J Chem Theory Comput. 2023 Sep 7. doi: 10.1021/acs.jctc.3c00787. Online ahead of print.

ABSTRACT

We evaluated the utility of a variant of the replica exchange method, a replica exchange with hybrid tempering (REHT), for all-atom explicit water biomolecular simulations and compared it with a more traditional replica exchange with the solute tempering (REST) algorithm. As a test system, we selected a 21-mer antimicrobial peptide PGLa binding to an anionic DMPC/DMPG lipid bilayer. Application of REHT revealed the following binding mechanism. Due to the strong hydrophobic moment, the bound PGLa adopts an extensive helical structure. The binding free energy landscape identifies two major bound states, a metastable surface bound state and a dominant inserted state. In both states, positively charged PGLa amino acids maintain electrostatic interactions with anionic phosphate groups by rotating the PGLa helix around its axis. PGLa binding causes an influx of anionic DMPG and an efflux of zwitterionic DMPC lipids from the peptide proximity. PGLa thins the bilayer and disorders the adjacent fatty acid tails. Deep invasion of water wires into the bilayer hydrophobic core is detected in the inserted peptide state. The analysis of charge density distributions indicated that peptide positive charges are nearly compensated for by lipid negative charges and water dipole ordering, whereas ions play no role in peptide binding. Thus, electrostatic interactions are the key energetic factor in binding cationic PGLa to an anionic DMPC/DMPG bilayer. Comparison of REHT and REST shows that due to exclusion of lipids from tempered partition, REST lags behind REHT in peptide equilibration, particularly, with respect to peptide insertion and helix acquisition. As a result, REST struggles to provide accurate details of PGLa binding, although it still qualitatively maps the bimodal binding mechanism. Importantly, REHT not only equilibrates PGLa in the bilayer faster than REST, but also with less computational effort. We conclude that REHT is a preferable choice for studying interfacial biomolecular systems.

PMID:37676235 | DOI:10.1021/acs.jctc.3c00787

J Med Chem. 2023 Sep 7. doi: 10.1021/acs.jmedchem.3c01252. Online ahead of print.

ABSTRACT

Protein or peptide cancer vaccines usually include immune potentiators, so-called adjuvants. However, it remains challenging to identify structurally simple, chemically accessible synthetic molecules that are effective and safe as vaccine adjuvant. Here, we present cholicamideβ (6), a self-assembling small-molecule vaccine adjuvant with an improved toxicity profile and proven efficacy in vivo. We demonstrate that cholicamideβ (6), which is less cytotoxic than its parent compound, forms virus-like particles to potently activate dendritic cells with the concomitant secretion of cytokines. When combined with a peptide antigen, cholicamideβ (6) potentiated the antigen presentation on dendritic cells to induce antigen-specific T cells. As a therapeutic cancer vaccine adjuvant in mice, a mixture of cholicamideβ (6) and a peptide antigen protected mice from the challenges of malignant cancer cells without overt toxicity. Cholicamideβ (6) may offer a translational opportunity as an unprecedented class of small-molecule cancer vaccine adjuvants.

PMID:37676021 | DOI:10.1021/acs.jmedchem.3c01252

mBio. 2023 Sep 7:e0094323. doi: 10.1128/mbio.00943-23. Online ahead of print.

ABSTRACT

Several intracellular pathogens, such as Mycobacterium tuberculosis, damage endomembranes to access the cytosol and subvert innate immune responses. The host counteracts endomembrane damage by recruiting repair machineries that retain the pathogen inside the vacuole. Here, we show that the endoplasmic reticulum (ER)-Golgi protein oxysterol-binding protein (OSBP) and its Dictyostelium discoideum homolog OSBP8 are recruited to the Mycobacterium-containing vacuole (MCV) dependent on the presence of the ESX-1 secretion system, suggesting that their mobilization is associated with membrane damage. Lack of OSBP8 causes a hyperaccumulation of phosphatidylinositol-4-phosphate (PI4P) on the MCV and decreased cell viability. OSBP8-depleted cells had reduced lysosomal and degradative capabilities of their vacuoles that favored mycobacterial growth. In agreement with a potential role of OSBP8 in membrane repair, human macrophages infected with M. tuberculosis recruited OSBP in an ESX-1-dependent manner. These findings identified an ER-dependent repair mechanism for restoring MCVs in which OSBP8 functions to equilibrate PI4P levels on damaged membranes. IMPORTANCE Tuberculosis still remains a global burden and is one of the top infectious diseases from a single pathogen. Mycobacterium tuberculosis, the causative agent, has perfected many ways to replicate and persist within its host. While mycobacteria induce vacuole damage to evade the toxic environment and eventually escape into the cytosol, the host recruits repair machineries to restore the MCV membrane. However, how lipids are delivered for membrane repair is poorly understood. Using advanced fluorescence imaging and volumetric correlative approaches, we demonstrate that this involves the recruitment of the endoplasmic reticulum (ER)-Golgi lipid transfer protein OSBP8 in the Dictyostelium discoideum/Mycobacterium marinum system. Strikingly, depletion of OSBP8 affects lysosomal function accelerating mycobacterial growth. This indicates that an ER-dependent repair pathway constitutes a host defense mechanism against intracellular pathogens such as M. tuberculosis.

PMID:37676004 | DOI:10.1128/mbio.00943-23

mBio. 2023 Sep 7:e0183623. doi: 10.1128/mbio.01836-23. Online ahead of print.

ABSTRACT

The recrudescence of Toxoplasma cysts is the cause of clinical disease in the immunocompromised. Although Toxoplasma has been a useful parasite model for decades because it is relatively easy to genetically modify and culture, attempts to generate and study the recrudescence of tissue cysts have come up short with cell culture-adapted strains generating low numbers of tissue cysts in vivo. Taking advantage of a new ex vivo model of Toxoplasma recrudescence that uses a Type II ME49 strain unadapted to cell culture, we determined the cell biology, gene expression, and host cell dependency that define bradyzoite-cyst reactivation. Bradyzoite infection of fibroblasts and astrocytes produced sequential tachyzoite growth stages with pre-programmed kinetics; thus, an initial fast-growing stage was followed by a slow-growing replicating form. In vivo infections demonstrated that only fast growth tachyzoites, and not parasites post-growth shift, led to successful parasite dissemination to the brain and peripheral organs. In astrocytes, cells that reside in the central nervous system (CNS), bradyzoites initiated an additional recrudescent pathway involving brady-brady replication, which is a pathway not observed in fibroblasts. To investigate the molecular basis of growth and cell-dependent reactivation pathways, single-cell mRNA sequencing was performed on recrudescing parasites, revealing distinct gene signatures of these parasite populations and confirming multifunctionality of the original ex vivo bradyzoite population. This revised model of Toxoplasma recrudescence uncovers previously unknown complexity in the clinically important bradyzoite stage of the parasite, which opens the door to further study these novel developmental features of the Toxoplasma intermediate life cycle. IMPORTANCE The classical depiction of the Toxoplasma lifecycle is bradyzoite excystation conversion to tachyzoites, cell lysis, and immune control, followed by the reestablishment of bradyzoites and cysts. In contrast, we show that tachyzoite growth slows independent of the host immune response at a predictable time point following excystation. Furthermore, we demonstrate a host cell-dependent pathway of continuous amplification of the cyst-forming bradyzoite population. The developmental plasticity of the excysted bradyzoites further underlines the critical role the cyst plays in the flexibility of the lifecycle of this ubiquitous parasite. This revised model of Toxoplasma recrudescence uncovers previously unknown complexity in the clinically important bradyzoite stage of the parasite, which opens the door to further study these novel developmental features of the Toxoplasma intermediate life cycle.

PMID:37675999 | DOI:10.1128/mbio.01836-23

J Integr Plant Biol. 2023 Sep 7. doi: 10.1111/jipb.13564. Online ahead of print.

ABSTRACT

Specialized plant metabolism is a rich resource of compounds for drug discovery. The acylated flavonoid glycoside melitidin is being developed as an anti-cholesterol statin drug candidate, but its biosynthetic route in plants has not yet been fully characterized. Here, we describe the gene discovery and functional characterization of a new flavonoid gene cluster (UDP-glucuronosyltransferases [CgUGTs], 1,2 rhamnosyltransferase [Cg1,2RhaT], acyltransferases [CgATs]) that is responsible for melitidin biosynthesis in pummelo (Citrus grandis (L.) Osbeck). Population variation analysis indicated that the tailoring acyltransferases, specific for bitter substrates, mainly determine the natural abundance of melitidin. Moreover, HMG-CoA reductase enzyme inhibition assays showed that the product from this metabolic gene cluster, melitidin, may be an effective anti-cholesterol statin drug candidate. Co-expression of these clustered genes in Nicotiana benthamiana resulted in the formation of melitidin, demonstrating the potential for metabolic engineering of melitidin in a heterologous plant system. This study establishes a biosynthetic pathway for melitidin, which provides genetic resources for the breeding and genetic improvement of pummelo aimed at fortifying the content of biologically active metabolites. This article is protected by copyright. All rights reserved.

PMID:37675654 | DOI:10.1111/jipb.13564

Plant Physiol. 2023 Sep 7:kiad492. doi: 10.1093/plphys/kiad492. Online ahead of print.

NO ABSTRACT

PMID:37675650 | DOI:10.1093/plphys/kiad492

Br J Clin Pharmacol. 2023 Sep 7. doi: 10.1111/bcp.15902. Online ahead of print.

ABSTRACT

Preclinical models have been the backbone of translational research for more than a century. Rats and mice are critical models in the preliminary stages of drug testing, both for determining efficacy and ruling out potential human-relevant toxicities. Historically, most preclinical pharmacological studies have used young, relatively healthy, inbred male models in highly-controlled environments. In the field of geriatric pharmacology, there is a growing focus on the importance of using more appropriate preclinical models both in the testing of therapeutics commonly used in older populations, and in the evaluation of potential geroprotective drug candidates. Here we provide a commentary on optimising preclinical models of ageing for translation to clinical trials. We will discuss approaches to modeling clinically-relevant contexts such as age, sex, genetic diversity, exposures and environment, as well as measures of clinically-relevant outcomes such as frailty and healthspan. We will identify the strengths and limitations of these approaches and areas for improvement. We will also briefly cover new preclinical models that move beyond rodents. We hope this commentary will be a springboard for larger discussions on optimising preclinical aging models for testing therapeutics.

PMID:37675638 | DOI:10.1111/bcp.15902

Front Microbiol. 2023 Aug 22;14:1238199. doi: 10.3389/fmicb.2023.1238199. eCollection 2023.

ABSTRACT

INTRODUCTION: Imbalances in gut microbes have been implied in many human diseases, including colorectal cancer (CRC), inflammatory bowel disease, type 2 diabetes, obesity, autism, and Alzheimer's disease. Compared with other human diseases, CRC is a gastrointestinal malignancy with high mortality and a high probability of metastasis. However, current studies mainly focus on the prediction of colorectal cancer while neglecting the more serious malignancy of metastatic colorectal cancer (mCRC). In addition, high dimensionality and small samples lead to the complexity of gut microbial data, which increases the difficulty of traditional machine learning models.

METHODS: To address these challenges, we collected and processed 16S rRNA data and calculated abundance data from patients with non-metastatic colorectal cancer (non-mCRC) and mCRC. Different from the traditional health-disease classification strategy, we adopted a novel disease-disease classification strategy and proposed a microbiome-based multi-view convolutional variational information bottleneck (MV-CVIB).

RESULTS: The experimental results show that MV-CVIB can effectively predict mCRC. This model can achieve AUC values above 0.9 compared to other state-of-the-art models. Not only that, MV-CVIB also achieved satisfactory predictive performance on multiple published CRC gut microbiome datasets.

DISCUSSION: Finally, multiple gut microbiota analyses were used to elucidate communities and differences between mCRC and non-mCRC, and the metastatic properties of CRC were assessed by patient age and microbiota expression.

PMID:37675425 | PMC:PMC10477591 | DOI:10.3389/fmicb.2023.1238199

Front Immunol. 2023 Aug 22;14:1211068. doi: 10.3389/fimmu.2023.1211068. eCollection 2023.

ABSTRACT

In recent years, the central role of cell bioenergetics in regulating immune cell function and fate has been recognized, giving rise to the interest in immunometabolism, an area of research focused on the interaction between metabolic regulation and immune function. Thus, early metabolic changes associated with the polarization of macrophages into pro-inflammatory or pro-resolving cells under different stimuli have been characterized. Tumor-associated macrophages are among the most abundant cells in the tumor microenvironment; however, it exists an unmet need to study the effect of chemotherapeutics on macrophage immunometabolism. Here, we use a systems biology approach that integrates transcriptomics and metabolomics to unveil the immunometabolic effects of trabectedin (TRB) and lurbinectedin (LUR), two DNA-binding agents with proven antitumor activity. Our results show that TRB and LUR activate human macrophages toward a pro-inflammatory phenotype by inducing a specific metabolic rewiring program that includes ROS production, changes in the mitochondrial inner membrane potential, increased pentose phosphate pathway, lactate release, tricarboxylic acids (TCA) cycle, serine and methylglyoxal pathways in human macrophages. Glutamine, aspartate, histidine, and proline intracellular levels are also decreased, whereas oxygen consumption is reduced. The observed immunometabolic changes explain additional antitumor activities of these compounds and open new avenues to design therapeutic interventions that specifically target the immunometabolic landscape in the treatment of cancer.

PMID:37675104 | PMC:PMC10479946 | DOI:10.3389/fimmu.2023.1211068

Front Plant Sci. 2023 Aug 22;14:1250878. doi: 10.3389/fpls.2023.1250878. eCollection 2023.

ABSTRACT

Changes in weather patterns with emerging drought risks and rising global temperature are widespread and negatively affect crop growth and productivity. In nature, plants are simultaneously exposed to multiple biotic and abiotic stresses, but most studies focus on individual stress conditions. However, the simultaneous occurrence of different stresses impacts plant growth and development differently than a single stress. Plants sense the different stress combinations in the same or in different tissues, which could induce specific systemic signalling and acclimation responses; impacting different stress-responsive transcripts, protein abundance and modifications, and metabolites. This mini-review focuses on the combination of drought and heat, two abiotic stress conditions that often occur together. Recent omics studies indicate common or independent regulators involved in heat or drought stress responses. Here, we summarize the current research results, highlight gaps in our knowledge, and flag potential future focus areas.

PMID:37674736 | PMC:PMC10478009 | DOI:10.3389/fpls.2023.1250878

J Phys Chem B. 2023 Sep 7;127(35):7556-7557. doi: 10.1021/acs.jpcb.3c05233.

NO ABSTRACT

PMID:37674460 | DOI:10.1021/acs.jpcb.3c05233

J Phys Chem B. 2023 Sep 7;127(35):7553-7555. doi: 10.1021/acs.jpcb.3c05234.

NO ABSTRACT

PMID:37674459 | DOI:10.1021/acs.jpcb.3c05234

Cell Rep Methods. 2023 Aug 30:100578. doi: 10.1016/j.crmeth.2023.100578. Online ahead of print.

ABSTRACT

Regulatory networks containing enhancer-gene edges define cellular states. Multiple efforts have revealed these networks for reference tissues and cell lines by integrating multi-omics data. However, the methods developed cannot be applied for large patient cohorts due to the infeasibility of chromatin immunoprecipitation sequencing (ChIP-seq) for limited biopsy material. We trained machine-learning models using chromatin interaction analysis with paired-end tag sequencing (ChIA-PET) and high-throughput chromosome conformation capture combined with chromatin immunoprecipitation (HiChIP) data that can predict connections using only assay for transposase-accessible chromatin using sequencing (ATAC-seq) and RNA-seq data as input, which can be generated from biopsies. Our method overcomes limitations of correlation-based approaches that cannot distinguish between distinct target genes of given enhancers or between active vs. poised states in different samples, a hallmark of network rewiring in cancer. Application of our model on 371 samples across 22 cancer types revealed 1,780 enhancer-gene connections for 602 cancer genes. Using CRISPR interference (CRISPRi), we validated enhancers predicted to regulate ESR1 in estrogen receptor (ER)+ breast cancer and A1CF in liver hepatocellular carcinoma.

PMID:37673071 | DOI:10.1016/j.crmeth.2023.100578

Biomed Pharmacother. 2023 Sep 4;166:115429. doi: 10.1016/j.biopha.2023.115429. Online ahead of print.

ABSTRACT

Neuroblastoma, a childhood cancer affecting the sympathetic nervous system, continues to challenge the development of potent treatments due to the limited availability of druggable targets for this aggressive illness. Recent investigations have uncovered that phosphoglycerate dehydrogenase (PHGDH), an essential enzyme for de novo serine synthesis, serves as a non-oncogene dependency in high-risk neuroblastoma. In this study, we show that homoharringtonine (HHT) acts as a PHGDH inhibitor, inducing intricate alterations in cellular metabolism, and thus providing an efficient treatment for neuroblastoma. We have experimentally verified the reliance of neuroblastoma on PHGDH and employed molecular docking, thermodynamic evaluations, and X-ray crystallography techniques to determine the bond interactions between HHT and PHGDH. Administering HHT to treat neuroblastoma resulted in effective cell elimination in vitro and tumor reduction in vivo. Metabolite and functional assessments additionally disclosed that HHT treatment suppressed de novo serine synthesis, initiating intricate metabolic reconfiguration and oxidative stress in neuroblastoma. Collectively, these discoveries highlight the potential of targeting PHGDH using HHT as a potent approach for managing high-risk neuroblastoma.

PMID:37673018 | DOI:10.1016/j.biopha.2023.115429

EBioMedicine. 2023 Sep 4;96:104777. doi: 10.1016/j.ebiom.2023.104777. Online ahead of print.

ABSTRACT

BACKGROUND: The cause and symptoms of long COVID are poorly understood. It is challenging to predict whether a given COVID-19 patient will develop long COVID in the future.

METHODS: We used electronic health record (EHR) data from the National COVID Cohort Collaborative to predict the incidence of long COVID. We trained two machine learning (ML) models - logistic regression (LR) and random forest (RF). Features used to train predictors included symptoms and drugs ordered during acute infection, measures of COVID-19 treatment, pre-COVID comorbidities, and demographic information. We assigned the 'long COVID' label to patients diagnosed with the U09.9 ICD10-CM code. The cohorts included patients with (a) EHRs reported from data partners using U09.9 ICD10-CM code and (b) at least one EHR in each feature category. We analysed three cohorts: all patients (n = 2,190,579; diagnosed with long COVID = 17,036), inpatients (149,319; 3,295), and outpatients (2,041,260; 13,741).

FINDINGS: LR and RF models yielded median AUROC of 0.76 and 0.75, respectively. Ablation study revealed that drugs had the highest influence on the prediction task. The SHAP method identified age, gender, cough, fatigue, albuterol, obesity, diabetes, and chronic lung disease as explanatory features. Models trained on data from one N3C partner and tested on data from the other partners had average AUROC of 0.75.

INTERPRETATION: ML-based classification using EHR information from the acute infection period is effective in predicting long COVID. SHAP methods identified important features for prediction. Cross-site analysis demonstrated the generalizability of the proposed methodology.

FUNDING: NCATS U24 TR002306, NCATS UL1 TR003015, Axle Informatics Subcontract: NCATS-P00438-B, NIH/NIDDK/OD, PSR2015-1720GVALE_01, G43C22001320007, and Director, Office of Science, Office of Basic Energy Sciences of the U.S. Department of Energy Contract No. DE-AC02-05CH11231.

PMID:37672869 | DOI:10.1016/j.ebiom.2023.104777

Thromb J. 2023 Sep 6;21(1):91. doi: 10.1186/s12959-023-00536-w.

ABSTRACT

BACKGROUND: Upon cellular injury, damage-associated molecular patterns (DAMPs) are released into the extracellular space and evoke proinflammatory and prothrombotic responses in animal models of sterile inflammation. However, in clinical settings, the dynamics of DAMP levels after trauma and links between DAMPs and trauma-associated coagulopathy remain largely undetermined.

METHODS: Thirty-one patients with severe trauma, who were transferred to Kagoshima City Hospital between June 2018 and December 2019, were consecutively enrolled in this study. Blood samples were taken at the time of delivery, and 6 and 12 h after the injury, and once daily thereafter. The time-dependent changes of coagulation/fibrinolysis markers, including thrombin-antithrombin complex, α2-plasmin inhibitor (α2-PI), plasmin-α2-PI complex, and plasminogen activator inhibitor-1 (PAI-1), and DAMPs, including high mobility group box 1 and histone H3, were analyzed. The relationship between coagulation/fibrinolysis markers, DAMPs, Injury Severity Score, in-hospital death, and amount of blood transfusion were analyzed.

RESULTS: The activation of coagulation/fibrinolysis pathways was evident at the time of delivery. In contrast, PAI-1 levels remained low at the time of delivery, and then were elevated at 6-12 h after traumatic injury. Histone H3 and high mobility group box 1 levels were elevated at admission, and gradually subsided over time. PAI-1 levels at 6 h were associated with serum histone H3 levels at admission. Increased histone H3 levels and plasmin-α2-PI complex levels were associated with in-hospital mortality. α2-PI levels at admission showed the strongest negative correlation with the amount of blood transfusion.

CONCLUSION: The elevation of histone H3 levels and fibrinolysis perturbation are associated with fatal outcomes in patients with traumatic injury. Patients with low α2-PI levels at admission tend to require blood transfusion.

PMID:37674235 | DOI:10.1186/s12959-023-00536-w

Mol Cancer. 2023 Sep 6;22(1):147. doi: 10.1186/s12943-023-01796-w.

ABSTRACT

Gastric adenocarcinoma (GAC) is a lethal disease characterized by genomic and clinical heterogeneity. By integrating 8 previously established genomic signatures for GAC subtypes, we identified 6 clinically and molecularly distinct genomic consensus subtypes (CGSs). CGS1 have the poorest prognosis, very high stem cell characteristics, and high IGF1 expression, but low genomic alterations. CGS2 is enriched with canonical epithelial gene expression. CGS3 and CGS4 have high copy number alterations and low immune reactivity. However, CGS3 and CGS4 differ in that CGS3 has high HER2 activation, while CGS4 has high SALL4 and KRAS activation. CGS5 has the high mutation burden and moderately high immune reactivity that are characteristic of microsatellite instable tumors. Most CGS6 tumors are positive for Epstein Barr virus and show extremely high levels of methylation and high immune reactivity. In a systematic analysis of genomic and proteomic data, we estimated the potential response rate of each consensus subtype to standard and experimental treatments such as radiation therapy, targeted therapy, and immunotherapy. Interestingly, CGS3 was significantly associated with a benefit from chemoradiation therapy owing to its high basal level of ferroptosis. In addition, we also identified potential therapeutic targets for each consensus subtype. Thus, the consensus subtypes produced a robust classification and provide for additional characterizations for subtype-based customized interventions.

PMID:37674200 | DOI:10.1186/s12943-023-01796-w

Nature. 2023 Sep 6. doi: 10.1038/s41586-023-06500-y. Online ahead of print.

ABSTRACT

Translational reprogramming allows organisms to adapt to changing conditions. Upstream start codons (uAUGs), which are prevalently present in mRNAs, have crucial roles in regulating translation by providing alternative translation start sites1-4. However, what determines this selective initiation of translation between conditions remains unclear. Here, by integrating transcriptome-wide translational and structural analyses during pattern-triggered immunity in Arabidopsis, we found that transcripts with immune-induced translation are enriched with upstream open reading frames (uORFs). Without infection, these uORFs are selectively translated owing to hairpins immediately downstream of uAUGs, presumably by slowing and engaging the scanning preinitiation complex. Modelling using deep learning provides unbiased support for these recognizable double-stranded RNA structures downstream of uAUGs (which we term uAUG-ds) being responsible for the selective translation of uAUGs, and allows the prediction and rational design of translating uAUG-ds. We found that uAUG-ds-mediated regulation can be generalized to human cells. Moreover, uAUG-ds-mediated start-codon selection is dynamically regulated. After immune challenge in plants, induced RNA helicases that are homologous to Ded1p in yeast and DDX3X in humans resolve these structures, allowing ribosomes to bypass uAUGs to translate downstream defence proteins. This study shows that mRNA structures dynamically regulate start-codon selection. The prevalence of this RNA structural feature and the conservation of RNA helicases across kingdoms suggest that mRNA structural remodelling is a general feature of translational reprogramming.

PMID:37674078 | DOI:10.1038/s41586-023-06500-y

Sci China Life Sci. 2023 Sep 1. doi: 10.1007/s11427-022-2382-0. Online ahead of print.

NO ABSTRACT

PMID:37673847 | DOI:10.1007/s11427-022-2382-0

J Cancer Res Clin Oncol. 2023 Sep 7. doi: 10.1007/s00432-023-05358-x. Online ahead of print.

ABSTRACT

PURPOSE: Skin cutaneous melanoma (SKCM) is a highly aggressive melanocytic carcinoma whose high heterogeneity and complex etiology make its prognosis difficult to predict. This study aimed to construct a risk subtype typing model for SKCM.

METHODS: The study proposes a deep learning framework combining early fusion feature autoencoder (AE) and late fusion feature AE for risk subtype prediction of SKCM. The deep learning framework integrates mRNA, miRNA, and DNA methylation data of SKCM patients from The Cancer Genome Atlas (TCGA), and clusters the screened multi-omics features associated with survival prognosis to identify risk subtypes. Differential expression analysis and functional enrichment analysis were performed between risk subtypes, while SVM classifiers were constructed between differentially expressed genes (DEGs) obtained by Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression screening and risk subtype labels inferred from multi-omics data, and the predictive robustness of risk subtypes inferred from the risk subtype classification prediction model was validated using two independent datasets.

RESULTS: The deep learning framework that combined early fusion feature AE with late fusion feature AE distinguished the two best risk subtypes compared to the multi-omics integration approach with single strategy AE or PCA. A promising C-index (C-index = 0.748) and a significant difference in survival (log-rank P value = 4.61 × 10-9) were found between the identified risk subtypes. The DEGs with the top significance values together with differentially expressed miRNAs provided the biological interpretation of risk subtypes on SKCM. Finally, the framework was applied to predict risk subtypes in two independent test datasets of SKCM patients, all of which showed good predictive power (C-index > 0.680) and significant survival differences (log-rank P value < 0.01).

CONCLUSION: The SKCM risk subtypes identified by integrating multi-omics data based on deep learning can not only improve the understanding of the molecular mechanisms of SKCM, but also provide clinicians with assistance in treatment decisions.

PMID:37673824 | DOI:10.1007/s00432-023-05358-x