Decoding the Role of H19 in Cholestatic Liver Injury Using snRNA-seq, Spatial Transcriptomics, and Machine Learning-Based Disease Prediction

Decoding the Role of H19 in Cholestatic Liver Injury Using snRNA-seq, Spatial Transcriptomics, and Machine Learning-Based Disease Prediction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Decoding the Role of H19 in Cholestatic Liver Injury Using snRNA-seq, Spatial Transcriptomics, and Machine Learning-Based Disease Prediction Grayson Welch Way, Xixian Jiang, Hongkun Lu, Nan Wu, Derrick Zhao, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8339668/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Background Primary Sclerosing Cholangitis (PSC) is a chronic obstructive biliary disease and remains a high-burden cholestatic liver disease with no approved therapies and a substantial recurrence rate following liver transplantation. The long non-coding RNA H19 (H19) has emerged as a potential driver of PSC progression, yet its cell-type-specific and spatially resolved mechanisms remain poorly defined. Results Age- and sex-matched wild type (WT), H19 knockout (H19KO), Mdr2 knockout (Mdr2KO), and double-knockout (DKO; Mdr2KO/H19KO) mice were used. The liver tissues were analyzed using single nucleus RNA sequencing (snRNAseq) and NanoString GeoMx spatial transcriptomics to elucidate H19-dependent cellular and spatial alternations in cholestatic liver injury. Machine learning models (logistic regression, XGBoost, neural network, and random forest) were developed to generate cell-type specific disease prediction signatures and validated using the publicly available human dataset GSE243981. Both spatial transcriptomics and snRNAseq identified a disease-associated cholangiocyte subcluster that was significantly expanded in Mdr2KO mice, but markedly diminished in DKO mice, demonstrating a requirement for H19 in sustaining pathogenic cholangiocyte state. SPP1 signaling was significantly dysregulated in cholestatic liver injury and ameliorated with H19 deletion. Novel murine markers were identified, including Gm13775 (healthy hepatocytes) and Clu and Spp1 (healthy cholangiocytes), all of which were markedly downregulaed in disease. Machine learning-based, cell type-specific disease prediction models achieved AUC values > 0.87 when validated in the GSE243981 human dataset. Noteably,Spp1 expression decreased in cholangiocytes but was ectopically upregulated in hepatocytes in diseased liver, highlighting disrupted intercellular signaling network. Spatial analyses showed that H19 deletion restored the disease-associated gene expression changes specifically within the bile duct region. Conclusion H19 deletion mitigates cholestatic injury by suppressing pathogenic cholangiocyte states, normalizing SPP1-mediated signaling, and restoring bile-duct-localized transcriptional programs. These findings position H19 as a critical regulator of cholangiocyte-driven pathology and a potential therapeutic target in PSC. Primary sclerosing cholangitis Cholestasis SPP1 Machine learning H19 Long non-coding RNA Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Primary sclerosing cholangitis (PSC) is a chronic, progressive hepatobiliary disease characterized by biliary fibrosis, obstructive cholestasis and inflammation of the intrahepatic and extrahepatic bile ducts( 1 – 3 ). The disruption of bile acid homeostasis contributes to a cascade of metabolic and immune dysregulation, as bile acids are important signaling molecules. PSC progresses along a spectrum with eventual development of end-stage liver failure( 4 ). PSC is also associated with a markedly increased risk of malignancies, such as cholangiocarcinoma, gallbladder carcinoma, hepatocellular carcinoma, and colorectal carcinoma( 5 – 10 ). In the United States, PSC is the fifth leading indication for liver transplantation( 11 ). Furthermore, recurrence occurs in up to 30% of transplant recipients( 11 ). Despite the severity of the disease, no proven pharmacotherapies exist, largely due to an incomplete understanding of the cellular and molecular mechanisms underlying PSC pathogenesis( 11 ). The long non-coding RNA H19 (H19) has emerged as an important regulator of PSC disease progression( 12 , 13 ). Previous research in our lab, and others, have shown that H19, which is normally not expressed in healthy hepatic tissues, is upregulated in PSC and exacerbates disease progression in Mdr2KO mice (the gold standard mouse model for PSC)( 12 , 14 – 19 ). However, the field lacks a mechanistic framework explaining how H19 reprograms specific hepatic cell populations, whether its effects are spatially restricted in the liver microenvironment, and which transcriptional networks are directly or indirectly governed by H19 in vivo. It is also unclear how diseae-assocaited cell states in mice correspond to those in human cholestatic liver disease, and whether concise molecular signatures can reliably classify cholangiocyte or hepatocyte health status across species. To address these gaps, we leveraged single-nucleus RNA sequencing (snRNA-seq) and spatial transcriptomic profiling across WT, H19KO, Mdr2KO and DKO mice. While single-cell RNA sequencing (scRNAseq) has advantages for profiling immune cells, snRNAseq offers several distinct benefits, including compatibility with frozen tissue, reduced dissociation-induced artifacts, and superior accuracy in profiling hepatocytes and cholangiocytes ( 20 ). Spatial transcriptomic technology, such as GeoMx® Digital Spatial Profiler (DSP), allows transcriptomic analysis at specific regions based on histological morphology. In our previous work, we demonstrated that genetic deletion of H19 significantly reduced liver fibrosis and slowed disease progression in Mdr2KO mice( 15 , 17 ). In this study, integration of snRNAseq and GeoMx DSP enabled us to 1) define disease-specific cholangiocyte and hepatocyte states associated with PSC, 2) map how h19 deletion reshapes these states at cellular and spatial levels, and 3) quantify the spatial restriction of H19-mediated transcriptional rescue. We further integrated publicly available human scRNA-seq and snRNA-seq datasets to evaluate cross-species conservation and built machine-learning prediction models based on minimal gene sets capable of classifying cholangiocyte and hepatocyte disease status with high accuracy. Materials and methods Animal Studies C57BL/6J Mdr2KO mice were gifted by Dr. Daniel Goldenberg, Department of Pathology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel. Maternal C57BL/6J H19 ΔExon1/+ (H19KO) mice were generated and gifted by Dr. Karl Pfeifer at the NIH. Dr. Jian-Ying Wang at the University of Maryland (Baltimore, MD, USA) provided the H19KO mice. Mdr2KO and H19KO (DKO) mice were generated as previously described ( 17 ). The gender and age-matched littermates of WT, H19KO, Mdr2KO and DKO (female, 6-month-old) were used. Mice were housed under 12 h/12 h light and dark cycle with unrestricted access to water and standard chow ad libitum . All animal experiments followed institutional guidelines for ethical animal studies and were approved by VCU institutional animal care and use committee. For GeoMx spatial transcriptomics, eight liver samples were collected from 6-month-old female mice (four Mdr2⁻/⁻ and four DKO). For single-nucleus RNA sequencing (snRNA-seq), liver tissues from 6-month-old female mice (one per genotype: WT, H19KO, Mdr2KO, and DKO) were flash-frozen in liquid nitrogen and stored at − 80°C until processing. scRNAseq and snRNAseq data analysis scRNAseq and snRNAseq analyses of both our dataset and the GSE243981 dataset were performed in R (version 4.3.3) utilizing the R package Seurat (version 5.1.0)( 23 , 24 ). For our mouse snRNA-seq data, nuclei with > 2.5% mitochondrial gene content were excluded. The R package SoupX (version 1.6.2) was used to remove ambient RNA contamination( 21 ). For GSE243981, mitochondrial content filters of 5% (snRNA-seq) and 25% (scRNA-seq) were applied. Data normalization and scaling were performed using Seurat’s SCTransform function and data integration was performed utilizing Harmony, accounting for known batch variables (i.e., different experimental techniques used: scRNAseq, snRNAseq, 3’ and 5’ sequencing) following the Seurat SCTransform workflow( 25 ). Predicted cell-cell communication were inferred using the R package CellChatv2 (version 2.1.0)( 26 ). Pseudotime analyses for cell type progression with disease was performed using Monocle3 (version 1.4.26)( 27 ). LASSO regression analysis was performed using the R package glmnet (version 4.1.10)( 28 , 29 ). Data visualization and statistical analyses were performed with SeuratExtend (version 1.2.5)( 30 ). Statistical significance was determined using the MAST algorithm via Seurat’s FindMarkers function. Machine learning Modeling Methods To assess the robustness and predictive accuracy of input features, four supervised machine learning models were developed and evaluated: a multilayer perceptron (MLP) neural network, an extreme gradient boosting (XGBoost) classifier, a random forest classifier, and a logistic regression model. For initial validation, models were trained and tested on mouse data using a 70/30 train–test split. For cross-species validation, each model was trained on the complete mouse dataset and subsequently tested on the entire human dataset (GSE243981). Mouse gene symbols were converted to their human orthologs using the gorth function in the gprofiler2 R package to enable translational comparison( 31 , 32 ). All models were trained to classify binary health status labels (healthy vs. diseased) and evaluated using the area under the receiver operating characteristic curve (ROC-AUC) as the primary performance metric with the R package pROC( 33 ). GeoMx Data Analysis GeoMx® data analysis was performed using the StandR R package following their pipeline( 34 ). QC was performed at both the region of interest (ROI) level and gene level to remove low-quality data. ROIs were filtered based on nuclear count, surface area, and library size using default thresholds. PCA was performed using the runPCA function from the scater package (version 1.36.0), and clustering was visualized using Uniform Manifold Approximation and Projection (UMAP)( 35 ). Different normalization (TMM, CPM, Upper quartile) and batch correction methods (RUV4, LimmaRemoveBatch, SVA) were tested. Ultimately, TMM normalization and RUV4 batch correction were used based on PCA and cluster separation statistics. Differential gene expression (DGE) analysis was performed using the limma R package (version 3.64.3)( 36 ). Significance was also tested and confirmed using DESeq2 (Version 1.48.1)( 37 ). Results Identification of a disease-specific cholangiocyte cluster Initial clustering and cell type annotation of the snRNAseq data identified 16 distinct clusters from 35,488 nuclei across all four genotypes. UMAP visualization confirmed proper integration across all genotypes and experimental techniques (Fig. 1 A). Cell identity was assigned based on the expression of canonical maker genes, demonstrating high fidelity clustering (Fig. 1 B). Compared to WT and H19KO, Mdr2KO and DKO mice exhibited drastic increases in cholangiocytes, fibroblasts, lymphocytes, and monocyte-derived macrophages (MdMQs) in terms of numbers and percent cell composition, accompanied by a reduction in hepatocytes and Kupffer cells, compared to their healthy counterparts (cholangiocytes, MdMQs and hepatocytes percentage of cells p < 0.05) (Fig. 1 C). Subclustering of cholangiocytes identified a disease-associated cholangiocyte cluster (cluster 2) (Fig. 2 A). Pseudotime trajectory analysis using monocle3 indicated that Cluster 2 corresponded to a late-stage disease state (Fig. 2 B). Notably, H19 deletion in Mdr2KO mice significantly reduced both the number and percentage of cholangiocytes within this cluster (Fig. 2 C). The top marker gene for cluster 2 was Csmd1 (CUB And Sushi Multiple Domains 1), which was markedly upregulated in disease and downregulated upon H19 deletion ( Supp. Figure 1 and Fig. 2 D). Cellular pathways altered in PSC and ameliorated by H19 deletion CellChatv2 analysis was performed to identify significant cell-cell signaling network and crosstalk interactions altered in Mdr2KO mice and to evaluate pathways modulated by H19 deletion. The total number of inferred interactions identified for WT, H19KO, Mdr2KO, and DKO were 255, 180, 573, and 438, respectively ( Supp. Figure 2 ). Among the pathways examined, Spp1 (osteopontin), collagen, laminin, and FN1 (fibronectin) signaling were all significantly upregulated in Mdr2KO compared to WT and were normalized toward WT in DKO mice (p-value < 0.05) (Fig. 2 A-B, Supp. Figures 3 – 5 ). Notably, Spp1 emerged as a major signaling pathway dysregulated in cholestatic liver injury and restored by H19 deletion. In healthy liver, Spp1 expression is predominately observed in cholangiocytes, mediating signaling to Kupffer cells. In contrast, during disease progression, Spp1 expression shifts to mid-lobular hepatocytes, indicating altered paracrine signaling (Fig. 3 C - D, Supp. Figure 6 A-B). H19 deletion significantly reduces Spp1 expression and outgoing Spp1-mediated signaling from mid-lobular hepatocytes (p < 0.05) (Fig. 2 C-D). Ligand-receptor analysis revealed that the top predicted interactions in the Spp1 pathway were Spp1-Cd44, Spp1-(Itgav + Itgb6), and Spp1-(Itga4 + Itgb1). Notably, Spp1-Cd44 signaling between cholangiocytes and Kupffer cells was present only in healthy liver and was lost in Mdr2KO mice, whereas the other Spp1 interactions were uniquely predicted in Mdr2KO mice (Fig. 4 A-B). Diseased cholangiocyte gene prediction model and identification of novel healthy markers Differential gene expression (DGE) analysis of subclustered cholangiocytes, combined with LASSO regression-based feature selection, identified six genes ( Clu , Spp1 , Slco3a1 , Cd44 , Anxa3 , Cftr ) for machine learning-based prediction modeling of cholangiocyte disease status (Fig. 4 C). Among these, Clu and Spp1 were strongly associated with healthy cholangiocytes, whereas Slco3a1 , Cd44 , Anxa3 , Cftr were enriched in diseased cholangiocytes in Mdr2KO and DKO mice (adj. p < 0.001) (Fig. 4 C). Three of the disease associated genes ( Slco3a1 , Anxa3 , Cftr ) were significantly reduced towards WT levels in DKO mice, indicating that H19 deletion ameliorates their dysregulation (Fig. 4 C). A striking spatial shift was observed for Spp1 and Clu , which transitioned from cholangiocyte-restricted expression in healthy liver to ectopic expression in mid-lobular hepatocytes in diseased liver, suggesting substantial rewiring of cellular identity and intercellular signaling during disease progression ( Supp. Figure 6 A–B). Cholangiocyte disease prediction model testing in humans The expression patterns of six model genes ( Clu , Spp1 , Slco3a1 , Cd44 , Anxa3 , Cftr ) were evaluated in human cholangiocytes using the GSE243981 dataset (both scRNAseq and snRNAseq). Five of the six genes demonstrated the same significant expression trends in PSC patient cholangiocytes as observed in mice (p < 0.05), with Cd44 as the exception when combining scRNAseq and snRNAseq datasets (Fig. 5 A), but was significantly different in scRNAseq data alone (Fig. 5 B). Cftr expression was significantly decreased in the combined data (sc/snRNAseq) (Fig. 5 A), but did not reach significance in the scRNAseq-only analysis (Fig. 5 B). These six genes were subsequently used to construct build cholangiocyte disease prediction models using four supervised machine learning algorithms: multilayer perceptron (MLP) neural network, XGBoost, random forest, and logistic regression. ROC–AUC analysis demonstrated robust performance, with AUC values > 0.92 in mouse datasets and > 0.87 in human cholangiocytes from GSE243981, supporting the translational relevance of the gene signatures and the predictive models (Fig. 5 C). Diseased Hepatocytes gene prediction model and identification of a novel healthy marker Hepatocytes were subclustered, zonally differentiated, and UMAPs generated to visualize proper data clustering and integration (Fig. 6 A-B, Supp. Figure 7 )( 38 , 39 ). Differential gene expression combined with LASSO regression identified six genes— Gm13775, Spp1, C6, Cdh1, Npas2 , and Cd74 —for construction of a hepatocyte disease prediction model using a random forest classifier, which achieved an AUC of 0.916 in mice ( Supp. Figure 8 A). Gm13775 emerged as the top-performing gene and a novel marker of healthy hepatocytes ( Supp. Figure 8 B–C). However, because no human homolog has been identified, Gm13775 was excluded from downstream cross-species model development. A revised six-gene model ( Spp1 , C6 , Npas2 , Cdh1, Cd74 , and Hamp ) was generated for comparative analyses. In Mdr2KO hepatocytes Spp1, C6, Npas2 , Cdh1 , and Cd74 were all significantly upregulated, while Hamp was significantly reduced relative to WT (adj. p < 0.001) (Fig. 6 C). In DKO mice, Spp1, Cdh1 , and Cd74 were significantly decreased and Hamp significantly increased toward WT levels (adj. p < 0.01), indicating that H19 deletion partially corrects hepatocyte dysregulation (Fig. 6 D). CellChat v2 pathway analysis further revealed that APP signaling was significantly reduced in DKO mice (Fig. 2 ). Across all genotypes, the dominant ligand–receptor interaction was App–Cd74, with mid-lobular hepatocytes showing increased CD74 expression and receiving App-derived signaling in disease states ( Supp. Figure 9 ). Hepatocyte disease prediction model testing in humans The six genes used in the hepatocyte disease prediction model (Hamp, Spp1, C6, Cdh1, Npas2, and Cd74 ) demonstrated strong translational concordance in human hepatocytes from the GSE243981 dataset. All six genes showed consistent and significant expression changes in both the combined scRNAseq/snRNAseq dataset (Fig. 7 A) and in the scRNAseq subset alone (Fig. 7 B). When tested across four supervised machine learning algorithms, these six genes achieved AUC values > 0.869 in both mouse and human datasets, confirming their cross-species robustness and strong predictive performance (Fig. 7 C). Deletion of H19 causes cell-type specific alterations in gene expression Random forest modeling was performed across multiple cell types to identify gene expression patterns associated with H19 deletion. Xist consistently emerged as the top predictive feature across all models ( Supp. Figure 10 ). Although Xist/Tsix dominated feature importance, each cell type also showed distinct H19 -dependent transcriptional signatures. In hepatocytes, a 13 gene model ( Xist , Gphn , Fga , Tsix , Malat1 , Ahsg , Apob , Ces1c , Spp1 , Cers6 , Cd74 , Srsf1 , and Hspa2 ) achieved an AUC score of 0.972 for distinguishing H19 deletion from WT mice ( Supp. Figure 10 ). Excluding Xist and Tsix modestly reduced performance (AUC = 0.902). In cholangiocytes, a 13 gene model ( Xist, Tsix, Csmd1, Chka, Frmd4b, Gnas, Npas2, Agmo, Ly6e, Gbp9, Sorbs3, Usp2 , and 1600012H06Rik ) achieved an AUC score of 0.934 (0.867 without Xist and Tsix ). In macrophages, a 12 gene model ( Xist, Tsix, Mat1a, Gm42418, Eef1a1, Rpl13a, Gm19951, Clec2d, Wwc1, Rnf128, Ugt2b5 , and Tnfrsf12a ) generated an AUC score of 0.861, decreasing to 0.764 without Xist and Tsix. Importantly, several genes within these models overlapped with those reversed by H19 deletion-that is, genes upregulated (or downregulated) in Mdr2KO vs. WT and reversed toward WT levels in DKO mice, indicating correction of PSC-associated dysregulation. Notable examples include Spp1 and Cd74 in hepatocytes, and Frmd4b , Gnas , Chka , Csmd1 , and Agmo in cholangiocytes ( Supp. Tables.1&2 ). Spatial transcriptomics reveal that H19 deletion-mediated amelioration in hepatocytes is restricted to hepatocytes in close proximity to bile ducts Spatial transcriptomics was utilized to investigate spatially defined effects of H19 deletion on hepatic gene expression. ROIs were selected to compare differences in isolated hepatocytes and hepatocytes in close proximity to bile ducts (Fig. 8 A). DGE analysis of the different ROIs comparing DKO vs Mdr2KO mice yielded several common and unique significant DEGs (Fig. 8 B). Common genes decreased in DKO mice vs Mdr2KO across both regions were genes associated with cellular stress responses, such as Hspa5 , Hspa8 , and Hsp90aa1 (p < 0.01 across both regions), and fatty acid synthesis gene Fasn (p < 0.01); however, several cholesterol metabolism genes were only significantly reduced in hepatocytes neighboring bile ducts (e.g., Apoa1 , Apoa2 , and Apoa5 ) (Fig. 8 B). Several genes related to significant disease-associated pathways (APP, Collagen, FN1, LAMININ) ameliorated by H19 deletion as well as hepatocyte-to-cholangiocyte transition marker genes ( Sox4 and Sox9 ) are only significantly reduced in regions with hepatocytes neighboring bile ducts (Fig. 8 B). Four of the five hepatocyte disease-associated prediction modeling genes were only significantly ameliorated in hepatocytes neighboring cholangiocytes ( Spp1, C6, Cdh1 , and Cd74 ) (Fig. 8 C), demonstrating that H19 deletion-mediated transcriptional recovery is spatially restricted to periductal hepatocytes. Discussion Understanding how PSC reshapes hepatobiliary cell states at single-cell and spatial resolution has been a longstanding barrier in the field. Although H19 has been implicated in cholestatic injury, the mechanisms by which it influences specific cellular trajectories and the spatial architecture of disease have remained unresolved. ( 12 , 14 – 19 ). In this study, supported by findings from Andrews, et. al , our integrative snRNA-seq, spatial transcriptomic, and machine-learning strategy addresses these gaps and provides several conceptual advances that reframe the role of H19 in PSC pathogenesis ( 32 ). A central finding of this study is the identification of a disease-specific cholangiocyte population defined by Csmd1 and Lama4 expression that emerges during PSC progression and is substantially reduced by H19 deletion. This subpopulation aligns with a late-stage pseudotime state, establishing a mechanistic link between H19 activity and cholangiocyte trajectory convergence toward a pathogenic cell identity. The enrichment of Lama4, an extracellular matrix component with known profibrotic functions, further connects H19 biology to matrix remodeling and fibrosis. Using multimodal approaches, we identified six genes that robustly distinguished healthy from diseased cholangiocytes across scRNA-seq and combined sc/snRNA-seq datasets. Notably, the simplest model, logistic regression, achieved strong performance, suggesting these genes define a relatively linear health vs disease axis. Several disease-specific genes ( Anxa3 , Slco3a1 , and Cftr ) were significantly reduced with H19 deletion (adj. p < 0.001, DKO vs Mdr2KO) (Fig. 4 F). Anxa3 encodes Annexin A3, a calcium-dependent phospholipid-binding protein elevated in liver cancers and explored as a diagnostic and therapeutic target ( 40 , 41 ). Slco3a1 (OATP3A1), a sodium-independent organic anion transporter linked to Crohn’s disease and NF-κB activation ( 42 ), also mediates bile-acid efflux and has been proposed as a protective adaptive response in cholestasis ( 43 ). Although Slco3a1 expression was reduced in DKO vs Mdr2KO (adj. p 0.05), indicating that its down-regulation likely reflects reduced cholestasis rather than direct H19 regulation. Two markers of healthy cholangiocytes, Clu and Spp1, showed striking disease-associated shifts: both were downregulated in cholangiocytes yet ectopically activated in hepatocytes in injured livers. Notably, reduced circulating CLU (clusterin) levels correlate with worse outcomes in biliary atresia, underscoring the clinical significance of losing these homeostatic markers ( 44 ). Despite its pathogenic role, H19 did not preserve healthy cholangiocyte identity and instead further suppressed Spp1 expression (p < 0.05). Beyond cholestatic liver disease, Spp1 signaling drives mesenchymal transition in pancreatic ductal adenocarcinoma, and its inhibition reduces tumor burden and improves survival ( 45 ). In healthy liver, cholangiocytes are the dominant source of Spp1 ( 46 ). Here, however, we find that during cholestatic injury Spp1 signaling is elevated but largely reprogrammed to originate from hepatocytes, suggesting a pathologic shift in intercellular communication. This hepatocyte-derived Spp1 may promote hepatic stellate cell (HSC) activation and myofibroblast transition, while suppression of this aberrant hepatocyte signal could limit fibrogenesis without compromising physiologic Spp1 function in cholangiocytes. snRNA-seq analysis uncovered distinct, cell-type–specific transcriptional alterations driven by cholestatic disease and modulated by H19 deletion. Diseased hepatocytes exhibited ectopic activation of genes normally restricted to other healthy cell types, including Spp1 and Clu (cholangiocytes) and Cd74 (antigen-presenting cells). H19 deletion reversed these aberrant expression patterns: Clu and Spp1 were significantly elevated in Mdr2KO compared with WT, but markedly reduced in DKO relative to Mdr2KO (adjusted p < 0.001) (Fig. 4 , Supp. Figure 6 ). In alignment with prior findings linking hepatic Cd74 up-regulation to Ikbkb loss, Mdr2KO hepatocytes demonstrated significantly decreased Ikbkb expression compared with WT (adjusted p < 0.001)( 47 ). Both Cd74 induction and Ikbkb suppression were normalized in DKO hepatocytes (adjusted p < 0.001) (Supp. Table 3), indicating that H19 influences hepatocyte immune signaling pathways. Finally, we identified a previously unrecognized healthy hepatocyte marker, Gm13775, a non-coding RNA with no known homologs. Its biological role remains undefined and warrants further functional investigation. A novel and unexpected discovery was the strong positive association between H19 and the X-chromosome inactivation regulators Xist and its antisense transcript Tsix. Xist expression was significantly reduced across all cell types in H19KO vs WT, DKO vs Mdr2KO, and combined DKO/H19KO vs Mdr2KO/WT comparisons (adjusted p 70% to < 6% in H19KO vs WT and DKO/H19KO vs Mdr2KO/WT comparisons. This relationship suggests a previously unrecognized regulatory connection between H19 and the Xist–Tsix axis. Perturbation of this epigenetic network could generate pseudo–copy number–like expression shifts in females and may contribute to sex-specific transcriptional differences in cholestatic liver disease ( 48 ).Further mechanistic dissection of this interaction is clearly warranted. We identified multiple cell type–specific differentially expressed genes influenced by H19. Machine-learning models distinguishing H19 WT from H19 knockout cells performed strongly across all major cell populations (AUC > 0.85 when including Xist), underscoring robust and highly cell type–specific transcriptional signatures. In hepatocytes, Spp1 and Cd74—two key disease-associated genes—were clearly regulated by H19, with H19 deletion restoring their expression toward WT levels. Of note, the APP–CD74 signaling axis has been implicated in promoting fibrosis in the kidney ( 49 ), raising the possibility of a conserved H19-dependent fibrogenic mechanism across organ systems. Additional hepatocyte-specific H19-responsive genes included the lipid metabolism regulators Cers6, Ces1c, and Apob, as well as Ahsg, all of which were downregulated following H19 deletion. Elevated Ahsg expression has previously been linked to tumor proliferation ( 50 ), suggesting broader relevance to liver tumorigenesis. In cholangiocytes, H19 deletion influenced several genes, including Frmd4b, Chka, Csmd1, Gnas, and Agmo. While increased Csmd1 expression has been associated with hepatocellular carcinoma ( 51 ), other studies propose that CSMD1 may function as a tumor suppressor depending on cellular and disease context ( 52 , 53 ), highlighting the complexity of its role in cholangiocyte biology. GeoMx spatial transcriptomics demonstrated that the improvement in hepatocyte disease-specific gene expression following H19 deletion was spatially restricted to hepatocytes adjacent to bile ducts. This pattern is consistent with prior reports showing that cholangiocytes are the primary source of H19 during cholestatic injury and that H19 can be transferred to neighboring cells via extracellular vesicles ( 14 , 15 , 17 ). Together, these findings suggest that H19-driven pathogenic signaling is mediated through local, bile duct–proximal intercellular communication, and that its removal selectively normalizes transcriptional programs in hepatocytes directly influenced by cholangiocyte-derived H19. Taken together, these findings demonstrate that H19 deletion induces both cell-type–specific and spatially resolved transcriptomic remodeling, restoring healthy gene-expression programs across multiple hepatic lineages and attenuating PSC-related pathology in both mouse and human datasets. While these results position H19 inhibition as a compelling therapeutic strategy for cholestatic liver injury, the newly uncovered regulatory interplay between H19 and the Xist/Tsix epigenetic axis underscores the need for careful evaluation of sex-dependent and broader epigenetic consequences prior to clinical translation. Short comings Although we were able to validate disease-associated differential gene-expression patterns and predictive modeling outputs using an independent human dataset, we could not directly validate the H19-deletion–specific findings due to the absence of comparable datasets. Additionally, the sequencing depth in both our mouse and publicly available human snRNA-seq datasets was insufficient to reliably detect H19, and H19 was not included in the GeoMx Whole Transcriptome Atlas (WTA) panel. These constraints limited our ability to perform direct cross-species validation of H19-dependent effects at either the single-cell or spatial resolution. Abbreviations AF Alexa Fluor aSMA alpha Smooth Muscle Actin AUC Area Under the Curve DEGs Differentially Expressed Genes DGE Differential Gene expression DKO Mdr2KOH19– / –Double Knockout DSP Digital Spatial Profiler HCC Hepatocellular Carcinoma H19 long non–coding RNA H19 H19KO H19 Knockout LASSO Least Absolute Shrinkage and Selection Operator Mdr2 Multidrug resistance 2 PBC Primary Biliary Cholangitis PSC Primary Sclerosing Cholangitis RNA Ribonucleic Acid ROC Receiver Operating Characteristic scRNAseq Single Cell RNA Sequencing snRNAseq Single Nuclear RNA sequencing WT Wild Type WTA Whole Transcriptome Atlas Declarations Conflict of Interest Statement: The authors have no conflict of interests to disclose. Financial Support This study was supported by VA Merit Award 5I01BX005730, VA ShEEP grants (1 IS1 BX004777-01 and 1IS1BX005517-01), National Institutes of Health Grant R01 DK115377, 2R56DK115377-05A1, 5R01AA030180, R01DK139587, NIH-NCI P01CA275740, 5T32AA029975. Dr. Zhou is the recipient of a Research Career Scientist Award from the Department of Veterans Affairs (IK6BX004477). Clinical Trial Number N/A Declarations Ethics approval and consent to participate All animal experiments were performed following institutional guidelines for ethical animal studies and approved by the Virginia Commonwealth University and Richmond VA Medical Center Institutional Animal Care and Use Committee, Virginia, USA. Consent for publication All authors reviewed and approved the final manuscript. All authors supported the publication of this manuscript. Competing interests The authors declare no competing financial interests. Funding This study was supported by VA Merit Award 5I01BX005730, VA ShEEP grants (1 IS1 BX004777-01 and 1IS1BX005517-01), National Institutes of Health Grant R01 DK115377, 2R56DK115377-05A1, 5R01AA030180, and NIH-NCI 1P01CA275740-01 and Cancer Center Support Grant P30 CA 016059. Dr. Zhou is the recipient of a Research Career Scientist Award from the Department of Veterans Affairs (IK6BX004477). Author Contribution Conceptualization and design, GWW, HZ; Methodology, GWW, XJ, HZ; Data Acquisition, GWW, XJ, HL, NW, DZ, YT, SB, XW; Data Analysis, GWW, XJ; Writing: Original Draft, GWW, XJ; Writing: Review & Editing, GWW, XJ, HZ; Funding Acquisition, HZ; Resources, HZ; Supervision, HZ. Data Availability All data and analysis methodology will be made available upon requests to the corresponding author. snRNAseq and spatial transcriptomic data will be uploaded to GEO. References Angulo P, Lindor KD. Primary sclerosing cholangitis. Hepatology. 1999;30:325–32. Hirschfield GM, Karlsen TH, Lindor KD, Adams DH. Primary sclerosing cholangitis. Lancet. 2013;382:1587–99. Karlsen TH, Folseraas T, Thorburn D, Vesterhus M. Primary sclerosing cholangitis - a comprehensive review. J Hepatol. 2017;67:1298–323. Eaton JE, Talwalkar JA, Lazaridis KN, Gores GJ, Lindor KD. Pathogenesis of primary sclerosing cholangitis and advances in diagnosis and management. Gastroenterology. 2013;145:521–36. Tabibian JH, Ali AH, Lindor KD. Primary Sclerosing Cholangitis, Part 2: Cancer Risk, Prevention, and Surveillance. Gastroenterol Hepatol (N Y). 2018;14:427–32. Chiang JYL, Ferrell JM. Bile Acids as Metabolic Regulators and Nutrient Sensors. Annu Rev Nutr. 2019;39:175–200. Chiang JYL, Ferrell JM. Bile acid receptors FXR and TGR5 signaling in fatty liver diseases and therapy. Am J Physiol Gastrointest Liver Physiol. 2020;318:G554–73. Hylemon PB, Takabe K, Dozmorov M, Nagahashi M, Zhou H. Bile acids as global regulators of hepatic nutrient metabolism. Liver Res. 2017;1:10–6. Jackson KG, Way GW, Zhou H. Bile acids and sphingolipids in non-alcoholic fatty liver disease. Chin Med J 2022: 10.1097/CM1099.0000000000002156 Way GW, Jackson KG, Muscu SR, Zhou H. Key Signaling in Alcohol-Associated Liver Disease: The Role of Bile Acids. Cells 2022;11. Tabibian JH, Ali AH, Lindor KD. Primary Sclerosing Cholangitis, Part 1: Epidemiology, Etiopathogenesis, Clinical Features, and Treatment. Gastroenterol Hepatol (N Y). 2018;14:293–304. Wang Y, Hylemon PB, Zhou H. Long Noncoding RNA H19: A Key Player in Liver Diseases. Hepatology. 2021;74:1652–9. Wang Y, Zeng J, Chen W, Fan J, Hylemon PB, Zhou H. Long Noncoding RNA H19: A Novel Oncogene in Liver Cancer. Noncoding RNA 2023;9. Li X, Liu R, Huang Z, Gurley EC, Wang X, Wang J, He H, et al. Cholangiocyte-derived exosomal long noncoding RNA H19 promotes cholestatic liver injury in mouse and humans. Hepatology. 2018;68:599–615. Li X, Liu R, Wang Y, Zhu W, Zhao D, Wang X, Yang H et al. Cholangiocyte-Derived Exosomal lncRNA H19 Promotes Macrophage Activation and Hepatic Inflammation under Cholestatic Conditions. Cells 2020;9. Li X, Liu R, Yang J, Sun L, Zhang L, Jiang Z, Puri P, et al. The role of long noncoding RNA H19 in gender disparity of cholestatic liver injury in multidrug resistance 2 gene knockout mice. Hepatology. 2017;66:869–84. Liu R, Li X, Zhu W, Wang Y, Zhao D, Wang X, Gurley EC, et al. Cholangiocyte-Derived Exosomal Long Noncoding RNA H19 Promotes Hepatic Stellate Cell Activation and Cholestatic Liver Fibrosis. Hepatology. 2019;70:1317–35. Wang Y, Tai YL, Way G, Zeng J, Zhao D, Su L, Jiang X, et al. RNA binding protein HuR protects against NAFLD by suppressing long noncoding RNA H19 expression. Cell Biosci. 2022;12:172. Xiao Y, Liu R, Li X, Gurley EC, Hylemon PB, Lu Y, Zhou H, et al. Long Noncoding RNA H19 Contributes to Cholangiocyte Proliferation and Cholestatic Liver Fibrosis in Biliary Atresia. Hepatology. 2019;70:1658–73. Wu H, Kirita Y, Donnelly EL, Humphreys BD. Advantages of Single-Nucleus over Single-Cell RNA Sequencing of Adult Kidney: Rare Cell Types and Novel Cell States Revealed in Fibrosis. J Am Soc Nephrol. 2019;30:23–32. Young MD, Behjati S. SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 2020;9. Lun ATL, Riesenfeld S, Andrews T, Dao TP, Gomes T, participants in the 1st Human Cell Atlas, Marioni J. JC. EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data. Genome Biol 2019;20:63. Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. 2024;42:293–304. Team} RC. R: A Language and Environment for Statistical Computing. In: R Foundation for Statistical Computing; 2024. Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. 2019;16:1289–96. Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, et al. Inference and analysis of cell-cell communication using CellChat. Nat Commun. 2021;12:1088. Cao J, Spielmann M, Qiu X, Huang X, Ibrahim DM, Hill AJ, Zhang F, et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature. 2019;566:496–502. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33:1–22. Tay JK, Narasimhan B, Hastie T. Elastic Net Regularization Paths for All Generalized Linear Models. J Stat Softw 2023;106. Hua Y, Weng L, Zhao F, Rambow F. SeuratExtend: streamlining single-cell RNA-seq analysis through an integrated and intuitive framework. Gigascience 2025;14. Kolberg L, Raudvere U, Kuzmin I, Vilo J, Peterson H. gprofiler2 -- an R package for gene list functional enrichment analysis and namespace conversion toolset g:Profiler. F1000Res 2020;9. Andrews TS, Nakib D, Perciani CT, Ma XZ, Liu L, Winter E, Camat D, et al. Single-cell, single-nucleus, and spatial transcriptomics characterization of the immunological landscape in the healthy and PSC human liver. J Hepatol. 2024;80:730–43. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, Muller M. pROC: an open-source package for R and S + to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. Liu N, Bhuva DD, Mohamed A, Bokelund M, Kulasinghe A, Tan CW, Davis MJ. standR: spatial transcriptomic analysis for GeoMx DSP data. Nucleic Acids Res. 2024;52:e2. McInnes L, Healy J, UMAP. Uniform Manifold Approximation and Projection for Dimension Reduction. ArXiv 2018;abs/1802.03426. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. Rosenberger FA, Thielert M, Strauss MT, Schweizer L, Ammar C, Madler SC, Metousis A, et al. Spatial single-cell mass spectrometry defines zonation of the hepatocyte proteome. Nat Methods. 2023;20:1530–6. Cunningham RP, Porat-Shliom N. Liver Zonation - Revisiting Old Questions With New Technologies. Front Physiol. 2021;12:732929. Yang L, Lu P, Yang X, Li K, Qu S. Annexin A3, a Calcium-Dependent Phospholipid-Binding Protein: Implication in Cancer. Front Mol Biosci. 2021;8:716415. Zhuang C, Wang P, Sun T, Zheng L, Ming L. Expression levels and prognostic values of annexins in liver cancer. Oncol Lett. 2019;18:6657–69. Wei SC, Tan YY, Weng MT, Lai LC, Hsiao JH, Chuang EY, Shun CT, et al. SLCO3A1, A novel crohn's disease-associated gene, regulates nf-kappaB activity and associates with intestinal perforation. PLoS ONE. 2014;9:e100515. Pan Q, Zhang X, Zhang L, Cheng Y, Zhao N, Li F, Zhou X, et al. Solute Carrier Organic Anion Transporter Family Member 3A1 Is a Bile Acid Efflux Transporter in Cholestasis. Gastroenterology. 2018;155:1578–92. e1516. Udomsinprasert W, Poovorawan Y, Chongsrisawat V, Vejchapipat P, Honsawek S. Decreased circulating clusterin reflects severe liver complications after hepatoportoenterostomy of biliary atresia. Sci Rep. 2020;10:19736. Li H, Lan L, Chen H, Zaw Thin M, Ps H, Nelson JK, Evans IM et al. SPP1 is required for maintaining mesenchymal cell fate in pancreatic cancer. Nature 2025. Song Z, Chen W, Athavale D, Ge X, Desert R, Das S, Han H, et al. Osteopontin Takes Center Stage in Chronic Liver Disease. Hepatology. 2021;73:1594–608. Koch KS, Leffert HL. Ectopic expression of CD74 in Ikkbeta-deleted mouse hepatocytes. Acta Histochem. 2011;113:428–35. Wang W, Min L, Qiu X, Wu X, Liu C, Ma J, Zhang D, et al. Biological Function of Long Non-coding RNA (LncRNA) Xist. Front Cell Dev Biol. 2021;9:645647. Liu B, Li F, Wang Y, Gao X, Li Y, Wang Y, Zhou H. APP-CD74 axis mediates endothelial cell-macrophage communication to promote kidney injury and fibrosis. Front Pharmacol. 2024;15:1437113. Xing X, Cao F, Gao L, Song M. AHSG, a Gene Promoting Tumour Proliferation, Migration and Invasion, is an Independent Prognostic Factor for Poor Overall Survival in Lung Adenocarcinoma. Mol Biol Rep. 2023;50:7659–66. Zhao Y, Yang B, Chen D, Zhou X, Wang M, Jiang J, Wei L, et al. Combined identification of ARID1A, CSMD1, and SENP3 as effective prognostic biomarkers for hepatocellular carcinoma. Aging. 2021;13:4696–712. Escudero-Esparza A, Bartoschek M, Gialeli C, Okroj M, Owen S, Jirstrom K, Orimo A, et al. Correction: Complement inhibitor CSMD1 acts as tumor suppressor in human breast cancer. Oncotarget. 2023;14:481–2. Escudero-Esparza A, Bartoschek M, Gialeli C, Okroj M, Owen S, Jirstrom K, Orimo A, et al. Complement inhibitor CSMD1 acts as tumor suppressor in human breast cancer. Oncotarget. 2016;7:76920–33. Additional Declarations No competing interests reported. Supplementary Files 20251211CellBioscienceSupplementaldata.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 30 Jan, 2026 Reviews received at journal 29 Jan, 2026 Reviews received at journal 25 Jan, 2026 Reviewers agreed at journal 14 Jan, 2026 Reviewers agreed at journal 12 Jan, 2026 Reviewers invited by journal 12 Jan, 2026 Editor assigned by journal 13 Dec, 2025 Submission checks completed at journal 13 Dec, 2025 First submitted to journal 11 Dec, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8339668","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":573330591,"identity":"1b73711d-f356-4cfe-a7a3-b7a674b6518b","order_by":0,"name":"Grayson Welch Way","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Grayson","middleName":"Welch","lastName":"Way","suffix":""},{"id":573330592,"identity":"4fbb769d-d109-4824-9582-aa7c9201069e","order_by":1,"name":"Xixian Jiang","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Xixian","middleName":"","lastName":"Jiang","suffix":""},{"id":573330595,"identity":"ccee75d9-d9bb-4127-a821-9aeb9b19bab1","order_by":2,"name":"Hongkun Lu","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Hongkun","middleName":"","lastName":"Lu","suffix":""},{"id":573330597,"identity":"d84dfb6c-0c38-44ba-9dfd-b42b5b3ec5df","order_by":3,"name":"Nan Wu","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Nan","middleName":"","lastName":"Wu","suffix":""},{"id":573330598,"identity":"69f761e9-3ae0-4f0a-a204-7386e93654ad","order_by":4,"name":"Derrick Zhao","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Derrick","middleName":"","lastName":"Zhao","suffix":""},{"id":573330599,"identity":"15577aa7-3d88-4121-84c3-74b0704c614d","order_by":5,"name":"Yun-ling Tai","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Yun-ling","middleName":"","lastName":"Tai","suffix":""},{"id":573330600,"identity":"b0418914-b54a-4541-b42e-7b0340d7944d","order_by":6,"name":"Sareh Bayatpour","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Sareh","middleName":"","lastName":"Bayatpour","suffix":""},{"id":573330602,"identity":"9811e990-762a-4906-8002-31c0543e7780","order_by":7,"name":"Xuan Wang","email":"","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":false,"prefix":"","firstName":"Xuan","middleName":"","lastName":"Wang","suffix":""},{"id":573330604,"identity":"a885360e-bbb9-4868-89dc-16d4e3806d0a","order_by":8,"name":"Huiping Zhou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEUlEQVRIiWNgGAWjYBACAwbGBoYEEEsCiD+ARRgYDwAFidPCOAOiheHAQbxaYACohZmHGC3mEsltDx7uqGWQn918TNq2zS5xO3vzg8MfdzDImfcvwKrFckZiu0HimeMMjHOOpUnntiUn7uw5ZnDg4BkGY5kbD7A77EZim0Ri2zEGZokcM+mcMwcSN9xIAGppY0icIXEAvxY2kBYLkJb7zz8Qo6WGgQekhaECZAsP1BZ+7GFgcOYh0C9tB3gkJNKSLXsqko03nMkpOHC2TcJYQgJ7iBkcT3/28GdbnZz8jOSDN34Y2MluOH5844PKNhs5CX7sDgMCNiA+zIMuCrRCIgGfljpsErhtGQWjYBSMgpEFAKVPagG5HDEKAAAAAElFTkSuQmCC","orcid":"","institution":"Virginia Commonwealth University","correspondingAuthor":true,"prefix":"","firstName":"Huiping","middleName":"","lastName":"Zhou","suffix":""}],"badges":[],"createdAt":"2025-12-11 19:23:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8339668/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8339668/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100229236,"identity":"4b0e2632-6004-4d72-a188-8ef992a5f009","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78306663,"visible":true,"origin":"","legend":"","description":"","filename":"20251211CellBioscience.docx","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/77b24a7b2f5001120c378e06.docx"},{"id":100229222,"identity":"20f60aca-281f-4645-8ea2-4956ffc43033","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10636,"visible":true,"origin":"","legend":"","description":"","filename":"a5000ec1d420450e9d521593b55dfd3f.json","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/ffeae5b8caafa15baa7ce658.json"},{"id":100229260,"identity":"4381707c-322a-4f73-a084-348d5458365b","added_by":"auto","created_at":"2026-01-14 11:10:00","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15646035,"visible":true,"origin":"","legend":"","description":"","filename":"20251211CellBioscienceSupplementaldata.docx","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/b00c207433efd194aaf8c0bd.docx"},{"id":100229219,"identity":"9d17f2a6-828b-4a86-9c19-a8b49e66a1db","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":133190,"visible":true,"origin":"","legend":"","description":"","filename":"a5000ec1d420450e9d521593b55dfd3f1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/457a2e46169762b16b967eec.xml"},{"id":100229221,"identity":"41a7c24d-e059-4f57-a729-bd6f28fe60bf","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10993812,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/01064c391deb2183324487ea.jpeg"},{"id":100229263,"identity":"f575e7e3-52fe-4b50-b483-70c1f86244d6","added_by":"auto","created_at":"2026-01-14 11:10:00","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12622224,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/5dfba14c1e10857a8db70bad.jpeg"},{"id":100369618,"identity":"d385e51c-6e0c-4a7b-a648-0b91444f2e2f","added_by":"auto","created_at":"2026-01-16 07:59:11","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10154852,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/ee6c8bc0b24c80866d8eaf86.jpeg"},{"id":100229238,"identity":"e2d76f33-9a8f-44f5-a0cd-a3d9de7f3483","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14406790,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/29dba21adccb597251a765e9.jpeg"},{"id":100229242,"identity":"8254a766-c7c7-410d-9a27-28c4edc16773","added_by":"auto","created_at":"2026-01-14 11:09:57","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13120748,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/4726d061586b1352f978b5e2.jpeg"},{"id":100229243,"identity":"09383acf-aefd-46ab-bd92-6324dc6a50ab","added_by":"auto","created_at":"2026-01-14 11:09:58","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14728830,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/3dd6da50c804a98137e0682c.jpeg"},{"id":100229284,"identity":"dcb71f98-6d0b-4f28-ba71-3b8d1083d8f4","added_by":"auto","created_at":"2026-01-14 11:10:04","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9315114,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/d51cdb6bea56275ca1a3cada.jpeg"},{"id":100370214,"identity":"3037a593-856d-4285-8b8e-cd9ce795f98a","added_by":"auto","created_at":"2026-01-16 08:00:33","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":24983776,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/3fc80ded3acd5f876be520a2.jpeg"},{"id":100229288,"identity":"06c0ef33-67d9-4d2d-9353-42cd82a7eb0d","added_by":"auto","created_at":"2026-01-14 11:10:05","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1271665,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/4463c46e829d0f135bd075e6.png"},{"id":100370433,"identity":"c39bb634-da52-40fb-92dd-8fc700a9cca8","added_by":"auto","created_at":"2026-01-16 08:05:48","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1174567,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/4318d836d8088530b3231c2f.png"},{"id":100229265,"identity":"bc7788e7-b977-4c52-b708-dd18cc62ba13","added_by":"auto","created_at":"2026-01-14 11:10:01","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":973201,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/5a809aefe860f8bb2a87c0e2.png"},{"id":100229241,"identity":"f0489ab8-2af7-4f3e-b7fb-2b5f9cea0794","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1188539,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/cf1cad3e09e494021cea71e0.png"},{"id":100229223,"identity":"c813ea59-72ef-45ca-b67b-2f5d88e02b4a","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":985301,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/760828229db50794e7d03e1e.png"},{"id":100229289,"identity":"0ec7a757-c304-4c95-b9b3-cef7bb35e9a8","added_by":"auto","created_at":"2026-01-14 11:10:05","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1562173,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/5fa59faa17ceefe698da4451.png"},{"id":100229261,"identity":"ad3cf276-0ac8-4ab6-a8ce-ad463aef6ffe","added_by":"auto","created_at":"2026-01-14 11:10:00","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":813551,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/7a212e4d358344123249095b.png"},{"id":100229228,"identity":"efb42d0a-67b7-4548-a059-d2a3a5fc8459","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2325107,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/07b9f52303f290232e3af938.png"},{"id":100229237,"identity":"42baa845-a503-4562-9de3-de83000c9f58","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"xml","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":130926,"visible":true,"origin":"","legend":"","description":"","filename":"a5000ec1d420450e9d521593b55dfd3f1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/ad076bd15eb3d4f2f8e18959.xml"},{"id":100229264,"identity":"a75d6993-d237-42e1-8792-e0a9c4a2f8c7","added_by":"auto","created_at":"2026-01-14 11:10:00","extension":"html","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":147367,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/7bd4cdbd5e6e904b8f94cea2.html"},{"id":100229268,"identity":"23fd00b5-f4df-4017-bcfa-5ea29965e6bf","added_by":"auto","created_at":"2026-01-14 11:10:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":15107222,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSingle-nucleus transcriptomic atlas of healthy and cholestatic mouse livers\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A)Uniform Manifold Approximation and Projection (UMAP) visualization of the clustering of 35,488 nuclei sequenced from all samples, WT, H19KO, Mdr2KO and DKO, respectively). (B) Dot plot showing canonical marker genes used to assign cell-type identities to each cluster. (C) Bar plots depicting (i) the proportional representation of each cluster by genotype and (ii) the absolute nuclei counts for each cell type.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/0e385515ccd7de3cb4b43a4e.png"},{"id":100229239,"identity":"c0eb9e27-7196-4972-9e96-fe44caeb4e02","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":11423961,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003esnRNAseq reveals disease-associated cholangiocyte subtype.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) UMAP visualization of subclustered cholangiocytes across all genotypes.\u003cbr\u003e\n (B) Pseudotime trajectory of cholangiocyte subclusters, with color denoting progression along the disease axis from healthy (purple) to late-stage disease (yellow).\u003cbr\u003e\n (C) Bar plots showing the proportional distribution of cholangiocyte subtypes across samples.\u003cbr\u003e\n (D) Violin plots illustrating \u003cem\u003eCsmd1\u003c/em\u003eexpression—the top marker of cluster 2—displayed (left) across all clusters, (middle) across clusters within each genotype, and (right) by cluster stratified by genotype. (*\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.05, **\u003cem\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/em\u003e\u0026lt; 0.01, ***\u003cem\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/em\u003e\u0026lt; 0.001, ****\u003cem\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/em\u003e\u0026lt; 0.0001; Wilcoxon test implemented \u003cem\u003evia \u003c/em\u003eSeurat Extend.)\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/c90fae6052058bba33600bc2.png"},{"id":100229220,"identity":"706ee83c-f881-4f24-bf88-8047e4a09774","added_by":"auto","created_at":"2026-01-14 11:09:54","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":12364907,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eH19 deletion significantly ameliorates disease-associated cell-signaling alterations.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRelative information flow plots showing pathways significantly altered between Mdr2KO and WT (A) or DKO (B) mice. Colored pathways denote significant changes (peach = increased, teal = decreased in Mdr2KO). Scatter (C) and circle (D) plots illustrating SPP1 signaling dynamics. In healthy mice, Spp1 is primarily expressed by cholangiocytes, but shifts to mid-lobular hepatocytes in Mdr2KO livers; H19 deletion markedly reduces this ectopic hepatocyte expression. Line color indicates ligand source, and line width represents interaction strength.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/56ddc110f6ae32ec5ef1e7ec.png"},{"id":100229244,"identity":"fea6bad6-4c39-450b-ab6d-39bb4f15d149","added_by":"auto","created_at":"2026-01-14 11:09:58","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":12257291,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAltered signaling pathways and genes distinguishing cholangiocyte disease status.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) (i) Relative contribution of the SPP1 signaling pathway comparing Mdr2KO and WT mice. (ii) Net-signaling role dot plot showing predicted SPP1 ligand-receptor interactions; dot color indicates communication probability (brown = maximum, blue = minimum). Blue and peach labels denote interactions enriched in WT and Mdr2KO, respectively. (B) Chord plots depicting cell-to-cell communication for each predicted ligand-receptor pair, separated by WT (i) and Mdr2KO (ii). Line color indicates ligand source; line thickness reflects interaction strength. Differential expression and LASSO analyses identified six genes for disease-prediction modeling. (C) Violin plots showing expression of these genes in mice. H19 deletion significantly reversed expression of disease-associated genes (\u003cem\u003eSlco3a1\u003c/em\u003e, \u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eCftr\u003c/em\u003e, \u003cem\u003eCd44\u003c/em\u003e) toward WT levels, except \u003cem\u003eCd44\u003c/em\u003e. (*p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001, ****p \u0026lt; 0.0001; Wilcoxon test, SeuratExtend).\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/20d3f853404cbf985deb87a0.png"},{"id":100369589,"identity":"e9bf05ea-899a-4f3e-afc1-595f7eac9037","added_by":"auto","created_at":"2026-01-16 07:59:09","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":8765683,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCross-species validation of cholangiocyte disease-prediction genes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMouse-derived model genes were tested on public human cholangiocyte data (GSE243981) subclustered from sc/snRNA-seq samples integrated with Harmony to correct for sequencing-method batch effects. (A) Violin plots showing expression of prediction-model genes across neurologically deceased donors (NDD), primary biliary cholangitis (PBC), and primary sclerosing cholangitis (PSC) patients using combined sc/snRNA-seq data. (B) Violin plots showing expression using only scRNA-seq samples (*p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001, ****p \u0026lt; 0.0001; Wilcoxon test, SeuratExtend). (C) ROC–AUC analysis showing machine-learning model performance in distinguishing healthy versus diseased cholangiocytes in mice (left) and humans (right). Human data testing set contained both snRNAseq and scRNAseq samples.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/e3cfcb0a5d5a7f0d17a324db.png"},{"id":100229240,"identity":"28ff8630-4932-4da5-9a73-86c01634d884","added_by":"auto","created_at":"2026-01-14 11:09:56","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":14823875,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHepatocyte disease status distinguishing genes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA) UMAP of mouse snRNAseq subclustered hepatocytes with all samples combined and B) split by genotype. C) Violin plots showing significantly different expression of each gene in mice comparing WT and H19KO mice versus Mdr2KO and DKO mice. D) Violin plots of hepatocyte disease prediction modeling genes split by genotype (*p\u0026lt;0.05, **p\u0026lt;0.01, ***p\u0026lt;0.001, ****p\u0026lt;0.0001, significance determined by Wilcoxon test using SeuratExtend).\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/2373721b6d5ecbd32f9c0a3c.png"},{"id":100371032,"identity":"653a8050-7a6d-4c4d-b778-c8f9f3e40f30","added_by":"auto","created_at":"2026-01-16 08:09:13","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":7673436,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCross-species validation of hepatocyte disease-prediction genes.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTesting of hepatocyte disease prediction model genes derived from mouse snRNAseq data on public human dataset GSE243981. Data was integrated using Harmony for known batch variables including RNA sequencing methods (scRNAseq and snRNAseq). A) Violin plots showing expression levels of hepatocyte disease prediction modeling genes in neurologically deceased donors (NDD), primary biliary cholangitis (PBC), and primary sclerosing cholangitis (PSC) patients combining snRNAseq and scRNAseq data. B) Violin plots depicting expression of hepatocyte prediction genes only using scRNAseq samples (*p\u0026lt;0.05, **p\u0026lt;0.01, ***p\u0026lt;0.001, ****p\u0026lt;0.0001, significance determined by Wilcoxon test using SeuratExtend). C) ROC-AUC analysis depicting MLMs effectiveness in distinguishing healthy and diseased hepatocytes in mice (left) and humans (right). Human data testing set contained both snRNAseq and scRNAseq samples.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/4c577e5770e2c6ecb51b33e1.png"},{"id":100229287,"identity":"5199e2cf-8629-4f0f-9d80-cab8da34bbe7","added_by":"auto","created_at":"2026-01-14 11:10:04","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":21963665,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGeoMx spatial transcriptomics reveals H19 deletion’s amelioration of disease associated gene expression is spatially restricted.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Representative immunofluorescence images illustrating GeoMx DSP region-of-interest (ROI) selection. (B) Volcano plots showing genes significantly altered by H19 deletion in Mdr2KO mice within hepatocytes adjacent to bile ducts (left) or isolated hepatocytes (right). (C) Volcano plots showing differential expression of hepatocyte disease-prediction genes depending on proximity to bile ducts (left) versus non-adjacent regions (right). Significance thresholds: adjusted p \u0026lt; 0.05 and |log₂FC| \u0026gt; 0.5, determined by both DESeq2 and StandR limma-voom analyses with RUV4 batch correction; p-values were adjusted using the Benjamini–Hochberg method.\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/e8859b507e1a0fa374b01941.png"},{"id":100229166,"identity":"c3c424db-3b54-40e0-bbfe-37d0b7ef9928","added_by":"auto","created_at":"2026-01-14 11:09:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":991698,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/824e96e0-6cf2-4ca1-8134-ac5776d8239a.pdf"},{"id":100229286,"identity":"45ff1aa7-c96d-4a1c-80e4-fb108adddd67","added_by":"auto","created_at":"2026-01-14 11:10:04","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":15646035,"visible":true,"origin":"","legend":"","description":"","filename":"20251211CellBioscienceSupplementaldata.docx","url":"https://assets-eu.researchsquare.com/files/rs-8339668/v1/06250b4993119d355cd890fc.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Decoding the Role of H19 in Cholestatic Liver Injury Using snRNA-seq, Spatial Transcriptomics, and Machine Learning-Based Disease Prediction","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePrimary sclerosing cholangitis (PSC) is a chronic, progressive hepatobiliary disease characterized by biliary fibrosis, obstructive cholestasis and inflammation of the intrahepatic and extrahepatic bile ducts(\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). The disruption of bile acid homeostasis contributes to a cascade of metabolic and immune dysregulation, as bile acids are important signaling molecules. PSC progresses along a spectrum with eventual development of end-stage liver failure(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). PSC is also associated with a markedly increased risk of malignancies, such as cholangiocarcinoma, gallbladder carcinoma, hepatocellular carcinoma, and colorectal carcinoma(\u003cspan additionalcitationids=\"CR6 CR7 CR8 CR9\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). In the United States, PSC is the fifth leading indication for liver transplantation(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Furthermore, recurrence occurs in up to 30% of transplant recipients(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Despite the severity of the disease, no proven pharmacotherapies exist, largely due to an incomplete understanding of the cellular and molecular mechanisms underlying PSC pathogenesis(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe long non-coding RNA H19 (H19) has emerged as an important regulator of PSC disease progression(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Previous research in our lab, and others, have shown that H19, which is normally not expressed in healthy hepatic tissues, is upregulated in PSC and exacerbates disease progression in Mdr2KO mice (the gold standard mouse model for PSC)(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan additionalcitationids=\"CR15 CR16 CR17 CR18\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). However, the field lacks a mechanistic framework explaining how H19 reprograms specific hepatic cell populations, whether its effects are spatially restricted in the liver microenvironment, and which transcriptional networks are directly or indirectly governed by H19 in vivo. It is also unclear how diseae-assocaited cell states in mice correspond to those in human cholestatic liver disease, and whether concise molecular signatures can reliably classify cholangiocyte or hepatocyte health status across species. To address these gaps, we leveraged single-nucleus RNA sequencing (snRNA-seq) and spatial transcriptomic profiling across WT, H19KO, Mdr2KO and DKO mice. While single-cell RNA sequencing (scRNAseq) has advantages for profiling immune cells, snRNAseq offers several distinct benefits, including compatibility with frozen tissue, reduced dissociation-induced artifacts, and superior accuracy in profiling hepatocytes and cholangiocytes (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Spatial transcriptomic technology, such as GeoMx\u0026reg; Digital Spatial Profiler (DSP), allows transcriptomic analysis at specific regions based on histological morphology.\u003c/p\u003e \u003cp\u003eIn our previous work, we demonstrated that genetic deletion of H19 significantly reduced liver fibrosis and slowed disease progression in Mdr2KO mice(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). In this study, integration of snRNAseq and GeoMx DSP enabled us to 1) define disease-specific cholangiocyte and hepatocyte states associated with PSC, 2) map how h19 deletion reshapes these states at cellular and spatial levels, and 3) quantify the spatial restriction of H19-mediated transcriptional rescue. We further integrated publicly available human scRNA-seq and snRNA-seq datasets to evaluate cross-species conservation and built machine-learning prediction models based on minimal gene sets capable of classifying cholangiocyte and hepatocyte disease status with high accuracy.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eAnimal Studies\u003c/h2\u003e \u003cp\u003eC57BL/6J Mdr2KO mice were gifted by Dr. Daniel Goldenberg, Department of Pathology, Hadassah-Hebrew University Medical Center, Jerusalem, Israel. Maternal C57BL/6J H19\u003csup\u003eΔExon1/+\u003c/sup\u003e (H19KO) mice were generated and gifted by Dr. Karl Pfeifer at the NIH. Dr. Jian-Ying Wang at the University of Maryland (Baltimore, MD, USA) provided the H19KO mice. Mdr2KO and H19KO (DKO) mice were generated as previously described (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). The gender and age-matched littermates of WT, H19KO, Mdr2KO and DKO (female, 6-month-old) were used. Mice were housed under 12 h/12 h light and dark cycle with unrestricted access to water and standard chow \u003cem\u003ead libitum\u003c/em\u003e. All animal experiments followed institutional guidelines for ethical animal studies and were approved by VCU institutional animal care and use committee.\u003c/p\u003e \u003cp\u003eFor GeoMx spatial transcriptomics, eight liver samples were collected from 6-month-old female mice (four Mdr2⁻/⁻ and four DKO). For single-nucleus RNA sequencing (snRNA-seq), liver tissues from 6-month-old female mice (one per genotype: WT, H19KO, Mdr2KO, and DKO) were flash-frozen in liquid nitrogen and stored at \u0026minus;\u0026thinsp;80\u0026deg;C until processing.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003escRNAseq and snRNAseq data analysis\u003c/h3\u003e\n\u003cp\u003escRNAseq and snRNAseq analyses of both our dataset and the GSE243981 dataset were performed in R (version 4.3.3) utilizing the R package Seurat (version 5.1.0)(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). For our mouse snRNA-seq data, nuclei with \u0026gt;\u0026thinsp;2.5% mitochondrial gene content were excluded. The R package SoupX (version 1.6.2) was used to remove ambient RNA contamination(\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). For GSE243981, mitochondrial content filters of 5% (snRNA-seq) and 25% (scRNA-seq) were applied. Data normalization and scaling were performed using Seurat\u0026rsquo;s SCTransform function and data integration was performed utilizing Harmony, accounting for known batch variables (i.e., different experimental techniques used: scRNAseq, snRNAseq, 3\u0026rsquo; and 5\u0026rsquo; sequencing) following the Seurat SCTransform workflow(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). Predicted cell-cell communication were inferred using the R package CellChatv2 (version 2.1.0)(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). Pseudotime analyses for cell type progression with disease was performed using Monocle3 (version 1.4.26)(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). LASSO regression analysis was performed using the R package glmnet (version 4.1.10)(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). Data visualization and statistical analyses were performed with SeuratExtend (version 1.2.5)(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). Statistical significance was determined using the MAST algorithm via Seurat\u0026rsquo;s FindMarkers function.\u003c/p\u003e\n\u003ch3\u003eMachine learning Modeling Methods\u003c/h3\u003e\n\u003cp\u003eTo assess the robustness and predictive accuracy of input features, four supervised machine learning models were developed and evaluated: a multilayer perceptron (MLP) neural network, an extreme gradient boosting (XGBoost) classifier, a random forest classifier, and a logistic regression model. For initial validation, models were trained and tested on mouse data using a 70/30 train\u0026ndash;test split. For cross-species validation, each model was trained on the complete mouse dataset and subsequently tested on the entire human dataset (GSE243981). Mouse gene symbols were converted to their human orthologs using the gorth function in the gprofiler2 R package to enable translational comparison(\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). All models were trained to classify binary health status labels (healthy vs. diseased) and evaluated using the area under the receiver operating characteristic curve (ROC-AUC) as the primary performance metric with the R package pROC(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eGeoMx Data Analysis\u003c/h3\u003e\n\u003cp\u003eGeoMx\u0026reg; data analysis was performed using the StandR R package following their pipeline(\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). QC was performed at both the region of interest (ROI) level and gene level to remove low-quality data. ROIs were filtered based on nuclear count, surface area, and library size using default thresholds. PCA was performed using the runPCA function from the scater package (version 1.36.0), and clustering was visualized using Uniform Manifold Approximation and Projection (UMAP)(\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e). Different normalization (TMM, CPM, Upper quartile) and batch correction methods (RUV4, LimmaRemoveBatch, SVA) were tested. Ultimately, TMM normalization and RUV4 batch correction were used based on PCA and cluster separation statistics. Differential gene expression (DGE) analysis was performed using the limma R package (version 3.64.3)(\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Significance was also tested and confirmed using DESeq2 (Version 1.48.1)(\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e).\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of a disease-specific cholangiocyte cluster\u003c/h2\u003e \u003cp\u003eInitial clustering and cell type annotation of the snRNAseq data identified 16 distinct clusters from 35,488 nuclei across all four genotypes. UMAP visualization confirmed proper integration across all genotypes and experimental techniques (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Cell identity was assigned based on the expression of canonical maker genes, demonstrating high fidelity clustering (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). Compared to WT and H19KO, Mdr2KO and DKO mice exhibited drastic increases in cholangiocytes, fibroblasts, lymphocytes, and monocyte-derived macrophages (MdMQs) in terms of numbers and percent cell composition, accompanied by a reduction in hepatocytes and Kupffer cells, compared to their healthy counterparts (cholangiocytes, MdMQs and hepatocytes percentage of cells p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Subclustering of cholangiocytes identified a disease-associated cholangiocyte cluster (cluster 2) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Pseudotime trajectory analysis using monocle3 indicated that Cluster 2 corresponded to a late-stage disease state (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Notably, H19 deletion in Mdr2KO mice significantly reduced both the number and percentage of cholangiocytes within this cluster (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The top marker gene for cluster 2 was \u003cem\u003eCsmd1\u003c/em\u003e (CUB And Sushi Multiple Domains 1), which was markedly upregulated in disease and downregulated upon H19 deletion (\u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e \u003cb\u003eand\u003c/b\u003e Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCellular pathways altered in PSC and ameliorated by H19 deletion\u003c/h3\u003e\n\u003cp\u003eCellChatv2 analysis was performed to identify significant cell-cell signaling network and crosstalk interactions altered in Mdr2KO mice and to evaluate pathways modulated by H19 deletion. The total number of inferred interactions identified for WT, H19KO, Mdr2KO, and DKO were 255, 180, 573, and 438, respectively (\u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Among the pathways examined, Spp1 (osteopontin), collagen, laminin, and FN1 (fibronectin) signaling were all significantly upregulated in Mdr2KO compared to WT and were normalized toward WT in DKO mice (p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA-B, \u003cb\u003eSupp.\u003c/b\u003e Figures\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Notably, Spp1 emerged as a major signaling pathway dysregulated in cholestatic liver injury and restored by H19 deletion. In healthy liver, \u003cem\u003eSpp1\u003c/em\u003e expression is predominately observed in cholangiocytes, mediating signaling to Kupffer cells. In contrast, during disease progression, Spp1 expression shifts to mid-lobular hepatocytes, indicating altered paracrine signaling (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC\u003cb\u003e- D, Supp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA-B). H19 deletion significantly reduces Spp1 expression and outgoing Spp1-mediated signaling from mid-lobular hepatocytes (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC-D). Ligand-receptor analysis revealed that the top predicted interactions in the Spp1 pathway were Spp1-Cd44, Spp1-(Itgav\u0026thinsp;+\u0026thinsp;Itgb6), and Spp1-(Itga4\u0026thinsp;+\u0026thinsp;Itgb1). Notably, Spp1-Cd44 signaling between cholangiocytes and Kupffer cells was present only in healthy liver and was lost in Mdr2KO mice, whereas the other Spp1 interactions were uniquely predicted in Mdr2KO mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA-B).\u003c/p\u003e\n\u003ch3\u003eDiseased cholangiocyte gene prediction model and identification of novel healthy markers\u003c/h3\u003e\n\u003cp\u003eDifferential gene expression (DGE) analysis of subclustered cholangiocytes, combined with LASSO regression-based feature selection, identified six genes (\u003cem\u003eClu\u003c/em\u003e, \u003cem\u003eSpp1\u003c/em\u003e, \u003cem\u003eSlco3a1\u003c/em\u003e, \u003cem\u003eCd44\u003c/em\u003e, \u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eCftr\u003c/em\u003e) for machine learning-based prediction modeling of cholangiocyte disease status (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Among these, \u003cem\u003eClu\u003c/em\u003e and \u003cem\u003eSpp1\u003c/em\u003e were strongly associated with healthy cholangiocytes, whereas \u003cem\u003eSlco3a1\u003c/em\u003e, \u003cem\u003eCd44\u003c/em\u003e, \u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eCftr\u003c/em\u003e were enriched in diseased cholangiocytes in Mdr2KO and DKO mice (adj. p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Three of the disease associated genes (\u003cem\u003eSlco3a1\u003c/em\u003e, \u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eCftr\u003c/em\u003e) were significantly reduced towards WT levels in DKO mice, indicating that H19 deletion ameliorates their dysregulation (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). A striking spatial shift was observed for \u003cem\u003eSpp1\u003c/em\u003e and \u003cem\u003eClu\u003c/em\u003e, which transitioned from cholangiocyte-restricted expression in healthy liver to ectopic expression in mid-lobular hepatocytes in diseased liver, suggesting substantial rewiring of cellular identity and intercellular signaling during disease progression (\u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA\u0026ndash;B).\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCholangiocyte disease prediction model testing in humans\u003c/h2\u003e \u003cp\u003eThe expression patterns of six model genes (\u003cem\u003eClu\u003c/em\u003e, \u003cem\u003eSpp1\u003c/em\u003e, \u003cem\u003eSlco3a1\u003c/em\u003e, \u003cem\u003eCd44\u003c/em\u003e, \u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eCftr\u003c/em\u003e) were evaluated in human cholangiocytes using the GSE243981 dataset (both scRNAseq and snRNAseq). Five of the six genes demonstrated the same significant expression trends in PSC patient cholangiocytes as observed in mice (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05), with \u003cem\u003eCd44\u003c/em\u003e as the exception when combining scRNAseq and snRNAseq datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA), but was significantly different in scRNAseq data alone (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Cftr expression was significantly decreased in the combined data (sc/snRNAseq) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA), but did not reach significance in the scRNAseq-only analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). These six genes were subsequently used to construct build cholangiocyte disease prediction models using four supervised machine learning algorithms: multilayer perceptron (MLP) neural network, XGBoost, random forest, and logistic regression. ROC\u0026ndash;AUC analysis demonstrated robust performance, with AUC values\u0026thinsp;\u0026gt;\u0026thinsp;0.92 in mouse datasets and \u0026gt;\u0026thinsp;0.87 in human cholangiocytes from GSE243981, supporting the translational relevance of the gene signatures and the predictive models (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDiseased Hepatocytes gene prediction model and identification of a novel healthy marker\u003c/h2\u003e \u003cp\u003eHepatocytes were subclustered, zonally differentiated, and UMAPs generated to visualize proper data clustering and integration (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA-B, \u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e)(\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). Differential gene expression combined with LASSO regression identified six genes\u0026mdash;\u003cem\u003eGm13775, Spp1, C6, Cdh1, Npas2\u003c/em\u003e, and \u003cem\u003eCd74\u003c/em\u003e\u0026mdash;for construction of a hepatocyte disease prediction model using a random forest classifier, which achieved an AUC of 0.916 in mice (\u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). Gm13775 emerged as the top-performing gene and a novel marker of healthy hepatocytes (\u003cb\u003eSupp.\u003c/b\u003e Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB\u0026ndash;C). However, because no human homolog has been identified, \u003cem\u003eGm13775\u003c/em\u003e was excluded from downstream cross-species model development. A revised six-gene model (\u003cem\u003eSpp1\u003c/em\u003e, \u003cem\u003eC6\u003c/em\u003e, \u003cem\u003eNpas2\u003c/em\u003e, \u003cem\u003eCdh1, Cd74\u003c/em\u003e, and \u003cem\u003eHamp\u003c/em\u003e) was generated for comparative analyses. In Mdr2KO hepatocytes \u003cem\u003eSpp1, C6, Npas2\u003c/em\u003e, \u003cem\u003eCdh1\u003c/em\u003e, and \u003cem\u003eCd74\u003c/em\u003e were all significantly upregulated, while \u003cem\u003eHamp\u003c/em\u003e was significantly reduced relative to WT (adj. p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC). In DKO mice, \u003cem\u003eSpp1, Cdh1\u003c/em\u003e, and \u003cem\u003eCd74\u003c/em\u003e were significantly decreased and \u003cem\u003eHamp\u003c/em\u003e significantly increased toward WT levels (adj. p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), indicating that \u003cem\u003eH19\u003c/em\u003e deletion partially corrects hepatocyte dysregulation (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). CellChat v2 pathway analysis further revealed that APP signaling was significantly reduced in DKO mice (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Across all genotypes, the dominant ligand\u0026ndash;receptor interaction was App\u0026ndash;Cd74, with mid-lobular hepatocytes showing increased CD74 expression and receiving App-derived signaling in disease states (\u003cb\u003eSupp. Figure\u0026nbsp;9\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eHepatocyte disease prediction model testing in humans\u003c/h2\u003e \u003cp\u003eThe six genes used in the hepatocyte disease prediction model (Hamp, Spp1, C6, Cdh1, Npas2, and Cd74\u003cb\u003e)\u003c/b\u003e demonstrated strong translational concordance in human hepatocytes from the GSE243981 dataset. All six genes showed consistent and significant expression changes in both the combined scRNAseq/snRNAseq dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA) and in the scRNAseq subset alone (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). When tested across four supervised machine learning algorithms, these six genes achieved AUC values\u0026thinsp;\u0026gt;\u0026thinsp;0.869 in both mouse and human datasets, confirming their cross-species robustness and strong predictive performance (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eDeletion of H19 causes cell-type specific alterations in gene expression\u003c/h2\u003e \u003cp\u003eRandom forest modeling was performed across multiple cell types to identify gene expression patterns associated with \u003cem\u003eH19\u003c/em\u003e deletion. Xist consistently emerged as the top predictive feature across all models (\u003cb\u003eSupp. Figure\u0026nbsp;10\u003c/b\u003e). Although Xist/Tsix dominated feature importance, each cell type also showed distinct \u003cem\u003eH19\u003c/em\u003e-dependent transcriptional signatures.\u003c/p\u003e \u003cp\u003eIn hepatocytes, a 13 gene model (\u003cem\u003eXist\u003c/em\u003e, \u003cem\u003eGphn\u003c/em\u003e, \u003cem\u003eFga\u003c/em\u003e, \u003cem\u003eTsix\u003c/em\u003e, \u003cem\u003eMalat1\u003c/em\u003e, \u003cem\u003eAhsg\u003c/em\u003e, \u003cem\u003eApob\u003c/em\u003e, \u003cem\u003eCes1c\u003c/em\u003e, \u003cem\u003eSpp1\u003c/em\u003e, \u003cem\u003eCers6\u003c/em\u003e, \u003cem\u003eCd74\u003c/em\u003e, \u003cem\u003eSrsf1\u003c/em\u003e, and \u003cem\u003eHspa2\u003c/em\u003e) achieved an AUC score of 0.972 for distinguishing H19 deletion from WT mice (\u003cb\u003eSupp. Figure\u0026nbsp;10\u003c/b\u003e). Excluding \u003cem\u003eXist\u003c/em\u003e and \u003cem\u003eTsix\u003c/em\u003e modestly reduced performance (AUC\u0026thinsp;=\u0026thinsp;0.902). In cholangiocytes, a 13 gene model (\u003cem\u003eXist, Tsix, Csmd1, Chka, Frmd4b, Gnas, Npas2, Agmo, Ly6e, Gbp9, Sorbs3, Usp2\u003c/em\u003e, and \u003cem\u003e1600012H06Rik\u003c/em\u003e) achieved an AUC score of 0.934 (0.867 without \u003cem\u003eXist\u003c/em\u003e and \u003cem\u003eTsix\u003c/em\u003e). In macrophages, a 12 gene model (\u003cem\u003eXist, Tsix, Mat1a, Gm42418, Eef1a1, Rpl13a, Gm19951, Clec2d, Wwc1, Rnf128, Ugt2b5\u003c/em\u003e, and \u003cem\u003eTnfrsf12a\u003c/em\u003e) generated an AUC score of 0.861, decreasing to 0.764 without \u003cem\u003eXist\u003c/em\u003e and \u003cem\u003eTsix.\u003c/em\u003e Importantly, several genes within these models overlapped with those reversed by H19 deletion-that is, genes upregulated (or downregulated) in Mdr2KO vs. WT and reversed toward WT levels in DKO mice, indicating correction of PSC-associated dysregulation. Notable examples include \u003cem\u003eSpp1\u003c/em\u003e and \u003cem\u003eCd74\u003c/em\u003e in hepatocytes, and \u003cem\u003eFrmd4b\u003c/em\u003e, \u003cem\u003eGnas\u003c/em\u003e, \u003cem\u003eChka\u003c/em\u003e, \u003cem\u003eCsmd1\u003c/em\u003e, and \u003cem\u003eAgmo\u003c/em\u003e in cholangiocytes (\u003cb\u003eSupp. Tables.1\u0026amp;2\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eSpatial transcriptomics reveal that H19 deletion-mediated amelioration in hepatocytes is restricted to hepatocytes in close proximity to bile ducts\u003c/b\u003e \u003c/p\u003e \u003cp\u003eSpatial transcriptomics was utilized to investigate spatially defined effects of H19 deletion on hepatic gene expression. ROIs were selected to compare differences in isolated hepatocytes and hepatocytes in close proximity to bile ducts (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). DGE analysis of the different ROIs comparing DKO vs Mdr2KO mice yielded several common and unique significant DEGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). Common genes decreased in DKO mice vs Mdr2KO across both regions were genes associated with cellular stress responses, such as \u003cem\u003eHspa5\u003c/em\u003e, \u003cem\u003eHspa8\u003c/em\u003e, and \u003cem\u003eHsp90aa1\u003c/em\u003e (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01 across both regions), and fatty acid synthesis gene \u003cem\u003eFasn\u003c/em\u003e (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01); however, several cholesterol metabolism genes were only significantly reduced in hepatocytes neighboring bile ducts (e.g., \u003cem\u003eApoa1\u003c/em\u003e, \u003cem\u003eApoa2\u003c/em\u003e, and \u003cem\u003eApoa5\u003c/em\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). Several genes related to significant disease-associated pathways (APP, Collagen, FN1, LAMININ) ameliorated by H19 deletion as well as hepatocyte-to-cholangiocyte transition marker genes (\u003cem\u003eSox4\u003c/em\u003e and \u003cem\u003eSox9\u003c/em\u003e) are only significantly reduced in regions with hepatocytes neighboring bile ducts (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). Four of the five hepatocyte disease-associated prediction modeling genes were only significantly ameliorated in hepatocytes neighboring cholangiocytes (\u003cem\u003eSpp1, C6, Cdh1\u003c/em\u003e, and \u003cem\u003eCd74\u003c/em\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eC), demonstrating that H19 deletion-mediated transcriptional recovery is spatially restricted to periductal hepatocytes.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eUnderstanding how PSC reshapes hepatobiliary cell states at single-cell and spatial resolution has been a longstanding barrier in the field. Although H19 has been implicated in cholestatic injury, the mechanisms by which it influences specific cellular trajectories and the spatial architecture of disease have remained unresolved. (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan additionalcitationids=\"CR15 CR16 CR17 CR18\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). In this study, supported by findings from \u003cem\u003eAndrews, et. al\u003c/em\u003e, our integrative snRNA-seq, spatial transcriptomic, and machine-learning strategy addresses these gaps and provides several conceptual advances that reframe the role of H19 in PSC pathogenesis (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eA central finding of this study is the identification of a disease-specific cholangiocyte population defined by Csmd1 and Lama4 expression that emerges during PSC progression and is substantially reduced by H19 deletion. This subpopulation aligns with a late-stage pseudotime state, establishing a mechanistic link between H19 activity and cholangiocyte trajectory convergence toward a pathogenic cell identity. The enrichment of Lama4, an extracellular matrix component with known profibrotic functions, further connects H19 biology to matrix remodeling and fibrosis.\u003c/p\u003e \u003cp\u003eUsing multimodal approaches, we identified six genes that robustly distinguished healthy from diseased cholangiocytes across scRNA-seq and combined sc/snRNA-seq datasets. Notably, the simplest model, logistic regression, achieved strong performance, suggesting these genes define a relatively linear health vs disease axis. Several disease-specific genes (\u003cem\u003eAnxa3\u003c/em\u003e, \u003cem\u003eSlco3a1\u003c/em\u003e, and \u003cem\u003eCftr\u003c/em\u003e) were significantly reduced with H19 deletion (adj. p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, DKO vs Mdr2KO) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF). \u003cem\u003eAnxa3\u003c/em\u003e encodes Annexin A3, a calcium-dependent phospholipid-binding protein elevated in liver cancers and explored as a diagnostic and therapeutic target (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). \u003cem\u003eSlco3a1\u003c/em\u003e (OATP3A1), a sodium-independent organic anion transporter linked to Crohn\u0026rsquo;s disease and NF-κB activation (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e), also mediates bile-acid efflux and has been proposed as a protective adaptive response in cholestasis (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). Although \u003cem\u003eSlco3a1\u003c/em\u003e expression was reduced in DKO vs Mdr2KO (adj. p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), it was unchanged in H19KO vs WT (p\u0026thinsp;\u0026gt;\u0026thinsp;0.05), indicating that its down-regulation likely reflects reduced cholestasis rather than direct H19 regulation.\u003c/p\u003e \u003cp\u003eTwo markers of healthy cholangiocytes, Clu and Spp1, showed striking disease-associated shifts: both were downregulated in cholangiocytes yet ectopically activated in hepatocytes in injured livers. Notably, reduced circulating CLU (clusterin) levels correlate with worse outcomes in biliary atresia, underscoring the clinical significance of losing these homeostatic markers (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). Despite its pathogenic role, H19 did not preserve healthy cholangiocyte identity and instead further suppressed Spp1 expression (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Beyond cholestatic liver disease, Spp1 signaling drives mesenchymal transition in pancreatic ductal adenocarcinoma, and its inhibition reduces tumor burden and improves survival (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e). In healthy liver, cholangiocytes are the dominant source of Spp1 (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e). Here, however, we find that during cholestatic injury Spp1 signaling is elevated but largely reprogrammed to originate from hepatocytes, suggesting a pathologic shift in intercellular communication. This hepatocyte-derived Spp1 may promote hepatic stellate cell (HSC) activation and myofibroblast transition, while suppression of this aberrant hepatocyte signal could limit fibrogenesis without compromising physiologic Spp1 function in cholangiocytes.\u003c/p\u003e \u003cp\u003esnRNA-seq analysis uncovered distinct, cell-type\u0026ndash;specific transcriptional alterations driven by cholestatic disease and modulated by H19 deletion. Diseased hepatocytes exhibited ectopic activation of genes normally restricted to other healthy cell types, including Spp1 and Clu (cholangiocytes) and Cd74 (antigen-presenting cells). H19 deletion reversed these aberrant expression patterns: Clu and Spp1 were significantly elevated in Mdr2KO compared with WT, but markedly reduced in DKO relative to Mdr2KO (adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Supp. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). In alignment with prior findings linking hepatic Cd74 up-regulation to Ikbkb loss, Mdr2KO hepatocytes demonstrated significantly decreased Ikbkb expression compared with WT (adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.001)(\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e). Both Cd74 induction and Ikbkb suppression were normalized in DKO hepatocytes (adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Supp. Table\u0026nbsp;3), indicating that H19 influences hepatocyte immune signaling pathways. Finally, we identified a previously unrecognized healthy hepatocyte marker, Gm13775, a non-coding RNA with no known homologs. Its biological role remains undefined and warrants further functional investigation.\u003c/p\u003e \u003cp\u003eA novel and unexpected discovery was the strong positive association between H19 and the X-chromosome inactivation regulators Xist and its antisense transcript Tsix. Xist expression was significantly reduced across all cell types in H19KO vs WT, DKO vs Mdr2KO, and combined DKO/H19KO vs Mdr2KO/WT comparisons (adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This reduction was evident both in magnitude (log-fold change) and in the fraction of cells expressing Xist. For example, in cholangiocytes, Xist-positive cells decreased from \u0026gt;\u0026thinsp;70% to \u0026lt;\u0026thinsp;6% in H19KO vs WT and DKO/H19KO vs Mdr2KO/WT comparisons. This relationship suggests a previously unrecognized regulatory connection between H19 and the Xist\u0026ndash;Tsix axis. Perturbation of this epigenetic network could generate pseudo\u0026ndash;copy number\u0026ndash;like expression shifts in females and may contribute to sex-specific transcriptional differences in cholestatic liver disease (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e).Further mechanistic dissection of this interaction is clearly warranted.\u003c/p\u003e \u003cp\u003eWe identified multiple cell type\u0026ndash;specific differentially expressed genes influenced by H19. Machine-learning models distinguishing H19 WT from H19 knockout cells performed strongly across all major cell populations (AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.85 when including Xist), underscoring robust and highly cell type\u0026ndash;specific transcriptional signatures. In hepatocytes, Spp1 and Cd74\u0026mdash;two key disease-associated genes\u0026mdash;were clearly regulated by H19, with H19 deletion restoring their expression toward WT levels. Of note, the APP\u0026ndash;CD74 signaling axis has been implicated in promoting fibrosis in the kidney (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e), raising the possibility of a conserved H19-dependent fibrogenic mechanism across organ systems. Additional hepatocyte-specific H19-responsive genes included the lipid metabolism regulators Cers6, Ces1c, and Apob, as well as Ahsg, all of which were downregulated following H19 deletion. Elevated Ahsg expression has previously been linked to tumor proliferation (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e), suggesting broader relevance to liver tumorigenesis. In cholangiocytes, H19 deletion influenced several genes, including Frmd4b, Chka, Csmd1, Gnas, and Agmo. While increased Csmd1 expression has been associated with hepatocellular carcinoma (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e), other studies propose that CSMD1 may function as a tumor suppressor depending on cellular and disease context (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e), highlighting the complexity of its role in cholangiocyte biology.\u003c/p\u003e \u003cp\u003eGeoMx spatial transcriptomics demonstrated that the improvement in hepatocyte disease-specific gene expression following H19 deletion was spatially restricted to hepatocytes adjacent to bile ducts. This pattern is consistent with prior reports showing that cholangiocytes are the primary source of H19 during cholestatic injury and that H19 can be transferred to neighboring cells via extracellular vesicles (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Together, these findings suggest that H19-driven pathogenic signaling is mediated through local, bile duct\u0026ndash;proximal intercellular communication, and that its removal selectively normalizes transcriptional programs in hepatocytes directly influenced by cholangiocyte-derived H19.\u003c/p\u003e \u003cp\u003eTaken together, these findings demonstrate that H19 deletion induces both cell-type\u0026ndash;specific and spatially resolved transcriptomic remodeling, restoring healthy gene-expression programs across multiple hepatic lineages and attenuating PSC-related pathology in both mouse and human datasets. While these results position H19 inhibition as a compelling therapeutic strategy for cholestatic liver injury, the newly uncovered regulatory interplay between H19 and the Xist/Tsix epigenetic axis underscores the need for careful evaluation of sex-dependent and broader epigenetic consequences prior to clinical translation.\u003c/p\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eShort comings\u003c/h2\u003e \u003cp\u003eAlthough we were able to validate disease-associated differential gene-expression patterns and predictive modeling outputs using an independent human dataset, we could not directly validate the H19-deletion\u0026ndash;specific findings due to the absence of comparable datasets. Additionally, the sequencing depth in both our mouse and publicly available human snRNA-seq datasets was insufficient to reliably detect H19, and H19 was not included in the GeoMx Whole Transcriptome Atlas (WTA) panel. These constraints limited our ability to perform direct cross-species validation of H19-dependent effects at either the single-cell or spatial resolution.\u003c/p\u003e \u003c/div\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAlexa Fluor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eaSMA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ealpha Smooth Muscle Actin\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eArea Under the Curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDEGs\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDifferentially Expressed Genes\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDGE\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDifferential Gene expression\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDKO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMdr2KOH19\u0026ndash;\u003csup\u003e/\u003c/sup\u003e\u0026ndash;Double Knockout\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDSP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDigital Spatial Profiler\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHCC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eHepatocellular Carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eH19\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003elong non\u0026ndash;coding RNA H19\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eH19KO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eH19 Knockout\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLASSO\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eLeast Absolute Shrinkage and Selection Operator\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMdr2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMultidrug resistance 2\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePBC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePrimary Biliary Cholangitis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePSC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePrimary Sclerosing Cholangitis\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRNA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRibonucleic Acid\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eROC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eReceiver Operating Characteristic\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003escRNAseq\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSingle Cell RNA Sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003esnRNAseq\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eSingle Nuclear RNA sequencing\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eWT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eWild Type\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eWTA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eWhole Transcriptome Atlas\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":" \u003cp\u003e \u003cstrong\u003eConflict of Interest Statement:\u003c/strong\u003e \u003cp\u003eThe authors have no conflict of interests to disclose.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFinancial Support\u003c/strong\u003e \u003cp\u003eThis study was supported by VA Merit Award 5I01BX005730, VA ShEEP grants (1 IS1 BX004777-01 and 1IS1BX005517-01), National Institutes of Health Grant R01 DK115377, 2R56DK115377-05A1, 5R01AA030180, R01DK139587, NIH-NCI P01CA275740, 5T32AA029975. Dr. Zhou is the recipient of a Research Career Scientist Award from the Department of Veterans Affairs (IK6BX004477).\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eClinical Trial Number\u003c/h2\u003e \u003cp\u003eN/A\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eDeclarations\u003c/h2\u003e \u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003e All animal experiments were performed following institutional guidelines for ethical animal studies and approved by the Virginia Commonwealth University and Richmond VA Medical Center Institutional Animal Care and Use Committee, Virginia, USA.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003eAll authors reviewed and approved the final manuscript. All authors supported the publication of this manuscript.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare no competing financial interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis study was supported by VA Merit Award 5I01BX005730, VA ShEEP grants (1 IS1 BX004777-01 and 1IS1BX005517-01), National Institutes of Health Grant R01 DK115377, 2R56DK115377-05A1, 5R01AA030180, and NIH-NCI 1P01CA275740-01 and Cancer Center Support Grant P30 CA 016059. Dr. Zhou is the recipient of a Research Career Scientist Award from the Department of Veterans Affairs (IK6BX004477).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization and design, GWW, HZ; Methodology, GWW, XJ, HZ; Data Acquisition, GWW, XJ, HL, NW, DZ, YT, SB, XW; Data Analysis, GWW, XJ; Writing: Original Draft, GWW, XJ; Writing: Review \u0026amp; Editing, GWW, XJ, HZ; Funding Acquisition, HZ; Resources, HZ; Supervision, HZ.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data and analysis methodology will be made available upon requests to the corresponding author. snRNAseq and spatial transcriptomic data will be uploaded to GEO.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAngulo P, Lindor KD. Primary sclerosing cholangitis. Hepatology. 1999;30:325\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHirschfield GM, Karlsen TH, Lindor KD, Adams DH. Primary sclerosing cholangitis. Lancet. 2013;382:1587\u0026ndash;99.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarlsen TH, Folseraas T, Thorburn D, Vesterhus M. Primary sclerosing cholangitis - a comprehensive review. J Hepatol. 2017;67:1298\u0026ndash;323.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEaton JE, Talwalkar JA, Lazaridis KN, Gores GJ, Lindor KD. Pathogenesis of primary sclerosing cholangitis and advances in diagnosis and management. Gastroenterology. 2013;145:521\u0026ndash;36.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTabibian JH, Ali AH, Lindor KD. Primary Sclerosing Cholangitis, Part 2: Cancer Risk, Prevention, and Surveillance. Gastroenterol Hepatol (N Y). 2018;14:427\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiang JYL, Ferrell JM. Bile Acids as Metabolic Regulators and Nutrient Sensors. Annu Rev Nutr. 2019;39:175\u0026ndash;200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiang JYL, Ferrell JM. Bile acid receptors FXR and TGR5 signaling in fatty liver diseases and therapy. Am J Physiol Gastrointest Liver Physiol. 2020;318:G554\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHylemon PB, Takabe K, Dozmorov M, Nagahashi M, Zhou H. Bile acids as global regulators of hepatic nutrient metabolism. Liver Res. 2017;1:10\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJackson KG, Way GW, Zhou H. Bile acids and sphingolipids in non-alcoholic fatty liver disease. Chin Med J 2022:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/CM1099.0000000000002156\u003c/span\u003e\u003cspan address=\"10.1097/CM1099.0000000000002156\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWay GW, Jackson KG, Muscu SR, Zhou H. Key Signaling in Alcohol-Associated Liver Disease: The Role of Bile Acids. Cells 2022;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTabibian JH, Ali AH, Lindor KD. Primary Sclerosing Cholangitis, Part 1: Epidemiology, Etiopathogenesis, Clinical Features, and Treatment. Gastroenterol Hepatol (N Y). 2018;14:293\u0026ndash;304.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, Hylemon PB, Zhou H. Long Noncoding RNA H19: A Key Player in Liver Diseases. Hepatology. 2021;74:1652\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, Zeng J, Chen W, Fan J, Hylemon PB, Zhou H. Long Noncoding RNA H19: A Novel Oncogene in Liver Cancer. Noncoding RNA 2023;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Liu R, Huang Z, Gurley EC, Wang X, Wang J, He H, et al. Cholangiocyte-derived exosomal long noncoding RNA H19 promotes cholestatic liver injury in mouse and humans. Hepatology. 2018;68:599\u0026ndash;615.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Liu R, Wang Y, Zhu W, Zhao D, Wang X, Yang H et al. Cholangiocyte-Derived Exosomal lncRNA H19 Promotes Macrophage Activation and Hepatic Inflammation under Cholestatic Conditions. Cells 2020;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Liu R, Yang J, Sun L, Zhang L, Jiang Z, Puri P, et al. The role of long noncoding RNA H19 in gender disparity of cholestatic liver injury in multidrug resistance 2 gene knockout mice. Hepatology. 2017;66:869\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu R, Li X, Zhu W, Wang Y, Zhao D, Wang X, Gurley EC, et al. Cholangiocyte-Derived Exosomal Long Noncoding RNA H19 Promotes Hepatic Stellate Cell Activation and Cholestatic Liver Fibrosis. Hepatology. 2019;70:1317\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, Tai YL, Way G, Zeng J, Zhao D, Su L, Jiang X, et al. RNA binding protein HuR protects against NAFLD by suppressing long noncoding RNA H19 expression. Cell Biosci. 2022;12:172.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao Y, Liu R, Li X, Gurley EC, Hylemon PB, Lu Y, Zhou H, et al. Long Noncoding RNA H19 Contributes to Cholangiocyte Proliferation and Cholestatic Liver Fibrosis in Biliary Atresia. Hepatology. 2019;70:1658\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu H, Kirita Y, Donnelly EL, Humphreys BD. Advantages of Single-Nucleus over Single-Cell RNA Sequencing of Adult Kidney: Rare Cell Types and Novel Cell States Revealed in Fibrosis. J Am Soc Nephrol. 2019;30:23\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoung MD, Behjati S. SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 2020;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLun ATL, Riesenfeld S, Andrews T, Dao TP, Gomes T, participants in the 1st Human Cell Atlas, Marioni J. JC. EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data. Genome Biol 2019;20:63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. 2024;42:293\u0026ndash;304.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeam} RC. R: A Language and Environment for Statistical Computing. In: R Foundation for Statistical Computing; 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKorsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. 2019;16:1289\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, et al. Inference and analysis of cell-cell communication using CellChat. Nat Commun. 2021;12:1088.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCao J, Spielmann M, Qiu X, Huang X, Ibrahim DM, Hill AJ, Zhang F, et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature. 2019;566:496\u0026ndash;502.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33:1\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTay JK, Narasimhan B, Hastie T. Elastic Net Regularization Paths for All Generalized Linear Models. J Stat Softw 2023;106.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHua Y, Weng L, Zhao F, Rambow F. SeuratExtend: streamlining single-cell RNA-seq analysis through an integrated and intuitive framework. Gigascience 2025;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKolberg L, Raudvere U, Kuzmin I, Vilo J, Peterson H. gprofiler2 -- an R package for gene list functional enrichment analysis and namespace conversion toolset g:Profiler. F1000Res 2020;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndrews TS, Nakib D, Perciani CT, Ma XZ, Liu L, Winter E, Camat D, et al. Single-cell, single-nucleus, and spatial transcriptomics characterization of the immunological landscape in the healthy and PSC human liver. J Hepatol. 2024;80:730\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRobin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, Muller M. pROC: an open-source package for R and S\u0026thinsp;+\u0026thinsp;to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu N, Bhuva DD, Mohamed A, Bokelund M, Kulasinghe A, Tan CW, Davis MJ. standR: spatial transcriptomic analysis for GeoMx DSP data. Nucleic Acids Res. 2024;52:e2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcInnes L, Healy J, UMAP. Uniform Manifold Approximation and Projection for Dimension Reduction. ArXiv 2018;abs/1802.03426.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLove MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenberger FA, Thielert M, Strauss MT, Schweizer L, Ammar C, Madler SC, Metousis A, et al. Spatial single-cell mass spectrometry defines zonation of the hepatocyte proteome. Nat Methods. 2023;20:1530\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCunningham RP, Porat-Shliom N. Liver Zonation - Revisiting Old Questions With New Technologies. Front Physiol. 2021;12:732929.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang L, Lu P, Yang X, Li K, Qu S. Annexin A3, a Calcium-Dependent Phospholipid-Binding Protein: Implication in Cancer. Front Mol Biosci. 2021;8:716415.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhuang C, Wang P, Sun T, Zheng L, Ming L. Expression levels and prognostic values of annexins in liver cancer. Oncol Lett. 2019;18:6657\u0026ndash;69.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei SC, Tan YY, Weng MT, Lai LC, Hsiao JH, Chuang EY, Shun CT, et al. SLCO3A1, A novel crohn's disease-associated gene, regulates nf-kappaB activity and associates with intestinal perforation. PLoS ONE. 2014;9:e100515.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan Q, Zhang X, Zhang L, Cheng Y, Zhao N, Li F, Zhou X, et al. Solute Carrier Organic Anion Transporter Family Member 3A1 Is a Bile Acid Efflux Transporter in Cholestasis. Gastroenterology. 2018;155:1578\u0026ndash;92. e1516.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUdomsinprasert W, Poovorawan Y, Chongsrisawat V, Vejchapipat P, Honsawek S. Decreased circulating clusterin reflects severe liver complications after hepatoportoenterostomy of biliary atresia. Sci Rep. 2020;10:19736.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H, Lan L, Chen H, Zaw Thin M, Ps H, Nelson JK, Evans IM et al. SPP1 is required for maintaining mesenchymal cell fate in pancreatic cancer. Nature 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong Z, Chen W, Athavale D, Ge X, Desert R, Das S, Han H, et al. Osteopontin Takes Center Stage in Chronic Liver Disease. Hepatology. 2021;73:1594\u0026ndash;608.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoch KS, Leffert HL. Ectopic expression of CD74 in Ikkbeta-deleted mouse hepatocytes. Acta Histochem. 2011;113:428\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang W, Min L, Qiu X, Wu X, Liu C, Ma J, Zhang D, et al. Biological Function of Long Non-coding RNA (LncRNA) Xist. Front Cell Dev Biol. 2021;9:645647.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu B, Li F, Wang Y, Gao X, Li Y, Wang Y, Zhou H. APP-CD74 axis mediates endothelial cell-macrophage communication to promote kidney injury and fibrosis. Front Pharmacol. 2024;15:1437113.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXing X, Cao F, Gao L, Song M. AHSG, a Gene Promoting Tumour Proliferation, Migration and Invasion, is an Independent Prognostic Factor for Poor Overall Survival in Lung Adenocarcinoma. Mol Biol Rep. 2023;50:7659\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Y, Yang B, Chen D, Zhou X, Wang M, Jiang J, Wei L, et al. Combined identification of ARID1A, CSMD1, and SENP3 as effective prognostic biomarkers for hepatocellular carcinoma. Aging. 2021;13:4696\u0026ndash;712.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEscudero-Esparza A, Bartoschek M, Gialeli C, Okroj M, Owen S, Jirstrom K, Orimo A, et al. Correction: Complement inhibitor CSMD1 acts as tumor suppressor in human breast cancer. Oncotarget. 2023;14:481\u0026ndash;2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEscudero-Esparza A, Bartoschek M, Gialeli C, Okroj M, Owen S, Jirstrom K, Orimo A, et al. Complement inhibitor CSMD1 acts as tumor suppressor in human breast cancer. Oncotarget. 2016;7:76920\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"cell-and-bioscience","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cbio","sideBox":"Learn more about [Cell \u0026 Bioscience](http://cellandbioscience.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/cbio/default.aspx","title":"Cell \u0026 Bioscience","twitterHandle":"@OACellBiology","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Primary sclerosing cholangitis, Cholestasis, SPP1, Machine learning, H19, Long non-coding RNA","lastPublishedDoi":"10.21203/rs.3.rs-8339668/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8339668/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003ePrimary Sclerosing Cholangitis (PSC) is a chronic obstructive biliary disease and remains a high-burden cholestatic liver disease with no approved therapies and a substantial recurrence rate following liver transplantation. The long non-coding RNA H19 (H19) has emerged as a potential driver of PSC progression, yet its cell-type-specific and spatially resolved mechanisms remain poorly defined.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eAge- and sex-matched wild type (WT), H19 knockout (H19KO), Mdr2 knockout (Mdr2KO), and double-knockout (DKO; Mdr2KO/H19KO) mice were used. The liver tissues were analyzed using single nucleus RNA sequencing (snRNAseq) and NanoString GeoMx spatial transcriptomics to elucidate H19-dependent cellular and spatial alternations in cholestatic liver injury. Machine learning models (logistic regression, XGBoost, neural network, and random forest) were developed to generate cell-type specific disease prediction signatures and validated using the publicly available human dataset GSE243981. Both spatial transcriptomics and snRNAseq identified a disease-associated cholangiocyte subcluster that was significantly expanded in Mdr2KO mice, but markedly diminished in DKO mice, demonstrating a requirement for H19 in sustaining pathogenic cholangiocyte state. SPP1 signaling was significantly dysregulated in cholestatic liver injury and ameliorated with H19 deletion. Novel murine markers were identified, including Gm13775 (healthy hepatocytes) and Clu and Spp1 (healthy cholangiocytes), all of which were markedly downregulaed in disease. Machine learning-based, cell type-specific disease prediction models achieved AUC values\u0026thinsp;\u0026gt;\u0026thinsp;0.87 when validated in the GSE243981 human dataset. Noteably,Spp1 expression decreased in cholangiocytes but was ectopically upregulated in hepatocytes in diseased liver, highlighting disrupted intercellular signaling network. Spatial analyses showed that \u003cem\u003eH19\u003c/em\u003e deletion restored the disease-associated gene expression changes specifically within the bile duct region.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eH19 deletion mitigates cholestatic injury by suppressing pathogenic cholangiocyte states, normalizing SPP1-mediated signaling, and restoring bile-duct-localized transcriptional programs. These findings position H19 as a critical regulator of cholangiocyte-driven pathology and a potential therapeutic target in PSC.\u003c/p\u003e","manuscriptTitle":"Decoding the Role of H19 in Cholestatic Liver Injury Using snRNA-seq, Spatial Transcriptomics, and Machine Learning-Based Disease Prediction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-14 11:08:22","doi":"10.21203/rs.3.rs-8339668/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-01-30T15:27:29+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-29T21:52:45+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-25T18:59:05+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"52028999012084550403438261020086151407","date":"2026-01-14T14:12:12+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"155134068486191548516187662150018327389","date":"2026-01-12T14:15:13+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-12T13:40:08+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-12-14T03:20:35+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-13T16:59:32+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cell \u0026 Bioscience","date":"2025-12-11T19:10:06+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cell-and-bioscience","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"cbio","sideBox":"Learn more about [Cell \u0026 Bioscience](http://cellandbioscience.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/cbio/default.aspx","title":"Cell \u0026 Bioscience","twitterHandle":"@OACellBiology","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6e4a7b52-8d4c-4246-8538-b910ebcdab3b","owner":[],"postedDate":"January 14th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-07T00:24:37+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-14 11:08:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8339668","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8339668","identity":"rs-8339668","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}