On the Presence and Role of Polar Lipids in Meibum

I read with interest the paper by Chen et al. ¹ published online on July 29, 2010. The paper is concerned with shotgun (or direct-infusion) lipidomic analysis of normal human meibum by means of electrospray ionization mass spectrometry. I noticed with pleasure that Chen et al., along with providing new and important information on the meibomian lipidome, largely confirmed our results reported in articles published from 2007 to 2010 and presented at ARVO Annual Meetings from 2006 to 2010 with regard to the lipid composition of normal human meibum, especially its wax esters (WEs), ^{2 –4} cholesteryl esters (CEs), ^5,6 (O-acyl)-omega-hydroxy fatty acids [OAHFAs]), ³ and phospholipids. ^2,3,7,8 I was also pleased that the authors arrived at the same idea that had been proposed, explained, and illustrated in a review paper published in 2009 about the putative role of OAHFAs as amphiphilic anionogenic compounds that may facilitate the spread of meibum on the ocular surface and thus be responsible for the tear film (TF) and TF lipid layer stabilization. ⁸ This function had been routinely ascribed to phospholipids, which, according to new data, apparently are not present in meibum in quantities sufficient to play such a role. 

However, I could not help but notice a few items that I want to comment on, as they are important for future progress in this area of ocular science, and, if left uncorrected, could lead to confusion and erroneous physiological conclusions. 

First, Chen et al. ¹ claimed that they observed a large presence of free fatty acids (FFAs) in normal human meibum, specifically those with C20 to C28 chain lengths. Of importance, the mass spectrometry (MS) signals of these FAs were observed in direct-infusion experiments, which was the basis for the claim. There is no doubt that these signals were correctly identified as those of FAs. However, the origin of these FA signals in normal human meibomian lipid samples seems to have been misjudged. Contrary to the statement by Chen et al., I am confident that these signals of FAs originated from a range of more complex lipids (such as CEs, triacylglycerols [TAGs], and di- and triesters), due to the spontaneous in-source losses of FFA residues during the MS analysis (Fig. 1). This effect is well known in MS of TAGs (which are known to spontaneously produce strong signals of their deacylated derivatives) and was recently discussed and used to identify and quantitate various CEs in human meibum. ⁶ The in-source behavior of these lipids is such that under the conditions of positive and negative ion mode atmospheric pressure ionization MS, part of those complex lipid esters lose some of their FA residues, which then become visible as ions of FFAs. This loss may be further facilitated by ammonium hydroxide, which was a reagent of choice in the experiments of Chen et al. However, in no way can these data be interpreted as a confirmation of the presence of the stated amounts of FFAs in the meibomian samples (up to 3% of the meibum weight, per Chen et al.) and thus should be treated as an experimental artifact. Notably, the FA signals reported by Chen et al. closely match those of the CEs reported in our recent paper ⁶ and the data on CEs presented by Chen et al. themselves. The other major FA signals, those of C16- and C18-FAs in particular, most likely originated from TAGs and even more complex lipids such as di- and triesters. Corroborating this interpretation of the data of Chen et al. are the results of our HPLC-MS experiments, in which only minor signals of true FFAs were observed, with their total amount being typically <0.1% of meibum (weight/weight), if present at all. We anticipated these problems with the inadvertent in-source fragmentation of complex lipids at the very beginning of our project. Thus, our HPLC-MS protocols differ from the shotgun experiments of Chen et al. in that the physical separation of meibomian lipids during the HPLC step occurred before the MS analysis. This approach allowed us to reliably distinguish between the signals of true FFAs present in meibum and those of other lipids, including those that are prone to spontaneous in-source fragmentation. In our experiments, the reverse-phase HPLC retention times of CEs, TAGs, and OAHFA (between 10 and 25 minutes) were dramatically different from those of the FFAs (retention times, ∼3–5 minutes), which made their correct structural assignments an easy task. Clearly, without the HPLC separation step, Chen et al. could not unequivocally determine the origin of these FFA signals and described them as true FFAs present in meibum, rather than the in situ–generated fragmentation products of more complex lipids. In addition, the HPLC-MS methods adopted in our laboratory are designed to provide quantitative results, as we use standard curves to quantify the analytes where possible (see, e.g., our recent paper on CEs ⁶ ). Our HPLC-MS approach used for FFA quantitation showed that their total content in normal human meibum, though varying from sample to sample, was typically <0.1% (weight/weight). 

Figure 1.

 

LC-MS analyses of human meibum and free fatty acids standards conducted in negative ion mode. Reverse phase HPLC separation of analytes on a C₁₈ column (Hypersil Gold, 2.1 × 150 mm, 5 μm) in an acetic acid–containing eluent with atmospheric pressure chemical ionization MS detection was conducted as described earlier. ⁶ (A) Total ion chromatogram of FFA standards recorded in the m/z range of 200 to 550. All FFA were detected as adducts with acetic acid (M + CH₃COO⁻). Nine saturated and unsaturated FFA were tested as an equimass mixture: palmitoleic acid (C_16:1, m/z 313), palmitic acid (C_16:0, m/z 315), linolenic acid (C_18:3, m/z 337), linolenic acid (C_18:2, m/z 339), oleic acid (C_18:1, m/z 341), stearic acid (C_18:0, m/z 343), arachidic acid (C_20:0, m/z 371), docosanoic acid (C_22:0, m/z 399), and tetracosanoic acid (C_24:0, m/z 427). (B) Extracted ion chromatogram of tetracosanoic acid (C_24:0, m/z 427, retention time RT = 4.4 min). (C) Averaged mass spectrum of the chromatographic peak of FFA standards eluted between 2 and 6 minutes. (D) Total ion chromatogram of normal human meibum recorded for the m/z range of 200 to 550. Note extremely small HPLC peaks with RT of 2.7 and 4.5 minutes in the elution region of FFA, and very intense HPLC peaks in the elution region of complex lipids. ⁶ (E) Extracted ion chromatogram of ion m/z 427 (tetracosanoic acid, C_24:0) detected in meibum. Note that the ion eluted as two separate HPLC peaks: the first peak had RT of 4.2 minutes and coeluted with authentic tetracosanoic acid standard, while the second peak had an RT of 18.1 minutes, similar to those of cholesteryl esters. ⁶ Also note that the HPLC peak area of free tetracosanoic acid in meibum (peak area 13866350 arbitrary units) is only 2.6% of the one of tetracosanoic acid formed in-source from more complex lipids due to their spontaneous fragmentation (546954237 arbitrary units). The absolute amounts of true FFA in meibum were quantified using calibration curves obtained with authentic FFA standards. (F) Averaged mass spectrum of the chromatographic peak of meibomian FFA eluted between 2 and 6 minutes. (G) Averaged mass spectrum of the chromatographic peak of complex meibomian lipids eluted between 10 and 25 minutes. Note that the spectra presented in panels (F) and (G) are quite similar.

Thus, the FA signals reported by Chen et al. most likely were in situ–generated artifacts—namely, products of the spontaneous in-source fragmentation of more complex meibomian lipids such as CEs, TAGs, and di- and tri-esters. This observation is critical for our understanding of the chemical composition of normal human meibum and the role of FFAs in tear film physiology and pathology. 

Another related problem seems to be the presence of a mixture of isobaric C16:1- and C18:1-based WEs reported by Chen et al. According to their report (see Fig. 3 of their paper online), C16:1-WE could be present in quantities approaching those of isobaric C18:1-WE. This was not the case in our HPLC-MS experiments, in which the palmitoleic acid–based WEs were indeed detected, but in quantities typically below 10% of their C18:1-based counterparts. In fact, the typical presence of C16:1-WE was so small that an analysis was deemed to be unnecessary at the time. Again, the HPLC step greatly facilitated identification of the WEs. Thus, the claimed observation of the intense product peaks of C16:1-WE could be partly a result of the presence of unknown isobaric compounds (not necessarily of the WE nature) with palmitoleic acid in their structures that had not been chromatographically separated from WE before the MS analysis. 

A few minor comments concern mostly the diligence in referring to the recent work done in the area. (1) The FA composition of meibomian CEs described by Chen et al. ¹ closely matched the one that had already been described in an earlier paper on the topic, ⁶ where exactly the same major saturated FAs (C24:0, C25:0, and C26:0) were reported and quantified. (2) The possible role of OAHFAs, exactly as described by Chen et al., had already been described in a recent review. ⁸ Unfortunately, these two papers were not referenced and discussed by Chen et al. 

In conclusion, it seems that the experimental approach described by Chen et al., though very informative, still has limitations that can lead to inadvertent but still erroneous interpretation of the data: In this case, a 10- to a 100-fold overestimation of the presence of endogenous FFAs and a large overestimation of the presence of C16:1-WE in normal human meibum. 

Footnotes

 Supported by NIH Grant R01EY019480.