Multicopper oxidase-1 is a ferroxidase essential for iron homeostasis in Drosophila melanogaster

Abstract

Multicopper ferroxidases catalyze the oxidation of ferrous iron to ferric iron. In yeast and algae, they participate in cellular uptake of iron; in mammals, they facilitate cellular efflux. The mechanisms of iron metabolism in insects are still poorly understood, and insect multicopper ferroxidases have not been identified. In this paper, we present evidence that Drosophila melanogaster multicopper oxidase-1 (MCO1) is a functional ferroxidase. We identified candidate iron-binding residues in the MCO1 sequence and found that purified recombinant MCO1 oxidizes ferrous iron. An association between MCO1 function and iron homeostasis was confirmed by two observations: RNAi-mediated knockdown of MCO1 resulted in decreased iron accumulation in midguts and whole insects, and weak knockdown increased the longevity of flies fed a toxic concentration of iron. Strong knockdown of MCO1 resulted in pupal lethality, indicating that MCO1 is an essential gene. Immunohistochemistry experiments demonstrated that MCO1 is located on the basal surfaces of the digestive system and Malpighian tubules. We propose that MCO1 oxidizes ferrous iron in the hemolymph and that the resulting ferric iron is bound by transferrin or melanotransferrin, leading to iron storage, iron withholding from pathogens, regulation of oxidative stress, and/or epithelial maturation. These proposed functions are distinct from those of other known ferroxidases. Given that MCO1 orthologues are present in all insect genomes analyzed to date, this discovery is an important step toward understanding iron metabolism in insects.

Multicopper ferroxidases catalyze the oxidation of ferrous iron (Fe²⁺) to ferric iron (Fe³⁺). Redox cycling between the two oxidation states is a critical aspect of iron metabolism because the two forms of iron have different properties: ferrous iron is soluble under most physiological conditions, but it participates in toxic radical formation, whereas ferric iron does not promote radical formation, but it is highly insoluble (1). Multicopper ferroxidases in yeast and algae function in the uptake of iron by oxidizing ferrous to ferric iron, which is then transported into the cell by a ferric iron permease (2, 3). The mammalian ferroxidases, hephaestin and ceruloplasmin, facilitate iron efflux by oxidizing extracellular ferrous iron that has been transported out of cells by a ferrous iron permease (ferroportin) (4, 5). Hephaestin is bound to the basolateral membrane of intestinal cells; thus, a reduction in hephaestin function leads to an accumulation of iron in the intestinal cells and a deficiency of iron in the rest of the body (6). A decrease in the function of ceruloplasmin (which is expressed as circulating and GPI-anchored isoforms) leads to an overaccumulation of iron in many tissues as a result of a decrease in iron efflux (7).

Iron metabolism in insects is not well understood, but some components of the iron metabolism pathways are known. Divalent metal transporter 1 (DMT1) transports iron from the diet into midgut epithelial cells (8). Ferritin has two functions in insects: iron storage (comparable to the function of mammalian ferritin) and iron transport (9). Insect ferritin is present in the endoplasmic reticulum and Golgi complex of a variety of cell types, including midgut cells, and iron-loaded ferritin is secreted into the hemolymph and is thought to transport iron to other parts of the body (9). Insect transferrin is present in the hemolymph, and, like mammalian transferrin, it may function as an iron transport protein (10, 11), although the mechanism of ferric-transferrin uptake is unknown (12). We wondered whether iron metabolism in insects requires the activity of a multicopper ferroxidase. Given that ferroxidases are necessary for iron metabolism in a diverse set of organisms (2, 3, 6, 7, 13, 14), it seemed likely that they would exist in insects. On the contrary, the multicopper ferroxidases with known physiological functions work in conjunction with an iron permease (2–5), whereas we and others have been unable to identify an iron permease gene in any insect genome (9).

To identify candidate insect ferroxidases, we evaluated the known set of insect multicopper oxidases (MCOs) (15). In addition to multicopper ferroxidases, the MCO family of enzymes includes laccases, which oxidize a broad range of substrates (including diphenols); ascorbate oxidases, which oxidize ascorbate; and a few less common oxidoreductases (16). All insect genomes analyzed to date contain at least two MCO genes, MCO1 and MCO2, and some insect species have more: for example, the Drosophila melanogaster genome encodes four (15). MCO1 orthologues have unknown enzymatic and physiological functions (15). MCO2 orthologues are laccases that participate in pigmentation and sclerotization of newly synthesized cuticle (15). D. melanogaster MCO3, which has no known orthologues outside of Drosophilidae, is predicted to be a ferroxidase because it contains a putative iron-binding residue, and loss of function leads to an increase in iron accumulation in the midgut; however, evidence of ferroxidase activity and information about when and where DmMCO3 is expressed are not available (8). Because MCO2 orthologues are known to be laccases, and most other insect MCOs, including DmMCO3, are found in only a subset of insect species, we decided to focus on MCO1.

The goal of this study was to determine whether MCO1 functions as a ferroxidase. Previous studies of gene expression in D. melanogaster found that MCO1 mRNA is most abundant in late stage embryos, feeding stage larvae, and adults, and that MCO1 is expressed predominantly in the digestive system (midgut, hindgut, and crop) and Malpighian tubules (17, 18). These expression profiles are similar to those of MCO1 orthologues in Anopheles gambiae and Manduca sexta (19–21). To establish whether MCO1 functions as a ferroxidase in D. melanogaster, we purified recombinant MCO1 and analyzed its enzymatic activity, performed immunohistochemistry to determine the location of MCO1 within the insect, and evaluated the loss-of-function phenotypes generated by RNAi-mediated knockdown.

Results and Discussion

MCO1 Orthologues Have Putative Iron-Binding Residues.

A typical MCO consists of three cupredoxin-like domains, and residues in domains II and III form the substrate binding pocket (22, 23). Multicopper ferroxidases have conserved acidic residues that bind to ferrous iron—for example, the substrate binding site of a yeast ferroxidase, Fet3p, has three acidic residues that coordinate iron binding: Glu185, Asp283 and Asp409 (24, 25). To identify possible iron binding residues in MCO1, we aligned the D. melanogaster MCO1 sequence with the Fet3p sequence. Two acidic residues in the MCO1 cupredoxin-like domain II, Asp380 and Glu552, align with Glu185 and Asp283 in Fet3p (Figs. S1 and S2). We verified that these two acidic residues are conserved in MCO1 orthologues from other insect species (Fig. S1). Given the high degree of conservation of these residues in MCO1 orthologues and given that Asp409 is not required for the ferroxidase activity of Fet3p (25), we hypothesized that MCO1 functions as a ferroxidase.

MCO1 Has Ferroxidase Activity.

To allow us to directly investigate the enzymatic activity of MCO1, we expressed a recombinant form of the enzyme in cultured insect cells. The MCO1 sequence contains a predicted signal peptide, two von Willebrand factor domains, a cysteine-rich region, three cupredoxin-like domains, and a putative carboxyl-terminal transmembrane segment (Fig. S2). Expression and purification of the full-length, WT protein was difficult as a result of low expression and proteolytic cleavage. To circumvent these problems, we expressed recombinant MCO1 without the von Willebrand factor domains and transmembrane segment, and we mutated Arg454 to alanine to reduce proteolytic cleavage (Fig. S2). The secreted enzyme was purified from serum-free medium with the use of lectin affinity, anion exchange, and size-exclusion chromatography. The identity of the purified enzyme was confirmed by immunoblot analysis, and adequate purity was verified by SDS/PAGE followed by Coomassie staining (Fig. 1 and Fig. S2).

Fig. 1.

Ferroxidase activity of MCO1. MCO1 and two mosquito laccases, MCO2A and MCO3, were analyzed by triplicate native gels. (A) Coomassie staining. (B) Activity assay using p-phenylenediamine, a substrate of laccases and multicopper ferroxidases (28). (C) Assay for ferroxidase activity using ferrous ammonium sulfate as the substrate. Note that all enzymes oxidized p-phenylenediamine, but only MCO1 had detectable ferroxidase activity.

Multicopper ferroxidases oxidize laccase substrates in addition to ferrous iron, but usually at a much lower catalytic efficiency (24). To analyze the substrate preferences of MCO1, we compared the activity of recombinant MCO1 with that of two mosquito laccases, A. gambiae MCO2A and MCO3 (26, 27). We determined kinetic constants associated with four laccase substrates, including a p-diphenol (hydroquinone) and three o-diphenols that are natural substrates of insect laccases (dopamine, N-acetyldopamine, and N-β-alanyldopamine) (26). We found that MCO1 had detectable laccase activity; however, the catalytic efficiencies (k_cat/K_m) were much lower than those of MCO2A and MCO3 (Table 1). These results suggest that MCO1 does not function as a laccase in vivo. To determine whether MCO1 has ferroxidase activity, an in-gel assay was used. Of the three enzymes, only MCO1 had detectable ferroxidase activity (Fig. 1). These results suggest that MCO1 functions as a ferroxidase.

Table 1.

Catalytic efficiencies of MCO1 and two insect laccases

Substrate	MCO1*	MCO2A*,†	MCO3*,†
Hydroquinone	6	114	302
Dopamine	2	96	171
NADA	Activity ND	307	155
NBAD	12	219	323

*Catalytic efficiencies (k_cat/K_m) are in min⁻¹ mM⁻¹.

^†The catalytic efficiencies of MCO2A and MCO3 were reported previously (26, 27).

NADA N-acetyldopamine, NBAD, N-β-alanyldopamine; ND, not detected.

MCO1 Is Present on Basal Surface of Digestive System and Malpighian Tubules.

To verify expression of MCO1 in the midgut and Malpighian tubules and to determine whether MCO1 expression is affected by the concentration of iron in the flies’ diet, we used quantitative RT-PCR to analyze MCO1 transcript abundance in whole flies, midguts, and Malpighian tubules. Flies were fed regular food or food supplemented with 0.5 mM or 10 mM iron. We found that iron supplementation had no detectable effect on MCO1 expression in whole flies, midguts, or Malpighian tubules (Fig. S3). Based on its amino acid sequence, we would expect MCO1 to enter the secretory pathway in cells of the midgut and Malpighian tubules and to remain attached to the cell membrane by a carboxyl-terminal transmembrane segment. Immunostaining supports this prediction. MCO1 immunoreactivity was along the basal surface of the midgut and Malpighian tubules (Fig. 2 and Fig. S4). Immunostaining was not restricted to particular regions of the digestive system or Malpighian tubules (Fig. S4). The basal location of MCO1 immunoreactivity suggests that MCO1 is extracellular, and thus positioned to oxidize ferrous iron in the hemolymph.

Fig. 2.

Immunolocalization of MCO1. Cryosections of larval (A) and adult (B) midguts and unsectioned larval (C) and adult (D) Malpighian tubules were immunostained with MCO1 antiserum. Anti-MCO1 antibodies were detected with Alexa Fluor 488-conjugated anti-rabbit IgG antibodies (green). Nuclei were stained with DAPI (blue). Immunoreactivity was detected along the basal surface of the epithelial cells of the midgut and Malpighian tubules. (Scale bars: 10 μm.)

Knockdown of MCO1 Is Correlated with Increased Longevity on High-Iron Food and Decreased Iron Accumulation.

To detect MCO1 loss-of-function phenotypes, we used RNAi-mediated knockdown. Strong knockdown was accomplished by crossing an actin-Gal4 (act-Gal4) driver line with a UAS-dsMCO1 responder line; weak knockdown was achieved with a daughterless-Gal4 (da-Gal4) driver. Negative controls were produced by crossing the driver with a line that is genetically similar to the RNAi line except for the lack of a dsMCO1 transgene. Knockdown was verified by quantitative RT-PCR (Fig. S5). We found that strong knockdown of MCO1 resulted in pupal lethality. In contrast, insects with a weak knockdown of MCO1 survived to adulthood, and they lived as long as negative controls (Fig. 3). The lethality associated with strong knockdown indicates that MCO1 is an essential gene.

Fig. 3.

Effect of MCO1 knockdown on longevity. Female (A) and male (B) flies were cultured on regular food or food supplemented with 10 mM ferric ammonium citrate (Fe). Longevity of MCO1 knockdown flies (KD) and control flies (indicated as “C”) was similar when the flies were fed normal food (gray and red curves). Flies cultured on food supplemented with 10 mM iron (blue and green curves) had decreased longevity compared with flies fed regular diet (P < 0.0001 for females and males), demonstrating that 10 mM iron was toxic to the flies. Most importantly, MCO1-knockdown flies lived longer than control flies when they were fed the high iron diet (green vs. blue curves; P < 0.0001 for females and males). Knockdown genotype was da-Gal4>UAS-dsMCO1a; and control genotype was da-Gal4>attP. Data are from at least four biological replicates. SEs are shown. Survival curves were compared by performing a log-rank (Mantel–Cox) test. (Note that, in A, the gray and red curves overlap.)

If MCO1 is involved in iron metabolism, we might expect knockdown to affect the longevity of flies cultured on food with a toxic concentration of iron. To test this hypothesis, we cultured da-Gal4>UAS-dsMCO1a and da-Gal4>attP (control) flies on food supplemented with 10 mM ferrous ammonium citrate, a concentration of iron that decreased longevity in control flies (Fig. 3). We found that MCO1-knockdown flies lived significantly longer than control flies on the iron-supplemented food (Fig. 3), indicating that knockdown of MCO1 protects the flies from iron toxicity.

Diminished iron accumulation in knockdown insects would provide an explanation for protection against iron toxicity. We used two methods to test the hypothesis that MCO1 knockdown leads to less iron accumulation: a histological method to detect iron in the midgut (29) and a FerroZine-based assay to estimate the iron content of whole insects (30). Both methods demonstrated a correlation between knockdown of MCO1 and a decrease in iron accumulation (Figs. 4 and 5 and Fig. S6). The decreased iron content in knockdown insects demonstrates that iron accumulation is influenced by MCO1 function.

Fig. 4.

Effect of MCO1 knockdown on iron accumulation in the midgut. Midguts of control (indicated as “C”) and knockdown (KD) genotypes were stained to detect iron. A and C show the iron cell region of a representative larval midgut from control (act-Gal4>attP) and knockdown (act-Gal4>UAS-dsMCO1a) genotypes. B and D show the iron cell region of a representative adult midgut from control (da-Gal4>attP) and knockdown (da-Gal4>UAS-dsMCO1a) genotypes. (Scale bars: 100 μm.) Knockdown was correlated with a reduction in the blue area of the midguts (E, larva; F, adult). Number of midguts analyzed were as follows: act-Gal4>attP, n = 21; act-Gal4>UAS-dsMCO1a, n = 23; da-Gal4>attP, n = 8; and da-Gal4>UAS-dsMCO1a, n = 8. Means ± SD are shown. Differences in areas were assessed by performing a Mann–Whitney test (E) or unpaired t test (F) (**P < 0.01 and ****P < 0.0001).

Fig. 5.

Effect of MCO1 knockdown on iron accumulation in whole insects. A FerroZine-based assay was used to determine the concentration of iron in whole larvae cultured on regular food (A) and adult females cultured on food supplemented with 10 mM ferric ammonium citrate (B). At both life stages, knockdown insects had less iron than controls. Larval genotypes were act-Gal4>attP (indicated as “C”) and act-Gal4>UAS-dsMCO1a (KD); adult genotypes were da-Gal4>attP (“C”) and da-Gal4>UAS-dsMCO1a (KD). Data are from three biological replicates. Means ± SD are shown. Differences in iron content were assessed by performing an unpaired t test (*P < 0.05 and **P < 0.01).

Possible Functions of MCO1.

The apparent absence of MCO1 on the apical surface of midgut epithelial cells argues against a role for MCO1 in the uptake of iron from the diet. The presence of MCO1 on the basal surface of midgut cells suggests a possible hephaestin-like function (i.e., a role in iron efflux from intestinal cells) (6). In support of this hypothesis is the observation that loss of MCO1 function is correlated with a decrease in whole body iron content. On the contrary, knockdown of MCO1 resulted in less midgut iron, not the overaccumulation observed in mammalian hephaestin mutants. In addition, as hephaestin functions together with an iron permease, whereas no insect iron permease has been identified, if MCO1 contributes to iron efflux from midgut cells, its mechanism is likely to differ from that of hephaestin. One explanation for the decrease in iron accumulation is an unidentified feedback mechanism. For example, knockdown of MCO1 may result in an excess of ferrous iron in the hemolymph, which may act as a signal to down-regulate iron uptake from the diet. If this explanation is correct, we might expect knockdown of MCO1 to lead to a down-regulation of DMT1 and a corresponding decrease in iron transport from the gut lumen to the midgut epithelial cells. As an initial test of this hypothesis, we used quantitative RT-PCR to analyze DMT1 expression and found that knockdown of MCO1 did result in lower DMT1 transcript abundance in the midgut (Fig. 6).

Fig. 6.

Effect of MCO1 knockdown on the expression of the iron importer DMT1. Quantitative RT-PCR was used to analyze DMT1 expression in the midguts of control and knockdown larvae (A) and adults (B). Knockdown of MCO1 resulted in a decrease in DMT1 transcript abundance in the midgut. Larval genotypes were act-Gal4>attP (indicated as “C”) and act-Gal4>UAS-dsMCO1a (KD); adult genotypes were da-Gal4>attP (indicated as “C”) and da-Gal4>UAS-dsMCO1a (KD). Data were normalized to rp49 expression. Data are from three biological replicates. Means ± SD are shown. Differences in expression were assessed by performing an unpaired t test (*P < 0.05 and **P < 0.01).

As mentioned earlier, the basal location of MCO1 suggests that the enzyme is positioned to oxidize ferrous iron to ferric iron in the hemolymph. The resulting ferric iron could be loaded onto transferrin and transported to iron-deficient cells for uptake, as is the case in mammals (31). Alternatively, oxidation of ferrous iron followed by transferrin loading may play a protective role by limiting toxic radical formation, which is generated by ferrous but not ferric iron (1), or, it may have an immune function, given that ferric-transferrin is less likely than free iron to serve as a source of iron for pathogens (32). The possibility of an immune function is supported by up-regulation of MCO1 in D. melanogaster (33) and A. gambiae (19) in response to infection. Finally, the oxidation of ferrous iron by MCO1 may supply ferric iron to melanotransferrin, which is required for septate junction formation during epithelial maturation in D. melanogaster (34). A developmental role for MCO1 would provide an explanation for the pupal lethality caused by strong MCO1 knockdown.

Conclusion.

The main goal of this study was to determine whether MCO1 functions as a ferroxidase. Several lines of evidence indicate that it does: presence of putative iron binding residues in cupredoxin-like domain II, ferroxidase activity of recombinant MCO1, increased longevity of knockdown flies (compared with control flies) when cultured on high-iron food, and decreased iron accumulation in knockdown insects. In addition, we determined that MCO1 is located on the basal surfaces of the digestive system and Malpighian tubules. We propose the following hypothetical model: MCO1 oxidizes ferrous iron in the hemolymph, and the resulting ferric iron is bound by transferrin or melanotransferrin, leading to iron storage, iron withholding from pathogens, regulation of oxidative stress, and/or epithelial maturation. Given that MCO1 orthologues have been identified in all insect genomes analyzed to date, our discovery that DmMCO1 is a ferroxidase should lead to a better understanding of iron metabolism in a wide range of insect species, including blood-feeding insects such as mosquitoes, which ingest a large amount of iron with each blood meal. Now that many of the components of the iron metabolism pathways in insects and mammals have been identified and studied, a picture of two distinct systems for metabolizing iron has emerged. Although the pathways in insects and mammals have many of the same elements (e.g., DMT1, ferritins, transferrins, and multicopper ferroxidases), some components that are critical in mammals, such as ferroportin and transferrin receptor, may not exist in insects, and even some of the homologous proteins, including multicopper ferroxidases, seem to have divergent functions. It is interesting to consider that insects (with their open circulatory system) and mammals (with their need for large amounts of iron-bound hemoglobin) may have evolved to metabolize iron differently.

Materials and Methods

Sequence Analyses.

Sequence alignments and analyses were performed as described previously (27). GenBank IDs for each sequence analyzed is listed in Table S1.

Recombinant Protein Expression.

A full-length MCO1 cDNA was obtained from the Drosophila Genomics Resource Center. This clone (RE34633) had a Gly331Ser mutation, which was reversed with site-directed mutagenesis by using the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene). Expression and purification of the full-length, WT protein was difficult as a result of low expression and proteolytic cleavage. To circumvent these problems, we generated a truncated cDNA that encoded only the cysteine-rich region and cupredoxin-like domains (Fig. S2), cloned it in-frame with the D. melanogaster Bip signal sequence from the pMT/Bip/V5/His vector (Invitrogen), and used site-directed mutagenesis to create an Arg454Ala mutant that was less susceptible to proteolytic cleavage. This cDNA was cloned into the pOET3 vector and the flashBAC system (Oxford Expression Technologies) was used to generate a recombinant baculovirus. Expression in Sf9 cells and purification by concanavalin-A–Sepharose and Q-Sepharose were done as described previously (27). Additional purification was accomplished with the use of a Superdex 200 HiLoad 16/60 column (GE Healthcare Life Sciences). The concentration of purified recombinant MCO1 was estimated as described previously (27). The yield was ∼0.5 mg MCO1 per liter of cell culture.

Activity Assays.

Laccase activity assays, determination of pH optima, and calculation of kinetic constants were performed as described previously (26). Briefly, reactions were done by mixing 0.5 μg MCO1 with substrate in a total volume of 200 μL and detecting product formation with a microplate spectrophotometer by observing the change in absorbance over time. All assays were done in triplicate. MCO1 had the highest activity at pH 6.0 to 8.0; therefore, pH 7.0 was used for determining kinetic constants. Ferroxidase activity was assayed with an in-gel method (27). Ten micrograms of recombinant MCO1, A. gambiae MCO2A (26), and A. gambiae MCO3 (27) were subjected to native PAGE in triplicate; one gel was stained with Coomassie blue, another was incubated in 9.2 mM p-phenylenediamine (Sigma) to detect laccase activity, and a third gel was incubated in 0.2 mM ferrous ammonium sulfate, followed by a brief incubation with 15 mM FerroZine (Acros Organics) to detect ferroxidase activity.

Drosophila Culture and Genetics.

Flies were cultured at 25 °C on K12 High Efficiency diet (USBiological) unless otherwise noted. Two ubiquitous Gal4 driver stocks were obtained from the Bloomington Drosophila Stock Center at Indiana University: y¹ w; act-Gal4/ CyO (no. 25374), and w¹¹¹⁸; da-Gal4 (no. 8641). A gut/salivary gland-specific Gal4 driver line (mg+sg-Gal4) was obtained from the Drosophila Genetics Resource Center in Kyoto: w*; P{w[+mW.hs]=GawB}NP3084 (no. 113–094). Two RNAi lines were obtained from the Vienna Drosophila RNAi Center (VDRC): w¹¹¹⁸; UAS-dsMCO1a/ CyO (no. 108677 from the KK library), and w¹¹¹⁸; UAS-dsMCO1b (no. 15602 from the GD library). The two RNAi lines have insertions at different chromosomal locations, target different regions of the MCO1 mRNA, and are predicted to have no off-targets. UAS-dsMCO1a is inserted into an attP site on chromosome 2; therefore, the “empty” attP line (VDRC no. 60100) was used as a negative control. UAS-dsMCO1b was inserted at random via a P-element–mediated event, and w¹¹¹⁸ (VDRC no. 60000) was used as the negative control. For experiments that required the genotypic identification of larval progeny from heterozygous parents, a GFP-marked CyO balancer chromosome was used (Bloomington Drosophila Stock Center no. 4523).

RNA Isolation and Quantitative RT-PCR.

Total RNA was isolated by using the Tri reagent method (Sigma), and genomic DNA was removed with DNaseI. The quality of DNase-treated total RNA was verified with a Bioanalyzer (Agilent). Pools of cDNA were synthesized with the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen) with oligo(dT) primers. Real-time PCR reactions were monitored on an iCycler (Bio-Rad) by means of SYBR Green (Bio-Rad) dye. Expression data were normalized to ribosomal protein 49 (rp49) or gapdh expression. Primers are listed in Table S2. Pairwise differences in expression were assessed by performing an unpaired t test, and multiple comparisons were analyzed by one-way ANOVA followed by Dunnett test (or Tukey test for data sets of n = 2). All statistical analyses were performed with GraphPad Prism software.

Production of Polyclonal Antiserum and Immunoblot Analysis.

Polyclonal antiserum was generated against residues Asp346 to Asp691 of mature MCO1 (Fig. S2) by using methods described previously (27) with minor modifications. Excised gel slices containing the recombinant protein were sent to Open Biosystems for the production of polyclonal antiserum in a rabbit. To remove antibodies that might react with conserved epitopes in other MCOs, the antiserum was incubated with purified A. gambiae MCO4 (19), and the unbound fraction was retained.

Immunohistochemistry.

To detect MCO1 in midguts, immunostaining of cryosections was performed as described previously (27) with minor modifications. Briefly, dissected guts from wandering larvae or adult females were fixed in 4% (wt/vol) paraformaldehyde in PBS solution for 1 h, sections were incubated for 2 h with partially purified MCO1 antiserum or preimmune serum, and Alexa Fluor 488-conjugated goat anti-rabbit IgG (Invitrogen) was used as the secondary antibody. To detect MCO1 in whole tissues, the digestive system plus Malpighian tubules were fixed with 4% (wt/vol) paraformaldehyde in PBS solution for 2 h, permeabilized in PBS solution plus 1.5% (vol/vol) Triton X-100 (PBS–Triton) for 5 min, and blocked for 1 h in PBS solution plus 1% Triton X-100 and 5% (wt/vol) BSA. Tissues were incubated with partially purified MCO1 antiserum (1:200–1:1,000) or preimmune serum for 24 h at 4 °C, washed with PBS-Triton buffer for 24 h at 4 °C, incubated in Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:500; Invitrogen) for 24 h at 4 °C, washed with PBS solution for 24 h at 4 °C, stained with DAPI for 15 min, and washed with PBS solution overnight at 4 °C. Sections and tissues were analyzed with a Zeiss LSM 510 META laser scanning confocal microscope by using excitation wavelengths of 405 and 488 nm and a 10× Plan-Neofluar objective with a 0.3 NA or a 40× Plan-Neofluar oil objective with a 1.3 NA. The specificity of MCO1 immunoreactivity was verified by immunostaining tissues from MCO1-knockdown insects (Fig. S4).

Longevity Assay.

Female da-Gal4 flies were crossed with UAS-dsMCO1a/ CyO or attP males. Progeny were cultured on unsupplemented food. Within 1 d after eclosion, control (da-Gal4>attP) and MCO1 knockdown (da-Gal4>UAS-dsMCO1a) female and male flies were selected and raised on food supplemented with 0 or 10 mM ferric ammonium citrate. Twenty-five flies were placed in a vial with food. Flies were cultured at 25 °C and transferred to vials with fresh food every 2 to 3 d. Dead flies were counted every 24 h. Four to six biological replicates were done. Survival curves were compared by performing a log-rank (i.e., Mantel–Cox) test with the use of GraphPad Prism software.

Histological Staining of Iron in Midgut Cells.

For detecting iron in the iron-storing cells of the midgut (35), a Prussian blue staining method was used (29). Midguts from wandering larvae or female adults were dissected and fixed in 4% (wt/vol) formaldehyde, permeabilized with 1% Tween-20, incubated with 2% (wt/vol) K₄Fe(CN)₆ in 0.24 N HCl, and rinsed with water. Larvae were cultured on regular food; newly eclosed adults were transferred to food supplemented with 1 mM ferric ammonium citrate for 1 d (for the purpose of increasing the amount of iron in the midgut to reliably detectable levels) and then transferred to regular food for 24 h to clear the high iron food from the gut lumen. Images were taken with a Zeiss Axio Imager microscope with a 10× objective and AxioCam camera. At least eight midguts of each genotype were analyzed. AxioVision software was used to outline the blue region of each midgut and to calculate the areas. Differences in areas were assessed by performing a Mann–Whitney test (for data sets with unequal variance) or an unpaired t test using GraphPad Prism software.

FerroZine-Based Assay.

To estimate whole-body iron concentrations, we used a FerroZine-based assay that has been described previously (30). Heads were removed from adults to avoid potential interference from eye pigments. Adults were cultured for 5 d on food supplemented with 10 mM ferric ammonium citrate to mimic the conditions used in the longevity experiment and then transferred to regular food for 24 h to clear the high-iron food from the gut lumen. We homogenized 13 wandering larvae or 10 female adults in 135 μL of lysis buffer (20 mM Tris, 137 mM NaCl, 1% Triton X-100, 1% glycerol). Samples were centrifuged to remove insoluble material. For blank controls, we substituted lysis buffer for the soluble fraction. Protein concentration was determined with a bicinchoninic acid assay. Proteins were hydrolyzed with concentrated hydrochloric acid at 95 °C, and ferric iron was reduced to ferrous iron with ascorbate. FerroZine was added to the samples, and the pH was increased with the addition of ammonium acetate. Complex formation was detected by absorbance at 562 nm. The concentration of iron was calculated based on a molar extinction coefficient of the iron–FerroZine complex of 27,900 M⁻¹ cm⁻¹ (36). Three biological replicates were performed, and differences in iron concentration were assessed by performing an unpaired t test by using GraphPad Prism software.