|Home | About | Journals | Submit | Contact Us | Français|
We investigated the population structure of 208 Candida dubliniensis isolates obtained from 29 patients (25 human immunodeficiency virus [HIV] positive and 4 HIV negative) as part of a longitudinal study. The isolates were identified as C. dubliniensis by arbitrarily primed PCR (AP-PCR) and then genotyped using the Cd25 probe specific for C. dubliniensis. The majority of the isolates (55 of 58) were unique to individual patients, and more than one genotype was recovered from 15 of 29 patients. A total of 21 HIV-positive patients were sampled on more than one occasion (2 to 36 times). Sequential isolates recovered from these patients were all closely related, as demonstrated by hybridization with Cd25 and genotyping by PCR. Six patients were colonized by the same genotype of C. dubliniensis on repeated sampling, while strains exhibiting altered genotypes were recovered from 15 of 21 patients. The majority of these isolates demonstrated minor genetic alterations, i.e., microevolution, while one patient acquired an unrelated strain. The C. dubliniensis strains could not be separated into genetically distinct groups based on patient viral load, CD4 cell count, or oropharyngeal candidosis. However, C. dubliniensis isolates obtained from HIV-positive patients were more closely related than those recovered from HIV-negative patients. Approximately 8% (16 of 194) of isolates exhibited itraconazole resistance. Cross-resistance to fluconazole was only observed in one of these patients. Two patients harboring itraconazole-resistant isolates had not received any previous azole therapy. In conclusion, longitudinal genotyping of C. dubliniensis isolates from HIV-infected patients reveals that isolates from the same patient are generally closely related and may undergo microevolution. In addition, isolates may acquire itraconazole resistance, even in the absence of prior azole therapy.
Candida dubliniensis was described for the first time in 1995 as a new species closely related to C. albicans (24). Although C. albicans is the clinically most important yeast (15), C. dubliniensis is frequently the causative agent of oropharyngeal candidosis (OPC) in human immunodeficiency virus (HIV)-infected and AIDS patients (3, 6, 21, 24).
Joly et al. developed a C. dubliniensis-specific semirepetitive probe (Cd25) to be used for DNA fingerprinting analysis and described the presence of two distinct groups of C. dubliniensis among 57 independent isolates (7). Later studies by Gee et al. confirmed these observations (5). These authors detected the same two distinct populations in a group of 98 independent C. dubliniensis isolates when using the Cd25 probe and karyotype analysis developed by Joly et al. (7). The majority of Cd25 group I isolates were from HIV-infected individuals, whereas the majority of Cd25 group II isolates were from HIV-negative individuals. These authors also showed that C. dubliniensis can exhibit microevolution in vivo and in vitro since it occurs in other yeast species. A third group of isolates which was generated by hybridization with the Cd25 probe was described by Al Mosaid et al. (2), which demonstrated a resistance against 5-flucytosine.
McManus et al. analyzed the population structure of C. dubliniensis by multilocus sequence typing and found a lower degree of diversity in this species in comparison to C. albicans (11).
According to our knowledge, there are only a few longitudinal studies analyzing the genotypic alterations in Candida spp. (18), and most of them focus on C. albicans (19). In one study, a number of HIV-infected patients enrolled in a longitudinal study of OPC were found to be infected with C. albicans at the beginning of the study and with C. dubliniensis at the end (10). There are no studies in which genotypic alterations in C. dubliniensis strains were examined over a prolonged period of time. Therefore, we performed Cd25 fingerprint analysis and PCR genotyping of C. dubliniensis isolates recovered from patients over several years. The aim of our study was the examination of the genotypic variability of C. dubliniensis isolates in relation to HIV infection, viral load, CD4 count, OPC, and resistance against antifungal drugs.
We selected 29 patients (25 were HIV positive and 4 were HIV negative) colonized or infected at least once with C. dubliniensis and collected 208 C. dubliniensis isolates over a period of 13 years (1989 to 2002). The maximum observation time was 9 years for a single patient. Each patient was sampled for Candida colonization in the oral cavity between 1 and 48 times over the course of the study. The majority (200 of 208) of the C. dubliniensis isolates were recovered from oral rinsings from HIV-positive patients. The isolates were grown on CHROMagar Candida to differentiate between different Candida species. One colony from each Candida isolate which demonstrated a green color (indicative of C. dubliniensis or C. albicans) was further characterized by arbitrarily primed PCR (AP-PCR) (25), and the ones demonstrating the pattern characteristic for C. dubliniensis were genotyped. The patients were arranged in groups with a high viral load (>36.00 copies/ml) and a low viral load (<36.000 copies/ml) according to the method of Coogan et al. (4). Patients with >200 CD4 cells/μl were separated from those with <200 CD4 cells/μl according to the method of Lilly et al. (8).
All isolates were tested according to the CLSI document M27-A2 using RPMI 1640 (with glutamine and phenol red, without bicarbonate) medium buffered to pH 7.0 with 0.165 M 3-[N-morpholino]-propanesulfonic acid, an inoculum of 103 cells/well, and incubation at 35°C (14). The MIC was defined as the lowest drug concentration in which no detectable growth was visible after 48 h using criteria described in CLSI document M27-A2.
Chromosomal DNA from all isolates was purified by a standard phenol-chloroform extraction (20) and used for AP-PCR and RFLP analysis. For the AP-PCR, we used the single primer RP02 (5′-gCg ATC CCC A-3′) (1) as described by Sullivan et al. (24). A total of 100 ng of the DNA was amplified in a PCR tube in a final volume of 50 μl, using the following cycling conditions: denaturation 5 min at 94°C, annealing 5 min at 36°C, and extension 5 min at 72°C. These steps were repeated four times. The next steps (94°C for 1 min, 36°C for 1 min, and 72°C for 2 min) were repeated 30 times. The final extension was done at 72°C for 10 min, and the tubes were cooled at 4°C until the products were loaded onto the 1.2% Tris-borate-EDTA agarose gel.
The PCR identification of isolates belonging to the different C. dubliniensis genotypes was performed as described by Gee et al. (5) using the primer pairs G1F-G1R and G2F-G2R. The PCR products were analyzed by running through an agarose gel stained with ethidium bromide.
The status of the C. dubliniensis CDR1 gene, i.e., wild-type versus mutated gene, was done as described by Moran et al. (12). Aliquots (10 ng) of genomic DNA were amplified in a total volume of 25 μl using Platinum Taq (Invitrogen, Germany), and 9-μl portions of the PCR products were cut with the restriction enzyme SspI (Fermentas, St. Leon-Rot, Germany) for 5 h in a volume of 11 μl. The whole digestion mix was loaded onto a 2% agarose gel and stained with ethidium bromide.
Restriction fragment length polymorphism (RFLP) patterns were obtained by overnight digestion of 10 μg of chromosomal DNA by using EcoRI (Invitrogen GmbH, Karlsruhe, Germany). Digested DNA was run through 0.8% agarose gels and transferred to nylon membranes (Hybond; Amersham). The probe consisted of a C. dubliniensis-specific sequence of ~16 kb which had been cloned in a lambda vector (7). The complete purified vector DNA (25 ng) was radioactive labeled with [32P]dTTP using a MegaPrime kit (GE Healthcare/Amersham) according to the instructions of the manufacturer. The hybridization buffer (MegaPrime kit; GE Healthcare) was prewarmed to 65°C, and denatured salmon sperm (final conc. 10 μg/ml) was added. The blots were prehybridized for 15 min and, after addition of the labeled probe, the hybridization was continued for additional 2 h at 65°C. The blots were washed once with washing buffer I (MegaPrime) at room temperature and twice at 65°C with washing buffer II (MegaPrime). The exposure to X-ray film was performed at room temperature without an intensifying screen, the time being dependent on the signal intensity.
For the analysis we used the Gelscan software package v5.1 (BioSciTec, Frankfurt, Germany). Most of the lanes and bands were detected automatically by the software. Bands that were obviously in the same position in different lanes but, due to slight differences in sample loading or migration, were not considered the same by the software were included manually (this was done by two different people independent from each other). For the analysis all bands visible between 23,130 and 2,027 bp were included. HindIII-digested lambda DNA was used as a marker which hybridized with the Cd25 probe, since we applied the complete probe, including the phage sequence for hybridization. The six marker bands which were used for the assignment of the banding pattern correspond to 23,130, 9,416, 6,557, 4,361, 2,322, and 2,027 bp. The Gelscan software was used to generate the cluster analyses based on the Euclidean distance d(Eu) according to the following formula:
where x1 is first band of isolate x and y1 is the first band of isolate y. The cluster analysis was generated by using the Dice coefficient and an unweighted pair group method. The size of the bands was removed from the original formula by an additional software package from BioSciTec. Therefore, the presence or absence of the bands was used to calculate a similarity coefficient between the different isolates. A value of d(Eu) = 0 indicates an identity of all bands. Increased values of d(Eu) demonstrate bands at different positions, i.e., the higher this value the less are the strains related to each other. Using these coefficient values, the software generates dendrograms according to the average linkage method, which means that the Southern blot hybridization patterns were compared to each other by calculating the average distance of all element pairs (bands) of both clusters.
Soll (22) defined several levels of relatedness, in which highly related means that the DNA fingerprints are highly similar but nonidentical and moderately related means that isolates group in a dendrogram in a cluster defined by a similarity coefficient threshold well above the average similarity coefficient for a set of presumed unrelated isolates. According to the values obtained with the Gelscan software (Euclidian distance), we set the threshold for a microevolution to a value of d(Eu) = 1.41, the value for highly related strains is d(Eu) = 2.24, and d(Eu) = 2.83 is used for moderately related strains. The results were also evaluated with a statistic register which is part of the Gelscan software package.
In all, 208 C. dubliniensis isolates recovered from 29 patients (25 HIV positive, 4 HIV negative) were subjected to Cd25 fingerprint analysis. This analysis could distinguish 58 different banding patterns, i.e., strains/genotypes, among the isolates (Fig. (Fig.1).1). A total of 53 of 58 strains were obtained from HIV-positive patients. Of the 58 strains, 55 were unique to individual patients, while 3 strains (genotypes 12, 25, and 43) were recovered from two different patients (Table (Table11 and Fig. Fig.11).
The cluster analysis showed two large groups at a node point of d(Eu) = 2.78: group A with 49 isolates and group B with 9 isolates. At d(Eu) = 2.64, group A isolates could be divided into two subgroups, with subgroup 1 containing 19 isolates and subgroup 2 containing 30 isolates (Fig. (Fig.1).1). The distance was in the range between 1 ≤ d(Eu) ≥ 2.78. The average distance for all isolates was d(Eu) = 1.67 ± 0.51, corresponding to a close relationship. The isolates of Cd25 group A had a mean d(Eu) = 1.68 ± 0.49, and the isolates of the Cd25 group B had a mean distance of d(Eu) = 1.53 ± 0.52.
We investigated whether these isolates could be assigned to the C. dubliniensis ITS genotypes described by Gee et al. (5). Genotype analysis was carried out by PCR and demonstrated that all strains could be assigned to C. dubliniensis genotype 1, and no genotype 2 isolates could be detected (Table (Table22).
In 21 of 29 patients, repeated cultures containing C. dubliniensis were recovered during follow-up visits (minimum of 2 weeks, maximum of 7 years and 8 months, mean = 32.3 months). In 6 of 21 patients, the same genotype of C. dubliniensis was detected from follow-up cultures. Strains exhibiting altered genotypes were recovered from the remaining 15 patients (Table (Table1).1). The genotypes of C. dubliniensis isolates from repeat cultures in seven of these patients were grouped in tight clusters, i.e., these isolates were highly related to the genotype identified in the initial sample and differed only through microevolutionary changes in fingerprint pattern (Fig. (Fig.2).2). However, isolates recovered from the remaining 8 patients exhibited a higher level of diversity [d(Eu) > 2.24]. Seven of these patients harbored C. dubliniensis strains from both subgroups 1 and 2 of Cd25 group A (Fig. (Fig.1)1) and could be considered to be independent, although closely related, clonal lineages. The remaining patient (i.e., patient 24) harbored less closely related strains belonging both Cd25 group A (genotype 29) and group B (genotypes 37 to 39, 42, 43, and 45).
The majority of the C. dubliniensis isolates (i.e., 200 of 208) were obtained from HIV-positive patients. The median distance d(Eu) = 1.71 ± 0.54 in this group was similar to the median distance for all isolates. The isolates from the HIV-positive patients grouped into 10 tight clusters. The eight isolates obtained from the four HIV-negative patients were arranged into five different groups, showing a median distance of d(Eu) = 2.02 ± 0.62. These isolates formed two loose clusters at a node of d(Eu) = 2.83. Therefore, isolates isolated from HIV-negative patients demonstrated a larger genetic variability than the isolates obtained from HIV-positive patients.
Four patients had an increased viral load (>36,000 copies/ml), and seven patients showed a lower viral load (<36,000 copies/ml). The median distance in the high viral load group was d(Eu) = 1.64 ± 0.65, and six tight clusters were found. In the low virus group the d(Eu) = 1.71 ± 0.56, and the C. dubliniensis genotypes were arranged in three tight clusters. So there was no correlation between the presence of genotypically different C. dubliniensis isolates and the viral load in HIV-positive patients.
Twelve patients with a CD4 count of <200/μl and seven patients with a CD4 count of >200/μl demonstrated median distances of d(Eu) = 1.82 ± 0.57 and d(Eu) = 1.68 ± 0.55, respectively. This difference was not statistically different. In both groups six tight clusters were observed. There was no correlation between the presence of genotypically different C. dubliniensis isolates and the Cd4 count in HIV-positive patients.
In nine patients (patients 4, 6, 8, 9, 13, 16, 19, 24, and 28) an OPC was observed during one of the follow-up visits at which oral rinsings were taken. The median distance of the isolates isolated from the OPC patients was d(Eu) = 2.10 ± 0.27, with two tight clusters and one loose cluster. The median distance of the isolates isolated from the non-OPC patients was d(Eu) = 1.67 ± 0.51, forming seven tight clusters and one single isolate not belonging to any cluster. We found a correlation between the detection of nine C. dubliniensis isolates belonging to eight different isolates and the presence of OPC in these patients.
The breakpoints for azole resistance were defined according to the 1997 NCCLS Guidelines, in which the MIC for fluconazole was ≥64 μg/ml and that for itraconazole was ≥1 μg/ml. We tested the azole susceptibility of 194 of 208 isolates and found 16 itraconazole-resistant isolates and 1 fluconazole-resistant isolate. These 17 resistant isolates represented 13 different Cd25 genotypes. Complete clinical information (OPC, antifungal therapy, etc.) was available for 14 of the 17 patients infected with azole-resistant strains. 88% (14 of 16) of the itraconazole-resistant C. dubliniensis isolates and 68% (114 of 167) of the sensitive isolates were obtained from patients who had received a prior antifungal therapy with fluconazole, but this difference was not statistically significant. In most cases, the development of azole resistance was associated with previous fluconazole usage. However, two patients harbored itraconazole resistant strains without any previous azole therapy (patients 25 and 29).
Itraconazole resistant isolates had MICs ranging from 1.56 to 50 μg/ml (Table (Table2).2). However, only one of these isolates exhibited cross-resistance to fluconazole (patient 10, fluconazole MIC > 40 μg/ml). However, three of these itraconazole resistant isolates exhibited reduced susceptibility to fluconazole (fluconazole MIC > 6 μg/ml) and could be considered susceptible dose dependent. However, the majority of itraconazole-resistant isolates did not exhibit cross-resistance to fluconazole and exhibited fluconazole-susceptible phenotypes (MIC < 4 μg/ml). The lack of cross-resistance could be due to a mutation in CDR1, the main mediator of azole cross-resistance in Candida spp. Indeed, genotype I C. dubliniensis isolates commonly carry a premature stop codon in CDR1, which renders this gene nonfunctional (12). We analyzed the itraconazole-resistant isolates described here for the presence of this mutation. The azole cross-resistant isolate from patient 10 harbored a functional copy of CdCDR1, indicating that CdCDR1 may be involved in cross-resistance to fluconazole and itraconazole in this isolate. However, four of the itraconazole-resistant isolates that were fluconazole susceptible also had functional copies of CdCDR1. The only isolate with two nonfunctional copies of CdCDR1 was the itraconazole-resistant isolate recovered from patient 16. This isolate exhibited the highest itraconazole MIC recorded here (50 μg/ml) and yet was fully susceptible to fluconazole (MIC 0.19 μg/ml).
We observed two different scenarios regarding genotype during the development of resistance: (i) In six patients (patients 3, 10, 13, 16, 19, and 23; Table Table2)2) resistance was associated with strain maintenance (i.e., the development of resistance in previously sensitive strains recovered form the same patient), and (ii) in three patients (patients 24, 25, and 29; Table Table2),2), resistance was associated with the emergence of a new genotype. One genotype associated with resistance was recovered from two patients (genotype 25 in patients 13 and 23). In patient 13, this strain was replaced during the development of resistance. However, in patient 23, this strain acquired resistance and was maintained during the development of resistance.
In contrast to previous publications in which independently collected isolates of C. dubliniensis were genotyped, we monitored a group of patients over several years (mean observation period, 32.3 months) and collected C. dubliniensis isolates at multiple visits. The aim of the present study was to determine (i) whether the patients were colonized with a single isolate only or with several isolates, (ii) whether there was a correlation between C. dubliniensis genotype and the clinical status of the patient (HIV status, viral load, immune status, and OPC), and (iii) whether there was a correlation between the appearance of genetically different strains and the development of resistance to azoles.
The median distance of all isolates was d(Eu) = 1.68 ± 0.51, i.e., all isolates showed little variability and were closely related. At a node of d(Eu) = 2.78, the dendrogram of all isolates could be split into two groups. Because of the low median distance in our isolates and the fact that the majority of our collection (53 of 58 isolates) was isolated from HIV-positive patients, it was likely that all isolates described in the present study belong to Cd25 group I as previously described (5, 7). PCR genotyping was carried out to confirm this using primers specific for the different C. dubliniensis genotypes (5). This demonstrated that all isolates in the present study belong to genotype 1. The majority of the 58 different isolates (55 of 58) were isolated from single patients and only 3 could be isolated independently from two different patients, respectively. These six patients had no relationship with each other, were seen independently from each other in the outpatient clinic and were all HIV positive. A similar observation was made by Gee et al. (5), but patients from whom identical isolates were isolated were treated primarily as inpatients in the same hospital.
Our longitudinal analysis of C. dubliniensis strain carriage identified three different scenarios of strain maintenance and replacement: (i) the same strain of C. dubliniensis could be recovered from 29% patients after repeat sampling, (ii) a strain exhibiting an altered but closely related fingerprint pattern could be recovered (i.e., microevolution of the infecting strain) in 33% of patients, and (iii) replacement of the infecting strain with a less related strain (38%). Only one of these patients (patient 24) harbored strains from both CD25 group A (genotype 29) and group B (genotypes 37 to 39, 42, and 43), indicating acquisition of a completely different clonal type. The mean distance, d(Eu), of clonal strains from one patient was generally ≤1, i.e., the old and the new strains in one patient were very closely related to each other, which is below the threshold for microevolutionary events [d(Eu) ≤ 1.41] as defined by Soll (22). An endogenous reinfection with an identical isolate originating in a different body part or an exogenous infection from a sexual partner may also explain how closely related genotypes arise in vivo. Only two sexual partners were analyzed in our cohort (patients 13 and 26), and they harbored isolates separated by a medium distance of d(Eu) = 1.43, i.e., the genetic relatedness of the isolates from the sexual partners was about the same as for all isolates. This means that we could not conclusively demonstrate exchange of C. dubliniensis isolates between sexual partners. Overall, these data suggest the notion that strain maintenance of the same or a closely related strain is the most common scenario in the oral cavity of HIV-infected patients colonized by C. dubliniensis and that minor changes in genotype are common over time. This is similar to the scenario outlined in C. albicans, with strain maintenance and “substrain shuffling” predominating (9).
A change of a strain could be due to the simultaneous presence of several clonal strains in the same clinical specimen but since only one colony was chosen for Cd25 fingerprint analysis, genetic diversity in the infecting population may have been overlooked. To examine this possibility, several yeast colonies (between 2 and 3) obtained from three different oral rinsings from one patient were taken and genotyped with Cd25 probe. It was shown that this patient was colonized with several genotypically different C. dubliniensis isolates, two of them closely related, whereas two other isolates were less related. We assume that during the course of the disease this patient was colonized with additional genotypes, while the original isolate persisted.
The C. dubliniensis isolates from HIV-positive patients were more closely related to each other than the C. dubliniensis isolates from HIV-negative patients [lower median distance of d(Eu) = 1.71 in the first group and d(Eu) = 2.02 in the latter group, respectively]. The isolates from patients without OPC were more homogeneous than the ones from the OPC group. However, a distinct genetic group was not formed by either the isolates of the patients with OPC or the isolates of the HIV-negative patients since they were all closely related. The isolates from patients with a higher viral load and from those with low CD4 cell count did not demonstrate a significant difference in genetic diversity than the isolates obtained from patients with a lower viral load and with high CD4 cell count, respectively.
It had been shown that C. dubliniensis is sensitive to azole antifungals such as fluconazole but may develop azole resistance in HIV-infected patients previously treated with fluconazole (23). In the present study, 1 isolate out of 194 was resistant to fluconazole, and 16 of 194 isolates were resistant to itraconazole. Interestingly, none of the patients from whom an itraconazole-resistant isolate was obtained had been treated with this drug. A total of 88% of the resistant C. dubliniensis isolates and 68% of the sensitive isolates were isolated from patients who were treated with fluconazole previously. This supports the notion that C. dubliniensis may develop an increased MIC for itraconazole after treatment with this drug (13, 16, 18). In addition, itraconazole-resistant isolates were recovered from patients with no previous azole exposure (patients 25 and 29). This finding suggests that resistance may develop in C. dubliniensis without previous azole exposure. Alternatively, resistant strains could be acquired from another patient that has been previously exposed to azole. However, we have no evidence of a relationship between patients 25 and 29 and patients who did receive azole therapy that would support the possibility of strain transmission.
In the majority of cases, the development of azole resistance occurred in strains already carried by the patient, i.e., maintenance of the original strain that acquired a resistant phenotype. The development of azole resistance was associated with a change in Cd25 genotype in only 3 of the patients examined. In two of these patients (patients 24 and 29) the new strain was closely related to the original infecting strain [d(Eu) < 2.24]. In addition to the new Cd25 genotype acquired in these two patients, they also maintained previous genotypes that independently acquired resistance. Overall, these data suggest that strain maintenance is more common than strain replacement or acquisition of new resistant strains in C. dubliniensis.
Reduced susceptibility to itraconazole has been observed previously in C. dubliniensis and was associated with increased expression of CdCDR1 or mutation of the ERG3 gene (12, 17). However, in these studies all of the itraconazole-resistant isolates were cross-resistant to fluconazole. In the present study, only one isolate exhibited cross-resistance to fluconazole (MIC > 40 μg/ml), while three isolates exhibited reduced susceptibility to fluconazole (MICs ≥ 6.25 μg/ml) and six isolates were fully susceptible to fluconazole (MIC < 4 μg/ml). Since cross-resistance to fluconazole and itraconazole is commonly mediated by the multidrug transporter CdCDR1, we hypothesized that the itraconazole-resistant isolates described in the present study may harbor nonfunctional copies of CdCDR1 containing a stop codon mutation described by Moran et al. (12). Analysis of the CdCDR1 gene in these isolates showed that four of the five itraconazole-resistant isolates that were fully fluconazole susceptible tested also had functional copies of CdCDR1. This finding was unexpected, since a functional copy of CDR1 would mediate azole cross-resistance in these isolates, suggesting that resistance in these isolates may involve non-CdCDR1 resistance mechanisms. Unexpectedly, the highest level of itraconazole resistance was observed in an isolate with two nonfunctional alleles of CdCDR1, indicating a CdCDR1-independent mechanism in this isolate. This resistance is unlikely to involve the mutations in ERG3, described by Pinjon et al. (17), since these mutations also result in azole cross-resistance, but could involve point mutations in ERG11. Further studies will be required in order to fully understand the resistance mechanisms at work in these isolates.
In conclusion, we found minor genetic alterations among our population of C. dubliniensis isolates. We found that strain maintenance was the most common scenario in these patients. Overall, all of the strains examined were closely related, although strains recovered from patients with HIV infection were more closely related than those from patients without HIV infection. Unexpectedly, ca. 8% of C. dubliniensis isolates exhibited resistance to itraconazole. This was often not associated with fluconazole cross-resistance and, in some patients, arose without previous azole therapy. Studies are currently under way to determine the mechanism of resistance in these isolates.
We thank S. Joly for the Cd25 probe, R. Jäger (BioSciTec) for help with the Gelscan software, and C. Radecke for excellent technical support.
Published ahead of print on 3 March 2010.