If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, Spain
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, Spain
Biomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, SpainGastroenterology and Hepatology Department, Hospital Universitario Ramon y Cajal, Instituto Ramon y Cajal de Investigacion Biosanitaria (IRYCIS), Universidad de Alcalá, Madrid, Spain
Biomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, SpainGastroenterology and Hepatology Department, Hospital Universitario Ramon y Cajal, Instituto Ramon y Cajal de Investigacion Biosanitaria (IRYCIS), Universidad de Alcalá, Madrid, Spain
Gastroenterology and Hepatology Department, Hospital Universitario Ramon y Cajal, Instituto Ramon y Cajal de Investigacion Biosanitaria (IRYCIS), Universidad de Alcalá, Madrid, Spain
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainPUCRS, Escola de Ciências, Laboratório de Pesquisa Em Biofísica Celular e Inflamação, Porto Alegre, Brazil
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, Spain
Grupo de Aplicaciones Biomédicas, Institut de Microelectrònica de Barcelona, IMB-CNM (CSIC), Esfera UAB, Bellaterra, SpainBiomedical Research Networking Center in Bioengineering, Biomaterials and Nanomedicine (CIBERBBN), Madrid, Spain
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, SpainDepartment of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, Switzerland
Biomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, SpainGastroenterology and Hepatology Department, Hospital Universitario Ramon y Cajal, Instituto Ramon y Cajal de Investigacion Biosanitaria (IRYCIS), Universidad de Alcalá, Madrid, Spain
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, Spain
Liver Vascular Biology Research Group, Barcelona Hepatic Hemodynamic Laboratory, IDIBAPS Biomedical Research Institute, Barcelona, SpainBiomedical Research Networking Center in Hepatic and Digestive Diseases (CIBEREHD), Madrid, SpainDepartment of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, Switzerland
Using a microfluidic liver-on-a-chip device we showed that pathological pressure is deleterious on LSEC phenotype.
•
RNAseq identified CBX7 as a key transcription factor in LSEC downregulated by pathological pressure.
•
Hepatic CBX7 downregulation by pressure was validated in patients with portal hypertension, and correlated with HVPG.
•
MiR-181a-5p was identified as a pressure induced upstream regulator of CBX7.
•
ECAD & SPINK1, targets of CBX7, were predictive of portal hypertension and clinically significant portal hypertension.
Abstract
Background and aims
Portal hypertension (PH) is a frequent and severe clinical syndrome associated to chronic liver disease (CLD). Considering the mechanobiological effects of hydrostatic pressure and shear stress on endothelial cells, we hypothesized that PH might influence the phenotype of liver sinusoidal endothelial cells (LSECs) during disease progression. The aim of this study was to investigate the effects of increased hydrodynamic pressure on LSECs and to identify endothelial-derived biomarkers of PH.
Methods
Primary LSECs were cultured under normal or increased hydrodynamic pressure within a pathophysiological range (1vs12 mmHg) using a microfluidic liver-on-a-chip device. RNAseq was used to identify pressure-sensitive genes, which were validated in liver biopsies from two independent cohorts of CLD patients with PH (n=73) vs subjects without PH (n=23). Biomarker discovery was performed in two additional independent cohorts of 104 patients with PH vs 18 w/o.
Results
Transcriptomic analysis revealed marked deleterious effect of pathological pressure in LSECs and identified chromobox 7 (CBX7) as key transcription factor diminished by pressure. Hepatic CBX7 downregulation was validated in patients with PH and significantly correlated with HVPG. MicroRNA 181a-5p was identified as pressure-induced upstream regulator of CBX7. Two downstream targets inhibited by CBX7, ECAD and SPINK1, were found increased in the bloodstream of patients with PH and were highly predictive of PH and clinically significant PH.
Conclusion
We characterize the detrimental effects of increased hydrodynamic pressure on the sinusoidal endothelium, identify CBX7 as a pressure-sensitive transcription factor, and propose the combination of two of its reported products as biomarkers of PH.
Lay summary
Increased pressure in the portal venous system that typically occurs during chronic liver disease (called portal hypertension) is one of the main drivers of related clinical complications, which are linked to a higher risk of death. In this study, we found that pathological pressure has a harmful effect on liver sinusoidal endothelial cells and identified CBX7 as a key protein involved in this process. CBX7 regulates the expression of E-cadherin and SPINK1 and, consequently, measuring these proteins in the blood of patients with chronic liver disease allows the prediction of portal hypertension and clinically significant portal hypertension.
This work was supported by the Instituto de Salud Carlos III (FIS PI20/00220 to J.G-S and PI20/PI20/01302 to A.A.), co-funded by the European Union and the Generalitat de Catalunya (AGAUR - 2021 SGR 01322). CIBEREHD is funded by Instituto de Salud Carlos III. Funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. M.O-R. has an iPFIS fellowship from the Instituto de Salud Carlos III (IFI16/00016). A.G-R has a Sara Borrell fellowship from the Instituto de Salud Carlos III (CD22/00097). M.M. is a recipient of a Río Hortega grant from Instituto de Salud Carlos III, Spain.
Disclosures
The authors declare no competing financial interests.
Data availability statement
All data are included in the main manuscript or supplementary files. Full RNAseq data deposited with accession number GSE181255.
Author contributions
M.O.-R. and A.G.-R. designed the research, performed the experiments, analysed the data, and wrote the manuscript; L.A.-J., L.P., E.C., R.P. and B.S. performed experiments. J.J.L. analysed data; M.M., L.T., L.O. and P.O. collected samples and analysed clinical data; R.V., J.B. A.A. and J.C.G.-P. critically revised the manuscript; J.G.-S. conceived the study, designed and directed the research, wrote the manuscript and obtained funding. All authors edited and reviewed the final manuscript.
Introduction
Advanced chronic liver disease (ACLD) is nowadays the 11th most common cause of death globally with approximately 1.16 million deaths per year.
Alcohol abuse, chronic viral hepatitis B or C infection and metabolic-associated fatty liver disease are amongst the most frequent causes of this condition. Portal hypertension (PH) is one of the main drivers of ACLD-related clinical complications including bleeding from gastro-esophageal varices, ascites, or hepatic encephalopathy,
The primary factor in the development of PH is a marked increase in the intrahepatic vascular resistance to portal blood flow, which occurs during the progression of fibrosis as a consequence of sinusoidal cell deregulation.
During persistent liver injury, liver sinusoidal endothelial cells (LSECs) become dysfunctional, losing their characteristic transmembrane pores named as fenestrae and increasing basement lamina deposition,
a process known as sinusoidal capillarization. Capillarization impairs the exchange of blood borne molecules with neighboring cells and contributes to hepatocyte dedifferentiation and death.
During liver injury, hepatic stellate cell (HSCs) activate and differentiate into extracellular matrix-secreting myofibroblasts, while Kupffer cells (KCs) polarize towards a pro-inflammatory phenotype recruiting monocytes to the site of injury for tissue repair.
Both microvascular dysfunction and fibrotic architectural distortion, imbalance of vasoactive mediators, and microthrombi formation leads to altered mechanical properties of the tissue, further increasing the vascular resistance and, ultimately, elevating the sinusoidal pressure. Despite the fact that LSECs, due to their particular sinusoidal location, are the first cells sensing changes in intravascular pressure, the possible direct contribution of this mechanobiological cue to LSECs dysfunctionality in the ACLD setting remains largely unknown.
Currently, hepatic venous pressure gradient (HVPG) is the most reliable method for diagnosing cirrhotic PH and allows stratification of patients with normal pressure (NP, HVPG ≤ 5 mmHg), PH (HVPG > 5 mmHg) and clinically significant PH (CSPH, HVPG ≥10 mmHg). Patients with CSPH are at risk of developing clinical complications, and are thus at increased risk of hepatic decompensation and death.
However, HVPG measurement is an invasive procedure with limitations including availability, affordability, and requiring specifically trained personnel.
prompted us to explore LSEC-derived pressure-associated biomarkers for PH and CSPH in ACLD patients. According to our hypothesis, since LSECs are directly in contact with blood flow, factors specifically secreted by this cell type in response to changes in portosinusoidal pressure might be detectable in the systemic bloodstream, being therefore potentially useful as non-invasive biomarkers.
Overall, the present study aimed at investigating the influence of pathological hydrodynamic pressure per se in LSECs function and to discover pressure-sensitive biomarkers that could be used for non-invasive assessment of portal hypertension in clinical practice.
Materials and methods
Animals
Male Wistar rats were kept at the University of Barcelona Faculty of Medicine facilities, housed three per cage, under controlled environmental conditions (19.7 ± 2 °C, 52 ± 5% humidity, 12 hours light/dark cycle) and free access to standard rodent food pellets and water. All the experimental procedures were approved by the Laboratory Animal Care and Use Committee of the University of Barcelona and performed in accordance with the European Community guidelines for the protection of animals used for experimental and other scientific purposes (EEC Directive 86/609).
Isolation of hepatic cells
Primary hepatocytes and non-parenchymal cells from healthy rats weighing 300-350 g and CCl4 induced cirrhotic rats
Boyer-Diaz Z, Aristu-Zabalza P, Andrés-Rozas M, Robert C, Ortega-Ribera M, Fernández-Iglesias A, et al. Pan-PPAR agonist lanifibranor improves portal hypertension and hepatic fibrosis in experimental advanced chronic liver disease. J. Hepatol. 74:1188–1199.
Briefly, liver was perfused, digested with 0.015% collagenase A (103586, Roche, Sant Cugat del Vallès, Spain) and mechanically disaggregated obtaining a multicellular suspension. Hepatocytes were purified by low-speed centrifugation and non-parenchymal cells were separated using a three-phase iodixanol (Optiprep™, Sigma-Aldrich, Saint Louis, MO, USA) density gradient centrifugation. Afterwards, the upper interphase, that contained the HSCs, was directly seeded, while the lower one, enriched in LSECs and KCs was further purified by differential adherence time to non-coated substrates. Highly pure (>95%) and viable (80%-95%) cells
were seeded at high density on conventional culture plates or on liver-on-a-chip devices.
Hydrodynamic pressure on a liver-on-a-chip
The study of hydrodynamic pressure in vitro was performed using an advanced sinusoid-mimicking microfluidic device (Exoliver; Supplementary Figure 1). The details of its fabrication, features and suitability for sinusoidal studies have been previously reported.
Boyer-Diaz Z, Aristu-Zabalza P, Andrés-Rozas M, Robert C, Ortega-Ribera M, Fernández-Iglesias A, et al. Pan-PPAR agonist lanifibranor improves portal hypertension and hepatic fibrosis in experimental advanced chronic liver disease. J. Hepatol. 74:1188–1199.
Ortega-Ribera M, Fernández-Iglesias A, Illa X, Moya A, Molina V, Maeso-Díaz R, et al. Resemblance of the human liver sinusoid in a fluidic device with biomedical and pharmaceutical applications. 2018;115.
Primary cells isolated from healthy rats were exposed to physiological (1 mmHg) or pathological (12 mmHg) pressures within the liver-on-chip device for 48 h. Laminar shear stress stimulus on LSECs cultures was maintained at 1.15 dynes/cm2 to preserve their functionality during long-term culture, as previously described.
Ortega-Ribera M, Fernández-Iglesias A, Illa X, Moya A, Molina V, Maeso-Díaz R, et al. Resemblance of the human liver sinusoid in a fluidic device with biomedical and pharmaceutical applications. 2018;115.
Desired hydrodynamic pressure within the device was ensured by modulating outflow height relative to LSECs culture and according to 1 mmHg = 1.36 cm H2O. Real-time assessment of pressure within the bioreactor was routinely performed in independent experimental settings with a constant flow of 1.5 ml/min and using a pressure probe both at the inflow and outflow of the device connected to a Powerlab (4 S P). Data was displayed into a LabChart v5.5.6 software file.
Device configuration and culture was performed as previously described in Ortega-Ribera et al..
Ortega-Ribera M, Fernández-Iglesias A, Illa X, Moya A, Molina V, Maeso-Díaz R, et al. Resemblance of the human liver sinusoid in a fluidic device with biomedical and pharmaceutical applications. 2018;115.
LSECs were cultured on a hydrophilic biocompatible polytetrafluoroethylene microporous membrane, and hepatocytes, KCs and HSCs at a lower layer, in contact with the endothelial fraction of the culture. Microfluidic cultures were maintained at 37 °C and 5% CO2, with 43 ml of recirculating media as follows: Dulbecco’s modified Eagle’s medium F12® was supplemented with 2.97% dextran (31392; Sigma-Aldrich, Darmstadt, Germany), 2% fetal bovine serum (04‐001‐1 A; Biological Industries, Kibbutz Beit-Haemek, Israel), 1% penicillin-streptomycin (10378-016, Biological Industries), 1% endothelial cell growth supplement (BT‐203; Biomedical Technologies, Kandel, Germany), 1% heparin (H3393; Sigma-Aldrich), 1% L‐glutamine (25030‐024; Gibco, Dublin, Ireland), 1% amphotericin B (03‐029‐1C; Biological Industries), 1 nM dexamethasone (D4902; Sigma-Aldrich), 10 ng/ml Epidermal Growth Factor (E4127; Sigma-Aldrich), 1.5 nM glucagon (16941‐32‐4, Novo Nordisk, Plainsboro, NJ, USA), 15 nM hydrocortisone (H0888, Sigma-Aldrich), and 1 μM insulin (Humulin S, Lilly S.A.).
mRNA sequencing
Primary LSECs transcriptome profile under physiological or increased hydrodynamic pressures was examined by mRNA sequencing from three independent experiments. Briefly, mRNA was isolated using the RNeasy® Micro Kit (Qiagen, Hilden, Germany) following manufacturer’s instructions. The sequencing library was prepared using 25 ng of total RNA by a Universal Plus mRNA-Seq NuGEN (0508, 9133, 9134, Tekan, Leek, The Netherlands). Single-end mRNA-seq was performed in the Illumina platform HiSeq2500. The dataset is available at the National Center for Biotechnology Information Gene Expression Omnibus database, accession number GSE181255. Genes were considered significantly deregulated when their fold-change was > 3 or < -3 and their p-value < 0.05. Venn diagrams were performed using the Venny2.1 free online software (by Juan Carlos Oliveros, BioinfoGP, CNB–CSIC). Human and murine genes were compared by homology (∼90% rat genes have an orthologue in the human genome). Canonical pathways analysis was performed using the Ingenuity Pathways Analysis software from Qiagen (content version: 49932394). Transcriptomics data from LSECs isolated from preclinical models of ACLD (CCl4; thioacetamide, TAA; and common bile duct ligation, cBDL) and alcohol associated cirrhotic patients were obtained from a previous publication from our group.
Primary rat LSECs were cultured on 12-well plate at 80-90% confluency. Two hours after the isolation, cells were transfected with Lipofectamine RNAiMAX Transfection Reagent (13778075, Thermofisher, Waltham, MA, USA) containing 50 nM mirVana™ miRNA inhibitor hsa-miR-181a-5p or scramble construct for 48 h following manufacturer’s instructions.
For miR-181a-5p expression analyses, Qiazol lysis reagent and miRNeasy® Micro Kit (Qiagen) were used for total RNA preservation and purification. MiR-181a-5p expression was assessed by using TaqMan MicroRNA Assays (Applied Biosystems, Madrid, Spain) with TaqMan Universal Master Mix II, no UNG (Applied Biosystems). Data was normalized to U6 snRNA expression.
Gene expression analysis
Total RNA was extracted and purified with TRItidy G™ (A4051,0200, Panreac, Castellar del Vallès, Spain) and TRIzol:chloroform for liver tissues or using Qiazol lysis reagent and RNeasy® Micro Kit (Qiagen) for primary cells according to the manufacturer instructions. RNA yield was quantified by Nanodrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). A total amount of 0.5 μg (tissue) and 0.15 μg (cells) RNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) in a thermal cycler (Eppendorf AG 22331, Hamburg, Germany), and qPCR was performed in a 7900 H T Fast Real-Time PCR System (Thermofisher), using the TaqMan Universal PCR Master Mix (Applied Biosystems).
RNA expression levels were normalized following the 2−ΔΔCt method, with beta actin (Actb) or glyceraldehyde-3-phosphate dehydrogenase (Gapdh) as housekeeping genes. The following TaqMan probes were used: chromobox 7 (CBX7, Rn01506264_m1 and Hs00545603_m1), Fc-gamma receptor IIb (CD32b, Rn00598391 m1), stabilin 2 (Stab2, Rn01503539_m1), Actb (Hs99999903_m1), Gapdh (Rn01775763_g1), hsa-miR-181a-5p (000480) and U6 snRNA (001973).
Scanning electron microscopy imaging
Primary rat LSECs were cultured on a circular coverglass with 18 mm of diameter and fixed overnight with 2% glutaraldehyde dissolved in 0.1 M cacodylate buffer (pH 7.4). Cells were washed 3 times with the cacodylate buffer for 5 min, incubated 1 h with 1% tannic acid and 2 h with 2% osmic acid. Afterwards, cells were dehydrated with an ethanol battery (50, 70, 90, 95 and 100%), critical-point dried with carbon dioxide, sputter coated with gold, and examined by scanning electron microscopy.
Four different cohorts of patients from the Hospital Clinic of Barcelona and the Ramon y Cajal Hospital Madrid (Spain) were included in this study:
-
Discovery and validation cohorts: these two cohorts included patients that had HVPG measurement during a transjugular liver biopsy procedure using an 18-G Tru-cut needle that yield a 20 mm length liver biopsy. Correlation between HVPG and tissue expression of pressure-sensitive candidate genes were first assessed in a retrospective discovery cohort and further validated in a larger cohort (validation cohort). The discovery cohort consisted in 12 patients without portal hypertension and 19 portal hypertensive cirrhotic patients due to chronic HCV infection. The validation cohort was prospectively collected and comprised 11 subjects without PH, 40 portal hypertensive cirrhotic patients due to HCV (n=16) and alcohol (OH, n=24) etiologies, and 14 HCV cirrhotic patients with sustained virological response (12 months after finalizing antiviral treatment). Non-portal hypertensive patients had biopsy and HVPG procedures due to previous suspicion of portal hypertension that was not confirmed and that had no signs of significant liver disease on liver biopsy examination.
-
Discovery and validation biomarker cohorts: These cohorts included patients having HVPG measurements and a peripheral blood sample collected just before HVPG measurements, and were used to discover and independently validate pressure-related non-invasive biomarkers for PH. The discovery biomarker cohort was a prospective one with 18 normotensive patients and 47 patients with CLD-derived PH enrolled at the Barcelona Hepatic Hemodynamic Unit. The validation biomarker cohort was external and retrospective, including 57 patients with CLD-derived PH from the Hepatology Unit of the Ramón y Cajal Hospital – Madrid.
Patients included in all cohorts underwent HVPG measurement due to clinical indications. The protocol of this study was reviewed and approved by the Ethical Committees for Clinical Investigation of the Hospital Clínic of Barcelona and the Ramón y Cajal Hospital, and was in according with the Helsinki Declaration of 1975, as revised in 1983. Written informed consent was obtained from each patient.
In short, a 7 F balloon-tipped catheter (“Fogarty” Edwards Lifesciences LLC, CA) was guided into the main right or middle hepatic vein for measurements of wedged and free hepatic venous pressures. The HVPG was calculated as the difference between both measurements. All measurements were taken by triplicate and averaged to obtain the baseline HVPG. Permanent tracings were obtained in each case in a multichannel recorder (GE Healthcare, Milwaukee, WI).
In addition, liver stiffness measurement was performed by transient elastography (Fibroscan; Echosens, Paris, France) together with determination of plasma albumin, bilirubin, INR, and presence of decompensation events for the calculation of the Child-Pugh and MELD scores.
Plasma collection
For the biomarker cohorts, blood was drawn from ACLD patients or healthy volunteers, collected in BD Vacutainer® K2 EDTA tubes (KFK286, Becton Dickinson), immediately kept at 4 °C, centrifuged at 1300 g for 10min at 4 °C to obtain the plasma, and further centrifuged at 3000 g for 15 min at 4 °C to remove contaminating platelets. Plasma was aliquoted and stored in the biobank of IDIBAPS or Ramon y Cajal Hospital at -80 °C following the internal agreement policy until used.
Enzyme linked immunosorbent assay (ELISA)
Levels of E-cadherin (ECAD) and serine protease inhibitor Kazal-type 1 (SPINK1) in plasma was measured by ELISA following manufacturer’s instructions. ECAD (ab233611) kits were purchased from Abcam (Cambridge, United Kingdom) and SPINK1 (DY7496-05) kits were purchased from R&D Systems (Abingdon, United Kingdom).
Statistical analyses
Data are presented as mean ± standard deviation (SD). Sample size for each experiment was evaluated according to previous and exploratory results. Statistical analysis was performed, first evaluating the normal distribution of the data using the Shapiro-Wilk test, with normality assumed for p-values greater than 0.05. One-Way ANOVA or Student's t-test was used for parametric variables or Mann-Whitney U test for non-parametric variables. The correlations were calculated by Pearson’s correlation. Statistical significance was set at a p-value <0.05. Areas under the receiver operating characteristic (AUROC) curve and cutoff values for the best specificity and sensitivity were established and positive predictive value (PPV) and negative predictive value (NPV) were calculated. Statistical analyses were performed using GraphPad Prism 8.0.2 (GraphPad Software, Inc; San Diego, CA, USA) or SPSS (IBM, Chicago, IL, USA) software.
In order to understand the effects of pathological hydrodynamic pressure in LSECs we performed bulk RNA sequencing on rat primary LSECs cultured under physiological (1 mmHg) or increased pressure (12 mmHg). Transcriptomic analysis revealed 201 deregulated genes (183 downregulated and 18 upregulated) with a fold change ≥ 3 and p-value < 0.05 in response to pathological pressure. Canonical pathways analysis showed a marked dysregulation of LSECs phenotype under increased hydrodynamic pressure, evidenced as an upregulation in pathways involved in cancer, matrix metalloproteinases inhibition, and angiogenesis, while T cell activation, eNOS signaling and antioxidant pathways were significantly downregulated (Figure 1A). Interestingly, classification of the differentially expressed genes (DEG) in molecular processes categories showed alterations in LSECs metabolism, cell death and survival, proliferation, angiogenesis/remodeling, and inflammation (Figure 1B), further demonstrating a global detrimental impact in LSECs homeostatic functions. Supplementary Table 1 shows the top 50 de-regulated genes in response to pathological hydrodynamic pressure, which includes known markers of LSEC dysfunction/capillarization like Kdr or Dll4.
Figure 1Effects of pathological pressure in LSECs cultured under pathological hydrodynamic pressure. Top upregulated and downregulated canonical pathways (A) and main molecular processes (B) modified due to pathological hydrodynamic pressure. (C) Venn diagrams comparing differentially expressed genes from LSECs submitted to pathological pressure (n=3) and LSECs isolated from 3 cirrhotic portal hypertension preclinical models and LSECs from human cirrhotic patients (n=6). Genes were considered significantly deregulated when their fold-change was > 3 or < -3 and their p-value < 0.05. Common bile duct ligation induced cirrhotic rats, cBDL; Carbon-tetrachloride induced cirrhotic rats, CCl4; Hydrodynamic pressure, HP; LSECs from human cirrhotic patients, hLSECs; Liver sinusoidal endothelial cells, LSECs; Thioacetamide‐induced cirrhotic rats, TAA.
As increased pressure is part of the complex and multifactorial pathophysiology of ACLD, we then compared the pressure specific DEG with those observed in LSECs isolated from 3 different preclinical models of ACLD: chronic CCl4, chronic TAA and cBDL. Additionally, pressure-specific DEG were also compared to those from primary LSECs isolated from alcohol-associated cirrhotic patients (hLSEC). 85% of the DEG found in human LSEC had a known rat orthologue and interestingly, 70% of the genes detected in human LSEC were also detected in rat LSEC transcriptomic analysis. Comparison between these groups showed that approximately 6% of the cirrhosis specific DEG were deregulated as well by increased pressure (specifically, DEG deregulated by pressure represented a 4.81% for CCl4, 5.91% for TAA, 6.62% for cBDL and 7.72% for hLSEC from their total DEG deregulated by cirrhosis) (Figure 1C), suggesting a contribution of pathological pressure in LSECs dysfunction in ACLD. Conversely, from the total DEG in LSECs exposed to increased hydrodynamic pressure, shared genes represented 56.7% in CCl4, 33.8% in TAA, 50.5% in cBDL and 15.9% in hLSEC.
Chromobox 7 is identified as a pressure-related factor in LSECs
We carefully analyzed the top 50 DEG list obtained from LSECs cultured under pathological pressure to identify a candidate with potential pleiotropic effects orchestrating LSECs phenotype in response to pressure. Chromobox 7 (CBX7) appeared as the only gene with transcription regulation activity that was downregulated in LSECs isolated from three preclinical models of ACLD and in LSECs isolated from cirrhotic patients (Supplementary Table 1). Analysis of CBX7 expression in preclinical models of LSECs capillarization or in human liver disease from publicly available transcriptomic studies agreed with our findings (Supplementary Table 2).
Characterization of CBX7 abundance in the main liver cell types (hepatocytes, LSECs, HSCs and KCs) revealed that LSECs are the main cell type expressing this transcription factor in the healthy liver, and the only one in which this gene has a downregulated expression in ACLD (Supplementary Figure 2).
Hepatic CBX7 downregulation correlates with HVPG in ACLD patients
To investigate the potential role of CBX7 as a pressure-sensitive biomarker we interrogated its gene expression in liver biopsies from two different cohorts of ACLD patients. The discovery cohort (Supplementary table 3) included array expression from patients without PH (n=12) and HCV-cirrhotic patients with PH (n=19) with similar age and gender distribution. The majority of patients belonged to Child Pugh class A. Six patients had an HVPG < 10 mmHg and 13 patients showed an HPVG over 10 mmHg. Liver biopsies from the validation cohort (Supplementary Table 4) included 40 patients with OH-associated (n = 24) and HCV (n = 16) etiologies and were analyzed by qPCR. Patients mean age was 58.8 ± 7.5 years for the ACLD group and 56.1 ± 16.4 for the non-PH group, both with a similar gender distribution. Child Pugh scores were evenly distributed, with 15, 16 and 9 patients with scores of A, B and C, respectively. Regarding HVPG, 7 patients had a HVPG below 10 mmHg and 33 patients above 10 mmHg.
CBX7 gene expression was analyzed in both cohorts comparing subjects without liver disease (HVPG ≤ 5 mmHg) against patients with PH (HVPG > 5 mmHg). CBX7 was significantly downregulated in patients with PH in both cohorts (Figure 2A and 2B, top; p<0.0001). CBX7 downregulation was independent of the etiology of ACLD (Figure 2C and 2D, top; p<0.0001) and significantly correlated with HVPG (Figure 2, middle panels). Table 1 and Figure 2 (bottom panels) show the AUROC values for the discovery and validation cohorts. Of note, these disclosed an excellent diagnostic performance, with AUROC above 0.964, with a sensitivity of 100% for the discovery and 92.5% for validation cohorts, and specificities of 75% and 90.9%, respectively. The performance was independent of etiology, being similar in HCV and OH patients (AUROC: 0.994 and 0.943, respectively). Thus, CBX7 expression was highly sensible detecting patients with PH. In addition, CBX7 gene expression was also assessed in a small subgroup of HCV patients with sustained virological response 12 months after antiviral therapy treatment (Supplementary Table 4). Interestingly, in these patients CBX7 was returning towards normal, being significantly higher than in un-treated HCV patients (+148% versus non-treated patients, p-value 0.0134).
Figure 2Hepatic CBX7 gene expression and correlation with HVPG. Hepatic CBX7 gene expression (Top), its correlation with HVPG (Middle) and ROC curves (Bottom) in the discovery and validation cohorts. The discovery cohort (A) included normal pressure (NP; n=12) and portal hypertension hepatitis C virus associated CLD patients (PH HCV; n=13). The validation cohort (B) included healthy normotensive individuals (NP; n=11), and portal hypertensive patients (n=30). The validation cohort is further split in patients with alcohol-associated cirrhosis (PH OH; n=18) (C) and HCV-related cirrhosis (PH HCV; n=12) (D). CBX7 gene expression data are presented as the mean ± SD, as the ratios of gene expression relative to β-actin and expressed as a percentage of the healthy group, set at 1. Data were compared with Student’s t-test (* p < 0.05) and correlations were calculated by Pearson’s correlation.
Altogether, these results suggest that CBX7 behaves as a pressure-sensitive gene in LSECs and provide the rational to further study the mechanisms regulating CBX7 expression.
miR-181 regulates CBX7 expression
Given the relevant role of microRNAs (miRNAs) as master regulators of gene expression, we used MiRmap tool
(University of Geneva, Swiss Institute of Bioinformatics) to identify miRNAs directly interacting with the CBX7 transcript. Different miRNAs were identified by the software as potential regulators of Cbx7 (Supplementary Figure 3A), however miR-181a-5p (named as mir-181a hereinafter) was identified as an upstream regulator for CBX7 both in rat and human and showed the higher predictive score (Supplementary Figure 3B). Interestingly, we found that miR-181a expression was significantly up-regulated in LSECs isolated from portal hypertensive patients with ACLD and from CCl4 cirrhotic rats (Figure 3A).
Figure 3Expression of miR181a in LSECs and its effects on CBX7. MiR-181a-5p expression in cirrhotic LSECs from rat (Rn) and human (Hs) (A), miR181a and CBX7 gene expression in LSECs after 48 h of in vitro capillarization (B, C) and in LSECs transfected with an anti-miR181a (D, E). Gene expression data are presented as the mean ± SD, as the ratios of gene expression relative to RNU6 or Gapdh and expressed in comparison to the 0 h group (B, C) or the vehicle group (D, E), set at 1. All experiments are performed with at least n=4 independent replicates. Data were compared with Student’s t-test (* p < 0.05).
In order to study the regulation of miR-181a on CBX7 we used an already established in vitro model for LSECs capillarization, consisting of in vitro culture of healthy LSECs in plastic Petri dishes to promote their dedifferentiation.
To assess the degree of dysfunction of in vitro capillarized LSEC, we analyzed the expression of landmark markers of functional LSECs, showing a significant reduction in the expression of scavenger receptors stabilin 2 (Stab2; p=0.004) and Fc-gamma receptor IIb (Cd32b; p<0.001), in addition to a marked decrease in the number of fenestrae, assessed by SEM (Supplementary Figure 4). Remarkably, the de-differentiated LSECs showed a significantly increased miR-181a expression (Figure 3B; p=0.015), while CBX7 expression was reduced (Figure 3C; p<0.0001). In order to demonstrate the interaction of miR181a with CBX7, we transfected LSECs with a specific miR181a inhibitor (anti miR181a), observing significantly higher levels of CBX7 mRNA expression (p=0.021) when miR181a was inhibited (p=0.004) (Figure 3 D-E).
CBX7 downstream targets as non-invasive biomarkers of PH
Because CBX7 is a nuclear transcription factor not expected to be found in plasma
(Protein Atlas), we wondered whether a non-invasive biomarker for assessing PH/CSPH could be identified using E-cadherin (ECAD) and serine peptidase inhibitor, Kazal type 1 (SPINK1), two proteins modulated by CBX7 according to previous studies.
ECAD and SPINK1 expression were analyzed in plasma samples from two independent biomarker cohorts. The discovery biomarker cohort included ACLD patients from three different etiologies (HCV, OH and metabolic-associated CLD) evenly distributed (38.3%, 34% and 27.7% respectively) with a mean age of 57 ± 7 years, a mean HVPG of 14.2 ± 5.6 mmHg. Most patients were Child Pugh A (31 out of 47 ACLD patients). Patient demographics, standard liver function tests and platelet count are shown in Supplementary Table 5.
Plasma levels of ECAD (Figure 4A) and SPINK1 (Figure 4B) were significantly upregulated (p<0.0001 and p=0.001, respectively) in patients with PH compared to patients with normal HVPG. AUROC curves to stratify patients between normal or increased portal pressure (Figure 4C-D) were calculated for both markers alone or combined using the following formula: Y = 1/1 + exp (+4.503 – (0.06*ECAD(ng/ml))-(0.055*SPINK1(ng/ml), obtained from a binary logistic regression.
Figure 4Analysis of ECAD, SPINK1 and their combination to predict portal hypertension. ECAD (A) and SPINK1 (B) in plasma of healthy portal normotensive humans (NP; n=18) compared to ACLD patients with portal hypertension (PH; n=47) of three different etiologies: metabolic-associated steatohepatitis (MASH), alcohol-associated (OH) or hepatitis C virus (HCV). ROC curve of ECAD, SPINK1 and the combination of ECAD+SPINK1 (ES) (C) and its performance (D). Data were compared with Student’s t-test (* p < 0.05). Values of area under the ROC curve (AUROC), the 95% confidence interval (CI), the cut-off value with better sensitivity (Sens) and specificity (Spec), positive predictive value (PPV) and negative predictive value (NPV). E-cadherin, ECAD; serine protease inhibitor Kazal-type 1, SPINK1.
A cut-off value of 0.579 for the combination of ECAD+SPINK1 exhibited the best AUROC (0.911), as compared to AUROC for the individual biomarkers (0.889 for ECAD and 0.747 SPINK1). Test performance statistics for ECAD+SPINK1 showed a 91.3% sensitivity and 83.3% specificity (Figure 4D).
Regarding CSPH, ECAD expression was significantly upregulated in patients with CSPH (Figure 5A, p=0.05) while SPINK1 upregulation was not (Figure 5B; p=0.32). AUROC curves (Figure 5C-D) for the performance identifying patients with HVPG below and above 10 mmHg were calculated for both markers and their combination using the formula as follows: Y = 1/1 + exp (+1.116 – (0.020*ECAD (ng/ml))-(0.013*SPINK1(ng/ml)). The combination of ECAD+SPINK1 showed the best predictive power; using a cut off of 0.647 the test sensitivity was 91.4% and specificity 63.6% (Figure 5D).
Figure 5Analysis of ECAD, SPINK1 and their combination to predict clinically significant portal hypertension (CSPH). ECAD (A) and SPINK1 (B) in plasma of patients with advanced chronic liver disease with sub-clinical portal hypertension (HVPG<10; n=11) or with CSPH (HVPG≥10; n=36) of three different etiologies: metabolic-associated steatohepatitis (MASH), alcohol-associated (OH) or hepatitis C virus (HCV). ROC curve of ECAD, SPINK1 and the combination of ECAD+SPINK1 (ES) (C) and its performance (D). Data were compared with Student’s t-test (* p < 0.05). Values of area under the ROC curve (AUROC), the 95% confidence interval (CI), the cut-off value with better sensitivity (Sens) and specificity (Spec), positive predictive value (PPV) and negative predictive value (NPV). E-cadherin, ECAD; serine protease inhibitor Kazal-type 1, SPINK1.
The AUROC of ECAD+SPINK1 was compared and combined with the AUROC of other known markers of liver fibrosis or PH, including transient elastography, ALT, AST, platelet count, the combination of platelets and bilirubin, the combination of transient elastography and platelets, and the Fib4 and APRI scores (Supplementary Figure 5A). These analyses showed that ECAD+SPINK1 has similar specificity as previously described markers, but had a higher sensitivity than most, which means that it performs better in detecting those patients that truly have CSPH, and is only outperformed by platelets and the combination of platelets and elastography. Importantly, ECAD+SPINK1 improves the overall predictive capacity of other markers when combined (Supplementary Figure 5B). ECAD showed a significant correlation with ALT, AST, platelets and the APRI score, while SPINK1 showed no significant correlation with any parameter (Supplementary Figure 6).
Finally, we evaluated the predictive capacity of ECAD and SPINK1 in an independent validation cohort of 57 patients with PH and CSPH that also included ACLD patients from three different aetiologies: 31.6% HCV, 24.6% OH, and 43.8% metabolic-associated CLD, and a mean HVPG of 15 ± 6.9 mmHg. Supplementary Table 6 shows patients demographics and standard liver function tests. The combination of both biomarkers had an AUROC of 0.941, with a sensitivity of 85.7%, specificity 94.7%, PPV of 96.77% and NPV 78.16%, at a cut-off value of 0.700.
Discussion
Portal hypertension is the main factor leading to clinical complications in patients with ACLD, triggering life-threatening manifestations and reducing survival.
Increased pressure in the portal venous system develops as a consequence of increased resistance to portal blood flow, which in ACLD is determined by microcirculatory dysfunction and structural abnormalities primarily driven by impaired function of sinusoidal cells.
Recent research suggests that cirrhosis-derived mechanobiological signals, such as increased liver stiffness or changes in the matrisome may, per se, represent key stimuli inducing and perpetuating sinusoidal dysfunction and further aggravating the disease in a vicious cycle.
Some previous studies have addressed the effects of increased pressure on the vascular endothelium, showing increased proliferation, release of endothelin 1, and cytoskeletal reorganization with decreased expression of adhesion molecules
In the liver field, a recent study described C-X-C motif chemokine ligand 1 upregulation in LSECs exposed to mechanical stretch, inducing a pro-inflammatory niche for neutrophil recruitment and microthrombi formation.
Nevertheless, it is important to note that data addressing the effect of pressure in major vessels (within a range of 50-150 mmHg) or biomechanical cyclic stretch might not be the best models to explore LSECs, since these highly specialized endothelial cells are subject to lower pressure and non-pulsatile flow.
In the present manuscript, we specifically investigate the impact of hydrodynamic pressure in primary LSECs using a physiologically relevant experimental approach.
Our findings demonstrate for the first time that the deleterious effect of increased pressure on LSECs is defined by a multifaceted deregulation in their phenotype and function, showing modifications in pathways involved in hepatic fibrosis, vasodilatation or inflammation, thus revealing the significant contribution of increased hydrodynamic pressure in LSECs dysfunction.
Transcriptomic analysis allowed us to identify CBX7 as an LSEC-specific pressure-sensitive transcription factor in humans, which has not been previously described. CBX7 is a member of the canonical polycomb repressive complex 1, an epigenetic regulator of histones that represses the transcription of genes involved in different processes including cell cycle, environmental stress response, cell fate transition, and cell proliferation and differentiation
Expression parameters of the polycomb group proteins BMI1, SUZ12, RING1 and CBX7 in urothelial carcinoma of the bladder and their prognostic relevance.
However, its role in any endothelium or as a pressure-sensitive molecule has never been outlined.
Considering CBX7 cellular functions as transcription regulator and its central expression in the liver endothelium, we explored its potential as a novel molecular marker of PH. Studies in human liver samples from two independent cohorts of portal hypertensive patients with ACLD of diverse etiologies showed a negative correlation between CBX7 and HVPG. Moreover, CBX7 was able to accurately stratify ACLD patients according to the presence and severity of portal hypertension. Interestingly, HCV patients having achieved a sustained virological response one year before by treatment with antivirals, exhibited lower HVPG than non-treated ACLD patients, which was associated with normalized CBX7 levels.
To further understand CBX7 biology in response to hydrodynamic pressure we explored its upstream regulation. MiR181a was identified as a specific regulator of CBX7 using miRmap tool.
We found three different target sites in the CBX7 mRNA for miR-181a-5p regulation in both rats and humans. Shear stress has already been reported to modulate the expression of specific miRNAs in endothelial cells
Nevertheless, we herein report for the first time that this miRNA responds to changes in hydrodynamic pressure in LSECs. Furthermore, in our study, miR181a was also found to be upregulated in murine LSECs in cirrhosis and during in vitro capillarization, which leads to a decrease in CBX7 gene expression. The interaction between miR181a and CBX7 was further confirmed by showing that miRNA181a silencing induced an upregulation of CBX7 levels in vitro.
As stated above, HVPG is the standard technique for assessing PH but, despite being the most trustable procedure, it has some limitations due to the fact that it is an invasive procedure with a non-negligible cost and that requires specifically trained personnel.
Therefore, the discovery of a non-invasive test for assessing this syndrome is of high interest. Recently, different elastography techniques have been in the spotlight as non-invasive methods for assessing PH and fibrosis.
Still, these techniques have limitations that do not apply to serum-based tests that could be measured in routine hospital laboratory. Because of this, we focused on two CBX7 downstream targets (ECAD and SPINK1) as potential plasma biomarkers to categorize ACLD patients based on their HVPG.
ECAD is a cell adhesion molecule expressed by different cell types that can be found in the plasma as soluble e-cadherin, resulting from the proteolytic cleavage of the cell surface ECAD and indicating several processes including a disruption in cell-cell interaction and increases in cell migration and proliferation.
Our results showed that ECAD could be a good indicator (and better than SPINK1) for both assessing the presence of PH and of CSPH independently of the etiology of the ACLD. Moreover, we found that combining ECAD and SPINK1 increased the sensitivity and specificity of the predictive ability of the test, with a better predictive power than each marker alone. ECAD + SPINK1 were able to discriminate patients with normal HVPG from those with portal hypertension with a sensitivity of 93.1% and specificity of 83.3%.
Taking into account that an HVPG ≥ 10 mmHg signals an increased risk of ACLD-related complications,
we examined if the use of this combination of biomarkers can detect patients with CSPH, at high risk of progressing to clinical decompensation, that may deserve initiating treatment with non-selective beta-blockers.
β blockers to prevent decompensation of cirrhosis in patients with clinically significant portal hypertension (PREDESCI): a randomised, double-blind, placebo-controlled, multicentre trial.
ECAD + SPINK1 classified patients with or without CSPH with 91.4% sensitivity and 63.6% specificity, which is in the upper range or superior to other non-invasive methods reported so far including transient elastography
Our findings were reinforced in an independent validation cohort, where we obtained a comparable predictive score. Additionally, we show that combining the expression of these CBX7-related targets with other parameters (including elastography) increases their predictive value. Therefore, this novel analytical approach, which uses simple tools such as ELISA, could be easily implemented in large studies to assess whether it may avoid HVPG measurement for patient selection for preventive therapy and potentially to assess the effect of therapy. Moreover, an increased HVPG is also an indicator of an increased risk of hepatocellular carcinoma (HCC) development. Since loss of CBX7 expression has been also described in HCC,
it is likely that measuring plasma biomarkers of CBX7 expression may also have a role in assessing the risk of HCC. Thus, our findings open new avenues for future research deciphering the predictive capacity of the herein described biomarkers for the detection of CSPH and HCC in patients with ACLD.
We are aware of the limitations of our study; the exact mechanism by which LSECs sense hydrodynamic changes in pressure remains unknown and forthcoming studies will be necessary to uncover this mechanosensing cue. Pressure-sensitive mechanoreceptor Piezo1
might be implicated in this process. Moreover, to fully consider the proposed novel biomarkers as a proper method to predict CSPH or HCC, further confirmatory studies with larger international populations are needed. On the other hand, and considering the results obtained with the different liver-related markers, we believe that a holistic non-invasive characterization of patients with portal hypertension should rely on the analysis of a combination of several markers, including biomarkers of endothelial de-differentiation.
In summary, our study sheds insight into the prominent role that pathologically relevant pressure exerts in the liver sinusoidal endothelium, inducing the dysregulation of endothelial-relevant pathways, with a direct implication of CBX7 and its upstream regulator miR-181a. Based on these findings, we propose the combination of the plasmatic levels of ECAD and SPINK1, two of CBX7 targets, as non-invasive biomarkers for PH onset and CSPH monitoring in ACLD patients.
Acknowledgements
This study was carried out at the Esther Koplowitz Center – IDIBAPS. The fabrication of Exoliver was performed by the platform of Production of Biomaterials and Biomolecules of the ICTS “NANBIOSIS”, more specifically by the U8 Unit of the CIBER in Bioengineering, Biomaterials & Nanomedicine (CIBERBBN) at the IMB-CNM (CSIC). Authors are indebted with former and current lab members from the Liver Vascular Biology Research Group and caregivers from the Barcelona Hepatic Hemodynamic Unit at the Hospital Clinic for their skilled technical assistance. We are indebted to the HCB-IDIBAPS and the Hospital Ramon y Cajal Biobanks for sample and data procurement.
Appendix A. Supplementary data
The following is/are the supplementary data to this article:
Boyer-Diaz Z, Aristu-Zabalza P, Andrés-Rozas M, Robert C, Ortega-Ribera M, Fernández-Iglesias A, et al. Pan-PPAR agonist lanifibranor improves portal hypertension and hepatic fibrosis in experimental advanced chronic liver disease. J. Hepatol. 74:1188–1199.
Ortega-Ribera M, Fernández-Iglesias A, Illa X, Moya A, Molina V, Maeso-Díaz R, et al. Resemblance of the human liver sinusoid in a fluidic device with biomedical and pharmaceutical applications. 2018;115.
Expression parameters of the polycomb group proteins BMI1, SUZ12, RING1 and CBX7 in urothelial carcinoma of the bladder and their prognostic relevance.
β blockers to prevent decompensation of cirrhosis in patients with clinically significant portal hypertension (PREDESCI): a randomised, double-blind, placebo-controlled, multicentre trial.
00053-8&id=ga1.jpg){kind=link}
00053-8&id=gr1.jpg){kind=link}
00053-8&id=gr2.jpg){kind=link}
00053-8&id=gr3.jpg){kind=link}
00053-8&id=gr4.jpg){kind=link}
00053-8&id=gr5.jpg){kind=link}