Study of pathological processes of meibomian gland dysfunction by in vitro culture airlifting conditions

ABSTRACT Meibomian gland dysfunction (MGD) is a group of disorders linked by functional abnormalities of the meibomian glands. Current studies on MGD pathogenesis focus on meibomian gland cells, providing information on a single cell’s response to experimental manipulation, and do not maintain the architecture of an intact meibomian gland acinus and the acinar epithelial cells’ secretion state in vivo. In this study, rat meibomian gland explants were cultured by a Transwell chamber-assisted method under an air-liquid interface (airlift) in vitro for 96 h. Analyses for tissue viability, histology, biomarker expression, and lipid accumulation were performed with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and TUNEL assays, hematoxylin and eosin (H&E) staining, immunofluorescence, Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR), transmission electron microscopy (TEM), and western blotting (WB). MTT, TUNEL, and H&E staining indicated better tissue viability and morphology than the submerged conditions used in previous studies. Levels of MGD biomarkers, including keratin 1 (KRT1) and 14 (KRT14) and peroxisome proliferator-activated receptor-gamma (PPAR-γ), along with oxidative stress markers, including reactive oxygen species, malondialdehyde, and 4-hydroxy-2-nonenal, gradually increased over culture time. The MGD pathophysiological changes and biomarker expression of meibomian gland explants cultured under airlift conditions were similar to those reported by previous studies, indicating that abnormal acinar cell differentiation and glandular epithelial cell hyperkeratosis may contribute to obstructive MGD occurrence.


Introduction
Meibomian glands (MGs) are important organs maintaining the health of the ocular surface. They are sebaceous glands specific to the eyelids that secrete meibum in a holocrine manner. Meibum is transported to the gland's opening through the MG duct system and finally spreads onto the ocular surface through orbicularis oculi muscle contractions during blinking. Meibum is the main lipid layer component of the tear film's surface that promotes its stability and prevents the evaporation of its liquid components.
MG dysfunction (MGD) plays a crucial role in dry eye development [1], which can lead to instability of the tear film, ocular surface damage, and visual disturbances [2]. A low meibum delivery state caused by MG duct blockage is the most common mechanism underlying MGD [3]. Keratin overexpression and lipid differentiation are involved in its pathogenesis, eventually leading to duct blockage [4][5][6]. Excessive active oxygen destroys cellular DNA, lipids, and proteins and promotes acinar epithelial cell dysfunction and death [7]. Oxidative stress is often associated with age-related chronic diseases, including age-related MGD. The role of oxidative stress in other MGD types has not been thoroughly defined [8].
To our knowledge, studies on the effects of factors like hormones [9], drugs [10], and fatty acids [11] on MGs have mainly employed immortalized human MG epithelial cell lines [12]. Most studies on MGD pathogenesis have used MG cells, which only provide information on a single cell's response to experimental manipulation and do not maintain the architecture of an intact MG acinus and the acinar epithelial cells' secretion state in vivo.
In this study, rat MG explants were cultured using a Transwell chamber-assisted method under air-liquid interface (airlift) conditions. The Transwell chamber is a penetrating chamber with a permeable microporous membrane, which was used to flatten and form an airliquid interface for the MG explants and deliver nutrients through the semi-permeable membrane. Changes in MG explants' histological structure, lipid aggregation state, MGD biomarker expression (keratins 1 [KRT1] and 14 [KRT14] and peroxisome proliferator-activated receptor-gamma [PPAR-γ]), and oxidative stress marker levels (reactive oxygen species [ROS], malondialdehyde [MDA], and 4-hydroxy-2-nonenal ) were examined to determine whether these levels were similar to those that occurred in MGD.

Animals
Male rats aged 4-6 weeks were obtained from Hunan Laysk Jingda Experimental Animal Co., Ltd. (Hunan, China) and reared in exhaust-ventilated closed-system and individual ventilation cages with sterile water and feed for three days. The design and implementation of animal-related experiments followed the guidelines of The Association for Research in Vision and Ophthalmology's Statement on the Use of Animals in Ophthalmic and Vision Research.

Diphenyl tetrazolium bromide salt assay
Explants were collected at the specified time points and weighed to detect viability as previously described [13]. Briefly, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; 884177; Sigma-Aldrich, U.S.A) was dissolved in 1× PBS to a final concentration of 5 mg/mL. Explants were incubated with 10 µl of MTT solution for 4 h under the described culture conditions. Next, the supernatant was removed and replaced with 100 µl dimethyl sulfoxide (D8371; Solarbio) for 10 min. Then, the eluate was transferred to a 96-well plate in triplicate before optical density readings were taken at 490 nm (ElX800; Biotek). were obtained by removing the skin, subcutaneous tissue, muscle, and palpebral conjunctiva from around the upper eyelid (dotted line = bisector). (c) Comparison of explant tissue viability in each culture method using the MTT assay. Key: #, P < 0.05 compared to the fresh tissue control group (0 h); ## , P < 0.01 compared to the fresh tissue control group (0 h); *, P < 0.05 with a two-sided test; 0h, fresh tissue control group.

H&E staining of explant sections
The collected explants were fixed in a 4% neutral paraformaldehyde (PFA; G1101; Jarvis, China) solution for 24 h. Next, they were dehydrated with gradients of ethanol (ET; 107017; Merck) and p-xylene (PX; 534056; Merck) with the following concentrations: 70% ET for 1 h, 80% ET for 50 min, 95% ET for 45 min, 100% ET for 75 min, a 50% ET and 50% PX mixture for 30 min, and 100% PX for 40 min. Next, these were embedded in paraffin (YA0012; Solarbio, China) at 60°C for 45 min and cut into 4 μm sections by a paraffin slicing machine (HM 340E; Thermo, U.S.A). For examination, sections were deparaffinized with gradient ET and PX concentrations as follows: 100% PX for 20 min, 100% ET for 20 min, 95% ET for 5 min, and 80% ET for 5 min. Next, the sections were stained with hematoxylin (ST2001; Saint Bio, China) for 10 min, washed under running tap water for 1 min, stained in eosin for 30 s, immersed in ethanol hydrochloride for 15 s, and washed under running tap water for 15 min. Finally, the morphology was observed under an optical microscope (Observer. A1; Zeiss, Germany).

Detecting apoptosis by TUNEL staining
Explants were collected at different time points and paraffin-embedded sections were made as described above. Then, tissue cell apoptosis was detected using a TUNEL Kit (G1502-50T; Servicebio, U.S.A). Briefly, explants were washed with 1× PBS for 5 min and then incubated with proteinase K (20 µg/mL) solution at 37°C for 20 min to hydrolyze tissue proteins. Next, these were suspended in 1× PBS, incubated with a cell permeabilizer at room temperature (RT) for 20 min to make them transparent, and incubated with the terminal deoxynucleotidyl transferase (TdT) enzyme buffer at 37°C for 2 h. Finally, their nuclei were counterstained with 0.1 µg/mL 4,6-diamino-2-phenylindole (DAPI) in 1× PBS (c0065; Solarbio, China). Images were randomly obtained with a fluorescence microscope (Axio image2; Zeiss) at 450 nm excitation frequency. Excitation fluorescence was analyzed using the ImageJ v.1.52a software (National Institutes of Health, U.S.A).

Meibum and keratin co-staining
Explants were collected as described above, fixed in a 4% PFA solution (G1101; Jarvis, China) for 1 h, and dehydrated with a sucrose series (10% for 10 min, 20% for 20 min, and 30% for 30 min) before embedding in OCT compound (4583, Sakura, Japan), frozen in liquid nitrogen, and cut into 10 μm sections. Next, the sections were socked in 1× PBS for 15 min to remove the excess embedding agent and then incubated with the KRT14 primary antibody as described above. Next, these were incubated with a goat anti-mouse secondary antibody as described above and stained with an appropriate amount of 1 µg/mL Nile Red solution at 37°C for 10 min in the dark. The Nile Red solution was prepared by mixing 1 mg Nile Red powder (N8440-100 mg; Solarbio) dissolved in 10 ml methanol solution (L13255; Thermo Fisher Scientific, U.S.A) to produce 0.1 mg/ml solution, which was diluted to 1 μg/ml with methanol. Finally, the nuclei of co-stained sections were counterstained with DAPI for 8 min at 37°C in the dark. Images were randomly obtained with the same Zeiss fluorescence microscope as described above, and their immunofluorescence was quantified using the ImageJ v.1.52a software.

Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR).
Tissue from the collected explants was cleaned, ground, and added to chloroform (151823; Sigma-Aldrich). Next, 1 mL of TRIzol reagent (15596018; Invitrogen) was added, and the samples were left undisturbed at RT for 3 min. Next, they were centrifuged at 4°C for 15 min, and their supernatant was transferred into 0.5 mL of isopropyl alcohol (67-63-0; Sigma-Aldrich) and incubated at −20°C overnight. Next, the samples were centrifuged in a low-temperature centrifuge (1-16 R; Kecheng, China) for 10 min. The precipitate was washed with 75% ET and centrifuged at a low temperature for 5 min to extract total RNA. Then, the total RNA samples were reverse transcribed into first-strand complementary DNAs (cDNAs) with a SureScript First-Strand cDNA Synthesis Kit (AORT-0020; GeneCopoeia, U.S.A) following the manufacturer's instructions. Next, qRT-PCR was performed using a BlazeTaq SYBR Green qRT-PCR Mix 2.0 Kit (Rn19071; GeneCopoeia) following the manufacturer's instructions. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the internal standardization control. The relative expressions of the target genes were calculated using the 2 −ΔΔCT method. The primers used in qRT-PCR were obtained from Changsha Qingcang Biotechnology Co., Ltd (Hunan, China). The primer sequences used for GAPDH and the target genes are listed in Table 1.

Western blotting
The tissues were collected at the specified time points, homogenized, and incubated with 150 µl of RIPA lysis buffer (P0013B; Beyotime, China) on ice for 1 h to lyse cells and obtain total proteins. Next, the supernatant was added to 5× protein buffer and boiled in a water bath for 5 min before storing overnight at −80°C. Total proteins were quantified using a bicinchoninic acid (BCA) Kit (P0010S; Beyotime) and separated on 12% sodium dodecyl-sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels at 90 V for 30 min, followed by 120 V for 60 min. Next, the separated proteins were transferred from the gel to a polyvinylidene difluoride (PVDF) membrane using a film transfer system comprising a sponge, three filter paper layers, PAGE glue, PVDF membrane, and a sponge. The transfer was performed in an appropriate amount of transfer buffer at 300 mA for 1 h in ice water. Then, the PVDF membranes were blocked in 5% milk (D8340; Solarbio) at RT for 2 h, incubated with primary antibodies against KRT1 (1:1000; Invitrogen), KRT14 (1:1000; Huabio), PPAR-γ (1:1000; Huabio), and β-actin (ACTB; 1:5000; Abcam) at RT for 2 h, and washed with Tris-buffered saline containing 0.1% Tween 20. Next, the sections were incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit (ab150077 Abcam) and HRP-conjugated goat anti-mouse (ab19195; Abcam) secondary antibodies for 1 h. The ImageJ v.1.52a software was used to normalize the corresponding protein bands.

Explant ROS assays
Tissues from explants collected at the specified time points were minced using a scalpel. Next, 5 mL of 0.25% trypsin (SM2003; Merck) was added, pipetted up and down several times with a 10 mL pipette, and incubated for 5 min at 37°C to digest the tissue before adding 5 mL of DMEM containing 10% FBS to deactivate the trypsin. Next, the samples were filtered (aperture: 0.0075 mm; FK-KN253-220; Corning) and centrifuged at 1000 rpm for 10 min to obtain a singlecell pellet. Next, the cell pellet was resuspended in 1 mL of DMEM and 1 μl of a dichlorodihydrofluorescein diacetate (DCFH-DA) antibody (S0033S; Beyotime). The cell suspension was subsequently incubated at 37°C for 20 min. Finally, ROS levels were detected using a flow cytometer (A24863; Thermo Fisher Scientific).

Expant MDA assays
Explants were collected as described above and washed twice with pre-chilled normal saline. The tissues were weighed, and a nine-times w/v of normal saline was added before these were ground to a homogenate using a tissue grinding apparatus (F-GT05; FKS, China). Next, the homogenates were centrifuged at 16000 g and 4°C for 15 min. Then, a working reagent was added according to the instructions specified in the MDA Kit (A003-1-2; NJJC, China), and the samples were heated in a water bath for 80 min before cooling Key: TNF-α, tumor necrosis factor-alpha.
to RT under running water. Next, the samples were centrifuged at 4000 rpm for 10 min, 3 ml of supernatant was transferred into a cuvette, and absorbance at 532 nm was obtained with a microplate reader (L1303006; JH). The MDA content was calculated using the formula provided in the kit's instructions as MDA (nmol/mg protein) = ([measured OD − control OD]/[standard OD − blank OD]) × (standard concentration/protein sample concentration).

Immunohistochemistry for lipid peroxidation damage
Paraffin sections were baked, dewaxed, hydrated, repaired, permeabilized, and sealed as described for immunofluorescence staining. Next, these were incubated with an anti-4-HNE monoclonal antibody (5% BSA diluted to 25 mg/mL; Genetex, U.S.A) overnight at 4°C in dark. Then, these were processed using the Universal Two-

Statistical analysis
All data are presented as mean ± standard deviation. Comparisons involving two groups were made using t-tests. Comparisons involving more than two groups were made by one-way analysis of variance with Dunnett's post hoc t-test. All data analyses were performed using the SPSS v.13.0 software (IBM, U.S.A). All results with P < 0.05 were considered statistically significant. Data illustrations were created using the GraphPad Prism 6.0 software (GraphPad, U.S.A).

Rat MG explant cultivation and viability
To examine the optimum method and appropriate time for explants, we detected tissue activity using the MTT assay. Values at 24 h were similar to those for fresh tissue (0 h) in both the Transwell chamber-assisted cultivation group and the non-Transwell cultivation group (P>0.05). However, values at 48, 72, and 96 h were all lower than those of fresh tissues in both groups (P < 0.05; Figure 1c). Values differed significantly between the Transwell chamber-assisted cultivation and non-Transwell cultivation groups at 24, 48, 72, and 96 h (P < 0.05; Figure 1c). These results showed that explant viability was greater in the Transwell chamber-assisted cultivation group than in the non-Transwell cultivation group (Figure 1c).

H&E and TUNEL assays
The pathological sections were assessed by three senior technicians under the double-blind standard. First, the whole picture and the resulting physical signs were observed in an orderly and comprehensive manner under low power (4× objective lens), and then the characteristics of cells (the integrity of ducts and glands and infiltration of inflammatory cells) were selectively observed under a high-power microscope. Explants cultured by both methods and fresh tissues had comparable morphology (Figure 2a). Complete structure of the MGs' morphology was seen with both culture methods within 72 h. However, greater inflammatory cell infiltration around the acinar cells was observed with the extension of the incubation time, and the central duct's epithelial layer had become thinner in both culture methods. At 48 h, pathological changes in acinar cells included chromatic agglutination, karyopyknosis in nuclei, and nuclear fragmentation while the space between the acinar gradually increased with acinar atrophies. At 96 h, the acinar cell arrangement was disordered; disintegration was missing, and the acinar space was further enlarged. Hyperplasia of connective tissue around the acinar and some cells in the glandular duct wall appeared as vacuoles. Nuclei fragmentation in the acinar cells disappeared by both culture methods. The structure of the cellular organelle alterations could be subtly observed in TEM (described later under the results of TEM).
The TUNEL assay showed that fluorescence intensity was significantly higher in the Transwell chamberassisted cultivation and non-Transwell cultivation groups than in the control group (0 h) at all time points between 24 h and 72 h (P < 0.05; Figure 2b,c). Intensities in the Transwell chamber-assisted cultivation group were comparable to those in the non-Transwell cultivation group at 72 h (P > 0.05; Figure 2b,c). Therefore, the H&E and TUNEL results indicated that MG explants in the Transwell chamber-assisted cultivation group maintained better structure and tissue activity than those in the non-Transwell cultivation group at time points between 24 and 48 h. Consequently, Transwell chamber-assisted cultivation was the better method for MG explant cultivation, and MG explants should be cultivated for 48 h.

Biomarker expression in rat MG explants
Changes in keratinization, lipid differentiation, and Nile Red biomarker in MG explants were detected by immunofluorescence according to previously described MGD studies [14][15][16] (Figure 3a-c). KRT1 and KRT14 expressions were significantly higher in MG explants than in fresh tissues (P < 0.05; Figure 3d,e). However, PPAR-γ expression was significantly lowered in MG explants than in fresh tissue (0 h) at 24 and 48 h (P < 0.05; Figure 3f).
The lipid distribution in MG explants in the Transwell chamber-assisted cultivation group was assessed at each time point by co-staining for Nile Red and KRT14. Nile Red staining showed that lipids in acinar cells were comparable to fresh tissues (0 h) at 24 and 48 h (P > 0.05; Figure 3g). According to the results for immunofluorescence-based KRT1 and KRT14 expression, MG explants cultured by the Transwell chamber-assisted cultivation method showed keratin overexpression. However, these showed decreased lipid differentiation and no differences in lipid production in acinar cells (Figure 3g).
Based on the H&E staining, TUNEL apoptosis, and biomarker expression patterns in rat MG explants cultured with each method, Transwell chamber-assisted cultivation appeared to be better. Therefore, qRT-PCR was used to detect the mRNA levels of KRT1, KRT14, PPAR-γ, and the inflammatory factor, TNF-α in the Transwell chamber-assisted cultivation group. mRNA levels of KRT1 and KRT14 were significantly higher in the Transwell chamberassisted cultivation group than in the fresh tissue (0 h) group at 24 and 48 h (P < 0.05; Figure 4a) but did not differ significantly (P > 0.05). The mRNA levels of PPAR-γ and TNF-α were significantly higher in the Transwell chamber-assisted cultivation group at 48 h (P < 0.05, Figure 4a) but not at 24 h (P > 0.05) compared to the fresh tissue (0 h) group.  Similarly, western blotting was performed to examine the protein levels of KRT1, KRT14, and PPAR-γ. KRT14 levels were significantly higher in the Transwell chamber-assisted cultivation group than in the fresh tissue (0 h) control group at 24 h (P = 0.0021) and 48 h (P < 0.0001) and differed significantly between 24 and 48 h (P < 0.05; Figure 4b). KRT1 levels were significantly higher in the Transwell chamber-assisted cultivation group than in the fresh tissue (0 h) control group at 24 h (P = 0.0132) and 48 h (P = 0.0383) but did not differ significantly between 24 and 48 h (P > 0.05).
The changes in protein and mRNA levels were consistent during keratinization. The keratin index showed an increasing trend but lipid differentiation marker expression showed a decreasing trend. PPAR-γ protein levels were significantly lower in the Transwell chamber-assisted cultivation group than in the fresh tissue (0 h) group at 24 and 48 h and differed significantly (P < 0.05; Figure 4b). In contrast, mRNA levels of PPAR-γ increased significantly from 0 to 48 h (P = 0.0484) but not from 0 to 24 h (P > 0.05).

Ultrastructural changes in acinar cells of MG explants
Acinar cells deep in MG explants were examined using a transmission electron microscope at 24 and 48 h and compared to fresh tissues (0 h). At 0 and 24 h, tight junctions were found between acinar cells but rarely observed at 48 h (Figure 5a,b, b2), and the numbers of organelles and mitochondria in acinar cells also decreased (Figure 5e). The cell membrane structure was unclear at 48 h, nuclei were irregularly shaped, and the nuclear membrane was partially broken. The chromatin appeared degraded, few mitochondria were left, the cristae disappeared, and the amorphous matrix had dissolved (Figure 5c, c3). Abundant secretory granules-lipid droplets could be found in acinar cells at all time points. In fresh tissue (0 h), the lipid droplets were small, and the capsule was complete, appearing round or oval. The area of secretory granules increased from 24 to 48 h (Figure 5e), and lipid vesicles broke into sheets and diffused in cell cytoplasm at 48 h, indicating abnormal acinar cell differentiation (Figure 5c, c3).

Oxidative stress in the explant model
To examine oxidative stress levels in rat MG explants, we detected ROS levels to assess overall oxidative stress, MDA to assess lipid oxidative stress, and 4-HNE to visualize oxidative stress sites. The MG explants cultured for 24 and 48 h differed significantly (P < 0.05; Figure 5a). The 4-HNE levels were significantly lower in fresh tissue (0 h) than in explants cultured for 24 and 48 h (P < 0.05; Figure 6b). Lipid peroxidation (MDA) was quantitatively analyzed in MG explants at different time points, and MDA levels were the highest at 24 h followed by 48 h (P < 0.05; Figure 6c).

Discussion
MGD comprises a group of disorders linked by functional abnormalities in MGs that can cause dry eye and other ocular surface diseases [10]. Animal experiments show that factors such as dry environment, bacterial infections, inflammation, and tears with high permeability can directly affect MG's function [5,13,14]. An immortalized MG epithelial cell line has been established but cannot be used to observe MG gland duct structure, lipid secretion, and changes in gland duct epithelial cells [13,14]. Various organs can be cultured in vitro for days [15][16][17] but a stable MG culture method in vitro is lacking.
Tissue culturing in airlifting conditions, such as cornea epithelial cells and conjunctiva, have been used as models to dissect pathophysiological changes occurring in dry eyes [18,19]. In this study, a Transwell chamberassisted method under airlift was developed for rat MG explant culture in vitro. This model imitates the body's physiological state wherein the meibomian's conjunctival surface contacts tears at the air interface. Results of the MTT assay, H&E staining, and TUNEL apoptosis [20,21] showed that this model maintained tissue viability and morphology better than those in submerged conditions [22] and more stable biomarker expression within 48 h in vitro was seen. Moreover, MG explants' histopathological morphology and acinar cells' microstructural changes observed in this research were similar to MGD pathological processes in animal studies [23]. This observation indicates that abnormal compensatory basal cell differentiation, increased meibum keratin content, and duct wall hyperkeratosis, ultimately led to obstructive MGD occurrence [5,13,14].
The normal MG central and secretory ducts comprise 4-6 layers of the nonkeratinized stratified squamous epithelium [24]. During keratinization, the ductal epithelium's nucleus changes from pre-keratinocytes to incomplete keratinocytes before disappearing finally. In this study, the MG explants' structure under both airlift and submerged culture conditions indicated changes in keratinization with extended culture times in H&E-stained samples.
The common keratins in MG ducts are KRT1, KRT5, KRT6, KRT10, KRT14, KRT16, and KRT17 [25]. Different keratin types are expressed in different glandular duct regions. Their overexpression can lead to glandular duct obstruction and changes in the meibum, leading to MGD [6]. Among them, KRT1 and KRT14 have been specifically targeted previously in MGD studies [23,25,26]. KRT14 is expressed in all stages of MG epithelial development which is mainly expressed in the MG acini and the central duct's basal layer as a main protein of basal keratinocytes and plays an important role in maintaining physical stability [27,28]. KRT1, which normally constitutes a keratinizing marker together with KRT10, is expressed primarily in the fully keratinized epithelium and is a typical ductal epithelium component [28,29]. In this study, KRT14 expression increased gradually from the basal to the epithelial layer in MG explants cultured in vitro, consistent with previous studies on MGD patients [25,26]. KRT1 expression increased, as evidenced by immunofluorescence assay while qRT-PCR and western blot results suggested that it first increased and then decreased. Previous studies have shown that the MG's normal central and connecting ducts contain hyaline keratin particles. Histology shows that the secreted meibum contained keratin components from the MG's epithelium [23]. KRT14 and Nile red co-staining in this study shows that lipids were uniformly distributed in the acinar cells' cytoplasm. However, acinar cells' lipid deposition did not change significantly at different time points. Human conjunctival culture exposure to air can induce squamous metaplasia, which is accompanied by the upregulation of K10, K14, and p63 expression and the conjunctival epithelial cells lose their normal phenotype and acquire an epidermal epithelial phenotype. Air exposure may also induce keratinization changes in MG explants [18,19]. During the culture, the MG explants showed compensatory proliferation of MG basal cells and excessive keratin expression in MG epithelial cells, potentially leading to changes in meibum compositional fluidity and obstructive MGD. Therefore, KRT14 may play a crucial role in MGD occurrence.
PPAR-γ has been constantly used as evidence of compensatory basal cell proliferation and differentiation. PPAR-γ is a lipid-activated nuclear hormone receptor that regulates lipid synthesis, plays a crucial role in adipose and sebaceous gland development [30], and is mainly responsible for the differentiation and maturation of lipid-forming cells [31]. In this study, mRNA and protein levels of PPAR-γ in MGs were inconsistent, possibly because the synthesis agonists of PPAR-γ are generally long-chain polyunsaturated fatty acids and eicosane derivatives [32]. The compensatory expression of PPAR-γ mRNA in cells increased but lacked synthetic raw materials and agonists to cause its protein levels to increase similarly with culture time. This finding further demonstrates the existence of compensatory differentiation into lipid-forming cells within acinar. Previous studies have shown increased concentrations of pro-inflammatory factors, like interleukin (IL)-1β and IL-6, in the tears and conjunctiva of MGD patients compared to healthy individuals [33][34][35]. Leukocyte infiltration in human MG acinar is related to MGD grade and severity [36]. However, in this study, IL-1β and IL-6 were negligibly expressed in rat MG explants cultured in vitro. TNF-α has a crucial role in inflammation and cell death [37][38][39] but its expression was negligible in fresh MG tissues from rats. Levels of TNF-α expression started to increase gradually when MG explants were cultured in vitro for 24 h and reached their peak at 48 h. However, the synthesis of TNF-α mRNA was low. These results indicate that MGD occurrence and development cannot be separated from the inflammatory factors that mainly are from the lacrimal gland and spread over the ocular surface with the help of the tear fluid. The two promote each other to form a vicious circle [40].
The ocular surface is in contact with the outside world, and the influence of oxidative stress is prominent, like in dry eye disease [41]. Studies have found that meibum undergoes a certain degree of lipid oxidation before it is secreted onto the ocular surface [8,21] ROS are metabolic by-products of oxygen consumption in tissues, organs, and cells. Mitochondria are the primary sites of oxidative damage. Mitochondrial damage can cause further damage to other cell organelles (e.g. the endoplasmic reticulum), ultimately leading to cell function damage and death [42]. Cornea epithelial cells cultured in airlifting conditions cause robust oxidative stress [43]. A similar phenomenon occurs in MG. ROS levels increase gradually over MG explant culture time in vitro. The acini cells' microstructure under a transmission electron microscope showed that the mitochondria were the first to be affected, as evidenced by a decrease in their numbers and collapsing structures with an overflowing matrix.
MDA and 4-HNE are stable decomposition end products of lipid peroxidation [44][45][46]. Accumulating oxidative damage and its effect on downstream enzyme reactions leads to metabolic dysfunction and apoptosis [47]. However, changes in 4-HNE and MDA levels were the highest at 24 h and decreased at 48 h. Combined with the above results, we speculated that changes in 4-HNE and MDA levels are related to decreased MG activity in vitro.
This model can be an important supplement to MGD research. However, this study had some limitations that should be addressed in the future. Since MG's functions in the body are affected by various factors, including hormones, sex, growth factors, and oculi muscle contraction [48][49][50][51], their culture conditions need to be optimized to extend survival time and improve physiological status.

Conclusions
In this study, we developed a Transwell chamberassisted method under airlift conditions for culturing rat MG explants in vitro, which showed better tissue viability and morphology than the previously used methods [22]. Moreover, this culture method for MG explants could act as an MGD model in vitro since a similar pathological process as MGD occurred during culturing. Elevated oxidative stress levels during culturing may aggravate the MGD pathological process. This in vitro MG explant model overcomes the cell models' inability to show the ducts of MG epithelial cells.

Authors' contributions
WJZ and SXH were responsible for study design, data acquisition, and analysis and were major contributors to writing the manuscript. HQK and ZYLB helped to perform the collection of specimens and experimental verification. HL and LK were responsible for the integrity of the entire study and manuscript review. All authors contributed to the article and approved the submitted version.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Data availability statement
The datasets uesd and/or analysed in the current study are available from the corresponding author upon reasonable request.

Ethical approval
The experiments involving animals were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the national Institutes of Health and were approved by the Laboratory Animal Center of Yunnan University (Kunming, China,YNUCARE20210203).