Mitochondria are unusually abundant in sensory terminals of mammalian muscle spindles; e.g., Landon (1966). We present the first quantitative morphological studies on these mitochondria, aiming to better understand their role in mechanosensory function. 3 adult C57/Bl6SJ mice were killed by Schedule 1 methods (Animals (Scientific Procedures) Act 1986 incorporating European Directive 2010/63/EU). Muscle-spindle-rich portions of deep masseter were removed in physiological saline, fixed (cacodylate buffer, 2.5 % glutaraldehyde/4 % paraformaldehyde; 4°C) for 24 hrs, postfixed (3 % potassium ferrocyanide and 4 % OsO₄), liganded with thiocarbohydrazide and 2 % OsO₄, block stained (1 % uranyl acetate and lead aspartate), and embedded in hard Epon. Sections (90 nm thick, transmission EM) were used to estimate the proportion of sensory-terminal volume occupied by mitochondria (V_V) and the areas of mitochondrial membranes per unit volume of mitochondria (S_V). Digital images (15000x magnification) were overlain with a 0.5mm sampling grid (ImageJ), for stereological analysis (Howard and Reed, 1998). Additional blocks were examined with serial block-face scanning EM (SBF-SEM) for 3-D reconstruction (Reconstruct; Fiala, 2005).

Mean V_V (% ± s.e.) for sensory terminals from the 3 mice were: 53.03 ± 4.38; 63.90 ± 7.37; and 47.88 ± 5.66; giving an overall average of 54.94 ± 3.49%, when estimated by stereology. To cross-validate this estimate, virtual reconstruction of 575 mitochondria in a sensory-terminal loop within a 15mm-long segment of a bag₂ fibre was made using SBF-SEM, revealing the mitochondria to be unbranched, more or less ovoid, with mean volume 0.358mm³ ± 0.015 s.e. (range, 0.015-3.57), highly skewed towards the smallest values (median = 0.260). Total mitochondrial volume was 209.41mm³ and terminal-loop volume was 376.79mm³, giving V_V of 55.58%, remarkably close to overall average V_V estimated by stereology. Mean surface area of the mitochondria was 2.14mm² ± 0.06 s.e., also skewed towards the smallest values (median = 1.80). Surface area: volume ratio was much less skewed, with mean 7.80mm^-1 ± 0.11 s.e. and median = 7.04. Overall values of S_V for the sensory-terminal mitochondria were: cristae, 13.67mm^-1 ± 0.58 s.e.; outer membrane, 4.47mm^-1 ± 0.22 s.e.; and inner membrane, 4.17mm^-1 ± 0.23 s.e. These values correspond to absolute areas per mean mitochondrial volume of: 4.90mm²; 1.60mm²; and 1.49mm², respectively.

These data show that volume proportion of mitochondria in the sensory terminals is extraordinarily high, at about 55%. By comparison a 21mm-long sarcomeric segment of a chain fibre, also reconstructed using SBF-SEM, contained 13.1% mitochondria by volume. These mitochondria were elongate, often branched, with mean volume 0.444mm³ ± 0.086 s.e. and mean surface area 4.557mm² ± 0.763 s.e. Using a similar reconstruction technique, Bleck et al., (2018) reported that small (approximately 1000mm³) volumes of glycolytic and oxidative extrafusal muscle fibres contained about 4 and 10% mitochondria by volume, and even the highly energetic cardiac muscle contained just about 34% in mouse. The functional importance of this superabundance in sensory terminals has yet to be determined, but it may underlie the observation that sensory ataxia is a commonly listed symptom in many mitochondrial genetic diseases (Ghaoui and Sue, 2018).