Candida duobushaemulonii: An Old But Unreported Pathogen

Candidiasis caused by species of the Candida haemulonii complex (Candida haemulonii and Candida duobushaemulonii) and closely related species, Candida auris and Candida pseudohaemulonii are increasing. These species often show reduced susceptibility to antifungal drugs, such as azoles and amphotericin B or, less frequently, echinocandins. However, conventional phenotypic identification methods are unable to accurately differentiate these species and, therefore, their prevalence may have been underestimated. In this study, 150 isolates that were probably misidentified were reanalyzed using two novel PCR approaches. We found that one isolate previously identified in 1996 as Candida intermedia was C. duobushaemulonii, being one of the oldest isolates of this species described to date. We also found that this isolate had reduced susceptibility to fluconazole, itraconazole, and amphotericin B.


Introduction
The genus Candida is demonstrated to be genetically heterogeneous. For that reason, many Candida species belong to a complex formed by various cryptic species that are phenotypically identical, and only molecular-based methods guarantee their accurate identification [1]. This is the case for Candida haemulonii, an opportunistic pathogen whose first clinical isolation, from a blood sample, was described in 1984 [2]. Prior to 2012, C. haemulonii was separated into two genetically distinct groups based on isoenzyme and protein profiles and DNA relatedness studies [3]. However, in the same year, as further studies revealed the heterogeneity in the DNA sequence of various genes, the complex was reclassified as two species (C. haemulonii sensu stricto and Candida duobushaemulonii) and one variety (C. haemulonii var. vulnera) [4]. Concurrently, two new species phylogenetically close to C. haemulonii complex, namely, Candida pseudohaemulonii and Candida auris, were described in 2006 and 2009, respectively [5,6].
Although infections caused by this complex of species remain rare, since they are not usually involved in outbreaks but in sporadic infections [7][8][9], they are increasingly being isolated worldwide and are reported as important fungi in clinical settings [10][11][12][13]. However, the case of C. auris stands out. Since its first description in 2009 in Japan, this fungus has been identified worldwide in a short

Sample Preparation and DNA Extraction
Both clinical isolates and the reference strain were cultured on Sabouraud dextrose agar (SDA) or yeast extract peptone dextrose (YEPD) medium at 37 • C during 24-48 h. Then, DNA extraction was performed using the DNeasy ® UltraClean ® Microbial Kit (QIAGEN, Hilden, Germany) following the manufacturer's instructions. DNA purity and quantity were assessed using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

PCR-Based Molecular Identification
The 150 isolates and the four positive controls were analyzed by two different PCR techniques, using primers and amplification conditions described by Ruiz-Gaitan et al. [35] and Arastehfar et al. [32] ( Table 1). One was a duplex C. auris-specific PCR that also amplifies the 5.8S rDNA gene as quality control band [35] and the other was a C. haemulonii complex-tetraplex PCR [32] (Figure 1).

ITS Region Sequencing and Species Identification
When robust amplification in any of the PCR or low-intensity bands were detected, sequencing of the ITS1-5.8S-ITS2 gene was performed for confirmation. This region was amplified with ITS1 and ITS4 primers as described previously [36]. Unpurified PCR products were sent to the Advanced Research Facilities of the University of the Basque Country (SGIker), where samples were purified and sequenced. Received sequences from both DNA strands were analyzed by Chromas v. 2.6.4 (Technelysium Pty Ltd., South Brisbane, Australia) and BioEdit v. 7.2.6.1 (Tom Hall, Raleigh, NC, USA), and sequence identity for species identification was determined using the Basic Local Alignment Search Tool (BLAST) at the NCBI database (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Phenotypical Characterization
All the isolates belonging to the C. haemulonii complex (n = 13) were plated on CHROMagar Candida medium supplemented with Pal's medium at both 37 • C and 42 • C during 24-48 h, in order to assess growth ability and colony characteristics [37].

Molecular Identification
Except for the two confirmed C. auris clinical isolates, no other C. auris was found using the C. auris-specific PCR. Nevertheless, as the quality control band of 5/8 C. rugosa, 7/8 C. lipolytica, 1/2 C. magnoliae, and 1/8 C. intermedia was absent, ITS sequencing was performed for these isolates. The identification of the seven C. lipolytica and the only C. magnoliae isolate was confirmed. On the contrary, the five C. rugosa and the only C. intermedia were all re-identified as Candida pararugosa ( Table 2). The C. haemulonii complex-tetraplex PCR method accurately identified the four isolates used as positive controls. Neither C. auris nor C. pseudohaemulonii were detected in the collection and the eight isolates previously identified as C. haemulonii were confirmed by the tetraplex PCR, as they all produced a 696 bp sized band. Interestingly, one isolate identified in 1996 as C. intermedia by API ® ID 32C produced a 115 bp sized band, corresponding to C. duobushaemulonii. Additionally, we re-analyzed this isolate by the current API ® ID 32C panel, which identified it as C. sake (87.4%, acceptable profile). The eight C. haemulonii isolates, the new C. duobushaemulonii, as well as the four positive controls, were all confirmed by ITS sequencing showing a percentage of identity with the reference strains higher than 99%.

Fungal Susceptibility
Although the C. duobushaemulonii isolate that was used as positive control showed susceptibility to all azole agents, the re-identified C. duobushaemulonii isolate was less susceptible to itraconazole (MIC = 0.5 µg/mL) and fluconazole (MIC = 32 µg/mL). Six C. haemulonii isolates also showed a lower susceptibility to itraconazole (MIC = 0.5 µg/mL) and fluconazole (MIC = 16-32 µg/mL), other one was inhibited by 1 µg/mL of itraconazole, and another one by 64 µg/mL of fluconazole, and another showed reduced susceptibility to both agents (MIC = 1 and 64 µg/mL, respectively, of itraconazole and fluconazole). Both C. auris isolates were less susceptible to fluconazole (MIC > 256 µg/mL) and itraconazole (MIC = 0.25 µg/mL). One of them also showed decreased susceptibility to voriconazole (MIC = 2 µg/mL). With regard to amphotericin B, all but one C. haemulonii isolate showed reduced susceptibility (MIC > 2 µg/mL). Finally, no reduced susceptibility was found to either echinocandins or 5-flucytosine (Table 2).

Phenotypic Characteristics
When grown on chromogenic agar, all but the C. pseudohaemulonii reference strain, whose colonies were purple, showed smooth, matte, and light-pink colored colonies (Figure 2). When cultured at 42 • C, only the two C. auris isolates were able to grow.

Discussion
Although Candida albicans remains the most often isolated species, a remarkable shift in the etiology of candidiasis is ongoing, as previously uncommon non-C. albicans species are arising or increasing in prevalence [38]. Part of this increase in prevalence of uncommon species of Candida might be due to the broadened use of molecular methods, such as PCR and MALDI-TOF for identification of clinical isolates. Examples of such pathogens are species related to the C. haemulonii complex, whose frequency has increased over the years. Between 1997 and 2007, the prevalence of C. haemulonii was very low (<0.01%), as observed by the ARTEMIS DISK Global Surveillance Study [39]. In a different retrospective study that was carried out between 2001 and 2005, this species was the sixth most common Candida species isolated (1.5%). Ramos et al. reported that C. haemulonii was the fourth most common species (12.1%) causing cutaneous candidiasis in Brazil from 2008 to 2009 [10]. A study in India reported that the prevalence of this species increased from 5.5% in 2006 to 18.2% in 2008 [40]. However, the identifications in those studies were performed using commercially available biochemical methods, such as API, MicroScan, and Vitek, which have been proved to misidentify these species. In addition, C. duobushaemulonii was not described until 2012, when Cendejas-Bueno et al. established C. haemulonii as a complex of species [4]. Therefore, whether candidiasis caused by these species is relatively new, as suggested by Frías-de-León et al. [13], or longstanding but unidentified remains unknown. Our results support this latter idea. We carried out a retrospective study in which we surveyed for species related to this complex in a culture collection that contains clinical isolates from 1993 to 2014 using novel molecular methods. From that year onwards, all the isolates with questionable identifications were sequenced. Interestingly, we discovered that one isolate that was identified in 1996 as C. intermedia by API ® ID 32C in fact was C. duobushaemulonii, and this type of misidentification has not been described before. In accordance with literature, this result shows the inaccuracy of biochemical methods towards correct identification of some infrequent Candida species and, consequently, that the real clinical incidence of the C. haemulonii complex might have been underestimated. For instance, Ramos et al. found that four C. haemulonii isolates that were collected between 2009 and 2013 in fact were C. duobushaemulonii [10,23].
Moreover, we show that the current API ® ID 32C panel is still unable to identify this species, which reinforces the importance of using molecular methods or MALDI-TOF to detect pathogens related to the complex. The PCR proposed by Ruiz-Gaitán et al. [31] that was tested in our study accurately identified both C. auris isolates and did not yield any C. auris-specific amplicon with other tested species. The 5.8S rDNA control amplicon was not produced with C. pararugosa, C. magnoliae, and C. lipolytica isolates. Accordingly, Ruiz-Gaitán et al. already reported that these control primers do not yield amplicons and therefore cannot be used in more distinct species such as C. lipolytica [31], which we confirmed in our study. However, as there are no studies of reported misidentifications of C. auris with those three species and their incidence of causing candidiasis is utterly low, we believe this PCR design is a powerful approach for clinicians to effectively detect C. auris infections. Likewise, the tetraplex-PCR proposed by Arastehfar et al. [32] is also an accurate and useful molecular technique. This was also demonstrated by Frias-de-León et al. [13], who showed correct identification of four C. auris control isolates, eight C. haemulonii isolates, and two C. duobushaemulonii, whereas no other isolate yielded any amplicon.
The re-identified C. duobushaemulonii isolate in the present study was obtained in 1996 from a toenail of a patient from Bizkaia (Spain), while the eight C. haemulonii isolates had been obtained in 2000 from various sources, including both invasive and superficial infections. Our results support the idea that species of this complex were primarily associated to superficial infections, prior to its emergence related to invasive candidiasis. Even though C. haemulonii was first clinically isolated in 1984 from the blood of a patient [2], most of the invasive infections caused by species belonging to this complex were reported in more recent years. In 1991, Gargeya et al. reported onychomycoses cases in the USA caused by C. haemulonii [41], and the five isolates used by Lehmann et al. in 1993 were also obtained from skin and nails [3]. In the study carried out by Ramos et al., all the isolates obtained from non-invasive sources (nails and skin) were collected prior to 2009, whereas all isolates collected between 2010 and 2013 originated from invasive sites [10]. Likewise, Hou et al. collected 31 clinical isolates (26 C. haemulonii and five C. duobushaemulonii) from 2010 to 2014, which were all recovered from invasive sources, mainly from blood and cerebrospinal fluid [22]. Although Lehmann et al. had already described in 1993 that C. haemulonii was a complex of two genetically different groups (I and II) [3], our re-identified isolate, together with the first isolate from a foot ulcer in 1990 [42] are, to our knowledge, the oldest C. duobushaemulonii isolates found. This underlines the importance of our study to understand the origin and spread of cryptic emergent species.
The C. duobushaemulonii isolate re-identified in our study, the eight C. haemulonii isolates and the two control C. auris isolates had reduced susceptibility to fluconazole and itraconazole, whereas seven C. haemulonii isolates had also reduced susceptibility to amphotericin B. This concurs with other studies, as clinical isolates of this complex usually show in vitro reduced susceptibility or resistance to these antifungal drugs [4,5,10,[16][17][18][19], often being associated with clinical failure [4,10,17]. Recently, a potential mechanism of resistance among the species of C. haemulonii complex to amphotericin B has been described. The membrane of the species of the complex is composed mainly of ergosterol pathway intermediates, thus there is no target for amphotericin B [43]. As all clinical isolates were susceptible to echinocandins, these antifungal drugs might be the best treatment option for infections caused by species related to the complex, as was also suggested elsewhere [12,17]. Nevertheless, the tendency of the complex to show resistance to azoles, echinocandins, and amphotericin B reinforces the necessity to accurately identify these uncommon species.
Finally, morphological characteristics such as colony color are not very useful for the identification of these species. Although in our study C. haemulonii, C. auris, and C. duobushaemulonii isolates showed smooth, light-pink colored colonies and the C. pseudohaemulonii isolate purple colonies on CHROMagar, other authors have reported that C. haemulonii complex species and C. auris can also show a dark pink color [21]. On the other hand, the ability of only C. auris isolates to grow well at 42 • C, which has also been described before [10,18,37], could comprise a helpful characteristic to differentiate C. auris from the other species.

Conclusions
Species of the C. haemulonii complex are emerging invasive pathogens that often show resistance to commonly used antifungal drugs. Therefore, their accurate and quick identification using novel molecular methods is essential as they can cause challenging invasive infections that pose difficulties in their therapeutic management. Although initially described as etiological agents of superficial candidiasis, surveillance should be conducted to clarify the mechanisms involved in developing invasive candidiasis.