Human muscle stem/progenitor cells are bi-potential adult stem cells

Tendon injury occurs at high frequency and is dicult to repair. Identication human stem cells being able to regenerate tendon will greatly facilitate the development of regenerative medicine for tendon injury. We identied human muscle stem/progenitor cells having tendon differentiation potential both in vitro and in vivo. Interestingly, the tendon differentiation potential is not present in mouse muscle stem cells. These ndings reveal that human muscle stem cells are bi-potential adult stem cells and can serve as a new source for tendon regeneration. and 1% Penicillin-Streptomycin were used to differentiate mouse muscle stem cells as 22,25 . DMEM containing 10% FBS, 100ng/ml GDF7 (R&D, cat#8386-G7-050), 0.2mM ascorbic acid (Sigma-Aldrich, cat#A4403) and 1% Penicillin-Streptomycin were used to induce tendon differentiation as described previously 26,27 . Primary tenocytes were cultured in DMEM containing 20% FBS, 2.5ng/ml bFGF (R&D, cat#233-FB-025) and 1% Penicillin-Streptomycin.


Introduction
Tendons is a tissue transmitting forces from skeletal muscle to bone to generate active movement. Tendonpathy affects more than 10% of the population under 45 and compromises the tendon functions 1,2 . Tendon injury healing is slow and incomplete due to the low number of cells in tendon and the hypovascular and anaerobic environment 3,4 . Tendon stem/progenitor cells (TDSCs) are a cell population derived from tendon and considered to be a subgroup of mesenchymal stem cells that have abilities to improve tendon injury healing 5,6 . However, the number of TDSCs in tendon is low and retrieving TDSCs is invasive and exacerbates tendon injury. Morphology and proliferation ability loss in the culture system also hampers the efforts to obtain su cient amounts of active TDSCs. To nd more cell types supporting tendon regeneration will facilitate the development of regenerative medicine to treat tendon injury.
Skeletal muscles directly connect to tendon and has strong regeneration ability due to the existence of unipotency muscle stem cells [7][8][9][10][11] . Functional muscle stem cells can be expanded in vitro to provide large amount of muscle stem cells for transplantation [12][13][14] . These features empower muscle stem cells to be a promising cell source for injury reparation.
Due to the different motion patterns, rodent tendon have different features from human 15 . Though rodent models have been widely used to study tendon regeneration, species related differences on regeneration have not been fully characterized. Here we found that human muscle stem/progenitor cells can be robustly differentiated to tendon cells both in vitro and in vivo. Transplantation of human muscle stem/progenitor cells to injured tendon improved the tendon regeneration, suggesting that human muscle stem/progenitor cells are bipotential adult stem cells. Interestingly, the tendon differentiation potential in mouse muscle stem cells is minimal and the differences on the cell intrinsic signaling network attributes to the species difference.

Results
It is a common clinical practice to harvest the whole peroneus longus tendon for the reconstruction of anterior cruciate ligament. The surgery completely removed the peroneus longus tendon. Surprisingly, the peroneus longus tendon regenerated well based on MRI images after 24 months (Fig. 1a) 16 . Another interesting clinical observation is that the removal of peroneus longus tendon is always accompanied by muscle injury (SFig. 1a). It seems that more severe muscle injury correlated with better tendon regeneration. Since muscle injury activates muscle stem cells, we therefore hypothesis that muscle stem cells which are activated in muscle injury caused by tendon removal could contribute to the tendon regeneration observed in clinics.
We rst analyzed the differentiation potential of human muscle stem/progenitor cells using the single cell sequencing data obtained from human muscle samples. The trajectory analysis revealed that Pax7 + muscle stem/progenitor cells displayed differentiation potentials towards two lineages. One is towards differentiated muscle cells, the other is towards tendon differentiation ( Fig. 1b and c). Scx, a key transcription factor determining tendon lineage, was expressed in muscle stem/progenitor cells as indicated by single cell sequencing data (Fig. 1c) and con rmed by RT-qPCR and RNA FISH ( Fig. 1d and e).
To further test this hypothesis, we next isolated human muscle stem/progenitor cells from human muscle biopsies by FACS sorting to isolate CD45-CD31-CD29 + CD56 + CXCR4 + cells as reported previously 17 . The majority of the FACS sorted cells showed high Pax7 expression that is the marker of muscle stem/progenitor cells ( Fig. 1f and SFig. 1b). As expected, the isolated human muscle stem/progenitor cells were differentiated to myotubes robustly (Fig. 1g). Consistently, the expression levels of genes related to muscle stem/progenitor cells like Pax7, Myf5, and MyoD were down-regulated after differentiation; while the myogenic differentiation related genes such as MyoG, Myh1, and Myh3 were up-regulated after differentiation ( Fig. 1h and i).
The isolated primary human muscle stem/progenitor cells were then induced to differentiate to tenocytes by GDF7 and ascorbic acid. After 12 days of tendon differentiation, the expression of tendon marker genes such as TNC and Scx was increased. In contrast, the expression of genes related to muscular differentiation was minimal and no enrichment was observed (Fig. 2a,b,c,and SFig. 2). Other genes expressed in more mature tendons, such as Col I, Mkx, Thbs4, and Comp, were also enriched after tendon differentiation induction (Fig. 2d). These results combined suggest that human muscle stem/progenitor cells are capable of tendon differentiation in vitro.
RNA sequencing was then performed to further determine the lineage of the differentiated cells. Myotubes and tendon differentiated from human muscle stem/progenitor cells displayed distinct expression patterns (Fig. 3a).
The differentiated tendon cells share high similarity to fresh isolated tenocytes from human tendons (Fig. 3a), suggesting that human muscle stem/progenitor cells are capable of tendon differentiation. Consistently, two distinct sets of genes were up-regulated in muscle stem/progenitor cells after myogenic induction and tenogenic induction, respectively ( Fig. 3b and c). Further GO analysis also displayed the activation of two distinct sets of cell features. Upon the skeletal muscle differentiation induction condition, terms related to skeletal muscle functions such as skeletal muscle thin lament assembly, skeletal muscle contraction, muscle organ development, and sarcomere organization were enriched (Fig. 3d), suggesting the muscular identity of the differentiated cells. In sharp contrast, terms related to tendon formation, tendon development, and tendon cell differentiation were enriched (Fig. 3e), suggesting that tendon identity is achieved after tenogenic induction of human muscle stem/progenitor cells. Together, these results suggest that human muscle stem/progenitor cells are capable of both myogenic and tenogenic differentiation in vitro.
We then investigated whether muscle stem cells have tendon differentiation potential and participate in tendon repair in vivo. Due to the di culty for lineage tracing in human, we performed the lineage tracing experiments in mice. We rst crossed Pax7CreERT2 to ox-Stop-ox-TdTomato mice to trace muscle stem cells and established a tendon injury mouse models mimicking the peroneus longus tendon removal surgery in human (SFig. 3). Mice were given 4 months to recover from the injury, respectively. Nevertheless, less than 0.5% tendon cells originated from mouse muscle stem cells were observed even after we induced muscle injury by injecting cardiotoxin (CTX) at the close proximate to the tendon injury site (SFig. 4a-c). These results suggest that murine muscle stem cells have poor tendon differentiation potentials. We thus isolated mouse muscle stem cells as reported previously to test their tendon differentiation ability in vitro. In sharp contrast to human muscle stem/progenitor cells, murine muscle stem cells failed to be induced to tendon cells upon the same induction condition as that for human muscle stem/progenitor cells (SFig. 4d and f), even though they displayed e cient myogenic differentiation (SFig. 4e). These results further indicate that murine muscle stem cells have poor tendon differentiation potential.
Due to the di culties in mouse lineage tracing, we set out to perform cell transplantation experiments to test whether ectopic muscle stem cells have tenogenic potential after being planted in tendon. We transplanted 50,000 Pax7 + human muscle stem/progenitor cells to the pre-injured tendon in SCID mice and harvested the tendon detecting the tendon cells originated from human cells (Fig. 4a). Two months after transplantation, the repaired tendon was harvested. The length of the tendon was obviously extended after transplantation of human stem/progenitor cells ( Fig. 4b and c). The presence of tendon cells with human origin was detected by immuno uorescence staining with antibody speci c for human laminA/C. At the PBS injection side, no human Lamin A/C signal was detected con rming the speci city of the antibody. In sharp contrast, human cells were detected after human muscle stem/progenitor cell transplantation ( Fig. 4d and e). The majority of the human cells co-stained with tendon marker TNC, Scx, and Tnmd ( Fig. 4e and f), suggesting that the transplanted human muscle stem/progenitor cells mainly differentiate to tendon cells in vivo when transplanted to the tendon. Only few transplanted human muscle stem/progenitor cells differentiated to MyHC + muscle cells ( Fig. 4d and g).
These results reveal that human muscle stem/progenitor cells have tendon differentiation potential in vivo. While when the same number of human muscle stem/progenitor cells were transplanted to pre-injured muscle, the same cell population differentiated predominantly to skeletal muscle (SFig. 5a-c), con rming their muscle stem/progenitor cell identities. Taken together, these results suggest that human muscle stem/progenitor cells are bi-potential adult stem cells and the differentiation potential is highly microenvironment dependent.
In sharp contrast to human muscle stem/progenitor cells, when 50,000 murine muscle stem cells constitutively expressing tdTomato were transplanted to pre-injured tendon in SCID mice, very low number of tdTomato + tendon cells were detected, though these cells can regenerate skeletal muscle e ciently (SFig. 6a-e). These results suggest that human muscle stem/progenitor cells have unique tenogenic differentiation potential both in vitro and in vivo.
We further examined the function of the regenerated tendons after human muscle stem/progenitor cell transplantation. The endurance time of the mice transplanted with human muscle stem/progenitor cells was longer than that of the mice with PBS injection while doing long distance run training (Fig. 4h). Consistently, the mice transplanted by muscle stem/progenitor cells can achieve higher maximum speed in running exercise ( Fig. 4i) and longer fatigue time (Fig. 4j). All together, these results suggest that the transplanted human muscle stem/progenitor cells can improve the functions of injured tendons by differentiate to mature tendon cells.
The above results revealed the cell intrinsic species difference on the differentiation potentials of muscle stem/progenitor cells between human and mouse. We further compared the expression pro les of mouse and human muscle stem/progenitor cells. Interestingly, Notch signaling were signi cantly enriched in human muscle stem/progenitor cells (Fig. 4k), indicating that Notch signaling could be the key node for maintaining the tendon differentiation potential.
Muscle stem cells have long been considered to be the typical unipotent adult stem cells. Here we show that the differentiation potential of muscle stem cells is species dependent. Human muscle stem/progenitor cells are bipotent adult stem cell with both muscle and tendon differentiation potentials and could serve as new seed cell type to treat tendon injury. Murine muscle stem cells barely have tendon differentiation potential. The species difference might be due to the higher Notch signaling level in human muscle stem/progenitor cells.
Notch signaling is critical for murine muscle stem cell quiescence maintenance by inhibiting differentiation 18-20 .
The elevated Notch signaling level in human muscle stem/progenitor cells may help maintaining a more naïve stage and enable the cells with more potentials to differentiate to multiple cell types. It will be interesting to explore the function of Notch signaling in human muscle stem/progenitor cells.
Studies using mouse models contributed to the majority of our knowledge about muscle stem cells and laid the foundation for our understanding of mammalian muscle stem cells. However, human is dramatically different from mouse in many aspects such as size, life span, and manners of motions. It is more demanding for humans to maintain the homeostasis of motion system with longer life span and bigger body size. Evolutionally, to evolve more differentiation potentials for a designated type of adult stem cells represents an economic aspects to maintain the longer life span and more complicated motion manner in human beings.

Human samples
Peroneal longus muscle and remanent tendon were obtained from the wastes of patients who underwent fullthickness peroneal longus tendon autograft for knee ligament reconstructive procedures. The study was approved by the ethical committee of Xinhua Hospital A liated to Shanghai Jiao Tong University, School of Medicine (Approval No. XHEC-D-2019-043) and written informed consents were obtained from all donors.

Animals
Animal care and use were in accordance with the guidelines of the animal facility hosted by Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, and the operations were approved by the ethical committee of Shanghai Institute of Biochemistry and Cell Biology. All mice were maintained in speci c pathogenfree (SPF) animal facility in individually ventilated cages (IVC) with controlled temperature (22±1℃) and light (12h light/dark cycle). SCID mice were purchased from Animal Model Research Center of Nanjing University. Pax7CreERT2 and Rosa26-Flox-Stop-Flox-tdTomato mice were purchased from Jackson Laboratory.
Two years after peroneus longus tendon harvest, bilateral legs were scanned with a 3.0-Tesla MRI (Ingenia; Philips Healthcare, Eindhoven, Netherlands). The proton density-weighted and spectral attenuated inversion recovery (PDW-SPAIR) fat suppressed images were obtained in the axial planes. Imaging was performed from the distal end of bula to the proximal cruris and 100 slices were obtained for each leg.

Primary tenocytes isolation
Tendon tissues obtained from tendon autograft surgery were washed with PBS. Epi-and peri-tendon sheath were completely removed. Tenocytes were isolated as described previously 5 . Brie y, the tendons were minced to 1mm 3 pieces and digested with 2 mg/ml collagenase II (Worthington biochemical, cat#LS004177) in DMEM (Gibco, cat#11965118) at 37℃ for 3hrs with gentle agitation. The digested tissue was ltered through a 40µm cell strainer (BD Falcon, cat#352340) and the isolated cells were plated in for subsequent analysis.

Cell transplantation in TA muscle
A single dose of 18 Grey irradiation was administered to the hind legs of the recipient SCID mice. TA muscle was injured by injecting 15μl of 10μM CTX (Sigma), and 50,000 human muscle stem/progenitor cells or 50,000 mouse muscle stem cells suspended in 10μl PBS were injected intramuscularly to the injury site as described 22,24 .
Tendon injury and cell transplantation A 1.5mm long and 0.3mm width piece was removed from the medial gastrocnemius tendon beginning from myotendinous junction in recipient SCID mice. 50,000 human muscle stem/progenitor or mouse muscle stem cells resuspended in 20μl Matrigel (BD Biosciences, cat#354277) were injected to the injury site with 28-gauge needles. PBS mixed with 20μl Matrigel were injected as control.

RNA-sequencing
Total RNA was isolated using TRIzol Reagent (Invitrogen, cat#15596-018) according to the manufacturer's instruction. mRNA was enriched with magnetic oligo (dT) beads (New England Biolabs, cat#S1419S). The cDNA library was constructed with mean inserts of 200bp with non-stranded library preparation using NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, cat#E7530L). Sequencing was performed by a pairedend 125 cycles rapid run on the Illumina HiSeq2500. Sequencing data were ltered by SolexaAQ (Q > 20 and length ≥ 25 bp) 28 .The adapter sequences and low quality segments (Phred Quality Score<20) were trimmed using Cutadapt 29 . Paired-end clean reads were then mapped to the reference genome GRCh38.98 using HISAT2 30 . Htseq-count was used to quantify the gene expression value 31 . Read count of each gene was normalized using FPKM (Fragments Per Kilo bases per Million fragments). Differential expression (DE) analysis were performed using DESeq and signi cant DE genes were de ned as those with absolute log2FoldChange>1 and p< 0.05. Heatmap and volcano plot of DE genes was generated by R package Pheatmap (https://cran.rproject.org/web/packages/pheatmap/index.html) and EnhancedVolcano (https://bioconductor.org/packages/release/bioc/vignettes/EnhancedVolcano/inst/doc/EnhancedVolcano.html), respectively. Gene enrichment analysis was conducted by R package topGO (Aibar et al., 2015) with p < 0.05 as the cut-off.

Trajectory analysis
Single cell RNA-seq data were analyzed by Seurat R (Version 3.2.0) package 32 . Cells with less than 200 genes, more than 6,000 genes detected, and more than 10% mitochondrial genes were excluded. Cells expressing Pax7, Scx, Tnmd, Mkx, Myh1, Myh3, Myh5 or Myl1 were selected for cell clustering and trajectory analysis. Sequencing reads for each gene were normalized to total UMIs in each cell to obtain normalized UMI values by "NormalizeData" function. The "ScaleData" function was used to scale and center expression levels in the data set for dimensional reduction. Total cell clustering was performed by "FindClusters" function at a resolution of 0.1 and dimensionality reduction was performed with "RunUMAP" function. Trajectory analysis was done by R package Monocle2 33 . The top 100 differentially expressed genes were identi ed to sort the cells in pseudo-time order, and cells were plotted along the inferred trajectory. Muscle stem cells were selected as the start of pseudotime and aligned via the "ordercells" function. 'UMAP' was applied to reduce dimensions and the visualization functions "plot_cell_trajectory" were used to plot each group along the same pseudotime trajectory.

Statistical analysis
The number of biological replicates and technical repeats in each experimental group are indicated in gure legends. Error bars indicated standard deviation. Two-tailed unpaired Student's t-tests were used when variances were similar (tested with F-test) for comparison paired comparation. One-way ANOVA tests followed by Dunnett's post-test or Tukey's post-test were used for multiple comparisons. Shapiro-Wilk tests were performed to determine data normality. Statistical analysis was performed in GraphPad Prism 7 (GraphPad Software, San Diego, USA) or SPSS version 19.0 for Windows (SPSS Inc., Chicago, IL, USA).
The normality of continuous data was determined with the. Independent sample Student t test was used to compare parameters between experiment and control group. It was considered signi cant with P value less than 0.05. Data were presented as mean ± S.E.M. unless stated.

Accession numbers
The complete mRNA-seq data used to produce intensity maps have been uploaded on to Sequence Read Archive (SRA) database. The Accession number is PRJNA721361.  Immunoblotting was performed with total proteins extracted from human muscle stem/progenitor cells before and after myogenic and tenogenic differentiation. Error bars indicated standard error and based on 4 independent experiments. GAPDH served as internal control for loading.

Figure 3
Human muscle stem/progenitor cells have tendon differentiation potential. a. Heat map of gene expression pro les of human muscle stem/progenitor cells, human muscle stem/progenitor cells after myogenic differentiation, human muscle stem/progenitor cells after tenogenic differentiation, and primary tenocytes. b.
Volcano plot of genes enriched in myogenic differentiation of human muscle stem/progenitor cells. c. Volcano plot of genes enriched in tenogenic differentiation of human muscle stem/progenitor cells. d. Bubble chart of GO analysis of cellular process up-regulated in myogenic differentiation of human muscle stem/progenitor cells. e.

Page 17/18
Bubble chart of GO analysis of cellular process up-regulated in tenogenic differentiation of human muscle stem/progenitor cells.

Figure 4
Human muscle stem/progenitor cells displayed tendon differentiation potential in vivo. a. Schem of human muscle stem/progenitor cell transplantation. b. Regenerated tendon after human muscle stem/progenitor cell transplantation. Tendon injury was induced in recipient SCID mice and 50,000 human muscle stem/progenitor cells were transplanted. Tendon was completely dissected out 2 months after human muscle stem/progenitor cell transplantation. Scale bars, 1cm. c. Statistic analysis of the length of the regenerated tendons after human muscle stem/progenitor cell transplantation. Error bars indicated standard error based on 10 independent experiments. *** indicated p<0.005. d. Immuno uorescence staining of MyHC and human Lamin A/C. The regenerated tendon and the connected skeletal muscle were dissected out and subjected for cryosection preparation. The cryosections were stained for MyHC which is speci cally expressed in skeletal muscle and human Lamin A/C which is speci cally expressed in human cells. Scale bars, 100µm. e. Immuno uorescence staining of TNC and human Lamin A/C. The cryosections generated from the regenerated tendon and the connected skeletal muscle were stained for TNC which is enriched in tendon and human Lamin A/C which is speci cally expressed in human cells. Scale bars, 100 µm. f. Immuno uorescence staining of Scx, Tnmd, and human Lamin A/C in cryosections derived from the regenerated tendon and the connected skeletal muscle 2 months after human muscle stem/progenitor cell transplantation. Scx and Tnmd were enriched in tendon cells. Bubble chart of differently activated biological pathways between human and mouse muscle stem cells.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.