Free-Radical-Mediated Formation of Trans-Cardiolipin Isomers, Analytical Approaches for Lipidomics and Consequences of the Structural Organization of Membranes †

Free-radical-mediated processes, such as peroxidation, isomerization and hydrogenation affecting fatty acid integrity and biological functions, have a trans-disciplinary relevance. Cardiolipins (CL, (1,3-diphosphatidyl-sn-glycerol)) and tetra-linoleoyl-CL are complex phospholipids, exclusively present in the Inner Mitochondrial Membrane (IMM) lipids, where they maintain membrane integrity and regulate enzyme functionalities. Peroxidation pathways and fatty acid remodeling are known causes of mitochondrial disfunctions and pathologies, including cancer. Free-radical-mediated isomerization with the change of the cis CL into geometrical trans isomers is an unknown process with possible consequences on the supramolecular membrane lipid organization. Here, the formation of mono-trans CL (MT-CL) and other trans CL isomers (T-CL) is reported using CL from bovine heart mitochondria and thiyl radicals generated by UV-photolysis from 2-mercaptoethanol. Analytical approaches for CL isomer separation and identification via 1H/13C NMR are provided, together with the chemical study of CL derivatization to fatty acid methyl esters (FAME), useful for lipidomics and metabolomics research. Kinetics information of the radical chain isomerization process was obtained using γ-irradiation conditions. The CL isomerization affected the structural organization of membranes, as tested by the reduction in unilamellar liposome diameter, and accompanied the well-known process of oxidative consumption induced by Fenton reagents. These results highlight a potential new molecular modification pathway of mitochondrial lipids with wide applications to membrane functions and biological consequences.


Optimization of the transesterification conditions
The procedure describes in the main text (section 2.1), was accurately determined in order to work under dry conditions. In fact, the inconsistency and not reproducibility of transesterification experiments of CL using HCL-MeOH and/or KOH/MEOH at different concentrations were caused probably by the presence of water (also in traces) formed during the reaction, slowing and negatively affecting the yield. The use of NaOMe sensibly reduced the problem of water-based hydrolysis, anyway a particular accuracy was required for the preparation of the NaOMe solution as described in the following procedure: molecular sieves were activated by drying at 160° in the oven for one night and left to reach the room temperature under a stream of nitrogen; MeOH (HPLC grade) was added to the molecular sieves always under nitrogen and left in this condition for two hours. Afterwards, the weighted amount of metallic sodium to reach 22 mM required for the reaction was added with care to the calculated volume of MeOH, keeping all materials under nitrogen. Fresh reagent in solution was used each time.
Table S1 describes the conditions used and the results of the cardiolipin transesterification into fatty acid methyl esters (FAME), evidencing the formation of free fatty acids as products of the hydrolysis in the experimental conditions. Table 2 describes the results using the NaOMe protocol, avoiding formation of free fatty acids.  The reaction mixture was first treated with hexane (3×2 mL) containing 0.1 mL of a standard solution of 17:0 methyl ester, to extract FAMEs, than the combined organic phases were evaporated under vacuum, the residue dissolved in 1 mL of nhexane and injected in GC (the GC areas were adjusted using a correction factor obtained by the recovery of the 17:0 standard).

Characterization of the reaction mix containing mono-trans cardiolipin
The reaction mix obtained after 4 min photolysis and work up as previously described was separated into two quantities: one for NMR spectra and the other one treated by GC transesterification for the identification of the cis and trans isomers of linoleic acid, as previously described.

S10
As detailed in the main text: For 9cis,12trans-18:2 isomer the literature reported a 0.7 ppm more deshielded resonance, 1 therefore it is possible to attribute the peak at 131.7 ppm (see Figure S6 for details) to the C-13 of a 9cis,12trans-18:2 CL isomer. By analogy with the CL structure, the signal at 131.6 ppm can be attributed to C-9 in the same molecule. Probably the alkenyl resonances of the other three fatty acid chains of this monotrans CL remain in the same positions than in the natural cis CL. Examining the resonances at 131.3 and 131.2 ppm again it is possible to attribute to C-9 and C-13 of a second mono-trans CL isomer, by analogy with the 9trans,12cis-18:2 data. 1 For the signals at 129.7 and 129.6, 128.8 and 128.7 ppm, the attribution of the less deshielded peak at 128.7 ppm is the C-10 of the 9trans,12cis-18:2 accompanied by the C-12 at 128.8 ppm in the same chain, whereas the resonances at 129.7 and 129.6 ppm individuate the 9cis,12trans-18:2 CL isomer. Figure S9. Enlargement of the 13 C NMR spectrum in CD3OD evidencing the region of the alkyl carbon atoms with the presence of the C-11 (bisallylic carbon atom) of the cis and mono-trans linoleic acid structures, which presents a consistent shift of the resonance going from cis to trans geometrical isomer, as described in the literature 5 .

Ag-complex of lipids and 1 H NMR spectra
The use of silver-TLC is well known to separate cis and trans alkenes and was used to separate cis and mono-trans fatty acid isomers in previous works. 2 After preparative Ag-TLC chromatography of the reaction mixture of CL isomerization carried out for 4 min under photolysis conditions, the separation of the Ag-monotransCL complex was achieved. The comparison of the 1 H NMR spectra is provided here below.