Faculty Bibliography

The motor neuron progenitor domain in the ventral spinal cord gives rise to multiple subtypes of motor neurons and glial cells. Here, we examine whether progenitors found in this domain are multipotent and which signals contribute to their cell-type-specific differentiation. Using an in vitro neural differentiation model, we demonstrate that motor neuron progenitor differentiation is iteratively controlled by Notch signaling. First, Notch controls the timing of motor neuron genesis by repressing Neurogenin 2 (Ngn2) and maintaining Olig2-positive progenitors in a proliferative state. Second, in an Ngn2-independent manner, Notch contributes to the specification of median versus hypaxial motor column identity and lateral versus medial divisional identity of limb-innervating motor neurons. Thus, motor neuron progenitors are multipotent, and their diversification is controlled by Notch signaling that iteratively increases cellular diversity arising from a single neural progenitor domain.

Imaging systems that allow visualization of specific loci and nuclear structures are highly relevant for investigating how organizational changes within the nucleus play a role in regulating gene expression and other cellular processes. Here we present a live imaging system for targeted detection of genomic regions. Our approach involves generating chimaeric transcripts of viral RNAs (MS2 and PP7) and single-guide RNAs (sgRNAs), which when co-expressed with a cleavage-deficient Cas9 can recruit fluorescently tagged viral RNA-binding proteins (MCP and PCP) to specific genomic sites. This allows for rapid, stable, low-background visualization of target loci. We demonstrate the efficiency and flexibility of our method by simultaneously labelling major and minor satellite regions as well as two individual loci on mouse chromosome 12. This system provides a tool for dual-colour labelling, which is important for tracking the dynamics of chromatin interactions and for validating epigenetic processes identified in fixed cells.

Polycomb and Trithorax group proteins encode the epigenetic memory of cellular positional identity by establishing inheritable domains of repressive and active chromatin within the Hox clusters. Here we demonstrate that the CCCTC-binding factor (CTCF) functions to insulate these adjacent yet antagonistic chromatin domains during embryonic stem cell differentiation into cervical motor neurons. Deletion of CTCF binding sites within the Hox clusters results in the expansion of active chromatin into the repressive domain. CTCF functions as an insulator by organizing Hox clusters into spatially disjoint domains. Ablation of CTCF binding disrupts topological boundaries such that caudal Hox genes leave the repressed domain and become subject to transcriptional activation. Hence, CTCF is required to insulate facultative heterochromatin from impinging euchromatin to produce discrete positional identities.

The emergence of limb-driven locomotor behaviors was a key event in the evolution of vertebrates and fostered the transition from aquatic to terrestrial life. We show that the generation of limb-projecting lateral motor column (LMC) neurons in mice relies on a transcriptional autoregulatory module initiated via transient activity of multiple genes within the HoxA and HoxC clusters. Repression of this module at thoracic levels restricts expression of LMC determinants, thus dictating LMC position relative to the limbs. This suppression is mediated by a key regulatory domain that is specifically found in the Hoxc9 proteins of appendage-bearing vertebrates. The profile of Hoxc9 expression inversely correlates with LMC position in land vertebrates and likely accounts for the absence of LMC neurons in limbless species such as snakes. Thus, modulation of both Hoxc9 protein function and Hoxc9 gene expression likely contributed to evolutionary transitions between undulatory and ambulatory motor circuit connectivity programs.

Regulatory proteins can bind to different sets of genomic targets in various cell types or conditions. To reliably characterize such condition-specific regulatory binding we introduce MultiGPS, an integrated machine learning approach for the analysis of multiple related ChIP-seq experiments. MultiGPS is based on a generalized Expectation Maximization framework that shares information across multiple experiments for binding event discovery. We demonstrate that our framework enables the simultaneous modeling of sparse condition-specific binding changes, sequence dependence, and replicate-specific noise sources. MultiGPS encourages consistency in reported binding event locations across multiple-condition ChIP-seq datasets and provides accurate estimation of ChIP enrichment levels at each event. MultiGPS's multi-experiment modeling approach thus provides a reliable platform for detecting differential binding enrichment across experimental conditions. We demonstrate the advantages of MultiGPS with an analysis of Cdx2 binding in three distinct developmental contexts. By accurately characterizing condition-specific Cdx2 binding, MultiGPS enables novel insight into the mechanistic basis of Cdx2 site selectivity. Specifically, the condition-specific Cdx2 sites characterized by MultiGPS are highly associated with pre-existing genomic context, suggesting that such sites are pre-determined by cell-specific regulatory architecture. However, MultiGPS-defined condition-independent sites are not predicted by pre-existing regulatory signals, suggesting that Cdx2 can bind to a subset of locations regardless of genomic environment. A summary of this paper appears in the proceedings of the RECOMB 2014 conference, April 2-5.

Efficient transcriptional programming promises to open new frontiers in regenerative medicine. However, mechanisms by which programming factors transform cell fate are unknown, preventing more rational selection of factors to generate desirable cell types. Three transcription factors, Ngn2, Isl1 and Lhx3, were sufficient to program rapidly and efficiently spinal motor neuron identity when expressed in differentiating mouse embryonic stem cells. Replacement of Lhx3 by Phox2a led to specification of cranial, rather than spinal, motor neurons. Chromatin immunoprecipitation-sequencing analysis of Isl1, Lhx3 and Phox2a binding sites revealed that the two cell fates were programmed by the recruitment of Isl1-Lhx3 and Isl1-Phox2a complexes to distinct genomic locations characterized by a unique grammar of homeodomain binding motifs. Our findings suggest that synergistic interactions among transcription factors determine the specificity of their recruitment to cell type-specific binding sites and illustrate how a single transcription factor can be repurposed to program different cell types.

Hox genes controlling motor neuron subtype identity are expressed in rostrocaudal patterns that are spatially and temporally collinear with their chromosomal organization. Here we demonstrate that Hox chromatin is subdivided into discrete domains that are controlled by rostrocaudal patterning signals that trigger rapid, domain-wide clearance of repressive histone H3 Lys27 trimethylation (H3K27me3) polycomb modifications. Treatment of differentiating mouse neural progenitors with retinoic acid leads to activation and binding of retinoic acid receptors (RARs) to the Hox1-Hox5 chromatin domains, which is followed by a rapid domain-wide removal of H3K27me3 and acquisition of cervical spinal identity. Wnt and fibroblast growth factor (FGF) signals induce expression of the Cdx2 transcription factor that binds and clears H3K27me3 from the Hox1-Hox9 chromatin domains, leading to specification of brachial or thoracic spinal identity. We propose that rapid clearance of repressive modifications in response to transient patterning signals encodes global rostrocaudal neural identity and that maintenance of these chromatin domains ensures the transmission of positional identity to postmitotic motor neurons later in development.

We present a method, LineageProgram, that uses the developmental lineage relationship of observed gene expression measurements to improve the learning of developmentally relevant cellular states and expression programs. We find that incorporating lineage information allows us to significantly improve both the predictive power and interpretability of expression programs that are derived from expression measurements from in vitro differentiation experiments. The lineage tree of a differentiation experiment is a tree graph whose nodes describe all of the unique expression states in the input expression measurements, and edges describe the experimental perturbations applied to cells. Our method, LineageProgram, is based on a log-linear model with parameters that reflect changes along the lineage tree. Regularization with L(1) that based methods controls the parameters in three distinct ways: the number of genes change between two cellular states, the number of unique cellular states, and the number of underlying factors responsible for changes in cell state. The model is estimated with proximal operators to quickly discover a small number of key cell states and gene sets. Comparisons with existing factorization, techniques, such as singular value decomposition and non-negative matrix factorization show that our method provides higher predictive power in held, out tests while inducing sparse and biologically relevant gene sets.

The study of developmentally regulated transcription factors by chromatin immunoprecipitation and deep sequencing (ChIP-seq) faces two major obstacles: availability of ChIP-grade antibodies and access to sufficient number of cells. We describe versatile genome-wide analysis of transcription-factor binding sites by combining directed differentiation of embryonic stem cells and inducible expression of tagged proteins. We demonstrate its utility by mapping DNA-binding sites of transcription factors involved in motor neuron specification.