Reorganization of mitochondria–organelle interactions during postnatal development in skeletal muscle

Skeletal muscle cellular development requires the integrated assembly of mitochondria and other organelles adjacent to the sarcomere in support of muscle contractile performance. However, it remains unclear how interactions among organelles and with the sarcomere relates to the development of muscle cell function. Here, we combine 3D volume electron microscopy, proteomic analyses, and live cell functional imaging to investigate the postnatal reorganization of mitochondria–organelle interactions in skeletal muscle. We show that while mitochondrial networks are disorganized and loosely associated with the contractile apparatus at birth, contact sites among mitochondria, lipid droplets and the sarcoplasmic reticulum are highly abundant in neonatal muscles. The maturation process is characterized by a transition to highly organized mitochondrial networks wrapped tightly around the muscle sarcomere but also to less frequent interactions with both lipid droplets and the sarcoplasmic reticulum. Concomitantly, expression of proteins involved in mitochondria–organelle membrane contact sites decreases during postnatal development in tandem with a decrease in abundance of proteins associated with sarcomere assembly despite an overall increase in contractile protein abundance. Functionally, parallel measures of mitochondrial membrane potential, NADH redox status, and NADH flux within intact cells revealed that mitochondria in adult skeletal muscle fibres maintain a more activated electron transport chain compared with neonatal muscle mitochondria. These data demonstrate a developmental redesign reflecting a shift from muscle cell assembly and frequent inter‐organelle communication toward a muscle fibre with mitochondrial structure, interactions, composition and function specialized to support contractile function.


Introduction
Skeletal muscle cellular development necessitates the coordination of physical interactions among mitochondria and other organelles interspersed within the contractile apparatus to meet the functional requirements of muscle contraction (Franzini-Armstrong, 1991).While alterations in mitochondrial content and composition (Glancy, 2020;Kaasik et al., 2001;Seppet et al., 2001).Consequently, spatial proximity among mitochondria and other organelles influences the frequency and efficiency of these interactions, thereby regulating functional capacity (Glancy et al., 2020).A comprehensive understanding of how organelle interactions change in response to varying functional demands can offer valuable insights into the role of intracellular organization in muscle cell function.
Several types of mitochondria-organelle interactions play a vital role in cellular energy distribution and metabolism in skeletal muscle (Eisner et al., 2013;Gemmink et al., 2018;Parry & Glancy, 2022).For example, mitochondria in striated muscle cells can form a reticulum, or network, of many adjacent mitochondria (Bakeeva et al., 1978;Bubenzer, 1966;Glancy et al., 2015;Kirkwood et al., 1986;Ogata & Yamasaki, 1985;Picard et al., 2013).These mitochondrial networks provide a structural link between mitochondria interspersed between the contractile apparatus and mitochondria located in large pools at the periphery of the cell (Bakeeva et al., 1978;Bubenzer, 1966;Glancy et al., 2015;Kirkwood et al., 1986) and are also thought to facilitate rapid cellular energy distribution throughout the relatively large skeletal muscle cell (Glancy et al., 2015;Skulachev, 1969Skulachev, , 1990)).Mitochondria can also form direct contact sites with their lipid droplet fuel source (Bosma, 2016;Gemmink et al., 2018;Zhang et al., 2011), and these lipid droplet-connected mitochondria are reported to be larger, longer, and have increased interactions with adjacent mitochondria than non-lipid droplet-connected mitochondria within the same cell (Bleck et al., 2018).Additionally, dynamic interactions between mitochondria and the sarcoplasmic reticulum (SR) enable tight regulation of mitochondrial and cytosolic calcium levels, which are essential for energy homeostasis, cell viability and other cellular processes (Boncompagni et al., 2009;Boncompagni et al., 2020;Eisner et al., 2013).Mitochondria are also closely associated with the energy-demanding contractile machinery occupying most of the cellular volume in skeletal muscle (Andrienko et al., 2003;Katti et al., 2022).Although organelle interactions in skeletal muscle cell culture, adult tissues and pathological conditions have been investigated (Bleck et al., 2018;Boncompagni et al., 2009;Boncompagni et al., 2020;Eisner et al., 2013), the role of inter-organelle connectivity during muscle cell development, a critical period for communication and coordination, remains largely unexplored.
Here, we characterize the physical reorganization of the cellular energy distribution system in skeletal muscle during postnatal development, focusing on mitochondria-organelle interactions.By leveraging the high-resolution and large field of view provided by focused ion beam scanning electron microscopy (FIB-SEM) (Glancy et al., 2015;Heymann et al., 2006) combined with machine-learning image segmentation (Sommer et al., 2011), we provide a high-throughput analysis of 3D mitochondria-organelle interactions within developing skeletal muscle.Our findings reveal that while tortuous mitochondria are loosely interspersed within the contractile machinery at birth (Bakeeva et al., 1981;Mishra et al., 2015), neonatal muscles display frequent interactions among mitochondria, lipid droplets and the SR.As muscles mature into oxidative or glycolytic fibre types, mitochondria undergo a structural transition, becoming more linear and tightly associated with the contractile apparatus (Bakeeva et al., 1981;Mishra et al., 2015).However, muscle maturation also involves less frequent interactions among mitochondria, lipid droplets and the SR, with glycolytic muscle exhibiting a near-complete loss of lipid droplets.Additionally, by performing new proteomic analyses, we report reduced expression levels of proteins implicated in mitochondria-organelle membrane contact sites during postnatal development, along with a decrease in proteins related to sarcomere and cellular assembly, despite a large increase in overall contractile protein abundance.Finally, functional analyses of intact muscle cell mitochondrial membrane potential, NADH redox status and NADH flux indicate that adult skeletal muscle mitochondria maintain a more active electron transport chain than neonatal muscle mitochondria.These results highlight a functional redesign of the skeletal muscle cell during postnatal development, transitioning from frequent organelle interactions in coordination with cellular assembly at birth to a cellular energy distribution system specifically tailored to facilitate contractile function in the mature skeletal muscle cell.

Ethical approval
All animal procedures were approved by the National Heart, Lung, and Blood Institute Animal Care and Use Committee and performed in accordance with the guidelines described in the Animal Care and Welfare Act (7 USC 2142 § 13).

Mice
C57BL6/N mice (∼6--8 weeks old) were obtained from Taconic Biosciences (Rensselaer, NY) and were used as breeding pairs.Their offspring was randomly assigned as postnatal (P) 1, P14 and P42 group (n = 3-4 per group).All pups were weaned at P21-P28, and both breeders and weaned animals had access to food and water ad libitum.The vivarium was maintained on a 12:12 h light and dark cycle at 20-26°C.Due to a technical limitation J Physiol 602.5 in using anogenital distance to ascertain sex in P1 pups, we randomly used male and female animals rather than separating experimental groups by sex.

Muscle sample preparation
As in a previous study (Bleck et al., 2018), both soleus (oxidative) and gastrocnemius (glycolytic) muscle fibres were carefully excised and fixed for FIB-SEM imaging acquisition.Briefly, skin-peeled hindlimbs were submerged in a fixative solution (2% glutaraldehyde in 0.1 m phosphate buffer, pH 7.2) for 30 min, while mice were under anaesthetization with 2% isoflurane by nosecone.Upon completion of the in vivo fixation protocol, the fixed muscles were excised and the animal killed by exsanguination.After 1 h incubation in standard fixative solution (2.5% glutaraldehyde, 1% formaldehyde, 0.12 m sodium cacodylate, pH 7.2-7.4), the excised tissues were subsequently post-fixed and en bloc stained (i.e. 2% aqueous osmium tetroxide incubation) following the published protocol with minor modifications (Bleck et al., 2018).After an overnight incubation in 1% uranyl acetate solution at 4°C, the samples were incubated at 60°C for 20 min with Walton's lead aspartate (0.02 m lead nitrate, 0.03 m aspartic acid, pH 5.5) and were thoroughly washed with distilled H 2 O at room temperature.Afterwards, the samples were gradually dehydrated with ethanol and were then incubated in 25%, 50%, 75% and 100% Embed 812 resin solutions for ∼36 h.Then, the tissue samples were placed on Zeiss SEM specimen mounts (Electron Microscopy Sciences, #75225, USA) and were polymerized at 60°C for 2-3 days.After the polymerization, the samples were cut and polished by Leica UCT Ultramicrotome (Leica Microsystems Inc., USA) that carried Trimtool 90 diamond knives (DiATOME, Switzerland).

FIB-SEM imaging
The Zeiss crossbeam 540 (Gemini II) was used to collect FIB-SEM images at 10 nm voxel size with Zeiss Atlas 5 software (Carl Zeiss Microscopy GmbH, Jena, Germany).The FIB milling (10 nm thickness) was conducted at 30 keV, while maintaining a beam current at 2-2.5 nA.All collected micrographs were aligned with a proprietary algorithm and then exported as 8 bit TIFF files for further imaging analysis.

Segmentation of cellular structures
All image processing was conducted on a desktop PC (New Tech Solutions, Fremont, CA) equipped with an Intel Xeon W-2155 (3.3 GHz processor, 10 cores/20 threads) and 256 GB RAM.As done previously (Bleck et al., 2018), Ilastik pixel classification software (Ilastik.org)was used for semi-automated image classification and segmentation of 3D structures of mitochondria and other subcellular organelles including lipid droplets, SR and t-tubules.Also as done previously (Bleck et al., 2018), individual mitochondria were segmented using the Multicut module in Ilastik.All data were exported as a 32 bit HDF file for imaging analysis with ImageJ (National Institutes of Health, Bethesda, MD, ImageJ.net).

Imaging analysis
After loading with the HDF5 plugin, all HDF image datasets were processed to investigate both mitochondrial network configurations and individual mitochondrial structures.Following the established analytical pipeline (Bleck et al., 2018), several ImageJ plugins and analytical tools were used to examine mitochondrial networks (OrientationJ Distributions plugin), individual mitochondrial structures (ROIManager3D plugin), intermitochondrial junctions (Image Calculator tool) and mitochondrial spatial interactions with other subcellular components (Image Calculator tool and 3D Geometrical Measure tool).All 3D mitochondrial and subcellular components were extracted and visualized using 3D viewer and volume viewer on ImageJ or in Imaris.
Mitochondrial nanotunnel abundance was assessed by determining the amount of mitochondrial volume with a diameter of 150 nm or less.This threshold value was chosen to ensure that only the thin, nanotunnel protrusions were selected since using a 200 nm value often selected entire mitochondria, particularly in P42 glycolytic cells.Eight bit binary mitochondrial segmentation files were opened in ImageJ and the Local Thickness plugin (Dougherty & Kunzelmann, 2007) was used to determine the local diameters of mitochondria.A 3D median filter of 100 nm × 100 nm × 100 nm was then applied and nanotunnels were selected as all pixels with a filtered local thickness of 150 nm or less using the manual threshold tool.The number of pixels within nanotunnels was divided by the total number of mitochondrial pixels and multiplied by 100 to determine the percentage nanotunnel volume.

Flexor digitorum brevis fibre isolation and imaging
Flexor digitorum brevis (FDB) muscles of mice were dissected and immediately placed into Tyrodes buffer (pH 7.4, 10 mm HEPES, 137 mm sodium chloride, 4.5 mm potassium chloride, 0.5 mm magnesium sulfate, 0.5 mm potassium phosphate, 10 mm glucose and 1.8 mm calcium chloride) and 3 mg/ml collagenase solution.The muscles were agitated in a water bath at 37°C for 20-75 min, depending on the age of the mouse (for younger age, less incubation time).The Tyrodes and collagenase solution was then removed and the digested muscle was resuspended in Tyrodes.Individual muscle fibres were released by gentle trituration (Park et al., 2014).Once cells were isolated, the solution was set aside to allow cells to settle to the bottom of the tube.The remaining Tyrodes solution was siphoned off and the cells were resuspended in either 2 μm Rhod-2 AM (Thermo Fisher, Waltham, MA) or 5 nm tetramethyl rhodamine, methyl ester (TMRM, Invitrogen, Waltham, MA) and 200 nm MitoTracker Green dye (MTG, Thermo Fisher, Waltham, MA).Rhod-2 stained cells were incubated at 37°C for 1 h with gentle agitation.The Rhod-2 dye was removed and then washed twice with 200 nm MitoTracker Green were performed prior to imaging.TMRM and MitoTracker Green solution incubated at 37°C for 10 min prior to imaging.
Muscle cells were imaged with a Leica upright SP8 microscope with a Nikon 25× (1.1 NA) water-immersion objective.Confocal sequential line scanning and internal HyD detectors were used during image capture.TMRM or Rhod-2 and MitoTracker Green were imaged with 488 nm and 552 nm excitation and 500-550 nm and 590-650 nm emission wavelengths, respectively.
Image analysis was performed on the whole fibre Z-stack using Fiji ImageJ.Thresholding of the MitoTracker Green signal created a mask of mitochondrial content.The mask was then applied to the TMRM or Rhod-2 images, therefore normalizing the variable measurements to the pixels contained within the mask.Using the Image Calculator in ImageJ, the mask was divided by itself to obtain 1.0 for mitochondrial pixels and NaN for non-mitochondrial pixels.The resultant image was then multiplied by each TMRM, Rhod-2 and MTG image to obtain masked images.The mean pixel values were then used to determine pixel intensity.TMRM and Rhod-2 values were then normalized to MTG pixel intensity to account for mitochondrial content.

MitoRACE
Mitochondrial flux capacity was measured using the MitoRACE method (Willingham et al., 2019).Briefly, FDB fibres were isolated as described above.The isolated fibres were then plated on Cell-Tak coated dishes.Optimal adhesion was achieved by allowing the cells to settle for upwards of 60 min before performing further experiments.
Imaging cells was performed using a Leica upright SP8 microscope.Multiphoton laser-scanning fluorescence was used to collect the autofluorescence of NADH.A microperfusion system (MultiChannel Systems, Reutlingen, Germany) was used to flush Tyrodes buffer through the coated cells at 7.8 ml/min.After 30 s of baseline imaging, 5 mm sodium cyanide was added to solution to inhibit oxidative phosphorylation.The immediate rise in NADH autofluorescence was then captured.NADH fluorescence was excited at 750 nm and emission was detected using a Leica HyD detector at 414-538 nm.Rapid imaging of 32 μm in depth was also employed to obtain measurements throughout the cell depth.Since the complete oxidation of NADH results in zero signal in our system (Willingham et al., 2019), the baseline redox status was calculated based on the baseline and fully reduced NADH intensities.

Fluorescence lifetime imaging (FLIM)
NADH fluorescence lifetime decay was measured using isolated FDB muscle fibres as described above.Cells were then imaged using a two-photon Leica (Germany) FALCON FLIM SP8 upright microscope.NADH autofluorescence was excited at 750 nm, and the emission was set at 420-440 nm.FLIM images were taken at a speed of 400 Hz, 50 frame repetitions per line.
All FLIM images were converted to a Picoquant BIN file format to integrate with Fiji ImageJ.The files were loaded into Fiji as 5D image stacks using the Bioformats library.Intensity images were produced from the fluorescence lifetime data by integrating the 5D stack across its time dimension.After manual segmentation to remove unwanted artefacts, a manual threshold was applied to the fibres to create a mask, which was then overlaid on top of the 5D image stack, then integrated spatially to produce a single decay curve for each fibre.
After fibres and solutions were converted into fluorescence decay curves, each curve was fitted to an artefact-controlled multi-exponential model convolved with the instrument response function from the Leica system.To determine the appropriate number of lifetimes for this model, components were added or removed to approach unity between the reduced sum of squared residuals.
Once appropriate models for the fibres were determined, curves from each dataset were simultaneously fitted to their respective models via global lifetime analysis (Knutson et al., 1983;Smirnov et al., 2009).Briefly, to establish a baseline test statistic, all curves from each dataset were first fitted to their model predetermined in the prior single curve analysis assuming unlinked lifetime parameters.To determine whether curves shared lifetime species, lifetime terms in the model were permuted to be linked or unlinked across all curves in the dataset, then were fitted keeping all other model options the same.To test whether the linkage assumptions were false, each linked permutation reduced sum of squared residuals was divided by the baseline sum of squared residuals and was mapped to the appropriate J Physiol 602.5 F-statistic for the number of degrees of freedom across the dataset.It was determined that all curves within each dataset shared molecular species, meaning that any subsequent calculations of relative species contribution would yield higher accuracy compared with results from a non-global lifetime analysis.
Using the global fit lifetime models, lifetime values of 4.14 ns, 1.36 ns and 0.39 ns were resolved.Lifetime values around 0.5 ns are known for free NADH, whereas values between 1.4 and 4 ns have been reported for bound NADH (Datta et al., 2020).The number of molecules contributing to each of NADH components was determined from the amplitude values.The sum of amplitudes for all the components provided a measure of total NADH molecules within each muscle fibre.The average amplitudes were then determined for each developmental stage (P1, P14, P42 oxidative and P42 glycolytic).A relative measure of total NADH per fibre was determined by finding the ratio of amplitude of each development stage to P42 glycolytic fibres.This value was then used to determine NADH concentration oxidized per second during the NADH flux analysis using a mitochondrial NADH concentration of 2.45 mm previously measured by HPLC in adult mouse hindlimb muscle mitochondria (Willingham et al., 2019).Importantly, the mitochondrial NADPH concentration measured by HPLC in the same mitochondria was <100 μM indicating that mitochondrial NADH contributes >95% of the overall NAD(P)H signal.

Statistical analysis
Using Excel 2016 (Microsoft, Redmond, WA) and Prism 8 (GraphPad, San Diego, CA), we conducted all statistical analyses.Two-way ANOVA was used to assess mean values of each dataset within and between groups (Fibre Type (Glycolytic, Oxidative) × Development (P1, P14, P42)).Tukey's honestly significant difference post hoc tests were used for multiple comparisons of muscle type and developmental stage.Sidak's multiple comparisons test was used for lipid droplet connected/not connected and high/low SR/T connected mitochondrial analyses.Statistical significance level was set at P < 0.05.

Dynamic reorganization of mitochondrial networks during postnatal development
To evaluate mitochondria-organelle interactions during postnatal muscle development, we used FIB-SEM to collect mouse skeletal muscle cell volumes with 10 nm pixel sizes in 3D at birth (postnatal day 1 (P1), Supplementary Movie 1), during the dynamic phase of the transition between neonatal and mature mitochondrial networks (Kim et al., 2019) (P14, Supplementary Movie 2), and after mitochondrial network structures had reached maturity (Kim et al., 2019) (P42, Supplementary Movies 3-4).To account for the muscle fibre-type differences in mitochondrial intra-and inter-organelle interactions previously observed in mature muscles (Bleck et al., 2018), we imaged cells from both the soleus and gastrocnemius muscles representing more oxidative and glycolytic muscles, respectively (Burkholder et al., 1994), and further confirmed adult muscle cell type based on mitochondrial content (Bleck et al., 2018).Machine learning segmentation of the FIB-SEM muscle volumes (Beier et al., 2017;Bleck et al., 2018;Sommer et al., 2011) allowed for high-throughput analyses of mitochondrial, lipid droplet and sarcotubular (sarcoplasmic reticulum + t-tubules; SR/T) structures as well as interactions among them.
Mitochondrial structure within a cell is coordinated across different spatial scales ranging from cell-wide networks to the size and shape of individual organelles to interactions with adjacent organelles (Bleck et al., 2018;Glancy et al., 2020) (Fig. 1A-F).Beginning at the cellular scale, networks of tortuous mitochondria were primarily aligned parallel to but loosely associated with the contractile apparatus in both newborn muscle types (Fig. 1A, D, I, J Supplementary Movie 1).Additionally, overall mitochondrial volume and number were similar between muscle types at birth (Fig. 1G-H, 6.3 ± 1.2% and 7.0 ± 0.4% of total muscle volume, 345 ± 20 and 313 ± 17 mitochondria/1000 μm 3 muscle, means ± SE, n = 3 muscle volumes, 618 and 276 mitochondria for P1 soleus and gastrocnemius, respectively).These data suggest that mitochondrial network configuration in neonatal muscles is driven by developmental status rather than muscle type at this stage.During the postnatal transition phase (P14), overall mitochondrial content and number were little changed from birth and were again no different between muscle types (Fig. 1G-H, 10.8 ± 0.2% and 7.7 ± 1.9% of total muscle volume, 400 ± 26 and 443 ± 6 mitochondria/1000 μm 3 muscle, means ± SE, n = 3 muscle volumes, 1124 and 1848 mitochondria for P14 soleus and gastrocnemius, respectively).However, mitochondrial networks in both muscles began to associate more closely with the contractile apparatus at P14 (Fig. 1B, E, J Supplementary Movie 2).In the soleus, mitochondrial networks became more linear and elongated along the parallel axis and began to form short branches along the perpendicular axis at the ends of the sarcomeres near the z-disc (Fig. 1B, I).In contrast, while the mitochondrial networks in the gastrocnemius muscle also became more linear and elongated than at birth (Fig. 1E), there was a greater contribution of perpendicularly oriented mitochondrial branches compared with the soleus muscle (Fig. 1I).By P42, the -462 mitochondria, three datasets.Points are means for each dataset.Bars are means ± SE. * P < 0.05, vs. P1; # P < 0.05, vs. P14; % P < 0.05, vs. O (SOL); α P < 0.05, main effect of development; β P < 0.05, main effect of fibre type; χ P < 0.05, interaction effect of development and fibre type.scale bar = 1 μm.[Colour figure can be viewed at wileyonlinelibrary.com] divergence of mitochondrial network configurations between muscle types was completed (Fig. 1C, F, I) as mitochondrial volume and number were significantly higher in the oxidative compared with glycolytic muscles (Fig. 1G-H, 10.8 ± 2.2% and 3.6 ± 0.3% of total muscle volume, 420 ± 37 and 247 ± 25 mitochondria/1000 μm 3 muscle, means ± SE, n = 3 muscle volumes, 1147 and 475 mitochondria for P42 oxidative and glycolytic muscles, respectively), and the mitochondrial networks reached their mature grid-like (oxidative, Supplementary Movie 3) and primarily perpendicular (glycolytic, Supplementary Movie 4) orientations (Fig. 1C, F, I).These data suggest that the formation of cell type-specific mitochondrial network configurations occurs in concert with the fibre-type specificity of myosin isoform composition that occurs during postnatal development (Agbulut et al., 2003;Gokhin et al., 2008).
At the single organelle level, the volume of individual mitochondria followed a similar time course to the overall mitochondrial content in both muscle types, where content gradually increased across development in the oxidative muscles and fell in the glycolytic muscles (Fig. 1K).While there were no significant differences in mitochondrial surface area to volume (SA/V) ratio among individual groups (Fig. 1L), gastrocnemius muscle mitochondria as a whole had greater SA/V ratios than soleus muscle mitochondria, consistent with previous reports on glycolytic versus oxidative muscle mitochondria (Bleck et al., 2018;Kirkwood et al., 1986;Kirkwood et al., 1987).To further investigate how individual muscle mitochondrial morphology is altered across postnatal development, we quantified the relative number of small (∼80-120 nm) doughnut-like holes in mitochondria (Fig. 2A), which have been suggested as a marker of oxidative stress (Liu & Hajnoczky, 2011).In soleus muscles, the number of mitochondrial doughnuts was highest at birth and then dropped significantly at P14 and remained low in the mature muscle (Fig. 2B, 20.0 ± 2.4, 3.3 ± 0.6 and 1.5 ± 0.4 mito doughnuts per 100 μm 3 mito volume, n = 3 muscle volumes, 26, 10 and 5 doughnuts for P1, P14 and P42, respectively).In the gastrocnemius, there were no differences in the number of doughnuts compared with the soleus at birth.However, the relative number of mitochondrial doughnuts rose significantly to more than 30-fold higher than in the oxidative muscles at P42 (Fig. 2B, 20.6 ± 1.6, 13.5 ± 4.0 and 45.8 ± 3.1 mito doughnuts per 100 μm 3 mito volume, n = 3 muscle volumes, 17, 24 and 32 doughnuts for P1, P14 and P42, respectively), suggesting that the increased formation of doughnuts reported previously in glycolytic muscles (Bleck et al., 2018) may be reflected at the mitochondrial level beginning during the late postnatal phase of development.Mitochondrial nanotunnels are another specialized mitochondrial morphology, which is also thought to be a marker for oxidative stress (Huang et al., 2013;Lavorato et al., 2017;Vincent et al., 2017).Formation of thin, elongated nanotunnels (Fig. 2C) was relatively rare early in development and made up less than 0.5% of mitochondrial volume at P1 and P14 (Fig. 2D).However, nanotunnel formation increased significantly by P42 in the glycolytic muscles (Fig. 2D, 0.35 ± 0.10, 0.13 ± 0.08 and 0.51 ± 0.15 %nanotunnel volume per mito volume, n = 3 muscle volumes for P1, P14 and P42 oxidative muscles, respectively; 0.30 ± 0.05, 0.37 ± 0.08 and 1.53 ± 0.43% nanotunnel volume per mito volume, n = 3 muscle volumes for P1, P14 and P42 glycolytic muscles, respectively).Thus, mature glycolytic muscles are characterized by the most frequent abundance of both mitochondrial doughnuts and nanotunnels compared with oxidative and developing muscles.
To determine whether interactions among mitochondria were altered during postnatal muscle development, we assessed the size, abundance and orientation of the intermitochondrial junctions (IMJs) between adjacent mitochondria (Fig. 2E, F), which it has been suggested allow for rapid communication and distribution of molecules among physically coupled mitochondria (Glancy et al., 2017;Picard et al., 2015).The size of individual IMJs and relative abundance of IMJs per mitochondrion were both largely similar across muscle types and developmental timepoints with the exception of an increase in both size and abundance observed at P14 in the soleus compared with the mature oxidative muscle (Fig. 2G, H).Conversely, while there were no differences in IMJ orientation detected between muscle types, each muscle type demonstrated a loss of parallel and a gain of perpendicularly oriented IMJs upon maturation (Fig. 2I).These data suggest that the putative physical coupling sites permitting the transfer of signalling molecules, metabolites and/or ions directly between adjacent mitochondria are primarily altered by changing their orientation within the cell rather than size or abundance during postnatal development.

Mitochondria-lipid droplet contact sites decrease in frequency across postnatal development
To evaluate how mitochondrial interactions with other organelles were altered during postnatal muscle  development, we began by assessing the size and content of the lipid droplets (Fig. 3A-C), which provide a direct fuel source for mitochondrial oxidative phosphorylation (Kanaley et al., 2009;Meex et al., 2015).Overall muscle content and the size of individual lipid droplets were highest at birth in both muscle types, followed by a significant decrease during the late postnatal stage that continued into maturation (Fig. 3D, E).Contact sites between mitochondria and lipid droplets (i.e.membranes within 30 nm (Gatta & Levine, 2017)) followed a similar pattern with a nearly 10-fold loss in lipid droplet contact site abundance per mitochondrion across postnatal development in the oxidative muscles and a complete loss of contact sites in the mature glycolytic muscle where no lipid droplets were found (Fig. 3F).These data suggest that physical interactions between mitochondria and lipid droplets, which facilitate the direct transfer of molecules between them (Herker et al., 2021), may be directly related to the metabolic fuel preferences of the skeletal muscle cell, as neonatal muscles are known to rely more heavily on fatty acids than adult muscles (Glatz & Veerkamp, 1982), while glycolytic muscles rely more on carbohydrate fuel sources relative to oxidative muscles (Hargreaves & Spriet, 2020).We previously found that mitochondria in contact with lipid droplets provided a structural capacity for greater energy distribution compared with non-lipid connected mitochondria within the same adult muscle cells (Bleck et al., 2018).To determine whether this apparent structural specialization of mitochondria within a network was present beginning at birth, we assessed individual mitochondrial structural characteristics for lipid droplet connected and non-connected mitochondria separately.Individual mitochondrial volume, a proxy for the internal capacity of a mitochondrion (Glancy et al., 2020), was consistently greater in lipid connected mitochondria for all muscles with lipid droplets present (Fig. 3G).Conversely, mitochondrial SA/V ratio, an indicator of the relative capacity to interact with the surrounding environment (Glancy et al., 2020), was lower in lipid droplet connected mitochondria across all timepoints and in both muscle types (Fig. 3H).Additionally, the total IMJ area per mitochondrion, an indicator of molecular transfer capacity to mitochondria, trended higher in lipid droplet connected mitochondria (Fig. 3I).Together, these data indicate that the specialization of lipid droplet connected mitochondria has already begun at birth and is maintained throughout the maturation process.

Mitochondria-sarcoplasmic reticulum interactions are highly abundant during early postnatal development
To begin our evaluation of mitochondrial interactions with the SR/T throughout postnatal development, we first assessed the total muscle cell volume occupied by the SR/T.At birth, the SR/T formed an unorganized mesh wrapping around the myofibrils with no difference in total volume between muscle types (Fig. 4A, C, E), similar to previous work in mice (Franzini-Armstrong, 1991;Luff & Atwood, 1971).During maturation, the SR/T became more organized into the well-known longitudinal SR mesh and SR/T triads wrapping around the myofibrils (Fig. 4B,  D) with SR/T in the mature glycolytic muscle occupying a greater volume than in newborn muscles or mature oxidative muscle (Fig. 4E), also in line with previous reports (Bleck et al., 2018;Eisenberg, 2010;Luff & Atwood, 1971).These data demonstrate the cell-type specification of the SR/T as well as the consistency between our 3D analysis and previous 2D assessments of SR/T volume.
Next, we assessed the physical interactions between the SR/T and 5332 individual mitochondria during postnatal development.Nearly every mitochondrion across all conditions had at least one contact (membranes within 30 nm) with the SR/T (99.6 ± 0.5, 99.6 ± 0.2, 99.4 ± 0.2, 99.3 ± 0.3, 99.6 ± 0.2, 98.3±1.2% of mitochondria in contact with SR/T, n = 300, 570, 1847, 1124, 468, 1023 mitochondria in three datasets for P1 gastrocnemius, P1 soleus, P14 gastrocnemius, P14 soleus, P42 glycolytic and P42 oxidative, respectively).The triadic nature of the interactions between the t-tubules and the SR means that most of the t-tubule surface is covered by the SR (Franzini-Armstrong, 1991) and, as a result, the mitochondrial interactions with the SR/T volume detected here are almost exclusively interactions between mitochondria and the SR.At birth, roughly one-sixth of the mitochondrial surface area was in direct contact with the SR/T on average with more than 80% of mitochondria having at least 10% of its surface area in contact with the SR/T (Fig. 4F, G, 16.6 ± 2.2 and 17.3 ± 2.3% mitochondrial surface area contact with SR/T, 83.5 ± 7.9 and 84.3 ± 10.7% of mitochondria with >10% surface area contact with SR/T, n = 300 and 570 mitochondria in three datasets for P1 gastrocnemius and P1 soleus, respectively).By 2 weeks of age, interactions between mitochondria and the SR/T fell by more than twofold for mean surface area contact and more than fourfold for percentage of mitochondria with at least 10% surface area in contact and were maintained at this level into maturity for both muscle types (Fig. 4F, G, 7.0 ± 1.1, 6.1 ± 0.6, 8.1 ± 0.6 and 6.8 ± 0.4% mitochondrial surface area contact with SR/T, 18.0 ± 6.7, 15.2 ± 1.9, 18.5 ± 1.1 and 17.1 ± 1.8% of mitochondria with >10% surface area contact with SR/T, n = 1847, 1124, 468 and 1023 mitochondria in three datasets for P14 gastrocnemius, P14 soleus, P42 glycolytic and P42 oxidative, respectively).These data demonstrate the ubiquitous yet dynamic nature of mitochondrial interactions with the SR in muscle cells.
Due to the unique structural characteristics of lipid droplet connected mitochondria described above (Fig. 3), we hypothesized that mitochondria with high SR/T contact areas may have different morphology compared with lower SR/T-contacting mitochondria.To investigate potential differences among mitochondria with high and low SR/T contact, we compared the structures of individual mitochondria with more or less than 10% surface area contact with the SR/T (Fig. 4G-J).No  three datasets; P1 glycolytic -252 mitochondria, three datasets; P14 glycolytic -354 mitochondria, three datasets; P42 glycolytic -110 mitochondria, three datasets; N values (<10% SRC): P1 oxidative -72 mitochondria, three datasets; P14 oxidative -980 mitochondria, three datasets; P42 -849 mitochondria, three datasets; P1 glycolytic -44 mitochondria, three datasets; P14 glycolytic -1457 mitochondria, three datasets; P42 glycolytic -348 mitochondria, three datasets.Points are means for each dataset.Bars represent means ± SE. * P < 0.05, vs. P1; % P < 0.05, vs. O (SOL); # P < 0.05, vs. P14; χ P < 0.05, interaction effect of development and fibre type; α P < 0.05, main effect of development.Scale bar = 1 μm.[Colour figure can be viewed at wileyonlinelibrary.com] differences in mitochondrial volume were detected between high and low SR/T contact mitochondria (Fig. 4H).Mitochondrial SA/V ratio was lower in high SR/T contact mitochondria at P14 in both gastrocnemius and soleus muscles and in P42 oxidative muscles (Fig. 4I).Additionally, there were no significant differences in mitochondrion-to-mitochondrion interactions through IMJs between high and low SR/T contact groups (Fig. 4J).These data are consistent with the lack of structural differences between high and low SR/T contact mitochondria across postnatal cardiac development (Kim et al., 2022) and suggest that SR/T contact level may not be associated with mitochondrial specialization as it is for lipid droplet contacts.

Mitochondria-organelle interactions are associated with cellular assembly during postnatal development
To evaluate the integrated role of mitochondrial interactions during postnatal muscle development, we assessed how the total mitochondrial outer membrane surface area in close proximity (30 nm) to other structures changed across time points in both muscle types (Fig. 5).At birth, nearly one quarter of mitochondrial surface area was in direct contact with adjacent organelles and one tenth of mitochondrial surface area was associated with adjacent sarcomeres in both muscle types (Fig. 5A, B, I, 24.6 ± 2.9 and 23.7 ± 1.7% surface area organelle contact per mitochondrion, 10.2 ± 1.1 and 9.1 ± 3.0% surface area associated with sarcomere per mitochondrion, n = 3 datasets each for P1 soleus and P1 gastrocnemius, respectively).However, by P14, membrane contact site abundance had dropped by nearly half and remained at that level throughout development (Fig. 4C-I, 13.0 ± 0.7, 11.9 ± 1.3, 9.6 ± 1.1 and 11.9 ± 1.0% surface area contact per mitochondrion, n = 3 datasets each for P14 soleus, P14 gastrocnemius, P42 oxidative and P42 glycolytic, respectively).Conversely, mitochondrial outer membrane associations with the sarcomere rose throughout development reaching the highest levels in mature muscles (Fig. 4C-I , 26.8 ± 3.8, 22.4 ± 2.6, 35.0 ± 8.3 and 28.0 ± 5.4% surface area associated with sarcomere per mitochondrion, n = 3 datasets each for P14 soleus, P14 gastrocnemius, P42 oxidative and P42 glycolytic, respectively).These data demonstrate the developmental dynamics of mitochondrial interactions where more frequent contacts with other organelles occur early in development followed by a transition to more direct associations with the contractile apparatus in mature muscles.
We previously reported that the developmental transition in glycolytic muscle mitochondrial network configuration occurred together with a transition in mitochondrial protein expression where mitochondrial assembly factors were highly abundant at birth and fell during maturation despite oxidative phosphorylation and total mitochondrial protein abundances increasing across development (Kim et al., 2019).To test the hypothesis that the mitochondrial interactions described above may also be associated with the developmental transition from cellular assembly towards specialized support for muscle contraction, we performed new analyses on our previously published proteomic database of mouse tibialis anterior skeletal muscle across postnatal development (Kim et al., 2019).As would be expected based on the proliferation of the contractile apparatus during growth (Deshmukh et al., 2015;Goldspink, 1970), muscle contractile proteins steadily increased in total abundance during maturation (Fig. 6A).Conversely, the total abundance of sarcomere assembly factors and other cellular assembly factors and chaperones all consistently decreased across postnatal development (Fig. 6C-D).Concomitantly, the abundance of reported tether proteins linking mitochondria and the SR (Csordás et al., 2018) also largely decreased during postnatal muscle development (Fig. 6E).Consistently, putative mitochondria-ipid droplet tethers (Cui & Liu, 2020) and lipid droplet-SR tether proteins (Hugenroth & Bohnert, 2020) also decreased in overall abundance during maturation (Fig. 6F, G).It is important to note that these decreases in SR tether proteins (Fig. 6E, G) occur despite the developmental increase in both SR volume (Fig. 4E) and overall protein abundance (Kim et al., 2019).Thus, the protein expression levels of cellular assembly factors and organelle contact sites each decrease during development whereas the expression of contractile, SR and mitochondrial proteins all increase with maturation.

The electron transport chain is activated during postnatal muscle development
To investigate the functional implications of the developmental reorganization of mitochondria-organelle interactions, we performed live cell functional imaging experiments on intact, isolated FDB muscle fibres from Y. Kim and others J Physiol 602.5 P1, P14 and P42 mice (Fig. 7).By assessing NADH autofluorescence (Fig. 7A-D) before and after rapid infusion of cyanide, we were able to assess the basal NADH redox state and NADH flux rate (Willingham et al., 2019).The rate of NADH production was significantly higher in both P42 oxidative and glycolytic muscles compared with P1 and P14 (Fig. 7R, 1.0 ± 0.1, 0.5 ± 0.1, 3.0 ± 0.2, 2.2 ± 0.2 μm NADH reduced per second, n = 5, 11, 9, 7 cells for P1, P14, P42 oxidative and P42 glycolytic muscles, respectively).While the NADH driving force into the electron transport chain (i.e.redox status) was highest in P42 glycolytic fibres, there was no difference between P42 oxidative, P14 and P1 muscles (Fig. 7Q, 59.2 ± 3.6, 52.8 ± 2.6, 56.4 ± 2.9, 71.1 ± 1.7% reduced NADH, n = 5, 11, 9, 7 cells for P1, P14, P42 oxidative and P42 glycolytic muscles, respectively).Thus, P42 oxidative fibres had a greater flux with a similar input driving force.To test whether an altered back pressure on the electron transport chain may explain the increased flux in P42 muscles, we assessed the mitochondrial membrane potential using exogenously added TMRM fluorescent dye (Fig. 7E-H).Since the intensity of TMRM fluorescence   is a product of the magnitude of the membrane potential and the amount of mitochondria within the cell, we simultaneously assessed mitochondrial content using membrane potential-insensitive MitoTracker Green dye (Pendergrass et al., 2004) (Fig. 7I-L) and normalized the TMRM intensity to the MTG signal.The mitochondrial membrane potential was significantly higher in both P42 oxidative and glycolytic muscles compared with P14 and P1 muscles (Fig. 7S, 0.32 ± 0.01, 0.26 ± 0.02, 0.47 ± 0.03, 0.51 ± 0.04 TMRM/MTG arbitrary fluorescent units, n = 31, 37, 18, 18 cells for P1, P14, P42 oxidative and P42 glycolytic muscles, respectively).Thus, the increased flux observed in P42 muscles occurred despite a greater membrane potential back pressure on the electron transport chain.To test whether altered calcium levels, a known activator of mitochondrial energetic flux (Glancy & Balaban, 2012;McCormack et al., 1990), may explain the increased flux in P42 muscles, we assessed mitochondrial calcium using Rhod-2 fluorescence (Fig. 7M-P) also normalized to the MTG signal.Mitochondrial calcium levels were highest in P42 glycolytic muscles followed by P14, P42 oxidative and P1 muscles (Fig. 7T, 0.17 ± 0.01, 0.33 ± 0.04, 0.23 ± 0.04, 0.40 ± 0.07 Rhod-2/MTG arbitrary fluorescent units, n = 32, 12, 5, 5 cells for P1, P14, P42 oxidative and P42 glycolytic muscles, respectively).Thus, mitochondrial calcium may explain the increased driving forces and energetic flux in the P42 glycolytic muscles compared with P1 and P14 muscles.However, mitochondrial calcium levels do not appear to explain how P42 oxidative muscles achieve greater electron transport chain flux with a similar input driving force (NADH redox) and greater back pressure (membrane potential).Thus, this increased conductance (Glancy et al., 2013), or activation state, of the electron transport chain in P42 oxidative muscles appears to occur during development by some other means.Overall, these results show that the mitochondrial electron transport chain becomes more active as the muscle reaches maturity.

Discussion
By evaluating the 3D mitochondria-organelle interactions within skeletal muscle cells across postnatal development, we demonstrate the physical reorganization of the cellular energy distribution system that occurs as the muscle transitions away from cellular assembly and towards focused support for contractile function.Skeletal muscle in newborn mice was characterized by tortuous mitochondrial networks arranged in parallel to the contractile axis of the cell with frequent contact sites between mitochondria and lipid droplet or SR membranes.However, despite being placed directly between the myofibrils, the worm-like appearance of the neonatal mitochondria resulted in relatively less of the mitochondrial surface area being located directly adjacent to the contractile apparatus.Thus, the organization of the newborn muscle cell reflects an increased structural capacity for direct communication between organelles (Eisner et al., 2014), lipid metabolism and calcium signalling, the latter two of which have been suggested to play key roles in cellular assembly and development (Debattisti et al., 2019;Hahn, 1982;Stiber et al., 2008;Tu et al., 2016;Yao et al., 2020).Conversely, regardless of whether the number of mitochondria increased (oxidative) or decreased (glycolytic) or whether mitochondria converted to grid-like (oxidative) or perpendicularly oriented (glycolytic) networks, interactions between mitochondria and other organelles fell nearly twofold during postnatal development while the more linear mitochondrial segments in mature muscles formed tighter associations with the adjacent contractile structures.These data suggest that the mitochondrial outer membrane surface area in mature muscle is specialized to directly support the energetic requirements of muscle contractions independent of mitochondrial content or network configuration.Consistently, our live cell functional assessments of mitochondrial energy conversion indicate that the electron transport chain becomes more active in both oxidative and glycolytic mature muscles compared with developing muscles.Thus, mitochondrial network structure (Fig. 1), function (Fig. 7) and protein composition (Kim et al., 2019) all appear to become specialized for energy metabolism in support of muscle contraction during postnatal development.
Individual mitochondria also appear capable of specialization relative to their adjacent neighbours within a mitochondrial network (Bleck et al., 2018;Kim et al., 2022).We show here that mitochondria connected to lipid droplets throughout postnatal skeletal muscle development are larger and more connected to adjacent mitochondria compared with non-lipid droplet connected mitochondria which have greater SA/V ratios.Thus, muscle mitochondria connected to lipid droplets have a greater structural capacity for energy conversion (i.e.greater internal volume) and rapid molecular transfer through mitochondrial networks (i.e.greater IMJ area) and less capacity for interactions with other cellular structures (i.e.lower available surface area relative to volume) (Glancy et al., 2020).This 7 μm for O,P.Q, NADH redox state across postnatal development.N = 5, 11, 9, 7 cells for P1, P14, P42 oxidative and P42 glycolytic, respectively.R, NADH flux rates across postnatal development.N = 5, 11, 9, 7 cells for P1, P14, P42 oxidative and P42 glycolytic, respectively.S, mitochondrial membrane potential across postnatal development.N = 31, 37, 18, 18 cells for P1, P14, P42 oxidative and P42 glycolytic, respectively.T, mitochondrial calcium levels across postnatal development.N = 32, 12, 5, 5 cells for P1, P14, P42 oxidative and P42 glycolytic, respectively.
[Colour figure can be viewed at wileyonlinelibrary.com] morphological distinction of lipid droplet-connected mitochondria is consistent across postnatal development and has also been reported across postnatal heart development (Kim et al., 2022), suggesting that lipid droplet-connected mitochondria in all striated muscles may be functionally specialized to utilize the adjacent lipid fuel source and distribute the converted energy throughout the mitochondrial network to adjacent mitochondria which are then structurally more suited for molecular distribution to other cellular components.However, while the relationship formed by membrane contact sites between mitochondria and lipid droplets appears consistent across striated muscles, a different type of functional specialization occurs in brown adipose tissue where lipid droplet-connected mitochondria appear to be functionally specialized to support building lipid droplets rather than breaking them down for fuel (Benador et al., 2018;Veliova et al., 2020).Conversely, lipid droplet connected mitochondria in liver were recently reported to be specialized for fatty acid oxidation (Talari et al., 2023).Thus, it is clear that there are tissue-specific differences in the functional specialization of lipid droplet-connected mitochondria.As such, future work to gain a better understanding of the molecular nature of the tethers linking lipid droplets and mitochondria and how they may differ among tissue types may offer some insight into how certain tethers facilitate specific functions.
Contact sites between mitochondria and the endo/SR have been shown to play important roles in many cellular processes, including mitochondrial dynamics (Friedman et al., 2011;Phillips & Voeltz, 2016), calcium signalling (Giacomello & Pellegrini, 2016;Naon & Scorrano, 2014), lipid synthesis (Giacomello & Pellegrini, 2016), ubiquinone synthesis (Subramanian et al., 2019) and cell death (Prudent & McBride, 2017).However, while the specific functional role of interactions between mitochondria and SR in skeletal muscle is not fully understood (Boncompagni et al., 2020), the focus of much of the muscle work to date has revolved around the role of these organelle interactions in calcium signalling (Eisner et al., 2013;Shkryl & Shirokova, 2006;Yi et al., 2011).Previous investigation into the postnatal relationship between mitochondria and SR/t-tubule triads found that the abundance of tethers linking mitochondria to these calcium release units (CRUs) was greater in mature than in 2-week-old mouse muscles (Boncompagni et al., 2009).Conversely, we found here that mitochondria-SR contact site abundance was greatest at birth and then fell significantly at 2 weeks of age and was maintained at the lower level until maturation.While the 3D nature of the current analyses versus the 2D nature of the previous study (Glancy et al., 2015) may play a role in this discrepancy, it is more likely due to the specificity of the previous analyses only to mitochondria-SR tethers located at CRUs whereas our analysis was not restricted to specific subcellular domains.The functional coupling of the SR and t-tubule triads likely specializes the calcium signalling in this region to support skeletal muscle excitation-contraction coupling.Thus, it is not surprising, and is consistent with the rest of this study, that mitochondrial interactions with CRUs would increase during the postnatal developmental reorganization period in order to more optimally support contraction in mature muscles.The current data combined with the previous study also suggest that it is the regions of the SR that are not part of the triads that account for the high abundance of mitochondria-SR contacts observed in the muscle at birth here.Indeed, in neonatal muscles, the SR is frequently associated with ribosomal granules (Schiaffino & Margreth, 1969), which suggests that the SR has a greater capacity for classical endoplasmic reticulum functions such as protein synthesis (Schwarz & Blower, 2016;Westrate et al., 2015) at this time point and it is consistent with a role in cellular assembly.Additionally, calcium signalling may still take place in the non-triadic mitochondria-SR contacts observed in neonatal muscles, as calcium activity beyond its role in muscle contraction is known to be important for postnatal muscle development and cellular assembly (Debattisti et al., 2019;Stiber et al., 2008;Tu et al., 2016) and the abundance of the mitochondrial calcium uniporter is higher in both myotubes and neonatal muscles than in mature skeletal muscle (Deshmukh et al., 2015;Kim et al., 2019).Again, detailing the molecular nature of the tethers linking the SR and mitochondria in skeletal muscle and how it may differ between subcellular domains would provide key insight into the major molecular exchanges occurring at these specialized contact sites.
Building a skeletal muscle cell into a specialized contractile fibre requires the coordination of several different organelles, first to assemble and synthesize the structures needed, and then to reorganize the cell into the optimal configuration to meet the given contractile demands.During the neonatal phase in which cellular assembly is ongoing, the mitochondrial outer membrane maintains abundant contact with both the SR and large lipid droplets.
In the dynamic phase of the transition towards functional specialization for contraction, the frequency of interactions between mitochondrial and SR membranes in the muscle cell decreases quickly, whereas the reduction in mitochondria-lipid droplet interactions drops more gradually.Finally, upon maturation, the mitochondrial outer membrane surface area becomes optimized to provide energetic support through molecular exchange with the surrounding contractile apparatus.Thus, these data reflect that, in addition to cellular organelle content and composition, organelle interactions are also highly tuned to the functional demands of the cell.

Figure 1 .
Figure 1.Dynamic reorganization of mitochondrial networks during postnatal development A-F, representative 3D rendering of individual mitochondria within networks in oxidative (O; SOL) and glycolytic (G; GAS) fibres of mice at postnatal (P) day 1, 14 and 42, respectively.Mitochondrial networks are arranged along muscle contraction axis (yellow arrow) and 90°rotated images are depicted in the upper-right corner.Each colour indicates individual mitochondria.G, total mitochondrial volume (% of muscle area).H, mitochondrial number per 1000 μm 3 of muscle.I, mitochondrial network orientations are calculated in a ratio of parallel to perpendicular arrangement.J, mitochondrial surface area within 30 nm of a sarcomere.K, individual mitochondrial volume (μm 3 ).L, mitochondrial surface area to volume ratio (μm −1 ).N values: P1 oxidative -609 mitochondria, three datasets; P14 oxidative -1115 mitochondria, three datasets; P42 oxidative -1414 mitochondria, three datasets; P1 glycolytic -274 mitochondria, three datasets; P14 glycolytic -1828 mitochondria, three datasets; P42 glycolytic

Published 2024 .
This article is a U.S. Government work and is in the public domain in the USA.The Journal of Physiology published by John Wiley & Sons Ltd on behalf of The Physiological Society.

Figure 5 .
Figure 5. Specialized mitochondria-organelle interactions transition from supporting cellular assembly to contractile function in skeletal muscle cells A, B, representative single P1 mitochondrion (green) alone (A) and in contact (B) with other mitochondria (magenta), lipid droplets (blue) and SR/T (gold).C, D, representative single P14 mitochondrion alone (C) and in contact (D) with other mitochondria, lipid droplets and SR/T.E, F, representative single P42 oxidative mitochondrion alone (E) and in contact (F) with other mitochondria, lipid droplets and SR/T.G, H, representative single P42 glycolytic mitochondrion alone (G) and in contact (H) with the SR/T.I, total mitochondria-organelle contact area (%) per mitochondrion.Data equivalent to sum of Figs.1N+2F+3F.[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 6 .
Figure 6.Abundance of contractile proteins, assembly factors, and organelle contact site tethers across postnatal muscle development A, Log2 ratio of tibialis anterior muscle contractile protein abundances relative to P1.Total abundance is the sum of all contractile proteins.Mean ± SE values weight each protein equally regardless of abundance differences across proteins.Green highlight indicates increase, black indicates no change, red indicates decrease.B, sarcomere assembly factor abundances. C, other assembly factor abundances. D, chaperone abundances.

Figure 7 .
Figure 7. Functional analysis of mitochondrial energetic function during skeletal muscle postnatal development A-D, representative images of isolated flexor digitorum brevis muscle fibre NADH autofluorescence across postnatal development.Scale bars -50 μm.E-H, representative images of tetramethyl rhodamine, methyl ester (TMRM) fluorescence across postnatal development.Scale bars -7 μm for E, H, 5 μm for F, G. I-L, representative images of MitoTracker Green (MTG) fluorescence across postnatal development.Scale bars -7 μm for I, L, 5 μm for J, K. M-P, representative images of Rhod-2 fluorescence across postnatal development.Scale bars -5 μm for M, N, E, mitochondria-sarcoplasmic reticulum tether abundances grouped by putative function.F, mitochondria-lipid droplet tether abundances.G, lipid droplet-sarcoplasmic reticulum protein abundances.N = 4 separate samples for each group.One sample was pooled tissues from 3-5 mice for P1 and P7 groups, and tissue from 1 mouse for P14, P21 and P42, respectively.[Colour figure can be viewed at wileyonlinelibrary.com]