Gene regulatory dynamics during craniofacial development in a carnivorous marsupial

Next, we asked what biological processes are associated with active regulatory regions in the dunnart face. To accomplish this, we linked peaks to genes in order to associate functional annotations of coding genes with the candidate regulatory elements that likely regulate their expression. To make use of resources available in model organisms such as GO databases, we converted all dunnart gene IDs to mouse orthologous genes for downstream applications. This reduced the dataset to 35,677 putative enhancers and 8589 promoters near (within 1 Mbp) genes with a one-to-one ortholog in mouse (Supplementary file 1a).

We found that gene annotations for both promoters and enhancers were enriched for 23% of the same GO terms, including cellular processes (protein localization to plasma membrane, protein localization to cell periphery, regulation of cell morphogenesis, positive regulation of cell migration) and development (axon development, camera-type eye development, muscle tissue development, striated muscle development). By contrast, 44% of GO terms were uniquely enriched among genes assigned to promoters and were related to mRNA processing, transcription, mRNA stability, cell cycle, and mRNA degradation (Figure 2A, Supplementary file 1b) and uniquely enriched GO terms for genes assigned to candidate enhancers corresponded to processes indicative of early embryonic development (Figure 2A, Supplementary file 1c).

Predicted functional enrichment for dunnart peaks.

(A) 304 significantly enriched GO terms clustered based on similarity of the terms. The function of the terms in each group is summarized by word clouds of the keywords. Rows marked by P were driven by genes linked to putative promoters, rows marked by E were driven by genes linked to putative enhancers. (B) Enriched TF motifs for transcription factor families (HOMER). PWM logos for preferred binding motifs of TFs are shown. The letter size indicates the probability of a TF binding given the nucleotide composition.

Terms related to facial skeleton development were enriched among genes assigned to putative dunnart enhancers, including bone cell development, muscle cell development, secondary palate development, roof of mouth development, and mesenchyme development, consistent with dunnart craniofacial morphology (Cook et al., 2021). Enhancers active near important palate genes (Won et al., 2023; Zarate et al., 2024) such as SHH, SATB2, MEF2C, SNAI2, and IRF6 in the dunnart at birth may highlight potential regulatory mechanisms driving early palatal closure. In addition, terms related to the development of the circulatory system, including regulation of vasculature development, circulatory system process, and blood circulation were enriched among genes linked to predicted enhancer regions (e.g., ACE, PDGFB, GATA4, GATA6, VEGFA). This is consistent with observations that show the oral region of newborn dunnarts is highly vascularized, with blood vessels visible through their translucent skin at birth (Cook et al., 2021).

To gain further insight into dunnart gene regulation at this developmental stage, we scanned putative enhancers and promoters for 440 known Homer vertebrate motifs and tested for enriched TFs (Heinz et al., 2010). Enhancers were significantly enriched for 170 TFs relative to a background set of random GC- and length-matched sequences (FDR-corrected, p < 0.01), including those with known roles in differentiation of cranial neural crest cells (TWIST, HOXA2), skeletal morphogenesis (DLX5, CREB5, HOXA2), bone development (ATF3, RUNX), cranial nerve development (ATOH1), and/or facial mesenchyme development (LHX2, FOXP1, MAFB; Figure 2B, Figure 2—figure supplement 1). Consistent with the GO enrichment, TFBS in promoter sequences were dominated by transcriptional initiation regulatory sequences, with significant enrichment for 13 TFs (FDR-corrected, p < 0.01) including RFX3, RFX2, NRF, NRF1, GRY, ZBTB33, RONIN, JUND, and GFX (Figure 2B, Figure 2—figure supplement 1).

Next, we assessed predicted target gene expression by performing bulk RNA-seq in dunnart face tissue collected on the day of birth. There were 12,153 genes reproducibly expressed at a level >1 TPM across three biological replicates, with the majority of genes expressed (67%; 8158/12,153) associated with an active enhancer and/or promoter peak. Most enhancers (78% of all enhancers, Supplementary file 1d) or promoters (87% of all promoters, Supplementary file 1e) were linked to a reproducibly expressed gene in the dunnart (Figure 3A). Additionally, the majority of reproducibly expressed genes near a regulatory peak (61%) were associated with both an enhancer and promoter region (Figure 3B), highlighting the correspondence in active regulatory elements and expressed genes at this time point in the dunnart face. Genes with a medium-to-high expression level at this stage (>10 TPM) and associated with at least one putative enhancer and promoter were enriched for biological processes including ‘in utero embryonic development’, ‘skeletal system development’, ‘muscle tissue development’, ‘skin development’, ‘vasculature development’, and ‘sensory organ development’ (Figure 3C, Supplementary file 1f). Enrichment for the term ‘in utero embryonic development’ is indicative of the altricial nature of the dunnart neonate. In a previous study, we showed that in the dunnart, ossification begins post-birth (day of birth young corresponding approximately to embryonic day (E) 12.5 in mouse) and that the dunnart neonate instead likely relies on the well-developed cranial muscles and an extremely large chondrocranium for structural head and feeding support during early pouch life (Cook et al., 2021; Clark and Smith, 1993). Consistent with this, we observed high expression (>20 TPM) of the key head myogenesis genes (Bentzinger et al., 2012; Buckingham, 2017; Lin et al., 2006), MYOD1, MYF6, MEF2C, PAX3, MYL1, and MYOG, essential genes regulating chondrogenesis (Lefebvre et al., 2019; Yip et al., 2019), SOX9, COL2A1, and FGFR1 and genes that act upstream of osteoblast differentiation (Qi et al., 2003; Huang et al., 2007), MSX1, MSX2, CEBPA/G, ALPL, DLX3, DLX5, FGFR1, and FGFR2 (Supplementary file 1d, e).

Genes linked to craniofacial enhancers and promoters in the dunnart are reproducibly expressed and involved in embryonic vasculature, muscle, skin, and sensory system development.

(A) The majority of nearest genes assigned to candidate enhancers and promoters were reproducibly expressed in dunnart face tissue, and (B) reproducibly expressed genes in dunnart were associated with both a promoter and enhancer region. (C) Biological processes enriched for genes medium to highly expressed (>10 TPM) and linked to both a promoter and enhancer region (FDR-corrected, p < 0.01).