Biology subjects

Zhao, C.

Publications and source records attributed to Zhao, C..

At least 19 recordsLinked to original sources

bioRxiv2026

Targeting the sinusoidal-to-capillarized transition to normalize vessels in breast cancer liver metastasis

Liver metastases exhibit marked resistance to anti-VEGF anti-angiogenic therapies, yet the underlying mechanisms remain poorly understood. By combining single-cell RNA sequencing of human breast cancer liver metastases with a liver sinusoidal endothelial cell (LSEC) lineage-tracing mouse model, we show that liver metastasis-associated vessels originate from Lyve1+ sinusoidal endothelial cells but undergo a sinusoidal-to-capillarized transition as they grow into the metastatic nodule, characterized by enhanced inflammatory signaling, loss of sinusoidal markers, and acquisition of capillarized markers. We functionally perturbed each of these abnormalities individually. Unexpectedly, ectopic Lyve1 overexpression paradoxically enhanced hyaluronan and other extracellular matrix deposition in the tumor microenvironment, suggesting a compensatory mechanism. Similarly, targeting TNF signaling, a hallmark inflammatory pathway in liver metastasis, dampened the inflammatory state of metastasis-associated vessels but promoted metastatic outgrowth, indicating that TNF blockade is not a viable strategy for metastasis treatment. Strikingly, inhibition of the capillary tip cell marker Esm1 improved vessel perfusion and enhanced chemotherapeutic drug delivery, two defining features of vascular normalization. Pan-cancer analysis further revealed that Esm1 expression is restricted to tip cells and correlates with poor patient prognosis, positioning Esm1 as a candidate target for promoting vascular normalization. Collectively, our findings reveal the vascular plasticity of liver metastasis and, although not all abnormal features are actionable, we propose targeting Esm1 to inhibit sprouting angiogenesis and to normalize dysfunctional liver metastasis vessels, offering a promising therapeutic avenue for liver malignancies.

cancer biology↗

bioRxiv2026

Distinct spatial organization of human hippocampal perfusion and microvascular pulsatility across age

Arterial pulsation can modulate neuronal activity through mechanosensitive pathways in mice, but whether vascular pulsatility is similarly related to neural organization in humans remains unclear. Here, using 7T MRI in 35 young and older adults, we mapped cerebral perfusion and microvascular volumetric pulsatility index (mvPI) across the unfolded hippocampus and integrated these measures with histological, electrophysiological, and transcriptomic reference maps. Perfusion and mvPI exhibited distinct spatial organization and age-related differences. In young adults, perfusion preferentially corresponded to parvalbumin-interneuron density and beta/gamma distribution, whereas mvPI was associated with theta-band distribution. A number of associations detected in young adults were not detected in the older adults maps. Spectral simulations showed qualitative similarity to heartbeat-related phase resetting as one possible explanation for the mvPI-theta relationship, while spatial transcriptomics revealed perivascular enrichment of PIEZO2, a candidate mediator of vascular mechanotransduction. Together, these findings establish perfusion and tissue-level pulsatility as distinct dimensions of human hippocampal vascular organization and provide a framework for investigating their relationships with neural dynamics across aging.

neuroscience↗

bioRxiv2026

Differential Survival of Retinal Ganglion Cell Types Following Branch Retinal Vein Occlusion

Background. Branch retinal vein occlusion (BRVO) is a common cause of retinal ischemia, rapidly inducing neuronal death and often producing irreversible vison loss. Interventions such as thrombolysis restore perfusion but cannot recover neurons already lost, and most patients retain permanent visual deficits. Identifying which neuronal populations die, and how rapidly, may therefore reveal timing windows and targets for neuroprotective treatment. Aims. To determine the timing of pathology, we evaluated vascular changes and neuronal degeneration in a photocoagulation-induced mouse model of BRVO. To determine differences in retinal ganglion cells (RGC) type susceptibility, we tracked the survival of RGC subpopulations using immunofluorescence markers. Methods. BRVO was induced by retro-orbital injection of Rose Bengal and FITC dextran followed by targeted photocoagulation of a branch vein. Obstruction was confirmed by fluorescein angiography and vascular damage was viewed by optical coherence tomography angiography. Cell survival was quantified by immunostaining at 3, 7, and 14 days using RNA Binding Protein with Multiple Splicing (RBPMS, pan RGC). RGC subtype markers for intrinsically photosensitive RGCs (TBR2), RGCs (SPP1), ON OFF direction-selective RGCs (CART), and F RGCs (FOXP2) were used to examine differences in survival across RGC types. Results. Occlusion selectively disrupted the superficial and intermediate capillary plexus, sparing the deep plexuses. All examined neuronal populations were significantly depleted, but with markedly different magnitude and timing. RGCs and F RGCs were the most vulnerable subtypes, both approaching complete loss by one week, whereas intrinsically photosensitive RGCs were the most resistant. Conclusion. RGCs, typically resilient in glaucoma, trauma, and excitotoxicity, were markedly vulnerable in BRVO, indicating that RGC subtype susceptibility is insult-specific.

neuroscience↗

bioRxiv2026

Auto-SMART: a web-based platform for repair-template design for CRISPR/Cas9 homology-directed editing

Genome engineering via CRISPR/Cas9 has been widely applied in biological research and developing gene therapies. However, genome editing flexibility is strongly constrained by the position and quality of guide RNAs relative to the intended edit: desired edit sites are frequently positioned far from efficient protospacer adjacent motifs (PAM). SMART (Silently Mutate And Repair Template) technique relaxes this constraint by silently mutating the gap between the Cas9 cut and the edit in the repair donor, preserving the encoded protein while expanding the usable guide-RNA design space. Designing such donors manually requires coordinated reasoning over transcript structure, reading frame, candidate protospacers, codon usage, homology arms and the expected edited allele. Auto-SMART is a web platform that automates this process for CRISPR/Cas9-mediated insertion, deletion, point mutation and sequence replacement in coding exons and in introns. Users choose a genome assembly, search a gene symbol or transcript, select a modification site and select an editing operation. Auto-SMART finds nearby Streptococcus pyogenes Cas9 (SpCas9) guides, ranks them by predicted on-target activity and cut-to-edit geometry, builds repair templates with synonymous substitutions, automatically recodes residual on-target sites to prevent re-cutting and renders interpretable panels for the target locus, donor template and locus after HDR, together with a downloadable report and a Primer3 genotyping handoff. The web application for Auto-SMART is freely available at https://auto-smart.chuanping.org/.

genetics↗

bioRxiv2026

Loss of 20-HETE-GPR75 Signaling Impairs Renal Hemodynamic Adaptation in Sepsis

BACKGROUD: Sepsis is a life-threatening condition characterized by a dysregulated host response to infection. It frequently precipitates sepsis-associated acute kidney injury (SA-AKI). 20-Hydroxyeicosatetraenoic acid (20-HETE), is an important regulator of vascular tone. We aim to investigate whether disruption of 20-HETE signaling contributes to septic renal dysfunction. METHODS: In the animal study, Sprague-Dawley rats underwent the cecal ligation and puncture to induce sepsis and SA-AKI. Comprehensive cardiac and renal ultrasonographic assessments were performed in all animals. Blood samples were collected for biochemical assessment of renal function and quantification of circulating 20-HETE levels. Kidney tissues were harvested for analysis of the 20-HETE-GPR75-PLC/PKC signaling pathway and histological evaluation by hematoxylin and eosin staining. In the clinical study, circulating 20-HETE levels and renal hemodynamic parameters derived from point-of-care ultrasound were evaluated in healthy individuals and patients with sepsis. Venous blood samples were collected for quantification of serum 20-HETE concentrations. RESULTS: Firstly, CLP-induced septic rats produced a biphasic hemodynamic response, with early compensation followed by late cardiovascular decompensation, accompanied by progressive renal dysfunction, tubular injury, depletion of circulating and renal 20-HETE, and reduced renal GPR75 expression. Secondly, Restoration of 20-HETE signaling with a pharmacological analog improved systemic and renal hemodynamics, increased MAP, reduced serum creatinine, blood urea nitrogen, and ameliorated renal pathological injury. These protective effects were partially attenuated by PKC inhibition, implicating PKC-dependent signaling in 20-HETE-mediated renal protection. Moreover, in septic patients, circulating 20-HETE concentrations were reduced compared with healthy individuals and were lowest among those with SA-AKI, paralleling alterations in renal hemodynamic parameters measured by point-of-care ultrasound. CONCLUSIONS: Sepsis-associated depletion of 20-HETE accompanies impaired systemic and renal hemodynamic regulation and is associated with the development of SA-AKI. Restoration of 20-HETE signaling improves hemodynamic function and attenuates renal injury, potentially involving GPR75-associated PLC/PKC signaling. KEY WORDS: sepsis-associated acute kidney injury, 20-hydroxyeicosatetraenoic acid, GPR75, renal hemodynamics, point-of-care ultrasonography

pathology↗

bioRxiv2026

Direct electrochemical cortisol detection via transition-resolved interrogation at a defect-engineered graphene interface

Continuous tracking of cortisol is central to understanding human stress physiology, yet direct electrochemical detection without biological receptors remains challenging despite its promise for stable and dynamic sensing. Electrochemical reduction of cortisol typically occurs at highly negative potentials, where parasitic interfacial currents, hydrogen evolution, and substantial capacitive background overlap with the cortisol reduction signal, preventing accurate quantification. We overcome this barrier through an integrated material-measurement strategy combining a defect-engineered graphene interface with a transition-resolved interrogation (TRI) measurement strategy. First, we engineered polybenzimidazole-derived laser-induced graphene containing nitrogen-rich defects while suppressing oxygen-derived functionalities which reduced parasitic background currents within the same cathodic potential regime. Next, we designed TRI to leverage differences in the time-dependent evolution of overlapping cathodic processes to isolate a localized cortisol-associated electrochemical reduction transition. Derivative-domain projection coupled with background estimation enables its reliable quantification. This integrated sensing architecture enables sensitive, selective, and dynamic cortisol detection in both artificial and biological interstitial fluids at low nanomolar concentrations. The response remains reproducible across physiologically relevant variations in pH, ionic strength, temperature, and repeated cycling. Together, these capabilities provide a basis for continuous electrochemical cortisol monitoring for future study of stress physiology.

bioengineering↗

bioRxiv2026

TLS-Sim: An Open-Source Toolkit for Individualized Transcranial Light Stimulation Modeling with Deep-Learning-Accelerated Simulation

Background: Accurate modeling of photon transport through the heterogeneous tissues of the human head is important for individualized transcranial light stimulation (tLS). However, high-precision Monte Carlo (MC) simulations can be computationally demanding, and existing general-purpose simulation tools do not provide an integrated, brain-oriented workflow. Methods: We developed TLS-Sim, a framework that integrates subject-specific head-model construction, multiple light-source configuration, GPU-accelerated MC photon simulation, visualization, and intracranial dosimetric analysis. Its central feature is a deep-learning-accelerated MC pathway, termed Flux-to-Flux, that estimates a high-photon-count energy-deposition field from a paired low-photon-count simulation. Sparse and full Monte Carlo fields were encoded using a three-dimensional variational autoencoder (VAE), and a three-dimensional velocity network was trained to transform the sparse latent representation toward the corresponding full-simulation representation. Results: We demonstrate the end-to-end toolkit and the Flux-to-Flux (F2F) pathway on a representative individualized head model evaluated across three source configurations. Relative to the sparse simulation, the learned prediction improved agreement with the full Monte Carlo reference by approximately +4 dB in signal-to-noise ratio on average, with correspondingly higher structural similarity and lower mean absolute error. Conclusions: TLS-Sim provides a modular, openly released computational framework for individualized tLS simulation and analysis. The Flux-to-Flux pathway is presented as an implemented capability that reduces the computational burden of high-photon-count simulation while improving spatial agreement with the full MC reference, most strongly at the low-fluence field boundaries where the sparse simulation is noisiest.

bioengineering↗

bioRxiv2026

Mnemonic Demand Reconfigures the Neural Architecture of Memory Encoding

Memory encoding must accommodate increasingly demanding loads, yet it remains unclear whether the underlying brain state is preserved or adaptively reconfigured as load increases. Here, we recorded EEG from 111 participants performing a visual memory task in which they encoded lists of real images ranging from 1 to 128 items, spanning low to high mnemonic demands, followed by a recognition test. We measured the similarity of set-size-specific interelectrode correlation patterns to a low-load reference. Similarity decreased monotonically as list length increased: patterns at set sizes below 4 remained close to the low-load configuration, whereas set sizes of 8 and above progressively shifted toward a distinct high-load configuration. This transition was captured by a single rotating eigenvector, revealing a low-dimensional trajectory from low- to high-load neural geometry. Smaller rotation angles from the low-load template predicted better memory performance at high loads. Together, these findings provide evidence for adaptive reconfiguration and reveal a structured, low-dimensional neural trajectory through which encoding architecture changes as mnemonic demands increase.

neuroscience↗

bioRxiv2026

Impaired proteostasis is an early feature of the diabetic heart in humans and mice

Diabetes and obesity increase cardiac lipid levels leading to cardiomyopathy and heart failure. We hypothesized that intermittent fasting would reduce cardiac lipid levels. Surprisingly, intermittent fasting increased myocardial triglyceride content, but rescued mortality and attenuated cardiomyopathy in mice overexpressing cardiomyocyte acyl-CoA synthetase 1 (MHC-ACSL1). Lipid overload caused cardiomyocyte accumulation of polyubiquitinated protein aggregates containing desmin, a scaffolding intermediate filament protein, which intermittent fasting prevented. Furthermore, intermittent fasting reversed elevated myocardial C16:0 ceramide content, and knockdown of ceramide synthase CerS5 and CerS6 reduced palmitate-induced protein aggregation, highlighting a role for C16:0 ceramides in this pathology. Conversely, impairing aggrephagy with cardiomyocyte-specific p62 ablation induced heart failure in mice fed a high-fat diet, with paradoxically reduced cardiac lipid content. Crucially, non-failing diabetic human hearts also exhibited protein aggregate pathology. Taken together, these results demonstrate that impaired proteostasis characterizes cardiomyopathy from cardiac lipid overload and identify a promising new therapeutic target for this condition.

molecular biology↗

bioRxiv2026

Live 4D monitoring of fungal infection in an ex vivo model of fungal keratitis reveals enhanced efficacy of combined PACK-CXL and natamycin treatment

Fungal keratitis (FK) is a major cause of visual impairment worldwide, yet the development and evaluation of new therapeutic approaches are limited by experimental models that inadequately reproduce deep stromal infection or permit longitudinal quantification of fungal burden. Here, we developed an ex vivo porcine corneal model of Aspergillus fumigatus keratitis that supports reproducible fungal invasion into the deeper corneal stroma and enables non-destructive monitoring of infection over time. Optimisation of the culture conditions demonstrated that reducing nutrient availability markedly promoted stromal penetration compared with nutrient rich conditions. Using a YFP expressing A. fumigatus strain combined with confocal microscopy and three-dimensional image analysis, fungal burden could be quantified repeatedly within individual corneas, allowing changes in established infection to be monitored longitudinally. We applied the model to evaluate natamycin and photoactivated chromophore for keratitis corneal cross-linking (PACK-CXL), alone and in combination. Both monotherapies significantly restricted the increase in fungal burden observed in untreated corneas, but neither consistently reduced pre-existing fungal burden. In contrast, combined natamycin and PACK-CXL treatment reduced fungal burden in all treated corneas and was significantly more effective than either treatment alone. Together, these findings establish a quantitative ex vivo platform for investigating invasive A. fumigatus corneal infection and demonstrate its ability to distinguish between treatments that restrict fungal expansion and those that reduce established fungal burden. The model provides a useful approach for the longitudinal evaluation and optimisation of therapeutic interventions for fungal keratitis.

microbiology↗

bioRxiv2026

Visual LLM-guided consensus spatial domain detection with L-STAR

Spatial domain detection is a central task in spatial transcriptomics, yet existing methods exhibit highly variable performance across datasets. We introduce L-STAR, a visual LLM-guided, consensus-based framework that leverages the visual reasoning capacity of large language models to adaptively rank and integrate spatial domain detection methods. L-STAR achieves robust and consistently improved performance, outperforming single spatial domain detection methods across diverse datasets.

bioinformatics↗

bioRxiv2026

High regeneration-associated stress defines a distinct HCC subgroup with therapeutically exploitable vulnerabilities

Background: To date, no precision oncology approach has been established for HCC. Despite the diverse underlying causes, HCC development exhibits a uniform pathophysiology characterised by chronic hyper-proliferation, resulting from hepatocyte apoptosis and compensatory liver regeneration. This chronic hyper-proliferative pressure, termed regeneration stress, drives genomic instability during HCC onset, yet its therapeutic potential remains poorly explored. This study aimed to identify targetable vulnerabilities tied to regeneration stress and establish clinically applicable markers for treatment stratification. Methods: Weighted gene co-expression network analysis (WGCNA) was applied on external bulk RNA-seq datasets to define a LIVer REgeneration Stress Signature (LIVRESS). The signature was functionally validated using HCC patient-derived organoids (HCC-Org), and vulnerabilities were mapped using mid-throughput drug screening, single-molecule and single-cell assays, and multi-omic integration. Results: High LIVRESS scores, characterised by enrichment in replication, mitotic and DNA damage repair pathways, identified a subset of HCC patients with aggressive disease and poorer survival across aetiologies. HCC-Org with high LIVRESS scores displayed exquisite sensitivity to multiple inhibitors of the checkpoint kinase ATR. Although HCC-Org models exhibited a baseline reduction in replication fork speed, sensitivity to ATR inhibitor (ATRi) was decoupled from replication fork dynamics and rather linked to intrinsic mitotic instability. ATR inhibition triggers mitotic failure and apoptosis in LIVRESSHigh HCC-Org. This killing effect was significantly potentiated by combining ATRi with PARPi or WEE1i. Multi-omic integration identified KPNA2 as a surrogate biomarker of ATRi sensitivity. Conclusion: Our findings demonstrate that a subset of HCC-Org, characterised by high liver regeneration-associated stress, is vulnerable to ATRi-based therapies. By focusing on a comprehensive regenerative stress model, we establish a framework to stratify HCC patients and implement biomarker-driven, ATR-based therapies for HCC patients with advanced disease. Impact and implications: Regeneration stress is a key factor that drives genomic instability in HCC, providing a basis for the LIVRESS to identify patients dependent on ATR-mediated checkpoints. These findings reveal a conceptual shift for researchers and trialists: ATRi efficacy is decoupled from replication fork dynamics and instead leverages mitotic fragility. Practically, the LIVRESS and its IHC surrogate marker (KPNA2) offer a scalable roadmap for physicians to improve patient stratification in ATRi-based precision oncology trials. While requiring prospective validation, these results pave the way toward biomarker-driven therapies for advanced HCC.

cancer biology↗

bioRxiv2026

CARD:Epi - Contextualizing Antimicrobial Resistance Determinants Using Deep Learning Language Models

Bacterial outbreak publications outline the key factors involved in the uncontrolled spread of infection. Such factors include the environment, pathogens, hosts, and antimicrobial resistance genes (ARGs). Individually, each paper published in this area gives a glimpse into the devastating impact drug resistant infections have on healthcare, agriculture, and livestock. When examined together, these publications provide contextual information on ARG transmission, from the discovery of new resistance genes to their dissemination to different pathogens, hosts, and environments. We have extracted this information from publications in PubMed by using the biomedical deep-learning language model, BioBERT. We trained BioBERT on two tasks: entity recognition to identify AMR-relevant terms (i.e., ARGs, taxonomy, environments, geographical locations, etc.) and relation extraction to determine which terms identified through entity recognition contextualize ARGs. By collating results from 204,094 antimicrobial resistance publications worldwide, we have generated interpretable results about the sources where genes are commonly found. To visualize the dataset, we have created two pipelines to analyze transmission patterns of ARGs across agriculture, environments, and human populations using a Confusogram and Uniform Manifold Approximation and Projection. Overall, we have taken a large-scale approach to collect antimicrobial resistance data from a commonly overlooked resource, i.e., the systematic examination of the large body of AMR literature and have visualized how scientific literature can be used to assess transmission patterns of ARGs.

genomics↗

bioRxiv2026

Divergent Entry, Convergent Trafficking in Defensin-Mediated Adenovirus Infection

Human -defensins can paradoxically inhibit or enhance human adenovirus (HAdV) infection. Defensin-resistant HAdVs exploit -defensins as molecular bridges to bypass canonical receptors, but whether this mode of entry redirects intracellular trafficking or contributes to defensin resistance remains unknown. Here we systematically compared intracellular trafficking after canonical receptor-mediated versus defensin-mediated entry using a defensin-resistant chimeric virus based on HAdV-C5 together with the human -defensins, HD5 and HNP1. Pharmacological perturbations, capsid mutants, and imaging revealed that both entry routes rapidly converge on common checkpoints. Defensin-mediated infection showed the same sensitivity to endosomal acidification blockade and comparable co-localization with early endosomes as canonical entry. Notably, the membrane-lytic activity of HD5 did not substitute for viral protein VI, indicating that endosomal escape still requires intact protein VI function. Downstream transport also overlapped, as defensin-mediated infection, like canonical entry, depended on dynein-driven microtubule transport and passage through the microtubule-organizing center (MTOC) for nuclear entry. Co-infection experiments further showed that defensin-sensitive and defensin-resistant viruses retain their intrinsic phenotypes within the same cell, indicating that defensin effects are virion-autonomous and determined at or before cell entry rather than by downstream trafficking. These findings support a model in which resistance to -defensin neutralization is governed by capsid features that control defensin binding and its effects on uncoating, not by access to an alternative intracellular pathway. Author SummaryHuman adenoviruses can be neutralized by -defensins, antimicrobial peptides that bind the capsid and block uncoating. Paradoxically, some defensin-resistant adenoviruses infect cells more efficiently in the presence of defensins, with the peptides acting as molecular bridges that allow the virus to attach independently of its normal receptors. We asked whether this alternative attachment route changes how adenoviruses traffic through the cell, which could contribute to defensin resistance. Using labeled viruses, inhibitors, capsid mutants, and receptor knockout cells, we compared trafficking after canonical versus defensin-mediated entry. Both routes converged at the earliest steps of infection, with identical endosomal acidification requirements, similar early endosome colocalization, and the same dependence on microtubule transport to the nucleus. Co-infection experiments demonstrated that sensitive and resistant viruses retained distinct phenotypes in the same cell, indicating susceptibility is set at or before entry. These results demonstrate that defensin effects are decided at the plasma membrane, where capsid features, not alternative entry routes, drive resistance. Adenovirus entry is flexible at the cell surface, where the virus can attach through multiple mechanisms, but narrows to a single, constrained trafficking route inside the cell. Our findings inform how defensin-resistant vectors may behave in defensin-rich tissues during gene therapy or virotherapy.

microbiology↗

bioRxiv2026

A cornea-specific role for the Aspergillus fumigatus carbon catabolite repressor, CreA, in tissue penetration and infection establishment

PurposeElucidate the influence of glucose metabolic pathways on A. fumigatus lung and corneal infection. MethodsThe A. fumigatus acuF and creA genes were deleted in an mcherry-expressing strain. The mutants were tested for alterations in radial growth, cell wall composition by fluorescence staining assays, and antifungal sensitivity through broth microdilution assays. Hyphal penetration of the strains through explanted porcine corneas was tracked by confocal microscopy using the mCherry signal. Virulence was evaluated in established models of invasive pulmonary aspergillosis (IPA) and fungal keratitis (FK) using C57BL/6J mice. ResultsDeletion of the A. fumigatus phosphoenolpyruvate carboxykinase (acuF) resulted in a dependency on exogenous glucose to support growth in vitro, but did not impact virulence in either the IPA or FK models. Loss of the carbon catabolite repressor CreA resulted in a broad dysregulation of carbon metabolic pathways and altered cell wall homeostasis. Surprisingly, whereas the{Delta} creA remained fully virulent in the lung, the mutant was unable to establish infection in the FK model. This in vivo phenotype corresponded to an inability of{Delta} creA to physically invade porcine corneal explants, which we attributed to a marked reduction in cell wall chitin content. ConclusionsGluconeogenesis is dispensable for A. fumigatus lung and corneal infection, suggesting tissue-derived glucose supports fungal growth in both environments. Loss of CreA disrupts glucose assimilation, its synthesis into chitin and, consequently, cell rigidity and hyphal invasion into the dense corneal stroma. Thus, CreA and other cell wall regulatory proteins may serve as targets for novel FK antifungals.

microbiology↗

bioRxiv2026

HaloMPNN: retraining ProteinMPNN on halophilic proteomes for salt-tolerant enzyme design

Machine learning-guided protein sequence redesign is now routinely used to optimize multiple properties relevant for protein engineering, most prominently thermostability and recombinant expression levels. Salt tolerance is a valuable property for "blue-biotechnology"- enabled biomanufacturing, yet no generative computational method exists to redesign proteins for increased salt tolerance. We hypothesized that a training dataset heavily biased toward salt-adapted proteomes would yield a model capable of designing proteins with halophilic properties. To test this, we retrained the sequence redesign model ProteinMPNN on proteins from "salt-in" extreme halophiles such as Haloarcula marismortui, a Dead Sea archaeon that grows optimally near 3-4 M NaCl, roughly six times the salinity of seawater, and accumulates molar concentrations of salts in its cytoplasm. Our model, HaloMPNN, redesigns non-halophilic proteins so that their properties shift towards those of natural halophilic proteins: lower predicted isoelectric point, greater surface acidity, and reduced surface and core hydrophobicity. Redesigning a broad range of non-halophilic proteins with SolubleMPNN, ProteinMPNN, and HyperMPNN shows that this shift is specific to HaloMPNN rather than a generic consequence of sequence redesign. HaloMPNN therefore offers both a route to designing candidate salt-tolerant enzymes and a means of identifying the characteristics that underlie halophilic adaptation.

bioengineering↗

bioRxiv2026

Evolving stem cell fate capacities and transcriptional priming in the developing olfactory epithelium

The olfactory epithelium of adult mammals contains two populations of stem cells that support its remarkable ability to regenerate neuronal and non-neuronal cell types throughout life. How olfactory epithelial cell types are established during development, however, is not well understood. Here, we use genetic lineage tracing and single-cell RNA sequencing of the perinatal mouse olfactory epithelium to construct a developmental trajectory consisting of multiple lineages. We identify transitional states and lineage relationships between individual cells and establish Ascl1+ cells as the primary multipotent progenitors in the perinatal olfactory epithelium. Further, Ascl1+ cells become progressively restricted in their cell fate capacity over developmental time and appear to be transcriptionally primed toward specific lineages. We also predict signaling pathways that may contribute to lineage plasticity and niche permissiveness. Together, these results contribute to our understanding of how cell-intrinsic and-extrinsic signals contribute to the establishment of a stem cell niche.

developmental biology↗

bioRxiv2026

Scalable single-cell isoform profiling with sequencing-by-expansion

Single-cell RNA sequencing has transformed our understanding of cellular systems, yet the reliance on short-read sequencing restricts analysis to gene-level quantification and obscures the immense biological diversity generated by alternative splicing. While long-read sequencing technologies can capture full-length RNA and resolve transcript isoforms, current platforms remain constrained by throughput and high per-base costs, rendering them impractical for modern million-cell applications. To address this critical limitation, we developed and optimized sequencing-by-expansion (SBX) chemistry for high-throughput single-cell RNA isoform profiling. Integrated within the AXELIOS 1 sequencing platform, SBX employs a unique biochemical conversion process that transforms complementary DNA into expanded surrogate high signal-to-noise polymers called Xpandomers which are sequenced via translocation through a dense nanopore array yielding over 9.5 billion reads in a two-hour run. To leverage this unique data type for long-read single-cell RNA isoform sequencing, we developed the Consensus UMI Deduplication using Longest Length (CUDLL) algorithm, which computationally consolidates variable-length raw SBX reads into single, high-fidelity consensus reads, elevating sequence accuracy to 99.83% and maximizing per transcript read length. We demonstrate that this consensus approach successfully captures the vast isoform diversity of single-cell libraries and enables the accurate measure of differential isoform expression across distinct cell types in peripheral blood mononuclear cells. Furthermore, SBX coupled with CUDLL efficiently resolves T-cell and B-cell receptor clonotypes directly from whole-transcriptome libraries without the need for VDJ-specific target enrichment. Ultimately, this work establishes SBX and the AXELIOS 1 as a transformative platform for high-scale single-cell isoform sequencing.

genetics↗