Shifts in embryonic oxygen levels cue heterochrony in limb initiation

Meng Zhu; ChangHee Lee; Rinaldo Catta-Preta; Clifford J Tabin

doi:10.1016/j.cell.2026.08.051

. Author manuscript; available in PMC: 2026 Oct 6.

Published before final editing as: Cell. 2026 Sep 24:S0092-8674(26)01062-7. doi: 10.1016/j.cell.2026.08.051

Shifts in embryonic oxygen levels cue heterochrony in limb initiation

Meng Zhu ¹, ChangHee Lee ¹, Rinaldo Catta-Preta ^1,^{^}, Clifford J Tabin ^1,^*

PMCID: PMC13636768 NIHMSID: NIHMS2212997 PMID: 42785299

Abstract

Heterochrony, or the alteration of developmental timing, is an important mechanism of evolutionary change. Avian species display synchronized growth of the forelimbs and hindlimbs, while mammalian species show delayed hindlimb development. We find that mammalian limb heterochrony is evident from the start of limb bud formation, and is associated with heterochronic expression of T-box transcription factors. This heterochronic shift is not due to changes in cis-regulatory sequences controlling T-box gene expression, but unexpectedly, is dependent upon differential oxygen levels to which avian and mammalian embryos are exposed prior to limb initiation. This is mediated, at least partially, by cRel, a transcription factor in the NFKB family, and by the oxygen-sensing transcription factor hypoxia inducible factor alpha (Hif1a). Together, these results provide mechanistic understanding of an important example of developmental heterochrony and exemplify how the maternal environment can regulate timing during embryonic development.

Graphical Abstract

graphic file with name nihms-2212997-f0001.webp

In Brief:

Mammalian embryos see a substantial increase in available oxygen once the placenta forms. This increase In oxygen exposure serves as a trigger for initiating the development of the hindlimb, thus controlling developmental timing, and more broadly linking maternal environment to evolutionary changes in body plan.

Introduction

Tetrapods have two pairs of limbs, located at rostral (forelimb) and caudal (hindlimb) positions along the trunk. Although the two pairs of appendages follow the same general developmental blueprint, they display differences in temporal pattern that align with evolutionary lineages¹. For example. in avian species, the early embryonic stages of limb outgrowth are synchronized; while in mammalian species, the hindlimb develops much later than that of the forelimb¹. The extent of this heterochrony varies among mammals such that the marsupial (metatherian) mammals display a more pronounced delay than placental (eutherian) mammals, and among the marsupial species, the ones that have shorter gestation time exhibit higher levels of heterochrony. The heterochronic delay in hindlimb development in marsupials has been postulated to represent an “energy trade-off” strategy^2,3, which prioritizes the growth of immediately required structures in the context of limited nutrient availability during early developmental stages. However, a molecular-level understanding of limb heterochrony regulation remains entirely unsolved, both for marsupial mammals and placental mammals. Indeed, in a more general sense, the control of timing during development is vastly under-studied relative to the control of spatially localized gene expression.

Here, we investigate the molecular basis of limb heterochrony between avian and mammalian species. We find that limb heterochrony in mammals starts at the limb initiation stage. The delayed onset of hindlimb development can be correlated with expression of Tbx4, a transcription factor known to regulate the first steps of hindlimb formation. By coupling RNA-sequencing analyses with gain-of-function screens, we identify a repressive role of the NFKB transcription factor cRel in regulating Tbx4 expression. Unexpectedly, we found that the low oxygen levels that early mammalian embryos experience, a by-product of viviparity, causes upregulation of cRel, together with hypoxia inducible factor 1 alpha (Hif1a), which attenuate Tbx4 expression in the early mammalian hindlimb. Our study presents mechanistic understanding of mammalian limb heterochrony, and illustrates how the maternal environment can instruct timing of tissue development in the fetus.

Results

Heterochrony in limb development between mammals and birds can be traced to the start of limb formation via epithelial-to-mesenchymal transition

To interrogate the regulation of limb heterochrony, we first asked whether differences in developmental timing were apparent from the earliest steps of limb development. To that end we characterized the timing of limb bud initiation in both chick and mouse embryos. Limb development starts with an epithelial-to-mesenchymal transition (EMT) of epithelial cells residing in the limb fields of the lateral plate mesoderm (LPM). This generates a mesenchymal population of limb progenitors and the emergence of a growing mound of cells that will transform into the paddle-like limb bud. Our previous work established that the EMT in the chicken embryo takes place in the limb fields at 45 to 52-hour post-gastrulation (Hamburger Hamilton (HH) stages 12–13 (in the forelimb, and 48 to 53-hour post-gastrulation (HH13–14) in the hindlimb⁴. EMT in the mouse forelimb field was previously characterized as occurring between 60-hour post-gastrulation (E8.5) and 66-hour post-gastrulation (E8.75)⁴. We extended this analysis to the mouse hindlimb by examining 72-hour post-gastrulation to 96-hour post-gastrulation (E9.0-E10.0) mouse embryos. We found that at 84-hour post-gastrulation (E9.5), majority of the cells in the hindlimb forming regions were still epithelial, as indicated by polarized F-actin and LAMININ⁴ and a monolayer organization in the LPM epithelium (Figure 1A). By contrast, at 90-hour post-gastrulation (E9.75), most of the cells in the hindlimb region showed mesenchymal morphology, with unpolarized distribution of both F-actin and Laminin and a multi-layer stacking of unorganized cells (Figure 1A). These results thus show that the EMT responsible for initiating the hindlimb in mouse embryos occurs between 84- to 90-hour post-gastrulation (E9.5 and E9.75). Together with previous results, quantifications pointed to a 2-hour difference in EMT timing between forelimb and hindlimb in chicken embryos, in contrast to an 18-hour difference in mice, thus, tracing limb heterochrony in mammals back to the EMT stage of limb development (Figure 1B).

Figure1. — (A) 84-hour post-gastrulation (E9.5) and 90-hour post-gastrulation (E9.75) mouse embryos hindlimb sections stained with F-actin, Laminin, DAPI to reveal the cellular epithelial/mesenchymal state. While at E9.5 the cells remain epithelial shapes, at E9.75, the cell completely transited to mesenchymal state. N=4 embryos for E9.5; N=2 embryos for E9.75. N=2 experiments. (B) Timeline of EMT in forelimb and hindlimb of mouse and chicken embryos. (C) Chicken hindlimb sections at 47-hour post-gastrulation (HH12) and 51.5-hour post-gastrulation (HH14) revealing the transcripts levels of Tbx4 by Hybridization chain reaction (HCR), and DAPI. (D) Quantification of chicken Tbx5 and Tbx4 transcripts levels across 27.5- to 72-hour post-gastrulation (HH8-HH18) in forelimb and hindlimb tissues. 0h(HH4), 27.5h(HH8), 35.5h(hh10), 47h(HH12), 51.5h(HH14), 53.5h(HH16) and 72h(HH18). Dotted lines indicated the EMT timings of forelimb and hindlimb respectively. N=2–6 embryos for each tissue/stage. (E) E9.0 and E9.75 mouse hindlimb sections revealing the transcripts levels of Tbx4 by HCR, and DAPI. (F) Quantification of mouse Tbx5 and Tbx4 transcripts levels across E7.5 to E10.0 in forelimb and hindlimb. 0h(E6.0), 36h(E7.5), 42h(E7.75), 48h(E8.0), 60h(E8.5), 72h(E9.0), 84h(E9.5), 90h(E9.75), 96h(E10.0) and 102h(E10.25). Dotted lines indicated the EMT timing of forelimb and hindlimb respectively. N=2–4 embryos are used for each tissue and each stage. N=2 experiments. (G) Images showing chicken hindlimb at 52.5-hour post-gastrulation (HH15) to stain Tbx4 transcripts by HCR, F-actin, DAPI to correlate with cellular state. Epithelial cells have fewer Tbx4 transcripts than mesenchymal cells. dotted lines=magnified areas. (H-I) Quantifications of Tbx5 (H) and Tbx4 (I) transcripts in epithelial/mesenchymal cells of chicken forelimb/hindlimb tissues. Dotted lines=transitioning stage when epithelial and mesenchymal cells both present. There are clear correlations between Tbx5 and Tbx4 transcripts number and epithelial/mesenchymal state. N=8 embryos, N=2 experiments. Scale bars (overview) = 150μm; scale bars (magnification) =15μm.

The heterochrony in EMT correlates with timing of expression of T-box transcription factors

We reasoned that the discrepancy in timing of EMT between mouse and chicken embryos would likely be attributable to the heterochronic expression of key limb initiation regulators. To explore this hypothesis, we analyzed the transcriptional dynamics of signaling pathway ligands and transcription factors previously described as regulating limb initiation, including Fgf10, Wnt8c, Wnt2b, Tbx5 and Tbx4^5–10. Fgf10 and Wnt2b act downstream of Tbx5 to initiate forelimb EMT, while Fgf10 and Wnt8c act downstream of Tbx4 to initiate hindlimb EMT⁶. To precisely determine their temporal correlation to the timing of EMT, we utilized Hybridization Chain Reaction (HCR)¹¹, a single molecule fluorescence in situ hybridization technology, to visualize transcripts of each gene and to determine their expression dynamics at spatial and temporal resolution. The designed probes show high specificity to the genes of interest as the expression profiles were consistent with results obtained using other methods^5,12,13 (Figure 1C,E,G, Figure S1A–F).

Our analysis across developmental stages in the forelimb and hindlimb regions of chicken and mouse embryos revealed that the dynamics of the T-box transcription factors, Tbx5 and Tbx4, exhibited the strongest correlation with the timing of EMT (Figure 1C–F; Figure S1A–I). Specifically, we found that the time when EMT occurs consistently corresponds to the time when half-maximum levels of the Tbx5 or Tbx4 transcripts are reached in the limb mesenchyme of both species (Figure 1D,F), a correlation not seen with the other genes tested. Furthermore, we found that this correlation is reflected at the cellular level: cells that transit earlier to the mesenchymal state display higher levels of Tbx5/4 transcripts than those yet to transit (Figure 1G–I; Figure S1G–I). Together, these results suggest T-box transcription factors as the plausible regulators of EMT heterochrony.

Genetic studies show that the depletion of Tbx5 or Tbx4 results in defects in forelimb or hindlimb outgrowth, while their overexpression can induce ectopic limb growth^8,9,12,14,15. To directly test whether Tbx5/4 expression regulates the timing of limb formation, we examined mouse embryos deficient for each of these genes. We had previously shown that depletion of Tbx5 impairs forelimb EMT⁴. We therefore turned our attention to an established Tbx4 knockout mouse line⁹. At 90-hour post-gastrulation (E9.75), the cells in hindlimb regions of heterozygous Tbx4 knockout embryos have already undergone EMT, as characterized by the localization of F-actin and Laminin. In contrast, a significant proportion of cells in the Tbx4 null embryos remained epithelialized and localized to the basal region of the LPM (34% showing polarization of both F-actin and Laminin in homozygous mutants, compared to 0.05% in Wildtype; Figure S2A–C). These results, together with the previous studies, show that the limb EMT heterochrony in the mouse embryo is associated with the heterochronic expression of Tbx5 and Tbx4.

Expression of known upstream activators of Tbx4 is concordant between chicken and mouse embryos

To investigate the mechanisms leading to delayed Tbx4 expression in the mouse embryo, we started by examining the dynamics of the signaling pathways and transcription factors already known to regulate Tbx4 expression. As gastrulation commences, the genes in the Hox clusters are expressed in a 5’ to 3’ order, resulting in the segmented anterior-posterior patterning of the trunk¹⁶. The forelimb regions are marked by expression of Hox4 paralogues, while the hindlimb regions are specified by Hox9 paralogues^17,18. Interactions with a set of signaling pathways involving Bone Morphogenetic Protein (BMP) activity, Wnt activity, and the inhibition of Retinoic acid (RA) signaling are also required for the specification of posterior LPM^19,20. These inputs together activate expression of the transcription factors Pitx1 and Isl1 that are enriched in posterior LPM, marking the regions that form the prospective hindlimb^21–23. The knockout of each of these transcription factors in the mouse embryo leads to downregulation of Tbx4 at the hindlimb initiation stage^24–26.

We first examined signaling essential for LPM specification. The existing literature does not suggest differences in the timing of Wnt or RA activation between the mouse and chick^27–30. Additionally, our examination of the temporal and spatial distribution of phosphorylated Smad proteins, indicators of BMP pathway activation, revealed no differences between species (Figure 2A–B). Thus, there do not appear to be temporal alterations in signaling pathways required to specify the hindlimb in the two species.

We then probed the transcriptional dynamics of Hoxb9, Pitx1 and Isl1 expression in mouse and chicken embryos by HCR (Figure 2C–F), in terms of hours post-gastrulation, and used the time to reach half-maximum as a metric for the timing of gene expression. We found that Hoxb9, Pitx1, and Isl1 activation times were similar in both species: In the chick, where the forelimb and hindlimb initiate at nearly the same time, expression of Tbx5, Tbx4, Pitx1 and Isl1 are all initiated simultaneously. In contrast, in the mouse, where hindlimb initiation is delayed relative to the forelimb, Hoxb9, Pitx1 and Isl1 are still activated together with Tbx5, despite the delay of Tbx4 (Figure 2D,F; Figure S2D–E). Consistently, we found that overexpression of Pitx1, a Tbx4 activator that directly binds to the Tbx4 enhancer region, does not induce precocious Tbx4 expression (Figure S2F–I). These data together suggest that the 22 hour delay in Tbx4 expression in the mouse hindlimb is not a consequence of delayed specification of the hindlimb lateral plate mesoderm (LPM), or a lag in the expression of known upstream activators.

Shifted timing of Tbx4 activation is not due to alterations in associated cis-regulatory elements

The modern view posits that most evolutionary change is attributable to mutations in cis-regulatory elements (CREs)³¹. To determine whether CRE changes proximal to Tbx4 are responsible for the shifted timing of that gene’s expression in the mouse, we compared Tbx4 enhancer activity between mouse and chicken. For the mouse, previous studies identified two genomic regions, named HLEA and HLEB, as necessary and sufficient to activate Tbx4 expression in the developing hindlimb³². To identify chicken Tbx4 enhancers, we took advantage of an ATAC-seq analysis previously carried out in the lab³³, and selected regions that showed accessibility in 72-hour post-gastrulation (HH18) hindlimb mesenchyme but not in the forelimb or flank. This criterion filtered the pool into 4 candidate genomic regions located within 100kb of the Tbx4 transcriptional start site (Figure 3A, Figure S3A). To test whether these regions can drive the transcription in a pattern resembling that of endogenous Tbx4, we electroporated enhancer:Citrine reporter constructs, each carrying a candidate region, into chicken forelimb, flank or hindlimb tissues at 51.5-hour post-gastrulation (HH14) (Figure 3B). When the electroporated embryos were examined at 72-hour post-gastrulation (HH18), 1 of the 4 candidate regions drove strong signal, restricted to the hindlimb tissue (Figure 3C–D; Figure S3A). This enhancer was designated as Tbx4-Rec1. To reveal the temporal activation pattern of Tbx4-Rec1, we quantified Citrine signal in hindlimb tissues of different stages by flow cytometry. This revealed that Tbx4-Rec1 activity is initiated at 47-hour post-gastrulation (HH12) and becomes strong by 53.5-hour post-gastrulation (HH16), paralleling the timing of endogenous Tbx4 expression (Figure 3E; Figure S3B–E). Thus, Tbx4-Rec1 activity resembles the expression profile of Tbx4 both spatially and temporally in the chicken embryo (Figure 1C–D). Finally, to determine whether Tbx4-Rec1 is functionally required for Tbx4 expression, we inhibited the endogenous Tbx4-Rec1 region by deploying the dCas9-KRAB assay, which utilizes a catalytically dead CAS9 fused to the Krüppel-associated box (KRAB) transcriptional repression domain. This system, guided by specifically designed gRNAs targeting Tbx4-Rec1, acts to methylate the genomic region, thereby repressing it³⁴. We electroporated a construct carrying a gRNA targeting the middle position of Tbx4-Rec1 and the coding sequence of dCas9-krab, into the 47-hour post-gastrulation (HH12) hindlimb field of chicken embryos. After 12 hours, the electroporated cells showed significant reduction of Tbx4 transcripts when compared to non-electroporated counterparts, or cells electroporated with a control gRNA (Figure 3F–G). Together, these experiments implicate Tbx4-Rec1 as a critical enhancer for Tbx4 expression in the embryonic chicken hindlimb.

Figure 3. — (A) Chromatin accessibility of chicken Tbx4 enhancer (Tbx4-Rec1). Data was extracted from ATAC-seq in Young et al., 2019³³. Tbx4-Rec1 genomic region (dotted blue square) has higher accessibility in hindlimb than forelimb or flank in 72-hour post-gastrulation (HH18) chicken embryos. (B) Scheme of enhancer bashing assay. Enhancer bashing constructs carrying each chicken Tbx4 enhancer candidate were electroporated in 47-hour post-gastrulation (HH12) chicken hindlimb fields. pCMV-mCherry construct was co-electroporated as control. Enhancer activity was examined at HH18 and 96-hour post-gastrulation (HH23). (C) Quantification of the 4 chicken Tbx4 enhancer candidates’ activity at HH18 by flow cytometry. Enhancer activity is calculated as the signal ratio between Green/Red. Mean value is shown for each condition. N=10 embryos for each group, N=3 experiments. Only Tbx4-Rec1 shows high activation. (D) Chicken hindlimb tissues electroporated with Tbx4-Rec1:Citrine and sectioned at HH18 and HH23 to examine its activity (Citrine) and DAPI. (E) Quantifications of mouse and chicken Tbx4 enhancers’ activity by flow cytometry. Data shown as mean± S.E.M. N=30 embryos for each condition, N=3 experiments. (F) 53.5-hour post-gastrulation (HH16) chicken hindlimb cells electroporated with dCas9-Krab constructs containing control gRNA or gRNA targeting Tbx4-Rec1. Embryos were fixed after 13h to reveal Tbx4 transcripts (HCR), ZsGreen and DAPI. Yellow circles=electroporated cells. (G) Quantifications of Tbx4 transcripts in conditions shown in F. Tbx4 transcripts are significantly downregulated in cells transfected with Tbx4-Rec1 gRNA. ****p<0.0001, ns, not significant, Mann-Whitney test. Data shown as individual points(cells) with mean± S.E.M. N=19 cells for Ctrl gRNA Citrine-; N=38 cells for Ctrl gRNA Citrine+; N=30 cells for Tbx4-Rec1 gRNA Citrine-; N=16 cells for Tbx4-Rec1 gRNA Citrine+. N=2–3 embryos for each condition, N=2 experiments. (H) Strategies for producing ES cell-embryo chimera. Wildtype ES cells were transfected with constructs containing piggybac transposase or piggybac LTRs enclosing enhancer:Citrine cassette and CMV:H2B-mCherry cassette. The transgenic ES cells were then aggregated with E2.5 (8-cell stage) mouse embryos and cultured until E3.5. The chimeric embryos were then transferred to pseudo-pregnant females and dissected at desired stages before and after Tbx4 expression commences. ES-embryo chimeric embryos were recovered and fixed at E8.5 and E10.0 to reveal enhancer activity (Citrine), mCherry, Tbx4 transcripts (HCR) and DAPI. N=6 embryos. N=2 experiments. Scale bars (overview)=150μm; scale bars (magnification)=15μm.

Sequence analyses suggest that the 3’ DNA sequence of Tbx4-Rec1 highly resembles that of the mouse Tbx4 enhancer, HLEB (Figure S3F). This suggests that the chicken Tbx4-Rec1 and mouse HLEB Tbx4 enhancers are homologous sequences. There is no obvious counterpart to HLEA in the chick genomic sequence near the Tbx4 gene³². To test whether the mouse and chick enhancers are functionally conserved, we performed enhancer-swap experiments to orthogonally express the Tbx4 enhancers across species. To introduce the mouse Tbx4 enhancer into the chicken embryo, we performed electroporation experiments with Citrine reporter constructs carrying either HLEA or HLEB, as above (Figure 3D–E). We found that both HLEA and HLEB were strongly activated, driving reporter expression at a similar level to Tbx4-Rec1 (Figure 3D). Moreover, the timing of activation of both mouse enhancers was similar to that of Tbx4-Rec1 (Figure 3E, Figure 1D). To introduce the chicken Tbx4 enhancer into the mouse embryo, we produced Embryonic Stem cells (ES)-embryo chimeras (Figure 3H). We first generated transgenic mouse ES cell lines each with a random genomic insertion of a Tbx4-Rec1:Citrine reporter cassette (Figure 3H). The transgenic ES cells were then aggregated to 8-cell stage mouse embryos, which were transferred to pseudo-pregnant female mice, and recovered at stages before and after Tbx4 expression. We optimized the ES cells aggregation protocol to achieve nearly full donor chimerism (Figure S3G, Methods). We first examined HLEB:Citrine activity as a positive control. As expected, no Citrine signal can be detected in the hindlimb LPM region from embryos recovered at 66-hour post-gastrulation (E8.75) (Figure S3H); However, strong activity can be detected only in hindlimb but not forelimb regions at 90-hour post-gastrulation (E9.75) (Figure S3I). Such an activation pattern recapitulates the endogenous mouse Tbx4 expression. We then examined Tbx4-Rec1:Citrine activity, and found that, similar to HLEB:Citrine, no Citrine signal was seen at 60-hour post-gastrulation (E8.5), but strong activation was detected at 96-hour post-gastrulation (E10.0) in the hindlimb (Figure 3H; Figure S3J). The inter-species enhancer analysis experiments thus demonstrate that the mouse and chicken Tbx4 enhancers are functionally equivalent, and that any sequence variation between them are insufficient to account for the temporal shift in Tbx4 expression observed in the mouse embryo.

NFKB transcription factors negatively regulate Tbx4 expression

We next turned into the alternative model that differentially expressed trans-acting factors are responsible for difference in the timing of Tbx4 expression between the two species. To address this hypothesis, we conducted bulk RNA-sequencing of forelimb and hindlimb tissues from 60-hour post-gastrulation (E8.5) and 90-hour post-gastrulation (E9.75) mouse embryos, as well as forelimb and hindlimb tissues of 47-hour post-gastrulation (HH12) chicken embryos (Figure S4A).

Consistent with the literature, our sequencing analyses identified Tbx5 and Raldh1a2 as significantly enriched in forelimb tissues^12,35, and Cdx2, Evx1, Pitx1 as enriched in the hindlimb tissues across both species^25,36,37 (Figure S4A). Comparing between stages, we found about 200 genes differentially expressed between 60-hour post-gastrulation (E8.5) and 90-hour post-gastrulation (E9.75) hindlimb tissues in the mouse.

To identify differentially expressed genes that might be functionally important for Tbx4 expression, we reviewed the documented function of each gene, and filtered out those with known discordant expression in HH12 chicken hindlimb, narrowing the focus to 22 factors related to metabolism, paracrine signaling, and transcriptional regulation (Table S1; Figure S5). To test whether the expression of any of these factors impact Tbx4 expression, we mis-expressed each gene in the chicken hindlimb. We designed two types of assays based on the gene’s expression profile: 1) potential inhibitors: genes with higher expression at 60-hour post-gastrulation (E8.5) than 84-hour post-gastrulation (E9.5) were overexpressed in the HH12 chicken hindlimb field, and their impact on Tbx4 expression assessed at 96-hour post-gastrulation (HH23). 2) potential activators: genes upregulated from E8.5 to 90-hour post-gastrulation (E9.75) were overexpressed in gastrulating HH4 chicken embryos to evaluate potential advancement in Tbx4 expression by 35.5-hour post-gastrulation (HH10).

This screen identified cRel as a potent repressor of Tbx4 expression (Figure 4A–F; Figure S4B–C). cRel is an effector of the NFKB signaling pathway involved in immune and stress response. We find that the expression of cRel inversely correlates with that of Tbx4 (Figure 4A–C; Figure S4C). Moreover, overexpression of cRel, but not the other candidates that we tested, results in a marked reduction in hindlimb bud size and consistently lower Tbx4 transcript levels only in electroporated cells, suggesting a cell-autonomous effect of REL in blocking Tbx4 expression (Figure 4D; Figure S4D).

Analysis of cRel knockout mouse embryos from 60-hour post-gastrulation (E8.5) to 84-hour post-gastrulation (E9.5) revealed that cRel-null embryos exhibited significant elevation of Tbx4 expression in hindlimb tissues at E8.5 compared to heterozygous littermates (Figure 4E–F). However, this ectopic expression of Tbx4 was not observed across all cells, and predominantly occurred on only one side of the embryo (Figure 4E). Despite this heterogeneous expression, the increased Tbx4 levels were sufficient to advance EMT timing by 72-hour post-gastrulation (E9.0), as cRel null hindlimb tissues show a significant higher proportion of mesenchymal cells than heterozygous at E9.0 (Figure 4G–H). Possibly due to the partial penetrance, we found no difference in hindlimb bud size between cRel null and heterozygous littermates at 84-hour post gastrulation (E9.5) (Figure S4F–G).

One possible explanation for the partial penetrance we observed could be redundancy with related factors co-expressed in the limb bud. Besides cRel, the NFKB family includes 4 other transcription factors, Rela, Relb, Nfkb1 and Nfkb2. Of the 5 factors, cRel, Rela and Nfkb1 were detectable in 60-hour post-gastrulation (E8.5) mouse hindlimb tissues in bulk RNA-seq (Figure S4B). To test whether other NFKB transcription factors similarly inhibit Tbx4 expression, we separately overexpressed Rela and Nfkb1 in the chicken hindlimb. We found that the overexpression of Rela, but not Nfkb1, led to significant reduction of hindlimb bud size (Figure 4D; Figure S4E). Examination of Tbx4 expression revealed a significant downregulation of Tbx4 transcripts in cells overexpressing Rela. By analyzing a recently published single-cell RNA sequencing data set, we found that cRel and Rela were similarly downregulated from 60-hour post-gastrulation (E8.5) to 84-hour post-gastrulation (E9.5) in the LPM³⁸. These data suggest Rela as a potential compensatory factor that may delay Tbx4 expression akin to cRel.

To further investigate the relationship between cRel expression and forelimb–hindlimb heterochrony, we assessed the impact of ectopic cRel expression on the forelimb regulator Tbx5. Although cRel overexpression led to a reduction in Tbx5 expression, this effect was comparatively modest (Figure S4H–J). These findings suggest that both differential cRel expression levels and an intrinsic difference in the sensitivity of forelimb versus hindlimb gene regulatory networks to cRel contribute to the heterochronic expression of Tbx5 and Tbx4, and consequently to the timing of EMT.

Taken together, these analyses revealed an unanticipated inhibitory role of NFKB transcription factors in delaying Tbx4 expression in the developing mouse hindlimb bud.

Maternal hypoxia impedes Tbx4 expression in the mouse hindlimb

Our identification of cRel and Rela as key genes responsible for the delayed activation of Tbx4 in the mouse hindlimb expands the set of known transcription factors in the early limb bud gene regulatory network. However, at a conceptual level, all this does is shift our analysis heterochrony is regulated from one transcription factor to a different upstream transcription factor. The underlying mechanism that establishes the time of hindlimb initiation remained elusive. Therefore, we continued to work our way upstream, shifting our focus to the regulatory mechanisms governing the temporal expression of the upstream factors cRel and Rela. In going through the literature on the regulation of these genes, we were struck by the reports that the expression of NFKB transcription factors can be induced by hypoxia in a number of contexts^39,40. This was intriguing because chicken eggs develop in environment close to normoxia (20% oxygen), whereas early mouse embryos are in hypoxia (2%-5% oxygen)^41–45. Notably, there is a transition in the mouse embryo from low oxygen levels to normoxia as the placental connection to the maternal circulation becomes robust, at a time closely correlating with hindlimb initiation. consistent with the direct measurement of oxygen concentration during early mouse development, reported by others^42–43, immunostaining for HIF1A (nuclear a HIF1A being a marker for cells in a hypoxic environment) revealed dynamic, stage-specific patterns. At HH12, chicken embryos exhibited low levels of nuclear HIF1A in the hindlimb field. In contrast, mouse embryos at a comparable stage (60 hours post-gastrulation, E8.5) displayed high levels of nuclear HIF1A. Nuclear HIF1A is diminished one day later, at 84-hour post-gastrulation (E9.5), coinciding with robust Tbx4 expression and hindlimb EMT initiation (Figure 5A–C).

Figure 5. — (A-B) HIF1A staining in forelimb and hindlimb of 47-hour post-gastrulation (HH12) chicken embryo (A) and E8.5 and E9.25 mouse embryos (B). Dotted lines=magnified regions. (C) Quantifications of nuclear Hif1a (normalized by DAPI) in conditions shown in A-B. ****p<0.0001, One way ANOVA test. Data shown as mean ± S.E.M, each dot=a cell. N=2 embryos for each stage. N=2 experiments. (D) Scheme of *ex-utero* culture. The cultured embryos morphologically resemble that of E8.5 mouse embryo. (E) *Ex-utero* cultured embryos in 5%/20% O₂ level as in (D) were fixed to examine Tbx5, Pitx1 and Tbx4 by HCR,and DAPI in forelimb and hindlimb after culture. Natural E8.5 embryos were compared as control. Dotted lines=magnified regions. (F-G) Quantifications of Tbx5 (F) and Tbx4 (G) in forelimb and hindlimb from conditions shown in E. ns, not significant, *p<0.05, ***p<0.001, ****p<0.0001, One way ANOVA test. Data shown as mean ± S.E.M. Each dot=an embryo. N=8 experiments for both 5% and 20% O₂. (H) *Ex-utero* cultured embryos in 5%/20% O₂ for 2 days and immunostained with laminin, F-actin and DAPI for examining epithelial/mesenchymal state of hindlimb cells. (I) Quantification of number of epithelial and mesenchymal cells for conditions shown in H. ****p<0.0001, Fisher’s exact test. Data is shown as bar chart, numbers in each bar represents the number of cells examined. N=4 embryos for 5% O₂; N=3 embryos for 20% O₂. N=2 experiments. Significant higher proportion mesenchymal cells are seen in hindlimb cells cultured in 20% O₂. (J) Quantitative PCR (qPCR) analysis of cRel, RelA, NFKB1 expression in embryos ex-utero cultured in 5%/20% O₂. **p<0.01, ns, not significant. Mann-Whitney test. Data shown as mean ± S.E.M. Each data point represents one experiment. Hypoxia culture induces significant upregulation of *cRel*. (K) *Ex-utero* cultured embryos from E6.5+2 days in normoxia or hypoxia, visualize cRel transcripts (HCR) and DAPI. (L) Quantification of *cRel* transcripts in conditions from K. ****p<0.0001, Mann-Whitney test. Data shown as mean ± S.E.M. Each datapoint=one embryo. N=5 experiments. Scale bars (overview) = 150μm; scale bars (magnification) =15μm.

To test whether oxygen levels could play a role in regulating the timing of Tbx4 expression in mouse embryos, we optimized an ex-utero culture system that allowed gastrulating embryos, dissected at 12-hour post-gastrulation (E6.5), to develop until approximately 66-hour post-gastrulation (E8.75) (Figure 5D; Figure S6A–B; Methods). The culture method is based on previous protocols^46,47 but with two modifications: 1) 5% FBS was added into the culture medium as an alternative to human cord serum; 2) the ecto-placental cone (EPC) was removed to enhance oxygen equilibration within and surrounding the embryonic tissues. The latter modification was based on the observation that in the presence of an intact EPC, Hif1a expression was not completely extinguished in embryos cultured in normoxic conditions (Figure S6C), likely due to inefficient gas penetration of the yolk sac layer. The cultured embryos generally attain a similar morphology to naturally developed E8.5 embryos in the formation of head-fold, allantois, and a clear distinction of anterior and posterior regions. There were occasional defects in the epithelization of somite tissues, which is likely due to the lack of sufficient mechanical cues in culture⁴⁸. Nonetheless, we found robust expression of forelimb field and hindlimb field markers Tbx5 and Pitx1, indicating normal differentiation of LPM lineages (Figure 5E–F; Figure S6D–E; Figure S7A–D).

Leveraging the ex-utero culture system, we compared gene expression under different oxygen conditions. We recovered embryos at E6.75, and cultured them for 1.5 days, in either 5% or 20% oxygen, which correspond, respectively, to hypoxic and normoxic conditions. Remarkably, we found consistent activation of Tbx4 transcripts in mouse embryos cultured in normoxic conditions, which did not occur in those cultured in hypoxic conditions (Figure 5E,G, Figure S6E). Moreover, this activation of Tbx4 in normoxia-grown embryos is more pronounced than seen in cRel knockout embryos. Nearly all these embryos showed premature, elevated, bilateral expression of Tbx4 (Figure 5E–G). The normoxia cultured embryos additionally show significant upregulation of Pitx1 in the posterior LPM (Figure S6D; Figure S7D). This effect was specific to these hindlimb genes as, in contrast, the normoxia cultured embryos showed no significant change in Tbx5 expression (if anything, there was a trend towards down-regulation) (Figure 5F). Consistent with the ectopic activation of Tbx4, we found that the Tbx4 enhancer Tbx4-Rec1, also showed scattered, yet significant upregulation in the hindlimb field of embryos cultured in normoxia, relative to those grown in hypoxia (Figure S6E–F).

We then asked whether the precocious expression of Tbx4 in normoxia-cultured mouse embryos would lead to an acceleration of limb bud initiation. To that end, we examined the status of EMT in normoxia-cultured embryos after 48 hours, the equivalent of 66-hour post-gastrulation (E8.75). Using F-actin and LAMININ polarization as markers, we found that hindlimb embryos cultured in normoxia condition show a significantly higher proportion of mesenchymal cells in the hindlimb field (Figure 5H–I). Moreover, we found a correlation between Tbx4 expression and EMT status, such that regions with higher Tbx4 expression display more multi-layered mesenchymal cells (Figure S6G), and consistently, normoxia cultured embryos showed increased thickness of the hindlimb tissue comparing to that of hypoxia (Figure S6H). Together, these results suggest that forced elevation of oxygen levels result in precocious hindlimb induction, including Tbx4 expression, EMT, and limb bud growth, in mouse embryos. We could not carry out the reciprocal experiment, and ask whether culture in hypoxia delays hindlimb initiation in avian embryos because early chick embryos fail to develop under hypoxic conditions.

To systematically evaluate how oxygen levels influence embryonic tissue development in the mouse, we performed droplet-based single-cell RNA sequencing (10x Genomics) on embryos cultured under normoxic and hypoxic conditions. Leveraging previously published, temporally highly resolved mouse single-cell whole embryo atlas data⁴⁹, we were able to identify the embryonic cell states expected during gastrulation, across both conditions. These included neuroectodermal derivatives (forebrain, midbrain, hindbrain, and spinal cord), mesodermal populations (paraxial, intermediate, and lateral plate mesoderm), endoderm, primordial germ cells, neural crest cells, and extra-embryonic tissues such as yolk sac (Figure S7A–B). The fidelity of tissue identity across both culture conditions highlights the capacity of our ex-utero system to support coordinated morphogenesis and lineage specification.

While the overall cellular composition remained broadly conserved between hypoxic and normoxic embryos (Figure S7A–B), differential gene expression analyses uncovered tissue-specific responses to oxygen availability (Figure S7C–D). Consistent with other contexts, elevation of oxygen induces significant downregulation of glycolysis pathway regulators, such as Gapdh, Hk2, Slc2a1, Ldha and Aldoa (Figure S7C). We further leveraged the full transcriptome of each cell to infer their stages, by aligning each cell to the reference atlas and observing the distribution of cell stages of its k-nearest neighbor references. The inferred stages of both conditions fall close to the expected 60–66 hour post-gastrulation (E8.5-E8.75) range (Figure S7E–F), with a slight overall delay in the relative developmental progression observed for embryos grown in normoxic conditions. Notably, some cell states exhibit more clear differences in developmental progression between hypoxic and normoxic conditions, with lateral plate mesoderm derivatives on average exhibiting potentially slight transcriptional acceleration (Figure S7F). This analysis, thus, highlights effects of oxygen levels on a variety of embryonic tissues. Most importantly in the current context, expression of canonical hindlimb specifiers such as Pitx1, Tbx4, together with Hox9 paralogues, were markedly upregulated in normoxic conditions, not only providing additional validation of the HCR results, but also suggesting that the impact of higher oxygen concentration on hindlimb development is not limited to Tbx4, but rather it affects the process on multiple levels (Figure S7D; Figure 5).

Maternal hypoxia represses Tbx4 expression through cRel and Hif1a

We next sought to understand the downstream pathways mediating hypoxia’s repressive effect on Tbx4 expression. While high oxygen culture normally induces a metabolic transition characterized by the downregulation of glycolysis and upregulation of oxidative phosphorylation (OXPHOS), in our context we found that neither the pharmacological inhibition of glycolysis, the activation of OXPHOS under hypoxia (Figure S8A–C), nor upregulation of glycolysis under normoxia, was sufficient to induce the expression of Tbx4 (Figure S8D–E). These results argue against the possibility that the shift in temporal expression of Tbx4 we observed is due to metabolic alterations.

To explore whether hypoxic conditions modulate the timing of Tbx4 activation by regulating NFKB transcription, we performed qPCR on dissected hindlimb field tissues from embryos cultured under normoxia or hypoxia conditions. Consistent with the bulk RNA-sequencing analyses, among the NFKB family members, only the transcripts of Rela, cRel and Nfkb1 can be readily amplified (Figure 5I–J, Figure S9A). Moreover, qPCR analysis of dissected hindlimb tissue showed significant elevation of cRel under hypoxia (Figure 5J), with whole-embryo single-cell RNA-seq pseudobulk analysis showing a similar trend (Figure S9A; underpowered for low-abundance transcription factors at this scale). These findings substantiate the hypothesis that transitioning from hypoxia to normoxia triggers the downregulation of cRel, thereby activating Tbx4 expression.

To better understand how oxygen and cRel regulate Tbx4 expression, we sought to dissect the regulatory relationships between cRel and the key transcriptional regulator of cellular responses to hypoxia, Hypoxia-inducible factor 1α (Hif1a). Hif1a is rapidly induced upon oxygen depletion to orchestrate adaptive gene expression programs. To determine whether oxygen-dependent regulation of Tbx4 is mediated by Hif1a, we examined Tbx4 expression in contexts where Hif1a levels were increased or decreased. We found that Hif1a overexpression in chicken hindlimb field induced significant downregulation of Tbx4 and smaller limb bud, phenocopying the effect of cRel (Figure S10A–E; Figure 4D). As with cRel, overexpression of Hif1a in chicken forelimb also resulted in downregulation of Tbx5, albeit to a lesser extent (Figure S10 D–E). In addition and consistently, Hif1a depletion had no effect on Tbx5 expression (Figure S10M–N). Conversely, we generated Hif1a knockout embryos using the ES cell-embryo chimera approach, based on the design of a previously established Hif1a knockout mouse model⁵⁰ (methods). Both ES cell colony genotyping and immunofluorescence in chimeric embryos confirmed the complete loss of HIF1A protein in all embryonic cells (Figure S10G–J). By E8.5, while wild-type embryos exhibited low Tbx4 expression, Hif1a knockout embryos displayed a marked upregulation of Tbx4 in the hindlimb field (Figure S10K–L). These findings indicate that Hif1a depletion phenocopies cRel loss-of-function, allowing premature Tbx4 activation and positioning Hif1a as a key mediator of oxygen-dependent repression of Tbx4.

To further delineate the epistatic relationship between cRel and Hif1a in repressing Tbx4, we manipulated the expression of each factor individually and assessed the impact on the other. We observed that depletion of Hif1a did not significantly alter cRel expression (Figure S10C,F). In contrast, modulation of cRel directly affected Hif1a levels: overexpression of cRel in the developing chicken hindlimb led to increased Hif1a expression, whereas cRel knockdown resulted in reduced Hif1a levels (Figure S11A–D). These findings indicate that, although both cRel and Hif1a are activated by hypoxia, cRel further promotes Hif1a expression, consistent with previous observations in other biological contexts⁴⁰. As genetic depletion of Hif1a results in a more pronounced increase in Tbx4 expression than occurs when cRel is depleted, it is plausible that multiple factors work in parallel with cRel to promote Hif1a expression, thereby collectively inhibiting Tbx4 expression.

Overall, our results support a model where the placenta-dependent embryonic transition from hypoxia to normoxia leads to downregulation of cRel and Hif1a, which allows the activation of Tbx4, and hence relieving a block on initiation of limb development in mouse embryos (Figure 6); and - from a broader perspective - explaining a critical aspect of heterochrony in limb formation between certain mammalian and avian species.

Figure 6. — Forelimb and hindlimb develop at different times from the limb initiation stage in the mouse embryo but much less so in the chicken, our results show that the limb initiation heterochrony in the mouse is mediated by hypoxia that the mouse embryo exposes to during the limb field specification stage. The hypoxia environment inhibits limb initiation by delaying hindlimb initiation factor’s expression such as Tbx4. Hypoxia’s effect on Tbx4 expression is at least partially mediated by NFKB transcription factors such as cRel, and Hif1a. Altogether, our findings reveal the mechanisms behind limb initiation heterochrony and suggest an additional role of the maternal environment in directing the timing of embryonic development during post-implantation stages.

Discussion

Despite growing interest in the study of developmental timing^43–45, mechanistic understanding of heterochronic change remains limited. Our study suggests a model explaining the observed heterochronic delay (post-displacement) in the onset of hindlimb formation in mouse embryos relative to chicken by hypoxia (Figure 6). Avian species primarily obtain oxygen via passive transport through the eggshell, which is limited in its oxygen-carrying capacity. As such, embryonic tissues gradually become hypoxic as cell numbers and tissue size increase⁵¹. In contrast, mammals initially develop under hypoxic conditions due to the inaccessibility of embryonic tissues to oxygen. Subsequently, the maturation of a functional placenta allows efficient maternal-fetus oxygen exchange through the allantois and blood circulation, and thus allows the transition to normoxia around mid-gestation stage^42–43. Eutherian mammals have evolved to exploit this increase in available oxygen during early post-implantation development, using it as a cue to alter the timing of hindlimb formation relative to that seen in bird embryos. In particular, in the chick embryo Pitx1 and Isl1 activate Tbx4 expression within 2-hour of their expression. But in the mouse embryo, Tbx4 (but not its functional forelimb equivalent Tbx5) has been brought under the strong repressive influence of the NFKB family members cRel and Rela, which, we speculate, have been co-opted for this purpose by being up-regulated under hypoxic conditions. As a consequence, although Pitx1 and Isl1 expression is activated in the hindlimb field at an equivalent developmental time as observed in the chick, they are unable to induce Tbx4 expression. This block in Tbx4 transcription lasts for 18–22 hours, until oxygen levels rise and hence NFKB activity is diminished. The temporal correlation in the expression of Tbx5, Pitx1,and Isl1 in both chicks and mice, and a similar correlation with Tbx4 in chicks, but not in the hindlimb of mice, suggests that the difference in relative timing of limb bud initiation between mammals and birds derives from a hindlimb delay, rather than forelimb acceleration. This evolutionary scenario assumes that the oxygen-dependent regulation of Tbx4 was not an ancestral condition. If it were, then the mechanism must have been lost during avian evolution.

Our results show that limb heterochrony in mammals (mouse) starts at the very first step of limb development, when Tbx4 induces EMT. This contrasts with the delay in forelimb (wing) development in emu embryos, in which the reduced forelimb growth rate relates not to the limb initiation, but rather to changes in post-initiation cell proliferation³³. In this case, Tbx5 expression occurs on time, but the expression of Fgf10, a key regulator of limb progenitor proliferation is reduced. This distinct mechanism is required in emus since their embryos, like chicks, lack an early hypoxic phase, and highlights the diverse levels of the limb developmental program that can be exploited by evolution to alter the tempo of limb outgrowth.

The delay in hindlimb development is even more pronounced in marsupial embryos, where it persists for longer period of time compared to placental mammals. As a result, the precociously born fetus displays an anatomically matured forelimb, capable of functioning in climbing to the pouch, but a still underdeveloped hindlimb³. Both forelimb acceleration and hindlimb delay have been posited as being responsible for limb heterochrony in marsupials^52–54. In this context, it is worth noting that, in spite of being much more extreme at later stages, the severity of marsupial limb heterochrony is similar to that of the mouse embryo at the limb initiation stage⁵⁵. Moreover, in the grey short-tailed opossum (Monodelphis domestica), both Tbx5 in the forelimb and Pitx1 in the hindlimb are detectable at least half a day earlier than Tbx4⁵², as is the case in mice. These results at least raise the possibility that the mechanism we identify here as underlying eutherian hindlimb heterochrony may also be utilized in marsupial embryos as one of the adaptive changes contributing to their more severe hindlimb developmental delay. If so, then limb heterochrony in placental mammals might be an evolutionary vestige from the non-placental common ancestor of both marsupials and eutherian mammals, rather than having arisen de novo within the placental stem lineage.

This study focused exclusively on differences in the timing of limb development among amniotes (mammals and birds). The extent to which these mechanisms apply to more evolutionarily distant tetrapods remains unclear. At minimum, the hypoxia-based pathway is not the only mechanism employed. For example, the early growth of the hindlimb in amphibians has been shown to be regulated by metamorphosis-associated hormonal signaling, a mechanism that is not conserved in mice⁵⁶.

One of the most intriguing findings of our study is the role of hypoxia in regulating the timing of hindlimb initiation. The transition from oviparity to viviparity dramatically alters the developmental environment from avian to mammalian species, with significant differences in oxygen levels as well as altering the level of accessibility of nutrients during the early embryonic development. As both parameters increase by placenta maturation, the level of oxygen can be viewed as a sensor of the level of nutrient availability. Mammalian embryos may have evolved to take advantage of this sensor to control the timing of development of less immediately required structures, such as the hindlimb, in an “energy trade-off strategy” to cope with limited nutrient availability during the earliest stages of eutherian development.

When ectopically expressed, both REL and RELA blocked the expression of Tbx4 and consequently the hindlimb bud formation. It is known that REL can competitively block the binding of RELA to the chromatin in other contexts⁵⁷. Therefore, although hypoxia does not significantly alter the expression of Rela, it is likely that Rela compensates for cRel activity only when cRel is depleted, explaining the weaker phenotype seen in cRel mutant embryos relative to that seen in embryos cultured in hypoxic conditions. In addition to cRel, we identified Hif1a as another hypoxia-responsive factor, having a repressive effect on Tbx4 expression under hypoxic conditions. Our functional analyses indicate that cRel can promote Hif1a expression, at least to a modest extent, consistent with previous reports in other systems. Due to the technical limitations of combining ex-utero culture with transgenic mouse lines, such as cRel knockouts, our dataset lacks the statistical power to fully resolve the regulatory relationships among cRel, Hif1a, and Tbx4. Nevertheless, our results point to a complex interplay of hypoxia-responsive factors in regulating the timing of limb development.

Besides the activation of NFKB transcription factors described here, hypoxia can influence the activities of multiple pathways including protein synthesis rate, oxidative phosphorylation levels, and mitochondrial biosynthesis—factors known to regulate developmental timing in various contexts.

Overall, our studies provide a comprehensive investigation of the molecular mechanisms regulating the observed heterochrony in limb initiation, an important embryonic adaptation to in-utero mammalian development. The mechanism we elucidate represents an unexpected use of hypoxia as a trigger for temporal events in the embryo. In a broader sense, this provides additional insight into how the maternal environment regulates developmental timing.

Limitations of the study

Because metabolic inhibitors can affect multiple aspects of basal cellular activity, conclusions regarding the direct role of metabolism in regulating Tbx4 expression and hindlimb development should be interpreted with caution.

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact Clifford Tabin (tabin@genetics.med.harvard.edu)

Materials availability

Plasmids and Mouse Embryonic Stem cell lines by this work are available upon request to the lead contact.

Data and code availability

The bulk RNA-sequencing data, as well as the single cell RNA-sequencing data has been deposited to NCBI GEO database (GSE298623 for bulk RNA seq; GSE298624 for single cell RNA-seq). Original images for figures are deposited to figshare (https://figshare.com/account/home#/projects/251516; DOI: 10.6084/m9.figshare.29287259). All other raw data is available upon reasonable request to the corresponding author. All processing code for the scRNA-seq analysis are available at https://github.com/chlee-tabin/limb-heterochrony and archived at Zenodo (DOI: 10.5281/zenodo.22074521). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR Methods

Experimental model and study participant details

Animals

Mice with various backgrounds were maintained and manipulated under the regulations of IACUC. Maximum 5 mice with the same sex were grouped in a cage. Wildtype lines, including CD1, C57BL/6J and F1 (mice are purchased from Charles River. cRel mutant line was imported from Stephen Smale lab (UCLA) and was described before⁷⁰. Tbx4 knockout line was purchased from Jackson Laboratory and was described previously⁹. Fertilized chicken eggs were ordered from Charles River farm and incubated at 38°C with 40% humidity.

Cell lines

Mouse embryonic stem cells (E14, or Hif1a mutant (generation procedure described in “Generation of Hif1a knockout ES cells” section) lines) were cultured in Fetal bovine serum (FBS)+Knockout 2iLif medium (DMEM (high glucose, 11955092), supplemented by 7.5% FBS (Thermofisher, A5669701) 7.5% Knockout Serum, Thermofisher, 10828028), 2mM GlutaMAX (Thermofisher, 35050061) 1mM Sodium Pyruvate (Thermofisher, 11360070), 0.1mM MEM NEAA (Thermofisher, 11140050), 0.1mM 2-mercaptoethanol (Thermofisher, 21985023), 10³ IU Human LIF recombinant protein (STEMCELL technologies, Cat No. 778855.1), 1uM PD0325901 (STEMCELL technologies, Cat No.72182), 3uM CHIR99021 (STEMCELL technologies, Cat No. 72052), 1x penicillin-streptomycin (Biological Industries, 030311B). The cells were cultured in 6-well or 12-well plates under 21% O₂, 5% CO₂, 37°C. The cell lines were not genotyped for sex.

Method details

Hybridization chain reaction (HCR)

Chicken embryos at HH8-HH23 stages and mouse embryos at E7.5-E10.0 stages were fixed in 4% PFA for 2h in room temperature or overnight at 4°C. Embryos were washed in PBST(PBS+0.1%Tween) for three times with 5mins each, and permeabilized in 70% ethanol in PBST for 2–5h at room temperature. The embryos were then washed by 37°C pre-warmed washing buffer (Molecular Instruments co.) for 5mins and pre-hybridized in pre-warmed hybridization buffer (Molecular Instruments co.) for 30mins in 38°C and then hybridized with hybridization buffer containing 60nM probes of targeted genes in 38°C overnight. On the next day, the embryos were washed in pre-warmed washing buffer for two times with 30mins each time, followed by two times washes in 5xSSCT (5xSSC with 0.1% Tween) with 30mins each time. Embryos were pre-amplified in amplification buffer (Molecular Instruments co.) for 5mins at room temperature and then amplified in amplification buffer containing 48nM amplifiers matching the amplification sequences of each probe set. Before amplification, the amplifiers were heated in 95°C for 90s and cooled down at room temperature for 30mins. The probes were designed following previous instructions.

Quantification of HCR signal

Transverse sections were collected across the entire limb bud region, encompassing both forelimb and hindlimb fields. From each sample, 2–3 representative sections were selected for imaging. Within each image, multiple 300 × 300 pixel regions were defined, and nuclei were manually counted to estimate cell numbers. Transcript quantification was performed using the ImageM MATLAB plugin, which detects and counts individual HCR puncta corresponding to transcript signals. From these measurements, we calculated the average number of transcripts per gene per cell at each developmental stage. For electroporation analyses, regions of interest (ROIs) were defined based on ZsGreen or mCherry fluorescence to identify electroporated areas in HCR images. Transcript numbers per cell were then quantified by manually examining electroporated cells throughout the Z-stack.

Chicken embryo electroporation and Early Chick (EC) culture

For HH13 lateral plate mesoderm electroporation, the chicken embryos were incubated for 50 hours, after which 5 ml albumin was removed from each egg and then windowed on the top. The electroporation was performed as previously described⁷¹. Briefly, plasmids with 1μg/ul mixed with FastGreen dye (0.2%) was injected into the coelom between two layers of lateral plate mesoderm of the forelimb or hindlimb and electroporated using BTX platinum needle L-shaped electrodes (45–0509). The electroporation was performed at 35 V, 5-ms pulse, 10-ms interval and 3–5 pulses. For HH4 chicken embryo electroporation, the eggs were cracked in 50-mm petri dishes and the albumin on the top of the embryo was gently removed using the blunt end of the forceps. The embryos were then transferred using pored filter papers (Whatman No.2) and was electroporated using Petri Dish Platinum Electrode using the following setting: 4 V, 50-ms pulse, 100-ms interval and 5 pulses.

The embryos were then transferred to 35-mm petri dish coated with albumin+0.6% bacto agar supplemented with 15 mg glucose and 121 mM sodium chloride (NaCl). The embryos were then cultured in 38 °C for 24 hours.

Immunostaining

The chicken or mouse embryos were first fixed in 4% PFA for 2h in room temperature or overnight at 4°C. The embryos were then dehydrated in 30% sucrose and embedded in 50% OCT in 15% sucrose solution and then processed with cryostat sectioning. The OCT in sections were washed by PBS for two times and then blocked by TNB solution (150 mM NaCl, 100 mM Tris-HCl (pH=7.5), 0.3% Triton X-100, 0.5% blocking reagent (Roche, 11096176001)) for 30mins at room temperature, and then incubated with TNB solutions containing primary antibodies in 4C for overnight. For Hif1a staining, primary antibody was diluted in PBS. The slides were then washed for 3×5mins in PBS and incubated with 2^nd antibodies for 1h at room temperature. After 2^nd antibody incubation, PBS or PBS containing DAPI and F-actin (diluted in 1:1000) were incubated for 5mins, followed by PBS wash for 5mins and mounted with Invitrogen^™ Fluoromount-G^™ Mounting Medium covered by cover slides. Dilutions for primary antibodies: Rabbit Anti-Laminin polyclonal antibody (Sigma-Aldrich, L9393), 1:500. Rabbit anti-HIF1a Polyclonal Antibody (Proteintech, 20960–1-AP), 1:100. Rabbit anti-phospho-Smad1 (Ser463/465)/Smad5 (Ser463/465)/Smad9 (Ser465/467) (D5B10) monoclonal antibody (Cell Signaling Technology, 13820S), 1:200. Secondary antibodies are diluted in TNB at 1:500 dilutions. List of antibodies include: donkey anti-Rabbit 488 (Jackson ImmunoResearch, 711-545-152), donkey anti-Rabbit 568(Invitrogen, A10042), donkey anti-Rabbit 647 (Jackson ImmunoResearch, 711-605-152).

Plasmids construction

Enhancer:reporter constructs: for constructing enhancer:reporter plasmids, pTK-RetE1-Citrine construct (a gift from Tatjana Sauka-Spengler (Addgene plasmid # 127779)) was double digested by KpnII (NEB, #R3142) and XmaI (NEB, R0180S) enzymes to create pTK-Citrine. Enhancer candidate regions were PCR amplified from chicken genomic DNA, with 5’ and 3’ primers carrying 18–22nt homologous sequences overlapping with the upstream and downstream of KpnI and XmaI sites of pTK--Citrine construct. The enhancer candidate regions were then cloned into pTK-Citrine vector via Gibson assembly. Positive clones were selected by both enzyme digestion and DNA sanger sequencing. The plasmids were maxi-preped before electroporation. Primers used for cloning are listed in Table S2.

Overexpression construct: pCAGGS-T2A-IRES-ZsGreen construct was double digested with XhoI (NEB, R1046) and EcoRI (NEB, R3101) enzymes. The protein coding sequence of cREL were amplified from chicken HH23 forelimb and hindlimb cDNA. 5’ and 3’ primers carrying 18–22nt homologous sequences overlapping with the upstream and downstream of XhoI and EcoRI sites of pCAGGS-T2A-IRES-ZsGreen construct. The coding sequences of genes of interest were then cloned into the construct via Gibson assembly. Positive clones were selected by both enzyme digestion and DNA sanger sequencing. The plasmids were maxi-preped before electroporation. The primer sequences were listed in Table S2.

Ex-utero culture and pharmacological treatments

E6.5 embryos were recovered from CD1 females mated with either CD1 or CB6F1 males, and were cultured in the medium containing 50% heat-inactivated rat serum (Biomed), 5% Fetal Bovine Serum 1× glutamax (GIBCO, 35050061), 100 units/ml penicillin and 100 μg/ml streptomycin (Biological Industries; 030311B) and 11mM HEPES (GIBCO, 15630056). Up to 3 embryos were placed in 12-well plate containing 2 ml medium each well. For each experiment, the embryos from the same batch were split into two groups and cultured for up to 2 days in 5% or 20% O₂ levels. All pharmacological inhibitors or activators were added directly into the ex-utero culture medium. 2-DG was treated at 0.1mM, Dichloroacetate (DCA) was treated at 0.15%.

For JC-1 staining, the hindlimb field from ex-utero cultured embryos were dissected manually and disassociated as single cells and run through flow cytometry. The Green (FL-2) to Red (FL-3) ratios were counted and compared.

Real-time quantitative PCR (RT-qPCR)

RNA was extracted from dissected hindlimb field of ex-utero cultured embryos from E6.5+1.5 days, under 5% or 20% O₂, using the Arcturus PicoPure RNA isolation kit (Arcturus Bioscience). RT-PCR was performed using a StepOne Plus Real-time PCR machine (BioRad). The expression level was calculated using ddCT methods, normalized to a Atp5f2 PCR reaction and the endogenous control group. The primers used for RT-PCR are displayed as Table S2.

For data analysis, ΔΔCt method was used to normalize the transcripts of inquiry’s expression to a control condition.

Making transgenic embryonic stem cells (ES)

Wild type E14 embryonic stem cells were transfected with two constructs, that one codes super-piggybac transposases (NovoPro, V012800), and one contains piggybac LTRs enclosing three cassette: a) enhancer:reporter cassette that contains enhancer:Hsv-tk:Citrine; b) CMV:H2B-mCherry; 3) Eif1a: puroR. Two days after transfection, the positive clones were first selected by puromycin treatment at 1ug/ml. After a week, the cells were double selected by manually picking the single cell formed colonies that do not show obvious citrine signal and detectable mCherry signal. Multiple clones were mixed and cultured together.

Making ES-embryo chimera

For making the chimera, F1 females (CB6F1, Charles River) were superovulated by PMSG injection on Day 1 and HCG injection on Day 3, and mated with F1 males to produce the embryos. The embryos were recovered at E1.5 and cultured until E2.5 in KSOM medium, on which day the zona pellucida was removed by acid Tyrode’s solution. The transgenic ES cells were trypsinized into single cells, and clamps with about 20 cells were aggregated with E2.5 mouse embryos before compaction. The chimeric embryos were then cultured in KSOM until E3.5, after which the chimeric embryos were transferred into the uterus of pseudo-pregnant CD1 females who were mated with sterilized CD1 males three days before the experiment. 20 chimeric embryos were transferred into each pseudo-pregnant female, and each experiment averagely 4 females were used for transfer.

Generation of Hif1a knockout ES cells

The region of depletion for Hif1a locus (containing a start coding site of Hif1a) was designed aas a previous study⁵⁰ (to keep the consistency, chromosome 12: 73973312 to 73973522). Four gRNAs targeting sequences upstream and downstream of the deletion region were designed using the CHOPCHOP website and transfected, together with Cas9 protein, using Lipofectamine CRISPRMAX Cas9 Transfection Reagent (Life Technologies Corporation, CMAX00001). Transfected ES cells were cultured for 48 h and then dissociated into single cells. Individual cells were isolated by mouth pipetting into 96-well plates for clonal expansion.

Once colonies reached approximately 70% confluency, genomic DNA was extracted for genotyping using primers flanking the deleted region: forward, 5′-ctttgttcatgttccatacaagtcc-3′; reverse, 5′-ccttcataagagcatattaatatgtgcac-3′. Homozygous deletions yielded a single band of ~336–364 bp, whereas wild-type and heterozygous clones produced an additional band at 502 bp. Homozygous knockout ES cell clones were expanded in 6-well plates prior to aggregation with CD1 host 8-cell stage embryos for chimera generation. Deletion of Hif1a was further confirmed by immunostaining for Hif1a protein. gRNA sequences are shown in Table S2.

Bulk RNA-sequencing and analyses

Tissues corresponding to the area of forelimb and hindlimb of chicken embryos at HH12, and mouse embryo at E8.5 and E9.5 were finely dissected and disassociated as single cell in TrypLE buffer. The RNA from each tissue was extracted using RNeasy Kit (Qiagen). 1ng of RNA was used for smart-seq v4 RNA sequencing. RNA was sequenced in pair-end. For the analyses, RNA sequencing reads were mapped to chicken genome Galgal6, and mouse genome mm10 by hisat2. FPKM (Fragments Per Kilobase per Million mapped reads) of genes were calculated by featureCounts, differentially expressed genes were calculated by DESeq2.

ATAC-seq identification of candidate CRE loci

The raw data deposited on GEO (GSE136775), as reported on Young et al (Curr Biol 2019 Nov 4;29(21):3681–3691.e5), was downloaded. After cleaning up and QC’ing the FASTQ files with TrimGalore (Krueger, 2015) and aligned them to the galGal6a (NCBI) genome using BWA (Li & Durbin, 2009), peaks were called with MACS2 (Zhang et al, 2008) using parameters recommended for ATAC-seq and genome coverage was determined using DeepTools (Ramirez et al, 2016). Differential accessibility between hindlimb, forelimb, and flank were calculated using the BDGDIFF function within MACS2 (signal cutoff: 10²).

Single cell RNA sequencing & analysis

As described above, ex-utero cultured embryos in hypoxia (2–5% O2) or normoxia (21% O2) for two days were dissociated in TrypLE at 37 degrees for 10 minutes with pipetting every 5 minutes. The dissociated cells were mixed with recovery medium (10% FBS in low glucose DMEM) and spun down for 5 minutes at 40 rcf, followed by one time wash in PBS, and ultimately placed in PBS with DRAQ5 (ThermoScientific, Cat # 62251, 10 uM) and DAPI (1 ug/mL), and placed on ice. The samples were subject to On-chip Sort microfluidic chip sorter (On-chip Biotechnologies) to enrich for viable cells (DRAQ5+DAPI−). During the sort, it was ensured that the cells were in ice cold condition. The resulting sorted suspensions in 1X PBS and 1% BSA were put into separate lanes for GEM generation according to the 10X Chromium GEM-X Single Cell (Dual Index) protocol. To minimize further any sources of batch effect and have an internal control for genotype demultiplexing, an additional lane was run with the hypoxic and normoxic condition samples combined. Subsequent post GEM-RT cleanup, cDNA amplification, and 3’ gene expression library construction steps were performed in accordance with the manufacturer’s instructions. Sample indices were chosen from the 10X Dual Index Kit TT Set A (PN-1000215). Sequencing was performed on a NovaSeq X flowcell at the Biopolymers Facility at Harvard Medical School according to the recommended 10X sequencing specification.

Generation of the count matrix (cellranger v9.0.0), SNP detection (cellsnp-lite v1.2.3), and genotype-based demultiplexing (vireoSNP v0.5.8) were all performed on the Harvard Medical School O2 computing cluster. To facilitate the computational time involved in genotype inference, all cells from the three libraries were randomly partitioned into four sets of ~40,000 cells that overlap 50% overlap with another partition, and vireo was run with 40 randomly generated seeds to find the best log-likelihood fit. The number of expected genotypes were set to 24 (N=24) to obtain inferred genotypes. Given all cells were at least two times inferred from separate partitions, the genotype for each partitions could be stitched bck for the whole dataset. Additional filtering was conducted to identify the cellular barcodes that conform to the expected pattern of exclusion for a specific library (presence exclusively in hypoxic condition library or normoxic condition library, with common presence in the shared library). These final genotype inferences were used to reconstruct 9 embryos from hypoxic conditions and 10 embryos for normoxic conditions. Ambiguous cellular barcodes (genotype inferences that were found in all three libraries, potential doublets that “cross-over” between genotypes in two overlapping partitions) were discarded. After generation of Seurat objects, downstream analysis using the Seurat package was performed in R⁶⁵ (Hao et al., 2023; Hao et al., 2021; Stuart et al., 2019; Butler et al., 2018; Satija et al., 2015).

The embryonic gastrulation atlas Imaz-Rosshandler et al.⁷² has been found to have some mis-annotations of the mesenchymal lineage (e.g. Cardiomyocyte FHF and SHF are swapped, based on the marker expressions). As such, the data was subsampled to contain E7.75 to E9.25 samples and were re-processed for batch-correction (harmony reference) and graph-based clustering using the Leiden algorithm (resolutions 0.2–2.0, consolidated by marker-based curation) that retains most of the mapping from the original annotation but fixed the paraxial, lateral plate mesoderm lineage and incorporated proper separation of the limb fields into forelimb and hindlimb clusters. The resulting reference annotation was used for LabelTransfer procedure in Seurat (v5) to label cells of the ex-utero dataset. The number of inferred cell state annotations were normalized against the number of deconvoluted genotype (embryo), and this relative fraction was compared between the two conditions. The reference dataset for stage inference Imaz-Rosshandler et al.⁷² was subsampled by random selection to equal stage representation across E7.75 to E9.25 prior to anchor-finding (Seurat::FindTransferAnchors, reference.reduction = “pca”, dims = 1:50).

Stage inference used Seurat::TransferData with k.weight = 50 (the Seurat v5 default) for the primary analysis; per-tissue significance was assessed by 1,000-iteration embryo-level permutation of the (embryo to condition) mapping, with the mean of pairwise (HYPO-embryo, NORM-embryo) Kolmogorov-Smirnov D statistics over the predicted-stage distributions as the test statistic. To verify that the choice of k.weight = 50 does not drive the inferred shift, the analysis was replicated at k.weight = 20, 50, 100; per-tissue p-values are reported in Figure S7F. Gene expression visualization was conducted by pseudobulking by embryo or embryo & predicted cell state. Two-sided Wilcoxon rank-sum tests with Benjamini Hochberg multiple testing adjustment were used for the per-embryo pseudobulk differential expression shown in the panels; parametric t-test values are provided in the source data. The size of the pseudobulk was only considered for filtering too small pseudobulk sets, given the average level of the gene in question (high expressing genes for Figure S7C, >100 cells; mid-expressing genes for S7D, >400 cells; for low-expressing genes for S9A, >2000 cells). The stage distribution analysis, the default parameters for ‘geom_density’ function in `ggplot2` package was used to display individual embryo and whole condition distributions. Kolmogorov-Smirnov test was performed on the stage distributions between hypoxic and normoxic conditions for difference.

Quantification and Statistical Analysis

Statistical methods are indicated for every experiment in the corresponding Figure legends. Statistics for sequencing datasets are indicated in above sections. Qualitative data is presented as a contingency table and was analyzed using Fisher’s exact test. Normality of quantitative data was first analyzed using D’Agostino’s K-squared test. A one-sample t-test was used to test whether an observed distribution followed a hypothetical mean. If data showed a normal distribution, then for comparison of two or multiple samples, an unpaired two-tailed Student’s t test (two experimental groups) or a One-way ANOVA test (more than two experimental groups) was used to analyze statistical significance. Differences in variances were taken into account by performing a Welch’s correction. For data that did not present a normal distribution, a Mann–Whitney U-test (two experimental groups) or a Kruskal–Wallis test with a Dunn’s multiple comparison test (more than two experimental groups) was used to test statistical significance. To determine the influence of different groups in multiple variants, two-way ANOVA was performed. Statistical analyses were performed using Prism software (http://www.graphpad.com)

Supplementary Material

Table S1. List of putative Chicken Tbx4 regulators that were functionally tested.

Table S2. List of DNA oligo sequences and plasmids used in this work.

Table S3. Summary of ex-utero culture experiments.

NIHMS2212997-supplement-1.pdf^{(26.3MB, pdf)}

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
HIF1A Polyclonal Antibody	Proteintech	20960-1-AP
Rabbit anti-phospho-Smad1 (Ser463/465)/Smad5 (Ser463/465)/Smad9 (Ser465/467) (D5B10) monoclonal antibody	Cell Signaling Technology	13820S
Rabbit Anti-Laminin polyclonal antibody	Sigma-Aldrich	L9393
donkey anti-Rabbit 488	Jackson ImmunoResearch	711-545-152
donkey anti-Rabbit 568	Invitrogen	A10042
donkey anti-Rabbit 647	Jackson ImmunoResearch	711-605-152
Bacterial and virus strains
DH-5a	New England Biolabs	C2987I
Chemicals, peptides, and recombinant proteins
Paraformaldehyde	Electron Microscopy Sciences	15170
Phosphate buffered saline (PBS)	Thermofisher	10010023
Ethanol	Sigma-Aldrich	459836-500mL
Wash Buffer	Molecular Instruments	Whole-mount 100ml
Hybridization buffer	Molecular Instruments	Whole-mount 50ml
Amplification buffer	Molecular Instruments	Whole-mount 50ml
B1, B2, B3 amplifiers, wavelength: 488, 546,594, 647	Molecular Instruments	50ul
UltraPure^™ SSC, 20X	Thermofisher	15557044
FastGreen FCF	Thermofisher	A16520.14, 25g
BTX platinum needle L-shaped electrodes	BTX	45-0509
OCT	Leica microsystems	FSC22
Blocking reagent	Millipore sigma	11096176001
super-piggybac transposases	NovoPro	V012800
HEPES	GIBCO	15630056
penicillin-streptomycin	Biological Industries	030311B
glutamax	GIBCO	35050061
Dichloroacetate	Sigma Aldrich	347795-10G
DMEM	Thermofisher	11955092
2-mercaptoethanol	Thermofisher	21985023
Fetal Bovine Serum	Thermofisher	A5669701
KnockOut serum	Thermofisher	10828028
Sodium Pyruvate	Thermofisher	11360070
MEM NEAA	Thermofisher	11140050
LIF recombinant protein	STEMCELL Technologies	708855.1
PD0325901	STEMCELL Technologies	72182
CHIR99021	STEMCELL Technologies	72052
KpnII	new england biolabs	R3142
XmaI	new england biolabs	R0180S
XhoI	new england biolabs	R1046
EcoRI	new england biolabs	R3101
Critical commercial assays
JC-1 Dye (Mitochondrial Membrane Potential Probe)	Thermofisher	T3168
PicoPure RNA isolation kit	Arcturus Bioscience	KIT0204
Deposited data
ATAC-seq data	Young et al., 2019³³	GEO (GSE136775)
Bulk RNA-seq data	New, this work	GEO (GSE298623)
Single cell RNA-seq	New, this work	GEO (GSE298624)
Experimental models: Cell lines
E14 mouse embryonic stem cells	Gift from Olivier Pourquie lab	E14
Experimental models: Cell lines
Hif1a KO Mouse Embryonic stem cell line	This work	E14
Experimental models: Organisms/strains
Mouse female and male; Strain: CD1	Charles River	CD1
Mouse female and male; Strain: C57Bl/6	Charles River	C57Bl/6
Mouse female and male; Strain: B6D2F1	Charles River	B6D2F1
Mouse female and male; Strain: B6C3F1	Charles River	B6C3F1
Transgenic mouse strain: Tbx4 mutant	The Jackson laboratory	Tbx4^tm1.1Pa/Mmjax
Transgenic mouse strain: cRel mutant	Gift from Stephen Smale lab (UCLA) which was originally created and maintained in Steven Gerondakis’s lab (Monash University).	cRel−/−
Oligonucleotides
Please see Table S2	This work	Table S2
Recombinant DNA
Please see Table S2 - Plasmids section	This work	Table S2 - Plasmids
Software and algorithms
Prism	Graphpad	Prism http://graphpad.com/
Rstudio	R	Rstudio https://posit.co/
FIJI	PBC	ImageJ https://imagej.net/ij/
FlowJo	BD biosciences	FlowJo www.flowjo.com
bcl-convert	Illumina	v4.2.7 https://support.illumina.com/sequencing/sequencing_software/bcl-convert.html
Cell Ranger	10x Genomics	(v9.0.0) https://www.10xgenomics.com/support/software/cell-ranger/latest
SAMtools	Danecek et al., 2021⁵⁸	SAMtools https://www.htslib.org/
cellSNP-lite	Huang & Huang, 2021⁵⁹	(v1.2.3) https://github.com/single-cell-genetics/cellsnp-lite
Vireo / vireoSNP	Huang et al., 2019⁶⁰	(v0.5.9) https://github.com/single-cell-genetics/vireo
scDblFinder	Germain et al., 2021⁶¹	(v1.20.2) https://bioconductor.org/packages/scDblFinder
scran	Lun et al., 2016⁶²	(v1.34.0) https://bioconductor.org/packages/scran
scater	McCarthy et al., 2017⁶³	(v1.34.1) https://bioconductor.org/packages/release/bioc/html/scater.html
SingleCellExperiment	Amezquita et al., 2020⁶⁴	(v1.28.0) https://bioconductor.org/packages/SingleCellExperiment
Seurat	Hao et al., 2024⁶⁵	(v5.3.0 / v5.4.0) https://satijalab.org/seurat
SeuratObject	Hao et al., 2024⁶⁵	(v5.2.0 / v5.3.0) https://cran.r-project.org/package=SeuratObject
harmony	Korsunsky et al., 2019⁶⁶	(v1.2.3 / v1.2.4) https://github.com/immunogenomics/harmony
leidenalg	Traag et al., 2019⁶⁷	(v0.10.1) https://github.com/vtraag/leidenalg
reticulate	Rdrr.io	(v1.43.0 / v1.45.0) https://rstudio.github.io/reticulate
cellxgene-census	CZI CELLxGENE Discover	(v1.17.0) https://chanzuckerberg.github.io/cellxgene-census/
tiledbsoma	TileDB / CZI	(v2.3.0) https://github.com/single-cell-data/TileDB-SOMA
anndata	JOSS	Anndata https://anndata.readthedocs.io/
tidyverse	Wickham et al., 2019⁶⁸	https://www.tidyverse.org/
R figure packages: patchwork ggrepel, ggrastr, ggbeeswarm, svglite, MASS	CRAN	patchwork ggrepel, ggrastr, ggbeeswarm, svglite, MASS https://cran.r-project.org/
R	R Core Team	(v4.4.3 / v4.4.2) https://www.r-project.org/
Python	Python Software Foundation	(v3.10 / v3.11) https://www.python.org/
Python packages: NumPy, Pandas, SciPy	Virtanen 2020⁶⁹	NumPy, Pandas, SciPy https://numpy.org/
Custom analysis & figure code	This work	https://github.com/chlee-tabin/limb-heterochrony; DOI: 10.5281/zenodo.22074521

Open in a new tab

Highlights:

Avian embryos form their forelimbs and hindlimbs simultaneously.
However, mammalian embryos initiate development of the hindlimb after the forelimb.
This delay may help cope with limited energy availability prior to placenta formation.
Oxygen levels, also rising after placentation, control timing of hindlimb development.

Acknowledgment

We thank all members in Cepko-Tabin labs and Susan Dymecki lab, and Dr. Emma Farley, Dr. Scott Edwards for giving constructive comments on the project. We thank Dr. Malcolm Logan and Dr. Ben Steventon for helpful discussions at the initial phase of the project. We also thank Dr. Nandan Nerurkar and Dr. Marianne Bronner for the training of early chick culture experiments, Dr. Lina Du from mouse engineering core at Dana-Farber with the help of embryo transfer, and Micron imaging core especially Dr. Paula Montero Llopis for imaging assistance. The cRel mutant line was imported from Stephen Smale (UCLA), which was originally created and maintained in Steven Gerondakis’s lab (Monash University). Work on this topic is supported by a grant from the NIH, HD032443 to C.J.T, and a Human Frontier Science program long-term post-doctoral fellowship LT000676/2020-L to M.Z.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests

The authors declare no competing interests.

Material availability

All materials are available through reasonable requests to the corresponding author.

References

1.Richardson MK, Gobes SM, van Leeuwen AC, Polman JA, Pieau C, and Sanchez-Villagra MR. (2009). Heterochrony in limb evolution: developmental mechanisms and natural selection. J Exp Zool B Mol Dev Evol 312, 639–664. 10.1002/jez.b.21250. [DOI] [PubMed] [Google Scholar]
2.Weisbecker V, Goswami A, Wroe S, and Sanchez-Villagra MR. (2008). Ossification heterochrony in the therian postcranial skeleton and the marsupial-placental dichotomy. Evolution 62, 2027–2041. 10.1111/j.1558-5646.2008.00424.x. [DOI] [PubMed] [Google Scholar]
3.Keyte AL, and Smith KK. (2014). Heterochrony and developmental timing mechanisms: changing ontogenies in evolution. Semin Cell Dev Biol 34, 99–107. 10.1016/j.semcdb.2014.06.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Gros J, and Tabin CJ. (2014). Vertebrate limb bud formation is initiated by localized epithelial-to-mesenchymal transition. Science 343, 1253–1256. 10.1126/science.1248228. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Sekine K, Ohuchi H, Fujiwara M, Yamasaki M, Yoshizawa T, Sato T, Yagishita N, Matsui D, Koga Y, Itoh N, and Kato S. (1999). Fgf10 is essential for limb and lung formation. Nat Genet 21, 138–141. 10.1038/5096. [DOI] [PubMed] [Google Scholar]
6.Takeuchi JK, Koshiba-Takeuchi K, Suzuki T, Kamimura M, Ogura K, and Ogura T. (2003). Tbx5 and Tbx4 trigger limb initiation through activation of the Wnt/Fgf signaling cascade. Development 130, 2729–2739. 10.1242/dev.00474. [DOI] [PubMed] [Google Scholar]
7.Barrow JR, Thomas KR, Boussadia-Zahui O, Moore R, Kemler R, Capecchi MR, and McMahon AP. (2003). Ectodermal Wnt3/beta-catenin signaling is required for the establishment and maintenance of the apical ectodermal ridge. Genes Dev 17, 394–409. 10.1101/gad.1044903. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Rallis C, Bruneau BG, Del Buono J, Seidman CE, Seidman JG, Nissim S, Tabin CJ, and Logan MP. (2003). Tbx5 is required for forelimb bud formation and continued outgrowth. Development 130, 2741–2751. 10.1242/dev.00473. [DOI] [PubMed] [Google Scholar]
9.Naiche LA, and Papaioannou VE. (2003). Loss of Tbx4 blocks hindlimb development and affects vascularization and fusion of the allantois. Development 130, 2681–2693. 10.1242/dev.00504. [DOI] [PubMed] [Google Scholar]
10.Kawakami Y, Capdevila J, Buscher D, Itoh T, Rodriguez Esteban C, and Izpisua Belmonte JC. (2001). WNT signals control FGF-dependent limb initiation and AER induction in the chick embryo. Cell 104, 891–900. 10.1016/s0092-8674(01)00285-9. [DOI] [PubMed] [Google Scholar]
11.Choi HMT, Schwarzkopf M, Fornace ME, Acharya A, Artavanis G, Stegmaier J, Cunha A, and Pierce NA. (2018). Third-generation in situ hybridization chain reaction: multiplexed, quantitative, sensitive, versatile, robust. Development 145. 10.1242/dev.165753. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Takeuchi JK, Koshiba-Takeuchi K, Matsumoto K, Vogel-Hopker A, Naitoh-Matsuo M, Ogura K, Takahashi N, Yasuda K, and Ogura T. (1999). Tbx5 and Tbx4 genes determine the wing/leg identity of limb buds. Nature 398, 810–814. 10.1038/19762. [DOI] [PubMed] [Google Scholar]
13.Ohuchi H, Nakagawa T, Yamamoto A, Araga A, Ohata T, Ishimaru Y, Yoshioka H, Kuwana T, Nohno T, Yamasaki M, et al. (1997). The mesenchymal factor, FGF10, initiates and maintains the outgrowth of the chick limb bud through interaction with FGF8, an apical ectodermal factor. Development 124, 2235–2244. 10.1242/dev.124.11.2235. [DOI] [PubMed] [Google Scholar]
14.Ng JK, Kawakami Y, Buscher D, Raya A, Itoh T, Koth CM, Rodriguez Esteban C, Rodriguez-Leon J, Garrity DM, Fishman MC, and Izpisua Belmonte JC. (2002). The limb identity gene Tbx5 promotes limb initiation by interacting with Wnt2b and Fgf10. Development 129, 5161–5170. 10.1242/dev.129.22.5161. [DOI] [PubMed] [Google Scholar]
15.Rodriguez-Esteban C, Tsukui T, Yonei S, Magallon J, Tamura K, and Izpisua Belmonte JC. (1999). The T-box genes Tbx4 and Tbx5 regulate limb outgrowth and identity. Nature 398, 814–818. 10.1038/19769. [DOI] [PubMed] [Google Scholar]
16.Duboule D. (1994). Temporal colinearity and the phylotypic progression: a basis for the stability of a vertebrate Bauplan and the evolution of morphologies through heterochrony. Dev Suppl, 135–142. [PubMed] [Google Scholar]
17.Minguillon C, Nishimoto S, Wood S, Vendrell E, Gibson-Brown JJ, and Logan MP. (2012). Hox genes regulate the onset of Tbx5 expression in the forelimb. Development 139, 3180–3188. 10.1242/dev.084814. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Moreau C, Caldarelli P, Rocancourt D, Roussel J, Denans N, Pourquie O, and Gros J. (2019). Timed Collinear Activation of Hox Genes during Gastrulation Controls the Avian Forelimb Position. Curr Biol 29, 35–50 e34. 10.1016/j.cub.2018.11.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Tanaka M, Jokubaitis V, Wood C, Wang Y, Brouard N, Pera M, Hearn M, Simmons P, and Nakayama N. (2009). BMP inhibition stimulates WNT-dependent generation of chondrogenic mesoderm from embryonic stem cells. Stem Cell Res 3, 126–141. 10.1016/j.scr.2009.07.001. [DOI] [PubMed] [Google Scholar]
20.Mori S, Sakakura E, Tsunekawa Y, Hagiwara M, Suzuki T, and Eiraku M. (2019). Self-organized formation of developing appendages from murine pluripotent stem cells. Nat Commun 10, 3802. 10.1038/s41467-019-11702-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Logan M, and Tabin CJ. (1999). Role of Pitx1 upstream of Tbx4 in specification of hindlimb identity. Science 283, 1736–1739. 10.1126/science.283.5408.1736. [DOI] [PubMed] [Google Scholar]
22.Duboc V, and Logan MP. (2011). Pitx1 is necessary for normal initiation of hindlimb outgrowth through regulation of Tbx4 expression and shapes hindlimb morphologies via targeted growth control. Development 138, 5301–5309. 10.1242/dev.074153. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Itou J, Kawakami H, Quach T, Osterwalder M, Evans SM, Zeller R, and Kawakami Y. (2012). Islet1 regulates establishment of the posterior hindlimb field upstream of the Hand2-Shh morphoregulatory gene network in mouse embryos. Development 139, 1620–1629. 10.1242/dev.073056. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Duboc V, Sulaiman FA, Feneck E, Kucharska A, Bell D, Holder-Espinasse M, and Logan MPO. (2021). Tbx4 function during hindlimb development reveals a mechanism that explains the origins of proximal limb defects. Development 148. 10.1242/dev.199580. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Szeto DP, Rodriguez-Esteban C, Ryan AK, O’Connell SM, Liu F, Kioussi C, Gleiberman AS, Izpisua-Belmonte JC, and Rosenfeld MG. (1999). Role of the Bicoid-related homeodomain factor Pitx1 in specifying hindlimb morphogenesis and pituitary development. Genes Dev 13, 484–494. 10.1101/gad.13.4.484. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Narkis G, Tzchori I, Cohen T, Holtz A, Wier E, and Westphal H. (2012). Isl1 and Ldb co-regulators of transcription are essential early determinants of mouse limb development. Dev Dyn 241, 787–791. 10.1002/dvdy.23761. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Ferrer-Vaquer A, Piliszek A, Tian G, Aho RJ, Dufort D, and Hadjantonakis AK. (2010). A sensitive and bright single-cell resolution live imaging reporter of Wnt/ss-catenin signaling in the mouse. BMC Dev Biol 10, 121. 10.1186/1471-213X-10-121. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Takemoto T, Abe T, Kiyonari H, Nakao K, Furuta Y, Suzuki H, Takada S, Fujimori T, and Kondoh H. (2016). R26-WntVis reporter mice showing graded response to Wnt signal levels. Genes Cells 21, 661–669. 10.1111/gtc.12364. [DOI] [PubMed] [Google Scholar]
29.Molotkov A, Molotkova N, and Duester G. (2005). Retinoic acid generated by Raldh2 in mesoderm is required for mouse dorsal endodermal pancreas development. Dev Dyn 232, 950–957. 10.1002/dvdy.20256. [DOI] [PubMed] [Google Scholar]
30.Cunningham TJ, Zhao X, Sandell LL, Evans SM, Trainor PA, and Duester G. (2013). Antagonism between retinoic acid and fibroblast growth factor signaling during limb development. Cell Rep 3, 1503–1511. 10.1016/j.celrep.2013.03.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Jeong S, Rebeiz M, Andolfatto P, Werner T, True J, and Carroll SB. (2008). The evolution of gene regulation underlies a morphological difference between two Drosophila sister species. Cell 132, 783–793. 10.1016/j.cell.2008.01.014. [DOI] [PubMed] [Google Scholar]
32.Menke DB, Guenther C, and Kingsley DM. (2008). Dual hindlimb control elements in the Tbx4 gene and region-specific control of bone size in vertebrate limbs. Development 135, 2543–2553. 10.1242/dev.017384. [DOI] [PubMed] [Google Scholar]
33.Young JJ, Grayson P, Edwards SV, and Tabin CJ. (2019). Attenuated Fgf Signaling Underlies the Forelimb Heterochrony in the Emu Dromaius novaehollandiae. Curr Biol 29, 3681–3691 e3685. 10.1016/j.cub.2019.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Alerasool N, Segal D, Lee H, and Taipale M. (2020). An efficient KRAB domain for CRISPRi applications in human cells. Nat Methods 17, 1093–1096. 10.1038/s41592-020-0966-x. [DOI] [PubMed] [Google Scholar]
35.Mic FA, Haselbeck RJ, Cuenca AE, and Duester G. (2002). Novel retinoic acid generating activities in the neural tube and heart identified by conditional rescue of Raldh2 null mutant mice. Development 129, 2271–2282. 10.1242/dev.129.9.2271. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Savory JG, Bouchard N, Pierre V, Rijli FM, De Repentigny Y, Kothary R, and Lohnes D. (2009). Cdx2 regulation of posterior development through non-Hox targets. Development 136, 4099–4110. 10.1242/dev.041582. [DOI] [PubMed] [Google Scholar]
37.Bell CC, Amaral PP, Kalsbeek A, Magor GW, Gillinder KR, Tangermann P, di Lisio L, Cheetham SW, Gruhl F, Frith J, et al. (2016). The Evx1/Evx1as gene locus regulates anterior-posterior patterning during gastrulation. Sci Rep 6, 26657. 10.1038/srep26657. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Qiu C, Martin BK, Welsh IC, Daza RM, Le TM, Huang X, Nichols EK, Taylor ML, Fulton O, O’Day DR, et al. (2024). A single-cell time-lapse of mouse prenatal development from gastrula to birth. Nature 626, 1084–1093. 10.1038/s41586-024-07069-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Yin K, and Juurlink BH. (2000). Regulation of expression of nuclear factor kappa B RelA in oligodendrocytes: effect of hypoxia. Neuroreport 11, 1877–1881. 10.1097/00001756-200006260-00015. [DOI] [PubMed] [Google Scholar]
40.Jiang Y, Zhu Y, Wang X, Gong J, Hu C, Guo B, Zhu B, and Li Y. (2015). Temporal regulation of HIF-1 and NF-kappaB in hypoxic hepatocarcinoma cells. Oncotarget 6, 9409–9419. 10.18632/oncotarget.3352. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Simon MC, and Keith B. (2008). The role of oxygen availability in embryonic development and stem cell function. Nat Rev Mol Cell Biol 9, 285–296. 10.1038/nrm2354. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Mitchell JA, and Yochim JM. (1968). Measurement of intrauterine oxygen tension in the rat and its regulation by ovarian steroid hormones. Endocrinology 83, 691–700. 10.1210/endo-83-4-691. [DOI] [PubMed] [Google Scholar]
43.Rodesch F, Simon P, Donner C, and Jauniaux E. (1992). Oxygen measurements in endometrial and trophoblastic tissues during early pregnancy. Obstet Gynecol 80, 283–285. [PubMed] [Google Scholar]
44.Lee YM, Jeong CH, Koo SY, Son MJ, Song HS, Bae SK, Raleigh JA, Chung HY, Yoo MA, and Kim KW. (2001). Determination of hypoxic region by hypoxia marker in developing mouse embryos in vivo: a possible signal for vessel development. Dev Dyn 220, 175–186. 10.1002/1097-0177(20010201)220:2<175::AID-DVDY1101>3.0.CO;2-F. [DOI] [PubMed] [Google Scholar]
45.Genbacev O, Zhou Y, Ludlow JW, and Fisher SJ. (1997). Regulation of human placental development by oxygen tension. Science 277, 1669–1672. 10.1126/science.277.5332.1669. [DOI] [PubMed] [Google Scholar]
46.Aguilera-Castrejon A, Oldak B, Shani T, Ghanem N, Itzkovich C, Slomovich S, Tarazi S, Bayerl J, Chugaeva V, Ayyash M, et al. (2021). Ex utero mouse embryogenesis from pre-gastrulation to late organogenesis. Nature 593, 119–124. 10.1038/s41586-021-03416-3. [DOI] [PubMed] [Google Scholar]
47.Culshaw LH, Savery D, Greene NDE, and Copp AJ. (2019). Mouse whole embryo culture: Evaluating the requirement for rat serum as culture medium. Birth Defects Res 111, 1165–1177. 10.1002/bdr2.1538. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Linde-Medina M, and Smit TH. (2021). Molecular and Mechanical Cues for Somite Periodicity. Front Cell Dev Biol 9, 753446. 10.3389/fcell.2021.753446. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Pijuan-Sala B, Griffiths JA, Guibentif C, Hiscock TW, Jawaid W, Calero-Nieto FJ, Mulas C, Ibarra-Soria X, Tyser RCV, Ho DLL, et al. (2019). A single-cell molecular map of mouse gastrulation and early organogenesis. Nature 566, 490–495. 10.1038/s41586-019-0933-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Iyer NV, Kotch LE, Agani F, Leung SW, Laughner E, Wenger RH, Gassmann M, Gearhart JD, Lawler AM, Yu AY, and Semenza GL. (1998). Cellular and developmental control of O2 homeostasis by hypoxia-inducible factor 1 alpha. Genes Dev 12, 149–162. 10.1101/gad.12.2.149. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Carroll C, Engström N, Nilsson PF, Haxen ER, Mohlin S, Berg P, Glud RN, and Hammarlund EU. (2021). Hypoxia Generated by Avian Embryo Growth Induces the HIF-α Response and Critical Vascularization. Frontiers in Ecology and Evolution 9. 10.3389/fevo.2021.675800. [DOI] [Google Scholar]
52.Keyte AL, and Smith KK. (2010). Developmental origins of precocial forelimbs in marsupial neonates. Development 137, 4283–4294. 10.1242/dev.049445. [DOI] [PubMed] [Google Scholar]
53.Harrison LB, and Larsson HC. (2008). Estimating evolution of temporal sequence changes: a practical approach to inferring ancestral developmental sequences and sequence heterochrony. Syst Biol 57, 378–387. 10.1080/10635150802164421. [DOI] [PubMed] [Google Scholar]
54.Lyne A. (1964). Observations on the breeding and growth of the marsupial Perameles nasuta Geoffroy, with notes on other bandicoots. Australian Journal of Zoology 12, 322–339. [Google Scholar]
55.Chew KY, Shaw G, Yu H, Pask AJ, and Renfree MB. (2014). Heterochrony in the regulation of the developing marsupial limb. Dev Dyn 243, 324–338. 10.1002/dvdy.24062. [DOI] [PubMed] [Google Scholar]
56.Brown DD, Cai L, Das B, Marsh-Armstrong N, Schreiber AM, and Juste R. (2005). Thyroid hormone controls multiple independent programs required for limb development in Xenopus laevis metamorphosis. Proc Natl Acad Sci U S A 102, 12455–12458. 10.1073/pnas.0505989102. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.de Jesus TJ, and Ramakrishnan P. (2020). NF-kappaB c-Rel Dictates the Inflammatory Threshold by Acting as a Transcriptional Repressor. iScience 23, 100876. 10.1016/j.isci.2020.100876. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, Whitwham A, Keane T, McCarthy SA, Davies RM, and Li H. (2021). Twelve years of SAMtools and BCFtools. Gigascience 10. 10.1093/gigascience/giab008. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Huang X, and Huang Y. (2021). Cellsnp-lite: an efficient tool for genotyping single cells. Bioinformatics 37, 4569–4571. 10.1093/bioinformatics/btab358. [DOI] [PubMed] [Google Scholar]
60.Huang Y, McCarthy DJ, and Stegle O. (2019). Vireo: Bayesian demultiplexing of pooled single-cell RNA-seq data without genotype reference. Genome Biol 20, 273. 10.1186/s13059-019-1865-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Germain PL, Lun A, Garcia Meixide C, Macnair W, and Robinson MD. (2021). Doublet identification in single-cell sequencing data using scDblFinder. F1000Res 10, 979. 10.12688/f1000research.73600.2. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Rubio-de Anda PF, Martinez-Orozco H, and Diaz Miranda SY. (2026). From Tissue to Morphometry: A Step-by-Step Golgi-Cox Workflow for Neuronal and Dendritic Spine Analysis in Adult Mice. J Vis Exp. 10.3791/71384. [DOI] [PubMed] [Google Scholar]
63.McCarthy DJ, Campbell KR, Lun AT, and Wills QF. (2017). Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics 33, 1179–1186. 10.1093/bioinformatics/btw777. [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Amezquita RA, Lun ATL, Becht E, Carey VJ, Carpp LN, Geistlinger L, Marini F, Rue-Albrecht K, Risso D, Soneson C, et al. (2020). Orchestrating single-cell analysis with Bioconductor. Nat Methods 17, 137–145. 10.1038/s41592-019-0654-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, and Satija R. (2024). Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304. 10.1038/s41587-023-01767-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh PR, and Raychaudhuri S. (2019). Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16, 1289–1296. 10.1038/s41592-019-0619-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Traag VA, Waltman L, and van Eck NJ. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Sci Rep 9, 5233. 10.1038/s41598-019-41695-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Wickham J, and Riitters KH. (2019). Influence of high-resolution data on the assessment of forest fragmentation. Landsc Ecol 34, 2169–2182. 10.1007/s10980-019-00820-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, Burovski E, Peterson P, Weckesser W, Bright J, et al. (2020). SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods 17, 261–272. 10.1038/s41592-019-0686-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Kontgen F, Grumont RJ, Strasser A, Metcalf D, Li R, Tarlinton D, and Gerondakis S. (1995). Mice lacking the c-rel proto-oncogene exhibit defects in lymphocyte proliferation, humoral immunity, and interleukin-2 expression. Genes Dev 9, 1965–1977. 10.1101/gad.9.16.1965. [DOI] [PubMed] [Google Scholar]
71.Nerurkar NL, Lee C, Mahadevan L, and Tabin CJ. (2019). Molecular control of macroscopic forces drives formation of the vertebrate hindgut. Nature 565, 480–484. 10.1038/s41586-018-0865-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Imaz-Rosshandler I, Rode C, Guibentif C, Harland LTG, Ton MN, Dhapola P, Keitley D, Argelaguet R, Calero-Nieto FJ, Nichols J, et al. (2024). Tracking early mammalian organogenesis - prediction and validation of differentiation trajectories at whole organism scale. Development 151. 10.1242/dev.201867. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1. List of putative Chicken Tbx4 regulators that were functionally tested.

Table S2. List of DNA oligo sequences and plasmids used in this work.

Table S3. Summary of ex-utero culture experiments.

NIHMS2212997-supplement-1.pdf^{(26.3MB, pdf)}

Data Availability Statement

All materials are available through reasonable requests to the corresponding author.

[R1] 1.Richardson MK, Gobes SM, van Leeuwen AC, Polman JA, Pieau C, and Sanchez-Villagra MR. (2009). Heterochrony in limb evolution: developmental mechanisms and natural selection. J Exp Zool B Mol Dev Evol 312, 639–664. 10.1002/jez.b.21250. [DOI] [PubMed] [Google Scholar]

[R2] 2.Weisbecker V, Goswami A, Wroe S, and Sanchez-Villagra MR. (2008). Ossification heterochrony in the therian postcranial skeleton and the marsupial-placental dichotomy. Evolution 62, 2027–2041. 10.1111/j.1558-5646.2008.00424.x. [DOI] [PubMed] [Google Scholar]

[R3] 3.Keyte AL, and Smith KK. (2014). Heterochrony and developmental timing mechanisms: changing ontogenies in evolution. Semin Cell Dev Biol 34, 99–107. 10.1016/j.semcdb.2014.06.015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Gros J, and Tabin CJ. (2014). Vertebrate limb bud formation is initiated by localized epithelial-to-mesenchymal transition. Science 343, 1253–1256. 10.1126/science.1248228. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Sekine K, Ohuchi H, Fujiwara M, Yamasaki M, Yoshizawa T, Sato T, Yagishita N, Matsui D, Koga Y, Itoh N, and Kato S. (1999). Fgf10 is essential for limb and lung formation. Nat Genet 21, 138–141. 10.1038/5096. [DOI] [PubMed] [Google Scholar]

[R6] 6.Takeuchi JK, Koshiba-Takeuchi K, Suzuki T, Kamimura M, Ogura K, and Ogura T. (2003). Tbx5 and Tbx4 trigger limb initiation through activation of the Wnt/Fgf signaling cascade. Development 130, 2729–2739. 10.1242/dev.00474. [DOI] [PubMed] [Google Scholar]

[R7] 7.Barrow JR, Thomas KR, Boussadia-Zahui O, Moore R, Kemler R, Capecchi MR, and McMahon AP. (2003). Ectodermal Wnt3/beta-catenin signaling is required for the establishment and maintenance of the apical ectodermal ridge. Genes Dev 17, 394–409. 10.1101/gad.1044903. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Rallis C, Bruneau BG, Del Buono J, Seidman CE, Seidman JG, Nissim S, Tabin CJ, and Logan MP. (2003). Tbx5 is required for forelimb bud formation and continued outgrowth. Development 130, 2741–2751. 10.1242/dev.00473. [DOI] [PubMed] [Google Scholar]

[R9] 9.Naiche LA, and Papaioannou VE. (2003). Loss of Tbx4 blocks hindlimb development and affects vascularization and fusion of the allantois. Development 130, 2681–2693. 10.1242/dev.00504. [DOI] [PubMed] [Google Scholar]

[R10] 10.Kawakami Y, Capdevila J, Buscher D, Itoh T, Rodriguez Esteban C, and Izpisua Belmonte JC. (2001). WNT signals control FGF-dependent limb initiation and AER induction in the chick embryo. Cell 104, 891–900. 10.1016/s0092-8674(01)00285-9. [DOI] [PubMed] [Google Scholar]

[R11] 11.Choi HMT, Schwarzkopf M, Fornace ME, Acharya A, Artavanis G, Stegmaier J, Cunha A, and Pierce NA. (2018). Third-generation in situ hybridization chain reaction: multiplexed, quantitative, sensitive, versatile, robust. Development 145. 10.1242/dev.165753. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Takeuchi JK, Koshiba-Takeuchi K, Matsumoto K, Vogel-Hopker A, Naitoh-Matsuo M, Ogura K, Takahashi N, Yasuda K, and Ogura T. (1999). Tbx5 and Tbx4 genes determine the wing/leg identity of limb buds. Nature 398, 810–814. 10.1038/19762. [DOI] [PubMed] [Google Scholar]

[R13] 13.Ohuchi H, Nakagawa T, Yamamoto A, Araga A, Ohata T, Ishimaru Y, Yoshioka H, Kuwana T, Nohno T, Yamasaki M, et al. (1997). The mesenchymal factor, FGF10, initiates and maintains the outgrowth of the chick limb bud through interaction with FGF8, an apical ectodermal factor. Development 124, 2235–2244. 10.1242/dev.124.11.2235. [DOI] [PubMed] [Google Scholar]

[R14] 14.Ng JK, Kawakami Y, Buscher D, Raya A, Itoh T, Koth CM, Rodriguez Esteban C, Rodriguez-Leon J, Garrity DM, Fishman MC, and Izpisua Belmonte JC. (2002). The limb identity gene Tbx5 promotes limb initiation by interacting with Wnt2b and Fgf10. Development 129, 5161–5170. 10.1242/dev.129.22.5161. [DOI] [PubMed] [Google Scholar]

[R15] 15.Rodriguez-Esteban C, Tsukui T, Yonei S, Magallon J, Tamura K, and Izpisua Belmonte JC. (1999). The T-box genes Tbx4 and Tbx5 regulate limb outgrowth and identity. Nature 398, 814–818. 10.1038/19769. [DOI] [PubMed] [Google Scholar]

[R16] 16.Duboule D. (1994). Temporal colinearity and the phylotypic progression: a basis for the stability of a vertebrate Bauplan and the evolution of morphologies through heterochrony. Dev Suppl, 135–142. [PubMed] [Google Scholar]

[R17] 17.Minguillon C, Nishimoto S, Wood S, Vendrell E, Gibson-Brown JJ, and Logan MP. (2012). Hox genes regulate the onset of Tbx5 expression in the forelimb. Development 139, 3180–3188. 10.1242/dev.084814. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Moreau C, Caldarelli P, Rocancourt D, Roussel J, Denans N, Pourquie O, and Gros J. (2019). Timed Collinear Activation of Hox Genes during Gastrulation Controls the Avian Forelimb Position. Curr Biol 29, 35–50 e34. 10.1016/j.cub.2018.11.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Tanaka M, Jokubaitis V, Wood C, Wang Y, Brouard N, Pera M, Hearn M, Simmons P, and Nakayama N. (2009). BMP inhibition stimulates WNT-dependent generation of chondrogenic mesoderm from embryonic stem cells. Stem Cell Res 3, 126–141. 10.1016/j.scr.2009.07.001. [DOI] [PubMed] [Google Scholar]

[R20] 20.Mori S, Sakakura E, Tsunekawa Y, Hagiwara M, Suzuki T, and Eiraku M. (2019). Self-organized formation of developing appendages from murine pluripotent stem cells. Nat Commun 10, 3802. 10.1038/s41467-019-11702-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.Logan M, and Tabin CJ. (1999). Role of Pitx1 upstream of Tbx4 in specification of hindlimb identity. Science 283, 1736–1739. 10.1126/science.283.5408.1736. [DOI] [PubMed] [Google Scholar]

[R22] 22.Duboc V, and Logan MP. (2011). Pitx1 is necessary for normal initiation of hindlimb outgrowth through regulation of Tbx4 expression and shapes hindlimb morphologies via targeted growth control. Development 138, 5301–5309. 10.1242/dev.074153. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Itou J, Kawakami H, Quach T, Osterwalder M, Evans SM, Zeller R, and Kawakami Y. (2012). Islet1 regulates establishment of the posterior hindlimb field upstream of the Hand2-Shh morphoregulatory gene network in mouse embryos. Development 139, 1620–1629. 10.1242/dev.073056. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Duboc V, Sulaiman FA, Feneck E, Kucharska A, Bell D, Holder-Espinasse M, and Logan MPO. (2021). Tbx4 function during hindlimb development reveals a mechanism that explains the origins of proximal limb defects. Development 148. 10.1242/dev.199580. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Szeto DP, Rodriguez-Esteban C, Ryan AK, O’Connell SM, Liu F, Kioussi C, Gleiberman AS, Izpisua-Belmonte JC, and Rosenfeld MG. (1999). Role of the Bicoid-related homeodomain factor Pitx1 in specifying hindlimb morphogenesis and pituitary development. Genes Dev 13, 484–494. 10.1101/gad.13.4.484. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Narkis G, Tzchori I, Cohen T, Holtz A, Wier E, and Westphal H. (2012). Isl1 and Ldb co-regulators of transcription are essential early determinants of mouse limb development. Dev Dyn 241, 787–791. 10.1002/dvdy.23761. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Ferrer-Vaquer A, Piliszek A, Tian G, Aho RJ, Dufort D, and Hadjantonakis AK. (2010). A sensitive and bright single-cell resolution live imaging reporter of Wnt/ss-catenin signaling in the mouse. BMC Dev Biol 10, 121. 10.1186/1471-213X-10-121. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] 28.Takemoto T, Abe T, Kiyonari H, Nakao K, Furuta Y, Suzuki H, Takada S, Fujimori T, and Kondoh H. (2016). R26-WntVis reporter mice showing graded response to Wnt signal levels. Genes Cells 21, 661–669. 10.1111/gtc.12364. [DOI] [PubMed] [Google Scholar]

[R29] 29.Molotkov A, Molotkova N, and Duester G. (2005). Retinoic acid generated by Raldh2 in mesoderm is required for mouse dorsal endodermal pancreas development. Dev Dyn 232, 950–957. 10.1002/dvdy.20256. [DOI] [PubMed] [Google Scholar]

[R30] 30.Cunningham TJ, Zhao X, Sandell LL, Evans SM, Trainor PA, and Duester G. (2013). Antagonism between retinoic acid and fibroblast growth factor signaling during limb development. Cell Rep 3, 1503–1511. 10.1016/j.celrep.2013.03.036. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Jeong S, Rebeiz M, Andolfatto P, Werner T, True J, and Carroll SB. (2008). The evolution of gene regulation underlies a morphological difference between two Drosophila sister species. Cell 132, 783–793. 10.1016/j.cell.2008.01.014. [DOI] [PubMed] [Google Scholar]

[R32] 32.Menke DB, Guenther C, and Kingsley DM. (2008). Dual hindlimb control elements in the Tbx4 gene and region-specific control of bone size in vertebrate limbs. Development 135, 2543–2553. 10.1242/dev.017384. [DOI] [PubMed] [Google Scholar]

[R33] 33.Young JJ, Grayson P, Edwards SV, and Tabin CJ. (2019). Attenuated Fgf Signaling Underlies the Forelimb Heterochrony in the Emu Dromaius novaehollandiae. Curr Biol 29, 3681–3691 e3685. 10.1016/j.cub.2019.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Alerasool N, Segal D, Lee H, and Taipale M. (2020). An efficient KRAB domain for CRISPRi applications in human cells. Nat Methods 17, 1093–1096. 10.1038/s41592-020-0966-x. [DOI] [PubMed] [Google Scholar]

[R35] 35.Mic FA, Haselbeck RJ, Cuenca AE, and Duester G. (2002). Novel retinoic acid generating activities in the neural tube and heart identified by conditional rescue of Raldh2 null mutant mice. Development 129, 2271–2282. 10.1242/dev.129.9.2271. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] 36.Savory JG, Bouchard N, Pierre V, Rijli FM, De Repentigny Y, Kothary R, and Lohnes D. (2009). Cdx2 regulation of posterior development through non-Hox targets. Development 136, 4099–4110. 10.1242/dev.041582. [DOI] [PubMed] [Google Scholar]

[R37] 37.Bell CC, Amaral PP, Kalsbeek A, Magor GW, Gillinder KR, Tangermann P, di Lisio L, Cheetham SW, Gruhl F, Frith J, et al. (2016). The Evx1/Evx1as gene locus regulates anterior-posterior patterning during gastrulation. Sci Rep 6, 26657. 10.1038/srep26657. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Qiu C, Martin BK, Welsh IC, Daza RM, Le TM, Huang X, Nichols EK, Taylor ML, Fulton O, O’Day DR, et al. (2024). A single-cell time-lapse of mouse prenatal development from gastrula to birth. Nature 626, 1084–1093. 10.1038/s41586-024-07069-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Yin K, and Juurlink BH. (2000). Regulation of expression of nuclear factor kappa B RelA in oligodendrocytes: effect of hypoxia. Neuroreport 11, 1877–1881. 10.1097/00001756-200006260-00015. [DOI] [PubMed] [Google Scholar]

[R40] 40.Jiang Y, Zhu Y, Wang X, Gong J, Hu C, Guo B, Zhu B, and Li Y. (2015). Temporal regulation of HIF-1 and NF-kappaB in hypoxic hepatocarcinoma cells. Oncotarget 6, 9409–9419. 10.18632/oncotarget.3352. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] 41.Simon MC, and Keith B. (2008). The role of oxygen availability in embryonic development and stem cell function. Nat Rev Mol Cell Biol 9, 285–296. 10.1038/nrm2354. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Mitchell JA, and Yochim JM. (1968). Measurement of intrauterine oxygen tension in the rat and its regulation by ovarian steroid hormones. Endocrinology 83, 691–700. 10.1210/endo-83-4-691. [DOI] [PubMed] [Google Scholar]

[R43] 43.Rodesch F, Simon P, Donner C, and Jauniaux E. (1992). Oxygen measurements in endometrial and trophoblastic tissues during early pregnancy. Obstet Gynecol 80, 283–285. [PubMed] [Google Scholar]

[R44] 44.Lee YM, Jeong CH, Koo SY, Son MJ, Song HS, Bae SK, Raleigh JA, Chung HY, Yoo MA, and Kim KW. (2001). Determination of hypoxic region by hypoxia marker in developing mouse embryos in vivo: a possible signal for vessel development. Dev Dyn 220, 175–186. 10.1002/1097-0177(20010201)220:2<175::AID-DVDY1101>3.0.CO;2-F. [DOI] [PubMed] [Google Scholar]

[R45] 45.Genbacev O, Zhou Y, Ludlow JW, and Fisher SJ. (1997). Regulation of human placental development by oxygen tension. Science 277, 1669–1672. 10.1126/science.277.5332.1669. [DOI] [PubMed] [Google Scholar]

[R46] 46.Aguilera-Castrejon A, Oldak B, Shani T, Ghanem N, Itzkovich C, Slomovich S, Tarazi S, Bayerl J, Chugaeva V, Ayyash M, et al. (2021). Ex utero mouse embryogenesis from pre-gastrulation to late organogenesis. Nature 593, 119–124. 10.1038/s41586-021-03416-3. [DOI] [PubMed] [Google Scholar]

[R47] 47.Culshaw LH, Savery D, Greene NDE, and Copp AJ. (2019). Mouse whole embryo culture: Evaluating the requirement for rat serum as culture medium. Birth Defects Res 111, 1165–1177. 10.1002/bdr2.1538. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Linde-Medina M, and Smit TH. (2021). Molecular and Mechanical Cues for Somite Periodicity. Front Cell Dev Biol 9, 753446. 10.3389/fcell.2021.753446. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Pijuan-Sala B, Griffiths JA, Guibentif C, Hiscock TW, Jawaid W, Calero-Nieto FJ, Mulas C, Ibarra-Soria X, Tyser RCV, Ho DLL, et al. (2019). A single-cell molecular map of mouse gastrulation and early organogenesis. Nature 566, 490–495. 10.1038/s41586-019-0933-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Iyer NV, Kotch LE, Agani F, Leung SW, Laughner E, Wenger RH, Gassmann M, Gearhart JD, Lawler AM, Yu AY, and Semenza GL. (1998). Cellular and developmental control of O2 homeostasis by hypoxia-inducible factor 1 alpha. Genes Dev 12, 149–162. 10.1101/gad.12.2.149. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Carroll C, Engström N, Nilsson PF, Haxen ER, Mohlin S, Berg P, Glud RN, and Hammarlund EU. (2021). Hypoxia Generated by Avian Embryo Growth Induces the HIF-α Response and Critical Vascularization. Frontiers in Ecology and Evolution 9. 10.3389/fevo.2021.675800. [DOI] [Google Scholar]

[R52] 52.Keyte AL, and Smith KK. (2010). Developmental origins of precocial forelimbs in marsupial neonates. Development 137, 4283–4294. 10.1242/dev.049445. [DOI] [PubMed] [Google Scholar]

[R53] 53.Harrison LB, and Larsson HC. (2008). Estimating evolution of temporal sequence changes: a practical approach to inferring ancestral developmental sequences and sequence heterochrony. Syst Biol 57, 378–387. 10.1080/10635150802164421. [DOI] [PubMed] [Google Scholar]

[R54] 54.Lyne A. (1964). Observations on the breeding and growth of the marsupial Perameles nasuta Geoffroy, with notes on other bandicoots. Australian Journal of Zoology 12, 322–339. [Google Scholar]

[R55] 55.Chew KY, Shaw G, Yu H, Pask AJ, and Renfree MB. (2014). Heterochrony in the regulation of the developing marsupial limb. Dev Dyn 243, 324–338. 10.1002/dvdy.24062. [DOI] [PubMed] [Google Scholar]

[R56] 56.Brown DD, Cai L, Das B, Marsh-Armstrong N, Schreiber AM, and Juste R. (2005). Thyroid hormone controls multiple independent programs required for limb development in Xenopus laevis metamorphosis. Proc Natl Acad Sci U S A 102, 12455–12458. 10.1073/pnas.0505989102. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.de Jesus TJ, and Ramakrishnan P. (2020). NF-kappaB c-Rel Dictates the Inflammatory Threshold by Acting as a Transcriptional Repressor. iScience 23, 100876. 10.1016/j.isci.2020.100876. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, Whitwham A, Keane T, McCarthy SA, Davies RM, and Li H. (2021). Twelve years of SAMtools and BCFtools. Gigascience 10. 10.1093/gigascience/giab008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] 59.Huang X, and Huang Y. (2021). Cellsnp-lite: an efficient tool for genotyping single cells. Bioinformatics 37, 4569–4571. 10.1093/bioinformatics/btab358. [DOI] [PubMed] [Google Scholar]

[R60] 60.Huang Y, McCarthy DJ, and Stegle O. (2019). Vireo: Bayesian demultiplexing of pooled single-cell RNA-seq data without genotype reference. Genome Biol 20, 273. 10.1186/s13059-019-1865-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R61] 61.Germain PL, Lun A, Garcia Meixide C, Macnair W, and Robinson MD. (2021). Doublet identification in single-cell sequencing data using scDblFinder. F1000Res 10, 979. 10.12688/f1000research.73600.2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Rubio-de Anda PF, Martinez-Orozco H, and Diaz Miranda SY. (2026). From Tissue to Morphometry: A Step-by-Step Golgi-Cox Workflow for Neuronal and Dendritic Spine Analysis in Adult Mice. J Vis Exp. 10.3791/71384. [DOI] [PubMed] [Google Scholar]

[R63] 63.McCarthy DJ, Campbell KR, Lun AT, and Wills QF. (2017). Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics 33, 1179–1186. 10.1093/bioinformatics/btw777. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Amezquita RA, Lun ATL, Becht E, Carey VJ, Carpp LN, Geistlinger L, Marini F, Rue-Albrecht K, Risso D, Soneson C, et al. (2020). Orchestrating single-cell analysis with Bioconductor. Nat Methods 17, 137–145. 10.1038/s41592-019-0654-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] 65.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, and Satija R. (2024). Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304. 10.1038/s41587-023-01767-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R66] 66.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh PR, and Raychaudhuri S. (2019). Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16, 1289–1296. 10.1038/s41592-019-0619-0. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R67] 67.Traag VA, Waltman L, and van Eck NJ. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Sci Rep 9, 5233. 10.1038/s41598-019-41695-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R68] 68.Wickham J, and Riitters KH. (2019). Influence of high-resolution data on the assessment of forest fragmentation. Landsc Ecol 34, 2169–2182. 10.1007/s10980-019-00820-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R69] 69.Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, Burovski E, Peterson P, Weckesser W, Bright J, et al. (2020). SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods 17, 261–272. 10.1038/s41592-019-0686-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R70] 70.Kontgen F, Grumont RJ, Strasser A, Metcalf D, Li R, Tarlinton D, and Gerondakis S. (1995). Mice lacking the c-rel proto-oncogene exhibit defects in lymphocyte proliferation, humoral immunity, and interleukin-2 expression. Genes Dev 9, 1965–1977. 10.1101/gad.9.16.1965. [DOI] [PubMed] [Google Scholar]

[R71] 71.Nerurkar NL, Lee C, Mahadevan L, and Tabin CJ. (2019). Molecular control of macroscopic forces drives formation of the vertebrate hindgut. Nature 565, 480–484. 10.1038/s41586-018-0865-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R72] 72.Imaz-Rosshandler I, Rode C, Guibentif C, Harland LTG, Ton MN, Dhapola P, Keitley D, Argelaguet R, Calero-Nieto FJ, Nichols J, et al. (2024). Tracking early mammalian organogenesis - prediction and validation of differentiation trajectories at whole organism scale. Development 151. 10.1242/dev.201867. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Shifts in embryonic oxygen levels cue heterochrony in limb initiation

Meng Zhu

ChangHee Lee

Rinaldo Catta-Preta

Clifford J Tabin

Abstract

Graphical Abstract

In Brief:

Introduction

Results

Heterochrony in limb development between mammals and birds can be traced to the start of limb formation via epithelial-to-mesenchymal transition

Figure1. Tbx5 and Tbx4 expression correlate with the timing of Epithelial-to-mesenchymal transition (EMT) and the limb heterochrony.

The heterochrony in EMT correlates with timing of expression of T-box transcription factors

Expression of known upstream activators of Tbx4 is concordant between chicken and mouse embryos

Figure 2. Regional delay of Tbx4 expression mediates EMT heterochrony.

Shifted timing of Tbx4 activation is not due to alterations in associated cis-regulatory elements

Figure 3. cis-regulatory elements responsible for Tbx4 activation are conserved between chicken and mouse embryos.

NFKB transcription factors negatively regulate Tbx4 expression

Figure 4. High cRel expression represses Tbx4 in the hindlimb of mouse embryo.

Maternal hypoxia impedes Tbx4 expression in the mouse hindlimb

Figure 5. Oxygen regulates the timing of Tbx4 expression partially through NFKB transcription factors.

Maternal hypoxia represses Tbx4 expression through cRel and Hif1a

Figure 6. Summary model.

Discussion

Limitations of the study

Resource Availability

Lead Contact

Materials availability

Data and code availability

STAR Methods

Experimental model and study participant details

Animals

Cell lines

Method details

Hybridization chain reaction (HCR)

Quantification of HCR signal

Chicken embryo electroporation and Early Chick (EC) culture

Immunostaining

Plasmids construction

Ex-utero culture and pharmacological treatments

Real-time quantitative PCR (RT-qPCR)

Making transgenic embryonic stem cells (ES)

Making ES-embryo chimera

Generation of Hif1a knockout ES cells

Bulk RNA-sequencing and analyses

ATAC-seq identification of candidate CRE loci

Single cell RNA sequencing & analysis

Quantification and Statistical Analysis

Supplementary Material

Highlights:

Acknowledgment

Footnotes

Material availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases