Poster Session B (+LUNCH AND OFFICE HOURS)

Single-cell RNA sequencing offers snapshots of whole transcriptomes but obscures the temporal RNA dynamics. Here we present single-cell metabolically labeled new RNA tagging sequencing (scNT-Seq), a method for massively parallel analysis of newly-transcribed ("new") and pre-existing ("old") mRNAs from the same cell. This droplet microfluidics-based method enables high-throughput chemical conversion on barcoded beads, efficiently marking newly-transcribed mRNAs with T-to-C substitutions. Using scNT-Seq, we jointly profiled new and old transcriptomes in tens of thousands of mouse primary cortical cells. These data revealed distinct patterns of newly synthesized mRNAs at single-cell level in response to brief or sustained neuronal activation. We further showed that measuring new RNA levels of target genes linked to a neuronal activity regulated transcription factor (TF) can temporally resolve TF regulatory network activity in single neurons. Using a novel computational model that explicitly incorporates metabolic labeling-based single-cell measurements, we computed time-resolved RNA velocity to infer cell state trajectories during the highly dynamic neuronal activation process (minutes to hours). Finally, with pulse-chase experiments, scNT-Seq can more accurately estimate RNA synthesis and degradation rates, revealing RNA regulatory strategies in rare cell populations. High-throughput time-resolved single-cell transcriptomics thus provides a broadly applicable strategy to investigate cell-type-specific RNA regulatory mechanisms in dynamic biological processes

Click Here for Poster

Brain computer interface (BCI) technology can be used to bypass damaged motor pathways, allowing people with movement disorders to directly control assistive devices with their thoughts. Current BCI systems are designed to detect information related to intended movements, while excluding information reflecting sensory inputs. However, sensory and motor information are tightly intertwined in cortical circuits, making it difficult to isolate signals encoding motor intention. The goal of this project is to generate new decoding models that simultaneously capture the motor and sensory components of neural activity.
Our approach is based on the analysis of three different types of data: single unit neural ensemble activity, 3D tracking of upper limb motion (from shoulder to fingertips), and estimates of the position and shape of objects in the environment. Neural data is collected using microelectrode arrays chronically implanted into the motor cortex of rhesus macaques. Limbs and objects are tracked using a state of the art markerless motion capture system which relies on transfer learning using deep neural networks (DeepLabCut, Mathis lab).These data streams will allow us to generate models that (1) predict the activity of individual neurons based on upper limb kinematics and object information derived from the visual scene, and (2) predict movement kinematics based on neural activity combined with object information, using camera inputs as surrogate sensory data. 
Our preliminary results validate our approach by demonstrating that models based solely on movement kinematics fail to capture the full variance in firing rates displayed by cortical neurons, and movement prediction significantly improves by taking into account object information derived from the visual scene. Our results will expand our understanding of visuomotor cortical networks, and will also contribute to the development of more effective BCI devices aiming to restore dexterous movement control to people with motor disorders.

Click Here for Poster

Peptidic natural products (PNP) are a major source of signal molecules and drug leads. The existing techniques for PNP discovery require isolation of bioactive molecules and structure elucidation, which are time consuming and expensive. Recent advances in high-throughput mass spectrometry (MS) and next generation sequencing has resulted in large MS/genomic datasets, which are gold mines for PNP discovery. However, currently there is no efficient algorithm to mine these datasets. We have developed computational tools to integrate MS/genomic data for automated discovery of PNPs from environmental isolates/communities.


HypoNPAtlas is a database of hypothetical natural products that is readily searchable against MS. Seq2ripp predicts the structure of ribosomally synthesized and post-translationally modified peptides (RiPPs) from microbial genome. MetaMiner (Cao, et al., 2019)  and NRPminer integrates MS/genomic data to discover RiPPs and non-ribosomal peptides (NRPs) respectively. MolDiscovery is a probabilistic model that efficiently searches small molecules mass spectra. Association networks (Cao, Shcherbin, & Mohimani, 2019) correlates metagenomic and metabolomic features to discover natural products and biotransformations. These tools have enabled discovery of various novel PNPs from public datasets. One of the NRPs discovered have shown anti-parasite activity.


Cao, et al., L. (2019). MetaMiner: A Scalable Peptidogenomics Approach for Discovery of Ribosomal Peptide Natural Products with Blind Modifications from Microbial Communities. Cell Systems, 9, 600-608.
Cao, L., Shcherbin, E., & Mohimani, H. (2019). A Metabolome- and Metagenome-Wide Association Network Reveals Microbial Natural Products and Microbial Biotransformation Products from the Human Microbiota. mSystems, 4, e00387-19.

Click Here for Poster

All proteins dynamically sample diverse folded, unfolded, and excited states with differing free energies. Although energy landscapes have been studied for decades, existing methods have been restricted to assaying one protein per sample, limiting the development of quantitative, global models of protein energy landscapes. We developed a new experimental approach to rapidly characterize hundreds to thousands of protein energy landscapes simultaneously by hydrogen exchange mass spectrometry. In this approach, the target protein library is expressed as a mixture from custom-synthesized DNA oligos, and individual intact proteins are resolved by LC-IMS-MS to measure the overall exchange at each timepoint. We applied this approach to examine the energy landscapes of over 1,000 de novo designed miniproteins (43 residues in length) and found wide variation in landscapes, even among designs with similar topologies. The size of our dataset enabled us to statistically analyze the structural origins of the varied landscapes, revealing how different interaction types modulate both stability and conformational fluctuations. Combining these new large-scale experiments with computational modeling should ultimately lead to a quantitative understanding of the structural determinants of protein energy landscapes.

Click Here for Poster

Human speech can be divided into short, rhythmically-timed elements, similar to syllables within words. Even our cries and laughs, as well as the vocalizations of other species, are periodic. However, the cellular and molecular mechanisms underlying the tempo of mammalian vocalizations remain unknown. Here we describe rhythmically-timed neonatal mouse vocalizations that occur within single breaths, and identify a brainstem node that structures these cries, which we name the intermediate reticular oscillator (iRO). We show that the iRO acts autonomously and sends direct inputs to key muscles in order to coordinate neonatal vocalizations with breathing, as well as paces and patterns these cries. These results reveal that a novel mammalian brainstem oscillator embedded within the conserved breathing circuitry plays a central role in the production of neonatal vocalizations.

Click Here for Poster

The brain is often bombarded by streams of information from multiple sources simultaneously. The goal of our lab is to understand the mechanisms that underlie the flexibility of information processing in cortical circuits, focusing on how inhibitory neurons gate the flow of information between sensory and association regions in a context-dependent manner. Head-fixed mice voluntarily running on a spherical treadmill are rapidly shifted between behavioral contexts within single sessions. In each context, two-photon imaging of calcium activity is used to monitor the responses of hundreds of genetically labeled inhibitory and excitatory neurons simultaneously. In some experiments, optogenetic stimulation is used to activate specific incoming projections to the imaged region, or to inactivate specific cell types. These tools are combined in three main projects. In the first project, the inhibitory mechanisms gating the flow of information between cortical regions will be dissected, to determine whether canonical rules define inhibitory operations across cortex, or if local specializations allow greater flexibility at different hierarchical levels of the cerebral cortex. In the second project, the roles of inhibitory neurons in setting network dynamics will be determined, and the consequences of shifting network dynamics on signal processing will be defined. In the third project, neuromodulatory recruitment of inhibitory circuits across the cortical hierarchy will be described, to determine how shifts in brain state affect information processing. To understand the neural underpinnings of perception, attention, and behavioral flexibility, it is critical to study the interaction between brain areas, rather than to focus on single brain regions in isolation. The experiments proposed here will use new tools to answer fundamental questions about how local circuits interact to process information, toward the goal of understanding how the distributed cortical network gives rise to cognitive processes such as attention and perception.

Click Here for Poster

The learning of motor skills relies on plasticity of the primary motor cortex as task acquisition drives the remodeling of cortical motor networks. Large scale cortical remodeling of evoked motor outputs occurs in response to the learning of skilled, corticospinal-dependent behavior, but not simple, unskilled tasks. We determined the response of corticospinal neurons to both skilled and unskilled motor training and assessed the role of corticospinal neuron activity in the execution of the trained behaviors. Using in vivo calcium imaging, we found that refinement of corticospinal activity correlated with the development of skilled, but not unskilled, motor expertise. Animals that failed to learn our skilled task exhibited a limited repertoire of dynamic movements and a corresponding absence of network modulation. Transection of the corticospinal tract and aberrant activation of corticospinal neurons show the necessity for corticospinal network activity patterns in the execution of skilled, but not unskilled, movement. We revealed a critical role for corticospinal network modulation in the learning and execution of skilled motor movements. The integrity of the corticospinal tract is essential to the recovery of voluntary movement after central nervous system injuries and these findings should help to shape translational approaches to motor recovery.

Click Here for Poster

Receptor-ligand interactions on cells mediate cell-cell and cell-environment communications in many biological processes. Adhesion frequency assay has unique capability of measuring the receptor-ligand interactions at the single-cell level. The measurement can provide important information on the quality of biological processes and for the selection of potent therapeutic cells. However, current adhesion frequency assay, which uses micropipettes to aspirate cells for both interaction measurement and cell transfer afterwards, is bulky, labor-intensive, manual-operative, and low-throughput. In order to maximize the potential of adhesion frequency assay in biomedical research and clinical applications, it is necessary to develop high-throughput miniature device. Along this line, we aim to develop on-chip adhesion frequency assay by merging optical manipulation and microfluidic technologies. Herein, I will share our progress on the assay.

Click Here for Poster