The myotendinous junction (MTJ) is a specialized domain of the multinucleated myofibre that is faced with the challenge of maintaining robust cell–matrix contact with the tendon under high mechanical stress and strain. Here, we profiled 24,124 nuclei in semitendinosus muscle–tendon samples from three healthy males by using single-nucleus RNA sequencing (snRNA-seq), alongside spatial transcriptomics, to gain insight into the genes characterizing this specialization in humans. We identified a cluster of MTJ myonuclei represented by 47 enriched transcripts, of which the presence of ABI3BP, ABLIM1, ADAMTSL1, BICD1, CPM, FHOD3, FRAS1 and FREM2 was confirmed at the MTJ at the protein level in immunofluorescence assays. Four distinct subclusters of MTJ myonuclei were apparent, comprising two COL22A1-expressing subclusters and two subclusters lacking COL22A1 expression but with differing fibre type profiles characterized by expression of either MYH7 or MYH1 and/or MYH2. Our findings reveal distinct myonuclei profiles of the human MTJ, which represents a weak link in the musculoskeletal system that is selectively affected in pathological conditions ranging from muscle strains to muscular dystrophies.

The myotendinous junction (MTJ) is highly specialized in both structure and molecular composition, and has evolved to meet the challenge of linking two distinct tissues and withstanding substantial forces (Knudsen et al., 2015; Kojima et al., 2008; Tidball, 1991). As the interface between muscle and the matrix-rich tendon, the MTJ is characterized by enrichment of certain proteins, the most prominent being collagen XXII (encoded by COL22A1; Koch et al., 2004). Collagen XXII is found in the basement membrane only at the myofibre tip and is seen in tendon attachment sites of muscles with diverse functions, such as limb and intercostal skeletal muscles, and the papillary muscles of the heart where they attach to the chordae tendineae (Koch et al., 2004). In a recent proteomics study, we identified 112 proteins that are enriched at the human semitendinosus MTJ when compared to levels in neighbouring muscle and tendon tissue (Karlsen et al., 2022); however, due to the nature of the MTJ as a muscle–tendon composite, this approach did not allow resolution of the origin of these proteins. Single-nucleus RNA sequencing (snRNA-seq) has led to major progress in the understanding of transcriptional diversity in the multinucleated skeletal muscle fibre, and recent studies in mice have consistently found a cluster of transcriptionally distinct myonuclei, which presumably express genes encoding MTJ proteins (Chemello et al., 2020; Dos Santos et al., 2020; Kim et al., 2020; Petrany et al., 2020; Wen et al., 2021).

The only other known transcriptionally distinct domain of the myofibre is the neuromuscular junction (NMJ). The main portion of the myofibre contains thousands of myonuclei, each of which are believed to manage their own domain, but under a common transcriptional programme, to provide mRNA for the major structural and metabolic functions of the myofibre. Whereas the NMJ has received extensive attention, much less focus has been directed to the MTJ, even though this myofibre domain is selectively affected in different pathological clinical conditions ranging from muscle strain injuries (Grange et al., 2023) to muscular dystrophies (Chemello et al., 2020; Morin et al., 2023). Interestingly, in mice with skeletal muscle-specific disruption of dystroglycan, dystroglycan staining was curiously preserved in the sarcolemma at the NMJ and MTJ (Cohn et al., 2002), further supporting the notion of segregated domains in the myofibre. Furthermore, muscle ends also appear to be the site of disease initiation (Heskamp et al., 2022), the reason for which remains unknown. Therefore, elucidation of the basic biological regulation of the MTJ domain is important in understanding the progression and treatment of a wide range of unsolved pathologies.

Here, we performed the first snRNA-seq analysis of the human muscle–tendon unit, revealing MTJ-enriched molecules and subspecialization of myonuclei at the tip of the myofibre. Previously published snRNA-seq analysis of human skeletal muscle is currently limited to tissue collected post-mortem, often with substantial ischaemic time and limited characterization of donors (Eraslan et al., 2022). To circumvent this issue, we collected surgical waste muscle–tendon tissue from three well-characterized healthy individuals, with the use of strict exclusion and inclusion criteria, and tissue was frozen within 40 min of surgical extraction. Freezing the tissue also facilitated running of the samples in the same batch. Other strengths of our approach include the avoidance of enzymatic tissue digestion, as is used in single-cell RNA sequencing (scRNA-seq) approaches, to improve preservation of the transcriptional state. Lastly, we applied a conservative statistical approach requiring differentially expressed genes (DEGs) to be significantly enriched in a cluster in the samples from each of the three individuals, thus accounting for biological variation. Therefore, in addition to our focus on the MTJ myonuclei, our data also provide high-quality profiling of nuclei in healthy human tendon and muscle tissue. Spatial transcriptomics and immunofluorescence assays supported our snRNA-seq data and together advance understanding of the MTJ myonuclei and the mononuclear cells present in healthy human muscle and tendon tissue.

snRNA-seq profile of the human MTJ in the context of muscle–tendon tissue

From three human samples (Fig. 1), cDNA libraries of a total of 25,556 nuclei were prepared using the 10× Genomics Chromium system, sequenced using droplet-based snRNA-seq and subjected to bioinformatical analysis (Table 1). Of these nuclei, 2–8% were identified as doublets and were removed, resulting in 6575, 10,140 and 7409 nuclei from the three participants for the downstream analyses (Table 1). The three datasets were integrated and subjected to unbiased clustering, resulting in 17 clusters (Fig. 2A). The clusters included all the expected cell types in tendon and skeletal muscle (Fig. 2A), and all clusters contained nuclei from all three samples (Fig. S1A). A list of DEGs for each cluster can be found in Table S1, and a complete overview of gene symbols, Ensembl gene IDs and gene names is provided in Table S2.

Fig. 1.

Overview of the experimental workflow. Semitendinosis muscle–tendon tissue was collected from patients during anterior cruciate ligament surgery, placed on ice within 8 min, cut into small pieces and snap frozen before being stored at −80°C. Frozen samples were then cut into smaller pieces in a cryochamber (−20°C) and stored at −80°C, followed by extraction and purification of nuclei using fluorescence-activated nuclei sorting (FANS). snRNA-seq libraries were constructed with the 10× Chromium platform, followed by sequencing. Bioinformatic analysis of the snRNA-seq data resulted in identification of 17 nuclei clusters, and 47 DEGs were discovered in the MTJ myonuclei cluster. Immunofluorescence staining was used to assay MTJ localization of DEG products at the protein level. Analysis of four MTJ subclusters and mapping of the snRNA-seq nuclei clusters onto spatial transcriptomics data obtained with the Visium Spatial Gene Expression kit (10x Genomics) was subsequently performed. Figure created with BioRender.com.

Fig. 1.

Overview of the experimental workflow. Semitendinosis muscle–tendon tissue was collected from patients during anterior cruciate ligament surgery, placed on ice within 8 min, cut into small pieces and snap frozen before being stored at −80°C. Frozen samples were then cut into smaller pieces in a cryochamber (−20°C) and stored at −80°C, followed by extraction and purification of nuclei using fluorescence-activated nuclei sorting (FANS). snRNA-seq libraries were constructed with the 10× Chromium platform, followed by sequencing. Bioinformatic analysis of the snRNA-seq data resulted in identification of 17 nuclei clusters, and 47 DEGs were discovered in the MTJ myonuclei cluster. Immunofluorescence staining was used to assay MTJ localization of DEG products at the protein level. Analysis of four MTJ subclusters and mapping of the snRNA-seq nuclei clusters onto spatial transcriptomics data obtained with the Visium Spatial Gene Expression kit (10x Genomics) was subsequently performed. Figure created with BioRender.com.

Fig. 2.

snRNA-seq-derived DEGs of the human MTJ. (A) snRNA-seq UMAP plot illustrating the 17 nuclei clusters in the snRNA-seq analysis of human muscle–tendon tissue (n=3). (B) Dot-plot of the expression profiles of the 47 DEGs identified in the MTJ myonuclei cluster. Dots are colour coded based on the average log-normalized expression value (expr), and the dot size indicates the percentage of nuclei in a cluster that express the DEG (pct_expr). (C) Venn diagram comparing transcripts identified as being enriched in MTJ myonuclei from mice (Dos Santos et al., 2020; Kim et al., 2020; Petrany et al., 2020; Wen et al., 2021) and humans (Karlsen et al., this study).

Fig. 2.

snRNA-seq-derived DEGs of the human MTJ. (A) snRNA-seq UMAP plot illustrating the 17 nuclei clusters in the snRNA-seq analysis of human muscle–tendon tissue (n=3). (B) Dot-plot of the expression profiles of the 47 DEGs identified in the MTJ myonuclei cluster. Dots are colour coded based on the average log-normalized expression value (expr), and the dot size indicates the percentage of nuclei in a cluster that express the DEG (pct_expr). (C) Venn diagram comparing transcripts identified as being enriched in MTJ myonuclei from mice (Dos Santos et al., 2020; Kim et al., 2020; Petrany et al., 2020; Wen et al., 2021) and humans (Karlsen et al., this study).

Table 1.

Overview of number of nuclei at all stages of the snRNA-seq process

Overview of number of nuclei at all stages of the snRNA-seq process
Overview of number of nuclei at all stages of the snRNA-seq process

Six non-myonuclei clusters were assigned through known marker genes for satellite cells (PAX7), fibroblasts (COL1A2, COL3A1, FAP, DCN, TCF7L2 and PDGFRA), tenocytes (COMP and MKX), smooth muscle cells (ACTA2, MYH11 and RGS5), endothelial cells (VWF and PECAM1) and immune cells (CD163 and F13A1) (Fig. 2A; Fig. S1B, Table S1). The remaining 11 clusters represented myonuclei, based on titin (TTN) expression, and clustered according to myofibre type. We detected two type I myosin clusters (expressing MYH7), one type IIA myosin cluster (expressing MYH2), two type IIX myosin clusters (expressing MYH1), two type IIAX myosin clusters (expressing MYH1 and MYH2) and one MTJ myonuclei cluster (expressing COL22A1, NCAM1 and SORBS2), which was found to express all three major myosin heavy chains (MYH1, MYH2 and MYH7) (Fig. 2A; Fig. S1B). Two myofibre clusters (Myo_9 and Myo_10) were not assigned an identity due to the lack of clear marker expression. The Myo_9 cluster was mainly found in one sample (Fig. S1A) and is a candidate NMJ cluster, since it was the only cluster containing genes previously identified in NMJ myonuclei clusters, such as the acetylcholine receptor CHRNA1 (Kim et al., 2020; Petrany et al., 2020), along with RAPH1 (Chemello et al., 2020; Dos Santos et al., 2020; Petrany et al., 2020), ITGA9 (Chemello et al., 2020; Dos Santos et al., 2020; Petrany et al., 2020) and PDGFC (Dos Santos et al., 2020; Petrany et al., 2020) (Table S1). The second unassigned cluster, Myo_10, was not fibre type specific (Fig. S1B). The last myonuclei cluster, the ‘Cytoplasm’ cluster, contained high levels of mitochondrial RNAs, including MT-CO1 (Fig. S1B). A large proportion of these ‘nuclei’ did not express the nuclear marker MALAT1, indicating that this cluster could represent cytoplasmic RNA and not nuclei (Fig. S1B). However, a similar Cytoplasm cluster has been observed by others, and it is currently under debate whether this cluster is an artefact or not (Eraslan et al., 2022; Tucker et al., 2020).

The main purpose of this study was to examine human MTJ myonuclei, which in our dataset were characterized by 47 DEGs. The top six most differentially expressed marker genes for the MTJ myonuclei cluster were SORBS2, BICD1, COL22A1, NCAM1, MED12L and ABLIM1 (Fig. 2B; Table S1). First, we compared these 47 DEGs with DEGs previously identified in MTJ myonuclei from mice (Fig. 2C). We used data from three mouse studies of uninjured muscles (Dos Santos et al., 2020; Petrany et al., 2020; Wen et al., 2021) and one mouse study of a mix of uninjured and regenerating muscles (Kim et al., 2020). Approximately half (23 out of 47) of the DEGs in human MTJ myonuclei were observed among the DEGs in mouse MTJ myonuclei (Fig. 2C), but only six DEGs were found in all five datasets (ABLIM1, COL22A1, FRAS1, LAMA2, PELI1 and SORBS2; Fig. 2C; Table S3).

Immunofluorescence staining confirms that MTJ myonuclei DEGs encode proteins with MTJ localization

Next, we used immunofluorescence staining to investigate whether the proteins encoded by DEGs in our human MTJ myonuclei are enriched in human MTJ tissue, and the combined snRNA-seq and immunofluorescence data led to the discovery of eight MTJ-enriched proteins (Fig. 3; Fig. S2). Most of these MTJ-enriched proteins were localized near the muscle fibre membrane in close proximity to the collagen XXII-defined MTJ (FRAS1, FREM2, BICD1, FHOD3, CPM and ADAMTSL1; Fig. 3A–F). The immunofluorescence signal for ABLIM1 was confined to the muscle fibre cytoplasm and was most intense at the tip of the fibres, gradually fading in the direction of the main portion of the muscle fibre (Fig. 3G). The ABI3BP signal was observed in the tendon and the extracellular matrix (ECM) between muscle and tendon, with a stronger immunofluorescence signal near the MTJ (Fig. 3H). Although ABI3BP transcripts were enriched in the MTJ myonuclei, it is possible that the site of ABI3BP protein deposition is determined by expression from nuclei assigned to the fibroblast and tenocyte clusters, where ABI3BP transcripts were also detected (Fig. 2B; Table S1). Notably, these eight positive immunofluorescence targets showed a varying degree of overlap with the DEGs reported for MTJ myonuclei in mice, as some were found in all four mouse datasets (ABLIM1 and FRAS1), some were only found in one or two of the mouse datasets (ABI3BP, BICD1, FHOD3 and FREM2) and some were only found in human MTJ myonuclei (ADAMTSL1 and CPM) (Fig. 2C; Table S3), underlining the good agreement as well as the diversity between these five datasets. In addition to the discovery of new components of the MTJ, the outcomes of our snRNA-seq analysis also demonstrate the robustness of our dataset from three subjects and more than 21,000 profiled nuclei.

Fig. 3.

FRAS1, FREM2, BICD1, FHOD3, CPM, ADAMTSL1, ABLIM1 and ABI3BP at the human MTJ. (A–F) Triple immunofluorescence labelling and widefield microscopy of 10-µm-thick sections of longitudinally cut human semitendinosus muscle–tendon tissue samples. FRAS1 (A), FREM2 (B), BICD1 (C), FHOD3 (D), CPM (E) and ADAMTSL1 (F) show strong enrichment at the tip of the muscle fibres in close association with the dystrophin (DYST)-stained sarcolemma (A–E), or with collagen IV (COL4)-stained basement membrane (F), and with the MTJ marker collagen XXII (COL22). Boxes indicate regions shown at higher magnification in the right-hand images. (G,H) Panels to the left show widefield microscopy images of ABLIM1 (G) and ABI3BP (H) in longitudinally cut muscle–tendon tissue sections. ABLIM1 labelling (G) is seen in the cytoplasm and is strongest near the MTJ, gradually fading along the length of the muscle fibre. ABI3BP labelling (H) is strongest in the MTJ region on the ECM side of the muscle fibres and in the tendon. The panels to the right show confocal microscopy images of isolated gracilis single muscle fibres. The graded increase in ABLIM1 intensity towards the MTJ in the cytoplasm of the muscle fibre is clearly depicted in G, and the extracellular localization of ABI3BP at the tip of the muscle fibre is clearly visible in H. Dystrophin (DYST) staining marks the sarcolemma. In A–H, the skeletal muscle fibre region is labelled with ‘M’ and the tendon is labelled with ‘T’. Nuclei were stained using DAPI (A–F) or Hoechst 33342 (G,H). Scale bars: 50 µm. Images shown are representative of n=2.

Fig. 3.

FRAS1, FREM2, BICD1, FHOD3, CPM, ADAMTSL1, ABLIM1 and ABI3BP at the human MTJ. (A–F) Triple immunofluorescence labelling and widefield microscopy of 10-µm-thick sections of longitudinally cut human semitendinosus muscle–tendon tissue samples. FRAS1 (A), FREM2 (B), BICD1 (C), FHOD3 (D), CPM (E) and ADAMTSL1 (F) show strong enrichment at the tip of the muscle fibres in close association with the dystrophin (DYST)-stained sarcolemma (A–E), or with collagen IV (COL4)-stained basement membrane (F), and with the MTJ marker collagen XXII (COL22). Boxes indicate regions shown at higher magnification in the right-hand images. (G,H) Panels to the left show widefield microscopy images of ABLIM1 (G) and ABI3BP (H) in longitudinally cut muscle–tendon tissue sections. ABLIM1 labelling (G) is seen in the cytoplasm and is strongest near the MTJ, gradually fading along the length of the muscle fibre. ABI3BP labelling (H) is strongest in the MTJ region on the ECM side of the muscle fibres and in the tendon. The panels to the right show confocal microscopy images of isolated gracilis single muscle fibres. The graded increase in ABLIM1 intensity towards the MTJ in the cytoplasm of the muscle fibre is clearly depicted in G, and the extracellular localization of ABI3BP at the tip of the muscle fibre is clearly visible in H. Dystrophin (DYST) staining marks the sarcolemma. In A–H, the skeletal muscle fibre region is labelled with ‘M’ and the tendon is labelled with ‘T’. Nuclei were stained using DAPI (A–F) or Hoechst 33342 (G,H). Scale bars: 50 µm. Images shown are representative of n=2.

snRNA-seq subcluster analysis reveals distinct myofibre domains

To further explore transcriptional heterogeneity within the MTJ myonuclei, we subjected the main clusters to subclustering by repeating the clustering selectively on the MTJ cluster; the two type I myosin clusters (MyHCI_1 and MyHCI_2); the type IIA, type IIX and type IIAX myosin clusters (MyHCIIA, MyHCIIAX_1, MyHCIIAX_2, MyHCIIX_1, MyHCIIX_2); the immune cell cluster (Immune); the fibroblast and tenocyte clusters (Fib, Teno; referred to here collectively as FibTen); the smooth muscle cell cluster (Smooth); and the endothelial cell cluster (Endo). This gave rise to a total of 30 subclusters, including four MTJ myonuclei subclusters; 26 were new subclusters (indicated by ‘Sub’ in the cluster names), and four corresponded to clusters from the original analysis [Cytoplasm, Myo_9, Myo_10 and satellite cells (SatCell)] (Fig. 4A; Fig. S1C,D, Table S4). To compare cell cluster markers we assessed the similarity between our 30 nuclei subclusters and cell types or nuclei types reported in the literature (Fig. S3) (Barruet et al., 2020; Chemello et al., 2020; De Micheli et al., 2020a,b; Dos Santos et al., 2020; Eraslan et al., 2022; Harvey et al., 2019; He et al., 2020; Kendal et al., 2020; Kim et al., 2020; Perez et al., 2022; Petrany et al., 2020; Rubenstein et al., 2020; Scripture-Adams et al., 2022; Wen et al., 2021; Yan et al., 2022). This revealed a good alignment between our clusters and those identified using snRNA-seq (Eraslan et al., 2022; Perez et al., 2022) and scRNA-seq (He et al., 2020; Rubenstein et al., 2020) in human skeletal muscle, and those identified using snRNA-seq (Chemello et al., 2020; Dos Santos et al., 2020; Wen et al., 2021) and scRNA-seq (De Micheli et al., 2020b) in mice.

Fig. 4.

snRNA-seq subcluster profile of the human MTJ. (A) UMAP plot illustrating the 30 snRNA-seq nuclei subclusters in human semitendinosus muscle–tendon tissue. In the inset panel, the position of nuclei in each of the four MTJ myonuclei subclusters is displayed separately in red. (B) Venn diagram showing the overlap of the 99 MTJ DEGs within the four MTJ subclusters. The 47 MTJ myonuclei DEGs from the MTJ myonuclei cluster are highlighted in bold. (C) Dot plot showing the expression profiles of the 99 DEGs in the four MTJ myonuclei subclusters. Dots are colour coded based on the average log-normalized expression value (expr), and the dot size indicates the percentage of nuclei in a cluster that express the DEG (pct_expr). (D) Violin plot of gene expression levels (log-normalized values) in the four MTJ myonuclei subclusters. Genes are MYH1 (myosin heavy chain IIX), MYH2 (myosin heavy chain IIA), MYH7 (myosin heavy chain I), COL22A1 (collagen XXII), NCAM1 (neural cell adhesion molecule 1).

Fig. 4.

snRNA-seq subcluster profile of the human MTJ. (A) UMAP plot illustrating the 30 snRNA-seq nuclei subclusters in human semitendinosus muscle–tendon tissue. In the inset panel, the position of nuclei in each of the four MTJ myonuclei subclusters is displayed separately in red. (B) Venn diagram showing the overlap of the 99 MTJ DEGs within the four MTJ subclusters. The 47 MTJ myonuclei DEGs from the MTJ myonuclei cluster are highlighted in bold. (C) Dot plot showing the expression profiles of the 99 DEGs in the four MTJ myonuclei subclusters. Dots are colour coded based on the average log-normalized expression value (expr), and the dot size indicates the percentage of nuclei in a cluster that express the DEG (pct_expr). (D) Violin plot of gene expression levels (log-normalized values) in the four MTJ myonuclei subclusters. Genes are MYH1 (myosin heavy chain IIX), MYH2 (myosin heavy chain IIA), MYH7 (myosin heavy chain I), COL22A1 (collagen XXII), NCAM1 (neural cell adhesion molecule 1).

The four MTJ myonuclei subclusters (Fig. 4A) together expressed 99 DEGs (Fig. 4B,C; Fig. S4, Table S4). Of these, 15 encode proteins belonging to the matrisome (Naba et al., 2012) and can be divided into: (1) core matrisome genes, including collagens (COL22A1), ECM glycoproteins (ABI3BP, FRAS1, LAMA2, LTBP1 and CCN2) and proteoglycans (HSPG2 and SPOCK1); and (2) matrisome-associated genes, including ECM regulators (ADAMTSL1 and P4HA1), ECM-affiliated proteins (FREM2 and GPC3) and secreted factors (ANGPT1, EGF and TGFB2). The 99 DEGs also encode proteins associated with actin assembly (FHOD3, ENAH and SORBS2), the microtubule–dynein complex (BICD1, DYNC1I1, GPSM2, KAZN and NCKAP5), TGF-β signalling (SMAD9, SNX25, TGFB2 and TCF7L1) and heparan sulfate chains (SPOCK1, EXT1, GPC3 and HS6ST3).

Transcriptional heterogeneity between the four MTJ myonuclei subclusters was evident in the form of two COL22A1-positive (COL22A1+) subclusters with predominantly fast-type myosin (MYH1 and/or MYH2, subsequently referred to collectively as MYH1/2) expression (MTJ_Sub2 and MTJ_Sub4), and two COL22A1-negative (COL22A1) subclusters with fibre type specificity (MTJ_Sub1 and MTJ_Sub3), predominantly expressing either fast-type (MYH1/2; MTJ_Sub1) or slow-type (MYH7; MTJ_Sub3) myosin (Fig. 4D). The uniform manifold approximation and projection (UMAP) cluster plot indicated greatest similarity between MTJ_Sub2 and MTJ_Sub4, which overlapped in the central part of the MTJ cluster (Fig. 4A). In addition, these two subclusters had a high prevalence of the DEGs (93 of 99 DEGs in the MTJ_Sub2 and MTJ_Sub4 cluster; Fig. 4B). Taken together, these observations suggest that MTJ_Sub2 and MTJ_Sub4 constitute the population of myonuclei residing at the MTJ domain, which is defined by the enrichment of collagen XXII protein. Notably, neither of these COL22A1+ subclusters had high expression of MYH7 (Fig. 4D). In contrast, MTJ_Sub1 and MTJ_Sub3 seemed to be less MTJ specific (i.e., COL22A1), and together expressed only 31 of the 99 DEGs (Fig. 4B; Table S4). In the UMAP plot these subclusters extended from the central part of the MTJ cluster towards the MyHCII cluster or the MyHCI cluster (Fig. 4A), in agreement with their differential expression of fast (MYH1/2) and slow myosin (MYH7), respectively (Fig. 4D).

A transitional COL22A1 myofibre domain

Based on these findings, we hypothesized that MTJ_Sub1 and MTJ_Sub3 belong to a population of myonuclei residing in a transitional MTJ domain, located between the collagen XXII-rich MTJ domain and the main portion of the muscle fibre. Accordingly, we define the transitional MTJ domain as a region of the muscle fibre without enrichment of the collagen XXII protein but with high expression of other proteins also enriched in the MTJ domain and not in the main portion of the muscle fibre. The only protein with this immunofluorescence staining pattern was ABLIM1 (Fig. 3G; Fig. S2), which demonstrated a similar staining pattern to that of NCAM1 at the MTJ (Daniloff et al., 1989; Jakobsen et al., 2018, 2021; Karlsen et al., 2022; Rieger et al., 1985). In line with this staining pattern, ABLIM1 and NCAM1 were among the few DEGs enriched in all four MTJ subclusters (Fig. 4B; Table S4). To further explore these observations, we determined the proportion of nuclei expressing ABLIM1 and/or NCAM1 in all 20,314 myonuclei in the MyHC and MTJ subclusters (Table S5). This analysis showed a marked increase in the proportion of nuclei expressing ABLIM1 and/or NCAM1 from the myonuclei in the main portion of the fibre (7.3%) to the transitional MTJ_Sub1 and MTJ_Sub3 myonuclei (83.6%), with the peak seen in the MTJ-specific MTJ_Sub2 and MTJ_Sub4 myonuclei (97.1%). These observations are schematically illustrated in Fig. 5A and D, together with immunofluorescence images of ABLIM1 and NCAM1, and their gradually increasing staining intensity towards the MTJ (Fig. 5C).

Fig. 5.

Schematic illustration of the main findings. (A) The myofibre shown is divided into three regions: the main portion of the myofibre, the transitional region and the MTJ region, which extends from the myofibre to the edge of the tendon. Each region is labelled to indicate the corresponding snRNA-seq nuclei subclusters, the protein localization confirmed by the immunofluorescence assays, and characteristic gene expression identified in the snRNA-seq data. See D for further details of symbols used. (B) Spatial transcriptomics mapping of the indicated snRNA-seq subclusters. Images show a 10-µm-thick section of human muscle–tendon tissue with autofluorescence in green (M, muscle; T, tendon). Each dot represents a single sampling site in the spatial transcriptomics grid corresponding to a 100×100 µm region. The colour intensity of each dot shows the level of overlap between snRNA-seq gene transcripts within a cluster on the spatial transcriptomic map. Dashed boxes link the subclusters to the relevant regions in A. Scale bar: 500 µm. (C) Single-channel immunofluorescence images with triple labelling of ABLIM1, NCAM1 and collagen XXII (COL22) proteins, with the signal intensity visualized by a heatmap overlay of the pixel intensity (mpl-inferno, ImageJ). M, muscle; T, tendon. Scale bar: 50 µm. Images shown are representative of n=3 (snRNA-seq), n=4 (spatial transcriptomics), n=2 (immunofluorescence). (D) Key for symbols used in A. Note that the MTJ is defined by high expression of collagen XXII protein and expresses several other membrane-associated proteins, including FRAS1, FREM2, BICD1, FHOD3, CPM and ADAMTSL1. In the myofibre cytoplasm there is a high abundance of ABLIM1 and NCAM1 protein in the MTJ region, as well as in the transitional region where the expression of collagen XXII protein is low. This protein expression corresponds to the diverse gene expression profile of nuclei in the MTJ_Sub2 and MTJ_Sub4 subclusters (MTJ myonuclei) versus the nuclei in the MTJ_Sub1 and MTJ_Sub3 subclusters (transitional myonuclei). In contrast, the abundance of all these MTJ-expressed proteins, including ABLIM1 and NCAM1, is very low in the main portion of the myofibre, corresponding to the gene expression profile of the myonuclei in the main portion of the fibre. The gene expression profile of the FibTen_Sub3 cluster includes expression of ABI3BP, THBS4, TNC, CILP and ITGA10, and the associated proteins are highly abundant in the ECM in the MTJ region and the tendon (ABI3BP, THBS4, TNC), or in the myofibre membrane at the MTJ (CILP, ITGA10; see Karlsen et al., 2022). This indicates that FibTen_Sub3 represents MTJ-related ‘myotenocytes’, in contrast to the tenocyte profile of FibTen_Sub5. Elements of this figure were created with BioRender.com.

Fig. 5.

Schematic illustration of the main findings. (A) The myofibre shown is divided into three regions: the main portion of the myofibre, the transitional region and the MTJ region, which extends from the myofibre to the edge of the tendon. Each region is labelled to indicate the corresponding snRNA-seq nuclei subclusters, the protein localization confirmed by the immunofluorescence assays, and characteristic gene expression identified in the snRNA-seq data. See D for further details of symbols used. (B) Spatial transcriptomics mapping of the indicated snRNA-seq subclusters. Images show a 10-µm-thick section of human muscle–tendon tissue with autofluorescence in green (M, muscle; T, tendon). Each dot represents a single sampling site in the spatial transcriptomics grid corresponding to a 100×100 µm region. The colour intensity of each dot shows the level of overlap between snRNA-seq gene transcripts within a cluster on the spatial transcriptomic map. Dashed boxes link the subclusters to the relevant regions in A. Scale bar: 500 µm. (C) Single-channel immunofluorescence images with triple labelling of ABLIM1, NCAM1 and collagen XXII (COL22) proteins, with the signal intensity visualized by a heatmap overlay of the pixel intensity (mpl-inferno, ImageJ). M, muscle; T, tendon. Scale bar: 50 µm. Images shown are representative of n=3 (snRNA-seq), n=4 (spatial transcriptomics), n=2 (immunofluorescence). (D) Key for symbols used in A. Note that the MTJ is defined by high expression of collagen XXII protein and expresses several other membrane-associated proteins, including FRAS1, FREM2, BICD1, FHOD3, CPM and ADAMTSL1. In the myofibre cytoplasm there is a high abundance of ABLIM1 and NCAM1 protein in the MTJ region, as well as in the transitional region where the expression of collagen XXII protein is low. This protein expression corresponds to the diverse gene expression profile of nuclei in the MTJ_Sub2 and MTJ_Sub4 subclusters (MTJ myonuclei) versus the nuclei in the MTJ_Sub1 and MTJ_Sub3 subclusters (transitional myonuclei). In contrast, the abundance of all these MTJ-expressed proteins, including ABLIM1 and NCAM1, is very low in the main portion of the myofibre, corresponding to the gene expression profile of the myonuclei in the main portion of the fibre. The gene expression profile of the FibTen_Sub3 cluster includes expression of ABI3BP, THBS4, TNC, CILP and ITGA10, and the associated proteins are highly abundant in the ECM in the MTJ region and the tendon (ABI3BP, THBS4, TNC), or in the myofibre membrane at the MTJ (CILP, ITGA10; see Karlsen et al., 2022). This indicates that FibTen_Sub3 represents MTJ-related ‘myotenocytes’, in contrast to the tenocyte profile of FibTen_Sub5. Elements of this figure were created with BioRender.com.

Unexpectedly low abundance of COL22A1+ MYH7+ MTJ myonuclei

The low level of MYH7 expression in COL22A1+ MTJ myonuclei was unexpected because it is generally believed, although not shown, that the collagen XXII protein is enriched at the MTJ in both slow and fast muscle fibres. Therefore, we first confirmed, using immunofluorescence, the clear presence of collagen XXII protein in slow and fast muscle fibres at the human MTJ (Fig. S5). Next, we ruled out that COL22A1+ MYH7+ nuclei were hidden within the non-MTJ myonuclei clusters, as there were only 30 COL22A1+ nuclei in the non-MTJ myonuclei subclusters, but none of these were MYH7 positive (Table S5). Notably, 19 of these nuclei belonged to the FibTen_Sub3 cluster and were negative for expression of MYH1/2 and MYH7. We then proceeded with a detailed examination of MYH1/2 and MYH7 expression in all COL22A1+ and COL22A1 MTJ myonuclei (Table S5), which is schematically illustrated in Fig. 6. Within the total number of 1206 MTJ myonuclei, 282 were COL22A1+ and 924 were COL22A1. The largest proportion (82%) of the COL22A1+ MTJ myonuclei was found in MTJ_Sub2 and MTJ_Sub4, whereas only 23.9% of the 924 COL22A1 MTJ myonuclei belonged to these MTJ subclusters. A small fraction of the 282 COL22A1+ MTJ myonuclei expressed MYH7 (16 nuclei, 5.7%), or a mix of MYH7 and either MYH1 or MYH2 (seven nuclei, 2.5%). In contrast, more than half of the COL22A1+ MTJ myonuclei expressed at least one of the fast myosins (MYH1/2; 150 myonuclei, 53.2%) without expressing MYH7. This resulted in the abundance of COL22A1+ MYH1/2+ myonuclei being 9.4 times that of COL22A1+ MYH7+ myonuclei (Fig. 6; Table S5). Because a larger proportion of the total myonuclei was assigned to the MyHCII cluster than to the MyHCI cluster, we computed correction factors for this fibre type distribution and other potential confounders (see Materials and Methods for correction factors), resulting in a 4.2 times greater abundance of COL22A1+ MYH1/2+ myonuclei than of COL22A1+ MYH7+ myonuclei after correction, suggesting that the underlying explanation for the difference in abundance is biological.

Fig. 6.

Schematic illustration of the low abundance of MTJ myonuclei expressing slow myosin and collagen XXII. (A) Schematic illustration of the expression of COL22A1, MYH1, MYH2 and MYH7 in MTJ myonuclei subclusters. The four MTJ myonuclei subclusters were pooled in pairs based on either high (MTJ_Sub2 and MTJ_Sub4, upper quadrants) or low (MTJ_Sub1 and MTJ_Sub3, lower quadrants) expression of COL22A1. All COL22A1+ nuclei are accounted for in the quadrants to the right, and all COL22A1 nuclei in the quadrants to the left. Note the low abundance of COL22A1+ MTJ myonuclei with MYH7 expression (highlighted in the two quadrants on the right). The raw data used in this analysis can be found in Table S5. (B) Illustration of the disproportional prevalence of MYH7+ versus MYH1/2+ MTJ myonuclei with COL22A1 expression. Note that the numbers in this figure are taken from the raw data, and application of appropriate correction factors resulted in a prevalence of COL22A1+MYH1/2+ nuclei that is 4.2 times that of COL22A1+MYH7+ nuclei (see Materials and Methods for details). Elements of this figure were created with BioRender.com.

Fig. 6.

Schematic illustration of the low abundance of MTJ myonuclei expressing slow myosin and collagen XXII. (A) Schematic illustration of the expression of COL22A1, MYH1, MYH2 and MYH7 in MTJ myonuclei subclusters. The four MTJ myonuclei subclusters were pooled in pairs based on either high (MTJ_Sub2 and MTJ_Sub4, upper quadrants) or low (MTJ_Sub1 and MTJ_Sub3, lower quadrants) expression of COL22A1. All COL22A1+ nuclei are accounted for in the quadrants to the right, and all COL22A1 nuclei in the quadrants to the left. Note the low abundance of COL22A1+ MTJ myonuclei with MYH7 expression (highlighted in the two quadrants on the right). The raw data used in this analysis can be found in Table S5. (B) Illustration of the disproportional prevalence of MYH7+ versus MYH1/2+ MTJ myonuclei with COL22A1 expression. Note that the numbers in this figure are taken from the raw data, and application of appropriate correction factors resulted in a prevalence of COL22A1+MYH1/2+ nuclei that is 4.2 times that of COL22A1+MYH7+ nuclei (see Materials and Methods for details). Elements of this figure were created with BioRender.com.

Spatial transcriptomics of human muscle–tendon tissue

Spatial transcriptomics was performed on cryosections from four individuals, where each predefined area contained a cross section and longitudinal section from the same sample. The high level of autofluorescence of tendon tissue was a useful guide for determining the position of muscle and tendon in the sections (Fig. S6Aa). Together, in all four samples, we identified 33,215 features across 6599 unique capture spots (Fig. S6Ab). Key muscle (MYH7) and tendon (COMP) transcripts were distributed clearly according to their respective tissues, whereas the position of COL22A1, the MTJ marker, was unclear due to a low number of reads (Fig. S6Ac). Six clusters were detected (Fig. S6Ad,e). DEGs were only found for clusters 2 and 4, which were characterized by gene expression associated with muscle (ACTA1, TTN, TPT1 and GAPDH) and tendon (COMP, DCN, PRELP and THBS4), respectively (Table S6).

Mapping snRNA-seq subclusters onto spatial transcriptomics data

The four snRNA-seq MTJ myonuclei subclusters corresponded with the spatial transcriptomics rendering of the muscle–tendon tissue interface (Fig. 5B; Fig. S6B). Furthermore, FibTen_Sub5 (and to a lesser extent FibTen_Sub3) clearly mapped to the tendon proper, which unequivocally identifies FibTen_Sub5 as a population of pure tenocyte nuclei (Fig. 5B; Fig. S6B). This is in line with the lack of concordance between FibTen_Sub5 and previously reported cell clusters derived from pure skeletal muscle (Fig. S3A,B), and with the observed similarity to cell clusters identified in human scRNA-seq data from muscle, tendon and MTJ (Fig. S3B; Yan et al., 2022), and in human scRNA-seq data from pure tendon (Fig. S3B; Kendal et al., 2020). For the remaining cell clusters (endothelial cells, smooth muscle cells, and immune cells), the spatial analysis showed a ubiquitous distribution across muscle and tendon tissue (Fig. S6B).

More than 30 years ago, protein enrichment of talin and desmin at the MTJ was reported by James G. Tidball (Tidball, 1992; Tidball et al., 1986), and over a decade later Manuel Koch and colleagues identified collagen XXII as a marker of the MTJ in skeletal and cardiac muscles (Koch et al., 2004). Unbiased discovery of new proteins at the MTJ is challenging due to difficulties in isolating this narrow anatomical region comprising two distinct tissues, and mapping of the proteins and RNAs present at the MTJ is still incomplete. Here, by integrated snRNA-seq and spatial transcriptomics of healthy human muscle–tendon tissue, we identify 99 DEGs across four distinct clusters of myonuclei at the MTJ. From these genes, we demonstrate MTJ enrichment for eight proteins using immunofluorescence staining. The specific roles of these proteins at the MTJ are unknown, but the proteins have a wide range of functions, including actin assembly and F-actin formation (ABLIM1 and FHOD3), sarcomere organization (FHOD3), basement membrane adhesion (FRAS1 and FREM2), intracellular cargo transport (BICD1), metallopeptidase activity (ADAMTSL1), macrophage differentiation (CPM), peptide hormone degradation at the cell surface (CPM), ECM organization (ABI3BP) and cell–substrate adhesion (ABI3BP). Based on nuclei profiles, we also identify a transitional region between the MTJ and the myofibre proper, in addition to a population of MTJ mononuclear cells.

The main focus on MTJ myonuclei in the present study was inspired by the consistent reporting of a specialized cluster of MTJ myonuclei in recent snRNA-seq studies in mice (Dos Santos et al., 2020; Kim et al., 2020; Petrany et al., 2020; Wen et al., 2021). Here, we present the first transcriptional profile of human MTJ myonuclei, constituting 47 DEGs, including transcripts for the known MTJ-enriched proteins collagen XXII (COL22A1) and neural cell adhesion molecule 1 (NCAM1). Notably, approximately half of these DEGs were not reported previously in studies including rodent MTJ myonuclei, and only six DEGs were found to be enriched in human MTJ myonuclei and all the rodent MTJ clusters (Dos Santos et al., 2020; Kim et al., 2020; Petrany et al., 2020; Wen et al., 2021), highlighting the importance of studying human tissue and underlining the differences that can be observed in snRNA-seq data from different studies, which are probably related to factors such as the method used to isolate nuclei, the choice of muscles and the bioinformatics approach taken.

According to our UMAP analysis and immunofluorescence staining patterns of proteins differentially expressed by the four MTJ subclusters, the COL22A1+ MTJ_Sub2 and MTJ_Sub4 subclusters represent a population of myonuclei belonging to the collagen XXII-defined MTJ domain. The low expression of COL22A1 we observed in MTJ_Sub1 and MTJ_Sub3 has previously been reported in some (Kim et al., 2020) or all (Wen et al., 2021) MTJ subclusters. However, by immunofluorescence we could couple the gene expression in MTJ_Sub2 and MTJ_Sub4 (ABLIM1, COL22A1, CPM, FRAS1, FREM2 and NCAM1) with protein enrichment specifically at the MTJ, and the gene expression in MTJ_Sub1 and MTJ_Sub3 (ABLIM1 and NCAM1) with protein enrichment in the transitional myofibre region. While the presence of a transitional region of the myofibre has been indicated at the protein level by consistent immunofluorescence staining of NCAM1 at the MTJ also in regions without collagen XXII (Jakobsen et al., 2018, 2021; Karlsen et al., 2022), this is to our knowledge the first evidence of a transcriptionally corresponding population of COL22A1 NCAM1+ myonuclei. The high expression of ABLIM1 in all four MTJ subclusters together with the immunofluorescence staining pattern of the ABLIM1 protein places ABLIM1 alongside NCAM1 in this transitional region and at the MTJ. Other DEGs in the four subclusters could share similar properties regarding the collagen XXII-defined MTJ domain and the transitional domain. It is worth noting that the snRNA-seq dataset appears to be validated by the immunofluorescence staining assays and spatial transcriptomics, spatially positioning the MTJ clusters (and others) as expected. Taken together, these findings expand our understanding of the compartmentalization at the MTJ from gene expression to protein enrichment and raise the question of whether such a transitional region also exists adjacent to the postsynaptic myonuclei at the NMJ.

In addition to our focus on the MTJ, a few observations are worth noting. Firstly, satellite cell nuclei were only found in a single well-defined cluster, indicating a lack of states or subtypes of satellite cells at the MTJ under basal conditions. Secondly, 15 of the 99 enriched genes in the MTJ myonuclei are designated as encoding matrisome proteins (Naba et al., 2012), which is interesting as these matrisome genes were enriched in myonuclei, in line with recent findings (Murach et al., 2022). In further support of a role for myonuclei in maintaining the muscle ECM, it is worth noting that LAMA2, which encodes the muscle-specific laminin known as merosin (Patton et al., 1997), was enriched in clusters representing the MTJ myonuclei, fibroblasts, satellite cells and tenocytes (Fig. 2B; Table S1), revealing a joint contribution of these cell types and myonuclei to regeneration of the basement membrane (Mackey and Kjaer, 2017a; Vracko and Benditt, 1972). Finally, enriched genes identified in our analysis of pooled type I myonuclei and type II myonuclei support proteomics profiles of single human myofibres (Murgia et al., 2021) and are associated with structures such as sarcoplasmic reticulum (ATP2A2 and ATP2A1) and sarcomeres (MYOM3, TNNC2, TNNT3, TNNI2, TPM1 and TPM3) (Table S1, type I and type II). MYOM3 was the only sarcomeric gene we found with enriched expression at the MTJ, where it was detected in the MTJ_Sub3 cluster, approaching the MYH7+ type I myosin clusters on the UMAP plot. We find the unexpectedly low prevalence of COL22A1+MYH7+ myonuclei intriguing. Whether this can be explained by a lower turnover rate of MTJ proteins in slow myofibres is unknown, but these data clearly suggest further specialization between fast and slow myofibres in the MTJ domain. In relation to sarcomeric proteins, we found that expression of RBM20 was enriched in all four MTJ subclusters (Table S1). RBM20 is a potent regulator of titin splicing (Riley et al., 2022), which suggests structural specialization of the myofibre tips at the level of the sarcomere.

The DEG signature of each of the FibTen subclusters serves to further underline the intrinsic differences between muscle fibroblasts and tenocytes, as has recently been reported at the protein level in extracellular vesicles isolated from human tenocytes and muscle fibroblasts (Yeung et al., 2020). The spatial transcriptomics and cross-comparison with cell clusters in human snRNA-seq and scRNA-seq data clearly pointed towards FibTen_Sub5 being a pure tenocyte cluster. FibTen_Sub3 transcripts seemed to be more spatially concentrated at the periphery of the tendon (Fig. S6B), indicating that this cluster is the best match to a recently reported muscle–tendon progenitor subpopulation (Fig. S3B) (Yan et al., 2022). The FibTen_Sub3 also demonstrated a high representation of transcripts encoding the MTJ-enriched proteins that we discovered previously using liquid chromatography–mass spectrometry-based proteomics, of which CILP, ITGA10 and THBS4 were also confirmed by immunofluorescence analysis (Karlsen et al., 2022), as well as HMCN1, which has recently been found at the MTJ (Welcker et al., 2021), and the muscle fibre-specific LAMA2 (Table S4). FibTen_Sub3 was also the non-myonuclei cluster containing the most COL22A1+ nuclei (Table S5). The presence of a COL22A1-expressing tenocyte population has previously been reported in scRNA-seq (Scott et al., 2019) and snRNA-seq studies using mice (Dos Santos et al., 2020; Petrany et al., 2020; Wen et al., 2021), and together our data indicate that the FibTen_Sub3 cluster represents a special population of peripheral tendon cells, or myotenocytes (Scott et al., 2019), that are responsible for maintenance of the tendon side of the MTJ. Taken together, these findings on fibroblasts and tenocytes provide insight into the cells responsible for maintaining the matrix of muscle, tendon and the specialized zone of the muscle–tendon interface.

In conclusion, through profiling the human muscle–tendon interface at the single-nucleus level, we have advanced the understanding of this specialized zone, which is a key region in muscle pathophysiology. Myonuclei in the MTJ cluster of dystrophin-deficient mice show many genes with an abnormal gene expression pattern compared to that of the other myonuclei in fibre type-specific clusters (Chemello et al., 2020), and the MTJ is resistant to experimental molecular disruption of muscle proteins (Cohn et al., 2002). Furthermore, MTJ strain injuries, induced by explosive high-force movements, have poor clinical prognosis with a high rate of re-injury. Clearly, many gaps remain in our understanding of MTJ regulation and repair. To this end, our dataset provides a detailed reference for gene expression at the single-nucleus level of the human MTJ, in the context of the full repertoire of mononuclear cells and myofibre types in healthy human muscle and tendon tissue.

Human subjects and sample collection

Semitendinosus muscle–tendon tissue from healthy patients was collected from four male patients for snRNA-seq and from two male and two female patients for spatial transcriptomics, and gracilis muscle–tendon tissue from healthy patients was collected from one male and one female patient for single-fibre microscopy (age 20–44 years, height 165–185 cm, weight 66–97 kg, body mass index 20.9–29.0, non-smokers with no known disease such as diabetes, arthritis or blood-borne diseases), all of whom were scheduled for anterior cruciate ligament reconstruction surgery with hamstring tendon grafts. None of the patients had performed regular heavy resistance exercise with the hamstring muscles in the months prior to the surgery. The study was approved by the Research Ethics Committees of the Capital Region of Denmark (ref. H–3–2010–070 and ref. H-20044907) and conformed to the Declaration of Helsinki II. All patients gave written informed consent before inclusion in the study.

On the day of the surgery, excess semitendinosus tissue was collected within 8 min in a precooled 50 ml falcon tube with 1% ice-cooled PBS and was subsequently kept on ice while being divided into smaller pieces (∼50 mg each; Fig. 1). Tissue for snRNA-seq was transferred to cryotubes and immediately frozen in liquid nitrogen within a maximum of 40 min from the time of extraction from the body before being stored at −80°C. All the samples contained a piece of tendon with the muscle fibres attached. Tissue for spatial transcriptomics and immunofluorescence staining was carefully aligned and embedded in Tissue-Tek (Sakura Finetek, Europe, AJ Alphen aan den Rijn, The Netherlands), then frozen in isopentane precooled in liquid nitrogen before storage at −80°C. Gracilis single fibres for immunofluorescence staining were fixed as described previously (Mackey and Kjaer, 2017b; Ralston et al., 1999).

Isolation of nuclei

The samples were removed from the −80°C freezer and transferred on dry ice to a cryochamber (−20°C), where a total of ∼500 mg tissue was cut into fine pieces with a scalpel (Fig. 1), before being divided evenly into ten 2 ml screwcap tubes (522-S, Techtum, Nacka, Sweden), each containing five 2.3 mm stainless steel balls (BioSpec Products, Bartlesville, OK), and stored at −80°C.

Nuclei were isolated using the Nuclei EZ Prep kit (NUC101; Sigma-Aldrich, St Louis, MO, USA), as follows. First, 1 ml Nucleic EZ lysis buffer was added to each tissue tube, and the tubes were immediately shaken for 20 s at 4 m/s in a FastPrep-24 (MP Biomedicals, Illkirch, France), followed by 5 min cooling on ice. Next, the shaking and cooling were repeated, and the homogenate was transferred to a new tube and gently mixed with 0.5 ml Nuclei EZ lysis buffer using a 1000 µl pipette. The tubes were incubated for an additional 5 min on ice, during which they were gently mixed twice. Then, the homogenate from all ten tubes was filtered through a 70 µm mini strainer (PluriStrainer Mini; pluriSelect Life Science, Leipzig, Germany), including a wash with 1.5 ml Nuclei EZ lysis buffer. The nuclei were spun down (500 g, 5 min, 4°C), and the pellet was resuspended in 1.5 ml Nuclei EZ lysis buffer by mixing ten times using a 1000 µl pipette. The nuclei were spun down (500 g, 5 min, 4°C), and 0.5 ml nuclei wash and suspension buffer [NWS-Buffer: 2% BSA, 2 mM MgCl2, 0.5% Protector RNAse Inhibitor (SKU3335399001; Merck, Soeborg, Denmark)] was carefully added, followed by 5 min incubation on ice before the nuclei were resuspended in 1.0 ml NWS-Buffer. The wash was repeated, and the nuclei were resuspended in 1 ml NWS-buffer, followed by filtering through a 40 µm mini strainer (PluriStrainer Mini), including a wash with 0.5 ml NWS-buffer. The nuclei were pelleted again and resuspended in 1.5 ml NWS-buffer. 1.2 ml was pelleted and resuspended in 300 µl NWS-buffer. The nuclei were stained with 7-AAD (1 mg/ml; A9400-1MG, Sigma-Aldrich) and sorted by fluorescence-activated nuclei sorting (FANS) to isolate single nuclei.

Single-nucleus RNA sequencing

Approximately 15,000 nuclei per sample were used to generate an snRNA-seq library using the Chromium Next GEM Single Cell 3′ Kit v3.1 (PN-1000269; 10× Genomics, Pleasanton, CA) according to the manufacturer's protocol.

The snRNA-seq libraries were sequenced on an Illumina NovaSeq 6000 machine, obtaining ∼450 million read pairs with 2×150 bp sequencing, by GENEWIZ (Leipzig, Germany). One sample failed the cDNA synthesis step, and therefore the final sequencing data represent single nuclei from three individuals.

Analysis of single-nucleus RNA sequencing data

The read sequences were paired and the read quality was filtered by using FastP v0.19.4 (Chen et al., 2018) with default settings (adaptor trimming disabled), followed by alignment to the transcriptome and read counting using Cell Ranger count v3.1.0 (10× Genomics). The transcriptome was generated from the human genome assembly GRCh38, with introns included in the transcripts to obtain pre-mRNA sequences.

Identification of nuclei-containing droplets and reduction of background from ambient RNA was achieved using CellBender v0.2.0 (Fleming et al., 2022 preprint) with recommended settings. Nuclei doublets within the droplets were removed by using Scrublet vDec2020 (Wolock et al., 2019) for three rounds, as described previously for tissue. The metrics are shown in Table 1.

The three samples were combined using the Seurat v4.1.0 package (Stuart et al., 2019) by first normalizing each dataset using SCTransform and then integrating the data using FindIntegrationAnchors and IntegrateData. Main nuclei types were defined using FindNeighbours, based on principal component analysis (PCA; 30 dimensions), and the Seurat graph-based method FindClusters, for clustering using manually selected resolution between 0.2 and 0.9 for best separations. Subtypes were found by repeating the clustering separately on selected main cluster types (MTJ, MyHCI, MyHCII, Immune, Fib+Ten, Smooth or Endo). The clusters were visualized in UMAPs using the Seurat DimPlot function. Markers for each of the clusters (DEGs) were found using Seurat FindConservedMarkers, selecting markers conserved for all three samples [fold change (FC)>2 and false discovery rate (FDR)<0.05 for all three samples]. Lists of marker genes for clusters identified in published scRNA-seq and snRNA-seq experiments were compared with the clusters identified in our data using the Seurat AddModuleScore function.

Calculation of correction factors for COL22A1+ MYH7+ nuclei

The observed imbalance between the 9.4 times higher proportion of COL22A1+ MYH1/2+ nuclei (150 nuclei) versus COL22A1+ MYH7+ nuclei (16 nuclei) was based on transcript counting at the level of single nuclei, which could lead to systematic biased conclusions. Therefore, we first applied a false-negative correction factor to account for the prevalence of MYH7-negative myonuclei within in the MyHCI clusters and MYH1/2-negative myonuclei within the MyHCII clusters. Within the total of 6862 myonuclei in the MyHCI clusters, 73.9% expressed MYH7. The MyHCI false-negative correction factor was calculated by dividing the total number of MyHCI myonuclei by the number of MyHCI myonuclei with MYH7 expression, giving a value of 1.35. Within the 12,245 myonuclei in the MyHCII clusters, 93.5% expressed MYH1 and/or MYH2, and a MyHCII false-negative correction factor for MYH1/2 myonuclei of 1.07 was similarly calculated.

After this, we applied a myonuclei-abundance correction factor to account for the number of MyHCI and MyHCII myonuclei in the dataset. There were 6862 MyHCI myonuclei and 12,245 MyHCII myonuclei in the dataset. This is presumably related to the fibre type distribution and/or fibre size in the samples. To account for this, we calculated the myonuclei-abundance correction factor by dividing the number of MyHCII myonuclei by the number of MyHCI myonuclei, giving a value of 1.78.

The raw data were then corrected as follows: (1) the raw number of COL22A1+ MYH7+ nuclei (16) was multiplied by the MyHCI false-negative correction factor (1.35) and the myonuclei-abundance correction factor (1.78), giving a corrected value of 38.6 COL22A1+ MYH7+ nuclei; (2) the raw number of COL22A1+ MYH1/2+ nuclei (150) was multiplied by the MyHCII false-negative correction factor (1.07), giving a corrected value of 160.4 COL22A1+ MYH1/2+ nuclei.

After correction, the proportion of COL22A1+ MYH1/2+ nuclei (160.4 nuclei in the dataset, corrected) versus COL22A1+ MYH7+ nuclei (38.6 nuclei in the dataset, corrected) was 160.4/38.6=4.2.

Spatial transcriptomics

Tendon–MTJ–muscle sections were subjected to spatial transcriptomics analysis using the Visium Spatial Gene Expression kit (10× Genomics). Embedded samples containing both tendon and muscle were cut in a cryostat (10 µm sections) parallel to the tendon as well as in transverse sections such that each capture area contained a sample cut in both directions. The sections were stained with Haematoxylin and Eosin (referred to collectively as H&E; catalogue numbers 860213 and 860049, respectively; locally produced by Region H Apoteket, Reagensafdeling, Bispebjerg Hospital, Bispebjerg Bakke 23, Copenhagen, Denmark) according to the manufacturer’s protocol and then imaged for H&E (brightfield) and autofluorescence (excitation 470 nm, emission 525 nm) using an Axio Scan.Z1 microscope (Zeiss, Birkerød, Denmark). Tissue was digested (15 min) and mRNA converted to a Spatial Gene Expression Library as recommended by the manufacturer's protocol. The libraries were sequenced (NovaSeq; 2×150 bp paired end, ∼700 million reads per library) by a commercial company (GeneWiz, Leipzig, Germany). The reads were aligned to the same transcriptome database as was used for the snRNA-seq and reads were counted using SpaceRanger v1.3.1 (10× Genomics). The counts were analysed using the Seurat v4.1.0 package (Stuart et al., 2019) by first log normalizing each dataset with NormalizeData and then clustering with FindNeighbours and FindClusters based on PCA (30 dimensions). Cell-type prediction scores were generated from the snRNA-seq clusters by using the spacexr/RCTD method (default setting on log-transformed spatial data with doublet_mode='Full’) (version 2.0.0, Cable et al., 2022). As the spatial transcriptomics data includes all cytoplasmic RNA, the snRNA-seq cluster ‘Cytoplasm’ was excluded from the analysis as it would otherwise be the best fit for all spots. Markers for each of the clusters (DEGs) were found using Seurat FindAllMarkers with default settings (wilcox, log2FC>0.25).

Immunofluorescence and microscopy

Tissue cryosections

For immunofluorescence of Tissue-Tek-embedded semitendinosus muscle–tendon samples, 10-µm-thick muscle–tendon sections were cut in the longitudinal plane of the muscle fibres and tendon in a cryostat at −20°C, collected on glass slides and stored at −80°C. For staining, slides were removed from the freezer and dried at room temperature. Detailed information of antibodies, dilution, blocking and fixation is provided in Table S7. All antibodies were diluted in 1% bovine serum albumin (BSA, IgG free) in Tris-buffered saline (TBS; Tris-HCl 0.05 M, sodium chloride 0.154 M, pH 7.4–7.6). In the cases of high background signal, a blocking step was included for 30 min prior to incubation with primary antibodies [5% goat serum (G9023; Sigma-Aldrich), 5% donkey serum (D9663; Sigma-Aldrich) and 5% BSA in TBS]. Fixation with 4% paraformaldehyde (5 min) or Histofix (12 min; 01000; Histolab, Gothenburg, Sweden) was applied either prior to application of primary antibodies (day 1, incubated overnight) or after application of secondary antibodies (day 2, 60 min incubation) – the optimal protocol was tested for each primary antibody. The samples were washed three times for 5 min in TBS between all protocol steps, and samples were finally mounted with cover glasses and DAPI in the mounting medium (Molecular Probes ProLong Gold anti-fade reagent, cat. no. P36931). Images were captured on an Olympus BX51 microscope, controlled by the Olympus cellSens Software (http://www.olympus-lifescience.com), using 10× (0.3 NA) or 20× (0.5 NA) objectives and a 0.5× camera (Olympus DP71, Olympus Deutschland GmbH, Hamburg, Germany). Image size was 4080×3072 pixels: 1747×1315 µm (2.33 pixels/µm) and 868×653 µm (4.70 pixels/µm) for images taken with the 10× and 20× objective respectively. Images were viewed and cropped for presentation in ImageJ (version 1.53c; National Institutes of Health, Bethesda, MD, USA), and colour-blind-friendly pseudocolours were applied to composite images.

Single muscle fibres

Adapting a previous protocol for isolating single muscle fibres (Mackey and Kjaer, 2017b; Ralston et al., 1999), individual myofibres were teased from the gracilis tendon of the fixed sample in 50% glycerol in PBS and incubated in immunobuffer (IB: PBS, 50 mM glycine, 0.25% BSA, 0.03% saponin and 0.05% sodium azide), with 0.1% Triton X-100 (cat. no. 9036-19-5, lot: STBJ9023, Sigma-Aldrich), in a 24-well plate. Primary antibodies were diluted in IB (containing 0.1% Triton X-100) and added to the wells, followed by incubation overnight (two nights for ABLIM1) at room temperature. Fibres were washed and incubated with secondary antibodies, along with Hoechst 33342 (cat. no. H1399, Invitrogen; 1 µg/ml, 1:100 dilution) for 2 h (4 h for ABLIM1). Fibres were washed and aligned in a drop of mounting medium (Molecular Probes Prolong Gold mounting, cat. no. P36930, Invitrogen) on a microscope slide, cover-slipped and then stored at −20°C. Confocal images were acquired with a Zeiss LSM710 (using Zeiss Zen Black 2012 software) using an EC Plan Neofluar 40×1.3 NA Oil DIC M27 objective – image size 4336×4336 pixels, 922×922 µm (4.70 pixels/µm). Hoechst 33342, Alexa Fluor 488, and Alexa Flour 594 were excited by a 405 nm diode laser (30 mW), a 458 nm argon laser (25 mW), and a 561 nm solid-state laser (2 mW), respectively. Images were viewed and cropped for presentation in ImageJ (version 1.53c), and colour-blind-friendly pseudocolours were applied to composite images.

We thank Anja Jokipii-Utzon and Ann-Christina Ronnié Reimann for technical assistance. We acknowledge Irina Korshunova and the University of Copenhagen Biotech Research and Innovation Centre (BRIC) Single-Cell Genomics core facility; Rajesh Somasundaram and the BRIC Flow Cytometry core facility; and the Core Facility for Integrated Microscopy, Faculty of Health and Medical Sciences, University of Copenhagen. We also acknowledge Casey Swoboda of the Millay laboratory at Cincinnati Children's Hospital Medical Center for incorporating our snRNA-seq data into the Myoatlas portal.

Author contributions

Conceptualization: A.K., C.C.Y., P.S., S.S., M. Kjaer, A.L.M.; Methodology: A.K., C.C.Y., P.S., L.D., A.L.M.; Software: P.S.; Formal analysis: A.K., C.C.Y., P.S., L.D.; Investigation: A.K., C.H., A.L.M.; Resources: J.R.J., M.R.K., M. Koch; Data curation: P.S., A.L.M.; Writing - original draft: A.K., P.S., S.S., A.L.M.; Writing - review & editing: C.C.Y.; Visualization: A.K., P.S., A.L.M.; Supervision: A.L.M.; Project administration: A.L.M.; Funding acquisition: A.K., M. Koch, M. Kjaer, A.L.M.

Funding

This work was supported by Nordea-fonden (Center for Healthy Aging Grant to M. Kjaer), Novo Nordisk Fonden (NNF20OC0064829 to A.L.M.), Lundbeckfonden (R303-2018-3427 to A.K.; R344-2020-254 to A.L.M.) and by the Deutsche Forschungsgemeinschaft (FOR2722 to M. Koch).

Data availability

The scRNA-seq and spatial transcriptomics data are deposited in the ArrayExpress database with accession numbers E-MTAB-12529 and E-MTAB-12530, respectively. snRNA-seq data have also been added to Myoatlas (https://research.cchmc.org/myoatlas), an interactive data portal for exploring gene expression at single-nucleus resolution. All other relevant data can be found within the article and its supplementary information.

Barruet
,
E.
,
Garcia
,
S. M.
,
Striedinger
,
K.
,
Wu
,
J.
,
Lee
,
S.
,
Byrnes
,
L.
,
Wong
,
A.
,
Xuefeng
,
S.
,
Tamaki
,
S.
,
Brack
,
A. S.
et al. 
(
2020
).
Functionally heterogeneous human satellite cells identified by single cell RNA sequencing
.
Elife
9
,
e51576
.
Cable
,
D. M.
,
Murray
,
E.
,
Zou
,
L. S.
,
Goeva
,
A.
,
Macosko
,
E. Z.
,
Chen
,
F.
and
Irizarry
,
R. A.
(
2022
).
Robust decomposition of cell type mixtures in spatial transcriptomics
.
Nat. Biotechnol.
40
,
517
-
526
.
Chemello
,
F.
,
Wang
,
Z.
,
Li
,
H.
,
Mcanally
,
J. R.
,
Liu
,
N.
,
Bassel-Duby
,
R.
and
Olson
,
E. N.
(
2020
).
Degenerative and regenerative pathways underlying Duchenne muscular dystrophy revealed by single-nucleus RNA sequencing
.
Proc. Natl. Acad. Sci. USA
117
,
29691
-
29701
.
Chen
,
S.
,
Zhou
,
Y.
,
Chen
,
Y.
and
Gu
,
J.
(
2018
).
fastp: an ultra-fast all-in-one FASTQ preprocessor
.
Bioinformatics
34
,
i884
-
i890
.
Cohn
,
R. D.
,
Henry
,
M. D.
,
Michele
,
D. E.
,
Barresi
,
R.
,
Saito
,
F.
,
Moore
,
S. A.
,
Flanagan
,
J. D.
,
Skwarchuk
,
M. W.
,
Robbins
,
M. E.
,
Mendell
,
J. R.
et al. 
(
2002
).
Disruption of DAG1 in differentiated skeletal muscle reveals a role for dystroglycan in muscle regeneration
.
Cell
110
,
639
-
648
.
Daniloff
,
J. K.
,
Crossin
,
K. L.
,
Pinçon-Raymond
,
M.
,
Murawsky
,
M.
,
Rieger
,
F.
and
Edelman
,
G. M.
(
1989
).
Expression of cytotactin in the normal and regenerating neuromuscular system
.
J. Cell Biol.
108
,
625
-
635
.
De Micheli
,
A. J.
,
Laurilliard
,
E. J.
,
Heinke
,
C. L.
,
Ravichandran
,
H.
,
Fraczek
,
P.
,
Soueid-Baumgarten
,
S.
,
De Vlaminck
,
I.
,
Elemento
,
O.
and
Cosgrove
,
B. D.
(
2020a
).
Single-cell analysis of the muscle stem cell hierarchy identifies heterotypic communication signals involved in skeletal muscle regeneration
.
Cell Rep.
30
,
3583
-
3595.e5
.
De Micheli
,
A. J.
,
Swanson
,
J. B.
,
Disser
,
N. P.
,
Martinez
,
L. M.
,
Walker
,
N. R.
,
Oliver
,
D. J.
,
Cosgrove
,
B. D.
and
Mendias
,
C. L.
(
2020b
).
Single-cell transcriptomic analysis identifies extensive heterogeneity in the cellular composition of mouse Achilles tendons
.
Am. J. Physiol. Cell Physiol.
319
,
C885
-
C894
.
Dos Santos
,
M.
,
Backer
,
S.
,
Saintpierre
,
B.
,
Izac
,
B.
,
Andrieu
,
M.
,
Letourneur
,
F.
,
Relaix
,
F.
,
Sotiropoulos
,
A.
and
Maire
,
P.
(
2020
).
Single-nucleus RNA-seq and FISH identify coordinated transcriptional activity in mammalian myofibers
.
Nat. Commun.
11
,
5102
.
Eraslan
,
G.
,
Drokhlyansky
,
E.
,
Anand
,
S.
,
Fiskin
,
E.
,
Subramanian
,
A.
,
Slyper
,
M.
,
Wang
,
J.
,
Van Wittenberghe
,
N.
,
Rouhana
,
J. M.
,
Waldman
,
J.
et al. 
(
2022
).
Single-nucleus cross-tissue molecular reference maps toward understanding disease gene function
.
Science
376
,
eabl4290
.
Fleming
,
S. J.
,
Chaffin
,
M. D.
,
Arduini
,
A.
,
Akkad
,
A.-D.
,
Banks
,
E.
,
Marioni
,
J. C.
,
Philippakis
,
A. A.
,
Ellinor
,
P. T.
and
Babadi
,
M.
(
2022
).
Unsupervised removal of systematic background noise from droplet-based single-cell experiments using CellBender
.
BioRxiv
.
Grange
,
S.
,
Reurink
,
G.
,
Nguyen
,
A. Q.
,
Riviera-Navarro
,
C.
,
Foschia
,
C.
,
Croisille
,
P.
and
Edouard
,
P.
(
2023
).
Location of hamstring injuries based on magnetic resonance imaging: a systematic review
.
Sports Health
15
,
111
-
123
.
Harvey
,
T.
,
Flamenco
,
S.
and
Fan
,
C.-M.
(
2019
).
A Tppp3+Pdgfra+ tendon stem cell population contributes to regeneration and reveals a shared role for PDGF signalling in regeneration and fibrosis
.
Nat. Cell Biol.
21
,
1490
-
1503
.
He
,
S.
,
Wang
,
L.-H.
,
Liu
,
Y.
,
Li
,
Y.-Q.
,
Chen
,
H.-T.
,
Xu
,
J.-H.
,
Peng
,
W.
,
Lin
,
G.-W.
,
Wei
,
P.-P.
,
Li
,
B.
et al. 
(
2020
).
Single-cell transcriptome profiling of an adult human cell atlas of 15 major organs
.
Genome Biol.
21
,
294
.
Heskamp
,
L.
,
Ogier
,
A.
,
Bendahan
,
D.
and
Heerschap
,
A.
(
2022
).
Whole-muscle fat analysis identifies distal muscle end as disease initiation site in facioscapulohumeral muscular dystrophy
.
Commun. Med.
2
,
155
.
Jakobsen
,
J. R.
,
Jakobsen
,
N. R.
,
Mackey
,
A. L.
,
Koch
,
M.
,
Kjaer
,
M.
and
Krogsgaard
,
M. R.
(
2018
).
Remodeling of muscle fibers approaching the human myotendinous junction
.
Scand. J. Med. Sci. Sports
28
,
1859
-
1865
.
Jakobsen
,
J. R.
,
Schjerling
,
P.
,
Svensson
,
R. B.
,
Buhl
,
R.
,
Carstensen
,
H.
,
Koch
,
M.
,
Rindom Krogsgaard
,
M. R.
,
Kjaer
,
M.
and
Mackey
,
A. L.
(
2021
).
RNA-sequencing and immunofluorescence of the myotendinous junction of mature horses and humans
.
Am. J. Physiol. Cell Physiol
.
321
,
C453
-
C470
.
Karlsen
,
A.
,
Gonzalez-Franquesa
,
A.
,
Jakobsen
,
J. R.
,
Krogsgaard
,
M. R.
,
Koch
,
M.
,
Kjaer
,
M.
,
Schiaffino
,
S.
,
Mackey
,
A. L.
and
Deshmukh
,
A. S.
(
2022
).
The proteomic profile of the human myotendinous junction
.
iScience
25
,
103836
.
Kendal
,
A. R.
,
Layton
,
T.
,
Al-Mossawi
,
H.
,
Appleton
,
L.
,
Dakin
,
S.
,
Brown
,
R.
,
Loizou
,
C.
,
Rogers
,
M.
,
Sharp
,
R.
and
Carr
,
A.
(
2020
).
Multi-omic single cell analysis resolves novel stromal cell populations in healthy and diseased human tendon
.
Sci. Rep.
10
,
13939
.
Kim
,
M.
,
Franke
,
V.
,
Brandt
,
B.
,
Lowenstein
,
E. D.
,
Schöwel
,
V.
,
Spuler
,
S.
,
Akalin
,
A.
and
Birchmeier
,
C.
(
2020
).
Single-nucleus transcriptomics reveals functional compartmentalization in syncytial skeletal muscle cells
.
Nat. Commun.
11
,
6375
.
Knudsen
,
A. B.
,
Larsen
,
M.
,
Mackey
,
A. L.
,
Hjort
,
M.
,
Hansen
,
K. K.
,
Qvortrup
,
K.
,
Kjaer
,
M.
and
Krogsgaard
,
M. R.
(
2015
).
The human myotendinous junction: an ultrastructural and 3D analysis study
.
Scand. J. Med. Sci. Sports
25
,
e116
-
e123
.
Koch
,
M.
,
Schulze
,
J.
,
Hansen
,
U.
,
Ashwodt
,
T.
,
Keene
,
D. R.
,
Brunken
,
W. J.
,
Burgeson
,
R. E.
,
Bruckner
,
P.
and
Bruckner-Tuderman
,
L.
(
2004
).
A novel marker of tissue junctions, Collagen XXII
.
J. Biol. Chem.
279
,
22514
-
22521
.
Kojima
,
H.
,
Sakuma
,
E.
,
Mabuchi
,
Y.
,
Mizutani
,
J.
,
Horiuchi
,
O.
,
Wada
,
I.
,
Horiba
,
M.
,
Yamashita
,
Y.
,
Herbert
,
D. C.
,
Soji
,
T.
et al. 
(
2008
).
Ultrastructural changes at the myotendinous junction induced by exercise
.
J. Orthop. Sci.
13
,
233
-
239
.
Mackey
,
A. L.
and
Kjaer
,
M.
(
2017a
).
Connective tissue regeneration in skeletal muscle after eccentric contraction-induced injury
.
J. Appl. Physiol.
122
,
533
-
540
.
Mackey
,
A. L.
and
Kjaer
,
M.
(
2017b
).
The breaking and making of healthy adult human skeletal muscle in vivo
.
Skelet. Muscle
7
,
24
.
Morin
,
A.
,
Stantzou
,
A.
,
Petrova
,
O. N.
,
Hildyard
,
J.
,
Tensorer
,
T.
,
Matouk
,
M.
,
Petkova
,
M. V.
,
Richard
,
I.
,
Manoliu
,
T.
,
Goyenvalle
,
A.
et al. 
(
2023
).
Dystrophin myonuclear domain restoration governs treatment efficacy in dystrophic muscle
.
Proc. Natl. Acad. Sci. USA
120
,
e2206324120
.
Murach
,
K. A.
,
Liu
,
Z.
,
Jude
,
B.
,
Figueiredo
,
V. C.
,
Wen
,
Y.
,
Khadgi
,
S.
,
Lim
,
S.
,
Morena Da Silva
,
F.
,
Greene
,
N. P.
,
Lanner
,
J. T.
et al. 
(
2022
).
Multi-transcriptome analysis following an acute skeletal muscle growth stimulus yields tools for discerning global and MYC regulatory networks
.
J. Biol. Chem.
298
,
102515
.
Murgia
,
M.
,
Nogara
,
L.
,
Baraldo
,
M.
,
Reggiani
,
C.
,
Mann
,
M.
and
Schiaffino
,
S.
(
2021
).
Protein profile of fiber types in human skeletal muscle: a single-fiber proteomics study
.
Skelet. Muscle
11
,
24
.
Naba
,
A.
,
Clauser
,
K. R.
,
Hoersch
,
S.
,
Liu
,
H.
,
Carr
,
S. A.
and
Hynes
,
R. O.
(
2012
).
The matrisome: in silico definition and in vivo characterization by proteomics of normal and tumor extracellular matrices
.
Mol. Cell. Proteomics
11
,
M111.014647
.
Patton
,
B. L.
,
Miner
,
J. H.
,
Chiu
,
A. Y.
and
Sanes
,
J. R.
(
1997
).
Distribution and function of laminins in the neuromuscular system of developing, adult, and mutant mice
.
J. Cell Biol.
139
,
1507
-
1521
.
Perez
,
K.
,
Ciotlos
,
S.
,
McGirr
,
J.
,
Limbad
,
C.
,
Doi
,
R.
,
Nederveen
,
J. P.
,
Nilsson
,
M. I.
,
Winer
,
D. A.
,
Evans
,
W.
and
Tarnopolsky
,
M.
(
2022
).
Single nuclei profiling identifies cell specific markers of skeletal muscle aging, sarcopenia and senescence
.
Aging
14
,
9393
-
9422
.
Petrany
,
M. J.
,
Swoboda
,
C. O.
,
Sun
,
C.
,
Chetal
,
K.
,
Chen
,
X.
,
Weirauch
,
M. T.
,
Salomonis
,
N.
and
Millay
,
D. P.
(
2020
).
Single-nucleus RNA-seq identifies transcriptional heterogeneity in multinucleated skeletal myofibers
.
Nat. Commun.
11
,
6374
.
Ralston
,
E.
,
Lu
,
Z.
and
Ploug
,
T.
(
1999
).
The organization of the Golgi complex and microtubules in skeletal muscle is fiber type-dependent
.
J. Neurosci.
19
,
10694
-
10705
.
Rieger
,
F.
,
Grumet
,
M.
and
Edelman
,
G. M.
(
1985
).
N-CAM at the vertebrate neuromuscular junction
.
J. Cell Biol.
101
,
285
-
293
.
Riley
,
L. A.
,
Zhang
,
X.
,
Douglas
,
C. M.
,
Mijares
,
J. M.
,
Hammers
,
D. W.
,
Wolff
,
C. A.
,
Wood
,
N. B.
,
Olafson
,
H. R.
,
Du
,
P.
,
Labeit
,
S.
et al. 
(
2022
).
The skeletal muscle circadian clock regulates titin splicing through RBM20
.
Elife
11
,
e76478
.
Rubenstein
,
A. B.
,
Smith
,
G. R.
,
Raue
,
U.
,
Begue
,
G.
,
Minchev
,
K.
,
Ruf-Zamojski
,
F.
,
Nair
,
V. D.
,
Wang
,
X.
,
Zhou
,
L.
,
Zaslavsky
,
E.
et al. 
(
2020
).
Single-cell transcriptional profiles in human skeletal muscle
.
Sci. Rep.
10
,
229
.
Scott
,
R. W.
,
Arostegui
,
M.
,
Schweitzer
,
R.
,
Rossi
,
F. M. V.
and
Underhill
,
T. M.
(
2019
).
Hic1 defines quiescent mesenchymal progenitor subpopulations with distinct functions and fates in skeletal muscle regeneration
.
Cell Stem Cell
25
,
797
-
813.e9
.
Scripture-Adams
,
D. D.
,
Chesmore
,
K. N.
,
Barthélémy
,
F.
,
Wang
,
R. T.
,
Nieves-Rodriguez
,
S.
,
Wang
,
D. W.
,
Mokhonova
,
E. I.
,
Douine
,
E. D.
,
Wan
,
J.
,
Little
,
I.
et al. 
(
2022
).
Single nuclei transcriptomics of muscle reveals intra-muscular cell dynamics linked to dystrophin loss and rescue
.
Commun. Biol.
5
,
989
.
Stuart
,
T.
,
Butler
,
A.
,
Hoffman
,
P.
,
Hafemeister
,
C.
,
Papalexi
,
E.
,
Mauck
,
W. M.
,
Hao
,
Y.
,
Stoeckius
,
M.
,
Smibert
,
P.
and
Satija
,
R.
(
2019
).
Comprehensive integration of single-cell data
.
Cell
177
,
1888
-
1902.e21
.
Tidball
,
J. G.
(
1991
).
Force transmission across muscle cell membranes
.
J. Biomech.
24
,
43
-
52
.
Tidball
,
J. G.
(
1992
).
Desmin at myotendinous junctions
.
Exp. Cell Res.
199
,
206
-
212
.
Tidball
,
J. G.
,
O'halloran
,
T.
and
Burridge
,
K.
(
1986
).
Talin at myotendinous junctions
.
J. Cell Biol.
103
,
1465
-
1472
.
Tucker
,
N. R.
,
Chaffin
,
M.
,
Fleming
,
S. J.
,
Hall
,
A. W.
,
Parsons
,
V. A.
,
Bedi
,
K. C.
,
Akkad
,
A.-D.
,
Herndon
,
C. N.
,
Arduini
,
A.
,
Papangeli
,
I.
et al. 
(
2020
).
Transcriptional and cellular diversity of the human heart
.
Circulation
142
,
466
-
482
.
Vracko
,
R.
and
Benditt
,
E. P.
(
1972
).
Basal lamina: the scaffold for orderly cell replacement. Observations on regeneration of injured skeletal muscle fibers and capillaries
.
J. Cell Biol.
55
,
406
-
419
.
Welcker
,
D.
,
Stein
,
C.
,
Feitosa
,
N. M.
,
Armistead
,
J.
,
Zhang
,
J.-L.
,
Lütke
,
S.
,
Kleinridders
,
A.
,
Brüning
,
J. C.
,
Eming
,
S. A.
,
Sengle
,
G.
et al. 
(
2021
).
Hemicentin-1 is an essential extracellular matrix component of the dermal–epidermal and myotendinous junctions
.
Sci. Rep.
11
,
17926
.
Wen
,
Y.
,
Englund
,
D. A.
,
Peck
,
B. D.
,
Murach
,
K. A.
,
Mccarthy
,
J. J.
and
Peterson
,
C. A.
(
2021
).
Myonuclear transcriptional dynamics in response to exercise following satellite cell depletion
.
iScience
24
,
102838
.
Wolock
,
S. L.
,
Lopez
,
R.
and
Klein
,
A. M.
(
2019
).
Scrublet: computational identification of cell doublets in single-cell transcriptomic data
.
Cell Syst.
8
,
281
-
291.e9
.
Yan
,
R.
,
Zhang
,
H.
,
Ma
,
Y.
,
Lin
,
R.
,
Zhou
,
B.
,
Zhang
,
T.
,
Fan
,
C.
,
Zhang
,
Y.
,
Wang
,
Z.
,
Fang
,
T.
et al. 
(
2022
).
Discovery of muscle-tendon progenitor subpopulation in human myotendinous junction at single-cell resolution
.
Research
2022
,
9760390
.
Yeung
,
C.-Y. C.
,
Schoof
,
E. M.
,
Tamáš
,
M.
,
Mackey
,
A. L.
and
Kjaer
,
M.
(
2020
).
Proteomics identifies differences in fibrotic potential of extracellular vesicles from human tendon and muscle fibroblasts
.
Cell Commun. Signal.
18
,
177
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information