Sleep Deprivation Induces Dry Eye Through Inhibition of PPARα Expression in Corneal Epithelium

Abstract

Purpose: To determine if sleep deprivation induces dry eye through altering peroxisome proliferator-activated receptor alpha (PPARα) expression in mice.

Methods: The “stick over water” sleep deprivation–induced dry eye (SDE) model evaluated PPARα involvement in inducing this condition. Scanning electron microscopy (SEM) examined microvilli morphology in superficial corneal epithelial cells (SCECs) in SDE and PPARα^−/− mice. Quantitative RT-PCR (qRT-PCR) and Western blot (WB) or immunostaining evaluated PPARα, carnitine palmitoyl transferase 1α (CPT1α), and transient receptor potential vanilloid 6 (TRPV6) expression levels and Ezrin phosphorylation status. Hematoxylin-eosin and Oil Red O staining characterized meibomian gland morphology and corneal lipid accumulation, respectively. Phenol red cotton threads measured tear production. In cultured corneal epithelial sheets, qRT-PCR, WB, and SEM determined the individual effects of fenofibrate and MK886 (PPARα agonist and antagonist, respectively) on PPARα, TRPV6 expression, and SCEC microvilli morphology.

Results: Corneal epithelial lipid accumulation, microvilli morphologic changes, and decreased tear production were associated with marked declines in PPARα, CPT1α, and TRPV6 expression levels as well as Ezrin phosphorylation status, whereas meibomian glands were unaltered in SDE mice. These effects of SDE mice mimicked those in their nonstressed PPARα^−/−counterpart. Topical application of fenofibrate reversed these effects in SDE corneas. In cultured corneal epithelial sheets, fenofibrate increased PPARα and TRPV6 gene and protein expression levels and restored microvilli morphology, whereas MK886 attenuated these changes.

Conclusions: Sleep deprivation induces dry eye through abnormal SCEC microvilli morphology, which is caused by sequential downregulation of PPARα, TRPV6 expression, and Ezrin phosphorylation status in mice.

Dry eye has become the most common ocular surface disease around the world. It is defined by the international TFOS Dry Eye Workshop II (TFOS DEWS II) as a “multifactorial disease of the tears and ocular surface that results in symptoms of discomfort, visual disturbance, and tear film instability with potential damage to the ocular surface.”¹ As of today, there is no selective dry eye therapeutic method. To deal with this limitation, additional insight is needed regarding the underlying pathophysiology of dry eye disease. 

Sleep deprivation (SD) can cause dry eye. A retrospective study of veterans with sleep-depriving apnea in the United States between 2005 and 2010 has found that this condition is associated with an increased risk of dry eye development.² In a clinical study, 24 hours of SD induced dry eye symptoms that include increased tear film hyperosmolarity, shortened tear break-up time, and reduced tear secretion.³ Furthermore, in a survey of a large Korean adult population, increases in dry eye incidence are associated with declines in sleep duration.⁴ Similarly, other studies^5,6 have reported an increased prevalence of dry eye in sleep-deprived patients. SD can be a consequence of a large amount of time individuals spend sitting in front of video terminals for both work and recreation, which can reduce sleep time below that required for good health. It is reported that adolescents and adults need for good health, on the average, 9 and 7 hours of sleep per day, respectively.^7,8 Such stress in combination with other environmental and genetic factors can contribute to development of this disease.⁹ The results of a survey of high-school students and adults shows that the distribution of sleep duration is 92.4% for <9 hours and 37.1% for <7 hours, respectively.^10,11 The mechanism underlying sleep deprivation–induced dry eye (SDE) remains unclear, even though there are therapeutic options merely providing symptomatic relief. 

We previously have described an SDE mouse model by using a “stick over water” method. This imposed stress keeps mice awake while they balance themselves on a stick to avoid falling down into the water owing to muscle atonia during sleep.¹² The pathologic changes that arise mimic those typically described in dry eye patients except that the mice do not develop ocular surface and lacrimal gland inflammation. Otherwise, conjunctival goblet cell abundance increases along with lacrimal gland hypertrophy. Furthermore, SCEC microvilli shorten and become less dense.¹² These findings suggest that a cause and effect relationship exists between SD and declines in microvilli morphology. 

TRPV6 is one of the 26 different isoforms in the transient receptor potential (TRP) channels superfamily expressed in mammals. It is the only highly Ca²⁺-selective channel in this superfamily.¹³ TRPV6 activation contributes to sustaining microvilli morphology by inducing Ca²⁺ transients maintaining epithelial intracellular calcium homeostasis.¹⁴ Such activity is dependent on adenosine triphosphate (ATP) availability, since TRPV6 channel activity is absent without MgATP.¹⁵ It is reported that TRPV6 expression can be regulated by fatty acid in the rat colorectum.¹⁶ This mechanosensitive channel is activated by fluid shear stress through increases in Akt-Ezrin phosphorylation inducing Ca²⁺ intracellular transients, which trigger microvilli elongation and increases in their formation in human placental trophoblastic cells.¹⁴ 

Corneal epithelial surface microvilli structural integrity promotes tear film adhesion to the SCECs. Such attachment is critical for hydrating and protecting this cellular layer from pathogenic infiltration.^17–19 Since SCEC microvilli integrity is sensitive to external irritants, their structural intactness and dimensions are used as indices for evaluating ocular surface health and eye drop toxicity.^20,21 Furthermore, as such changes affect the tear film adherence, reversing losses in SCEC layer microvilli structure and integrity occurring in dry eye remains a challenge.²² The lipid layer secreted by meibomian glands reduces tear film evaporative losses, which is essential for preserving the underlying corneal epithelial health and integrity.²³ This secretion contains wax esters, triglycerides, free fatty acids, as well as neutral diesters, which are needed to protect the underlying tissue from becoming desiccated.²⁴ One characteristic of meibomian gland dysfunction (MGD) is progressive obstruction of its terminal ducts, which reduces its secretory activity. In turn, such changes can alter epithelial microvilli structural integrity as well as induce lipid-deficient dry eye.²⁵ This condition is the leading cause of dry eye.²⁶ However, whether SD causes MGD to induce lipid-deficient dry eye remains unknown. 

PPARα, is a member of the peroxisome proliferator-activated receptors (PPARs) family.²⁷ This nuclear transcription factor plays a crucial role in regulating fatty acid β-oxidation (FAO) and homeostatic control of lipid metabolism.²⁷ Activation of PPARα promotes uptake, utilization, and catabolism of fatty acids by upregulating genes involved in fatty acid transport, fatty acid binding, and activation of peroxisomal as well as mitochondrial FAO.²⁸ FAO metabolites are oxidative phosphorylation substrates, which on a per mole basis yield more ATP than glucose oxidation. It is possible that ATP availability can be rate limiting for TRPV6 activation.^15,29 With the realization that PPARα modulates FAO and ATP generation, which induces mechanosensitive TRPV6 activation, we hypothesized that a decrease in PPARα-modulated TRPV6 expression disrupts SCEC microvilli morphology and gives rise to dry eye symptomology, since the changes of SCEC microvilli morphology have been described in SDE mice and dry eye patients.^12,30,31 

We describe here the mechanism that accounts for how dry eye develops in the “stick over water” SDE mouse model. The results indicate that SD induces dry eye through disrupting SCEC microvilli morphology, which is caused by sequential declines in PPARα, TRPV6 expression, and Ezrin phosphorylation status. 

Materials and Methods

Sleep Deprivation and Treatment

Both wild-type (WT) and PPARα-knockout (PPARα^−/−) mice in the C57BL/6J background were bred in-house. Adult male C57BL/6J and PPARα^−/− mice (20–22 g) were originally purchased from Shanghai SLAC Laboratory Animal Company Ltd. (Shanghai, China) and The Jackson Laboratory (Bar Harbor, ME, USA), respectively. Animal experiments were performed in accordance with the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vison Research. The Xiamen University Experimental Animal Ethics Committee approved the experimental protocol. All mice were free of ocular surface defects before their use in experiments. C57BL/6J mice were sleep deprived for 20 hours per day for 5 or 10 days as previously described.¹² Thirty mice were randomly divided into three equal-sized groups (n = 10). One group served as an untreated normal control (Ctrl), whereas the other two groups were sleep deprived for either 5 or 10 days. 

To determine an optimal fenofibrate concentration range for topically increasing corneal PPARα expression, 30 animals were randomly divided into six equal groups (n = 5), one of them served as a Ctrl group, whereas the other five groups were sleep deprived for 10 days. These SD groups received eye drops with a pipette (twice per day, 10 μL each time) containing either 50, 100, 200, or 500 μM fenofibrate (dissolved in dimethyl sulfoxide [DMSO] and further diluted with saline) or vehicle (saline containing same volume of DMSO as 500 μM fenofibrate) on the sixth day after SD for 5 consecutive days. After the last treatment, the animals were euthanized by cervical dislocation, their eyes were enucleated, and the corneas were dissected for total RNA extraction as previously reported.³² Quantitative RT-PCR (qRT-PCR) quantified PPARα mRNA expression level. 

To determine the effect of fenofibrate on microvilli morphology of SCECs in SDE mice, 100 animals were randomly divided into four equal groups (n = 25). One of them served as a Ctrl group, whereas the other three groups were sleep deprived for 10 days. One of these three SD groups was left untreated. Both eyes of other animals were treated either with 200 μM fenofibrate or with vehicle (saline containing same volume of DMSO as 200 μM fenofibrate). The fenofibrate and vehicle groups received eye drops with a pipette (twice per day, 10 μL each time) on the sixth day after SD for 5 consecutive days. After termination of the SD period, the animals were euthanized by cervical dislocation, and both of their eyes were enucleated. A small number of the enucleated eyes (n = 80) were fixed with cold 4% paraformaldehyde and then paraffin or frozen sections were prepared for immunohistochemistry (IHC) or immunofluorescence (IF) studies or Oil Red O (ORO) staining. Other enucleated eyes (n = 120) were used to extract protein, isolate RNA, or for fixation in cold 2.5% glutaraldehyde for SEM. 

Materials

Rabbit anti-PPARα, anti-CPT1α, anti-TRPV6, and anti-Ezrin antibodies were obtained from Proteintech Group, Inc. (Wuhan, China). Rabbit anti–phospho-Ezrin-T567 antibody was obtained from Abclonal Biotechnology Co. (Woburn, MA, USA). Horseradish peroxidase (HRP)–conjugated mouse anti-rabbit β-actin, HRP-conjugated goat anti-rabbit IgG, HRP-conjugated rabbit anti-goat IgG antibodies, Hoechst 33342 dye, MK886, Oil Red O dye, and DMSO were obtained from Sigma-Aldrich Corp. (St. Louis, MO, USA). Alexa Fluor 488 donkey anti-rat IgG antibody, Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), and Dispase II were obtained from Invitrogen Corp (Carlsbad, CA, USA). N-Histofine simple stain MAX PO (anti-rabbit) was obtained from Nichirei Biosciences (Tokyo, Japan); the Liquid DAB+ substrate chromogen system was obtained from Dako (Glostrup, Denmark). Fenofibrate was purchased from Abcam (Cambridge, England). 

Hematoxylin-Eosin and Immunohistochemistry Staining

After euthanization, the eyeballs and eyelids were fixed immediately in cold 4% paraformaldehyde for 24 hours followed by embedding in paraffin wax blocks that were cut into 4-μm-thick sections. Hematoxylin-eosin (H&E) staining of eyelids and PPARα and CPT1α IHC of eyeballs were performed after the sections were deparaffinized and rehydrated, respectively.^32,33 The images were observed under a light microscope (Eclipse 50i; Nikon Instruments, Melville, NY, USA), and captured for comparison and analysis. To analyze the relative mean density of positive immunostaining in the cornea, the relative integrated optical density (IOD) analysis was performed.³⁴ 

Immunofluorescence and Oil Red O Staining

After euthanization, rapid eyeball removal, and sample fixation in cold 4% paraformaldehyde for 24 hours, tissue specimens were embedded in OCT. A cryostat microtome prepared 6-μm-thick frozen sections, which was followed by IF and ORO staining.¹² The images of the IF- and ORO-stained tissues were viewed under a confocal microscope (Zeiss LSM 780; Carl Zeiss MicroImaging GmbH, Jena, Germany) and saved for comparison and analysis. Analysis software (NIS Elements version 4.1; Nikon) calculated the mean staining intensity in each section.³⁵ 

RNA Isolation and qRT-PCR

TRIzol reagent (Invitrogen) extracted total corneal RNA according to the manufacturer's instructions. cDNA was synthesized with a reverse transcription kit (RR047A; TaKaRa, Shiga, Japan). QRT-PCR was performed on StepOne Real-Time PCR System (Applied Biosystems, Alameda, CA, USA) by using a SYBR Premix Ex Taq Kit (TaKaRa).³² The Table provides the primer sequences. The amplification program included an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of denaturation at 95°C for 10 seconds and annealing and extension at 60°C for 30 seconds. The comparative threshold cycle (Ct) method analyzed the results of the relative qRT-PCR, which were normalized to β-actin expression.³² 

Table

 

Mouse Primer Sequences Used for Quantitative Real-Time PCR

 

Western Blot Assay

Proteins from mouse corneas and primary corneal epithelial sheets were extracted with cold RIPA buffer containing a proteinase inhibitor cocktail (Roche Applied Science, Mannheim, Germany). Equal amounts of proteins were subjected to electrophoresis with 10% sodium dodecyl sulfate–polyacrylamide gels. The standard WB assay protocol was applied. Specific primary antibodies (anti-PPARα, anti-TRPV6, anti–phospho-Ezrin, and anti-Ezrin) and secondary antibodies (HRP-conjugated goat anti-rabbit IgG) were used. The specific bands were resolved with an enhanced chemiluminescent reagent and recorded by a Molecular Imager ChemiDoc XRS (Bio-Rad, Hercules, CA, USA). Each of the corresponding β-actin expression levels normalized band staining intensity.³² 

Corneal Epithelial Microvilli Morphology

The corneas were carefully dissected along the limbus and quickly fixed in cold 2.5% glutaraldehyde for 6 hours. The samples were then prepared for SEM.³⁶ The images were observed under a SEM (LEO, Oberkochen, Germany) and captured for comparison and analysis. The number and length of SCEC microvilli of captured random fields at the higher power (magnification = 50,000) were calculated by using Image J software (http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA).^37,38 

Culture and Treatment of Mouse Corneal Epithelia

Under aseptic conditions, corneas were kept overnight in DMEM containing Dispase II (2 mg/mL) at 4°C. Corneal epithelial sheets were isolated and cultured in six-well plates with DMEM containing 10% FBS in a humidity-controlled incubator at 37°C in a 95% air and 5% CO₂ mixture and treated with 25 μM fenofibrate, 10 μM MK886, or vehicle (containing the same volume of DMSO as 25 μM fenofibrate or 10 μM MK886). The untreated corneal epithelial sheets served as a normal control. After 24-hour treatment, these corneal epithelial sheets were used to either extract protein, isolate RNA, or for fixation in cold 2.5% glutaraldehyde for SEM to view their microvilli morphology. 

Measurement of Tear Production

Tear production was measured by using phenol red cotton threads (Zone-Quick; Yokota, Tokyo, Japan) at a same time point (8 AM).³⁵ 

Statistical Analysis

Each of the aforementioned experiments was performed in triplicate. Summary data were reported as mean ± SD. Student's t-test evaluated the significance of a difference between two groups. One-way ANOVA analyzed differences between three or more groups in conjunction with the post hoc Tukey test. P < 0.05 was considered to indicate a statistically significant difference. 

Results

Lipid Accumulation in SDE Mouse Corneas

ORO is a fat-soluble diazol dye that mainly stains neutral lipids and cholesteryl esters and is used to detect lipid accumulation.^39,40 ORO staining showed that corneal lipid accumulation increased more in either 5-day or 10-day sleep-deprived than in non–sleep deprived normal mouse corneas (Figs. 1A, 1B). To determine whether this difference is associated with altered PPARα expression levels, we evaluated its gene and protein expression levels and localization. PPARα immunostaining markedly declined throughout the cornea, which was even more dramatic in the epithelium of the 5-day and 10-day groups. These effects accompanied significant declines in both PPARα mRNA and protein expression levels (Figs. 1C–G). To validate such an association between increased lipid accumulation and SD, we determined if CPT1α expression was also altered, since it is a PPARα target gene and a rate-limiting enzyme of FAO.^41–43 Both CPT1α gene expression and IHC staining intensity declined, which is consistent with declines in PPARα expression induced by SD (Figs. 1H–J). This effect allegedly leads to declines in FAO followed by increases in lipid accumulation in SDE corneas. 

Figure 1

 

Sleep deprivation increases corneal lipid accumulation. Representative images (A) and mean fluorescence intensity (B) of ORO staining (red labeling, which is indicated by white arrows) in normal and SDE mouse corneas. Scale bars: 100 μm. Representative mRNA (C) and protein (D, E) expression levels, IHC images (F), and relative IOD analysis (G) of PPARα in normal and SDE mouse corneas (PPARα IHC is labeled brown and indicated by black arrows). Scale bars: 50 μm. Representative mRNA expression level (H), IHC images (I), and relative IOD analysis (J) of CPT1α in normal and SDE mouse corneas (CPT1α IHC is labeled brown and indicated by black arrows). Scale bars: 50 μm. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Meibomian Gland Morphology Unaltered by SD

The MGD animal model is characterized by eyelid histopathology involving inflammatory cells, hyperkeratinization of the ductal epithelium, and cystic dilatation of the ducts in the meibomian glands. Besides, the orifices are plugged with a proteinaceous substance and the acinar cells are hypertrophic and/or hyperplastic. ORO-stained meibum is seen in and amongst the dilated ducts.^44,45 To determine whether SD induces MGD resulting in corneal abnormal lipid metabolism, we performed H&E and ORO staining of the meibomian gland in SDE mice. Its morphology was the same as that of normal mice. Specifically, the meibomian duct and acini were clearly visible (Fig. 2A) and exhibited ORO-positive staining (Fig. 2B). ORO stained meibum in the dilated ducts, and other histopathologic features of MGD-like disease were absent. Therefore, SD had no effect on meibomian gland function and histology. 

Figure 2

 

Meibomian gland morphology in SDE mice. (A) H&E staining of duct (asterisks) and acini (triangles) of meibomian gland in SDE and normal mice. Scale bars: 50 μm. (B) ORO staining of meibomian gland of SDE and normal mice. The duct and acini of meibomian gland are indicated by black asterisks and triangles, respectively. Scale bars: 100 μm.

Abnormal TRPV6 Expression and Microvilli Morphology in SDE Mouse Corneas

We determined if TRPV6 modulation is involved in mediating SD-induced changes in microvilli structure, since it has been reported that changes in its activity affect microvilli structure in human placental trophoblastic cells.¹⁴ Furthermore, MgATP availability modulates this channel whose activity can be affected by changes in fatty acid levels interacting with positive response elements on the TRPV6 gene promoter.¹⁶ Accordingly, we hypothesized that SD modulates TRPV6 expression, since this stress increased epithelial lipid accumulation and presumably decreased both fatty acid levels and MgATP availability. To determine if SD-induced increases in lipid accumulation affect TRPV6 activity, the effect of this stress was measured on TRPV6 gene and protein expression levels in sleep-deprived and normal corneas. Both corneal TRPV6 mRNA and protein expression levels significantly declined after 5 days and 10 days of SD, as compared with those in normal corneas (Figs. 3A–C). Similarly, TRPV6 IF was downregulated (Figs. 3D, 3E). In human placental trophoblastic cells, fluid flow stress promotes microvilli formation and elongation through sequential increases in TRPV6 expression and Ezrin phosphorylation status. Ezrin is a microvilli-delimited protein linker between the microvillus plasma membrane and its actin cytoskeleton whose phosphorylation status affects microvilli formation.^14,46 Furthermore, microvilli structure affects the tear film stability.⁴⁷ Accordingly, we determined if such an effect also occurs in SDE mouse corneas. The SCEC microvilli of normal mice had closely apposed finger-like membrane processes that were more than 700 nm long with a smooth surface, whereas these projections were shorter, rougher, and less dense in the SCECs of SDE mice (Figs. 3F–H). Only the Ezrin phosphorylation status rather than its protein expression level declined in SDE compared to normal corneas (Figs. 3I, 3J). These results were associated with decreased tear production (Fig. 3K) in SDE mice. Therefore, change in the Ezrin phosphorylation status contributes to mediating TRPV6 control of corneal epithelial microvilli structural integrity and tear film stability in mice. 

Figure 3

 

TRPV6 expression and microvilli morphology in SDE mouse corneas. Representative mRNA (A) and protein (B, C) expression levels, IF images (D), and mean IF intensity (E) of TRPV6 in normal and SDE mouse corneas (TRPV6 IF is labeled green and indicated by white arrows). Scale bars: 20 μm. SEM comparison of SCEC microvilli number (F), length (G), and morphology (H) in normal and SDE mice. Scale bars: 2 μm (upper row) and 0.2 μm (lower row). Ezrin phosphorylation status (p-Ezrin) and total-Ezrin expression (I, J) in normal and SDE mouse corneas. (K) Phenol red cotton thread measurement of tear production in normal and SDE mice. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Lipid Accumulation in PPARα^−/− Mouse Corneas

To validate PPARα involvement in controlling epithelial lipid accumulation, we compared the effects of loss of PPARα function with that of SDE mice on ORO staining and CPT1α gene and protein expression. In PPARα^−/− mice, ORO staining intensity (Figs. 4A, 4B) was larger, whereas the CPT1α expression (Figs. 4C−E) was less than in normal corneas. Therefore, the declines in CPT1α gene and protein expression along with increases in lipid accumulation caused by the loss of PPARα gene function confirm that PPARα expression is essential for promoting lipid metabolism in corneas. 

Figure 4

 

Lipid accumulation in PPARα^−/− mouse corneas. Representative images (A) and mean fluorescence intensity (B) of ORO staining (red labeling, which is indicated by white arrows) in normal and PPARα^−/− mouse corneas. Scale bars: 100 μm. Representative mRNA expression level (C), IHC images (D), and relative IOD analysis (E) of CPT1α in normal and PPARα^−/− mouse corneas (CPT1α IHC is labeled brown and indicated by black arrows). Scale bars: 50 μm. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Abnormal TRPV6 Expression and Microvilli Morphology in PPARα^−/− Mouse Corneas

Similar to SDE mice, TRPV6 expression (Figs. 5A−C) was significantly reduced in PPARα^−/− mouse corneas, which is consistent with declines in TRPV6 IF (Figs. 5D, 5E). Furthermore, the association of altered microvilli morphology (Figs. 5F–H) and decreased Ezrin phosphorylation status (Figs. 5I, 5J) with declines in tear production (Fig. 5K) also occurred in PPARα^−/− mice. This correspondence between SDE and PPARα^−/− mice indicates that the declines in PPARα function occurring in SDE corneas contribute to declines in TRPV6 activation and disruption of microvilli morphology. 

Figure 5

 

TRPV6 expression and microvilli morphology in PPARα^−/− mouse corneas. Representative mRNA (A) and protein (B, C) expression levels, IF images (D), and mean IF intensity (E) of TRPV6 in normal and PPARα^−/− mouse corneas (TRPV6 IF is labeled green and indicated by white arrows). Scale bars: 20 μm. SEM comparison of SCEC microvilli number (F), length (G), and morphology (H) in normal and PPARα^−/− mice. Scale bars: 2 μm (upper row) and 0.2 μm (lower row). Ezrin phosphorylation status (p-Ezrin) and total-Ezrin expression (I, J) in normal and PPAR^−/− mouse corneas. (K) Phenol red cotton thread measurement of tear production in normal and PPAR^−/− mice. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Fenofibrate Reverses Declines in Lipid Metabolism in SDE Mouse Corneas

Fenofibrate is a PPARα agonist widely used in the clinic that promotes FAO and lipid metabolism through a wide range of effects.^48,49 For example, fenofibrate can be compounded as an ophthalmic solution for treating corneal alkali burn in rats.⁵⁰ Accordingly eye drops containing either fenofibrate (50, 100, 200, 500 μM) or just its vehicle were administered daily to determine if they could reverse declines in lipid metabolism in 10-day corneas. Only 200 and 500 μM fenofibrate significantly increased PPARα gene expression (data not shown). Therefore, 200 μM fenofibrate was chosen as an appropriate dose for topical administration. The ORO staining intensity (Figs. 6A, 6B) was weaker, while PPARα (Figs. 6C–G) and CPT1α (Figs. 6H–J) expression levels were significantly greater in fenofibrate-treated corneas than in vehicle-treated corneas. These differences suggest that fenofibrate treatment reversed declines in lipid metabolism in SDE corneas through increasing PPARα expression. 

Figure 6

 

Fenofibrate reversal of increased lipid accumulation in SDE mouse corneas. Representative images (A) and mean fluorescence intensity (B) of ORO staining (red labeling, which is indicated by white arrows) in normal, 10-day SDE, vehicle-treated, and fenofibrate-treated mouse corneas. Scale bars: 100 μm. Representative mRNA (C) and protein (D, E) expression levels, IHC images (F), and relative IOD analysis (G) of PPARα in normal, 10-day SDE, vehicle-treated, and fenofibrate-treated mouse corneas (PPARα IHC is labeled brown and indicated by black arrows). Scale bars: 50 μm. Representative mRNA expression level (H), IHC images (I), and relative IOD analysis (J) of CPT1α in normal, 10-day SDE, vehicle-treated, and fenofibrate-treated mouse corneas (CPT1α IHC is labeled brown and indicated by black arrows). Scale bars: 50 μm. Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

PPARα Regulates SCEC TRPV6 Expression and Microvilli Morphology

We determined if such reversal is accompanied by changes in TRPV6 and SCEC microvilli morphology. Fenofibrate (200 μM) significantly increased TRPV6 expression (Figs. 7A–E) and partially reversed the changes in SCEC microvilli morphology (Figs. 7F–H) in SDE corneas. Fenofibrate (25 μM) increases PPARα activation in cultured ECC-1 endometrial cancer cells.⁵¹ On the other hand, the PPARα-specific antagonist MK886 (10 μM) reduces its expression in cultured A549 cells.⁵² We determined if this dosage of fenofibrate for 24 hours upregulated PPARα expression and improved cultured SCEC microvilli morphology through TRPV6 expression upregulation. This PPARα agonist significantly increased PPARα and TRPV6 expression levels (Figs. 8A–C) and maintained normal SCEC microvilli morphology, whereas its vehicle had no effect (Fig. 8B). On the other hand, MK886 (10 μM) exposure for 24 hours decreased PPARα and TRPV6 expression (Figs. 8D–F). In addition, this PPARα antagonist shortened SCEC microvilli and caused them to become rougher and less dense (Figs. 8G–I), whereas the effect of vehicle treatment on microvilli morphology was similar to that in normal SCECs group. These opposing effects of fenofibrate and MK886 confirm PPARα involvement in preserving microvilli morphology through sustaining TRPV6 expression. 

Figure 7

 

Fenofibrate rescues declines in TRPV6 expression and altered microvilli morphology in SDE mouse corneas. Representative mRNA (A) and protein (B, C) expression levels, IF images (D), and mean IF intensity (E) of TRPV6 in normal, 10-day SDE, vehicle-treated, and fenofibrate-treated mouse corneas (TRPV6 IF is labeled green and indicated by white arrows). Scale bars: 20 μm. A SEM comparison of microvilli numbers (F), length (G), and morphology (H) in normal, 10-day SDE, vehicle-treated, and fenofibrate-treated mice. Scale bars: 2 μm (upper row) and 0.2 μm (lower row). Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 8

 

PPARα modulation affects SCEC TRPV6 expression and microvilli morphology. Representative mRNA (A) and protein (B, C) expression levels of PPARα and TRPV6 in normal, vehicle-treated, and fenofibrate-treated cultured mouse corneal epithelial sheets. Representative mRNA (D) and protein (E, F) expression levels of PPARα and TRPV6 in normal, vehicle-treated, and MK886-treated cultured mouse corneal epithelial sheets. SEM comparison of microvilli number (G), length (H), and morphology (I) in SCECs of normal, vehicle-treated, fenofibrate-treated, and MK886-treated cultured mouse corneal epithelial sheets. Scale bars: 1 μm (upper row) and 0.2 μm (lower row). Data are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

We used the SDE mouse model to gain insight into the underlying mechanisms inducing dry eye symptomology in sleep-deprived individuals. The agreement between the effects of SD and loss of PPARα function indicates that declines in PPARα and CPT1α expression lead to increases in corneal lipid accumulation, which may in turn decrease fatty acid availability and inhibit TRPV6 expression. Such declines in TRPV6 expression presumably suppress Ca²⁺ influx to levels that decrease the Ca²⁺-dependent Ezrin phosphorylation status and disrupt microvilli morphology of SCECs in SDE mice. 

As fenofibrate reversed SDE-induced declines in PPARα and CPT1α expression, targeting this transcription factor may provide a novel approach for treating this condition in sleep-deprived patients. This conjecture agrees with the findings that either MK886 or loss of PPARα gene function had opposite effects to those of fenofibrate on lipid accumulation, PPARα, CPT1α, and TRPV6 expression levels, and microvilli morphology and ocular surface health. Owing to TRPV6 functional impairment, such an effect may disrupt epithelial microvilli formation, which in turn reduces tear film adherence to the epithelial surface, leading to desiccation and compromise of its barrier function. This suggestion is in accord with a report in which mechanosensitive TRPV6 activation triggers microvilli formation in human placental trophoblastic cells.¹⁴ If future clinical studies confirm such an association, it is conceivable that these surface changes contribute to sleep-deprived dry eye disease. 

We determined if PPARα expression modulation affects changes in corneal lipid accumulation because increases in its expression level stimulate lipid metabolism in fatty liver disease.⁴² Our results indicate such an association also exists in the cornea, since SD induced declines in its expression, which caused epithelial lipid accumulation to rise. On the other hand, increasing PPARα expression with fenofibrate suppressed the SD-induced increases in lipid accumulation and reversed declines in PPARα, CPT1α, and TRPV6 gene and protein expression. MK886 instead inhibited PPARα and TRPV6 gene and protein expression levels, which validates PPARα involvement in mediating SDE symptomology. Furthermore, their expression levels fell to levels similar to those in SDE corneas and PPARα^−/− corneas. Therefore, SD-induced declines in PPARα expression promote dry eye symptomology by disrupting microvilli structure through sequential declines in CPT1α and TRPV6 gene and protein expression levels, leading to decreases in the Ezrin phosphorylation status. 

Even though declines in both PPARα and CPT1α gene and protein expression levels account for increases in lipid accumulation and abnormal microvilli morphology in SDE and PPARα^−/− mouse corneas, it is unclear how resulting increases in lipid accumulation suppressed TRPV6 expression and affected microvilli morphology. One possibility is that increases in lipid accumulation reduced fatty acid availability. This condition could lead to declines in FAO substrates that are large enough to cause ATP generation to fall below a level adequate for sustaining ATP-dependent functional TRPV6 activity. This is conceivable, since on a per mole basis fatty acids generate more ATP than glucose oxidation through glycolysis.^{14–16,53,54} Another possibility is that reduced fatty acid availability due to lipid accumulation may directly suppress TRPV6 gene expression, since fatty acids stimulate TRPV6 expression through interacting with fatty acid–positive response elements on the TRPV6 gene promoter.¹⁶ Increases in TRPV6 expression and activation have a corresponding effect on the Ezrin phosphorylation status, which is an essential factor for microvilli formation.¹⁴ Accordingly, these two possibilities may account for how SD-induced increases in lipid accumulation suppress the expression and activity of TRPV6 that disrupts SCEC microvilli morphology through declines in Ezrin phosphorylation status. 

Dry eye can arise from tear aqueous deficiency and evaporative losses that in turn increase tear film osmolarity and inflammation of the ocular surface.¹ As tears provide nutrient support and lubricate the ocular surface, tear film instability resulting from impaired adherence to the ocular surface may result in dry eye.^1,55 As the SDE model simulates many dry eye disease characteristics that include losses in microvilli morphology, this ocular surface change can contribute to the development of this condition. Other possible factors that can compromise tear film adherence and stability and lead to dry eye include pathologic changes in conjunctival epithelial microvilli as well as alterations in tear film composition.⁴⁷ It is possible that the described increases in conjunctival goblet cell density and lacrimal gland hypertrophy along with lipid accumulation in acini cells are adaptive responses to a decline in tear production in the SDE mice.¹² Nevertheless, we could not identify any meibomian gland morphologic changes suggesting constancy of lipid layer secretion into the tear film. 

Evaporative dry eye resulting in increases in tear film osmolarity can in some cases lead to ocular surface inflammation.¹ This inflammatory response results from inflammatory cells infiltrating into the ocular surface tissues and lacrimal gland in various types of Sjögren's syndrome–related dry eye.^56,57 Stress factors including environmental challenges, infections, endogenic stress, autoimmunity, and genetic factors may also disturb the homoeostatic balance of the ocular surface and activate an acute inflammatory response.^58–60 Increased production and activation of proinflammatory cytokines, such as interleukin 1 and TNF-α, as well as inflammatory cell infiltration into the ocular surface, occur in dry eye.⁶¹ Even though there is evidence indicating that dry eye–related ocular surface inflammation is mediated by lymphocytes,⁶² we failed to detect any proinflammatory cytokine expression and inflammatory cell infiltration into both corneas and/or lacrimal glands in our previous study establishing the SDE mice model.¹² Furthermore, SD also strongly influences neuroendocrine function. Declines in their secretions may also affect lipid metabolism because hormones, neurotransmitters, and neuropeptides, including insulin, glucagon, catecholamines, cortisol, growth hormone, neuropeptide Y, and somatostatin, also mediate control of lipid metabolism.^63–67 It remains an open question warranting future study as to whether SD-induced disruption of oxidative lipid metabolism is due to inhibition of PPARα expression resulting from changes in hormonal, neurotransmitter, and neuropeptide levels. 

In conclusion, the similar disruptive effects of SDE and loss of PPARα function on SCEC microvilli morphology suggest that declines in this transcription factor expression induce these changes through reducing fatty acid availability, TRPV6 expression, and Ezrin phosphorylation status (Fig. 9). Therefore, the maintenance of PPARα expression may be essential for maintaining TRPV6 expression and activity at levels that are high enough to sustain normal microvilli morphology. Under this condition, tear film adherence to the underlying epithelial cell layer promotes ocular surface health. Stimulating PPARα function in the SDE mouse model alleviated ocular surface alterations that are similar to those occurring in some dry eye patients. If future studies show that our working model is translatable to sleep-deprived patients, this transcription factor is a potential target to ameliorate their dry eye symptomology. 

Figure 9

 

Schema describing how SD induced dry eye trough declines in PPARα expression. In normal mice, PPARα expression contributes to supporting normal microvilli morphology formation through promoting fatty acid metabolism, TRPV6 expression, and Ezrin phosphorylation status. This chain of events prevents dry eye by maintaining adequate ocular surface tear film adherence. In contrast, in SDE mice, the inhibition of PPARα function induces abnormal microvilli formation, which disrupts adequate ocular surface tear film adherence and causes dry eye, through decreases in fatty acid availability, which sequentially reduces TRPV6 expression and Ezrin phosphorylation status.

Acknowledgments

The authors thank Sanming Li and Changkai Jia for technical support. 

Supported by National Key R&D Program of China (No. 2018YFA0107304 [ZLiu]) and National Natural Science Foundation of China (No. 81570818 [YC], 81330022 [ZLiu], 81470602 [WL], 81770894 [WL], 81370991 [YC], and 81570816 [ZLin]). 

Disclosure: L. Tang, None; X. Wang, None; J. Wu, None; S. Li, None; Z. Zhang, None; S. Wu, None; T. Su, None; Z. Lin, None; X. Chen, None; X. Liao, None; T. Bai, None; Y. Qiu, None; P.S. Reinach, None; W. Li, None; Y. Chen, None; Z. Liu, None