DNA

Duncan Graham, Benjamin J. Mallinder, David Whitcombe, Nigel D. Watson, and ... Adrian M. T. Linacre, Calum H. Munro, Nigel D. Watson, and Peter C. Wh...
0 downloads 0 Views 788KB Size
J. Phys. Chem. 1995,99, 1608-1613

1608

Does Adsorption on the Surface of a Silver Colloid Perturb DrugDNA Interactions? Comparative SERS, FT-SERS, and Resonance Raman Study of Mitoxantrone and Its Derivatives Igor Nabiev,*J$*Alexandre BaranovJ95 Igor ChourpaJ Abdel BeljebbarJ Ganesh D. SockalingumJ and Michel Manfait*,+ Laboratoire de Spectroscopie BiomolLculaire, UFR de Pharmacie, UniversitL de Reims Champagne-Ardenne, 51096 Reims Cedex, France; Optical Spectroscopy Division, Shemyakin & Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences, I I7871 Moscow, Russia; and Vavilov State Optical Institute, I99034 St. Petersburg, Russia Received: July 8, 1994; In Final Form: September 28, 1994@

SERS spectra of the potent antitumour agent mitoxantrone in aqueous hydrosol and in the hydrosol prepared

from deuterium oxide and of its complexes with DNA have been recorded and compared with the corresponding preresonance Raman (pre-RR) and FT-SERS spectra of the same species. SERS and pre-RR spectra obtained M, respectively, were and 5 x at the same excitation wavelength and at concentrations of 5 x found to be nearly identical both in the frequencies and the relative intensities of the bands. Moreover, interactions between the drug and calf thymus DNA induced identical effects in the pre-resonance Raman, surface-enhanced Raman scattering (SERS), and Fourier transform SERS spectra of the drugs. An analysis of these spectral changes showed that an interaction involves preferential intercalation of the ring A and, in part, ring B of the chromophore inside the DNA double-stranded helix. The structural specificity of the mitoxantrone intercalation has been studied by SERS analysis of the complexes between the drug and DNA duplexes [d(CpG)9]2 and [d(ApT)&. Mitoxantrone was found to be intercalated preferentially within the CG-rich regions of the double-stranded helix. The data show that the adsorption of the drugDNA complex on the surface of silver hydrosol does not induce detectable perturbations of the molecular interactions within the complex and thus demonstrate the applicability of SERS for the analysis of drugDNA interactions under conditions preserving the structure of the complexes and at extremely low concentrations.

Introduction Surface-enhanced Raman scattering (SERS) spectroscopy has been used as a powerful method to obtain information for drug chromophores and their interactions with the targets (DNA, intracellular Models of the intercalation between some drugs and DNA have been proposed and were found to be consistent with NMR and X-ray crystallography data.1-3s5 Moreover, the extra-high sensitivity of the technique enabling the recording of highly resolved spectra from the pico- and even femtomolar quantities of some compounds (when manipulating with 10-8-10-10 M concentrations) and quenching of the chromophore fluorescence both allowed the chromophores to be followed within the living cell, hence, to develop a technique of micro-SERS s p e c t r o ~ c o p y . ~ . ~ . ~ The main question of SERS applicability for resolving problems in chemistry, biochemistry, and biomedicine concems the possibility of making measurements preserving the native conformation of the molecule or supramolecular complex. Some research teams’-3 have published data undoubtedly demonstrating that the interaction with the metal surface in hydrosols is quite smooth and does not disturb the native structure of the molecules or their interactions within the complexes. On the other hand, significant perturbations in the

’UniversitC de Reims Champagne-Ardenne.

*Shemyakin & Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences. + Vavilov State Optical Institute. *To whom correspondence should be addressed, at Laboratoire de Spectroscopie BiomolCculaire, UFR de Pharmacie, UniversitC de Reims Champagne-Ardenne, 51 rue Cognacq Jay, 51096 Reims Cedex, France. @Abstractpublished in Advance ACS Absrracrs, January 1, 1995.

0022-365419512099- 1608$09.oOlO

structure of heme-containing proteins have been found in some SERS-active systems.8 The difficulty of independent control of the influence of the silver (or in general metallic) surface on the biomolecular structure concems the lack of “surfaceenhanced” methods other than SERS spectroscopy: it is quite easy to lose the signal from the molecules located in the vicinity of the surface when the strong background signal from the molecules in the bulk is detected by traditional (normally non“surface-enhanced”) methods of control. Hence, it seems necessary to use SERS spectroscopy itself for the control of molecular structure in the vicinity to the metal surface by comparison to the corresponding Raman (or RR) and SERS spectra. But in reality this is not easy. SERS spectra usually differ markedly from the corresponding Raman spectra of free molecules due to the probable changes in the selection rules and the short-range character of the Raman cross section enhancement in SERS-active systems. This is revealed by the selective enhancement of certain vibrations, as well as the appearance of new bands in a SERS ~ p e c t r u m .Moreover, ~ the Raman (and even RR) spectra are obtained only at concentrations at least 2 orders of magnitude higher than that for the SERS spectrum of the same compound and usually have significant contribution from parasitic fluorescence. Finally, very few examples of close correlation between SERS and Raman spectra of the same biological compound are known from the l i t e r a t ~ r e . ~ Among .~ them, no successful attempts have been made to compare the SERS and Raman spectra for the evaluation of the influence of the silver surface on the molecular interactions within the drugDNA complexes. In the present paper, we report on a detailed comparative preRR, SERS, and FT-SERS study of mitoxantrone (Figure l), a 0 1995 American Chemical Society

DrugDNA Interactions

J. Phys. Chem., Vol. 99, No. 5, 1995 1609

Mitoxantrone

Solvent Blue 35

RYCHz)zHH

.-c

(r:

C

a:

+C S C

E C

LT

u i L IN

5 Yi

1600

1400

1200

1000

800

600

400

wavenumbers (cm-1) 1600

1400

1200

1000

800

600

400

wavenumbers (cm- 1 )

Figure 2. SERS (1, 2) and preresonance Raman (3-6) spectra of mitoxantrone (1-4) and solvent blue 35 (5,6) in aqueous (1,3,5) and in deuterium oxide (2.4, 6) solutions. Excitation: 514.5 nm. PreresoM; laser power 150 mW; nance Raman (3-6): concentration 5 x 100 accumulations (1 s each). SERS (1, 2): concentration of the drug ca. 5 x lo-* M; laser power 15 mW; 64 accumulations (1 s each).

tions.15J6 All these modes are allowed in the Raman spectra but only totally symmetric vibrations (AI) are active in RR excitations. Very few B2-type vibrations could be observed when preresonance excitation is used or when some factors induce distortion of symmetry of the molecule. Hence, enhancement of the B2-type vibrations can be seen in the FTSERS spectrum of the drug at the very far-off-resonance preresonance conditions of excitation at 1064 nm (Figure 3). 1,4,5,8-Tetrahydroxyanthraquinone(AQoH,Figure 1) can be considered as a simplified model chromophore for mitoxantrone. Calculation of the normal modes for AQOHand AQODhas been carried 0ut,I53l6and the assignment has been obtained from the analysis of the polarized light infrared spectra and 4.2 K emission fluorescence spectra in solid ~ o l u t i o n . 'Hence, ~ ~ ~ ~the vibrational modes of A Q ~ H are relatively well-known (Table 1). Considering that AQOHbelongs to D7h. symmetry group, the assignment of the in-plane vibrations in the RR and SERS spectra of mitoxantrone could be carried out from the following correlation diagram:

Comparative analysis of the calculated normal modes, polarized IR spectra of the AQOHand AQNHZ , I 7 calculated normal modes, and infrared and 4.2 K emission fluorescence spectra of the 1,4-(OH)2AQ; 1,4(NHz)zAQ, and I-NH2P-OHAQ c h r o m o p h ~ r e s ~ ~showed ~ ' ~ - ~that ~ the frequencies of the major vibrations do not exhibit significant changes upon 0 N substitutions. The substitution of two OH groups in the 1- and

-

Figure 3. Normalized FT-SERS spectra of mitoxantrone (l), its complex with calf-thymus DNA (2), and their difference spectrum (3 = 1-2). DrugDNA complex: ratio 1 molecule of drug for 200 base pairs of DNA. Concentration of the free drug 5 x lo-' M; wavelength of excitation (40 mW): 1064 nm; 300 accumulations. Positive peaks in the difference spectrum (3) correspond to the bands exhibiting a

decrease of intensity upon interaction of the drug with CT-DNA. Absolute concentration of the drug/DNA complex in the hydrosol (ca. M in the drug concentration) was adjusted to have a spectrum of quality comparable to that of spectrum 1. 4-positions of AQOHfor NHR groups (where R is a heavy radical, Figure 1) should lead to the appearance of the v(C-N) and S(CC-N) modes belonging to the A1 symmetry. The frequencies of these vibrations in the hydroxyamino derivatives of AQ are located in the same spectral regions as v(C-0) and S(CC-0) motions. l 7 s 2 I However, the quite heavy (e.g., methoxy) radical in mitoxatrone is anticipated to lead to a pronounced downward shift ca. 10-15 cm-I of the v(C-N) mode as it has been found in the infrared and fluorescence emission spectra of the 1,4(NHCH&AQ and 1,4(NH2)2AQ derivative^.'^ SB35 (Figure 1) seems to be a suitable model for the separation of the Raman bands which are most closely related to v(C-N) and &CC-N) motions, and those related to v(C0) and S(CC-0) vibrations. The SB35 chromophore belongs to the same (C2") symmetry group and has a heavy radical at the same position as does mitoxantrone. Therefore, RR spectra of SB35 in H20 and D20 at 514.5 nm excitation have been recorded (Figure 2) and used for vibrational assignments (Table 1). Absorption spectra of mitoxantrone at pH 7 and 10, and of its derivative SB35 (without hydroxyl substituents in ring A) are presented in Figure 1. An increase in pH disturbs the conjugation of the chelate system of mitoxantrone due to the weakness of the intramolecular hydrogen bonds. This leads to a remarkable degradation of the absorption band related to the lowest energy electronic transition. The weakness of the intramolecular hydrogen bonds results in a decrease in the bond order. So, an upward shift of v(C=O) mode and slight downshifts of v(C-0), v(C-N), and some of the other vibrations are expected in the Raman spectra of mitoxantrone (Table 1). Deuteration does not induce noticeable changes in the absorption spectra of mitoxaritrone and SB35 (data not shown) indicating that there is no significant distortion of the

J. Phys. Chem., Vol. 99, No. 5, 1995 1611

DrugDNA Interactions

TABLE 1: Resonance Raman and SERS Frequencies (cm-') of Mitoxantrone and Calculated Frequencies of 1,4,5,8-Tetrahydroxyanthraquinone (AQoH)and Resonance Raman Frequencies of SB35 for Vibrational Assignment@ SB35 RR,HzO, A Q ~ H ~ pH=7.2 1646 s 1609 w 1555 vw 1516 vw 1466 w 1412 m 1353 w 1311 w 1252 vs 1162m 1096 m

RR,HzO, pH=7.2

mitoxantrone RR,HzO, SERS,HzO, p H = 10 pH=7.2

I646 s 1600 vw

1646 m

1491 w I447 m 1405 vw 1361 w 1338 w 1300 vs 127&w 1256dw II80m 1106m

1491 vw 1447 vw 1420 vw 1308 vs 1256 sh 1196 w lll0w 838 vw

492 vw 433 w

470 vw 433 w

480 vw 435 vw

SERS, DzO

I646 s 1603 vw 1563 vw 1491 w 1447 m 1405 vw 1361w 1338w 1300 vs 1 2 7 8w 1256dw 1176m I103 m 984 w 813 w 553 vw 505 vw

1646 s 1603 w 1563 vw 1491 vw 1447 s 1400 vw 1361 sh 1338 vs ca. 1270 w 1270 m 1232 s 1152 vw 1107 w 1020 vw 812 w

467e w 435 w 339 vw

462 w 431 vw 339 vw

assignments mode, symmetry, [fragment of the chromophore: A-C (Figure 1)1 v(C=O), AI, P

I

}ring stretch, AI ring stretch coupled with v(C-O), Bz ring stretch coupled with v(C-0), AI ring stretch, AI v(C-C) coupled with chelate system, AI, [B] }ring stretch,coupled with v(CO), AI ~ ( c - 0 )AI, ~ [AI v(C-N), AI, [Cl }d(CC-H) coupled with v(C-0) and v(C-N), AI v(C-C), AI, [Bl }def ring, AI, [A and C] d(C=O), AI, [Bl def ring, AI, [A and C] S(CC-O), AI, [AI d(CC-N), AI, [Cl d(CCC), AI, P I

a Frequencies in cm-I; vs, very strong; s, strong; m, medium; w, weak; vw, very weak; sh, shoulder. The frequencies in italic correspond to the bands that decrease in intensity upon complexation of mitoxantrone with DNA. Calculated and experimental (normal coordinate analysis,16 IR," and 4.2 K fluorescence1*).Symmetry groups are indicated in brackets. Frequency values obtained as a result of the deconvolution of the overlapped bands in the 1400-1200 cm-' region. See text for details. dFrequency values obtained from depolarized RRS spectra. e Strong effect of DNA binding on the FT-SERS spectra of mitoxantrone (Figure 3).

lowest energy electronic transition related to the chelate system. Hence, the changes in the pre-RR and SERS spectra of the drugs on going from H20 to D20 solution are determined by the effect of heavy mass atom (deuterium) substitution (frequencies) and by the alteration of the symmetry of some vibrations (relative band intensities). As can be seen from Figure 2, deuteration causes similar modifications in the pre-RR and SERS spectra of mitoxantrone. Table 1 shows the frequencies and assignment of the vibrations of AQOH(experimental and calculated), SB35, and mitoxantrone in H20 and D2O. The v(C=O) frequency is situated at the 1675-1670 cm-l region for AQ and for its derivatives with non-hydrogen-bonded carbonyl groups (e.g., ~ , ~ - ( O H ) Z A Q The ) . ~ frequency ~,~~ of the v(C=O) mode was found to be ca. 1630 cm-l in the spectra of the hydroxy derivatives of AQ15,20922 and ca. 1650 cm-' for its amino derivatives 1 7 a and for SB35 (Table 1). The band at 1646 cm-' in the spectra of mitoxantrone is assigned to the Y (C=O) mode. The presence of this band, with the same frequency and relative intensity in the pre-RR spectra of SB35 and mitoxantrone (Figure 2), reflects the dominant influence of the intramolecular bond C=O-H-N-C between rings B and C of the chromophores on the frequency of this vibration. Deuteration does not affect either the frequency or relative intensity of v(C=O) vibration for all (ca. 20) anthraquinone derivatives studied before.16J7 The group of bands in the 1610-1400 cm-' region is assigned to the v(C-C) vibrations of the AI-type symmetry in the phenolic rings A and C of the chromophore. Thereafter, we shall call these vibrations "ring stretches", because it is difficult to localize them on certain bonds in view of their complicated form.17 These bands are not strongly dependent on the deuteration and pH (Table 1). An exception is the band at 1447 cm-', which is more intense in D20 than in H20 (due, probably, to an increase of the local symmetry, Figure 2) and exhibits a decrease in intensity at pH = 10 (Table 1): this demonstrates the coupling of the vibration with the chelate

system of the chromophore. The bands at 1491 and 1447 cm-' disappear or exhibit pronounced shifts in frequency on going to solvent blue (Figure 2). Hence, an assignment of these two bands to the ring stretch vibrations coupled to the v ( C - 0 ) motions seems to be reasonable. The region 1400-1200 cm-' seems to be the result of overlapping of several bands. The bands at 1361 and 1300 cm-l are well resolved, and the band at 1256 cm-' has been easily detected as an individual band in the depolarized pre-RR spectrum of mitoxantrone (Table 1). The deconvolution of the 1400-1200 cm-' spectral region of the pre-RR and SERS spectra into five Gaussians, three of which with the known positions of the maxima (1361, 1300, and 1256 cm-l) enabled us to determine the position of two other bands at ca. 1338 and 1278 cm-' (Table 1). The band at 1361 cm-I is assigned to v(C-C) motions localized on the B ring of the chromophore (Table 1). The intensity of this band decreases at pH = 10 showing some coupling with the chelate system of the chromophore but is not strongly influenced by deuteration. The frequency of this mode changes only slightly on going to SB35, supporting its skeletal character. The band at 1338 cm-' can be attributed to the A1 symmetry ring stretching involving Y ( C - 0 ) and v(C-N) motions.16J7 Deuteration leads to a strong increase in its intensity: it becomes a dominant band in the both pre-RR and SERS spectra (Figure 2). This could be explained in the frame of an increase of local symmetry of corresponding motions. Similar spectral features were observed in the RR spectra of deuterated a d r i a m y ~ i n . ~ s ~ - ~ ~ It seems as if the contribution of v ( C - 0 ) in this band is more significant than the Y(C-N) motion: no other intense bands have been found in the corresponding region of the pre-RR spectrum of deuterated SB35. The band at ca. 1300 cm-' is the most intense one in the pre-RR, SERS, and FT-SERS spectra of mitoxantrone. It could be assigned to the ring stretching mode coupled with the v(C0) motion^.^^^^^ This vibration is very sensitive to the existence

1612 J. Phys. Chem., Vol. 99,No. 5, 1995 of the C-0 group in ring A of the mitoxantrone chromophore: the band completely disappears on going from mitoxantrone to SB35 (Figure 2). The band at ca. 1300 cm-', is not sensitive to C N substitution in ring C: it remains unchanged on going from mitoxantrone to aclacinomycin or adriamycin (the chromophores of the last two drugs have C-0 instead of C-N substitutions in ring C).2,3,24Deuteration leads to a dramatic decrease in the intensity of this band, probably, accompanied by a downward frequency shift to ca. 1270 cm-I (Figure 2). The last two bands of the 1400-1200 cm-' region, namely, ca. 1278 and 1256 cm-', seems to be related to the v(C-0) and v(C-N) vibrations, respectively. The low frequency component at ca. 1256 cm-I should be assigned to v(C-N) vibration, since N atoms of ring C of the chromophore are coupled to heavy radicals and also taking into consideration the correlation diagram (see above). This is confirmed by the analysis of the SERS spectra of SB35 (Figure 2, Table 1). The band at ca. 1278 cm-I, attributed to v(C-0) mode, does not exist in the spectrum of SB35, whereas a very strong band at ca. 1252 cm-I due to the v(C-N) vibration is observable. Frequency shifts of the 1278 and 1256 cm-' bands upon deuteration (to ca. 1270 and 1232 cm-', respectively) are compatible with those expected for the stretching vibrations coupled with the motions of the chelate system of the c h r o m ~ p h o r e . ' ~ -An * ~ increase in the relative intensity of the v(C-N) upon deuteration (Figure 2) could be related to an increase of the local symmetry of this vibration: inclusion of deuterium provides partial compensation of the influence of the heavy radical (R) in ring C (Figure 1) of the chromophore. The band at ca. 1256 cm-' is more intense in the RR spectrum of mitoxantrone at pH 10 than at pH 7 (Table 1). This fact confirms an asymmetrical pH-dependent degradation of the intramolecular bonds within the chromophore of mitoxantrone mentioned above. The bands at ca. 1176 and 1103 cm-' have frequencies characteristic of d(CC-H) vibrations coupled with both C - 0 and C-N motion^.'^.'^,^^ Bands with similar frequencies were detected in the RR spectrum of SB35 (Figure 2). Variations of their frequencies upon mitoxantrone deuteration and changes in pH (Table 1) correlate well with those expected for the model chromophores. I 6 x 1 It seems that the d(CC-0) motion contributes largely to the band at ca. 467 cm-'. The other possible assignments of this band, e.g., the d(CC=O) vibrationi6 or ring deformation are less probable: the band disappears in the pre-RR spectrum of SB35 (Figure 2, Table 1). The d(C=O) vibration is located at ca. 505 cm-I in the SERS spectra of mitoxantrone and at 492 cm-' in the pre-RR spectrum of SB35. The d(CC-N) vibration is expected to be at a lower frequency compared to the S(CC-0) motion (due to the influence of heavy radical R in ring C of mitoxantrone thus, the 435 cm-' band in the spectra of both mitoxantrone and SB35 (Table 1) could be assigned to the d(CC-N) vibration. The increase in intensities of the bands at ca. 467 and 435 cm-' (S(CC-0) and d(CC-N) modes, respectively) is a prominent feature of the FT-SERS spectra of mitoxantrone at the far-off-resonance (1064 nm) preresonance excitation (Figure 3). This fact is compatible with the strong coupling of the longwave absorption band of mitoxantrone at 660 nm (Figure 1) with the chelate structure of the drug. A rise in the relative intensity of the bands, related to the motions involving oxygen atoms, on going to the preresonance excitation in the region of the long-wavelength electronic transition of the chromophore has been also observed in the pre-RR spectra of a d r i a m y ~ i n . ~ ~

Nabiev et al.

-

N

'ii, ' I

I

1600

1400

1200

1000

800

600

400

wovenumbers (cm-1)

Figure 4. Effects of calf thymus DNA binding on the preresonance Raman (1) and SERS (2) spectra and effects of [d(CpG)& (3) and [d(ApT),]z (4) binding on the SERS spectrum of mitoxantrone (difference spectra: spectrum of the free drug minus spectrum of the complex). Drug/CT-DNA complex for the pre-RR spectrum and drug/ oligonucleotides complexes for the SERS spectra: ratio 1 molecule of drug for 40 base pairs of DNA; for the SERS spectra of complexes with CT-DNA: 1 molecule of drug for 200 bp of DNA. Experimental conditions for recording of the original spectra were as in Figure 2. The procedure of normalization and subtraction was as in Figure 3.

SEW Spectra of Complexes of Mitoxantrone with DNA and with Double-Stranded Oligodeoxyribonucleotide. A comparison between the difference (free drug minus drugDNA complex) SERS, pre-RR (Figure 4), and FT-SERS (Figure 3) spectra of mitoxantrone and mitoxantroneDNA complexes shows the loss in intensity, upon complexation, of the bands at 1302 and 1275 cm-' assigned to the ring stretching modes coupled with C - 0 motions and to the v(C-0) vibration, respectively. On the other hand, the bands due to the C-N vibrations were found to be unchanged upon complexation. This clearly indicates that the interaction with the base pairs involves the intramolecularly hydrogen bonded rings A and B of the mitoxantrone chromophore. The fact that only particular vibrations are affected indicates that only a portion of the chromophore is involved in the interaction. This portion is formed by rings A and B of mitoxantrone. The vibrations localized on ring C of the chromophore were not changed upon intercalation of the drug. These effects do not provide new information about mitoxantroneDNA interactions but correlate well with the data published before. NMR and optical spectroscopic analyses showed that in the mitoxantroneDNA complex, the A and, probably, the B ring of the chromophore is partially intercalated within the DNA duplex with the terminal OH groups of the side chains bound to the central phosphate groups of DNA.I2st4 The most important fact concerning SERS application is that the effect of the complexation in the pre-RR spectra was found to be identical to that observed in the SERS (Figure 4) spectra (at the same wavelength of excitation). The bands due to C - 0 motions are also reduced in intensity (and to nearly the same extent as that of the SERS) upon complexation, while the modes coupled with the C-N group of ring C were found to be the same both in the pure drug and in the drugDNA complex. Similar spectral effects when binding with the DNA have also been noticed in the FT-SERS spectra of the drugDNA

J. Phys. Chem., Vol. 99, No. 5, 1995 1613

DrugDNA Interactions complexes. As was mentioned before, excitation at 1064 nm corresponds to the preresonance conditions of excitation from the long-wavelength side of the electronic transition of the mitoxantrone chromophore, whereas excitation at 5 14.5 nm (applied to record RR and SERS spectra) corresponds to preresonance conditions of excitation on the low-wavelength side of the electronic transition of the drug (Figure I). It is the long-wavelength electronic transition of mitoxantrone which was found to be most sensitive to the drug interactions: an absorption band at ca. 660 nm exhibits pronounced long-wavelength shifting upon binding to DNA (Figure 1). An analysis of the FT-SERS spectra of the drug and drugDNA complex provides additional c o n f i a t i o n of the model for mitoxantrone binding proposed from the pre-RR and SERS spectra with the visible excitation: intercalation induces a strong effect on the band at ca. 466 cm-' (related to the S(CC-0) vibrations in ring A), whereas the band due to the G(CC-N) motions coupled with ring C (ca. 440 cm-I) was found to be unchanged. These bands are very weak in the pre-RR and SERS spectra with the 514.5 nm excitation (Figure 2), being related to the long-wavelength electronic transition of mitoxantrone (Figure I), but are quite strong in the FT-SERS spectra excited at 1064 nm (Figure 3). Hence, IT-SERS spectroscopy has enabled us to obtain additional information conceming drugDNA interactions. The SERS spectra of mitoxantrone and of its complexes with CT-DNA, [d(CpG)&, and [d(ApT)& have been compared (Figure 4). It can be noticed that the drug interaction with [d(CpG),]z induces nearly the same effect in the SERS spectra as that found for the mitoxantrone/CT-DNA complex. On the other hand, only negligible changes were found for the drug! [d(ApT),]z complex, as compared with the SERS spectrum of free mitoxantrone (Figure 4). Hence, it is certain that mitoxantrone is intercalated into the CpG and/or GpC sequence but not (or with a much lower binding constants) into a TpA or ApT sequence. On the basis of what is known about its structure and bond moments, it is clear that the mitoxantrone chromophore has its net dipole moment nearly along its long axis. The electrostatic potential on the interface of the two successive CG and GC base-pairs is such that a dipole moment is directed nearly perpendicularly to the long axes of both base pairs? This coincides with the direction along which the long axis of the intercalating chromophore is placed. However, the electrostatic potential is rather flat near the interface of the two successive TA and AT base pairs and no preferable dipole-moment direction seems to be produced.25 Therefore, the electrostatic potential on the interface of CG and GC base pairs just mentioned should give additional stability of a few kcal mol-' to the guest mitoxantrone compared with AT and TA base pairs. The results of our SERS study of structural specificity of mitoxantrone intercalation within the double-stranded helix show that the chromophore is partially intercalated into the CpG site. It is fixed there probably by electrostatic and van der Waals forces. This would cause little change in the ring stretching force constants but could cause appreciable changes in the effective C-0 bending force constants.2 The problem of the investigation of drugDNA interaction remains one of great significance, since most of the drugs are supposed to induce their effects at the DNA level. Compared with the techniques available for selective studies of molecular interactions within high-molecular-weight complexes, SERS spectroscopy is extremely sensitive and has excellent fingerprinting capabilities. One of the main questions of the applicability of SERS for resolving problems in physical and biological chemistry concems the possibility of making measurements under experimental conditions preserving the native conformation of the molecule. The present study shows that

the SERS spectra of mitoxantrone recorded on the hydrosol prepared by reduction of silver by trisodium citrate are identical to the corresponding pre-RR spectra obtained with the same excitation. Moreover, interaction with the silver surface does not induce any perturbations due to mitoxantrone/DNA interactions: the same spectral effects were induced by the drug! DNA binding in the pre-RR, SERS, and FT-SERS spectra. These effects show that rings A and B of the mitoxantrone chromophore (Figure 1) are intercalated within the DNA, whereas ring C is localized outside the double-stranded helix: the data correlate well with the results obtained by NMR spectroscopyi2 and biochemical techniques.' i,13,14 Acknowledgment. The research was supported by Grant 6225 from the Association pour la Recherche contre le Cancer (France), Lederlk Laboratories (Rungis, France), and, in part, by Grants MHAOOO from the Intemational Science Foundation (U.S.A.) and INTAS (E.C.). The authors wish to thank Dr. N. Bystrov (Shemyakin & Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Sciences) for the generous gift of samples of [d(CpG)& and [d(ApT)gIz. Registry No. Mitoxantrone, 65271-80-9; solvent blue 35, 64862-96-0; Ag, 7440-22-4. References and Notes (1) Smulevich, G.; Feis, A. J. Phys. Chem. 1986, 90, 6388-6392. (2) Nonaka, Y.; Tsuboi, M.; Nakamoto, K. J. Raman Spectrosc. 1990, 21, 133-141. (3) Nabiev, I. R.; Sokolov, K. V.; Manfait, M. Eiomolecular Spectroscopy; Wiley: London, 1993; Chapter 7, pp 267-338. (4) Morjani, H.; s o u , J.-F.; Nabiev, I.; Lavelle, F.; Manfait, M. Cancer Res. 1993, 53, 4784-4790. (5) Nabiev, I.; Chourpa, I.; Manfait, M. J. Phys. Chem. 1994,98, 13441350. (6) Nabiev, I.; Chourpa, I.; Riou, J.-F.; Nguyen, C. H.; Lavelle, F.; Manfait, M. Biochemistry 1994, 33, 9013-9023. (7) Nabiev, I.; Mojani, H.; Manfait, M. Eur. Eiophys. J. 1991, 19, 31 1-316. (8) Smulevich, G.; Spiro, T. G. J. Phys. Chem. 1985.89, 5168-5173. (9) Otto, A,; Mrozek, I.; Grabhom, H.; Akermann, W. J. Phys., Condens. Matter 1992,4, 1143-1168. (10) Shenkenberg, T. D.; Von Hoff, D. D. Ann. Znt. Med. 1986, 105, 67-81. (11) Lown, J. W.; Hanstock, C. C.; Bradley, R. D.; Scraba, D. G. Mol. Pharmacol. 1984, 25, 178-184. (12) Lown, J. W.; Hanstock, C. C. J. Eiomol. Struct. Dyn. 1985, 2, 1097- 1106. (13) Fox, K. R.; Waring, M. J.; Brown, J. R.; Neidle, S. FEBS Lett. 1986,202, 289-293. (14) Kapuscinski, J.; Darzynkiewicz, 2.Proc. Natl. Acad. Sci. U.S.A. 1985, 34, 4203-4213. (15) Gastilovich, E. A.; Golitzina, L. V.; Kryuchkova, G. T.; Shigorin, D. N. Opt. Spektrosk 1976,40, 800-808. (16) Anoshin, A. N.; Gastilovich, E. A.; Mishenina, K. A,; Shigorin, D. N. Zh. Fiz. Khim. 1983, 57, 1046-1049. (17) Anoshin, A. N.; Gastilovich, E. A,; Klemenko, V. G.; Kopteva, T. S.; Mikhailova, K. V.; Rodionov, A. N.; Strokach, N. S.; Shigorin, D. N. Vibrational spectra of multiatom molecules; Kolotyrkin, Ya. M., Ed.; Nauka: Moskow, 1986; pp 166-188. (18) Gastilovich, E. A.; Kryuchkova, G. T.; Shigorin, D. N. Opt. Spektrosk. 1975, 38, 282-284. (19) Anoshin, A. N.; Gastilovich, E. A,; Gorelik, M. V. Zh. Fiz. Khim. 1982, 56,1665-1670. (20) Anoshin, A. N.; Gastilovich, E. A.; Shigorin, D. N. Zh. Fiz. Khim. 1980, 54, 2474-2480. (21) Mikhailova, K. V.; Anoshin, A. N.; Gastilovich, E. A. Zh. Fiz. Khim 1985, 59, 1453-1457. (22) Singh, S. N.; Singh, R. S. Spectrochim. Acta, Part A 1968, 24, 1591- 1596. (23) Smulevich, G.; Marzocchi, M. P. Chem. Phys. 1986, 105, 159171. (24) Manfait, M.; Bernard, L.; Theophanides, T. J. Raman Spectrosc. 1981, 11, 68-74. (25) Millefiori, S ; Millefiori, A. Spectrochim. Acta, Part A 1988, 44, 17-22. JP94 17337