RNA and membraneless organelles: Membraneless compartments can form in cells through liquidliquid phase separation (see the Perspective by Polymenidou). But what prevents these cellular condensates from randomly fusing together? Using the RNA-binding protein (RBP) Whi3, Langdon et al. demonstrated that the secondary structure of different RNA components determines the distinct biophysical and biological properties of the two types of condensates that Whi3 forms. Several RBPs, such as FUS and TDP43, contain prion-like domains and are linked to neurodegenerative diseases. These RBPs are usually soluble in the nucleus but can form pathological aggregates in the cytoplasm. Maharana et al. showed that local RNA concentrations determine distinct phase separation behaviors in different subcellular locations. The higher RNA concentrations in the nucleus act as a buffer to prevent phase separation of RBPs; when mislocalized to the cytoplasm, lower RNA concentrations trigger aggregation. Science , this issue p. 922 , p. 918 ; see also p. 859

Phenotypic drug discovery requires billions of cells for high-throughput screening (HTS) campaigns. Because up to several million different small molecules will be tested in a single HTS campaign, even small variability within the cell populations for screening could easily invalidate an entire campaign. Neurodegenerative assays are particularly challenging because neurons are post-mitotic and cannot be expanded for implementation in HTS. Therefore, HTS for neuroprotective compounds requires a cell type that is robustly expandable and able to differentiate into all of the neuronal subtypes involved in disease pathogenesis. Here, we report the derivation and propagation using only small molecules of human neural progenitor cells (small molecule neural precursor cells; smNPCs). smNPCs are robust, exhibit immortal expansion, and do not require cumbersome manual culture and selection steps. We demonstrate that smNPCs have the potential to clonally and efficiently differentiate into neural tube lineages, including motor neurons (MNs) and midbrain dopaminergic neurons (mDANs) as well as neural crest lineages, including peripheral neurons and mesenchymal cells. These properties are so far only matched by pluripotent stem cells. Finally, to demonstrate the usefulness of smNPCs we show that mDANs differentiated from smNPCs with LRRK2 G2019S are more susceptible to apoptosis in the presence of oxidative stress compared to wild-type. Therefore, smNPCs are a powerful biological tool with properties that are optimal for large-scale disease modeling, phenotypic screening, and studies of early human development.

Amyotrophic lateral sclerosis (ALS) is a rapidly progressive neurodegenerative disorder due to selective loss of motor neurons (MNs). Mutations in the fused in sarcoma (FUS) gene can cause both juvenile and late onset ALS. We generated and characterized induced pluripotent stem cells (iPSCs) from ALS patients with different FUS mutations, as well as from healthy controls. Patient-derived MNs show typical cytoplasmic FUS pathology, hypoexcitability, as well as progressive axonal transport defects. Axonal transport defects are rescued by CRISPR/Cas9-mediated genetic correction of the FUS mutation in patient-derived iPSCs. Moreover, these defects are reproduced by expressing mutant FUS in human embryonic stem cells (hESCs), whereas knockdown of endogenous FUS has no effect, confirming that these pathological changes are mutant FUS dependent. Pharmacological inhibition as well as genetic silencing of histone deacetylase 6 (HDAC6) increase α-tubulin acetylation, endoplasmic reticulum (ER)–mitochondrial overlay, and restore the axonal transport defects in patient-derived MNs.

Modeling Parkinson’s disease (PD) using advanced experimental in vitro models is a powerful tool to study disease mechanisms and to elucidate unexplored aspects of this neurodegenerative disorder. Here, we demonstrate that three-dimensional (3D) differentiation of expandable midbrain floor plate neural progenitor cells (mfNPCs) leads to organoids that resemble key features of the human midbrain. These organoids are composed of midbrain dopaminergic neurons (mDANs), which produce and secrete dopamine. Midbrain-specific organoids derived from PD patients carrying the LRRK2-G2019S mutation recapitulate disease-relevant phenotypes. Automated high-content image analysis shows a decrease in the number and complexity of mDANs in LRRK2-G2019S compared to control organoids. The floor plate marker FOXA2, required for mDAN generation, increases in PD patient-derived midbrain organoids, suggesting a neurodevelopmental defect in mDANs expressing LRRK2-G2019S. Thus, we provide a robust method to reproducibly generate 3D human midbrain organoids containing mDANs to investigate PD-relevant patho-mechanisms.