LC/ESR/MS study of spin trapped carbon-centred radicals formed from in vitro lipoxygenase-catalysed peroxidation of gamma-linolenic acid

Abstract

Gamma-linolenic acid (GLA) has been reported as a potential anti-cancer and anti-inflammatory agent and has received substantial attention in cancer care research. One of the many proposed mechanisms for GLA biological activity is free radical-mediated lipid peroxidation. However, no direct evidence has been obtained for the formation of GLA-derived radicals. In this study, a combination of LC/ESR and LC/MS was used with alpha-[4-pyridyl-1-oxide]-N-tert-butyl nitrone (POBN) to profile the carbon-centred radicals that are generated in lipoxygenase-catalysed GLA peroxidation. A total of four classes of GLA-derived radicals were characterized including GLA-alkyl, epoxyallylic, dihydroxyallylic radicals and a variety of carbon-centred radicals stemming from the beta-scissions of GLA-alkoxyl radicals. By means of an internal standard in LC/MS, one also quantified each radical adduct in all its redox forms, including an ESR-active form and two ESR-silent forms. The results provided a good starting point for ongoing research in defining the possible biological effects of radicals generated from GLA peroxidation.

Figure 1

Off-line ESR spectra from the complete GLA-peroxidation system and the relevant control experiments. (A) ESR spectrum of the complete system at 30 min reaction time. The complete system (50 mm phosphate buffer, pH 7.5) contained 20 mm POBN, 1 mm GLA (in 1% ethanol) and 2 × 10⁴ Units/ml LOX. An ESR field scan (70 G) was performed and the hyperfine couplings of this spectrum were a^N ≈ 15.69 G and a^H ≈ 2.68 G; (B and C) ESR spectra of POBN and LOX control experiments. ESR field scans were performed at 30 min reaction time for the complete reaction mixtures absent POBN and LOX, respectively; (D) ESR spectrum of control experiment of GLA. Reaction mixture excluding GLA (adding GLA stock solution was replaced by adding the same volume of ethanol) was subjected to ESR field scan at 30 min. Hyperfine coupling constants of this spectrum were a^N≈15.71 G, a^H≈2.62 G.

Figure 2

On-line LC/ESR and LC/MS chromatogram of the enzyme-free, condensed ACN-sample mixture from the experiment in Figure 1A. (A) UV chromatographic separation was performed at an absorption of 265 nm with a C₁₈ column (ZORBAX Eclipse-XDB, 3.0 × 75 mm, 3.5 µm) equilibrated with solvent A (H₂O-0.1% HOAc). Conditions of gradient elution were described in Methods; (B) ESR chromatogram was obtained in an ESR spectrometer equipped with an Aquax ESR cell. A time scan was performed with the magnetic field fixed on the maximum of the first line of Figure 1A. There was a 9 s offset due to the connections between the UV detector and the ESR. On-line ESR settings were described in Methods; (C) On-line MS (full scan or total ion chromatogram, TIC, m/z 150 to m/z 600) was obtained with chromatographic conditions identical to those in on-line ESR. The LC flow rate (0.8 ml/min) was adjusted to 30 ~40 µl/min into the MS inlet with a splitter; the first three min of LC eluants were always by-passed and not analysed for their MS. Between UV and MS measurements there was a 35 s offset due to the connection settings. On-line MS settings were described in Methods; (D) Extracted ion current (EIC) chromatogram of four ions of (m/z 240, m/z 336, m/z 266 and m/z 296) from the above full scan was obtained for the MS profile matching the POBN adducts that were monitored as ESR-active peaks in on-line ESR.

Figure 3

LC/MS² spectra of selected POBN adducts that were ESR-active peaks in Figure 2B. (A) LC/MS² of m/z 240 ion for ESR-active peak 2; (B) LC/MS² of m/z 296 ion for ESR-active peak 3; (C) LC/MS² of m/z 336 ion for ESR-active peak 4 (similar LC/MS² spectra were also observed for ESR-active peaks 5 and 6); (D) LC/MS² of m/z 266 ion for ESR-active peak 7; (E) LC/MS² of m/z 506 ion for ESR-active peak 1; and (F) LC/MS² of m/z 515 ion from D₉-POBN experiment relevant to ESR-active peak 1 in E. Note that the fragmentation patterns and ‘a’, ‘b’, ‘c’ and ‘d’ ions of all tested POBN adducts were consistent with the LC/MS² of POBN adducts published elsewhere [30,31] as well as the pattern described in Scheme 1. D₉-POBN-related products always have retention times ~12 s shorter than its POBN counterpart under our chromatography conditions. Some unique characters were observed between each counterpart of POBN vs D₉-POBN experiment due to fragmentations with/without the loss of the tert-butyl group (‘a’ and ‘b’ ions/‘c’ ion).

Figure 4

EICs and LC/MS² of ESR-silent forms of POBN adducts that were identified in Figure 3. (A) EIC and LC/MS² (inset) of m/z 267 ion as the reduced form of POBN/^•C₅H₁₁; (B) EIC and LC/MS² (inset) of m/z 265 ion as the oxidized form of POBN/^•C₅H₁₁; (C) EIC and LC/MS² (inset) of m/z 297 ion as the reduced form of POBN/^•C₅H₉O₂; (D) EIC and LC/MS² (inset) of m/z 295 ion as oxidized form of POBN/^•C₅H₉O₂; (E) EIC and LC/MS² (inset) of m/z 337 ion as the reduced form of POBN/^•C₈H₁₃O₂; (F) EIC and LC/MS² (inset) of m/z 335 ion as the oxidized form of POBN/^•C₈H₁₃O₂; and (G and H) EICs and LC/MS² (insets) of m/z 505 and m/z 514 ion (D₉-POBN counterpart) as oxidized form of POBN adduct of ^•L(OH)₂, respectively. Note that a D₉-POBN-related product always has retention times ~12 s shorter than its POBN counterpart under our chromatography conditions. Fragmentation patterns and a, b, c and d ions of all tested POBN adducts (a’, b’, c’, d’ ions of all tested D₉-POBN adducts) were consistent with LC/MS² of POBN adducts published elsewhere [30,31] as well as those described in Scheme 1. The redox forms of the m/z 240 ion (m/z 241 and m/z 239) were not analysed because they are GLA-unrelated radicals. ‘×’ represents POBN-unrelated EIC peak; ‘*’ represents a portion of the ESR-active form(s) being reduced/oxidized during MS ionization.

Figure 5

EICs and LC/MS² of ESR-silent forms of POBN/L^• and POBN/OL^•. (A and B) EIC and LC/MS² of m/z 473 for the reduced product of POBN/L^•, as well as LC/MS² of the m/z 482 ion from the D₉-POBN experiment (inset); (C and D) EIC and LC/MS² of m/z 489 for the reduced product of POBN/OL^•, as well as LC/MS² of the m/z 498 ion from the D₉-POBN experiment (inset); (E and F) EIC and LC/MS² of the m/z 471 ion for the oxidized product of POBN/L^•, as well as LC/MS² of the m/z 480 ion from D₉-POBN experiment (inset); and (G and H) EIC and LC/MS² of the m/z 487 ion for the oxidized product of POBN/OL^•, as well as LC/MS² of the m/z 496 ion from the D₉-POBN experiment (inset). Note that fragmentation patterns and a, b, c and d ions of all tested products (a’, b’, c’ and d’ ions for the D₉-POBN experiments) were consistent with the LC/MS² of POBN products published elsewhere [30,31] as well as the pattern described in Scheme 1. ‘×’ represents POBN-unrelated EIC peak.

Figure 6

Comprehensive quantification of POBN/^•C₅H₁₁ via LC/MS. (A) EIC abundance comparison of m/z 204 (3.8 µg/ml internal standard of D₉-POBN) vs m/z 266 (ESR-active form of POBN/^•C₅H₁₁) at t_R≈21.2 min as well as at t_R≈11.5 and 19.1 min (generated from its two oxidized forms due to MS ionization); (B) EIC abundance comparison of m/z 204 vs m/z 267 (the reduced form of POBN/^•C₅H₁₁) at t_R≈15.1 min as well as at t_R≈ 11.5 and 21.2 min (generated from the oxidized form m/z 265 and the adduct form m/z 266 due to MS ionization, respectively); and (C) EIC abundance comparison between m/z 204 vs m/z 265 (oxidized forms of POBN/^•C₅H₁₁) at t_R≈11.5 and 19.1 min, as well as at t_R≈15.1 and 21.2 min (generated from m/z 267 and m/z 266 due to MS ionization, respectively). Quantanalysis version 1.8 for Agilent 6300 Series Ion trap LC/MS was used to process the integration and calculation of peaks.

Scheme 1

Chemistry of formation of spin adducts of GLA-derived carbon-centred radicals in LOX-catalysed GLA peroxidation, the subsequent redox reaction and relationship of the three redox forms of the POBN adduct.