Biology subjects

Lee, K.

Publications and source records attributed to Lee, K..

At least 199 records · Page 11Linked to original sources

bioRxiv2021

Flow Cytometric Evaluation of Yeast-Bacterial Cell-Cell Interactions

Synthetic cell-cell interaction systems can be useful for understanding multicellular communities or for screening binding molecules. We adapt a previously characterized set of synthetic cognate nanobody-antigen pairs to a yeast-bacteria coincubation format and use flow cytometry to evaluate cell-cell interactions mediated by binding between surface-displayed molecules. We further use fluorescence-activated cell sorting (FACS) to enrich for a specific yeast-displayed nanobody within a mixed yeast-display population. Finally, we demonstrate that this system supports characterization of a therapeutically relevant nanobody-antigen interaction: a previously discovered nanobody that binds to the intimin protein expressed on the surface of enterohemorrhagic E. coli. Overall, our findings indicate that the yeast-bacteria format supports efficient evaluation of ligand-target interactions. With further development, this format may facilitate systematic characterization and high throughput discovery of bacterial surface-binding molecules.

synthetic biology↗

bioRxiv2021

Role of EXO1 nuclease activity in genome maintenance, the immune response and tumor suppression in Exo1D173A mice

DNA damage response pathways rely extensively on nuclease activity to process DNA intermediates. Exonuclease 1 (EXO1) is a pleiotropic evolutionary conserved DNA exonuclease involved in various DNA repair pathways, replication, antibody diversification, and meiosis. But, whether EXO1 facilitates these DNA metabolic processes through its enzymatic or scaffolding functions remains unclear. Here we dissect the contribution of EXO1 enzymatic versus scaffolding activity by comparing Exo1DA/DA mice expressing a proven nuclease-dead mutant form of EXO1 to entirely EXO1-deficient Exo1-/- and EXO1 wild type Exo1+/+ mice. We show that Exo1DA/DA and Exo1-/- mice are compromised in canonical DNA repair processing, suggesting that the EXO1 enzymatic role is important for error-free DNA mismatch and double-strand break repair pathways. However, in non-canonical repair pathways, EXO1 appears to have a more nuanced function. Next-generation sequencing of heavy chain V region in B cells showed the mutation spectra of Exo1DA/DA mice to be intermediate between Exo1+/+ and Exo1-/- mice, suggesting that both catalytic and scaffolding roles of EXO1 are important for somatic hypermutation. Similarly, while overall class switch recombination in Exo1DA/DA and Exo1-/- mice was comparably defective, switch-switch junction analysis suggests that EXO1 might fulfill an additional scaffolding function downstream of class switching. In contrast to Exo1-/- mice that are infertile, meiosis progressed normally in Exo1DA/DA and Exo1+/+ cohorts, indicating that a structural but not the nuclease function of EXO1 is critical for meiosis. However, both Exo1DA/DA and Exo1-/- mice displayed similar mortality and cancer predisposition profiles. Taken together, these data demonstrate that EXO1 has both scaffolding and enzymatic functions in distinct DNA repair processes and suggest a more composite and intricate role for EXO1 in DNA metabolic processes and disease.

cancer biology↗

bioRxiv2021

Characterizing Genetic Mechanisms Involved in Measuring Day-length in Drosophila melanogaster

Many organisms are known to regulate seasonal behaviors and physiological processes in response to day length changes through photoperiodism. Extreme changes in photoperiods have detrimental effects on human health, which can impair development and serve as the origin of adult diseases. Since the seminal work by Bunning in 1936, there are studies supporting the view that organisms can measure the day length through an endogenous 24-hour cellular circadian clock. However, the mechanisms involved in measuring seasonal or day-length changes are not understood. In the current study, we performed a genome-wide association study (GWAS) on photoperiodism using the Drosophila Genetic Reference Panel. The GWAS identified 4 top candidate genes responsible for photoperiodic regulations. The knockout mutants of the four candidate genes (Protein Kinase C delta (Pkcdelta), Glucuronyltransferase-P (GlcAT-P), Brain-specific homeobox (Bsh), and Diuretic hormone 31 Receptor (Dh31-R1)) were analyzed for their photoperiod and circadian period phenotypes. PKCdelta and GlcAT-P mutants show a significantly different photoperiod response compared to that of the wild type strain, and also had an altered circadian period phenotype. Further molecular characterization revealed two independent mutant alleles of PKCdelta with a defective catalytic domain had distinct photoperiod responses. Taken these data together, we concluded that there is overlap between the circadian clock and photoperiodic regulations in Drosophila, and PKCdelta is a component that is involved in both circadian and photoperiodic regulations. By identifying novel molecular components of photoperiod, the current study provides new insights into the genetic mechanisms of determining the seasonal changes. Author SummaryExtreme changes in photoperiods have detrimental effects on human health, which can impair development and serve as the origin of adult diseases. The molecular and genetic mechanisms of how an organism interprets and adapts to seasonal environmental changes are not well understood. Drosophila Genetic Reference Panel (DGRP) is a community resource of 205 inbred lines created for studying population genomics and quantitative traits. Using DGRP we performed a genome-wide association study (GWAS) to find potential photoperiod candidate genes. PKCdelta, a candidate gene from the GWAS study was identified to have both photoperiod and circadian effects. This data supports the view that circadian clock and photoperiodism have a shared regulatory circuit. Our study sheds light onto potential genes that could be further studied to characterize the mechanisms of photoperiodism, and its relationship to the circadian clock.

genetics↗

bioRxiv2021

Effect of dietary zinc supplementation on the gastrointestinal microbiota and host gene expression in the Shank3B-/- mouse model of autism spectrum disorder

Shank genes are implicated in ~1% of people with autism and mice with Shank3 knock out mutations exhibit autism-like behaviours. Zinc deficiency and gastrointestinal problems can be common among people with autism, and zinc is a key element required for SHANK protein function and gut development. In Shank3B-/- mice, a supplementary zinc diet reverses autism behaviours. We hypothesise that dietary zinc may alter the gut microbiome, potentially affecting the gut-microbiome-brain axis, which may contribute to changes in autism-like behaviours. To test this, four types of gastrointestinal samples (ileum, caecum, colon, faecal) were collected from wild-type and knock-out Shank3B-/- mice on either control or supplemented-zinc diets. Cage, genotype and zinc diet each contributed significantly to bacterial community variation (accounting for 12.8%, 3.9% and 2.3% of the variation, respectively). Fungal diversity differed significantly between wild-type and knock-out Shank3B-/- mice on the control zinc diet, and the fungal biota differed among gut locations. RNA-seq analysis of host (mouse) transcripts revealed differential expression of genes involved in host metabolism that may be regulated by the gut microbiota and genes involved in anti-microbial interactions. By utilising the Shank3B-/- knock-out mouse model we were able to examine the influence of - and interactions between - dietary zinc and ASD-linked host genotype. These data broaden understanding of the gut microbiome in autism and pave the way towards potential microbial therapeutics for gastrointestinal problems in people with autism. ImportancePreviously, supplemental dietary zinc in the Shank3B-/- mouse model of autism spectrum disorder resulted in observations of ASD behaviours reversal; in this study we also used the Shank3B-/- mouse model to examine the influence of - and interaction between - dietary zinc and ASD-linked host genotype. Sample location along the gastrointestinal tract, genotype and zinc diet explained some of the variation in the microbiota data, with notable bacterial differences between treatment groups. Differential expression of host genes between treatment groups, including antimicrobial interaction genes and gut microbiota-regulated host metabolism genes, suggests that the interplay between gut microbes, the gastrointestinal tract and the brain may play a major role towards the observed amelioration of ASD behaviours seen previously with supplemented dietary zinc. These results widen the scope towards manipulating both dietary zinc and the microbiota itself to ameliorate ASD-related behaviours and associated gastrointestinal issues.

microbiology↗

bioRxiv2021

Precision engineering of an anti-HLA-A2 chimeric antigen receptor in regulatory T cells for transplant immune tolerance

Infusion of regulatory T cells (Tregs) engineered with a chimeric antigen receptor (CAR) targeting donor-derived human leukocyte antigen (HLA) is a promising strategy to promote transplant tolerance. Here, we describe an anti-HLA-A2 CAR (A2-CAR) generated by grafting the complementarity-determining regions (CDRs) of a human monoclonal anti-HLA-A2 antibody into the framework regions of the Herceptin 4D5 single-chain variable fragment and fusing it with a CD28-{zeta} signaling domain. The CDR-grafted A2-CAR maintained the specificity of the original antibody. We then generated HLA-A2 mono-specific human CAR Tregs either by deleting the endogenous T-cell receptor (TCR) via CRISPR/Cas9 and introducing the A2-CAR using lentiviral transduction or by directly integrating the CAR construct into the TCR alpha constant locus using homology-directed repair. These A2-CAR+TCRdeficient human Tregs maintained both Treg phenotype and function in vitro. Moreover, they selectively accumulated in HLA-A2-expressing islets transplanted from either HLA-A2 transgenic mice or deceased human donors. A2-CAR+TCRdeficient Tregs did not impair the function of these HLA-A2+ islets, whereas similarly engineered A2-CAR+TCRdeficientCD4+ conventional T cells rejected the islets in less than 2 weeks. A2-CAR+TCRdeficient Tregs delayed graft-versus-host disease only in the presence of HLA-A2, expressed either by co-transferred peripheral blood mononuclear cells or by the recipient mice. Altogether, we demonstrate that genome-engineered mono-antigen-specific A2-CAR Tregs localize to HLA-A2-expressing grafts and exhibit antigen-dependent in vivo suppression, independent of TCR expression. These approaches may be applied towards developing precision Treg cell therapies for transplant tolerance.

immunology↗

bioRxiv2021

Genome sequencing of 196 Treponema pallidum strains from six continents reveals additional variability in vaccine candidate genes and dominance of Nichols clade strains in Madagascar

In spite of its immutable susceptibility to penicillin, Treponema pallidum (T. pallidum) subsp. pallidum continues to cause millions of cases of syphilis each year worldwide, resulting in significant morbidity and mortality and underscoring the urgency of developing an effective vaccine to curtail the spread of the infection. Several technical challenges, including absence of an in vitro culture system until very recently, have hampered efforts to catalog the diversity of strains collected worldwide. Here, we provide near-complete genomes from 196 T. pallidum strains - including 191 T. pallidum subsp. pallidum - sequenced directly from patient samples collected from 8 countries and 6 continents. Maximum likelihood phylogeny revealed that samples from most sites were predominantly SS14 clade. However, 99% (84/85) of the samples from Madagascar formed two of the five distinct Nichols subclades. Although recombination was uncommon in the evolution of modern circulating strains, we found multiple putative recombination events between T. pallidum subsp. pallidum and subsp. endemicum, shaping the genomes of several subclades. Temporal analysis dated the most recent common ancestor of Nichols and SS14 clades to 1717 (95% HPD: 1543-1869), in agreement with other recent studies. Rates of SNP accumulation varied significantly among subclades, particularly among different Nichols subclades, and was associated in the Nichols A subclade with a C394F substitution in TP0380, a ERCC3-like DNA repair helicase. Our data highlight the role played by variation in genes encoding putative surface-exposed outer membrane proteins in defining separate lineages, and provide a critical resource for the design of broadly protective syphilis vaccines targeting surface antigens. Author SummaryEach year, millions of new cases of venereal and congenital syphilis, caused by the bacterium Treponema pallidum (T. pallidum) subsp. pallidum, are diagnosed worldwide, resulting in significant morbidity and mortality. Alongside endemic circulation of syphilis in low-income countries, disease resurgence in high-income nations has underscored the need for a vaccine. Due to prior technological limitations in culturing and sequencing the organism, the extent of the genetic diversity within modern strains of T. pallidum subsp. pallidum remains poorly understood, hampering development of a broadly protective vaccine. In this study, we obtained 196 near-complete T. pallidum genomes directly from clinical swabs from eight countries across six continents. Of these, 191 were identified as T. pallidum subsp. pallidum, including 90 Nichols clade genomes. Bayesian analysis revealed a high degree of variance in mutation rate among subclades. Interestingly, a Nichols subclade with a particularly high mutation rate harbors a non-synonymous mutation in a putative DNA repair helicase. Coupling sequencing data with protein structure prediction, we identified multiple novel amino acid variants in several proteins previously identified as potential vaccine candidates. Our data help inform current efforts to develop a broadly protective syphilis vaccine.

genomics↗

bioRxiv2021

Petascale neural circuit reconstruction: automated methods

3D electron microscopy (EM) has been successful at mapping invertebrate nervous systems, but the approach has been limited to small chunks of mammalian brains. To scale up to larger volumes, we have built a computational pipeline for processing petascale image datasets acquired by serial section EM, a popular form of 3D EM. The pipeline employs convolutional nets to compute the nonsmooth transformations required to align images of serial sections containing numerous cracks and folds, detect neuronal boundaries, label voxels as axon, dendrite, soma, and other semantic categories, and detect synapses and assign them to presynaptic and postsynaptic segments. The output of neuronal boundary detection is segmented by mean affinity agglomeration with semantic and size constraints. Pipeline operations are implemented by leveraging distributed and cloud computing. Intermediate results of the pipeline are held in cloud storage, and can be effortlessly viewed as images, which aids debugging. We applied the pipeline to create an automated reconstruction of an EM image volume spanning four visual cortical areas of a mouse brain. Code for the pipeline is publicly available, as is the reconstructed volume.

neuroscience↗

bioRxiv2021

Functional connectomics spanning multiple areas of mouse visual cortex

To understand the brain we must relate neurons functional responses to the circuit architecture that shapes them. Here, we present a large functional connectomics dataset with dense calcium imaging of a millimeter scale volume. We recorded activity from approximately 75,000 neurons in primary visual cortex (VISp) and three higher visual areas (VISrl, VISal and VISlm) in an awake mouse viewing natural movies and synthetic stimuli. The functional data were co-registered with a volumetric electron microscopy (EM) reconstruction containing more than 200,000 cells and 0.5 billion synapses. Subsequent proofreading of a subset of neurons in this volume yielded reconstructions that include complete dendritic trees as well the local and inter-areal axonal projections that map up to thousands of cell-to-cell connections per neuron. Here, we release this dataset as an open-access resource to the scientific community including a set of tools that facilitate data retrieval and downstream analysis. In accompanying papers we describe our findings using the dataset to provide a comprehensive structural characterization of cortical cell types1-3 and the most detailed synaptic level connectivity diagram of a cortical column to date2, uncovering unique cell-type specific inhibitory motifs that can be linked to gene expression data4. Functionally, we identify new computational principles of how information is integrated across visual space5, characterize novel types of neuronal invariances6 and bring structure and function together to decipher a general principle that wires excitatory neurons within and across areas7, 8.

neuroscience↗

bioRxiv2021

Spontaneous Mutants of Streptococcus sanguinis with Defects in the Glucose-PTS Show Enhanced Post-Exponential Phase Fitness

Genetic truncations in a gene encoding a putative glucose-PTS protein (manL, EIIABMan) were identified in subpopulations of two separate laboratory stocks of Streptococcus sanguinis SK36; the mutants had reduced PTS activities on glucose and other monosaccharides. Using an engineered mutant of manL and its complemented derivative, we showed that the ManL-deficient strain had improved bacterial viability in stationary phase and was better able to inhibit the growth of the dental caries pathogen Streptococcus mutans. Transcriptional analysis and biochemical assays suggested that the manL mutant underwent reprograming of central carbon metabolism that directed pyruvate away from production of lactate, increasing production of hydrogen peroxide (H2O2) and excretion of pyruvate. Addition of pyruvate to the medium enhanced the survival of SK36 in overnight cultures. Meanwhile, elevated pyruvate levels were detected in the cultures of a small, but significant percentage ([~]10%), of clinical isolates of oral commensal bacteria. Furthermore, the manL mutant showed higher expression of the arginine deiminase system than the wild type, which enhanced the ability of the mutant to raise environmental pH when arginine was present. Significant discrepancies in genome sequence were identified between strain SK36 obtained from ATCC and the sequence deposited in GenBank. As the conditions that are likely associated with the emergence of spontaneous manL mutations, i.e. excess carbohydrates and low pH, are those associated with caries development, we propose that the glucose-PTS strongly influences commensal-pathogen interactions by altering the production of ammonia, pyruvate, and H2O2. ImportanceA health-associated dental microbiome provides a potent defense against pathogens and diseases. Streptococcus sanguinis is an abundant member of a health-associated oral flora that antagonizes pathogens by producing hydrogen peroxide. There is a need for a better understanding of the mechanisms that allow bacteria to survive carbohydrate-rich and acidic environments associated with the development of dental caries. We report the isolation and characterization of spontaneous mutants of S. sanguinis with impairment in glucose transport. The resultant reprograming of central metabolism in these mutants reduced the production of lactic acid and increased pyruvate accumulation; the latter enables these bacteria to better cope with hydrogen peroxide and low pH. The implications of these discoveries in the development of dental caries are discussed.

microbiology↗

bioRxiv2021

Arousal impacts distributed hubs modulating the integration of brain functional connectivity

Even when subjects are at rest, it is thought that brain activity is organized into distinct brain states during which reproducible patterns are observable. Yet, it is unclear how to define or distinguish different brain states. A potential source of brain state variation is arousal, which may play a role in modulating functional interactions between brain regions. Here, we use simultaneous resting state functional magnetic resonance imaging and pupillometry to study the impact of arousal levels indexed by pupil area on the integration of large-scale brain networks. We employ a novel sparse dictionary learning-based method to identify hub regions participating in between-network integration stratified by arousal, by measuring k-hubness, the number (k) of functionally overlapping networks in each brain region. We show evidence of a brain-wide decrease in between-network integration and inter-subject variability at low relative to high arousal, with differences emerging across regions of the frontoparietal, default mode, motor, limbic, and cerebellum networks. State-dependent changes in k-hubness relate to the actual patterns of network integration within these hubs, suggesting a brain state transition from high to low arousal characterized by global synchronization and reduced network overlaps. We demonstrate that arousal is not limited to specific brain areas known to be directly associated with arousal regulation, but instead has a brain-wide impact that involves high-level between-network communications.

neuroscience↗

bioRxiv2021

A Comprehensive Overview of the Physical Health of the Adolescent Brain Cognitive Development Study (ABCD) Cohort at Baseline

Physical health in childhood is crucial for neurobiological as well as overall development, and can shape long-term outcomes into adulthood. The landmark, longitudinal Adolescent Brain Cognitive Development StudySM (ABCD study(R)), was designed to investigate brain development and health in almost 12,000 youth who were recruited when they were 9-10 years old and will be followed through adolescence and early adulthood. The overall goal of this paper is to provide descriptive analyses of physical health measures in the ABCD study at baseline, including but not limited to sleep, physical activity and sports involvement, and body mass index, and how these measures vary across demographic groups. This paper outlines how the physical health of the ABCD sample corresponds with that of the US population and highlights important avenues for health disparity research. This manuscript will provide important information for ABCD users and help guide analyses investigating physical health as it pertains to adolescent and young adult development.

developmental biology↗

bioRxiv2021

A Unified Model for Entrainment by Circadian Clocks: Dynamic Circadian Integrated Response Characteristic (dCiRC)

Circadian rhythm is a ubiquitous phenomenon, and it is observed in all biological kingdoms. In nature, their primary characteristic or phenotype is the phase of entrainment. There are two main hypotheses related to how circadian clocks entrain, parametric and non-parametric models. The parametric model focuses on the gradual changes of the clock parameters in response to the changing ambient condition, whereas the non-parametric model focuses on the instantaneous change of the phase of the clock in response to the zeitgeber. There are ample empirical data supporting both models. However, only recently has a unifying model been proposed, the circadian integrated response characteristic (CiRC). In the current study, we developed a system of ordinary differential equations, dynamic CiRC (dCiRC), that describes parameters of circadian rhythms and predicts the phase of entrainment in zeitgeber cycles. dCiRC mathematically extracts the underlying information of velocity changes of the internal clock that reflects the parametric model and the phase shift trajectory that reflects the non-parametric model from phase data under entraining conditions. As a proof of concept, we measured clock parameters of 26 Neurospora crassa ecotypes in both cycling and constant conditions using dCiRC. Our data showed that the morning light shortens the period of the clock while the afternoon light lengthens it. We also found that individual ecotypes have different strategies of integrating light effects to accomplish the optimal phase of entrainment, a model feature that is consistent with our knowledge of how circadian clocks are organized and encoded. The unified model dCiRC will provide new insights into how circadian clocks function under different zeitgeber conditions. We suggest that this type of model may be useful in the advent of chronotherapies.

systems biology↗

bioRxiv2021

Stopping vs Resting state during motor imagery paradigm

Current non-invasive Brain Machine interfaces commonly rely on the decoding of sustained motor imagery activity (MI). This approach enables a user to control brain-actuated devices by triggering predetermined motor actions. One major drawback of such strategy is that users are not trained to stop their actions. Indeed, the termination process involved in BMI is poorly understood with most of the studies assuming that the end of an MI action is similar to the resting state. Here we hypothesize that the process of stopping MI (MI termination) and resting state are two different processes that should be decoded independently due to the exhibition of different neural pattens. We compared the detection of both states transitions of an imagined movement, i.e. rest-to-movement (onset) and movement-to-rest (offset). Our results shows that both decoders show significant differences in term of performances and latency (N=17 Subjects) with the offset decoder able to detect faster and better MI termination. While studying this difference, we found that the offset decoder is primarily based on the use of features in Beta band which appears earlier. Based on this finding, we also proposed a Random Forrest based decoder which enable to distinguish three classes (MI, MI termination and REST).

neuroscience↗

bioRxiv2021

Oligodendrocyte precursor cells prune axons in the mouse neocortex

Neurons in the developing brain undergo extensive structural refinement as nascent circuits adopt their mature form1. This transformation is facilitated by the engulfment and degradation of excess axonal branches and inappropriate synapses by surrounding glial cells, including microglia and astrocytes2,3. However, the small size of phagocytic organelles and the complex, highly ramified morphology of glia has made it difficult to determine the contribution of these and other glial cell types to this process. Here, we used large scale, serial electron microscopy (ssEM) with computational volume segmentation to reconstruct the complete 3D morphologies of distinct glial types in the mouse visual cortex. Unexpectedly, we discovered that the fine processes of oligodendrocyte precursor cells (OPCs), a population of abundant, highly dynamic glial progenitors4, frequently surrounded terminal axon branches and included numerous phagolysosomes (PLs) containing fragments of axons and presynaptic terminals. Single- nucleus RNA sequencing indicated that cortical OPCs express key phagocytic genes, as well as neuronal transcripts, consistent with active axonal engulfment. PLs were ten times more abundant in OPCs than in microglia in P36 mice, and declined with age and lineage progression, suggesting that OPCs contribute very substantially to refinement of neuronal circuits during later phases of cortical development.

neuroscience↗

bioRxiv2021

Deep Learning-Based Phenotyping of Breast Cancer Cells Using Lens-free Digital In-line Holography

Lens-free digital in-line holography (LDIH) offers a wide field of view at micrometer-scale resolution, surpassing the capabilities of lens-based microscopes, making it a promising diagnostic tool for high-throughput cellular analysis. However, the complex nature of holograms renders them challenging for human interpretation, necessitating time- consuming computational processing to reconstruct object images. To address this, we present HoloNet, a novel deep learning architecture specifically designed for direct analysis of holographic images from LDIH in cellular phenotyping. HoloNet extracts both global features from diffraction patterns and local features from convolutional layers, achieving superior performance and interpretability compared to other deep learning methods. By leveraging raw holograms of breast cancer cells stained with well-known markers ER/PR and HER2, HoloNet demonstrates its effectiveness in classifying breast cancer cell types and quantifying molecular marker intensities. Furthermore, we introduce the feature-fusion HoloNet model, which extracts diffraction features associated with breast cancer cell types and their marker intensities. This hologram embedding approach allows for the identification of previously unknown subtypes of breast cancer cells, facilitating a comprehensive analysis of cell phenotype heterogeneity, leading to precise breast cancer diagnosis.

bioinformatics↗

bioRxiv2021

Deep Learning-based Subcellular Phenotyping of Protrusion Dynamics Reveals Fine Differential Drug Responses at Subcellular and Single-Cell Levels

Live cell imaging provides unparallel insights into dynamic cellular processes across spatiotemporal scales. Despite its potential, the inherent spatiotemporal heterogeneity within live cell imaging data often obscures critical mechanical details underlying cellular dynamics. Uncovering fine-grained phenotypes of live cell dynamics is pivotal for precise understandings of the heterogeneity of physiological and pathological processes. However, this endeavor introduces formidable technical challenges to unsupervised machine learning, demanding the extraction of features that can faithfully preserve heterogeneity, effectively discriminate between different molecularly perturbed states, and provide interpretability. While deep learning shows promise in extracting useful features from large datasets, it often falls short in producing such high-fidelity features, especially in unsupervised learning settings. To tackle these challenges, we present DeepHACX (Deep phenotyping of Heterogeneous Activities of Cellular dynamics with eXplanations), a self-training deep learning framework designed for fine-grained and interpretable phenotyping. This framework seamlessly integrates an unsupervised teacher model with interpretable features to facilitate feature learning in a student deep neural network (DNN). Significantly, it incorporates an autoencoder-based regularizer, termed SENSER (SENSitivity-enhancing autoEncoding Regularizer), designed to prompt the student DNN to maximize the heterogeneity associated with molecular perturbations. This approach enables the acquisition of features that not only discriminate between different molecularly perturbed states but also faithfully preserve the heterogeneity linked to these perturbations. In our study, DeepHACX successfully delineated fine-grained phenotypes within the heterogeneous protrusion dynamics of migrating epithelial cells, uncovering specific responses to pharmacological perturbations. Remarkably, DeepHACX adeptly captured a minimal number of highly interpretable features uniquely linked to these fine-grained phenotypes, each corresponding to specific temporal intervals crucial for their manifestation. This unique capability positions DeepHACX as a valuable tool for investigating diverse cellular dynamics and comprehensively studying their heterogeneity.

bioinformatics↗

bioRxiv2021

A locomotor neural circuit persists and functions similarly in larvae and adult Drosophila

Individual neurons can undergo drastic structural changes, known as neuronal remodeling or structural plasticity. One example of this is in response to hormones, such as during puberty in mammals or metamorphosis in insects. However, in each of these examples it remains unclear whether the remodeled neuron resumes prior patterns of connectivity, and if so, whether the persistent circuits drive similar behaviors. Here, we utilize a well-characterized neural circuit in the Drosophila larva: the Moonwalking Descending Neuron (MDN) circuit. We previously showed that larval MDN induces backward crawling, and synapses onto the Pair1 interneuron to inhibit forward crawling (Carreira-Rosario et al., 2018). MDN is remodeled during metamorphosis and regulates backward walking in the adult fly. We investigated whether Pair1 is remodeled during metamorphosis and functions within the MDN circuit during adulthood. We assayed morphology and molecular markers to demonstrate that Pair1 is remodeled during metamorphosis and persists in the adult fly. In the adult, optogenetic activation of Pair1 resulted in arrest of forward locomotion, similar to what is observed in larvae. MDN and Pair1 are also synaptic partners in the adult, showing that the MDN-Pair1 interneuron circuit is retained in the adult following hormone-driven pupal remodeling. Thus, the MDN-Pair1 neurons are an interneuronal circuit - i.e. a pair of synaptically connected interneurons - that persists through metamorphosis, taking on new input/output neurons, yet generating similar locomotor behavior at both stages.

neuroscience↗

bioRxiv2021

Detection of sleep apnea from single-channel electroencephalogram (EEG) using an explainable convolutional neural network

Sleep apnea (SA) is a common disorder involving the cessation of breathing during sleep. It can cause daytime hypersomnia, accidents, and, if allowed to progress, serious, chronic conditions. Continuous positive airway pressure is an effective SA treatment. However, long waitlists impede timely diagnosis; overnight sleep studies involve trained technicians scoring a polysomnograph, which comprises multiple physiological signals including multi-channel electroencephalography (EEG). Therefore, it is important to develop simplified and automated approaches to detect SA. We have developed an explainable convolutional neural network (CNN) to detect SA from single-channel EEG recordings which generalizes across subjects. The network architecture consisted of three convolutional layers. We tuned hyperparameters using the Hyperband algorithm, optimized parameters using Adam, and quantified network performance with subjectwise 10-fold cross-validation. Our CNN performed with an accuracy of 76.7% and a Matthews correlation coefficient (MCC) of 0.54. This performance was reliably above the conservative baselines of 50% (accuracy) and 0.0 (MCC). To explain the mechanisms of our trained network, we used critical-band masking (CBM): after training, we added bandlimited noise to test recordings; we parametrically varied the noise band center frequency and noise intensity, quantifying the deleterious effect on performance. We reconciled the effects of CBM with lesioning, wherein we zeroed the trained networks 1st-layer filter kernels in turn, quantifying the deleterious effect on performance. These analyses indicated that the network learned frequency-band information consistent with known SA biomarkers, specifically, delta and beta band activity. Our results indicate single-channel EEG may have clinical potential for SA diagnosis.

neuroscience↗