Abstract

   Skeletal muscle is crucial for lifelong metabolic health, with its
   development initiating during embryogenesis and guiding later growth.
   Single‐cell RNA sequencing of E13.5 mouse embryos suggests that
   maternal obesity impairs embryonic myogenesis, primarily during myotube
   formation, and is correlated with the downregulation of myogenin (Myog)
   expression. Spatial transcriptomic sequencing further confirms these
   findings. Conversely, maternal obesity induces the upregulation of
   RelA, a transcription factor and inflammatory signaling effector, in
   the myogenic cells of obese embryos. Additionally, single‐cell ATAC
   sequencing suggests that a cis‐regulatory region, located 4 kb proximal
   to the Myog promoter, exhibits coordinated expression with Myog and
   acts as an enhancer of Myog. In obese embryonic myogenic cells, RelA is
   recruited to this region, inhibiting Myog expression and myotube
   formation. Notably, in vitro RelA knockdown or inhibition rescues Myog
   expression and promotes myogenic differentiation. Similarly, metformin
   treatment of obese mothers restores embryonic Myog expression and
   myogenesis by suppressing RelA. Together, the multi‐omics analyses
   demonstrate that RelA targets a distal Myog enhancer to coordinately
   inhibit Myog expression and embryonic myogenesis.

   Keywords: embryonic myogenesis, enhancer, inflammation, maternal
   obesity, myogenin, RelA
     __________________________________________________________________

   Maternal obesity (MO) accounts for more than 40% of pregnancies in the
   United States, induces inflammation, and impairs embryonic myogenesis,
   known to program muscle dysfunction in offspring. Inflammation enhances
   the binding of RelA to a novel distal enhancer located 4 kb upstream of
   myogenin promoter to inhibit embryonic myogenesis, which is recovered
   by metformin treatment during early pregnancy.

   graphic file with name ADVS-12-e03712-g003.jpg

1. Introduction

   Skeletal muscle is the major tissue for locomotion and metabolism,
   constituting 30–40% of body mass.^[ [42]^1 , [43]^2 , [44]^3 ^]
   Myogenesis starts during early embryogenesis, and the majority of the
   primary and secondary muscle fibers form during the prenatal stage
   through a sequential progression of myogenic progenitor cells to
   multinucleated myotubes and fibers. During the early stage of
   embryogenesis, myogenic progenitor cells originate from the
   dermomyotome of embryonic mesoderm and migrate to their designated
   locations in the body to develop into various muscle groups.^[ [45]^4
   ^] Embryonic myogenesis starts around embryonic day 9.5 (E9.5) in the
   mesoderm of mice when PAX3 and PAX7‐positive myogenic progenitor cells
   first appear.^[ [46]^5 , [47]^6 ^] During E10.5 to E12.5, these
   progenitor cells express the first myogenic regulatory factor (MRF),
   MYF5, which initiates myogenic lineage commitment.^[ [48]^7 ^] A
   portion of MYF5+ cells also contributes to the brown adipocyte
   lineage.^[ [49]^8 ^] Subsequently, MYOD1, another member of MRFs,
   appears and enforces the differentiation of myogenic progenitor cells
   to myoblasts.^[ [50]^9 ^] Later, myoblasts undergo terminal
   differentiation and fuse to form multinucleated myotubes, which is
   governed by myogenin (Myog).^[ [51]^10 ^] The myotubes form during this
   stage become primary myofibers and work as a scaffold for secondary
   myofiber formation during the fetal stage.^[ [52]^11 ^] These,
   altogether, determine the post‐natal muscle development and its
   long‐term metabolic function.^[ [53]^12 ^]

   The expression of MRFs is regulated by essential regulatory factors,
   including other MRFs and non‐coding RNAs, with concomitant epigenetic
   modifications of their promoters.^[ [54]^13 ^] Distinct distal
   enhancers have been identified for Myf5 and Myod1, which possess the
   binding motifs of regulatory factors.^[ [55]^14 , [56]^15 , [57]^16 ^]
   Consistently, Myog expression is initiated by MYOD1, a co‐transcription
   factor myocyte enhancer factor 2C (MEF2C), as well as regulated by
   myogenin itself as a positive feedforward loop, followed by the removal
   of inhibitory epigenetic marks in the promoter.^[ [58]^13 , [59]^17 ^]
   Available studies suggest that, MYOD1 can modulate 3D chromosome
   architecture to alter interactions between regulatory elements and
   target genes including Myog.^[ [60]^18 , [61]^19 ^] However, the
   dynamics of Myog expression and the involvement of possible
   cis‐regulatory elements and enhancer have not been defined.

   The worldwide steady escalation of the obesity rate in reproductive‐age
   women over the past decades (currently ≈39.8% in the United States) has
   turned it into a pandemic.^[ [62]^20 , [63]^21 , [64]^22 , [65]^23 ,
   [66]^24 ^] Maternal obesity (MO) not only affects maternal health but
   also exerts a negative impact on the development of their offspring,
   including myogenesis.^[ [67]^25 , [68]^26 ^] Recent studies, including
   our studies, have reported that MO induces a low‐grade chronic
   inflammation in the uterine microenvironment and upregulates nuclear
   factor kappa B (NF‐kB) signaling in embryos.^[ [69]^25 , [70]^27 ,
   [71]^28 ^] As effectors of NF‐kB signaling, the NF‐kB family consists
   of RelA (P65), c‐Rel, RelB, P50 (NFKB1), and P52, which form RelA‐P50
   and RelB‐P52 heterodimers to activate gene expression.^[ [72]^29 ^] As
   the key mediatory pathway of inflammation, NF‐kB signaling is involved
   in various pathophysiological processes, affecting myoblast
   proliferation and myogenic differentiation.^[ [73]^30 , [74]^31 ^] In
   satellite cells, NF‐kB enhances the cell cycle progression from G1 to S
   phase, while inhibiting myogenic differentiation.^[ [75]^32 ^] However,
   the impact of NF‐kB in embryonic myogenesis has yet to be explored,
   partially due to the technical challenges for studying early embryos.
   Using single‐cell (sc) multi‐omics, we explored the effects of
   inflammation and its effector, RelA, in impairing embryonic myogenesis
   due to MO.

   We conducted scRNA and scATAC sequencing of E11.5 and E13.5 embryos,
   and spatial transcriptomic profiling of the myotome of E13.5 embryos
   from control and obese mice. We found that RelA binds with a novel
   distal enhancer that also possesses binding motifs of MYOD1, MEF2C, and
   MYOG, which suppresses Myog expression in embryos due to MO. Both in
   vitro knockdown and in vivo inhibition of RelA rescued embryonic
   myogenesis. These studies suggest that RelA and the associated NF‐kB
   signaling is a potential therapeutic target to rescue embryonic
   myogenesis in offspring from mothers with obesity and other
   inflammatory conditions.

2. Results

2.1. Single‐Cell Transcriptomic and ATAC Sequencing of the Mouse Embryos

   To induce obesity, we randomly divided female mice into control (CT)
   and MO groups. CT group was exposed to a control diet containing 10%
   energy from fat, and the MO group was exposed to a High‐fat diet (HFD)
   containing 45% energy from fat until there was a 20% difference in the
   body weight between the two groups in ≈10 weeks (Figure [76]1A; Figure
   [77]S1A, Supporting Information). We also conducted an intraperitoneal
   glucose tolerance test (IpGTT) and fasting blood glucose test, where
   obese mice exhibited higher fasting blood glucose and less glucose
   tolerance (Figure [78]S1B, Supporting Information).

   During the first wave of myogenesis, myo‐progenitor cells highly
   express Myf5 to initiate myogenic commitment, and Myog expression
   escalates later and reaches its peak at E13.5 for terminal
   differentiation of myoblasts and their fusion to form myotubes.^[
   [79]^33 ^] To investigate the impact of MO on myoblast differentiation
   and myotube formation, we conducted scRNA‐sequencing of embryonic
   tissue at E11.5 and 13.5 and scATAC‐seq at E13.5 from C57BL/6 control
   and obese female mice. For scRNA‐seq data processing, after filtering
   out the low‐quality cells, we recovered a total of 43,813 cells from
   where 27,359 cells were from E11.5 to 18,454 cells from the E13.5
   group. Using Seurat V5, we combined all the cells from two
   developmental time points and clustered them following the non‐linear
   dimensional reduction technique: UMAP. We identified 20 different cell
   clusters from our integrated data set and annotated them based on the
   expression of specific marker genes (Figure [80]1B; Tables [81]S1 and
   [82]S2, Supporting Information). Among the major cell clusters, cluster
   0 was embryonic fibroblast cells, expressing fibrogenic marker genes
   Col6a1, Col6a3, Col1a1, and Pdgfra, accounting≈15% of all cells;
   clusters 1 and 2, two next major clusters, accounted for ≈13% of cells
   and expressed embryonic mesenchymal marker genes Prrx1 and Twist1
   (Figure [83]1C,D; Table [84]S1, Supporting Information).^[ [85]^34 ,
   [86]^35 ^] On the other hand, clusters 9 and 17 expressed myogenic
   markers Myod1 and Cdh15; clusters 6 and 7 highly expressed
   osteochondrogenic marker genes Msx1 and Sox9 (Figure [87]1B,C; Table
   [88]S1, Supporting Information).^[ [89]^26 , [90]^36 ^]

Figure 1.

   Figure 1
   [91]Open in a new tab

   Single‐cell RNA and ATAC‐seq of embryos from control (CT) and obese
   (MO) mice. A) Schematic diagram of experimental design for induction of
   obesity in female mice and scRNA‐seq of E11.5 & E13.5 embryos and
   scATAC‐seq of E13.5 embryos. B) Uniform manifold approximation and
   projection (UMAP) plot showing 20 major clusters from the scRNA‐seq
   data sets of both time points. Names of individual clusters are listed
   next to the figure along with their corresponding identification
   number. C) Dot plot showing top differentially expressed
   lineage‐specific markers across all the 20 clusters identified from the
   integrated scRNA‐seq dataset. D) Percentage of each cell cluster in the
   integrated scRNA‐seq data. E) UMAP plot of scATAC‐seq showing 27 cell
   clusters from E13.5 embryos of CT and MO mice. F) Predicted cell
   clusters in scATAC‐seq from the scRNA‐seq by cross‐modality integration
   and label transfer. G) The percentage of each predicted cell cluster in
   the integrated scATAC‐seq data.

   With the progression of embryonic days from E11.5 to E13.5, there was
   a significant decline in the mesenchymal cell population (clusters 1
   and 2), and a gradual increase in the fibrogenic (cluster 0),
   osteochondrogenic (clusters 6 and 7) and myogenic cells (clusters 9 and
   17; Figure [92]S1C,D, Supporting Information). The gradual decrease in
   mesenchymal progenitor cell population and concomitant increase in the
   lineage‐specific cell types showed an ongoing progression of
   organogenesis and illustrated a landscape of embryonic development
   during E11.5 and E13.5.^[ [93]^37 ^]

   In scATAC‐seq, we recovered ≈16 283 cells after initial pre‐processing
   and quality control to remove the low‐quality cells. Clustering and
   UMAP visualization of those cells by following non‐linear dimension
   reduction methods using Signac 2.1.0, we identified 27 unique cell
   clusters (Figure [94]1E). To interpret the cell type, we conducted the
   cross‐modality integration and label transfer from our scRNA‐seq data
   and predicted the cell types of scATAC‐seq clusters (Figure [95]1F).
   Major cell clusters such as neurogenic, chondrogenic, and myogenic
   clusters had very high (>90%) prediction scores, while the mesenchymal
   and fibrogenic cell clusters had a comparatively low prediction score
   (<80%) which could be due to their heterogeneity during this
   developmental stage (Figure [96]S1E, Supporting Information). The
   predicted clusters exhibited similar gene activity patterns when
   compared to the marker gene expression from scRNA‐seq (Figure [97]S1F,
   Supporting Information). For instance, the fibrogenic cluster (Cluster
   0) had higher expression of the Col6 gene in both scRNA and scATAC‐seq
   data; the myogenic clusters 17 and 9 expressed Myod1, Cdh15, Myog, and
   Myh3 for both scRNA and scATAC‐seq (Figure [98]S1F, Supporting
   Information). Similar to scRNA‐seq, mesenchymal cell cluster (cluster
   1) and fibrogenic cell cluster (cluster 0) possessed the majority of
   cells in scATAC‐seq (Figure [99]1G). We identified 266 959 variable
   peaks across the cells, of which 41% overlapped with the promoter, 15%
   with exons, and 28% with introns and intergenic regions (Figure
   [100]S1G, Supporting Information). Altogether, our scRNA and scATAC‐seq
   captured the landscape of early embryonic development and the
   developmental trajectories of myogenic, osteogenic, and fibrogenic
   cells.

2.2. Myogenic Developmental Trajectory

   During early embryonic development, the myotome‐derived Pax3, Pax7, and
   Myf5+ myogenic progenitor cells differentiate into myocytes and fuse to
   form Myh3, Myl1+ multinucleated myotubes under the governance of MRFs
   Myf5, Myod1 and Myog (Figure [101]2A). Cell clusters 9 and 17 from our
   scRNA‐seq expressed those marker genes and MRFs (Figure [102]S2A,
   Supporting Information). Therefore, we extracted all cells from both
   clusters and subclustered them into 7 unique clusters (clusters 0 to 6)
   to reconstruct the myogenic developmental trajectory (Figure [103]2B).
   During embryonic myogenesis, the PAX3+ mesenchymal progenitor cells are
   the first cells to appear from the dorsomedial lips of dermomyotome
   that express Pax7 to form myotome, which was identified in cluster 1
   (Figure [104]2C).^[ [105]^8 ^] MYF5 is the earliest MRF that determines
   their myogenic lineage specification.^[ [106]^7 ^] MYOD1 is the later
   transcription factor that starts expressing ≈E10.5 and overlapping Myf5
   expression which were identified in cluster 0 (Figure [107]2B,C). Myod1
   expression induces differentiation of myoblasts into committed myogenic
   lineage cells and peaked among cells of clusters 4 and 5
   (Figure [108]2B,C).^[ [109]^38 ^] MYOG is the last regulatory factor in
   embryonic myogenesis required for the fusion of mononucleated myoblasts
   to multinucleated Myh3 and Myl1+ myotubes which were observed in
   clusters 3 and 6 (Figure [110]2B,C).^[ [111]^39 ^] Brown adipocytes,
   contributed by multiple tissue lineages including Pax3 and Myf5+
   mesenchymal progenitor cells, are positive for Ebf2 which were present
   in cluster 2 (Figure [112]2B,C).^[ [113]^8 ^]

Figure 2.

   Figure 2
   [114]Open in a new tab

   MO inhibits embryonic myogenesis via inhibition of Myog expression. A)
   Schematic diagram illustrating the gradual progress of embryonic
   myogenesis from myogenic progenitor cells to myotube formation under
   the regulation of different myogenic regulatory factors. B) Isolation
   of myogenic cluster and reconstruction of myogenic development
   trajectory from integrated scRNA‐seq of E11.5 and E13.5 embryos. C)
   Myogenic cell type‐specific marker gene expression in the reconstructed
   developmental trajectory. D) Identification of myogenic cell types from
   E13.5 scATAC‐seq based on the marker gene activity. E) Comparison
   between percentages of myogenic cell populations from scRNA‐seq. F)
   Comparison of myogenic marker genes and myogenic regulatory factors in
   differentiated myoblasts and myotubes from integrated scRNA‐seq. G)
   Accessibility and activity of myogenic marker genes in myogenic cell
   population from scATAC‐seq.

   Additionally, compared to E11.5, E13.5, myogenic cells exhibited a
   lower prevalence of progenitor cells and a higher prevalence of
   differentiated myoblasts and myotubes (Figure [115]S2B, Supporting
   Information). The expression of Pax3, a marker of progenitor cells, was
   not identifiable at E13.5, while Myh3 and Myl1 were highly expressed in
   E13.5 myogenic cells (Figure [116]S2B, Supporting Information).

   To identify similar cellular heterogenicity in the myogenic cell
   population from our scATAC‐seq at E13.5, we extracted cells from
   clusters 9 and 17. Both clusters exhibited higher activity and
   accessibility of myogenic marker genes Pax7, Myod1, Myog, and Myh3
   (Figure [117]2D; Figure [118]S2C, Supporting Information). Further
   clustering of both clusters 9 and 17 exhibited 5 sub‐clusters. Two
   sub‐clusters from cluster 17 exhibited higher activities of Myod1,
   Myog, Myh3, and Myl1, which denotes as differentiated myoblasts and
   multinucleated myotubes respectively (Figure [119]2D; Figure [120]S2D,
   Supporting Information). On the other hand, three sub‐clusters from
   predicted cluster 9 expressed Pax3, Pax7, Myod1, and Ebf2 that
   identified them as the mesenchymal progenitors, proliferating
   myoblasts, and BAT lineage cells, respectively (Figure [121]2D; Figure
   [122]S2E, Supporting Information). Additionally, Myod1 was only found
   in proliferating myoblasts not in the other two clusters
   (Figure [123]2D).

   Cumulatively, these data identify the heterogenic cell types present in
   the myogenic cell populations of embryos and portray the gradual
   progression of myogenic developmental trajectory based on their gene
   expression and chromatin accessibility. In addition, a portion of BAT
   progenitor cells were also identified with a mesenchymal progenitor
   origin.

2.3. Myog and Myotube Formation were Inhibited in Differentiated Myoblasts
due to Maternal Obesity

   The commitment of progenitor cells to the myogenic lineage and their
   fusion to multinucleated myotubes are two major steps of myogenesis. We
   compared the expression and accessibility of MRF genes between cells
   from CT and MO groups in both scRNA and scATAC‐seq (Tables [124]S3 and
   [125]S4, Supporting Information) Combined E11.5 and E13.5 scRNA‐seq
   data didn't exhibit difference in the expression of Myf5 and other
   early myogenic marker genes such as Pax7 (Figure [126]S2F, Supporting
   Information). Similarly, the expression of Ebf2, the earliest gene
   expressed in the BAT progenitor cells didn't differ (Figure [127]S2G,
   Supporting Information). We observed a lower expression of Myod1 in
   proliferating myoblasts, however, the difference was not significant
   and percentage of cells didn't differ between the two groups
   (Figure [128]2E; Figure [129]S3A, Supporting Information). MO impaired
   the expression of Myod1 and Myog in the differentiated myoblast cells
   (Figure [130]2F; Figure [131]S3B, Supporting Information). The markers
   of the terminal stage of embryonic myogenesis, Myog, Myh3 and Myl1 were
   expressed much lower in myotubes from MO group (Figure [132]2F; Figure
   [133]S3B, Supporting Information). The percentages of differentiated
   myoblasts and myotubes were also ≈32% and ≈25% lower in MO embryos,
   respectively (Figure [134]2E). In term of embryonic days, Myod1 and
   Myog both were downregulated in myoblasts and myotubes during E11.5
   however only Myog exhibited lower expression in both cell types during
   E13.5 (Figure [135]S3C, Supporting Information). Additionally, gene
   ontology (GO) analysis of the downregulated genes identified their
   association with major biological processes including “Muscle cell
   differentiation”, “Myoblast Differentiation”, “Myoblast fusion” and
   “Myotube cell Development”, suggesting the downregulation of myogenic
   differentiation and myotube formation. (Figure [136]S3D, Supporting
   Information). In short, the negative impact of MO on embryonic
   myogenesis was highly evident in the differentiated myoblasts and
   myotubes.

   In our scATAC‐seq at E13.5, the number of cells in both myogenic
   clusters from the CT and MO groups was comparable (Figure [137]S3E,F,
   Supporting Information). Gene accessibility profiles from our scATAC‐
   seq also exhibited a similar impact of MO on the myogenic developmental
   trajectory. The accessibility of Pax7 and Myf5 in the myogenic
   progenitor cell population and Ebf2 from brown adipocyte lineage didn't
   vary between the progenitor cell populations from CT and MO groups
   along with their gene activity (Figure [138]S3G,H, Supporting
   Information). However, the accessibility of Myod1 differed in the
   proliferating myoblasts between the two groups (Figure [139]2G).
   Furthermore, Myog accessibility in the differentiated myoblasts and
   myotubes especially in the promoter region was lower in MO
   (Figure [140]2G).

   Taken together, the scRNA and scATAC‐seq analysis of embryonic tissue
   samples showed that MO impairs Myog expression and embryonic
   myogenesis, mainly during the terminal differentiation and myotube
   formation.

2.4. Maternal Obesity Induces an Inflammatory Response and Activates NF‐kB
Signaling in Embryos

   Maternal obesity elevates pro‐inflammatory cytokine levels in maternal
   circulation and induces an inflammatory response in the embryos.^[
   [141]^27 , [142]^40 ^] To explore the impact of MO on inflammatory
   response in embryos, we conducted the Gene Ontology (GO) analysis of
   upregulated genes from the DEGs between CT and MO embryos. The GO
   analysis suggests the association of upregulated genes in MO with
   biological processes “Regulation of inflammatory response”, “Canonical
   NF‐kB signaling”, “IL‐6 and IL‐8 production”, and “IL‐6 mediated
   signaling pathway” (Figure [143]3A; Figure [144]S4A,B, Supporting
   Information). In addition, GO analysis identified the involvement of
   upregulated genes in MO with molecular functions: “TNF receptor
   superfamily binding”, “NF‐kB binding”, and “TNF receptor binding”
   (Figure [145]3B; Figure [146]S4C, Supporting Information).
   Additionally, analysis of transcription factors (TF) associated with
   the upregulated genes by ChEA3 identified members of the NF‐kB family
   (NFKB1, RelA) among the top associated TF lists. (Figure [147]3C).
   These data showed the up‐regulation of inflammation in MO embryos.

Figure 3.

   Figure 3
   [148]Open in a new tab

   MO induces an inflammatory response and activates NF‐kB signaling in
   embryos. A) Enriched GO biological processes associated with the
   upregulated genes in MO embryos. B) Enriched GO molecular function
   associated with the upregulated genes in MO embryos. C) Top ten
   transcription factors regulating upregulated genes in myogenic clusters
   from scRNA‐seq. D) Top ten enriched motifs in myogenic cell clusters
   from scATAC‐seq. E) Motif activity and fold changes of NF‐kB family
   genes in MO embryos. F) Comparative motif activity of RelA in myogenic
   clusters from CT and MO embryos.

   NF‐kB family, consisting of RelA, c‐Rel, RelB, P50, and P52, exists as
   RelA homodimer, RelA‐P50, and RelB‐P52 heterodimers when activated,^[
   [149]^29 ^] mediating the canonical and non‐canonical NF‐kB signaling
   pathways respectively.^[ [150]^30 ^] Proinflammatory cytokines,
   including TNF‐a (tumor necrosis factor‐alpha), IL‐1 (interleukin‐1),
   IL‐6, and IFN‐I (type I interferons) are major activators of canonical
   NF‐kB signaling.^[ [151]^31 ^] However, the expression of the NF‐kB
   family genes didn't alter due to MO (Figure [152]S4D, Supporting
   Information).^[ [153]^32 , [154]^41 ^] In canonical NF‐kB signaling,
   RelA homodimer or RelA and P50 heterodimer form a complex with IκBα
   (inhibitor of nuclear factor kappa B) and remain inactive. Upon
   activation by different external stimuli, IκB kinase (IKK)
   phosphorylates IκBα and triggers its degradation by ubiquitination.
   This process leads to the rapid nuclear translocation of the canonical
   RelA dimer and triggers the expression of their targeted genes.^[
   [155]^42 ^] To further identify the most active transcription factors
   of the NF‐kB family, we performed motif enrichment analysis in myogenic
   cell clusters based on scATAC‐seq (Table [156]S5, Supporting
   Information). RelA (P65) and P50 (NFKB1) were among the top active
   motifs in the myogenic cells (Figure [157]3D). Additionally, among the
   members of the NF‐kB family, RelA motif activity was upregulated (Fold
   change: 2.26) in the MO embryos (Figure [158]3E), which was confirmed
   by the differential motif activity analysis (Figure [159]3F; Figure
   [160]S4E–G, Supporting Information). In short, MO activates NF‐kB
   signaling in myogenic cells demonstrated through the upregulation of
   RelA dimer activity.

2.5. Spatial Transcriptomic Profiling of Myotome Reveals Myog Inhibition and
NF‐kB Upregulation Due to MO

   To gain deep insight into the embryonic myotome due to MO, we conducted
   GeoMx DSP transcriptomic profiling of myotome (Figure [161]4A; Figure
   [162]S5A, Supporting Information) and principal component analysis
   (PCA) grouped samples from CT and MO into distinct clusters, indicating
   differences in their transcriptional profiles. (Figure [163]4B). A list
   of down‐regulated genes and up‐regulated genes were identified between
   CT and MO embryos (Figure [164]4C; Figure [165]S5B, Supporting
   Information). Similar to scRNA‐seq, the GO analysis with the
   differentially expressed genes revealed the association of the
   biological process “Regulation of inflammatory response” with the
   up‐regulated genes and “Muscle cell differentiation” with the
   down‐regulated genes (Figure [166]4D,E). Furthermore, significant
   down‐regulation of Myod1 and Myog, as well as Myh3 was observed in the
   MO embryos (Figure [167]4F). On the other hand, Pax7 was higher in MO
   embryos, whereas Pax3 and the levels of Mef2a, and Mef2c didn't differ
   (Figure [168]S5C, Supporting Information). In terms of canonical NF‐kB
   signaling, a higher relative abundance of RelA and P50 was observed in
   the MO group (Figure [169]4G; Figure [170]S5D, Supporting Information).

Figure 4.

   Figure 4
   [171]Open in a new tab

   Spatial transcriptomic sequencing of embryonic myotome from CT and MO
   embryos. A) Fluorescence image of embryonic myotome in cross‐section of
   embryos visualized by MYH3+ cells. B) UMAP plot of principal component
   analysis (PCA) of embryonic myotome from CT and MO embryos. C) Volcano
   plot of differentially expressed genes in the myotome of MO group
   compared with CT group. D) Enriched GO biological processes associated
   with the upregulated genes in embryonic myotome. E) Enriched GO
   biological processes associated with the downregulated genes in
   embryonic myotome. F) Relative mRNA abundance of myogenic regulatory
   factors and marker genes in myotome between CT and MO embryos. G)
   Relative mRNA abundance of members from NF‐kB family in myotome between
   CT and MO embryos.

   Overall, similar to our findings from scRNA and scATAC‐seq, the GeoMx
   spatial transcriptional profiling confirmed the suppression of
   myogenesis and the up‐regulation of NF‐kB family members RelA and P50
   due to MO in the myotome.

2.6. NF‐kB Binds to an Enhancer of Myog to Down‐Regulate Its Expression

   Enhancers are cis‐regulatory regions that enhance the expression of
   nearby genes.^[ [172]^43 ^] The promoter‐enhancer interaction, i.e.,
   cis‐regulatory interactions can be identified from scATAC‐seq data by
   analyzing the cis‐co‐accessible networks (CCANs).^[ [173]^44 ^] CCAN
   analysis of our current scATAC‐seq datasets revealed a co‐accessible
   region (Chr1: 134285400‐134286400) proximal to the Myog promoter
   (Figure [174]5A). The accessibility of this co‐accessible enhancer
   region was cell‐type specific and was only found in myogenic cell
   clusters (Figure [175]5A). Also, the activity of this enhancer region
   followed a similar pattern of changes as the Myog promoter and its
   expression (Figure [176]5B). The accessibility of this enhancer region
   was highly correlated with Myog promoter accessibility (R^2 = 0.82) and
   gene activity (R^2 = 0.83) (Figure [177]5C,D). In alignment with this
   finding, the accessibility of this identified enhancer region was lower
   in both MO myoblasts and myotubes where Myog expression was inhibited
   (Figure [178]5E). This enhancer region possesses several binding sites
   of NF‐kB family members: RelA, P50, and MYOD1and MYOG, and their
   co‐factor MEF2C from JASPER database (Figure [179]5F; Figure [180]S6A,
   Supporting Information).^[ [181]^45 ^] In ReMap, a database of
   regulatory regions based on DNA binding experiments like ChIP‐seq,
   exhibits active MYOG and MYOD1 binding activities in this region in
   C2C12 cells, embryonic myoblasts and myotubes (Figure [182]S6B,
   Supporting Information). We observed lower accessibility of the MYOD1,
   MYOG, and MEF2C binding sites on the enhancer region, as well as their
   lower motif binding activity in MO myogenic cells (Figure [183]5F;
   Figure [184]S6C–E, Supporting Information). We didn't observe changes
   in the motif binding activity of other MRFs (Figure [185]S6E,F,
   Supporting Information).

Figure 5.

   Figure 5
   [186]Open in a new tab

   A regulatory enhancer region regulates Myog promoter activity and is
   inhibited by RelA binding. A) A distal co‐accessible enhancer region
   (E) associated with the cell type specific activity of the Myog
   promoter (P). B) Comparison of relative activity of co‐accessible
   enhancer region along with Myog promoter and gene activity. C)
   Correlation between the Myog enhancer and promoter accessibility. D)
   Enhancer accessibility and Myog expression from scATAC‐seq of E13.5
   embryos. E) Relative comparison of Myog enhancer accessibility and
   subsequent impact on its promoter and gene activity in scATAC‐seq from
   CT and MO embryos at E13.5. F) Location of RelA, P50, MYOG, MYOD1 and
   MEF2C binding motifs in the Myog enhancer region and relative
   accessibility from scATAC‐seq of CT and MO embryos.

   Together, we identified an enhancer region of Myog ≈4 kb upstream of
   its promoter, which is highly correlated with its promoter activity and
   expression. Additionally, the presence of a binding site on the
   enhancer and upregulation of motif activity suggest enhanced
   recruitment of RelA dimer to the enhancer region and the
   down‐regulation of Myog expression by inhibiting the enhancer function
   in MO embryos.

2.7. Binding of RelA to the Myog Enhancer Inhibits Its Expression and Impacts
Myotube Formation

   To confirm our findings from scRNA, scATAC‐seq, and spatial sequencing,
   we conducted RT‐qPCR analysis of E13.5 CT and MO embryonic tissue
   samples. The expression of Myod1 and Myog was lower in MO embryos
   (Figure [187]6A). Similarly, the protein levels of MYOG and MYH3 were
   lower in MO embryos, which aligns with the findings from scRNA and
   spatial Seq (Figure [188]6B). However, the mRNA expression of Myf5 did
   not deviate in the MO embryo (Figure [189]6A). In addition,
   immunofluorescence staining of E13.5 embryos with MYH3 antibody showed
   a smaller myotome in the cryosections of MO embryos relative to CT
   embryos (Figure [190]6C).

Figure 6.

   Figure 6
   [191]Open in a new tab

   Inhibition of Myog impaired myogenesis E13.5 embryonic tissues. A)
   Relative mRNA expression of myogenic regulatory factors Myog, Myod1 and
   Myf5 in embryos from CT and MO. B) Cropped Western Blots of MYOG and
   MYH3 (β‐Tubulin as loading control). C) Fluorescence image of embryonic
   myotome at E13.5 from CT and MO embryos visualized by MYH3+ cells. D)
   Cropped Western Blots of p‐RelA, and RelA (β‐Tubulin as loading
   control). E) Binding enrichment of RelA with the enhancer and promoter
   region of Myog from CT and MO embryos during E11.5 and E13.5. CT:
   Control group; MO: Maternal Obesity group. Data are shown as mean ±
   SEM, and each dot represents one litter (n = 5). ^∗ p < 0.05, ^∗∗ p <
   0.01, and ^∗∗∗ p < 0.001 in CON versus MO by one‐tailed unpaired
   Student's t‐test (a–e).

   RelA is a key mediator of canonical NF‐kB signaling, and its
   phosphorylation is an indication of its active state, which enhances
   its nuclear translocation.^[ [192]^31 ^] E13.5 embryos from MO mice had
   a higher ratio of p‐RelA/RelA (Figure [193]6D). By ChIP assay using an
   antibody against RelA, we found higher recruitment of RelA to the
   enhancer region of both E11.5 and E13.5 MO embryos (Figure [194]6E),
   which might suppress the binding of other transcription factors and
   inhibit Myog expression.

   Together, these analyses verify our findings from scRNA, scATAC and
   spatial seq that MO elevated the binding of RelA dimer to the enhancer
   region of Myog and inhibition of myogenesis.

2.8. Myog Expression and Myotube Formation are Impaired in Differentiated
Myoblasts Due to RelA

   The nuclear translocation of the RelA homodimer, or a heterodimer of
   RelA with another member of the NF‐κB family, is required for the
   activation of NF‐κB signaling and for binding to the regulatory regions
   of target genes to exert its transcriptional effects.^[ [195]^46 ^] The
   release of sequestrated RelA dimer by phosphorylation and
   ubiquitination of IκBα triggers its rapid translocation to nucleus,
   however the overexpression of RelA is also associated with its nuclear
   positivity in both in vivo and in vitro condition, independent of its
   phosphorylation status in the canonical NF‐kB signaling cascade.^[
   [196]^47 , [197]^48 ^] Therefore, to explore the role of RelA on
   myogenesis, we overexpressed RelA in C2C12 cell, as well as activated
   the NF‐kB signaling TNF‐a and evaluated their impact on myogenesis
   (Figure [198]S7A, Supporting Information). We used scrambled plasmid
   with GFP as a positive control to evaluate the transfection efficiency
   and non‐specific effect on myogenesis (Figure [199]S7B, Supporting
   Information). In the RelA overexpression (RelA OE) group, we observed a
   higher level of total RelA without changing its phosphorylation level,
   and the RelA level was unchanged in the TNF‐a group (Figure [200]7A).
   Consistently, we observed lower levels of MYOG and MYH3 in t‐a and RelA
   OE groups (Figure [201]7A). Immunofluorescence imaging also exhibited a
   lower number of myotubes in both the RelA OE group and TNF‐a treated
   group (Figure [202]7B). Similarly, RT‐qPCR also showed the inhibition
   of Myod1 and Myog expression in RelA OE and TNF‐a groups
   (Figure [203]7C). Activation of NF‐kB signaling by TNF‐a or RelA OE
   increased the translocation of RelA homodimer or RelA‐P50 heterodimer
   into the nucleus which increased their binding with enhancer region of
   Myog and inhibited myogenesis in C2C12 cells. Together, these data show
   that activation of canonical NF‐kB signaling impairs Myog expression
   and myotube formation.

Figure 7.

   Figure 7
   [204]Open in a new tab

   RelA inhibits Myog and myotube formation in differentiated myoblasts
   from C2C12 cells in vivo. A) Cropped western blots of MYOG, MYH3, RelA
   and p‐RelA (β‐tubulin was used as loading control) B) Immunostaining of
   myosin heavy chain (MHC). C) Relative mRNA expression of myogenic
   marker gene Myod1 and Myog. D) Binding enrichment of RelA at the distal
   enhancer region of Myog in C2C12 cells after activation of NF‐kB
   signaling activation by TNF‐a and RelA overexpression (RelA OE) by
   plasmid DNA. E) Immunostaining of myosin heavy chain (MHC). F) Relative
   mRNA expression of myogenic marker genes Myod1 and Myog. G) Cropped
   western blots of MYOG, MYH3, and RelA (β‐tubulin was used as loading
   control) in C2C12 cells with/without siRNA‐mediated RelA knockdown and
   TNF‐a in C2C12 myoblast cells. CON: Control, RelA KD: RelA knockdown.
   Data are presented as mean ± SEM, and each dot represents one
   independent experiment (n = 3). ^∗ p < 0.05, ^∗∗ p < 0.01, and ^∗∗∗ p <
   0.001 groups by one‐way ANOVA (n = 3).

   To further verify the role of canonical NF‐kB signaling on myogenesis,
   we knocked down RelA by siRNA following the activation of NF‐kB
   signaling by TNF‐a in C2C12 cells (Figure [205]S7C,D, Supporting
   Information). In the absence of TNF‐a, RelA KD only had minor effects
   on Myod1, and Myog expression, and myotube formation
   (Figure [206]7E,F). Consistently, NF‐kB activation by TNF‐a reduced the
   myotube formation in CON cells, which was absent in RelA KD cells
   (Figure [207]7E). Similarly, the mRNA levels of Myod1 and Myog were not
   impacted in RelA KD cells after TNF‐a addition (Figure [208]7F).
   Knockdown of RelA didn't activate the canonical NK‐kB signaling in RelA
   KD cells and didn't exhibit deleterious effects on myoblast
   differentiation nor myotube formation (Figure [209]7G; Figure [210]S7E,
   Supporting Information). Collectively, these data verify the crucial
   role of RelA in inhibiting Myog expression and myotube formation in
   C2C12 myogenic cells due to inflammation.

2.9. Metformin Treatment Attenuates NF‐kB Signaling and Rescues Myogenesis in
MO Embryos

   Proinflammatory cytokines, including TNF‐a, IL‐1, IL‐6, and IFN‐I are
   major activators of canonical NF‐kB signaling.^[ [211]^31 ^] To explore
   the impact of NF‐kB activation in committed myoblasts and their
   myogenic differentiation, we used TNF‐a to induce inflammation in
   confluent C2C12 myoblasts (Figure [212]S8A,B, Supporting Information).
   Furthermore, we used Caffeic Acid Phenethyl Ester (CAPE) and Metformin
   (MET) to inhibit the canonical NF‐kB pathway (Figure [213]S8A,
   Supporting Information). CAPE specifically inhibits the ubiquitination
   and phosphorylation‐mediated activation of the RelA dimer and prevents
   their translocation into the nucleus to regulate their target genes.^[
   [214]^49 ^] On the other hand, MET effectively inhibits the TNF‐a,
   HIF‐1a, and IL‐6‐mediated activation of inflammatory signaling in both
   AMP‐activated protein kinase (AMPK) and AMPK‐independent pathways.^[
   [215]^50 ^] Immunofluorescence imaging exhibited a lower number of
   myotubes in TNF‐a treated group, on the other hand, the number of
   myotubes were increased after addition of CPAE and MET (Figure
   [216]8A). Similarly, RT‐qPCR showed the inhibition of Myod1 and Myog
   expression in TNF‐a group, which was rescued by MET and CAPE
   (Figure [217]8B). Activation of NF‐kB signaling by TNF‐a increased the
   p‐RelA/RelA ratio, which reduced MYOG and MYH3 contents in the myogenic
   cells (Figure [218]8C), and CAPE and MET rescued the expression of MYOG
   and MYH3 by reducing the p‐RelA/RelA ratio (Figure [219]8C). Notably,
   the number of differentiated myoblasts was comparatively lower in the
   TNF‐a group, indicating the negative impact of inflammation on myogenic
   differentiation (Figure [220]S8C, Supporting Information). Together,
   these data show that activation of canonical NF‐kB signaling impairs
   Myog expression and myotube formation.

Figure 8.

   Figure 8
   [221]Open in a new tab

   Metformin attenuates RelA activation and rescues Myog expression and
   myogenesis in vitro and E13.5 embryos. A) Immunostaining of myosin
   heavy chain (MHC). B) Relative mRNA expression of myogenic marker genes
   Myod1 and Myog. C) Cropped western blots of MYOG, MYH3, p‐RelA and RelA
   (β‐tubulin was used as loading control) after activation of NF‐kB
   signaling by TNF‐a and its inhibition with CAPE (Caffeic Acid Phenethyl
   Ester) and MET (Metformin) treatments in C2C12 myoblast cells. D)
   Metformin treatment of obese mice and collection of E13.5 embryos. E)
   Cropped Western Blots of p‐RelA, and RelA (β‐Tubulin as loading
   control). F) Relative mRNA expression of myogenic regulatory factors
   Myog and Myod1 in embryos. G) Cropped Western Blots of MYOG and MYH3
   (β‐Tubulin as loading control) from CT (n = 6 mice per group), MO (n =
   6 mice per group), and MO‐MET (n = 6 mice per group) groups. H) A
   schematic diagram summarizes the RelA dimer‐dependent mechanism
   impairing Myog expression and embryonic myogenesis due to maternal
   obesity. Data are presented as mean ± SEM, and each dot represents one
   litter. ^∗ p < 0.05, ^∗∗ p < 0.01, and ^∗∗∗ p < 0.001.

   Metformin (MET) has long been used to control pre‐gestational and
   gestational diabetes, and prevailing data suggest additional health
   benefits of in‐utero exposure to offspring.^[ [222]^51 , [223]^52 ^]
   Therefore, we treated MO mice (Figure [224]S8D, Supporting Information)
   with MET from one week before breeding to E13.5 for embryonic tissue
   collection (Figure [225]8E). Embryos from Met‐treated embryos didn't
   differ in their body weight (Figure [226]S8E,F, Supporting
   Information), however, it reduced the inflammatory response in MO
   embryos as evidenced by the reduction of the p‐RelA /RelA ratio
   (Figure [227]8E). Excitingly, MET recovered Myod1 and Myog expression
   in MO embryos (Figure [228]8F). Similarly, the protein contents of MYOG
   and MYH3 were elevated in MO embryos treated with Met (Figure [229]8G).
   In short, MET treatment of obese maternal mice suppressed inflammation
   and the canonical NF‐kB signaling in embryos, which rescued embryonic
   myogenesis (Figure [230]8H).

3. Discussion

   The obesity pandemic in pregnant women is well documented to exert
   negative impacts on the development of skeletal muscle, including
   reduced muscle strength, endurance capacity, and smaller muscle mass of
   offspring.^[ [231]^26 , [232]^53 , [233]^54 , [234]^55 , [235]^56 ,
   [236]^57 ^] The prenatal stage is critically important for skeletal
   muscle development.^[ [237]^58 ^] The primary myofibers form during the
   embryonic stage and lay the foundation for later muscle fiber
   development. Here, using multiomics tools, including scRNA‐seq,
   scATAC‐seq, and spatial transcriptomic profiling, we comprehensively
   analyzed the myogenic lineage specification and differentiation in
   mouse embryos. Our data suggest the existence of a distal
   cis‐regulatory enhancer region of Myog, which mediates the inhibition
   of embryonic myogenesis induced by MO.

   Embryonic myogenesis can be separated into myogenic commitment and
   differentiation. During E11.5, we identified the downregulation of
   Myod1 in myoblasts, but didn't observe a change in proliferating
   myoblast cell populations in MO embryos. Proliferating myoblasts also
   express Myf5, which precedes Myod1 expression and rapidly downregulates
   during differentiation; Myf5 and Myod1 share similar binding sites and
   functionalities to induce the transcription of genes related to
   myogenic differentiation.^[ [238]^59 ^] We identified a lower
   percentage of differentiated myoblasts in the MO embryos.^[ [239]^19 ,
   [240]^60 ^] More importantly, we observed downregulation of Myog in
   differentiated myoblasts at both E11.5 and E13.5 stages due to MO.^[
   [241]^9 ^] We also identified a lower percentage of differentiated
   myoblasts and myotubes in the MO embryos. Consequently, our data
   suggest that the MO mainly affects the terminal differentiation and
   myotube formation during embryonic myogenesis while having less effect
   on the myogenic commitment and proliferation.

   Expression of Myog is regulated by various transcription factors
   including MYF5 and MYOD1.^[ [242]^10 , [243]^61 , [244]^62 ^] These
   transcription factors bind to different conserved DNA binding motifs on
   the Myog promoter, which regulate its context‐dependent expression.^[
   [245]^62 ^] In scATAC‐seq, we identified an enhancer region that is
   −4 kb upstream to the Myog TSS, and its location overlaps with a region
   recently confirmed as an active motif binding site by Self‐transcribing
   active regulatory region sequencing (STARR‐seq) in murine embryonic
   stem cells.^[ [246]^63 ^] Global Run‐on sequencing (GRO‐seq),
   STARR‐seq, and several other methods can measure enhancer activity, but
   have limitations in identifying the enhancer‐promoter pair.^[ [247]^64
   ^] Our CCAN analysis identified the Myog enhancer‐promoter pair, which
   is highly correlated with Myog activity. Additionally, the
   accessibility of this enhancer region was highly cell‐type‐specific and
   only accessible in the myogenic cells. These enhancer regions also
   possess binding sites of transcription factors MYOD1, MEF2C, and even
   MYOG to regulate Myog expression. One of the potential roles of this
   enhancer region is likely to fine‐tune the Myog expression depending on
   cell states. During the decline of Myod1 expression after the myogenic
   commitment, co‐transcription factors like MEF2C bind to these regions
   to up‐regulate Myog expression. Additionally, MYOG has its
   autoregulatory positive feedforward mechanism in combination with
   MEF2C.^[ [248]^9 , [249]^65 ^] We observed downregulation of MEF2C
   expression in both differentiated myoblasts and Myotubes, as well as
   lower activity in their binding sites on the enhancer due to MO, which
   might reduce the overall Myog expression. These observations in
   combined with other reports confirm the existence of an enhancer region
   that is correlated with Myog expression.

   Obesity, metabolic diseases, and several pathological conditions in
   pregnant mothers including pelvic inflammatory disease (PID) and
   inflammatory bowel disease (IBD) can induce an inflammatory condition
   in the intrauterine microenvironment and activate inflammatory response
   in embryos.^[ [250]^25 ^] In this study, we found that RelA activity
   was highly up‐regulated among the members of the NF‐kB family due to MO
   and inhibition of RelA with Metformin both in vivo and in vitro rescued
   myogenesis. These data are concordant with in vitro studies that
   revealed increased myogenic activity in RelA^−/− C2C12 cells and
   primary mouse embryonic fibroblast cells.^[ [251]^31 ^] Importantly, we
   discovered a novel enhancer containing RelA and P50 binding sites in
   the vicinity of Myog promoter and ChIP‐PCR confirmed higher binding of
   RelA dimer to the enhancer region in MO embryos. These findings suggest
   a potential mechanism leading to Myog down‐regulation during
   embryogenesis.

   Metformin has been widely used in the treatment of gestational diabetes
   in women and comes with multiple benefits to both mothers and their
   embryos because of its ability to cross‐placenta.^[ [252]^66 , [253]^67
   ^] Pre‐pregnancy and pregnancy treatment with metformin showed variable
   effects on offspring at different stages of their life.
   Periconceptional metformin treatment improves embryonic myogenesis
   which was associated with the inhibition of canonical NF‐kB signaling,
   consistent with our previous discovery that metformin suppresses
   fibrogenesis in E13.5 MO embryos.^[ [254]^27 , [255]^68 ^] Recent
   studies showed that metformin administration induced fetal growth
   restriction in murine and primate models during late gestation,^[
   [256]^69 , [257]^70 ^] however, metformin administration during the
   first trimester has not been associated with developmental
   abnormalities or adverse outcomes at both fetal and adult stages.^[
   [258]^71 , [259]^72 , [260]^73 ^] These results highlight the
   periconceptional use of metformin as a potential therapeutic agent to
   mitigate maternal inflammatory responses and improve embryonic
   myogenesis for obese mothers.

   In summary, we found that MO impairs embryonic myogenesis via
   inhibition of Myog and terminal myogenic differentiation. Canonical
   NF‐kB signaling is a key mediator that inhibits Myog expression through
   the recruitment of RelA dimer to the enhancer region. Our findings also
   show that NF‐kB is a potential therapeutic target to rescue embryonic
   myogenesis and prevent the long‐term impact of MO on offspring health.

4. Experimental Section

Mice

   Wild‐type C57BL/6J female mice were purchased from The Jackson
   Laboratory (Bar Harbor, ME, USA). To induce obesity, wild‐type C57BL/6J
   female mice (Jackson Lab., Bar Harbor, ME, USA) were assigned to two
   groups randomly at 8 weeks of age. CT group was provided a control diet
   containing 10% of energy from fat (D12450H, Research Diets, New
   Brunswick, NJ, USA), and the MO group was provided a high‐fat diet
   containing 45% of energy from fat (D12451, Research Diets) ad libitum
   for 10 weeks. Both groups were then mated with randomly selected
   C57BL/6J males, and mating was confirmed by the presence of a vaginal
   plug in the morning. At E11.5 and E13.5, embryos were collected after
   euthanizing the mice by CO[2] inhalation and cervical dislocation
   following AVMA Guidelines (2020). All animal studies were conducted in
   WSU following IACUC‐approved protocol: 6704.

Single‐Cell RNA Sequencing

   Embryos at the same developmental stage were collected during each of
   the time points, E11.5, and E13.5, and the developmental stages were
   confirmed following the Theiler stages 19 and 21 with somite numbers
   45–47 and 52–55, respectively. By using a dissection microscope (Leica
   DFC450C), the head and first 7 somite pairs and internal organs were
   removed. Embryonic tissues were incubated at 37 °C for 15 min with
   TryLE Express dissociation reagent (Gibco; Cat# 12 605.010) after
   mincing finely, and the digesta was filtered through a 40 µm cell
   strainer. According to 10 × Genomics protocol, libraries were
   constructed by using Chromium Single Cell 3′ Library & Gel Bead Kit v3
   and sequenced in an Illumina NovaSeq 600S4 as described in our previous
   study.^[ [261]^27 ^]

Processing of scRNA‐Seq Data and Differential Gene Expression Analysis

   Raw sequence was preprocessed by using Cell Ranger V7.0 and mapped with
   the mouse mm10 reference transcriptome. R package Seurat V 4.3.0 was
   used to merge libraries from two embryonic stages, and all cells
   possessing UMIs >200 000 and <500 genes were removed; cells with >20%
   mitochondrial genes were removed, including doublets, for further
   analysis. After global normalization, scaling, and reducing the linear
   dimensionality, top 35 components were processed using Seurat to
   visualize in a 2D map via Uniform Manifold Approximation (UMAP). The
   pseudo‐bulk single cell aggregation technique with edgeR assessed gene
   expression differences between cell clusters and two groups over all
   clusters. All readings were aggregated using “Seurat2PB” to create the
   pseudo‐bulk sample. To determine DEGs between CT and MO groups, we
   employed quantile‐adjusted conditional maximum likelihood (qCML) with a
   threshold value of p < 0.05 after normalizing aggregated data.

Gene Ontology and KEGG Pathway Enrichment Analysis

   For gene ontology analysis, we used the R package “clusterProfiler” and
   the “gseGO” function for identifying enriched biological and molecular
   functions associated with the differentially expressed genes between
   the CT and MO groups. “ChEA3” was used to identify enriched
   transcription factors associated with the upregulated genes in myogenic
   clusters of MO embryos.

Single‐Cell ATAC Seq

   Single nuclei were isolated by following protocol [262]CG000124 from
   10x genomics. Briefly, the single‐cell suspension from 13.5 embryos was
   prepared using a similar protocol for scRNA‐seq as described above and
   diluted to a concentration of 100 cells uL^−1. To obtain 2  ×  10^6
   cells, 200 uL of the cell suspension was taken and pelleted by
   centrifuging at 300 ×g for 5 min at 4 °C. Cells were resuspended in
   200uL lysis buffer (10 mm Tris‐HCl pH 7.4, 10 mm NaCl, 3 mm MgCl[2] and
   1x PBS) and incubated on ice for 1 min. The suspended cells were then
   washed with Nuclei wash buffer (10x Genomics PN‐ [263]CG000124),
   resolved in resuspension buffer (10x Genomics PN‐ [264]CG000124), and
   centrifuged at 500xg for 10 min at 4 °C twice. The pellets were
   resuspended in nuclei wash buffer and resuspension buffer, and isolated
   nuclei were inspected visually, and the nuclei concentration was
   assessed and proceeded for library preparation according ×10 Genomics
   single‐cell ATAC‐Seq protocol and sequenced.

Processing of scATAC‐Seq Data

   Raw scATAC‐seq data were analyzed by Cell Ranger ATAC (V 2.1.0) and
   aligned with the mouse genome reference (mm10). R package Signac
   (V1.14.0) was used to process data from the CT and MO groups. Data from
   both groups were merged by using the “merge” function of Signac and
   normalized by using Term frequency‐inverse document frequency (TF‐IDF).
   Dimensionality reduction was conducted by singular value decomposition
   (SVD), and cells were plotted in non‐linear dimension reduction for
   visualization by the “RunUmap” function, similar to scRNA seq. Gene
   activity matrices were created by “GeneActivity” function of Signac and
   merged with ATAC data set as a new assay. To predict the cell types of
   scATAC‐seq clusters, we transferred the cell label from our scRNA‐seq
   data by cross‐modality integration. The differentially accessible peaks
   between cell types were computed with the “FindMarkers” function in
   Signac and filtered by logfc.threshold  =  0.2 and p‐value level p <
   0.05 were visualized by the default coverage plot function.

Motif Activity Analysis

   Transcription factor activity was analyzed using the positional weight
   matrix of the motif in the JASPER2020 database and using ChromVAR.
   Motif information from the JASPER database was merged with the Signac
   Seurat object and the differential activity of motifs was calculated
   using the “FindMarekrs” function. An adjusted p‐value level P < 0.05
   was used to filter the top differentially active motifs, and the
   “FindMotifs” function was used to test their fold enrichment and
   plotted by the “MotifPlot” function.

Finding Co‐Accessible Sites

   Cicero package was used for identifying CCANs using the scATAC‐seq data
   of myogenic clusters. The myogenic cell cluster data set extracted from
   the Signac Seurat object were converted to CellDataSet format and
   cicero object, respectively. The “run_cicero” function was used to find
   the CCAN network and merged with the dataset to visualize along with
   the gene accessibility coverage plot.

Spatial Transcriptomic Sequencing

   For special transcriptomic sequencing, a total of 6 embryos from each
   of the CT and MO groups were selected, where each embryo was randomly
   selected from one pregnant mouse. Samples were fixed with 10% formalin
   and embedded with paraffin following formalin‐fixed paraffin‐embedded
   tissue (FFPE) protocol. Embedded tissues were sectioned at a thickness
   of 5 µm and mounted on glass slides, dewaxed and incubated overnight
   with GeoMx RNA detection probes. MYH3 fluorescent‐tagged antibody were
   used to identify myotome and locate regions‐of‐interest (ROIs). The
   photocleaved indexing oligos were then collected in the GeoMx Digital
   Spatial Profiler (Nanostring, Seattle, WA, USA) by profiling ROIs using
   region‐specific cleaving. To generate a digital quantification of the
   RNA expression, cleaved indices were sequenced and quantified using
   Illumina Next Sequeq 2K (San Diego, CA, USA). The downstream analysis
   was conducted by GeomxTools 3.10.0 through following similar procedure
   used in our previous study.^[ [265]^4 ^]

Cell Culture

   C2C12 (ATCC CRL‐1772) myoblast cells were incubated with a growth
   medium composed of DMEM with 10% FBS (Gibco # 10.439.001) and 1%
   penicillin‐streptomycin (Sigma, #P0781). When reached 100% confluency,
   growth medium was supplemented with 2% horse serum to induce myogenic
   differentiation. The canonical NF‐kB signaling was activated in the
   differentiating myoblasts by TNF‐a (5 ng mL^−1). In addition, 500 µm
   MET (Metformin; MP Biomedicals, Cat# 151 691) or 10 µm CAPE (Caffeic
   acid phenethyl ester) were used to inhibit the canonical NF‐kB
   signaling. After 6 days, the myotubes were visualized by
   immunohistochemical staining using myosin heavy chain (MHC) antibody
   (MF‐20, DSHB, Iowa City, IA).

RelA Overexpression

   To overexpress RelA, RelA cFlag pcDNA3 (Addgene, Plasmid#20012) were
   used in C2C12 cells and GFP plasmid as a negative control for
   evaluation of transfection efficiency. Plasmid transfection was
   performed using Lipofectamine 3000 transfection kit (Invitrogen)
   according to manufacturer's protocol. Briefly, in 24‐well plate when
   the cell reached to 60–70% confluency, 500 ng of plasmid DNA for each
   well were mixed with P3000 reagent in Opti‐MEM reduced serum media and
   diluted with equal amount of lipofectamine 3000. The DNA‐lipid complex
   was incubated for 15 min in RT and added to cells for transfection. The
   transfection media were replaced with regular growth medium after 24 h
   of incubation and the transfection efficiency were evaluated based on
   the expression of GFP.

RelA Knockdown

   For RelA knockdown in C2C12 myoblast cells, we used RelA Pre‐designed
   Silencer Selected SiRNA (ID: s72857, Invitrogen, CA, USA) and a
   non‐targeting negative control siRNA. Briefly, 5 nmol of oligo siRNA
   was diluted to a final concentration of 20 µm. C2C12 cells in the
   second day of incubation were incubated with a Lipofectamine 3000
   Transfection Reagent (Invitrogen) and siRNA (25 nm) in OPTI‐MEM Reduced
   Serum Medium (Gibco). The transfection medium was replaced by the
   growth medium after 24 h of incubation.

RT‐qPCR

   Following DNase treatment, total mRNA was extracted from tissue or
   cells by TRIzol reagent (Sigma, St. Louis, MO), and cDNA was prepared
   using an iScript cDNA Synthesis Kit (Cat# 1708841, BIO‐RAD) by reverse
   transcription. The RT‐PCR was conducted using SsoAdvanced Universal
   SYBR Green Supermix (Bio‐Rad, Cat# 172570) by iQ5 RT‐PCR detection
   instrument (Bio‐Rad Laboratories, Hercules, CA). The relative
   expression of mRNA was determined using the ΔΔ− Ct method after
   normalization to the 18S RNA. Primer sequences used in this study are
   listed in Table [266]S6 (Supporting Information).

Western Blotting

   Western blot analyses were used to analyze the protein content.
   Proteins from homogenized tissues and cells were extracted using lysis
   buffer (100 mm Tris–HCl, pH 6.8, 2.0% SDS, 20% glycerol, 0.02%
   bromophenol blue, 5% 2‐mercaptoethanol, 100 mm NaF and 1 mm Na3VO4),
   and separated by 10% sodium dodecyl sulfate‐polyacrylamide gels
   electrophoresis (SDS‐PAGE) as described previously.^[ [267]^27 ^]
   Antibodies used for western Blotting are listed in the key resources
   table.

ChIP‐qPCR Assay

   For ChIP‐qPCR, sample fixation, chromatin isolation,
   immunoprecipitation, and DNA isolation were done by using SimpleChIP
   Enzymatic Chromatin IP Kit (Cell Signaling Technology, #9003) following
   the manufacturer's protocol. Chromatin was fragmented through
   sonication, using 6 × 10s impulses with 1 min pauses. RelA (NF‐κB p65;
   PCRP‐RELA‐2B6‐s, DSHB, Iowa City, IA) antibody was used for overnight
   incubation, and spin‐column purification method was used for DNA
   purification. The purified DNA was then used as a template for qPCR to
   assess the enrichment of specific target regions. Primers used to
   amplify the Myog promoter and enhancer regions in ChIP‐qPCR are listed
   in the Table [268]S7 (Supporting Information).

Statistical Analysis

   A total of 15 randomly selected embryos from three pregnant mice for
   the CT and MO groups were collected respectively for single‐cell RNA
   sequencing at E11.5 and E13.5 time points and single‐cell ATAC
   sequencing at E13.5. The transformation, normalization, quality control
   method and software used for pre‐processing of sequencing data are
   described above in the “Processing of scRNA‐seq data” and “Processing
   of scATAC‐seq data” sections. On the other hand, a total of five
   pregnant mice from each of the CT, MO, and MO‐MET groups were used to
   collect embryos at E 11.5 and E13.5 for biochemical analysis. Each
   pregnancy was considered as an experimental unit. For cell culture
   studies, three separate experiments (n = 3) were carried out with 3
   technical replicates for each experiment. Data were analyzed using R
   version 4.3.1, and student's t‐test or one‐way ANOVA was used to
   compare statistical significance between or among groups. Data are
   reported as mean ± SEM and visualized using GraphPad Prism version
   8.0.0 for Windows (GraphPad Software, Boston, Massachusetts, USA).
   Statistical differences were indicated as ^∗ p < 0.05, ^∗∗ p < 0.01,
   and ^∗∗∗ p < 0.001.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   M.N.H., M.D. designed the study methodology, M.N.H., Y.G., S.I., L.C.,
   X.L., Z.K, and J.M.D.A. conducted the experiments M.N.H., and N.L.
   analyzed the sequencing data N.L and M. Z. provided key experimental
   resources. M.N.H. prepared the original draft and M.D. reviewed &
   edited with input from all the other authors.

Supporting information

   Supporting Information
   [269]ADVS-12-e03712-s001.pdf^ (2.1MB, pdf)

   Supporting Information
   [270]ADVS-12-e03712-s002.xlsx^ (31.3MB, xlsx)

Acknowledgements