Traditional animal-based drug discovery has high failure rates, prompting the search for and adoption of human-centered new approach methodologies (NAMs). Rapid advances in stem cell-, organoid-, and in silico-based NAMs, supported by evolving frameworks from the National Institutes of Health (NIH) and the Food and Drug Administration (FDA), now span the entire drug-discovery continuum, from disease modeling and drug design to efficacy testing. Our review highlights recent progress across these domains, including the identification of therapeutic candidates, the development of cutting-edge models and technologies, and the potential of NAMs themselves as treatments in preclinical and clinical contexts, while examining the key biological, technical, and regulatory bottlenecks that need to be addressed to enable robust translational adoption. We conclude by discussing the translational and societal considerations essential to the responsible adoption of NAMs and outlining future human-centric pipelines poised to redefine the landscape of drug discovery.

Artificial intelligence has rapidly advanced molecular science, yet progress has largely unfolded through specialized models tailored to specific tasks. In this issue of Cell, Peng et al. introduce PocketXMol, a unified 3D generative framework that reframes molecular design problems as conditional reconstruction of atomic interactions. Demonstrations across design scenarios highlight its potential in molecular engineering.

Nicotine, tobacco's addictive and potent insecticidal alkaloid, has shaped human history, agriculture, and the plants that produce it. However, the enzymatic steps and reaction mechanisms involved in nicotine biosynthesis remain elusive. Here, we reveal that the final coupling reaction is stabilized by glycosylation via a uridine diphosphate (UDP)-glycosyltransferase, reduced and activated by an A622, condensed through a stereoselective intermolecular Mannich-like reaction, sequentially oxidized by a berberine bridge enzyme-like (BBL), and finally deglycosylated by a β-glucosidase to yield nicotine. A 5-component metabolon assembles at vacuolar membranes to channel both nicotine biosynthesis and its transport. We reconstituted this metabolon both in vitro and heterologously in vivo. Abrogating any of these components depletes nicotine accumulations. A multidrug and toxic compound extrusion (MATE) transporter is essential for efficiently engineering nicotine production in heterologous plant species, which confers pest resistance. This work completes the nicotine biosynthesis pathway and provides critical insights into the intermolecular Mannich-like reaction, a fundamental mechanism for scaffold formation in many plant alkaloids.

The tissue-level processes underpinning metastatic outgrowth remain unclear. We combined single-cell RNA sequencing, spatial transcriptomics, and AI-supported 3D imaging in human breast cancer with functional investigations in mice to uncover a 3D morphogenetic process essential for macrometastatic expansion. Macrometastases pervasively activate a metastatic trabecular morphogenesis (MTM) gene-expression program that redeploys developmental branching morphogenesis to build macrometastases as a 3D trabecular lattice of epithelial cords. MTMHIGH cells pre-exist in primary tumors destined to metastasize, whereas MTMLOW primaries are non-metastatic and display a compact, expansile growth architecture. Chromatin immunoprecipitation sequencing (ChIP-seq) on metastatic organoids identifies ETV1/4/5 as master regulators of MTM and branching cancer morphogenesis, required for metastatic outgrowth but dispensable for primary tumor take, bulk growth, and initial metastatic dissemination. Spatial and functional analyses reveal stromal fibroblast growth factor (FGF)→fibroblast growth factor receptor (FGFR) signaling as an actionable MTM dependency. Thus, we link metastatic outgrowth to a 3D developmental morphogenetic process, exposing therapeutic vulnerabilities specific to the lethal macrometastatic stage.

Many viruses can switch between lytic replication and dormancy (or lysogeny). It was recently discovered that some viruses that infect bacteria (known as bacteriophage or phage) employ peptide-based ("arbitrium") communication systems to optimize their lysis/lysogeny switch; high peptide concentrations signal a lack of susceptible hosts and trigger lysogeny, while low peptide concentrations signal an abundance of uninfected hosts and prompt lysis. Here, we demonstrate that arbitrium phages belonging to different species and genera can influence each other's infection dynamics by secreting similar communication peptides, leading to early lysogenization of the signal-receiving phage and elevated fitness of the signal-emitting phage. Antagonistic coevolution between signal-emitting and signal-receiving phages to manipulate each other's infection behaviors may explain the rapid diversification of arbitrium systems and their frequent horizontal exchange to escape the noise of crosstalk.

Arbitrium is a communication system that helps bacteriophages to decide between lysis and lysogeny through secreted peptides. In this system, the arbitrium communication peptide (AimP) binds its cognate arbitrium receptor (AimR) to repress aimX (a negative regulator of lysogeny) expression, promoting lysogeny. It has been assumed that each AimR responds exclusively to its own AimP. Here, we challenge this view by demonstrating cross-communication between arbitrium systems. Using prototypical arbitrium phages, we show that AimP peptides can bind and repress non-cognate AimR receptors, promoting lysogeny and reducing prophage induction. Structural and biochemical analyses reveal conserved receptor features that permit cross-recognition of non-cognate peptides while preserving recognition of cognate partners. In mixed lysogenic cultures, these interactions alter induction outcomes, underscoring their ecological significance. Extending to infection contexts, we demonstrate that crosstalk favors lysogeny of incoming phages in cells harboring compatible systems. These findings establish that phages engage in cross-species communication via peptide signaling, reshaping microbial communities in unexpected ways.

Although kinase activators hold significant therapeutic promise, their development remains challenging and rarely achieved. Here, we report the discovery of direct small-molecule activators of p21-activated kinase-1 (PAK1), a key regulator of cardiac homeostasis, using a rational peptide-guided strategy. Targeting PAK1 autoinhibitory regulation, we identified a previously unrecognized autoinhibition-release site between the autoregulatory region and the kinase domain. Subsequent high-throughput screening and medicinal chemistry optimization yielded selective allosteric activators that enhance PAK1 activity with micromolar potency and isoform selectivity. Structural and mechanistic analyses indicate that these activators disrupt autoinhibitory regulation and promote local and global conformational transitions to the active state. Enhanced PAK1 signaling was confirmed in cardiac cells, and in vivo studies demonstrated therapeutic efficacy in both inherited and acquired cardiac hypertrophy. Collectively, these findings establish rational modulation of kinase autoinhibitory regulation as a potential strategy for the broader discovery of kinase activators, a largely unexplored area of therapeutic development.

Delineating how acquired nutrients are partitioned into different intracellular pathways and how these various fates support distinct functions in T cells is limited. We show that CD8+ T cells acquire cysteine to serve both as a substrate for glutathione (GSH) production, which modulates effector functions, and to cede its sulfur for NFS1-dependent FeS cluster synthesis, which supports proliferation. NFS1 deletion in activated CD8+ T cells promotes exhaustion and dampens anti-cancer immunity, whereas blocking cysteine flux into GSH or enforcing FeS metabolism enhances tumor control. This role for disrupted FeS metabolism in T cell exhaustion is echoed in data from human hepatocellular carcinoma. Elucidating how different intracellular pathways use cysteine enables targeted control of cysteine flux to retain the beneficial effects of cysteine while abolishing those that restrain function. We illustrate this concept for one metabolite, cysteine, but it is likely to apply to other metabolites relevant for immune cell function.

Chimeric antigen receptor (CAR) T cells have transformed hematologic cancer therapy but remain limited in solid tumors by antigen heterogeneity and a suppressive, pro-fibrotic microenvironment. We previously identified the urokinase plasminogen activator receptor (uPAR) as upregulated in senescent, pro-fibrotic cells and showed that uPAR-directed CAR T cells could safely reverse fibrosis in mice. Integrative analyses now reveal that uPAR is broadly expressed in solid tumors enriched for TP53 and RAS pathway mutations. These tumors adopt a progenitor-like state supported by a niche of uPAR-positive stromal cells with senescence features. Human uPAR CAR T cells eliminate tumor cells and their stromal support, induce durable regressions across diverse models, eradicate systemic metastases, and are potentiated by senescence-inducing therapies. Importantly, these cells achieve robust antitumor activity without sustained myelosuppression in mice reconstituted with human immune systems. Together, these findings establish uPAR as a broadly applicable CAR T target capable of overcoming major barriers in solid tumor therapy.

The laboratory mouse is a key model system for biomedical research, yet body-wide measurement tools are lacking. Here, we generate spatiotranscriptomics profiles of whole-mouse sections that accurately capture histological regions. We spatially assign 379 cell types across whole-mouse section profiles by building a reference dataset of 59M single cells coupled with a scalable computational method for cell-type assignment. Moreover, we exploit these whole-mouse profiles to create a machine learning pipeline, LABEL, which enables pan-body annotation of tissues and cell types on histology images from H&E-stained sections. Lastly, we apply whole-mouse spatial profiling to map systemic inflammation in endotoxemia, which delineates organism-wide changes in gene expression programs across tissues and cell types. Together, our work paves the way for body-wide studies of the molecular and cellular processes that govern the organismal biology of the laboratory mouse across space, time, and conditions.

Linking genetic data with electronic health records in hospital biobanks promises to advance precision medicine, but limited ancestral diversity constrains discovery and generalizability. We analyzed 93,936 participants from the UCLA ATLAS Community Health Initiative to inform disease prevalence and genetic risk across five continental and 36 fine-scale ancestry groups. We discovered numerous unreported gene-phenotype associations, including FN3K with intestinal disaccharidase deficiency in Europeans and admixed Americans. Polygenic scores (PGS) robustly predicted common diseases, with effects markedly diminished in non-Europeans. Furthermore, we reduced the pronounced European bias in curated clinical variants using computational predictors, uncovering unreported disease-gene associations, including ANKZF1 and peripheral vascular disease in African Americans. Longitudinal data revealed that semaglutide efficacy varies across ancestries, is associated with PGS for type 2 diabetes, and is modulated by genetic variation in PTPRU. These findings illustrate how ancestrally diverse biobanks from a single health system yield robust disease associations and pharmacogenomic insights.

Stimulator of interferon genes (STING) activation requires coat protein complex II (COPII)-mediated endoplasmic reticulum (ER) exit, but the mechanism remains elusive. Here, we identify EEΦxΦ (339EEVTV343 in human STING) as the ER-exit motif recognized by SEC24 homolog C (SEC24C). Using AlphaFold3, we present a predicted structure of SEC24C binding to a STING dimer, revealing the EEΦxΦ motif in a previously structurally unresolved region. Mutations in this motif or the SEC24C cargo-binding site disrupt STING trafficking and signaling. Our findings support a STING oligomerization and avidity threshold model that explains regulated ER exit. The EEΦxΦ motif is conserved in vertebrate STING homologs and is sufficient to mediate ER exit of unrelated proteins. Interestingly, the STING ER-exit motif is suboptimal compared with known SEC24C cargos, which is crucial for preventing immune overactivation. An engineered "super-ER-exit" STING is constitutively active and induces potent antitumor immunity. Tandem repeats of this motif competitively inhibit endogenous STING signaling. Collectively, this study elucidates the STING-ER-exit mechanism and presents strategies for modulating STING signaling.

Piaget's theory posits that children develop structured knowledge schemas for inferring and assimilating new information, yet the underlying neural mechanisms remain unclear. In 203 participants aged 8-25 years, we investigated how maturation of a two-dimensional (2D) knowledge map underpins inferential reasoning and knowledge assimilation. Grid-cell-like codes in the entorhinal cortex (EC) strengthened with age, reflecting schema representations in non-spatial conceptual spaces, and predicted improved inferential reasoning. These grid-like codes also supported the medial prefrontal cortex (mPFC) in encoding distance relationships between objects on the 2D map. As participants assimilated new information, they integrated it into existing grid patterns in the EC. Moreover, the maturation of these neural codes tracked real-world intelligence measures, particularly reasoning abilities. Our findings demonstrate that the development of non-spatial grid-like neural codes offers a mechanistic account of cognitive development, bridging psychological theory with a fundamental cellular representation of the cognitive map.

Neurodegenerative diseases (NDs) pose clinical challenges due to their complexity and molecular heterogeneity. Here, we present a pan-neurodegeneration atlas (PanNDA) from multilayer, deep proteomic analysis of 2,279 human brain samples spanning 6 major NDs: Alzheimer's disease (AD), Lewy body dementia (LBD), frontotemporal lobar degeneration with TDP-43 pathology, progressive supranuclear palsy with tau pathology, vascular dementia, and Parkinson's disease. PanNDA integrates data from whole proteome, detergent-insoluble proteome, and posttranslational modifications (phosphorylation and ubiquitination), enabling intra- and inter-disease comparisons. Intra-disease analyses uncover distinct molecular subtypes (e.g., three in AD and four in LBD), reveal dysregulated pathways, and prioritize top-ranked proteins. Inter-disease comparisons identify shared alterations in NDs, such as GPNMB in microglial and lysosomal activation and NPTX2 in synaptic regulation, alongside disease-specific changes and hub regulators within protein networks. Overall, PanNDA provides a systems-level framework for understanding ND mechanisms and serves as a foundational resource that is accessible via an interactive website: https://penglab.shinyapps.io/pannda.

Some mammalian tissues can replace lost cells within one lineage, but organ-level regeneration-restoring diverse cell types across lineages-remains rare. Here, we show that late embryonic full-thickness skin injuries heal by regenerating epithelial, mesenchymal, neuronal, and vascular tissues with proper connectivity. However, this ability is lost soon after birth, resulting in failure to restore most cell types and hyperinnervation within the wound bed. Single-cell sequencing identified a postnatal wound-specific fibroblast (PWF) population absent after embryonic wounding. Through an in vivo screen, we discovered that three PWF-enriched genes-Timp1, Cxcl12, and Ccl7-inhibit organ-level regeneration and cause hyperinnervation when overexpressed in embryonic wounds. Reducing hyperinnervation in postnatal wounds through the depletion of Cxcl12 in fibroblasts or nerve ablation enables regeneration of diverse lineages after injury. Our study identifies mechanisms that transition an organ from regenerative to non-regenerative, discovers fibroblast-driven hyperinnervation as a key barrier, and demonstrates that removing this barrier unlocks organ-level regeneration.

To define and systematically characterize the human E3 ubiquitin ligase (E3) landscape, we generated the E3-ome, a compendium of E3s encoded by the human genome. The E3-ome integrates experimental data, bioinformatics, and published research, revealing 672 high-confidence E3s. We standardized E3 classifications to create a unified framework for annotation and comparative analysis. The E3-ome identified several previously unrecognized domains, motifs, E3 candidates, and relationships, expanding the diversity of E3s. Furthermore, the E3-ome mapped the spatial and physiological organization of E3s across human tissues and cell types, revealing context-dependent E3s. Genetic analyses identified disease-associated variants across the E3-ome, linking E3s to diverse human pathologies. Together, these analyses define the human E3 landscape at high resolution and deliver a foundational resource to drive mechanistic and therapeutic discovery.

Nucleotide-binding, leucine-rich repeat (NLR) receptors are widespread intracellular immune sensors across kingdoms. Plant G10-type coiled-coil (CCG10)-NLRs constitute a distinct phylogenetic clade that remains poorly characterized. Here, we identified a gain-of-function mutant of wheat autoimmunity 3 (WAI3GOF), which encodes a constitutively active CCG10-NLR resulting from a residue substitution in the leucine-rich repeat (LRR) domain. Cryo-electron microscopy (cryo-EM) analysis reveals that activated WAI3 assembles into a distinctive octameric resistosome. Arabidopsis RPS2, another CCG10-NLR, also forms an octamer, indicating a conserved structural property across monocot and dicot plants. The WAI3 resistosome induces a prolonged and sustained increase in cytosolic calcium, likely facilitated by a unique channel architecture arising from its divergent coiled-coil (CC) domain configuration. Notably, this domain arrangement may be shared by plant NLRs that lack the conserved EDVID (Glu-Asp-Val-Ile-Asp) motif in their CC domains. Together, our findings uncover a conserved yet previously uncharacterized NLR resistosome structure and provide insights into the plant immune receptor plasticity.

How morphological diversity arises from variations in biomechanical processes remains an open question. Although forces shape tissues, how force-generating systems differ across species to create diverse forms is unclear. Here, we combine comparative morphogenesis and active matter theory across six cnidarian species spanning 500 million years of divergence to identify the mechanical basis of larval shape diversity. We define species-specific configurations of mechanical modules-termed mechanotypes-that quantitatively predict larval shapes across taxa. We find that shape elongation is a simple trait at the mesoscale level, as its variation depends on one mechanical module, whereas shape polarity is a complex trait dependent on several modules. Perturbations mimicking interspecies regulatory differences reshape these modules, reprogramming larval morphology into forms resembling sister species. By establishing a mesoscale mechanical framework for cross-species comparison, this work reveals how variations in a limited set of tissue-scale parameters generate morphological diversity.

Influenza virus is transmitted via respiratory expulsions, but detecting infectious virus in expulsions is challenging. Here, we describe quantification and genotyping of infectious virus in respiratory particles using a modular influenza sampling tunnel (MIST). The particles deposit on cell monolayers, enabling culture, quantification, and sequencing of viruses. Concomitantly, water-sensitive paper and fine particle samplers yield respiratory particle counts over a broad size range. Using the MIST, we captured infectious virus from humans experimentally infected with the influenza virus on multiple days post-inoculation. The recovered respiratory particles varied in quantity over three orders of magnitude and contained viral variants also detected in samples from infected individuals. Expulsion of infectious virus was associated with infectious viral load in saliva and nasopharyngeal swabs and with clinical symptoms. These data reveal maintenance of viral diversity in expelled aerosols and suggest heterogeneity among individuals in the magnitude of infectious expulsions, impacting forward transmission potential.