In recent years, genome-wide profiling approaches have begun to uncover the molecular programs that drive developmental processes. In particular, technical advances that enable genome-wide profiling of thousands of individual cells have provided the tantalizing prospect of cataloging cell type diversity and developmental dynamics in a quantitative and comprehensive manner. Here, we review how single-cell RNA sequencing has provided key insights into mammalian developmental and stem cell biology, emphasizing the analytical approaches that are specific to studying gene expression in single cells.

Leptin is an adipose tissue hormone that maintains homeostatic control of adipose tissue mass by regulating the activity of specific neural populations controlling appetite and metabolism¹. Leptin regulates food intake by inhibiting orexigenic agouti-related protein (AGRP) neurons and activating anorexigenic pro-opiomelanocortin (POMC) neurons². However, whereas AGRP neurons regulate food intake on a rapid time scale, acute activation of POMC neurons has only a minimal effect^3–5. This has raised the possibility that there is a heretofore unidentified leptin-regulated neural population that rapidly suppresses appetite. Here we report the discovery of a new population of leptin-target neurons expressing basonuclin 2 (Bnc2) in the arcuate nucleus that acutely suppress appetite by directly inhibiting AGRP neurons. Opposite to the effect of AGRP activation, BNC2 neuronal activation elicited a place preference indicative of positive valence in hungry but not fed mice. The activity of BNC2 neurons is modulated by leptin, sensory food cues and nutritional status. Finally, deleting leptin receptors in BNC2 neurons caused marked hyperphagia and obesity, similar to that observed in a leptin receptor knockout in AGRP neurons. These data indicate that BNC2-expressing neurons are a key component of the neural circuit that maintains energy balance, thus filling an important gap in our understanding of the regulation of food intake and leptin action.

Although exocytosis is critical for the proper trafficking of materials to the plasma membrane, relatively little is known about the mechanistic details of post-Golgi trafficking in plants. Here, we demonstrate that the DENN (Differentially Expressed in Normal and Neoplastic cells) domain protein STOMATAL CYTOKINESIS DEFECTIVE1 (SCD1) and SCD2 form a previously unknown protein complex, the SCD complex, that functionally interacts with subunits of the exocyst complex and the RabE1 family of GTPases in Arabidopsis thaliana. Consistent with a role in post-Golgi trafficking, scd1 and scd2 mutants display defects in exocytosis and recycling of PIN2-GFP. Perturbation of exocytosis using the small molecule Endosidin2 results in growth inhibition and PIN2-GFP trafficking defects in scd1 and scd2 mutants. In addition to the exocyst, the SCD complex binds in a nucleotide state-specific manner with Sec4p/Rab8-related RabE1 GTPases and overexpression of wild-type RabE1 rescues scd1 temperature-sensitive mutants. Furthermore, SCD1 colocalizes with the exocyst subunit, SEC15B, and RabE1 at the cell plate and in distinct punctae at or near the plasma membrane. Our findings reveal a mechanism for plant exocytosis, through the identification and characterization of a protein interaction network that includes the SCD complex, RabE1, and the exocyst.

Red teaming, the practice of adversarially exposing unexpected or undesired model behaviors, is critical towards improving equity and accuracy of large language models, but non-model creator-affiliated red teaming is scant in healthcare. We convened teams of clinicians, medical and engineering students, and technical professionals (80 participants total) to stress-test models with real-world clinical cases and categorize inappropriate responses along axes of safety, privacy, hallucinations/accuracy, and bias. Six medically-trained reviewers re-analyzed prompt-response pairs and added qualitative annotations. Of 376 unique prompts (1504 responses), 20.1% were inappropriate (GPT-3.5: 25.8%; GPT-4.0: 16%; GPT-4.0 with Internet: 17.8%). Subsequently, we show the utility of our benchmark by testing GPT-4o, a model released after our event (20.4% inappropriate). 21.5% of responses appropriate with GPT-3.5 were inappropriate in updated models. We share insights for constructing red teaming prompts, and present our benchmark for iterative model assessments.