Transcriptional and epigenetic dynamics during specification of human embryonic stem cells.

Differentiation of human embryonic stem cells (hESCs) provides a unique opportunity to study the regulatory mechanisms that facilitate cellular transitions in a human context. To that end, we performed comprehensive transcriptional and epigenetic profiling of populations derived through directed differentiation of hESCs representing each of the three embryonic germ layers. Integration of whole-genome bisulfite sequencing, chromatin immunoprecipitation sequencing, and RNA sequencing reveals unique events associated with specification toward each lineage. Lineage-specific dynamic alterations in DNA methylation and H3K4me1 are evident at putative distal regulatory elements that are frequently bound by pluripotency factors in the undifferentiated hESCs. In addition, we identified germ-layer-specific H3K27me3 enrichment at sites exhibiting high DNA methylation in the undifferentiated state. A better understanding of these initial specification events will facilitate identification of deficiencies in current approaches, leading to more faithful differentiation strategies as well as providing insights into the rewiring of human regulatory programs during cellular transitions.

scBasset: sequence-based modeling of single-cell ATAC-seq using convolutional neural networks.

Single-cell assay for transposase-accessible chromatin using sequencing (scATAC) shows great promise for studying cellular heterogeneity in epigenetic landscapes, but there remain important challenges in the analysis of scATAC data due to the inherent high dimensionality and sparsity. Here we introduce scBasset, a sequence-based convolutional neural network method to model scATAC data. We show that by leveraging the DNA sequence information underlying accessibility peaks and the expressiveness of a neural network model, scBasset achieves state-of-the-art performance across a variety of tasks on scATAC and single-cell multiome datasets, including cell clustering, scATAC profile denoising, data integration across assays and transcription factor activity inference.

Semisupervised adversarial neural networks for single-cell classification.

Annotating cell identities is a common bottleneck in the analysis of single-cell genomics experiments. Here, we present scNym, a semisupervised, adversarial neural network that learns to transfer cell identity annotations from one experiment to another. scNym takes advantage of information in both labeled data sets and new, unlabeled data sets to learn rich representations of cell identity that enable effective annotation transfer. We show that scNym effectively transfers annotations across experiments despite biological and technical differences, achieving performance superior to existing methods. We also show that scNym models can synthesize information from multiple training and target data sets to improve performance. We show that in addition to high accuracy, scNym models are well calibrated and interpretable with saliency methods.

Effective gene expression prediction from sequence by integrating long-range interactions.

How noncoding DNA determines gene expression in different cell types is a major unsolved problem, and critical downstream applications in human genetics depend on improved solutions. Here, we report substantially improved gene expression prediction accuracy from DNA sequences through the use of a deep learning architecture, called Enformer, that is able to integrate information from long-range interactions (up to 100 kb away) in the genome. This improvement yielded more accurate variant effect predictions on gene expression for both natural genetic variants and saturation mutagenesis measured by massively parallel reporter assays. Furthermore, Enformer learned to predict enhancer-promoter interactions directly from the DNA sequence competitively with methods that take direct experimental data as input. We expect that these advances will enable more effective fine-mapping of human disease associations and provide a framework to interpret cis-regulatory evolution.

Predicting 3D genome folding from DNA sequence with Akita.

In interphase, the human genome sequence folds in three dimensions into a rich variety of locus-specific contact patterns. Cohesin and CTCF (CCCTC-binding factor) are key regulators; perturbing the levels of either greatly disrupts genome-wide folding as assayed by chromosome conformation capture methods. Still, how a given DNA sequence encodes a particular locus-specific folding pattern remains unknown. Here we present a convolutional neural network, Akita, that accurately predicts genome folding from DNA sequence alone. Representations learned by Akita underscore the importance of an orientation-specific grammar for CTCF binding sites. Akita learns predictive nucleotide-level features of genome folding, revealing effects of nucleotides beyond the core CTCF motif. Once trained, Akita enables rapid in silico predictions. Accounting for this, we demonstrate how Akita can be used to perform in silico saturation mutagenesis, interpret eQTLs, make predictions for structural variants and probe species-specific genome folding. Collectively, these results enable decoding genome function from sequence through structure.

Spectral widths and Stokes shifts in InP-based quantum dots.

InP-based quantum dots (QDs) have Stokes shifts and photoluminescence (PL) line widths that are larger than in II-VI semiconductor QDs with comparable exciton energies. The mechanisms responsible for these spectral characteristics are investigated in this paper. Upon comparing different semiconductors, we find the Stokes shift decreases in the following order: InP > CdTe > CdSe. We also find that the Stokes shift decreases with core size and decreases upon deposition of a ZnSe shell. We suggest that the Stokes shift is largely due to different absorption and luminescent states in the angular momentum fine structure. The energy difference between the fine structure levels, and hence the Stokes shifts, are controlled by the electron-hole exchange interaction. Luminescence polarization results are reported and are consistent with this assignment. Spectral widths are controlled by the extent of homogeneous and inhomogeneous broadening. We report PL and PL excitation (PLE) spectra that facilitate assessing the roles of homogeneous and different inhomogeneous broadening mechanisms in the spectra of zinc-treated InP and InP/ZnSe/ZnS particles. There are two distinct types of inhomogeneous broadening: size inhomogeneity and core-shell interface inhomogeneity. The latter results in a distribution of core-shell band offsets and is caused by interfacial dipoles associated with In-Se or P-Zn bonding. Quantitative modeling of the spectra shows that the offset inhomogeneity is comparable to but somewhat smaller than the size inhomogeneity. The combination of these two types of inhomogeneity also explains several aspects of reversible hole trapping dynamics involving localized In3+/VZn2- impurity states in the ZnSe shells.

Murine single-cell RNA-seq reveals cell-identity- and tissue-specific trajectories of aging.

Aging is a pleiotropic process affecting many aspects of mammalian physiology. Mammals are composed of distinct cell type identities and tissue environments, but the influence of these cell identities and environments on the trajectory of aging in individual cells remains unclear. Here, we performed single-cell RNA-seq on >50,000 individual cells across three tissues in young and old mice to allow for direct comparison of aging phenotypes across cell types. We found transcriptional features of aging common across many cell types, as well as features of aging unique to each type. Leveraging matrix factorization and optimal transport methods, we found that both cell identities and tissue environments exert influence on the trajectory and magnitude of aging, with cell identity influence predominating. These results suggest that aging manifests with unique directionality and magnitude across the diverse cell identities in mammals.

Biexciton and trion dynamics in InP/ZnSe/ZnS quantum dots.

Transient absorption (TA) and time-resolved photoluminescence (PL) spectroscopies have been used to elucidate the hole tunneling and Auger dynamics in biexcitons and negative trions in high-quality InP/ZnSe/ZnS quantum dots (QDs). In a previous paper [Nguyen et al., J. Phys. Chem. C 125, 15405-15414 (2021)], we showed that under high-intensity photoexcitation, two types of biexcitons are formed: those having two conduction band electrons and two valence band holes (designated as an XX state) and those having two conduction band electrons, one valence band hole, and an additional trapped hole (designated as an XT state). In the present paper, we show that both types of biexcitons can undergo Auger processes, with those of the XT state being a factor of four to five slower than those of the XX state. In addition, the trapped holes can undergo tunneling into the valence band, converting an XT state to an XX state. The relative amplitudes of the fast (XX) and slow (XT) components are different in the TA and PL kinetics, and these differences can be quantitatively understood in terms of oscillator strengths and electron-hole overlap integrals of each state. XT to XX hole tunneling rates are obtained from the comparison of the XT state lifetimes with those of the negative trions. This comparison shows that the tunneling times decrease with decreasing core size and shell thickness. These times are about 2 ns for the thinnest shell red-emitting QDs and decrease to 330 ps for QDs that luminesce in the yellow.

Identity of the reversible hole traps in InP/ZnSe core/shell quantum dots.

Density functional theory calculations are combined with time-resolved photoluminescence experiments to identify the species responsible for the reversible trapping of holes following photoexcitation of InP/ZnSe/ZnS core/shell/shell quantum dots (QDs) having excess indium in the shell [P. Cavanaugh et al., J. Chem. Phys. 155, 244705 (2021)]. Several possible assignments are considered, and a substitutional indium adjacent to a zinc vacancy, In3+/VZn 2-, is found to be the most likely. This assignment is consistent with the observation that trapping occurs only when the QD has excess indium and is supported by experiments showing that the addition of zinc oleate or acetate decreases the extent of trapping, presumably by filling some of the vacancy traps. We also show that the addition of alkyl carboxylic acids causes increased trapping, presumably by the creation of additional zinc vacancies. The calculations show that either a single In2+ ion or an In2+-In3+ dimer is much too easily oxidized to form the reversible traps observed experimentally, while In3+ is far too difficult to oxidize. Additional experimental data on InP/ZnSe/ZnS QDs synthesized in the absence of chloride demonstrates that the reversible traps are not associated with Cl-. However, a zinc vacancy adjacent to a substitutional indium is calculated to have its highest occupied orbitals about 1 eV above the top of the valence band of bulk ZnSe, in the appropriate energy range to act as reversible traps for quantum confined holes in the InP valence band. The associated orbitals are predominantly composed of p orbitals on the Se atoms adjacent to the Zn vacancy.

Effects of using an abdominal simulator to develop palpatory competencies in 3rd year medical students.

BACKGROUND: Medical school faculty are hard pressed to provide clerkship students with sufficient opportunity to develop and practice their capacity to perform a competent clinical examination, including the palpatory examination of the abdomen. We evaluated the impact of training with an abdominal simulator, AbSim, designed to monitor the depth, location, and thoroughness of their palpation and to provide concurrent and summative feedback regarding their performance. METHODS: All third-year medical students were given the opportunity to develop their palpatory skills with the AbSim simulator during the family medicine rotation. The performance of those who studied with the simulator was measured by its sensors, before and after a training session that included visual feedback regarding the depth and coverage of the student's manual pressure. Additionally, all students reported their confidence in their evolving abdominal palpation skills at the beginning and end of the rotation. RESULTS: 119 (86.9%) of 137 students filled out the initial questionnaire, and 73 (61.3%) studied with the abdominal simulator. The training produced a highly significant improvement in their overall performance (4 measures, p's < 0.001). Pre-training performance (depth calibration and thoroughness of coverage) was not related to the number of months of previous clinical rotations nor to previous internal medicine or surgery rotations. There was little relation between students' confidence in their abdominal examination skills and objective measures of their palpatory performance; however, students who chose the training started with less confidence, and became more confident after training. CONCLUSIONS: Guided abdominal simulator practice increased medical students' capacity to perform an abdominal examination with more appropriate depth and thoroughness of palpation. Interpretation of changes in confidence are uncertain, because confidence was unrelated to objectively measured performance. However, students with low initial confidence in their abdominal examination seemed to be more likely to choose to study with the abdominal simulator.

Sequential regulatory activity prediction across chromosomes with convolutional neural networks.

Models for predicting phenotypic outcomes from genotypes have important applications to understanding genomic function and improving human health. Here, we develop a machine-learning system to predict cell-type-specific epigenetic and transcriptional profiles in large mammalian genomes from DNA sequence alone. By use of convolutional neural networks, this system identifies promoters and distal regulatory elements and synthesizes their content to make effective gene expression predictions. We show that model predictions for the influence of genomic variants on gene expression align well to causal variants underlying eQTLs in human populations and can be useful for generating mechanistic hypotheses to enable fine mapping of disease loci.

Cross-species regulatory sequence activity prediction.

Machine learning algorithms trained to predict the regulatory activity of nucleic acid sequences have revealed principles of gene regulation and guided genetic variation analysis. While the human genome has been extensively annotated and studied, model organisms have been less explored. Model organism genomes offer both additional training sequences and unique annotations describing tissue and cell states unavailable in humans. Here, we develop a strategy to train deep convolutional neural networks simultaneously on multiple genomes and apply it to learn sequence predictors for large compendia of human and mouse data. Training on both genomes improves gene expression prediction accuracy on held out and variant sequences. We further demonstrate a novel and powerful approach to apply mouse regulatory models to analyze human genetic variants associated with molecular phenotypes and disease. Together these techniques unleash thousands of non-human epigenetic and transcriptional profiles toward more effective investigation of how gene regulation affects human disease.

Radiative dynamics and delayed emission in pure and doped InP/ZnSe/ZnS quantum dots.

We have used time-correlated single photon counting to elucidate the radiative dynamics of InP/ZnSe/ZnS core/shell/shell quantum dots (QDs) that differ in the amount and distribution of excess indium. Stoichiometric QDs having an In:P atom ratio very near unity exhibit simple luminescence kinetics. The photoluminescence (PL) rises with the 40 ps instrument response function and exhibits a decay that is close to a single exponential with a time constant that decreases from 32 to 28 ns with increasing shell thickness. QDs having excess indium (In:P ratio of 1.15-1.63) show a significant component of a slower rise time assigned to transient population of indium-based hole traps in the ZnSe shell. They also have a slower PL decay, attributed to an equilibrium between these traps, which are optically dark, and the emissive valence-band state. This results in a radiative lifetime that increases from 32 to 48 ns with increasing shell thickness. Different treatments of the InP cores prior to shell deposition result in different core/shell interfaces as indicated by resonance Raman spectroscopy, as well as differences in the amplitude and timescale of the slow PL rise and the PL decay time. These are interpreted in terms of different radial distributions of the indium-based hole traps, which can be related to differences in the interfacial lattice strain.

Muscle insulin resistance in type 1 diabetes with coronary artery disease.

AIMS/HYPOTHESIS: The risk for coronary artery disease (CAD) is substantially increased in type 1 diabetes and it has been postulated that insulin resistance may contribute to this risk. The current study measured insulin resistance in type 1 diabetes with vs without CAD and with a focus upon skeletal muscle, to test the hypothesis that insulin resistance is more severe in participants who have type 1 diabetes and CAD. Additionally, in type 1 diabetes, we examined the hypothesis that insulin resistance is more severe in soleus (an oxidative type muscle) vs tibialis anterior (a more glycolytic type of muscle). METHODS: Insulin resistance was measured in participants with type 1 diabetes with (n = 9, CAD+) and without CAD (n = 10, CAD-) using euglycaemic insulin infusions combined with positron emission tomography (PET) imaging of [18F]fluorodeoxyglucose (FDG) uptake into soleus and tibialis anterior skeletal muscles. Coronary artery calcium (CAC) score was quantified by electron beam tomography. RESULTS: CAD+ participants with type 1 diabetes had a >100-fold higher CAC score than did CAD- participants with type 1 diabetes but groups did not differ in HbA1c or insulin dose. During clamp studies, CAD+ and CAD- groups had similar glucose disposal but were insulin resistant compared with historical non-diabetic participants (n = 13). FDG uptake by soleus muscle was similarly reduced, overall, in individuals with type 1 diabetes with or without CAD compared with non-diabetic individuals. However, FDG uptake by tibialis anterior muscle was not reduced in CAD- participants with type 1 diabetes while in CAD+ participants with type 1 diabetes it was 75% greater (p < 0.01). Across all participants with type 1 diabetes, FDG uptake by tibialis anterior muscle correlated positively with CAC severity. CONCLUSIONS/INTERPRETATION: Our study confirms that systemic and skeletal muscle-specific insulin resistance is seen in type 1 diabetes but found that it does not appear to be more severe in the presence of CAD. There were, however, sharp differences between soleus and tibialis anterior muscles in type 1 diabetes: while insulin resistance was clearly manifest in soleus muscle, and was of equal severity in CAD+ and CAD- participants, tibialis anterior did not suggest insulin resistance in participants with type 1 diabetes, as FDG uptake by tibialis anterior correlated positively with CAC severity and was significantly increased in participants with type 1 diabetes and clinical CAD. Graphical abstract.

Associations Between the Specialty of Opioid Prescribers and Opioid Addiction, Misuse, and Overdose Outcomes.

OBJECTIVE: To examine associations between opioid prescriber specialty and patient likelihood of opioid use disorder (OUD), opioid misuse, and opioid overdose. DESIGN: Longitudinal retrospective study using Pennsylvania Medicaid data (2007-2015). METHODS: We constructed an incident cohort of 432,110 enrollees initiating prescription opioid use without a history of OUD or overdose six months before opioid initiation. We attributed patients to one of 10 specialties using the first opioid prescriber's specialty or, alternatively, the specialty of the dominant prescriber writing the majority of the patient's opioid prescriptions. We estimated adjusted rates for OUD, misuse, and overdose, adjusting for demographic variables and medical (including pain) and psychiatric comorbidities. RESULTS: The unadjusted incidence rates of OUD, misuse, and overdose were 7.13, 4.73, and 0.69 per 100,000 person-days, respectively. Patients initiating a new episode of opioid treatment with Pain Medicine/Anesthesiology (6.7 events, 95% confidence interval [CI] = 5.5 to 8.2) or Physical Medicine and Rehabilitation (PM&R; 6.1 events, 95% CI = 5.1 to 7.2) had higher adjusted rates for OUD per 100,000 person-days compared with Primary Care practitioners (PCPs; 4.4 events, 95% CI = 4.1 to 4.7). Patients with index prescriptions from Pain Medicine/Anesthesiology (15.9 events, 95% CI = 13.2 to 19.3) or PM&R (15.8 events, 95% CI = 13.5 to 18.4) had higher adjusted rates for misuse per 100,000 person-days compared with PCPs (9.6 events, 95% CI = 8.8 to 10.6). Findings were largely similar when patients were attributed to specialty based on dominant prescriber. CONCLUSIONS: Differences in opioid-related risks by specialty of opioid prescriber may arise from differences in patient risk factors, provider behavior, or both. Our findings inform targeting of opioid risk mitigation strategies to specific practitioner specialties.

Spatial and temporal studies of metabolic activity: contrasting biochemical kinetics in tissues and pathways during fasted and fed states.

The regulation of nutrient homeostasis, i.e., the ability to transition between fasted and fed states, is fundamental in maintaining health. Since food is typically consumed over limited (anabolic) periods, dietary components must be processed and stored to counterbalance the catabolic stress that occurs between meals. Herein, we contrast tissue- and pathway-specific metabolic activity in fasted and fed states. We demonstrate that knowledge of biochemical kinetics that is obtained from opposite ends of the energetic spectrum can allow mechanism-based differentiation of healthy and disease phenotypes. Rat models of type 1 and type 2 diabetes serve as case studies for probing spatial and temporal patterns of metabolic activity via [2H]water labeling. Experimental designs that capture integrative whole body metabolism, including meal-induced substrate partitioning, can support an array of research surrounding metabolic disease; the relative simplicity of the approach that is discussed here should enable routine applications in preclinical models.

Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks.

The complex language of eukaryotic gene expression remains incompletely understood. Despite the importance suggested by many noncoding variants statistically associated with human disease, nearly all such variants have unknown mechanisms. Here, we address this challenge using an approach based on a recent machine learning advance-deep convolutional neural networks (CNNs). We introduce the open source package Basset to apply CNNs to learn the functional activity of DNA sequences from genomics data. We trained Basset on a compendium of accessible genomic sites mapped in 164 cell types by DNase-seq, and demonstrate greater predictive accuracy than previous methods. Basset predictions for the change in accessibility between variant alleles were far greater for Genome-wide association study (GWAS) SNPs that are likely to be causal relative to nearby SNPs in linkage disequilibrium with them. With Basset, a researcher can perform a single sequencing assay in their cell type of interest and simultaneously learn that cell's chromatin accessibility code and annotate every mutation in the genome with its influence on present accessibility and latent potential for accessibility. Thus, Basset offers a powerful computational approach to annotate and interpret the noncoding genome.

PAN-AMPK Activation Improves Renal Function in a Rat Model of Progressive Diabetic Nephropathy.

Metabolic dysregulation and mitochondrial dysfunction are important features of acute and chronic tissue injury across species, and human genetics and preclinical data suggest that the master metabolic regulator 5'-adenosine monophosphate-activated protein kinase (AMPK) may be an effective therapeutic target for chronic kidney disease (CKD). We have recently disclosed a pan-AMPK activator, MK-8722, that was shown to have beneficial effects in preclinical models. In this study we investigated the effects of MK-8722 in a progressive rat model of diabetic nephropathy to determine whether activation of AMPK would be of therapeutic benefit. We found that MK-8722 administration in a therapeutic paradigm is profoundly renoprotective, as demonstrated by a reduction in proteinuria (63% decrease in MK-8722 10 mg/kg per day compared with vehicle group) and a significant improvement in glomerular filtration rate (779 and 430 µl/min per gram kidney weight in MK-8722 10 mg/kg per day and vehicle group, respectively), as well as improvements in kidney fibrosis. We provide evidence that the therapeutic effects of MK-8722 may be mediated by modulation of renal mitochondrial quality control as well by attenuating fibrotic and lipotoxic mechanisms in kidney cells. MK-8722 (10 mg/kg per day compared with vehicle group) achieved modest blood pressure reduction (10 mmHg lower for mean blood pressure) and significant metabolic improvements (decreased plasma glucose, triglyceride, and body weight) that could contribute to renoprotection. These data further validate the concept that targeting metabolic dysregulation in CKD could be a potential therapeutic approach. SIGNIFICANCE STATEMENT: We demonstrate in the present study that the pharmacological activation of AMPK using a small-molecule agent provided renoprotection and improved systemic and cellular metabolism. We further indicate that modulation of renal mitochondrial quality control probably contributed to renoprotection and was distinct from the effects of enalapril. Our findings suggest that improving renal mitochondrial biogenesis and function and attenuating fibrosis and lipotoxicity by targeting key metabolic nodes could be a potential therapeutic approach in management of CKD that could complement the current standard of care.

The Role of Primary Care in Improving Access to Medication-Assisted Treatment for Rural Medicaid Enrollees with Opioid Use Disorder.

BACKGROUND: The opioid epidemic has disproportionately affected rural areas, where a limited number of health care providers offer medication-assisted treatment (MAT), the mainstay of treatment for opioid use disorder (OUD). Rural residents with OUD may face multiple barriers to engagement in MAT including long travel distances. OBJECTIVE: To examine the degree to which rural residents with OUD are engaged with primary care providers (PCPs), describe the role of rural PCPs in MAT delivery, and estimate the association between enrollee distance to MAT prescribers and MAT utilization. DESIGN: Retrospective cohort study. PARTICIPANTS: Medicaid-enrolled adults diagnosed with OUD in 23 rural Pennsylvania counties. MAIN MEASURES: Primary care utilization, MAT utilization, distance to nearest possible MAT prescriber, mean distance traveled to actual MAT prescribers, and continuity of pharmacotherapy. KEY RESULTS: Of the 7930 Medicaid enrollees with a diagnosis of OUD, a minority (18.6%) received their diagnosis during a PCP visit even though enrollees with OUD had 4.1 visits to PCPs per person-year in 2015. Among enrollees with an OUD diagnosis recorded during a PCP visit, about half (751, 50.8%) received MAT, most of whom (508, 67.6%) received MAT from a PCP. Enrollees with OUD with at least one PCP visit were more likely than those without a PCP visit to receive MAT (32.7% vs. 25%; p < 0.001), and filled more buprenorphine and naltrexone prescriptions (mean = 11.1 vs. 9.3; p < 0.001). The median of the distances traveled to actual MAT prescribers was 48.8 miles, compared to a median of 4.2 miles to the nearest available MAT prescriber. Enrollees traveling a mean distance greater than 45 miles to MAT prescribers were less likely to receive continuity of pharmacotherapy (OR = 0.71, 95% CI = 0.56-0.91, p = 0.007). CONCLUSIONS: PCP utilization among rural Medicaid enrollees diagnosed with OUD is high, presenting a potential intervention point to treat OUD, particularly if the enrollee's PCP is located nearer than their MAT prescriber.