Background: microRNAs (miRNAs) have been shown to play essential roles in a wide range of biological processes. Many computational methods have been developed to identify targets of miRNAs. However, the majority of these methods depend on pre-defined features that require considerable efforts and resources to compute and often prove suboptimal at predicting miRNA targets. Results: We developed a novel hybrid deep learning-based (DL-based) approach that is capable of predicting miRNA targets at a higher accuracy. This approach integrates convolutional neural networks (CNNs) that excel in learning spatial features and recurrent neural networks (RNNs) that discern sequential features. Therefore, our approach has the advantages of learning both the intrinsic spatial and sequential features of miRNA:target. The inputs for our approach are raw sequences of miRNAs and genes that can be obtained effortlessly. We applied our approach on two human datasets from recently miRNA target prediction studies and trained two models. We demonstrated that the two models consistently outperform the previous methods according to evaluation metrics on test datasets. Comparing our approach with currently available alternatives on independent datasets shows that our approach delivers substantial improvements in performance. We also show with multiple evidences that our approach is more robust than other methods on small datasets. Our study is the first study to perform comparisons across multiple existing DL-based approaches on miRNA target prediction. Furthermore, we examined the contribution of a Max pooling layer in between the CNN and RNN and demonstrated that it improves the performance of all our models. Finally, a unified model was developed that is robust on fitting different input datasets. Conclusions: We present a new DL-based approach for predicting miRNA targets and demonstrate that our approach outperforms the current alternatives. We supplied an easy-to-use tool, miTAR, at https://github.com/tjgu/miTAR. Furthermore, our analysis results support that Max Pooling generally benefits the hybrid models and potentially prevents overfitting for hybrid models.
4 February 2025 in Journal of Clinical Investigation
Constitutively active mutations of KRAS are prevalent in non–small cell lung cancer (NSCLC). However, the relationship between these mutations and resistance to platinum-based chemotherapy and the underlying mechanisms remain elusive. In this study, we demonstrate that KRAS mutants confer resistance to platinum in NSCLC. Mechanistically, KRAS mutants mediate platinum resistance in NSCLC cells by activating ERK/JNK signaling, which inhibits AlkB homolog 5 (ALKBH5) N6 -methyladenosine (m 6 A) demethylase activity by regulating posttranslational modifications (PTMs) of ALKBH5. Consequently, the KRAS mutant leads to a global increase in m 6 A methylation of mRNAs, particularly damage-specific DNA-binding protein 2 ( DDB2 ) and XPC , which are essential for nucleotide excision repair. This methylation stabilized the mRNA of these 2 genes, thus enhancing NSCLC cells’ capability to repair platinum-induced DNA damage and avoid apoptosis, thereby contributing to drug resistance. Furthermore, blocking KRAS-mutant–induced m 6 A methylation, either by overexpressing a SUMOylation-deficient mutant of ALKBH5 or by inhibiting methyltransferase-like 3 (METTL3) pharmacologically, significantly sensitizes KRAS -mutant NSCLC cells to platinum drugs in vitro and in vivo. Collectively, our study uncovers a mechanism that mediates KRAS-mutant–induced chemoresistance in NSCLC cells by activating DNA repair through the modulation of the ERK/JNK/ALKBH5 PTM-induced m 6 A modification in DNA damage repair–related genes.
The Streptococcus mutans Cid/Lrg system is central to the physiology of this cariogenic organism, affecting oxidative stress resistance, biofilm formation and competence. Previous transcriptome analyses of lytS (responsible for the regulation of lrgAB expression) and cidB mutants have revealed pleiotropic effects on carbohydrate metabolism and stress resistance genes. In this study, it was found that an lrgAB mutant, previously shown to have diminished aerobic and oxidative stress growth, was also much more growth impaired in the presence of heat and vancomycin stresses, relative to wild-type, lrgA and lrgB mutants. To obtain a more holistic picture of LrgAB and its involvement in stress resistance, RNA sequencing and bioinformatics analyses were used to assess the transcriptional response of wild-type and isogenic lrgAB mutants under anaerobic (control) and stress-inducing culture conditions (aerobic, heat and vancomycin). Hierarchical clustering and principal components analyses of all differentially expressed genes revealed that the most distinct gene expression profiles between S. mutans UA159 and lrgAB mutant occurred during aerobic and high-temperature growth. Similar to previous studies of a cidB mutant, lrgAB stress transcriptomes were characterized by a variety of gene expression changes related to genomic islands, CRISPR-C as systems, ABC transporters, competence, bacteriocins, glucosyltransferases, protein translation, tricarboxylic acid cycle, carbohydrate metabolism/storage and transport. Notably, expression of lrgAB was upregulated in the wild-type strain under all three stress conditions. Collectively, these results demonstrate that mutation of lrgAB alters the transcriptional response to stress, and further support the idea that the Cid/Lrg system acts to promote cell homeostasis in the face of environmental stress.
MicroRNAs (miRNA) are short non-coding RNAs widely implicated in gene regulation. Most metazoan miRNAs utilize the RNase III enzymes Drosha and Dicer for biogenesis. One notable exception is the RNA polymerase II transcription start sites (TSS) miRNAs whose biogenesis does not require Drosha. The functional importance of the TSS-miRNA biogenesis is uncertain. To better understand the function of TSS-miRNAs, we applied a modifiedCrosslinking,Ligation,andSequencing ofHybrids on Argonaute (AGO-qCLASH) to identify the targets for TSS-miRNAs in HCT116 colorectal cancer cells with or withoutDROSHAknockout. We observed that miR-320a hybrids dominate in TSS-miRNA hybrids identified by AGO-qCLASH. Targets for miR-320a are enriched for the eIF2 signaling pathway, a downstream component of the unfolded protein response. Consistently, in miR-320a mimic- and antagomir- transfected cells, differentially expressed gene products are associated with eIF2 signaling. Within the AGO-qCLASH data, we identified the endoplasmic reticulum (ER) chaperone calnexin as a direct miR-320a down-regulated target, thus connecting miR-320a to the unfolded protein response. During ER stress, but not amino acid deprivation, miR-320a up-regulates ATF4, a critical transcription factor for resolving ER stress. In summary, our study investigates the targetome of the TSS-miRNAs in colorectal cancer cells and establishes miR-320a as a regulator of unfolded protein response.
5-hydroxymethylcytosine, also known as the sixth DNA base of the genome, plays an important role in brain aging and neurological disorders such as Alzheimer’s disease. However, little is known about its genome-wide distribution and its association with Alzheimer’s disease pathology. Here, we report a genome-wide profiling of 5-hydroxymethylcytosine in 1079 autopsied brains (dorsolateral prefrontal cortex) of older individuals and assess its association with multiple measures of Alzheimer’s disease pathologies, including pathological diagnosis of Alzheimer’s disease, amyloid-β load, and PHFtau tangle density. Of 197,765 5-hydroxymethylcytosine regions detected, we identified 2821 differentially hydroxymethylated regions associated with Alzheimer’s disease neuropathology after controlling for multiple testing and covariates. Many differentially hydroxymethylated regions are located within known Alzheimer’s disease loci, such as RIN3, PLCG2, ITGA2B, and USP6NL. Integrative multi-omics analyses support a potential mechanistic role of 5-hydroxymethylcytosine alterations in Alzheimer’s disease. Our study presents a large-scale genome-wide atlas of 5-hydroxymethylcytosine in Alzheimer’s brain and offers insight into the mechanism underlying Alzheimer’s disease pathogenesis.
Tumor-associated macrophages (TAMs) are abundant in the tumor microenvironment (TME) and dampen the immune response, negatively affecting patient survival. Therefore, targeting TAMs could address the limitations of current cancer treatments. However, drug development in this area remains limited. The leukocyte-associated Ig-like receptor 1 (LAIR1), also called CD305, is prominently expressed on the surface of TAMs. We have uncovered what we believe to be a previously unrecognized immunosuppressive LAIR1/factor XIII A/collagen IV pathway across various cancer types. Inhibition of LAIR1, either through knockout ( Lair1–/– ), antibody blockade (anti-Lair1 antibody), or a chimeric antigen receptor (CAR) design (3-in-1 CAR by combining tumor targeting, T cell trafficking, and remodeling of the immunosuppressive TME in 1 CAR construct) provided an enhanced antitumor response. LAIR1 inhibition enhanced peripheral and intratumoral CD8 memory T cell populations, induced a phenotypic shift of M2-like macrophages toward M1 macrophages, and normalized tumor collagen IV and structural components in the TME, facilitating effective tumor–T cell interactions and tumor suppression. Enhanced antitumor responses were observed when Lair1–/– or anti-Lair1 antibody was used alone or in combination with CAR T cells or when the 3-in-1 CAR T cells were used solely in tumor models resistant to chemotherapy–radiation–programmed cell death protein 1 (PD-1) blockade. These findings position LAIR1 inhibition as a promising strategy for cancer immunotherapies.