Publications

Mafb and c-Maf transcription factors (TFs) are expressed in medial ganglionic eminence (MGE) lineages, beginning in progenitors and continuing into mature GABAergic parvalbumin-positive (PV+) and somatostatin-positive (SST+) cortical interneurons (CINs). Deleting Mafb and c-Maf in MGE before SST versus PV fate specification causes developmental anomalies, including altered numbers of CINs and seizure phenotypes, but the specific contributions of these TFs in postmitotic SST+ and PV+ CINs to epilepsy remain unknown. To address this, we conditionally deleted Mafb or c-Maf in SST+ or PV+ interneurons after interneuron fate specification in female and male mice. Deletion of c-Maf, but not Mafb, in SST+ cells was associated with reduced synaptic excitation onto these cells and with spontaneous spike-and-wave discharges, consistent with absence-like seizures. In contrast, deletion of Mafb in SST+ CINs reduced their density in superficial cortical layers but did not induce epilepsy. Neither c-Maf nor Mafb deletion in PV+ CINs produced major electrophysiological or histological abnormalities in the somatosensory cortex. These findings identify a specific requirement for c-Maf in modulating synaptic excitation of SST+ interneurons and show that its loss in SST+ cells is associated with the development of absence-like epileptiform activity in vivo. Together, our results refine the understanding of how transcriptional programs shape interneuron function in the mature cortex and highlight c-Maf/MAF-dependent pathways as candidates for investigation in epilepsy genetics.

DNA double-stranded breaks (DSBs) are toxic events that can be reversed without genetic information loss by homology-directed repair (HDR), wherein information is copied from an intact template molecule. Finding a correct template within millions to billions of other DNA bases is termed homology search and is mediated by the protein RAD51. To monitor transient search in cellulo, we develop RAD51 proximity identification sequencing (RaPID-seq), a highly sensitive method marking all DNA searched regardless of whether it is chosen as the final template. We find that HDR in human cells is hierarchical with DSB proximity constraining the search space from which sequence homology determines the chosen template. Exogenously introduced DNA templates, such as those used during genome editing, are unconstrained and efficiently searched by the DSB, thereby competing with endogenous template search. Our data reveal the invisible process of homology search and shed new light on fundamental mechanisms underlying genome editing.

Maternal iron deficiency (ID) affects nearly 85% of all pregnancies worldwide by the third trimester and is strongly linked with maternal morbidity, poor birth outcomes, and cognitive and behavioral developmental deficits in children. Despite its burden on maternofetal health, no guidelines exist for routine screening of ID in pregnant individuals in the United States. The World Health Organization uses a diagnostic (ie, nonscreening) ferritin threshold of 15 μg/L for ID in the first trimester of pregnancy, with no guidance for the second and third trimesters. Routine screening for ID in pregnancy could identify a large proportion of mothers who remain undiagnosed and untreated under the current clinical status quo. Accordingly, we conducted, to our knowledge, the first cost-effectiveness analysis of ferritin-based ID screening in pregnant women in the United States, comparing thresholds of (1) 30 μg/L vs (2) 15 μg/L vs (3) no screening from a modified US societal perspective accounting for wages lost to infusion time, infusion administration costs, and the cost of annual hematology follow-up. We found that screening at 30 μg/L was the preferred strategy across extensive base-case, threshold, and sensitivity analyses and across all examined willingness-to-pay thresholds, with an incremental cost-effectiveness ratio of $34 000/quality-adjusted life year (QALY; 95% credible interval, $32 000-$36 200/QALY) compared with no screening. No parameter variation changed this outcome, and screening at 30 μg/L was favored in 100% of 10 000 Monte Carlo iterations. Population-level regular second- and third-trimester screening for ID using physiologically informed ferritin ranges should be considered for all pregnant women in the United States.

Artificial intelligence (AI) is rapidly transforming stem cell and developmental biology, offering new strategies to analyze, interpret and optimize complex, dynamic systems such as organoids and stem cell-derived embryo models. In this Perspective, we chart the integration of AI into image-based analysis of stem cell systems, highlighting how deep learning, convolutional neural networks and emerging foundation models enable automated classification, segmentation and phenotyping at increasing scale and precision. We showcase applications in phenotyping, drug screening and mechanistic discovery, including real-time fate prediction and the identification of hidden morphological signatures linked to differentiation and disease. Practical challenges, including limited annotated data, model interpretability and live imaging constraints, are examined alongside future opportunities, such as multimodal integration, real-time experimental steering and protocol optimization. Altogether, we argue that AI is not merely an analytical tool, but a discovery engine that enhances reproducibility, accelerates insight and brings us closer to a mechanistic understanding of self-organization in complex stem cell-derived systems.

Nonmyeloablative-related haploidentical allogeneic stem cell transplantation (NMAC-HID allo-HSCT) has emerged as an additional treatment to achieve durable remission in sickle cell disease (SCD), a prevalent blood disorder characterized by painful vaso-occlusive crises and chronic anemia. The standard of care (SOC) for SCD includes hydroxyurea, pain management, and blood transfusion, but patients with SCD still lose several decades of life expectancy. Gene therapy (GT) for SCD is the other treatment for lifelong disease amelioration in SCD, with accessibility limited by cost and manufacturing capacity in the United States and globally. Two recent prospective studies that evaluated NMAC-HID allo-HSCT validated haploidentical allotransplantation as an efficacious and accessible treatment option in the era of GT. Given the ongoing price negotiation across jurisdictions for GT implementation and the absence of cost-effectiveness data comparing NMAC-HID allo-HSCT and GT, we conducted a cost-effectiveness analysis of NMAC-HID allo-HSCT vs GT vs SOC for adults and children living with SCD. The primary outcomes were the incremental cost-effectiveness ratio and the net monetary benefits across these 3 strategies. The secondary outcome was the maximum cost-effective threshold price for GT compared with NMAC-HID allo-HSCT. Treatment with SOC, NMAC-HID allo-HSCT, and GT accrued 14.3, 20.1, and 22.1 quality-adjusted life-years at costs of $1.22 million, $1.15 million, and $2.75 million, respectively. NMAC-HID allo-HSCT was the cost-effective strategy compared with GT in 100% of 10 000 Monte Carlo iterations across the base case and all scenario analyses. The maximum cost-effective thresholds for GT vs SOC were $1.4 million in the United States and $4200 to $22 000 across India, Nigeria, and Tanzania, depending on willingness-to-pay thresholds.

Neanderthals had a robust build and distinctive skeletal features. The hypothalamic-pituitary-somatotropic (HPS) axis, with growth hormone (GH) as a central signaling molecule, plays a crucial role in regulating skeletal development. To explore whether Neanderthal-specific genetic variations in the HPS axis contributed to their physical robustness, we examined genes encoding the relevant receptors and hormones. We find that the Neanderthal growth hormone receptor (GHR) carried two amino acid changes and one deletion. When the Neanderthal GHR is expressed in a GH-dependent cell line, the cells proliferate faster than cells expressing the modern human GHR when stimulated by pituitary GH but not by placental GH. We also show that some present-day humans have inherited the gene encoding the Neanderthal GHR. These individuals tend to have more muscle mass and exhibit some craniofacial traits reminiscent of Neanderthals. Thus, aspects of Neanderthal anatomy live on in people today.

How the spatial arrangement of synaptic inputs shapes neuronal feature selectivity remains a fundamental question. Here, we map the three-dimensional distribution of excitatory and inhibitory synapses across the dendritic arbor of CA1 pyramidal neurons in vivo and build biophysical models to probe their impact on place-cell emergence. Excitatory synapses are non-uniformly distributed, forming structural clusters preferentially on terminal apical and basal dendrites, whereas inhibitory synapses are uniformly arranged. Relative to dispersed configurations, clustered inputs generate higher-quality, stable place fields while recruiting ∼13% fewer active synapses for equivalent somatic output, and cause elevated voltage-gated calcium influx and NMDA-receptor activation. Notably, disrupting clustering permits recovery of somatic excitability but not dendritic calcium dynamics, implicating clustering in calcium-dependent plasticity. Synaptic organization further determines integration strategy: clustered inputs preferentially engage apical dendritic nonlinearities, whereas distributed inputs rely on basal summation. These results establish synaptic clustering as a core mechanism for efficient, compartmentalized spatial computation.

A central goal in sensory science is to establish quantitative mappings between physical stimuli and perceptual experience. Although such mappings are well defined in vision and audition, they remain elusive in olfaction, particularly for complex odor mixtures. Here, we show that perceptual distances between odor mixtures can be predicted with high fidelity and are unexpectedly well captured by a compact semantic space derived from single-molecule representations. In the Dialogue for Reverse Engineering Assessment and Methods Olfactory Mixtures Prediction Challenge, we assembled a unified dataset of odor-mixture pairs, benchmarked predictions on a hidden test set of 46 pairs, and integrated the top-performing models into a postchallenge ensemble. This model outperformed existing state-of-the-art approaches on the hidden test set, reducing RMSE by about 33% to 0.08 and increasing Pearson correlation by 53% to 0.57, and maintained strong performance on an independent validation set of 50 newly designed mixture pairs. An ensemble, retaining only olfactory semantic features for each model included, further improved predictions, raising the Pearson correlation by 7% to 0.61 on the test set and by 15% to 0.54 on the validation set. Given that semantic features were extracted from pure molecules, it suggests that mixture perception may not require fundamentally different representational principles from single-molecule olfaction. Together, these results establish a reproducible quantitative framework for olfactory mixture perception and advance efforts to measure, model, and engineer smell.