A consortium of inhibitory neurons control the firing patterns of pyramidal cells, but their specific roles in the behaving animal are largely unknown. We performed simultaneous physiological recordings and optogenetic silencing of either perisomatic (parvalbumin (PV) expressing) or dendrite-targeting (somatostatin (SOM) expressing) interneurons in hippocampal area CA1 of head-fixed mice actively moving a treadmill belt rich with visual-tactile stimuli. Silencing of either PV or SOM interneurons increased the firing rates of pyramidal cells selectively in their place fields, with PV and SOM interneurons having their largest effect during the rising and decaying parts of the place field, respectively. SOM interneuron silencing powerfully increased burst firing without altering the theta phase of spikes. In contrast, PV interneuron silencing had no effect on burst firing, but instead shifted the spikes’ theta phase toward the trough of theta. These findings indicate that perisomatic and dendritic inhibition have distinct roles in controlling the rate, burst and timing of hippocampal pyramidal cells.

We describe the anatomy of all the primary motor neurons in the fly proboscis and characterize their contributions to its diverse reaching movements. Pairing this behavior with the wealth of genetic tools offers the possibility to study motor control at single-neuron resolution, and soon throughout entire circuits. As an entry to these circuits, we provide detailed anatomy of proboscis motor neurons, muscles, and joints. We create a collection of fly strains to individually manipulate every proboscis muscle through control of its motor neurons, the first such collection for an appendage. We generate a model of the action of each proboscis joint, and find that only a small number of motor neurons are needed to produce proboscis reaching. Comprehensive control of each motor element in this numerically simple system paves the way for future study of both reflexive and flexible movements of this appendage.

We reconstructed, from a whole CNS EM volume, the synaptic map of input and output neurons that underlie food intake behavior of larvae. Input neurons originate from enteric, pharyngeal and external sensory organs and converge onto seven distinct sensory synaptic compartments within the CNS. Output neurons consist of feeding motor, serotonergic modulatory and neuroendocrine neurons. Monosynaptic connections from a set of sensory synaptic compartments cover the motor, modulatory and neuroendocrine targets in overlapping domains. Polysynaptic routes are superimposed on top of monosynaptic connections, resulting in divergent sensory paths that converge on common outputs. A completely different set of sensory compartments is connected to the mushroom body calyx. The mushroom body output neurons are connected to interneurons that directly target the feeding output neurons. Our results illustrate a circuit architecture in which monosynaptic and multisynaptic connections from sensory inputs traverse onto output neurons via a series of converging paths.

Cerebellar granule cells constitute the majority of neurons in the brain and are the primary conveyors of sensory and motor-related mossy fiber information to Purkinje cells. The functional capability of the cerebellum hinges on whether individual granule cells receive mossy fiber inputs from multiple precerebellar nuclei or are instead unimodal; this distinction is unresolved. Using cell-type-specific projection mapping with synaptic resolution, we observed the convergence of separate sensory (upper body proprioceptive) and basilar pontine pathways onto individual granule cells and mapped this convergence across cerebellar cortex. These findings inform the long-standing debate about the multimodality of mammalian granule cells and substantiate their associative capacity predicted in the Marr-Albus theory of cerebellar function. We also provide evidence that the convergent basilar pontine pathways carry corollary discharges from upper body motor cortical areas. Such merging of related corollary and sensory streams is a critical component of circuit models of predictive motor control. DOI:http://dx.doi.org/10.7554/eLife.00400.001.

We have demonstrated that it is possible to radically change the specificity of maltose binding protein by converting it into a zinc sensor using a rational design approach. In this new molecular sensor, zinc binding is transduced into a readily detected fluorescence signal by use of an engineered conformational coupling mechanism linking ligand binding to reporter group response. An iterative progressive design strategy led to the construction of variants with increased zinc affinity by combining binding sites, optimizing the primary coordination sphere, and exploiting conformational equilibria. Intermediates in the design series show that the adaptive process involves both introduction and optimization of new functions and removal of adverse vestigial interactions. The latter demonstrates the importance of the rational design approach in uncovering cryptic phenomena in protein function, which cannot be revealed by the study of naturally evolved systems.

Segmenting an image into multiple components is a central task in computer vision. In many practical scenarios, prior knowledge about plausible components is available. Incorporating such prior knowledge into models and algorithms for image segmentation is highly desirable, yet can be non-trivial. In this work, we introduce a new approach that allows, for the first time, to constrain some or all components of a segmentation to have convex shapes. Specifically, we extend the Minimum Cost Multicut Problem by a class of constraints that enforce convexity. To solve instances of this NP-hard integer linear program to optimality, we separate the proposed constraints in the branch-and-cut loop of a state-of-the-art ILP solver. Results on photographs and micrographs demonstrate the effectiveness of the approach as well as its advantages over the state-of-the-art heuristic.

Many image segmentation algorithms first generate an affinity graph and then partition it. We present a machine learning approach to computing an affinity graph using a convolutional network (CN) trained using ground truth provided by human experts. The CN affinity graph can be paired with any standard partitioning algorithm and improves segmentation accuracy significantly compared to standard hand-designed affinity functions. We apply our algorithm to the challenging 3D segmentation problem of reconstructing neuronal processes from volumetric electron microscopy (EM) and show that we are able to learn a good affinity graph directly from the raw EM images. Further, we show that our affinity graph improves the segmentation accuracy of both simple and sophisticated graph partitioning algorithms. In contrast to previous work, we do not rely on prior knowledge in the form of hand-designed image features or image preprocessing. Thus, we expect our algorithm to generalize effectively to arbitrary image types.

Deep neural networks have been applied to improve the image quality of fluorescence microscopy imaging. Previous methods are based on convolutional neural networks (CNNs) which generally require more time-consuming training of separate models for each new imaging experiment, impairing the applicability and generalization. Once the model is trained (typically with tens to hundreds of image pairs) it can then be used to enhance new images that are like the training data. In this study, we proposed a novel imaging-transformer based model, Convolutional Neural Network Transformer (CNNT), to outperform the CNN networks for image denoising. In our scheme we have trained a single CNNT based backbone model from pairwise high-low SNR images for one type of fluorescence microscope (instance structured illumination, iSim). Fast adaption to new applications was achieved by fine-tuning the backbone on only 5-10 sample pairs per new experiment. Results show the CNNT backbone and fine-tuning scheme significantly reduces the training time and improves the image quality, outperformed training separate models using CNN approaches such as - RCAN and Noise2Fast. Here we show three examples of the efficacy of this approach on denoising wide-field, two-photon and confocal fluorescence data. In the confocal experiment, which is a 5 by 5 tiled acquisition, the fine-tuned CNNT model reduces the scan time form one hour to eight minutes, with improved quality.

Deep neural networks can improve the quality of fluorescence microscopy images. Previous methods, based on Convolutional Neural Networks (CNNs), require time-consuming training of individual models for each experiment, impairing their applicability and generalization. In this study, we propose a novel imaging-transformer based model, Convolutional Neural Network Transformer (CNNT), that outperforms CNN based networks for image denoising. We train a general CNNT based backbone model from pairwise high-low Signal-to-Noise Ratio (SNR) image volumes, gathered from a single type of fluorescence microscope, an instant Structured Illumination Microscope. Fast adaptation to new microscopes is achieved by fine-tuning the backbone on only 5-10 image volume pairs per new experiment. Results show that the CNNT backbone and fine-tuning scheme significantly reduces training time and improves image quality, outperforming models trained using only CNNs such as 3D-RCAN and Noise2Fast. We show three examples of efficacy of this approach in wide-field, two-photon, and confocal fluorescence microscopy.

Covalent inhibitors are an emerging class of therapeutics, but methods to comprehensively profile their binding kinetics and selectivity across the proteome have been limited. Here we introduce COOKIE-Pro (COvalent Occupancy KInetic Enrichment via Proteomics), an unbiased method for quantifying irreversible covalent inhibitor binding kinetics on a proteome-wide scale. COOKIE-Pro uses a two-step incubation process with mass spectrometry-based proteomics to determine k and K values for covalent inhibitors against both on-target and off-target proteins. We validated COOKIE-Pro using BTK inhibitors spebrutinib and ibrutinib, accurately reproducing known kinetic parameters and identifying both expected and unreported off-targets. The method revealed that spebrutinib has over 10-fold higher potency for TEC kinase compared to its intended target BTK. To demonstrate the method's utility for high-throughput screening, we applied a streamlined two-point strategy to a library of 16 covalent fragments. This approach successfully generated thousands of kinetic profiles, enabling the quantitative decoupling of intrinsic chemical reactivity from binding affinity at scale and validating the method's broad applicability. By providing a comprehensive view of covalent inhibitor binding across the proteome, COOKIE-Pro represents a powerful tool for optimizing the potency and selectivity of covalent drugs during preclinical development.

bioRxiv preprint: https://doi.org/10.1101/2025.06.19.660637