Differential effects of dietary cholesterol and triglycerides on the lipid homeostasis in Meibomian glands

Abstract

Exocrine Meibomian glands (MG) play a central role in the ocular surface physiology by producing meibum – a lipid secretion composed of cholesteryl esters (CE), cholesterol (Chl), triacylgycerols (TAG), waxes and other types of lipids. MG were previously shown to synthesize Meibomian lipids (ML) in situ via a complex array of reactions termed meibogenesis. However, questions remain about the role of dietary lipids in meibogenesis. To establish if dietary Chl (DC) and TAG (DT) can participate in meibogenesis, we studied mice whose diet was supplemented with trace amounts of deuterated Chl (²H-Chl) and ¹³C-labeled triolein (¹³C-TO), and the products of their biosynthetic transformations were analyzed using LC/MS. We demonstrated that ²H-Chl, but not ¹³C-TO, could be directly incorporated into meibum. Furthermore, ²H-Chl was esterified into MG-specific ultra long ²H-CE, which were vastly different from plasma CE and ²H-CE. The measured ²H-Chl/Chl and ²H-CE/CE ratios in meibum increased in a time-dependent manner reaching ~5% and ~1.2%, respectively. The ²H-Chl/²H-CE ratio was about 3.5x higher than that for endogenous unlabeled Chl and CE, indicating accumulation of ²H-Chl in meibum. The elongation pattern of Meibomian ²H-CE closely replicated that of unlabeled CE. On the other hand, ¹³C-TO was not detected in any of the ML samples as an intact lipid or its metabolized/hydrolyzed products. We conclude that DC can be directly esterified into MG-specific CE, while DT undergo extensive catabolic transformations before reaching MG. These findings demonstrate that DC can have a direct impact on MG and ocular surface lipid homeostasis and pathophysiology.

1. INTRODUCTION

Lipid homeostasis in the exocrine lipid-secreting Meibomian glands (MG) [1, 2] is maintained via a network of concerted biosynthetic reactions which was termed meibogenesis [3, 4]. The MG secretion (also known as meibum [5]) is a complex mixture of a large number of various wax esters (WE), free cholesterol (Chl) and cholesteryl esters (CE), triacylglycerols (TAG) and other, more complex and unique, lipids [3, 6–8]. There are several features of Meibomian lipids (ML): most of their fatty acids (FA) and fatty alcohols are of extremely long chain nature (up to C34 or so), often iso- and anteiso-branched, and odd-chain [5, 9, 10]. Under normal conditions, the lipid composition of meibum is extremely invariable [3, 11, 12] which might be one of the most important requirements for the proper ocular surface homeostasis: As shown recently, changes in ML profiles in several mouse knockout models that targeted genes of meibogenesis led to severe MG and ocular abnormalities [13–17]. About a third (w/w) of ML is comprised of CE in which Chl is esterified to saturated and unsaturated FA ranging from C₁₄ to C₃₄. Importantly, the CE composition of meibum shows only minor inter-donor variability in humans [12, 18] and mice [11]. The conservative nature of meibum implies a tightly controlled biosynthetic environment in MG. Recent transcriptomic studies revealed that MG of humans and mice express a range of genes that are known to encode FA elongases, desaturases, reductases, WE and CE synthases, and other enzymes directly related to lipid metabolism in general [3, 11, 12]. Among those, key genes of Chl biosynthesis – such as Dhcr7, Dhcr24, Lss, Fdft, and others, – were found to be expressed at high levels. Thus, it seems that MG are equipped with the molecular machinery that is capable of biosynthesizing Chl in situ and de novo and, in normal conditions, maintaining its proper levels. However, in certain pathologies such as chalazia [19, 20] and high triglyceride-low wax syndrome (HTLW) [14] the levels of Chl in MG and meibum were considerably increased. Also, population-wide studies revealed that incidences of Dry Eye – a common pathology that affects hundreds of millions of people worldwide – were linked to systemic lipid disorders such as dyslipidemia, e.g. high levels of Chl and/or TAG in human blood [21, 22]. Thus, it was reasonable to speculate that increased levels of Chl in meibum and MG may be a factor in the onset and development of those pathophysiological conditions.

However, the role of blood Chl that may originate either from diet or from other biosynthetic loci (such as intestines and liver) in meibogenesis remains unknown. Also unknown is the origin of TAG in meibum. Earlier, we reported prominent differences between TAG of normal meibum and those in human blood [23]. To determine if there is a direct link between dietary fat (Chl and TAG) and meibum, we conducted this pilot study using deuterated free Chl and ¹³C-labeled triolein as food supplements.

2. MATERIALS AND METHODS

1.1. Labeled lipids:

Cholesterol-2,2,3,4,4,6-d₆ (d₆-Chl; Cat. #488577-CONF) and glyceryl tri(oleate-1,2,3,7,8-¹³C₅) (¹³C₅-TO; Cat. #772941-CONF) were purchased from Isotec, Inc. (Miamisburgh, OH, USA) via SigmaAldrich/MilliporeSigma, St. Louis, MO, USA).

1.2. Other reagents:

Lipids standards were obtained from MilliporeSigma and Nu-Chek Prep., Inc. (Elysian, MN, USA). Commercial grade peanut oil was used as a vehicle for experiments with labeled lipid tracers. Organic solvents for analytical experiments were of HPLC or MS grade. The solvents were obtained from SigmaAldrich, ThermoFisher (Waltham, MA, USA), or Burdick and Jackson (Muskego, MI, USA). Formic acid and ammonium formate were of >99.99% purity from SigmaAldrich.

1.3. Animals and Animal Procedures:

Animal procedures used in this study were approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center (UTSW) and were conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. Two months old C57Bl/6J male mice were purchased from Wakeland Mouse Breeding Core facility at UTSW. The animals were kept in the Animal Research Center (ARC) of UTSW under supervision of ARC technicians and veterinarians on a 12-h light/dark cycle with free access to water and food (the 2016 Teklad global 16% protein rodent diet; from Envigo, Indianapolis, IN). During the experiments with tracers, the animals were gavaged daily at around 8 am with 0.1 mL of commercial grade peanut oil as vehicle with 5 mg/mL d₆-Chl and 5 mg/mL ¹³C₅-TO as labeled tracers.

1.4. Tissue Collection and Sample Preparation:

Treated mice were euthanized using ketamine/xylazine cocktail at Days 7, 14, and 28 (n=4 for each time point) 2 hrs post-gavage and tarsal plate (TP) tissue samples were collected for lipidomic analysis as described earlier [14]. Four untreated mice were used as controls (Day 0 samples). Briefly, TP (four from each mouse) were excised under a surgical microscope, dissected free from other tissues, and placed in 0.5 mL of chloroform/methanol = 2/1 (v/v) solvent mixture for an overnight storage at −20°C. Then, the lipids were extracted from TP three times using the same volumes of the same solvent mixture, all extracts were pooled, the solvent was evaporated under a stream of nitrogen at 36°C using a FlexiVap WorkStation evaporator (from Glas-Col, Terre Haute, IN, USA), and the oily residue was dissolved in iso-propanol (0.2 mL per 4 TP). The samples were stored in sealed vials at −80°C before the analyses.

Mouse blood samples (between 0.4 and 0.8 mL) were collected by cardiac puncture after the mice were euthanized and placed in BD Vacutainer 13×75 mm 3 mL whole blood collection tubes with K₂EDTA Additive (from Becto, Dickinson and Co., Franklin Lakes, NJ, USA). The tubes were inverted multiple times to allow for mixing of the sample with the additive, placed on ice for an hour, and then centrifuged at 4°C and 1,500 g for 10 min to obtain plasma. The plasma lipids were extracted using a published protocol [24] by mixing a 100μL plasma specimen with 2 mL of a methanol/methyl-tert-butyl ether/chloroform = 1.33/1/1 (v/v/v) solvent mixture placed in glass tubes with PTFE-lined caps, vortexing for 30 sec, shaking for 20 min at room temperature, and centrifuging at 1,000 RPM for 30 min. Then, the supernatant was collected, the solvent evaporated as described above for Meibomian lipids, the final oily residue dissolved in 1 mL of iso-propanol and stored in crimped 2-ml autoinjector vials at −80°C before the analyses. No sample deterioration was observed in stated conditions for at least three months.

1.5. Analytical Instrumentation and Procedures:

Extracted intact lipids were analyzed “as is” without any further manipulations. A previously published liquid chromatography/mass spectrometry (LC/MS) protocol was used [11]. A high resolution Synapt G2-Si QToF mass spectrometer was equipped with a ZSpray atmospheric pressure ionization LC interface, an IonSABRE-II atmospheric pressure chemical ionization (APCI) ion source and a LockSpray unit with 2μg/mL Leucine-Enkephalin as calibrant (all from Waters Corp., Milford, MA). An Acquity UPLC C18 BEH column (1 × 100 mm, 1.7 μm) was used. Between 0.5 and 1 μL of sample was injected. Chromatographic separation of analytes was performed on an Acquity M-Class binary ultra-high performance liquid chromatograph (also from Waters), in reverse-phase (RP) mode at a flow rate of 20μL/min at 30°C. A linear binary acetonitrile/iso-propanol gradient of two solvent mixtures was used: Solvent A, composed of an acetonitrile : iso-propanol = 45 : 55 (v/v) solvent mixture (95%) and 10mM ammonium formate in water (5%), and Solvent B, comprised of a mixture of iso-propanol (95%) and 10mM ammonium formate in water (5%). The elution cycle started with 100% of Solvent A and progressed to 100% of Solvent B over the period of 50 min. Then, the column was washed with Solvent B for 10 min and re-equilibrated with Solvent A for 15 min, which completed the analytical cycle.

The lipid analytes were detected in positive ion mode, mostly as highly visible (M + H)⁺, (M − H₂O + H)⁺ and (M – fatty acid + H)⁺ adducts. The following settings of the APCI detector were used: Sensitivity mode with resolution 10,000 FWHP; scan rate of 1 scan/sec; corona current 1.7 μA; probe temperature 400°C; source temperature 150°C; sampling cone 30V; cone gas flow (nitrogen) 100 L/hr; desolvation gas flow 400 L/hr; nebulizer gas flow 3 Bar; trap gas flow (helium) 2 mL/min.

1.6. Transcriptomic analysis of mouse tarsal plates:

Mouse TP specimens from five untreated wild-type animals were subjected to transcriptomic analysis using mRNA microarrays (Clariom™ D from Affymetrix, Santa Clara, CA, USA) as described recently [3, 11]. Analysis of data was performed in Transcriptomic Analysis Console (v.4.0.2; from Affymetrix, Inc./Applied Biosystems/ThermoFisher Scientific, Santa Clara, CA, USA). The expression levels of genes were computed as log(2) values on a 0–20 scale. The levels of housekeeping and reference genes Chmp2A, Emc2, Gpi1, Psmb4, Rab4, Reep5, Snrpt3, Vcp, and Vps29 were used as internal controls [11, 25].

1.7. Data Computation and Statistical Analyses:

Experimental results were analyzed using MassLynx software package v.4.1. (from Waters) and SigmaStat v.3.5 (from Systat Software, Inc., San Jose, CA, USA). Results obtained for treated and non-treated groups were compared using Student’s t-test. The p-value of less than 0.05 was considered statistically significant.

3. RESULTS

Two-month old wild type C57BL/6J male mice were selected for the study. Previously, we demonstrated that at that age mice of both sexes had no MG abnormalities, produced meibum of the same composition, and the transcription levels of meibogenesis-related genes were identical implying no gender-specific effects at that age [11]. The choice of the vehicle (peanut oil) was based on a favorable solubility of labeled lipids in the oil and its sufficient stability to oxidation in storage conditions at room temperature [26, 27].

The overall design of the experiment is shown in Fig. 1. Lipid tracers – d₆-Chl and ¹³C₅-TO – are stable, isotopically labeled forms of natural Chl and TO that were expected to be absorbed from blood with the same rate and efficacy as their non-labeled counterparts. This type of labeling allowed for an easy detection of tracers in the study samples. d₆-Chl provided a characteristic signal of its dehydrated product (M − H₂O + H)⁺ with a m/z 375.385 ± 0.006 amu (theoretical m/z value of 375.3892), while ¹³C₅-TO produced a (M + H)⁺ adduct with m/z 900.8372 (theor. 900.7911) and two major fragments (M − FA + H)⁺ with m/z 613.5698 (theor. 613.5352) and (M − 2FA + H)⁺ with m/z 344.3021 (theor. 344.2899). These signals were used to detect the tracers in study samples.

Figure 1.

Lipid tracers and the overall design of the experiment. (A) Chemical structure of 1,2,3,7,8-¹³C₅-glyceryl trioleate (¹³C₅-TO) and 2,2,3,4,4,6-d₆-cholesterol (d₆-Chl). (B) Timeframe of sample collection and lipid analysis. (C) Sample preparation and analysis steps. Red arrow points at tarsal plates in the sample vial. (D) High resolution time-of-flight APCI mass spectrum of ¹³C₅-TO obtained in positive ion mode (TOF MS AP+). Note spontaneous in-source formation of (M + H)⁺, (M − FA + H)⁺, and (M − 2FA + H)⁺ adducts that were used to detect ¹³C₅-TO and evaluate its metabolism in mouse MG. (E) High resolution APCI mass spectrum of d₆-Chl (TOF MS AP+).

As an example, Fig. 2 illustrates results of a typical LC/MS experiment with the TP of a mouse gavaged for 14 days. A partial mass spectrum of the Chl/d₆-Chl fraction from the same mouse sample is shown in Fig. 2A. A normal Chl and CE profile of the TP lipid extract, shown as an extracted ion chromatogram of their common analytical ion m/z 369.35 (Fig. 2B), closely replicated profiles published before [3, 11]. These CE are based on normal, non-labeled Chl and were predominantly of a very and extremely long chain nature whose structures were determined before [3, 11]. Thus, Chl for these CE was either biosynthesized in the mouse tissues, or originated from the diet. Note, however, that the 2016 Teklad diet does not list Chl and CE among its components [29]. Also, the vehicle (peanut oil) did not have any appreciable amounts of Chl – its content in the former was about 0.3% of the tracer, or 0.015 mg/mL. Thus, one could presume that all unlabeled Chl and CE detected in the mouse TP sample were biosynthesized in situ.

Figure 2.

Concurrent metabolic transformations of unlabeled and labeled cholesterol in Meibomian glands of mice result in nearly identical profiles of unlabeled and labeled cholesteryl esters. Samples of Meibomian lipids were analyzed using reverse phase LC/APCI-MS. (A) Partial mass spectra of unlabeled Chl and labeled d₆-Chl simultaneously present in the tarsal plate extract. Note that the region of the d₆-Chl mass spectrum between m/z 375 and 380 is magnified ×24. (B) Extracted ion chromatogram of unlabeled Meibomian Chl and CE (m/z 369.35). Peak assignments are shown at the end of the Legend. (C) Extracted ion chromatogram of labeled Meibomian Chl and CE (m/z 375.39) (D) Superimposed extracted ion chromatograms of unlabeled (green trace) and labeled (red trace) cholesteryl esters. Peak assignments: Chl (1); C_12:0-CE (2); C_13:0-CE (3); C_14:0 and C_16:1-CE (4); C_16:0 and C_18:1-CE (5); C_17:0-CE (6); C_18:0 and C_20:1-CE (7); C_19:0-CE (8); C_20:0 and C_22:1-CE (9); C_21:0-CE (10); C_22:0 and C_24:1 (11); C_23:0-CE (12); C_24:0 and C_26:1-CE (13); C_25:0-CE (14); C_26:0 and C_28:1 (15); C_27:0 (16); C_28:0 and C_30:1-CE (17); C_29:0-CE (18); C_32:1-CE (19), C_34:1-CE (20). Peaks (21) and (22) were those of Chl-OAHFA [28].

An extracted ion chromatogram of ion m/z 375.39 revealed a range of CE that were based on the deuterated tracer d₆-Chl (Fig. 2C). The retention times of these newly formed d₆-CE and their mass spectra were used for analytes identification as described before for regular Meibomian CE [11]. The profile of d₆-CE observed in treated mice closely replicated the molecular profile of unlabeled CE, except for a slightly higher presence of shorter chain d₆-CE with FA in the C₁₄-C₂₀ range which eluted between 15 and 25 min of the run (Fig. 2D). Importantly, the d₆-Chl/d₆-CE ratio of 0.49 was much higher than the Chl/CE ratio of 0.10 calculated for unlabeled compounds. This disproportionate accumulation of free, non-esterified d₆-Chl clearly indicated that its esterification into d₆-CE proceeded slower than its absorption from the blood.

To validate the analyte identification in biological samples, these compounds were fragmented in MS/MS experiments as shown in Figure 3. Characteristic pairs of fragments m/z 215.1781/221.2181, 161.1333/167.1659 and 147.1160/152.1443 that were produced by Chl and d₆-Chl respectively, were used to confirm the elution profiles of unlabeled and labeled Meibomian Chl and CE (Figure 3, Panels C-F). This approach let to an unequivocal identification of these analytes in the mouse ML samples.

Figure 3.

LC/MS/MS validation of the d₆-Chl incorporation into Meibomian lipids. Analytical ions obtained in MS/MS fragmentation experiments and used in panels D and F are labeled with asterisk in Panels A and B. (A) MS/MS fragmentation of the analytical ion m/z 369.35 (unlabeled Chl and CE); (B) MS/MS fragmentation of the analytical ion m/z 375.39 (labeled Chl and CE); (C) Extracted ion chromatogram of ion m/z 369.35 (MS experiment); (D) Extracted ion chromatogram of ion m/z 147.12 (fragment of unlabeled Chl and CE; MS/MS experiment; trap collision energy ramping from 20 to 75 eV); (E) Extracted ion chromatogram of ion m/z 375.39 (MS experiment); (F) Extracted ion chromatogram of ion m/z 152.12 (fragment of d₆-Chl and d₆-CE; MS/MS experiment; trap collision energy ramping from 20 to 75 eV). Note how closely chromatograms of ions 147/369 and 152/375 replicated each other in both experiments. A Day 28 sample is shown.

Next, the enrichment of TP lipids with d₆-Chl and d₆-CE was measured (Table 1). The total presence of Chl and CE in mouse TP extracts were calculated as described earlier [3, 11]. Extracted ion chromatograms of ions m/z 369.35 and 375.39 were integrated using the MassLynx built-in routines and the results were compared using Student’s t-test for treated and untreated mice. No statistically significant differences between the total levels of unlabeled Chl and CE in untreated and treated mice were found. The total (d₆-Chl + d₆-CE)/(Chl + CE) ratio for this sample was 0.014, i.e. the overall Meibomian Chl and CE enrichment level with d₆-Chl was 1.4%. For d₆-CE/CE ratio only, the enrichment level approached 0.012 (or 1.2%), while the d₆-Chl/Chl ratio was a substantially higher at 0.05 (or 5%). Considering that free Chl in mouse TP extracts was present in less than 10% of CE, and was less than 5% in expressed meibum, the actual levels of d₆-Chl in those samples were less than 0.3%. The natural variability in the Meibomian lipid makeup was previously shown to be between approximately 5 and 25% for various lipids, and was the highest for free Chl [11]. Thus, the enrichment of meibum with d₆-Chl and d₆-CE was well within those boundaries and was unlikely to affect the lipid homeostasis of the ocular surface. Notably, no noticeable differences between ocular phenotypes of treated and untreated mice were detected, which was in line with low incorporation of labeled lipids in meibum.

Table 1.

Incorporation of d₆-Chl in mouse meibum shown as fractions of the corresponding lipid groups.

Mouse group	(Chl + CE) / TIN*	d6(Chl+CE) / TIN	Chl / CE	Chl / TIN	CE / TIN	d6-Chl / d6-CE	d6-Chl / TIN	d6-CE / TIN
Untreated	0.114 ± 0.018	n/a	0.074 ± 0.007	0.008 ± 0.001	0.106 ± 0.018	n/a
Treated, 28 days	0.128 ± 0.004	0.0018 ± 0.0001	0.084 ± 0.015	0.009 ± 0.001	0.120 ± 0.005	0.295 ± 0.052	0.00046 ± 0.00001	0.00138 ± 0.00011
p-value (Untreated vs. Treated)	0.18	n/a	0.272	0.207	0.185	n/a

^*TIN – total ion number (proportional to the total lipid content), obtained by integrating total ion chromatograms. Corresponding ion numbers for Chl, CE, d₆-Chl, and d₆-CE were obtained by integrating extracted ion chromatograms. Four mice in each group were evaluated. n/a – not applicable. Samples collected on Day 28 of gavaging are shown.

These findings also suggested that the rate of d₆-Chl absorption from blood was higher than the rate of its esterification in MG leading to accumulation of nonesterified d_6-Chl in MG. As these changes were time-dependent and progressed from Week 1 to Week 4, there is little doubt that the achievable degrees of meibum enrichment with d₆-Chl and d₆-CE could be even higher than the ones reported above, but this was not tested in the current experiments because of the high cost of labeled lipid tracers.

Thus, it became evident that d₆-Chl was successfully absorbed by the MG of the mouse and was esterified into typical Meibomian-type CE at detectable levels without being catabolized first. However, the rate of d₆-Chl esterification into d₆-CE was slower that the rate of its absorption from the blood leading to an increase in the Chl/CE ratio in the MG secretions.

Then, incorporation of ¹³C₅-TO was tested. The tracer had the same retention time as the natural unlabeled TO (Fig. 4A). When ¹³C₅-TO was dissolved in the vehicle at 5mg/mL concentration, the signal of its (M + H)⁺ adduct was about 12% of the signal of the natural TO present in the vehicle, which was sufficient for its reliable detection. However, when the TP extract was tested, neither the extracted ion chromatograms nor the mass spectra indicated any presence of ¹³C₅-TO above the noise level (Figs. 4B-4D). Though an extremely weak elution peak with a retention time of about 19.7 min was detected (green trace, Fig. 4B), it, apparently, belonged to another compound with m/z 897.7883 (Compound X). Indeed, its retention time was considerably slower than that of ¹³C₅-TO (19.15 min), but the same as the retention times of a triplet of ions m/z 898.7953, 899.8029 and 900.7980. No signals of (M + H)⁺ adducts (Fig. 4C and 3D), or (M − FA + H)⁺ and (M − 2FA + H)⁺ derivatives of ¹³C₅-TO (not shown) were observed. These experiments established that the Compound X could not be ¹³C₅-TO, but rather a proton adduct of a tetraunsaturated TAG with a molecular formula C₅₈H₁₀₄O₆ (theoretical monoisotopic m/z 897.7911) with four isotopic peaks (M + H + 1)⁺ through (M + H + 4)⁺. Thus, a conclusion was made that neither intact ¹³C₅-TO nor its hydrolyzed fragments could accumulate in MG of the mouse above the noise level.

Figure 4.

Dietary 1,2,3,7,8-¹³C₅-glyceryl trioleate and the Meibomian lipids. Samples of Meibomian lipids and ¹³C₅-TO were analyzed using reverse phase LC/MS. (A) Extracted ion chromatograms of unlabeled (m/z 885.78; red trace) and labeled (m/z 900.84; green trace) triolein in the vehicle (peanut oil). The major peak (labeled with *) is triolein. A signal of a triacylglycerol with the same isobaric m/z value of 885.78 and retention time of ~17 min is labeled with **. (B) Extracted ion chromatograms of unlabeled endogenous triolein of the mouse Meibomian lipids (m/z 885.78; red trace; marked with *), and Compound X (m/z 900.79; green trace). (C) High resolution mass spectra of natural unlabeled (spectrum on the left; m/z 885.7871) and dietary labeled (spectrum on the right; m/z 900.8372) triolein in the tracer lipid mixture. (D) High resolution mass spectra of unlabeled (left; m/z 885.7871) triolein and Compound X (right; m/z 897.7883; 898.7953; 899.8029 and 900.7980) in the mouse Meibomian lipids. Compound X was tentatively identified as a tetraunsaturated triacylglycerol with a molecular formula of C₅₈H₁₀₄O₆. Note that neither ¹³C₅-TO nor its metabolized/hydrolyzed products – diolein and monoolein – were detected above the noise level in any of the TP samples.

The experiments described above were repeated for all tested mice in all tested conditions. Daily oral gavage of mice with labeled lipid tracers ensured a regular influx of the tracers for up to 28 days, which allowed us to estimate the dynamics of their accumulation in ML.

First, incorporation of d₆-Chl was evaluated. The efficacy and reproducibility of d₆-Chl incorporation in Meibomian CE were sufficient to reach preliminary conclusions about its metabolic transformations in mouse MG. First, the CE and d₆-CE profiles of mice were compared. The profiles were found to be very similar and reproducible regardless of the treatment (Figs. 2B-2D and 5A). The only repeatable difference between the pools of CE and d₆-CE was a marginally elevated ratio of shorter chain d₆-CE similar in FA length to common dietary C₁₂-C₂₀ FA, at the expense of C₂₅-C₃₂ d₆-CE (compare Figs. 5A and 2D).

Figure 5.

Incorporation of labeled dietary d₆-Chl in Meibomian lipids. (A) Representative normalized metabolic profiles of unlabeled and labeled Meibomian cholesteryl esters of four mice M1 – M4 on Day 14 of the experiment; data obtained by integrating peaks shown in Fig. 2D. (B) Dynamics of the enrichment of the Chl fraction with d₆-Chl. (C) Dynamics of the enrichment of the CE fraction with d₆-CE.

Next, we determined the dynamics of d₆-Chl and ¹³C₅-TO absorption and incorporation in ML. A time course of meibum enrichment with d₆-Chl and d₆-CE (calculated as the Chl/d₆-Chl and CE/d₆-CE ratios and expressed in %) is shown in Figs. 5B and 5C. Though d₆-Chl and d₆-CE were already easily detectable on Day 7, the degree of meibum enrichment with the tracer continued to rise in a time-dependent manner for both types of lipids and reached ~5.2% for d₆-Chl and ~1.1% for d₆-CE on Day 28 (Fig. 5C). On the other hand, no enrichment of meibum with ¹³C₅-TO or its hydrolyzed metabolites above the noise level was detected in any of the ML samples.

The demonstrated enrichment of meibum with d₆-Chl and d₆-CE implied the existence of a mechanism of Chl intake into MG from the blood. To explore this possibility, we performed mRNA microarray analyses of excised TP as described earlier [11]. A list of genes that are involved in Chl signaling and facilitate its absorption (i.e. influx) by cells and tissues, was obtained from Mouse Genome Informatix Database (MGI, http://www.informatics.jax.org/) as Gene Ontology Annotations GO:0030299, GO:0070508, GO:0045797 and GO:0032376, among others. The following prominent genes that regulate Chl influx were detected in mouse TP and are shown in an alphabetical order with corresponding log(2) expression values (on a 0-to-20 scale) with SD values of <0.5 for all genes: Abca1 (13.8), Abcg1 (9.5), Acat2 (12.8), Anxa1 (15.3), Anxa2 (15.3), Apo1bp (10.1), Apoe (15.4), Cav1 (9.7), Cd36 (15.4), Ldlr (10.7), Ldlrap1 (9.1), Lima1 (10.8), Lpcat3 (14.7), Lrp1 (10.7), Lrp6 (11.8), Npc1 (10.8), Nus1 (12.1), Scarb1 (10.6), Scp2 (11.3), Stard4 (15.6), Stard5 (9.5), and Vps4b (11.9).

Importantly, the plasma CE profiles were entirely different from the Meibomian CE profiles (Figure 6): The former were based almost exclusively on medium chain length C₁₂-C₁₈ FA, while FA residues of the latter varied from C₁₂ to C₃₄, with the most prominent FA being in the C₂₄-C₂₈ range. Dietary d₆-Chl was detected in the mouse plasma samples as free, non-esterified compound, and d₆-CE of the same types as non-labeled CE (Figures 6C and 6D). Note that the molecular profiles of non-labeled and labeled compounds closely resembled each other, which revealed that d₆-Chl was converted into d₆-CE with the same efficacy as unlabeled Chl was converted into CE (Table 2), though the overall abundance of deuterated lipids in plasma and meibum was expectedly lower than that of unlabeled Chl and CE (as discussed above and shown in Figures 2B, 2C, 5B, 5C, and Table 1). The overall degree of incorporation of d₆-Chl in plasma lipids in the form of free d₆-Chl and d₆-CE was calculated to be (9.8 ± 0.6) % of the total (Chl + CE) fraction of mouse plasma (Figure 7 and Table 2).

Figure 6.

Incorporation of labeled dietary d₆-Chl in mouse plasma and meibum. Typical samples obtained from an individual mouse on Day 14 of gavaging are shown as normalized extracted ion chromatograms. (A) LC/MS analysis of an equimolar mixture of authentic Chl and four CE with C_16:1, C_18:1′ C_22:1, and C_24:1 fatty acid residues (50μM each). An extracted ion chromatogram of ion m/z 369.35 is shown. The fatty acid residues, retention times (in minutes), and peak areas (in arbitrary units) are shown next to each peak. (B) An extracted ion chromatogram of d₆-Chl. Analytical ion m/z 375.39 was used. (C) Chl and CE profile of mouse plasma. An extracted ion chromatogram of the analytical ion m/z 369.35 is shown. Plasma CE were based primarily on C₁₂-C₁₈ FA with 0 to 3 double bonds. (D) d₆-Chl and d₆-CE profile of mouse plasma shown as an extracted ion chromatogram of ion m/z 375.39. (E) Chl and CE profile of mouse meibum detected as ion m/z 369.35. (F) d₆-Chl and d₆-CE profile of mouse meibum shown as an extracted ion chromatogram of ion m/z 375.39. The peak labels in Panels E and F are the same As in Figure 2.

Table 2.

Incorporation of d₆-Chl in mouse plasma on Day 28 shown as fractions of the corresponding lipid groups.

Mouse group	(Chl + CE) / TIN*	d6(Chl+CE) / TIN	Chl / CE	Chl / TIN	CE / TIN	d6-Chl / d6-CE	d6-Chl / TIN	d6-CE / TIN
Untreated	0.312 ± 0.076	n/a	0.251 ± 0.030	0.062 ± 0.004	0.250 ± 0.023	n/a
Treated, 28 days	0.235 ± 0.076	0.023 ± 0.009	0.268 ± 0.015	0.049 ± 0.015	0.185 ± 0.061	0.269 ± 0.010	0.021 ± 0.002	0.019 ± 0.007
p-value (Untreated vs. Treated)	0.242	n/a	0.416	0.149	0.102	n/a

^*TIN – total ion number (proportional to the total lipid content), obtained by integrating total ion chromatograms. Corresponding ion numbers for Chl, CE, d₆-Chl, and d₆-CE were obtained by integrating extracted ion chromatograms. Four mice in each group were evaluated. n/a – not applicable.

Figure 7.

Mass spectra of dietary d₆-Chl and unlabeled Chl in mouse plasma. A representative sample obtained from an individual mouse on Day 28 of gavaging is shown. An apparent molecular Chl to d₆-Chl ratio is proportional to the intensity of their signals.

4. DISCUSSION

This study was to answer two specific questions: 1) “Can dietary lipids, such as Chl and TAG, be incorporated in meibum without being extensively catabolized first?”, and 2) “If such incorporation does occur, how does this process affect the lipid balance in MG?”.

As LC/MS is currently the most sensitive and specific method for identification of minor analytes in complex mixtures, this approach was chosen to study metabolic transformations of ²H- and ¹³C-labeled lipids in mouse MG. Other approaches, such as nuclear magnetic resonance, infrared and Raman spectroscopies, or using radioactively labeled precursors (e.g., ³H- or ¹⁴C-labeled lipids) were deemed less suitable because of either inferior sensitivity/selectivity of those approaches, or because of poor compatibility/integration with LC instrumentation. In addition, radioactive precursors would represent higher health and environmental risks, which using non-radioactive labels completely eliminated.

Next, the roles of Chl in MG lipid homeostasis need to be discussed. Lipids of meibum are complex mono, di-, and tri-esters that are vastly different from those found in other tissues, organs, and secretions. This fact implies that most of those lipids are formed in situ and/or de novo via a process that was termed meibogenesis [3, 4]. However, the origins of the major precursors of ML – cholesterol and fatty acids – remain unclear as they, in different forms, freely circulate with blood, and can possibly be incorporated in ML as is. At the same time, Chl and FA can be biosynthesized in situ due to high expression levels of related key genes and enzymes of steroidogenesis and FA metabolism in MG [3, 4, 12–15]. Free Chl is an important signaling factor in inflammation and immune response [30], so increase in its level may promote inflammatory responses in MG. Indeed, such a multifold increase was observed in chalazia [20] and in meibum of a patient with HTLW syndrome [23]. Chl was also increased in meibum of Elovl3 knockout mice [14, 31], which had severely inflamed and swollen eyelids, and so were eyelids of Soat1-null mice [32]. Thus, excessive accumulation of Chl in MG may be a trigger, or at least a part, of inflammatory processes of the eyelid tissue.

Apart from possible pro-inflammatory functions of Chl, the lipid induced undesirable physical changes in meibum that altered its behavior in in vitro models of the human tear film (TF) [30]. The overall effect of the increase in Chl levels was a clear destabilization of the lipid layers formed of meibum. Specifically, it was shown in direct in vitro Langmuir trough experiments that an increase in Chl led to a multifold increase in the rigidity of ML films. Also, Chl made the films easily collapsible, having abnormal surface morphology, and producing compression-decompression isotherms with high hysteresis [33]. These findings, extrapolated onto the in vivo conditions, depict high levels of Chl in meibum as a clearly detrimental factor that may play a major role in various pathophysiological processes in the eye and adnexa.

Our experiments demonstrated that dietary d₆-Chl could, while dietary ¹³C₅-TO, apparently, could not, be found among ML as intact molecules or partially hydrolyzed products (in case of labeled TO). Dietary d₆-Chl was observed in mouse plasma as free unesterified sterol and its esters with C₁₂-C₁₈ FA. Thus, the four rings of dietary d₆-Chl (A, B, C, and D) remained intact and the compound did not lose any of its six deuterium labels in rings A and B. However, the compound underwent partial esterification before reaching MG. Moreover, dietary d₆-Chl (and, possibly, d₆-CE) were absorbed by MG and (trans)esterified into very, extremely, and ultra-long chain C₁₆- to C₃₄-FA alongside unlabeled Chl, producing d₆-CE profiles that were virtually indistinguishable from those of unlabeled CE (Fig. 2D), but drastically different from plasma CE (Figure 6). The only noticeable difference between the pools of labeled and unlabeled Meibomian CE was a much higher Chl/CE ratio for deuterated compounds (Figs. 2B and 2C). This increase may reflect a higher rate of d₆-Chl/d₆-CE absorption by MG compared to their esterification/transesterification with various endogenously generated Acyl-CoA in the meibocytes of MG, making the latter a rate-limiting step in the biosynthesis of Meibomian CE. The other, less obvious, difference was a minor increase in shorter chain d₆-CE with FA in the C₁₄-C₂₀ range at the expense of their longer chain variants (Fig. 2D and Fig. 5A). This effect was observed for all tested samples and may be related to a mechanism of CE intake from blood by MG. Interestingly, the kinetics of incorporation of the tracer was rather slow as the degree of enrichment with deuterated lipids rose gradually during four weeks of gavaging (Fig. 5B and 5C). This observation may be an indication of either a rather slow lipid metabolism in MG, or a slow intake of the tracer by MG. These two possibilities, among others, seem plausible and should be evaluated in future studies.

Incorporation of dietary Chl, but not TAG (Fig. 4), in ML implies that the diet may have a direct, but selective, impact on meibogenesis as there is a direct transporting pathway for dietary Chl and/or CE, but not for TAG. This finding may explain, in part, beneficiary therapeutic effects of omega-3 fatty acids used as human food supplements in our earlier experiments [34] and comparable therapeutic effects that were observed in later studies by Bhargava et al. [35], Olenik et al. [36] and in a range of other projects: The treatments resulted in improvement in clinical signs and symptoms of Dry Eye due to the increased tear production, but, in our hands, no changes in the composition of major ML (such as WE and CE) were detected [34]. Yet, in a similar study [37], Mascai et al. reported a decline in saturated FA in meibum in response to omega-3 FA supplementation. It was not possible to establish from that study whether the treatment led to a local change in lipid homeostasis in MG, or caused more systemic effects on the lipid homeostasis in other tissues and organs. Therefore, our direct experiments with ¹³C₅-TO attempted to separate those possibilities and established that no incorporation of ¹³C₅-TO in the mouse meibum occurred even after a 4-week of daily treatment. It seems that dietary TAG undergo extensive metabolism (possibly, via peroxisomal β-oxidation of their FA in the intestinal tissues, liver or elsewhere) before reaching MG, and thus can have only systemic and/or mediated effect on MG functions and the ocular surface health.

Transcriptomic experiments with mouse TP revealed that MG of mice are equipped with necessary machinery for Chl intake from the blood: The main genes that facilitate Chl intake by cells and tissues – Abca1, Abcg1, Acat2, and others listed in the Results section – were found to be highly expressed in mouse TP in a 9-to-15 range on a log(2) scale. Moreover, a side-by-side comparison of plasma Chl and CE with their Meibomian counterparts revealed that the d₆-Chl and plasma-type d₆-CE could be effectively esterified/transesterified into very- and extremely long chain MG-specific d₆-CE that are not present in mouse plasma in detectable quantities. The detailed investigation and discussion of such a mechanism of Chl/CE absorption and metabolism go beyond the scope of this paper and will be explored in the future.

Finally, direct incorporation of dietary Chl into MG lipids that was demonstrated in our study implies that pathological systemic diseases such as hyper- or hypocholesterolemia may have a direct impact on the quality of meibum not only by influencing the in situ biosynthesis of Chl and CE in MG, but also via alterations of their intake from the blood. The magnitude of these potential effects of Chl and other dietary lipids on meibogenesis is currently unknown, which justify their comprehensive evaluation in future studies.

HIGHLIGHTS.

• Dietary cholesterol accumulates in Meibomian glands in a time-dependent manner.

• Dietary cholesterol is incorporated into typical Meibomian cholesteryl esters (CE).

• Dietary cholesterol can increase the cholesterol to CE ratio in meibum.

• Dietary triacylglycerols do not accumulate in Meibomian glands in intact form.