1. Developmental Biology
Download icon

An autoregulatory cell cycle timer integrates growth and specification in chick wing digit development

  1. Joseph Pickering
  2. Kavitha Chinnaiya
  3. Matthew Towers  Is a corresponding author
  1. University of Sheffield, United Kingdom
Research Article
  • Cited 0
  • Views 624
  • Annotations
Cite this article as: eLife 2019;8:e47625 doi: 10.7554/eLife.47625

Abstract

A fundamental question is how proliferation and growth are timed during embryogenesis. Although it has been suggested that the cell cycle could be a timer, the underlying mechanisms remain elusive. Here we describe a cell cycle timer that operates in Sonic hedgehog (Shh)-expressing polarising region cells of the chick wing bud. Our data are consistent with Shh signalling stimulating polarising region cell proliferation via Cyclin D2, and then inhibiting proliferation via a Bmp2-p27kip1 pathway. When Shh signalling is blocked, polarising region cells over-proliferate and form an additional digit, which can be prevented by applying Bmp2 or by inhibiting D cyclin activity. In addition, Bmp2 also restores posterior digit identity in the absence of Shh signalling, thus indicating that it specifies antero-posterior (thumb to little finger) positional values. Our results reveal how an autoregulatory cell cycle timer integrates growth and specification and are widely applicable to many tissues.

https://doi.org/10.7554/eLife.47625.001

Introduction

How proliferation and growth are controlled during embryonic development is a difficult question to address because these processes often involve the interaction between intrinsic mechanisms and extrinsic signals, as revealed in early studies, in which limb buds were transplanted between differently-sized salamander species (Twitty and Schwind, 1931). Although much is known about how proliferation and growth are controlled by extrinsic signals, little is known about how these processes are controlled intrinsically. One mechanism involves changing the level of a factor beyond a threshold at which it can stimulate or inhibit cell cycle entry. For instance, the progressive depletion of DNA replication factors triggers the mid-blastula transition in Xenopus embryos after a defined number of cell cycles (Collart et al., 2013). In addition, the increase in the levels of a G1-S phase inhibitor causes cultured adult oligodendrocyte progenitor cells to differentiate after a defined duration of proliferation (Durand and Raff, 2000).

We previously showed that polarising region cells of the chick wing bud measure their duration of proliferation in a similar manner to that of cultured oligodendrocyte progenitor cells (Chinnaiya et al., 2014). The polarising region is an important developmental organiser of the early limb bud that produces the secreted signal, Sonic hedgehog (Shh) (Riddle et al., 1993), which is implicated in providing cells with antero-posterior (thumb to little finger) positional values (Yang et al., 1997; Tickle and Towers, 2017). The grafting of polarising regions from young to older wing buds showed that proliferation rates are intrinsically determined (Chinnaiya et al., 2014). However, both polarising region and oligodendrocyte progenitor cells require permissive signals to maintain them in a proliferative state. In the case of the polarising region, this signal is based on Fgfs (Fibroblast growth factors) emanating from the overlying apical ectodermal ridge—a thickening of the epithelium that rims the distal tip of the wing bud (Niswander et al., 1993; Laufer et al., 1994; Niswander et al., 1994). Similarly, the application of Fgfs or Pdgf (Platelet-derived growth factor) maintains the proliferation of cultured oligodendrocyte progenitor cells (Durand et al., 1997). In addition, both polarising region and oligodendrocyte progenitor cells require an instructive hydrophobic signal. For polarising region cells, this signal is considered to be retinoic acid (Chinnaiya et al., 2014), which emanates from the trunk of the embryo (Mercader et al., 2000). The growth of the wing bud away from the influence of retinoic acid signalling then starts the polarising region cell cycle timer (Chinnaiya et al., 2014). However, in the case of oligodendrocyte progenitor cells, retinoic acid or thyroid hormone is required to stop the cell cycle timer after a defined duration (Barres et al., 1994; Tokumoto et al., 1999). The timing of oligodendrocyte progenitor cell proliferation involves the cell-intrinsic control of p27kip1 (Durand et al., 1997Durand et al., 1998), an inhibitor of the D cyclins, which are the rate-limiting effectors of the G1-S phase transition (Ohtsubo and Roberts, 1993). Although these important in vitro studies on oligodendrocyte progenitor cells provide an exemplar for understanding the temporal control of proliferation, it is unclear if similar mechanisms function during embryogenesis.

In this paper, we reveal that polarising region and oligodendrocyte progenitor cells use similar molecular mechanisms to control their duration of proliferation. We provide evidence that Shh signalling stimulates the proliferation of polarising region cells via Cyclin D2 and then inhibits proliferation via p27kip1. We also demonstrate how this timer restricts posterior digit development in the chick wing. In addition, our data suggest that Bmp2 is an important component of the timer, playing a dual role in preventing digit development, while simultaneously specifying posterior digit identity.

Results

Shh regulates cell cycle gene expression in the polarising region

Previously we demonstrated that Shh signalling intrinsically controls the rate and duration of polarising region cell proliferation (Chinnaiya et al., 2014; Pickering and Towers, 2016). Thus, the transient inhibition of Shh signalling at HH20 (for approximately 72 h) using cyclopamine increases the percentage of polarising region cells in the G1-phase of the cell cycle within 24 h, indicative of a decreased rate of proliferation (Figure 1 – data from Chinnaiya et al., 2014; Pickering and Towers, 2016). This is followed by an acute stimulation in the proliferation rate between 24 and 48 h. During this time, the proliferation of distal mesenchyme cells adjacent to the polarising region is unaffected by the loss of Shh signalling, thus indicating specific effects on polarising region cell proliferation (Pickering and Towers, 2016). The proliferation rate then diminishes in the posterior mesenchyme following the termination of Shh expression, but it is still more rapid than in untreated wings (Figure 1) The over-proliferation of polarising region cells is associated with the formation of an extra posterior digit, so that the normal pattern of digits running in the anterior to posterior sequence of 1-2-3, becomes 1-2-2-2, or 1-2-2-3 (Pickering and Towers, 2016).

Shh signalling influences polarising region proliferation.

In normal development (blue line - HBC carrier control) the percentage of polarising region cells in G1-phase of the cell cycle slightly decreases over 6 hr between HH20 and HH21 and then increases throughout wing outgrowth (Note, Shh expression terminates at HH27/28 and data and statistics taken from Chinnaiya et al., 2014; Pickering and Towers, 2016. In all cases, between 10 and 14 polarising regions were dissected and data were obtained from 5 to 10,000 cells). The inhibition of Shh signalling using cyclopamine/HBC at HH20 (red line) increases the percentage of G1-phase cells over 24 hr until HH24, then there is a sharp decrease over the next 24 hr until HH27 when Shh expression normally terminates. Over the next 24 hr until HH29, G1-phase cell numbers recover in posterior mesenchyme after Shh expression has terminated (unpaired Pearson’s χ2 test **- =< 0.01, and ****- =< 0.0001; see Chinnaiya et al., 2014; Pickering and Towers, 2016).

https://doi.org/10.7554/eLife.47625.002

Our previous data provided insights into how polarising region proliferation could be intrinsically controlled. The gene encoding Cyclin D2—an important effector of the G1-S phase transition—is specifically expressed in the polarising region (Towers et al., 2008) (Figure 2a,b - see expression of Shh in polarising region cells - Figure 2c,d). Shh signalling is required for the expression of Cyclin D2 (Towers et al., 2008) and this could account for the rapid proliferation of polarising region cells in the early wing bud (Figure 1). Therefore, to identify factors intrinsic to the polarising region that could act downstream of Shh signalling to decelerate proliferation at later stages, we determined the expression patterns of the three D cyclin-dependent kinase inhibitors of the Cip/Kip group. Neither expression of p21cip1 (Figure 2e,f) nor p57kip2 (Figure 2g,h) is detectable in the chick wing bud. However, p27kip1 is specifically expressed in the polarising regions of both chick wings and legs over the period that Shh is expressed (Figure 2i–k), although it is undetectable in the posterior mesenchyme at later stages (Figure 2l—note expression in differentiating myogenic cells at HH27/28). In addition, p27kip1 protein specifically localises to the polarising region of HH24 wing buds (Figure 2m).

The polarising region expresses G1-S phase regulators.

Polarising region cells express Cyclin D2 and Shh at HH21/22 (a, c) and at HH25/26 (b, d); but not p21cip1 and p57kip2 (e-h). Polarising region cells express p27kip1 at HH21/22 (i) and HH25/26 (j). p27kip1 is also expressed in the leg (L) polarising region at HH22, as well as in the wing (W) (k), and in differentiating myogenic cells in the wing at HH27/28 (l). p27kip1 protein is present in the polarising region at HH24 (m, n = 3/3). Note, gene expression patterns were observed in all embryos analysed (n => 12). Scale bars: a, c, e, g, i, l – 150 μm; b, d, f, h, j – 300 μm; k – 250 μm; - m – 75 μm.

https://doi.org/10.7554/eLife.47625.003

To examine if Shh signalling is required for Cyclin D2 and p27kip1 expression, we systemically applied cyclopamine or control carrier (HBC) to HH20 embryos. Cyclin D2 expression is undetectable in the polarising region at both 8 and 30 h after cyclopamine is applied (Figure 3a–d— both n = 4/4, note, Shh signalling is attenuated within 4 h; Pickering and Towers, 2016; and see also Towers et al., 2008). However, although expression of p27kip1 is unaffected at 8 h (Figure 3e, fn = 6/6), it is undetectable at 30 h (Figure 3g, hn = 5/5), coinciding with the time that polarising region cells over-proliferate (Figure 1). It should also be noted that cyclopamine does not affect the expression of Cyclin D1 (Towers et al., 2008).

Shh signalling regulates Cyclin D2 and p27kip1 expression.

Application of cyclopamine at HH20 causes loss of Cyclin D2 expression in the polarising region at 8 hr (a, b, n = 4/4 in each experiment) and 30 hr (c, d n = 4/4 in each experiment). Although p27kip1 expression is unaffected at 8 hr (e, f, n = 6/6 in each experiment) it is undetectable at 30 hr (g, h, n = 5/5 in each experiment). Scale bars: a, b, e, f – 150 μm; c, d, g, h – 300 μm.

https://doi.org/10.7554/eLife.47625.004

Taken together, these data show that Shh signalling regulates Cyclin D2 and p27kip1 expression in a manner that reflects the temporal parameters of polarising region cell proliferation (Figure 1). Thus, the requirement of Shh signalling for Cyclin D2 expression correlates with increased proliferation and the requirement for p27kip1 expression correlates with decreased proliferation (Figure 1).

D cyclin inhibition prevents polarising region over-proliferation caused by loss of Shh signalling

In wing buds in which Shh signalling is blocked, the loss of p27kip1 expression could result in the D cyclin-dependent over-proliferation of polarising region cells and additional digit formation (Figure 1). To test this hypothesis, we treated chick embryos with PD0332991—a selective inhibitor of all D cyclin/cyclin-dependent kinase complexes (Toogood et al., 2005; Finn et al., 2009) (note, PD0332991 is marketed as Palbociclib and used to treat breast cancer). In control experiments, the systemic application of PD0332991 to HH20 chick embryos significantly attenuates polarising region proliferation by increasing the number of cells in G1-phase by 31.2% after 6 h (Figure 4a, Figure 4—source data 1A and B, Pearson’s χ2 test, =<0.0001, n=12 polarising regions in each experiment). However, this inhibition is short-lived, as the number of G1-phase cells in PD0332991-treated wings is less than 1% different by 24 h (Figure 4a, Figure 4—source data 1C and D, Pearson’s χ2 test, => 0.05, = 12 polarising regions in each experiment). Note, for all χ2 analyses, the percentage of cells in G1-phase in control and experimental samples were compared to the percentage of cells in S/G2 and M phase.

After validating a dose of PD0332991 that is sufficient to inhibit G1-S phase progression in the chick embryo, we asked if this dose of D cyclin/cyclin dependent kinase inhibitor could prevent the over-proliferation of polarising region cells caused by loss of Shh signalling. We systemically applied PD0332991 to control HBC carrier-treated embryos, and also to cyclopamine-treated embryos when polarising region cells begin to over-proliferate at HH24. In both experiments, this causes a significant increase in the number of G1-phase cells after 6 h: a 24.9% increase in HBC-treated wings and a 26.6% increase in cyclopamine-treated wings (Figure 4b and c, Figure 4—source data 1 - source data 1E–H, Pearson’s χ2 test, =< 0.0001 and = 12 polarising regions in each experiment) and by 24 h at HH27, there is still a significant difference, but the values are much closer: approximately 4% in both cases (Figure 4b and c, Figure 4—source data 1 - source data 1I–L, Pearson’s χ2 test, =< 0.0001 and = 12 polarising regions in each experiment).

These findings show that D cyclin inhibition can prevent the over-proliferation of polarising region cells caused by the loss of Shh signalling.

D cyclin inhibition prevents digit formation caused by loss of Shh signalling

To examine the outcome of transient D cyclin inhibition on wing bud growth, we applied PD0332991 at HH24 and then measured the antero-posterior axis across its widest point after 24 h. In control and cyclopamine-treated wing buds, PD0332991 significantly reduces the width of the antero-posterior axis by 17% and 7%, respectively (Figure 4d–h, unpaired student’s t-test =< 0.0001 and =< 0.05, respectively, = 7 wing buds in each case—the apical ectodermal ridge is marked by the expression of Fgf8). Note, there is no significant difference between control wings and cyclopamine/PD0332991-treated wings (Figure 4d). The application of PD0332991 also reduces the pronounced posterior-distal outgrowth seen in cyclopamine-treated wings, which is associated with the formation of an extra digit (Pickering and Towers, 2016) (compare asterisks in Figure 4g and h).

D cyclin inhibition prevents polarising region over-proliferation.

Application of PD0332991 at HH20 (green line) increases the percentage of polarising region cells in G1-phase to 88.4% after 6 hr compared with 57.2% in DMEM-treated controls (a, blue line - Pearson’s χ2 test - < 0.0001, n = 12 polarising regions in both experiments; 11,745 and 11,310 cells analysed, respectively – note HH20 G1-phase data taken from Chinnaiya et al., 2014; Pickering and Towers, 2016). After 24 hr, G1-phase cell numbers are similar: 65.1% in DMEM-treated polarising regions and 65.9% in PD0332991-treated polarising regions (a - Pearson’s χ2 test – p > 0.05 and are not significantly different, n = 12 polarising regions in both experiments; 13,062 and 11,833 cells analysed, respectively). Application of PD0332991 at HH24 (green line) to control HBC-treated wing buds (blue line) significantly increases the percentage of polarising region cells in G1-phase to 87.8% (11,932 cells analysed) after 6 hr compared with 62.9% in DMEM/HBC-treated controls (b, blue line - Pearson’s χ2 test - p < 0.0001; 5,235 cells analysed n = 12 polarising regions in both experiments – note HH20, HH21 and HH24 G1-phase data taken from Chinnaiya et al., 2014; Pickering and Towers, 2016). After 24 hr, G1-phase cell numbers are still significantly different: 72.2% in DMEM/HBC-treated polarising regions and 68.7% in PD0332991-treated polarising regions (b - Pearson’s χ2 test – < 0.0001, n = 12 polarising regions in both experiments; 11,472 and 10,669 cells analysed, respectively). Application of PD0332991 at HH24 to cyclopamine-treated wing buds (green line – note cyclopamine added at HH20) significantly increases the percentage of polarising region cells in G1-phase to 85.9% after 6 hr compared with 59.3% in cyclopamine-treated wing buds (c, blue line - Pearson’s χ2 test - < 0.0001, n = 12 polarising regions in both experiments; 10,940 and 11,637 cells analysed, respectively – note HH20, HH21 and HH24 G1-phase data taken from Chinnaiya et al., 2014; Pickering and Towers, 2016). After 24 hr, G1-phase cell numbers are still significantly different: 64.1% in cyclopamine-treated polarising regions and 60.1% in cyclopamine/PD0332991-treated polarising regions (c - Pearson’s χ2 test – p < 0.0001, n = 12 polarising regions in both experiments; 11,118 and 10,001 cells analysed, respectively). PD0332991 reduces expansion of the wing bud by 7% compared with DMEM-treated wings, and by 17% compared with cyclopamine-treated wings at 48 hr (d unpaired t test *- = <0.05, and ****- =< 0.0001, n => 7 in all cases). Note loss of posterior overgrowth in PD0332991-treated wing buds (asterisks in g and h). Examples of wing buds from which measurements are taken are shown hybridised for Fgf8 (e-h). Scale bars: 300 μm.

https://doi.org/10.7554/eLife.47625.005

To assess if the transient inhibition of D cyclin activity at HH24 affects digit patterning, we analysed wings at day 10 of incubation (HH36). Both PD0332991-treated and control DMEM-treated wings always have the normal pattern of digits (1-2-3 - Figure 5a, b, n = 12/12 in both cases). We then determined if PD0332991, when applied at HH24 as polarising region cells over-proliferate, affects the pattern of digits in wings treated with cyclopamine at HH20. The wings of embryos treated with only cyclopamine frequently produce an extra posterior digit—either a digit 2 (14% - Figure 5c, f, = 6/42), or a digit 3 (36% - Figure 5d, f, = 15/42, note that this digit arises from the polarising region; Pickering and Towers, 2016). The remaining cyclopamine-treated wings have a normal digit pattern (50%, = 21/42 - Figure 5f). A normal pattern of digits is also found in the wings of approximately half of the embryos (52%) treated systemically with cyclopamine and then PD0332991 (Figure 5fn = 12/23). However, following these treatments, the 1-2-2 digit pattern is found in 40% of wings (Figure 5e, f, = 9/23). The remaining two wings have four digits, suggesting that PD033299 treatment was ineffective (Figure 5fn = 2/23). These results show that 50% of cyclopamine-treated wings have four digits, but only 8% of cyclopamine/PD033299-treated wings have four digits (Figure 5g, χ2 test p =< 0.0001).

Therefore, D cyclin inhibition prevents the formation of an additional digit following the loss of Shh signalling.

D cyclin inhibition prevents digit development caused by loss of Shh signalling.

Application of DMEM (a) and PD0332991 at HH24 (b) does not affect digit patterning (n = 12/12 in both cases). Application of cyclopamine at HH20 results in a 1-2-2-2 digit pattern (c, f - n = 6/42, 14%), a 1-2-2-3 pattern (d, f - n = 15/42, 36%) and a 1-2-3 pattern (f, - n = 21/42, 50%) in day 10 wings. Application of cyclopamine at HH20 and PD0332991 at HH24 results in a 1-2-2 digit pattern (e, f, n = 9/23–40%), a 1-2-3 pattern (f - n = 12/23, 52%) a 1-2-2-2 pattern and 1-2-2-3 pattern (– both n = 1/23, 4%) in day 10 wings. 50% of cyclopamine-treated wings have four digits, 8% of cyclopamine/PD033299-treated wings have four digits (g, χ2 test =< 0.0001). Scale bars: 1 mm.

https://doi.org/10.7554/eLife.47625.007

Bmp2 signalling can restore p27kip1 expression in the polarising region following loss of Shh signalling

To understand why the inhibition of Shh signalling takes 24 h to attenuate p27kip1 expression, we considered the involvement of an intermediate factor. Since Bmp (Bone morphogenetic protein) signalling is known to regulate Cip/Kip transcription (Sharov et al., 2006; Nakamura et al., 2003), we investigated if Bmp2—which is a transcriptional target of Shh signalling in the posterior part of the wing bud (Yang et al., 1997)—could control p27kip1 expression (Note, Bmp2 expression is lost within 8 h of cyclopamine treatment; Scherz et al., 2007). To examine this, we treated HH20 embryos with cyclopamine, and after 16 h, we implanted beads soaked in recombinant Bmp2 protein into the posterior mesenchyme of right-hand wing buds. In control right-hand wing buds treated with PBS-soaked beads for 8 h, p27kip1 expression remains undetectable (Figure 6a – n = 3/3). However, in right-hand wing buds treated with Bmp2-soaked beads, p27kip1 expression is strongly induced in the polarising region, but not in other regions around the bead (Figure 6b – n = 6/6). This result provides evidence that Shh signalling controls p27kip1 expression via Bmp2.

Bmp2 regulates p27kip1 expression and inhibits proliferation.

Application of cyclopamine at HH20 and a PBS-soaked bead to the right-hand wing bud 16 hr later causes loss of p27kip1 expression in the polarising region at 24 hr (a, n = 3/3). However, a Bmp2-soaked bead induces p27kip1 in the polarising region (b, n = 6/6). Application of Bmp2-soaked beads to HH21 wing buds significantly increases the percentage of cells in G1-phase to 74.3% (green line) compared with 64.1% in PBS-soaked bead controls (blue line) at 24 hr (c, Pearson’s χ2 test - p < 0.0001 – note HBC carrier added also, n = 14 polarising regions in both experiments; 7,309 and 9,339 cells analysed, respectively– note HH20, HH21 G1-phase data taken from Chinnaiya et al., 2014; Pickering and Towers, 2016). After 48 hr, the percentages of cells are very similar (75% and 73.6%, respectively) and not significantly different (c, Pearson’s χ2 test – p > 0.05; 8,402 and 6,576 cells analysed, respectively). Application of Bmp2-soaked beads to cyclopamine-treated HH20 wing buds (green line) significantly increases the percentage of G1-phase cells to 78.3% compared with 62.4% in PBS-bead/cyclopamine-treated controls (blue line) at 24 hr, and to 74.3% compared with 65.9% at 48 hr. (d, n = 14 polarising regions in all experiments; 6.924, 9,363, 10,869 and 9,790 cells analysed, respectively - Pearson’s χ2 test - p < 0.0001, – note HH20, HH21 G1-phase data taken from Chinnaiya et al., 2014; Pickering and Towers, 2016). Bmp2-soaked beads reduce expansion of the wing bud by 8.3% compared to PBS bead-treated wings at 24 hr and by 18.9% compared with PBS-soaked bead-cyclopamine-treated wings at 48 hr (e-i, unpaired t test *- p =< 0.05, and ***- p =< 0.0005, n = 7 in all cases). Note reduced posterior/distal growth in Bmp2-treated wing buds (asterisks in h and i). Examples of wing buds from which measurements are taken are shown hybridised for Fgf8 (f-i). Scale bars: a, b – 150 μm, e-h - 300 μm.

https://doi.org/10.7554/eLife.47625.008

Bmp2 prevents polarising region over-proliferation caused by loss of Shh signalling

To examine if Bmp2 can reverse the over-proliferation of polarising region cells following the loss of Shh signalling, we administered cyclopamine at HH20, and after 6h at HH21, we implanted beads soaked in Bmp2 protein into the posterior mesenchyme of right-hand wing buds. In HBC-treated wing buds, Bmp2 protein beads significantly increase the number of polarising region cells in G1-phase of the cell cycle by 10.2% compared with control PBS-bead treated buds at 24 h (Figure 6c, Figure 6—source data 1A and B, Pearson’s χ2 test, p =< 0.0001 and n = 14 polarising regions in each experiment), and by 48 h, there is only a 1.4% change which is not significantly different (Figure 6c, Figure 6—source data 1E and F, Pearson’s Chi-squared testχ2 test, p => 0.05 and n = 14 polarising regions in each experiment). Similarly, in cyclopamine-treated wing buds, Bmp2 protein beads significantly increase the number of polarising region cells in G1-phase of the cell cycle by 15.9% compared with control PBS/cyclopamine-treated buds at 24 h (Figure 6d, Figure 6—source data 1C and D, Pearson’s χ2 test, p =< 0.0001 and n = 14 polarising regions in each experiment), and there is still a significant 8.4% increase by 48 h (Figure 6d, Figure 6—source data 1G and H, Pearson’s Chi-squared testχ2 test, p =< 0.0001 and n = 14 polarising regions in each experiment).

The application of Bmp2 protein also significantly affects the expansion of the wing bud across the antero-posterior axis by 8.3% in control wings buds, and by 18.9% in cyclopamine-treated wing buds (Figure 6e-i, unpaired student’s t-test p =< 0.05 and p =< 0.0001, respectively, n = 7 wing buds in each case—note apical ectodermal ridge is marked by the expression of Fgf8). In addition, Bmp2 prevents the pronounced posterior-distal outgrowth seen in cyclopamine-treated wings, which is associated with the formation of an extra digit (Pickering and Towers, 2016) (compare asterisks in Figure 6h and i).

Taken together, these results suggest that Bmp2 can reverse the effects of the loss of Shh signalling on polarising region cell proliferation.

Bmp2 prevents digit formation caused by loss of Shh signalling

To determine if the application of Bmp2 to cyclopamine-treated wing buds affects digit patterning, we analysed wings at day 10 of incubation. The application of PBS- or Bmp2-soaked beads to the posterior region of control HH21 wing buds does not affect digit pattern (Figure 7a,b, n = 8/8 and n = 10/10, respectively). Since cyclopamine has identical effects on digit patterning in the left and right-hand wings of approximately 90% of the same embryos (Pickering and Towers, 2016), we used this as a control for testing the effects of Bmp2 protein applied to the right-hand wing bud. All of the cyclopamine-treated embryos with three digits in their left-hand wings have a 1-2-3 pattern (n = 14/14), and their contralateral wings treated with Bmp2 protein also have a 1-2-3 pattern (n = 14/14). In addition, out of the cyclopamine-treated embryos with four digits in their left wings, 47% have the digit pattern of 1-2-2-2 (Figure 7c,d, = 8/18) and 53% have the pattern 1-2-2-3 (Figure 7c,en = 10/18). However, all of their contralateral wings in which Bmp2-soaked beads are implanted have three digits, and 94% of these have a digit 3 in a 1-2-3 pattern (Figure 7c,f, = 17/18 – note one wing had a 1-2-2 pattern). Therefore, the incidence of a posterior digit 3 increases from 53% in cyclopamine-treated wings to 94% in cyclopamine/Bmp2-treated wings (Figure 7g, χ2 test p =< 0.0001).

Taken together, these findings reveal that Bmp2 can prevent additional digit formation and can also specify the digit 3 identity in the absence of Shh signalling.

Bmp2 inhibits digit development and specifies digit 3.

Application of HBC/PBS-soaked beads (a) or HBC/Bmp2-soaked beads at HH21 (b) do not affect digit patterning (n = 8/8 and 10/10, respectively). Wings with four digits following cyclopamine-treatment at HH20 have a 1-2-2-2 digit pattern (c, d - 47%, n = 8/18) or a 1-2-2-3 pattern (c, e - 53%, n = 10/18). Application of Bmp2-soaked beads 2 hr later into the right-hand wings buds of the same cyclopamine-treated embryos results in a 1-2-3 digit pattern (c, f - 94%, n = 17/18) or a 1-2-2 pattern (c, 6%, n = 1/18). 53% of cyclopamine-treated wings have a posterior digit 3, but 94% of cyclopamine/Bmp2-treated wings have a posterior digit 3 (g, χ2 test p =< 0.0001). Scale bars: 1 mm.

https://doi.org/10.7554/eLife.47625.010

Discussion

We have shown that Shh signalling controls an intrinsic programme of chick wing polarising region cell proliferation. Our data suggest that this is mediated via antagonistic regulators of the G1-S phase transition: Cyclin D2 and p27kip1. We have provided evidence that Bmp2 signalling functions as part of this timer, both to restrict posterior digit development, and to integrate this process with the specification of antero-posterior positional values.

An autoregulatory cell cycle timer controls polarising region proliferation

We previously showed that Shh signalling is required for Cyclin D2 expression in the chick wing polarising region at early stages, which potentially regulates the number of Shh-expressing cells (Towers et al., 2008) (Figure 8a). In addition, Cyclin D1 is expressed throughout the posterior part of the wing bud, as well as in the polarising region (Towers et al., 2008). Since the D cyclins are rate-limiting for G1-S-phase progression (Ohtsubo and Roberts, 1993), it is likely that both Cyclins D1 and D2 influence polarising region proliferation, but in being restricted to the polarising region, Cyclin D2 plays a fine-tuning role. We have now demonstrated that Shh signalling is also required for the expression of p27kip1 in the chick wing polarising region, and we have provided evidence that this occurs via an intermediate signal—Bmp2 (Figure 8a). The other two members of the Cip/Kip group of inhibitors, p21cip1 and p57kip2, are not expressed in the early chick wing bud, therefore indicating that, unless another inhibitor of the G1-S transition is expressed in the polarising region, p27kip1 is responsible for restraining the proliferation of polarising region cells. We propose that Shh signalling activates the Bmp2-p27kip1 pathway to prevent the over-proliferation of polarising region cells and the formation of an additional digit (Figure 8a,b). The finding that Shh signalling is not required for the maintenance of Cyclin D1 expression (Towers et al., 2008), indicates that the product of this gene is likely to be responsible for the over-proliferation of polarising region cells in the absence of both Cyclin D2 and p27kip1 (Figure 8b). Therefore, the fact that Shh signalling controls the expression of Cyclin D2 and p27kip1 with different temporal dynamics, provides a robust auto-regulatory feedback loop for modulating polarising region proliferation in the chick wing (Figure 8a). Our findings are consistent with many studies demonstrating that Shh signalling can either stimulate proliferation via D cyclins (Kenney and Rowitch, 2000; Komada et al., 2013; Fink et al., 2018; Lobjois et al., 2004; Mill et al., 2005; Seifert et al., 2010) or inhibit proliferation via D cyclin inhibitors (Shkumatava and Neumann, 2005; Fu et al., 2017; Parathath et al., 2010). It could be that Shh signalling fulfils both functions in most tissues, but at different times, and this could be the focus of future endeavours.

An autoregulatory polarising region cell cycle timer.

(a) Shh signalling is required for Cyclin D2 expression in the polarising region (green). Bmp2 signalling acts downstream of Shh to control p27kip1 that inhibits the D-cyclin-dependent formation of a digit from the polarising region (X), and that also promotes the digit 2 to digit 3 positional value in adjacent cells. Shh signalling also stimulates Cyclin D1-dependent proliferation in the digit-forming field, which is required for the progressive specification of antero-posterior digit positional values - 1, 2 and then 3, over 12 hr. (b) Inhibition of Shh signalling during digit specification prevents Bmp2 signalling from promoting the digit 2 to digit 3 positional value in the digit-forming field adjacent to the polarising region. In addition, loss of Bmp2 signalling prevents p27kip1 from inhibiting polarising region proliferation and digit development. Since blocking Shh signalling attenuates Cyclin D2 function, Cyclin D1 is likely to be responsible for the over-proliferation of polarising region cells.

https://doi.org/10.7554/eLife.47625.011

Bmp2 signalling specifies posterior positional values

The chick wing polarising region produces a concentration gradient of Shh signalling that is considered to directly specify adjacent cells with antero-posterior positional values (Riddle et al., 1993). During this process, cells are specified with digit 1 positional values, which are sequentially promoted to digit 2 and then digit 3 positional values over 12 h (Yang et al., 1997; Towers et al., 2011) (Figure 8a). A pattern of anterior digits (1-2-2-2) is often found in wings in which Shh signalling has been blocked during the promotion phase (Pickering and Towers, 2016). We have now provided evidence that digits with posterior identity fail to form in such wings, because the loss of Bmp2 signalling prevents the promotion of the digit 2 to the digit 3 positional value (Figure 8a,b). Therefore, this finding supports the overlooked suggestion that Bmps function downstream of Shh signalling in the specification of antero-posterior values (Drossopoulou et al., 2000). This is an attractive hypothesis because the genes operating downstream of the positional information gradient of Shh signalling in the specification of digit identity have remained elusive (Tickle and Towers, 2017). However, the manipulation of Bmp signalling during hand-plate stages in the chick leg can change digit identity (Dahn and Fallon, 2000). Therefore, we suggest that Bmp signalling encodes antero-posterior positional values at an early stage of chick limb development, the positional memory of which could be interpreted at later stages of development by the second wave of Bmp signalling in the expanded hand-plate (Suzuki et al., 2008).

One of the problems with Bmp2 acting as a long-range signal in digit patterning, is that when it is applied to the anterior margin of the chick wing bud in the form of a recombinant protein, it fails to duplicate the pattern of digits across the antero-posterior axis (Francis et al., 1994). However, this can be explained by the requirement for Shh signalling acting as a growth signal to establish the digit-forming field (Towers et al., 2008) over which Bmps could then operate (Drossopoulou et al., 2000). Another line of evidence against Bmp signalling playing a role in the specification of antero-posterior positional values was the finding that mice lacking Bmp2/4/7 function in their limbs appear to have no overt changes in digit identity (Bandyopadhyay et al., 2006). However, we have suggested that the digits of mouse limb (digits 2-5) have similar ‘anterior’ character in terms of phalanx number and proportions (Pickering and Towers, 2016; Towers, 2018). Note, that the mouse limb has five digits in the anterior to posterior sequence of 1, 2, 3, 4, and 5. One possibility is that the Bmp2-p27kip1 pathway is not as active in mammalian limbs as it is in avian wings, and this could allow the polarising region to produce two digits (Harfe et al., 2004). In addition, diminished Bmp2 signalling might only be sufficient to specify anterior positional values, thus resulting in a cryptic ‘1-2-2-2-2’ digit pattern in mammalian limbs. A model such as this could provide an explanation for the digit patterns obtained when Shh function is removed at different time-points during mouse limb development (Zhu et al., 2008). We have previously suggested that the digit patterns interpreted as 1-2-4 and 1-2-4-5 (Zhu et al., 2008) could instead be ‘1-2-2’ or ‘1-2-2-2’ patterns, because of the failure of Shh signalling to promote sufficient expansion of the digit-forming field to allow four digits to form with a ‘digit 2’ character (Pickering and Towers, 2016; Tickle and Towers, 2017). Therefore, we conclude that Shh functions primarily as a mitogen, and that Bmps function downstream of Shh signalling in specifying antero-posterior positional values in the chick wing.

Implications of the autoregulatory polarising region cell cycle timer

Cell cycle timers have been predicted in several developmental contexts (Newport and Kirschner, 1982; Durand and Raff, 2000; Burton et al., 1999; Primmett et al., 1989; Lewis, 1975). Recently, we provided evidence that a progressive increase in Bmp signalling is responsible for the slowing of the cell cycle (Pickering et al., 2018), which is linked to the acquisition of proximo-distal positional values in the distal mesenchyme of the chick wing bud (humerus to digit tips) (Saiz-Lopez et al., 2015). It is tempting to speculate that this could involve a D cyclin-dependent kinase inhibitor acting downstream of Bmp signalling, although it is unlikely to be p27kip1, which is only expressed in myogenic cells at later stages of chick wing development. However, important work showed that the intrinsic control of p27kip1 is required for the timely differentiation of cultured adult oligodendrocyte progenitor cells after a defined duration of proliferation (Durand et al., 1997; Durand et al., 1998). Our work extends these earlier studies by providing evidence for a role of p27kip1 in cell cycle timing during embryogenesis.

It is of note that mice lacking the function of p27kip1 display postnatal overgrowth, which is consistent with cells in various tissues failing to restrain proliferation (Fero et al., 1996; Kiyokawa et al., 1996; Nakayama et al., 1996). However, human overgrowth syndromes involving p27KIP1 are rare (Grey et al., 2013), and growth syndromes involving loss-of-function mutations in D CYCLINS have not been reported. In addition, neither p27kip1 (Fero et al., 1996; Kiyokawa et al., 1996; Nakayama et al., 1996) or cyclin D2 mice (Huard et al., 1999) have been reported to have limb patterning defects, which could be considered inconsistent with the data presented in this report. This could reflect differences between how chick and mice limbs develop. Alternatively, it is likely that robust mechanisms compensate for the germline deletion of individual cell cycle components. For instance, mice lacking all three D cyclins progress to quite late stages of embryogenesis, because other cyclins normally associated with different cell cycle phases replace their functions (Kozar et al., 2004). On the other hand, a role for CYCLIN D2 has been predicted in human limb development from the analyses of gain-of-function mutations, which might not be readily compensated for by other cell cycle regulators (Mirzaa et al., 2014). Thus, de novo mutations in CYCLIN D2—that produce a stable form of the protein—have been identified in individuals who have a rare overgrowth disorder named Megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome (MPPH) (Mirzaa et al., 2014). Interestingly, individuals with MPPH syndrome frequently produce an extra posterior digit in both their upper and lower limbs, which in light of our data, could suggest that the stable form of CYCLIN D2 has overridden the SHH-BMP2-p27KIP1 pathway (Figure 8a,b). Therefore, we suggest that the patterning and growth defects found in individuals with MPPH, and related overgrowth syndromes (Mirzaa et al., 2013), are caused by the loss of control over intrinsic cell cycle timing mechanisms that operate in multiple tissues.

D cyclins and limb growth

Our earlier work (Towers et al., 2008) and present data suggest that Shh signalling specifies two growth compartments during chick wing development. This growth does not specify digit number per se, which is determined at later stages by processes considered to involve self-organisation (Newman and Frisch, 1979; Oster et al., 1983; Wolpert, 1994). Thus, long-range Shh signalling by the polarising region promotes expansion of the adjacent anterior digit-forming field, likely via Cyclin D1, to provide sufficient tissue for three digits (Towers et al., 2008). Conversely, short-range Shh signalling restricts expansion of the posterior polarising region, likely via Cyclin D2/p27kip1, to prevent it producing a digit. It is unclear how these two growth outputs of Shh signalling are mediated simultaneously, but it is likely to involve the spatial distribution of the Gli transcription factors: Gli1 is expressed in posterior cells, whereas Gli2 and Gli3 are expressed more-anteriorly (Marigo et al., 1996; Schweitzer et al., 2000). The anterior growth response to Shh signalling appears evolutionarily conserved in vertebrates because three digits develop from cells adjacent to the polarising region in the chick wing (Towers et al., 2011), chick leg (Towers et al., 2011) and mouse limb (Harfe et al., 2004). However, the ability of the chick leg and mouse limb polarising region to produce more posterior digits than the chick wing, suggests a high degree of growth plasticity, which could be a consequence of the altered dynamics of the auto-regulatory polarising region cell cycle timer.

Similarities and differences can be drawn between chick wing development and Drosophila wing development: Hedgehog signalling emanates from the posterior compartment of the Drosophila wing to promote growth of the anterior compartment via the Gli orthologue, Cubitus interruptus (Ci) and its regulation of Cyclin D (Duman-Scheel et al., 2002). However, cells in the posterior compartment are unable to respond to Hedgehog signalling because they do not express Ci (Lecuit et al., 1996). Instead, Hh signals to the posterior compartment through the Bmp2 orthologue, Decapentaplegic, which is expressed at the antero-posterior compartment boundary (Lecuit et al., 1996). As well as having only one Gli transcription factor (Ci) (Orenic et al., 1990), Drosophila has one D cyclin (Finley et al., 1996) and one D cyclin inhibitor (Dacapo) (Lane et al., 1996). Thus, by extending these gene families, vertebrates appear to have enabled Shh signalling to elicit multiple growth responses in both a spatial and temporal manner.

Materials and methods

Chick husbandry

Request a detailed protocol

Bovans Brown chicken eggs were incubated and staged according to Hamburger Hamilton (Hamburger and Hamilton, 1951) HH19 is incubation day 3, HH24 - day 4, HH27 - day 5, HH29 - day 6, HH30 - day 7, and HH36 is day 10.

Whole mount in situ hybridisation

Request a detailed protocol

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 mins (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

Request a detailed protocol

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 mins 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 (p27Kip1 1:500 Cell Signalling Technology – 2552) 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 mins 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 mins and mounted in Vectashield/DAPI medium. Slides were left at 4°C and imaged the following day.

Flow cytometry

Request a detailed protocol

Polarising regions were dissected in ice cold PBS under a LeicaMZ16F microscope using a fine surgical knife, pooled from replicate experiments (12-15), and digested into single cell suspensions with trypsin (0.5%, Gibco) for 30 mins 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 mins, 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 (5,000–12,000 cells in each case). Statistical significance of numbers of cells in different phases of the cell cycle (G1 vs. S, G2 and M) between pools of dissected wing bud polarising region tissue (12–15 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 wing buds of the same and different embryos showed that less than 2% difference in percentages of G1 phase cells).

Shh signalling and D cyclin inhibition

Request a detailed protocol

Cyclopamine (Sigma) was suspended in control carrier (45% 2-hydropropyl-β-cyclodextrin in PBS, Sigma, to a concentration of 1 μg/μl) and 4 μl pipetted directly onto embryos over the limb bud, after removal of vitelline membranes. PD0332991 was resuspended in DMEM at a concentration of and 0.1 μg/μl and 20 μl pipetted after removal of vitelline membranes. Note in all cases, control wings were treated with 2-hydropropyl-β-cyclodextrin (HBC) or DMEM only.

Alcian blue skeletal preparations

Request a detailed protocol

Embryos fixed in 90% ethanol for 2 days then transferred to 0.1% alcian blue in 80% ethanol/20% acetic acid for 1 day, before being cleared in 1% KOH.

Bead implantation

Request a detailed protocol

Control affigel blue beads (Biorad) were soaked in PBS/4 mM HCl. Affigel beads were soaked in human Bmp2 protein (0.05 μg/μl1 - R and D) dissolved in PBS/4 mM HCl. Beads were soaked for 2 hr and implanted into posterior mesenchyme using a sharp needle.

References

  1. 1
  2. 2
    A novel role for Thyroid-Hormone, glucocorticoids and retinoic acid in timing oligodendrocyte development
    1. BA Barres
    2. MA Lazar
    3. MC Raff
    (1994)
    Development 120:1097–1108.
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
    A model for anteroposterior patterning of the vertebrate limb based on sequential long- and short-range shh signalling and bmp signalling
    1. G Drossopoulou
    2. KE Lewis
    3. JJ Sanz-Ezquerro
    4. N Nikbakht
    5. AP McMahon
    6. C Hofmann
    7. C Tickle
    (2000)
    Development 127:1337–1348.
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
    Bone morphogenetic proteins and a signalling pathway that controls patterning in the developing chick limb
    1. PH Francis
    2. MK Richardson
    3. PM Brickell
    4. C Tickle
    (1994)
    Development 120:209–218.
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
    Cerebellar histogenesis is disturbed in mice lacking cyclin D2
    1. JM Huard
    2. CC Forster
    3. ML Carter
    4. P Sicinski
    5. ME Ross
    (1999)
    Development 126:1927–1935.
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
    Fate maps and the pattern of cell division: a calculation for the chick wing-bud
    1. JH Lewis
    (1975)
    Journal of Embryology and Experimental Morphology 33:419–434.
  30. 30
  31. 31
  32. 32
    Opposing RA and FGF signals control proximodistal vertebrate limb development through regulation of meis genes
    1. N Mercader
    2. E Leonardo
    3. ME Piedra
    4. C Martínez-A
    5. MA Ros
    6. M Torres
    (2000)
    Development 127:3961–3970.
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
    Mechanical aspects of mesenchymal morphogenesis
    1. GF Oster
    2. JD Murray
    3. AK Harris
    (1983)
    Journal of Embryology and Experimental Morphology 78:83–125.
  45. 45
  46. 46
  47. 47
  48. 48
    Periodic segmental anomalies induced by heat shock in the chick embryo are associated with the cell cycle
    1. DR Primmett
    2. WE Norris
    3. GJ Carlson
    4. RJ Keynes
    5. CD Stern
    (1989)
    Development 105:119–130.
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
    Relationship between dose, distance and time in sonic Hedgehog-mediated regulation of anteroposterior polarity in the chick limb
    1. Y Yang
    2. G Drossopoulou
    3. PT Chuang
    4. D Duprez
    5. E Marti
    6. D Bumcrot
    7. N Vargesson
    8. J Clarke
    9. L Niswander
    10. A McMahon
    11. C Tickle
    (1997)
    Development 124:4393–4404.
  66. 66

Decision letter

  1. Marianne E Bronner
    Senior and Reviewing Editor; California Institute of Technology, United States

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.

Thank you for sending your article entitled "A Sonic hedgehog-Bmp2-p27 kip1 timer controls chick wing digit identity and number" for peer review at eLife. Your article has been evaluated by Marianne Bronner as the Senior Editor and Reviewing Editor, and two reviewers.

Below is summary of the required revisions. In addition, we include the full reviews for your reference.

Essential revisions:

1) The Introduction and the discussion should be substantially expanded. There is a missed opportunity to discuss what is known about the function of Shh to shorten the G1 phase and to accelerate cell proliferation, not just in limbs, but also in other tissues beyond the oligodendrocyte cell culture system that they cite as inspiration for this work. Given the fact that eLife has no limitation on word count, the authors could place these findings in a much broader context that could better illustrate why these results represent a substantial advance.

2) Please refrain from over-interpretation. For example, as cyclin/cdk inhibitor arrests cells in G1 with or without cyclopamine, please change the last sentence of that section to a softer interpretation (e.g. "Overproliferation in response to loss of Shh signaling is cyclin/cdk-dependent."). There are similar problems with their interpretations in multiple places like stating that Shh prevents cyclin-dependent phenotypes rather than cyclin inhibition reverses the effects of loss of Shh?

3) There should be statistical analysis of digit number. Since their statement is "these findings demonstrate that Shh signaling prevents the D cyclin-dependent formation of an additional digit", it should be possible to determine whether 21/42 with four digits (regardless of identity) in cyclopamine treated embryos is different from 2/23 with four digits with cyclopamine and PD0332991.

4) Please present the statistics more clearly throughout the text rather than in figure legends and methods. For p values, please clarify how these relate to specific time points and clarify reproducibility of the experiments. For all experimental manipulations, please report the actual numbers of individuals in the affected and unaffected groups.

5) The PD0332991 exposure was systemic, so there wasn't a way to say which ones should have had a cyclopamine phenotype to "rescue". For this reason it makes sense to include a "1-2-3" category. Contrast that with the BMP2 bead implantation experiment that is very local and one-sided, with the contralateral side for comparison, so it makes sense that the 1-2-3 category was removed from the analysis. However, the number of individuals unaffected by cyclopamine should be reported.

6) They provide evidence that Bmp2 is sufficient to inhibit proliferation after cyclopamine treatment and that Bmp2 induces p27 after cyclopamine treatment, but the argument that p27 inhibits proliferation in this context is indirect. Is there another way to halt proliferation in the limb in absence of p27? Even if p27 is sufficient, it may not be necessary. Please rewrite the manuscript to provide a more cautious set of conclusions and acknowledge the various interpretations that cannot be excluded by the data.

Reviewer #1:

Here, Pickering and colleagues present the next chapter in a thoughtful and comprehensive series of manuscripts that address longstanding questions about limb outgrowth and its integration with the anterior-posterior specification of digits. Specifically, they build on their 2016 Development paper wherein they demonstrated a surprising role for Shh signaling to restrict distal limb expansion; late exposure to cyclopamine causes digit bifurcations and polydactyly, paradoxical to the early requirement for Shh to expand the limb field. In this manuscript submitted to eLife, they take inspiration from in vitro studies of a cell cycle timer in cultured oligodendrocytes to test the hypothesis that a Cyclin D inhibitor, p27kip1, acts downstream of Shh to inhibit Cyclin D-dependent cell cycle progression. They end with a model that ties multiple works together with these new data to suggest Shh initially expands the limb field by activating Cyclin D while at the same time inducing Bmp2 expression. Bmp2 in turn activates p21Kip to 'put the brakes' on distal tissue expansion to establish the appropriate number, and identity, of resulting digits.

The authors first reconfirm a previous observation that Cyclin D2 is expressed in the polarizing region and show that its expression is dependent on Shh signaling. They then show p21Kip1 is also expressed in the polarizing region and downregulated by cylopamine, albeit with a temporal delay relative to Cyclin D2. They next test the hypothesis that CyclinD/cdk activity is required for the increased proliferation after late cyclopamine exposure by showing that a cyclin/cdk inhibitor is sufficient to stall cells in G1 and reduce distal limb width to near control in the presence of cyclopamine. The authors then carry these experiments out longer to test the hypothesis that Shh inhibition of cyclin/cdk prevents formation of extra digits. Indeed, inhibition of cyclin/cdk together with shh inhibition frequently reduces the number of digits from four to three.

Last, the authors investigate the mechanism that results in a delayed responsiveness of p21Kip1 downregulation after cyclopamine exposure. Inspired by the fact that Bmps are known to regulate Cip/Kip expression, and the responsiveness of Bmp2 to cyclopamine, they test the hypothesis that restoring Bmp2 expression by bead implantation after cyclopamine treatment will prevent the loss of p21Kip1 expression. I think that this simple and elegant experiment is the greatest strength of the paper. It allows the authors to present a more complete model of early cyclin D2 activation followed by p21Kip1 expression that is delayed by dependence on the intermediate upregulation of Bmp2.

My major concerns are minimal as I think these data are complete and compelling. However, a few corrections are essential.

1) The final paragraph of the Introduction is abruptly short and should be expanded a little to set the stage for the rest of the manuscript. This is a stylistic criticism.

2) The graphed results of FACS in Figure 1, Figure 4A and B, and Figure 6C have no standard deviations per time point and no indication of how many times these experiments were repeated to show reproducibility.

3) Please calculate and report the p value for Cyc+PD versus DMEM in Figure 4C, since I suspect there is no difference.

Reviewer #2:

Description of this manuscript:

In this manuscript, Pickering and colleagues report that Sonic hedgehog (Shh) stimulates proliferation of cells in the polarising region of the chick wing bud by controlling the cell cycle gene Cyclin D2, and at later stages Shh inhibits overproliferation by activating Bmp2-p27kip1. In a series of experimental manipulations of chick wing buds, they show that blocking hedgehog signalling causes polarising region cells to over-proliferate, leading to polydactyly (extra digits), and that this affect can be prevented by either application of Bmp2 protein or by inhibition of cyclin D. Inhibition of Shh by cyclopamine results in loss of Cyclin D2 within 8 hours and p27kip1 within 30 hours, Application of Bmp2 blocks cyclopamine-induced over-proliferation and extra digit formation, and this is associated with a rescue of p27kip1 expression.

Summary:

Overall, I found this to be an interesting study that takes advantage of some unique advantages of the chick limb system and builds on the previous work from the Towers lab. However, there are several major weaknesses, including over-interpretation of the results in several places, holes in the experimental logic, lack of statistical support for conclusions, and deficiencies in scholarship. These issues prevent me from recommending publication in eLife. There are also some concerns that this study is an incremental advance over previous work, and therefore it may be of interest to a specialized but not a general readership.

Essential revisions:

1) Previous work, some published more than a decade ago, showed that disruption of Shh signaling causes an increase in the percentage of cells in G1, and that Shh directly regulates entry of cells into S-phase of the cell cycle. Surprisingly, the authors cite only their own studies from 2016. It would enhance the paper and improve the scholarship if the authors could acknowledge previous studies and discuss how their results fit within the context of this previous work. Some of these papers have addressed very similar questions to those examined in this manuscript, both in the limb and in other developing systems. Examples of such publications are Zhu et al., 2008; Seifert et al., 2018; Komada et al., 2013; Fink et al., 2018.

2) They show that application of PD0332991, a D cyclin/cdk inhibitor, increases the percentage of G1 cells in control and cyclopamine-treated limbs. In the Results section, they conclude "These findings suggest that Shh signalling prevents the D cyclin-dependent over-proliferation of polarising region cells." This is an over-interpretation of the results. All that this experiment shows is that the PD0332991 inhibits cell cycle progression, which is the known function of the drug.

3) Results section: The experiments in which PD0332991 and cyclopamine were applied in succession resulted in 1-2-2 digit patterns in 40% of the wings. Normal digit pattern is 1-2-3. 5. Treatment with cyclopamine alone produces either an extra digit 2 (14%), an extra digit 3 (36%), or a normal digit pattern (50% – Figure 5F). The variable effects of cyclopamine alone, and the frequency of digit pattern changes under each condition, calls out for a statistical analysis to determine the significance of the effect. In their interpretation of these results, they state, "these findings demonstrate that Shh signalling prevents the D cyclin-dependent formation of an additional digit". This is an overly strong conclusion based on an overinterpretation of some rather indirect evidence.

4) Results section. "The percentage of polarising region cells in G1-phase of the cell cycle increases to 74.3% in Bmp2-treated wing buds compared with 64.1% in control PBS bead treated buds at 24 h". Is this a significant change? Throughout the paper, the absence of statistical analysis underlines the results.

5) In the cyclopamine + Bmp2 experiments (Results section), it is stated, "of the cyclopamine-treated wings with four digits in their left wings, 47% have the digit pattern of 1-2-2-2…and 53% have the pattern 1-2-2-3…" The problem here is that they report percentages of the…treated wings that developed four digits. How many of the treated embryos developed four digits? It is critical to know the size/percentage/number of individuals in the affected and unaffected groups in order to interpret the distribution of digit patterns within the affected group.

6) If p27kip1 is integral to the limb polarising region timer, then one would expect deletion of p27kip1 enhance to perturb limb patterning. However, this experiment was done in 1996 by Fero et al., Kiyokawa et al., and Nakayama et al., but there were no reported effects on the limbs. How does the finding that p27kip1 is dispensable for limb development affect the conclusion (Discussion section) that "a cell cycle timer involving p27kip1 also operates in vivo to control tissue growth and patterning of a major signalling centre during embryogenesis"?

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

Thank you for submitting your article "An autoregulatory cell cycle timer integrates growth and specification in chick wing digit development" for consideration by eLife. Your article has been reviewed by Marianne Bronner as the Senior Editor and Reviewing Editor, and two reviewers. The reviewers have opted to remain anonymous.

While the manuscript has been much improved, reviewer 2 continues to worry about statistical analysis. I have included the full reviews and ask you to fix these remaining issues. Although eLife does not normally allow for multiple rounds of review, I am willing to make an exception in this case since I think you can deal with this remaining concerns in a relatively short period of time without further experimentation but with better analysis. However, no further rounds of revision will be possible, so I urge you to address in detail each of the points raised by the reviewer below.

Reviewer #1:

The resubmitted manuscript by Pickering and colleagues is improved by an expanded Introduction and Discussion section. I also appreciate the revised summary statements of the conclusions that explicitly state the results without over-interpreting the data and the more rigorous presentation of data.

Reviewer #2:

1) In the review of the original submission, I noted that this advance was incremental relative to their previous (2014 and 2016) papers on this topic, but my comment was not addressed during the open comment period or in the editor's letter to the authors. Therefore, in fairness to the authors, I considered that point to be moot when reviewing the revised version of the paper and my comments are restricted to the data and interpretation. No further action is required.

2) I have reconsidered my comments on Figure 5. I was critical of the analysis of the variable digit phenotypes that were obtained in the two treatments and I suggested that they examine the statistical significance of the variation. Although the authors report the differences in specific digit patterns, their conclusions are based entirely on whether the limbs have 3 digits or 4 digits; the variation in digit patterns was ignored. I accept that they have chosen to bin the data this way, and the effect of PD0332991 treatment on digit number is clear. If the range of variation produced by cyclopamine is not considered in the interpretation of results, then there is no reason to perform the statistical test that I suggested in my previous reviews. I view this as the authors' prerogative. No further action is required.

3) In Figure 7, the use of 3 graphs to display the results of Bmp treatment is unnecessary. The number of individuals unaffected by cycloplamine (panel C) can be stated in the text and the graph eliminated. To make it more clear that panel h has broken-out the 1223 and 123 categories of panel D, I suggest using the same colors.

4) For the experiments that compare the number of cells in G1 between two treatment groups (such as cyclopamine+Bmp2 vs. cyclopamine+PBS) in Figure 1, Figure 4 and Figure 6, I suggested that they use the Student's t-test, which is the appropriate test to compare a quantitative, single dependent variable (the number of cells in G1) between 2 groups. Instead, the authors used a Pearson's chi-squared test and argued that carrying out a t-test would require additional experiments.

I found it difficult to understand how running a different statistical test on the same data set could require additional experiments. However, after reading this paper several times, I now suspect that the central problem concerns a disparity between the description of the data in the results and the description of the data in the methods. Throughout the text, they refer to changes in "the number of G1-phase cells" or "the number of cells in G1". This implies that they compared the numbers of cells in G1 between groups. However, the methods section describes the use of DNA content to assign cells to G1, S, or G2/M, and these categorical assignments were expressed as a percentage of the total cell number (Materials and methods section). Therefore, they are actually comparing between 2 groups the percentage of cells in 1 of the 3 categories, which is not the same as comparing the number of cells in G1 in 2 treatment groups. In other words, it appears that they are comparing categorical variables rather than a quantitative variable. If this is the case, then they can probably fix the problem by providing a more accurate description of the data in the Results section.

5) Finally, although I have offered recommendations on the appropriate statistical tests, I believe that reviewers should not dictate how authors conduct their studies. The authors have my comments and the reasoning behind my advice, so I will now leave it to the authors to proceed as they see fit.

https://doi.org/10.7554/eLife.47625.014

Author response

Essential revisions:

1) The Introduction and the discussion should be substantially expanded. There is a missed opportunity to discuss what is known about the function of Shh to shorten the G1 phase and to accelerate cell proliferation, not just in limbs, but also in other tissues beyond the oligodendrocyte cell culture system that they cite as inspiration for this work. Given the fact that eLife has no limitation on word count, the authors could place these findings in a much broader context that could better illustrate why these results represent a substantial advance.

We overlooked the fact there are no limits on word count and we will expand our Introduction and Discussion section.

2) Please refrain from over-interpretation. For example, as cyclin/cdk inhibitor arrests cells in G1 with or without cyclopamine, please change the last sentence of that section to a softer interpretation (e.g. "Overproliferation in response to loss of Shh signaling is cyclin/cdk-dependent."). There are similar problems with their interpretations in multiple places like stating that Shh prevents cyclin-dependent phenotypes rather than cyclin inhibition reverses the effects of loss of Shh?

We have strived to amend incidences flagged up by the reviewers (and elsewhere) where we have over-interpreted our data.

3) There should be statistical analysis of digit number. Since their statement is "these findings demonstrate that Shh signaling prevents the D cyclin-dependent formation of an additional digit", it should be possible to determine whether 21/42 with four digits (regardless of identity) in cyclopamine treated embryos is different from 2/23 with four digits with cyclopamine and PD0332991.

We have performed chi-squared tests showing that this result is significant. We also performed chi-squared tests for the BMP2 rescue of the cyclopamine-treated wings and this shows that the most-posterior digit 3 identity is a significant result.

4) Please present the statistics more clearly throughout the text rather than in figure legends and methods. For p values, please clarify how these relate to specific time points and clarify reproducibility of the experiments. For all experimental manipulations, please report the actual numbers of individuals in the affected and unaffected groups.

We have documented the p-values for each time-point in both the text and the figure panels and included n-numbers.

5) The PD0332991 exposure was systemic, so there wasn't a way to say which ones should have had a cyclopamine phenotype to "rescue". For this reason it makes sense to include a "1-2-3" category. Contrast that with the BMP2 bead implantation experiment that is very local and one-sided, with the contralateral side for comparison, so it makes sense that the 1-2-3 category was removed from the analysis. However, the number of individuals unaffected by cyclopamine should be reported.

We have included this data for cyclopamine-treated left-hand wings and also the digit patterns in the contralateral BMP-treated wings.

6) They provide evidence that Bmp2 is sufficient to inhibit proliferation after cyclopamine treatment and that Bmp2 induces p27 after cyclopamine treatment, but the argument that p27 inhibits proliferation in this context is indirect. Is there another way to halt proliferation in the limb in absence of p27? Even if p27 is sufficient, it may not be necessary. Please rewrite the manuscript to provide a more cautious set of conclusions and acknowledge the various interpretations that cannot be excluded by the data.

This is a good point and we have strived to avoid over-interpreting our findings.

Reviewer #1:

[…] My major concerns are minimal as I think these data are complete and compelling. However, a few corrections are essential.

1) The final paragraph of the Introduction is abruptly short and should be expanded a little to set the stage for the rest of the manuscript. This is a stylistic criticism.

We have amended this.

2) The graphed results of FACS in Figure 1, Figure 4A and B, and Figure 6C have no standard deviations per time point and no indication of how many times these experiments were repeated to show reproducibility.

Figure 1 represents previously published experiments performed in the same manner as the ones in our current paper (Chinnaiya, Tickle and Towers, 2014, Pickering and Towers, 2016). The 0-24 hour time-points in Figure 4B and 0-6 hour time-points in Figures6C also show some of this previous data. This data is not essential for these figures, but we felt that it would help the reader understand the experiments. All other time-points show new data. We have explained this in the revised paper.

In our flow cytometry experiments we analyse between 5,000 and 12,000 cells from 12-15 polarising regions (we need to pool this many polarising regions to get enough cells to produce accurate counts). Therefore, this ensures that our results are based on many replicas of the same experimental manipulation and the large number of cells counted provides very accurate data. For instance, we have shown that G1-phase numbers between left and right wing buds of the same embryos differ by less than 1% (Chinnaiya et al., 2014). In addition, when we carefully stage-match tissue from separate embryos incubated at the same time, we also achieve similar values (Chinnaiya et al., 2014). Thus, when we perform Pearson’s chi-squared analyses, a significantly different result generally equates to a >1-2% difference in G1-phase numbers between the control and experimental tissue (Chinnaiya, Tickle and Towers, 2014). The differences in G1 numbers in the key experiments in this paper range between 10% and 32%, which represent extremely significant changes. In our revised paper we have documented the p-values for each time-point in both the text and the figure panels and added n numbers for the cells and polarising regions analysed.

3) Please calculate and report the p value for Cyc+PD versus DMEM in Figure 4C, since I suspect there is no difference.

We have done this.

Reviewer #2:

[…] Overall, I found this to be an interesting study that takes advantage of some unique advantages of the chick limb system and builds on the previous work from the Towers lab. However, there are several major weaknesses, including over-interpretation of the results in several places, holes in the experimental logic, lack of statistical support for conclusions, and deficiencies in scholarship. These issues prevent me from recommending publication in eLife. There are also some concerns that this study is an incremental advance over previous work, and therefore it may be of interest to a specialized but not a general readership.

Essential revisions:

1) Previous work, some published more than a decade ago, showed that disruption of Shh signaling causes an increase in the percentage of cells in G1, and that Shh directly regulates entry of cells into S-phase of the cell cycle. Surprisingly, the authors cite only their own studies from 2016. It would enhance the paper and improve the scholarship if the authors could acknowledge previous studies and discuss how their results fit within the context of this previous work. Some of these papers have addressed very similar questions to those examined in this manuscript, both in the limb and in other developing systems. Examples of such publications are Zhu et al., 2008; Seifert et al., 2018; Komada et al., 2013; Fink et al., 2018.

We have broadened our review of the literature and we thank the reviewer for these helpful suggestions. It is widely regarded that Shh is a mitogen and it was difficult to decide which studies to concentrate on. It was also known previously that Shh could inhibit proliferation. We did not intend to focus on our previous work but much of it created the premise of the current paper. However, we have now cited papers that describe links between Shh and D cyclins and their inhibitors. In terms of discussion related to these points, we have suggested that it is important to analyse Shh signalling over time, as it is likely to play both anti-proliferative and proliferative roles in may tissues.

2) They show that application of PD0332991, a D cyclin/cdk inhibitor, increases the percentage of G1 cells in control and cyclopamine-treated limbs. In the Results section, they conclude "These findings suggest that Shh signalling prevents the D cyclin-dependent over-proliferation of polarising region cells." This is an over-interpretation of the results. All that this experiment shows is that the PD0332991 inhibits cell cycle progression, which is the known function of the drug.

This is a good point that we have now amended the Results section. ‘These findings show that D cyclin inhibition can prevent the over-proliferation of polarising region cells caused by the loss of Shh signalling.’

3) Results section: The experiments in which PD0332991 and cyclopamine were applied in succession resulted in 1-2-2 digit patterns in 40% of the wings. Normal digit pattern is 1-2-3. 5. Treatment with cyclopamine alone produces either an extra digit 2 (14%), an extra digit 3 (36%), or a normal digit pattern (50% – Figure 5F). The variable effects of cyclopamine alone, and the frequency of digit pattern changes under each condition, calls out for a statistical analysis to determine the significance of the effect. In their interpretation of these results, they state, "these findings demonstrate that Shh signalling prevents the D cyclin-dependent formation of an additional digit". This is an overly strong conclusion based on an overinterpretation of some rather indirect evidence.

We understand that this point refers to essential requirement outlined by the Editor. We have used Chi squared tests ‘to determine whether 21/42 wings with four digits (regardless of identity) in cyclopamine treated embryos is different from 2/23 wings with four digits in cyclopamine and PD0332991’. We have also softened our conclusion – Line 195-197. ‘Therefore, these findings demonstrate that D cyclin inhibition prevents the formation of an additional digit following the loss of Shh signalling.’

4) Results section. "The percentage of polarising region cells in G1-phase of the cell cycle increases to 74.3% in Bmp2-treated wing buds compared with 64.1% in control PBS bead treated buds at 24 h". Is this a significant change? Throughout the paper, the absence of statistical analysis underlines the results.

Previously, we have shown that G1-phase numbers between left and right wing buds of the same embryos differ by less than 1% (Chinnaiya et al., 2014). In addition, when we carefully stage-match tissue from separate embryos incubated at the same time, we also achieve similar values (Chinnaiya et al., 2014). When we perform Pearson’s chi-squared analyses, a significantly different result generally equates to a >1-2% difference in G1-phase numbers between the control and experimental tissue. The differences in G1 numbers in the key experiments in this paper range between 10% and 32%, which represent extremely significant changes. We have added these p values to both the text and figures.

5) In the cyclopamine + Bmp2 experiments (Results section), it is stated, "of the cyclopamine-treated wings with four digits in their left wings, 47% have the digit pattern of 1-2-2-2…and 53% have the pattern 1-2-2-3…" The problem here is that they report percentages of the…treated wings that developed four digits. How many of the treated embryos developed four digits? It is critical to know the size/percentage/number of individuals in the affected and unaffected groups in order to interpret the distribution of digit patterns within the affected group.

We have included this data.

6) If p27kip1 is integral to the limb polarising region timer, then one would expect deletion of p27kip1 enhance to perturb limb patterning. However, this experiment was done in 1996 by Fero et al., Kiyokawa et al., and Nakayama et al., but there were no reported effects on the limbs. How does the finding that p27kip1 is dispensable for limb development affect the conclusion (Discussion section) that "a cell cycle timer involving p27kip1 also operates in vivo to control tissue growth and patterning of a major signalling centre during embryogenesis"?

There have been many reports of the deletion of cell cycle genes including D cyclins and their inhibitors producing no effects on mouse limb patterning. It is possible that the embryonic cell cycle is extremely robust to such germline knockouts. For instance, mice with deletion of all three D cyclin genes develop to quite late stages of embryogenesis because of compensation by other cyclins. Additionally, Cyclin D2 knockout mice have not been reported to have limb defects. However, the predicted constitutive expression of Cyclin D2 does cause limb defects in MPPH patients, suggesting the cell cycle is unable compensate. Therefore, knockout analyses in mice would have also suggested that Cyclin D2 is not important for limb development, but our data and interpretation of MPPH patients would indicate this is not the case. We have discussed all of these points in relation to p27 and Discussion section. We have also softened our conclusion. ‘Our work extends these earlier studies, by providing evidence for a role of p27kip1 in cell cycle timing during embryogenesis’.

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

While the manuscript has been much improved, reviewer 2 continues to worry about statistical analysis. I have included the full reviews and ask you to fix these remaining issues. Although eLife does not normally allow for multiple rounds of review, I am willing to make an exception in this case since I think you can deal with this remaining concerns in a relatively short period of time without further experimentation but with better analysis. However, no further rounds of revision will be possible, so I urge you to address in detail each of the points raised by the reviewer below.

Reviewer #2:

1) In the review of the original submission, I noted that this advance was incremental relative to their previous (2014 and 2016) papers on this topic, but my comment was not addressed during the open comment period or in the editor's letter to the authors. Therefore, in fairness to the authors, I considered that point to be moot when reviewing the revised version of the paper and my comments are restricted to the data and interpretation. No further action is required.

2) I have reconsidered my comments on Figure 5. I was critical of the analysis of the variable digit phenotypes that were obtained in the two treatments and I suggested that they examine the statistical significance of the variation. Although the authors report the differences in specific digit patterns, their conclusions are based entirely on whether the limbs have 3 digits or 4 digits; the variation in digit patterns was ignored. I accept that they have chosen to bin the data this way, and the effect of PD0332991 treatment on digit number is clear. If the range of variation produced by cyclopamine is not considered in the interpretation of results, then there is no reason to perform the statistical test that I suggested in my previous reviews. I view this as the authors' prerogative. No further action is required.

3) In Figure 7, the use of 3 graphs to display the results of Bmp treatment is unnecessary. The number of individuals unaffected by cycloplamine (panel C) can be stated in the text and the graph eliminated. To make it more clear that panel h has broken-out the 1223 and 123 categories of panel D, I suggest using the same colors.

We have eliminated the graph (panel C) and we have changed the colour of panel H (now G) to purple to avoid confusion with the categories.

4) For the experiments that compare the number of cells in G1 between two treatment groups (such as cyclopamine+Bmp2 vs. cyclopamine+PBS) in Figure 1, Figure 4 and Figure 6, I suggested that they use the Student's t-test, which is the appropriate test to compare a quantitative, single dependent variable (the number of cells in G1) between 2 groups. Instead, the authors used a Pearson's chi-squared test and argued that carrying out a t-test would require additional experiments.

I found it difficult to understand how running a different statistical test on the same data set could require additional experiments. However, after reading this paper several times, I now suspect that the central problem concerns a disparity between the description of the data in the results and the description of the data in the methods. Throughout the text, they refer to changes in "the number of G1-phase cells" or "the number of cells in G1". This implies that they compared the numbers of cells in G1 between groups. However, the methods section describes the use of DNA content to assign cells to G1, S, or G2/M, and these categorical assignments were expressed as a percentage of the total cell number (Materials and methods section). Therefore, they are actually comparing between 2 groups the percentage of cells in 1 of the 3 categories, which is not the same as comparing the number of cells in G1 in 2 treatment groups. In other words, it appears that they are comparing categorical variables rather than a quantitative variable. If this is the case, then they can probably fix the problem by providing a more accurate description of the data in the Results section.

When we have first described flow cytometry we have added: “Note for all Chi-square analyses the percentage of cells in G1-phase in control and experimental samples were compared to the percentage of cells in S/G2 and M phase”.

5) Finally, although I have offered recommendations on the appropriate statistical tests, I believe that reviewers should not dictate how authors conduct their studies. The authors have my comments and the reasoning behind my advice, so I will now leave it to the authors to proceed as they see fit.

We have continued to use Chi-square tests to maintain consistency with our previous papers.

https://doi.org/10.7554/eLife.47625.015

Article and author information

Author details

  1. Joseph Pickering

    Department of Biomedical Science, University of Sheffield, Sheffield, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Investigation, Writing—original draft, Performing and planning most experiments, Analysis and validation of the data, Editing and approving the paper
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5892-5159
  2. Kavitha Chinnaiya

    Department of Biomedical Science, University of Sheffield, Sheffield, United Kingdom
    Contribution
    Formal analysis, Validation, Investigation, Performing p27 cyclopamine experiments, Analysis and validation of the data, Approving the paper
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3375-420X
  3. Matthew Towers

    Department of Biomedical Science, University of Sheffield, Sheffield, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    m.towers@sheffield.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2189-4536

Funding

Wellcome (202756/Z/16/Z)

  • Joseph Pickering
  • Kavitha Chinnaiya
  • Matthew Towers

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

Acknowledgements

We thank the Wellcome Trust for funding, Cheryll Tickle, Marysia Placzek and Jasmina Tarpy for critical reading, and the University of Sheffield flow cytometry core facility.

Senior and Reviewing Editor

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

Publication history

  1. Received: April 11, 2019
  2. Accepted: September 23, 2019
  3. Accepted Manuscript published: September 23, 2019 (version 1)
  4. Version of Record published: October 4, 2019 (version 2)

Copyright

© 2019, Pickering et al.

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

Metrics

  • 624
    Page views
  • 99
    Downloads
  • 0
    Citations

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

Download links

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

Downloads (link to download the article as PDF)

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

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