Acta pharmacol. et toxicol. 1975, 36, 13-24.
From the Department of Clinical Chemistry, Arhus kornrnunehospital, DK-8000 Arhus C, Denmark
Binding of Digitoxin to Human Serum Proteins: Influence of pH on the Binding of Digitoxin to Human Albumin BY Axel Brock (Received May 31, 1974, accepted August 27, 1974)
Abstract: The binding of digitoxin to albumin was studied under various conditions with regard to electrolytes and pH. Considerable variations of Na, K, Ca and Mg within the range of clinical relevance did not influence the binding of digitoxin to albumin. Under the influence of excessive concentrations of digitoxin this binding followed the simple law of mass action with a n average number of binding sites of about 0.5 independent of pH. The intrinsic association constants were about twice the apparent association constants, calculated in another way and with the assumption of one binding site per molecule of albumin. The binding of digitoxin to albumin was shown to depend on the pH of the medium with a maximum association constant at pH = 4.8 (intrinsic association constant, K = 5 X 105 I/mol). The binding of digitoxin to human serum proteins was shown to be equal to that of the binding to albumin. Under conditions relevant for clinical interpretations the binding of digitoxin to other proteins was insignificant. Key-words: Digitoxin - protein binding
-
albumin.
Fourty years ago it was demonstrated, that serum proteins under certain conditions were able to bind digitoxin and in this way eliminate the effect of this drug on isolated hearts (HOEKSTRA 1931). After this, many studies have been performed to characterize this drug-protein interaction. For review of the literature before 1965, SCHOLTAN et al. (1966) should be consulted. Real quantitative information concerning the binding of cardiac glycosides to plasma proteins has only been available in the last few years (SCHOLTAN et al. 1966; LUKAS& DE MARTINO 1969), and recently it has been shown, that the binding of digoxin to albumin is influenced by variations of pH (BROCK1974). The purpose of the present study has been to investigate the influence
14
AXEL BROCK
of pH on the digitoxin-albumin interaction and the influence of Na, K, Ca and Mg on the binding of digitoxin to human albumin under conditions relevant for clinical interpretation.
Materials and methods Equilibrium dialysis. The binding of digitoxin to human serum proteins was studied in an equilibrium
system similar to that used for the study of the digoxin-albumin interaction (BROCK 1974). The studies of the binding of digitoxin to albumin were performed on lyophilized human albumin (“Albumin, Grade A, KABI and “Crystallic Human Albumin”, Statens Seruminstitut, Denmark), showing one distinct electrophoretic fraction of albumin and an insignificant content of transferrin. The studies of the binding of digitoxin to native serum proteins were performed on non-fractionated serum, eventually diluted with 0.9 % saline. As the actual concentrations of albumin under the conditions studied were most frequently only 0.5-1 g/l, gelatine (Merck 4078), 5 g/l, was added as a carrier (in preliminary studies it was not possible to demonstrate any binding of digitoxin to this protein). In order to minimize the transport of water across the membrane gelatine, 5 g/1, was also added to the albumin free phase outside the membrane. Digitoxin (Merck 3044) was dissolved in 70 % alcohol (100 @ml) and added to the phase outside the membrane together with electrolytes and SH-digitoxin (NEN, NET-211). Even in the studies of the digitoxin-albumin interaction under the influence of excessive concentrations of digitoxin the concentration of alcohol in the system did not exceed 3.5 % , which in preliminary studies was shown not to influence the binding. Equilibrium conditions were established within 2-3 days, independent of the presence of gelatine. The measurements of pH and albumin concentrations were performed at equilibrium. pH was measured at the actual temperature (most frequently 37”). For the measurement of the albumin concentrations, a quantitative immunoelectrophoretic technique was used according to the principles of Laurel1 (standard: Standard-Human-Serum, Behringwerke AS). When necessary the registered 3H activity was corrected for varying quenching by an external y source, using a quenching correcting curve made from quenched SH-standards containing the same amounts of water and scintillation medium as the samples. In each study the establishment of the equilibrium conditions was controlled by measurement of the 3H activity on each side of the membrane in blanks without albumin. Calculations.
In the studies of the digitoxin albumin interaction under the influence of excessive concentrations of digitoxin, the average number of moles of digitoxin bound by one mole Acpm 1000 1 of albumin was calculated as ij X X--, where A cpm = cpm/ml inside a 765 b the membrane - cpm/ml outside the membrane, a = specific activity of SH-digitoxin, cpm/pg, and b = concentration of albumin inside the membrane (pmoV1, MW = 68.000). In the same way the molar concentration of unbound digitoxin was calculated cpm outside the membrane x -X1000 10-0 moV1. In the typical as [digitoxin] = a 165
--
15
DIGITOXIN BINDING TO SERUM PROTEIN
2 x Idigi t o x i n 1
L/rno[
Y
0.25
0.50
0.75
Fig. 1. Scatchard plot, demonstrating the binding of digitoxin to human albumin, 0.5 g f l ( 0 ) and 1.0 g/l (0)at pH = 4.6 and pH = 7.2. Medium: 0.15 M-NaCI.
case “a” was about 18,000 cpm/pg, “b” = 7.35 or 14.7 pmoV1 and [digitoxin] = 0.05 - 5 x 10-6 mol/l. Evaluating the data in a Scatchard plot, the straight lines were drawn according to the principle of least squares. For these calculations as well as for the calculations of the S. D . of the slope of the lines and the unexplained variance was used a standard computer program (REGRE, IBM 1130 Scientific Subroutine Package). The S. D. of the intercepts were calculated in conventional ways ( S o u & ROHLF 1969). In the studies of the binding of digitoxin to albumin under conditions far from saturation the apparent association constant with the assumption of one binding site per molecule of albumin, K , was calculated like that of digoxin (BROCK 1974).
AXEL BROCK
16
Table 1 . Binding of digitoxin to human albumin. Albumin
Carrier protein
A A A,dialyzed
5 5 5 5
pH
K x 10-5 Vmol
N
n 23 23 16
4.6
1.2k 0.4 4.6 f 0 . 4
4.7
4.5 f0.3
0.44 k 0.13 0.49 f0.05 0.52 f 0.04
4.7
4.5 k 1.1
0.50* 0.12
10
4.4
5.8 f0.7
0.42 f 0.06
11
g/l g/l g/l g/l
7.2
A, dialyzed
Gelatine Gelatine Gelatine Gelatine dialyzed
B
Gelatine 5 g/1
K = intrinsic association constant Vmol (fS. D.). N = average number of binding sites per molecule of albumin (k S. D.). n = number of points of the Scatchard plot. Medium: 0.15 M-NaCl.
Results Binding of digitoxin to albumin using excessive concentrations of digitoxin. The successive saturation of albumin with digitoxin was studied under equilibrium conditions, using excessive concentrations of digitoxin. Data obtained from these experiments were evaluated in Scatchard plots. Fig. 1 shows such a plot, demonstrating the digitoxin-albumin interaction at pH = 4.6 and pH = 7.2. Within the whole range straight lines were obtained in this graphical presentation indicating homogeneity of the binding sites. The average number of binding sites per molecule of albumin, N, was found to be independent of pH and, under the conditions of this study, about 0.5 (0.49 and 0.44). The intrinsic association constants, K, defined as the slope of the straight lines, were found to be about 1-5 x 50” h o l . Table 1 shows some results obtained in the same way using purified albumin and albumin obtained from different sources (albumin A: KABI, albumin B: Statens Seruminstitut). As demonstrated in the table, the results of such experiments were independent of the albumin material used for the study, neither did a purification of the albumin or the carrier previous to the study influence the binding of digitoxin to albumin. Influence of p H and the temperature on the binding of digitoxin to albumin. The influence of pH and the temperature on the binding of digitoxin to albumin was studied using pharmacological concentrations of digitoxin (10-100 ng/ml).
DIGITOXIN BINDING TO SERUM PROTEIN
17
log K ’ l/rnol
/
4
6
8
PH
Fig. 2. The influence of pH on the binding of digitoxin (10-100 ng/ml) t o human albumin, 0.5 g/l. Medium: 0.15 M-NaCI. K’: apparent association constant calculated with the assumption of one binding site. 0 : human albumin, 37”. 0 : human albumin tris 0.01 M, 37”, X : human albumin tris 0.01 M, 27”.
+ +
Fig. 2 shows that the binding of digitoxin to albumin is dependent on the pH. The apparent association constant calculated with the assumption of one binding site per molecule of albumin, K’, was found to be of maximum value at pH = 4.8 with a K’ = 2.5 x lo5 l/mol, corresponding to an intrinsic association constant K = 5 x lo5 I/mol. In the range of p H = 4.8-8.0 the apparent association constant was found to decrease with increasing pH from K’ = 2.5 x lo” Vmol at pH = 4.8 to K’ = 2.8 x lo4 at pH = 8.0. Within the range of p H = 5.3-8.0 this relationship was adequately described by the following equation: log K’ = -0.338 x pH 7.13 (r = -0.973, n = 26). Corresponding intrinsic association constants = 2 x apparent association constants (K = 2 x K’, N = 0.5).
+
18
AXEL BROCK
Table 2. Influence of Na, K, Ca and Mg on the binding of digitoxin to human albumin at 37", pH = 7.4 k 0.1 assuming one binding site per molecule of albumin. Na mmol/l 121 225 122 122 121 121 122 121 121 122 124 122
K mmol/l
2.6 5.2 10.2
5.5 10.5 2.9
Ca mmoVl
Mg mmol/l
2.1 3.2 4.4
-
3.3 1.1 4.4
0.8 0.8 0.8 0.8
K x 10-4 Vmol 4.1 4.9 5.0 4.3 5.5 3.4 7.0 5.8 5.8 5.3 4.0 7.0
Diffusible anion: CI. Duplicates. Albumin: 0.5 g/l (7.35 pmol/l). Digitoxin: 9-80 ng/ml (12-105 nmol/l).
The binding of digitoxin to albumin was found to be independent of a decrease in temperature from 37' to 27'. Therefore, within this range of temperature, AH = 0, independent of pH. Within the same range of temperature and pH = 5-8 the change in entropy, AS', was found as 20-25 cal/mol/degree. (-AGO = 6.4-7.8 Kcal/mol, calcu1at:ons based on intrinsic association constants). Influence of Nu, K , Ca and M g on the binding of digitoxin to albumin. The influence of Na, K, Ca and Mg on the binding of digitoxin to albumin was studied for pharmacological concentration of digitoxin. Table 2 shows, that even considerable variations in the concentrations of Na, K, Ca and Mg did not significantly influence the binding of digitoxin to albumin. Binding of digitoxin to serum proteins, using excessive concentrations of digitoxin. Fig. 3 shows the binding of digitoxin to the serum proteins of a healthy subject under the influence of excessive concentrations of digitoxin (unbound mol/l, p H = 4.6), mol/l, albumin: 8-16 x digitoxin: 0.15-5.2 x
DIGITOXIN BINDING TO SERUM PROTEIN
7 Idigi t o x i n ]
x
19
\/mol
Fig. 3. Scatchard plot, demonstrating the binding of digitoxin to human serum proteins, assuming albumin as the only protein able to bind digitoxin. Concentration of albumin: 0.55 g/l ( 0 ) and 1.1 g/l (O), pH = 4.6, 37".
suggesting that albumin is the only protein capable of binding digitoxin. The figure demonstrates a binding of digitoxin to serum proteins equal to that of the simple digitoxin-albumin interaction, Like this the average number of binding sites per molecule of albumin was found to be around 0.5. The intrinsic association constant was determined as 6 x lo5 l/mol, a value equal to that of the simple digitoxin-albumin interactions under the same conditions.
Binding of digitoxin to native serum. The binding of digitoxin to human serum diluted with 0.9 yo saline (albumin concentrations under the conditions studied: 0.4-0.6 g/l) was in-
20
AXEL BROCK
Table 3. Apparent associations constants for the binding of digitoxin to serum proteins assuming that albumin is the only protein able to bind digitoxin. 37", pH = 7.6. Serum albumin, g/l (original sample)
K x 10-4 Vmol
15 20 24 38 39 42 45
2.7 2.1 1.8 2.5 3.4 3.2 2.7
Apparent association constant of the simple digitoxin-albumin interaction under the same conditions: 3.6 X lo4 I/mol. Actual albumin concentrations under the conditions used: 0.4-0.6 g/l. Medium: 0.15 M-NaCI.
vestigated as 37O, pH = 7.6 k 0.1, using serum samples from normo- and hypo-albuminaemic subjects (y-globulins: 9-15 g.4). Table 3 shows that the apparent association constants, calculated with the assumption that albumin is the only protein capable of binding digitoxin, were independent of the albumin concentrations of the original serum samples and equal to that of the simple digitoxin-albumin interaction. Like the simple digitoxinalbumin interaction the corresponding intrinsic association constants = 2 X apparent association constants. The binding of digitoxin to undiluted serum proteins was studied at 37O, using pharmacological concentrations of digitoxin. Fig. 4 shows the influence of the albumin concentration on the unbound fraction of digitoxin. For the study serum samples from 38 hypo- and normo-albuminaemic subjects were used. The albumin concentrations given in fig. 4 are the actual concentrations under equilibrium conditions. The figure demonstrates a close relation between the albumin concentration and the unbound fraction of digitoxin.
Discussion The binding of digitoxin to human albumin and native serum was studied in an equilibrium dialysis system similar to that used for the study of the digoxin-albumin interaction (BROCK1974), a system with great advantages, as partial adsorption to the membrane or to the glass surface does not
DIGITOXIN BINDING TO SERUM PROTEIN
21
Unbound f r a c t i o n o f d i g i t ox i n 0.1 a a
,
0
I
@
I
a
I
0.1 1
a
\
0.0'
10
20
30 LO A l b u m i n g/L
Fig. 4. Influence of the albumin concentration on the binding of digitoxin to human serum proteins (digitoxin: 90-145 ng/rnl, pH = 7.2 f 0.2, 37"). Curve: unbound fractions of digitoxin calculated from the actual association constant according to fig. 2.
influence the calculations (EISEN1964). In the present study the slight solubility of digitoxin in aqueous solutions has limited the ranges of albumin and digitoxin concentrations to be studied. Most frequently albumin concentrations of 0.5-1 g/l were chosen, as these concentrations caused a distribution of digitoxin making the calculations maximum independent of radiochemical impurities not bound by the albumin. It was demonstrated, that the binding of digitoxin to human albumin was influenced by variations in pH. The association constant reached the maximum value around pH = 4.8. In the range of pH = 5.3-8 the relationship between pH and the apparent association constant of the digitoxin-
22
AXEL BROCK
albumin interaction was found to be adequately described by the simple equation: log K' = -0.338 X pH 7.13. This equation permits calculations of the apparent association constants of digitoxin-albumin systems with known pH, but interpretations with respect to the nature of this interaction should not be done from this quite empirical relation between the binding energy and pH. In the study of the digitoxin-albumin interaction made by LUKAS& DE MARTINO (1969) this effect of pH was not demonstrated, but precise experimental data were not presented. LUKAS& DE MARTINO (1969) demonstrated the binding of digitoxin to albumin to be influenced by the temperature. In the present study the association constants were found to be independent of decreasing the temperature from 37' to 27'. Within this range the binding process was followed by a positive change in entropy = 20-25 cal/mol/degree. Because of the very narrow range of the temperature, all these thermodynamic constants should be accepted with some reservations. However, the binding process is adequately followed by a moderate positive gain in entropy, probably indi1959). cating the binding process to be of hydrophobic nature (KAUZMANN In the study by LUKAS& DE MARTINO (1969) the maximum average number of moles of digitoxin bound by one mole of albumin was found to be about one. In the present study the average number of binding sites was found to be about 0.5 and independent of pH. This value of about 0.5 was found for two different preparations of albumin as well as for nonfractionated serum, suggesting serum albumin as the only protein able to bind digitoxin. In each case the apparent association constants, calculated in another way and with the assumption of one binding site per molecule of albumin, were found to be about half the intrinsic association constants. This fact is consistent with the presence of an average of 0.5 binding sites per molecule of albumin. A possible source of error, able to explain the discrepancy between the observations concerning the number of binding sites made in the present study and in that of LUKAS& DE MARTINO (1969), should be the presence of large amounts of impurities of the albumin preparations, giving the average number of binding sites = 0.5, but different electrophoretic techniques have failed to disclose such impurities. Inhomogeneity of the albumin preparations, for example partial denaturation, or significant differences between the SH-digitoxin-albuminand the digitoxin-albumin interaction may also be possible sources of error. The present study cannot give any acceptable explanation for the discrepancy between these two sets of observations, and further studies must be required if any conclusions concerning the molecular mechanism can be drawn. Using native serum from subjects with varying concentrations of albumin and suggesting albumin as the only protein able to bind digitoxin, the as-
+
DIGITOXIN BINDING TO SERUM PROTEIN
23
Table 4 . Calculated fraction of unbound digitoxin in serum at 37" with the assumption of albumin as the only binding protein.
dl
pH = 6.8 (K= 6.8 X lo4)
10 20 30 40 50
0.091 0.048 0.032 0.024 0.020
Albumin
pH = 7.1
pH = 8.0
5.4 X lo4)
pH = 7.4 (K = 4.3 X 104)
pH = 7.7 (K= 3.4 X lo")
2.7 X 104)
0.112 0.059 0.040 0.031 0.025
0.137 0.073 0.050 0.038 0.031
0.167 0.091 0.063 0.048 0.038
0.201 0.112 0.077 0.059 0.048
(K' =
(K' =
sociation constants were found to be equal to that of the simple digitoxinalbumin interaction under the same conditions, indicating that the influence of other proteins than albumin is negligible under conditions relevant for clinical interpretation. Assuming that albumin is the only protein able to bind digitoxin, the unbound fractions of the drug in serum can be calculated from the apparent association constants obtained from the study of the simple digitoxinalbumin system. Table 4 demonstrates the expected influence of the albumin concentrations and the p H based on such theoretical considerations, demonstrating for example an increase in the unbound fraction of digitoxin from 3 to 13 yo, when the concentration of albumin is decreased from 50 to 10 8/1 (pH = 7.4). These calculations should be valid within the whole therapeutic range of digitoxin. Using undiluted serum samples the observed unbound fractions of digitoxin were found to be higher than those calculated. This discrepancy might be explained by the presence of radiochemical impurities unable to be bound by albumin or other proteins. Under the conditions of these studies such radiochemical impurities, present in concentrations of 1-2 yoof the total radioactivity, would explain this discrepancy between observed and predicted values. Uncertainty of the determinations of the albumin concentrations may be another source of error. Finally deviations form the ideal conditions assumed in the use of the simple law of mass action should be expected, as the apparent association constants used for the calculations were obtained under quite different conditions with regard to the albumin concentrations. In any case, for clinical purposes a possible binding of digitoxin to the y-globulins can be neglected in the calculations of the unbound fraction of digitoxin. Calculations based on albumin concentrations and pH show, that variations of these two parameters influence the unbound fraction of
24
AXEL BROCK
digitoxin highly significantly. Under conditions relevant for clinical interpretation the effect of variations in the albumin concentration is the main one, even if variations of pH will also cause material variations in the unbound fraction of digitoxin.
REFERENCES Brock, A.: Binding of digoxin to human serum proteins: Influence of pH on the binding of digoxin to human albumin. Acta pharmacol. et toxicol. 1974, 34, 260266. Eisen, H. N.: Equilibrium dialysis for measurement of antibody - hapten affinities. Meth. med. Res. 1964, 10, 106-114. Hoekstra, R. A.: Das Verhalten von Digitalisglykosiden in Blut und Gewebsfliissigkeit. Arch. exp. Path. Pharmacol. 1931, 162, 649-662. Kauzmann, W.:Some factors in the interpretation of protein denaturation. Adv. Protein Chem. 1959, 14, 1-63. Lukas, D. S. & A. G. De Martino: Binding of digitoxin and some related cardenolides to human plasma proteins. J . clin. Invest. 1969, 48, 1041-1053. Scholtan, W.,K. Schlossman & H. Rosenkranz: Bestimmung der Eiweissbindung von Digitalis-Praparaten mittels der Ultracentrifuge. Arzneimittelforsch. 1966, 16, 109-118. Sokal, R. R. & F. J. Rohlf: Biometry. W. H. Freeman and Company, San Francisco 1969.
From the Department of Clinical Chemistry, Arhus kornrnunehospital, DK-8000 Arhus C, Denmark
Binding of Digitoxin to Human Serum Proteins: Influence of pH on the Binding of Digitoxin to Human Albumin BY Axel Brock (Received May 31, 1974, accepted August 27, 1974)
Abstract: The binding of digitoxin to albumin was studied under various conditions with regard to electrolytes and pH. Considerable variations of Na, K, Ca and Mg within the range of clinical relevance did not influence the binding of digitoxin to albumin. Under the influence of excessive concentrations of digitoxin this binding followed the simple law of mass action with a n average number of binding sites of about 0.5 independent of pH. The intrinsic association constants were about twice the apparent association constants, calculated in another way and with the assumption of one binding site per molecule of albumin. The binding of digitoxin to albumin was shown to depend on the pH of the medium with a maximum association constant at pH = 4.8 (intrinsic association constant, K = 5 X 105 I/mol). The binding of digitoxin to human serum proteins was shown to be equal to that of the binding to albumin. Under conditions relevant for clinical interpretations the binding of digitoxin to other proteins was insignificant. Key-words: Digitoxin - protein binding
-
albumin.
Fourty years ago it was demonstrated, that serum proteins under certain conditions were able to bind digitoxin and in this way eliminate the effect of this drug on isolated hearts (HOEKSTRA 1931). After this, many studies have been performed to characterize this drug-protein interaction. For review of the literature before 1965, SCHOLTAN et al. (1966) should be consulted. Real quantitative information concerning the binding of cardiac glycosides to plasma proteins has only been available in the last few years (SCHOLTAN et al. 1966; LUKAS& DE MARTINO 1969), and recently it has been shown, that the binding of digoxin to albumin is influenced by variations of pH (BROCK1974). The purpose of the present study has been to investigate the influence
14
AXEL BROCK
of pH on the digitoxin-albumin interaction and the influence of Na, K, Ca and Mg on the binding of digitoxin to human albumin under conditions relevant for clinical interpretation.
Materials and methods Equilibrium dialysis. The binding of digitoxin to human serum proteins was studied in an equilibrium
system similar to that used for the study of the digoxin-albumin interaction (BROCK 1974). The studies of the binding of digitoxin to albumin were performed on lyophilized human albumin (“Albumin, Grade A, KABI and “Crystallic Human Albumin”, Statens Seruminstitut, Denmark), showing one distinct electrophoretic fraction of albumin and an insignificant content of transferrin. The studies of the binding of digitoxin to native serum proteins were performed on non-fractionated serum, eventually diluted with 0.9 % saline. As the actual concentrations of albumin under the conditions studied were most frequently only 0.5-1 g/l, gelatine (Merck 4078), 5 g/l, was added as a carrier (in preliminary studies it was not possible to demonstrate any binding of digitoxin to this protein). In order to minimize the transport of water across the membrane gelatine, 5 g/1, was also added to the albumin free phase outside the membrane. Digitoxin (Merck 3044) was dissolved in 70 % alcohol (100 @ml) and added to the phase outside the membrane together with electrolytes and SH-digitoxin (NEN, NET-211). Even in the studies of the digitoxin-albumin interaction under the influence of excessive concentrations of digitoxin the concentration of alcohol in the system did not exceed 3.5 % , which in preliminary studies was shown not to influence the binding. Equilibrium conditions were established within 2-3 days, independent of the presence of gelatine. The measurements of pH and albumin concentrations were performed at equilibrium. pH was measured at the actual temperature (most frequently 37”). For the measurement of the albumin concentrations, a quantitative immunoelectrophoretic technique was used according to the principles of Laurel1 (standard: Standard-Human-Serum, Behringwerke AS). When necessary the registered 3H activity was corrected for varying quenching by an external y source, using a quenching correcting curve made from quenched SH-standards containing the same amounts of water and scintillation medium as the samples. In each study the establishment of the equilibrium conditions was controlled by measurement of the 3H activity on each side of the membrane in blanks without albumin. Calculations.
In the studies of the digitoxin albumin interaction under the influence of excessive concentrations of digitoxin, the average number of moles of digitoxin bound by one mole Acpm 1000 1 of albumin was calculated as ij X X--, where A cpm = cpm/ml inside a 765 b the membrane - cpm/ml outside the membrane, a = specific activity of SH-digitoxin, cpm/pg, and b = concentration of albumin inside the membrane (pmoV1, MW = 68.000). In the same way the molar concentration of unbound digitoxin was calculated cpm outside the membrane x -X1000 10-0 moV1. In the typical as [digitoxin] = a 165
--
15
DIGITOXIN BINDING TO SERUM PROTEIN
2 x Idigi t o x i n 1
L/rno[
Y
0.25
0.50
0.75
Fig. 1. Scatchard plot, demonstrating the binding of digitoxin to human albumin, 0.5 g f l ( 0 ) and 1.0 g/l (0)at pH = 4.6 and pH = 7.2. Medium: 0.15 M-NaCI.
case “a” was about 18,000 cpm/pg, “b” = 7.35 or 14.7 pmoV1 and [digitoxin] = 0.05 - 5 x 10-6 mol/l. Evaluating the data in a Scatchard plot, the straight lines were drawn according to the principle of least squares. For these calculations as well as for the calculations of the S. D . of the slope of the lines and the unexplained variance was used a standard computer program (REGRE, IBM 1130 Scientific Subroutine Package). The S. D. of the intercepts were calculated in conventional ways ( S o u & ROHLF 1969). In the studies of the binding of digitoxin to albumin under conditions far from saturation the apparent association constant with the assumption of one binding site per molecule of albumin, K , was calculated like that of digoxin (BROCK 1974).
AXEL BROCK
16
Table 1 . Binding of digitoxin to human albumin. Albumin
Carrier protein
A A A,dialyzed
5 5 5 5
pH
K x 10-5 Vmol
N
n 23 23 16
4.6
1.2k 0.4 4.6 f 0 . 4
4.7
4.5 f0.3
0.44 k 0.13 0.49 f0.05 0.52 f 0.04
4.7
4.5 k 1.1
0.50* 0.12
10
4.4
5.8 f0.7
0.42 f 0.06
11
g/l g/l g/l g/l
7.2
A, dialyzed
Gelatine Gelatine Gelatine Gelatine dialyzed
B
Gelatine 5 g/1
K = intrinsic association constant Vmol (fS. D.). N = average number of binding sites per molecule of albumin (k S. D.). n = number of points of the Scatchard plot. Medium: 0.15 M-NaCl.
Results Binding of digitoxin to albumin using excessive concentrations of digitoxin. The successive saturation of albumin with digitoxin was studied under equilibrium conditions, using excessive concentrations of digitoxin. Data obtained from these experiments were evaluated in Scatchard plots. Fig. 1 shows such a plot, demonstrating the digitoxin-albumin interaction at pH = 4.6 and pH = 7.2. Within the whole range straight lines were obtained in this graphical presentation indicating homogeneity of the binding sites. The average number of binding sites per molecule of albumin, N, was found to be independent of pH and, under the conditions of this study, about 0.5 (0.49 and 0.44). The intrinsic association constants, K, defined as the slope of the straight lines, were found to be about 1-5 x 50” h o l . Table 1 shows some results obtained in the same way using purified albumin and albumin obtained from different sources (albumin A: KABI, albumin B: Statens Seruminstitut). As demonstrated in the table, the results of such experiments were independent of the albumin material used for the study, neither did a purification of the albumin or the carrier previous to the study influence the binding of digitoxin to albumin. Influence of p H and the temperature on the binding of digitoxin to albumin. The influence of pH and the temperature on the binding of digitoxin to albumin was studied using pharmacological concentrations of digitoxin (10-100 ng/ml).
DIGITOXIN BINDING TO SERUM PROTEIN
17
log K ’ l/rnol
/
4
6
8
PH
Fig. 2. The influence of pH on the binding of digitoxin (10-100 ng/ml) t o human albumin, 0.5 g/l. Medium: 0.15 M-NaCI. K’: apparent association constant calculated with the assumption of one binding site. 0 : human albumin, 37”. 0 : human albumin tris 0.01 M, 37”, X : human albumin tris 0.01 M, 27”.
+ +
Fig. 2 shows that the binding of digitoxin to albumin is dependent on the pH. The apparent association constant calculated with the assumption of one binding site per molecule of albumin, K’, was found to be of maximum value at pH = 4.8 with a K’ = 2.5 x lo5 l/mol, corresponding to an intrinsic association constant K = 5 x lo5 I/mol. In the range of p H = 4.8-8.0 the apparent association constant was found to decrease with increasing pH from K’ = 2.5 x lo” Vmol at pH = 4.8 to K’ = 2.8 x lo4 at pH = 8.0. Within the range of p H = 5.3-8.0 this relationship was adequately described by the following equation: log K’ = -0.338 x pH 7.13 (r = -0.973, n = 26). Corresponding intrinsic association constants = 2 x apparent association constants (K = 2 x K’, N = 0.5).
+
18
AXEL BROCK
Table 2. Influence of Na, K, Ca and Mg on the binding of digitoxin to human albumin at 37", pH = 7.4 k 0.1 assuming one binding site per molecule of albumin. Na mmol/l 121 225 122 122 121 121 122 121 121 122 124 122
K mmol/l
2.6 5.2 10.2
5.5 10.5 2.9
Ca mmoVl
Mg mmol/l
2.1 3.2 4.4
-
3.3 1.1 4.4
0.8 0.8 0.8 0.8
K x 10-4 Vmol 4.1 4.9 5.0 4.3 5.5 3.4 7.0 5.8 5.8 5.3 4.0 7.0
Diffusible anion: CI. Duplicates. Albumin: 0.5 g/l (7.35 pmol/l). Digitoxin: 9-80 ng/ml (12-105 nmol/l).
The binding of digitoxin to albumin was found to be independent of a decrease in temperature from 37' to 27'. Therefore, within this range of temperature, AH = 0, independent of pH. Within the same range of temperature and pH = 5-8 the change in entropy, AS', was found as 20-25 cal/mol/degree. (-AGO = 6.4-7.8 Kcal/mol, calcu1at:ons based on intrinsic association constants). Influence of Nu, K , Ca and M g on the binding of digitoxin to albumin. The influence of Na, K, Ca and Mg on the binding of digitoxin to albumin was studied for pharmacological concentration of digitoxin. Table 2 shows, that even considerable variations in the concentrations of Na, K, Ca and Mg did not significantly influence the binding of digitoxin to albumin. Binding of digitoxin to serum proteins, using excessive concentrations of digitoxin. Fig. 3 shows the binding of digitoxin to the serum proteins of a healthy subject under the influence of excessive concentrations of digitoxin (unbound mol/l, p H = 4.6), mol/l, albumin: 8-16 x digitoxin: 0.15-5.2 x
DIGITOXIN BINDING TO SERUM PROTEIN
7 Idigi t o x i n ]
x
19
\/mol
Fig. 3. Scatchard plot, demonstrating the binding of digitoxin to human serum proteins, assuming albumin as the only protein able to bind digitoxin. Concentration of albumin: 0.55 g/l ( 0 ) and 1.1 g/l (O), pH = 4.6, 37".
suggesting that albumin is the only protein capable of binding digitoxin. The figure demonstrates a binding of digitoxin to serum proteins equal to that of the simple digitoxin-albumin interaction, Like this the average number of binding sites per molecule of albumin was found to be around 0.5. The intrinsic association constant was determined as 6 x lo5 l/mol, a value equal to that of the simple digitoxin-albumin interactions under the same conditions.
Binding of digitoxin to native serum. The binding of digitoxin to human serum diluted with 0.9 yo saline (albumin concentrations under the conditions studied: 0.4-0.6 g/l) was in-
20
AXEL BROCK
Table 3. Apparent associations constants for the binding of digitoxin to serum proteins assuming that albumin is the only protein able to bind digitoxin. 37", pH = 7.6. Serum albumin, g/l (original sample)
K x 10-4 Vmol
15 20 24 38 39 42 45
2.7 2.1 1.8 2.5 3.4 3.2 2.7
Apparent association constant of the simple digitoxin-albumin interaction under the same conditions: 3.6 X lo4 I/mol. Actual albumin concentrations under the conditions used: 0.4-0.6 g/l. Medium: 0.15 M-NaCI.
vestigated as 37O, pH = 7.6 k 0.1, using serum samples from normo- and hypo-albuminaemic subjects (y-globulins: 9-15 g.4). Table 3 shows that the apparent association constants, calculated with the assumption that albumin is the only protein capable of binding digitoxin, were independent of the albumin concentrations of the original serum samples and equal to that of the simple digitoxin-albumin interaction. Like the simple digitoxinalbumin interaction the corresponding intrinsic association constants = 2 X apparent association constants. The binding of digitoxin to undiluted serum proteins was studied at 37O, using pharmacological concentrations of digitoxin. Fig. 4 shows the influence of the albumin concentration on the unbound fraction of digitoxin. For the study serum samples from 38 hypo- and normo-albuminaemic subjects were used. The albumin concentrations given in fig. 4 are the actual concentrations under equilibrium conditions. The figure demonstrates a close relation between the albumin concentration and the unbound fraction of digitoxin.
Discussion The binding of digitoxin to human albumin and native serum was studied in an equilibrium dialysis system similar to that used for the study of the digoxin-albumin interaction (BROCK1974), a system with great advantages, as partial adsorption to the membrane or to the glass surface does not
DIGITOXIN BINDING TO SERUM PROTEIN
21
Unbound f r a c t i o n o f d i g i t ox i n 0.1 a a
,
0
I
@
I
a
I
0.1 1
a
\
0.0'
10
20
30 LO A l b u m i n g/L
Fig. 4. Influence of the albumin concentration on the binding of digitoxin to human serum proteins (digitoxin: 90-145 ng/rnl, pH = 7.2 f 0.2, 37"). Curve: unbound fractions of digitoxin calculated from the actual association constant according to fig. 2.
influence the calculations (EISEN1964). In the present study the slight solubility of digitoxin in aqueous solutions has limited the ranges of albumin and digitoxin concentrations to be studied. Most frequently albumin concentrations of 0.5-1 g/l were chosen, as these concentrations caused a distribution of digitoxin making the calculations maximum independent of radiochemical impurities not bound by the albumin. It was demonstrated, that the binding of digitoxin to human albumin was influenced by variations in pH. The association constant reached the maximum value around pH = 4.8. In the range of pH = 5.3-8 the relationship between pH and the apparent association constant of the digitoxin-
22
AXEL BROCK
albumin interaction was found to be adequately described by the simple equation: log K' = -0.338 X pH 7.13. This equation permits calculations of the apparent association constants of digitoxin-albumin systems with known pH, but interpretations with respect to the nature of this interaction should not be done from this quite empirical relation between the binding energy and pH. In the study of the digitoxin-albumin interaction made by LUKAS& DE MARTINO (1969) this effect of pH was not demonstrated, but precise experimental data were not presented. LUKAS& DE MARTINO (1969) demonstrated the binding of digitoxin to albumin to be influenced by the temperature. In the present study the association constants were found to be independent of decreasing the temperature from 37' to 27'. Within this range the binding process was followed by a positive change in entropy = 20-25 cal/mol/degree. Because of the very narrow range of the temperature, all these thermodynamic constants should be accepted with some reservations. However, the binding process is adequately followed by a moderate positive gain in entropy, probably indi1959). cating the binding process to be of hydrophobic nature (KAUZMANN In the study by LUKAS& DE MARTINO (1969) the maximum average number of moles of digitoxin bound by one mole of albumin was found to be about one. In the present study the average number of binding sites was found to be about 0.5 and independent of pH. This value of about 0.5 was found for two different preparations of albumin as well as for nonfractionated serum, suggesting serum albumin as the only protein able to bind digitoxin. In each case the apparent association constants, calculated in another way and with the assumption of one binding site per molecule of albumin, were found to be about half the intrinsic association constants. This fact is consistent with the presence of an average of 0.5 binding sites per molecule of albumin. A possible source of error, able to explain the discrepancy between the observations concerning the number of binding sites made in the present study and in that of LUKAS& DE MARTINO (1969), should be the presence of large amounts of impurities of the albumin preparations, giving the average number of binding sites = 0.5, but different electrophoretic techniques have failed to disclose such impurities. Inhomogeneity of the albumin preparations, for example partial denaturation, or significant differences between the SH-digitoxin-albuminand the digitoxin-albumin interaction may also be possible sources of error. The present study cannot give any acceptable explanation for the discrepancy between these two sets of observations, and further studies must be required if any conclusions concerning the molecular mechanism can be drawn. Using native serum from subjects with varying concentrations of albumin and suggesting albumin as the only protein able to bind digitoxin, the as-
+
DIGITOXIN BINDING TO SERUM PROTEIN
23
Table 4 . Calculated fraction of unbound digitoxin in serum at 37" with the assumption of albumin as the only binding protein.
dl
pH = 6.8 (K= 6.8 X lo4)
10 20 30 40 50
0.091 0.048 0.032 0.024 0.020
Albumin
pH = 7.1
pH = 8.0
5.4 X lo4)
pH = 7.4 (K = 4.3 X 104)
pH = 7.7 (K= 3.4 X lo")
2.7 X 104)
0.112 0.059 0.040 0.031 0.025
0.137 0.073 0.050 0.038 0.031
0.167 0.091 0.063 0.048 0.038
0.201 0.112 0.077 0.059 0.048
(K' =
(K' =
sociation constants were found to be equal to that of the simple digitoxinalbumin interaction under the same conditions, indicating that the influence of other proteins than albumin is negligible under conditions relevant for clinical interpretation. Assuming that albumin is the only protein able to bind digitoxin, the unbound fractions of the drug in serum can be calculated from the apparent association constants obtained from the study of the simple digitoxinalbumin system. Table 4 demonstrates the expected influence of the albumin concentrations and the p H based on such theoretical considerations, demonstrating for example an increase in the unbound fraction of digitoxin from 3 to 13 yo, when the concentration of albumin is decreased from 50 to 10 8/1 (pH = 7.4). These calculations should be valid within the whole therapeutic range of digitoxin. Using undiluted serum samples the observed unbound fractions of digitoxin were found to be higher than those calculated. This discrepancy might be explained by the presence of radiochemical impurities unable to be bound by albumin or other proteins. Under the conditions of these studies such radiochemical impurities, present in concentrations of 1-2 yoof the total radioactivity, would explain this discrepancy between observed and predicted values. Uncertainty of the determinations of the albumin concentrations may be another source of error. Finally deviations form the ideal conditions assumed in the use of the simple law of mass action should be expected, as the apparent association constants used for the calculations were obtained under quite different conditions with regard to the albumin concentrations. In any case, for clinical purposes a possible binding of digitoxin to the y-globulins can be neglected in the calculations of the unbound fraction of digitoxin. Calculations based on albumin concentrations and pH show, that variations of these two parameters influence the unbound fraction of
24
AXEL BROCK
digitoxin highly significantly. Under conditions relevant for clinical interpretation the effect of variations in the albumin concentration is the main one, even if variations of pH will also cause material variations in the unbound fraction of digitoxin.
REFERENCES Brock, A.: Binding of digoxin to human serum proteins: Influence of pH on the binding of digoxin to human albumin. Acta pharmacol. et toxicol. 1974, 34, 260266. Eisen, H. N.: Equilibrium dialysis for measurement of antibody - hapten affinities. Meth. med. Res. 1964, 10, 106-114. Hoekstra, R. A.: Das Verhalten von Digitalisglykosiden in Blut und Gewebsfliissigkeit. Arch. exp. Path. Pharmacol. 1931, 162, 649-662. Kauzmann, W.:Some factors in the interpretation of protein denaturation. Adv. Protein Chem. 1959, 14, 1-63. Lukas, D. S. & A. G. De Martino: Binding of digitoxin and some related cardenolides to human plasma proteins. J . clin. Invest. 1969, 48, 1041-1053. Scholtan, W.,K. Schlossman & H. Rosenkranz: Bestimmung der Eiweissbindung von Digitalis-Praparaten mittels der Ultracentrifuge. Arzneimittelforsch. 1966, 16, 109-118. Sokal, R. R. & F. J. Rohlf: Biometry. W. H. Freeman and Company, San Francisco 1969.
