Electrochemical Studies of the Interfacial Behavior of Insulin

JOURNAL OF COLLOID AND INTERFACE SCIENCE ARTICLE NO.

184, 449–455 (1996)

0640

Electrochemical Studies of the Interfacial Behavior of Insulin STEPHANIE M. MACDONALD AND SHARON G. ROSCOE 1 Chemistry Department, Acadia University, Wolfville, Nova Scotia, Canada B0P 1X0 Received January 17, 1996; accepted July 22, 1996

The interfacial behavior of insulin and chain A and chain B of insulin was investigated at the platinum electrode in a phosphate buffer, pH 7.0, using cyclic voltammetry. The enthalpy of adsorption, DHADS , calculated from a linear van’t Hoff relationship over the temperature range 299 to 333 K gave values of 022 { 1, 017 { 1, and 09 { 1 kJ mol 01 for insulin, chain B and chain A, respectively. Above these temperatures denaturation of insulin occurs, and for all three molecules the surface adsorption measured by the surface charge densities showed an immediate decrease followed by a slight increase. The surface concentrations of insulin of 2.9 { 0.2 mg m02 at 299 K and 3.2 { 0.3 mg m02 at the physiological temperature of 310 K agreed well with the calculated value determined from geometrical dimensions. Similar calculations from experimental surface charge densities for chain A and chain B indicated that a more efficient packing prevailed with the individual polypeptides. From a consideration of the mechanism of adsorption based on cyclic voltammetric measurements, an estimation of the number of carboxylate groups on insulin was determined to be 6 { 2, which agrees with the known number of acidic residues on the protein. Similar calculations for chain A gave 2 { 0.4, which indicates that this peptide remains as a monomer in the phosphate buffer. However, the value obtained for chain B under similar experimental conditions was 6 { 2, indicating that this peptide appears to dimerize in the phosphate buffer. Dimer formation of insulin is known to occur through hydrophobic interactions and four hydrogen bonds between the B chains. q 1996 Academic Press, Inc.

Key Words: insulin; cyclic voltammetry; adsorption; interfacial behavior.

INTRODUCTION

Recently there has been interest in the mechanism of protein interactions with surfaces of artificial and biocompatible materials used in the areas of biosensors and medical implant devices, as well as the methods to control these processes (1, 2). Studies have been made on the adsorption of insulin to infusion systems and infusion containers in an attempt to understand the interaction with hydrophilic and hydrophobic surfaces in order to predict accurate amounts of delivered 1

To whom correspondence should be addressed.

insulin (3–7). This has also been important in the design of the artificial pancreas because of the need to preserve the native constitution of the protein (8). Adsorption of insulin appeared to occur more readily at hydrophobic polydimethylsiloxane (3) or Teflon (9) surfaces than on hydrophilic quartz (3) or polyacrylamide (9) surfaces. A pronounced initial drop in delivered insulin occurred when polyethylene tubing was used, but steady-state adsorption resulted after 3 h of continual use (4). More insulin was found to adsorb to a dialysis bag at 310 K than at 297 K (6). Addition of human albumin to the infusion reduced the adsorption of insulin to the surfaces of the infusion bottles and systems (7). Inactivation of insulin also occurs by the formation of fibrils, which has been suggested to occur through hydrophobic interactions between the monomers (10). Fibril formation can be induced by heat (10), physical means such as shaking (11) or using high shear rates (12), denaturing chemicals such as urea (10), and the interaction with hydrophobic materials such as silicon rubber and polypropylene (13). However, adsorption does not always result in denaturation, as insulin adsorbed on controlled pore-glass surfaces for 3 months was found to be almost fully active after desorption (14). The bovine insulin molecule consists of two polypeptide chains designated as A and B, and joined by two disulfide linkages (15–17). Chain A, with 21 amino acids, is composed of two acidic residues and an internal disulfide linkage in addition to the two cystine residues which link with chain B. Chain B is the larger of the two polypeptides containing 30 amino acids, of which two are also acidic residues. Thus, the insulin molecule, formed by the disulfide linkages between the two chains, contains a total of six carboxylate groups from the two acidic residues and carboxylate end group on each chain. In aqueous solutions association of insulin molecules may occur as a result of hydrophobic interactions and hydrogen bonding (18–20). A large proportion of free molecules exist in solution at low concentrations ( õ16 mM, 0.1 mg ml 01 ) and in the absence of zinc ions (21, 22). At relatively high concentrations the molecules exist predominantly as dimers and hexamers. The insulin dimer is a compact, oblong unit, similar in size and overall shape to the enzyme lysozyme (15). Only

449

AID

JCIS 4500

/

6g18$$$761

11-20-96 15:05:14

0021-9797/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

coida

AP: Colloid

450

MACDONALD AND ROSCOE

the B chain residues participate in the contacts between the two molecules. These contacts involve residues from one side of the B chain a helix, B12 valine and B16 tyrosine, and also the last eight residues of the extended B chain. The C terminal chains of the two molecules run closely antiparallel to one another and interact through hydrogen bonds, as in a b pleated sheet, between the peptide CO and NH groups, B24 and B26. Thus, dimer formation includes a number of the hydrophobic groups on the surface of the individual molecules. This chain arrangement can have perfect twofold symmetry, however, departures from symmetrical arrangement result from packing of the residues. Three dimers interact in the presence of zinc ions to form hexamers (15). The groups attached directly to the zinc ions have been identified as the B10 histidines. In the electron density map, six insulin molecules are very intricately packed in a hexamer around the two zinc ions. Although the insulin hexamer in isolation is very smooth around its circumference, it is deeply grooved within its upper and lower surfaces. The grooves are formed between the projecting A chain residues of the three insulin dimers which compose the hexamer. The grooves are lined with the functionally active residues, histidine, tyrosine, and serine. The A chain region is six times repeated over its smooth circumference, and therefore any changes made within the A chain are known to affect the activity of insulin. When the hexamer passes through the membrane into dilute solution, it dissociates into dimers and monomers, at each stage opening for possible action with other strategic residues and in particular with nonpolar residues. Insulin in plasma would be a monomer under physiological conditions if the physiological concentration of insulin is around 10 011 mol liter 01 (15). A number of investigations have been previously carried out with a variety of proteins in order to determine whether the application of the technique of cyclic voltammetry could give information regarding the interfacial behavior of the proteins as a function of their structural conformation in the bulk solution. Proteins studied in this laboratory have been b-lactoglobulin ( 23, 24 ) , lysozyme ( 24 ) , ribonuclease ( 24 ) , a-lactalbumin ( 25 ) , cytochrome c ( 26 ) , myoglobin ( 26 ) , hemoglobin ( 26 ) , and microperoxidase ( 26 ) . The present study therefore extends these investigations to determine the mechanisms of adsorption of insulin, and individually the chain A and the chain B of insulin. The interfacial behavior of these polypeptides has been studied at the platinum electrode surface as a function of temperature of the bulk solution. EXPERIMENTAL

Reagents and Solutions Solutions of bovine insulin (Sigma Chemical Company, product number I 5500, zinc content Ç0.5%, MR 5733.5),

AID

JCIS 4500

/

6g18$$$762

11-20-96 15:05:14

bovine chain A of insulin (Sigma Chemical Company, product number I 1633, MR 2531.6), and bovine chain B of insulin (Sigma Chemical Company, product number I 6383, MR 3495.9) were dissolved in 0.05 mol liter 01 phosphate buffer (pH 7.0) prepared using anhydrous monobasic potassium phosphate (KH2PO4 ) (Sigma Chemical Company, cell culture tested) and sodium hydroxide (prepared from standardized BDH Chemical Company concentrate). Conductivity water (Nanopure, resistivity 18.2 megohm cm01 , Barnstead/Thermolyne, 4 module unit) was used in the preparation of all aqueous solutions. Electrochemical Cell and Electrodes Three-compartment all-glass electrochemical cells were constructed with two glass-sleeved stopcocks as two of the separate compartments. The working and counter electrode compartments were purged with nitrogen gas (Linde, commercial purity 99.7%) which was deoxygenated by passage through a copper furnace at 573 K. This also provided a well-mixed bulk solution. Electrodes of high-purity platinum wire or mesh (99.99%, Johnson, Matthey and Mallory) were degreased by refluxing in acetone, sealed in soft glass, electrochemically cleaned by potential cycling in 0.5 mol liter 01 sulfuric acid and stored in 98% sulfuric acid. The real surface area of the platinum electrode was obtained from the charge under the hydrogen underpotential deposition peaks (27). Saturated calomel reference electrodes were constructed by a standard procedure (28). Their potentials were checked frequently against a standard hydrogen electrode and compared with the literature value. The saturated calomel electrodes were found to be reproducible to within {1 mV. Methods Cyclic voltammograms were obtained using a Hokuto Denko potentiostat (Model HA-301) and a Hokuto Denko function generator (Model HB-111) and recorded on an Allen Datagraph, Inc. X-Y recorder-plotter (Model 720M). The sweep rate used throughout was 500 mV s 01 , since diffusion processes predominate at low scan rates and adsorption processes predominate at high scan rates (29). The current and potential response was monitored by an ammeter ( Keithley, 195 System DMM) and a voltmeter (Keithley 126 Autoranging multimeter). The electrochemical cell containing the phosphate buffer and the solutions were placed in a constant temperature bath fitted with a Jubalo P temperature regulator for one half hour prior to the start of each experiment. During this time and throughout the experiment, the phosphate buffer in the electrochemical cell was purged with deoxygenated nitrogen to ensure the removal of dissolved oxygen and provided a wellmixed bulk solution. Oxygen removal was verified by the profile of the cyclic voltammogram recorded at intervals before the start of each experiment. When steady state condi-

coida

AP: Colloid

451

INTERFACIAL BEHAVIOR OF INSULIN

tions were attained, a series of cyclic voltammograms was recorded from 00.80 V to a specific anodic end potential which was changed in a decreasing stepwise manner from 1.0 to 0.10 V in increments of 0.10 V. These cyclic voltammograms were recorded for the phosphate buffer solution before the addition of any protein. The protein was then added in specific aliquots giving stepwise increasing concentrations ranging from 7 1 10 04 to 0.2 g liter 01 in the bulk solution. Once steady state conditions were attained, which was verified by continuously identical tracings after potential cycling for about 4 min, the cyclic voltammograms were recorded again for a series of anodic end potentials. The areas of the anodic oxidation and oxide reduction regions of the cyclic voltammograms were determined by scanning the entire voltammogram with a Hewlett Packard Scan Jet 11cx scanner. The figure was then digitized using Unscanit software (Silk Scientific Software), and areas of the specific regions were determined by a computer integration program. The surface charge density, QADS , resulting from the deposition of the protein on the electrode following additions of specific aliquots of analyte was determined from the difference between the integrated areas of anodic oxidation and oxide reduction in the presence of the adsorbing species. The small surface charge density which resulted from the phosphate buffer in the absence of protein was subtracted from that measured for the protein and the remainder was attributed to the protein adsorption, p

p

QADS Å (Q Oo 0 Q Or ) 0 (QOo 0 QOr ),

[1]

where QOo is the anodic oxidation charge density in phosphate buffer, QOr is the oxide reduction charge density in p phosphate buffer, Q Oo is the anodic oxidation charge density p in the presence of the protein, and Q Or is the oxide reduction charge density in the presence of the adsorbing species. Cyclic voltammograms were recorded after each aliquot of protein was added to the well-mixed buffer solution in the electrochemical cell after steady state conditions were achieved as indicated by continuously identical tracings (ca. ú3 min). An example of a change in the cyclic voltammetric profile to anodic end potential 0.50 V is shown in Fig. 1 for an addition of insulin to give a concentration of 0.105 g liter 01 . The anodic oxidation region (Oo ) shows a blocking of oxide formation (f ) and the anodic oxide reduction region ( Or ) is therefore diminished (F ). The reproducibility of determining the anodic oxidation and anodic reduction areas from the same cyclic voltammogram tracing using the Unscanit software and the computer integration program was within {0.005%. The reproducibility of determining areas of different cyclic voltammogram tracings from different experiments for phosphate buffer at anodic end potential 0.45 V was within {3%. The standard

AID

JCIS 4500

/

6g18$$$762

11-20-96 15:05:14

FIG. 1. Cyclic voltammogram of 0.1 mol liter01 phosphate buffer (dashed line) and the cyclic voltammogram after addition of insulin to give a concentration of 0.105 g liter 01 (solid line).

deviation of the surface charge densities of the plateau values for protein adsorption were found to be {6%. From the surface charge densities, the surface concentration, G in mg m02 , was calculated using G Å QADS M/nF,

[2]

where M is the molar mass, n is the total number of electrons transferred per protein molecule, and F is the Faraday constant. The concentration of free sulfhydryl groups, as a measurement of the accessibility of the sulfhydryl groups in the native and denatured protein, was determined under conditions of varying temperature using 5,5 *-dithiobis(2-nitrobenzoate) or DTNB (Ellman’s reagent) (30, 31). The procedure was modified slightly in order to determine the free sulfhydryl groups under the present experimental conditions with phosphate buffer and in the absence of EDTA and denaturing solvents. The solutions were held at the experimental temperature for 30 min. DTNB was then added and allowed to react for an additional 30 min before absorbance measurements were made at 412 nm using a Beckman DU65 UV/vis spectrophotometer. Calculations of concentrations were made using the value of e412 Å 13600 liter mol 01 cm01 (30, 31). RESULTS AND DISCUSSION

The surface charge density, QADS , was measured versus the concentration of insulin at 299 K at pH 7.0 with anodic end potentials which normally correspond to a monolayer of OH in aqueous solution at these temperatures (Fig. 2). Measurements of protein adsorption at these low anodic end potentials allow a comparison of results with other techniques. As the concentration of insulin was increased in the bulk solution, the surface charge density increased to a pla-

coida

AP: Colloid

452

MACDONALD AND ROSCOE

FIG. 2. Surface charge density as a function of concentration of insulin: ( s ) 299 K, ( L ) 310 K, ( n ) 333 K.

teau level of Ç60 mC cm02 . This experiment was repeated at the physiological temperature of 310 K and at 333 K. The surface charge densities increased and reached plateau values of Ç65 and 156 mC cm02 , respectively (Fig. 2). Two large aliquots of insulin were added to a fresh solution of phosphate buffer at 299 K to determine the effect that a large protein addition to the phosphate buffer might have on the surface charge density. An addition to give a concentration in the bulk solution of 0.0554 g liter 01 gave a surface charge density of 56.3 mC cm02 , and a second large addition to give a total concentration of 0.170 g liter 01 gave a surface charge density of 56.0 mC cm02 . These values agreed favorably with the average plateau value of 55.7 { 3 mC cm02 obtained with small incremental additions of protein (Fig. 2). In order to compare the behavior of the insulin protein with the two segments which compose the molecule, similar experiments were made separately with chain A and with chain B of insulin. In two separate experiments, the surface charge density measurements with chain B gave a plateau in surface charge density as the concentration of the protein segment in the bulk solution was increased (Fig. 3). Both plateau values were very close to 100 mC cm02 and within experimental uncertainty of {5%. The two samples of chain

FIG. 3. Surface charge density as a function of concentration: (a) at 299 K, ( s ) insulin, ( n ) ( m ) chain A insulin, ( h ) ( j ) chain B insulin; (b) at 333 K, ( l ) chain A insulin, ( L ) chain B insulin.

AID

JCIS 4500

/

6g18$$$762

11-20-96 15:05:14

FIG. 4. The effect of temperature on the surface charge density: ( s ) insulin, ( n ) chain A insulin, ( h ) chain B insulin.

B were taken from different sample vials from the commercial distributor. When similar experiments were made with chain A, the response was quite different. A very large surface charge density resulted which approached plateau values at Ç220 mC cm02 for the two runs with different samples. Separate runs were also made at 333 K, which gave surface charge densities of 324 and 216 mC cm02 for chain A and chain B, respectively. In order to compare the effect of temperature on the surface charge density resulting from surface adsorption, plateau values obtained from the additions of small increments of protein in separate experiments at 299, 310, and 333 K (Figs. 2 and 3) were used in a plot of ln (QADS ) versus 1/T to show the van’t Hoff relationship in the initial electron transfer equilibrium process (Fig. 4). The results at 323, 333, 343, and 353 K were obtained from the experiment at 310 K following the last aliquot of protein, and then the temperature of the bulk solution was increased in a stepwise manner from 310 to 353 K. After the temperature in the electrochemical cell had equilibrated for 30 min at each temperature, the cyclic voltammogram was recorded. The two values at 333 K for each insulin, chain A and chain B, showed the agreement obtained between the two types of experiments. At each temperature the surface charge densities were found to increase in the order insulin, chain B, chain A, with chain A showing the highest value over the temperature range 299 to 353 K. However, with increasing temperature the enthalpy for adsorption, DHADS , calculated from the slopes of the van’t Hoff relationship was similar for insulin and chain B (i.e., 022 { 1 kJ mol 01 and 017 { 1 kJ mol 01 , respectively) compared with the value for chain A of 09 { 1 kJ mol 01 . A decrease in surface charge density was observed when the temperature of denaturation of insulin was reached between 333 and 343 K. At 353 K, the surface charge density increased slightly for the protein segments and more substantially for the whole protein. Insulin has six sulfhydryl groups which are all dimerized in the native structure. Experiments were made to determine

coida

AP: Colloid

INTERFACIAL BEHAVIOR OF INSULIN

FIG. 5. Reaction of insulin with Ellman’s reagent as a function of time for the temperatures: ( s ) ( l ) 299 K, ( n ) 310 K, ( , ) 323 K, ( h ) 333 K, ( m ) 343 K, ( j ) 353 K.

if, over the temperature range studied, reduction of the disulfide bonds occurred, particularly at the temperature of denaturation. The formation of a free sulfhydryl group was monitored as a function of temperature using Ellman’s reagent and spectroscopic measurements. The results for the insulin molecule are shown in Fig. 5 and Table 1. From the percentage of free sulfhydryl, the number of disulfide bonds in the reduced state was calculated as 15% at 299 K. At a physiological temperature of 310 K the result showed a decrease to only 2% for the number of reduced disulfide bonds. As the temperature was increased to the range 323 to 343 K, the number of reduced disulfide bonds dropped to 0. At 353 K, which is above the temperature for denaturation of the protein, it appeared that after prolonged heating at this temperature, 62% of the disulfide bonds underwent reduction. This may account for a much more open structure which resulted in an increased surface charge density when the temperature was increased to 353 K, following the decrease observed near the denaturation temperature (Fig. 4). In order to be complete, measurements were also made with samples of chain A and chain B at 299 K, and these results are compared with those obtained with insulin (Fig. 6). Since the cysteine residues were oxidized to sulfonic acid functional groups, negligible absorbance was observed, as was expected. TABLE 1 Percentage of Free Sulfhydryl of Insulin

AID

Temperature (K)

Percentage of reduced disulfide bonds ({2%)

299 310 323 333 343 353

15 2 0 0 0 62

JCIS 4500

/

6g18$$$762

11-20-96 15:05:14

453

FIG. 6. Reaction of chain A and chain B insulin with Ellman’s reagent as a function of time at 299 K: ( s ) ( l ) insulin, ( n ) chain A insulin, ( h ) chain B insulin.

The surface concentrations ( G ) were calculated from the surface charge densities of insulin, chain A and chain B measured to anodic end potential 0.50 V, which correspond to a monolayer of OH in aqueous solutions. Surface concentrations resulting from adsorption of insulin at 299 K from a phosphate buffer, pH 7.0, on a platinum electrode gave a plateau value of 2.9 { 0.2 mg m02 . This value agrees well with the reported surface concentration value of 2.7 mg m02 calculated for a hexagonally close-packed monolayer of hexamers for the adsorption of insulin assuming end-on adsorption of hexagonally packed cylinders with a height of 3.5 nm, a diameter of 5.0 nm, and a molar mass of the unit of 34 800 (32). Razumas et al. (33) reported a surface concentration of 1.27 mg m02 using ellipsometry measurements on Pt. Insulin adsorption was observed over the potentials of 00.40 to /0.40 V vs SCE. At the lower potential, adsorption of insulin was 0.6 mg m02 but was observed to diminish to essentially zero at 00.2 V, which is in the double-layer region for platinum. The adsorption then continued to increase with increasing anodic potential to a value of 1.55 mg m02 at /0.4 V. Higher potentials were not reported. From the present experiments, the surface concentration at 299 K was also calculated to give 0.312 molecules nm02 or 3.20 nm2 molecule 01 . From the geometry of the molecule in the hexamer conformation with a reported diameter of 5.0 nm (15), the adsorbed hexamer would give a surface area of 19.6 nm2 molecule 01 , or for a single monomer in the same position, a value of 3.26 nm2 molecule 01 . This value agrees very well with the value obtained from electrochemical measurements in the present study. The insulin used in the present study contained 0.05% Zn, or about 1 Zn/2 insulin molecules. The hexamer state is stabilized with two Zn /2 ions, which makes the aggregate even more hydrophilic (16). At physiological temperature, 310 K, a slightly higher surface concentration of 3.2 { 3 mg m02 was obtained. From the experimental surface charge densities, the surface concentration of insulin chain A was calculated to be 8.7 mg m02 , which gave a surface area per molecule of

coida

AP: Colloid

454

MACDONALD AND ROSCOE

0.48 nm2 molecule 01 . Similarly for chain B, the surface concentration of 4.9 mg m02 gave a calculated surface area per molecule of 1.18 nm2 molecule 01 . Both values for surface area per molecule were low, which suggests a more efficient packing of each of the individual polypeptide chains on the electrode surface than when the two chains are bound together to form insulin monomers, which gave a value of 3.20 nm2 molecule 01 . Previous studies have suggested that the adsorption of the proteins at the platinum electrode surface results from the interaction of the carboxylate functional groups of the acidic amino acid residues (23, 24). These conclusions were based on kinetic measurements of the electron transfer reaction mechanisms of amino acids (34, 35). The initial steps of adsorption and electron transfer process may be represented by the equation (13)

and taking natural logarithms ) can then be represented as ( 23 )

P / nM r P(M)n / ne, where P represents the protein adsorption on a metal (M) electrode with n representing the number of sites for carboxylate interaction with the metal surface accompanied by the transfer of a total of n electrons. The second step, involving decarboxylation is represented as P(M)n r P * (M)n / nCO2 , where P * represents the modified protein. This step has been determined to be rate limiting. The rate expressions for the forward and reverse reactions for the first step are written as (23) i 1 Å nzFk1 (1 0 uG • ) nC exp( 0 ( DG ‡ 1 0 bVF / br1u )/RT )

[3]

i01 Å nzFk01 ( uG • )exp( 0 ( DG ‡ 01 / (1 0 b )VF 0 (1 0 b )r1u )/RT ),

[4]

where i 1 is the current for reaction 1 in the forward direction, i01 the current for reaction 1 in the reverse direction, n the number of electrons transferred, z is the charge, F is the Faraday constant, k1 and k01 are the rate constants for the forward and reverse of reaction 1, u is the fractional surface coverage, G • refers to the radical species on the electrode surface, C is the protein concentration in the bulk solution, DG ‡ and DG ‡ are the Gibbs free energy of activation, b 1 01 represents the symmetry factor, V is the potential, r1 allows for variation in the heat of adsorption with coverage, R is the gas constant, and T is the temperature in Kelvin. Step 1 can be considered in quasi-equilibrium if step 2 is rate determining. The rate expression (after rearrangement

AID

JCIS 4500

/

6g18$$$763

FIG. 7. Determination of the number of carboxylate groups in the insulin molecule and its peptides: ( s ) insulin, ( n ) ( m ) chain A insulin, ( h ) ( j ) chain B insulin.

11-20-96 15:05:14

ln(C/ u ) Å n[ 0ln(1 0 u )] 0 ln K / ( s ),

[5]

where K Å k1 /k01 and s Å ( DG 0 VF / r1u )/RT. The slope of a plot of ln(C/ u ) versus 0ln(1 0 u ) gives the value of n, the number of electrons transferred and the number of carboxylate groups interacting with the metal surface. From the plot of ln(C/ u ) versus 0ln(1 0 u ) (Fig. 7), the value for n was determined from the slope to be 6 { 2 for insulin, which corresponds very well with the number of acidic residues present on the protein. This result suggests that a significant concentration of the insulin molecule exists as a monomer in the low concentrations used in these experiments in the phosphate buffer solution at pH 7.0. All the measurements were made at concentrations õ0.10 g liter 01 , for which monomers may be present in significant quantities (21, 22). In two different experiments, the value of n for chain A was determined to be 2 { 0.4. Since chain A contains two carboxylate groups plus the end carboxylate, this number agrees quite favorably with those on the molecule. However, the number obtained for chain B was 6 { 2. Chain B only contains three carboxylate groups including the C terminal group. However, this chain is responsible for the formation of dimers in the associations of the insulin molecule to form hexamers. Chain B is known to dimerize through the hydrophobic interactions of amino acids B12 valine and B16 tyrosine and the last eight residues of the extended B chain, rather than through disulfide or ionic linkages (16). In addition, four hydrogen bonds result in a tight pairing of the dimers, and the authors concluded that this form of the dimer exists in solution (19). Therefore in the present study, dimerization could occur in these aqueous solutions, which would account for the value of n, since there would then be a total of six carboxylates including the two terminal C groups for the dimer.

coida

AP: Colloid

INTERFACIAL BEHAVIOR OF INSULIN

CONCLUSIONS

Studies of the interfacial behavior of insulin, chain A and chain B of insulin at the platinum electrode over the temperature range 299 to 363 K, gave a linear van’t Hoff relationship from 299 to 333 K. In this temperature region all three peptides adsorbed strongly, although the individual chains consistently gave higher surface charge densities than the whole molecule of insulin. This is not unexpected as the smaller chains are able to pack more efficiently on the electrode surface due to flexibility in their structures. The enthalpy of adsorption, DHADS , calculated from the linear van’t Hoff relationships over the temperature range 299 to 333 K gave values of 022 { 1, 017 { 1, and 09 { 1 kJ mol 01 for insulin, chain B, and chain A, respectively. Above these temperatures denaturation of insulin occurs, and for all three molecules the surface adsorption measured by the surface charge densities showed an immediate decrease followed by a slight increase. The surface concentrations of insulin of 2.9 { 0.2 mg m02 at 299 K and 3.2 { 0.3 mg m02 at the higher physiological temperature of 310 K agreed well with the calculated value determined from geometrical dimensions. Similar calculations from the experimental surface charge densities for chain A and chain B indicated that a more efficient packing prevailed with the individual polypeptides. From a consideration of the mechanism of adsorption based on cyclic voltammetric measurements, an estimation of the number of electrons transferred and hence the number of carboxylate groups on insulin was determined to be 6 { 2, which agrees with the known number of acidic residues on the protein. Similar calculations for chain A gave 2 { 0.4, which indicates that this peptide remains as a monomer in the phosphate buffer. However, the value obtained for chain B under similar experimental conditions was 6 { 2, indicating that this peptide appears to dimerize. When insulin forms dimers, chain B holds the monomers together through a series of hydrophobic interactions and four hydrogen bonds. Thus, under these conditions, chain B polypeptides appear to dimerize readily in the phosphate buffer. ACKNOWLEDGMENTS Grateful acknowledgment is made to the Natural Sciences and Engineering Research Council, Canada for support of this research. The authors thank James R. Roscoe for writing the computer integration program for determining the cyclic voltammogram areas.

REFERENCES 1. Brash, J. L., and Lyman, D. J., J. Biomed. Mater. Res. 3, 175 (1969). 2. Reynaud, J. A., Malfoy, B., and Bere, A., Bioelectrochem. Bioenerg. 7, 595 (1980).

AID

JCIS 4500

/

6g18$$$763

11-20-96 15:05:14

455

3. Iwamoto, G. K., Van Wagenen, R. A., and Andrade, J. D., J. Colloid Interface Sci. 86, 581 (1982). 4. Tol, A., Quik, R. F. P., and Thyssen, J. H. H., Pharm. Weekly Sci. Ed. 10, 213 (1988). 5. Minemoto, M., Asahara, T., Maeda, H., and Noda, H., Byoin Yakugaku 11, 431 (1985). 6. Twardowski, Z. J., Nolph, K. D., McGary, T. J., and Moore, H. L., Am. J. Hosp. Pharm. 40, 583 (1983). 7. Feyerherd, F., Lorenz, D., Neumann, J., Lippert, H., and Ziegler, M., Anaesthesiol. Reanim. 6, 94 (1981). 8. Baier, R. E., Meyer, A. E., Natiella, J. R., Natiella, R. R., and Carter, J. M., J. Biomed Mater. Res. 18, 337 (1984). 9. Sefton, M. V., and Antonacci, G. M., Diabetes 33, 674 (1984). 10. Waugh, D. F., Wilhelmson, D. F., Commerford, S. L., and Sackler, M. L., J. Am. Chem. Soc. 75, 2592 (1953). 11. Bringer, J., Heldt, A., and Grodsky, G. M., Diabetes 30, 83 (1981). 12. Sato, S., Ebert, C. D., and Kim, S. W., J. Pharm. Sci. 72, 228 (1983). 13. Thurow, H., and Geisen, K., Diabetologia 27, 212 (1984). 14. Mizutani, T., J. Pharm. Sci. 69, 279 (1980). 15. Hodgkin, D. C., Blundell, T. L., Cutfield, J. F., Cutfield, S. M., Dodson, G. G., Dodson, E. J., Mercola, D. A., and Vijayan, M., in ‘‘Insulin Action’’ (I. B. Fritz, Ed.), p. 1. Academic Press, New York, 1972. 16. Adams, M. J., Blundell, T. L., Dodson, E. J., Dodson, G. G., Vijayan, M., Baker, E. N., Harding, M. M., Hodgkin, D. C., Rimmer, B., and Sheat, S., Nature 224, 491 (1969). 17. Blundell, T. L., Cutfield, J. F., Cutfield, S. M., Dodson, E. J., Dodson, G. G., Hodgkin, D. C., Mercola, D. A., and Vijayan, M., Nature 231, 506 (1971). 18. Jeffrey, P. D., and Coates, J. H., Biochemistry 5, 3820 (1966). 19. Blundell, T. L., Dodson, G. G., Hodgkin, D. C., and Mercola, D. A., Adv. Protein Chem. 26, 279 (1972). 20. Pocker, Y., and Biswas, S. B., Biochemistry 19, 5043 (1980). 21. Pocker, Y., and Biswas, S. B., Biochemistry 20, 4354 (1981). 22. Pekar, A. H., and Frank, B. H., Biochemistry 11, 4013 (1972). 23. Roscoe, S. G., Fuller, K. L., and Robitaille, G., J. Colloid Interface Sci. 160, 243 (1993). 24. Roscoe, S. G., and Fuller, K. L., J. Colloid Interface Sci. 152, 429 (1992). 25. Rouhana, R., Budge, S. M., MacDonald, S. M., and Roscoe, S. G., Food Res. International, in press. 26. Hanrahan, K.-L., MacDonald, S. M., and Roscoe, S. G., Electrochim. Acta 41, 2469 (1996). 27. Angerstein-Kozlowska, H., in ‘‘Comprehensive Treatise of Electrochemistry’’ (E. Yeager, J. O’M. Bockris, B. E. Conway, and S. Sarangapani, Eds.), Vol. 9, p. 15. Plenum, New York, 1984. 28. Kennedy, J. H., ‘‘Analytical Chemistry,’’ p. 482. Harcourt Brace Jovanovich, San Diego, 1984. 29. Brown, E. R., and Large, R. F., in ‘‘Physical Methods of Chemistry, Part IIA, Electrochemical Methods’’ (A. Weissberger and B. W. Rossiter, Eds.), Vol 1, p. 506. Wiley-Interscience, New York, 1971.) 30. Glazer, A. N., De Lange, R. J., and Sigman, D. S., in ‘‘Laboratory Techniques in Biochemistry and Molecular Biology, Volume 4, Part I, Chemical Modification of Proteins, Selected Methods and Analytical Procedures’’ (T. S. Work and E. Work, Eds.), p. 113. North-Holland, Amsterdam, 1975. 31. Janatova, J., Fuller, J. K., and Hunter, M. J., J. Biol. Chem. 243, 3612 (1968). 32. Arnebrant, T., and Nylander, T., J. Colloid Interface Sci. 122, 557 (1988). 33. Razumas, V., Kulys, J., Arnebrant, T., Nylander, T., and Larsson, K., translated from Electrokhimiya 24, 1518 (1988). 34. Marangoni, D. G., Smith, R. S., and Roscoe, S. G., Can. J. Chem. 67, 921 (1989). 35. Marangoni, D. G., Wylie, I. G. N., and Roscoe, S. G., Bioelectrochem. Bioenerg. 28, 269 (1991).

coida

AP: Colloid