Prostaglandin EP2 receptor downstream of Notch signaling inhibits differentiation of human skeletal muscle progenitors in differentiation conditions

Understanding the signaling pathways that regulate proliferation and differentiation of muscle progenitors is essential for successful cell transplantation for treatment of Duchenne muscular dystrophy. Here, we report that a γ-secretase inhibitor, DAPT (N-[N-(3,5-difluorophenacetyl-L-alanyl)]-S-phenylglycine tertial butyl ester), which inhibits the release of NICD (Notch intercellular domain), promotes the fusion of human muscle progenitors in vitro and improves their engraftment in the tibialis anterior muscle of immune-deficient mice. Gene expression analysis revealed that DAPT severely down-regulates PTGER2, which encodes prostaglandin (PG) E2 receptor 2 (EP2), in human muscle progenitors in the differentiation condition. Functional analysis suggested that Notch signaling inhibits differentiation and promotes self-renewal of human muscle progenitors via PGE2/EP2 signaling in a cAMP/PKA-independent manner. Fusako Sakai-Takemura et al. report that self-renewal of human muscle progenitors is regulated by NOTCH/EP2 signaling. The findings deepen our understanding on self-renewal of muscle stem cells and contribute to the development of treatments for Duchenne muscular dystrophy.

D uchenne muscular dystrophy (DMD) is a devastating muscle disease caused by mutations of the DMD gene, which encodes dystrophin. Currently there is no effective treatment for DMD 1 . Transplantation of muscle progenitors/ precursors is a therapeutic strategy for DMD 2 . However, clinical trials of myoblast transfer in the 1990s were all unsuccessful. Experiments using mouse models suggested that the majority of transplanted myoblasts were lost immediately after transplantation [3][4][5] .
Human induced pluripotent stem cells (hiPSCs) can be induced to differentiate into skeletal muscle cells even after extensive expansion [6][7][8][9][10] . Therefore, hiPS cells are expected to provide sufficient amounts of muscle progenitors for cell therapy. Recently, we reported an improved sphere culture-based protocol for induction of muscle progenitors from hiPSCs 10 . Induced muscle progenitors efficiently formed multinucleated myotubes in vitro and differentiated into myofibers in immune-deficient dystrophin-deficient mdx mice. However, the number of dystrophin-positive myofibers in mdx muscle was not satisfactory 10 , requiring further investigation to clarify why myogenic cells, which differentiate efficiently into myotubes in vitro, do not form myofibers in vivo after engraftment.
Notch is a key regulator of myogenesis during development and postnatal life [11][12][13][14][15] . Recently, Low et al. reported that Dll4 activates Notch3 to regulate self-renewal in mouse C2C12 cells and mouse primary myoblasts 16 . Baghdadi et al. revealed that Notch keeps the satellite cells in their niche partly via collagen Vcalcitonin receptor signaling 17 . These reports using mouse models emphasize again that Notch is indispensable for generation and maintenance of muscle satellite cells. On the other hand, the effects of Notch activation on engraftment remain controversial. Parker et al. reported that activation of Notch signaling during ex vivo expansion enhanced the efficiency of engraftment in a canine-to-murine xenotransplantation model 18 . In contrast, Sakai et al. reported that mouse muscle stem cells and human myoblasts treated with Notch ligands in vitro restored PAX7 expression but did not improve regeneration capacity after transplantation into mice 19 .

Results
A Notch inhibitor, DAPT, promoted myotube formation by human muscle progenitors. First, to explicate the effects of Notch signaling on differentiation of human muscle progenitors, we added DAPT, which specifically inhibits the γ-secretase complex and, as a result, blocks Notch signaling (Fig. 1a), to the cultures of human muscle progenitors. DAPT increased both the fusion index and myotube diameter of Hu5/KD3 cells, a human muscle progenitor cell line 20 (Fig. 1b-e), hiPS-derived myogenic cells (Fig. 1f-i), and adult human primary myoblasts (Supplementary Fig. 1), suggesting that Notch inhibition stimulated the recruitment of hiPS-derived muscle progenitors and postnatal myogenic cells, which otherwise do not fuse, to fusion.
DAPT improved engraftment of human muscle progenitors. Next, we tested whether DAPT improves engraftment of human muscle progenitors by promoting differentiation of engrafted cells (Fig. 2). DAPT was added to a suspension of Hu5/KD3 cells just before transplantation into pre-injured tibialis anterior (TA) muscles of immunodeficient NOD/Scid mice. Interestingly, DAPT improved the efficiency of cell transplantation of Hu5/ KD3 cells (Fig. 2a-c). We then tested whether DAPT improved the efficiency of cell transplantation of hiPSC-derived muscle progenitors. DAPT was added to the cell suspension just before direct injection into the TA muscle of NSG-mdx 4CV mice, then injected into the engrafted TA muscle four times with 2-day intervals (Fig. 2d). DAPT treatment increased the numbers of human lamin A/C-positive dystrophin-positive myofibers ( Fig. 2d-f).
Identification of Notch signal-responsive genes in human muscle progenitors in differentiation conditions. To clarify the Notch target genes that inhibit or augment myotube formation, we examined the gene expression in Hu5/KD3 human myoblasts treated with DAPT for 4 days using RNA-seq analysis (Fig. 3a, b). We found that relatively limited numbers of the genes were up-(60 genes) or downregulated (67 genes) more than twofold by DAPT (Fig. 3b). In addition to protein-coding mRNA, 14 noncoding RNAs were found to be differentially expressed (Fig. 3b). We list the 10 most upregulated and the 10 most downregulated genes after DAPT treatment in Table 1. NOTCH3 and two wellknown Notch target genes, HES1 and HEY1, were listed as the most downregulated genes, confirming that DAPT successfully inhibited Notch signaling. In the RNA-seq analysis, NOTCH4 mRNA was extremely low in both groups. NOTCH1 expression was also low, and it was not up-or downregulated with DAPT treatment (Supplementary DATA). Among Notch ligands, only JAG1 was differentially expressed (downregulated) by DAPT treatment (Supplementary DATA).
Except LINC00948 (Linc-RAM) and IGFN1, most of the genes listed in Table 1 as "upregulated" have not been reported to be involved in skeletal muscle differentiation. Linc-RAM is a recently identified long non-coding RNA, which codes a SERCA activity-regulating small molecule, MYOREGULIN. Full-length Linc-RAM, but not Myoregulin, is reported to enhance myogenic differentiation in mice 21 . IGFN1 is also reported to be required for fusion of C2C12 myoblasts 22 .
Next, by qPCR, we confirmed the downregulation of NOTCH3, HEY1, HEYL, PTGER2(EP2), COL6A3, APOE, CMKLR1, UNC5B, and SCG2, and upregulation of ID1 in Hu5/KD3 cells treated with DAPT ( Fig. 3c, Supplementary Fig. 2). There was no significant difference between the expression levels of NOTCH2 in DAPTand non-treated cells at day 4. The expression of NOTCH3 was very low at day 0 and gradually increased in control cells, but the increase was not observed in DAPT-treated cells. The expression of PTGER2 was transiently increased at day 4. In contrast, the expression of PTGER2 was gradually decreased in DAPT-treated cells (Fig. 3c). The same expression patterns were observed in human primary myoblasts ( Supplementary Fig. 3).
NOTCH3 and EP2 were highly expressed in self-renewing human muscle progenitors. Next, we examined NOTCH3 expression in differentiation of Hu5/KD3 cells by FACS. NOTCH3 expression was detected in a fraction of the cells in high cell-density culture, but not on proliferating cells cultured at a low density ( Supplementary Fig. 4b). The induction was completely abolished by DAPT treatment (Supplementary Fig. 4c). To clarify the function of NOTCH3, NOTCH3-high, and NOTCH3negative cells were sorted by FACS (Fig. 4b). Then, total RNA was extracted from halves of these two fractions to perform qPCR. The other halves of the cells were plated onto collagen-I-coated plates at nearly confluency.
Upregulation of EP2 blocked the differentiation of Hu5/KD3 cells ( Fig. 6a-d, Supplementary Fig. 5d-f). In contrast, knockdown of EP2 improved fusion of muscle progenitors ( Fig. 6e-i). These results suggest that the PGE2/EP2 signal inhibits differentiation of human muscle progenitors in differentiation conditions.
Blockage of EP2 signaling promoted differentiation of iPSCderived myogenic cells. Next, we examined the effects of activation and blockage of EP2 receptor on hiPSC-derived muscle progenitors (Fig. 7). As expected, DAPT and TG6-10-1, an antagonist of EP2, greatly improved the fusion of the cells to the same degree. Unexpectedly, PGE2 and butaprost, an agonist of EP2, had no significant effects on the fusion index (Fig. 7), probably because differentiation of hiPSC-derived myogenic cells is severely suppressed by TGF-β signaling, as previously reported 10 . DAPT slightly suppressed the proliferation of hiPSCderived muscle progenitors, but there was no significant difference in cell numbers among the other groups (Fig. 7d).
Contribution of EP4 to self-renewal of human myogenic cells. Prostaglandin E2 has four seven-transmembrane G proteincoupled receptors (GCPRs), EP1-4. Among them, EP2 and EP4 activate adenyl cyclase, but the two receptors have different structures and functions. Therefore, we tested the effects of an EP4 antagonist, ONO-AE3-208, on the differentiation of myogenic cells. ONO-AE3-208 did not increase the fusion index of Hu5/KD3 myogenic cells (Fig. 8d), suggesting that the contribution of EP4 to self-renewal of myogenic progenitors is small.
PGE2 produced by COX-2 activated EP2 signaling. Next, we examined the upstream of EP2 using COX inhibitors. Indomethacin, which inhibits both COX-1 and COX-2, improved the fusion of Hu5/KD3 cells and hiPSC-derived myogenic progenitors (Fig. 8b, Supplementary Fig. 6d). COX-1-selective SC-560 did not promote differentiation of Hu5/KD3 cells and hiPSC-derived muscle progenitors (Fig. 8c, Supplementary Fig. 6f). In contrast, the COX-2-selective valdecoxib promoted the differentiation of Hu5/KD3 cells and hiPSC-derived muscle progenitors (Fig. 8c, Supplementary Fig. 6e), suggesting that COX-2 is involved mainly in EP2-mediated suppression of muscle differentiation. We measured the concentration of PGE2 by an enzyme-linked immunosorbent assay (ELISA). Unexpectedly, the PGE2 level was quite low in the culture medium of hiPSCs-derived myogenic  RelaƟve mRNA expression  Relative mRNA expression cells and Hu5/KD3 cells, and their concentration was not significantly changed by DAPT treatment (Supplementary Fig. 7). The expression level of COX-2 mRNA in Hu5/KD3 cells was also extremely low when examined by RT-qPCR ( Supplementary  Fig. 7d). Interestingly, PGE2 levels rapidly increased in the later stage of differentiation ( Supplementary Fig. 7e). These results and EP2 overexpression experiments (Fig. 6a-d, Supplementary  Fig. 5d-f) suggest that the PGE2-EP2 receptor signaling is mainly regulated by the expression levels of the receptor rather than the PGE2 concentration in the early phase of differentiation. To understand the roles for PGE2 in the later stage of muscle differentiation, which is likely produced by multinucleated myotubes, further analysis needs to be performed. signaling of EP2 receptor, we examined the effects of forskolin (adenylyl cyclase activator), cell membrane-permeable dibutyryl cyclic AMP (dbcAMP), which mimics endogenous cyclic adenosine monophosphate (cAMP), H-89, a PKA inhibitor, and ESI-09 (EPAC inhibitor), on the differentiation of Hu5/KD3 myogenic cells and hiPSC-derived myogenic cells. Unexpectedly, forskolin did not suppress the differentiation of muscle progenitors (Fig. 8e). Contrarily, dbcAMP stimulated the differentiation of muscle progenitors (Fig. 8f, Supplementary  Fig. 6g). H-89 and ESI-09 did not improve the differentiation of muscle progenitors (Fig. 8g, h, Supplementary Fig. 6h). 8-CPT-2Me-cAMP (selective activator of Epac) and 8-bromo-cAMP (selective activator of protein kinase A) had no significant effects on differentiation of myogenic cells (Supplementary Fig. 8a, b)  CAY10684-treated myogenic cells fused well to form multinucleated myotubes (Fig. 8i-l). The results suggest that activation of the cAMP-PKA pathway does not promote self-renewal of myogenic progenitors.

Discussion
DAPT improved myotube formation by a human muscle progenitor cell line, Hu5/KD3 cells, human primary myoblasts, and hiPSC-derived muscle progenitors. Gene expression analysis of DAPT-treated cells revealed that DAPT treatment downregulated NOTCH3 and Notch effector genes, HES1, HEYL, and HEY1, indicating that Notch signaling is highly active in human muscle progenitors and inhibits their differentiation (Fig. 3, Table 1, Supplementary Fig. 2). Importantly, Kitzman et al. previously reported that inhibition of Notch signaling induces myotube hypertrophy in primary human myoblasts by recruiting a subpopulation of reserve cells 24 . In the present study, we observed that DAPT not only promoted differentiation of myogenic progenitors, but also promoted the efficiency of cell transplantation in xenotransplantation experiments (Fig. 2). We speculate that Notch inhibition improved the efficiency of cell transplantation by augmentation of fusion between donor cells and host myofibers, possibly at the expense of self-renewal of muscle progenitors. If DAPT treatment suppresses replenishment of the satellite cell pool, the beneficial effects of DAPT might be shortterm. The long-term effects of DAPT in cell transplantation, for example, after repeated muscle injury remain to be shown. We found that DAPT treatment drastically downregulated NOTCH3 expression in human muscle progenitors. Recently, Notch3 was reported to mediate self-renewal of mouse C2C12 cells and primary myoblasts during differentiation and prevent their progression into the cell cycle 16 . To confirm the role of NOTCH3 in differentiation of human muscle progenitors, we examined NOTCH3 expression on muscle progenitors by FACS. NOTCH3 was induced in a fraction of the cells only when the cells were cultured at a high cell density, but not at a low cell density (Supplementary Fig. 4). As predicted, NOTCH3-negative cells quickly and robustly formed myotubes. In contrast, most NOTCH3-high cells remained mononuclear (Fig. 4). Thus, our results support Low's report on NOTCH3 function: NOTCH3 suppresses differentiation of a fraction of muscle progenitors to spare the stem cell fraction when a majority differentiate. Low   was almost undetectable in our RNA-seq analysis (Supplementary DATA), which Notch ligands serve as a ligand of NOTCH3 in human muscle progenitors remains to be shown. EP2 was exclusively expressed in NOTCH3-positive cells (Fig. 4d). Therefore, we examined whether NOTCH3-NICD directly upregulated the expression of the EP2 gene or not. Unexpectedly, overexpression of NOTCH3-NICD did not upregulate EP2 expression (Fig. 4e), suggesting that EP2 is upregulated by NOTCH members other than NOTCH3 in self-renewing cells.
NOTCH2 is a candidate that upregulates the expression of the EP2 gene because NOTCH2 is highly expressed in NOTCH3positive cells compared with NOTCH3-negative cells (Fig. 4d) and downregulated by DAPT treatment (Supplementary Fig. 9). Prostaglandin E2 is a potent bioactive lipid messenger, which mediates diverse signals in physiological and pathological conditions. Recently, two groups reported that PGE2 activates proliferation of muscle myoblasts or muscle satellite cells via the EP4 receptor 25,26 , but the roles of PGE2 in muscle differentiation are not well understood. Our results suggest that Notch signaling upregulated EP2 in a human muscle progenitor cell line, Hu5/ KD3. A natural EP2 ligand, PGE2, and an EP2-specific agonist, butaprost, suppressed muscle differentiation. In contrast, a specific antagonist of EP2, TG6-10-1, improved myotube formation by Hu5/KD3 cells ( Fig. 5d-g) and hiPSC-derived muscle progenitors (Fig. 7), suggesting that the NOTCH-EP2 axis regulates self-renewal of human muscle progenitors in differentiationpromoting conditions (summarized in Fig. 9).
Although our data suggest that the PGE2 produced by COX-2 regulated the cell fate decision, differentiation or self-renewal, the concentration of PGE2 in the culture medium was quite low and did not significantly change upon DAPT administration (Supplementary Fig. 7). Furthermore, the level of COX-2 mRNA was also extremely low and did not change upon DAPT administration ( Supplementary Fig. 7). On the other hand, transient overexpression of EP2 by plasmid vectors mimicked the effects of butaprost or PGE2 (Fig. 6a-d, Supplementary Fig. 5d-f). In addition, downregulation of EP2 by shRNA plasmid vectors improved fusion of muscle progenitors (Fig. 6e-i). These data suggest that the EP2 signaling was regulated mainly at the expression level of EP2 receptors.
Both EP2 and EP4 activated adenyl cyclase and increased the intracellular cAMP concentration in the cells. However, an antagonist of EP4, ONO-AE3-208, did not improve the fusion index of Hu5/KD3 myogenic cells (Fig. 8d), suggesting that EP4 did not make a major contribution to self-renewal of myogenic cells.
Forskolin, an activator of adenylyl cyclase, did not inhibit muscle differentiation. H-89, a protein kinase A (PKA) inhibitor, or ESI-09 (Epac inhibitor) did not promote the fusion of myogenic cells, and dbcAMP promoted muscle differentiation of the cells (Fig. 8, Supplementary Fig. 6). 8-CPT-2Me-cAMP (selective activator of Epac) and 8-Bromo-cAMP (selective activator of protein kinase A) had no significant effects on differentiation of myogenic cells ( Supplementary Fig. 8a, b). Together, these results suggest that EP2 promoted self-renewal of myogenic cells via cAMP-independent pathways.
In conclusion, we showed that Notch inhibition promoted differentiation of human muscle progenitors in vitro and in vivo. We also found an important role for Notch/PGE2/EP2 receptor signaling in regulation of the cell fate of myogenic progenitors. Molecular mechanisms by which EP2 promotes self-renewal of myogenic progenitors remain to be determined.

Methods
Ethical statement. The research plans using human iPS cells and all experimental procedures using mice were approved by the Ethical Committees of the National Institute of Neuroscience, National Center of Neurology and Psychiatry (NCNP), Japan, and performed according to the guidelines.
Adult primary myoblasts (human skeletal muscle myoblasts) were purchased from Lonza (catalog #CC-2580) and cultured on collagen type I-coated dishes in 10% FBS/DMEM.
We routinely checked mycoplasma contamination by a PCR method (PCR Mycoplasma Detection Set cat#6601, Takara).
To inhibit Notch signal, cells were cultured in DMEM/10% FBS containing 10 μM DAPT (Sigma-Aldrich). Because DAPT was dissolved in DMSO (10 mM stock solution), 0.1% DMSO was added to the control culture. For inhibition of prostaglandin EP2 receptors, an EP2-specific antagonist, TG6-10-1 (Calbiochem), was added to the cell culture at 1-10 µM. To inhibit EP4 receptors, ONO-AE3-208 (1-50 nM; a gift from Ono Pharmaceutical Co. LTD.) was used. To stimulate EP2, prostaglandin E2 (Nacalai Tesque) (0.5-100 ng/ml) or an EP2 agonist, butaprost (Cayman) (0.01-5 µg/ml) was added to the culture. To stimulate EP4, CAY10684 (Cayman) was added to the culture. The medium was changed every day. For inhibition of PGE2 production, SC-560 (Cayman), valdecoxib (Tokyo Chemical Industry Co. Ltd. (TCI)), or indomethacin (Wako) was added to the culture. For inhibition of protein kinase A, H-89 (Cayman) was added to the culture. To examine the effects of cAMP on muscle differentiation, dbcAMP (Fujifilm) was used. Forskolin (TCI) was used to activate adenylyl cyclase. ESI-09 (Cayman) was Fig. 9 NOTCH3 and EP2 are key signaling molecules for self-renewal of muscle progenitors in differentiation phases. In differentiation conditions, the majority of myogenic cells fuse to form multinucleated myotubes. NOTCH signaling upregulates the expression of EP2 and NOTCH3 in a fraction of muscle progenitors. Prostaglandin E2 activates EP2, and promotes self-renewal of muscle progenitors. The interplay between NOTCH3 signal and EP2 signals remains to be clarified. Fusako Sakai-Takemura et al. report that self-renewal of human muscle progenitors is regulated by NOTCH/EP2 signaling. The findings deepen our understanding on self-renewal of muscle stem cells and contribute to the development of treatments for Duchenne muscular dystrophy.
Statistics and reproducibility. We did not exclude any data from the analysis. Data were expressed as the mean ± standard deviation (SD), and analyzed and plotted using the GraphPad Prism 8 software. The significance of differences between two groups was analyzed by the unpaired two-tailed Student's t-test. Comparisons of multiple experimental groups were performed by two-way ANOVA followed by Sidak's multiple comparisons, or one-way ANOVA followed by Dunnett's or Tukey-Kramer's multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Effect sizes were calculated using the following formula: r ¼ Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All the data of this study are shown in the main text and supplementary information files.
The RNA-seq data were deposited in ArrayExpress with accession No. E-MTAB-8825. The source data underlying the main figures are presented in Supplementary Data. Any additional source data or material used in this study can be obtained from the corresponding author upon reasonable request.