Meibum Sphingolipid Composition is altered in individuals with Meibomian Gland Dysfunction- a side by side comparison of Meibum and Tear Sphingolipids

Abstract

Purpose:

Sphingolipids (SPL) play a role in cell signaling, inflammation, and apoptosis. The purpose of this study was to examine meibum and tear SPL composition in individuals with poor versus good meibum quality.

Methods:

Individuals were grouped by meibum quality (n=25 with poor quality, case group and n=25 with good quality, control group). Meibum and tears were analyzed with liquid chromatography-mass spectrometry (LC-MS) to quantify SPL classes. Semiquantitative and relative composition (mole percent) of SPL and major classes, Ceramide (Cer), Hexosyl-Ceramide (Hex-Cer), Sphingomyelin (SM), Sphingosine (Sph), and sphingosine 1-phosphate (S1P) were compared between groups.

Results:

Demographic characteristics were similar between the two groups. Overall, individuals with poor meibum quality had more SPL pmole in meibum and tears than controls. Relative composition analysis revealed that individuals with poor meibum quality had SPL composed of less Cer, Hex-Cer, and Sph and more SM compared to individuals with good quality meibum. This pattern was not reproduced in tears as individuals with poor meibum quality had SPL composed of a similar amount of Cer, but more Hex-Cer, Sph and SM compared to controls. In meibum, SPL pmole and relative composition most strongly correlated with MG metrics while in tears, SPL pmole and relative composition most strongly correlated with tear production. SPL in both compartments, specifically Cer pmole in meibum and S1P% in tears, correlated with DE symptoms.

Conclusion:

SPL composition differs in meibum and tears in patients with poor vs good meibum quality. These findings may be translated into therapeutic targets for disease.

1. Introduction

Dry eye (DE) is a multifactorial process manifesting with a variety of symptoms and signs caused by tear film modification in any of the three major tear constituents (mucin, aqueous, and lipid) with the potential for ocular surface damage and inflammation [1, 2]. The prevalence of DE varies based on disease definition, geographic location, and among races, but the International Dry Eye Workshop estimated the overall prevalence to be between 5-30% in patients 50 years of age or older [3]. DE is classified into two main subgroups; aqueous deficient dry eye (ADDE) and evaporative dry eye (EDE), with EDE being the more prevalent sub-type in both clinic and population based studies [3]. ADDE manifests with decreased tear production and is closely related to inflammation and systemic immune abnormalities, whereas EDE manifests with tear film instability, often due to an unhealthy tear film lipid layer (TFLL) [4, 5].

Abnormalities in Meibomian glands (MG) and their secreted product meibum often underlie TFLL abnormalities [4, 6]. Several pathophysiologic mechanisms are thought to contribute to MG dysfunction (MGD) including obstruction (in the setting of keratinization, fibrosis, or neutrophilic trapping) [7, 8], inflammation [8], altered expression of peroxisome proliferator-activated receptor gamma (PPARγ, a master regulator of meibocyte differentiation, function, and lipid synthesis), and meibocyte stem cell depletion [9, 10], to name a few. Regardless of the underlying mechanisms, clinically, patients with MGD are noted to have alterations in the quality and quantity of meibum [11, 12].

Meibum is composed of non-polar lipids (wax and sterol esters, diglycerides, triglycerides, hydrocarbons, and trace amounts of monoglycerides and fatty alcohols), polar lipids (cerebrosides, phospholipids including phosphatidylcholine and sphingomyelin, and trace amounts of ceramides) and bacterial degradation products (free-fatty acids) [6]. Non-polar lipids comprise the largest portion of meibum (~77%) and act as a lubricant and barrier to water. Polar lipids constitute a lower percent (~8%) of meibum and act as surfactants to help spread the TFLL across the ocular surface and also serve as structural supports for the non-polar lipids [6]. Specifically, polar meibum lipids have been found to sit between the aqueous-mucin phase and non-polar meibum lipids, providing crucial structural support to the non-polar lipids as well as generating a direct evaporation resistant effect [13-15].

Compositional changes in meibum, notably, an increase in the relative proportion of polar lipids, have been noted with aging and in the setting of ocular surface diseases such as MGD [16, 17]. This is accompanied by a decreased relative proportion of non-polar surface lipids such as wax ester to cholesteryl ester ratios, and these changes have been associated with TFLL instability (though changes in wax esters and cholesteryl esters have not been observed unanimously) [18-20]. Differential expression of specific polar (O-acyl)-omega-hydroxy fatty acids has been observed in the tear and meibum profiles of individuals with MGD, reinforcing the role of polar lipid species [21]. Sphingolipids (SPL) are a minor component of polar lipids in meibum, found to comprise ~30% of polar lipids [22]. SPL are found in every cell, particularly in the cell membrane and intra-cellular organelles. SPL have received significant attention in last two decades because they not only provide structural support but also play a crucial role in cell signaling [23]. Specifically, ceramide (Cer) and sphingosine (Sph) are pro-apoptotic; while ceramide 1-phosphate (C1P) and sphingosine 1-phosphate (S1P) are anti-apoptotic and play important role in inflammatory signaling. Alterations in these ‘bioactive SPLs’ and their associated pathways have been found in a myriad of human diseases including cancer, metabolic, and neurological diseases [24-26]. In the eye, alterations in SPL bioactivity have been associated with optic neuritis, uveitis, age-related macular degeneration, retinitis pigmentosa, cataracts, and diabetic retinopathy [18, 27-30].

With respect to ocular surface diseases, in vitro studies have demonstrated that increasing Cer concentration in meibum led to an increase in meibum melting temperature, rigidity, and TFLL destabilization [16, 31]. Furthermore, SPL may have a functional role in MGD, as bioactive SPL are involved in cell signaling, inflammation, and apoptosis[32], factors implicated in the pathophysiology of MGD [33]. In our prior study, we found that individuals with poor meibum quality had SPL composed of less Cer, Hex-Cer, and S1P, and more SM and Sph compared to individuals with good quality meibum [34]. Overall, individuals with poor meibum quality had an increased ratio of Cer (pro-apoptotic) to S1P (pro-survival) in their meibum. Missing from this study was a concomitant examination of SPL in tears and an analysis of how SPL levels effect TFLL function. As such, in the current study, we aimed to validate our findings in a new population as well as examine the relationship between meibum and tear SPL and their impact on ocular surface findings.

2. Methods

Study Population:

Individuals with and without MGD were prospectively recruited from the Miami Veterans Administration Medical Center (VAMC) between October 2013 and August 2016. Participants were excluded if they had anatomic abnormalities of the eyelids or cornea (e.g. ectropion, pterygium), wore contact lenses, had a history of refractive, glaucoma, or retina surgery, used ocular medications other than artificial tears (held for two hours prior to participation in the study), had active corneal infection or inflammation, or had cataract surgery within the last 6 months. Patients with human immunodeficiency virus, sarcoidosis, graft-versus-host disease, or a collagen vascular disease were also excluded. The Miami VAMC approved the prospective study, informed consent was obtained from all patients, and the study was adherent with the principles of the Declaration of Helsinki.

Data Collection:

Demographics, past medical history and medication information were collected by self-report and verified by medical records.

Classification of MGD:

The quality of forcefully expressed meibum was graded on a scale of 0 to 4 (0, clear; 1, cloudy; 2, granular; 3, toothpaste; 4, no meibum extracted) [35]. Individuals were first divided into two groups based on this grade. Good meibum quality was defined as grade 0 or 1 and this population served as the control group. Poor meibum quality was defined as grade, 2, 3, or 4 and this population served as the case group.

Ocular surface evaluation:

All individuals underwent a standardized tear film assessment of both eyes which included, in the order performed (1) tear osmolarity (TearLAB, San Diego, CA), (2) tear breakup time (TBUT), evaluated after 5μl of fluorescein was placed on the superior conjunctivae, (3) corneal staining with fluorescein using the National Eye Institute scale [2], (4) Schirmer’s strips with anesthesia, and (5) meibomian gland (MG) assessment including characterization of meibum quality, inferior meibomian gland plugging and dropout. MG plugging was graded on a scale from 0 to 3, defined as: 0, none; 1, less than 1/3 lid involvement; 2, between 1/3 and 2/3 lid involvement; 3, greater than 2/3 lid involvement [36, 37]. MG dropout was assessed by retroillumination and graded to the Meiboscale (five-point scale: 0, area of loss 0%; 1, area of loss <25%; 2, area of loss 25%-50%; 3, area of loss 51-75%; 4, area of loss >75%) [38]. The quality of forcefully expressed meibum was graded as noted above [35].

Symptoms:

Participants in the study completed two standardized DE questionnaires, the Ocular Surface Disease Index (OSDI, range 0-100)[39] and 5-Item Dry Eye Questionnaire (DEQ-5, range 0-22) [40].

Sample collection:

We opted to use cotton buds for sample extraction and collection based on previous research showing increased retrieval of ocular surface lipids using this methodology compared to meibomian gland forceps and meibomian gland evaluators [41]. A drop of proparacaine was placed on the ocular surface and on two cotton tip applicators. Ten minutes after administration of proparacaine on the ocular surface, one applicator was placed behind and one in front of the interior tarsal plate and pressure was applied to forcefully express meibum from the inferior orifices while moving the tips back and forth. On average, we tried to express meibum from at least five glands from each eye and collect all expressed meibum. The expressed meibum was collected by swiping the cotton tip applicator across the inferior lid margin. Based on this collection method, our ‘meibum samples’ may consist not only of expressed meibum, but also of meibum plugs and other components including epithelial cells, cellular debris, tears, and/or microbes. Meibum collection was first performed in the right eye and then repeated in the left eye with the same applicator. The cotton tip applicator was then broken, the bud placed in an Eppendorf tube, and the tube immediately placed in −80°C. Schirmer strips (collected prior to meibum sampling) were also placed in Eppendorf tubes and frozen at −80°C.

Extraction and Analysis of Sphingolipids:

Lipids (meibum) from the entire cotton bud, tears from entire Schirmer’s strip were extracted using a modified Bligh & Dyer method as previously described [34]. Total recovery of lipids was assessed using total volumes of meibum on cotton tips or tears in Schirmer’s strips, thus differences in total amounts were dependent on initially recovered volumes. For analysis, frozen samples were placed in borosilicate glass tubes and spiked with 250 pmole of C1P, sphingomyelin, ceramide, and monohexosyl ceramide (d18:1/12:0 species), and sphingosine (So), sphinganine (Sa), sphingosine-1-phosphate (S1P), sphinganine-1-phosphate (Sa1P) (d17:0 sphinganine/d17:1 sphingosine) as internal standard (Avanti Polar Lipids). Following addition of internal standard, MeOH:CHCl3 (2:1) was added to the tubes and the mixture was sonicated for approximately 2 minutes. Samples were then incubated for 6h at 48°C. Extracts were then centrifuged at 5000 rpm for 20 min, transferred to a new glass tube, dried down and reconstituted in methanol (500 μl) by sonication. Extracts were again centrifuged at 5000 rpm for 20 min and transferred to injection vials for mass spectrometry analysis.

Liquid chromatography/mass spectrometry:

Analysis of sphingolipids was carried out by using UPLC ESI-MS/MS. Sphingolipids were separated using a Shimadzu Nexera X2 LC-30AD coupled to a SIL-30AC auto injector, coupled to a DGU-20A5R degassing unit in the following way. An 8 min, reversed phase LC method utilizing an Acentis Express C18 column (5cm x 2.1mm, 2.7μm) was used to separate the sphingolipids at a 0.5 ml/min flow rate at 60°C. The column was equilibrated with 100% Solvent A [methanol:water:formic acid (58:44:1, v/v/v) with 5mM ammonium formate] for 5 min and then 10 μl of sample was injected. 100% Solvent A was used for the first 0.5 min of elution. Solvent B [methanol:formic acid (99:1, v/v) with 5mM ammonium formate] was increased in a linear gradient to 100% Solvent B from 0.5 min to 3.5 min. Solvent B was held constant at 100% from 3.5 min to 6 min. From 6 min to 6.1 min solvent B was reduced to 0%, and solvent A returned to 100%. Solvent A was held constant at 100% from 6.1 min to 8 min.

Sphingolipids were analyzed via mass spectrometry using an AB Sciex Triple Quad 5500 Mass Spectrometer via ‘targeted’ assay. Q1 and Q3 were set to detect distinctive precursor and product ion pairs specific for sphingolipids only as published previously [42]. Ions were fragmented in Q2 using N2 gas for collisionally induced dissociation. Analysis used multiple-reaction monitoring in positive-ion mode. Sphingolipids were monitored using precursor → product MRM pairs. The mass spectrometer parameters used were: Curtain Gas: 30 psi; CAD: Medium; Ion Spray Voltage: 5500 V; Temperature: 500°C; Gas 1: 60 psi; Gas 2: 40 psi; Declustering Potential, Collision Energy, and Cell Exit Potential vary per transition. The species of SPL were identified based on their retention time and m/z ratio and quantity of each (in pmole) was determined semi-quantitatively using the peak areas of the internal standards that were spiked in each sample, as described in previous publications [42-45]. SPL values were reported as total pmole and as mole% to control for tear and meibum volume and provide consistent data for comparison. The call for a positive detection was based on the identified peak and signal to noise ratio (S/N) for the area under the curve. The cutoff for the specifies of Cer, SM, and Hex-Cer was >100 S/N. For low expressing Sa, So, and S1P it was >10 S/N. Non detectable values were set as “0” or presented as not detected (ND) throughout the paper. Mass spectrometry was specific for non-dihydro species of Cer, Hex-Cer, and SM. We did not measure dihydro species of these. Our SPL specific methodology also did not measure other meibum lipids such as non-polar wax ester and cholesterol ester species, nor phospholipids besides SM, such as phosphatidylcholine and phosphatidylethanolamine which together make up ~50% of phospholipid species in meibum [22, 46].

Statistical Analysis:

Descriptive statistics were performed to summarize patient baseline demographics and clinical information. Non-parametric tests were used to examine differences pmole SPL and relative mole percent by demographics and clinical metrics. After inspecting residuals, multiple linear regression analyses were carried out to assess the impact of demographics, medication use, and MGD metrics on the outcome variables. All statistical analyses were performed using SPSS 22.0 (SPSS, Inc., Chicago, IL). The sample size was calculated to detect an effect size of 0.8. 25 subjects per group provided 80% power for an effect size of 0.8, using a two-sided t-test with type I error probability of 0.05. In this paper, we opted to give information on all variables being compared as opposed to correcting the p-value (e.g. Bonferroni) since the latter methodology has its own limitations [47].

3. Results

Fifty individuals participated in the study (mean age 57 ± 9 years; 84% men; 38% white, 60% black; 18% Hispanic). Comorbidities and systemic drug use were similar in the poor vs good quality meibum groups with the exception of NSAID use which was significantly more frequent in the good meibum quality group (Table 1). We first quantified total SPL and major SPL classes (Cer, Hex-Cer, SM, S1P, and Sph) in meibum and tears collected from all 50 patients (Table 2). We found that the IQR (25^th percentile - 75^th percentile) for total meibum and tear SPL (895 pmole and 1339 pmole), and for SPL sub-types (SM, Cer, Hex-Cer, Sph, Sa, and S1P) were fairly wide, reflecting variance in amount of sample collected (Supplementary Table 1). Overall, a greater pmole of SPL was recovered from tears than from meibum, but with wider IQR for tear vs meibum samples.

Table 1.

Demographics grouped by meibum quality.

	Good Meibum Quality (N=25)	Poor Meibum Quality (N=25)	P-Value
Age, mean ± SD	55 ± 11 years	59 ± 7 years	0.16
Sex, male % (n)	88% (22)	80% (20)	0.44
Race, White % (n)	48% (12)	28% (7)	0.24
Black % (n)	52% (13)	68% (17)
Ethnicity, Hispanic % (n)	16% (4)	20% (5)	0.71
Co-morbidities
PTSD, % (n)	32% (8)	16% (4)	0.19
Depression, % (n)	72% (18)	80% (20)	0.51
Arthritis, % (n)	52% (13)	46% (11)	0.67
Sleep Apnea, % (n)	20% (5)	20% (5)	1.00
Hypercholesterolemia, % (n)	44% (11)	52% (13)	0.57
Hypertension, % (n)	56% (14)	56% (14)	1.00
Rosacea, % (n)	0% (0)	0% (0)	0.00
Benign Prostastic Hyperplasia, % (n)	16% (4)	16% (4)	1.00
Medication Use
NSAID, % (n)	44% (11)	16% (4)	0.03*
Anti-Depressant, % (n)	52% (13)	72% (18)	0.15
Anti-Anxiety, % (n)	60% (15)	72% (18)	0.37
Antihistamine, % (n)	24% (6)	16% (4)	0.48
Gabapentin, % (n)	36% (9)	20% (5)	0.21
Multivitamin, % (n)	64% (16)	56% (14)	0.56
Beta Blocker, % (n)	20% (5)	8% (2)	0.22
Statin, % (n)	48% (12)	40% (10)	0.57
Aspirin, % (n)	48% (12)	24% (6)	0.08
Sildenafil, % (n)	12% (3)	28% (7)	0.16

SD=standard deviation, n=number in group, PTSD=post-traumatic stress disorder, NSAID=nonsteroidal anti-inflammatory drug

^*= p value < 0.05

Table 2.

Amount (pmole) of major sphingolipid classes in meibum and tears grouped by meibum quality.

Sphingolipid Species	Good Quality (N=25)		Poor Quality (N=25)		P-Value
	Median	IQR	Median	IQR
Meibum
Cer	79.6	99.2	202	146	0.002
Hex-Cer	27.1	25.3	43.2	32.0	0.007
SM	335	310	957	480	<0.0005
Sph	38.6	28.2	39.2	31.6	0.96
Sa	15.3	9.4	19.4	18.5	0.07
SPL	480	437	1189	530	<0.0005
Tears
Cer	100	273	192	212	0.61
Hex-Cer	28.0	55.6	37.0	84.7	0.18
SM	639	908	951	1034	0.29
Sph	52.0	70.9	100*	101	0.07
S1P	66.4	88.5	26.8	82	0.07
SPL	829	1325	1276	1351	0.49
Cer/S1P	2.58	3.39	5.95	14.11	0.045

SPL=total sphingolipid; Cer=ceramide; Hex-Cer=hexosyl ceramide; SM=sphingomyelin; Sph=sphingosine; Sa=sphinganine; S1P=sphingosine 1-phosphate; IQR=interquartile range. Bioactive SPL.

^*Median in 8 individuals with detectable levels (all individuals with good meibum quality had detectable levels).

We first examined SPL pmole in meibum and tears by demographic data. Though our measurement is semi-quantitative in nature, the pmole values can be compared between controls and cases and among various groups. Overall, age was positively associated SPL in meibum [total SPL pmole (rho=0.36, p=0.01), ceramides pmole (rho=0.39, p=0.005), Hex-Cer pmole (rho=0.47, p=0.001), SM pmole (rho=0.33, p=0.02), and Sph pmole (rho=0.28, p=0.045)] but not in tears. Blacks had higher SM pmole (618 vs 390, p=0.03) and total SPL pmole (898 vs 594, p=0.04) in meibum than whites but no differences in tear SPL were noted by race. No differences in meibum or tear SPL were noted by gender or ethnicity (i.e. Hispanic vs non-Hispanic).

We next compared amounts (pmole) of SPL, Cer, Hex-Cer, SM, Sph, Sa, and S1P between the two groups. We found that the distribution of total ceramides, Hex-Cer, SM, and SPL were significantly different between the groups. Overall, individuals with poor meibum quality had higher levels of SPL in meibum and tears (Cer, Hex-Cer, SM) compared to individuals with good meibum quality (Table 2). The difference reached statistical significance in the meibum but not the tear samples, likely due to the wider variability of SPL pmole recovered from tears. When examining the Cer/S1P ratio in tears, individuals with poor meibum quality had a higher ratio (5.95 IQR 14.11 vs 2.58 IQR 3.39) than individuals with good meibum quality, p=0.045.

Next, we obtained mole percentages of each SPL sub-type by dividing the pmole of the specific SPL by the total pmole of SPL extracted from the individual. This resulted in a metric that was not dependent on volume of meibum extracted or the amount of tears collected. These relative values are also consistent and not affected by minor analytical metrics and batch effects. When examined as percentages, age significantly correlated with Sph% (rho=−0.33, p=0.02) in meibum. Blacks had higher tear Hex-Cer% (3.7% vs 2.5%, p=0.02) and Sph% (4.6% vs 3.1%, p=0.001) but no differences in SPL% were noted by gender or ethnicity.

We then compared the mole percent of Cer, Hex-Cer, SM, and Sph between the groups. We found that individuals with poor meibum quality had SPL composed of less Cer (15.1% vs 17.1%), less Hex-Cer (3.8% vs 4.9%), less Sph (4.3% vs 8.2%) and more SM (75.9% vs 65.8%) as compared to individuals with good meibum quality (Table 3). In addition, individuals with poor meibum quality had a lower Cer/SM and Hex-Cer/SM ratios. This pattern was not reproduced in tears as individuals with poor meibum quality had SPL composed of a similar amount of Cer, but more Hex-Cer (3.6% vs 2.5%, p=0.049), Sph (4.5% vs 3.4%, p=0.07) and SM (77% vs 75%, p=0.07) compared to individuals with good quality meibum. We evaluated the impact of potential confounders (demographics and medication use) beyond meibum quality on meibum pmol and relative SPL % using forward stepwise multi-variable linear regression analyses. Even when considering confounders, meibum quality remained in the final model for total SPL (β= 646.85, SE= 131.72 , p<0.001, R²=0.58) , Hex-Cer% (β=−2.43, SE=0.99 , p=0.018, R²=0.34), SM% (β= 17.00, SE= 4.45, p<0.001, R²=0.50), and Sph% (β=−5.70, SE= 1.49, p=, R²=0.49).

Table 3.

Relative composition of major sphingolipid classes in meibum and tears grouped by meibum quality.

Sphingolipid Species	Good Quality (N=25) Mole Percent		Poor Quality (N=25) Mole Percent		P-Value
	Median	IQR	Median	IQR
Meibum
Cer	17.1	8.6	15.1	6.7	0.09
Hex-Cer	4.9	3.8	3.8	1.0	0.02
SM	65.8	13.2	75.9	6.3	<0.005
Sph	8.2	6.2	4.3	2.7	<0.005
Sa	3.3	3.1	1.8	1.7	0.006
Cer/SM*	0.25	0.18	0.20	0.10	0.02
Hex-Cer/SM*	0.07	0.09	0.05	0.02	0.007
Cer:Hex-Cer*	3.45	2.02	3.90	2.97	0.11
Tears
Cer	11.9	7.2	12.6	6.6	0.98
Hex-Cer	2.5	2.6	3.6	2.8	0.049
SM	75.0	8.1	77.4	10.0	0.07
Sph	3.4	2	4.5	5	0.07
S1P	4.6	3.9	2.0	5.6	0.02
Cer/SM*	0.17	0.10	0.16	0.09	0.48
Hex-Cer/SM*	0.03	0.03	0.05	0.04	0.09
Cer:Hex-Cer*	4.76	5.39	3.86	2.64	0.19

SPL=total sphingolipid; Cer=ceramide; Hex-Cer=hexosyl ceramide; SM=sphingomyelin; Sph=sphingosine; Sa=sphinganine; S1P=sphingosine 1-phosphate; IQR=interquartile range. Bioactive SPL.

^*Ratios were first calculated for each individual and then mean ratios were compared between groups

Next, we examined relationships between SPL pmole and various DE symptoms and signs, including signs of MGD. In meibum, the measure that most closely correlated with DE symptoms was Cer pmole (rho=−0.37, p=0.008 for DEQ5; rho=−0.38, p=0.006 for OSDI), linking lower Cer pmole with more severe symptoms (Table 4). Examining DE signs, several significantly correlated with SPL pmole including meibum quality (rho=0.64, p<0.005), Meibomian gland drop out (rho=0.34, p=0.02), and Schirmer (rho=0.31, p=0.03). A similar relationship was noted when examining SPL pmole sub-types (Table 4). This indicates that higher meibum SPL pmole correlated with more abnormal MG metrics but with more normal tear production. In contrast, tear SPL pmole negatively correlated with Schirmer scores (rho=−0.32, p=0.03).

Table 4.

Spearman’s Rho (ρ) coefficients between sphingolipid pmole and dry eye (DE) symptoms and signs.

Spearman’s ρ	DE Symptoms		MGD/DE Signs
	DEQ-5	OSDI	Meibomian Quality	Eyelid Vascularity	Meibomian Dropout	Meibomian Plugging	TBUT	Fluorescein Stain	Schirmer Test	MMP
In meibum (total pmole)
Cer	−0.37**	−0.38**	0.46***	−0.16	0.37***	0.23	0.03	−0.02	0.32*	0.12
Hex-Cer	−0.21	−0.25	0.40**	−0.09	0.28	0.15	0.13	−0.15	0.37***	−0.05
SM	−0.30*	−0.32*	0.66***	0.03	0.37***	0.03	−0.08	0.0	0.15	−0.03
Sph	0.02	−0.09	0.13	−0.06	0.10	0.31*	0.02	0.06	0.22	0.19
Sa	−0.21	−0.23	0.31*	−0.28	0.17	0.23	0.17	−0.06	0.30*	−0.03
SPL	−0.27	−0.21	0.64***	−0.28	0.34*	0.20	0.003	−0.12	0.31*	−0.02
In tears (total pmole)
Cer	−0.12	−0.06	0.04	−0.09	0.15	0.09	−0.19	0.12	−0.42	−0.046
Cer/SM	0.08	0.08	−0.14	−0.01	−0.08	−0.11	−0.06	−0.006	−0.40**	0.04
Hex-Cer	−0.10	0.005	0.17	−0.28	0.25	0.22	−0.18	0.21	−0.03	0.03
Hex-Cer/SM	−0.02	0.07	0.07	−0.20	0.20	0.20	−0.007	0.18	−0.01	0.09
SM	−0.20	−0.11	0.11	−0.14	0.23	0.16	−0.21	0.14	−0.33*	−0.09
Sph	−0.19	0.005	0.32	−0.32	0.20	0.18	−0.02	0.14	0.10	−0.14
S1P	0.18	0.31	−0.17	−0.08	−0.01	0.03	0.03	0.07	−0.003	−0.03
SPL	−0.16	−0.05	0.07	−0.14	0.19	0.14	−0.19	0.14	−0.32*	−0.07

DE=Dry Eye, SPL=total sphingolipid, Cer=ceramide, Hex-Cer=hexosyl ceramide, SM=sphingomyelin, Sph=sphingosine, Sa=sphinganine, S1P=sphingosine 1-phosphate, TBUT=Tear Break-Up Time, MMP=Matrix Metallopeptidase-9, OSDI=Ocular Surface Disease Index, DEQ-5=Dry Eye Questionnaire, ρ Rho=coefficient for dependent variables

^*Statistically significant difference at p-values <0.05

^**p-value <0.01

^***p-value <0.001.

We subsequently examined relationships between relative composition of SPL and various DE symptoms and signs (Table 5). In meibum, the measure most strongly correlated with relative SPL composition was meibum quality. The other DE and MGD metrics examined did not significantly correlate with relative SPL composition. However, several DE measures correlated with relative tear SPL composition. For symptoms, the strongest correlations were with S1P% (rho=0.31, p=0.03 for DEQ5; rho=0.41, p=0.003 for OSDI), indicating higher symptoms severity with increasing relative composition of S1P in tears. For signs, the strongest correlation was with tear production, assessed via Schirmer score [%Cer (rho=−0.50, p<0.005), %Hex-Cer (rho=0.51, p<0.005), %Sph (rho=0.59, p<0.0005), and %S1P (rho=0.29, p=0.04)]. This suggests that lower tear production was associated with higher %Cer and lower % of other SPL sub-types.

Table 5.

Spearman’s Rho (ρ) coefficients between relative composition of sphingolipids and dry eye (DE) symptoms and signs

Spearma n’s ρ	DE Symptoms		MGD/DE Signs
	DEQ-5	OSDI	Meibomian Quality	Eyelid Vascularity	Meibomian Dropout	Meibomian Plugging	TBUT	Fluorescein Stain	Schirmer Test	MMP
In meibum (% pmole)
Cer	−0.12	−0.24	−0.21	0.10	0.17	0.10	0.03	0.13	0.11	0.18
Cer/SM	−0.10	−0.25	−0.29*	0.13	0.13	0.09	0.03	0.13	0.08	0.20
Hex-Cer	0.03	−0.10	−0.34*	0.15	0.06	0.003	0.15	0.0	0.09	−0.08
Hex-Cer/SM	0.0	−0.16	−0.39**	0.18	0.05	0.04	0.14	0.06	0.08	0.005
SM	0.03	0.23	0.51***	−0.25	0.02	0.0	−0.11	−0.18	−0.004	−0.22
Sph	0.28	0.11	−0.56***	0.19	−0.23	−0.01	0.08	0.19	−0.14	0.13
Sa	0.07	−0.08	−0.42**	0.10	−0.16	−0.02	0.07	0.10	−0.19	−0.002
In tears (% pmole)
Cer	−0.008	−0.06	−0.07	0.04	−0.03	−0.11	−0.12	−0.03	−0.49***	0.02
Cer/SM	0.08	0.08	−0.14	−0.01	−0.08	−0.11	−0.06	−0.006	−0.40**	0.04
Hex-Cer	0.05	0.12	0.27	−0.30*	0.27	0.25	−0.09	0.19	0.51***	0.18
Hex-Cer/SM	0.09	0.17	0.25	−0.31*	0.26	0.25	−0.06	0.19	0.51***	0.12
SM	−0.20	−0.36*	0.18	0.19	0.03	−0.03	−0.08	−0.20	−0.04	−0.02
Sph	0.07	0.22	0.46**	−0.27	0.28	0.20	−0.05	0.41*	0.59***	0.002
S1P	0.31*	0.41**	−0.21	−0.06	−0.07	−0.007	0.15	0.06	0.29*	−0.005

DE=Dry Eye, SPL=total sphingolipid, Cer=ceramide, Hex-Cer=hexosyl ceramide, SM=sphingomyelin, Sph=sphingosine, Sa=sphinganine, S1P=sphingosine 1-phosphate, TBUT=Tear Break-Up Time, MMP=Matrix Metallopeptidase-9, OSDI=Ocular Surface Disease Index, DEQ-5=Dry Eye Questionnaire, ρ Rho=coefficient for dependent variables

^*Statistically significant difference at p-values <0.05

^**p-value <0.01

^***p-value <0.001.

We then examined for correlations between meibum and tear SPL pmole (Supplementary Table 2). Overall, no significant correlations were found with the exception of S1P pmole in tears which was lower in individuals with higher Cer, Hex-Cer, and SM in meibum (rho=−0.35, p=0.01 for Cer; rho=−0.29, p=0.04 for Hex-Cer, rho=−0.52, p<0.0005 for SM). In a similar manner, mostly insignificant relationships were noted between relative composition of SPL in meibum and tears (Supplementary Table 3).

Each sphingolipid class is composed of lipids of different fatty acid chain lengths. In humans, six isoforms of ceramide synthase (CerS) have been identified and each produces a different subset of Cer, containing fatty acid chains of 14 to 32 carbons [48]. We investigated the mole percentages of major Cer species and found some significant differences between the two groups. Specifically, individuals with poor meibum quality had significantly higher median mole percent of Cer C16:0 and C24:1, but significantly lower median mole percent of C18:0, C20:0, and C22:0 (Table 6). In contrast, no significant differences in relative composition of Cer chain lengths were noted between the groups in tears. Checking ratios of short-chain ceramides, which are recognized as biomarker for metabolic diseases [49], we found the ratio of C16:0/C18:0 is significantly higher in poor quality meibum.

Table 6.

Composition of major ceramide species (mole percent) in meibum and tears grouped by meibum quality.

	Meibium			Tears
Fatty Acid Chain Length	Good Quality	Poor Quality	P-value	Good Quality	Poor Quality	P-value
Cer
Cer C14:0	0.09	0.12	0.21	0.12	0.11	0.15
Cer C16:0	2.12	4.54	0.006	4.69	5.30	0.49
Cer C18:1	0.15	0.24	0.05	0.54	0.65	0.18
Cer C18:0	3.27	1.86	0.001	0.62	0.58	0.47
Cer C20:0	2.23	1.26	0.001	0.52	0.54	0.50
Cer C22:0	2.75	1.79	0.001	1.30	1.08	0.29
Cer C24:1	0.75	1.40	<0.0005	1.45	1.47	0.50
Cer C24:0	3.21	2.55	0.22	2.02	1.78	0.06
Cer C26:1	0.32	0.20	0.19	0.29	0.24	0.07
Cer C26:0	1.34	1.08	0.12	0.58	0.44	0.06
Cer C16:0/C18:0	0.84*	2.10	<0.0005	7.70	10.30	0.03
Cer C16:0/C18:1	18.38*	18.94	0.60	8.81	8.31	0.05
SM
SM C14:0	1.56	1.80	0.02	1.69	1.74	0.65
SM C16:0	33.97	32.45	0.24	33.05	35.07	0.01
SM C18:1	0.49	0.66	<0.0005	0.58	0.63	0.08
SM C18:0	3.08	4.55	<0.0005	4.53	4.35	0.57
SM C20:0	2.61	3.66	<0.0005	3.91	3.38	0.04
SM C22:0	4.56	6.88	<0.0005	8.19	8.01	0.81
SM C24:1	8.69	11.77	<0.0005	8.92	10.49	0.05
SM C24:0	5.32	9.03	<0.0005	10.25	10.07	0.81
SM C26:1	0.65	0.73	0.04	0.69	0.61	0.13
SM C26:0	0.92	1.10	0.008	1.43	1.30	0.65

Cer=ceramide, SM=sphingomyelin

^*n=18 and 20, respectively due to division by 0

We also compared mole percentages of SM species between the groups and again found significant differences that mirrored what we found with regards to total SM%. Specifically, individuals with poor quality meibum had significantly more C14:0, C18:1, C18:0, C20:0, C22:0, C24:1, C24:0, C26:1, and C26:0 compared to individuals with good meibum quality. Overall, these differences were not as robust in tears.

4. Discussion

In this study, we examined relationships between total SPL and SPL composition in meibum and tears with various MGD and DE parameters. Overall, individuals with poor meibum quality had a higher quantity of SPL recovered from meibum and tears. Compositionally, individuals with poor meibum quality had meibum SPL composed of less Cer, Hex-Cer, and Sph and more SM compared to individuals with good quality meibum. Interestingly, relative SPL composition in tears did not mirror that of meibum, with greater SPL variability in tears. For example, while we did not detect the bioactive SPL S1P in meibum, we found that individuals with good meibum quality had a significantly greater % relative composition of tear S1P compared to those with poor meibum quality. In addition, unlike in meibum, individuals with poor meibum quality had tear SPL comprised of more Hex-Cer by % composition compared to individuals with good quality meibum. Finally, different relationships were noted between ocular surface parameters and SPL in the two compartments. In meibum, SPL pmole and relative composition most strongly correlated with MG metrics while in tears, SPL pmole and relative composition most strongly correlated with tear production. In both compartments, specifically Cer pmole in meibum and S1P% in tears, correlated with DE symptoms.

When comparing our findings with our prior work, some differences and some similarities are noted. Previously, we found that individuals with poor vs good meibum quality had similar pmol of SPL recovered from meibum [34], while in our current study, we found significantly higher levels of SPL pmol recovered from individuals with poor quality meibum. Variability in methodology between the two different facilities that performed the analyses may have contributed to the noted difference. Along with methodology, other potential variables, such as demographics and ocular surface status, may have contributed to the differences noted between studies. However, both studies are consistent in that relative composition of different SPL classes did not depend on the amount of meibum collected, with lower Cer, Hex-Cer, and Sph and higher SM in individuals with poor vs good quality meibum. Along with this important observation in meibum, the novel contribution of this study was a concomitant examination of tear SPL. The finding of similar SPL in both compartments suggest that the major source of tear SPL are the meibomian glands. However, differences in relative proportions of some SPL sub-types also suggest that tear SPL may also be generated or modified locally by cells and enzymes present at the level of the ocular surface [46, 50, 51]. Further data that support this hypothesis include the greater complexity of tear lipids compared to meibum, the local production of lipids in other locations, such as by the type II alveolar epithelial cells in lungs [52, 53], and the presence of SPL modifying enzymes (neutral SMase and ceramidase) in tears [50].

SPL metabolism is complex and consists of many enzymes, alternate pathways, and feedback loops, with ceramides as a central hub.[54]. The most bioactive sphingolipids are ceramide (Cer), sphingosine (Sph), and sphingosine 1-phosphate (S1P) and they are readily interconvertable.[55, 56]. Overall, Cer, Sph, and S1P play varied and important roles in inflammatory pathway at cellular and organ levels (directly or indirectly); additionally, Cer and Sph are known to be pro-apoptotic, and S1P is anti-apoptotic [54]. Ceramide is produced via three pathways, a de novo pathway, through breakdown of SM by sphingomyelinases (SMases), or by a salvage pathway that uses Sph generated in lysosomal degradation to synthesize Cer [27, 57]. Cer, in turn, can be converted to Sph (in lysosome) and then S1P (in cytoplasm), to Hex-Cer, or back into SM (in Endoplasmic reticulum) (Figure 1) [54]. The ratio of Cer and S1P composition is considered as one of the most important metrics when assessing the bioactive role of SPLs. In our prior study we found that individuals with poor meibum quality had a higher ratio of Cer to S1P in meibum [34]. In this study, we could not detect S1P in meibum but found a similar pattern in tears, possibly indicating a relative increase in pro-apoptotic lipids (Cer) in MGD.

Figure 1. Ceramide metabolic pathways and bioactive sphingolipids (SPLs).

The three major pathways of ceramide production are as follows: De novo pathway, sphingomyelinase (SMase) pathway, and Salvage pathway. Bioactive SPLs are shown in Red. Ceramide is synthesized through a de novo pathway catalyzed by serine palmitoyltransferase and ceramide synthase (CerS) or through the breakdown of sphingomyelin by sphingomyelinases (SMase). Sphingosine can be generated from ceramide by ceramidases (CDase) and further converted into sphingosine 1-phosphate (S1P) by sphingosine kinase (SphK). Ceramide and sphingosine are overall pro-apoptotic lipids, whereas S1P has overall anti-apoptotic effects. Ceramide is phosphorylated to Ceramide-1 phosphate (C1P) by cearmide kinase (CerK) plays role in inflammation.

Based on the biology of SPL, it is plausible that specific SPL profiles in meibum and tears have both structural and functional consequences. Structurally, SPLs help maintain the integrity of the tear film by their integration in the polar lipid monolayer that interfaces between the non-polar lipids and aqueous-mucin layer [13]. Additionally, SPLs, especially, SM may act as a scaffold at the interface of non-polar bulk wax and the cholesterol esters above it [58]. In an in vitro study, increasing the relative percentage of Cer increased rigidity, melting temperature, and tear film instability [31], while in an epinephrine induced MGD rabbit model, hyperkeratinization was associated with increased Cer% in meibum [59]. Another group examined meibum SPL levels in individuals with chronic anterior and/or posterior blepharitis with (n=47) and without (n=84) aqueous tear deficiency (ATD, corneal staining, low tear meniscus). They found lower mean SM% in individuals with chronic blepharitis and ATD to those without ATD (7.2% SD 4.8 vs 12.6% SD 5.6, p<0.05) [17]. Yet, another study of meibum lipid composition in Asian participants found a trend for higher meibum SM% in individuals with DE symptoms (OSDI>21, n=27) as compared to controls (OSDI <12.9, n=10) (0.061%±0.014 vs. 0.032%±0.005, p=0.06) [60]. Taken together, it is difficult to compare findings across studies given the use of different models and disease definitions but there may be an optimal composition of SPL metabolites, with disease manifesting when this balance is awry.

Our study findings must be considered in light of its limitations including our MGD definitions and our collection and analysis methodology. In a previous study, meibum quality was graded on a 0-2 scale (clear, cloudy, and yellow) with increasing quantities of unsaturated free fatty acid (FFA) noted with increasing meibum grade [61]. In our study, we examined various classes of SPL but not FFA and graded meibum quality using a clinically standard scale (range 0-4: clear, cloudy, granular, toothpaste, no meibum extracted) [35, 62]. Furthermore, we subclassified meibum into “poor” vs “good” categories based on our clinical interpretation of the grading [34].

Additional methdologic considerations are that we used topical anesthetic and forceful expression to obtain our samples and therefore the extracted meibum may not be fully representative of the physiological meibum available to the ocular surface during normal blinking. While no studies have directly compared the effect of meibum expression methods on SPL sample recovery and composition, a study involving 5 healthy non-contact lens wearers, found that forceful expression and collecton on cotton buds led to a greater recovery of phospholipids, including SM, compared to meibomian gland forceps and meibomian gland evaluators [41]. This study provided the rationale for our use of cotton buds in our current and past studies [34]. However, future studies are needed to examine the impact of the collection instrument, anesthetic use prior to collection, and periodic artificial tear use on recovered SPL composition.

Beyond collection, we used a methodology that only examined for SPL in meibum via targeted analysis which is different from prior studies that examined a wide range of lipids, including SPL [14, 17, 46, 63]. Both methods have strengths and weaknesses. We chose to focus specifically on SPL as obtaining information on the complete lipidome in meibum is challenging methodologically. In fact, examining studies that have attempted to do this has resulted in diversified and sometimes conflicting outcomes [64, 65]. Overall, past studies have detected SPL and specifically sphingomyelins (SM) in both meibum and tears [46, 60, 65]. However, the % have varied between studies. For example, one group found SM to comprise 41% of phospholipids in meibum and 15% of phospholipids in tears [46]. Another group assessed SM as a proportion of polar lipids and found it to comprise 7-14% of meibum polar lipids [14, 17]. Wide ranges have been similarly found in tear lipidomes, with one group finding phosphatidylcholine to comprise 70% of all observed tear film lipids [52], while another paper proposes modest rates of 20% of phospholipid classes [46]. With our analysis, we detected a number of species of Cer, Hex-Cer, and SM along with only long-chain base containing species of SPL such as Sph, sphinganine and S1P [34]. However, a limitation of this analysis is that we cannot comment on SPL as a % of other lipids as those were not examined in our study. Overall, additional research is needed to characterize the effects of specific protocols on recovery and composition of SPL [64]. In the future, we can consider alternative methods, such as monophasic extraction rather than the Bligh and Dryer extraction, which has been found to produce higher yields in phospholipids from human lens (and could show similar benefits in meibum) [66]. Additional limitations that need to be considered are the study’s cross-sectional design which precludes the ability to comment on temporal variability in SPL profiles, lack of concomitant data on tear inflammatory markers, and a specific statistical methodology. On the other hand, the strength of our study was a concomitant comparison of both pmole and relative composition (%) of SPL in tears and meibum and their relationships to a wide variety of symptoms and signs.

Our findings have potential therapeutic implications in MGD. The increased SM we observed the meibum of individuals with MGD could be a key factor in the pathophysiology of disease and its subsequent tear film abnormalities. Specifically, SMases, that may be produced by a variety of bacteria living on the ocular surface [50, 67], can convert SM to Cer and Cer to Hex-Cer, both of which were noted to be elevated in tears. In addition, SMase activation can contribute to ocular surface inflammation, another component of MGD and DED [50]. As such, SMases may be a future potential target in MGD and DED, as they have become other diseases including cystic fibrosis (CF), cancer, and atherosclerosis [68]. For example, a heterogenous group of compounds, some of which are already approved Federal Drug Administration (FDA) functionally inhibit acid SMase (aSMase) [69]. In CF, inhalation of SMase inhibitory tricyclic antidepressants decreased Cer concentration and reduced inflammation in treated mice compared to controls [70]. Further, CF patients treated with amitriptyline (25 mg, 50 mg, and 75 mg doses) daily for 14 days had decreased nasal epithelial cell Cer concentrations on immunofluorescence compared to those treated with placebo (0.92 ± 0.67 relative intensity vs 2.0 ± 1.55 relative intensity) [71]. This translated into improved forced expiratory volume in 1 second (FEV1) in 20 individuals compared to the 39 individuals treated with placebo (+7.6 ± 7.0%, p=<0.001 vs. −1.8 ± 3.3%, p=0.010) [72].

In a similar manner, fingolimod (FTY720), a Sph analogue, is approved for the treatment of multiple sclerosis. Fingolimod acts as an immunosuppressant by internalizing S1P receptors critical in T lymphocyte signaling and is believed to have off-target antitumor effects [73]. In previous studies, we demonstrated that FTY720 can inhibit Cer generation in retina and human corneal epithelial cells by inhibiting either de novo or SMase pathway [74, 75]. Applied to DED, In a NOD DED mouse model, mice treated with fingolimod 0.005% topically, 3 times a day for 4 weeks had higher TBUT compared to controls (0.005%: 3.89s ± 0.21 vs. controls: 1.86s ± 0.16s, p < 0.05). Concomitantly, tear levels of the inflammatory marker IL-1B were decreased in the case and increased in the control group, compared to baseline [76]. Our current findings and prior studies support the need for future research examining the role of topical SPL modulation as a treatment of MGD and DED.

Highlights.

• Individuals with poor meibum quality had more sphingolipids (SPL) recovered from meibum and tears compared to individuals with good meibum quality.

• The bioactive SPLs Ceramide (Cer) and Sphingosine (Sph) were detected both in meibum and tears while sphingosine 1-phosphate (S1P) was additionally detected in tears.

• SPL profiles in individuals with poor meibum quality differed between meibum and tears.

• SPL profiles in both meibum and tears correlated with various dry eye symptoms and signs.

Abbreviations:

DE

dry eye

ADDE

aqueous deficient dry eye

EDE

evaporative dry eye

TFLL

tear film lipid layer

MG

meibomian glands

MGD

meibomian glands dysfunction

SPL

Sphingolipids

Cer

ceramide

Sph

sphingosine

C1P

ceramide 1-phosphate

S1P

sphingosine 1-phosphate

Hex-Cer

hexosyl ceramide

SM

sphingomyelin

Sa

sphinganine

SMase

sphingomyelinase

ATD

aqueous tear deficiency

aSMase

acid sphingomyelinase

FFA

free fatty acid

pmole

pico mole