Binding of d -glucose to insulin

Biochimica et Biophysica Acta, 386 (1975) 603-607

© Elsevier ScientificPublishing Company, Amsterdam-- Printed in The Netherlands BBA 36996 BINDING OF D-GLUCOSE TO INSULIN

P.

ANZENBACHERa and V.

KALOUS b

aDepartment of Biochemistry and bDepartment of Physical Chemistry, Charles University, Albertov 2030, 128 40 Prague 2 (Czechoslovakia)

(Received October 2nd, 1974)

SUMMARY Binding of D-glucose to insulin has been studied by equilibrium dialysis. The binding is not very specific and probably takes place in two steps. The average amount of glucose molecules bound per insulin molecule is eight, two molecules in the first and six during the second step of binding. The intrinsic binding constants for both steps are almost the same (6-102 M -1 and 1.10 a M -1) which can be explained by assuming: (1) that after binding of the first two molecules a conformational change of insulin occurs which facilitates the binding of the next six molecules of o-glucose; or (2) that in the second step of binding the glucose binds to hydrophobic regions which are unmasked by dissociation of the insulin dimer. Using a three-dimensional model of the insulin molecule areas of the protein molecule where binding of glucose can occur were selected. The glucose-binding site very probably involves the area at the insulin surface where most of the invariant and modification-selective residues are present.

INTRODUCTION The binding properties of a number of proteins have been studied during the last 20 years with the aim of a better understanding of their biological function [1, 2]. Small molecules and ions (ligands) bind to proteins with a greater or lesser specificity. A mathematical description of binding [2, 3] includes the evaluation of the average number of ligand molecules bound per protein molecule, ~, and the intrinsic binding constant, k. It is an entirely different question at which site of the protein molecule the binding takes place and what kind of interaction may be involved. In the case of the insulin molecule, one can compare experimental results with information provided by crystallographic studies [4]. Moreover, it is known that conformational changes of insulin in solution are negligible and its tertiary structure remains unchanged [5, 6]. In the present communication, binding of glucose to insulin is studied.

604 MATERIALS AND METHODS Materials

Insulin was obtained from L6~iva, Prague (Lot No. 62961/4/70; 24.1 units/mg); D-glucose was obtained from Spofa, Prague, and was of reagent-grade quality. All other chemicals were of reagent-grade quality and obtained from Lachema, Brno. o-[U-~4C]Glucose was from the Institute for Research, Production and Uses of Radioisotopes, Prague. Methods

Binding was measured by equilibrium dialysis in highly polished Plexite cells, similar to those described by Furlong et al. [7]. One half-cell contained 80 /~1 of 1.10 -4 M insulin in an ammonia buffer of pH 7.9; the other half-cell was provided with 80/~1 glucose solution in the same buffer and they were separated by an LKB/ 6390-1 dialysis membrane presoaked in NaHCO3. The cells were rotated for 24 h at 20 4- 1 °C, after which duplicate or triplicate samples of 10/al were withdrawn from each half-cell. At glucose concentrations lower than 5.10 -2 M the samples were dried on aluminium planchets and counted in a Tesla NZQ 717-T Geiger-Miiller counter. With glucose concentrations higher than 5.10 -2 M, the samples were counted in a toluene/dioxane scintillation liquid in a Nuclear Chicago Mark I scintillation spectrometer. No penetration of insulin through the membrane was observed. Adsorption of protein on the plexite as well as quenching due to presence of insulin in the sample were negligible. Difference spectrum measurements were recorded at room temperature with a Specord UVVIS spectrophotometer (Carl Zeiss, Jena) in combination with a EZ 7 type pen recorder (Labora, Prague). A pHm4e type apparatus (Radiometer, Copenhagen) was used for measuring the pH. RESULTS AND DISCUSSION Results of equilibrium dialyses of insulin (1.10 -4 M with respect to the monomer) against different concentrations of D-glucose are shown in Fig. 1. Binding of D-glucose to the insulin molecule starts to be pronounced at a concentration of free glucose of about 1.10 -a M. The average number of glucose molecules bound per insulin monomer, ~, reaches the value of eight at a free glucose concentration of about 5.10 -2 M. The course of the experimental curve suggests that binding occurs in two steps. The values of the intrinsic binding constants (kx, k2) and the number of binding sites which belong to each step (nl, n2) have been obtained by mathematical analysis (as the best fit of experimental data for the case of two independent classes of binding sites [8, 9]) and are nl ---- 2 and n2 -----6, respectively; the values of the binding constants are approximately the same (kx -----6.102 M -x, k2 ----- 1.103 M-~). The fact that D-glucose binds to insulin in two steps with almost the same affinity can be explained by the hypothesis that binding of the first two molecules of D-glucose facilitates binding of the following six. This is most probably due to a change of the insulin conformation after binding of the first two molecules of glucose (which allows the next six glucose molecules to bind). Another alternative explanation of the sigmoidal course of the curve is based on the fact that binding of ligands to proteins which

605

8 7 6 5 4 3 2 1 0 -4.5

-4.0

I

-3.0

-21o

Ioo c

I-.] -1"°

010

1.0

Fig. 1. Binding of D-glucoseto insulin. Dependence of the average number of bound D-glucosemolecules per insulin monomer (~) on the concentration of free D-glucose (c). The insulin concentration was 1-10-4 M, pH 7.9. exist in several polymeric forms results in various shapes of sigmoidal binding curves [10]. Considering that insulin in solution at neutral pH forms aggregates which are in equilibrium with the monomer [11] the shape of experimental binding curve can be explained by unmasking the hydrophobic region between monomers in the dimer molecule (B 16 tyrosine, B 26 tyrosine, B 25 phenylalanine; see Table I) during the second step of binding. TABLEI POTENTIAL CARBOHYDRATE-BINDING SITES AT THE INSULIN MOLECULE The additional amino acid residues are supposed to lie in the neighbourhood of the main (central) residue and may be involved in the binding due to their functional groups (e.g. -NH2, -CH(CHa)2, -OH). The amino acid residues are numbered in accordance with Blundell et al. [4]. Area No.

Main residue

Additional residues

I II 111 IV V VI VII

A 19 Tyr A 14 Tyr B 26 Tyr B 16 Tyr B I Phe B 25 Phe B 5 His

A1 Gly, A 18Asn, B27Thr A 12 Ser, B 3 Asn B 11 Leu, B 12 Val, B 15 Leu, B 28 Pro B 13 Glu, B 17 Leu, B 24 Phe B 4 Gin, B 2 Val, B 17 Leu A 21 Asn B 4 Gln, A 8 Thr, B 7 Cys

The ultraviolet difference spectrum of insulin obtained by adding D-glucose is shown in Fig. 2. The "tandem cell" arrangement [12] has been used with four cells filled with the buffer, a solution of insulin and o-glucose, a solution of D-glucose, and a solution of insulin alone in concentrations corresponding to that used in the equilibrium dialysis experiments. The changes in the spectrum at about 250 nm are consistent with the perturbation of the cystine chromophore; in the region of 260 nm the changes indicate perturbation of phenyl groups and those between 270 and 290 nm

606 0.02 ,dA

0.01

-0.01 240

i

I

,

I 260

I

,

i

A [nm'~

I 280

i

I

I 300

Fig. 2. Difference spectrum of insulin. Four cells ("tandem cells" system [12]) were used; 7.10 -2 M D-glucose, 1.10 -4 M insulin, pH 7.9. are due to interaction of the tyrosine hydroxyl group in the protonized and the dissociated form, respectively [13]. The value o f p H 7.9 used in this work is near to that used by Yu et al. [6], who showed that insulin retains its structure in solution of pH 8.3. The ammonia buffer was chosen for its pH which differs only slightly from that of most extracellular fluids (e.g. blood plasma 7.35-7.45, pancreatic juice 7.0-8.0, liver bile 7.4-8.0; all values taken from White et al. [14]). Moreover, the ammonia buffer contains no components which can compete with glucose in binding to the protein. In comparison with a Tris buffer or other commonly used buffers an ammonia buffer contains only relatively small particles, whose interaction with insulin is very probably of a character different (i.e. ionic or strongly polar) from that expected in the case of glucose. It is known from the binding studies of carbohydrates to concanavalin A [15] that the binding site very probably contains two tyrosyl, one aspartyl and one arginyl I TYR A 19

l

"iT TYR A14

I / "V" PHE B1

ALA B30 ~

]E TYR B 2 6

% TYR B16

Fig. 3. Simplified schematic picture of an insulin monomer with indicated A- and B-chains and the potential carbohydrate-binding areas.

607 residue. Having used a three-dimensional model of the insulin dimer [16] we tried to select areas of the insulin molecule where binding of glucose can possibly occur (see Table I and Fig. 3). The "suspected" Area I contains tyrosine A 19 and additional residues which, with the surrounding oxygen atoms from peptide bonds, form a large site which is rather polar. Areas I I - V I do not markedly differ in their hydrophobicity. Probably the most hydrophobic region is around tyrosine B 26 (Area III). The importance of the region around tyrosine A 19, and, moreover, around the terminal A 21 asparagine and the adjacent non-polar B 24, B 25, B 26, B 12, B 16 has been emphasized by Blundell et al. [4] in their review of the structure of insulin. This area is formed by a large number of residues, which are invariant with respect to various biological species and which are sensitive to modification or deletion. Perturbation of the aromatic amino acids as well as that of the cystine residues (which are at the top and the b o t t o m of this region) due to binding of D-glucose has also been demonstrated by the difference spectrum (Fig. 2). The binding occurs at concentrations of free D-glucose which are near to those in the blood [17] (90 rag/100 ml, i.e. 5.10 -a M). The binding of other small molecules (especially amino acids) to insulin and the investigation of interactions involved is an object of further study. REFERENCES 1 Edsall, J. T. and Wyman, J. (1958) in Biophysical Chemistry, Vol. 1, Chapter 11, pp. 591-660, Academic Press, New York 2 Steinhardt, J. and Reynolds, J. A. (1969) Multiple Equilibria in Proteins, Academic Press, New York 3 Laiken, N. and N6methy, G. (1970) J. Phys. Chem. 74, 4421-4431 4 Blundell, T. L., Dodson, G. G., Dodson, E., Hodgkin, D. C. and Vijayan, M. (1971) in Recent Progr. Horm. Res. (Astwood, E. B., exl.), Vol. 27, pp. 1-40 Academic Press, New York 5 Mercola, D. A., Morris, J. W. S. and Arquilla, E. R. (1972) Biochemistry 11, 3860-3874 6 Yu, N. T., Liu, C. S. and O'Shea, D. E. (1972) J. Mol. Biol. 70, 117-132 7 Furlong, C. E., Morris, R. G., Kandrach, M. and Rosen, B. P. (1972) Anal. Biochem. 47, 514-526 8 Edsall, J. T. and Wyman, J. (1958) in Biophysical Chemistry, Vol. I, p. 613, Academic Press, New York 9 Steinhardt, J. and Reynolds, J. A. (1969) Multiple Equilibria in Proteins, p. 15, Academic Press, New York 10 Nichol, L. W., Jackson, W. J. H. and Winzor, D. J. (1967) Biochemistry 6, 2449-2456 11 Pekar, A. H. and Frank, B. H. (1972) Biochemistry 11, 4013--4016 12 Donovan, J. W. (1969) in Physical Principles and Techniques of Protein Chemistry (Leach, S. J., ed.), Part A, pp. 162-164 Academic Press, New York 13 Donovan, J. W. (1969) in Physical Principles and Techniques of Protein Chemistry (Leach, S. J., ed.), Part A, p. 116 Academic Press, New York 14 White, A., Handier, P. and Smith, E. L. (1973) in Principles of Biochemistry, 5th edn, p. 910, McGraw-Hill, New York 15 Hardman, K. D. and Ainsworth, C. F. (1973) Biochemistry 12, 4442--4448 16 Lewitov~i, A. (1971) Thesis, Charles University, Prague 17 White, A., Handler, P. and Smith, E. L. (1973) in Principles of Biochemistry, 5th edn, p. 480, McGraw-Hill, New York