Comparative Pathogenicity of United Kingdom Isolates of the Emerging Pathogen Candida auris and Other Key Pathogenic Candida Species

The incidence of invasive candidiasis, which includes candidemia and deep tissue infections, continues to rise and is associated with considerable mortality rates. Candida albicans remains the most common cause of invasive candidiasis, although the prevalence of non-albicans species has increased over recent years. Since its first description in 2009, Candida auris has emerged as a serious nosocomial health risk, with widespread outbreaks in numerous hospitals worldwide. However, despite receiving considerable attention, little is known concerning the pathogenicity of this emerging fungal pathogen. Here, using the Galleria mellonella insect systemic infection model, we show strain-specific differences in the virulence of C. auris, with the most virulent isolates exhibiting pathogenicity comparable to that of C. albicans, which is currently accepted as the most pathogenic member of the genus.

The first 2 United Kingdom isolates of C. auris were received at the UK National Mycology Reference Laboratory (MRL) in 2013, from blood cultures from 2 unrelated patients in distant geographical localities (MRL unpublished data). Since 2013, we have received a further 19 isolates from at least 6 different hospitals, including 14 isolates suspected of being part of an outbreak. Here we have compared the pathogenicities of 12 United Kingdom isolates of C. auris from 6 different referring National Health Service (NHS) hospitals with the pathogenicities of equivalent isolates of other common pathogenic Candida species, using the Galleria mellonella insect systemic infection model.

RESULTS AND DISCUSSION
The characteristics of the 12 isolates of C. auris employed in the current study are detailed in Table 1, with antifungal MIC values determined at the MRL. Initial attempts to generate suspensions of C. auris isolates in phosphate-buffered saline (PBS) for larval inoculation revealed striking strain-specific differences in phenotypic behavior. While most isolates readily formed homogeneous suspensions upon thorough vortex mixing, the resulting suspensions seen with 4 independent isolates from 3 different referring hospitals were grossly particulate and contained individual yeast cells mixed with large aggregations ("aggregate" strains) ( Table 1 and Fig. 1). For these 4 isolates, aggregates could not be physically disrupted by vigorous vortex mixing or by detergent treatments (data not shown). Since the aggregates were too large to permit larval inoculation and since cell numbers within the aggregates could not be accurately quantified, homogeneous suspensions were instead achieved by allowing initial suspensions to settle for 10 min, followed by removal of the supernatant containing individual yeast cells that had remained in suspension and adjustment of these individual cells to the appropriate concentration for injection into larvae.
In agreement with previous reports (10,24), the pathogenicity of the common Candida species at 37°C in G. mellonella was directly related to the ability of individual species to produce hyphal filaments or pseudohyphae ( Fig. 2; see also Fig. S1 in the supplemental material), with very little strain-to-strain variation in virulence within each species (see Fig. S1). Thus, C. albicans and C. tropicalis exhibited greater virulence than C. lusitaniae, C. guilliermondii, and members of the C. parapsilosis species complex, and virtually no larval killing was induced by those organisms that form only rudimentary pseudohyphae or no pseudohyphae (C. glabrata, C. nivariensis, C. krusei, C. kefyr, C. bracarensis, and Saccharomyces cerevisiae) ( Fig. 2; see Table 2 for full statistical analyses).
Strikingly, despite most reports suggesting that C. auris does not form significant pseudohyphae in vitro (14,15,21), C. auris strains exhibited virulence in G. mellonella that was significantly higher (in terms of the kinetics of larval death and the number of larvae killed) than that seen with most other common pathogenic yeast species, with overall pathogenicity approaching that observed with C. albicans and C. tropicalis isolates ( Fig. 2 and Table 2). Dissection of representative larvae that had been inoculated with the various strains and incubated at 37°C for 18 h revealed significant hyphal proliferation in hemolymph form larvae inoculated with C. albicans (Fig. 3A). However, no hyphal or pseudohyphal formation was observed in larvae infected with any C. auris strains at 18 h or any time postinfection (Fig. 3B to D). Interestingly, in larvae that had received nonaggregating strains of C. auris, larval dissection revealed large numbers of individual budding yeast cells, including in phagocytic cells ( Fig. 3B and E). However, in larvae inoculated with individual yeast cells prepared from aggregate-forming strains of C. auris, hemolymph contained large aggregates of C. auris cells, with few individual yeast cells, indicating that the ability to produce large aggregates had been maintained in vivo (Fig. 3C and E). In the light of this differential behavior of C. auris isolates in G. mellonella, further experiments compared larval killing with aggregate-forming versus non-aggregate-forming strains, with larvae incubated at both 30°C and 37°C. Strikingly, nonaggregate strains exhibited significantly greater virulence than aggregate-forming strains at both temperatures ( Fig. 4 and Table 2) (P ϭ 0.02), with nonaggregate isolates showing virulence that was indistinguishable from that of C. albicans strains at 37°C (Fig. 4).
In the current report, we present for the first time a comparative study of the pathogenicities of isolates of Candida auris and those of other common pathogenic Candida species and the somewhat surprising finding that C. auris virulence is comparable to that seen with C. albicans in the invertebrate G. mellonella model, despite the fact that C. auris isolates do not undergo significant filamentation in this model organism. This finding is all the more striking since C. auris yeast cells are  more comparable in size and growth rate to C. glabrata than to C. albicans ( Fig. 2  and data not shown). Moreover, we have demonstrated the novel finding that certain C. auris isolates form large aggregates of cells both in vitro and in vivo in inoculated larvae, even when larvae were inoculated with individual cells prepared from aggregating isolates. Microscopic examination of these aggregates suggests that they form due to reduced daughter cell liberation after budding (see, for example, Fig. 3C), rather than due to flocculation of individual budding cells. This contention would certainly be supported by our inability to disrupt the aggregates with intense vortex mixing and detergent treatments. In G. mellonella, aggregateforming strains exhibit less virulence than those strains that exist as single budding cells. Further studies will be required to determine if aggregate-forming strains produce less dissemination during infections in humans or, conversely, whether the ability to form large aggregates protects those strains against phagocytic attack or the effects of antifungal agents or detergents used to clean hospital environments.

MATERIALS AND METHODS
Fungal strains. All C. auris isolates were identified by ribosomal DNA (rDNA) gene sequencing targeting the 28S rRNA or by internal transcribed spacer 1 (ITS1) regions and matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) analysis or by a combination of the two methods exactly as described previously (25). For the other Candida species included for comparison, where possible, clinical isolates were from deep-seated infections. Identity to the species level was confirmed by sequencing or MALDI-TOF analysis in all cases. Killing assays in G. mellonella. Killing assays were performed in Galleria mellonella exactly as described previously (10), using final (sixth) instar larvae (Livefood UK Ltd., Rooks bridge, Somerset, United Kingdom) weighing approximately 300 mg each that were free of gray markings and that had been maintained at room temperature in the dark and inoculated within 48 h of receipt. Suspensions of individual Candida isolates that had been grown on Sabouraud's agar for 24 h at 37°C were harvested by gentle scraping of colony surfaces with sterile plastic loops, washed twice in sterile PBS, counted in hemocytometers, and adjusted to 10 5 cells/l in sterile PBS. Individual larvae were inoculated in the left rear proleg with 1 ϫ 10 6 yeast cells-PBS (final inoculum volume, 10 l) using a 10-l Hamilton syringe fitted with a 26-gauge blunt needle. At least 10 larvae were inoculated per isolate per experiment (experiments employed 4 independent isolates of each Candida test species [12 isolates in the case of C. auris]). Control groups of larvae received 10 l of sterile PBS in exactly the same manner. Inoculated larvae were incubated at 30°C or 37°C and scored for viability at 8-h intervals as described previously (10). Differences in resulting Kaplan-Meier survival plots were evaluated using the Mann-Whitney (two-sample Wilcoxon) test. In some experiments, fungal cell filamentation postinfection was assessed by sacrificing representative larvae from each inoculum group at 24 h postinfection and aseptic collection of the fat body/solid internal structures and hemolymph followed by microscopic examination (10).
Antifungal susceptibility testing of C. auris isolates. Broth microdilution determination of yeast MICs was performed according to CLSI method M27-A3 (26) in round-bottomed 96-well plates with yeast blastospore suspensions prepared in saline solution and then diluted into RPMI 1640 and adjusted to a final concentration of 2.5 ϫ 10 3 CFU/ml. Inoculated plates were incubated for 24 to 48 h at 35°C. MICs were read at 24 and 48 h as the concentration of drug that elicited 100% inhibition of growth (amphotericin B) or significant (approximately 50%) inhibition of growth compared with that of a drug-free control (fluconazole, voriconazole, posaconazole, anidulafungin, and flucytosine).

ACKNOWLEDGMENT
We are grateful to the other members of the United Kingdom MRL for their assistance with data collation and phenotypic and molecular analyses of isolates.