Clonal Diversity of Candida auris, Candida blankii, and Kodamaea ohmeri Isolated from Septicemia and Otomycosis in Bangladesh as Determined by Multilocus Sequence Typing

Candida auris, Candida blankii, and Kodamaea ohmeri have been regarded as emerging fungal pathogens that can cause infections with high mortality. For genotyping of C. auris, a multilocus sequence typing (MLST) scheme based on four locus sequences has been reported, while there is no typing scheme for C. blankii and K. ohmeri. In the present study, the existing MLST scheme of C. auris was modified by adding more locus types deduced from sequence data available in the GenBank database. Furthermore, MLST schemes of C. blankii and K. ohmeri were developed using the four cognate loci (ITS, RPB1, RPB2, D1/D2) and similar sequence regions to those of C. auris. These MLST schemes were applied to identify the ST (sequence type) of clinical isolates of C. auris (n = 7), C. blankii (n = 9), and K. ohmeri (n = 6), derived from septicemia or otomycosis in Bangladesh in 2021. All the C. auris isolates were classified into a single ST (ST5) and clade I, having a Y132F substitution in ERG11p, which is associated with azole resistance. Similarly, all the C. blankii isolates belonged to a single type (ST1). In contrast, six K. ohmeri isolates were assigned to five types (ST1-ST5), suggesting its higher genetic diversity. These findings revealed the availability of MLST schemes for these three fungal species for understanding their clonal diversity among clinical isolates.


Introduction
Candida species are the most common fungal pathogens in humans, causing superficial and systemic infections. Among the forms of candidiasis, candidemia is among the leading causes of nosocomial infections, associated with significant mortality, and its incidence has been increasing worldwide in the past few decades [1]. A notable recent trend in epidemiology of Candida is the emergence of novel Candida species such as C. auris [2], and the prevalence of uncommon pathogenic yeast [3], which has been suggested to be due mostly to an increase in immunocompromised patients.
Candida auris is a newly described species that was first isolated from the external ear canal of a patient in Japan in 2009 [4]. Subsequently, the global spread of this species has occurred rapidly [5,6]. However, it has also been noted that the genetically different clades (I through V) of C. auris emerged nearly simultaneously in multiple countries across six continents, for an unknown reason [7]. Bloodstream infection is the most common type of infection by C. auris in children, with a high mortality rate [6]. This species shows high resistance to fluconazole, with variable susceptibility to other antifungals and an increase in multi-drug resistance (MDR) [5,6]. The spread of C. auris with MDR has been recognized as a public health concern in Europe [8] and the Middle East [9], and its sudden increase in prevalence in 2021 was reported in the US [10].
Candida blankii, which was first described in mink in 1968 [11], has been isolated from dairy products. However, since its identification in a neonate with cystic fibrosis in 2015 [12], C. blankii has been considered an emerging human pathogen that primarily infects immunocompromised humans, mostly causing bloodstream infections [3,13,14]. Reduced susceptibility to azoles in this species was reported [15], and an outbreak of fungemia was described in India [16]. Recently, C. blankii bloodstream infection in the elderly was reported as a complication of COVID-19 in the US [17].
Kodamaea ohmeri, a heterotypic synonym of C. guilliermondii var. membranaefaciens, is also regarded as an emerging fungal pathogen, primarily causing fungemia, endocarditis, and onychomycosis [18,19]. Although amphotericin B and fluconazole are commonly used as antifungal therapy, the occurrence of fluconazole resistance is noted [19,20]. Mortality due to K. ohmeri infection is high in cases of fungemia, but low in other infection types. Infections with this species have been found worldwide, with nearly 70% being reported in Asia (particularly East and Southeast Asia) [19].
In Bangladesh, candidemia is estimated to constitute a serious burden, with an estimated 8100 cases per year (5/100,000 rate) [21]. We previously characterized the species and antimicrobial susceptibility of 109 presumptive Candida isolates from 3 infection types in Mymensingh, Bangladesh, and revealed 13 species, including C. auris and K. ohmeri [22]. We describe herein the detection of more isolates of C. auris and K. ohmeri, and the identification of C. blankii, firstly in Bangladesh, among isolates at the same study site. Furthermore, to clarify the clonal diversity, isolates of the three emerging fungal species were analyzed using multilocus sequence typing (MLST), based on partial nucleotide sequences of four housekeeping genes.
An MLST scheme of C. auris using four loci (ITS, RPB1, RPB2, and D1/D2) of housekeeping genes was originally proposed by Cendejas-Bueno and coworkers [23]. Using these sequences of the four loci, Prakash and coworkers reported a multilocus phylogenetic analysis of C. auris [24]. Subsequently, employing the sequence data of the same four loci, Kwon et al. [25] designed an MLST scheme in which the locus type was assigned to the sequences, and the sequence type (ST) was assigned to the combination of the four locus types (allelic profiles). This scheme was utilized for 61 Korean isolates, revealing a single ST (ST cluster) among them. However, this MLST analysis for C. auris has not yet been well utilized, and has not been reported for more clinical isolates at other study sites. Furthermore, no MLST scheme has been established for C. blankii and K. ohmeri. In the present study, the MLST scheme of C. auris was improved through the addition of more types of four loci, the sequences of which could be retrieved from the GenBank database. For C. blankii and K. ohmeri, provisional MLST schemes were created, based on the four loci, which are same as those of C. auris, using the sequence data retrieved from the GenBank database. These MLST schemes were applied for genotyping recent clinical isolates of C. auris, C. blankii and K. ohmeri in Bangladesh.

Clinical Isolates of C. auris, C. blankii, and K. ohmeri
In our previous study, clinical isolates of presumptive Candida spp. collected from Mymensingh Medical College hospital, during 10-month period in 2021, were characterized [22]. In the present study, we analyzed additional 22 isolates that were collected in the same period but could not be identified phenotypically, and by the PCR-RFLP method for the ITS region. These isolates were recovered from the blood and aural swabs (discharge) of patients with suspected candidemia and otomycosis, respectively. Isolation of Candida from specimens was performed as described previously [22].
The nucleotide sequence of the ITS region was determined using Sanger sequencing, with PCR products and Candida species universal primers ITS1 and ITS4 [26]. Species were determined by searching for an identical or highly similar sequence in the GenBank database, using the BLAST web tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed on 20 April 2023). Isolates identified as C. auris, C. blankii, and K. ohmeri were further genotyped using MLST schemes.

Antifungal Susceptibility Testing
Susceptibility to antifungal agents was determined using the broth microdilution method, as described previously [22]. The minimum inhibitory concentration (MIC) was determined for fluconazole and amphotericin B. The susceptibility of C. auris to these antifungals was judged according to the criteria defined by the US Centers for Disease Prevention and Control (CDC) (https://www.cdc.gov/fungal/candida-auris/c-auris-antifungal.html, accessed on 26 May 2023).

MLST Schemes of C. auris, C. blankii and K. ohmeri
MLST was performed according to the procedure described previously by Kwon et al. [25]. In this scheme, partial nucleotide sequences are determined for four loci: the ITS region of the ribosomal subunit, the D1/D2 large ribosomal subunit region, RPB1 encoding the largest subunit of RNA polymerase II, and RPB2 encoding the DNAdependent RNA polymerase II. Fungal DNA was extracted using the phenol-chloroform method, and the ITS, D1/D2 regions and RPB1 sequences (i.e., the same sequence regions among the three species) were amplified by PCR, using the primers and conditions described previously [24] (Table 1). For PCR amplification of RPB2, newly designed primers were used for the three species (Table 1). Nucleotide sequences of amplified PCR products were determined via Sanger sequencing on an automated sequencer. TCTCCCACTCGATTTCTGCT [-] For C. auris, obtained sequence data were collated with those published previously [25] to assign type of each locus. Furthermore, for all the three fungal species, sequences repre-senting different types of the four loci were searched using the BLAST web tool, utilizing the obtained sequence data for the present study isolates. The sequence data of individual allelic types were compared through alignment using the Clustal Omega program (https://www.ebi.ac.uk/Tools/msa/clustalo/, accessed on 10 April 2023). Phylogenetic dendrograms of the sequences of four MLST loci were constructed using the maximum likelihood method and MEGA.6 software, together with sequence data available in the GenBank database. ST (sequence type) was assigned to the combination of the four locus types, i.e., the allelic profile.

Sequence Analysis of ERG11 Gene of C. auris and Clade Typing
The nucleotide sequence of C. auris ERG11 gene was determined as described previously [27]. To detect mutation associated with azole resistance, deduced amino acid sequences were compared with those of the fluconazole-susceptible C. auris strain B11220 (a wild-type strain of clade II, GenBank accession no. XP_028891800) [28]. The major geographical clade of C. auris (I-IV) was determined using clade-specific PCR, as described previously [29].

GenBank Accession Number
The nucleotide sequences of ITS and D1/D2 regions, RPB1 and RPB2 determined in the present study were deposited in the GenBank database under the accession numbers shown in Table S1.

MLST Scheme of C. auris
The nucleotide sequences of the four MLST loci were determined for C. auris isolates, and the sequence ranges were adjusted according to those reported previously [25]. Sequences of ITS, RPB1, D1/D2 of all the seven isolates were identical to those of allelic types "b", "b", "b" designated by Kwon et al. [25], respectively. The RPB2 sequence was identical among the seven isolates, but did not coincide with those of the alleles "a" or "c", as described previously [25]. In the present study, the allelic designations using the alphabet (a-c) were converted to Arabic numerals, because the allelic profiles of all the established MLST schemes of the bacterial species and four Candida species are shown as combinations of Arabic numerals (https://pubmlst.org/, accessed on 10 April 2023). Accordingly, four alleles (ITS-RPB1-RPB2-D1/D2) of the present isolates were assigned to 1-1-1-1 (Table 2), and the sequences of the four loci are shown in Tables S2-S5. To assign more allelic types for the C. auris MLST scheme, divergent sequences of the four loci were searched among the GenBank database using the BLAST web tool and the allelic type 1 sequence of the four loci as standards. Allelic type numbers were assigned according to higher frequency, as detected in the GenBank database. Thus, the allelic types of four loci representing divergent sequences were provisionally assigned ( Table 2 and Tables  S2-S5). Although less than four alleles were found in D1/D2, RPB1, and RPB2, a larger number of divergent sequences were retrieved for ITS. Therefore, in this study, allelic types were assigned to only six sequences of ITS that showed a frequency of two or more in the GenBank database (Table 2). Sequence identities within ITS and D1/D2 loci were somewhat lower (95.19-99.74% and 96.46-99.26%, respectively) than those within RPB1 and RPB2 (≥98.52%) ( Table S6). The allelic types of four loci were evidently distinguished by phylogenetic dendrograms (Figure 1).

MLST Schemes of C. blankii and K. ohmeri
MLST schemes for C. blankii and K. ohmeri were designed using a range of sequences in the four loci which are similar to those of C. auris scheme [25] (Tables S2-S5). As described above, the divergent sequences of the four loci were searched using the BLAST web tool and the sequences obtained in the present isolates; allelic types were assigned as shown in Table 2, and also schematically represented by dendrograms (ITS locus of K. blankii and four loci of K. ohmeri) in Figure 1. For C. blankii, four alleles were identified for ITS, while only one or two alleles were assigned to other loci. In contrast, eight ITS and RPB2 allelic types were found in K. ohmeri, while the ITS region was more divergent (94.52-99.74% identity) than RPB2 (98.69-99.8% identity) (Table S6).

Assignment of ST for C. auris, C. blankii, and K. ohmeri Isolates
To assign ST for isolates of individual species, allelic profiles representing a combination of the four loci (ITS-RPB1-RPB2-D1/D2) were summarized in Table 3. All the C. auris isolates in the present study had a 1-1-1-1 profile, which was assigned to ST5, following ST1 through ST4 (ST cluster), which had been already described [25]. Nine C. blankii isolates exhibited an identical profile of 1-1-2-1, which was indicated by ST1. Six K. ohmeri isolates showed five different allelic profiles, representing ST1 through ST5, due to more diverse allelic types in RPB1 and RPB2. Only two isolates from septicemia belonged to the same type, ST1 (Table 4).  Table 2, and those representing each locus type retrieved from GenBank database are indicated with accession numbers. Trees were statistically supported by bootstrapping with 1000 replicates, and genetic distances were calculated using the Kimura two-parameter model. The variation scale is presented at the bottom. Locus type numbers are shown in brackets on the right of the dendrogram with vertical bars.  Table 2, and those representing each locus type retrieved from GenBank database are indicated with accession numbers. Trees were statistically supported by bootstrapping with 1000 replicates, and genetic distances were calculated using the Kimura two-parameter model. The variation scale is presented at the bottom. Locus type numbers are shown in brackets on the right of the dendrogram with vertical bars.   3.5. Characteristics of C. auris, C. blankii, and K. ohmeri and Their Infections All the C. auris isolates in the present study were classified into clade I (the South Asia clade), and mostly showed resistance to fluconazole and amphotericin B (Table 4). In the ERG11p, all isolates had a Y132F substitution, without a K143R mutation. Except for a case of otomycosis with C. blankii, all the C. auris and C. blankii infections were septicemia that occurred in neonates and was often associated with premature birth, a low birth weight, and an intravenous canula as risk factors. Most septicemia cases with C. auris (6/7), and half of septicemia cases with C. blankii (4/8) were fatal. Two C. blankii isolates showed higher MIC to fluconazole and amphotericin B than other isolates. K. ohmeri infections were septicemia in neonates (n = 3) and otomycosis in adults (n = 3), without a fatal case. Otomycosis due to K. ohmeri was recurrent and associated with chronic suppurative otitis media. Three isolates from otomycosis showed higher MIC to fluconazole than blood isolates. The two infection types were caused by different STs of K. ohmeri.

Discussion
In our previous study, C. auris and K. ohmeri were genetically confirmed in a few among 109 isolates of presumptive Candida species in Bangladesh [22]. However, in the present study, further characterization of those remaining among the same batch of isolates revealed the presence of more isolates of these species, and C. blankii was first confirmed in Bangladesh. Consequently, C. auris, C. blankii, and K. ohmeri were identified at a rate of approximately 6% each (8-9 among a total 131 isolates), accounting for nearly 20% of the total Candida-like yeast isolates. Particularly, C. auris and C. blankii were isolated from fatal septicemia, and K. ohmeri was associated with recurrent otomycosis as well as neonatal septicemia. These findings underscore the importance of identification of the three emerging species as fungal pathogens in Bangladesh. However, these species might have been underestimated or not recognized, because they are easily misidentified as other Candida species using biochemical testing [2,5,16,18]. Therefore, sequencing analysis of the ITS region should be employed for notable cases, both for appropriate therapeutic management and epidemiological study.
C. auris isolates in the present study were revealed to be a single clone belonging to ST5 (locus profile, 1-1-1-1) and clade I (South Asia clade), having an identical substitution in ERG11p (Y132F). The similar single clonality of C. auris was reported in Korea for 61 isolates from different sources and hospitals over a 10-year period [25]. These isolates belonged to ST2 (locus profile, 2-2-2-2) and were grouped into clade II (East Asia clade), and associated with a low incidence of the ERG11p substitution. Taken together, these findings using MLST indicated the distinctive nature of C. auris in global regions (represented by clades), and the single clonality of isolates at each study site. The distinction of isolates from South Asia and East Asia was also described using a phylogenetic analysis of the C. auris gene sequence (a concatenated sequence of four MLST loci), revealing cluster 1 (India, South Africa), cluster 2 (Korea and Japan), and cluster 3 (Brazil) [24]. Similarly, MLST-based phylogenetic analysis showed the differentiation of clades I-III among global isolates [30], and the distinction between South African isolates and Korean isolates [31].
Kwon and coworkers also analyzed ten C. auris strains representing each of four clades, provided by the US CDC, using the MLST scheme, and showed that ST1, ST2, ST3, and ST4 correspond to clade I, II, III, and IV, respectively [25]. However, in our present study, all the C. auris isolates in Bangladesh, which belonged to clade I, were assigned to ST5. This difference in the ST of clade I may imply a diversity of clones within a clade. In addition, the ST1 CDC isolate AR0382 had no mutation in Erg11p [25], unlike our ST5 isolates. Therefore, it is suggested that the C. auris strains of clade I may have evolved, presumably through exposure to antifungal agents, while a single C. auris clone is distributed throughout a region/country.
In the present study, in Bangladesh, C. blankii was first identified, and the isolates were revealed to comprise a single clone (ST1), causing 50% of fatal cases among septicaemia. Although MLST was not available previously, a single clonal origin of C. blankii isolates was described in a study in Delhi, India, as revealed by an amplified fragment length polymorphism pattern [16]. These isolates were derived from fungemia in neonates (with a case fatality rate of 45%), showing similar observations to our study, which suggests a prevalence of C. blankii in South Asia. Although limited information is available on this subject, human infections due to C. blankii have been reported in South Asia, Brazil, Argentina, Kuwait, Iran, and India [13][14][15][16][17]32,33]. Therefore, South Asia appears to be one of the global hotspots of the prevalence of C. blankii. Indian C. blankii isolates, which were collected mostly in 2016, were all susceptible to fluconazole [16]. However, in our study, two among nine C. blankii isolates (22%) showed higher MICs to fluconazole and amphotericin B, causing fatal outcomes. Additionally, in Kuwait, fluconazole-resistant C. blankii was reported to cause a fatal bloodstream infection in a neonate [13]. These findings suggest the occurrence of a mutation associated with azole resistance in C. blankii, as observed for C. auris. Therefore, the prevalence and spread of azole resistance in C. blankii should be also carefully monitored.
In contrast to C. auris and C. blankii, the K. ohmeri isolated in our present study was genetically divergent, as defined by ST, despite the small number of isolates. The isolates were derived from septicemia in children and otomycosis in adults, with the latter showing higher MICs to fluconazole. This species has been isolated from patients with various types of infections and background, with fungemia in neonates and elderly being dominant [18,19].
Although molecular epidemiological information is not available for K. ohmeri isolates, it is suggested that the wide range of infection types may be related to potentially higher genomic diversity, which might allow adaptation to various human body sites and tissues. This point may be further understood using genotypic analysis of K. ohmeri, depending on the infection type and/or host attributes. ST1 was assigned to two isolates from bloodstream infection in a neonate and in an infant, which may suggest an association of this type with septicemia. Though five STs were identified for the present K. ohmeri isolates, the D1/D2 locus was only assigned to a single type (type 1). This may imply that this locus type is the one prevalent in Bangladesh, or has higher conservation in D1/D2 than other loci within this species.
To date, MLST schemes of Candida have been established for four species (C. albicans, C. glabrata, C. krusei, and C. tropicalis), employing seven loci (C. albicans) or six loci (other species) [34][35][36][37]. According to the PubMLST website (https://pubmlst.org/organisms/, accessed on 29 April 2023), >4000 types of C. albicans, and >1400 types of C. tropicalis were registered, but there are still <300 types for C. krusei and C. glabrata. In the present study, we modified the existing MLST scheme of C. auris [25], and designed new MLST schemes of C. blankii and K. ohmeri based on the C. auris scheme with four loci. With these MLST schemes, we identified only a single ST for C. auris and C. blankii, and five STs for K. ohmeri. The lesser variety in identified STs in the present study may be due to the limited numbers of clinical isolates available for the analysis, and also possibly to the less divergent nature of these species within a single study site. However, there are many sequences representing various locus types of the four loci that could be retrieved from GenBank, and many of these types were not assigned to the present isolates in Bangladesh. Therefore, the presence of numerous STs may actually be estimated for global isolates. To evaluate the discrimination ability and usefulness of the MLST schemes for the epidemiological study of C. auris, C. blankii, and K. ohmeri, the accumulation of more sequence data from clinical isolates will be necessary. Above all, because the sequence data of C. blankii are substantially lacking, it is essential to have medical professionals recognize the presence of this emerging species to facilitate its genetic identification in clinical laboratories. For this, the spread of accurate identification of Candida species using sequencing of ITS region is necessary.
A limitation of the present study is that short tandem repeat (STR) typing was not performed for C. auris [38] because we focused on the design and application of MLST schemes for the three fungal species. STR typing has been applied for the genotyping of C. auris as a rapid and reliable method with high discrimination ability [39]. Therefore, a comparison of MLST with STR would be also of significance for the development of the molecular epidemiology of C. auris.