ELSEVIER
Bioelectrochemistry and Bioenergetics 44 (1997) ! 3-21
Review
The reduction of disulfide bonds in proteins at mercury electrodes Michael J. t-loneychurch
!
DepaJ~tment of Molecular Sciences, James Cook Unicersity of North Queensland, Tm~vzscille, 48! 1 Queensland, Australia Received 14 August 1996; received in revised form 21 March 1997
Abstract The adsorption of disulfide containing proteins on mercury and subsequent reduction is reviewed. Methods for determining the protein surface excess and number of electroactive disulfide bonds are discussed. Peaks in up to three potential regions (I, II and lid are observed depending on experimental conditions. Peaks in regions I and lll have not been as extensively studied as peak II which is attributed to the reversible or quasi-reversible reduction of a mercury-protein thiolate bonds which are formed from disulfide bonds located in hydrophobic regions of a protein as well as certaio sulfhydryl groups and thioethers. © 1997 Elsevier Science S.A. Keywords: Protein surface excess; Electroactive disulfide bonds
1. Introduction
2. Adsorption of proteins on mercury
Proteins have been studied e~ectrochemically for many years and several reviews ~ ,J the adsorption and reduction of proteins on mercury have appeared [1-3]. Since these reviews appeared the development of adsorptive stripping methods [4] have stimulated renewed interest in the accumulation and reduction of proteins at mercury electrodes. With this in mind, it seems timely to once again review the subject of the adsorption of proteins and subsequent reduction of disulfide bonds on mercury and critically re-examine earlier conclusions in light of recently published work. The aim of this review is to detail, for someone intending to work in this area, the voltammetric response that is likely when proteins are reduced at mercury electrodes and to explain the origins of that response and how to obtain information electrochemically from which one may make conclusions about the number of electroactive disulfide bonds, the surface excess of the protein and hence the extent of denaturation of the adsorbed protein.
Much of the current knowledge about protein adsorption stems from the work of MacRitchie [5] and Norde [6] on solid surfaces and Kuznetsov and Shumakovich [7]. Kutnetsov et al. [8-10], and Scheller et al. [11-13] on mercury electrodes. MacRitchie has concluded that only a small portion of the protein molecule needs to enter the interface in order for adsorption to proceed spontaneously. On reaching the interface, the protein molecule has an opportunity to lower the free energy of the system by orienting its non-polar residues toward the non-aqueous phase. At solid surfaces, adsorption increases with increased hydrophobicity of the surface a n d / o r protein. Hydrophobic interactions are very significant and appear to be a major driving force for protein adsorption [6]. On hydrophilic surfaces, protein adsorption is often at least partly reversible whereas adsorption on hydrophobic surfaces is often irreversible. At fluid/fluid interfaces, it is well established that proteins lose their tertiary structure [5]. Studies of polymer adsorption have shown that an adsorbed flexible polymer molecule may be divided into three parts: (1) the trains, which are segments adsorbed at the interface; (2) the loops, which are segments of the chain that extend into the adjacent bulk phase; and (3) the tails, the two s~:gments at the ends of each polymer chain. Kuznetsov ct al. [10] proposed that segments of the protein which adsorb (i.e.,
i Present address. Department of Chemistry, University of Hawaii, 2545 The Mall, H o n o l u l u . HI 9 6 8 2 2 , USA. E - m a i h
[email protected] 0302-4598/97/$17.00 © 1997 Elsevier Science S.A. All rights reserved. PI! S 0 3 0 2 - 4 5 9 8 ( 9 7 ) 0 0 0 6 2 - 7
16
M.J. Honeychurch/ Bioelectrochemisto, and Bioenergetics 44 (1997) 13-21
trains) are in relatively hydrophobic regions whereas the loops are made up primarily of hydrated hydrophilic regions. It has been established that proteins denature or unfold following adsorption onto mercury and that adsorption is irreversible. According to the data of Scheller et al. [12], proteins spread ono mercury electrodes to give a final thickness of 8-10 A, the thickness of a single polypeptide chain, at the interface. This has often been termed 'surface denaturation'. The monolayer of protein has a porous structure. Approximately 20% of the electrode area remains available for reactions with smaller molecules diffusing through the porous adsorbed layer. Since an electrode with a monolayer protein coverage is now a hydrophilic surface, it has been shown that native protein molecules can adsorb reversibly onto this surface [8]. For a protein adsorbed on the electrode with a density, p, in g cm-3, and layer thickness, l, in cm, the adsorption area can be related to the molecular weight by:
s = NAtp
(1)
where S is the adsorption area per molecule in cm 2, M r the molecular weight and NA Avagadro's number. If the adsorbed protein layers have similar densities and thicknesses, then there should be a linear relationship between the protein molecular weight and the adsorption area. The adsorption areas of proteins at the dropping mercury electrode (DME) has been calculated from capacitance time curves [ 12]: AC Co = 7.36 x I04NAScr*,DI/2t I/2 ACm
Co
(2)
CO is the capacitance of the supporting electrolyte and AC m is the decrease in capacitance at maximum coverage of the electrode, c~ is the bulk concentration of the protein in tool cm-3, Dp is the diffusion coefficient of the protein in cm 2 s-l, and t is the drop time. Razumas et al. [14] showed that a linear relationship existed between the adsorption area, determined by ac polarography, of alkaline phosphatase and the free energy of hydration of the anions in the supporting electrolyte. For example, the adsorption area of alkaline phosphatase in a supporting electrolyte of iodide was less than half that in phosphate. Adsorption of the protein is accompanied with desorption of water and specifically adsorbed anions. Therefore, adsorption area and hence denaturation is decreased with increasing specific adsorption of the anion. The dependency of the adsorption area on the free energy of hydration of the anions observed by Razumas et al. [14] is apparently due to the relationship between hydration energy and the degree of specific adsorption of the anion [14].
The validity of Eq. (1) was tested by plotting protein molecular weight for thirteen proteins vs. literature values of adsorption area determined at the DME by Eq. (2) [I 5] (Fig. 1). For proteins of molecular weight below 100 kD, a good correlation is obtained. The equation for the line of best fit is M r ( k D ) = - 3 . 2 + 7 . 1 2 X 1 0 - 3 S (,g2) ( r = 0 . 9 7 8 ) . Taking the layer thickness to be 8-10 ,A, this corresponds to adsorbed layer density of 1.2-1.5 g cm -3. This is consistent with the literature value of 1.3 g cm -3 [8,12]. It follows from this discussion on the denaturing of proteins on mercury that the use of protein crystallographic areas to estimate the surface excess would be expected to lead to errors. Despite the evidence that proteins denature upon adsorption onto mercury, Santhanam et al. [17] found that urease retained its enzymatic activity when adsorbed on a HMDE. The enzyme activity could be switched on or off by alternating the potential of the HMDE to - 0 . 4 0 V or - 0 . 5 8 V, respectively. They concluded that the reduction at - 0 . 5 8 V allowed free movement of sulfhydryl groups which was sufficient to alter the distant active site and decrease activity. The restoration of enzymatic activity following the switching of the electrode potential to - 0 . 4 0 V and then subsequent loss of enzymatic activity following the switching of the electrode potential to - 0 . 5 8 V took approximately 10 rain. This is consistent with the slow movement of peptide segments observed by Kuznetsov et al. [10]. However the findings of Santhanam et al. are not necessarily inconsistent with the above model. The only difference between the two outlying proteins in Fig. 1 and the others which make up points on the line of best fit their relative molecular weight. Fig. I shows 200
150
100
~
J
50
0
0
!
5000 10000 15000 Adsorption Area I Az
20000
Fig. 1. Plot of molecular weight vs. adsorption area obtained by ac polarography in at pH 7.0. (O) Kuznetsov et al. [8], hemoglobin in 0. ! M phosphate; ((3) Scheller et al. [12], ribonuclease A, insulin, iysozyme, ovalalbumin, BSA in 0.1 M KCI, (A) Scheller et al. [11], cytochrome c, metmyoglobin, metmyoglobin dimers, glycogen phosphorylase 0.1 M KCI; (m) Razurnas et al. [16], ceruloplasmin, lactoperoxidase and alkaline phosphatase in O.1 M phosphate.
M.J. Honeychurch/ Bioelectrochemisto, and Bioenergetics 44 (1997) 13-21
that the two larger molecular weight proteins do not appear to denature and spread on the electrode surface to the same extent as the smaller proteins that were studied. Larger proteins exist as aggregates of polypeptide chains called subunits which are usually folded into a globular conformation and interact with other monomers [18]. Urease has a molecular weight of 480 kD and is made up of two subunits. It is possible that when larger proteins adsorb only the subunit(s) that gain a foothold on the surface denature. This would result in a smaller adsorption area than would be predicted by extrapolating the molecular weight vs. adsorption area line of best fit obtained from lower molecular weight proteins. If this is the case, one might expect the active site to be preserved in some percentage of the adsorbed urease.
3. Determination of the protein surface excess
In an electrochemical experiment where O and R are the oxidized and reduced form of the adsorbed protein, the total amount of adsorbed protein at time t is equal to the amount initially adsorbed (t = 0) plus the total amount brought to, or removed from, the electrode by diffusion after the commencement of the experiment (t > 0). When the protein is accumulated on a HMDE of radius, r o, by an adsorptive preconcentration procedure, the mass balance equation is: ro(t) +/"R(t)
+ no
+ DR
ar
f{~( OcR(r't) ) .Ot. Or r= r,,
(3)
F o and F R are the surface ex,:.e~ses of the oxidized and reduced protein, respectively, and F T is the total surface excess of protein at the completion of the accumulation procedure. In order to simplify Eq. (3), one chooses experimental conditions such that the amount of protein accumulated during the preconcentration procedure will always be significantly greater than the amount of protein accumulated during the analysis step, i.e., voltage sweep. This can be done by extending the preconcentration period or by increasing the voltage sweep rate. The diffusion terms in Eq. (3) can then be ignored and it reduces to:
/~O(t) +/"R(t) ----/-'V.
(4)
The flux of protein to the electrode during the accumulation procedure is:
Dp
Or
r = ro
=opt;
+-- , ro
(5)
where tr is the period of accumulation from a quiescent solution in s. In the case of proteins, strong adsorption is
15
known to take place on mercury. If it is assumed that all of the protein reaching the electrode is adsorbed then:
Dp
[ac(r,~)] dF(t) Or
,.=to =
d--'---~
(6)
The surface excess of protein at a HMDE, FT, can be determined by integrating Eq. (6): r , = ¢; 2
~-
+ ,,
ro
(7)
Several workers [19-23] have calculated the surface excess while failing to take into account the additional term (the fight hand term in the brackets in Eq. (7)) needed due to the sphericity of the HMDE. These workers have made calculations by assuming that the surface excess varies directly with the square root of time. This is the case for a DME or for diffusion to a planar surface, but leads to errors in the determination of the protein surface excess at a HMDE. It should be noted that in these papers, the experimental capacitance-time curves during the accumulation procedure are measured. It is reported that the capacitance decreases directly with the square root of time for small accumulation times and varies directly with time for longer duration's. It is unnecessary and incorrect to divide the capacitance-time curve into two regions. Consider the following equation:
C=Co(I --O} +CmO,
(8)
where C is the capacitarice, CO is the capacitance of the supporting electrolyte; 0 is the fractional coverage given by F J F m, Fm is lhe maximum surface excess and C m is the minimum capacitance, which is measured at Fm. It can be seen from F~I. (7) that at short accumulation times 2(°-~/)1/2>>-~-~ ro and from Eq. (8), it follows that the capacitance will decrease proportionally to the square root of time. At longer accumulation times ~r O >> 2(°-~/) 1/2 and the capacitance decreases directly with the increase in accumulation time. Honeychurch [15], and Honeychurch and Ridd [24] have recently determined the adsorption area of several proteins at the HMDE using Eq. (7). The results are shown in Fig. 2. The plot of protein molecular weight vs. the adsorption area was linear as predicted by Eq. (1). The equation for the line of best fit is Mr (kD) = 0.26 + 8.29 × 10 -3 S (,~2) (r = 0.999). Taking the density of the adsorbed layer to be 1.3 g cm -3 [8,12], this corresponds to an adsorbed layer thickness of 10.6 ,~, which is consistent with the value obtained from ac polarographic data. The fact that similar results are obtained at both the HMDE and DME indicates that following the initial foothold being gained by the protein on the electrode, the denaturation and spreading to a thickness of approximately 10 ',~ takes place within the timescale of the DME, thereafter movement of
16
M.J. Honeychurch / Bioelectrochemisto' and Bioenergetics 44 (1997) 13-21 100
,
80
I!
,
Transfer~
10 n A
--~e~60 40
Trypsin,/ / Papain _/ RNase A
20 eL o
Y 0
'"~"1"" L t , 2000 4000 6000 8000 Adsorption Area I ~2
+o.',
i
-,.e
II
-m.e
E/V
Fig. 3. Cyclic voltammetricbehaviorof 10 mg i- ~ humanserum albumin following 800 s accumulation at a HMDE at +0.10 V. Scan rate = 200 mV s -I . (Taken from Ref. [26]).
10000
Fig. 2. Plotof molecularweight, M r, v s . adsorption area of proteins, S, at a HMDEfollowing60 s accumulationin unstirred solution.
peptide segments within the adsorbed layer is slow (relative to the time to complete the experiment). If the number of electroactive disulfide bonds per protein molecule is known, then the surface excess may be calculated from the total charge obtained by integrating cyclic voltammograms: a = nFr T ,
|
.o 2
(9)
where Q is the charge in C cm -2. For many proteins, however, the number of electroactive disulfide groups is not known, therefore it is not possible to use Eq. (9) to determine the surface excess and adsorption area. The value of n obtained from bulk electrolysis experiments would not necessarily be reliable for use in shorter timescale voltammetric measurements. A bulk electrolysis experiment is typically very much longer than a voltammetric experiment. Following the initial reduction (disulfide bond breakage), the peptide chains in the adsorbed protein would obtain more 'freedom' to move and therefore increase the possibility that other disulfide bonds may now be stericly able to be reduced. This process would lead to a larger value of n than would be observed in the shorter timescale voltammetric experiments. Also, a protein, or any macromolecule, can adopt several conformations at the electrode following adsorption, depending on which part of the molecule gains the initial foothold on the electrode surface [5]. Therefore, we may expect that the experimentally determined number of electroactive disulfide bonds will be a mean of the number of electroactive bonds from all the different conformations. Studies with metalloproteins such as cytochrome c produce integral values of n. These proteins may also adopt many different conformations at the electrode surface. The difference between the reduction of a metal center in metalloproteins and disulfide bond(s) in proteins is that the reduction of the later is catalyzed by the mercury electrode (see below) and requires the disulfide bond(s) to be in
direct contact with the electrode. In addition to this, only disulfide bonds in hydrophobic regions of the protein are likely to give rise to an electrochemical response at approximately - 0 . 6 V [10,25]. Therefore, conformational variations of adsorbed disulfide containing proteins lead to variations in the accessibility of the disulfide bonds to the electrode surface and may produce non-integer n values for the reduction process. The experimentally determined value of n therefore represents a mean value for the various conformations adopted by the protein adsorbing on the electrode surface.
4. The nature o f t h e e l e c t r o c h e m i c a l r e s p o n s e o f p r o t e i n s at m e r c u r y e l e c t r o d e s
In the study of protein, reduction at mercury electrodes peaks maybe observed in up to three potential regions depending on the experimental conditions employed (Figs. 3 and 4). A voltammetric peak is sometimes observed between - 0 . 2 to - 0 . 5 V (peak I). Peak I occurs following accuII
III
0.4
0.3
0.2
-.
0.1
llq
o
-0.1
-0.~i
-0.~7
"0:9 "1.;
-1.3~
E/V
-0.2
Fig. 4. Cyclic voltammogramof 0.078 mM insulin at the HMDE following 60 s accumulation at open circuit from a stirred solution. Scan rate = 200 mV s-n. (Taken from Ref. [27]).
M.J. Hoaeychurch/ Bioelectrochemistry and Bioenergetics 44 {1997) 13-21
17
Table 1 Selected CV studies o f peak II in proteins. Measurements carded out in 0. I M phosphate buffer unless stated otherwise. All potentials are in Volts vs. the NHE Protein
pH
Ei
Chymotrypsin Ferredoxin GDH
7.4 a 7.4 7.4 0.9 b 6.9 7.0
HSA
7.4
Anti-HSA Insulin IgM
Scylla Serrata M T
7.4 7.4 7.4 c 7.4
+0.25 - 0.1 - 0.8 + 3.2 + 0.14 + 0.35 +0.25 +0.05 + 0.30 + 0.30 - 0. ! +0.30 - 1.0
Human M T Prothrombin fragment 2 Prothrombin fragment I Prothrombin Ribonuclease Somatostatin Trypsin Urease
7.3 d 7.8 e 7.8 e 7.8 e 4.6 f 4.6 f 0.9 b 7.3
BSA
....
+ + + + +
Epc
Epc
Epa
Peak I
Peak II
Peak U
-0.45 - 0.43 - 0.34 - 0.10 - 0.36 - 0.3 --0.3 -0.48 - 0.35 - 0.35 sh - 0.43 -0.38 - 0.41
-0.49 -0.40
0 -0.1 -0.33 - 0.08 - 0.06 -
1.2 0.1 0. ! 0.1 0.3 0.3 0.2
v (mV s - t )
200 200
[28] [29]
-o.35
200
[3o]
--0.09 -0.36
50 90 10-100 100 200
[31] [19] [32]
100 200 100 200
[26] [27] [33] [34]
200 200 200 200 200 200 50 45
[34] [22] [20] [21 ] [35] [35] [31] [17]
-0.3 -0.35 - 0.32 - 0.43
- 0.46 - 0.45
- 0.59 -0.41 - 0.46 - 0.45
-
-
0.45 0.23 0.25 0.08 0.34
Ref.
0.48 0.27 0.27 0.07 0.28
[261
a0.05 M phosphate; b0.1 M HCI; c0.025 M phosphate; d0.15 M NaCI; e0.001 M Tris; f0.05 M acetic acid/acetate. E i, initial potential; sh, shoulder; ALT, alanine aminotransferas¢; GDH, glutamate dehydrogenase; HSA, human serum albumin; IgG, immunogiobulin G; lgM, immunoglobulin M. Table 2 Selected pulse and alternating current voltammetric studies of peak II in proteins and polypeptides. Measurements carried out in 0.1 M phosphate buffer unless stated otherwise. All potentials are in Volts vs. the NHE Protein
Alkaline Phosphatase ALT Amino acid Oxidase Avidin a'b BSA BSA BSA Ceruloplasmin Con A Ferritin GDH Haemoglobin HSA Anti-HSA Mouse IgG Antimouse IgGb IgG Lactoperoxidase Phosphorylase b Xanthine Oxidase
Method
pH
Ei
Ep~
Epc
Peak I
Peak II
ac
7.0
- 0.4
DP ac
7. !
+ 0.25
DP DP
7.4
SW ac ac DP ac DP ac DP DP DP DP DP ac ac ac
7.4
+ 0.1 +0.35 + 0.25 + 0.3
7.0 7.4
- 0.4 + 0.25
- 0.01
7.0 7 7.4 7.4 7.4 7.4 7 7.0 7
+ 0.25
- 0. l
0 +0.30 + 0.2 + 0.25 0 - 0.4
- 0.44 + 0.03
-0.14 - 0.03
- 0.05 - 0.03
- 0.29 - 0.44 - 0.30 -0.35 - 0.35, - 0.45 sh - 0.35 - 0.45 - 0.48 - 0.33 - 0.48 - 0.35 - 0.41
-
0.36 0.33 0.35 0.44 0.4 0.48
u (mY s - t )
A E (DP) (mV)
1.7
f (ac, SW) (Hz)
Ref.
60
[ 16]
l0
50
[36] [8]
l0 5 5 240
100 50 50 50
[37] [28]
1.7 5
60 50
10
50
5 5 10 l0 nr 1.7
50 50
nr 60
[38] [8] [16] [39] [8] [32] [8] [26] [26] [40] [40] [41] [16] [8] [8]
a0.05 M phosphate; bopen circuit accumulation. DP, differential pulse; SQ, square wave; ac, alternating current; oc, open circuit, var, variable DE, pulse width, f , ac or SW frequency; sh, shoulder; nr, not reported, E i, initial potential energy.
M.J. Honeychurch / Bioelectrochemistr3.'and Bioenergetics 44 (1997) 13-21
mulation at positive potentials and at pH > 5.0 but has not been the subject of detailed studies. The experimental conditions under which peak I appears would indicate that it may be a capacitative peak due to interactions between the electrode and carboxylate groups in the protein. A pH-dependent reversible or quasi-reversible voltammetric peak (or polarographic wave) is often observed at a potential of approximately - 0 . 6 V vs. SCE at pH 7. This will be referred to as peak II in this discussion. This has been investigated by a number of electroanalytical techniques and is generally attributed to the protein disulfide bonds [1,2] as well as certain sulfhydryl groups (e.g., cysteine residues in ferredoxin and metallothionein) and thioethers (e.g., methionine residues in concanavalin A). This peak is discussed in more detail in Section 5. The third voltammetric peak which is found at higher protein solution concentrations ( > p,M) is irreversible and found between - 0 . 9 to - 1 . 2 V. Peak III is a diffusion peak since in linear sweep voltammetric experiments the dependency of the peak height with square root of the sweep rate is linear. It has been attributed to the reduction of disulfide bonds in hydrophilic 'loops' which diffuse to the electrode surface [10] or the reduction of disulfide bonds in proteins diffusing from the bulk [27]. Tables 1 and 2 show some selected examples of the studies carded out on the reduction of disulfide containing adsorbed proteins.
5. The reduction of disulfide bonds in adsorbed proteins
Stankovich and Bard [27,29] carried out a detailed study of the reduction of zinc-free bovine insulin and BSA at the HMDE in 0.1 M phosphJte buffer at pH 7.4 following an accumulation period. P e ~ II was interpreted as being an adsorption peak due to the linear dependence of the peak current on the sweep rate. For the reduction of insulin, they concluded that peak II was due to the reduction of two disulfide bonds and ruled out the possibility of a reaction between the adsorbed insulin and the electrode to form a thiolate salt. To test for any direct reaction between mercury and BSA at open circuit, a BSA solution was stirred over a mercury pool for 8 h. A spot test for mercuric ion was negative. They concluded that no reaction takes place between BSA and mercury and no denaturation on stirring over a mercury pool occurs. Given the likelihood that the solubility product of the mercuric thiolate is very low a negative spot test is not unusual. Moreover, their interpretation of their results was heavily influenced by their conclusion in their previous paper [42] that no chemical reaction took place between cystine and mercury. That conclusion, which differs from all other studies that have been published on cystine reduction on mercury, was subsequently questioned by Florence [43].
Cystine has been studied extensively but its does have a drawback as an analogue for the study of the reduction of protein disulfide bonds becau,~e the reduction product, cysteine, desorbs whereas a protein remains adsorbed following its reduction. However, studies with small polypeptides containing disulfide bonds seem to yield CVs which are decidedly more complex than those of proteins [44]. The most comprehensive study of cystine reduction on mercury has been published recently by Heyrovsky et al. [45], and Vavricka and Heyrovsky [46]. Using several different electrochemical techniques, these workers found that cystine reacts chemically with mercury, forming a surface bound mercuric cysteine thiolate. This compound is transformed in a currentless process to cysteine mercurous thiolate analogous to the reaction which occurs for glutathione [47]. The surface reactions of cystine in acidic solutions are relatively slow due to the repulsion caused by the extra positive charge carded by the molecule. Mercuric ions can also react with cystine in solution to form mercuric cysteinate by complexing with the carboxyl groups. From these studies by Heyrovsky et al., one may conclude that by analogy with cystine, a chemical reaction with mercury does take place between protein disulfide bonds which are accessible to the surface. Kuznetsov et al. [10] proposed that peak II is due to the mercury catalyzed reduction of disulfide bonds in hydrophobic regions of the protein. In other words, only the disulfides in protein regions which form adsorbed 'trains' react to form the mercury-protein thiolate salt that is subsequently reduced. This proposition has recently been tested by Honeychurch and Ridd [25]. Honeychurch and Ridd determined the value of n for five proteins adsorbed on a HMDE. A hydrophobicity scale was developed from the free energy of transfer of amino acids from water to cyclohexane and the relative hydrophobicity of each of the disulfide bonds in each protein was determined. The number of 'hydrophobic' disulfide bonds in each protein was determined based on the hydrophobicity of its neighboring amino acids. Assuming only hydrophobic disulfides are electroactive, the predicted number of electrons that would be transferred in the reduction is twice the number of hydrophobic disulfide bonds. A good correlation was obtained between the number of electrons transferred in the reduction determined experimentally and the value predicted from the hydrophobicity scale. Pavlovic and Miller [48] studied the reduction of denatured and native ribonuclease by ac polarography. The reduction process, for which the apparent number of electrons was one, appeared to be impeded at higher surface concentrations of the protein. They proposed the following mechanism to account for their results: RSSR' + Hg + H ++ e - ~ RSHg + R'SH,
(10)
RSHg 4- H ÷+ e - ~ RSH + Hg,
(ll)
M.J. Honevchurch / Bioelectrochemistr3' and Bioenergetics 44 (1997) 13-21
where R and R' are peptide segments in the protein. Eq. (10) was considered to be the rate determining step in the above mechanism. Due to the mechanism, including a slow rate determining step plots of E vs. log (Ox/Red) would have slopes (RT/nF) of 60 mV. This indicates an apparcnt one electron transfer instead of the expected 30 mV for a two electron transfer and is analogous to the observation of Stricks and Kolthoff [49] for the reduction of glutathione on mercury. Peak II is therefore due to the one electron reduction of a mercury-protein thiolate bond. Since the electrode reaction was reversible, they proposed that the slow access of the electroactive groups to the surface was responsible for the frequency dependence at high surface concentrations. The diffusion coefficient of the electroactive groups within the adsorbed layer was calculated to be in the order of 10 - ~ to 10 -~2 cm 2 s -I Further evidence to support the conclusion that a chemical reaction takes place between the protein disulfide bonds and the mercury electrode were obtained by Honeychurch and Ridd [50]. They studied the chemical oxidation of electrolytically reduced disulfide bonds by potentiometric stripping analysis (PSA). The analytical response, which was a derivative chronopotentiometric (DCP) peak, was due to the oxidation of sulfhydryl groups by dissolved oxygen. It is well known in the biochemical literature that this process is catalyzed by mercury [51]. For comparison, the sulfhydryl groups were oxidized by applying a constant current the absence of dissolved oxygen. The oxidative DCP peaks had a similar shape and peak potential to those obtained by the mercury catalyzed oxidation leaving one to conclude that a mercury-thiolate salt was formed in this experiment also. During the electrolysis period, in the presence of dissolved oxygen, the disulfide bonds are reduced at the electrode. The sulthydryi pairs may then be oxidized by oxygen, in a mass transport limited chemical reaction, back to the disulfide bond, since reorientation of the suifhydryl pairs is slow relative to the transport of oxygen to the electrode. The resulting disulfide bond is then immediately reduced under the applied potential and the cycle is repeated. In the PSA studies of Honeychurch and Rid& the presence of dissolved oxygen stabilized most of the proteins against further conformational changes following there electrolytic reduction. Insulin, which c3nsists of two peptide chains linked by two disulfide bonds, was an exception. Stankovich and Bard [27] had previously noted that CV peaks for adsorbed insulin decreased with successive cycles. They attributed this to a reorientation ~,f sulfhydryl groups to a geometry that prohibited their oxidation. In the PSA studies, the insulin signal decreased with increasing electrolysis time, even in the presence of the stabilizing effect of dissolved oxygen, but only at monolayer coverage. Since this was not observed for any of the other proteins in their experiments, which were all
19
single chain proteins~ Honeychurch and Ridd concluded that the peptide chains were moving relative to one another rather than the mdlvluual sulfliydJ~yl groups. There have been literature reports of proteins devoid of disulfide bonds giving rise to a voltammetric peak similar to peak II. Ferredoxins are iron sulphur proteins containing cysteine residues but devoid of disulfide bonds. The reduction of apo-ferredoxins (iron free) have been studied by CV and ac voltammetry by lkeda et ai. [19]. CVs showed symmetrical peaks at - 0 . 6 V vs. SCE which were attributed to the R S H / R S H g redox couple. The proposed mechanism was: RSHg + e - ~ RS- + Hg,
(12)
R S - + H + ~ RSH.
(13)
The apparent rate constant for the overall reaction, determined for two types of ferredoxin at several pHs, was in the range 101-152 s -i. Olafson and Sim [30] and Olafson [34] have studied a variety of metallothioneins. Metallothioneins are non-disulfide containing proteins with a high cysteine content whose biological role is to bind metal ions, primarily cadmium, copper and zinc. In the absence of metal ions and by beginning with an anodic sweep, the CVs were similar to those of disulfide-containing proteins. In the presence of metal ions, the voltammem'c peaks are shifted according to the difference in the stability of the metal ion-thiolate complex relative to that of mercuric-thiolate. In ferredoxins and metallothioneins, the cysteine residues are oriented specifically to bind to an appropriate metal ion. There do not appear to have been any reports of free cysteine residues in other proteins that are giving rise to peak II. Rodriguez Fiores et al. [39] studied the interaction of the lectin concanavalin A (Con A) with mannose. Con A is a tetramer (M r = 102,500), each monomer contains one Ca :+ and one Mn :+ ion and has two methionine residues but no cysteine residues. Con A has a binding site for sugars such as mannose for which the metal ions are necessary [52]. The adsorptive stripping response is similar to that for disulfide containing proteins with peak I at -0.21 and peak II -0.53. Addition of mannose to the cell resulted in a linear increase in the peak current of the main peak presumably due to an increase in the surface excess (decrease in adsorption area) following a conformational change due to the sugar binding. If it is assumed that the reduction of methionine residues in Con A account for peak |I then the similarities between the reduction peaks of Con A with those recorded for cystine/cysteine containing proteins are analogous to the similarities in voltammetric signals resulting from the reduction of metbionine with that of cystine. yon Wandruska and Yuan [53] recently reported on the cathodic square wave stripping of several sulfur species
M.J. Honeychurch / Bioelectrochemisto' and Bioenergetics 44 (1997) 13-21
20
including cysteine, cystine and methionine. Cysteine and cystine gave peaks at approximately -0.55 V vs. SCE in mixed borate phosphate buffer pH 8.5 following accumulation at 0 V. A broad peak was also present at approximately -0.20 V which corresponds to the Hg(II) --) Hg(l) transition observed by Heyrovsky et al. [45]. Methionine and thioproline also gave cathodic stripping peaks. Methionine gave a peak at -0.47 V of similar shape to cystine and cysteine under the same conditions. A capacitance peak (peak I) was also present at -0.05 V. A reaction scheme for the reduction of methionine was proposed: 2 R C H 2 C H 2 S C H 3 + Hg + 2 H 2 0 (RCH2CH2S)2Hg + 2CH3OH + 2H + + 2e-.
(14)
Recently, Buckley et al. [37] reported that streptavidin, a protein with no sulphur containing amino acids produced a voltammetric response similar to disulfide containing proteins (i.e., peak II). The nature of the peak was not investigated in any detail and since no other reports of non-sulphur containing proteins exhibiting peak II exist the voltammetric response of streptavidin would appear to be anomalous.
6. Conclusion (i) Proteins denature upon adsorption on mercury electrodes to give adsorbed layers of uniform thickness. In dilute solutions, adsorption is mass transport limited. The denaturation is sufficiently rapid that adsorption areas determined at a DME and a HMDE are similar. (ii) Peak II (at approximately - 0 . 6 V) is due to the reversible or quasi-reversible reduction of a mercury-protein thiolate bond which are formed from disulfide bonds located in hydrophobic regions of a protein as well as certain sulfhydryl groups (e.g,, cysteine residues in ferredoxin and metallothioneins) and thioethers (e.g., methionine residues in concanavalin A). (iii) Peak III, which is found at protein solution concentrations of micro-molar or greater, (between - 0.9 to - 1.2 V) is a diffusion peak attributed to the irreversible reduction of disulfide bonds in hydrophilic 'loops' or the reduction of disulfide bonds in proteins diffusing from the bulk.
Acknowledgements M.J.H. would like to acknowledge the receipt of a James Cook University, Department of Molecular Sciences scholarship.
References [1] G. Dryhurst, K.M. Kadish, F. Scheller, R. Renneberg, Biological Electrochemistry, Academic Pre~s, New York, 1982, p. 398.
[2] H. Berg, Electrochemistry of Biopolymers, in: S. Srinivasan, Y.A. Chizmadzhev, J.O.M. Bockris, B. Conway, E. Yeager (Eds.), Comprehensive Treatise of Electrochemistry, Vol. 10, Plenum, New York, 1985, p. 189. [3] E. Palecek, Modem Polarographic (Voltammetric) Techniques in Biochemistry and Molecular Biology, in: G. Milazzo (Ed.), Topics in Bioelectrochemistry and Bioenergetics, Vol. 5, Wiley, Chichester, 1983, p. 65. [4] J. Wang, Voltammetry following nonelectrolyfic preconcentration, in: A.J. Bard (Ed.), Electroanalytical Chemistry, Vol. 16, Marcel Dekker, New York, 1989, p. 1. [5] F. MacRitchie, Proteins at Interfaces, in: C.B. Anfinsen, J.T. Edsall, F.M. Richards (Eds.), Advances in Protein Chemistry, Vol. 32, Academic Press, New York, 1978, p. 283. [6] W. Norde, Adsorption of Proteins at Solid Surfaces, in: L.-H.Y. Lee (Ed.), Adhesion and Adsorption of Polymers, Voi. 2, Plenum Press, New York, 1980, p. 801. [7] B.A. Kuznetsov, G.P. Shumakovich, Bioelectrochem. Bioenerg. 2 (1975) 35. [8] B.A. Kuznetsov, N.M. Mestechkina, G.P. Shumakovich, Bioelectrochem. Bioenerg. 4 (1977) 1. [9] B.A. Kuznetsov, G.P. Shumakovich, N.M. Mestechkina, Bioelectrochem. Bioenerg. 4 (1977) 5 ! 2. [10] B.A. Kuznetsov, G.P. Shumakovich, N.M. Mestechkina, J. Electroanal. Chem. 248 (1988) 387. [11] F. Scheller, M. Jansen, G. Etzold, H. Will, Bioelectrochem. Bioenerg. ! (1974) 478. [12] F. Scheller, M. J~inchen, H.J. Priimke, Biopolymers 14 (1975) 1553. [13] F. Scheller, G. Strnad, B. Neuman, M. Kuln, W. Ostroski, Bioelectrochem. Bioenerg. 6 (1979) 117. [14] V.J. Razumas, G.-J.A. Vidugiris, J.J. Kulys, Bioelectrochem. Bioenerg. 22 (1989) 9. [15] M.J. Honeychurch, Ph.D. thesis, James Cook University of North Queensland, Townsville, Australia, 1995. [16] V.J. Razumas, G.-J.A. Vidug,;ris, A.A. Zapaiskyte, A.V. Gudavicius, J.J. Kulys, Bioelectrochem. Bioenerg. 15 (1986)407. [17] K.S.V. Santhanam, N. Jespersen, A.J. Bard, J. Am. Chem. Soc. 99 (! 977) 274. [18] T.E. Creighton, Proteins Structure and Molecular Properties, 2nd edn., Freeman, New York, 1993, p. 98. [19] T. Ikeda, K. Toriyama, M. Senda, Bull. Chem. Soc. Jpn. 52 (1979) 1937. [20] M.F. Lecompte, J. Clavilier, C. Dode, J. Elion, I.R. Miller, Bioelectrochem. Bioenerg. 13 (1984) 211. [21] M.F. Lecompte, J. Clavilier, C. Dode, J. Elion, I.R. Miller, J. Electroanal. Chem. 163 (1984) 345. [22] M.F. Lecompte, N. Azzouz, Bioelectrochem. Bioenerg. 24 (1990) 13. [23] I.R. Miller, L. Doll, Bioelectrochem. Bioenerg. 24 (1990) 131. [24] M.J. Honeychurch, M.J. Ridd, J. Electroanal. Chem. 418 (1996) 185. [25] M.J. Honeychurch, M.J. Ridd, Bioclectrochem. Bioenerg. 41 (1996) 115 [26] J. Rodriguez Flores, M.R. Smyth, J. Electroanal. Chem. 235 (1987) 317. [27] M.T. Stankovich, A.J. Bard, J. Electroanal. Chem. 85 (1977) 173. [28] J. Rodriguez Flores, R. O'Kennedy, M.R. Smyth, Anal. Chim. Acta 212 (1988) 355. [29] M.T. Stankovich, A.J. Bard, J. Electroanal. Chem. 86 (1978) 189. [30] R.W. Olafson, R.G. Sim, Anal. Biochem. 100 (1979) 343. [31] J. Wang, V. Villa, T. Tapia, Bioelectrochem. Bioenerg. 19 (1988) 39. [32] P. Carty, R. O'Kennedy, E. Lorenzo Abad, J.M. Fernandez Alvarez, J. Rodriguez Flores, M.R. Smyth, K. Tipton, Analyst 115 (1990) 617. [33] J.M. Fernandez Alvarez, M.R. Smyth, R. O'Kennedy, Talanta 38 (1991) 391.
M.J. Honeychurch / Bioelectrochemistr3' and Bioenergetics 44 (1997) 13-21 [34] R.W. Olafson, Bioelectrochem. Bioenerg. 19 (1988) 11 I. [35] U. Forsman, Anal. Chim. Acta 166 (1984) 141. [36] J.M. Fernandez Aivarez, C. O'Fagain, R. O'Kennedy, C.G. Kilty, M.R. Smyth, Anal. Chem. 62 (1990) 1022. [37] E. Buckley, J.M. Fernandez Alvarez, M.R. Smyth, R. O'Kennedy, Electroanalysis 3 ( 1991 ) 43. [38] J. Rodriguez Flores, C. Martin, E. Gonzales, F. Pariente, E. Lorenzo, Electroanalysis 3 ( 1991 ) 405. [39] J. Rodriguez Flores, R. O'Kennedy, M.R. Smyth, Analyst 113 (1988) 525. [40] M.R. Smyth, E. Bucldey, J. Rodriguez Floreso R. O'Kennedy, Analyst 113 (1988) 31. [41] H. Emons, G. Werner, W.R. Heineman, Analyst ! 15 (1990) 405. [42] M.T. Stankovich, A.J. Bard, J. Electroanal. Chem. 75 (1977)487. [43] T.M. Florence, J. Electroanal. Chem. 97 (1979) 219.
21
[44] P. Mader, V. Vesela, M. Heyrovsky, M. Lebi, M. Braunsteinova, Collect. Czech. Chem. Comm. 53 (1988) 1579. [45] M. Heyrovsky, P. Mader, V. Vesela, M. Fedurco, J. Electroanal. Chem. 369 (1994) 53. [46] S. Vavricka, M. Heyrovsky, J. Electroanal. Chem. 375 (1994) 371. [47] I.M. Kolthoff, W. Stricks, N. Tanaka, J. Am. Chem. Soc. 77 (1952) 4739. [48] O. Pavlovic, I.R. Miller, Exp. Suppl. 18 (1971) 513. [49] W. Stricks, I.M. Kolthoff, J. Am. Chem. Soc. 74 (1952) 4646. [50] M.J. Honeychurch, M.J. Ridd, Electroanalysis 8 (1996) 654. [51] P.C. Jocelya, Biochemistry of the SH group, Academic Press, London, 1972, p. 105. [52] H. Bittiger, H.P. Schnebli, Concanavalin A as a Tool, Wiley, London, 1976. [53] R. von Wandruska, X. Yuan, Talanta 40 (1993) 37.