Biology subjects

Lam, A.

Publications and source records attributed to Lam, A..

At least 19 recordsLinked to original sources

bioRxiv2026

NanoPro - a Framework for Probiotically-Guided Nanovector Therapy

The 'antigen problem', i.e. the disconnect between the broad spectrum of therapeutic modes enabled by the range of viral and synthetic nanoparticles for which workable receptor-targeting architectures exist and the relative dearth of addressable target receptors on most tumors, is a long-standing challenge in targeted tumor nanotherapy. Here we provide in vitro-proof-of-principle for NanoPro, a platform for guiding nanoscale therapeutics to otherwise antigenically indistinct tumors via probiotically-released tumor cell-binding tags, ProTags. These consist of an orthogonal antigen (GFP or ALFA) fused to the heparin-binding domain of placenta growth factor-2 and serve as attachment receptors for ProTag-targeted nanovectors. ProTag release from Escherichia coli Nissle 1917 was facilitated by deletion of the membrane-tethering lpp gene, which increased ProTag supernatant concentrations 16.5-fold compared to the parental strain without impacting viability. HBD-dependent decoration with ProTag in bacterial supernatant followed by treatment with receptor-targeted aGFP- or aALFA-lentiviral vectors resulted in majority transduction of MC-38 and MDA-MB-468 tumor cell lines with antigen specificities of 96.5% and 98.5%, respectively, depending on the choice of nanobody. This ProTag-mediated interaction was notably robust, with selective transduction still observed at 625-fold supernatant dilution, warranting further exploration of the utility of the NanoPro system for tumor-restricted delivery of nanotherapeutics.

synthetic biology↗

bioRxiv2026

Lateral Propagation from Sensory Cortex Provides Differential Gain for Sensory Selection

Selectively responding to target stimuli while ignoring distractors is a fundamental cognitive ability. In neocortex, such selection is associated with enhanced responses to target stimuli relative to distractor stimuli, yet the gain mechanisms underlying this differential responsivity are poorly understood. Most proposed gain mechanisms emphasize modulation within the center response field of a stimulus. In contrast, here we investigate modulation of lateral propagation from the center response field as a potential gain mechanism during a whisker-based selective detection task in mice. Both target (Go) and distractor (NoGo) stimuli evoked activity that propagated beyond their center response fields, with substantially greater lateral propagation for target stimuli. Target-evoked propagation increased with task learning and, in expert mice, exhibited the spatiotemporal characteristics of a traveling cortical wave preferentially involving infragranular layers. Furthermore, we demonstrate that frontal cortex actively suppresses distractor-evoked sensory responses, including both center region activity and lateral cortical recruitment. Computational modeling showed that modulation of lateral excitation could reproduce key features of these experimental findings. Together, our results support the regulation of lateral cortical propagation as a mechanism for differentially amplifying behaviorally relevant sensory representations during sensory selection.

neuroscience↗

bioRxiv2026

Social and neuroendocrine phenotypes reprogrammed by endocrine-disrupting chemicals can be mitigated by Limosilactobacillus reuteri modulation of the gut microbiome-thyroid-oxytocin axis

IntroductionEnvironmental factors are increasingly implicated in the etiology of autism spectrum disorder (ASD). Polybrominated diphenyl ethers (PBDEs) are anthropogenic toxicants added as flame retardants to consumer products that have become ubiquitous environmental contaminants and disrupt thyroid hormone (TH) and neuroendocrine systems. We have previously shown that developmental PBDE exposure produces ASD-like traits with involvement of oxytocin (OXT)-thyroid hormone signaling. Limosilactobacillus reuteri (LR), a widely used probiotic bacterium, has been shown to improve social functioning and increase TH and OXT levels in murine models. Therefore, we tested the hypothesis that LR supplementation (LR) prevents PBDE-induced deficits in socioemotional behavior with concomitant modulation of TH signaling genes on hypothalamic OXT neurons. MethodsC57BL/6N mouse offspring were exposed to a commercial penta-mixture of PBDE congeners, DE-71, at an environmentally realistic concentration, 0.1 mg/kg/d (DE-71), or to corn oil vehicle (VEH/CON) via their mothers during gestation and lactation. Offspring received supplementation with LR ATCC PTA 6475 (107-108 CFU/mL, po) indirectly via the dam or continuation directly through adulthood. Unsupplemented controls were given saline. ResultsFecal microbiome analysis in offspring confirmed colonization of LR at postnatal day (P) 40 and depletion by P104. LR treatment increased plasma total thyroxine in DE-71 and plasma OXT in VEH/CON dams. In DE-71 offspring of both sexes, LR normalized deficient scores on social novelty preference and emotional recognition in adult females and males and deficient long-term social recognition memory (SRM) in adult DE-71 females; DE-71 males were normal. Reduced olfactory dishabituation between two social odors may partly explain the compromised socioemotional behavior produced by DE-71 in an LR-dependent manner. Multiplex RNA in situ hybridization performed on immunoreactive OXT-ergic neurons in the paraventricular hypothalamic nucleus (PVH) revealed significant upregulation of TH transporter monocarboxylate transporter 8 (Mct8) and downregulation of iodothyronine deiodinase 3 (Dio3) in DE-71 relative to VEH/CON females. This toxicant-induced reprogramming was prevented by probiotic treatment. DE-71 males expressed reduction in Mct8 and Dio3 transcripts on OXT-ergic neurons with minimal LR protection. In the female supraoptic nucleus (SON), Mct8 and Dio3 were downregulated by DE-71 and normalized in DE-71+LR; there were no group effects on transcript levels in male SON. Results of fecal 16S rRNA sequencing indicated reduced -diversity and altered {beta}-diversity in the gut bacterial community of female but not male DE-71 exposed offspring; most changes were correctable by LR. Alterations in taxa-level abundance caused by DE-71 and reversed by LR were observed in both sexes. These involved Bifidobacterium, Coprococcus, Desulfovibrio, Oscillospira, and Peptococcaceae in females and Desulfovibrionaceae, Rikenella, and Turicibacter in males. Exposed dams showed no detriment in - and {beta}-diversity while showing reduced abundance of several Firmicutes and Proteobacteria taxa that could be rescued by LR. The relative abundance of Lactobacillus was upregulated in DE-71 males and DE-71+LR males and dams. ConclusionsThese results indicate that developmental probiotic supplementation effectively mitigated organohalogen-induced ASD-like deficits in socioemotional behavior and partially corrected dysbiosis of gut bacterial communities in exposed offspring of both sexes. Concomitantly, PBDEs altered the expression of TH regulatory genes Mct8 and Dio3 in PVH OXT neurons in a sex-dependent manner, suggesting that TH regulation of OXT neuroendocrine cells may modulate the emergence of toxicant-induced ASD-relevant behavior. While LR reinstated normal behavioral outcomes in PBDE-exposed offspring of both sexes, coincident normalization of hypothalamic TH signaling transcripts occurred more broadly in females, indicating the existence of unique parallel processes influencing the preventive effects of LR on ASD-relevant behavioral deficits in both sexes.

neuroscience↗

bioRxiv2026

In vivo gene disruption and homology-directed repair in muscles and muscle stem cells using CRISPR/Cas9

Programmable endonucleases such as CRISPR/Cas9 provide powerful tools to edit mammalian genomes by engaging cellular mechanisms of DNA double-strand break (DSB) repair. CRISPR-catalysed homology-directed repair (CRISPR-HDR), though generally less efficient than other modes of DNA repair, holds particular promise to enable precise sequence replacement by targeted insertion of a homologous DNA template1,2. While recent studies have reported appreciable levels of HDR in cardiomyocytes in vivo3, skeletal muscle myofibres have historically been considered refractory to HDR-mediated genome editing4. Furthermore, how repair outcomes differ across tissues after systemic delivery of CRISPR/Cas9 editors, whether precise HDR editing can be achieved in regenerative tissue stem cells, and how developmental timing influences accessibility to CRISPR-induced repair remain unclear. Here, we use an adeno-associated virus (AAV)-delivered in vivo GFP-to-BFP colour-switching reporter system (AAV-GFP-to-BFP) to examine in vivo CRISPR-HDR with cellular- and tissue-level resolution. We find that postnatal cardiac muscle, skeletal muscle, and muscle stem cells undergo templated HDR at different rates across discrete developmental stages in mice. While HDR-edited muscle stem cells and myofibres were readily detectable after in vivo editing in juvenile mice, editing in neonatal mice yielded more efficient HDR in cardiac tissue. Based on these results, we adapted the CRISPR-HDR approach to rescue the therapeutically relevant Dmd mutation in mdx mice, demonstrating recoding to the wild-type protein sequence in both skeletal and cardiac muscles. These results provide a framework for advancing donor-templated DNA repair in living postnatal animals, and reveal unexpected cellular, developmental, and disease-related constraints on precise, therapeutic in vivo gene correction.

cell biology↗

bioRxiv2026

In vivo base editing via single myotrophic adeno-associated viruses in dystrophic mouse muscle and satellite cells

Duchenne muscular dystrophy (DMD) is the most common, lethal X-linked neuromuscular disorder of childhood and is caused by mutations in the Dmd gene that disrupt dystrophin expression. Although adeno-associated virus-mediated gene therapies hold tremendous promise for DMD treatment, their clinical applications have been limited by dose-dependent vector and genome-level toxicities. Here, we developed and tested a single-vector adenine base editing strategy as a potentially safer genome editing approach to recode the pathogenic nonsense mutation into a benign missense mutation in mdx4cvDMD mouse model. Delivered using a muscle-tropic adeno-associated virus (MyoAAV) at a clinically-feasible dose (4E13 VG/kg), this strategy enabled detectable molecular recoding of the mdx4cv mutation in mice ranging in age from 3 days to 6 months. Yet, the overall efficiency and therapeutic impact of in vivo base editing with this system was highest in mice treated at the juvenile stage, with animals administered MyoAAV vectors at 3 weeks of age showing robust recovery of dystrophin expression and significant improvement in muscle contractile properties only one month later. Notably, introduction of adenine base editors either earlier in development, in neonatal mice, or later, in adulthood, yielded substantially lower editing efficiencies, particularly in muscle satellite cells whose editing is essential to ensure durable rescue of dystrophin expression in growing and regenerating muscle. Taken together, these results demonstrate the therapeutic potential of single-vector adenine base editing for DMD and underscore the importance of recipient age and disease stage in achieving optimal treatment outcomes for this and other genetic muscle disorders.

cell biology↗

bioRxiv2026

Antibiotic Exposure Through Human Milk Influences the Infant Gut Microbiome

Infant antibiotic treatment is associated with increased risk of developing non-communicable diseases, potentially through disruption of the gut microbiome. However, the impact of indirect antibiotic exposure via human milk remains largely unexplored. Here, we investigate a cohort (n=80) of antibiotic-treated breastfeeding mother-infant dyads and untreated matching controls using integrative multi-omics analyses of fecal, milk, and skin samples (n=1,455). Maternal antibiotic treatment was associated with different infant fecal microbiome and metabolome profiles, including lower abundance of Bacteroides, Lactobacillus, and Bifidobacterium, and higher levels of antimicrobial resistance gene reads. Further, fecal metabolic alterations associated with indirect antibiotic exposure were exacerbated by formula milk supplementation. In a subset of infants (n=61), indirect exposure was associated with higher body mass index (BMI). These findings suggest that maternal antibiotic treatment during lactation may influence the early-life infant gut microbiome with potential long-term implications.

microbiology↗

bioRxiv2026

Precise measurement of rodent drinking using CLiQR (Capacitive Lick Quantification in Rodents)

Accurate quantification of rodent licking behavior is essential for studies of fluid intake, including investigations of alcohol use disorder and obesity. Lickometry systems vary widely in sensing modality, cost, scalability, and data resolution. Also, many available systems require specialized housing or store only binary lick/no lick data based on thresholding. Here, we present CLiQR (Capacitive Lick Quantification in Rodents), an open-source capacitive lickometry system designed for high-throughput recording of licking behavior in home-cage environments while preserving the full capacitance time series. The system uses MPR121 capacitive sensors connected to custom metal-tipped serological pipette sippers and a centralized desktop computer to record data from up to 24 mice concurrently, with capacity for two-bottle choice experiments. Validation experiments demonstrated that the capacitive signals reliably distinguish licking from non-licking interactions. Total lick counts showed a positive correlation with measured fluid consumption (R2 = 0.849, p < 0.0001), confirming that detected events provide a meaningful proxy for intake. All information necessary to reproduce the system is shared openly in this manuscript and online. By combining scalability, full-trace data acquisition, and low cost, CLiQR provides a flexible and extensible platform for high-throughput behavioral neuroscience experiments and enables retrospective improvement of lick-detection algorithms.

animal behavior and cognition↗

bioRxiv2026

E-Selectin Orchestrates IL-1B-Dependent Neuroinflammation via NLRP3 in Vincristine-Induced Neuropathy

Vincristine-induced peripheral neuropathy (VIPN) is a frequent and dose-limiting complication of cancer therapy, yet the upstream mechanisms coupling vascular activation to neuroinflammation remain poorly defined. Here we identify E-selectin as a critical orchestrator of vincristine-induced neuropathy. Systematic interrogation of endothelial adhesion molecules in a murine model of VIPN revealed that blockade of E-selectin, but not ICAM-1, PECAM-1 or P-selectin, completely prevented mechanical hypersensitivity and markedly reduced F4/80 immune cell accumulation in dorsal root ganglia and peripheral nerves. Genetic deletion of E-selectin conferred equivalent protection, despite the absence of structural loss of intraepidermal or myelinated fibres, indicating a predominantly functional neuroimmune pathology. Spatial transcriptomics demonstrated that vincristine induces a conserved stress and neuroinflammation-associated transcriptional programme in dorsal root ganglia, with immune and stromal populations acting as dominant signalling hubs. Genetic or pharmacological perturbation of E-selectin did not abolish injury-associated pathways but redistributed cell-cell communication networks, reducing immune-cell dominance and reshaping interferon and metabolic signalling states without inducing Sele expression. Mechanistically, E-selectin exerted non-canonical effects beyond endothelial adhesion. Local E-selectin administration was sufficient to induce macrophage-dependent mechanical hypersensitivity that was abolished in Fut4/7-deficient mice and following phagocyte depletion. In macrophages, E-selectin enhanced vincristine-driven NF-{kappa}B activation, NLRP3 inflammasome assembly and IL-1{beta} release. Together, these findings position E-selectin as an upstream regulator of IL-1{beta}-dependent neuroinflammation in VIPN and identify selective targeting of E-selectin-mediated immune-neuron interactions as a therapeutic strategy for chemotherapy-induced neuropathy.

neuroscience↗

bioRxiv2026

Hormone signaling and immune programs define differential endocrine responsiveness in high-risk breast tissue

Hormone therapies are frequently used to reduce breast cancer risk in individuals at increased risk for primary or subsequent disease; however, tissue-level responses to these therapies are heterogeneous and incompletely understood. Background parenchymal enhancement (BPE) on breast magnetic resonance imaging (MRI) provides a non-invasive radiologic readout of breast tissue features associated with endocrine responsiveness and cancer risk. Although BPE is associated with hormonal exposure, a subset of patients with BPE do not show a response to preventive endocrine therapy and therefore may remain at increased breast cancer risk. In this study, we integrated single-nucleus RNA sequencing and spatial transcriptomics to define the determinants of endocrine responsiveness in the setting of BPE. We identify hormone-driven epithelial cells with high levels of estrogen signaling and endocrine responsiveness, together with immune-associated epithelial programs characterized by diminished luminal identity and increased expression of immune-modulatory pathways, including major histocompatibility complex (MHC) class II and CD74. Functional organoid assays validate that these epithelial states exhibit differential sensitivity to tamoxifen and demonstrate that inflammatory signals can induce immune-modulatory epithelial programs. Together, our findings identify hormone signaling and immune programs as key determinants of endocrine responsiveness in breast tissue and provide a biological basis for interpreting radiologic markers relevant to cancer prevention.

cancer biology↗

bioRxiv2026

Volumetric montaging of optical coherence tomography in human retinas

Optical coherence tomography (OCT) has transformed clinical eye care by providing high-resolution volumetric imaging of the retina. Recently, ultrawide-field-of-view (FOV) OCT played an increasingly significant clinical role; however, most clinical OCT systems offer only a rather limited FOV. We increased the FOV of clinical OCT by volumetrically montaging multiple OCT datasets in three dimensions (3D). We performed volumetric montaging by representing the internal limiting membrane (ILM) and retinal pigment epithelium (RPE) in each volume as point clouds and using these point clouds to compute transformations that map each volume to a common reference frame. We validated our methodology using datasets from three institutions with different OCT hardware and data-acquisition procedures. Using the mean surface distance between point clouds, we found the error in montaging was less than the lateral pixel size. Our method enabled existing clinical OCT to achieve ultrawide FOV imaging without any hardware modification.

bioengineering↗

bioRxiv2026

The relationship between changes in psychosocial domains with a virtual salsa class and personality, perceived performance and enjoyment

This pre-post-test study investigated 1) pre-post changes in psychosocial domains with a single virtual salsa class; 2) effect sizes relative to an in-person class, and 3) individual factors, including personality, perceived performance, and enjoyment. An experimental group (n=33) of novice dancers 18-30 years old, participated in a single virtual salsa class. Positive and Negative Affect Scale (PAS, NAS), Perceived Stress Scale (PSS), and the Inclusion in Community and Self-Scale (ICS) were administered before and after class. Participants completed the Big Five Inventory-10 (BFI-10) before, and rated their performance and enjoyment (ordinal scale 1-5) after class. Effect sizes were calculated, and pre-post changes were analyzed with Wilcoxon signed-rank tests. Relationships between pre-post changes and individual factors were analyzed with Spearmans rank correlations. PAS, NAS, PSS, and ICS significantly improved and effect sizes were larger than those for an in-person salsa class except for ICS. Change in NAS was negatively correlated with neuroticism. These results suggest that a virtual salsa class may improve mood, stress, and social connection similar to in-person classes and change in mood may be influenced by personality traits such as neuroticism. Understanding the psychosocial effects of virtual dance and the influence of individual factors will facilitate implementation of dance as an accessible rehabilitation intervention to improve psychosocial well-being.

neuroscience↗

bioRxiv2026

Community Curation of Microbial Metabolites Enables Biological Insights of Metabolomics Data

Microbial metabolites play a critical role in regulating ecosystems, including the human body and its microbiota. However, understanding the physiologically relevant role of these molecules, especially through liquid chromatography tandem mass spectrometry (LC-MS/MS)-based untargeted metabolomics, poses significant challenges and often requires manual parsing of a large amount of literature, databases, and webpages. To address this gap, we established the Collaborative Microbial Metabolite Center knowledgebase (CMMC-KB), a platform that fosters collaborative efforts within the scientific community to curate knowledge about microbial metabolites. The CMMC-KB aims to collect comprehensive information about microbial molecules originating from microbial biosynthesis, drug metabolism, exposure-related molecules, food, host-derived molecules, and, whenever available, their known activities. Molecules from other sources, including host-produced, dietary, and pharmaceutical compounds, are also included. By enabling direct integration of this knowledgebase with downstream analytical tools, including molecular networking, we can deepen insights into microbiota and their metabolites, ultimately advancing our understanding of microbial ecosystems.

bioinformatics↗

bioRxiv2025

In silico reconstruction of primary and metastatic tumor architecture using GIS-augmented spatial transcriptomics

The tumor microenvironment (TME) comprises different cell populations that interact, contributing to tumor heterogeneity and therapy response. Spatial transcriptomics offers valuable insights into transcriptional complexity and heterogeneity of the TME. We established Geographic Information System (GIS)-augmented In-Silico Reconstruction of Tumor Architecture (GIS-ROTA), a biologically informed analytic framework that integrates pathway or cell type-based enrichment analysis with local Morans I to uncover functional spatial domains. In our Visium dataset of primary and metastatic estrogen receptor-positive breast tumor samples, GIS-ROTA revealed extensive co-localization of estrogen response with metabolic pathway gene sets and mutual exclusivity with metastasis-related and specific immune-related pathway gene sets. The novelty of our approach lies in considering biological functions prior to identifying any spatial domains, providing direct interpretability and minimizing the subjectivity of interpreting clusters observed from conventional analytic methods. Overall, our GIS-ROTA framework integrates biological knowledge first, yielding spatial patterns with functional relevance and enabling identification of novel targets for development of therapeutic strategies.

cancer biology↗

bioRxiv2025

Structure-Activity Relationship and Target Investigation of Thiophen-2-yl-Pyrimidines against Schistosoma species

Chemotherapeutic options for schistosomiasis, a prevalent infectious disease of poverty, are limited to just one drug, praziquantel (PZQ), and alternatives are needed. Our previous studies identified thiophen-2-yl pyrimidines (TPPs), which are structurally derived from microtubule (MT)-active phenylpyrimidines, as potent paralytics of Schistosoma mansoni. Although relatively non-toxic to mammalian cells, the progenitor compound, 3, had poor aqueous solubility and was lipophilic potentially hindering preclinical advancement. To address these issues and expand on the structure-activity and structure-property relationships, 43 new TPP analogs were designed and synthesized, their lipophilicity calculated (cLogP), and their anti-schistosomal activity evaluated in culture. This effort yielded compound 38, which possessed an oxetane-containing amine moiety at C5, and an ortho, ortho-difluoroaniline at C6 of the TPP scaffold. Compared to 3, compound 38 had better aqueous solubility (46 vs. < 0.5 {micro}M) and decreased lipophilicity (logP calc. 4.48 vs. 6.81), with toxicity CC50 values > 20 {micro}M against three mammalian cell lines. Further, paralytic potency, as measured by the EC50 value for adult S. mansoni motility, was increased 14.5-fold (538 vs. 37 nM), and plasma half-life (t1/2) was improved 3-fold, from 0.48 to 1.51 h for a 40% loss in maximum plasma concentration (Cmax). In washout experiments, 38 produced a sustained paralysis of both juvenile and adult S. mansoni, possibly suggesting a broader in vitro efficacy spectrum compared to PZQ, which is inactive against the juvenile parasite. Also, the two other medically important species, Schistosoma haematobium and Schistosoma japonicum, were susceptible to 38. Finally, to identify potential protein targets, we synthesized a TPP photoaffinity labeling (PAL) probe that labeled several S. mansoni proteins by SDS-PAGE fluorescence analysis, although, notably, not tubulin, suggesting that the antischistosomal activity of 38 is a function of engaging other targets. Future work with the TPP series will aim to decrease toxicity further while improving PK properties to better support in vivo efficacy testing.

microbiology↗

bioRxiv2025

ACSS2-Mediated Metabolic-Epigenetic Crosstalk Drives Fulvestrant Resistance and Represents a Novel Therapeutic Target

1.Endocrine therapies target hormone-dependent cancer cells, primarily through estrogen receptor alpha (ER), expressed in [~]70% of breast cancers (ER+). Despite treatment advances, 30-40% of ER+ breast cancer patients experience recurrence and metastasis, with 5-year survival rates of only 31.9%. We validated poor outcomes for liver metastasis patients treated with Fulvestrant (Fulv) using the local Carle Foundation Hospital cohort and examined metabolic pathways in liver metastatic patient-derived xenograft (PDX) models, revealing upregulated lipid and acetyl-CoA production. Our previous work demonstrated that combining Fulv with acetyl-CoA synthase inhibitor (ACSI) targeting Acyl-CoA Synthetase Short Chain Family Member 2 (ACSS2), synergistically reduced ER+ metastatic breast cancer (MBC) cell viability in vitro. Using multiple analytical approaches-isotope tracing, CUT&RUN sequencing, immunofluorescence, western blot, and RNA sequencing-we characterized the effects of acetyl-CoA synthesis inhibition on Fulv-induced alterations. Fulv treatment of MBC cells increased ACSS2 expression and acetate utilization. Isotope tracing revealed that Fulv decreased acetate flux to the TCA cycle while promoting fatty acid synthesis. Importantly, ACSS2 was predominantly nuclear and CUT&RUN sequencing showed that Fulv treatment increased ACSS2 chromatin occupancy and ER/ACSS2/H3K27ac overlapping sites near genes associated with tumor progression, which was eliminated by combination of ACSI and Fulv. RNA sequencing revealed reduction of Fulv-induced expression of genes involved in cancer cell metabolism and key signaling pathways in cancer with the Fulv+ACSI combination. In a therapy-resistant xenograft model, combining Fulv and ACSI reduced Fulv-dependent increase in metastatic burden. Our findings indicate ACSS2 contributes to endocrine therapy resistance through nuclear acetyl-CoA provision for epigenetic alterations. Targeting these cancer cell adaptations represents a novel therapeutic approach potentially reducing metastasis-related mortality and improving breast cancer treatment outcomes.

cancer biology↗

bioRxiv2025

GLIS3 Marks a Neural-like Progenitor Cell State that Drives Metastasis in Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) has a high rate of recurrence and metastasis despite intensive therapy. The neural-like progenitor (NRP) transcriptional program is enriched in residual disease after neoadjuvant chemotherapy and radiotherapy; however, the mechanisms for increased NRP expression in the post-treatment setting remain unclear. Here, we find that NRP signatures are strongly enriched in tissue injury and regeneration, and NRP cancer cells co-express transcription factors involved in pancreatic development. We identify both cell autonomous and ligand mediated mechanisms for inducing NRP expression in vitro. We establish and characterize isogenic mouse organoid overexpression models for transcription factors associated with the NRP, classical, and basal-like states. We discovered that Glis3 is a key NRP-associated transcription factor that drives malignant properties including greater clonogenicity, tumor growth, and metastasis compared to isogenic models for the classical and basal-like states. Our work highlights the emergence of clinically relevant developmental regeneration programs in the post-treatment context.

cancer biology↗

bioRxiv2025

Interpretable biophysical neural networks of transcriptional activation domains separate roles of protein abundance and coactivator binding

Deep neural networks have improved the accuracy of many difficult prediction tasks in biology, but it remains challenging to interpret these networks and learn molecular mechanisms. Here, we address the interpretability challenges associated with predicting transcriptional activation domains from protein sequence. Activation domains, regions within transcription factors that drive gene expression, were traditionally difficult to predict due to their sequence diversity and poor conservation. Multiple deep neural networks can now accurately predict activation domains, but these predictors are difficult to interpret. With the goal of interpretability, we designed simple neural networks that incorporated biophysical models of activation domains. The simplicity of these neural networks allowed us to visualize their parameters and directly interpret what the networks learned. The biophysical neural networks revealed two new ways that arrangement (i.e. the sequence grammar) of activation domain controlled function: 1) hydrophobic residues both increase activation domain strength and decrease protein abundance, and 2) acidic residues control both activation domain strength and protein abundance. Notably, the biophysical neural networks helped us to recognize the same signatures in complex interpreters of the deeper neural networks. We demonstrate how combining biophysical and deep neural networks maximizes both prediction accuracy and interpretability to yield insights into biological mechanisms.

systems biology↗

bioRxiv2025

Active learning enables discovery of transcriptional activators across fungal evolutionary space

Biological discovery and design are increasingly being guided by predictive models in place of costly experimentation. However, existing datasets are often biased by overrepresentation from model organisms, leading to failures in evolutionary studies of non-model species. We present a hybrid framework that leverages high-throughput molecular assays and active learning to quantify biological properties across evolutionary space. We focus on transcriptional activators, which contain activation domains (ADs) that promote gene expression. ADs are intrinsically disordered and poorly conserved, which limits their study using comparative genomics. Here, we developed ADhunter, a high-capacity regression model that outperforms state-of-theart algorithms in identifying and quantifying the strength of transcriptional activators. Model uncertainty was used to guide evolutionary sampling across 7.8 million proteins from 2,400 fungal genomes. We functionally characterized 9,836 ADs from 1,071 fungal genomes, providing a 15.5-fold expansion in genome representation compared to existing datasets. Comprehensive sampling from non-model genomes improved model generalizability and provides the first functional annotation for 3,416 proteins from 670 non-model fungi. Model interpretability analysis aligns with the biophysical model of AD function and reveals novel, underrepresented protein codes, highlighting the importance of sampling from non-model organisms to build evolutionarily robust models for predicting biological properties.

genomics↗