Development and Validation of an in-House Library of Colombian Candida auris Strains with MALDI-TOF MS to Improve Yeast Identification

Abstract

Background: Candida auris is characterized for having a high genetic variability among species. MALDI-TOF MS library contains spectra from only three strains of C. auris, which makes difficult the identification process and gives low scores at the species level. Our aim was to construct and validate an internal library to improve C. auris identification with Colombian clinical strains. Methods: From 30 clinical strains, 770 mass spectra were obtained for the construction of the database. The validation was performed with 300 strains to compare the identification results in the BDAL and C. auris Colombia libraries. Results: Our library allowed a complete, 100% identification of the evaluated strains and a significant improvement in the scores obtained, showing a better performance compared to the Bruker BDAL library. Conclusions: The strengthening of the database is a great opportunity to improve the scoring and C. auris identification. Library data are available via ProteomeXchange with identifier PXD016387.

1. Introduction

Invasive candidiasis is the most common fungal disease in hospitalized patients, with more than 50,000 deaths worldwide [1]. The main etiological agent is Candida albicans. However, the prevalence of non-albicans species (e.g., C. glabrata, C. parapsilosis, C. tropicalis, C. krusei, C auris) is increasing [2,3].

C. auris is a multi-drug resistant yeast (MDR), which was first recovered in 2009 from the external ear canal of a 70-year-old female Japanese patient in Tokyo [4]. Henceforth, it has been reported in over 20 countries on five continents, including Colombia, with reports made by our group [5,6]. C. auris candidemia is associated with mortality rates of about 30–60%, depending on the setting [7].

The identification of C. auris is a challenge. Only a few phenotypic methods detect this species correctly. It is often classified mistakenly as C. haemulonii, C. famata, C. lusitaniae, and C. parapsilosis by VITEK 2; as R. glutinis by API 20C AUX systems; as C. catenulate and C. haemulonii by BD Phoenix, and as C. guilliermondii and C. tropicalis by MicroScan [5,7]. The unique biochemistry methodology that is capable to detect it with a high reliability is the version VITEX 2 XL (bioMérieux version 8.01). However, specialized techniques, such as DNA sequencing and matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) are the most efficient methodologies for a reliable identification [8,9,10].

In the last decade, MALDI-TOF MS has been proven as a solid option for microbial identification through the comparison of its profiles of cytoplasmic protein spectra. This technique provides faster identification than traditional molecular methods, and its usefulness has been evaluated and used in many taxonomic groups, including fungi [11].

The capabilities of this method depend mainly on the content of the protein spectral libraries that are used. Currently, the identification of C. auris by MALDI-TOF MS Biotyper OC 3.1.66 BRUKER-BDAL contains spectra of only three C. auris strains (CBS KCTC 17809, CBS KCTC 17810 from Korea) and C. auris (DSM 21092T CBS from Japan), which made it difficult to obtain a high identification score in our experience with Colombians strains. However, Vatanshenassan et al. published good scores obtained in the identification of 50 C. auris strains from United States, Netherlands, India, Israel, and South Africa using Bruker MBT Compass Library, Revision E MBT 7854 [12,13].

In 2018, Escandón et al. performed a phylogenetic analysis based on the sequencing of the complete genome of C. auris Colombian strains. They made a comparison with countries such as South Africa, Japan, Pakistan, and India, and observed polymorphism variants of nucleotides with 140 Venezuelan strains. This indicated the great difference that exists in each geographic region of C. auris strains [14,15]. It is important to create a library with strains from different geographical locations, which will make it possible to strengthen libraries for greater and better identification, as well as to facilitate epidemiological studies and clinical diagnoses. Herein, we constructed and validated an internal library for the identification of C. auris with Colombian clinical strains.

2. Methods

2.1. Ethics’ Statement

The authors confirm that the ethical policies of the journal, as noted on the journal’s author guidelines page, have been adhered to. In this investigation, no ethical approval was required. The data used were strains spectra obtained by MALDI-TOF MS. All patients are anonymized and only the code of strains was transferred for this investigation. Therefore, no informed consent was required.

2.2. Characteristics of Strains Used to Make the in-House Library

The library was created with 30 Colombian clinical strains: The ones that were used to make the first reports of C. auris in Colombia, from the outbreak reported in the north and center of the country previously identified as a C. auris by MALDI-TOF MS [5,6].

2.3. Construction of In-House Library “C. auris Colombia”

Strains preserved in our bank were cultured from glycerol stocks stored at −80 °C, after incubation of strains at 35 °C for 24–36 h on Sabouraud dextrose agar (SDA). Protein extraction was performed using the protein extraction in formic acid/ethanol method, according to the Bruker Daltonics’ protocol with minor modifications as reported by Marklein in 2009 [16]. Briefly, two or three colonies were mixed with 300 μL of HPLC-grade water until homogenization and then 900 μL of 100% ethanol (Sigma-Aldrich, St Louis, MO) were added. After centrifugation at 15,000 g for 2 minutes, the pellet was dried at 25 °C, reconstituted in equal volumes of 100% formic acid (Sigma- Aldrich) and acetonitrile (Sigma-Aldrich) (20 µL each), mixed thoroughly, and centrifuged at 15,000 g for 2 min. One microliter of supernatant was spotted onto a 96-spot steel plate (Bruker Daltonic) and left to dry at room temperature before the addition of 1 µL of the α-Cyano-4-hydroxycinnamic acid HCCA matrix (provided by the supplier). For yeast identification, each sample was tested in duplicate. Only the spot reporting the highest probability score of identification was considered [17]. MALDI-TOF MS mass range for measuring spectra was made between 2000 and 20,000 Daltons, results were then compared, and a score was obtained according to the manufacturer’s technical specifications, as follows: Highly probable species identification (≥2300–3000), secure genus identification, probable species identification (≥2000–2299), probable genus identification (1700–1999), and not reliable identification (<1700).

For the library construction, each protein extract from strains was placed on 10 different spots on the plate to generate mass spectra data using FlexControl version 3.4. Only higher quality spectra (low noise, recognizable peaks, and more than 10⁴ units of intensity) analyzed with FlexAnalysis software were considered for the creation of combined Main Spectra (MSP) library with a minimum of 20 spectra per strain. Compound spectra were created for the library using the “MSP creation” function of the MALDI Biotyper version 3.1 7311 reference spectra (main spectra) (Bruker Daltonics, Bremen, Germany), a library named “Candida auris Colombia”. For phylogenetic analysis of the mass spectra of C. auris, the dendrogram was generated using the respective functionality of the MALDI-TOF MS Biotyper 3.1 7311 offline client. The mass spectra of 30 respective strains with a score value of >2 was considered for the preparation of the dendrogram. The spectra of all strains tested were analyzed as a core-oriented dendrogram using an arbitrary distance level of 1000 as the cut-off [18,19,20].

The mass spectrometry proteomics data were deposited into the ProteomeXchange Consortium via the PRIDE [21] partner repository with the dataset identifier PXD016387.

2.4. Validation of Colombia Candida auris Library

The library evaluation was effectuated with 300 C. auris strains from different national institutions that were identified in our laboratory. The spectra obtained for these identifications were used to evaluate our library [22], following the same protein extraction protocol cited before, comparing results between BDAL (Bruker) and Candida auris Colombia (in house) libraries (Figure 1). Potential false positives from non-C. auris strains, such as C. catenulata, C. duobushaemulonii, C. famata, C. guilliermondii, C. haemulonii, C. krusei, C. lusitaniae and C. parapsilosis, were evaluated. None of the yeasts was misidentified.

3. Results

3.1. Construction of the in-House Library “C. auris Colombia”

The MSPs for the in-house library construction was generated through the accumulation of 777 mass spectra from 30 C. auris clinical strains (Figure S1). Subsequently, the three MSPs from C. auris in the BDAL library and our 30 MSPs were used for the dendrogram construction (Figure 2). We observed that our MSPs were different from the BDAL MSPs, which led the identification of C. auris Colombian strains to present scores of <2.0 in the BDAL library. The dendrogram obtained showed two clades (Colombian-Oriental strains) when compared to the mass spectrum of each isolate. This library is available for everyone and those interested only need to write to the corresponding author; validating with strains from other geographic areas is important for us. The C. auris Colombia library showed a greater MSP diversity compared with BDAL (one representative MSP) (Figure 3).

3.2. Candida auris Colombia Library Validation

With the BDAL library, 199/300 strains (66%) were identified, belonging to four distinct identification levels: Highly probable species identification (0/300); secure genus, probable species identification (62/300); probable genus identification (137/300); and not reliable identification (101/300).

When performing MALDI-TOF MS identification with the Candia auris Colombia library, better results were obtained, identifying the totality of the isolations and improving the identification score 300/300 (100%). Strains were identified belonging to three distinct identification levels: Highly probable species identification (81/300); secure genus, probable species identification (196/300); and probable genus identification (23/300) (Figure 4).

4. Discussion

Since the appearance of C. auris in the world in 2009 [4], the field of medical mycology has been revolutionized. C. auris is a yeast that not only presents an atypical susceptibility profile but also a laborious identification [10]. Direct protein analysis using MALDI-TOF MS offers a great opportunity to identify microorganisms that are difficult or impossible to detect with the biochemical methods [23]. In this study, we constructed a C. auris library to improve the identification of the yeasts due to the lack of MSPs in the BDAL library. When comparing the mass spectra of BDAL with the C. auris Colombia library, we observed that the latter had significant differences in the spectra. This probably occurred due to protein extraction protocol or the geographical origin of these isolations, which is consistent with Escandón et al., who observed a phylogenetic difference between Colombian strains and those from other countries [14]. This distinction also occurs with phenotypic characteristics, such as a strain’s susceptibility, where the resistance for antifungal molecules is different depending on the geographic area [24]. Also, resistance mechanisms varied in each country. A multicentric study stated that the mutations in ERG11 implicated in azole resistance varied according to the geographic region of the strains [25]. The variation in the characteristics of the different strains around the world made the identification with traditional methods difficult, even with MALDI-TOF MS. Therefore, when we identified our strains using MALDI-TOF MS with BDAL library, the score of identification was <2 and when we built the dendrogram with our strains’ spectra and BDAL strain spectra, we observed a deep separation that was represented in the clear division between the Colombian and the Eastern strains.

Candida auris Colombia library presented multiple spectra of C. auris that improved the identification scores in 100% of the strains used for the validation of our library. This confirmed that a greater diversity and the robustness of the database improved the identification accuracy of C. auris based on MSPs in MALDI-TOF MS.

We know that a weakness of our study was not having C. auris strains from the four phylogenetic clades, but as the CDC says, “supplemental MALDI-TOF databases that include additional C. auris strains from all four of the phylogenetic clades may enable users to overcome identification challenges by providing consistently higher MALDI identification scores. MicrobeNet is one example of a free online MALDI database of rare and unusual pathogens that is curated by CDC experts [8].” With this Colombia library, we want to provide spectra from South American strains (clade IV) and contribute to the construction of a robust database of all clades.

5. Conclusions

The Bruker’s library version 3.1 7311 reference spectra (main spectra) had only three C. auris strains from Korea and Japan, which entailed identification problems and a low identification score. Our database had a remarkable improvement in the identification of the 300 strains, evidencing that the strengthening of the database is a great opportunity to improve the scoring and the identification of C. auris. We are interested in sharing this library and contributing to the construction of a robust database.