THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 236, No. 12, December 1961

Printed in U.S.A.

Thyroxine Binding by Human Serum Albumin*

GEORGE L. TRITSCH, CHARLES E. RATHKE,~ NORMA E. TRITSCH, AND CATHARINE M. WEISS

From the Roswell Park Memorial Institute, Bufalo 3, New York

(Received for publication, February 10, 1961)

Thyroxine and triiodothyronine are transported in the blood as noncovalent complexes with certain serum proteins. This has been shown by many investigators and was the subject of an extensive recent review by Robbins and Rall (1). The bind- ing proteins are an a-globulin (called thyroxine-binding globu- lin), a prealbumin, and serum albumin. The first appears to be the most significant binder of the thyroxine normally present under physiological conditions. Of these three, only albumin is presently available in a state of reasonable purity, and for this reason we have studied the binding of thyroxine by this protein. The present paper presents our findings with regard to the in- tensity and the optimal conditions for this interaction.

The measurements of the binding are based on the observa- tion by Klebanoff (2) that thyroxine has a stimulatory effect on the oxidation of reduced diphosphopyridine nucleotide by hy- drogen peroxide in the presence of peroxidase. This stimula- tory effect is obliterated by albumin, and it will be shown that this is a quantitative measure of thyroxine binding by albumin. This is in agreement with the preliminary observation by Kle- banoff that plasma proteins can inhibit the thyroxine-potentiated oxidation of diphosphopyridine nucleotide by the peroxidase system, although an ultrafiltrable inhibitor was found to be present as well.

EXPERIMENTAL PROCEDURE

Methods-The rate of oxidation of DPNH to DPN was deter- mined by the decrease in optical density of a 1 cm depth of the reaction mixture at 340 rnH in a Beckman model DU spectro- photometer at room temperature. The blank contained only the buffer used in the reaction cell. Volumes of 0.1 ml or less were measured in calibrated micropipets of the Linderstrem- Lang-Holter type.l Thyroxine was dissolved in 0.155 M sodium hydroxide, and microliter quantities of this solution (0.010 M in thyroxine) were added to the various buffered solutions. The effect of this amount of base on the pH was negligible.

The pH was measured in a Beckman model G pH meter. Unless the effect of pH was under study, all experiments were performed at pH 7.35. Appropriate mixtures of only NaHzP04 and Na2HP04 were used to prepare buffers between pH 5.4 and 8.6. Phosphate was selected because it is a good buffer in the physiological range and because phosphate has been shown to

* This investigation was supported in part by a Research Grant (A-40301 from the National Institute of Arthritis and Metabolic Disease; of the United States Public Health Service.

t Participant in Roswell Park Memorial Institute’s Seventh Summer Program in Science. 1960. Mr. Rathke’s stioend was m&de availaGe through an Experimental Training Graht (2R-6) from the Public Health Service to the University of Rochester.

1 Arthur H. Thomas Company, No. 2464-J.

cause minimal interference in the binding of small molecules by albumin (3). At pH 5.37 and below, acetate buffers were used. With the exception of those instances in which the effect of ionic strength was under investigation, the reaction mixtures were of 0.155 ionic strength, so as to approximate the ionic strength of physiological fluids. Buffer solution of ionic strength 0.233 (2.0 ml) was placed in the spectrophotometer cuvette. The other reactants, except for the minute amount of thyroxine in NaOH as noted above, were dissolved in water and added to the buffer. The &al volume was brought to 3.0 ml with water to give an ionic strength of 0.155.

Protein concentration was determined by the biuret method (4), which was calibrated by means of Kjeldahl determinations. Albumin, in the presence of other proteins, was measured by the method of Bracken and Klotz (5).

MaterialsHorse-radish peroxidase (lot No. P 60B-213; 68 purpurogallin units per mg), DPNH, and the sodium salt of L-thyroxine were purchased from the Sigma Chemical Company. The human serum albumin was the crystallized Pentex product (lot No. 45F04). It was shown by electrophoresis at pH 8.6 to be free from globulins, and showed the expected heterogeneity on electrophoresis at pH 4.7. A molecular weight of 69,000 was assumed for the albumin.

The crystallized bovine serum albumin and the Cohn frac- tions were purchased from Pentex, Inc. The normal human plasma was obtained through the courtesy of Dr. Elias Cohen and the Roswell Park Memorial Institute blood bank. Poly- vinylpyrrolidone was purchased from the Oxford Laboratories.

RESULTS AND CONCLUSIONS

Conditions for Binding-Fig. 1 describes an experiment de- signed to demonstrate the binding of L-thyroxine to human al- bumin. The addition of 0.05 pmole of L-thyroxine to the per- oxidase system caused an immediate decrease in optical density at a rate of 0.320 optical density unit per minute (solid circles, Fig. 1). When 0.04 pmole of albumin was added to such a re- action mixture before all the DPNH was oxidized, (open circles, Fig. l), the rate of reaction immediately decreased to 0.019 optical density unit per minute. This slope was identical with the one observed when the same amounts of L-thyroxine and albumin were mixed beforehand and added to the reaction mix- ture (solid sG%ures, Fig. 1). These results indicate that the binding of thyroxine by albumin is instantaneous, and that the order of addition of thyroxine and albumin is of no consequence. The fact that the slopes were identical also indicated that the concentration of unbound thyroxine was the same in both cases. As will be shown below, the slope of the curve is a direct meas- ure of the amount of unbound L-thyroxine. In the presence of

3163

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from

3164 Thyroxine Binding by Albumin

0,600

0.500

% g 0.400

5

c g 0.300 x

8 0200 8

0.100

peroxidase only (open squares, Fig. 1) the slope was found to be 0.0035 optical density unit per minute. The fact that in the presence of albumin, the rate was 5.5 times faster (0.019 versus 0.0035 optical density unit per minute) indicated that some of the L-thyroxine was not bound by the added amount of albumin. It has thus been demonstrated that albumin, by binding L- thyroxine, can slow down the rate of reaction, and that this effect is identical in magnitude whatever the order of addition of albumin and n-thyroxine.

Quantitative Determination of L-Thyroxine Binding at pH ?‘.%i -It was shown by Klebanoff that the increase of the rate of the peroxidase reaction was directly proportional to the n-thyroxine concentration. We have repeated his experiment at pH 7.35 and an ionic strength of 0.155 and have also observed a linear relationship (Fig. 2). Thus, Fig. 2B may be used as a calibra- tion curve for the determination of the concentration of free n-thyroxinein solutionsof pH 7.35 phosphatebuffer, ionic strength of 0.155. The observations illustrated in Fig. 1 were therefore extended as shown in Fig. 3. Here, various amounts of human albumin were added to n-thyroxine in the assay mixture, and the decrease in the rate of the reaction was observed. The picture obtained was similar to Fig. 2A in which the effect of varying the amount of n-thyroxine alone was observed. Since the rate of the reaction is proportional to the concentration of free n-thyroxine, Fig. 2B allows the determination of the con- centration of unbound n-thyroxine in each of the reaction mix- tures used. Fig. 4 shows the data plotted according to the

Vol. 236, No. 12

0 I 2 3 4 5 6 7 8 9 IO

TIME (minutes)

FIG. 1. Effect of human albumin on the thyroxine-potentiated oxidation of DPNH. All reaction mixtures contained 2.0 ml of phosphate buffer, pH 7.35, ionic strength 0.233; 10 pmoles of Hz02 in 0.1 ml of water; and 0.40 pmole of DPNH in 0.1 ml of water and enough water (added before the peroxidase) to make the total volume 3.0 ml after the last addition. The timer was started, and readings were taken at 30-second intervals. At 2 minutes, 0.1 mg of horse-radish peroxidase in 0.1 ml of water was added to each cuvette. At 4 minutes, the following materials were added: O-0, 0.05 pmole of thyroxine in 0.005 ml of 0.005 M NaOH; m---m, a mixture of 0.05 pmole of thyroxine in 0.005 ml of 0.155 M NaOH plus 0.049 pmole of albumin in 0.1 ml of water; O-O, 0.05 pmole of thyroxine in 0.005 ml of 0.155 M NaOH fol- lowed by0.049 pmole of albumin in0.1 ml of water at 64 minutes; III-•I, no further additions.

0.600 -0.300 10

m :: t 0.400 0.200 L1

z $

z 5 2 0.300 0.01 6

: 2 0.200 - 0.100 5 z

0 t

::

0

0.500 /’ 0.005

0.100 2

0.05 0.04 0.02 2 0.03

b 4 5 6 7 8 9 10 II 0 0.01 0.02 0.03 0.04 0.05

TIME (minutes) L-THYROXINE ( ymoles)

FIG. 2. Effect of thyroxine concentration on the DPNH oxida- tion at pH 7.35. A, the result when different quantities (the numbers next to the curves indicate micromoles of L-thyroxine added) of thyroxine are added to the mixtures of buffer, HzOz, DPNH, and peroxidase indicated in Fig. 1. B, the data from A yield a linear calibration curve.

0.700

0.600

P 0 0.500 I%

G

5 z 0.400 it!

d I;? t O 0.300

0.200

I 4 6 7 8

TIME (mmutes)

FIG. 3. Effect of variation in the amount of thyroxine and human albumin on the thyroxine-potentiated oxidation of DPNH at pH 7.35. Quantities of buffer, HzOz, DPNH, and peroxidase, are the same as in Fig. 1. The numbers next to the curves refer to the following mixtures of micromoles of thyroxine and albumin: 1, 0.025 + 0.0432; 8, 0.050 + 0.0432; 3, 0.050 + 0.0281; 4, 0.100 + 0.0432; 6, 0.125 + 0.0432; 6, 0.150 + 0.0432; 7, 0.175 + 0.0432; 8, 0.200 + 0.0432.

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from

December 1961 G. L. Tritsch, C. E. Rathke, N. E. Tritsch, and C. M. Weiss

3 r ent in the reaction mixture at the various pH values. The molar concentrations of thyroxine bound at a constant initial concen- tration of reactants were plotted against pH, and the points obtained are shown in Fig. 5.

3165

t o 123456

r

FIG. 4. Thyroxine binding plotted according to Scatchard (6,7). 0, human serum albumin at pH 7.35; 0, bovine serum albumin at pH 7.35; n , human serum albumin at pH 4.55; 0, acetylated human serum albumin at pH 7.35.

general equation developed by Scatchard (6,7):

r/T = kn - kr (1)

where r is the moles of n-thyroxine bound per mole of protein, T is the concentration of unbound L-thyroxine, n is the theo- retical maximal number of n-thyroxine molecules that can be bound, and k is the intrinsic association constant. A plot of r/T versus r would give a straight line which intercepts kn (as r --t 0) and n (as r/T -+ 0). When interactions between the bound molecules are significant, or if the association constants for the sites differ, the plot will not be linear. The intercepts on the axes are then Zniki and Zni. In Fig. 4, the values of ni and kc for each set of sites may be determined with reasonable precision by fitting a curve to the plot through calculation of T and T for different values of ni and kc from the equation:

n,k,Wl r=C-

1 + kiD'1 (2)

In this study, the curved line shown in Fig. 4 was obtained by assuming that albumin has two thyroxine binding sites, nl = 1, and n2 = 5 with association constants kl = 2.5 X 106 M-l and kp = 0.060 X lo6 M-~, respectively. From these constants, the AFO values at 25’ may be calculated; they were found equal to -8.7 and -6.5 kilocalories per mole, respectively.

Effect of pH-In order to obtain an insight into the role of ionized groups in the binding process, the effect of the pH of the reaction mixture was investigated. To this end, n-thyroxine and human albumin, in an arbitrarily selected molar ratio of approximately 2: 1, were mixed first and then allowed to react with the remaining components of the enzymatic assay system in buffers of identical ionic strength (0.155) but of different pH. To eliminate the effect of pH on the activity of the peroxidase and on the ability of L-thyroxine to affect the rate of the re- action in the absence of albumin, a calibration curve similar to that in Fig. 2B was prepared at each pH. Thus, it was possible

L-Thyroxine Binding at pH 4.56-From an inspection of Fig. 5 it is apparent that at the normal pH of serum (pH 7.3 to 7.4), n-thyroxine binding by human albumin is close to the maximal value that is attained at approximately pH 8. It is also evident that at pH 5 and below, only approximately half as much L-

thyroxine is bound by the albumin. Thus, it was of interest to repeat the experiments illustrated in Fig. 4 at a pH below 5. The binding was therefore studied in a pH 4.55 acetate buffer of an ionic strength of 0.155, and the results were plotted (solid squares) in Fig. 4. From the results it is evident that the straight line shown in Fig. 4 accounts for the experimental data, and that n equals 2 and k is equal to 0.08 x lo6 M-l. This is of the same order of magnitude as kp obtained at pH 7.35 above. Since the assay system is much less sensitive to thyroxine at pH 4.55 than at pH 7.35, there is probably more error in these binding data than in the ones obtained at pH 7.35.

Efect of Ionic Strength-Mixtures of NaH2P04 and Na2HP04 of pH 7.35 but differing in ionic strength were prepared, and the ability of 0.05 pmole of human albumin to inhibit the po- tentiation of the DPNH oxidation by 0.1 pmole of L-thyroxine was determined at several ionic strengths (Fig. 6, soZid circles). A calibration curve was constructed at each ionic strength to determine the effect of ionic strength on the ability of 0.1 pmole of n-thyroxine in the absence of albumin to potentiate the rate of DPNH oxidation (Fig. 6, open circles). From the calibration curve at the appropriate ionic strength, the percentage of the 0.1 pmole of L-thyroxine bound by the 0.05 pmole of albumin was determined. Fig. 6 indicates that L-thyroxine alone is best able to potentiate the oxidation of DPNH by the peroxidase system at low or high ionic strengths. Thus, ionic forces are of great importance in that reaction. The ability of albumin to bind n-thyroxine, with the peroxidase system as a measurement of binding, is sensitive only to low ionic strength. Although the efficacy of n-thyroxine alone to act as a catalyst was quite sensi

r 1.6 i

. . . . ! . 1

4 5 6 7 8 9

PH

FIG. 5. Effect of pH upon thyroxine binding. Peroxidase, HzOz, and DPNH as in Fig. 1. Buffers and amounts of thyroxine _. . . . . . .._ to estimate the amount of unbound and bound n-thyroxine pres- and human albumin used are described in the text.

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from

3166 Thyroxine Binding by Albumin Vol. 236, No. 12

- 0.20 t t I I I 1 t 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

IONIC STRENGTH

FIG. 6. Effect of ionic strength at pH 7.35. Peroxidase, HzOz, and DPNH as in Fig. 1. Description of the buffers and the amounts of thyroxine and human albumin used are described in the text. O-0, slopes (optical density units per minute) of the calibration curves (as Fig. 2B) at different ionic strengths; a---•, per cent of total amount of thyroxine bound at different ionic strengths.

tive to ionic strengths above 0.3, the ability of albumin to bind n-thyroxine was not changed significantly above an ionic strength of 0.3. Thus, it appears that salts can compete with n-thyroxine for the albumin binding sites up to an ionic strength of 0.3, but beyond this, additional salt does not affect the binding signifi- cantly. On the whole, the effect of ionic strength is not great, and no specific information about the binding process can be derived from these data.

Binding by Other Substances-Bovine serum albumin behaved like the human protein (see Fig. 4). Polyvinylpyrrolidone was only approximately one-fourth as effective a n-thyroxine binder at a weight equal to that of albumin, and was not investigated in detail.

Since the purification of the specific thyroxine-binding globu- lin has not yet been achieved, it was not possible to compare the behavior of this protein with that of albumin. However, frac- tions of human serum proteins which have been shown to be rich in thyroxine-binding globulin (8) are available (Cohn Frac- tion IV-4). The ability of this fraction to bind n-thyroxine, as measured by the peroxidase assay, could be accounted for com- pletely by the 27%, by weight, of albumin present in this frac- tion. Normal human plasma that was found to be 0.5 mM in albumin (3.45 g of albumin per 100 ml) behaved as though a solution of pure albumin of this concentration were used. When the results of the experiments with the plasma were plotted as in Fig. 4 (for these calculations, the plasma was treated as though it were a 0.5 mM albumin solution), the points fell on a curve essentially identical to the one in Fig. 4. Somewhat greater scattering of the data was observed with this sensitive method of plotting. Similar results were obtained with the Cohn frac- tion. Thus, by the use of the peroxidase assay system, thyroxine binding by the thyroxine-binding globulin which is present in plasma and in the Cohn fraction cannot be detected. The im- plications of this finding will be discussed below.

Human albumin, to which different amounts of n-thyroxine were added, was subjected to equilibrium dialysis with methyl

orange at pH 7.35. The binding of methyl orange by the thy- roxine-albumin mixtures (1 to 4 moles of n-thyroxine per mole of albumin) was identical to that of albumin without thyroxine. This indicated that n-thyroxine and methyl orange are bound at different sites and do not compete for the same sites.

Human albumin was acetylated with acetic anhydride (9) until six and one-half amino groups remained unacetylated as compared to 56 free amino groups in the native protein. These values were obtained from Van Slyke amino nitrogen determina- tions2 The ability of this acetylated protein to bind n-thyroxine at pH 7.35 was determined, and the results were plotted (open sluures) in Fig. 4. From the results it is evident that acetyla- tion destroys some, but not all, of the thyroxine-binding sites on the albumin molecule. Acetylated albumin behaves at pH 7.35 as does the native protein at pH 4.55.

DISCUSSION

The objective of this study was to determine the nature of the site and the forces involved in the binding of thyroxine to serum albumin. There are many studies with similar general objectives in the literature (lo), but the particular experimental approach herein employed has not been applied in published binding studies. In the present study, the concentration of bound thyroxine was not measured experimentally. It was cal- culated from a knowledge of the total amount of thyroxine added and the amount remaining unbound in the experiment. It was also assumed that the extent to which albumin can inhibit the thyroxine-potentiated peroxidase reaction is a measure of thy- roxine binding by albumin.

Great caution must be exercised in making use of the data herein presented to explain thyroxine binding under physiological conditions: In a typical assay experiment, the total protein con- rentration was 0.117 g per 100 ml, and the iodine level (meas- ured as butanol-extractable iodine) was 865 pg of 1~ per 100 ml. The former is & of the normal concentration found in serum and the latter is 200 times the normal level.

The finding that plasma behaved as a thyroxine binder, as though it contained only albumin, requires elaboration. The capacity of the thyroxine-binding globulin is, on the average, 0.2 pg of thyroxine per ml of plasma (1). In our experiments, 0.1 ml of plasma was used. The thyroxine-binding globulin in this amount of plasma can thus bind no more than 0.000025 pmole of thyroxine. This is too small an amount to be detected when 0.05 pmole of thyroxine is added, and the bulk of this is bound by the albumin. Even though it has been shown (1) that the association constant for the thyroxine-binding globulin is approximately 3000 times (7.9 x 10Q) as great as the one ob- tained in this work for the strong binding site on albumin, the amount of thyroxine that can be bound by the low concentration of the thyroxine-binding globulin in plasma is too small to be detected in the presence of the large amount of albumin by the assay employed, Therefore, no conclusions can be drawn from the data presented as to the manner in which the thyroxine- binding globulin binds thyroxine.

The study of the pH dependence of thyroxine binding indicates that some of the binding sites are affected by the hydrogen ion concentration, and that the amount of thyroxine bound by a given quantity of albumin increases as the pH changes from 5 to 8. Thus, the removal of protons favors thyroxine binding.

2 We are grateful to Dr. Allan L. Grossberg under whose super- vision these determinations were performed.

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from

December 196 1 G. L. Tritsch, C. E. Rathke, N. E. Tritsch, and C. M. Weiss 3167

However, binding still occurs below pH 5. Fig. 4 shows that at pH 4.55 1 molecule of albumin can bind 2 thyroxine molecules. At pH 7.35, a total of 6 thyroxine molecules can be bound: The first one is held relatively firmly, and the remaining 5 are held with forces comparable to the ones that are operative at pH 4.55. This suggests that the assumptions made to calculate the curved line in Fig. 4 do not explain the total picture. The sites that function at pH 4.55 presumably also function at pH 7.35. Five sites with a kz of 0.06 X 106 M-l were assumed in the cal- culation of the curve in Fig. 4. There must therefore be three sites with a similar k that are not functional at pH 4.55. This presents a possible picture at pH 7.35 of one strong pH-dependent site, three pH-dependent weak sites, and two pH-independent weak sites. All these sites would have to be taken into account to calculate a theoretical curve to explain the variation of r with pH. This would involve the fitting of so many parameters that the results would not be convincing. Therefore, no at- tempt was made to calculate a theoretical curve to account for the data presented in Fig. 5, and, as a result, no conclusions are being drawn from the experimental data as to what chemical groups are involved in the binding process. The fact that acet- ylation destroys the single strong pH-dependent binding site (Fig. 4) eliminates imidazole groups as being part of this bind- ing site since acetyl imidazole compounds are rapidly hydrolyzed (11) even in neutral solution.

SUMMARY

The ability of human serum albumin to bind thyroxine has been measured by determining the inhibition by albumin of the thyroxine-accelerated oxidation of reduced diphosphopyridine nucleotide by hydrogen peroxide in the presence of horse-radish peroxidase. A maximum of approximately 6 molecules of thy- roxine can be bound per molecule of albumin. The data pre- sented suggest three different kinds of thyroxine binding sites on the albumin molecule.

The first site holds 1 thyroxine molecule approximately 40 times as tightly as the other sites bind the remainder. The data may be interpreted to indicate a strong pH-dependent binding site (k = 2.5 x 106 M-I), three pH-dependent weak sites, and two pH-independent weak sites. A study of the pH dependence of the binding revealed that binding decreases but is not abolished as the pH is lowered from 7.35 to 5. Studies with acetylated albumin eliminated the imidazole group as the ionizable group at the pH-dependent strong site.

Acknowledgments-The authors are grateful to their many colleagues at the Roswell Park Memorial Institute for their helpful criticism and suggestions during this work. Particular thanks are due to Dr. Avery Sandberg for his many suggestions and encouragement during all phases of this work and to Dr. Gabor Markus for his help and advice in interpreting some of the results.

REFERENCES

1. ROBBINS, J., AND RALL, J. E., Physiol. Revs., 40,415 (1960). 2. KI.EBAN&F; S. J., J. l&01. &em.; 234, 2437,.2480 (1959). 3. KLOTZ, I. M., in H. NEURATH AND K. C. BAILEY (Editors),

The proteins, Vol. I, Part B, Academic Press, Inc., New York, 1953, p. 780.

4. GORNALL. A. G.. BARDAWILL. C. J.. AND DAVID. M. M.. J. Biol. &em., li7, 751 (1949): ’

5. BRACKEN, J. S., AND KLOTZ, I. M., Am. J. Clin. Pathol., 23, 1055 (1953).

6. SCATCHARD, G., Ann. N. Y. Acad. Sci., 61,660 (1949). 7. EDSALL, J. T., AND WYMAN, J., Biophysical chemistry, Vol. I,

Academic Press, Inc., New York, 1958, p. 617. 8. FREINKEL, N., DOWLING, J. T., AND INGBAR, S. H., J. Clin.

Invest., 34, 1698 (1955). 9. FRAENKEL-CONRAT, H., BEAN, R. S., AND LINEWEAVER, H.,

J. Biol. Chem., 177, 385 (1949). 10. KLOTZ, I. M., in H. NEURATH AND K. C. BAILEY (Editors),

The proteins, VoZ. I, Part B, Academic Press, Inc., New York, 1953, p. 727.

11. STAAB, H. A., Chem. Ber., 89, 1927 (1956).

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from

George L. Tritsch, Charles E. Rathke, Norma E. Tritsch and Catharine M. WeissThyroxine Binding by Human Serum Albumin

1961, 236:3163-3167.J. Biol. Chem. 

  http://www.jbc.org/content/236/12/3163.citation

Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/236/12/3163.citation.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on July 11, 2019http://w

ww

.jbc.org/D

ownloaded from