Surface enhanced Raman spectroscopic evidence that adsorption on

Joseph Deere, Edmond Magner, J. Gerard Wall, and B. Kieran Hodnett ... C. Roland Wolf , John S. Miles , Sepalie Seilman , M. Danny Burke , Bernard N...
6 downloads 0 Views 615KB Size
5168

J . Phys. Chem. 1985,89, 5168-5173

Surface Enhanced Raman Spectroscopic Evidence That Adsorption on Sllver Particles Can Denature Heme Proteins Giulietta Smulevicht and Thomas G. Spiro* Department of Chemistry, Princeton University, Princeton, New Jersey 08544 (Received: February 15, 1985; In Final Form: June 7 , 1985)

Raman spectra are reported for micromolar solutions of the heme proteins hemoglobin (Hb), cytochrome b5 (cyt b), and cytochrome c (cyt c), absorbed on aqueous silver sols, using 413.1-nm excitation, in resonance with the heme Soret band. The surface enhanced Raman scattering (SERS) spectra of HbOz and of cyt b show distinct changes relative to the solution resonance Raman spectra of these molecules. These changes clearly indicate formation of surface bound hemin pox0 dimers, implying that the heme prosthetic groups have been extracted from their binding pockets in at least some of the protein molecules. The extent of dimer formation is preparation-dependent and may depend on silver particle aggregation, as well as the surface potential, which are difficult to control. Dimer formation is less pronounced for cytochrome c but is still readily observed; it is suggested that Ag' ions at the surface catalyze the cleavage of the heme-protein thioether bonds. The reduced forms of these proteins showed SERS spectra which were similar to the solution RR spectra, although some conversion to Fe"' was generally observed; no w-oxo dimer formation was detected. These results indicate that heme extraction is facilitated under oxidizing conditions, perhaps via increased surface charge on the Ag surface. The heme environment is unperturbed for reduced cyt c, as judged by the rich low-frequency SERS spectrum. For deoxy hemoglobin, however, the Fe-imidazole stretching band appears to shift from 215 to 200 cm-' in the SERS spectrum, suggestingsome perturbation of the hemeprotein linkage.

Introduction The discovery that Raman scattering cross sections can be enormously enhanced when the molecules under study are adsorbed on roughened silver electrodes' or silver colloids2 has generated great interest in the potential of surface enhanced Raman scattering (SERS) spectroscopy for chemical and biochemical studies. In the case of biological molecules, the advantages of high sensitivity and quenching of fluorescence by the metal surface3 are attractive. A case in point is the recent demonstration from this laboratory that flavoproteins adsorbed on silver colloids produce high-quality resonance Raman (RR) spectra at micromolar concentration, with no interference from the flavin fluorescence that normally obscures R R spectra of the molecules in s ~ l u t i o n .While ~ some spectral perturbations were observed, suggesting direct interaction of the flavin chromophores with the silver surface, it was possible to demonstrate nearly undiminished activity for glucose oxidase while bound to the colloid, implying that the enzyme active site remained i n t a ~ t . ~ While this result is encouraging for biological applications, the possibility of structural alteration induced by surface forces remains a concern that must be addressed in each individual casc. In the present study we find that this concern is justified for heme proteins. In several cases, the SERS spectra produced by heme proteins adsorbed on silver colloids are not those of the native protein; they are instead identified as arising from surface bound p-oxo-bridged iron porphyrin dimers, which have been well characterized in previous work.5 These species can only have been formed by disruption of the heme-binding pockets of the proteins. SERS spectra for heme proteins6 and bacterial photoreaction centers' adsorbed on roughened silver electrodes have been reported by Cotton et al. In the case of myoglobin and ~ ~ spectral differences were observed, relative cytochrome c , some to solution RR spectra, but it was unclear whether these differences were induced by the field of the electrode or were indicative of protein structural alterations. The latter might have been produced by adsorption per se or by laser-induced damage to the electrode surface.* In the present study we examine silver colloid suspensions, which are less subject to laser damage since individual particles are moved rapidly out of the laser beam in the spinning sample cell employed. The results establish that heme proteins are susceptible to denaturation by adsorption to the silver surface, 'Permanent address: Laboratorio di Spettroscopia Molecolare, Dipartimento Di Chimica. Universitl Di Firenze, 50121 Firenze, Italy.

0022-3654/85/2089-5168$01 SO10

at least in their oxidized forms. The reduced forms appear to be largely intact in our experiments, although the examination of the low-frequency SERS spectrum of deoxyhemoglobin suggest that the Fe-imidazole linkage is significantly perturbed.

Experimental Section Hemoglobin was prepared from fresh human red blood cells by the method of Perutze9 The protein was deoxygenated by stirring under N2 at 4 "C. Sperm whale skeletal muscle myoglobin (11) was purchased from Sigma and purified by preparative isoelectric focussing. Cytochrome b5 was a gift from Dr. John Schenkman, University of Connecticut, Farmington. Cytochrome c was purchased from Sigma. Oxidized heme proteins were converted to reduced forms by passing N2 over the surface of the solution for 10-15 min and adding a small amount of sodium dithionite. The p-oxo dimer of Fe"' protoporphyrin(1X) dimethyl ester was prepared by dissolving Fe"'(PPDME)Cl in CH2C12 through 5% deactivated alumina and eluting with 10% CH,OH in CH2C12. Silver sols were prepared according to Creighton,2 by mixing 5 mL of 1 X M A g N 0 3 with 10 mL of ice cold 8 X M NaBH,. After 2 min, 0.2 pL of 0.1 M sodium citrate was added to stabilize the sol; after 1 h, 5 pL of heme protein solution (3 X 10-5-3 X M) was added to 1.5 mL of the colloid, which changed from opalescent yellow to orange. For Fe" proteins, N, was passed gently over the surface of the sols for 3 h to remove 0 2 , before the samples were added in a glovebag containing N,, M) was added to after which 20 pL of fresh NaBH, (3 X increase the Fe"/Fe"' ratio; a mixture of oxidation states was nevertheless usually observed. (1) (a) Jeanmaire, D. L.; VanDuyne, R. P. J . Elecfroanal. Chem. 1977, 84, 1-20. (b) Albrecht, H. G.; Creighton, J. A. J . Am. Chem. SOC.1977, 99, 52 15-52 18. (2) Creighton, J. A,; Blatchford, C. G.; Albrecht, M. G. J . Chem. Soc., Faraday Trans 2 1979, 75, 790-798. (3) Lippitsch, M. E. Chem. Phys. L e f f .1981, 79, 2. (4) Copeland, R. A,; Fodor, S. P. A,; Spiro, T. G. J . Am. Chem. SOC.1984, 106, 3872-3874. (5) Sanchez, L. A.; Spiro, T. G. J . Phys. Chem. 1985, 89, 763. (6) (a) Cotton, T. M.; Schultz, S. G.; VanDuyne, R. P. J . Am. Chem. SOC. 1980,102,796C-7962. (b) Cotton, T. M.; Timkovich, R.; Cork, M. S . , FEES Lett. 1981, 133, 39-44. (7) Cotton, T. M.; VanDuyne, R. P. FEES Left. 1982, 147, 8 1 . (8) Mahoney, M. R.; Cooney, R. P. J . Phys. Chem. 1983,87,459&4591. (9) Perutz, M. F. J . Crysf.Growth 1968, 2, 54-56.

0 1985 American Chemical Society

The Journal of Physical Chemistry, Vol. 89, No. 24, 1985 5169

Adsorption of Heme Proteins on Ag' Particles

h0~413.1nm

(Fam PPI 0/ CH,CI,

F

?!

-00 C CH 2 CHp CH3 Figure 2. Structural diagram for a metalloprotoporphyrin.

11

1 1300

1

oxyHb SER.S ! '

1

I

A

cm-1

I

i

I

1500

Figure 1. Raman spectra with 413.1-nm Kr' laser excitation of (a) oxyHb (3 X IO4 M) in aqueous solution, (b) and (c) oxyHb (1 X IOd M) on different Ag sols, (d) the p o x 0 dimer of Fe"' protoporphyrin(1X) dimethyl ester (3 X IO4 M) in CH2C12. Experimental conditions: laser power 60mW at the source, spectral slit width 5 cm-I, accumulation time 1 s/cm-'. Dashed lines indicate bands due to oxyHb and dot-dashed lines indicate bands due to p-oxo dimer.

Raman spectra were obtained by backscattering from slowly rotating NMR tubes (rapid rotation tended to settle the sols), using a 413.1-nm line of a Kr+ laser (Spectra Physics 171). The scattered light was collected and focussed into a computer-controlled Spex 1401 double monochromator, equipped with a cooled photomultiplier (RCA) and photon-counting electronics.

Results and Discussion Hemoglobin and Myoglobin. Figure 1 compares a solution RR spectrum of oxyhemoglobin ( H b 0 2 (a)), with two colloid SERS spectra. Since the protein concentration was 3 X lo4 M for the M for the colloid spectra, solution R R spectrum, but 1 X it is evident that a large surface enhancement is being observed. The two SERS spectra were obtained with different colloid preparations and show marked alterations from each other, and from the aqueous H b 0 2 R R spectrum. Spectrum c, however, can readily be attributed to a p-oxo heme dimer, as can be seen by comparison with the R R spectrum (d) of (Fe1''PPDME)02. This spectrum is the same as that for aqueous hemin p-oxo dimer (nonesterified) and of its SERS spectrum obtained on a silver ele~trode.~ Particularly characteristic of these species is the intense band at 1489 cm-l, assigned to the porphyrin skeletal mode v3; the frequency is typical for high-spin 5-coordinate Fe"' hemes,I0 while the intensity is especially marked for the p-oxo dimer. The other bands, 1371, 1568, 1585, and 1622 cm-I, also arise from skeletal modeslo (v4, v2, v3*, and vl0) and are found at the same positions in spectrum c. For Hb02, which contains low-spin heme, these modes are found at distinctly shifted frequencies: 1374, 1500, 1580, 1600, and 1635 cm-l for v4, v3, v 2 , v3*, and vlo. The other bands seen in the H b 0 2 spectrum are vinyl modes (1423 and 1617 cm-I) and an additional porphyrin skeletal mode 1550 cm-I). The SERS spectrum of the second colloid, spectrum b, can be seen to be a composite of H b 0 2 and p-oxo dimer contributions. Two v 3 bands are clearly seen at the expected frequencies, 1488 and 1500 cm-I, while the broad envelopes centered at 1575 and 1621 cm-l contain overlapping contributions from the two species. Other colloid preparations gave spectra containing varying contributions from H b 0 2 and the p-oxo dimer. Since the heme group in H b 0 2 is low spin, the appearance of a high-spin Fe"' signal in the surface spectra means that the dioxygen ligand has been lost, apparently as superoxide, leaving Fe"'. Superoxide generation from oxyHb is a known mode of autoxidation and is accelerated by the addition of exogeneous agents, e.g. CuZ+.lla It is possible that interaction with the positively charged Ag surface likewise induced superoxide generation and autoxidation. The identification of the high-spin Fe"' signal with p-oxo dimer further implies that the autoxidized protein releases heme from its binding pockets to the surface, where two hemes can interact to form dimer. This surface reaction is plausibly mediated by the propionate peripheral substituents of the protoporphyrin ring (see Figure 2), which can facilitate heme adsorption to the silver surface. The carboxylate groups might be expected t o produce a SERS band near 1400 cm-I, as is observed for citrate on silver sols,IIbbut the carboxylate mode is not subject to resonance enhancement and is probably too weak to observe in the presence of the enhanced porphyrin spectrum. (IO) Choi, S.; Spiro, T. G.; Langry, K. C.; Smith, K. M.; Budd, D. L.; LaMar, G . N. J. Am. Chem. SOC.1982, 104, 4345-4351. ( I 1) (a) Winterbourn, C. C.; McGrath, B. M.; Carrel], R. W. Biochem. J . 1976, 155, 493-502. (b) Siiman, 0.; Bumm, L. A,; Callaghan, R.; Blatchford, C. G.; Kerker, M. J . Phys. Chem. 1983,87, 1014.

Smulevich and Spiro

5170 The Journal of Physical Chemistry, Vol. 89, No. 24, 1985

Xon

(D

-Em

413.1 nm

d

n

I

oxyHb SERS

I

1. deoryHb SERS f . . !

- 1 150

I

450

450

!

! !

I

I

I

750

A

j

i~ . .

I

I

I

!

.-

.

.

I 1050

I

,

--

I

I I350

I

I 1650

cm-1

M) on Ag M) in aqueous solution, (b) deoxyHb (1 X Figure 3. Raman spectra with 413.1-nm Kr+ laser excitation of (a) deoxyHb (3 X sol, and (c) oxyHb (1 X 10" M) on Ag sol. Experimental conditions: laser power 60 mW at the source, spectral slit width 5 cm-', accumulation time 5 s/cm-' between 150 and 50 cm-l and 1 s/cm-l between 450 and 1700 cm-'. The dashed lines indicate bands due to deoxyHb and the dot-dashed lines indicate bands due to Fe"' heme.

It is likely that the rate of the surface reaction depends on the state of aggregation of the Ag particles, and that particle aggregation in turn varies from sample to sample, thus accounting for the variation in the spectra. All of the sols eventually precipitated, but at different times. We attempted to monitor the time course of the protein denaturation but were frustrated by the relatively rapid decay of the spectra, associated with sol precipitation. It is notable that spectrum b, showing a mixture, is much weaker (see the lower signal/noise) than spectrum c, in which the p-oxo dimer dominates. A reasonable interpretation is that native protein shows much smaller surface enhancement than p-oxo dimer, perhaps because of less favorable orientation or contact with the surface. If this is the case, then the relative intensities of the Raman bands are misleading with respect to the extent of denaturation; a small fraction of released hemes would be capable of dominating the SERS spectrum. Figure 3 compares the R R spectrum of aqueous deoxyHb with the SERS spectra of oxy- and deoxyHb on silver sols. This oxyHb sol gave a p-oxo dimer spectrum, as seen by the characteristic appearance of the high-frequency bands. The SERS spectrum of the deoxyHb sol shows all of the bands seen in the deoxyHb R R spectrum, but additional bands are seen, which are readily attributable to a low-spin Fe"' heme. The indicators for this assignment are the 1370-cm-I shoulder, the Fe"' position for v4 (oxidation state marker) band, and the clearly resolved band

-

at 1500 cm-I, the vj position for low-spin Fe"' heme.l0 We found it impossible to produce a pure Fe" spectrum, even with careful degassing and the presence of a small excess of NaBH, (a large excess leads to sol precipitation). Evidently, the surface potential of the silver particles is sufficient to maintain the Felll/rlratio at nonnegligible values, or else there is strong specific absorption of the Fe"' species to the silver. A high surface charge might also be associated with high local pH, a factor that could accelerate p-oxo dimer formation for the oxyHb sols. Since oxidation (as opposed to oxygenation) of deoxyHb produces the high-spin met-aquoHb, the observation of a low-spin Fe"' signal means either that sufficient O2 is retained, perhaps by surface absorption, to produce HbO,, or that a low-spin hemichrome is formed at the surface, perhaps by coordination of the distal imidazole of Hb. It is notable that there is no trace in the deoxyHb SERS spectrum of the 1488-cm-I band which is characteristic of the p-oxo dimer. This implies that the oxidized heme which is present has not been released from the protein. If any of the reduced heme had been released from the protein, it probably would have been oxidized to the p o x 0 dimer, whose reduction potential at silver is quite n e g a t i ~ e .It~ therefore appears that deoxyHb is less prone to releasing its heme group to the silver surface than is (autoxidized) oxyHb. The left-hand panel of Figure 3 shows the spectra between 150 and 450 cm-I. The relatively prominent band in the deoxyHb RR

The Journal of Physical Chemistry, Vol. 89, No. 24, 1985

Adsorption of Heme Proteins on Ag+ Particles

SER S

5171

"20

h0=413.1nm

I I300

I

1

1500

1

I 1700

I

A cm-1

1.300

I

I

1500

I

I 1700

Figure 4. Raman spectra at the indicated excitation wavelengths of (a) Fe"' cyt b (1 X 10" M) on Ag sol, (b) Fe" cyt b (1 X lod M) on Ag sol, (c) Fe" cyt b (1 X 10-4 M) in 20 mM tris acetate buffer (pH 8.1), and (d) Fe"' cyt b (1 X lo4 M) in 20 mM tris acetate buffer (pH 8.1). Experimental conditions: (a) and (b) laser power 70 mW at the source, spectral slit width 5 cm-l, accumulation time 1 s/cm-', (c) and (d) laser power 35 mW at the sample, spectral slit width 6 cm-I, accumulation time 2 s/0.5 cm-I. The (c) and (d) samples were kept at 4 OC with a flow of cold N,.

spectrum at 21 5 cm-I has been assigned12to the stretching mode of the bond between the heme Fe and the imidazole ring of the proximal histidine. It is quite sensitive to the details of the protein structure and is known to shift up to -220 cm-' when the quaternary structure is switched from T (tense) to R (relaxed) by chemical m o d i f i ~ a t i o nor~ ~kinetic methods.I4 This region of the oxyHb R R spectrum is featureless. Therefore, while the deoxyHb SERS spectrum contains both Fe" and Fe"' contributions, only the Fe"-imidazole stretch is expected between 200 and 250 cm-I. This spectrum (b) shows a band at 200, not 215 cm-I. The apparent shift suggests an appreciable weakening of the Feimidazole bond. Thus, although deoxyHb appears to retain its heme groups upon adsorption to the silver surface, there is evidence, nevertheless, for significant perturbation of the hemeprotein linkage. The 200-cm-' band is quite broad, suggesting heterogeneity among the four heme groups of the H b tetramer; this is reasonable, since it seems unlikely that all four heme groups of a particular H b molecule could be simultaneously in contact with the surface. Myoglobin showed the same SERS characteristics as hemoglobin. Met-aquo Mb showed facile k-oxo dimer formation, while deoxyMb did not, but its Fe-imidazole stretching band was lowered to 200 cm-', as in the case of deoxyHb. Cytochrome b,. Like hemoglobin, cytochrome bS (cyt b) contains noncovalently bound protoporphyrin(IX), but instead of an open coordination site the heme Fe is bound to a second his(12) (a) Kincaid, J.; Stein, P.; Spiro, T. G . Proc. Narl. Acad. Sci. U.S.A. 1979, 76,4156. (b) Kitagawa, T.; Nagai, K.; Tsubaki, M. FEBS Left. 1979, 104, 376. (13) Nagai, K.; Kitagawa, T. Proc. Narl. Acad. Sci. U.S.A. 1980, 77, 2033-2037. (14) Stein, P.; Terner, J.; Spiro, T. G. J . Phys. Chem. 1982,86, 168-170.

tidine ligand, and is low spin in both oxidized and reduced forms.I5 The R R spectra of cytochrome b5 in aqueous solution, shown in the right-hand panel of Figure 4, are typical of low-spin Fe" and Fe"' hemes. The left-hand panel shows SERS spectra for silver sols loaded with oxidized and reduced cyt b at micromolar concentration. The oxidized spectrum (a) is entirely different from the R R spectrum of the oxidized protein (c), and it is again identical with the pox0 dimer spectrum (compare Figure 1). Thus oxidized cyt b, like HbOz, is prone to lose its heme group to the silver surface, although no prior autoxidation step is needed in this case. As with deoxyHb, the SERS spectrum of reduced cyt b shows a mixture of Fe" and FelI1contributions, the latter being particularly evident in the 1371-cm-' v, shoulder and the 1500-cm-I v 3 band. This is the expected spectrum for oxidized cyt b, and it is notable that there is no evidence for p o x 0 dimer formation in this spectrum (no 1490-cm-' band) even though spectrum a shows Fe"' cytochrome b5 to be converted to the oxo dimer at the silver surface. Evidently the reducing conditions under which spectrum b were obtained had something to do with the stability of the oxidized protein a t the silver surface. Cyrochrome c. Like cytochrome b5,cytochrome c (cyt c) is low spin in oxidized and reduced forms; it has two endogenous ligands, in this case histidine and methionine.16a In addition, the protoporphyrin ring is covalently linked to the protein via condensation of its peripheral vinyl groups with cysteine side chains. The saturation of the vinyl groups produces a considerable simplification of the high-frequency Raman spectra,l' since the vinyl (15) Ikeda, M.; Iizuka, T.; Takao, H.; Hagihara, B. Biochim. Biophys. Acta 1974, 336, 15-24. (16) (a) Dickerson, R. E.; Timkovick, R. In "The Enzymes", Boyer, P., Ed., 3rd ed.,Vol. 11; Academic Press: New York, 1975: pp 397-567. (b) Paul, K-G. Acta Chem. Scand. 1950, 4, 239-244.

5172 The Journal of Physical Chemistry, Vol. 89, No. 24, 1985

Smulevich and Spiro

A,= 406.7nm

Cyt c

c"

P

Fern/ SER S

Fex/ H20

Feu SERS

-- -

150

350

300

I300

I500

no0

Figure 5. Raman spectra with 406.7-nm Kr+ laser excitation of (top to bottom) Fe"' cyt c on Ag sol (30pM) and in H 2 0 (2.2 mM), Fe" cyt c on Ag sol (30 pM) and in H20(2.2 mM). Experimental conditions: laser power 50 mW at the source, spectral slit width 5 cm-I, accumulation time 3 s/cm-' between 150 and 350 cm-' for the sol spectra and 1 s / l cm-' for all the other regions.

modes, and also the vinyl-activated infrared (K) skeletal modes,I0 are missing. This is illustrated in the right-hand panel of Figure 5, which shows high-frequency RR and SERS spectra for reduced and oxidized cyt c. The five peaks in the Fe" RR spectrum, 1359, 1490, 1554, 1589, and 1620 cm-I, are skeletal modes v4, v3, v I I , vZ, and vIo. They all show characteristic frequency shifts in the Fe"' RR spectrum (whose 1359-cm-I v4 shoulder reveals a small Fell component, reflecting the high redox potential of this protein). The Fe" SERS spectrum is the same as the Fe" RR spectrum, although vIo and v I 1 are too weak to identify clearly. Likewise the Fe"' SERS spectrum is the same as the RR spectrum (although there is no longer a Fe"-derived v4 shoulder, again suggestive of a more oxidizing environment at the silver surface) except that an additional band appears a t 1490 cm-l. Figure 6 shows Fe"' SERS spectra for four different sols, arranged in order of increasing 1490-cm-' contribution. As the 1490-cm-' band grows, so do additional bands at 1570 and 1622 cm-I. These are all characteristic frequencies for the p-oxo dimer (see Figure l), and we conclude that Fell' cyt c, like Fe"' cyt b and H b 0 2 , is susceptible to p-oxo dimer formation at the silver surface, albeit (17) Strekas, T.C.;Spiro, T.G.Eiochim. Eiophys. Acta 1972, 278, 188.

to a smaller extent, since none of the cyt c SERS spectra are completely dominated by p-oxo dimer bands as was the case with the other two proteins. Production of p-oxo dimer from cytochrome c is a surprising result, considering that the porphyrin is attached covalently to the protein. The heme groups must emerge from the protein pockets, and if the thioether links are maintained then p-oxo dimer formation implies close contact of two protein molecules, perhaps requiring appreciable denaturation. More likely the thioether links are hydrolyzed, releasing the hemes to form p-oxo dimer. We note in this connection that silver salts under mild conditions promote thioether bond cleavage and the release of heme from cytochrome c , and ~ that ~ silver ~ sols are thought to be coated with Ag' ions.'Ib Thus catalysis of heme release at the silver surface is plausible. The Ag' ions might migrate to the interior of the protein and attack the thioether bonds, or the emergence of the heme from the crevice via interactions with the surface might bring the thioether bonds into contact with surface-bound Ag+. The latter mechanism seems more probable. We note that glucose oxidase retains most of its enzymatic activity when bound to silver sol, even though Ag+ salts are potent inhibitors of the enzyme, suggesting that the surface-bound Ag+ ions are not m ~ b i l e .As ~

The Journal of Physical Chemistry, Vol. 89, No. 24, 1985 5173

Adsorption of Heme Proteins on Ag' Particles

Cytc Fe'

SERS

h,,= 406.7 nm

are simpler, and the extra bands observed in the protein are probably due to activation of out-of-plane skeletal modes and perhaps peripheral substituent modes due to protein induced distortions.'O It is therefore significant that the SERS spectrum of Fe" cytochrome c (Figure 4) retains all of the low-frequency bands of the R R spectrum, with only a slight loss in resolution near 400 cm-I. This result establishes that there is very little, if any, disturbance of the native protein structure, at least in the vicinity of the heme group, upon binding of Fe" cyt c to the silver particles. In the case of Fe"' protein, however, substantial alterations are noted in the low-frequency SERS spectrum (Figure 5). The prominent 397-cm-I R R band is halved in intensity, while the 303-cm-I band has disappeared, and the 265-cm-I band has grown. Figure 6 shows that further changes are associated with increasing p-oxo dimer contribution; both the 395- and 412-cm-' bands lose intensity. The p-oxo dimer itself shows only weak scattering in the low-frequency region, on the scale of the protein SERS bands, and it is likely that these intensity alterations are associated with heme groups remaining in the protein crevice, and reflect perturbations of the protein structure itself.

II

Conclusions

d

C I

I

b

-

I

I

I

I

I I

t 1

I

I

I I I

I

y I

group can be released from the protein, not only for H b and cyt b, which contain noncovalently bound protoporphyrin(IX), but also cyt c, in which the heme periphery is linked to the proton by two covalent bonds. We surmise that this process is facilitated by interaction of the heme propionate groups with the positively charged silver surface. Since both propionate groups are on the

300

ULIII '

Figure 6. Raman spectra with 406.7-nm Kr' laser excitation of Fe"' cyt c (3 X IO-' M) on Ag sols. (a), (b), and (c) are three different colloids. (d) is the same sol as (c) after 2 h of laser illumination. Experimental conditions: laser power 50 mW at the source,spectral slit width 5 cm-I, accumulation time 1 s / l cm-I. The dashed lines indicate the bands attributed to p o x 0 dimer.

with cyt b, heme release appears to be specifically associated with the oxidized protein. Were the heme to be released from its crevice in the Fer*protein, its redox potential would be lowered to that of aqueous heme, and oxidation to the p-oxo dimer would have been observed. It is known that reduced cytochrome c is less compressible,I8 and exchanges amide protons more slowly,19than oxidized protein. The low-frequency R R spectra of cytochrome c are richly as is shown by the center and left-hand panels in Figure 4. In the same region, the spectra of aqueous heme complexes (18) Eden, D.; Matthew, J. B.; Rosa, J. J.; Richards, F. M. Proc. Natl. Acad. Sci. U.S.A. 1982, 79, 815-819. (19) ,(a) Ulmer, D. D.; Kagi, J. H. R. Biochemistry 1968, 7, 2710-2717. (b) Kaai, J. H. R.;Ulmer. D. D. Biochemistry 1968, 7. 2718-2723. (c) Patel, D.'J.; canuel, L. L. Proc. Natl. Acad. Sc;. U.S.A. 1976, 73, 1398-1402. (20) Yu, N-T.; Srivastava, R. B. J . Raman Spectrosc. 1980,9, 166-171. (21) Choi, S.; Spiro, T. G. J . Am. Chem. SOC.1983, 105, 3683-3692.

though the surface-bound reduced proteins are revealed by the SERS spectra to be partially oxidized. It may be that the surface charge, and perhaps the local pH, of the particles are lowered sufficiently under these conditions that heme extraction is inhibited. Although the heme group appears to remain imbedded in the proteins when they are reduced, clear evidence for perturbation of the heme-protein linkage is seen in the Fe-imidazole Raman band of deoxyHb, which is down shifted and broadened in the SERS spectrum. Only in the case of Fe" cyt c does the R R spectrum, which contains several protein-induced low-frequency bands, show positive evidence for an unperturbed heme environment when the protein is bound to silver particles. This may be associated with the exceptional compactness and rigidity of this form of the protein."

Acknowledgment. This work was supported by a grant from the Italian Consiglio Nazionale Delle Ricerche (to G.S.) and by N I H Grant G M 33576 (to T.G.S.). W e thank Dr. John Schenkman for the generous gift of cytochrome b5, Dr. Ruby Evangelists-Kirkup for help with the cytochrome b5spectra, and a referee for bringing to our attention the possibility of Ag' catalysis of heme release from cytochrome c. C,

Registry No. Ag, 7440-22-4; cytochrome b5,9035-39-6; cytochrome 9007-43-6.