Subtype-specific accumulation of intracellular zinc pools is associated with the malignant phenotype in breast cancer

We first utilized X-ray fluorescence microscopy to determine the spatial distribution of Zn in Luminal and Basal tumors and adjacent normal tissue (Fig. 1). Spatial analysis revealed differences in Zn content and localization within the malignant regions. In Luminal breast tumors, Zn primarily accumulated around the tumor periphery. In Basal breast tumors, Zn was more evenly distributed throughout the malignant tissue. When compared with the distribution of calcium (Ca), some differences were noted such that Zn overlapped closely with Ca in Basal tumors, but this was less consistent in Luminal tumors.

Due to the differences in Zn distribution, we postulated that Basal and Luminal breast tumors would have unique patterns of Zn transporter expression. We analyzed microarray gene expression data derived from human breast tumors and noted a significant effect of subtype on Zn transporter expression (Fig. 2a). Pair-wise comparison of the subtypes show that Basal and ER⁺ tumors had a distinct pattern of SLC30A, SLC39A, and MT gene expression. Significant differences were detected in the expression of MT1, MT2, SLC30A5-6, SLC30A8-9, and SLC39A1-2, SLC39A4, SLC39A6-9, SLC39A11, and SLC39A14 with a false discovery rate (FDR) of p < 0.05. In addition, we noted more similarity between the Basal and HER2⁺ subtypes and less similarity between the Basal and ER⁺ subtypes with respect to the expression of the SLC30A, SLC39A, and MT genes (Table 1). Of particular note, we found that MTs, SLC39A4, and SLC39A14 were consistently over-expressed while SLC39A6, SLC39A9, and SLC39A11 were consistently under-expressed in the Basal subtypes (fold-change >1.5). Finally, among ER⁺ tumors, increasing tumor grade was associated with increased expression of SLC39A8. Taken together our data indicated that gene expression of members of the Zn transporting network was substantially different between Basal and Luminal breast tumor subtypes.

We next analyzed microarray gene expression data of all breast cancer cell lines that were available (Fig. 2b). Statistical analysis of Zn transporter expression indicated that similar to what was observed in breast tumor tissues, basal-like cancer cells also had significantly higher expression of MTs and SLC39A14 and lower expression of SLC39A6, SLC39A9, and SLC39A11 (fold-change >1.5; FDR, p < 0.05) (Table 1). In sum, genes regulating Zn transport have a subtype-specific pattern of expression in cell lines under normal culture conditions that is similar to that observed in breast tumors, suggesting that malignant cell lines are reliable models for dissecting Zn metabolism and the consequences of dysregulation in breast cancer.

Based upon these data, we chose T47D and MDA-MB-231 cells as common models representing Luminal and Basal subtypes of breast cancer, respectively. First, we measured the relative gene expression of all Zn transporters and MTs using relative real-time PCR (Fig. 3 and Table 1). Similar to our microarray analysis, we identified significant subtype-specific differences in the expression of numerous Zn management proteins. MT was only expressed in basal-like MDA-MB-231 cells. Expression of four SLC30A (SLC30A2, SLC30A5, SLC30A6, and SLC30A10) and two SLC39A (SLC39A4 and SLC39A11) genes were significantly lower in basal-like MDA-MB-231 cells compared to luminal T47D cells (p < 0.05). Expression of SLC30A9, SLC39A2, and SLC39A8 were significantly higher in MDA-MB-231 cells relative to expression in T47D cells (p < 0.05).

We next compared the expression of Zn transporters in T47D and MDA-MB-231 cells to the non-transformed mammary epithelial cell line MCF10A (Fig. 3). Luminal cells: Consistent with reports that MT over-expression is associated with invasiveness [23], we detected minimal mRNA expression of MTs in non-invasive, T47D cells. T47D cells had significantly lower expression of five SLC39A (SLC39A1, SLC39A3, and SLC39A8-10) and three SLC30A (SLC30A1, SLC30A7, and SLC30A9) genes relative to expression in MCF10A cells (p < 0.05). Expression of two SLC39A (SLC39A4 and SLC39A7) and two SLC30A (SLC30A2 and SLC30A10) genes were higher relative to expression in MCF10A cells (p < 0.05). Next, protein expression profiles were generated for most of the ZIP (SLC39A gene family) and ZnT (SLC30A gene family) proteins except for ZIP2, ZIP9, and ZnT7, which we were unable to identify suitable antibodies (Fig. 4). Changes in gene expression did not always parallel changes in protein abundance (Table 2). The protein abundance of ZnT9 and several ZIP proteins (ZIP1, ZIP4, ZIP7, and ZIP12) was significantly lower in T47D compared to MCF10A cells, while numerous ZIP (ZIP5, ZIP6, ZIP8, ZIP10, and ZIP14) and ZnT (ZnT1, ZnT2, ZnT3, ZnT4, ZnT5, ZnT8 and ZnT10) proteins were significantly over-expressed in T47D cells compared with MCF10A cells. In addition, the presence of an additional protein much larger than the predicted molecular mass was detected for ZIP3 (~130 kD), ZIP14 (~76 kD), and ZnT6 (~110 kD) in T47D cells. (Only the larger molecular mass of the ZIP14 protein has been reported previously [in U251 cells, Abcam], thus it was included in the quantification). Additionally, ZIP4 and ZIP11 proteins were only robustly detected in MCF10A cells and ZIP13 protein was not detected in human breast cells (data not shown). Basal-like cells: Consistent with reports that MT over-expression is associated with invasiveness [23], mRNA expression of MTs was ~200-fold higher in basal-like cells relative to expression in MCF10A cells (Fig. 3). MDA-MB-231 cells had lower (>4-fold) expression of six SLC30A (SLC30A1, SLC30A5, SLC30A6, SLC30A7, SLC30A8, and SLC30A10) and five SLC39A (SLC39A1, SLC39A2, SLC39A3, SLC39A6, and SLC39A9) genes. Only the expression of SLC39A5 was significantly higher (~5-fold) relative to its expression in MCF10A cells (p < 0.05). Again, changes in gene expression did not parallel changes in protein abundance (Table 2). The protein abundance of four ZnT (ZnT2, ZnT5, ZnT8, and ZnT9) and six ZIP (ZIP4, ZIP5, ZIP7, ZIP8, ZIP11, and ZIP12) proteins was significantly lower, while only two Zn transporters (ZIP10 and ZnT1) were significantly over-expressed at the protein level in MDA-MB-231 compared with MCF10A cells (Fig. 4). A larger isoform (~2 kD larger than expected) of ZIP1 and ZIP10 was detected in MDA-MB-231 cells; however, these isoforms were not detected in MCF10A cells and were not included in the densitometric analysis. ZnT10 and ZIP14 were not detected in MCF10A or MDA-MB-231 cells.

As described above, we noted that while mRNA expression of numerous Zn transporters in MDA-MB-231 cells was similar to/greater than expression in MCF10A cells (ZIP4, ZIP5, ZIP7, ZIP8, ZIP11, ZIP12, ZnT2, and ZnT9), the protein levels were greatly reduced. To provide some mechanistic understanding behind this finding we chose to focus on ZnT2. To determine why ZnT2 protein was minimally detectable in MDA-MB-231 cells, we inhibited proteasomal and lysosomal degradation and measured ZnT2 abundance by immunoblotting. Treatment of MDA-MB-231 cells with the proteasomal inhibitor MG132, but not the lysosomal inhibitor chloroquine, resulted in greater abundance of ZnT2 compared to untreated MDA-MB-231 cells (Fig. 5), suggesting that ZnT2 is proteosomally degraded in MDA-MB-231 cells.

We next assessed differences in Zn content and distribution within these breast cancer and non-malignant cells. We first quantified the total amount of Zn in cell lysates using atomic absorption spectroscopy. Similar to observations in breast tumors, T47D cells accumulated ~5-fold greater Zn (0.276 ± 0.01 μg Zn/mg protein) while MDA-MB-231 cells accumulated ~2.5-fold greater Zn (0.107 ± 0.02 μg Zn/mg protein) relative to MCF10A cells (0.057 ± 0.06 μg Zn/mg protein), p < 0.05. To visualize the localization of Zn pools within these distinct subtypes, we used the Zn-responsive fluorophore FluoZin-3 [23] which fluoresces upon Zn binding (kD = 4–15 nM) [24]. We noted that the staining pattern of FluoZin-3 was distinctly different in each cell line (Fig. 6a). Labile Zn pools in MCF10A cells were confined to small peri-nuclear vesicles, consistent with our previous observation that Zn accumulates in the Golgi apparatus/endoplasmic reticulum in normal mammary cells [25]. In contrast, labile Zn pools in T47D cells were enriched in numerous large punctate intracellular vesicles, while in MDA-MB-231 cells, hazy fluorescence was noted throughout the cytoplasm and in a large, amorphous subcellular compartment(s). Importantly, consistent with the measurements of total Zn concentration noted above, quantification of FluoZin-3 fluorescence suggested that T47D cells accumulated significantly greater labile Zn compared with MCF10A and MDA-MB-231 cells (Fig. 6b). Collectively, Zn transporter protein profiling (setting a threshold of a >1.4-fold change in protein abundance) combined with the visualization of sub-cellular Zn pools using FluoZin-3, allowed us to generate a model of the Zn ionome (the profile of subcellular Zn transporters and Zn distribution) in luminal and basal-like breast tumor cells (Fig. 6c). This modeling illustrates that increased abundance of numerous ZIP and vesicular ZnT proteins (ZnT2, ZnT3, ZnT4, ZnT5, ZnT8, and ZnT10) is associated with enhanced vesicular Zn accumulation in luminal cells, while increased expression of only two ZIP proteins and MT is associated with more moderate Zn accumulation in basal-like cells.

Because ZnT2 expression is positively associated with vesicular Zn accumulation in malignant breast cancer cells, we next used immunofluorescent imaging in Luminal and Basal tumors to determine if the abundance of ZnT2 was similar to our observations in luminal and basal-like breast cancer cells. Indeed, we found that ZnT2 was primarily detected in Luminal tumors and in some cases was enhanced around the periphery (Fig. 6d), similar to the pattern of Zn accumulation observed using Xray fluorescence microscopy (Fig. 1). However, this peripheral staining pattern in Luminal tumors was not consistently observed. Moreover, minimal ZnT2 was detected in Basal tumors, collectively reflecting the results obtained in cultured breast cancer cells.

To test the hypothesis that the ability to sequester Zn in vesicles via ZnT2 is an important driver of the molecular phenotype, we expressed ZnT2 in MDA-MB-231 cells and assessed the functional response (Fig. 7). We chose this approach because ZnT2 is a vesicular transporter important for Zn accumulation [26] and protection against Zn cytotoxicity [27], this approach complemented our previous report that ZnT2 attenuation in T47D cells reduces tumorigenesis [18], and ZnT2 was significantly over-expressed in luminal cells and Luminal tumors and not expressed in basal-like cells and Basal tumors. In the current set of studies, we found that ZnT2 over-expression in MDA-MB-231 cells (Fig. 7a) led to Zn vesicularization and decreased cell viability. FluoZin-3 imaging revealed that in MDA-MB-231 cells transfected to express ZnT2, labile Zn accumulated in vesicles throughout the cell (Fig. 7b). Moreover, there were significantly more vesicles (Fig. 7c) and vesicles of a larger size (Fig. 7d) compared to vesicles in the mock transfected cells. In addition, the overall fluorescence intensity of vesicles per cell (Fig. 7e) was significantly greater suggesting that ZnT2 over-expressing cells accumulated more vesicular Zn. This was associated with a concurrent change in morphology (Fig. 7f) and a ~25 % reduction in cell viability (Fig. 7g).

To better understand the effects of ZnT2 over-expression on cell viability, we analyzed the effects of ZnT2 over-expression on cell cycle and CDK2 activity. CDK2 is a cell cycle protein that is critical for the G1/S transition. We observed distinct shifts in cell cycle in ZnT2 over-expressing cells compared to mock-transfected cells (Fig. 8a). Immunoblot analysis and kinase activity assays for CDK2 revealed that Zn sequestration in ZnT2 over-expressing cells had no effect on CDK protein abundance (Fig. 8b; top panel); however, CDK2 activity was significantly reduced by ~80 % (Fig. 8b; bottom panel). Shifts in cell cycle in ZnT2 over-expressing cells were also associated with decreased cell proliferation (Fig. 8c) and increased apoptotic cell death (Fig. 8d and e). However, most importantly, the molecular phenotype was profoundly affected, as invasion was significantly reduced (80 %) in ZnT2 over-expressing MDA-MB-231 cells (Fig. 8f).

A breast cancer cell line dataset (GSE12777, [58]) and a breast tumor dataset (GSE5460, [59]) was used in the analysis. The analysis was performed using the PARTEK software [60]. The cel files were normalized using RMA and the batch effect was corrected using scan date and subtype as ANOVA factors within PARTEK batch-correction implementation. Finally, each array was scaled to the grand median to ensure accuracy of inter-array comparisons. Breast tumors were classified into Basal, ER-High Grade (ER HG), ER-Low Grade (ER LG) and HER [61]. Breast cancer cell lines were classified into basal and luminal based on previously published work and the unassigned cell lines were classified based on cluster membership after Hierarchical clustering of the top 10 % highly variant genes [62, 63]. The Jetset definitions for “best-probeset” (jetset.scores.hgu133plus 2_0.99.3.csv) available for download at [64] was used to restrict the analysis to one probe-set per gene [39]. Subsequent analysis was performed on the SLC30A1-A10 (ZnT proteins), SLC39A1-A14 (ZIP proteins) and the metallothionein genes MT1E, MT1F, MT1G, MT1H and MT2A. The latest Affymetrix annotation (June 2011) assigns probeset 219215_s_at to SLC39A4 and 202667_s_at to SLC39A4 and SLC39A7. However both Jetset and GeneAnnot [65] assigned probeset 202667_s_at to SLC39A7 and 219215_s_at to SLC39A4, therefore we followed the Jetset/GeneAnnot definitions and retained both probesets in our analysis. ANOVA was used to identify differentially expressed genes and the FDR (step-up) corrected p-value <0.05 was used as the cut-off criteria for selecting significant genes. Hierarchical clustering based on Spearman rank dissimilarity of gene expression values and complete linkage was used to generate the heatmaps.

Human malignant luminal ER⁺/PR⁺/HER2⁻ (T47D), basal-like ER⁻/PR⁻/HER2⁻ (MDA-MB-231) and non-malignant (MCF10A) breast cells were chosen to represent three different breast cell subtypes. Cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA). T47D cells were maintained in growth medium containing, RPMI 1640 (SIGMA, St. Louis, MA) supplemented with fetal bovine serum (10 %), insulin (0.2 units/mL), sodium pyruvate (1.0 mM) and penicillin/streptomycin (1 %). MDA-MB-231 cells were maintained in L15 medium containing penicillin/streptomycin (1 %) and horse serum (10 %). MCF10A cells were maintained in 171 Medium supplemented with Mammary Epithelial Growth Supplement (Invitrogen, Carlsbad, CA). All culture mediums contained ~5 μM Zn as assessed by atomic absorption spectroscopy. Cells were routinely cultured in plastic 75 cm² flasks and sub-cultured every 4–5 days. Cells were maintained in a humidified chamber in 5 % CO₂ at 37 °C.

Cells were cultured on 15 cm² polycarbonate dishes in growth medium until 90–100 % confluent. Cells were initially rinsed with PBS, and then rinsed with PBS plus EDTA (1 mM) to remove any loosely bound Zn. Cells were collected by gentle scraping and pelleted by centrifugation at 2000 g for 10 min at 4 °C. Cellular protein concentration was determined by the Bradford assay. Cells were resuspended in Ultrex II Nitric Acid (0.5 mL, VWR, West Chester, PA) in mineral-free polypropylene vials and digested at room temperature overnight. Zn concentration was analyzed by atomic absorption spectroscopy using an Atomic Absorption Analyst 400 (Perkin Elmer, Waltham, MA) with WinLab32 software. Data was normalized to total protein content measured by the Bradford assay.

Labile Zn pools were characterized and visualized as previously described [25]. Cells were seeded onto glass coverslips and cultured overnight until 60–90 % confluent. Cells were rinsed twice with PBS then loaded with FluoZin-™3 AM (1 μM in DMSO containing pluronic acid 127 to a final concentration of 0.02 %; Invitrogen, USA) following manufacturer’s instructions in Opti-MEM for 1 h at 37 °C. Cells were briefly rinsed twice with PBS and washed with PBS for 30 min at 25 °C with constant shaking. Images were collected from live cells using a FV-1000 confocal microscope (Penn State Microscopy and Cytometry Facility). Number, size and the fluorescence intensity of vesicles were analyzed using Imaris® software (Connecticut, USA).

Discarded human tissue samples were collected under Dana-Farber Harvard Cancer Center institutional review board protocol #93-085. Tissue samples were plunge-frozen in ice-cold isopentane bath and processed for microscopy as previously described [25]. Frozen, unfixed tumors were sectioned (5 μm), mounted on positively-charged glass slides, dried at room temperature, post-fixed in phosphate-buffered 4 % paraformaldehyde, washed three times with 1× PBS at 4 °C for 5 min and stained with hematoxylin and eosin (H & E). Briefly, sections were air dried for 4 min, stained with 0.1 % hematoxylin for 2 min and rinsed in ddH₂O for 5 min. Sections were then stained with 0.5 % Eosin times and then rinsed in distilled H₂O 3 times. Sections were then dehydrated in ethanol and rinsed in xylenes. Regions of malignant tissue were identified. Serial sections (20 μm) were used for X-ray fluorescence microscopy and imaged with the scanning x-ray microprobe at beamline 2-ID-E at the Advanced Photon Source (Argonne, IL), quantified and processed as previously described [25].

Frozen, unfixed tumors were sectioned (5 μm) with a Microm HM 505 E cryostat (GMI, Ramsey, MN) at −25 °C and briefly dried at room temperature onto positively-charged slides. Sections were then post-fixed in phosphate-buffered 4 % paraformaldehyde at room temperature for 15 min. Following fixation, sections were washed with 1× PBS at 4 °C for 5 min; this was repeated 3 times. Sections were treated with 3 % H₂O₂ for 10 min and washed twice in 1× PBS for 5 min. Sections were then permeabilized with 0.2 % triton X-100 in 1× PBS for 45 min at room temperature. Sections were blocked (0.1 % heat inactivated goat serum, 1 % BSA, 0.3 % Triton X-100 in 1× PBS) for 1 h at room temperature in a humidified chamber. ZnT2 antibody [66] was diluted in blocking buffer (4 μg/mL) and sections were incubated overnight at 4 °C in a humidified chamber. Sections were washed 3 times with 1× PBS for 5 min and then incubated with DAPI (175 μg/mL) for 10 min at RT, then washed three times with 1× PBS for 5 min. Tissue was mounted in ProLong® Diamond Antifade (ThermoFisher Scientific, USA) and the coverslips were sealed with nail polish.

Total RNA was isolated from cells using Trizol (Invitrogen, Carlsbad, CA) according to manufacturer’s instructions. RNA was quantified by spectrophotometry and the integrity was assessed by examination of 28S and 18S bands in 2 % agarose gel electrophoresis. cDNA was synthesized from 1.0 μg of total RNA using TaqMan® reverse transcription kit (Applied Biosystems, Foster City, CA) in 25 μl reaction mixture following manufacturer’s instructions. The reaction mixture was incubated at 25 °C for 10 min, then at 48 °C for 30 min and heated to 95 °C for 5 min. cDNA products were stored at −20 °C until used for semi-quantitative PCR. Semi-quantitative PCR was performed using the DNA Engine Opticon 2 System real-time thermocycler (BioRad, Hercules, CA) coupled with SYBR Green technology (BioRad) and gene-specific primers to human Zn transporters (SLC39A1-14 and SLC30A1-10), MT (predicted to detect MT1A, MT1C, MT1D, MT1E, MT1F, MT1H, MT1L, MT1S, MT1X and MT2A) and human β-actin (Primer 3 Input v4.0). The PCR cycling parameters were as follows: 95 °C for 10 min, and 40 cycles of 95 °C for 15 s, 60 °C for 30 s and 72 °C for 30 s. The linearity of the dissociation curve was analyzed by Opticon 2 System software and the mean cycle time of the linear part of the curve was designated Ct. Each sample was analyzed in duplicate and normalized to β-actin using the following equation: ΔCt_gene = Ct_gene − Ct_β-actin. The difference in expression between MDA-MB-231 and T47D cells was calculated using the following equation: 2^(ΔΔCt), (ΔΔCt = mean ΔCt_gene in MDA-MB-231 cells − mean ΔCt_gene in T47D cells. Values represent mean fold change ± SD, relative to MDA-MB-231 cells (set to 100 %). The difference in expression between MDA-MB-231 or T47D cells and non-malignant MCF10A cells was calculated using the following equation: 2^(ΔΔCt), (ΔΔCt = mean ΔCt_gene in MCF10A cells − mean ΔCt_gene in MDA-MB-231 or T47D cells. Values represent mean fold change ± SD, relative to MCF10A cells (set to 100 %).

Total membrane proteins were isolated from cultured cells, electrophoresed (20–100 μg/sample) and transferred to nitrocellulose as previously described [66]. Additional file 1: Table S1 identifies the antibodies used and the molecular mass of Zn transporters in various cell lines and tissues that have been reported in the literature. Antibodies used in this study are noted. Membranes were blocked for 1 h in 5 % non-fat milk in PBS/0.1 % Tween-20 (PBS-T) and washed 3 times in PBS-T, followed by incubation with antibodies directed against Zn transporters for 45 min then washed 3 times in PBS-T. We were unable to identify suitable antibodies for ZnT7, ZIP2 and ZIP9. Proteins were detected following incubation with donkey, anti-rabbit IgG (1:30,000) conjugated to horseradish peroxidase (Amersham Pharmacia Biotech), visualized with Super Signal Femto Chemiluminescent Detection System (Pierce, Rockford, IL) and exposed to autoradiography film. Membranes were stripped and reprobed for β-actin to control for equal protein loading. Relative band density was quantified using the Carestream Gel Logic 212 Pro and the ratio of Zn transporter: β-actin was used for analysis. Samples were run in duplicate or triplicate and immunoblots were repeated 3–5 times. ZnT2 over-expression in transfected cells was confirmed by incubating membranes with anti-HA antibody (0.8 μg/mL) for 1 h, detected with secondary antibody labeled with IRDye (1:20,000) for 1 h protected from light. ZnT2-HA was detected using the LI-COR® Odyssey CLx System (LI-COR; Lincoln, NE).

MDA-MB-231 cells were plated in 6-well plates at a cell density of 5 ×10⁵. Twenty-four h later, MDA-MB-231 cells were treated with either 30 μM of MG132 (Sigma-Aldrich) for 6 h or 10 μM of Chloroquine Diphosphate salt (MP Biomedicals, LLC; Solon, OH) for 24 h. Total membrane preps and immunoblot analysis were executed as described above. Briefly, ZnT2 was detected following incubation with anti-rabbit IgG (1 μg/mL) conjugated to horseradish peroxidase and visualized with Super Signal Femto Chemiluminescent Detection System (Pierce, Rockford, IL) on a FluorChem System (ProteinSimple, San Jose, CA). Nitrocellulose membranes were stripped once and reprobed for β-actin. Relative densitometry was quantified using the AlphaView® Software (ProteinSimple, San Jose, California). ZnT2 densitometry was normalized to β-actin and used for analysis. Samples were run in triplicate and immunoblots were repeated 2–3 times.

ZnT2 over-expression in MDA-MB-231 cells utilized the Lipofectamine® 2000 Transfection Reagent delivery system (Thermo Scientific, Grand Island, NY). MDA-MB-231 cells were plated at a cell density of ~5 × 10⁵ in 6-well plates (Corning®, USA) and cultured until ~80–90 % confluence. Cells were transfected with Lipofectamine® 2000 and plasmid containing ZnT2-HA [26] (2.5 μg) at a ratio of 2.8:1 (transfection reagent: DNA ratio) in Opti-MEM® (Life Technologies, USA) per manufacturer’s instructions. Cells were transfected for 5 h, after which the transfection medium was removed and replaced with antibiotic-free, normal growth medium. All experiments were done 24 h post transfection. Transfection was confirmed by immunoblotting as described above.

Proliferation was determined with the MTT Cell Growth Determination Kit (Sigma-Aldrich, USA). MDA-MB-231 cells were transfected as described previously in a 6-well plate. Cells were trypsinized and plated in four 96-well plates at a cell density of 5 × 10³. Cells were allowed to grow for up to 48 h. Time 0 represents cells plated and analyzed on the same day (~6 h after cells attach). Cells were treated with 3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyl tetrazolium bromide (MTT) as instructed by the manufacturer. Briefly, 10 μL of MTT was added aseptically per well and incubated for 3 h. Medium was removed and 100 μL of MTT solvent added. Absorbance was read at 570 nm. Data represent mean ± SD, n = 3 samples/genotype from 3 independent experiments.

Cells were plated (3 × 10⁵) in a 6-well plate and transfected as described previously. Cells were collected 24 h post-transfection and suspended in cold 70 % ethanol and stored at −20 °C for 24 h. Cells were stained with propidium iodide (BD Biosciences; San Jose, CA) and analyzed by flow cytometry using a FACSCALIBUR (BD Biosciences; Penn State Hershey Flow Cytometry Core Facility). Data represent mean ± SD, n = 3 sample/genotype from 2 independent experiments.

Immunoprecipitation of CDK2 protein containing complexes and determination of associated kinase activity were both determined as previously described [67]. Briefly, protein extracts were prepared from 2 × 10⁶ MDA-MB-231 mock-transfected cells and cells over-expressing ZnT2 using the glass-bead breakage method, followed by immunoprecipitation of CDK2-containing protein complexes from 200 μg of protein, with the exception of the pre-clearing step which was performed using a 1:1000 dilution of normal rabbit IgG (Upstate). Immunoprecipitated CDK2 complexes and CDK2 protein expression levels in total cell lysates were determined using immunoblot analysis. Immunoprecipitated CDK2-containing protein complexes were subjected to a Histone H1 kinase assay. Briefly, immunoprecipitated CDK2 complexes were washed once in immunoprecipitation buffer and then twice with kinase reaction buffer. Kinase reactions were performed in a final volume of 20 μL consisting of kinase buffer supplemented with 20 μM ATP, 10 μCi of [γ-³²P] ATP, and 1 μg of histone H1 (Roche) as substrate. Kinase assay mixtures were incubated at 30 °C for 30 min. Reactions were stopped with the addition of loading buffer, and samples were boiled for 10 min. Reactions were resolved by electrophoresis on a 10 % SDS-polyacrylamide gel and dried, followed by autoradiograpy.

Apoptosis was determined via the Annexin V-FITC kit (Trevigen, Gaithersburg, MD). Briefly, 3 × 10⁵ cells were transfected in a 6-well plate; 24 h later, cells were trypsinized (2 min), washed and centrifuged at 300 × g for 5 min at room temperature. Cells were stained with both Annexin V-FITC and propidium iodide for 15 min in the dark at room temperature. Cells were washed and processed by flow cytometry. Data represent mean ± SD, n = 3 samples/genotype from 3 independent experiments.

Cells were transfected as previously described and 4 × 10⁴ cells were added to 200 μL of serum- and antibiotic-free growth medium per Boyden insert while 600 μL of antibiotic-free growth medium, with 10 % FBS, was added to the bottom of the well (in a 24-well plate). Cells were incubated in a humidified chamber for 24 h at 37 °C. Inserts were removed and submerged in PBS several times to remove unattached cells. Non-invading cells were removed by gently scraping with a wet cotton applicator. Cells were fixed by submerging the inserts into 4 % paraformaldehyde for 10 min. The insert was washed with 1× PBS and stained with hematoxylin and 3 % glacial acetic acid for 30 min. The insert was washed gently, several times with distilled water. Migrated cells were visualized under light microscopy at 40× by cutting out the porous membrane of the insert and mounting on a slide (migrated side down). Data represent mean ± SD, n = 3 samples/genotype from 2 independent experiments.

Gene names (SLC30A and SLC39A) and concomitant proteins (ZnT and ZIP, respectively) are used to differentiate between gene and protein expression. Results of studies in cultured cells are presented as mean ± SD or SEM where indicated. Statistical comparisons were performed using Student’s t-test or one-way ANOVA as indicated (Prism Graph Pad, Berkeley, CA). A significant difference was demonstrated at p < 0.05.