Balancing Positive and Negative Selection: In Vivo Evolution of Candida lusitaniae MRR1

Abstract

The evolution of pathogens in response to selective pressures present during chronic infections can influence their persistence and virulence and the outcomes of antimicrobial therapy. Because subpopulations within an infection can be spatially separated and the host environment can fluctuate, an appreciation of the pathways under selection may be most easily revealed through the analysis of numerous isolates from single infections. Here, we continued our analysis of a set of clonally derived Clavispora (Candida) lusitaniae isolates from a single chronic lung infection with a striking enrichment in the number of alleles of MRR1 Genetic and genomic analyses found evidence for repeated acquisition of gain-of-function mutations that conferred constitutive Mrr1 activity. In the same population, there were multiple alleles with both gain-of-function mutations and secondary suppressor mutations that either attenuated or abolished the constitutive activity, suggesting the presence of counteracting selective pressures. Our studies demonstrated trade-offs between high Mrr1 activity, which confers resistance to the antifungal fluconazole, host factors, and bacterial products through its regulation of MDR1, and resistance to hydrogen peroxide, a reactive oxygen species produced in the neutrophilic environment associated with this infection. This inverse correlation between high Mrr1 activity and hydrogen peroxide resistance was observed in multiple Candida species and in serially collected populations from this individual over 3 years. These data lead us to propose that dynamic or variable selective pressures can be reflected in population genomics and that these dynamics can complicate the drug resistance profile of the population.IMPORTANCE Understanding microbial evolution within patients is critical for managing chronic infections and understanding host-pathogen interactions. Here, our analysis of multiple MRR1 alleles in isolates from a single Clavispora (Candida) lusitaniae infection revealed the selection for both high and low Mrr1 activity. Our studies reveal trade-offs between high Mrr1 activity, which confers resistance to the commonly used antifungal fluconazole, host antimicrobial peptides, and bacterial products, and resistance to hydrogen peroxide. This work suggests that spatial or temporal differences within chronic infections can support a large amount of dynamic and parallel evolution and that Mrr1 activity is under both positive and negative selective pressure to balance different traits that are important for microbial survival.

Keywords: Candida albicans; Candida auris; Candida lusitaniae; Mrr1; chronic infection; cystic fibrosis; drug resistance; evolution; fluconazole; hydrogen peroxide; yeast.

FIG 1

Constitutively active and low-activity Mrr1 variants naturally evolved in a single C. lusitaniae population. (A) Maximum likelihood-based phylogeny constructed from SNPs identified in the whole-genome sequences of 20 C. lusitaniae isolates; modified from Demers et al. (10). Select branchpoints are marked with the Mrr1 variants (text colored to match branches) present in subsequent isolates. Mrr1 variants are identified by amino acid changes that resulted from SNPs or indels; * indicates a stop codon. The one-nucleotide indel in codons P1174 (insertion) and K912 (deletion) cause frameshift mutations that resulted in early termination, denoted with “(t),” at N1176 and L927, respectively. Gray star at the root of the tree denotes the “ancestral” MRR1 sequence, which lacks any of the mutations listed. U04 and U05, which are used in panels B and C, are highlighted. FLZ MICs (μg/ml) as determined in reference are listed. (B) FLZ MICs for unaltered, mrr1Δ and MRR1 complemented strains in the FLZ-resistant U04 (native allele MRR1^Y813C) strain background. (C) Same as in panel B, but in the FLZ-sensitive U05 strain background (native allele MRR1^L1191H+Q1197*). Strains containing the same MRR1 allele in panels B and C are represented by circles of the same color. Data shown represent at least four independent assays on different days. Each sample was statistically compared to every other sample; the same lowercase letters indicate samples that are not significantly different, and different lowercase letters indicate significant differences (P < 0.0001 [B] or P < 0.001 [C]) as determined by one-way ANOVA with Tukey’s multiple-comparison test of log₂-transformed values. (D) Differentially expressed genes between strains harboring the constitutively active Mrr1-Y813C variant (U04 and U04 mrr1Δ+MRR1^Y813C) and those lacking MRR1 (U04 mrr1Δ) or harboring low-activity variants (U04 mrr1Δ+MRR1^ancestral and U04 mrr1Δ+MRR1^L1191H+Q1197*) when grown in liquid YPD medium; statistical cutoffs used were FDR of <0.05 and fold change of ≥2 (see Table S1 in the supplemental material). Normalized counts per million (CPM) from RNA-Seq are scaled by row (gene) with hierarchical clustering by Euclidean distance. Complemented strains are denoted by their respective MRR1 allele. Predicted C. albicans homologs are listed next to C. lusitaniae gene names (Table S1).

FIG 2

Premature stop codons in Mrr1 differentially impact MDR1 induction by benomyl. (A) Mean FLZ MICs for each of the 20 clinical C. lusitaniae isolates in Fig. 1A separated by the number of nonsynonymous mutations within MRR1 (as defined in reference 10); mean of each group is shown. Two-tailed unpaired t test of log₂-transformed MIC values; ***, P < 0.001. (B) Schematic of C. lusitaniae MRR1 annotated with putative regulatory domains determined by sequence analysis and homology to C. albicans (22) and locations of truncating (top) and putative activating (bottom) mutations. Putative domains include a DNA binding domain with a zinc cluster motif (Zn₂Cys₆; amino acids 33 to 61), a transcriptional regulatory middle homology region (MHR; amino acids ∼607 to 1023), an inhibitory domain (ID; amino acids 1123 to 1217), and an activating domain (AD; amino acids 1218 to 1265). L927 and N1176 are the locations of stop codons caused by indels in codons K912 and P1174, respectively. (C and D) MDR1 expression normalized to ACT1 in the absence (solid bars) or presence (striped bars) of 50 μg/ml benomyl. Means ± standard deviations (SDs) of representative data in biological triplicates are shown; similar trends observed on at least three different days. Two-way ANOVA with Sidak’s multiple-comparison test; **, P < 0.01; ****, P < 0.0001; ns, not significant. In panel D, strain names are highlighted corresponding to the number of mutations in MRR1, yellow for one and green for two, as in Fig. 2B. The colors of the circles (A), lines (B), and bars (D) correspond to MRR1 alleles shown Fig. 1A.

FIG 3

Premature stop codons repeatedly arose in constitutively active Mrr1 variants resulting in reduced Mrr1 activity but, in some cases, restored Mrr1 inducibility. Schematic of inferred evolution of MRR1 alleles in the L1191H+Q1197* (A) and Y1126N (B) lineages. (C) FLZ MICs for U04, U04 mrr1Δ and MRR1 complemented strains in the U04 mrr1Δ background. Means ± SDs from three independent assays on different days shown. Each sample was statistically compared to every other sample; the same lowercase letters indicate samples that are not significantly different, and different lowercase letters indicate significant differences (P < 0.01) as determined by one-way ANOVA with Tukey’s multiple-comparison test of log₂-transformed values. (D) MDR1 expression normalized to ACT1 from culture grown in YPD (bars, left y axis). Means ± SDs from three independent assays on different days; data from each day were normalized to the expression of U04 mrr1Δ+MRR1^ancestral. Each sample was statistically compared to every other sample; the same lowercase letters indicate samples that are not significantly different, and different lowercase letters indicate significant differences (b to d, P < 0.05; all other pairwise comparisons, P < 0.01) as determined by one-way ANOVA with Tukey’s multiple-comparison testing of log₂-transformed data. Overlaid with log₂-transformed mean ± SD fold change (FC) in normalized MDR1 expression following exposure to 50 μg/ml benomyl (squares, right y axis); full data presented with statistics in Fig. S3D. (C and D) FLZ MICs and MDR1/ACT1 expression data are colored to match; the sample names are shown on the x axis of panel D. Comparison of mean basal MDR1 (E), MGD1 (F), or FLU1 (G) expression from panel D and Fig. S3B or C, respectively, excluding strains lacking functional MRR1, and mean ± SD log₂-transformed FC of the induction following benomyl exposure from Fig. S3D to F; circles colored to match those in panel C. Goodness of fit r² value for nonlinear regression shown.

FIG 4

Constitutive Mrr1 activity decreases resistance to H₂O₂ in multiple Candida species. (A) Growth curve of U04 mrr1Δ+MRR1^Y813C (teal) and U04 mrr1Δ+MRR1^Y813C mdr1Δ (black) grown at 37°C in YNB medium supplemented with the indicated carbon source: glucose (circles), amino acids (triangles), or glycerol (squares). Means from representative data acquired in triplicates shown. (B) Quantification of cytokines IL-8 and IL-1β in BAL fluid from the CF patient with (red) or seven patients without (black) C. lusitaniae in their lungs. Two-way ANOVA with Sidak’s multiple-comparison testing found no significant differences. (C) Serial dilution assays of C. lusitaniae, C. albicans, and C. dubliniensis strains on YPD or YPD supplemented with the indicated concentration of diamide or H₂O₂. Strain names in bold font were shown to contain GOF mutations in Mrr1 resulting in increased FLZ resistance (Fig. 3C and references , , and 46). Plates were imaged after 24 or 48 h of growth at 37°C, as indicated. (D) Percent growth in well-aerated 5 ml YPD plus 1 mM H₂O₂ was calculated relative to that of the vehicle only control after 22 to 24 h growth at 37°C. These data represent six independent assays performed on different days. Significance determined by paired t test; **, P < 0.01.

FIG 5

Trade-off between FLZ and H₂O₂ resistance persists in evolving C. lusitaniae populations during a chronic lung infection. (A) Schematic of sampling timeline (top) and histogram of the number of isolates that (i) were mostly uninhibited on FLZ, but were inhibited by H₂O₂ (red), (ii) were mostly uninhibited on H₂O₂ but were inhibited by FLZ (blue), or (iii) were uninhibited under both conditions (black). For the schematic, the gray bar represents the 6 to 10 months before the BAL during which this patient was identified as being colonized by non-albicans Candida (NAC) species. C. lusitaniae was determined to be the dominate microbe in the upper and lower lobe (UL and LL, respectively) BAL samples (red arrow), which marks the start of the green bar. Sp1 was obtained 1 month before the BAL and was retrospectively also found to contain abundant C. lusitaniae. Sp1.5, Sp2, Sp4, Sp5, and Sp6 were obtained 3, 9, 32, 35, and 38 months, respectively, after the BAL and all contained C. lusitaniae. A 4-month course of FLZ therapy was given after the BAL. Scale bar indicates 6 months. Multiple isolates were collected from each sample/timepoint (n = 38 to 80) and assayed for growth on YPD supplemented with 8 μg/ml FLZ or 4 mM H₂O₂. Growth was scored as completely inhibited, partially inhibited or uninhibited compared to that of a YPD-only control. (B) Model for the evolution of C. lusitaniae MRR1 in this population. Whole-genome sequencing and mutation analyses suggest that following the initial infection with C. lusitaniae harboring the Mrr1-ancestral variant, a combination of exposure to different stimuli that changed overtime or by locations within the CF lung environment led to the selection for a heterogeneous population. Multiple constitutively active Mrr1 variants arose, and while some persisted over time, others were subsequently mutated again. The secondary mutations causing premature stop codons (represented by shortened bars) resulted in reversion to low Mrr1 activity that was inducible or complete loss of Mrr1 activity. The balance between selective pressures resulted in a heterogeneous population of isolates with varied resistance (^R) to biologically and clinically important compounds.