F.solani (No.3.1791) was obtained from China General Microbiological Culture Collection Center (CGMCC, Beijing, China). After two or three subcultures, two milliliters of stroke-physiological saline solution(NS) were added to the slant medium containing well-grown F. solani. The fungal hyphae were ground with a sterile glass rod for making the fungal suspension, solution of which was adjusted by turbidimeter to get 0.5 Mx fungal suspension.
In vitro viability test by CXL
Suspension of 10ul was placed to grow in each well of a sterile 96-well plate (Corning Life Science, Lowell, MA, USA) filled with 100 μl of potato dextrose agar (PDA, Beijing Sanyao Technique Development Co. China) per well for 24 h of incubation at 25 °C [16, 17]. The microplate wells have an internal diameter of 6.40 mm, so that the entire area can be exposed to UVA from 7.00 mm diameter light source.
Under sterile conditions, with the mass fraction of 0.1% riboflavin photosensitizer 20ul after 30-min, the cultures of F.solani were exposed to irradiation for different durations (respectively two, five, ten, 20 and 30 min). Further, the fungal viability in plate cultures irradiated with UVA (30 min) were evaluated under four conditions: no treatment (control), CXL: UVA (365 nm)/riboflavin, riboflavin and UVA (365 nm). Each process was repeated three times.
The standard procedure applied in the treatment of progressive keratoconus was followed to assess the antifungal effects of riboflavin and long-wave UVA irradiation. At a distance of 5 cm, fungal was exposed to laser irradiation. The size of the laser beam spot was 7 mm. The UVA (365 nm) has a power density of 45 mW/cm^2 and fungal beyond the irradiation scope was sheltered against light. After irradiation, the culture dishes were placed at 25 °C in an incubator for 48 h, photos of which were taken with a digital camera. Fungal gray values were analyzed by Image J software , and then the quantification of fungal cells viability was calculated.
In total, 150male C57BL/6 J mice, weighing from 24 g to 27 g were used. The animals were fed and handled in strict compliance with the ARVO Statement For the Use of Animals in Ophthalmic and Vision Research. The study was approved by the Henan Eye Hospital Institutional Committee. All mice were anesthetized with an intraperitoneal injection of pentobarbital sodium (80 mg/kg.b.w.) (Sigmae Aldrich, USA) before interventions. Topical anesthetic (1% tetracaine hydrochloride drops) was applied for local anesthesia of corneal. The cornea of each mouse was scarified using a sterile scalpel to create a superficial wound of intersecting marks in a grid pattern, as referred . A sharpened bamboo toothpick (0.30 mm tip diameter, 1.10 mm tip length) was used to scrape along the scratch 2 to 3 times to create a rough surface. The scarified cornea was subsequently smeared with fungi for fungal infection. A positive model was confirmed by an observation from a confocal microscope detection of fungal hyphae and/or spores 24 h post-inoculation.
Corneal CXL treatment
The mice were divided randomly into three groups: group A. control group; group B. treated with CXL for two minutes; group C. treated with CXL for three minutes. Treatment started 24 h after fungal was smeared. The corneal epithelium of mice was scraped each group of 2-mm fields (which is a standard method). Then group B and group C were initiated CXL treatment. The lesions were instilled immediately with 0.1% riboflavin solution at 5-min intervals until a yellow dye in the aqueous humor were confirmed in the anterior chambers under a slit lamp with a cobalt blue filter. The corneas were continuously exposed to UVA irradiation (365 nm; irradiance 45 mW/cm^2; CCL-365VARIO instrument: guangteng technology xiamen co.LTD).
All mice were observed under a slit-lamp biomicroscopic and were scored accordance with the Schreiber scoring system  daily for 7 days and 10th day after modeling. Corneal lesion, the depth of ulceration, corneal opacity, edema, neovascularization were clinically evaluated on the 10 days. Corneal ulcers were graded as follows : stage 0, no ulcer; stage 1, superficial ulcer; stage 2, medium-depth ulcer; stage 3, deep ulcer; stage 4, descemetocele; and stage 5, corneal perforation.
Corneal fungal plate count
Ten mice were randomly sacrificed in each group in 1, 3, 7, 10 days after modeling respectively, eyes were enucleated for quantitative fungal recovery confirmed by plate counts , 0.85% of the stroke-physiological saline solution(NS) was added to the cornea which was grounded with a grinding stick to release the fungus. The suspension was centrifuged at 2000 rpm for 10 min for collecting supernatant, then cultured on potato dextrose agar(PDA) at 25 °C for 2 days, the quantity of colonies was determined in all cultures.
Pathological observation of corneal tissue in mice
On the 10th day after treatment, mice went through cervical dislocation and the corneals were extracted for histopathological examination. The corneas of mice, with the mass fraction of 4% formaldehyde fixed after 24 h. Then the samples were embedded in paraffin and sagittally cut into 5-um-thick sections. Hematoxylin–eosin (HE) staining was used for histomorphological analysis of all groups. Observation was performed with the aid of a Nikon Eclipse E100 light microscope (Nikon, Sendai, Japan) under × 20 magnification. The disposition of collagen fibers, and inflammatory infiltration, were analyzed in terms of the microcorneal slices. Based on a scale of 0 to 4, the previously described inflammation score  was modified to the following: 0, no inflammation; 1, minimum change; 2. mild changes; 3. moderate changes; 4. severe changes.
Statistical analyses were made by applying the triplicate values of each experimental condition. The data were expressed as the mean ± SD. Fungal gray values for different durations were compared with the control group was determined by T-test. Wilcoxon matched sign rank test for 2 related samples was used to compare the 30-min intra-group values. Clinical score and corneal fungal plate count obtained for each mouse were given One-way ANOVA for multiple comparisons. When homogeneity of variance was detected, the least significant difference (LSD) was used. In all other instances, Tamhane’s T2 test was performed. Pearson product correlation analysis was used to study the relationship between clinical score and corneal fungal plate count Nonparametric Kruskal–Wallis 1-way analysis of variance by rank were used for the corneal transformation. P < 0.05 was judged to be statistically significant. SPSS21.0 software (IBM, Inc., Chicago, IL, USA) was used for data analysis.