Heterogeneity in form and function of the rat extensor digitorum longus motor unit

Abstract The motor unit comprises a variable number of muscle fibres that connect through myelinated nerve fibres to a motoneuron (MN), the central drivers of activity. At the simplest level of organisation there exist phenotypically distinct MNs that activate corresponding muscle fibre types, but within an individual motor pool there typically exists a mixed population of fast and slow firing MNs, innervating groups of Type II and Type I fibres, respectively. Characterising the heterogeneity across multiple levels of motor unit organisation is critical to understanding changes that occur in response to physiological and pathological perturbations. Through a comprehensive assessment of muscle histology and ex vivo function, mathematical modelling and neuronal tracing, we demonstrate regional heterogeneities at the level of the MN, muscle fibre type composition and oxygen delivery kinetics of the rat extensor digitorum longus (EDL) muscle. Specifically, the EDL contains two phenotypically distinct regions: a relatively oxidative medial and a more glycolytic lateral compartment. Smaller muscle fibres in the medial compartment, in combination with a greater local capillary density, preserve tissue O2 partial pressure (PO2) during modelled activity. Conversely, capillary supply to the lateral compartment is calculated to be insufficient to defend active muscle PO2 but is likely optimised to facilitate metabolite removal. Simulation of in vivo muscle length change and phasic activation suggest that both compartments are able to generate similar net power. However, retrograde tracing demonstrates (counter to previous observations) that a negative relationship between soma size and C‐bouton density exists. Finally, we confirm a lack of specificity of SK3 expression to slow MNs. Together, these data provide a reference for heterogeneities across the rat EDL motor unit and re‐emphasise the importance of sampling technique.


| INTRODUC TI ON
Physiology is inherently variable, e.g. anatomical heterogeneities of endothelial cells across specific vascular beds (Vanlandewijck et al., 2018), sites of reactive oxygen species production within skeletal muscle (Staunton et al., 2019) and distribution of calcium exchange channels within cardiomyocytes (Jayasinghe et al., 2009).
Understanding the origin of such heterogeneities within an experimental system is essential to be able to accurately assess the integrated response to physiological or pathological perturbations.
The EDL is active during the swing phase of walking and thought to be involved in dorsiflexion of the ankle (Nicolopoulos-Stournaras & Iles, 1984). This fusiform muscle comprises a single proximal tendon, with the distal end subdividing into four separate muscle-tendon compartments, with tendons that insert onto the phalanges of digits II-V. These individual compartments represent different proportions of muscle mass, with heads II-V connected to 23%, 17%; 15% and 45% of the EDL muscle mass, respectively (Huijing et al., 1998).
Interestingly, this highly compartmentalised mixed hindlimb muscle is often treated as a relatively homogeneous entity. However, a brief report (Egginton, 1990) on phenotypic variation in the rat EDL described regional heterogeneity where, depending on the sampling indices adopted, different conclusions may be drawn about muscle composition.
Little progress has been made to better characterise EDL structural and functional composition in the last three decades. We confirmed regional heterogeneity in fibre type composition of the EDL (Deveci et al., 2001;Kissane & Egginton, 2019), linked with regionalspecific adaptations to systemic hypoxia (Deveci et al., 2001), and demonstrated distinct optimal frequency to generate peak power during cyclical contractions within EDL compartments associated with different responses to fatigue . However, there lacks a comprehensive description of the local capillary supply across this supposedly heterogeneous muscle. Additionally, there remains a paucity of information regarding distribution of the central drivers of activity (i.e. the motoneuron; MN) that determine the phenotypic arrangement of muscle. Understanding fine scale heterogeneities in structure and function among the whole motor unit is needed to meaningfully quantify maladaptive changes associated with pathologies like ageing and motor neuron disease, where Type IIb/x units are especially susceptible to denervation and dysfunction (Hepple & Rice, 2016;Kanning et al., 2010).
Here we quantified the structural composition of the most medial (head II) and lateral (head V) compartments of the rat EDL using histological assessment of muscle phenotype and local vascular supply, with functional consequence inferred from mathematical modelling of oxygen consumption. Using the muscle work loop technique to replicate in vivo cyclical muscle length change trajectories and phasic activation, we provide an assessment of in situ capacity to generate power. Finally, compartmentally injected retrograde tracers in combination with histological assessment of C-bouton complex SK3 (Deardorff et al., 2013(Deardorff et al., , 2014Smith & Brownstone, 2020) was used to characterise phenotypically distinct MN populations.

| Animals
Male Wistar rats (6-8 weeks old) were used in this study, housed under a 12:12 light-dark cycle in a temperature-controlled 21°C environment, with ad libitum access to food and water. Six animals were used to characterise the isolated muscle mechanical properties (226 ± 11 g) with seven (261 ± 8 g) used for MN tracing and histological assessment of muscle fibre type and vascular composition.

| Isolated muscle experiments
Briefly, animals were anaesthetised with isoflurane (5% in 100% O 2; IsoFlo ® ; Zoetis UK Ltd) and maintained at 2% during the careful dissection of the hindlimb EDL muscle. The distal tendons that insert onto the phalanges of digits II-V (Figure S1a) were located, released by transecting the retinaculum ligament in the ankle, and the tibialis anterior muscle removed to expose the EDL ( Figure S1b).
The isolated EDL was placed immediately into chilled (4°C), oxygenated (95% O 2 , 5% CO 2 ) Krebs-Henseleit solution (in mmol L −1 : 117 NaCl, 4.7 KCl, 2.5 CaCl 2 , 1.2 MgSO 4 , 24.8 NaHCO 3 , 1.2 KH 2 PO 4 and 11.1 glucose) (Burton, 1975), and the animal culled by an approved schedule one technique. The whole EDL was pinned out on a Sylgard ® (761036; Sigma) lined Petri dish at approximately in vivo resting length ( Figure S1c). The most medial (21.1 ± 0.79 mg) and lateral (51.1 ± 9.5 mg) compartments of the muscle were dissected free under a dissection microscope following fascicles that insert onto each distal tendon towards the proximal tendon. These remained in oxygenated Krebs-Henseleit solution until experimentation ( Figure   S1d) while the two intermediate compartments were discarded. At random, muscle compartments were attached to a metal rod fixed to a polychlorotrifluoroethylene lid and placed into a glass chamber containing fresh oxygenated Krebs-Henseleit solution. The muscle was then connected via a light stainless steel rod to the arm of an ergometer (305B-LR; Aurora Scientific Inc.) mounted to a Digimatic height gauge (Mitutoyo UK, Ltd) to control muscle length, and allowed to recover for 30 min before the experiments.

| Isometric muscle mechanics
Muscle length was incrementally increased in 0.5 mm steps to determine the twitch force-length relationship and therefore identify the length at which maximum isometric twitch force was generated (L 0 ). The muscle was activated using a supramaximal stimulus with a 0.2 ms pulse width using parallel platinum electrodes. A 200 ms isometric tetanus (200 Hz stimulus frequency) was performed and used to calculate maximum isometric tetanic stress. This length was used in the subsequent work loop experiment.

| Mechanical power during cyclical contractions
Using the work loop technique (Josephson, 1985), muscles were subjected to cyclical length changes and phasic electrical stimulation to activate muscles such that force generation occurred predominantly during shortening, maximising the net work and power generated by the muscle. The functional capacity of individual muscle compartments to generate power during five sinusoidal length trajectory at 7 Hz was compared with mean length L 0 and a strain amplitude of ±5% fibre length, as previously reported . Net power was calculated from the average of the two greatest cycles across the five work loop cycles.

| MN tracing
Animal surgery was completed by a competent Home Office approved PIL holder, under aseptic conditions. Surgical anaesthesia was induced and maintained with isoflurane (5% and 2%, respectively, in 100% O 2 ; IsoFlo; Zoetis UK Ltd). Two retrograde fluorescent tracers were injected into the right EDL 5 days prior to muscle sampling. The medial injections are positioned more proximally in the EDL muscle compared with the lateral (see Figure S1) with the medial compartments deep tendon/aponeurosis used to direct the needle for tracers, compared with the lateral injections which are more distally placed. A volume of 1 μl of 1.5% 647 nm cholera toxin subunit b (CTB) Alexa Fluor™ Conjugate (Invitrogen) was injected into both medial and lateral compartments and 3 μl of 1.5% 555 nm Fast Blue (FB; Polyscience, Inc.) was injected only into the medial EDL compartment. The combinatorial tracer approach allows us to differentially identify MNs from the lateral compartment (CTB + .FB − ) and medial compartment (CTB + .FB + ). The skin was sutured using 5-0 Mersilk (Ethicon; Johnson & Johnson Medical Ltd). Animals received analgesic (0.015 mg/kg, Vetagesic ® ; Ceva) and antibiotic (2.5 mg/kg, Baytril ® ; Baye) subcutaneously for 2 days post-surgery.

| Tissue preparation
Animals were anaesthetised and the right EDL was dissected, the mid-portion was coated in optimum cutting temperature compound (OCT, Agar Scientific) then snap frozen in isopentane cooled in liquid nitrogen. All muscle tissue was stored at −80°C until cryosectioning. Next, animals were transcardially perfused with 0.1 M phosphate buffer and fixed with 4% paraformaldehyde (PFA). Spinal columns were removed immediately after perfusion and post-fixed in 4% PFA for 24 h. Spinal cords were dissected and cryoprotected in 30% sucrose at 4°C for 7 days. Next, lumbar segments were isolated, frozen in OCT (Agar Scientific) and stored at −20°C.

| Muscle histology
Extensor digitorum longus muscles were cryo-sectioned (−20°C, 12 µm), mounted on polylysine-coated slides (VWR International) and stored at −20°C until staining. Muscle fibre type composition was determined using methods previously described (Al-Shammari Indices for capillary-to-fibre ratio (C:F) and capillary density (CD) and fibre cross-sectional area (FCSA) were derived from histological sections. The global indices presented here are most common among basic and pathological studies describing the relationship between capillaries and muscle fibres; C:F may provide information relating to capillary growth/rarefaction, while CD may provide information on the diffusive capacity of the capillary bed. However, while these global indices describe gross changes in capillary supply they lack descriptive power for capillary distribution, which has a signifi- and capillary locations (lectin positive microvessels), the OTM generates a digital mask of the skeletal muscle cross section, which incorporated individual fibre types to provide fibre type-specific oxygen demand (Sullivan & Pittman, 1984;Wüst et al., 2009) allowing us to model local oxygen consumption and estimate tissue O 2 partial pressure (PO 2 ) distribution.

| Spinal cord immunohistochemistry
Spinal cord immunohistochemistry was performed as previously described (Kissane, Al-Shammari, et al., 2021;Smith et al., 2017). In brief, the L3-L6 segments were sectioned at 50 µm on a cryostat, and free-floating sections were collected and stored in phosphate-

| Confocal microscopy and quantitative analysis of the spinal cord
Images were acquired with a Zeiss LSM 800 confocal microscope Three-dimensional (3D) reconstructions of each MN were rendered from the confocal image Z-stacks, utilising Imaris Software.
In the 3D isometric view, solid surfaces of the MN soma with dendrites, C-boutons and SK3 were created via surface rendering and thresholding. CTB or FB was used to model the MN surface. A masking feature was then used to select SK3 clusters contacting the MN surface and/or proximal to the C-bouton. Imaris was then used to generate volume and surface area data for each MN, C-bouton and SK3 cluster. To determine the MN cross-sectional area, the soma perimeter was outlined using ImageJ at the central plane of the nucleus, as illustrated elsewhere .
Since all alpha-MNs contain C-boutons, cells with no C-bouton labelling were removed (Deardorff et al., 2014). To identify medial vs.
lateral MNs, a fluorescence signal intensity threshold was measured in IMARIS. Cells with a fast blue intensity greater than 2785 intensity units (AU) were classified as FB-positive, and therefore medial EDL MNs. All other CTB + , FB − MNs were classed as lateral MNs.
The threshold used for segregating MN was chosen by finding the greatest drop off in intensity using histograms created using the following equation: All data are expressed as mean ± standard error. Statistical test (Shapiro-Wilks) for normality was completed on all data, where either an independent t-test (normally distributed) or Mann-Whitney U tests (not normally distributed) were completed. Relationships between local capillary indices and fibre size, as well as synapse density and soma area were assessed using linear regressions. All statistical analysis was performed using IBM SPSS Statistics, v.25, with statistical significance set at p < 0.05.

| Phenotypically distinct compartments of the EDL
The EDL has a distinctly heterogeneous fibre type distribution, graded from the oxidative medial compartment laterally to the more glycolytic portion of the muscle (Figure 1a). Fibre size varies according to location (Figure 1b) with the medial compartment average fibre area 39% smaller than that of the lateral [1308 ± 65 vs. 1824 ± 145 µm 2 , respectively; t(10) = −3.251, p = 0.009]. In addition, there exists a significant phenotypic difference in fibre type composition between medial and lateral compartments (Figure 1c; Table 1). The lateral compartment comprises entirely of Type II fibres, with a numerical composition of 19.8 ± 2.5% Type IIa and 80.2 ± 2.5% Type IIb/x compared with the more oxidative medial compartment containing 9.4 ± 0.5% Type I, 32.3 ± 4.2% Type IIa and 58.4 ± 4.3% Type IIb/x. Interestingly, when looking at the areal composition of the muscle fibres, the lateral compartment consists of only 10.5 ± 1.7% Type IIa fibres, with the remaining 89.5 ± 1.7% Type IIb.

| Heterogeneity in capillary distribution and oxygen tension profile
Each muscle fibre type has a different level of oxygen demand; therefore, a heterogeneity in muscle fibre type distribution implies a gradient in oxygen demand across the muscle. This presumably requires a tailored microvascular supply. Indeed, the lack of difference in C:F in combination with a significantly elevated CD in the more oxidative medial compartment suggests that the microvascular supply is determined by local fibre morphology and fibre type distribution (Table 2). However, to resolve functional heterogeneities in oxygen supply at an individual fibre level we utilised scale-independent local indices in combination with mathematical modelling of resulting oxygen tension, using anatomically accurate capillary and fibre spatial distributions (Figure 2a). There was a significantly lower capillary  Note: Values presented are mean ± SEM (n = 6/group).
Abbreviations: CD, capillary density; C:F, capillary to fibre ratio; LCD, local capillary density; LCFR, local capillary to fibre ratio; SEM, standard error of the mean.
relationship with FCSA across the medial (R 2 = 0.304, p < 0.001) and lateral compartments (R 2 = 0.249, p < 0.001, Figure 2e). Fibre typespecific local indices of capillary supply are presented in Table 2. The combination of lower CDA and higher LCD (Table 2)

| Mechanical properties of the EDL compartments
Given inherent heterogeneities in oxidative demand (fibre type composition) and oxygen delivery (capillary supply), we sought to explore muscle functional capacity in an ex vivo environment.
The active force-length relationship for both compartments of the EDL followed similar profiles, increasing in force approaching L 0 , with a defined plateau maintained at +1 mm over L 0 (Figure 3b).

| Heterogeneities within a single motor pool
Individual compartments were labelled with retrograde tracers  Figure 4h); however, the pooled SK3 density was negatively correlated with MN surface area (R 2 = 0.061, p = 0.003, Figure 4i).

| DISCUSS ION
Through histological assessment of muscle fibre type and capillary supply, mathematical modelling of oxygen transport, assessment of muscle mechanics and retrograde labelling of MNs, our comprehensive overview of the rat EDL motor unit has identified heterogeneity in form and predicted function across multiple organisational levels of a single motor pool ( Figure 5). These data emphasise that the motor unit is a tightly regulated system, from the supply of oxygen to individual muscle fibres down to the distribution of ion channels across a MN, and cautions that inadequate sampling techniques (e.g.

F I G U R E 3 Mechanical properties of the extensor digitorum longus (EDL). (a) The compartmental arrangement of the EDL and sites of retrograde tracer injection. (b)
The force-length relationship for the medial (blue) and lateral (red) compartment, with both active (solid lines) and passive (dashed lines) force profiles. (c) Peak length for (d) maximum isometric tetanic stress. (e) An example work loop for the medial (blue) and lateral (red) compartment muscle preparations and (f) the net power generated. Mean ± standard error of the mean (SEM), *p < 0.05. L 0 , optimum muscle length to generate force; P/P 0 , force relative to the peak force during the work loop cycle insufficient random fields of view across histological sample, use of only scale-dependent indices and inconsistencies in retrograde tracer injections) could lead to false conclusions about motor unit remodelling and plasticity. Understanding these heterogeneities is critical to accurately quantify deficits across the motor pool in response to pathological perturbations, and to aid the development of effective treatment strategies.

| Implications of a heterogeneous capillary supply and oxygen demand
The EDL has long been described as a phenotypically fast muscle (Egginton, 1990), and has commonly been used in experimental re- (h) Probability density for SK3 density across the entire EDL with Gaussian plots for the whole EDL (black), medial cells (blue) and the lateral cells (red). Finally (i), the lack of any significant relationship between SK3 density and cell surface area. Scale bar = 150 µm. Mean ± standard error of the mean (SEM) muscles into a single phenotype. Having demonstrated a phenotypic gradient across this mixed muscle and shown regional adaptations to hypoxia (Deveci et al., 2001), we now have refined our approach to identify heterogeneities at the level of an individual capillary and muscle fibre. Our data suggest that within an individual muscle, the oxygen transport system may diverge from assumed symmorphosis (Weibel et al., 1991) where, for example, capillary supply is adequate to meet metabolic demand, and would prioritise maintenance of a relatively constant tissue PO 2 across the entire muscle. This, however, does not appear to be true of the EDL. With a substantially more dense and tailored vascular supply, the medial compartment appears to prioritise oxygen supply appropriate for tissue oxidative demand, whereas the distinct arrangement of capillaries in the lateral compartment is likely optimised for removal of metabolites (e.g. lactate), a characteristic previously reported between distinctly oxidative and glycolytic muscles (Hudlicka et al., 1987). Future work selectively activating individual compartments through stimulation of individual nerve branches (Balice-Gordon & Thompson, 1988) in combination with blood gas analysis (Hudlicka et al., 1987) and plasma metabolomics (Hauton et al., 2015) would begin to unpick the role of differential metabolite removal.

| Mechanical considerations of heterogenous muscles
Understanding the complexity of muscle fibre heterogeneity on the form and function of a whole muscle is an emerging research area. Recent work has shown fibre length (Bolsterlee et al., 2018;Charles et al., 2019;Kissane et al., 2018) and sarcomere length (Moo et al., 2016) to be heterogeneously distributed throughout the skeletal muscle, and that sarcomeres respond differently across regions in response to whole muscle perturbations (Moo et al., 2016(Moo et al., , 2020. Additionally, we have shown that the individual compartments of EDL present distinct fatigue responses during sustained work, with the medial compartment preserving the capacity to maintain force, with impaired relaxation kinetics, whereas the lateral compartment maintains relaxation kinetics with a compromised ability to generate sustained force . In contrast, the present data demonstrate that despite considerable phenotypic and morphometric differences between the two compartments, they are able to generate comparable force and net power, under the specified experimental conditions. Unfortunately, to date, there are no published compartmental recordings of in vivo activity in the EDL, making it difficult to compare task-specific functional capacity.

| The inherent heterogeneity within an individual motor pool
Few experiments report compartmentally traced MNs (Ishihara et al., 1995), partly due to accessibility challenges at the extreme ends of a muscle, with some researchers opting to crudely divide MN populations based on numerical fibre type composition, e.g. splitting the diaphragm motor pool into tertiles (Rana et al., 2019(Rana et al., , 2020. This type of pseudo differentiation of a MN population is challenging, despite 33% of the diaphragm fibres being Type IIb/x fibres (Warren et al., 2020), faster MNs typically innervate more extrafusal fibres than those of slow MNs (Kanning et al., 2010), and therefore exists a great potential to incorrectly classify MNs through the use of equal tertile boundaries.
While histological assessment of phenotypically distinct MNs is problematic we were able to identify MNs from two muscle compartments within the same motor pool. The large proportion of Type IIb/x fibres within both medial and lateral compartments are served by MNs of similar soma area, unlike the difference between compartments seen in the tibialis anterior muscle (Ishihara et al., 1995). However, a trend for smaller MNs in the medial compartment F I G U R E 5 Overview of the extensor digitorum longus (EDL) motor unit heterogeneities. Labelled motoneurons across the EDL motor pool highlight the negative relationship between soma area and VAChT Density and SK3 Density. These motoneurons innervate phenotypically distinct portions of the EDL muscle that is synchronously recruited to provide power for movement. At the level of individual fibres and capillaries heterogeneities in oxygen supply and demand are evident, which when scaled up combine to meet the demands of the muscle, principally maintenance of oxidative phosphorylation in the medial and removal of lactate in the lateral compartment is consistent with the smaller FCSA there. Interestingly, we found a negative relationship between MN size and C-bouton density, in contrast to that described by Kanning et al. (2010). Therefore, the previous opinion that fast MNs contained higher C-bouton densities (Hellström et al., 2003) does not hold true for the EDL.
Finally, in light of recent single-cell transcriptomics of cholinergic neurons, it is perhaps surprising that we were unable to selectively identify slow MNs using SK3 (KCNN3 gene) , given its supposed enrichment (Blum et al., 2021) and previously reported expression profile (Deardorff et al., 2013(Deardorff et al., , 2014(Deardorff et al., , 2021. While gene expression does not directly translate to protein expression, the discrepancy between the number of positive cells in the current analysis implies that such specificity does not exist. The lack of differential labelling of SK3 within the EDL compartments shows that this small conductance calcium-activated potassium channel is an inappropriate marker of slow MNs, and reiterates the importance of understanding the heterogeneity of form and function, down to the level of the MN ion channel organisation (Deardorff et al., 2021).
In conclusion, our data demonstrate a significant difference in fibre type composition across the rat EDL, accompanied by a tailored vascular supply to meet metabolic demands. Though quantifiable structural heterogeneities exist, isolated individual compartments were able to produce equivalent power under our experimental conditions. Finally, across the whole motor pool C-bouton synaptic input correlate negatively with soma size. Together, these data provide a source of reference for the heterogeneities within the rat EDL motor unit and, given the inherent heterogeneities in form and function, emphasise the importance of sampling technique across multiple levels of the motor unit.

ACK N OWLED G EM ENTS
The authors thank the help from Calvin Smith, Arash Ghaffari-Rafi and Robert Brownstone for their previous collaborative efforts and support of this additional work.

CO N FLI C T O F I NTE R E S T
All authors confirm that there is no conflict of interest.