Differential regulation of cranial and cardiac neural crest by serum response factor and its cofactors

  1. Colin J Dinsmore
  2. Philippe Soriano  Is a corresponding author
  1. Department of Cell, Development and Regenerative Biology, Icahn School of Medicine at Mount Sinai, United States

Abstract

Serum response factor (SRF) is an essential transcription factor that influences many cellular processes including cell proliferation, migration, and differentiation. SRF directly regulates and is required for immediate early gene (IEG) and actin cytoskeleton-related gene expression. SRF coordinates these competing transcription programs through discrete sets of cofactors, the ternary complex factors (TCFs) and myocardin-related transcription factors (MRTFs). The relative contribution of these two programs to in vivo SRF activity and mutant phenotypes is not fully understood. To study how SRF utilizes its cofactors during development, we generated a knock-in SrfaI allele in mice harboring point mutations that disrupt SRF-MRTF-DNA complex formation but leave SRF-TCF activity unaffected. Homozygous SrfaI/aI mutants die at E10.5 with notable cardiovascular phenotypes, and neural crest conditional mutants succumb at birth to defects of the cardiac outflow tract but display none of the craniofacial phenotypes associated with complete loss of SRF in that lineage. Our studies further support an important role for MRTF mediating SRF function in cardiac neural crest and suggest new mechanisms by which SRF regulates transcription during development.

Editor's evaluation

This carefully executed study demonstrates new mechanisms by which Serum Response Factor (Srf) regulates transcription. The authors report the effects that loss of Srf function has on different neural crest lineages in the mouse. The results convincingly show that the main function of Srf within neural crest is in the cardiac neural crest lineage where it regulates cytoskeletal genes.

https://doi.org/10.7554/eLife.75106.sa0

Introduction

Multicellular development requires the precise management of cellular behaviors including proliferation, migration, and differentiation. These are coordinated through intercellular communication pathways, such as growth factor signaling, that couple extracellular information with internal effectors, including transcription factors (TFs) (Fantauzzo and Soriano, 2015; Lemmon and Schlessinger, 2010). The balance between opposing transcription programs is tuned by signaling pathways which activate specific TFs or in some cases cofactors that direct the behavior of a common TF. One example of the latter is the essential transcription factor Serum Response Factor (SRF) (Posern and Treisman, 2006). SRF is necessary for the expression of immediate early genes (IEGs) in cells stimulated with serum or growth factors, as well as many genes related to the actin cytoskeleton, contractility, and muscle differentiation.

SRF binds a conserved DNA regulatory sequence known as a CArG box, a motif found at many cytoskeletal and growth-factor inducible gene promoters (Mohun et al., 1991; Norman et al., 1988; Sun et al., 2006b). SRF can, however, effect at least two unique transcriptional programs by coupling with two families of cofactors that compete for a common binding site on SRF itself (Miano, 2003; Wang et al., 2004 Figure 1A). The ternary complex factors (TCFs) are E26 transformation-specific (ETS) family proteins activated by extracellular signal-regulated kinase 1/2 (ERK1/2) phosphorylation (Mylona et al., 2016). Once activated, they bind DNA and promote cellular proliferation by transcribing IEGs in coordination with SRF (Esnault et al., 2017; Gualdrini et al., 2016). There are three TCF members in mouse and human: ELK1, ELK3/NET, and ELK4/SAP1 (Posern and Treisman, 2006). Opposing SRF-TCF activity are the Myocardin Related Transcription Factors (MRTFs). These cofactors rely on SRF to bind DNA, promote cytoskeletal gene expression, and are particularly important in muscle differentiation (Posern and Treisman, 2006). MRTFs bind to and are inhibited by G-actin. Polymerization of G-actin into F-actin liberates MRTFs to translocate to the nucleus and bind SRF (Miralles et al., 2003). This can be promoted by multiple signaling pathways, including phosphoinositide 3-kinase (PI3K), that stimulate guanine nucleotide exchange factors to activate F-actin-promoting Rho-family GTPases (Brachmann et al., 2005; Hanna and El-Sibai, 2013; Jiménez et al., 2000; Vasudevan and Soriano, 2014). MRTFs are also positively and negatively regulated by extensive phosphorylation (Panayiotou et al., 2016). Three MRTFs are known to interact with SRF: Myocardin itself, MRTF-A/MKL1/MAL, and MRTF-B (Parmacek, 2007). Myocd is expressed specifically in muscle while Mrtfa and Mrtfb are more broadly expressed (Posern and Treisman, 2006). A fourth MRTF, MAMSTR/MASTR, interacts with MEF2 proteins and is not known to bind SRF (Creemers et al., 2006).

Figure 1 with 2 supplements see all
Loss of Srf in NC affects cytoskeletal gene expression.

(A) Diagram depicting SRF, its TCF and MRTF cofactors, and the upstream signals that regulate them. (B) DAPI stained embryos at E11.5 and E12.5 show a facial cleft following loss of Srf in NC. Scale bar represents 1 mm. (C) Diagram depicting RNA-sequencing strategy. (D) Volcano plot showing DEGs in Srf NC conditional mutants. Genes with a P-value < 0.01 and log2 fold change (FC) >0.25 are colored. Select genes are labeled. (E) A heatmap of the top 25 DEGs by q value. The samples cluster by genotype and are color-coded by Z-score. (F) Gene set enrichment analysis (GSEA) using a list of DEGs with q ≤ 0.05 and Log2FC ≤–0.25. Enrichment for ENCODE TF ChIP-Seq, GO Cellular Component, and GO Biological Process are shown. (G) GSEA for known SRF, MRTF, and TCF ChIP targets from previous datasets (Esnault et al., 2014; Gualdrini et al., 2016) across our entire dataset. (H) Overlap of known MRTF and TCF targets with DEGs q ≤ 0.05. (I) Absolute value of log2 FC for DEGs that overlap with each category. Horizontal bar indicates the mean (0.385 Shared, 0.378 MRTF, 0.219 TCF).

Srf and its cofactors have been extensively studied genetically. Srf-/- mutant mice die between E6.5 and E8.5 showing defects in mesoderm formation (Arsenian et al., 1998; Niu et al., 2005). Cofactor knockouts are comparatively mild. Single TCF mutants are all fully or partially viable (Ayadi et al., 2001; Cesari et al., 2004; Costello et al., 2004; Weinl et al., 2014) and Elk1; Elk3; Elk4 triple null embryos have not been described in detail but survive until E14.5 without obvious defects (Costello et al., 2010; Gualdrini et al., 2016). Mrtfa-/- mutant mice are viable (Li et al., 2006; Sun et al., 2006a) whereas Mrtfb-/- mice are inviable between E13.5-E15.5, exhibiting cardiovascular defects (Li et al., 2012; Oh et al., 2005). Myocd-/- mice have the most severe phenotype and die at E10.5, also from cardiovascular defects (Espinoza-Lewis and Wang, 2014; Li et al., 2003). Mrtfa; Mrtfb double null mice have not been described, but conditional double mutants have shown these factors exhibit redundancy and broadly phenocopy loss of Srf in several tissues and cell types (Cenik et al., 2016; Guo et al., 2018; Li et al., 2005a; Trembley et al., 2015). However, studies comparing Srf and Mrtfa; Mrtfb mutants are not always identical. In megakaryocytes, loss of Mrtfa and Mrtfb is more severe than loss of Srf and there are large gene expression differences in the two models (Smith et al., 2012; Halene et al., 2010). Indeed, there is evidence that MRTFs may regulate genes independent of SRF or act as cofactors for TFs other than SRF (Asparuhova et al., 2011; Kim et al., 2017). Whether the differences in Srf versus Mrtfa; Mrtfb loss-of-function studies are due to SRF-TCF activity, SRF-independent MRTF activity, or TCF/MRTF-independent SRF activity remains uncertain. These studies are summarized in Supplementary file 1.

One tissue in which SRF was found to be essential is the neural crest (NC) (Newbern et al., 2008; Vasudevan and Soriano, 2014). The NC is a transient developmental population of cells that arises from the dorsal neural tube, migrates ventrally throughout the embryo, and gives rise to numerous cell types including the bone and connective tissue of the face, as well as smooth muscle cells in the cardiac outflow tract (Bronner and Simões-Costa, 2016). The extensive migration, proliferation, and various differentiation outcomes these cells undergo requires accurate coordination, and decades of study have revealed a panoply of signaling pathways and transcription factors important in these processes, including the fibroblast growth factor (FGF) pathway, platelet-derived growth factor (PDGF) pathway, and SRF itself (Brewer et al., 2015; Dinsmore and Soriano, 2018; Newbern et al., 2008; Rogers and Nie, 2018; Tallquist and Soriano, 2003; Vasudevan and Soriano, 2014). We and others have previously shown that Srf is required in the NC for craniofacial and cardiovascular development (Newbern et al., 2008; Vasudevan and Soriano, 2014). Intriguingly, mice homozygous for a hypomorphic allele of Mrtfb die shortly after birth with cardiac outflow tract defects and can be rescued by a neural crest-specific transgene (Li et al., 2005b). Assays in mouse embryonic palatal mesenchyme cells (MEPMs) indicated that stimulation with the secreted ligands fibroblast growth factor (FGF) or platelet-derived growth factor (PDGF) promoted SRF-TCF complex formation, but only PDGF promoted SRF-MRTF interactions, in a PI3K-dependent manner (Vasudevan and Soriano, 2014). Supporting the importance of SRF-MRTF interactions, Srf interacted genetically with Pdgfra but not Fgfr1 in NC (Vasudevan and Soriano, 2014). However, the contributions of each SRF-cofactor transcriptional program to the overall Srf NC phenotype are unclear.

In this study, we further characterize the molecular consequences of losing Srf expression in the NC through marker analysis and expression profiling, finding the most highly downregulated genes to be cytoskeletal in nature. We then test the presumed requirement of SRF-MRTF interactions using a novel SrfaI allele carrying mutations that prevent SRF-MRTF-DNA ternary complex formation to circumvent MRTF redundancy and control for possible SRF-independent MRTF activity. These embryos have striking developmental defects that are outwardly similar to Myocd mutant mice. Conditional NC mutants reveal an essential role for optimal SRF-MRTF activity in the cardiac crest, whereas the mutation is well-tolerated in the cranial NC. These observations raise the possibility that non-cardiovascular tissue may be able to develop with only minimal SRF-MRTF activity or that SRF can support cytoskeletal gene expression on its own or with other cofactors.

Results

Srfflox/flox; Wnt1-CreTg/+ mice develop a midfacial cleft and bleb, characterized by reduced cytoskeletal gene expression

To establish a phenotypic baseline for embryos lacking Srf in NC, we first examined conditional null embryos at E11.5 and E12.5 and assessed their morphology. Consistent with our previous study, a midfacial cleft develops from E10.5 to E11.5, becoming prominent at E11.5 as a failure of the medial nasal process and lateral nasal process (MNP and LNP, respectively) to converge at the midline (Figure 1B–C; Vasudevan and Soriano, 2014). By E12.5, a fluid-filled bleb develops at the midline, often with hemorrhaging into the midfacial cavity (Figure 1B). Embryos turned necrotic starting at E12.5 and did not survive past E13.5.

We sought to better understand the molecular defects that underly this outcome. As Srf has been implicated in mediating cell differentiation, we asked whether early craniofacial patterning was affected. However, expression of the differentiation markers Msx1 (craniofacial mesenchyme), Alx3 (MNP and LNP mesenchyme, medial mandibular mesenchyme), and Six3 (ventral forebrain, nasal placode, eye), as well as the markers of patterning centers Shh (ventral forebrain, weak oral MNP, and mandibular epithelium), and Fgf8 (ventral forebrain, oral MNP, and epithelium) were all unaffected at E10.5 as assessed by in situ hybridization, suggesting craniofacial patterning was largely normal at this stage (Figure 1—figure supplement 1).

We next sought to identify differentially expressed genes (DEGs) through bulk RNA-sequencing of control and mutant frontonasal prominences (FNP, i.e. MNP+ LNP) and mandibles at E11.5. To confirm the quality of the dataset and suitability of the analysis pipeline, we first compared mandible versus FNP gene expression among all samples and identified differentially expressed transcripts encoding 4084 DEGs (q ≤ 0.05, Wald test), among them known regulators of mandible or FNP identity, such as Hand2 and Six3 (Figure 1—figure supplement 2A-B). Principal component analysis showed strong separation of the samples by tissue (Figure 1—figure supplement 2C). We next identified DEGs in control versus Srfflox/flox; Wnt1-CreTg/+ mandibles and FNPs. Mandibles showed 40 DEGs and FNPs 219 (q ≤ 0.05, Wald test). A joint model including both tissue samples and accounting for tissue-of-origin identified 381 DEGs (Figure 1D–E). Srf itself was among the top DEGs, confirming efficient conditional deletion in the cranial NC (Figure 1D–E, Figure 1—figure supplement 2D), but SRF cofactors were not affected (Figure 1—figure supplement 2E).

The most differentially expressed genes primarily encoded cytoskeletal genes that were known targets of SRF-MRTF activity, including Actg1, Cnn2, Vcl, Actb, and Cfl1 (Figure 1D–E). We subjected a more stringent list of 43 downregulated and 36 upregulated genes with q ≤ 0.05 and Log2FC ≥ 0.25 to gene set enrichment analysis using the online tool Enrichr (Xie et al., 2021). Downregulated genes were enriched for cytoskeletal GO terms and SRF-binding motifs (Figure 1F), whereas upregulated genes showed little enrichment for either TF motifs or GO terms and may not be direct SRF targets (Figure 1—figure supplement 2G). We then used gene set enrichment analysis to compare our results with known targets of SRF, MRTF, and TCF (Esnault et al., 2014; Gualdrini et al., 2016; Supplementary file 1). All three gene lists showed enrichment in our dataset, but the MRTF list was most significantly enriched (Figure 1G). Furthermore, limiting this comparison to DEGs with q ≤ 0.05, genes bound by MRTF or MRTF and TCF were more significantly affected (greater fold change) than those bound by TCF alone (Figure 1H–I). We also performed these analyses on the individual FNP and Mandible datasets and found similar enrichment for SRF motifs, cytoskeleton-related GO terms, and a stronger enrichment for known MRTF targets than for TCF targets (Figure 1—figure supplement 2H-I). The one major difference between the tissues was that the FNP dataset contained a group of uniquely affected genes that enriched for terms related to cholesterol metabolism, but these were not further investigated (Figure 1—figure supplement 2H, GO Biological Process).

In summary, our gene expression analysis found that the genes most affected by loss of Srf in both mandible and FNP were enriched for cytoskeleton-related established SRF-MRTF targets. These data, coupled with our previous observations that Pdgfra interacted genetically with Srf in NC and PDGF stimulation promoted SRF-MRTF complex formation, led us to hypothesize that SRF-MRTF interactions would be critical for midfacial development.

SrfaI/aI succumb during early organogenesis with cardiovascular defects

In order to test the requirement for SRF-MRTF interactions genetically, we introduced four knock-in point mutations to the αI helix of the SRF DNA-binding domain previously shown to disrupt SRF-MRTF-DNA ternary complex formation while leaving SRF-TCF-DNA complex formation unaffected (Figure 2—figure supplement 1A-C; Hipp et al., 2019; Zaromytidou et al., 2006). Underscoring their importance, we found these residues are conserved in Srf orthologs from human to sponge, although they are intriguingly less well-conserved in clades lacking a readily identifiable Mrtf ortholog, namely flatworms (Platyhelminthes) and Placazoa (Figure 2—figure supplement 2). We included an N-terminal 3xFLAG tag and refer to the allele as SrfaI. As a control, we generated a separate SrfFLAG tagged line without the αI helix mutations.

SrfFLAG/FLAG mice were viable and fertile, confirming that neither the FLAG tag nor targeting strategy affected development. In contrast, no SrfaI/aI mice were found at weaning age (Table 1). Because we observed no stillborn or dying neonates, we examined embryos at different stages. SrfaI/aI embryos were recovered in Mendelian ratios until E10.5 but were easily identifiable from E9.5 onward due to their obvious morphological differences from control littermates. Mutant embryos were slightly smaller at E9.5 and most had turning defects that ranged in severity from incompletely turned to totally unturned (Figure 2A). This was accompanied by a wavy neural tube, as seen in many embryos with a deficiency in mesoderm (Figure 2A, middle embryo) and some embryos showed a delay in anterior neural tube closure, indicated by the open midbrain (Figure 2A). Mutant embryos also had a missing or hypoplastic second pharyngeal arch (Figure 2A–B, asterisks). Additionally, the yolk sac showed a crinkled appearance with numerous red blood cells, but no obvious mature blood vessels, indicating the onset of primitive hematopoiesis but not vasculogenesis (Figure 2C).

Figure 2 with 2 supplements see all
SrfaI/aI embryos succumb at E10.5 with numerous defects.

(A) DAPI stained E9.5 embryos of the indicated genotypes show that compared to SrfaI/+ embryos, SrfaI/aI embryos are growth retarded, incompletely turned, have short and disorganized tails, a wavy neural tube (red arrowhead, trunk), delayed anterior neural tube closure (red arrowhead, head), and a hypoplastic or missing second pharyngeal arch (red asterisk). Scale bar represents 500 μm. (B) Higher magnification confocal images of the first two embryos in (A). Scale bar represents 500 μm. (C) Brightfield images of E9.5 yolk sacs indicate defective vasculogenesis in mutant embryos. (D) DAPI-stained E10.5 embryos show more extensive growth retardation, a distended heart tube, and pericardial edema (red arrow). Scale bar represents 500 μm. (E) Immunofluorescent staining of E10.5 yolk sacs shows that mutant yolk sacs lack a remodeled vascular plexus or any large vessels. Images are representative of n = 4 embryos of each genotype. Scale bar represents 250 μm. (F) Transverse sections through E10.5 embryos at the level of the heart show reduced F-actin intensity via phalloidin staining. Images are representative of n = 4 embryos of each genotype. Scale bar represents 50 μm. (G) A similar pattern Is seen in the neural tube. Scale bar represents 50 μm. (H) Cell proliferation, indicated through phospho-Histone H3 (Ser10) (pHH3) staining, is reduced in mutant embryos. Scale bar represents 50 μm. (I) Quantitation of (H), n = 4 each genotype. p = 0.0005, Student’s unpaired two-tailed t-test. (J) Cell death, revealed through cleaved caspase three staining, is dramatically increased in mutant embryos. Scale bar represents 50 μm. (K) Quantitation of (J), n = 3 control embryos and n = 4 mutant embryos. p = 0.0033, Student’s unpaired two-tailed t-test. Columns are the mean and error bars represent the standard deviation in (I) and (K).

Table 1
SrfaI/aI embryos are not recovered at weaning.

Expected and recovered numbers of embryos of each genotype at weaning (P21). No homozygous mutant embryos were recovered.

GenotypeExpectedObserved
Srf+/+7.7512
SrfαI/+15.519
SrfaI/aI7.750
  1. X2 Test = 0.0044.

By E10.5, the anterior neural tube had closed but other defects remained or became apparent. Mutant embryos were much smaller than their wild-type or heterozygous littermates (Figure 2D). The developing heart tube appeared distended and thin, and most embryos showed pericardial edema (Figure 2D, arrows). The overall length of mutant embryos was shorter (Figure 2D) and a subset failed to turn, remaining inflected similar to the rightmost embryo in Figure 2A (data not shown). Whole-mount immunostaining of the yolk sac with the endothelial marker CD31/PECAM1 revealed that while wild-type littermates had an extensively remodeled capillary plexus, including the presence of larger vessels, mutant yolk sacs had only a crude primitive capillary plexus, despite the presence of CD31-positive cells (Figure 2E). A reduction in F-Actin levels throughout mutant embryos, including the heart (Figure 2F) and neural tube (Figure 2G), was consistent with reduced SRF-MRTF-mediated transcription of cytoskeletal genes. Additionally, mutant embryos showed reduced cell proliferation and significantly increased cell death (Figure 2H–K).

While striking, the phenotype of SrfaI/aI embryos is less severe than that reported for Srf null mutants, which succumb from E6.5-E8.5 and do not induce expression of the mesoderm marker T (Arsenian et al., 1998). We generated homozygous Srf-/- embryos and found them to be delayed at E6.5 and E7.5 and were not recovered at later stages (data not shown), verifying the early lethality on our genetic background and thus confirming the difference in severity between the Srf- and SrfaI alleles.

The Srf- and SrfaI alleles cause similar defects in the anterior mesodermal lineage

In order to make a second comparison between the SrfaI and Srf- alleles, we generated Srfflox/flox; Mesp1Cre/+ and SrfaI/flox; Mesp1Cre/+ embryos and assessed them at E9.5 and E10.5. Mesp1-Cre directs recombination in anterior mesoderm, including cardiac mesoderm. This is a tissue where SRF-MRTF interactions are known to be required, particularly through SRF-Myocardin activity in the developing heart and vascular smooth muscle (Li et al., 2003; Miano et al., 2004; Niu et al., 2005; Parlakian et al., 2004). In addition, the wavy neural tube phenotype that we observed in SrfaI/aI is often associated with mesoderm deficiency.

Both Srfflox/flox; Mesp1Cre/+ and SrfaI/flox; Mesp1Cre/+ embryos were inviable after E10.5 and exhibited similar phenotypes. Mutant embryos were small, had turning defects (or arrested prior to or during the turning process), pericardial edema, and hypoplastic hearts. They appeared quite similar to SrfaI/aI embryos and phenocopied Myocd mutants (Figure 3A–B; Li et al., 2003). At E9.5, we observed a wavy neural tube in SrfaI/flox; Mesp1Cre/+ embryos, indicating this phenotype is at least partially attributable to defects in mesoderm. Notably, although SrfaI/flox; Mesp1Cre/+ and Srfflox/flox; Mesp1Cre/+ embryos were broadly similar, Srfflox/flox; Mesp1Cre/+ embryos were more strongly affected, being reproducibly smaller and completely unturned. Although these embryos were generated from separate crosses, precluding direct comparisons, the observations were consistent across multiple litters. We conclude from this analysis that the SrfaI allele is less severe than the Srf- allele, but nevertheless represents a significant curtailment of SRF activity. Moreover, because SrfaI/flox; Mesp1Cre/+ and SrfaI/aI embryos are so similar, the SrfaI/aI phenotype is not a secondary consequence of placental insufficiency, a common cause of cardiovascular and neural phenotypes (Perez-Garcia et al., 2018), as Mesp1Cre/+ labels anterior embryonic and extraembryonic (i.e. yolk sac) mesoderm, but not the trophectoderm-derived placenta (Saga et al., 1999).

The SrfaIand Srfflox alleles exhibits similar defects in anterior mesoderm.

(A) E10.5 littermate embryos were stained with DAPI and imaged. Loss of Srf in the Mesp1-Cre lineage causes embryos to be undersized with pericardial edema, hypoplastic hearts, and turning defects. Phenotype observed in n = 3/3 mutant embryos. (B) A similar experiment in which SrfaI is the only Srf allele expressed in the Mesp1-Cre lineage. These embryos appear comparable to the mutant embryos in (A), although they are clearly less severely affected as they are slightly larger and partially turned. Phenotype observed in n = 3/3 mutant embryos. Scale bar represents 500 μm in all images. Note the 2 x higher crop in mutant embryos to better illustrate phenotypes.

SrfaI/flox; Wnt1-CreTg/+ embryos do not display craniofacial defects at E13.5

We next asked whether SrfaI/flox; Wnt1-CreTg/+ embryos would display similar defects to Srfflox/flox; Wnt1-CreTg/+ embryos, as we expected. Surprisingly, these embryos appeared completely normal at E13.5 (Figure 4A), when Srfflox/flox; Wnt1-CreTg/+ embryos are already dying and display obvious craniofacial abnormalities.

Figure 4 with 1 supplement see all
NC SrfaIconditional mutants are normal at E13.5.

(A) DAPI stained E13.5 littermate embryos carrying a ROSA26TdT/+ Cre reporter show no apparent craniofacial defects in conditional mutants. Scale bar represents 500 μm. (B) Diagram illustrating the culture of MEPM cells. (C) Immunofluorescent staining of passage 2 MEPM cells shows reduced F-actin and SMA fluorescence in mutant cells compared to cells from heterozygous littermate control embryos. Scale bar represents 50 μm. (D) RT-qPCR from serum-starved and serum-stimulated MEPM cells indicates no difference in IEG expression (Egr1, Fos) but a significant defect in Srf, Tagln,, and Acta2 expression and a downward trend in Vcl expression. Values are fold expression of control starved cells. N = 3 control lines and n = 4 mutant lines. (F) RT-qPCR from E11.5 FNPs and mandibles shows a significant reduction in Actg1, Vcl, and Srf expression in Srf NC conditional knockouts compared to control conditional heterozygous littermates. N = 5 each genotype. (G) Assaying the same genes and tissues as (F) using SrfaI NC conditional mutants shows a downward trend or significant reduction approximately half as large as (F). N = 7 controls and n = 6 mutants. For (E–G) significance was determined by Student’s unpaired t-test with two-stage step-up correction (Benjamini, Krieger, and Yekutieli) for multiple comparisons. q-values are indicated on the graphs. Columns are the mean and error bars represent the standard deviation.

We tested whether the SrfaI allele was behaving as expected in the NC lineage using MEPM cells cultured from E13.5 SrfaI/flox; Wnt1-CreTg/+; ROSA26TdT/+ (mutant) and Srfflox/+; Wnt1-CreTg/+; ROSA26TdT/+ (control) palatal shelves (Figure 4B). We assessed the expression of genes preferentially regulated by SRF-TCF activity, such as the IEGs, and those regulated by SRF-MRTF activity, namely cytoskeletal genes. At the protein level, immunofluorescent staining of MEPM cells for F-actin and smooth muscle actin (SMA) showed reduced intensity in mutant cell lines compared to control lines (Figure 4C). We also assessed gene expression at the mRNA level by qPCR in starved and serum-stimulated lines. While we noted no significant changes in the expression of the IEGs Egr1 and Fos (Figure 4D), levels of Tagln (SM22) and Acta2 (the gene encoding SMA) were significantly downregulated in both conditions, and we noted a downward trend in Vcl expression (Figure 4D). Srf itself was also significantly downregulated, likely due to autoregulation via the several CArG elements at the Srf locus (Figure 4D, Figure 2—figure supplement 1A). We confirmed that MRTF-A translocated to the nucleus in response to serum in control and mutant cell lines, indicating that the regulated localization of MRTF-A is not affected but that SRFaI fails to productively interact with MRTF-A despite its nuclear localization, thereby affecting transcription of MRTF targets (Figure 4—figure supplement 1A-B).

Since we saw striking changes in these cells in culture but no phenotype up to this stage in the embryo, we sought to compare gene expression differences in SrfaI/flox; Wnt1-CreTg/+ and Srfflox/flox; Wnt1-CreTg/+ embryos. We assessed two of the most strongly affected DEGs from our RNA-Seq, Actg1 and Vcl, and Srf itself in both models. In E11.5 Srfflox/flox; Wnt1-CreTg/+ facial prominences, we found a strong reduction in expression of all three genes, confirming our RNA-Seq data (Figure 4E). The same genes were downregulated in SrfaI/flox; Wnt1-CreTg/+ prominences, but to a lesser degree than the conditional knockouts (Figure 4F), potentially accounting for the lack of phenotype at this stage.

Because Srf interacts genetically with Pdgfra in the NC and because SRF-MRTF transcriptional targets were suggested to be of particular importance downstream of PDGFRA signaling, we reasoned that the SrfaI allele might also interact genetically with Pdgfra in this tissue (Vasudevan and Soriano, 2014). To test this possibility, we generated SrfaI/+; PdgfraH2B-EGFP/+; Wnt1-CreTg/+ triple heterozygous male mice and crossed them with Srfflox/flox; ROSA26TdT/TdT mice but did not observe facial clefting in embryos of any genotype (data not shown).

In summary, cells from SrfaI/flox; Wnt1-CreTg/+ embryos show the expected changes in gene expression, yet the embryos themselves show no outward signs of the severe craniofacial phenotypes observed in Srfflox/flox; Wnt1-CreTg/+ embryos at this stage and have milder defects in gene expression.

SrfaI/flox; Wnt1-CreTg/+ mice succumb in the early postnatal period with outflow tract defects

SrfaI/flox; Wnt1-CreTg/+ embryos survived until birth, after which they died within the first 2 days of life with visible cyanosis (Figure 5A). We examined E18.5 skeletal preparations for defects in patterning or ossification that may arise after E13.5. Mutant skulls were smaller than in control littermates, but the craniofacial skeleton was patterned normally indicating a developmental delay at this timepoint (Figure 5—figure supplement 1). To determine the underlying cause of cyanosis, we examined the cardiac outflow tract at P0 as the smooth muscle in this region is NC-derived and responsible for proper remodeling of the aortic arch vessels during development. We found a highly penetrant (9/14) patent ductus arteriosus (PDA) exclusively in SrfaI/flox; Wnt1-CreTg/+ neonates (Figure 5B–C). In this condition, the embryonic shunt from the pulmonary artery to the aorta, the ductus arteriosus, fails to close after birth, making circulation to the lungs inefficient and likely explaining the postnatal cyanosis. We also noted one instance of aberrant right subclavian artery, in which the right subclavian artery originates from the descending aorta instead of the brachiocephalic artery, which only supplies the right common carotid artery in this condition. We also inspected P0 mice from a similar cross on a PdgfraH2B-EGFP/+ background to assess whether heterozygosity for Pdgfra would exacerbate phenotypes at this stage, but neonates were recovered in the expected Mendelian ratios with similar outflow tract defects (Figure 5—figure supplement 2A-B). Two SrfaI/flox; Wnt1-CreTg/+; Pdgfra+/+ from this set of crosses displayed a more severe outflow tract defect: right aortic arch with mirror image branching (2/9; Figure 5C, far right).

Figure 5 with 2 supplements see all
NC SrfaI conditional mutants succumb postnatally with outflow tract defects.

(A) Kaplan-Meyer survival curve for neonatal SrfaI/flox; Wnt1-CreTg/+ (Mutant) compared to littermates of all other genotypes (Control). Significance was computed at each timepoint using a Mantel-Cox log-rank test. P0 p = 0.0142, P1 p < 0.0001, P2 p < 0.0001. (B) Stacked columns showing the distribution of PDA-related phenotypes in SrfaI/flox; Wnt1-CreTg/+ neonates compared to littermates of all other genotypes. (C) DAPI-stained postnatal day 0 (P0) hearts carrying a ROSA26TdT/+ Cre lineage reporter showing the entire heart (top row) and the outflow tract region (bottom row). Examples of mutant phenotypes such as PDA, ARSA, and RAA. An asterisk indicates where the missing right subclavian artery should be. Note the ARSA mouse had succumbed prior to dissection and the image is dimmer due to the presence of clotted blood. The outflow tract defects are schematized below. Scale bar represents 500 μm. (D) Hematoxylin and eosin-stained frontal sections through P0 hearts showing mild to severe PDA and an example of VSD and PTA in mutants. Scale bar represents 500 μm. (E) Summary of our results, showing the requirements for SRF versus SRFaI in different tissues and timepoints. A, aorta; ARSA, aberrant right subclavian artery; DA, ductus arteriosus; LCA, left common carotid artery; LSA, left subclavian artery; LV, left ventricle; P, pulmonary artery; PDA, patent ductus arteriosus; PTA, persistent truncus arteriosus; RAA, right aortic arch with mirror image branching; RCA, right common carotid artery; RSA, right subclavian artery.

Sections through SrfaI/flox; Wnt1-CreTg/+ and control Srfflox/+; Wnt1-CreTg/+ hearts confirmed the macroscopically observed PDAs and also revealed one instance of ventricular septal defect (VSD) with persistent truncus arteriosus (PTA), a failure of the truncus arteriosus to fully septate into the aorta and pulmonary artery, out of ten mutant hearts examined (Figure 5D). Mutant embryos that survived the first several hours of life had a milk spot, indicating they did not have pronounced defects in olfaction or the craniofacial bones, nerves, and muscles required for suckling. Thus, while the SrfaI allele is surprisingly well-tolerated in the cranial NC lineage throughout most of development, it is required in the cardiac NC-derived smooth muscle of the cardiac outflow tract to support postnatal life, highlighting a critical role for SRF-MRTF interactions in this particular NC lineage (Figure 5E).

Discussion

SRF is a ubiquitously expressed transcription factor whose transcriptional output is strongly influenced by its cofactors, the TCFs and MRTFs. These cofactors, in turn, are regulated both by specific expression patterns (e.g. Myocd specifically in muscle) and by signaling pathways, such as ERK1/2 and PI3K (Posern and Treisman, 2006; Figure 1A). We sought to better understand the relationship between SRF, its cofactors, and mutant phenotypes in both a general and a tissue-specific manner.

Our previous study demonstrated a requirement for Srf in the NC lineage (Vasudevan and Soriano, 2014), so we first investigated how loss of SRF in this tissue affected gene expression. We found normal early NC patterning but misexpression of cytoskeleton-related genes in Srfflox/flox; Wnt1-CreTg/+ embryos (Figure 1C–F, Figure 1—figure supplements 1 and 2E). Comparing the DEGs with known SRF and cofactor targets showed particular enrichment for MRTF targets in our datasets (Figure 1G–H, Figure 1—figure supplement 2F-G), consistent with SRF’s well-known function regulating cytoskeletal genes and the dominant role of MRTF cofactors in the serum response (Esnault et al., 2014). To test the supposition that SRF-MRTF activity would be the main driver of NC SRF activity, we made a new SrfaI mouse model harboring point mutations that specifically disrupt SRF-MRTF-DNA complex formation (Zaromytidou et al., 2006).

Homozygous SrfaI/aI embryos died at E10.5 with defects in the yolk sac vasculature, heart, and neural tube and exhibited reduced F-actin levels (Figure 2A–F). Both the timing of lethality and gross appearance of the embryos strongly resemble Myocd-/- mutant mice (Li et al., 2003). The phenotype is much less severe than Srf-/- embryos, however, which die around E7.5 and fail to induce mesoderm (Arsenian et al., 1998; Niu et al., 2005). Expression of the SrfaI allele specifically in the Mesp1Cre mesodermal lineage also resulted in lethality at E10.5 and defects almost as severe as complete loss of Srf in this same lineage (Figure 3). The similarity of the two alleles in the Mesp1Cre lineage compared against their strikingly different phenotypes in gastrula-stage embryos suggested there may be time or tissue specific requirements for SRF-MRTF activity.

We went on to test the SrfaI allele in the NC lineage, where we presumed it would be critically important. Similar to the early embryo, expression of the SrfaI allele was well tolerated in NC and we found no facial cleft or bleb in our SrfaI/flox; Wnt1-CreTg/+ mice, as in Srfflox/flox; Wnt1-CreTg/+ mice (Figure 4A). The allele also failed to interact genetically with Pdgfra in this lineage, suggesting PDGF signaling may rely on MRTF-independent SRF activity (Figure 5—figure supplement 2A-B). Nevertheless, this mutation in NC does have a profound developmental effect as SrfaI/flox; Wnt1-CreTg/+ mice completed gestation but died by postnatal day 2 with visible cyanosis (Figure 5A). Examination of the cardiac outflow tract revealed numerous defects including PDA, right aortic arch, and one instance of VSD with PTA (Figure 5C–D). This result is reminiscent of mice with a hypomorphic gene trap mutation in Mrtfb and mice carrying a conditional deletion of Myocd in the NC (Huang et al., 2008; Li et al., 2005b), both of which result in early postnatal lethality due to defects in outflow tract development. Similarly, NC conditional Srf mutants on a genetic background that permitted later development exhibit outflow tract defects at E16.5 (Newbern et al., 2008). These studies and our own results together highlight a critical role for SRF-MRTF interactions in cardiac NC development. A summary of the tissue-specific sensitivities we found to loss of Srf or SrfaI expression is depicted in Figure 5E.

We considered three possible explanations for the tolerance of the SrfaI allele in NC and early embryo, which are not mutually exclusive: (i) that TCFs are more important than previously recognized or are redundant with MRTFs in some respects, (ii) that different tissues may have different thresholds for SRF-MRTF activity and that our point mutant may interact differently with different MRTF family members, and (iii) that SRF may utilize cofactors other than MRTFs and TCFs or have cofactor-independent (i.e. basal) activity sufficient to support development in some contexts.

The first possibility is that TCF factors play a more important role than previously thought or can act redundantly with MRTFs. Because TCF triple mutant embryos survive until E14.5 without obvious morphological defects (Costello et al., 2010), we find it unlikely that TCF-specific activity could explain the tolerance of SrfaI during gastrulation or in NC. However, it has been shown that some SRF targets can be bound and regulated by both MRTF and TCF factors (Esnault et al., 2014), raising the possibility of cofactor redundancy. Yet, most studies to date indicate that rather than acting redundantly, MRTFs and TCFs mediate distinct and opposing phenotypic outcomes, contractility and proliferation, respectively (Gualdrini et al., 2016; Wang et al., 2004). Indeed, a potential result of this competition is that the consequences of suppressing SRF-MRTF interactions could vary according to the level of TCF expression in a given cell type. While we cannot rule out the possibility that TCFs are the primary SRF cofactors in NC or function redundantly to MRTFs without further genetic experiments, this explanation is difficult to reconcile with the existing literature. Although there is no known Srf allele that blocks TCF binding while leaving MRTF binding unperturbed, the V194E mutation (V189E in mouse) disrupts SRF interaction with both cofactors (Ling et al., 1998). Generating a mouse model of this allele would shed considerable light on the question of TCF/MRTF redundancy.

A second explanation is that non-muscle lineages may be able to function with minimal, but not zero, SRF-MRTF activity. It is possible that the SrfaI allele substantially impairs but does not eliminate SRF-MRTF-DNA complex formation and functions as a hypomorph. In vitro gel shift assays for SRF-MRTF-DNA complex formation using purified components and single molecule imaging of SRF in cells both indicate a substantial disruption of SRF-MRTF activity for this allele, but the gel shift experiments that first characterized these mutations detected 5–10% residual complex formation (Hipp et al., 2019; Zaromytidou et al., 2006). Our own data demonstrate that homozygous SrfaI/aI embryos grossly phenocopy Myocd-/- mutants (Figure 2) and SrfaI/flox; Mesp1-CreTg/+ and Srfflox/flox; Mesp1-CreTg/+ embryos were similar, though not identical (Li et al., 2003). On the other hand, double conditional mutants for Mrtfa/Mrtfb largely do phenocopy Srf conditional mutants in several tissues (Cenik et al., 2016; Guo et al., 2018; Li et al., 2005a; Trembley et al., 2015). Therefore, we may be observing differing dosage requirements for SRF-MRTF activity in distinct tissues. Muscle lineages, such as cardiovascular cells affected by Mesp1Cre and NC-derived outflow tract smooth muscle affected by Wnt1-Cre, may need optimal SRF-MRTF output and are therefore strongly affected by the SrfaI allele. Non-muscle lineages such as the cranial NC may survive and develop properly with only residual SRF-MRTF transcription. The SrfaI allele might more strongly suppress Myocardin compared to MRTFA/B, which could also contribute to the tissue-specific requirements we observe. Two predictions of this threshold model are that conditional ablation of Mrtfa/Mrtfb in NC would phenocopy conditional loss of Srf and that Mrtfa/Mrtfb double mutant embryos should succumb around E7.5 as Srf null embryos do. Conversely, non-muscle lineages where conditional ablation of Mrtfa/Mrtfb yields Srf-like phenotypes, such as podocytes and epicardium, should be indifferent to the SrfaI mutations.

The issue of SRF tissue-specific dosage effects may have relevance to human disease. A recent study performed targeted sequencing of SRF in nonsyndromic conotruncal heart defect patients and identified two novel mutations with reduced transcriptional output, one from a patient with VSD and the other with Tetralogy of Fallot with right aortic arch (Mengmeng et al., 2020). Thus, tuning of SRF output may modulate disease in a tissue-specific manner. Along these lines, mutations in MYOCD cause congenital megabladder and associated cardiovascular phenotypes such as PTA and VSD in humans, but monoallelic mutations affect only males whereas biallelic mutations affect both sexes (Houweling et al., 2019). Furthermore, heterozygosity for FLNA, a gene we found strongly affected by loss of Srf in mouse NC, causes the human disease Periventricular Heterotopia I and affected females present with PDA, whereas hemizygous males die during gestation (Fox et al., 1998). Intriguingly, NC-specific conditional knockout of Flna causes perinatal lethality with outflow tract defects in mice (Feng et al., 2006). It would be interesting to further explore the notion of tissue-specific thresholds for SRF-cofactor complexes in future studies.

A third explanation is that SRF functions with additional factors that regulate gene expression independent of or redundantly with MRTFs (and/or TCFs) or has sufficient basal activity to support development in some tissues. Although MRTFs and TCFs are the most well-studied SRF cofactors, many other TFs have been shown to interact with SRF, including but not limited to Homeodomain proteins (Chen et al., 1996; Chen et al., 2002; Grueneberg et al., 1992; Shin et al., 2002), GATA factors (Belaguli et al., 2000; Morin et al., 2001), and Forkhead-family transcription factors (Freddie et al., 2007; Liu et al., 2005), as well as the Initiator-binding protein TFII-I (Grueneberg et al., 1997; Kim et al., 1998). Many of these studies were performed in muscle cells and it is unclear which cofactors might act independently of MRTFs. However, several of these genes or their orthologues are expressed in the cranial NC at E10.5 and E11.5, when cleft formation begins in Srfflox/flox; Wnt1-CreTg/+ embryos (Minoux et al., 2017). One candidate is the homeodomain protein PRRX1/PHOX1/MHOX, which was shown to form complexes with SRF mediated by TFII-I (Grueneberg et al., 1997). Double mutants for Prrx1 and its orthologue Prrx2 have defects of the craniofacial skeleton, aortic arch arteries, and ductus arteriosus (Bergwerff et al., 2000; Lu et al., 1999). Gtf2i mutants (the gene encoding TFII-I) rarely survive past E10.5 but can exhibit a facial cleft, hemorrhaging, and hypoplastic pharyngeal arches (Enkhmandakh et al., 2009). Whether SRF mediates transcription on its own, independently of MRTFs and TCFs, or perhaps using additional tissue-specific cofactors, would be exciting to determine.

In conclusion, we found that the primary transcriptional consequence of losing Srf in NC was a defect in actin cytoskeleton-related gene expression. Using a novel SrfaI allele to perturb SRF’s interactions with MRTFs, the primary cofactors regulating the cytoskeletal transcription program, we uncovered a crucial role for SRF-MRTF activity in the cardiac NC, but surprisingly found the mutation well-tolerated in the cranial NC. Further study will be necessary to determine the relevant SRF-cofactor ensembles in different developmental contexts.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (Mus musculus)SrfMGI, Ensembl, UniProtSrf, ENSMUSG
00000015605, Q9JM73
Strain, strain
background
(Mus musculus)
129S4/SvJaeJIMSR, Jackson Labs009104All the subsequent
genetic reagents
were made on or backcrossed
to this strain
Genetic reagent (Mus musculus)SrfFLAGThis paperTo be submitted to Jackson Labs3x-FLAG tag knocked in
to
SRF’s
N-terminus
Genetic reagent (Mus musculus)SrfaIThis paperTo be submitted to Jackson LabsAlpha-I helix mutations
knocked into
SRF and
identical 3x-
FLAG to
above
Genetic reagent (Mus musculus)Wnt1-CreIMSR, Jackson LabsH2az2Tg(wnt1-cre)11RthRRID:IMSR_JAX:003829
Genetic reagent (Mus musculus)Mesp1CreIMSR, Jackson LabsMesp1tm2(cre)YsaRRID:IMSR_HAR:3358
Genetic reagent (Mus musculus)SrffloxIMSR, Jackson LabsSrftm1RmnRRID:IMSR_JAX:006658
Genetic reagent (Mus musculus)MORE-CreIMSR, Jackson LabsMeox2tm1(cre)SorRRID:IMSR_JAX:003755
Genetic reagent (Mus musculus)R26RTdTIMSR, Jackson LabsGt(ROSA)
26Sortm14(CAG-tdTomato)Hze
RRID:IMSR_JAX:007914
Genetic reagent (Mus musculus)R26RmTmGIMSR, Jackson LabsGt(ROSA)
26Sortm4(ACTB-tdTomato,-EGFP)Luo
RRID:IMSR_JAX:007676
Cell line (Mus musculus)MEPMThis paperPrimary cell line derived in lab.
Used and extinguished by passage 2.
OtherDMEM High GlucoseInvitrogen11965118
OtherPenicillin-Streptomycin (10,000 U/mL)Gibco15140122100 x stock
used at 0.5 x
OtherL-Glutamine
(200 mM)
Gibco2 5030081100 x stock
used at 1 x
OtherCharacterized Fetal Bovine Serum,
CA Origin
HyCloneSH30396.03Lot AC10235406
AntibodyRat anti-CD31
(rat monoclonal)
BD BiosciencesBD Biosciences
Cat# 553370, RRID:AB_394816
IF(1:50)
Antibodyanti-Cleaved Caspase 3 (rabbit monoclonal)Cell SignalingCell Signaling Technology
Cat# 9665, RRID:AB_2069872
IF(1:400)
Antibodyanti phospho-Histone H3 (Ser10) (rabbit polyclonal)MilliporeMillipore
Cat# 06–570, RRID:AB_310177
IF(1:500)
Antibodyanti-MKL1 (rabbit polyclonal)ProteintechProteintech
Cat# 21166–1-AP, RRID:AB_2878822
IF(1:100)
Antibodyanti-SMA (rabbit monoclonal)Cell SignalingCell Signaling Technology
Cat# 19245, RRID:AB_2734735
IF(1:200)
AntibodyGoat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 488Thermo FisherThermo Fisher
Scientific
Cat# A32723, RRID:AB_2633275
IF(1:500)
AntibodyGoat anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 555Thermo FisherThermo Fisher
Scientific
Cat# A32732, RRID:AB_2633281
IF(1:500)
OtherPhalloidin-Alexa Fluor 647InvitrogenCat A22287IF(1:400)
Recombinant DNA reagentAlx3 (plasmid, in situ probe)This paperdoi.10.5061/dryad.mgqnk9916
Recombinant DNA reagentFgf8 (plasmid, in situ probe)Crossley and Martin, 1995pBS-SK-Fgf8doi.10.5061/dryad.mgqnk9916
Recombinant DNA reagentMsx1 (plasmid, in situ probe)Hong and Krauss, 2012doi.10.5061/dryad.mgqnk9916
Recombinant DNA reagentShh (plasmid, in situ probe)This paperdoi.10.5061/dryad.mgqnk9916
Recombinant DNA reagentSix3 (plasmid, in situ probe)This paperdoi.10.5061/dryad.mgqnk9916
Sequence-
based reagent
Srf_flox_FHoltz and Misra, 2008genotyping primersTGCTTACTGG
AAAGCTCATGG
Sequence-based reagentSrf_flox_RHoltz and Misra, 2008genotyping primersTGCTGGTTTG
GCATCAACT
Sequence-based reagentSrf_null_RThis papergenotyping primersCTAACCCTGC
CTGTCCTTCAUse with
Srf_flox_F
Sequence-
based reagent
Srf_flag_FThis papergenotyping primersGATGAACGA
TGTGACCTCGC
Sequence-
based reagent
Srf_flag_RThis papergenotyping primersAGGGAGGA
GCCAACTCCTTA
Sequence-
based reagent
aR4Hamilton et al., 2003genotyping primersCCCTTGTGG
TCATGCCAAACFor PdgfraEGFP
Sequence-
based reagent
aR5Hamilton et al., 2003genotyping primersGCTTTTGCC
TCCATTA
CACTGGFor PdgfraEGFP
Sequence-
based reagent
loxHamilton et al., 2003genotyping primersACGAAGTTAT
TAGGTCC
CTCGACFor PdgfraEGFP
Sequence-
based reagent
Cre_800This papergenotyping primersGCTGCCACGAC
CAAGTGACA
GCAATG
Sequence-
based reagent
Cre_1200This papergenotyping primersGTAGTTATTC
GGATCATCAG
CTACAC
Sequence-
based reagent
moreforTallquist and Soriano, 2000genotyping primersGGGACCACC
TTCTTTTGGCTTC
Sequence-
based reagent
morerevTallquist and Soriano, 2000genotyping primersAAGATGTGGAG
AGTTCGGGGTAG
Sequence-
based reagent
morecreTallquist and Soriano, 2000genotyping primersCCAGATCCTC
CTCAGAA
ATCAGC
Sequence-
based reagent
R26mTmG_FMuzumdar et al., 2007genotyping primersCTCTGCTGC
CTCCTGGCTTCT
Sequence-
based reagent
R26mTmG_wt_RMuzumdar et al., 2007genotyping primersCGAGGCGG
ATCACAA
GCAATA
Sequence-
based reagent
R26mTmG_mut_RMuzumdar et al., 2007genotyping primersTCAATGGGCG
GGGGTCGTT
Sequence-
based reagent
R26Tdt_wt_FMadisen et al., 2010genotyping primersAAGGGAGCT
GCAGTGGAGTA
Sequence-
based reagent
R26Tdt_wt_RMadisen et al., 2010genotyping primersCCGAAAATC
TGTGGGAAGTC
Sequence-
based reagent
R26Tdt_mut_FMadisen et al., 2010genotyping primersGGCATTAAAGC
AGCGTATCC
Sequence-
based reagent
R26Tdt_mut_RMadisen et al., 2010genotyping primersCTGTTCCTGT
ACGGCATGG
Sequence-
based reagent
Acta2_qPCR_FThis paperqPCR primersGGCACCACT
GAACCCTAAGG
Sequence-
based reagent
Acta2_qPCR_RThis paperqPCR primersACAATACCAG
TTGTAC
GTCCAGA
Sequence-
based reagent
Actg1_qPCR_FThis paperqPCR primersATTGTCAATG
ACGAGTGCGG
Sequence-
based reagent
Actg1_qPCR_RThis paperqPCR primersCTTACACTGC
GCTTCTTGCC
Sequence-
based reagent
Egr1_qPCR_FThis paperqPCR primersTGGGATAACTC
GTCTCCACC
Sequence-
based reagent
Egr1_qPCR_RThis paperqPCR primersGAGCGAACAA
CCCTATGAGC
Sequence-
based reagent
Fos_qPCR_FThis paperqPCR primersTCCTACTACCAT
TCCCCAGC
Sequence-
based reagent
Fos_qPCR_RThis paperqPCR primersTGGCACTAGAG
ACGGACAGA
Sequence-
based reagent
Hprt_qPCR_FThis paperqPCR primersTCCTCCTCAG
ACCGCTTTT
Sequence-
based reagent
Hprt_qPCR_RThis paperqPCR primersCATAACCTGG
TTCATCATCGC
Sequence-
based reagent
Srf_qPCR_FThis paperqPCR primersGTGCCACTGG
CTTTGAAGA
Sequence-
based reagent
Srf_qPCR_RThis paperqPCR primersGCAGGTTGGT
GACTGTGAAT
Sequence-
based reagent
Tagln_qPCR_FThis paperqPCR primersGACTGCACTTC
TCGGCTCAT
Sequence-
based reagent
Tagln_qPCR_RThis paperqPCR primersCCGAAGCTAC
TCTCCTTCCA
Sequence-
based reagent
Vcl_qPCR_FThis paperqPCR primersTCTGATCCT
CAGTGG
TCTGAAC
Sequence-
based reagent
Vcl_qPCR_RThis paperqPCR primersAAAGCCATTC
CTGACCTCAC
OtherBM-PurpleRocheCat. #11442074001
Commercial
assay or kit
Luna Universal qPCR Master MixNew England BiolabsCat. #M3003L
Commercial
assay or kit
NEBuilder HiFi DNA Assembly Master MixNew England BiolabsCat. #E2621S
Commercial
assay or kit
RNeasy
Plus Mini Kit
QiagenCat. #74,134
Chemical compound,
drug
Gibco
Geneticin
(G-418)
Thermo FIsherCat. #11811031
Software, algorithmSalmonSalmonSalmon, RRID:SCR_017036
Software, algorithmWasabiWasabihttps://github.com/COMBINE-lab/wasabi;
Patro, 2019
Software, algorithmSleuthSleuthsleuth, RRID:SCR_016883
Software, algorithmGraphPad
Prism
GraphPad
Prism
GraphPad Prism, RRID:SCR_002798
OtherDAPI stainSigmaCat. #D9542-10mg(1–5 µg/mL)

Animal husbandry

Request a detailed protocol

All animal experimentation was conducted according to protocols approved by the Institutional Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai (LA11-00243). Mice were kept in a dedicated animal vivarium with veterinarian support. They were housed on a 13 hr-11hr light-dark cycle and had access to food and water ad libitum.

Mouse models

Request a detailed protocol

The following previously described mouse lines were used: H2az2Tg(wnt1-cre)11Rth referred to as Wnt1-Cre (Danielian et al., 1998), Mesp1tm2(cre)Ysa referred to as Mesp1Cre (Saga et al., 1999), Srftm1Rmn referred to as Srfflox (Miano et al., 2004), Meox2tm1(cre)Sor referred to as MORE-Cre (Tallquist and Soriano, 2000), Gt(ROSA)26Sortm14(CAG-tdTomato)Hze referred to as R26RTdT (Madisen et al., 2010), and Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo referred to as R26RmTmG (Muzumdar et al., 2007). SrfFLAG and SrfaI mice were generated by gene targeting. Homology arms of 2 kb and 6.4 kb were cloned into the pPGKneoF2L2DTA backbone. The longer arm was assembled in three fragments using HiFi assembly cloning (NEB) and included a 3 x FLAG tag introduced with a primer. Fragments were amplified from 129S4 genomic DNA using Q5 polymerase (NEB) except for the middle segment of the long arm (i.e. the coding sequence of exon 1), which was amplified from a cDNA clone before or after introducing the αI helix mutations via site-directed mutagenesis. The targeting constructs were linearized and electroporated in AK7 (129S4 lineage) embryonic stem cells. Clones were selected with G418, screened by long-range PCR, and verified by Southern blot. Correctly targeted clones were injected into C57BL6/J E3.5 blastocysts, transferred to pseudopregnant F1 (C57BL6/J X 129S4) surrogates, and chimeras selected based on coat color. Founders were crossed to MORE-Cre mice to remove the NeoR cassette (Tallquist and Soriano, 2000). All mice were analyzed on a 129S4 co-isogenic background. Genotyping primers are available in Supplementary file 2.

Conservation

Request a detailed protocol

Representative species from the various taxa were subject to BLASTP searches with default parameters using the amino acid sequences for mouse SRF, MKL2, and ELK1. Potential hits were then confirmed by reciprocal BLASTP back to mouse. If a species lacked an ELK1 homolog, the mouse ETS1 sequence was used to search for ETS-domain containing genes. If the given species lacked a hit for a particular search, the search was repeated for the entire taxon. Ctenophora were searched using amino acid sequences for mouse SRF and MEF2C, yeast MCM2, snapdragon Deficiens, and Arabidopsis Agamous. No homolog was identified, suggesting loss of MADS proteins in this lineage.

RNA sequencing

Request a detailed protocol

FNPs (LNP+ MNP) and mandibles were carefully removed from E11.5 embryos in ice-cold PBS using fine forceps. A total of eight embryos across two litters representing four mutants and four controls were collected. Total RNA was immediately extracted using the RNeasy Plus Mini kit (Qiagen). RNA quality was assessed by Tapestation and all samples had RIN scores ≥ 9.8. Samples were sent to GeneWiz for PE150 sequencing. There were 14.6–30 m reads / sample and an average of 24.2 m reads / sample.

Reads were pseudo-aligned to the mouse transcriptome (mm10 partial selective alignment method, downloaded from refgenie) using salmon 1.5.0 and the flags –validateMappings –gcBias –numBootstraps 30. Pseudoalignments were processed with wasabi 1.0.1 and analyzed with sleuth 0.30.0–4 with the flag gene_mode = true. Analysis was performed using a full model that accounted for genotype, litter/batch, and tissue-of-origin (for combined tissue analysis only) versus a reduced model consisting only of litter/batch (and tissue-of-origin). Fold-changes and q-values were computed using the Wald test. Volcano plots were made with VolcaNoseR. Heat maps were generated using the Shinyapp HeatMappr. Gene set enrichment analysis for GO terms, ENCODE datasets, etc. were done with the web utility Enrichr (Xie et al., 2021). Enrichment for a custom list of targets was performed using GSEA software 4.10 and normalized read counts for the entire dataset. Analysis for the joint tissue model was run in phenotype mode ( > 7 samples per condition) and for the individual tissue samples in gene_set mode ( < 7 samples per condition) according to the software developer. The maximum number of genes per set was raised to 800 to accommodate the target lists. All other parameters were default.

MEPM culture

Request a detailed protocol

Mouse embryonic palatal mesenchyme cells were generated as described (Fantauzzo and Soriano, 2017). Briefly, palatal shelves were dissected from E13.5 embryos in ice cold PBS using fine forceps. Yolk sac tissue was used for genotyping. Palates from individual embryos were held on ice until dissection was complete and palates were then dissociated using 0.125% Trypsin-EDTA at 37°C for 10 min with occasional trituration using a P1000 pipet. Trypsin was neutralized with an equal volume of growth media (DMEM High Glucose supplemented with Glutamine, Penicillin-Streptomycin, and 10% Fetal Calf III serum) and plated onto culture dishes coated in 0.1% gelatin. Cells were passaged as they approached confluency, every 2–3 days, and used for experiments at passage 2. MEPM cell lines were not screened for mycoplasma as they were used and extinguished by passage 2, but there was no evidence of mycoplasma by DAPI staining and other immortalized cell lines used in the lab tested negative for mycoplasma, so the risk of contamination was negligible.

Immunofluorescence

Request a detailed protocol

MEPM cells were seeded on #1.5 coverslips coated with 0.1% gelatin. For starvation experiments, cells were starved overnight in 0.1% serum then stimulated 30’ with 10% serum. Cells were fixed using 4% PFA in PBS for 10’ at 37 °C. Embryos were dissected in ice cold PBS, fixed one hour in 4% PFA in PBS at 4 °C, rinsed in PBS, cryoprotected in 30% sucrose, and embedded in OCT. Sections were cut at 10 µm thickness using a Leica cryostat. Yolk sacs were fixed one hour in 4% PFA in PBS at 4 °C and stained whole. All samples were rinsed in PBS, blocked and permeabilized in blocking media (PBS, 0.3% TritonX-100, 1% BSA, 5% calf serum) one hour at RT, primary antibody was diluted in fresh blocking media and samples treated overnight at 4 °C, washed 3 x PBS at RT, incubated in Alexa Fluor Plus-conjugated secondary antibodies (Invitrogen) diluted 1:500 in fresh blocking media with 1 μg/ml DAPI for 1–2 hr at RT or overnight at 4 °C, and finally washed 3 x in PBS at RT. Samples were mounted in Prolong Diamond (Invitrogen) mounting media and imaged on a Zeiss AxioObserver inverted fluorescence microscope or a Zeiss 780 upright confocal microscope. Thresholding was performed and scalebars added in the FIJI implementation of ImageJ. All images for a given experiment were processed identically with the exception of Figure 2A where the mutant embryos were brightened compared to the control embryo to better illustrate phenotypes.

Quantitation of cell proliferation and cell death was performed by staining frozen sections with the indicated antibodies. Sections at the level of the heart were imaged using a 10 x objective on a Zeiss 780 confocal microscope, a 1024 × 1024 pixel count and 6 μm step size. Tiling was used with 10% overlap when necessary to image the entire section. Maximum intensity projections were made in the FIJI implementation of ImageJ and identical thresholds used for each embryo to calculate the DAPI-positive and cleaved Caspase 3-positive or phospho-Histone H3 (Ser 10)-positive area on sections at the level of the heart. Any of the embryo’s posterior present in the section was ignored as this region was not present in all sections for all embryos.

Antibodies used were, rat anti-CD31 (BD Pharmingen, 553370) 1:50, rabbit anti-cleaved Caspase 3 (Cell Signaling 9665, 1:400), rabbit anti-phospho Histone H3 (Ser10) (Millipore 06–570, 1:500), rabbit anti-MKL1 (Proteintech, 21166–1-AP) 1:100, rabbit anti-SMA (Cell Signaling, 19245) 1:200. Phalloidin-Alexa Fluor 647 (Invitrogen) was included where indicated during secondary antibody staining at 1:400.

RT-qPCR

Request a detailed protocol

Cells were seeded, passaged, starved, and stimulated as for immunofluorescence except in 12-well tissue culture plates. Embryo facial prominences were dissected in cold PBS and transferred to 1.5 ml microfuge tubes on ice. Following the indicated stimulation regimes when applicable, cells/tissue were lysed in 300 μl RLT buffer supplemented with BME, and RNA isolated using the RNeasy Plus Kit (Qiagen) according to the manufacturer’s instructions. RNA concentration was quantified using a Nanodrop. One μg total RNA was used for reverse transcription. RNA was primed using a 2:1 molar ratio of random hexamer and polydT (Invitrogen) and reverse transcribed with Superscript IV (Invitrogen) according to the manufacture’s protocol. Resulting cDNA was diluted 5 x with water and stored at –20 °C. One μl cDNA was used per qPCR reaction. qPCR was performed using Luna 2 x Master Mix (NEB) on an iQ5 thermocycler (Bio-Rad) in triplicate. Differences in gene expression were calculated by ΔΔCT using Hprt for normalization. Primer sequences are listed in Supplementary file 3.

Nuclear translocation analysis

Request a detailed protocol

Cells were starved, treated, and stained as described above and imaged on an inverted Zeiss AxioObserver microscope. Z-stacks were maximum intensity projected in the FIJI implementation of ImageJ, background subtracted, and the DAPI channel used to create a nuclear mask. This mask was then used to measure the average nuclear intensity in the MRTF-A channel for each nucleus. The data presented are the pooled results from two cell lines of each genotype where each dot is an individual nucleus. At least 70 cells were analyzed per condition.

In situ hybridization

Request a detailed protocol

E10.5 embryos were dissected in ice-cold PBS and fixed overnight in 4% FA in PBS at 4 °C, rinsed in PBS, dehydrated through a MeOH series and stored in 100% MeOH at –20 °C. Embryos were stained using standard techniques for the indicated transcripts using published, DIG-labeled probes, and were developed in BM Purple (Roche | Sigma-Aldrich). For Fgf8 the proteinase-K digestion was omitted in order to maintain integrity of the ectoderm. A detailed protocol is included as Supplementary file 2 and probe sequences are available as SnapGene files at Dryad. https://doi.org/105061/dryadmgqnk9916.

Histology

Request a detailed protocol

P0 hearts were fixed overnight in 4% FA in PBS at 4 °C, rinsed in PBS, dehydrated through an ethanol series, and embedded in paraffin. Five μm sections were cut using a Leica microtome. After drying, sections were stained with Harris modified hematoxylin (Fisher) and Eosin Y using a standard regressive staining protocol.

Skeletal preparations

Request a detailed protocol

Skeletons were stained by standard techniques. Briefly, E18.5 embryos were skinned, eviscerated, fixed in ethanol, stained with.015% alcian blue and.005% alizarin red overnight at 37 ° C, cleared in 1% KOH, processed through a glycerol:KOH series, and photographed in 80% glycerol in PBS.

Statistical methods

Request a detailed protocol

Specific statistical methods, significance values, and n are detailed in the figure legends. For RNA-Seq, statistics were computed using the built-in Wald Test function in the Sleuth analysis package. All other statistics were performed using GraphPad Prism 9.

Data availability

The NGS data is available on GEO. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE186770.

The following data sets were generated
    1. Dinsmore CJ
    2. Soriano P
    (2021) NCBI Gene Expression Omnibus
    ID GSE186770. Differential regulation of cranial and cardiac neural crest by Serum Response Factor.

References

  1. Conference
    1. Axel Visel CA
    (2015) FaceBase Consortium
    RNA-Seq analysis in E13.5 mouse palates.
    https://doi.org/10.25550/TN2
    1. Crossley PH
    2. Martin GR
    (1995)
    The mouse Fgf8 gene encodes a family of polypeptides and is expressed in regions that direct outgrowth and patterning in the developing embryo
    Development (Cambridge, England) 121:439–451.
    1. Mohun TJ
    2. Chambers AE
    3. Towers N
    4. Taylor MV
    (1991)
    Expression of genes encoding the transcription factor SRF during early development of Xenopus laevis: identification of a CArG box-binding activity as SRF
    The EMBO Journal 10:933–940.
    1. Saga Y
    2. Miyagawa-Tomita S
    3. Takagi A
    4. Kitajima S
    5. Miyazaki J i
    6. Inoue T
    (1999)
    MesP1 is expressed in the heart precursor cells and required for the formation of a single heart tube
    Development 126:3437–3447.

Decision letter

  1. Marianne E Bronner
    Senior and Reviewing Editor; California Institute of Technology, United States
  2. Richard Treisman
    Reviewer; Francis Crick Institute, United Kingdom

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Differential regulation of cranial and cardiac neural crest by Serum Response Factor" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Marianne Bronner as the Senior and Reviewing Editor. The following individual involved in review of your submission has agreed to reveal their identity: Richard Treisman (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

All three reviewers expressed great enthusiasm for this study and feel that it will make a valuable contribution to the literature. That said, they also include extensive suggestions for improvements and clarifications to the text and figures. While these suggestions are numerous, all of the proposed changes are editorial and do not require further experimentation. I refer you to the comments of the reviewers below for details on suggested modifications.

Reviewer #1 (Recommendations for the authors):

1) Page 4: it would be nice to show a supplementary a table with the summary of the phenotypes of the co-factor knock outs.

2) Figure 1B: Please label upper and lower panels with mutant and sibs. The difference in labeling (flox/+ and flox/flox is difficult to see).

3) Please provide an excel file with the genes that are associated with the different Enrichment terms.

4) Page 9: 'Underscoring their importance, we found these residues are conserved in Srf homologs from human to sponge, though they are intriguingly less well-conserved in clades lacking a readily identifiable Mrtf ortholog'.

Please add: '.., such as… (list the clades)'.

5) Page 11: In Srfα mutants the authors detected an increase in b-Catenin levels. Please discuss this finding, as there are studies published regarding the interaction between Srf and Wnt signaling.

6) Page 13, line 285. Please add a sentence describing the interpretation of the findings in this paragraph.

7) Line 398: typo: but not does…

8) In the same paragraph it is unclear if starved is compared to serum, or mutant to sibling. Please clarify.

9) How do the authors explain that the gene expression changes are the same between the null and the point mutation mutant but the phenotypes are different? Are very subtle gene expression changes enough to cause the phenotypic differences?

Reviewer #2 (Recommendations for the authors):

I strongly support publication of this paper, with some modifications. Specific points to address:

Figure 1. This revisits the 2014 GD data for the Wnt1-cre SRF knockout. Its real value for this study would be to act as a control for analysis of the Wnt1-cre SRF-alpha1 mutant, but this is not done. This is important, because it may give insights into potential gene expression differences between the null and SRF-alpha1. The authors should do this experiment.

Figure 4. While serum stimulation is of interest, a more pertinent analysis, given the fact that there is no genetic interaction bewteen SRF-alpha1 and PDGFR, would be to look at PDGF stimulation, either in this system or MEPMs. What happens? Are there differences between Wnt1-cre SRF KO and SRF-alpha1?

Reviewer #3 (Recommendations for the authors):

1. The authors suggest in the discussion that TCF factors may compensate for loss of SRF-MRTF activity. Could the authors provide some in vivo staining to show the expression patterns of genes involved in these two complexes? Although bulk-RNAseq data shows that these two complexes are both involved in samples collected from the craniofacial region, it's hard to conclude whether these two complexes could compensate for each other without information about their in vivo expression patterns.

2. Defects of the neural tube are not visible in Figures 2A and 2C. Could the authors adjust the orientation of the samples for better visualization and clarification of the phenotype?

3. Since MRTF-A translocated to the nucleus in response to serum in control and mutant cell lines, the authors need to provide explanation as to why there are no differences observed in control groups with/without serum?

4. Bulk RNA-seq analysis using tissue from Srfflox/flox;Wnt1-CreTg/+ mice strongly suggests that the SRF-MRTF complex plays a critical role in craniofacial development. Therefore, the authors generated Srf αl mice, in which they mutated the αI helix of the SRF DNA-binding domain to disrupt SRF-MRTF complex formation while leaving SRF-TCF-DNA complex formation unaffected. However, in the neural crest conditional mutant (SrfαI/flox;Wnt1-CreTg/+) mice, the defects were detected not in the craniofacial tissues but in cardiac development. Could the authors provide more explanation for these results? Are there downstream genes targeted by both SRF-MRTF and SRF-TCF in the craniofacial tissues so that the SRF-TCF complex can compensate for the lost function of the SRF-MRTF complex?

5. Why did the authors study the function of SRF-MRTF in the mesoderm-derived lineage? Is this an attempt to show the efficiency of the SrfαI/al model? Do Srfflox/flox;Wnt1-CreTg/+ mice have outflow tract defects?

https://doi.org/10.7554/eLife.75106.sa1

Author response

Reviewer #1 (Recommendations for the authors):

1) Page 4: it would be nice to show a supplementary a table with the summary of the phenotypes of the co-factor knock outs.

We thank the reviewer for this suggestion and now include these phenotypes as Supplementary Table 1. The other tables have been renumbered accordingly.

2) Figure 1B: Please label upper and lower panels with mutant and sibs. The difference in labeling (flox/+ and flox/flox is difficult to see).

We have clarified the labeling in this panel as well as panel 1E.

3) Please provide an excel file with the genes that are associated with the different Enrichment terms.

These data are now included as tabs in the Supplementary File. We regret our earlier omission.

4) Page 9: 'Underscoring their importance, we found these residues are conserved in Srf homologs from human to sponge, though they are intriguingly less well-conserved in clades lacking a readily identifiable Mrtf ortholog'.

Please add: '.., such as… (list the clades)'.

We now highlight the pertinent clades in the manuscript and thank the reviewer for the suggestion.

5) Page 11: In Srfα mutants the authors detected an increase in b-Catenin levels. Please discuss this finding, as there are studies published regarding the interaction between Srf and Wnt signaling.

We also found this quite interesting and worthy of future study but were unsure how broadly to speculate on its implications within the manuscript. One could imagine the change in beta-catenin levels to be caused through direct transcriptional regulation, through altered cytoskeletal behavior (speculatively, by increased recruitment and/or stabilization at junctions in an attempt to compensate for reduced F-actin levels or through altered cadherin levels), or by changes in beta-catenin protein’s stability, most obviously through a change in canonical Wnt activity. Increased canonical Wnt signaling should lead to beta-catenin stabilization or alternately, increased beta-catenin stabilization might ectopically activate or increase sensitivity to Wnt. We note that the enhanced beta-catenin expression remained enriched at cell junctions and did not build up within the nucleus. Further, we did not detect a strong Wnt signature in our RNA-seq studies. However, we realize nuclear beta-catenin levels can be difficult to detect (Apc mutants can show stabilized junctional, not nuclear, beta-catenin, e.g. Buchert et al., 2010) and a Wnt phenotype may be specific to the SrfαI allele or the early embryo and not neural crest. However, the existence, strength, and mechanism of this effect in the context of our Srf mutant remains unclear. Several reports have suggested a link between Srf and the Wnt pathway including the suppression of SRF-mediated IEG expression by Wnt, Wnt regulation of Mrtfa expression, SRF-mediated regulation of Wnt pathway targets, and phosphorylation of SRF by GSK3 promoting SRF-MRTF activity. Most intriguingly, one study found an effect on Wnt through SRF-mediated expression of a microRNA that regulates EMT and E-Cadherin, indirectly affecting beta-catenin localization. However, after considering these numerous potential explanations, we feel this result might be more distracting than helpful in the present manuscript. We hope to include this result in a future manuscript where the connections between SRF, adherens junctions, beta-catenin, and potentially the Wnt pathway can be explored – and tested – in greater detail. Therefore, we have opted to remove those two panels from the present manuscript.

6) Page 13, line 285. Please add a sentence describing the interpretation of the findings in this paragraph.

We added a clause explaining that we expected MRTF-A’s localization to be unaffected by our mutations in Srf. The upstream actin dynamics and signaling pathways that regulate MRTFs should remain intact in our point mutant background. There was a formal possibility that mutating Srf could cause an indirect change in the pathways that control MRTF localization. If this were true, we might interpret transcriptional changes as being due to the failure of SRFαI to interact with MRTF, when in fact it may be due to the unexpected failure of MRTF to translocate to the nucleus at all. We agree that in the original manuscript, ending the paragraph with a rather bluntly (un)explained control experiment was confusing.

7) Line 398: typo: but not does…

Fixed. We thank the reviewer for their careful reading.

8) In the same paragraph it is unclear if starved is compared to serum, or mutant to sibling. Please clarify.

We were unsure which experiment this comment was referring to, but if it is in reference to the residual 5-10% complex formation discussed later in the same paragraph, that experiment was comparing wild type to mutant protein in an in vitro complex formation (gel shift) assay. We added further description of the experiment to clarify this matter.

9) How do the authors explain that the gene expression changes are the same between the null and the point mutation mutant but the phenotypes are different? Are very subtle gene expression changes enough to cause the phenotypic differences?

We thank the reviewer for highlighting this confusing point. We found strong phenotypic and gene expression changes in cultured MEPM cells derived from the SrfαI conditional mutants (Figure 4). Srfflox/flox conditional (Wnt1-Cre) mutants do not survive long enough to culture similar MEPM cells. The gene expression experiments in Figure 1 were from E11.5 craniofacial tissue. To conduct a comparison of these two genetic models using an identical tissue source and assay, we performed RT-qPCR from E11.5 SrfαI/flox and Srfflox/flox Wnt1-Cre conditional mutants and littermate controls. We find that the cytoskeletal genes we assayed were downregulated in both models, but more strongly in the conditional knockouts than in the conditional point mutants. This likely explains the different phenotypes: there is one threshold for failure in the craniofacial crest (and possibly also the gastrula-stage embryo) and another for failure in the cardiac crest (which forms smooth muscle) and the Mesp1 mesodermal lineage. However, whether this expression is due to residual MRTF activity or transcription of the same targets supported by other cofactors remains to be determined. We have now added these important data as figure 4E-F and commented where relevant, although these data ultimately do not substantially change the possible mechanistic interpretations covered in the Discussion.

Reviewer #2 (Recommendations for the authors):

I strongly support publication of this paper, with some modifications. Specific points to address:

Figure 1. This revisits the 2014 GD data for the Wnt1-cre SRF knockout. Its real value for this study would be to act as a control for analysis of the Wnt1-cre SRF-alpha1 mutant, but this is not done. This is important, because it may give insights into potential gene expression differences between the null and SRF-alpha1. The authors should do this experiment.

This is an excellent suggestion. The Srflox/lox; Wnt1-Cre mice do not survive long enough on our genetic background to generate MEPMs and compare directly with the cells in Figure 4. We were also concerned that repeating the RNA-Seq analysis on the conditional point mutants would take an excessive amount of time. However, we agree that a side-by-side assessment of the two models in NC at the level of gene expression, rather than phenotype, was warranted and therefore performed RT-qPCR on several of the most differentially regulated genes from our RNA seq on freshly isolated E11.5 craniofacial prominences from both conditional point mutant and conditional knockout embryos (Vasudevan and Soriano, 2014). We find a reduction in gene expression in both models, but the effect is roughly 2x more profound in the conditional knockouts. We think this is consistent with the observed phenotype. The residual gene expression must be sufficient in craniofacial NC (and perhaps gastrula) but not in Mesp1 or cardiac NC lineages. The qPCR data are now included as Figure 4E-F and the Discussion has been updated.

Figure 4. While serum stimulation is of interest, a more pertinent analysis, given the fact that there is no genetic interaction bewteen SRF-alpha1 and PDGFR, would be to look at PDGF stimulation, either in this system or MEPMs. What happens? Are there differences between Wnt1-cre SRF KO and SRF-alpha1?

This is an interesting suggestion, but a difficult experiment to do. The Srf NC conditional mutants do not survive long enough on our genetic background to generate MEPMs. Some experiments were performed in our previous study using cultured Srf conditional knockout cells derived from an earlier timepoint, but these were only immunofluorescent experiments on a small number passage 0 cells, as these cells did not take well to culture. The experiment could best be done by crossing the relevant alleles into an immortalized background such as Ink4a or generating Srf null and point mutant cell lines in an already immortalized but otherwise wild type cell line via CRISPR. However, generating these materials would take a longer amount of time than can be expected for the manuscript revision.Reviewer #3 (Recommendations for the authors):

1. The authors suggest in the discussion that TCF factors may compensate for loss of SRF-MRTF activity. Could the authors provide some in vivo staining to show the expression patterns of genes involved in these two complexes? Although bulk-RNAseq data shows that these two complexes are both involved in samples collected from the craniofacial region, it's hard to conclude whether these two complexes could compensate for each other without information about their in vivo expression patterns.

We thank the reviewer for raising this point. Some of these data were presented in a previous publication (Vasudevan and Soriano, 2014). We also now include previously published RNA-Seq data on sorted NC cells, whole palate, and passage 2 MEPM cells showing co-expression of Srf and its cofactors in these cells and tissues in Figure 1 Supplement 2E, as well as our own bulk RNA seq data for the same genes, showing a similar pattern. We attempted to perform immunofluorescent staining for cofactors in frozen sections but did not obtain convincing staining, possibly due to the lower relative expression levels of ELK1 and MRTFA, which have the most well-validated antibodies.

2. Defects of the neural tube are not visible in Figures 2A and 2C. Could the authors adjust the orientation of the samples for better visualization and clarification of the phenotype?

We agree the images were rather dim and small in the initial version of the figure. Therefore, we have first readjusted the brightness of the mutant embryos in figure 2A. They were initially imaged and thresholded identically to the control, but as they are smaller with fewer cells, this resulted in a dimmer image. We readjusted them (and added a note to the Methods section) and added arrowheads pointing out the phenotypes. We think the open neural tube/exencephaly and wavy neural tube are now much easier to see. Second, we added a new set of panels where two of the E9.5 embryos were reimaged at a higher magnification using a confocal microscope. The pharyngeal arch defect and open midbrain are now much clearer. We thank the reviewer for the suggestion.

3. Since MRTF-A translocated to the nucleus in response to serum in control and mutant cell lines, the authors need to provide explanation as to why there are no differences observed in control groups with/without serum?

See response to reviewer 1, point 6. Briefly, MRTF translocation should be unaffected, but once in the nucleus it should have reduced or absent formation of a productive SRF-MRTF-DNA ternary complex with our SRF point mutant compared to wild type SRF. The experiment was a control that nothing upstream was unexpectedly affected. We added clarification in the manuscript on this point.

4. Bulk RNA-seq analysis using tissue from Srfflox/flox;Wnt1-CreTg/+ mice strongly suggests that the SRF-MRTF complex plays a critical role in craniofacial development. Therefore, the authors generated Srf αl mice, in which they mutated the αI helix of the SRF DNA-binding domain to disrupt SRF-MRTF complex formation while leaving SRF-TCF-DNA complex formation unaffected. However, in the neural crest conditional mutant (SrfαI/flox;Wnt1-CreTg/+) mice, the defects were detected not in the craniofacial tissues but in cardiac development. Could the authors provide more explanation for these results? Are there downstream genes targeted by both SRF-MRTF and SRF-TCF in the craniofacial tissues so that the SRF-TCF complex can compensate for the lost function of the SRF-MRTF complex?

We performed RT-qPCR testing of SrfαI/flox;Wnt1-CreTg/+ and Srfflox/flox;Wnt1-CreTg/+ to get a better comparison of gene expression differences in NC among the two alleles and find downregulation of actin cytoskeletal genes in both models, but to a lesser degree in the point mutant compared to the conditional knockout. Whether this is due to TCF compensation (it has been suggested that TCF and MRTF coregulate some targets), residual MRTF function, and/or TCF/MRTF-independent SRF function is unknown at this time, but we discuss these possibilities within the Discussion. We have reorganized the Discussion section to incorporate ideas raised by reviewer 2 and we hope this makes these points more clearly. Reviewer 2 suggested several excellent experiments to test these ideas including generation of an SrfV194E allele that should disrupt both TCF and MRTF interactions. Another good test would be conditional ablation of Mrtfa/b in the NC. We are excited to pursue these possibilities going forward but agree with reviewer 2 that these experiments are beyond the scope of this manuscript. Finally, to address the reviewer’s question concerning craniofacial vs cardiac phenotypes, it seems likely given the strong effect in the Mesp1 lineage and SRF’s known role in muscle that the SrfαI allele has a particularly strong effect in the cardiac crest because these cells mostly differentiate into smooth muscle, whereas craniofacial NC cells become other cell types such as bone, cartilage, neurons, glia, melanocytes, and dermal mesenchyme. Therefore, a threshold effect is likely at work where cells requiring high levels of actin-related gene expression (i.e. muscle) are more affected by SrfαI than those that require more modest expression levels. Further testing of this idea with new genetic models will be key.

5. Why did the authors study the function of SRF-MRTF in the mesoderm-derived lineage? Is this an attempt to show the efficiency of the SrfαI/al model? Do Srfflox/flox;Wnt1-CreTg/+ mice have outflow tract defects?

The reviewer is correct. Because vascular and cardiac mesoderm are known to rely on SRF-Myocardin activity, we wanted to test the efficiency of the SrfαI allele compared to the Srf- allele (and published mutants such as Myocd-/-) in this context. Further, we wondered if the wavy neural tube was (at least partially) non-autonomous and due to mesoderm defects. We did see this phenotype in SrfαI/flox; Mesp1Cre/+ embryos at E9.5, supporting this interpretation. We failed to mention this observation in the initial manuscript and have now added this rationale and finding to the Results section.

Srfflox/flox; Wnt1-Cre mice do have cardiac outflow tract defects, as reported by Newbern et al. We mention this finding within the Discussion.

https://doi.org/10.7554/eLife.75106.sa2

Article and author information

Author details

  1. Colin J Dinsmore

    Department of Cell, Development and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6404-1202
  2. Philippe Soriano

    Department of Cell, Development and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Project administration, Supervision, Writing - review and editing
    For correspondence
    philippe.soriano@mssm.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0427-926X

Funding

National Institute of Dental and Craniofacial Research (R01 DE022363)

  • Philippe Soriano

National Institute of Dental and Craniofacial Research (F32 DE026678)

  • Colin J Dinsmore

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the members of the Soriano lab, Robert Krauss, Sergei Sokol, and Harish Vasudevan for critical reading of the manuscript. We are appreciative of Kevin Kelley and the Mouse Transgenic Core for stable tissue culture facilities. We thank the Mt. Sinai Microscopy core facility for consistent advice and top-notch instrumentation. CJD was supported by F32 DE026678 from National Institutes of Health (NIH)/National Institute of Dental and Craniofacial Research (NIDCR). This work was supported by R01 DE022363 from NIH/NIDCR to PS.

Ethics

All animal experimentation was conducted according to protocols approved by the Institutional Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai under protocol LA11-00243. Mice were kept in a dedicated animal vivarium with veterinarian support. They were housed on a 13hr-11hr light-dark cycle and had access to food and water ad libitum.

Senior and Reviewing Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Reviewer

  1. Richard Treisman, Francis Crick Institute, United Kingdom

Publication history

  1. Preprint posted: October 28, 2021 (view preprint)
  2. Received: October 28, 2021
  3. Accepted: January 18, 2022
  4. Accepted Manuscript published: January 19, 2022 (version 1)
  5. Version of Record published: February 1, 2022 (version 2)

Copyright

© 2022, Dinsmore and Soriano

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 845
    Page views
  • 94
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Colin J Dinsmore
  2. Philippe Soriano
(2022)
Differential regulation of cranial and cardiac neural crest by serum response factor and its cofactors
eLife 11:e75106.
https://doi.org/10.7554/eLife.75106

Further reading

    1. Cancer Biology
    2. Developmental Biology
    Maja Solman et al.
    Research Article Updated

    Gain-of-function mutations in the protein-tyrosine phosphatase SHP2 are the most frequently occurring mutations in sporadic juvenile myelomonocytic leukemia (JMML) and JMML-like myeloproliferative neoplasm (MPN) associated with Noonan syndrome (NS). Hematopoietic stem and progenitor cells (HSPCs) are the disease propagating cells of JMML. Here, we explored transcriptomes of HSPCs with SHP2 mutations derived from JMML patients and a novel NS zebrafish model. In addition to major NS traits, CRISPR/Cas9 knock-in Shp2D61G mutant zebrafish recapitulated a JMML-like MPN phenotype, including myeloid lineage hyperproliferation, ex vivo growth of myeloid colonies, and in vivo transplantability of HSPCs. Single-cell mRNA sequencing of HSPCs from Shp2D61G zebrafish embryos and bulk sequencing of HSPCs from JMML patients revealed an overlapping inflammatory gene expression pattern. Strikingly, an anti-inflammatory agent rescued JMML-like MPN in Shp2D61G zebrafish embryos. Our results indicate that a common inflammatory response was triggered in the HSPCs from sporadic JMML patients and syndromic NS zebrafish, which potentiated MPN and may represent a future target for JMML therapies.

    1. Developmental Biology
    Yulong Liu et al.
    Research Article

    Zebrafish are an established research organism that has made many contributions to our understanding of vertebrate tissue and organ development, yet there are still significant gaps in our understanding of the genes that regulate gonad development, sex, and reproduction. Unlike the development of many organs, such as the brain and heart that form during the first few days of development, zebrafish gonads do not begin to form until the larval stage (≥5 dpf). Thus, forward genetic screens have identified very few genes required for gonad development. In addition, bulk RNA sequencing studies which identify genes expressed in the gonads do not have the resolution necessary to define minor cell populations that may play significant roles in development and function of these organs. To overcome these limitations, we have used single-cell RNA sequencing to determine the transcriptomes of cells isolated from juvenile zebrafish ovaries. This resulted in the profiles of 10,658 germ cells and 14,431 somatic cells. Our germ cell data represents all developmental stages from germline stem cells to early meiotic oocytes. Our somatic cell data represents all known somatic cell types, including follicle cells, theca cells and ovarian stromal cells. Further analysis revealed an unexpected number of cell subpopulations within these broadly defined cell types. To further define their functional significance, we determined the location of these cell subpopulations within the ovary. Finally, we used gene knockout experiments to determine the roles of foxl2l and wnt9b for oocyte development and sex determination and/or differentiation, respectively. Our results reveal novel insights into zebrafish ovarian development and function and the transcriptome profiles will provide a valuable resource for future studies.