Mixed Candida albicans strain populations in colonized and infected mucosal tissues

Multilocus sequence typing of six Candida albicans colonies from primary isolation plates revealed instances of colony-to-colony microvariation and carriage of two strain types in single oropharyngeal and vaginal samples. Higher rates of colony variation in commensal samples suggest selection of types from mixed populations either in the shift to pathogenicity or the response to antifungal treatment.


Introduction
Candida albicans is the most common fungal species found as a gut commensal and opportunistic human pathogen. By multilocus sequence typing (MLST) of two or more sets of C. albicans isolates from 44 patients, we showed previously that microvariation, typically seen in the form of apparent gain and loss of sequence heterozygosity in one or more of the seven genes used for MLST, was demonstrable in 36 of the patients, and strain replacement was evidenced unequivocally in one patient between two hospital admissions (Odds et al., 2006). The observation of microvariation between multiple C. albicans isolates from single sources has been demonstrated previously by other genotyping methods (Schroppel et al., 1994;Lockhart et al., 1995;Pujol et al., 1999;Chong et al., 2003;Samaranayake et al., 2003a, b;Forche et al., 2005;Sampaio et al., 2005). We suggested that one explanation for the high level of microvariation encountered was that natural C. albicans populations in vivo comprise a mixture of closely related strain types, with a high level of genetic diversity maintained by mechanisms such as recombination and chromosomal ploidy shifts (Odds et al., 2006).
Phenotypic evidence for mixed C. albicans strain populations in vivo exists in terms of variations of antifungal susceptibility (Johnson et al., 1995;Le Guennec et al., 1995) and of colony form (Hellstein et al., 1993) between individual colonies in the primary isolation cultures. Hellstein et al. (1993) used Ca3 DNA fingerprinting to examine pairs of different colony forms among oral isolates from four patients and showed that one of the four patients carried two different, but related types. Le Guennec et al. (1995) used multilocus enzyme electrophoresis (MLEE) to characterize 10-colony sets cloned from 10 primary oral isolation plates for four AIDS patients. They found no MLEE variation among 40 colonies on four isolation plates from one patient, while MLEE differences consistent with microvariation were found among colonies from the other six plates.
To explore the prevalence of mixed C. albicans populations among individual patients further, we undertook a prospective survey by MLST of six individual C. albicans colonies chosen at random from 32 primary isolation plates. We obtained primary oral isolation plates, presumed to represent commensal carriage, from healthy volunteers and primary oral and vaginal isolation plates from patients with symptomatic oral and vaginal C. albicans infections. MLST for six randomly selected colonies on each isolation plate revealed that microvariation differences between colonies occurred more commonly in the commensal samples than the samples from infected patients. One of the 12 healthy volunteers was found to carry two entirely different strain types on the same isolation plate.

Materials and methods
Mouthwash samples positive for C. albicans were obtained from 12 undergraduate student volunteers who provided samples anonymously. The students rinsed their mouths with 10 mL of sterile distilled water, returned the fluid to a sterile container, and 100 mL was plated on Sabouraud agar. Plates bearing from 6 to 50 yeast colonies (mode = 10 colonies) were selected for further study. From the yeast growth, six well-separated colonies were chosen and propagated separately on Sabouraud agar slants. Samples from 10 female patients with symptomatic vaginitis and from 10 patients (five females) with various forms of oral Candida infection were handled similarly, except that the yeast growth came from plates inoculated from vaginal swabs and whole saliva samples, respectively. For the patients with vaginitis, the separate colonies had been streaked out from confluent yeast growth. For those with oral infection, the salivary C. albicans counts ranged from 20 to 4 10 4 yeasts mL À1 saliva, with most samples containing 4 10 3 yeasts mL À1 : these samples were plated with dilutions of saliva to provide the separated colonies sampled for this study. The antifungal treatment status of the patients at the time of sampling was unknown in most cases.
Presumptive identification of the isolates as C. albicans was based on colony colour on CHROMagar Candida (Odds & Bernaerts, 1994) and confirmed using PCR with primers that amplified the ITS1 region of ribosome-encoding DNA, which also designated the ATP-binding cassette (ABC) type of each isolate (McCullough et al., 1999). All 192 colonies were further typed by MLST and for homozygosity at the mating-type locus as described previously (Bougnoux et al., 2003;Tavanti et al., 2003). The MLST data, representing 1344 bidirectional sequence determinations, were assigned to genotypes for the seven loci sequenced and to diploid sequence types (DSTs) by reference to the Internet database for C. albicans MLST (http://test1.mlst.net/).

Results and discussion
Details of the 32 subjects whose primary isolation plates were the sources of six random colonies for strain typing are given in Table 1. For five of the 12 healthy volunteers, all six colonies from mouthwash isolation plates were the same DST and ABC type and all were heterozygous at the MTL. For Student04, five colonies were indistinguishable by DST and ABC type, but the sixth colony was a different DST and ABC type and even represented a different clade of strains from the other five colonies (Table 1). This result was interpreted as evidence of carriage of two unrelated strain types in a single individual. For Student10, one colony was a different but closely related DST from the other five colonies, and its ABC type (A) differed from that of the other five colonies (type C). This equivocal result may indicate either a microvariation in strain types or carriage of distinct types in a single individual. For the remaining five primary isolations from the healthy volunteers, either one or two colonies differed from the remainder in the sequence for just one of the seven DNA fragments used for MLST, a level of disparity we regard as indicating sequence microvariation. ABC types and MTL data were the same for all six colonies tested despite the variations in DST.
Among the 10 primary isolation plates from patients with vaginitis, two examples of one colony DST differing from the other five tested were found (Vag03 and Vag06, Table 1). For patient Vag05, four different but closely related DSTs were found among the six colonies tested. This represents the most disparate example of microvariation encountered among all 32 sets of isolates: all ABC and MTL data for this patient were the same. Among the 10 isolation plates from patients with oral Candida infections, three instances of single-colony DST differences were found (Oral03, Oral08 and Oral09, Table 1). ABC and MTL data were indistinguishable, patient per patient, for the colonies isolated from oral and vaginal infections.
From the 32 subjects, overall, we isolated strains representing seven C. albicans clades (Odds et al., 2007) plus one singleton. Discounting Student04, who carried strains from clades 1 and 2, 15 patients had isolates from clade 1, nine from clade 2, two from clade 8 and one each from clades 3, 4, 6 and 9 and a singleton (Table 1). Clade 1 is the most ubiquitous C. albicans clade, containing one-third of all isolates among 1391 studied from all global sources (Odds et al., 2007), and 70% of the 171 clade 2 isolates in the same study came from the UK. Because all the isolates in the present study were of UK origin, the heavy representation of isolates from clades 1 and 2 is consistent with existing epidemiologic information and the clade distribution of isolates in our sample appears to be typical for a set from 32 subjects in the UK.
The finding of a difference in ABC type between colonies for one of the isolates (from Student10) was unusual but not entirely unexpected; previously we found variability in ABC types within two sets of isolates from 43 sources (Odds et al., 2006).
Examples of DNA sequence microvariation, also called 'micro-evolution', between colonies have been evidenced previously by DNA fingerprinting (Hellstein et al., 1993) and by MLEE (Le Guennec et al., 1995), and microvariation after measured numbers of population generations has been documented as a consequence of exposure of C. albicans to azole antifungal agents (Cowen et al., 2000(Cowen et al., , 2001. Our data represent a much larger sample size than was used  infrequently, but our unpublished results show that MLST types can alter with prolonged exposure to antifungal agents, as already found by Cowen et al. (2000Cowen et al. ( , 2001. We also have preliminary (unpublished data) evidence of selective overgrowth of some DSTs when different strains are cocultured in competition. Our data showed a higher prevalence of colony-to-colony variation in primary isolations from healthy volunteers (six of 12 subjects showing microvariation based on DST, plus a seventh subject with two definitively different strain types in the single sample) than in the samples from patients with superficial infections (six of 20 subjects showed microvariation based on DST). If a colonizing population of C. albicans comprises a mixture of types, which was the case for 7/12 of our healthy volunteers, it seems reasonable to hypothesize that increased cell numbers and epithelial invasion associated with superficial infections may result from the selective proliferation of a single subtype, or a set of fewer subtypes, that was present in the mixed commensal population before invasive infection. This conclusion is qualified by the fact that the proportion of all isolated colonies tested was higher for the commensal samples, where numbers of colonies initially isolated were small compared with the high saliva counts and confluent growth from swabs for the patients. Because the antifungal treatment status was unknown for the majority of the patients, we cannot distinguish between selection of a less variable colony population arising from differences in invasiveness and selection in response to antifungal exposure.