Evaluation of Vitek MS for Differentiation of Cryptococcus neoformans and Cryptococcus gattii Genotypes

Cryptococcus neoformans and Cryptococcus gattii are the main pathogenic species of invasive cryptococcosis among the Cryptococcus species. Taxonomic studies have shown that these two taxa have different genotypes or molecular types with biological and ecoepidemiological peculiarities.

ABSTRACT

Cryptococcus neoformans and Cryptococcus gattii are the main pathogenic species of invasive cryptococcosis among the Cryptococcus species. Taxonomic studies have shown that these two taxa have different genotypes or molecular types with biological and ecoepidemiological peculiarities. Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been proposed as an alternative method for labor-intensive methods for C. neoformans and C. gattii genotype differentiation. However, Vitek MS, one of the commercial MALDI-TOF MS instruments, has not been yet been evaluated for this purpose. Thus, we constructed an in-house database with reference strains belonging to the different C. neoformans (VNI, VNII, VNIII, and VNIV) and C. gattii (VGI, VGII, VGIII, and VGIV) major molecular types by using the software Saramis Premium (bioMérieux, Marcy-l’Etoile, France). Then, this new database was evaluated for discrimination of the different genotypes. Our in-house database provided correct identification for all C. neoformans and C. gattii genotypes; however, due to the intergenotypic mass spectral similarities, a careful postanalytic evaluation is necessary to provide correct genotype identification.

INTRODUCTION

The genus Cryptococcus comprises encapsulated yeasts that are divided in two main pathogenic species, Cryptococcus neoformans and Cryptococcus gattii (1) However, phylogenetic analyses have shown intraspecific genomic diversity among C. neoformans and C. gattii strains, dividing them into eight major molecular types or genotypes, as follows: Cryptococcus neoformans VNI, VNII, VNIII (AD hybrid), and VNIV and C. gattii VGI, VGII, VGIII, and VGIV (2, 3). Recent studies have shown that these eight major molecular types have differences in their biology, ecoepidemiology, antifungal susceptibility, and disease characteristics, making genotypic differentiation relevant (4).

Molecular methods, such as restriction fragment length polymorphism (RFLP), amplified fragment length polymorphism (AFLP), and multilocus or whole-genome sequence analysis, have been used to characterize these major eight molecular types; however, these methods are labor costly, labor-intensive, and restricted to reference laboratories (5). In 2011 and 2012, studies using the Bruker’s matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) instrument showed that this technique was able to discriminate the major molecular types of C. neoformans and C. gattii (5–7). Thus, MALDI-TOF MS, an easy, fast, and inexpensive method used in research and clinical laboratories, became an attractive alternative for C. neoformans and C. gattii genotype differentiation. The Vitek MS (bioMérieux, Marcy-l’Etoile, France), another MALDI-TOF MS instrument that has been used in clinical laboratories, has shown performance for yeast identification similar to that of Bruker’s instrument (8, 9). However, there are differences regarding mass spectrum acquisition, identification algorithms, and reference spectrum creation between the instruments, which makes relevant the evaluation of Vitek MS for C. neoformans and C. gattii major molecular type differentiation (10, 11). Thus, aiming to fill this gap and to consolidate the previous promising findings, we evaluated the Vitek MS (bioMérieux) and provided technical details for future application of this technology for the differentiation of C. neoformans and C. gattii molecular types.

MATERIALS AND METHODS

Fungal organisms.

A total of 44 isolates belonging to the major C. neoformans and C. gattii molecular types were used to construct in-house SuperSpectrum library, with 24 isolates from the National Institute of Quality Control in Health (INCQS; https://portal.fiocruz.br/en/unidade/national-institute-quality-control-health-incqs) and 20 clinical isolates maintained at the culture collection from the Tropical Medical Institute from São Paulo (University of São Paulo). To test the performance of the new database, 32 other isolates also representative of the major C. neoformans and C. gattii molecular types were included in the study, with 26 isolates from the mycological collection of the Federal University from Rio Grande do Norte, Brazil, and 6 additional isolates from the Tropical Medicine Institute (see Table S1 in the supplemental material). Clinical isolates were genotyped by URA5 restriction fragment length polymorphism (RFLP) analysis, as previously described (12).

Sample preparation for MALDI-TOF MS analysis.

Strains and isolates maintained as frozen stocks at −80°C in yeast extract-peptone-dextrose medium were subcultured on Sabouraud’s dextrose agar (SDA) plates and incubated for 48 h at 30°C before MALDI-TOF MS analysis. Then, a standard protein extraction protocol with ethanol and formic acid was carried out. In brief, one loop of yeast biomass was transferred into a 1.5-ml microcentrifuge tube containing 300 µl of purified water and mixed thoroughly. Subsequently, 900 µl of absolute ethanol was added to each tube and mixed for 1 min. The samples were centrifuged for 2 min at 13,000 rpm, and the supernatant was removed. The pellet was dried at room temperature, and 50 µl of formic acid (70%) was added. In addition, an equivalent volume of acetonitrile was added, and the mixture was centrifuged for 2 min at 13,000 rpm. Finally, 1 μl of the clear supernatant was spotted in quadruplicate onto a disposable MALDI target slide composed of a polypropylene carrier with a stainless steel layer (bioMérieux). After air-drying, each spot was overlaid with 1 µl of α-cyano-4-hydroxycinnamic acid (HCCA) matrix (bioMérieux).

MALDI-TOF MS analysis.

Measurements were performed on a Vitek MS instrument (bioMérieux). For each acquisition group, a standard (Escherichia coli ATCC 8739) was included to calibrate the instrument and validate the run. The spectra were generated using the Launchpad version 2.8 software (bioMérieux) and analyzed using the research use only (RUO) software Saramis Premium version 4.12 (bioMérieux). The in vitro diagnostic (IVD) system was not evaluated due to the lack of C. neoformans and C. gattii genotype-specific mass spectrum profiles in the database.

Construction of an in-house SuperSpectrum library.

For SuperSpectrum construction, mass spectra of 44 isolates belonging to the major C. neoformans and C. gattii genotypes (see Table S1) were imported into the Saramis Premium software package (bioMérieux). Then, 10 high-quality mass spectrum replicates (≥120 masses, ≥70% similarity) of a given organism were selected to create a genotype-specific SuperSpectrum with the Saramis Premium SuperSpectrum tool (bioMérieux), according to the manufacturer’s instructions. The specificity of the potential biomarker masses was determined by comparison against the whole Saramis spectral archive (bioMérieux), and each SuperSpectrum had 40 biomarkers (masses) selected. Then, each biomarker was designated 35 points, given a total of 1,400 points for the final SuperSpectrum. Mass spectrum profiles from other Cryptococcus species (e.g., Cryptococcus laurentii [n = 11], Cryptococcus curvatus [n = 8], Cryptococcus uniguttulatus [n = 12]) and from the genus Trichosporon (n = 94) that are present in the original Saramis spectral archive version 4.12 (bioMérieux) were submitted to identification against the new in-house library to attest its specificity.

Genotype identification by MALDI-TOF MS.

We initially evaluated the clustering of the different genotypes by mass spectrometry by creating a SuperSpectrum dendrogram and correlation matrix through the Saramis Premium software (bioMérieux). The dendrogram was generated based on whole spectra, with a single-link clustering algorithm and a binary mass list with an error of 800 ppm. Then, the 44 isolates that were used to construct the in-house SuperSpectrum library were subcultured once again and submitted to identification. Self-identifications (mass spectra identified by SuperSpectra from the same isolate) were not taken into account. Finally, the second set of 32 isolates, which were not used to construct the in-house SuperSpectrum library, were also submitted to identification (performance analysis step).

The Saramis Premium software (bioMérieux) compares the sample’s mass spectrum against SuperSpectra, and the sum of peak weights is computed after matching mass signals of each SuperSpectrum and is transformed into a confidence value and points as follows (13): dark green with 99.9% confidence values or 1,000 to 1,400 points, light green with 90% to 99% confidence values or 900 to 999 points, yellow with 80% to 89.9% confidence values or 800 to 899 points, and white with 75% to 79.9% confidence values or 750 to 799 points.

RESULTS

The dendrogram split the SuperSpectra of C. neoformans and C. gattii into two species-specific branches (Fig. 1). These main branches showed a clear distinction between the two species, with SuperSpectra showing more than 20 species-specific biomarkers (Fig. 1). Moreover, the cryptococcal isolates were separated into clusters corresponding to the main genotypes of the two Cryptococcus species (Fig. 1). No single mass spectrum or SuperSpectrum from Cryptococcus species other than C. neoformans and C. gattii or from Trichosporon species showed cross-identification with the new in-house library.

FIG 1.

Saramis Premium dendrogram clustering of SuperSpectra from each Cryptococcus neoformans and Cryptococcus gattii organism, with distances displayed in number of identical masses. Filter was tolerance of 0.08%, absolute intensity of ≥0, relative intensity of ≥0, and mass range 3,000 to 20,000 Da.

The correlation matrix of the C. neoformans SuperSpectra (each one with 40 biomarkers) showed higher (twice) intragenotype than intergenotype similarity, as there was a mean of 24 intragenotype common biomarkers compared with only 12 intergenotype common biomarkers (Fig. S1A). The correlation matrix of the C. gattii SuperSpectrum (each one with 40 biomarkers) also showed higher intragenotype than intergenotype similarity; there were means of 27 and 13 intragenotype and intergenotype common biomarkers, respectively (Fig. S1B).

All isolates that were used to construct the in-house SuperSpectrum library had correct species assignment as C. neoformans and C. gattii by the new in-house SuperSpectrum library. Regarding the results for genotype identification, the provided confidence intervals were not discriminatory, and ambiguous results with >90% confidence level values occurred in 20 of the 24 (83%) and in 20 of the 20 (100%) C. neoformans and C. gattii organisms, respectively (Fig. 2A). However, the identification points were able to correctly discriminate all isolate genotypes, with the higher point values corresponding to the accurate genotype.

FIG 2.

Identification points provided for each isolate by the Saramis Premium software (bioMérieux). The values for the correct genotype identifications are depicted in a continuous line, while those for incorrect genotype identifications are provided in a dashed line. Color boxes correspond to the confidence level identifications, as follows: dark green with 99.9% confidence values, light green with 90% to 99% confidence values, and yellow with 80% to 89.9%. As noticed in the light and dark green boxes, most of the isolates had cross-identification with other genotypes with 90% to 99.9% confidence levels. However, higher identification points provided correct genotype identifications. (A) Identifications for the isolates that were used for the construction of the in-house library. (B) Identifications for the isolates that were used during the performance step.

Among the 32 additional isolates used in the performance analysis step, all had correct species assignment as C. neoformans and C. gattii by the new in-house SuperSpectrum library. For genotype identification, the provided confidence intervals were also not discriminatory, and ambiguous results with >90% confidence level values occurred in 5 of the 19 (26%) and in 7 of the 13 (54%) C. neoformans and C. gattii organisms, respectively (Table 1). Yet, genotype identification of all isolates was efficiently achieved by the reported point values, with higher point values corresponding to the accurate genotype (Table 1).

TABLE 1.

Cryptococcus neoformans and Cryptococcus gattii genotypes identification provided by the Saramis Premium software equipped with the in-house SuperSpectrum library during the performance step

Genotype (no. of isolates)	Unambiguousa identifications with green confidence levels (%)	Confidence levels of (%):		Correct identifications with higher point values (%)	Point values of:
		Correct identifications	Incorrect identifications		Correct identifications	Incorrect identifications
C. neoformans
VNI (8)	87.5	99.9–99.9	80.5–91	100	1,225–1,400	805–910
VNII (6)	67	99.9–99.9	80.5–98	100	1,295–1,400	805–980
VNIII (1)	100	99.9	87.5	100	1260	875
VNIV (4)	50	99.9–99.9	80.5–94.5	100	1,050–1,260	805–945
C. gattii
VGI (2)	0	99.9–99.9	94.5–98	100	1,295–1,365	945–980
VGII (8)	62	99.9–99.9	77–98	100	1,065–1,365	770–980
VGIII (1)	0	99.9	98	100	1,330	980
VGIV (2)	100	99.9–99.9	80.5–84	100	1,295–1,330	805–840

^aOnly the correct genotype was named with confidence levels above 90%.

Figure 2 illustrates that interpretation of the results provided by Saramis Premium software (bioMérieux) using point values instead of confidence levels is able to correctly discriminate all C. neoformans and C. gattii genotypes.

DISCUSSION

In the last 20 years, taxonomic studies with the help of molecular tools have recognized several new (cryptic) species among the pathogenic fungi (14, 15). The new taxonomic evidence added to biological traits, and to epidemiological and clinical data, has resulted in powerful and holistic information that is helping better understand transmission routes and mechanisms of acquisition, implement disease control measures, and finally, achieve a better organism-targeted therapy (16). For example, it was shown that VGII and VGIV isolates show higher MICs for azole derivatives than do isolates of C. neoformans and the other C. gattii (17) genotypes, and VGII isolates have been linked to outbreaks of invasive disease in immunocompetent hosts in North America (18) and Brazil (19). Indeed, the assembly of these relevant data with genealogical analyses has led experts to propose that the C. neoformans and C. gattii major molecular types are indeed at least seven different species, including C. neoformans (genotypes VNI and VNII), C. deneoformans (VNIV), C. gattii (VGI), C. deuterogattii (VGII), C. bacillisporus (VGIII), C. tetragattii (VGIV), and C. decagattii (2, 20).

Since MALDI-TOF MS is becoming the method of choice for microorganism identification in clinical laboratories all over the world, systematic identification of C. neoformans and C. gattii major molecular types by using this technology may be achieved in the future. However, validated reference spectrum libraries including several representatives of each genotypes/species are necessary to produce accurate results and to minimize the number of misidentifications (21). Moreover, the identification algorithms of the different MALDI-TOF systems may not have sufficient discriminatory power to separate some closely related taxa. Thus, additional postanalytical steps may be required for the correct organisms’ identification (8, 22). Dendrograms generated with the RUO tools (e.g., BioTyper and Saramis) provide preliminary yet useful information regarding the relatedness of the different species by its main mass spectrum profiles. As previously reported by studies that used the Bruker’s MALDI-TOF MS apparatus (5–7), the dendrogram created with the Saramis Premium software segregated the major molecular types of C. neoformans and C. gattii into different and genotype-specific clades. Despite clear genotype distinction provided by the dendrogram, initial evaluation showed that 91% of the organisms had ambiguous identifications with >90% confidence values. As shown in the correlation matrix, some SuperSpectra belonging to different genotypes had more than 20 common biomarkers. These intergenotypic similar mass spectrum profiles when analyzed by the Saramis’s matching signals method may result in ambiguous identifications if only the confidence level percentage is taken into account. However, by using an alternative algorithm that took into account only the higher identification points, all organisms had unambiguous genotype assignment by the Saramis software.

Alternatively, the Vitek MS IVD system (bioMérieux) uses an identification algorithm method based on a process called “mass binning” that has shown better performance for microorganisms identification than Saramis’s method of matching mass signals (13, 23, 24). Thus, although not evaluated here, an upgrade of the IVD database with the major C. neoformans and C. gattii genotypes by the manufacturer could enhance the power of Vitek MS to well discriminate these microorganisms.

In conclusion, we showed that Vitek MS (bioMérieux) instrument has the potential to well discriminate the major molecular types of C. neoformans and C. gattii. The Saramis software enables the construction of genotype-specific SuperSpectra, but a robust in-house library and an alternative identification algorithm are necessary to achieve accurate performance.