Final Report Summary - NEURAL TUBE NETWORK (Systematic analysis and modeling of the gene regulatory network underlying neural tube patterning)
In the developing spinal cord, the secreted molecule Sonic hedgehog (Shh) specifies 5 molecularly distinct progenitor domains in a precise spatial order along the dorsoventral axis of the developing spinal cord. These domains can be distinguished by the expression of different combinations of transcription factors (TFs) and generate functionally distinct neuronal subtypes. During their specification, the progenitors transit through several metastable states, exhibited by the expression of different combinations of TFs. The gene regulatory network controlling transition through these states and final progenitor fate is poorly understood.
The main goal of this project is to define the transcriptional states during progenitor specification and model the gene regulatory network controlled by Shh in this process. As outlined in section B1 of this proposal, the project is separated in three main steps: 1) establish a profiling platform as a tool to iteratively assay the operation of the transcriptional network during neural tube patterning in the developing chick embryo; 2) identify sets of transcription factors differentially expressed upon perturbation of the transcriptional network underlying patterning of the ventral neural tube; 3) reconstruct, refine and analyze a model of the GRN underlying patterning of the ventral neural tube.
Summary of progress towards each objective:
1. Relevant techniques were developed for the in vivo transfection and FACs purification of cells from the chick neural tube. To identify the transcriptome of neural progenitors responding to different periods of Shh signaling we performed transcriptome analyses using high throughput mRNA sequencing (HTS) of neural tube cells electroporated with SmoM2, an activated verison of the Shh transducing protein, and isolated at 12hpe and 18hpe. This time points represent progenitors that are predominantly in pMN and pV3 identities, respectively. We obtained biological triplicates for each condition. The acquired reads were mapped to chick cDNAs and ESTs, and differentially expressed genes identified by data analysis using the R package DESeq. We validated the results using a second set of data collected using Nanostring nCounter this provides a more accurate estimate of gene expression levels using a small number of genes..
2. I constructed a database of chick transcription factors based on collating data from the available mammalian databases and extrapolating this to chick. I have used this to annotate the data from the transcriptome analysis. In addition, to complement previous experiments on the second objective, we obtained HTS transcriptome data from RNA samples obtained by transfection of chick neural tubes in vivo with GFP control, SmoM2 and , Ptc1Deltaloop2 at 12hpe and 18hpe, and Nkx2.2 Olig2, Irx3, Nkx6.1 and combinations of these TFs at 12hpe. By comparing the profiles of differentially expressed genes between the conditions of TF misexpression and constitutive activation of Shh (SmoM2) for different amounts of time, we have uncovered that gene expression for the p3 and pMN states is predominantly defined by combinatorial regulation by Nkx2.2 and Nkx6.1 for the p3 state and Olig2 and Nkx6.1 for the pMN state.
3. Towards objective 3, we are collaborating with computational statisticians at UCL, London and Royal Holloway University, London. This involves developing dimensionality reduction algorithms to visualize and cluster the data sets (Bushati et al, 2011) and the development of machine learning algorithms to infer regulatory relationships from correlated patterns of expression in the datasets. In addition we are intersecting our data with data obtained by others in the lab to identify genes directly regulated combinatorially by Nkx2.2 Nkx6.1 and Olig2 using ChIP-Seq and the transcriptome data obtained in this project. We have continued our collaboration with computer scientists at UCL to develop probabilistic models for gene regulation from our transcriptome data. We are currently working on integration of the data to identify genes regulated by individual and combinatorial TFs during ventral neural tube patterning.
Summary of main results achieved so far:
Using in ovo electroporation of constructs that change the level of Shh signalling and/or the expression of specific TFs, we experimentally perturbed the gene regulatory network in the developing chick neural tube. We found that the transcription factors Olig2 and Nkx6.1 and Nkx2.2 and Nkx6.1 act in a combinatorial manner to contribute to the gene expression profiles of a large number of genes expressed in the pMN and p3 progenitor states.
These findings suggest that while individual TFs such as Nkx6.1 and Nkx2.2 can establish progenitor-like states, the combinatorial action of these resembles the states obtained by constitutive Shh activation more closely. These findings provide a contrast to the ‘master regulator’ view of cell type specification.
Expected final results and potential impact and use:
Our data suggests both combinatorial and non-combinatorial actions of neural tube TFs in the establishment of progenitor gene expression profiles. Ongoing collaborative analyses are designed to test and validate this hypothesis. Potentially, this finding would have impact for understanding the gene regulatory network underlying dynamic establishment of progenitor domains in the ventral neural tube and gene regulatory networks operating in other developing tissues. This could be applicable to medically relevant tissue-engineering approaches and the field of regenerative medicine, as well as to stem cell research.