The TOR and cell wall integrity MAPK pathways play opposite roles during CO2 stress
Protein kinases (PK) and transcription factors (TF) are key regulators of the cellular response to extracellular stresses. To identify PKs and TFs that affect CO2 tolerance, we screened systematic deletion mutant libraries covering the majority of non-essential PKs and TFs in the CO2 tolerant C. neoformans strain background H9911,12. Host levels of CO2 are fungistatic to intolerant strains and mutants rather than fungicidal10. To generate quantitative data on the relative fitness of these mutants, we developed a competitive fitness assay (Fig. 1a). H99 expressing mNeonGreen was used as a reference strain and inoculated into a 96-well microtiter plate with an unlabeled PK or TF mutant at a 1:1 ratio in host-relevant RPMI 1640 medium buffered to pH 7 at 30 °C under either ambient or 5% CO2 for 24 h. The screen was performed at 30 °C rather than at the host temperature of 37 °C so that the CO2 phenotype of temperature sensitive mutants could be assessed. The relative ratio of mutant to reference was determined by flow cytometry and the ratio at 5% CO2 was normalized to growth under ambient air to generate a CO2 fitness score. Control experiments comparing H99 to CO2-intolerant strains demonstrated that the assay identified strains with fitness defects at 5% CO2 (Supplementary Fig. 1a).
We screened a total of 129 PK and 155 TF mutants in duplicate11,12. Scatter plots summarizing the screens are shown in Fig. 1b & c. A total of 21 PK mutants showed statistically significant alterations in competitive fitness relative to H99 with 14 mutants hyper-susceptible to CO2 while 7 mutants displayed increased fitness. Importantly, the cbk1∆ mutant showed reduced fitness which corroborates Chadwick et al. who showed that the RAM pathway is important for CO2 tolerance13. In contrast, all but one of the TF mutants with altered CO2 fitness were more resistant than H99. The full results of the screen are provided in Supplementary Data 1 and 2. We re-constructed 7 PK and 4 TF deletion mutants with altered CO2 fitness and determined their competitive fitness in independent experiments (Fig. 1d, e). The majority of mutants tested on agar plates showed reduced growth, although some mutants (e.g., sch9∆ bwc2∆) only showed a phenotype in liquid media; this is likely due to the increased sensitivity of direct competition assay (Supplementary Fig. 1b).
Two major protein kinase signaling pathways emerged from this data set. First, deletion mutants of YPK1, SCH9, and GSK3 were hypersusceptible to elevated CO2 concentrations and all are part of the TOR pathway11,14, strongly suggesting that TOR is required for CO2 tolerance (Fig. 1b, d). Although Ark1 has not been characterized in C. neoformans, its S. cerevisiae homolog negatively regulates Tor-dependent endocytosis, tying it to the TOR pathway as well15. Second, deletion of the MAPKKK, MAPK, and MAPK kinases (MKK1, BCK1, and MPK1) of the cell wall integrity (CWI) pathway increased CO2 fitness (Fig. 1b, d), suggesting that this pathway may be maladaptive for CO2 stress. Importantly, the TOR and CWI pathways have been shown to negatively regulate the other in both S. cerevisiae and C. neoformans14,16.
The only TF mutant that showed reduced fitness (albeit modest) in 5% CO2 was BWC2, a component of the blue-white light sensing system in C. neoformans17. In contrast, multiple TF deletion mutants showed increased fitness in 5% CO2 relative to H99. Among these, a previously defined pathway of Yap1, Gat201 and Gat20418 emerged (Fig. 1c). Gat201 and Gat204 are GATA TFs that regulate multiple virulence traits in C. neoformans19. Additionally, Gat201 binds the GAT204 promoter19 and regulates its expression of Gat204 (Fig. 1c). Furthermore, Jang et al. found that Yap1 functions upstream of Gat201 during capsule induction18. The consistent phenotypes observed across components of previously described regulatory pathways provide confidence that these genes are important for CO2 responses.
Finally, the deletion mutant of Rim101, a TF that is critical for adaptation to host pH20, is also CO2 resistant (Fig. 1c, e). The cAMP-PKA pathway functions through Rim101 under some conditions21 and the pka1∆ mutant is modestly resistant to CO2 as well (Fig. 1b). From these genetic data, it appears that signaling pathways that are required for adaptation to other host-related environmental stresses such as elevated temperature (CWI pathway) and alkaline pH (Rim101 pathway) may reduce adaptation to host CO2 concentrations. It is also important to note that 9 PKs not related to the TOR pathway with diverse functions are also required for CO2 tolerance (Fig. 1b), further supporting the notion that it represents a significant cellular stress for C. neoformans.
The TOR pathway is required for C. neoformans tolerance of elevated CO2 concentrations
A single, essential Tor kinase is present in C. neoformans and, consequently, the construction of TOR1 deletion mutants is not possible14. To test the hypothesis that CO2 tolerance is dependent on TOR, we compared the antifungal activity of the TOR inhibitor rapamycin (RAP) under ambient and 5% CO2 conditions. RAP was more potent in 5% CO2 compared to ambient conditions (Fig. 2a). RAP was also more active against an environmental strain, A7-35-23, at elevated CO2 (Fig. 2b). Because deletion mutants of YPK1 and SCH9 are both CO2 sensitive and hypersensitive to RAP14, we asked if resistance to CO2 of TF mutants correlated with RAP resistance. Interestingly, rim101∆ mutants were relatively resistant to RAP at elevated CO2 compared to H99 but not under ambient conditions (Fig. 2c). Thus, it is possible that these TF mutants have elevated TOR pathway activity at baseline as a compensatory response to the defects induced by the mutations and this leads to increased fitness in elevated CO2.
The TOR pathway is a conserved regulator of many cellular processes including ribosome biosynthesis, protein translation, amino acid transport, autophagy, actin polarization, and membrane homeostasis among others22. Previous studies of the C. neoformans TOR pathway confirmed these conserved functions14. Specifically, inhibition of TOR with RAP down-regulated genes involved in rRNA processing, ribosome biogenesis, and actin cytoskeleton while transmembrane transporters and carbohydrate metabolic genes were up-regulated14. To further test the hypothesis that CO2 triggers a TOR-mediated cellular response, we characterized the transcriptional profile of H99 cells grown in buffered RPMI medium at 37 °C under ambient or 5% CO2 concentrations (Supplementary Data 3; (Wald test p values and Benjamini-Hochberg adjustment for multiple comparisons)). Somewhat surprisingly, CO2 induced a relatively slow transcriptional response with only 9 genes differentially expressed (adjusted P value < 0.05, log2 ± 1) relative to ambient conditions after a 4-hour exposure to 5% CO2 (Fig. 3a).
Eight hours after exposure to 5% CO2, 199 genes were differentially expressed with 192 of those genes downregulated. This set of downregulated genes was enriched for membrane (FDR .012, Benjamini-Hochberg) and integral membrane proteins (FDR 0.024); indeed, 60 of the 199 differentially expressed genes were membrane-associated proteins. By 24 h; however, a total of 1204 genes were differentially expressed (log2 ± 1, FDR < 0.05) with the expression of 528 genes downregulated and 677 upregulated (Fig. 3a). GO term analysis (FDR < 0.05, Benjamini-Hochberg) indicates that the upregulated genes are enriched for rRNA processing, ribosome biogenesis, DNA replication and aromatic compound synthesis (Fig. 3b and Supplementary Data 4) while the down-regulated genes were enriched for transmembrane transport, carbohydrate metabolism, redox process, and cellular response to heat (Fig. 3c and Supplementary Data 4). As indicated by the Venn diagram shown in Fig. 3d, 22% of the genes upregulated in CO2 (151/677) are downregulated in cells treated with RAP while 33% (176/528) of genes downregulated in CO2 are upregulated in the presence of RAP. Overall, the transcriptional response to CO2 occurs over multiple hours and appears to have two phases: an early phase that is highly enriched for membrane-associated genes and a late phase that is consistent with the activation of the TOR pathway. The delayed nature of this response would suggest that gene expression may be compensating for the physiological and biophysical effects of elevated CO2 rather than through a direct sensing of the elevated CO2 concentrations.
Elevated CO2 suppresses CWI MAPK signaling in CO2-tolerant but not intolerant strains
Genes involved in the cellular response to heat are downregulated when cells are shifted from ambient to host-like concentrations of CO2 (Fig. 3c). The CWI pathway is a key positive regulator of the C. neoformans response to elevated temperature23,24,25. Because deletion mutants of PKs in the CWI pathway (BCK1, MKK2, MPK1) in H99 background are resistant to CO2 (Fig. 1b, d), it appears that the expression of temperature stress-related genes may be maladaptive during CO2 stress. Indeed, we have previously reported that elevated temperature exacerbates CO2 stress13. In S. cerevisiae and C. neoformans, activation of the TOR pathway negatively regulates the CWI pathway14,16, suggesting the hypothesis that 5% CO2 may blunt CWI pathway activation. To test this hypothesis, we used a phospho-specific antibody to monitor the phosphorylation of Mpk1, the terminal MAPK of the CWI pathway, following temperature shift from 30 °C to 37 °C in ambient or 5% CO2 (Fig. 4a). Consistent with previous literature23,24, Mpk1 phosphorylation is increased in CO2-tolerant H99 at 37 °C in ambient air conditions and this level of phosphorylation is maintained for the 4-hour time course. In 5% CO2, temperature-induced Mpk1 phosphorylation was delayed and was consistently reduced relative to ambient CO2 conditions over the time course (Fig. 4a, left panel). These data clearly demonstrate that the activation of the CWI pathway is blunted by host levels of CO2.
We next asked if temperature-induced Mpk1 phosphorylation was also blunted by 5% CO2 in a CO2-sensitive strain. To do so, we performed the same temperature-shift experiment with A7-35-233,10, a CO2 sensitive environmental isolate (Fig. 4a, right panel). Mpk1 phosphorylation increased at 37 °C in A7-35-23 but the extent of Mpk1 phosphorylation was unaffected by growth in 5% CO2. These data, combined with the genetic and transcriptional results discussed above, support a model in which CO2 suppresses temperature-induced activation of the CWI pathway and that this suppression may improve tolerance to host CO2 concentrations.
Reduced Rim101 pathway activity leads to CO2 and rapamycin resistance
The Rim101 pathway regulates cellular responses to alkaline pH, capsule formation, and cell wall biosynthesis20,21,26; our data indicate that it, too, is maladaptive during CO2 stress (Fig. 1c, e). Rim101 is activated by proteolysis which in turn is mediated by a well-defined protein complex in C. neoformans (Fig. 4b). Deletion mutants of two additional components of the Rim101 pathway, the calpain protease Rim13 and the scaffolding protein Rim20, were also resistant to CO2 relative to H99 (Fig. 4c). Next, we characterized the effect of CO2 on the proteolytic processing of Rim101 using western blot analysis of cells containing a Rim101-GFP allele as previously reported20,21. Under ambient CO2 in buffered RPMI medium, both unprocessed (~150 kD band; FL) and processed (~100 kD band; P) Rim101-GFP are detectable (Fig. 4d); additional degradation products are also present in the blot. In 5% CO2, neither the processed nor the unprocessed forms of Rim101 are detectable and increased amounts of a band corresponding to free GFP are present. These data suggest that Rim101 is degraded in 5% CO2 and are consistent with the idea that Rim101 activity is reduced to compensate for CO2 stress.
We next characterized the effect of the Rim101 pathway on the transcriptional response to growth at 5% CO2. The rim101∆ mutant was incubated in buffered RPMI medium in ambient or 5% CO2 for 24 h. The Rim101 TF has a broad effect on gene expression with 680 genes upregulated and 859 genes downregulated in the rim101∆ mutant compared to H99 in 5% CO2 (log2 ± 1, FDR < 0.05, Supplementary Data 5; (Wald test p values and Benjamini-Hochberg adjustment for multiple comparisons)). Because the rim101∆ mutant is resistant to RAP, we hypothesized that genes upregulated in the mutant may correspond to genes downregulated by RAP. Indeed, 150 of the 360 genes (41%) downregulated by inhibition of TOR14 are upregulated in the rim101∆ mutant (Fig. 4e). GO term analysis of this set demonstrates that ribosome biogenesis (40 genes; p = 8.3 × 10−37), ribonucleoprotein complex processing (40 genes, p = 1.8 ×10−33), and RNA processing (38 genes; p = 5.8 × 10−22) are strongly enriched and represent 25% of the upregulated genes in rim101∆. These genes are key effectors of the TOR pathway and are upregulated beyond the level of expression induced by CO2 in H99. These transcriptional data support the hypothesis that deletion of Rim101 leads to compensatory activation of the TOR pathway and, consequently, increased CO2 tolerance.
Elevated expression of putative ABC transporter CNAG_07799/PDR9 reduces CO2 fitness
Next, we sought to identify non-regulatory genes that modulate CO2 fitness. To do so, we examined the expression of CO2-responsive genes in CO2-tolerant and -intolerant strains using Nanostring nCounter technology. A focused set of 118 genes covering a diverse set of functions that showed differential regulation in CO2 ( ± 1 log2, FDR < 0.05) based on the RNA seq data at the 24-hour time point above (see Supplementary Data 6 for full set of genes, raw and processed data). The transcriptional profiles of CO2-tolerant and intolerant strains were distinct as shown in Fig. 5a. The sensitive strains show a more consistent profile across different strains than the CO2-tolerant strains. For example, the profile for CO2-tolerant strain C23 is quite distinct from the other two CO2-tolerant strains. In contrast, the CO2-sensitive strains have sets of genes that are consistently highly expressed or lowly expressed. No genes were consistently highly expressed in all tolerant strains relative to intolerant genes. However, a set of 13 genes was expressed higher in the CO2-sensitive stains relative to the tolerant strains (Fig. 5a inset). Five out of these 13 genes have predicted functions related to membrane and lipid homeostasis: ECM2201, CNAG_03227, HAPX, CNAG_07799/PDR9, IPC1, and SRE1. ECM2201, HAPX, and SRE1 are TFs with a role in regulating ergosterol biosynthesis genes. Deletion mutants of the three TFs showed CO2 fitness that was similar to WT (Supplementary Data 2). IPC1 codes for the enzyme that generates inositol-phosphatidyl-ceramide which is the target of the antifungal molecule aureobasidin27; increased CO2, however, did not affect the antifungal activity of aureobasidin (Supplementary Fig. 2).
CNAG_07799 is one of 10 PDR/ABCG family ABC transporters in the C. neoformans genome and has been named PDR9 by Winski et al.28. Phylogenetic analysis places PDR9 in Clade III of fungal PDR/ABC transporters and its closest C. neoformans homologue is AFR1 which is involved in azole resistance28. In addition to mediating drug efflux, PDR/ABC transporters have significant roles in lipid transport and membrane homeostasis29. We were, therefore, interested in its potential effect on CO2 tolerance.
The expression of PDR9 is increased by CO2 exposure in both tolerant and sensitive strains (Fig. 5b). However, the absolute expression of PDR9 in ambient and 5% CO2 conditions is much higher in sensitive strains. For tolerant strains, the expression of PDR9 following induction by 5% CO2 remains well below the baseline levels of intolerant strains. These data suggest that it is the expression level of PDR9 that is important for CO2 tolerance and not the fold change from ambient to 5% CO2. Nanostring profiling of TF mutants with altered CO2 tolerance also showed that PDR9 was downregulated in the resistant mutant rim101∆ but was upregulated in the hypersensitive mutant bwc2∆ (Supplementary Fig. 3a). These data suggested us that elevated expression of PDR9 may reduce CO2 fitness.
To test this hypothesis, we constructed a derivative of the CO2 tolerant strain H99 in which PDR9 was placed under the control of the Histone 3 promoter which has been used by So et al. to overexpress other genes14; importantly, these strains have increased expression of PDR9 by RT-PCR analysis at both ambient and 5% CO2 with the expression in CO2 slightly higher (<1.5 fold, Supplementary Fig. 3b). Consistent with our hypothesis, the PH3–PDR9 strain show reduced fitness in 5% CO2 relative to its H99 parental strain (Fig. 5c). Similar results were obtained on spot dilution assays (Supplementary Fig. 3c). We also introduced the PH3–PDR9 allele into a second C. neoformans background (A1-84-14) and observed an increase in CO2-susceptibilty, indicating that the phenotype is not dependent on strain background (Supplementary Fig. 3d). Thus, absolute expression of PDR9 correlates with reduced growth in host concentrations of CO2. It is likely that the increased expression of PDR9 in H99 and other relatively tolerant strains induced by CO2 contributes to their modest reduction in growth in 5% CO2 relative to ambient CO2 conditions (see Fig. 2a and Supplementary Fig. 3c). However, our data indicate that the absolute expression levels and not the difference in expression of PDR9 between ambient and elevated CO2 correlate with sensitivity to CO2.
We previously reported that CO2 increases C. neoformans susceptibility to the ergosterol inhibitor fluconazole and myriocin, an inhibitor of the first step of sphingolipid biosynthesis10. Since S. cerevisiae homologs of PDR9 affect susceptibility to these drugs, we tested the susceptibility of the PDR9 overexpression strain (Supplementary Fig. 3e). There was no difference in the zone of inhibition for fluconazole, indicating that PDR9 does not act as a fluconazole efflux pump and that it is unlikely that the strains have reduced ergosterol content28,30. In contrast, myriocin susceptibility was increased in the overexpression strain, suggesting that increased expression of PDR9 may alter sphingolipid homeostasis.
C. neoformans remodels phospholipid asymmetry through the TOR pathway in response to CO2 stress
The TOR pathway regulates phospholipid asymmetry at the plasma membrane (PM) by modulating the activity of aminoglycerolipid (AGL) flippases via Ypk1 in S. cerevisiae31. The flippases transport phosphatidylethanolamine (PE) and/or phosphatidylserine (PS) from the outer leaflet of the PM to the inner, cytosolic face of the PM and floppases transport PE/PS in the opposite direction32. Based on the phenotypes of TOR mutants and their function in S. cerevisiae, we hypothesized that PM asymmetry may play an important role in C. neoformans tolerance of host CO2 stress.
To test this hypothesis, we first examined the effect of CO2 and mutants with altered CO2 tolerance on susceptibility to the antifungal duramycin. Duramycin binds PE on the PM outer leaflet as part of its mechanism of action and cells with increased outer leaflet PE exposure show increased susceptibility to the drug31. Although the susceptibility of H99 to duramycin is not affected by 5% CO2, the ypk1∆ mutant is sensitive to duramycin, relative to H99 (Fig. 6a, b). In S. cerevisiae, the TOR pathway inhibits PE flippases though activation of Ypk1 which, in turn, is a negative regulator of Fpk1; accordingly, overexpression of YPK1 increases duramycin susceptibility while the fpk1∆ deletion mutant is resistant to duramycin31. These data suggest that Ypk1 has a distinct role in PE homeostasis in C. neoformans. Supporting this conclusion, the duramycin susceptibility of the fpk1∆ mutant is unchanged relative to H99 (Fig. 6a). Interestingly, deletion of SCH9, a second TOR dependent kinase that is also hypersusceptible to CO2, has increased susceptibility to duramycin. If Ypk1 and Fpk1 have the same lipid asymmetry functions in C. neoformans and S. cerevisiae, then the ypk1∆ mutant would be resistant to duramycin and the fpk1∆ mutant would be hypersusceptible. These results indicate that the Tor-Ypk1 axis has a completely different effect on PE asymmetry in C. neoformans compared to S. cerevisiae. Ypk1 is required for CO2 tolerance while deletion of FPK1 has no effect on this phenotype (Supplementary Data 1), further supporting the idea that the relationship between these two TOR-regulated kinases is different in C. neoformans
One of the proposed roles of PDR/ABC transporters in lipid homeostasis is as floppases29. Because TOR-related mutants with increased PM outer leaflet PE exposure show increased CO2 susceptibility, we tested the duramycin susceptibility of the PH3–PDR9 strain. As shown in Fig. 6b, overexpression of PDR9 increases susceptibility to duramycin at both ambient and 5% CO2. The changes in zones of clearance observed with the PH3–PDR9 strain are similar to those observed in S. cerevisiae mutants involved in PE homeostasis31. These data are consistent with PDR9 having a possible PE floppase activity and that its effect on CO2 tolerance may involve a role in membrane lipid homeostasis. The definitive biochemical characterization of floppases has been technically difficult and somewhat controversial. In S. cerevisiae, Pdr5 has been proposed to be a floppase33. Pdr5 is primarily localized to the PM33. We, therefore, tagged PDR9 with mNeonGreen under both the endogenous and PH3 promoter; the latter strain was hypersusceptible to CO2, confirming the mNeonGreen-tagged allele is functional (Supplementary Fig. 4a). In both strains and under multiple growth conditions, Pdr9-mNeonGreen localizes to intracellular puncta and not to the plasma membrane (Supplementary Fig. 4b). Since validated organelle markers in C. neoformans are limited, we have not conclusively localized Pdr9-mNeonGreen, but the pattern of signal is characteristic of late- Golgi/endosomes and not the PM. Thus, Pdr9 affects PE homeostasis in a manner consistent with a floppase. However, Pdr9 does not localize to the PM, which is where most proposed floppases are found, raising the possibility that this effect on PE distribution may be indirect. As such, additional biochemical and cell biological studies will be required before a definitive lipid asymmetry-related function for Pdr9 can be made with confidence.
Flippases and floppases also affect the distribution of PS between the inner and outer leaflets of the PM32. To test the effect of CO2 on the distribution of PS, we used the cell impermeant molecule, annexin V, to stain outer leaflet PS following recently procedures reported by Huang et al. for the use of this assay with C. neoformans34. Consistent with the previous report34, outer leaflet PS is low but detectable under ambient air conditions (Fig. 6c, d). Exposure to 5% CO2 in RPMI at 37 °C significantly increases outer leaflet PS staining in H99. Overexpression of PDR9 reduced the level of outer leaflet PS induced by 5% CO2 as did deletion of YPK1. Together, these data clearly indicate that adaptation to 5% CO2 is associated with increased outer membrane PS relative to PE and that disruption of this balance by increased expression of PDR9 or by deletion of YPK1 leads to reduced fitness in 5% CO2.
To further test the hypothesis that elevated CO2 concentrations increase PM outer leaflet PS concentrations by altering phospholipid asymmetry, we performed PS uptake assays using fluorescently labeled NBD-PS following a previously reported approach with some modifications34. Mid-log phase H99 cells grown in ambient conditions or 5% CO2 were exposed to NBD-PS for 30 min and cellular uptake of NBD-PS quantitated by flow cytometry; intracellular uptake of NBD-PS was confirmed by microscopy (Supplementary Fig. 5a). As shown in Fig. 6e, cells exposed to 5% CO2 show ~50% reduction NBD-PS uptake relative to cells grown in ambient air conditions. Cdc50 regulates the activity of flippases and, consequently, PS asymmetry in S. cerevisiae. Multiple labs have shown that Cdc50 is also required for PS flip activity in C. neoformans34,35. We, therefore, tested the effect of a cdc50∆ mutant on CO2 susceptibility. Consistent with previous data showing that the cdc50∆ mutant has increased annexin V staining, it is resistant to 5% CO2 relative to WT (Fig. 6f).
An alternative possibility is that total cellular PS is increased in the presence of 5% CO2 leading to the apparent increase in outer leaflet PS localization. To test that possibility, we modulated the total cellular synthesis of PS using a strain containing a copper-repressible allele of CHO1, an enzyme required for the synthesis of PS36. If CO2 tolerance is dependent on an increase in total PS, then preventing that increase by incubating the PCTR4–CHO1 strain on media containing high copper concentrations should reduce CO2 tolerance. As expected, the PCTR4–CHO1 strain has a growth defect on copper-replete media compared to copper-deficient media at ambient air (Supplementary Fig. 6); however, there is no further reduction in the growth of the strain when it is incubated at 5% CO2. The results of this experiment are more consistent with CO2 tolerance being dependent on lipid asymmetry remodeling than with a mechanism involving a global increase in PS synthesis.
Finally, increased expression of PDR9 leads to increased duramycin sensitivity which indicates an increase in outer membrane PE (Fig. 6b). To test if this phenotype is associated with alterations in PE transport, we compared the uptake of fluorescently labeled PE as well; intracellular uptake of NBD-PE was also confirmed by microscopy (Supplementary Fig. 5b). As shown in Fig. 6g, overexpression of PDR9 leads to a 2.5-fold reduction in uptake of NBD-PE. These data further support, but do not conclusively establish, the possible function of Pdr9 as a phospholipid floppase. Taken together, these data strongly support a model in which host-relevant concentrations of CO2 lead to remodeling of PS lipid asymmetry and that disruption of this remodeling reduces tolerance of those conditions.
Overexpression of PDR9 reduces C. neoformans virulence
Environmental strains with elevated expression of PDR9 are not only intolerant of host-relevant concentrations of CO2 but are also less virulent in mouse models of cryptococcosis3,10. We, therefore, hypothesized that the reduced CO2 tolerance of the PH3–PDR9 strain would translate to reduced infectivity and virulence. To test this hypothesis, we used two infection models. First, we carried out a competitive fitness experiment in which a 1:1 mixture of H99 and PH3–PDR9 was used to infect the respiratory tract of mice. The mice were euthanized at DPI 14 and a ratio of the two strains in the lung tissue was determined by quantitative plating on YPD and YPD + NAT (selective for the overexpression strain). Consistent with in vitro studies, the PDR9 overexpression strain was 10-fold less fit in the lung (Fig. 7a). We also compared the virulence of H99 and the PH3–PDR9 strain using single-strain infection experiments. As shown in the survival curve in Fig. 7b, infection with H99 led to 70% moribundity by DPI 40 while no animals infected with the PH3–PDR9 strain showed signs of distress to day 86. The fungal burden at day of sacrifice was determined for five H99-infected animals and compared to the fungal burden of PH3–PDR9 strain-infected animals at DPI 86. Consistent with expectations, H99-infected animals showed high fungal burden in both the lung (Fig. 7c) and brain tissue (Fig. 7d). Interestingly, 4 out of five animals infected with the PH3–PDR9 strain had ~3 log10 organisms in the lung but none in the brain. This observation suggests that overexpression of PDR9 reduces lung burden and reduces dissemination to the brain.
To determine if the reduced virulence of the PH3–PDR9 strain was due to alterations in the three canonical C. neoformans virulence traits2, we compared its growth at 37 °C, melaninization, and capsule formation to H99 under both ambient and 5% CO2 conditions. The PH3–PDR9 strain grew similar to H99 at 37 °C (Supplementary Fig. 7a) and showed no changes in melaninization (Supplementary Fig. 7b). Unexpectedly, the capsule formed by the PH3–PDR9 strain was 1.5-fold thicker relative to H99 in 5% CO2 (Fig. 7e, f). In mouse models, reduced capsule formation is generally associated with reduced virulence. One possible explanation for this discordance is that the capsule formed by PH3–PDR9 is quantitatively larger but may have qualitative or structural defects that affect its function. One of the best characterized functions of a capsule is to interfere with phagocytosis by macrophages37. Using the murine macrophage cell line J774, we compared the phagocytosis of the PH3–PDR9 strain to H99. Consistent with its increased capsule formation, phagocytosis of the PH3–PDR9 strain was reduced relative to H99 (Fig. 7g). This suggests that, to a first approximation, the capsule formed by this mutant functions as expected. Thus, we assert that the best explanation for the reduced fitness of the PH3–PDR9 strain in vivo is its reduced ability to tolerate host levels of CO2.