Impaired skeletal muscle fatigue resistance during cardiac hypertrophy is prevented by functional overload‐ or exercise‐induced functional capillarity

Capillary rarefaction is hypothesized to contribute to impaired exercise tolerance in cardiovascular disease, but it remains a poorly exploited therapeutic target for improving skeletal muscle performance. Using an abdominal aortic coarctation rat model of compensatory cardiac hypertrophy, we determine the efficacy of aerobic exercise for the prevention of, and mechanical overload for, restoration of hindlimb muscle fatigue resistance and microvascular impairment in the early stages of heart disease. Impaired muscle fatigue resistance was found after development of cardiac hypertrophy, but this impairment was prevented by low‐intensity aerobic exercise and recovered after mechanical stretch due to muscle overload. Changes in muscle fatigue resistance were closely related to functional (i.e. perfused) microvascular density, independent of arterial blood flow, emphasizing the critical importance of optimal capillary diffusion for skeletal muscle function. Pro‐angiogenic therapies are an important tool for improving skeletal muscle function in the incipient stages of heart disease.

r Using an abdominal aortic coarctation rat model of compensatory cardiac hypertrophy, we determine the efficacy of aerobic exercise for the prevention of, and mechanical overload for, restoration of hindlimb muscle fatigue resistance and microvascular impairment in the early stages of heart disease.
r Impaired muscle fatigue resistance was found after development of cardiac hypertrophy, but this impairment was prevented by low-intensity aerobic exercise and recovered after mechanical stretch due to muscle overload.
r Changes in muscle fatigue resistance were closely related to functional (i.e. perfused) microvascular density, independent of arterial blood flow, emphasizing the critical importance of optimal capillary diffusion for skeletal muscle function.
r Pro-angiogenic therapies are an important tool for improving skeletal muscle function in the incipient stages of heart disease. Peter G. Tickle is working on aspects of skeletal muscle physiology and biomechanics as a postdoctoral researcher at the University of Leeds. He is particularly interested in the underlying factors that regulate muscle performance such as muscle blood flow, the microcirculation, energy consumption and the influence of disease. His current work is focused on characterizing the mechanics of very high frequency insect flight muscle contractions using the work loop technique with Dr Graham Askew. Paul W. Hendrickse received his PhD from Manchester Metropolitan University for his research into the role of the microcirculation in skeletal muscle function and plasticity. He currently works as a postdoctoral research associate under the direction of Professor Hans Degens, studying the effect of countermeasures to unloading in bed rest on muscle morphology.

Introduction
Persistent exercise intolerance is a hallmark of chronic heart failure (CHF), compromising quality of life and contributing to a poor clinical prognosis. Skeletal muscle performance is, however, poorly correlated with left ventricular ejection fraction in CHF (Franciosa et al. 1981;Rogers, 2001), suggesting that peripheral factors determine the severity of exercise intolerance (Pandey et al. 2015). Further pathological changes in CHF contributing to impaired exercise performance include sarcopenia and a slow-to-fast change in muscle fibre type composition (Drexler et al. 1992). In addition, there is growing evidence for constraints imposed by structural and functional changes in the microcirculation in the muscle (Duscha et al. 1999;Richardson et al. 2003;Tickle et al. 2020). Reduced skeletal muscle microvascular density occurs in both experimentally induced (Kindig et al. 1999;Nusz et al. 2003;Richardson et al. 2003;Bowen et al. 2017) and clinical (Schaufelberger et al. 1995;Duscha et al. 1999;Wadowski et al. 2018) CHF, and this rarefaction imposes a limit on the transfer of oxygen and nutrients, and removal of waste products, to and from respiring muscle tissue. Indeed, there is strong evidence for a key role of the perfused microcirculation in determining performance of cardiac (Hauton et al. 2015) and skeletal muscle (Tickle et al. 2020). The perfused, or functional, microcirculation is expected to be equivalent to the total, or anatomical, microcirculation in healthy tissue, whereby every vessel is available for blood flow. Experimentally induced arteriolar blockade to simulate the early stages of functional rarefaction indicates that muscle performance is strongly associated with the degree of perfused vessels, regardless of total underlying capillarity (Tickle et al. 2020). Skeletal muscle capillary rarefaction after imposition of cardiac dysfunction indicates that systemic effects may occur soon (<21 days) after development of perturbed heart function (Nusz et al. 2003), potentially due to enhanced endothelial apoptosis (Rossig et al. 2000) or increased vasoconstriction leading to arteriolar closure (Prewitt et al. 1982). Coupled to the deleterious effects of rarefaction, a disturbed anatomical distribution of capillaries may also compromise muscle function because increased heterogeneity of capillary spacing can reduce tissue oxygenation (Piiper & Scheid, 1991;Degens et al. 2006b). Furthermore, animal models of CHF demonstrate reduced red blood cell velocity and flux (Richardson et al. 2003), in addition to an increased proportion of capillaries with intermittent RBC flux at rest and during contractile activity (Kindig et al. 1999). The consequent increase in transit time potentially explains the higher oxygen extraction observed in CHF (Katz et al. 2000), but also suggests that muscle oxygenation is impaired, leading to suboptimal mitochondrial function (Behnke et al. 2007).
Improving muscle and exercise performance in CHF may therefore depend upon restoration of optimal topology of the microcirculation. However, it remains unknown whether capillary distribution and/or capillary perfusion is changed during CHF, and whether these putative changes affect skeletal muscle oxygenation, performance and capacity for adaptive remodelling.
Capillary proliferation from pre-existing vessels (angiogenesis) in skeletal muscle is an essential adaptation to match tissue performance to changing functional demands (Brodal et al. 1977;Zumstein et al. 1983) that is triggered by both physical and metabolic stimuli (Hudlicka, 1991). For example, the increased longitudinal strain on the abluminal capillary surface (Egginton et al. 2001) during functional overload of a muscle (via extirpation of a muscle synergist, hereafter 'overload') not only results in hypertrophy (Zhou et al. 1998;Deveci & Egginton, 2002;Tickle et al. 2020) and enhanced fatigue resistance (i.e. slower rate of active force decline) (Frischknecht & Vrbova, 1991;Ballak et al. 2016;Tickle et al. 2020), but also induces angiogenesis (Egginton et al. , 2011Williams et al. 2006;Ballak et al. 2016). The significance of angiogenesis for function is illustrated by the observation that even after experimentally induced functional capillary rarefaction, overload-mediated angiogenesis can recover capillarization and fatigue resistance in otherwise healthy muscle (Tickle et al. 2020).
In addition to the pro-angiogenic effects of longitudinal strain, capillary proliferation can be induced by enhanced vascular shear stress via application of a hyperaemic stimulus such as occurs during chronic electrical stimulation of muscle (Hudlicka et al. 1977;Egginton & Hudlicka, 1999) or vasodilator administration Mandel et al. 2016). While the hyperaemic stimulus undoubtedly contributes to expansion of the microvascular bed (Waters et al. 2004;Olesen et al. 2010) with prolonged aerobic exercise training (Andersen & Henriksson, 1977), exercise-induced angiogenesis is also mediated by metabolic (e.g. hypoxia, glucose metabolism) factors (Olfert et al. 2016). In clinical (Gustafsson et al. 2001;Esposito et al. 2010Esposito et al. , 2018 and experimental (Ranjbar et al. 2017) CHF, enhanced capillarity and angiogenic signalling (e.g. via vascular endothelial growth factor, VEGF) are found after exercise therapy, indicating that the capacity for remodelling is still present despite established disease. Improvements in peak oxygen consumption after exercise therapy in CHF patients are also considered to be driven by improvements in peripheral microvascular function (Haykowsky et al. 2012). In CHF patients without exercise therapy, contradictory data on angiogenic marker expression have been reported. For instance, elevated VEGF (Valgimigli et al. 2004) is hypothesized to drive repair of endothelial damage rather than expanding microvascular density per se (Chong et al. 2004). In another study, however, reduced VEGF expression (Arakawa et al. 2003) was indicative of regulatory dysfunction and the potential of multiple factors (e.g. severity and specific type of CHF) to affect capillary proliferation in muscle tissue. Nevertheless, the beneficial effects of exercise-induced angiogenesis for restoring muscle function suggest that intrinsic remodelling capacity may be harnessed, but this remains a poorly exploited therapeutic target.
The principal objective of this paper was therefore to quantify the functional and structural effects of angiogenic stimuli on skeletal muscle during compensatory cardiac hypertrophy (CCH) induced by coarctation of the abdominal aorta (Degens et al. 2006a), a model of incipient cardiac dysfunction. We tested in this rat model the following specific hypotheses: (1) CCH has a deleterious effect on skeletal muscle microcirculation and fatigue resistance; (2) exercise prevents the adverse effects of CCH on skeletal muscle; (3) overload has a restorative effect on impaired skeletal muscle capillarization and fatigue resistance in rats with CCH. By quantifying the effects of cardiac dysfunction on skeletal muscle function and microcirculation, we provide further evidence that the perfused microvascular bed is a therapeutic target for skeletal muscle dysfunction in chronic diseases such as CHF.

Ethical approval
All experimental work complied with the UK Animals (Scientific Procedures) Act 1986, and local approval (70/08674) was granted by the University of Leeds Animal Welfare and Ethical Review Committee. All experiments conformed to the principles and regulations described by guidelines published in the Journal of Physiology (Grundy, 2015).

Aortic constriction and EDL overload
Experiments were designed to enable comparison of the effects of aerobic exercise and/or functional overload on muscle morphology, performance and blood flow after establishment of CCH. In-house bred male Wistar rats were used throughout. While heart disease is not specific to either sex, male rats were used as an extensive body of complementary data is available, and constraints of time and funding precluded an expansion of the study. Compensatory cardiac hypertrophy was induced via abdominal aortic constriction using a titanium band ('banded' groups). Groups of aortic banded animals were allocated as follows: (i) banded control (Aob: 304 ± 6 g, N = 7); (ii) unilateral overload of the EDL (Aob+OV: 376 ± 6 g, N = 7); (iii) voluntary wheel running (Aob+EX: 305 ± 12 g, N = 7); (iv) acute banded, in which the J Physiol 599.15 immediate effects of banding were quantified (Acute-Aob: 245 ± 40 g, N = 6). Corresponding non-banded groups were also used: (i) intact (Control: 257 ± 8 g; N = 7); (ii) overload (OV: 343 ± 9 g, N = 7); (iii) wheel running (EX: 316 ± 41 g, N = 8). Only EX and Aob+EX had access to a cage wheel, all other groups were sedentary. All groups had ad libitum access to food and water and were housed under a 12 h light/dark cycle.
Under isoflurane anaesthesia (induction 4%, maintenance 2.5%) rats underwent surgery to constrict the abdominal aorta immediately cranial to the renal artery bifurcations using a titanium clip (post-mortem measured clip diameter: 0.37 ± 0.07 mm Ligaclip; Ethicon Endo-Surgery Inc., Cincinnati, OH, USA) (Cornelussen et al. 1994;Levy et al. 1996;Degens et al. 2006a). Depth of anaesthesia was regularly assessed by testing the pedal withdrawal reflex. Aob and Aob+EX were otherwise intact. Unilateral mechanical overload of the extensor digitorum longus (EDL) (Zhou et al. 1998;Deveci & Egginton, 2002;Tickle et al. 2020) was performed in Aob+OV 4 weeks after clip placement. In this procedure, extirpation of the tibialis anterior induces stretch and functional overload of the EDL. Post-operative analgesia (buprenorphine (Vetergesic, Ceva, Amersham, UK) 0.05 mg kg −1 ) and antibiotic (Enrofloxacin (Baytril, Bayer, Reading, UK) 2.5 mg kg −1 ) were provided after all surgical procedures. Muscle fatigue resistance and hindlimb perfusion in Aob and Aob+EX was quantified 4 weeks after aortic constriction, and in Aob+OV after an additional 2 weeks.
The overload model of muscle angiogenesis was selected because the success, or lack thereof, depends upon weight-bearing and voluntary activity; actions that may be easily translated to the clinical setting. In contrast, other effective angiogenic stimuli such as imposition of a clean stretch or elevated shear stress (e.g. via electrical stimulation or α 1 -blocker administration) were considered less suitable in this context because they either require unwieldy treatment (e.g. fixing muscle length for long periods by limb immobilization (Goldspink et al. 1995)), potentially incur pharmaceutical side-effects, may drive angiogenesis without corresponding functional benefit (Kissane et al. 2021), and/or may compound CHF-derived hypotension via vascular dilatation.

Wheel running
EX and Aob+EX animals were housed in a cage with ad libitum access to an instrumented running wheel that enabled real-time monitoring of night-time exercise performance (distance moved, exercise duration and running speed) via bespoke LabVIEW software ('Rodent voluntary exercise analysis system' developed in conjunction with the University of Manchester, UK). Access to the wheel was provided immediately after aortic constriction surgery for Aob+EX. To ensure access to the running wheel was the same for each animal and exclude erroneous data due to disturbance, all cage wheels were locked during weekly cage cleaning and welfare checks.

Muscle fatigue resistance and hindlimb perfusion
Protocols for measuring muscle fatigue resistance and arterial blood flow are described in Tickle et al (2020). In brief, anaesthesia was induced with isoflurane (4% in 100% O 2 ) and thereafter maintained by constant alfaxalone (Jurox, Crawley, UK) infusion (30-35 mg kg hr −1 ) delivered via a catheter in the external jugular vein. Implanted carotid and tail artery catheters allowed continuous measurement (blood pressure transducer: AD Instruments, UK) of central and peripheral blood pressure, and heart rate. Adequate depth of surgical anaesthesia was regularly assessed by reference to heart rate and blood pressure in addition to tests of the pedal withdrawal reflex.
Bilateral EDL isometric twitch force at optimal length was measured by linking each muscle to a lever arm force transducer (305B-LR: Aurora Scientific, Aurora, ON, Canada) and providing supramaximal electrical stimulation via the popliteal nerve (Hudlicka et al. 1977). Unimpeded access to the EDL was facilitated by extirpation of the overlying synergist tibialis anterior. To determine EDL fatigue resistance, a 30 s period of 1 Hz twitches, to activate the metabolic machinery, was followed by 10 Hz impulses (0.3 ms pulse width, supramaximal voltage) for 180 s (Egginton & Hudlicka, 1999;Tickle et al. 2020). No evidence of muscle ischaemia (rapid decline in twitch force, slowing of twitch time course) was noted during experiments, indicating that the stimulation protocol did not constrain muscle blood flow via increased intra-muscular pressure. Fatigue resistance was quantified as a 'fatigue index' (FI) that was calculated as muscle force at the end of the test divided by peak muscle tension at the beginning of the test. A mean of five consecutive twitches was selected to represent end-stimulation and peak force data.
The femoral artery blood flow (FBF) was measured bilaterally throughout each experiment with perivascular flow probes (0.7PSB; Transonic, Ithaca, NY, USA) at the proximal aspect of the profunda femoris arterial bifurcation (Tickle et al. 2020). The muscle mass-specific increase in flow, i.e. hyperaemic increment above resting, was calculated to account for potential differences in EDL mass and transformed to femoral vascular conductance using arterial pressure measured in the tail. While measured continuously, for clarity we present blood flow at rest and immediately post-stimulation (when peak blood flow was typically evident).
Following the conclusion of each experiment, capillary perfusion was visualized by an injection of fluorescein isothiocyanate (FITC) -labelled dextran solution (50 mg kg −1 , MW = 500,000; Sigma, Poole, UK) via the carotid artery (Tickle et al. 2020). Before and after injection, the EDL in both legs were stimulated at 4 Hz for 30 s to induce functional hyperaemia and capillary recruitment. EDL were left in situ for a minimum of 1 min after injection to maximize dye localization in perfused capillaries (Snyder et al. 1985). While still under general anaesthesia, animals were then killed by cervical dislocation.

Histology
Immediately after animals were killed, the EDL muscles were removed, blotted dry and weighed. The muscle was then frozen on cork with OCT embedding medium (Thermo Scientific, Loughborough, UK) using liquid nitrogen-cooled isopentane and stored at −80°C. Ten-micrometre cryostat sections were prepared and labelled with biotinylated Griffonia simplicifolia lectin I (5 μl ml −1 FL-1101 GSL I; Vector Labs, Peterborough, UK) and streptavidin Alexa Fluor 350 conjugate. For fibre typing, serial sections were blocked in 10% goat serum in PBS for 60 min (Vector Laboratories, USA) then incubated for 120 min with monoclonal antibodies BAD-5 (1:600), SC-71 (1:600), 6H1 (1:50) on one slide and BF-F3 (1:100) on another in blocking solution for fibre types I, IIa, IIx and IIb, respectively (Developmental Studies Hybridoma Bank, USA). After three 5 min washes in PBS the sections were incubated in the dark with secondary antibodies Alexa Fluor IgG2b for type I (1:500), Alexa Fluor IgG1 for type IIa (1:500) and Alexa Fluor 555 IgM for types IIb and IIx (1:500) (Thermofisher Scientific, USA) in blocking solution. After three further 5 min washes the slides were mounted with ProLong Diamond Antifade mountant (Thermofisher Scientific, USA). Whole-muscle cross-sectional images were acquired with a confocal microscope (Leica TCS SP5) at ×20 magnification. Subsequent image analysis with BTablet and Anatis (BaLoH Software, Ooij, The Netherlands) enabled determination of fibre type, fibre cross-sectional area and whether capillaries were perfused (i.e. FITC labelled). To account for the regional heterogeneity of capillary distribution and fibre-type composition (Kissane et al. 2018) three regions of interest (475 × 475 μm 2 ) of the whole-muscle image were used to establish an unbiased counting frame with which to quantify muscle fibre type composition (numerical fibre type proportions and the specific fibre cross-sectional area, FCSA), capillarization (expressed as capillary to fibre ratio, C:F, or capillary density, CD) and capillary domain area (CDA). CDA quantifies the area surrounding a capillary defined by equidistant boundaries from adjacent capillaries and is an index of the area of tissue to which a capillary may supply adequate oxygen to working muscle (Hoofd et al. 1985;Al-Shammari et al. 2014). Heterogeneity of capillary distribution is provided by Log RSD . Capillary localization was completed as described in Tickle et al. (2020). Sample sizes for histological parameters were reduced in EX (N = 5) and OV (N = 5) due to tissue damage.
Mathematical modelling of tissue PO 2 distribution was performed in representative (N = 1 in each group) EDL histological sections using a custom MATLAB program ('Oxygen Transport Modeler') (Al-Shammari et al. 2019). Model input assumptions include capillary radius, muscle oxygen demand, myoglobin concentration, oxygen solubility and diffusivity; and since measurements of these parameters were not possible in this study, values were applied uniformly across groups (Tickle et al. 2020). Therefore, while the absolute values produced by this model should be interpreted with caution, the calculated tissue PO 2 serves as an indicative comparison of the effects of functional capillary rarefaction between groups.

Statistical analyses
In all cases, differences in FI, blood flow and histological parameters between groups were quantified using ANOVA with Tukey's post hoc tests. ANCOVA with Sidak's post hoc tests were used to test whether aortic banding influenced heart mass after controlling for the scaling effect of body mass (Solomon & Bengele, 1973). Statistical testing was conducted in SPSS (v.25). Data are presented throughout as means ± SD unless stated otherwise and statistical tests were considered significant where P < 0.05.

Cardiovascular effects of aortic banding
Acute-Aob had an mean arterial (carotid) pressure that did not significantly differ from any group, while tail pressure was lower than all other groups (P < 0.002), thereby producing a greater carotid:tail BP ratio (Table 1; P < 0.001). Carotid pressure was higher after chronic banding compared with all unbanded groups ( Fig. 1; Con vs. Aob: P = 0.001; EX vs. Aob+EX: P = 0.026; OV vs. Aob+OV: P = 0.031). Tail blood pressure was higher in EX (P = 0.039) and Aob+EX (P = 0.016) than Control (Table 1). There were no significant differences in heart rate between groups.

Muscle performance
The FI was impaired compared with Control immediately after band application (Acute-Aob (P < 0.001)) and remained impaired after 4 weeks (P = 0.039; Fig. 3, Table 2). The FI was comparable to Control in Aob+EX (P = 0.999) and Aob+OV (P = 0.999) and was significantly higher than Aob (P = 0.039) While overload improved FI in OV (P = 0.036), FI in EX was similar to Control (P = 0.954).

Femoral artery perfusion
Resting femoral artery flow (ml . min −1 ) was significantly lower in Acute-Aob compared with all other groups (vs. Control: P = 0.003; vs. OV: P = 0.036; vs. all others: P < 0.001), Fig. 4A). End-stimulation FBF was   . Impaired muscle fatigue resistance accompanied compensatory cardiac hypertrophy but recovered after intervention Fatigue index (FI) was reduced immediately after band application and remained impaired for 4 weeks without intervention (Aob: P = 0.039). Exercise (Aob+EX) and overload (Aob+OV) recovered fatigue resistance to control levels (both P = 0.999) despite the chronic effects of aortic banding. Group sizes were the same as in Fig. 1. Statistically significant differences between groups are denoted by letters; groups sharing the same letter were not different, those sharing none were statistically different (P < 0.05) as determined by ANOVA with Tukey's post hoc tests. lower in the Acute-Aob than EX (P = 0.022), Aob+EX (P = 0.018) and Aob+OV (P = 0.037) groups, but not significantly different from Control (P = 0.255), OV (P = 0.081) and Aob (P = 0.177) ( Fig. 4B; Table 2). Hyperaemic scope was similar in all groups. The magnitude of hyperaemic flow per gramme of EDL did not differ significantly between groups, even when accounting for femoral vascular conductance per gramme of muscle tissue (FVC) ( Table 2).

Muscle mass and histology
EDL mass as a proportion of body mass was increased in OV compared with Control (P = 0.036), Aob (P = 0.023) and Aob+EX (P = 0.038), but was similar to EX (P = 0.544) and Aob+OV (P = 0.938) ( Table 3). The magnitude of muscle hypertrophy after overload was, however, greater in OV compared with Aob+OV (P = 0.027; Table 3). Anatomical CD did not differ significantly between Control and any aortic banded animals (Aob: P = 0.999; Aob+EX: P = 0.938; Aob+OV: P = 0.465). OV had an increased anatomical CD compared with Control (P = 0.029), thereby indicating angiogenesis had occurred that exceeded the degree of fibre hypertrophy. Development of CCH precipitated a significant reduction in perfused CD in Aob (P < 0.001), Aob+EX (P = 0.001) and Aob+OV (P < 0.001) (compared with non-banded controls). Perfused CD was not significantly different between banded groups (Table 3).
While anatomical C:F was unchanged across groups compared with Control, it was increased in Aob+OV compared with Aob (P = 0.028) and Aob+EX (P = 0.002). Compared with unbanded groups, perfused C:F was reduced in Aob (P = 0.009). In contrast, perfused C:F was recovered in Aob+EX (P = 0.065) and Aob+OV (P = 0.329) and was not significantly different (P < 0.05) from Control (Table 3).
Compared with Control, anatomical CDA was increased in Aob (P = 0.009) while similar in Aob+EX (P = 0.845) and Aob+OV (P = 0.999) ( Table 3). In contrast, a more pronounced effect was obvious in the

. Resting (A) and end-stimulation (B) mean femoral artery blood flow
A pronounced reduction in femoral blood flow occurred immediately after aortic stenosis (Acute-Aob) but was restored in chronically banded animals. Group sizes were the same as in Fig. 1. Statistically significant differences between groups are denoted by letters; groups sharing the same letter were not different, those sharing none were statistically different (P < 0.05) as determined by ANOVA with Tukey's post hoc tests. perfused CDA where a significant increase was found in Aob compared with all other groups (P < 0.001). Intervention in Aob+EX (P = 0.285) and Aob+OV (P = 0.103) reduced this value to levels statistically non-significant from Control (Fig. 5, Table 3). Neither anatomical Log RSD nor perfused Log RSD were significantly different from Control in any group, and no differences were found amongst Aob, Aob+EX or Aob+OV.
Fibre-type composition did not differ significantly from Control with any intervention. Types I, IIa and IIb/IIx FCSA were, however, larger in Aob+OV when compared with Control (Type I: P = 0.013; Type IIa: P < 0.001; Type IIb/IIx: P = 0.001) and Aob+EX (Type I: P = 0.035; Type IIa: P = 0.013; Type IIb/IIx: P = 0.046), while IIb/IIx fibres were larger than Control in EX (P = 0.020), Aob (P = 0.024) and Aob+OV (P = 0.001) ( Table 3). Overload did not change FCSA in OV relative to Control and did not increase in Aob+OV relative to either OV or Aob.

PO 2 modelling
Increased development of potential hypoxic regions within the muscle was observed in Aob, with a pronounced effect at moderate and high-intensity tissue oxygen demand (Fig. 6). Compared with Aob, mean tissue PO 2 was improved in Aob+EX and Aob+OV (Fig. 6B), resulting in a qualitative PO 2 distribution similar to that of Control. Within control groups, there was a modest enhancement of tissue PO 2 in EX and OV compared with Con (Fig. 6A).

Discussion
In this paper we have demonstrated that CCH coincides with rarefaction of functional capillaries and a reduction in muscle fatigue resistance. Chronic mechanical stretch via overload (analogous to resistance exercise) and voluntary aerobic exercise recovered or prevented loss of perfused capillaries in muscle tissue, thereby contributing to the maintenance of fatigue resistance. These data indicate that the skeletal muscle microcirculation is a major therapeutic target to restore muscle function in patients with chronic diseases such as CHF.

Effects of acute aortic banding
There was a striking effect of banding and overload or exercise on EDL fatigue resistance, highlighting the sensitivity of skeletal muscle to both local and systemic influences. Aortic banding precipitated an immediate decline in muscle performance that remained reduced in the long term. In the acute situation, reduced muscle fatigue resistance may be due to an attenuated arterial blood supply (corresponding to the low tail blood pressure and reduced exercise-induced FBF), consistent with impaired ischaemic muscle performance in conditions of constrained feed artery perfusion (Fulgenzi et al. 1998;Murthy et al. 2001).

Effects of chronic aortic banding
The left ventricular pressure overload induced by abdominal aortic coarctation resulted, as expected, in CCH as a consequence of chronic hypertension above the stenosis (Cornelussen et al. 1994;Levy et al. 1996;De Sousa et al. 2002;Degens et al. 2006a). Reversal of the observed acute reduction in tail pressure was realized by an increased hypertension indicating that homeostatic mechanisms to maintain peripheral blood pressure were effective during CCH. Yet, muscle fatigue resistance was still reduced 4 weeks after banding despite a normal exercise-induced increase in FBF. We infer that limitations of muscle performance after long-term aortic banding were due to changes within the muscle itself rather than reduced arterial blood supply.
An earlier study using abdominal aortic coarctation-induced CCH reported a muscle-specific effect of CCH on FI, with functional impairments occurring in the soleus and tibialis anterior, but not in the EDL (Levy et al. 1996). Although we have no explanation for the discrepancy between EDL performance in this paper and Levy et al. (1996), it could be related to methodological differences in the way in which fatigue resistance was determined; 10 Hz stimulation used here elicits a slower decline in force that likely accommodates resolution of fine changes in force, which may be overlooked during the rapid onset of fatigue during 40 Hz stimulation (Levy et al. 1996).
We recently demonstrated that perfused capillary density is a major determinant of fatigue resistance in healthy skeletal muscle (Tickle et al. 2020), emphasizing that microvascular impairment can contribute to attenuated muscle performance. Capillary rarefaction is associated with impaired exercise tolerance in clinical (Gerovasili et al. 2009) heart disease (Duscha et al. 1999;Nusz et al. 2003) and given the absence of changes in arterial flow in our CCH rodent model, reduction in the perfused muscle microcirculation was a potential constraint on muscle function. Indeed, functional indices of EDL capillarization were negatively affected after development of CCH. Reduced functional (perfused) C:F and CD in Aob also parallels the impaired skeletal muscle blood flow reported in advanced CHF (Wilson et al. 1984;Sullivan et al. 1989;Musch & Terrell, 1992;Kindig et al. 1999). Furthermore, we infer that a larger CDA in Aob compromised fatigue resistance due to diffusion limitations of oxygen (see Fig. 6). The pronounced decline in modelled tissue PO 2 in Aob indicates that rarefaction of perfused vessels imposes a limitation on oxygen diffusion that is exacerbated by exercise of increasing intensity, leading to more rapid muscle fatigue than seen in healthy tissue. That tissue PO 2 is improved after

Figure 5. Deleterious enlargement of capillary domain area is restored by angiogenic intervention
Perfused capillary domain area (CDA) is increased after banding (Aob: N = 7; P < 0.01) but treatment via exercise (Aob+EX: N = 7) or overload (Aob+OV: N = 7) recovers CDA back to control level (P > 0.05). Control (N = 7), EX (N = 5) and OV (N = 5). Statistically significant differences between groups are denoted by letters; groups sharing the same letter were not different, those sharing none were statistically different (P < 0.05) as determined by ANOVA with Tukey's post hoc tests. J Physiol 599.15 intervention following aortic constriction highlights the potential efficacy of the functional microcirculation for improving fatigue resistance. Interestingly, the PO 2 model outputs broadly correspond to experimental evidence from exercising muscle tissue, whereby a large gradient occurs between capillary and myoglobin-associated PO 2 (Richardson et al. 1995;Poole et al. 2020).
Despite chronic groups undergoing at least 4 weeks of abdominal aortic constriction, leading to functional microvascular rarefaction, the total (anatomical) C:F Figure 6. Simulation of muscle PO 2 at rest, moderate and high rates of oxygen consumption A, based on perfused capillary localization, areas of muscle hypoxia (deep blue shading, where PO 2 < 0.5 mmHg) increase after chronic application of an aortic band (Aob: D, J, P) and this pattern corresponds to functional capillary rarefaction and impaired muscle fatigue resistance. Provision of exercise (Aob+EX: E, K, Q) or overload (Aob+OV: F, L, R) improves muscle PO 2 (warmer colours) and provides functional restoration of capillary domain area and muscle fatigue resistance. There is consequently strong evidence for a directional change in PO 2 although simulated absolute PO 2 values should be interpreted with caution due to the inherent limitations of modelling. B, average tissue PO 2 during resting (red), moderate (green) and high (blue) intensity tissue O 2 demand. Data were calculated using the exemplar images in (A), N = 1 in each group. [Colour figure can be viewed at wileyonlinelibrary.com] and CD were undiminished. This contrasts with other reports of reduced anatomical CD in CHF (Nusz et al. 2003), hypertension (Serne et al. 2001) and hindlimb ischaemia (Dai et al. 2002). While the determinant(s) of this observed reduction in perfused capillaries cannot be specified from the histological data presented in this paper, it appears consistent with an initial (reversible) phase of enhanced vasoconstriction-derived arteriolar closure that, given more time, is followed by structural capillary rarefaction because prolonged hypertension leads to capillary apoptosis (Prewitt et al. 1982); therefore our data are likely representative of the incipient phase of microvascular deficit.
Adverse changes in functional capillarity and fatigue resistance in Aob manifested during otherwise normal FBF. Maintenance of femoral flow despite chronic hypertension has been reported in other studies of abdominally coarcted rats with impaired peripheral circulatory structure or function (Overbeck, 1980;Boegehold et al. 1991;Levy et al. 1996). For example, Boegehold et al. (1991) reported peripheral arteriolar rarefaction in the hindlimb of rats subjected to abdominal aortic constriction, despite normal femoral arterial pressure. Similarly, peripheral vascular resistance and impaired hindlimb vasodilation develop after surgically imposed abdominal aortic stenosis, despite no increase in local femoral blood pressure (Bell & Overbeck, 1979;Overbeck, 1980;Ungvari et al. 2004). Together, these findings suggest that the mechanism(s) determining functional rarefaction is pressure-independent and therefore derived from systemic factors related to the development of cardiac dysfunction. The established anti-angiogenic effects of reactive oxygen species (Ungvari et al. 2004) and inflammatory cytokines (Agnoletti et al. 1999;Sun et al. 2007), which are both increased during development of cardiac dysfunction, may determine the progression of microvascular rarefaction. This may be further compounded by disuse, as observed in the shorter distanced travelled by Aob+EX, since inactivity leads to microvascular supply deficiency .
Despite impairments in the functional microcirculation, Log RSD was maintained in Aob, suggesting that, at least in this early phase of CCH and hypertension, heterogeneity of functional and anatomical capillary spacing is unaffected, preventing further declines in oxygen transport and muscle performance (Degens et al. 2006b;Al-Shammari et al. 2014). Shifts from slow to fast in skeletal muscle fibre type and muscle atrophy may develop with CHF and also contribute to exercise intolerance (Mancini et al. 1992;De Sousa et al. 2002;Carvalho et al. 2003), but these changes did not occur in banded groups. We can therefore exclude these factors as underlying the reduced fatigue resistance in our aortic-banded rats. Changes in fibre type composition are reported to occur 12 weeks after imposition of ascending aortic stenosis and fibre atrophy by 24 weeks (Carvalho et al. 2003), so the relatively short 4-to 6-week duration of these experiments may not have been sufficient to observe development of such impairments.

Exercise-mediated preservation of fatigue resistance after aortic banding
Exercise induced an unexpectedly high peripheral blood pressure, contrary to an expected exercise-induced hypotensive effect (Pescatello et al. 1991). Compared with Control, there was also an adaptive cardiac enlargement with exercise (EX) that did not significantly differ in magnitude from the reactive hypertrophy in Aob+EX and Aob+OV, indicating the potential contributions of adaptive (Moreira-Goncalves et al. 2015) and pathological (Grossman et al. 1975) stimuli to cardiac growth.
Wheel-running exercise did not significantly enhance FI in EX animals, but maintained muscle performance in Aob+EX despite development of CCH. This suggests that the level of exercise completed by animals was not sufficient to register an improvement in the exercise tolerance of the EDL muscle in EX, but was of sufficient intensity to offset the pathological effects of CCH on skeletal muscle in Aob. While inclination to exercise was undiminished in Aob+EX, the duration and intensity of each running bout was lower than in EX, leading to a lower overall distance travelled, mirroring previous work in banded rats (De Sousa et al. 2002). Interestingly, mean running velocity was reduced relative to EX in Aob+EX indicating that onset of volitional exercise fatigue contributed to declining (expected) locomotor muscle power output, despite no differences in maximal isometric twitch tension, i.e. muscle force generation was not impaired per se, but other factors compromised functional output. Nevertheless, the lesser amount of exercise was sufficient to restore EDL function to normal levels, thereby highlighting the beneficial effect of limited aerobic activity to reverse pathological changes that occur in skeletal muscle with CCH. Abnormalities in mitochondrial function and muscle metabolism are also reversed following voluntary exercise in clinical CHF (Adamopoulos et al. 1993;Stratton et al. 1994;Hambrecht et al. 1995) and an aortic coarctation model (De Sousa et al. 2002;Gomes et al. 2016), so together these data indicate that aerobic exercise provides a potent restorative effect on skeletal muscle structure and function in various stages of disease.
Voluntary wheel-running did not confer obvious benefit on structural or functional muscle microvasculature in unbanded rats, and in line with this, muscle FI in EX was not significantly different from Control. It is recognized that the pro-angiogenic effects of exercise may not be activated below a critical threshold J Physiol 599.15 (Prior et al. 2003), so it is perhaps not surprising that low-intensity wheel-running did not stimulate significant capillary proliferation in EX. In contrast, aerobic exercise after banding (Aob+EX) preserved functional CDA and perfused C:F, thereby maintaining tissue oxygenation and normal muscle FI (Al-Shammari et al. 2014). Therefore, low-intensity aerobic exercise may stimulate the reopening of hypertension-derived vasoconstricted arterioles to allow for utilization of the underlying functional microcirculation. The vasodilatory effects of an increased production of nitric oxide (Koller et al. 1994) and prostaglandins (Murrant et al. 2014) during exercise may determine this potential for increased capillary availability. Interestingly, although log RSD was maintained, perfused CD remained below unbanded levels in Aob+EX indicating that microvascular remodelling after exercise may be targeted to restore capillarity to meet local functional requirements. Establishment of local hypoxic regions within muscle upon microvascular rarefaction with CCH ( Fig. 6) may also contribute to a potentiated angiogenic response because hypoxic and exercise stimuli occur simultaneously, activating multiple proliferation pathways (Adair, 2005). Many of the reported benefits of exercise therapy in CHF (Belardinelli et al. 1999) may therefore be in part due to microvascular remodelling.

Recovery of fatigue resistance by overload after aortic banding
OV enhanced FI, as found in previous experimental studies (Frischknecht & Vrbova, 1991;Degens et al. 1993;Tickle et al. 2020). This benefit extended to animals with CCH, where an FI improvement of ca 32% was recorded in Aob+OV compared with a 25% enhancement in OV. In this case, therefore, a targeted mechanical stimulus without (obvious) systemic reciprocal effects leads to a rapid restoration of muscle performance. This parallels the enhanced muscle performance in CHF patients after resistance exercise therapy (Selig et al. 2004;Braith et al. 2005;Braith & Beck, 2008;Giuliano et al. 2017).
Improved functional capillarity with overload was accompanied by enhanced fatigue resistance, as reported previously (Degens et al. 1992;Zhou et al. 1998;Deveci & Egginton, 2002;Egginton et al. 2011;Ballak et al. 2016;Tickle et al. 2020), and parallels the upregulation of angiogenic growth factor expression and capillarity observed after resistance exercise in healthy subjects (Gavin et al. 2007;Ferguson et al. 2018;Holloway et al. 2018) and CHF patients (Williams et al. 2007). Alleviation of symptoms associated with heart disease may be hampered by coincident pathologies such as impaired peripheral blood flow, reducing the angiogenic effects of hyperaemic stimuli for expansion of the microcirculation. Interestingly, angiogenesis following overload may occur in conditions of experimentally reduced blood flow Deveci & Egginton, 2002), highlighting the potential of flow-independent mechanical stimuli to improve muscle capillarity and fatigue resistance, even in advanced disease. Therefore, angiogenesis may contribute to improved capillary perfusion by increasing the anatomical supply of microvessels fed by patent arterioles, or elevated angiogenic growth factor levels may attenuate capillary apoptosis. This may occur in the sedentary overload model despite a relative lack of shear stress-derived nitric oxide and prostaglandin upregulation, which may otherwise contribute to a reopening of constricted arterioles with aerobic exercise intervention. The successful application of muscle overload to rats with CCH provides further evidence to support therapeutic resistance exercise, which has the benefit of imparting additional cardiovascular strain for short bouts in contrast to the long durations required with aerobic exercise (Duncker et al. 2014) while providing an effective method for improving muscle function (Giuliano et al. 2017).
While EDL hypertrophy occurred in OV and Aob+OV, relative mass gain was lower after development of CCH although there were no significant differences in maximal twitch force or FCSA. Impaired muscle hypertrophy with induced CCH is similar to that reported in a rat model of metabolic syndrome, where limited overload-induced muscle growth was attributed to dysfunctional mTOR signalling, a central regulator of growth and proliferation Paturi et al. 2010). Raised levels of inflammatory cytokines, particularly TNFα, suppress mTOR activity in heart failure (Seiler et al. 2016) and may in part explain reduced gain in muscle mass in Aob+OV (Schiaffino et al. 2013).
Greater capillarization is associated with enhanced skeletal muscle fibre hypertrophy after resistance exercise training (Snijders et al. 2017;Moro et al. 2019), highlighting the critical importance of the microcirculation to muscle growth. Reduced functional capillarization with CCH may therefore contribute to the blunted hypertrophic response. Given the potential for regrowth of the functional microcirculation in Aob+OV, a longer period of overload may accommodate a full restoration of capillary parameters.

Conclusion
We have demonstrated that functional microvascular rarefaction with CCH underlies the impaired skeletal muscle fatigue resistance in CCH. Loss of functional capillarization can, however, be prevented through aerobic exercise and restored by functional overload-derived angiogenic stimuli, contributing to the successful recovery of muscle performance. Therefore, improved fatigue resistance in Aob+EX and Aob+OV as a consequence of improved functional capillarity supports the hypothesis that skeletal muscle performance in CHF is primarily determined by peripheral factors, specifically the muscle microcirculation. These interpretations bear a caveat: we have determined the effects on skeletal muscle function of incipient cardiac dysfunction rather than advanced disease, when co-morbidities add complexity to the mechanisms of muscle dysfunction. Nevertheless, the key determining role of muscle microcirculation in the development of exercise impairments during cardiac dysfunction underscore how angiogenic restorative growth should be a key therapeutic target.