We collected 45 Candida isolates from postoperative ocular infections between January 2010 and December 2024, including 22 C. parapsilosis, 16 C. albicans, 5 C. tropicalis, and 2 C. glabrata. C. parapsilosis ATCC 22019 was used as a reference control strain. It is used as a quality control strain for mass spectrometry identification and biofilm formation assays. In this study, all C. parapsilosis isolates were identified to the species level, not merely as part of the complex. Each isolate was identified by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). A small amount of colony material from a fresh culture was picked with a sterile swab and smeared evenly onto a MALDI-TOF target plate, then air-dried at room temperature. After drying, 1μL of 70% formic acid was added to the spot, followed by 1μL of matrix solution (saturated α-cyano-4-hydroxycinnamic acid in 50% acetonitrile + 47.5% ultrapure water + 2.5% trifluoroacetic acid). After the sample dried again, identification was performed using a Smart MS 5020 automated microbial MALDI-TOF MS system (Zhuhai DL Biotech). A score≥2.0 was considered a reliable species-level identification (Xie et al., 2019; Rodríguez-Temporal et al., 2023).

Each Candida isolate was inoculated on Sabouraud dextrose agar and grown at 35°C for 48 h, then washed twice with sterile phosphate-buffered saline (PBS) (10, 000 g, 5 min). The cells were resuspended in Sabouraud dextrose broth and adjusted to 2×10⁶ cells/mL. Then, 100μL of this suspension was added to each well of a sterile 96-well plate (three replicate wells per strain) and incubated at 37°C for 24 h. After incubation, the medium was discarded and the wells were gently washed three times with PBS. Biofilm formation was quantified by measuring the absorbance at 450 nm (OD450) after crystal violet staining (described below). According to Tavanti et al., biofilm-forming ability was categorized as follows: OD450 < 0.03, no biofilm; 0.03 ≤ OD450 < 0.08, weak biofilm; 0.08 ≤ OD450 < 0.16, moderate biofilm; OD450 ≥ 0.16, strong biofilm (Tavanti et al., 2010).

To assess biofilm biomass and metabolic activity at different temperatures, 100μL of the same cell suspension was added to each well of a 96-well plate (three replicates per strain) and incubated either at 37°C for 24 h or at 4°C for 14 days. Two different methods were then used to evaluate biofilm production (Pierce et al., 2008). For biomass quantification, 100 μL of 10% formaldehyde was added to each well to fix the biofilm at room temperature for 2 min. Then, 100μL of 20 mg/mL crystal violet solution was added to each well and incubated for 30 min. The wells were washed with 95% ethanol (100μL) to destain, and absorbance was measured at 620 nm using a spectrophotometer. For metabolic activity, an XTT reduction assay was performed: a solution of 0.5 mg/mL XTT and 1 mmol/L menadione was freshly prepared and 100μL of this mixture was added to each well, followed by incubation at 37°C for 2 h. The absorbance of the XTT formazan product was measured at 450 nm with a microplate reader.

The microbial adhesion to hydrocarbons (MATH) assay was used to evaluate cell surface hydrophobicity (CSH). Candida cells were cultured in Sabouraud dextrose broth at 37°C for 24 h and washed twice with PBS. The cells were resuspended in PBS and adjusted to an optical density (OD600) of 0.4–0.5 (initial OD, A0). Then 0.4 mL of n-hexadecane was gently layered over 3 mL of the cell suspension. The two-phase system was vortexed vigorously and allowed to separate, after which the OD600 of the aqueous phase was measured (A1). The percentage of cell surface hydrophobicity was calculated as: Hydrophobicity (%) = [1–(A1/A0)]×100% (Harjai et al., 2014). Each strain was tested in at least three independent experiments.

Candida isolates were grown on Sabouraud agar at 35°C for 48 h and washed twice with sterile PBS (10, 000 g, 5 min). The cells were resuspended in sterile PBS and adjusted to 2×10⁶ cells/mL. For the adhesion assay (Silva-Dias et al., 2015), 300μL of the yeast suspension was mixed with 300μL of fluorescent microspheres (2×10⁸ microspheres/mL) in a sterile tube and incubated on a shaker at 150 rpm at room temperature for 30 min. The mixture was then analyzed by flow cytometry, collecting 50, 000 yeast cell events per sample. According to Silva-Dias et al., the percentage of yeast cells with bound fluorescent microspheres (gated as population P2) was used as an indicator of adhesion ability. Adhesion was categorized as weak (P2 ≤ 20%), moderate (20% < P2 ≤ 30%), strong (30% < P2 ≤ 50%), or very strong (P2 > 50%). Additionally, the adhesion patterns were classified as homogenic or heterogenic based on the distribution of microspheres per cell: a homogenic distribution (single peak in flow cytometry) indicates each yeast cell is bound to only one microsphere, whereas a heterogenic distribution (multiple peaks) indicates individual yeast cells have more than one microsphere attached. All adhesion experiments were performed at least three times.

Each Candida isolate, after identification, was streaked for isolation on CHROMagar Candida medium and incubated at 35°C for 5 days to observe colony morphology. Smears of the colonies were then prepared and stained with Gram stain, lactophenol cotton blue, and a fungal fluorescence stain, to observe and record spore morphology.

1, 3-β-D-glucan is a polysaccharide component widely present in fungal cell walls. The G-test (β-glucan test) uses a kinetic chromogenic assay to detect 1, 3-β-D-glucan. Given that serum 1, 3-β-D-glucan is an early indicator of invasive fungal infection, we explored the relationship between biofilm formation and 1, 3-β-D-glucan release by Candida isolates (Giacobbe et al., 2017). Each Candida strain was inoculated into five replicate wells of a 96-well plate (100μL of suspension per well at 2×10⁶ cells/mL) and incubated at 37°C for 24 h. The culture supernatants from the five wells were pooled and centrifuged (3000 rpm, 10 min). The samples were processed using a Fungitell 1, 3-β-D-glucan detection kit, and 1, 3-β-D-glucan concentrations were measured with a kinetic microplate reader at 75 minutes into the assay.

A Transwell chamber system was used to evaluate neutrophil chemotaxis. Neutrophils were isolated from fresh human whole blood using a neutrophil isolation kit and diluted in RPMI 1640 medium. Based on preliminary tests, we found that a neutrophil concentration of 5×10⁵/mL in the upper chamber and a Candida suspension of 5×10⁴ cells/mL in the lower chamber allowed clear observation of neutrophil migration through the membrane. Therefore, 200μL of the neutrophil suspension (5×10⁵/mL) was added to the upper chamber, and 500μL of a Candida culture (5×10⁴ cells/mL, grown for 24 h in RPMI 1640) was added to the lower chamber. After incubating at 37°C with 5% CO₂ for 5 h, the Transwell insert was removed, and neutrophils that had migrated into the lower chamber were counted (Justus et al., 2023).

Susceptibility to nine antifungal agents commonly used in ophthalmology was tested for each Candida isolate in both planktonic (free-living) and biofilm-forming states. The antifungal agents and concentration ranges were: amphotericin B (AMB, 0.125–16μg/mL), chlorhexidine (CDI, 0.03–16μg/mL), natamycin (NAT, 0.12–128μg/mL), terbinafine (TDN, 0.015–16μg/mL), voriconazole (VOR, 0.015–16μg/mL), posaconazole (POS, 0.015–16 μg/mL), itraconazole (ITR, 0.03–16μg/mL), and caspofungin (CAS, 0.016–16μg/mL). The MIC values of planktonic Candida were determined according to the CLSI standard protocols M27-S4 and M27-A3 (Clinical and Laboratory Standards Institute. (CLSI), 2008; Clinical and Laboratory Standards Institute. (CLSI), 2012). According to the experiment by Fernanda M. Bezerra et al (Bezerra et al., 2023), the Candida suspension concentration was adjusted to 0.5 McFarland and inoculated into 96-well cell culture plates (100μL/well), followed by incubation at 37°C for 24 h. A negative control group without antifungal drugs and a blank control group without Candida were established. The wells were rinsed three times with 200 μL PBS to remove non-adherent fungal cells. Antifungal drugs with serial two-fold dilutions were added and incubated at 35°C for 24 h, followed by three additional PBS washes (200 μL each) to eliminate non-adherent cells. Biofilm MICs (BMIC) were defined as the drug concentrations that reduced metabolic activity by 50% compared to drug-free controls. The metabolic activity of biofilms was assessed using the XTT reduction assay as previously described. Each drug-isolate combination was tested in triplicate, and mean values were calculated.

Genomic DNA of C. parapsilosis isolates was extracted using a bacterial DNA extraction kit (Vazyme, Nanjing). Four polymorphic microsatellite loci (CP1, CP4, CP6, and B5) with high discriminatory power were selected based on the method of Sabino et al (Sabino et al., 2010). The primer sequences and fluorescent labels for these loci are shown in Table 1 (Sabino et al., 2010). PCR amplification was performed in a 25 μL reaction containing 0.5 μL forward primer (10 μmol/L), 0.5 μL reverse primer (10 μmol/L), 1 μL DNA template, 0.5 μL dNTP mix (each 5 μmol/L), 2.5 μL 10× Taq buffer (with MgCl₂), 0.2 μL Taq DNA polymerase (5 U/μL), and sterile water to 25 μL. The PCR cycling program was: 95°C for 5 min; 30 cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 30 s; and a final extension at 72°C for 10 min. PCR products were analyzed by high-resolution capillary electrophoresis (Sangon Biotech, Shanghai). Allele sizes were determined for each locus, and multilocus genotypes were assigned based on the combination of alleles at the four loci. Allele and genotype frequencies were calculated using GeneMapper software.

GraphPad Prism 10 (GraphPad Software, San Diego, CA) was used for statistical analyses and graphing. Quantitative data are expressed as mean ± standard error of the mean (Mean ± SEM). For non-normally distributed data, the Kruskal-Wallis test was used to compare groups, and p < 0.05 was considered statistically significant. Correlations were analyzed using Pearson’s correlation coefficient or Spearman’s rank correlation coefficient, as appropriate. For microsatellite data, cluster analysis of C. parapsilosis genotypes was performed using the UPGMA method in MVSP (v3.13n) software, and genetic distances between isolates were calculated.

A total of 45 Candida isolates from postoperative ocular infections were identified by MALDI-TOF MS. These included 22 C. parapsilosis (48.9%), 16 C. albicans (35.6%), 5 C. tropicalis (11.1%), and 2 C. glabrata (4.4%). All identification scores were ≥ 2.000, indicating reliable species-level identification.

Among the 45 patients with postoperative ocular Candida infection, 26 (57.8%) were male and 19 (42.2%) were female. The mean age was 51.1 ± 14.8 years. Fourteen patients (31.1%) were from Northeast China, 27 (60.0%) were from North China, and 4 (8.9%) were from other regions. The most common underlying systemic disease was type 2 diabetes mellitus (22.2%, 10/45), followed by hypertension (13.3%, 6/45). The most frequent ophthalmic surgery associated with Candida infection was corneal transplantation (33.3%, 15/45), followed by cataract surgery (24.4%, 11/45) and vitrectomy (15.6%, 7/45). We grouped the patients by the infecting Candida species (C. parapsilosis group, C. albicans group, C. tropicalis group, C. glabrata group) and analyzed the types of prior ocular surgery in each group. Corneal transplantation was the most common surgery in the C. parapsilosis group (54.5%, 12/22), while cataract surgery was most common in the C. albicans group (31.3%, 5/16). Due to the small case numbers, the predominant surgery type in the C. tropicalis and C. glabrata groups was not clearly determined. In terms of clinical outcomes, 25 patients (55.6%) had their infections controlled with medication alone, whereas 20 patients (44.4%) required therapeutic corneal transplantation. Notably, in the C. parapsilosis group, 72.7% (16/22) of patients ultimately required surgical intervention, and 3 patients (13.6%) underwent a repeat corneal transplant (Table 2).

C. parapsilosis and C. tropicalis exhibited significantly higher biofilm-forming ability than C. albicans and C. glabrata (p < 0.001). Among all 45 isolates, 18 (40.0%) were classified as strong biofilm formers (OD450 ≥ 0.16), 6 (13.3%) as weak biofilm formers (OD450 < 0.08), and 21 (46.7%) as non-biofilm-forming. As shown in Figure 1, at both 37°C and 4°C, the total biofilm biomass and metabolic activity of C. parapsilosis biofilms were significantly higher than those of C. albicans(p<0.001, p<0.0001)(p<0.001, p<0.01). At 37°C, the biofilm biomass of C. parapsilosis was positively correlated with its biofilm metabolic activity (r=0.807**), whereas no such correlation was observed for the other species. At 4°C, there was no correlation between biofilm biomass and metabolic activity for any of the Candida species.

Candida parapsilosis and C. tropicalis showed significantly higher cell surface hydrophobicity percentages than C. albicans and C. glabrata (p < 0.001) (Figure 2). In C. parapsilosis and C. tropicalis, the cell surface hydrophobicity was positively correlated with biofilm biomass (r=0.48* and r=0.90*), whereas no correlation was observed for C. albicans or C. glabrata.

C. parapsilosis demonstrated the strongest adhesion ability among the species, with an adhesion percentage of (38.54 ± 3.97)%, and 73.7% of C. parapsilosis isolates were classified as having strong or very strong adhesion. C. tropicalis had the second highest adhesion (19.98 ± 7.50%), with 20% of isolates classified as strong/very strong. C. albicans and C. glabrata exhibited weak adhesion, with adhesion percentages below 10%. The adhesion of C. parapsilosis was significantly higher than that of C. albicans (p < 0.0001). A heterogeneous adhesion distribution pattern (multiple peaks in flow cytometry) was observed in 63.2% of C. parapsilosis isolates, which was significantly higher than in the other species (Figure 3). Additionally, C. parapsilosis adhesion was positively correlated with biofilm biomass (r=0.60**), a correlation significantly stronger than that of the other species. C. albicans also showed a positive correlation between adhesion and biofilm biomass (r=0.56*), as did C. tropicalis (r=0.90*).

In preliminary microscopic examinations of clinical corneal exudate smears, we observed that Candida could appear in two forms. In one form (non-biofilm-forming Candida), fungal spores were distributed sparsely and could be phagocytosed by neutrophils; in the other form (biofilm-producing Candida), fungal spores appeared in sheets that neutrophils could not phagocytose (Figure 4). This finding corroborated that some Candida can form biofilms upon infecting donor corneas.

After 5 days of culture on CHROMagar, the colony and spore morphologies of the Candida isolates were recorded and compared based on their biofilm-forming abilities (Figure 5). We found that isolates with strong biofilm-forming ability had rough, wrinkled and dry colonies. On Gram staining, these isolates showed spores in pseudohyphal form, and on lactophenol cotton blue staining their spores were uniformly and deeply stained. In contrast, isolates with weak biofilm-forming ability had smooth, moist colonies; on Gram stain their spores appeared as oval yeasts, and on lactophenol cotton blue their spores showed uneven staining intensity.

Our results showed that Candida strains with strong biofilm formation released 1, 3-β-D-glucan at 374.2 ± 295.8 pg/mL, whereas non-biofilm-forming strains released 714.3 ± 447.0 pg/mL (Figure 6). The non-biofilm group released significantly more 1, 3-β-D-glucan than the strong biofilm group (p < 0.05), indicating that non-biofilm-forming Candida strains exhibit higher antigenicity. Moreover, the amount of 1, 3-β-D-glucan released was negatively correlated with the biofilm’s metabolic activity (r = –0.64*).

In the Transwell chemotaxis assay, the control group (no Candida) induced the migration of 4.8×10⁴ ± 8.4×10³ neutrophils. The high biofilm-producing group induced 7.8×10⁵ ± 1.1×10⁵ neutrophils to migrate, while the non-biofilm-producing group induced 8.76×10⁵ ± 3.7×10⁵ neutrophils (Figure 7). In the non-biofilm group, one C. parapsilosis isolate attracted as many as 1.49×10⁶ neutrophils; this strain’s 1, 3-β-D-glucan level was 410.2 pg/mL, without a notably elevated antigenic response.

Planktonic antifungal susceptibility tests were successfully performed on all 45 Candida isolates. However, due to disruption of biofilms during the washing steps for weak biofilm producers, only 18 strong biofilm-forming isolates (15 C. parapsilosis and 3 C. tropicalis) were evaluated for antifungal susceptibility in the biofilm state. Except for C. tropicalis (which had terbinafine, voriconazole, posaconazole, and itraconazole MICs > 16 μg/mL in both planktonic and biofilm states), all Candida species showed markedly higher MICs in the biofilm state than in the planktonic state. Across all isolates, caspofungin maintained low MIC values in both states with no significant differences. The planktonic MIC results for all isolates are presented in Table 3, and the biofilm MIC (BMIC) results for the 18 strong biofilm-forming isolates are presented in Table 4.

Microsatellite multilocus typing results for the 22 C. parapsilosis isolates are shown in Table 5. The 22 isolates were resolved into 18 distinct genotypes. Types I, II, III, and IV each contained two isolates with identical allele profiles, while each of the remaining genotypes was represented by a single isolate, indicating a high genetic diversity among the isolates(Figure 8). Notably, all isolates from the year 2024 were grouped into genotype III or IV, and all of these were high biofilm-producing strains. This suggests a potential clustering trend for C. parapsilosis genotypes in 2024.