Generation of cell diversity in the developing brainstem and its modeling in vitro

Abstract

The intersection of spatial and temporal patterning programs underlies the formation of neural cell type diversity in the developing central nervous system (CNS). Deciphering the molecular mechanisms regulating positional and temporal patterning provides a basic framework for how the developing brain can be functionally assembled, and the growing knowledge about these patterning mechanisms offer a powerful tool to effectively control the differentiation of pluripotent stem cells (PSCs) into different neuronal subtypes of clinical importance.

The research described in this thesis is part of a continuing effort to define the molecular mechanisms that modulate the positional and temporal identity of immature neural stem cells (NSCs) and enables these to differentiate into specific subtypes of neurons at defined positions and over specific time-windows in the developing CNS. We have examined the sequential specification of motor neurons (MNs) and serotonergic neurons (5HTNs) by a Nkx2.2+ temporal lineage in the ventral hindbrain (HB) and outlined a three-node timer network that conceptually explains how time can be encoded by NSCs. Additionally, and unexpectedly, we show that timed exposure of retinoic acid (RA) can be applied to effectively pattern human PSC (hPSC)-derived neural progenitors into forebrain (FB), midbrain (MB), and HB regional identities. Based on this finding, we developed novel and robust differentiation protocols for production of mesencephalic dopaminergic (mDA) neurons and 5HTNs of the HB. We show that RA-specified human mDA neurons restore motor function after transplantation into a rat model of Parkinson’s disease (PD), and that mouse and human 5HTNs can be utilized as cellular platforms to screen small molecules for their capacity to modulate serotonin (5-HT) signaling in 5HTNs.

In Paper I, we address how time is encoded by NSCs in temporal patterning processes in the CNS. We focused on a region in the ventral HB where NSCs, defined by expression of the transcription factor (TF) Nkx2.2, sequentially generate MNs, 5HTNs and oligodendrocyte precursors (OLPs). Shh signaling induces the initiation of MN production through induction of the MN-determining TF Phox2b while a delayed activation of transforming growth factor β (Tgfβ) suppresses Phox2b, terminates MN production and trigger the birth of late-born 5HTNs (Dias et al., 2014). In Paper I we present a three-node incoherent feed-forward loop (IFFL) circuitry that conceptually explains how time can be measured and set in the Nkx2.2+ lineage. By applying a series of in vivo and in vitro experiments, in combination with computational modeling, we reveal a progressive decline of Gli1-3 transcription and bifunctional Gli2-3 TFs over time. Tgfβ is sensitive to transcriptional repressor forms of Gli proteins (GliR) which prohibit Tgfβ induction by Gli activators (GliA) until GliR has been titrated out. Once activated, the cell non-autonomous activity of Tgfβ counterbalances noise and facilitates a synchronous fate switch of Nkx2.2+ NSCs at the population levels.

In Paper II, we show that timed delivery of RA can be effectively applied to regionally pattern hPSCs into FB, MB and HB regional territories, in a manner resembling the previously established activity of WNT signaling. However, while WNT signaling is concentration sensitive, it is the duration of RA exposure that is crucial for regional patterning and the response of cells is relatively insensitive to altered RA concentrations. By combining RA- and Shh-signaling we could robustly direct the differentiation of hPSCs into mDA neurons, whose loss underlie motor deficits in PD. When grafted into the striatum of parkinsonian rats, RA-specified cell preparations engraft, differentiate into functional mDA neurons and relieve motor deficits. These data provide proof-of-concept that RA-based protocol for mDA generation could provide a new and alternative route for cell replacement therapy for PD. In addition to mDA neurons, we show that extended exposure of hPSCs to RA results in efficient generation of cranial human MNs and human 5HTNs, suggesting that RA-based regional patterning can be applied to generate several types of clinically relevant neurons from hPSCs.

In Paper III, we utilized protocols developed by Dias et al., 2014 and in Paper II to produce mouse and human 5HTNs, which dysfunction is strongly linked to various neuropsychiatric disorders and are the target of most prescribed antidepressants. After adaptation of protocols to a screenable format and high-content imaging, we performed an unbiased phenotypic screen to identify small molecules that modulate 5-HT signaling in mouse and human 5HTNs. Out of ~5200 annotated small molecules, we identified and confirmed ~200 hits that modulated 5-HT content in mouse neurons with ~70% of these showing a similar phenotypic response on human 5HTNs. Many hits had previously been associated with the monoaminergic system, but many compounds were not obviously connected to 5HTNs. Among those were the muscarinic acetylcholine receptor (mAChR) antagonist oxybutynin which promoted a notable increase of neuronal 5-HT content and which in subsequent secondary assays acted as an inhibitor of monoamine oxidases in 5HTNs. These data provide proof-of-concept that phenotypic screening on stem cell-derived 5HTNs is a powerful tool to identify new types of compounds with potential antidepressant properties.