Skip to main content
PMC home page
eLife logoLink to eLife
. 2018 Sep 3;7:e37429. doi: 10.7554/eLife.37429

An intrinsic cell cycle timer terminates limb bud outgrowth

Joseph Pickering 1, Constance A Rich 1, Holly Stainton 1, Cristina Aceituno 2, Kavitha Chinnaiya 1, Patricia Saiz-Lopez 2, Marian A Ros 2,3, Matthew Towers 1,✉
Editors: Marianne E Bronner4, Marianne E Bronner5
PMCID: PMC6143340  PMID: 30175958

Abstract

The longstanding view of how proliferative outgrowth terminates following the patterning phase of limb development involves the breakdown of reciprocal extrinsic signalling between the distal mesenchyme and the overlying epithelium (e-m signalling). However, by grafting distal mesenchyme cells from late stage chick wing buds to the epithelial environment of younger wing buds, we show that this mechanism is not required. RNA sequencing reveals that distal mesenchyme cells complete proliferative outgrowth by an intrinsic cell cycle timer in the presence of e-m signalling. In this process, e-m signalling is required permissively to allow the intrinsic cell cycle timer to run its course. We provide evidence that a temporal switch from BMP antagonism to BMP signalling controls the intrinsic cell cycle timer during limb outgrowth. Our findings have general implications for other patterning systems in which extrinsic signals and intrinsic timers are integrated.

Research organism: Chicken

Introduction

Detailed models of pattern formation often involve cells being informed of their position in the embryo by interpreting gradients of extrinsic signals, or by measuring intrinsic time. Recently, it has become evident that both processes can operate together - either in parallel, such as in antero-posterior patterning of the limb (thumb to little finger) (Harfe et al., 2004; Towers et al., 2011) and in dorso-ventral patterning of the neural tube (Dessaud et al., 2007) - or sequentially, such as in proximo-distal patterning of the limb (shoulder to finger tips) (Cooper et al., 2011; Roselló-Díez et al., 2014; Saiz-Lopez et al., 2015; Saiz-Lopez et al., 2017). Thus, understanding the precise requirement for extrinsic signals and intrinsic time in pattern formation is a challenging task. In addition, little is known about how these mechanisms operate together to terminate growth following pattern formation.

During early stages of proximo-distal patterning of the chick wing (stage Hamburger Hamilton HH19 – see Materials and methods for embryonic days), signals from the main body of the embryo specify proximal identity (humerus) (Cooper et al., 2011; Mercader et al., 1999; Mercader et al., 2000; Roselló-Díez et al., 2011). Once distal mesenchyme cells of the wing bud are displaced by growth away from proximal signals, there is a switch to an intrinsic timing mechanism that specifies distal identity (elbow to digits) (Saiz-Lopez et al., 2015). This intrinsic timer involves cells executing a programme of proliferation, expressing regulators of distal position (i.e. Hoxa13) and altering their cell surface properties, which are considered to encode position along the proximo-distal axis. In addition, distal limb bud mesenchyme cells produce a factor that maintains the overlying apical ectodermal ridge (Zwilling and Hansborough, 1956) - a thickening of the distal epithelium, which is required for bud outgrowth until HH29/30 (Saiz-Lopez et al., 2015; Saunders, 1948; Summerbell, 1974). This factor is the Bone Morphogenetic Protein (BMP) antagonist Gremlin1 (Grem1), which maintains expression of genes encoding Fibroblast Growth Factors (FGFs) in the apical ectodermal ridge until HH27 (Merino et al., 1999; Pizette and Niswander, 1999; Zúñiga et al., 1999). Reciprocal signalling between the apical ectodermal ridge and the underlying mesenchyme is considered to maintain outgrowth, first by repressing apoptosis, and then by permissively maintaining cell proliferation (Dudley et al., 2002; Saiz-Lopez et al., 2015). Thus, the breakdown of the epithelial-mesenchyme (e-m signalling) feedback loop at HH27, and the subsequent loss of apical ectodermal ridge signalling, is considered necessary to terminate limb bud outgrowth at the end of the patterning phase (Zwilling and Hansborough, 1956; Scherz et al., 2004; Verheyden and Sun, 2008). Note that the final distal phalanx is laid down at around HH28 in the chick wing (Saunders, 1948; Summerbell, 1974). This could suggest that a switch from an intrinsic timer in the mesenchyme back to extrinsic signalling by the apical ectodermal ridge occurs at later stages of proximo-distal patterning to terminate outgrowth. Later growth of the limb after the apical ectodermal ridge has regressed is caused by the expansion of osteogenic progenitor cells in the developing long bones.

However, we recently showed that grafts of HH27 mesenchyme cells expressing Green Fluorescent Protein (GFP) that were made to younger HH20 buds, and which were left to develop for 24 hr until HH24 (now referred to as HH24 graft or HH24g), completed their proliferation programme in the presence of an overlying host apical ectodermal ridge (Saiz-Lopez et al., 2017) red lines - Figure 1a, compare similar trajectories between HH27 to HH29 of normal development – black lines). Thus, HH24g mesenchyme (distal 150 microns of graft) has a similar cell cycle profile to HH29 mesenchyme (stage of donor limb), rather than contralateral HH24 mesenchyme (stage of host – Figure 1a - red lines). Although wing buds that develop with these grafts produce the three main segments of the wing, the autopod (digits and wrist) is very stunted due to the reduced growth of the grafted tissue relative to host tissue (Saiz-Lopez et al., 2017). Therefore, an interpretation of these findings is that proliferative outgrowth terminates intrinsically in the distal mesenchyme (curved arrows in graft in green, Figure 1b) independently of e-m signalling (arrows between ridge and mesenchyme, Figure 1b). An alternative possibility is that extrinsic e-m signalling had irreversibly broken down in donor tissue at the time of grafting and causes loss of proliferation in the graft (arrows absent in HH27 wing in Figure 1c).

Figure 1. Cell cycle and RNA-seq analyses of chick wing distal tips.

(a) Decline in cell cycle rate determined by proportions of cells in G1-, S- and G2/M-phases between HH20 and HH30 in the distal chick wing bud (black lines Saiz-Lopez et al., 2017). Red lines show maintenance of cell cycle program in grafts of HH27 distal tips made to HH20 wing buds left for 24 hr until HH24 (HH24g - tissue would have progressed from HH27 to HH29 in donor) – note trajectories follow same pattern as black lines between HH27 and HH29. (b–c) Procedure for making HH27-20 grafts (Saiz-Lopez et al., 2017) and predictions for loss of proliferative growth in HH24g mesenchyme. (b) Intrinsic termination: e-m signalling (arrows in HH27 wing bud) maintained between mesenchyme and apical ridge but proliferation declines intrinsically in mesenchyme independently of e-m signalling (curved arrows in HH24g mesenchyme). (c) Extrinsic termination: e-m signalling irreversibly lost between graft and apical ridge in HH27 bud (arrows absent) and proliferation lost in graft. (d) Heat-map showing the correlation (Pearson) of the normalised RNA-seq data collapsed to the mean expression per group and the degree of correlation indicated by the colour (red: higher, blue: lower). (e) KEGG analyses across pairwise contrasts with degree in pathway change indicated by the colour (red: up-regulated, blue: down-regulated). (f–h) Clustering of RNA-seq data across pairwise contrasts with degree of gene expression change indicated by the colour (red: higher, blue: lower). (f) Cluster 1: genes that increase between HH24 and HH29 (black line) and are maintained in HH24g (red line - > 2 fold higher in HH24g than HH24). (g) Cluster 2: genes that increase between HH24 and HH29 (black line) and reset in HH24g (red line - > 2 fold lower in HH24g than HH29). (h) Cluster 3: genes that decrease between HH24 and HH29 (black line) and reset in HH24g (red line - > 2 fold higher in HH24g than HH29).

Figure 1.

Figure 1—figure supplement 1. Cluster 1 Clustering of RNA-seq data across pairwise contrasts with degree of gene expression change indicated by the colour (red: higher, blue: lower).

Figure 1—figure supplement 1.

Cluster 1: genes that increase between HH24 and HH29 (black line) and are maintained in HH24g distal mesenchyme (red line - > 2 fold higher in HH24g than HH24).

Figure 1—figure supplement 2. Cluster 2 Clustering of RNA-seq data across pairwise contrasts with degree of gene expression change indicated by the colour (red: higher, blue: lower).

Figure 1—figure supplement 2.

Cluster 2: genes that increase between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme (red line - > 2 fold lower in HH24g than HH29).

Figure 1—figure supplement 3. Cluster 3 Clustering of RNA-seq data across pairwise contrasts with degree of gene expression change indicated by the colour (red: higher, blue: lower).

Figure 1—figure supplement 3.

Cluster 3: genes that decrease between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme (red line - > 2 fold higher in HH24g than HH29).
Open in a new tab

In this study, we show that proliferative outgrowth at the end of the patterning phase terminates intrinsically in the distal mesenchyme of the chick wing bud in the presence of e-m signalling. Our data provide evidence that an intrinsic switch from autocrine BMP antagonism to autocrine BMP signalling controls the cell cycle programme in distal mesenchyme during limb bud outgrowth.

Results

RNA sequencing reveals reversible and stable gene expression

To gain insights into how intrinsic cell cycle timing is maintained in limb development, we used RNA-seq to determine how the transcriptome changes over time in the distal part of the wing bud during normal development and then compared this to the HH24g transcriptome. To achieve this, we performed RNA-seq on pooled blocks of distal cells from HH24, HH24g, HH27 and HH29 chick wing buds (three replicates of 12 blocks taken from distal tissue excluding the polarizing region– note one HH24 sample failed quality control - see Materials and methods). The region of undifferentiated mesenchyme is considered to extend proximally beneath the apical ectodermal ridge by 200 – 300 microns and to remain at a consistent length throughout outgrowth (Summerbell et al., 1973), but for our analyses, we used RNA extracted from the distal 150 microns of control buds and grafts to ensure consistency between all samples. We compared HH24g with contralateral HH24 datasets and HH24g with HH29 datasets (the stage the graft would have developed to if left in situ in the donor wing). This identified 154 genes that are differentially expressed (>2 fold difference at a p-value of <0.0005) between the two comparisons: 55 genes between the HH24g-HH24 pair and 99 genes between the HH24g-HH29 pair (Supplementary files 1 and 2). Pearson correlation of the normalised data reveals that overall gene expression levels in HH24g distal cells are closer to HH24 than to HH29 cells, suggesting a good degree of resetting behaviour (Figure 1d). This is unexpected considering the idea that older mesenchyme cells maintain both their proliferative and patterning potential when grafted to a younger bud (Saiz-Lopez et al., 2017).

We performed KEGG (Kyoto Encyclopaedia of Genes and Genomes) analyses to identify up- and down-regulated pathways between the two pairwise comparisons (Figure 1e, Supplementary file 1 and 2 - the pathways identified were those that contained two or more genes at a >2 fold difference at a p-value of 0.05 – it should be noted that this analysis is limited since only approximately 20% of chicken genes belong to verified KEGG pathways). There are five down-regulated pathways identified by the KEGG analyses in HH24g compared to HH29 distal cells - focal adhesion, ECM (extracellular matrix), gap junction, MAPK (Mitogen-Activated Protein Kinase) and actin cytoskeleton - suggesting a general reset of genes in these pathways that increase in expression over time between HH24 and HH29. In addition, there are four up-regulated pathways identified by the KEGG analyses in HH24g compared to HH24 distal cells - ECM, focal adhesion, adrenergic signalling, TGFβ Transforming Growth Factor Beta) - suggesting maintained expression of genes in these pathways that generally increase in expression over time. In addition, the MAPK pathway is also up-regulated in HH24g compared to HH29 distal cells, showing the complex nature of this signalling pathway, and suggesting some genes in this pathway that decrease in expression over time can be reset. Interestingly, pathways represented by genes whose decrease in expression over time is maintained in HH24g distal cells were not found.

To identify genes that share similar behaviour across the two pairwise contrasts, we performed hierarchical clustering analyses of the 154 genes significantly altered in expression between the HH24g/HH24 and HH24g/HH29 datasets. This divided the genes into three clusters (See Materials and methods – note three genes were excluded that fell into more than one cluster). Cluster 1 (Figure 1f, Figure 1—figure supplement 1, Supplementary file 3) contains 54 genes that increase in expression between HH24 and HH29 during normal development (black line) and are maintained in HH24g distal cells (maintain a >2 fold increase in expression compared to equivalent area in contralateral HH24 buds – red line). Cluster 2 (Figure 1g, Figure 1—figure supplement 2 and Supplementary file 4) contains 44 genes that normally increase in expression between HH24 and HH29 (black line) and are not maintained in HH24g distal cells (>2 fold decrease in expression compared to HH29 buds - red line). Cluster 3 (Figure 1h, Figure 1—figure supplement 3 and Supplementary file 5) contains 53 genes that decrease in expression between HH24 and HH29 (black line) and are reset in HH24g grafts (>2 fold increase in expression compared to HH29 buds - red line). Thus, genes in cluster one reflect stability of expression while genes in clusters 2 and 3 show reversibility of expression. In consensus with the KEGG analyses, a cluster containing genes whose decrease in expression over time is maintained in HH24g distal cells was not found, indicating that the young environment potently reactivates gene expression. These data suggest that many genes that are preferentially expressed during early stages of wing development can be reset to some degree (Cluster 3). However, later expressed genes show differential resetting behaviour in an earlier environment (Clusters 1 and 2).

Bmp2/4/7 expression is maintained in HH24g mesenchyme

To identify signalling pathways associated with the maintenance of cell cycle timing in HH24g mesenchyme, we analysed the KEGG and clustering analyses for pathways/genes predominantly active at late stages of development, and that are maintained in distal mesenchyme transplanted to an earlier environment. This highlighted the TGFβ pathway (Figure 1e) and Bmp2 and Bmp7 in cluster 1 that encode members of the Bone Morphogenetic Protein (BMP) family, which are signalling ligands of the TGFβ superfamily (Figure 1f, Figure 1—figure supplement 1, Supplementary file 3). Notably, BMP signalling is implicated in terminating outgrowth at the end of the patterning phase by inhibiting FGF signalling by the apical ectodermal ridge (Pizette and Niswander, 1999). Read-counts of the RNA-seq data show that the expression of Bmp2 increases by 5.3-fold between HH24 and HH29 – similar to the difference in expression in HH24g tissue compared to HH24 tissue (6.2-fold - log2 scale is shown), thus showing that late expression levels of Bmp2 are maintained in an earlier environment (Figure 2a). Similarly, a 1.5-fold increase in Bmp7 expression occurs between HH24 and HH29, and the difference between expression levels in HH24g tissue compared to HH24 tissue is 2.5-fold (Figure 2c). Analyses by RNA in situ hybridization are consistent with these observations (Figure 2b and d). We also analysed the expression of Bmp4 which has a prominent role in limb development (Selever et al., 2004; Bénazet et al., 2009). The RNA-seq data indicates that Bmp4 expression levels decrease by two-fold between HH24 and HH29 and that expression levels are approximately equivalent between HH24 and HH24g tissue (Figure 2e, note stronger expression in ridge at HH24 compared to Bmp2/7). Despite this, slightly enhanced expression could be detected by in situ hybridization in the region of the grafted tissue in HH24g buds (asterisk - Figure 2f) in comparison to the equivalent position in contralateral buds. However, cluster 3, which comprises genes that decrease over time and that demonstrate resetting behaviour, contains (Gremlin1) Grem1, which encodes the BMP antagonist that maintains the apical ectodermal ridge (Figure 1—figure supplement 3, Supplementary file 5). Between HH24 and HH29, Grem1 expression levels are reduced by 17.2-fold until they are almost undetectable by in situ hybridization (Figure 2g and h). However, Grem1 expression levels are approximately the same in HH24 and HH24g mesenchyme (Figure 2g and h). This result suggests that Grem1 could antagonise BMP2/4/7 signalling in HH24g mesenchyme.

Figure 2. Bmp2/4/7 and Grem1 expression in HH24g mesenchyme.

Figure 2.

Open in a new tab

(a) Histogram showing expression levels of Bmp2 as normalised log2 values of RNA sequencing read-count intensities. (b) In situ hybridization showing Bmp2 expression - note intense expression in HH24g mesenchyme in area of graft (n = 3/3, HH24 is the contralateral bud flipped horizontally). (c) Expression levels of Bmp7 determined by RNA-seq. (d) In situ hybridization showing Bmp7 expression - note intense region of expression in HH24g mesenchyme (n = 3/3). (e) Expression levels of Bmp4 determined by RNA-seq. (f) In situ hybridizations showing Bmp4 expression - note enhanced expression in HH24g mesenchyme in area of graft (asterisk - n = 2/3). (g) Expression levels of Grem1 determined by RNA-seq. (h) In situ hybridizations showing resetting of Grem1 expression (n = 7/9). Scale bars: HH24 buds - 500 μm; HH29 buds - 200 μm.

BMP signalling is maintained in HH24g mesenchyme

To analyse BMP signalling in HH24g mesenchyme, we selected a bona-fide target, Msx2 (Pizette et al., 2001). RNA-seq data shows that Msx2 has a similar profile to both Bmp2 and Bmp7: a 2.8-fold increase between HH24 and HH29 and a two-fold difference in HH24g tissue compared to HH24 tissue (Figure 3a and b). Consistent with this finding, immunofluorescence on sections using an antibody that recognizes phosphorylated SMAD1/5/9 – the signal transducer for BMP2/4/7 – reveals increased levels in later wing buds (Figure 3c–e). However, whereas weak expression is detected in HH24 wing buds (Figure 3c), strong expression is observed in the distal part of grafts (the area processed for RNA-seq) in contralateral HH24g buds (Figure 3f). Control grafts of HH20 distal mesenchyme made to host HH20 wing buds show no obvious differences in pSMAD1/5/9 signal intensity after 24 hr compared to stage-matched HH24 buds (Figure 3g–compare to Figure 3c). Enhanced pSMAD1/5/9 signal is also detected in the host apical ectodermal ridge above HH24g mesenchyme (asterisk Figure 3h), in contrast to adjacent regions (arrow – Figure 3h). This result suggests that the level of BMP signalling is maintained in mesenchyme transplanted from older buds to a younger environment, and that it is also enhanced in the host apical ectodermal ridge in response to the older mesenchyme.

Figure 3. BMP signalling in HH24g mesenchyme.

Figure 3.

Open in a new tab

(a) Expression levels of Msx2 determined by RNA-seq. (b) In situ hybridization showing Msx2 expression - note intense expression in HH24g mesenchyme in grafted area (n = 3/4). (c–h) Immunostaining of pSMAD1/5/9: signal increases in the distal part of chick wing buds over time (c–e) and is enhanced in HH24g mesenchyme (f) compared to contralateral HH24 buds (n = 3/3 c) but not in HH20-HH20 grafts (n = 2/2 g); signal is enhanced in apical ectodermal ridge above grafts in HH24g mesenchyme (asterisk-h) but not adjacent regions of ridge (n = 2/2 - arrow-h). (i–j) Lysotracker staining of apoptotic cells reveals no difference in HH24 (i) and HH24g (j) mesenchyme (n = 5/5) – asterisks indicate anterior necrotic zone. (k) Flow cytometry of wing bud distal mesenchyme: PBS- and Noggin-soaked beads do not significantly affect proportion of cells in G1-phase after 8 hr compared to left wing controls (Pearson’s χ2 test – p=0.1 and p=0.4, respectively); Noggin-soaked beads implanted at HH20 (+Nog) produce a significant decrease in proportion of cells in G1-phase in HH24g grafts compared to PBS-treated controls (+PBS) after 8 hr (Pearson’s χ2 test - p<0.0001); BMP2-soaked beads implanted at HH20 significantly increase G1-phase cells after 8 hr compared to left wing controls (Pearson’s χ2 test - p<0.0001), note boxes indicate separate experiments. (l–m) Lysotracker staining of apoptotic cells reveals no difference in PBS (l) - n = 2/2) and BMP2 (m) - n = 2/2) treated mesenchyme after 8 hr. Scale bars: HH24 buds - 500 μm, HH29 buds - 200 μm in b; 150 μm in c-g; 75 μm in h; 300 μm in i-j and l-m.

Figure 3—source data 1. Flow cytometry graphs for cell cycle analyses.
elife-37429-fig3-data1.pdf (962.7KB, pdf)
DOI: 10.7554/eLife.37429.008

BMP signalling is associated with regions of apoptosis in the developing limb, including anterior and posterior necrotic zones, as well as the interdigital mesenchyme (Zuzarte-Luis and Hurle, 2005). To determine if apoptosis could reduce proliferation in HH24g mesenchyme, we performed lysotracker staining of apoptotic cells. This reveals no appreciable difference in HH24 control (Figure 3i) and HH24g mesenchyme (Figure 3j) in whole-mount preparations. Note, the anterior necrotic zone can be observed in both examples (asterisks in Figure 3i and j).

To investigate if BMP signalling is responsible for the reduction in proliferation in HH24g mesenchyme (Saiz-Lopez et al., 2017), we implanted beads soaked in the BMP antagonist, Noggin, into HH20 wing buds at the time of grafting. After 8 hr, we used flow cytometry to determine the proportion of cells in the different phases of the cell cycle in distal tissue (150 microns from the distal tip). Control PBS- and Noggin-soaked beads implanted into HH20 wing buds for 8 hr do not significantly change distal mesenchyme cell cycle parameters compared to control left wing buds (Figure 3k). In HH24g grafts treated with PBS-soaked beads, 69% of distal mesenchyme cells are in G1-phase compared to 60.8% in the equivalent region of the contralateral wing bud (Figure 3k). This is consistent with the reduced proliferative potential we previously observed in such grafts (Figure 1a, [Saiz-Lopez et al., 2017]). However, in HH24g mesenchyme treated with Noggin-soaked beads, the proportion of cells in G1-phase decreases from 69% to 62.4% - a value significantly closer to the control value of 60.8% in the contralateral bud (Figure 3k). This result shows that the inhibition of BMP signalling in HH24g mesenchyme can significantly enhance proliferation.

To test if BMP signalling has an instructive effect on cell proliferation in early wing buds, we implanted beads soaked in two concentrations of recombinant BMP2 protein (BMP2high and BMP2low - see Materials and methods) into the distal mesenchyme of HH20 buds, and compared cell cycle profiles to those obtained in contralateral wing buds. After 8 hr, the proportion of cells in G1-phase in control wing buds is 57.8%, but this increases to 60.5% and 66.2% in BMP2low and BMP2high buds, respectively (Figure 3k). This finding shows that enhanced BMP signalling inhibits proliferation of distal mesenchyme cells in the early chick wing bud in a dose-dependent fashion. In addition, lysotracker staining shows that both PBS- (Figure 3l) and BMP2high -soaked beads (Figure 3m) do not cause apoptosis after 8 hr, thus showing that BMP signalling does not inhibit proliferation by compromising cell survival. Taken together, these results suggest that BMP signalling in HH24g mesenchyme maintains an intrinsic cell cycle programme.

FGF response is reset in HH24g mesenchyme

The maintenance of BMP signalling in HH24g mesenchyme could also reduce proliferation via inhibition of FGFs produced by the apical ectodermal ridge. However, consistent with the observation that Grem1 expression is maintained in HH24g mesenchyme (Figure 2g,h), we have previously shown that Fgf8 expression is maintained in the apical ectodermal ridge above grafts of distal mesenchyme from much older wing buds - including in tissue that would have developed to stage HH36 if left in situ (or day 10 of incubation – the ridge normally regresses at HH29/30 in the chick wing at day 6–6.5 of incubation - Saiz-Lopez et al., 2017). To determine if HH24g distal mesenchyme cells are responsive to apical ectodermal ridge signals, we analysed a well-defined read-out of FGF signalling in the limb, Dusp6 (Eblaghie et al., 2003) - also known as Mkp3 and Pyst1 - see also Figure 4g–j). RNA-seq read-counts show that Dusp6 expression decreases by 2.1-fold between HH24 and HH29, but that no significant difference in expression is observed between HH24g tissue and the equivalent region of contralateral control buds, indicating that the level of FGF signalling is approximately the same in the grafted cells as in the host wing bud (Figure 4a and b). We also analysed the expression of Sprouty2 (Spry2) that is another well-defined downstream target of FGF signalling (Chambers and Mason, 2000). Similar to Dusp6, expression of Spry2 decreases between HH24 and HH29 (1.7-fold), but is approximately equivalent in HH24g compared to control HH24 mesenchyme (Figure 4c and d). These data show that Dusp6 and Spry2 expression levels decrease over time, consistent with the maturation and regression of the apical ectodermal ridge, and that both are expressed in HH24g mesenchyme at similar level as in surrounding host tissue, thus indicating that grafted cells are responding to FGF signalling. Notably, cluster 3 contains genes encoding other members of the FGF family, including Fgf13 and Fgf18, which are expressed in distal mesenchyme at early stages (Munoz-Sanjuan et al., 1999; Mok et al., 2014) and that can be reset in a younger environment (Figure 1—figure supplement 3, Supplementary file 5). RNA-seq read-counts show that Fgf18 decreases by 90-fold between HH24 and HH29, by which point it is undetectable by in situ hybridization (Figure 4e and f), but expression is only 3.4-fold higher in HH24 tissue compared to HH24g tissue. Therefore, taken together, these results show that HH24g mesenchyme is responsive to FGF signalling.

Figure 4. FGF signalling in HH24g mesenchyme.

Figure 4.

Open in a new tab

(a-e) Resetting of gene expression in HH24g mesenchyme. Expression levels of Dusp6 (a), Spry2 (c) and Fgf18 (e) determined by RNA-seq. (b) In situ hybridization showing equivalent levels of Dusp6 (n = 3/3, (b), Spry2 (n = 3/3, (d), and Fgf18 (n = 3/3, (f) in HH24 and contralateral HH24g mesenchyme. (g–j) FGF-soaked beads (j) n = 3/3), but not PBS-soaked beads (h) n = 3/3), up-regulate expression of Dusp6 after 6 hr (control left wing buds flipped horizontally – (g) and (i). (k) Flow cytometry of wing bud distal mesenchyme: PBS-soaked beads do not significantly affect proportion of cells in G1-phase after 6 hr compared to left wing controls (Pearson’s χ2 test – p=0.5); FGF-soaked beads implanted at HH27 significantly increase G1-phase cells compared to left wing control after 6 hr (Pearson’s χ2 test - p<0.0001), note boxes indicate separate experiments. Scale bars: HH24 buds, HH27 buds - 300 μm; HH29 buds - 200 μm.

Figure 4—source data 1. Flow cytometry graphs for cell cycle analyses.
elife-37429-fig4-data1.pdf (399.4KB, pdf)
DOI: 10.7554/eLife.37429.010

Our data suggest that autocrine BMP signalling is responsible for the intrinsic decline in proliferation in distal mesenchyme of the chick wing bud. However, we also examined if FGF signalling could also have an instructive role. To do this, we implanted beads soaked in recombinant FGF8 protein to the distal mesenchyme of chick wing buds at HH27, the stage at which proliferation rapidly starts to decline (Figure 1a) (Saiz-Lopez et al., 2017) and determined cell cycle profiles. RNA in situ hybridisation shows that FGF-soaked beads, but not PBS-soaked beads, implanted into HH27 wing buds for 6 hr increase expression of Dusp6 (Figure 4g–j). Flow cytometry reveals that FGF-soaked beads, but not control PBS-soaked beads, significantly increase the proportion of cells in G1-phase compared to control left wing buds (Figure 4k– 70.8 vs. 67.8%). Therefore, FGF signalling appears to have a slight inhibitory effect on proliferation, rather than instructively regulating it in a positive manner.

Discussion

By grafting late mesenchyme cells into an early epithelial environment, we have determined that the progressive loss of e-m signalling is not required for the decline and cessation of chick wing bud outgrowth at the end of the patterning phase. Our data reveals that the duration of mesenchyme cell proliferation is intrinsically determined but requires permissive signals from the apical ectodermal ridge: a distinction not recognized in previous approaches. For instance, surgical removal of the apical ectodermal ridge in the chick wing before HH24 causes apoptosis in the distal mesenchyme, but at later stages decreases proliferation (Dudley et al., 2002). In addition, genetic approaches that compromise apical ectodermal ridge signalling in the mouse limb also cause defective outgrowth (Lu et al., 2008; Mariani et al., 2008; Yu and Ornitz, 2008). The fact that mesenchyme expansion was naturally inhibited in these experiments therefore masked the intrinsic control of proliferation in this tissue.

A schematic detailing the relationship between e-m signalling and mesenchyme proliferation during normal wing outgrowth is shown in Figure 5. At early stages of wing development (HH20), the Grem1-mediated inhibition of BMP signalling fulfils two main roles: to maintain the apical ectodermal ridge until HH28 as classically-described (Summerbell, 1974; Summerbell et al., 1973) and to permit a rapid rate of proliferation in distal mesenchyme cells as we have shown here. Around HH20, the mechanism of positional specification switches in mesenchyme progenitors from extrinsic signalling (humerus) (Cooper et al., 2011; Roselló-Díez et al., 2011) to an intrinsic timer (distal to elbow) (Saiz-Lopez et al., 2015). This intrinsic mechanism involves a progressive reduction in the proliferation rate as the elements are laid down in a proximal to distal sequence until HH28 (Saiz-Lopez et al., 2015) (Figure 5). Here, we showed that the deceleration in proliferation rate is caused by an increase in autocrine BMP signalling between HH24 and HH29 that overcomes Grem1-mediated antagonism in distal mesenchyme cells. Once the intrinsic cell cycle timer has run its course at HH29 and distal mesenchyme expansion ceases, high BMP signalling causes the apical ectodermal ridge to regress at HH29/30 (Figure 5). Interestingly, recent genetic evidence in the mouse has provided evidence that BMP signalling terminates the extension of the main body axis (Anderson et al., 2016).

Figure 5. An intrinsic cell cycle timer terminates limb bud outgrowth.

Figure 5.

Open in a new tab

HH20-HH24: antagonism of BMP signalling by Grem1 permits rapid mesenchyme proliferation and maintenance of the apical ectodermal ridge (purple) that permissively supports outgrowth. HH24-HH27: intrinsic rise in BMP signalling counters BMP antagonism and instructively decelerates mesenchyme proliferation rate. HH27-HH30: mesenchyme proliferation diminishes as patterning completes at HH28 and then BMP signalling inhibits FGF signalling in apical ectodermal ridge that then regresses.

It was unexpected that all the genes identified in the RNA-seq analysis that are predominantly expressed at early stages could be reset to some degree by the signalling environment (Cluster 3). This is consistent with the observations that early proximo-distal patterning is controlled by extrinsic signals – possibly retinoic acid signalling from the flank of the embryo (Cooper et al., 2011; Roselló-Díez et al., 2011). Notably, this cluster contains several genes encoding products known to be involved in maintaining a stem-cell-like state including Plzf (Zbtb16) (Liu et al., 2016) Sall4 (Yang et al., 2008) (also Sall1 and 3), Gbx2 (Tai and Ying, 2013) and Lin28b (Zhang et al., 2016). This raises the possibility that a gene regulatory network operates during early limb development to maintain the multipotency of mesenchyme cells and warrants further investigation.

Opposing gradients of BMP ligands and their antagonists specify positional identities and regulate growth in a concentration-dependent manner in many developmental contexts (Bier and De Robertis, 2015). Thus, the intrinsic switch from BMP antagonism to BMP signalling in limb development that we describe here provides a mechanism by which cells can be instructed by autocrine BMP signalling in the absence of spatial gradients. It will be important to identify if this mechanism operates in other patterning systems.

Materials and methods

Chick husbandry and tissue grafting

Wild-type and GFP-expressing Bovans Brown chicken eggs were incubated and staged according to Hamburger Hamilton (Hamburger and Hamilton, 1951) HH19 is embryonic day 3, HH24 - day 4, HH27 - day 5, HH29 – day 6 and HH30, day 7. For tissue grafting, the polarizing region was discarded and a stripe of 150 µm of distal sub-AER mesenchyme about 500 µm long was dissected from HH27 wing buds. The ectoderm was removed after incubation in 0.25% trypsin at room temperature for 2 min and the mesenchyme was then cut into three cuboidal pieces. These were then placed in slits made using a fine sharpened tungsten needle between the apical ectodermal ridge and underlying mesenchyme in the mid-distal region of HH20 wing buds.

RNA sequencing analyses and clustering

Grafts were made as described above and after 24 hr the 150 µm distal part of the graft was carefully dissected under a UV scope. Equivalent tissue was dissected from the contralateral HH24 wing bud and also 150 µm of tissue was dissected from three positions at the distal tips of HH24, HH27 and HH29 wing buds. Three replicate experiments were performed from each condition and the tissue was pooled before the RNA was extracted using Trizol reagent (Gibco), and sequenced on a HiSeq 2000 (Paired end readings of 50 bp - Instrument: ST300). Sequencing data was mapped using Hisat v2.0.3 (bam file generation). Salmon output was used to quantify the data (number of reads) (Patro et al., 2017). The raw data has been deposited in array express (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6437/). Twelve samples (three for each of the four conditions HH24, HH24g, HH27 and HH29) were checked for quality issues. This was done by manually inspecting the density plot, boxplots, PCA plots, correlation heatmap and distance plot, as well as using several automatic outlier tests, namely distance, Kolmogorov-Smirnov, correlation and Hoeffding's D. Based on this, one of the HH24 samples was excluded from the analysis.

The count data for the samples were normalised using trimmed mean of m-value normalisation and transformed with Voom, resulting in log2-counts per million with associated precision weights (Robinson and Oshlack, 2010; Law et al., 2014). A heat-map was made showing the correlation (Pearson) of the normalised data collapsed to the mean expression per group. A statistical analysis using an adjusted p-value<0.0005 and a fold-change >2 identified 151 genes as differentially expressed in the two contrasts evaluated. Gene clusters were identified from the set of 151 unique genes that were differentially expressed in at least one of the statistical comparisons. The evaluation considered between two and 35 clusters using hierarchical, k-means, and PAM clustering methods based on the internal, stability and biological metrics provided from the clValid R package. The majority of internal validation and stability metrics indicated that either the lowest possible number or conversely the highest number evaluated were preferable. The metrics giving more nuanced information in the intermediate range were the Silhouette measure, and the Biological Homogeneity Index (BHI). Based on manual inspection, it was decided to use hierarchical clustering with k = 3 gene clusters, which showed favourable properties for both these measures. For KEGG pathway analyses, significant genes with adjusted p-value<0.05 and fold-change >= 2 from each comparison were analysed for enrichment of KEGG pathway membership using a hypergeometric test. Enrichment (p<0.05) was assessed for up and down-regulated genes separately.

Apoptosis analyses

Chick wing buds were dissected in PBS and transferred to Lysotracker (Life Technologies, L-7528) PBS solution (1:1000) in the dark. Wing buds were incubated for 1 hr at 37°C, washed in PBS, and fixed overnight in 4% PFA at 4°C. Wing buds were then washed in PBS and progressively dehydrated in a methanol series.

Whole mount in situ hybridisation

Embryos were fixed in 4% PFA overnight at 4°C, dehydrated in methanol overnight at −20°C, rehydrated through a methanol/PBS series, washed in PBS, then treated with proteinase K for 20 min (10 μg/ml−1), washed in PBS, fixed for 30 mins in 4% PFA at room temperature and then prehybridised at 69°C for 2 hr (50% formamide/50% 2x SSC). 1 μg of antisense DIG-labelled mRNA probes were added in 1 ml of hybridisation buffer (50% formamide/50% 2x SSC) at 69°C overnight. Embryos were washed twice in hybridisation buffer, twice in 50:50 hybridisation buffer and MAB buffer, and then twice in MAB buffer, before being transferred to blocking buffer (2% blocking reagent 20% lamb serum in MAB buffer) for 2 hr at room temperature. Embryos were transferred to blocking buffer containing anti-digoxigenin antibody (1:2000) at 4°C overnight, then washed in MAB buffer overnight before being transferred to NTM buffer containing NBT/BCIP and mRNA distribution visualised using a LeicaMZ16F microscope.

Immunohistochemistry

Embryos were fixed in 4% PFA for 2 hr on ice, washed in PBS and transferred to 30% sucrose overnight at 4°C. Embryos were frozen in OCT mounting medium, fixed to a chuck and cryosectioned immediately. Sections were dried for 2 hr and washed in PBS containing 0.1% triton for 10 min at room temperature. Sections were blocked in 0.1% Triton, 1 – 2% hings/serum for 1 – 1.5 hr at room temperature. Primary antibody solution was added (1:500 rabbit anti-PSMAD1/5/9 Cell Signalling Technology – D5B10) in PBS/0.1% triton/1 – 2% hings, before coverslips were added and slides placed in a humidified chamber at 4°C for 72 hr. Slides were then washed for 3 × 5 min in PBS at room temperature. Secondary antibody (1:500 anti-rabbit Alexa Fluor 594 Cell Signalling Technology – 8889S) was added in PBS 0.1% triton/1 – 2% hings for 1 hr at room temperature. Slides were washed in PBS for 3 × 5 min and mounted in Vectashield/DAPI medium. Slides were left at 4°C and imaged the following day.

Flow cytometry

Distal 150 μm blocks of mesenchyme tissue were dissected in ice cold PBS under a LeicaMZ16F microscope using a fine surgical knife pooled from replicate experiments (12), and digested into single-cell suspensions with trypsin (0.5%, Gibco) for 30 min at room temperature. Cells were briefly washed twice in PBS, fixed in 70% ethanol overnight, washed in PBS and re-suspended in PBS containing 0.1% Triton X-100, 50 µg/ml−1 of propidium iodide and 50 µg/ml−1 of RNase A (Sigma). Dissociated cells were left at room temperature for 20 min, cell aggregates were removed by filtration and single cells analysed for DNA content with a FACSCalibur flow cytometer and FlowJo software (Tree star Inc). Based on ploidy values cells were assigned in G1-, S-, or G2/M-phases and this was expressed as a percentage of the total cell number (>7500 cells in each case). Statistical significance of numbers of cells between pools of dissected wing bud tissue (10–12 in each pool) was determined by Pearson’s χ2 tests to obtain two-tailed p-values (significantly different being a p-value of less than 0.05 - see (Chinnaiya et al., 2014) - statistical comparisons of cell cycle parameters between the left and right wing buds embryos showed that differences are less than 1%).

Bead implantation

Control affigel blue beads (Biorad) were soaked in PBS/4 mM HCl. Affigel beads were soaked in Human Noggin (0.025 μg μl−1 - R and D) or human BMP2 protein (0.025 μg μl−1 and 0.05 μg μl−1 - R and D) dissolved in PBS/4 mM HCl. Control heparin beads (Biorad) were soaked in PBS. Heparin beads were soaked in human FGF8 (0.5 μg μl−1 - R and D dissolved in PBS. Beads were soaked for 2 hr and implanted into the distal mesenchyme using a sharp needle.

Acknowledgements

We thank the Wellcome Trust for funding, Cheryll Tickle for critical reading, Adrian Sherman/Helen Sang for GFP-expressing chicken embryos, the University of Sheffield flow cytometry core facility and Max Bylesjo and Paul McAdam at FIOS Genomics for bioinformatics.

Funding Statement

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

Contributor Information

Matthew Towers, Email: m.towers@sheffield.ac.uk.

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

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

Funding Information

This paper was supported by the following grants:

  • Wellcome Trust 202756/Z/16/Z to Joseph Pickering, Constance A Rich, Holly Stainton, Kavitha Chinnaiya, Matthew Towers.

  • Spanish Ministerio de Economia BFU2017-88265- P to Cristina Aceituno, Patricia Saiz-Lopez, Marian A Ros.

Additional information

Competing interests

No competing interests declared.

Author contributions

Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft.

Formal analysis, Validation, Investigation.

Formal analysis, Investigation.

Formal analysis, Investigation.

Formal analysis, Investigation.

Conceptualization, Investigation, Methodology.

Formal analysis, Supervision, Validation, Investigation, Methodology, Project administration, Writing—review and editing.

Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Methodology, Writing—original draft, Project administration, Writing—review and editing.

Additional files

Supplementary file 1. RNA sequencing of HH24g vs.

HH24 distal mesenchyme 55 genes are differentially expressed between HH24g and HH24 distal mesenchyme. 53 genes are increased in expression in HH24g distal mesenchyme and two genes are decreased in expression.

elife-37429-supp1.xls (6.7KB, xls)
DOI: 10.7554/eLife.37429.012
Supplementary file 2. RNA sequencing of HH24g vs.

HH29 distal mesenchyme 99 genes are differentially expressed between HH24g and HH29 distal mesenchyme. 55 genes are increased in expression in HH24g distal mesenchyme and 44 genes are decreased in expression.

elife-37429-supp2.xls (12KB, xls)
DOI: 10.7554/eLife.37429.013
Supplementary file 3. Cluster 1 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

Cluster 1: genes that increase between HH24 and HH29 (black line) and are maintained in HH24g distal mesenchyme.

elife-37429-supp3.xls (1.5KB, xls)
DOI: 10.7554/eLife.37429.014
Supplementary file 4. Cluster 2 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

Cluster 2: genes that increase between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme.

elife-37429-supp4.xls (1.3KB, xls)
DOI: 10.7554/eLife.37429.015
Supplementary file 5. Cluster 3 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

Cluster 3: genes that decrease between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme.

elife-37429-supp5.xls (1.5KB, xls)
DOI: 10.7554/eLife.37429.016
Transparent reporting form
elife-37429-transrepform.docx (249.1KB, docx)
DOI: 10.7554/eLife.37429.017

Data availability

RNA sequencing data has been deposited (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6437/).

The following dataset was generated:

Joseph Pickering, author; Matthew Towers, author. RNA sequencing of chicken limb distal mesenchyme cells to compare gene expression over time and in heterochronic grafted cells. 2018 https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6437/ Publicly available at the Electron Microscopy Data Bank (accession no: E-MTAB-6437)

References

  1. Anderson MJ, Schimmang T, Lewandoski M. An FGF3-BMP signaling Axis regulates caudal neural tube closure, neural crest specification and Anterior-Posterior Axis extension. PLoS Genetics. 2016;12:e1006018. doi: 10.1371/journal.pgen.1006018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bénazet JD, Bischofberger M, Tiecke E, Gonçalves A, Martin JF, Zuniga A, Naef F, Zeller R. A self-regulatory system of interlinked signaling feedback loops controls mouse limb patterning. Science. 2009;323:1050–1053. doi: 10.1126/science.1168755. [DOI] [PubMed] [Google Scholar]
  3. Bier E, De Robertis EM. EMBRYO DEVELOPMENT. BMP gradients: a paradigm for morphogen-mediated developmental patterning. Science. 2015;348:aaa5838. doi: 10.1126/science.aaa5838. [DOI] [PubMed] [Google Scholar]
  4. Chambers D, Mason I. Expression of sprouty2 during early development of the chick embryo is coincident with known sites of FGF signalling. Mechanisms of Development. 2000;91:361–364. doi: 10.1016/S0925-4773(99)00288-9. [DOI] [PubMed] [Google Scholar]
  5. Chinnaiya K, Tickle C, Towers M. Sonic hedgehog-expressing cells in the developing limb measure time by an intrinsic cell cycle clock. Nature Communications. 2014;5:4230. doi: 10.1038/ncomms5230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cooper KL, Hu JK, ten Berge D, Fernandez-Teran M, Ros MA, Tabin CJ. Initiation of proximal-distal patterning in the vertebrate limb by signals and growth. Science. 2011;332:1083–1086. doi: 10.1126/science.1199499. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Dessaud E, Yang LL, Hill K, Cox B, Ulloa F, Ribeiro A, Mynett A, Novitch BG, Briscoe J. Interpretation of the sonic hedgehog morphogen gradient by a temporal adaptation mechanism. Nature. 2007;450:717–720. doi: 10.1038/nature06347. [DOI] [PubMed] [Google Scholar]
  8. Dudley AT, Ros MA, Tabin CJ. A re-examination of proximodistal patterning during vertebrate limb development. Nature. 2002;418:539–544. doi: 10.1038/nature00945. [DOI] [PubMed] [Google Scholar]
  9. Eblaghie MC, Lunn JS, Dickinson RJ, Münsterberg AE, Sanz-Ezquerro JJ, Farrell ER, Mathers J, Keyse SM, Storey K, Tickle C. Negative feedback regulation of FGF signaling levels by Pyst1/MKP3 in chick embryos. Current Biology. 2003;13:1009–1018. doi: 10.1016/S0960-9822(03)00381-6. [DOI] [PubMed] [Google Scholar]
  10. Hamburger V, Hamilton HL. A series of normal stages in the development of the chick embryo. Journal of Morphology. 1951;88:49–92. doi: 10.1002/jmor.1050880104. [DOI] [PubMed] [Google Scholar]
  11. Harfe BD, Scherz PJ, Nissim S, Tian H, McMahon AP, Tabin CJ. Evidence for an expansion-based temporal Shh gradient in specifying vertebrate digit identities. Cell. 2004;118:517–528. doi: 10.1016/j.cell.2004.07.024. [DOI] [PubMed] [Google Scholar]
  12. Law CW, Chen Y, Shi W, Smyth GK. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biology. 2014;15:R29. doi: 10.1186/gb-2014-15-2-r29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Liu TM, Lee EH, Lim B, Shyh-Chang N. Concise review: balancing stem cell Self-Renewal and differentiation with PLZF. Stem Cells. 2016;34:277–287. doi: 10.1002/stem.2270. [DOI] [PubMed] [Google Scholar]
  14. Lu P, Yu Y, Perdue Y, Werb Z. The apical ectodermal ridge is a timer for generating distal limb progenitors. Development. 2008;135:1395–1405. doi: 10.1242/dev.018945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Mariani FV, Ahn CP, Martin GR. Genetic evidence that FGFs have an instructive role in limb proximal-distal patterning. Nature. 2008;453:401–405. doi: 10.1038/nature06876. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Mercader N, Leonardo E, Azpiazu N, Serrano A, Morata G, Martínez C, Torres M. Conserved regulation of proximodistal limb axis development by Meis1/Hth. Nature. 1999;402:425–429. doi: 10.1038/46580. [DOI] [PubMed] [Google Scholar]
  17. Mercader N, Leonardo E, Piedra ME, Martínez-A C, Ros MA, Torres M. Opposing RA and FGF signals control proximodistal vertebrate limb development through regulation of Meis genes. Development. 2000;127:3961–3970. doi: 10.1242/dev.127.18.3961. [DOI] [PubMed] [Google Scholar]
  18. Merino R, Rodriguez-Leon J, Macias D, Gañan Y, Economides AN, Hurle JM. The BMP antagonist gremlin regulates outgrowth, chondrogenesis and programmed cell death in the developing limb. Development. 1999;126:5515–5522. doi: 10.1242/dev.126.23.5515. [DOI] [PubMed] [Google Scholar]
  19. Mok GF, Cardenas R, Anderton H, Campbell KH, Sweetman D. Interactions between FGF18 and retinoic acid regulate differentiation of chick embryo limb myoblasts. Developmental Biology. 2014;396:214–223. doi: 10.1016/j.ydbio.2014.10.004. [DOI] [PubMed] [Google Scholar]
  20. Munoz-Sanjuan I, Simandl BK, Fallon JF, Nathans J. Expression of chicken fibroblast growth factor homologous factor (FHF)-1 and of differentially spliced isoforms of FHF-2 during development and involvement of FHF-2 in chicken limb development. Development. 1999;126:409–421. doi: 10.1242/dev.126.2.409. [DOI] [PubMed] [Google Scholar]
  21. Patro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon provides fast and bias-aware quantification of transcript expression. Nature Methods. 2017;14:417–419. doi: 10.1038/nmeth.4197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Pizette S, Abate-Shen C, Niswander L. BMP controls proximodistal outgrowth, via induction of the apical ectodermal ridge, and dorsoventral patterning in the vertebrate limb. Development. 2001;128:4463–4474. doi: 10.1242/dev.128.22.4463. [DOI] [PubMed] [Google Scholar]
  23. Pizette S, Niswander L. BMPs negatively regulate structure and function of the limb apical ectodermal ridge. Development. 1999;126:883–894. doi: 10.1242/dev.126.5.883. [DOI] [PubMed] [Google Scholar]
  24. Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biology. 2010;11:R25. doi: 10.1186/gb-2010-11-3-r25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Roselló-Díez A, Ros MA, Torres M. Diffusible signals, not autonomous mechanisms, determine the main proximodistal limb subdivision. Science. 2011;332:1086–1088. doi: 10.1126/science.1199489. [DOI] [PubMed] [Google Scholar]
  26. Roselló-Díez A, Arques CG, Delgado I, Giovinazzo G, Torres M. Diffusible signals and epigenetic timing cooperate in late proximo-distal limb patterning. Development. 2014;141:1534–1543. doi: 10.1242/dev.106831. [DOI] [PubMed] [Google Scholar]
  27. Saiz-Lopez P, Chinnaiya K, Campa VM, Delgado I, Ros MA, Towers M. An intrinsic timer specifies distal structures of the vertebrate limb. Nature Communications. 2015;6:8108. doi: 10.1038/ncomms9108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Saiz-Lopez P, Chinnaiya K, Towers M, Ros MA. Intrinsic properties of limb bud cells can be differentially reset. Development. 2017;144:479–486. doi: 10.1242/dev.137661. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Saunders JW. The proximo-distal sequence of origin of the parts of the chick wing and the role of the ectoderm. The Journal of Experimental Zoology. 1948;108:363–403. doi: 10.1002/(SICI)1097-010X(19981215)282:6&#x0003c;628::AID-JEZ2&#x0003e;3.0.CO;2-N. [DOI] [PubMed] [Google Scholar]
  30. Scherz PJ, Harfe BD, McMahon AP, Tabin CJ. The limb bud Shh-Fgf feedback loop is terminated by expansion of former ZPA cells. Science. 2004;305:396–399. doi: 10.1126/science.1096966. [DOI] [PubMed] [Google Scholar]
  31. Selever J, Liu W, Lu MF, Behringer RR, Martin JF. Bmp4 in limb bud mesoderm regulates digit pattern by controlling AER development. Developmental Biology. 2004;276:268–279. doi: 10.1016/j.ydbio.2004.08.024. [DOI] [PubMed] [Google Scholar]
  32. Summerbell D, Lewis JH, Wolpert L. Positional information in chick limb morphogenesis. Nature. 1973;244:492–496. doi: 10.1038/244492a0. [DOI] [PubMed] [Google Scholar]
  33. Summerbell D. A quantitative analysis of the effect of excision of the AER from the chick limb-bud. Journal of Embryology and Experimental Morphology. 1974;32:651–660. [PubMed] [Google Scholar]
  34. Tai CI, Ying QL. Gbx2, a LIF/Stat3 target, promotes reprogramming to and retention of the pluripotent ground state. Journal of Cell Science. 2013;126:1093–1098. doi: 10.1242/jcs.118273. [DOI] [PubMed] [Google Scholar]
  35. Towers M, Signolet J, Sherman A, Sang H, Tickle C. Insights into bird wing evolution and digit specification from polarizing region fate maps. Nature Communications. 2011;2:426. doi: 10.1038/ncomms1437. [DOI] [PubMed] [Google Scholar]
  36. Verheyden JM, Sun X. An Fgf/Gremlin inhibitory feedback loop triggers termination of limb bud outgrowth. Nature. 2008;454:638–641. doi: 10.1038/nature07085. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Yang J, Chai L, Fowles TC, Alipio Z, Xu D, Fink LM, Ward DC, Ma Y. Genome-wide analysis reveals Sall4 to be a major regulator of pluripotency in murine-embryonic stem cells. PNAS. 2008;105:19756–19761. doi: 10.1073/pnas.0809321105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Yu K, Ornitz DM. FGF signaling regulates mesenchymal differentiation and skeletal patterning along the limb bud proximodistal Axis. Development. 2008;135:483–491. doi: 10.1242/dev.013268. [DOI] [PubMed] [Google Scholar]
  39. Zhang J, Ratanasirintrawoot S, Chandrasekaran S, Wu Z, Ficarro SB, Yu C, Ross CA, Cacchiarelli D, Xia Q, Seligson M, Shinoda G, Xie W, Cahan P, Wang L, Ng SC, Tintara S, Trapnell C, Onder T, Loh YH, Mikkelsen T, Sliz P, Teitell MA, Asara JM, Marto JA, Li H, Collins JJ, Daley GQ. LIN28 regulates stem cell metabolism and conversion to primed pluripotency. Cell Stem Cell. 2016;19:66–80. doi: 10.1016/j.stem.2016.05.009. [DOI] [PubMed] [Google Scholar]
  40. Zúñiga A, Haramis AP, McMahon AP, Zeller R. Signal relay by BMP antagonism controls the SHH/FGF4 feedback loop in vertebrate limb buds. Nature. 1999;401:598–602. doi: 10.1038/44157. [DOI] [PubMed] [Google Scholar]
  41. Zuzarte-Luis V, Hurle JM. Programmed cell death in the embryonic vertebrate limb. Seminars in Cell & Developmental Biology. 2005;16:261–269. doi: 10.1016/j.semcdb.2004.12.004. [DOI] [PubMed] [Google Scholar]
  42. Zwilling E, Hansborough LA. Interaction between limb bud ectoderm and mesoderm in the chick embryo. III. experiments with polydactylous limbs. Journal of Experimental Zoology. 1956;132:219–239. doi: 10.1002/jez.1401320203. [DOI] [Google Scholar]

Decision letter

Editor: Marianne E Bronner1
Reviewed by: Neil Vargesson2

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: the authors were asked to provide a plan for revisions before the editors issued a final decision. What follows is the editors’ letter requesting such plan.]

Thank you for sending your article entitled "An intrinsic cell cycle timer terminates limb bud outgrowth" for peer review at eLife. Your article is being reviewed by Marianne Bronner as the Reviewing and Senior Editor and three reviewers.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond with an action plan and timetable for the completion of the additional work. We plan to share your responses with the reviewers and then issue a binding recommendation.

Summary:

In this paper, the authors attempt to show that the breakdown of reciprocal extrinsic signaling between the mesenchyme and the overlying epithelium is not required for the termination of proliferative outgrowth of limb. For this purpose, they performed RNA-seq of distal cells from HH24, HH27, HH29 wing buds as well as HH24g (grafts of old mesenchyme cells (HH27) to early HH20 wing buds, which is left until HH24). RNA-seq analyses revealed that, in HH24g, Bmp2/4/7 expression is maintained, whereas FGF and Grem1 expression is reset. Based on these results, authors draw conclusions that the mesenchyme cells terminate growth by an intrinsic cell cycle timer in the presence of e-m signalling. While the results are interesting, many changes are required to bring the manuscript to publishable form. Most critical is to perform further functional experiments to test the role of genes identified by RNA-seq to validate the model. It would also be critical to provide evidence that the permissive signals from the AER is required for the duration of mesenchyme cell proliferation. The major issues raised by the reviewers are listed below. If you think you can make these changes in 2-3 months, we would be willing to entertain a revised version of the manuscript. To this end, it would be very helpful if you could provide us with an outline of how you would address the reviewers' criticisms.

Essential revisions:

1) Three criteria for demonstrating a class of factors (e.g., BMPs) is responsible for an event (e.g., loss of proliferation in HH24g cells). (1) Is it present at the right place and time? Figure 2 addresses this. (2) Can it perform this function? Figure 3 K-L addresses this. (3) Removal of the factor should block the event. An argument can be made that the last criterion is most important. There is no experiment in this study that meets this criterion. Therefore, additional functional data are essential; otherwise this study remains much too inconclusive. Perhaps a Noggin bead, an siRNA-inhibition of Bmpr1A experiment, or a chemical inhibition of BMP signaling such as LDN-193189 can be used to test this idea. I think this is the biggest issue and needs to be addressed before publication.

2) The authors never use the term "autocrine" in regard to BMP signaling in their model. But is it not the case? Please either use the term or explain why it is not appropriate.

3) As the authors point out, it is important to exclude that a loss of AER-FGF is affecting cell cycle in the graft- especially because they find increased pSmad staining in the AER over their grafts. Previous work (Development 144, 479) demonstrates that AER-FGF8 expression persists in such grafts. But it is possible that within the mesenchyme an intrinsic timer regulates its competence to respond to the AER-FGF signal. Is Dusp6 enough to demonstrate these cells receive a normal FGF signal? Different targets likely respond to different signaling thresholds. The authors should also examine Spry2 and Spry4 in their grafts.

4) Given the well-known effect of limb-bud BMP signalling on cell death, the authors should report what cell death occurred in their bead/flow cytometry experiment (Figure 3K-L).

5) The authors need to expound some about why they propose BMP signaling is instructive and FGF signals are permissive. The "young" AER-FGF signal over the graft has no effect on the cell cycle – hence it is permissive. The authors need to provide further evidence that the permissive signals from the AER is required for the duration of mesenchyme cell proliferation. Why is the BMP signal instructive – a dose response experiment, where more BMP has a greater effect, would help clarify the use of this term.

6) In this work, authors showed that, in HH24g, (a) Bmp2/4/7 expression is maintained (as late stage), (b) FGF and Grem1 expression is reset (as early stage) and (c) implanted (not endogenous) BMP represses cell proliferation in limb buds. Based on these data, authors concluded "an intrinsic switch from BMP antagonism to BMP signalling controls the cell cycle program in distal mesenchyme during limb bud outgrowth". There is a wild leap of logic in this conclusion from limited data presented here.

7) Regarding the Bmp implantation experiment, concentration of Bmp or method was not included. Does this experiment reflect the function of endogenous Bmp on cell proliferation in limb buds?

8) The term "intrinsic cell cycle timer" needs more explanation. Did authors mean that cells memorize the positional information as an epigenetic memory, gene expression, protein accumulation, or cell adhesion? If they consider all these possibilities, other conclusions should also be considered.

9) There are no statistical or scientific reasons why authors selected Bmp, FGF, and Grem1 for further analyses. Please justify why you are looking at BMPs, FGFs, Gremlin. If these are the "usual suspects", one could disregard the RNA seq data. If other enriched genes, such as cyclinD1, BAMBI, SPON1, HOXC13, Sall and EphA4, were taken into consideration, their conclusions might have been changed.

From Cluster 1, authors selected Bmp2 and Bmp7. In the text, it was said that "This cluster notably includes genes encoding members of the Bone Morphogenetic Protein (BMP) family-Bmp2 and Bmp7. Genes encoding other known signaling molecules are not present in cluster1". I do not understand why authors did not perform GO, or similar analyses to select candidates. As authors are interested in the mechanisms of intrinsic cell cycle timer, genes related to proliferation or epigenetic modifications should be also considered. Following genes seem to be highly relevant.

CCND1 (cyclin1)

SPON1(spondin1, a heparin-binding extracellular matrix glycoprotein, which can control bone metabolism, via possible disruption of latent TGF-β activation.)

ARID5B (Transcription coactivator that make complex with PHF2, which mediates demethylation of H3K9me2)

CREB5 (Binds DNA as a homodimer or as a heterodimer with JUN or ATF2/CREBP1)

From Cluster 2, authors selected Fgfr2, but again they did not explain the statistical reasons. Cluster 2 contains the following interesting genes, but not selected for further analyses.

BAMBI (BMP and Activin Membrane Bound Inhibitor)

HoxC13 (Controls transcription of genes, involved cell proliferation, such as CCND1)

FBLN1 (a calcium binding ECM and plasma glycoprotein. Required for normal limb morphogenesis)

RDH10 (retinol dehydrogenase 10. Required for normal limb morphogenesis)

LGR5 (GPCR class A receptor. A member of the Wnt signaling pathway. R-spondin proteins are their ligands)

From cluster 3, they selected FGF and Grem1, again without any statistical reasons. Cluster 3 contains other genes known to be involved in setting positional values of limbs or limb morphogenesis, such as Sall1, 3, 4, EphA4 (receptor for ephrin-A2, which regulates position-specific cell affinity in the chicken limb bud), Dlx2 (TF, Required for limb development). Sall1 and Sall3 contribute to the autopod development, by competing with Hoxa13-d13 for a target sequence in the EphA4 upstream region. Such an antagonistic interaction between Hox and Sall in the limb suggested to be contributed to the fine-tuning of local Hox activity leading to proper morphogenesis of cartilage element (i.e. controlling the positional value of the limb along the proximo-disal axis).

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "An intrinsic cell cycle timer terminates limb bud outgrowth" for further consideration at eLife. Your revised article has been favorably evaluated by Marianne Bronner (Senior Editor) and three reviewers.

The manuscript has been improved but there are some issues remaining that need to be addressed before acceptance, as outlined below:

1) The paper ends with "It will be important to identify if this mechanismoperates in other patterning systems." Considering "this mechanism" as BMP signaling that terminates growth, I strongly suggest the authors consider citing Anderson et al., 2016. This paper provides genetic evidence that BMP signals terminate axis extension (Figure 10 is a summary model). The paragraph at the top of page 20 discusses this idea.

2) KEGG analyses

i) It's not clear if KEGG is suitable for identifying pathways in Gallus gallus. According to Supplementary file 2, KEGG recognizes pathways for only 24 genes out of 99 genes (fail to recognize SALL3, GREM1, DLX2, EPHA4 etc.). This fact is not even mentioned in the text. ii) Supplementary file names are still not organized. For example, Supplementary file 2 is named "5159_1_supp_876279_pc44ty.xls", and thus I was not able to tell which file is for HH24g-HH24 pair or HH24g-HH29 pair. iii) Description of KEGG analyses is incomplete in Supplementary files. For example, in Supplementary file 2, KEGG Pathways of NPY2R is described as "Neuroactive ligand-receptor in…". iv) Description of KEGG pathways in Supplementary fileSupplementary file 1, Supplementary file 2 and Figure 1E is not consistent. For example, in Supplementary file 2, KEGG Pathways of KCNQ1 is described as "Adrenergic signaling in cardio…", but in Figure 1E, it was "Adrenergic". This should be explained at least in Figure legend. v) In Figure 1E, HH24g vs. HH29 is on the left side in Up-regulated pathways, but on the right side in Down-regulated pathways. It is confusing. They should be on the same side.

3) BMP and Noggin beads implantation.i) In Figure 3K, control for HH24g+Nog+8h is not appropriate. Controls should be limb buds implanted with PBS/4mM HCl (i.e. vehicle)-soaked beads (i.e. HH24g-PBS/4mM HCl+8h). ii) In Figure 3K, controls for HH20+BMP2low/higH+8h are not appropriate. Controls should be limb buds implanted with PBS/4mM HCl -soaked beads (i.e. HH20-PBS/4mM HCl+8h). iii) In Figure 3L, control is not appropriate. Again, control should be limb buds implanted with PBS/4mM HCl-soaked beads.

4) FGF beads implantation. i) Vehicle for FGF beads implantation is not mentioned in MM. ii) In Figure 4G, control should be limb buds implanted with vehicle-soaked beads. iii) In Figure 4I, again, control should be HH27+vehicle+6h.

eLife. 2018 Sep 3;7:e37429. doi: 10.7554/eLife.37429.022

Author response

[Editors’ notes: the authors’ response after being formally invited to submit a revised submission follows.]

Summary:

In this paper, the authors attempt to show that the breakdown of reciprocal extrinsic signaling between the mesenchyme and the overlying epithelium is not required for the termination of proliferative outgrowth of limb. For this purpose, they performed RNA-seq of distal cells from HH24, HH27, HH29 wing buds as well as HH24g (grafts of old mesenchyme cells (HH27) to early HH20 wing buds, which is left until HH24). RNA-seq analyses revealed that, in HH24g, Bmp2/4/7 expression is maintained, whereas FGF and Grem1 expression is reset. Based on these results, authors draw conclusions that the mesenchyme cells terminate growth by an intrinsic cell cycle timer in the presence of e-m signalling. While the results are interesting, many changes are required to bring the manuscript to publishable form. Most critical is to perform further functional experiments to test the role of genes identified by RNA-seq to validate the model. It would also be critical to provide evidence that the permissive signals from the AER is required for the duration of mesenchyme cell proliferation. The major issues raised by the reviewers are listed below. If you think you can make these changes in 2-3 months, we would be willing to entertain a revised version of the manuscript. To this end, it would be very helpful if you could provide us with an outline of how you would address the reviewers' criticisms.

Essential revisions:

1) Three criteria for demonstrating a class of factors (e.g., BMPs) is responsible for an event (e.g., loss of proliferation in HH24g cells). (1) Is it present at the right place and time? Figure 2 addresses this. (2) Can it perform this function? Figure 3 K-L addresses this. (3) Removal of the factor should block the event. An argument can be made that the last criterion is most important. There is no experiment in this study that meets this criterion. Therefore, additional functional data are essential; otherwise this study remains much too inconclusive. Perhaps a Noggin bead, an siRNA-inhibition of Bmpr1A experiment, or a chemical inhibition of BMP signaling such as LDN-193189 can be used to test this idea. I think this is the biggest issue and needs to be addressed before publication.

We will graft distal mesenchyme cells from HH27 wing buds to HH20 wing buds and implant Noggin-soaked beads into the host to investigate if this enhances cell cycle progression in the graft. Cell cycle profiles will be compared to the contralateral wing bud, to wings with grafts without Noggin, and to wings treated with Noggin alone.

Revision: We have done this experiment (Figure 3K) and found that after 8h, Noggin-soaked beads significantly reverse the proliferative defect in HH24g mesenchyme. This provides strong evidence that BMP signalling is responsible for loss of proliferation in this tissue.

2) The authors never use the term "autocrine" in regard to BMP signaling in their model. But is it not the case? Please either use the term or explain why it is not appropriate.

The autocrine effect of BMP signalling will be considered and discussed.

Revision: We realise now that we used autocrine synonymously with intrinsic. We have now used autocrine in the text where necessary.

3) As the authors point out, it is important to exclude that a loss of AER-FGF is affecting cell cycle in the graft- especially because they find increased pSmad staining in the AER over their grafts. Previous work (Development 144, 479) demonstrates that AER-FGF8 expression persists in such grafts. But it is possible that within the mesenchyme an intrinsic timer regulates its competence to respond to the AER-FGF signal. Is Dusp6 enough to demonstrate these cells receive a normal FGF signal? Different targets likely respond to different signaling thresholds. The authors should also examine Spry2 and Spry4 in their grafts.

Dusp6 is a good read-out of FGF signalling since its expression is regulated by active FGF signalling during development and induced in response to an FGF-soaked bead in the distal mesenchyme (we will show this data). We have looked at the RNA sequencing data and Spry1, 2 and 4 behave like Dusp6. We will analyse the expression of Spry genes in HH24g wings.

Revision: We have now shown that Dusp6 is induced in response to FGF8-soaked beads (Figures 4G and H). We decided to analyse the expression of Spry2 in HH24g wing buds and we now show that this behaves like Dusp6 (Figures 4C and D).

4) Given the well-known effect of limb-bud BMP signalling on cell death, the authors should report what cell death occurred in their bead/flow cytometry experiment (Figure 3K-L).

We will perform apoptosis analyses on HH21 wing buds treated with BMP2-soaked beads for 6 hours.

Revision: We have now done this experiment and shown that BMP2 beads do not induce cell death (Figures 3L and M). The only thing we changed was to implant beads at HH20 for 8h to maintain consistency with the Noggin-bead experiments.

5) The authors need to expound some about why they propose BMP signaling is instructive and FGF signals are permissive. The "young" AER-FGF signal over the graft has no effect on the cell cycle – hence it is permissive. The authors need to provide further evidence that the permissive signals from the AER is required for the duration of mesenchyme cell proliferation. Why is the BMP signal instructive – a dose response experiment, where more BMP has a greater effect, would help clarify the use of this term.

We will perform an experiment to test if BMP-soaked beads have a dose-dependent effect on cell cycle progression. To further test the permissive role of FGF signalling we will maintain FGF signalling using beads at late stages of chick wing development when the apical ridge begins to regress and see if this can enhance cell cycle progression in an instructive manner. The results of experiments 1 and 3 will also contribute to distinguishing between the permissive and instructive effects of FGF signalling.

Revision: We have now shown that BMP2 beads implanted into HH20 wing buds for 8h have a dose-dependent effect on cell cycle progression suggesting an instructive role (Figure 3K). We have also implanted beads soaked in FGF just as proliferation starts to decline rapidly (HH27) and found that this does not increase cell cycle progression (Figure 4I). This suggests that FGFs play a permissive role in proliferation.

6) In this work, authors showed that, in HH24g, (a) Bmp2/4/7 expression is maintained (as late stage), (b) FGF and Grem1 expression is reset (as early stage) and (c) implanted (not endogenous) BMP represses cell proliferation in limb buds. Based on these data, authors concluded "an intrinsic switch from BMP antagonism to BMP signalling controls the cell cycle program in distal mesenchyme during limb bud outgrowth". There is a wild leap of logic in this conclusion from limited data presented here.

Depending on the results of the above experiments we might have to consider revising these statements, which we agree are too strong at present. We will also further discuss our 2015 paper (Saiz-Lopez et al., 2015) in which we showed that cell cycle progression is intrinsically timed in the distal mesenchyme.

Revision: In our previous paper we showed that the cell cycle is intrinsically regulated although it was unclear how it is terminated (Introduction, Results section). Having now shown that BMP signalling/antagonism controls the cell cycle programme over time we are confident with our speculation. However, we will still say ‘provide evidence’ as to not be too unnecessarily definitive.

7) Regarding the Bmp implantation experiment, concentration of Bmp or method was not included. Does this experiment reflect the function of endogenous Bmp on cell proliferation in limb buds?

We will include information about BMP concentration. The result of the Noggin-bead experiment (point 1) should allow us to speculate more on the role of endogenous BMP signalling.

Revision: We have included details of concentration now and we feel that experiment 1 provides strong evidence for the endogenous role of BMP signalling in the limb.

8) The term "intrinsic cell cycle timer" needs more explanation. Did authors mean that cells memorize the positional information as an epigenetic memory, gene expression, protein accumulation, or cell adhesion? If they consider all these possibilities, other conclusions should also be considered.

We use the term “intrinsic timer” to explain that cell cycle parameters are controlled intrinsically in the mesenchyme, but still require permissive signals from the apical ridge. This was a major part of our 2015 paper and we will explain this better in our revision. We do not yet know what the nature of the clock is.

Revision: At this stage we do not feel we have enough data to speculate upon the potential mechanism of the timer.

9) There are no statistical or scientific reasons why authors selected Bmp, FGF, and Grem1 for further analyses. Please justify why you are looking at BMPs, FGFs, Gremlin. If these are the "usual suspects", one could disregard the RNA seq data. If other enriched genes, such as cyclinD1, BAMBI, SPON1, HOXC13, Sall and EphA4, were taken into consideration, their conclusions might have been changed.

From Cluster 1, authors selected Bmp2 and Bmp7. In the text, it was said that "This cluster notably includes genes encoding members of the Bone Morphogenetic Protein (BMP) family-Bmp2 and Bmp7. Genes encoding other known signaling molecules are not present in cluster1". I do not understand why authors did not perform GO, or similar analyses to select candidates. As authors are interested in the mechanisms of intrinsic cell cycle timer, genes related to proliferation or epigenetic modifications should be also considered. Following genes seem to be highly relevant.

CCND1 (cyclin1)

SPON1(spondin1, a heparin-binding extracellular matrix glycoprotein, which can control bone metabolism, via possible disruption of latent TGF-β activation.)

ARID5B (Transcription coactivator that make complex with PHF2, which mediates demethylation of H3K9me2)

CREB5 (Binds DNA as a homodimer or as a heterodimer with JUN or ATF2/CREBP1)

From Cluster 2, authors selected Fgfr2, but again they did not explain the statistical reasons. Cluster 2 contains the following interesting genes, but not selected for further analyses.

BAMBI (BMP and Activin Membrane Bound Inhibitor)

HoxC13 (Controls transcription of genes, involved cell proliferation, such as CCND1)

FBLN1 (a calcium binding ECM and plasma glycoprotein. Required for normal limb morphogenesis)

RDH10 (retinol dehydrogenase 10. Required for normal limb morphogenesis)

LGR5 (GPCR class A receptor. A member of the Wnt signaling pathway. R-spondin proteins are their ligands)

From cluster 3, they selected FGF and Grem1, again without any statistical reasons. Cluster 3 contains other genes known to be involved in setting positional values of limbs or limb morphogenesis, such as Sall1, 3, 4, EphA4 (receptor for ephrin-A2, which regulates position-specific cell affinity in the chicken limb bud), Dlx2 (TF, Required for limb development). Sall1 and Sall3 contribute to the autopod development, by competing with Hoxa13-d13 for a target sequence in the EphA4 upstream region. Such an antagonistic interaction between Hox and Sall in the limb suggested to be contributed to the fine-tuning of local Hox activity leading to proper morphogenesis of cartilage element (i.e. controlling the positional value of the limb along the proximo-disal axis).

We performed the RNA sequencing experiments to gain some general insights into the reversibility and stability of gene expression. When we realised that Grem1 was in a cluster of ‘reset’ genes, our focus turned to the mechanism of limb bud growth termination, as this was an unexpected finding. We are very much interested in looking at further classes of genes – particularly Hox genes and Ephrins – that could help us understand the programme that specifies autopod development. In our previous paper in 2017, we showed that autopod fate is maintained in HH24g cells.

We have performed both GO and KEGG analyses and we should have shown this data. There are four up-regulated pathways in HH24g wings compared to HH24 wings. These are: ECM, focal adhesion, adrenergic signalling, and TGF-β signalling. This is why we focussed on BMPs. Surprisingly the RNA sequencing has not provided any indication of cell cycle behaviour – even during the interval that cell proliferation declines in normal development (HH24 to HH29). We think this is because the activities of key cell cycle regulators, such as cyclins, are generally controlled post-translationally.

Revision: We included the KEGG analyses (Figure 1E) as this provided a clearer and simplified representation of altered pathway behaviour than the GO analyses.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some issues remaining that need to be addressed before acceptance, as outlined below:

1) The paper ends with "It will be important to identify if this mechanismoperates in other patterning systems." Considering "this mechanism" as BMP signaling that terminates growth, I strongly suggest the authors consider citing Anderson et al., 2016. This paper provides genetic evidence that BMP signals terminate axis extension (Figure 10 is a summary model). The paragraph at the top of page 20 discusses this idea.

2) KEGG analyses

i) It's not clear if KEGG is suitable for identifying pathways in Gallus gallus. According to Supplementary file 2, KEGG recognizes pathways for only 24 genes out of 99 genes (fail to recognize SALL3, GREM1, DLX2, EPHA4 etc.). This fact is not even mentioned in the text.

In the chicken genome ~20% of genes belong to verified KEGG pathways. This is slightly less than some model organisms (human, mouse and rat have ~30% annotated), but on par with others (e.g. zebrafish ~20%). Since our KEGG analysis successfully and independently identified the TGFβ and FGF signalling pathways that are the focus of this work, we would prefer to keep the analysis. We also manually inspected the RNA-seq gene lists for other possible candidates involved in maintaining cell cycle progression in HH24g wing buds and we did not find stronger ones than Bmp2 and Bmp7. However, we have mentioned in the Results section that the KEGG analysis is limited.

ii) Supplementary file names are still not organized. For example, Supplementary file 2 is named "5159_1_supp_876279_pc44ty.xls", and thus I was not able to tell which file is for HH24g-HH24 pair or HH24g-HH29 pair.

We apologise for this but we correctly labelled the file names so the upload process must be scrambling them. We have added a note during the upload for this to be corrected if it happens again.

iii) Description of KEGG analyses is incomplete in Supplementary files. For example, in Supplementary file 2, KEGG Pathways of NPY2R is described as "Neuroactive ligand-receptor in…".

We have manually corrected all of the pathway names.

iv) Description of KEGG pathways in Supplementary file 1, Supplementary file 2 and Figure 1E is not consistent. For example, in Supplementary file 2, KEGG Pathways of KCNQ1 is described as "Adrenergic signaling in cardio…", but in Figure 1E, it was "Adrenergic". This should be explained at least in Figure legend.

We have corrected and simplified pathway names, for instance, ‘Adrenergic signalling’ rather than ‘Adrenergic signalling in cardiomyocytes’.

v) In Figure 1E, HH24g vs. HH29 is on the left side in Up-regulated pathways, but on the right side in Down-regulated pathways. It is confusing. They should be on the same side.

We have amended this.

3) BMP and Noggin beads implantation. i) In Figure 3K, control for HH24g+Nog+8h is not appropriate. Controls should be limb buds implanted with PBS/4mM HCl (i.e. vehicle)-soaked beads (i.e. HH24g-PBS/4mM HCl+8h).

Although we labelled this HH24g-Nog+8h the control was an affigel bead soaked in PBS/4mM HCl. For brevity, we have mentioned the full details of the controls in the Materials and methods section.

ii) In Figure 3K, controls for HH20+BMP2low/higH+8h are not appropriate. Controls should be limb buds implanted with PBS/4mM HCl -soaked beads (i.e. HH20-PBS/4mM HCl+8h).

For our flow cytometry experiments the most accurate control is the left wing buds of the same embryos. Over several papers we have shown that proportions of G1-phase cells differ by less that 1% between equivalent tissue in left and right wing buds of the same embryos. However, we agree with the reviewer that it is important to verify if control PBS beads affect the cell cycle. Therefore, we have treated HH20 wing buds with PBS/4mM HCl-soaked beads for 8h and shown that this does not significantly affect cell cycle parameters compared to the equivalent tissue in the contralateral wing bud.

iii) In Figure 3L, control is not appropriate. Again, control should be limb buds implanted with PBS/4mM HCl-soaked beads.

This control was PBS/4mM HCl-soaked beads and we have mentioned this in the Materials and methods section.

4) FGF beads implantation. i) Vehicle for FGF beads implantation is not mentioned in MM.

Heparin beads were used and the recombinant FGF protein was dissolved in PBS.

ii) In Figure 4G, control should be limb buds implanted with vehicle-soaked beads.

We have included this experiment and shown that heparin/PBS beads implanted at HH27 do not up-regulate Dusp6 after 6h.

iii) In Figure 4I, again, control should be HH27+vehicle+6h.

Following on from point 3ii, we have performed an experiment in which we implanted control heparin beads soaked in PBS to HH27 wing buds and shown that this does not significantly affect cell cycle parameters after 6h compared to the equivalent tissue in the contralateral wing bud.

Associated Data

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

    Supplementary Materials

    Figure 3—source data 1. Flow cytometry graphs for cell cycle analyses.
    elife-37429-fig3-data1.pdf (962.7KB, pdf)
    DOI: 10.7554/eLife.37429.008
    Figure 4—source data 1. Flow cytometry graphs for cell cycle analyses.
    elife-37429-fig4-data1.pdf (399.4KB, pdf)
    DOI: 10.7554/eLife.37429.010
    Supplementary file 1. RNA sequencing of HH24g vs.

    HH24 distal mesenchyme 55 genes are differentially expressed between HH24g and HH24 distal mesenchyme. 53 genes are increased in expression in HH24g distal mesenchyme and two genes are decreased in expression.

    elife-37429-supp1.xls (6.7KB, xls)
    DOI: 10.7554/eLife.37429.012
    Supplementary file 2. RNA sequencing of HH24g vs.

    HH29 distal mesenchyme 99 genes are differentially expressed between HH24g and HH29 distal mesenchyme. 55 genes are increased in expression in HH24g distal mesenchyme and 44 genes are decreased in expression.

    elife-37429-supp2.xls (12KB, xls)
    DOI: 10.7554/eLife.37429.013
    Supplementary file 3. Cluster 1 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

    Cluster 1: genes that increase between HH24 and HH29 (black line) and are maintained in HH24g distal mesenchyme.

    elife-37429-supp3.xls (1.5KB, xls)
    DOI: 10.7554/eLife.37429.014
    Supplementary file 4. Cluster 2 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

    Cluster 2: genes that increase between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme.

    elife-37429-supp4.xls (1.3KB, xls)
    DOI: 10.7554/eLife.37429.015
    Supplementary file 5. Cluster 3 Clustering of RNA-seq data across HH24g/HH24 and HH24g/HH29 pairwise contrasts.

    Cluster 3: genes that decrease between HH24 and HH29 (black line) and reset in HH24g distal mesenchyme.

    elife-37429-supp5.xls (1.5KB, xls)
    DOI: 10.7554/eLife.37429.016
    Transparent reporting form
    elife-37429-transrepform.docx (249.1KB, docx)
    DOI: 10.7554/eLife.37429.017

    Data Availability Statement

    RNA sequencing data has been deposited (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6437/).

    The following dataset was generated:

    Joseph Pickering, author; Matthew Towers, author. RNA sequencing of chicken limb distal mesenchyme cells to compare gene expression over time and in heterochronic grafted cells. 2018 https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6437/ Publicly available at the Electron Microscopy Data Bank (accession no: E-MTAB-6437)

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES