Skip to main content

Investigation of the molecular mechanisms ensuring precision and reproducibility of Sonic Hedgehog mediated patterning in the vertebrate ventral neural tube

Final Report Summary - NEURALNETWORKNOISE (Investigation of the molecular mechanisms ensuring precision and reproducibility of Sonic Hedgehog mediated patterning in the vertebrate ventral neural tube)

Proper development of an organism depends on the establishment of precise and reproducible patterns of gene expression and growth. In many tissues these two processes are controlled by morphogens, locally produced secreted molecules that form concentration gradients and induce target genes at a distance from the source of the signal. This led to the idea that the concentration of morphogens is the main determinant for their biological activity. This view has been challenged recently by findings in various model systems that highlighted the importance of other factors, especially the interpretation of signaling duration by downstream gene regulatory networks (GRNs), for differential target gene induction.
The vertebrate spinal cord is one of the best-understood model systems in terms of morphogen signaling. Here, neuronal progenitor domains are specified along the dorsal-ventral (DV) axis by the distinct activities of different morphogen systems. Ventrally, Sonic Hedgehog (Shh) emanating first from the notochord and later from the floor plate (FP) forms a gradient that subdivides the neural tube into six progenitor domains (FP, p3, pMN, p2, p1, and p0), which generate distinct subtypes of interneurons (V0-V3) and motor neurons (MN). Progenitor identity is conferred by a combinatorial code of transcription factors (TFs), whose expression patterns are determined in response to graded Shh activity and cross-repression between the TFs themselves. This process is best understood for the three most ventrally located neural progenitor domains (p3, pMN, p2), whose boundaries are determined by cross-repressive interactions between four core members of the Shh-controlled GRN (Pax6, Irx3, Olig2 and Nkx2.2).
The relative contribution of the GRN to the precision of pattern formation and target gene expression and to buffer noise in Shh signaling had not been addressed, when we started this project. Furthermore, the extent to which other signaling pathways and biological processes, e.g. neurogenesis, affect the expression levels of these GRN components was unclear. The main questions addressed throughout the course of the project were:

1.) Quantification of spatiotemporal noise in Shh pathway activity and target gene expression
2.) Establish single cell sequencing to assay cell state, heterogeneity and noise in gene expression and to reveal exogenous effects on the Shh GRN
3.) Integrate the experimental data into computational models to quantify gene expression dynamics

To measure the spatiotemporal dynamics in Shh signaling we validated a previously constructed Shh reporter and established fluorescent reporter lines for core components of the Shh-controlled GRN. Analysis of Shh signaling and GRN expression dynamics at the population level confirmed earlier findings that higher levels and longer duration of Shh signaling promote ventralization of neural progenitors. Surprisingly, however, this correlation was not obvious at the single cell level. This suggested to us that potentially other signaling pathways and biological processes may affect expression levels of these genes.
To further investigate the molecular basis of this phenomenon, we used single cell transcriptomics of ES cell derived spinal motor neuron progenitors in conjunction with in vivo molecular and genetic experiments. Using methods from graph theory and dimensionality reduction we developed computational tools that allowed us to reconstruct, from the single cell RNAseq data, changes in the transcriptional profile of cells as they progress from naive neural progenitors to MN progenitors and post-mitotic MNs. The resulting transcriptional dynamics identified signatures of other signaling pathways as well as novel factors involved in the process. More strikingly, it defined several distinct phases of MN differentiation including a stage, previously unrecognized, in which the MN progenitor transcription factor Olig2 increases as cells commit to MN differentiation. This correlated with the direct repression of the Notch signaling effectors Hes1 and Hes5, known inhibitors of neurogenesis. Conversely, inhibition of Notch signaling increased expression levels of Olig2. These results suggest that besides activity of the Shh pathway the activities of other developmental signaling pathways have to be taken into account to understand the expression dynamics of components of this GRN.
Taken together, our data reveal a tight coupling between the transcriptional networks that control patterning and differentiation and highlight the pivotal role of Olig2 not only as an important organizer of dorsal-ventral pattering, but also as the pacemaker for MN generation. These results also explain the spatial and temporal patterns of neurogenesis observed in the neural tube.

Our results are likely to be relevant for researchers studying neuronal differentiation in other developmental and clinical contexts. Olig2 is not only important for MN formation but is also required for oligodendrocyte formation and implicated in the formation of gliomas. Notch signaling and Hes genes are implicated in regulating cell fate decisions throughout animal development and in cancer. Thus, Olig2 (and other Olig family members) may regulate Notch signaling in other developmental contexts through a similar mechanism. Furthermore, MNs are clinically relevant as their degeneration causes amyotrophic lateral sclerosis (ALS). Hence, there is substantial interest in understanding the mechanisms underlying MN formation for disease modeling and improving reprogramming strategies for cell replacement. To our knowledge this is the first reconstruction of transcriptional dynamics by single cell sequencing along the endogenous trajectory from neural progenitors to MNs and provides previously unavailable detail of the gene expression changes that lead to MN differentiation. Thus, our single cell sequencing data will serve as an important reference for future studies comparing transcriptional dynamics during reprogramming or endogenous differentiation of MNs. More generally, the analytical methods we develop are directly applicable to similar problems. Finally, the surprising demonstration that the reconstructed developmental trajectories not only accurately predicts the known sequence of gene expression but provides higher temporal resolution than available from in vivo observations is likely to be of broad interest to the field.