Neuronal connectivity in the circadian clock network is essential for robust endogenous timekeeping. In the Drosophila circadian clock network, the four pairs of small ventral lateral neurons (sLNvs) serve as main pacemakers. Peptidergic communication via sLNv, which release the key output neuropeptide, Pigment Dispersing Factor (PDF), has been well characterized. In the absence of PDF, flies become largely arrhythmic, similar to the phenotype associated with the loss of the mammalian circadian peptide, VIP. In contrast, little is known about the role of the synaptic connections that sLNvs form with downstream neurons. Connectomic analyses revealed that despite their role as key pacemaker neurons within the clock network, the sLNvs form few connections with other clock neurons. However, they form strong synaptic connections with a small group of previously uncharacterized neurons, SLP316, which in turn synapse onto dorsal clock neurons. Here, we show that silencing SLP316 neurons via tetanus toxin (TNT) expression shortens the free-running period, whereas hyper-exciting them by expressing the constitutively open Na[+] channel, NaChBac, results in period lengthening. Under light-dark cycles, silencing SLP316 neurons also causes lower daytime activity and higher daytime sleep. Our results revealed that the main postsynaptic partners of the Drosophila pacemaker neurons are a non-clock neuronal cell type that regulates the timing of sleep and activity.

Taste detection and hunger state dynamically regulate the decision to initiate feeding. To study how context-appropriate feeding decisions are generated, we combined synaptic resolution circuit reconstruction with targeted genetic access to specific neurons to elucidate a gustatory sensorimotor circuit for feeding initiation in Drosophila melanogaster. This circuit connects gustatory sensory neurons to proboscis motor neurons through three intermediate layers. Most of the neurons in this pathway are necessary and sufficient for proboscis extension, a feeding initiation behavior, and respond selectively to sugar taste detection. Hunger signals act at select second-order neurons to increase feeding initiation in food-deprived animals. In contrast, a bitter taste pathway inhibits premotor neurons, illuminating a central mechanism that weighs sugar and bitter tastes to promote or inhibit feeding. Together, these studies reveal the neural circuit basis for the integration of external taste detection and internal nutritive state to flexibly execute a critical feeding decision.

In cap-dependent translation initiation, the open reading frame (ORF) of mRNA is established by the placement of the AUG start codon and initiator tRNA in the ribosomal peptidyl (P) site. Internal ribosome entry sites (IRESs) promote translation of mRNAs in a cap-independent manner. We report two structures of the ribosome-bound Taura syndrome virus (TSV) IRES belonging to the family of Dicistroviridae intergenic IRESs. Intersubunit rotational states differ in these structures, suggesting that ribosome dynamics play a role in IRES translocation. Pseudoknot I of the IRES occupies the ribosomal decoding center at the aminoacyl (A) site in a manner resembling that of the tRNA anticodon-mRNA codon. The structures reveal that the TSV IRES initiates translation by a previously unseen mechanism, which is conceptually distinct from initiator tRNA-dependent mechanisms. Specifically, the ORF of the IRES-driven mRNA is established by the placement of the preceding tRNA-mRNA-like structure in the A site, whereas the 40S P site remains unoccupied during this initial step.

The RNA polymerase II core promoter is a structurally and functionally diverse transcriptional module. RNAi depletion and overexpression experiments revealed a genetic circuit that controls the balance of transcription from two core promoter motifs, the TATA box and the downstream core promoter element (DPE). In this circuit, TBP activates TATA-dependent transcription and represses DPE-dependent transcription, whereas Mot1 and NC2 block TBP function and thus repress TATA-dependent transcription and activate DPE-dependent transcription. This regulatory circuit is likely to be one means by which biological networks can transmit transcriptional signals, such as those from DPE-specific and TATA-specific enhancers, via distinct pathways.

Single-molecule localization fluorescence microscopy constructs super-resolution images by sequential imaging and computational localization of sparsely activated fluorophores. Accurate and efficient fluorophore localization algorithms are key to the success of this computational microscopy method. We present a novel localization algorithm based on deep learning which significantly improves upon the state of the art. Our contributions are a novel network architecture for simultaneous detection and localization, and new loss function which phrases detection and localization as a Bayesian inference problem, and thus allows the network to provide uncertainty-estimates. In contrast to standard methods which independently process imaging frames, our network architecture uses temporal context from multiple sequentially imaged frames to detect and localize molecules. We demonstrate the power of our method across a variety of datasets, imaging modalities, signal to noise ratios, and fluorophore densities. While existing localization algorithms can achieve optimal localization accuracy at low fluorophore densities, they are confounded by high densities. Our method is the first deep-learning based approach which achieves state-of-the-art on the SMLM2016 challenge. It achieves the best scores on 12 out of 12 data-sets when comparing both detection accuracy and precision, and excels at high densities. Finally, we investigate how unsupervised learning can be used to make the network robust against mismatch between simulated and real data. The lessons learned here are more generally relevant for the training of deep networks to solve challenging Bayesian inverse problems on spatially extended domains in biology and physics.

Single-molecule localization fluorescence microscopy constructs super-resolution images by sequential imaging and computational localization of sparsely activated fluorophores. Accurate and efficient fluorophore localization algorithms are key to the success of this computational microscopy method. We present a novel localization algorithm based on deep learning which significantly improves upon the state of the art. Our contributions are a novel network architecture for simultaneous detection and localization, and new loss function which phrases detection and localization as a Bayesian inference problem, and thus allows the network to provide uncertainty-estimates. In contrast to standard methods which independently process imaging frames, our network architecture uses temporal context from multiple sequentially imaged frames to detect and localize molecules. We demonstrate the power of our method across a variety of datasets, imaging modalities, signal to noise ratios, and fluorophore densities. While existing localization algorithms can achieve optimal localization accuracy at low fluorophore densities, they are confounded by high densities. Our method is the first deep-learning based approach which achieves state-of-the-art on the SMLM2016 challenge. It achieves the best scores on 12 out of 12 data-sets when comparing both detection accuracy and precision, and excels at high densities. Finally, we investigate how unsupervised learning can be used to make the network robust against mismatch between simulated and real data. The lessons learned here are more generally relevant for the training of deep networks to solve challenging Bayesian inverse problems on spatially extended domains in biology and physics.

Small-molecule fluorophores, such as fluorescein and rhodamine derivatives, are critical tools in modern biochemical and biological research. The field of chemical dyes is old; colored molecules were first discovered in the 1800s, and the fluorescein and rhodamine scaffolds have been known for over a century. Nevertheless, there has been a renaissance in using these dyes to create tools for biochemistry and biology. The application of modern chemistry, biochemistry, molecular genetics, and optical physics to these old structures enables and drives the development of novel, sophisticated fluorescent dyes. This critical review focuses on an important example of chemical biology-the melding of old and new chemical knowledge-leading to useful molecules for advanced biochemical and biological experiments. Expected final online publication date for the Annual Review of Biochemistry Volume 86 is June 20, 2017. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.

Neural circuitry has evolved to form distributed networks that act dynamically across large volumes. Conventional microscopy collects data from individual planes and cannot sample circuitry across large volumes at the temporal resolution relevant to neural circuit function and behaviors. Here we review emerging technologies for rapid volume imaging of neural circuitry. We focus on two critical challenges: the inertia of optical systems, which limits image speed, and aberrations, which restrict the image volume. Optical sampling time must be long enough to ensure high-fidelity measurements, but optimized sampling strategies and point-spread function engineering can facilitate rapid volume imaging of neural activity within this constraint. We also discuss new computational strategies for processing and analyzing volume imaging data of increasing size and complexity. Together, optical and computational advances are providing a broader view of neural circuit dynamics and helping elucidate how brain regions work in concert to support behavior.

In this paper, we present a novel error measure to compare a computer-generated segmentation of images or volumes against ground truth. This measure, which we call Tolerant Edit Distance (TED), is motivated by two observations that we usually encounter in biomedical image processing: (1) Some errors, like small boundary shifts, are tolerable in practice. Which errors are tolerable is application dependent and should be explicitly expressible in the measure. (2) Non-tolerable errors have to be corrected manually. The effort needed to do so should be reflected by the error measure. Our measure is the minimal weighted sum of split and merge operations to apply to one segmentation such that it resembles another segmentation within specified tolerance bounds. This is in contrast to other commonly used measures like Rand index or variation of information, which integrate small, but tolerable, differences. Additionally, the TED provides intuitive numbers and allows the localization and classification of errors in images or volumes. We demonstrate the applicability of the TED on 3D segmentations of neurons in electron microscopy images where topological correctness is arguable more important than exact boundary locations. Furthermore, we show that the TED is not just limited to evaluation tasks. We use it as the loss function in a max-margin learning framework to find parameters of an automatic neuron segmentation algorithm. We show that training to minimize the TED, i.e., to minimize crucial errors, leads to higher segmentation accuracy compared to other learning methods.

The spatial distribution of diverse biomolecules in multicellular organisms is essential for their physiological functions. High-throughput in situ mapping of biomolecules is crucial for both basic and medical research, and requires high scanning speed, spatial resolution, and chemical sensitivity. Here, we developed a Tissue Expansion method compatible with matrix-assisted laser desorption/ionization Mass spectrometry Imaging (TEMI). TEMI reaches single-cell spatial resolution without sacrificing voxel throughput and enables the profiling of hundreds of biomolecules, including lipids, metabolites, peptides (proteins), and N-glycans. Using TEMI, we mapped the spatial distribution of biomolecules across various mammalian tissues and uncovered metabolic heterogeneity in tumors. TEMI can be easily adapted and broadly applied in biological and medical research, to advance spatial multi-omics profiling.

Preprint: 10.1101/2025.02.22.639343