl/6J mice, wheat germ agglutinin conjugated to Alexa Fluor 647 (WGA-647, 1 mg/ml, 200 μl; no. W32466; Fisher) was injected into the retroorbital sinus to access the systemic circulation and label the endothelial glycocalyx. The WGA-647 distributed throughout the vascular compartment for 10 min. Thereafter, the GM was removed, cleaned and secured with pins as above. The muscle was fixed overnight at 4°C (4% paraformaldehyde in PBS) avoiding direct light. Tissues were washed in PBS, cleared in 100% glycerol overnight at 4°C, then rinsed in PBS, mounted onto a slide and coverslipped.

Whole mount GM preparations were viewed using a Nikon E600 microscope with a Nikon Plan Fluor ×10 objective (NA 0.3) coupled to a CMOS camera (Orca Flash 4.0; Hamamatsu Photonics K.K., Hamamatsu City, Japan) and personal computer. The XYZ translational stage (Ludl Electronic, Hawthorne, NY, USA) was controlled by stepper motors coupled to an integrated joystick for constant repositioning of the tissue during imaging. The tracing icon was adjusted by the operator to match vessel diameter; spatial resolution was <2 μm. This strategy is similar to that first developed for analysis of arteriolar networks in aged mice (Bearden et al. 2004). A grid slide (MBF Bioscience, Williston, VT, USA) was used for parcentric and parfocal calibration when linking the hardware with the software. A stage micrometre (100 × 0.01 = 1 mm; Graticules Ltd, Tonbridge, UK) validated the spatial calibration.

3D reconstruction and analysis of resistance networks

Vesselucida Microscope Edition software (RRID: SCR_017320; MBF Bioscience) was used to trace entire resistance networks in the GM, from the inferior gluteal (feed) artery to terminal arterioles. Some tracings contained collateral branches to the resistance network fed by the superior gluteal artery. Occasionally, in thicker regions of the GM, reduced visibility resulted in discontinuous or incomplete traced networks, which were not used in the present analyses. Furthermore, networks containing fewer than 100 vessel segments in the completed tracing were excluded from analyses. Vesselucida Explorer software (RRID: SCR_017674; MBF Bioscience) was used to calculate number of segments, their diameter and length along with the number of anastomoses in each tracing. To generate summary histograms across GM preparations during regeneration, segments were binned into diameter increments of 1 μm; for cumulative length, the length of those within each bin were summed. Anastomoses were defined as the shortest length to return to a given branch point.

Statistics

Results are displayed as means (standard deviation, SD). A power analysis using parameters α = 0.05, power (1 − β) = 80% indicated a minimum group size of n = 4 for isolated vessels and n = 5 for experiments completed in the GM as the appropriate number of replicates for criterion measurements. Statistical analyses were performed using Prism 9 software (RRID: SCR_002798; GraphPad Software Inc., La Jolla, CA, USA). For ex vivo experiments on isolated vessel preparations, V_m, [Ca²⁺]_i and diameter were analysed with a paired two-tailed Student’s t-test; vascular cell death experiments were analysed using a two-way ANOVA with Šídák’s multiple comparisons test. For capillary network morphology, variables evaluated during regeneration were analysed by one-way ANOVA with Tukey’s multiple comparisons post hoc tests. Frequency distributions for dimensions of resistance networks were analysed with the Kolmogorov–Smirnov test. P < 0.05 was considered statistically significant. Values of n refer to the number of vessels or images evaluated across four to seven mice.

Results

Previous studies evaluating microvascular injury did not resolve whether damage and loss of perfusion were a direct effect of BaCl₂ on microvascular cells or consequential to disruption of myofibres and leukocyte infiltration (Hardy et al. 2016; Fernando et al. 2019; Morton et al. 2019). Therefore, to evaluate the effect of BaCl₂ on microvascular SMCs and ECs independent of surrounding myofibres or inflammation, superior epigastric arteries (SEAs), which have a single layer of each cell type, were isolated from the abdominal musculature and exposed to BaCl₂ ex vivo. SMCs were studied in the wall of intact vessels and ECs were evaluated in endothelial tubes following dissociation of SMCs (Socha & Segal, 2013). Controls performed without BaCl₂ confirmed that V_m and [Ca²⁺]_i of SMCs and of ECs remained stable for the duration of recordings (Norton & Segal, 2018; Norton et al. 2019, 2020).

BaCl₂ depolarizes microvascular SMCs with increased [Ca²⁺]_i inducing vasoconstriction

Adding 1.2% BaCl₂ during superfusion of SEAs progressively depolarized SMCs from −48 (7.9) to −19 (2.9) mV over several minutes (P = 0.002, n = 4 vessels from separate mice; Fig. 1A and B). Electrical responses plateaued in 7 (2) min. In SMCs, membrane depolarization is accompanied by a rise in [Ca²⁺]_i through the activation of voltage gated (L-type) Ca²⁺ channels in the plasma membrane (Lipscombe et al. 2004) and subsequent vasoconstriction. As such, exposure to BaCl₂ increased SMC [Ca²⁺]_i (P = 0.03, n = 4 vessels from separate mice; Fig. 1C) and constricted SEAs from 76 (15) to 34 (19) μm (P = 0.001, n = 4 vessels from separate mice; Fig. 1D).

Figure 1. BaCl₂ induces depolarization, SMC Ca²+ influx and vasoconstriction in microvessels without cell death.

A, representative recording of V_m from a SMC in a pressurized SEA. Addition of BaCl₂ initiated progressive depolarization that stabilized after several min. B, BaCl₂ depolarizes SMCs in intact vessels (left) and ECs in endothelial tubes (right). For SMCs, ΔV_m = −29 (6) mV (P = 0.002, n = 4); for ECs, ΔV_m = −21 (3) mV (P = 0.001, n = 4). C, [Ca²+]_i increases in SMCs of intact microvessels (P = 0.03, n = 4) but not in ECs of endothelial tubes (P = 0.31, n = 4) during exposure to BaCl₂. Summary data of [Ca²+]_i measured with Fura-2 fluorescence (F_340/380) before and during BaCl₂ application. D, intact microvessels constrict 55% after 1 h exposure to BaCl₂. Paired average diameter before (baseline) and after BaCl₂ addition (P = 0.001, n = 4). E, BaCl₂ exposure does not induce cell death of SMCs or ECs ex vivo (P > 0.978, n = 3–6). Summary data presented as means (SD); n-values denote vessels each from a separate mouse. For B–D, each colour represents a separate experiment with data analysed using a paired two-tailed Student’s t-test. For E, a 2-way ANOVA with Šídák’s multiple comparisons test was used.

BaCl₂ depolarizes microvascular ECs with no change in [Ca²⁺]_i

In endothelial tubes, addition of BaCl₂ depolarized ECs from −41 (1.9) to −20 (4.2) mV over a similar time course to that in SMCs (P = 0.001, n = 4 endothelial tubes from separate mice; Fig. 1B). In the absence of voltage gated Ca²⁺ channel expression in ECs (Moccia et al. 2012), depolarization of ECs with BaCl₂ had no effect on [Ca²⁺]_i (P = 0.319, n = 4 endothelial tubes from separate mice; Fig. 1C).

BaCl₂ does not directly cause microvascular cell death

Exposing the mouse extensor digitorum longus (EDL) muscle to 1.2% BaCl₂ promptly depolarized myofibres and increased [Ca²⁺]_i, leading to disruption of the sarcolemma and proteolysis culminating in cell death within 1 h (Morton et al. 2019). We hypothesized that BaCl₂ affects microvascular cells in a similar manner. However, contrary to our hypothesis, 1 h exposure to 1.2% BaCl₂ was not lethal to SMCs or ECs in intact vessels or to ECs in endothelial tubes (P > 0.978, n = 3–6 vessels from separate mice; Fig. 1E). Furthermore, extending BaCl₂ exposure to 3 h had no further effect on SMC or EC viability (n = 3 from separate mice, not shown). These data suggest that, rather than a direct action by BaCl₂, the adverse microenvironment within an injured skeletal muscle created by myofibre damage and degeneration leads to disruption of capillary ECs.

Arterioles and venules are spared from BaCl₂-induced muscle injury

Capillaries are fragmented and perfusion is abolished within 24 h of muscle injury (Hardy et al. 2016; Fernando et al. 2019; Morton et al. 2019). At this early time point, leakage of a 70 kDa dextran from residual microvessels suggests a loss of structural integrity of arterioles or venules (Fernando et al. 2019). To investigate whether pre- and post-capillary microvessels are also damaged by muscle injury, the GM of C57Bl/6J mice was injured by injection of 75 μl of 1.2% BaCl₂ under the muscle (Fernando et al. 2019). Whole mount immunostaining for Myh11 labelled SMCs that encircle arterioles and venules embedded within the GM. While other cell types may express Myh11, they neither localize to nor circumscribe the abluminal surface of pre- or post-capillary microvessels, thereby enabling positive identification of SMCs (Miano et al. 1994; Babu et al. 2000). As indicated for isolated microvessels exposed to BaCl₂ ex vivo (Fig. 1E), the structural integrity of SMCs within arteriolar networks of the GM remained intact at 1 dpi, though sporadic localized damage was observed (Fig. 2, n = 5–12 images from 4 mice per time point). Continuous circumferential SMC coverage along arteriolar segments with uniform diameter was not different at 1 dpi compared to uninjured muscle. The integrity of venular segments and their polygonal SMCs was also preserved (Fig. 2). Our quantitative analyses therefore centred on capillary and pre-capillary (resistance) networks.

Figure 2. BaCl₂ injection does not disrupt SMCs of arterioles (a) or venules (v) embedded in the GM.

Representative maximum projection confocal z-stacks showing SMCs identified by Myh11 expression at 0 dpi (A and B) and 1 dpi (C and D). Note example of rare local damage to arteriolar SMCs at 1 dpi. Boxed regions within A and C are enlarged in B and D. Red: immunostaining for Myh11. Brightness of enlarged images of venules is enhanced to optimize visualization. Images are representative of n = 5–12 images from 4 mice per time point.

Role of neutrophils in BaCl₂-induced capillary damage

Muscle injury initiates a stereotypical inflammatory response, with neutrophils invading within 1–2 h of injury and peaking at 12–24 h post-injury (Fielding et al. 1993). Neutrophils release cytolytic and cytotoxic molecules that may damage other resident cell types, along with cytokines that attract monocytes and macrophages to remove cellular debris (Tidball, 2017). Given the disruption of capillaries following intramuscular injection of BaCl₂ (Hardy et al. 2016; Fernando et al. 2019; Morton et al. 2019) when compared to the integrity of ECs exposed to BaCl₂ ex vivo (Fig. 1E), we tested whether invading neutrophils are necessary for capillary fragmentation after BaCl₂ injury. Differential blood counts determined that circulating neutrophils comprised 12.6 (3.6)% (n = 4 mice) of white blood cells in uninjured C57BL/6J mice. At 1 dpi, neutrophils increased to 49.5 (6.4)% of circulating white blood cells (P < 0.0001, n = 4 mice). Intraperitoneal injections of anti-Ly6G ab prior to BaCl₂ injury reduced circulating neutrophils at 0 dpi and at 1 dpi (4.3 (5.9)% and 2.9 (4.0)%, respectively; P < 0.0001; n = 4 mice per group).

Using an EC-specific Cre driver (Cdh5-Cre^ERT2; Wang et al. 2010) and tamoxifen-induced recombination of the Rosa26^mTmG locus to genetically label the endothelium with membrane-bound eGFP (Cdh5-mTmG mice), we evaluated capillary network structure in whole mount preparations of the GM. The orderly arrangement of capillary networks that course along myofibres in uninjured muscle (Fig. 3A) was not impacted by neutrophil depletion alone. Following BaCl₂ injury, capillaries were fragmented at 1 dpi (P < 0.0001, n = 8 images from 4 mice; Fig. 3A), which explains the loss of perfusion at this time (Fernando et al. 2019). While neutrophil depletion did not prevent capillary fragmentation, damage was attenuated as indicated by partial preservation of tissue area occupied by capillaries (P = 0.285, n = 8 images from 4 mice; Fig. 3B). As an index of capillary fragmentation with fewer pathways for blood flow, the maximum connected length of networks was decreased at 1 dpi compared to 0 dpi (P = 0.002, n = 8 images from 4 mice; Fig. 3C); neutrophil depletion prior to muscle injury did not rescue this effect (P = 0.624, n = 8 images from 4 mice). Nevertheless, neutrophil-depleted mice exhibited greater capillary perfusion as visualized by FITC-dextran circulating in the bloodstream of C57Bl/6 mice (n = 4 mice, Fig. 3D).

Figure 3. Skeletal muscle injury with BaCl₂ disrupts capillary networks.

A, representative maximum projection z-stacks of GM from Cdh5-mTmG mice (n = 4–5 mice per condition). Capillary ECs (green; eGFP) are well organized and align with myofibres (oriented vertically, not shown) in uninjured muscle at 0 dpi. At 1 dpi after BaCl₂ injection, capillaries are fragmented with few intact segments. Neutrophil depletion with Ly6G antibody (+ Ly6G ab) affords partial protection against capillary damage at 1 dpi. Scale bar applies to all panels. B, muscle injury with BaCl₂ decreases the percentage vascular area occupied by capillary ECs at 1 dpi. Neutrophil depletion did not prevent capillary fragmentation at 1 dpi (P = 0.285); 1 dpi (P < 0.0001, n = 8) and 1 dpi + Ly6G ab (P < 0.0001, n = 8) vs. 0 dpi (n = 5). C, injury by BaCl₂ decreases the maximum connected length of capillary networks, indicating fragmentation; 1 dpi (P = 0.011, n = 8) and 1 dpi + Ly6G ab (P = 0.002, n = 8) vs. 0 dpi (n = 4). D, representative images from 4 C57Bl/6 mice showing intravascular FITC-dextran (green, 70 kDa) labelling of perfused residual microvascular networks with some dye leakage into surrounding tissue. At 1 dpi, more capillaries remain perfused by FITC-dextran following neutrophil depletion using Ly6G ab. Scale bar applies to both panels. Summary data are means (SD); n-values denote images analysed from 4–5 mice per condition. Data were analysed by 1-way ANOVA with Tukey’s multiple comparisons tests.

Regenerating capillary networks are dilated and disorganized

By 2–3 dpi, endothelial sprouts emerged from surviving capillary segments at multiple initiation points. The ensuing angiogenesis re-established perfused networks at 5 dpi (Fig. 4A) as reported (Fernando et al. 2019). After a ~65% reduction at 1 dpi (Fig. 3B), the percentage vascular area recovered to uninjured levels at 5 dpi (P = 0.994, n = 10 images from 5 mice; Fig. 4B) and did not change thereafter (P > 0.49, n = 10–11 images from 5–7 mice). However, capillary networks appeared disorganized (Fig. 4A) and dilated (Fig. 4C) during their regeneration, with diameter increasing from 4.7 (0.5) μm at 0 dpi (n = 12 images from 5 mice) to 6.9 (0.4) μm at 5 dpi (P < 0.0001, n = 7 images from 5 mice) and 6.1 (1.4) μm at 10 dpi (P = 0.009, n = 6 images from 4 mice). By 21 dpi, capillary diameter (5.7 (0.8) μm) was not different from uninjured controls (P = 0.094, n = 6 images from 4 mice).

Figure 4. Regenerated capillaries are initially dilated when reperfused, but microvascular density is similar throughout regeneration.

A, maximum projections of confocal z-stacks acquired from GM of Cdh5-mTmG mice before injury and during regeneration (ECs: green; eGFP). Representative images from 4–5 mice. Scale bar applies to all panels. B, following capillary fragmentation at 1 dpi (see Fig. 3A), the percentage vascular area has returned to control levels by 5 dpi and is not different throughout regeneration (P > 0.49, n = 10–11 images from 4–7 mice per time point). C, capillaries are dilated at 5 dpi (P < 0.0001, n = 7 images from 5 mice) and 10 dpi (P = 0.009, n = 6 images from 4 mice) but return to uninjured diameter by 21 dpi (P = 0.094, n = 6 images from 4 mice); 0 dpi (n = 12 images from 7 mice). Summary data are means (SD) and analysed by 1-way ANOVA with Tukey’s multiple comparisons tests.

To quantify morphological changes in regenerating capillary networks, we first evaluated the number of branch points. In uninjured muscle (0 dpi), there were 26 (4) branch points/mm of network length (n = 6 images from 4 mice) and this value did not change throughout regeneration (P > 0.672, n = 5–6 images from 4–5 mice per time point; Fig. 5A). However, capillary regeneration was non-uniform. Poorly vascularized regions (not shown) were located adjacent to regions of high angiogenic activity that often had elongating sprouts that connected with neighbouring capillaries to form a dense microvascular mesh (Fig. 5B) resembling a developmental vascular plexus (Risau & Flamme, 1995). Trifurcations not found in uninjured muscle were also present within regenerating networks at 10 dpi, albeit with low incidence (Fig. 5B).

Figure 5. Regenerating capillary networks are disorganized.

A, branch point frequency (number/total capillary length, mm) throughout regeneration is not different from 0 dpi (P > 0.672, n = 5–6 images from 4–5 mice). B, confocal z-stacks of GM from Cdh5-mTmG mice highlighting distinct capillary structures during regeneration. At 5 dpi, multiple sprouts (*) elongate towards adjacent capillaries (a) creating a dense mesh (box) (b). By 10 dpi, trifurcations (∗) are present, which were not found in uninjured muscle (c). Furthermore, multiple anastomoses create plexus-like structures (box) similar to those during development (d) (ECs: green; eGFP). Scale bar applies to all panels. C, regenerated capillaries are less well aligned with myofibres (oriented vertically at 0°) at 5 dpi (P = 0.005, n = 5 images from 4 mice) and 10 dpi (P = 0.008, n = 6 images from 5 mice) compared to uninjured controls at 0 dpi (n = 6 images from 5 mice). By 21 dpi, capillaries have remodelled such that their orientation is no longer different from 0 dpi (P = 0.18, n = 5 images from 4 mice). Grey bar: ±1 SD of Gaussian fit at 0 dpi for capillaries oriented parallel to myofibres (−5 to +5°) superimposed at each time point for reference. D, a microvascular unit (MVU) constitutes a terminal arteriole (*) and the capillaries it supplies in both directions. MVUs are readily identified at 0 dpi but cannot be resolved before 21 dpi because of disoriented segments (e.g. panels in B). Summary data are means (SD) and analysed by 1-way ANOVA with Tukey’s multiple comparisons tests.

Nascent capillaries grew in multiple directions and orientation analyses confirmed that fewer capillaries aligned with regenerating myofibres. To calculate the percentage of capillary segments parallel to myofibres, the distribution of segment angles was plotted relative to myofibres, which were oriented vertically at 0° for reference and horizontally defined as −90° and +90°. Capillary segments in uninjured GM oriented within 1 SD of vertical were considered parallel to myofibres (−5 to +5°). This same interval was used as a reference for 5, 10 and 21 dpi in Fig. 5. In uninjured muscle (0 dpi), 47 (8)% of capillary segments were parallel to myofibres (n = 6 images from 5 mice; Fig. 5C). However, at 5 dpi, only 34 (4)% of regenerated capillaries were parallel with myofibres (P = 0.005, n = 5 images from 4 mice); this structural disorientation persisted through 10 dpi (P = 0.008, n = 6 images from 5 mice). However, by 21 dpi the percentage of capillaries orientated parallel to myofibres was not different from 0 dpi (P = 0.18, n = 5 images from 4 mice). Capillary network disorganization was manifested by the loss of identifiable microvascular units (MVUs), defined as a group of capillaries oriented bidirectionally along myofibres supplied by a common terminal arteriole (Emerson & Segal, 1997). Restoration of organized MVUs characteristic of uninjured skeletal muscle was not apparent until 21 dpi (Fig. 5D), which corresponds with recovery of flow control and maturation of regenerating myofibres in the mouse GM following BaCl₂ injury (Fernando et al. 2019).

Terminal arterioles remodel during muscle regeneration

Finding minimal cellular damage in arterioles (Fig. 2) but extensive remodelling of capillaries and MVUs (Fig. 5), we questioned whether the architecture of pre-capillary resistance networks underwent remodelling during regeneration. At criterion time points, the vasculature of C57BL/6J mouse GM stained with fluorescent wheat germ agglutinin (WGA) was manually traced from the inferior gluteal (feed) artery to terminal arterioles (Fig. 6). Resulting traces included intra-network anastomoses and collateral connections to arterioles originating from the superior gluteal artery. Summary data are based on GM preparations from five to seven separate mice at each time point.

Figure 6. Representative maps of resistance networks from feed artery to terminal arterioles stained with WGA.

0 dpi (A), 5 dpi (B), 10 dpi (C), 21 dpi (D). n = 5–6 GM per time point. Scale bar applies to all panels.

Uninjured muscle contained 452 (175) microvessel segments (n = 5 mice; Fig. 7A). During regeneration, the total number of segments was not different compared to 0 dpi; the total vascular length was also not different at the time points studied. However, when categorized by microvessel diameter, the number of arterioles 5–10 μm in diameter (i.e. terminal arterioles) increased at 5 dpi compared to 0 dpi and remained elevated through 21 dpi (n = 5–7 mice; Fig. 7B). The corresponding cumulative vascular length of the smallest arterioles was also increased at 5 dpi vs. 0 dpi indicating proliferation of terminal arterioles (P = 0.031, n = 7 mice; Fig. 5C). In addition, more anastomoses were present in microvascular networks at 5 dpi than other time points (P = 0.048, n = 7 mice; Fig. 7D), reflecting pronounced angiogenesis during the early stages of regeneration.

Figure 7. Terminal arterioles in resistance networks proliferate after muscle injury.

A, the total number of segments in resistance networks was not significantly different between uninjured GM and those during regeneration (P > 0.863, as depicted in Fig. 6). B, the number of segments 5–10 μm in diameter increase at 5 dpi (P = 0.003), which persists through 21 dpi (P = 0.024, n = 5–7). Error bars removed for clarity; ∗P vs. 0 dpi, #P vs. 5 dpi, †P vs. 10 dpi. C, the cumulative length of terminal arterioles in GM at 5 dpi is increased compared to uninjured GM (n = 5–7). Error bars omitted for clarity. ∗P = 0.031 vs. 0 dpi, #P = 0.03 vs. 5 dpi. D, anastomoses are more prevalent in arteriolar networks at 5 dpi compared to 0 dpi. By 10 dpi, the number of anastomoses returned to uninjured levels (∗P = 0.048 vs. 0 dpi, n = 5–7). Summary data are means (SD); n-values denote GM preparations from separate mice. A and D were analysed by 1-way ANOVA with Tukey’s multiple comparisons tests; Kolmogorov–Smirnov tests were used in B and C.

Discussion

Second only to myofibres, microvascular cells represent the largest cell population in healthy adult skeletal muscle (Rubenstein et al. 2020). This relationship underscores the reliance of skeletal muscle on a robust microvascular supply to support the metabolic demands of myofibres and their survival. Understanding how respective cellular components of intact muscle respond to injury and work in concert during regeneration is essential to developing effective therapies for restoring tissue structure and function following injury. In defining the nature of myofibre damage after acute exposure of the GM to BaCl₂, we found concurrent disruption of capillary perfusion (Fernando et al. 2019; Morton et al. 2019). While our findings of capillary fragmentation are consistent with those described in the mouse tibialis anterior (TA) muscle for a variety of injuries (Hardy et al. 2016), we are the first to test whether direct exposure to BaCl₂ results in microvascular cell death and to evaluate regeneration of the entire microcirculation following capillary fragmentation from myofibre degeneration.

Microvascular cell damage after muscle injury

The depolarization of SMCs and ECs in response to BaCl₂ (Fig. 1A and B) is consistent with the action of Ba²⁺ as a broad-spectrum K⁺ channel inhibitor (Bonev & Nelson, 1993; Nelson & Quayle, 1995). However, the consequences of such an effect differ between microvascular cells and myofibres. In myofibres, BaCl₂-induced depolarization is accompanied by a rise in [Ca²⁺]_i resulting in proteolysis, membrane disruption and cell death within 1 h (Morton et al. 2019). Elevation of [Ca²⁺]_i can increase mitochondrial Ca²⁺ content to trigger cytochrome c release and intrinsic apoptosis (Pinton et al. 2008). For SMCs exposed to the same conditions, membrane depolarization also increased [Ca²⁺]_i, which led to sustained vasoconstriction (Fig. 1D), as seen in rabbit aorta (Karaki et al. 1986). However, since SMC death did not occur after the same duration of exposure as myofibres (or even when extended to 3 h), the present findings indicate that either longer exposure to BaCl₂ is necessary (which is unlikely given the equal probability for BaCl₂ diffusion into both cell types) or that the rise in [Ca²⁺]_i was not sufficient to induce SMC death. Finding negligible disruption of arteriolar or venular SMCs at 1 dpi (Fig. 2) when myofibres have degenerated in the GM supports the latter explanation. We suggest that differences in mitochondria content, expression of distinct voltage gated Ca²⁺ channel isoforms, the ability for the sarcoplasmic/endoplasmic reticulum to sequester Ca²⁺ without dysfunction or the expression of pro-apoptotic proteins may explain the disparity in cell death between myofibres and microvascular SMCs exposed to BaCl₂ (Park et al. 2014; Cho et al. 2017).

In light of their disruption in vivo, it has not been possible to isolate intact capillaries for ex vivo studies. As an alternative approach, we isolated endothelial tubes, which retain properties of the intact endothelium (Behringer et al. 2012. Recording from these microvascular ECs demonstrated that depolarization in response to BaCl₂ occurred without an increase in [Ca²⁺]_i (Fig. 1B and C), a finding consistent with their lack of voltage operated Ca²⁺ channels in the EC membrane (Moccia et al. 2012). Furthermore, no cell death was evident in endothelial tubes exposed to BaCl₂ (Fig. 1E). That ECs comprising capillaries are damaged following BaCl₂ injection in vivo (Fig. 3A; Hardy et al. 2016; Morton et al. 2019) but ECs directly exposed to BaCl₂ ex vivo are not (Fig. 1E), suggests that ECs in arteriolar and venular networks may be protected by SMCs and that the otherwise exposed capillary ECs are injured secondary to myofibre degeneration from BaCl₂ exposure. Upon exposure to BaCl₂, resting force of the mouse EDL muscle increases ~40% over 15 min, which then returns to baseline before the end of an hour as proteolysis disrupts the integrity of contractile proteins (Morton et al. 2019). Given that there is no evidence of capillary damage during this time, the present data imply that skeletal muscle tissue (myofibre) degeneration induces capillary fragmentation many hours after injury (e.g. by 1 dpi), rather than as a consequence of persistent (<1 h) myofibre contraction or by direct action of BaCl₂.

Within 1–2 h of trauma to myofibres, skeletal muscle is invaded by neutrophils that generate pro-inflammatory cytokines, chemokines and reactive oxygen species which exacerbate tissue damage (Tidball, 2017). While neutrophil depletion prior to BaCl2 injection did not prevent capillary damage at 1 dpi (Fig. 3A and B), more capillaries remained perfused when compared to mice in which neutrophils were not depleted (Fig. 3D). This outcome suggests that while neutrophil aggregation and the resulting hypoxia may contribute to capillary damage (e.g. generation of reactive oxygen species or plugging of capillaries; Schmid-Schönbein, 1993), additional mechanisms lead to EC death when skeletal muscle is injured. Candidates include toxic substances released from degenerating myofibres, calpain-dependent protein degradation, along with recruitment of proinflammatory cells including monocytes and macrophages (Diez-Roux & Lang, 1997; Clanton, 2007; Tidball, 2017; Zhang et al. 2017; Silvis et al. 2020).

Restoration of capillary networks during muscle regeneration

Previous studies investigating capillarity during regeneration evaluated muscle cross sections (Luque et al. 1995; Hardy et al. 2016) and therefore could not address the dynamics of microvascular network organization. Cross sections are inherently biased towards vessels oriented parallel to myofibres (Skalak & Schmid-Schönbein, 1986), which compromises the resolution of transverse microvessels and anastomoses. Moreover, studies that have examined the intact microcirculation during skeletal muscle regeneration have focused on hindlimb muscles such as the TA (Hardy et al. 2016) and EDL (Arpino et al. 2017), which restricts observations to the superficial portion of the muscle due to the thickness of the tissue. While capillaries, terminal arterioles and collecting venules can be observed, proximal resistance networks reside deeper within the muscle. To overcome these limitations, we evaluated microvascular regeneration in the GM, a thin, flat skeletal muscle well suited to high resolution imaging and analysis of entire microvascular networks when reflected away from the torso to observe the ventral surface (Bearden et al. 2004; Fernando et al. 2019).

Triggered by the release of such growth factors as vascular endothelial growth factor (VEGF), fibroblast growth factor and insulin-like growth factor 1 from the hypoxic milieu (Krock et al. 2011), nascent capillaries sprout and elongate from surviving microvessel fragments at 2–3 dpi (Hardy et al. 2016; Morton et al. 2019). These regions of angiogenic activity within the tissue highlight the selective roles of individual ECs during angiogenesis. As shown during development, endothelial tip and stalk cells are selected after stimulation by VEGF and Notch signalling (Blanco & Gerhardt, 2013). Tip cells guide capillary sprouts as they sense attractive and repulsive signals in the microenvironment. Stalk cells trail tip cells and proliferate to elongate nascent capillaries such that by 5 dpi, capillary networks have reformed (Fig. 4A) and perfusion is restored (Fernando et al. 2019). While others have reported no change or even a reduction in the diameter of capillaries during the early stages of muscle regeneration (Hardy et al. 2016), we observed a ~30% increase in capillary diameter at 5 and 10 dpi (Fig. 4B). Vasodilatation is one of the earliest steps in both physiological and pathological angiogenesis (Ziche & Morbidelli, 2000). That the entire resistance network is also dilated during this time (Fernando et al. 2019) suggests that regenerating skeletal muscle requires elevated blood flow afforded by a dilated microvasculature to provide ample nutrients for cellular regeneration. Capillary dilatation may reflect aberrant coverage by mural cells (pericytes) or dysregulated inter- and intracellular signalling (McGahren et al. 1998; Peppiatt et al. 2006; Hamilton et al. 2010).

Following injury of the GM, adjacent to sites of angiogenesis and regeneration are areas with extensive damage that are poorly perfused or avascular at 5 dpi, implying that microvascular regeneration after skeletal muscle injury is asynchronous between neighbouring regions of tissue. Heterogeneity amongst ECs exists between vascular beds, microvessels and even along the length of a single microvessel in healthy skeletal muscle (Aird, 2012; Jambusaria et al. 2020). Corresponding differences in gene expression and nuances in EC function may explain the asynchronous regenerative response of the microvasculature. In response to tissue hypoxia, angiogenesis occurs primarily in areas containing glycolytic (type II) myofibres (Deveci et al. 2001; Badr et al. 2003), suggesting that differential requirements for oxygen and metabolites further affect the angiogenic response. That the GM is a muscle comprised of mixed fibre types (Lampa et al. 2004) is consistent with the observed heterogeneity in angiogenesis during regeneration of the GM.

In addition to the density of capillary networks, optimal distribution of capillaries is crucial for adequate muscle oxygenation. Heterogeneous capillary spacing negatively impacts muscle oxygenation due to local increases in diffusion distances (Rakusan & Turek, 1985; Degens et al. 2006). Microvascular structures resembling a developmental vascular plexus were present throughout the GM at 5 and 10 dpi (Fig. 5B), with disorganized capillary networks containing abnormal branching (e.g. trifurcations) and more segments that were not aligned with regenerating myofibres (Fig. 5C). Similar abnormal microvascular morphogenesis has been described after grafting the hamster TA muscle (Messina & Carlson, 1991) and ischaemic injury to mouse EDL muscle (Arpino et al. 2017).

A microvascular unit constitutes a terminal arteriole and the group of capillaries it supplies (Emerson & Segal, 1997; Segal, 2005). Following BaCl₂ injury of the GM, the organization of capillaries into discernible MVUs apparent in uninjured muscle was not identifiable until 21 dpi (Fig. 5D), the time point which coincides with the restoration of myofibre number and size in addition to the recovery of blood flow control in the resistance vasculature (Fernando et al. 2019). In uninjured, healthy skeletal muscle, terminal arterioles orient obliquely to myofibres, whereas capillaries primarily run parallel with myofibres. When a terminal arteriole constricts or dilates, flow through all capillaries it supplies diminishes or increases, respectively. We anticipated that nascent microvessels would repopulate residual basement membranes (Hansen-Smith et al. 1980; Hansen-Smith, 2000; Mancuso et al. 2006; Kelly-Goss et al. 2013) following microvascular damage from skeletal muscle injury (Vracko & Benditt, 1972; Hardy et al. 2016). However, in contrast to the organization of MVUs in healthy muscle, regenerated capillary networks were less aligned with myofibres, contained irregular structures and required 21 dpi to approximate normal structure. That network organization was not fully recovered at 21 dpi is consistent with findings of others that microvascular network structure remains flawed upwards of 6 months after injury (Hardy et al. 2016; Arpino et al. 2017). Such a delay in reorganization implies that nascent capillaries need not regrow along basement membranes of old microvessels. In turn, these structural abnormalities in regenerating capillary networks may contribute to non-uniform perfusion during the early stages of myofibre regeneration. Nevertheless, the persistence of arteriolar dilatation through 10 dpi (Fernando et al. 2019) promotes capillary perfusion, which may help compensate for limitations imposed by aberrant capillary network organization.

Remodelling of arteriolar networks during regeneration

Muscle injury by BaCl₂ did not induce vascular cell death or structurally damage the resistance (or adjacent venular) network upstream from terminal arterioles (Fig. 2). Evaluating network architecture during regeneration revealed an increase in the number of the smallest (i.e. terminal) arterioles 5–10 μm in diameter that persisted through 21 dpi (Fig. 7B and C). A similar increase in small arterioles was observed after ischaemic injury in the mouse spinotrapezius muscle after arterial ligation (Bailey et al. 2008). We also observed an increase in the number of anastomoses in arteriolar networks at 5 dpi (Fig. 7D). These direct connections ensure redundancy in blood flow paths to limit underperfusion while promoting maximum perfusion of regenerating tissue before local blood flow control recovers. Although the origin of new arterioles in microvascular networks during regeneration remains incompletely understood, nascent capillaries may become ‘arterialized’ by recruiting mural cells (Price et al. 1994; Mac Gabhann & Peirce, 2010). Together, the present data suggest that the repair, recovery and maintenance of new myofibres requires a robust supply of oxygen and nutrients from the regenerating microcirculation preceding the recovery of blood flow regulation.

Conclusion

Skeletal muscle is highly vascularized. Capillaries are located parallel to and in close association with myofibres and are essential for providing oxygen and nutrients while removing metabolic byproducts. The present study shows that capillaries, but not SMCs or ECs of larger microvessels, are structurally damaged by the microenvironment created by degenerating myofibres. When perfusion is restored, nascent capillary networks are dilated, disorganized and associated with more terminal arterioles. These early morphological adaptations may compensate for lack of blood flow regulation during myofibre regeneration (Fernando et al. 2019). Reorganization of capillaries and terminal arterioles into MVUs at 21 dpi coincides with restoration of the number of myofibres, their cross-sectional area and the recovery of blood flow regulation in resistance networks. Understanding how microvascular structure and function are restored following skeletal muscle injury provides new insight for developing therapeutic interventions for the treatment of acute muscle trauma in adults.

Translational perspective.

Despite the integral role of the microcirculation in supporting normal tissue function, little is known of what happens to the microvasculature in the context of skeletal muscle injury and regeneration. We used complementary ex vivo and in vivo approaches to show that capillary endothelial cells are killed secondary to myofibre degeneration rather than direct action by the myotoxin BaCl₂. Therefore, controlling the noxious substances present in the tissue microenvironment during conditions of myofibre degeneration (e.g. trauma, overexertion or disease) may preserve microvascular structure and function. After the fragmentation and destruction of capillaries, networks regenerate in a disorganized manner, are dilated and are associated with an increased number of terminal arterioles. These initial outcomes may compensate for the lack of blood flow control during the early stages of regeneration and suggest that regenerating myofibres require maximal perfusion as inflammation subsides and nascent myofibres mature. Capillary networks remodel into discernible microvascular units coincident with the recovery of the number and cross-sectional area of nascent myofibres. Understanding how intrinsic cellular components recover in concert and resolving the time course of microvascular repair provide a novel foundation for developing strategies to attenuate tissue damage and to restore skeletal muscle function following injury in humans.

Key points.

• Skeletal muscle regenerates after injury; however, the nature of microvascular damage and repair is poorly understood. Here, the myotoxin BaCl₂, a standard experimental method of acute skeletal muscle injury, was used to investigate the response of the microcirculation to local injury of intact muscle.

• Intramuscular injection of BaCl₂ induced capillary fragmentation with myofibre degeneration; arteriolar and venular networks remained intact. Direct exposure to BaCl₂ did not kill microvascular endothelial cells or smooth muscle cells.

• Dilated capillary networks reformed by 5 days post-injury (dpi) in association with more terminal arterioles. Capillary orientation remained disorganized through 10 dpi.

• Capillaries realigned with myofibres and reorganized into microvascular units by 21 dpi, which coincides with the recovery of vasomotor control and maturation of nascent myofibres.

• Skeletal muscle injury disrupts its capillary supply secondary to myofibre degeneration. Reorganization of regenerating microvascular networks accompanies the recovery of blood flow regulation.

Biography

Nicole Jacobsen is currently a postdoctoral fellow in Steven Segal’s laboratory at the University of Missouri, Columbia. Her interest in microvascular injury and repair developed as an undergraduate and PhD student studying channel-independent functions of cardiovascular gap junction channels in determining cell death vs. survival at the University of Arizona. Her long-term interests include understanding how heterocellular communication regulates pathological microvascular remodelling. She is currently exploring the role of satellite cells in microvascular regeneration after skeletal muscle injury.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.