Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Protein interaction is critical molecular regulatory activity underlining cellular functions and precise cell fate choices. Using TWIST1 BioID-proximity-labeling and network propagation analyses, we discovered and characterized a TWIST-chromatin regulatory module (TWIST1-CRM) in the neural crest cells (NCC). Combinatorial perturbation of core members of TWIST1-CRM: TWIST1, CHD7, CHD8, and WHSC1 in cell models and mouse embryos revealed that loss of the function of the regulatory module resulted in abnormal differentiation of NCCs and compromised craniofacial tissue patterning. Following NCC delamination, low level of TWIST1-CRM activity is instrumental to stabilize the early NCC signatures and migratory potential by repressing the neural stem cell programs. High level of TWIST1 module activity at later phases commits the cells to the ectomesenchyme. Our study further revealed the functional interdependency of TWIST1 and potential neurocristopathy factors in NCC development. eLife digest Shaping the head and face during development relies on a complex ballet of molecular signals that orchestrates the movement and specialization of various groups of cells. In animals with a backbone for example, neural crest cells (NCCs for short) can march long distances from the developing spine to become some of the tissues that form the skull and cartilage but also the pigment cells and nervous system. NCCs mature into specific cell types thanks to a complex array of factors which trigger a precise sequence of binary fate decisions at the right time and place. Amongst these factors, the protein TWIST1 can set up a cascade of genetic events that control how NCCs will ultimately form tissues in the head. To do so, the TWIST1 protein interacts with many other molecular actors, many of which are still unknown. To find some of these partners, Fan et al. studied TWIST1 in the NCCs of mice and cells grown in the lab. The experiments showed that TWIST1 interacted with CHD7, CHD8 and WHSC1, three proteins that help to switch genes on and off, and which contribute to NCCs moving across the head during development. Further work by Fan et al. then revealed that together, these molecular actors are critical for NCCs to form cells that will form facial bones and cartilage, as opposed to becoming neurons. This result helps to show that there is a trade-off between NCCs forming the face or being part of the nervous system. One in three babies born with a birth defect shows anomalies of the head and face: understanding the exact mechanisms by which NCCs contribute to these structures may help to better predict risks for parents, or to develop new approaches for treatment. Introduction The cranial neural crest cell (NCC) lineage originates from the neuroepithelium (Vokes et al., 2007; Groves and LaBonne, 2014; Mandalos and Remboutsika, 2017) and contributes to the craniofacial tissues in vertebrates (Sauka-Spengler and Bronner-Fraser, 2008) including parts of the craniofacial skeleton, connective tissues, melanocytes, neurons, and glia (Kang and Svoboda, 2005; Blentic et al., 2008; Ishii et al., 2012; Theveneau and Mayor, 2012). The development of these tissues is affected in neurocristopathies, which can be traced to mutations in genetic determinants of NCC specification and differentiation (Etchevers et al., 2019). As an example, mutations in transcription factor TWIST1 in human are associated with craniosynostosis (El Ghouzzi et al., 2000) and cerebral vasculature defects (Tischfield et al., 2017). Phenotypic analyses of Twist1 conditional knockout mouse revealed that TWIST1 is required in the NCCs for the formation of the facial skeleton, the anterior skull vault, and the patterning of the cranial nerves (Soo et al., 2002; Ota et al., 2004; Bildsoe et al., 2009; Bildsoe et al., 2016). To comprehend the mechanistic complexity of NCC development and its implication in a range of diseases, it is essential to collate the compendium of genetic determinants of the NCC lineage and characterize how they act in concert in time and space. During neuroectoderm development, transcriptional programs are initiated successively in response to morphogen induction to specify neural stem cell (NSC) subdomains along the dorsal-ventral axis in the neuroepithelium (Briscoe et al., 2000; Vokes et al., 2007; Kutejova et al., 2016). NCCs also arise from the neuroepithelium, at the border with the surface ectoderm through the pre-epithelial-mesenchymal transition (pre-EMT) which is marked by the activation of Tfap2a, Id1, Id2, Zic1, Msx1 and Msx2 (Baker et al., 1997; Mayor et al., 1997; Saint-Jeannet et al., 1997; Marchant et al., 1998; Etchevers et al., 2019). In the migratory NCCs, gene activity associated with pre-EMT and NCC specification is replaced by that of EMT and NCC identity (Marchant et al., 1998). NCC differentiation progresses in a series of cell fate decisions (Lasrado et al., 2017; Soldatov et al., 2019). Genetic activities for mutually exclusive cell fates are co-activated in the progenitor population, which is followed by an enhancement of the transcriptional activities that predilect one lineage over the others (Lasrado et al., 2017; Soldatov et al., 2019). However, more in-depth knowledge of the specific factors triggering this sequence of events and cell fate bias is presently lacking. Twist1 expression is initiated during NCC delamination and its activity is sustained in migratory NCCs to promote ectomesenchymal fate (Soldatov et al., 2019). TWIST1 mediates cell fate choices through functional interactions with other basic-helix-loop-helix (bHLH) factors (Spicer et al., 1996; Firulli et al., 2005; Connerney et al., 2006) in addition to transcription factors SOX9, SOX10, and RUNX2 (Spicer et al., 1996; Hamamori et al., 1997; Bialek et al., 2004; Laursen et al., 2007; Gu et al., 2012; Vincentz et al., 2013). TWIST1 therefore constitutes a unique assembly point to identify the molecular modules necessary for cranial NCC development and determine how they orchestrate the sequence of events in this process. To decipher the molecular context of TWIST1 activity and identify functional modules, we generated the first TWIST1 protein interactome in the NCCs. Leveraging the proximity-dependent biotin identification (BioID) methodology, we captured TWIST1 interactions in the native cellular environment including previously intractable transient and low-frequency events which feature interactions between transcription regulators (Roux et al., 2012; Kim and Roux, 2016). Integrating prior knowledge of protein associations and applying network propagation analysis (Cowen et al., 2017), we uncovered modules of highly connected interactors as potent NCC regulators. Among the top-ranked candidates were histone modifiers and chromatin remodelers that constitute the functional chromatin regulatory module (TWIST1-CRM) in NCC. Genome occupancy, gene expression, and combinatorial perturbation studies of high-ranked members of the TWIST1-CRM during neurogenic differentiation in vitro and in embryos revealed their necessity in stabilizing the identity of early migratory NCC and subsequent acquisition of ectomesenchyme potential. This study also highlighted the concurrent activation and cross-repression of the molecular machinery that governs the choice of cell fates between neural crest and neurogenic cell lineages. Results Deciphering the TWIST1 protein interactome in cranial NCCs using BioID The protein interactome of TWIST1 was characterized using the BioID technique which allows for the identification of interactors in their native cellular environment (Figure 1A). We performed the experiment in cranial NCC cell line O9-1 (Ishii et al., 2012) transfected with TWIST1-BirA* (TWIST1 fused to the BirA* biotin ligase). In the transfected cells, biotinylated proteins were predominantly localized in the nucleus (Figure 1—figure supplement 1A,B; Singh and Gramolini, 2009). The profile of TWIST1-BirA* biotinylated proteins were different from that of biotinylated proteins captured by GFP-BirA* (Figure 1—figure supplement 2A). Western blot analysis detected TCF4, a known dimerization partner of TWIST1, among the TWIST1-BirA* biotinylated proteins but not in the control group (Figure 1—figure supplement 2A). These findings demonstrated the utility and specificity of the BioID technology to identify TWIST1-interacting proteins. Figure 1 with 2 supplements see all Download asset Open asset TWIST1 interactome in cranial NCCs revealed using BioID and network propagation. (A) BioID procedure to identify TWIST1-interacting partners in neural crest stem cells (NCCs). TWIST1-BirA* (TWIST1 fused to the BirA* biotin ligase) labeled the proteins partners within the 10 nm proximity in live cells. Following cell lysis and sonication, streptavidin beads were used to capture denatured biotin-labeled proteins, which were purified and processed for mass spectrometry analysis. (B) TWIST1-specific interaction candidates identified by BioID mass-spectrometry analysis in NCC cell line (p<0.05; Fold-change >3; PSM#>2) overlap with all reported TWIST1 interactions on the Agile Protein Interactomes DataServer (APID) (Alonso-López et al., 2019). (C) Networks constructed from stringent TWIST1-specific interaction at a significant threshold of adjusted p-value (adjp) <0.05 and Fold-change >3. Unconnected nodes were removed. Top GO terms for proteins from three different clusters are shown. Node size = -Log10 (adjp). Genes associated with human and mouse facial malformation (HP:0001999, MP:0000428) were used as seeds (dark red) for heat diffusion through network neighbors. Node color represents the heat diffusion score. (D) Expression of candidate interactor genes in cranial neural crest from E9.5 mouse embryos; data were derived from published transcriptome dataset (Fan et al., 2016). Each bar represents mean expression ± SE of three biological replicates. All genes shown are expressed at level above the microarray detection threshold (27, red dashed line). We characterized all the proteins biotinylated by TWIST1-BirA* and GFP-BirA* followed by streptavidin purification using liquid chromatography combined with tandem mass spectrometry (LC-MS/MS) (Supplementary file 1). Differential binding analysis of TWIST1 using sum-normalized peptide-spectrum match (PSM) values (Figure 1—figure supplement 2B,C; see Materials and methods) revealed 140 putative TWIST1 interactors in NCCs (p<0.05; Fold-change >3; PSM#>2; Figure 1B, Supplementary file 1). These candidates included 4 of 56 known TWIST1 interactors, including TCF3, TCF4, TCF12, and GLI3 (overlap odds ratio = 18.05, Chi-squared test p-value=0.0005; Agile Protein Interactomes DataServer [APID]) (Alonso-López et al., 2019; Fan et al., 2020). Despite that the APID database covers a broad spectrum of protein interaction in different cell line models, it was noted that the TWIST1 partners, TCF3, 4, and 12 that were recurrently identified in yeast-two-hybrid, immunoprecipitation and in vitro interaction assays were recovered by BioID (El Ghouzzi et al., 2000; Firulli et al., 2007; Fu et al., 2011; Teachenor et al., 2012; Sharma et al., 2013; Kotlyar et al., 2015; Li et al., 2015). This finding has prompted us to explore the rest of the novel partners identified in the BioID analysis. Network propagation prioritized functional modules and core candidates in TWIST1 interactome We invoked network propagation analytics to identify functional modules amongst novel TWIST1 BioID-interactors and to prioritize the key NCC regulators (see Materials and methods). Network propagation, which is built on the concept of ‘guilt-by-association’, is a set of analytics used for gene function prediction and module discovery (Sharan et al., 2007; Ideker and Sharan, 2008; Cowen et al., 2017). By propagating molecular and phenotypic information through connected neighbors, this approach identified and prioritized relevant functional clusters while eliminating irrelevant ones. The TWIST1 functional interaction network was constructed by integrating the association probability matrix of the BioID candidates based on co-expression, protein-interaction, and text mining databases from STING (Singh and Gramolini, 2009; Szklarczyk et al., 2015). Markov clustering (MCL) was applied to the matrix for the inference of functional clusters (Figure 1—figure supplement 2D, Supplementary file 2). Additionally, data from a survey of the interaction of 56 transcription factors and 70 unrelated control proteins were used to distinguish the likely specific interactors from the non-specific and the promiscuous TF interactors (Li et al., 2015). Specific TF interactors (red) and potential new interactors (blue; Figure 1—figure supplement 2D–i) clustered separately from the hubs predominated by non-specific interactors (gray; Figure 1—figure supplement 2D–ii). The stringency of the screen was enhanced by increasing the statistical threshold (adjusted p-value [adjp]<0.05) and excluding the clusters formed by non-specific interactors such as those containing heat-shock proteins and cytoskeleton components. Gene Ontology analysis revealed major biological activities of proteins in the clusters: chromatin organization, cranial skeletal development, and ribosome biogenesis (Figure 1C; Supplementary file 2; Chen et al., 2009). Heat diffusion was applied to prioritize key regulators of NCC development. The stringent TWIST1 interaction network comprises proteins associated with facial malformation phenotypes in human/mouse (HP:0001999, MP:0000428), which points to a likely role in NCC development. These factors were used as seeds for a heat diffusion simulation to find near-neighbors of the phenotype hot-spots (i.e. additional factors that may share the phenotype) and to determine their hierarchical ranking (Figure 1C, Supplementary file 2). As expected, that disease causal factors are highly connected and tend to interact with each other (Jonsson and Bates, 2006), a peak of proteins with high degrees of connectivity emerged among the top diffusion ranked causal factors, most of which are from the chromatin organization module (Figure 1—figure supplement 2F). TWIST1 and these interacting chromatin regulators were referred to hereafter as the TWIST1-chromatin regulatory module (TWIST1-CRM). Among the top 30 diffusion ranked BioID candidates, we prioritized nine for further characterization. These included chromatin regulators that interact with TWIST1 exclusively in NCCs versus 3T3 fibroblasts: the chromodomain helicases CHD7, CHD8, the histone methyltransferase WHSC1 and SMARCE1, a member of the SWI/SNF chromatin remodeling complex (Figure 1C, candidates name in red; Figure 1—figure supplement 2E,F; Supplementary file 3). We also covered other types of proteins, including transcription factors PRRX1, PRRX2, TFE3 and the cytoplasmic phosphoprotein DVL1 (Dishevelled 1). The genes encoding these proteins were found to be co-expressed with Twist1 in the cranial NCCs of in embryonic day (E) 9.5 mouse embryos (Figure 1D, Supplementary file 1; Bildsoe et al., 2016; Fan et al., 2016). The chromatin regulators interact with the N-terminus domain of TWIST1 Co-immunoprecipitation (co-IP) assays showed that CHD7, CHD8, PRRX1, PRRX2, and DVL1 could interact with TWIST1 like known interactors TCF3 and TCF4, while TFE3 and SMARCE1 did not show any detectable interaction (Figure 2A). Fluorescent immunostaining demonstrated that these proteins co-localized with TWIST1 in the nucleus (Figure 1—figure supplement 1C). The exceptions were DVL1 and TFE3, which were localized predominantly in the cytoplasm (Figure 1—figure supplement 1C). Among these candidates, CHD7 and CHD8 are known to engage in direct domain-specific protein-protein interactions (Batsukh et al., 2010). Three sub-regions of CHD7 and CHD8 were tested for interaction with TWIST1 (Figure 2B). For both proteins, the p1 region, which encompasses helicases and chromodomains, showed no detectable interaction with partial or full-length TWIST1. In contrast, the p2 and the p3 regions of CHD7 and CHD8 interacted with full-length TWIST1 as well as with its N-terminal region (Figure 2C). Reciprocally, the interaction was tested with different regions of TWIST1 including the bHLH domain, the WR domain, the C-terminal region and the N-terminal region (Figure 2B). CHD7, CHD8, and WHSC1 interacted preferentially with the TWIST1 N-terminus whereas the TCF dimerization partners interacted specifically with the bHLH domain (Figure 2D). Consistent with the co-IP result, SMARCE1 and TFE3 did not interact with TWIST1. Interestingly, the other known factor that binds the TWIST1 N-terminal region is the histone acetyltransferase CBP/P300 which is also involved in chromatin remodeling (Hamamori et al., 1999). These findings demonstrated the direct interaction of TWIST1 with a range of epigenetic factors and transcriptional regulators and identified the TWIST1 N-terminal region as the domain of contact. Figure 2 Download asset Open asset The chromatin regulators interact with the N-terminus domain of TWIST1. (A) Detection of HA-tagged proteins after immunoprecipitation (IP) of TWIST1 (IP: α-TWIST1) from lysates of O9-1 cells transfected with constructs expressing TWIST1 (input blot: α-TWIST1) and the HA-tagged proteins partners (input blot: α-HA). (B) Schematics of CHD7, CHD8, and TWIST1 proteins showing the known domains (gray blocks) and the regions (double arrows) tested in the experiments shown in panels C and D. (C, D) Western blot analysis of HA-tagged proteins (α-HA antibody) after GST-pulldown with different TWIST1 domains (illustrated in B). Protein expression in the input is displayed separately. T, full-length TWIST1; N, N-terminal region; C, C-terminal region; bHLH, basic helix-loop-helix domain; TA, transactivation domain. Genetic interaction of Twist1 and chromatin regulators in craniofacial morphogenesis The function of the core components of the TWIST1-CRM was investigated in vivo using mouse embryos derived from ESCs that carried single-gene or compound heterozygous mutations of Twist1 and the chromatin regulators. Mutant ESCs for Twist1 and the three validated NCC-exclusive chromatin regulatory partners Chd7, Chd8, and Whsc1 were generated by CRISPR-Cas9 editing (Figure 3—figure supplement 1A,B; Ran et al., 2013). ESCs of specific genotype (non-fluorescent) were injected into 8 cell host wild-type embryos (expressing fluorescent DsRed.t3) and chimeras were collected at E9.5 or E11.5 (Figure 3A; Sibbritt et al., 2019). Figure 3 with 1 supplement see all Download asset Open asset Genetic interaction of Twist1 and chromatin regulators in craniofacial morphogenesis. (A) Experimental strategy for generating chimeric mice from WT and mutant ESCs (see Materials and methods). (B) Lateral and dorsal view of mid-gestation chimeric embryos with predominant ESC contribution (embryos showing low or undetectable red fluorescence). Genotype of ESC used for injection is indicated. Scale bar: 1 mm. Heterozygous embryos of single genes (Twist1+/-, Chd8+/-, Whsc1+/-) showed mild defects including hemorrhages and mild neural tube defect (white arrowheads). Compound heterozygous embryos displayed open neural tube and head malformation (orange arrowheads, n ≥ 6 for each genotype, see panel 3C), in addition to heart defects. (C) Proportions of normal and malformed embryos (Y-axis) for each genotype (X-axis). Severity of mutant phenotypes was determined based on the incidence of developmental defects in the neuroepithelium, midline tissues, heart and vasculature: Normal (no defect); Mild (1–2 defects); Severe (3–4 defects), and early lethality. The number of embryos scored for each genotype is in parentheses. (D) Whole-mount immunofluorescence of E11.5 chimeras derived from wildtype ESCs, shows the expression of TFAP2A (red) and neurofilament (NF, green) and cell nuclei by DAPI (blue). Schematic on the right shows the neuroepithelium structures: f, forebrain; m, midbrain; h, hindbrain; tv, telencephalic vesicle; fn, frontonasal region. (E) (i) NCC cells, marked by TFAP2A, and neuroepithelial cells, marked by SOX2, are shown in (ii) sagittal, and (iii) transverse view of the craniofacial region (red line in ii: plane of section). (F) Quantification of frontal nasal TFAP2A+ tissues (mean normalized area ± SE) of three different sections of embryos of each genotype. (G) SOX2 intensity (mean ± SE) in the ventricular zone of three sections of embryos of each genotype were quantified using IMARIS. (H) (i) Cranial nerves visualized by immunostaining of neurofilament (NF). (ii–v) Maximum projection of cranial nerves in embryos. Missing or hypoplastic neurites are indicated by arrowheads. (ii'–v') Cross-section of neurofilament bundles in the trigeminal ganglion. Red dashed line in i: plane of section. Bar: 500 μm; V, trigeminal ganglion; III, IV, VII, VIII; rio, infraorbital nerve of V2; rmd, mandibular nerve; ropht, ophthalmic profundal nerve of V1; rfr, frontal nerve. (I) Thickness of neural bundle in the trigeminal ganglion was measured by the GFP-positive area, normalized against area of the trigeminal ganglion (TFAP2A+). Values plotted represent mean fold change ± SE. Each condition was compared to WT. p-Values were computed by one-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ns, not significant. Only embryos with predominant contribution of mutant ESCs, indicated by the absence or low level of DsRed.t3 fluorescence, were analyzed. The majority of embryos derived from single-gene heterozygous ESCs (Twist1+/-, Chd8+/-, and Whsc1+/-) displayed mild deficiency in the cranial neuroepithelium and focal vascular hemorrhage (Figure 3B,C). Compound heterozygous embryos (Twist1+/-;Chd7+/-, Twist1+/-;Chd8+/-, and Twist1+/-;Whsc1+/-) displayed more severe craniofacial abnormalities and exencephaly (Figure 3B,C). Given that CHD8 was not previously known to involve in craniofacial development of the mouse embryo, we focused on elucidating the impact of genetic interaction of Chd8 and Twist1 on NCC development in vivo. While Chd8+/- embryos showed incomplete neural tube closure, compound Twist1+/-;Chd8+/- embryos displayed expanded neuroepithelium, a phenotype not observed in the single-gene mutants (Figure 3B,E). The population of NCCs expressing Tfap2α, a Twist1-independent NCC marker (Brewer et al., 2004) was reduced in the frontonasal tissue and the trigeminal ganglion (Figure 3E–i,F). In contrast, SOX2 expression was upregulated in the ventricular zone of the neuroepithelium of mutant chimeras (Figure 3E–ii,iii,G). Furthermore, Twist1+/-, Chd8+/- and Twist1+/-;Chd8+/- embryos displayed different degrees of hypoplasia of the NCC-derived cranial nerves (Figure 3H). Cranial nerves III and IV were absent, and nerve bundle in the trigeminal ganglia showed reduced thickness, most evidently in the Twist1+/-;Chd8+/- compound mutant embryos (Figure 3H,I). Altogether, these results suggested that TWIST1 genetically interaction with epigenetic regulators CHD7, CHD8, and WHSC1 to guide the formation of the cranial NCC and downstream tissue genesis in vivo. Genomic regions co-bound by TWIST1 and chromatin regulators are enriched for early migratory NCC signatures in the open chromatin region The loss of NCC progenitors and neural tube malformation indicate that the combined activity of TWIST1-chromatin regulators might be required from early in NCC development. To understand the molecular function of TWIST1-chromatin regulators in early NCC differentiation, we performed an integrative ChIP-seq analysis. The ChIP-seq dataset for TWIST1 was generated from the ESC-derived neuroepithelial cells (NECs) which are the source of early NCCs (Figure 4—figure supplement 1 and Materials and methods). We retrieved published NEC ChIP-seq datasets for CHD7 and CHD8 and the histone modifications and reanalyzed the data following the ENCODE pipeline (ENCODE Project Consortium, 2012; Sugathan et al., 2014; Ziller et al., 2015; Figure 4—figure supplement 1A). Two H3K36me3 ChIP-seq datasets for NECs were included in the analysis (Du et al., 2017; Chai et al., 2018) on the basis that WHSC1 trimethyl transferase targets several H3 lysine (Morishita et al., 2014) and catalyzes H3K36me3 modification in vivo (Nimura et al., 2009). Genome-wide co-occupancies of TWIST1, CHD7, and CHD8 showed significant overlap (Fisher's exact test) and clustered by Jaccard Similarity matrix (Figure 4A). ChIP-seq peaks were correlated with active histone modifications H3K27ac and H3K4me3 but not the inactive H3K27me3, or the WHSC1-associated H3K36me3 modifications (Figure 4A). TWIST1, CHD7, and CHD8 shared a significant number of putative target genes (Figure 4B). TWIST1 shared 63% of target genes with CHD8 (odds ratio = 16.93, Chi-squared test p-value<2.2e-16) and 18% with CHD7 (odds ratio = 8.26, p-value<2.2e-16; Figure 4B; Supplementary file 4). Compared with genomic regions occupied by only one factor, greater percentage of regions with peaks for two or all three factors (TWIST1, CHD7, and CHD8) showed H3K27ac and H3K4me3 signal (Figure 4C). This trend was not observed for the H3K27me3 modification. Similarly, the co-occupied transcription start sites (TSS) showed open chromatin signatures with enrichment of H3K4me3 and H3K27Ac and depletion of H3K27me3 (Ernst et al., 2011; Rada-Iglesias et al., 2011; Figure 4D,E). We also did not observe H3K36me3 modifications near the overlapping TSSs, suggesting that WHSC1 may have alternative histone lysine specificity in the NECs. Figure 4 with 3 supplements see all Download asset Open asset Genomic region showing overlapping binding of TWIST1 and partners are enriched for active regulatory signatures and neural tube patterning genes. (A) Top panel: Trajectory of ESC differentiation to neuroepithelial cells (NECs) and NCCs. Bottom panel: Jaccard Similarity matrix generated of ChIP-seq data of TWIST1, CHD7, CHD8, and histone modifications from NE cells. The Jaccard correlation is represented by a color scale. White squares indicate no significant correlation (p<0.05, fisher’s exact test) or odds ratio <10 between the two datasets. (B) Venn diagram showing overlaps of putative direct targets of TWIST1, CHD7, and CHD8, based on ChIP-seq datasets for NECs (Sugathan et al., 2014). (C) Percent genomic region that is marked by H3K4me3, H3K27ac, and H3K27me3 among regions bound by one, two or all three factors among TWIST1, CHD7, and CHD8. Randomized peak regions of similar length (1 kb) were generated for hg38 as a control. (D) Heatmaps of genomic footprint of protein partners at ± 5 kb from the TSS, based on the ChIP-seq datasets (as in A) and compared with histone marks H3K4me3, H3K27ac, and H3K27me3 in human neural progenitor cells (Ziller et al., 2015). TSS lanes with no overlapping signals were omitted. (E) ChIP-seq density profile (rpkm normalized) for all TSS flanking regions shown in D. (F) Gene Ontology analysis of genomic regulatory regions by annotations of the nearby genes. Regions were grouped by presence of binding site of individual factor (TWIST1, CHD7, and CHD8), or by 2–3 factors in combination. The top non-redundant developmental processes or pathways for combinatorial binding peaks or individual factor binding peaks are shown. p-Value cut-off: 0.05. (G) Enrichment of TWIST-chromatin regulator targets among regulons of different NCC single-cell clusters at E8.5-E10.5 (Soldatov et al., 2019). Number of overlapping genes with DNA binding peaks (TSS ± 1 kb) for each TF combination are represented by dot size and -log(p-value) is represented by color gradient. Gene modules with significant enrichment (p<0.05) are labeled with asterisk. A random set of genes from the scRNA-seq, with number comparable to the largest TF binding group (1000 genes) were used as control. (H) IGV track (Robinson et al., 2011) showing ChIP-peak overlap (red arrows) at common transcriptional target genes in cell mobility (Sox9, Pdgfra, Snai1, Lamb1, Dlx2) in NCC development and neurogenesis (Jag1, Foxb1, Sox2, En1, Zic3, Dll3) repressed at early migration. Gene diagrams are indicated (bottom row). The top Gene Ontology enriched for the co-occupied regulatory regions of two or three core components included neural tube patterning, cell migration and BMP signaling pathway (Figure 4F). Overlapping peaks of the partners were localized within ± 1 kb of the TSS of common target genes (Figure 4H; Supplementary file 4). This integrative analysis revealed that the TWIST1-chromatin regulators shared genomic targets that are harbored in open chromatin in the NECs. To pinpoint more specifically when TWIST1-chromatin regulators are required and better interpret their transcriptional activities in light of the in vivo dynamics and timing of target gene activity, we examined relevant gene activities in the E8.5- E10.5 mouse NCCs scRNA-seq datasets of NCCs traced by Wnt1-Cre and Sox10-Cre reporters (Soldatov et al., 2019). The first clue came from the expression of Twist1, Chd7, Chd8, and Whsc1 in NCCs clusters that are ordered in developmental pseudotime: neural tube, delaminatory, early migratory, migratory 1, migratory 2, sensory, autonomic and mesenchyme (Figure 4—figure supplement 2A–C). Twist1 displayed stage-specific dynamics and initially peaked in the early migratory NCC followed by exponential activation while progressing to the mesenchyme. On the other hand, the three interacting partners expressed rather ubiquitously throughout all NCC populations (Figure 4—figure supplement
Abstract This study evaluated the association between body pH value and preoperative deep vein thrombosis (DVT) in geriatric hip fractures. Older adult patients with hip fractures were screened between January 2015 and September 2019. The demographic and clinical characteristics of the patients were collected. Multivariate binary logistic regression and generalized additive models were used to identify the linear and nonlinear associations between pH value and preoperative DVT. Analyses were performed using EmpowerStats and R software. A total of 1465 patients were included in the study. DVT occurred in 476 (32.6%) of these admitted older adults. We observed a nonlinear association between the serum pH value and preoperative DVT in geriatric patients with hip fractures. A pH value of 7.39 was the inflection point in the curve, with pH highly correlated with DVT at pH < 7.39 (odds ratio [OR] 19.47; 95% confidence interval [CI] 1.45–260.91; P = 0.0249). Patients with lower pH had a lower chance of preoperative DVT formation, and the risk of DVT increased 18.47-fold for every 0.1 unit change in pH. Although at pH > 7.39, pH was not correlated with DVT (OR 1.26; 95% CI 0.85–1.86; P = 0.2561), the odds of DVT did not vary with pH, and the highest risk of thrombosis was reached. The body pH value is nonlinearly associated with preoperative DVT in geriatric patients with hip fractures, and it could be considered a predictor of the risk of DVT. Registered information This study is registered in the website of Chinese Clinical Trial Registry (ChiCTR: ChiCTR2200057323).
As a preliminary exploration, based on the historical order data of the battery-swap stations, this paper predicts the order quantity of the stations in the future period, which can be used as a reference for the battery scheduling of battery-swap stations. Compared with the existing studies, we establish the network of battery-swap stations based on the real large-scale order data, and build the graph model combined with multihead attention to predict. The challenges in this paper include similarity analysis, building the adjacency matrix of battery-swap stations, and the model for distributed stations fine-grained order volume prediction. To address the above challenges, we build the GAT model to accurately predict the order quantity of batteryswap stations. Our work consists of two parts. First, we construct three features and calculate the similarity among the features based on the spatio-temporal correlation data of battery-swap stations. The similarity will be used to construct the association matrix of stations and fuse it with the distance adjacency matrix to construct the topological network of battery-swap stations. Second, we build a graph neural network model and input the network structure of battery-swap stations and historical order data for prediction. In order to capture the correlation features of data, we introduce the multi-attention mechanism, which uses attention to capture the features of different graph nodes, and finally achieves better prediction effect through multi-connection. Our model performs well in hourly order prediction, with an average MAE of about 1.07. We believe that this work can provide reference and ideas for enterprises in order balance, battery scheduling and other aspects.
We describe PROPER-seq (protein-protein interaction sequencing) to map protein-protein interactions (PPIs) en masse. PROPER-seq first converts transcriptomes of input cells into RNA-barcoded protein libraries, in which all interacting protein pairs are captured through nucleotide barcode ligation, recorded as chimeric DNA sequences, and decoded at once by sequencing and mapping. We applied PROPER-seq to human embryonic kidney cells, T lymphocytes, and endothelial cells and identified 210,518 human PPIs (collected in the PROPER v.1.0 database). Among these, 1,365 and 2,480 PPIs are supported by published co-immunoprecipitation (coIP) and affinity purification-mass spectrometry (AP-MS) data, 17,638 PPIs are predicted by the prePPI algorithm without previous experimental validation, and 100 PPIs overlap human synthetic lethal gene pairs. In addition, four previously uncharacterized interaction partners with poly(ADP-ribose) polymerase 1 (PARP1) (a critical protein in DNA repair) known as XPO1, MATR3, IPO5, and LEO1 are validated in vivo. PROPER-seq presents a time-effective technology to map PPIs at the transcriptome scale, and PROPER v.1.0 provides a rich resource for studying PPIs.
Abstract Chromatins are pervasively attached by RNAs. Here, we asked whether global RNA-chromatin contacts are altered in a given cell type in a disease context, and whether these alterations impact gene expression and cell function. In endothelial cells (ECs) treated by high-glucose and TNFα, we employed single-cell RNA-sequencing and in situ mapping of RNA-genome interaction (iMARGI) assay to delineate temporal changes in transcriptome and RNA-chromatin interactome. ECs displayed dramatic and heterogeneous changes in single cell transcriptome, accompanied by a dynamic and strong increase in inter-chromosomal RNA-DNA interactions, particularly among super enhancers (SEs). These SEs overlap with genes contributing to inflammatory response and endothelial mesenchymal transition (EndoMT), two key aspects of endothelial dysfunction. Perturbation of a high-glucose and TNFα-activated interaction involving SEs overlapping LINC00607 and SERPINE1 attenuated the pro-inflammatory and pro-EndoMT gene induction and EC dysfunction. Our findings highlight RNA-chromatin contacts as a crucial regulatory feature in biological and disease processes, exemplified by endothelial dysfunction, a major mediator of numerous diseases.
PTL-1 is the sole homolog of the MAP2/MAP4/tau family in Caenorhabditis elegans. Accumulation of tau is a pathological hallmark of neurodegenerative diseases such as Alzheimer's disease. Therefore, reducing tau levels has been suggested as a therapeutic strategy. We previously showed that PTL-1 maintains age-related structural integrity in neurons, implying that excessive reduction in the levels of a tau-like protein is detrimental. Here, we demonstrate that the regulation of neuronal ageing by PTL-1 occurs via a cell-autonomous mechanism. We re-expressed PTL-1 in a null mutant background using a pan-neuronal promoter to show that PTL-1 functions in neurons to maintain structural integrity. We next expressed PTL-1 only in touch neurons and showed rescue of the neuronal ageing phenotype of ptl-1 mutant animals in these neurons but not in another neuronal subset, the ventral nerve cord GABAergic neurons. Knockdown of PTL-1 in touch neurons also resulted in premature neuronal ageing in these neurons but not in GABAergic neurons. Additionally, expression of PTL-1 in touch neurons alone was unable to rescue the shortened lifespan observed in ptl-1 mutants, but pan-neuronal re-expression restored wild-type longevity, indicating that, at least for a specific group of mechanosensory neurons, premature neuronal ageing and organismal ageing can be decoupled.
Objective: In this paper, we tested and studied the dynamic plantar pressure distribution of normal subjects in Xi'an and its surrounding region, and extract feature date of subjects of different ages. In order to provide the basis for the next assessment of diseased feet and the study of individual-adapted foot orthopedic appliances, we execute preliminary analysis the effects of age and body mass index on the distribution of plantar pressure based on our study.; Method: We collected data of 784 walking plantar pressures from 392 volunteers using RsScan footscan plantar pressure assessment system, calculated indicators of pressure, impulse and center-of-pressure(COP) trajectories, grouped according to ages, compared the difference and carried out correlation analysis and summarized the law of the change by means of statistic.
Result: According to our study, the plantar peak pressure was weakly positive correlation with age, weight, and BMI index, the variation of the curve of the foot pressure center in the direction of the x axis(DetaCOPx) was weakly positive correlation with and the height, weight and BMI index, and the average impulse was weakly positive correlation with the age, weight and BMI index. The changes of plantar dynamic parameters DetaCOPx, and the change speed of the foot pressure center curve in the direction of the X axis (VCOPx), and the coordinates of the foot pressure center curve in the direction of the x axis (COPx) on the left and right feet were basically the same, with no significant difference. There was no significant difference in the peak pressure, DetaCOPx and average impulse among all age groups.
To investigate the dynamic expression and role of vitamin D receptor (VDR) in the myocardium of mice with viral myocarditis (VMC).One hundred and twenty 4-week-old male BALB/c mice were selected and assigned into control (n=40) and experimental groups (n=80). The mice in the experimental group were injected intraperitoneally with Coxsackievirus B3 to establish the model of VMC, while the mice in the control group were injected intraperitoneally with an equal volume of DMEM solution. Fifteen mice in the experimental group and ten mice in the control group were sacrificed at 3, 7, 14, or 28 days after injection, and the myocardial specimens were obtained. The dynamic expression of VDR in the myocardium was determined by the immunohistochemical technique. The pathological changes in the myocardium were examined using hematoxylin and eosin staining.In the experimental group, the mice had significantly increased expression of VDR after virus injection (P<0.01); the expression of VDR reached the peak at 7 days after injection, and then declined gradually; the expression of VDR remained high at 28 days after injection. At 3, 7, 14, and 28 days after injection, the expression of VDR in the experimental group was significantly higher than that in the control group (P<0.01). Moreover, in the experimental group, the changes in the pathological score of the myocardium were in accordance with the changes in the expression of VDR; the expression level of VDR in the myocardium was positively correlated with the pathological changes in the myocardium in the experimental group (P<0.01).VDR may be involved in the inflammatory-immune process in the pathogenesis of VMC.
'/%3e%3c/defs%3e%3cpath fill='%23012169' d='M0 0h10080v5040H0z'/%3e%3cpath stroke='%23fff' stroke-width='.6' d='m0 0 6 3m0-3L0 3' clip-path='url(%23b)' transform='scale(840)'/%3e%3cpath stroke='%23e4002b' stroke-width='.4' d='m0 0 6 3m0-3L0 3' clip-path='url(%23c)' transform='scale(840)'/%3e%3cpath stroke='%23fff' stroke-width='840' d='M2520 0v2520M0 1260h5040'/%3e%3cpath stroke='%23e4002b' stroke-width='504' d='M2520 0v2520M0 1260h5040'/%3e%3cg fill='%23fff'%3e%3cuse xlink:href='%23d' x='2520' y='3780'/%3e%3cuse xlink:href='%23a' x='7560' y='4200'/%3e%3cuse xlink:href='%23a' x='6300' y='2205'/%3e%3cuse xlink:href='%23a' x='7560' y='840'/%3e%3cuse xlink:href='%23a' x='8680' y='1869'/%3e%3cuse xlink:href='%23e' x='8064' y='2730'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)