Cholesteryl esters of ω-(O-acyl)-hydroxy fatty acids in vernix caseosa[S]

Cholesteryl esters of ω-(O-acyl)-hydroxy FAs (Chl-ωOAHFAs) were identified for the first time in vernix caseosa and characterized using chromatography and MS. Chl-ωOAHFAs were isolated using adsorption chromatography on silica gel and magnesium hydroxide. Their general structure was established using high-resolution and tandem MS of intact lipids, and products of their transesterification and derivatizations. Individual molecular species were characterized using nonaqueous reversed-phase HPLC coupled to atmospheric pressure chemical ionization. The analytes were detected as protonated molecules, and their structures were elucidated in the negative ion mode using controlled thermal decomposition and data-dependent fragmentation. About three hundred molecular species of Chl-ωOAHFAs were identified in this way. The most abundant Chl-ωOAHFAs contained 32:1 ω-hydroxy FA (ω-HFA) and 14:0, 15:0, 16:0, 16:1, and 18:1 FAs. The double bond in the 32:1 ω-HFA was in the n-7 and n-9 positions. Chl-ωOAHFAs are estimated to account for approximately 1–2% of vernix caseosa lipids.

breakdown products (monoacylglycerols, diacylglycerols, and free FAs) are found instead (10,11). On the contrary, the skin surface lipids of many animals are rich in diesters. Mammals like rabbit, cat, or cow biosynthesize type I diesters, whereas mouse, hamster, guinea-pig, or gerbil produce mostly type II diesters (1,2-DDE) (8,(12)(13)(14). A special sort of type II diester, with one FA replaced by isovaleric acid, has been found in the dog (15) and macaque (16). Uropygial (preen) glands of birds mostly produce 2,3-DDE (17). The existence of various types of esters in the sebum of humans and animals has been discussed in terms of the physicochemical properties of the lipids and their likely roles in the skin protection and control of the microbiome (10,11).
Recently, we reported on the analysis of molecular species of 1,2-DDE (18) from vernix caseosa. Diesters were isolated from a total lipid extract using TLC and separated by nonaqueous reversed-phase HPLC. The 1,2-DDE molecular species were identified using tandem MS with atmospheric pressure chemical ionization (APCI). In addition to 1,2-DDE, molecular species of an unknown lipid class appeared in the chromatogram at higher retention times. In this work, we isolated the unknown lipids from a fresh material using multistep chromatography and identified them as cholesteryl esters of -(O-acyl)-hydroxy FAs (Chl-OAHFAs). To the best of our knowledge, these diesters represent a new lipid class for vernix caseosa. A HPLC/ APCI-MS 2 method was developed and applied for comprehensive characterization of Chl-OAHFA molecular species in vernix caseosa.

Materials
Vernix caseosa (1-2 g) was collected from healthy newborn subjects delivered at full term (gestation weeks [39][40][41][42] immediately after the delivery. The samples were stored in amber glass vials at 80°C. The study was approved by the Ethics Committee of the General University Hospital, Prague (910/09 S-IV) and the samples were collected with a written informed parental consent.

Standard synthesis
Cholesteryl ester of 16-{[(9Z)-octadec-9-enoyl]oxy}hexadecanoic acid was synthesized by stirring (Z)-octadec-9-enoyl chloride and 16-hydroxyhexadecanoic acid in anhydrous pyridine. The oily product was purified by flash chromatography and treated with cholesterol in dichloromethane in the presence of N,N'dicyclohexylcarbodiimide and 4-(dimethylamino)pyridine (19). The final purified product was obtained in 39% overall yield and its structure was verified by NMR. The details on the synthetic procedure and NMR data are given in the supplemental data (Section 1).

Isolation of total lipids
Lipids were extracted from 20 samples equally representing gender of newborns (10 boys, 10 girls). Each sample was processed separately as follows: The sample (300 mg) was suspended in methanol:chloroform (2:1, by volume; 3 ml) in a conical-bottom glass centrifuge tube, and homogenized using a vortex shaker followed by a 2 min treatment in an ultrasonic bath. Then, chloroform (1 ml) and water (1.8 ml) were added and the suspension was shaken for 1 h. The mixture was centrifuged at 400 g for 5 min. The chloroform layer was transferred into a new glass tube, vortexed with 1.8 ml of water and collected after centrifugation. The water layer was re-extracted with 1 ml of chloroform. The chloroform extracts were combined, treated with anhydrous magnesium sulfate to remove the water residues, and filtered through precleaned cotton wool. The lipid extracts from all 20 samples were combined and concentrated on a rotary evaporator (37°C, 170 mbar) to approximately one-third of the original volume. The rest of the solvent was evaporated under a stream of nitrogen. In total, 6.0 g of vernix caseosa yielded 561.1 mg of total lipids. The lipids were reconstituted in chloroform:methanol (19:1 by vol) at the concentration of 30 mg/ml and stored at 20°C.

Fractionation of lipids
Approximately half of the total lipid extract was fractionated in two steps using semi-preparative TLC. In the first step, lipids were separated on glass plates coated with silica gel using hexane:diethyl ether (93:7, by volume) mobile phase. The zones were visualized under UV light after spraying with rhodamine 6G (0.05% in ethanol). Silica gel with diesters (R f = 0.35-0.48) was scraped off the plates and the lipids were extracted with freshly distilled diethyl ether. The solvent was evaporated under a nitrogen stream. The procedure was used repeatedly (approximately 3 mg of lipids separated in each step) and yielded 27.6 mg of diesters (F-1). In the second step, F-1 was reconstituted in chloroform:methanol (2:1, by volume) at a concentration of 30 mg/ml and separated on glass plates coated with magnesium hydroxide (for details on the TLC plate preparation, see supplemental data, Section 2). Hexane:ethyl acetate (99.95:0.05, by volume) was used as a mobile phase. Prior to the separation, a filter paper was inserted into the developing chamber. After the filter paper was fully soaked with the solvent, each plate was developed twice to focus the zones; in the first step to 3/4 of the plate height and then, after air-drying, to the top. After air-drying, the zones were sprayed with 0.05% primuline in ethanol and then visualized under UV light. A synthesized standard [18:1(n-9)/16:0-Chl] was used to verify the R f of Chl-OAHFAs. The sorbent layer corresponding to R f = 0.66-0.76 was collected and extracted with diethyl ether. The procedure was used repeatedly to process the whole F-1 (approximately 3 mg separated in each step) and yielded 5.8 mg of Chl-OAHFAs (F-2). The lipids were dissolved in chloroform:methanol (9:1, by volume) at a concentration of 5 mg/ml and stored at 20°C. The whole isolation and fractionation procedure is depicted in Fig. 1.

Transesterification
Chl-OAHFAs were transesterified using an acid catalyst (20). Briefly, F-2 was dissolved in chloroform:methanol (2:3, by volume) in a small glass ampoule. After adding acetyl chloride, the ampoule was sealed and heated at 70°C for 60 min. The reaction mixture was neutralized with silver carbonate and the organic layer was used for further analyses.

Trimethylsilylation
TMS derivatives of HFA were prepared according to a published procedure (21). FA methyl esters (FAMEs) were dissolved in dried acetonitrile (0.4 mg/ml) and treated with excess of N,Obis(TMS)acetamide at 40°C for 10 min. The solvent was evaporated under a stream of nitrogen, the residues were dissolved in chloroform (250 l), and the sample was injected onto the GC column.

Oxidation with pyridinium dichromate in DMF
HFAs were oxidized with pyridinium dichromate (PDC) in N,N-dimethylformamide (DMF) (22). The solution of FAMEs in chloroform:methanol, 2:3, by volume (100 l) was shaken with 5 mg of PDC and 50 l of DMF for 24 h at laboratory temperature. The reaction mixture was transferred into a small glass column with silica gel, and the reaction products were eluted with chloroform (2 × 200 l). The sample volume was reduced to 30 l under a stream of nitrogen.

GC/EI-MS
The analyses were performed on a 7890N gas chromatograph coupled to a 5975C mass spectrometer, equipped with EI and quadrupole analyzer (Agilent Technologies, Santa Clara, CA). The sample (2 l) was injected in the split mode with a split ratio of 5:1. The injector and transfer line temperatures were set to 350°C and 340°C, respectively. A DB-5HT fused silica capillary column (15 m × 250 m; a film thickness 0.10 m) from Agilent Technologies was used. The carrier gas was helium at a constant flow rate of 1.5 ml/min. The temperature program was set as follows: 100°C (2 min), then 6°C/min to 370°C (3 min). The ion source and quadrupole temperatures were 230°C and 150°C, respectively. EI spectra (70 eV) were recorded from m/z 20 to 700.

Direct infusion MS and HPLC/APCI-MS
Direct infusion and HPLC/MS experiments were performed using an LTQ Orbitrap XL hybrid FT mass spectrometer equipped with an Ion Max source and controlled by Xcalibur (all Thermo Fisher Scientific). The mass spectrometer was coupled to a HPLC system consisting of a Rheos 2200 quarternary gradient pump (Flux Instruments, Reinach, Switzerland), a PAL HTS autosampler (CTC Analytics, Zwingen, Switzerland), and a DeltaChrom CTC 100 column oven (Watrex, Prague, Czech Republic). Molecular species of Chl-OAHFAs were separated in nonaqueous reversed-phase HPLC and detected by APCI-MS 2 using the conditions as follows: The temperature of the sample tray was set at 10°C. The autosampler injected 5 l of the sample and the injection system was washed with chloroform/acetonitrile (1:1, by volume). Two Nova-Pak C18 stainless-steel columns connected in series (150 and 300 mm × 3.9 mm, particle size 4 m; Waters, Milford, MA) were placed in a column oven set at 40°C. The mobile phase was prepared from acetonitrile (A) and ethyl acetate (B) using the following linear gradient program: 0 min, 30% of A and 70% of B; 74.

Shorthand nomenclature
Considering previous literature (23,24), sterol esters of -(Oacyl)-hydroxy FAs are abbreviated St-OAHFAs. In the case of cholesterol, Chl-OAHFAs is used. Molecular species within this class are abbreviated using a FA/HFA-Chl format, which is consistent with the established shorthand nomenclature for OAHFAs (25) and reflects the chemical structure of Chl-OAHFAs. Thus, for instance, a molecular species identified in this work as 16:0/32:1-Chl corresponds to cholesteryl ester of HFA with 32 carbons and 1 double bond, with the hydroxyl esterified to a saturated FA with 16 carbons. In the ion description, "Chl" is considered a neutral cholesterol molecule. Thus, for instance, protonated OAHFA, which is an ion resulting by neutral loss of dehydrated cholesterol from Chl-OAHFA (M),

General structure elucidation
As shown in our previous report (18), the unknown lipids eluted in the reversed-phase HPLC at higher retention times than aliphatic 1,2-DDEs. For protonated molecules, high-resolution APCI mass spectra revealed elemental compositions, C n H 2nx O 4 , where x = 11, 13, 15, or 17. The existence of four oxygen atoms and similar retention on silica gel as 1,2-DDE pointed out diester lipids. The formulas corresponded to ring plus double bond equivalents (RDBEs) of 6.5-9.5, showing an unusually high degree of unsaturation and/or presence of rings (the RDBE values for 1,2-DDEs ranged from 1.5 to 3.5). Protonated molecules of unknown lipids were found at higher m/z values (m/z 1,000-1,250) than 1,2-DDE (m/z 800-1,000). The APCI spectra of all molecular species showed neutral loss of 368 Da, typical for sterol-containing compounds. The occurrence of a sterol corresponded well with the high RDBE values. The presence of a sterol in the unknown diesters and its absence in 1,2-DDE made it possible to separate these two diester lipid classes from each other using TLC on magnesium-based sorbents (26,27). Commercial production of such sorbents was mostly discontinued, which forced us to use common reagent-grade chemicals. Nevertheless, a fraction of unknown sterol-containing diesters with satisfactory purity was obtained. In the next step, the unknown lipids were transesterified and analyzed by high-temperature GC/EI-MS and high-resolution MS. The GC/EI-MS data showed cholesterol and a rich mixture of FAMEs (supplemental Fig. S4). A group of FAMEs at unusually high retention times was identified as methyl esters of HFAs (HFAMEs). Their TMS derivatives (Fig. 2) provided spectra consistent with a hydroxy group at the terminal carbon (28,29) and revealed saturated and monounsaturated HFAs with 29-34 carbons. Interestingly, saturated HFAs eluted in two chromatographically separated peaks, obviously differing by chain branching. The mass spectra of earlier eluting isomers of 30:0 and 32:0 showed significantly more abundant m/z 103, which could indicate methyl branching in the iso position. The chromatographic peaks of TMS derivatives of monounsaturated HFAMEs were distorted, likely because of double bond positional isomers. The original transesterified mixture was also analyzed by high-resolution APCI-MS (supplemental Fig. S5a). The positive ion spectra displayed abundant m/z 369.3510 (cholesterol; [Chl + H  H 2 O] + ) and showed ions consistent with protonated molecules of saturated and monounsaturated hydroxy FAMEs (exact masses within 2 ppm error). To confirm the position of the hydroxy group in an independent experiment, oxidation with PDC in DMF was performed. As known from the literature (22), the oxidation of primary alcohol gives carboxyl, whereas the reaction of secondary alcohol provides a carbonyl group. High-resolution electrospray mass spectrum of negative ions showed reaction products with one extra oxygen, i.e., carboxyl derivatives (supplemental Fig. S5b). Therefore, the hydroxy group on the terminal carbons of HFAs was confirmed. Double bond position in unsaturated HFAs was investigated using our previously developed method based on gas phase reactions of acetonitrile (30)(31)(32). CID MS/MS spectra of the [M + 55] +• adduct of HFAME (32:1) revealed double bonds in two positions, n-7 (more abundant) and n-9 (less abundant), see supplemental Fig. S5c. The existence of two isomers corresponded well with the distorted peaks in GC chromatogram discussed above. The entire analytical strategy used for the lipid class identification is summarized on a flowchart in supplemental Fig. S6.
All the results directed us to a hypothesis of diesters composed of HFAs having cholesterol attached to the carboxyl and common FA to the hydroxyl. The unknown diester lipids in vernix caseosa were identified as Chl-OAHFAs with the general structure shown in Fig. 3.

Optimization of HPLC/APCI-MS
The mass spectra of a synthetic standard were studied with the aim to develop an HPLC/MS method for comprehensive characterization of Chl-OAHFA molecular species. The APCI spectrum of 18:1(n-9)/16:0-Chl in the positive ion mode (Fig. 4A) (Fig. 4D). Unluckily, the fragmentation channels proceeded almost exclusively via elimination of water. Thus, positive ion mode turned out to be useless for structure elucidation of Chl-OAHFAs. The standard did not provide [M  H]  in the negative ion mode because of the absence of groups prone to deprotonation. Fortunately, elevated temperature in the ion source induced thermal degradation to OAHFA that easily deprotonated in the corona discharge (Fig. 5A). The process was not very efficient under the conditions used and required optimization of the ion source parameters. The ion source temperature and the mobile phase flow rate, as well as ion optics voltages, were tuned to maximize signal of deprotonated OAHFA (supplemental data, Section 6). Fragmentation of deprotonated OAHFAs is known to give structural information on FAs and HFAs; the MS/MS spectrum of 18:1(n-9)/16:0 (Fig. 5B) showed the same fragments as in previously published work (33). The chromatography was optimized using the lipid sample from vernix caseosa, with the aim to achieve the highest possible resolution for molecular species in a reasonable time. Based on our previous experience with 1,2-DDEs (18), a nonaqueous reversed-phase system with the two Nova-Pak C 18 columns connected in series with a total length of 45 cm was developed. Various binary mobile phases containing methanol, propan-2-ol, acetonitrile, ethyl acetate, and acetone were studied. Finally, we ended up with separation conditions similar to those used for 1,2-DDE (18), i.e., a linear increase of ethyl acetate in acetonitrile in 140 min. The flow rate of the mobile phase in the elution window of Chl-OAHFAs was reduced from 0.6 to 0.15 ml/min to increase detection sensitivity.

Chl-OAHFAs in vernix caseosa
The base peak chromatogram of Chl-OAHFAs showed many overlapping peaks (Fig. 6A). The extracted chromatograms displayed up to several peaks for each m/z value, obviously representing isomers. The elution order followed the equivalent carbon number concept (34); retention increased with the length of the chain and decreased with the number of double bonds. As shown in Fig. 7, the molecular species appeared on a band rising almost linearly with the retention time (at higher retention times the curve rose exponentially as the flow rate increased from 150 l/min back to 600 l/min).
The molecular species of Chl-OAHFAs were detected and identified using APCI-tandem MS with data-dependent scanning. The high-resolution full-scan mass spectra of positively charged ions (  In total, 295 molecular species of Chl-OAHFAs were fully characterized in 59 chromatographic peaks, and Chl-OAHFAs in an additional 11 peaks were characterized by the total number of carbons and double bonds. The list of the 50 most abundant species is given in Table 1 (for all identified Chl-OAHFAs see supplemental Table S1). The relative proportions of identified molecular species were estimated from peak areas integrated in the chromatograms reconstructed for [M + H] + and relative intensities of deprotonated FA ions in MS 2 spectra. It is important to note that response factors of lipids depend on the number of double bonds and carbon chain length (35,36). As neither standards nor response factors were available for quantification, the relative proportions must be considered merely as an estimate.

DISCUSSION
Indirect evidence of cholesterol-containing diesters in vernix caseosa appeared in 1969 when a subclass of type I diesters [also called "type III diester waxes" (11)], composed of HFAs esterified on the hydroxyl group with an unsubstituted FA and on the carboxyl group with a sterol, was proposed (9). However, a diester lipid class with a sterol moiety has never been substantiated in vernix caseosa. Our work brings evidence that the sterol-containing diesters are, in fact, Chl-OAHFAs comprised of long-chain HFAs.
Ester lipids containing HFAs are important constituents of meibum, which is a secretion of holocrine meibomian glands in the eyelids of humans and most animals. Meibum lipids form the outermost layer of the tear film that protects the ocular surface from desiccating and bacterial infection (42). The existence of long-chain HFAs in meibum is known from the early eighties when they were found in steer and human samples (43,44). Mostly monounsaturated straight-chain HFAs with 30-36 carbons formed approximately 10% of the total acids. Soon after, St-OAHFAs (Type I-St) were isolated and identified in steer and human meibum. It was the first report on these diesters in animal samples (45). In addition to the HFAs, meibomian St-OAHFAs appeared to contain predominantly FAs with 18:1 and 16:1 chains and cholesterol (60%) or lathosterol (35%). St-OAHFAs formed approximately 5% of steer meibomian lipids. The interest in St-OAHFAs increased again after discovering OAHFAs in meibum (33). The OAHFAs are likely either precursors or degradation products of St-OAHFAs, and they are considered to be one of the amphiphilic compounds that separate and stabilize the interfacial layer between the very hydrophobic lipids of meibum and the aqueous layer of the tear film (46). They are linked to dry eye disease and might represent biomarkers of the disease progression (47). The complex mixture of OAHFAs in human meibum contains molecular species composed mostly of monounsaturated FAs with 18 and 16 carbons and 28:1-34:1 HFAs (46, 48). Recently, 61 OAHFAs with 34-56 carbon atoms and one to seven double bonds in human meibum have been reported (25). The occurrence of St-OAHFAs in meibum was reconfirmed by the analysis of their intact molecules (49). The OAHFA moieties of St-OAHFAs and OAHFAs in meibum are closely related (50). Their abundances are also similar, with each lipid class accounting for 3-5% of meibomian lipids (24,47). St-OAHFAs have also been found in meibum of several animal species, including dogs, mice, and rabbits (23). Very recently, OAHFAs similar to the meibomian ones have been detected in equine sperm and the essential role of these amphiphilic lipids in sperm function was hypothesized (51). It is worth noting that mammals also biosynthesize isomeric OAHFAs derived from HFAs with nonterminal hydroxyl. These recently discovered endogenous lipids (also known as fatty acyl esters of hydroxy FAs) exhibit antiinflammatory and anti-diabetic effects (52,53).
In this work, Chl-OAHFAs comprised approximately 20% of the nonpolar diester fraction. As diesters form 3-9% of the total lipids (6, 7), Chl-OAHFAs are estimated to account for approximately 1-2% of vernix caseosa lipids, which is less than in human meibum (3-5%) (24,45,47). Chl-OAHFAs were detected and comprehensively characterized from a pooled sample representing 10 male and 10 female full-term babies. Moreover, some experiments were also performed with samples from individual subjects (data not shown). Chl-OAHFAs were detected in all samples at a similar level, suggesting that these esters are commonly present in vernix caseosa and that they are not related to fetus gender. Our experiments showed that the sterol moiety in Chl-OAHFAs is represented by cholesterol; lathosterol, reported in addition to cholesterol in meibomian St-OAHFAs (45), has been detected in this work only in traces. HFAs in Chl-OAHFAs from vernix caseosa resembled those existing in meibomian St-OAHFAs and OAHFAs. The most abundant HFAs in vernix caseosa (32:1, 34:1, and 30:1) were also the most abundant in St-OAHFAs (44,45) and OAHFAs (25) from meibum. High proportions of these HFAs have also been reported for equine sperm (51). Saturated HFAs of meibum are both straight chain and methyl branched (44), which is also in agreement with our findings. The level of similarity can be illustrated by i-30:0 and i-32:0 HFAs identified in vernix caseosa based on the mass spectra of their TMS derivatives (Fig. 2). The earlier work showed that meibomian even-carbon HFAs are iso-methyl branched, whereas oddcarbon HFAs are predominantly of the anteiso type (44). As regards double bond position in unsaturated HFAs in meibum, the available data are scarce. The action of 9desaturases was proposed (44) and n-9 and n-7 double bonds were assumed (46). We clearly demonstrated that the double bonds in 32:1 HFAs in vernix caseosa are in  Table S1. a The values of relative peak area were calculated in two steps. In the first step, full scan trace in the positive ion mode was used to create reconstructed ion chromatograms for individual m/z values ([M + H] + ). The peak areas were integrated and put in relative values. In the second step, relative proportions of the isomers within each peak were established from relative intensities of the deprotonated FA ions in the negative ion MS/MS spectra. b Molecular species with the relative peak area higher than 1.5% are bolded.
both positions, n-7 (more abundant) and n-9 (less abundant). As regards FAs, the largest signals provided Chl-OAHFAs with 14:0, 15:0, 16:0, 16:1, and 18:1 chains, which is not surprising for a sample of vernix caseosa origin (54,55). The distribution of FAs in Chl-OAHFAs ( Fig. 8A) was similar to the overall representation of FAs in the vernix caseosa lipidome reported earlier (56) (compare supplemental Fig. S8). The higher levels of 16:1 and 18:1 observed for Chl-OAHFAs has to be interpreted with caution because, as discussed above, APCI-MS is somewhat more sensitive to unsaturated lipids (35,36). Nevertheless, the results indicate that Chl-OAHFAs are biosynthesized from a pool of all FAs, likely with a certain preference for 16:1 and 18:1 chains. The meibomian St-OAHFAs and OAHFAs tend to incorporate fatty acyls with 16 and 18 carbons more exclusively (24,25,45,47). Meibomian OAHFAs have been reported to comprise a relatively large proportion of FA 18:2 (24), which is not the case of vernix caseosa Chl-OAHFAs. Nevertheless, 18:2 was the most abundant diunsaturated FA in Chl-OAHFAs from vernix caseosa. Considering the intact molecules, this study showed the ratio of unsaturated and saturated Chl-OAHFAs as 89%:10%, which is in a good agreement with meibum (91 ± 3%:9 ± 1%) (57). The main species in vernix caseosa are represented by 15:0/32:1-Chl, 14:0/32:1-Chl, 18:1/32:1-Chl, 16:1/32:1-Chl, and 16:0/32:1-Chl, which reflects the higher incorporation of saturated shorter chain FAs than in meibum. The biological role of Chl-OAHFAs in vernix caseosa remains to be clarified. The diesters are structurally related to cholesteryl esters, important constituents of neutral lipids in vernix caseosa (6). Chl-OAHFAs might be a storage or inactivated form of OAHFAs, similarly to cholesteryl esters of very long chain FAs and OAHFAs in meibum (46,58). However, it is not known whether nonesterified OAHFAs exist in vernix caseosa (or eventually in amniotic fluid) or not. One can imagine that the amphiphilic properties of OAHFAs might be important for proper functions and/or cohesion of vernix caseosa. It seems reasonable to assume that Chl-OAHFAs are products of the sebaceous gland, taking into account that they are produced by a meibomian gland (a type of sebaceous gland).  Table S1 (the extended version of Table 1).
Sebaceous glands in the developing skin are a presumed source of structurally related cholesteryl esters in vernix caseosa (2). On the other hand, OAHFA structural motives in certain ceramides (5,37,38) cannot be overlooked, and the epidermal origin of Chl-OAHFAs should be considered as well.
We finally wish to briefly comment on the analytical methodology used in this work. Although positive ion mode of APCI-MS provided intense signals of protonated Chl-OAHFAs, it has turned out to be unusable for detailed structural elucidation because of virtually missing fragments related to FAs and HFAs. Negative ion mode is unlikely to be efficient for diester-type structures lacking easily deprotonatable groups. Although deprotonated molecules of St-OAHFAs have been reported (49), we found the negative ion mode to be insensitive. On-line thermal degradation of Chl-OAHFAs to OAHFAs in the APCI source worked well and made it possible to detect about 300 molecular species of Chl-OAHFAs, by far the highest number of reported Chl-OAHFAs to date.