Biochimica et Biophysica Acta, 1091 (1991) 41-45 © 1991 Elsevier Science Publishers B.V. (Biomedical Division) 0167-4889/91/$03.50 ADONIS 016748899100072Z
41
BBAMCR 12833
The effect of bile salts on human vascular endothelial cells C.M. Garner, C.O. Mills, E. Elias and J.M. Neuberger Liver Research Laboratories, Queen Elizabeth Hospital, Edgbaston, Birmingham (U.K.)
(Received 21 May 1990)
Key words: Cytotoxicity; Bile salt; Vascular endothelial cell; (Human)
The uptake and release of radiochromium from adult human vascular endothelial cells in culture was employed to determine the relative toxicity of different bile salts. Endothelial cells after pre-incubation with SICr for 18 h were incubated with bile salts for 24 h and percentage chromium release was taken as a measure of toxicity to cells. Lithocholic acid (LC) (potassium salt) was cytotoxic at concentrations greater than 50/tM. However, LC glucuronide, sulfate and the fl-epimer were progressively less toxic with toxicity seen at concentrations of 60, 110 and 180 pM, respectively. The greatest cytotoxic effect was observed with glycolithocholic acid (GLC) (potassium salt) which was toxic at every concentration tested (20-200/tM). Sulfation abolished the toxic effect of GLC. At the concentrations employed for the assay (between 20 and 240/tM) GLC sulfate (disodium salt), taurolithocholic acid sulfate (disodium salt), cholic acid (sodium salt), glycoeholic acid (sodium salt), deoxycholic acid (sodium salt) and ursodeoxycholic acid (sodium salt) were not cytotoxic. The StCr release cytotoxicity assay was vr.lidated with lactate dehydrogenase leakage from endothelial cells with a good correlation (r = 0.87). These data confirm in a human cellular system that LC and its conjugates were the most toxic of the bile salts tested and explains its pathophysiological importance in hepatobiliary disease. It also suggests that biotransformation by either sulfation or fl-epimerisation of bile salts especially of LC, as occurs in patients with intrahepatic or extrahepatic biliary obstruction or severe cholestasis, is hepatoprotective.
Pntroduction
Sulfation a~d glucuronidation of bile salts are important in the pathophysiology of hepatobiliary disease with cholestasis. Sulfation has been reported to represent a biotransformation of considerable protective advantage [1], especially in man who cannot further hydroxylate lithocholate conjugates [2,3]. Sulfation of lithocholic acid (LC) may diminish or abolish its cytotoxicity although the efficiency of this cytoprotective pathway has been disputed by Carey [4], who demonstrated that sulfoglycolithocholate (GLC stdfate) (disodium salt) is not more soluble than its non-sulfated form and suggested that sulfated glycolithocholate (GLC sulfate) may be potentially cytotoxic. This was confirmed by Yousef et al. [5] who found sulfoglycolithocholate (GLC sulfate) to be cholestatic in rats.
Glucuronidation is also important in liver disease. Glucuronidation of LC makes it more polar and water soluble [6], and yet LC glucuronide was found to exert cholestatic effects comparable to those exerted by unconjugated LC [6]. Another monohydroxy bile acid, 3fl-hydroxycholenoic acid, present in serum and urine of patients with biliary atresia, is known to be tholestatic; sulfation does not prevent its cholestatic potential [7], although sulfation makes the compound more polar and water soluble. Because of these various discrepancies of LC, we have employed a cytotoxic assay based on a human endothelial cell culture model to determine the relative toxicity of various bile salts, especially LC, including the effect of glucuronidation and sulfation on their cytotoxicity. Materials and Methods
Preparation of vascular endothelial cells Abbreviations: LC, lithocholic acid; GLC, glycolithocholic acid; TLC, taurolithocholic acid; CA, cholic acid; GC, glycocholie acid; DC deoxycholic acid; UDC, ursodeoxycholic acid; CDC, chenodeoxycholic acid; LDH, lactate dehydrogenase; E199, Earle's 199 medium. Correspondence: C.M. Garner, Liver Research Laboratories, Queen Elizabeth Hospital, Edgbaston, Birmingham, B15 2TH, U.K.
Aorta or iliac vessels were removed from human organ donors and were used to produce vascular endothelial cells as described by Freshney [8]. The vessels were rinsed with Earle's 199 (E199) (Gibco, Paisley, U.K.) to remove blood and cell debris and incubated for 15 minutes with Worthington Type la collagenase
42 (Lorne Diagnostics, Bury St. Edmunds, U.K.) in phosphate buffered saline (1 mg/ml). Cells were established in primary culture using complete media which consisted of Earle's 199 (E199) (Gibco) with AB human serum 2070 (Blood Transfusion Service, Vincent Drive, Edgbaston, Birmingham, U.K.), preservative free heparin (50 units/ml) (CP Pharmaceuticals, Wrexham, U.K.), glutamine 2 mM (Flow Labs, Irvine, U.K.), benzylpenicillin (sodium) (24 #g/ml) (Glaxo Laboratories, U.K.), streptomycin sulphate (40 #g/ml) (Evans Medical, Horsham, U.K.) and endothelial cell growth factor (0.15 mg/ml) (produced from bovine brains using a modification of the method of Maciag [9]). The cells were plated onto 75 cm2 flasks pretreated with gelatin (170 w/v) for one h and were incubated at 37 °C in a 570 CO, humidified incubator. When the cells became confluent, they were subcultured using collagenase (1 mg/ml). In this experiment, cells were always used between passages 2 and 6. Immunofluorescence staining for Factor VIII related antigen showed 9970 endothelial cells.
Cytotoxicity assay Endothelial cells were plated onto 96-well flat-bottomed tissue culture plates (Gibco) precoated with 1% gelatin (Sigma Chemical Co, Dorset, U.K.), at a density of 1.104 cells/well and were left overnight to adhere at 37°C in a 570 CO2 humidified incubator. 51Cr (1.5/xCi) (Amersham International, Aylesbury, U.K.) was added per well and the plates were incubated for 18 h. After rinsing, the control wells were incubated with E199 and the test wells with E199 and bile salts. Triplicate wells were harvested immediately to assess the percentage chromiam uptake. After 24 h incubation, the cells were harvested by removing and retaining the supernatants. All wells were then rinsed again with E199 and treated with Triton X-100 170 (Sigma) for 5-10 nan in order to lyse the cells. After another Triton rinse, the supernatants and the lysates were separately counted (cpm) on a Muitigamma II gamma counter (LKB). Values obtained were used to calculate percentage uptake and release using the Triton supematants to obtain a maximum level of chromium release. The uptake of chromium over 18 h was between 15 and 2070 of the total chromium placed on the wells. Release of chromium from untreated wells (control release) after 24 h incubation in E199 was between 30 and 50% of the total chromium in each well. The standard deviation between the duplicate or triplicate wells of each concentration was normally in the range of 0.5-4.070. Lactate dehydrogenase assay Lactate dehydrogenase (LDH) activity in supernatants taken during the cytotoxicity assay was measured by a modification of the method of Bergmeyer and Bernt [10]. The supernatants were stored at - 20 ° C.
Supernatants (1 ml) were used in the assay and the concentration of buffer was modified accordingly.
Statistics Results are expressed as means ± S.E. Intra-assay variation was 2.5% and interassay variation was 13.7% (n = 21). Results Results are shown as the net percentage chromium release (i.e. the percentage gross chromium release after incubation with the test compound minus the percentage control release). Fig. 1A shows that LC (potassium salt) was not cytotoxic at concentrations up to 20 /~M, but had a cytotoxic effect at 50 ~tM or greater. However, as shown in Fig. 1B, glucuronidation of LC led to a slight decrease in cytotoxicity. It became cytotoxic at levels greater than 60 #M, but was not cytotoxic below this level. Sulfation of LC (Fig. 1C) further lowered the cytotoxicity to endothelial cells which was seen at levels above 90/tM. 3fl-Hydroxy-cholenoic acid, the sulfated fl-epimer of LC was less cytotoxic to endothelial cells than LC sulfate (Fig. 1D) and cytotoxicity was not seen at concentrations up to 110 #M but was seen at concentrations greater than 180 #M. Fig. 1E illustrates that glycination of LC increased the cytotoxicity at every concentration tested ranging from 20200 #M. However, sulfation of glycolithocholic acid (GLC) abolished the cytotoxic effect (cf Figs. 1E and 1F). It is also apparent from comparison of Figs. 1C and 1F that glycination of LC sulphate appeared to abolish its cytotoxicity to the endothelial cells at 110 ~tM. Fig. 1G illustrates that taurination of lithocholic acid sulfate to give taurolithocholic acid (TLC) sulfate also resulted in a decrease in cytotoxicity and was not toxic at any of the tested concentrations (20-210 #M). Dihydroxy bile salts appeared to be noncytotoxic. Figs. 1 H - K illustrate that cholic acid (CA), glycocholic acid (GC), deoxycholic acid (DC) and ursodeoxycholic acid (UDC) were not cytotoxic within the range 20-240 #M. Fig. 3a shows confluent endothelial cells grown in the complete culture medium and Fig. 3b illustrates the effect seen when the cells are cultured for 24 h in E199 with 40/~M glycolithocholic acid (GLC).
Comparison of chromium release with LDH activity The net percentage amount of LDH activity (i.e., the percentage gross activity after incubation with the test compound minus the percentage control activity) present in supernatants obtained from cells treated with LC sulfate, LC and 3fl-hydroxy cholenoic acid sulfate were compared to the net percentage chromium release (as before) obtained from the same wells using regression analysis. The correlation coefficient of LDH activity compared to chromium release was 0.87 + 2.5, P = 0.000017. These results are shown in Fig. 2.
43 Discussion
damage [11]. We applied this technique to assess the relative cytotoxicity of different bile salts on human vascular endothelial cells and obtained a good correlation with LDH release by the cells. LDH release is an
Radiochromium uptake and release from cells has been reported to be a good indicator of membrane B
A
C
LITHOCHOLIC ACID (K salt)
.°I
= |
%Or
°1
%Cr J Release 0
0
,i| 1
20
Net
20
50
70
Net
20
30
50
3B-HYDROXY CHOLEIVOIC ACID
0
110
130
140
160
40
70
90
110
130
150
7o
%Cr
0o
110
13o
1,~o los
200 220
Bile salt cone (tJM)
l/
20
Release 0
GLYCOLITHOCHOLIC ACID SULFATE
4O
Net
(Na 2 salt)
20
%Cr Release
0
I |ll
- 20
180 200 220
G
20
40
60
80 100 120 140 160 Bile salt conc (IJM)
180 200
20
40
50
70
90
110 130
140
1,50 180
Bile salt conc (pM)
H
I
4st
TAUROLITHOCHOLIC ACID SULPHATE
DEOXYCHOLIC ACID (Na salt)
40
URSODEOXYCHOLIC ACID (Na salt)
(Na 2salt)
20
Net
201
Net
g === | g
0
Release 0 U
Release O
=
-20,, ~ ,
-20
-20 20
40
60
80
100
130
1,50 170
190
210
20
%Cr
%Cr
%Or Release
40
20
GLYCOLITHOCHOLIC ACID (K salt)
Bile salt c o n c (IJM)
r
20
,50
7'0
100
120
140
,70
190
220 240
2o
, so
70
g
n =
I
1oo 12o 14o
,70
190 220 240
Bile salt cone (IJM)
Bile salt c o n c (luM)
Bile salt c o n c (pM)
K
J
GLYCOCHOLIC ACID (Na salt)
CHOLIC ACID (Na salt)
4O
Net
100
- 20
20
Net
B0
Ne,
-20 l
4O
60
40
i
n
I
F
60
SULFATE (Na salt)
%Cr
I
Bile salt conc (IJM)
E
20
Release
O
-2 0
100 ,~)O 140 170 100 190 220 240
D
°l
%Cr Release
"7
Bile salt c o n c (IJM)
4O
(Na salt)
20
Net
-20
-20
LITHOCHOLIC ACID SULFATE
4O
(Na salt)
Net
Release
LITHOCHOLIC ACID GLUCURONIDE
r '°) i
20
%Cr
Net
20
%Cr
g
I I .........am =o
,50
.o
90
.o
.o
,is
Bile salt c o n e (IJM)
,is
200 230
~'o
4'0
8'o
.'o
,so
,2o
,;o
,is
,:o
2o0
Bile salt c o n e (gM)
Fig. 1. Chromium release for different bile salts. Human adult vascular endothelial cells were cultured with Eades 199 complete media and incubated with SlCr (1.5 pCi) for 18 h at 37 °C in a 5% CO2 humidified incubator. After rinsing, bile salts (20-200/~M) were added and then incubated for a further 24 h at 37°C in a 5% CO2 humidified incubator. The plates were rinsed and treated with Triton X-100 (1%) to lyse the cells. Chromium release was measured using a gamma counter (LKB). The net percentage 5ZCr release (i.e., the percentage gross chromium release after incubation with the test compound minus the percentage control release) was then calculated. The results are expressed as means+ S.E. (A), lithocholic acid (potassium salt); (B), lithocholic acid glucuronide (sodium salt); (C), ]ithocholic acid sulfate (sodium salt); (D), 3-fl-hydroxy choler~oic acid sulfate (sodium salt); (E), glycolithocholic acid (potassium salt); (F), glycolithocholic acid sulfate (sodium salt); (G) taurolithochollc acid sulfate (sodium salt); (H), dcoxycholic acid (sodium salt); (I), ursodeoxycholic acid (sodium salt); 0), cholic acid (sodium salt); (K), glycocholic acid (sodium salt).
44 10
° ~
o
o "10 e" (D '10 ¢¢
..t -10 L
-lO
•
o
lo 20 Chromium-51 release
3o
4o
Fig. 2. Correlation of SlCr release with lactate dehydrogenase activity. The net percentage lactate dehydrogenase activity (LDH) determined by the method of Bergmeyer [10] and net percentage S;Cr release were measured from supematants taken from adult human vascular endothelial cells treated with LC sulfate (175 ~=M, 220/~M), LC (50/~M, 75 ~M, 170 ~M) and 3/~-hydroxycholenoic acid sulfate (45/~M, 90 #M, 120/~M, 175 ~aM, 220/~M). Regression analysis was used to compute the correlation coefficient of the LDH and 5tCr release. The correlation coefficient was 0.87 ± 2.5, P = 0.000017. The net ~ LDH activity is the ~ gross LDH activity after incubation with the test compound minus the ~ activity in the control wells. The net • SlCr release is t.~:e gross SICr release after incubation with the test compound minus the ~ 51Cr release in the control wells.
established marker of cytotoxicity in terms of membrane damage. The data show that percentage radiochromium release reflects cytotoxicity of trihydroxy, dihydroxy and monohydroxy bile salts. Both trihydroxy and dihydroxy bile salts, either unconjugated or conjugated with glycine, showed no cytotoxicity at the concentrations employed, which were below the critical micellar concentration cf the bile salts. Indeed, these bile salts had a 'protective' effect on endothelial cells relative to controls, suggesting that they probably maintain endothelial cells in culture. It seemed that CA, GC and UDC had a greater protective effect than DC. The greater protective effect of UDC relative to DC may be attributed to the orientation of the hydroxyl group to form the fl-epimer (3a, 7fl) which decreases the detergency of a bile salt and therefore has the effect of lowering the interaction of the bile salt to the membrane of the cell. Because of its lack of cytotoxicity UDC has been employed in the treatment of patients with primary biliary cirrhosis [12]. At the concentrations employed we did not observe chenodeoxycholate (CDC) to be cytotoxic; however it was found to be toxic when assessed on patients who develop evidence of liver injury while ingesting CDC for cholesterol gall stones
~scular endothelial cells. Phase contrast photomicrographs (magnification x 120) of: (a) Control, confluent cells incubated at 37°C in a 5~ CO2 humidified incubator in Earle's 199 complete media. (b) Cells incubated at 37 o C in a 5~ CO2 humidified incubator in Earle's 199 and treated for 24 h with 40/~M GLC (potassium salt).
45 dissolution [13]. In contrast, our observations plus others [14,15] suggest that UDC is not cytotoxic. The radiochromium model of looking at cytotoxicity could not differentiate clearly between the cytotoxic effect of UDC and GA at the concentrations employed. However, there might be a possib!e differentiation of the relative cytotoxicity if higher concentrations were used. At the concentrations employed, LC and its esters had a far greater effect than dihydroxy and trihydroxy bile acids. This may be associated with the relative insolubility of LC and its derivatives compared with polyhydroxylated bile salts. For example, CA (3~t, 7a, 12a) is many-fold ( > 4000) more soluble than the monohydroxy bile acid, LC (3a) [16]. Glucuronidation of LC increases polarity and solubility [6], and produces cholestasis comparable to that exerted by unconjugated LC in vivo. Intravenous administration of LC glucuronide, resulted in partial or complete cholestasis [6] in the rat at concentrations as low or lower than those at which lithocholate produced cholestasis [1]. Sulfation of LC also decreased its toxicity (Figs. 1A and C), which may be explained in part by the 3-fold increase in solubility of sulfolithocholic acid compared with LC [6]. However, there is no direct relationship between the solubility and cytotoxicity of the various conjugates of LC. Thus, although LC glucuronide is more soluble (7-fold) than LC sulfate [6], the threshold dose for cytotoxicity in our assay was lower for LC glucuronide than for LC sulfate. In addition glycinated trihydroxy bile acids were non-toxic (Fig. 1J), but glycination of [C increased toxicity several-fold. Therefore the increased toxic effect of GLC may be explained, in part, in terms of its interaction with the membrane [17]. Sulfation of GLC to give sulfolithoglycine abolished the toxic effect of GLC (cf Figs. 1E and F). One surprising finding in our assays was that test cytotoxicity for certain bile acids was consistently less than controls. For example, GLC sulfate and TLC sulfate had a protective effect (Figs. 1F and G) and fl-epimefization of LC and unsaturation at the 5' position to give 3fl-hydroxycholenoic acid sulfate diminished the cytotoxicity of LC sulfate (of Figs. 1C and D). This re-iterates the importance of fl-epimerisation in rendering bile salts, especially the monohydroxy bile salts, more soluble [6] and confi~'ms studies by Kimura et al. [18] and SchSlmerich et al. [19], that the fl-position of the 7-hydroxy group reduces toxicity [18]. In conclusion, sulfation and glucuronidation of bile salts such as LC decreased or abolished their toxicity to endothelial cells and suggests that these two metabolic pathways are important in hepatobiliary disease with chronic cholestasis where sulfation and glucuronidation increase with progression of the disease. Furthermore, fl-epimerisation and unsaturation of the bond at C-5 to form 3 fl-hydroxycholenoic acid also
decreased toxicity and therefore like sulfation and glucuronidation this metabolic process may be important in biliary obstruction especially in patients with biliary atresia who excrete large amounts of 3-fl-hydroxycholenoic acid in urine and serum [20]. The lack of cytotoxic effect observed with trihydroxy bile salts and the fl-hydroxy bile salt (UDC) illustrated that orientation and the number of hydroxyl groups on bile salts decreases or progressively inhibits their toxic potential. The good correlation obtained when the radiochromium assay was compared with LDH release suggested that the former is also a measure of membrane damage and that the least hydroxylated bile salts (monohydroxy bile salts) induced the greatest membrane damage to endothelial cells. Acknowledgments This work was supported by the Jules Thorn Charitable Trust. The authors would like to thank the surgical teams headed by Mr. P. McMaster and Mr. J. Buckels for obtaining the tissue samples. References 1 Javitt, N.B. and Emerman, S. (1968) J. Clin. Invest. 47, 1002-1014. 2 Kuss, E., Fernbacher, S. and Sies, H. (1973) Acta Endocrinol. (Suppl.) 173, 1. 3 Cowen, A.E., Korman, M.G., Hofmann, A.F., Cass, O.W. and Coffin, S.B. (1975) Gastroenterology 69, 67-76. 4 Carey, M.C., Wu, S.F.T. and Watkins, J.B. (1979) Biochim. Biophys. Acta. 575, 16-26. 5 Yousef, I.M., Tuchweber, B., Vonk, R.J., Masse, D., Audet, M. and Roy, C.C. (1981) Gastroenterology 80, 233-241. 60elberg, D.(3., Chari, M.V., Little, J.M., Adcock, E.W. and Lester, R. (1984) J. Clin. lnvest. 73, 1507-1514. 7 Mathis, U., Karlaganis, (3. and Preisig, R. (1983)Gastroenterology 85, 674-681. 8 Freshney, R.I. (1987) Culture of Animal Ceils. A Manual of Basic Technique, Ch. 20, pp. 257-288, Alan R. Liss, Ne~ York. 9 Maeiag, T., Cerundolo, J., llsley, S., Kelley, P.R. and Forand, R. (1979) Proc. Natl. Acad. Sci. USA. 76 (11), 5674-5678. 10 Bergmeyer, H.U. (1974) Methods Enzym. Anal. 2, 574-579. 11 Zawydiwski, R. and Duncan, G.R. (1978) In Vitro 14 (8), 707-713. 12 Poupon, R., Poupon, R.E., Calmus, Y., Chrieten, Y., Ballet, F. and Darnis, F. (1987) Lancet i, 834-835. 13 Fisher, R.L., Anderson, D.W., Boyer, J.L., Ishak, K., Klatskin, (3., Lachin, J.M. and Phillips, M.J. (1982) Hepatology 2 (2), 187-201. 14 Tint, G.S., Salen, G., Colalillo, A., Graber, D., Verga, D., Speck, J. and Shefer, S. (1982) Ann. Intern. Med. 97 (3), 351-356. 15 Bachrach, W.H. and Hofmann, A.F. (1982) Dig. Dis. Sci. 27 (9), 833-856. 16 Roda, A. and Fini, A. (1984) Hepatology 4 (5), 728-768. 17 Rudman, D. and Kendall, F.E. (1957) J. Clin. Invest. 36, 538-542. 18 Kimura, T., Shimamura, M. and Yamaguchi, A. (1981) Acta Hepat. Japan 22, 1-7. 19 Schtilmerich, J., Becher, M-S., Schmidt, K... Schubert, R., Kremer, B., Feldhaus, S. and Gerok, W. (1984) Hepatology 4 (4), 661-666. 20 Makino, I., Sjovali, J., Norman, A. and Strandvik, B. (1977) Fed. Eur. Biochem. Soc. Lett. 15-1,161-4.
41
BBAMCR 12833
The effect of bile salts on human vascular endothelial cells C.M. Garner, C.O. Mills, E. Elias and J.M. Neuberger Liver Research Laboratories, Queen Elizabeth Hospital, Edgbaston, Birmingham (U.K.)
(Received 21 May 1990)
Key words: Cytotoxicity; Bile salt; Vascular endothelial cell; (Human)
The uptake and release of radiochromium from adult human vascular endothelial cells in culture was employed to determine the relative toxicity of different bile salts. Endothelial cells after pre-incubation with SICr for 18 h were incubated with bile salts for 24 h and percentage chromium release was taken as a measure of toxicity to cells. Lithocholic acid (LC) (potassium salt) was cytotoxic at concentrations greater than 50/tM. However, LC glucuronide, sulfate and the fl-epimer were progressively less toxic with toxicity seen at concentrations of 60, 110 and 180 pM, respectively. The greatest cytotoxic effect was observed with glycolithocholic acid (GLC) (potassium salt) which was toxic at every concentration tested (20-200/tM). Sulfation abolished the toxic effect of GLC. At the concentrations employed for the assay (between 20 and 240/tM) GLC sulfate (disodium salt), taurolithocholic acid sulfate (disodium salt), cholic acid (sodium salt), glycoeholic acid (sodium salt), deoxycholic acid (sodium salt) and ursodeoxycholic acid (sodium salt) were not cytotoxic. The StCr release cytotoxicity assay was vr.lidated with lactate dehydrogenase leakage from endothelial cells with a good correlation (r = 0.87). These data confirm in a human cellular system that LC and its conjugates were the most toxic of the bile salts tested and explains its pathophysiological importance in hepatobiliary disease. It also suggests that biotransformation by either sulfation or fl-epimerisation of bile salts especially of LC, as occurs in patients with intrahepatic or extrahepatic biliary obstruction or severe cholestasis, is hepatoprotective.
Pntroduction
Sulfation a~d glucuronidation of bile salts are important in the pathophysiology of hepatobiliary disease with cholestasis. Sulfation has been reported to represent a biotransformation of considerable protective advantage [1], especially in man who cannot further hydroxylate lithocholate conjugates [2,3]. Sulfation of lithocholic acid (LC) may diminish or abolish its cytotoxicity although the efficiency of this cytoprotective pathway has been disputed by Carey [4], who demonstrated that sulfoglycolithocholate (GLC stdfate) (disodium salt) is not more soluble than its non-sulfated form and suggested that sulfated glycolithocholate (GLC sulfate) may be potentially cytotoxic. This was confirmed by Yousef et al. [5] who found sulfoglycolithocholate (GLC sulfate) to be cholestatic in rats.
Glucuronidation is also important in liver disease. Glucuronidation of LC makes it more polar and water soluble [6], and yet LC glucuronide was found to exert cholestatic effects comparable to those exerted by unconjugated LC [6]. Another monohydroxy bile acid, 3fl-hydroxycholenoic acid, present in serum and urine of patients with biliary atresia, is known to be tholestatic; sulfation does not prevent its cholestatic potential [7], although sulfation makes the compound more polar and water soluble. Because of these various discrepancies of LC, we have employed a cytotoxic assay based on a human endothelial cell culture model to determine the relative toxicity of various bile salts, especially LC, including the effect of glucuronidation and sulfation on their cytotoxicity. Materials and Methods
Preparation of vascular endothelial cells Abbreviations: LC, lithocholic acid; GLC, glycolithocholic acid; TLC, taurolithocholic acid; CA, cholic acid; GC, glycocholie acid; DC deoxycholic acid; UDC, ursodeoxycholic acid; CDC, chenodeoxycholic acid; LDH, lactate dehydrogenase; E199, Earle's 199 medium. Correspondence: C.M. Garner, Liver Research Laboratories, Queen Elizabeth Hospital, Edgbaston, Birmingham, B15 2TH, U.K.
Aorta or iliac vessels were removed from human organ donors and were used to produce vascular endothelial cells as described by Freshney [8]. The vessels were rinsed with Earle's 199 (E199) (Gibco, Paisley, U.K.) to remove blood and cell debris and incubated for 15 minutes with Worthington Type la collagenase
42 (Lorne Diagnostics, Bury St. Edmunds, U.K.) in phosphate buffered saline (1 mg/ml). Cells were established in primary culture using complete media which consisted of Earle's 199 (E199) (Gibco) with AB human serum 2070 (Blood Transfusion Service, Vincent Drive, Edgbaston, Birmingham, U.K.), preservative free heparin (50 units/ml) (CP Pharmaceuticals, Wrexham, U.K.), glutamine 2 mM (Flow Labs, Irvine, U.K.), benzylpenicillin (sodium) (24 #g/ml) (Glaxo Laboratories, U.K.), streptomycin sulphate (40 #g/ml) (Evans Medical, Horsham, U.K.) and endothelial cell growth factor (0.15 mg/ml) (produced from bovine brains using a modification of the method of Maciag [9]). The cells were plated onto 75 cm2 flasks pretreated with gelatin (170 w/v) for one h and were incubated at 37 °C in a 570 CO, humidified incubator. When the cells became confluent, they were subcultured using collagenase (1 mg/ml). In this experiment, cells were always used between passages 2 and 6. Immunofluorescence staining for Factor VIII related antigen showed 9970 endothelial cells.
Cytotoxicity assay Endothelial cells were plated onto 96-well flat-bottomed tissue culture plates (Gibco) precoated with 1% gelatin (Sigma Chemical Co, Dorset, U.K.), at a density of 1.104 cells/well and were left overnight to adhere at 37°C in a 570 CO2 humidified incubator. 51Cr (1.5/xCi) (Amersham International, Aylesbury, U.K.) was added per well and the plates were incubated for 18 h. After rinsing, the control wells were incubated with E199 and the test wells with E199 and bile salts. Triplicate wells were harvested immediately to assess the percentage chromiam uptake. After 24 h incubation, the cells were harvested by removing and retaining the supernatants. All wells were then rinsed again with E199 and treated with Triton X-100 170 (Sigma) for 5-10 nan in order to lyse the cells. After another Triton rinse, the supernatants and the lysates were separately counted (cpm) on a Muitigamma II gamma counter (LKB). Values obtained were used to calculate percentage uptake and release using the Triton supematants to obtain a maximum level of chromium release. The uptake of chromium over 18 h was between 15 and 2070 of the total chromium placed on the wells. Release of chromium from untreated wells (control release) after 24 h incubation in E199 was between 30 and 50% of the total chromium in each well. The standard deviation between the duplicate or triplicate wells of each concentration was normally in the range of 0.5-4.070. Lactate dehydrogenase assay Lactate dehydrogenase (LDH) activity in supernatants taken during the cytotoxicity assay was measured by a modification of the method of Bergmeyer and Bernt [10]. The supernatants were stored at - 20 ° C.
Supernatants (1 ml) were used in the assay and the concentration of buffer was modified accordingly.
Statistics Results are expressed as means ± S.E. Intra-assay variation was 2.5% and interassay variation was 13.7% (n = 21). Results Results are shown as the net percentage chromium release (i.e. the percentage gross chromium release after incubation with the test compound minus the percentage control release). Fig. 1A shows that LC (potassium salt) was not cytotoxic at concentrations up to 20 /~M, but had a cytotoxic effect at 50 ~tM or greater. However, as shown in Fig. 1B, glucuronidation of LC led to a slight decrease in cytotoxicity. It became cytotoxic at levels greater than 60 #M, but was not cytotoxic below this level. Sulfation of LC (Fig. 1C) further lowered the cytotoxicity to endothelial cells which was seen at levels above 90/tM. 3fl-Hydroxy-cholenoic acid, the sulfated fl-epimer of LC was less cytotoxic to endothelial cells than LC sulfate (Fig. 1D) and cytotoxicity was not seen at concentrations up to 110 #M but was seen at concentrations greater than 180 #M. Fig. 1E illustrates that glycination of LC increased the cytotoxicity at every concentration tested ranging from 20200 #M. However, sulfation of glycolithocholic acid (GLC) abolished the cytotoxic effect (cf Figs. 1E and 1F). It is also apparent from comparison of Figs. 1C and 1F that glycination of LC sulphate appeared to abolish its cytotoxicity to the endothelial cells at 110 ~tM. Fig. 1G illustrates that taurination of lithocholic acid sulfate to give taurolithocholic acid (TLC) sulfate also resulted in a decrease in cytotoxicity and was not toxic at any of the tested concentrations (20-210 #M). Dihydroxy bile salts appeared to be noncytotoxic. Figs. 1 H - K illustrate that cholic acid (CA), glycocholic acid (GC), deoxycholic acid (DC) and ursodeoxycholic acid (UDC) were not cytotoxic within the range 20-240 #M. Fig. 3a shows confluent endothelial cells grown in the complete culture medium and Fig. 3b illustrates the effect seen when the cells are cultured for 24 h in E199 with 40/~M glycolithocholic acid (GLC).
Comparison of chromium release with LDH activity The net percentage amount of LDH activity (i.e., the percentage gross activity after incubation with the test compound minus the percentage control activity) present in supernatants obtained from cells treated with LC sulfate, LC and 3fl-hydroxy cholenoic acid sulfate were compared to the net percentage chromium release (as before) obtained from the same wells using regression analysis. The correlation coefficient of LDH activity compared to chromium release was 0.87 + 2.5, P = 0.000017. These results are shown in Fig. 2.
43 Discussion
damage [11]. We applied this technique to assess the relative cytotoxicity of different bile salts on human vascular endothelial cells and obtained a good correlation with LDH release by the cells. LDH release is an
Radiochromium uptake and release from cells has been reported to be a good indicator of membrane B
A
C
LITHOCHOLIC ACID (K salt)
.°I
= |
%Or
°1
%Cr J Release 0
0
,i| 1
20
Net
20
50
70
Net
20
30
50
3B-HYDROXY CHOLEIVOIC ACID
0
110
130
140
160
40
70
90
110
130
150
7o
%Cr
0o
110
13o
1,~o los
200 220
Bile salt cone (tJM)
l/
20
Release 0
GLYCOLITHOCHOLIC ACID SULFATE
4O
Net
(Na 2 salt)
20
%Cr Release
0
I |ll
- 20
180 200 220
G
20
40
60
80 100 120 140 160 Bile salt conc (IJM)
180 200
20
40
50
70
90
110 130
140
1,50 180
Bile salt conc (pM)
H
I
4st
TAUROLITHOCHOLIC ACID SULPHATE
DEOXYCHOLIC ACID (Na salt)
40
URSODEOXYCHOLIC ACID (Na salt)
(Na 2salt)
20
Net
201
Net
g === | g
0
Release 0 U
Release O
=
-20,, ~ ,
-20
-20 20
40
60
80
100
130
1,50 170
190
210
20
%Cr
%Cr
%Or Release
40
20
GLYCOLITHOCHOLIC ACID (K salt)
Bile salt c o n c (IJM)
r
20
,50
7'0
100
120
140
,70
190
220 240
2o
, so
70
g
n =
I
1oo 12o 14o
,70
190 220 240
Bile salt cone (IJM)
Bile salt c o n c (luM)
Bile salt c o n c (pM)
K
J
GLYCOCHOLIC ACID (Na salt)
CHOLIC ACID (Na salt)
4O
Net
100
- 20
20
Net
B0
Ne,
-20 l
4O
60
40
i
n
I
F
60
SULFATE (Na salt)
%Cr
I
Bile salt conc (IJM)
E
20
Release
O
-2 0
100 ,~)O 140 170 100 190 220 240
D
°l
%Cr Release
"7
Bile salt c o n c (IJM)
4O
(Na salt)
20
Net
-20
-20
LITHOCHOLIC ACID SULFATE
4O
(Na salt)
Net
Release
LITHOCHOLIC ACID GLUCURONIDE
r '°) i
20
%Cr
Net
20
%Cr
g
I I .........am =o
,50
.o
90
.o
.o
,is
Bile salt c o n e (IJM)
,is
200 230
~'o
4'0
8'o
.'o
,so
,2o
,;o
,is
,:o
2o0
Bile salt c o n e (gM)
Fig. 1. Chromium release for different bile salts. Human adult vascular endothelial cells were cultured with Eades 199 complete media and incubated with SlCr (1.5 pCi) for 18 h at 37 °C in a 5% CO2 humidified incubator. After rinsing, bile salts (20-200/~M) were added and then incubated for a further 24 h at 37°C in a 5% CO2 humidified incubator. The plates were rinsed and treated with Triton X-100 (1%) to lyse the cells. Chromium release was measured using a gamma counter (LKB). The net percentage 5ZCr release (i.e., the percentage gross chromium release after incubation with the test compound minus the percentage control release) was then calculated. The results are expressed as means+ S.E. (A), lithocholic acid (potassium salt); (B), lithocholic acid glucuronide (sodium salt); (C), ]ithocholic acid sulfate (sodium salt); (D), 3-fl-hydroxy choler~oic acid sulfate (sodium salt); (E), glycolithocholic acid (potassium salt); (F), glycolithocholic acid sulfate (sodium salt); (G) taurolithochollc acid sulfate (sodium salt); (H), dcoxycholic acid (sodium salt); (I), ursodeoxycholic acid (sodium salt); 0), cholic acid (sodium salt); (K), glycocholic acid (sodium salt).
44 10
° ~
o
o "10 e" (D '10 ¢¢
..t -10 L
-lO
•
o
lo 20 Chromium-51 release
3o
4o
Fig. 2. Correlation of SlCr release with lactate dehydrogenase activity. The net percentage lactate dehydrogenase activity (LDH) determined by the method of Bergmeyer [10] and net percentage S;Cr release were measured from supematants taken from adult human vascular endothelial cells treated with LC sulfate (175 ~=M, 220/~M), LC (50/~M, 75 ~M, 170 ~M) and 3/~-hydroxycholenoic acid sulfate (45/~M, 90 #M, 120/~M, 175 ~aM, 220/~M). Regression analysis was used to compute the correlation coefficient of the LDH and 5tCr release. The correlation coefficient was 0.87 ± 2.5, P = 0.000017. The net ~ LDH activity is the ~ gross LDH activity after incubation with the test compound minus the ~ activity in the control wells. The net • SlCr release is t.~:e gross SICr release after incubation with the test compound minus the ~ 51Cr release in the control wells.
established marker of cytotoxicity in terms of membrane damage. The data show that percentage radiochromium release reflects cytotoxicity of trihydroxy, dihydroxy and monohydroxy bile salts. Both trihydroxy and dihydroxy bile salts, either unconjugated or conjugated with glycine, showed no cytotoxicity at the concentrations employed, which were below the critical micellar concentration cf the bile salts. Indeed, these bile salts had a 'protective' effect on endothelial cells relative to controls, suggesting that they probably maintain endothelial cells in culture. It seemed that CA, GC and UDC had a greater protective effect than DC. The greater protective effect of UDC relative to DC may be attributed to the orientation of the hydroxyl group to form the fl-epimer (3a, 7fl) which decreases the detergency of a bile salt and therefore has the effect of lowering the interaction of the bile salt to the membrane of the cell. Because of its lack of cytotoxicity UDC has been employed in the treatment of patients with primary biliary cirrhosis [12]. At the concentrations employed we did not observe chenodeoxycholate (CDC) to be cytotoxic; however it was found to be toxic when assessed on patients who develop evidence of liver injury while ingesting CDC for cholesterol gall stones
~scular endothelial cells. Phase contrast photomicrographs (magnification x 120) of: (a) Control, confluent cells incubated at 37°C in a 5~ CO2 humidified incubator in Earle's 199 complete media. (b) Cells incubated at 37 o C in a 5~ CO2 humidified incubator in Earle's 199 and treated for 24 h with 40/~M GLC (potassium salt).
45 dissolution [13]. In contrast, our observations plus others [14,15] suggest that UDC is not cytotoxic. The radiochromium model of looking at cytotoxicity could not differentiate clearly between the cytotoxic effect of UDC and GA at the concentrations employed. However, there might be a possib!e differentiation of the relative cytotoxicity if higher concentrations were used. At the concentrations employed, LC and its esters had a far greater effect than dihydroxy and trihydroxy bile acids. This may be associated with the relative insolubility of LC and its derivatives compared with polyhydroxylated bile salts. For example, CA (3~t, 7a, 12a) is many-fold ( > 4000) more soluble than the monohydroxy bile acid, LC (3a) [16]. Glucuronidation of LC increases polarity and solubility [6], and produces cholestasis comparable to that exerted by unconjugated LC in vivo. Intravenous administration of LC glucuronide, resulted in partial or complete cholestasis [6] in the rat at concentrations as low or lower than those at which lithocholate produced cholestasis [1]. Sulfation of LC also decreased its toxicity (Figs. 1A and C), which may be explained in part by the 3-fold increase in solubility of sulfolithocholic acid compared with LC [6]. However, there is no direct relationship between the solubility and cytotoxicity of the various conjugates of LC. Thus, although LC glucuronide is more soluble (7-fold) than LC sulfate [6], the threshold dose for cytotoxicity in our assay was lower for LC glucuronide than for LC sulfate. In addition glycinated trihydroxy bile acids were non-toxic (Fig. 1J), but glycination of [C increased toxicity several-fold. Therefore the increased toxic effect of GLC may be explained, in part, in terms of its interaction with the membrane [17]. Sulfation of GLC to give sulfolithoglycine abolished the toxic effect of GLC (cf Figs. 1E and F). One surprising finding in our assays was that test cytotoxicity for certain bile acids was consistently less than controls. For example, GLC sulfate and TLC sulfate had a protective effect (Figs. 1F and G) and fl-epimefization of LC and unsaturation at the 5' position to give 3fl-hydroxycholenoic acid sulfate diminished the cytotoxicity of LC sulfate (of Figs. 1C and D). This re-iterates the importance of fl-epimerisation in rendering bile salts, especially the monohydroxy bile salts, more soluble [6] and confi~'ms studies by Kimura et al. [18] and SchSlmerich et al. [19], that the fl-position of the 7-hydroxy group reduces toxicity [18]. In conclusion, sulfation and glucuronidation of bile salts such as LC decreased or abolished their toxicity to endothelial cells and suggests that these two metabolic pathways are important in hepatobiliary disease with chronic cholestasis where sulfation and glucuronidation increase with progression of the disease. Furthermore, fl-epimerisation and unsaturation of the bond at C-5 to form 3 fl-hydroxycholenoic acid also
decreased toxicity and therefore like sulfation and glucuronidation this metabolic process may be important in biliary obstruction especially in patients with biliary atresia who excrete large amounts of 3-fl-hydroxycholenoic acid in urine and serum [20]. The lack of cytotoxic effect observed with trihydroxy bile salts and the fl-hydroxy bile salt (UDC) illustrated that orientation and the number of hydroxyl groups on bile salts decreases or progressively inhibits their toxic potential. The good correlation obtained when the radiochromium assay was compared with LDH release suggested that the former is also a measure of membrane damage and that the least hydroxylated bile salts (monohydroxy bile salts) induced the greatest membrane damage to endothelial cells. Acknowledgments This work was supported by the Jules Thorn Charitable Trust. The authors would like to thank the surgical teams headed by Mr. P. McMaster and Mr. J. Buckels for obtaining the tissue samples. References 1 Javitt, N.B. and Emerman, S. (1968) J. Clin. Invest. 47, 1002-1014. 2 Kuss, E., Fernbacher, S. and Sies, H. (1973) Acta Endocrinol. (Suppl.) 173, 1. 3 Cowen, A.E., Korman, M.G., Hofmann, A.F., Cass, O.W. and Coffin, S.B. (1975) Gastroenterology 69, 67-76. 4 Carey, M.C., Wu, S.F.T. and Watkins, J.B. (1979) Biochim. Biophys. Acta. 575, 16-26. 5 Yousef, I.M., Tuchweber, B., Vonk, R.J., Masse, D., Audet, M. and Roy, C.C. (1981) Gastroenterology 80, 233-241. 60elberg, D.(3., Chari, M.V., Little, J.M., Adcock, E.W. and Lester, R. (1984) J. Clin. lnvest. 73, 1507-1514. 7 Mathis, U., Karlaganis, (3. and Preisig, R. (1983)Gastroenterology 85, 674-681. 8 Freshney, R.I. (1987) Culture of Animal Ceils. A Manual of Basic Technique, Ch. 20, pp. 257-288, Alan R. Liss, Ne~ York. 9 Maeiag, T., Cerundolo, J., llsley, S., Kelley, P.R. and Forand, R. (1979) Proc. Natl. Acad. Sci. USA. 76 (11), 5674-5678. 10 Bergmeyer, H.U. (1974) Methods Enzym. Anal. 2, 574-579. 11 Zawydiwski, R. and Duncan, G.R. (1978) In Vitro 14 (8), 707-713. 12 Poupon, R., Poupon, R.E., Calmus, Y., Chrieten, Y., Ballet, F. and Darnis, F. (1987) Lancet i, 834-835. 13 Fisher, R.L., Anderson, D.W., Boyer, J.L., Ishak, K., Klatskin, (3., Lachin, J.M. and Phillips, M.J. (1982) Hepatology 2 (2), 187-201. 14 Tint, G.S., Salen, G., Colalillo, A., Graber, D., Verga, D., Speck, J. and Shefer, S. (1982) Ann. Intern. Med. 97 (3), 351-356. 15 Bachrach, W.H. and Hofmann, A.F. (1982) Dig. Dis. Sci. 27 (9), 833-856. 16 Roda, A. and Fini, A. (1984) Hepatology 4 (5), 728-768. 17 Rudman, D. and Kendall, F.E. (1957) J. Clin. Invest. 36, 538-542. 18 Kimura, T., Shimamura, M. and Yamaguchi, A. (1981) Acta Hepat. Japan 22, 1-7. 19 Schtilmerich, J., Becher, M-S., Schmidt, K... Schubert, R., Kremer, B., Feldhaus, S. and Gerok, W. (1984) Hepatology 4 (4), 661-666. 20 Makino, I., Sjovali, J., Norman, A. and Strandvik, B. (1977) Fed. Eur. Biochem. Soc. Lett. 15-1,161-4.
