Evaluation of a 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet-induced mouse model in a comparative experimental study of portal hypertension

Abstract

BACKGROUND

Portal hypertension (PHT) is a life-threatening complication of chronic liver disease, necessitating reliable animal models that mimic its clinical heterogeneity. Classical mouse models like bile duct ligation (BDL) exhibit a low 4-week survival (35%), while carbon tetrachloride (CCl₄) models have delayed pathogenesis, requiring ≥ 8 weeks for PHT development, limiting their efficiency.

AIM

To evaluate the 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet-induced mouse model as a biliary PHT model, comparing it to BDL and CCl₄ models.

METHODS

Mice were assigned to DDC diet, BDL, or CCl₄ groups. Assessments included portal pressure, histological examination of biliary fibrosis and hepatic stellate cell (HSC) activation (desmin expression), scanning electron microscopy for sinusoidal fenestrae and capillarization, endothelial nitric oxide synthase (eNOS) regulation (phosphorylated-eNOS/total eNOS ratio and total eNOS level), ductular reaction, inflammatory infiltration, and portosystemic shunting. Survival rates and operational feasibility (feed-based administration) were evaluated.

RESULTS

At 4 weeks, DDC induced portal pressure comparable to BDL and CCl₄. The DDC model showed moderate biliary fibrosis (similar to BDL but less than CCl₄) and greater HSC activation than the other two models. Sinusoidal fenestrae reduction and capillarization in DDC matched BDL and CCl₄ models. DDC had decreased phosphorylated-eNOS/total eNOS ratio, while BDL and CCl₄ models exhibited reduced total eNOS. DDC demonstrated robust ductular response, inflammation, and shunting, a hallmark of PHT. Survival was 100% (vs 35% BDL, 58.3%-66.6% CCl₄), with simpler feed-based induction.

CONCLUSION

The DDC model offers strong biliary PHT relevance, high survival, and efficiency, making it a superior alternative to BDL and CCl₄ models for biliary PHT research.

Key Words: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; Portal hypertension; Mouse model; Portal pressure; Ductular reaction; Chronic liver disease

Core Tip: This study validates the 0.1% 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet mouse model for biliary-related portal hypertension (PHT). It matches bile duct ligation (BDL) and carbon tetrachloride (CCl₄) models in PHT pathologies (e.g., elevated portal pressure, sinusoidal changes) but outperforms them with 100% 4-week survival (vs 35% BDL, 58.3%-66.6% CCl₄), simpler feed-based use, and better recapitulation of PHT hallmarks, serving as a superior alternative.

INTRODUCTION

Portal hypertension (PHT) is a common and severe complication of chronic liver disease (CLD), particularly cirrhosis, resulting in life-threatening outcomes such as gastroesophageal variceal bleeding, ascites, hepatorenal syndrome, and hepatic encephalopathy[1]. This pathology is central to cirrhosis-related morbidity, contributing to 1 million of the 2 million annual global liver disease-related deaths[2]. As the primary driver of hepatic decompensation in cirrhosis, PHT significantly worsens mortality rates; for instance, cirrhosis accounted for 2.4% of worldwide deaths in 2019, with an age-standardized prevalence of 1.4%[3]. This substantial burden underscores the gaps in our understanding of disease mechanisms and the scarcity of effective treatments, underscoring the critical need for deeper mechanistic insights and novel therapeutic strategies.

Animal models serve as primary tools for investigating disease pathogenesis and conducting drug research[4-6]. These models can simulate disease onset, progression, and drug responses to a significant extent. Classic PHT models, such as rat bile duct ligation (BDL), carbon tetrachloride (CCl₄ or thioacetamide), and partial portal vein ligation (PPVL) models, each have distinct characteristics[7]. Researchers studying treatments for related diseases or PHT-related complications often choose the BDL model[8,9] (characterized by early-stage high portal pressure and biliary fibrosis), the PPVL model[10] (featuring early visible portal vein shunting), and the CCl₄ model[11,12] (known for its simple modeling process). It is widely acknowledged that the larger body size and robust vascular system of the rat model have significantly facilitated PHT research for decades[8,13-15]. A recent study on silencing the 5-hydroxytryptamine receptor 1a gene in rats confirmed that in vivo gene intervention effectively alleviates PHT[16]. However, due to immature gene editing technology in rats, mice with genomes more homologous to humans and advanced gene editing tools have become the preferred model for PHT research[17]. Increasingly, researchers are adapting rat PHT model construction methods to explore PHT pathogenesis and treatment in mice, with or without gene interventions. Examples include studies showing how mechanical stretch increases chemokine (C-X-C motif) ligand 1 (Cxcl1) expression in liver sinusoidal endothelial cells (LSEC) to promote PHT[18] and how endothelial p300 promotes PHT and hepatic fibrosis through C-C motif chemokine ligand 2-mediated angiocrine signaling[19].

Etiologies of cirrhosis include viral infections (hepatitis B, C, and D), alcohol-related liver disease, metabolic and genetic disorders, autoimmune diseases, biliary diseases, vascular disorders, drug-induced cases, and cryptogenic cases[20]. Given the diverse pathogeneses of cirrhosis, BDL and CCl₄ mouse models fail to capture the full range of PHT mechanisms across etiologies. There is an urgent need to develop mouse models simulating multiple liver diseases that induce PHT, better reflecting clinical heterogeneity. For instance, Mdr2 knockout mice exhibited early portal pressure elevation as early as 2015[21]. Recently, we found that a 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet mouse model of primary biliary cholangitis (PBC) showed portal pressure comparable to BDL- and CCl₄-induced PHT models at 4 weeks. As previously reported, DDC-fed mice serve as a toxicant-induced model of sclerosing cholangitis and biliary fibrosis[22]. Chronic DDC feeding for 4 weeks induces biliary porphyrin secretion, resulting in intraductal plug formation, ductular proliferation, onion-skin periductal fibrosis, and subsequent biliary fibrosis[23]. This approach is simpler than the CCl₄ model and addresses liver diseases leading to PHT such as PBC not covered by BDL or CCl₄ models. Therefore, we aim to investigate the mechanisms of early-onset PHT in the DDC diet model and evaluate its potential as a primary pathogenic model for PHT research by assessing fibrosis, abnormal liver sinusoidal endothelial function, and inflammation. These analyses will support future studies on pathogenic mechanisms.

MATERIALS AND METHODS

Animal models

All animal experiments involving mice were reviewed and approved by the Ethics Committee of the Ninth People’s Hospital Affiliated to Shanghai Jiao Tong University (approval No. SH9H-2021-A233-SB).

DDC diet mouse model

C57BL/6 male mice (8-10 weeks old) were fed a normal chow diet (NCD) supplemented with 0.1% DDC to induce ductular reaction (DR) and biliary fibrosis[24]. For the 2-8 week feeding experiment and subsequent pathological characterization, five mice per group were used to measure portal pressure and body weight at 2, 4, 6, and 8 weeks. In follow-up comparative experiments, four mice per group were sacrificed at 4 weeks for sample collection.

BDL mouse model

C57BL/6 male mice (8-10 weeks old) were anesthetized with isoflurane and subjected to a midline laparotomy (approximately 1 cm) to expose the common bile duct. Two knots were tied using 5-0 silk sutures without transecting the duct. The peritoneum was reapproximated, and the muscle layer and skin were separately sutured with 3-0 silk sutures. Sham-operated mice underwent laparotomy and bile duct dissection only. This procedure induces biliary obstruction, leading to cholestasis and PHT for comparative studies. In previous studies, we used 8 sham and 20 BDL mice; in comparative studies with the DDC diet model, we initially used 8 sham and 25 BDL mice (sample sizes based on expected survival rates and power analysis). Survival improvement measures for BDL mice included intraperitoneal vitamin K injections (50 μg/kg, once a week), switching to 4-0 silk sutures for ligation, and postoperative feeding with digestible jelly rich in sugar and water. At 4 weeks post-surgery, all surviving mice had portal pressure and body weight measured, with liver and mesentery samples collected for analysis.

CCl₄ mouse model

C57BL/6 male mice (8-10 weeks old) were housed in a pathogen-free facility at 22 °C under a 12-hour light/dark cycle. Liver fibrosis was induced by intraperitoneal injection of CCl₄ (1 μL/g body weight; Sigma-Aldrich, catalog No. 319961) diluted in oil, administered twice weekly for 8 or 12 weeks. Control mice received an equivalent volume of oil. At the end of the 8-week treatment period, control (oil) and 8-week CCl₄-injected mice were euthanized after measuring portal pressure and body weight, with tissues harvested for analysis. The same protocol was applied to 12-week CCl₄-injected mice at study termination. In prior cohorts, the control (oil) group consisted of 8 mice, whereas the 8-week and 12-week CCl₄ groups each included 12 mice (sample sizes justified by prior power calculations). This configuration was maintained in comparative studies with the DDC diet model.

Measurement of portal pressure

In accordance with our previous study[8], mice were anesthetized by inhalation of isoflurane (1.5%-2%) and monitored for respiratory rate to ensure adequate anesthesia. A 23-gauge needle was carefully inserted into the portal vein. The catheter attached to the needle was connected to a pressure transducer to detect portal pressure using an ALC-MPA multi-channel biological information analysis system (Shanghai Alcott Biotech Co., Ltd.).

Biochemical analysis

Mice were sacrificed after hemodynamic measurements. Plasma was obtained from mouse blood samples by centrifugation at 1500 × g for 15 minutes at 4 °C. The serum levels of alanine aminotransferase (ALT) (catalog No. 105-020579-00), aspartate aminotransferase (catalog No. 105-020580-00), total bilirubin (T-Bil) (catalog No. 105-020584-00), direct bilirubin (D-Bil) (catalog No. 105-00021A-00), and total bile acid (TBA) (catalog No. 105-020596-00) were measured using commercial kits (Mindray, Shenzhen, China) following the manufacturer’s instructions. All assays were performed in triplicate.

Scanning electron microscopy

Mice were euthanized, and livers were harvested, perfused with phosphate-buffered saline (potential of hydrogen = 7.4) at 37 °C to maintain tissue integrity, and minced into 1 mm³ cubes. Samples were incubated in 1.5% glutaraldehyde at 4 °C overnight, washed three times with 0.05 M sodium cacodylate buffer, post-fixed with 1% osmium tetroxide for 1 hour at 4 °C, and washed three times with distilled water. Tissues were dehydrated via an ethanol series, critical point dried, mounted on stubs with copper tape, sputter-coated with 15 nm platinum, and observed under a field-emission scanning electron microscope (AURIGA^®, Carl Zeiss) to assess sinusoidal fenestrae and capillarization.

Immunohistochemistry

Liver pathological morphology and portosystemic shunting were evaluated using hematoxylin and eosin (HE), Masson, and Sirius red staining protocols. For immunohistochemistry, sections underwent antigen retrieval with citrate buffer, followed by blocking of endogenous peroxidase activity with 0.3% hydrogen peroxide in methanol and preincubation with 10% normal horse serum to prevent nonspecific binding. Primary antibodies against collagen 1 (1:2500, catalog No. 67288-1-Ig, Proteintech), α-smooth muscle actin (SMA) (1:200, catalog No. ab5694, Abcam), desmin (1:4000, catalog No. 16520-1-AP, Proteintech), lymphatic vessel endothelial hyaluronan receptor 1 (LyVE-1) (1:100, catalog No. ab219556, Abcam), cluster of differentiation (CD) 34 (1:1000, catalog No. 14486-1-AP, Proteintech), von Willebrand factor (vWF) (1:200, catalog No. 27186-1-AP, Proteintech), vascular endothelial growth factor receptor 2 (VEGFR2) (1:150, catalog No. ab2349, Abcam), vascular endothelial growth factor A (VEGF-A) (1:100, catalog No. ab52917, Abcam), and CD31 (1:5000, catalog No. 11265-1-AP, Proteintech) were applied overnight at 4 °C, followed by 60-minute incubation with secondary antibodies at room temperature. Sections were counterstained with hematoxylin, dehydrated through ethanol gradients, cleared with xylene, and mounted for visualization under a fluorescence microscope to analyze protein expression and localization. Staining was quantified using ImageJ software.

Western blotting analysis

Liver and mesenteric artery tissues from mice were immediately stored at -80 °C until use. For protein extraction, samples were homogenized in radio immunoprecipitation assay buffer (Beyotime) according to the manufacturer’s protocol. Homogenates were centrifuged at 12000 × g for 15 minutes at 4 °C, and supernatants were collected to determine total protein concentration using a bicinchoninic acid assay kit (Beyotime). Equal protein amounts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electro-transferred to polyvinylidene difluoride membranes. Membranes were probed with primary antibodies against endothelial nitric oxide synthase (eNOS) (1:1000, Cell Signaling), phosphorylated eNOS (p-eNOS) (1:1000, Affinity), and glyceraldehyde 3-phosphate dehydrogenase (1:5000, Proteintech). After washing, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies for 60 minutes at room temperature and visualized using enhanced chemiluminescence detection (Millipore). Bands were quantified using densitometry (ImageJ).

Preparation and reverse transcription quantitative polymerase chain reaction

Total RNA was isolated from liver tissues using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. RNA purity and concentration were assessed via absorbance ratios at 260/280 nm and 260/230 nm using a NanoDrop spectrophotometer. First-strand complementary DNA was synthesized from 1 μg of total RNA using the PrimeScript RT Reagent Kit (Takara) with genomic DNA Eraser to eliminate genomic DNA contamination. Reverse transcription quantitative polymerase chain reaction was performed on an ABI 7900HT fast real-time polymerase chain reaction system using SYBR green master mix (Yeasen). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the endogenous control. Reactions were run in triplicate with the following thermal cycling conditions: 95 °C for 30 seconds, followed by 40 cycles of 95 °C for 5 seconds and 60 °C for 30 seconds. Melt curve analysis verified amplicon specificity. Relative gene expression levels were calculated using the 2^-ΔΔCt method, normalized to GAPDH, and calibrated to control samples. The primer sequences are listed in Table 1. Primer efficiency was validated (90%-110%).

Table 1 Primer sequences.

Gene name	Sequence (5’-3’) forward	Sequence (5’-3’) reverse
TNF-α (mouse)	CCCTCACACTCAGATCATCTTCT	GCTACGACGTGGGCTACAG
IL-1β (mouse)	GCAACTGTTCCTGAACTCAACT	ATCTTTTGGGGTCCGTCAACT
IL-6 (mouse)	TAGTCCTTCCTACCCCAATTTCC	TTGGTCCTTAGCCACTCCTTC
GAPDH (mouse)	AGGTCGGTGTGAACGGATTTG	TGTAGACCATGTAGTTGAGGTCA

IL: Interleukin; TNF: Tumor necrosis factor; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

Statistical analysis

All data are presented as mean ± SD. Experiments were conducted in a randomized (software-generated random number randomization) and blinded (code-based blinding) manner to minimize subjective bias. Normality was assessed using the Shapiro-Wilk test. Statistical differences between two groups were evaluated using an unpaired Student’s t-test, while differences among three or more groups were assessed with one-way analysis of variance. Survival curve analysis incorporated the Gehan-Breslow-Wilcoxon test and Mantel-Haenszel risk ratio estimation. Statistical significance was set at P < 0.05. Data analysis and visualization were performed using SPSS^® (Version 23.0, IBM) and GraphPad Prism^® (Version 7.0, GraphPad Software).

RESULTS

DDC diet mouse model (4 weeks) showed portal pressure comparable to BDL and CCl₄ mouse models

Building on previous literature[19,25,26], we successfully established PHT mouse models using BDL and CCl₄ methods. The timelines and procedures are shown in Figure 1A-C. CCl₄ protocols vary widely across institutions, so our group standardized two durations (8 or 12 weeks) to explore PHT progression. BDL mice reached approximately 10 mmHg portal pressure by 4 weeks, far exceeding sham controls (Figure 1D), but survival dropped to just 35% post-surgery (Figure 1E). Even with modifications to the surgical approach and additional postoperative care methods, the survival rate did not improve significantly (Supplementary Figure 1). In the CCl₄ mouse model, toxic effects elevated portal pressure to approximately 10 mmHg at 8 weeks, with minimal further rise at 12 weeks (Figure 1F). Survival rates hovered at 58.3%-66.6% for both groups (Figure 1G). To comprehensively characterize phenotypic changes in DDC diet mouse models, we collected data from mice fed DDC for 2, 4, 6, and 8 weeks. At 2 weeks, no differences emerged in body weight or portal pressure vs NCD controls. By 4 weeks, DDC mice showed marked weight loss compared to the NCD group, with portal pressure reaching approximately 10 mmHg (Figure 1H and I). Extending feeding to 6-8 weeks brought no additional changes, with only one death noted (Figure 1J), underscoring the model’s safety.

Figure 1 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet mouse model (4 weeks) exhibited portal pressure comparable to bile duct ligation and carbon tetrachloride mouse models. A-C: Schematics of the modeling procedures and timelines for 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet (A), bile duct ligation (BDL) (B), and carbon tetrachloride (CCl₄) mouse models (C), respectively; D and E: Portal pressure measurements and survival curve of sham (n = 8) and BDL (n = 7) groups; F and G: Portal pressure and survival curve in oil (n = 8), 8-week CCl₄ (n = 7), and 12-week CCl₄ (n = 8) groups; H-J: Longitudinal comparisons of body weight, portal pressure, and survival curve between normal chow diet and DDC groups at 2 (n = 5), 4 (n = 5), 6 (n = 5), and 8 (n = 4) weeks post-modeling. ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; HR: Hazard ratio; CI: Confidence interval; NS: No significant.

DDC diet mouse model (4 weeks) exhibited moderate biliary fibrosis in the portal vein region

To investigate mechanisms underlying portal pressure elevation in the DDC diet mouse in comparison with other PHT animal models, we examined portal vein fibrosis via HE, Masson, and Sirius red stains. In HE staining, liver portal tracts of DDC diet mice exhibited intraductal plug formation, DR, onion-skin periductal fibrosis, and secondary biliary fibrosis. In contrast, BDL-induced bile duct hyperplasia around the portal veins was less extensive but more densely cellular. Additionally, the CCl₄ mouse model showed minimal bile duct proliferation, with predominant hepatocellular ballooning changes. The DDC diet mouse model and the BDL mouse model exhibited comparable biliary fibrosis in the periductal region. Periductal fibrosis aligned closely between DDC and BDL, contrasting CCl₄’s broader fibrous deposition in the portal tracts and interlobular spaces (Figure 2A). The DDC diet mouse model and the BDL mouse model exhibited comparable biliary fibrosis in the periductal region. In contrast, the CCl₄-induced model predominantly showed fibrous deposition in the portal tracts and interlobular spaces (Figure 2A). Masson quantification showed comparable fibrosis across models (Figure 2B-D), although Sirius red staining revealed that DDC-induced fibrosis was moderately less severe than in the CCl₄ and BDL mouse models (Figure 2E-G).

Figure 2 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet mouse model (4 weeks) exhibited moderate biliary fibrosis in the portal vein region. A: Hematoxylin-eosin, Masson, and Sirius red staining of mouse livers; B-G: Statistical analysis of Masson and Sirius red staining results; H: Immunohistochemistry staining for collagen-1, α-smooth muscle actin (SMA), and desmin in mouse livers; I-Q: Statistical analysis of collagen-1, α-SMA, and desmin staining results. Sample sizes (n): 8:8 (normal chow diet vs 3,5-diethoxycarbonyl-1,4-dihydrocollidine), 8:7 (sham vs bile duct ligation), and 8:7:8 (oil vs 8-week carbon tetrachloride vs 12-week carbon tetrachloride). ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. NS: No significant; NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; HE: Hematoxylin-eosin; COL1: Collagen-1; SMA: Smooth muscle actin.

The expression levels of hepatic stellate cell (HSC) activation markers including collagen 1, α-SMA, and desmin key proteins implicated in liver fibrogenesis, were also assessed (Figure 2H). Quantification of collagen 1 protein levels showed that both CCl₄ and BDL mouse models exhibit significantly increased collagen fiber deposition (Figure 2I-K). However, α-SMA and desmin staining indicated that HSC activation in the DDC diet mouse model was significantly higher than in the other two models (Figure 2L-Q). These results suggest that the DDC diet induces biliary fibrosis comparable to the other two models, contributing to portal pressure elevation and PHT-associated complications[27].

DDC diet mouse model (4 weeks) exhibited significant LSEC dysfunction

Liver vascular integrity depends on LSEC and HSC homeostasis. Shear stress is critical for LSEC function, promoting LSECs to vasodilators (e.g., nitric oxide) for sinusoidal pressure control[28]. In PHT, LSEC dysfunction involves loss of fenestrations, dysregulated paracrine factor secretion[29], impaired sinusoidal cell crosstalk, activated mechanosensitive pathways, and altered membrane proteomics, which increase intrahepatic vascular resistance (IHVR) and simulate downstream effects[1]. First, we assessed liver sinusoidal endothelial fenestration disruption using scanning electron microscopy (SEM) (Figure 3), which revealed sharp fenestrae losses in DDC, BDL, and CCl₄ models (Figure 3A, C, and E), with 12-week CCl₄ showing the deepest declines (Figure 3B, D, and F). There were subtle variations in eNOS protein alterations across these models. DDC preserved total eNOS but decreased the p-eNOS/total eNOS ratio (Figure 3G and H), while BDL and CCl₄ reduced total eNOS (Figure 3I-M). Collectively, these results suggest dysfunctional abnormalities in LSECs. Capillarization markers LyVE-1, CD34, and vWF serve as classic markers for assessing liver sinusoidal capillarization[30], with staining shown in Figure 3N. LyVE-1 staining indicated that the DDC diet or BDL significantly suppressed LyVE-1 protein expression in hepatic sinusoids, leading to sinusoidal dysfunction. In contrast, 8-week CCl₄ injection failed to stably inhibit LyVE-1 protein, whereas 12-week injection potently induced liver sinusoidal capillarization (Figure 3O, R, and U). CD34 yielded consistent results: 8-week CCl₄ injection failed to stably impair liver sinusoidal function, whereas 12-week injection recapitulated dysfunction comparable to DDC diet and BDL mouse models (Figure 3P, S, and V). vWF staining revealed subtle discrepancies: 8-week CCl₄ injection exacerbated LSEC dysfunction, comparable to DDC diet and BDL mouse models, with more profound impairment at 12 weeks (Figure 3Q, T, and W). Despite marker inconsistencies in these functional markers, all models disrupt LSEC homeostasis and contribute to PHT pathogenesis.

Figure 3 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet mouse model (4 weeks) exhibited significant liver sinusoidal endothelial cell dysfunction. A: Scanning electron microscopy (SEM) of liver sinusoids in normal chow diet (NCD) and 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet mouse models; B: Statistical analysis of fenestrae number and porosity; C: SEM of liver sinusoids in sham and bile duct ligation (BDL) mouse models; D: Statistical analysis of fenestrae number and porosity; E: SEM of liver sinusoids in oil, 8-week carbon tetrachloride (CCl₄), and 12-week CCl₄ mouse models; F: Statistical analysis of fenestrae number and porosity; G and H: Protein expression and quantitative analysis of total endothelial nitric oxide synthase (eNOS) and phosphorylated eNOS (p-eNOS) in NCD and DDC diet mouse livers; I and J: Protein expression and quantitative analysis of total eNOS and p-eNOS in sham and BDL mouse livers; K-M: Protein expression and quantitative analysis of total eNOS and p-eNOS in oil, 8-week CCl₄, and 12-week CCl₄ mouse livers; N: Immunohistochemistry staining for lymphatic vessel endothelial hyaluronan receptor 1 (LyVE-1), cluster of differentiation (CD) 34, and von Willebrand factor (vWF) in livers of the three model groups; O-W: Quantitative analysis of positive staining areas for LyVE-1, CD34, and vWF. Sample sizes (n): 8:8 (normal chow diet vs 3,5-diethoxycarbonyl-1,4-dihydrocollidine), 8:7 (sham vs bile duct ligation), and 8:7:8 (oil vs 8-week carbon tetrachloride vs 12-week carbon tetrachloride). ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. NS: No significant; NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; eNOS: Endothelial nitric oxide synthase; p-eNOS: Phosphorylated endothelial nitric oxide synthase; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; LyVE-1: Lymphatic vessel endothelial hyaluronan receptor 1; vWF: Von Willebrand factor; CD: Cluster of differentiation.

Dysfunction DDC diet induces abnormal proliferation of intrahepatic vascular structures and extrahepatic portosystemic shunting

Contrary to expectations, augmented intrahepatic angiogenesis may represent a compensatory repair mechanism, yet it exacerbates liver pathology without alleviating PHT[31]. This paradox arises because newly formed vasculature diverges from the sinusoidal vascular system, failing to enhance blood perfusion in cirrhotic livers. Instead, these abnormal vessels disrupt hepatocellular homeostasis, deteriorate liver injury, and potentiate fibrogenesis and inflammation[32]. Vascular markers VEGFR2, CD31, and VEGF-A revealed that the DDC diet, BDL, and CCl₄ injection all significantly promoted intrahepatic vascular proliferation in the portal vein region at sampling (Figure 4A-J). CD31 staining quantified BDL as marginally less pronounced than in the other two models (Figure 4C, F, and I). Intriguingly, VEGFR2 and VEGF-A expression levels were elevated in the 12-week CCl₄ injection group compared to the oil group, validating the ligand-receptor interplay between VEGF-A and VEGFR2 (Figure 4H and J). Additionally, we observed extensive extrahepatic portosystemic shunting (Figure 4J-M), reflecting the PHT pathology and thus validating model establishment. These collateral vessels augment blood inflow into the liver, exacerbating portal pressure and precipitating other complications.

Figure 4 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet induces abnormal proliferation of intrahepatic vascular structures and extrahepatic portosystemic shunting. A: Immunohistochemistry (IHC) staining for vascular endothelial growth factor receptor 2, cluster of differentiation 31, and vascular endothelial growth factor A in mouse livers; B-J: Statistical analysis of positive staining areas from IHC results; K-M: Hematoxylin-eosin staining of mesenteric tissues. Sample sizes (n): 8:8 (normal chow diet vs 3,5-diethoxycarbonyl-1,4-dihydrocollidine), 8:7 (sham vs bile duct ligation), and 8:7:8 (oil vs 8-week carbon tetrachloride vs 12-week carbon tetrachloride). ^bP < 0.01. ^cP < 0.001. NS: No significant; NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; CD: Cluster of differentiation; VEGFR2: Vascular endothelial growth factor receptor 2; VEGF-A: Vascular endothelial growth factor A.

DDC-induced intrahepatic inflammatory responses contribute to elevated portal pressure

Hepatocyte damage in CLD causes injury, triggering a pro-inflammatory state in parenchymal and non-parenchymal liver cell types, leading to liver fibrosis, cirrhosis, PHT, and hepatic failure[33]. Based on macrophage marker F4/80 detection in mice (Figure 5A), the DDC diet mouse model exhibited the most robust inflammatory cell infiltration (Figure 5B). To evaluate hepatic inflammation and injury, we assessed inflammatory cytokines and liver enzymes across the three models. In the CCl₄ mouse model, tumor necrosis factor-α elevation was marginally less pronounced than in the other two models, while interleukin-1β activation showed no significant inter-model differences (Figure 5C-G). Notably, interleukin-6 exhibited the most substantial increase in the BDL mouse model (Figure 5C, E, and G). In the DDC diet mouse model, ALT demonstrated the most pronounced elevation among liver enzymes (Figure 5D). In the BDL mouse model, serum T-Bil and D-Bil levels increased by nearly three orders of magnitude, potentially accounting for the higher mortality rate (Figure 5F). Under CCl₄ intoxication, hepatic tissues exhibited a nearly 100-fold increase in TBA levels (Figure 5H). Despite extensive inflammatory cell infiltration, upregulation of inflammatory cytokines and liver enzymes remained relatively modest in the DDC model, warranting further mechanistic exploration.

Figure 5 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine-induced intrahepatic inflammatory responses contribute to elevated portal pressure. A: Immunohistochemistry staining for F4/80 in mouse livers; B: Statistical analysis of F4/80-positive staining areas; C and D: Detection of hepatic inflammatory index messenger RNA and liver enzymes in normal chow diet and 3,5-diethoxycarbonyl-1,4-dihydrocollidine groups; E and F: Detection of hepatic inflammatory index messenger RNA and liver enzymes in sham and bile duct ligation groups; G and H: Detection of hepatic inflammatory index messenger RNA and liver enzymes in oil, 8-week carbon tetrachloride (CCl₄), and 12-week CCl₄ mouse models. Sample sizes (n): 8:8 (normal chow diet vs 3,5-diethoxycarbonyl-1,4-dihydrocollidine), 8:7 (sham vs bile duct ligation), and 8:7:8 (oil vs 8-week carbon tetrachloride vs 12-week carbon tetrachloride). ^bP < 0.01. ^cP < 0.001. NS: No significant; NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; IL: Interleukin; TNF: Tumor necrosis factor; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; T-Bil: Total bilirubin; D-Bil: Direct bilirubin; TBA: Total bile acid.

DDC diet induces DRs comparable to those in the BDL model

DR, a periportal compensatory response to various liver injuries, including biliary diseases, viral hepatitis, non-alcoholic fatty liver disease, and acute fulminant liver failure[34], appeared comparable in DDC and BDL portal areas. Cytokeratin 7 (CK7) analysis showed DDC surpassing the BDL mouse model. Moreover, CK7 protein analysis revealed that DR induced by the DDC diet was even higher than in the BDL model, a severe model (Figure 6A-G). Focusing on the CCl₄ model, the advantages of the DDC diet-induced biliary disease-related PHT model become more prominent. Although 12 weeks of CCl₄ injection can stably induce DR in the portal area, 8 weeks of CCl₄ failed to significantly trigger biliary epithelial hyperplasia as assessed by CK7 (Figure 6H-J). Studies have shown that DR is associated with PHT and liver disease prognosis[35], and the DDC diet mouse model outperforms BDL and CCl₄ models in inducing DR and biliary epithelial proliferative responses, particularly in short-term modeling. The DDC-induced model may therefore be used to study PHT pathogenesis primarily involving biliary disorders.

Figure 6 The 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet induces ductular reactions comparable to those in the bile duct ligation model. A: Immunohistochemistry staining for cytokeratin (CK) 7, CK19, and SOX9 in mouse livers; B-J: Statistical analysis of positive staining areas for CK7, CK19, and SOX9. Sample sizes (n): 8:8 (normal chow diet vs 3,5-diethoxycarbonyl-1,4-dihydrocollidine), 8:7 (sham vs bile duct ligation), and 8:7:8 (oil vs 8-week carbon tetrachloride vs 12-week carbon tetrachloride). ^bP < 0.01. ^cP < 0.001. NS: No significant; NCD: Normal chow diet; DDC: 3,5-diethoxycarbonyl-1,4-dihydrocollidine; BDL: Bile duct ligation; CCl₄: Carbon tetrachloride; CK: Cytokeratin.

DISCUSSION

Cholestasis is a pathological condition defined by the obstruction of bile flow from the liver to the duodenum, often due to reduced bile-forming capacity, defective secretion by hepatocytes or cholangiocytes, or mechanical blockage of bile ducts[36]. Among cholestatic liver diseases, PBC and primary sclerosing cholangitis (PSC) are major causes of CLD in adults, leading to biliary cirrhosis, end-stage liver disease, and the need for liver transplantation[37,38]. Due to limitations in studying human patients, rodent models replicating cholestatic histopathology have been developed. Common models include BDL and diet-based approaches (e.g., DDC diet). Although both are validated for PBC studies, the DDC diet model is particularly valuable for investigating DR in liver regeneration[39]. In clinical PBC, early inflammation spreads to the periportal area[40], with cholangiocyte proliferation acting as a “pacemaker” for periportal fibrosis[41] and PHT complications[42]. This pattern mirrors our DDC mouse model findings, where DR is comparable to BDL and hyperproliferative cholangiocytes surround the portal vein, triggering early PHT. Notably, portal pressure in 4-week DDC mice reaches approximately 10 mmHg, comparable to BDL and CCl₄ models. However, due to natural aversion to toxins, mice reduce DDC intake over time, preventing further portal pressure increases at 6-8 weeks differing slightly from long-term PBC progression. Thus, we selected the 4-week DDC model for PHT pathogenesis studies, as it shows peak portal pressure and the greatest difference from controls.

Classic PHT syndromes involve fibrosis from HSC activation[32], liver sinusoidal dysfunction[43], angiogenic effects, including vascular endothelial cell growth and proliferation that exacerbate intrahepatic resistance and fibrosis[34], and liver inflammatory infiltration with enzyme abnormalities due to liver injury. Most classical animal models of PHT reproduce similar pathological phenotypes. In fibrosis, classic BDL and CCl₄ mouse models show similar collagen deposition but in different locations. The DDC model mirrors BDL in portal region deposition but with less overall fibrosis, aligning with non-cirrhotic PHT in clinical PBC. However, HSC activation in the DDC mouse model is equivalent to that observed in the BDL and CCl₄ models, with even higher fold changes in the expression of the marker desmin. These comparative results indicate that significant liver fibrosis or cirrhosis has been established in the DDC diet mouse model, substantially increasing IHVR and thereby elevating portal pressure.

As the primary channel for fluid and substance exchange in the liver, the fenestrations of LSECs, along with their morphological and functional homeostasis, are critical factors in maintaining portal pressure[32]. Recent studies on LSEC function and homeostasis have further confirmed their pivotal role in PHT. For instance, increased expression of Cxcl1 in LSECs recruits neutrophils, generates sinusoidal microthrombi, and promotes PHT[18]. Additionally, endothelial p300 promotes PHT and hepatic fibrosis through C-C motif chemokine ligand 2-mediated angiocrine signaling[19]. In this study, electron microscopy revealed that the DDC diet mouse model effectively induces LSEC capillarization, characterized by reductions in both the number of fenestrae and the total fenestral area. These changes were comparable to those in the BDL and CCl₄ models. Furthermore, as reported in previous studies[44], loss of LyVE-1 expression, leading to increased α-SMA a sign of LSEC dysfunction was detected in all three models. Meanwhile, significant upregulation of LSEC capillarization markers, such as CD34 (indicated by single-cell sequencing[45] and vWF (associated with sinusoidal contraction in hepatitis C virus induced liver fibrosis[46]), was observed in all three models. eNOS, which mediates LSEC-derived nitric oxide, serves as both a classic marker of LSEC capillarization and a key protein for LSEC function. Consistent with prior reports, the DDC mouse model showed a decreased ratio of p-eNOS to total eNOS, which, alongside vascular cell adhesion molecule 1 and intercellular adhesion molecule 1, induces LSEC dysfunction[44]. In contrast, the BDL and CCl₄ mouse models exhibited a direct reduction in total eNOS levels, another indicator of impaired LSEC function and worsening PHT[47]. Despite these differing patterns of eNOS alterations across models, all models impair eNOS function, thereby exacerbating LSEC dysfunction and PHT. Notably, pharmacological rescue of eNOS protein signaling pathway activation alleviates the pathological phenotypes of LSECs and PHT[48,49].

Abnormal angiogenesis in the portal area, driven by the proliferation of vascular endothelial cells, exacerbates fibrosis[50]. For example, pharmacological inhibition of VEGFR2 protein[13] or its related signaling pathways[11] suppresses abnormal hepatic vascularization and ameliorates PHT phenotypes. In this study, the DDC mouse model exhibited even more pronounced portal vascular proliferation than the other two models. Similar to the BDL and CCl₄ models, DDC diet-induced liver disease triggered extensive portosystemic shunting a hallmark of PHT providing compelling evidence of elevated portal pressure. Furthermore, in the DDC diet mouse model, we observed significantly increased infiltration of inflammatory cells and DR phenotypes compared to the NCD group, with more pronounced elevations than in the other two models. This represents a key advantage of the DDC model to serve as an animal model for studying clinical diseases such as biliary tract disorders or immune-related cirrhotic PHT.

Over the years, animal models of PHT have evolved from initial rat-based systems to more advanced mouse-derived models. However, pathological characteristics vary between species. For instance, BDL in rats produces significant, even palpable, liver cirrhosis, whereas BDL in mice leads to milder cirrhosis but more frequent extensive liver necrosis a feature rarely seen in rat models (Supplementary Figure 1A). Furthermore, reviews of relevant literature and our team’s research indicate that BDL mouse models, when applied to clinical disease studies[51], suffer from low 4-week survival rates, limiting their use in mechanistic investigations involving drugs, proteins, or genetic interventions to alleviate PHT. The potential for survivor bias remains uncertain. Despite our efforts to enhance survival through procedural adjustments such as switching ligation sutures from 5-0 to 4-0, administering weekly vitamin K1 injections for hemostasis, and providing postoperative feeding with easily digestible jelly (Supplementary Figure 1B); however, these measures failed to significantly improve 4-week outcomes in BDL mice. In stark contrast, the DDC-induced biliary model offers substantially higher survival rates, highlighting a major advantage for advancing PHT research with DDC mouse models.

Another PHT model the CCl₄ mouse model is favored by many researchers for its simplicity and reliable success rate[52]. However, compared to the DDC diet model, the entirely feed-based method of the DDC diet is even more straight forward. The DDC diet model achieves portal pressure levels comparable to those in the CCl₄ model at 4 weeks, at least 4 weeks earlier than CCl₄ injection. Additionally, DR cannot be stably increased for 8 weeks in the CCl₄ model, and even intraperitoneal injection-induced DR at 12 weeks is slightly inferior to that in the BDL and DDC diet models. This limits the CCl₄ model’s utility as an animal model for studying biliary-related PHT syndromes.

In this study, we analyzed the advantages and disadvantages of three PHT animal models. Among them, the DDC mouse model though not originally developed for PHT research was introduced to address the insufficient etiological coverage of existing PHT models. The Mdr2-knockout mouse model, a classic for studying DR-associated PSC, also exhibits PHT pathological phenotypes, as reported in previous studies[21]. While the Mdr2-knockout model retains notable value, with ongoing exploration of its phenotypic variations to identify distinct PHT pathophysiological mechanisms and inform clinical treatments, its modeling and long-term maintenance costs are significantly higher than those of the DDC model. It should be acknowledged that this study had inherent limitations: It did not explore variables like mouse strain, gender, or age, and mouse size prevented measuring systemic hemodynamics, focusing only on intrahepatic PHT changes. Future research should address these, such as validating DDC in diverse strains or combining with genetic models for broader PHT etiologies. Additionally, while DDC excels in short-term biliary PHT, long-term studies could assess progression to cirrhosis. Overall, the DDC model advances preclinical PHT research by overcoming key limitations of BDL and CCl₄, potentially informing therapies for cholestatic diseases.

CONCLUSION

This study demonstrates that the DDC diet mouse model effectively recapitulates PHT, with portal pressure reaching approximately 10 mmHg at 4 weeks comparable to classic BDL and CCl₄ models. Notably, the DDC model exhibits a 100% 4-week survival rate, surpassing BDL (35%) and exceeding CCl₄ (58.3%-66.6%). Mechanistically, DDC induces biliary fibrosis in the portal region similar to BDL but with less deposition than CCl₄, while showing more pronounced HSC activation (desmin). SEM revealed comparable reductions in liver sinusoidal fenestrae and capillarization across models, though eNOS regulation differed. The DDC model also induced robust DR and inflammatory infiltration, outperforming BDL and CCl₄ in short-term modeling, alongside extensive extrahepatic portosystemic shunting. Its feed-based administration offers operational simplicity compared to surgical (BDL) or injectable (CCl₄) methods. These findings establish the DDC model as a superior tool for biliary disorder-related PHT research, integrating pathological relevance and high survival. Future studies should explore its application in diverse etiologies to further advance therapeutic development.