Carbon Dioxide Potentiates Flucytosine Susceptibility in Cryptococcus neoformans

ABSTRACT

Cryptococcal meningoencephalitis remains a global health threat with limited treatment options. Currently, the most effective treatment regimens are based on a combination therapy of flucytosine with either amphotericin B or fluconazole. Slow but steady progress is being made toward universal access to flucytosine-based therapies. The broadening access to flucytosine combination therapies will be accompanied by the need for microbiological methods that reliably determine strain susceptibility. This is especially true considering that flucytosine susceptibility can vary widely across clinical isolates. Identifying culture conditions that best represent the host environment are likely optimal and may even be required for accurately determining in vivo flucytosine susceptibility. Here, we report that culture conditions incorporating host-like concentrations of carbon dioxide (CO₂) potentiated flucytosine susceptibilities across clinical isolates (10 of 11) that exhibited a range of MIC values under ambient growth conditions (2 to 8 μg/mL) by standard Clinical and Laboratory Standards Institute susceptibility testing. CO₂ induced a dose-dependent increase in flucytosine susceptibility between 2- and 8-fold over standard conditions. The CO₂-dependent increase in flucytosine susceptibility did not correspond to an increase in fluorouracil susceptibility, indicating a central role for flucytosine uptake through the cytosine permease in the presence of host-like CO₂ concentrations. Indeed, the expression of the cytosine permease gene (FCY2) was induced 18- to 60-fold in the mouse lung environment. Therefore, the activity of flucytosine is likely to be very dependent upon host environment and may not be well represented by standard in vitro susceptibility testing.

IMPORTANCE Cryptococcus neoformans causes life-threatening infections of the brain. The most effective treatment regimens are based on flucytosine-based combination therapy, which has led to increasingly successful broadening of access to flucytosine globally. Wider use of flucytosine-based therapies for cryptococcal infections will require the ability to reliably determine clinical isolate susceptibilities. We showed that host-like carbon dioxide stress affected flucytosine susceptibility, and this likely occurred through flucytosine uptake. We further showed that the gene encoding the permease, FCY2, and that is responsible for flucytosine uptake was strongly induced during cryptococcal infection. Our data provide insights into the distinctions between the activity of flucytosine in the host environment and during in vitro susceptibility testing.

OBSERVATION

Cryptococcus neoformans is a frequent cause of life-threatening fungal infections that primarily occur in immunocompromised patient populations (1). Although cryptococcosis most commonly affects people living with HIV, its prevalence is also a concern among patients receiving immune-suppressive therapies to manage autoimmune disease or solid organ transplantation. Amphotericin B (a polyene) is the long-standing gold standard treatment for cryptococcosis. Unfortunately, amphotericin B often requires hospitalization and is therefore a less amenable treatment option in resource-limited regions of endemicity. This has left the more tolerated yet far less effective fluconazole (an azole) as a frontline therapy for some regions of the world. Beyond polyenes and azoles, no other class of antifungals serves as a viable anticryptococcal monotherapy. A third antifungal drug, flucytosine (5-FC), improves treatment outcomes when combined with either amphotericin B or fluconazole. Clinical trials investigating either amphotericin B or fluconazole in combination with flucytosine showed a 1.5- to 3-fold increase in survival over the respective monotherapies (2, 3). Moreover, flucytosine has been reported to suppress heteroresistance to fluconazole (4). The accumulating clinical data strongly support the use of flucytosine as an adjunctive therapy. Fortunately, flucytosine is becoming more accessible and more widely adopted. Understanding the mechanisms that direct flucytosine as an effective adjuvant will be important for identifying the underlying factors that determine treatment outcome.

The apparent in vivo synergistic effects offered by flucytosine combination therapy have been somewhat perplexing compared to the in vitro data. Clinical isolates show a lower rate of synergism in vitro compared to the positive in vivo outcomes observed in combination with amphotericin B or fluconazole (5, 6). This disconnect between in vitro and in vivo flucytosine efficacy has also been observed in Aspergillus fumigatus and less so for Candida albicans (7–9). Discrepancies between in vitro testing conditions and the in vivo infection environment may contribute to this discordance. As part of a project to identify in vitro Cryptococcus susceptibility testing conditions for flucytosine that better correlated with in vivo outcomes, Viviani and colleagues found that yeast nitrogen base rather than the standard RPMI medium and pH 5.4 rather than pH 7.0 were conditions that best correlated with in vivo flucytosine efficacy (10). Separately, Verweij and colleagues observed that testing flucytosine susceptibility against Aspergillus at pH 5.0 rather than pH 7.0 was a far better predictor of in vivo efficacy (9). Later, the pH dependence of flucytosine activity against Aspergillus fumigatus was determined to result from a pH-dependent regulation of the cytosine permease, encoded by FCYB, that mediates flucytosine uptake (11). We present data that this mechanism is also present for C. neoformans but that an alternative stress derived from host carbon dioxide (CO₂) concentrations may contribute to enhanced flucytosine susceptibility in vivo.

For environmental fungi such as C. neoformans and A. fumigatus to initiate infection, they must first adapt to the lung environment. We have recently shown that an elevated CO₂ concentration (~5%) in the lung relative to ambient air (0.001%) is an independent stress, separate from pH, that C. neoformans must overcome to establish infection in mammalian hosts (12). Additionally, we reported that host-like concentrations of CO₂ (5%) potentiated the antifungal activity of azoles, including fluconazole, toward C. neoformans. In contrast, CO₂ does not affect susceptibility to polyenes such as amphotericin B (12). To understand the underlying mechanisms contributing to these observations, our ongoing work (to be reported elsewhere) includes RNA sequencing (RNA-seq) analysis to assess the transcriptional response of the C. neoformans reference strain H99 to CO₂ stress in RPMI-morpholinepropanesulfonic acid (MOPS) medium at pH 7.0, with or without 5% CO₂. This work uncovered the fact that expression of the cytosine permease gene FCY2 was elevated in cultures incubated at 5% CO₂ relative to those under ambient CO₂ conditions (fold increase of 5.2 ± 0.6 under 5% CO₂ versus ambient conditions, mean ± standard deviation [SD]; n = 3) (Fig. 1B). We repeated this experiment and confirmed this observation through quantitative PCR (qPCR) analysis (4.1 ± 2.5 increase under CO₂ versus ambient; n = 3) (Fig. 1B). We were curious if this CO₂-mediated phenomenon could potentiate flucytosine susceptibility. Multiple clinical isolates were tested using the standard CLSI method of antifungal susceptibility testing under either ambient air conditions or under CO₂ (5%) supplemented air. Across 11 clinical isolates of C. neoformans (generously provided by John Perfect at Duke University), 8 isolates exhibited a 2-fold reduction in the flucytosine MIC, indicating a modest but consistent increase in flucytosine susceptibility in the presence of host CO₂ concentrations (Fig. 1C). This CO₂ effect was dose dependent, and 6/11 isolates exhibited a 2- to 8-fold reduction in the MIC under elevated CO₂ (10%) conditions at 37°C, while the remaining 5 isolates grew too slowly under 10% CO₂ for MIC determinations (Fig. 1C). We then repeated 5-FC determinations under ambient and 10% CO₂ with 30°C, and 9/11 isolates exhibited a 2- to 64-fold reduction in MIC, with an average reduction of 15-fold and with 1 isolate unable to grow under 10% CO₂ (Fig. 1D). The increase in flucytosine susceptibility occurred across strains that exhibited a range of MIC values under ambient growth conditions (2 to 8 μg/mL). The single clinical isolate that did not exhibit increased flucytosine susceptibility under CO₂ culture conditions had a very high MIC and was likely resistant (64 μg/mL).

FIG 1.

Carbon dioxide potentiates 5-FC susceptibility through FCY2 in C. neoformans. (A) Schematic of 5-FC uptake through the cytosine permease (FCY2) and processing by the cytosine deaminase via FCY1 to 5-FU. (B) Relative expression of FCY2 in RPMI-MOPS (pH 7.0) with RPKM reads in 5% CO₂ relative to that under ambient air conditions for RNA-seq and threshold cycle (ΔΔC_t) of FCY2 normalized those of ACT1 for qPCR. (C) MICs determined via CLSI standards modified with 5% or 10% CO₂ and normalized to ambient air conditions for 5-FC. Inferential statistics were performed using GraphPad Prism and a one-sample t test against ambient conditions. ***, P < 0.001. (D) MICs determined via CLSI standards with ambient air or 10% CO₂ at 30°C; asterisk denotes no growth under tested conditions. (E) MICs determined via CLSI standard methods under ambient air for 5% CO₂ for 5-FU; asterisk denotes no growth under tested conditions. (F) Relative expression of FCY2 on day 4 of intranasal pulmonary murine infection, normalized to expression of ACT1 and compared to in vitro ambient RPMI-MOPS (pH 7.0) growth conditions via qPCR. Inferential statistics were performed using GraphPad Prism and a one-sample t test against ambient in vitro conditions. *, P < 0.05; **, P < 0.01. All clinical strains were C. neoformans var. grubii and were generously provided by John Perfect at Duke University. Numbers denote strain identifiers.

5-FC is a synthetic fluorinated analog of cytosine that relies on uptake through the cytosine permease (FCY2) (Fig. 1A). Once in the cell, flucytosine is a prodrug that is first converted to 5-fluorouracil (5-FU) and is either incorporated into RNA to inhibit protein synthesis or into DNA to inhibit replication (13). The most common mechanism of flucytosine resistance is loss of Fcy2 function, which prevents flucytosine uptake (13). To assess whether CO₂ exerts its effect on flucytosine susceptibility through modulation of Fcy2, we characterized the activity of 5-FU against the same clinical isolates in the presence and absence of CO₂ (5%). Uptake of 5-FU was Fcy2 independent. Therefore, CO₂ concentrations would have no effect on 5-FU activity if it is acting through its permease function. CO₂ did not increase the 5-FU susceptibility of the tested strains, consistent with an Fcy2-dependent mechanism for CO₂-induced flucytosine potentiation (Fig. 1E). This observation is consistent with our RNA-seq data, which showed a >2-fold induction of the cytosine deaminase Fcy1 responsible for converting 5-FC to 5-FU (data not shown). Together, these data suggested that CO₂ modulates 5-FC susceptibility by modulating cellular uptake rather than processing and incorporation.

Next, we characterized how in vivo conditions within the lung affected FCY2 expression. To assess this, we inoculated A/J mice intranasally with 5 × 10⁵ CFU of either the reference strain H99 or the clinical isolate C23. Lungs of the infected mice were harvested on day 4 postinoculation, and RNA was extracted for qPCR analysis of FCY2 expression in vivo. The H99 (17.7 ± 2.9 increase, in vivo versus in vitro; n = 3) and C23 (60 ± 18.4 increase in vivo/in vitro; n = 5) exhibited higher FCY2 expression in vivo under in vitro culture conditions than under ambient conditions (Fig. 1F). This increased FCY2 expression in vivo exceeded the increased expression due to CO₂ stress alone (Fig. 1B). This suggested that CO₂ as well as other factors within the in vivo environment contribute to elevated expression of FCY2 in infected lung relative to in vitro susceptibility testing conditions.

To further explore the role of host levels of CO₂ in modulating flucytosine susceptibility, we asked whether CO₂ rather than pH could also serve as the in vivo signal potentiating flucytosine susceptibility against A. fumigatus. We assessed this by using a radial growth assay in glucose minimal medium (GMM) with MOPS buffered at pH 7.0 with 125 μg/mL flucytosine at 37°C under ambient or 5% CO₂ supplemented air (Fig. 2). No reduction in radial growth was observed under changed CO₂ conditions, leaving pH-dependent flucytosine potentiation as a the currently supported mechanism. It is possible that exposure to the low-pH environment of the phagolysosome of alveolar macrophages is a driving force behind the pH dependence of flucytosine efficacy. Alternatively, tight control of lung pH has also been shown to be greatly depend on rapid and efficient gas exchange when injured lungs experience a significant drop in pH (14). Perhaps immune infiltration and lung damage associated with disease progression are enough to disrupt the gas exchange and buffering capacity of the lung. Repeating our radial growth assay as before, but without MOPS buffer, revealed 5-FC potentiation with reduced radial growth under 5% CO₂ relative to plates incubated at ambient CO₂ concentrations at later time points (144 h) (Fig. 2).

FIG 2.

Carbon dioxide does not potentiate 5-FC susceptibility in A. fumigatus. Radial growth assay of 10³ conidia of A. fumigatus strain CEA10 on GMM-MOPS buffered at pH 7.0 versus unbuffered GMM plates grown under ambient or 5% CO₂ supplemented air. Marked circles outline the edge of radial growth for the matched ambient growth conditions.

In vitro flucytosine susceptibility is highly dependent on the specific culture conditions and does not necessarily correlate well with its activity in animal models or for patient outcomes. Recently, this susceptibility was linked to pH-dependent expression of the permease required for flucytosine uptake in A. fumigatus (11). Our data indicate that CO₂ concentrations affect 5-FC activity toward A. fumigatus but that this effect indirect through modulation of medium pH. In contrast, we also found that the activity of flucytosine toward C. neoformans was increased by host-relevant concentrations of CO₂ but was independent of pH. Finally, compared to standard in vitro susceptibility testing conditions, the expression of the Fcy2 permease required to import 5-FC appears to be expressed at much higher levels during mammalian infection, possibly explaining its significant in vivo efficacy compared to standard in vitro susceptibility testing Taken together, these observations suggest that antifungal testing under conditions that better approximate the host environment may be useful for some antifungal drugs, such as flucytosine.