Synthesis of triphenylphosphonium vitamin E derivatives as mitochondria-targeted antioxidants

A series of mitochondria-targeted antioxidants comprising a lipophilic triphenylphosphonium cation attached to the antioxidant chroman moiety of vitamin E by an alkyl linker have been prepared. The synthesis of a series of mitochondria-targeted vitamin E derivatives with a range of alkyl linkers gave compounds of different hydrophobicities. This work will enable the dependence of antioxidant defence on hydrophobicity to be determined in vivo. (cid:1) 2015 The Authors. Published Ltd. an the CC license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Mitochondria are essential to the functioning of most eukaryotic cells because they provide the energy necessary for cell activities in the form of an elevated adenosine triphosphate/diphosphate ratio by oxidative phosphorylation. 1 Because free radicals are produced as a side product of this respiration, mitochondrial localised oxidative damage accumulates faster than in the rest of the cell. Mitochondrial dysfunction due to oxidative damage has been implicated in a wide range of conditions from ageing, 2e4 ischaemia-reperfusion injury, 5e7 cancer, 8e10 epilepsy 11 and to neurodegenerative diseases 12e15 such as amyotrophic lateral sclerosis 16 and Alzheimer's 17,18 and Parkinson's 19 diseases. Because of the roles mitochondria play in a wide range of pathologies, the engineering of molecules to prevent mitochondrial oxidative damage has therapeutic potential. 20 Attaching a bioactive moiety to a lipophilic cation, such as triphenylphosphonium (TPP), enables non-mediated, membrane potential driven, accumulation of the active group within the mitochondrial matrix. 21e23 These molecules have been shown to actively accumulate within mitochondria in tissues following oral delivery. 24 Numerous examples in the literature, including in vitro, 22,23,25e27 ex vivo, 28 in vivo 29 as well as human trials 30 have demonstrated that mitochondria-targeted antioxidants are significantly more effective than non-targeted analogues at preventing mitochondrial oxidative damage in a range of pathologies.
To date, mitochondria-targeted antioxidants based on the natural antioxidants Coenzyme Q, 22 lipoic acid, 31 and vitamin E 23 have been synthesised. A series of compounds based on coenzyme Q (viz. MitoQ) have been synthesised with 3-, 21 5-, 21 [10][11][12][13][14][15][16][17][18][19][20][21]22 and 15carbon 21 alkyl chains linking the TPP and quinone functional groups (MitoQ 3e15 ). This set of compounds displayed a wide range of lipophilicities 21 and there was significant dependence of the protective behaviour in vitro on the linking chain length. 32 The maximum antioxidant efficacy for the MitoQ n series against mitochondrial oxidative damage was obtained with a 10-carbon alkyl chain (viz. MitoQ 10 ). This intriguing chain-length dependence was found to be due to increased hydrophobicity which enhanced the extent of uptake into mitochondria by favouring adsorption to the mitochondrial inner membrane; longer linker chain-length which allowed the antioxidant quinol moiety to penetrate deeper into the core of the mitochondrial inner membrane relative to the TPP moietydwhich was localized close to the membrane surface; and longer chain length which allowed access of the ubiquinone moiety to the active site of mitochondrial complex II thereby facilitating its rapid reduction to the active ubiquinol antioxidant. 32 Only one targeted vitamin E molecule (MitoE) has been reported to date and this contains a 2-carbon chain linking the functional, antioxidant chroman moiety of vitamin E to the targeting TPP cation and is thus termed MitoE 2 (1). 23 As the antioxidant efficacy of the MitoQ n compound depended on alkyl chain length, we have carried out a synthetic study to obtain mitochondria-targeted compounds based on vitamin E with varying alkyl chain lengths, and consequently varying lipophilicities. Previously some of the compounds obtained from this study have been used to create a series of vitamin E succinate derivatives to assay for anticancer activity 33 and this report also provides full experimental support for the precursors involved in that work.
The previous synthesis 23, 25 of MitoE 2 involved a large number of steps, was not amenable to the creation of analogues, and was not adaptable to scale-up. Retrosynthetic analysis of a generic MitoE n with chain length n (A) (Fig. 1) shows it can be formed from B by displacement of a leaving group Y. The structure B is the key intermediate in the synthetic scheme as this establishes the basic carbon framework of the target molecule. The substituted heterocyclic ring in B can be derived from 2,3,5-trimethyl-p-hydroquinone (C) and a tertiary allylic alcohol (D) which in turn can be derived from a methyl ketone (E) and a vinyl organometallic species (F). The appropriate methyl ketone (E) can be formed from the corresponding u-hydroxy alkyne (G).
As a variety of u-hydroxy alkynes are available, especially by utilising acetylene zipper chemistry, 31 this provides the required flexibility to form any MitoE n . Using this general approach we report the synthesis of MitoE n as the mesylate salts with n¼2, 4, 6 (1, 2, 3) and also MitoE 10 (4) and MitoE 11 (5) using related chemistry.

Results and discussion
The synthesis of MitoE 6 (3) (Scheme 2), will be described in detail as an exemplar of the methodology. For this the required u-   (7) in 95% yield (Scheme 1). The primary hydroxyl group was then converted into a THP ether (8) in 86% yield followed by reaction with vinylmagnesium chloride to readily afford the tertiary allylic alcohol 9, following chromatography with 0.1% Et 3 N in the elutant, in 96% yield. Reaction of 9 with 2,3,5-trimethyl-phydroquinone (10) in acid 35 dpreferably formic acid 36 dafforded diol 11 in 53% yield.
The synthesis of MitoE 2 and MitoE 4 followed the same basic route although some specific modifications for producing the key intermediate allylic tertiary alcohol precursors (14,19) were required reflecting availability of suitable starting materials or undesired intramolecular reactions. Thus the synthesis of the MitoE 2 hydroxychroman intermediate (15, Scheme 1) was completed starting with 4-hydroxy-2-butanone (12) and proceeding via the THP ether (13) and tertiary allylic alcohol (14). The synthesis of MitoE 4 required the use of 5-hexyn-1-ol (16) as the starting material. Reaction of 16 with aqueous Hg(OTf) 2 $(TMU) 2 resulted in significant formation of the cyclic hemiketal, 2-methyl-tetrahydro-2H-pyran-2-ol. 37 To overcome this undesired cyclization the primary alcohol was converted to an acetate (17) before transformation into the acetoxy methyl ketone (18). 32 Treatment of 18 with excess vinylmagnesium chloride simultaneously formed the required tertiary allylic alcohol functionality and hydrolysed the acetate ester affording 19.
Preliminary studies on the formation of triphenylphosphonium salts from chromanols with an appropriate primary leaving group and triphenylphosphine showed substantial degradation of the heterocyclic system, particularly when the rate of phosphonium salt formation was relatively slow. While the nature of these undesired side reactions was not fully elucidated it was considered useful to protect the electron rich phenol function with an electron withdrawing group during phosphonium salt formation. There are several literature reports of the use of the methanesulfonyl (mesyl) group to protect phenols with subsequent removal using basic reagents. 38 The mesyl group can therefore serve two functions, as an electrophilic activator of the primary alcohol by forming a mesyloxy group while also providing phenol protection during phosphonium salt formation.
Diol 11 was treated with two equivalents of MsCl in the presence of Et 3 N and afforded the bis-mesylate 21 in 78% yield (Scheme 2). The bis-mesylate was then reacted with PPh 3 at 90 C for 48 h to give 22 in 92% yield. Removal of the aryl mesyl ester was trialled with the mesyl derivative of vitamin E and, while reactions using NaBH 4 , Cs 2 CO 3 or MeONa/MeOH showed no change, reaction with three equivalents of LDA resulted in complete conversion to the phenol vitamin E (Supplementary data). To compensate for the likely ylide formation with the phosphonium substrates, the amount of LDA used in the reaction with 22 was increased to 6 equiv and this protocol afforded MitoE 6 (3) mesylate in 49% yield.
The synthesis of MitoE 4 (2) mesylate was also completed from 20 following the same sequence (viz. 20/23/24/2) (Scheme 2). For the synthesis of MitoE 2 (1) the bis-mesylate (25) was readily obtained from 15, but direct reaction of 25 with PPh 3 at 80e90 C produced no phosphonium salt. However addition of 5 equiv of NaI to the melt gave 26, which was then treated with LDA to afford MitoE 2 (1) mesylate after anion exchange.
During the developmental phase the syntheses of MitoE 10 (4) and MitoE 11 (5) as the mesylate salts were also completed using less efficient earlier variations on the optimised route presented above (Supplementary data). MitoE 10 (4) was synthesised starting from 1hydroxy-11-dodecyne, obtained by the addition of 1-bromononane to lithiated propargyl THP ether followed by triple bond migration using NaH and ethylenediamine. 39 In both cases the final step involved reaction of triphenylphosphine with the mesyloxy phenol (S7, S11, Supplementary data) and the adverse effects of extended exposure of the unprotected phenol moiety to these reaction conditions was evident.
The NMR spectra for the mesylates MitoE 2 (1), MitoE 4 (2), MitoE 6 (3), MitoE 10 (4) and MitoE 11 (5) displayed a number of common features which are summarized in Fig. 2 and Table 1. As an example the 1 H NMR spectrum of MitoE 6 (3) contained a triplet at d 2.55 (J¼6.8 Hz), consistent with the signal from the methylene group at position 4. NOESY and gCOSY correlations from this resonance to a multiplet at d 1.66e1.80 allowed assignment to the methylene protons at 3. Both of these signals had NOESY correlations to a 3-proton singlet at d 1.17 therefore assigned to the methyl protons at 12. The resonances from the aryl methyl groups were identified from NOESY correlations as from the protons at 10 (d 2.10) and 11 (d 2.00) and a correlation between the signal from the remaining methyl group 9 (d 2.07) and the methylene resonance of 4. The signal from the protons at position 6 0 was clearly evident as a multiplet at d 3.22e3.30.  Further structural confirmation for MitoE 2 (1) was provided by X-ray diffraction of the crystalline bromide (Fig. 3). In this structure, the sixemembered heterocyclic ring of MitoE 2 adopted an envelope conformation with q¼54.5 (lit. 40 54.7 ) and with C2 deviating from the O1eC3eC4eC4aeC8a plane by 0.699 A, towards the phosphonium group.
To determine whether the biochemical properties and antioxidant efficacy of the chromanol moiety of MitoE were affected by changing the length of alkyl chain conjugating it to the TPP cation, the ability of the most (MitoE 10 ) and least (MitoE 2 ) lipophilic members of the MitoE n series to prevent lipid peroxidation was measured and compared. The ability of the compounds to act as chain breaking antioxidants in the rat brain homogenate system of lipid peroxidation 23 was assessed. This system was chosen as comparison of the effects of the antioxidants on preventing lipid peroxidation would not be confounded by differential uptake into mitochondria. Rat brain homogenates were allowed to undergo spontaneous lipid peroxidation, which was assessed by the production of thiobarbituric reactive species (TBARS). 22 The effect of MitoE 2 , and MitoE 10 on preventing this peroxidation was then assessed (Fig. 4) and showed that both compounds were of comparable efficacy in preventing lipid peroxidation. Therefore it is concluded that conjugation to the TPP cation by differing alkyl chain lengths to the chromanol moiety does not significantly alter their intrinsic antioxidant efficacy. Consequently any changes in antioxidant efficacy seen in mitochondrial or cell studies can be assigned to differences in uptake, adsorption or recycling. The antioxidant efficacy of the MitoE compounds was far greater than that of Trolox, which contains the same antioxidant chromanol moiety as MitoE, connected to a short-chain carboxylic acid rather than a TPP function, making it far more hydrophilic. This suggests that the lipophilic nature of the TPP moiety enhances the interaction of the antioxidant moiety with the phospholipid bilayer.
This aspect was extended by measuring whether MitoE 2 and MitoE 10 compounds were accumulated by mitochondria in response to the mitochondrial membrane potential, as is expected for a compound linked to a TPP compound. To measure the uptake of the MitoE compounds an ion-selective electrode was used that responds to the concentration of the TPP cation in solution (Fig. 5).   When the MitoE compounds were added to a mitochondrial suspension the ion-selective electrode responded to the increase in concentration of the MitoE compound in the extracellular environment. When the mitochondria were energised with the respiratory substrate succinate a large membrane potential across the mitochondrial inner membrane was established. This led to the extensive accumulation of the compounds within mitochondria, thus lowering the extracellular concentration which is detected by the electrode. Addition of the uncoupler FCCP resulted in the dissipation of the membrane potential and consequent release of the compounds back in to the supernatant and this is evident from the electrode response.
Therefore these experiments showed that, as expected, both MitoE 2 and MitoE 10 were taken up by energised mitochondria in response to the membrane potential. To see if this accumulation within mitochondria enhanced the ability of the mitochondriatargeted compounds to decrease oxidative damage to isolated mitochondria the activity was assessed and compared with the antioxidant efficacy of Trolox, a chroman containing antioxidant molecule, which is not taken up by mitochondria (Fig. 6).
This showed that both MitoE 2 and MitoE 10 were comparably protective against mitochondrial lipid peroxidation and that both were more protective against oxidative damage that Trolox. Finally, the ability of the most effective version of MitoE, MitoE 10 to protect against oxidative damage to mitochondrial DNA caused by the redox cycling molecule menadione was determined (Fig. 7). This also showed that MitoE 10 was able to protect against this form of mitochondrial oxidative damage more effectively that the control compound decylTPP. MitoE 2 was not protective in this assay (data not shown) consistent with the greater protection of MitoE 10 against lipid peroxidation.

Conclusion
The development of this generalised route to MitoE analogues has allowed a suite of targeted analogues to be prepared. We have assessed the biochemical properties of the most (MitoE 10 ) and least (MitoE 2 ) lipophilic members of the MitoE compounds and found that they are effective. By analogy with the MitoQ suite of compounds the series of MitoE compounds we have made will also have a range of lipophilicities and can be used to assess the effect of lipophilicity on the biological effects of MitoE. This strategy is both efficient and can be systematically varied to create a range of MitoE compounds. These compounds accumulate in mitochondria and preliminary biological data demonstrate that MitoE analogues show greater efficacy in preventing lipid peroxidation, mitochondrial oxidative damage and damage to mitochondrial DNA than non-targeted compounds. Further work needs to be conducted to fully understand the trend of biological behaviour of the MitoE series and determine the activity of the intermediate MitoE compounds and enable the dependence of the antioxidant efficacy of MitoE on chain length to be assessed in vivo.

General procedures
Thin Layer Chromatography (TLC) was performed with silica gel (Merck) 60F 254 coated on aluminium roll and were developed in solvent mixtures as indicated. Plates were visualised first with UV light (254 nm) then stained with vanillin or phosphomolybdic acid and heated. Column chromatography was performed using Merck 60 Silica (200e400 mesh, 40e63 mm) as the adsorbent. Columns were pre-equilibrated with the starting solvent before use and 50 g of adsorbent per g of crude product was used. Anion Exchange Chromatography was performed using Amberlite Ò IRA-400(Cl) ion exchange resin. The column was pre-equilibrated with 10% aqueous MsOH before use. Material was loaded in MeOH and the column was eluted with 1:   on a Varian INOVA-500 spectrometer at 11.74T and 298 K, operating at 499.74 MHz and 125.67 MHz, respectively as indicated. 31 P NMR spectra were acquired on a Varian INOVA-300 spectrometer at 7.05 T and 298 K operating at 121.40 MHz. Spectra were acquired in CDCl 3 , CD 2 Cl 2 and CD 3 OD as indicated. The solvent peak was used as an internal reference for 1 H and 13 C NMR spectra. In CDCl 3 the 1 H and 13 C NMR spectra were referenced to 7.26 ppm and 77.08 ppm, respectively; in CD 2 Cl 2 1 H and 13 C NMR spectra were referenced to 5.31 ppm and 53.8 ppm, respectively and CD 3 OD to 3.31 ppm and 49.0 ppm, respectively. Phosphoric acid (85%) was used as an external reference at 0 ppm for 31 P NMR spectra. Spectra were processed using standard Varian software. For each 1 H NMR signal, chemical shifts (d), relative integral, multiplicity, coupling constant (J), and assignment information are given unless otherwise stated. 13  X-ray diffraction data were collected on a Bruker APEX II CCD diffractometer, with graphite monochromated Mo-Ka (l¼0.71073 A) radiation. Intensities were corrected for Lorentz polarisation effects and a multiscan absorption correction was applied. 41 The structure was solved by direct methods (SIRe97 41 ) and refined on F 2 using all data by full-matrix least-squares procedures (SHELXL 97 42 ).

Syntheses
4.2.1. General synthetic procedures. Argon was used for reactions requiring an inert atmosphere. Standard vacuum line Schleck techniques were employed; glassware was flame-dried before use and cannula were used for transferring liquids between reaction vessels. Removal of solvents was achieved by either rotary evaporation at temperatures of up to 50 C, or by evaporation under a stream of argon. Some syntheses were carried out a 15 mL Kimax brand round bottomed 10 mm id borosilicate test-tubes fitted with a screw-top that enables a convenient sealed system under argon to be established.
Diisopropylamine was distilled from NaOH before use. Trifluoromethanesulfonic anhydride (Tf 2 O) was distilled from P 2 O 5 immediately before use. 2,3,5-Trimethyl-p-hydroquinone (10) was stirred as a suspension in hexane (10 equiv w/v) was stirred for 30 min. The solvent was removed by filtration and the pale yellow precipitate was dried at <0.5 mmHg for 2 h. Pyridine was refluxed over KOH and distilled onto 4 A molecular sieves before use. All other reagents were used without purification.
Absolute alcohol was dried over 4 A molecular sieves. Acetone was refluxed then distilled onto 3 A molecular sieves. Acetonitrile (CH 3 CN) was refluxed over CaH 2 and distilled directly before use. Dichloromethane (CH 2 Cl 2 ) was refluxed over P 2 O 5 and distilled onto 3 A molecular sieves. Diethyl ether (Et 2 O) was distilled before use and stored over sodium wire. Tetrahydrofuran (THF) was refluxed over KOH then distilled onto Na wire. The pre-dried solvent was then freshly distilled under an argon atmosphere over a benzophenone-K-Na amalgam (4:1) prior to use. All other solvents were used without purification. 4.2.2. 9-(Tetrahydro-2H-pyran-2-yloxy)-3-hydroxy-3-methyl-non-1-ene (9). Vinylmagnesium chloride (1.4 M in THF, 6.3 mL, 8.82 mmol) was added to a solution of 8 (Supplementary data) (1.006 g, 4.41 mmol) in anhydrous THF (60 mL) stirring at À78 C. This was stirred for 2 h and then allowed to warm to room temperature over 30 min. To the reaction mixture was added dropwise saturated aqueous NH 4 Cl (50 mL) and then this was extracted with Et 2 O (3Â50 mL). The combined organic phase was washed with saturated aqueous NaCl (50 mL), dried over anhydrous MgSO 4 , filtered and concentrated to give 9 as a pale yellow oil (1.086 g, 4.24 mmol, 96%) which was used without further purification. Analysis calcd for C 15 (11). A solution of 9 (0.959 g, 3.74 mmol) and freshly prepared 2,3,5trimethyl-p-hydroquinone (10, 0.460 g, 3.02 mmol) in formic acid (90 mL) were heated to reflux and refluxed for 3 h under an atmosphere of argon. The reaction was poured into crushed ice (w180 mL) and this was extracted with Et 2 O (3Â100 mL) under argon. The combined organic phase was washed with H 2 O (3Â100 mL) under argon, dried over anhydrous MgSO 4 and concentrated. The oily brown residue was dissolved in MeOH (90 mL), conc. HCl (0.090 mL) was added and the reaction refluxed for a further 30 min under argon. The reaction was diluted with ice cold H 2 O (200 mL) and this was extracted with Et 2 O (3Â100 mL) under argon. The combined organic phase was washed under argon with H 2 O (3Â400 mL), saturated aqueous Na 2 CO 3 (3Â100 mL) and H 2 O again (3Â100 mL), dried over anhydrous MgSO 4 , filtered and concentrated to give a brown oil. The crude product was purified by column chromatography on silica gel and elution with 1:9 Et 2 O:CH 2 Cl 2 followed by crystallisation from CH 2 Cl 2 and afforded 11 as a pale yellow solid (0.490 g, 1.60 mmol, 53%   (14). Vinylmagnesium chloride (1.40 M in THF, 9.5 mL, 13.3 mmol) was added to a solution of 13 (Supplementary data) (1.006 g, 5.84 mmol) in anhydrous THF (50 mL) stirring at À78 C. This was stirred for 2 h and then allowed to warm to room temperature over 30 min. Saturated aqueous NH 4 Cl (50 mL) was added dropwise and this was extracted with Et 2 O (3Â50 mL). The combined organic phase was washed with saturated aqueous NaCl (50 mL), dried over anhydrous MgSO 4 , filtered and concentrated to give a pale yellow oil. The crude product was purified by column chromatography on silica gel and elution with 1:3 Et 2 O:CH 2 Cl 2 containing 0.1% Et 3 N afforded pure 14 as a colourless liquid (1.091 g, 5.45 mmol, 93%). Analysis calcd for C 11   4.2.6. 2-(4-Hydroxybutyl)-2,5,7,8-tetramethyl-chromen-6-ol (20). A solution of 19 (1.062 g, 7.37 mmol) and freshly prepared 2,3,5trimethyl-p-hydroquinone (10, 1.031 g, 6.78 mmol) in formic acid (200 mL) were heated to reflux and refluxed for 3 h under an atmosphere of argon. The reaction was poured onto crushed ice (w600 mL) and this was extracted with Et 2 O (3Â200 mL) under argon. The combined organic phase was washed with H 2 O (3Â200 mL) under argon, dried over anhydrous MgSO 4 and concentrated. The oily brown residue was dissolved in MeOH (200 mL), conc. HCl (0.200 mL) was added and the reaction refluxed for a further 30 min under argon. The reaction was diluted with icecold H 2 O (400 mL) and this was extracted with Et 2 O (3Â200 mL) under argon. The combined organic phase was washed under argon with H 2 O (3Â200 mL), saturated aqueous Na 2 CO 3 (3Â200 mL) and H 2 O again (3Â200 mL), dried over anhydrous MgSO 4 , filtered and concentrated to give a brown oil. Column chromatography on silica gel and elution with 1:9 Et 2 O:CH 2 Cl 2 afforded 20 which recrystallised from CH 2 Cl 2 to give a pale yellow solid (1.061 g, 3.81 mmol, 52%). Analysis calcd for C 17

4-(6-Methanesulfonyloxy-2,5,7,8-tetramethyl-chromen-2-yl)
hexyl methanesulfonate (21). A solution of 11 (0.323 g, 1.05 mmol) and Et 3 N (0.70 mL, 0.511 g, 5.05 mmol) was stirred in anhydrous CH 2 Cl 2 (60 mL) at room temperature for 5 min. Methane sulfonyl chloride (MsCl) (0.20 mL, 0.136 g, 1.19 mmol) was added and the reaction was stirred for a further 1 h. The reaction mixture was washed with H 2 O (5Â50 mL), and saturated aqueous NaHCO 3 (50 mL), dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a yellow oil. The crude oil was purified by crystallisation from EtOH to give 21 as a white solid (0.378 g, 0.82 mmol, 78%). Analysis calcd for C 15 (23). A solution of 20 (0.468 g, 1.68 mmol) and Et 3 N (1.12 mL, 0.818 g, 8.08 mmol) was stirred in anhydrous CH 2 Cl 2 (30 mL) at room temperature for 5 min. MsCl (0.315 mL, 0.463 g, 4.04 mmol) was added and the solution was stirred for 1 h. The reaction mixture was washed with H 2 O (5Â30 mL) and saturated aqueous NaHCO 3 (30 mL), dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a yellow solid which was recrystallised twice from EtOH to afford 23 as a white solid (0.543 g, 1.25 mmol, 74%). Analysis calcd for C 19  and Et 3 N (424 mL, 0.310 g, 3.06 mmol) was stirred in anhydrous CH 2 Cl 2 (6 mL) at room temperature for 5 min. MsCl (88.0 mL, 0.129 g, 1.13 mmol) was added and the reaction was stirred for a further 1 h. The reaction mixture was diluted with CH 2 Cl 2 (5 mL). This was washed with H 2 O (5Â20 mL), dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a white solid (0.224 g) which, after recrystallisation from EtOH to give 25 as white crystals (0.167 g, 0.41 mmol, 81%). Analysis calcd for C 17 (26). A mixture of 25 (0.300 g, 0.74 mmol), NaI (0.558 g, 3.74 mmol) and triphenylphosphine (1.55 g, 3.69 mmol) was flushed with argon in a Kimax tube then sealed and the reaction was stirred as a melt at 90 C for 48 h. The crude product was dissolved in CH 2 Cl 2 (w2 mL) and precipitated from petroleum ether 40e60 (100 mL) three times. The product was dissolved in methanol and passed down an anion exchange column loaded with À OMs. The residual solvents were removed to give 26 as a white solid (0.450 g, 0.67 mmol, 91%). The UV spectrum of this compound has been reported previously. 23 (1). A solution of lithium diisopropylamide was prepared by adding diisopropylamine (0.30 mL, 0.216 g, 2.13 mmol) to anhydrous THF (4 mL) at À78 C followed by n-BuLi (1.8 M in hexane, 1.0 mL, 1.8 mmol). The solution was stirred at À78 C for 30 min and then allowed to warm to 0 C. The lithium diisopropylamide solution was then added to a solution of 26 (0.195 g, 0.29 mmol) in anhydrous THF (4 mL) with stirring at 0 C. After 30 min the solution was allowed to warm to room temperature and then aqueous saturated NH 4 OMs (10 mL) was added. The aqueous layer was extracted with CH 2 Cl 2 (3Â10 mL) and the combined organic phases dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a pale yellow oil (1.003 g). The crude product was dissolved in CH 2 Cl 2 (0.5 mL) and precipitated from Et 2 O (20 mL) twice then chromatographed on silica gel with elution with 1:9 EtOH:CH 2 Cl 2 and finally passed through an anion exchange column loaded with À OMs in methanol. The residual solvents were removed by freeze drying to give 1 as a white solid (52. 8

4.2.
14. (4-(6-Hydroxy-2,5,7,8-tetramethyl-chroman-2-yl)-butyl)triphenylphosphonium methanesulfonate, MitoE 4 (2). A solution of lithium diisopropylamide (2.07 mmol) in anhydrous THF (4.0 mL) was prepared as for 2 above and added to a solution of 24 (0.192 g, 0.28 mmol) in anhydrous THF (4.0 mL) stirring at 0 C. After 30 min the solution was allowed to warm up to room temperature and then aqueous saturated NH 4 OMs (10 mL) was added. The aqueous layer was extracted with CH 2 Cl 2 (3Â10 mL). The combined organic phases were dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a pale yellow oil. The crude product was dissolved in CH 2 Cl 2 (0.5 mL) and precipitated from Et 2 O (20 mL) twice and then purified by column chromatography on silica gel and elution with 1:9 EtOH:CH 2 Cl 2 to afford 2 as a white solid (0.091 g, 0.15 mmol, 53%  lithium diisopropylamide (1.49 mmol) in anhydrous THF (5.0 mL) was prepared as for 2 above and added to a solution of 22 (0.152 g, 0.21 mmol) in anhydrous THF (10 mL) with stirring at 0 C. After 30 min the solution was allowed to warm to room temperature and then aqueous saturated NH 4 OMs (10 mL) was added. The aqueous layer was extracted with CH 2 Cl 2 (3Â1 mL). The combined organic phases were dried over anhydrous MgSO 4 , filtered and concentrated in vacuo to give a pale yellow oil which was dissolved in CH 2 Cl 2 (0.5 mL) and precipitated from Et 2 O (20 mL) twice. Purification by column chromatography on silica gel and elution with 1:9 EtOH:CH 2 Cl 2 followed by freeze drying gave 3 as a white solid (0.067 g, 0.10 mmol, 49%). TLC: R f 0.19 (1:9 EtOH:CH 2 Cl 2 ); HRMS (þve ESI) m/z calcd for [M] þ : 551.3073, found: 551.3078; 1 H NMR Rat brain homogenates were prepared and incubated for 30 min at 37 C as described 23 with a range of concentration of MitoE compounds, Trolox or ethanol carrier. The extent of lipid peroxidation was assessed by measuring TBARS. 22 Values were corrected for the background level at t¼0 and were expressed as a % of the control with EtOH carrier. Data are the meansAESD of 4e6 independent experiments.

Mitochondrial and cell experiments
Rat liver mitochondria were prepared as described. 22 Mitochondrial uptake of MitoE 2 , MitoE 10 was measured using an ionselective electrode for TPP, as previously described 21 Briefly, isolated rat liver mitochondria (1 mg protein/mL) were incubated in 120 mM KCl, 10 mM HEPES, 1 mM EGTA, pH 7.2, KOH supplemented with 4 mg/mL rotenone at 37 C with stirring. The TPP-electrode response was calibrated with five successive additions of 1 mM MitoE 2 or MitoE 10 (from a 1 mM stock in EtOH). The mitochondria were energised with 10 mM succinate as a respiratory substrate to induce membrane potential-dependent uptake of the MitoE compound into the mitochondrial matrix. Finally the mitochondria were uncoupled with 1 mM FCCP, dissipating the membrane potential and resulting in release of MitoE.
To measure mitochondrial lipid peroxidation, isolated rat liver mitochondria (2 mg protein/mL) were incubated in KCl medium as above, supplemented with 10 mM succinate and 4 mg/mL rotenone at 37 C and pretreatedAEMitoE for 2 min. Lipid peroxidation was then initiated with 1 mM cumene hydroperoxide, and the mitochondria were incubated for a further 15 min. Lipid peroxidation was assayed as TBARS as above. Values were corrected for background level without cumene hydroperoxide and were expressed as a % of the control with EtOH carrier. Data are the meansAESD of 3e4 independent experiments.
To measure mitochondrial DNA damage, C2C12 cells were seeded at 20,000 cells/cm 2 in six-well culture plates and grown overnight at 37 C. Cells were then pre-incubated in 100 nM MitoE 10 or n-decyltriphenylphosphonium bromide (decylTPP) for 30 min, and then 25 mM menadione was added to generate oxidative damage within the cells and incubated for a further 1 h. Cells were washed and DNA was isolated using the DNeasy Blood and Tissue kit from Qiagen, quantitated using the Picogreen Assay (Invitrogen), and then the extent of oxidative damage was assessed by a quantitative PCR assay comparing the relative amplification of a long (w10 kbp) of mitochondrial DNA normalised to a short (w100 bp) section as described. Data are expressed relative to untreated cells and are meansAESEM of three independent experiments. Statistical significance was determined by Student's ttest. **p<0.01 compared against menadione treated samples.