Changes in mammalian copper homeostasis during microbial infection

Abstract

Animals carefully control homeostasis of Cu, a metal that is both potentially toxic and an essential nutrient. During infection, various shifts in Cu homeostasis can ensue. In mice infected with Candida albicans, serum Cu progressively rises and at late stages of infection, liver Cu rises, while kidney Cu declines. The basis for these changes in Cu homeostasis was poorly understood. We report here that the progressive rise in serum Cu is attributable to liver production of the multicopper oxidase ceruloplasmin (Cp). Through studies using Cp−/− mice, we find this elevated Cp helps recover serum Fe levels at late stages of infection, consistent with a role for Cp in loading transferrin with Fe. Cp also accounts for the elevation in liver Cu seen during infection, but not for the fluctuations in kidney Cu. The Cu exporting ATPase ATP7B is one candidate for kidney Cu control, but we find no change in the pattern of kidney Cu loss during infection of Atp7b−/− mice, implying alternative mechanisms. To test whether fungal infiltration of kidney tissue was required for kidney Cu loss, we explored other paradigms of infection. Infection with the intravascular malaria parasite Plasmodium berghei caused a rise in serum Cu and decrease in kidney Cu similar to that seen with C. albicans. Thus, dynamics in kidney Cu homeostasis appear to be a common feature among vastly different infection paradigms. The implications for such Cu homeostasis control in immunity are discussed.

Graphical Abstract

During infection, liver Cu rises while kidney Cu falls, and liver production of Cu-ceruloplasmin(Cp) helps restore serum levels of Fe

INTRODUCTION

Transition metals in biology have a dual role: they are essential micronutrients and co-cofactors for many vital enzymes, but also have the potential to become toxic. Cu is a prime example of such double-edged functionality. The metal is a required co-factor for many enzymes involved in oxygen chemistry and electron transfer, including cytochrome C oxidase for mitochondrial respiration, superoxide dismutase enzymes for anti-oxidant defense, and multi-Cu oxidases that couple oxygen reduction to oxidation of a variety of substrates ^{1, 2}. At the same time, Cu is potentially toxic through its ability to generate reactive oxygen species (ROS) via Fenton chemistry ³, to mis-metallate non-cupro enzymes ⁴, and to disrupt Fe-S clusters ^5–7. Unicellular microbial pathogens that are in direct contact with their environment can be particularly vulnerable to extreme highs and lows in Cu, and during infection the animal host can exploit Cu as weaponry against these microbes ^8–11.

The best studied example of Cu in innate immunity involves the Cu burst of macrophages. Macrophages engulf pathogens into the phagolysosome that is essentially an encasement of chemical toxins, including ROS ¹², reactive nitrogen species ^{13, 14} and elevated Cu ^{10, 15}. The elevation in phagolysosome Cu involves both increases in Cu uptake through the cell surface CTR1 Cu importer and pumping of Cu into the phagolysosomal compartment by ATP7A ¹⁶, one of two mammalian Cu-transporting P-type ATPases ¹⁷. In response to Cu elevation in the phagolysosome, successful pathogens have evolved elaborate machineries for Cu tolerance, many of which are virulence factors ^18–23.

The animal host can also withhold Cu from invaders and this appears to be especially true in the case of eukaryotic pathogens such as fungi, which heavily rely on Cu as a micronutrient due to numerous intracellular cuproenzymes ^{8, 24}. Interestingly, the macrophage that can attack pathogens with excess Cu can also starve certain microbes of Cu nutrients. When the pulmonary pathogen Histoplasma capsulatum invades macrophages, intraphagolysomal Cu is initially elevated, but at later stages, phagolysosomal Cu diminishes to levels that induce Cu starvation stress in the fungus ^{25, 26}. Another pulmonary fungal pathogen, Cryptococcus neoformans, also encounters extreme highs and lows in Cu during invasion of host. In the lung, pulmonary macrophages attack the C. neoformans with high Cu, but upon dissemination to the brain and cerebrospinal fluid, the fungus is deprived of Cu and activates Cu limitation stress responses ^27–29. The ability to sense Cu limitation and overcome Cu starvation is also important for Aspergillus fumigatus virulence in mice ³⁰.

Another fungal pathogen that is subject to fluctuations in host Cu is C. albicans. C. albicans is a common component of human flora that can cause life-threatening systemic infections ^{31, 32}. In a murine model of disseminated candidiasis, the kidneys, which are the main target organ of infection ³³, display biphasic changes in Cu. At early stages of C. albicans infection, kidney Cu rises, then progressively falls at later stages ^34–36. These changes in host Cu are sensed by the invading fungi, and C. albicans responds first by upregulating Cu export, followed by a Cu starvation stress response involving upregulation of Cu import and downregulation of the major Cu protein, Cu/Zn SOD1 ^{34, 35}. C. albicans also infects the spleen and liver in this model of candidiasis, and like the kidney, spleen Cu also displays a biphasic response of initially elevated Cu, followed by Cu loss, while liver Cu initially remains stable then rises later during infection ³⁴. The fluctuations in kidney Cu during infection are particularly surprising, in that in uninfected animals, kidney Cu typically remains constant relative to other tissues in cases of Cu deficiency or Cu excess ^37–40. The mechanism by which kidney Cu fluctuates during C. albicans infection is not known. ATP7B, one of two Cu transporting ATPase of mammals, is seen to rise in the kidney during C. albicans invasion ³⁵, although the contribution of ATP7B to changes in kidney Cu during infection has not been examined. Furthermore, it is not known whether the fluctuations in kidney Cu are the direct result of tissue infiltration by the pathogen or are part of a more global immune response of the host. It is noteworthy that the loss in kidney Cu is paralleled by a dramatic and progressive rise in serum Cu ³⁴. The basis of this rise is unknown, and whether it is connected to fluctuations in the kidney is unclear.

Here we explore the basis for these changes in host Cu during C. albicans infection. Our findings show that the progressive rise in serum Cu is due to increased liver production of ceruloplasmin (Cp), a multi-Cu oxidase and ferroxidase produced in the liver. This increased Cp production was found to be unconnected to changes in kidney Cu but rather was necessary to control circulating pools of Fe during infection. Furthermore, we demonstrate that the changes in kidney Cu are not due to direct invasion of the tissue by a microbe, since similar fluctuations in Cu without kidney invasion were seen with the malaria parasite P. berghei which remains intravascular. Thus, these changes in host kidney Cu appear to be part of a global immune response. Our findings reveal organ-specific changes in host Cu homeostasis as a generalized component of host immunity.

EXPERIMENTAL

Mouse models of infection

All mouse studies were carried out in accordance with the National Institutes of Health guidelines for the ethical treatment of animals. The protocol was approved by the Institutional Animal Care and Use Committee of the Johns Hopkins University medical institutions, mouse protocol number MO13M264 (C. albicans infection) and number MO19H11 (P. berghei).

For the lateral tail vein model of disseminated candidiasis, studies involved both male and female mice and no sex differences were noted within any experiments presented in this work, consistent with previous findings on mouse tissue and serum Cu during C. albicans infection ^{34, 36}. C. albicans strain SC5314 was used for all studies. Fungal cells were grown to stationary phase overnight in a yeast extract, peptone, and 2% (wt/vol) dextrose-based media (YPD). Cells were harvested and washed 2x in phosphate buffered saline (PBS) and cell density determined by optical density at 600 nm (OD₆₀₀). The cell concentration of the inoculum was confirmed by cell enumeration on a hemocytometer (Hausser Scientific). 5×10⁵ C. albicans cells suspended in 100 μL PBS were injected into the lateral tail vein of 9–11 week old BALB/c or C5BL/6J strains of mice or the indicated mutants. Mice were sacrificed at 24, 48 and 72 hours after infection. Blood was collected and allowed to clot on ice before isolation of serum by centrifugation at 1000 x g for 10 minutes. Organs were perfused with PBS as previously described ³⁴ and kidneys and liver were collected and processed for CFUs and metal analysis. The Cp^−/− mice have been previously described ⁴¹. The liver specific Atp7b^ΔHep and global knockout Atp7b^−/− mice have been previously described ^{40, 42}.

To measure colony forming units during C. albicans infection, perfused tissues were weighed before homogenization by mechanical force in 1 mL of PBS. After homogenization, samples were serially diluted in PBS and plated onto YPD medium with 1% Penicillin, 1% Streptomycin (Quality Biological Inc., 120–095-721).

For P. berghei experiments, 7-week-old female BALB/C mice were infected by injection of 1×10⁶ parasites of P. berghei strain ANKA into the peritoneum of mice as previously described ⁴³. Mice were sacrificed at 7 and 14 days post infection. Blood and PBS perfused kidneys were collected at the time of sacrifice and processed for metal analysis identically to C. albicans infected samples.

Biochemical analyses of serum and tissues

Serum was diluted 1:50 in MilliQ water and Cu content was measured on a PerkinElmer Life Sciences AAnalyst 600 graphite furnace atomic absorbance spectrometer (AAS). Tissue samples were weighed and digested overnight at 90^oC in 1 mL of 20% (vol/vol) Ultrex II Ultrapure nitric acid. Samples were diluted to 2% nitric acid and Cu content was measured via AAS.

To monitor serum oxidase activity, o-dianisidine dihydrochloride was used as substrate as previously described ⁴⁴. In a 96 well plate (Falcon, 353072) the following was combined in duplicate samples: 3.75 μL serum, 56.5 μL of 10 μM diethylenetriameinepentaacetic acid in a 0.1 M sodium acetate buffer at pH 5.5, and 30 μL of 2.5 mg/mL o-dianisidine dihydrochloride (Sigma Aldrich, D3252–25G). Samples were incubated at 37^oC and following 5 and 60 minute time points, 150 μL 9 M sulfuric acid was added and absorbance measured at 540 nm using Synergy HT (Biotek) to yield A₅ and A₆₀, respectively. Oxidase activity in units/L was calculated as (A₆₀ – A₅) x 113.6.

Serum Fe (SI) and unsaturated iron binding capacity (UIBC) were both determined using a ferrozine based assay from Pointe Scientific (I7504–60). To measure SI, 100 μL of 220 mM hydroxylamine hydrochloride in pH 4.5 acetate buffer with surfactant was combined with 20 μL of either serum sample, water, or 500 μg/dL ferrous chloride Fe standard in hydroxylamine hydrochloride. The absorbance at 560 nm (A₅₆₀) was measured on an Eon plate reader (Biotek) (A₁ reading). 2 μL of 16.7 mM ferrozine in hydroxylamine hydrochloride was then added, following by incubation at 37^oC for 10 minutes and A560 measurements (A₂ reading). SI in μg/dL was calculated as (A_2serum -A_1serum)/(A_2standard -A_1standard) x 500. For UIBC, 80 μL of 500 mM Tris, pH 8.1 with surfactant and 0.05% (w/v) sodium azide was mixed with 60 μL of water, 20 μL of serum sample and 40 μL of the aforementioned Fe standard, or 20 μL of water and 40 μL of Fe standard. A₅₆₀ was measured (A1). 2 μL of 16.7 mM ferrozine in hydroxylamine hydrochloride was added and incubation proceeded at 37^oC for 10 minutes followed by A560 measurements (A2). UIBC in μg/dL was calculated as 500-(A_2sample-A_1sample)/(A_2standard-A_1standard) x 500. Total iron binding capacity (TIBC) was calculated as SI + UIBC. Transferrin saturation was then calculated as SI/TIBC x 100.

Ceruloplasmin protein levels were determined by immunoblot. 5 μL of serum was loaded onto a 4–12% Bis-Tris gel (Thermo Fisher). Blots were probed with anti-ceruloplasmin (Abcam, Ab19171) at a 1:10,000 dilution and secondary Alexa Fluor 680 donkey anti-goat (Invitrogen, #A-21084) at 1:10,000 dilution. For albumin, 1 μL of serum was run on a 4–12% Bis-Tris gel (Thermo Fisher) and stained with Coomassie brilliant blue (Fisher Biotech, BP101–50). Densitometry quantification of Cp and albumin was determined using ImageJ version 1.49 (NIH software).

All statistical calculations were made using RStudio. One-way ANOVA with Tukey post-test was used when comparing multiple groups with one independent variable, and two-way ANOVA with Tukey post-test was used to compare group means with two independent variables. Two-tailed student’s t-tests were used when comparing two groups. Least squares analysis using the lm() function was used to determine the slope and intercept of the line in Fig. 1E. In all graphs, the values plotted are from individual mice; the central line and error bars represent the mean and standard error of the mean (SEM), respectively.

Fig. 1.

The rise in serum Cu in C. albicans-infected mice is attributable to holo-ceruloplasmin. Nine week old female BALB/c mice were infected with C. albicans strain SC5314 and sacrificed at the indicated time points. Blood was harvested at sacrifice and serum was isolated. (A) Serum Cu during infection for 5–8 mice per group. ***P<0.001, ****P< 0.0001 as determined by one-way ANOVA with Tukey post-test. (B) Cp levels in serum from three mice per time point were assayed via immunoblot (B top) and normalized to albumin levels visualized by Coomassie blue staining (B bottom). (C) The Cp to albumin ratio was measured by densitometric tracings of the images in part B. *P<0.05, ***P<0.001 as determined by one-way ANOVA. The slight loss in albumin at 72 hours does not affect the statistical significance of Cp increases at 48 and 72 hours. (D) Oxidase activity of Cp was assayed via o-dianisidine dihydrochloride oxidation. Results shown are from 5 mice per group. ***P<0.001,**** P< 0.0001 as determined by one-way ANOVA with Tukey post-test. (E) Correlation between oxidase activity (from D) and serum Cu (from A), slope= 1.76+/− 0.14, intercept=−3.23+/− 2.66, R²=0.894, p=2.03×10⁻¹⁰. Line parameters and statistics determined by lm() function in RStudio. Values from individual mice are plotted with middle line representing the mean with error bars showing +/− SEM. Brackets indicate comparisons between time 0 (uninfected) and either the 48 or the 72 hour time point. The increases at 24 hours compared to time 0 controls are not statistically significant.

RESULTS AND DISCUSSION

The accumulation of Cu-ceruloplasmin in serum of mice infected with C. albicans

Infection of 9 week old BALB/c mice with C. albicans via lateral tail vein injection was associated with a progressive increase in total serum Cu over 72 hours post-infection (Fig. 1A). The progressive rise in serum Cu was observed in males and females (Fig. 1A) and in different strains of mice (BALB/c and C57BL/6) ^{34, 36}. The basis for serum Cu increase during this model of candidiasis was not understood. We considered that this rise in Cu might reflect elevations in the multicopper oxidase ceruloplasmin (Cp), an acute phase protein in the serum ⁴⁵ that normally accounts for 40–70% of total serum Cu ⁴⁶. Analysis of serum Cp by immuno-blotting revealed an increase in Cp up to 72 h post infection (Fig. 1B,C). Compared to Cp, serum albumin stayed relatively constant and only decreased slightly (≈12%) at 72 h (Fig. 1B). Since Cp circulates as both apo and Cu-loaded protein ⁴⁷, we measured Cp activity as an indicator of metallated protein using the o-dianisidine dihydrochloride oxidation assay ^{44, 48}. As seen in Fig. 1D, there was a dramatic rise in o-dianisidine reactivity during infection that increased progressively up to 72 h. Comparison of serum Cu and serum o-dianisidine reactivity revealed a strong correlation (Fig. 1E), indicating that the elevated serum Cu seen during C. albicans infection is attributable to serum Cp.

Although serum o-dianisidine reactivity is typically used as indicator of Cp activity ⁴⁸, other factors can react with o-dianisidine including the closely related multicopper oxidase hephaestin ⁴⁹ or diamine oxidases ⁴⁸. To determine whether Cp is the main contributor to the observed oxidase activity, we infected Cp^−/− null mice with C. albicans and measured changes in Cu and oxidase activity in the mouse serum. The infected Cp^−/− mice showed fungal burden in kidneys similar to the infected WT C57BL/6 mice (Fig. 2A), but had larger weight loss compared to the WT at 72 h (Fig. 2B), indicating Cp may play some role in guarding against infection in this model. Nevertheless, mouse survival was not substantially affected by Cp^−/− deletion, and these mice therefore provided a valuable model to directly test the contribution of Cp to changes in serum Cu. As seen in Fig. 2C, the WT C57BL/6 mice exhibited the characteristic rise in serum Cu over 72 h of infection. By comparison, the Cp^−/− mice had very low basal serum Cu that failed to rise during the course of infection. These findings support the results of Fig. 1, and demonstrate that the strong elevation in serum Cu during disseminated candidiasis can be attributed to Cu containing Cp.

Fig. 2.

The effects of Cp null mutations on serum Cu, weight loss and renal lesions during infection. Male (triangles) and female (circles) 11 week old Cp^−/− null mice and age matched WT (“Wt’) controls were infected with C. albicans as in Fig. 1. (A) Colony forming units (CFU) in the kidneys harvested at the indicated time points in hours; n= 4–7 mice per group. Analysis via two-way ANOVA showed no statistically significant difference between WT and Cp^−/− mice. (B) Weight loss at 72 hours post infection is significantly higher in Cp^−/− mice as determined via two-tailed Student’s t-test, n=8 mice per group, *P=0.0281. (C) Total serum Cu levels are significantly higher in WT mice compared to Cp^−/− mice at all time points as determined by two-way ANOVA with Tukey post-test; 4–9 mice per group. ****P<0.0001 where comparisons are made between serum Cu from WT versus Cp^−/− mice at each time point. This two-way ANOVA test also showed that in Cp^−/− mice, there was no statistically significant difference in serum Cu levels at 24 or 72 hours compared to the 0 time point control. Values for individual mice are plotted with middle line representing the mean with error bars showing +/− SEM.

Where is the Cu-ceruloplasmin coming from in this mouse model of candidiasis? Cp can be secreted from a variety of tissues and cells including macrophages, mononuclear cells, the mammary gland, the choroid plexus, and the kidney ^50–53, although a principle source is the liver ⁵⁴. To test whether the Cu-loaded Cp that hyperaccumulates during disseminated candidiasis is derived from the liver, we used mice containing a liver specific deletion in ATP7B, the P-type Cu ATPase in the secretory pathway that provides Cp with Cu ⁵⁵. These Atp7b^ΔHep mice hyperaccumulate Cu in the liver and Cp circulates in largely the apo form as determined by a native gel assay ⁴². Consistent with these previous findings, we observe that uninfected Atp7b^ΔHep mice show low Cp activity as revealed by o-dianisidine reactivity (Fig. 3A). These mice also exhibit very low basal levels of serum Cu in uninfected mice (Fig. 3B). Figure 3C shows that during infection with C. albicans, Atp7b^ΔHep mice have no significant changes in kidney fungal burden. Interestingly, serum Cu and Cp activity in Atp7b^ΔHep mice remained exceptionally low during the course of infection compared to the dramatic rises in WT litter mate controls (Fig. 3A, B). Despite no increase in Cp activity, the Atp7b^ΔHep mice showed elevations in Cp protein by immunoblot, similar to levels seen in WT mice (Fig. 3D). Thus, the elevated Cp that accumulates in infected Atp7b^ΔHep mice appears to be apo and inactive. These studies show that the high level of Cu-loaded and enzymatically active Cp that accumulates in the serum of C. albicans infected mice is derived from the liver and is dependent on hepatic ATP7B.

Fig. 3.

During infection with C. albicans, serum ceruloplasmin is produced from the liver. Male (triangles) and female (circles) 11 week old Atp7b^ΔHep (7b^ΔHep, red) mice and WT littermate controls (black) were infected with C. albicans. (A) Oxidase activity of mouse serum was measured by o-dianisidine dihydrochloride oxidation; n=3–6 mice per group. *P=0.013; **P=0.0030; **** P< 0.0001. (B) Serum Cu levels are shown for 3–6 mice per group. ***P=0.00061; ****P<0.0001. (C) There was no significant difference in kidney CFUs between Wt and Atp7b^ΔHep mice in 3–4 mice per group analyzed. (D) CP in serum at 0 and 72 hours was detected by immunoblot with albumin as loading control. Graph at bottom shows band intensity quantified by densitometry. (A,B) Statistical significance was determined by two-way ANOVA with Tukey post-test where comparisons are made between WT and Atp7b^ΔHep samples at each time point. Values plotted are individual mice with middle line representing the mean with error bars showing +/− SEM.

Ceruloplasmin and transferrin Fe during C. albicans infection

While many roles have been ascribed to Cp, a major function is to act as a ferroxidase for ferrous iron to facilitate loading of Fe⁺³ onto serum transferrin ^{41, 56}. A previous study has shown that in the mouse tail vein model for disseminated candidiasis, infection with C. albicans led to a great reduction in transferrin-Fe levels at 24 h post-infection, presumably due to release of the pro-inflammatory hormone hepcidin that triggers an anemia of inflammation response ^{57, 58}. We corroborated these findings and observed a 3 fold drop in total Fe and transferrin-Fe 24 h post infection in BALB/c mice (Fig. 4A,B). Unexpectedly however, the mice seemed to recover over time and serum Fe and transferrin saturation were partially restored 72 h post infection (Fig. 4A–B). A similar U-shape curve in serum Fe during C. albicans infection could be seen in WT C57BL/6 mice (Fig. 4C). This recovery in serum Fe at later stages of fungal infection had not been previously described and we tested whether Cp was involved. As seen in Fig. 4C, the basal level of serum Fe in uninfected Cp^−/− mice was much lower than age-matched controls, consistent with previous results ^{41, 59}. Notably, this low level of serum Fe in Cp^−/− mice appeared to remain relatively constant over the course of infection (Fig. 4C). Based on these findings, it appears that the marked elevation of Cp during infection can help in the recovery of serum Fe deficiency during later stages of infection.

Fig. 4.

Changes in serum Fe during infection. Total serum Fe and transferrin Fe were measured in 9 week old BALB/C (A-B) and 11 week old C57BL/6 mice and Cp ^−/− mutants (C). (A-B) Total serum Fe and transferrin saturation were determined in serum of 4–9 mice per group at the indicated time points. *P <0.05; ***P<0.001; ****P<0.0001 (C). Total serum Fe was determined in the serum of 2–5 WT and Cp ^−/− mice per group at the designated time points. *P <0.05; ***P<0.001. Statistical significance was determined by one-way ANOVA with Tukey post-test for (A-B) and two-way ANOVA with Tukey post-test for (C). Sex of mice denoted by given markers: male (triangles) and female (circles). Values plotted are individual mice with middle line representing the mean with error bars showing +/− SEM.

The parallel inductions of Cp and hepcidin during infection seem at odds with one another: while hepcidin works to reduce serum Fe levels, Cp can increase serum Fe through transferrin loading. Our findings suggest a model in which Cp induction serves to restrict the extent of anemia of inflammation induced by hepcidin, which might otherwise be detrimental to the host. However, we cannot exclude other possible roles for Cp during infection. For example, Cp has been shown to act as an antioxidant ^{45, 60–62} and may therefore protect cells from the free radical attack of host immunity. Cp can also act in the oxidation of molecules such as hormones ⁶³ and in nitric oxide oxidation ^{64, 65}. One or more of these alternative biochemical activities may also contribute to Cp function as an acute phase protein. Regardless of its precise activity during infection (recovery in serum Fe versus pro-oxidant or anti-oxidant reactions), Cp seems to contribute to animal fitness during disseminated candidiasis, as witnessed by the larger weight loss in Cp^−/− mice compared to WT controls (Fig. 2B).

Changes in tissue Cu during C. albicans infection and ceruloplasmin

Aside from the rise in serum Cu, infection with C. albicans leads to specific changes in Cu content of tissues including the liver and kidney ^34–36. We sought to test whether these changes in tissue Cu are connected to the strong elevation in serum Cu and serum Cp.

With regard to the liver, total Cu levels significantly increase by 72 h in the murine model of disseminated candidiasis ³⁴. Why the liver accumulates elevated Cu during infection was unknown. As one possibility, Cu may be elevated to meet the demand for liver production of Cu-Cp. We tested this hypothesis using Cp^−/− null mice. As seen in Fig. 5A, WT C57BL/6 and Cp^−/− null mice have similar basal levels of total liver Cu prior to infection. Following 72 h of infection, C57BL/6 mice exhibited a rise in liver Cu similar to that previously seen with infected BALB/c mice ³⁴. By comparison, liver Cu 72 h post infection remained essentially unchanged from basal levels in the Cp^−/− null mice (Fig. 5A). Thus, liver production of Cp occurs in parallel with the increase in liver Cu observed.

Fig. 5.

Presence of ceruloplasmin and ATP7B affect liver Cu but not kidney Cu. (A) Liver Cu content in WT and Cp^−/− mice was measured in 3–4 mice per group. *P=0.038, where comparisons are made between WT and Cp^−/− mice at 72 hours. (B) Kidney Cu levels in 4–8 WT C57BL/6 mice per group (left panel) and 4–9 Cp^−/− mice per group (right panel). *P<0.05; ***P<0.001. (C) Kidney Cu levels in 4–5 WT and Atp7b^−/− mice per group during infection. *P<0.05. Statistical significance was determined by two-tailed Student’s t-test (A,C) and one-way ANOVA with Tukey post-test for (B). Brackets in part B and C indicate statistically significant differences comparing two time points for a particular mouse group (WT, Atp7b^−/− or Cp^−/−). Values plotted are individual mice with middle line representing the mean with error bars showing +/− SEM.

While hepatic Cu steadily increases during C. albicans infections, renal Cu levels increase at 24 h followed by a decrease at 72 h ^34–36, and a similar phenomenon has been reported for the spleen ³⁴. The reduction in renal Cu is not due to urinary losses during infection ³⁴, and is concomitant with high liver production of Cu-Cp (Fig. 1B–D). Could these events be linked? There is precedence for cross tissue communication in Cu homeostasis, e.g., cardiac Cu deficiency signals Cu mobilization from other tissues ^{66, 67}. It was therefore possible that during infection, the need to produce Cu-Cp in the liver might signal compensatory Cu losses from other tissues. We used Cp^−/− mice to address whether CP production is sufficient to trigger Cu changes in extra-hepatic tissues. In WT C57BL/6 mice, kidney Cu levels fall at 72 h, as previously reported ^34–36 (Fig. 5B, left). However, the same change in kidney Cu during infection was also observed in Cp^−/− mice (Fig. 5B, right). These studies exclude liver production of Cu-Cp as a signal to drive Cu loss from the kidney.

The decreases in kidney Cu accumulation at late stages of C. albicans infection could result from either reductions in Cu uptake or increases in Cu export through Cu exporting ATPases, ATP7A or ATP7B ^{17, 68, 69}. Brown and colleagues have shown that levels of ATP7B mRNA and protein are increased in the kidney during C. albicans invasion³⁵. We examined whether this elevated ATP7B was responsible for the fluctuations in kidney Cu by examining kidney Cu levels in whole animal Atp7b ^−/− mice. As seen in Fig. 5C, kidney Cu levels in the Atp7b ^−/− mice were observed to drop at 72 h infection, similar to wild type littermate controls. Thus, ATP7B by itself is not responsible for the changes in kidney Cu homeostasis during infection.

If ATP7B is not the key, the loss in kidney Cu during infection may result from inhibition of Cu uptake by CTR1 or activation of Cu export by ATP7A. Inhibition of CTR1 is not likely, since kidney levels of this Cu importer are seen to increase, not decrease, during infection with C. albicans ³⁵. Furthermore, in kidneys of Atp7b ^−/− mice, CTR1 levels are unaffected, while the Cu exporter ATP7A shows a dramatic shift in localization ^{70, 71}. Specifically, ATP7A localizes in the basolateral membranes of cortical tubules in kidneys of Atp7b ^−/− mice to facilitate Cu export and compensate for loss of ATP7B ⁷¹. This ability of ATP7A to correct for ATP7B deficiency may explain why Atp7b ^−/− mice still loose kidney Cu during C. albicans infection. Hence, it is possible that ATP7A (and/or ATP7B) drives kidney Cu loss during infection, and in support of this notion, both transporters increase in the kidney upon C. albicans invasion ³⁵. Future investigations into ATP7A await the generation of kidney-specific deletions or conditional alleles, as ATP7A is essential for mouse viability ⁷².

Host Cu responses in a distinct model of infection: P. berghei

The fluctuations in renal Cu during C. albicans infection are remarkable, in that under non-infectious conditions, kidney Cu remains relatively constant compared to other tissues in cases of Cu deficiency or Cu excess ⁷¹. As one possibility, the loss in kidney Cu could represent a special case of C. albicans infiltrating and damaging the tissue. The kidney is the site of highest fungal burden in the mouse model of disseminated candidiasis ³³ and the Cu response may simply be a localized effect of colonization. To address whether kidney infiltration triggers Cu loss, we investigated host Cu changes in mice infected with a pathogen that does not predominantly invade the kidney. We chose the intravascular malaria parasite P. berghei that infects erythrocytes, preferentially reticulocytes ⁷³ but not kidney cells nor endothelium, and does not produce kidney lesions as found with C. albicans infections ⁷⁴. During blood stream infection with P. berghei, we observed elevations in serum Cu (Fig. 6A) and Cp activity (Fig. 6B) similar to that seen with C. albicans infection (Fig. 1A, D). Most interestingly, we also observed a decrease in kidney Cu levels at later stages of infection (Fig. 6C). The loss in kidney Cu with P. berghei is similar that seen during C. albicans infections (Fig. 5B,C and ^34–36), albeit on a longer timeline. Thus, the kidney Cu response to infection does not appear to be a specific effect of infiltration of this organ by the fungal pathogen, but rather a more global response to systemic infections.

Fig. 6.

Changes in serum and kidney Cu during infection with P. berghei. Seven week old female BALB/c mice were infected with P. berghei and sacrificed at the indicated time points. Results shown are pooled from two experiments with 8 control mice, 3 mice sacrificed at day 7, and 8 mice after 14 days. (A) Serum Cu measurements of P. berghei infected mice. ****P<0.0001. (B) Oxidase activity of P. berghei infected serum in 19 mice. **** P< 0.0001. (C) Kidney Cu levels of P. berghei infected mice. **P<0.01. Statistical significance was determined by one-way ANOVA with Tukey post-test for (A-C). Values plotted are individual mice with middle line representing the mean with error bars showing +/− SEM.

Nutritional immunity for Copper?

Why do kidney Cu levels decrease during infection? This appears to represent a specific host response to infection or inflammation because such changes in kidney Cu are typically not seen under non-infection conditions of Cu excess or Cu deficiency ^{37, 39, 40, 71, 75}. The loss in tissue Cu could be a nutritional immunity response of the host, i.e., an intentional attempt to thwart pathogen growth by starving the microbe of its Cu nutrients. Indeed C. albicans is subject to Cu starvation when it invades the kidney and responds by increasing fungal Cu uptake and sparing utilization of Cu as a co-factor ^{34, 35}. This withholding of Cu from C. albicans reflects a type of nutritional immunity that is unlike similar mechanisms involving calprotectin sequestration of Mn and Zn ^{76, 77}, in that the Cu limitation is tissue wide rather than localized to sites of infections ^{34, 35}. However, since the loss in kidney Cu is also observed with an infectious agent that does not infiltrate the kidney (Fig. 6C), it more likely represents a global response to infection and inflammation rather than localized nutritional immunity. As an alternative possibility, the loss in kidney Cu may not reflect a bona fide nutritional immunity response, but rather a shift in prioritization of whole animal Cu away from the kidney and towards the liver to meet demands for Cu-Cp production. If this model is correct, the mobilization of kidney Cu is not a direct response to increases in liver Cu and liver Cp, as kidney Cu still declines in the Cp^−/− mouse. A similar model invoking cross-tissue redistribution of Cu has been used to explain mobilization of Cu from the liver towards the heart during cardiac Cu deficiency ^{66, 67}. How cross-tissue Cu homeostasis is communicated during infection is still unknown, but a likely possibility involves cytokine release as part of the acute phase response. The identification of Cu homeostasis signals during inflammation in future studies would greatly enhance our understanding of the complex roles this metal plays during microbial infection.

CONCLUSIONS

Together, our findings underscore the importance of host Cu homeostasis during infection. During disseminated candidiasis, murine Cu homeostasis undergoes several dynamic changes, including fluctuations in Cu levels of the kidney, liver and serum. We find that changes in serum and liver Cu are solely due to liver production of Cu-Cp and that this Cp acts in the recovery of serum Fe levels at late stages of infection. Kidney Cu levels fall as infection progresses and these changes occur independent of the rise in serum Cp or of the Cu transporting ATPase ATP7B. While the mechanism is still unknown, we propose a model in which the loss in kidney Cu is a global response to infection and inflammation, perhaps involving ATP7A and cytokines of the acute phase response. Kidney Cu loss is not specific to fungal invasion of the tissue but also occurs with infectious agents where the kidney itself is not colonized. Future studies to uncover the signals governing these dynamics in host Cu could greatly improve our understanding of how Cu homeostasis integrates into the larger picture of host inflammation.