Bile acids trigger cholemic nephropathy in common bile-duct–ligated mice
Research Unit for Experimental and Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine, Medical University of Graz, Graz, Austria
Department of Pathology, Medical University of Graz, Graz, Austria
Address reprint requests to: Peter Fickert, M.D., Research Unit for Experimental and Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine, Medical University of Graz, Auenbruggerplatz 15, 8036 Graz, Austria. E-mail: email@example.com; fax: +43 316-385-17108; Michael Trauner, M.D., Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Währinger Gürtel 18-20, 1090 Wien, Austria. E-mail: firstname.lastname@example.org; fax: +43 1-40400-4735.
Potential conflict of interest: Nothing to report.
Tubular epithelial injury represents an underestimated but important cause of renal dysfunction in patients with cholestasis and advanced liver disease, but the underlying mechanisms are unclear. To address the hypothesis that accumulation and excessive alternative urinary elimination of potentially toxic bile acids (BAs) may contribute to kidney injury in cholestasis, we established a mouse model for detailed in vivo time course as well as treatment studies. Three-day common bile duct ligation (CBDL) induced renal tubular epithelial injury predominantly at the level of aquaporin 2–positive collecting ducts with tubular epithelial and basement membrane defects. This was followed by progressive interstitial nephritis and tubulointerstitial renal fibrosis in 3-, 6-, and 8-week CBDL mice. Farnesoid X receptor knockout mice (with a hydrophilic BA pool) were completely protected from CBDL-induced renal fibrosis. Prefeeding of hydrophilic norursodeoxycholic acid inhibited renal tubular epithelial injury in CBDL mice. In addition, we provide evidence for renal tubular injury in cholestatic patients with cholemic nephropathy. Conclusion: We characterized a novel in vivo model for cholemic nephropathy, which offers new perspectives to study the complex pathophysiology of this condition. Our findings suggest that urinary-excreted toxic BAs represent a pivotal trigger for renal tubular epithelial injury leading to cholemic nephropathy in CBDL mice. (Hepatology 2013; 58:2056–2069)
ethylenediaminetetraacetic acid; FXR, farnesoid X receptor
FXR KO mice
hematoxylin and eosin
monocyte chemoattractant protein 1
multidrug-resistance protein 2
periodic acid Schiff
transforming growth factor beta 1
unilateral ureter ligation
vascular cell adhesion molecule
Acute kidney injury (AKI) is a common complication in patients with end-stage liver disease and represents a high-risk situation. Because of the fact that hepatorenal syndrome (HRS), an important and principally reversible cause of renal failure in patients with liver cirrhosis, may be difficult to differentiate from other causes of AKI in clinical practice, a revised clinical classification has been proposed. Interestingly, recent studies revealed a high proportion of structural abnormalities, including vascular and tubular epithelial injuries, on renal biopsies in patients with cirrhosis with impaired renal function without proteinuria and hematuria.[3, 4] In addition, chronic cholestatic liver diseases are frequently associated with tubulointerstitial nephropathies.[5, 6] Likewise, patients with obstructive jaundice have an increased incidence of AKI and renal failure in the perioperative phase[7, 8] and frequently show acute tubular epithelial injury on renal biopsy, despite careful volume replacement therapy. Such renal alterations in cholestasis were previously also referred to as cholemic nephropathy.
Cholestasis, characterized by increased hepatic and serum bile acid (BA) levels, has also been linked to organ dysfunction in cirrhosis. Cholestatic hepatocytes attempt to limit intracellular accumulation of BAs by induced basolateral hepatocellular export and adaptive changes in the proximal renal tubule collectively facilitating their renal elimination at the expense of increasing the BA burden for the renal tubular system.[12, 13] This could cause kidney injury by BA-induced oxidative stress, endotoxemia caused by increased translocation from the intestine resulting from the enteral lack of BAs, increased production, or expression of vasoactive mediators and their receptors as well as volume depletion.[14-18] However, little is known whether and how increased urinary excreted BAs may be causally linked to AKI in cholestatic patients. Long-term common bile duct ligation (CBDL) in mice was shown to be associated with chronic cholestasis, ascites formation, and hyperaldosteronemia, but it remains undefined whether this is associated with renal pathology. The aims of this study were (1) to establish and characterize a mouse model of combined cholestatic liver and kidney disease enabling detailed time-course studies in cholemic nephropathy and (2) to identify potential mechanisms focusing on the enigmatic role of BAs.
Material and Methods
Experiments were performed with 2-month-old male C57/BL6 mice (25-30 g body weight), housed with a 12-hour light/dark cycle and permitted ad libitum consumption of water. Experimental protocols were approved by the local animal care and use committees according to criteria outlined by the National Academy of Sciences (BMWF-66.010/0045-II/10b/2010).
CBDL was performed as described previously. Before harvesting, some groups of mice were housed in metabolic cages for 24 hours for urine sampling. For detailed time-course studies of cholemic nephropathy, mice were harvested at 3 and 7 days as well as 3, 6, and 8 weeks after CBDL. In addition, the effects of CBDL were compared in farnesoid X receptor (FXR) knockout (KO) mice (FXR−/−; congenic C57/BL6; obtained from Frank J. Gonzalez, National Cancer Institute, National Institutes of Health, Bethesda, MD) and respective wild-type (WT) controls.
norUrsodeoxycholic Acid Feeding
To test the hypothesis that prefeeding of hydrophilic norursodeoxdycholic acid (norUDCA) protects mice from toxic BA-induced renal tubular injury, 7-day norUDCA-fed (0.5%) CL57/BL6 mice were subjected to CBDL and diets were continued until harvesting 3 days thereafter.
Unilateral Ureter Ligation
This model was used as a positive control to induce tubulointerstital kidney fibrosis in mice. After midline abdominal incision under general anesthesia (isoflurane; Abbott Laboratories, Maidenhead, UK), the left ureter was double ligated close to the kidney and mice were harvested 7 days thereafter.
Serum Biochemical Analysis and Liver Histology
Serum samples were stored at −80°C and subsequently analyzed for alanine aminotransferase (ALT), alkaline phosphatase (ALP), total serum BA, and urea levels by a cobas 6000 analyzer (Roche Diagnostics Corporation, Indianapolis, IN). For conventional light microscopy, livers were fixed in 3.7% neutral buffered formaldehyde solution and embedded in paraffin. Sections (2 µm thick) were stained with hematoxylin and eosin (H&E), periodic acid Schiff (PAS), and Sirius Red.
Immunohistochemistry for Vascular Cell Adhesion Molecule 1, Macrophage Marker F4/80, Aquaporin 2, and Na+/K+−2Cl− Cotransporter
Immunohistochemistry (IHC) for vascular cell adhesion molecule (VCAM)−1 was performed on acetone-fixed (−20°C for 10 minutes) cryosections (1.5 µm thick) of kidney tissue by using the purified rat anti-mouse CD106 (VCAM-1) antibody (Ab; catalog no.: 550547; dilution, 1:100; BD Pharmingen, San Diego, CA). Cells of the macrophage/dendritic lineage were detected by staining 0.1% protease XXIV–treated paraffin sections (2 µm thick) of kidney tissue with an Ab recognizing the macrophage antigen, F4/80 (rat anti-mouse F4/80; catalog no.: MCA497GA; dilution, 1:50; AbD Serotec, Oxford, UK). IHC for aquaporine 2 (AQP2) was performed on microwave-treated (ethylenediaminetetraacetic acid [EDTA]; sodium buffer, pH 8.0) paraffin sections (2 µm thick) of kidney tissue using rabbit anti-AQP2 (catalog no.: ab85876; dilution, 1:1,000; Abcam plc, Cambridge, UK). IHC for Na+/K+−2Cl− cotransporter (NKCC2) was performed on 0.1% protease XXIV–treated paraffin sections (2 µm thick) using rabbit anti-SLC12A1 Ab (catalog no.: AV41388; dilution, 1:100; Sigma-Aldrich, St. Louis, MO). Primary Abs were detected using appropriate secondary Abs, including polyclonal rabbit anti-rat immunoglobulins/biotinylated (catalog no.: E0468; Dako Denmark A/S, Glostrup, Denmark) and rabbit on rodent HRP-Polymer (catalog no.: RMR622; Biocare Medical, Concord, CA) for VCAM-1 and F4/80 and rabbit on rodent HRP-Polymer (catalog no.: RMR622; Biocare Medical) for AQP2 and NKCC2. Binding of Abs was visualized using β-amino-9-ethyl-carbazole (Dako AEC + High Sensitivity Substrate Chromogen Ready-to-Use; catalog no.: K3461; Dako Denmark A/S) as substrate.
Immunofluorescence Microscopy for Laminin, AQP2, Anion Exchanger 1, and Pendrin
Immunofluorescence (IF) for basement membrane protein laminin and AQP2 was performed on formaldehyde/methanol/acetone–fixed cryosections of kidney tissue (3 µm thick) using rabbit anti-laminin (catalog no.: ab11575; dilution, 1:200; Abcam plc) and either rabbit anti-AQP2 (catalog no.: ab85876; dilution, 1:500; Abcam plc) or goat anti-AQP2 (catalog no.: ab105171; dilution, 1:500; Abcam plc). Double-labeling IF was performed on formaldehyde/methanol/acetone–fixed cryosections of kidney tissue (3 µm thick) using rabbit anti-laminin (catalog no.: ab11575; dilution, 1:200; Abcam plc) and goat anti-AQP2 Ab (catalog no.: ab105171; dilution, 1:500; Abcam plc) as well as goat anti-AQP2 Ab (catalog no.: ab105171; dilution, 1:500; Abcam plc) combined with rabbit anti-AE1 (anion exchanger 1) Ab for staining of type A intercalated cells (dilution, 1:750) or goat anti-AQP2 Ab (catalog no.: ab105171; dilution, 1:500; Abcam plc) combined with an Ab against mouse pendrin for identification of non-type-A intercalated cells (guinea pig anti-pendrin; dilution, 1:500[21, 22]). Primary Abs were detected using the following fluorophore-conjugated secondary Abs: Alexa Fluor 594 Goat Anti-Rabbit IgG (immunoglobulin G; catalog no.: A-11037; Life Technologies, Carlsbad, CA); Alexa Fluor 488 Donkey Anti-Goat IgG (catalog no.: A-11055; Life Technologies); and Rhodamine Red-X-AffiniPure F(ab')2 Goat Anti-Guinea Pig IgG (catalog no.: Jackson ImmunoResearch Europe Ltd., Newmarket, UK). Slides where counterstained with 4',6-diamidino-2-phenylindole (DAPI; 1.5 µg/mL).
Fluorescence Microscopy for Fluorescent Ursodeoxycholic Acid
Fluorescent-labeled ursodeoxycholic acid (UDCA; ursodeoxycholyl [UDC]/Nε-4-nitrobenzo-2-oxa-1,3diazol [NBD]/lysine; molecular weight, 684; kindly provided by Prof. Dr. Alan Hofmann, University of California, San Diego, La Jolla, CA) was injected into the portal vein of 3-day CBDL mice and sham-operated controls. After relaparotomy, the portal vein was cannulated using a 27-G needle. Thereafter, UDC/Nε-NBD/lysine (100 µmol per mouse solubilized in 200 mL physiologic saline solution) was given continuously over 10 minutes. Subsequently, kidneys were excised and frozen immediately in liquid nitrogen. Unfixed cryosections of kidney tissue (3 µm thick) were mounted on slides, air-dried, and immediately examined with a fluorescence microscope (Olympus BX 51; Olympus, Tokyo, Japan).
Measurement of Renal Hydroxyproline Content
Quantification of kidney fibrosis was carried out by measurement of renal hydroxyproline concentration by a calorimetric method. In brief, at least 50 mg of frozen kidney tissue was homogenized in 6 N of HCl and hydrolyzed overnight at 110°C. After 16 hours, hydrolysates were filtered, neutralized with NaOH, and oxidized with chloramine-T. This was followed by a reaction with perchloric acid and p-dimethylaminobenzaldehyde, resulting in the formation of a chromophore quantified photometrically at 565-nm wavelengths.
Preparation of Kidney Protein and Analysis of VCAM-1 and β-Actin by Western Blotting
Protein was isolated by sonication of kidney tissue in a homogenization buffer (0.25 mol/L of sucrose, 10 mmol/L of HEPES [pH 7.5], and 1 mmol/L of EDTA [pH 8.0], containing the protease inhibitors, phenylmethylsulfonyl fluoride, aprotinin, leupeptin, and pepstatin). Protein (30 µg) was run on a 10% sodium dodecyl sulphate/polyacrylamide gel, transferred to nitrocellulose, and blotted with respective Abs (mouse VCAM-1/CD106 Ab; catalog no.: AF643; dilution, 1:1,500; R&D Systems, Minneapolis, MN; monoclonal anti-β-actin Ab; catalog no.: A5441; dilution, 1:5,000; Sigma-Aldrich). Binding was detected by using peroxidase-conjugated respective immunoglobulins (Dako), and peroxidase activity was visualized by using the enhanced chemiluminescence method western blotting detection system.[23, 24]
Determination of Hepatic Messenger RNA Levels Using Quantitative Polymerase Chain Reaction
RNA was extracted and reverse transcribed into complementary DNA (cDNA). Polymerase chain reaction reaction (20 μL) contained 12.5 ng of cDNA, 330 nM of each primer, and 10.5 μL of SYBR Green Master mix (Applied Biosystems, Foster City, CA). Expression levels of all transcripts were normalized to the housekeeping gene, 36b4. Primers used are summarized in Supporting Table 1.
Data are reported as arithmetic means ± standard deviation (SD) of 5-10 animals in each group. Statistical analysis included the Student t test, when appropriate, Mann-Whitney's nonparametric U test, or analysis of variance with Bonferroni's post-testing when three or more groups were compared, using SPSS statistics (SSPS, Inc., Chicago, IL) with the generous help of Prof. Dr. Andrea Berghold (Institute for Medical Informatics, Statistics and Documentation; Medical University Graz, Graz, Austria). A P value <0.05 was considered significant.
Long-Term CBDL Leads to Cholemic Nephropathy in Mice
To mimic chronic cholestasis, CBDL was performed for a duration of up to 8 weeks, leading to significantly elevated serum parameters for liver injury (ALT) and cholestasis (ALP and BA; Supporting Table 2), together with histological evidence for biliary fibrosis demonstrating chronic cholestatic liver injury (not shown). At the time of harvesting, 8-week CBDL mice frequently showed an impressively dilated common bile duct and obvious loss of abdominal and, especially, epididymal fat, but no signs of bile leakage or peritonitis (Fig. 1A). The kidney surface was irregular and showed a greenish color, most likely attributable to bilirubin (Fig. 1B). Kidney size and weight were significantly reduced in CBDL mice (Fig. 1C), whereas kidney/body weight ratio (not shown) did not differ significantly, compared to sham-operated controls. These structural changes were associated with increased serum urea levels (Fig. 1C) and an increased urinary volume indicative of polyuric renal failure in 8-week CBDL mice (5.2 ± 2.0 versus 1.3 ± 0.7 mL/24 hours in controls; P = 0.009). Kidneys of 8-week CBDL mice showed dilated tubuli and renal tubulointersititial nephritis and pronounced fibrosis on H&E-stained sections (Fig. 1E). Cytologic urinalysis revealed characteristic findings for tubular injury in CBDL mice, reflected by a significantly increased number of tubular cell cylinders and urinary casts (Fig. 1F). In contrast, the urine of 8-week sham-operated controls was almost free of cells and debris. Consequently, the characteristic kidney phenotype of cholemic nephropathy in long-term CBDL mice called for more-detailed mechanistic time-course studies.
The Progression of Tubular Epithelial Injury in CBDL Mice Over Time
Already after 3 day CBDL, kidneys showed tubular epithelial injury at the border region between the outer and the inner strip and in the inner medulla, with small foci of coagulation necrosis and tubular casts detectable only on PAS-stained sections at that early time point (Fig. 2B). From day 7, we observed dilated tubules and an increasing number of protein and cell casts occasionally in distal tubules and most prominent in collecting ducts in the inner medulla (Fig. 2C). In addition, kidneys frequently showed progressive partial occlusion and dilatation of distal tubules and collecting ducts in 3-, 6-, and 8-week CBDL mice (Fig. 1D-F). Concomitantly, we observed an increasing number of atrophic glomeruli over time with a dilated Bowman's space. Additional support for the conclusion that the predominant injury in response to 3-day CBDL was to collecting ducts was achieved by IHC and IF staining of AQP2 (specifically expressed in the apical plasma membrane and apical vesicles of collecting duct cells[25, 26]), showing a partial lack of AQP2 positivity and parallel loss of nuclear staining in necrotic collecting duct epithelial cells (Fig. 3A-D). In addition, serial sections convincingly showed that injured tubuli observed on PAS-stained sections corresponded nicely to AQP2-positive collecting ducts (Fig. 3D-F), whereas NKCC2-positive cells of the thick ascending limb of Henle appeared normal (Supporting Fig. 1). We found no evidence that the observed reduced AQP2 staining of collecting ducts observed in CBDL mice was the result of an increased relative number of intercalated cells, as demonstrated by double IF staining for AQP2 and AE1 for type A intercalated cells or pendrin for non-type-A intercalated cells[21, 22] (Fig. 3C and Supporting Fig. 2). Together, these findings were indicative of a loss of epithelial barrier continuity of collecting ducts in 3-day CBDL mice (Fig. 3A,C,D). Thus, we next searched for basement membrane defects of collecting ducts by double staining with anti-laminin and anti-AQP2 Abs. In contrast to controls, showing collecting ducts with continuous basement membrane (Fig. 4A), kidneys of 3-day CBDL mice showed severely altered collecting ducts, with ulceration of the epithelium and exfoliation of epithelial cells coalescing to cell casts within the lumens of collecting ducts, and frequent loss of basement membrane continuity corresponding to these areas (Fig. 4B). The functional relevance of these findings was demonstrated by leakage of portal-vein–injected and urinary-excreted UDCA/NBD/lysine in CBDL mice from the lumens of collecting ducts (Fig. 5), whereas sham-operated controls stained almost negative (not shown). In line and principally reflecting BA leakage in a well-established experimental condition, hepatic bile infarcts of the same CBDL mice stained positive with fluorescent UDC/NBD/lysine (Supporting Fig. 3). Taken together, these findings demonstrate that collecting ducts represent early targets in cholemic nephropathy in CBDL mice and suggest that a potential tubulotoxic agent with the presumably highest local concentration or toxicity at the level of collecting ducts may lead to epithelial cell injury and basement membrane damage already as early as 3 days after CBDL.
CBDL Leads to Progressive Interstitial Nephritis
Because inflammation is a well-known trigger for renal fibrosis and subgroups of peripheral monocytes may also significantly contribute to kidney fibrosis, we next tested renal VCAM-1 and macrophage/dendritic cell marker F4/80 expression. Renal VCAM-1 and F4/80 protein expression was induced over time. Because sham-operated controls did not significantly differ over the various time points studied, only 8-week sham-operated animals are shown. VCAM-1 expression was primarily induced in tubular epithelial cells and, to a lesser degree, in endothelial cells as well as the interstitium increasing over time (Supporting Fig. 4A), accompanied by significant increased levels of VCAM-1 messenger RNA (mRNA) and protein expression (Supporting Fig. 4B,C). In contrast, F4/80 expression was primarily induced in cells of the renal interstitium and within glomeruli (Fig. 6A). This renal inflammatory response in CBDL mice was accompanied by pronounced overexpression of F4/80 and monocyte chemoattractant protein 1 (Mcp-1) mRNA already at week 3 after CBDL (Figs. 6B,C).
CBDL Leads to Tubulointerstitial Kidney Fibrosis
To closely follow up the development of kidney fibrosis in CBDL mice, we compared renal collagen α1(I)) and transforming growth factor beta 1 (tgf-β1) mRNA, hydroxyproline levels, and Sirius Red–stained kidney sections at several time points after CBDL to controls. Significant renal fibrosis was already developed 3 weeks after CBDL and increased later (Fig. 7A). This was accompanied by induction of collagen α1(I) and tgf-β1 mRNA expression (Fig. 7B) and significantly elevated renal hydroxyproline levels (Fig. 7C).
CBDL FXR−/− Mice Are Protected From Renal Fibrosis
PAS staining after 3-day CBDL revealed tubular epithelial injury in both genotypes (Fig. 8A). However, we hypothesized that long-term CBDL FXR−/− mice could be protected against full-blown cholemic nephropathy resulting from a more hydrophilic BA pool composition (compared to WT controls), as shown previously, and much lower BA levels at later time points after CBDL, as demonstrated herein (Supporting Table 3). In fact, serum BA levels in 8-week CBDL mice were 20-fold higher in WT mice (2,851 ± 1,097 μmol/L), compared to 8-week CBDL FXR−/− mice (110 ± 21 μmol/L), and kidney fibrosis in long-term CBDL FXR−/− mice was indeed ameliorated, as demonstrated by Sirius Red staining and significantly lower renal hydroxyproline levels (Fig. 8B). To additionally rule out a general resistance of FXR−/− mice to renal fibrosis, we compared the degree of renal fibrosis in response to unilateral ureter ligation (UUL; as a noncholestatic condition). Notably, we found no differences in fibrotic response after UUL in FXR−/− mice, compared to WT mice (Fig. 8C). Collectively, these findings indicate that FXR−/− mice are protected from long-term, but not early, CBDL-induced tubular epithelial injury, which can be explained by declining and less hydrophobic serum BA levels in FXR−/− mice (but not WT) over time.
norUDCA Prefeeding Protects 3-Day CBDL Mice From Tubular Epithelial Injury
Next, we hypothesized that prefeeding hydrophilic norUDCA 7 days before CBDL could protect against tubular epithelial injury, because norUDCA represents the main BA in serum and urine after feeding and undergoes substantial renal elimination. Indeed, norUDCA prefeeding was able to prevent tubular epithelial injury in 3-day CBDL mice, as demonstrated by PAS- and AQP2-stained kidney sections (Supporting Fig. 5).
Cholestatic Patients With Cholemic Nephropathy Show Tubular Epithelial Injury
To substantiate our concept with human pathological findings, we screened our pathological archives for patients with cholestatic liver disease and cholemic nephropathy in whom both tissues were available for examination. Kidney histology frequently showed tubular casts, interstitial nephritis, and renal fibrosis as well as characteristic collecting duct lesions (Supporting Fig. 6).
Renal tubular injury is a major, perhaps underestimated, and still poorly understood cause of renal dysfunction in advanced liver diseases.[3, 4, 30] We hypothesized that cholestatic liver dysfunction with systemic accumulation of potentially toxic BAs, together with their exaggerated compensatory urinary elimination, may contribute to AKI in such patients. Therefore, we established and characterized a mouse model of progressive cholestatic liver disease associated with tubulointerstitial nephritis and renal fibrosis and impaired renal function, which offers novel perspectives to study the mechanisms and novel treatment strategies for cholemic nephropathy.
Experimental studies from the 1940s and 1950s frequently used dogs and indeed showed structural renal alterations including interstitial nephritis and renal fibrosis in cholestasis. However, in subsequent years, most laboratories changed to rats and the renal alterations reported in CBDL rats are notoriously mild.[32, 33] In light of our current findings in CBDL mice, it is tempting to speculate that these species differences may be, at least in part, related to differences in metabolism and transport of BAs. In addition, this may also represent a BA concentration issue because we did not observe cholemic nephropathy in 8-week 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC)-fed and 8-week multidrug-resistance protein 2 (Mdr2)−/− mice with sclerosing cholangitis and biliary type of liver fibrosis, which, however, show lower serum BA levels (data not shown). Though the degree of liver injury, ductular reaction, and fibrosis in CBDL, DDC-fed, and Mdr2−/− mice may be more or less comparable, serum BA levels are 5-fold (versus 8-week DDC) and up to 94-fold (versus Mdr2−/−) higher in CBDL mice. In addition, differences in the temporal dynamics of rising serum BA levels may come into play with a sudden increase in CBDL mice (i.e., a 40-fold increase within 24 hours), whereas serum BA levels rise continuously and slowly in DDC-fed mice with normal serum BA levels even after 4 weeks. Because the observed kidney phenotype is specific for CBDL mice with the far highest SBA levels among the tested models, it is tempting to hypothesize that BAs may represent an important trigger for the observed renal pathology in CBDL mice.[37, 38] Of interest, 3-day CBDL FXR−/− mice, which were previously shown to have high urinary BA levels 3 days after surgery, but which, later, excrete mainly polyhydroxylated/nontoxic BAs, showed tubular injury after 3-day CBDL, but were protected from renal fibrosis in the long-term course. Taken together, these findings argue for the critical role of BAs in the development of renal tubular epithelial injury. This concept is further supported by the protective effects of pretreatment of CBDL mice with hydrophilic norUDCA (Supporting Fig. 5), which is extensively excreted by the urinary route. However, alternative pathogenetic factors, such as the production of vasoactive endothelin-1 by reactive cholangiocytes or activation of Toll-like receptor pathways through increased gut permeability and bacterial translocation, together with pronounced elevation of alternatively renal-excreted cholephiles, such as BAs, need to be considered and have to be studied in detail in the future.
The association between jaundice and renal failure was first described in 1911 and is a well-known clinical phenomenon with still unresolved underlying mechanisms.[41-43] Jaundiced patients undergoing invasive procedures are at markedly increased risk of renal dysfunction.[7, 8] Morphological studies revealed severe alterations of renal tubules in patients with acute obstructive jaundice previously referred to as cholemic nephropathy. In addition, cholestatic liver diseases are associated with progressive tubulointerstitial nephropathy in early childhood.[6, 44-46] Clinical features were comparable, including early onset of tubulointerstitial nephritis and cholestasis and a renal histology characterized by tubular atrophy/dilatation, and interstitial and periglomerular fibrosis, finally leading to end-stage renal disease. Moreover, chronic cholestatic liver diseases have been reported to be associated with tubulointerstitial nephritis and Fanconi's syndrome.[47-49] Severe alcoholic steatohepatitis carries a notoriously high risk for renal failure. These usually deeply jaundiced patients frequently have AKI, and serum bilirubin levels have a major effect on patient survival. BAs have also been implicated in organ (including renal) damage in liver cirrhosis. Finally, the combination of cholestatic jaundice and renal failure is also prominent in patients with systemic inflammatory response syndrome/sepsis, raising the question of whether retained cholephiles may contribute to organ dysfunction. Interestingly, we could identify morphological evidence for collecting duct lesions and interstitial nephritis in postmortem samples from cholestatic patients with cholemic nephropathy. However, the herein presented human findings are limited by the low number of identified cases, because it was extremely difficult to obtain access to both liver and renal tissue from the same patients and numerous potential confounding factors in such critically ill patients must be acknowledged (e.g., use of high-dose vasoconstrictors, shock, and potentially nephrotoxic antibiotics). Importantly, the critical lesions at the level of collecting ducts located on the boarder at the inner and outer strip and, especially, in the inner medulla are most likely to be missed by conventional percutaneous kidney biopsy, which is, in addition, rarely performed in this clinical context.
Based on our combined findings, we suggest the following pathogenetic working model for cholemic nephropathy (Supporting Fig. 7). Severe forms of cholestasis may lead to excessive alternative renal excretion of cholephiles with (1) tubular epithelial injury, basement membrane alterations, and leakage of urine into the kidney parenchyma and obstruction of collecting ducts resulting from cell detritus and protein casts leading to (2) induction of a reactive phenotype of tubular epithelial cells with overexpression of proinflammatory and -fibrogenetic cytokines resulting in interstitial nephritis, and, consequently, (3) tubulointerstitial kidney fibrosis. This concept may have major implications for diagnosis and management of renal problems in liver disease, in particular, cholestasis.
The authors thank Dr. W. Erwa (Graz) and colleagues for performing liver tests and Andrea Deutschmann, Katharina Kinslechner, Judith Gumhold, Sabine Halsegger, and Dr. Alexander Kirsch for their excellent technical assistance. The authors also express their sincere thanks to the participants of the 13th Pichlschloss Transport Meeting (July 2012), organized by Prof. Gustav Paumgartner, Munich, for their fruitful discussion and helpful suggestions.