Integrating deep learning and unbiased automated high-content screening to identify complex disease signatures in human fibroblasts.

Drug discovery for diseases such as Parkinson's disease are impeded by the lack of screenable cellular phenotypes. We present an unbiased phenotypic profiling platform that combines automated cell culture, high-content imaging, Cell Painting, and deep learning. We applied this platform to primary fibroblasts from 91 Parkinson's disease patients and matched healthy controls, creating the largest publicly available Cell Painting image dataset to date at 48 terabytes. We use fixed weights from a convolutional deep neural network trained on ImageNet to generate deep embeddings from each image and train machine learning models to detect morphological disease phenotypes. Our platform's robustness and sensitivity allow the detection of individual-specific variation with high fidelity across batches and plate layouts. Lastly, our models confidently separate LRRK2 and sporadic Parkinson's disease lines from healthy controls (receiver operating characteristic area under curve 0.79 (0.08 standard deviation)), supporting the capacity of this platform for complex disease modeling and drug screening applications.

Disease-associated mutation in SRSF2 misregulates splicing by altering RNA-binding affinities.

Serine/arginine-rich splicing factor 2 (SRSF2) is an RNA-binding protein that plays important roles in splicing of mRNA precursors. SRSF2 mutations are frequently found in patients with myelodysplastic syndromes and certain leukemias, but how these mutations affect SRSF2 function has only begun to be examined. We used clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein-9 nuclease to introduce the P95H mutation to SRSF2 in K562 leukemia cells, generating an isogenic model so that splicing alterations can be attributed solely to mutant SRSF2. We found that SRSF2 (P95H) misregulates 548 splicing events (<1% of total). Of these events, 374 involved the inclusion of cassette exons, and the inclusion was either increased (206) or decreased (168). We detected a specific motif (UCCA/UG) enriched in the more-included exons and a distinct motif (UGGA/UG) in the more-excluded exons. RNA gel shift assays showed that a mutant SRSF2 derivative bound more tightly than its wild-type counterpart to RNA sites containing UCCAG but bound less tightly to UGGAG sites. Thus in most cases the pattern of exon inclusion or exclusion correlated with stronger or weaker RNA binding, respectively. We further show that the P95H mutation does not affect other functions of SRSF2, i.e., protein-protein interactions with key splicing factors. Our results thus demonstrate that the P95H mutation positively or negatively alters the binding affinity of SRSF2 for cognate RNA sites in target transcripts, leading to misregulation of exon inclusion. Our findings shed light on the mechanism of the disease-associated SRSF2 mutation in splicing regulation and also reveal a group of misspliced mRNA isoforms for potential therapeutic targeting.