Fatty acids are the simplest and most fundamental lipids found in biology, and they are readily analyzed by mass spectrometry. A rich literature exists concerning the analysis of these structures, primarily using electron ionization (EI); from saturated, straight chain species to more complex unsaturated and alkyl-substituted examples. EI is quite an energetic mode of ionization that leads to substantial ion decomposition from which, when followed by subsequent mass spectrometric analysis, a wealth of structural information can be deduced. The ion chemistry of these events was covered extensively in the first monograph on this topic by this author.¹ 

Electrospray ionization (ESI) of free fatty acids (RCOOH) effectively yields molecular ion species at higher efficiency than EI; both positive [M + H]⁺ and [M + cation]⁺ as well as negative ions [M − H]⁻, but in a manner that leads to little, if any, subsequent fragmentation.²  Therefore, to glean any structural details concerning fatty acids it is necessary to impart internal energy to these fatty acid molecular ion species, sufficient to break covalent bonds. In almost all cases this is accomplished by collisional activation (collision induced dissociation, CID) with a neutral gas molecule in a collision cell, such as Rf-only quadrupole sector, or by increasing the angular momentum of the ion in a Paul-type ion trap or ion cyclotron resonance instrument where it can collide with neutral gas molecules in the cell.

Fatty acids exist in biological matrices as the deprotonated, ionized carboxylic acid (carboxylate anion) or esterified to glycerolipids, glycerophospholipids, wax esters, various saccharolipids, as well as cholesterol esters (steroid alcohol such as ergosterol in fungi) along with other less abundant esters of complex lipids. Many sphingolipids have long chain fatty acyl substituents derived from fatty acids, but exist in this class of lipids as N-acyl amides that are very stable to base hydrolysis. Analysis of fatty acids can involve electrospray ionization of the free fatty acids or the fatty acids liberated from the various esterified fatty acids after saponification.

The isolation of free fatty acids is very facile and can be carried out by solvent extraction techniques. A typical experiment might involve isolation of all simple and complex fatty acids using techniques such as the Folch or Bligh/Dyer extraction followed by saponification using 0.5 M NaOH at 37 °C for 1 hour.^3,4  This solution is then neutralized and extracted by hexane or ethyl acetate to isolate the very nonpolar fatty acids in the organic layer in high yield. Subsequent electrospray analysis involves dissolving these isolated fatty acids in a solvent system that permits ESI, such as a methanol/water system containing ammonium acetate (5 mM) to provide electrical conductivity to the spray solvent.

As discussed below, the saturated fatty acids are difficult to collisionally activate to yield structural informative ions, unlike the case for EI mass spectrometry. While molecular weight information is obtained, a definitive structure such as possible branching points are difficult, if not impossible, to assign from electrospray generated positive or negative ions. When high resolution mass analysis is employed to detect these fatty acid molecular species, the exact elemental composition of the ion can be readily calculated, since these are usually below m/z 500 and high resolution mass analyzers currently marketed readily achieve 5 ppm or less mass error measurement. When using ESI, definitive analysis of a fatty acid, even for saturated fatty acids, requires additional information such as chromatographic retention times relative to standard fatty acids to unambiguously assign the structure of these molecules. This is due to potential methyl branched alkyl chains and double bond regioisomers. Alternatively, making a derivative such as a methyl ester followed by GC-MS analysis remains a powerful strategy, as previously described.¹ 

Most common fatty acids typically have a single, polar functional group, namely, the carboxylic acid moiety. In aqueous solution, the pKa of most fatty acids is between 4 and 5, making them highly ionized at neutral pH as the RCOO⁻ form and, not surprisingly, when electrosprayed in a solvent system buffered to pH 5 or above, yields a robust appearance of the carboxylate anion [M − H]⁻ emerging from the ion source. The collisional activation of fatty acids is exemplified by four different fatty acids that are typically encountered in biological extracts (Figure 1.1). The [M − H]⁻ ions derived from common fatty acids found in most biological samples are listed in Table 1.1 along with product ions observed following collisional activation. As can be seen from these examples, very few product ions are observed for saturated and monounsaturated fatty acids; however, there are product ions that have been reported when alkali attachment cations are collisionally activated, as will be discussed below.

Collisional activation of the [M − H]⁻ from saturated fatty acids typically only yields [M − H − H₂O]⁻,⁵  which is not structurally definitive. Yet this ion is interesting, since it must involve the loss of two protons from the alkyl chain of the fatty acid. Study of this ion using techniques such as resonance capture ionization has suggested a cyclic structure for this ion (Scheme 1.1),⁶  but in theory it could proceed from a negative ion charge localized at a site along the alkyl chain (migration of H⁺ to the carboxylate anion) followed by the loss of the small neutral water molecule. This does occur to some extent with saturated fatty acids, but is more favorable for polyunsaturated fatty acids when the charge site on the alkyl chain can be delocalized over many atoms.

Structural characterization of saturated fatty acids has been extensively pursued by electron ionization mass spectrometry, where ion formation is much more energetic and decomposition processes that lead to structurally informative product ions are more favorable. In addition, chromatographic separation, in particular using capillary gas chromatography or high performance liquid chromatography, are powerful strategies to combine with molecular weight information (mass spectrometry) to assign a more definitive structure to the observed molecular anions through strategies such as co-elution with authentic standards. An example is the liquid chromatographic separation of a series of common fatty acids revealed by the observed molecular anions (Figure 1.2).

ESI can also be used to generate positive ions from fatty acids and this mode of ion formation has largely been employed by making alkali metal adducts such as [M − H + Na₂]⁺, [M − H + Li₂]⁺, and [M − H + Ba]⁺. Such ions are interesting in that they have a complete closed shell of electrons and are thus quite stable. The formation of these adduct ions requires dissolving the free fatty acid in an appropriate solvent typically containing 1–3 mM alkali metal salt as the acetate or formate form of the corresponding cation. The yield of the attachment ion products can be less than quantitative in that multiple adduct species are often observed such as [M + Na]⁺, [M + Li]⁺, as well as the desired [M − H + 2Li]⁺ for lithiated species (Table 1.2). In such cases the origin of the sodium ions is likely from the borosilicate glass surfaces used in vials to prepare the samples. There have not been detailed studies reported to optimize a single adduct of fatty acid from such alkali metal salts and multiple species of cations from a single fatty acid is a complication not seen in negative ion ESI where just a single [M − H]⁻ is readily observed.

An advantage of the mass spectrometric analysis of such positive ions has been reports of product ions being formed following collisional activation, even at relatively low energies. For saturated fatty acids a series of alkali ions formed by fast atom bombardment ionization was reported a number of years ago following collisional activation of [M − H + 2Li] and [M − H + Ba]⁺ by charge site remote fragmentation mechanisms (CRF), also termed “remote site fragmentation”.⁷  This CRF process took place when precursor ions were accelerated to high velocities just prior to collision with a neutral gas molecule in a collision cell, initiating a high energy-decomposition reaction (Scheme 1.2). It now appears that low energy collisional activation of [M − H + 2Li]⁺ ions generated by ESI in a tandem quadrupole instrument can also undergo CRF-driven decomposition.⁸  Mechanisms for ion formation had been reported using isotope labeled fatty acids and include charge-driven as well as charge-remote rearrangements of carbon–carbon bonds. The general utility of these alkali attachment ions and interpretation of CID events for structural characterization of unknown fatty acids has not been widely reported, but this technique does offer some promise when applied to the analysis of unsaturated and polyunsaturated fatty acids.

The situation for analysis of unsaturated fatty acids is, on one hand, much better in terms of obtaining abundant product ions following collisional activation of molecular ion species, but on the other hand, it is made more difficult because of the introduction of stereochemical uncertainty such as double bond position and double bond geometry for unsaturated fatty acids. The addition of the π-orbitals to the molecular structure of these fatty acids facilitates carbon–carbon bond cleavage for either the carboxylate anion or the metalated carboxylate cation. In some cases product ion formation is relevant to the position of the double bonds and as the number of double bonds increases, additional pathways become open for ion decomposition events. The mechanisms for the reactions have only been partially studied but nonetheless, they have become definitive for structural identification of biologically derived fatty acids.

The major product ions from a number of monounsaturated fatty acid molecular anions [M − H]⁻ observed at low collision energy, is a loss of water [M − H − H₂O]⁻; the exact origin or origins of the protons lost along with the carboxylate oxygen atom have not been fully elucidated, but most likely they are the result of formation of a cyclic anion (Scheme 1.1).

Additional product ions include cleavage of the carbon–carbon bond adjacent (α) or vinylic (β) to the double bond (Figure 1.3). A mechanism consistent with octadec-9-enoate (18 : 1 n − 9) and the unsaturated fatty acids presented in this figure could involve a change remote fragmentation.⁵  The driving force for such a mechanism would involve breaking of the C–H bond allylic to the double bond by the electron withdrawing properties of the π-orbital with concerted formation of a stable conjugated double bond on one fragment, loss of hydrogen gas (H₂) and a stable olefin (Scheme 1.2). An alternative reaction for the abundant ion at m/z 127 (Figure 1.3A) is an “ene” reaction, seen quite frequently in lipid collisional spectra which is a proton abstraction reaction driven by a double bond, as shown in Scheme 1.3. This gas phase behavior is surprisingly common for many unsaturated fatty acids, either as carboxylate anions or metalated cations, and since the carbon–carbon bond formation does not involve those atoms bearing the charge site, the term “remote site fragmentation” is relevant.

The dilithiated octadec-9-enoate [M − H + Li₂]⁺ cation can be used as an example of the behavior of monounsaturated fatty acids that have a closed shell of electrons (Figure 1.4A). These species are readily formed by ESI when alkali metal salts such as LiCl or Li-acetate are added to the electrospray solvent system. Unfortunately, the formation of this cationized carboxylate may not be quantitative and other cationic species are often observed, including [M + H]⁺, [M + Li]⁺, [M − H + Li₂]⁺², and even [M − H + Na₂]⁺. The exact alkali metal salt, ESI conditions and even the ESI ion source employed, can alter the relative ratios of all of these ions. While this makes it less attractive to study these species, the advantage of making these adducts is that detailed studies have been reported concerning their behavior and thus they provide specific information that is useful in predicting subtle changes in structure with an unknown fatty acid.

A common abundant product ion for the dilithiated cation from monounsaturated and polyunsaturated fatty acids with a first double bond at least seven carbons from the carboxyl group appears at an odd mass and corresponds to a charge remote fragmentation involving the double bond nearest the cationized carboxyl group. This arises from oleate (18 : 1 n-9) following cleavage of carbon 4-5 and a single proton rearrangement, presumably from carbon-4. A mechanism consistent with this ion formation has been suggested to be a charge remote elimination reaction of H₂ and formation of two double bonds after cleavage of carbon 4-5 (Scheme 1.4).⁷ 

Another consistent remote site mechanism that does not involve the loss of H₂ would be driven by proton abstraction by the double bond. This ene reaction drives cleavage of the carbon–carbon bond, 5-carbon atoms removed, leading to a cyclopentane neutral and a terminal olefin attached to the dilithiated carboxylate cations (Scheme 1.5).

A series of radical cation ions (even m/z) are observed at a low m/z that corresponds to homolytic bond cleavage near the metalated carboxylate cation. Virtually all [M − H + 2Li]⁺ fatty acids (saturated, monounsaturated, and polyunsaturated) have these characteristic ions,⁸  but their abundance varies in an interesting manner that can be used to deduce a proximal position of the first double bond when it is quite near the carboxylate moiety. For Δ⁴-fatty acids (e.g. docosahexaenoate-lithium adduct) the most abundant ion appears at m/z 72, likely abundant because of the loss of a neutral radical species delocalized by the Δ⁴ double bond shown in Scheme 1.6.

The Δ⁵-series, such as is found in arachidonic acid and EPA, has the metalated carboxylate radical cation as the most abundant low mass ion (m/z 58), perhaps due to a stabilized loss of the highly delocalized radical illustrated in Scheme 1.6. These Δ⁵-species still have an abundant m/z 72 radical cation that can be stabilized by the resonance structure illustrated (Scheme 1.6). The Δ⁶ unsaturated fatty acids yield a radical cation of highest abundance observed at m/z 72 due to both delocalization the radical site in the neutral hydrocarbon chain that is lost, and delocalization of the radical by the carboxylate electrons. For the location of the nearest double bond more distal than Δ⁶, remote site fragmentation appears to be the most abundant process occurring in the production of these low mass product ions (Scheme 1.4 and 1.5).

These same ion series of low mass ions (Scheme 1.4) are also observed following CID of the barium ion adduct [M − H + Ba]⁺ (Figure 1.4B) at m/z 196, 209, and 223, presumably following the same mechanism of decomposition of the molecular ion species at m/z 419 from bariated oleate.⁹  The most abundant ion is seen at m/z 209 for this barium adduct that corresponds to charge remote cleavage of carbon bond 3-4, as illustrated for m/z 85 for lithiated carboxylate cations and may reflect unique weakening of the C–H bond at C-2 due to the affinity of Ba⁺⁺ for the electrons at the carboxylate oxygen atoms (Scheme 1.4). An ion at m/z 196 is found quite abundant and likely an analog of the structure suggested in Scheme 1.6 for m/z 72, again due to the possible resonance structures of this ion.

Considerably more structurally relevant information is generated following collisional activation of polyunsaturated (as homoconjugated double bonds) fatty acid anions, as exemplified by the CID of arachidonic acid (AA) anion (Figure 1.5). In addition to the loss of H₂O (m/z 285) and CO₂ (m/z 259), the next most abundant product ion could, at first glance, correspond to cleavage of a double bond between carbon atoms 5 and 6 (m/z 205). This site of cleavage is consistent with the ion at m/z 212 from D₈-AA (which has all double bond methine hydrogen atoms labeled with deuterium) (Figure 1.5B), which is shifted to a +7 ion. However, the origin of this ion at m/z 205 must be more complex since there is also a significant abundance at +6 (m/z 211) from the D₈-AA. In addition, high resolution analysis of the ion at m/z 205 yields a doublet at m/z 205.196 (C₁₅H₂₄) and 205.123 (C₁₃H₁₇O₂).

Two possible pathways that would account for the carbon 5-6 double bond cleavage are presented in Scheme 1.7. One would be a charge remote double 1[3]-sigmatropic arrangement of the double bond to the Δ^3,4 position then loss of neutral CO₂ and butadiene, yielding a carbanion at carbon-5 at m/z 205.196. This double bond rearrangement could also be charge driven by the carboxylate anion that could readily remove a proton from carbon-4, generating a conjugated carbanion site carbon-6 and shifting the double bond to Δ⁴. This carbanion structure could undergo additional proton abstraction at carbon-13, then carbon-3, reverting to the same Δ^3,4 carboxylate anion structure in Scheme 1.7. While the carbanion site would seem to be rather energetic, proton abstraction at carbon-13 would lead to a much more stable delocalized anion through conjugation with the two adjacent double bonds (Scheme 1.7).

The deuterium labeled arachidonate, as well as high resolution measurements of m/z 205, revealed formation of a second ion at m/z 205. The origin of this second product that is consistent with this data involves loss of the terminal 7 carbon atoms (along with both protons at C-14,15 in deuterium labeled in D₈-AA as well as the 19,19,20,20-D₄-AA),⁵  but shifted to m/z 211 for D₈-AA, as in Figure 1.5B (Scheme 1.8). This mechanism involves an allylic proton transfer and vinylic bond cleavage described by Hsu and Turk and is another example of an ene proton transfer reaction.⁸  The loss of this C₇H₁₄ neutral species would leave a highly stable conjugated triene anion at m/z 205.123.

Docosahexaenoate anion behaves in a similar manner in terms of cleavage of susceptible double bonds, which are both allylic and vinylic. The product ion at m/z 229 (Figure 1.1) is likely formed by a similar mechanism as in Scheme 1.7 after shift of the double bond at Δ⁴ to Δ³ and charge driven loss of CO₂ and butadiene.

Metal ion attachment to the carboxylate moiety of fatty acids has been used to generate positive ions from polyunsaturated fatty acids. The advantage of this approach is to render a very stable closed shell ion (cationic) where this charge site does not easily leave the carboxyl moiety, even when ions receive additional energy. When collisions of this stable cation occur in a tandem quadrupole or ion trap-type mass spectrometer, excitation of the ion takes place, which is often relaxed by charge remote fragmentation mechanisms. Structurally useful reactions occur for the mono- and polyunsaturated fatty acid species as described for the Li⁺ adducts by Hsu and Turk.⁸  Since biosynthesis of polyunsaturated fatty acids (PUFA) leads to homoconjugated double bonds that are classified as to the double bond closest to the methyl terminus, indicated by “n- terminology” (e.g. n-6 or n-3 PUFAs). These positions are largely determined by two carbon atom elongases, such as EOVL1.¹⁰  The position of the first double bond, counting from the carboxyl carbon atom, is designated by the “delta nomenclature” (e.g. Δ⁵, Δ⁹, Δ¹¹, etc.) that indicates the number of carbon atoms between the carboxyl group and the first double bond. The exact position of the first homoconjugated double bonds is controlled by fatty acyl CoA ester desaturases such as Δ⁵-desaturase or Δ⁶-desaturase.¹⁰  These double bond positions become important targets for structure elucidation and CID mass spectrometry of these closed shell cations offered by alkali metal adducts.

Several examples of diverse PUFAs have been examined as Li⁺ adducts and mechanism of product ion formation suggested (8). While these mechanisms have not been rigorously tested, they are consistent with high resolution measurements and limited availability of isotopically labeled PUFAs. The CID of three example PUFAs (arachidonic acid, Δ⁵, ω-6; docosatetraenoic acid, Δ⁷, ω-6; and docosahexaenoic acid, Δ⁴, ω-3) illustrate major mechanisms of ion formation (Figure 1.6).

Allylic/vinylic carbon bond cleavages (A–V cleavage) with hydrogen transfer and formation of conjugated product ions or neutrals dominate the decomposition of these molecules after collisional activation (Scheme 1.9) and the observed ion depends upon which side (R₁ or R₂) the lithiated carboxyl group is located. The C–C bond cleaved in this scheme is both allylic to carbon A and vinylic to carbon B. Fragment X or Y can either be the resultant product ion since the charge remote fragmentation is driven by the stability of both products.

The ion of highest mass-to-charge ratio, which is also the most abundant in this region for each PUFA in Figure 1.6 correspond to this A–V cleavage at the n-terminus. This ion can classify the last double bond position in the PUFA and, therefore, the n-family type.

From the highest mass n-series ion formed by the A–V mechanism, the other members in this series can be seen at sequential 40 Da intervals. For CID at m/z 341 from the [M − H + 2Li]⁺ of DHA,⁸  the n-series ions are observed at m/z 285, 245, 205, 165, 125, and 85. The mechanism of formation of m/z 285 is illustrated in Scheme 1.10 for this A–V series of charge remote fragmentations.

The other ion series that differs also by 40 Da (m/z 113, 153, 193, and 233) would arise from the same general mechanism, but initiated at a different allylic/vinylic site for the hydrogen transfer (Scheme 1.11). As discussed previously, in the saturated and monounsaturated fatty acids, the family members with the first double bond in a similar position relative to the carboxyl group can also be readily discerned. The mechanisms of formation of either the radical cations at low mass have been presented (Scheme 1.6) or the cations derived from the alternative A–V cleavage (m/z 233) are indicated in Scheme 1.11.

Many different types of hydroxy fatty acids are formed as intermediates of fatty acid biosynthesis and metabolism, as well as products of lipid peroxidation, to name just a few examples. Even the simplest monohydroxy fatty acid offers an additional channel for ion decomposition after collisional activation of either positive or negative ions in the tandem quadrupole mass spectrometer due to the lower activation energy imparted by the additional oxygen heteroatom.¹¹  As the complexity of hydroxy fatty acids increases with additional functional groups such as double bonds and additional carbinol moieties, the number of product ions increase as well as the amount of structural information that is readily gleaned from understanding the mechanism of ion formation. Most detailed mechanistic studies have been carried out following the collisional activation of negative ions (carboxylate anions) of these saturated species, but decomposition mechanisms for the carboxylate metalated cations have also been reported.⁸ 

The singly hydroxylated fatty acid yields abundant fragment ions after collisional activation of [M − H]⁻ and except for the 3-hydroxy fatty acids, the ions which are formed are surprisingly consistent, no matter at which position the hydroxyl group is present on the fatty alkyl chain. This renders structural information difficult to glean since the same mass losses are typically observed. This is illustrated for three positional isomers (Figure 1.7) 2-hydroxyhexadecanoate, 3-hydroxyhexadecanoate, and 15-hydroxypentadecanoate. The product ions for 2-hydroxyhexadecanoate and 15-hydroxypentadecanoate are surprisingly similar with a loss of 18, 46 and 74 Da from the molecular anion [M − H]⁻. While the high mass ions formed following the collisional activation of 3-hydroxyhexadecanoate molecular anion yield high mass product ions corresponding to the losses of 18 and 46 (very low abundance) Da, the most abundant ion is observed at m/z 59, which is unique to the 3-hydroxy position. This product ion is likely due to a favorable cleavage between carbon-2 and carbon-3 as a result of a charge driven proton rearrangement (ene reaction) (Scheme 1.12) seen for carboxylic acids even under electron ionization conditions.¹ 

This fragmentation mechanism has been shown to yield valuable structural information from α-alkyl-β-hydroxy fatty acids typically encountered as the mycolic acids unique to mycobacteria. A loss of a neutral aldehyde cleaving the α-alkyl group attached to the two carbon fragment of Scheme 1.12 was used to define the position of unsaturation and carbon-chain distribution in their branched-chain fatty acids using neutral loss and precursor ion scanning of a naturally occurring mixture of mycolic acids.¹² 

A mechanism operating in the decomposition of these hydroxy fatty acids yielding the very common ions was studied in detail by Claeys and coworkers using ¹⁸O-labeled carboxyl moieties and H/D-exchange of the proton on the carbinol oxygen atom.¹³  The mechanism common to all these hydroxy fatty acids (Scheme 1.13) likely involves a charge-driven loss of CO₂ then loss of H₂ driven by the adjacent hydride ion (H⁻) from carbon-3, attacking the proton attached to the hydroxyl group (no matter what position), yielding a stable alkoxide anion.

Even if the carbinol site is quite remote from the carboxyl moiety (e.g. 15-hydroxypentadecanoic fatty acid, Figure 1.7C) this is quite an efficient process. When the hydroxyl group is closer to the carbinol group, such as for 12-hydroxyoctadecanoic acid,¹³  ions can also be seen from a charge driven loss of a neutral hydrocarbon chain alpha to the nascent alkoxide anion at m/z 169. A mechanism which can account for the formation of the neutral hydrocarbon would involve formation of an alkoxide anion that could be stabilized as an aldehydic anion (Scheme 1.14).

This type of α-cleavage to the hydroxyl group is even more abundant for dihydroxy fatty acids, where the anion site can be on two separate carbinol moieties. These vicinal diols can be readily made from olefins by OsO₄ oxidation to locate double bonds in an unsaturated fatty acid.¹⁴  The facile formation of these product ions is consistent with the mechanism shown in Scheme 1.14 and supported by stable isotope labeled molecules.¹³ 

The introduction of a double bond into the fatty acid acyl chain that also has a hydroxyl group has a significant impact on the ease of decomposition of carboxylate anions [M − H]⁻ in the tandem quadrupole (low energy) mass spectrometer (Figure 1.8). Since biochemical as well as chemical oxidation of unsaturated fatty acids, and in particular polyunsaturated fatty acids, leads to many different hydroxylated fatty acids, a large number of lipids are found in this class both from enzymatic as well as nonenzymatic products of lipid peroxidation. Major fragment ions are observed, corresponding to cleavage adjacent to the carbinol C–C bond, but allylic to a double bond. Several general mechanisms have been suggested to operate in the collision induced decomposition of unsaturated hydroxy fatty acid molecular ions generated by electrospray ionization.^11,15–17  Both charge driven and charge remote mechanisms (Scheme 1.15) appear to operate and have been proposed to account for the majority of the most abundant product ions.

Either R₁ or R₂ could retain the charge site (carboxylate anion or even the closed-shell alkali metal attachment positive ion) and many examples are available to illustrate this reaction, as well as isotope labeling and high resolution measurements to support this mechanism.¹⁶  A variant of this highly favorable allylic reaction is cleavage of the carbon–carbon bond (vinylic to the alcoholic moiety) carbon atom (Scheme 1.16), but this decomposition pathway is likely preceded by double bond rearrangements. For conjugated dienyl fatty acids, which are common products of lipoxygenase reactions and peroxidation of polyunsaturated fatty acid, a 1[5]-sigmatropic shift likely precedes the carbon–carbon cleavage step which alters the position of the closest double bond to an allylic position, which is much more favorable to cleave, in that a very stable aldehyde and olefin result.

A major driving force for these decomposition reactions would appear to be related to the slightly weakened bond allylic to the double bond, which is rendered this way in part by the electronegativity of the oxygen atom at the immediately adjacent carbon atom. In specific cases, multiple double bond migration steps precede the C–C allylic cleavage step (e.g. 1[5]- sigmatropic shift). It would appear that increasing conjugation of the nascent olefin being formed from a diene to triene further enhances the probability for this fragmentation taking place.¹⁵ 

A second general mechanism for carbon–carbon cleavage of unsaturated hydroxy fatty acid following collisional activation involves charge driven events. Such events are often observed in a negative ion mode, where the rather stable carboxylate anion abstracts the proton of the alcohol functional group, leaving an alkoxide anion at that position. This more reactive charge localized species can readily undergo aldehyde formation (neutral species) on either side of the carbon–oxygen bond, but typically it is most abundant when a stabilized anion can form by charged delocalization by one or more double bonds in conjugation with the bond being cleaved and formation of the carbon-centered negative charge at this site (Scheme 1.17).

Again, the initial carboxylate anion can either be in R₁ or R₂ and the most abundant product ion would be the anion with electron density delocalized by an allylic double bond. A variant of this mechanism would involve an initial charge remote double bond migration, to render a double bond allylic to the carbon–oxygen alkoxide structure. Again, the bond allylic to the double bond is the one most readily cleaved, yielding the highest ion current, in the illustrated case (Scheme 1.18) with the R₂ portion of the starting unsaturated fatty alcohol.

Another interesting mechanism reported to occur for the CID of unsaturated alcohol fatty acids is the oxy-Cope type rearrangement (Scheme 1.19) found for specific polyunsaturated hydroxy fatty acids with a 3-hydroxy-1,5-diene structural unit embedded in the carboxylic acid structure.¹⁵  The unique arrangement of molecular orbitals and charge localization appearing as an alkoxide anion render this cyclic double bond rearrangement quite favorable.¹⁸  The formation of such product ions have been described previously, and are remarkable in that the product ions are radical anions.¹⁶  This means for non-nitrogen-containing hydroxy, unsaturated fatty acids, this ion would appear at an even mass-to-charge ratio.

Specific examples of each of these decomposition mechanisms can be observed for the 18-carbon hydroxy fatty acids in Figure 1.8. The abundant fragment ion observed following CID of 9-hydroxyoctadeca[10,12]dienoate at m/z 171 (Figure 1.8A), could be a result of the terminal aldehyde formation that would be a charge remote allylic fragmentation preceded by migration of both double bonds in a 1[5]-sigmatropic shift as described (Scheme 1.16). A charge-driven mechanism for decomposition of this hydroxy unsaturated fatty acid anion (e.g. Scheme 1.18) would lead to migration of the charge site to the methylene terminal hydrocarbon chain. This ion would occur at m/z 123, but is not observed (Figure 1.8A).

The isomeric fatty acid 13-hydroxyoctadeca[9,11]dienoate also has a rather simple collisional mass spectrum with a major product ion at m/z 195 (Figure 1.8B). This ion could result by both charge-driven and charge-remote mechanisms after 1[5]-sigmatropic shifts of the Δ^9,11-double bond to the Δ^8,11 positions which renders the carbon–carbon bond cleaved being allylic to the conjugated diene (charge-driven mechanism, Scheme 1.18) or alternatively the newly formed Δ¹⁰-double bond being in a favorable position for the charge remote allylic fragmentation mechanism (Scheme 1.17).

These same reaction pathways can be evoked to rationalize the product ions observed following collisional activation of monohydroxy arachidonic acid isomers as negative ions (Figure 1.9). The most abundant ions, aside from the loss of small neutral molecules such as H₂O and CO₂, correspond to carbon–carbon cleavage adjacent to the hydroxyl group. It is important to follow the movement of the proton when one additional or one less mass unit appears in the observed fragment ion. For the CID of 5-HETE, 8-HETE, and 9-HETE the loss of the proton is a result of charge remote double bond migration prior to allylic fragmentation as specifically indicated in the suggested mechanism to rationalize the appearance of m/z 115 from 5-HETE (Scheme 1.20).

The ion at m/z 203 observed from 5-HETE (Figure 1.9A) would most likely be the result of a charged drive reaction, since the site of ionization is moved from the carboxyl moiety and transferred to the alkyl chain (Scheme 1.21).

The addition of a proton to the observed fragment ion following CID of the [M − H]⁻ can be illustrated by 12-HETE major product ion at m/z 179 (Figure 1.9B). This likely involves a charged driven mechanism where the carbinol proton at C-12 is transferred to the carboxylate anion prior to carbon–carbon bond cleavage, but this was likely preceded by a 1[5]-sigmatropic proton shift to render a saturated carbon atom allylic to the hydroxyl group as well as making a conjugated triene group from carbon-5 to carbon-10 (Scheme 1.22). The ion observed at m/z 208 (even mass) suggests that it could be a radical anion and a product of the oxy-Cope reaction indicated in (Figure 1.9B).

It is interesting to note that not all of the hydroxy arachidonate species yield this oxy-Cope reaction product in major abundance and this due to the fact that 5- and 15-HETE do not have the structural requirements necessary for the oxy-Cope reaction and that only 8, 9, and 12-HETEs have the possibility of further delocalization of the radical site over a number of carbon atoms because of the specific arrangement of the initial hydroxy group. However, only with the 12-HETE structure can the radical site be delocalized over 7-carbon atoms in the product ion. This is illustrated in Scheme 1.23 for a 1[3]-sigmatropic shift of an initially formed m/z 208 to further delocalize the radical site. This extended conjugation of the radical anion would be possible for the 8-HETE (Figure 1.9C) structure with the radical delocalized carbon atoms, but only a very small ion is observed at m/z 192. This suggests that there is additional competition for the anion charge site, resulting in shifting of the carboxyl group negative charge prior to radical cleavage. The delocalized radical would then be lost as an uncharged neutral species and therefore not observed. The 9-HETE structure would appear to fit the structural requirements for the initial oxy-Cope rearrangement, but the radical anion product could not be delocalized by extended conjugation as for the other structures.

The addition of an epoxy group into the fatty acyl chain, especially with one or more double bonds, also present in the fatty acid, introduces sufficient reactivity after collisional activation to drive a number of bond cleavage pathways. These reactions have not been systematically studied, but based upon a structural variation of product ions for many different epoxides, mechanisms have been suggested to account for the major species observed. Often the ions rationalized are indicated by drawing lines to indicate cleavage across the 3-membered epoxide ring, but details are not discussed as to the mechanisms actually operating. A likely process involves collision induced carbon–oxygen bond cleavage of the already strained epoxide (indicated as bond a and b in Scheme 1.24), leaving a radical-centered oxygen atom at one site and another site with a carbon-centered radical. The carbon-centered radical could readily abstract hydrogen atoms from methylene groups remote from the original epoxide with the most likely being four or five carbon atoms removed and in this way separate the two free radicals from each other, terminating any reversible C–O bond cleavage/formation (Scheme 1.24 proceeds after radical cleavage of the b-bond).

The oxygen-centered radical could then pair with an electron from the original carbon–carbon bond of the epoxide, cleaving that bond to a neutral aldehyde moiety and reforming a carbon-radical at the adjacent epoxide carbon atom. This radical could then pair with the remote carbon radical to cyclize the formation of a neutral 5- or 6-membered ring (Scheme 1.25).

The site of the remote carboxylate anion would determine the actual observed product ion. Initial cleavage of the a-bond (Scheme 1.24) would lead to a separate ion by the same mechanism, again in the case of Scheme 1.25 the carboxylate anion at R₁ or R₂ would determine the ion observed. The summary of these mechanisms is shown in Scheme 1.26 for a specific 9(10)-epoxyoctadeanoic acid. This type of rationalization is often depicted when the CID mass spectra of fatty acid epoxides are presented (Figure 1.10) and in fact the major product ions from collisional activation of 9(10)-epoxyoctadecanoate anion were m/z 171 and 155.¹⁹  These authors have also published that trans-9(10)-epoxystearic acid behave identically, but abundance ratios differ from that of the cis-9(10)-epoxystearic acid for m/z 155/171.

The presence of one or more double bonds in the fatty acid chain provides additional decomposition pathways. The formation of epoxides by the enzymatic reaction catalyzed by various cytochrome P-450 is a common occurrence in many cells with polyunsaturated fatty acids, rendering formation of epoxides, homoconjugated to one or more double bonds, sometimes on both sides of the epoxide. A commonly observed fragment ion comes from an additional decomposition mechanism corresponding to facile hydrogen atom rearrangement following homolytic cleavage of an epoxide carbon–oxygen bond, likely facilitated by the double bond that can participate in the hydrogen rearrangement (Scheme 1.27).

A good example of this mechanism is seen in the epoxides of linoleic acid termed leukotoxin A and leukotoxin B (Figure 1.11), which show the cleavage reactions across the epoxide ring described in Schemes 1.24–1.26, as well as the hydrogen rearrangements on the side of the epoxide homoconjugated to the double bond (Scheme 1.27). In the case of leukotoxin A, the ion m/z 183 loses a hydrogen atom during the rearrangement while for leukotoxin B a similar fragment ion at m/z 183 gains a hydrogen atom although the exact mass and elemental compositions of the same nominal m/z 183 ions are different (Scheme 1.28).

These same mechanisms of epoxide decompositions are observed for the collisional activation of the carboxylate anions of the P-450 metabolites of arachidonic acid which have been termed EETs (epoxyeicosatrienoic acids) and the CID mass spectra for the four regioisomers are presented in Figure 1.12. The cleavage reactions whose mechanisms are discussed above are indicated on the structural annotations of each mass spectrum.

An interesting ion observed following CID of 5(6)-EET appears at m/z 191 (Figure 1.12A) that does not contain the carboxylate anion and therefore arises following a charge driven reaction where the location of anionic site appears to be at carbon-7. The proximity of the 5(6)-epoxide to the carboxylate anion facilitates this latter group to attack the epoxide at carbon-5 followed by epoxide ring opening and alkoxide anion formation (Scheme 1.29). This can then form an aldehyde following scission of carbon-6, leaving the anion site at carbon-7 delocalized by the adjacent double bond. This ion further fragments to yield the hydrocarbon anions at m/z 163, 137, and 99.²⁰ 

Polyunsaturated fatty acids that contain a homoconjugated 1,4-diene moiety in the alkyl chain is a common occurrence in biology due to the mechanism of unsaturated fatty acid biosynthesis and the specific desaturases expressed in plant and animal cells. Such fatty acyl groups readily undergo radical abstraction of a methylene proton in biological systems leading to formation of a conjugated pentadienyl radical, which rapidly reacts with diatomic oxygen to form an oxygen-centered, conjugated hydroperoxy radical that eventually abstracts a hydrogen atom to form a conjugated lipid hydroperoxide species. These lipid hydroperoxides are surprisingly stable and can be isolated from biological matrices as such. Ubiquitous peroxidases and hydroperoxidases, however, often reduce these hydroperoxides to a conjugated diene fatty acid alcohol. Such conjugated alcohols can be enzymatically and chemically oxidized to unsaturated keto fatty acids.

This section discusses the mass spectrometry of both keto fatty acids and hydroperoxy fatty acids when these functional groups are conjugated with double bonds since they undergo very similar collision induced decomposition reactions even though it would appear that the functional groups are quite different. The discussions here will center around the collisional activation of negative molecular ions [M − H]⁻, but lipid hydroperoxides have been analyzed as positive [M + NH₄]⁺ ions as well.²⁰ 

The unsaturated hydroperoxy fatty acids behave virtually identically to the keto fatty acids and the only abundant ion which distinguishes these two classes of molecules is the molecular anion [M − H]⁻ in which the hydroperoxy species is 18 amu higher in mass than the corresponding keto unsaturated fatty acid. Collisional activation of the [M − H]⁻ from the unsaturated hydroperoxy fatty acids (Figure 1.13) results in facile loss of H₂O and all subsequent product ions suggest this dehydration step involves the loss of the hydrogen atom on the same carbon atom bearing the hydroperoxy moiety.²¹  This would result in formation of a ketone at that position.

An alternative mechanism would involve the loss of a hydrogen atom from another site such as a saturated carbon atom 3 or 4 removed from the hydroperoxy oxygen atom, resulting in formation of a stable epoxide, furan, or pyran, respectively. In fact, loss of H₂O to form a keto-like structure is a dominant reaction based on isotope labeling studies,²²  but formation of the ring structure appears to be a reasonable option as well. The mechanism for the formation of the keto moiety (Scheme 1.30) has been proposed based upon isotope labeling studies.²² 

The similar behavior of unsaturated keto fatty acids is illustrated in Figure 1.14 for different, but isomeric, ketones derived from linoleic acid. These molecules have been studied in some detail by collisional activation aided by the availability of isotope labeled material (three deuterium atoms at the methine carbon atoms that make up the double bonds).

The formation of the abundant product ion from both 9-hydroperoxy-octadeca-10,12-dienoate (not shown) as well as 9-oxo-octadeca-10,12-dienoate appear at m/z 197 and 185 (Figure 1.14A). These ions at first glance would appear quite unlikely since they involve cleavage between carbon atoms 10-11 (at double bond location) and in between the conjugated diene at carbon atoms 11-12 as well as proton transfer reactions. It was suggested that the mechanism of formation of these ions proceeded after an initial enolization of the oxyl group to form a vinyl alcohol based upon the isotope labeling studies.²³ 

A fragmentation mechanism for m/z 185 also consistent with the isotope labeled analogs and the sequential 1[5]-sigmatropic proton shifts of the double bonds, proceeds from the diene no longer conjugated to the keto moiety, that can readily decompose to very stable neutral species in product ion fragments by remote site fragmentation mechanisms (Scheme 1.31).

If one invokes reversibly of these double bond migrations, deuterium scrambling would result specifically for the [10,11,12-²H₃]-9-oxo-octadeca-10,12-dienoic acid that reversibly rearrange following collisional activation to 9-oxo-octadeca-13,15-dienoate anion. Reversible 1[5]-sigmatropic shift would exchange protons for deuterium labeled atoms at positions between carbons 10 and 13. The mechanism leading to the scission of the carbon 10-11 bond (m/z 185) from the transient 9-oxo-octadeca-13,15-dienoate anion could proceed by an energetically favorably ene reaction often (Scheme 1.32). The carbon-12 proton (γ-proton to oxo moiety) would be quite favorably transferred due to it being vinylic to the diene at carbon-13. The resulting ion would result in a vinyl alcohol carboxylate anion product and a neutral, conjugated triene. The same 9-oxo-octadeca-13,15-dienoate anion could undergo a separate proton transfer reaction leading to carbon 11-12 scission (Scheme 1.33).

Alternatively, the second product ion at m/z 197 would result after a further 1[5]-sigmatropic shift to form the 9-oxo-[14,16]-octadecadienoate anion that could undergo the previously discussed (Scheme 1.2) charge remote loss of H₂ forming m/z 197 as the dienone and formation of a 7-carbon, conjugated triene neutral hydrocarbon. These mechanisms are consistent with the observed ions in the deuterium labeled species that allow one to track proton transfer reactions when specific protons are deuterium atoms. However, the difference in products formed are the neutral molecules because the charge site remains on the product identical to both mechanisms.

These same mechanisms appear to be operating to some extent for the regioisomer 13-oxo-[9,11]-octadecadienoate anion after collisional activation (Figure 1.14B). After two 1[5]-sigmatropic proton shifts, the 13-oxo-[6,8]-octadecadienoate anion could undergo the facile γ-proton transfer (ene reaction) (Scheme 1.34), but this mechanism would lead to a conjugated triene on the portion of the molecule containing the carboxylate anion and the ion observed at m/z 179 which is not that abundant.²¹ 

The most abundant product ion observed for 13-oxo-[9,11]-octadecadienoic acid could also be a result of the intermediate rearranged ion discussed above that moves the conjugated diene closer to the carboxylate anion. This charge driven product ion could then arise from cleavage of carbon atoms 11-12 and the alkoxide anion formation at m/z 113 (Scheme 1.35).