The Journal of
Physical Chemistry
0 Copyright, 1987, by ihe American Chemical Society
VOLUME 91, NUMBER 6 MARCH 12, 1987
LETTERS Direct Electrical Communication between Chemically Modified Enzymes and Metal Electrodes. 1. Electron Transfer from Glucose Oxidase to Metal Electrodes via Electron Relays, Bound Covalently to the Enzyme Y. Degani and A. Heller* AT& T Bell Laboratories, Murray Hill, New Jersey 07974 (Received: December 15, 1986)
Glucose-reduced glucose oxidase does not directly transfer electrons to conventional electrodes because the distance between its redox centers and the electrode surface exceeds, even on closest approach, the distance across which electrons are transferred at sufficient rates. Therefore, electrical communication between the redox centers of this enzyme and electrodes required either the presence, and diffusion to and from the enzyme’s redox center, of 0, and H202,or the presence of members of a redox couple, or the use of special electrodes like TTFITCNQ. We show here that direct electrical communication between the redox center of a large enzyme molecule and a simple metal electrode can be established through chemical modification of the enzyme. When a sufficient number of electron-relaying centers are attached through covalent bonding to the protein of glucose oxidase, electrons are transferred from the enzyme’s redox centers to relays that are closer to the periphery of the enzyme. Because some of the relays are located sufficiently close to the enzyme’s surface, electrons are transferred at practical rates to the electrode. As a result, a glucose-concentration-dependentcurrent flows in an electrochemical cell made with conventional electrodes when the electrolytic solution contains the relay-modified enzyme. Such a current does not flow when the solution contains the natural enzyme. Specifically, electrical communication is established between the FADIFADH, centers of glucose oxidase and gold, platinum, or carbon electrodes through the covalent bonding of an average of 12 molecules of ferrocenecarboxylic acid per glucose oxidase molecule. The electron-relaying centers are amides of ferrocenecarboxylic acid and primary protein amines. Over 50% of the catalytic activity of the enzyme is retained after the centers for electron transfer are introduced. In pH 7.2 solutions containing the chemically modified enzyme, glucose is electrochemically oxidized at potentials that are more oxidizing than +0.44 Y (SHE).
Introduction In homogeneous solutions, redox enzymes, such as glucose oxidase, accept electrons from and transfer electrons to small oxidizable/reducible ions or molecules but do not exchange electrons with simple metal electrodes. In the absence of adequately fast electron transfer, these electrodes do not respond to variations in substrate (e.g., glucose) concentration. Thus, the current flowing through a gold, platinum, or carbon electrode immersed in a glucose oxidase containing electrolytic solution does not vary with the concentration of glucose, even though glucose 0022-365418712091-1285$01SO10
reduces the FAD (oxidized flavin adenine dinucleotide) center of the enzyme to FADHz (reduced flavin adenine dinucleotide), and the reduced enzyme diffuses to the surface of the electrode. The reason that FADH, is not oxidized electrochemically through the electrode reaction
-
FADH, FAD + 2H+ + 2eEo = 0.05 V (SHE) at pH 7
(1)
is that the FAD/FADH2 redox centers are located deep in the enzyme, and even if the enzyme is adsorbed on the electrode, the 0 1987 American Chemical Society
1286 The Journal of Physical Chemistry, Vol. 91, No. 6,1987
distances between either of its two FAD/FADH2 centers and the surface of the electrode exceed the distance across which electrons are transferred at measurable rates. To enable electron transfer from redox centers of enzymes of metal electrodes, the common practice of the past 60 years has been to add to the enzyme solution a redox couple with an appropriate redox potential, or to introduce oxygen, which is reduced by FADH2 to hydrogen peroxide, a chemically and electrochemically reactive species. In these cases the electron-shuttling species penetrates the hydrated enzyme and after approaching its redox centers, transfers or accepts electrons. The oxidized or reduced species then diffuses to the electrode, where it is reduced or oxidized. In the case of glucose oxidase, the ferrocene carboxylate/ferrocinium carboxylate couple is one of several effective electron shuttles.' Glucose sensors based on this shuttle, on the assay of hydrogen peroxide, and on electrodes consisting of members of redox couples such as TTF/TCNQ have been rep ~ r t e d . ~In. ~addition, glassy carbon electrodes that are modified to contain boronic acid functions can electrochemically reduce and oxidize the enzyme but do not respond to g l u c o ~ e . ~ We show here that electrical communication between the redox center of an enzyme and conventional gold, platinum, and carbon electrodes can be established by chemical modification of the enzyme. We also show that in solutions of modified glucose oxidase simple metal electrodes respond to glucose even though the solutions do not contain freely diffusing, electron-shuttling species. The required modification of the enzyme involves the chemical attachment of electron relays to its protein. After the attachment of these relays, electrons at the glucose oxidase FAD/FADH2 centers, which are reduced by glucose to glucose oxidase-FADH2 according to reaction 2
-
0-D-glucose + glucose oxidase-FAD 6-D-gluconolactone + glucose oxidase-FADH2 (2) are transferred to electron relays at the periphery of the enzyme. When the reduced enzyme diffuses to the proximity of the metal electrode, electrons are transferred from the peripheral relays to the metal and the enzyme is electrochemically oxidized through the reaction
-
glucose oxidase-FADH2 . glucose oxidase-FAD
+ 2H' + 2e-
(3)
Key to electrical communication between glucose oxidase and the electrodes is the spacing, Le., the density, of the relays. The relays must be sufficiently close to both the FAD/FADH2 centers and to the metal electrodes (and possibly also to each other) for the electron transfer to be rapid. The distance dependence of electron-transferrates in biosystems , has been the subject of intensive theoretical and experimental research in recent It is now evident that electron transfer (1) Cass, A. E. G.; Davis G.; Francis, G. D.; Hill, H. A. 0.;Aston, W. J.; Higgins, J. I.; Plotkin, E. V.; Scott, L. D. L.; Turner, A. P. F. Anal. Chem. 1984, 56, 667. (2) Kulys, J. T.; Cenas, N . K. Biochim. Biophys. Acta 1983, 744, 57. (3) Albery, W. J.; Bartlett, D. N.; Craston, D. H. J . Electroanal. Chem. 1985,194, 223. (4) Narashimhan, K.; Wingard, L. B., Jr. Anal. Chem. 1986, 58, 2984. (5) Mayo, S. L.; Ellis, W. R., Jr.; Crutchley, R. J.; Gray, H. B. Science (Washington D.C.) 1986, 233, 948. (6) McLendon, G.; Miller, J. R.; Simolo, K.; Taylor, K.; Grant, A. G.; English, A. M. ACS Symp. Ser. 1986, 307, 150. (7) Miller, J. R. In Antennas and Reaction Centers of Photosynthetic Bacteria; Michel-Beyerle, M. E., Ed.; Springer-Verlag: Berlin, 1985; p 234. (8) Boxer, S. G. Antennas and Reaction Centers of Photosynthetic Bacteria; Michel-Beyerle, M. E., Ed.; Springer-Verlag: Berlin, 1985; p 306. (9) Bixon, M.; Jortner, J. J. Phys. Chem. 1986,90,3795; FEBS Lett. 1986, 200, 303. (10) Isied, S. S. Prog. Znorg. Chem. 1984, 32, 443. (1 1) McLendon, G.; Guarr, T.; McGuire, M.; Simolo, K.; Strauch, S.; Taylor, K. Coord. Chem. Rev. 1985, 64, 113. (12) Peterson-Kennedy, S. E.; McGourty, J. L.; Ho, P. S.; Sutoris, C. J.; Liang, N.; Zemel, H.; Blough, N . V.; Margoliash, E.; Hoffman, B. M. Coord. Chem. Rev. 1985,64, 125. (13) Takaka, T.; Takenaka, K.; Kawamura, H.; Beppu, Y. J. Biochem. (Tokyo) 1986,99, 833.
Let ters
y POLYSACCHARIDE
E
Figure 1. Schematic drawing of the glucose oxidase molecule, showing the electron-transfer distances involved in the various steps of moving an electron from its two FAD/FADH2 centers to a metal electrode. Left: the enzyme before modification. Right: the modified enzyme, after chemical attachment of an array of electron-transfer relays.
rates (k) within and between molecules, as well as between electrodes and ions or molecules in their proximity, or between the ions decay exponentially with the distance (d) between the involved centers, Le., k 0: e-ad. In proteins, the electron-transfer rates drop by lo4 when the distance between an electron donor and an acceptor is increased from 8 to 17 A.5 In view of the rapid decay in electron-transfer rates with distance, it is not surprising that glucose oxidase does not communicate electrically with simple metal (i.e., gold or platinum) or semimetal (carbon) electrodes. The enzyme, a structurally rigid glycoprotein of 160 000 Da, has a hydrodynamic radius of 43 A32 and consists of two identical polypeptide chains, each containing a FAD/FADH2 center.33 The rigidity and therefore the ruggedness of the enzyme derive in part from the polysaccharide that forms its outer hydrophilic envelope. A schematic representation of the enzyme before its modification is shown on the left side of Figure 1. In the unmodified enzyme the distances between the FAD/FADH2 redox centers and the electrode, shown as dm, are excessive for electron transfer, and the incremental current, which flows in an electrochemical cell at a given potential when glucose is added to the glucose oxidase containing electrolyte, is too small to be measured. We find that by incorporating electron accepting/transferring relays in the enzyme, without causing damaging structural change, the rate of electron transfer to the metal electrode is greatly enhanced. As shown schematically on the right side of Figure 1, the distances between the FAD/FADH2 centers and the relays (dm),between the relays themselves (dRR),and between the relays and the metal electrodes (dRE)can all be shorter than dm. When dFR,dRR, and dRE or dm and dRE are sufficiently short, electrical
-
~~
(14) Crutchley, R. J.; Ellis, W. R.; Gray, H. B. J . Am. Chem. SOC.1985, 107, 5002. (15) Larsson, S. J . Chem. Soc., Faraday Trans. 2 1983, 79, 1375. (16) McGourty, J. L.; Blough, N. V.; Hoffman, B. M. J . Am. Chem. SOC. 1983, 105,4470. (17) Marcus, R. A. Znt. J . Chem. Kinet. 1981, 13, 865. (18) Marcus, R. M.; Sutin, N. Biochim. Biophys. Acta 1985, 81, 265. (19) Beratan, D. N.; Hopfield, J. J. J . Am. Chem. SOC.1984, 106, 1584. (20) Isied, S. S.; Worosila, G.; Atherton, S. J. J. Am. Chem. SOC.1982, 104,7659. (21) Hopfield, J. J. Proc. Natl. Acad. Sci. U.S.A. 1974, 71, 3640. (22) Jortner, J. Biochim. Biophys. Acta 1980, 594, 193; J. Chem. Phys. 1976, 64,4860; J. Am. Chem. SOC.1980, 102, 6676. (23) Sutin, N . Prog. Znorg. Chem. 1983, 30, 441. (24) Sutin, N.; Newton, M. D. Annu. Rev. Phys. Chem 1984, 35, 437. (25) Kuznetsov, A.; Ulstrup, J. J . Chem. Phys. 1981, 75, 2047. (26) Li, T. T. T.; Weaver, M. J. Am. Chem. Chem. SOC,1984,106,6107. (27) Weaver, M. J.; Li, T. T. T. J. Phys. Chem. 1986,90, 3923. (28) Krishtalik, L. I. Elektrokhimiya 1975, 11, 184. (29) Kuhn, H. Phys. Rev. A 1986, 34, 3409. (30) Mann, B.; Kuhn, H. J . Appl. Phys. 1971,42,4398. (31) Dogonadze, R. R.; Kuznetsov, A. M.; Ulstrup, J. Electrochim. Acta 1977, 22, 967. (32) Nakamura, S.; Hayashi, S.; Koga, K. Biochim. Biophys. Acta 1976, 445, 294. (33) The Worthington Manual; Worthington Biochemical: Freehold, NJ, 1977; p 37-39.
The Journal of Physical Chemistry, Vol. 91, No. 6, 1987
Letters
CH3\
/CH3 N
I
I
NH
(CH, )3
I
N
+
II C II
PROTEIN-NH2 pH 7.2 3M UREA
*
PH 7.2)
N
I I
CH2
II
NH-PROTEIN
e& 'C
/
N O
(CHZ)~
I
NH
+
I
c=o I NH I CH 2 I CH3
CH3
I
1287
III
lY
P
Figure 2. Steps involved in attaching the electron relays to the protein backbone of the enzyme.
communication between the FAD/FADH, centers and the metal electrodes becomes possible. Thus, after electron-transfer relays are incorporated in the protein part of the enzyme, glucose reduces the FAD to FADH, and electrons flow, via the relays, to the metal electrodes. Therefore, the addition of glucose produces an anodic current at a potential corresponding to the redox potential of the protein-bound relay. The incorporation of an array of covalently bound electron relays in glucose oxidase and the resulting observation of a glucose-concentration-dependent anodic current in an electrochemical cell containing the relay-modified enzyme are the subject of this study.
Experimental Section Chemicals. Glucose oxidase [EC 1.1.3.41 from Aspergillus niger (catalogue no. G-8135) was purchased from Sigma, St. Louis, MO. The activity of this enzyme was 118 units/mg. N a - H E P e [sodium 4-(2-hydroxyethy1)- 1-piperazineethanesulfonate], DEC [ 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride], and ferrocenecarboxylic acid (I) were purchased from Aldrich. Sephadex G-15 was purchased from Pharmacia Fine Chemicals. Catalase [EC 1.11.1.61 was purchased from Sigma (catalogue no. C-10). Preparation of the Modified Enzyme. All reactions were carried out in an ice bath. Ferrocenecarboxylic acid (80 mg) was dissolved in 4 mL of a 0.15 M Na-HEPES solution to form a slightly turbid solution of pH 7.3 f 0.1. (When necessary, the pH was lowered to this value by adding dropwise, with stirring, an 0.1 M HCl solution.) DEC (100 mg) was then added and dissolved, followed by 810 mg of urea. After the urea dissolved the pH was again adjusted to 7.2-7.3 and 60 mg of glucose oxidase was added and dissolved. If the pH moved out of the 7.2-7.3 range, it was again adjusted by adding either 0.15 M Na-HEPES or 0.1 M HCl. The solution was then placed in a glass vial, sealed with a paraffin (Parafilm) foil, and left immersed in an ice and water containing dewar overnight (- 15 h). The turbid solution was centrifuged, and the supernatant liquid was filtered under mild (-2 atm) pressure through a 0.2-~m-porefilter. The modified enzyme was separated from the unreacted ferrocenecarboxylate and from the reaction product of ferrocenecarboxylate and DEC by gel filtration through a 2-cm-diameter, 22-cm-long column of Sephadex G-15. Prior to the preparation of the column, the Sephadex was soaked for >2 h in a pH 7.0 buffer that was prepared by adding drops of concentrated HC1 to a 0.085 M solution of Na,HPO,. This buffer was used as the eluent in the separation. The enzyme was eluted in the first, orange fraction, of 4 mL or less. Assay of Enzyme Activity before and after Chemical Modifications. The activities of the unmodified and modified enzyme solutions were determined colorimetrically by measuring the rate of bleaching of 2,6-dichloroindophenol in the presence of 50 mM glucose under N 2 a t m ~ s p h e r e . ~ ~ Assay of Iron in the Enzyme. The iron content of the enzyme was measured, before and after modification, by atomic absorption spectroscopy. (34) Yoshimura, T.; Isemura, T. J . Biochem. (Tokyo) 1971, 69, 839.
Electrochemical Measurements. The electrochemical cell had two compartments. The inner, enzyme-containing, compartment was a 1.0-cm4.d. Pyrex tubing with a VF glass frit. The working electrode was an epoxy-embedded gold, platinum, or carbon disk of 1.5" diameter. The working electrode was polished sequentially with 1.O-, 0.3-, and 0.05-km A1203,rinsed in deionized water, and dried in a nitrogen stream prior to each measurement. The reference electrode was NaC1-saturated calomel, and the counter electrode was a spectroscopic graphite rod. The electrolyte in the outer compartment was the buffer used for gel filtration. The measurements were performed under a nitrogen atmosphere, using a PAR 173 potentiostat/galvanostat with a PAR 175 programmer and an H P 7090A recorder. Preparation of the Modified but Deactivated Enzyme. The modified enzyme was prepared as described earlier in this section, except that 1.44 g of urea was added to the enzyme solution that was reacted with DEC. The urea concentration was thus 6 M. The resulting chemically modified enzyme was subjected to gel filtration on Sephadex G-15. The modified enzyme was eluted first, while part of the FAD coenzyme was retained on the column. The remaining FAD was removed by gel filtration after the enzyme solution was incubated with urea (6 M) for 15 min. The enzyme, though yellow brown because of the covalently attached ferrocene/ferrocinium groups, catalyzed neither electrochemical nor conventional glucose oxidation. Results Chemical Modification of the Enzyme. Hoare and Koshland have shown that carbodiimide is an effective agent for forming amides by coupling amines and carboxylic a ~ i d s . The ~ ~ reaction ,~~ sequence applied to form the amide bonds between the electron-transfer relays and the protein amines of the enzyme is shown in Figure 2. In this sequence ferrocenecarboxylic acid (I) is first reacted with the carbodiimide (DEC) (11) at pH 7.2 to form the 0-acylisourea (111). The latter acylates the primary amine groups of the protein to form amides (IV) and the substituted urea (V). The reaction requires the presence of urea for some batches of the enzyme. Iron Content of the Enzyme before and after Its Chemical Modification. Analysis of solutions of the enzyme by atomic absorption spectroscopy shows that prior to modification there are two iron atoms per enzyme molecule and that after modification there are 14 iron atoms per molecule. Cyclic Voltammetric Characteristics before and after Chemical Modification of the Enzyme. The cyclic voltammograms shown in Figure 3 were obtained for an enzyme concentration of 10 mg/mL (1200 units/mL in the case of the unmodified enzyme) in 0.085 M Na2HP04brought to pH 7.0 by adding concentrated HC1. The solution also contained 6000 units/mL of catalase to decompose any hydrogen peroxide that might be formed in the presence of traces of oxygen. Curve a of Figure 3 show the cyclic voltammogram of solutions of the enzyme before modification with or without 5.0 mM glucose. The cyclic voltammograms observed under identical conditions (35) Hoare, D.; Koshland, D. E. J . Am. Chem. SOC.1966, 88, 2057. (36) Hoare, D.; Koshland, D. E. J . B i d . Chem. 1967, 242, 2447.
1288
The Journal of Physical Chemistry, Vol. 91, No. 6, 1987
20c
a C
+
z 100 W
LI LI 3 0
0.4 0.5 06 POTENTIAL, VOLTS ( S H E )
Figure 3. Cyclic voltammograms of glucose oxidase containing buffer solutions, with and without glucose. Prior to modification, either without glucose or with 5.0 mM glucose (a); after attaching the electron relays, but without glucose (b); after attaching the electron relays, with 0.8 mM glucose, (c); with 5 mM glucose, (d).
with the modified enzyme are seen in curves b (no glucose), c (0.8 mM glucose) and d (5.0 mM glucose) of Figure 3 . It is evident that while the unmodified enzyme does not respond to the addition of glucose, the modified enzyme does. The potential at which half the glucose-concentration-dependent limiting current is reached is 0.5 V (SHE), which is consistent with the redox potential of the ferrocenecarboxylate/ferrociniumcarboxylatecouple. To differentiate between the transport of electrons in the enzyme and the long-known process of mediated electrochemical oxidation of the enzyme by soluble, unattached, low molecular weight redox couples, we must confirm that the electrochemical activation of the enzyme is caused by its chemical modification and not by transport of soluble ferrocene-containing species, which are known mediators of electrochemical oxidation of glucose by glucose oxidase.' Such species could originate from noncovalently attached ferrocene moieties that would dissociate and be removed in the gel filtration step. To this end, the solution of the ferrocenemodified glucose oxidase was incubated with 50 mM glucose at 25 "C under nitrogen for 24 h and again subjected to gel filtration. This time there was no colored band except that of the modified enzyme. The electrochemical behavior of the solution after the second gel filtration was identical with that after the first, Le., with that shown in curves b, c, and d of Figure 3. Enzyme Activity before and after Chemical Modification. After covalent attachment of an average of 12 ferrocene centers per enzyme molecule, the original activity of the enzyme (1 20 units/mg) changed to 70 & 10 units/mg. Thus, to 2/3 of the original activity was retained after chemical modification. Disproof of Conventional Mediation of Charge Transport by the Enzyme. Conventional, low molecular weight redox species, such as oxygen/hydrogen peroxide or ferrocene carboxylate/ ferrocinium carboxylate, mediate the electron-transfer process from the FAD/FADH coenzyme sites to an electrode in a process that involves the following steps: first, glucose reduces the FAD to FADH; next, the oxidized mediator (0, or ferrocinium carboxylate) diffuses into the enzyme and approaches the FADH center sufficiently closely for electron transfer to take place (Le,, for reduction of 0, to peroxide or of ferrocinium carboxylate to
Letters ferrocene carboxylate); then, the reduced species diffuses out of the enzyme and eventually reaches the electrode, where it is reoxidized. Because the structure of glucose oxidase and the positions of the two FAD/FADH centers are not known, one cannot estimate the depth to which the mediator must diffuse into the macromolecule. If only shallow penetration is required, it is not inconceivable that covalently attached ferrocene/ferrocinium centers of a chemically modified, but enzymatically inactive enzyme molecule might penetrate an unmodified, but enzymatically active molecule. If this were the case, the chemically modified enzyme might act as a conventional diffusing mediator and the observed electrical communication between the enzyme and metal electrode could be explained without invoking a decrease in tunneling distance, i.e., without invoking tunneling relays. To disprove such a mechanism, we measured the glucose concentration dependence of the current-voltage characteristics of a cell with a solution containing equal amounts of two different enzymes. One was the native, unmodified, active enzyme. The other was the chemically modified enzyme, with covalently attached ferrocene/ferrocinium centers, which has been intentionally deactivated through removal of its FAD/FADH coenzymes. The current-voltage characteristics of the cell with the two enzymes, in sharp contrast with that made with the chemically modified enzyme but active enzyme, did not change upon the addition of glucose. The characteristics of the cell with the modified but deactivated enzyme did, however, show the normal reduction and oxidation waves of the ferrocene/ferrocinium couple (Figure 3, curve b). This proves that, although electrochemical reduction/oxidation was taking place and although the deactivated enzyme could freely diffuse to the active enzyme, electron exchange between the relays of one enzyme and FAD/FADH centers of the other did not take place and that the chemically modified enzyme did not act as a conventional diffusing mediator.
Discussion Organized arrays for intermolecular electron transfer abound in biological systems. For example, structurally and energetically characterized, theoretically understood arrays are involved in bacterial p h o t o ~ y n t h e s i s . ~ - ~In~ ~these ~ ~ ~ ~arrays, ~ ' long-range electron transfer results from tunneling. The tunneling current depends on the distance between the electron-accepting and -transferring centers, on the chemistry (i.e., the phonon spectrum) of the centers, on the medium through which the electrons tunnel, on the energetics of the electron-transfer process, and on the electric field between the centers. Although nature has precisely microengineered numerous such arrays and there has been important work on insertion of electron donors and acceptors into biomolecules, chemical modification has done little to improve on functional charge-transfer systems. Our results show, however, that such improvement is possible: Electrical communication between glucose oxidase and metal electrodes, of potential relevance to glucose sensors and related glucose-level control systems, has been established by attaching electron relays to the protein part of the enzyme molecules. After such modification, a significant glucose-concentration-dependent current flows at the redox potential of the relay. The current is controlled by the rate of either reaction 2 or 3. In the presence of glucose oxidase and at adequate glucose concentrations the rate of reaction 2 is rapid. However, because glucose oxidase is a 43-A radius molecule and because the FAD/FADH2 centers are deep inside the enzyme, reaction 2 proceeds far too slowly with the unmodified enzyme
(37) Michel-Beyerle, M. E., Ed. Antennas and Reaction Centers of Photosynthetic Bacteria; Springer Verlag: Berlin, 1985. (38) Govindjee, Ed. Photosynthesis: Energy Conuersion by Plants and Bacteria; Academic: New York, 1982. (39) Zinth, W.; Knapp, E. W.; Fischer, S. F.; Kaiser, W.; Deisenhofer, J.; Michel, H. Chem. Phys. Lett. 1985, 119, 1. (40) Zinth, W.; Sander, M.; Dobler, J.; Kaiser, W.; Michel, H. In Antennas and Reaction Centers of Photosynthetic Bacteria; Michel-Beyerle, M. E., Ed.; Spinger-Verlag: Berlin, 1985; p 97. (41) DeVault, D.; Chance, B. Biophys. J . 1966, 6, 825.
J. Phys. Chem. 1987, 91, 1289-1292 enzyme for an electrochemically detectable current to flow to gold, platinum, or carbon electrodes because the distance between the FAD/FADH2 centers and the metal electrodes is excessive (Figure 3, curve a). Reaction 3 does, however, proceed rapidly and a significant current does flow, after an average of 12 ferrocene/ ferrocinium centers are bound to the enzyme molecules by amide links between the ferrocenecarboxylic acid and protein amine groups. To establish electrical communication between an enzyme and a metal electrode, it is necessary that the density of relays be adequate, i.e., that the electron-tunneling distances involved in communicating between the redox center of the enzyme (FAD/FADH2 in the case of glucose oxidase) and the electrode be substantially shortened by incorporation of the relays (Figure 1). Furthermore, in the case of glucose oxidase, the redox potential of the relays must be oxidizing relative to the FAD/FADH2 couple. The +OS1 V (SHE) potential' of the ferrocene carboxylate/ferrocinium carboxylate couple assures the oxidation of FADH2 to FAD, because the redox potential of the FAD/ FADH2 couple is 0.05 V (SHE) at pH 7.2.42 The greatest difficulty that we encountered in this work was, not unexpectedly, covalently attaching a large number of relay molecules to the inner protein part of the enzyme without causing a deactivating structural change. In a very large number of experiments, involving different redox couples and bonding reactions, we were unable to reproducibly attach the electron-transfer relays to the enzyme's protein. In most cases our hopes of success, raised by the observation of electrochemical glucose oxidation, were dashed following the gel filtration step, which readily differentiates between electrochemical oxidation involving diffusing mediators (known already to Thunberg and to T h e 0 r e 1 1 ~ and ~~~) electrochemical oxidation based on covalently attached nondif(42) Fieser, L. F.; Fieser, M. Organic Chemistry; Reinhold: New York, 1956; p 460. (43) Thunberg, T. Skand. Arch. Physiol. 1925, 46, 339. (44) Theorell. H. Biochem. 2. 1935. 278, 263. (45) Theorel1,'H. In Sumner, J. B.; Myrback, K. The Enzymes; Academic: New York, 1951.
1289
fusing electron-transfer relays. (Because the diffusing mediators are small molecules, they are retained longer on the gel filtration columns. The large enzyme molecules do not penetrate the column's pores and, therefore, move rapidly.) Our early failures taught us that the polysaccharide envelope of the enzyme can prevent reaction between the 0-acylisourea (111) and the inner protein in some batches of the enzyme. The results became, however, reproducible when we used 3 M urea solutions, where the inner protein was just sufficiently exposed and unfolded for proper relay attachment, yet the FAD was not lost and the enzyme could recover its structure in a urea-free aqueous buffer. Thus we obtained an enzyme having to 2 / 3 of the activity of the unmodified enzyme in catalyzing the classical reaction of glucose oxidase (eq 4), yet able to communicate with metal electrodes. 8-D-glucose
+ O2
-
6-D-gluconolactone
+ H202
(4)
Direct electrical communication between electron-relay modified enzymes and metal electodes opens a new route to electrochemical and bioelectronic sensors and to the direct electrochemical synthesis of biochemicals. Work in progress in our laboratory shows that there are several families of useful electron-transfer relays and that these can be attached to different functional groups of enzyme proteins. Although the ferrocene relays are at this time the most effective, we find that R ~ " ( N H ~ ) ~ ~ ~ , ~ ' relays, - b a s e attached d to the histidine functions of the protein of glucose oxidase, are short-lived but also effective. Furthermore, we find that the establishment of electrical communication between enzymes and metal or graphite electrodes is not limited to glucose oxidase but is feasible also in other enzymes such as amino acid oxidase.
Acknowledgment. We are grateful to Robert Bittman of Queens College, CUNY, for sharing with us his insight into conformational changes in proteins, to Anthony M. Williams for iron assays by atomic absorption spectroscopy, and to Tetsuo Yamane for many enlightening discussions. (46) Matthews, C. R.; Erickson, P. M.; Van Vliet, D. L.; Petersheim, M. J . Am. Chem. SOC.1978, 100, 2260. (47) Sandberg, R. J.; Gupta, G. Bioinorg. Chem. 1973, 3, 39.
Optical Studies of Silver Clusters and Larger Surface Structure Evolution on a Silver Electrode during Electrochemical Cycling C. D. Marshall and G. M. Korenowski* Department of Chemistry, Rensselaer Polytechnic Institute, Troy, New York 121 80-3590 (Received: September 1 1 , 1986; In Final Form: January 5, 1987)
Optical second-harmonic generation and laser-induced luminescence are used to follow the formation and removal of silver electrode surface structure during an oxidation-reduction cycle in a NaF electrolyte. Small silver clusters, detected by luminescence, form during surface reorganization following dissolution and deposition of silver. The clusters slowly evolve into larger structures that enhance second harmonic generation. The changing character of the surface is found to persist for a few minutes following silver deposition. These studies point out the potential importance of metal clusters and evolving surface structure in silver electrode chemistry and a method of studying this structure.
Introduction both macroAn oxidation-reduction cycle (ORC) can scopic and atomic scale roughness on the silver electrode surface. This roughness can have major effects on the physical and chemical behavior of the electrode. MacroscoDic roughness of 1-100-nm size is known to give rise to a classical electromagnetic
-
0022-3654/87/2091-1289$01.50/0
enhancement of optical fields and is a source of enhancement for optical processes such as surface-enhanced Raman scattering (SERS) and reflected optical second-harmonic generation (SHG).'" Atomic scale roughness or surface silver clusters can (1) Gersten, J.; Nitzan, A. J . Chem. Phys. 1980, 73, 3023.
0 1987 American Chemical Society
