Mast Cell Activation Protects Cornea by Promoting Neutrophil Infiltration via Stimulating ICAM-1 and Vascular Dilation in Fungal Keratitis

The role of mast cells (MCs) in fungal infection is largely unknown. This study was to explore a protective role and mechanism of MCs in fungal keratitis. Experimental fungal keratitis (FK) mouse model was developed. Mice untreated (UT) or receiving corneal wound without fungal infection (Mock) were used as controls. Large number of connective tissue MCs was found in normal mice. MC activation with degranulation was largely observed, and the percentage of degranulated/total cells was high in FK. Dilated limbal vasculature with increased permeability, as well as largely infiltrated neutrophils with stimulated ICAM-1 protein levels were observed in corneas of FK mice, when compared with Mock and UT mice. Interestingly, pretreatment with cromolyn sodium (Block) significantly blocked MC degranulation, dramatically suppressed vascular dilation and permeability, and markedly reduced neutrophil infiltration with lower ICAM-1 levels in FK mice at 6–24 hours. Furthermore, the Block mice manifested prolonged disease course, increased pathological damage, and vigorous fungus growth, with much higher corneal perforation rate than FK mice at 72 h. These findings reveal a novel phenomenon that MCs play a vital role in protecting cornea against fungal infection through degranulation that promotes neutrophil infiltration via stimulating ICAM-1 production and limbal vascular dilation and permeability.

and scarring formation. MCs also induce the expression of ICAM-1 in endothelial cells through the release of cytokines to promote neutrophil chemotaxis 16 and perform antimicrobial actions by secreting antimicrobial peptides defensins 17 . Studies on corneas infected by virus in MC-knockout mice manifest increased virus load and leukocyte infiltrations, but decreased corneal transparency 18 . When MCs encountered Candida albicans, they can digest spores by endocytosis 19 . After toe pads were inoculated with the fungus, MC activation-associated proteins can be detected in the serum of guinea pigs 20 . However, no direct data demonstrated the role of MCs or their anti-fungal mechanism in fungal keratitis.
To investigate how MCs influence corneal fungal infection, we established the mouse model of fungal keratitis with Fusarium Solani to explore possible functions and mechanisms of MCs during fungal infection in the cornea. We found that MCs activation initiated by fungi infection contributed to the neutrophil infiltration by influencing the expression of ICAM-1 to the infected cornea and result in protection in FK.

Results
Connective tissue mast cells were observed to be located in the corneal limbus. MCs are distributed in the ocular surface [21][22][23] . To reveal the characteristics of these cells, we used classic staining methodstoluidine blue staining combining the whole-mount technique to verify the distribution of MCs in the cornea and conjunctiva. Consistent with previous studies, we found the presence of a large number of MCs in the conjunctiva (Fig. 1A) and limbus. These cells were toluidine blue and CD117 positive (Fig. 1B), which was a constructive expressed molecular on the MC during maturation. MCs also located near to the blood vessels and nerves (Fig. 1C).
MCs have been known to consist of at least two types, connective tissue mast cells (CTMCs) and mucosal mast cells (MMCs) based on to their locations and enzymes. Some researchers show both types of cells may co-exist in some areas. Studies show that mouse MC protease (mMCPT)-6 was specifically expressed in CTMC while mMCPT-1 existed only in MMCs. To determine the MC subtypes in the limbus, the corneal whole mounting and immunofluorescence staining technique was applied. Our results found that the MCs in limbus were mMCPT-6 positive but mMCPT-1 negative cells (Fig. 1D), indicating they were CTMCs.
MCs were largely activated in fungal keratitis. MC cytoplasm contains a large number of particles.
When MCs are irritated by parasites, bacteria, and other pathogens, granules wrapped in cytoplasm are released into the matrix of surrounding tissue 24 . In order to observe whether the fungal infection trigger MC degranulation in the cornea, we established FK models to observe the presentation of activated (degranulated) MC in corneal limbus. We found that a large number of particles were around the MCs after 6 hours of corneal irritation by fungal infection or trauma (Fig. 2A). The number of MCs was 645 ± 26 cells per cornea in normal mice, and no significant change of the MC number was observed in Mock, FK and Block group. The percentage of degranulated  Mast cell degranulation stimulated vessels dilation in corneal limbus. The vascular diameter controls the speed of blood flow and numbers of immune cell infiltration. In order to determine the effect of MCs on blood vessel, we utilized cromolyn sodium, immunofluorescence staining, and Evans blue (EB) to detect changes in the vessel diameter and vessel permeability in the process of FK.
Mast cell degranulation increased ICAM-1 expression. In addition to vascular changes, ICAM-1 is crucial for neutrophil migration into the sites of infection through the vascular wall 25,26 . An enzyme-linked immune-sorbent assay was used to investigate whether ICAM-1 protein expression changed by MC activation.
We found immune cell infiltration at the early stage of fungal keratitis. In wild type mice, neutrophils were only distributed in the corneal limbus with a density of 248 ± 68 cells per cornea (Figs 3B and 4D). Neutrophil infiltration increased to 1085 ± 57 cells at 6 h in the Mock group, and also increased slightly to 1216 ± 298, 1987 ± 340 cells at 12 h and 24 h, respectively. Neutrophil infiltration in the FK group significantly increased to 1258 ± 98 cells, in contrast to the Mock group (P = 0.023) at 6 h; it was increased to 1847 ± 258 cells at 12 h (P = 0.015) and 5046 ± 258 cells at 24 h (P < 0.001), significantly higher than that of Mock group. When MC activation was blocked, the number of neutrophils was decreased to 511 ± 92 cells at 6 h, lower than the Mock group (P < 0.001) and FK group (P < 0.001), but similar to normal corneas (P = 0.094, Figs 3B and 4D). After 12 h and 24 h stimulation by fungi, the neutrophil infiltration in the Block group showed a very small increase, but was still significantly lower than the trauma group (P = 0.040, P = 0.011) and FK group (P < 0.001, P < 0.001) (Figs 3B and 4D).

Mast cell activation was related to clinical and pathological process of fungal keratitis.
To investigate the effects of MC activation on the clinical process and pathological change of FK, we treated animals In the Block group, the corneal clinical manifestations were similar as that of the FK group and the Mock group at 6 h and 12 h, with no consequential differences in clinical scores (P = 1, n = 6 for each time point of the cornea) between the FK group and Block group (Fig. 5A,C). However at 24 h (Fig. 5A): the Block group was characterized as transparent cornea, the iris was clearly visible, with significant differences with FK group in clinical score (Fig. 5A). The differences consisted at 36 h, with no hypopyon formations. Although the Block group manifested corneal opacity at 72 h, it differed from the FK group in lesion characteristics under the slit lamp microscope: restricted dry and dense lesions in the FK group suggested the recovery process of corneal inflammation; however, moist and crisp lesions in the Block group revealed a severe inflammatory reaction. By a total of three batches of experiments (six eyes per group), the clinical score in the Block group was greater than that of the FK group at 72 h (Fig. 5C), and also perforated rates were higher than FK group (P = 0.023) (Fig. 5D).
Along with the slit lamp images, histological staining was also performed to understand the histopathology changes underlying the corneal clinical symptom (Fig. 5B). In the FK group, only a small amount of inflammatory cell infiltration were present in the cornea at 6 h and 12 h, and the organizational structure showed no difference from the normal corneas. At 24 h, there was a great deal of inflammatory cell infiltration in the cornea with immune cell underneath the endothelium, as well as increased corneal thickness, and the matrix structure disappeared. Although corneas were damaged, scar tissue within the cornea was formed at 72 h. In the Block group, HE staining showed it underwent no damage in the structure when comparing with the FK group at 6 h and12 h (Fig. 5B), with almost no inflammatory cell infiltration, and it only showed increased corneal thickness without structural destruction at 24 h (Fig. 5B). The differences between the FK group and the Block group were also manifested by clinical scores (P < 0.001) (Fig. 5C). After 72 h, the normal corneal morphology disappeared, and all that remained was the organization framework with large number of immune cells adhering to the endothelium.
HRT3 confocal microscopy, GMS (Gomori's methenamine silver) and immunofluorescence staining were performed to examine the growth of fungi. HRT3 images (Fig. 6A) showed that fungus growth was vigorously in 12-36 h, but significantly reduced at 48 h, and there was few fungi detected at 72 h in the FK model. The same results were also showed by GMS staining (Fig. 6B). Fungal volume from immunofluorescence staining (Fig. 6C) was at low level at 6-12 h (P = 1.000, P = 1.000), increased significantly at 24 h (P < 0.001), reached the peak at 36 h (P = 0.019), then significantly decreased to low level at 48 h, and very few at 72 h in the FK group. However, the strong growth of fungi continued and lasted from12 to 72 h in the Block group (Fig. 6C), much higher than FK group at 48-72 h, (P < 0.001), which suggested that block of blocking MC activation decreased neutrophil suppression of fungal growth in keratitis.
These findings suggest that MC activation by corneal central fungi may cause early neutrophil infiltration from the limbus site and thus lead to severe clinical manifestations in the early stage of fungal keratitis, indirect anti-fungal activity, and finally preventing the late tissue damage caused by fungi growth, decreasing the perforation rate in the late stage.

Discussion
MCs play a key role in the fight against bacterial infection and viral infection 28,29 , however their role remains poorly understood in the process of corneal fungal infection. This research found that MCs, which are mainly distributed in the corneal limbus, play a dominant role in controlling the fungus infection of the cornea by several mechanisms. This conclusion is based on following evidence: (1) Fungi infection can induce activation and degranulation of MCs in the limbus; (2) cromolyn, the MC stabilizer for degranulation, can significantly prolong the course of disease, increase pathological damage, and increase the perforation rate of fungal keratitis; (3) cromolyn reduces neutrophil infiltration to the site of the fungal infection in the central cornea; (4) MC activation can increase ICAM-1 expression.
Bacteria, parasites, fungi and other physical stimuli can promote MCs activation through the recognition receptors on the cell surface. In the vitro studies, these microbes can directly stimulate the MC degranulation. In this study, the central cornea was irritated by fungi causing MC activation in limbus signaling there are some mechanisms during cell activation. Neurons can secrete nerve growth factors and inflammatory factors, which can bind to the receptors on MC surface following different stimuli 30 . So we conjecture that fungi irritated corneal central nerve fibers to secrete nerve factors and inflammatory factors, then induction of MC degranulation via receptors on MC surface. However, detailed studies are required to establish the definitive mechanism. Many immune cells participate in fungal infection. There are few evidences on MC in FK. Studies show that MCs can directly kill microorganisms 26 , and also kill bacteria indirectly by chemotaxis of neutrophils [31][32][33] . MCs can also make direct contact with the bacteria to produce a variety of antibacterial substances, and they can phagocytize and subsequently display the bacterial antigens on their cell surfaces 28,34 . Recent findings have revealed that MCs form the extra-cellular traps which can trap the bacteria to some location to facilitate the anti-bacterial reaction 35,36 . We demonstrated that MCs have an indirect effect on killing fungi in the pathogenesis of FK, via reducing fungal growth by promoting neutrophil infiltration. MC activation is needed for controlling the invasion of fungi was also similar to the study on rat model of acute invasive fungal rhino sinusitis 37 . In a HSV-1 keratitis animal model, research shows that MCs knock-out can reduce the infiltration of neutrophils in the cornea, so as to promote the elimination of virus tilter 18 . In our study, effects of MC activation blocked by cromolyn on neutrophil infiltration is also similar, but the effect on the elimination of fungi is opposite to that of HSV-1, and further studies are needed.
MCs are necessary for corneal wound healing and coordinating neutrophil and platelet recruitment 38 . Eric Pearlman research team confirmed that in the infectious keratitis, corneal haze formation was consistent with neutrophils infiltration 27,39 . The fungal load in neutrophil-knockout mice was higher than the controls 40 , showing that neutrophils are crucial to produce a fungicidal effect. Neutrophils engulf the spores and then kill the fungi or spore through oxygen-dependent or oxygen-independent mechanisms; it can also be formed by extracellular traps to limit fungal hyphae and spore wavering, and then ultimately digest the hyphae or spore through enzymes. In the present study, MCs were located only in the limbus, while the fungal hyphae in the cornea were mainly located in the central cornea. Neutrophils were attracted to the corneal center, where MCs were activated by the fungi. Therefore, it is clear that MCs play an important role in the chemotaxis of neutrophils in fungal keratitis, which have antifungal effects indirectly.
MCs which locate around the blood vessels can promote changes in vascular diameter 41 . Changes of blood vessels were different when MCs activated by different stimuli 42 : After intraperitoneal injection of saline, there were no change in the diameter of the peritoneal blood vessels; after 48/80 stimulation, the vascular diameter decreased at the first 9 days, and then gradually return to normal. The different effect of the two stimuli on vessel diameter may be due to the degree of MC stimulation. Our results suggest that MC activation is less likely to be caused by trauma than by fungal stimulation. This may explain why trauma stimulus does not cause a change in the vessel diameter, whereas a persistent stimulation of the fungus causes increases. Experimental studies showed that 48/80 stimulation of MCs in the early tissue culture (6-8 days) did not increase the vascular permeability of macromolecules, while promotion of vascular permeability of macromolecules occurred in the late stage (9-10 days) 43 . Strbian et al. found that cromolyn could restrict hemorrhage formation induced by plasminogen activator 44 . Evans blue can bind with albumin to form Evans blue-albumin complex. In this study, vascular permeability detected by Evans blue concentration was increased by fungal irritation and was higher than trauma irritation but can be suppressed by cromolyn sodium in Block group.
MC activation promotes neutrophil infiltration, blood vessel diameter increases, and ICAM-1 changes. In fungal keratitis, corneal infiltration of inflammatory cells is mainly due to neutrophils. ICAM-1 plays an important role in the chemotaxis of neutrophils during inflammation 45,46 . The infiltration of neutrophils in the blood vessel is divided into three steps: rolling, adhesion, and penetrating. ICAM-1 is crucial for neutrophil amoeboid movement through the vascular wall to the damaged site. We confirmed that the increased expression of ICAM-1 in the cornea initially then decreases in the late stage in the FK group were coherent with the changes in the number of infiltrated neutrophils. The neutrophil infiltration and ICAM-1 expression were blocked by MC inhibitors during the early stage of fungal keratitis.
There are two factors inducing perforation in FK: one is the virulence or mechanical penetration of the fungus itself, and the other is the destruction of the inflammatory cells to the cornea. When neutrophils kill fungi, the intracellular enzyme is released, causing the destruction of corneal tissue 47 . Neutrophil chemotaxis was inhibited in the MC non-activation group, which made it unable to kill fungi. Consequently, the fungal virulence effect is far greater than the destructive effects of neutrophils on the tissue organization, which may be the reason that the perforation rate of the block group was higher than that of the FK group. This demonstrated that MC activation was required for controlling fungal keratitis.
In this experiment, the choice of cromolyn sodium as a MC stabilizer, has certain limitations. Cromolyn sodium can affect macrophages and other inflammatory cells, and it is unclear whether there is an immunosuppressive effect on other cells of the corneal tissue. In addition, studies have shown that in mice, effects of cromolyn sodium on MC membrane stability is lower than that of the other animal in vivo effect 48 . Cromolyn may have an effect on the degranulation of other granule cells, such as neutrophils, basophils. To investigate this role, our next step will be to use MC knockout mice to verify this effect. In addition, the role of MCs in mice may be different with that in the human body, and the specific role in the human body still needs further experimental study.
In the present study, we showed connective MCs are found in the corneal limbus and conjunctiva. Corneal fungal infection led to degranulation of these MCs. MC activation can promote vascular changes and increase ICAM-1 expression, thereby promoting neutrophil chemotaxis, as well as affect corneal transparency in the early stage and decrease the perforation rate. The specific reason for MC degranulation caused by fungi infection still needs further investigation. This study adds to evidence on the pathogenesis of fungal keratitis, and may provide information for the treatment of fungal keratitis in the future.

Materials and Methods
Animals. All experimental protocols used in these studies were maintained and animals euthanized according University (MARC) (Nanjing, China). The mice were maintained on a 12:12-day cycle in a temperature-controlled room at 25 °C.
Mast cell stabilization and the FK model. Before injection, cromolyn sodium (Sigma-Aldrich, St. Louis, MO) was dissolved in phosphate-buffered saline (PBS, 0.01 M) at a concentration of 4 mg/ml. The cromolyn sodium solution was protected from light during preparation. To stabilize the mast cells, cromolyn sodium (100 mg/kg body weight; Sigma-Aldrich) or vehicle (0.01 M PBS) was administered to the mice by a single intraperitoneal (i.p.) injection for 30-120 min before and after the model 49 , and then additional injections were given every six hours for the duration of the experiment.
The C57BL/6 J mice were randomly divided into three groups, which were the UT group, Mock group, FK group, and Block group, according to the littermate grouping method. The mice were anesthetized by an intraperitoneal injection of sodium pentobarbital (80 mg/kg body weight), and l% (weight/volume) tetracaine hydrochloride eye drops were used for corneal surface anesthesia. Then, under a dissecting microscope, the FK model infected with Fusarium solani (F. 3.1791) was built referring to previous methods 50 : using sterile knife to make crossing scratch on the corneal central with the depth of under Bowman's membrane, then bamboo sticks with fungi hypha scratch 2 to 3 times following the crossing bottom to mimic the natural fungi infection in cornea. Corneas of the Mock group were processed the same procedure above, except fungi infection.
Corneal clinical score and confocal microscopy. Corneas of three groups were continuously observed and photographed at the corresponding time points under a slit-lamp microscope (SL-8Z, TOPCON, Japan) and a HRT3 confocal microscope (Heidelberg, Germany). Clinical scores were recorded under slit-lamp microscope according to lesion area, depth, corneal neovascularization and others 50 . Parameters of lesion area, depth, corneal neovascularization were given a grade from 0 (normal) to 4 (very severe), and others that contained hypopyon, hyphema, descemetocele and perforation grade from 1 to 4. The fungi hypha and immune cells infiltration were observed by the HRT3 confocal microscope.

Histologic preparations.
For the histologic analysis, mice were euthanized by cervical dislocation at different time points, and the whole cornea and partial sclera were removed and then fixed in 10% neutral formalin (F) for 120 min. Two-micrometer sections were separately stained with Gomori methenamine silver (GMS), or hematoxylin-eosin (HE), and reviewed by light microscopy.
Observation of Flat-Mounted Cornea. The mice were euthanized by cervical dislocation at different time points. Whole eyes were removed and then fixed with an alcohol-formaldehyde solution (AF) for 120 min. The cornea was trimmed to retain the limbus, but the iris and ciliary body removed. After washing five times with 0.01 M PBS (10 min each), the corneas were dipped in the following solutions in sequence: a 0.2% Triton-BSA solution for 30 min, a 2% BSA solution for 30 min, and the primary antibodies (anti-mouse CD117-FITC provided by Zhijie Li, anti-mouse Tubulin-PE, anti-mouse CD31-PE, anti-mouse MCPT-1, anti-mouse MCPT-6, and anti-mouse Gr-1-FITC, were purchased from eBioscience) incubation for one night. Then, the secondary antibodies (Alexa568, Alexa488, eBioscience) were incubated for 6 hours, washed; incubated in CFW (Sigma-Aldrich, St. Louis, MO) for 30 min, washed; incubated with DAPI for 30 min, and washed. The corneas were then flat mounted with four or eight radial cuts (all of incubation sustained at 4 °C). After multiple immunofluorescence and CFW staining procedures, the flat-mounted corneas of the model groups were photographed by confocal laser scanning microscopy (Z axis, Z = 5 μm). The image analysis was performed by visual inspection of the individual image sections. The different corneal layers were identified by their characteristic morphometric features. Representations of the morphological analyses of the neutrophils and fungal volumes were performed using the Imaris software, version 7.3.1 × 64 (Bitplane, Zurich, Switzerland).
Vessel diameter and Limbal Vascular permeability. Vessel diameter detection: The Vessel diameter detection was based on the corneal whole-mount technique. Vessel diameter was detected under fluorescent microscope. The limbal microartery in the four equal parts of the cornea was select and captured under 40× objective lens. Then the diameter of limbal microartery was detected on AR software. The limbal microartery is located between two capillary veins. There were four values per cornea, six corneas in every group.
Limbal Vascular permeability: Evans Blue (EB, Sigma, St Louis, MO) in normal saline (2%, 4 ml/kg) was injected intravenously after operation. The mice were killed by cervical dislocation 1 h later and corneas were trimmed. After the tissue was then dried at 100 °C for 24 h, the samples were homogenized in methylformamide (Yongda, Tianjin, China) with 200 μl per cornea, incubated for 24 h at 60 °C, and centrifuged for 5 min at 1000 g. The absorbance (A) of supernatants was analyzed at 632 nm using Envision Multi label plate reader 2104 (Perkin Elmer, USA). The amount of EB (μg/g) was calculated through standard curve established by known concentrations of EB.
Statistical analysis. The statistical significance of the differences between groups at different times was determined by the one-way analysis of variance test, followed by the least significant difference t (LSD-t) test or Kruskal-Wallis Rank Sum test using SPSS V.17.0. A P value of < 0.05 was considered significant. The data are expressed as the means ± STD.