F4 - Isoprostanes as Specific Marker of Docosahexaenoic Acid Peroxidation in Alzheimer's Disease


  • Jaffar Nourooz-Zadeh,

  • Edwin H. C. Liu,

  • B. Yhlen,

  • Erik E. Änggåard,

  • Barry Halliwell

  • The present address of Dr. B. Halliwell is Department of Biochemistry, National University of Singapore, Kent Ridge Crescent, Singapore, 119260.

  • Abbreviations used : AAPH, 2,2' -azobis(2-amidinopropane) ; AD, Alzheimer's disease ; BHT, butylated hydroxytoluene ; DHA, docosahexaenoic acid ; GC-MS, gas chromatography-mass spectroscopy ; NBB, N-butylboronate ; NICI, negative ion chemical ionisation ; PFB, pentafluorobenzyl bromate ; PG, prostaglandin ; PMI, postmortem interval ; TMS, trimethylsilane.

Address correspondence and reprint requests to Dr. J. Nouroozzadeh at Centre for Clinical Pharmacology and Therapeutic Toxicology, Department of Medicine, University College, 5 University Street, WC1E 6JJ, London, U.K.


Abstract : F2-isoprostanes are prostaglandin-like compounds derived from free radical-catalysed peroxidation of arachidonic acid. Peroxidation of eicosapentaenoic acid produces F3-isoprostanes, whereas peroxidation of docosahexaenoic acid would give F4-isoprostanes. This study demonstrates the presence of esterified F4-isoprostanes in human brain and shows that levels are elevated in certain brain cortex regions in Alzheimer's disease. Our data with Alzheimer's disease suggest that analysis of F4-isoprostanes will provide new opportunities to study lipid peroxidation in the neurodegenerative diseases.

It has been postulated that increased production of reactive oxygen species and damage to docosahexaenoic acid (DHA) may contribute to the pathophysiology of neurodegenerative diseases (Reiter, 1995 ; Beal, 1996 ; Markesbery, 1997). DHA is an essential constituent of nervous tissue and an essential requirement for the development of the brain (Martinez, 1992). DHA is mainly present esterified to phospholipids in the brain and accounts for 40% of the polyunsaturated fatty acids (Söderberg et al., 1991 ; Wilson and Tocher, 1991 ; Martinez, 1992). The DHA content in brain is decreased in several neurodegenerative diseases (Söderberg et al., 1991 ; Wilson and Tocher, 1991 ; Martinez, 1992 ; Wilson and Sargent, 1993 ; Guan et al., 1994). However, it is not clear whether the decline in the content of DHA is a result of changes in fatty acid metabolism or oxidative modification of DHA. DHA is very peroxidisable. Peroxidisability of DHA is five times greater than that of linolenic acid (18:2ω6) (Cosgrove et al., 1987). It is uncertain whether DHA in nervous tissue is highly peroxidisable in vivo (Rehncrona et al., 1980 ; Halliwell, 1992).

A major advance in the measurement of lipid peroxidation was the discovery of F2-isoprostanes (Morrow et al., 1990), a series of prostaglandin (PG)-like compounds formed in vivo by free radical-catalyzed peroxidation of arachidonic acid. Peroxidation of arachidonic acid yields four subfamilies of F2-isoprostanes, and the 8-epi-PGF isomer has received most attention (Roberts and Morrow, 1997) as a reliable index of arachidonic acid oxidation in vivo (Gopaul et al., 1994a,b, 1995 ; Nouroozzadeh et al., 1995 ; Reilly et al., 1996). Peroxidation of eicosapentaenoic acid-liposomes produces F3-isoprostanes (Nourooz-Zadeh et al., 1997), and F4-isoprostanes are formed during peroxidation of liposomes containing DHA (Nourooz-Zadeh et al., 1998). Eight subfamilies of F4-isoprostanes could be formed from DHA owing to free radical attack at positions C6, C9, C12, C15, and C18(Fig. 1). These different families of isoprostanes should be unique biomarkers of peroxidation of individual fatty acids and would be very useful in neurological studies if they could be quantified in human brain, because many techniques to measure lipid peroxidation do not distinguish among oxidation of individual fatty acids (Halli-well, 1996).

Figure 1.

Proposed pathways for peroxidation of DHA and formation of F4-isoprostane regioisomers.

FIG. 1.

In this study, we describe methodology for the determination of levels of free and total (sum of free and esterified) isoprostanes in human brain, report the presence of esterified F4-isoprostanes in normal brain cortex, and show that levels are elevated in Alzheimer's disease (AD).



Reagents were obtained from the suppliers used previously (Nourooz-Zadeh et al., 1998).

Tissue samples

Human brain tissue was obtained from Dr. Nigel Cairns (Brain Bank, Institute of Psychiatry, Denmark Hill, London, U.K.). AD was clinically diagnosed according to the criteria of the National Institute of Neurological Disorders and Stroke and AD and Related Disorders Association and confirmed pathologically (for details, see Lyras et al., 1997).

Preparation of brain homogenate

Brain tissue (1 g) was added to 10 ml of ice-cold phosphate-buffered saline (10 mM potassium phosphate and 10 mMNaCl, pH 7.4) containing butylated hydroxytoluene (BHT ; 100 μM) to inhibit lipid peroxidation, and homogenate was prepared on ice using a Potter homogenizer.

2,2′-Azobis(2-amidinopropane) (AAPH)-dependent peroxidation

Brain homogenate was prepared in phosphate-buffered saline alone (no butylated hydroxytoluene), and AAPH was added to give a 20 mM final concentration. Incubation was at 37°C, and aliquots were removed at specific times for analysis of F4-isoprostanes. Control samples were incubated at 37°C without AAPH.

Extraction procedure

Brain tissue (100 mg) was transferred into a glass vial, methanol (1 ml) was added, and homogenate was prepared using a Potter homogenizer. The sample was transferred into a glass tube, chloroform (2 ml) was added, and the sample was vortex-mixed for 30 s. After centrifugation (5 min, 2,000 g), the organic (lower) layer was transferred into a new glass tube. Chloroform (2 ml) and methanol (1 ml) were added to the remaining aqueous phase, the extraction procedure was repeated, and the organic layers were pooled. Chloroform and methanol were added to the remaining aqueous phase, and the extraction procedure was repeated for a third time.

Analysis of isoprostanes

Total (sum of free and esterified) isoprostanes.

An aliquot (1 ml) of brain homogenate (100 mg/ml) was transferred into a glass tube, and aqueous KOH (4 M ; 500 μl) was added. After incubation at 45°C for 45 min, the pH was adjusted to 3 by adding 4 M HCl (500 μl). The internal standard PGF2-d4 (5 ng in 100 μl of ethanol) was then added and thoroughly vortexmixed. Ethyl acetate (10 ml) was added, and the sample was vortex-mixed for 20 s and then centrifuged at 1,000 g for 5 min at room temperature. The upper (organic) layer was transferred into a new glass tube. Total lipid extract from the previous step was applied onto an NH2 cartridge preconditioned with hexane (10 ml). The cartridge was washed with 10 ml of hexane/ethyl acetate (30 : 70, vol/vol), acetonitrile/water (90 : 10, vol/vol), and acetonitrile, respectively. Isoprostanes were eluted by washing the column with 5 ml of ethyl acetate/methanol/acetic acid (10/85/5, by volume) (Nourooz-Zadeh et al., 1995).

Free (unesterified) isoprostanes.

An aliquot (1 ml) of brain homogenate (100 mg/ml) was transferred into a glass tube. Total lipids were partitioned with ethyl acetate, and isoprostanes were isolated as described above.

Pentafluorobenzyl bromate (PFB) ester and trimethylsilane (TMS) ether derivatisation.

The final eluate from the NH2 chromatography step was dried under a stream of N2 at 45°C. PFB (40 μl, 10% in acetonitrile) and 20 μl of diisopropyl ethylamine (10% in acetonitrile) were added, and the sample was transferred to a screw-cap vial with a Teflon-lined cap and incubated at 45°C for 30 min. The solvent was removed under N2 at 45°C. N,O-Bis(trimethylsily)trifluoroacetamide (50 μl) and diisopropyl ethylamine (5 μl) were added to the dried residue, and the sample was incubated for 30 min at 45°C. The solvent was dried under a stream of N2, and the residue was reconstituted in 40 μl of N,O-bis(trimethylsilyl)trifluoroacetamide (0.1% in isooctane).

N-Butylboronate (NBB) derivatives.

The final extract from the NH2 chromatography step was converted to PFB ester derivatives as described above. Boronic acid [2% (wt/vol)] in dimethoxypropane (50 μl) was added to the dry extract, and the sample was incubated for 15 min at 60°C. The solvent was removed under a stream of N2, and TMS ether derivatisation was carried out as described above but without diisopropyl ethylamine.


The final extract from the NH2 chromatography step was resuspended in methanol (2 ml). Platinum (IV) oxide (10 mg) was added as catalyst, and the sample was flushed with H2 at room temperature for 2 h. The sample was centrifuged, the methanol layer was transferred into a new glass vial, and the solvent was dried under N2 at 40°C. PFB and TMS derivatization steps were carried out as described above.

Gas chromatography-mass spectroscopy (GC-MS)/negative ion chemical ionisation (NICI).

This was carried out on a Hewlett Packard model 5890 GC apparatus linked to a VG70SEQ using the NICI with ammonia as reagent gas, essentially as described by Nourooz-Zadeh et al. (1998). Samples (2 μl) were injected onto an SPB-1701 column (30 m × 0.25 mm i.d. ; film thickness, 0.25 μm ; Supelco, Dorset, U.K.). Separation was carried out using the following temperature programme : initial temperature, 175°C ; initial time, 2 min ; rate, 30°C/min ; final temperature, 270°C ; final time, 30 min. Samples (2 μl) were injected into a temperature-programmed Gerstel injector (Thames Chromatography, Maidenhead, U.K.). Quantitative analysis was performed using selected ion monitoring of the carboxylate anion [M-181]- at m/z 569, 593, and 573 for F2-isoprostanes, F4-isoprostanes, and PGF2-d4, respectively. For structural validation of the PG4-like compounds as NBB/PFB/TMS derivatives, the signals were monitored at m/z 514. The signal for the hydrogenated PG4-like compounds as PFB/TMS derivatives was monitored at m/z 601.

Fatty acid analysis.

The fatty acid profile of the brain samples was determined by the analysis of total fatty acids as their methyl esters. Two hundred microlitres (20 mg) of brain homogenate was mixed with 250 μl (100 μg) of heptadecanoic acid as an internal standard. Five hundred microlitres of ethyl acetate and 300 μl of water were then added to the homogenate, vortex-mixed, and centrifuged at 2,500 g for 5 min. The upper organic layer was the transferred to a clean glass vial. A further 500 μl of ethyl acetate was added to the remaining aqueous phase, vortex-mixed, and centrifuged. The upper organic layers were pooled and dried under a stream of nitrogen. Fatty acid methyl esters were prepared by adding 500 μl of boron trifluoride/methanol (14%) solution to the dried lipid, and the tubes were capped and incubated at 60°C for 30 min. The samples were allowed to cool before water and hexane were added. The mixtures were vortex-mixed and centrifuged. The hexane layer was collected and dried. The residue was redisolved in 100 μl of hexane, of which 1 μl was injected onto an Omegawax 320 column (0.32 μm × 30 m ; file thickness, 0.25 μm) using a temperature gradient of 120-240°C at 6°C/min. The signal was detected by a flame ionisation detector.


Optimization of total lipid extraction procedure for brain cortex

Folch extraction (chloroform/methanol, 2 : 1, vol/vol) is routinely used for extraction of isoprostanes from tissues. Initial studies with [3H]PGF added as a tracer to brain tissue revealed that only 64 ± 5% (n = 3) of it was recovered after homogenisation and three extractions with chloroform/methanol (2 : 1, vol/vol). Better extraction was achieved using ethyl acetate. A quantitative recovery (98 ± 3% ; n = 3) of the [3H]PGF added to brain homogenate was achieved at a ratio of ethyl acetate/aqueous phase of 5 : 1 (vol/vol), which was used in further experiments.

NH2 chromatography recovery experiment

Brain homogenate (l ml) was spiked with [3H]PGF as a tracer, and total lipids were extracted with ethyl acetate. The ethyl acetate layer was applied to an NH2 cartridge prewashed with hexane (5 ml), and the cartridge was treated as described in Materials and Methods. The radiolabel was quantitatively retained on the NH2 cartridge ; <2% was lost after washing the cartridge with 10 ml of hexane/ethyl acetate (30 : 70, vol/vol), acetonitrile/water (90 : 10, vol/vol), and acetonitrile. The majority of the radiolabel (83 ± 11% ; n = 3) was present in the final eluate from the NH2 cartridge using ethyl acetate/methanol/acetic acid (10 : 85 : 5, by volume ; 5 ml). Overall recovery of the radiolabel after total lipid extraction and chromatography on an NH2 cartridge was reproducibly 74 ± 4% (n = 3).

Analysis of F4-isoprostanes in brain cortex

Exposure of brain homogenate to AAPH, an agent that generates peroxyl radicals capable of initiating lipid peroxidation (Noguchi et al., 1993), at 37°C for 24 h produced an array of peaks (retention times, 20-26 min ; (Fig. 2). Components I and II were the major products. Minor components were also detected in the region between 26 and 32 min. Formation of components I and II peaked at 4 h (Fig. 3).

Figure 2.

Representative GC-MS/NICI chromatograms of total (sum of free and esterified) F4-isoprostanes (PFB ester/TMS ether derivatives) in human brain cortex (parietal lobe). The signals at m/z 593 and 573 represent the carboxylate anion [M - 180]- for the F4-isoprostanes and the PGF2-d4 (internal standard), respectively. The homogenate was incubated with AAPH (50 mM) at 37°C for 24 h, and aliquots were removed. Upper panel : Brain homogenate (without AAPH). Middle panel : AAPH-oxidized homogenate (37°C, 24 h). Bottom panel : PGF2-d4 (internal standard).

Figure 3.

Time course of F4-isoprostane formation during incubation of parietal cortex homogenate, without or with AAPH (final concentration, 50 mM) at 37°C. Aliquots were removed at the stated times and assayed for total (sum of free and esterified) F4-isoprostanes. Upper panel : Component I. Lower panel : Component II. Data are mean ± SD (bars) values (n = 6). Time courses were similar for other brain regions.

FIG. 2.

FIG. 3.

The chromatographic profile for total F4-isoprostanes in brain homogenate in the absence of AAPH was very similar to those from the AAPH-oxidized samples (Fig. 2). No component I and II could be detected if the hydrolysis step was omitted, i.e., no free F4-isoprostanes could be detected. Inclusion of the chain-breaking antioxidant BHT or of indomethacin as a cyclooxygenase inhibitor (each at 100 μM final concentration) before homogenisation had no effect on the total levels of components I and II detected in brain homogenate. BHT was added to all tissues before homogenisation (except in the AAPH experiment). There was no significant correlation between the levels of component I or II and age, postmortem interval (PMI), or storage time of the tissues.

Components I and II were identified as PGF4-like compounds by the following criteria : (a) GC-MS chromatographic profiles as PFB ester/TMS ether for the components in the extracts from freshly prepared brain homogenates were virtually identical to those obtained from AAPH-oxidized brain homogenate (Fig. 2). (b) Formation of cyclic boronate (NBB) derivatives caused loss of the signal at m/z 593 and appearance of new peaks at m/z 514, consistent with cis-oriented hydroxyl groups on the cyclopentane ring. (c) Hydrogenation abolished the signal at m/z 593, coincident with the appearance of new peaks at m/z 601, as expected from reduction of four double bonds. (d) Relative retention times and mass spectrometric characteristics of compounds I and II were identical in brain homogenate and after oxidation of DHA-liposomes by AAPH or by addition of Cu2+ (Nourooz-Zadeh et al., 1998). (e) Mass spectrometry analysis of F4-isoprostanes at high resolution (8,000 times) revealed that the chromatographic patterns for the components eluted between 20 and 26 min were identical to those measured at lower resolution (1,000 times).

Analysis of F2- and F4-isoprostanes in AD

Samples of occipital, temporal, and parietal lobe from patients who had died with clinically and pathologically confirmed AD were obtained and analyzed in parallel with pathologically normal tissue samples matched as far as possible for age (control, 74.1 ± 16.0 years ; AD, 79.9 ± 6.1 years ; p > 0.05 ; n = 8 and 10, respectively), storage time (p > 0.05), and PMI (control, 46.7 ± 12.0 h ; AD, 29.9 ± 18.2 h ; p = 0.025). No correlation of F4-isoprostane levels with age, storage time, or PMI was observed in this limited number of samples for either AD or control. Analysis of fatty acid residues in the brain tissue lipids showed that the percentages of arachidonic acid and DHA were not significantly different in normal and AD tissues. Levels of F2-isoprostanes were also identical (Fig. 4). However, in temporal and occipital lobes there was a rise in content of F4-isoprostanes, measured as levels of compound I or II. The rise was significant whether data were expressed per unit mass of tissue or on the basis of fatty acid residues (Fig. 4). However, no rise was observed in parietal lobe.

Figure 4.

Comparison of levels of total F2 - and F4-isoprostanes in cortical regions from AD and healthy brain tissues. Data are mean ± SD (bars) values (n = 8 and 10, respectively). *p < 0.05, **p < 0.01, by Student's t test. NS, not significant.

FIG. 4.


Assessment of the role of lipid peroxidation in human neurodegenerative diseases requires sensitive and reliable methods that can detect specific oxidation products (Halliwell, 1996). In this study we show that the assay we have developed for quantification of F4-isoprostanes is applicable to human brain as marker of DHA peroxidation. Final quantification of F4-isoprostanes is carried out by GC-MS/electron capture detection, a specific and sensitive technique.

Established procedures for the isolation of isoprostanes from biological tissues involve Folch extraction (chloroform/methanol, 2 : 1, vol/vol) and chromatography on a C18 and silica cartridge followed by TLC (Morrow and Roberts, 1994). However, only 64% of [3H]PGF as a tracer was recovered from brain homogenate after three extractions with chloroform/methanol (2 : 1, vol/vol). We found that replacement of the Folch technique with ethyl acetate extraction gave 98% recovery of the radiolabelled standard from brain homogenate. In addition, we replaced chromatography on C18 and silica cartridges and a TLC step by chromatography on an NH2 cartridge, which functions as an ion exchanger selectively binding organic compounds containing a carboxylate anion (Nourooz-Zadeh et al., 1995). The recovery of [3H]PGF following the NH2 chromatography step was 83 ± 11%, in agreement with the finding of Nourooz-Zadeh et al. (1995). Overall recovery of the radiolabelled standard from brain homogenate following ethyl acetate extraction and NH2 chromatography was 75 ± 11%. The combined ethyl acetate extraction and NH2 chromatography procedure works well for the isolation of PGF-like compounds from brain.

Analysis of brain homogenate revealed the presence of an array of peaks monitored at m/z 593 when analysed for F4-isoprostanes. Two peaks (I and II) with retention times between 24.4 and 25.0 min were present in abundance (Fig. 2). Exposure of brain homogenate to the free radical generator AAPH gave an identical array of peaks. Formation of components I and II during AAPH oxidation of the brain homogenate increased with time (Fig. 3) ; after 24 h the levels of components I and II had increased by 23- and 18-fold over controls, respectively. The criteria demonstrating that components I and II in brain are F4-isoprostanes include the following : (a) GC-MS profiles as PFB ester/TMS ether and relative retention times are identical to those obtained from AAPH- or copper-oxidized DHA-liposomes. (b) Analysis of the extract as NBB/PFB/TMS derivatives gave no peaks at m/z 593 but new peaks at m/z 514. PGF-like compounds only form cyclic boronate (NBB) derivatives when the -OH groups on the cyclopentane ring are cis-oriented. (c) Catalytic hydrogenation abolished the signal at m/z 593, and new peaks appeared at m/z 601, indicative of the presence of four double bonds.

DHA is highly susceptible to peroxidation in vitro because of the number of double bonds in the molecule (Halliwell, 1992). Therefore, it is important to minimise artefact formation during storage and/or during sample processing. In the present study, the following precautions were taken : (a) Brain tissue was kept at -70°C after tissue collection. Time of storage did not affect the F4-isoprostane level. Lyras et al. (1997) summarised the evidence that postmortem lipid peroxidation does not occur in human brain samples under such conditions. (b) BHT, a chain-breaking antioxidant, was added to the samples before homogenisation at a level that prevented brain peroxidation as measured by other assays. (c) The samples were kept on ice during homogenization.

The presence of F4-isoprostanes in human brain cortex suggests that they are formed in vivo, presumably by free radical attack on DHA residues esterified into lipids, because no free F4-isoprostanes were detected. Values of F2-isoprostanes were broadly comparable in occipital, temporal, and parietal lobes and were not elevated in AD samples (Fig. 4). Levels of F4-isoprostanes were low in occipital and temporal lobes but higher in the parietal lobe control samples examined. Levels in AD tissue were not significantly different in parietal lobe (although there was a very high variation between samples) but were clearly elevated in occipital and temporal lobes. These data, combined with the lack of changes in levels of F2-isoprostanes, suggest that peroxidation of DHA residues is elevated in AD. The AD and control samples were cell-matched for subject age and storage time. Although less well-matched for PMI, the mean PMI was shorter in AD, so this is unlikely to be a source of artefact.

In conclusion, we have introduced a technique that should be of value in investigating DHA peroxidation in human brain and illustrated its application using AD. The technique will also be of value in many other areas (Halliwell, 1996 ; Nourooz-Zadeh et al., 1998) and in further exploring the significance of DHA peroxidation in AD. Since our article was submitted, Roberts et al. (1998) have reported the presence of F4-isoprostanes in rat and pig brain and that levels in CSF are elevated in AD patients, confirming our data.


We thank the British Heart Foundation and the Ministry of Agriculture, fisheries, and Food for financial support and the Brain Bank of the Institute of Psychiatry, London, for supplying samples, especially Dr. Nigel Cairns. We also thank Dr. Leonidas Lyras for assisting with the provision of tissue.