Chem. Res. Toxicol. 1989,2, 181-185
181
Structure and Dynamics of a Lipoic Acid-Arsenical Adduct Kilian Dill,**?Earle R. Adams,? Richard J. O'Connor,§ and Evelyn L. McGown*,s Department of Chemistry, Clemson University, Clemson, South Carolina 29634, and Chemistry Branch, Biophysical Research Division, Letterman Army Institute of Research, Presidio of San Francisco, California 94129 Received January 16, 1989 The lipoic acid-phenyldichloroarsine adduct was prepared in methanol, and the structure and molecular motions of this adduct were studied. T h e results showed t h a t a six-membered heteroatom adduct was formed. One-dimensional and two-dimensional NMR spectroscopy was used to confirm the structure and assign some of the resonances in the proton and carbon spectra. Spin-lattice relaxation times of the various carbon atoms indicated that the overall molecular reorientation time (TR)of the molecule is 0.02 ns at 30 "C. An Arrhenius plot of the data showed that the activation energy (E,) for molecular tumbling is 13.4 kJ/mol.
Organic arsenicals containing a single lipophilic alkyl or aryl group are potent metabolic poisons (1)because of their high affinity for sulfhydryls. They are transported systemically as glutathione adducts within the erythrocytes (2-4). In metabolizing tissues, there are many potential sulfhydryl targets, especially those involved in the active sites of enzymes (5-11). Among the enzymes most sensitive to arsenicals are a-keto acid dehydrogenases (8-1 1). For example, the pyruvate dehydrogenase complex contains lipoamide dehydrogenase. It acts on a lipoamide moiety that is linked [via the carboxyl group of lipoic acid, 6,8-dithiooctanoic acid (l)]to the e-amino group of a lysyl residue (12) of the acyl transferase. In order to investigate the mechanism of cytotoxicity of organic arsenicals toward enzymes containing lipoic acid, we present herein a detailed study of the reaction, stoichiometry, structure, and molecular dynamics of the cis lipoic acid-phenyldichloroarsine adduct (structure 2). One-dimensional and two-dimensional multinuclear NMR spectroscopy was used, and the effect of temperature on the spin-lattice relaxation times (Ti's) of the carbon atoms was examined.
1
Materials and Methods Lipoic acid (reduced form, 98%) was purchased in 100-mg lots in sealed ampules from Sigma Chemical Co. Methanol-d4 (99.96 atom %) was obtained from Merck, Sharpe and Dohme. Phenyldichloroarsine (PhAsC12)was purchased and purified as previously described (2). The lipoic acid-PhAsC12 adduct was prepared by dissolving 100 mg of lipoic acid in 4 mL of methanol-d4 contained in a constricted 10" NMR tube. This was followed by the addition of 65 pL of PhAsC12 (1:l stoichiometry). The sample was then put through three freeze-thaw cycles under vacuum in order to ensure that all of the gases had been expelled before sealing the NMR tube under vacuum. Lipoic acid samples were prepared for NMR spectral studies by dissolving 100 mg of lipoic acid in methanol-d4 in a 10-mm NMR tube. The head space over the sample was saturated with gaseous nitrogen before capping the sample. All sample temperatures were maintained a t 30 "C, except for the temperature studies. NMR data were obtained from a Varian XL-300 spectrometer operating a t 299.945 and 75.429 MHz for the observation of 'H and 13Cnuclei, respectively. Spectral windows of 4000 and 16000 Hz and 90" pulse widths of 20 and 15 ps were typically used for 'H and 13C 1D spectra, respectively. The 2D double quantum filtered COSY (DQCOSY) spectra were acquired as a 256 X 1024 data matrix with a spectral window of 4000 Hz. The data matrix was processed with a phase-shifted sine-bell function in both dimensions and zero filled to 1024 X 1024. 2D HETCOR spectra were acquired in a 512 X 2048 matrix. The data matrix was processed with a phase-shifted sine-bell function, and the matrix was zero filled to 2048 X 2048. The spectral windows for the 2D spectra (DQCOSY and HETCOR) were the same as those used for the acquisition of the simple 'H and 13C 1D spectra. Distortionless enhancement of polarization transfer (DEPT) experiments were carried out by using Varian programs. Spinlattice relaxation times were determined by using the inversion-recovery method with 14 7 values, ranging from 0.10 to 10 s, and the data were fit to an exponential function. The temperature of the probe was verified by the proton chemical shifts of methanol. Nuclear Overhauser enhancement values were obtained by using the gated decoupling technique with the pulse delay set to 5 times the longest 5"' value observed for the aliphatic carbon atoms of the adduct.
Results and Discussion
8 2
* To whom correspondence should be addressed. + Clemson
University.
* Letterman Army Institute of Research.
We have previously shown that PhAsCla reacts with vicinal thiol groups on a molecule to form a stable fivemembered heteroatom ring; this work is relevant to the development of antidotes to arsenic poisoning (13,14). All the arsenical adducts we have studied so far by NMR spectroscopy (13,14) and X-ray crystallography1prefer-
0893-228x/89/2702-0181$01.50/0@ 1989 American Chemical Society
Dill et al.
182 Chem. Res. Toxicol., Vol. 2, No. 3, 1989
solvent
I '
8
' ' I ' ' " I "
7
' ~ " " l " " ~ " " l " " " " l " " ~ ' ' ' ' l ' " ' ~ ' " ' l " " " "
6
5
4
2
3
p p m from M e 4 Si
Figure 1. 'H NMR spectra of lipoic acid and lipoic acid-PhAsCl, adduct. Time domain data were collected in 64 000 addresses with a recycle time of 15-18 s. (A) Lipoic acid; spectrum required 16 transients. (B) Lipoic acid-PhAsCl, adduct; spectrum required 150 transients.
entially form the five-membered ring adduct having the trans configuration (between the phenyl ring and the functional group on the alkyl chain). Lipoic acid, on the other hand, contains l,&dithiol groups, and if a one-to-one adduct is formed between the arsenical and lipoic acid, a six-membered heteroatom ring should be produced. Figures 1 and 2 show the lH NMR spectra and the proton-decoupled, natural abundance 13C NMR spectra of lipoic acid and the lipoic acid-PhAsCl, adduct. The proton spectra appear extremely complex, as would be expected for a long aliphatic side chain of a molecule such as lipoic acid. The carbon spectra, on the other hand, show well-resolved single-carbonresonances, and these are more amenable to direct assignments. In both the proton and carbon spectra, the addition of PhAsCl, to the lipoic acid solution provided new signals in regions of the spectra that could be attributable to aromatic resonances (-7.5 and -130 ppm in the lH and 13Cspectra, respectively). Some of the resonances for unreacted lipoic acid (at 22.7, 25.7, 27.6,34.8,39.8,40.0, and 44.2 ppm in the aliphatic region) have disappeared (or actually shifted) in the PhAsC1,treated sample, which indicates probable formation of a 1:l complex. Some minor peaks have also appeared which may represent a byproduct of the reaction or the formation of a minor isomer. Due to 1,3-substitutent interactions in a cyclohexyl ring, the cis isomer (2) would most likely be the major product because it places the substituents in the equatorial-equatorial positions. The trans isomer (3) is probably the minor product, and in this case the substituents could be in the axial-equatorial positions. Formation of the adduct causes a number of substantial changes in the 13CNMR spectrum of lipoic acid. The resonances that appear to be most affected are those at 22.7,40.0, and 44.2 ppm; they represent C8, C6, and C7, respectively (15). E. R. Adams, K. Dill, D. Jeter, A. W. Cordes, and J. K. Kolis, unpublished results.
Other resonances show minor changes in A6 or do not appear to be affected at all.
3
The specific assignments of the resonances in the 13C spectra of lipoic acid were taken from the data published by Paukstelis and co-workers (15). The proton and carbon assignments in the lipoic acid-PhAsCl, spectra were obtained by 1D and 2D NMR methods, and the spectral analyses are given below. As indicated by structure 2, there are three types of proton and carbon atoms present in the lipoic acidPhAsCl, adduct: the atoms of the phenyl ring, the atoms within the six-membered heteroatom ring, and the atoms of the aliphatic portion of lipoic acid (Cl-C5). For studies of the molecular motion of various portions of the adduct, 13Cspectral assignments were necessary. The proton resonance assignments were made from the DQCOSY spectrum and from the related assignments in the carbon spectrum. Since the H8 protons (2.81 ppm) are at the end of the molecule, they should be coupled only to the H7 protons. Note in Figure 3 that there is only one cross-peak associated with H8 and this has been assigned
Chem. Res. Toxicol., Vol. 2, No. 3, 1989 183
Lipoic Acid-Arsenical Adduct
I
I
8
I
,
70
F2 IPPM) B O I
90
4
100
-
110;
130P
- 1 O 2 l140 8
7
6
5
4
3
1
Fl IPPM)
F i g u r e 4. 13C-lH HETCOR contour plot for the lipoic acidPhAsClz adduct. Spectral windows of 4000 and 16 000 Hz were used for 'H and 13C,respectively. The data were collected in a 512 X 2048 matrix by using 32 scans per trace and a 7-s delay between scans. A line-broadening factor of 5 Hz was used in processing the data in the 13C dimension. Table I. 140
110
100
60
80 ppm l m m
40
Me, SI
Figure 2. Proton-decoupled natural abundance 13CNMR spectra of lipoic acid and lipoic acid-PhAsC12 adduct. The time domain data were collected in 52 352 addresses with a recycle time of 16.6 s. A line-broadening factor of 2 Hz was used in processing the data. (A) Lipoic acid; spectrum required 528 transients. (B) Lipoic acid-PhAsClz adduct; spectrum required 355 transients.
A
8
Chemical Shift and Relaxation Data for the Lipoic Acid-PhAsC12 Adduct" at 30.6 OC NT,,b s chemical shift Aromatic Region 140.5 (nonprotonated) 133.0 2.55 (2.77) 130.3 2.58 (2.82) 129.9 2.58 (2.80) Aliphatic Region 41.1 (C6) 39.7 (C5) 36.9 (C7) 34.6 (C2) 28.8 (C8) 26.1 (C4 or C3) 25.5 (C4 or C3)
2.28 (2.94) 3.24 (2.88) 2.66 (3.04) 5.34 (2.94) 2.61 (2.96) 3.86 (2.90) 3.86 (2.76)
'For the major isomer only. Some resonance assignments are provided. The carbonyl carbon atom is not given, but it was found to resonate at 171.6 ppm. bOnly determined for protonated carbon atoms. NOES are given in parentheses. The TI of the nonprotonated aromatic carbon atom was found to be >30 s.
1.4-
1.6-
4
1.8-
.
0.
2.0 -
3.2
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
1.4
F1 (PPM)
Figure 3. 'H DQCOSY contour plot for the lipoic acid-PhAsClz adduct. Spectral windows of 4OOO Hz were used in each dimension. The data were processed in a 256 X 1024 matrix by using 20 scans per trace and a 7-s delay between scans.
to H7 (1.87 ppm). H7 must be coupled to H6 as well as H8. A cross-peak is observed in Figure 3 at 2.61 ppm that can be assigned to H6. We have only indicated some of the resonance assignments in Figure 3, but others can be made with some certainty. For example, H6 is coupled to a resonance centered in the complex multuplet(s) at about 1.4 ppm; this is probably H5. H4 resonates at 1.5 ppm, and this can only be determined from the HETCOR contour plot. The H4 resonance appears to be coupled also to a resonance at 2.3 ppm (H3). Further assignments are not possible because of the complexity of the spectrum and the fact that the intensity of the cross-peaks also depends upon the respective J values. Figure 4 shows the HETCOR contour plot for lipoic acid-PhAsC12. This contour plot provided many of the assignments given in Table I. These assignments were corroborated by the DEPT experiment and the Tl values obtained (nonnormalizedvalues). The DEPT experiment clearly showed that the resonance at 41.1 ppm was a methine carbon atom (C6). Furthermore, the T1value of
184 Chem. Res. Toxicol., Vol. 2, No. 3, 1989
this carbon atom is at least twice as long as that of C7 and C8; this would be consistent with the expected results due to the number of attached protons. Further, but not concrete, assignments could also be made on the basis of the DQCOSY results. For example, a cross-peak exists in the HETCOR contour plot for the assigned H5 proton and a resonance at 39.7 ppm in the carbon spectrum; this must be C5. H4 (1.5 ppm) appears to show a cross-peak to a resonance a t about 26 ppm (C4). The assignments of C8-C6 are made likewise. Note that the results of Figures 1,3, and 4 indicate that the two H8 protons (cis and trans with respect to the phenyl ring) appear to be magnetically equivalent; this is also observed for the two H7 protons (structure 2). In all our studies dealing with the five-membered heteroatom ring adducts (using 2,3-dithiopropanol and 2,3-dithiopropanesulfonic acid) we have found that the two protons attached to the same carbon atom within the ring system (trans and cis with respect to the phenyl ring) exhibited rather substantial chemical shift differences (up to 2 ppm) (13,14). Apparently, in the six-membered heteroatom ring system, the phenyl ring has little steric influence (or ring current effects) on the H7 and H8 protons. In the fivemembered heteroatom system it was also assumed that chair interconversion was slow when compared to overall molecular reorientation of the molecule or the rotation of the phenyl ring about the As-C1’-C4’ axis. Spin-lattice relaxation times can be used to obtain information about the structure and fluctional dynamics of a molecule, provided that some of the molecular motions are understood and that the Tl-relaxation process is dominated by a dipole-dipole (‘H-13C) mechanism. For a small molecule under the extreme narrowing conditions and a Tl-relaxation process that is dominated by the dipolar mechanism, the Tl’s for the carbon atoms are defined by the equation: (TiDD)-’ = h2yC2yH2NrCH4(TR) In this equation, T1 is a function of nuclear constants, carbon-hydrogen distances (dipole-dipole interactions), the number of directly bonded hydrogens (N), and the molecular reorientation time of the molecule. In order to substantiate that the dipolar mechanism is dominant for the T,-relaxation of the carbon atoms of the adduct, the NOE’s of the carbon atoms were measured a t various temperatures. Under extreme narrowing conditions, the NOE value for the carbon atoms should approach 3.0, if the dipolar relaxation mechanism is dominant. The NOE values that we observed for the various carbon atoms of the adduct at 30 “C are given in Table I. NOE values of 2.8-3.0 clearly indicate the predominance of a lH-13C dipolar mechanism in the spin-lattice relaxation process for the carbon atoms. Similar NOE values were observed at the same temperatures that T1measurements were made. The three molecular motions of interest are the molecular reorientation rate (activation parameters), ring inversion rate, and rotation of the phenyl ring. These three motional parameters were obtained by Aksnes and coworkers (16-18) for several five-membered ring systems containing a phenylarsenic moiety, using spin-lattice relaxation data. For their adducts, they assumed that the overall reorientation was much faster than ring inversion. Furthermore, their data clearly showed that the NT1 values of the protonated carbon atoms of the phenyl moiety were nonequivalent and these in general were longer than the ring carbon N T , values. They rationalized their results on the basis that the phenyl ring exhibited fast ”internal” rotation about the As-C1’-C4’ axis, which in turn was
Dill et al. 1.3
1.2
--c
1.1
C
1.0
3.0
3.1
(11~)
3.2
3.3
io3
Figure 5. Arrhenius plot for the aromatic carbon atoms of lipoic acid-PhAsClz. Two of the aromatic carbon resonances showed almost identical relaxation rates at all temperatures.
faster than the overall molecular reorientation of the adduct. This is definitely not the case for the lipoic acidphenyldichloroarsine adduct, where the NT, values of the phenyl carbon atoms are equivalent and equal to the NT, values of the adduct ring carbon atoms. The fact that the NT, values of the phenyl carbon atoms of the lipoic acid-phenyldichloroarsine adduct are equal to the NT1 values of the carbon atoms of the heteroatom ring system (C6-W indicates that their reorientation occurs at approximately the same rate (assuming an isotropic rigid rotor at 30 “C, 7R = 0.02 ns) or that the overall molecular reorientation of the molecule dominates the motional process. In contrast to the NT1 values exhibited by the phenyl ring carbon atoms and heteroatom ring carbon atoms, the aliphatic side chain carbon atoms of lipoic acid exhibit somewhat longer NT, values. This probably results from the greater flexibility of the aliphatic side chain. Note that, due to their greater flexibility, the carbon atoms furthest away from the heteroatom ring system (C2) tend to show longer NT, values than those closer to the heteroatom ring system (C5). In order to gain further information about the energies of the molecular motions, we monitored the Tl’s of the respective carbon atoms as a function of the absolute temperature (Arrhenius plots). This could be accomplished because the T1 values are directly proportional to the values of T~ and are also temperature dependent. Figure 5 shows the Arrhenius plot for the carbon atoms of the phenyl moiety of the lipoic acid-phenyldichloroarsine adduct. The slope of the line of the Arrhenius plot is equal to -E,.R. The value obtained from the plot shows that the activation energy for the overall molecular reorientation of the molecule is 13.4 kJ/mol. Similar results were also obtained from the Arrhenius plots of the carbon atoms of the six-membered heteroatom ring system and the carbon atoms of the aliphatic side chain of the lipoic acid adduct. The value obtained for the activation energy of our molecule appears to be much lower (in some cases by a factor of 2.5) than those obtained by Aksnes and coworkers for “simple” diarsolanes and dithioarsolanes in benzene-d6. However, the authors clearly indicated in a series of elaborate papers that the very simple dioxoarsolanes and dithioarsolanes probably dimerize a t lower
Lipoic Acid-Arsenical Adduct temperatures and that this may indeed affect their activation energy values. Our value is closest to one of their systems that did not dimerize to a great extent and therefore represents a true activation parameter which is not influenced by other factors. Consistent with the observations of Aksnes (16-18), our system evidently did not show dimerization because the bulky side chain of lipoic acid apparently prevents dimerization.
Acknowledgment. K.D. acknowledges the support of the National Research Council and Battelle Memorial Institute. Registry No. 2, 120789-75-5.
References (1) Webb, J. L. (1966) in Enzyme and Metabolic Inhibitors (Webb,
J. L., Ed.) Vol. 3, Academic Press, New York. (2) Dill, K., Adams, E. R., O’Connor, R. J., Chong, S., and McGown, E. L. (1987) One-dimensional and two-dimensional nuclear magnetic resonance studies of the reaction of phenyldichloroarsine with glutathione. Arch. Biochem. Biophys. 257, 293-301. (3) Dill, K., O’Connor, R. J., and McGown, E. L. (1987) Spin-echo NMR investigation of the interaction of phenyldichloroarsine with intact erythrocytes. Znorg. Chin. Acta 138, 95-97. (4) Chong, S., Dill, K., and McGown, E. L. (1989) Interaction of phenyldichloroarsine with erythrocytes. J. Biochem. Toxicol. In press. (5) Drummond, G. I. (1981) Inactivation of cardiac adenylate cyclase by oxidation, trivalent arsenicals, and N-ethylmaleimide. Arch. Biochem. Biophys. 211, 30-38. (6) Voordouw, G., Van der Vies, S. M., Veeger, C., and Stevenson, K. J. (1981) Modification of the thiol residues of pyridine nucleotide transhydrogenase from Azobacter vinelandii. Activity modulation by the divalent thiol reagent p-aminophenylarsenoxide. Eur. J. Biochem. 118,541-546. (7) Brown, S. B., Turner, R. J., Roche, R. S., and Stevenson, K. J. (1987) Spectroscopic characterization of thioredoxin covalently
Chem. Res. Toxicol., Vol. 2, No. 3, 1989 185 modified with monofunctional organoarsenical reagents. Biochemistry 26, 863-871. (8) Stevenson, K. J., Hale, G., and Perham, R. N. (1978) Inhibition of pyruvate dehydrogenase multienzyme complex from Escherichia coli with mono- and bifunctional arsenoxide. Biochemistry 17,2189-2192. (9) Adamson, S. R., Robinson, J. A., and Stevenson, K. J. (1984) Inhibition of pyruvate dehydrogenase multienzyme complex from Escherichia coli with a radiolabeled bifunctional arsenoxide: evidence for an essential histidine residue at the active site of lipoamide dehydrogenase. Biochemistry 23, 1269-1274. (10) Knowles, F. C. (1985) Reactions of lipoamide dehydrogenase and glutathione reductase with arsonic acids and arsonous acids. Arch. Biochem. Biophys. 242, 1-10, (11) Danson, M. J., McQuattie, A., and Stevenson, K. J. (1986) Dihydrolipoamide dehydrogenase from halophilic bacteria: purification and properties of the enzyme from Halobacterium halobium. Biochemistry 25, 3880-3884. (12) Reed, L. J. (1974) Multienzyme complexes. Acc. Chem. Res. 7, 40-46. (13) Dill, K., Adams, E. R., O’Connor, R. J., and McGown, E. L. (1987) 2D NMR studies of the phenyldichloroarsine-British anti-lewisite adduct. Magn. Reson. Chem. 25, 1074-1077. (14) OConnor, R. J., Dill, K., McGown, E. L., and Hallowell, S. F. (1989) 2D-NMR studies of arsenical-sulfhydryl adducts. Magn. Reson. Chem. In press. (15) Paukstelis, J. V., Byrne, E. F., Connor, T. P., and Roche, T. E. (1977) Carbon-13 nuclear magnetic resonance spectra of thiols and thioacetals: lipoic acid and derivatives. J . Org. Chem. 42, 3941-3944. (16) Aksnes, D. W., and Holak, T. A. (1981) Carbon-13 NMR spinlattice relaxation in dithioarsolane. Org. Magn. Reson. 17, 285-289. (17) Aksnes, D. W., and Ramstad, K. (1984) Carbon-13 NMR spinand its lattice relaxation studies on 2-chloro-1,3,2-dioxarsolane ring-substituted methyl derivatives. Acta Chem. Scand. A30, 781-787. (18) Aksnes, D. W., and Ramstad, K. (1985) Carbon-13 NMR spinlattice relaxation and molecular motion in 2-phenyl-l,3,2-dioxarsolane and its 4-methyl and 4,5 dimethyl derivatives. Magn. Reson. Chem. 23, 253-258.