BigNeuron: a resource to benchmark and predict performance of algorithms for automated tracing of neurons in light microscopy datasets.

BigNeuron is an open community bench-testing platform with the goal of setting open standards for accurate and fast automatic neuron tracing. We gathered a diverse set of image volumes across several species that is representative of the data obtained in many neuroscience laboratories interested in neuron tracing. Here, we report generated gold standard manual annotations for a subset of the available imaging datasets and quantified tracing quality for 35 automatic tracing algorithms. The goal of generating such a hand-curated diverse dataset is to advance the development of tracing algorithms and enable generalizable benchmarking. Together with image quality features, we pooled the data in an interactive web application that enables users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and tracing data, and benchmarking of automatic tracing algorithms in user-defined data subsets. The image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. We observed that diverse algorithms can provide complementary information to obtain accurate results and developed a method to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms in noisy datasets. However, specific algorithms may outperform the consensus tree strategy in specific imaging conditions. Finally, to aid users in predicting the most accurate automatic tracing results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic tracings.

CREBB repression of protein synthesis in mushroom body gates long-term memory formation in Drosophila.

Learned experiences are not necessarily consolidated into long-term memory (LTM) unless they are periodic and meaningful. LTM depends on de novo protein synthesis mediated by cyclic AMP response element-binding protein (CREB) activity. In Drosophila, two creb genes (crebA, crebB) and multiple CREB isoforms have reported influences on aversive olfactory LTM in response to multiple cycles of spaced conditioning. How CREB isoforms regulate LTM effector genes in various neural elements of the memory circuit is unclear, especially in the mushroom body (MB), a prominent associative center in the fly brain that has been shown to participate in LTM formation. Here, we report that i) spaced training induces crebB expression in MB α-lobe neurons and ii) elevating specific CREBB isoform levels in the early α/ß subpopulation of MB neurons enhances LTM formation. By contrast, learning from weak training iii) induces 5-HT1A serotonin receptor synthesis, iv) activates 5-HT1A in early α/ß neurons, and v) inhibits LTM formation. vi) LTM is enhanced when this inhibitory effect is relieved by down-regulating 5-HT1A or overexpressing CREBB. Our findings show that spaced training-induced CREBB antagonizes learning-induced 5-HT1A in early α/ß MB neurons to modulate LTM consolidation.

Deep-computer-generated holography with temporal-focusing and a digital propagation matrix for rapid 3D multiphoton stimulation.

Deep learning-based computer-generated holography (DeepCGH) has the ability to generate three-dimensional multiphoton stimulation nearly 1,000 times faster than conventional CGH approaches such as the Gerchberg-Saxton (GS) iterative algorithm. However, existing DeepCGH methods cannot achieve axial confinement at the several-micron scale. Moreover, they suffer from an extended inference time as the number of stimulation locations at different depths (i.e., the number of input layers in the neural network) increases. Accordingly, this study proposes an unsupervised U-Net DeepCGH model enhanced with temporal focusing (TF), which currently achieves an axial resolution of around 5 µm. The proposed model employs a digital propagation matrix (DPM) in the data preprocessing stage, which enables stimulation at arbitrary depth locations and reduces the computation time by more than 35%. Through physical constraint learning using an improved loss function related to the TF excitation efficiency, the axial resolution and excitation intensity of the proposed TF-DeepCGH with DPM rival that of the optimal GS with TF method but with a greatly increased computational efficiency.

CREBA and CREBB in two identified neurons gate long-term memory formation in Drosophila.

Episodic events are frequently consolidated into labile memory but are not necessarily transferred to persistent long-term memory (LTM). Regulatory mechanisms leading to LTM formation are poorly understood, however, especially at the resolution of identified neurons. Here, we demonstrate enhanced LTM following aversive olfactory conditioning in Drosophila when the transcription factor cyclic AMP response element binding protein A (CREBA) is induced in just two dorsal-anterior-lateral (DAL) neurons. Our experiments show that this process is regulated by protein-gene interactions in DAL neurons: (1) crebA transcription is induced by training and repressed by crebB overexpression, (2) CREBA bidirectionally modulates LTM formation, (3) crebA overexpression enhances training-induced gene transcription, and (4) increasing membrane excitability enhances LTM formation and gene expression. These findings suggest that activity-dependent gene expression in DAL neurons during LTM formation is regulated by CREB proteins.

Sparse-view synchrotron X-ray tomographic reconstruction with learning-based sinogram synthesis.

Synchrotron radiation can be used as a light source in X-ray microscopy to acquire a high-resolution image of a microscale object for tomography. However, numerous projections must be captured for a high-quality tomographic image to be reconstructed; thus, image acquisition is time consuming. Such dense imaging is not only expensive and time consuming but also results in the target receiving a large dose of radiation. To resolve these problems, sparse acquisition techniques have been proposed; however, the generated images often have many artefacts and are noisy. In this study, a deep-learning-based approach is proposed for the tomographic reconstruction of sparse-view projections that are acquired with a synchrotron light source; this approach proceeds as follows. A convolutional neural network (CNN) is used to first interpolate sparse X-ray projections and then synthesize a sufficiently large set of images to produce a sinogram. After the sinogram is constructed, a second CNN is used for error correction. In experiments, this method successfully produced high-quality tomography images from sparse-view projections for two data sets comprising Drosophila and mouse tomography images. However, the initial results for the smaller mouse data set were poor; therefore, transfer learning was used to apply the Drosophila model to the mouse data set, greatly improving the quality of the reconstructed sinogram. The method could be used to achieve high-quality tomography while reducing the radiation dose to imaging subjects and the imaging time and cost.

Towards stimulated Raman scattering spectro-microscopy across the entire Raman active region using a multiple-plate continuum.

Stimulated Raman scattering (SRS) has attracted increasing attention in bio-imaging because of the ability toward background-free molecular-specific acquisitions without fluorescence labeling. Nevertheless, the corresponding sensitivity and specificity remain far behind those of fluorescence techniques. Here, we demonstrate SRS spectro-microscopy driven by a multiple-plate continuum (MPC), whose octave-spanning bandwidth (600-1300 nm) and high spectral energy density (∼1 nJ/cm-1) enable spectroscopic interrogation across the entire Raman active region (0-4000 cm-1), SRS imaging of a Drosophila brain, and electronic pre-resonance (EPR) detection of a fluorescent dye. We envision that utilizing MPC light source will substantially enhance the sensitivity and specificity of SRS by implementing EPR mode and spectral multiplexing via accessing three or more coherent wavelengths.

Light field microscopy based on structured light illumination.

In this Letter, we present the modeling, design, and characterization of a light sheet-based structured light illumination (SLI) light field microscopy (LFM) system for fast 3D imaging, where a digital micromirror device is employed to rapidly generate designed sinusoidal patterns in the imaging field. Specifically, we sequentially obtain uniformly illuminated and structured light field images, followed by post-processing with a new, to the best of our knowledge, algorithm that combines the deconvolution and HiLo algorithms. This enables fast volumetric imaging with improved optical cross-sectioning capability at a speed of 50 volumes per second over an imaging field of 250×250×80µm3 in the x, y, and z axis, respectively. Mathematical models have been derived to explain the performance enhancement due to suppressed background noises. To verify the results, imaging experiments on fluorescence beads, fern spore, and Drosophila brain samples, have been performed. The results indicate that the light sheet-based SLI-LFM presents a fast 3D imaging solution with substantially improved optical cross-sectioning capability in comparison with a standard light sheet-based LFM. The new light field imaging method may find important applications in the field of biophotonics.

Three-Dimensional Tracking of Multiple Small Insects by a Single Camera.

Many systems to monitor insect behavior have been developed recently. Yet most of these can only detect two-dimensional behavior for convenient analysis and exclude other activities, such as jumping or flying. Therefore, the development of a three-dimensional (3D) monitoring system is necessary to investigate the 3D behavior of insects. In such a system, multiple-camera setups are often used to accomplish this purpose. Here, a system with a single camera for tracking small insects in a 3D space is proposed, eliminating the synchronization problems that typically occur when multiple cameras are instead used. With this setup, two other images are obtained via mirrors fixed at other viewing angles. Using the proposed algorithms, the tracking accuracy of five individual drain flies, Clogmia albipunctata (Williston) (Diptera: Psychodidae), flitting about in a spherical arena (78 mm in diameter) is as high as 98.7%, whereas the accuracy of 10 individuals is 96.3%. With this proposed method, the 3D trajectory monitoring experiments of insects can be performed more efficiently.

Millisecond two-photon optical ribbon imaging for small-animal functional connectome study.

We developed a high-speed two-photon optical ribbon imaging system, which combines galvo-mirrors for an arbitrary curve scan on a lateral plane and a tunable acoustic gradient-index lens for a 100 kHz-1 MHz axial scan. The system provides micrometer/millisecond spatiotemporal resolutions, which enable continuous readout of functional dynamics from small and densely packed neurons in a living adult Drosophila brain. Compared to sparse sampling techniques, the ribbon imaging modality avoids motion artifacts. Combined with a Drosophila anatomical connectome database, which is the most complete among all model animals, this technique paves the way toward establishing whole-brain functional connectome.

Additive Expression of Consolidated Memory through Drosophila Mushroom Body Subsets.

Associative olfactory memory in Drosophila has two components called labile anesthesia-sensitive memory and consolidated anesthesia-resistant memory (ARM). Mushroom body (MB) is a brain region critical for the olfactory memory and comprised of 2000 neurons that can be classified into αß, α'ß', and γ neurons. Previously we demonstrated that two parallel pathways mediated ARM consolidation: the serotonergic dorsal paired medial (DPM)-αß neurons and the octopaminergic anterior paired lateral (APL)-α'ß' neurons. This finding prompted us to ask how this composite ARM is retrieved. Here, we showed that blocking the output of αß neurons and that of α'ß' neurons each impaired ARM retrieval, and blocking both simultaneously had an additive effect. Knockdown of radish and octß2R in αß and α'ß' neurons, respectively, impaired ARM. A combinatorial assay of radish mutant background rsh1 and neurotransmission blockade confirmed that ARM retrieved from α'ß' neuron output is independent of radish. We identified MBON-ß2ß'2a and MBON-ß'2mp as the MB output neurons downstream of αß and α'ß' neurons, respectively, whose glutamatergic transmissions also additively contribute to ARM retrieval. Finally, we showed that α'ß' neurons could be functionally subdivided into α'ß'm neurons required for ARM retrieval, and α'ß'ap neurons required for ARM consolidation. Our work demonstrated that two parallel neural pathways mediating ARM consolidation in Drosophila MB additively contribute to ARM expression during retrieval.

Untangling the Tauopathy for Alzheimer's disease and parkinsonism.

Tau is a microtubule-associated protein that mainly localizes to the axon to stabilize axonal microtubule structure and neuronal connectivity. Tau pathology is one of the most common proteinopathies that associates with age-dependent neurodegenerative diseases including Alzheimer's disease (AD), and various Parkinsonism. Tau protein undergoes a plethora of intra-molecular modifications and some altered forms promote the production of toxic oligomeric tau and paired helical filaments, and through which further assemble into neurofibrillary tangles, also known as tauopathy. In this review, we will discuss the recent advances of the tauopathy research, primarily focusing on its association with the early axonal manifestation of axonal transport defect, axonal mitochondrial stress, autophagic vesicle accumulation and the proceeding of axon destruction, and the pathogenic Tau spreading across the synapse. Two alternative strategies either by targeting tau protein itself or by improving the age-related physiological decline are currently racing to find the hopeful treatment for tauopathy. Undoubtedly, more studies are needed to combat this devastating condition that has already affected millions of people in our aging population.

Q&A: Why use synchrotron x-ray tomography for multi-scale connectome mapping?

To understand how information flows and is used in the human brain, we must map neural structures at all levels, providing visualizations similar to those of Google Earth for continents, countries, cities, and streets. Unfortunately, the imaging and processing techniques currently used in connectomics projects cannot achieve complete mapping for the brains of large animals within the timespan of a typical research career. However, feasible improvements in x-ray imaging would change this situation. This Q&A discusses synchrotron x-ray tomography, an exciting new approach for in situ mapping of whole-brain wiring diagrams at multiple levels of spatial resolution.

Toward Whole-Body Connectomics.

Recent advances in neuro-technologies have revolutionized knowledge of brain structure and functions. Governments and private organizations worldwide have initiated several large-scale brain connectome projects, to further understand how the brain works at the systems levels. Most recent projects focus on only brain neurons, with the exception of an early effort to reconstruct the 302 neurons that comprise the whole body of the small worm, Caenorhabditis elegans However, to fully elucidate the neural circuitry of complex behavior, it is crucial to understand brain interactions with the whole body, which can be achieved only by mapping the whole-body connectome. In this article, we discuss the current state of connectomics study, focusing on novel optical approaches and related imaging technologies. We also discuss the challenges encountered by scientists who endeavor to map these whole-body connectomes in large animals.

Optimizing depth-of-field extension in optical sectioning microscopy techniques using a fast focus-tunable lens.

Volume imaging based on a fast focus-tunable lens (FTL) allows three-dimensional (3D) observation within milliseconds by extending the depth-of-field (DOF) with sub-micrometer transverse resolution on optical sectioning microscopes. However, the previously published DOF extensions were neither axially uniform nor fit with theoretical prediction. In this work, complete theoretical treatments of focus extension with confocal and various multiphoton microscopes are established to correctly explain the previous results. Moreover, by correctly placing the FTL and properly adjusting incident beam diameter, a uniform DOF is achieved in which the actual extension nicely agrees with the theory. Our work not only provides a theoretical platform for volumetric imaging with FTL but also demonstrates the optimized imaging condition.

Selection of motor programs for suppressing food intake and inducing locomotion in the Drosophila brain.

Central mechanisms by which specific motor programs are selected to achieve meaningful behaviors are not well understood. Using electrophysiological recordings from pharyngeal nerves upon central activation of neurotransmitter-expressing cells, we show that distinct neuronal ensembles can regulate different feeding motor programs. In behavioral and electrophysiological experiments, activation of 20 neurons in the brain expressing the neuropeptide hugin, a homolog of mammalian neuromedin U, simultaneously suppressed the motor program for food intake while inducing the motor program for locomotion. Decreasing hugin neuropeptide levels in the neurons by RNAi prevented this action. Reducing the level of hugin neuronal activity alone did not have any effect on feeding or locomotion motor programs. Furthermore, use of promoter-specific constructs that labeled subsets of hugin neurons demonstrated that initiation of locomotion can be separated from modulation of its motor pattern. These results provide insights into a neural mechanism of how opposing motor programs can be selected in order to coordinate feeding and locomotive behaviors.

Optogenetic control of selective neural activity in multiple freely moving Drosophila adults.

We present an automated laser tracking and optogenetic manipulation system (ALTOMS) for studying social memory in fruit flies (Drosophila melanogaster). ALTOMS comprises an intelligent central control module for high-speed fly behavior analysis and feedback laser scanning (∼40 frames per second) for targeting two lasers (a 473-nm blue laser and a 593.5-nm yellow laser) independently on any specified body parts of two freely moving Drosophila adults. By using ALTOMS to monitor and compute the locations, orientations, wing postures, and relative distance between two flies in real time and using high-intensity laser irradiation as an aversive stimulus, this laser tracking system can be used for an operant conditioning assay in which a courting male quickly learns and forms a long-lasting memory to stay away from a freely moving virgin female. With the equipped lasers, channelrhodopsin-2 and/or halorhodopsin expressed in selected neurons can be triggered on the basis of interactive behaviors between two flies. Given its capacity for optogenetic manipulation to transiently and independently activate/inactivate selective neurons, ALTOMS offers opportunities to systematically map brain circuits that orchestrate specific Drosophila behaviors.

Drosophila ORB protein in two mushroom body output neurons is necessary for long-term memory formation.

Memory is initially labile and gradually consolidated over time through new protein synthesis into a long-lasting stable form. Studies of odor-shock associative learning in Drosophila have established the mushroom body (MB) as a key brain structure involved in olfactory long-term memory (LTM) formation. Exactly how early neural activity encoded in thousands of MB neurons is consolidated into protein-synthesis-dependent LTM remains unclear. Here, several independent lines of evidence indicate that changes in two MB vertical lobe V3 (MB-V3) extrinsic neurons are required and contribute to an extended neural network involved in olfactory LTM: (i) inhibiting protein synthesis in MB-V3 neurons impairs LTM; (ii) MB-V3 neurons show enhanced neural activity after spaced but not massed training; (iii) MB-V3 dendrites, synapsing with hundreds of MB α/ß neurons, exhibit dramatic structural plasticity after removal of olfactory inputs; (iv) neurotransmission from MB-V3 neurons is necessary for LTM retrieval; and (v) RNAi-mediated down-regulation of oo18 RNA-binding protein (involved in local regulation of protein translation) in MB-V3 neurons impairs LTM. Our results suggest a model of long-term memory formation that includes a systems-level consolidation process, wherein an early, labile olfactory memory represented by neural activity in a sparse subset of MB neurons is converted into a stable LTM through protein synthesis in dendrites of MB-V3 neurons synapsed onto MB α lobes.

Automated in situ brain imaging for mapping the Drosophila connectome.

Mapping the connectome, a wiring diagram of the entire brain, requires large-scale imaging of numerous single neurons with diverse morphology. It is a formidable challenge to reassemble these neurons into a virtual brain and correlate their structural networks with neuronal activities, which are measured in different experiments to analyze the informational flow in the brain. Here, we report an in situ brain imaging technique called Fly Head Array Slice Tomography (FHAST), which permits the reconstruction of structural and functional data to generate an integrative connectome in Drosophila. Using FHAST, the head capsules of an array of flies can be opened with a single vibratome sectioning to expose the brains, replacing the painstaking and inconsistent brain dissection process. FHAST can reveal in situ brain neuroanatomy with minimal distortion to neuronal morphology and maintain intact neuronal connections to peripheral sensory organs. Most importantly, it enables the automated 3D imaging of 100 intact fly brains in each experiment. The established head model with in situ brain neuroanatomy allows functional data to be accurately registered and associated with 3D images of single neurons. These integrative data can then be shared, searched, visualized, and analyzed for understanding how brain-wide activities in different neurons within the same circuit function together to control complex behaviors.

Auditory circuit in the Drosophila brain.

Most animals exhibit innate auditory behaviors driven by genetically hardwired neural circuits. In Drosophila, acoustic information is relayed by Johnston organ neurons from the antenna to the antennal mechanosensory and motor center (AMMC) in the brain. Here, by using structural connectivity analysis, we identified five distinct types of auditory projection neurons (PNs) interconnecting the AMMC, inferior ventrolateral protocerebrum (IVLP), and ventrolateral protocerebrum (VLP) regions of the central brain. These auditory PNs are also functionally distinct; AMMC-B1a, AMMC-B1b, and AMMC-A2 neurons differ in their responses to sound (i.e., they are narrowly tuned or broadly tuned); one type of audioresponsive IVLP commissural PN connecting the two hemispheres is GABAergic; and one type of IVLP-VLP PN acts as a generalist responding to all tested audio frequencies. Our findings delineate an auditory processing pathway involving AMMC→IVLP→VLP in the Drosophila brain.