Cloning and Characterization of a Novel Mammalian Zinc Transporter, Zinc Transporter 5, Abundantly Expressed in Pancreatic β Cells

Intracellular homeostasis for zinc is achieved through the coordinate regulation of specific transporters engaged in zinc influx, efflux, and intracellular compartmentalization. We have identified a novel mammalian zinc transporter, zinc transporter 5 (ZnT-5), by virtue of its similarity to ZRC1, a zinc transporter ofSaccharomyces cerevisiae, a member of the cation diffusion facilitator family. Human ZnT-5 (hZnT-5) cDNA encodes a 765-amino acid protein with 15 predicted membrane-spanning domains. hZnT-5 was ubiquitously expressed in all tested human tissues and abundantly expressed in the pancreas. In the human pancreas, hZnT-5 was expressed abundantly in insulin-containing β cells that contain zinc at the highest level in the body. The hZnT-5 immunoreactivity was found to be associated with secretory granules by electron microscopy. The hZnT-5-derived zinc transport activity was detected using the Golgi-enriched vesicles prepared from hZnT-5-induced HeLa/hZnT-5 cells in which exogenous hZnT-5 expression is inducible by the Tet-on gene regulation system. This activity was dependent on time, temperature, and concentration and was saturable. Moreover, zinc at a high concentration (10 mm) inhibited the growth of yeast expressing hZnT-5. These results suggest that ZnT-5 plays an important role for transporting zinc into secretory granules in pancreatic β cells.

Zinc is a trace element that is indispensable for life, because it is a key structural component of a large number of proteins such as metalloenzymes and zinc-dependent transcriptional factors (1,2). In mammals, zinc is absorbed from the diet through intestinal epithelial cells and is transported into the blood, where most of it is bound by albumin and ␣ 2 -macroglobulin (3). The circulating zinc is incorporated into all types of cells, and the cytosolic zinc is taken up by intracellular organelles.
The kinetic studies and characterization of zinc transport using a variety of both animal cell systems and membrane vesicles have suggested the transporter-mediated movement of zinc (3). Recently, cDNAs of several zinc transporters have been molecularly cloned from mammals, and most of them have been assigned to two metal transporter families, the cation diffusion facilitator (CDF) 1 family that functions in zinc efflux from cytoplasm and the ZRT/IRT-related protein (ZIP) family that functions in zinc influx into cytoplasm (4 -7). CDF family members, ZnT-1 to -4, have characteristics of membrane topology such as six predicted membrane-spanning domains, a cytoplasmic histidine-rich loop between membrane-spanning domains IV and V, and both amino and carboxyl termini thought to reside intracellularly (5). The histidine-rich loop may construct a metal-binding site. In mammals, ZnT-1 is a probable zinc transporter responsible for zinc efflux across the plasma membranes (8). ZnT-2 is found on late endosomes, where it appears to facilitate zinc transport into its vesicle compartment (9,10). ZnT-3 is expressed in the brain (11) and is responsible for the accumulation of zinc in synaptic vesicles (12). ZnT-4 is expressed in intestinal epithelial cells (13) and mammary gland epithelia, where it is required for zinc transport into milk (14). Thus, ZnT-1 to -4 appear to be engaged in zinc efflux or organelle compartmentalization from cytoplasm.
The members of the ZIP family are predicted to have eight potential membrane-spanning domains and an intracellular loop with a histidine-rich region between membrane-spanning domains III and IV that may function as a metal-binding site as does the similar loop in CDF family members. Thus far, three transporters (human Zip1 to -3) were identified in mammals (15)(16)(17). hZip1 and hZip2 are thought to be involved in zinc influx across the plasma membranes, because forced expression of these transporters in mammalian cells increased their zinc uptake (16,17).
In addition to the biological functions in metalloenzymes and transcription factors, zinc plays unique roles in some specialized cells (18 -20). In hippocampal neurons and pancreatic ␤ cells, a large amount of zinc is accumulated in cytoplasmic vesicles/granules (11,21). Zinc in synaptic vesicles is co-Zinc Transporter Abundantly Expressed in Pancreatic ␤ Cells secreted with neurotransmitters upon excitation; zinc is thought to modulate synaptic transmission (18). In pancreatic ␤ cells, zinc binds with proinsulin in the secretory pathway to form zinc proinsulin hexamers, which are subsequently converted into zinc insulin microcrystals for storage in secretory granules (19,22). It is believed that formation of zinc proinsulin hexamers facilitates intracellular transport and that formation of microcrystals causes accumulation of insulin at a supersaturating concentration, thereby providing a mechanism for an acute secretion of insulin in response to glucose. ZnT-3 has been shown to be responsible for zinc accumulation in synaptic vesicles (12), but nothing is known about the mechanism by which zinc enters secretory granules in ␤ cells.
We previously showed transepithelial transport of several metal ions including zinc in polarized Madin-Darby canine kidney cells (23). This finding prompted us to search the mouse expressed sequence tag (EST) sequences homologous to ZRC1 (24), an S. cerevisiae zinc transporter that belongs to the CDF family. Further data base analysis and subsequent experiments resulted in identifying a novel zinc transporter, ZnT-5, that has a very long unique amino-terminal sequence, followed by carboxyl-terminal sequence significantly homologous to CDF family members. ZnT-5 was expressed abundantly in the human pancreas in association with secretory granules. Here we report the structural characteristics and localization of ZnT-5. We propose that ZnT-5 may be a transporter of zinc into secretory granules in ␤ cells.
Isolation of the Human and Mouse ZnT-5 Gene-We found three mouse EST clones (AI020400, AA409021, and AA797841) with homology to ZRC1, a zinc transporter of S. cerevisiae. EST clones derived from mouse and human tissues with sequences contiguous with these three fragments were identified from the data base. Using two gene-specific primers designed based on these EST clones (forward primer (1231F), 5Ј-CTGTCTGGAGGAGTGGTAGT-3Ј; reverse primer (1400R), 5Ј-GG-GATCGATTGAGAGCTATG-3Ј; numbering is defined as the transcription starting site as ϩ1), the human placenta cDNA library was screened with the Rapid cDNA PCR Screening Service (TAKARA, Kyoto, Japan). Thus, the 3Ј portion of a cDNA that hereafter is called human ZnT-5 (hZnT-5) was isolated. The 5Ј portion of hZnT-5 cDNA was isolated by primer extension techniques using human liver poly(A) RNA (OriGene Technologies) and reverse primer (961R, 5Ј-TGAC-CCATGGGCACAA-3Ј) as previously described (25). These two portions were ligated to construct full-length hZnT-5. We discovered the mouse ZnT-5 (mZnT-5) gene by searching the EST data base using hZnT-5 cDNA as a query (AW910349). Sequence reactions on both strands were performed using DNA Sequencing Kit (PerkinElmer Life Sciences). All homology searches were performed using the Basic Local Alignment Search Tools (BLAST; National Center for Biotechnology Information).
Northern Blot Analysis-Northern blot membranes for human multiple tissues and endocrine system purchased from CLONTECH were hybridized to hZnT-5 cDNA probe (from ϩ930 to the poly(A) site) 32 P-radiolabeled by using a random priming method (TAKARA) in Hybrisol I solution (Oncor), washed with 0.1ϫ SSC and 0.1% SDS at 42°C twice, and then autoradiographed for 24 h.
To produce an antigen, we constructed the plasmid for expression of a fused protein consisting of the carboxyl-terminal tail of hZnT-5 and maltose-binding protein. hZnT-5 cDNA corresponding to 133 amino acid residues from Ile 633 to Met 765 (see Fig. 1B) was inserted, in-frame, downstream of maltose-binding protein in pMAL-C2X vector (New England Biolabs).
Monoclonal Antibody against hZnT-5-An anti-hZnT-5 monoclonal antibody (mAb) was produced as described previously (27). The hybridoma that produces anti-hZnT-5 mAb was subcloned by the limiting dilution method, and a clone named 1-7B was established. 1-7B mAb was IgG1 subclass. A ascites was generated by injection of 1 ϫ 10 7 hybridoma cells into pristine-primed mice. 1-7B mAb was precipitated with 50% ammonium sulfate and purified with a protein G column.
Immunofluorescence and Electron Microscopic Immunohistochemistry-Double immunofluorescence for ZnT-5 and insulin was detected using paraffin sections from paraformaldehyde-fixed human pancreas. Dewaxed sections were incubated with 1-7B mAb (34 g/ml), Cy3labeled donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories), and guinea pig anti-insulin (5 g/ml; Zymed Laboratories Inc.) in this order, followed by fluorescein isothiocyanate-labeled donkey antiguinea pig IgG (Jackson ImmunoResearch Laboratories). The stained sections were observed under a confocal laser-scanning microscope (Fluoview; Olympus). For electron microscopy, the paraformaldehydefixed cells were rinsed in phosphate-buffered saline and processed by the pre-embedding silver-intensified immunogold method. After preincubation with normal goat serum, the samples were immunoreacted with 1-7B mAb (17 g/ml) and further incubated with anti-mouse IgG covalently linked to 1.4-nm gold particles (Nanogold; Nanoprobes). Following silver enhancement (HQ silver; Nanoprobes), specimens were osmificated, dehydrated, and embedded in Quetol 812. Ultrathin sections were prepared with an ultramicrotome (Leica Ultracut UCT; Nissei Sangyo, Tokyo, Japan), and stained with an aqueous solution of 2% uranyl acetate for observation under a transmission electron microscope.
Preparation of Membrane Proteins-Crude membrane proteins were prepared according to the method described by McMahon and Cousins (28). Briefly, cells (ϳ1 ϫ 10 7 /15-cm dish) were scraped into 5 ml of cold phosphate-buffered saline. These cells were pelleted and resuspended in 3 ml of cold homogenizing buffer (0.25 M sucrose, 20 mM HEPES, and 1 mM EDTA containing protease inhibitors) and homogenized with 10 strokes of a very tight fitting 5-ml Dounce homogenizer. The homogenate was centrifuged at 2200 rpm at 4°C to remove nucleus. The postnuclear supernatant was centrifuged for 30 min at 30,000 rpm at 4°C. The pellet was resuspended in 500 l of homogenizing buffer and stocked at Ϫ70°C as membrane proteins. The postnuclear supernatant was centrifuged on a 0.7-1.7 M sucrose gradient to fractionate the membrane proteins with density. Membrane proteins were prepared from each fraction.
Preparation of hZnT-5-induced Vesicles-A crude homogenate was prepared by homogenizing HeLa/hZnT-5 cells cultured for 36 h with 2.5 g/ml Dox (hZnT-5-induced condition) or without Dox (uninduced condition) with 10 strokes of a very tight fitting 10-ml Dounce homogenizer. The crude homogenate (6 ml) in homogenizing buffer was adjusted to   (29). The gradients were centrifuged for 2.5 h at 30,000 rpm. The turbid band at the 0.8 M/1.2 M sucrose interface was harvested by syringe puncture. These fractions were diluted to 0.25 M sucrose by the addition of 20 mM HEPES, 1 mM EDTA and recentrifuged for 30 min at 30,000 rpm. The pellet resuspended in uptake buffer was used for 65 Zn transport assays.

65
Zn Transport Assay-Uptake of zinc into hZnT-5-induced vesicles was measured by the rapid filtration method (30). Vesicles (20 g) were added to 120 l of uptake buffer (10 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , and 1 mM CaCl 2 ) containing 65 ZnCl 2 (PerkinElmer Life Science) and incubated at 37°C for the indicated periods. The reaction was stopped by filtrating aliquots of the reaction mixture through 0.45-m pore size nitrocellulose filters (Millipore Corp.), followed by washing twice with 1 ml of stopping buffer (uptake buffer containing 1 mM cold ZnCl 2 ). The filters used in each assay were pretreated with 5 ml of stopping buffer. The rate of 65 Zn uptake was estimated on an auto-␥ counter, COBRA II (Packard Instrument Co.).
Immunoblot Analysis-Membrane proteins (20 g) were lysed using 6ϫ SDS sample buffer at 37°C for 5 min and electrophoresed through 7.5% SDS-polyacrylamide gels. After transfer to polyvinylidene difluoride membrane (Pall Corporation), the blot was blocked with blocking solution (5% skim milk and 0.1% Tween 20 in phosphate-buffered saline) and then incubated with 1-7B mAb (1:3000 dilution) or anti-GM130 polyclonal antibody (gift from Dr. Nobuaki Nakamura; 1:1500 dilution) in blocking solution. Horseradish peroxidase-conjugated secondary antibodies (Amersham Pharmacia Biotech) were used at a 1:3000 dilution, and Super Signal Chemiluminescent Substrate (Pierce) was used for the detection. The intensity of bands was measured using Image Gauge (Fuji Photo Film, Tokyo, Japan).
Yeast Strains and Growth Conditions-The yeast strain used in this study was BY4742 (MAT␣ his3 leu2 ura3). DNA transformations were performed using the lithium acetate procedure. Yeast was transformed with the control plasmid (pYES2), the plasmid containing hZnT-5 cDNA, or the plasmid containing 5Ј-deleted hZnT-5 cDNA (⌬hZnT-5) in which the amino-terminal nine membrane-spanning domains had been deleted. The transformed yeasts were grown at 30°C for 3 days either on the YPD plates containing 2% galactose or on the same plates supplemented with 10 mM ZnSO 4 , 3 mM FeSO 4 , or 0.5 mM CoCl 2 .

Identification of Human and Mouse cDNA Encoding ZnT-5-
Data base comparisons of the mouse EST sequences with ZRC1 (24), a zinc transporter that belongs to CDF family, revealed several nucleotide fragments homologous to this transporter. Three of them were not found in known genes and partially overlapped. A further data base search of contiguous sequences identified a 2.3-kb fragment that contained a putative open reading frame (ORF), an in-frame termination codon, and a canonical polyadenylation signal. The 2.3-kb fragment did not contain a translation initiation codon, suggesting that this fragment is the 3Ј portion of a longer cDNA. We obtained human and mouse full-length cDNAs by the combination of rapid cDNA PCR screening and primer extension technique. Human cDNA is shown in Fig. 1A. Because this cDNA encodes a zinc transporter with homology to CDF family as described below, we hereafter refer to this protein as ZnT-5.
The ATG at position ϩ202 was assumed to initiate an ORF encoding hZnT-5 of 765 amino acids (Fig. 1A). This initiation codon was preceded with a 200-bp untranslated region that contains two small ORFs that can encode two small peptides (3 and 13 amino acids long) by initiating at positions ϩ103 and ϩ152, respectively. Interestingly, this structure in the untranslated region is completely conserved in mZnT-5. mZnT-5 had 761 amino acids and showed high homology to hZnT-5 (94%; Fig. 1B). Hydropathy plot analysis of hZnT-5 showed the potential 15 membrane-spanning domains (Fig. 1C). The aminoterminal portion (ϳ400 amino acid residues) containing nine membrane-spanning domains showed little homology to known proteins.
Northern blot analysis indicated that the hZnT-5 gene was ubiquitously expressed as 3-and 4-kb transcripts in human tissues (Fig. 2). High expression was seen in the pancreas, followed by the kidney and liver. hZnT-5 cDNA that we identified in this study corresponds to the 3-kb isoform. Analysis of the 4-kb transcript is under way in our laboratory.
Detection of hZnT-5 Protein by Anti-hZnT5 Monoclonal Antibody, 1-7B-A region of the carboxyl-terminal end of hZnT-5 was used as an antigen for antibody production. This carboxylterminal region was fused with maltose-binding protein, and the fused protein was expressed in E. coli. Purified fusion protein was used as an antigen, and a hybridoma producing anti-hZnT-5 monoclonal antibody (1-7B mAb) was established. To demonstrate that hZnT-5 is a membrane protein, we carried out immunoblot analysis using the membrane and cytosolic fractions derived from some human cell lines (HEp-2, Hep3B, JAR, and HeLa cells) and HeLa/hZnT-5 cells in which exogenous hZnT-5 is inducible when cultured in the presence of Dox (Fig. 3A). The 1-7B mAb recognized a 55-kDa protein in membrane fractions, and no positive bands were found in cytosolic fractions. The molecular size of hZnT-5 calculated from cDNA is 84 kDa. A significant difference in the calculated molecular size and that found by immunoblot analysis raised the possibility that hZnT-5 may be cleaved to yield the carboxyl-terminal protein detected by 1-7B mAb. To examine whether this is the case, we expressed the amino-terminal FLAG-tagged hZnT-5 protein in HeLa cells (HeLa/FLAG-hZnT-5) and detected the expressed hZnT-5 by an anti-FLAG antibody, M5. As Fig. 3B shows, FLAG-hZnT-5 migrated with a size of 55-60 kDa, which is similar to the size detected by the 1-7B mAb. Thus, the 55-kDa protein that reacts with 1-7B mAb is the intact hZnT-5.
Preferential Expression of hZnT-5 in Human Pancreatic ␤ Cells-Since hZnT-5 mRNA was abundantly expressed in human pancreas (see Fig. 2B), we immunohistochemically examined the localization of hZnT-5. As shown in Fig. 4A, expression of hZnT-5 in the human pancreas can be almost completely superimposed on that of insulin. ␤ cells expressed ZnT-5 much more abundantly than the other types of pancreatic cells; hZnT-5 was undetectable in other endocrine cell types includ- ing ␣ cells and was weakly detected in a limited number of acinar cells under the conditions used here. Moreover, the electron microscopic studies revealed that the immunoreactivity of hZnT-5 was associated with secretory granules (Fig. 4B). It appears that hZnT-5 is located in the perimeter of granules, which suggests the localization of hZnT-5 in granular membrane.
hZnT-5 Acts as a Zinc Transport Protein-Since the carboxyl-terminal portion of hZnT-5 has a strong homology to zinc transporters belonging to the CDF family and hZnT-5 was abundantly expressed in pancreatic ␤ cells in association with secretory granules, we tested whether hZnT-5 functions as a zinc transporter. We chose HeLa cells to find zinc transport activity of hZnT-5 because of their lesser expression of endogenous hZnT-5 among all tested human cell lines (data not shown). First, the subcellular localization of hZnT-5 was determined using HeLa/hZnT-5 cells cultured in the presence of Dox (hZnT-5-induced condition). The localization of hZnT-5 was indistinguishable from that of a marker protein of Golgi apparatus, GM130, by indirect immunoreactive staining (Fig. 5A) and immunoblot analysis (Fig. 5B), indicating that most of hZnT-5 is present in Golgi apparatus.
Next, because expression of hZnT-5 was increased 9-fold when HeLa/hZnT-5 cells were cultured in the induced condition (see Fig. 3A), we examined whether the induced HeLa/hZnT-5 cells accumulate zinc in Golgi apparatus more than did the uninduced cells using Zinquin, a zinc-specific fluorescence probe. No significant difference was found between the induced and uninduced cells (data not shown). Then we tried to detect zinc transport activity of hZnT-5 using Golgi-enriched vesicles prepared from total membranes of HeLa/hZnT-5 cells cultured in either hZnT-5-induced or uninduced condition. The membranes were fractionated by sucrose gradient centrifugation, and 5-fold enrichment of Golgi membranes was confirmed by immunoblot analysis (data not shown). hZnT-5-induced or uninduced vesicles were prepared from the enriched Golgi membranes (see "Experimental Procedures") and used for 65 Zn transport assays. Time course of 65 Zn uptake into vesicles was investigated using the hZnT-5-induced vesicles or uninduced vesicles (Fig. 6A). The 65 Zn uptake into the hZnT-5-induced vesicles was higher than that of uninduced vesicles at 37°C, although no difference in the uptake was seen at 4°C. The increased uptake observed in hZnT-5-induced vesicles at 37°C is probably caused by the induced hZnT-5 protein. The uptake at 37°C was much larger than that at 4°C, indicating that 65 Zn uptake is likely transporter-mediated accumulation into vesicles rather than cell surface binding. Fig. 6B shows that the rate of 65 Zn uptake at 37°C is concentration-dependent and saturable in both hZnT-5-induced vesicles and uninduced vesicles. The uptake rate in hZnT-5-induced vesicles at 37°C was higher than that in uninduced vesicles. The exogenous hZnT-5-derived activity calculated by subtracting uptake rates of uninduced vesicles from those of hZnT-5-induced vesicles was also concentration-dependent and saturable. The saturation curve fitted Michaelis-Menten kinetics with an apparent K m of 0.25 M and a V max of 15 nmol/mg protein/min.
In another study, we examined the growth dependence of hZnT-5-induced and uninduced HeLa/hZnT-5 cells on zinc in the medium but found no differences with the concentration of zinc (data not shown). Then, we examined the effects of hZnT-5 expression on yeast. The yeast expressing hZnT-5 grew normally in the normal medium (Fig. 6C, left) but showed poor growth in medium containing 10 mM zinc, which is the upper limit for the control yeast to grow (Fig. 6C, middle). A high iron concentration had no effect (Fig. 6C, right). It is noted that the growth of the yeast expressing the 5Ј-deleted hZnT-5 cDNA (⌬hZnT-5) in which the amino-terminal nine membrane-spanning domains have been deleted is unaffected by zinc at a high concentration (Fig. 6C, middle). This result indicates that hZnT-5 acts as a zinc transporter, although the mechanism responsible for the zinc-sensitive phenotype remains to be identified. A similar sensitive phenotype was seen in the presence of a high cobalt concentration (0.5 mM), suggesting that cobalt is a potential substrate of hZnT-5 (data not shown).  3. Immunoblot analysis with anti-hZnT-5 monoclonal antibody, 1-7B. A, hZnT-5 was present in membrane proteins. Immunoblot analysis was performed using the membrane proteins (mb) and cytosolic proteins (Cy) derived from some human cell lines (larynx carcinoma (HEp-2), hepatoma (Hep3B), placental trophoblast-derived choriocarcinoma (JAR), and cervix carcinoma (HeLa)) and HeLa/ hZnT-5 cells cultured in either induced or uninduced conditions. hZnT-5 was detected only in the lanes of membrane proteins. B, the 55-kDa protein was intact hZnT-5 protein. Immunoblot analysis was performed with 1-7B mAb (lanes 1 and 2) or anti-FLAG antibody, M5 (lanes 3 and 4), using membrane proteins prepared from HeLa/hZnT-5 cultured in the induced condition (lanes 1 and 4) and HeLa/FLAG-hZnT-5 cells cultured in the induced condition (lanes 2 and 3). 20 g of protein was loaded in each lane. The position of a protein size marker is indicated on the left.

FIG. 4. Preferential expression of hZnT-5 in human pancreatic ␤ cells.
A, double immunofluorescence staining of human pancreas was performed using 1-7B mAb and anti-insulin antibody. Note that hZnT-5 protein is specifically detected in insulin-containing ␤ cells but not in other endocrine cell types or acinar cells. Scale bar, 50 m. B, electron microscopic immunolabeling of ␤ cells shows that hZnT-5 was localized in association with secretory granules. Immunogold particles for hZnT-5 are present on the perimeter of secretory granules (arrows). Scale bar, 1 m.

DISCUSSION
A data base search of DNA sequences homologous to a yeast zinc transporter and cloning of full-length cDNA revealed a novel protein, which appeared to be a zinc transporter consisting of 765 amino acid residues. The primary sequence was well conserved in human and mouse. Since its carboxyl-terminal portion (365 amino acid residues) contains six membrane-spanning domains, a structural characteristic of metal transporters in the CDF family, and is highly homologous to mammalian zinc transporters (ZnT-1 to -4) that belong to the CDF family, we refer to this transporter as ZnT-5. Functional assays of this protein using Golgi-enriched vesicles and yeast cells supported this classification. ZnT-5, however, has a very long aminoterminal portion (400 amino acid residues) with nine membrane-spanning domains. MSC2 (YDR205W) protein (a CDF family member) with zinc homeostasis function in yeast has been shown to contain 12 membrane-spanning domains (31). MSC2 and ZnT-5 are homologous in the carboxyl-terminal portions with six membrane-spanning domains but not in the amino-terminal portions. To our knowledge, there is only one protein homologous to the amino-terminal portion of ZnT-5 in Caenorhabditis elegans (CAB81912). The function of this C. elegans protein is unknown.
The functions of most members of the CDF family have been examined by overexpression experiments; i.e. the cells gain resistance to metal toxicity, indicating that the six membranespanning domain is sufficient to exhibit transporter activity. These facts raised the possibility that ZnT-5 may posttranslationally be cleaved to yield the carboxyl-terminal fragment for manifesting transporter activity. This is unlikely because immunoblot analysis by the use of 1-7B mAb against the carboxylterminal peptide of hZnT-5 and the antibody against FLAG tagged at the amino terminus detected hZnT-5 with a similar molecular size. Furthermore, zinc at a high concentration in-hibited the growth of yeast expressing the full-length hZnT-5 cDNA but not that of yeast expressing the 5Ј-deletion form of hZnT-5 cDNA (⌬hZnT-5).
CDF family members are found in both eukaryotes and prokaryotes (5,32), and complementation assays between different species have been successful (14,33). For example, expression of mouse ZnT-4 in ⌬zrc1 yeast that cannot grow in medium containing zinc at a high concentration due to insufficient efflux of zinc allows the yeast to grow. Expression of hZnT-5 in ⌬zrc1 yeast, however, could not reduce zinc toxicity (data not shown). Instead, wild-type yeast expressing hZnT-5 showed limited growth at a high zinc concentration. We attempted to determine the localization of hZnT-5 expressed in yeast, but unfortunately yeast contained protein(s) that cross-reacts with 1-7B mAb; the high background made it difficult to localize hZnT-5 in yeast. The mechanism making the yeast hypersensitive to zinc remains unknown. Nevertheless, a zinc-dependent phenotype of yeast expressing hZnT-5 provides additional support that ZnT-5 transports zinc. Gaither and Eide (17) have reported a similar finding that yeast expressing hZIP1, a mammalian zinc transporter belonging to the ZIP family, shows limited growth at a high zinc concentration. hZIP1 does not complement the ⌬zrt1/zrt2 double mutant yeast strain, which cannot grow in a zinc-limiting condition (17).
The intracellular concentration of free zinc is estimated to be at a nanomolar level (34), because the major part of zinc in cytoplasm is bound to proteins such as methallothionein. A K m value of ϳ0.25 M that we calculated from the in vitro zinc transporter activity of hZnT-5 is relatively high, suggesting that free zinc may not be a real substrate of ZnT-5. A transport assay under conditions that mimic intracellular environments more faithfully is necessary. The addition of H ϩ and ATP showed no effects in zinc uptake by hZnT-5-induced vesicles (data not shown).

FIG. 5. hZnT-5 was localized in Golgi apparatus in HeLa/hZnT-5 cells cultured in hZnT-5-induced condition.
A, hZnT-5 was co-localized with a marker protein of Golgi apparatus, GM130. HeLa/hZnT-5 cells cultured in hZnT-5-induced condition were fixed and double-immunostained with 1-7B mAb or GM130 polyclonal antibody, followed by secondary antibodies conjugated with Alexa 488 or Alexa 594, respectively. B, hZnT-5 protein prepared from HeLa/ hZnT-5 cells cultured in hZnT-5-induced condition was fractionated into Golgi membrane fraction. The postnuclear supernatant derived from HeLa/hZnT-5 cells was separated into 11 fractions by centrifugation on a 0.7-1.7 M sucrose gradient. Membrane protein from each fraction was analyzed by the immunoblotting technique, and the distribution of hZnT-5 in the gradient was compared with that of GM130. The major part of hZnT-5 was detected in fractions 4 and 5, where GM130 was also concentrated, although the fractions with lower densities also contained small amounts of hZnT-5.
Zinc is not only a key structural component of a large number of zinc-containing proteins but also possesses important modulator functions in some specialized cells (1, 2, 18 -20). One of these is the formation of zinc insulin microcrystals in secretory granules in pancreatic ␤ cells. Zinc in these granules is believed to play two indispensable roles to store insulin in crystal forms (19,22,35); first, two zinc ions are bound with six proinsulin molecules to form a hexamer, which occurs just after endoplasmic reticulum export of proinsulin. Second, zinc promotes the granule core (crystal) formation, which occurs after proteolytic conversion from proinsulin to insulin in secretory granules. The extensive accumulation of zinc in pancreatic ␤ cells suggests that these cells are equipped with the powerful machinery that drives zinc influx into secretory granules. However, which transporter handles zinc into endoplasmic reticulum and secretory granules is unknown. Recently, mRNAs of ZnT-1 to -4 in pancreatic islets were detected by RT-PCR techniques (36), but neither the cellular nor subcellular distribution of these transporter proteins in the pancreas is known. We found that hZnT-5 protein is abundantly expressed in human pancreatic ␤ cells but not in glucagon-secreting ␣ cells and most acinar cells. It is noteworthy that hZnT-5 is associated with secretory granules. From these results, we propose that ZnT-5 is responsible for transporting zinc into secretory granules to form insulin crystals. Since ZnT-5 is ubiquitously expressed, however, ZnT-5 probably plays a role in the movement of zinc in a variety of tissues.