Detection and Characterization of Two Phenotypes of Candida parapsilosis in South Korea: Clinical Features and Microbiological Findings

We believe that this is the first study describe the microbiological and molecular characteristics of bloodstream isolates of C. parapsilosis in Korea exhibiting two phenotypes (sinking and floating). An important aspect of our findings is that the sinking phenotype was observed predominantly in isolates harboring a Y132F substitution in the ERG11 gene (92.9%), fluconazole-nonsusceptible (FNS) isolates (86.7%), and clonal BSI isolates (74.4%) of C. parapsilosis. ABSTRACT We newly detected two (sinking and floating) phenotypes of Candida parapsilosis among bloodstream infection (BSI) isolates from Korean hospitals and assessed their microbiological and clinical characteristics. During the performance of a Clinical and Laboratory Standards Institute (CLSI) broth microdilution antifungal susceptibility testing, the sinking phenotype had a characteristic smaller button-like appearance because all yeast cells sank to the bottoms of the CLSI U-shaped round-bottom wells, whereas the floating phenotype comprised dispersed cells. Phenotypic analysis, antifungal susceptibility testing, ERG11 sequencing, microsatellite genotyping, and clinical analysis were performed on C. parapsilosis isolates from 197 patients with BSI at a university hospital during 2006 to 2018. The sinking phenotype was detected in 86.7% (65/75) of the fluconazole-nonsusceptible (FNS) isolates, 92.9% (65/70) of the isolates harboring the Y132F ERG11 gene substitution, and 49.7% (98/197) of all isolates. Clonality was more frequently observed for the Y132F-sinking isolates (84.6% [55/65]) than for all other isolates (26.5% [35/132]; P < 0.0001). Annual incidence of Y132F-sinking isolates increased 4.5-fold after 2014, and two dominant genotypes, persistently recovered for 6 and 10 years, accounted for 69.2% of all Y132F-sinking isolates. Azole breakthrough fungemia (odds ratio [OR], 6.540), admission to the intensive care unit (OR, 5.044), and urinary catheter placement (OR, 6.918) were independent risk factors for BSIs with Y132F-sinking isolates. The Y132F-sinking isolates exhibited fewer pseudohyphae, a higher chitin content, and lower virulence in the Galleria mellonella model than the floating isolates. These long-term results illustrate the increasing BSIs caused by clonal transmission of the Y132F-sinking isolates of C. parapsilosis. IMPORTANCE We believe that this is the first study describe the microbiological and molecular characteristics of bloodstream isolates of C. parapsilosis in Korea exhibiting two phenotypes (sinking and floating). An important aspect of our findings is that the sinking phenotype was observed predominantly in isolates harboring a Y132F substitution in the ERG11 gene (92.9%), fluconazole-nonsusceptible (FNS) isolates (86.7%), and clonal BSI isolates (74.4%) of C. parapsilosis. Although the increase in the prevalence of FNS C. parapsilosis isolates has been a major threat in developing countries, in which the vast majority of candidemia cases are treated with fluconazole, our long-term results show increasing numbers of BSIs caused by clonal transmission of Y132F-sinking isolates of C. parapsilosis in the period with an increased echinocandin use for candidemia treatment in Korea, which suggests that C. parapsilosis isolates with the sinking phenotype continue to be a nosocomial threat in the era of echinocandin therapy.

The clinical implications of phenotypes of C. parapsilosis are also largely unexplored, as are the mechanisms driving the increased clonal transmission of BSIs caused by FNS isolates. We recently detected two phenotypes (sinking and floating) among BSI isolates of C. parapsilosis during the performance of Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) antifungal susceptibility testing. Because two distinct phenotypes of C. parapsilosis isolates had not been described previously, here, we investigated the microbiological and molecular characteristics of C. parapsilosis BSI isolates with these two phenotypes and their role in the clonal transmission of C. parapsilosis BSI isolates harboring the ERG11 mutation Y132F. We assessed all 197 C. parapsilosis BSI isolates collected over a 13-year period (2006 to 2018) at one tertiary-care hospital in South Korea, because FNS isolates of C. parapsilosis were persistently recovered in this hospital over a period of several years, as reported in our previous multicenter study (3).

RESULTS
Detection of the sinking phenotype. The two phenotypes of C. parapsilosis was first detected by visual examination of the CLSI antifungal BMD plate wells after 48 h of incubation (Fig. 1A). Cells with the floating phenotype were dispersed in plate wells like other common Candida species isolates, whereas cells with the sinking phenotype had a smaller, button-like appearance. For each isolate, the phenotypic results obtained from CLSI susceptibility testing against six antifungal agents (fluconazole, voriconazole, amphotericin B, caspofungin, micafungin, and anidulafungin) were the same, as were those in the drug-free positive-control wells and the drug-containing (subinhibitory concentrations of antifungal agents) wells in the plate. All yeast cells with the sinking phenotype sank to the bottom of the tubes, resulting in clear supernatants, but cells with the floating phenotype showed slightly turbid supernatants in tubes (Fig. 1B). When plated on cornmeal-Tween 80 agar, all sinking phenotype isolates produced either no or only a few pseudohyphae, whereas all floating-phenotype cells produced pseudohyphae (Fig. 1C). In contrast to the CLSI microplates, the sinking and floating phenotypes could not be differentiated in the flat-bottom wells of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) BMD microplates (Fig. 1D).
Antifungal susceptibility, phenotyping, and ERG11 sequencing. All isolates of the same clonal microsatellite genotype had the same (sinking or floating) phenotype. Among the clonal isolates, all 60 isolates with SY1-6 genotypes exhibited a sinking phenotype, and they had the Y132F mutation (with the exception of three SY2 and two  ) were found to be a distinct group without close relationships between clonal genotypes of the other countries (Fig. 2B). A comparison of the number of clonal isolates between period A (2006 to 2013) and period B (2014 to 2018) showed that clonal clusters of the FNS (n = 13) and FS (n = 15) isolates occurred at similar low rates during period A but clonal clusters of the FNS isolates (n = 47; mostly Y132F-sinking isolates) were more frequent than those of FS isolates (n = 16) during period B (Fig. 2C). Two dominant genotypes, SY1 and SY2, persistently recovered for 6 and 10 years, respectively, accounted for 69.2% (45/65) of all Y132F-sinking isolates (Fig. 2D). Clinical characteristics. The overall crude 30-day rate of mortality due to C. parapsilosis BSI was 27.4%. The crude 30-day rates of mortality from BSI caused by Y132Fsinking, non-Y132F-sinking, Y132F-floating, and non-Y132F-floating isolates were 32.3%, 21.2%, 20.0%, and 26.6%, respectively; the differences were not significant. Table 3 summarizes the results of a baseline characteristics of patients with C. parapsilosis BSIs and the risk factors for BSIs with Y132F-sinking isolates assessed by uni-and multivariate logistic regression analyses. Several characteristics were more frequently found in the patients with candidemia due to Y132F-sinking isolates, including vasopressor use, multifocal colonization, intensive care unit (ICU) admission, concomitant bacteremia, urine catheter use, central venous catheter (CVC) use, prior antifungal exposure (especially for azole), and breakthrough fungemia (especially for azole). In a multivariate analysis, azole breakthrough fungemia (odds ratio [OR], 6.540), ICU admission (OR, 5.044), and urinary catheter placement (OR, 6.918) were independent risk factors for candidemia caused by Y132F-sinking isolates. Microbiological characteristics and virulence traits. When plated on potato dextrose agar (PDA), cells with a sinking phenotype formed mostly large smooth colonies with round margins, whereas the colony size of those with a floating phenotype varied, with most colonies having irregular margins (Fig. 3A). Electron microscopy of sinkingphenotype isolates revealed spherical cells that were stacked but did not stick to each other and a cell wall that was thicker (median, 209.8 nm; range, 148.9 to 314.3 nm) than that of the floating cells (median, 163.9 nm; range, 131.8 to 186.2 nm) (P , 0.0001) (Fig. 3B). There were no significant differences in the mean biofilm-forming ability among Y132F-sinking, non-Y132F-sinking, and non-Y132F-floating isolates (Fig. 3C). The mean survival rate of Galleria mellonella larvae infected with Y132F-sinking isolates was significantly higher than that of larvae infected with non-Y132F-sinking isolates (log-rank test, hazard ratio [HR] of Y132F-sinking versus non-Y132F-sinking, 0.3477 [95% confidence interval [CI], 0.2268 to 0.5530], P , 0.0001) or non-Y132F-floating isolates (log-rank test, HR of Y132Fsinking versus non-Y132F-floating, 0.2858 [95% CI, 0.1989 to 0.4106], P , 0.0001) (Fig. 3D). When the relative chitin content was measured using a flow-cytometric assay based on calcofluor white staining, the mean (standard deviation [SD]) fluorescence intensity of Y132F-sinking (6.500 6 1.597) and non-Y132F-sinking (5.633 6 0.9114) isolates was significantly higher than non-Y132F-floating isolates (3.600 6 1.457) (Mann-Whitney a For microsatellite typing, each strain was characterized by a genotype resulting from the combination of the sizes of the four markers (CP1, CP4, CP6, and B) (3,14,(23)(24)(25)(26)(27). b Antifungal susceptibility and interpretative categories of resistance were determined using the CLSI M27-4ED broth microdilution method and CLSI M60-ED, respectively (26,27). FR, fluconazole resistant (MIC, $8 mg/mL); F-SDD, fluconazole susceptible-dose dependent (MIC, 4 mg/mL); FS, fluconazole susceptible (MIC, #2 mg/mL). c Genotypes (total of 107 isolates) derived from a single isolate. d Clonal strains were defined by the isolation of an identical microsatellite genotype from more than one patient. test, all sinking versus non-Y132F-floating, P = 0.0003; Y132F-sinking versus non-Y132Ffloating, P = 0.0008; nonY132F-sinking versus non-Y132F-floating, P = 0.0127) (Fig. 3E).

DISCUSSION
Our study is the first to show that long-term clonal transmission of C. parapsilosis BSI isolates harboring a Y132F ERG11 mutation in Korean hospitals involved isolates with sinking phenotypes, which are quite different from those with floating phenotypes. Unlike isolates with the floating phenotype, those with the sinking phenotype were easily detected by their smaller button-like appearance in CLSI antifungal BMD plate wells, because all yeast cells sank to the round bottoms of the U-shaped wells. All isolates of the same microsatellite genotype had the same (sinking or floating) phenotype, suggesting that phenotype may align with the genotype of those isolates. Almost all Y132F-sinking isolates were clonally related, and two dominant clonal genotypes persistently recovered over 6 or 10 years accounted for 69.2% of all Y132F-sinking isolates. Given that the sinking phenotype was observed predominantly in Y132F (91.5%), FNS (86.7%), and clonal (74.4%) BSI isolates of C. parapsilosis and that azole breakthrough fungemia, ICU admission, and urinary catheter use were independent risk factors for fungemia caused by Y132F-sinking isolates, we postulated that strains with a sinking phenotype may be prone to having an ERG11 Y132F substitution after azole exposure, may be more enriched, and may persist for a long time in the ICU environment, where they may cause BSIs in vulnerable patients with indwelling catheters or azole exposure.
Kuhn et al. found that clonal isolates of C. parapsilosis had a higher ability to form biofilms than did unrelated strains, suggesting that biofilm production plays a role in C. parapsilosis outbreaks and provides an index of persistence in the hospital setting (15). In contrast, Thomaz et al. reported a low biofilm-forming ability of all clonal Y132F isolates of C. parapsilosis (7). We did not find any significant differences in biofilm formation between clonally related Y132F-sinking and other nonclonal isolates, indicating that biofilm formation is not a critical causal factor of clonal transmission or outbreaks caused by FNS isolates of C. parapsilosis. However, we did note that clonality was more frequent among sinking (68.4%) than among floating (23.2%) phenotypes of C. parapsilosis. In addition, cells with the sinking phenotype were considerably different from cells with the floating phenotype. The microbiological characteristics of the sinking isolates, including colony morphology and cornmeal agar morphology, resemble those previously reported for smooth-colony isolates (16), but we could not differentiate sinking from floating phenotypes based on colony morphology alone. Our results therefore suggest that the unique characteristics of the sinking phenotype likely enable this yeast to survive in the presence of multiple stressors, such as environmental pressure and antifungal exposure, which plays an important role in C. parapsilosis regarding clonal spread, rather than biofilm formation.
Recent studies have shown that BSIs caused by FR Y132F isolates of C. parapsilosis are associated with a higher 30-day mortality (50.0% or 63.8%) (9,11). In the present study, the Y132F-sinking isolates exhibited both FR (48.2%) and F-SDD (50.8%), but there was no statistical difference in the 30-day mortality of patients with BSIs cause by Y132F-sinking (32.3%) versus by other (24.8%) isolates, which may be partly due to the lower virulence of Y132F-sinking isolates. In the present study, in comparison with floating phenotype isolates, the Y132F-sinking phenotypes of C. parapsilosis formed fewer pseudohyphae, considered a major virulence factor (17), and were less virulent than other isolates in the G. mellonella model, in agreement with a recent report (18). In this study, Y132F-sinking BSI isolates were significantly associated with ICU admission and vasopressor use, although the latter was not an independent risk factor for BSIs caused by Y132F-sinking isolates. Considering their lower virulence, Y132F-sinking isolates may not frequently cause severe fungemia, which results in hemodynamic instability and vasopressor use in ICU patients. Alternatively, the long-term persistence of Y132F-sinking strains in the hospital environment may cause BSIs in vulnerable patients on vasopressor therapy in ICUs, where vasopressors are typically administered via a CVC (19). The increase in the prevalence of FR C. parapsilosis isolates has been a major threat in developing countries, in which the vast majority of candidemia cases are treated with fluconazole (1). However, a recent Korean single-center study showed a significant increase in FR C. parapsilosis candidemia in the period with an increased echinocandin use for candidemia treatment (35.3% versus 0.0%, respectively) after the approval of echinocandin  (20). Similarly, we also found increases in FNS (4.7-fold), Y132F (5.1-fold), sinking (3.6-fold), and Y132F-sinking (4.6-fold) isolates of C. parapsilosis after 2014, which were paralleled by an increase in the number of candidemia patients who received echinocandin therapy from 2014 at Severance Hospital, as reported in a previous study (21). All of these data suggest that Y132F-sinking isolates of C. parapsilosis continue to be a nosocomial threat in the era of echinocandin therapy. Unlike other common Candida species, C. parapsilosis has a naturally occurring amino acid substitution, P660A, in the FKS1 hot spot 1 region, which might reduce susceptibility to echinocandins in vitro (17). In this study, sinking isolates had significantly higher chitin contents than floating isolates. In addition, the walls of sinking isolates were thicker than those of floating isolates, although electron micrographs of only a few isolates were obtained. The thicker wall and elevated wall chitin content have been found in Candida albicans isolates with FKS hot spot mutations, which show reduced fitness and attenuated pathogenicity (22,23). Candida populations that survive echinocandin exposure develop tolerance to Electron microscopic images of sinking phenotype isolates reveal many spherical cells without pseudohyphae and a thicker cell wall compared with floating phenotype isolates. (C) Mean biofilm-forming abilities, assessed using the XTT reduction assay, were not significantly different among the three isolate groups. OD, optical density. (D) The mean survival rate of larvae infected with Y132F-sinking isolates was significantly higher than that of larvae infected with non-Y132F-sinking or non-Y132F-floating isolates. (E) Mean chitin contents of Y132F-sinking and non-Y132F-sinking isolates were significantly higher than that of floating isolates, as determined in a flow cytometric assay based on calcofluor white staining.
FNS C. parapsilosis with Sinking Phenotype Microbiology Spectrum echinocandin agents and have a high chitin content in the cell wall (24,25). Further studies are needed to elucidate the factors contributing to the recent rise in incidence of BSIs due to Y132F-sinking isolates of C. parapsilosis in Korea and to evaluate whether the selection of Y132F-sinking subpopulations of C. parapsilosis coincides with echinocandin introduction. The present study demonstrates that the same clonal Y132F-sinking isolates were persistently recovered in two hospitals (Severance Hospital and SMC) in South Korea over several years, which was consistent with our previous multicenter study (3). A recent study performed at the SMC demonstrated the continued evolution of azole resistance mechanisms (MRR gene mutation) and thus increasing fluconazole MICs over time in clonal Y132F isolates (14). Further increases in the fluconazole MICs of these clonal Y132F-sinking isolates may pose significant challenges to antifungal management, with an increasing rate of azole breakthrough fungemia. Therefore, we suggest that the detection of the sinking phenotype of FNS C. parapsilosis BSI isolates serves as an indicator of ongoing undetected long-term clonal transfer of Y132F-sinking isolates within a hospital.

MATERIALS AND METHODS
Fungal isolates, antifungal susceptibility testing, phenotyping, and molecular analysis. From January 2006 to December 2018, 197 nonduplicate C. parapsilosis BSI isolates were collected at Severance Hospital (2,400 beds; Seoul, South Korea). The Candida isolates were identified to the species level using matrix-assisted laser desorption ionization-time of flight mass spectrometry and/or molecular methods (3,4). CLSI M27-A3 BMD tests for susceptibility to fluconazole, voriconazole, amphotericin B, caspofungin, micafungin, and anidulafungin were performed, and the interpretative guidelines in the CLSI M60-ED1 document were used to classify the isolates based on clinical breakpoints (26,27). Two phenotypes (sinking and floating) were determined by visually examining the CLSI BMD plate wells after incubating the isolates, as well as by assessing the morphology of the yeasts on cornmeal-Tween 80 agar. ERG11 was sequenced for all isolates as described previously (3). Microsatellite typing using four markers (CP1, CP4, CP6, and B) was performed for genotype determination also as described previously (3,14,(28)(29)(30)(31)(32). Clonal strains were defined by the isolation of an identical microsatellite genotype from more than one patient. We further evaluated the genetic relationships of genotypes of clonal strains from our cohort with 35 genotypes of clonal strains available in the previous publications (14,(28)(29)(30)(31)(32). Microsatellite dendrograms were constructed using the unweighted pair-group method with arithmetic mean (UPGMA), as described elsewhere (33).
Clinical data collection. The collected data included patient demographics, comorbidities, severity of infection, clinical status at the time of positive culture, and therapeutic measures (34)(35)(36). Breakthrough fungemia was the development of candidemia during antifungal therapy (34)(35)(36)(37). This study was approved by the Institutional Review Board of Yonsei University College of Medicine (YUHS 4-2019-0970).
Statistical analyses. SPSS 27.0 (IBM Corp., Armonk, NY, USA) was used for the statistical analysis and Prism 9.3.1 (GraphPad Software Inc., San Diego, CA, USA) for the graphical work. Univariate analyses were based on the chi-squared or Fisher's exact test, as appropriate, for discrete variables. The predictive factors for C. parapsilosis fungemia caused by Y132F-sinking isolates were analyzed using univariate and multivariate logistic regression analyses. Potential predictive factors in the univariable comparisons (P , 0.1) were included in an initial multivariable regression model. P values of ,0.05 were considered statistically significant.