Effect of 2, 5-Substituents on the Stability of Cyclic Nitrone Superoxide Spin Adducts: A Density Functional Theory Approach

Abstract

To design efficient spin traps for superoxide radicals, interest in the elucidation of substituent effects on the stability of superoxide spin adducts has become a necessary priority. In the present study, five cyclic nitrone superoxide spin adducts, i.e. DMPO-OOH, M₃PO-OOH, EMPO-OOH, DEPMPO-OOH, and DEPDMPO-OOH, were chosen as model compounds to investigate the effect of 2,5-subsitituents on their stability, through structural analysis and decay thermodynamics using density functional theory (DFT) calculations. Analysis of the optimized geometries reveals that none of the previously proposed stabilizing factors, including intramolecular H-bonds, intramolecular nonbonding interactions, bulky steric protection, nor the C(2)–N(1) bond distance can be used to clearly explain the effect of 2,5-substituents on the stability of the spin adducts. Additionally the effect of the 2,5-substituents on the stability of the superoxide spin adducts cannot be simply clarified by Milliken charges on both atoms (nitroxyl nitrogen and nitroxyl oxygen). Subsequent study found that spin densities on the nitroxyl nitrogen and oxygen are well correlated with the half-life times of the spin adducts, and consequently are the proper parameters to characterize the effect of 2,5-substituents on their stability. Examination of the decomposition thermodynamics further supports the effect of the substituents on the persistence of cyclic nitrone superoxide spin adducts.

I. Introduction

Electron spin resonance (ESR) in combined with the spin trapping technique, as one of the useful methods to detect superoxide radical and other reactive oxygen species, has been extensively employed to probe and characterize the free radical processes in chemical and biological systems. To efficiently detect the free radicals in biological systems, the spin trap used should trap radicals rapidly, resulting in a long-lived radical spin adduct. The cyclic nitrone 5,5-dimethyl-1-pyrroline N-oxide (DMPO) is one of the most commonly used spin traps.¹ However, its extensive application in biological milieu has been restricted due to the low superoxide trapping rate (1.2 M⁻¹s⁻¹ at pH 7.4)² and the low stability of the corresponding superoxide spin adduct (t_1/2 ≈ 1.0 min)^{1, 2}. Therefore, the molecular structure of DMPO has been systematically modified and various cyclic nitrone analogues obtained. For example, introducing the substituents P(O)(OCH₂CH₃)₂ or COOCH₂CH₃ gives birth to alkoxyphosphoryl-nitrone 5-diethoxyphosphoryl-5-methyl-1-pyrroline N-oxide (DEPMPO)³ and alkoxycarbonyl-nitrone 5-ethoxycarbonyl-5-methyl-1-pyrroline N-oxide (EMPO),⁴ respectively. Both DEPMPO and EMPO exhibit a significant improvement in the spin-trapping rate for superoxide radical and the stability of their corresponding superoxide spin adducts,⁵ and both have been widely used in radical detection in chemistry and biology research.⁶

The persistence of the superoxide spin adduct is an even more important factor in the above-mentioned criterion (i.e. the superoxide spin-trapping rate and the stability of the superoxide adduct). Therefore, many efforts have been made toward elucidating the stabilizing role of the strong electron-withdrawing alkoxyphosphoryl and alkoxycarbonyl groups regarding the superoxide spin adducts. Tordo and Nohl⁷ experimentally suggested that the strong electron-withdrawing effect and the large steric hindrance of the alkoxyphosphoryl group or alkoxycarbonyl group were responsible for the stability of the linear and cyclic nitrone spin adducts. Vellamena’s theoretical data⁸ further revealed that the intramolecular H-bond played an important role in the stability of the superoxide and hydroxyl spin adducts. In analysis of the phosphoryl effect on the stability of linear nitrone superoxide spin adducts, we found that besides the aforementioned stabilizing factors, the intramolecular nonbonding interaction may be another contributing factor to the stabilization of the linear nitrone superoxide spin adducts.⁹ Taken together, the large steric hindrance, intramolecular H-bonding, and nonbonding interactions induced by the strong electron-withdrawing groups are all possible structural factors that contribute to the stabilization of superoxide spin adducts.

In vivo, the radical spin-trapping capability of nitrones is not only dependent on the static stabilizing effect on the resulting adducts, but is also associated with the spin adducts’ surrounding environments. Samuni and co-workers¹⁰ found that radical (^·CH₃, ^·CO2⁻, ^·OH, etc.) adducts of 2,5,5-trimethyl-1-pyrrole-N-oxide (M₃PO), obtained by replacing the β-H of DMPO with a methyl group, are far more stable and less susceptible to cellular-induced destruction than the corresponding DMPO adducts. However, the M₃PO superoxide spin adduct is too unstable to be directly detected by ESR at room temperature.¹⁰ To design a better spin trap for superoxide radical on the basis of M₃PO molecular skeleton, we previously synthesized its phosphoryl analogue, 5-(Diethoxybenzene phosphoryl) 2,5-dimethyl-1-pyrrole–N-oxide (DEPDMPO), through incorporation of one phosphoryl group at the C-5 position.¹¹ ESR results indicated that the half-life of the superoxide spin adduct DEPDMPO-OOH is 2.6 min, which is much longer than that of M₃PO-OOH. Comparative analysis of the half-lives of DMPO-OOH/M₃PO-OOH and DEPMPO-OOH/DEPDMP-OOH further indicates that the introduction of a methyl group at the C-2 position is not beneficial to the stability of DMPO-type superoxide spin adducts.

Consequently, in order to have a better understanding of the effects of having a 2-methyl group and 5-strong electron-withdrawing groups, briefly referred to as 2,5-substituents, on the persistence of the superoxide spin adducts and to reveal the correlation of such substitutions with the intrinsic structural features of the spin adducts, we herein comparatively analyze and discuss the geometric and electronic structures, as well as several possible decomposition pathways of five superoxide spin adducts DMPO-OOH, M₃PO-OOH, DMPO-OOH, EMPO-OOH, DEPMPO-OOH, and DEPDMPO-OOH. Our goal in this paper was accomplished by virtue of the density functional theory (DFT) calculation method CPCM-B3LYP/6-311+G(d,p)//B3LYP/6-31G(d).

II. Experimental Materials and Methods

A. Materials

According to the reported literature methods, DMPO,¹ M₃PO,¹ EMPO,⁴ DEPMPO³, and DEPDMPO¹¹ were synthesized in our laboratory. Xanthine oxidase (XOD), hypoxanthine (HX), nitrogen blue tetrazolium (NBT), four vinyl cyanide (TCNE), and Diethylenetriamine five acetic acid (DTPA) were purchased and used without further purification. PSII membranes obtained from market spinach were isolated by the method of Berthold et al. with the modification of Yruela et al.¹²

B. Decay Dynamics of Superoxide Spin Adducts

The light-PSII system was utilized to investigate the kinetics of decay for the superoxide spin adducts.¹³ Phosphate buffer solution (0.1M, pH 7.0) containing PSII (0.45 mg/ml), spin trap (50 mM), TCNE (1 mM), and DTPA (1 mM) was illuminated by He-Ne laser (25mW, 633 nm) for 2min. Once the light source was shut off, the decay of the superoxide spin adduct was followed by monitoring the decrease of the intensity of the first low field ESR line. Simulation of an approximate first-order decay process was carried out by Origion7.0 software. The peak intensity of the detected ESR signal is related to the actual concentration of the spin adduct by a scaling factor. The ESR spectra were recorded with the following parameters: frequency modulation, 100 kHz; modulation amplitude, 2 Gauss; scan time, 42 s; time constant, 0.16 s; microwave power, 12.8 mW.

C. Computational Methods

Considering the accuracy and convenience of density functional theory (DFT) ¹⁴, the B3LYP function on the basis set of 6-31G (d) was employed to carry out calculations. First, molecular geometries were optimized by semiempirical quantum chemical method AM1. Then, B3LYP/6-31G (d) was used for the full geometry optimization in gas phase. All stable structures predicted without virtual vibrational frequencies were gained. Zero point vibrational energy (ZPVE) and the vibrational contribution to the enthalpy and entropy were scaled by a factor of 0.9806.¹⁵ Single point energies were obtained using B3LYP/6-311+G (d, p). Therein, the solvent effects were also considered by employing the self-consistent reaction field (SCRF) method with conductor-like polarizable continuum model (CPCM).¹⁶ All quantum chemical calculations were performed with Gaussian 98.¹⁷

III. Results and Discussion

A. Decay Kinetics of Superoxide Spin Adducts

The stability of the superoxide spin adduct is one of the most crucial factors in designing a desirable superoxide spin trap, and the length of its half-life (t_1/2) is commonly used to evaluate its stability. However, the previously reported t_1/2’s regarding the five superoxide spin adducts studied in this work were not obtained in an identical experimental conditions.^{1b, 2c, 3, 4, 11,18} In order to comparatively investigate the substituent effects on their stability, as listed in the first row of Table 1, the t_1/2’s were re-examined under uniform conditions. As expected, M₃PO-OOH was too unstable to be detected by ESR spectroscopy. In the case of DEPDMPO-OOH, only the trans-isomer was observed due to the steric hindrance of the alkoxyphosphoryl and methyl groups, which forces the superoxide radical to attack the C=N bond only from the opposite side of the alkoxyphosphoryl group. Thus, the introduction of a methyl group at the C-2 position not only eliminates the presence of the cis-isomer, but also removes the β-H splitting from its ESR spectrum. Both factors lead to a simplified ESR spectrum with higher resolution, as shown in Figure 1. Remarkable differences in t_1/2 values among DEPMPO-OOH (14.8min), EMPO-OOH (8.6min), and DMPO-OOH (56s) indicates that the incorporation of an electron-withdrawing group at the C-5 position of cyclic nitrones increases the stability of the corresponding superoxide spin adduct. Considering that DEPMPO-OOH has a much longer half-life than EMPO-OOH, it seems that the stabilizing effect of the alkoxyphosphoryl group is stronger than that of the alkoxycarbonyl group. Comparative investigation revealed that the introduction of an electron-donating methyl group at the C-2 position of nitrones decreases the stability of the spin adducts. For example, Both DEPDMPO-OOH (2.1min) and M₃PO-OOH (undetectable) have smaller t_1/2’s in comparison with DEPMPO-OOH and DMPO-OOH, respectively.

TABLE 1.

Half-Life Times (t_1/2) and ESR Hyperfine SplittingConstants (A_N) for the Superoxide Spin Adducts in Phosphate Buffer (0.1 M, pH=7.0)

Spin adduct	DMPO-OOH	M3PO-OOH	EMPO-OOH	DEPMPO-OOH	DEPDMPO-OOH
t1/2(min)	56s; this work	-	8.0; this work	14.0; this work	2.1; this work
			8.0; ref. 4a	14.0; ref. 3a
	1.0; ref. 1b		4.8; ref. 4b	14.2; ref. 2c	2.6; ref. 11
	54s; ref.2c	-
	50s; ref.18		8.6; ref. 4c	14.8; ref. 3b
				13.0; ref. 18
AN (G)	14.3	-	13.3	13.2	13.4

FIGURE 1.

ES RSpectrum of DEPDMPO -OOH

B. Structural Stability of Superoxide Spin Adducts

(a) Analysis of the Optimized Geometry of Superoxide Spin Adducts

Intramolecular H-bonding, nonbonding interactions, as well as large steric hindrance induced by strong electron-withdrawing groups, as mentioned in the Introduction, are all possible structural factors which can stabilize linear nitrone superoxide spin adducts. Thus, in order to theoretically elucidate the effect of 2,5-substituents on the stability of the superoxide spin adducts, in the following sections we attempt to analyze these three stabilizing factors based on their optimized geometries. Two optimized configurations, i.e., cis- and trans-isomers, are assigned for DEPMPO-OOH and EMPO-OOH, indicating the position of the alkoxyphosphoryl or alkoxycarbonyl groups relative to the HOO group .

The calculated intramolecular H-bonds (IHB) between hydrogen of the OOH group and oxygen of the nitroxyl, alkoxyphosphoryl, and alkoxycarbonyl groups are in the range of 1.814 Å ~2.076 Å (Table 2). Herein, IHBs for DMPO-OOH, EMPO-OOH, and DEPMPO-OOH are similar to previously reported values.^{8b, 19} The IHB for cis-DEPMPO-OOH (1.814 Å) is shortest, indicating that the alkoxyphosphoryl group is the best among the studied groups to stabilize the superoxide spin adduct s through IHB. Further analysis shows that EMPO-OOH, however, has weaker IHBs (1.987 Å for trans-isomer and 2.000 Å for cis-isomer) than DMPO-OOH (1.959 Å), paradoxically inconsistent with their stability. This implies that intramolecular H-bonding is unable to explain the effect of the substituent at C-5 on the stability of the superoxide spin adducts. A similar result was also observed for the methyl at C-2. For instance, less stable M₃PO-OOH (1.933 Å) affords a stronger IHB than DMPO-OOH (1.959 Å) does. Therefore, although IHBs may play an important role in stabilizing the superoxide spin adducts, it cannot be used to fully interpret the substituent effect at 2,5-positions.

TABLE 2.

Intramolecular H-Bonds (Including O-H---O-N, O-H---O=P and O-H---O=C) Calculated by B3LYP/6-31G(d).

Spin adduct	H-bond distance (Å)
	This work	Ref. 8b
DMPO-OOH	1.96	1.99
M3PO-OOH	1.933
EMPO-OOH	1.99-trans;a2.00-cisb	1.99-trans;a2.00-cisb
DEPMPO-OOH	1.98-trans;a1.81-cisc	2.00-trans;a1.91-cisc
DEPDMPO-OOH	1.96

^aO-H---O-N.

^bO-H---O=C.

^cO-H---O=P.

Examination of intramolecular nonbonding interactions indicates that many interactions are present in the molecular structure of the superoxide spin adducts and their distances vary from 2.293 to 2.973 Å (Table S-1), which falls within the range of previously proposed nonbonding interactions (2.591~2.963 Å).²⁰ The alkoxyphosphorylated and alkoxycarbonylated spin adducts, such as DEPMPO-OOH and EMPO-OOH afford stronger intramolecular nonbonding interactions compared to DMPO-OOH, implying that the electron-withdrawing groups at the C-5 position increase the stability of the superoxide spin adducts through intramolecular nonbonding interactions. However, the introduction of a methyl group at the C-2 position enhances the intramolecular nonbonding interaction in DEPDMPO-OOH or M₃PO-OOH when compared to DEPMPO-OOH or DMPO-OOH, respectively. As a result, the intramolecular nonbonding interaction is obviously in contradiction with the destabilizing role of the methyl group for the spin adducts.

Inspection into the optimized geometry of DEPMPO-OOH shows that the alkoxyphosphoryl group is held in close proximity to the hydroperoxyl and nitroxyl groups. The steric proximity can be evidently demonstrated by comparison of some spatial angles around C(5). For example, the angle of N(1)-C(5)-P (106.6° for trans-isomer and 106.7° for cis-isomer) is smaller than the usual bond angle of sp³ hybrid orbital (109.5°). It can be thus determined that the large alkoxyphosphoryl group provides steric protection towards two vulnerable groups, -OOH and nitroxyl, by preventing or inhibiting the attack from other molecules (e.g., water) in the solution, and therefore stabilizesthe spin adduct. Similarly, this steric protection also occurs in the structures of the other four superoxide spin adducts. As shown in Table 3, the order for the corresponding spatial angle N(1)-C(5)-P (or C) is DEPDMPO-OOH (106.3°) < DEPMPO-OOH (106.6° for trans-isomer and 106.7° for cis-isomer) < EMPO-OOH (106.4° for trans-isomer and 108.9° for cis-isomer) < DMPO-OOH.(108.9°) ~ M₃PO-OOH (108.9°). It is therefore concluded that an alkoxyphosphoryl group substituted at the C-5 position has stronger protection than either alkoxycarbonyl or methyl groups, because of the larger steric volume, explaining the better stability of DEPMPO-OOH, compared to that of EMPO-OOH and DMPO-OOH, and also the better stability of DEPDMPO-OOH when compared to that of M₃PO-OOH. However, in comparison to the C-5 substitution, the methyl group substituted at the C-2 position has almost no effect on varying the spatial angles (N(1)-C(5)-P of DEPDMPO-OOH and N(1)-C(5)-C of M₃PO-OOH), which probably means that the C-2 substitution does not effect the stability via the means of steric protection .

TABLE 3.

Thermodynamic Parameters for Superoxide Spin Adducts Calculated by CPCM-B3LYP/6-311+G (d,p)//B3LYP/6-31G(d)

	DMPO-OOH	M3PO-OOH	EMPO-OOH	DEPMPO-OOH	DEPDMPO-OOH
Milliken Spin Density on N (1)	0.4992	0.5090	0.4836-trans 0.4707-cis	0.4461-trans 0.4499-cis	0.4886
Milliken Spin Density on O (1′)	0.4697	0.4658	0.4784-trans 0.5095-cis	0.4829-trans 0.5114-cis	0.4719
Milliken Charge on N (1)	0.1305	0.2476	0.1638-trans -0.0071-cis	0.2268-trans 0.0863-cis	0.3027
Milliken Charge on O (1′)	−0.2638	−0.2305	−0.2044-trans −0.1229-cis	−0.1443-trans −0.0834-cis	−0.1068
Bond Angles (°) of N(1) -C(5) -P (C)	108.9	108.9	106.4-trans 108.9-cis	106.6-trans 106.7-cis	106.3
Bond Lengths (Å) of N (1)-C (2)	1.481	1.492	1.481-trans 1.477-cis	1.482-trans 1.474-cis	1.493
Bond Lengths (Å) of O (2′)-O (3′)	1.455	1.455	1.454-trans 1.456-cis	1.454-trans 1.452-cis	1.454
Dihedral Angles (°) of O (1′)-N (1)-C (2) -H	73.6		−62.4-trans 59.1-cis	−63.3-trans 43.6-cis

Above analyses demonstrate that intramolecular H-bonds, intramolecular nonbonding interactions, as well as steric protection may be important factors that contribute to stabilizing the superoxide spin adducts, but they cannot be used to fully elucidate the effects of 2,5-subsituents on the stability of the spin adducts. This encouragesus to seek other better explanations.

The C(2)–N(1) bond has been reported to play a key role in the unimolecular decomposition process of nitrone radical adducts.^{8a, 21} Thus, the C(2)–N(1) bond distance was chosen as another candidate to represent the 2, 5-substituents’effect. As listed in Table 3, the order of C(2)–N(1) bond distances is as follows: DEPDMPO-OOH (1.493Å) > M3PO-OOH (1.492Å) > DMPO-OOH (1.481Å) > EMPO-OOH (1.477Å-cis) > DEPMPO-OOH (1.474Å-cis). Except for DEPDMPO-OOH, the order is the same as that for their stability profile. That is to say, introduction of an electron-withdrawing group at C-5 position shortens the C(2)–N(1) bond distance and then stabilizes superoxide spin adducts, but the addition of a methyl group at the C-2 position has an opposite effect.

(b) Analysis of the Optimized Electronic Structure of Superoxide Spin Adducts

Chemical reactivity is not only closely related to the geometric structures of the reactants, but is also dominated by their electronic structures. It may therefore be expected that the changes in the electronic structures of superoxide spin adducts will vary the stabilities according to the effect from 2,5 -substituents.

It has been recently reported that a relatively negative charge on the nitroxyl nitrogen of the hydroxyl spin adducts, caused by a strong electron-withdrawing group at C-5 position, can stabilize the C–N bond and significantly increases the electronegativity of nitrogen through an inductive effect.^8a,22 Our calculation (Table 3) reveals that either the charge on the nitroxyl nitrogen for EMPO-OOH (−0.0071-cis) or for DEPMPO-OOH (0.0863-cis) is more negative than that for DMPO-OOH (0.1305), consistent with the previous hypothesis,^8a,22 but the negative charge on the nitroxyl nitrogen for EMPO-OOH is obviously lower than that for DEPMPO-OOH, which cannot be interpreted according to their stability differences. The methyl substituent at the C-2 position affords more positive charge on the nitroxyl nitrogen, as demonstrated by the following order: DEPDMPO-OOH (0.3027) > M₃PO-OOH (0.2476) > DMPO-OOH (0.1305) > DEPMPO-OOH (0.0863-cis). However, the order is not identical to the stability sequence of all four adducts. As a result, the effect of the 2,5-substituents on the stability of the superoxide spin adducts cannot be simply explained by the charge on the nitroxyl nitrogen. Similarly, we found that the charge on nitroxyl oxygen (Table 3) has no correlation with the stability of the superoxide spin adducts, either.

It is well known that the delocalization of the unpaired electron between the nitrogen and oxygen atoms of nitroxyl radical (N–O^·) presents a resonance structure with N^·+–O⁻. Electron-withdrawing substituents at adjacent carbons increase spin density at the oxygen atom due to the increased role of the resonance structure with N–O^·, and as a result the oxygen has the lower charge.²³ There exists a lower spin density and higher charge at the nitrogen atom when a strong electron-withdrawing group bind at C-5 position. Considering the stabilizing effect of an electron-withdrawing substituent, such as an alkoxyphosphoryl or alkoxycarbonyl group, it is proposed that a relatively low spin density on the nitrogen atom probably stabilizes the superoxide spin adduct. The order of spin densities on the nitroxyl nitrogen (summarized in Table 3) can be obtained as follows: M₃PO-OOH (0.5090) > DMPO-OOH (0.4992) > DEPDMPO-OOH (0.4886) > EMPO-OOH (0.4836 for trans-isomer; 0.4707 for cis-isomer) > DEPMPO-OOH (0.4461 for trans-isomer; 0.4499 for cis-isomer). It is worth noting that the order is nearly identical to the stability sequence of all five spin adducts, which is evidence supporting the proposal about the stabilizing effect of spin density on the nitroxyl nitrogen. A quantitative linear correlation between spin density on the nitrogen and the length of the half-life for the superoxide spin adducts is shown in Figure 2a (r = −0.9920). Therefore, it is concluded that the more stable the superoxide spin adduct, the lower the spin density on the nitroxyl nitrogen. Comparatively, the spin density on nitroxyl oxygen has a similarly positive correlation with the stability of the spin adducts (Figure 2b, r = 0.9253). Consequently, both spin densities on nitroxyl nitrogen and nitroxyl oxygen are good parameters that can be used to estimate the stability of the superoxide spin adducts.

FIGURE 2.

Plots of Half-Life Times (t_1/2) of Superoxide Spin Adducts Versus Spin Density on Nitroxyl Nitrogen (a) or Nitroxyl Oxygen (b) and Versus Experimental Hyperfine Splitting Constant of Nitroxyl Nitrogen (A_N) (c)

It has been reported that the isotropic ESR hyperfine splitting constant (hfsc) is proportional to the unpaired electron density at the nucleus (Fermi contact terms) with an equation of A_N= (8π/3h)g_Nβ_Ng_eβ_eρ(0), where ρ(0) = |ψ(0)|².²⁴ In actuallity, only the s-type orbital has nonzero electron density at the nucleus; all other-type orbitals (p-orbital, d-orbital and so on) have at least one node passing through the nucleus (zero density). Considering that the isotropic hyperfine splitting constant is proportional to the partial unpaired electron density (s-orbital), we may reasonably deduce that the half-lifes (t_1/2) of the spin adducts would be negatively correlated with their hfs constants (A_N). This speculation is demonstrated by the plot of experimental A_N with t_1/2 of the adducts. As shown in Figure 2c, A_N exhibits an exponential correlation with t_1/2 of the spin adducts. This nonlinearity probably arises from the fact that A_N is only proportional to the unpaired electron density of s-orbitals on the nitrogen atom.

C. Decay Thermodynamics of Superoxide Spin Adducts

Although an exhaustive empirical elucidation has been made of the decay mechanisms for cyclic nitrone radical adducts, especially superoxide spin adducts, ^{2c, 25} its theoretical study has not received much attention. The interpretation of these decay mechanisms is valuable because it can be applied to the future development of efficient spin traps. Until recently only one study using DFT theory has shed light on the thermodynamics of decay for DMPO-OOH and DEPMPO-OOH.^21a To gain further insight into the effect of 2,5-substituents on the stability of cyclic nitrone superoxide spin adducts and to design some new nitrones with better spin trapping properties, we therefore conducted DFT calculations for the possible decay thermodynamics of these five superoxide spin adducts.

C–N bond cleavage, as mentioned above, is generally involved in the unimolecular decomposition reaction of nitrone superoxide spin adducts. ²¹ Moreover, a few reports have revealed that C–H^β bond cleavage also occurs, both in the unimolecular decomposition of DMPO-OOH²⁶ or PBN-OOH²⁷, and in the bimolecular decay route for DEPMPO-OOH.²⁸ Meanwhile, the unimolecular reduction of nitroxide to the hydroxylamine should be taken into account when both an electron and a H⁺ donor are present in the solution.^9,23 This type of reduction is a favorable route especially in biological systems.²⁹ As a result, four possible decay routes, including three unimolecular pathways and one bimolecular pathway, have been analyzed separately, as illustrated in Scheme 1. The calculated varieties of free energy (ΔG) including the effects of solvation, using the conductor-like polarizable continuum model (CPCM) are listed in Table 4. The schematic diagram of energy levels for ΔG in the rate-limiting steps of these decay routes are shown in Figure 3. The ΔG values described in the discussion are based on the most stable conformations (cis-isomers) for all five spin adducts. In addition, the solvent contribution to ΔG is listed in Table S-2.

SCHEME 1.

Various Possible Decomposition Pathways for Superoxide Spin Adducts

TABLE 4.

Reaction Free Energies (kcal/mol) of Various Decomposition Pathways for Superoxide Spin Adducts at the CPCM-B3LYP/6-311+G (d,p)//B3LYP/6-31G(d) Level

Reaction Scheme	DMPO-OOH	M3PO-OOH	EMPO-OOH	DEPMPO-OOH	DEPDMPO-OOH
Mechanism A
Step 1	−2.77	−10.31	−3.81-trans	−3.10-trans	−12.15
			2.41-cis	0.85-cis
Step 2	−1.29	−4.85	−1.97-trans	−1.33-trans	−2.89
			−3.30-cis	−3.39-cis
Step 3	−4.05	−15.16	−5.78-trans	−3.88-trans	−15.04
			−0.89-cis	−2.54-cis
Mechanism B
Step 4	−86.25		−83.55-trans	−84.65-trans
			−83.49-cis	−82.76-cis
Mechanism C
Step 5	11.93	10.08	12.03-trans	9.41-trans	9.99
			9.45-cis	15.72-cis
Step 6	−128.37	−125.37	−129.72-trans	−127.33-trans	−128.31
			−124.33-cis	−133.29-cis
Mechanism D
Step 7	−39.12		−9.79-trans	−11.16-trans
			−9.68-cis	−7.37-cis

FIGURE 3.

Energy Level for Reaction Free Energies (kcal/mol) in Key Steps of Possible Decomposition Pathways (Mechanisms A, B, C and D)

In the unimolecular decay process involving C–N bond cleavage (Mechanism A), superoxide spin adducts decompose through two steps.^9,21a,21c The hemolytic cleavage of the hydroperoxyl O–O bond (step 1) produces a diradical intermediate 2 (in singlet) and a hydroxyl radical, and then the diradical 2 undergoes a C–N bond cleavage to yield a nitrosoaldehyde 3 (step 2). ^21a As usually observed during the decay of some superoxide spin adducts, the HO^· generated may then be trapped by the original nitrone to form a hydroxyl spin adduct.³⁰ Structural analysis of the superoxide spin adducts finds that the O–O bond lengths for all of the adducts are ~1.45Å, which is similar to the previously calculated values of cyclic nitrone superoxide spin adducts.^8b This implies that the likelihood of hydroxyl radical production from the O–O bond is the same for all of these superoxide spin adducts. Comparatively, as discussed above, the C(2)–N(1) bond distance varies dramatically and is simultaneously cleaved when the O–O bond is ruptured in step 1 on the basis of the structural analysis for diradical 2 (Figure S-1). Accordingly, the C(2)–N(1) bond may play a key role in mechanism A, consistent with the previous proposal. ^{8a, 21} This implicitly suggests that step 1 is a rate-limiting step in Mechanism A. This result is further ascertained by the fact that the diradical 2 and the final product, nitrosoaldehyde 3, share the same structure except for a difference in conformation. Thus, mechanism A was modified: the hemolytic cleavages of the hydroperoxyl O-O bond and C–N bond produce a diradical intermediate 2 (in singlet) and a hydroxyl radical (step 1), and then the diradical 2 undergoes a configuration adjustment to yield a nitrosoaldehyde 3 (step 2). The order of ΔG₁ is as follows: EMPO-OOH (2.41kcal/mol) > DEPMPO-OOH (0.85kcal/mol) > DMPO-OOH (−2.77kcal/mol) ≫ M₃PO-OOH (−10.31kcal/mol) > DEDPMPO-OOH (−12.15kcal/mol). Based on the order and the values of ΔG₁ (shown in Table 4 and Figure 3), we can roughly speculate that an electron-withdrawing group at the C-5 position of a cyclic nitrone stabilizes its corresponding superoxide spin adduct by decreasing the reaction tendency described in mechanism A, where as the methyl group at the C-2 position plays an opposite role.

The unimolecular decomposition process via the C–H^β bond cleavage (mechanism B), possibly induced by HO⁻ or H₂O, yields a nitroxyl-ketone 4 through the elimination of a H₂O molecule (step 4).^21a The decomposition only occurs for DMPO-OOH, EMPO-OOH, or DEPMPO-OOH, bearing a β-H at C-2 position. All calculated ΔG_4,S are less than −80.00 kcal/mol, significantly lower than ΔG_1,S and ΔG_3,S in mechanism A, which means that mechanism B is thermodynamically more feasible than the decomposition process via C–N cleavage. In Mechanism B, the ΔG₄ is DMPO-OOH < EMPO-OOH < DEPMPO-OOH, which is identical with the observed decomposition tendency of the superoxide spin adducts. In other words, an electron-withdrawing group at the C-5 position can lessen the tendency to decay via C–H^β cleavage, and therefore stabilize the superoxide spin adducts. It is shown that ease of the C–H^β bond cleavage for the HO^· adduct was dependent on the conformation of the β-H relative to the singly occupied orbital on the nitroxyl nitrogen.^8a Only a low level of activation energy for the C–H^β cleavage is required when the singly occupied orbital on the nitroxyl nitrogen is in the same plane as the H^β atom to be abstracted. At this point, the dihedral angle of O(1′)-N(1)-C(2)-H is close to 90°. Structural analysis shows that the more stable EMPO-OOH and DEPMPO-OOH have a smaller dihedral ′O(1′)-N(1)-C(2)-H (59.1° for cis-EMPO-OOH; 43.6° for cis-DEPMPO-OOH) than that of DMPO (73.6°). Correspondingly, an electron-withdrawing group at C-5 position decreases the likelihood of the decayvia C–H^β bond cleavage.

In the unimolelcular reduction process (mechanism C), formation of a H-bond between an O atom from the nitroxyl group and a H⁺ from H₃O⁺ (step 5) is a prerequisite, and then, the produced intermediate 5 accepts one electron and gives rise to hydroxylamine 6 (step 6). Although step 5 is endoergic (9.45~15.72 kcal/mol), the subsequent step 6 has a high exothermicity with ΔG₆ values of −124.33~−133.29 kcal/mol. A similar reduction mechanism has been proposed to possibly occur in the decay of the linear nitrone superoxide spin adducts, and the protonation process is a rate-limiting step.⁹ In step 5 (a protonation process), however, the order of ΔG₅ is EMPO-OOH < DEPDMPO-OOH~M₃PO-OOH < DMPO-OOH < DEPMPO-OOH (as shown in Table 4 and Figure 3), largely inconsistent with their stability sequence. This result indicates that mechanism C probably is not the proper decay pathway that is responsible for elucidating the 2,5-substituents’ effect on the stability of cyclic nitrone superoxide spin adducts.

In the bimolecular decay pathway (Mechanism D), the products were diamagnetic hydroxylamine 6 and compound 7. This is similar to the step 4 reaction in Mechanism B, where the β-H abstraction reaction (step 7) takes place only in the decomposition processes of DMPO-OOH, EMPO-OOH, and EDPMPO-OOH and, more interestingly, the ease of the β-H abstraction by another superoxide spin adduct molecule was postulated to be similarly dependent on the dihedral ′O(1′)-N(1)-C(2)-H.^8a Therefore, we can reasonably estimate the bimolecular decay tendency according to the dihedral angle (DMPO-OOH > EMPO-OOH > DEPMPO-OOH, see discussion for mechanism B). A comparison of ΔG₇ values calculated for mechanism D shows that the order of ΔG₇ is as follows: DEPMPO-OOH (−7.37kcal/mol) > EMPO-OOH (−9.68 kcal/mol) > DMPO-OOH (−39.12 kcal/mol), which indicates that, in the bimolecular decomposition process, the introduction of a strong electron-withdrawing group at the C-5 position thermodynamically stabilizes the superoxide spin adduct.

IV. Conclusions

Five cyclic nitrones superoxide spin adducts were utilized to investigate the effects of 2,5-subsitituents on the stability of the superoxide spin adducts by employing DFT calculations. Analysis of their geometric structures indicates that the previously suggested three factors, including intramolecular H-bonds, intramolecular nonbonding interactions, and steric protection, may be important stabilizing factors for the superoxide spin adducts, but they are unable to fully predict the effect of 2,5-substituents on the stability. Further investigations on the stabilizing factors for the superoxide spin adducts indicate that neither the C(2)–N(1) bond distance, nor the charges on the nitroxyl nitrogen and the nitroxyl oxygen, are good parameters that can be used to explain the effects of 2,5-substituents.

Nevertheless, an inspection of the spin densities on nitroxyl nitrogen and nitroxyl oxygen reveals that both of their spin densities are linearly correlated with the stability of the superoxide spin adducts and thus can be used as proper parameters to interpret and predict the effects of 2,5-substituents. Additional experimental analysis on the correlation between the hyperfine splitting constant of nitroxyl nitrogen (A_N) and the half-lifes (t_1/2) of the spin adduct strongly support the hypothesis regarding the spin densities. On the other hand, thermodynamic calculations of the decay pathways, including three possible unimolecular decomposition processes and one possible bimolecular decomposition process, demonstrate that all the decay pathways, except for the unimolecular reduction of nitroxide to hydroxylamine, can be used to explain the effects of 2,5-substituentson spin adduct stability.

CHART 1.

Molecular Structures of Superoxide Spin Adducts