BMSC reduces ROS and inflammation levels by inhibiting TLR4/MYD88/NF-κB signaling axis to alleviate dry eye

Objective To investigate the therapeutic effect of Bone marrow mesenchymal stem cells (BMSCs) on dry eye mice, and to investigate the mechanism of TLR4/MYD88/NF-κB signaling pathway on corneal injury repair in dry eye mice. Methods To establish a hypertonic dry eye cell model. Western blot for measureing the protein expressions of caspase-1, IL-1β NLRP3 and ASC and Rt-qpcr for mRNA expression. Flow cytometry for detecting the ROS content and apoptosis rate. CCK-8 for detecting the proliferation activity of cells, and ELISA for the levels of inflammation-related factors. The levels of inflammation-related factors were detected by ELISA. The dry eye mouse model of benzalkonium chloride was established. Three clinical parameters used to evaluate ocular surface damage, namely tear secretion, tear film rupture time and corneal sodium fluorescein staining, were measured with phenol cotton thread. Flow cytometry and TUNEL staining are both for he apoptosis rate. Western blot also for detecting the protein expressions of TLR4, MYD88, NF-κB, inflammation-related factors and apoptosis-related factors. The pathological changes were evaluated by HE and PAS staining. Results In vitro, BMSCs and inhibitors of TLR4, MYD88 and NF-κB showed decreased ROS content, decreased inflammatory factor protein level, decreased apoptotic protein level and increased mRNA expression compared with NaCl group. BMSCS partially reversed cell apoptosis induced by NaCl and improved cell proliferation. In vivo, it reduces corneal epithelial defects, goblet cell loss and inflammatory cytokine production, and increases tear production. In vitro, BMSC and inhibitors of TLR4, MYD88 and NF-κB could protect mice from apoptosis induced by hypertonic stress. In terms of mechanism, NACL-induced NLRP3 inflammasome formation, caspase-1 activation and IL-1β maturation can be inhibited. Conclusion BMSCs treatment can reduce ROS and inflammation levels and alleviate dry eye by inhibiting TLR4/MYD88/NF-κBsignaling pathway.


Introduction
Dry eye disease (DED) is recognized as a multifactorial chronic in ammatory disease, it has a serious impact on patients' visual function and quality of life. The disease can cause various symptoms and visual impairment, and may be accompanied by ocular surface damage, characterized by unstable tear lm [1] . At present, arti cial tears and anti-in ammatory drugs are mainly used as the treatment strategies for dry eye [2] , but anti-in ammatory drugs cannot distinguish pathological and protective immune responses and have many toxic side effects [3,4] . Therefore, it is of great signi cance to study the mechanism of dry eye in clinical treatment.
In recent years, with increasing research, there is a growing tendency to consider in ammation as the central driver of dry eye pathogenesis, the so-called in ammatory "vicious cycle" [5] . A growing body of research also suggests that changes in tear composition can lead to epithelial damage and neurological irritation, further destabilizing the tear lm, amplifying the in ammatory response and creating a vicious cycle [3] . Oxidative stress is another factor that in uences the development of dry eye and plays an important role in ocular diseases, including ocular surface lesions [6] . Studies have shown that continuous exposure with oxidative stress are associated with the initiation and progression of cell damage, leading to dry eye, UV-induced ocular surface epithelial damage [7] . Therefore, reducing ROS and in ammation levels is essential to alleviate dry eye.
In recent years, signi cant breakthroughs have been made in the use of mesenchymal stem cells (MSC) as a pluripotent stem cell for the treatment of autoimmune diseases, such as severe corneal epithelial injury, corneal transplantation and uveitis [8][9][10] . Among them, the earliest discovered and most clinically used are bone marrow-derived MSCs, which are called bone marrow mesenchymal stem cells (BMSC).
Previous Studies have shown that local application of BMSC in dry eye animal models using different methods (local eye drop/periorbital injection/bulbar subconjunctival injection) can protect corneal epithelial microvilli structure, reduce corneal epithelial injury, and increase the number of conjunctival goblet cells [11][12][13][14] . The expression levels of leukocyte differentiation antigen 4 (CD4), interleukin (IL-6, IL-1), tumor necrosis factor (TNF-α) and other in ammatory factors were decreased, and the secretion function of some lacrimal gland tissues was restored [14,15] . Although the treatment e cacy of BMSC in dry eye has been initially con rmed, the molecular mechanism by which it exerts a repairing effect on ocular surface tissue damage during the treatment remains unclear.
TLR4/MYD88/NF-κB signaling pathway is closely related to the pathogenesis of DED. Hyperosmosis, lack of tear lm or increased evaporation will produce proin ammatory cytokines and MMP on the eye surface, activate immune cells, cause in ammatory cycle, and damage the functional units of the lacrimal gland, which has been con rmed by studies [16] . TLR signaling induces the expression of proin ammatory mediators in human corneal cells mainly through NF-κB transcriptional activity in response to ligand binding. Moreover, TLR activation can also cause the chemokines production and proin ammatory cytokines are produced and secreted [17] .
Baseline levels of MMPs and proin ammatory cytokines are reduced when MyD88 de ciency occurs [18] .
It was found that the levels of CXCL1, TNFα, IL-2, IL-9, IL-1βand IL-1αwere decreased in the ocular surface tissues which MyD88-de cient mice during EDE. This suggested that TLR signaling inhibits in ammatory molecules during dry eye when MYD88-dependent is lacking and points to potential targets for DED antiin ammatory therapy [19,20] .
Combined with the already clear BMSC, it can effectively alleviate ocular surface tissue damage and dry eye in ammation, and the proven role of BMSC in effectively downregulating ROS in other elds, and the preliminary study that ROS may cause the production of IL-1β and subsequent in ammatory response through activation of NLRP3 in ammatory vesicles that eventually lead to the development of dry eyes, together with the related studies that TLR4 activation leads to cellular secretion of ROS, NLRP3 We hypothesize that the therapeutic effects of BMSC on dry eye may be related to its immunomodulatory, antioxidant effects and anti-in ammatory, and may be related to its paracrine signaling and potential effects on corneal immune regulation. -KB signaling pathway and thus reduce ROS and in ammation levels for the relief and treatment of dry eye corneal injury. Therefore, this study will investigate the effects of BMSC on dry eye by reducing ROS and in ammation levels, and to investigate whether its mechanism of action is related to the TLR4/MYD88/NF-κB signaling pathway.

MCECs cell culture
Mouse corneal epithelial cells (MCECs) were purchased from Wuhan Saios Biotechnology Co., LTD. The lyophilization tube of MCECs was removed and opened after the tube mouth was sterilized with alcohol.
The cells were centrifuged for 3 min at a speed of 1000 RPM /min after the cells were transferred to the centrifuge tube. The supernatant was removed and the cell culture medium was added and blown repeatedly with a pipette gun to fully mix the cells and culture medium. The cells were inoculated into DMEM/F12 complete medium containing 1% dual antibody and 10% FBS, and then cultured in a cell incubator with 5% CO 2 at 37℃. Changing the culture mediumonce once two days.

Cells treated with different osmotic pressure solutions
MCECs cells which during the logarithmic growth phase were seeded into serum-free medium with corresponding osmotic pressure at300, 340, 380, 420, 460 and 500mOsM, and incubated with 5% CO 2 for another 24 h at 37℃.

Protein imprinting analysis
Added 50µl of RIPA Lysis and Extraction Buffer to the cultured cells and ground tissues, placed on ice for 20min, centrifuged at 10,000rpm for 10min, transfered the supernatant to a pre-cooled EP tube, determined the protein content by BCA, boiled water bath for 10 min after added the same volume of 2×SDS loading buffer, and conducted subsequent experiments or stored at -80℃.
Different concentrations of drugs were added when the cells proliferated to 60% of the culture plate area, and incubated at 37°C for 24 h. The drug solution and culture medium were then discarded, and extracted total proteins from the cells which for subsequent protein blotting experiments by the RIPA lysate. Cellular proteins were transferred to polyvinylidene di urane (PDVF) membranes after separated by 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), and blocked with 5% (w/V) skim milk for 1 h. PDVF membranes were incubated overnight with primary antibodies which include IL-1β, IL-10, TNF-α, ASC, Caspase-1, NLRP3, MYD88, NF-κB and TLR4 at 4°C.And then, the membranes were then incubated at room temperature with secondary antibody which labeled by horseradish peroxidase for 1h.
ECL color development, gel system imaging analysis, and Image J analysis of grayscale values were performed.

Real-time uorescence quantitative PCR
Extracting RNA by Trizol reagent and RNeasy® Plus Micro Kit, respectively. Reverse transcription to cDNA was performed according to the instructions of the cDNA reverse transcription kit manufacturer. Real-time PCR performed by SYBR Green real-time PCR kit with U6 as internal reference. Repeat three times, and the results were calculated by the 2 − ΔΔCt method. Detailed primer information is shown in Table 1.

Flow cytometry detection of ROS content
After trypsin digestion, the cell density of each group of cells was adjusted to 2×10 5 cells with cell culture medium, then seeded into 6-well plates and cultured for 24 h in a cell incubator containing 5% CO 2 . The medium was discarded after centrifugation, and the cells were resuspended by adding D-Hanks solution.
Add 500 µL of D-Hanks solution containing DCFH-DA (20 µL) to each sample and incubate for 30 min in a cell culture incubator protected from light, shaking every 5 min to allow the cells to fully bind to the probe. Cells were resuspended by adding D-Hanks solution again, and detected the level of ROS in the cells by ow cytometry.
2.6 CCK-8 assay for cell proliferation viability Cells (5×10 3 cells/well) were inoculated in 96-well plates and incubated for 24 h at 37℃in a 5% CO 2 incubator. Adding 10 µL CCK-8 to each well after the dosing process of each group. After 30 min, measured the 450 nm absorbance value of each well by microplate reader.

Detection of apoptosis by ow cytometry
Firstly, the trypsin-digested cells in each group were incubated for 24 h (1×10 5 cells/dish) in 60 mm dishes, Annexin-V-FITC/PI kit was for detecting the apoptosis rate when the two washes with pre-cooled PBS were complete.

ELISA for in ammatory factors
Supernatants of the cell cultures were collected and cryostored at -80℃ to allow quanti cation of cytokines using a speci c ELISA kit.

Dry eye mouse model construction
Ninety mice were randomly selected to construct a dry eye model, and 5 µL of 0.2% benzalkonium chloride solution was dropped to each eye once in the morning and once in the evening. All lasted for 7 days, and clinical assessment of dry eyes in mice was performed on the 7th day. The dry eye mouse model was successfully constructed when the tear secretion was ≤ 10 mm/5 min and the tear lm rupture time was ≤ 5 s. Treatment was started at the end of the evaluation on day 7.
2.10 Tear secretion of each group of mice was measured by phenol red cotton thread After 7 days of treatment, a su ciently long phenol red cotton thread was cut, the mouse eyelid was propped open with one hand, and putting the phenol red cotton thread on the conjunctival sac which at the junction of the middle third of the outer of lower eyelid. The time was timed with a stopwatch for 30 s. The phenol Khmer thread was removed to measure the red part of the cotton thread, and the measurement unit was mm. The measurements were repeated twice for each eye of the mice, and the results were averaged and recorded.

Tear lm rupture time detection
The conjunctival sacs of both eyes of each group of mice were rinsed with 0.9% saline for 1 ~ 3 times, and cotton swabs were used to wipe away excess water around the corners of the mice's eyes. A pipette gun was used to drop 1 µL sodium uorescein (100 g/L) into the conjunctival sac of mice, and the mice were arti cially assisted to transient eyes for 3 times after 10 s. Subsequently, separated the upper and lower eyelids of the mice by hand, and the time to the rst rupture in the tear lm was observed under the cobalt blue light of the slit lamp. The time point was recorded as the time of tear lm rupture.

Sodium corneal uorescein staining test
The mice in each group were safely xed, 1.5 µL 0.5% uorescein sodium dye was slowly dripped into the inferior conjunctival sac of each group with a pipetting gun, and the eyes of the mice were closed. 2 min later, the corneas of the mice were observed under the slit lamp for the appearance of punctate or lamellar or fused staining, and the corneal epithelium was rated for sodium uorescein staining. a 12point method was used to score uorescein staining : cornea was divided into four quadrants, and each quadrant was divided into four grades which depending on the degree of staining and the area.The score of no staining was 0, the score of 1 to 30 punctate staining was 1, the score of > 30 punctate staining was 2, and the score of lamentous, lamellar or block staining fusion was 3 [21] . Each mouse was scored three times, and the mean of the three scores was taken. The score was 3. Each mouse was scored as the mean of the three scores performed.

TUNEL detection of apoptosis
Mice were sacri ced with cervical dislocation and eyeballs collected. First x the specimen with paraformaldehyde solution for 24 hours. Dehydrated and transparent, and the sections were embedded for 5 µm. After dewaxed hydration, antigen repair solution was repaired. The samples were incubated at 37°C after addition of TUNEL mixture, and then rinsed with PBS. After adding 50 µL alkaline phosphatase antibody, the samples were placed in an incubator at 37°C for 30min and then rinsed with PBS for 3min×3 times. mounted with water-based sealant, and then dried at 60°C before viewing the slice under a light microscope.

Pathology studies
The tissue samples were rst routinely embedded in para n. HE staining: staining sections with hematoxylin for 5min after dewaxing, rinsed with tap water to return to blue, stained in eosin dye solution for 1min, dehydrated in ethanol, and sealed with neutral gum for observation and analysis. PAS staining:The collected eyeballs were embedded in para n and xed with 4% paraformaldehyde (PFA), then Sections with a thickness of 5 µm were prepared, and Staining was performed according to the manufacturer's PAS staining kit instructions. Finally, the upper and lower conjunctival membranes were observed and photographed by digital light microscope.

Statistical analysis
GraphPad Prism7.0 was used to analyze the experimental data and plot the pictures, The analysis results are presented in the form of "mean + standard deviation". Among them, T-test was for comparison of two groups, one-way analysis of variance was for multiple groups, respectively. p < 0.05 indicates statistical signi cance.

Response of mouse corneal epithelial cells (MCECs) in different concentrations of hypertonic solutions
The expression of ASC, IL-1β, caspase-1 and NLRP3 in MCECs cells which cultured at 300, 340, 380, 420, 460 and 50 osmotic pressures for 24 hours was measured byWestern blot. The results showed that, the expression levels of ASC, IL-1β, caspase-1 and NLRP3 in MCECs cells were signi cantly increased than 300 mOsM group with the increase of osmolality. The expression levels of ASC, IL-1β, caspase-1 and NLRP3 at 500 mOsM were higher than those at other osmotic pressures (Fig. 1A). The results of the RT-qPCR assay were consistent with Western blot assay (Fig. 1B). The results of ROS cotent showed that the ROS content increased signi cantly with the increase of osmotic pressure, and the content of ROS in cells at 500 mOsM was signi cantly higher than other osmotic pressures (Fig. 1C). Detected the proliferation activity of cells in each group by CCK-8. it was found that the proliferation activity of MCECs in the hypertonic environment was signi cantly lower than that in the 0 mOsM group (Fig. 1D). The results of apoptosis showed that the apoptosis level of MCECs signi cantly increased with the increase of osmotic pressure (Fig. 1E). Therefore, 500 mOsM was chosen as the osmotic pressure for the mouse dry eye model cell construct.

BMSC Affects ROS, NLRP3, and IL-1β In ammatory Signaling to Mitigate Dry Eye Cell Damage
Mouse BMSCs were co-cultured with MCECs treated by 500 mOsM NaCl, and detected the ROS content.
The results showed that the ROS content was signi cantly lower in NaCl + BMSCS co-culture group than NaCl group (Fig. 2A). The result of expressions which apoptosis-related and in ammatory cytokines proteins and mRNA in MCECs turns out that, the expression of Bcl-2 was signi cantly increased ,and the expression of ASC, IL-1β, caspase-1, NLRP3and Bax proteins in NaCl + BMSC-treated group were decreased signi cantly compared with NaCl group (Fig. 2B-2C). The detection result of ELISA showed that the levels of 1L-1β and TNF-αwere signi cantly lower in NaCl + BMSC-treated group than those in NaCl group (Fig. 2D). CCK-8 and ow cytometry results showed that NaCl + BMSCS could promote the proliferation viability of MCECs and inhibit the level of cell apoptosis compared with NaCl group (Fig. 2E-F). The detection results of NF-κB, TLR4, and MYD88 protein expression showed that, the expressions of NF-κB, TLR4, and MYD88 in NaCl + BMSC group and NaCl + ST2825 group were signi cantly decreased compared with NaCl group (Fig. 2G). In summary, BMSC co-culture signi cantly inhibited apoptosis and in ammatory responses and promoted cell proliferation in MCECs of 500 mOsM NaCl-induced mouse dry eye model.

BMSC reduces ROS and in ammation levels by inhibiting TLR4/MYD88/NF-κB signaling axis to slow down dry eye
The results of Western blot showed that, the expressions of NF-κB, TLR4, and MYD88 were signi cantly lower in NaCl + BMSCS group compared with NaCl group. Compared with NaCl + BMSCS group, the expressions of NF-κB, TLR4, and MYD88 were decreased after TAK-242 addition; after addition of ST2825, only the expression of MYD88 was decreased; after the addition of IMD-0345, only the expression of NF-κB decreased (Fig. 3A). The results of ROS detection in the cells showed that the ROS content in cells of NaCl + BMSC group and NaCl + TAK-242 group was signi cantly decreased than the NaCl group. And there was a further decresion about the ROS content in NaCl + BMSC + TAK-242 group than NaCl + BMSC group (Fig. 3B). The result about protein expressions of ASC, IL-1β, caspase-1, NLRP3 and Bax, Bcl-2 showed that, the expressions of ASC, IL-1β, caspase-1, NLRP3 and Bax were signi cantly lower in NaCl + BMSC group and NaCl + TAK-242 group than NaCl group, while the expression of Bcl-2 was higher than NaCl group. The expressions of NLRP3, ASC, caspase-1, IL-1βand Bax in NaCl + BMSCS + TAK-242 group were further decreased than NaCl + BMSCS group, while the expression of Bcl-2 was opposite (Fig. 3C). The detection results of ELISA showed that the contents of IL-1β and TNF-αwere signi cantly decreased in NaCl + BMSC group and NaCl + TAK-242 group than NaCl group, and signi cantly lower in NaCl + BMSCS + TAK-242 group than NaCl + BMSCS group (Fig. 3D).

Pathological changes in the dry eye mouse model
In the dry eye model mice, mice were treated with BMSCS or RA, The determination results of the tear secretion of mice which by phenol red cotton thread showed that, the tear secretion of mice was signi cantly reduced in the dry eye group compared with the NC group, while the tear secretion of mice in the BMSCS treatment group was increased compared with the dry eye group (  Fig. 4B). The corneal uorescein sodium staining experiment showed that the staining score in DED group was 9.58 ± 1.03, it was signi cantly higher than NC group (0.92 ± 0.77). The corneal epithelium of mice in NC group was intact and transparent, and The corneas of DED mice showed punctate uorescein sodium staining. The staining score of mice in BMSC-treated group was 3.51 ± 0.38, which was signi cantly lower than DED group (Fig. 4C). In conclusion, BMSC treatment can alleviate corneal injury in dry eye mice. Detection results of ROS and in ammatory factors IL-1β and TNF-α in corneal tissue of mice showed that the contents of ROS, IL-1β and TNF-α were signi cantly higher in dry eye mice than NC group. BMSCS treatment inhibited the production of ROS, IL-1β and TNF-α (Fig. 5A-B). The results which Levels of in ammasome and apoptosis-related proteins in corneal tissues of mice showed that the expressions of ASC, IL-1β, caspase-1, NLRP3 and Bax were signi cantly higher in dry eye group than NC group, while Bcl-2 was signi cantly lower. The expression levels of ASC, IL-1β, caspase-1, NLRP3 and Bax in BMSC group were lower than dry eye group, while the results of Bcl-2 were opposite (Fig. 5C-D). The TUNEL staining assay of apoptosis level showed that the apoptosis level was highern in dry eye group than NC group, and lower in BMSC group than dry eye group (Fig. 5E). The results which protein expressions of NF-κB, MYD88 and TLR4 in the dry eye mouse model showed that the protein expressions of NF-κB, MYD88 and TLR4 were signi cantly up-regulated in the dry eye animal model, and consistently down-regulated after BMSCS treatment (Fig. 5F). HE staining and PAS staining showed that after BMSC treatment, the corneal epithelium of mice was atter, the epithelial cells were more regularly arranged, the keratinization of the epithelial layer was reduced, the collagen bers in the corneal stroma were slightly aligned, swelling was reduced, and the number of conjunctival cup cells was increased compared with the dry eye group (Fig. 5G-5H).
3.6 BMSC reduces ROS and in ammation levels by inhibiting TLR4/MYD88/NF-κB signaling axis to slow down dry eye The result of Western blot showed that the expressions of NF-κB, MYD88, TLR4, ASC, IL-1β, caspase-1, NLRP3 and Bax were signi cantly lower in NaCl + BMSC group and NaCl + TAK-242 group than NaCl group, while the expression of Bcl-2 was signi cantly higher than NaCl group. The expressions of NF-κB, MYD88, TLR4, ASC, IL-1β, caspase-1, NLRP3 and Bax in the NaCl + BMSCS + TAK-242 group were further decreased than the NaCl + BMSCS group, while the expression of Bcl-2 was contrary ( Fig. 6A-6C). Using ELISA to detect in ammatory factors and ow cytometry to detect ROS, the levels of ROS, TNF-α, IL-1β were signi cantly lower in cells of NaCl + BMSCS group and NaCl + TAK-242 group than NaCl group. The levels of ROS, TNF-α, IL-1β were signi cantly lower in NaCl + BMSCS + TAK-242 group than NaCl + BMSCS group (Fig. 6D-E). The results of HE staining showed that NF-κB, MYD88, TLR4 inhibitor group and BMSCtreated group had similar corneal HE staining. After BMSC was superimposed with TLR4 inhibitor, the corneal pathological condition was similar to that of the control group, but there were still more epithelial cell layers and there was slight keratinization of the epithelial layer (Fig. 6F). PAS staining results showed that the conten of conjunctival goblet cells in NaCl group was signi cantly reduced and increased after BMSCS treatment. And higher in NaCl group than dry eye group after treatment with NF-κB, MYD88, TLR4 inhibitors. The content of conjunctival goblet cells was increased signi cantly in BMSC plus TLR4 inhibitor compared to in MSC alone treatment group (Fig. 6G).

Discussion
Dry eye syndrome (DED) affects 5-35% of the global populationis. It is congsidered to be one of the most common eye diseases, and its prevalence will increase with age [22] . Its main pathogenesis is the instability, hyperosmolarity, in ammation and damage of the tear lm, which can lead to severe epithelial damage, corneal neovascularization and even ulcer formation, resulting in vision loss [23] . Therefore, it is very important to study the speci c molecular mechanism of dry eye for the development of new dry eye treatments.
Chronic in ammation of the ocular surface is closely related to dry eye [24] . It has been documented that dry eye-related in ammation occurs in tissues such as the cornea, conjunctiva, meibomian, and lacrimal glands [25][26][27] , resulting in loss of conjunctival goblet cells, loss of limbal stem cells, and tear de ciency. In this study, we observed that the expression of IL-1β and TNF-α corneal epithelial cells of dry eyes was signi cantly increased, which was consistent with previous ndings. ROS, which are potentially damaging to tissues, are by-products of oxidative metabolic processes [28] . In recent years, the critical role of oxidative stress in the pathogenesis of dry eye is supported by more and more evidence. Liu [29] et al. found that ROS production increased signi cantly in hypertonic medium. A water-soluble CENPembedded contact lens (CENP-CL) was developed by Cui et al. for the prevention of ocular surface diseases, which can remove ROS [30] . The protective effect of wearing CENP-CLS on ocular surface has also been shown in mouse models.The signi cant increase in ROS production by MECEs cells was found in this study, which is consistent with previous related ndings, suggesting that oxidative damage caused by increased ROS is related to the development of high tear osmolar pressure and dry eye.
MSC, as a pluripotent stem cell, has good immunomodulatory and anti-in ammatory properties, and Yang et al. [31] reported that BMSC injection at the peak of disease or onset in patients with primary dry syndrome inhibited the proliferation of T and B cells. Li Ying et al. [32] found that MSC could be induced to differentiate into corneal epithelial cells and reconstruct the normal eye surface. At the same time, MSC could also reduce the immune in ammatory response and improve the microenvironment of the eye surface by secreting soluble factors, so as to promote the repair of corneal epithelium [33] . Therefore, MSC may be better than hormones and immunosuppressants in the anti-in ammatory and immunomodulatory treatment of dry eyes [34,35] . Recently, Rhian et al. [36] also proposed a new ability of MSC-antioxidant. They suggested that the antioxidant properties of MSC include multiple mechanisms, including scavenging ROS, promoting endogenous antioxidant defense, immunomodulation through ROS inhibition, and improving mitochondrial dysfunction, which may signi cantly reduce in ammatory responses and oxidative damage in diabetic nephropathy, severe acute pancreatitis, Alzheimer's disease and metabolic renal vascular disease [37][38][39][40] . The antioxidant effect of BMSC was con rmed in this study. We found that BMSC effectively down-regulated ROS levels in both hypertonicity-induced MCECs cells and benzalkonium-induced dry eye model mice, and alleviated the promotional effects of ROS inducer (RA) on ROS, NLRP3, Caspase-1, IL-1β, ASC levels and apoptosis as well as the inhibitory effects on cell proliferation, suggesting that BMSC may exert its antioxidant effects by scavenging ROS to exert its antioxidant effects, thereby downregulating pro-in ammatory cytokine levels and repairing dry eye corneal damage.
TLR4/MyD88/NF-κB signaling pathway is involved in a variety of diseases. For example, alleviating LPSinduced acute lung injury in mice by regulating TLR4/MyD88/NF-κB pathway [41] . TLR4/MyD88/NF-κB signal participates in myocardial in ammatory response after CME, activates NLRP3 in ammasome, promotes in ammatory cascade, and further aggravates myocardial injury [42] . A large body of evidence suggests that TLR4 activation leads to ROS secretion by neutrophils [43] and it has been shown that inhibition of NF-κB activation decreases ROS [44][45][46] . In addition, in a previous study, it was found that BMSCS exosomal Mir-146a could regulate the in ammatory response of diabetes mellitus (DR) by inhibiting the TLR4/MyD88/NF-κB pathway. In this study, we investigated whether BMSC could alleviate dry eye by inhibiting TLR4/MYD88/NF-κB signaling axis to reduce ROS and in ammation levels through establishing a dry eye cell model and mouse model. The results showed that the expressions of NF-κB, MYD88, TLR4 were increased in the dry eye cell model and the dry eye mouse model, and the expressions of ROS, TNF-α, IL-1β, caspase-1, ASC and NLRP3, were down-regulated after adding inhibitors of NF-κB, MyD88 and TLR4, respectively. The levels of ROS, TNF-α, IL-1β, caspase-1, ASC and NLRP3 were signi cantly down-regulated after superimposing inhibitors of NF-κB, MyD88 and TLR4, on BMSCS. In conclusion, BMSC can inhibit the secretion of proin ammatory cytokines by inhibiting TLR4/MYD88/NF-κB signaling pathway, promote the proliferation of mouse corneal epithelial cells, inhibit the level of apoptosis, inhibit in ammatory response, and alleviate corneal injury in dry eye model mice.
This study is based on the mechanism of BMSC-mediated TLR4/MyD88/NF-κB signaling pathway to alleviate dry eye, which provides a certain theoretical basis for the treatment of dry eye. However, the speci c practical mechanism has not been further veri ed in clinical practice, and further in-depth and detailed research is needed in the future.

Declarations
Ethical approval and consent to participate All animal experimental protocols were approved by the Experimental Animal Ethics Committee of Kunming Yan 'an Hospital (2021070). All methods are performed in accordance with the relevant guidelines and regulations and in accordance with ARRIVE guidelines.

Consent to publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

4Competing interests
The authors declare that they have no competing interests.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Fulllengthgelimages.pdf