Deep Learning Sequence Models for Transcriptional Regulation.

Deciphering the regulatory code of gene expression and interpreting the transcriptional effects of genome variation are critical challenges in human genetics. Modern experimental technologies have resulted in an abundance of data, enabling the development of sequence-based deep learning models that link patterns embedded in DNA to the biochemical and regulatory properties contributing to transcriptional regulation, including modeling epigenetic marks, 3D genome organization, and gene expression, with tissue and cell-type specificity. Such methods can predict the functional consequences of any noncoding variant in the human genome, even rare or never-before-observed variants, and systematically characterize their consequences beyond what is tractable from experiments or quantitative genetics studies alone. Recently, the development and application of interpretability approaches have led to the identification of key sequence patterns contributing to the predicted tasks, providing insights into the underlying biological mechanisms learned and revealing opportunities for improvement in future models.

Protective Effect of Betulin on Streptozotocin-Nicotinamide-Induced Diabetes in Female Rats.

Type 2 diabetes is characterized by hyperglycemia and a relative loss of ß-cell function. Our research investigated the antidiabetic potential of betulin, a pentacyclic triterpenoid found primarily in birch bark and, intriguingly, in a few marine organisms. Betulin has been shown to possess diverse biological activities, including antioxidant and antidiabetic activities; however, no studies have fully explored the effects of betulin on the pancreas and pancreatic islets. In this study, we investigated the effect of betulin on streptozotocin-nicotinamide (STZ)-induced diabetes in female Wistar rats. Betulin was prepared as an emulsion, and intragastric treatments were administered at doses of 20 and 50 mg/kg for 28 days. The effect of treatment was assessed by analyzing glucose parameters such as fasting blood glucose, hemoglobin A1C, and glucose tolerance; hepatic and renal biomarkers; lipid peroxidation; antioxidant enzymes; immunohistochemical analysis; and hematological indices. Administration of betulin improved the glycemic response and decreased α-amylase activity in diabetic rats, although insulin levels and homeostatic model assessment for insulin resistance (HOMA-IR) scores remained unchanged. Furthermore, betulin lowered the levels of hepatic biomarkers (aspartate aminotransferase, alanine aminotransferase, and alpha-amylase activities) and renal biomarkers (urea and creatine), in addition to improving glutathione levels and preventing the elevation of lipid peroxidation in diabetic animals. We also found that betulin promoted the regeneration of ß-cells in a dose-dependent manner but did not have toxic effects on the pancreas. In conclusion, betulin at a dose of 50 mg/kg exerts a pronounced protective effect against cytolysis, diabetic nephropathy, and damage to the acinar pancreas and may be a potential treatment option for diabetes.

Accelerated Generation of Extra-Islet Insulin-Producing Cells in Diabetic Rats, Treated with Sodium Phthalhydrazide.

ß-cells dysfunction plays an important role in the pathogenesis of type 2 diabetes (T2D), partially may be compensated by the generation of extra-islet insulin-producing cells (IPCs) in pancreatic acini and ducts. Pdx1 expression and inflammatory level are suggested to be involved in the generation of extra-islet IPCs, but the exact reasons and mechanisms of it are unclear. Macrophages are key inflammatory mediators in T2D. We studied changes in mass and characteristics of extra-islet IPCs in rats with a streptozotocin-nicotinamide model of T2D and after i.m. administration of 20 daily doses of 2 mg/kg b.w. sodium aminophthalhydrazide (APH). Previously, we found that APH modulates macrophage production and increases the proliferative activity of pancreatic ß-cells. Expressions of insulin and Pdx1, as well as F4/80 (macrophage marker), were detected at the protein level by immunohistochemistry analysis, the concentration of pro- and anti-inflammatory cytokines in blood and pancreas-by ELISA. Diabetic rats treated with APH showed an increasing mass of extra-islet IPCs and the content of insulin in them. The presence of Pdx1+ cells in the exocrine pancreas also increased. F4/80+ cell reduction was accompanied by increasing TGF-ß1 content. Interestingly, during the development of diabetes, the mass of ß-cells decreased faster than the mass of extra-islet IPCs, and extra-islet IPCs reacted to experimental T2D differently depending on their acinar or ductal location.

Type 1 Diabetes Impairs Cardiomyocyte Contractility in the Left and Right Ventricular Free Walls but Preserves It in the Interventricular Septum.

Type 1 diabetes (T1D) leads to ischemic heart disease and diabetic cardiomyopathy. We tested the hypothesis that T1D differently affects the contractile function of the left and right ventricular free walls (LV, RV) and the interventricular septum (IS) using a rat model of alloxan-induced T1D. Single-myocyte mechanics and cytosolic Ca2+ concentration transients were studied on cardiomyocytes (CM) from LV, RV, and IS in the absence and presence of mechanical load. In addition, we analyzed the phosphorylation level of sarcomeric proteins and the characteristics of the actin-myosin interaction. T1D similarly affected the characteristics of actin-myosin interaction in all studied regions, decreasing the sliding velocity of native thin filaments over myosin in an in vitro motility assay and its Ca2+ sensitivity. A decrease in the thin-filament velocity was associated with increased expression of ß-myosin heavy-chain isoform. However, changes in the mechanical function of single ventricular CM induced by T1D were different. T1D depressed the contractility of CM from LV and RV; it decreased the auxotonic tension amplitude and the slope of the active tension-length relationship. Nevertheless, the contractile function of CM from IS was principally preserved.

Contrastive pre-training for sequence based genomics models.

MOTIVATION: In recent years deep learning has become one of the central approaches in a number of applications, including many tasks in genomics. However, as models grow in depth and complexity, they either require more data or a strategic initialization technique to improve performance. RESULTS: In this project, we introduce cGen, a novel unsupervised, model-agnostic contrastive pre-training method for sequence-based models. cGen can be used before training to initialize weights, reducing the size of the dataset needed. It works through learning the intrinsic features of the reference genome and makes no assumptions on the underlying structure. We show that the embeddings produced by the unsupervised model are already informative for gene expression prediction and that the sequence features provide a meaningful clustering. We demonstrate that cGen improves model performance in various sequence-based deep learning applications, such as chromatin profiling prediction and gene expression. Our findings suggest that using cGen, particularly in areas constrained by data availability, could improve the performance of deep learning genomic models without the need to modify the model architecture.

Atlas of primary cell-type-specific sequence models of gene expression and variant effects.

Human biology is rooted in highly specialized cell types programmed by a common genome, 98% of which is outside of genes. Genetic variation in the enormous noncoding space is linked to the majority of disease risk. To address the problem of linking these variants to expression changes in primary human cells, we introduce ExPectoSC, an atlas of modular deep-learning-based models for predicting cell-type-specific gene expression directly from sequence. We provide models for 105 primary human cell types covering 7 organ systems, demonstrate their accuracy, and then apply them to prioritize relevant cell types for complex human diseases. The resulting atlas of sequence-based gene expression and variant effects is publicly available in a user-friendly interface and readily extensible to any primary cell types. We demonstrate the accuracy of our approach through systematic evaluations and apply the models to prioritize ClinVar clinical variants of uncertain significance, verifying our top predictions experimentally.

Hepatic insulin synthesis increases in rat models of diabetes mellitus type 1 and 2 differently.

Insulin-positive (+) cells (IPCs), detected in multiple organs, are of great interest as a probable alternative to ameliorate pancreatic beta-cells dysfunction and insulin deficiency in diabetes. Liver is a potential source of IPCs due to it common embryological origin with pancreas. We previously demonstrated the presence of IPCs in the liver of healthy and diabetic rats, but detailed description and analysis of the factors, which potentially can induced ectopic hepatic expression of insulin in type 1 (T1D) and type 2 diabetes (T2D), were not performed. In present study we evaluate mass of hepatic IPCs in the rat models of T1D and T2D and discuss factors, which may stimulate it generation: glycaemia, organ injury, involving of hepatic stem/progenitor cell compartment, expression of transcription factors and inflammation. Quantity of IPCs in the liver was up by 1.7-fold in rats with T1D and 10-fold in T2D compared to non-diabetic (ND) rats. We concluded that ectopic hepatic expression of insulin gene is activated by combined action of a number of factors, with inflammation playing a decision role.