Potential conflict of interest: Nothing to report.
See Editorial on Page 9
Renal sodium retention in experimental liver cirrhosis originates from the distal nephron sensitive to aldosterone. The aims of this study were to (1) determine the exact site of sodium retention along the aldosterone-sensitive distal nephron, and (2) to evaluate the role of aldosterone and mineralocorticoid receptor activation in this process. Liver cirrhosis was induced by bile duct ligation in either adrenal-intact or corticosteroid-clamped mice. Corticosteroid-clamp was achieved through adrenalectomy and corticosteroid supplementation with aldosterone and dexamethasone via osmotic minipumps. 24-hours renal sodium balance was evaluated in metabolic cages. Activity and expression of sodium- and potassium-dependent adenosine triphosphatase were determined in microdissected segments of nephron. Within 4-5 weeks, cirrhosis induced sodium retention in adrenal-intact mice and formation of ascites in 50% of mice. At that time, sodium- and potassium-dependent adenosine triphosphatase activity increased specifically in cortical collecting ducts. Hyperaldosteronemia was indicated by increases in urinary aldosterone excretion and in sgk1 (serum- and glucocorticoid-regulated kinase 1) mRNA expression in collecting ducts. Corticosteroid-clamp prevented induction of sgk1 but not cirrhosis-induced sodium retention, formation of ascites and stimulation of sodium- and potassium-dependent adenosine triphosphatase activity and expression (mRNA and protein) in collecting duct. These findings demonstrate that sodium retention in cirrhosis is independent of hyperaldosteronemia and of the activation of mineralocorticoid receptor. Conclusion: Bile duct ligation in mice induces cirrhosis which, within 4-5 weeks, leads to the induction of sodium- and potassium-dependent adenosine triphosphatase in cortical collecting ducts, to renal sodium retention and to the formation of ascites. Sodium retention, ascites formation and induction of sodium- and potassium-dependent adenosine triphosphatase are independent of the activation of mineralocorticoid receptors by either aldosterone or glucocorticoids. (HEPATOLOGY 2007;46:173–179.)
Patients with liver cirrhosis and portal hypertension develop renal sodium retention which promotes formation of ascites and peripheral edema. The “underfill theory” initially attributed sodium retention to secondary hyperaldosteronism accounted for by renal hypoperfusion brought about by intraabdominal fluid sequestration.1
However, the underfill theory cannot explain the early phase of cirrhosis as sodium retention may precedes formation of ascites.2–7 Thus, the “overflow theory” postulated that edema and ascites formation resulted from primary renal sodium retention promoting plasma volume expansion.8 This theory is supported by the fact that abolishing portal hypertension in cirrhotic dogs by side-to-side portocaval shunt prevented ascites formation but not sodium retention.5, 6 Later, the “revised underfill theory” proposed that peripheral arterial vasodilatation leading to decreased effective blood volume was the primary determinant of intravasal underfilling.9–11 Although it is consistent with experimental findings in different models of portal hypertension,12, 13 several studies performed mostly in humans failed to support the concept of arterial vasodilatation as a main afferent mechanism in pre-ascitic sodium retention.11
Activation of the mineralocorticoid receptor (MR) in sodium retention in cirrhosis is also controversial. On the one hand, several clinical trials showed that the MR antagonists induced resolution of ascites in a majority of patients with cirrhosis.14–18 Spironolactone also prevented renal sodium retention and ascites formation in experimental cirrhosis.4 On the other hand, several reports indicated normal renin and aldosterone plasma levels in cirrhotic patients.2, 4, 7, 19–22 Furthermore, saline loading decreased plasma renin and aldosterone in patients with cirrhosis without resolving sodium retention.23
To reconcile these findings, it has been proposed that aldosterone sensitivity was increased in cirrhosis.24–27 This apparent high sensitivity to aldosterone might be accounted for in fact by the abnormal activation of MR by glucocorticoids. Supporting this hypothesis was the finding that 11β hydroxysteroid dehydrogenase type 2 (11β -HSD2) activity, which prevents access of glucocorticoids to MRs,28 is decreased in cirrhosis.29–31 Again, this hypothesis is supported by some but not all experimental results. Thus, inhibition of corticosterone production or pharmacological inhibition of MR prevented sodium retention in cirrhotic rats,32 but adrenalectomy failed to induce natriuresis in cirrhotic rats.33
Along with the uncertainties concerning the role aldosterone and MR, there is no consensus either on the renal site of sodium retention. Although several reports in experimental models of cirrhosis show that sodium retention originates along the aldosterone-sensitive nephron segments,29, 32, 34–39 other reports concluded to the thick ascending limb of Henle's loop as the main site of sodium retention in rat models.40, 41
The present study aimed at (1) localizing the site of sodium retention along the nephron, and (2) at determining the role of aldosterone and MR in sodium retention. We therefore developed a mouse model of cholestatic cirrhosis: sodium- and potassium-dependent adenosine triphosphatase (Na+,K+-ATPase) activity was used as a marker of sodium reabsorption, and the role of aldosterone and MR was evaluated in corticosteroid-clamped animals.
Male CD1 mice (Charles River, L'Abresle, France) weighing 20-25g at onset of experiments were housed and handled according to french legislation. Animals were anesthetized by intraperitoneal injection of a mix containing Dormitor (Pfizer, Karlsruhe, 0.5μg/g bw), Climasol (Gräub, Bern, 5μg/g bw) and Fentanyl (Janssen Cilag, Issy les Moulineaux, 50ng/g bw), and were awaken with subcutaneous injection of a mix containing Antisedan (Pfizer, 2.5μg/g bw), Sarmasol (Gräub, 0.5μg/g bw) and Narcan (Serb Laboratoires, Paris, 1.2 μg/g bw).
Experiments were performed either in mice in which bile duct was ligated and partially excised (BDL mice) or in sham-operated mice. Two experimental series were carried on adrenal-intact and corticosteroid-clamped mice respectively. Corticosteroid-clamp was achieved by bilateral adrenalectomy and supplementation with 20μg/kg/day aldosterone and 25μg/kg/day dexamethasone through subcutaneous osmotic pump (Alzet 1002, Charles River). Adrenalectomy was performed 7 days after bile duct ligation (BDL) or sham operation and a first minipump was implanted at that time. It was replaced 13 days later by a second minipump delivering the same doses of steroids.
Mice were fed the usual laboratory chow ad libitum with free access to tap water. Starting 10 days after BDL, mice were housed in individual metabolic cages (Phymep, Paris) and daily food intake and urine excretion were monitored. Urine sodium and creatinine concentrations were determined using an automatic analyzer (Konelab 20i, Thermo, Cergy). Daily urinary sodium balance was calculated as the difference between dietary sodium intake and urinary excretion. Urine aldosterone concentration was determined by radioimmunoassay on the week prior animal sacrifice in series 1 and in the middle of the stage of each minipump in series 2. The two measurements gave similar results and only data of the second measurements are presented.
Animals were sacrificed 26-32 days after surgery. Ascites was quantitated by absorbing it on a paper and weighing, and nephron segments were dissected from collagenase-treated kidneys.42 For RNA extraction, tubules were isolated under “RNase-free conditions”,43 and for immunoblotting, antiproteases (phenylmethylsulfonyl fluoride 0.6mM; leupeptin 0.2μM) were added to the dissection solution and bovine serum albumin was omitted.
Na+,K+-ATPase activity was determined on pools of 4-6 nephron segments as previously described.42 Total ATPase activity was determined in a solution containing 100mM NaCl, 5mM KCl, 10mM MgCl2, 1mM ethylenediaminetetraacetic acid, 100mM Tris-HCl, 10mM Na2ATP, and 5nCi/μl -32P]-ATP (Dupont, Boston, MA) (2-10Ci/mmol) at pH 7.4. For Na+,K+-independent ATPase activity measurements, NaCl and KCl were omitted, Tris HCl was 150mM, and 2mM ouabain was added. Na+,K+-ATPase activity was taken as the difference between total and Na+,K+-independent ATPase activities.
RT-Real Time PCR.
RNAs were extracted from pools of 20-50 microdissected cortical collecting ducts (CCDs).43 Tubules were transferred into 400μl of denaturating solution (4M guanidium thiocyanate, 25mM sodium citrate, pH 7.0, 0.5% sarcosyl, 0.1M β-mercaptoethanol and 20μg glycogen). After phenol-chloroform extraction and isopropyl alcohol precipitation, the final pellet was vacuum-dried and resuspended in 5μl of water containing 40U/ml ribonuclease inhibitor (Promega). Reverse transcription was performed using first strand cDNA synthesis kit for RT-PCR (Roche Diagnostics), according to the manufacturer's protocol. Real time PCR was performed using a cDNA quantity corresponding to 0.1mm of nephron on a LightCycler (Roche Diagnostics) with the DyNAmo Capillary SYBR Green qPCR kit (Finnzymes, Saint Quentin en Yvelines) according to the manufacturer's protocol, except that the reaction volume was reduced to 8μl. PCR products were calculated as percent of a standard. Results (arbitrary unit per mm tubule length) are expressed as means ± SE from several animals. Specific primers (sequences available on request) were designed using ProbeDesign (Roche Diagnostics).
Pools of CCDs (20-30mm) were transferred into 400μl of microdissection solution and centrifuged for 5min at 500g. The cell pellet was resuspended in 20μl of dissection solution and solubilized at 60°C for 20 minutes after addition of one volume of 2X Laemmli. Sodium dodecyl sulphate-polyacrylamide gel electrophoresis was performed on 7% polyacrylamide gels. Proteins were electro-transfered to polyvinylidine difluoride membranes (Polyscreen, Dupont de Nemours). After blocking, blots were successively incubated with anti-Na+,K+-ATPase antibody (dilution 1/10,000) raised against the holoenzyme purified from rat kidney44 and goat anti-rabbit IgG antibody (dilution 1/20,000) coupled to horseradish peroxidase (Promega), and revealed by chemiluminescence with SuperSignal West Pico Substrate (Pierce Biotechnology, France). Densitometry was quantitated by digital image analysis (Image Scan, Molecular Dynamics), expressed as a function of the tubular length, and thenafter as percent of the mean densitometry of the corresponding control.
Results are expressed as means ± SE from several animals. Comparison between groups was performed by variance analysis followed by PLSD Fisher test.
Series 1: Adrenal-intact mice
Approximately 10% of BDL mice died from surgery or liver disease. In all surviving BDL mice, liver histology revealed established liver cirrhosis with significant infiltration of connective tissue in the portal zone and enhanced proliferation of bile duct epithelial cells and hepatocytes. However, only 50% of BDL mice displayed ascites at the time of sacrifice. Although the volume of ascites was highly variable (range: 0.1-12ml; mean ± SE: 2.1 ± 0.8ml; n = 14), no correlation was found between the volume of ascites and biological or biochemical parameters. In contrast, these parameters were different in BDL mice with or without ascites, justifying that cirrhotic mice were allocated a posteriori either to an ascitic or a non-ascitic group (BDL+ and BDL− mice respectively).
Urinary excretion of creatinine remained constant throughout the study and was not different in the three groups of mice (mmol/day ± SD: 4.8 ± 0.4; 5.1 ± 0.5 and 4.5 ± 0.4 in sham-operated, BDL− and BDL+ mice respectively). Sham-operated mice displayed positive daily sodium balance (≈100μmol/day) over the whole period of study, and consequently their cumulative sodium balance increased linearly with time over the 2-week period of metabolic study (Fig. 1). BDL+ mice also displayed positive sodium balance throughout the study but, starting on day 21 after BDL, their cumulative sodium balance became higher than in sham-operated mice. Cumulative sodium balance in BDL− mice was higher but not statistically different to sham-operated mice.
As compared with sham-operated mice, excretion of aldosterone increased significantly in BDL− and BDL+ mice (pmol/day ±SE: Sham-operated, 14.7 ± 3.2, n = 10; BDL−, 21.5 ± 1.7, n = 9, P < 0.001 versus sham; BDL+, 35.6 ± 8.6, n = 4, P < 0.001 versus sham and P <0.025 versus BDL−).
Na+,K+-ATPase activity increased markedly specifically in CCDs of BDL mice, and this stimulation was more pronounced in BDL+ than in BDL− mice (Fig. 2). Na+,K+-ATPase activity was reduced in proximal convoluted tubules (PCTs) and proximal straight tubules (PSTs) of BDL mice (Fig. 2).
Series 2: Corticosteroid-Clamped Mice
The overall mortality in corticosteroid-clamped mice was also ≈10%, and approximately half of the surviving BDL mice displayed ascites at the time of sacrifice (range: 0.1-13ml; mean ± SE: 1.4 &lusmn; 0.6ml; n = 23). Accordingly, animals were distributed a posteriori in BDL+ and BDL− groups depending on the presence or absence of ascites at the time of sacrifice.
Urinary excretion of aldosterone was similar in corticosteroid-clamped and adrenal-intact mice (Fig. 3). In corticosteroid-clamped animals, BDL did not modify excretion of aldosterone, attesting the efficiency of the model (Fig. 3). Corticosteroid-clamp efficiency was also attested by the lack of over-expression of serum- and glucocorticoid-induced kinase 1 (sgk1) mRNA in CCDs of cirrhotic corticosteroid-clamped mice, conversely to adrenal-intact ones (Fig. 3).
Corticosteroid-clamped BDL+ mice displayed higher cumulative sodium balance than corticosteroid-clamped sham operated mice (Fig. 4). Corticosteroid-clamped BDL− mice had also slightly higher sodium balance than corresponding controls, but the difference did not reach statistical significance.
In CCDs of corticosteroid-clamped animals, Na+,K+-ATPase activity was increased in BDL mice and the effect was higher in BDL+ than in BDL− mice (Fig. 5A). Stimulation of Na+,K+-ATPase activity in CCDs of BDL mice was similar in adrenal-intact and corticosteroid-clamped mice (compare Fig. 2 and Fig. 5A). Stimulation of Na+,K+-ATPase was associated with parallel increases in the mRNA and protein amounts of its α1 subunit (Fig. 5B-D), suggesting transcriptional induction. Conversely, the amount of mRNAs encoding α, β and γ subunits of epithelial socium channel (ENaC) was not increased in corticosteroid-clamped cirrhotic mice (not shown).
This work first aimed at developing a mouse model of cirrhosis to benefit, in the future, from genetic tools to elucidate the mechanism of sodium retention. Comparison with BDL rats shows that liver histological lesions were similar in the two models.3 The renal response was more precocious in mice than rats since about 50% of BDL mice already displayed ascites 4.5 wk after surgery whereas no rat was ascitic at 5 weeks.32 In both BDL rat and mouse models, the volume of ascites was highly variable.32
The second aim was to localize the site of sodium retention along the nephron. For this purpose, we searched for increased Na+,K+-ATPase activity in discrete nephron segments of cirrhotic mice. As a matter of fact, Na+,K+-ATPase is the motor of sodium reabsorption in all nephron segments,45 and its activity is correlated with sodium transport. Results confirmed the critical role of CCD in sodium retention, previously proposed in various models of cirrhosis.34, 36, 38 Curiously, however, no change in Na+,K+-ATPase activity was observed in distal convoluted tubule (DCT), connecting tubule (CNT) and outer medullary collecting duct (OMCD), the other targets sites of aldosterone, although increased apical targeting of ENaC (CNT and OMCD) or overexpression of the sodium chloride cotransporter (DCT) were previously reported in cirrhotic animals.34, 36, 38 This suggests that, except in the CCD, the stimulatory effect of aldosterone on apical sodium transporters is counterbalanced by a post MR mechanism that prevents induction of Na+,K+-ATPase and thereby limits sodium retention. Na+,K+-ATPase was not altered in the medullary and cortical thick ascending limb of Henle's loop (MTAL and CTAL), although this nephron segment was proposed as a main site of sodium retention in BDL rats.40, 41
Na+,K+-ATPase activity was decreased in the proximal tubule of BDL mice, confirming previous findings in cirrhotic BDL rats showing decreased proximal sodium reabsorption and expression of Na+,K+-ATPase and of the sodium proton exchanger, the main apical sodium transporter in proximal tubule.3
The third aim was to evaluate the role of hyperaldosteronemia in sodium retention in cirrhosis. Although BDL mice displayed significant hyperaldosteronemia, attested to by increased excretion of aldosterone and induction of sgk1 in CCD, experiments in corticosteroid-clamped animals definitely demonstrate that hyperaldosteronemia is not required for sodium retention, ascites formation and induction of Na+,K+-ATPase in CCD in cholestatic mice. This is consistent with clinical observations of sodium retention in cirrhotic patients with normal renin activity and plasma aldosterone level2, 4, 7, 19, 20, 22 and experimental findings showing that adrenalectomy does not promote natriuresis in cirrhotic rats.33 Altogether, these findings invalidate the underfill theories of sodium retention in liver cirrhosis and are compatible with a primary activation of sodium retention.
The last issue concerned the potential role of abnormal MR activation by glucocorticoids in sodium retention. MR displays similar affinities for aldosterone and for glucocorticoids in vitro, and cortisol (in human) or corticosterone (in rat and mice) is present at 100 times the level of aldosterone in plasma. In mineralocorticoid-sensitive cells, saturation of MR by glucocorticoids is prevented by 11β-HSD2 which converts cortisol and corticosterone in 11-keto metabolites displaying poor affinity for MR. Along with its prominent role for mineralocorticoid selectivity, loss of function mutations of 11β-HSD2 gene46 or inhibition of 11β-HSD2 activity by liquorice47 induces sodium retention and hypertension. Because chenodeoxycholic acid is increased in cholestasis and dose-dependently inhibits 11β-HSD2 activity in CCD allowing activation of the MR by endogenous corticosterone,29, 48 this pathway was proposed as the initial stimulator of sodium retention during liver cirrhosis.32, 49, 50 However, present data in corticosteroid-clamped BDL mice demonstrate that inhibition of 11β-HSD2 activity is not essential for sodium retention in liver cirrhosis. First, activation of MR should stimulate Na+,K+-ATPase and induce over-expression of sgk1 and ENaC mRNAs all along the distal nephron and not specifically in the CCD. Second, the use of dexamethasone for glucocorticoid complementation prevented promiscuous activation of MR. Indeed, (1) because dexamethasone is approximately 100-fold more efficient than corticosterone for transactivating glucocorticoid receptors,51 it could be administered at concentration of the same order as aldosterone, and (2) at this low concentration, even in the absence of 11β-HSD2 activity, dexamethasone does not transactivate the MR because it is 1000-fold less efficient than aldosterone.51
In summary, bile duct ligation in mice induces cirrhosis which, within 4-5 weeks, leads to the induction of Na+,K+-ATPase in CCDs, to renal sodium retention and to ascites formation. Sodium retention, ascites formation and induction of Na+,K+-ATPase are independent of the activation of mineralocorticoid receptors by either aldosterone or glucocorticoids, supporting the “overflow theory” of sodium retention in liver cirrhosis in the cholestatic mice model.
We gratefully acknowledge Dr Leviel and Prof Bruneval (Hôpital Européen Georges Pompidou, Paris) for aldosterone radioimmunoassay and liver histology respectively.