Photochemistry and Photobiology Vol. 56, No. 2, pp. 157-162, 1992 Printed in Great Britain. All rights reserved
0031-8655/92 $05.00+0.00 Copyright @ 1992 Pergamon Press Ltd
RESONANCE RAMAN AND SURFACE-ENHANCED RESONANCE RAMAN SPECTROSCOPY OF HYPERICIN LYDIAN. RASER,STEPHEN V. KOLACZKOWSKI and THERESE M. COWON* Department of Chemistry, Iowa State University, Ames, IA 50011, USA (Received 14 October 1991; accepted 27 January 1992)
Abstract-Hypericin has been found to exhibit a variety of photodynamic effects. To correlate biological activity with molecular structure, complete physical characterizationof hypericin is required. The vibrational spectrum has been determined and resonance Raman and surface enhanced resonance Raman scattering spectra are reported. In addition, the Raman spectrum of a model compound has been studied to facilitate assignment of the vibrational modes of hypericin.
and its mode of action be understood as fully as possible (Carpenter and Kraus, 1991;Hudson et al., Hypericin (structural formula, Fig. l), a multi-ring 1991;Lopez-Bazzocchi er al., 1991). quinoidal compound found in certain species of the The photodynamic action of hypericin is thought plant genus Hypericium, has been shown to exhibit to be the result of either the formation of radicals, a variety of photodynamic effects (Blum, 1941;Pace a Type I mechanism, or through the formation of and Mackinney, 1941; Knox and Dodge, 1985; singlet oxygen, a Type I1 mechanism. Both mechanDuran and Song, 1986;Knox et al., 1987). Hyperiisms have been observed in in virro studies (Heitz, cism, a condition of severe sensitivity to light, was observed in a variety of animals after ingestion of 1987;Knox et al., 1987). The singlet oxygen quanhypericin containing plants (Pace, 1942; Brock- tum yield has been determined and is quite large, mann, 1952;Thomson, 1971;Giese, 1971 and 1980). 0.74 (Jardon et al., 1986,1987). A value such as this In addition to inducing photosensitivity, it was dis- usually indicates an efficient intersystem crossing covered that hypericin is the chromophore of the mechanism, common for aromatic ketones, involvphotoreceptor (stentorin) that is responsible for the ing conformational changes (difficult in the expectnegative phototactic response of Stenfor (Song et edly rigid hypericin structure) or aggregation (Turro, 1978). Following the aromatic ketone argual., 1980;Walker et al., 1979;Yang et al., 1986). It ment, the high triplet yield suggests that significant is also known to have bactericidal properties and electron density is present on the carbonyls of was used at one time as an antidepressant (Duran hypericin in the excited singlet state. Before a comand Song, 1986). Recent findings that hypericin plete understanding of the mode of action is posshows a photodynamic effect against several retroviruses, including human immunodeficiency virus sible, the relevant physical, chemical and biological (H1V)t have made it essential that this compound parameters must be established. Absorption, fluorescence and phosphorescence spectra of hypericin have been previously reported (Pace and Mackinney, 1939; Scheibe and Schontag, 1942;Walker et OH 0 HO al., 1979;Jardon et al., 1986, 1987;Racinet et al., 1988;Jardon and Gautron, 1989). In this study, Raman spectroscopy has been employed to elucidate the vibrational spectrum of hypericin. Raman spectroscopy has several advantages over infra red (IR) for biological studies, the most important being the lack of interference from water. Also, visible wavelengths are used in Raman and, therefore, no special optics or sample containers are needed. The vibrational and other physical data for hypericin are essential for complete OH 0 HO characterization of hypericin in order to correlate Figure 1. Structure of hypericin. the biological activity with the molecular structure. Surprisingly little work has been done in this area 'To whom correspondence should be addressed. and, therefore, studies on model compounds are ?Abbreviations: DMSO, dimethyl sulfoxide; HIV, human immunodeficiency virus; Hyp, hypericin; PQ, 9,lO- indicated. In addition, absorption experiments, phenanthrenequinone;RR, resonance Raman; SERRS, designed to assess the amount of aggregation, were also performed. surface-enhanced (resonance) Raman scattering. INTRODUCTION
LYDIAN. RASERet al.
enhanced Raman scattering) spectra of PQ were obtained and compared with hypericin. MATERIALS AND METHODS
The hypericin used in this study was synthesized and purified in Dr. G. Kraus’ laboratory at Iowa State University. 9,lO-phenanthraquinone (Pa) was recrystallized from ethanol prior to use. Surface-enhanced Raman scattering experiments involving hypericin were carried out using a solution of hypericin (approximately 10+ M ) in DMSO, DMSO in chloroform or ethanol. The DMSO was dried over molecular sieves (3 A) to remove residual water. The chloroform was HPLC grade and was used without further OH / d \ H O purification; the ethanol was 200 proof (Barton Solvent). For SERS of PQ, PQ was dissolved in benzene (HPLC Grade, Omnisolve) prior to adsorption from the solution onto the SERS substrate. The absorption spectra of hypericin and PQ, in benzene, were obtained with a Perkin Elmer Lambda 6 spectrophotometer. Fluorescence spectra for hypericin were determined using an Aminco SLM 8000 spectrofluorometer. The excitation wavelength was 550 nm and the emission was monitored between 590 and 850 nm. The fluorescence spectra were detected perpendicular to the excitation source. The resolution of the excitation and emission monochromaters was 4 nm. Absorption and fluorescence spectra were obtained at room temperature. SERRS spectra for hypericin and PO were collected Figure 2. (a) Structure of 9,lO-phenanthrenequinone using an electrochemically roughened Ag electrode as the (Pa). (b) Structure of hypericin with a PQ ‘‘like’’ structure SERRS substrate. The electrode was dipped into a soloutlined. ution of the appropriate compound for approximately 15 min, removed from the solution and placed, without rinsing, in a liquid nitrogen optical dewar. The scattered Two variants of Raman spectroscopy have been radiation was collected in a backscattering geometry. utilized in this study. These are resonance Raman Hypericin SERRS spectra were taken with various wave406.7, 457.8, 488.0 and 514.5 nm. A Coherent (RR) and surface-enhanced (resonance) Raman lengths: Innova Kr+ 100 (406.7 nm) laser and an Ar+ 200 series scattering (SERRS) spectroscopy. The resonance (457.8, 488.0 and 514.5 nm) laser were used as the excieffect arises when the wavelength of light used for tation sources. The laser power at the sample was approxiRaman excitation is in resonance with an electronic mately 20 mW and the integration time was 20 s. The absorption band, resulting in enhancement of the SERS spectrum of PQ was obtained using 488.0 nm excitation and is compared to SERRS spectra of hypericin at vibrational modes that are coupled to the electronic this wavelength. transition. This enhancement can be as much as loh Resonance Raman spectra of hypericin and Raman times. Surface enhancement arises when a molecule spectra of PQ were obtained from solid samples packed is adsorbed onto a roughened metal surface (most in capillary tubes. Hypericin spectra were obtained at 457.8, 488.0 and 514.5 nm. Raman spectra of PQ were often Ag, A u or Cu). The roughened metal gives acquired using 488.0 and 514.5 nm excitation. The RR rise to increased Raman scattering through either and Raman spectra of hypericin and PQ, respectively, an electromagnetic o r chemical enhancement mech- were compared at 488.0 nm. All spectra of solid samples anism. Often, a decrease in fluorescence also occurs were taken at room temperature. Spectra at all excitation wavelengths except 406.7 nm and this leads to a further increase in the Raman were recorded using a Spex Triplemate spectrometer with signal to noise ratio (Cotton, 1988). a Princeton Applied Research Corp. (PARC) intensified In any application of SERRS, it is necessary to SiPD detector (model 1421-R-1024HD)cooled to -40°C. consider the nature of the molecule-surface interac- For detection at 406.7 nm, a Spex Triplemate spections. In these experiments, the Raman spectrum of trometer with a PARC model 1420 detector was used. detector was cooled with tap water. All spectra hypericin in solution was compared with the SERRS This reported here were plotted using Spectra Calc (Galactic spectrum to determine whether a strong chemical Software) and were baseline corrected and smoothed. interaction is present between hypericin and the Ag surface. If such were the case, the surface spectrum RESULTS should exhibit large shifts and intensity changes for some bands as compared to the R R spectrum. The absorption and fluorescence spectra [Figs. The Raman spectrum of 9,lO-phenanthrenequin- 3(a) and 3(b)] of hypericin obtained in D M S O are one, (Fig. 2), has also been determined in an effort identical with those previously reported for ethanol to assist in assignment of the vibrational modes of solutions with respect to peak position and shape hypericin. The excitation wavelengths used were (Scheibe and Schontag, 1942; Walker et al., 1979; not in resonance with any electronic absorption Jardon and Gautron, 1989). Pace and Mackinney transition of PQ. The Raman and SERS (surface (1939) described the effect of solvents on the
Raman spectroscopy of hypericin
500 700 WAVELENGTH (NM)
EMISSION WAVELENGTH (NM) Figure 3. (a) Absorption spectrum of hypericin in DMSO (approx. M). Scan rate = 600 nm/min, slit width = 1 nm, path length = 1 cm. Raman excitation wavelengths are marked by arrows. (b) Fluorescence spectrum of hypericin in DMSO. Excitation wavelength = 550 nm. Excitation and emission monochromator resolution = 4 nm.
absorption spectrum of hypericin in the visible region (450-600 nm). DMSO was not used in their study. Table 1 compares the positions of the major absorption and emission bands in DMSO to those in other solvents. The extinction coefficient in DMSO was calculated to be 43000. This is very close to the value of 41600 (in ethanol) reported by Scheibe and Schontag (1942). Serial dilutions (spectra not shown) were performed in an effort to assess whether aggregation occurs in ethanol. There was no change in the hypericin spectrum on diluting the sample from 4 x M to 4 x M. This
result suggests that hypericin is monomeric in polar solvents between and M. In non-polar solvents, chlorofodhexane (8:2) vol/vol, there are small differences in the peak positions compared to those in polar solvents. Our results are in agreement with those published by Pace and MacKinney (1939). The fluorescence spectrum was affected in a manner similar to that of the absorption spectra. The emission maxima are blue shifted by approximately 10 nm in ethanol as compared to those in DMSO. The excitation wavelengths used for RR and SERRS are indicated by the arrows on the absorption spectrum in Fig. 3(a). The extinction coefficients for these transitions are relatively low but are still sufficient to obtain large resonant enhancement. Excitation wavelengths closer to the more intense optical transitions could not be used. For example, at 568 nm, the fluorescent background completely overwhelmed the Raman signal and at 351.0 nm surface enhancement is negligible. Attempts to measure resonance Raman spectra in DMSO solution at several wavelengths were unsuccessful due to the large fluorescence background. However, RR spectra were obtained from the solid sample. The RR and SERRS spectra of hypericin are shown in Figs. 4 and 5. In Table 2 the peak positions in the high frequency region of the spectra are compared. Small frequency shifts occur with adsorption of the hypericin onto the silver substrate and there are also some differences in band intensities between the two sets of spectra. The largest difference between the RR and SERRS spectra is the greater intensity of the 1370 cm-* band at all excitation wavelengths in the RR spectra. This intensity difference is much less pronounced at 514.5 nm. The 1370 cm-I band has been tentatively assigned to an in-plane vibration of the central fragment, the portion of the molecule containing the carbonyl groups (rings I1 and VII). This assignment is based on comparison with PQ spectra (Gastilovich et al., 1986). Other differences in the RR vs SERRS spectra include greater intensity of the band at approximately 1315 cm-I at 488.0 and 514.5 nm excitation and the absence of the band at 1359 cm-' at 457.9 nm excitation in the RR spectrum. The
Table I Absorption maxima (nm)
Solvent DMSO Aq. DMSO Ethanol Chlorofodhexane (8:2) Pyridine'
599 597 591 601 603
555 554 547 563 558
514 516 510 525 520
480 481 477 489 483
'Absorption maxima taken from reference (Pace and Mackinney. 1939).
384 386 383 388
Emission maxima (nm) 343 342 337
333 332 326 332
LYDIAN. RASERel al. Table 2. Hypericin RR and SERRS at different wavelengths 406.7
SERRS 1643 1617 1581 1553 (1510)
1632 1604 1563 1500 1446 1419
1620 1590 1561 14% 1465 1436
1636 1605 1556 1498 1450
1622 1595 1563 1499 1459
1636 15% 1557 1499 1456
(1621) 1589 1558 11490) (1463) (1436) 1393 1374
1330 1314 1293 1248
1328 1313 1292 1245
(1130) (1012) (942) (922)
1392 1353 1315 1291 1251
1299 1184 1143 1025 919
1014 94 1 923
1012 941 920
(Parentheses indicate a weak signal).
SERRS spectra contain bands at 1359 (457.9nm excitation), 1343 (488nm excitation) and 1342 Figure 4. Resonance Raman spectra of solid hypericin. (a) (514.5nm excitation) that are not present in the 457.9 nm, (b) 488.0 nm, (c) 514.0 nm. Integration time, RR spectra. These bands are in the region of C-C 20 s, laser power, 20 mW; spectral resolution, 3 cm-I. in-plane stretching modes (Abasbegovic et al., 1964; Dollish et al., 1974). The similarity between the spectra indicate the absence of strong chemical " m interactions between the Ag surface and hypericin. 0 The carbonyl bands, normally present between 1660 and 1700 cm-', are not apparent in the SERRS or RR spectra of hypericin. The SERRS effect produces the largest enhancement of vibrational modes coupled to electronic transitions that are perpendicular to the surface. Absence of the carbonyl bands in the SERRS spectra suggests three possii bilities: (1) the hypericin adsorbs through one cari bony1 group but the excitation wavelengths are not I c in resonance with an electronic transition dipole 2 along the axis of the carbonyl group; (2) the hypericin adsorbs to the surface in a different orientation, k c e.g. the four hydroxy groups are closest to the surE face and the C=O bonds of the carbonyl groups E are parallel to the surface; and (3)the hypericin is C adsorbed with the ring plane parallel to the surface. Langmuire-Blodgett monolayer studies of hypericin (to be published) have indicated an area of approximately 55 A2 per hypericin molecule. This area is smaller than the calculated area of hypericin if the rings are parallel to the surface. Therefore, the hypericin must be oriented at some angle relative to the surface. SERRS spectra of hypericin RAMAN SHIFT (cm -1) adsorbed from a L-B monolayer are similar to those formed from self-adsorption. This implies that the Figure 5. SERRS spectra of hypericin. (a) 406.7 nm, (b) hypericin adsorbs in the same manner using these 457.9 nm, (c) 488.0 nm, (d) 514.0 nm. Experimental con- two methods and eliminates the third possibility. ditions are the same as in Fig. 3. Preliminary results suggest that the first hypoth-
RAMAN SHIFT (cm -1)
Raman spectroscopy of hypericin
Table 3. Comparison of hypericin and PQ spectra
Hypericin SERRS 488.0 nm
1677 1622 15%
1622 1595 1563
1636 1605 1556
1592 1561 1539 1502 I
1441 1371 1343 1295 1254
1370 1330 1314 1293 1248
1287 (1248) 1039 1016
RAMAN SHIFT (cm -1)
(Parentheses indicate a weak signal).
esis is the correct one, based on the observation of a strong band at 1643 cm-I with 406.7 nm excitation. The band is somewhat low in frequency for a C=O stretch but a decrease in frequency may result due to interaction of this group with the surface. To assist in the assignment of the Raman bands of hypericin, both Raman and SERS spectra of PQ were determined. The Raman spectrum of PQ had a very high fluorescence background and some Raman bands were difficult to distinguish. The SERS spectra of PQ showed higher signal to noise ratios compared to the Raman spectra obtained from the solid sample. The improvement in spectral quality is due to significant fluorescence quenching by the Ag surface. A comparison of the vibrational bands for hypericin and PQ is given in Table 3. The hypericin and PQ spectra are shown in Fig. 6. The spectra are quite similar in the spectral region above lo00 cm-’. The similarity is somewhat surprising because the structure of hypericin can be approximated by two fused anthraquinones. Based on the vibrational spectrum, however, it appears that the hypericin structure can be viewed as four interconnecting PQ-like structures, with each hypericin carbonyl fragment common to two PQlike structures [Fig. 2(b)]. Close proximity of oxygen atoms, whether as carbonyl or hydroxy groups, may also account for the similar features between PQ and hypericin. Because of uncertainty in the assignments for PQ, the hypericin bands cannot be assigned in an unambiguous manner. Bands in the region of 1650 to 1250 cm-’ are assigned as C-C stretching vibrations, most likely due to the outer benzene
Figure 6. SERS spectra of hypericin (top) and 9,10-phenanthrenequinone (bottom) at 488 nm. Experimental conditions as in Figure 3.
fragments of the four “corner” rings (see Fig. 1) (Gastilovich et al., 1986). The band at 1331 cm-’ by comparison to PQ is a C-C in-plane bending vibration of the outer benzene fragments and the 1370 cm-I band is due to in-plane vibrations of the central fragment, the portion of the molecule containing the carbonyl group (Gastilovich et al., 1986). Bands between 1160 and 1100 cm-I are from C-CH in-plane bending modes and bands in the region from 1030 to 680 cm-I are due to C-C inplane bending vibrations (Abasbegovic et al., 1964; Dollish et al., 1974; Lehmann et al., 1979; Dutta and Hutt, 1987). Hypericin has C2” symmetry and has 152 normal modes distributed between four symmetry species A,, A2, BI and B2. All modes are Raman and IR allowed except for the A2 modes which are IR forbidden. Infrared studies of hypericin are needed to assist in the assignments of the vibrational bands. CONCLUSIONS
The RR and SERRS spectra of hypericin show a great deal of similarity, indicating a weak interaction between the hypericin and the Ag substrate. High quality spectra were obtained using this technique. In addition, fluorescence was quenched by the SERS substrate allowing spectra of PQ to be obtained. Because of these two factors, SERRS should prove useful as a sensitive technique for the study of hypericin-containing systems, e.g. study of model membrane systems or native systems (viruses). Due to the complexity of the hypericin structure and, hence, the vibrational spectrum, exact assignments of vibrational bands are not possible without a normal mode analysis. Further experiments are necessary including the use of IR spectroscopy, additional model compounds and isotopic substitution. To facilitate the assignments of the vibrational spectrum, identification of the electronic
LYDIAN. RASERet al.
ttes du premier &at triplet de l'hypericine etude par spectroscopie laser. J. Chim. Phys. Phys. -Chim. Biol. 83, 311-315. Jardon, P., N. Lazorchak and R. Gautron (1987) FormaAcknowledgements-The authors appreciate the assistance tion d'oxygene singulet 'A, photosensibilisee par I'hypof Drs. Jae-Ho Kim and Jeanne Wynn with the prepericine. Characterisation et etude du mecanisme par aration of this manuscript. The authors are grateful for spectroscopie laser. J . Chim. Phys. Phys.-Chim. Biol. the financial support of the National Institutes of Health 84, 1141-1145. (GM 35108) (TMC). TMC is the recipient of a NIEHS Knox, J. P. and A. D. Dodge (1985) Isolation of the Career Development Award (ES10069). photodynamic pigment hypericin. Plant Cell Environ. 8, 19-25. REFERENCES Knox, J. P., R. I. Samuels and A. D. Dodge (1987) Photodynamic action of hypericin. In Light Activated Pesticides (Edited by J. R. Heitz and K. R. Downum), Abasbegovic N., N. Vukotic and L. Colombo (1964) pp. 265-270. American Chemical Society, Washington, Raman spectrum of anthracene. J. Chem. Phys. 41, DC. 2575-2577. Brockmann, H. H. (1952) Photodynamically active natural Lehmann, K. K., J. Smolarek, 0. S. Khalil and L. Goodman (1979) Vibrational spectra of anthraquinone. J . pigments. Prog. Org. Chem. 1, 64-82. Blum, H. F. (1941) Photodynamic Action and Diseases Phys. Chem. 83, 12W1205. Caused by Light. Reinhold, New York. Lopez-Bauocchi, I., J. B. Hudson and G. H. N. Towers Carpenter, S. and G. A. Kraus (1991) Photosensitization (1991) Antiviral activity of the photoactive plant pigment hypericin. Phorochem. Photobiol. 54, 95-98. is required for inactivation of Equine Infectious Anemia Neto, N., M. Scrocco and S. Califano (1966) A simplified Virus by hypericin. Phorochem. Phorobiol. 53, 169-174. Cotton, T. M. (1988) The application of surface-enhanced valence force field of aromatic hydrocarbons-I. Normal Raman scattering to biochemical systems. In Speccoordinate calculations for C6H6,,C6D6, C,,,H,, C,,D,, rroscopy of Surfaces (Edited by R. J. H. Clark and R. CL4HI0and C14D,,,.Specrrochim. Acra 22, 1981-1998. E. Hester), Vol. 16, pp. 91-153. John Wiley, New York. Pace, N. (1942) The etiology of hypericism, a photosensitivity produced by St Johns-wort. Am. J . Physiol. Dollish, F. R., W. G. Fateley and F. F. Bentley (1974) 136, 650-656. Characteristic Raman Frequencies of Organic ComPace, N. and G. Mackinney (1939) On the absorption pounds. John Wiley, New York. spectrum of hypericin. J. Am. Chem. SOC. 61, Duran, N. and P . 4 . Song (1986) Hypericin and its photodynamic action. Photochem. Phorobiology. 43,677-680. 3594-3595. Dutta, P. K. and J . A. Hutt (1987) Infrared and resonance Pace, N. and G. Mackinney (1941) Hypericin, the photodynamic pigment from St Johns-wort. 1. Am. Chem. Raman spectroscopic studies of 1-hydroxy-9,lO-anthraSOC. 63, 2570-2574. quinone and its metal complexes. J. Raman Specrros. Racinet, H., P. Jardon and R. Gautron (1988) Formation 18, 339-344. d'oxygegen singulet 'A, photosensibilisee par I'hyperiGastilovich, E. A., V. G. Klimenko, T. S. Kopteva, I. cine Ctude cinttique en milieu micellaire non ionique. A. Migel' and D. N. Shigorin (1986) Detailed interpretation of the vibrations of polyatomic molecules with J . Chim. Phys. Phys.-Chim. Biol. 85, 971-977. a low-symmetry nuclear configuration. Russ. J . Phys. Scheibe, G. and A. Schontag (1942) Lichtabsorption und Fluorescnz des Hypericins. Chem. Ber. 75, 2019-2026. Chem. (Engl. trans.) 60,69-73. Giese, A. C. (1971) Photosensitization by natural pig- Singh, S, N. and R. S. Singh (1968) Vibrational spectra of condensed ring quinones-I. 1,Cnaphthoquinone and ments. Photophysiology 6, 77-129. Giese, A. C. (1980) Hypericism. Phorochem. Photobiol. 9,lO-anthraquinone. Specrrochim. Acta MA,1591-1597. Song, P.-S., D.-P. Hader and K. L. Poff (1980) Phototactic Rev. 5, 229-255. orientation by the iliate, Stentor coeruleus. Photochem. Heitz, J. R. (1987) Development of photoactivated comPhotobiol. 32, 781-786. pounds as pesticides. In Light Activated Pesticides Stenman, F. (1969) Raman scattering from powdered 9, (Edited by J. R. Heitz and K.R.Downum) pp. 22-38. 10-anthraquinone. J. Chem. Phys. 51, 3413-3414. American Chemical Society, Washington, DC. Hudson, J. B., I. Lopez-Bauocchi and G. H. N. Towers Thomson, R. H. (1971) Naturally Occurring Quinones. Academic Press, New York. (1991) Antiviral activities of hypericin. Antiviral Res. Turro, N. J. (1978) Modern Molecular Photochemistry. 15, 101-112. Ito, T. (1978) Cellular and subcellular mechanisms of BenjamidCummings, Reading, MA. photodynamic action: the lo2hypothesis as a driving Walker, E. B., T. Y. Lee and P A . Song (1979) Spectroscopic characterization of the stentor photoreceptor. force in recent research. Phorochem. Photobiol. 28, Biochim. Biophys. Acta 587, 129-144. 493-508. Jardon, P. and R. Gautron (1989) Photophysical proper- Yang, K.-C., R. K. Prusti, E. B. Walker, P A . Song, M. Watanabe and M. Furuya (1986) Photodynamic action ties of hypericin in solution and in micellar dispersion. in Srenror coeruleus sensitized by endogenous pigment J. Chim. Phys. Phys.-Chim. Biol. 86, 2173-2190. stentorin. Phorochem. Phorobiol. 43, 305-310. Jardon, P., N. Lazorchak and R. Gautron (1986) Propritransitions giving rise to the absorption spectrum would also be useful.