R A M A N SPECTROSCOPY OF BIOPOLYMERS
lyzed there are no other proteins in the reference set with a very high helix content. W h e n this entire set of reference proteins is used to analyze secondary structure for another protein, the actual errors are likely to be smaller than those s h o w n in Table XI, and a s / 3 - s h e e t and helical proteins are added to this set of data the statistics are more likely to reflect this lower error. It is n o t e w o r t h y that the correlation coefficient for turn is relatively high in Table XI. This m a y be due to the inclusion of data a b o v e 1310 cm-l.8 Acknowledgments I am pleased to have this opportunity to thank Dr. Warner Peticolas without whose help and ideas this work would not have been started. It is also a pleasure for me to thank Dr. Keith Dunker for his ideas and support in the beginning of this work. I thank also Richard Honzatko, Janet Smith, Wayne Hendrickson, Marian Szebenyi, Alex Wlodawer, Brian Matthews, Martha Teeter, Terry Cutrera, Richard Priest, and Joel Schnur. This work was supported by a National Research Council-Naval Research Laboratory Research Associateship, NSF Grant PCM-8302893, ONR Grant WR30342, USUHS Grant R07139, and USUHS Grant C07147 awarded to Robert Williams.
 U l t r a v i o l e t
Introduction R a m a n s p e c t r o s c o p y of biopolymers is conveniently divided into two types, resonant and nonresonant, on the basis of proximity of the excitation radiation f r e q u e n c y to electronic transitions of c o m p o n e n t s of the sample. N o n r e s o n a n t , or classical, R a m a n s p e c t r o s c o p y has p r o v e n very useful in the determination of protein, nucleic acid, and m e m b r a n e conformational states. R e s o n a n c e R a m a n s p e c t r o s c o p y has provided considerable information on h e m e proteins and other metalloproteins and visual pigments. T h e s e substances have c h r o m o p h o r e s absorbing in the visible region readily accessible to conventional lasers c o m m o n l y used for Raman s p e c t r o s c o p y . T h e y are also nonfluorescent so that the relatively w e a k R a m a n signals can be o b s e r v e d without contamination from this broad fluorescence emission. R e s o n a n c e of the excitation frequency with an electronic transition METHODS IN ENZYMOLOGY, VOL. 130
Copyright © 1986by Academic Press, Inc. All rights of reproduction in any form reserved.
MACROMOLECULAR CONFORMATION: SPECTROSCOPY
results in several important changes in the Raman spectrum of a biopolymer. Under resonance conditions, the intensity is often greatly enhanced permitting studies of relatively dilute solutions. The enhancement of Raman scattering is restricted to vibrational motions of the component with an electronic transition and, thus, the resonance Raman spectrum is often greatly simplified relative to that observed off-resonance. Resonance Raman spectra are often dominated by the properties of the chromophore in the particular excited electronic state resulting in the resonance. This means that the resonance Raman spectrum may be quite different from that observed off-resonance. The observation of combinations and overtones, for example, is common under resonance conditions but not offresonance. Furthermore, the dominance of the resonance Raman spectrum by the proximal electronic state means that the spectrum will depend on which electronic state is nearest resonance. Thus, one expects distinct spectra depending on excitation conditions. This will be illustrated below. These features of resonance Raman spectra make it a very useful technique for the study of electronic excited states as well as for the establishment of conformational features of chromophoric biopolymer components. Ultraviolet resonance Raman scattering in the 250 to 300 nm region has been used to investigate the electronic excitations of the nucleic acid bases.l-7 Use of radiation in this region to investigate proteins results in intense fluorescence from the aromatic residues. The level of light scattered by the Raman process is generally very weak by the standards used in fluorescence spectroscopy. This means that the broad, featureless fluorescence emission can completely obscure the vibrational features of a Raman spectrum. Subtraction of the background due to the fluorescence is generally not sufficient to reveal the Raman peaks because the noise contributed by statistical contributions to the fluorescence cause the background-subtracted spectra to have more noise than signal by orders of magnitude. We have recently applied methods for generation of ultraviolet radiaY. Nishimura, A. Y. Hirakawa, and M. Tsuboi, Adv. Infrared Raman Spectrosc. 5, 217 (1979). 2 D. C. Blazej and W. L. Peticolas, Proc. Natl. Acad. Sci. U.S.A. 74, 2639 (1977). 3 W. L. Peticolas and D. C. Blazej, Chem. Phys. Lett. 63, 604 (1979). 4 y . Nishimura, H. Haruyama, K. Nomura, A. Hirakawa, and M. Tsuboi, Bull Chem. Soc. Jpn. 52, 1340 (1979). 5 y . Nishimura, A. Y. Hirakawa, and M. Tsuboi, Chem. Lett. 907 (1977). 6 M. Tsuboi, S. Takahashi, and I. Harada, in "Physico-chemical Properties of Nucleic Acids" (J. Duchesne, ed.), Vol. 2, p. 91. Academic Press, New York, 1973. 7 L. Chinsky, A. Laigle, W. L. Peticolas, and P.-Y. Turpin, J. Chem. Phys. 76, 1 (1982).
RAMAN SPECTROSCOPY OF BIOPOLYMERS
tion in the 200-250 nm region to Raman spectroscopy. 8-21 This has been used in studies of small molecules8-15 as well as biopolymers.16-19 This work has been reviewed. 2°.2~The use of radiation in this spectral region has several significant advantages. The first is that it provides resonance enhancement, and thus sensitivity and selectivity, for a new set of "chromophores" that are common in biopolymers. The problem of fluorescence from protein components is eliminated because this emission is at longer wavelengths than those used to collect the Raman spectrum. Thus, a 4000 cm -1 interval to longer wavelengths of, say, 215 nm ends at 235 nm, well short of the onset of fluorescence. Another advantage of this procedure is that it often results in new Raman peaks and therefore increases the total information available. This is due, in some cases, to resonance with electronic states higher than that responsible for the lowest transition. In others, it is due to the fact that the off-resonance spectrum is dominated by contributions from many excited states at very high energies with properties different from those of states resonant in this UV excitation region. This article describes the technology needed to perform these UV resonance Raman experiments. Emphasis is placed on the laser methods used to generate the excitation radiation. This is followed by some discussion of the primitive state of our present sample handling methods. The signal collection methods for the rest of the Raman experiment are, for the most part, conventional and will only be outlined. This method will then be illustrated with several examples including ultraviolet resonance Raman spectra of the nucleic acid bases 16A7and several protein compos L. D. Ziegler and B. Hudson, J. Chem. Phys. 74, 982 (1981). 9 L. D. Ziegler and B. Hudson, J. Chem. Phys. 79, 1134 (1983). 10 L. D. Ziegler and B. Hudson, J. Chem. Phys. "/9, 1197 (1983). 11 L. D. Ziegler and B. Hudson, J. Phys. Chem. 88, 1110 (1984). ~2L. D. Ziegler, P. B. Kelly, and B. Hudson, J. Chem. Phys. 81, 6399 (1984). 13 R. A. Desiderio, D. P. Gerrity, and B. Hudson, Chem. Phys. Lett. 115, 29 (1985). 14 R. R. Chadwick, D. P. Gerrity, and B. Hudson, Chem. Phys. Lett. 115, 24 (1985). 15 D. P. Gerrity, L. D. Ziegler, P. B. Kelly, R. A. Desiderio, and B. Hudson, J. Chem. Phys. 83, 3209 (1985). ~6 L. D. Ziegler, B. Hudson, D. P. Strommen, and W. L. Peticolas, Biopolymers 23, 2067 (1984). 17 W. L. Kubasek, B. Hudson, and W. L. Peticolas, Proc. Natl. Acad. Sci. U.S.A. 82, 2369 (1985). 18 L. C. Mayne, L. D. Ziegler, and B. Hudson, J. Phys. Chem. 89, 3395 (1985). 19 L. C. Mayne and B. Hudson, J. Am. Chem. Soc., submitted. 20 B. Hudson, P. B. Kelly. L. D. Ziegler, R. A. Desiderio, D. P. Gerrity, W. Hess, and R. Bates, in "Advances in Laser Spectroscopy" (B. A. Garetz and J. R. Lombardi, eds.), Vol. 3, p. 1. Wiley, New York, 1986. 21 B. Hudson, Spectroscopy 1, 22 (1986).
M A C R O M O L E C U L A R C O N F O R M A T I O N : SPECTROSCOPY
,06, 532 266
FIG. 1. A schematic illustration from above of the laser apparatus used to generate ultraviolet radiation for resonance Raman studies. The Nd:Yag laser is a Quanta-Ray DCR-2 30 Hertz model. The harmonic generating crystals are contained in a temperature-controlled metal housing (HG-1 with added heating tape and temperature controller; the current Quanta-Ray commercial model is an HG-2). The harmonics are separated with a 60 degree prism. The selected beam is sent to the lens of the hydrogen Raman shifting cell. The other beams, especially the 1064 nm IR beam, are sent to beam stops. The length of our Raman shifting cell is 0.75 m. The input and output lenses have focal lengths of 50 cm. The emerging Raman shifted beams are dispersed with a prism and one beam is selected for Raman excitation. All of the optical components are supracil quartz.
nents, including a model for the peptide bond, N-methylacetamide, TMhistidine, and some proline-containing peptides. 19The qualitative interpretation o f these spectra is well developed since most of the bands have been previously identified as to the types of vibrational motion involved. The quantitative understanding of the intensities of these bands is still in its infancy because of the complexity of the excited states of these species. H o w e v e r , the principles are well established and will be illustrated. Experimental Methods
Laser Technology A schematic illustration of the laser system used for these ultraviolet resonance R a m a n experiments is presented in Fig. 1. The basic c o m p o nent o f this system is a commercial N d : Y A G laser (Quanta-Ray DCR-2). This device produces pulsed radiation at 1064 nm with an average p o w e r of roughly 6 - 7 W. The pulse duration is - 8 nsec and the repetition rate is typically 30 Hz. This c o r r e s p o n d s to an energy per pulse of about 200 mJ and a peak p o w e r at the pulse maximum of 20-30 MW. When focused, this laser is easily able to generate electric fields that are sufficient to cause dielectric b r e a k d o w n o f air. These high peak powers result in very efficient c o n v e r s i o n to shorter wavelengths using nonlinear optical crystals. In our applications we use K D P (potassium dihydrogen phosphate) or K D * P (potassium dideutero phosphate) crystals. A commercially avail-
RAMAN SPECTROSCOPY OF BIOPOLYMERS
able combination of these crystals is used to generate harmonics of the fundamental 1064 nm infrared radiation at 532, 355, and 266 nm. The average power levels at these wavelengths are typically 3.0 W, 1.2 W, and 600 mW. This corresponds to 20 mJ/pulse at 266 nm. Similar Nd:YAG lasers with an amplifier stage can produce 266 nm pulse energies of 100 mJ at I0 Hz. The higher harmonic at 213 nm (1064 nm/5) can be generated by summing the 266 nm radiation with the fundamental. This is achieved with 90° phase matching using a special temperature controlled KDP crystal held at --400. 22 The power generated at 213 nm is limited by temporary damage to the KDP crystal if the full power of the YAG laser is used. This damage is apparently due to the generation of absorbing f-centers from iron impurities in these crystals. Illumination at high intensity results in a brown discoloration of the crystal which attenuates the 213 and 266 nm radiation. Warming the crystal to room temperature removes this discoloration. The maximum output of 213 nm radiation is roughly 10 mW average power. Recent advances in nonlinear optical materials promise to eliminate this limitation on the intensity of 213 nm radiation. The radiation produced at 532, 355 and 266 nm is sufficient to generate stimulated Raman scattering in molecular hydrogen gas, its isotopic variants and other gases. Higher pulse energies at 213 nm obtainable with new nonlinear crystals should be sufficient to permit similar shifting of this radiation. The result of the stimulated Raman shifting process is a series of collimated, powerful beams of radiation with frequencies shifted from the incident radiation by the vibrational frequency of the gas. For molecular hydrogen this frequency is 4155 cm -j. The new radiation is displaced in both the Stokes (lower frequency) and anti-Stokes (higher frequency) directions relative to the incident radiation. This stimulated Raman shifting process is the key to the generation of a wide variety of frequencies with sufficient power for resonance Raman scattering well into the vacuum ultraviolet region. The stimulated Raman process can be understood by analogy with the stimulated emission process occurring in laser action. In the Raman case, the incident radiation results in normal spontaneous Raman scattering producing, primarily, Stokes-shifted output with no preferential propagation direction. For polyatomic molecules, there will be several shifted frequencies. Some of this secondary radiation travels in the direction of propagation of the incident beam. This radiation is able to enhance the spontaneous Raman process because it provides radiation at the correct frequency to stimulate the downward step. For a polyatomic molecule, one or perhaps two of all of the modes are more efficient in this 2z G. H. L e s c h , J. C. J o h n s o n , and G. A. M a s s e y , IEEEJ. Quantum Electron. 12, 83 (1976).
stimulated process and begin to dominate the intensity. The resulting stimulation process results in amplification of the Raman-shifted radiation traveling in-phase with the excitation beam. For molecular hydrogen and 266 nm excitation radiation, this will result in a new beam at 299 nm, 4155 cm-~ lower in energy. The two beams can now interact with each other via four-wave mixing processes. This nonlinear interaction results in the production of radiation at the frequency to3
Either of the two propagating frequencies, 266 or 299 nm, may act as to~ with the other being tOE. The result is the generation of two new frequencies with wavelengths of 240 and 341 nm. These beams will also be colinear with the pump beam in the limit of a nondispersive Raman medium (with constant refractive index). Repetition of this process to produce further Stokes and anti-Stokes shifted beams is surprisingly efficient. Molecular hydrogen is a particularly good Raman shifting medium because of its large Raman cross section and its low refractive index dispersion. The important point about this process is that it has a threshold energy level. Below a certain energy per pulse, no shifted radiation is produced. Near threshold, the output is very unstable so that there are large fluctuations in the intensity of the shifted beams. At very high power levels the process tends to saturate because of thermal refractive effects in the medium. A list of the wavelengths that can be generated using hydrogen as the shifting medium and the 532, 355, and 266 nm Nd:YAG harmonics is given in the table. Raman spectra have been obtained with wavelengths as short as 184 nm. The main point is that this one, simple method is able to generate a wide variety of wavelengths. The density of these wavelengths is probably sufficient for coverage of most biopolymer Raman excitations, i.e., sharp resonances falling between these values are not expected. Other media can be used to generate alternate frequencies but molecular hydrogen has the lowest threshold and the largest Raman shift. The Raman shifting cell used for this process is commercially available (Quanta-Ray) or can be readily fabricated. The cell should be 1/2 to 1 m in length and fitted with thick (1/4 in.) Suprasil quartz or calcium fluoride windows. The window holder flanges should be lined with a gasket on the outside of the window to permit compression in response to the internal pressure. A valve and a gauge complete the cell. The optimum pressure of hydrogen gas depends on the particular Raman-shifted line of interest. Lower pressures (5-10 atm) are used for anti-Stokes scattering to minimize beam walk-off due to dispersion. A prism is used to select the particular Nd:YAG harmonic (532,355, or 266 nm) that is to be used for Raman shifting. This beam is focused into the Raman shifter with a long
RAMAN SPECTROSCOPY OF BIOPOLYMERS
STIMULATED RAMAN LINES FROM H2
Number of quanta of H2 vibration
266.04 252.67 245.97 239.55 228.66 223.13 217.86 212.83 208.81 204.19 199.76 192.13 188.21 184.45
4th 2nd 3rd 4th 2nd 3rd 4th 5th 2nd 3rd 4th 2nd 3rd 4th
0 5th 3rd 1st 6th 4th 2nd 0 7th 5th 3rd 8th 6th 4th
focal length lens. The emerging beams are collimated with a matching long focal length lens and then dispersed with another prism. A particular beam is selected and directed to the sample. One point worth mentioning is the inadvertant generation of stimulated rotational Raman scattering with this apparatus. The radiation produced by the KDP harmonic generating crystals is linearly polarized. Linear polarized radiation will only generate vibrationally shifted stimulated Raman scattering. Circularly polarized light also excites rotational and vibrational-rotational stimulated Raman radiation with frequencies quite close to the vibrationally shifted band selected for the resonance Raman excitation. The presence of crystallinity or strain-induced birefringence in the entrance window of the stimulated Raman shifting cell can induce ellipticity in the excitation beam and thus generate these rotational components. An alternative procedure for generation of ultraviolet radiation is the use of a doubled dye laser. This has the advantage that continuously tunable radiation is available. It has the disadvantage that large changes in the excitation frequency require changing the dye. There is also a problem of dye degradation with prolonged use. This is particularly a problem for the blue coumarin dyes which lase in the 440 to 500 nm region. The shortest wavelength that can be generated by direct doubling of dye laser output is 217 nm. A modification of this procedure uses the summation of the 1064 nm radiation with the doubled output of efficient red dye laser
MACROMOLECULAR CONFORMATION: SPECTROSCOPY
from circuloting pump
to circuloting pump
FIG. 2. A schematic illustration from the side of the sample geometry showing the excitation and collection arrangement. The liquid sample stream is directed downward into a collection tube. The excitation beam is directed upward and then back by prisms and the scattered radiation is collected by large quartz lenses. The use of a visibly fluorescent sample facilitates alignment of the collection optics.
radiation. An apparatus for ultraviolet resonance Raman scattering based on this approach has been described. 23 The most efficient method for generation of tunable far ultraviolet radiation is to combine the two processes described above. The idea is to generate sufficient power in the doubled dye laser radiation that the pulse energy is above the stimulated Raman threshold. Doubled dye laser or dye laser plus 1064 nm radiation at, say, 220 nm is shifted to 202 nm by one anti-Stokes conversion in molecular hydrogen. Devices of this sort based on excimer lasers are commercially available. The use of Nd:YAG lasers for this procedure requires an amplifier stage following the oscillator.
Sample and Scattering Geometry The radiation produced by the methods described above is directed to the sample using quartz prisms. (Most inexpensive mirrors will not withstand continued illumination with this UV radiation.) A final lens is used to focus the radiation on the sample. Backscattering geometry is used for most of our experiments. This configuration is obtained by using a very small Suprasil prism mounted on a post in the center of a large quartz lens which serves as the primary collection element (see Fig. 2). This geometry is used because backscattering is relatively insensitive to the value of the optical density of the sample at the particular excitation wavelength. Our earlier experiments used a Spex 1401 half-meter double monochromator. More recently we have used a McPherson 1-m vacuum UV 23 S. A. Asher, C. R. Johnson, and J. Murtaugh, Rev. Sci. Inst. 54, 1657 (1983).
RAMAN SPECTROSCOPY OF BIOPOLYMERS
single monochromator. This device has a single curved grating as its only optical component, resulting in very high throughput in the ultraviolet region. A fairly large monochromator is needed for work in the far ultraviolet because of the increased number of wavenumbers/nm in this region. The vibrational Raman bands of biopolymers are usually somewhat over 5 cm-~ wide. This is a reasonable number to use for adequate resolution if no significant features are to be missed. At 500 nm this bandwidth of 5 cm -~ corresponds to a monochromator bandpass of 0.13 nm. At 200 nm the corresponding bandpass is 0.02 nm. Our experiments so far have dealt with homogeneous, nonscattering solutions. For these samples, stray light rejection with a single monochromator appears to be adequate for Raman shifts greater than about 200 cm -~. Scattering samples require double or triple monochromators. Another important consideration in UV Raman scattering is order sorting. If the excitation wavelength is, say, 210 nm then the 4000 cm -~ interval spanning the fundamental region of the Raman spectrum extends to 230 nm. If this region is covered in second order to increase resolution, the monochromator will also pass 420 to 460 nm light in first order. This is in the tail of the strong fluorescence of many proteins. This fluorescence will overwhelm the Raman signal. There are no inexpensive high efficiency filters for the far UV region. Our solution to this problem is to use a "solar blind" photomultiplier which only responds to light in the 200 to 300 nm region. Even with this detector, however, second order operation is not possible when there is intense fluorescence. The use of this detector in first order reduces stray fluorescence and room light signals. Optical multichannel detection offers considerable benefits with respect to rate of data acquisition and rejection of laser pulse to pulse fluctuations as a source of noise. However, presently available detectors suffer somewhat in terms of resolution. Because of "bleed-through" between the detector elements, the narrowest lines are about three channels wide. This corresponds to a slit width of 75/xm. With a (high efficiency) 600 groove/mm grating, a one-meter monochromator has a dispersion in first order of about 1.6 nm/mm slit width. This means that the spectral bandwidth is about 0.12 nm for these multichannel devices. At 200 nm this corresponds to about 33 cm -~. Use of a 1200 groove/mm grating reduces this to 16 cm -~. This is probably adequate for most applications but often spectra at higher resolution will be needed in order to see if additional features can be resolved. The pulsed output of the photomultiplier is processed with a box car amplifier (Stanford Research Systems) triggered by a signal from the laser. In most cases, 30 or 100 pulses are averaged corresponding to a time constant of 1 or 3 sec. The analog output of the box car amplifier is sent
MACROMOLECULAR CONFORMATION: SPECTROSCOPY
both to a strip chart recorder and to the ADC input of a minicomputer. A reference channel is not used in most cases. The samples used for studies of biopolymers are circulating streams of aqueous solutions. A small dye laser pump circulates a 5 to 50 ml sample producing a downward directed column of liquid. The ultraviolet radiation strikes the front surface of the stream. This arrangement is used because it avoids all windows. We have found that Suprasil quartz and all other materials tested become fluorescent under prolonged illumination by focused UV radiation. Also, there seems to be a significant increase in sample damage at the window-solution interface. The other advantage of this sampling method is that it continuously exchanges the sample, minimizing the effects of degradation. The same arrangement can be used in a "once through" geometry if sample damage is a significant problem. The disadvantages of this sample system are the relatively large volume, the exposure of the sample to the atmosphere, and the relative difficulty of accurate temperature control. It is expected that these restrictions can be overcome by the use of a shielding layer of temperature controlled, humidified inert gas. In recent work, we have developed a fiat-surface guided flow device in which a stream - 2 mm thick with exposed surfaces 5.7 mm wide is created by flow between two parallel glass plates. The fiat surface of the stream permits reduction of stray light by directing the reflected incident radiation in a direction off the optic axis. The concentration used in most of our work on small peptides has been 1-10 mM. Proteins have been examined at 100/~M and it is possible to use concentrations as low as 10 ttM with spectral averaging and probably 1/zM using multichannel detection. Water is a very poor Raman scatterer and has excellent UV transmission and therefore makes a good solvent for these studies. Phosphate, sulfate, or cacodylate can be used as buffers and internal frequency standards. The scattering from sodium sulfate in aqueous solution has been shown to provide a good calibrated intensity standard. 24 Additional frequency standards are provided by the excitation frequency (zero wavenumber shift) and by a small amount of the adjacent stimulated Raman shifted line (at 4155 cm -J for hydrogen). Lines corresponding to atmospheric oxygen (at 1556 cm -1) and nitrogen at (2330 cm -j) are also often observed in spectra. Frequency calibration can also be provided by spectrometer wavelength standardization during each scan. Our method zl uses a chopped low pressure mercury lamp synchronized with the Nd:YAG laser but out of phase. A second box car amplifier collects this calibration scan. 24 j. M. Dudik, C. R. Johnson, and S. A. Asher, J. Chem. Phys. 82, 1732 (1985).
SPECTROSCOPY OF BIOPOLYMERS
200 209 218 229 240 253 266 "T
E u 299 500 e~ L b_
200 209 218 229 240 253 266
Shift (cm -1)
F I 6 . 3 . The resonance Raman spectra of GMP, CMP, AMP, and UMP obtained with the excitation wavelengths indicated at the left of each panel. The concentration is 1 m M in each case except for the 299 nm spectra where a concentration of 10 m M was used. The spectra have been scaled to a constant value for the 994 cm -j band o f phosphate present at 1 M concentration, p H 7.0.~7
It is unfortunate that both of the materials usually used for protein denaturation, guanidine and urea, are very strong UV Raman scatterers. This means that proteins will have to be denatured using other solvent systems such as extremes of temperature or pH or, perhaps, lithium or other salts. This has yet to be investigated. Illustrative Examples Nucleic Acid Bases
Raman spectra are presented in Figs. 3-5 for dilute (1 mM) aqueous solutions of the four ribonucleotides AMP, GMP, UMP, and CMP ob-
MACROMOLECULAR CONFORMATION" SPECTROSCOPY
k o.o~ - - - - e , 7
, ~, , ~
Excitatien Fmlmcy(m) FIG. 4. Raman excitation profiles for the four ribonucleotides.17
tained with laser excitation at 299, 266, 253,240, 229, 218, 209, and 200 nm. 16,17These spectra were obtained in 1 M potassium phosphate at pH 7 (Fig. 3) or at the indicated pHs (Fig. 5). In most cases, 3-5 scans each of about 40 min duration have been averaged. Distinct evidence of strong, selective resonance enhancement is obtained in that the spectra have very different appearance at different excitation wavelengths. These spectra are presented on the same relative intensity scale by using the phosphate band at 994 cm -~ as an internal reference. Excitation profiles have been constructed for the strongest bands (Fig. 4). The excitation spectra for many of the vibrational bands are dominated by a peak corresponding to the lowest energy electronic transition near 260 nm. Smaller peaks are
RAMAN SPECTROSCOPY OF BIOPOLYMERS
GMP 266 nm
11 i pH
, ~^/~A 266 A Mnm P ~ / /
FIG. 5. The resonance Raman spectra of AMP and GMP obtained with 266 and 218 nm radiation. The pHs of the solutions are indicated? 7
seen for higher energy electronic transitions. For some modes, the resonance enhancement is dominated by the higher energy transitions. A full description of the resonance Raman profiles of the nucleic acids will have to include several excited electronic states. 17,25Conversely, these excitation profiles can be used as a method for sorting out overlapping electronic transitions in these complicated chromophores. The scattering from the different bases is quite different in intensity with the strongest bands decreasing roughly in the order GMP > AMP > UMP > CMP. However, differential enhancement is possible by the proper choice of excitation wavelength. For example, between 220 and 230 nm the scattering of the 1529 cm -1 band of CMP is stronger than any of the bands of GMP. Similarly, the relative contribution of bands that overlap because of low resolution or intrinsic bandwidth, such as the 1585 cm -1 band of GMP and the 1583 cm ~band of AMP, can be sorted out by comparison of the spectra obtained at two different wavelengths. These spectra are most easily obtained using 266 nm radiation because this is near the peak of the excitation profile for the strongest modes and because of the relatively high average power of the radiation at this wave25 p. Callis, Ann. Rev. Phys. Chem. 34, 329 (1984).
In H20£~ol I
I I $00 [email protected]
FIG. 6. The resonance Raman spectra of N-methylacetamide in water (solid curve) and deuterium oxide (dashed curve) obtained with excitation at 218 rim. The concentration was 10 mM. length. On the other hand, two examples are given in Fig. 5 of cases where ionic species can be easily distinguished using far UV excitation (218 nm) but the spectra of these species are indistinguishable with 266 nm excitation. This demonstrates the utility of far UV resonance Raman spectroscopy for obtaining structural information. Simple Peptides Ultraviolet resonance Raman spectra of the simple peptide compound N-methylacetamide are shown in Figs. 6-8. TM Three isotopic species are shown: the normal proto species, the N-deutero form, and the N(15)deutero-C(13) form. The in-plane vibrations of the peptide band, amide I, II, III, etc. are modified on deuterium substitution. The modes of the deuterated form are designated amide I', II', etc. These spectra are of interest in several respects. The first is the essential absence of the C ~ O stretch (amide I) near 1680 cm -1. This is one of the stronger bands in offresonance spectra. It is believed that the apparent absence of this band in these spectra is due, at least in part, to a wide distribution of frequencies for this vibration due to a variety of hydrogen bonded structures. The major evidence for this is that this band is quite strong and sharp in
RAMAN SPECTROSCOPY OF BIOPOLYMERS NMAC~oLId) I
J t +6-
IdQ v e n u m b e r
FIG. 7. The resonance Raman spectra of C(12), N(14) N-methylacetamide (solid line) in deuterium oxide compared to that for the isotopically substituted species with carbon-13 and nitrogen-15 in the peptide bond (also in deuterium oxide).
acetonitrile solution. The amide I band is also seen in protein resonance Raman spectra. The second feature of interest is the high intensity of the amide II and II' bands (Fig. 6). This result was anticipated in preresonance studies of this compound using longer wavelengths where some enhancement of the amide II band was observed. 26,27 In the deuterated form, the amide II' band is by far the strongest component of the far UV spectrum. This band is very strong in infrared spectra but very weak in Raman spectra obtained with visible excitation. The enhancement with ultraviolet excitation can be understood in terms of the geometry changes associated with excitation of the peptide chromophore to its ~'-~r* state. The resonance Raman spectra of these low symmetry species are dominated by those modes that are along the geometry change associated with the excitation. In this case the analysis can be put on a quantitative basis because of the outcome of the comparison of the spectra for normal and isotopically 26 I. Harada, Y. Sugawa, H. Matsuura, and T. Shimanouchi, J. Raman Spectrosc. 4, 91 (1975). _,7y . Sugawara, I. Harada, H. Matsuura, and T. Shimanouchi+ Biopolymers 17, 1405 (1978).
MACROMOLECULAR CONFORMATION" SPECTROSCOPY NMACsolld) t .0-
h I gh
3000 3500 Wov®number
FIG. 8. A n e x p a n d e d view o f the r e s o n a n c e R a m a n s p e c t r u m of the same N - m e t h y l a c e tamide species o f Fig. 7 obtained with 200 n m excitation. T h e h a r m o n i c s of the amide II' vibration are indicated as are the solvent bands. The strong sharp line at 4155 c m ~ is the n e x t lower stimulated R a m a n line at 218 n m .
substituted species shown in Fig. 7. The shift upon replacement of the amide carbon and nitrogen by higher mass isotopes is roughly 47 cm ~. The shift expected if the motion of this mode were a pure C - - N stretch would be 55 cm -1. This observation, in combination with the intensity observations, indicates that the displacement of the amide group on excitation to its 7r-Tr* excited state is predominantly along the C - - N bond. This seems to be in disagreement with simple molecular orbital calculations. The assignment of the 7r-~r* of this peptide as the dominant state responsible for resonance enhancement has recently been confirmed by determination of the excitation profiles for the strongest bands, z8 The replacement of an amide deuteron by a hydrogen splits the intensity of the amide II' band between the amide II and III transitions (Fig. 6). Experiments performed in a 50-50 mixture of HzO and D20 show that the sum of the intensities of these two bands is equal to the intensity of the original amide II' band. 18This is easily explained by the hypothesis 28 j. M. Dudik, C. R. J o h n s o n , and S. A. A s h e r , J. Phys. Chem. 89, 3805 (1985).
RAMAN SPECTROSCOPY OF BIOPOLYMERS I I neo," I GLY-PRO(=oll I d )
I c GLY-PRO(,=IOSlI',ed).I
347 H20. I 218nm
? l Y
F[o. 9. ResonanceRaman spectra of N-GIy-Pro-COOH (solid (dashed line) in w a t e r obtained with 218 nm excitation.
line) and cyclic-Gly-Pro
that in the proto form the C - - N stretching motion is spread between these two modes because of mixing with the in-plane bending motion of the C - - N - - H group. In the deutero form this motion occurs at lower frequency and is not coupled to the C - - N stretch. The spectrum of N-methylacetamide observed with 200 nm excitation (Fig. 8) exhibits a characteristic feature of resonance Raman spectra in the appearance of overtone bands. This is due to the dominance of the Raman scattering process by vibronic levels well up in the excited electronic state having large transition amplitudes to highly excited levels of the ground state. The resonance Raman spectrum of the X-proline peptide bond obtained with ultraviolet radiation is similar to that of deuterated N-methylacetamide with one strong band near 1500 cm -l (Fig. 9). ~9 This is the expected result in that the absence of an amide proton eliminates coupling of the C - - N and C - - N - - H motions in the same fashion as deuterium substitution. In this linear dipeptide the strong band appears at 1485 cm -1. Restriction of the X-proline peptide bond into a cis configuration raises the dominant peak to a value of about 1515 cm -1. Although further studies are needed, it appears that this method may be capable of detecting the cis
MACROMOLECULAR CONFORMATION: SPECTROSCOPY PDP in I'1201ot 246, 218, I& 299n= 3.6"
i i 1489 1669 V~onu~
FIG. 10. Resonance Raman spectra of N-acetylprolinemethylamide obtained with excitation at 240, 218, and 200 nm (top to bottom). The band at 1485 cm -1 is due to the Nacetylproline linkage (see Fig. 9). The band at 1577 cm -I is due to the prolinemethylamide linkage. This band is preferentially enhanced at shorter wavelengths. A study of the individual species acetylproline and prolyglycine shows that the optimum relative excitation of X-Pro/Pro-Y occurs at about 230 nm.
isomer of the X-proline peptide bond. In order to be useful in studies of complex structures such as proteins, it will be necessary to enhance the signal from the X-proline bond relative to that from other peptide linkages. It has been found 19that excitation near 230 nm results in a maximum relative scattering of X-proline to other linkages of about a factor of 30. This is due to the relatively red shifted absorption of the X-proline linkage. An example of this behavior is shown in Fig. 10. Several of the amino acid side chains have been investigated using this new method. Histidine is one of the most interesting in terms of protein structural studies and enzymatic mechanism. The spectra of Fig. 11 show that deuterium substitution and the state of ionization of this group have a considerable influence on the form of these spectra.
I N .6 T E N S T T .4 Y
1200 1400 WAVENUMBER
Fro. 11. Resonance Raman spectra of histidine obtained with 213 nm excitation in (A) D20 at pH 11.8, (B) D20 at pH 4.4, and (C) H20 at pH 4.0.
Summary Ultraviolet resonance Raman scattering promises to be a useful technique for investigating the structure, refolding, and isotope exchange behavior o f proteins and nucleic acids. Protein-nucleic acid interactions may be particularly amenable to this new method. These preliminary results and many others not reported here have demonstrated that these spectra are quite strongly enhanced, are often distinct from those obtained with visible excitation, are very sensitive to isotopic substitution and conformation, and, in many cases, are sensitive to the detailed wavelength used for excitation. This last observation should prove useful in sorting out complex overlapping bands in proteins by providing a check
MACROMOLECULAR CONFORMATION" SPECTROSCOPY
on any proposed assignment in the form of confirmatory intensity changes. Ultraviolet resonance Raman spectroscopy is also clearly a useful technique for probing the geometries of excited electronic states of these species. It is also likely that this method can be useful in sorting out the complex pattern of electronic excited states of biopolymers. This is particularly needed if other well developed optical methods, such as circular dichroism, are to be put on a firm theoretical foundation. The limits of sensitivity of UV Raman scattering have yet to be determined. Considerable improvements in the quality of spectra have resulted over the past year due primarily to changes in the laser hardware, the use of a high throughput monochromator and signal collection and processing methods. The use of multichannel detection will probably permit studies of micromolar concentration protein solutions. Advances in sample handling methods are clearly needed and reliable internal intensity standards at shorter wavelengths would be useful. Clearly, however, laser technology is no longer the limiting factor in such studies. Methods are now available that permit extension of this technique to wavelengths as short as 150 nm. Acknowledgments This work was supported by NIH Grant GM32323, NSF Grant PCM8308529, and NIH predoctoral trainingGrant GM07759-05. We thank Drs. LawrenceZieglerand DanielGerpity who performed some of the early experiments on peptide components.
 R e s o n a n c e R a m a n S t u d i e s o f L i g a n d B i n d i n g By NAI-TENG YU Introduction A complete elucidation of the mechanism of function of a biological metal center requires knowledge of the exact nature of the metal-ligand bonds and their dependence on the protein environment. For understanding the cooperativity in hemoglobin (Hb) it is essential to know the effect of quaternary structure on the iron-ligand bond strength in both ligated and unligated states ~,2 in order to determine if the free energy of i M . F . P e r u t z , Annu. Rev. Biochem. 4 8 , 327 (1979). 2 D . L . R o u s s e a u a n d M . R . O n d r i a s , Annu. Rev. Biophys. Bioeng. 12, 357 (1983).
METHODS IN ENZYMOLOGY, VOL. 130
Copyright © 1986 by Academic Press, Inc. All rights of reproduction in any form reserved.