Gene regulatory logic for reading the Sonic Hedgehog signaling gradient in the vertebrate neural tube

The regulatory logic of the Pax6-Olig2-Nkx2.2 circuit explains why both the level and duration of Shh signaling affect gene expression. Two factors dictate the response of a cell: the current level of Gli activity, and the existing gene expression profile in the cell (Figure 7A). Because the current state of gene expression in a cell is a consequence of prior Gli activity, which results from exposure to Shh, it provides a memory of the signaling experienced by a cell. In this way the circuit acts as the timer that measures the duration, as well as the level, of Gli activity and Shh exposure. The consequence of this mechanism is that the response to different levels of signal is produced by the same mechanism that generates the different temporal responses of the genes. Thus, the temporal and graded responses to Shh are inseparable properties of the regulatory circuit. This reconciles experimental results that have indicated that either the level or the duration of signaling is critical for the control of gene expression (Ahn and Joyner, 2004; Dessaud et al., 2007; 2010; Harfe et al., 2004; Pagès and Kerridge, 2000). Moreover, the observation that temporally changing levels of signaling can generate the same spatial patterns of gene expression suggests that different modes of signaling—temporal versus graded—could be responsible for the profile of gene expression at different positions within the tissue.

In addition to explaining the mechanism that results in differential spatial patterns of gene expression, we provide in silico and in vivo evidence that the network confers both robustness to signal fluctuation and hysteresis. The insensitivity of the circuit to transient changes in the level of signaling provides a means to achieve reliable patterning despite the inherent noisiness of development (Bollenbach et al., 2008; Gregor et al., 2007). This suggests a solution to the apparent discrepancy between the accuracy of patterning processes and the limits on the precision of morphogen gradients (Bollenbach et al., 2008; Gregor et al., 2007; Jaeger and Reinitz, 2006; Manu et al., 2009a). Moreover, this might help explain why, in the absence of Gli3, although there is a marked increase in signaling, only minor neural tube patterning defects are observed. We provide evidence that the transcriptional circuit is able to buffer the short-lived elevation in signaling levels that result from loss of Gli3. The increased levels of Ptch1, induced by the increased signaling, are then likely to contribute to restoring signaling to normal levels. This represents an example of system-level feedback and, from the perspective of Waddington's epigenetic landscape (Waddington, 1942), suggests a molecular explanation for the phenomenon of “canalization,” although further studies will be necessary to identify additional mechanisms for the robustness of neural tube patterning.

The in silico and experimental analysis also revealed that the transcriptional circuit confers hysteresis to Nkx2.2 expression. This offers an explanation for how the pattern of gene expression is maintained during development even as the signaling gradient recedes. The maintenance of gene expression patterns during the elaboration of tissue development is a key feature of patterning. The finding that the same mechanism is responsible for both the initial interpretation and the maintenance of Nkx2.2 provides an elegant solution to this problem.

Finally, examination of the regulatory logic of the circuit revealed that it consists of an overlapping arrangement of positive and negative feedback (Figure 7B). Together with the experimental analysis, this provides an intuitive understanding of the performance of the transcriptional circuit. Moreover, the logic can be generalized and extended to regulate additional target genes in a morphogen-like manner (Figures 7C and 7C′). As such, this mechanism might represent a general strategy for morphogen interpretation. Together, the study highlights the information-processing power of transcriptional networks (Alon, 2007; Davidson, 2010; Jaeger and Reinitz, 2006; Vokes et al., 2007), and the simplicity and adaptability of this mechanism suggest that it is likely to be relevant for the control of patterning of tissues other than the neural tube.

Mice containing mutant alleles for Pax6 (Small Eye allele), Olig2, and Gli3 (XtJ allele) have been described previously (Ericson et al., 1997; Hui and Joyner, 1993; Zhou and Anderson, 2002). To generate the Tg(GBS-GFP) line, eight concatemerized fragments of a FoxA2 enhancer that contains a Gli binding site (Sasaki et al., 1997) were cloned upstream of the hsp68 minimal promoter and eGFP (details are provided in Extended Experimental Procedures). To stage mice, we used somite number and converted these to standardized times (Table S1) expressed as hours postheadfold stage (hph). All procedures were carried out with the approval of the Institute Ethical and Biological Services Animal Research Committees under Home Office Project License (PPL 80/2091).

Shh null allele was generated from a Cre-dependent conditional allele and genotyping was carried out as reported (Lewis et al., 2001). Genotyping for all other mouse lines was carried out as reported (Buscher et al., 1998; Lu et al., 2002; Schedl et al., 1996).

8 concatemerized fragments of a FoxA2 enhancer that contains a Gli Binding Site (GBS; TTATGACGGAGGCTAACAAGCAGGGAACACCCAAGTAGAAGCTGGCTGTC) (Sasaki et al., 1997) were cloned upstream of the hsp68 minimal promoter and eGFP in pBluescript SK+. To protect the activity of the 8GBS-hsp68-eGFP from position effects when integrated into the mouse genome, this fragment was then subcloned in the BamH1 site of pJC13-1(Chung et al., 1993), which contains 2 insulators on each side of the BamH1 site. Tg(GBS-GFP) transgenic mice were generated by pro-nuclear injection into fertilized eggs from FVBN mice, and founder animals were genotyped for the eGFP sequence by PCR using the following primers: 5′TGCAGTGCTTCAGCCGCTAC3′; 5′CCAGCAGGACCATGTGATCG3′. Two transient transgenic embryos at 70 hr postheadfold stage (hph) (e10.0) and 2 stable lines were obtained and all displayed similar patterns of eGFP expression (data not shown).

Mouse embryos from timed pregnant females were staged by counting somite number and fixed by immersion in 4% paraformaldehyde for 45min to 2hrs at 4°C. Fixed embryos were cryoprotected by equilibration in 15% sucrose, cryosectioned (14 μm) and processed for immunostaining or in situ hybridization (ISH). The following primary antibodies were used: rabbit anti-Olig2 (1:1000, Chemicon), sheep anti-GFP (1:1000, Biogenesis), rabbit anti-GFP (1:1000, Invitrogen), mouse against Nkx2.2 (1:25), rabbit anti-Ptch1 (Gift from Dr. Argraves [Morales et al., 2009]), guinea-pig anti-Gli2 (Gift from Dr Eggenschwiler [Ko et al., 2010]). Secondary antibodies were from Jackson Immuno Research: FITC donkey anti-sheep or donkey anti-rabbit IgG (H+L); Cy5 donkey anti-goat IgG (H+L) and Cy3 donkey anti-mouse, guinea-pig or rabbit IgG (H+L). ISH was performed as described (Yamada et al., 1993), using as probes against mouse Ptch1 (Ding et al., 1998), mouse Gli1 and GFP (kindly provided by C.C. Hui and S. Gerety), respectively. Analyses were carried out using a Leica TCS SP2 and/or SP5 confocal microscope or a Zeiss Axioplan 2 and images processed with Adobe Photoshop CS4 software (Adobe Systems).

All chick misexpression constructs were based on pCAGGS expression vector (Niwa et al., 1991) engineered to bicistronically express nuclear targeted GFP. Gli3AHIGH (Stamataki et al., 2005), Gli3AMED (Stamataki et al., 2005), SmoM2 (Hynes et al., 2000), Olig2 (Novitch et al., 2001), and the two Pax6-RNAi (Das et al., 2006) constructs have been described previously. Hamburger and Hamilton (HH) stage 8-12 chick embryos were electroporated and incubated in ovo before dissecting and processing for immunohistochemistry.

Gli3AMED (Stamataki et al., 2005), Pax6-RNAi (Das et al., 2006) constructs or pCAGGS as control were electroporated in chick embryos along with GBS-Luc, a firefly luciferase reporter construct containing eight repeats of the Gli binding site sequence (Sasaki et al., 1997) and a Renilla-luciferase reporter carrying the CMV immediate early enhancer promoter (Promega) for normalization. Embryos were homogenized with a douncer in Passive Lysis Buffer on ice and measurement of firefly and Renilla luciferase activities was performed using the Dual Luciferase Reporter Assay System (Promega).

Neural plate tissue was isolated from 8-12hph mouse embryos and cultured as described in (Yamada et al., 1993). The medium for mouse explants was supplemented with N2 and B27 (GIBCO). For the GFP intensity quantification, at least five images containing approximately 150 cells each that had been randomly chosen from five to six explants were quantified with ImageJ (NIH) and the data were analyzed using MATLAB (Mathworks). Each experiment was performed independently more than once and gave reproducible results.

Mouse embryos from 8 to 12hph stages with intact yolk sacs, dissected from timed pregnant females, were cultured for 12h or 24h in medium (rat serum, Tyrode solution; 1:1). Cyclopamine (Sigma) dissolved in EtOH was used at a concentration of 10μM, while Purmorphamine (Calbiochem) dissolved in DMSO at a concentration of 5-10μM as indicated. Cultures were performed in a water-saturated roller-tube incubator at 37°C, 5% CO2 and 20% O2. After culture, embryos were fixed and processed. Gene expression patterns and the activity of the Tg(GBS-GFP) were always compared between embryos processed in the same culture experiment in appropriate control conditions.

Images were obtained using a Leica SP2 or SP5 confocal system. Stepwise bleaching was performed to test linearity of the GFP signal. Each image was the average of 3 optical sections, 0.2μm apart, taken from the middle of a 14μm cryosection. The fluorescence intensity of Nkx2.2, Olig2, GFP and Ptch1 expression was measured in rectangles of 16μm wide positioned from the ventral to dorsal midline along the apical side of the neural tube. Image J v.1.43 g image analysis software (NIH) was used. For comparison neural progenitor cells are on average ∼6μm wide in the dorsal-ventral axis and ∼12μm from apical to basal. Background measurements were obtained from mesoderm and/or the dorsal neural tube and these were subtracted from each assayed profile. To be able to compare the different sets of experiments, we normalized the values of GFP intensities in each set of experiments to those found in wild-type embryos (Figures 2B, S1D, and S1E). The positions of the Nkx2.2 and Olig2 expression domain boundaries were defined as the point at which fluorescent intensity reached 30% of the average of the fluorescence intensity within a domain, in which the fluorescence intensity was two folds higher than the background signal. The heat maps of GBS-GFP and Ptch1 intensity profiles along the D-V axis as well as the plots in Figure 1C, were generated from measures of mean fluorescence intensity after background subtraction in bins of 4μm. All the cell positions were recorded as the distance from the floor plate given as a % of the neural tube size. Data analysis and image generation were performed using Matlab (Mathworks, Natick, MA) or Excel (Microsoft Office).

The heat maps provide a color-coded scale for individual datasets by associating the lowest value in the data matrix with a dark blue color and the highest value with the dark red color and interpolating linearly between these values.

In the system of differential Equations 1–3, P, O and N represent the levels of Pax6, Olig2 and Nkx2.2 expression, respectively and G the levels of Shh-Gli signaling. The maximum rate of expression of each factor P, O, N is given by α, β, γ, respectively, and its degradation rate by ki, (i = 1-3). Michaelis-Menten kinetics describes the induction of O and N by G. The cross-repressive interactions between the TFs are parameterized in the equations by Hill coefficients hi, (i = 1-5) and critical values OcritP, NcritP, OcritN, PcritN and NcritP. In particular hi influences the steepness of the sigmoidal curve of the repression function, which characterizes the change in expression of a given TF in response to its repressor. The higher the value of hi, the steeper is the sigmoidal curve. The critical values dictate the value of the repressor for which the level of expression of the repressed TF is half the maximal level. Note that the dynamical system is designed to represent the epistatic relationships between the transcription factors and does not require the regulatory interactions to be direct.

The complexity of the system of differential Equations 1–3 precludes analytical solutions, thus the system was solved numerically using Matlab ode45 solver (Mathworks). The simulations give the P, O and N output values as a function of time for different values of the model parameters. For given values of G we calculated the values of P, O and N either prior to the system reaching equilibrium or once the system has reached a steady state. A full mathematical analysis of the system, including numerical analyses and a Heaviside simplification (in the limit where the Hill coefficients are infinitely large) that provides analytical solutions for the steady states and their stability, are described separately (J.P., K.M.P., and J.B., unpublished data).

We identified sets of model parameter values in which the system adopted the biologically relevant behavior: that is the levels of P, O, N switch from PHIGH→OHIGH→NHIGH states in response to progressively higher values of G (the parameters used throughout this study are shown in Table S2 and the states PHIGH, OHIGH, NHIGH are defined as the condition in which the levels of P, O, N are above 1, respectively). A sensitivity analysis, halving or doubling each parameter value in turn, indicated that the parameters that describe the strength of the cross repression had a significant influence on the output of the system (Table S3). In the model, the repressive activity of the TFs is described by the combination of two independent sets of parameters: Hill coefficients and critical values. Stronger repressors are characterized by higher values of hi and/or lower critical values. Keeping one set of parameters fixed, we determined the values of the other set that produced the relevant multistate switch. For both sets of parameters, the switch from OHIGH to NHIGH required N to inhibit O more strongly than O inhibited N (Tables S2 and S3). In addition, N had to strongly inhibit P, when O was removed from the system (β = 0).

For the extended system shown in Figure 7, the equations in C were solved numerically with the parameters given by: a = 3, b = 5, c = 5, d = 5, h13 = 4, h14 = 6, h21 = 2, h23 = 1, h24 = 1, h31 = 6, h32 = 5, h34 = 1, h41 = 6, h42 = 5, h43 = 5, X2crit3 = 2, X2crit4 = 0.2, X3crit2 = 0.5, X3crit4 = 3, Xjcriti = 1; [Xjcriti = critical value that describes the effect of j on i], ki = 1; i = 1,2,3,4; M = 1).

Mouse embryos from timed pregnant females were staged and fixed in 4% paraformaldehyde for 45 min to 2 hr at 4°C. Fixed embryos were cryoprotected by equilibration in 15% sucrose, cryosectioned (14 μm), and processed for immunostaining (Briscoe et al., 2000) or in situ hybridization (ISH) (Yamada et al., 1993). Details of the reagents and the quantification are provided in the Extended Experimental Procedures.

Neural plate tissue was isolated from HH10 stage chick embryos and cultured as described (Yamada et al., 1993). Shh protein was generated as described (Ericson et al., 1997). Cyc (Toronto Research Chemicals) was dissolved in 100% ethanol. Luciferase assays in explants were performed as previously described (Dessaud et al., 2007). Each experiment was performed independently more than once and gave reproducible results.

We are grateful to N. Bushati, A. Kicheva, T.M. Jessell, E. Kutejova, S. Tozer, and J.P. Vincent for discussions. We thank NIMR Biological Services staff for help with the mouse colonies. This work was supported by the MRC (U117560541) and Wellcome Trust (080630). E.D. was supported by EMBO and Marie Curie Fellowships, N.S. by a Marie Curie Fellowship, V.R. by an EMBO Fellowship, and A.R. by the FCT, Portugal. K.M.P. thanks the National Institute of Mathematical and Biological Synthesis for a sabbatical fellowship (NSF Grant EF-0832858 and University of Tennessee, Knoxville).

N.B. and E.D. conceived and initiated the study. V.R. generated Tg(GBS-GFP) mice. A.R. and V.R designed and performed the quantitative analyses of mouse embryos. N.B and V.R performed mouse genetic experiments. N.S. designed and performed explant experiments. J.P. and K.M.P formulated and analyzed the mathematical model. V.R., N.B., and J.B. wrote the manuscript.