Search bioRxiv⌕ Search

Biology subjects

Zhai, K.

Publications and source records attributed to Zhai, K..

7 recordsLinked to original sources

DNA-LLM2Vec: Architecture-Aware Task-Conditioned Adaptation of Decoder DNA Foundation Models

Decoder DNA foundation models pretrain with dense next-token prediction and scale to long sequences, but their representations are directional. This is a poor fit for genomic tasks that need sequence context both upstream and downstream of a locus. Bidirectional masked-language-model encoders supply this context, but they pretrain more sparsely and scale less easily to long windows. We introduce DNA-LLM2Vec, a staged decoder-to-encoder framework that opens bidirectional context, repairs the resulting mismatch with masked next-token prediction, and applies contrastive views encoding strand and positional invariance. We instantiate it on two architecturally distinct backbones, DNAGPT and HyenaDNA, and evaluate seven biological task families (grouped from GB, NT, GUE, and GUE+EPI) with frozen linear probing. The benefits of conversion are architecture-aware and task-conditioned rather than uniform. Crop-based positional contrastive learning helps most for the extended-context families (enhancer-promoter interaction and Virus-COVID), raising frozen Avg-EPI MCC to 27.5% for DNAGPT and 26.7% for HyenaDNA, whereas for shorter, local families the results depend on backbone and tokenisation. Beyond classification, bidirectional context and MNTP improve causal-eQTL variant-effect prediction specifically at distance, and a base-pair mutagenesis diagnostic shows that adaptation organises motif-aware contact structure while reverse-complement training imprints strand-consistent geometry, yet neither predicts the best downstream view. DNA-LLM2Vec therefore produces frozen bidirectional representations whose most effective adaptation view depends on the task structure and the backbone. Code and checkpoints are available at https://github.com/semi-nothing/dna-llm2vec.

bioinformatics↗

Somatostatin Receptors Shape Insulin and Glucagon Output within the Pancreatic Islet through Direct and Paracrine Effects

Aims/HypothesisPancreatic delta cells secrete somatostatin (SST), which can inhibit both alpha and beta cells of the pancreatic islet. By controlling insulin and glucagon release, delta cells play an important role in maintaining nutrient homeostasis. However, the mechanism by which a single inhibitory hormone inhibits both alpha and beta cells, which are often considered as functional antagonists in the counterregulatory control of blood glucose, has been a physiological riddle. Here, we solve this riddle through assessment of the contributions of alpha and beta cell specific somatostatin receptors to cell intrinsic behaviors and hormone release. MethodsIslets from mice constitutively expressing fluorescent sensors reporting on cyclic AMP and Ca2+ in both alpha and beta cells were imaged using stimuli to mimic the post-prandial state of a meal consisting of glucose and amino acids. This approach was coupled with cell specific somatostatin receptor antagonists to identify how somatostatin inhibits alpha and beta cell hormone output through modulation of cAMP and Ca2+ secondary messengers and paracrine interactions. ResultsOur results support and extend prior observation that somatostatin receptor 2 (SSTR2) is the only somatostatin receptor expressed by alpha cells, while somatostatin receptor 3 (SSTR3) is the only receptor expressed by mouse beta cells. Interestingly, SSTR2 and SSTR3 regulate downstream cAMP and Ca2+ signaling cascades differently within alpha and beta cells of intact islets. Stimulation of somatostatin receptors robustly inhibits cyclic AMP in alpha or beta cells. In contrast, stimulation of SSTR2 inhibits alpha cell Ca2+ with significantly greater potency compared to inhibition of beta cell Ca2+ via SSTR3. Despite the absence of SSTR2 on beta cells, blocking alpha cell SSTR2 during nutrient stimulation resulted in a significant increase in insulin release downstream of local release of glucagon. Conclusions/InterpretationOur observations address the physiological riddle of the delta cells role during the post-prandial phase where we demonstrate that somatostatin primarily inhibits alpha cell cAMP and Ca2+ via SSTR2, preventing glucagon release. Blocking SSTR2 resulted in an increase in locally released glucagon, which coupled with muted ability for SSTR3 to inhibit beta cell calcium under strong nutrient stimulation, results in potentiation of glucose stimulated insulin secretion from the beta cell. We conclude that the role of delta cells under nutrient stimulation is to modulate the volume of insulin release by tuning the strength of intra-islet paracrine potentiation of insulin secretion by glucagon, mediated via beta cell GLP1R. Research in ContextO_ST_ABSWhat is already known about this subject?C_ST_ABSO_LISomatostatin released by delta cells can attenuate glucagon release from the alpha cells and insulin release from the beta cells of the islet through inhibitory somatostatin receptors. C_LIO_LISomatostatin receptor 2 is a major receptor expressed on the mouse alpha cell surface, while mouse beta cells express SSTR3 on their primary cilia. C_LIO_LIParacrine signaling by alpha cell glucagon potentiates insulin release through the engagement of stimulatory glucagon-like peptide 1 receptors on beta cells. C_LI What are the key questions?O_LIHow do the cell specific somatostatin receptors of the alpha cell (SSTR2) and beta cell (SSTR3) differentially regulate cell-intrinsic Ca2+ and cAMP signaling and the indirect paracrine pathways that shape insulin secretion in the post-prandial state? C_LI What are the new findings?O_LIAlpha cells of the mouse islet express exclusively SSTR2 and not SSTR3. C_LIO_LISomatostatin is a more potent inhibitor of cytosolic calcium activity via alpha cell SSTR2 compared to beta cell SSTR3. C_LIO_LIPreventing somatostatin inhibition of the alpha cell in the presence of elevated glucose and amino acids potentiates insulin release through the engagement of glucagon-like peptide 1 receptors on the beta cell by glucagon. C_LI How might this impact on clinical practice in the foreseeable future?O_LIOur findings indicate that local feedback inhibition of beta cells is accomplished by tuning the strength of the paracrine potentiation of locally released glucagon. This is relevant in the context of developing treatments for hypoglycemia with alpha cell specific SSTR2-specific antagonists that could potentiate insulin release and potentially exacerbate hypoglycemic episodes. C_LI Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=177 SRC="FIGDIR/small/688371v1_ufig1.gif" ALT="Figure 1"> View larger version (71K): org.highwire.dtl.DTLVardef@1138ddaorg.highwire.dtl.DTLVardef@c613f9org.highwire.dtl.DTLVardef@49806eorg.highwire.dtl.DTLVardef@18ece86_HPS_FORMAT_FIGEXP M_FIG C_FIG

physiology↗

A Phosphorelay Circuit Drives Extracellular Alkalinization in Plant Receptor Kinase Signaling

Extracellular alkalinization has long been recognized as a hallmark of plant cell-surface receptor activation, including during pattern-triggered immunity (PTI); yet the mechanisms driving elicitor-induced alkalinization and its role in immune signaling remain unclear. Here, we demonstrate that inhibition of autoinhibited H+-ATPases (AHAs) is required for elicitor-induced extracellular alkalinization. This alkalinization is essential for immune signaling mediated by diverse plasma membrane-localized receptor kinases (RKs) through modulation of ligand-receptor interactions. Notably, RKs transduce elicitor-triggered signaling via BOTRYTIS-INDUCED KINASE 1 (BIK1), which inhibits AHA activity by disrupting AHA-GENERAL REGULATORY FACTOR (GRF) interactions through a conserved phosphorylation event. Interestingly, this pathway is crucial for cell wall damage (CWD) responses involving the RK MALE DISCOVERER 1-INTERACTING RECEPTOR LIKE KINASE 2 (MIK2) and its ligand, SERINE RICH ENDOGENOUS PEPTIDE 18 (SCOOP18). Our findings reveal a conserved phospho-regulatory pathway that governs extracellular alkalinization to coordinate plant immune signaling, offering new insights into plant stress resilience.

plant biology↗

P(all-atom) Is Unlocking New Path For Protein Design

We introduce Pallatom, an innovative protein generation model capable of producing protein structures with all-atom coordinates. Pallatom directly learns and models the joint distribution P(structure, seq) by focusing on P(all-atom), effectively addressing the interdependence between sequence and structure in protein generation. To achieve this, we propose a novel network architecture specifically designed for all-atom protein generation. Our model employs a dual-track framework that tokenizes proteins into residue-level and atomic-level representations, integrating them through a multi-layer decoding process with "traversing" representations and recycling mechanism. We also introduce the atom14 representation method, which unifies the description of unknown side-chain coordinates, ensuring high fidelity between the generated all-atom conformation and its physical structure. Experimental results demonstrate that Pallatom excels in key metrics of protein design, including designability, diversity, and novelty, showing significant improvements across the board. Our model not only enhances the accuracy of protein generation but also exhibits excellent sampling efficiency, paving the way for future applications in larger and more complex systems.

bioinformatics↗

Integrated multi-omics data reveals the molecular subtypes and the contruction of novel signature to aid Immunotherapy

Malignant cancer exhibits a severe imbalance between the active of immune response and tolerance. And the different and connection between the establishment and deficiency of the tumor immune response is still unknown entirely. Here, we used ten clustering algorithms based on five immune cells infiltration matrix to identified two specific molecular subtypes and determine its potential clinical outcomes and immune response effect. The cluster A subtype can be called the cold inflammatory subtype, which revealed the worst prognosis. And the cluster B can be identified as hot inflammatory subtype with high activated immune-related pathways. Then we used the WGCNA algorithms and limma to confirm the driven genes between the cluster A and B. 67 genes were identified the cluster A signature genes, and based on that we used the ssgsea algorithm to quantize the two subtypes. We confirmed the high score group had high activated level of immune activity pathways and better survival rate. The tumorigenesis-associated pathways were analysed in low score group, indicating active cell proliferation may be associated with poor prognosis. An in-silico drug screen predicted the erlotinib can be an effective compound in low score group based on the lung cancer cell line. Our research confirmed that the score can worked as an effective immunotherapy index, which provide a promising therapeutic strategies on immune therapeutics for lung adenocarcinoma.

cancer biology↗

ERp44 is Required for Endocardial Cushion Development by Regulating VEGFA Secretion in Myocardium

Rationale: Endocardial cushions are precursors of the valvoseptal complex that separates the four heart chambers and control blood flow through the heart. Abnormalities in endocardial cushion development lead to atrioventricular septal defects (AVSDs), which affect 1 in 2,100 live births. Several genes have been implicated in the development of endocardial cushions. Specifically, endoplasmic reticulum-resident protein 44 (ERp44) has been found to play a role in the early secretory pathway, but its function in heart development has not been well studied. Objective: The goal of this study was to investigate the role of ERp44 in heart development in mice. Approach and Results: Using conventional and tissue-specific knockout mouse models, we demonstrated that ERp44 plays a specific role in heart development. ERp44 knockout (KO) mice were smaller in size, and most mice died during early postnatal life. KO hearts exhibited the typical phenotypes of congenital heart diseases, such as abnormal heart shapes as well as severe septal and valvular defects. Similar phenotypes were found in cTnt-cre+/-; Erp44fl/fl mice, which indicated that myocardial ERp44 principally controls endocardial cushion formation. Further studies demonstrated that the deletion of ERp44 significantly decreased the proliferation of cushion cells and impaired the endocardial-mesenchymal transition (EndMT), which was followed by endocardial cushion dysplasia. Finally, we found that ERp44 directly bound to VEGFA and controlled its release. Conclusions: ERp44 contributes to the development of the endocardial cushion by affecting the EndMT of cushion cells by regulating VEGFA release in myocardial cells.

developmental biology↗

HAUSP Stabilizes SOX2 through Deubiquitination to Maintain Self-renewal and Tumorigenic Potential of Glioma Stem Cells

Glioblastoma (GBM) is the most lethal brain tumor containing glioma stem cells (GSCs) that promote malignant growth and therapeutic resistance. The self-renewal and tumorigenic potential of GSCs are maintained by core stem cell transcription factors including SOX2. Defining the posttranslational regulation of SOX2 may offer new insights into GSC biology and potential therapeutic opportunity. Here, we discover that HAUSP stabilizes SOX2 through deubiquitination to maintain GSC self-renewal and tumorigenic potential. HAUSP is preferentially expressed in GSCs in perivascular niches in GBMs. Disrupting HAUSP by shRNA or its inhibitor P22077 promoted SOX2 degradation, induced GSC differentiation, impaired GSC tumorigenic potential, and suppressed GBM tumor growth. Importantly, pharmacological inhibition of HAUSP synergized with radiation to inhibit GBM growth and extended animal survival, indicating that targeting HAUSP may overcome GSC-mediated radioresistance. Our findings reveal an unappreciated crucial role of HAUSP in the GSC maintenance and provide a promising target for developing effective anti-GSC therapeutics to improve GBM treatment. HighlightsO_LIHAUSP deubiquitinates and stabilizes SOX2 in glioma stem cells (GSCs). C_LIO_LIHAUSP is preferentially expressed by GSCs in perivascular niches in GBMs. C_LIO_LIHAUSP is required for maintaining GSC self-renewal and tumorigenic potential. C_LIO_LITargeting HAUSP inhibited malignant growth in GSC-derived GBM xenografts. C_LIO_LIInhibition of HAUSP synergized with radiation to suppress GBM tumor growth. C_LI

cancer biology↗