Spatial proteomics reveals heterogeneity in neural markers underpinning high-fat diet-induced myopathy in male mice

Metabolic dysfunction in skeletal muscle disturbs its contractile response as well as its innervation and vascular networks. The molecular drivers responsible for affecting decline in function remain poorly defined. To provide insight and locate these, we mapped changes in the spatial proteome occurring as a result of impaired metabolic health. We exposed male mice (C57/BL6J) to diet induced obesity to investigate the impairment of muscle metabolic function and myopathy. We conducted digital spatial profiling using the NanoString GeoMx® platform on recovered skeletal muscle (tibialis anterior) comparing it to standard fed controls. Digital spatial profiling revealed areas with shifts in the contractile protein desmin and CD31 expression, a marker of tissue stress and cellular maladaptation. We find increased expression of proinflammatory markers were identified in areas of elevated Desmin in obese samples compared to controls. Our data suggest a dietary-driven relationship between the spatial abundance of the sarcomere protein desmin and the influx of neural and inflammatory mediators to muscle. This supports the concept of pro-inflammatory events underpinning the muscle metabolic dysfunction associated with chronic non-communicative diseases such as type 2 diabetes, metabolic syndrome, and chronic obstructive pulmonary disorder.

. Regional changes in desmin abundance have been recorded and are linked 63 to shifts in mechanical force and diets high in fat reduce desmin levels (Palmisano et al., 2015). 64 Desmin abundance in human skeletal muscle is also significantly reduced in those living with 65 type 2 diabetes (Hwang et al., 2010). Using high fat diet to induce obesity in mice damages 66 temperature for 1 hour. Following from this, primary antibodies of 1:200 Goat anti-CD31 107 (R&D systems, U.K, goat anti-PECAM1) and 1:500 anti-desmin (Abcam, U.K, Alexa Fluor TM 647 108 anti-desmin, ab195177) diluted in 5% BSA were incubated on tissue for 72 hour at 4°C. 109 Following this, tissues were washing thrice with PBS and incubated with Alexa Fluor TM 488 110 Donkey anti-Goat IgG (ThermoFisher Scientific, U.K) for 2 hours and washed again with a final 111 DAPI staining. Coverslips were mounted onto slides and imaged using the Leica SP5 confocal. 112 The oil emersion x20 magnification objective was used to image the skeletal tissue sections. 113 X20 magnification was used as representative images. 20 frames per images within a Z-stack 114 image were taken. Images were transported and processed within Fiji software for further 115 analysis. 116

Confocal image analysis 117
Confocal captured images were processed within ImageJ v. 2.0.0. software. Images were 118 opened as Z-stack hyperstacks. All Images to be analysed were calibrated with a reference 119 image including a scale bar. Stacks with no fluorescent signalling were removed and the rest 120 of the hyperstack were converted into a Z projection to form a 2-dimensional image. The 121 contract, background, minimal and maximal signal on a 0-250 grey value scale were adjusted 122 in each signal channel and kept consistent between each image in each treatment group 123 analysed. Within ImageJ, integrated density of the CD31 and desmin separated channels were 124 analysed by drawing a square from the selection tool around 1024x1024 pixel images. The 125 integrated density measurement setting was applied and values were extracted from results 126  oligonucleotide-labels from the geometric segment ROIs selected. The oligo-labels from each 146 ROI were counted using nCounter analysis system using a high-resolution scan setting (555 147 FoV). 148

Data processing and analysis 149
The count data (RCC files) generated using the nCounter digital analyser transferred into the 150 GeoMX DSP analyser for further quality controls and analysis. The data was normalised to 151 the internal controls of each ROI (ribosomal protein S6, histone H3 and GAPDH) by GeoMx® 152 DSP software (v2.4). Differential expression plotted as Log2 transformed and normalized 153 counts. 154

Statistical analysis 155
All data were analysed in GraphPad Prism Version 9 and reported as the mean ± standard Tibialis anterior muscle was collected from these mice and subject to spatial proteomics 173 described in the schematic workflow ( Fig. 2A). Targeted quantification of a panel of neural 174 markers was used to measure proteins responsible for the major driving events of a loss of 175 contractile function. Digital protein expression data were recovered from muscle (Fig. 2B), 176 this was converted to log2 scaled data (Fig. 2C). Abundance of the endothelial protein CD31 177 was shown to be significantly elevated in muscle of high fat diet mice, compared to normal 178 chow. Similarly, the microtubule associated protein 2 (MAP2) neurofilament protein is also 179 induced by high fat. Using the quantitative analysis of the DSP we also detected significantly 180 increased changes across tissue in the abundance of neuronal cell markers CD68, NeuN, 181 S100B, TMEM119 and Olig2. The proliferation mark Ki-67 was also increased. These data 182 establish spatially resolved changes in the abundance of neuronal proteins as a response to a 183 high fat diet. 184 Skeletal muscle tissue exhibits spatial heterogeneity in response to high fat diet. 185 We validated and reproduced the DSP outcomes through immunological staining for CD31 in 186 normal and high fat diet fed muscle tissue (Fig. 3A-B). Counts for CD31 were significantly 187 higher in the high fat diet mice when assessed by digital spatial proteomics and 188 immunostaining using conventional antibody generated data (Fig. 3C). This agrees with DSP 189 obtained data and is supportive of a CD31 mediated response to the high fat diet induced 190 changes in muscle architecture and endothelial infiltration during metabolic stress (Pi et al., 191 2018). Cell nuclei were also stained for using SYTO83 and desmin was used as an indicator of 192 myofiber structure. Intensity of desmin staining was reduced in these mice compared to 193 normal chow (Fig. 3D). Using in situ spatial proteomics we identify these specific regions of 194 interest with desmin enrichment or CD31 depletion. To document and quantify these changes 195 in muscle and endothelial markers we divided loci dependent upon their abundance of CD31 196 or desmin (Fig. 3E). 197

Diet influences localisation of neuronal and muscle markers in skeletal muscle. 198
In quantifying these regions of interest, we identified elevated CD68, Neun, TMEM119 and 199 Myelin basic protein, CD45, CD39 and CD163 (Fig 4B). The responses recorded here by DSP 204 reflect widespread muscle inflammation during obesity (Fink et al., 2014). 205 When measuring loci of high desmin it was found that there was significantly elevated 206 expression of MAP2, synaptophysin, Ki-67 and IBA1 (Fig. 4 C). These proteins were all found 207 to have significantly increased abundance in the high fat diet muscle and were spatially 208 resolved to areas of high desmin, not CD31. This indicates that morphologically distinct areas 209 of the response to high fat diet exist within skeletal muscle. 210

Discussion 211
Spatial changes in neural markers in skeletal muscle tissue exposed to high fat diet. 212 Here we show digital spatial mapping of targeted proteins in murine skeletal muscle. We also 213 document changes in protein abundance in response to high fat diet. We used an established 214 method of dietary induced metabolic stress, through a 60% kcal fat feeding regimen. This