|Home | About | Journals | Submit | Contact Us | Français|
Singlet oxygen attack on cholesterol (Ch), a prominent monounsaturated lipid of mammalian cell plasma membranes, gives rise to three hydroperoxide (ChOOH) isomers, 5α-OOH, 6α-OOH and 6β-OOH, the latter two in lower yield than 5α-OOH, and 6α-OOH in lowest yield. A third possible positional isomer, 7α-OOH and 7β-OOH, is produced by free radical attack. In the presence of iron and ascorbate (Fe/AH), 5α-OOH or 6β-OOH in phosphatidylcholine/Ch/ChOOH (20:15:1 by mol) liposomes was reduced to its corresponding alcohol, the rate constant being approximately the same for both ChOOHs. Using [14C]Ch as an in situ probe, we found that liposomal 5α-OOH readily set off free radical-mediated (chain) peroxidation reactions when exposed to Fe/AH, whereas 6β-OOH under the same conditions did not. Moreover, liposomal 5α-OOH triggered robust chain peroxidation in [14C]Ch-labeled L1210 cells, leading to cell death, whereas 6β-OOH was essentially inert in this regard. Thus, 5α-OOH and 6β-OOH undergo iron-catalyzed reductive turnover, but only the former can provoke toxic membrane damage. These novel findings have important implications for UVA-induced photodamage in Ch-rich tissues like skin and eye, where1O2 often plays a major role.
At least 80 % of the cholesterol (Ch) of most mature mammalian cells is located in the plasma membrane, where it comprises ~45 mol % of the total lipid (1,2). Like other unsaturated lipids, Ch is subject to membrane-damaging oxidative modification under conditions of oxidative stress, including photodynamic stress (2,3). Among the many different products/intermediates generated by Ch (photo)oxidation, hydroperoxide species (ChOOHs) are of great interest not only because of their importance as mechanistic indicators, but also because of their diverse fates, e.g. damage-expanding one-electron reduction catalyzed by iron, damage-attenuating two-electron reduction catalyzed by selenoperoxidase GPx4, and delocalized redox signaling via intermembrane translocation (3-6). In singlet oxygen (1O2)-mediated Ch photooxidation (Type II chemistry), three primary ChOOHs are generated, 3β-hydroxy-5α-cholest-6-ene-5-hydroperoxide (5α-OOH), 3β-hydroxycholest-4-ene-6α-hydroperoxide (6α-OOH), and 3β-hydroxycholest-4-ene-6β-hydroperoxide (6β-OOH), each of which arises via ene-addition of 1O2 (7,8). The observed accumulation rates of these hydroperoxides during photooxidation of Ch in solution, liposomal membranes, or cells varies as follows: 5α-OOH 6β-OOH > 6α-OOH (5,8) An earlier study provided rational explanations for this trend based on steric factors in the Ch A/B rings (5). In free radical-mediated Ch photooxidation (Type I chemistry), as in ordinary autoxidation, the most prominent ChOOHs formed are 3β-hydroxycholest-5-ene-7α-hydroperoxide (7α-OOH) and 3β-hydroxycholest-5-ene-7β-hydroperoxide (7β-OOH) (9). These species, along with end-products such as 5,6-epoxide, 7-ketone, and the diols 7α-OH and 7β-OH, also appear when 1O2-derived 5α-OOH undergoes free radical chain-initiating one-electron reduction in membranes (4,5). When comparing the ability of liposomal 5α-OOH and 6β-OOH to trigger such reactions upon exposure to a lipophilic iron chelate (Fe(HQ)3) and ascorbate (AH-), we found that both peroxides underwent reductive decay at essentially the same rate (5). In the process, 5α-OOH induced robust chain peroxidation of other liposomal lipids. Surprisingly, however, 6β-OOH produced little, if any, such effect. Consistently, 5α-OOH triggered lethal chain peroxidation of plasma membrane lipids in L1210 cells, whereas 6β-OOH did not. These novel findings, along with possible explanations for the observed differences, are reported for the first time in this paper.
Unlabeled Ch, Chelex-100, 8-hydroxyquinoline, ascorbic acid, RPMI 1640 growth medium, and fetal bovine serum were obtained from Sigma Chemical Co. (St. Louis, MO). Amersham Biosciences (Arlington Heights, IL) supplied the [4-14C]Ch (~55 mCi/mmol in toluene), referred to as [14C]Ch. Before incorporation into liposomes or cells, the [14C]Ch was separated from any preexisting oxidation products by normal-phase HPLC, as described (10). Avanti Polar Lipids (Alabaster, AL) supplied the 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC). HPLC-grade organic solvents were from Mallinkrodt (Paris, KY). The primary ChOOHs 5α-OOH and 6β-OOH were prepared by aluminum phthalocyanine disulfonate-sensitized photooxidation of Ch in pyridine, followed by reverse-phase HPLC separation of the two isomers (11). After confirmation of identity by NMR (12), a stock solution of each ChOOH in isopropanol was analyzed for peroxide concentration by iodometric assay (13) and stored at -20 °C. We decided not to isolate and test 6α-OOH in study because its photooxidation yield is considerably lower than that of 6β-OOH (5). 14C-labeled 7α-OOH, 7β-OOH, 7α-OH, and 7β-OH, used as standards for HPTLC-PI-based ChOX analyses, were prepared as described previously (11,14); 5,6-epoxide, also used as a standard, was obtained from Steraloids Inc. (Wilton, NH). The following stock solutions were used for inducing ChOOH one-electron reduction: 1.0 mM ferric-8-hydroxyquinoline (Fe(HQ)3) (15), and 100 mM ascorbic acid in Chelex-treated PBS, prepared immediately before an experiment.
Large (100 nm) unilamellar vesicles (LUVs) were fabricated by an extrusion process, using an apparatus from Lipex Biomembranes (Vancouver, BC) and 0.1 μm polycarbonate filters (5, 10). The typical composition of a liposome preparation for assessing Fe(HQ)3/AH--induced ChOOH reductive decay and [14C]ChOX accumulation was as follows: 1.0 mM DMPC or POPC, 0.75 mM (~0.5 μCi/ml) [14C]Ch, and 0.05 mM 5α-OOH or 6β-OOH in bulk phase PBS. Before use the PBS was treated with Chelex-100 to deplete redox metal ions that might catalyze premature ChOOH loss (10). Other details were as described previously (5,10). The LUVs were stored under argon at 4 °C and used experimentally within 48 h.
Murine L1210 leukemia cells, obtained from the American Type Culture Collection (Rockville, MD), were grown in RPMI 1640 medium supplemented with 10% serum, growth factors, and antibiotics, using standard culture conditions (14). Cells in serum-free medium were labeled with [14C]Ch by spontaneous transfer from liposomes, using a published procedure (15). The [14C]Ch delivered by this approach is located predominantly in the plasma membrane. All experiments were carried out on exponentially growing cells.
PC/[14C]Ch/ChOOH (20:15:1 by mol) LUVs in PBS, where PC is DMPC or POPC and ChOOH is 5α-OOH or 6β-OOH, were incubated at 37 °C in the presence of 1.0 μM Fe(HQ)3 and 1.0 mM AH- (added last). Controls without Fe(HQ)3 or AH- were incubated alongside. At various time points, samples were removed, extracted with chloroform/methanol (2:1 by vol) and recovered lipid fractions were analyzed for remaining ChOOH by high-performance liquid chromatography with mercury cathode electrochemical detection [HPLC-EC(Hg)], using previously specified conditions (5,15).
Lipid extracts from Fe(HQ)3/AH--treated PC/[14C]Ch/ChOOH LUVs or [14C]Ch-labeled L1210 cells exposed to ChOOH-containing LUVs were analyzed for [14C]ChOX species (7α/β-OOH, 7α/β-OH, 5,6-epoxide) by means of high performance thin layer chromatography with phosphorimaging radiodetection (HPTLC-PI). Well-separated Ch, the loss of which to ChOX was usually negligible, was used as an internal loading standard. Additional details about methodology and evaluation of [14C]ChOX yields were as described previously (15).
L1210 cells (~7.5 × 105/ml in 1% serum-containing RPMI medium) were treated with POPC/Ch/5α-OOH or POPC/Ch/6β-OOH LUVs in a 96-well tissue culture plate such that the starting concentration of ChOOH in bulk suspension was 50 μM. All mixtures were standardized to the same concentration of total LUV lipid to normalize any background toxicity due to lipid alone. After LUV addition, the cells were returned to the incubator for 24 h and then assessed for survival by clonal assay (14).
In initial experiments we compared liposomal 5α-OOH and 6β-OOH for susceptibility to one-electron reduction and ability to set off chain lipid peroxidation reactions. As shown in Fig. 1A, 5α-OOH and 6β-OOH, each at ~2.8 mol % in separate POPC/[14C]Ch/ChOOH LUVs, were reduced at the same apparent first order rate (k ~0.087 min-1) when incubated with Fe(HQ)3 and AH-. No reaction occurred when iron was omitted (Fig. 1A), in agreement with earlier evidence that iron plays an essential catalytic role in shuttling an electron from AH- to the hydroperoxide (15). As shown previously for 5α-OOH and 7α-OOH (15), this one-electron step is rate-limiting in overall reduction of ChOOH to ChOH (diol), its rate increasing proportionately with iron concentration. When incubated with Fe(HQ)3/AH-, 5α-OOH and 6β-OOH in separate DMPC/[14C]ChOOH LUVs (where non-oxidizable DMPC was substituted for oxidizable POPC) once again decayed at the same rate (k ~0.05 min-1), as shown in Fig. 1B. The reason for the approximately 40% lower rate constant in the DMPC-containing vs. POPC-containing LUVs is not clear, but one possibility is that it reflects a fluidity difference which decreases access of AH- to membrane-bound ferric iron or ChOOH to ferrous iron.
In addition to tracking ChOOH decay in Fe(HQ)3/AH--treated PC/[14C]Ch/ChOOH liposomes, we tracked [14C]ChOX formation as a measure of ChOOH chain-initiating potency. As shown by the HPTLC-PI results in Fig. 2A, reductive activation of POPC/[14C]Ch/5α-OOH LUVs resulted in progressive accumulation of [14C]ChOX species, including (in order of increasing mobility) 7α-OH, 7β-OH, 5,6-epoxide, 7α-OOH, and 7β-OOH. No significant formation of these species was observed in a control that contained AH-, but not Fe(HQ)3 (results not shown), consistent with the non-decay of control 5α-OOH seen in Fig. 1A. In striking contrast to 5α-OOH, 6β-OOH in POPC/[14C]Ch/6β-OOH LUVs produced few, if any, net ChOX during incubation with Fe(HQ)3/AH- (Fig. 2A). The trace amount of 5,6-epoxide (band 3) was present from the outset and remained constant. The concentration of total ChOX in bulk LUV suspension increased progressively with reaction time in the 5α-OOH system, reaching ~25 μM by 30 min, but it remained at <1 μM throughout in the 6β-OOH system (Fig. 2B).
We also examined ChOX buildup in Fe(HQ)3/AH-- activated DMPC/[14C]Ch/ChOOH liposomes, where Ch was the only oxidizable lipid. As shown in Fig. 3, the concentration of total net ChOX in DMPC/[14C]Ch/5α-OOH LUVs increased rapidly with incubation time out to about 10 min and then more slowly, reaching ~22 μM by 30 min. However, the ChOX yield in DMPC/[14C]Ch/6β-OOH LUVs was barely detectable throughout, i.e. <1 μM. In both types of liposome, therefore, 5α-OOH and 6β-OOH underwent one-electron reduction at the same rate (Fig. 1 A, B), only 5α-OOH was able to initiate free radical-mediated lipid peroxidation (Figs. 2 and and3).3). The fact that POPC is oxidizable, but DMPC is not, probably accounts the greater initial rate of 5α-OOH-induced ChOX accumulation in the Fig. 3 experiment compared with the Fig. 2. In the latter case, POPC would have competed with [14C]Ch for a chain-initiating oxyl radical, whereas in the former case no phospholipid competition was possible, permitting faster early ChOX buildup. The later stage slowdown in 5α-OOH-induced ChOX (Fig. 3) is attributed to increasing termination due to combination of Ch-derived radicals, events that would be less frequent in the POPC system (Fig. 2).
We asked whether the contrasting effects of 5α-OOH and 6β-OOH one-electron turnover in liposomal membranes would also be manifested at the cellular level. To examine this, we exposed [14C]Ch-labeled L1210 cells to POPC/Ch/5α-OOH or POPC/Ch/6β-OOH LUVs, using initial ChOOH concentrations of 20 μM and 50 μM, and assessed [14C]ChOX accumulation and cell survival after an incubation period of 24 h. As shown in Fig. 4, a substantial yield of total ChOX was generated by 5α-OOH, 20 μM and 50 μM giving ~6 μmol/106cells and ~11 μmol/106cells, respectively. In striking contrast, 6β-OOH, even at the higher concentration, produced no more than ~0.5 μmol/106cells. Using a clonogenicity assay, we found that cell survival after 24 h exposure to 50 μM LUV ChOOH was as follows: 2 ± 1% for 5α-OOH; 95 ± 2% for 6β-OOH; means ± SE (n=3). Clearly, therefore, 5α-OOH was much more cytotoxic than 6β-OOH, and this correlated with 5α-OOH's ability to induce vigorous chain peroxidation and 6β-OOH's inability to do so.
Most lipid hydroperoxides (LOOHs) in membrane environments, including 5α-OOH and 7α/β-OOH, are susceptible to iron-catalyzed one-electron reduction, and this can exacerbate membrane damage by triggering chain lipid peroxidation (3,4,15). In the case of 5α-OOH, the effects of a non-radical primary reaction (1O2 addition to Ch) are intensified by secondary free radical reactions, whereas with 7α/β-OOH, free radical reactions are propagated. We found unexpectedly in the present study that another 1O2 adduct, 6β-OOH, though reduced by Fe(HQ)3/AH- at the same rate as 5α-OOH in liposomal membranes, lacked the latter's ability to set off damaging chain peroxidation cascades. In consistent fashion, liposomal 6β-OOH was found to be minimally prooxidant and cytotoxic toward L1210 cells compared with liposomal 5α-OOH. In an earlier study with oxidizable liposomes prepared with the same amount of either 5α-OOH or 7α-OOH, we showed that these ChOOHs decayed at the same initial rate during Fe(HQ)3/AH- treatment and that the rate of [14C]ChOX accumulation was also the same, indicating that 5α-OOH and 7α-OOH have the same chain-initiating potency (15). Both of these rates were dependent on Fe(HQ)3 concentration, consistent with a two one-electron reduction steps, the first (rate-limiting) step being ChOOH reduction to an oxyl radical (ChO·) by Fe2+, and the second step being reduction of ChO· to the alcohol (ChOH) by an oxidizable lipid. Ascorbate reduces Fe3+ to Fe2+ for the first step and competes with a lipid reductant in the second step. In another study (5), we showed that preexisting 5α-OOH, 6α-OOH, and 6β-OOH in photoperoxidized liposomes all decayed at the same rate during incubation with Fe(HQ)3/AH-, suggesting equal first-step reducibility. Although [14C]ChOX buildup was not determined in this work (5), 6β-OOH was shown to be much less cytotoxic toward L1210 cells than 5α-OOH; e.g. at 100 μM ChOOH, 6β-OOH had an 8-fold lower log10 reduction value in a clonogenicity assay. In seeking an explanation for this, we compared the susceptibilities of 5α-OOH, 6β-OOH, and 7α-OOH to GPx4-catalyzed reductive inactivation. (GPx4 remains the only enzyme known to be capable of detoxifying ChOOHs.) Kinetic experiments with both isolated and cell-borne enzyme revealed that 6β-OOH was the most reactive GPx4 substrate and 5α-OOH the least reactive, possibly due to its unique tertiary structure (14). This evidence suggested that 6β-OOH's limited cytotoxicity compared with 5α-OOH was simply due to its rapid reductive inactivation by GPx4. Although this could be a contributing factor, we now know, based on the present study, that 6β-OOH lacks the strong prooxidant properties of 5α-OOH and 7α/β-OOH, and this could be the most important reason for it being relatively innocuous. We examined only the β-epimer of the 6-positional hydroperoxide in this study due to the relatively low yield of the α-epimer in the photopreparative procedure used (5,8,11).
How can one account for the insignificant chain-initiating potency and cytotoxicity of 6β-OOH compared with 5α-OOH? A reasonable deduction based on the data presented is that both hydroperoxides in Fe(HQ)3/AH-- treated liposomes were equally prone to initial one-electron reduction by ferrous iron, but differed in their susceptibility to the second one-electron reduction involving hydrogen abstraction from an oxidizable lipid molecule, i.e. the actual chain-initiating step. This difference might be explained on thermodynamic grounds, i.e. the redox potential of 6β-O· might be significantly lower than that of 5α-O· such that the former lacks lipid oxidizing ability. Alternatively, a kinetic argument might apply, i.e. 6β-O· may abstract lipid hydrogens at a far lower rate than 5α-O·, possibly due to some structural factor, e.g. unusual orientation in the membrane bilayer. Distinguishing between these possibilities will require additional study; however, viz. determination of redox potentials for the two ChOOHs. At present, the kinetic explanation is tentatively more appealing because liposomal 6β-OOH does appear to give rise to at least some ChOX after prolonged incubation with iron and ascorbate.
This study places additional emphasis on the importance of 5α-OOH as a 1O2 adduct with cytotoxic and redox signaling properties (3-5). In mammalian tissues under 1O2 attack, e.g. long-wavelength UVA-exposed skin (16,17) or blue light-exposed retinal pigment epithelium (18), 5α-OOH would be generated more rapidly than any other ChOOH isomer, but detoxified most slowly by the GSH/GPx4 system. Accordingly, one would expect 5α-OOH to have a longer lifetime and accumulate to higher steady state levels than other ChOOHs and possibly also phospholipid hydroperoxides, given that the latter are much better GPx4 substrates than ChOOHs (19). Although 5α-OOH is known to undergo allylic rearrangement to 7α-OOH, this is typically insignificant at low levels of membrane peroxidation, as in a moderately 1O2-stressed cell (5). In addition to undergoing one- and two-electron turnover in a membrane of origin, LOOHs can desorb and transfer to other membranes, where such turnover may occur. If antioxidant capacity of acceptor compartments is insufficient, transfer will expand LOOH cytotoxic and redox signaling action (6). All ChOOHs translocate faster than phospholipid counterparts, but among the former, 5α-OOH moves significantly faster than 6β-OOH, both spontaneously and when assisted by a lipid transfer protein (6). Based on these established characteristics of 5α-OOH (fastest 1O2-mediated generation, slowest enzymatic detoxification, relatively rapid intermembrane transfer) and our evidence herein that 6β-OOH is incapable of launching membrane-damaging free radical peroxidation, 5α-OOH stands out as the most dangerous 1O2 adduct of Ch and as potentially more dangerous than 1O2 adducts of natural phospholipids.
This work was supported by NIH Grants CA72630 and HL85677 (AWG) and Polish Ministry of Science Grant N301-08332/3232 (WK).
†This invited paper is part of the Symposium-in-Print: “Phototoxicity of the Skin and Eye”, in honor of Dr. Colin Chignell