Cryptococcosis is a deadly opportunistic fungal infection primarily caused by Cryptococcus neoformans
and Cryptococcus gattii
. HIV infection is a major risk factor associated with the development of cryptococcosis and cryptococcal meningitis (CM), which cause 44% of HIV infection-related deaths in South Africa (1
). Cryptococcal meningitis is diagnosed in approximately 90% of HIV-infected patients who have cryptococcosis (2
) and results in a mortality rate of about 70% in low-income countries (4
) where access to antifungal drugs other than fluconazole is limited.
Antifungal treatment of cryptococcosis is limited to three classes of antifungal drugs: polyenes (amphotericin B), pyrimidine analogues (flucytosine), and azoles (fluconazole) (5
). Amphotericin B (AMB) binds to fungal membrane ergosterol, causing changes in membrane permeability, leakage of ions, and cell death (6
). Resistance to AMB is rare, from 0% to 5.8% for C. neoformans
and C. gattii
, respectively (reviewed in reference 6
). Decreased membrane ergosterol content or changes in membrane sterol composition are the major causes of AMB resistance (6
). Despite its renal toxicity, AMB remains the primary drug of choice for the treatment of cryptococcal infections in Western hospitals (7
). However, because of the need to administer AMB intravenously and its high price, AMB is rarely used in Sub-Saharan Africa, where HIV-associated cryptococcal meningitis is prevalent.
Flucytosine (5FC) is a prodrug: it is imported by cytosine permease and converted to 5-fluorouracil (5FU) by cytosine deaminase, which is a fungus-specific enzyme. 5FU is further processed by uracil phosphoribosyl transferase, which inhibits both DNA and protein synthesis (8
). 5FC resistance arises rapidly (7
), via mutations either in genes encoding the cytosine permease or the uracil phosphoribosyl transferase or in UXS1
, which encodes an enzyme that converts UDP-glucuronic acid to UDP-xylose for capsule biosynthesis (9
). Thus, 5FC monotherapy is discouraged, but according to the new WHO guidelines in 2018, 1 week of 5FC and AMB or 2 weeks of 5FC and fluconazole are recommended as induction regimens for cryptococcal meningitis (10
). Another report found that a combination of 5FC and fluconazole was more fungicidal against C. neoformans
than fluconazole alone (11
Fluconazole (FLC) is one of the most widely available and commonly used antifungal drugs. It is generally fungistatic, rather than fungicidal. FLC inhibits Erg11p, a cytochrome P450-dependent 14α-sterol demethylase that is essential for sterol biosynthesis. Inhibiting Erg11p prevents the conversion of lanosterol to ergosterol, depletes ergosterol, and causes the accumulation of toxic 14α-methyl sterols (12
), which are synthesized via an Erg3p-dependent bypass pathway. Cryptococcal resistance to FLC can be caused by point mutations of ERG11
Sionov and coworkers studied the acquisition of FLC heteroresistance, a transient increase in strain MIC, in strain H99 and found a stepwise acquisition of aneuploidies due to sequential duplication of specific chromosomes (15
). H99 adapted to 32 μg/ml FLC via Chr1 disomy, and this strain also acquired Chr1 disomy plus Chr4 disomy at 64 μg/ml FLC. Further adaptation to 128 μg/ml FLC was achieved by disomy of additional chromosomes. However, the aneuploids were unstable. After daily passage in the absence of FLC, aneuploid chromosomes, as well as the resistance phenotype, were lost. This unstable resistance due to aneuploidy in C. neoformans
is termed “heteroresistance” (16
). In vivo
, Chr1 disomy was frequently found in FLC-resistant colonies and was associated with clinical relapse during FLC monotherapy of HIV-associated cryptococcal meningitis patients (17
). Chr1 harbors ERG11
, which encodes the target of azole antifungals, as well as AFR1
, which encodes an efflux pump that reduces intracellular FLC concentrations (16
). Thus, in C. neoformans
, as in Candida albicans
, aneuploidy of one chromosome can mediate increased resistance by increasing the copy number of two genes that act additively (18
Here, we found that C. neoformans strain H99 simultaneously adapted to FLC via diverse aneuploidies when exposed to supra- or sub-MICs of FLC. Unexpectedly, some aneuploids were not heteroresistant to FLC, but rather, they became tolerant to AMB. Heteroresistant cells have high MICs and thus test as resistant to the drug, although the resistance phenotype is unstable; tolerance is defined as the ability of susceptible isolates to grow in the presence of supra-MIC drug levels. Other aneuploids became heteroresistant to FLC and tolerant to 5FC. Importantly, a short time of exposure to sub-MIC FLC concentrations was sufficient to select aneuploid adaptors that, depending upon the karyotype, were heteroresistant to FLC and tolerant to 5FC or FLC and hydrogen peroxide (H2O2). Thus, the plasticity of the C. neoformans genome challenges combinations of therapeutic strategies used to treat cryptococcosis.
By isolating FLC adaptors following direct exposure to supra- and sub-MICs of FLC, we detected rapid adaptation within a single cycle of drug exposure. Furthermore, independent adaptors derived from one FLC concentration acquired diverse aneuploid karyotypes, with multiple isolates that had the same karyotype exhibiting very similar phenotypes with respect to FLC heteroresistance and other drugs. Interestingly, over 90% of the adaptors were aneuploid, and the majority carried Chr1 disomy and/or Chr4 disomy, irrespective of the FLC concentration used for selection. Importantly, most of the tested aneuploid colonies that had been selected on FLC were also cross-adapted to 5FC and/or AMB, without prior exposure to either 5FC or AMB. This raises a serious concern, given that exposure to the most widely used antifungal drug, FLC, rapidly yielded isolates tolerant to at least two and even all three of the antifungal drug classes available to treat cryptococcal meningitis.
The rapid adaptation to FLC and its association with diverse aneuploid karyotypes may be explained by the types of mitotic defects incurred when either C. albicans
or C. neoformans
is exposed to FLC. Specifically, a subpopulation of FLC-exposed cells undergoes a transient arrest in cell cycle progression that includes defective cytokinesis. This, in turn, yields tetraploid cells with chromosome instability that subsequently undergo high levels of chromosome loss, resulting in high levels of aneuploidy (21
). An alternative possibility is that aneuploids are formed directly upon exposure to FLC (24
Karyotypic diversity arises from C. neoformans
polyploid titan cells exposed to FLC as well. The genome sequences of four progeny of a single titan cell revealed one with Chr1 disomy, one with 2 copies of the right arm of Chr1, one with Chr4 disomy and Chr6 disomy, and one with Chr4 disomy and Chr10 disomy (25
). These polyploid titan cells are intrinsically unstable, rapidly returning to haploidy in the absence of stress (25
). Whether this mechanism is similar to or distinct from the mechanism of FLC-induced aneuploidy in haploid cells remains to be determined.
Almost all aneuploid adaptors (33 out of 36) were heteroresistant to FLC, with specific karyotypes causing higher heteroresistance. Colonies selected from 32 μg/ml FLC mostly (18 out of 27) had Chr1 disomy alone and a MIC of 32 μg/ml, colonies selected from 64 μg/ml FLC mostly (7 out of 8) had a MIC of 64 μg/ml, and colonies selected from 128 μg/ml FLC all had a MIC of 128 μg/ml. All three strains selected on 128 μg/ml FLC were disomic for Chr1 and Chr4 or a segment of Chr4, with two having additional chromosomes, including Chr6 and Chr10. These results are consistent with a previous report that sequential exposure of strain H99 to elevated concentrations of FLC resulted in a stepwise accumulation of disomies of different chromosomes (Chr1 disomy, then Chr1 disomy plus Chr4 disomy, and then additional Chr10 disomy with or without Chr14 disomy) (16
). However, in this study, H99 was directly exposed to different concentrations of FLC until the first colonies appeared, with longer times required for colony appearance at higher FLC concentrations. This may be because the arrest of cell cycle progression in cells was dose dependent. Of note, some karyotypes caused FLC MICs higher than the selective conditions. For example, Chr4 disomy alone and Chr4 disomy plus segmental disomy of Chr1 appeared under 32-μg/ml FLC selection and enabled growth in up to 64 μg/ml of FLC, and one isolate disomic for Chr1, 2, -4, -5, -10, and -14 appeared under 64 μg/ml FLC selection and was able to grow in up to 128 μg/ml FLC. Therefore, instead of a stepwise increase of resistance, some aneuploid karyotypes enabled increased FLC tolerance in a single selective step.
Adaptors with the same karyotypes exhibited similar profiles of antifungal drug tolerance, which is consistent with the hypothesis that altered drug responses are due to aneuploidies in these strains. Nonetheless, not all disomic chromosomes have an additive effect on the ability to grow in FLC. For example, isolates with whole-Chr1 disomy or segmental disomy of 1.92 Mb of Chr1 had an FLC MIC of 32 μg/ml. Strains with Chr1 disomy plus Chr6 disomy retained the same MIC as strains with Chr1 disomy alone, and an isolate with Chr6 disomy alone had the same MIC as the parent. These results imply that Chr6 disomy does not affect the FLC MIC and that it is neutral, providing little fitness benefit or cost in FLC. Chr4 seemed to have different effects in different karyotypic contexts: when Chr4 disomy was the only aneuploidy, the strain had a MIC of 64 μg/ml. In addition, the following karyotypes that included Chr4 had the same MIC as when Chr4 disomy was the only aneuploidy: Chr1 disomy plus Chr4 disomy, segmental-Chr1 disomy plus Chr4 disomy, or Chr1 disomy plus Chr4 disomy plus Chr14 disomy. This suggests that Chr14 may also be neutral, with little or no benefit in FLC, However, Chr1 disomy plus segmental-Chr4 disomy had a higher MIC (128 μg/ml) than either Chr1 disomy or Chr4 disomy alone or disomy of both chromosomes, illustrating the potential for multiple genes on the same chromosome to modulate the dose-responsiveness of growth on FLC. Adaptors with more-disomic chromosomes usually had high MICs. For example, adaptors with disomy of 4, 6, or 7 chromosomes (Chr1, -4, -6, and -10; Chr1, -2, -4, -6, -10, and -14; or Chr1, -3, -4, -5, -10, -12, and -14, respectively) were able to grow in up to 128 μg/ml FLC.
Unexpectedly, three isolates selected on FLC did not grow better when retested on FLC. These included isolates with the following three karyotypes: Chr6 disomy alone, Chr2 disomy plus Chr4 disomy, and Chr3 disomy plus Chr14 disomy. One possible explanation for their selection and subsequent loss of the ability to grow in FLC is that they initially carried aneuploid chromosomes that conferred resistance and subsequently lost them when they were propagated without drug (between the selection and retesting experiment). Transient resistance (heteroresistance) to FLC due to unstable aneuploidy is well documented in C. neoformans
(reviewed in reference 16
). In Candida albicans
, drug tolerance due to aneuploidy is also unstable in the absence of drug selection (26–28
Importantly, all 13 karyotypes that caused heteroresistance to FLC were cross-tolerant to 5FC. Among them, 11 karyotypes were Chr1 disomic alone or in combination with at least one other disomic chromosome. In a recent study of H99, as well as other C. neoformans
and C. gattii
strains, 5FC selection caused Chr1 disomy that was attributed to AFR1
, a gene on Chr1 that encodes an efflux pump. However, the effect of increased AFR1
expression on 5FC tolerance was strain dependent (29
). Taken together, these studies indicate that Chr1 disomy can be selected by growth on FLC or 5FC and that it confers cross-tolerance to both FLC and 5FC, irrespective of the drug used for the initial selection. This is reminiscent of Chr5 monosomy in C. albicans
, which causes cross-tolerance to l
-sorbose, caspofungin (fungicidal drug of the echinocandin family), and 5FC (26
), and Chr2 trisomy, which causes cross-tolerance to caspofungin and hydroxyurea (a commonly used chemotherapeutic drug) (27
). Thus, cross-tolerance to unrelated drugs via aneuploidy is a concern common to many pathogenic fungi and, in these studies, to both basidiomycete and ascomycete yeasts.
Of particular concern is the Chr1-plus-Chr4 disomy karyotype, which appeared twice among the isolates studied here. This karyotype conferred cross-tolerance to drugs from three classes of antifungals: FLC, AMB, and 5FC. Since these three drugs hail from the three major drug classes in clinical use, this type of multidrug tolerance has the potential to impede effective clinical treatment of cryptococcal infection. Importantly, the acquisition of two aneuploid chromosomes, whether in a single selection step or in a more gradual manner, has the potential to reduce the efficacy of the only available treatment options. Furthermore, given the use of AMB in induction therapy protocols, it will be important to investigate whether primary exposure to AMB gives rise to cross-tolerance to FLC and 5FC as well.
Aneuploid adaptors were selected after as little as 48 h of exposure to sub-MIC FLC concentrations (8 μg/ml). Among 120 random colonies, 5 had Chr1 disomy and one (segmental disomy of Chr1) was disomic for 62 kb of Chr1 that includes ERG11
but not AFR1.
Therefore, aneuploidy enabled rapid adaptation to both sub- and supra-MICs of FLC, and a short time of exposure to sub-MIC FLC was sufficient to select for aneuploidies that enabled tolerance to supra-MIC FLC. The heteroresistance in the segmental-Chr1 disomy was lower than for whole-Chr1-disomic aneuploids, probably because that segment includes only ERG11
and not AFR1
. Furthermore, duplication of whole Chr1 caused FLC and 5FC cross-tolerance, while duplication of one 62 kb region on Chr1 (adaptor L8-6) caused FLC and H2
cross-tolerance but not 5FC tolerance. Given that an extra copy of Erg11 on Chr1 is a major cause of FLC heteroresistance (16
), this lack of 5FC tolerance in L8-6 implies that the ability to grow on 5FC resides elsewhere on Chr1, outside the 62 kb chromosomal segment. Whether the ability to survive H2
resides within the 62 kb retained in strain L8-6 remains to be determined.
In conclusion, C. neoformans strain H99 directly adapted to FLC at a range of concentrations. The adapted isolates had a range of karyotypes, with the most recurrent being Chr1 disomy, which conferred tolerance to FLC as well as to 5FC. Chr4 was the next-most-recurrent aneuploidy and, most importantly, two independent isolates with disomy of both Chr1 and 4 were able to grow in drugs from all major anticryptococcosis drug classes, raising concerns about the speed with which multidrug tolerance can arise in C. neoformans.