Oxylipin analysis methods


  • Martin J. Mueller,

    1. Julius-von-Sachs-Institute for Biosciences, Pharmaceutical Biology, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany
    Search for more papers by this author
    • These authors contributed equally.

  • Laurent Mène-Saffrané,

    1. Gene Expression Laboratory, Plant Molecular Biology, University of Lausanne, Biology Building, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
    • These authors contributed equally.

  • Christoph Grun,

    1. Julius-von-Sachs-Institute for Biosciences, Pharmaceutical Biology, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany
    Search for more papers by this author
  • Kathrin Karg,

    1. Julius-von-Sachs-Institute for Biosciences, Pharmaceutical Biology, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany
    Search for more papers by this author
  • Edward E. Farmer

    Corresponding author
    1. Gene Expression Laboratory, Plant Molecular Biology, University of Lausanne, Biology Building, CH-1015 Lausanne, Switzerland
    Search for more papers by this author

*(fax +0041 21 692 4195; e-mail edward.farmer@unil.ch).


Practical guidelines for monitoring and measuring compounds such as jasmonates, ketols, ketodi(tri)enes and hydroxy-fatty acids as well as detecting the presence of novel oxylipins are presented. Additionally, a protocol for the penetrant analysis of non-enzymatic lipid oxidation is described. Each of the methods, which employ gas chromatography/mass spectrometry, can be applied without specialist knowledge or recourse to the latest analytical instrumentation. Additional information on oxylipin quantification and novel protocols for preparing oxygen isotope-labelled internal standards are provided. Four developing areas of research are identified: (i) profiling of the unbound cellular pools of oxylipins; (ii) profiling of esterified oxylipins and/or monitoring of their release from parent lipids; (iii) monitoring of non-enzymatic lipid oxidation; (iv) analysis of unstable and reactive oxylipins. The methods and protocols presented herein are designed to give technical insights into the first three areas and to provide a platform from which to enter the fourth area.


Oxylipin analysis is a powerful and under-utilized approach with which to probe cell physiology. The goal of this work was to provide straightforward protocols suitable for use in non-specialized laboratories. The methods presented are all based on gas chromatography (GC)/mass spectrometry (MS) and take advantage of the two basic types of detection: electron ionization and chemical ionization. GC is not the best analytical method for all oxylipins as some, although stable in the fridge, degrade or rearrange when analysed at high temperatures, causing losses and/or the artifactual generation of other oxylipins. Good examples of this are fatty acid divinyl ethers. In some cases these problems are greatly reduced by using short GC columns rather than the more standard long columns. However, compounds such as fatty acid (FA) hydroperoxides and some other low-stability oxylipins are often better analysed by liquid chromatography (LC). In addition, many large or polar oxylipins (e.g. oxylipin glycosides) are difficult to fractionate with GC and are better separated by LC. However, for the majority of oxylipins GC is the most powerful platform.

There are many protocols available for oxylipin analysis. We have selected two. The first protocol, that of Schmelz et al. (2003, 2004), is based on a metabolomic approach whereby hundreds or thousands of volatilized molecules are first extracted. This mixture is separated by GC. The user then selects the molecules of interest for quantification. This method is excellent for the routine measurement of many well-known and abundant oxylipins. Several phytohormones can be measured simultaneously. The second protocol, from Weber et al. (1997), is different in that it is dedicated to the extraction and analysis of oxylipins. Fewer molecules are extracted for GC separation. This means that the background is reduced; low-abundance oxylipins often migrate as individual and distinct peaks, and the method has permitted the discovery of several novel oxylipins. In a third section of the paper, a highly target-oriented method is described for a specific class of oxylipins: hydroxy fatty acids (OH-FAs). These molecules can be used as markers to quantify oxidative stress.

An important issue is chemical derivatization. Whenever possible, the simplest protocols with the minimum possible chemical derivatization of oxylipins are provided. In most cases (e.g. the analysis of jasmonic acid and 12-oxo-phytodienoic acid) it is sufficient to modify carboxyl groups on the molecules using one of two simple chemical reactions provided herein. Many hydroxylated molecules (e.g. trihydroxy-fatty acids) need a second type of derivatization to increase their volatility. In the presentation of the three protocols, other published oxylipin methods (e.g. Montillet et al., 2004; Müller et al., 2002; Stenzel et al., 2003) have been useful sources of additional information. Prior to describing the methods, some discussion of sample preparation is provided and attention is drawn to several common pitfalls in oxylipin analysis. Here, oxylipins are defined simply as oxygenated fatty acids (FAs).

Getting started

Basic versatile equipment

The methods described here require a low-polarity capillary gas chromatograph column such as the Hewlett Packard 5MS (30 m, 0.25-mm internal bore; Agilent, Wilmington, DE, USA). For some applications, particularly if the compounds to be analysed are unstable (e.g. epoxy alcohols), shorter GC columns 12 m in length can be useful. For the mass spectrometer, a simple quadrupole detector is sufficient. For redistilling solvents, extracting tissues and carrying out derivatization reactions you need access to a well-ventilated chemical hood with a gas line carrying nitrogen. This gas line is used to evaporate solvents during extraction and derivatization. A heating block facilitates solvent evaporation. Some other equipment specific to each protocol we discuss is indicated within each protocol. Although not necessary for the protocols described herein, a simple high-performance liquid chromatograph equipped with an ultraviolet light detector provides a useful tool that is complementary to gas chromatography/mass spectrometry (GC-MS, see Box 1 for glossary) and can be used for purifying certain internal standards, unstable oxylipins etc. Note that wherever possible glass pipettes and vessels should be used for handling and storing oxylipins. Plastics usually contain plasticizing agents (such as phthalates) that almost invariably complicate analysis. Worse, plasticizers can contaminate oxylipin samples used for biological activity tests.

Sample preparation: optional de-esterification

Many publications contain data on oxylipins that are assumed to be ‘free’, i.e. not esterified into cell membranes. These pools are dynamic and particularly responsive to stress. However, many if not most cellular oxylipins are esterified into cell membranes and it is important that researchers decide if the liberation of these esterified oxylipins is desirable. If it is, the following protocol for de-esterification of oxlipins/FAs from membrane preparations or tissues (including seeds) can be used. FA hydroperoxides are first reduced to their hydroxides with trimethyl phosphite (trimethyl phosphine). When combined with a good profiling method this step can provide insights into oxylipin pools to the sub-organelle level. Researchers can also let insects ingest leaves. In this case, as we report herein, the insect does the de-esterification and free oxylipins can be recovered in its regurgitant.

A de-esterification protocol

Start with 0.25 g of tissue ground as a frozen powder in liquid nitrogen or 0.1 g of membranes. Suspend in 4 ml of MeOH and 2 ml of CHCl3 in a Pyrex tube and add 12.5 μl of trimethyl phosphite. Rotate for 2 h at 4°C and then add 3 ml of H2O and 1 ml of CHCl3. Centrifuge at 1000 g for 15 sec and recover the lower phase. Dry down with N2 gas at 40°C.

Resuspend immediately in 200 μl of MeOH/CHCl3 (2:1). Add 5 ml of MeOH followed by 150 μl of H2O. Now add 0.5 g of K2CO3 and shake (most of the salt will not dissolve). Incubate at 60°C for 1 h in a sealed vial in which air has been purged with nitrogen (N2) gas. Remove excess K2CO3 by filtration then add 2 ml of H2O to the filtrate. Slowly adjust the pH to 8–9 with 4 m hydrochloric acid. Dry down in a vacuum centrifuge for the Schmelz et al. protocol or use directly in the Weber et al. protocol. Note that there will be a large excess of free FAs and oxylipins relative to methods where de-esterification is not employed. This means that less sample will be required for processing. Alternatively, doubling the volume of the first column in the Weber et al. protocol will ensure that the first column is not overloaded.

Handling, storage and weighing issues

When frozen powders of plant tissues are generated, handled and stored they should not be allowed to thaw, even momentarily, prior to use as this will permit lipase activity, causing the unwanted de-esterification of esterified oxylipin pools. Another problem can arise if plant samples or crude extracts are stored prior to use: non-enzymatic oxidation leading to the production of artefacts. The problem of lipase activity even in organic solvents is ever present, so the user needs to bear this in mind. Ideally, fresh samples should be extracted and worked up immediately, but in practice plant samples will often need to be stored. Samples should be purged with N2 gas and stored at as low a temperature as possible (liquid N2 is best; otherwise at −80°C). Once isolated as pure molecules or as complex mixtures, oxylipins are best stored in ethyl acetate.

A word about internal standards

Internal standards allow quantification. They should be as closely related structurally to the compound being quantified as possible and have similar physicochemical properties and chemical stability. The standard has to mimic the analyte throughout the analysis. Firstly, there are internal standards that cannot be produced naturally in plants: for example, identical molecules containing one or more heavy isotopes such as deuterium, 13C or 18O. These are clearly the best option and should be used whenever possible. Secondly, there are those standards that might theoretically be produced but that have not been detectable in control experiments: for example, oxylipins where a single or multiple double bonds have been reduced, thus changing the mass of the compounds or model oxylipins with carbon skeleton lengths not found in plant samples. These standards are often easy to synthesize but their absence in the sample to be analysed must always be verified. A case in point is dihydrojasmonic acid (dJA). This can be readily prepared by platinum-catalyzed reduction of jasmonic acid (JA; see the protocol for hydrogenating FAs in the non-enzymatic oxidation section). However, because it can also occur naturally in some plants (e.g. cereal seeds; see Liechti and Farmer, 2002: the jasmonate biochemical pathway connections map; http://stke.sciencemag.org/cgi/cm/CMP_7361), care must be taken when such compounds are used as standards in new situations. Below, Box 2 describes the preparation of oxygen isotope-labelled standards while Box 3 explains how internal standards are used.

The ionization method

At this point, it is necessary to raise an important point concerning the detection method used. There are two common modes for ionizing molecules in the ion source of the MS: electron ionization (EI) and chemical ionization (CI). Electron ionization yields a positively charged molecular ion radical [M·]+ that is prone to undergo extensive fragmentation reactions in the ion source which may dramatically decrease the intensity of the molecular ion and generate diagnostic high-molecular-mass fragments. These yield ‘molecular fingerprints’; fragment ions help to identify the compound of interest and can also be used for quantification. Because of the high-resolution power and specificity of GC-MS run in the EI mode, this method is still the gold standard in compound identification which has not been surpassed by any other analytical method. EI is the mode of choice for the Weber et al. (1997) method. Table S1 presents parent ion and fragment masses for some of commonly encountered oxylipins. Note that many compounds elute as two peaks from the GC. This is the case for JA [JA, oxo(pentenyl)cyclopentanes (OPCs), 12-oxo-phytodienoic acid (OPDA) and dinor OPDA; see Table S1 for more details].

Chemical ionization is especially valuable for high-sensitivity target compound analysis in a very complex background of molecules such as that often generated with the Schmelz et al. (2003) method. During chemical ionization, oxylipins (as well as other analytes) become protonated at one of the oxygen atoms. The proton donor is a reagent gas. To start with, we recommend methane for CI. For softer ionization, experienced users may prefer other gases (isobutane or ammonia).

Because under CI conditions fragmentation is suppressed, only the [M + H]+ (or one of the fragment ions) can be used for identification and quantification. The intensity of the quasi molecular ion can be more than 10 times higher than that of the molecular ion in the EI mode, while simultaneously the background noise level is dramatically reduced. We recommend positive CI for quantitative target analysis, especially when the modified Schmelz protocol is used. In this case, the GC uses helium as the mobile phase and, for example, methane for ionization. Note that CI is very useful for determining the molecular masses of unknown compounds which can easily be obtained by comparison of GC-MS runs recorded under CI and EI conditions.

The highest detector sensitivity, however, can be achieved with another technique related to CI; the negative ion CI mode (NCI). The sensitivity obtained with NCI can be up to three orders of magnitude higher than that obtained with the EI mode. Sensitivity and selectivity become important issues when either the target compound is rare or the tissue sample is small. A major drawback of NCI is that samples require a special derivatization that favours the formation of anions in the ion source of the MS. Notably, anion formation is not only a highly efficient but also a highly selective process, as predominately derivatised compounds will form anions, thereby greatly reducing the background noise level. As with positive CI, molecular mass information can easily be obtained but structural information provided by fragment ions is lost. The high selectivity of NCI is also of great advantage when non-selective sample preparation procedures are used (e.g. the Schmelz et al. protocol).

In summary, for routine oxylipin analyses we recommend the Schmelz et al. (2003) protocol with positive CI. The Weber et al. (1997) protocol can be used with any form of detection but is usually coupled to electron ionization.

Box 1 – Abbreviations and glossary


N,O-bis(trimethylsilyl)-trifluoroacetamide (sylilating agent, TMS donor)


dihydrojasmonic acid


dinor 12-oxo-phytodienoic acid


fatty acid


gas chromatography


high-performance liquid chromatography


internal standard


jasmonic acid


liquid chromatography (HPLC in this article)




malondialdehyde (breakdown product of certain oxidized lipids)


mass spectrometry


oxo(pentenyl)cyclopentanes (precursors of JA derived from OPDA)


12-oxo-phytodienoic acid


salicylic acid


target compound


trimethylsilyl group (protecting group for hydroxyl functions)


pentafluorobenzyl group (protecting group for carboxyl functions)


chemical ionization: the process whereby new ionized species are formed when a reactant gas introduced into the ion source interacts with ions corresponding to the molecules being analysed. The process may involve the transfer of an electron, a proton, or other charged species between the reactants. When a positive ion results, the term CI may be used without qualification. When a negative ion results, the term negative ion CI (NCI) should be used. Many oxylipins analysed with CI and NCI do not fragment so that single ions corresponding to [M + H]+ (CI) or [M-PFB] (NCI) are detected


electron ionization: this term is used to describe ionization of molecules by electrons accelerated between 50 and 150 eV (where ‘eV’ is the energy acquired by a particle containing 1 unit of charge through a potential difference of 1 V). Usually 70 eV electrons are used to produce positive ions. The largest ion produced in EI corresponds to the positively charged parent molecule and is referred to as ‘M+’. Quite often the abundance of this ion is too low for easy detection, whereas smaller fragment ions are seen. Examples are given in Table S1


molecule or molecular ion containing a single isotope of each element: the exact mass of the ion is calculated by adding the masses of the lightest isotopes of all elements contributing to the molecule (i.e. 12C = 12.00000, 1H = 1.00782, 16O = 15.99492 and 28Si = 27.97693). Because of the low resolution of most MS instruments and for convenience, generally only the nominal masses are indicated. Nominal masses are calculated by simply adding the integer masses of the lightest isotopes of all elements (i.e. 12C = 12, 1H = 1, 16O = 16 and 28Si = 28). In contrast, molecular weights (as specified in catalogues) represent masses calculated by adding atomic weights (i.e. C = 12.0107, H = 1.00794, O = 15.9994 and Si = 28.0855) of all elements


mass, the unified atomic mass unit (u), or Dalton (Da), is used to express atomic and molecular masses. It is defined to be 1/12 of the mass of one atom of 12C


mass/charge ratio (dimensionless): ions are separated by the MS according to their m/z ratios. Because the charge (z) of ions generated from small molecules (such as oxylipins) in the ion source is generally 1, the observed m/z ratio is usually identical to the ion mass


selected ion current monitoring: ion current of all ions with a specified m/z ratio


total ion current: sum of all the separate ion currents carried by the different ions contributing to the mass spectrum. For quantification, selected ion currents are usually monitored

Response factor

In this article, the response factor is a correction factor used for quantification based on peak area ratios of single ions from the internal standard and the target oxylipin. The same quantities of internal standard and target oxylipin may yield different ion peak areas (detector responses) after injection into the GC-MS system. The response factor is the ratio of the response of the internal standard and the response of the target oxylipin

E, Z

designation for naming double bond isomers. Double bonds have two groups at each end. Their priority is based on the atomic number of the atom attached to the double bond. Double bonds with higher priority groups on the same side of the bond are Z (from the German zusammen = together). When the higher priority groups at each end of the bond are on opposite sides they are E (from the German entgegen = opposite). In all examples in this article, E corresponds to trans and Z to cis. Note that E and Z isomers of oxylipins are often easy to separate by gas chromatography. Occasionally, Z double bonds form E double bonds as a result of heating in the gas chromatograph

R, S

R/S notation for enantiomers (a type of stereoisomer). To apply this notation, identify the stereogenic centres (most commonly a ‘chiral’ C atom with four different groups attached). Assign the priority to each group (high = 1, low = 4) based on the atomic number of the atom attached to the stereogenic centre. Position the lowest priority group away from you. If you are using a model, grasp this group in your fist. For the other three groups, determine the direction of high to low priority (1 to 3) If this is clockwise, then the centre is R (from the Latin rectus = right); if this is counter-clockwise, then it is S (from the Latin sinister = left). With the methods described herein, R and S enantiomers of oxylipins cannot be separated or distinguished. However, isomers with more than one chiral carbon can potentially be separated when one of the stereogenic centres in each isomer has the same R/S notation (as for JA and OPDA which both have two chiral carbons). Furthermore, some groups may epimerize in the gas chromatograph, changing R to S or vice versa

Box 2 – How to prepare heavy oxygen-labelled internal standards

The availability of good internal standards for oxylipin quantification (Box 3) can be a limitation, and simple, general methods for their production are needed. Fortunately, most oxylipins have a carboxyl group in which both naturally occurring oxygen-16 atoms can be exchanged against oxygen-18, converting them to molecules that will be four mass units heavier. Oxygen-18 labelled standards are often easy to make and are almost ideal standards, but they must be handled with some special precautions, as described below. To the best of our knowledge, oxygen-18 labelled standards have not been used in the plant oxylipin field although they have been utilized elsewhere for the quantification of several types of molecule (Hood et al., 2005; Murphy and Clay, 1990).

Below are two simple protocols optimized for the preparation of oxygen-18 labelled standards on the milligram scale. For these protocols, H218O (>98 atom %18O from Isotec, Miamisburg, OH, USA) is used as the source of oxygen-18. While each method has its limitations, most oxylipins can be labelled with at least one of the protocols. The oxygen exchange mechanism for the two methods (and for an additional method presented as supplementary material) is shown in Figure S1. Formulas to calculate the theoretical incorporation of oxygen-18 using the three methods can also be found in the legend to Figure S1.

I. Acid-catalyzed oxygen-18 exchange

Principle and uses.  Both oxygen-16 atoms in the carboxyl groups of oxylipins are remarkably resistant to exchange with oxygen-18 in [18O]H2O under neutral and alkaline conditions. Even in water at pH > 2, no significant exchange takes place within at least 5 h. However, the oxygens can be readily exchanged under strong acidic conditions in 4 m HCl. In addition, all oxylipins with keto groups (e.g. JA and OPDA) exchange the oxygen in the keto group far more readily than in the carboxyl group yielding triply labeled compounds. In this case, the keto group label should be re-exchanged to yield stable doubly labeled standards (see the Web supplement for a protocol). Acid-labile oxylipins such as hydroxylated and epoxy fatty acids (i.e. phytoprostanes, hydroxy-FAs, etc.) are dehydrated or hydrolysed under acidic conditions and cannot be labelled this way. An advantage of this simple protocol is that mixtures of chemically similar compounds can be labeled in one pot. This method is good for labelling JA starting from MeJA.

Protocol.  Dissolve the water-free unlabeled compound -(up to 0.5 mg) in 50 μL of 4 M HC1 in 1,4-dioxane (premade solution, Aldrich Chemicals) and 25 μL of H218O. After an incubation period of 2 to 24 h at 40°C to 90°C (examples can be found in the supplement), the sample is dried under vacuum. If the compound to be labelled contains a keto group (e.g. MeJA) the oxygen-1 8 in this group needs to be back-exchanged with oxygen-16 (see Web supplement). If not, the labelled oxylipin is immediately dissolved in anhydrous acetonitrile for storage. Oxygen-18 incorporation is determined by GC-MS. In a separate GC-MS run, the doubly labelled compound is quantified against a known amount of unlabelled compound.

Note.  Under equilibrium conditions in [18O]water (98 atom % oxygen-18) 96.04% of the compound remains unlabelled, 3.92% is singly labelled and 96.04% is doubly labelled. Thus, the internal standard will contain a very small amount of unlabelled target compound which is, however, below the limit of detection. Strong acids such as salicylic acid may need much longer incubation times, higher temperatures or stronger acids to reach the calculated equilibrium. For further details, see the material associated with Figure S1.

II. Enzyme-catalyzed oxygen-18 exchange

Principle and uses.  Serine hydrolases catalyse ester (and amide) hydrolysis through a mechanism involving transesterification of the acid moiety (oxylipin) to a serine residue in the enzyme active site. This covalent intermediate is subsequently hydrolysed by water in a reaction that is, importantly, reversible, i.e. the enzyme may bind free acids to the active-site serine and this covalent intermediate is rapidly hydrolysed by water again. In the presence of [18O]water, both oxygen-16 atoms of the carboxyl group are exchanged against oxygen-18. The protocol presented below is based on inexpensive enzyme preparations which work well in organic solvents. Here, two different broad-specificity enzyme preparations can be used; this allows use with many oxylipins. In our hands, the hydrolases selected have worked well with JA, OPDA, FAs, hydroxy FAs, and all types of phytoprostanes.

Protocol.  Dissolve the unlabeled compound (up to 0.1 mg) in 50 μL of anhydrous ethylene glycol (Aldrich Chemicals). Either of two enzyme preparations can be used: (Lipase II, crude from porcine pancreas, 100 to 400 units/mg, Sigma, Buchs, Switzerland) or liver esterases (Esterase, crude from porcine liver, approx. 20 units/mg, Sigma). The enzyme (about 1 mg) is suspended in the solvent and the mixture is sonicated for one minute to disperse the enzyme before [18O]water (25 μL) containing buffer salts (dried solution of 2 μL 150 mM CaCl2 in 1 M Tris-HCl buffer pH 7.4) is added. After an incubation period of 5 to 24 h at 40°C, the sample is extracted with 1 mL of hexane/diethyl ether (2:1, v/v) containing 10 μL of acetic acid. The organic phase is dried under a stream of nitrogen, dissolved in an anhydrous solvent (i.e. acetonitrile), and stored under the same conditions as the unlabeled compound.

Note.  The enzyme catalyses an equilibrium reaction similar to the acid-catalyzed reaction, and thus the theoretical oxygen-18 atom percentage incorporation is the same as described above. In the case of OPDA, the incubation period should not exceed 12 h as the oxygen in the keto group may become partially labelled in an uncatalyzed fashion (in this case triply labelled oxylipin is obtained; see the discussion in the material associated with Figure S1).

Essential precautions

Oxygen-18 labelled standards are generally stable but they should not be stored for prolonged periods (i.e. several days) in aqueous solutions and are best stored in anhydrous solvents. In the presence of plant extracts, however, a serious drawback is the very real possibility of back-exchange of the oxygen-18 label catalyzed by plant lipases and esterases (Pollard and Ohlrogge, 1999) by which the internal standard can be converted into the target oxylipin. Some plant lipases may be very resistant to denaturation and are even active in organic solvents. Oxygen-18 labelled standards should never be added to plant material or extracts directly. It is first recommended to extract the plant material with methanol. The extract is then incubated for 5 min with phenylmethanesulphonyl fluoride (PMSF, dissolved in isopropanol; 5 mm final concentration) in order to inactivate serine hydrolases irreversibly (note: PMSF is highly toxic!). After removal of insoluble material by centrifugation, the extract is dried and the internal standard is added. Alternatively, plant material can be boiled in a microwave oven for a few seconds prior to addition of the internal standard. Although there should be no internal standard breakdown when following these guidelines, a [18O]C17:0 fatty acid can be used as an internal control standard. The internal control standard is added together with the internal standard to the plant extract. The integrity of the control standard after the work-up procedure can be checked by GC-MS. Finally, the oxylipin sample that is injected into the GC must not contain any alcohols as considerable transesterification takes place in the injector port.

Box 3 – The Quantification of Oxylipins

The quantification of oxylipins (target compounds, TCs) in biological samples always requires an internal standard (IS, discussed in Box 2 and in the text) to correct for losses occurring during sample preparation. The following text uses the measure of JA in plant tissues to illustrate how the strategy used for quantification depends on whether or not the IS is an ideal chemical mimic of the TC. In the first of two examples the IS used is considered to be ideal. The best option is always to use an isotopically labelled version of the TC. New oxygen-18 ISs (described in Box 2) would be considered near ideal ISs and could be used in place of the carbon-13 standard used in the first example.

For JA quantification, [1,2-13C]JA, obtained from chemical synthesis (Müller et al., 2002), is two mass units heavier than JA and can be used as an IS. This two mass unit difference is, of course, conserved after methylation. This IS is an ideal mimic of JA as it cannot be separated from JA through conventional sample purification and chromatographic procedures. Only the mass spectrometer is able to physically separate and detect light MeJA (TC) and heavy MeJA (IS). For identification and quantification, molecular ions or fragment ions that are unique to both compounds are first selected. In the CI mode, for instance, the [M + H]+ ion of MeJA is detected at m/z = 225 and that of [1,2-13C]MeJA at m/z = 227. These ions are monitored and appear as co-migrating peaks in ion current chromatograms. The ion peak area ratio MeJA/[1,2-13C]MeJA is thus obtained. As [1,2-13C]JA is a perfect mimic of JA throughout the protocol from the plant sample to the MS detector, all losses are quantitatively the same for light and heavy JA. In other words, the molar ratio of JA to [1,2-13C]JA in the plant sample is the same as the peak area ratio of m/z = 225 to m/z = 227. As the amount of [1,2-13C]JA added to the plant sample prior to extraction is known, we can now calculate the amount of JA in the sample from the peak area ratio. Importantly, the absolute recovery of [1,2-13C] JA or JA is not of interest as long as sufficient molecules for accurate detection and quantification are recovered. However, the absolute recovery of JA through the purification method can be determined by processing samples with known amounts of JA. After processing, [1,2-13C]MeJA is added and the amount of recovered MeJA is determined by GC-MS. This information (loss of JA during sample preparation or absolute recovery) is useful to optimize extraction and purification protocols but not necessary for quantification.

Unfortunately, perfect ISs such as [1,2-13C]MeJA or [18O]2-JA are not always available and one sometimes has to use a chemically related compound that is less than ideal. For JA analysis, dihydro-JA (dJA) has often been used. dJA is easily prepared (see the experimental section) and is two mass units heavier than JA. In this case, however, additional experiments are mandatory to establish a quantification method because dJA is not a perfect (but still an appropriate) mimic of JA. At the beginning of the experiment, a known amount of dJA is added to the plant extract. Bit by bit, small amounts of this IS and endogenous JA will be lost during work-up, derivatization and GC-MS analysis. It cannot be assumed beforehand that all losses are quantitatively the same for the two chemically different compounds. Firstly, the detector response for dMeJA and MeJA may differ because of uneven losses in the GC injector as well as different ionization efficiencies and ion stabilities in the GC-MS system. Therefore, a correction factor, the response factor, has to be determined. Inject at least three times a 1:1 mixture of MeJA and dMeJA (100 pmol of each) into the GC-MS. In the CI mode, [M + H]+ ions are detected at m/z = 225 and 227, respectively. The peak area detected at m/z = 227 divided by the peak area at m/z = 225 is the response factor. Alternatively, the response factor can be derived from calibration curves by injecting a fixed amount of dMeJA together with increasing amounts of MeJA into the GC-MS system as shown in Figure B1.

Figure B1.

A calibration curve for comparing detector responses of a target compound (e.g. MeJA) and a non-ideal IS (e.g. dMeJA).

After injection of a 1:1 mixture of MeJA/dMeJA, the peak area ratio of MeJA/dMeJA is not 1 (as expected in the case of a perfect, isotopically labelled IS) but about 14/10 = 1.4. The response factor is therefore 1/1.4 = 0.714. This means that dMeJA is less sensitively (about 0.71-fold) detected than MeJA. When analysing mixtures with known amounts of dMeJA (e.g. 10 pmol) and unknown amounts of MeJA, the peak area ratio of MeJA/dMeJA is determined by GC-MS. If the peak area ratio is, for example, 21, the amount of MeJA can be calculated by multiplying the peak area ratio (= 21) by the response factor (= 0.71) and the amount of internal standard (10 pmol). Thus, the sample contained at the point of injection 21 × 0.71 × 10 pmol = 149 pmol MeJA. The response factor may depend on the operation conditions and the design of the GC-MS instrument used and should be checked regularly either by analysis of 1:1 mixtures of TC and IS or by establishing a new calibration curve.

The response factor, however, does not correct for uneven losses of MeJA and dJA during sample preparation before the time of injection (or derivatization). For instance, dMeJA has a slightly higher boiling point than MeJA, and thus potentially evaporates more slowly during sample drying, i.e. may become enriched during sample preparation. This possibility must be taken into account and the sample preparation method must be optimized with careful drying of samples to avoid MeJA loss by evaporation. In order to check that the method does not discriminate between dMeJA and MeJA, blank samples are spiked with dMeJA (i.e. 100 pmol) and increasing amounts of MeJA (i.e. 10–1000 pmol). Samples (in triplicate) are processed through the whole work-up procedure and measured by GC-MS. Amounts of MeJA are calculated from the peak area ratios of MeJA and dMeJA using the response factor. The method is correct and the dMeJA is an appropriate standard if the calculated amounts of MeJA are equal to the amounts of MeJA added to the blank sample. From the calibration curves obtained in this way, additional important information can be derived, such as the lower and upper limits of quantification and the precision of the method. Note that in real samples contaminants may increase the limit of quantification and reduce the precision of the method. If no trace of MeJA is observed it is ‘below the limit detection’. The precision of the method is also dependent on the ratio of dMeJA and MeJA and is optimal in the range of 0.1–10. Thus, the amount of IS added to samples should be in the range of the expected amount of the TC.

Finally, in the example given herein, CI is used for quantification. EI is also perfectly good and, while often less sensitive, allows the user a choice of fragment ions with which to quantify the oxylipin in question.


Whichever protocol is chosen, oyxlipins need to be made more volatile in order to separate them by GC. For EI analysis, carboxyl groups on oxylipins are converted to methyl esters in a one-step methylation procedure described below. Similarly, for CI analysis carboxyl group are methylated (positive CI) or esterified to pentaflurobenzyl (PFB) esters (NCI). In addition, other chemical derivatizations are often used. These are applied for two reasons. Firstly, they can make polar oxylipins more volatile so that they migrate on the CG column. Secondly, derivatization can be used as a diagnostic tool to determine which functional groups exist on a molecule (hydroxyl groups, carbonyl groups etc.). The most useful when using the EI mode of detection is the silylation of hydroxyl groups using reagents such as N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA; Pierce, Rockford, IL, USA). This does not affect the major jasmonates (e.g. JA, OPDA and OPCs) but will, of course, modify hydroxy-jasmonates. Chemical derivatization reactions can sometimes have unwanted side effects. In practice these are rarer with methylation and somewhat more common with silylation. A ‘blank’ derivatization on a sample containing no oxylipins is a useful control to include.

Derivatization protocols

Methylation (carboxyl group modification)

Compatibility: all protocols.  (Trimethylsilyl)diazomethane (2 m in diethyl ether (Et2O); Aldrich Chemical Company, Buchs, Switzerland) is a liquid that can be stored at room temperature. For use, the reagent is freshly diluted in methanol (1:20). The methanolic solution (20 μl) is added to the sample reconstituted in 250 μl of Et2O. After 5 min, excess reagent is destroyed by addition of 2 μl of acetic acid. Thereafter the sample is concentrated under a stream of N2. Note that some methyl esters with low boiling points, such as short-chain FAs, methyl salicylate and methyl jasmonate, may evaporate when the sample is brought to complete dryness. Methylation of carboxyls with (trimethylsily)diazomethane is generally fast and quantitative. Hydroxyls are not methylated except if they are acidic. For instance, phenols (e.g. salicylic acid) can be partially or completely methylated. In addition, α,β-unsaturated carbonyls (e.g. OPDA, abscisic acid, γ-ketols, keto-dienes and ketotrienes) may react when exposed to high reagent concentrations for prolonged periods. For these compounds, very brief (10-sec) methylations should be effected and diazomethane rapidly destroyed with acetic acid.

Silylation protocol (hydroxyl group modification)

Compatibility: use with all protocols given herein.  The sample is dissolved in 100 μl of chloroform. After addition of 50 μl of BSTFA the sample is heated at 40°C for 1 h. Thereafter, the sample is dried under a stream of N2. BSTFA has been most widely used for simultaneous silylation of carboxyl, hydroxyl and amino groups. However, many trimethylsilyl (TMS) derivatives are sensitive to hydrolysis and degradation in the presence of water or polar contaminants during the collection of volatiles (a second derivatization with BSTFA after collection of volatiles often greatly improves the recovery of analytes). It is to be expected that silylation with N-methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide (MTBSTFA) might be more compatible with the volatilization protocol as the tert-butyldimethylsilyl derivatives are more stable (Birkemeyer et al., 2003).

To summarize, for most oxylipins methylation is sufficient. Hydroxylated oxylipins require further derivatization either by silylation or acetylation. For the first protocol we describe, that of Schmelz et al. (2003), acetylation (described below) is perhaps most useful. For the Weber et al. (1997) protocol silylation is commonly used.

General tricks of the trade for oxylipin analysis

When MS is used in the EI mode, parent ions have masses with even numbers if molecules are composed of carbon, hydrogen and oxygen. Molecules containing a nitrogen atom yield parent ions of uneven mass units. Quite often, pairs of related compounds are found and this almost always provides useful information. Oxylipins may, for example, be derived from linoleic acid (18:2) and linolenic acid (18:3). In the case of plants containing 16:3 FA (Mongrand et al., 1998), oxylipin pairs may originate from 16:3 and 18:3 (e.g. dinor OPDA and OPDA). In these cases, compounds are usually of chloroplast origin. Other paired peaks may reflect various cis and trans epimerizations and isomerizations.

When comparing two samples, for example wild-type and mutant, never base results on a single analysis. Always perform biologically independent replicates. One must always bear in mind that increases in recovered oxylipins may be attributable to increased synthesis, reduced turnover, increased lipase activity or combinations of these. Also, large differences in the levels of oxylipins can occur between organs and between species; this can affect the amount of starting material required. The general health of plants can greatly affect oxylipin levels: control plants should be maintained correctly. When working with pathogens it is a good idea to collect an oxylipin signature of the microorganism grown in culture to investigate whether it might produce molecules of interest found in the plant. Note that the column lifetime is decreased if large quantities of low-purity samples are injected. If in doubt, first inject highly diluted samples. The older the GC column, the less sensitive the detection.

Two protocols for oxylipin analysis

Protocol 1. Rapid and efficient oxylipin analysis: the modified Schmelz et al. protocol

The principle of the method.  The Schmelz et al. (2003) method is not a specialized oxylipin analysis procedure, and oxylipins are embedded in a pool of other compounds. The revolutionary sample preparation method removes all non-volatile materials before injection into the GC by cleverly trapping them, leaving behind a complex mixture of non-volatile compounds that might otherwise interfere with analysis. The original protocol from Schmelz et al. (2003) was designed to capture volatiles plus other metabolites. We have modified the protocol in order to extract a broader range of metabolites and to speed up the sample preparation. However, natural volatiles are now lost. To use the modified Schmelz et al. (2003) protocol, a small and simple device for trapping volatile molecules as described in Engelberth et al. (2003) is necessary. A vibrating ball mill and some ceramic beads, all described in Engelberth et al. (2003), must be purchased.

In this method, a crude methanolic plant extract is first dried and then derivatized. Heating the mixture under a stream of N2 allows the capture of the entire spectrum of metabolites which can be analysed by GC. These compounds are trapped on a simple homemade filter device containing the adsorbent Super Q as described in Engelberth et al. (2003). The trapped analytes can then be eluted with a small volume of an organic solvent and analysed directly by GC. The great advantage of the protocol is that it requires no special expertise and samples are ready for analysis within 2 h. Moreover, it is a general method that can be used for a broad spectrum of metabolites.

The basic protocol (modified Schmelz protocol)

Step 1: extraction. Plant tissue (100 mg, stored at −80°C) is mixed with 1 ml of methanol in a 2-ml vial. Internal standards, 50 mg of polystyrene-supported triphenylphosphine (1.4–2 mmol g−1; Lancaster Synthesis, Windham, NH, USA) and 50 μl of acetic acid are added. After addition of a ceramic bead (6 mm diameter) the tissue is homogenized and extracted using a vibrating ball mill for 3 min. Centrifugation is carried out at 1000 g for 1 min, and the supernatant is transferred to a 2-ml glass vial and dried in a vacuum centrifuge (Speed-Vac, Kleinen, Wohlen, Switzerland). The residue must be completely dry (no remaining acetic acid) before proceeding to the derivatization step. During drying some volatiles will be lost.

Step 2: derivatization. Several derivatization protocols can now be applied to the sample. If high-abundance metabolites are to be analysed, samples are methylated and analysed in the EI mode. Most users, however, will employ the positive CI mode. However, when the amount of sample is limiting or the analyte is rare, NCI analysis should be used after PFB bromide derivatization (see below). Whichever detection mode you use, you can employ an optional acetylation to modify hydroxyl and amino groups.

Step 3: sample purification (vapour phase extraction). The sample dissolved in diethyl ether (Et2O) is pipetted into a 2–4-ml glass vial. The vial has a GC septum which is held in place by an open-top metal screw cap that must be very tight. The septum is penetrated with two luer-lock needles (the inlet and the outlet needles, respectively) which are not inserted into the solution. A 1-ml glass solid-phase extraction column (luer tip, glass fibre frits) is packed with 250 mg of Super Q 80/100 (Alltech Associates Inc., Lexington, KT, USA) material and connected with the luer lock of the outlet needle. The inlet needle is connected with a source of inert gas (helium or nitrogen) adjusted to 100 ml min−1. As the packed column does not generate a significant back pressure, a vacuum line connected to the open end of the adsorbent column (as proposed in the original method) is not required. The solvent of the sample is evaporated until the vial is dry. The vial may be heated to speed up the evaporation, but the solvent must not boil in order to prevent transfer of non-volatiles in the vapour to the adsorbent. Solvents will pass through the adsorbent while the analytes are trapped. After drying, the vial is inserted into a heating block (250°C) and volatilized molecules are collected for an additional 3 min. Thereafter, the adsorbent column is removed from the vial and eluted with 500 μl of Et2O. The eluate is collected in a high-performance liquid chromatography (HPLC) vial with a conical insert and concentrated to about 5 μl under a stream of N2. A volume of 2 μl is injected into the GC-MS.

The GC programme.

A shallow GC temperature programme that starts at a low temperature is used. Metabolites are analysed on an HP-5MS column (30 m × 0.25 mm; Milian, Geneva, Switzerland), using a linear helium flow at 23 cm sec−1 and a column temperature step gradient starting at 80°C for 3 min; 80–300°C at 10°C min−1; 300°C for 5 min. The injector of the splitless mode is set at 280°C.

Step 4: analysis. This method traps hundreds of compounds, many of which co-elute from the GC column. Mixed mass spectra are therefore usually obtained. In the EI mode, characteristic ions (molecular ions and fragment ions) of the target molecules found in Table 1 or mass spectra libraries can be used for identification and quantification. In the CI mode, target molecules usually yield only one intensive ion: the [M + H]+ ion (add 1 mass unit to the molecular mass in the positive CI mode to calculate the mass of the ion). Usually, no or little fragmentation occurs; however, in the case of hydroxylated metabolites (either free or silylated), neutral loss of TMS-OH or H2O is a common fragmentation reaction which becomes predominant (the quasi molecular ion peaks [M + H]+ may not even be detectable). As characteristic ions can often be detected at different times in the chromatograms, it is essential to determine the correct retention times of all analytes with authentic reference compounds.

Table 1.  A complete list of hydroxy fatty acids formed by singlet oxygen-mediated, free radical-catalyzed and lipoxygenase-catalyzed oxidation of C18 fatty acids (FAs). The configuration of the oxygenated carbon is (R,S) in the case of the singlet oxygen- and radical-dependent products and exclusively (S) in the case of the plant LOX products
Oxygenation at carbon1O2RadicalLOX
 910(E), 12(Z),15(Z)10(E), 12(Z),15(Z)
10(E), 12(E),15(Z)
10(E), 12(Z),15(Z)

Alternative derivatization for the analysis of trace compounds: pentafluorobenzyl (PFB) esterification of carboxyl groups.  PFB derivatives are used for analysis in negative CI-MS. This protocol replaces methylation and is compatible with all sample purification procedures presented herein. The sample is dissolved in 200 μl of chloroform and 20 μl of N,N-diisopropylamine and derivatized for 60 min at 50°C with 10 μl of PFB-bromide. It is then dried under a stream of N2, reconstituted in 2 ml of hexane and loaded onto a silica solid-phase extraction column (500 mg). It is eluted immediately with 5 ml of Et2O. (The column can be re-used after washing with 5 ml of methanol, ether and hexane.) The sample is then concentrated to dryness under a stream of N2. The silica column is required to remove salts formed by the reaction of PFB-bromide with N,N-diisopropylamine. This step is not required when the sample is further purified by the volatilization protocol. Compared to the methyl esters, PFB esters have a higher boiling point and are not lost during drying. PFB esterification is sufficient when the target compounds have no free hydroxyl groups. Compounds with hydroxyls have to be silylated or acetylated. We recommend the acetylation protocol, which will acetylate hydroxyl and amino groups, for several reasons. The derivatization is highly efficient, is reproducible and covers a wide range of metabolites (sugars, amino acids and citric acid cycle intermediates). For acetylation, the sample is dissolved in pyridine/acetic anhydride [1:1, volume/volume (v/v)] and incubated for 1 h at 60°C. It is then dried down with nitrogen and the sample vapour extraction procedure is used. The PFB derivatives fragment in the ion source of the MS and yield intense [M-PFB] anion mass peaks (subtract 181 mass units from the oxylipin-PFB ester mass to calculate the fragment mass of the ion in the NCI mode).

The results: what to expect. Figure 1 shows what to expect when healthy Arabidopsis leaves are analysed with the modified Schmelz et al. (2003) protocol and analysed with positive CI. The three upper panels show selected ions corresponding to the methyl esters of three compounds (SA, JA and OPDA, see Box 1), while the lower trace shows the reconstructed total ion current from which these data were extracted. In a 100-mg leaf sample from Arabidopsis, basal levels of many phytohormones can easily be determined in the EI mode. However, because of the high background level, basal JA, abscisic acid and indolacetic acid levels are often close to or just below the limit of quantification. In the positive CI mode, the signal intensity is often 10-fold higher and generally sufficient to determine levels of SA, JA, OPDA, abscisic acid, indolacetic acid (Figure 1; see Table S1 for abbreviations) and FAs. The positive CI mode is also recommended to analyse broad metabolite profiles. After methylation and acetylation, most amino acids, several carbohydrates and citric acid cycle intermediates can be easily determined. Highest sensitivity can be reached in the NCI mode. One to five mg of tissue is generally sufficient to obtain high signal intensities for all acidic phytohormones (SA, JA, OPDA, etc.) and FAs. Often, it is necessary to inject the sample in both a concentrated and a diluted form as the dynamic range of the GC-MS is not sufficient to analyse low- and high-abundance metabolites simultaneously.

Figure 1.

Analysis of a methylated sample from a healthy 6-week-old Arabidopsis Columbia-0 (Col-0) leaf (100 mg).
The sample was prepared according to the modified Schmelz protocol. Detection was in the positive chemical ionization mode. The three upper panels show selected ion monitoring traces for methyl salicylate (SA, m/z = 153), methyl jasmonate (JA, m/z = 225), and methyl OPDA (m/z = 307). The total ion current trace over the entire mass range (m/z = 100–350) is shown in the lower panel.

Trouble-shooting.  The sample preparation procedure was found to be very robust to a variety of experimental conditions. There are two main problems associated with the non-selective sample preparation procedure. One is the presence of high-abundance (primary) metabolites and low-abundance trace metabolites (often oxylipins and phytohormones) which may be up to four orders of magnitude less abundant) in the same sample. Quantification (see Box 3) of trace metabolites may be complicated by contaminant peaks or the high noise level. In this case, a more selective sample preparation procedure may be necessary. Another potential problem with the Schmelz procedure is the artificial formation of metabolites from structurally related endogenous metabolites. Isomerization and dehydration reactions which may occur to some extent in the injector of the GC are more pronounced in the Schmelz protocol. For instance, cis-trans isomerization of the side chains at the cyclopentenone ring system of cis-JA or OPDA occurs almost quantitatively when using the Schmelz protocol. Thermal stress during sample preparation also enhances the formation of an isomer of OPDA with a double bond between the two side chains. Also, incompletely reduced FA hydroperoxides may produce an array of metabolites (including ketodienes and ketotrienes, epoxides and short-chain aldehydes). However, this may also occur when using conventional procedures.

Protocol 2. A route to finding novel oxylipins: Weber et al. (1997)

A specialist method for oxylipin profiling was developed by Weber et al. (1997). This protocol does not require one to choose which oxylipins one wishes to survey and has allowed the identification of several novel oxylipins as well as providing large amounts of quantitative data. Here, baseline separation of oxylipins is a great advantage in identifying and/or quantifying low-abundance compounds. However, the protocol requires larger tissue samples (0.5 to 1.0 g) and is time-consuming and thus best suited to low-throughput use. However, its continued use will probably lead to the discovery of new oxylipins, especially in signalling mutants.

The principle of the method.  Oxylipins at high pH are passed through a reverse-phase column, titrated to low pH, and then re-applied to the same column from which they are then eluted (after washing), derivatized, and applied to a silica column for further purification. At the start of the protocol, trimethyl phosphite is added to reduce FA hydroperoxides (OOH-FAs) to more stable FA hydroxides (OH-FAs). (Note: Et2O should be re-distilled before use to remove antioxidants.)

The basic protocol

Step 1: extraction. Start with 0.5–1.0 g of tissue. Add 3.5 ml of ice-cold methanol to a 15-ml Falcon tube, kept on ice. Then add 10 μl of an internal standard of your choice (e.g. 100 ng of dihydro-jasmonic acid and 100 ng of tetrahydro 12-oxo-phytodienoic acid in EtOH). Add 12.5 μl of trimethyl phosphite (toxic). Grind the tissue in liquid nitrogen to a fine powder and transfer the frozen powder to the Falcon tube and homogenize immediately with a Polytron (Kinematica, Luzern, Switzerland) on ice for 1 min. Extract by rotating on a rotary shaker for 2 h at 4°C and then add 1.5 ml of ice-cold water, mixing well. Keep for 5 min on ice, and then centrifuge for 15 min at 4°C and 3500 g. Transfer the supernatant to a new 15-ml Falcon tube on ice. Now add 90 μl of 1 m NH4OH (resulting pH should be 8–9), mix well, and keep on ice.

Step 2: first column. Purification is performed on reverse-phase extraction cartridges (Bakerbond SPE glass octadecyl 8 ml, 500 mg filling; J. T. Baker, Deventer, Holland; catalogue number 7334–06). First, condition the column with 10 ml of MeOH, then with 70% MeOH (10 ml). Pass the sample over the column and collect the run-through in a 50-ml measuring flask, and then elute the remaining oxylipins with 75% MeOH (7 ml) into the same measuring flask.

Add to the eluate 120 μl of 10% formic acid (resulting pH 3–4). Dilute with H2O to a final volume of 50 ml (the final MeOH concentration should be <20%). Keep on ice. Wash and condition the used column first with 6 ml MeOH containing 40 μl formic acid, then with 5 ml MeOH, then with 6 ml Et2O and finally with 5 ml MeOH followed by 2 × 7 ml H2O.

Step 3: first column, second pass . Load the sample on the column and wash with 15% EtOH in H2O (7 ml) then with H2O (7 ml). Next, elute the oxylipins into a 20-ml Pyrex tube with two successive washes each of 5 ml Et2O. Remove residual water on the bottom of the tube with a Pasteur pipette and then dry the organic phase by adding a ‘pinch’ of anhydrous MgSO4; mix well.

Centrifuge for 5 min, 15°C at 720 g. Transfer the supernatant (organic phase) to a new glass tube, wash MgSO4 with fresh Et2O (2 ml) and combine the organic phases. Evaporate Et2O with N2 at 40°C. The samples can now be derivatized. The dried sample should be processed immediately. Note that, after this step, samples should not come into contact with any plastic surfaces.

Step 4: methylation. Dissolve the oxylipins in 60 μl of MeOH, and add 200 μl of Et2O and 5 μl of (trimethylsilyl)diazomethane (a 2 m solution in Et2O). Vortex and incubate for 20 min at room temperature. Evaporate with N2. (Stop immediately after the solvent is evaporated. Beware loss of volatiles!)

Step 5: optional second derivatization. This derivatization is used to capture more polar oxylipins such as polyhydroxy FA. If the derivatization is not performed, load compounds directly onto the second column in 500 μl of hexane.

Dissolve oxylipins in 120 μl of acetonitrile, add 60 μl of BSTFA (Pierce, Rockford, IL, USA), vortex and then incubate for 30 min at 60°C. Evaporate with N2 (stop immediately after the solvent is evaporated!). Redissolve the sample in 500 μl of hexane, and vortex.

Step 6: Second column. Use a Bakerbond Florisil SPE glass 8-ml column, 500 mg packing (Baker, catalogue number 7420–06).

Condition the column with 3 ml of Et2O followed by 10 ml of hexane. Then load the sample on the column. Rinse the sample tube with another 500 μl of hexane and load this on the column. Wash the column with 1 ml of hexane and then another 4 ml of hexane. Then elute oxylipins with two washes of 3.5 ml of Et2O/ethyl acetate (1:1) into a glass tube. Concentrate the solvent under N2 to approximately 200 μl, transfer to the sample vial, rinse the tube with another 200 μl of Et2O and transfer to the sample vial. Evaporate the solvent completely with N2. Resuspend the sample in 10 μl of hexane for injection.

The GC programme.  FAs are analysed with the EI mode (70 eV electron potential; 11 psi column head pressure; HP-5MS column 30 m; internal bore 0.25 mm). Methylated samples that have not been derivatized further are separated with the following programme: 100°C for 1 min, 100–160°C at 20°C min−1, 160–238°C at 3°C min−1, and 238–300°C at 30°C min−1. Samples that have been both methylated and silylated with BSTFA are separated with the following programme: 100°C for 1 min, 100–160°C at 20°C min−1, 160–280°C at 3°C min−1, and 280–300°C at 30°C min−1.

The results: what to expect.  For leaves, most oxylipins give baseline separation. Healthy wild-type Arabidopsis leaves (Figure 2, upper panel) analysed without the de-esterification protocol give relatively few peaks; 18:3, 18:2 and 16:0 FAs dominate these profiles, and OPDA is observed, as are lower levels of JA which, at about 0.7 ng g−1 fresh weight, are about three to four times the detection limit for this compound with EI. The parallel analysis of a lesion mimic mutant accelerated cell death 2-2 (acd2-2; Greenberg et al., 1994) revealed a far more complex oxylipin signature (Figure 2, lower panel) with high levels of free 18:2, 18:3, OPDA and JA. In this particular case, one attractive interpretation is that this can partly be explained by increased in vivo lipase activity during lesion formation. Infection with avirulent pathogens yields profiles similar to those recorded for acd2-2. Dry seeds (e.g. cereal and Arabidopsis) give spectra dominated by FAs. However, upon germination, a large variety of oxylipins and oxylipin-like compounds accumulate.

Figure 2.

Oxylipin signatures of healthy wild-type (Col-0) Arabidopsis leaves and leaves from the accelerated cell death 2-2 (acd2-2) mutant (Greenberg et al., 1994), undergoing lesion formation.
Leaves were from 6-week-old plants grown at 22°C with 10 h light. Samples were analysed with the Weber et al. (1997) protocol after methylation with no de-esterification step. Note the massive accumulation of oxylipins such as jasmonate (JA) and also the phytoalexin camalexin. Similar profiles are obtained when leaves are infiltrated with avirulent bacterial pathogens (not shown). See Table S1 for abbreviations.

The method can be used in many ways. Illustrated in Figure 3 is a strategy to examine the leaf/insect interface, to answer the question of which oxylipins the insect encounters when it feeds. By analysing regurgitant from insects fed on both wild-type and alleneoxide synthase (aos) mutant (Park et al., 2002) plants, the provenance of many oxylipins can be determined. The experiment shown in Figure 3 revealed the presence of 13-keto-octadecadienoic acid (13-KOTE), a biologically active and electrophilic ketotriene that may have been released from plant membranes by wound-induced and/or insect-mediated de-esterification from membrane lipids. Its absence in aos samples suggests that jasmonates were necessary for its production in the wild-type plant. This example underlines the advantage, indeed necessity, of incorporating mutants in oxylipin analyses.

Figure 3.

Exploring oxylipin profiles at the plant–insect interface. Fourth instar Pieris rapae larvae were allowed to feed for 24 h on the leaves of 6-week-old Arabidopsis plants: (a) wild-type Col-0; (b) alleneoxide synthase (aos) in Col-0 (Park et al., 2002). Regurgitant from the insects (15 μl) was extracted and analysed with the method of Weber et al. (1997) using electron ionization. Samples were methylated prior to gas chromatography/mass spectrometry (GC-MS) analysis. Jasmonate (JA) was at the limit of detection. 13-Keto-octadecadienoic acid (13-KODE) eluted as two isomers; the second to elute may be an isomer with a 12-cis double bond. Note the presence of jasmonates, 13-keto-octadecadienoic acid (13-KODE) and 13-keto-octadecatrienoic acid (13-KOTE) in (a) (wild-type plant) but not in (b) (aos). See Table S1 for abbreviations.

Trouble-shooting.  The three most common problems we have encountered (apart from poor preparation of plant material) are as follows. (i) Peaks from impure solvents or from purification media. Table 1 gives mass spectral values for the antioxidant butylated hydroxytoluene (BHT) and for two amides that occasionally elute from reverse-phase columns. (ii) Derivitization artifacts. Compounds containing α,β-unsaturated carbonyl groups, for example γ-ketols and some cyclopentenones, are prone to rearrange during silylation. Some hydroxylated molecules (e.g. hydroxy-cyclopentanones) can dehydrate during silylation and/or on column to yield cyclopentenones. (iii) Concluding that oxylipins are absent or of low abundance when they are in fact present at high concentration but esterified into cell membranes. Both oxylipin protocols we present have their limitations in that transiently produced and/or unstable oxylipins are hard to trap and extract from living tissues. Specialized methods, beyond the scope of this article, are necessary for these sorts of compounds. Also, compounds too polar to be extracted or volatilized will escape detection.

The estimation of non-enzymatic oxidation.  The products of non-enzymatic lipid oxidation provide vital information on tissue status and can be used as quantitative indices of oxidative processes in vivo. As outlined below, certain oxidized lipids provide specific information about the nature of the oxidation process. In addition, many oxidation products such as phytoprostanes and malondialdehyde (MDA) are chemically reactive electrophile species which display a variety of potent biological activities. However, electrophiles and short FA fragments are not the best markers of non-enzymatic lipid oxidation because they can escape detection as a result of their high chemical reactivity (conjugation to biomolecules), chemical instability and rapid metabolism in vivo. MDA levels are often presented as an indication of non-enzymatic lipid oxidation. If MDA is measured we draw attention to potential difficulties in using thiobarbituric acid reactivity (TBAR) protocols with plant samples (see Janero, 1990 for a discussion of these methods). The presence of true MDA-TBA adducts should be demonstrated. Most TBAR assays require heating samples over 70°C and may, in part, measure susceptibility to non-enzymatic lipid oxidation rather than lipid oxidation per se (L. Dubugnon and EEF, unpublished). For MDA measurement, the method of Yeo et al. (1999) is appropriate if an internal standard and suitable controls are used (Weber et al., 2004). At present the best and most sensitive markers for free radical-mediated non-enzymatic lipid oxidation are non-reactive isoprostanes such as F1-phytoprostanes (Imbusch and Mueller, 2000a) as they cannot be produced by any other oxidation mechanism [i.e. lipoxygenase (LOX) or 1O2-mediated oxidation]. However, as basal isoprostane levels are low, time-consuming sample preparation and derivatization protocols have to be applied (Imbusch and Mueller, 2000b).

Fortunately, good alternatives exist in the form of OH-FAs. OH-FA analysis, especially when combined with organelle or membrane fractionation, can provide great insights into the location and mechanism of lipid oxidation and whether it is a result of enzymatic processes, radical-catalyzed processes, or singlet oxygen effects.

A word about hydroxy-fatty acids.  Biological membranes can act as ‘molecular lenses’ which significantly enhance non-enzymatic FA peroxidation in hydrophobic lipid bilayers. Peroxidation of polyunsaturated FAs can also be catalyzed by lipoxygenases. Whatever their mode of generation, esterified and free OOH-FAs can be reduced in vivo to OH-FAs. In vivo, a complex mixture of OH-FAs of enzymatic and non-enzymatic origin is present both esterified in membranes and as free compounds (Table 1). This complexity contains much useful information.

Unsaturated but not saturated FAs are prone to non-enzymatic oxidation. It should be noted that the non-enzymatic reaction of FAs with oxygen, hydrogen peroxide and superoxide anion radical is slow while that with singlet oxygen and especially with hydroxyl radicals is fast. Although autoxidation may occur spontaneously, the majority of reactive oxidants involved in non-enzymatic lipid peroxidation in vivo are probably formed as side products of metabolic processes. As shown in Table 1, the enzymatic and non-enzymatic oxidation of C18-unsaturated FAs may yield 24 racemic products which differ with respect to the number and configuration of the double bonds as well as the carbon at which they are oxygenated. Currently, there is no single analytical method available that allows quantification of all isomers, but the single method we present already yields much useful information on the in vivo mechanisms of OH-FA production.

Singlet oxygen (1O2).  Photo-oxidation is an important mechanism of lipid peroxidation which involves formation of reactive 1O2 from normal triplet O2 in the presence of light and a photosensitizer. Singlet oxygen, which is not a radical, reacts directly with isolated double bonds of unsaturated FAs. For each double bond present, two hydroperoxides with equal abundance can be formed. The rate of reaction is proportional to the number of double bonds present in the FA. Singlet oxygen reacts with dienoic FAs approximately 1500 times faster than normal triplet oxygen (Frankel, 2005). As no free radical intermediates are formed, the reaction is clean and cannot be inhibited by radical chain-breaking antioxidants. However, peroxides formed by singlet oxygen or other oxidation processes may play an important part in initiation of a self-propagating (hydroxyl) radical-catalyzed oxidation process. Oxidation of FAs by singlet oxygen and free radicals yields, in part, the same hydroperoxides. However, as shown in Table 1 for the C18-FA series, 10- and 15-OH-FAs are exclusive markers of 1O2-dependent FA oxidation.

Hydroxyl radical.  In terms of lipid oxidation, hydroxyl radicals are the most reactive oxygen species. Hydroxyl radicals abstract a hydrogen radical from unsaturated FAs which may then combine with oxygen, yielding a peroxyl radical which may abstract a hydrogen radical from another FA. Thereby the radical chain reaction is propagated and hydroperoxy FAs are generated. The intermediate radical species may produce a great variety of secondary products including MDA and phytoprostanes. The sensitivity of FAs to free radical-catalyzed oxidation increases disproportionally to the number of double bonds. When the reactivity of a monoene (such as in oleate, 18:1) is set to 1, the reactivity of a diene (as in linoleate, 18:2) increases to 40 and that of a triene (as in linolenate, 18:3) to 80 (Frankel, 2005). From oleate, 8-OH-FAs and 11-OH-FAs can be formed and these OH-FAs represent exclusive but not very sensitive markers of radical-dependent FA oxidation. From linoleic acid, a racemic mixture of 9-OOH- and 13-OOH-FA in a 1:1 ratio is formed. In contrast, linolenate autoxidation yields a racemic mixture of 9-OOH-, 12-OOH-, 13-OOH- and 16-OOH-FA with an uneven distribution ratio of 25, 8, 10 and 57%, respectively, as the intermediate peroxyl radicals may undergo secondary reactions leading to different products. 12- and 16-OH-FAs are considered to be good general markers for non-enzymatic oxidation (1O2- and radical-dependent).

Lipoxygenases.  From a chemical point of view, lipoxygenases catalyse a reaction sequence similar to that of free radicals. During the enzymatic oxidation, the radical chain reaction is initiated and terminated by the enzyme, giving 9(S)- and 13(S)-OOH-FAs as products in plants. With the exception of some seed-specific and a few specialized lipoxygenases, these enzymes generally act only on free FAs (Feussner and Wasternack, 2002). The enzymatic reduction of these hydroperoxides yields exclusively free 9(S)- and 13(S)-OH-FAs. 9- and 13-OH-FAs are markers for both enzymatic (LOX-dependent) and non-enzymatic (1O2- and radical-dependent) FA oxidation. In order to fully explore the nature and extent of lipoxygenase activity, hydroxyl group stereochemistry must be examined (Montillet et al., 2004).

Choice of method for OH-FA analysis.  The GC-MS protocol described below is especially useful to characterize non-enzymatic lipid peroxidation. The advantage of this protocol is that sample complexity is greatly reduced by hydrogenation of OH-FAs. Moreover, analysis of the resulting OH-stearates is easy and does not require special expertise. A disadvantage of the method is that direct information about the precursor FAs (18:1, 18:2 or 18:3) and the double bond geometry is lost. However, with the protocol singlet oxygen- and radical-catalyzed lipid oxidation can be distinguished.

Alternatively, HPLC-based methods are widely used for the determination of OH-FAs. However, authentic reference compounds and at least two HPLC runs using reverse-phase, straight-phase or chiral columns are usually required to resolve OH-FAs. Special expertise is thus required. As these methods are based on HPLC coupled to an optical detector at 243 nm, OH-FAs with non-conjugated double bonds (10-OH- and 15-OH-FAs) and low-abundance OH-FAs cannot be detected. The Montillet et al. (2004) method is, however, especially useful for the determination of abundant enzymatic OH-FAs and is highly recommended for this.

Protocol 3 for the analysis of non-enzymatic oxidation

The basic protocol.  The principle of the protocol is to convert (by methylation and hydrogenation) all C18 OH-FAs into methyl hydroxy-stearates. These saturated FAs are derivatized with BSTFA to their TMS ethers which are then separated by GC. Hydrogenation is performed for several reasons. The number of isomers (24 racemic OH-FAs) produced by different double bond geometries is reduced to eight. In addition, the hydrogenated OH-FAs are chemically and thermally more stable and yield few, but characteristic and intense, fragment ions in the EI mode. The fragmentation pattern easily allows identification and quantification of hydroxy stearates (Figure 4). A previously published protocol (Mueller and Brodschelm, 1994) provides very clean samples that are especially suited to this method. It has been used for oxylipin and phytohormone analysis by many groups (Müller et al., 2002; Stumpe et al., 2005).

Figure 4.

Analysis of total hydroxy fatty acids in leaves. Healthy Arabidopsis (Col-0) leaf tissue (100 mg) was extracted, hydrolysed and analysed as described in the protocol for hydroxy fatty acids (OH-FA) analysis. Detection in the electron ionization (EI) mode was employed. (R,S)-Hydroxy-stearates were detected by selected ion monitoring of the A-fragments (Table 2). B-fragments can also be used for identification and quantification (not shown).

Step 1: extraction. Frozen plant material (0.1–0.5 g of tissue) is transferred into a 50-ml Falcon tube. 15-hydroxy-eicosatetraenoic acid (15-HETE; Cayman Chemicals, Ann Arbor, MI, USA) is added as an internal standard (50 or 500 ng for the analysis of free or total HO-FAs, respectively). Saturated NaCl solution (10 ml), 1 m citric acid (0.5 ml) and diethyether [25 ml of Et2O containing 0.005% (weight/volume; w/v) butylated hydroxytoluene as antioxidant] are added to the frozen sample. Triphenylphosphine (50 mg) is then added. The mixture is homogenized for 3 min with a high-performance disperser (Ultra-Turrax T25 at 24 000 r.p.m.; IKA-Werk, Staufen, Germany). After centrifugation (10 min at 2000 g), the ether phase is removed and directly used in the next step. For the determination of total OH-FAs, the de-esterification protocol is applied after extraction.

Step 2: sample purification. The ether extract from step 1 is applied to an aminopropyl solid-phase extraction column (500 mg). The column is washed with 3 ml of chloroform:isopropanol, 2:1 (v/v), and eluted with 6 ml of Et2O:acetic acid, 98:2 (v/v). [The column can be regenerated by successive washes with 5 ml of methanol:12 N hydrochloric acid (95:5, v/v), methanol, methanol:triethylamine (95:5), methanol and Et2O.] The sample is taken to dryness at 40°C under a stream of N2 and derivatized organic acids (acidic oxylipins) are trapped on the aminopropyl silica material as ammonium salts. Subsequent washes with organic solvents (e.g. chloroform/methanol) will remove most of the non-acidic contaminants, including chlorophyll. For elution, acetic acid is added to the organic solvent (Et2O) to protonate the carboxy anions of the oxylipins in order to break the strong ionic interactions.

Step 3: methylation. The sample is taken to dryness and methylated by adding 50 μl of methanol, 200 μl of Et2O and 5 μl of (trimethylsilyl)diazomethane (2 m solution in hexane). The sample is incubated at room temperature for 5 min and then taken to dryness for hydrogenation.

Step 4: hydrogenation. The sample is taken to dryness in a 25-ml round-bottomed flask and 2–5 mg of PtO2 hydrate catalyst is added. The sample is suspended in 5 ml of ethanol and stirred slowly with a magnetic bar. The flask is connected via a three-way tap to a water pump and slowly evaporated. The tap is closed as soon as the ethanol starts to boil. Next, the flask is connected to a reservoir of hydrogen (e.g. a balloon that has been flushed twice with hydrogen) and vented with hydrogen. The sample is stirred vigorously for 20 min under the atmosphere of hydrogen. At the end of this time, the hydrogen supply is disconnected, and the flask is evaporated again and vented with air. The catalyst can be filtered off by passing the suspension through a pipette tip packed with a homemade glass fibre filter. The entire hydrogenation procedure must be performed in a hood. Precautions should be taken to prevent the dry catalyst coming into contact with hydrogen and air, as this almost always causes an oxyhydrogen detonation. When the solution has passed the filter, the hydrogen-loaded catalyst on the filter may eventually become dry and start to glow in the presence of air. Therefore, highly flammable chemicals (e.g. Et2O) should be removed from the hood beforehand. The ethanol solution containing the hydrogenated sample is taken to dryness before proceeding to the next step.

Step 5: silylation. The sample is then suspended in 100 μl of chloroform. After addition of 50 μl of BSTFA the sample is heated at 40°C for 1 h. The sample is taken to dryness and reconstituted in 20 μl of hexane prior to injection (1 μl) into the GC-MS.

Programming the GC.  OH-FA methyl esters/TMS ethers are analysed on a HP-5 column (30 m × 0.25 mm), with a linear helium flow at 23 cm sec−1 and a column temperature step gradient of 150–225°C at 20°C min−1, 225–275°C at 5°C min−1, 275–300°C at 20°C min−1, and 300°C for 5 min. The injector of the splitless mode is set at 280°C. Data are collected in the EI mode.


Quantification is performed by comparing the area of characteristic fragment ion peaks with that of the area of the fragment ion peak of the 15-hydroxy-eicosatetraenoic acid (15-HETE) derivative. Note that one disadvantage is that 9- and 13-hydroxy derivatives of linoleic and linolenic acid (of enzymatic and non-enzymatic origin) are converted into the same molecule and thus can no longer be distinguished.

What to expect

All OH-FA derivatives listed in Table 2 occur in healthy Arabidopsis leaves and can be quantified by GC-MS in the EI mode. A chromatogram from an analysis of Arabidopsis leaves is shown in Figure 4. C18-OH-FAs other than those listed in Table 2 could not be detected, although they might occur under some conditions. 2-OH-FAs that are derived from FAs via the α-dioxygenase pathway have been reported to occur abundantly in tobacco (Nicotiana tabacum; Hamberg et al., 2003) and are probably also present in Arabidopsis. 2-OH-FAs are not detected by the method presented as they are polar molecules that do not elute from the aminopropyl column. However, elution of the column with 9 ml of methanol:acetic acid 98:1 (v/v) recovers these metabolites together with di- and trihydroxylated FAs (Hamberg et al., 2003).

Table 2.  Hydroxy-fatty acid levels in healthy Arabidopsis leaves
OH-FA derivativem/zResponseOH-FA levels in Arabidopsis leaves
inline image(18:0)A/(20:0)A (18:0)B/(20:0)BFree OH-FA (pmol g−1)Esterified OH-FAs (nmol g−1)
  1. Hydroxy fatty acids were methylated, hydrogenated and silylated prior to gas chromatograph/mass spectrometer (GC-MS) analysis in the electron ionization (EI) mode. Two characteristic ions, fragments A and B, were monitored and used for quantification. Response factors were calculated from chromatograms of a 1:1 mixture (mol/mol) of the C18 hydroxy fatty acids (OH-FAs) and the C20 OH-FA (internal standard). Peak areas of fragment ions A, (18:0)A and (20:0)A, or fragment ions B, (18:0)B and (20:0)B, were used to calculate the area ratio. OH-FA levels (per gram fresh weight) in Arabidopsis leaves were determined using 15-hydroxyeicosateraenoic acid as the internal standard. Not included are OH-FAs found in cutin and sphingolipids.

15-OH-(20:0)173343int. Standardint. Standard

In leaves of 6-week-old Arabidopsis plants, levels of free OH-FAs are in the range of 30 to 350 pmol g−1 of fresh weight. Concentrations of esterified hydroxy FAs are typically 10- to 100-fold higher than those of free OH-FAs and are found in the range of 1 to 39 nmol g−1 of fresh weight. Dependent on the growth conditions and the age of plants, levels of OH-FAs may vary considerably. The relatively high levels of 13-OH-FAs in Arabidopsis leaves (Table 2) indicate the formation of 13-OH-FAs by 13-LOXs (Montillet et al., 2004). Typically, levels of 12- and 16-HO-FAs in Arabidopsis are <10–20% of the levels of 9- and 13-OH-FAs. During leaf aging, the relative proportion of non-enzymatic oxidation products increases. In senescent leaves, non-enzymatic oxidation products may become dominant (Berger et al., 2001). During the hypersensitive response in potato leaves, the rates of non-enzymatic and enzymatic lipid oxidation appear to be interdependent (Göbel et al., 2003).

In healthy Arabidopsis leaves, levels of 8-OH- and 11-OH-FAs are generally low. As they are derived exclusively from free radical-catalyzed oxidation of oleate, they could potentially be used as markers of radical-catalyzed lipid peroxidation. However, oleate oxidizes much more slowly than linolenate (see above) and thus is not a very sensitive marker. 15-OH-FAs are exclusively derived from 1O2-dependent oxidation of linolenate. The relatively high levels of 15-OH-FAs in Arabidopsis leaves indicate that singlet oxygen produces significant amounts of OH-FAs. In Arabidopsis roots, 15-OH-FAs are barely detectable, their levels being <1% of those in leaf tissue. Therefore, 15- OH-FAs may be used as a marker of singlet oxygen-mediated lipid peroxidation. Traditionally, 12-OH-FAs and 16-OH-FAs have been used as markers for non-enzymatic lipid peroxidation of linoleate/linolenate and linolenate, respectively (Berger et al., 2001; Montillet et al., 2004). As these HO-FAs can be derived from 1O2-dependent as well as free radical-catalyzed oxidation, the relative contributions of the two processes to the formation of these OH-FAs cannot be differentiated. However, under conditions where 1O2-mediated lipid peroxidation can be neglected (such as in roots), 12- and 16-OH-FAs are accurate markers of free radical-catalyzed lipid peroxidation.

Future challenges

The limiting factor in oxylipin analysis is good experimental design. With this, we can soon expect to know whether membrane subdomains such as lipid rafts harbour biologically active oxylipins, which oxylipins propagate information over long distances in the plant, and which oxylipins are produced in specialized cells such as those in reproductive tissues. We can also look deeper into the mechanisms and consequences of membrane lipid oxidation down to subcellular levels.


We thank members of our laboratories past and present for critical comments and advice. Aurore Chéletat is thanked for technical assistance. Christian Triantaphylides and Jean-Luc Montillet provided useful suggestions. Scott Sattler and Mads E. Nielsen tested and elaborated part of the de-esterification protocol given herein. This work was supported by grants from the DFG (SFB 567) to MJM, and Swiss NSF (NCCR Plant Survival programme) to EEF.