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.

This study examined the role of DAPT, which promotes myoblast fusion to promote engraftment of myoblasts and the manner by which DAPT promotes myoblast fusion. The authors conclude that DAPT enhances engraftment in a mouse model and acts by downregulating NOTCH3 and prostaglandin E2 receptor 2. The results are potentially novel, but further clarification is needed to support the claims.
The statistical analysis was appropriate although details are needed about the analysis of the RNASeq results.
The two goals of the paper detract from the overall impact of the study. One set of experiments examine engraftment of myoblasts after treatment with DAPT which is known to promote myoblast fusion. The second goal, and the title of the paper, examined signaling pathways and provides evidence that DAPT acts by blocking NOTCH3 and prostaglandin EP2 signaling. While there is some relationship between the two objectives, there are a number of important issues to address with respect to engraftment that may or may not be relevant to a study of signaling pathways. These issues are: 1. The choice of 10 µM DAPT caused aberrant fusion of myoblasts as shown by the large number of nuclei and larger myofiber size in Figures 1e and 1I with hiPSCs and a human myoblast line. Fusion looks somewhat better with primary human myoblasts treated with 10 µM DAPT (Supplementary Figure 1b), although there is still significant variability in myotube diameter. 2. For engraftment studies, important to know whether the increased engraftment due to addition of DAPT leads to myofiber maturation and functional improvements in the muscle. 3. Ideally a dose-response study should be performed to identify conditions that improve engraftment and function. Thus, studies are inconclusive as to whether the use of DAPT is suitable to enhance engraftment for therapeutic applications.
The results with respect to the role of NOTCH3 and PGE2 receptor 2 are suggestive but additional support is needed. For the experiments in Figure 4, it was unclear whether the FACS and RT-PCR were performed after sorting and at the time of fusion to ensure that recovery of NOTCH 3 did not occur. Further, NOTCH2 levels had declined in the Notch3 negative cells, which could affect fusion.
The studies on the role of PGE2 need further clarification. There are reports the PGE2 promotes proliferation of myoblasts but Butaprost had no effect and EP4 seems to be responsible, (e.g. Cell Cycle 14(10): 1507-1516 (2015). Also, the dose-response of PGE2 on the fusion index is very different than that observed by Beaulieu et al. (2012. Neurobiology of Diseases 45 (1) 122-129), Figure 3A, which showed a 50% inhibition of human wild type myoblast fusion at 0.5 ng/mL PGE2. The response of the hiPS-derived muscle progenitors differs from the results with primary cells (Figure 6c). Measurement of the PGE2 receptors and knockdown of specific PGE2 receptors may help to clarify the role of PGE2 in proliferation Specific Comments 1. Figure 1d. Clarify when the fusion index was measured. 2.Figure 5e-5g. Provide an explanation as to why the fusion index for low concentrations of drug or no drug is high relative to results reported in Figure 1d for the Hu5/KF3 cells. 3.For the RNA-Seq analysis, how well does the Hu5/KD3 human myoblast line compare with primary human myoblasts? The response to PGE2 and Butaprost may go through a maximum and differ from that of TG6-10-1 but the high baseline may make it difficult to discern an effect due to the drugs on myoblast fusion. 4.Additional details are needed about RNA Sequencing including method of library preparation, read alignment, software used and FDR levels used.
Reviewer #3 (Remarks to the Author): Summary: The manuscript entitled "Prostaglandin EP2 receptor downstream of NOTCH signaling inhibits differentiation of human skeletal muscle progenitors in differentiation condition" by Sakai-Takemura et al. shows that PGE2-EP2 signaling promotes self-renewal of human muscle progenitor cells downstream of NOTCH. They showed that the inhibition of EP2 receptor activation as well as NOTCH signaling facilitated differentiation of muscle progenitors with cell fusion as an index. However, there is insufficient evidence, the logic is only superficial, and manuscript is not clearly written as pointed out below. To this reviewer, this manuscript remains too preliminary, and therefore not suitable for publication in the Communications Biology.
Major concerns: 1) Lack of results regarding the PG synthesis pathway: The authors concluded that NOTCH signaling suppresses differentiation of muscle progenitor cells, and that such effect is mediated by PGE2-EP2 receptor signaling. However, the authors did not observe the PGE2 production and its synthetic pathway. What cell type (muscle progenitor or neighboring cell) produces PGE2? In these cells, is PGE2 release as well as cyclooxygenase (COX; COX-1 or COX2) gene expression increased upon some stimuli or DAPT treatment? In any case, the treatment of cell with an aspirin-like drug such as indomethacin should mimic the effect of DAPT. How about COX-1-selective and COX-2-selective inhibitor?
In addition, the authors should show the significant amount of PGE2 is really produced in the culture media at the corresponding levels showing pharmacologically inhibitory effect on fusion index (25 ng/mL, approximately 70 nM) as shown in Fig. 5e.
2) Only EP2 but not EP4 is really involved? Why the effective doses were different between PGE2 and butaprost? To this reviewer, PGE2 appeared to be more effective than butaprost. Since the authors detected the expression of EP4, another cAMPcoupled PGE receptor, in NOTCH3-positive cells, the authors should evaluate potential contribution of EP4 receptor signaling to PGE2-induced effect. Moreover, the authors should show whether EP2 receptor is functionally expressed in muscle progenitor cells: PGE2 as well as butaprost (and also an EP4 agonist) is able to increase cAMP levels? Furthermore, such a cAMP-producing activities of PGE2 and butaprost were not detected in DAPT-treated cells?
3) Lack of verification of cAMP and downstream pathways:

Answers to the reviewers' comments:
We deeply appreciate the reviewers' valuable comments on our manuscript. According to the comments, we performed additional experiments and obtained several new findings, which further support our hypothesis. The revised parts are in red in the text. This is a list of the figures we revised or newly added to the revised manuscript.  RT-qPCR was performed on mRNAs extracted from freshly sorted NOTCH3-positive (N3+) and NOTCH3-negative cells (N3-). The fusion index was evaluated 3 days after re-plating sorted cells. We revised the experimental design shown in Fig. 4a.
To clarify the relationship between NOTCH3 and EP2, we overexpressed the intracellular domain of NOTCH3 (NICD3) in Hu5/KD3 cells. NICD3 did not increase EP2 expression, suggesting that NOTCH3 is not upstream of EP2. NOTCH2 is a candidate for upregulation of EP2, because NOTCH2 is expressed at a much higher level in myoblasts than in myotubes (our supplementary Fig. 7 and Kitzmann et al., 2006), and, as the reviewer pointed out, NOTCH2 expression was higher in NOTCH3-positive cells than in NOTCH3negative cells (Fig.4d). Our previous model of the NOTCH3-EP2 axis was revised (new Fig.   9).
The studies on the role of PGE2 need further clarification. There are reports the PGE2 promotes proliferation of myoblasts but Butaprost had no effect and EP4 seems to be responsible, (e.g. Cell Cycle 14(10): 1507-1516 (2015).
Another EP2 agonist, ONO-AE1-259-1 did not stimulate the proliferation of hiPSC-derived myogenic cells and human primary myoblasts in growth condition (data not shown). The results are consistent with previous reports using mouse primary myoblasts (e.g. Cell Cycle 14: 1507-1516, 2015). Butaprost did not increase the numbers of myogenic cells in the differentiation conditions (Fig. 6). Thus, EP2 does not stimulate proliferation of the myogenic cells in either proliferation or differentiation condition. We think that EP2 can promote self-renewal of muscle progenitors without promoting cell proliferation.
In this study, we added EP4 agonist. ONO-AE3-208, to the cell cultures at full confluency in order to evaluate the effects on differentiation of muscle progenitors but not on proliferation. ONO-AE3-208 did not promote muscle differentiation of the cells (Fig. 7).
The result suggests that the contribution of EP4 to decision of cell fate (differentiate or stay undifferentiated) of muscle progenitors in differentiation-promoting condition (i.e., full confluency and low levels of growth factors) is relatively small. We think the dose response to PGE2 in their paper is not different from ours.
The response of the hiPS-derived muscle progenitors differs from the results with primary cells (Figure 6c). Measurement of the PGE2 receptors and knockdown of specific PGE2 receptors may help to clarify the role of PGE2 in proliferation.
We appreciate the reviewer's comments. The much lower fusion index in hiPSC-derived muscle progenitors compared with Hu5/KD3 or primary myoblasts is probably due to strong inhibition of their muscle differentiation by TGF-β signaling, which can be reversed Instead of a knockdown of EP2, which did not work well in our hands, we are now generating conventional knockout mice in which the EP2 gene can be inactivated in PAX7positive muscle progenitors and muscle satellite cells. We should be able to report more details on the EP2-mediated signaling that regulates the cell fates of muscle progenitors in the near future.

Specific Comments
1. Figure 1d. Clarify when the fusion index was measured.
The fusion index was measured at Day 10. We added this description to the figure legend to Fig. 1 of the revised manuscript. We showed the RT-qPCR results of primary human myoblasts (supplementary Fig. 3),

Figure
confirming that the genes differently expressed in DAPT-treated Hu5/KD3 human myoblasts were also differently expressed in primary human myoblasts.
Indeed, the baseline of the fusion index of the Hu5/KD3 human myoblast line was high, but we successfully evaluated effects due to the drugs on myoblast fusion (Fig. 7). We also evaluated the effects of TG6-10-1 on hiPS cell-derived myogenic cells (Fig. 6). TG6-10-1 improved fusion of hiPSC-derived muscle progenitors. We think that the results support the findings using Hu5/KD3 cells.

Additional details are needed about RNA Sequencing including method of library preparation, read alignment, software used and FDR levels used.
We added details about RNA sequencing to the Methods of the revised manuscript. To this reviewer, this manuscript remains too preliminary, and therefore not suitable for publication in the Communications Biology.
Major concerns:

1) Lack of results regarding the PG synthesis pathway:
The authors concluded that NOTCH signaling suppresses differentiation of muscle progenitor cells, and that such effect is mediated by PGE2-EP2 receptor signaling.
However, the authors did not observe the PGE2 production and its synthetic pathway.

What cell type (muscle progenitor or neighboring cell) produces PGE2?
TG6

In these cells, is PGE2 release as well as cyclooxygenase (COX; COX-1 or COX-2) gene expression increased upon some stimuli or DAPT treatment?
We measured PGE2 levels using a prostaglandin E2 ELISA kit (Cayman, Item No.514010).
PGE2 levels were extremely low in the culture medium of Hu5/KD3 cells (13-17 pg/ml), or in the culture medium of differentiating hiPSC-derived muscle cells (8-15 pg/ml) (supplementary Fig. 8). Therefore, it was difficult to determine whether DAPT downregulates or up-regulates PGE2 production. The expression levels of COX-1 and COX-2 mRNA were also extremely low when examined by RT-qPCR, and down-or up-regulation by DAPT treatment was not evident in Hu5/KD3 cells (supplementary Fig. 8). These results are consistent with RNAseq data, in which the signal levels of COX-1 and COX-2 genes were low and not down-regulated or up-regulated with DAPT treatment (data not shown).
We speculate that the PGE2-EP2 receptor signaling is mainly regulated at the expression levels of the receptors.

In any case, the treatment of cell with an aspirin-like drug such as indomethacin should mimic the effect of DAPT. How about COX-1-selective and COX-2-selective inhibitor?
Indomethacin improved the fusion of Hu5/KD3 cells and hiPSC-derived muscle progenitors at a level similar to TG6-10-1, suggesting that the EP2 receptor was stimulated with PGE2 produced by COX-1 and/or COX-2 (Fig. 7). COX-2-selective valdecoxib, but not COX-1selective SC-560, promoted differentiation of Hu5/KD3 myogenic cells (Fig. 7) and hiPSCderived myogenic progenitors (supplementary Fig.6), suggesting that COX-2 is mainly involved in NOTCH signal-mediated suppression of muscle differentiation.  .8). We added much higher levels of PGE2 in our experiments, but experiments using specific inhibitors indicated that PGE2 activated the EP2 receptor to promote self-renewal of muscle progenitors.

2) Only EP2 but not EP4 is really involved?
Why the effective doses were different between PGE2 and butaprost? To this reviewer, PGE2 appeared to be more effective than butaprost. shown), suggesting the importance of EP2 in the regulation of muscle differentiation.

Since the authors detected the expression of EP4, another cAMP-coupled PGE receptor, in NOTCH3-positive cells, the authors should evaluate potential contribution of EP4
receptor signaling to PGE2-induced effect.
CAY10684, an agonist of EP4, increased the cAMP level in Hu5/KD3 cells, but did not reduce the fusion index (Fig. 7). In addition, an EP4 antagonist, ONO-AE3-208, did not significantly increase the fusion index of Hu5/KD3 myogenic cells (Fig. 7) and hiPSCderived muscle progenitors (data not shown), suggesting that the contribution of EP4 to PGE2-induced effects downstream Notch signaling is relatively small.

Moreover, the authors should show whether EP2 receptor is functionally expressed in muscle progenitor cells: PGE2 as well as butaprost (and also an EP4 agonist) is able to
increase cAMP levels? Furthermore, such a cAMP-producing activities of PGE2 and butaprost were not detected in DAPT-treated cells?
The cAMP level was elevated by PGE2, butaprost, or CAY10684. These data indicate that both EP2 and EP4 receptors are functionally expressed in muscle progenitors. The cAMPproducing activities of PGE2 and butaprost were slightly reduced in the cells treated with DAPT for 3 days, although there was no statistically significant difference due to the high variations in the values (Fig. 7, supplementary Fig. 9). Importantly, CAY10684 increased the intracellular cAMP level as butaprost did, but the fusion index of CAY10684-treated cells was not reduced, suggesting that cAMP might not be the mediator for PGE2-induced self-renewal-promoting activities.

3) Lack of verification of cAMP and downstream pathways:
The authors discussed EP2-derived cAMP may participate in self-renewal of muscle progenitor cells (page 5, line 3 from bottom). If so, the effect of PGE2 or butaprost should be mimicked by membrane-permeable cAMP such as dibutyryl cAMP (dbcAMP).

Conversely, the effects of PGE2 and butaprost are inhibited by H-89, an PKA inhibitor?
We tested the effects of dbcAMP and H-89, a PKA inhibitor, on the differentiation of myogenic cells. Unexpectedly, dbcAMP stimulated the differentiation of muscle progenitors. H-89 did not improve the differentiation of myogenic cells. We confirmed the results using both Hu5/KD3 myogenic cells and hiPSC-derived muscle progenitors (Fig. 7   and supplementary Fig. 6). These results suggest that the cAMP-PKA pathway does not regulate self-renewal of muscle progenitors downstream of Notch signaling.

The effect of DAPT is reversed by PGE2, butaprost, an EP4 agonist, and dbcAMP?
Furthermore, even in the absence of NOTCH3, is dbcAMP able to maintain the selfrenewal state in the muscle progenitors?
The effects of DAPT were not reversed by butaprost in Hu5/KD3 cells, probably because DAPT down-regulated EP2. Only a high dose of PGE2 reversed the effects to some extent (data not shown).
CAY10684, an EP4 agonist, increased the cAMP concentration but did not inhibit fusion of myogenic cells (Fig. 7). In addition, dbcAMP promoted differentiation of Hu5/KD3 myogenic cells and hiPSC-derived muscle progenitors. H89, an inhibitor of PKA, did not promote the differentiation of Hu5/KD3 and hiPSC-derived muscle progenitors.
Collectively, these results suggest that the Notch/EP2-mediated inhibition of muscle differentiation is not regulated by the cAMP-PKA signal pathway.
cAMP did not seem to mediate EP2 signaling for self-renewal of muscle progenitors (Fig.7).
Therefore antagonized the effects of butaprost (Fig. 8, supplementary Fig.10), suggesting that Src is one of major signaling molecules that inhibit muscle differentiation downstream of the EP2 receptor. LY294002 (PI3K inhibitor) showed mildly antagonistic effects against butaprost (supplementary Fig. 10). SP600125 (JNK inhibitor), erlotinib (EGFR inhibitor), and CHIR99021 (GSK-3β inhibitor) had little or no effect on the muscle differentiation of human muscle progenitors (supplementary Fig. 10) . 10), and butaprost further inhibited differentiation of muscle progenitors (Fig. 8). We added these results to the revised Ms. In any case, to this reviewer, there still exists fatal discrepancy in this manuscript. The authors failed to certify that effective dose of PGE2 is endogenously synthesized in this culture system.

PGE2 levels increase with differentiation
We appreciate reviewer's comments. In the new experiments, we sampled not only culture media but also cells, and measured the PGE2 concentration at five timepoints during muscle differentiation. Until day 4, PGE2 levels were very low, but when cells started to form myotubes, the PGE2 level increased up to 200 pg/mL (approx. 0.5 nM) (Supplementary Figure 8).
Although the concentration was less than 1 nM in the early phase of differentiation, COX inhibition experiments (Fig.7) and EP2 overexpression and knockdown experiments (Fig.5) suggest that an effective dose of PGE2 is endogenously synthesized in this culture system.
We found the upregulation of PGE2 in the later stage of differentiation. Therefore, we examined COX1 and COX2 mRNA levels in myoblasts and myotubes by RT-qPCR. Although there was no statistical significance, COX1 was found to be expressed at higher levels in myotubes rather than in myoblasts. In contrast, COX2 tended to be highly expressed in myoblasts. The results suggest that COX2 in myoblasts produced PGE2 in an early stage of differentiation to regulate cell fate (this study). Later, COX1 in myotubes would produce PGE2 to keep satellite cells in an undifferentiated state in their niche. The latter hypothesis needs further experiments.

Upregulation of EP2 is enough to inhibit differentiation of myogenic progenitors
To clarify the roles of EP2, we up-regulated the expression of EP2 in Hu5/KD3 myogenic cells using two different plasmid vectors. Interestingly, mild upregulation of EP2 inhibited differentiation of the cells (new Figure 5h-j and supplementary Figure 5d). In contrast, modest downregulation of EP2 by shRNA increased the fusion index (Fig. 5). The results indicate that EP2 is a key molecule to regulate self-renewal of myogenic cells. The results also suggest that the expression level of EP2 is tightly regulated and that even in a low concentration of prostaglandinE2, upregulation of EP2 is enough for signal transduction.
Of course, there is no doubt about the requirement of PGE2 for EP2 signaling as shown by the experiments using cox inhibitors (Fig.7).

2) Another problem in this manuscript is weak conclusion. The authors concluded that
the downstream signaling pathway of EP2 receptor is not cAMP-PKA pathway, but c-Src/PI3K/inhibition of p38-MAPK. However, they concluded this mechanism only from experiments using inhibitors. To this reviewer, if the PGE2-EP2 receptor activation really affects muscle differentiation via the mechanism other than cAMP/PKA pathway, the signaling pathway should be the crucial part of this manuscript. The authors should show that PGE2 as well as butaprost but not CAY10684 (an EP4 agonist) really activates c-Src and inhibits p38 MAPK in muscle progenitor cells.

cAMP-PKA signaling
We further tested the effects of three activators and one inhibitor of the AC-cAMP-PKA and AC-cAMP-Epac signaling pathways on muscle differentiation.

EP2 and Src signaling
Although our previous data showed that Src signaling negatively regulated muscle differentiation and a Src inhibitor antagonized butaprost's effects, we did not detect phosphorylation of Src by butaprost or PGE2 by using Western blotting (data not shown) or a human phospho-kinase array (Abcam, ARY003B). The array also suggested that Src 4 family kinases (Src, Lyn, Lck, Fyn, Yes , Fgr, and Hck), ERK1/2, JNK1/2, EGFR, GSK-3β, Akt, and CREB are not direct targets of EP2 signaling (please see attached figure). Therefore, we deleted the data on Src and PI3K from the manuscript. On a human phospho-kinase array, butaprost slightly reduced phosphorylation of p38α-MAPK, a key signaling molecule for muscle differentiation. Western blotting also showed lower levels of phosphorylation of p38α-MAPK in butaprost-treated cells than in control cells (please see attached figure).
Therefore, p38α-MAPK might be a target of EP2 signaling. In near future, we will clarify the molecules that link EP2 and p38α MAPK. Thus, how EP2 regulate self-renewal of muscle progenitors is not clear. We stated this limitation of our study in the Discussion.