University of Stuttgart
ORCID: 0000-0002-3676-0293Publishes on Topic Modeling, Surface Modification and Superhydrophobicity, Natural Language Processing Techniques. 31 papers and 704 citations.
Add your photo, update your bio, and get notified when your ranking changes.
Amyotrophic lateral sclerosis (ALS) is a lethal disease characterized by motor neuron degeneration and associated with aggregation of nuclear RNA-binding proteins (RBPs), including FUS. How FUS aggregation and neurodegeneration are prevented in healthy motor neurons remain critically unanswered questions. Here, we use a combination of ALS patient autopsy tissue and induced pluripotent stem cell-derived neurons to study the effects of FUS mutations on RBP homeostasis. We show that FUS' tendency to aggregate is normally buffered by interacting RBPs, but this buffering is lost when FUS mislocalizes to the cytoplasm due to ALS mutations. The presence of aggregation-prone FUS in the cytoplasm causes imbalances in RBP homeostasis that exacerbate neurodegeneration. However, enhancing autophagy using small molecules reduces cytoplasmic FUS, restores RBP homeostasis and rescues motor function in vivo. We conclude that disruption of RBP homeostasis plays a critical role in FUS-ALS and can be treated by stimulating autophagy.
Despite impressive performance in many benchmark datasets, AI models can still make mistakes, especially among out-of-distribution examples. It remains an open question how such imperfect models can be used effectively in collaboration with humans. Prior work has focused on AI assistance that helps people make individual high-stakes decisions, which is not scalable for a large amount of relatively low-stakes decisions, e.g., moderating social media comments. Instead, we propose conditional delegation as an alternative paradigm for human-AI collaboration where humans create rules to indicate trustworthy regions of a model. Using content moderation as a testbed, we develop novel interfaces to assist humans in creating conditional delegation rules and conduct a randomized experiment with two datasets to simulate in-distribution and out-of-distribution scenarios. Our study demonstrates the promise of conditional delegation in improving model performance and provides insights into design for this novel paradigm, including the effect of AI explanations.
In amyotrophic lateral sclerosis (ALS) motor neurons (MNs) undergo dying-back, where the distal axon degenerates before the soma. The hexanucleotide repeat expansion (HRE) in C9ORF72 is the most common genetic cause of ALS, but the mechanism of pathogenesis is largely unknown with both gain- and loss-of-function mechanisms being proposed. To better understand C9ORF72-ALS pathogenesis, we generated isogenic induced pluripotent stem cells. MNs with HRE in C9ORF72 showed decreased axonal trafficking compared with gene corrected MNs. However, knocking out C9ORF72 did not recapitulate these changes in MNs from healthy controls, suggesting a gain-of-function mechanism. In contrast, knocking out C9ORF72 in MNs with HRE exacerbated axonal trafficking defects and increased apoptosis as well as decreased levels of HSP70 and HSP40, and inhibition of HSPs exacerbated ALS phenotypes in MNs with HRE. Therefore, we propose that the HRE in C9ORF72 induces ALS pathogenesis via a combination of gain- and loss-of-function mechanisms.
Amyotrophic lateral sclerosis (ALS) arises from an interplay of genetic mutations and environmental factors. ssRNA viruses are possible ALS risk factors, but testing their interaction with mutations such as in FUS, which encodes an RNA-binding protein, has been difficult due to the lack of a human disease model. Here, we use isogenic induced pluripotent stem cell (iPSC)-derived spinal neurons to investigate the interaction between ssRNA viruses and mutant FUS. We find that rabies virus (RABV) spreads ALS phenotypes, including the formation of stress granules with aberrant composition due to increased levels of FUS protein, as well as neurodegeneration and reduced restriction activity by FUS mutations. Consistent with this, iPSC-derived spinal neurons harboring mutant FUS are more sensitive to HIV-1 and Zika viruses. We demonstrate that RABV and HIV-1 exacerbate cytoplasmic mislocalization of FUS. Our results demonstrate that viral infections worsen ALS pathology in spinal neurons with genetic risk factors, suggesting a novel role for viruses in modulating patient phenotypes.