Calcium binding by bile acids: in vitro studies

In this study, we compared in vitro calcium binding by the taurine and glycine conjugates of the major bile acids in human bile: cholic (CA), chenodeoxycholic (CDCA) and deoxycholic (DCA) acids, together with the cholelitholytic bile acids unodeoxycholic (UDCA) and unocholic (UCA) acids. At physiological total calcium ( C a m ) (1-15 mM) and bile acid (BA) (10-50 mM) concentrations, all the bile acids caused concentration-dependent falls in [Ca"], suggesting calcium binding. Except for glycine-conjugated CDCA, all the other calcium-bile acid cemplexes were soluble in 150 mM NaCl. The calcium binding affinities followed the pattern: dhydroxy (CDCA, UDCA and DCA) >trihydroxy (CA and UCA) bile acids, and glycine conjugates >taurine conjugates. The glycine conjugate of UDCA, which increases during UDCA treatment, had the highest calcium binding affinity. Ten-20 mM phospholipid modestly increased calcium binding by CA conjugates, but not by CDCA, UDCA, and DCA conjugates. Phospholipid also prevented the precipitation of glyco-CDCA in the presence of calcium. Bile acid-calcium binding was pH-independent over the range 6.5-8.5. The different calcium binding affinities of the major biliary bile acids may partly explain their varying effects on biliary calcium secretion. The results also suggest that neither precipitation of calcium-bile acid complexes nor impaired calcium binding by bile acids is important in the pathogenesis of human calcium gallstone formation. Gleeson, D., G. M. Murphy, and R. H. Dowling. Calcium binding by bile acids: in vitro studies using a calcium ion electrode. J. Lipid Res. 1990. 31: 781-791. Supplementary key w o r d s acid/phospholipid calcium binding calcium in electrode phospholipids biliary calcium bile The calcium salts, calcium bilirubinate, carbonate, and phosphate are the major components of noncholesterol gallstones (1, 2). They are also found at the center of cholesterol-rich stones (3) and may play an important role in nucleating cholesterol crystals from bile which is supersaturated with cholesterol (4). Calcium salts precipitate out of solution only when the concentration product of their constituent ions exceeds their solubility product. Therefore, precipitation of calcium salts in bile should depend, in part, on the biliary free ionized calcium concentration Bile acid-calcium binding in vitro has been the subject of several recent studies (4-8). However, there arc? still few ([C a'+]>. quantitative data on the comparative calcium binding affinities of the major bile acids in human bile. We embarked on such a study for several reasons. First, only 20-30% of calcium in human hepatic bile and 10-15% in human gallbladder bile is in the free ionized form, the remaining 70-90% being bound (9-11). The results of equilibrium dialysis studies (10) suggest that up to 80% of calcium binding in hepatic bde and 40% of that in gallbladder bile is accounted for by bile acid micelles. Bile acids have been proposed as major calcium buffers which, by lowering [Ca2+], minimize the risk of calcium precipitation in bile (4). The in vitro calcium binding properties of bile acids, therefore, have major implications for [Ca"] in bile. Second, there are two situations in which changes in biliary bile acid composition are associated with a tendency to calcium salt precipitation in bile. 9 Following terminal ileal resection in man, there is an increased risk of calcified gallstone formation (12) and bile is enriched with chenodeoxycholic acid (13) and with glycine-conjugated bile acids (14). it) During ursodeoxycholic acid (UDCA) treatment, bile becomes enriched with the glycine conjugates of UDCA (15) and acquired gallstone calcification is common (16-20). The mechanism for this calcification may be increased carbonate ion concentration following UDCA-induced bicarbonate secretion by the bile ductular epithelium (21, 22), but the role of induced changes in biliary [Ca2+] i s still unclear. In the initial description of UDCA-associated gallstone calcification, it was suggested that calcium combined with glycoursodeoxycholic acid and that the resultant salt, being poorly soluble, precipitated onto the surface of the stone (16). We now know, Abbreviations: CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; UDCA, unodeoxycholic acid; UCA, ursocholic acid; BA, bile acid; CMC, critical micellar concentration; E A , taurocholic acid; GCA, glycocholic acid; E D C A , taurochenodeoxycholic acid; TDCA, taurodeoxycholic acid. 'Present address: University Department of Gastroenterology, Royal Infirmary, Oxford Road, Manchester 13, England. *To whom requests should be addressed at: Gastroenterology Unit, 18th Floor, Guy's Tower, Guy's Hospital, London SE1 9RT, England. Journal of Lipid Research Volume 31, 1990 781 by gest, on O cber 8, 2017 w w w .j.org D ow nladed fom however, that calcium glycoursodeoxycholate is quite soluble in bile (23), suggesting that some other mechanism must be involved. If, for example, UDCA and its conjugates were poor calcium binding agents, the free ionized [Ca"] in bile would remain high, thereby risking calcium salt precipitation. Third, bile acids increase biliary calcium output in dogs (24) and in humans (22, 25, 26). Different bile acids increase biliary calcium output to differing degrees (24, 26) which may be due to differences in their calcium binding affinities. MATERIALS AND METHODS Preparation of bile acids and phospholipid solutions Sodium salts of the glycine (G) and taurine (T) conjugates of cholic (CA), chenodeoxycholic (CDCA), and deoxycholic (DCA) acids (approximately 98% pure) were obtained from Sigma Laboratories, Dorset, UK. The G and T conjugates of ursodeoxycholic acid (UDCA) were gifts from Roussel UCLAF, Paris, and the conjugates of ursocholic acid (UCA) were from Gipharmex SPA, Milan. The sodium salts of the bile acids were made up in concentrations ranging from 0 to 100 mM in 150 mM NaCl and 100 mM Na-1,4-piperazinediethanesulfonic acid (PIPES) buffered at pH 7.0. Solutions of CaC12 (2-20 mM and, in some experiments, 4-30 mM) in 150 mM NaCl were also prepared. At every bile acid concentration, 1 ml of each CaC1, solution was added to 1 ml of bile acid solution in glass vials to give final total calcium concentrations of 1-10 mM (and in some experiments 2-15 mM), and final bile acid (BA) concentrations of 0-50 mM, in 50 mM Na-PIPES and 150 mM NaCl (saline-PIPES). These calcium concentrations encompass the range found in human gallbladder and hepatic biles. All solutions were made up in duplicate. After the addition of CaCl,, the solutions were agitated gently for 30-60 min before measurement of free ionized calcium concentration [Ca"]. For the studies involving phospholipid, 2 g phosphatidylcholine dissolved in ethanol was obtained from Sigma Laboratories, Dorset, UK, and air-dried to a powder. This procedure was repeated twice after washing with deionized water. The powder was dispersed in 150 mM NaCl by sonication for 30 min which yielded a milky suspension, and the volume made up to 10 ml. Volumes of the suspension were then added to the bile acid solutions to give final concentrations of 50 mM BA and 0, 10, and 20 mM phospholipid, in saline-PIPES. After adding the phospholipid to the CA, CDCA, and DCA conjugates, the phospholipid dispersed immediately to give a clear solution. With G-UDCA and T-UDCA, however, dispersion of the phospholipid to a clear solution occurred only after gentle heating. Measurement of free ionized calcium: the calcium selective electrode The electrode membrane incorporated a highly calcium-selective neutral carrier ligand, ETH1001, dissolved in a polyvinyl chloride base, as previously described (27, 28). The electrode was filled with 10 mM CaC12 in saline-PIPES, the internal solution. The internal solution was connected via a silver chloride electrode to an AVO voltmeter. The circuit was completed by a calomel electrode containing concentrated KCI which was also in contact with the test solution. Preliminary experiments indicated that, over the range of [Ca] studied in both bile acid and saline solutions, the electrode exhibited near-ideal Nernstian behavior, that is, the recorded potential AE was closely and linearly proportional to the log of the calcium concentration. Correlation coefficients between recorded potential and [Ca"] by linear regression analysis routinely exceeded 0.998. Furthermore, a AE of between 26.5 and 28.5 mV was seen for each 10-fold rise in [Ca]. At [CaCl,] 1-10 mM, electrode readings were 11.0 f 0.3 mV lower in 150 mM NaCl than in H20, corresponding to a reduction in calcium ion activity (aCa2') of 61 f 2%. This is as expected, because the calcium ion activity coefficient, y, which relates (aCa2') to [Ca"], falls as total ionic strength increases. However, this fall seemed maximal at [NaCl] 150 mM, since the electrode readings were not affected by further increases in [NaCl] from 150 to 250 mM. Nor were the electrode readings affected by pH changes over the range 6.5-8.5, in unbuffered solutions. All studies were carried out at room temperature. Each day, a standard curve was constructed using CaC1, (1-10 mM) or, in some experiments 2-15 mM, in saline-PIPES, without bile acid (or CaC12 in150 mM NaCl for the "unbuffered" studies). AE was plotted against log [Ca] and the slope of the line was estimated graphically. Electrode readings were then taken in bile acid * phospholipid solutions; stable readings were attained with 2 min. Because of a small drift in baseline readings between repeated measurements in standard solutions ( < 3 mV in 1 day), each reading in bile acid solutions was bracketed between two readings in a standard solution (usually 1.0 mM CaCl,). After completing the readings in the bile acid solutions, standard curves were again constructed. The slopes were identical to those generated initially. At all [ C a m ] values, the electrode potentials were invariably lower in the bile acid solutions than in the corresponding standards, indicating lower calcium ion activity (aCa"). This was not due to precipitation of a calcium-bile acid complex because a) , the solutions, with the exception of calcium-GCDCA (see Results section), remained clear for up to 6 months, and b), after centrifugation (3000 g for 20 min), supernatant [ C a m ] values, measured by atomic absorption spectrophotometry, corresponded closely with the expected stoichiometric values. 782 Journal of Lipid Research Volume 31, 1990 by gest, on O cber 8, 2017 w w w .j.org D ow nladed fom Neither was the fall in (a Ca") in bile acid solutions likely to result from bile acid induced increases in total ionic strength, and subsequent falls in y. Fig. 1 illustrates calculated [Ca"] values (see formula below) at [ C a m ] 1-10 mM in 50 mM X A dissolved in (a) saline-PIPES and (b) 50 mM PIPES in which the sodium concentration had been adjusted to keep total ionic strength (1/2 C cz') constant. The differences in electrode readings, and therefore in calculated [Ca"] between (a) and (b), are very small indicating that y changes little in BA solutions over the range of calculated total ionic strength values encountered in these studies. We concluded that the depression of (a Ca") reflected a proportional fall in [Ca2'] consequent on BA binding of Ca in a soluble form. For all subsequent studies, therefore, y was assumed to be identical in BA and standard solutions and [Ca"] was calculated from the formula: