with no sign of 2-OOH-3-(1,2-dioxane)-SQ’s peak, which is usually registered around 21.9 min 21 (Fig. S2, ESI), further confirming their non-reactivity with 1O2. The generation of 1O2 was investigated from the decomposition of EP in the presence and absence of SQ-OOHs in deuterated chloroform/methanol (1:1) over 1 h with spectra taken at 5 min intervals. In the absence of SQ-OOHs the formed 1O2 was constant from 0 to 1 h (Fig. S3A, ESI), in their presence however, 1O2 was not detected over the same period of time (Fig. S3B, ESI). To better elucidate the absence of 1O2 signal in the presence of SQ-OOHs, the changes occurring on EP were investigated using LC-UV, Q1 MS scan, and 1H NMR. Changes in EP’s LC-UV peaks in the presence and absence of SQ-OOHs gave similar patterns (Fig. S4, ESI), indicating the decomposition of EP under both conditions. Q1 scans of EP’s peaks (Fig. S5 to S8, ESI) show at 0 h the presence of both intact and decomposed EP, while they show mostly fully decomposed EP in the presence and absence of SQ-OOHs after incubation at 25°C from 1 to 18 h. Likewise, 1H NMR spectra interpretation of EP (Table S1 and Fig. S9 to S11, ESI) in the presence and absence of SQ-OOHs show evidence of EP’s decomposition in both conditions, further confirming the generation of 1O2 in both cases. The absence of 1O2 phosphorescence peaks in the presence of SQ-OOHs is not well understood and may be either somehow quenched, or due to the presence of trace amounts of radical decomposition species generated from SQ-OOHs at 25°C. The interaction of these radicals with EP would induce in parallel, a radical decomposition mechanism of EP, generating 3O2 instead of 1O2 (Fig. S12, ESI), making the quantity of 1O2 less and more difficult to detect. Nevertheless, we consider these results to be sufficient evidence to refute the involvement of 1O2 in the generation of SQ hydroperoxy cyclic peroxides from SQ-OOHs.
Under elevated temperatures, both in the presence and absence of the radical initiator MeO-AMVN (Fig. 2B), the generation of 2-OOH-3-(1,2-dioxane)-SQ was observed, exhibiting an initial increase reaching 5.30 ± 0.27 ng and 8.62 ± 0.56 ng respectively after 1 h. Subsequently, the levels gradually decreased and eventually disappeared after 4 h. The overall generated amount was lower in the case of the presence of the radical initiator (MeO-AMVN) (Fig. 3). These observations indicate that the thermal oxidation of SQ-OOHs is not the ideal condition for the generation of SQ hydroperoxy cyclic peroxides due to their instability and fast degradation under heat, nor is it solely instigated or facilitated by a radical attack since MeO-AMVN exhibited a decrease in the formed amount of 2-OOH-3-(1,2-dioxane)-SQ compared to its absence. The reason for the appearance of the compound under thermal oxidation is proposed to be that the reaction requires a certain amount of activation energy; however, under the same conditions, modifications of LOOHs in general can take place and may be more prone to destructive modifications. Breakdown products are more abundant as a result of the homolytic cleavage of hydroperoxides’ O-O bond which is more favored under elevated temperatures specifically 30–32, explaining the low generated quantity, the fast decrease and disappearance of 2-OOH-3-(1,2-dioxane)-SQ, as its and SQ-OOHs’ O-O bonds undergo cleavage, yielding mostly decomposition products. For instance, the cleavage of peroxide’s O-O bond to give alkoxyl and hydroxyl radicals is often associated with and measured under thermal oxidation 33. Under photooxidation and in the presence of 3O2, the generation of 2-OOH-3-(1,2-dioxane)-SQ was noticed to be the highest among all conditions and continued to increase up until 8 h reaching an amount of 24.07 ± 0.76 ng (Fig. 3). Further confirming that the oxidation mechanism leading to the formation of SQ hydroperoxy cyclic peroxides from SQ-OOHs is photochemically favored. Overall, we noticed that higher molecular weight secondary oxidation products are more likely to be formed under photooxidation with minor breakdown products, while the latter are the major observed secondary oxidation products under thermal oxidation. This further indicates that under the present conditions, thermal oxidation of SQ-OOHs has a higher tendency for “destructive modifications” while their photooxidation has more of “constructive modifications” pattern.
Both unoxidized SQ and SQ-OOHs’ exposure to ozone resulted mainly in breakdown products, 2-OOH-3-(1,2-dioxane)-SQ was not one of the major products in both cases (Fig. S13, ESI). Hence, ozone, which is in constant contact with human skin lipids, especially in polluted environments, has no role or effect on the mechanism generating 2-OOH-3-(1,2-dioxane)-SQ from SQ-OOHs or directly from SQ.
From the above set of comparative oxidative conditions, we could confirm that the serial cyclization of SQ-OOHs to give SQ hydroperoxy cyclic peroxides is ideally induced and generated by photooxidation in the presence of 3O2, making them the optimal conditions for this reaction.
Characterization of the intermediate radical species generated during the formation of SQ hydroperoxy cyclic peroxides
ESR spectroscopy was employed to identify and monitor the formation of radicals during the photooxidation of SQ-OOHs and the generation of SQ hydroperoxy cyclic peroxides, specifically, 2-OOH-3-(1,2-dioxane)-SQ. To identify the radical species formed during the reaction, reference radicals were used, and their spectra, including hyperfine splitting and coupling constants, along with the literature, were compared to those obtained from SQ-OOHs’ oxidation to determine the generated intermediate radical species. Spectra obtained from the UV irradiation of tBuOOH generated •OR and HO• radicals over 30 min (Fig. 4A1 and A2) based on the hyperfine splitting interpretation (Fig. S14, ESI), which were determined as DMPO-OH/OR adducts (Fig. 2C2). Its characteristic four peaks of approximately 1:2:2:1 intensities and the coupling constants aN=14.29 G and aßH≈14.21 G were also consistent with the DMPO-OH spectra reported in the literature 34–37. Spectra obtained from the UV irradiation of H2O2 in the presence of DMPO over 30 min (Fig. 4B1 and B2) generated majorly HOO• (adduct presented in Fig. 2C3) following the interpretation of the hyperfine splitting (Fig. S15, ESI), which was consistent with DMPO peroxyl radical adducts reported in the literature 38–40, with the coupling constants aN=15.75 G and aßH=22.41 G. And minorly HO• with the constants aN=14.23 G and aßH≈14.20 G.
O2•− spectrum presented in Fig. 4C (interpretation in Fig. S16, ESI), shows a DMPO-OO− adduct with a similar 12 splitting pattern as reported previously 41,42.
DMPO presented clear characteristic and distinguishable adduct-based hyperfine splitting patterns, depending on the type of the radical. However, this was hardly the case when POBN was used as the spin trap (adducts presented in Fig. 2D2-D4). The spectra shown in Fig. 5 did not exhibit significant differences among the POBN radical adducts that could help identify the radical species. Moreover, over a period of 13 h and 30 min, SQ-OOHs derived radicals quenched by POBN were detected between 0 and 4 h only (Fig. S17, ESI), while they were detected between 0 and 6 h in the case of DMPO with higher intensities. Consequently, DMPO was chosen as the primary spin trap for the identification and analysis of the radicals formed during the cyclization of SQ-OOHs to SQ hydroperoxy cyclic peroxides.
The oxidation of SQ-OOHs in the presence of DMPO provided significant information as shown in the side profile and 3D representations of the radicals’ turnover overnight (13 h and 30 min) in Fig. 6A1 and A2. Radicals were observed throughout the oxidation process, with the highest intensities between 0 h and 6 h. A distinct change in the pattern of the radicals was observed exclusively during the initial 12.8 min, to give similar constant spectra for the rest of the oxidation (Fig. 7). At 1.6 min, the registered spectrum (Fig. 7A) was compatible with DMPO-OO− reported in several previous studies 43–45. Its hyperfine pattern appeared to differ from the one obtained from the pure O2•− in Fig. 4, this discrepancy can be attributed to the solvent composition in the case of SQ-OOHs, which contained mainly hexane and isopropyl alcohol instead of DMSO. This is consistent with the previous reports that gave similar spectra when the solvent used contained alcohols 43,44. Following the initial 1.6 min of photooxidation of SQ-OOHs, the signal of O2•− radicals decreased but remained detectable, this can be partially due to its half-life (order of 10− 6 s) compared to that of ROO• (order of 17 s) 46. Notably, a prominent hyperfine structure emerged and continued to intensify over time, as observed in Fig. 7B to 7E. Additionally, another minor unidentified three peaks hyperfine structure of 1:1:1 intensities and the coupling constant: aN=15.48 G (Fig. 8) was registered. A similar observation was reported by Walger et al. in their investigation of the radicals generated from H2O2/CuII/phenanthroline system 47, where the hyperfine structure could also not be identified and was referred to as “triplet radical”. The major hyperfine structure was consistent with DMPO-OOH and DMPO-OOR (SQ-OO• radicals) with the coupling constants aN=15.20 G and aßH=22.65 G (Fig. 8), which are very close to the coupling constants of DMPO-OOH observed from the photolysis of H2O2 (aN=15.75 G and aßH=22.41 G) indicating the presence of both SQ-OO• and HOO• radicals. The slight differences can be explained by both the effect of the radical’s substitution and the presence of multiple overlapping radical signals in the case of SQ-OOHs photooxidation. Moreover, carbon radicals present similar hyperfine splitting structure with coupling constants that vary depending on the substitution of the radical 48,49, for this reason, the possibility of the presence of a carbon centered radical from SQ-OOHs during this photooxidation cannot be overruled.
Upon addition of SOD, the duration in which the radicals were observed to form in high intensities in its absence, appeared to have no signal at first or a very low signal (i.e., no radicals) (Fig. 6C) up until 4 h. This is apparent from the comparison of the 3D and side profile data (Fig. 6B1 and B2). SOD is a well-known O2•− scavenger, the elimination of radicals’ signals during the peak period of radical formation resulting from the oxidation of SQ-OOHs upon its addition, highly confirms the involvement of O2•− as a key first radical intermediate in the serial cyclization of SQ-OOHs to SQ hydroperoxy cyclic peroxides. The pattern of the radicals formed during SQ-OOHs photooxidation in the presence of SOD, which were generated gradually and that could be clearly seen after 4 h, presented one species compatible with the HOO• spectra (Fig. S18, ESI) (which can also be interpreted as ROO• and/or R•). Moreover, minor radicals (O2•−, and triplet radical) previously detected during the oxidation of SQ-OOHs were not observed in the presence of SOD throughout the oxidation. Hence, it can be inferred that the detected radical is solely HOO•, which is generated through the photolysis of H2O2, product of SOD’s O2•− quenching activity. Although O2•− has been previously hypothesized to be involved in lipid peroxidation by acting both in the initiation and termination steps in in vivo and in vitro studies 50–53, to the best of our
knowledge, there are no reports that demonstrated its implication in the formation of higher molecular weight secondary oxidation products from LOOHs, especially, in their serial cyclization. We, therefore, present the first evidence of the crucial involvement of O2•− in the formation of SQ hydroperoxy cyclic peroxides from SQ-OOHs. Furthermore, LC-MS/MS analysis of 2-OOH-3-(1,2-dioxane)-SQ generated from the photooxidation of SQ-OOHs in the presence of SOD (Fig. S19, ESI) showed no formation of 2-OOH-3-(1,2-dioxane)-SQ and a decline in the starting trace amount of the secondary oxidation product. This further confirms that SOD suppresses the generation of 2-OOH-3-(1,2-dioxane)-SQ and that O2•− plays a critical role in the formation of SQ hydroperoxy cyclic peroxides from SQ-OOHs.
Chemical calculations of the O-O’s overall negative electrostatic charge and the proton’s positive charge of each isomer’s hydroperoxide’s moiety as determined by Spartan 18 software are expressed in Table 1. The difference between the negative and positive electrostatic charges was found to be most pronounced in the isomers 6-OOH-SQ, 10-OOH-SQ and 2-OOH-SQ with the values: −0.296, − 0.292 and − 0.293 respectively. This indicates that tertiary SQ-OOHs exhibit a greater disparity between negative and positive electrostatic charges compared to the secondary isomers (which had the values of − 0.246, − 0.275 and − 0.286). This can be attributed to factors such as hyperconjugation, the inductive effect, and the stabilizing steric effect (reduced steric hindrance) which are more pronounced in the tertiary isomers. The enhanced inductive effect promotes the stabilization of the hydroperoxide-bearing quaternary carbon, as a result, the hydroperoxide moiety of tertiary SQ-OOHs exhibits increased reactivity. Heterolytic cleavage is characterized by a more common occurrence when there is a significant disparity between the negative and positive electrostatic charges at the two ends of a bond. The greater difference in electrostatic charges between the oxygen and hydrogen moieties in tertiary SQ-OOHs allows
Table 1
Chemical calculations of the difference between the negative electrostatic charges (-O-O-) and positive electrostatic charges (-H) of SQ-OOHs hydroperoxide moiety.
Electrostatic charges |
Isomer | 11-OOH-SQ | 10-OOH-SQ | 7-OOH-SQ | 6-OOH-SQ | 3-OOH-SQ | 2-OOH-SQ |
R-(O-O)-H | -0.612 | -0.652 | -0.64 | -0.659 | -0.59 | -0.65 |
R-O-O-(H) | 0.337 | 0.36 | 0.354 | 0.363 | 0.344 | 0.357 |
Charge difference | -0.275 | -0.292 | -0.286 | -0.296 | -0.246 | -0.293 |
for a higher likelihood of heterolytic cleavage of the O-H bond, compared to the secondary isomers (Illustration in Fig. S20, ESI), explaining why 6-OOH-SQ and 10-OOH-SQ are the primary targets of this photooxidation mechanism 21. The above mentioned criteria are all influential factors that contribute to the enhanced stability of tertiary peroxyl radicals. Consequently, tertiary hydroperoxides are more prone to decomposition via heterolytic cleavage than secondary hydroperoxides in a mixture, followed by the release of a photoelectron to produce a more stable peroxyl radical. These factors also affect the geometry of the different isomers 54, which may, in turn, affect the stability of the resulting ions and radicals, the reaction, and its rate. Notably, this reaction was observed to exhibit significant differences when conducted in different solvents. Specifically, when methanol was used as the solvent for the oxidation of total SQ-OOHs, of which 6-OOH-SQ and 10-OOH-SQ are the first targets, LC-UV analysis revealed no decrease in the peaks’ intensities of these two isomers, the peak corresponding to 2-OOH-3-(1,2-dioxane)-SQ was not detected. In contrast, when the oxidation was carried out in hexane, a noticeable decrease in the peaks’ intensities was observed for the two isomers, accompanied by the appearance of the peak corresponding to 2-OOH-3-(1,2-dioxane)-SQ (Fig. S21, ESI), the observed difference can be attributed to the fact that methanol, unlike hexane, functions as a proton donor, hindering cyclization by facilitating proton addition to the formed SQ-OO• radical. This further highlights the significance of the choice of the reactive conditions on its outcomes. The above O-H heterolytic cleavage route also supports the formation and detection of O2•− radicals by ESR as the first radical species formed from the photooxidation of SQ-OOHs. Moreover, Q1 analyses of DMPO adducts with radicals resulting from SQ-OOHs photooxidation and thermal oxidation (Fig. S22, ESI), confirm the presence of both DMPO-O-SQ/OH and DMPO-OO-SQ under thermal oxidation with high decomposition products, while it shows the detection of only DMPO-OO-SQ under photooxidation. Further confirming our proposed O-H cleavage under photooxidation and the statement made in the discussion of SQ-OOHs thermal oxidation in the previous section. Consequently, we suggest the hereinafter mechanism (Fig. 9). In the case of 6-OOH-SQ, following cleavage, the formed SQ-OO− further undergoes release of an electron upon exposure to photons under the described conditions, which is subsequently accepted by 3O2, generating O2•− and SQ-OO•. The latter undergoes cycloaddition on the adjacent C3 carbon to give a 1,2-dioxane ring and a carbon radical on C2. While O2•− reacts with H+ to give a peroxyl radical, which subsequently reacts with the C2 carbon radical to give a hydroperoxide, generating 2-OOH-3-(1,2-dioxane)-SQ. In the case of the remaining isomers, the same mechanism applies with serial cyclization from the firstly formed SQ-OO• to give multiple 1,2-dioxane rings and a hydroperoxide on C2 (as shown in Fig. 1). While prior investigations have proposed the initiation of the serial cyclization in the formation of lipid cyclic peroxides from LOOH via the generation of a peroxyl radical (LOO•) and a proton radical (H•) 8–16, the actual formation mechanisms of these radicals have not been substantiated or demonstrated. In the current study, we present compelling proof regarding the unforeseen participation of O2•− in the formation of these radicals, the serial cyclization of SQ-OOH, and the strongly probable heterolytic cleavage of the O-H bond as the initial step in this reaction under photocatalytic conditions.
In the present work, it was observed that despite the general misconception arising from the generalization of the values of bond dissociation energies (BDEs), which would indicate a higher likelihood of O-O homolytic cleavage, the cleavage of O-H bonds, both heterolytic and homolytic, is not inherently forbidden by any laws. In fact, based on the totalitarian principle 55, it is still quite probable, which is what was observed in our case study. Moreover, although heterolytic cleavage energies are in general considered to be higher than BDEs for the same type of bond, the accurate bond cleavage pathway may not always abide by the values provided in the literature, as BDEs and bond ionization energies (BIEs) are estimated calculative and experimental approximations under certain reactive conditions, and may not reflect the behavior of molecules in different experimental procedures. Changing one parameter, such as the solvent, which can introduce a distinct cage effect, has the potential to significantly influence the dynamics of the reactive molecules, resulting in unexpected patterns of bond cleavage, reaction outcomes, and rates 56–59. This highlights that the general applicability of BDEs and BIEs for a specific bond cannot be assumed ipso facto based on the values available in the literature, unless calculated under identical reactive conditions. This becomes particularly important when investigating unfamiliar reaction mechanisms; hence, these values should be regarded as circumstantial rather than definitive. Several studies have in fact pointed out the limitations and deficiencies associated with this prevailing misapprehension, as they demonstrated different bond cleavage values and reaction pathways based on the specific reaction’s conditions and substitutions of the functional group in question for a variety of compounds 33,60–63.