Search bioRxiv⌕ Search

Biology subjects

Wu, T.

Publications and source records attributed to Wu, T..

At least 91 records · Page 5Linked to original sources

Glioblastoma Treatment by Systemic Actinium-225 alpha-particle Dendrimer-radioconjugates is Improved by Chemotherapy

RATIONALEThe poor prognosis of glioblastoma is largely due to drug resistance and tumor location that, together, make it difficult to treat aggressively without affecting the rest of the brain. METHODOLOGYHigh-energy, short-range (40-80{micro}m) dendrimer-delivered -particles could address both challenges, because (1) they cause complex, highly cytotoxic double-strand DNA breaks, and (2) irradiation of the neighboring brain is minimal, since dendrimers selectively delivers them to tumors. Since cancer cells that are not directly hit by -particles will likely not be killed, the patterns of tumor irradiation affect efficacy. Systemically injected dendrimers extensively accumulate in glioblastomas, where they are taken up by tumor associated macrophages (TAMs), which tend to infiltrate tumors. We hypothesized that dendrimers labeled with -particle emitters, when being carried by TAMs, could more evenly irradiate glioblastomas, improving survival. In this study, the efficacy of dendrimers radiolabeled with the -particle emitter actinium-225 (dendrimer-radioconjugates) was evaluated when administered alone and/or after temozolomide, in a syngeneic immune-competent orthotopic GL261-C57BL/6 mouse model. RESULTSSystemically-administered dendrimer-radioconjugates, at activities that did not result in long-term toxicities, prolonged survival of mice with orthotopic GL261 tumors, compared to standard-of-care temozolomide (39 vs 31 days mean survival, p=0.0061) and non-treated animals (30 days, p=0.0009). Importantly, injection of temozolomide 24 hours before administration of dendrimer-radioconjugates further improved survival remarkably (44 days). This improvement in efficacy was attributed to: (1) the significant increase (by 33%) in tumor absorbed doses delivered by dendrimer-radioconjugates when injected after chemotherapy, without altering normal organ dosimetry, while sparing the tumor-surrounding healthy brain; (2) the potentially deeper tumor penetration of dendrimer-radioconjugates, suggested by the enhancement of dendrimer penetration within GL261-spheroids, employed as model tumor-avascular regions and/or TAM-free regions; and/or (3) the formation of a more lethal cocktail when both modalities acted on same cancer cells, that was correlated with increased levels of dendrimer-radioconjugates associating with GL261 cells in vitro and with greater incidences of karyomegaly in vivo. CONCLUSIONSThis study demonstrates the potential of a brain tumor targeted systemic actinium-225 radiopharmaceutical therapy that inhibits growth of glioblastoma cells and prolongs survival of mice with orthotopic brain tumors, further improved by standard-of-care temozolomide, without notable toxicities.

bioengineering↗

Systems genetic dissection of Alzheimer's disease brain gene expression networks

In Alzheimers disease (AD), changes in the brain transcriptome are hypothesized to mediate the impact of neuropathology on cognition. Gene expression profiling from postmortem brain tissue is a promising approach to identify causal pathways; however, there are challenges to definitively resolve the upstream pathologic triggers along with the downstream consequences for AD clinical manifestations. We have functionally dissected 30 AD-associated gene coexpression modules using a cross-species strategy in fruit fly (Drosophila melanogaster) models. Integrating longitudinal RNA-sequencing and behavioral phenotyping, we interrogated the unique and shared transcriptional responses to amyloid beta (A{beta}) plaques, tau neurofibrillary tangles, and/or aging, along with potential links to progressive neuronal dysfunction. Our results highlight hundreds of conserved, differentially expressed genes mapping to human AD regulatory networks. To confirm causal modules and pinpoint AD network drivers, we performed systematic in vivo genetic manipulations of 357 conserved, prioritized targets, identifying 141 modifiers of A{beta}- and/or tau-induced neurodegeneration. We discover an up-regulated network that is significantly enriched for both AD risk variants and markers of immunity / inflammation, and which promotes A{beta} and tau-mediated neurodegeneration based on fly genetic manipulations in neurons. By contrast, a synaptic regulatory network is strongly downregulated in human brains with AD and is enriched for loss-of-function suppressors of A{beta}/tau in Drosophila. Additional experiments suggest that this human brain transcriptional module may respond to and modulate A{beta}-induced glutamatergic hyperactivation injury. In sum, our cross-species, systems genetic approach establishes a putative causal chain linking AD pathology, large-scale gene expression perturbations, and ultimately, neurodegeneration.

systems biology↗

Identification and characterization of ugpE required for the full virulence of Streptococcus suis

Streptococcus suis (S. suis) is an emerging zoonotic pathogen that threatens both animal and human health worldwide. UgpE is a protein subunit of the Ugp (uptake of glycerol phosphate) transporter system that is involved in glycerophospholipid synthesis in bacterial membranes. In this study, an ugpE deletion mutant was constructed and the effects of ugpE deletion on cell morphology, biofilm formation, and virulence were investigated. Deletion of ugpE did not affect bacterial growth but impaired cell chain formation and capsular synthesis by downregulating the mRNA levels of the capsular regulon genes cps-2B, cps-2C, and cps-2S. Deletion of ugpE also led to decreased tolerance to heat, oxidative, and acid-base stress. Crystal violet staining and scanning electron microscopy demonstrated ugpE negatively regulated biofilm formation in liquid culture and the rdar biofilm morphotype on agar plates. Moreover, ugpE deletion not only reduced hemolysin activity, survival in whole human blood, and anti-phagocytosis ability against porcine alveolar macrophages (PAM) but also enhanced bacterial adhesion and invasion of human cerebral microvascular endothelial cells (hCMEC/D3) by upregulating the expression of multiple genes associated with cell adhesion. In a mouse infection model, ugpE deletion significantly attenuated virulence and lowered the number of viable bacteria in the blood and major organs, as well as distribution of macrophages. In conclusion, this study identified UgpE as a novel virulence factor that plays a pivotal role in the regulation of virulence and biofilm formation of S. suis.

microbiology↗

Somatic Transthyretin-Related Proteins in C. elegans Govern Reproductive Longevity by Sustaining Sperm Integrity and Timely Ovulation

The decline in reproductive capability during adult life is critical for health, but its mechanism is elusive. We systematically analyzed the developmental role of an expanded TTR family of proteins, structurally analogous to mammalian thyroid hormone-transporting Transthyretin, and identified three paralogous proteins, TTR-15, TTR-16, and TTR-17, differentially expressed in somatic cells of the gonads and secreted around gametes in C. elegans. Simultaneous inactivation of TTR-15, TTR-16, and TTR-17 leads to a rapid reduction in reproductive capacity in middle age. While oocyte and sperm production remain unaffected in the triple mutants, late-onset infertility results from stalled ovulation. Mechanistically, the absence of TTR-15, TTR-16, and TTR-17 causes sperm to prematurely deplete the cytoplasmic pool of major sperm protein (MSP), released via non-conventional vesicle budding as a signal for ovulation. We propose that the somatic gonads play a central role in maintaining sperm integrity post-production and determining the duration of the reproductive age. HighlightsO_LISystematic analysis of TTR family proteins reveals diverse expression and critical functions. C_LIO_LITTR-15, TTR-16, and TTR-17 are secreted around gametes in C. elegans. C_LIO_LITTR-15/16/17 triple KO exhibits middle-age onset infertility due to stalled ovulation. C_LIO_LIMSP, a signal for ovulation, is prematurely depleted from sperm in the absence of TTR-15/16/17. C_LI

developmental biology↗

53BP1 deficiency leads to hyperrecombination using break-induced replication (BIR)

Break-induced replication (BIR) is mutagenic, and thus its use requires tight regulation, yet the underlying mechanisms remain elusive. Here we uncover an important role of 53BP1 in suppressing BIR after end resection at double strand breaks (DSBs), distinct from its end protection activity, providing insight into the mechanisms governing BIR regulation and DSB repair pathway selection. We demonstrate that loss of 53BP1 induces BIR-like hyperrecombination, in a manner dependent on Pol-primase-mediated end fill-in DNA synthesis on single-stranded DNA (ssDNA) overhangs at DSBs, leading to PCNA ubiquitination and PIF1 recruitment to activate BIR. On broken replication forks, where BIR is required for repairing single-ended DSBs (seDSBs), SMARCAD1 displaces 53BP1 to facilitate the localization of ubiquitinated PCNA and PIF1 to DSBs for BIR activation. Hyper BIR associated with 53BP1 deficiency manifests template switching and large deletions, underscoring another aspect of 53BP1 in suppressing genome instability. The synthetic lethal interaction between the 53BP1 and BIR pathways provides opportunities for targeted cancer treatment.

molecular biology↗

CRISPR-CLEAR: Nucleotide-Resolution Mapping of Regulatory Elements via Allelic Readout of Tiled Base Editing

CRISPR tiling screens have advanced the identification and characterization of regulatory sequences but are limited by low resolution arising from the indirect readout of editing via guide RNA sequencing. This study introduces CRISPR-CLEAR, an end-to-end experimental assay and computational pipeline, which leverages targeted sequencing of CRISPR-introduced alleles at the endogenous target locus following dense base-editing mutagenesis. This approach enables the dissection of regulatory elements at nucleotide resolution, facilitating a direct assessment of genotype-phenotype effects.

genomics↗

Therapeutically targeting the classical complement pathway with antisense oligonucleotides in Alzheimer's disease

The complement classical pathway (CP) is a key mediator of synapse loss and neurodegeneration in mouse models of Alzheimers (AD) and other neurodegenerative diseases. We analyzed human brain proteomics and found consistent elevations of all CP proteins, but not other complement pathways, in AD patient brains. We performed human genetics analysis that identified a rare variant in the C1S gene within the Finnish population that is associated with AD and we found that a common AD-associated C1S variant correlates with increased C1S protein levels. A targeted assay detected elevated C1S activation in AD patient CSF. Given this specific implication of the CP in AD, we next evaluated the therapeutic approach of targeting the CP in the brain using antisense oligonucleotides (ASOs). To identify promising CP targets for knockdown using ASOs we first tested for rescue of synapse loss in an AD mouse model using heterozygous and homozygous complement knockout mice and examined the relative brain expression levels of different CP genes. Based on these experiments we prioritized C1r, C1s and C4 as promising targets for therapeutic knockdown using ASOs. We then screened for ASOs for each target, evaluating in vitro and in vivo knockdown and toxicity, and identified optimal ASOs targeting C1r, C1s and C4. Experiments with AD model mice demonstrated significant rescue of synapse loss following treatment with C1r, C1s or C4 ASOs. Overall, our findings provide proof of concept for using nucleic acid-based medicine to target the CP in AD and demonstrate the translational potential of this approach.

neuroscience↗

Differential interactions determine anisotropies at interfaces of RNA-based biomolecular condensates

Biomolecular condensates form via macromolecular phase separation, giving rise to coexisting phases delineated by interfaces. Here, we characterize the structures of interfaces formed by phase separation driven by heterotypic interactions in ternary mixtures of two types of RNA molecules and polyethylene glycol. We find that purine-rich RNAs are scaffolds that drive phase separation via strong heterotypic interactions. Conversely, pyrimidine-rich RNA molecules are defined by weaker heterotypic interactions. They function as adsorbents that accumulate at and wet the interfaces of coexisting phases formed by phase separation of scaffolds. Our computations predict that scaffolds and adsorbents have different non-random orientational preferences at interfaces. We tested these predictions using single-molecule super resolution imaging that tracks the motions of fluorogenic probes that are bound to RNA molecules. Motions parallel to the interface were found to be faster than motions perpendicular to the interface. These findings support previous predictions regarding anisotropies of motions at interfaces.

biophysics↗

Defensive symbiont genotype distributions are linked to parasitoid attack networks

Facultative symbionts are widespread in arthropods and can provide important services such as protection from natural enemies. Yet what shapes associations with defensive symbionts in nature remains unclear. Two hypotheses suggest that either interactions with antagonists, or host plants, may explain the prevalence of symbionts through shared selective pressures and routes of horizontal transmission. Here we investigate the factors driving similarities in the Hamiltonella defensa symbiosis shared among host species within field collected aphid communities. We show that, Hamiltonellas genotype distribution strongly aligns with sharing the same parasitoids, rather than host plants, highlighting parasitoids as a key selective agent shaping the symbiosis across host species. Our data indicates parasitoid host-specificity drives the prevalence of specific aphid-Hamiltonella associations, suggesting defensive symbioses are maintained by the selective pressure imposed by dominant parasitoid species. These findings underscore the importance of interactions with natural enemies in explaining patterns of defensive symbiosis in nature.

ecology↗

Break-induced replication is activated to repair R-loop-associated double-strand breaks in SETX-deficient cells

The primary role of break-induced replication (BIR) is to repair single-ended double-strand breaks (seDSBs) generated at broken replication forks and eroding telomeres. In this study, we demonstrated that when senataxin (SETX), an RNA/DNA helicase, is defective, hyper-recombination using the BIR mechanism is induced at R-loops/hybrids-accumulated double-ended DSBs (deDSBs), uncovering a role for BIR in repair of R-loops/hybrids-associated deDSBs. Intriguingly, the loss of SETX not only triggers non-canonical hyper-end resection requiring RAD52 and XPF, but also stalls Pol-primase-initiated end-fill DNA synthesis due to the accumulation of RNA/DNA hybrids on single-strand DNA (ssDNA) overhangs at deDSBs. This conflict between fill-in DNA synthesis and accumulated hybrids induces PCNA ubiquitination and PIF1 loading, thereby initiating the BIR mechanism at deDSBs. Hyper-resection further enhances PCNA ubiquitination and PIF1 loading, driving BIR-mediated hyper-recombination. Moreover, SETX is synthetic lethal with PIF1, RAD52, and XPF, offering new strategies for targeted treatment of SETX-deficient tumors.

molecular biology↗

NRF2-dependent regulation of the prostacyclin receptor PTGIR drives CD8 T cell exhaustion

The progressive decline of CD8 T cell effector function--also known as terminal exhaustion--is a major contributor to immune evasion in cancer. Yet, the molecular mechanisms that drive CD8 T cell dysfunction remain poorly understood. Here, we report that the Kelch-like ECH-associated protein 1 (KEAP1)-Nuclear factor erythroid 2-related factor 2 (NRF2) signaling axis, which mediates cellular adaptations to oxidative stress, directly regulates CD8 T cell exhaustion. Transcriptional profiling of dysfunctional CD8 T cells from chronic infection and cancer reveals enrichment of NRF2 activity in terminally exhausted (Texterm) CD8 T cells. Increasing NRF2 activity in CD8 T cells (via conditional deletion of KEAP1) promotes increased glutathione production and antioxidant defense yet accelerates the development of terminally exhausted (PD-1+TIM-3+) CD8 T cells in response to chronic infection or tumor challenge. Mechanistically, we identify PTGIR, a receptor for the circulating eicosanoid prostacyclin, as an NRF2-regulated protein that promotes CD8 T cell dysfunction. Silencing PTGIR expression restores the anti-tumor function of KEAP1-deficient T cells. Moreover, lowering PTGIR expression in CD8 T cells both reduces terminal exhaustion and enhances T cell effector responses (i.e. IFN-{gamma} and granzyme production) to chronic infection and cancer. Together, these results establish the KEAP1-NRF2 axis as a metabolic sensor linking oxidative stress to CD8 T cell dysfunction and identify the prostacyclin receptor PTGIR as an NRF2-regulated immune checkpoint that regulates CD8 T cell fate decisions between effector and exhausted states. One Sentence SummaryThe KEAP1-NRF2 pathway is hyperactivated in terminally exhausted CD8 T cells and drives T cell dysfunction via transcriptional regulation of the prostacyclin receptor, Ptgir.

immunology↗

MEK signaling represents a viable therapeutic vulnerability of KRAS-driven somatic brain arteriovenous malformations

Brain arteriovenous malformations (bAVMs) are direct connections between arteries and veins that remodel into a complex nidus susceptible to rupture and hemorrhage. Most sporadic bAVMs feature somatic activating mutations within KRAS, and endothelial-specific expression of the constitutively active variant KRASG12D models sporadic bAVM in mice. By leveraging 3D-based micro-CT imaging, we demonstrate that KRASG12D-driven bAVMs arise in stereotypical anatomical locations within the murine brain, which coincide with high endogenous Kras expression. We extend these analyses to show that a distinct variant, KRASG12C, also generates bAVMs in predictable locations. Analysis of 15,000 human patients revealed that, similar to murine models, bAVMs preferentially occur in distinct regions of the adult brain. Furthermore, bAVM location correlates with hemorrhagic frequency. Quantification of 3D imaging revealed that G12D and G12C alter vessel density, tortuosity, and diameter within the mouse brain. Notably, aged G12D mice feature increased lethality, as well as impaired cognition and motor function. Critically, we show that pharmacological blockade of the downstream kinase, MEK, after lesion formation ameliorates KRASG12D-driven changes in the murine cerebrovasculature and may also impede bAVM progression in human pediatric patients. Collectively, these data show that distinct KRAS variants drive bAVMs in similar patterns and suggest MEK inhibition represents a non-surgical alternative therapy for sporadic bAVM.

molecular biology↗

Histone deacetylase 1 and 2 play essential roles in alveolar macrophage homeostasis

Alveolar macrophages (AMs) are the primary lung-resident macrophages that play a pivotal role in pathogen clearance and surfactant homeostasis. Although the molecular regulation of AM development at the transcriptional level has been more widely studied, the epigenetic mechanisms that maintain AM homeostasis remain incompletely understood. Here, we demonstrate that histone deacetylases HDAC1 and HDAC2 play essential roles in preserving AM integrity. Dual deletion of HDAC1 and HDAC2 resulted in severe defect in AM development, leading to pulmonary alveolar proteinosis (PAP) and increased susceptibility to influenza infection in mice. Mechanistically, HDAC1 and HDAC2 act through their deacetylase activity to suppress the expression of pro-apoptotic protein Bim, thereby enhancing cell survival. Furthermore, HDAC1 and HDAC2 form a functional complex with the histone demethylase LSD1 (KDM1A) to co-regulate a transcriptional program governing AM differentiation. Our findings identify HDAC1 and HDAC2 as central epigenetic regulators essential for alveolar macrophage homeostasis.

immunology↗

A Conserved Inhibitory Interdomain Interaction Regulates DNA-binding Activities of Hybrid Two-component Systems in Bacteroides

Hybrid two-component systems (HTCSs) comprise a major class of transcription regulators of polysaccharide utilization genes in Bacteroides. Distinct from classical two-component systems in which signal transduction is carried out by intermolecular phosphotransfer between a histidine kinase (HK) and a cognate response regulator (RR), HTCSs contain the membrane sensor HK and the RR transcriptional regulator within a single polypeptide chain. Tethering the DNA-binding domain (DBD) of the RR with the dimeric HK domain in an HTCS could potentially promote dimerization of the DBDs and would thus require a mechanism to suppress DNA-binding activity in the absence of stimulus. Analysis of phosphorylation and DNA-binding activities of several HTCSs from Bacteroides thetaiotaomicron revealed a DBD-suppression mechanism in which an inhibitory interaction between the DBD and the phosphoryl group-accepting receiver domain (REC) decreases autophosphorylation rates of HTCS-RECs as well as represses DNA-binding activities in the absence of phosphorylation. Sequence analyses and structure predictions identified a highly conserved sequence motif correlated with a conserved inhibitory domain arrangement of REC and DBD. Presence of the motif, as in most HTCSs, or its absence, in a small subset of HTCSs, is likely predictive of two distinct regulatory mechanisms evolved for different glycans. Substitutions within the conserved motif relieve the inhibitory interaction and result in elevated DNA-binding activities in the absence of phosphorylation. Our data suggest a fundamental regulatory mechanism shared by most HTCSs to suppress DBD activities using a conserved inhibitory interdomain arrangement to overcome the challenge of the fused HK and RR components. ImportanceDifferent dietary and host-derived complex carbohydrates shape the gut microbial community and impact human health. In Bacteroides, the prevalent gut bacteria genus, utilization of these diverse carbohydrates relies on different gene clusters that are under sophisticated control by various signaling systems, including the hybrid two-component systems (HTCSs). We have uncovered a highly conserved regulatory mechanism in which the output DNA-binding activity of HTCSs is suppressed by interdomain interactions in the absence of stimulating phosphorylation. A consensus amino acid motif is found to correlate with the inhibitory interaction surface while deviations from the consensus can lead to constitutive activation. Understanding of such conserved HTCS features will be important to make regulatory predictions for individual systems as well as to engineer novel systems with substitutions in the consensus to explore the glycan regulation landscape in Bacteroides.

biochemistry↗

Quantitative analysis of cis-regulatory elements in transcription with KAS-ATAC-seq

Cis-regulatory elements (CREs) are pivotal in orchestrating gene expression throughout diverse biological systems. Accurate identification and in-depth characterization of functional CREs are crucial for decoding gene regulation network and dynamics during cellular processes. In this study, we developed Kethoxal-Assisted Single-stranded DNA Assay for Transposase-Accessible Chromatin with Sequencing (KAS-ATAC-seq) to provide quantitative insights into transcriptional activity of CREs. A main advantage of KAS-ATAC-seq lies in its precise measurement of ssDNA levels within both proximal and distal ATAC-seq peaks, enabling the identification of transcriptional regulatory sequences in genomes. This feature is particularly adept at defining Single-Stranded Transcribing Enhancers (SSTEs). SSTEs are highly enriched with nascent RNA transcription and specific transcription factors (TFs) binding sites that determine cellular identity. Moreover, KAS-ATAC-seq provides a detailed characterization and functional implications of various SSTE subtypes; KAS-ATAC-seq signals on SSTEs exhibit more robust correlation with enhancer activities when compared with ATAC-seq data and active histone mark profiles. Our analysis of promoters and SSTEs during mouse neural differentiation demonstrates that KAS-ATAC-seq can effectively identify immediate-early activated CREs in response to retinoic acid (RA) treatment. We further discovered that ETS TFs and YY1 are critical in initiating early neural differentiation from mESCs to NPCs. Our findings indicate that KAS-ATAC-seq provides more precise annotation of functional CREs in transcription. Future applications of KAS-ATAC-seq would help elucidate the intricate dynamics of gene regulation in diverse biological processes and biomedical applications.

genomics↗

LXA4 promotes the browning of white adipose through miR-133a-3p/SIRT1 pathway

Lipoxin A4 (LXA4) promotes the browning of white adipose and energy consumption. The specific mechanism of which involved in white adipose browning is less clear. A high-fat diet (HFD) mouse model was constructed. Different groups of mice were treated with LXA4 accordingly. The body weight of mouse, subcutaneous and visceral fat, and food intake were recorded. The effect of LXA4 was examined by observing changes in pathology, serum insulin and lipid accumulation indices. The effects of LXA4/miR-133a-3p/Sirtuin1 on lipid droplet formation, fat browning-related genes, and the insulin receptor-AKT pathway in cells were examined after induction of adipocyte differentiation in 3T3-L1 precursors. At the cellular level, LXA4 promoted lipid droplet formation, expressions of fat browning genes and activation of the insulin receptor-AKT pathway in differentiated 3T3-L1 cells. MiR-133a-3p agomir partially offset the effects of LXA4. SIRT1 was a downstream target gene of miR-133a-3p, participating in the promotive effects of LXA4 on fat browning. LXA4 promotes white adipose browning and relieves insulin resistance through miR-133a-3p/SIRT1 pathway.

biochemistry↗

Tension-induced adhesion mode switching: the interplay between focal adhesions and clathrin-containing adhesion complexes

Integrin-based adhesion complexes are crucial in various cellular processes, including proliferation, differentiation, and motility. While the dynamics of canonical focal adhesion complexes (FAs) have been extensively studied, the regulation and physiological implications of the recently identified clathrin-containing adhesion complexes (CCACs) are still not well understood. In this study, we investigated the spatiotemporal mechanoregulations of FAs and CCACs in a breast cancer model. Employing single-molecule force spectroscopy coupled with live-cell fluorescence microscopy, we discovered that FAs and CCACs are mutually exclusive and inversely regulated complexes. This regulation is orchestrated through the modulation of plasma membrane tension, in combination with distinct modes of actomyosin contractility that can either synergize with or counteract this modulation. Our findings indicate that increased membrane tension promotes the association of CCACs at integrin V{beta}5 adhesion sites, leading to decreased cancer cell proliferation, spreading, and migration. Conversely, lower membrane tension promotes the formation of FAs, which correlates with the softer membranes observed in cancer cells, thus potentially facilitating cancer progression. Our research provides novel insights into the biomechanical regulation of CCACs and FAs, revealing their critical and contrasting roles in modulating cancer cell progression.

cancer biology↗

Predicting RNA Sequence-Structure Likelihood via Structure-Aware Deep Learning

MotivationThe active functionalities of RNA are recognized to heavily dependent on the structure and sequence. Therefore, A model that can accurately evaluate a design by giving RNA sequence-structure pairs would be a valuable tool for many researchers. Machine learning methods have been explored to develop such tools, showing promising results. However, two key issues remain. Firstly, the performance of machine learning models is affected by the features used to characterize RNA. Currently, there is no consensus on which features are the most effective for characterizing RNA sequence-structure pairs. Secondly, most existing machine learning methods extract features describing entire RNA molecule. We argue that it is essential to define additional features that characterize nucleotides and specific sections of RNA structure to enhance the overall efficacy of the RNA design process. ResultsWe develop two deep learning models for evaluating RNA sequence-structure pairs. The first model, NU-ResNet, uses a convolutional neural network architecture that solves the aforementioned problems by explicitly encoding RNA sequence-structure information into a 3D matrix. Building upon NU-ResNet, our second deep learning model, NUMO-ResNet, incorporates additional information derived from the characterizations of RNA, specifically the 2D folding motifs. In this work, we introduce an automated method to extract these motifs based on fundamental secondary structure descriptions. To assess the robustness of our models, we conduct 10-fold cross validation. Furthermore, we evaluate the performance of both models on two independent testing datasets. Our proposed models demonstrate excellent performance across both datasets and surpass the performance of the ENTRNA approach. Availability and ImplementationThe corresponding source code and data for this research is available at https://github.com/yzhou617/NU-ResNet_and_NUMO-ResNet. ContactGiulia.Pedrielli@asu.edu Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗