Local capillary supply in muscle is not determined by local oxidative capacity

Highly oxidative muscles have a denser capillary network than those with a high glycolytic capacity, and within a given muscle, such as rat plantaris, fibres in the oxidative region have a higher capillary density than those in the more glycolytic region (Wüst et al., 2009). This correlation between anatomical capillary supply and tissue oxidative capacity also seems to apply at a smaller scale of biological control, where the local capillary supply to an individual fibre appears to be positively related to its oxidative capacity (Bekedam et al., 2003). Importantly, in such studies, fibre size was not considered and hence the influence of local diffusion distances on cellular oxygenation cannot be assessed. We have previously demonstrated that the local capillary supply to a fibre correlates with its cross-sectional area and is only slightly modulated by its oxidative capacity, but not by phenotype (Egginton and Gaffney, 2010; Wüst et al., 2009). However, if the principle of symmorphosis, which states that structures are matched to functional demand, is valid, then local capillarisation in a muscle should be arranged so that maximal oxygen demand per capillary is tightly regulated. To explore whether local feedback results in each capillary serving a similar maximal demand for oxygen, we estimated the supply areas (domains) of individual capillaries (Al-Shammari et al., 2014). Capillary domains provide a good estimate of the tissue oxygenation capacity of a capillary, even in muscles containing a mixture of fibres with different metabolic demand (Al-Shammari et al., 2014), whereas the total volume of mitochondria, as reflected by succinate dehydrogenase activity, in a domain is a reflection of the maximal oxygen demand served by that capillary. We hypothesised that if the primary determinant of local capillary supply was local oxygen demand, then the maximal oxygen demand (M_O2,max) supported by each capillary should be similar for each capillary.

Muscle biopsies were aseptically taken with a Rongeur forceps (Zepf Medizintechnik, Germany) under local anaesthesia (2 ml of 1% lidocaine) from the vastus lateralis (VL) and soleus of young men (N=10; 23–43 years old), following local ethical approval by the independent ethics committee Ärztekammer Nordrhein, Düsseldorf, Germany (No. 2010426) and written informed consent. The biopsies were taken as part of a bed rest study (ClinicalTrials.gov registration number NCT01655979) using baseline samples only. To facilitate longitudinal orientation, samples were embedded in a silicone tube filled with Optimal Cutting Temperature compound (Scigen^® Gardena), frozen in liquid nitrogen and stored at −80°C until analysis. All participants underwent an extensive health screening.

Serial frozen sections (8 µm) were co-stained with biotinylated lectin (Ulex europaeus agglutinin I, Vector Laboratories, UK; 1 h, 50 µl ml⁻¹ in 1% BSA HEPES) and anti-mouse myosin type I (1:100; Novocastra, Leica Biosystems, UK; product code: NCL-MHCs) to reveal capillary locations and Type I fibres, respectively. Sections were subsequently incubated with a secondary goat anti-mouse horseradish peroxidase labelled antibody (30 min, 1:200; Dako, UK) and stained (Vector VIP HRP substrate kit), as described by the manufacturer. The sections were mounted in glycerol-gelatine and stored at 4°C. Serial sections were stained for succinate dehydrogenase (SDH) activity as described previously (Wüst et al., 2009). Briefly, sections were incubated in the dark (20 min, 37°C in 37 mmol l⁻¹ phosphate buffer, 74 mmol l⁻¹ sodium acetate and 0.4 mmol l⁻¹ tetranitroblue tetrazolium, pH 7.6) (Fig. 1A,B). The optical density at 660 nm (OD₆₆₀) was determined (ImageJ; National Institutes of Health, Bethesda, USA) as an index of the aerobic capacity/mitochondrial content of muscle fibres because OD₆₆₀ is linearly related to fibre M_O2,max (Wüst et al., 2009). For each image, a separate calibration curve was constructed with a series of filters with a known OD₆₆₀ to remove potential optical bias related to differences in background intensity and lighting between sections.

Step-wise regression was performed to assess the impact of fibre type, size and mass-specific M_O2,max on LCFR and DAF. The correlations between M_O2,max and domain area were determined by Spearman correlation coefficients because Shapiro–Wilk tests indicated that the data were not normally distributed.

We confirm that the LCFR correlated positively with fibre cross-sectional area (FCSA) in both human VL (Fig. 2A; R=0.576 and R=0.625 when also including mass-specific M_O2,max in the model; both P<0.001) and soleus muscle (Fig. 2B; R=0.578 and R=0.591, respectively; P<0.001). LCFR per fibre perimeter, a measure of the capillary–fibre contact area, was positively related to the mass-specific M_O2,max in both VL (Fig. 2C; R=0.329; P<0.001) and soleus (Fig. 2D; R=0.138; P=0.002) muscles. Step-wise linear regression revealed that the number of DAF – a functionally more realistic index than ‘capillaries around a fibre’ – was primarily determined by FCSA; correlations improved when mass-specific M_O2,max was also included in the model (VL: R=0.470 versus 0.508; both P<0.001; soleus: R=0.497 versus 0.510; both P<0.001). Only in the soleus did inclusion of fibre type improve the correlation further (R=0.521; P=0.043). Intriguingly, in both VL (Fig. 2E) and soleus (Fig. 2F), the M_O2,max per capillary varied from almost 0 to more than 1000 pl mm⁻¹ min⁻¹. Also, M_O2,max was positively correlated with domain area in VL (Fig. 2E) and soleus (Fig. 2F). Thus, capillaries with larger oxygen supply areas supply a larger volume of mitochondria, and hence support a potentially larger maximum oxygen flux. These observations may require a fundamental review of ideas about determinants of local muscle capillary supply.

Even at the whole muscle level, a poor correlation with gross capillarisation was found in several species across a 17-fold range in muscle oxidative capacity (Maxwell et al., 1980). Capillary growth can occur without an increase in oxidative capacity, e.g. following selective stimulation of fast fibres (Egginton and Hudlická, 2000). While there may still be temporal coupling, because not all capillaries are perfused at any given moment and perfused vessels may have different flows adapted to the local demand for oxygen, it is unlikely that such perturbations during exercise can account for the more than 1000 pl mm⁻¹ min⁻¹ range in local M_O2,max among capillaries, as even at rest, all capillaries will have been perfused in as little as 20 s (Hargreaves et al., 1990). Another possibility is that the positive relationship between local M_O2,max and capillary domain size reflects a compensation for a reduced oxygen diffusion gradient due to the decreasing microvascular P_O2 from the arteriolar to venular end of capillaries (Egginton and Gaffney, 2010). While these factors may help to match temporal oxygen supply and demand, they do not explain why local structural capillary supply correlates poorly with local maximal oxygen demand, a violation of the symmorphosis principle that assumes structures are matched to functional demand. While the heterogeneity of capillary spacing does have an impact on tissue oxygenation (Egginton and Gaffney, 2010), it is apparently not regulated to maintain M_O2,max per capillary. The capillary–fibre exchange area, however, may be a better reflection of the capacity for oxygen flux, as suggested by the greater correlation between peak oxygen consumption with the capillary–fibre contact area than with capillary density (Hepple et al., 1997). Lack of significant correlations between fibre oxidative capacity and the local capillary density in both rat and human muscle (Wüst et al., 2009), but significant correlations between LCFR or DAF per fibre perimeter and fibre oxidative capacity, support this suggestion. Finally, the high volume of mitochondria in larger domains may serve to enhance flux of oxygen by exerting an extraction pressure, hence increasing total respiration rate, even if individual mitochondria are working submaximally under conditions of reduced oxygen tension. It is striking, however, that the local capillary supply was more tightly related to FCSA than fibre oxidative capacity. Consistent with these observations, fibre hypertrophy and angiogenesis during muscle overload have a similar time course (Plyley et al., 1998), further supporting a coupling between fibre size and local muscle capillarisation. Such a coupling is, in part, explicable by the fact that not only muscle fibres and satellite cells, but also endothelial cells act as mechanotransducers and may secrete factors with reciprocal effects in response to mechanical deformation (Christov et al., 2007).

Whatever the cause of this novel observation, the data indicate that: (1) maximal oxygen demand supported by individual capillaries varies enormously (more than 1000 pl mm⁻¹ min⁻¹) and (2) muscle fibre size rather than local maximal oxygen demand is a prime determinant of local muscle capillarisation.

We thank all the participants for providing muscle biopsies.