954
Chem. Res. Toxicol. 1996, 9, 954-964
Radical-Initiated Lipid Peroxidation in Low Density Lipoproteins: Insights Obtained from Kinetic Modeling A. Reginald Waldeck† and Roland Stocker* Biochemistry Unit, The Heart Research Institute, 145 Missenden Road, Camperdown, Sydney, NSW 2050, Australia Received March 28, 1996X
We present kinetic models of various complexity for radical-initiated lipid peroxidation in low density lipoproteins (LDL). The models, comprised of simultaneous differential equations programmed in Mathematica, were used to evaluate the concentration profiles of the reactants of interest. Single-phase reaction schemes describing lipid peroxidation and antioxidation according to the “conventional” and tocopherol-mediated peroxidation (TMP) model were simulated for conditions of low and high radical fluxes produced by thermolabile azo initiators. The results show that the particular dependencies of the rates of lipid peroxidation (Rp) on the rates of initiation (Ri) for the two reaction schemes were accurately predicted by the simulations. Both models qualitatively predicted inhibition of lipid peroxidation in the presence of R-tocopherol (R-TOH) under high radical flux conditions, suggesting that both can describe inhibited lipid peroxidation in solution under these conditions. TMP, but not the conventional model, could also predict the experimentally observed complex behavior of LDL lipid peroxidation induced with different concentrations of azo initiators. Specifically, TMP faithfully reproduced the observed kinetic chain length of lipid peroxidation of .1 at low and ,1 at high concentration of the initiator (i.e., 0.2 and 10 mM, respectively for LDL at 1 µmol apoB100/L) during the R-TOH-containing period of oxidation. It also demonstrated the experimenon Ri. Kinetic analysis of radical generation and tally observed nondependence of RTMP p initiation of lipid peroxidation in an extended, two-compartment model of TMP showed that phase separation of bimolecular reactions in a suspension of LDL particles can lead to a ∼400fold increase in the rate of lipid hydroperoxide formation. The experimentally observed coantioxidant action of water-soluble ascorbate and lipid-soluble ubiquinol-10 were verified using this model. A simple biophysical model constituting the reactions of TMP and incorporating the compartmental nature of an LDL suspension is proposed. Together, the results demonstrate that TMP is the only model that fits the experimental data describing the early stages of LDL lipid peroxidation under various oxidizing conditions. The implications of our findings are discussed in relation to atherogenesis and a recently proposed alternative model of LDL lipid peroxidation (Abuja and Esterbauer (1995) Chem. Res. Toxicol. 8, 753).
Introduction There is increasing evidence suggesting that free radical-mediated oxidation of low density lipoproteins (LDL)1 contributes to atherogenesis in humans (1). Oxidation of LDL’s lipids often precedes and can contribute to the chemical modification of apolipoprotein B-100, which can lead to the “high uptake” form of the lipoprotein (2). Uptake of such modified LDL may cause transformation of macrophages into “foam cells”, a hallmark of developing atherosclerotic lesions. As a result, there has been much interest in LDL lipid oxidation, and specifically LDL antioxidation, because of the potential beneficial actions of natural and synthetic antioxidants. An issue of ongoing debate is whether, and under what conditions, R-tocopherol (R-TOH; biologically the most * To whom correspondence should be addressed: Roland Stocker, The Heart Research Institute, 145 Missenden Rd., Camperdown, Sydney, NSW 2050, Australia. E-mail:
[email protected]; Tel: +61-2-550-3560; Fax: +61-2-550-3302. † Present address: Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, CA 91125. X Abstract published in Advance ACS Abstracts, July 15, 1996. 1 Abbreviations: AAPH, 2,2′-azobis(amidinopropane hydrochloride); AH-, ascorbate; LDL, low density lipoproteins; LH, bisallylic methylene groups; LOOH, lipid hydroperoxide; LOO•, lipid peroxyl radical; NRP, nonradical products; QH2, ubiquinol-10; TMP, tocopherol-mediated peroxidation; R-TOH, R-tocopherol; R-TO•, R-tocopheroxyl radical.
S0893-228x(96)00057-4 CCC: $12.00
active form of vitamin E (3) and most abundant antioxidant in organic extracts of LDL (4)) inhibits or promotes the peroxidation of unsaturated lipids in intact LDL. In the more generally accepted model (Scheme 1; referred herein as “conventional” lipid peroxidation) R-TOH functions solely as an antioxidant (5-16). This model appears to successfully describe the kinetics of oxidation of active bisallylic methylene groups (LH) of lipids in homogeneous solution (5, 6, 10, 11), liposomes (5, 8, 12, 13, 16), and micelles (7, 15, 16). As a result, this mechanism is generally thought to account for lipid peroxidation in LDL (4). More recently, however, tocopherol-mediated peroxidation (TMP; Scheme 2), where R-TOH acts as both a pro-oxidant and antioxidant, has been suggested as a new mechanism describing the molecular action of the vitamin in oxidizing LDL. TMP invokes properties of the heterophasic nature of a suspension of LDL particles (17-21) and is based on experimental findings that are either contrary to or cannot be explained by the “conventional” scheme of lipid peroxidation and antioxidation (Table 1). Kinetic modeling is useful for elucidating aspects of complex biochemical pathways (22-24). Models of liposomal (25), mitochondrial (26), and cellular (27) lipid peroxidation have been reported, and these have been © 1996 American Chemical Society
Kinetic Modeling of LDL Lipid Peroxidation Scheme 1. Single-Phase Reaction Scheme of Azo Initiator-Induced “Conventional” LDL Lipid Peroxidationa
Chem. Res. Toxicol., Vol. 9, No. 6, 1996 955
29-34) and ubiquinol-10 (QH2) (18, 35-37) during LDL oxidation, yielding results consistent with previous experimental data (29).
Experimental Procedures
a The rate constants are denoted by k, and the subscripts g, iLH, perox, p, inh, and t denote generation, initiation via reaction with LH, peroxidation, propagation, inhibition, and termination, respectively. The azo initiator (2,2′-azobis(2,4-dimethylvaleronitrile) (AMVN) or AAPH; i.e., R-NdN-R, abbreviated in figures and schemes as R) produces alkyl radicals (R•) (not shown) to which O2 adds to generate an alkyl peroxyl radical (ROO•). The latter initiates lipid peroxidation by abstracting a hydrogen atom from LH, yielding a lipid alkyl radical (L•), to which O2 adds. The resulting lipid peroxyl radical (LOO•) propagates the chain reaction by reacting with another LH thereby re-forming L• and a LOOH. R-TOH suppresses lipid peroxidation by trapping LOO•, generating LOOH and R-TO•. Thus, in this model the vitamin functions solely as an antioxidant. The radical chain is terminated by the reaction of LOO• with R-TO• or with itself, leading to nonradical products (NRP) NRP1 and NRP2, respectively.
Scheme 2. Single-Phase Reaction Scheme of TMP of LDL Lipidsa
a R-TO•, formed by hydrogen abstraction from R-TOH by ROO•, is the peroxidation chain initiating and carrying species, as opposed to L• and LOO•, respectively, in Scheme 1. Termination proceeds via reactions of R-TO• with ROO• or LOO• (19). Thus, in this scheme, vitamin E functions as a pro- and an antioxidant. For abbreviations, see legend to Scheme 1; kiTOH denotes the rate of initiation via reaction with R-TOH.
instructive in highlighting the salient features of the particular systems. Here we describe the kinetic modeling of radical-initiated lipid hydroperoxide (LOOH) formation in LDL according to the conventional and TMP models, and evaluate our conclusions against the ones reached in a recent paper presenting simulations of LDL lipid peroxidation based on the conventional model (28). The differences between the two models are highlighted under low-versus-high radical flux conditions. Expanding the TMP model by incorporating phase separation revealed a ∼400-fold increase in the extent of reactions taking place within the lipoprotein particle, compared with that in the aqueous phase, and allowed modeling of the actions of the co-antioxidants ascorbate (AH-) (19,
Computer Simulations and Regression Analysis. The computer models were programmed in Mathematica (Wolfram Research Inc. Champaign, IL; version 2.2), and they consisted of five “units”: the first one specified the values of the rate constants to be used in the simulations; the second one defined the rates of each reaction according to principles of chemical kinetics (38); the third one specified the solution to the system in the form of integrated rate equations under initial conditions; the fourth one specified the variables (i.e., reactants) of the system; and the fifth unit specified the output of the reactants versus time as concentration profiles. The function “NDSolve” was used to find numerical solutions to the systems (Schemes 1-4) of simultaneous (coupled) differential equations. Initial conditions (i.e., concentrations at “t ) 0”) were given for every reactant in the system, so that the solutions were specified completely. NDSolve uses an adaptive step-length procedure to solve “stiff” differential equations. For some of the simulations shown in Figures 1A and 2A, a ratelimiting step was applied to the scission reaction of the azo initiator as well as pseudo-first-order reaction conditions with respect to the oxygen addition reactions to facilitate numerical evaluation (see Results). To ensure that the system of differential equations could be solved even with “non-smooth” solutions, the option MaxSteps[] was set to “Infinity”. It was also found necessary to specify the absolute error that was allowed in the solution to the systems of differential equations describing Schemes 1 and 2 (simulations shown in Figures 1A and 2A), in order to “track” solutions that closely approached zero (e.g., the concentration profile of R-TOH). This was performed using the options AccuracyGoal[] and WorkingPrecision[]. Typically, using a 12-digit precision in the calculations (AccuracyGoal ) 12) yielded sufficiently accurate results. Simulations were carried out on Macintosh FX and Quadra 605 computers. Evaluations of the reactant concentration profiles typically required kiLH × [LH]. 3
V. Zammit, R.T. Dean, and R. Stocker, unpublished.
962 Chem. Res. Toxicol., Vol. 9, No. 6, 1996
In addition to these oversimplifications, two findings reported in ref 28 can be taken as support for TMP, while interpreted as evidence against it, by the authors. Including R-TO• recombination in our simulations of TMP using values between 102 and 104 M-1 s-1 for the rate constant (as used in ref 28) did not affect appreciably the reactant concentration profiles presented in this work. First, the authors concluded that the rate constant for R-TO• recombination (which they included in their reaction scheme) needed to be
