Search bioRxiv⌕ Search

Biology subjects

Chen, W.

Publications and source records attributed to Chen, W..

At least 253 records · Page 14Linked to original sources
bioRxiv2024

Self-inspired learning to denoise for live-cell super-resolution microscopy

Every collected photon is precious in live-cell super-resolution (SR) fluorescence microscopy for contributing to breaking the diffraction limit with the preservation of temporal resolvability. Here, to maximize the utilization of accumulated photons, we propose SN2N, a Self-inspired Noise2Noise engine with self-supervised data generation and self-constrained learning process, which is an effective and data-efficient learning-based denoising solution for high-quality SR imaging in general. Through simulations and experiments, we show that the SN2Ns performance is fully competitive to the supervised learning methods but circumventing the need for large training-set and clean ground-truth, in which a single noisy frame is feasible for training. By one-to-two orders of magnitude increased photon efficiency, the direct applications on various confocal-based SR systems highlight the versatility of SN2N for allowing fast and gentle 5D SR imaging. We also integrated SN2N into the prevailing SR reconstructions for artifacts removal, enabling efficient reconstructions from limited photons. Together, we anticipate our SN2N and its integrations could inspire further advances in the rapidly developing field of fluorescence imaging and benefit subsequent precise structure segmentation irrespective of noise conditions.

bioengineering↗
bioRxiv2024

Cbfβ regulates Wnt/β-catenin, Hippo/Yap, and TGFβ signaling pathways in articular cartilage homeostasis and protects from ACLT surgery-induced osteoarthritis

As the most common degenerative joint disease, osteoarthritis (OA) contributes significantly to pain and disability during aging. Several genes of interest involved in articular cartilage damage in OA have been identified. However, the direct causes of OA are poorly understood. Evaluating the public human RNA-seq dataset showed that Cbf{beta}, (subunit of a heterodimeric Cbf{beta}/Runx1,Runx2, or Runx3 complex) expression is decreased in the cartilage of patients with OA. Here, we found that the chondrocyte-specific deletion of Cbf{beta} in tamoxifen-induced Cbf{beta}f/fCol21-CreERT mice caused a spontaneous OA phenotype, worn articular cartilage, increased inflammation, and osteophytes. RNA-sequencing analysis showed that Cbf{beta} deficiency in articular cartilage resulted in reduced cartilage regeneration, increased canonical Wnt signaling and inflammatory response, and decreased Hippo/YAP signaling and TGF-{beta} signaling. Immunostaining and western blot validated these RNA-seq analysis results. ACLT surgery-induced OA decreased Cbf{beta} and Yap expression and increased active {beta}-catenin expression in articular cartilage, while local AAV-mediated Cbf{beta} overexpression promoted Yap expression and diminished active {beta}-catenin expression in OA lesions. Remarkably, AAV-mediated Cbf{beta} overexpression in knee joints of mice with OA showed the significant protective effect of Cbf{beta} on articular cartilage in the ACLT OA mouse model. Overall, this study, using loss-of-function and gain-of-function approaches, uncovered that low expression of Cbf{beta} may be the cause of OA. Moreover, Local admission of Cbf{beta} may rescue and protect OA through decreasing Wnt/{beta}-catenin signaling, and increasing Hippo/Yap signaling and TGF{beta}/Smad2/3 signaling in OA articular cartilage, indicating that local Cbf{beta} overexpression could be an effective strategy for treatment of OA.

pathology↗
bioRxiv2024

Blocker-SELEX: A Structure-guided Strategy for Developing Inhibitory Aptamers Disrupting Undruggable Transcription Factor Interactions

Despite the well-established significance of transcription factors (TFs) in pathogenesis, their utilization as pharmacological targets has been limited by the inherent challenges associated with modulating their protein-protein and protein-DNA interactions. The lack of defined small-molecule binding pockets and the nuclear localization of TFs do not favor the use of small-molecule inhibitors, or neutral antibodies, in blocking TF interactions. Aptamers are short oligonucleotides exhibiting high affinity and specificity for a diverse range of targets. Large molecular weights, expansive blocking surfaces and efficient cellular internalization make aptamers a compelling molecular tool for use as traditional TF interaction modulators. Here, we report a structure-guided design strategy called Blocker-SELEX to develop inhibitory aptamers (iAptamer) that selectively block TF interactions. Our approach led to the discovery of iAptamers that cooperatively disrupts SCAF4/SCAF8-RNA Polymerase II (RNAP2) interactions, thereby dysregulating RNAP2-dependent gene expression and splicing and, in turn, leading to the impairment of cell proliferation. This approach was further applied to develop iAptamers to efficiently block WDR5-MYC interaction with a nexus in cancer. Taken together, our study highlights the potential of Blocker-SELEX in developing iAptamers that effectively disrupt potentially pathogenic TF interactions with attendant implications for iAptamers as chemical tools for use in the study of biological functions of TF interactions, but also for potential use in nucleic acids drug discovery.

biochemistry↗
bioRxiv2024

Plaat1l1 controls feeding induced NAPE biosynthesis and contributes to energy balance regulation in zebrafish.

Dysregulation of energy balance leading to obesity is a significant risk factor for cardiometabolic diseases such as diabetes, non-alcoholic fatty liver disease and atherosclerosis. In rodents and a number of other vertebrates, feeding has been shown to induce a rapid rise in the intestinal levels of N-acyl-ethanolamines (NAEs) and the chronic consumption of a high fat diet abolishes this rise. Administering NAEs to rodents consuming a high fat diet reduces their adiposity, in part by reducing food intake and enhancing fat oxidation, so that feeding-induced intestinal NAE biosynthesis appears to be critical to appropriate regulation of energy balance. However, the contribution of feeding-induced intestinal NAE biosynthesis to appropriate energy balance remains poorly understood because the specific enzymes responsible for feeding-induced NAE biosynthesis have not been identified. The rate-limiting step in the intestinal biosynthesis of NAEs is formation of their immediate precursors, the N-acyl-phosphatidylethanolamines (NAPEs), by phosphatidylethanolamine N-acyltransferases (NATs). At least six NATs are found in humans and multiple homologs of these NATs are found in most vertebrate species. In recent years, the fecundity and small size of zebrafish (Danio rerio), as well as their similarities in feeding behavior and energy balance regulation with mammals, have led to their use to model key features of cardiometabolic disease. We therefore searched the Danio rerio genome to identify all NAT homologs and found two additional NAT homologs besides the previously reported plaat1, rarres3, and rarres3l, and used CRISPR/cas9 to delete these two NAT homologs (plaat1l1 and plaat1l2). While wild-type fish markedly increased their intestinal NAPE and NAE levels in response to a meal after fasting, this response was completely ablated in plaat1l1-/- fish. Furthermore, plaat1l1-/-fish fed a standard flake diet had increased weight gain and glucose intolerance compared to wild-type fish. The results support a critical role for feeding-induced NAE biosynthesis in regulating energy balance and suggest that restoring this response in obese animals could potentially be used to treat obesity and cardiometabolic disease.

biochemistry↗
bioRxiv2023

Macrophages as Key Mediators of Trp-1-Specific CD4+ T Cell Therapy for Murine Melanoma

There is a growing body of evidence indicating that CD4+ T cells, alongside CD8+ T cells, can effectively combat cancer. However, the mechanisms underlying the anti-tumor properties of CD4+ T cells are complex and not yet fully understood. To investigate these mechanisms, we utilized a murine model of tyrosinase-related protein 1 (Trp1)-specific CD4+ T cell adoptive transfer therapy for treating melanoma. By employing single-cell RNA sequencing (scRNA-seq), we analyzed the immune cells present in the tumors of mice that received adoptive transfer of Trp1-specific CD4+ T cells. Unexpectedly, within the tumor-infiltrating immune cells, the Trp1 CD4+ T cell population was relatively small, displaying characteristics indicative of exhaustion. In contrast, the most prominent cell cluster comprised macrophages, expressing high levels of the T cell inhibitory receptor ligand PD-L1 and the pro-inflammatory cytokine IL-1{beta}, suggesting a distinct M1 phenotype. Systemic depletion of macrophages following Trp1 CD4+ T cell transfer therapy compromised the antitumor effectiveness and resulted in tumor recurrence. These findings highlight the crucial role of innate macrophages as an effector cell population in Trp1-specific CD4+ T cell adoptive cell transfer therapy.

immunology↗
bioRxiv2023

Chlamydial protease-like activating factor targets SLC7A11 for degradation to induce ferroptosis and facilitate dissemination.

Chlamydia trachomatis, the most prevalent bacterial agent of sexually transmitted infections, possesses remarkable capacities for dissemination within the host, leading to reproductive health complications. The release of progeny through the orchestrated lysis of host cells plays a crucial role in Chlamydia dissemination, but the underlying molecular mechanisms remain largely elusive. Here, we uncovered a novel mechanism by which Chlamydia induces host cells ferroptosis to facilitate its dissemination. This process involves the degradation of host protein SLC7A11 by the chlamydial protease-like activating factor (CPAF), resulting in glutathione depletion, oxidative damage, and subsequent host cell lysis characterized by lipid peroxidation. Infection with CPAF-deficient strain fails to induce host cells ferroptosis, leading to restricted progeny release. Importantly, inhibiting ferroptosis effectively limits the release of Chlamydia progeny, highlighting its potential as a therapeutic strategy for controlling Chlamydia dissemination. These findings provide insights into the chlamydial conserved dissemination strategy and enhance understanding of its pathogenesis.

microbiology↗
bioRxiv2023

eRNA-IDO: a one-stop platform for identification, interactome discovery and functional annotation of enhancer RNAs

Increasing evidence proves the transcription of enhancer RNA (eRNA) and its important role in gene regulation. However, we are only at the infancy stage of understanding eRNA interactions with other biomolecules and the corresponding functionality. To accelerate eRNA mechanistic study, we present the first integrative computational platform for human eRNA identification, interactome discovery, and functional annotation, termed eRNA-IDO. eRNA-IDO comprises two modules: eRNA-ID and eRNA-Anno. Functionally, eRNA-ID identifies eRNAs from de novo assembled transcriptomes. The bright spot of eRNA-ID is indeed the inclusion of 8 kinds of enhancer makers, whose combination enables users to personalize enhancer regions flexibly and conveniently. In addition, eRNA-Anno provides cell/tissue specific functional annotation for any novel and known eRNAs through discovering eRNA interactome from the prebuilt or user-defined eRNA-coding gene networks. The pre-built networks include GTEx-based normal co-expression networks, TCGA-based cancer co-expression networks, and omics-based eRNA-centric regulatory networks. Our eRNA-IDO carries sufficient practicability and significance for understanding the biogenesis and functions of eRNAs. The eRNA-IDO server is freely available at http://bioinfo.szbl.ac.cn/eRNA_IDO/.

bioinformatics↗
bioRxiv2023

Saturation genome editing-based functional evaluation and clinical classification of BRCA2 single nucleotide variants

Germline BRCA2 loss-of function (LOF) variants identified by clinical genetic testing predispose to breast, ovarian, prostate and pancreatic cancer. However, variants of uncertain significance (VUS) (n>4000) limit the clinical use of testing results. Thus, there is an urgent need for functional characterization and clinical classification of all BRCA2 variants. Here we report on comprehensive saturation genome editing-based functional characterization of 97% of all possible single nucleotide variants (SNVs) in the BRCA2 DNA Binding Domain hotspot for pathogenic missense variants that is encoded by exons 15 to 26. The assay was based on deep sequence analysis of surviving endogenously targeted haploid cells. A total of 7013 SNVs were characterized as functionally abnormal (n=955), intermediate/uncertain, or functionally normal (n=5224) based on 95% agreement with ClinVar known pathogenic and benign standards. Results were validated relative to batches of nonsense and synonymous variants and variants evaluated using a homology directed repair (HDR) functional assay. Breast cancer case-control association studies showed that pooled SNVs encoding functionally abnormal missense variants were associated with increased risk of breast cancer (odds ratio (OR) 3.89, 95%CI: 2.77-5.51). In addition, 86% of tumors associated with abnormal missense SNVs displayed loss of heterozygosity (LOH), whereas 26% of tumors with normal variants had LOH. The functional data were added to other sources of information in a ClinGen/ACMG/AMP-like model and 700 functionally abnormal SNVs, including 220 missense SNVs, were classified as pathogenic or likely pathogenic, while 4862 functionally normal SNVs, including 3084 missense SNVs, were classified as benign or likely benign. These classified variants can now be used for risk assessment and clinical care of variant carriers and the remaining functional scores can be used directly for clinical classification and interpretation of many additional variants. SummaryGermline BRCA2 loss-of function (LOF) variants identified by clinical genetic testing predispose to several types of cancer. However, variants of uncertain significance (VUS) limit the clinical use of testing results. Thus, there is an urgent need for functional characterization and clinical classification of all BRCA2 variants to facilitate current and future clinical management of individuals with these variants. Here we show the results from a saturation genome editing (SGE) and functional analysis of all possible single nucleotide variants (SNVs) from exons 15 to 26 that encode the BRCA2 DNA Binding Domain hotspot for pathogenic missense variants. The assay was based on deep sequence analysis of surviving endogenously targeted human haploid HAP1 cells. The assay was calibrated relative to ClinVar known pathogenic and benign missense standards and 95% prevalence thresholds for functionally abnormal and normal variants were identified. Thresholds were validated based on nonsense and synonymous variants. SNVs encoding functionally abnormal missense variants were associated with increased risks of breast and ovarian cancer. The functional assay results were integrated into a ClinGen/ACMG/AMP-like model for clinical classification of the majority of BRCA2 SNVs as pathogenic/likely pathogenic or benign/likely benign. The classified variants can be used for improved clinical management of variant carriers.

genomics↗
bioRxiv2023

Prediction of DNA i-Motifs Via Machine Learning

i-Motifs (iMs), are secondary structures formed in cytosine-rich DNA sequences and are involved in multiple functions in the genome. Although putative iM forming sequences are widely distributed in the human genome, the folding status and strength of putative iMs vary dramatically. Much previous research on iM has focused on assessing the iM folding properties using biophysical experiments. However, there are no dedicated computational tools for predicting the folding status and strength of iM structures. Here, we introduce a machine learning pipeline, iM-Seeker, to predict both folding status and structural stability of DNA iMs. The programme iM-Seeker incorporates a Balanced Random Forest classifier trained on genome-wide iMab antibody-based CUT&Tag sequencing data to predict the folding status and an Extreme Gradient Boosting regressor to estimate the folding strength according to both literature biophysical data and our in-house biophysical experiments. iM-Seeker predicts DNA iM folding status with a classification accuracy of 81% and estimates the folding strength with coefficient of determination (R2) of 0.642 on the test set. Model interpretation confirms that the nucleotide composition of the C-rich sequence significantly affects iM stability, with a positive correlation with sequences containing cytosine and thymine and a negative correlation with guanine and adenine. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=140 SRC="FIGDIR/small/571121v1_ufig1.gif" ALT="Figure 1"> View larger version (31K): org.highwire.dtl.DTLVardef@11f4398org.highwire.dtl.DTLVardef@18eb7e9org.highwire.dtl.DTLVardef@cc49c6org.highwire.dtl.DTLVardef@1ff91a8_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗
bioRxiv2023

An automatic glaucoma grading method based on attention mechanism and EfficientNetB3 network

Deep learning has received considerable attention in the computer vision field and has been widely studied, especially in recognizing and diagnosing ophthalmic diseases. Currently, glaucoma recognition algorithms are mostly based on unimodal OCT, the visual field for glaucoma auxiliary diagnosis. Such algorithms have poor robustness and limited help for glaucoma auxiliary diagnosis; therefore, this experiment is proposed to use a 2D fundus image and 3D-OCT scanner two modal data as the experimental dataset and use the EfficientNet-B3 network and ResNet34 network models for feature extraction and fusion to improve automatic glaucoma grading accuracy. Since fundus images usually contain a large number of meaningless black background regions, this may lead to feature redundancy. Therefore, this experiment employs an attention mechanism that focuses the attention of the convolutional neural network on eye subject features to improve the performance of the glaucoma autoclassification model.

bioengineering↗
bioRxiv2023

Linking root-associated fungal and bacterial functions to root economics

Tree roots form symbioses with soil microbes to acquire nutrients, but the relationships between root nutrient acquisition strategies and microbial community composition remain poorly understood. Here, we measured root traits and root-associated fungal and bacterial guilds in 336 trees of 52 species from a subtropical forest. We found a fungal gradient from ectomycorrhizal to saprotrophic dominance, which corresponded with a shift from organic to mineral nutrient economics. This fungal gradient was aligned with the increase of root nitrogen concentration, suggesting a linkage from simple root trait to fungal-mediated carbon-nutrient cycling. We also found that the functional composition of fungal and bacterial communities was closely correlated with host root-zone pH, which often varied among coexisting trees. Root-zone pH was independent of the common root traits, underpinning a potential new gradient in the root trait space. Our findings integrate microbial functions into the root economics framework, thereby advancing the understanding of diversity of nutrient acquisition strategies across forest trees.

ecology↗
bioRxiv2023

Pre-mRNA fate decision safeguards the fidelity of the inflammatory response

The fidelity of immune responses depends on a timely controlled and selective mRNA degradation that is largely driven by RNA-binding proteins (RBPs)[1, 2]. It remains unclear whether the selection of an individual mRNA molecule for degradation is governed by stochastic or directed processes. Here, we show that tristetraprolin (TTP, also known as ZFP36), an essential anti-inflammatory RBP[3], destabilizes target mRNAs via a hierarchical molecular assembly. The formation of the assembly strictly relies on the interaction of TTP with RNA. The TTP homolog ZFP36L1 exhibits similar requirements, indicating a broader relevance of this regulatory program. Unexpectedly, the assembly of the cytoplasmic mRNA-destabilization complex is licensed in the nucleus by TTP binding to pre-mRNA, while mature cytoplasmic mRNA does not constitute a de novo TTP target. Hence, the fate of an inflammation-induced mRNA is decided concomitantly with its synthesis. This decision mechanism prevents the translation of superfluous and potentially harmful inflammation mediators, ensuring an efficient cessation of the immune response, irrespective of transcriptional activity.

molecular biology↗
bioRxiv2023

Exploring the family feud: a fungal endophyte induces local cell wall-mediated resistance in wheat roots against the closely related "take-all" pathogen

Take-all disease, caused by the ascomycete fungus Gaeumannomyces tritici, is one of the most important root diseases of wheat worldwide. The fungus invades the roots and destroys the vascular tissue, hindering the uptake of water and nutrients. Closely related non-pathogenic species in the Magnaporthaceae family, such as Gaeumannomyces hyphopodioides, occur naturally in arable and grassland soils and have previously been reported to reduce take-all disease in field studies. However, the mechanism of take-all protection has remained unknown. Here, we characterise the root infection biologies of G. tritici and G. hyphopodioides in wheat. We investigate the ultrastructure of previously described "subepidermal vesicles" (SEVs), produced in wheat roots by non-pathogenic G. hyphopodioides, but not by pathogenic G. tritici. We show that G. hyphopodioides SEVs share key characteristics of fungal resting structures; containing a greater number of putative lipid bodies and a significantly thickened cell wall compared to infection hyphae. We demonstrate that take-all control is achieved via local but not systemic host changes in response to prior G. hyphopodioides root colonisation. A time-course wheat RNA sequencing analysis revealed extensive transcriptional reprogramming in G. hyphopodioides colonised tissues, characterised by a striking downregulation of key cell-wall related genes, including cellulose synthase (CESA), and xyloglucan endotransglucosylase/hydrolase (XTH) genes. In the absence of take-all resistant wheat cultivars or non-virulent G. tritici strains, studying closely related non-pathogenic G. hyphopodioides provides a much-needed avenue to elucidate take-all resistance mechanisms in wheat.

plant biology↗
bioRxiv2023

High Frequency Longitudinal RNAseq Reveals Temporally Varying Genes and Recovery Trajectories in Rats

When living organisms are exposed to potentially toxic chemicals, they respond via changes in gene expression. Traditional differentially expressed gene analysis based on before/after blood samples does not reveal the responses temporal dynamics, and often produces false positives and negatives. Here, we performed longitudinal daily RNA sequencing on rats dosed with tetracycline, isoniazid, carbon tetrachloride, or valproate. We identified 4,302 temporally variable genes (TVGs) with statistically strong change in expression following dosing. Projecting TVG expressions into a 3-dimensional principal component (PC) space reveals consistent trajectories for recovery following dosing, and enables separation of healthy from recovering states with 91% to 99% area under the receiver operator curve. Finally, we observed that Fast Recovery vs. Slow Recovery rats exhibited distinct temporal expression patterns in the TVGs, suggesting that individual variations could be potentially captured via longitudinal RNAseq analysis.

molecular biology↗
bioRxiv2023

Preclinical Characterization of the Omicron XBB.1.5-Adapted BNT162b2 COVID-19 Vaccine

As SARS-CoV-2 continues to evolve, increasing in its potential for greater transmissibility and immune escape, updated vaccines are needed to boost adaptive immunity to protect against COVID-19 caused by circulating strains. Here, we report features of the monovalent Omicron XBB.1.5-adapted BNT162b2 vaccine, which contains the same mRNA backbone as the original BNT162b2 vaccine, modified by the incorporation of XBB.1.5-specific sequence changes in the encoded prefusion-stabilized SARS-CoV-2 spike protein (S(P2)). Biophysical characterization of Omicron XBB.1.5 S(P2) demonstrated that it maintains a prefusion conformation that adopts a flexible and predominantly open one-RBD-up state, with high affinity binding to the human ACE-2 receptor. When administered as a 4th dose in BNT162b2-experienced mice, the monovalent Omicron XBB.1.5 vaccine elicited substantially higher serum neutralizing titers against pseudotyped viruses of Omicron XBB.1.5, XBB.1.16, XBB.1.16.1, XBB.2.3, EG.5.1 and HV.1 sublineages and the phylogenetically distant BA.2.86 lineage than the bivalent Wild Type + Omicron BA.4/5 vaccine. Similar trends were observed against Omicron XBB sublineage pseudoviruses when the vaccine was administered as a 2-dose primary series in naive mice. Strong S-specific Th1 CD4+ and IFN{gamma}+ CD8+ T cell responses were also observed. These findings, together with prior experience with variant-adapted vaccine responses in preclinical and clinical studies, suggest that the monovalent Omicron XBB.1.5-adapted BNT162b2 vaccine is anticipated to confer protective immunity against dominant SARS-CoV-2 strains. ONE-SENTENCE SUMMARYThe monovalent Omicron XBB.1.5-adapted BNT162b2 mRNA vaccine encodes a prefusion-stabilized spike immunogen that elicits more potent neutralizing antibody responses against homologous XBB.1.5 and other circulating sublineage pseudoviruses compared to the bivalent Wild Type + Omicron BA.4/5 BNT162b2 vaccine, thus demonstrating the importance of annual strain changes to the COVID-19 vaccine.

immunology↗
bioRxiv2023

Methyl-TWAS: A powerful method for in silico transcriptome-wide association studies (TWAS) using long-range DNA methylation

In silico transcriptome-wide association studies (TWAS) are commonly used to test whether expression of specific genes is linked to a complex trait. However, genotype-based in silico TWAS such as PrediXcan, exhibit low prediction accuracy for a majority of genes because genotypic data lack tissue- and disease-specificity and are not affected by the environment. Because methylation is tissue-specific and, like gene expression, can be modified by environment or disease status, methylation should predict gene expression with more accuracy than SNPs. Therefore, we propose Methyl-TWAS, the first approach that utilizes long-range methylation markers to impute gene expression for in silico TWAS through penalized regression. Methyl-TWAS 1) predicts epigenetically regulated/associated expression (eGReX), which incorporates tissue-specific expression and both genetically- (GReX) and environmentally-regulated expression to identify differentially expressed genes (DEGs) that could not be identified by genotype-based methods; and 2) incorporates both cis- and trans- CpGs, including various regulatory regions to identify DEGs that would be missed using cis-methylation only. Methyl-TWAS outperforms PrediXcan and two other methods in imputing gene expression in the nasal epithelium, particularly for immunity-related genes and DEGs in atopic asthma. Methyl-TWAS identified 3,681 (85.2%) of the 4,316 DEGs identified in a previous TWAS of atopic asthma using measured expression, while PrediXcan could not identify any gene. Methyl-TWAS also outperforms PrediXcan for expression imputation as well as in silico TWAS in white blood cells. Methyl-TWAS is a valuable tool for in silico TWAS, leveraging a growing body of publicly available genome-wide DNA methylation data for a variety of human tissues.

bioinformatics↗
bioRxiv2023

Investigation of contributions from cortical and subcortical brain structures for speech decoding

Language impairments often arise from severe neurological disorders, prompting the development of neural prosthetics based on electrophysiological signals for the restoration of comprehensible language information. Previous decoding efforts have focused mainly on signals from the cerebral cortex, neglecting the potential contributions of subcortical brain structures to speech decoding in brain-computer interfaces (BCIs). This study aims to explore the role of subcortical structures for speech decoding by utilizing stereotactic electroencephalography (sEEG). Two native Mandarin Chinese speakers, who underwent sEEG implantation for pharmaco-resistant epilepsy, participated in this study. sEEG contacts were primarily located in the superior temporal gyrus, middle temporal gyrus, inferior temporal gyrus, thalamus, hippocampus, insular gyrus, amygdala, and parahippocampal gyrus. The participants were asked to read Chinese text, which included 407 Chinese characters (covering all Chinese syllables), displayed on a screen after receiving prompts. 1-30, 30-70 and 70-150 Hz frequency band powers of sEEG signals were used as key features. A deep learning model based on long short-term memory (LSTM) was developed to evaluate the contribution of different brain structures during encoding of speech. Prediction of speech characteristics of consonants (articulatory place and manner) and tone within single words based on the selected features and electrode contact locations was made. Cortical signals were generally better at articulatory place prediction (86.5% accuracy, chance level = 12.5%), while cortical and subcortical signals predicted articulatory manner at similar level (51.5% vs 51.7% accuracy, respectively, chance level = 14.3%). Subcortical signals generated better prediction for tone (around 58.3% accuracy, chance level = 25%). Superior temporal gyrus remains highly relevant during speech decoding for both consonants and tone. Prediction reached the highest level when cortical and subcortical inputs were combined, especially for tone prediction. Our findings indicate that both cortical and subcortical structures can play crucial roles for speech decoding, each contributing to different aspects of speech.

neuroscience↗