Interaction of Benzo[ alpyrene with Caffeine
The Journal of Physical Chemistry, Vol. 82, No. 26,
1978
2829
Interaction of Benzo[ a Ipyrene and 6-Oxybenzo[ a Ipyrene with Caffeine. Structures of the Complexes as Studied by Nuclear Magnetic Resonance Chemical Shift and Relaxation Yoshio Nosaka, * Kazuyuki Akasaka, and Hiroyuki Hatano Department of Chemistry, Faculty of Science, Kyoto University, Kyoto 606, Japan (Received January 26, 1978, Revised Manuscript Received June 21, 1978)
The interaction of caffeine with benzo[a]pyrene (BP) and with 6-oxybenzo[a]pyrene (OBP), a carcinogenic free-radical intermediate, has been studied with the proton nuclear magnetic resonance (NMR) method. Observation of ring current shift for both caffeine and BP molecules in CDC13led to the conclusion that caffeine associates with BP by a stacking arrangement in this solvent. The magnitude of the chemical shifts observed for BP solubilized by caffeine-dloin D20 suggests the formation of a 1:2 BP-caffeine complex. The formation of a 1:2 complex is independently supported by the results of measurements of the spin-lattice relaxation time (T,) which shows that the rotational correlation time of the caffeine-BP complex is a factor of 3-4 longer than that of caffeine itself. The OBP radical is solubilized by caffeine in D20 and enhances the l/T1 of the four kinds of protons of caffeine. With these enhancements, the average distances between the protons of caffeine and the electron spin of the radical in the complex were calculated. To determine probable structures of the complex, these distances were computer simulated for all possible intermolecular orientations. The most probable structure is such that an OBP molecule is sandwiched between two caffeine molecules in a parallel arrangement with a plane-to-plane distance of 0.33 nm, similar to that of the 1:2 complex between BP and tetramethyluric acid crystals. This structure is in qualitative agreement with the structure of the complex between the diamagnetic BP and caffeine expected from chemical shift measurements.
Introduction Benzo[a]pyrene (BP) has been known as a potent carcinogen for the past few decades. Recently, it has been shown that B P is activated enzymatically to 6-hydroxyb e n z ~ [ a ] p y r e n e ,or ~ $to ~ a kind of e p o ~ i d e . ~The , ~ 6hydroxybenzo[a]pyrene, which is the precursor of the OBP radical, reacts spontaneously and covalently with a guanine base of nucleic acids, and this kind of reaction seems to be the initial stage for the outbreak mechanism of cancer.5,6 Caffeine is an important compound in terms of its basic purine structure and its good solubilizing characteristic^.^ In addition, it is also an important compound in the modification of the carcinogenic effects of physical as well as chemical agents in living system^.^,^ This modification of carcinogenesis seems to be closely related to the solubilizing activity, and the study of the interaction between caffeine and B P or OBP would be important for the understanding of the carcinogenic activity of BP. Interactions between purines and aromatic hydrocarbons have been investigated in the solid state and also in solutions. For the structure of the complexes, a parallel arrangement of the two molecular planes has been reported from X-ray data of the mixed crystals,lOJ1but this arrangement was not supported directly in solution by means of NMR s p e c t r o ~ c o p y . ~ Hence, ~J~ we investigate the detailed structure of the complex between caffeine and BP in solution by means of pulse Fourier transform NMR spectroscopy and a computer-simulation technique developed recently by the authors.14 Experimental Section Benzo[a]pyrene and caffeine were purchased from Nakarai Chemical Co. and used without further purification. The B P radical crystals were prepared from 6hydroxybenzo[a]pyrene by chemical oxidation with chloranil in benzene. The 6-hydroxybenzo[a]pyreneused was a generous gift from Dr. C. Nagata. Caffeine-d,, was synthesized by the methylation of xanthine with di0022-3654/78/2082-2829$0 1.OO/O
methyl-d, sulfate15 and the successive deuteration of H-8 by heating in DzO. For NMR T , measurement the samples were prepared according to the following procedure. Chemically produced OBP radicals were dissolved in a D 2 0 solution of caffeine (40 mM) and EDTA (1 mM) by grinding in a mortar, unresolved crystals were removed by ultracentrifugation, and finally the samples were deaerated by bubbling nitrogen gas. EDTA was added to reduce the effect of possible paramagnetic metal impurities. The concentrations of the radicals were estimated from the integrated intensity of the electron spin resonance (ESR) signals, using a 2,2,6,6-tetramethyl-4-piperisinol 1-oxy1 aqueous solution as a standard. ESR spectra were measured a t X-band with a JEOL PE-3X ESR spectrometer. All NMR measurements were carried out using a JEOL PS-100 NMR spectrometer equipped with a pulse Fourier transform unit operating a t 100 MHz. Simulation of the structure of the molecular complex was carried out with a FACOM M-190 computer system in the Kyoto University Data Processing Center. The simulation program used was almost the same as program I in our preceding work.14
Results and Discussion 1. Complex Formation between BP and Caffeine in CDCl,. NMR chemical shifts have been used to determine the structure of the complex between aromatic hydrocarbons and purines in organic solvents.12J3 Hanna and Sandoval pointed out that the five-membered ring of caffeine interacts with benzene in CC14.12Recently, Donesi and his co-workers proposed a perpendicular arrangement for the complex between 1,3,7,9-tetramethyluric acid (TMU) and aromatic hydrocarbons in CDCI3.l3 However, they discussed the structure only from the proton chemical shift of TMU induced by aromatic hydrocarbons, without considering the shift of aromatic hydrocarbons induced by TMU. Hence, we have examined the interaction between C 1978 American Chemical Society
2830
The Journal of Physical Chemistry, Vol. 82,
No. 26,
Y. Nosaka, K. Akasaka, and H. Hatano
1978
TABLE I: Association Constant ( K )and Limiting Induced Shifts (6 ) Obtained for Different Protons of Caffeine for t h e 1:1 Association between Caffeine and Benzo[a] pyrene in CDCl,n proton
H-8
K, M-' 6 PPm
1.24 2.02
9
a
?
i
0.33 0.44
H,C-7
H3C-3
1 . 1 6 i 0.20 2.03 i 0.29
1.39 i 0.23 0.94 i 0.13
H,C-l 1.44 0.69
k f.
0.31 0.12
Calculated by non-linear least-squares fit to the shifts of caffeine ( 5 mM) induced by benzo[a]pyrene ( 0 - 2 0 0 m M )at 29
"C. TABLE 11: Observed Upfield Shifts (6 ) of Benzo[a]pyrene (9 mM) in t h e Presence of Caffeine-d,, (0.75 M) in CDCI, at 29 "Ca proton 1 0 0 6 , ppm
1
2
3
4
5
6
7
8
9
10
11
12
5.9
3.6
5.6
8.0
10.6
13.2
6.1
2.9
3.0
11.4
13.5
9.5
Chemical shifts were calculated with computer simulation of the spectrum, and are believed t o be accurate t o *0.01 ppm. Limiting shifts are estimated from the association constant (Table I) t o be twice these values.
B P and caffeine in CDC13 by conventional shift analysis. A t first, the chemical shifts of protons of caffeine (5 mM) were measured in the presence of BP (eight different concentrations, 0-200 mM). The NMR spectrum of caffeine consists of four well-separated signals which have already been assigned.16 By assuming an equilibrium equation to form a 1:l complex, the association constants and the limiting chemical shifts were calculated by the non-linear least-squares fit method, and are listed in Table I. The fact that the association constants obtained for the four kinds of protons agree with each other within the standard deviation well supports the assumption of a unique structure for the complex. The chemical shift induced by the formation of the complex is large for H-8 and for the methyl protons at N-7 (H3C-7),indicating BP associates preferentially with the five-membered ring of caffeine as pointed out by Hanna and Sandoval for the case of benzene in CC14.12 In an alternate experiment, the chemical shifts of BP (9 mM) were measured with and without caffeine in CDC1,. The NMR spectrum of BP itself is rather complicated as shown in Figure IC, however the assignment of the spectrum has already been done by Haigh and Mallion at 220 MHz.l' The chemical shifts from an external reference (tetramethylammonium chloride solution) could therefore be estimated from the simulation of the recorded spectra. Table I1 shows the calculated values of the induced chemical shifts in a 0.75 M caffeine-dlo solution. The fact that the NMR signals not only of caffeine (Table I) but also of BP (Table 11) shift significantly upfield by mutual association clearly indicates that the complex is formed by a stacking interaction in CDC1,. Hence, the perpendicular arrangement,13 which would cause the low field shifts of BP signals, is not plausible for the caffeine-BP complex. The association constant for the 1:l complex given in Table I (averaged value of 1.30 M-I) indicates that only one-half of the B P is complexed with one caffeine at the concentration of Table I1 (0.75 M). Then the limiting chemical shifts of BP as a result of complexation with caffeine would be approximately a factor of 2 larger than those in Table 11. Hence, limiting shifts of some of the protons which are located at the central part of BP molecules (Le., H-5, -6, -10, -11,and -12) are expected to be as large as 0.2 ppm or more. This shows that, in the stacking complex, caffeine interacts preferentially with the central part of the BP molecule, but it is difficult to discuss a detailed geometry for the complex based on these chemical shifts. 2. NMR Spectrum of BP Solubilized in a Caffeine Aqueous Solution. It had been well known that BP is solubilized by caffeine in aqueous solution and the solu-
I
9.0
83
7 3 ?pm from TMS
Flgure 1. (a) The NMR spectrum of benzolalpyrene solubilized in 40 mM caffeine-d,, D,O solution, 4000 scans, 180'-(4 s)-90' pulse sequence, 6-s repetition time. The shadowed area shows the inverted signal for H-8 of the caffeine. (b) Calculated spectrum of benzo[alpyrene for the 1.2 complex between benzo[a] pyrene and caffeine from the shift data of the 1:l complex in CDCI,. (c) 10 mM benzo[alpyrene in CDC13. The arrow shows the position of the caffeine H-8 signal in CDCI,.
bility was reported by Boyland and Green with fluorescence measurements of BP.' Substantial amounts of BP molecules form either 1:l or 1:2 complexes with caffeine in the solubilized state. However, the structure of the complex has not been studied in detail. In order to obtain information on the structure of the complex, we have attempted to measure the NMR spectrum of B P solubilized in a D 2 0 solution of caffeine. Caffeine-dIowas again used to reduce the large signals from the methyl protons of caffeine. In spite of the experimental difficulty, we were able to obtain an NMR spectrum of BP solubilized by caffeine-dlo in aqueous (D,O) solution as shown in Figure la. The concentration of BP was estimated to be 0.2 mM from integration of the signal intensity, using the EDTA signal as a standard. The H-8 signal of caffeine in this spectrum was employed to compare the chemical shift for B P in the two medium, since the chemical shift of H-8 is not significantly affected by the dimerization of caffeine itself.18 The pattern of the spectrum was different from that of BP itself in the absence of caffeine in CDC13 (Figure IC),
Interaction of Benzo[a] pyrene with Caffeine
The Journal of Physical Chemistry, Vol. 82, No. 26, 1978 2831
but it was readily recognized that the positions of whole peaks were shifted to higher field by about 0.5 ppm. The magnitude of the chemical shift change is approximately twice the limiting shift in CDC13, suggesting that a sandwich type complex is formed in the solubilized state. Based on the observed shifts in the presence of caffeine in CDC1, (Table 11),we can simulate the NMR spectrum of B P for a 1:2 complex as shown in Figure lb, by assuming a symmetrical arrangement of two caffeine molecules around B P in the complex. It is rather surprising that the overall pattern of tlhe spectrum of the solubilized B P (Figure l a ) could be fairly synthesized from the data in CDC13. The agreement of the two spectra (Figures l a acd l b ) suggests that the orientation of the caffeine molecule on B P in DzO is similar to that in CDCl,, except that in aqueous solution a stacking sandwich type arrangement (caffeine-BP-caffeine) is favored. If one assumes that the orientation of caffeine relative to B P in the complex is governed mainly by the intermolecular forces but not by the interaction with tlhe solvents, a similar orientation of the two molecules would be expected between the aqueous environment and organic solvents. The prediction of the 1:2 sandwich type complex in aqueous solution has been well supported by other techniques; the optical absorption and fluorescence spectra of BP solubilized in caffeine aqueous solutions are identical with those in DNA s o l ~ t i o n in ’ ~which strong evidence is given for the flow-dichroism measurement for the intercalation of B P into base pairs of DNA.20 In the following section, we will show that the sandwich type complex is also supported from the consideration of the rotational correlation time. 3. Rotational Correlation T i m e of the Complex in D20. In order to estimate the rotational correlation time of the complex, we measured the spin-lattice relaxation time ( Tl) of protons of B P solubilized in 40 mM caffeine-dlo D:,O solution. Although the proton signals of solubilized EIP were not separated from each other (see Figure l a ) , the Tl’s were almost the same for all higher field parts of the spectrum, and the average value was 1.14 f 0.11 s. In a separate experiment, we found that all protons, except H-10 and H-11, of BP in CDC13 also have almost the same relaxation times. The spin-lattice relaxation time of proton i would ’be expressed as a function of the proton-proton distance rLh and the isotropic rotational correlation time ~ , . ~Under l the extreme narrowing condition (7,
