Systematic functional characterization of putative zinc transport genes and identification of zinc toxicosis phenotypes in Drosophila melanogaster

The heavy metal zinc is an essential component of the human diet. Adequate, bioavailable zinc is crucial for correct functioning of the immune response, protein and DNA synthesis, retinal development, liver function, blood clotting, metalloenzyme function and olfaction (Berg and Shi, 1996; Tubek et al., 2008). In biological systems, zinc, a group IIB transition metal, is redox inert and remains in the Zn²⁺ ionic state (Maret, 2010). Sequence data mining projects have estimated that up to 10% of human proteins bind zinc (Andreini et al., 2006). The ability of Zn²⁺ ions to form structure-stabilizing cross-links makes the metal an ideal structural cofactor for small proteins and the flexible coordination geometry of zinc ions allows zinc binding proteins to shift conformations quickly in order to carry out complex biological reactions (Berg and Shi, 1996; Brown et al., 2002; Wellinghausen et al., 1997). As such, zinc is an essential catalytic cofactor of some key metabolic proteins such as carbonic anhydrase and alkaline phosphatase (Vallee and Auld, 1990a; Vallee and Auld, 1990b). Symptoms of zinc deficiency in humans include the impairment of eyesight, taste (hypogeusia) and smell (anosmia), a decreased immune response, skin lesions, hyperkeratinization of the epidermis and testicular atrophy as well as decreased cognitive and motor functions in children (Brown et al., 2002; Hambidge, 2000; Tubek, 2006; Tubek, 2007).

Symptoms of zinc toxicosis include manifestation of the immune disorder neutropenia, a haematological disorder characterized by an abnormally low number of neutrophil granulocytes and anaemia (Willis et al., 2005). In addition, inhalation of zinc salts has been shown to cause metal fume fever, which manifests as an acute respiratory tract infection and can lead to chronic asthma (Fuortes and Schenck, 2000). The link between zinc homeostasis and widespread or debilitating diseases such as diabetes mellitus, Alzheimer's disease, asthma, acrodermatitis enteropathica and Crohn's disease has come under intense scrutiny in recent years (Chen et al., 2000; Frederickson et al., 2005; Friedlich et al., 2004; Gyulkhandanyan et al., 2008; Smidt et al., 2009; Sturniolo et al., 2001).

At the cellular level, zinc homeostasis is achieved through the interplay of proteins involved in zinc uptake and export. These processes have been shown to be regulated by two gene families that are common across many phyla; the Zrt/IRT-like (Zip) gene family (solute carrier family 39, SLC39A) and the cation diffusion facilitator (CDF) family (SLC30A) – commonly described as ZnT genes (Kambe et al., 2008). In vivo studies have produced a large amount of data regarding zinc transporter function (Falcón-Pérez, Dell'Angelica, 2007; Mao et al., 2007; Murakami and Hirano, 2008; Ohana et al., 2009). The lethal milk (lm) mutant mice carry a loss of function mutation in the ZnT4 gene which results in reduced zinc secretion into the milk produced in epithelial mammary gland cells (Dufner-Beattie et al., 2006). As a result, homozygous lm/lm pups suckling from lm/lm dams do not survive past the first 2 weeks of neonatal development (Murgia et al., 2006). Mice with impaired ZIP1, ZIP2 or ZIP3 function display a decreased ability to absorb zinc from the diet, particularly during pregnancy when zinc absorption is normally increased (Dufner-Beattie et al., 2005; Dufner-Beattie et al., 2006). The ZnT8 knockout (KO) mouse model, which displays decreased insulin secretion, has provided insights into the endogenous function of ZnT8 as an importer of zinc into insulin-containing granules in pancreatic β-cell islets (Hardy et al., 2011). Additionally, ZnT5 KO mice have defects in mast cell delayed-type allergic reactions, illustrating the potentially important role of zinc transport proteins in immune response pathways (Nishida et al., 2009).

KO mouse models have also provided evidence for important signalling roles for some Zip genes, suggesting that there is an even broader range of functions carried out by zinc transport proteins than previously thought. Zip13 KO mice display impaired connective tissue development – this transporter has been suggested to play a role in localizing SMAD to the nucleus (Fukada et al., 2008) – whilst Zip14 KO mice display impaired gluconeogenesis, presumably due to the disruption of endogenous ZIP14 function as a regulator of G protein-coupled receptor signalling (Hojyo et al., 2011). Furthermore, it has recently been shown that proteins other than ZIP and ZnT may be able to transport zinc. Mammalian TRPM3 has been demonstrated to mediate the influx of zinc into pancreatic β-cells (Wagner et al., 2010) and loss of function of the Drosophila TRPM channel gene, dTRPM, resulted in larval growth inhibition and cell autonomous growth defects associated with defects in mitochondrial structure and function (Georgiev et al., 2010). All of the dTRPM mutant phenotypes could be phenocopied by zinc depletion and partially rescued by zinc supplementation (Georgiev et al., 2010), indicating a key role for dTRPM in regulating zinc homeostasis in a specific cellular organelle, although the cellular localization of this channel is not yet known.

In addition to transport and channel genes, several other gene classes are known to play an integral role in regulating cellular and systemic zinc levels. The metallothionein (mtn) genes encode small metallophilic proteins whose production is stimulated upon an increase in cellular levels of metals such as zinc, copper or cadmium. Mtns are thought to sequester and detoxify these metals in the cytosol while also providing a storage option. Genetic ablation of all four Mtn genes (A–D) in the vinegar fly Drosophila results in viable flies that are exquisitely sensitive to elevated dietary metal levels (Egli et al., 2006). Transcriptional induction of the Mtn genes requires the activity of metal transcription factor 1 (Mtf-1), which is also essential for viability under metal stress in both Drosophila and mouse (Egli et al., 2003; Wang et al., 2004b).

In Drosophila, disruption of metal transport gene expression has previously been shown to result in severe and easily recognizable phenotypes. Copper-related phenotypes that result from over-expression or suppression of important copper transport genes, such as DmATP7 and Ctr1a, commonly result in abnormal pigment phenotypes, neuronal defects and impaired survival when larvae and adults are reared on modulated copper diets (Binks et al., 2010; Burke et al., 2008; Norgate et al., 2006). In addition to the previously characterized Drosophila zinc transport genes ZnT35c (Yepiskoposyan et al., 2006), dZnT63C (Wang et al., 2009), fear of intimacy (foi) (Mathews et al., 2006; Mathews et al., 2005) and Catchecholamines up (Catsup) (Stathakis et al., 1999), nine more putative zinc transport genes were identified for analysis during this study with an additional four being recently identified by homology to mammalian proteins. Thus, with 17 predicted zinc transport genes compared with the 25 in mammals, the Drosophila system is clearly still complex yet has reduced scope for redundancy compared with higher organisms and, more importantly, provides rapid and flexible gene manipulation tools for sophisticated in vivo functional analyses.

Here, we used the UAS-Gal4 system to analyse Drosophila zinc transporter function in vivo. In particular, we focused on characterizing the functions of zinc transporter genes in two distinct tissue types, the developing eye and midline epithelial tissue. We generated tissue-specific zinc toxicosis situations after genetically modifying the expression levels of two zinc transporter genes in a combinatorial manner and are able modify these toxicosis phenotypes by controlling dietary zinc uptake during development.

The following fly stocks were used: w¹¹¹⁸ (BL3605, Bloomington Stock Centre, IN, USA). gmr-Gal4, P{GMR-GAL4.w⁻}2 (BL9146). pannier-Gal4, P{GawB}pnr^MD237 (BL3039). Tubulin-Gal4, P{tubP-GAL4}LL7/TM3, Sb¹ (BL5138). RNA interference (RNAi) lines (Dietzl et al., 2007) obtained from the Vienna Drosophila RNAi Centre (VDRC) include: v13311 (dZnT41F), v7461 (dZnT63C), v105145 (dZnT63C), v7688 (dZnT33D), v103398 (dZnT33D), v3836 (dZnT35C), v103263 (dZnT35C), v5390 (dZnT77C), v12132 (dZnT86D), v107388 (dZnT86D), v49329 (dZip88E), v106785 (dZip88E), v7338 (dZip42C.2), v3986 (dZip42C.1), v37358 (dZip89B), v10102 (fear of intimacy), v1362 (dZip99C) and v10095 (Catchecholamines up). CG numbers for zinc transport genes are listed in supplementary material Table S1. Over-expression lines generated for this study were: pUAST dZnT41F^flag, pUAST dZnT63C^flag, pUAST dZnT33DPB^flag, pUAST dZnT35C^flag, pUAST dZnT77C^flag, pUAST dZnT86D^flag, pUAST dZip88E^flag, pUAST dZip42C.2^flag, pUAST dZip42C.1^flag, dZip42C.1^eGFP, pUAST dZip89B^flag, pUAST fear of intimacy (FOI)^flag, pUAST dZip99C^flag and pUAST Catchecholamines up (Catsup)^flag.

Drosophila full-length open reading frames for all putative zinc transport genes were PCR amplified from cDNA extracted from w1118 third instar larvae using primer pairs listed in supplementary material Table S2. In each case the termination codon was omitted to allow C-terminal tagging with either a FLAG epitope or enhanced green fluorescent protein (eGFP tag). Amplified gene fragments were subcloned in-frame with FLAG and eGFP tags into the pUAST-attB vector. This construct was injected into PhiC31 attP 51C and 96E or 86Fb strains (provided by Konrad Basler, Institute of Molecular Life Sciences, University of Zurich). Microinjections utilized an Eppendorf Femtojet apparatus with Femtotips II pre-pulled glass needles (Eppendorf, Hamburg, Germany). The C-terminus was chosen in each case so that expression of full-length cDNA in vivo could be confirmed by western blotting or immunohistochemical analysis. Addition of C-terminal epitope tags or fluorescent protein can interfere with the function of the tagged protein; therefore, caution must be taken when interpreting over-expression results using such transgenic constructs.

The online source MUSCLE (MUltiple Sequence Comparison by Log-Expectation) was used to align zinc transporter amino acid sequences with default settings. Highly divergent alignments were manually edited using BioEdit Sequence Alignment Editor V.7.0.9.0. Phylogenies were estimated using PhyML 3.0 by maximum likelihood. Phylogenetic trees were annotated using FigTree v1.3.1. Similarity and identity values were calculated using EMBOSS Matcher (Bill Pearson's lalign application, version 2.0u4) (Guindon et al., 2010).

All Drosophila strains and crosses were maintained on standard medium at 25°C unless stated otherwise. Experiments were performed under the same conditions with zinc- or chelator-supplemented media as described. Standard medium was supplemented with 3–8 mmol l⁻¹ zinc chloride (ZnCl₂; Sigma-Aldrich, St Louis, MO, USA) to make zinc-supplemented medium, or with N,N,N′,N′-tetrakis (2-pyridylmethyl)-ethylenedidiamine (TPEN; Sigma-Aldrich) or histidine (Ralph et al., 2010) to make zinc-deficient medium. Drosophila embryos were transferred between stages 7 and 16 onto medium supplemented with aqueous prediluted ZnCl₂ or TPEN. Feeding experiments were density controlled with 100 embryos transferred per vial in triplicate.

Adult flies were partially dissected then mounted directly onto Plasticine and monitored with a Leica MZ6 stereomicroscope. All images were recorded with a Leica DC300 digital camera using Leica Application Suite (LAS) software. Salivary glands used for localization analyses were freshly dissected in 0.1% phosphate-buffered saline (PBS)-Tween 20 and mounted onto glass slides in 70% glycerol. Fluorescence was detected using a Nikon C1 Upright microscope and an oil-emersion objective lens (×40), using excitation energies of 405, 488 and 561 nm.

Wandering third instar larvae were dissected in cold PBS then fixed in 4% paraformaldehyde in PBS for 30 min. Fixed tissues were washed in PBS then rinsed briefly in H₂O prior to two final 2 min washes in 0.1 mol l⁻¹ ammonium acetate. Imaginal discs were transferred to a small drop of ammonium acetate on a silicon nitride window. Remaining liquid was allowed to air dry and tissues were left to air dry overnight at room temperature. Silicon nitride windows (product code SiRN-7.5-200-3.0-500; Silson, Northampton, UK) were used: frame size 7.5×7.5 mm; frame thickness 200 m; membrane size 3.0×3.0 mm; membrane thickness 500 nm.

X-ray fluorescence (XRF) microscopy was carried out as per a previously published protocol (Lye et al., 2011), with the following changes: excitation of K-shell fluorescence emission from first row transition metals was carried out using X-rays of 12.73 keV energy. Secondary X-ray fluorescence and scattering was collected using a 384-element Maia detector, which was oriented in back scatter geometry to the beam.

A series of Blastp searches identified 17 genes in the Drosophila melanogaster genome whose encoded proteins share amino acid sequence homology to mammalian Zip and ZnT genes (supplementary material Table S1). Four of these genes, catsup, foi, dZnT63C (also known as dZnT1) and dZnT35C have previously been characterized and the last three have been confirmed as functional zinc transporters. DZnT35C and DZnT63C have been shown to transport zinc through cells lining the lumen of the Malphigian tubules and midgut, respectively (Wang et al., 2009; Yepiskoposyan et al., 2006), whilst FOI has been shown to take up zinc through the outer plasma membrane in yeast and mammalian cell culture assays (Mathews et al., 2005). Note that while we have retained the long-held names for Catsup and foi, for the remaining 15 genes we have adopted the nomenclature of Yepiskoposyan (Yepiskoposyan et al., 2006), calling each gene dZnT or dZip followed by the gene's cytological location. We have avoided assigning a number to each gene to avoid the risk of assuming orthology to mammalian genes that may not prove accurate.

Our phylogenetic analysis shows that individual putative Drosophila ZnT and ZIP transporters often bear greater sequence similarity to related human zinc transporters than to each other (supplementary material Figs S1, S2). Drosophila and human ZnTs cluster into four distinct clades containing: (1) hZnT2–4, hZnT8, dZnT35C, dZnT33D and dZnT41F; (2) hZnT5–7 and dZnT86D; (3) hZnT1, hZnT10, dZnT63C and dZnT77C; and (4) hZnT9 and dZnT49B. The Drosophila and human ZIPs cluster into five distinct clades containing: (1) hZIP1–3, dZip42C.1, dZip42C.2, dZip89B and dZip88E; (2) hZIP9 and dZip102B; (3) hZIP5–6, hZIP10, FOI and dZip71B; (4) hZIP7, hZIP13, dZip99C and CATSUP; and (5) hZIP11 and dZip48C.

It has previously been shown that dZnT63C is required for viability, particularly under conditions of zinc stress (Wang at al., 2009). We therefore hypothesized that viability would be challenged by the ubiquitous suppression or ectopic expression of zinc transport genes that play major roles in zinc homeostasis and have little or no functional redundancy with alternative transporters. Therefore a small-scale screen was carried out, which utilized over-expression lines and suppression lines (VDRC RNAi lines) for 13 of the 17 putative zinc transport genes in Drosophila. These lines were initially crossed to tubulin-Gal4, which drives expression ubiquitously throughout the larval and adult stages, in order to identify zinc transport genes that are important for viability in Drosophila.

At least six zinc transport genes appear to be important for viability (Table 1), although possible off-target effects cannot be discounted without rescue experiments with RNAi-resistant over-expression constructs. foi has previously been shown to play an important role during development (Van Doren et al., 2003). Suppression of foi using tubulin-Gal4 resulted in lethality at the early larval stage. Ubiquitous suppression of Catsup, dZnT33D and dZip42C.2 or enhanced expression of dZip42C.1 also resulted in lethality during the early larval stages. Ubiquitous suppression of dZip42C.1 generated a phenotype whereby adult flies display moderate to severe melanotic defects (J.C.L., unpublished) and flies carrying two copies of the dZip42C.1^RNAi suppression construct under tubulin-Gal4 control did not advance beyond the L1 larval stage. Suppression of dZnT86D resulted in an extended larval second instar period (up to several weeks) and an inability to progress through the pupal stage of development, resulting in eventual lethality at the wandering third instar or early pupal stage. These larvae displayed distinct reductions in body size in comparison to wild-type larvae and pupae (J.C.L., unpublished). Similar delays in larval development due to disrupted zinc homeostasis by suppression of dZnT63C (Wang et al., 2009) or mutation of the ion channel gene dTRPM (Georgiev et al., 2010) have previously been reported, with melanotic accumulations attributed to prolonged larval growth phases rather than a specific disruption to zinc levels (Georgiev et al., 2010).

This suppression/over-expression screen was repeated using two different Gal4 driver lines. gmr-Gal4 drives expression in all cells of the eye disc posterior to the morphogenetic furrow, as well as the larval Bolwigs organ and the corpus cardiaca neurons of the larval ring gland (J.C.L., unpublished). The pannier-Gal4 expression domain encompasses the dorsal midline epidermal cells from the occipital head region to the end of the abdomen but excludes the terminalia. It is also expressed in the dorsal region of the larval wing disc, which eventually will become the notum of the adult fly. The gmr-Gal4 driver commonly produces morphological phenotypes whilst still retaining larval and adult viability. The pannier-Gal4 driver has been shown to present phenotypes ranging from abnormalities in pigment, bristle and cuticular arrangements to a loss of viability (Mummery-Widmer et al., 2009). The results from these experiments are summarized in Table 1. Whilst suppression of dZip42C.1 produced no phenotypes, enhanced expression of dZip42C.1 using pannier-Gal4 resulted in a lethal phenotype but had no effect on eye development using gmr-Gal4 under normal conditions.

We attempted to gauge the sensitivity of eye tissue to dietary zinc in flies over-expressing gmr>dZip42C.1^flag. The effect of raising these flies on zinc-supplemented diets was noticeable. Severe reductions in total eye tissue, as well as disorganized ommatidial and sensory bristle arrangements were observed in a zinc dose-dependent manner (Fig. 1). When raised on 3 mmol l⁻¹ zinc-supplemented media, gmr>dZip42C.1^flag adults displayed a mild rough eye phenotype (Fig. 1F) compared with untreated media (NF, Fig. 1E). gmr>dZip42C.1^flag flies raised on 5 mmol l⁻¹ zinc-supplemented media displayed a moderate rough eye (Fig. 1G) whilst development on 8 mmol l⁻¹ zinc-supplemented media resulted in highly irregular adult eye morphology and an obvious decrease in total eye tissue (Fig. 1H). At all zinc doses, wild-type control eyes were unaffected (Fig. 1A–D). Because of sequence homology between dZip42C.1 and hZIP1, we predict that this protein is involved in cellular zinc uptake and that enhanced expression may lead to increased uptake and subsequent cellular toxicity when dietary zinc concentrations are increased.

We aimed to produce an obvious but non-lethal zinc toxicosis phenotype within the fly, which was morphologically evident under normal dietary conditions but could be modified by altering dietary zinc load. As our zinc feeding experiment results suggested that dZip42C.1 is a strong zinc transporter, we used dZip42C.1 enhanced expression as a baseline for achieving a modifiable zinc toxicosis phenotype. We hypothesized that the phenotype gained from an increased zinc supply due to dZip42C.1^flag over-expression could be emulated in vivo via co-expression of other Zip genes or suppression of ZnT genes. As pannier>dZip42C.1^flag produced a lethal phenotype, we used the attenuated dZip42C.1^eGFP line for this screen as it produced only a mild thoracic phenotype when over-expressed in the midline and did not compromise viability.

The screen yielded several interesting phenotypes, shown in Fig. 2 and summarized in Table 2. Suppression of dZnT63C coupled with over-expression of dZip42C.1 produced the strongest phenotype. In the eye, sensory bristles are visibly misaligned and the entire eye, particularly in the centre, appears highly de-pigmented when the gmr-Gal4 driver was used (Fig. 2C). When crossed to pannier-Gal4, this combination produced severe thoracic clefts, loss of cuticle pigmentation, thoracic bristle misalignment and loss of scutellum (Fig. 2H). The stronger dZip42C.1^flag line was also crossed to the dZnT63C^RNAi suppression line in the eye. The result was a pupal lethal phenotype and almost a complete lack of eye development under gmr-Gal4 control. Co-over-expression of the related Zip genes dZip89B or dZip42C.2 with dZip42C.1^eGFP also resulted in a mild abnormal eye and midline phenotype (J.C.L., unpublished), further suggesting that dZip42C.1 over-expression is promoting increases in cellular zinc uptake that can be boosted by co-expression of related zinc importers. Phenotypes caused by the combination of dZip42C.1^eGFP expression and dZnT63C^RNAi suppression in either the eye or the midline could be rescued completely by the additional suppression of dZip42C.1 (Fig. 2D,H) or over-expression of dZnT63C^flag (Fig. 2E,I). This indicates that the RNAi lines for both these genes are highly specific and unlikely to be exhibiting off-target effects and that the over-expression lines for both genes are indeed active despite the presence of the C-terminal FLAG epitope.

Ectopic expression and visualization of the dZIP42C.1^eGFP fusion protein verified that this line was expressing the dZip42C.1^eGFP transgene. This fusion protein localized to the inner apical (lumenal) membrane and some intracellular compartments of third instar larval salivary gland cells (Fig. 3A–C), consistent with the previously reported apical localization of mammalian ZIP1 (Wang et al., 2004a). Additionally, ectopic expression of a dZNT63C^eGFP fusion protein in the larval salivary gland indicated that this protein localized to the outer basolateral plasma membrane of gland cells (Fig. 3D–F) consistent with dZNT63C localization at the outer membrane of larval midgut enterocyte cells (Wang et al., 2009). Some dZNT63C was also observed at the apical membrane. In the light of a previous report that identified dZnT63C as a zinc transport gene (Wang et al., 2009), and our own dZip42C.1^eGFP/dZnT63C^RNAi phenotypes, we then combined the dZnT63C^RNAi suppression line with all Zip^flag over-expression lines in an attempt to produce another zinc toxicosis condition; however, no additional phenotypes were observed.

Verification of the zinc-specific nature of the eye and midline phenotypes was achieved by raising these crosses on zinc- or TPEN-supplemented media. Raising control w¹¹¹⁸ flies on media supplemented with 1–14 mmol l⁻¹ zinc chloride from the embryo stage does not affect morphology post-eclosion (J.C.L., unpublished). In contrast, expression of dZip42C.1^eGFP coupled with dZnT63C suppression under gmr-Gal4 control resulted in increasingly acute ommatidial degradation and loss of sensory bristles upon dietary zinc supplementation (Fig. 4A–D). Raising these flies on TPEN-supplemented media rescued the rough eye/loss of pigment phenotype in a dose-dependent manner (Fig. 4E–G). The therapeutic effects of dietary TPEN intake at 50 μmol l⁻¹ was minimal (Fig. 4E); however, flies raised on 100 μmol l⁻¹ TPEN-supplemented media displayed partial rescue of eye pigmentation (Fig. 4F). When these flies were raised on 150 μmol l⁻¹ TPEN-supplemented media, eye morphology and pigmentation appeared almost wild-type (Fig. 4G). The pupal lethal/loss of eye phenotype that resulted from gmr>dZip42C.1^flag/dZnT63C^RNAi was also rescued with dietary TPEN treatment as well as by treatment with 1 mmol l⁻¹ dietary histidine or 300 μ>mol l⁻¹ EDTA – two other zinc chelators (Fig. 4I–L). The rescued adults displayed eyes that were a stronger variation of the phenotype observed in gmr>dZip42C.1^eGFP/dZnT63C^RNAi flies.

Raising pannier>dZip42C.1^eGFP/dZnT63C^RNAi suppression flies on zinc-supplemented media resulted in an exacerbation of the previously observed phenotype (Fig. 5). When raised on zinc-supplemented diets, adult flies from this cross began to display the formation of large melanotic masses (Fig. 5B–D). These masses were observed primarily in the abdominal regions of both male and female adult flies and became more numerous when flies were raised on increasing concentrations of zinc (5 and 8 mmol l⁻¹). At 5 mmol l⁻¹ supplemented zinc, partially penetrant lethality at the pupal stage became evident. When flies were raised on 8 mmol l⁻¹ zinc-supplemented media, almost complete lethality was observed. Dissection of partially developed pupae from these cases revealed an extremely high occurrence of melanotic masses in pupal tissue and a complete absence of notal/scutellar structures in the pannier domain (Fig. 5D).

Decreased dietary zinc uptake resulted in a reversion of the abnormal thoracic phenotype to wild-type (Fig. 5E–H). This occurred in a TPEN dose-dependent manner. Development on 50 μmol l⁻¹ TPEN-supplemented food had a minimal effect on the phenotype in the form of partially corrected bristle alignment and occasional evidence of pigment production in the thoracic midline (Fig. 5E). However, raising these flies on 100 μmol l⁻¹ TPEN-supplemented food resulted in a reduced thoracic cleft, slight development of the scutellum and scutellar bristles, as well as almost wild-type thoracic bristle structure and alignment (Fig. 5F). These flies also displayed reduced pigment loss in the midline region of the thorax. A partial- to-full rescue of the abnormal zinc toxicosis phenotype was observed when these flies were raised on 150 μmol l⁻¹ TPEN supplemented food (Fig. 5G). Adults displayed an almost wild-type appearance, the only evidence of any phenotypic abnormality being a slight loss of pigment in the midline thoracic region and occasionally a reduction in scutellum size. Additionally, despite the severity of the pannier>dZip42C.1^flag/dZnT63C suppression phenotype, viability could be restored when flies were raised on 150 μmol l⁻¹ TPEN-supplemented media (Fig. 5I). These flies displayed only subtle thoracic pigmentation defects and morphologically normal sensory bristle patterns and scutellae.

The cleft phenotype caused by pannier>dZip42C.1^eGFP/dZnT63C^RNAi and the rough eye phenotype caused by the gmr>dZip42C.1^eGFP/dZnT63C^RNAi combination may be due to increased apoptosis within targeted tissues during larval–pupal development. To test this possibility, we inhibited apoptosis by crossing flies displaying zinc toxicosis phenotypes to a UAS p35 line. While expression of p35 alone had no phenotypic effect (Fig. 6A,D), we found that suppressing apoptosis in the gmr domain mildly exacerbated the original zinc toxicosis phenotype (Fig. 6B,C). Additionally, suppressing apoptosis in the pannier domain noticeably exacerbated the midline phenotype (Fig. 6E,F). This combination caused lethality at the late pupal stage. Dissection of these flies from pupal cases revealed an almost total abolition of thoracic tissue.

Expression of the MtnB gene in the fly midgut enterocytes is induced by increases in cellular zinc (and copper and cadmium) levels and a reporter construct driving enhanced yellow fluorescent protein (eYFP) under the control of the MtnB regulatory region (MtnB-eYFP) can be used as an in vivo reporter of zinc levels (Selvaraj et al., 2005). MtnB-eYFP displays moderate expression levels in specific sections of the midgut (Fig. 7A). This expression is significantly increased by increasing dietary zinc levels (Fig. 7C) and is suppressed almost to background fluorescence levels in larvae raised on TPEN food (Fig. 7B). To establish the effect of manipulation of zinc transport genes on enterocyte zinc levels, we drove transgene expression using the pan-midgut driver HR-Gal4 (Chung et al., 2007) in the presence of the MtnB-eYFP reporter. Significant increases in MtnB-eYFP expression were observed under the suppression of dZnT63C (Fig. 7D) and Catsup (Fig. 7E) and over-expression of dZip89B^flag (Fig. 7F), whereas decreases were observed with the suppression of dZip89B (Fig. 7G), suppression of dZnT33D (Fig. 7H) and over-expression of dZnT63C^flag (Fig. 7I). The dZnT63C and dZip89B results are consistent with genes involved in zinc efflux and uptake, respectively, whereas the Catsup and dZnT33D results indicate a role in intracellular zinc distribution. dZip89B^flag over-expression could even induce MtnB-eYFP expression in TPEN-fed larvae (Fig. 7J). Interestingly, midgut-specific expression of dZnT86D^flag, dZip42C.1^flag and dZip42C.2^flag all resulted in larval death.

The effect of the zinc toxicosis combination on zinc levels was harder to determine. The dZip42C.1^eGFP construct could not be used as the eGFP would interfere with the MtnB-eYFP reporter and the dZip42C.1^flag construct caused early lethality under HR-Gal4 control, both alone and in combination with dZnT63C suppression. However, continuous, high-level zinc chelation (150 μmol l⁻¹ TPEN) rescues both these genotypic combinations to small but viable larvae. HR-gal4>dZip42C.1^flag larvae displayed minimal MtnB-eYFP activity under these high-TPEN conditions (Fig. 7K) but the addition of dZnT63C suppression resulted in increased MtnB-eYFP signal (Fig. 7L) indicating that even in these low zinc-availability conditions, a section of the midgut is accumulating high zinc levels.

To further examine the effect of zinc transport gene manipulation on cellular zinc content, the spatial distribution of zinc was analysed using larval wing imaginal discs (Fig. 8) by Synchrotron XRF microscopy. The pannier-Gal4 driver expresses strongly in the dorsal-most region of the wing disc, making this an ideal tissue in which to observe changes in metal distribution when zinc transport gene expression is modified (Lye et al., 2011). In control wing discs, zinc was distributed ubiquitously throughout the entire wing disc, whilst copper was distributed primarily at the ventral margins (Fig. 8A). Calcium was widespread throughout the dorsal region of the disc. Over-expression of dZip42C.1^flag resulted in a reduction of copper in the ventral margins of the disc and redistribution of zinc and calcium to the dorsal pannier region (Fig. 8B). This trend was observed in three out of four wing discs. The effect of dZnT63C suppression in the pannier domain on zinc distribution appeared to be negligible in all five wing discs observed (Fig. 8C), suggesting there may be an ancillary efflux mechanism at play in response to suppression of dZnT63C. Interestingly, pannier>dZnT63C^RNAi wing discs did display a redistribution of copper to the pannier domain and a less polarized distribution of calcium in three wing discs. One out of three wing discs of flies expressing pannier>dZip42C.1^eGFP/dZnT63C^RNAi displayed similar zinc, copper and calcium distributions to those observed in pannier>dZnT42C.1^flag discs; however, copper was still observed along the ventral margins (Fig. 8D). Additional images for each genotype are shown in supplementary material Fig. S3.

While foi has previously been identified as an important factor during Drosophila development (Mathews et al., 2006) and Catsup has been implicated in pigment formation by acting as a negative regulator of tyrosine hydroxylase activity (Stathakis et al., 1999), very few putative Drosophila ZIP and ZnT transporters have so far been characterized in vivo, and focus on the potential zinc-transporting nature of these proteins has been minimal. However, work conducted by Gutiérrez and colleagues (Gutiérrez et al., 2010) does suggest that foi is mildly upregulated in zinc-stressed fumble mutants. Our findings suggest that dZip42C.1 is a strong Drosophila zinc uptake transporter. Additionally, our data reaffirm the previously proposed function of dZnT63C as an important zinc efflux protein. dZnT63C expression in enterocyte cells lining the midgut lumen is particularly important in the maintenance of correct zinc homeostasis (Wang et al., 2009). Ubiquitous suppression of dZnT63C resulted in decreased lifespan, presumably as a result of insufficient systemic supplies of zinc. Our data have added to the proposed function of dZnT63C by extending the known functional requirements of this gene to the eye and dorsal epithelial cells, as suppression of dZnT63C in these cells results in strong zinc toxicosis phenotypes when combined with dZip42C.1 over-expression.

Six of the 13 zinc transport genes characterized here appear to be essential for fly viability, as seen by the lethality caused by ubiquitous RNAi suppression, highlighting the broad requirement for zinc homeostasis in numerous tissues. Of the remaining seven genes, three demonstrated over-expression phenotypes indicating that tight regulation of their activity is needed for animal health. The lack of suppression phenotypes for these seven genes could be due to either redundancy between the Drosophila transporters or a conditional requirement for these genes under zinc deficiency or excess conditions. It must be noted that RNAi constructs rarely generate complete suppression of gene activity and therefore residual transcripts could also be responsible for the absence of phenotype. Conclusive loss-of-function analysis will require the generation of null alleles of each of these genes.

Our targeted genetic interaction results indicate that multiple zinc transport genes must be suppressed or enhanced in a combinatorial manner before target tissues are visibly affected by zinc toxicosis. This buffering effect may be a result of one or a combination of cellular mechanisms. Transcriptional up- or down-regulation of alternative zinc transport genes within the same tissue could offset potential zinc toxic effects, until a certain concentration of cytosolic zinc is reached. In particular, increased activity of ZnT proteins at vesicular zinc storage sites such as zincosomes may confer protective benefits. However, Drosophila expression studies have found little evidence that zinc transport genes are strongly up- or down-regulated in response to changes in environmental zinc availability. Indeed, Drosophila micro-array analyses have identified only two zinc transport genes that were marginally induced by feeding larvae 5 mmol l⁻¹ zinc (Yepiskoposyan et al., 2006).

Our findings suggest that a genetically based zinc toxicosis situation can be therapeutically treated or exacerbated by modifications to the larval diet. Variations in dietary zinc availability had noticeable effects on the zinc toxicosis phenotypes in both eye and midline tissue. Raising wild-type larvae on a diet supplemented with 5 mmol l⁻¹ ZnCl₂ has previously been shown to increase adult total body zinc content 2.4-fold (Gutiérrez et al., 2010). In the same experiment, dietary supplementation with 100 μmol l⁻¹ TPEN was shown to decrease total body zinc in adults by almost 60%. Studies by Yepiskoposyan and colleagues also showed a strong correlation between zinc feeding and significant increases in total body zinc (Yepiskoposyan et al., 2006). These results, coupled with our own, suggest that mechanisms for trafficking zinc from the diet, through enterocyte cells lining the midgut lumen and into the haemolymph are extremely efficient. The sudden rescue of the more severe pannier>dZip42C.1^flag phenotype on 150 μmol l⁻¹ TPEN-supplemented media, in contrast to the continued lethality observed when these flies are raised on 100 μmol l⁻¹ TPEN-supplemented media, further suggests that there is a sensitive threshold effect at play. This threshold effect is also evidenced by the wild-type phenotypes of flies over-expressing dZip42C.1^flag in contrast to the highly irregular phenotypes observed when dZnT63C is additionally suppressed.

The lethality observed when dZip42C.1^flag over-expression was combined with dZnT63C suppression under gmr-Gal4 control was unexpected as genetic ablation of eye tissue in gmr>reaper flies has previously been shown to have no effect on viability (Ditzel et al., 2008). However, we have also found gmr-Gal4 to drive expression in the corpus cardiacae neuroendocrine cells of the larval ring gland (J.C.L., unpublished), which regulate secretion of the eclosion hormone ecdysone. Lethality at the late pupal stage is therefore likely to be the result of disrupted hormone signalling in the ring gland rather than loss of eye tissue. While we did not observe lethality using this combination under control of a CC-Gal4 driver (J.C.L., unpublished), which expresses specifically in the corpus cardiacae cells (Kim and Rulifson, 2004), this might be attributable to lower expression levels compared with the gmr-Gal4 driver.

The zinc toxicosis phenotypes described here have strong potential as sensitive zinc sensors, providing a quick and effective in vivo model system for the further characterization of mammalian zinc transport genes. Additionally, treatment of the zinc toxicosis lines with various therapeutics could help to validate the chelating nature of these drugs. Our results suggest that Drosophila can be used as an effective model for investigating zinc homeostasis in vivo. The relative ease with which combinatorial genetic manipulations can be carried out in targeted tissues will allow a comprehensive analysis of the functional interactions between the various transporter proteins that control cellular zinc levels. In addition, Drosophila, and the zinc toxicosis lines in particular, may prove an effective model, which may facilitate the discovery of novel genes that regulate zinc homeostasis. Drosophila could also provide an exceptional model for testing therapeutic treatments for diseases that commonly display zinc dys-homeostasis. The implications of these findings to the treatment of zinc disorders such as acrodermatitis enteropathica, a disease characterized by decreased absorption of ingested zinc as a result of a null mutation in the human zinc transport protein ZIP4 (Dufner-Beattie et al., 2003), could be far reaching with respect to the further study of zinc-related diseases as well as to the development of therapeutic agents to control the symptoms of these diseases.

Drosophila stocks were provided by The Australian Drosophila Research Support Facility www.ozdros.com. All Drosophila RNAi lines were provided by the Vienna Drosophila RNAi Centre. This research was undertaken on the XFM beamline at the Australian Synchrotron, Victoria, Australia.