Biology subjects

Cheng, L.

Publications and source records attributed to Cheng, L..

At least 55 records · Page 3Linked to original sources

bioRxiv2025

Synchronized Latency Reversal and Immune Clearance by a Multifunctional Fusion Protein Enables HIV-1 Reservoir Reduction

Strategies that concurrently reactivate latent reservoirs and enhance immune-mediated clearance hold significant promise for achieving an HIV cure. Here, we developed hyperIL-15xsCD4-Fc (15xsCD4-Fc), a fusion protein that integrates latency reactivation, targeted immune engagement, and effector-mediated killing into a single molecule. This agent not only potently reverses HIV-1 latency in CD4+ T cells from people living with HIV-1 (PLWH) through coordinated IL-15 receptor signaling and sCD4-mediated HIV-1 envelope (Env) engagement, but also enhances antigen-specific CD8+ T cell response in PBMCs derived from PLWH. Furthermore, 15xsCD4-Fc enables Env-specific elimination of reactivated latently infected cells by NK cells while preventing off-target cytotoxicity. In PBMCs from PLWH, 15xsCD4-Fc reduced replication-competent HIV-1 DNA by 93.8%. In antiretroviral treated HIV-1-infected humanized mice, the molecule demonstrated both safety and efficacy in diminishing viral reservoir in lymphoid organs. This spatiotemporally coupled approach to reservoir exposure and immune recognition establishes a clinically viable strategy for clearing the HIV-1 reservoir. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=195 SRC="FIGDIR/small/693434v1_ufig1.gif" ALT="Figure 1"> View larger version (44K): org.highwire.dtl.DTLVardef@8f8f0org.highwire.dtl.DTLVardef@10f63f7org.highwire.dtl.DTLVardef@16c1de2org.highwire.dtl.DTLVardef@11e1787_HPS_FORMAT_FIGEXP M_FIG C_FIG

immunology↗

bioRxiv2025

Cross-Species Cortical Geometry Reveals Conserved Gradients Across Primates and Human-Specific Expansion

The primate cerebral cortex, characterized by its complex structural geometry, underlies advanced cognitive functions and represents a defining feature distinguishing primates from other mammals. However, cross-species patterns of cortical geometry and the links between human cortical geometry and transcriptional architecture remain poorly understood. We developed a geometry-based cross-species cortical alignment framework to systematically investigate the similarities and differences in structural connectivity and cortical expansion characteristics among macaques, chimpanzees, and humans, and additionally explored the transcriptional underpinnings of human cortical geometry. Our analysis revealed conserved spatial patterns of cortical geometric features across species, providing the foundation for constructing a cross-species structural common space to support the alignment framework. We found that primary sensory, somatomotor, and face-selective regions exhibited high structural connectivity similarity across species, whereas prefrontal and parietal association cortices displayed significant divergence. We also identified disproportionate cortical expansion in the default mode network, with a consistent expansion trend across different evolutionary lineages in primates. Furthermore, neuroimage-transcription analysis indicated that cortical geometric features were correlated with transcriptional profiles enriched in neurodevelopmental and connectivity-related pathways. These results highlight a conserved yet hierarchically differentiated organization of the cerebral cortex in primates, providing new insights into the biological basis of human brain evolution.

neuroscience↗

bioRxiv2025

CDH3 as a Novel Therapeutic Target in Basal-like Double-Negative Prostate Cancer

PurposeBasal-like (also known as double-negative) prostate cancers are aggressive tumors that lack effective targeted therapies. We aimed to delineate the role of CDH3 (P-cadherin) in basal-like prostate cancer and evaluate CDH3-directed therapeutic strategies. MethodsWe integrated genetically engineered mouse models (GEMMs) of prostate cancer, bulk and single-cell transcriptomic analyses, and a suite of in vitro and in vivo experiments. CDH3 expression and associated signaling pathways were examined in Pten/Apc double-knockout mouse prostates and human datasets. Functional studies included antibody-drug conjugate (ADC) cytotoxicity assays and the development of chimeric antigen receptor (CAR) T cells targeting CDH3, tested in prostate cancer cell lines and xenograft models. ResultsPten/Apc double deletion in prostate GEMMs led to highly aggressive tumors with markedly elevated CDH3 expression and enrichment of non-canonical WNT signaling components. Transcriptomic analyses of patient-derived prostate tumors confirmed that CDH3 is significantly upregulated in basal-like prostate cancer subtypes relative to luminal subtypes. Single-cell RNA sequencing revealed CDH3 expression predominantly in basal epithelial cells. Mechanistically, we found that active YAP1 signaling and a WNT5A-ROR2 non-canonical WNT axis drive CDH3 expression. Targeting CDH3 with a CDH3-specific ADC induced potent, antigen-dependent killing of CDH3 prostate cancer cells in vitro and significantly suppressed tumor growth in in vivo metastatic prostate cancer models. Likewise, CDH3-targeted CAR T cells specifically recognized and lysed CDH3-expressing prostate tumor cells while sparing CDH3-negative cells, leading to tumor regression and improved survival in mouse models, especially when combined with PD-1 checkpoint blockade. ConclusionsCDH3 is a key marker and functional driver of basal-like prostate cancer. Therapeutic strategies leveraging CDH3, including ADCs and CAR T cells, demonstrate strong preclinical efficacy, supporting the development of CDH3-targeted treatments to overcome resistance in aggressive prostate cancer.

cancer biology↗

bioRxiv2025

Complete and haplotype-resolved maps of genomic and epigenetic discordance in monozygotic twins

Telomere-to-telomere (T2T) genome assemblies are indispensable for accurate detection of genetic variation and for resolving complex repetitive regions. Monozygotic (MZ) twin pedigrees provide a powerful model to investigate de novo mutations (DNMs), however, comprehensive, haplotype-resolved analyses of structural variation (SV), allele-specific inheritance in complex regions, and DNA methylation in diploid human genomes remain limited. Here, we generated complete, haplotype-resolved T2T assemblies for two female twins (C33 and C35) from a Han Chinese pedigree by integrating complementary, state-of-the-art sequencing technologies. The resulting T2T-C33 and T2T-C35 assemblies are highly contiguous and complete, with Genome Continuity Inspector (GCI) scores of 74.94 (maternal) and 77.94 (paternal), and consensus quality values (QV) >75 (k = 21). We comprehensively cataloged 62 inter-twin single-nucleotide variants (SNVs), 15 small indels, and identified both shared and private DNMs, revealing nascent genomic divergence between the MZ twins. Focused interrogation of complex regions uncovered pronounced haplotype-specific length polymorphisms and structural heterogeneity within centromeric higher-order repeat (HOR) arrays. Notably, we observed extensive HOR copy-number variation between haplotypes, including a large copy-number difference on maternal chromosome 18, underscoring dynamic HOR array evolution even among genetically identical individuals. Concurrently, genome-wide DNA methylation profiling delineated allele-specific epigenetic variation that may contribute to phenotypic discordance. Together, these high-quality, diploid T2T genomes from a Han Chinese pedigree provide a valuable resource for population-aware genomics and reveal fine-scale, haplotype-specific divergence in MZ twins. Our results advance understanding of repeat dynamics, centromeric architecture, epigenetic variation and the spectrum of human genomic variation at single-base and structural scales.

genomics↗

bioRxiv2025

Neuroinflammatory Stress Preferentially Impacts Synaptic MAPK Signaling and Mitochondria in Excitatory Neurons

BackgroundUnderstanding synapse-specific effects of neuroinflammation can provide mechanistic and therapeutically relevant insights across the spectrum of neurological diseases. MethodsWe applied neuron-specific proteomic biotinylation in vivo, differential centrifugation of brain for crude synaptosome enrichment (P2 fraction) and mass spectrometry (MS) analysis of biotinylated proteins to derive native-state proteomes of Camk2a-positive neurons and their corresponding P2 synaptic compartments. Next, in an in vivo model of systemic lipopolysaccharide (LPS) dosing, we examined the effects of neuroinflammation on whole neuron and synaptic compartments using a combination of MS, network analysis, confirmatory biochemical and ultrastructural assays and integrative approaches across our mouse-derived and existing human datasets. ResultsUltrastructural and biochemical analyses of P2 fractions verified enrichment in synaptic elements, including synaptic vesicles and mitochondria. MS of biotinylated proteins from Camk2a-specific bulk brain homogenates (whole neuron) and P2 fractions (synaptosome) showed enrichment of >1000 proteins, consistent with neuron-specific biotinylation, also confirmed by immunofluorescence microscopy. Camk2a-specific synaptic proteome revealed molecular signatures related to mitochondrial function, synaptic transmission, protein translation. LPS-treated mice displayed body weight loss and neuroinflammation, characterized by glial activation, increased pro-inflammatory cytokine levels and upregulated expression of Alzheimers disease (AD)-related microglial genes. LPS-induced neuroinflammation exerted distinct effects on the synaptic proteome, including increased mitochondrial and reduced cytoskeletal-synaptic proteins, while suppressed synaptic MAPK signaling. Importantly, these changes were not observed at the whole neuron level, indicating unique vulnerability of the synapse to neuroinflammation. In line with synapse proteomic and signaling changes, LPS altered the ultrastructure of asymmetric synapses, suggesting dysregulation of excitatory neurotransmission. Co-expression network analysis of Camk2a neuronal proteins further resolved mitochondria- and synapse-specific protein modules, some of which were neuroinflammation-dependent. Neuroinflammation increased levels of a mitochondria-enriched module, and decreased levels of a pre-synaptic vesicle module, without impacting a post-synaptic membrane module. LPS-dependent mitochondrial and LPS-independent post-synaptic modules in mouse neurons mapped to post-mortem human AD brain proteomic modules which were decreased in cases with AD dementia and positively correlated to cognitive function, including pro-resilience markers for AD. ConclusionOur findings using native-state proteomics of Camk2a neurons combined with synaptosome enrichment identify proteome-level mechanisms of early synaptic vulnerability to neuroinflammation relevant to AD.

neuroscience↗

bioRxiv2025

Single-cell proteomics: a powerful new tool to study kidney cell heterogeneity

BackgroundTechnological advancements in protein mass spectrometry have significantly enhanced analytical sensitivity and throughput, enabling single-cell proteomics by mass spectrometry (SCP-MS) to become reality. SCP-MS allows high-resolution analysis of cellular heterogeneity and function, bypassing bulk analysis limitations. Here, we used SCP-MS to document at the protein level the cellular diversity of mouse kidney cells. We further focused SCP-MS on low abundance distal convoluted tubule (DCT) cells that are essential for body electrolyte homeostasis and blood pressure control. MethodsMouse kidney cells were isolated via enzymic digestion and single-cells isolated using fluorescence- activated cell sorting (FACS). DCT cells were isolated from mice with GFP expression specifically in the DCT (parvalbumin-GFP) using a similar workflow. SCP-MS analyses was performed using an Orbitrap Ascend Mass Spectrometer with a FAIMS Pro Duo interface coupled with Wide ISolation window High-resolution MS1 Data Independent Acquisition (WISH-DIA). Spectronaut was used for protein identification and quantification, and cell-type annotation and clustering were performed in Python by scanpy implementation. ResultsWorkflow benchmarking using Hela cells confirmed a successful SCP-MS setup. From 768 single mouse kidney cells, SCP-MS identified 2626 unique proteins. Computational approaches resolved various nephron segments, including distinct DCT populations. From enriched DCT cells, 1912 proteins were identified, enabling classification of three populations - DCT1, DCT2, and a proliferative subset (ProLIF) that represented a transient state between DCT1 and DCT2. ProLIF cells had elevated abundance of the sodium-chloride cotransporter NCC and represented approximately 33% of DCT cells, substantially exceeding previous transcriptomic estimates of ProLIF cell frequency (~0.1%). High proliferation (~39%) of DCT cells was confirmed using immunohistochemistry. ConclusionsThis SCP-MS analysis of mouse kidney uncovered significant cellular heterogeneity not captured effectively using transcriptomics. Despite imitations in proteome depth and throughput, SCP-MS provides a powerful approach for investigating kidney cellular dynamics at the protein level. Key pointsO_LIWe developed a single-cell proteomics by mass spectrometry (SCP-MS) workflow identifying up to ~2,000 proteins in a single mouse kidney cell C_LIO_LISCP-MS identified a substantial proliferative subset of cells in the DCT underestimated by transcriptomics C_LIO_LISCP-MS could inform on clinical strategies for kidney disease monitoring, offering robust indicators of tubular injury or regeneration. C_LI

biochemistry↗

bioRxiv2025

Crosslinking of Lipid Nanoparticles Enhances the Delivery Efficiency and Efficacy of mRNA Vaccines

Lipid nanoparticles (LNPs) have enabled the effective delivery of RNA therapeutics and mRNA vaccines. However, their broader applications are limited by the suboptimal stability and endosomal escape efficiency. Here, we present an easy-to-adopt post-assembly crosslinking approach to enhance the structural and functional stability of mRNA LNPs. By leveraging a series of cholesterol derivatives and crosslinking methods, we induce crosslinks of the lipid components following mRNA LNP assembly to form the crosslinked LNPs (cLNPs). We systematically evaluated crosslinking parameters and identified optimal conditions that enhance both the physical stability and transfection efficiency of cLNPs. Our findings demonstrate that cLNPs exhibit improved structural integrity under storage and lyophilization conditions, as well as increased extracellular stability and endosomal escape efficiency, resulting in improved performance of mRNA LNPs both in vitro and in vivo. This crosslinking strategy represents a critical advance in LNP engineering, enabling more resilient LNPs and broadening the applicability of LNP-based therapies for gene therapy and vaccine delivery. Our work lays the foundation for developing next-generation LNPs with superior stability and delivery efficiency, broadening the impact of RNA therapeutics and vaccines.

bioengineering↗

bioRxiv2025

Buffer Valency Engineering Enables High-concentration and Shelf-stable DNA Transfection Particles for Viral Vector Production

Cost-effective and scalable production is critical for advancing the clinical translation of adeno-associated virus (AAV)-mediated gene therapy. The widely used transient transfection method using plasmid DNA (pDNA)-loaded transfection particles for AAV production faces technical challenges due to instability of the particles and the concentration limits for particle preparation, hindering reproducibility and scalability. Here, we report a streamlined and scalable strategy to generate shelf-stable, highly concentrated pDNA/poly(ethylenimine) (PEI) transfection particles. By incorporating trivalent citrate ions in the dilution buffers, we kinetically modulate electrostatic complexation to achieve uniform nanoparticle assembly and prevent aggregation at high concentrations. This enables a tenfold increase in pDNA concentration in stabilized transfection particles from a typical range of 10-20 g/mL to 200 g/mL, while reducing the required dosing volume from 5-10% to 0.5% of the cell culture medium. The particle assembly process is robust to changes in mixing scale and timing and is compatible with standard workflows. We demonstrate equivalent AAV production efficiencies to standard methods and consistent performance in various production scales, which confirms the practical utility of this assembly method in developing robust, scalable, and cost-effective AAV manufacturing processes.

biochemistry↗

bioRxiv2025

Tumour-driven lipid accumulation in oenocytes reflects systemic lipid alterations

Cancer cachexia is a multifactorial syndrome characterized by systemic metabolic dysfunction, including liver steatosis. In this study, we examined the role of larval oenocytes - hepatocyte-like cells, in a Drosophila model of cancer cachexia. We found that oenocytes in tumour-bearing larvae accumulate lipid droplets in response to tumour-secreted signals, Gbb and ImpL2. This lipid accumulation reflects systemic changes in lipid metabolism, responding to lipid metabolism manipulations in either the fat body or the muscle. Disrupting lipid synthesis (via FASN1 and DGAT1), storage (via Lsd2), or trafficking (via apolipoproteins) in these tissues significantly modulated lipid droplet accumulation in oenocytes. Moreover, oenocyte-specific knockdown of FASN1 reduced their lipid content and non-autonomously affected lipid droplet size in the fat body, suggesting cross-regulatory interactions between these tissues. Cachectic oenocytes also exhibited altered signaling profiles, characterized by reduced PI3K and elevated Wnt and Ecdysone activity. Enhancing PI3K signaling through Akt overexpression restored oenocyte size and reduced lipid levels; however, these changes did not significantly improve muscle integrity. Together, our data suggests that dynamic exchange of lipids occur between the fat body, oenocytes and the muscle during cancer cachexia. While the fat body and muscle lipid pools are key regulators of muscle integrity, oenocytes - despite their metabolic responsiveness, do not appear to play an active role in preserving muscle function during cachexia.

developmental biology↗

bioRxiv2025

Impact of G-tract RNAs and the DHX36 helicase on stress granule composition and formation

Stress granules are RNA-protein condensates that form in response to an increase in untranslating mRNPs. Stress granules form by the condensation of mRNPs through a combination of protein-protein, protein-RNA, and RNA-RNA interactions. Several reports have suggested that G-rich RNA sequences capable of forming G-quadruplexes promote stress granule formation. Here, we provide three observations arguing that G-tracts capable of forming rG4s do not promote mRNAs partitioning into stress granules in human osteosarcoma cells. First, we observed no difference in the accumulation in stress granules of reporter mRNAs with and without G-tracts in their 3 UTRs. Second, in U-2 OS cell lines with reduced DHX36 expression, which is thought to unwind G-quadruplexes, the partitioning of endogenous mRNAs was independent of their predicted rG4-forming potential. Third, while mRNAs in stress granules initially appeared to have a higher probability of forming rG4s than bulk mRNAs, this effect disappeared when rG4 motif abundance was standardized by mRNA length. However, we observe that in a G3BP1/2 double knockout cell line, reducing DHX36 expression rescued stress granule-like foci formation. This indicates that DHX36 can limit stress granule formation, potentially by unwinding trans rG4s, or limiting other intermolecular RNA-RNA interactions that promote stress granule formation. Key Points- G-tract RNAs with quadruplex forming potential in an mRNA do not affect its partitioning into stress granules - mRNA partitioning to stress granules is dependent on mRNA length rather than rG4-forming potential - DHX36, a DEAH-box helicase that unwinds RNA G-quadruplexes, limits stress granule formation O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=72 SRC="FIGDIR/small/659950v1_ufig1.gif" ALT="Figure 1"> View larger version (16K): org.highwire.dtl.DTLVardef@a24c58org.highwire.dtl.DTLVardef@1422ce9org.highwire.dtl.DTLVardef@1929677org.highwire.dtl.DTLVardef@d48fa6_HPS_FORMAT_FIGEXP M_FIG C_FIG

cell biology↗

bioRxiv2025

ROMO1 loss in cholinergic neurons induces mitochondrial ultrastructural damage and ALS-like neuromuscular degeneration

Mitochondrial dysfunction is strongly associated with the pathogenesis of amyotrophic lateral sclerosis (ALS), a neurodegenerative disorder characterized by progressive motor neuron degeneration. However, it remains obscure whether mitochondrial abnormities are sufficient to drive ALS development. Here, we show that selective depletion of reactive oxygen species modulator 1 (ROMO1), an inner mitochondrial membrane-delimited protein, in cholinergic neurons leads to adult-onset, progressive locomotor deficits in mice that closely resemble ALS pathology. ROMO1 ablation in cholinergic neurons induces ALS-like neuromuscular degeneration, as evidenced by age-dependent motor neuron loss, axon degeneration, disrupted cholinergic transmission, neuromuscular junction denervation, and the resultant muscle atrophy. Notably, ROMO1 loss induces early and progressive mitochondrial cristae deformation in motor neurons, preceding the onset of ALS-like syndromes. Our findings support that mitochondrial impairment in vulnerable motor neurons is a sufficient contributor to ALS etiology, positioning mitochondria as a potential therapeutic target.

neuroscience↗

bioRxiv2025

Chinmo defines the region-specific oncogenic competence in the Drosophila central nervous system

While genetic mutations can promote hyperplastic growth, they do not always result in oncogenic outcomes. We and others have previously identified the transcription factors Nerfin-1 and Lola as inhibitors of dedifferentiation. Here, we investigate how the oncogenic potential of dedifferentiation varies across different neural lineages in the Drosophila central nervous system (CNS). We found that Nerfin-1 inactivation causes tumorigenic phenotypes in the central brain (CB) and the ventral nerve cord (VNC) but not the optic lobes (OLs). In contrast, Lola inactivation leads to tumour overgrowth specifically in the OLs. We identify Chinmo, a temporal transcription factor, and its regulation by ecdysone signalling as key determinants of the oncogenic competence in different regions of the brain, influencing the tumorigenic outcome of dedifferentiation. This work provides a fundamental framework to understand how oncogenic competence arises beyond genetic mutations. Significance statementIn the CNS, the same tumorigenic mutation has been shown cause differential oncogenic outcomes in different regions of the brain. The mechanism underlying this phenomenon remains largely unknown. We have previously demonstrated that malignant brain tumours can be induced via neuronal dedifferentiation in the Drosophila CNS. Here, we demonstrate that dedifferentiated neural stem cells drive tumorigenesis in a region-specific manner, accounted for by region-specific expression and regulation of temporal factors by cell-intrinsic and hormonal signals. Together, this work extends our understandings of how brain regionalisation can be a constraint to oncogenesis.

developmental biology↗

bioRxiv2025

The blood-brain barrier regulates brain tumour growth specifically via the SLC36 amino acid transporter Pathetic in Drosophila

Tumours adapt their metabolism to sustain increased proliferation, rendering them particularly vulnerable to fluctuations in nutrient availability. However, the role of the tumour microenvironment in modulating sensitivity to nutrient restriction (NR) remains poorly understood. Using a Drosophila brain dedifferentiation neural stem cell (NSC) tumour model induced by Prospero (Pros) inhibition, we show that tumour sensitivity to NR is governed by the perineural glial (PG) cells of the blood-brain barrier (BBB), a major component of the glial niche surrounding the tumour. We identify the SLC36 amino acid transporter Pathetic (Path) as a crucial regulator of nutrient sensitivity. Under NR, while wildtype buffers against low nutrient levels by upregulating Path, tumour glia downregulate Path. Furthermore, Path is specifically required by the tumour (but not wildtype) PG; its downregulation causes reduced proliferation of PG cells and, in turn, restricts NSC tumour growth. Path influences PG proliferation via the mTor-S6K pathway, and its expression is controlled by Ilp6 levels and the Insulin/PI3K pathway. Overexpression of Path is sufficient to counteract the inhibitory effects of NR on tumour growth. These findings suggest that Path levels at the BBB play a key role in determining tumour sensitivity to NR.

cancer biology↗

bioRxiv2025

Vib2Sound: Separation of Multimodal Sound Sources

Understanding animal social behaviors, including vocal communication, requires longitudinal observation of interacting individuals. However, isolating individual-level vocalizations in complex environments is challenging due to background noise and frequent overlaps of coincident signals from multiple vocalizers. A promising solution lies in multimodal recordings that combine traditional microphones with animal-borne sensors, such as accelerometers and directional microphones. These sensors, however, are constrained by strict limits on weight, size, and power consumption and often lead to noisy or unstable signals. In this work, we introduce a neural network-based system for sound source separation which leverages multi-channel microphone recordings and body-mounted accelerometer signals. Using a dataset of zebra finches recorded in a social setting, we demonstrate that contact sensing largely outperforms conventional microphone-array recordings. By enabling the separation of overlapping vocalizations, our approach offers a valuable tool for studying animal communication in complex naturalistic environments.

animal behavior and cognition↗

bioRxiv2025

Arcuate Fasciculus: Evolutionary Convergence in Marmosets and Humans

The marmoset is a highly vocal platyrrhine monkey that shares key anatomical and functional features with humans, offering insights into the evolution of brain connectivity. Although similarities in vocalization features with humans have been reported, it remains unclear whether marmosets possess an arcuate fasciculus (af) homolog. This study delineated white matter tracts in marmosets, establishing homologies with those observed in other primates, including macaques, chimpanzees, and humans. The presence of an af homolog in marmosets was confirmed by tracer and ultra-high-resolution diffusion magnetic resonance imaging datasets. We compared cortical connectivity patterns across these species and found the af in marmosets terminates in the ventral frontal cortex, with greater similarity to humans than macaques. Furthermore, we linked af connectivity with vocalization-related brain activation in both marmosets and humans. Collectively, our findings suggest that a dorsal pathway, which emerged early in marmoset evolution, has evolved convergently with humans, despite their distant phylogenetic kinship.

neuroscience↗

bioRxiv2025

Mapping the coupling between tract reachability and cortical geometry of the human brain

The study of cortical geometry and connectivity is prevalent in research on the human brain. However, these two aspects of brain structure are usually examined separately, leaving the essential connections between the brains folding patterns and white matter connectivity unexplored. In this study, we aimed to elucidate fundamental links between cortical geometry and white matter tract connectivity. We developed the concept of tract-geometry coupling (TGC) by optimizing the alignment between tract connectivity to the cortex and multiscale cortical geometry. Specifically, spectral analyses of the cortical surface yielded a set of geometrical eigenmodes, which were then used to explain the locations on the cortical surface reached by specific white matter tracts, referred to as tract reachability. In two independent datasets, we confirmed that tract reachability was well characterized by cortical geometry. We further observed that TGC had high test-retest ability and was specific to each individual. Interestingly, low-frequency TGC was found to be heritable and more informative than the high-frequency components in behavior prediction. Finally, we found that TGC could reproduce task-evoked cortical activation patterns. Collectively, our study provides a new approach to mapping coupling between cortical geometry and connectivity, highlighting how these two aspects jointly shape the connected brain.

neuroscience↗

bioRxiv2025

CDK12 condensation in nuclear speckles confers sensitivity to cyclin K molecular glue degraders

Transcriptional machinery often operates in the configuration of biomolecular condensates, presenting emerging opportunities to develop therapies targeting transcriptionally addicted malignancies. In this study, we discovered that cyclin-dependent kinase 12 (CDK12), a key regulator of transcription elongation, formed dynamic liquid-like droplets within nuclear speckles, a process enhanced by high CDK12 expression and CDK12 inhibitor treatment. Recognizing this unique property, we rationally designed ZSQ253, a modular cyclin K molecular glue degrader that coupled binding moieties for both CDK12 and DDB1. By promoting phase separation-driven assembly and subsequent proteasomal degradation of the CDK12-cyclin K complex, ZSQ253 interfered with oncogenic transcription and demonstrated potent antitumor capability. Further structure-guided optimization yielded HQY1428, a derivative with a multi-site gluing effect and improved drug activity. Given the frequent genomic co-amplification of CDK12 and ERBB2 across diverse human cancers, we showed that combining HQY1428 with HER2 inhibitors provided synergistic therapeutic benefits. Collectively, our findings establish CDK12 condensation within nuclear speckles as an exploitable vulnerability, and introduce a novel mechanism of action underlying cyclin K molecular glue degraders.

cancer biology↗

bioRxiv2025

AI-Driven Efficient De Novo design of GLP-1RAs with Extended Half-Life and Enhanced Efficacy

Peptide drug has revolutionized modern therapeutics, offering novel treatment avenues for various diseases. However, low efficacy, time consumption, and high cost hinder peptide drug design and discovery. We present an efficientin approach that integrates deep learning-based protein design with efficient functional screening, enabling the rapid design of biotechnologically important peptides with improved the stability and efficacy. We designed 10,000 de novo GLP-1 receptor agonists (GLP-1RAs), of which 60 met the stability, efficacy, and diversity criteria in the vitral functional screening. In vitro validation revealed a 62% success rate, while in vivo experiment demonstrated that two designed GLP-1RAs exhibited significantly extended half-lives, approximately three times longer than Semaglutide. In diabetic and obesity mouse models, the most competitive candidate showed superior therapeutic efficacy over Semaglutide. Our AI driven peptide design pipline integrats protein design, functional screening, and experiment validation, reducing the number of iterations required to find novel peptide candidates. The entire process, from design to screening, can be completed in a single cycle within two weeks.

bioengineering↗