Differentiation between Atypical Isolates of Candida lusitaniae and Candida pulcherrima by Determination of Mating Type

The variety of non-albicans Candida species involved in human pathology, their rising contribution to invasive infections, and the unusual antifungal susceptibility profiles of some of these species make their identification to the species level essential for epidemiological investigations and for optimizing therapy and patient management (8). In clinical laboratories, routine identification is usually performed with commercial systems that have more or less broad databases and that often combine classical phenotypic tests, i.e., assimilation, biochemical, and morphology tests. However, as these phenotypic characteristics may be variable, these methods may fail to discriminate between closely related species, leading to false or doubtful identifications, and may fail to detect new emergent pathogens (3, 6, 7). Candida lusitaniae, which is characterized by its propensity to develop resistance to amphotericin B during therapy, belongs to this important group of opportunistic pathogens (1, 4). Interestingly, it is one of the opportunistic Candida species that have a known sexual cycle (4). Its teleomorph, Clavispora lusitaniae, is a heterothallic ascomycetous yeast from the Metschnikowiaceae family (11). A mating type test has thus been developed to unambiguously identify C. lusitaniae isolates (2).

We report on five clinical isolates that the ID 32C system (bioMérieux, Marcy l'Etoile, France) suggested were either C. lusitaniae or Candida pulcherrima, another member of the Metschnikowiaceae family, which is a potential pathogen (4). These isolates were definitely identified as Clavispora lusitaniae through mating tests.

The five clinical isolates were obtained from five patients hospitalized in four medical centers by partner clinical laboratories. They were recovered from stool (isolates 74 and 147), the upper respiratory tract (isolate 87), and blood (isolates 93 and 155). They were presumptively identified as C. lusitaniae by routine testing. Before they were added to our collection, their identities were checked with the ID 32C system. The test was performed in accordance with the manufacturer's instructions. Growth in each well was determined by visual reading after 48 h of incubation. Results were recorded as a 10-digit numerical profile (biocode), which was analyzed by the API Lab software (bioMérieux). Depending on the results, identification can be made to the species or genus level or can be considered unreliable. The results were given as an identification percentage (%id) that represents the relative proximity of the profile obtained for the taxon to those of all of the other taxa in the database. A T index, representing the relative proximity of the profile obtained to the most typical profile for each taxon, was also given. Identification was regarded by the manufacturer as excellent (%id, ≥99.9; T index, ≥0.75), very good (%id, ≥99.0; T index, ≥0.5), good (%id, ≥99.0; T index, ≥0.25), or acceptable (%id, ≥80.0; T index, ≥0). The test results that did not support the identity found for the species (so-called “test against” results), followed by their percentage of expected occurrence for the species, were also given. Each identification assay was performed twice, in independent experiments. Candida guilliermondii ATCC 6260 and Candida glabrata ATCC 90030 were used as quality control strains. C. lusitaniae CBS 6936 and C. pulcherrima IP82363 were used as reference strains. Additional phenotypic tests, i.e., growth at 42°C and production of pseudomycelium, were performed with Sabouraud dextrose agar (bioMérieux) and potato-carrot-bile medium (Bio-Rad, Marnes la Coquette, France), respectively. The sexual reproduction ability of the five clinical isolates was studied by the mating test with yeast carbon base (Difco Laboratories) agar as previously described (2). C. lusitaniae strains 6936 MATa and CL38 MATα (ATCC MYA-2630) were used as mating type tester strains, and C. pulcherrima IP82363 was used as a negative control. To carry out a mating test, cells of an isolate and of the two tester strains were cultivated to stationary phase in 2 ml of a medium containing 1% yeast extract, 2% peptone, and 2% glucose under agitation (250 rpm) at 35°C. Equal volumes of cell suspension (typically 500 μl) from the isolate and from each of the tester strains were mixed separately in two 1.5-ml microtubes. The cell mixtures were centrifuged (3 min, 3,000 rpm [1,000 × g], 20°C), and the cellular pellet was resuspended in 500 μl of distilled sterile water. Aliquots of 5 μl from each mixture were spotted onto yeast carbon base solid medium. After 24 to 48 h of incubation at room temperature, cells were removed from the spots and examined with a phase-contrast microscope (magnification, ×400). The presence of asci and ascospores in only one of the two mixtures indicated that the isolate tested belonged to the species C. lusitaniae and that it had a mating type opposite to that of the tester. Confirmation should be obtained by verifying carefully the lack of asci and ascospores in the second mating reaction, indicating in this case that the two tested strains had the same mating type.

For all of the isolates but one (isolate 87), the first identification assay with the ID 32C system generated biocodes that corresponded to a good-to-excellent identification for the Candida genus level only (Table 1). The biocodes suggested that these isolates were either C. lusitaniae or C. pulcherrima. Major test results that did not support identification as C. lusitaniae were negative cellobiose (isolates 74 and 93), xylose (isolate 147), and methyl-d-glucose (isolate 155) test results. Identification of isolate 155 as Candida melibiosica was also proposed, without test against identification. Isolate 87, which did not assimilate rhamnose, was well identified to the species level, i.e., as C. pulcherrima. These disturbing results led us to perform a second assay. Overall, the same three species were proposed but the variable use of some carbon sources by three isolates generated new biocodes. Those for isolate 74, which never assimilated cellobiose, corresponded again to very good identification to the Candida genus level only. Isolate 87 assimilated rhamnose and was thus well identified as C. lusitaniae. For isolate 93, the identification comment was nonreliable because it assimilated neither cellobiose nor methyl-d-glucose.

When standard assimilation tests fail to discriminate between closely related species, discrimination may be obtained through certain culture characteristics. Among the three proposed species, C. lusitaniae is the only one to both grow at 42°C and produce elaborate pseudomycelium on potato-carrot-bile medium (5). All of the clinical isolates but one (isolate 155) grew at 42°C. Only isolates 74 and 87 produced an abundant, well-developed pseudomycelium after 48 h and could be thus identified as C. lusitaniae. Concurrently, the five isolates were subjected to mating type tests with two reference tester strains of C. lusitaniae, one of each mating type, MATa or MATα. Positive and negative mating type control tests were performed with C. lusitaniae reference strains 6936 and CL38 and with C. pulcherrima strain IP82363 (Table 2). Results unequivocally identified the five isolates as C. lusitaniae and allowed attribution of the MATa genotype to isolate 74 and of the MATα genotype to isolates 87, 93, 147, and 155.

Although C. lusitaniae is included in the databases of all of the identification systems commercially available in Europe, ID 32C was chosen as a checking system because of its extensive database (63 species). In addition, it is the only system that includes both the rhamnose and cellobiose assimilation tests, which are highly discriminant for C. lusitaniae (7, 10). In the course of our identification checking, we noted that interpretation of the biocodes sometimes yielded C. lusitaniae plus one or more species with often close identification percentages. Generally, discrimination was done easily by supplemental tests. However, the occurrence of C. pulcherrima on several occasions was of concern because it is a rare non-albicans Candida species that might become an opportunistic pathogen (4). In addition, its morphology and physiology are very close to those of C. lusitaniae, and C. pulcherrima is not included in the database of other identification systems (5). It must be underscored that there is no one phenotypic test that can discriminate definitely between C. lusitaniae and C. pulcherrima (5). Identification therefore relies on the relative frequency of a given profile for each taxon, rather than on the individual frequency of the response to each test. In the present study, both identification checks gave atypical assimilation profiles and the ID 32C system failed to identify the five isolates accurately. This failure is probably due, in part, to the database design. Note that the misidentification of isolate 87 would have been avoided if the database had taken into account the fact that some strains of C. lusitaniae fail to utilize rhamnose (5). Additional tests, i.e., pseudomycelium production and growth at 42°C, seem to be unreliable diagnostic tools for discrimination between C. lusitaniae and C. pulcherrima. An alternative test consists of using sexual reproduction, which has been described in Clavispora lusitaniae (11) and Metschnikowia pulcherrima (9), as a discriminating diagnostic tool. The two species are not cross fertile, and their meiotic products are very dissimilar; M. pulcherrima yields long, slender, acicular ascospores, whereas C. lusitaniae yields thick, clavate ascospores (2, 5, 11). We previously described an easy-to-use and efficient method to readily observe mating and ascospore production in C. lusitaniae (2). In the present study, this method was used successfully to discriminate between C. lusitaniae and closely related species, i.e., C. pulcherrima and C. melibiosica. As the results were obtained within 48 h, the test was not more time-consuming than the usual additional phenotypic tests. In conclusion, our study demonstrates the usefulness of the mating type test for identification, in a reference laboratory, of atypical strains of Candida (Clavispora) lusitaniae, which is otherwise problematic with the ID 32C system.

• Copyright © 2005 American Society for Microbiology