Gustatory systems drive critical survival behaviors such as feeding, foraging, and social interactions. However, gustation remains one of the least mapped sensory modalities at the connectome level. Here, we present the first complete wiring diagram of the male Drosophila adult gustatory system, comprehensively reconstructing gustatory receptor neurons (GRNs) from peripheral organs in a contiguous electron microscopy volume spanning brain, cervical connective, and ventral nerve cord. Integrating this with existing datasets, we generated a pan-central nervous system (CNS), cross-sex connectome that reveals GRN diversity through connectivity-based clustering, molecular identity mapping, and sexual dimorphism analysis. We mapped all feeding motor neurons and traced complete sensory-to-motor pathways to feeding, foraging, endocrine, and social behavior circuits. The emerging circuit architectures reveal distinct circuits for nutrient assessment, motor control, neuroendocrine regulation, and courtship. This work defines the gustatory system's organization at synaptic resolution and provides a framework for understanding how internal states modulate sensory-driven decisions across behavioral contexts.

Visual systems transform photoreceptor inputs into rich perceptual representations through hierarchical networks that extract features along parallel pathways. Although this architecture is conserved across species, how visual information is routed throughout an entire brain remains elusive in any animal. Using the male Drosophila connectome, we trace signals from photoreceptors through the optic lobes—layered, retinotopic regions containing two-thirds of the brain’s neurons—and onward into the central brain. Network-based analyses reveal a multilayered architecture of pathway classes with distinct input mixtures. Signals from visual-input channels spread broadly yet converge in focal regions apparently specialized for particular features and fine spatial sampling. Predictions of receptive-field structure and feature-related input biases are consistent with physiological data and extend to thousands of uncharacterized neuron types. These analyses provide a neuron-by-neuron account of how a visual system organizes and integrates information across an entire brain.

DNA double-strand breaks (DSBs) require rapid signalling and physical stabilisation of broken DNA ends to preserve genome integrity. Here, we identify myosin VI (MVI) as an ATM-regulated component of the DSB response. DNA damage induces rapid nuclear accumulation and nanoscale reorganisation of MVI across multiple cell models, in an ATM-dependent manner. Pharmacological or genetic perturbation of MVI attenuates γH2AX signalling and disrupts Ku80 organisation, while DNA damage persists. This leads to increased sensitivity to cisplatin and bleomycin. Super-resolution imaging reveals spatial association of MVI with Ku80-containing repair structures, implicating MVI in non-homologous end joining (NHEJ). In a minimal reconstituted system, MVI and actin enhance the proximity of Ku70/80-bound DNA ends. Together, our findings identify MVI as a regulator of DSB repair that links ATM signalling to Ku-associated DNA-end stabilisation and suggest that targeting MVI may sensitise tumour cells to genotoxic therapy.

Many insects manipulate plants by injecting effector proteins. In one extreme example of this molecular "hijacking," aphids inject bicycle proteins into , contributing to the development of novel organs called galls. Bicycle proteins share no amino acid sequence similarity with proteins of known function. Here, we report the crystal structures of two divergent bicycle proteins. Both proteins contain saposin-like folds: one with multiple disulfide bonds exhibits a swapped domain topology; the other has no disulfide bonds and possesses two distinct, tandem domains. To explore the structural evolution of bicycle proteins, we attempted to predict bicycle protein structures with Alphafold2 (AF2) and other deep learning programs. While AF2 did not recover the two experimental structures using existing databases, it succeeded when provided with multiple sequence alignments (MSAs) of protein sequences from newly sequenced closely related species. Using this approach, we generated 2,400 high-confidence bicycle protein predictions from seven aphid species. While all aphid bicycle proteins contain predicted saposin-like folds, they display a vast diversity of structural and physicochemical properties. While this diversity thwarts prediction of conserved functions encoded in structure, it suggests that bicycle proteins have evolved to target diverse plant processes and/or to evade plant immune surveillance. Our extension of AF2 with custom MSAs of proteins from closely related species provides a generalizable, powerful approach for predicting structures of rapidly evolving protein families.

Summary Fluorescent imaging in live cells is a cornerstone of life sciences. While natural fluorescent proteins have been engineered to enhance individual features, no existing tag combines ideal properties into a single system: high brightness, reversible binding, compact size, and stability across diverse conditions. Here, we achieve this through de novo design of rhodamine binders (Rhobin). To harness the broad repertoire of rhodamine fluorophores, we developed a generalizable design strategy for a pan-rhodamine binder compatible with diverse wavelengths and applications. Rhobin enables live- and fixed-cell imaging of various subcellular targets in mammalian cells, showing brightness surpassing existing tags. Its reversible fluorophore binding supports super-resolution stimulated emission depletion (STED) and live-cell single-molecule imaging for extended durations compared with HaloTag. Beyond conventional systems, Rhobin enables live imaging of the extremophile Sulfolobus acidocaldarius at 75°C, previously inaccessible with current tags. Together, these results establish Rhobin as a versatile platform for next-generation imaging and biosensor design.

A developmental increase in cerebrospinal fluid (CSF) production during development is essential for neuronal growth and ventricular expansion. A key regulator of CSF production is the specialized sensory multicilia of the choroid plexus (ChP), which mediate non-canonical Sonic hedgehog (Shh) signaling to suppress water channel and ion transporter expression, thereby limiting CSF production. ChP multicilia progressively shortens during development, attenuating Shh signaling and promoting CSF production. Here, we identify miR-34/449 miRNAs as essential regulators of ChP multiciliogenesis. Whereas mutations in canonical ciliogenesis genes elevate CSF production and contribute to hydrocephaly, deletion of miR-34/449 reduces CSF volume and causes microcephaly. Loss of miR-34/449 miRNAs causes excessive basal body amplification, defective basal body docking, and failure of developmental multiciliary shortening. Consequently, miR-34/449-deficient ChP cilia remain abnormally long and fail to attenuate Shh signaling, resulting in sustained repression of water channel and ion transporter expression and reduced CSF production. Mechanistically, miR-34/449 miRNAs directly target Gmnc, a master transcriptional regulator of multiciliogenesis, to restrain basal body amplification and promote basal body docking. Together, our findings identify miR-34/449 miRNAs as critical regulators of ChP multiciliogenesis and establish the developmental remodeling of ChP multicilia as a mechanism to couple Shh signaling dynamics to developmental control of CSF production.

T4 and T5 neurons are the first direction-selective neurons in the visual pathway. They are the most numerous cell types in the fly brain (~6000 within each optic lobe) and, as a population, their compact dendritic arbours span the entire visual field. They are classified into four subtypes (a, b, c, and d). Each subtype encodes one of four orthogonal motion directions (up, down, forwards, backwards). Crucially, the dendrites of these neurons are oriented inversely to the functional direction of motion which they encode. This dendritic orientation is what ultimately determines their functional directional encoding. The development of these neurons is well characterised up to the point of neuropil innervation. However, the full population of these neurons innervate their target neuropil prior to the emergence of directionality within their dendrites. As it stands, development prior to the emergence of dendritic orientations, and the adult oriented dendrite are both well understood, but the key components relating to the emergence of orientation itself are missing. Recent whole-brain electron microscopy (EM) connectomes of Drosophila melanogaster provide an unprecedented level of resolution and completeness when considering the morphology of neurons. Utilising this, we isolate the dendritic arbour of every T4 and T5 neuron within a female adult Drosophila brain, made available through FAFB-FlyWire. In doing so we are able to rigorously quantify the morphology of these dendrites in order to understand their similarities and differences. In doing so we aim to shed light on the origins of dendritic directionality. We reason that either this emerges through a tightly controlled, subtype specific mechanism, or is the result of a subtype agnostic mechanism and external factors. In the former case, we would expect evidence of this in differences between the morphological structure of individual dendrites between T4 and T5, and their subtypes. Our analysis however reveals a high degree of structural similarity between T4 and T5, and within their subtypes. Particularly, the geometry of branching, section orientation, and tree-graph structure of these dendrites show only minor variability, with no consistent separation between T4 and T5, or their subtypes. These results indicate that, despite forming in different neuropils, and serving distinct motion directions, T4 and T5 dendrites follow closely aligned morphological patterns. This suggests a shared mechanism of directed outgrowth, as opposed to symmetry breaking emerging through neuron type or subtype specific mechanisms.

Recent advances in field technology have led to a massive influx of in-the-wild video data for ecological science. The primary bottleneck in leveraging this data is the high cost of expert annotation. While computer vision offers a potential solution, current models frequently fail when deployed in complex marine environments. To characterize these failures, we introduce WildFin, a novel benchmark for fish behavior recognition collected and annotated by ecologists. WildFin spans two critical real-world paradigms: stationary cameras monitoring groups of fish and dynamic divers following individual subjects. The dataset represents a massive curation effort, involving 1,350 hours of fieldwork and 600 hours of expert annotation to produce 9 hours of behavioral data with over 2 million frame-by-frame labels. We benchmark modern vision foundation models and quantify tradeoffs between static and spatiotemporal architectures, revealing the substantial gap that remains between current model capabilities and the demands of real-world underwater behavioral analysis. Project website: this https URL.

A central challenge in systems neuroscience research is to elucidate how distinct neuronal cell types coordinate to produce complex behaviors: an endeavor that requires linking their molecular identity, connectivity and activity patterns in behaving animals. Recent advances in single-cell RNA sequencing and spatial transcriptomics have revealed remarkable molecular diversity among neurons. However, most functional recording techniques, including in vivo two-photon calcium imaging, do not reveal the molecular cell identities of recorded cells. Although manual one-to-one matching between functional imaging and post hoc molecular profiling has been achieved for dozens to a few hundred neurons, these approaches are typically labor-intensive, difficult to scale, and capture only a limited fraction of cell types. Integrating these two modalities at large scale remains challenging, limiting our ability to fully understand the general logic of neural computation. Here we present a cross-modal workflow that integrates in vivo two-photon calcium imaging in behaving mice with Expansion-Assisted Iterative Fluorescence In Situ Hybridization (EASI-FISH), a thick-tissue spatial transcriptomic approach. As a proof of concept, we apply this workflow to the mouse dorsal hippocampus, a brain region with high neuronal density and small cell size where manual cell matching is impractical, making it a stringent test of the method’s accuracy and scalability. Using our approach, we achieved highly accurate alignment between in vivo neural activity and ex vivo molecular cell-type identity across hundreds to thousands of neurons, enabling direct mapping of neural dynamics to molecularly defined neuronal populations. We showed that longitudinal neural recordings followed by molecular mapping reveal distinct neural activity patterns that emerge as animals learn a spatial navigation task. By linking activity patterns to molecularly defined cell types, this method provides a framework for testing whether neural representations are organized through cell-type-specific coding or distributed population activity, offering insight into the computational logic by which neural circuits generate complex behaviors.MOTIVATION The mammalian brain contains an immense diversity of neurons exhibiting substantial morphological, functional, and molecular diversity. Understanding how this molecular heterogeneity supports neural computation and shapes behavior remains a central challenge in systems neuroscience. Yet, most approaches for recording neural activity provide limited access to molecular identity, making it difficult to directly relate functional dynamics to cell-type-specific features across many neurons in the same tissue. Here, we present a cross-modal workflow that recovers, aligns, and matches in vivo neural activity with targeted molecular profiling, enabling functional readouts and RNA profiling to be coupled at the individual cell level. This method provides a powerful platform for uncovering how molecularly defined neurons contribute to complex activity patterns and, ultimately, animal behavior.

RNA design has been hindered by the limited accuracy of three-dimensional (3D) structure prediction. In this study, we show that intricate RNA structures can be generated with current deep learning tools through accurate de novo design of pseudoknot secondary structures. In an Eterna competition involving 57 pseudoknots, generative artificial intelligence (AI) methods matched experienced human designers in solving most blind challenges, evaluated by single nucleotide-resolution chemical mapping, compensatory mutagenesis, and cryo-electron microscopy. AI-generated molecules with accurate secondary structures formed well-ordered 3D folds stabilized by noncanonical tertiary interactions not modeled during design. Success was guided by an RNet foundation model trained on prior chemical mapping data, suggesting that some difficult RNA design tasks may be tractable without first solving RNA 3D structure prediction.