Mrs4 loss of function in fungi during adaptation to the cystic fibrosis lung

Abstract

The genetic disease cystic fibrosis (CF) frequently leads to chronic lung infections by bacteria and fungi. We identified three individuals with CF with persistent lung infections dominated by Clavispora (Candida) lusitaniae. Whole genome sequencing analysis of multiple isolates from each infection found evidence for selection for mutants in the gene MRS4 in all three distinct lung-associated populations. In each population, we found one or two unfixed, non-synonymous mutations in MRS4 relative to the reference allele found in multiple environmental and clinical isolates including the type strain. Genetic and phenotypic analyses found that all evolved alleles led to loss of function of Mrs4, a mitochondrial iron transporter. RNA Seq analyses found that Mrs4 variants with decreased activity led to increased expression of genes involved in iron acquisition mechanisms in both low iron and replete iron conditions. Furthermore, surface iron reductase activity and intracellular iron was much higher in strains with Mrs4 loss of function variants. Parallel studies found that a subpopulation of a CF-associated Exophiala dermatiditis infection also had a non-synonymous loss of function mutation in MRS4. Together, these data suggest that MRS4 mutations may be beneficial during chronic CF lung infections in diverse fungi perhaps for the purposes of adaptation to an iron restricted environment with chronic infections.

Introduction

Evolution of pathogens in infections can lead to the rise of isolates with increased resistance to host defenses or drugs, improved fitness, or enhanced access to nutrients. An understanding of pathoadaptive mutations may improve therapies and treatments. The repeated rise of specific mutations in bacteria and fungi associated with chronic infections has been particularly well-documented in the context of lung infections in people with cystic fibrosis (CF). The genetic mutations that cause CF lead to chronic infections. Several studies have shown that nutritional immunity, mediated by innate immune effectors such as calprotectin which restrict access to metals, force microbes to employ diverse strategies to acquire essential nutrients such as iron and zinc (1–3).

Many of the mutations that repeatedly arise in CF-related bacterial and fungal pathogens are in regulators (e.g. lasR and mucA in Pseudomonas aeruginosa, agr in Staphylococcus aureus). Analysis of Candida CF isolates has found similar regulatory mutations. A study of Candida albicans CF infection identified six instances of loss-of-function (LOF) mutations in NRG1, which encodes a repressor of filamentation; nrg1 LOF mutants are resistant to the suppression of filamentation by the frequently co-infecting bacterium Pseudomonas aeruginosa (4). We previously published that a single Clavispora (Candida) lusitaniae infection, with no detectable co-infecting bacteria, had numerous activating and subsequent suppressing mutations in MRR1 (5) that lead to heterogenous resistance to the antifungal fluconazole, the toxic metabolite methylglyoxal, and P. aeruginosa toxins. Longitudinal collections of Aspergillus fumigatus isolates from a single individual with CF showed acquired mutations that lead to HOG pathway hyperactivation and improved fitness in the presence of oxidative and osmotic stress (6).

In this work, we describe a locus in C. lusitaniae (7) that was independently mutated in three separate subjects with CF. C. lusitaniae, a haploid member of the CTG clade within the Saccharomycetaceae family, is known to readily develop Amphotericin B resistance (8) and can develop resistance to caspofungin and azoles (9, 10). C. lusitaniae has been reported in association with plant and food products, and is a less commonly found to be an abundant member of microbiome communities. Notably, C. lusitaniae is closely related to Candida auris which is also known for the repeated development of multi-drug resistance, and is a critical threat on the WHO priority pathogens list (11, 12).

C. lusitaniae evolution in the CF lung may provide an opportunity to study fungal adaptation to host environments. We found non-synonymous mutations in MRS4 arose independently in three different CF lung infections. Mrs4 is a high affinity inner mitochondrial membrane iron transporter that brings iron into the inner lumen. In other fungi, Mrs4-mediated iron transport is necessary for robust growth in low iron environments, resistance to Cons and menadione, and its function supports the synthesis of iron-sulfur clusters for incorporation into diverse enzymes and regulators (13–18). Previous studies found that MRS4 deletion significantly reduces virulence of C. albicans in a murine model for systemic candidiasis (19). We found that each of the Mrs4 variants had decreased function. RNA seq analysis of isolates with different MRS4 alleles in iron replete and iron restricted conditions demonstrated that Mrs4 LOF led to significant increases in expression of multiple iron uptake methods. Furthermore, strains with MRS4 LOF alleles demonstrated increased accumulation of intracellular iron. A LOF mutation in MRS4 was also found in CF infection isolates of Exophiala dermatitidis. Taken together, these data highlight that in two distinct environmental fungi, chronic infection leads to the selection for Mrs4 variants with decreased function and an increased capacity for iron uptake. Future studies will determine if these host-adapted strains have new vulnerabilities that can be exploited therapeutically.

Results

MRS4 mutations are found in C. lusitaniae from CF lung infections

Analysis of the microbiota in bronchoalveolar lavage (BAL) fluids collected at Dartmouth Health found three subjects CF with lung infections dominated by C. lusitaniae ((7) and in a manuscript in preparation). We sequenced the genomes of 12–20 isolates from each population (see Methods section for Accession numbers). To identify mutations that likely arose during infection, non-synonymous single nucleotide polymorphisms (SNPs) that were not fixed within the population from each individual were determined (Table S1). Nineteen genes had two alleles with non-synonymous differences in more than one population, defined as isolates from a single subject at a single time point, and only one gene had alleles with non-synonymous differences within in all three populations: CLUG_02526 (Figure 1A). CLUG_02526 encodes an amino acid sequence with 71% identity with C. albicans SC5314 Mrs4 (19), and 51% with Saccharomyces cerevisiae S288C Mrs3 and Mrs4 (Figure S1). Mrs4 functions as high affinity mitochondrial iron importers in both species (16, 20). Due to the high percent sequence identity and our phenotypic characterization of CLUG_02526 deletion mutants (described below), we will heretofore refer to CLUG_02526 as MRS4.

Figure 1. Non-synonymous SNPs in MRS4 were found in whole genome sequence data from 12–20 C. lusitaniae isolates from each of three subjects with chronic CF infections.

A) Analysis of loci that were heterogeneous in three C. lusitaniae populations in three separate individuals (Subjects A, B, and C) found that only CLUG_02526 (MRS4) had subpopulations with non-synonymous substitutions in all three infections. B) Two MRS4 alleles were detected in each population. “REF” indicates the MRS4 sequence in environmental and acute infection isolates of C. lusitaniae. C) The MRS4 sequence encodes a barrel-structure iron transporter on the inner mitochondrial membrane; the protein is 318 amino acids long and comprised of six transmembrane alpha-helices denoted by light blue bars. Each mutation is predicted to disrupt or truncate one of these transmembrane domains (see SUSPECT analysis, Fig S3). D) Pooled sequencing was performed on isolates from bronchoalveolar lavage fluid taken from specific lobes of subjects A and B. The relative abundances of MRS4 alleles was quantified by analysis of individual reads.

To compare the MRS4 sequences from CF C. lusitaniae isolates to MRS4 sequences to a broader collection of C. lusitaniae strains, we also analyzed MRS4 in ten C. lusitaniae isolates from diverse clinical and environmental sources. We found that the predicted Mrs4 amino acid sequences were identical and the genes differed by only a small number of synonymous SNPS that varied among alleles (Figure S2). The conserved Mrs4 amino acid sequence will be referred to as the “reference” or Mrs4^REF sequence. Isolates with MRS4 alleles that encoded the Mrs4^REF sequence were found in both Subject A and Subject C (Figure 1B). In addition to the reference allele, Subject A and Subject C populations each had isolates with mutant alleles that differed by single non-synonymous SNPs, and encoded MRS4^A235V and MRS4^G138V, respectively (Figure 1B). Subject B isolates carried one of two mutant alleles that encoded MRS4^A147D and MRS4^Q254* (Figure 1B), suggesting that two independent MRS4 mutant lineages arose within that population. The Mrs4 substitutions or terminations occurred in predicted transmembrane alpha helices of the protein (Figure 1C) and occurred at residues that were conserved across diverse species (Figure S1). Each MRS4 mutation found in the CF C. lusitaniae isolates had a high likelihood of affecting function based on the SuSPECT analysis method which estimates the probability for single amino acid variants to impact phenotype (Figure S3) (21).

Spatial and longitudinal analyses show that MRS4 alleles may have arisen independently in different regions of the lung and that Mrs4 LOF isolates persisted over time and across compartment.

Using whole genome sequence data for pools of 75–96 isolates from the upper and lower lobes of the right lung of Subject A and the upper, middle and lower lobes of the left lung of Subject B, we determined the fraction of reads that encoded the MRS4 SNPs described above within each pool. For the pooled isolates of the upper lobe of Subject A, the reads contained only the MRS4^REF allele, while ~43% of sequenced population from the lower lobe had the MRS4^A235V allele (Figure 1D). Both alleles were detectable in the upper, middle, and lower lobe isolate pools of Subject B, with the MRS4^A147D allele present at a higher percentage in the upper and lower lobe (~83 and 89%, respectively), while MRS4^Q254* was found in ~93% of reads in the middle lobe. Sequence analysis of four sputum isolates from a sample donated by subject A, collected ~1 year after the BAL isolates were recovered, found two isolates with MRS4^REF and two isolates with MRS4^A235V suggesting that the MRS4 variant-containing isolates persisted over time. Similarly, we obtained 6 respiratory sputum and stool isolates from Subject B more than one year after the initial isolation and amplified and sequenced the MRS4 allele. We found that they all contained the MRS4^A147D allele. These longitudinal isolates indicate that the MRS4 mutations persisted over time and possibly in multiple compartments, and thus were not transiently present at the time of isolation.

MRS4 variants confer LOF phenotypes

In other fungi, the deletion of MRS4 impairs growth in low iron media or the presence of an iron chelator (17, 20, 22). To determine the activity of Mrs4 variants in the C. lusitaniae CF isolates, we first constructed an mrs4Δ derivative of Subject B isolate B_L01, which had an MRS4^Q254* allele, then complemented back either the native B_L01 MRS4^Q254* allele or MRS4^REF at the native locus (Figure 2A). We chose the MRS4 allele from strain ATCC 42720 as the source of the MRS4^REF sequence. We observed that all four isogenic strains (B_L01 parental isolate, the mrs4Δ mutant, and the mrs4Δ mutant complemented with MRS4^REF or MRS4^Q254*) reached similar yields in YPD (Figure 2B). Yields were reduced in YPD amended with a high-affinity ferrous iron chelator bathophenanthroline disulfonate (BPS), and we observed differences dependent on MRS4 allele (Figure 2B). Deletion of mrs4 in the B_L01 background reduced final yield. Complementation with the MRS4^REF allele restored growth to a significantly greater degree than the parent strain, or to the B_L01 mrs4Δ complemented with the native allele. These results indicate that the truncated variant has decreased function relative to the reference allele, but that the truncated variant may retain some function.

Figure 2.

Mrs4^Q254* confers loss of function in C. lusitaniae. A) An mrs4Δ mutant and mrs4Δ mutants complemented with MRS4^Q254* or the MRS4^REF were constructed in the B_L01 clinical isolate background B) Strains were assessed for growth in a 96-well plate after 24 h at 37° in YPD or YPD with 80 μM BPS iron chelator. Columns labelled with a are non-significantly different from each other, and are significantly different from columns labelled with b and c. C) Indicated strains were grown for 24 h at 37° in YPD supplemented with 2.5 mM CoCl₂ (left) and 12.5μM CdCl₂ (right). There were at least three replicates per sample. Indicated p-values are from a one-way ANOVA with Tukey’s post-hoc correction, ns, not significant.

Deletion mutants lacking MRS4 in S. cerevisiae and C. albicans have altered metal sensitivities such that mutants are more resistant to cobalt and more sensitive to cadmium when compared to their Mrs4+ counterparts (17, 20) due to activation of transcription factors such as Aft1 in S. cerevisiae (15). We found that cadmium and cobalt affected the growth of C. lusitaniae in an MRS4 allele dependent manner. The B_L01 mrs4Δ derivative and the mrs4Δ complemented with the allele encoding Mrs4^Q254* were more resistant to cobalt than the mrs4Δ mutant complemented with MRS4^REF (Figure 2C). Conversely, the mrs4Δ strain complemented with the Mrs4^REF variant was more resistant to cadmium than the mrs4Δ mutant or the mutant complemented with Mrs4^Q254*. Together, these data further suggest that the Mrs4^Q254* variant is less functional than the reference allele.

To characterize the levels of function for the other three Mrs4 variants found in the CF clinical C. lusitaniae isolates, we created mrs4Δ derivatives of isolates A_U05 (MRS4^A235V), B_L04 (MRS4^A147D), and C_M06 (MRS4^G138V) which were then complemented with the MRS4^REF allele. In each case, complementation with the functional MRS4^REF allele made cells more sensitive to cobalt (Figure 3A) as was the case for strain B_L01. Similarly, replacement of MRS4 with the MRS4^REF allele increased cadmium sensitivity in B_L04 (MRS4^A147D), and C_M06 (MRS4^G138V) (Figure 3B). Unexpectedly, the A_U05 (MRS4^A235V) isolate had greater resistance to cadmium than the same background with the MRS4^REF allele, perhaps due to other genetic differences. When we expressed MRS4^A235V in B_L01 mrs4Δ, the resultant strain was more sensitive to cadmium than the isogenic strain with MRS4^REF supporting the conclusion that the Mrs^A235V variant had low or no activity (Figure S4). We also made an mrs4Δ mutation in an outgroup C. lusitaniae strain RSY284 (DH2383) (23) with a native MRS4^REF allele, and confirmed that the mutant had the expected resistance to cobalt and sensitivity to cadmium (Figure S5A). Together, these data suggest that LOF mutations in MRS4 arose four independent times across three chronic CF infections indicating possible selection for phenotypes associated with loss of Mrs4 function.

Figure 3. MRS4 mutations in each clinical population demonstrate LOF phenotypes.

Representative parent isolates of each mutation from Subject B (B_L04, MRS4^A147D), Subject A (A_U05, MRS4^A235V), and Subject C, (C_M06, MRS4^G138V) and their mrs4Δ derivatives that were then complemented with the MRS4^REF allele, were grown in YPD supplemented with A) 2.5 mM CoCl₃ and B) 12.5 μM CdCl₂. Data represent the endpoint OD600 measured by a Synergy Neo2 plate reader after 24 h of growth at 37°. Indicated p-values are from one-way ANOVA with Tukey’s post-hoc , ns, not significant.

Characterization of MRS4 growth phenotypes.

Mitochondrial metabolism is highly dependent on the availability of metals such as iron. To assess whether the MRS4 loss-of-function alleles affected mitochondrial activity, we examined the growth of isogenic strains with either MRS4^REF or Mrs4^Q254* in medium with either glucose (which can be fermented) or glycerol as the major carbon source. In both yeast extract peptone medium or yeast nitrogen base media with 2% glucose, there was a slightly faster growth for the strain with Mrs4^REF compared to the mrs4 null mutant or strains with Mrs4^Q254* but the differences were not significant (Figure 4A and C). Similar results were obtained when grown in medium with glycerol as the dominant or sole carbon source (Figure 4B and D). These results suggested that there was no major defect in respiratory metabolism.

Figure 4. Lack of MRS4 growth phenotypes in complex and minimal media with glycerol and glucose.

Cells from exponential phase cultures of B_L01, its mrs4Δ derivative, and mrs4Δ mutants complemented with MRS4^Q254* or the MRS4^REF were inoculated into a 96-well plate and grown for 24 h at 37° in YP medium supplemented with A) 2% glucose or B) 2% glycerol or in YNB defined medium without amino acids supplemented with C) 2% glucose or (D) 2% glycerol. OD₆₀₀ was measured over time using a Synergy Neo2 plate reader.

As a more sensitive indicator of changes in levels of respiration and fermentation, we quantified glucose consumption relative to fermentation product production by HPLC. Analysis of supernatants from cultures of the mrs4Δ + MRS4^REF strain grown in medium with glucose as the sole carbon source found that the dominant fermentation product was acetate followed by ethanol; glycerol was not detected. Under the same conditions, C. albicans strain SC5314 produced mainly ethanol with low levels of acetate and glycerol. Normalized to moles of carbon, C. lusitaniae converted more than twice as much glucose to fermentation products than C. albicans (Table S2). When the mass balance of glucose consumed and fermentation products made for C. lusitaniae mrs4Δ + MRS4^Q254* and mrs4Δ + MRS4^REF were compared, we found no significant differences suggesting comparable levels of respiration and fermentation (Table S2). We also constructed a C. albicans mrs4Δ/Δ homozygous mutant and found no significant difference in fraction of carbon used for fermentation when the SC5314 wild type was compared to the mrs4Δ/Δ homozygous mutant (Table S2).

To further assess metabolic differences associated with Mrs4 LOF, we analyzed growth of B_L01 parent and mrs4Δ::MRS4^REF on different sole carbon sources using the Biolog^™ carbon utilization phenotype microarray plates. We inoculated equal concentrations of each strain in YNB minimal medium and observed growth over the course of 48 h. Across the 192 carbon sources tested, there were no differences in growth that were confirmed in secondary analysis (Supplementary dataset 1). Analysis of B_L01 parent and B_L01::MRS4^REF did not show any differences in minimal inhibitory concentrations for commonly used antifungals including fluconazole and amphotericin B (data not shown). Consistent with published studies in Cryptococcus neoformans (24), in C. lusitaniae DH2383 and B_L01, functional Mrs4 was necessary for full H₂O₂ resistance (Figure S5).

MRS4 LOF impacts expression of iron homeostasis regulators and metal storage

To gain insight into how Mrs4 LOF affected C. lusitaniae, we performed a transcriptomics analysis of B_L01 mrsΔ with MRS4^REF and MRS4^Q254*. Because cells with defective Mrs4 (e.g. MRS4^Q254*) showed reduced growth on medium with chelator (Figure 2B), we performed an RNA-seq analysis of cells from mid-exponential phase cultures growing in YPD and parallel cultures that received a 1 h exposure to the iron chelator BPS (Figure 5A for experimental scheme). The short exposure to BPS was used to limit the pleiotropic effects associated with growth differences. Comparison of the strains with MRS4^REF without or with exposure to iron chelator found eleven genes that had a fold difference greater than log₂ 1 and a false discovery corrected p-value less than 0.05 (Figure 5B and Supplementary dataset S2). Among the genes that were differentially expressed were four cell surface high affinity ferric iron uptake genes (FTR1, FRE9, FRE10, and FET31) and genes that encode regulators involved in iron utilization SFU1 and HAP43 (Figure 5B) which are known to be transcriptionally regulated in response to iron limitation response in other Candida species (25, 26).

Figure 5. Loss of Mrs4 function leads to increased expression of iron acquisition genes.

A) Design of RNA-seq sample preparation. Sextuplicate cultures of B_L01 mrs4Δ complemented with either REF or Q254* MRS4 alleles were grown overnight, then sub-cultured into YPD and grown for 5 h. Cultures were grown for an additional hour with either 80μM BPS or vehicle prior to RNA isolation. Gene expression heatmaps of differentially expressed genes (P < 0.05 and a log₂ fold-change ≥|1|) in a comparison between B) the B_L01::MRS4^REF strain grown in YPD (iron replete) or YPD with BPS (iron deplete) C) B_L01::MRS4^REF and B_L01::MRS4^Q254* grown in YPD with BPS, and D) B_L01::MRS4^REF and B_L01::MRS4^Q254* grown in YPD.

Comparison of the transcriptomes of B_L01 mrsΔ+MRS4^REF to B_L01 mrs4Δ+MRS4^Q254* after chelator exposure found that the same ferric reductases that were induced upon exposure to chelator in the Mrs4^REF strain were higher in cells with Mrs4^Q254* (Figure 5C). We again found differential expression of the SFU1 and HAP43 regulators, and also found greater fold induction of the gene encoding pro-iron acquisition transcription factor SEF1 in the mrs4Δ+MRS4^Q254* relative to mrsΔ+MRS4^REF with chelator. In addition, we found differential expression of putative orthologs of iron uptake genes (SIT1, CSA1) and other metal transporters (CCC1, CCC2) (27) when Mrs4 function was low. MRS4 transcript levels were log₂ 1.7 fold higher in the mrs4Δ+MRS4Q254* strain, and the MRS4-adjacent gene GOR1, predicted to encode a glyoxylate reductase, was also significantly higher in the Mrs4^Q254* bearing strain.

We also compared transcriptomes of mrsΔ+MRS4^REF and mrs4Δ+MRS4^Q254* in control conditions without chelator. Thirty-seven genes were differentially expressed across the two strains in the presence of chelator (Figure 5C), and 27 of them were still differentially expressed in its absence (Figure 5D). Ferric reductase encoding genes and their regulators (e.g. HAP43) were again differentially expressed, and the fold difference between strains was higher than in the chelator condition (Figure 5C and D). An independent experiment, using qRT-PCR analysis of RNA from mrsΔ+MRS4^REF and mrs4Δ+MRS4^Q254* also found transcript levels of HAP43 and FRT1 to be significantly higher in both iron replete and iron chelated conditions (Figure S6). We also observed that mrsΔ+MRS4^REF had ~4.5 fold higher levels of INO1, the ortholog of inositol-3-phosphate synthase, than mrs4Δ+MRS4^Q254* in YPD (Figure 5D).

Loss of Mrs4 function leads to higher surface ferric reductase activity

We sought to determine if the higher levels of in transcripts encoding cell surface ferric reductases in MRS4^Q254* strains, even in iron-replete conditions, translated into higher levels of iron acquisition activity. To do so, we utilized tetrazolium chloride (TTC), a substrate for ferric reductases (28). To avoid complications associated with growth inhibition by TTC, we grew colonies on YNB-glycerol agar for 24 h, then overlaid with molten 1% agar containing TTC (Figure 6A). Upon TTC reduction by ferric reductases, an insoluble red pigment forms. After ten minutes, there was a strong red coloration associated with colonies formed by MRS4^Q254* and the mrs4Δ strains, but not the MRS4^REF strain (Figure 6A). The TTC reduction phenotype was abrogated by the addition of excess ferric iron in the agar overlay (Figure 6A). These data suggest greater ferric reductase activity on the cell surface when Mrs4 activity is low. Similar results though to a lesser degree were observed on YPD medium (Figure 6B).

Figure 6. Decreased Mrs4 function increases ferric reductase activity and intracellular iron content.

A) B_L01 derived strains mrs4Δ, mrs4Δ::MRS4^REF, mrs4Δ::MRS4^Q254* were spotted on YNB-glycerol plates. Plates were incubated for 24 h at 37°C. Each plate was overlayed with a 10 ml solution of 1 mg/ml tetrazolium chloride (TTC) and incubated for 5 min prior to imaging. Red pigmentation indicated represents greater levels of ferric iron reduction. Inclusion of 10 mM of FeCl₃ (+Fe) as a competitor eliminates TTC reduction. 24 B) B_L01 derived strains mrs4Δ, mrs4Δ::MRS4^REF, mrs4Δ::MRS4^Q254* and hap43Δ were serially diluted from 1 OD and spotted on YPD plates, then allowed to grow for 24 hours at 37° C then analyzed as in panel A. C) C. lusitaniae DH2383 and its mrs4Δ mutant and C. albicans SC5314 with single and double knockouts of mrs4 were analyzed for surface ferric reductase activity on YNB-glycerol. D) B_L01 mrs4Δ strains complemented with MRS4^REF and MRS4^Q254* alleles were grown in YPD or YPD + 80 μM BPS as outlined in Fig. 5A. Whole cell iron was quantified using ICP-MS. Data represents the averages of three technical replicates for two experiments done on separate days. Indicated p-values are from Student’s t-tests.

Previously studies showed that decreased mitochondrial iron levels induced the activity of C. lusitaniae transcription factors by modulating cytosolic Fe-S-containing regulators. We predicted that Hap43 (within the Hap43-Sfu1-Sef1 transcription factor network) was part of this response that led to increased expression of iron uptake genes. Thus, we analyzed a hap43Δ mutant in the B_L01 background with the defective Mrs4^Q254* variant, and found that Hap43 indeed contributed to the increased levels of iron reductase activity created by Mrs4 LOF (Figure 6B). We also found that surface ferric reductase activity was higher in the mrs4Δ strains if C. lusitaniae strain 2383 and C. albicans SC5314 mrs4Δ/Δ. In C. albicans, both copies of MRS4 to be deleted for manifestation of the increased surface ferric reductase phenotype.

Mrs4 LOF leads to the accumulation of intracellular iron

To evaluate the consequences of differential expression of iron acquisition genes in strains with and without Mrs4 function, we analyzed the concentrations of cellular iron by ICP-MS. In iron replete YPD medium, the strain mrs4Δ+ MRS4^Q254* had significantly higher levels of intracellular iron than mrs4Δ::MRS4^REF(Figure 6D). In cells subjected to chelator treatment for 1 h prior to harvest, intracellular iron was lower than in the untreated cells, but still significantly higher in the mrs4Δ::MRS4^Q254* strain. These data suggest that the observed increase in iron acquisition gene transcripts was concomitant with higher intracellular iron which may be advantageous in vivo and may explain the repeated LOF of Mrs4 in clinical C. lusitaniae populations.

MRS4 mutations in other fungi of clinical interest

In light of variation in MRS4 in three different C. lusitaniae populations, we investigated the consequences of two other naturally-occurring MRS4 allelic variations that we observed in other species. First, we assessed the activity of Mrs4 variants in Candida auris, a fungus that emerged within the past forty years (29) and is closely related to C. lusitaniae. Multidrug resistant strains that caused localized outbreaks emerged independently within genetically distinct clades. Analysis of Candida auris MRS4 alleles within and between clades, available from published sequences (30), found that the encoded Mrs4 sequences were identical, with one exception. Clade I strains differed from strains in the other clades in that the MRS4 allele encoded Mrs4^31V while others encoded Mrs4^31A. This difference was confirmed by Sanger sequencing. To assess the activity of these two Mrs4 variants, we expressed them in a C. lusitaniae mrs4Δ mutant. Both C. auris Mrs4^31A and Mrs4^31V alleles were able to complement mrs4Δ for growth with iron chelator and cadmium sensitivity to a similar extent as the functional C. lusitaniae Mrs4^REF protein (Figure S7), indicating that both alleles were functional.

The second analysis of Mrs4 variants was in the black yeast Exophiala dermatiditis. We had previously identified a CF infection predominated by E. dermatiditis. We performed whole genome sequencing of twenty-three isolates and found a subpopulation of isolates with a non-synonymous SNP in the MRS4 (Figure 7A) (31). Subsequent Sanger sequencing confirmed that seven of the twenty three isolates carried an MRS4 ortholog with a sequence identical to the E. dermatiditis type strain, NIH8656 (referred to here as encoding Mrs4^40E-REF). The other sixteen sequenced isolates had a variant MRS4 with an E40G substitution. To characterize the MRS4 alleles in these E. dermatitidis isolates, both were synthesized and heterologously expressed in C. lusitaniae mrs4Δ and assessed for function relative to the C. lusitaniae MRS4 alleles. The E. dermatiditis alleles (EdMRS4^40E-REF and EdMRS4^40G) were codon optimized for Candida spp. and the spliced introns were removed. The EdMRS4 alleles were introduced at the native MRS4 site and expressed under the control of the C. lusitaniae MRS4 promoter.

Figure 7. An Mrs4 loss-of-function subpopulation also emerged in Exophiala dermatiditis during a chronic CF lung infection.

A) Two alleles of MRS4 were found in E. dermatitidis isolates from in a single chronic CF lung infection. Of the 23 isolates sequenced, seven genomes encoded the reference Mrs4^40E (E.d. REF), which is identical to previously sequenced E. dermitiditis strains, and sixteen isolates encoded an Mrs4^40Gvariant (E40G). B) C. lusitaniae B_L01 mrs4Δ strains complemented with the two E. dermatitidis MRS4 alleles grown in YPD with 80 μM BPS for 24 h. C. lusitaniae B_L01 mrs4Δ expressing functional MRS4^REF or MRS4^Q254* were included for comparison. Indicated p-value are from a one-way ANOVA with Tukey’s post-hoc , ns, not significant.

Complementation of the mrs4Δ strain with Mrs4^40E-REF fully complemented the mrs4Δ mutant to the levels of C. lusitaniae MRS4^REF. In the presence of the BPS iron chelator, the mrs4Δ strain with the E. dermatiditis Mrs4^40G variant had significantly reduced growth compared to a strain with E. dermatiditis Mrs4^40E, which fully restored the Mrs4 function to levels observed for the C. lusitaniae reference gene (Figure 7B). This indicates that a Mrs4 LOF mutation also arose in a population of E. dermatitidis during a CF lung infection. The repeated occurrence of MRS4 mutations in fungal CF infections strongly suggests a selective benefit for MRS4 LOF mutations in the CF lung.

Discussion

In this work, we showed the repeated loss of Mrs4 activity across two fungal species, C. lusitaniae and E. dermatiditis, in the context of chronic CF lung infections. Each acquired non-synonymous mutations that resulted in reduced or loss of Mrs4 function, which led to defects in iron import into the mitochondrial inner lumen and increased expression and activity of iron acquiring pathways. The selection for MRS4 LOF mutations in chronic lung infection may inform future studies on the mechanism of fungal persistence within the host and resistance to therapeutic strategies. The emergence of Mrs4 LOF in two diverged species of Ascomycota, both environmental fungi which colonized chronic CF lung infections, highlights the possibility that mutations in MRS4 may be important for the shift to commensal colonizing yeast. Work by Kim et al. (4) found C. albicans NRG1 LOF mutations in isolates from different individuals with CF suggesting that inactivating mutations in this locus increased fitness. Interestingly, mutants in Nrg1 have increased expression of iron uptake genes (32).

In other species, mutation of Mrs4 leads impairs the mitochondrial synthesis of Fe-S clusters, and Fe-S cluster levels modulate the iron starvation response through their insertion into specific transcription regulators in ways that modulate activity of the iron response network that includes Hap43, Sef1, Sfu1 (14, 15, 33). Thus, Mrs4 mutation broadly promotes the induction of iron acquisition pathways even in iron replete conditions as we observed (Figure 4D). Thus, MRS4 mutations may be a key mechanism for a simultaneous increase in activity of multiple regulators. Because inactivating mutations in genes encoding Sfu1, the transcriptional repressor of iron uptake or Yfh1, the iron-sulfur exporter, would only activate a subset of pathways or have other pleiotropic effects on the cell. Unlike LOF or gain-of-function mutations in iron regulators themselves, the MRS4 mutation did not result in constitutive derepression of the complete iron uptake regulon, as we observed a reduction in expression level of genes involved in the low iron response in iron-replete media and this level of regulation may be beneficial in preventing the accumulation of toxic concentrations of iron and other metals.

In chronic infections, such as those in the CF lung, the host restricts the availability of essential nutrients such as iron via nutritional immunity (34). Host proteins such as lactoferrin, transferrin, and calprotectin sequester iron, making it less accessible to pathogens. Enhanced expression of siderophore acquisition pathways (e.g. SIT1) and surface ferric reductases (e.g. those encoded by CLUG_02348 and CLUG_04344) by C. lusitaniae likely provides a significant advantage in acquiring iron from iron sequestering molecules. The increased accumulation of cellular iron may aid cells during fluctuations in iron availability, such as what might occur in the lung environment that experiences cycles of increased inflammation associated with disease exacerbations and inflammation resolution. Candida species also have mechanisms to acquire iron from heme which represents approximately 80% of iron within the body. In chronic bacterial infections, evidence suggests that heme utilization is active (35, 36). Candida species employ a three-factor system for the acquisition of heme, using the secreted factor Csa2 to bind and ferry heme to the cell surface, where it is brought into the cell by Pga7 and Rbt5. Heme acquisition pathways (CLUG_4093, CLUG_4096, and CLUG_4097) were at significantly higher levels in MRS4^Q254* than in MRS4^REF (Figure 5C and D) in both control and iron chelated conditions. A role for Mrs4 in chronic conditions is an interesting contrast to the importance of Mrs4 function in systemic fungal infections and it is interesting to consider different iron demands and sources in different types of infections (19).

Decreased Mrs4 activity may allow cells to accumulate high levels of iron without jeopardizing mitochondrial function. Iron is highly regulated by all cells due to its reactive properties. MRS4 mutation may not only enhance iron uptake, but also limit iron concentrations in mitochondria which protects mitochondria from damage. Reduced iron uptake by mitochondria may be key for the accumulation of total cellular iron.

One unique feature of the three CF C. lusitaniae infections was that there were no detectable bacterial pathogens as the time of the BAL sample collection. When other Candida species are detected in CF, the fungi are usually part of a mixed bacterial-fungal infection. The reason for these intriguing differences is not known and a current area of study. Here, we showed that when compared to C. albicans SC5314, C. lusitaniae produced significantly less ethanol and more acetate as fermentation products. Ethanol stimulates biofilm formation and virulence factor production in CF bacterial pathogens (37, 38), and thus metabolic differences between fungi may be a differentiating factor. Mutations in MRS4 may also be an important factor in competing with other microbes for iron. Lastly, C. albicans and C. lusitaniae differ in the degree to which they stimulate macrophages and thus differences in immune response may play an important role. Future studies will determine how different species persist in chronic infections, the roles specific mutations that are repeatedly under selection, and whether there are common themes across different pathogens. Chronic infection by microbes that are not also human commensals, such as C. lusitaniae and E. dermatiditis, provide an opportunity to study initial adaptations to the host environment and factors that are most critical for survival and persistence.

Materials and Methods

Clinical isolate collection from respiratory samples.

Clinical isolates were acquired from sputum and bronchoalveolar lavage (BAL) fluid samples that were plated on YPD (1% yeast extract, 2% peptone, 2% glucose, 1.5% agar) containing gentamycin, blood agar or CHROMagar Candida media then restruck on YPD to obtain single isolates which were then saved in 25% glycerol. C. lusitaniae clinical isolates were obtained in accordance with the study protocol approved by Dartmouth Health Institutional Review Board (#22781) using methods described in (5).

DNA isolation, genome sequence analysis and variant calling.

Genomic DNA was extracted from cultures grown in YPD (2% peptone, 1% yeast extract, and 2% glucose) for ~16 hours; extractions were performed using the MasterPure yeast DNA purification kit (Epicentre). Genomic libraries, for single and pooled isolate DNA, were prepared using the KAPA HyperPrep Kit and sequenced using paired-end 150 bp reads on the Illumina NextSeq500 platform, to a depth of 100–150x coverage per sample as described in Demers et al. (5). The pipeline for genome analyses is available in a github repository (https://github.com/stajichlab/PopGenomics_Clusitaniae; doi: 10.5281/zenodo.7800401). The short read sequences were aligned to a modified version (5) of the Candida lusitaniae ATCC 42720 genome (39). The ATCC 42720 genome was altered to remove mitochondrial fragments inserted into the nuclear assembly and the mitochondrial contig (Supercontig_9) was replaced by a complete mitochondrial genome from strain C. lusitaniae CBS 6936 (NC_022161.1). The following regions were masked out due to unusually high coverage and likely mitochondrial origin: (Supercontig_1.2:1869020-90 1869184,1664421-1664580; Supercontig_1.3:1076192-1076578,1324802- 1324956,1353096-1353260; Supercontig_1.6:126390-126604; Supercontig_1.8:29199-92 29370). Alignments were made using bwa (0.7.17-r1188) (40) and stored as a sorted, aligned read CRAM file with Picard (2.14.1, http://broadinstitute.github.io/picard/) to assign read groups and mark duplicate reads (script 01_align.sh). CRAM files were processed to realign reads using GATK’s RealignerTargetCreator v4.1.8.1 and IndelRealigner following best practices of GATK (41). Each realigned CRAM file was processed with GATK’s HaplotypeCaller (script 02_call_gvcf.sh) followed by joint calling of variants on each Chromosome using GATK’s GenotypeGVCF method script 03_jointGVCF_call_slice.sh). This step also removed low quality variant positions: low quality SNPs were filtered based on mapping quality (score <40), quality by depth (<2 reads), Strand Odds Ratio (SQR>4.0), Fisher Strand Bias (>200), and Read Position Rank Sum Test (<−20). These files were combined to produce a single variant call format (VCF) file of the identified variants to produce list of high quality polymorphisms (script 04_combine_vcf.sh). The quality filtered VCF file containing only variants among the clinical isolates was categorized by SnpEff (5.1) (42) and the ATCC 42720 gene annotation. Genome assemblies of the strains was performed with SPAdes (v3.12.0) (42) after trimming and adaptor cleanup of the reads was performed with AdapatorRemoval (v2.0) (43) and quality trimming with sickle (v1.33) (44). De novo assemblies were further screened for vector contamination with vecscreen step of AAFTF v0.3.1 (https://github.com/stajichlab/AAFTF) (doi: 10.5281/zenodo.1620526). The metadata for the strains corresponding to the genome sequences for isolates from these three subjects will be described in a Microbial Resource Announcement submitted prior to publication of this work. The raw sequence reads for whole genome sequencing of Subject A, B, and C isolates have been deposited into NCBI sequence read archive under BioProject # PRJNA948351. Details for the isolates and isolate pools from different regions of the lung for Subject A are described in (5). For the analysis of MRS4 sequences from environmental C. lusitaniae strains (Table S3) and from clinical isolates obtained from subsequently obtained sputum or stool sample was performed by amplifying MRS4 using primers DRM031 and DRM032 and sequenced using these primers along with primer ED157 which binds within the MRS4 sequence (see Table S4 for primer sequences).

Strains and mutant construction.

Fungal strains and plasmids used in this study are listed in Table S3. Fungi were maintained on YPD medium. CRISPR-Cas9 knockout of MRS4 from clinical isolates was performed using previously described methods (45). For complementation of the reference MRS4 allele, 5’ UTR and coding region were amplified from ATCC 42740 and assembled into a complementation plasmid with a marker encoding hygromycin resistance (HYG) and 3’ UTR using yeast recombination cloning (46). The complementation plasmid was digested with NotI and KasI resulting in a ~3500 bp fragment which was transformed into mrs4Δ derivatives of representative isolates along with Cas9 and a crRNA targeting the NAT marker.

Analysis of Candida auris MRS4 alleles.

Clade I (strain B8441), Clade II (B11220), Clade III (B11221), and Clade IV (B11245) were used in the MRS4 sequence comparisons of C. auris isolates. C. auris MRS4 was amplified from one of two isolates from the CDC Antibiotic Resistance Isolate bank: AR bank #0382 of Clade I (Biosample Accession # SAMN18754596), and AR bank #0383 of Clade III (Biosample Accession # SAMN05379609). C. auris MRS4 alleles were amplified using primers with 20 base pairs of overlap with C. lusitaniae MRS4 5’ UTR and the HYG resistance cassette. E. dermatitidis MRS4 alleles were synthesized de novo by Genscript with the omission of introns, and 20 base pairs of overlap with C. lusitaniae MRS4 5’ UTR and the HYG resistance cassette. These sequences were then reintroduced into the native locus by complementation cassette by restriction digest, replacing the reference MRS4 allele with the heterologous sequences. Complementation of heterologous sequences was then performed by the same method as complementation of the reference allele. Primers are listed in Table S4.

Growth Assays.

Unless otherwise stated, strains were grown as 5 ml cultures in YPD overnight (~16 h), exponential growth aliquots were washed three times and subcultured into experimental medium. For spot titer growth comparisons, cultures were diluted to 1 OD in diH₂O, diluted 1:10 serially, and spotted in 5 μl volumes on plates. For growth assays in 96 well plates, a starting concentration of 0.005 OD in YPD was used and stated concentrations of BPS, cobalt chloride, or cadmium chloride were added from stock solutions in water. BPS (Sigma CAS# 52746-49-3), cobalt chloride (Sigma CAS# 7791-13-1), and cadmium chloride (Sigma CAS# 654054-66-7) stocks were 100 mM, 100 mM, and 100 μM respectively. Final yield was measured by OD600 at 24 h post-inoculation. For H₂O₂ sensitivity, fresh aliquots of 9.8 M H₂O₂ were diluted into YPD at the time of inoculation, and growth was measured over the course of 24 h as previously described. Biolog assays were conducted by suspending 0.01 OD of each strain in YNB, and aliquoting 200μl of culture into 192 wells of two proprietary Biolog^™ plates PM1 and PM2A with a diverse array of carbon sources. Growth was measured by OD600 over the course of 48 hours, and the final yield is represented in Supplementary Dataset 1.

TTC analysis of surface iron reductase activity.

After 24 h growth on the indicated medium, a 10 ml solution containing 0.5 mg/ml or 1 mg/ml tetrazolium chloride, 1% molten agar, and 10 mM FeCl₃ chloride (from 100 mM stock made fresh), if indicated, was carefully pipetted using a 10 ml serological pipette to cover the entirety of the plate. Plates were incubated for the specified time (10 min to 1 h) prior to imaging.

Transcriptome analysis of the effects MRS4 mutation.

For RNA isolation, isolates were sub-cultured from overnight cultures into fresh YPD and grown for 6 h which corresponds to cultures in mid-exponential growth phase. Cultures were sub-cultured into six replicate 5 ml cultures of each strain which were incubated at 37C on a rollerdrum. After 5 h of growth, three replicate cultures of each strain were dosed with BPS to a final concentration of 80 μM while the other cultures received water only. Samples were spun down in 15 ml conical tubes, snap-frozen with ethanol and dry ice, and stored for at least 1 h at −80°C. RNA extraction was performed using to MasterPure Yeast RNA Purification kit protocol (Epicentre) according to manufacturer instructions. RNA was submitted to MiGS for RNA Seq analysis. EdgeR was used for normalization and differential gene analysis of raw counts provided by MiGS. RNA-seq data have been submitted to the SRA database: #SUB10993521.

qRT-PCR analysis.

Culture growth and RNA extraction was performed as described above for the RNA-seq analyses. RNA was DNAse treated with the Turbo DNA-free Kit (Invitrogen). cDNA was synthesized from 500 ng DNAse-treated RNA using the RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific), following the manufacturer’s instructions for random hexamer primer (IDT) and GC rich template. qRT-PCR was performed on a CFX96 Real-Time System (Bio-Rad), using SsoFast Evergreen Supermix (Bio-Rad) with the primers listed in Table S4. Thermocycler conditions were as follows: 95 °C for 30 s, 40 cycles of 95 °C for 5 s, 65 °C for 3 s and 95 °C for 5 s. Transcripts were normalized to ACT1 expression.

Intracellular iron quantification.

Cells from overnight cultures were subcultured into YPD and grown for 5 h before the addition of 80 μM of BPS iron chelator. Samples were taken before adding chelator, and 1 h after iron restriction. Samples were spun down in pre-weighed Eppendorf tubes, washed, and pellets were dried in a vacuum centrifuge for 3 h. Final dry weight was calculated for each pellet, and each was digested with 100μl of 70% HNO₃. After overnight digestion, samples were heated to 90°C to ensure complete digestion. After dilution with 3.9 ml of diH₂O, samples were submitted to the Dartmouth Trace Metal Core for ICP-MS analysis of iron content.

HPLC Supernatant Analysis.

Strains grown in overnight cultures were washed three times in dH₂O and subcultured in triplicate into YNB minimal media supplemented with 100 mM glucose at a final OD of 0.01. Cultures were allowed to grow for 6 hours at 37° C, then centrifuged at 13,200 RPM for 5 minutes for the separation of insoluble solids and collection of supernatant. Four blank media samples of YNB were also prepared with known concentrations of added carbon sources. One blank was supplemented with 100 mM glucose, 5 mM sodium acetate, 100 mM ethanol, 5 mM sodium citrate, 500 μM sodium lactate, 5 mM sodium succinate, 200 μM sodium pyruvate, and 100 mM glycerol, with two other blanks containing 1:10 and 1:100 dilutions of these carbon sources for the creation of a standard curve within a linear range. The final blank was prepared without any additional carbon sources. For each sample, 400 μl of supernatant was centrifuged at 10,000 RPM for 2.5 minutes through Corning nonsterile nylon 0.22 X-spin filters (#8169), then 20 μl of 10% sulfuric acid as added. Samples were transferred to 2 ml polypropylene snap top microvials for HPLC analysis. Samples were analyzed for levels of various sugars and organic acids utilizing a Shimadzu HPLC (LC-2030) with Biorad Aminex HPX-87H column, LC-20AD pump system, SPD-20AV detector, SIL20AC autosampler, and CTO-20AC column oven.

Statistical Analysis.

All data were analyzed using Graph Pad Prism 8. The data represent the mean standard deviation of at least three independent experiments with three technical replicates unless stated otherwise. Comparisons were made using a two-tailed, unpaired Student’s T-Test or ANOVA as indicated. One-way ANOVA tests were performed across multiple samples with Tukey’s multiple comparison test for unpaired analyses.