Chem. Res. Toxicol. 1991,4, 295-299 (25) Grant, J. L., Kramer, V. J., Ding, R., and Kispert, L. D. (1988) Carotenoid cation radicals: electrochemical, optical, and EPR study. J. Am. Chem. SOC.110,2151-2157. (26) Samokyszyn,V. M., and Marnett, L. J. (1990) Inhibition of liver
295
microsomal lipid peroxidation by 13-cis-retinoicacid. Free Radical Biol. Med. 8, 491-496. (27) Burton, G. W., and Ingold, K. U. (1984) &Carotene: an unusual type of lipid antioxidant. Science 224, 569-573.
Activation Energies and Formation Rate Constants for Organic Arsenicai-Antidote Adducts As Determined by Dynamic NMR Spectroscopy Kilian Dill,*,?Lihua Huang,? Daniel W. Bearden,? Evelyn L. McGown,* and Richard J. O'Connor* Department of Chemistry, Clemson University, Clemson, South Carolina 29634-1905, and Chemistry Branch, Letterman Army Institute of Research, Presidio of San Francisco, California 94129-6800 Received November 8, 1990 Phenyldichloroarsine reacts with 1,3-dimercapto-2-propanol and 1,2-dimercaptopropane to form 1:l adducts in the form of a six-membered and five-membered heteroatom rings. Two geometric isomers for each compound are present in dynamic equilibrium. Rate constants and the activation barriers for the interconversion of the geometric isomers were determined by dynamic NMR spectroscopy. The activation barriers indicate that the fivemembered heteroatom ring is more stable than the six-membered heteroatom ring. Trivalent arsenic reacts with 1,2- and l,&dithiol compounds (dimercaptans) to form five-membered and sixmembered heteroatom rings (I, 2). Because the adducts are extremely stable relative to the adducts formed with biological molecules, dithiols can serve as antidotes to arsenic poisoning by extracting tissue-bound arsenic. Nearly 50 years ago, British antilewisite (2,3-dimercaptol-propanol; BAL) was selected as a topical antidote to lewisite ( 3 ) . It has remained the customary antidote for systemic poisoning in the United States, despite several drawbacks, including its inherent toxicity. In our search for improved arsenic antidotes, we have employed multinuclear NMR spectroscopy to study the structures, solution dynamics, binding phenomena, and chemistries of various arsenic-dithiol adducts (4-8). In the present study, we compared properties of five-membered and six-membered cyclic adducts. The BAL analogues that we exam(DMP) and 1,2-diined were 1,3-dimercapto-2-propanol mercaptopropane (PDT), which react with trivalent arsenic to form six- and five-membered rings, respectively (Schemes I and 11). In this paper we present the preparation, structure(s), and solution dynamics of the adducts formed between phenyldichloroarsine (PhAsClJ and the two related dithiol-containing antidotes DMP and PDT. We noticed unexpected cross-peaks between the major and minor resonances in our 2D-NOESY spectra for both adducts. These cross-peaks suggested the presence of two conformers in equilibrium. Using lH NMR relaxation rate spectroscopy, we were able to determine the interconversion rates and activation energies for the geometric isomen present for each adduct (Schemes I and 11).
Experimental Section
Chemical Co. and purified as described earlier (9). Methanol-d, (99.96%) was a product of Merck, Sharpe & Dohme. Methods. The PhAs(DMP) and PhAs(PDT) adducts were formed by the addition of an equimolar quantity of PhAsC12 to DMP or PDT in a 10-mm NMR tube which contained 3 mL of methanold,. The sample was flushed with gaseous nitrogen and capped. Graphic models of the compounds were obtained with the Chem 3D program (Cambridge Scientific Computing, Inc.) using bond angles and bond lengths obtained from a recently published crystal structure of an arsenic- and sulfur-containingrelated heteroatom system (10). NMR spectra were recorded on a Bruker AC300E and variable-temperature studies on an IBM 200. For PhAs(DMP) adduct, kinetic parameters and activation energies were obtained by using saturation-transfer techniques (see below) (11). For the PhAs(PDT) adduct, kinetic parameters and activation energies were obtained by using the selective inversion method via a DANTE pulse sequence (see below) (12). The second method used was found to be superior because not as many experiments were required. Relaxation Analysis. The theory, methodology, and equations used to ascertain the physical rate constants (via saturation transfer) for the interconversion of the two geometric isomers of PhAs(DMP) from the NMR experiments are given in ref 11with one important modification. Our model exhibits significant chemical shift differences between the protons in the two observed geometric isomers. Thus, for the two spins of interest (a and b), the spectrum displays four separate multiplet patterns: a and b in one conformation and a' and b' in another conformation. These are identified in Figure 1. For our case, the z-component of the magnetization becomes
d dt
E];
bz
Materials. DMP and PDT were purchased from Aldrich Chemical Co. and used without purification. *HNMR spectroscopy showed that DMP and PDT were 299% pure. Phenyldichloroarsinewas purchased from Research Organic/Inorganic + Clemson
University.
* Letterman Army Institute of Research. 0893-228~/91/2704-0295$02.50/0 0 1991 American Chemical Society
296 Chem. Res. Toxicol., Vol. 4, No.3, 1991
Dill et al. Scheme I
w
Scheme I1
Q
a
b
c & c'
1
tI-
3.8
3.7
L
I
I
I
I
2.9
2.8
2.7
2.6
L
7 2.5 2.4
PPM
Figure 1. 'H NMR (at 300 MHz) spectrum of 624 mM DMP and PhAsCl2 (1:l molar ratio) at 293.6 K. Spectrum required 16 scans. where a:, a,, b:, and b, are the instantaneous values of the z-magnetizationof the four multiplets and ab, ao,bb, and bo are the equilibrium values. The quantities 6,d, pa, pat, pb, and pw are the unknown relaxation rate constants for intramolecular relaxation; kl and k2 are the unknown chemical exchange rates
for the conformationalexchange process. The rates kl and k2 are related to the equilibrium rate constant by K = kl/k2,which can be determined from the integrals of the NMR spectrum in Figure 1. Thus our model has eight unknown quantities as shown in eq 1. The hydrogen atoms of c and c' are not included in our calculations because our inversion-recovery ()'2 results suggest a small 6 value. The general solution to eq 1is quite complicated, so we have applied approximate solutions in a manner similar to ref 11. For our first set of experiments, we applied the standard inversionrecovery pulse sequence: 180°-t-900-observe. If we evaluate eq 1within the limit of small values oft, the initial recoveries can be written:
This operation is particularly useful because it produces two independent cross relaxing systems, (a, b) and (a', b'), which can be treated exactly like the simple system in ref 11. Thus, if 2(p, - pblt
