Ptch1 and Gli regulate Shh signalling dynamics via multiple mechanisms

Abstract

In the vertebrate neural tube, the morphogen Sonic Hedgehog (Shh) establishes a characteristic pattern of gene expression. Here we quantify the Shh gradient in the developing mouse neural tube and show that while the amplitude of the gradient increases over time, the activity of the pathway transcriptional effectors, Gli proteins, initially increases but later decreases. Computational analysis of the pathway suggests three mechanisms that could contribute to this adaptation: transcriptional upregulation of the inhibitory receptor Ptch1, transcriptional downregulation of Gli and the differential stability of active and inactive Gli isoforms. Consistent with this, Gli2 protein expression is downregulated during neural tube patterning and adaptation continues when the pathway is stimulated downstream of Ptch1. Moreover, the Shh-induced upregulation of Gli2 transcription prevents Gli activity levels from adapting in a different cell type, NIH3T3 fibroblasts, despite the upregulation of Ptch1. Multiple mechanisms therefore contribute to the intracellular dynamics of Shh signalling, resulting in different signalling dynamics in different cell types.

Figure 1. Shh levels increase in the neural tube.

(a) Cartoon of the Shh pathway. In the absence of Shh, Ptch1 inhibits Smo. GliFL is processed into its repressive form GliR, which switches off the expression of target genes. In the presence of Shh, Ptch1 receptors bind to the ligand, derepressing Smo and converting GliFL to its active form, GliA, which promotes the target gene expression including ptch1. (b) Sections from embryos were immunostained for either GBS-GFP, Ptch1 and Nkx2.2 or GBS-GFP, Shh and Nkx2.2 (Nkx2.2 data not shown). Two sections from the same embryo at the 16-somite stage. Scale bar, 20 μm. GBS-GFP and Ptch1 data has been reproduced from ref. . (c) A neural tube section at the 30-somite stage immunostained for Shh. The mean Shh fluorescence intensity across a 16-μm wide region of interest (ROI) extending from ventral to dorsal midline, in steps of 1 μm, was quantified next to the apical lumen. The floor plate (FP) and notochord (NC) provide the source of Shh. Scale bar, 20 μm. (d) Shh immunostaining in embryos of the indicated somite stages (ss) from E8.5 (~10ss) to E10.5 (~40ss) Scale bar, 20 μm. The number in brackets represents the dorsal–ventral size of the neural tube.

Figure 2. Quantitation of Shh signalling levels.

(a) Mean Shh intensity for the indicated developmental stages along the dorsal–ventral axis. The width of the curves shows the 95% confidence interval for the mean. Yellow crosses indicate the mean position of the Nkx2.2 ventral boundary (s.e. <1.1 μm, not shown). The shift in the position of the maximum intensity over time is attributed to the changes in the size and position of the floor plate. (b,c) The decay lengths (b) and amplitudes (c) with linear trendlines from exponential functions fit to the Shh fluorescence intensity profiles. The average correlation coefficient for the exponential fits using Matlab’s non-linear least-squares method was R²=0.74. For the decay length, there was no correlation with neural tube tissue size, for the amplitude the correlation coefficient, R²=0.8 (P<0.0001). (d–f) Heat maps of mean Shh, GBS-GFP and Ptch1 fluorescence intensity (colour-coded) profiles. Embryonic stage on the x axis and distance from the ventral midline, relative to the total neural tube length, on the y axis. All profiles normalized to maximum intensity. Each panel represents data from 68 embryos, with at least three embryos per stage (Shh (n=185), GBS-GFP (n=206), Ptch1 (n=102)). The GBS-GFP and Ptch1 data are reanalysed data from ref. . (g–i) The data in d–f, plotted for several DV position ranges. On the x axis are different time points. The data were binned along the x axis with bin size of 40 μm and the curves interpolated by fitting a spline at 0.1 μm intervals. The curve width indicates 95% confidence intervals for the mean values.

Figure 3. Ptch1 feedback is not required for adaptation.

(a,b) Examples of mouse embryos at 4–6-somite stage cultured for 24 h in the presence (b) or absence (a) of Pur. The media was replaced after 12 h to ensure against degradation of Pur, and the expression of GBS-GFP was recorded at 12 and 24 h. The embryos were also stained for the neural progenitor transcription factors Nkx2.2 and Olig2. Scale bars, 20 μm. (c,d) The GBS-GFP fluorescence intensity quantified as in Fig. 1c. The plots show the average profiles from six experiments, normalized to the maximum GFP levels in each experiment at 12 h, separately for Control and Pur treatment. In both cases, there is an adaptation in the ventral region of the neural tube demonstrating that Ptch1 feedback is not required for adaptation.

Figure 4. Three distinct mechanisms could explain adaptation.

(a) Diagram of the full model used in the analysis. Shh binds to Ptch1, freeing the inhibition of GliFL conversion to GliA. GliA and GliR both compete to bind at two binding sites on the target genes gfp, ptch1 and x. X inhibits the production of GliFL. The coloured boxes highlight the three different mechanisms. Adaptation occurs by the regulation of Gli degradation (red) or adaptation by ptch1 feedback (blue) or by the transcriptional downregulation of Gli (green). (b,c) The experimental data (Fig. 2 and Supplementary Fig. 1c) was stereotyped into two time courses representing the average levels of Shh (b) and GBS-gfp and ptch1 (c) in two ventral regions of the neural tube (0–10% and 10–20% of relative DV length). The Shh time course (b) (from Fig. 2g) was linearly fitted and extrapolated back in time to zero levels. (c) The GBS-gfp and ptch1 target time dynamics were approximated from the protein expression in Fig. 2h,i. (d–k) Simulated signal trajectories of the posterior parameter distributions obtained by the Bayesian analysis. The Figures show the median trajectory (red) and 40–60 percentile ranges (blue) for either gfp (for the Gli stability and transcription models) or ptch1 (in the full model and ptch1 feedback model—where gfp followed similar trajectories). The response to the higher Shh input (d–g) and the lower Shh input (h–k) for the full model (d,h), Gli stability (e,i), Ptch1 transcription (f,j) and Gli transcription (g,k) are shown.

Figure 5. Posterior parameter distributions for adaptation mechanisms.

(a–m) Box plots of the marginal posterior distributions for the key parameters for the indicated models. These comprise: the transcription rate of ptch1 (b), the degradation rate of ptch1 mRNA (c), the translation rate of ptch1 (d), the degradation rate of Ptch1 protein (d), the activation rate of Ptch1 (f), the degradation rates of GliFL (g), GliR (h), GliA (i), the conversion rates of GliFL to GliR (j) and GliFL to GliA (k), and the binding affinites of Polymerase to gfp (l) and to x (m).

Figure 6. Gli2 is downregulated in the ventral neural tube.

(a–d) Transverse sections of the neural tube between somite stages 5ss and 25ss were immunostained for GBS-GFP, Olig2 (a marker of the pMN domain), and Gli2. Gli2 protein levels are significantly downregulated by 15ss in the Nkx2.2 domain (ventral to Olig2) and across the entire neural tube by 25ss, when GBS-GFP expression has also been downregulated. Identical magnification in all panels; scale bar, 50 μm.

Figure 7. Different dynamics of Shh signalling in NIH3T3 cells.

(a–d) NIH3T3 cells were cultured with Shh, Pur or control medium for the indicated amounts of time. Gli1, Gli2, Gli3 and Ptch1 RNA expression was determined by qPCR and normalized to actin levels. Signalling is induced as shown by the upregulation of Ptch1 and Gli1 however there was no adaptation. In these cells Gli2 and to a lesser extent Gli3 were upregulated by Shh signalling. (e,f) The model (described in Fig. 4a and Supplementary Fig. 2) was implemented with a fixed pulse of Shh (f). A set of parameters was selected for which adaptation would occur in ptch1 (and gfp—not shown; solid line in e). The parameter c_X, defining the transcriptional strength of factor X, was varied and the model output compared in each case (I): c_X=0 (X acts a repressor—solid line); c_X=1 (X has no effect—dotted line); c_X=10 (X acts as an activator—dashed line). The remaining model parameters were: tl_Ptc=100, tl_X=1, tl_GliFL=100, conv_GliR=0.01, tr_gliFL=100, tr_x=1, deg_ptc=2, tr_ptc=100, [Pol]=1, deg_Ptc=0.1, deg_GliFL=0.1, deg_GliA=1.5, deg_GliR=0.01, deg_X=0.5, deg_x=1, K_Gli_ptc=1, deg_gliFL=0.03, conv_GliA=10, K_X_gli=10, K_Pol_ptc=1, K_Pol_x=1, K_Pol_gli=0.01, c_GliA=10, c_GliR=0, k_ShhPtc=100, Km_Ptch1=1, K_Gli_x=10, act_Ptc=10, (units are defined in Supplementary Fig. 2B).