Human population history at the crossroads of East and Southeast Asia since 11,000 years ago.

Past human genetic diversity and migration between southern China and Southeast Asia have not been well characterized, in part due to poor preservation of ancient DNA in hot and humid regions. We sequenced 31 ancient genomes from southern China (Guangxi and Fujian), including two ∼12,000- to 10,000-year-old individuals representing the oldest humans sequenced from southern China. We discovered a deeply diverged East Asian ancestry in the Guangxi region that persisted until at least 6,000 years ago. We found that ∼9,000- to 6,000-year-old Guangxi populations were a mixture of local ancestry, southern ancestry previously sampled in Fujian, and deep Asian ancestry related to Southeast Asian Hòabìnhian hunter-gatherers, showing broad admixture in the region predating the appearance of farming. Historical Guangxi populations dating to ∼1,500 to 500 years ago are closely related to Tai-Kadai and Hmong-Mien speakers. Our results show heavy interactions among three distinct ancestries at the crossroads of East and Southeast Asia.

Recognition of methamphetamine and other amines by trace amine receptor TAAR1.

Trace amine-associated receptor 1 (TAAR1), the founding member of a nine-member family of trace amine receptors, is responsible for recognizing a range of biogenic amines in the brain, including the endogenous ß-phenylethylamine (ß-PEA)1 as well as methamphetamine2, an abused substance that has posed a severe threat to human health and society3. Given its unique physiological role in the brain, TAAR1 is also an emerging target for a range of neurological disorders including schizophrenia, depression and drug addiction2,4,5. Here we report structures of human TAAR1-G-protein complexes bound to methamphetamine and ß-PEA as well as complexes bound to RO5256390, a TAAR1-selective agonist, and SEP-363856, a clinical-stage dual agonist for TAAR1 and serotonin receptor 5-HT1AR (refs. 6,7). Together with systematic mutagenesis and functional studies, the structures reveal the molecular basis of methamphetamine recognition and underlying mechanisms of ligand selectivity and polypharmacology between TAAR1 and other monoamine receptors. We identify a lid-like extracellular loop 2 helix/loop structure and a hydrogen-bonding network in the ligand-binding pockets, which may contribute to the ligand recognition in TAAR1. These findings shed light on the ligand recognition mode and activation mechanism for TAAR1 and should guide the development of next-generation therapeutics for drug addiction and various neurological disorders.

Bump-and-Hole Engineering Identifies Specific Substrates of Glycosyltransferases in Living Cells.

Studying posttranslational modifications classically relies on experimental strategies that oversimplify the complex biosynthetic machineries of living cells. Protein glycosylation contributes to essential biological processes, but correlating glycan structure, underlying protein, and disease-relevant biosynthetic regulation is currently elusive. Here, we engineer living cells to tag glycans with editable chemical functionalities while providing information on biosynthesis, physiological context, and glycan fine structure. We introduce a non-natural substrate biosynthetic pathway and use engineered glycosyltransferases to incorporate chemically tagged sugars into the cell surface glycome of the living cell. We apply the strategy to a particularly redundant yet disease-relevant human glycosyltransferase family, the polypeptide N-acetylgalactosaminyl transferases. This approach bestows a gain-of-chemical-functionality modification on cells, where the products of individual glycosyltransferases can be selectively characterized or manipulated to understand glycan contribution to major physiological processes.

Metabolic remodeling maintains a reducing environment for rapid activation of the yeast DNA replication checkpoint.

Nucleotide metabolism fuels normal DNA replication and is also primarily targeted by the DNA replication checkpoint when replication stalls. To reveal a comprehensive interconnection between genome maintenance and metabolism, we analyzed the metabolomic changes upon replication stress in the budding yeast S. cerevisiae. We found that upon treatment of cells with hydroxyurea, glucose is rapidly diverted to the oxidative pentose phosphate pathway (PPP). This effect is mediated by the AMP-dependent kinase, SNF1, which phosphorylates the transcription factor Mig1, thereby relieving repression of the gene encoding the rate-limiting enzyme of the PPP. Surprisingly, NADPH produced by the PPP is required for efficient recruitment of replication protein A (RPA) to single-stranded DNA, providing the signal for the activation of the Mec1/ATR-Rad53/CHK1 checkpoint signaling kinase cascade. Thus, SNF1, best known as a central energy controller, determines a fast mode of replication checkpoint activation through a redox mechanism. These findings establish that SNF1 provides a hub with direct links to cellular metabolism, redox, and surveillance of DNA replication in eukaryotes.

Latent feature extraction with a prior-based self-attention framework for spatial transcriptomics.

Rapid advances in spatial transcriptomics (ST) have revolutionized the interrogation of spatial heterogeneity and increase the demand for comprehensive methods to effectively characterize spatial domains. As a prerequisite for ST data analysis, spatial domain characterization is a crucial step for downstream analyses and biological implications. Here we propose a prior-based self-attention framework for spatial transcriptomics (PAST), a variational graph convolutional autoencoder for ST, which effectively integrates prior information via a Bayesian neural network, captures spatial patterns via a self-attention mechanism, and enables scalable application via a ripple walk sampler strategy. Through comprehensive experiments on data sets generated by different technologies, we show that PAST can effectively characterize spatial domains and facilitate various downstream analyses, including ST visualization, spatial trajectory inference and pseudotime analysis. Also, we highlight the advantages of PAST for multislice joint embedding and automatic annotation of spatial domains in newly sequenced ST data. Compared with existing methods, PAST is the first ST method that integrates reference data to analyze ST data. We anticipate that PAST will open up new avenues for researchers to decipher ST data with customized reference data, which expands the applicability of ST technology.

Deep reinforcement learning identifies personalized intermittent androgen deprivation therapy for prostate cancer.

The evolution of drug resistance leads to treatment failure and tumor progression. Intermittent androgen deprivation therapy (IADT) helps responsive cancer cells compete with resistant cancer cells in intratumoral competition. However, conventional IADT is population-based, ignoring the heterogeneity of patients and cancer. Additionally, existing IADT relies on pre-determined thresholds of prostate-specific antigen to pause and resume treatment, which is not optimized for individual patients. To address these challenges, we framed a data-driven method in two steps. First, we developed a time-varied, mixed-effect and generative Lotka-Volterra (tM-GLV) model to account for the heterogeneity of the evolution mechanism and the pharmacokinetics of two ADT drugs Cyproterone acetate and Leuprolide acetate for individual patients. Then, we proposed a reinforcement-learning-enabled individualized IADT framework, namely, I$^{2}$ADT, to learn the patient-specific tumor dynamics and derive the optimal drug administration policy. Experiments with clinical trial data demonstrated that the proposed I$^{2}$ADT can significantly prolong the time to progression of prostate cancer patients with reduced cumulative drug dosage. We further validated the efficacy of the proposed methods with a recent pilot clinical trial data. Moreover, the adaptability of I$^{2}$ADT makes it a promising tool for other cancers with the availability of clinical data, where treatment regimens might need to be individualized based on patient characteristics and disease dynamics. Our research elucidates the application of deep reinforcement learning to identify personalized adaptive cancer therapy.

MS-BACL: enhancing metabolic stability prediction through bond graph augmentation and contrastive learning.

MOTIVATION: Accurately predicting molecular metabolic stability is of great significance to drug research and development, ensuring drug safety and effectiveness. Existing deep learning methods, especially graph neural networks, can reveal the molecular structure of drugs and thus efficiently predict the metabolic stability of molecules. However, most of these methods focus on the message passing between adjacent atoms in the molecular graph, ignoring the relationship between bonds. This makes it difficult for these methods to estimate accurate molecular representations, thereby being limited in molecular metabolic stability prediction tasks. RESULTS: We propose the MS-BACL model based on bond graph augmentation technology and contrastive learning strategy, which can efficiently and reliably predict the metabolic stability of molecules. To our knowledge, this is the first time that bond-to-bond relationships in molecular graph structures have been considered in the task of metabolic stability prediction. We build a bond graph based on 'atom-bond-atom', and the model can simultaneously capture the information of atoms and bonds during the message propagation process. This enhances the model's ability to reveal the internal structure of the molecule, thereby improving the structural representation of the molecule. Furthermore, we perform contrastive learning training based on the molecular graph and its bond graph to learn the final molecular representation. Multiple sets of experimental results on public datasets show that the proposed MS-BACL model outperforms the state-of-the-art model. AVAILABILITY AND IMPLEMENTATION: The code and data are publicly available at https://github.com/taowang11/MS.

RETICULON-LIKE PROTEIN B2 is a proviral factor co-opted for the biogenesis of viral replication organelles in plants.

Endomembrane remodeling to form a viral replication complex (VRC) is crucial for a virus to establish infection in a host. Although the composition and function of VRCs have been intensively studied, host factors involved in the assembly of VRCs for plant RNA viruses have not been fully explored. TurboID-based proximity labeling (PL) has emerged as a robust tool for probing molecular interactions in planta. However, few studies have employed the TurboID-based PL technique for investigating plant virus replication. Here, we used Beet black scorch virus (BBSV), an endoplasmic reticulum (ER)-replicating virus, as a model and systematically investigated the composition of BBSV VRCs in Nicotiana benthamiana by fusing the TurboID enzyme to viral replication protein p23. Among the 185 identified p23-proximal proteins, the reticulon family of proteins showed high reproducibility in the mass spectrometry data sets. We focused on RETICULON-LIKE PROTEIN B2 (RTNLB2) and demonstrated its proviral functions in BBSV replication. We showed that RTNLB2 binds to p23, induces ER membrane curvature, and constricts ER tubules to facilitate the assembly of BBSV VRCs. Our comprehensive proximal interactome analysis of BBSV VRCs provides a resource for understanding plant viral replication and offers additional insights into the formation of membrane scaffolds for viral RNA synthesis.

Phytochromes enhance SOS2-mediated PIF1 and PIF3 phosphorylation and degradation to promote Arabidopsis salt tolerance.

Soil salinity is one of the most detrimental abiotic stresses affecting plant survival, and light is a core environmental signal regulating plant growth and responses to abiotic stress. However, how light modulates the plant's response to salt stress remains largely obscure. Here, we show that Arabidopsis (Arabidopsis thaliana) seedlings are more tolerant to salt stress in the light than in the dark, and that the photoreceptors phytochrome A (phyA) and phyB are involved in this tolerance mechanism. We further show that phyA and phyB physically interact with the salt tolerance regulator SALT OVERLY SENSITIVE2 (SOS2) in the cytosol and nucleus, and enhance salt-activated SOS2 kinase activity in the light. Moreover, SOS2 directly interacts with and phosphorylates PHYTOCHROME-INTERACTING FACTORS PIF1 and PIF3 in the nucleus. Accordingly, PIFs act as negative regulators of plant salt tolerance, and SOS2 phosphorylation of PIF1 and PIF3 decreases their stability and relieves their repressive effect on plant salt tolerance in both light and dark conditions. Together, our study demonstrates that photoactivated phyA and phyB promote plant salt tolerance by increasing SOS2-mediated phosphorylation and degradation of PIF1 and PIF3, thus broadening our understanding of how plants adapt to salt stress according to their dynamic light environment.

SALT OVERLY SENSITIVE2 stabilizes phytochrome-interacting factors PIF4 and PIF5 to promote Arabidopsis shade avoidance.

Sun-loving plants trigger the shade avoidance syndrome (SAS) to compete against their neighbors for sunlight. Phytochromes are plant red (R) and far-red (FR) light photoreceptors that play a major role in perceiving the shading signals and triggering SAS. Shade induces a reduction in the level of active phytochrome B (phyB), thus increasing the abundance of PHYTOCHROME-INTERACTING FACTORS (PIFs), a group of growth-promoting transcription factors. However, whether other factors are involved in modulating PIF activity in the shade remains largely obscure. Here, we show that SALT OVERLY SENSITIVE2 (SOS2), a protein kinase essential for salt tolerance, positively regulates SAS in Arabidopsis thaliana. SOS2 directly phosphorylates PIF4 and PIF5 at a serine residue close to their conserved motif for binding to active phyB. This phosphorylation thus decreases their interaction with phyB and posttranslationally promotes PIF4 and PIF5 protein accumulation. Notably, the role of SOS2 in regulating PIF4 and PIF5 protein abundance and SAS is more prominent under salt stress. Moreover, phyA and phyB physically interact with SOS2 and promote SOS2 kinase activity in the light. Collectively, our study uncovers an unexpected role of salt-activated SOS2 in promoting SAS by modulating the phyB-PIF module, providing insight into the coordinated response of plants to salt stress and shade.

The RAC/ROP GTPase activator OsRopGEF10 functions in crown root development by regulating cytokinin signaling in rice.

RAC/Rho of plant (ROP) GTPases are major molecular switches that control diverse signaling cascades for plant growth, development, and defense. Here, we discovered a signaling node that connects RAC/ROPs to cytokinins. Rice (Oryza sativa) plants develop a fibrous root system mainly composed of crown roots. Cytokinin signaling via a phosphorelay system is critical for crown root development. We show that OsRopGEF10, which activates RAC/ROPs, acts upstream of the cytoplasmic-nuclear shuttling phosphotransfer proteins AHPs of the cytokinin signaling pathway to promote crown root development. Mutations of OsRopGEF10 induced hypersensitivity to cytokinin, whereas overexpressing this gene reduced the cytokinin response. Loss of OsRopGEF10 function reduced the expression of the response regulator gene OsRR6, a repressor of cytokinin signaling, and impaired crown root development. Mutations in OsAHP1/2 led to increased crown root production and rescued the crown root defect of Osropgef10. Furthermore, auxin activates the ROP GTPase OsRAC3, which attenuates cytokinin signaling for crown root initiation. Molecular interactions between OsRopGEF10, OsRAC3, and OsAHP1/2 implicate a mechanism whereby OsRopGEF10-activated OsRAC3 recruits OsAHP1/2 to the cortical cytoplasm, sequestering them from their phosphorelay function in the nucleus. Together, our findings uncover the OsRopGEF10-OsRAC3-OsAHP1/2 signaling module, establish a link between RAC/ROPs and cytokinin, and reveal molecular crosstalk between auxin and cytokinin during crown root development.

RhoGDI phosphorylation by PKC promotes its interaction with death receptor p75NTR to gate axon growth and neuron survival.

How receptors juggle their interactions with multiple downstream effectors remains poorly understood. Here we show that the outcome of death receptor p75NTR signaling is determined through competition of effectors for interaction with its intracellular domain, in turn dictated by the nature of the ligand. While NGF induces release of RhoGDI through recruitment of RIP2, thus decreasing RhoA activity in favor of NFkB signaling, MAG induces PKC-mediated phosphorylation of the RhoGDI N-terminus, promoting its interaction with the juxtamembrane domain of p75NTR, disengaging RIP2, and enhancing RhoA activity in detriment of NF-kB. This results in stunted neurite outgrowth and apoptosis in cerebellar granule neurons. If presented simultaneously, MAG prevails over NGF. The NMR solution structure of the complex between the RhoGDI N-terminus and p75NTR juxtamembrane domain reveals previously unknown structures of these proteins and clarifies the mechanism of p75NTR activation. These results show how ligand-directed competition between RIP2 and RhoGDI for p75NTR engagement determine axon growth and neuron survival. Similar principles are likely at work in other receptors engaging multiple effectors and signaling pathways.

Author Correction: The Apostasia genome and the evolution of orchids.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

The water lily genome and the early evolution of flowering plants.

Water lilies belong to the angiosperm order Nymphaeales. Amborellales, Nymphaeales and Austrobaileyales together form the so-called ANA-grade of angiosperms, which are extant representatives of lineages that diverged the earliest from the lineage leading to the extant mesangiosperms1-3. Here we report the 409-megabase genome sequence of the blue-petal water lily (Nymphaea colorata). Our phylogenomic analyses support Amborellales and Nymphaeales as successive sister lineages to all other extant angiosperms. The N. colorata genome and 19 other water lily transcriptomes reveal a Nymphaealean whole-genome duplication event, which is shared by Nymphaeaceae and possibly Cabombaceae. Among the genes retained from this whole-genome duplication are homologues of genes that regulate flowering transition and flower development. The broad expression of homologues of floral ABCE genes in N. colorata might support a similarly broadly active ancestral ABCE model of floral organ determination in early angiosperms. Water lilies have evolved attractive floral scents and colours, which are features shared with mesangiosperms, and we identified their putative biosynthetic genes in N. colorata. The chemical compounds and biosynthetic genes behind floral scents suggest that they have evolved in parallel to those in mesangiosperms. Because of its unique phylogenetic position, the N. colorata genome sheds light on the early evolution of angiosperms.

Reconstitution of phytochrome A-mediated light modulation of the ABA signaling pathways in yeast.

Abscisic acid (ABA), a classical plant hormone, plays an essential role in plant adaptation to environmental stresses. The ABA signaling mechanisms have been extensively investigated, and it was shown that the PYR1 (PYRABACTIN RESISTANCE1)/PYL (PYR1-LIKE)/RCAR (REGULATORY COMPONENT OF ABA RECEPTOR) ABA receptors, the PP2C coreceptors, and the SnRK2 protein kinases constitute the core ABA signaling module responsible for ABA perception and initiation of downstream responses. We recently showed that ABA signaling is modulated by light signals, but the underlying molecular mechanisms remain largely obscure. In this study, we established a system in yeast cells that was not only successful in reconstituting a complete ABA signaling pathway, from hormone perception to ABA-responsive gene expression, but also suitable for functionally characterizing the regulatory roles of additional factors of ABA signaling. Using this system, we analyzed the roles of several light signaling components, including the red and far-red light photoreceptors phytochrome A (phyA) and phyB, and the photomorphogenic central repressor COP1, in the regulation of ABA signaling. Our results showed that both phyA and phyB negatively regulated ABA signaling, whereas COP1 positively regulated ABA signaling in yeast cells. Further analyses showed that photoactivated phyA interacted with the ABA coreceptors ABI1 and ABI2 to decrease their interactions with the ABA receptor PYR1. Together, data from our reconstituted yeast ABA signaling system provide evidence that photoactivated photoreceptors attenuate ABA signaling by directly interacting with the key components of the core ABA signaling module, thus conferring enhanced ABA tolerance to light-grown plants.

LIM domain only 7 negatively controls nonalcoholic steatohepatitis in the setting of hyperlipidemia.

BACKGROUND AND AIMS: Hyperlipidemia has been extensively recognized as a high-risk factor for NASH; however, clinical susceptibility to NASH is highly heterogeneous. The key controller(s) of NASH susceptibility in patients with hyperlipidemia has not yet been elucidated. Here, we aimed to reveal the key regulators of NASH in patients with hyperlipidemia and to explore its role and underlying mechanisms. APPROACH AND RESULTS: To identify the predominant suppressors of NASH in the setting of hyperlipidemia, we collected liver biopsy samples from patients with hyperlipidemia, with or without NASH, and performed RNA-sequencing analysis. Notably, decreased Lineage specific Interacting Motif domain only 7 (LMO7) expression robustly correlated with the occurrence and severity of NASH. Although overexpression of LMO7 effectively blocked hepatic lipid accumulation and inflammation, LMO7 deficiency in hepatocytes greatly exacerbated diet-induced NASH progression. Mechanistically, lysine 48 (K48)-linked ubiquitin-mediated proteasomal degradation of tripartite motif-containing 47 (TRIM47) and subsequent inactivation of the c-Jun N-terminal kinase (JNK)/p38 mitogen-activated protein kinase (MAPK) cascade are required for the protective function of LMO7 in NASH. CONCLUSIONS: These findings provide proof-of-concept evidence supporting LMO7 as a robust suppressor of NASH in the context of hyperlipidemia, indicating that targeting the LMO7-TRIM47 axis is a promising therapeutic strategy for NASH.

Dysregulation of Gut Microbiota-Derived Neuromodulatory Amino Acid Metabolism in Human Immunodeficiency Virus-Associated Neurocognitive Disorder: An Integrative Metagenomic and Metabolomic Analysis.

OBJECTIVE: Although accumulating evidence implicating altered gut microbiota in human immunodeficiency virus (HIV) infection and neurodegenerative disorders; however, the association between dysbiosis of the gut microbiota and metabolites in the pathogenesis of HIV-associated neurocognitive disorder (HAND) remains unclear. METHODS: Fecal and plasma samples were obtained from 3 cohorts (HAND, HIV-non-HAND, and healthy controls), metagenomic analysis and metabolomic profiling were performed to investigate alterations in the gut microbial composition and circulating metabolites in HAND. RESULTS: The gut microbiota of people living with HIV (PLWH) had an increased relative abundance of Prevotella and a decreased relative abundance of Bacteroides. In contrast, Prevotella and Megamonas were substantially decreased, and Bacteroides and Phocaeicola were increased in HAND patients. Moreover, untargeted metabolomics identified several neurotransmitters and certain amino acids associated with neuromodulation, and the differential metabolic pathways of amino acids associated with neurocognition were depleted in HAND patients. Notably, most neuromodulatory metabolites are associated with an altered abundance of specific gut bacteria. INTERPRETATION: Our findings provide new insights into the intricate interplay between the gut and microbiome-brain axis in the pathogenesis of HAND, highlighting the potential for developing novel therapeutic strategies that specifically target the gut microbiota. ANN NEUROL 2024;96:306-320.

COP1 positively regulates ABA signaling during Arabidopsis seedling growth in darkness by mediating ABA-induced ABI5 accumulation.

CONSTITUTIVELY PHOTOMORPHOGENIC1 (COP1), a well-characterized E3 ubiquitin ligase, is a central repressor of seedling photomorphogenic development in darkness. However, whether COP1 is involved in modulating abscisic acid (ABA) signaling in darkness remains largely obscure. Here, we report that COP1 is a positive regulator of ABA signaling during Arabidopsis seedling growth in the dark. COP1 mediates ABA-induced accumulation of ABI5, a transcription factor playing a key role in ABA signaling, through transcriptional and post-translational regulatory mechanisms. We further show that COP1 physically interacts with ABA-hypersensitive DCAF1 (ABD1), a substrate receptor of the CUL4-DDB1 E3 ligase targeting ABI5 for degradation. Accordingly, COP1 directly ubiquitinates ABD1 in vitro, and negatively regulates ABD1 protein abundance in vivo in the dark but not in the light. Therefore, COP1 promotes ABI5 protein stability post-translationally in darkness by destabilizing ABD1 in response to ABA. Interestingly, we reveal that ABA induces the nuclear accumulation of COP1 in darkness, thus enhancing its activity in propagating the ABA signal. Together, our study uncovers that COP1 modulates ABA signaling during seedling growth in darkness by mediating ABA-induced ABI5 accumulation, demonstrating that plants adjust their ABA signaling mechanisms according to their light environment.

Hydrogen Sulfide Modulates Endothelial-Mesenchymal Transition in Heart Failure.

BACKGROUND: Hydrogen sulfide is a critical endogenous signaling molecule that exerts protective effects in the setting of heart failure. Cystathionine γ-lyase (CSE), 1 of 3 hydrogen-sulfide-producing enzyme, is predominantly localized in the vascular endothelium. The interaction between the endothelial CSE-hydrogen sulfide axis and endothelial-mesenchymal transition, an important pathological process contributing to the formation of fibrosis, has yet to be investigated. METHODS: Endothelial-cell-specific CSE knockout and Endothelial cell-CSE overexpressing mice were subjected to transverse aortic constriction to induce heart failure with reduced ejection fraction. Cardiac function, vascular reactivity, and treadmill exercise capacity were measured to determine the severity of heart failure. Histological and gene expression analyses were performed to investigate changes in cardiac fibrosis and the activation of endothelial-mesenchymal transition. RESULTS: Endothelial-cell-specific CSE knockout mice exhibited increased endothelial-mesenchymal transition and reduced nitric oxide bioavailability in the myocardium, which was associated with increased cardiac fibrosis, impaired cardiac and vascular function, and worsened exercise performance. In contrast, genetic overexpression of CSE in endothelial cells led to increased myocardial nitric oxide, decreased endothelial-mesenchymal transition and cardiac fibrosis, preserved cardiac and endothelial function, and improved exercise capacity. CONCLUSIONS: Our data demonstrate that endothelial CSE modulates endothelial-mesenchymal transition and ameliorate the severity of pressure-overload-induced heart failure, in part, through nitric oxide-related mechanisms. These data further suggest that endothelium-derived hydrogen sulfide is a potential therapeutic for the treatment of heart failure with reduced ejection fraction.

mRNA-CLA: An interpretable deep learning approach for predicting mRNA subcellular localization.

Messenger RNA (mRNA) is vital for post-transcriptional gene regulation, acting as the direct template for protein synthesis. However, the methods available for predicting mRNA subcellular localization need to be improved and enhanced. Notably, few existing algorithms can annotate mRNA sequences with multiple localizations. In this work, we propose the mRNA-CLA, an innovative multi-label subcellular localization prediction framework for mRNA, leveraging a deep learning approach with a multi-head self-attention mechanism. The framework employs a multi-scale convolutional layer to extract sequence features across different regions and uses a self-attention mechanism explicitly designed for each sequence. Paired with Position Weight Matrices (PWMs) derived from the convolutional neural network layers, our model offers interpretability in the analysis. In particular, we perform a base-level analysis of mRNA sequences from diverse subcellular localizations to determine the nucleotide specificity corresponding to each site. Our evaluations demonstrate that the mRNA-CLA model substantially outperforms existing methods and tools.