Lack of biliary lipid excretion in the little skate, Raja erinacea, indicates the absence of functional Mdr2, Abcg5, and Abcg8 transporters

Transcription

1 Am J Physiol Gastrointest Liver Physiol 286: G762 G768, First published December 30, 2003; /ajpgi Lack of biliary lipid excretion in the little skate, Raja erinacea, indicates the absence of functional Mdr2, Abcg5, and Abcg8 transporters Ronald P. J. Oude Elferink, 1,2 Roelof Ottenhoff, 2 Gert Fricker, 1,3 David J. Seward, 1,4 Nazzareno Ballatori, 1,4 and James Boyer 1,5 1 Mount Desert Island Biological Laboratory, Salsbury Cove, Maine 04672; 2 Academic Medical Center, Liver Center, 1105 BK Amsterdam, The Netherlands; 3 Institute for Pharmacological Technology and Biopharmac, Ruprecht-Karls University, Heidelberg, Germany; 4 Department of Environmental Medicine, University of Rochester School of Medicine, Rochester, New York 14642; and 5 Liver Center, Yale University School of Medicine, New Haven, Connecticut Submitted 26 September 2003; accepted in final form 25 December 2003 Oude Elferink, Ronald P. J., Roelof Ottenhoff, Gert Fricker, David J. Seward, Nazzareno Ballatori, and James Boyer. Lack of biliary lipid excretion in the little skate, Raja erinacea, indicates the absence of functional Mdr2, Abcg5, and Abcg8 transporters. Am J Physiol Gastrointest Liver Physiol 286: G762 G768, First published December 30, 2003; /ajpgi The ABC transporters bile salt export pump (BSEP; encoded by the ABCB11 gene), MDR3 P-glycoprotein (ABCB4), and sterolin 1 and 2 (ABCG5 and ABCG8) are crucial for the excretion of bile salt, phospholipid, and cholesterol, respectively, into the bile of mammals. The current paradigm is that phospholipid excretion mainly serves to protect membranes of the biliary tree against bile salt micelles. Bile salt composition and cytotoxicity, however, differ greatly between species. We investigated whether biliary phospholipid and cholesterol excretion occurs in a primitive species, the little skate, which almost exclusively excretes the sulphated bile alcohol scymnolsulphate. We observed no phospholipid and very little cholesterol excretion into bile of these animals. Conversely, when scymnolsulphate was added to the perfusate of isolated mouse liver perfusions, it was very well capable of driving biliary phospholipid and cholesterol excretion. Furthermore, in an erythrocyte cytolysis assay, scymnolsulphate was found to be at least as cytotoxic as taurocholate. These results demonstrate that the little skate does not have a system for the excretion of phospholipid and cholesterol and that both the MDR3 and the two half-transporter genes, ABCG5 and ABCG8, have evolved relatively late in evolution to mediate biliary lipid excretion. Little skate plasma membranes may be protected against bile salt micelles mainly by their high sphingomyelin content. mouse; bile; ABC transporters; phospholipid cholesterol MDR2 P-GLYCOPROTEIN IN THE mouse and MDR3 Pgp in humans mediate the translocation of phosphatidylcholine across the canalicular membrane of the hepatocyte, thereby making the phospholipid available for bile salt-induced secretion into primary bile (2). The exact mechanism by which this process occurs is unknown. These transporters are indispensable for biliary phospholipid secretion (23). Similarly, it was recently discovered that biliary cholesterol is critically dependent on the function of the two half transporters Abcg5 and Abcg8. Mice in which these two genes have been knocked out, have a 10-fold reduction in their biliary cholesterol secretion (27). Biliary lipid secretion has two important functions: first, it represents the main route of cholesterol elimination, and second, phospholipids and cholesterol protect cells along the biliary tree from the detergent action of bile salts. Absence of phospholipid secretion leads to extensive damage of hepatocytes and bile duct epithelial cells: Mdr2 / mice develop a liver disease characterized by proliferation of bile duct epithelial cells, portal inflammation, and fibrosis (16, 25). A similar but more severe disease is observed in pediatric patients with progressive familial intrahepatic cholestasis type 3, caused by mutations in the MDR3 gene, the human ortholog of rodent Mdr2. In 50% of these patients, the disease process progresses toward liver failure, which makes orthotopic liver transplantation necessary (7, 14). MDR3/Mdr2 is also thought to contribute to biliary elimination of hydrophobic organic anions or cations that bind to or partition into biliary vesicles or (21). Sequestration of hydrophobic compounds into these structures decreases the concentration of the monomeric forms in bile and thus stimulates further export from the cell. In support of this hypothesis, a recent study by Huang and Vore (13) reported that Mdr2 is required for biliary excretion of a tricarbocyanine dye containing two polar sulfonate groups and a quaternary ammonium group [indocyanine green (ICG)]. Although ICG is likely to be substrate for the anion transporter Mrp2, previous studies of ICG distribution in bile indicated that it is extensively (90 100%) associated with phospholipid vesicles and mixed lipid/bile salt micelles (21). Huang and Vore (13) observed that biliary excretion of ICG (0.4 mmol) was reduced by 90% in Mdr2 / mice relative to wild-type mice, whereas the biliary excretion of estradiol-17 -D-glucuronide [E(2)17G] was increased by 30% in Mdr2 / mice, indicating that the absence of Mdr2 differentially influences the biliary excretion of these organic anions and that the presence of phospholipid vesicles and mixed micelles in bile stimulate biliary excretion of ICG. From all these observations, a clear picture emerges in which Mdr2/MDR3 Pgp serves to protect hepatocytes and bile duct epithelial cells against bile salt-induced cytotoxicity. Because these genes are highly related to the MDR1 gene, which exist already in lower eukaryotes, it can be hypothesized that Mdr2 has evolved from the Mdr1 gene during evolution as a system of self defense against bile salts. To test this hypothesis, we studied phospholipid secretion in the little skate, a more primitive animal than mammals. The little skate (Raja erina- Address for reprint requests and other correspondence: R. P. J. Oude Elferink, AMC Liver Center, Academic Medical Center S1-162, Meibergdreef 69-71, 1105 BK Amsterdam ( r.p.oude-elferink@amc.uva.nl). The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. G /04 $5.00 Copyright 2004 the American Physiological Society

2 cea) is an elasmobranch that evolved 200 million years ago. Bile formation in this animal has been well studied in the past (1, 3, 8 10, 19). Various types of bile salts and bile alcohols exist in the animal kingdom, some of which are much less cytotoxic than the hydrophobic mammalian bile salts (12). The little skate secretes scymnolsulphate into bile, almost exclusively (15). This is a C27-sulphated bile alcohol, which has the same nucleus as cholate but bears a longer side chain with three hydroxyl groups. Because this bile salt is not found in mammalian bile, we investigated whether perfusing skate liver with mammalian bile salts [taurocholate (TC) and taurochenodeoxycholate (TCDC)] would elicit lipid secretion and, conversely, we studied the effect of scymnolsulphate on phospholipid secretion in mouse livers. These experiments demonstrated that the skate liver is neither capable of secreting phospholipid nor cholesterol, although scymnolsulphate is fully capable of driving excretion of both lipids in mouse livers that express Mdr2, Abcg5, and Abcg8. The absence of lipid excretion was surprising because we found that scymnolsulphate is at least as cytolytic as TC. MATERIALS AND METHODS Skate liver perfusion. Male little skates (Raja erinacea, kg body wt) were collected by net from Frenchman s Bay in Maine and maintained for up to 4 days in tanks equipped with flowing sea water (15 C) at the Mount Desert Island Biological Laboratory (Salsbury Cove, ME). Livers were removed from the skates and perfused in an erythrocyte-free, recirculating perfusion system at 15 C as previously described (20, 22). The perfusion medium consisted of well-oxygenated, heparinized elasmobranch Ringer solution containing 5 mm glucose and 5 mm HEPES/TRIS (ph 7.5). The bile duct was cannulated with a 17-cm segment of polyethylene tubing (PE-90). Because the proximal gallbladder and cystic duct are intrahepatic and cannot be ligated, the cystic duct was excluded by inserting a plug at the neck of the gallbladder through an incision at the gallbladder apex. The plug, which consisted of a plastic cap of an 18-G hypodermic needle covered with two layers of Parafilm, was secured into the gallbladder with sutures. Next, the collateral tributaries of the portal vein were ligated, and the portal vein was cannulated with a 2- to 3-cm segment of polyethylene tubing (PE-205) attached to an equal length of latex tubing. After the portal vein was cannulated, the liver was flushed with ml of heparinized elasmobranch Ringer solution. The liver was then excised and perfused at a rate of 30 ml/min, which produced a perfusion pressure of 2 4 cmh 2O, which is optimal for bile production and O 2 consumption in the isolated perfused skate liver (20). Liver weights of the animals were g(n 25). The first 150 ml perfusate were discarded after a single passage; subsequently, a recirculating perfusion was performed with a reservoir containing 100 ml of perfusion medium. The medium was continuously filtered and aerated with humidified air. The filter upstream from the perfusate reservoir consisted of a 200- m silk-screen mesh stretched over a small funnel, whereas the downstream filter was a Millipore filter holder containing a prefilter (AP ) and a 1.2- m filter (RAWP ). Bile was collected in 30-min intervals while a 0.5-ml sample was also taken from the perfusate. The perfusion was carried out for 7 h; after 2 h, 5 mol of the indicated bile salts were added to the recirculating medium every 30 min (10 mol/h). Bile volume was measured gravimetrically, assuming a density of one. ICG or dibromosulphoftalein (DBSP) were added after 1 h of perfusion at an initial concentration of 10 M (1 mol/liver). Bile and perfusate samples were collected every hour for 8 h. ICG concentrations were measured spectrophotometrically at 805 nm, after dilution of the samples in elasmobranch Ringer containing 0.25% bovine serum albumin. DBSP was measured spectrophotometrically G763 at 585 nm, after dilution with 0.1 M sodium pyrophosphate buffer (ph 8.2). cdna library construction and screening. RNA was isolated from skate liver (R. erinacea) as described (6) and used for construction of a cdna library in a ZAP expression vector (Stratagene). The library was screened under low stringency with two probes that corresponded to the killifish ABC region of the bile salt export pump (Bsep) as previously described (6). PCR was used to identify DNA fragments of appropriate size that were subsequently sequenced for identification of ABC transporter orthologues. Mouse liver perfusion. Male mice of FVB/N genetic background were bred in our own colony (Academic Medical Center, The Netherlands). Isolated liver perfusions were carried out in a recirculating fashion (exactly as described in Ref. 11). Bile samples were collected in 10-min intervals. Ten minutes after the start of the perfusion, the bile salt infusions were started and continuously infused in the perfusion medium at a rate of 600 nmol min g body wt 1. Bile salt, phospholipid, and cholesterol were determined using fluorescent enzymatic assays (as described in Ref. 17). In all experiments, bile flow was measured by weighing the bile samples, assuming a specific density of 1 g/ml. Purification and analysis of scymnolsulphate. Scymnolsulphate (Fig. 1) was purified from pooled gallbladder bile from spiny dogfish (Squalus acanthias; basically following the procedure described in Ref. 15). Briefly, lyophilized bile was dissolved in chloroform/methanol/acetic acid (33:15:1) and subjected to consecutive silica gel 60 chromatographies. The final purification was performed by reversephase adsorption chromatography using Serdolit PAD 1-resins (Serva Electrophoresis, Heidelberg, Germany). Desorption was achieved by washing the resins with methanol. By this procedure, 10 g lyophilized dogfish bile yielded 500 mg scymnolsulphate. The bile alcohol was characterized by negative ion first atom bombardment mass spectrometry (JEOL JMS-700, JEOL, Eching, Germany) and 13 C-nuclear magnetic resonance (360, MHz-spectrometer Aspect 3000; Bruker, Karlsruhe, Germany). All bile salts were purchased from Sigma Aldrich (St. Louis, MO). Bile salt-mediated cytolysis was performed Fig. 1. Structural formulas of taurocholate (TC; A) and scymnolsulphate (B).

3 G764 Table 1. Composition of skate bile compared with that of mouse bile exactly as described by Velardi et al. (26) using washed human erythrocytes. All our experiments on animals complied with guidelines of the institutional care and use committee. Statistical differences were analyzed by two-tailed Student s t-test with unequal variance. RESULTS Skate Mouse Bile flow, l h 1 g liver [Bile salt], mm [Phospholipid], M [Cholesterol], M Values are means SE. The lipid concentrations were determined in the first bile sample after cannulation. In the skate, this represents a 30-min sample, and in the mouse this is a 15-min sample. Bile salt concentration ([Bile salt]) is defined here as reaction equivalents in the enzymatic assay using 3-OH steroid dehydrogenase. [Phospholipid], phospholipid concentration; [cholesterol], cholesterol concentration. Endogenous bile secretion parameters in the little skate. Analysis of bile samples from the little skate (Table 1) revealed that these animals do not secrete significant amounts of phospholipid or cholesterol into bile. For comparison, the composition of mouse bile is given as well. Because bile flow rates in the skate are extremely low compared with the mouse, the values are given as biliary concentrations rather than as secretion rates. Using standards of known concentration (as determined by HPLC-mass spectrometry), we could show that scymnolsulphate is quantitatively measured in our enzymatic assay with 3-OH steroid dehydrogenase. Although bile salt concentrations are comparable in bile from skate and mouse, the amounts of cholesterol and phospholipid are several orders of magnitude lower in skate bile. Analysis of skate gallbladder bile by thin-layer chromatography confirmed the absence of phosphatidylcholine and aminophospholipids, phosphotidylserine, and phosphatidylethanolamine. A small quantity of sphingomyelin was observed. The question therefore arose whether the skate expresses a functional phospholipid secretion system. Fig. 2. Bile formation in little skate during infusion of TC or taurochenodeoxycholate (TCDC). Isolated skate livers were perfused as described. At 2 h after the start of the perfusion, bile salt was added to the perfusate (10 mol/h). A: bile flow; B: bile salt output; C: phospholipid (PL) output; D: cholesterol output. F, Control perfusion (no addition of bile salt to perfusate); E, addition of TCDC to the perfusate;, addition of TC to the perfusate; CHOL, cholesterol. Data are given as means SE and are derived from experiments with 7 animals in each group. Significant difference of the TC perfusion with the control perfusion (*P 0.05). Skate liver perfusion with TC and TCDC. To investigate this, isolated skate livers were perfused with TC and TCDC, two different mammalian bile salts (Fig. 2). These bile salts were added to the perfusate at 2 h after the start of the perfusion at a rate of 10 mol/h. As a control, perfusion was carried out without addition of bile salt. TC was efficiently secreted by perfused skate livers and maintained an increase in bile flow compared with livers that were not perfused with bile salt (Fig. 2A). At the highest secretion level, 60% of the administered TC was secreted into bile (Fig. 2B). Note that there is a 2- to 3-h lag before the administered bile salts appear from the end of the biliary cannula because of the slow bile flow rates. TCDC was much less well secreted and led to a reduction in bile flow toward the end of the experiment. Neither of the two mammalian bile salts stimulated significant phospholipid secretion (Fig. 2C), which did not exceed 200 pmol h 1 g liver 1. Hence the phospholipid-to-bile salt ratio in the little skate is 100-fold lower than in rodent liver. Cholesterol secretion was barely detectable in the control perfusions of skate livers (Fig. 2D). On perfusion with either TC or TCDC, the cholesterol concentration in skate bile tended to rise, but the values were variable and not significantly different from the control perfusion. In Fig. 3, the relationship between bile flow and biliary bile salt output is given for the skate (during TC perfusion) and for mice (during endogenous bile salt secretion). Please note that for skate, the data are expressed per hour and gram of liver, whereas for the mouse, the data are expressed per minute and 100 g body wt. When expressed as such, the relationships are quite similar. Because mice are 25 g and have a liver of 1 g, the bile salt output in mice is 15 times higher than in the skate. Irrespective of the different rates, the water volume generated per micromole of bile salt is virtually identical (12 and 16 l/ mol in skate and mouse, respectively, see equations in Fig. 3, A and B). When phospholipid output is plotted against bile salt output for both skate and mice (Fig. 3C), it becomes clear that there is no bile salt-driven phospholipid secretion in the skate. When the same is done for cholesterol (Fig. 3D), some cholesterol secretion is observed, but this is only slightly dependent on bile salt secretion. These data suggest that the skate does not have a

4 G765 Fig. 3. Relationship between biliary excretion parameters in skate and mouse. Relationship between bile salt output and bile flow in skate (A) and mouse (B). The formula in the figure describes the linear relationship between the 2 parameters. The bile salt-dependent fraction of bile flow is similar in skate and mice (12 and 16 l/ mol resp.). C: relationship between bile salt and PL output in skate and mice. D: relationship between bile salt and cholesterol output. E, skate; F, mice. Please note that because of the large difference in output rates between skate and mice, the output rates from skates are expressed as nmol h 1 g liver 1, whereas those from mice are expressed as nmol min g body wt. The phospolipid and bile salt output rates in individual samples from the skate liver perfusion with TC (n 7, Fig. 2) are compared with those from experiments with mice (n 3). F, mouse bile samples. E, skate bile samples. functional homolog of the mammalian Mdr2/MDR3 P-glycoprotein nor ABCG5/ABCG8, which is essential for biliary excretion of phospholipid and cholesterol, respectively. This possibility was confirmed by RT-PCR of skate mrna, which identified orthologs of Mdr1, Bsep, and Mrp2 (5, 6) but not for Mdr2, Abcg5, and Abcg8. Mouse liver perfusion with scymnolsulphate. To assess whether scymnolsulphate is capable of driving phospholipid secretion in the presence of Mdr2 Pgp, we purified scymnolsulphate from gallbladder bile of the spiny dogfish and perfused mouse livers with the purified bile alcohol. In the perfused mouse liver, this bile alcohol was readily secreted (Fig. 4B), indicating that the mouse ABC transporters, Bsep and possibly Mrp2, recognize and transport this substrate. Secretion of scymnolsulphate caused substantial phospholipid secretion (Fig. 4C), demonstrating that in the presence of the canalicular phospholipid translocator Mdr2, scymnolsulphate is entirely capable of stimulating phospholipid secretion. In addition, scymnolsulphate elicited significant cholesterol secretion (Fig. 4D). In Fig. 5, the relationship between bile salt Fig. 4. Biliary excretion of scymnolsulphate by isolated perfused mouse liver. The gallbladder of mice was cannulated, and the liver was perfused with tauroursodeoxycholate (TUDC; F) or with scymnolsulphate (E) at a rate of 600 nmol min g body wt. Infusion of bile salt into the perfusate was started after 10 min of perfusion. A: bile flow; B: bile salt output; C: PL output; D: cholesterol output. Data are presented as means SE (n 3 in both groups). Significant difference (*P 0.05) between scymnolsulphate and tauroursodeoxycholate perfusion.

5 G766 Fig. 5. Relationship of PL excretion (A) and cholesterol excretion (B) to bile salt excretion in the perfused mouse liver. The concentrations of bile salt, PL, and cholesterol in individual samples illustrated in Fig. 4 are plotted. Mouse livers were perfused with TUDC (F) or with scymnolsulphate (E). and phospholipid secretion was compared for scymnolsulphate and tauroursodeoxycholate in mice. Cytotoxicity of scymnolsulphate. We subsequently analyzed the cytotoxicity of scymnolsulphate by incubation of human erythrocytes with increasing concentrations of the conjugated bile alcohol and measured the extent of cell lysis. Several known mammalian bile salts were analyzed in parallel, so that a direct comparison of cytotoxicity could be made. Washed erythrocytes were incubated with the indicated concentrations of scymnolsulphate, TC, tauroursodeoxycholate, and taurodeoxycholate (TDC), and cell lysis was assessed by centrifugation and measurement of the hemoglobin concentration in the supernatant (Fig. 6). Quite surprisingly, it was found that scymnolsulphate induced erythrocyte lysis at lower concentrations than TC. Hence, scymnolsulphate is more cytolytic than TC. Because the little skate lives at considerably lower temperatures than 37 C, we also repeated the experiment at 15 C. At this lower temperature, the cytotoxicity was considerably lower, but the relative cytotoxicity of the various bile salts remained the same (data not shown). Biliary excretion of DBSP and ICG by perfused skate liver. Because Mdr2 is also thought to contribute to biliary exrection of hydrophobic organic compounds that bind to or partition into biliary vesicles or micelles [such as ICG (13)], additional studies measured biliary ICG excretion in the perfused skate liver and compared its excretion with that of another anionic dye, DBSP, which is a relatively hydrophilic substrate for the canalicular Mrp2 transport protein. ICG and DBSP were added to perfusate at the 1-h time interval, and 1 h later, 4% of the DBSP remained in the perfusate ( 0.4 M), whereas 10 15% of the ICG ( M) remained, indicating a slightly faster clearance of DBSP. Given the slow bile flow rate in the skate, these dyes did not appear in collected bile until the third collection interval (from 1 2 h after administering the dyes; Fig. 7). DBSP was excreted into bile more efficiently than ICG such that over this collection interval the amount of DBSP in bile was four times that of ICG (Fig. 7). However, both of these compounds were concentrated in bile, indicating active hepatobiliary transport. DBSP reached its maximum biliary concentration at the seventh collection interval and started declining by the eighth hour, whereas ICG concentration continued to rise until the end of the experiment (Fig. 7). DISCUSSION The present study analyzed the ability of skate liver to secrete phospholipid and cholesterol. We observed that under basal conditions, the little skate excretes hardly any phospholipid and cholesterol into bile. The skate virtually excrete only scymnolsulphate [3,7,12, 24,26,27-hexahydroxy-5 -cholestane-26(27)-sulphate] into bile (15), and it is not known whether this bile alcohol is capable of driving lipid excretion. We therefore administered mammalian bile salts, which drive lipid excretion in mammals, to the perfusate of perfused skate livers. TC was readily secreted by skate livers, as we have shown previously for 3 H-TC. (4). More recently, the skate homolog of the BSEP was cloned and characterized; this transporter has similar affinities for TC and TCDC as the rat Bsep (6). Interestingly, TCDC, when Fig. 6. Cytolysis of human erythrocytes by different bile salts including scymnolsulphate. Washed erythrocytes were incubated with the indicated concentration of TUDC (squares), TC (diamonds), scymnolsulphate (triangles), or TDC (circles). The cells were incubated with the bile salts at 37 C for 5 min and subsequently centrifuged. The amount of lysed cells was determined by measuring the concentration of hemoglobin in the supernatant. One hundred percent lysis was achieved by incubating the cells with 50 mm TDC. The data are representative for a set of 3 independent experiments. Fig. 7. Biliary excretion of dibromosulphoftalein (DBSP; n 4) and indocyanine green (ICG; n 6) in isolated perfused skate livers. ICG or DBSP was added after 1 h perfusion at an initial concentration of 10 M (1 mol/liver). Bile and perfusate samples were collected every hour for 8 h. Squares, DBSP excretion; triangles, ICG excretion.

6 administered to the perfusate, was very poorly excreted; the amount of bile salt in bile was not significantly different from that in livers without bile salt. In addition, bile flow in TCDCtreated liver tended to be lower. This may indicate that the hydrophobic TCDC either is not easily transported by skate liver or that it could induce cholestasis. The first possibility is not very likely because TCDC was found to compete with TC for transport by skate liver plasma membrane vesicles (1). Hence, it seems likely that TCDC is also transported. The excretion of TC was not accompanied by lipid excretion, suggesting that the skate does not have a phospholipid translocating system, similar to the rodent Mdr2 or the human MDR3 P-glycoprotein. The possibility exists that, due to their physical properties, conjugated bile alcohols such as scymnolsulphate are not able to extract phospholipid or cholesterol from the canalicular membrane. We therefore also investigated whether scymnolsulphate could drive biliary lipid secretion in mice. Infusion of scymnolsulphate in normal mice led to a strong excretion of the conjugated bile alcohol into bile. This again is not surprising, because it was recently shown that rat Bsep has a similar affinity for scymnolsulphate as skate Bsep (6). In mice, scymnolsulphate elicited lipid secretion of similar magnitude as TC. The scymnolsulphate-driven phospholipid and cholesterol secretion in perfused mouse livers demonstrate that this conjugated bile alcohol is quite capable of extracting lipids from the membranes of mice liver. This adds evidence to the suggestion that the skate does not have the capability of translocating phospholipid. The surprising finding was that scymnolsulphate was at least as cytolytic as TC. Mice have a bile salt pool that consists of 30 70% of TC, the remainder being tauromuricholate. The latter bile salt is much less cytolytic than TC and comparable with tauroursodeoxycholate. This creates the rather paradoxical situation that mice, which have a less cytolytic bile salt pool than skate, are susceptible to bile salt-induced liver damage if they do not secrete lipids (as is the case in the Mdr2 / model), whereas the skate does not seem to have this problem. Similar to the Mdr2 / mouse, the little skate does not excrete significant amounts of lipid. Several possible explanations exist. The skate might excrete other, as yet unidentified, compounds that interact with micelles of scymnolsulphate so as to reduce the phospholipid solubilizing capacity. We have not made an extensive search for phospholipids other than PC, PS, PE, and SM, by thin-layer chromatography. None of these phospholipids that are normally found were present in significant concentrations in skate bile, and only a minor quantity of sphingomyelin was detected. Alternatively, the outer leaflet of the skate canalicular membrane might be adapted in such a way that it can withstand the detergent effects of scymnolsuphate. It has been reported (24) that skate liver membranes have an exceptionally high anisotropy, compared with rat membranes, when each is measured at their respective body temperatures. This indicates that skate liver membranes are exceptionally rigid, which could result from a relatively high sphingomyelin content. If this analysis of liver plasma membranes also pertains to the canalicular membrane, it would confer a stronger resistance of this membrane toward the bile alcohol. We also found that cholesterol excretion into bile is very low in the skate, even on administration of TC or TCDC to the perfusate. Although it was thought for a long time that the excretion of these two lipids is highly coupled, several lines of evidence now indicate that this is not the case, at least in terms of their transport mechanisms. First, although cholesterol excretion in the Mdr2 / mouse is nearly absent, this can be increased by infusion of more hydrophobic bile salts such as TC or taurodeoxycholate or by feeding the animals a diet containing cholate so as to increase the percentage of hydrophobic bile salt (18). In this situation, cholesterol excretion is induced, but phospholipid excretion remains absent and the excretion of the two lipids can be uncoupled. Second, it has recently been discovered that cholesterol excretion into bile is critically dependent on the expression of the two half-transporter genes Abcg5 and Abcg8 (27). Knockout mice for these two genes have a 10-fold reduced cholesterol excretion, whereas phospholipid excretion is not affected. Conversely, mice that overexpress Abcg5 and Abcg8 in the liver have a marked increase in biliary cholesterol excretion, again without a big change in phospholipid excretion (28). Hence, although the excretion of both phospholipid and cholesterol is exquisitely dependent on bile salt excretion, the translocation/extrusion mechanisms seem to be mediated by separate transporter proteins. The skate does not excrete cholesterol to any significant extent, even under conditions that drive phospholipidindependent cholesterol excretion in the mouse. This suggests that the skate, an evolutionarily ancient vertebrate, may not only lack an Mdr2 homolog but also the functional homologs of Abcg5 and Abcg8. These important mammalian lipid transporters presumably developed later in evolution. Evidence has been presented in the past to suggest that hydrophobic drugs such as ICG are also transported via Mdr2 Pgp, because biliary ICG excretion is impaired in Mdr2 / mice. This phenomenon can, however, also be a secondary consequence of the absence of lipid in the bile of these animals, which acts as a sink for this hydrophobic drug. The presence of high concentrations of ICG in bile of the little skate, lacking an Mdr2 homolog, suggests that biliary elimination, at least for a large part, is achieved by active transporters other than Mdr2 Pgp. ACKNOWLEDGMENTS G767 We thank Catherine Bennet and Courtney Brooks for assistance with the liver perfusion studies. GRANTS This work was supported, in part, by United States Public Health Service Grants P30-ES-03828, P30-DK-34989, DK-25636, and DK REFERENCES 1. Ballatori N, Rebbeor JF, Connolly GC, Seward DJ, Lenth BE, Henson JH, Sundaram P, and Boyer JL. Bile salt excretion in skate liver is mediated by a functional analog of Bsep/Spgp, the bile salt export pump. Am J Physiol Gastrointest Liver Physiol 278: G57 G63, Borst P and Oude Elferink RPJ. Mammalian ABC transporters in health and disease. Annu Rev Biochem 71: , Boyer JL, Schwarz J, and Smith N. Biliary secretion in elasmobranchs. I. Bile collection and composition. Am J Physiol 230: , Boyer JL, Schwarz J, and Smith N. Biliary secretion in elasmobranchs. II. Hepatic uptake and biliary excretion of organic anions. Am J Physiol 230: , Cai SY, Soroka CJ, Ballatori N, and Boyer JL. Molecular characterization of a multidrug resistance-associated protein, Mrp2, from the little skate. Am J Physiol Regul Integr Comp Physiol 284: R125 R130, Cai SY, Wang L, Ballatori N, and Boyer JL. Bile salt export pump is highly conserved during vertebrate evolution and its expression is inhib-

7 G768 ited by PFIC type II mutations. Am J Physiol Gastrointest Liver Physiol 281: G316 G322, De Vree JM, Jacquemin E, Sturm E, Cresteil D, Bosma PJ, Aten J, Deleuze JF, Desrochers M, Burdelski M, Bernard O, Oude Elferink RP, and Hadchouel M. Mutations in the MDR3 gene cause progressive familial intrahepatic cholestasis. Proc Natl Acad Sci USA 95: , Fricker G, Dubost V, Finsterwald K, and Boyer JL. Characteristics of bile salt uptake into skate hepatocytes. Biochem J 299: , Fricker G, Hugentobler G, Meier PJ, Kurz G, and Boyer JL. Identification of a single sinusoidal bile salt uptake system in skate liver. Am J Physiol Gastrointest Liver Physiol 253: G816 G822, Fricker G, Wossner R, Drewe J, Fricker R, and Boyer JL. Enterohepatic circulation of scymnol sulfate in an elasmobranch, the little skate (Raja erinacea). Am J Physiol Gastrointest Liver Physiol 273: G1023 G1030, Frijters CM, Tuijn CJ, Ottenhoff R, Zegers BN, Groen AK, and Oude Elferink RPJ. The role of different P-glycoproteins in hepatobiliary secretion of fluorescently labeled short-chain phospholipids. J Lipid Res 40: , Hofmann AF, Schteingart CD, and Hagey LR. Species differences in bile acid metabolism. In: Bile Acids in Liver Disease, edited by Paumgartner G and Beuers U. Dordrecht The Netherlands: Kluwer Academic, 1995, p Huang L and Vore M. Multidrug resistance P-glycoprotein 2 is essential for the biliary excretion of indocyanine green. Drug Metab Dispos 29: , Jacquemin E, de Vree JML, Cresteil D, Sokal E, Sturm E, Dumont M, Burdelski M, Bosma PJ, Bernard O, Hadchouel M, and Oude Elferink RPJ. The wide spectrum of MDR3 deficiency in patients with progressive familial intrahepatic cholestasis type 3: from neonatal cholestasis to cirrhosis of adulthood. Gastroenterology 120: , Karlaganis G, Bradley SE, Boyer JL, Batta AK, Salen G, Egestad B, and Sjovall J. A bile alcohol sulphate as a major component in the bile of the small skate (Raja erinacea). J Lipid Res 30: , Mauad TH, van Nieuwkerk CMJ, Dingemans KP, Smit JJM, Schinkel AH, Notenboom RGE, van den Bergh Weerman MA, Verkruisen RP, Groen AK, Oude Elferink RPJ, Van der Valk MA, Borst P, and Offerhaus GJA. Mice with homozygous disruption of the mdr2 P- glycoprotein gene: a novel animal modle for studies of nonsuppurative inflammatory cholangitis and hepatocarcinogenesis. Am J Pathol 145: , Oude Elferink RP, Ottenhoff R, van Marle J, Frijters CM, Smith AJ, and Groen AK. Class III P-glycoproteins mediate the formation of lipoprotein X in the mouse. J Clin Invest 102: , Oude Elferink RPJ, Ottenhoff R, van Wijland M, Frijters CMG, van Nieuwkerk C, and Groen AK. Uncoupling of biliary phospholipid and cholesterol secretion in mice with reduced expression of mdr2 P-glycoprotein. J Lipid Res 37: , Reed JS, Smith ND, and Boyer JL. Determinants of biliary secretion in isolated perfused skate liver. Am J Physiol Gastrointest Liver Physiol 242: G319 G325, Reed JS, Smith ND, and Boyer JL. Hemodynamic effects on oxygen consumption and bile flow in isolated skate liver. Am J Physiol Gastrointest Liver Physiol 242: G313 G318, Scharschmidt BF and Schmid R. The micellar sink: a quantitative assessment of the association of organic anions with mixed micelles and other macromolecular aggregates in rat bile. J Clin Invest 62: , Simmons TW, Hinchman CA, and Ballatori N. Polarity of hepatic glutathione and glutathione S-conjugate efflux, and intraorgan mercapturic acid formation in the skate. Biochem Pharmacol 42: , Smit JJM, Schinkel AH, Oude Elferink RPJ, Groen AK, Wagenaar E, Van Deemter L, Mol CAAM, Ottenhoff R, Van der Lugt NMT, van Roon MA, Van der Valk MA, Offerhaus GJA, Berns AJM, and Borst P. Homozygous disruption of the murine mdr2 P-glycoprotein gene leads to a complete absence of phospholipid from bile and to liver disease. Cell 75: , Smith DJ and Ploch SA. Isolation of Raja erinacea basolateral liver plasma membranes: characterization of lipid composition and fluidity. J Exp Zool 258: , Van Nieuwkerk CM, Oude Elferink RP, Groen AK, Ottenhoff R, Tytgat GN, Dingemans KP, Van Den Bergh Weerman MA, and Offerhaus GJ. Effects of ursodeoxycholate and cholate feeding on liver disease in FVB mice with a disrupted mdr2 P-glycoprotein gene. Gastroenterology 111: , Velardi AL, Groen AK, Oude Elferink RP, van der Meer R, Palasciano G, and Tytgat GN. Cell type-dependent effect of phospholipid and cholesterol on bile salt cytotoxicity. Gastroenterology 101: , Yu L, Hammer RE, Li-Hawkins J, Von Bergmann K, Lutjohann D, Cohen JC, and Hobbs HH. Disruption of Abcg5 and Abcg8 in mice reveals their crucial role in biliary cholesterol secretion. Proc Natl Acad Sci USA 99: , Yu L, Li-Hawkins J, Hammer RE, Berge KE, Horton JD, Cohen JC, and Hobbs HH. Overexpression of ABCG5 and ABCG8 promotes biliary cholesterol secretion and reduces fractional absorption of dietary cholesterol. J Clin Invest 110: , 2002.