Host-Imposed Copper Poisoning Impacts Fungal Micronutrient Acquisition during Systemic Candida albicans Infections

Nutritional immunity is a process whereby an infected host manipulates essential micronutrients to defend against an invading pathogen. We reveal a dynamic aspect of nutritional immunity during infection that involves copper assimilation. Using a combination of laser ablation inductively coupled mass spectrometry (LA-ICP MS) and metal mapping, immunohistochemistry, and gene expression profiling from infected tissues, we show that readjustments in hepatic, splenic and renal copper homeostasis accompany disseminated Candida albicans infections in the mouse model. Localized host-imposed copper poisoning manifests itself as a transient increase in copper early in the kidney infection. Changes in renal copper are detected by the fungus, as revealed by gene expression profiling and fungal virulence studies. The fungus responds by differentially regulating the Crp1 copper efflux pump (higher expression during early infection and down-regulation late in infection) and the Ctr1 copper importer (lower expression during early infection, and subsequent up-regulation late in infection) to maintain copper homeostasis during disease progression. Both Crp1 and Ctr1 are required for full fungal virulence. Importantly, copper homeostasis influences other virulence traits—metabolic flexibility and oxidative stress resistance. Our study highlights the importance of copper homeostasis for host defence and fungal virulence during systemic disease.


Introduction
Micronutrients such as ferrous or cuprous ions are scarce, yet essential for life. They play central roles in many biological processes, executing both structural and catalytic functions [1]. However, these ions are toxic in excess. For example, they promote free radical generation via immunity. We conclude that the maintenance of cellular copper homeostasis during periods of copper excess and subsequent starvation is essential for C. albicans virulence.

Candida albicans strains and growth conditions
Experiments were performed with C. albicans clinical isolate SC5314 [23] or mutants constructed in this background (below). Unless specified otherwise, cells were grown at 30°C in YPD medium [24] or YNB (0.67% yeast nitrogen base without amino acids) medium, supplemented with 2% appropriate carbon sources, or with stressor at the specified concentration.
C. albicans mutants were constructed using the Clox system with nourseothricin selection [25]. To create the homozygous ctr1/ctr1 deletion mutant, the NAT1-Clox cassette was PCR amplified using primer pairs CTR1_CLOx_F_JP plus CTR1_CLOx_R_JP, and CTR1_CLOx_ F3_JP plus CTR1_CLOx_R3_JP (S1 Table) and used to transform SC5314 to sequentially delete both CTR1 alleles. To construct the ctr1/CTR1 reintegrant strain, the wild type CTR1 sequence was reintroduced into its native chromosomal locus. The CTR1 locus was PCR amplified using primers SpeI_F_CTR1_ReINT_JP and Not1_R_CTR1_ReINT_JP (S1 Table), digested with SpeI and NotI, and cloned into pLNMCL [25]. The plasmid was then digested with BstAPI and used to transform the ctr1/ctr1 deletion mutant selecting for nourseothricin resistance.
To create the homozygous crp1/crp1 null mutant, the NAT1-Clox cassette was PCR amplified using primer pairs CRP1_CLOx_F_JP plus CRP1_CLOx_R_JP, and CRP1_CLOX_F2_JP plus CRP1_CLOX_R2_JP (S1 Table), with both CRP1 alleles deleted sequentially in C. albicans SC5314. To construct the crp1/CRP1 reintegrant strain, the wild type CRP1 locus was PCR amplified using primers Not1_F_CRP1_ReINT_JP plus Not1_R_CRP1_ReINT_JP (S1 Table), digested with NotI and cloned into pLNMCL. This plasmid was then digested with Kpn2I, and used to transform the crp1/crp1 null mutant.
The genotype of each mutant and the ploidy of the target locus were verified by qRT-PCR with Roche LightCycler 480 system with specific primer pairs (S2 Table).
(MOI = 1) with M-1 mouse kidney cortical collecting duct epithelial cells (CRL-2038, ATCC) for up to 24 h in DMEM:Ham's F12 supplemented with 2 mM glutamine (Gibco Life Technologies), 5 μM dexamethasone (Hspira UK Limited), 5% foetal bovine serum with 5% CO 2 at 37°C. KC production was measured by ELISA [29]. Renal epithelial cell damage was assessed by assaying lactate dehydrogenase release into the culture supernatant [29]. For histology, the cells were fixed in 4% formaldehyde for 12 h at 8°C, washed three times with PBS and stained using Periodic acid and Schiff [28]. Mann-Whitney U test with two-tailed t-test was used to assess statistical differences between the two groups.
For fluorescent detection of ATP7B, the secondary antibody used was goat anti-rabbit Alexa Fluor 647 (IgG H&L) conjugate (Abcam, Cambridge, UK). The staining was performed in the dark, after which the sections were mounted using VectaShield Mounting Medium with DAPI (VectorLabs, Orton Southgate, UK).

Gene expression studies
Mouse and fungal transcripts from tissues fixed with RNAlater (Qiagen, Crawley, UK) or in vitro C. albicans cultures were quantified as described elsewhere [30]. Roche LightCycler 480 and Universal Probes were used in monocolour hydrolyses reactions in qRT-PCR, according to the manufacturer's instructions. Primer pairs and specific probes are listed in S3 Table and elsewhere [12,30]. Fungal transcript levels were normalised to the ACT1 mRNA control, and mammalian transcript levels to the GAPDH mRNA. Differences were considered statistically significant for p0.05 in two-tailed t-test, using the Mann-Whitney U test. For multiclass comparisons in Figs 1A Laser ablation inductively coupled plasma mass spectrometry (LA-ICP MS) and ICP-MS for total tissue metal content LA-ICP MS experiments were performed as described previously [12]. Sequential 22-μm-thick cryosections from the same tissue were prepared for LA-ICP MS, histology and immunohistochemistry. Elemental distribution mapping was performed with a laser ablation system (UP-213, New Wave) coupled to an Agilent 7500c ICP-MS, as described previously [12]. Data were background subtracted and plotted using Microsoft Excel v14.2.4. The images are representative of at least three biological replicates. Total tissue copper measurements were conducted using standard methods [31]. The data are averages from kidneys of four different animals at the indicated infection stages, pooled and measured in samples of two.

Ethics
All animal experiments were approved by the University of Aberdeen Welfare and Ethical Review Board. The experiments were conducted in compliance with United Kingdom Home Office licenses for research on animals (project license number PPL 60/4135) and in accordance with EU Directive 2010/63/EU. Animals were euthanized by cervical dislocation. All experiments are reported in accordance with the ARRIVE guidelines [32]. As the infection progresses, the abundance of hepatic transcripts for the metal storage protein metallothionein MT1 and the copper-containing ferroxidase ceruloplasmin (CP) increase, while the abundance of the copper importer Ctr1 transcript decreases (A). In contrast, hepatic ceruloplasmin protein levels do not increase with progressive C. albicans infection, while CTR1 protein levels diminish (B), as assessed by immunohistochemistry (Materials and Methods). All transcript data were acquired in duplicate from four biological replicates: healthy mice (replicates H1-H4), animals at early (24 h, replicates E1-E4) or late (96 h, replicates L1-L4) infection stages, following injection with saline (controls, H) or C. albicans SC5314. Transcript abundances were normalised against the GAPDH mRNA. Fold differences in expression are given when p0.05. Numerical data are provided in S4 Table. 63 Cu measurements using LA-ICP MS. 13 C levels remain relatively unchanged throughout, whereas there is a transient increase in tissue 63 Cu early in the infection (far right). Histology of the corresponding tissue sections is shown ('pas_h', middle right). The sequential transverse kidney sections shown in (A-C) are representative of at least two technical replicates, with at least three biological replicates. (D) The shifts in copper distribution are reinforced by changes in the expression of genes encoding copper-associated functions. These transcript abundance data were acquired in duplicate from at least three biological replicates. Transcript abundances were normalised to the GAPDH mRNA. Fold differences in expression are given when p0.05. Numerical data are given in S5 Table. The fungus senses and responds to the copper redistribution in the colonised kidney, as evidenced by the significant up-regulation of the CRP1 copper exporter gene early in the infection, and the CTR1 copper importer gene at the later infection stage. These transcript data were acquired in duplicate from at least seven biological replicates: early (24 h, replicates E1-E8); late (96 h, replicates L1-L7) infection stage. Fungal transcript abundances were normalised to the ACT1 mRNA. Fold differences of expression are given when p0.05. Numerical data are given in S6 Table. (B) The C. albicans gene expression pattern in fungal cells isolated from the colonised kidneys is distinct from the fungal response to the blood, supporting tissue-specific response of the CTR1 and CRP1 transcripts in (A). Blood was obtained from three healthy animals. Transcript abundances for blood and in vitro experiments were acquired in duplicate from three biological replicates. Colour scale is as in (A). Numerical data are given in S7 Table. (C) C. albicans transcriptional response to the blood from healthy animals (HB) and animals 24 h post infection (24h) differs from that of in vitro (NGY) grown cells, and is largely unchanged for the subset of genes analysed. Colour scale is as in (A). Numerical data are given in S8 Table. (D) The inactivation of either CRP1 or CTR1 compromises C. albicans virulence in the mouse model, as assessed by determining fungal loads in the kidney 72 h post infection. Mann-Whitney U test with two-tailed ttest was used to assess statistical differences between the groups (n = 6 mice per group; the detection limit of our CFU determining method is ca. 100 fugal cells/kidney; CFU counts were below the detection limit in four animals infected with crp1 null mutant and in five animals infected with ctr1 null mutant). Differences were considered statistically significant when p0.05, and the corresponding p values are given.

C. albicans infections affect hepatic copper metabolism
The mammalian liver is involved in iron and copper homeostasis. It is a major site of iron storage, with ferric iron trapped in ferritin [33], and regulates iron fluxes in other organs through the production and excretion of the soluble hormone hepcidin [34,35]. The liver also produces ceruloplasmin, the major plasma cuproprotein, and ferroxidase that promotes iron mobilisation [13,36]. The liver receives the bulk of dietary copper from the duodenum, conjugated to albumin or α2-macroglobulin. Ceruloplasmin also promotes the delivery of copper to other organs [37,38].
During disseminated C. albicans infections, hepcidin production and hepatic iron stores increase dramatically [12]. Given the documented relationship between copper and iron homeostases [13], we profiled key hepatic copper-related transcripts to gain insight into the effect of C. albicans infection upon hepatic copper metabolism. These included transcripts encoding the mammalian copper storage protein metallothionein MT1, the high-affinity copper importer CTR1, and ceruloplasmin. We sampled the livers of healthy animals and those of animals infected systemically for 24 or 96 h (henceforth referred to as 'early' or 'late' infection stages) (Fig 1A, S4 Table). The Mt1 transcript was significantly up-regulated while the Ctr1 transcript was significantly down-regulated at both infection time points. This suggested increased copper storage and decreased copper consumption by the liver [39,40]. Indeed, recently Li et al [41] reported perturbations in hepatic copper content during the course of candidiasis infections in mice. Ceruloplasmin production is induced during infections as part of the generic acute phase response of the host [42]. Accordingly, the Cp transcript was significantly up-regulated in the livers of infected, but not healthy, animals ( Fig 1A). Interestingly, this was not accompanied by an increase in hepatic ceruloplasmin protein levels, as determined by immunohistochemistry (Fig 1B). This might suggest that Cp gene expression is regulated at a posttranscriptional level [43]. CTR1 protein levels decreased as infection progressed, which correlated with the decline in Ctr1 transcript levels (Fig 1A and 1B). We conclude that systemic C. albicans infection perturbs host copper homeostasis in the liver.

C. albicans infections perturb splenic copper homeostasis
We showed previously that systemic candidiasis affects metal (iron) metabolism in the spleen [12]. Therefore, we investigated the effects of C. albicans infection on copper appropriating proteins of the splenic red pulp. We used specific antibodies to probe splenic levels of ceruloplasmin, CTR1 and ATP7B (a trans-Golgi copper efflux ATPase which supplies copper to various cuproproteins [13,44]). While ceruloplasmin was barely detectable and remained unchanged (not shown), both CTR1 and ATP7B decreased over the course of systemic infection ( Fig 1C). These data are consistent with a lowered splenic copper consumption. Indeed, splenic copper levels were recently shown to decrease during candidiasis in the mouse model [41], further supporting the view that copper appropriation by the spleen declines during disseminated C. albicans infections.
In certain cell types, depending on copper availability, ATP7B and the closely related copper transporter ATP7A relocalise to facilitate copper export rather than trans-Golgi delivery, migrating to the apical or basolateral membranes, respectively [13,45,46]. We observed no change in the subcellular localisation of ATP7B, which was detected as defined punctate structures at each infection stage (S1 Fig). However, we did observe a gradual decrease in ATP7B levels in the splenic red pulp, suggesting quantitative rather than qualitative effects of infection on ATP7B.
Collectively, our data indicate readjustments of hepatic and splenic copper homeostatic functions during the development of systemic C. albicans infections.
Adjustment of renal copper homeostasis coincides with increased renal iron acquisition during C. albicans infections During disseminated candidiasis, the kidney takes over some erythrocyte recycling functions from the spleen, and during the latter stages of infection this is reflected in increased renal transferrin receptor (iron acquisition), HO-1 and HO-2 (haem iron extraction), and hepcidin and ferritin (iron retention) [12]. Renal iron medullary deposits increase as a result. Simultaneously, the host imposes iron nutritional immunity in the renal cortex through infiltrating immune cells [12]. Given the well-documented mechanistic link between the iron and copper homeostases [13], we reasoned that kidney copper metabolism would also be affected by systemic candidiasis. Therefore, we compared renal ceruloplasmin, CTR1, ATP7A and ATP7B protein levels in animals at different stages of infection (Fig 2A). Indeed, the levels of these copper homeostasis proteins increased during infection. Furthermore, the abundance of ATP7B progressively increased in both the renal cortex and medulla during infection, but its subcellular localisation remained unchanged over the course of the disease (Fig 2B). Although ATP7B is generally thought to be liver-specific, this result is consistent with a previous report on ATP7B dynamics in an in vitro model [46]. Collectively, these data suggest increased renal copper acquisition and/or redistribution during systemic fungal infection.
We next mapped 63 Cu in kidneys from animals at different stages of infection and observed a transient increase in the overall copper in early infection (Fig 2C). We measured the total amount of kidney copper from organs of healthy animals, and animals at early and late infection stages. There was a ca. 10% increase in total kidney copper at the 24 h time point, when compared with healthy animals (16.6±0.5 mg/kg dry weight 'early infection' vs. 15.3±0.8 mg/ kg dry weight 'healthy'). Late in the infection, the total kidney copper levels fell by more than ca. 10% compared to the healthy controls (13.9±0.4 mg/kg dry weight 'late infection'). The recent data of Li et al [41] reinforce these observations. The kidney is more resilient to global metal content changes than other organs [47], and therefore the detected changes in copper may suggest a physiologically relevant role for the infection process.
Our data indicated changes in renal copper loading and distribution during disease progression, suggesting possible alterations in intracellular renal copper pools. We therefore profiled the expression of mouse genes encoding copper metallochaperones and some of their clients: Ccs and Sod1; Cox17 and Cox1; Atox1 for Atp7B; ceruloplasmin (Cp), the main cuproprotein involved in iron loading of transferrin; and the mitochondrial Mn Sod2 (Fig 2D, S5 Table). The abundances of the Cox1, Cox17 and Sod1 transcripts all decreased in abundance during disease progression. Interestingly, SOD1 protein was present at lesion sites late in the infection (S2 Fig). Ccs and Sod2 transcript levels remained unchanged. Remarkably, Cp transcript was up-regulated about seven-fold in infected kidney, relative to the healthy controls. This suggests that the ceruloplasmin protein detected late in the infection (Fig 2A) may have been synthesised by the kidney [48], rather than being transported from the liver. This corroborated our observation that hepatic ceruloplasmin levels do not increase during the course of infection (Fig 1B).
Since the kidney is relatively resilient to changes in metal content compared to other organs [47], we surmised that changes in Cox1 and Cox17 mRNA levels during infection could either signify kidney injury or infection-triggered shifts in renal copper dynamics. Mitochondrial dysfunction and fragmentation occur during kidney injury [49,50]. However, Cox1 transcript levels decreased early, i.e., 24 h post infection, when only modest numbers of C. albicans cells were detectable in the kidneys. Furthermore, the mRNA levels for genes encoding mitochondrial fission and fusion proteins remained unchanged during the infection (Fig 2D). Thus, the changes in renal copper homeostasis gene expression during systemic C. albicans infection appear to reflect an active response to infection rather than tissue damage.

Renal copper fluxes trigger micronutrient adaptation in C. albicans during systemic infection
The detected changes in renal copper content during disseminated C. albicans infections ( Fig  2C) temporally coincided with the mobilisation of renal iron toward the medulla and away from fungal lesions during the infection [12]. We previously showed that the fungus senses the reduction in iron availability and responds by activating haem iron acquisition alongside its reductive iron acquisition pathway [12]. Therefore, we reasoned that the dynamic cupric/ cuprous renal microenvironment should also be detected by the fungus. We examined the expression of copper-associated genes in fungal cells from renal lesions, focusing on copper transporters, chaperones, regulators and storage proteins (Fig 3A, S6 Table). Transcript levels for the high affinity copper efflux ATPase Crp1 [51] were significantly higher at early infection stages, and then declined. This was consistent with a recent report indicating that fungal CRP1 expression increases early during systemic infection [52]. In contrast, expression of the high affinity copper importer CTR1 [53] was lower at early infection stage and then increased during the course of infection (Fig 3). These expression patterns differed from those of C. albicans cells exposed to blood (Fig 3B, S7 Table) or those grown in vitro (Fig 3C, S8 Table). Rather, they were consistent with fungal copper efflux coinciding with the transient increase in renal copper levels early in infection. Furthermore, the data suggested an increase in fungal copper import in response to the subsequent decline in renal copper levels (Fig 2C).
We examined the regulation of C. albicans CRP1 (copper efflux pump) and CTR1 (copper importer) in response to copper in vitro. CRP1 gene expression increased while CTR1 expression decreased in response to elevated copper concentrations, in agreement with earlier reports [51,53,54] (S3 Fig). In addition, CTR1 was induced under iron limitation, consistent with the interaction of iron and copper micronutrients and in agreement with previous studies [55] (S4 Fig, S9 Table). Regulation of these C. albicans genes in response to copper and iron in vitro reflected their expression patterns in vivo in response to the observed changes in renal metal ion availability (Fig 2C and Potrykus et al [12]). We next asked whether the C. albicans cells were responding to copper content in the bloodstream or in the renal tissue. The bloodstream contains copper-containing entities such as ceruloplasmin, macroglobulin, albumin and amino acids [13], and therefore it was possible that the C. albicans cells were reacting to copper levels in the blood. Indeed, a progressive increase in serum copper has been reported over the course of systemic candidiasis in the mouse model [41]. Hence we examined the impact of blood from infected mice and healthy controls upon the expression of copper-(and also iron-) related genes in C. albicans. C. albicans cells were incubated for 30 min with blood from healthy animals and animals 24 h postinfection (Fig 3C) as this time point coincided with high CRP1 expression levels in vivo ( Fig  3A) and fungal cells in vitro respond to copper within this timescale (S4 Fig). Significant differences were observed between the blood-incubated samples and the corresponding in vitro controls for most of the genes analysed. However, the only gene displaying a significant difference between C. albicans cells exposed to infected or uninfected blood was CSA2, which encodes a haem receptor (Fig 3C). Given the minimal differences between infected and uninfected blood, we inferred that the observed changes in renal copper distribution during systemic candidiasis, and the accompanying C. albicans gene expression patterns, reflect changes in copper retention by the kidney rather than bloodstream copper accessible to the fungus.

The virulence of C. albicans during systemic infection depends on both CRP1 and CTR1
The CRP1 and CTR1 genes encoding the copper efflux pump and importer, respectively, are differentially regulated during systemic infection (Fig 3A). To test whether these genes are important for infection, we constructed isogenic C. albicans deletion mutants in the genetic background of the clinical isolate SC5314 [33] using our Clox gene disruption system [36] and tested their virulence in the mouse model of systemic infection [12]. We inoculated mice with similar numbers of fungal cells, based on haemocytometer measurements of total cell numbers. Subsequent measurements of the actual CFUs revealed differences in the number of live fungal cells injected into the mice possibly due to different strain viabilities in vitro. This might have affected the final kidney CFU outcomes. Nevertheless, the data strongly suggest that CRP1 and CTR1 were both required for the full virulence of C. albicans (Fig 3D). Thus, during infection, C. albicans cells require the capacity to extrude, as well as import and assimilate, copper.
The main phenotype of the CRP1 null mutant in vitro was its exquisite sensitivity to copper. In rich YPD medium, crp1 cells were unable to grow at Cu 2+ concentrations over 7.5 mM, whereas wild type cells tolerated concentrations of 12.5 mM (Fig 4A). This recapitulated the phenotype of C. albicans crp1 mutants constructed in other genetic backgrounds [51,54].
We tested whether the virulence defect of crp1 cells resulted from its role in the C. albicansrenal cell interaction. To achieve this we exploited our ex vivo model of renal infection, which is a proxy for C. albicans virulence studies [29]. The interaction of C. albicans with the renal epithelial cells in this ex vivo model was not compromised by CRP1 inactivation (Fig 4B and  4C). No significant differences between C. albicans crp1 and wild type cells were observed with respect to their ability to stimulate lactate dehydrogenase (LDH) release or KC cytokine production by the renal cells (Fig 4B and 4C). Furthermore, after an overnight incubation, the crp1 mutant formed a matt of fungal cells over the renal cells, comparable to the wild type control (not shown). Therefore, the transient CRP1 up-regulation observed in vivo during developing systemic infection (Fig 3A) is not simply an outcome of fungus-renal cell interactions.
The deletion of the Ctr1 importer impaired copper acquisition by the fungus and also exerted pleiotropic effects on C. albicans (Fig 5). Deletion of CTR1 in the SC5314 background rendered C. albicans cells sensitive to copper deprivation (Fig 5A), thereby recapitulating the phenotype of ctr1 cells in a different, ura3-, genetic background [53]. Furthermore, the ctr1 null mutant was sensitive to iron chelating agents (Fig 5B), thereby underscoring the role of copper in iron acquisition [55]. We note that while BPS is generally viewed as an iron-specific chelator, it can also bind other metals. Therefore, we do not exclude the possibility that the chelation of other metals may contribute to the observed effects of BPS. Mutant ctr1 cells were also sensitive to oxidative (H 2 O 2 ) stress. Interestingly, these phenotypes were suppressed by copper supplementation, indicating that they were caused by intracellular copper deprivation. Iron scavenging and oxidative stress resistance promote C. albicans pathogenicity [56][57][58], and, therefore, it was not surprising that ctr1 cells displayed attenuated virulence during systemic infection (Fig 3D).
Interestingly, C. albicans ctr1 cells failed to utilise glycerol or ethanol, and their ability to assimilate glucose and oleic acid was also compromised (Fig 5D). These defects in carbon assimilation were rescued by copper supplementation, but not by ferric ions or haemoglobin iron (Fig 5D). Thus they were mediated by copper, not iron, insufficiency resulting from the lack of the high affinity copper importer. Furthermore, the growth defects of C. albicans ctr1 cells were accompanied by changes in core metabolic pathway gene expression (Fig 5E, S10  Table). It has been established that C. albicans dynamically readjusts its core metabolism during the course of systemic infections. For example, the CDC19 gene encoding the glycolytic enzyme pyruvate kinase is down-regulated, while transcripts for the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (PCK1) and the glyoxylate cycle enzyme malate synthase (MLS1) are up-regulated (Fig 5F, S11 Table) [52,59,60]. Indeed, it has been demonstrated that the ability to assimilate alternative carbon sources is critical for fungal virulence [59,61,62]. Therefore, the influence of Ctr1-mediated copper sequestration upon C. albicans virulence is probably mediated in part through promotion of efficient carbon assimilation.

Discussion
Infection sites comprise perpetually changing microcosms, shaped by the dynamic interactions between the pathogen and the host. Micronutrients play a central role in this dynamic interplay during disease progression. Consequently, nutritional immunity has evolved as a mechanism by which the host manipulates local micronutrient concentrations to the detriment of the invading microbe in an attempt to contain an infection [4,9]. The case for nutritional immunity is well documented for iron, and bacterial and fungal pathogens including C. albicans require effective iron acquisition mechanisms for full virulence [56,58]. Here we demonstrate another layer of complexity governing the pathogen-host tug of war-the regulation of copper homeostasis.
Our data clearly indicate that disseminated fungal infections affect copper metabolism in organs that promptly clear the infection (the liver and spleen), as well as organs that become colonized (the kidney). These changes coincide with previously described [12] infection-associated perturbations in iron metabolism in these organs. Unsurprisingly, the local changes in copper levels are sensed by C. albicans. Similarly to iron acquisition [12,56,58], intact copper acquisition pathways (Fig 3D) are required for full C. albicans virulence. Early in the infection, high expression of fungal Crp1 high affinity efflux pump accompanies increased copper levels, Fig 5. High affinity copper acquisition influences fungal pathogenicity factors. C. albicans ctr1 cells are susceptible to copper deprivation imposed using the copper chelator BCS (95 μM) (A). The C. albicans ctr1 mutant is also sensitive to iron chelation by BPS (B). The C. albicans ctr1 strain is highly sensitive to H 2 O 2 (C). Both of these defects are reversed by copper supplementation. Further, the C. albicans ctr1 mutant displays growth defects on various carbon sources, a phenotype that is suppressed by copper supplementation, but not by haem or Fe 3+ addition (D). Images were taken after 72 h incubation at 30°C. Number of cells spotted in every subset is given at the bottom of the panel. (E) The transcript levels for key metabolic genes are perturbed in ctr1 cells, in the absence (top) and presence (bottom) of copper. These data represent averages of three biological replicates with two technical replicates. Numerical data are given in S10 Table. (F) Changes in C. albicans transcript levels during renal colonisation. These data represent duplicate measurements from at least four biological replicates from the kidneys of animals at early (24 h, replicates E1-E4, 4 animals) or late (96 h, replicates L1-L5, 5 animals) infection stage. Colour scale is as in (E). Fungal transcript abundances were normalised to the ACT1 mRNA. Numerical data are given in S11 Table. doi:10.1371/journal.pone.0158683.g005 possibly to evade copper poisoning [51] (Fig 3). As immune infiltrates create zones of ironexclusion around fungal cells and iron becomes limiting for the fungus [12], C. albicans increases the expression of its high affinity copper importer Ctr1. It is tempting to speculate that the down-regulation of fungal copper-dependent reductive iron acquisition in the course of infection [12] is partly driven by changes not just in iron but also copper availability. Access to copper also impacts other key factors that enhance the virulence of C. albicans, namely, oxidative stress resistance [57,63,64], carbon assimilation and the expression of core metabolic genes [42,59,[65][66][67].
In conclusion, we demonstrate the impact of systemic infection on copper homeostasis in the host, the interplay between copper and other essential micronutrients such as iron in infected tissues in situ, and the effects of host copper on fungal copper homeostasis and carbon assimilation. It is clear that "nutritional immunity" is more than just a static limitation of a single nutrient. Rather, this phenomenon is a cumulative outcome of dynamic, simultaneous spatio-temporal alterations of multiple micronutrients impacting the progression of infection. Early exponential phase C. albicans SC5314 cells grown at 30°C in YNB-Glucose medium were exposed to 170 μM iron chelator BPS or 250 μM copper chelator BCS for 30 min, before fixing in RNAlater. Gene expression was analysed by qRT-PCR, as described in Materials and Methods. Relative transcript abundances are normalised to ACT1, and expressed as ratios to control (i.e., no iron or copper chelation). Values are averages of duplicate measurements from three independent experiments. Numerical data are given in S9 Table. (TIF) S1  Table. Normalised qRT-PCR data for Fig 1A. (XLSX) S5 Table. Normalised qRT-PCR data for Fig 2D. (XLSX) S6 Table. Normalised qRT-PCR data for Fig 3A. (XLSX) S7  Table. Normalised qRT-PCR data for Fig 5E. (XLSX) S11 Table. Normalised qRT-PCR data for Fig 5F. (XLSX)