2714
J. Phys. Chem. 1981, 85, 2714-2717
Picosecond and Nanosecond Resonance Raman Studies of Bacteriorhodopsin. Do Configurational Changes of Retinal Occur in Picoseconds? Chung-Lu Hsleh, Mark Nagumo,+Malcolm Nicol, and M. A. El-Sayed’ Department of Chemistry, University of California, Los Angeles, California 90024 (Received: June IS, 198 1; In Final Form: July 27, 198 1)
The picosecond and nanosecond resonance Raman spectra of bacteriorhodopsin (bR) have been determined by microbeam and flow techniques. The 40-ps pulses were generated by a synchronously pumped mode-locked, cavity-dumped dye laser. The pulse width is extended to 15 ns by CW pumping the cavity-dumped dye laser. The spectrum of the picosecond transient in the fingerprint region resembles the bRFh spectrum. The retinal in the latter species is believed to be in the 13-cis configuration. These results suggest that the retinal changes from an “all-trans”- to a “13-cis”-typeof configurationin less than 40 ps (the pulse width of the photolyzing laser). Slight differences are observed between the picosecond and nanosecond retinal spectra. These observations suggest that more than one species exist in the 40-ps to 100-ns time scale.
Introduction
cm-l from spectra taken at 40-ps, 15-ns, and 100-ns time scales is discussed.
Bacteriorhodopsin (bR) is a chromoprotein and the pigment of the purple membrane of some halophilic bacExperimental Section teria. The chromophore of bR is retinal in either the all-trans or 13-cis configuration.lO2 Other isomers of retinal Bacteriorhodopsin was produced from the culture meare not f ~ u n d . ~ BUpon illumination, bR undergoes a dium of the H. halobium S-9 mutant strain. Growth and isolation of the purple membrane were based on the known photochemical cycle (Figure 1)involving proton release and uptake from opposite sides of the membrane. Several procedure.16 The sample was suspended in deionized reviews give detailed descriptions of the functional, water at -50 ELM.The carotenoid content of the sample structural, kinetic, and spectroscopic properties of bR.4-7 was monitored by the resonance Raman spectra; only those The primary event in the photochemical cycle has been samples with no detectable carotenoid bands at 1155 and the focus of many recent studies. The energy stored in this 1515 cm-’ were used for these studies. step is used to complete the light-induced cycle of bR and The experimental arrangement for recording the Raman to support the energy for transport processes across the spectra involved mi~robeam’~ and flow techniques as decell membrane. While biochemical e x t r a c t i ~ n and ~ ~ ~ * ~ scribed previously,14except that the CW laser was replaced resonance Raman spectroscopic studies’O suggest that by a pulsed laser system18 that was used to obtain the all-trans to 13-cis isomerization of the retinal may occur picosecond and the nanosecond resonance Raman spectra during the photochemical cycle, the exact step in the cycle of bR in a manner similar to the one used to obtain the at which isomerization occurs is still controversial. In resonance Raman spectra of carboxyhem~globin.~~ The particular, the belief that isomerization occurs during the pulsed laser system was a mode-locked argon-ion laser first step on the picosecond time scale has been questioned recently.” (1)Oesterhelt, D.; Meentzen, M.; Schuhmann, L. Eur. J . Biochem. Time-resolved resonance Raman spectroscopy is a ver1973,40,453. satile and useful technique for examining the dynamics (2) Pettei, M. J.; Yudd, A. P.; Nakanishi, K.; Henselman, R.; Stoeckand structures of biological systems (for a general review, enius, W. Biochemistry 1977,16,1955. (3)Dencher, N. A. Rafferty, C. N.; Sperling, W. Ber. Kernforchungsee ref 12 and 13). Previous studies have suggested that slage Julich 1976,Jul-1374,1-42. retinal undergoes a large isomeric change in the 100-ns (4)Henderson, R. Annu. Reu. Biophys. Bioeng. 1977,6, 87. time scale.14 In the work described here, we attempted (5)Lanyi, J. K.Microbiol. Reu. 1978,42,682. (6)Stoeckenius, W.; Lozier, R. H.; Bogomolni, R. A. Biochim. Biophys. to determine whether configurationalchanges of the retinal Acta 1979,505, 215. occur on a shorter time scale. Resonance Raman spectra (7)Ottolenghi, M. In “Advances in Photochemistry”, Pitts, J. N. Jr.; in the 40-ps and the 15-11s time scales have been obtained; Hammond, G. S.: Gollnick.. K.:. Grosiean, - D., Eds.: Wilev: New York, 1980;pp 97-200. and, in this Letter, we describe the fingerprint region of (8) Mowery, P. C.; Stoeckenius, W. Biochemistry 1980,20, 2302. these spectra. Furthermore, in order to identify the (9)Tsuda, M.; Glaccum, M.; Nelson, B.; Ebrey, T. G. Nature (London) isomeric form of the retinal in the picosecond intermediate, 1980,287,351. (10)Braiman, M.; Mathies, R. Biochemistry 1980,19, 5421. we compared its fingerprint spectrum with that of bRib. (11) Applebury, M. L.; Peters, K. S.; Rentzepis, P. M. Biophys. J . The picosecond spectrum is found to be similar to the 1978,23, 375. spectrum of bRk$, which was previously determined15 by (12)Mathies, R. In “Chemical and Biological Applications of Lasers”, Moore, C. B., Ed.; Academic Press: New York, 1979;Vol. 4, pp 55-99. subtraction techniques from the unphotolyzed spectrum El-Sayed, M. A. ACS Symp. Ser. 1979,102,215-27. (13) of li ht-adapted bR,,, (all-trans retinal) and dark-adapted (14)Terner, J.;Hsieh, C.-L.; Burns, A. R.; El-Sayed, M. A. Roc. Natl. bR[%5 !equal amounts of all-trans- and 13-cis-retinal). Acad. Sci. U.S.A. 1979, 76, 3046. (15)Terner, J.;Hsieh, C.-L.; El-Sayed, M. A. Biophys. J. 1979,26,527. bR,,, is thought to contain “13-cis” retinal. The time (16)Becher, B. M.; Cassim, J. Y. Prep. Biochem. 1965,5, 161. evolution of certain Raman bands in the region 950-1260 (17)Berns, M. W. E x p . Cell Res. 1971,65,470. ’
‘Department of Chemistry, Columbia University, New York, NY 10027. 0022-3654/8 1/2085-27 14$0 1.25/0
(18)Nicol, M.; Hara, Y.; Wiget, J. M.; Anton, M. J. Mol. Struct. 1978, 47,371. (19) Terner, J.; Strong, J. D.; Spiro, T. G.; Nagumo, M.; El-Sayed, M. A. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 1313.
0 1981 Arnerlcan Chemical Soclety
The Journal of Physical Chemistry, Vol. 85, No. 19, 1981 2715
Letters h y L j R r ( d o r k adopted)
Q
bR
576
(RLL-TRANS)
40-PS TRANSIENT
2psec
15-NS TRANSIENT
Flgure 1. The proton-pumping photochemical cycle of bacteriorhodopsin. The subscripts on each intermediate are the absorption maxima in nanometers. Half-times and absorption maxima are given for room temperature. (A recent proposes that the first few steps are to be modified to bR570 -+ I -+ J625 bK610, see text.)
100-NS TRANSIENT
-
(Spectra-Physics, Model 171) which synchronously pumped an extended dye laser cavity (Spectra-Physics, Model 375) fitted with a cavity dumper. At a repetition rate of 0.84 MHz, the average output power of the dye laser at 587 nm (Rhodamine 6G, Kodak) was 12 mW (15 nJ/ pulse). The pulse duration was 40 ps as measured by the two-photon mixing autocorrelation method.z0 The nanosecond pulses (-15 ns, 20 nJ/pulse) were generated by switching off the mode locker and cavity dumping the dye laser only. Each laser pulse was simultaneously used for photolysis and Raman excitation. The signal detection and the data processing methods have been described e1~ewhere.l~ The resonance Raman spectra are derived by computer subtraction of the spectrum of the unphotolyzed bR5,,, sample, obtained by using diffuse focus of a low-intensity CW dye laser, from the spectrum of photolyzed bR sample, obtained by using tight focus of the pulsed laser. The excitation wavelength of the laser is exactly the same for these two spectra. The 1256-cm-l band of bR57015was used to monitor the extent of the subtraction. Further detail will be discussed in a future publication. It is believed that the difference spectrum contains information of the intermediate during the photolysis period. Problems and some precautions about the subtraction techniques have been previously discussed.14J5 The reported vibrational frequencies are estimated to be accurate to f 3 cm-l in the bR570 spectrum and to *5 cm-l in all other spectra derived from computer subtraction.
Results The spectra to be analyzed are shown in Figure 2, in which all but spectrum a are obtained by using subtraction techniques. The spectra can be divided in three different regions: region a, the fingerprint region, is betweeen 1100 and 1260 cm-l; region b, centered about 1010 cm-'; and region c, the low region, below 1000 cm-l. The band near 1010 cm-l has been assigned to C-CH3 stretching vibrations,z1rz2mainly from the C(9) and C(13) positions of polyene chain (Figure 3). The 950-990-cm-l region contains mainly C-C-H bending vibrations,21although C-CH, stretching modes may also ~ o n t r i b u t e . ~ ~ The fingerprint region is sensitive to the isomeric con(20) Nagumo, M.Ph.D. Dissertation, 1981, University of California, Los Angeles. (21) Rimai, L.; Gill, D.; Parsons, J. L. J. Am. Chem. SOC. 1971, 93, 1353. (22) Cookingham, R. E.; Lewis, A. J. Mol. Bid. 1978, 119, 569. (23) Warshel, A.; Karplus, M. J . Am. Chem. SOC.1974,96, 5677.
Flgure 2. Resonance Raman spectra of different bacteriorhodopsin forms and photointermediates on the following time scales: (a) Unphotolyzed bR570with 587.0-nm excitation (thls work). This spectrum Is essentially the same as that taken with 514.5-nm e~cltation.'~(b) 40 ps with 587.0-nm excitation (this work). (c) 15 ns with 587.0-nm excitation (this work). (d) 100 ns with 552.3-nm excitation." (e) bR:to spectrum, 514.5-nm excitation.''
Figure 3. Configurations of the all-trans (bottom) and 13-cis (top) protonated Schiff base retinals.
figuration of the retinal.21 The main features in this region have been assigned to C-C stretching vibrations and other vibrations strongly coupled with the C-C mode^.^^^^^ Studies of the visual pigment in this region lent strong support for isomerization26nfrom cis- to trans-retinal upon photon absorption. The Raman studies have shown that the spectra of rhodopsin, isorhodopsin,26 and metarhodopsinz7closely resemble the spectra of the protonated Schiff bases of the 11-cis-, 9-cis-, and all-trans-retinal, respectively, in solution. These observations lead to the conclusion that the retinal chromophore in the visual pigment undergoes a photoinduced isomerization from an 11-cis or 9-cis form to an all-trans form. However, because of the stronger protein-retinal interactionz8in bR and the large similarity of the Raman spectra of the all-trans and ~ ~ comparisons S , ~ ~ ~ of ~ this ~ 13-cis model C O ~ ~ O U direct (24) Heyde, M. E.; Gill, D.; Kilponen, R. G.; Rimai, L. J.Am. Chem. SOC.1971,93,6776. (25) Cookingham, R. E.; Lewis, A.; Lemley, A. T. Biochemistry 1978, 17, 4699. (26) Mathies, R.; Freeman, T.B.; Stryer, L. J. Mol. Biol. 1977, 109, 267 , --.
(27) Doukas, A. G.; Aton, B.; Callender,R. H.; Ebrey, T. G. Biochemistry 1978, 17, 2430. (28) El-Sayed, M. A.; Terner, J. Photochem. Photobiol. 1979,30, 125. (29) Aton, B.; Doukas, A. G.; Callender, R. H.; Becher, B.; Ebrey, T. G. Biochemistry 1977, 16, 2995.
2716
Letters
The Journal of Physical Chemistry, Vol. 85, No. 19, 1981
kind are less conclusive in determiningthe type of isomeric change involved in the bR system. For this reason, it becomes imperative to make the comparison of the spectra of these stereoisomers within the protein pocket as outlined below. It is generally believed that bR570 contains all-transand bRF4 consists of an equimolar mixture of all-trans- and 13-crs-retinal.lJ If the all-trans-retinal is in the same form in bRF4 and in bR570, the spectrum remaining after subtraction of the proper amount of the spectrum of bR570 from that of bR& should represent a “13-cis” species which is designated as bRFh. The Raman spectrum of bRF& obtained in this manner is depected in Figure 2e. The following observations and conclusions can be drawn from the comparison of the different spectra shown in Figure 2. (1)The picosecond resonance Raman spectrum (spectrum b) is very different from that of bR570 (spectrum a). The bands change their relative intensities between these two spectra in all regions (950-1250 cm-l). This is strong evidence that the configuration of retinal changes in the 40-ps time scale. (2) The Raman spectrum of bRFh (spectrum e) is remarkably similar to the picosecond spectrum in the fingerprint region (spectrum b). The energies of the bands at 1170, 1183, and 1202 cm-l (spectrum e) appear to be increased by 6-10 cm-l in the picosecond spectrum to 1178, 1192, and 1208 cm-’ (spectrum b). These changes are on the order of solvent shifts. The results in the fingerprint region suggest that the retinal configurational changes may take place from an “all-trans”-type to a “13-cis”-type within the protein pocket. The observed band shifts indicate that the retinal environment of the “13-cis” form of the picosecond transients is not exactly the same as that in the bRF$ species. Similar conclusions concerning the isomerization of the retinal in the rhodopsin system in the picosecond time scale have recently been reported.30 (3) As the photolysis period increased (spectra b-d), the following changes in the spectra are observed: (i) In the fingerprint region, the 1192-cm-’ band remains as a strong feature from the 40-ps to the 100-ns spectra, while the bands at 1178 and 1208 cm-l in the picosecond spectrum merely appear as shoulders in the 15- and 100-ns spectra. (ii) In region b, the C-CH3 band is a relatively sharp, prominent feature at 1010 cm-’ in bR570 (spectrum a), bLm,15 bM412,” and bRFh (spectrum e). However, this band is rather broad and diffuse for the early stages of the photoinduced cycle in the 40-ps to 100-ns time scale (spectra b-d). These results suggest that one or more methyl group may adapt several low-energy sites within the protein pocket as soon as the retinal changes its configuration but before the protein has time to adjust. (iii) The spectra of both bR570 (spectrum a) and bRFh (spectrum e)15p32933 have a weak band around 960 cm-’; all other intermediates (spectra b-d and in ref 10 and 15) in the photochemical cycle except b064031 possess a distinct feature around 970 cm-’. The shift of this band to 970 cm-’ and the increase in its intensity are particularly evident in the spectrum of the picosecond transient (spectrum b). (30)Hayward, G.;Carlsen, W.; Siegman, A,; Stryer, L. Science 1981, 211,942. (31)Terner, J.;Hsieh, C.-L.; Burns, A. R.; El-Sayed, M. A. Biochernistry 1979,18, 3629. (32)Aton, B.; Doukas, A. G.; Callender, R. H.; Becher, B.; Ebrey, T. G.Biochirn. Biophys. Acta 1979,576,424. (33)Stockburger, M.; Klusmann, W.; Gattermann, H.; Massig, G.; Petters, R. Biochemistry 1979,18, 4886.
Since the bands in this low-energy region have not been completely assigned, any conclusion should be viewed as tentative. However, if these bands result from out-of-plane C-C-H vibrations, the observed relative intensification of these bands might suggest that the retinal configurations in the intermediates are more nonplanar than in their parent bR570.Such nonplanarity should increase resonance enhancement due to U,T* mixing.34 The twist configuration of the bathorhodopsin has been previously suggested from the Raman studies of the photochemistry of the rhodopsin s y ~ t e m . ~ ~ J ~ (iv) Although a 984-cm-l band is seen in the 100-ns transient (spectrum d), not much intensity can be observed in this region for the earlier transients (spectra b and c). to be deuterium Since the 984-cm-’ band has been sensitive, similar studies are currently under way in the 15-ns and 40-ps time scales. (v) The changes of the spectra observed when the photolysis time is changed from 40 ps to 100 ns (spectra b-d) provides evidence for the existence of more than one distinct intermediate during this interval. Recently, a new scheme has been p r ~ p o s e d for ”~~ the ~ early stages of the photocycle: bR570 -k. I
-
- 1Pa
11 pa
JGZ5
bK610
where the primary transient I is the precursor of J625and the dark reaction I 5625 precedes the formation of the deuteration-sensitive b&1@ Since our technique records an integrated Raman scattering from transients formed during the pulse duration, the differences in these spectra could result from different contributions to the observed scattering signal from the early intermediates (I, 5626, and bK610)*
In summary, our 40-ps resonance Raman spectrum of bacteriorhodopsin is similar to, but shifted from, the spectrum of bRi& in the fingerprint region. This result suggests that retinal changes from an “all-trans”- to a “13-cis”-type of configuration inside the protein in less than 40 ps. This shift is suggested to result from slight differences of the protein environments of the “134sretinal” between the picosecond transient and bRih. The increased intensity of the C-C-H out-of-plane bending vibrations suggests that the retinal become more nonplanar. The diffuse nature of the C-CH3 stretching modes might suggest that, in the short-lived transient, more than one minimum within the protein pocket are available to the methyl group(s). Finally, the fact that the spectrum changes as the photolysis period increases from 40 ps to 100 ns suggests that, in agreement with recent optical absorption r e s ~ l t s , more ~ ~ ~than ~ ~ one ~ ~ distinct * intermediate are formed during the first 100 ns of the photochemical cycle. The detail of this work, including the picosecond and the nanosecond resonance Raman spectrum in the ethylenic stretching region as well as the deuteration effects, will be published elsewhere soon.
(34)El-Sayed, M.A. In “Visual Pigments and Purple Membranes”, Methods in Enzymology Series, Biomembranes, Lester Packer, Ed.; Academic Press: New York, 1981. In press. (35)Eyring, G.; Curry, B.; Mathies, R.; Fransen, R.; Palings, I.; Lugtenburg, J. Biochemistry 1980,19, 2410. (36)Dinur, U.;Honig, B.; Ottolenghi, M. “Proceddings of the Italian Society for Pure and Applied Biophysics, Parma, Italy, October 1979”; Plenum Press: New York, 1980. (37)Ippen, E. P.;Shank, C. V.; Lewis, A.; Marcus, M. A. Science 1978, 200, 1279. (38) Kaufmann, K. J.; Sundstrom, V.; Yamane, T.; Rentzepis, P. M. Biophys. J. 1978,22,121.
The Journal of Physical Chemistry, Vol. 85,No. 19, 1981 2717
Letters
Acknowledgment. The authors thank Dr. R. A. Bogomolni and Professor W. Stoeckenius for the S-9 mutant strain and some samples. We are also grateful to Professor P. D. Boyer and his group for the use of their facilities to prepare the sample, and to Professor M. W. Berns for
lending us the microscope objectives. Support provided by the Department of Energy (Office of Basic Energy Sciences), National Science Foundation, the University Research Committee, and Spectra-Physics, Inc. is gratefully acknowledged.
ARTICLES An Analysis of the Hard-Soft Lewis Acid-Base Concept and the Drago Equation Employing ab Initio Molecular Orbital Theory John Douglast and Peter Kollman” Depattment of Pharmaceutical Chemistry, School of Pharmacy, Universi& of California, San Francisco, California 94 143 (Recelved: April 7, 1980; In Final Form: March 19, 1981)
We present ab initio SCF calculations employing the Morokuma component analysis on a number of Lewis acid-base interactions. Examples of hard and soft acids and bases have been studied in order to analyze the Pearson idea that hard acids interact preferentially with hard bases and that soft acids interact preferentially with soft bases. We conclude, on the basis of the noncovalent interactions studied here, that the empirical preference of soft acids for soft bases maybe largely a solvent effect rather than an intrinsic property of the direct interactions themselves. We also applied our calculations to a number of systems previously analyzed by the Drago equation, in which each molecule’s hardness and softness is taken into account in estimating experimental interaction enthalpies. Here the parallel between the predictions of the Drago equation and the calculations is reasonable for closely related complexes, but not so satisfactory when rather different complexes are compared. Unfortunately, it is not certain at this point whether deficiencies in the theory (431-G ab initio SCF), deficiences in the Drago equation, or solvation effects are the reason for the poor agreement.
Introduction The chemical literature contains a vast amount of experimental data on the stabilities of complexes formed by the association of electron donors and acceptors. In the past 30 years there have been a number of efforts to systematize and to understand this body of data from an empirical and a theoretical viewpoint. In 1958, Ahrland, Chatt, and Daviesl noted that electron acceptors fall into two classes: those that form the strongest complex with electron donors from the first row of the periodic table and those that form the strongest complex with second row donors. This idea was extended by Pearson2to the principle of hard and soft Lewis acids and bases. This principle states that hard acids bind more strongly to hard bases and that soft acids bind more strongly to soft bases. The original classification was based primarily on the Ahrland, Chatt, and Davies empirical criterion, although subsidiary criteria were also introduced. It is found that soft bases are those in which the donor is highly polarizable, and has low electronegativity, while hard bases have the opposite properties. Soft acids have properties similar to those of soft bases, i.e., high polarizability, low positive charge, and low electronegativity. Hard acids are the opposite. Important and inherent in the hard-soft principle is the fact what when the strength of a series of Lewis bases is ranked, reversals in strength may occur in the order, depending t Eastern
Washington University, Cheney, WA 99004. 0022-3654/81/2085-27 17$01.25/0
upon the Lewis acid which is being used. drag^,^ recognizing this interdependence of donor and acceptor properties, in 1965 proposed a four-parameter equation to represent the enthalpy of interaction of donor-acceptor pairs in poorly solvating media: -AH = EAEB C A C B (1) Here A and B represent the electron acceptor and donor, respectively, while E (“electrostatic”) and C (“covalent”) are empirically determined parameters for each. The Drago relationship provides a useful means of predicting enthalpies for the formation of complexes for which experimental values are not available. It relates to the hard-soft principle in that, for there to be a strong donor-acceptor interaction, the two molecules must “match” in the sense that both must have a large E parameter or a large C. It goes beyond the hard-soft principle in that the E and C parameters are not exclusive so that both the E and C parameters for a given molecule might be either large or small. The theoretical basis of the Drago equation has been of some interest. Klopman4 has shown that eq 1is consistent
+
(1)S. Arhland, J. Chatt, and N. Davies, Quart. R., Chem. Soc., 12, 265 (1958). (2)R.G.Pearson, J. Am. Chem. SOC.,85,3533 (1963). See also R. G. Pearson, J. Chem. Educ., 45,481, 643 (1968). (3)R.S.Drago and B. B. Wayland, J.Am. Chem. SOC.,87,3571 (1965). For a more recent review and a table of parameters for 33 acids and 48 bases, see R. S. Drago, Struct. Bonding, 15,73 (1973).
0 198 1 American Chemical Society
