Metallothionein Is Part of a Zinc-scavenging Mechanism for Cell Survival under Conditions of Extreme Zinc Deprivation*

Metallothionein (MT) is a small cysteine-rich protein thought to play a critical role in cellular detoxification of inorganic species by sequestering metal ions that are present in elevated concentrations. We demonstrate here that metallothionein can play an important role at the other end of the homeostatic spectrum by scavenging an essential metal in a mouse fibroblast cell line that has been cultured under conditions of extreme zinc deprivation (LZA-LTK−). These cells unexpectedly produce constitutively high levels of metallothionein mRNA; however, the MT protein accumulates only when high concentrations of zinc are provided in the media. Until this MT pool is saturated, no measurable zinc remains in the external media. In this case, zinc deprivation leads to amplification of the MT gene locus in the LZA-LTK− cell line. Furthermore, the intracellular zinc levels in the fully adapted cells remain at the normal level of 0.4 fmol zinc/cell, even when extracellular zinc concentration is decreased by 2 orders of magnitude relative to normal media.

Metallothionein (MT) is a small cysteine-rich protein thought to play a critical role in cellular detoxification of inorganic species by sequestering metal ions that are present in elevated concentrations. We demonstrate here that metallothionein can play an important role at the other end of the homeostatic spectrum by scavenging an essential metal in a mouse fibroblast cell line that has been cultured under conditions of extreme zinc deprivation (LZA-LTK ؊ ). These cells unexpectedly produce constitutively high levels of metallothionein mRNA; however, the MT protein accumulates only when high concentrations of zinc are provided in the media. Until this MT pool is saturated, no measurable zinc remains in the external media. In this case, zinc deprivation leads to amplification of the MT gene locus in the LZA-LTK ؊ cell line. Furthermore, the intracellular zinc levels in the fully adapted cells remain at the normal level of 0.4 fmol zinc/cell, even when extracellular zinc concentration is decreased by 2 orders of magnitude relative to normal media.
Zinc is an ubiquitous and essential component in biological systems. Iron is the only other transition metal that is more abundant in humans; however, if one subtracts the amount of iron in hemoglobin, zinc becomes the most abundant transition metal (1). Zinc has been identified as a central component of over 300 enzymes and plays an essential structural function in an entire class of transcription factors. The biological essentiality of zinc implies the existence of homeostatic mechanisms that regulate its absorption, distribution, cellular uptake, and excretion (2). However, until recently little was known about how these processes occur within the cell or about the molecules that mediate their action (3).
Metallothionein (MT) 1 is an abundant zinc-binding protein and one of the few eukaryotic proteins identified as having an essential role in heavy metal detoxification. Copious levels of MT protein and mRNA are found in organisms and tissues exposed to high levels of zinc or cadmium (4). Transcriptional activation of the MT genes in response to high concentrations of these potentially toxic metals is mediated through trans-acting proteins that bind to DNA regulatory elements located upstream of the MT gene coding sequences (5)(6)(7)(8). These metalresponsive elements (MREs) are necessary and sufficient to confer zinc-and cadmium-responsive activation of MT genes (9). Each MT protein molecule may bind up to seven atoms of either zinc or cadmium, where each metal is tetrahedrally coordinated to cysteine residues. Because newly synthesized MT proteins sequester the inducing metals, elevated tissue concentrations of the ions are often detected upon metal exposure (10 -12). Conversely, MT protein and mRNA levels generally decrease in a dose-dependent manner in tissues of animals fed zinc-restrictive diets (13)(14)(15). Additional studies showed that moderate maternal zinc deficiency in rats during pregnancy and lactation results in the reduced expression of MT in the livers of their pups (12, 16 -18). Such results have led to suggestions that the MT protein may serve to buffer intracellular zinc levels (12).
We have described the methods used to generate a cell line derived from mouse fibroblasts (L-M(TK Ϫ )), herein designated as LTK Ϫ , which survives conditions of extreme zinc deprivation (19). Previous attempts at establishing similar zinc-deficient cell lines by other laboratories were unsuccessful in producing a system that would survive beyond a few passages of growth in medium containing less than 0.1 M zinc (20 -24). This cell line, designated as low zinc-adapted (LZA-LTK Ϫ ) survives continuous passage in media that contain less than 60 nM zinc and thus provides a useful system to study fundamental components of the cellular machinery responsible for zinc homeostasis. Furthermore, despite continuous culture in these conditions of extreme zinc deprivation, these cells remarkably maintained intracellular zinc at levels comparable with the parental LTK Ϫ cell line grown in normal media containing 1-5 M zinc (19).
In this paper, we present evidence that MT plays a role in the maintenance of intracellular zinc at the standard 0.4 fmol zinc/cell level in the LZA-LTK Ϫ cells, thus identifying a zincscavenging role for the protein. Unlike the studies in which animals were fed zinc-restrictive diets (12, 16 -18), MT mRNA is not degraded in the LZA-LTK Ϫ cell line; instead, it is constitutively present at high levels. Our results suggest that the overexpression of MT is one step that enhances the survival of the LZA-LTK Ϫ cells in conditions of extreme zinc deprivation. cations. The Chelex resin was prepared according to the manufacturer's instructions with neutralization of the resin to physiological pH and multiple washings with 0.25 M HEPES, pH 7.4, before its addition to media (at 40 g/liter) and other solutions (at 20 g/liter) with constant stirring at 4°C for 20 -24 h. Aliquots of the solutions were removed prior to and after Chelex treatment, and the concentrations of divalent cations in the samples were determined using an inductively coupled plasma-atomic emission spectrophotometer (ICP-AES) on an Atomscan 25 (Thermo Jarrell Ash) with Thermospec software (version 5.02). Divalent cations other than zinc were restored to their normal concentrations by the addition of appropriate metal chloride salts. Sterilization of the media following Chelex treatment was accomplished by filtering through a 0.2 M Nylon Media Plus filter unit (Nalgene).
Measurement of Intracellular Zinc-LZA-LTK Ϫ cells were passaged into zinc-deficient medium supplemented with various concentrations of zinc chloride from a 5 mM stock solution. Following growth, the media were decanted and saved for zinc analysis by ICP-AES. The cells were washed three times with metal-stripped 1ϫ phosphate-buffered saline (PBS) (136 mM NaCl, 2.6 mM KCl, 10.14 mM Na 2 HPO 4 , 1.76 mM KH 2 PO 4 , pH 7.4) to remove the contaminating traces of zinc. The cells were scraped into metal-stripped 1ϫ PBS and were pelleted at 14,000 ϫ g for 1 min. The cell pellet was dried overnight in a 65°C oven, digested with 150 l of concentrated nitric acid at 80°C for 15 min, and diluted to 1.5 ml with distilled H 2 O prior to analysis on the ICP-AES.
RNA and DNA Filter Hybridization-RNA was isolated from cell cultures lysed with guanidinium thiocyanate by sedimentation through a cesium chloride cushion (25). Equivalent amounts of RNA were fractionated on formaldehyde-agarose gels (26), transferred to nitrocellulose, and hybridized with radiolabeled cDNA probes for mouse MT-I (a gift from Dr. Richard Palmiter) or GAPDH (a gift from Dr. Richard Morimoto). Genomic DNA was purified from cells lysed with SDS/ proteinase K solution by phenol/chloroform extraction (27). DNA digested with EcoRI was fractionated by agarose gel electrophoresis, transferred to nitrocellulose, and probed with radiolabeled mouse MT-I or GAPDH cDNA. The nitrocellulose filters were washed and exposed to x-ray film at Ϫ70°C with an intensifying screen.
[ 35 (27). After brief centrifugation to rid the extract of membranes, the soluble extract was carboxymethylated with 12.5 mM iodoacetate, 50 mM Tris, pH 8.0, at 50°C for 15 min (28). Equivalent amounts of protein per lane, as determined by the Bradford method, were electrophoresed utilizing Tricine-based PAGE (29). Following electrophoresis, the gels were fixed, treated with En 3 hance (Du-Pont), and dried prior to autoradiography.
Subcellular Fractionation-LZA-LTK Ϫ cells that had been grown in various [Zn] media were washed with metal-stripped 1ϫ PBS. Cells were scraped into metal-stripped 1ϫ PBS, pelleted by centrifugation, and resuspended in 0.25 M sucrose buffered with 5 mM MOPS, pH 7.4. Whole cell zinc levels were measured by ICP-AES using an aliquot of the resuspended pellet that had been digested with nitric acid. Other cell aliquots were passed three times through a stainless steel syringe filter holder that was lined with two polycarbonate 16.0-m filters (Osmonics). ␤-Hexosaminidase activity assays, which test for the integrity or latency of the organelles, were conducted on aliquots of each homogenate (30). Differential velocity sedimentation resulted in four fractions: 1) P1 (pellet from homogenate spun at 1,000 ϫ g, 10 min), which typically contains nuclei and unbroken cells; 2) P2 (pellet from the P1 supernatant spun at 10,000 ϫ g, 10 min) containing large intracellular membranes; 3) P3 (pellet from the P2 supernatant spun at 100,000 ϫ g, 60 min), which typically contains smaller intracellular membranes and vesicles; and 4) the soluble fraction or cytosol (the supernatant from the 100,000 ϫ g, 60-min spin) (31). As with the total intracellular zinc measurements, the pellets from each of the fractions were dried overnight, dissolved in concentrated nitric acid, diluted in distilled H 2 O, and analyzed for zinc content by ICP-AES.
Gel Filtration Column Chromatography-Cytosols from subcellular fractionation experiments were chromatographed on a G-75 Sephadex superfine size exclusion column (Amersham Pharmacia Biotech) utilizing 20 mM Tris, pH 7.8, as the running buffer. In the cadmium saturation analyses, a portion of the cytosol or purified horse MT (Sigma) was incubated with a 5-fold molar excess of cadmium chloride, relative to zinc concentration of the sample, in the presence of 5 mM ␤-mercaptoethanol at 37°C for 1 h immediately prior to sample application to the column (32, 33). Zinc and/or cadmium content of the resultant fractions was measured via ICP-AES.
DNA Transfection and Reporter Construct Assays-RSV-CAT and MRE-␤Geo (a gift of Richard Palmiter) plasmids were transiently transfected into cells via the DEAE-dextran method (34). After 3 days of growth, cells were harvested and split into samples for the reporter construct assays. The Hoechst 33258 staining assay was used to measure the amounts of DNA per plate of cells (7). ␤-Galactosidase activity from the MRE-␤Geo was measured as cleavage of an O-nitrophenyl ␤-galactopyranoside substrate and expressed as A 405 per g of DNA from the harvested cells. CAT activity from the RSV-CAT construct was measured as the percentage of acetylation of a radiolabeled chloramphenicol substrate (35).

LZA-LTK Ϫ Cells Maintain Basal Zinc Levels When Grown in Zinc-deficient Conditions and Are Efficient Scavengers of the
Metal-The ability of the LTK Ϫ and LZA-LTK Ϫ cell lines to accumulate Zn(II) was examined in response to the addition of a wide range of zinc concentrations in external media ([Zn] media ). LTK Ϫ cells and LZA-LTK Ϫ cells were seeded into unmanipulated media and zinc-stripped media (19), respectively, supplemented with various concentrations of ZnCl 2 . Following growth to near confluency, samples of the media were collected, and the cells were washed several times with PBS, which had been treated with Chelex to remove residual traces of zinc. Cells from each plate were harvested, counted, and then lysed in concentrated nitric acid. The amount of zinc per cell was determined at each [Zn] media by ICP-AES. Basal levels of intracellular zinc in the parental LTK Ϫ cells were measured to be 0.44 Ϯ 0.09 fmol zinc/cell in n ϭ 4 experiments (Fig. 1A). intracellular zinc concentration was observed in the presence of the maximum amount of zinc added into the medium (50 M [Zn] media ). Consistent with previous observations (19), LZA-LTK Ϫ cells grown in unsupplemented zinc-deficient medium maintained basal intracellular zinc levels (0.54 Ϯ 0.13 fmol zinc/cell n ϭ 4 experiments) similar to those measured in the LTK Ϫ cells (Fig. 1A). As [Zn] media was increased, however, the intracellular zinc levels within the LZA-LTK Ϫ cells increased substantially, reaching a maximal level of 8.68 Ϯ 0.86 fmol zinc/cell when [Zn] media was 3.5 M (a 16-fold increase). Additional zinc supplemented into the media, up to 50 M, resulted in no further increase in intracellular zinc levels.
The concentration of zinc remaining in the spent medium following growth was also measured by ICP-AES (Fig. 1B). The dashed line represents the concentration of zinc expected in spent media if the growth of the cells did not significantly perturb [Zn] media ; this was the case with the LTK Ϫ cell line. Intriguingly, no measurable zinc was observed in the spent media from LZA-LTK Ϫ cells grown with [Zn] media Ͻ 3.5 M. Furthermore, there was a striking relationship between the amount of intracellular zinc in LZA-LTK Ϫ cells and the concentration of extracellular zinc remaining in spent media (Fig.  1, A and B). Although measurable amounts of metal were added, zinc was undetectable in spent media until the intracellular zinc levels in the LZA-LTK Ϫ cells reached the "plateau" level of approximately 8.6 fmol zinc/cell. One interpretation of these data is that zinc-scavenging mechanisms are present in the LZA-LTK Ϫ cells that enhance uptake and intracellular sequestration of zinc from the extracellular medium and that this mechanism fills the zinc storage molecules with all available environmental zinc until these sites are saturated.
High Constitutive Levels of MT RNA Are Present in LZA-LTK Ϫ Cells at All [Zn] media -Intracellular metal concentrations of zinc or cadmium are often directly proportional to the amount of MT protein present within cells (36 -38). In the LTK Ϫ cells, the rate of increase in intracellular zinc levels as a function of [Zn] media is similar to previously published patterns for metal-induced MT gene transcription (39). However, the LZA-LTK Ϫ cells exhibit a strikingly different intracellular zinc profile in response to increased extracellular zinc. MT mRNA levels were therefore examined in both cell lines to determine what role, if any, MT plays in assisting the cell to scavenge zinc from the media. As expected from previous studies, MT mRNA levels rise in LTK Ϫ cells in response to elevated [Zn] media (Ͼ20 M) in a dose-dependent manner that is presumably mediated through transcriptional activation of the MREs ( Fig. 2A). studies were conducted. The high cysteine content of MT and relatively low molecular weight (ϳ10 kDa) make the protein readily detectable in [ 35 S]cysteine labeling experiments. LTK Ϫ and LZA-LTK Ϫ cells were grown in [Zn] media similar to those used in the RNA and intracellular zinc measurements, and [ 35 S]cysteine was added 24 h prior to harvesting to achieve steady-state labeling. Extracts were carboxymethylated to prevent oxidation of MT sulfhydryl groups (28), and equivalent amounts of protein were loaded and resolved on Tricine-based SDS-PAGE gels (29). In the parental LTK Ϫ cell line (Fig. 3A) (Fig. 1B). Grown at a wide range of [Zn] media, the amount of MT protein in cells grown in the lower zinc concentrations was relatively low. It is possible that differential translation rates of the MT mRNA might contribute to the differences in MT protein abundance in the LZA-LTK Ϫ cell line at the different [Zn] media conditions. However, it is more likely that differences in the rates of degradation of the apoprotein and the metal-bound protein account for the observed steadystate protein levels. Specifically, Klaassen et al. (40) have shown in proteolysis experiments performed in lysosomal extracts that apo-MT, MT complexed with Cd(II), or MT com-plexed with Zn(II) had rates of degradation of 50, 200, 35, and 20 pmol/mg of lysosomal protein/min, respectively. Thus, in the absence of enough saturating metal to stabilize the protein, it is expected that apo-MT will be rapidly proteolyzed in the LZA-LTK Ϫ cells.
To test whether restriction of zinc availability leads to rapid proteolytic turnover of MT protein, LZA-LTK Ϫ cells grown in 20 M [Zn] media and labeled for 24 h with [ 35 S]cysteine were exposed to the intracellular metal-ion chelator TPEN or the extracellular agents EDTA and EGTA for various lengths of times. Dose-dependent and time-dependent degradation of MT protein occurred in cells treated with the intracellular chelator TPEN (Fig. 4). The t 1/2 of MT in cells treated with TPEN was approximately 90 min, while the t 1/2 for MT in the untreated cells was greater than the 4-h experiment. Other labeled proteins were not affected by TPEN, suggesting that MT degradation is due to metal chelation and not general degradation of all proteins. EDTA and EGTA, unable to traverse the plasma membrane because of negatively charged moieties, had no effect on the labeled MT although both have high affinities for zinc.
The disruption of the plasma membrane by extrusion through polycarbonate membranes with pore sizes slightly smaller than the diameters of the cells. Latency, a measurement of the structural integrity of cellular organelles, was examined by ␤-hexosaminidase enzyme activity measurements on aliquots from the postnuclear supernatants (30). Only samples with greater than 90% latency were used in subsequent procedures to ensure that the distribution of zinc among the fractions was not altered by contamination from disrupted organelles. Additionally, a wide variety of buffers were utilized to ensure that zinc was not being chelated from the organelle fractions (data not shown). Subcellular fractionation by differential velocity centrifugation was used to partition the homogenate into four fractions (31). ICP-AES analyses ( Fig. 5A and Table I) demonstrated that the majority of the zinc taken up by the cells at increasing [Zn] media is found in the cytosol. Additionally, there were small, but measurable, increases in the other membranous compartments ( Table I).
The cytosolic fraction was further resolved by gel filtration chromatography. The majority of zinc in the cytosolic fraction was associated with a single peak that increased in parallel with [Zn] media (Fig. 5B) and eluted with an apparent molecular mass of 10,000 daltons. This peak coincides with the elution profile of purified cadmium-saturated MT (data not shown). The problems associated with the detection of MT by Western blot analysis have been well documented, and specific modifications have been suggested to improve the performance of these assays (41). Even employing these changes, we were still unable to detect MT by Western analysis in fractions obtained by gel filtration or from commercially available purified MT (data not shown). However, competitive displacement of MTbound zinc by Cd(II), which relies upon a rapid exchange of Cd(II) into the protein, has frequently been used to quantify MT (42). Aliquots of the cytosols were incubated with excess cadmium in the presence of thiols followed by gel filtration chromatography. Zinc was completely removed from each of the 10,000 molecular weight fractions and replaced with cadmium (Fig. 5C). The zinc displaced from MT (data not shown), along with excess cadmium, could be detected in the flow-through fractions (peaks 28 -33). These data indicate that the principle zinc-binding species in the 10,000 fraction is metallothionein.
MT mRNA Levels Do Not Rapidly Increase in LTK Ϫ Cells Transferred to Zinc-deficient Medium-It is possible that the LZA-LTK Ϫ cell line results from the selection of a subset of cells that grow in conditions of low zinc by increasing MT expression. Alternatively, higher MT mRNA and protein levels may result in most or all LTK Ϫ cells as a rapid response to conditions of zinc deprivation. A rapid induction of MT in response to zinc deficiency would suggest that overexpression of MT is a primary response to deal with the starvation conditions. If the levels of MT mRNA remain relatively constant in the initial hours of exposure to zinc deficiency, then the high constitutive levels of MT mRNA probably represent a long term adaptation of the LZA-LTK Ϫ cell line and suggest that other components of zinc homeostasis machinery are altered to respond to the zincdeficient conditions. Northern blot analysis was performed on mRNA from LTK Ϫ cells isolated after transfer of the cells into zinc-deficient medium for several days. Individual plates were harvested on each of the first 3 days. On the third day, an additional plate of the LTK Ϫ cells was passaged into several more plates containing zinc-deficient medium for time points extending to 124 h. No significant increase in MT mRNA occurred (Fig. 6), indicating that increased MT expression is not a simple adaptive response of LTK Ϫ cells to low zinc conditions and is likely to have arisen from a genetic event.
MT MRE Elements Are Not Constitutively Activated in LZA-LTK Ϫ Cells-Increased metallothionein gene expression in cell lines adapted to growth in media containing high concentrations of zinc and cadmium has been shown to result, in part, from increased binding of factors to the metal-responsive elements (43). Thus, we tested the possibility that the long term adaptation leading to metallothionein message elevation in LZA-LTK Ϫ cells may be the result of strong constitutive activation of the MT promoter through the metal-responsive elements. An MRE-driven ␤-galactosidase reporter construct (MRE-␤Geo) (7) was transiently transfected into the LTK Ϫ and LZA-LTK Ϫ cell lines, and ␤-galactosidase activity was examined as a function of increasing [Zn] media . Differences in transfection efficiencies between the two cell lines were controlled by cotransfection with a RSV-CAT construct, which is under the regulation of the strong, constitutively active Rous sarcoma virus long terminal repeat (44). In the parental LTK Ϫ cell line, zinc-dependent activation of the MRE-␤Geo construct (Fig. 7A) paralleled the Northern blot results in Fig. 2, indicating that the increase of metallothionein message in this case is probably mediated through the metal-responsive elements. In the LZA-LTK Ϫ cell line, however, there was no indication of ␤-galactosidase activity for any of the [Zn] media conditions (Fig. 7A). While CAT activity assays demonstrated that the LZA-LTK Ϫ cells had lowered transfection efficiencies relative to the LTK Ϫ cells (data not shown), it was not sufficiently decreased to account for the lack of ␤-galactosidase activity. At 50 M [Zn] media , LTK Ϫ cells possessed 100-fold more ␤-galactosidase activity than the LZA-LTK Ϫ cells grown at similar [Zn] media . The lack of responsive transfected MRE elements led us to examine other mechanisms for elevation of MT mRNA.
MT Gene Copy Number Is Amplified in LZA-LTK Ϫ Cells-Another mechanism that can lead to increased levels of specific

FIG. 4. Effect of chelators of extracellular and intracellular metals on MT levels in LZA-LTK ؊ cells grown in 20 M [Zn] media .
Following steady-state labeling with 10 Ci of [ 35 S]cysteine, cells were treated with chelators for the indicated times, the cells were harvested, and extracts were prepared. Following carboxymethylation, equivalent amounts of protein per lane were separated on a Tricine-based polyacrylamide gel. The gel was treated with En 3 hance prior to autoradiography. mRNAs, selective gene amplification, was also examined. Cell lines, including LTK Ϫ cells, have been shown to amplify the copy number of the MT locus in response to continuous exposure to high concentrations of cadmium or copper (27,(45)(46)(47)(48). Increased amounts of MT protein produced from the amplified genes provide a selective advantage to the cell by binding and sequestering toxic metals. To determine if the copy number of the MT-I gene had been increased in the LZA-LTK Ϫ cell line relative to the parental LTK Ϫ cells, Southern blot analysis was performed. Fig. 8  A, the amount of zinc within each fraction was determined by ICP-AES and normalized to a per cell basis. P1 is pellet from a low speed spin, typically containing nuclei, large membranes, and unbroken cells; P2 and P3 are pellets from supernatants obtained at increased spinning speeds, containing organelles, intracellular membranes, and vesicles. The supernatant obtained from the P3 spin is labeled cytosol. Cytosol fractions were split into two halves and were further separated by gel filtration. B, chromatographic profiles (Sephadex G-75) of the zinc content of one-half of the cytosolic fraction measured by ICP-AES. C, the other half of the cytosol was incubated with a 5-fold molar excess of cadmium (as compared with zinc) in the presence of thiols for 1 h and then separated on the G-75 Sephadex column. The resultant fractions were measured for cadmium content by ICP-AES. implications for metal-responsive regulation. If genomic amplification of the MT locus in the LZA-LTK Ϫ cell line leads to multiple endogenous MREs, these could sequester a limiting MRE-binding factor and thus prevent zinc-responsive expression from the transfected reporter construct (Fig. 7A). Alternatively, the overexpression of the metallothionein protein in the LZA-LTK Ϫ cell line may abrogate activation of the metal-responsive element regulatory system. Since the high constitutive levels of MT would immediately bind labile metal as it entered the cell, the availability of the cellular Zn(II) would be limited and perhaps prevented from reaching levels necessary to activate MRE-binding factors such as MTF-1. Experiments were conducted in which increasing amounts of the MRE-␤Geo plasmid were transfected into the parental LTK Ϫ and LZA-LTK Ϫ cell lines grown in 50 M [Zn] media . Increased ␤-galactosidase activity was observed as the amount of plasmid transfected into the LTK Ϫ cell line was increased (Fig. 7B), indicating that increased levels of MRE from the transfected constructs can recruit the requisite zinc-responsive activating factors. Conversely, there was a lack of increased ␤-galactosidase activity in LZA-LTK Ϫ cells grown in 50 M [Zn] media as the amount of transfected construct was increased. Although it is possible that the endogenous, amplified MREs within the LZA-LTK Ϫ cell line monopolize MRE-binding activation factors such as MTF-1 and thus deprive the transfected constructs, we consider it more likely that the overexpressed MT protein suppresses labile intracellular zinc to levels below the threshold needed to activate MTF-1. Supporting evidence for this conclusion comes from previous studies, which demonstrated that cell lines in which the MT locus had been amplified 10 -12-fold still retain the ability to induce metallothionein by heavy metals at the transcriptional level (49,50). Furthermore, transgenic mice that carry 56 additional copies of the mouse MT-I locus also retain the ability to induce MT protein levels in response to high levels of zinc and cadmium (51,52). The simplest interpretation is that the high constitutive levels of MT mRNA in the LZA-LTK Ϫ cells are not regulated at the level of MRE/ MTF-1-dependent transcription. LTK Ϫ Status of the LZA-LTK Ϫ Cells Does Not Contribute to Their Survivability in Zinc-deficient Conditions-Experiments were conducted to determine if the LTK Ϫ cells have a proliferative advantage in zinc-deficient media because of their independence from thymidine kinase, an enzyme that has been reported to be extremely sensitive to zinc-deficient conditions (53,54). The parental L cell line is dependent on this enzyme for efficient growth. Fresh aliquots of both the L-M(TK Ϫ ) (ATCC CCL 1.3) and L (ATCC CCL 1) cell lines were obtained through the American Type Culture Collection (Rockville, MD) and established as continuous cultures in unmanipulated media. Each cell line was then transferred into the zinc-stripped media (Ͻ60 nM zinc) and was passaged every fourth day in the media. Periodically, extracts generated from the cells were measured for intracellular zinc content. The levels of the metal immediately dropped to 0.1 fmol zinc/cell in each cell line within 24 h of being placed into the zinc-deficient media (Fig.  9). Unlike the previously cited attempts at establishing a cell line in zinc-deficient conditions, neither culture stopped proliferating, although the morphology of both cell lines did rapidly change to an elongated, spindly appearance as reported for the LTK Ϫ cells (19). Furthermore, both cell lines were able to regain the original level of 0.4 fmol zinc/cell after 36 days of continuous culture. These results indicate the thymidine kinase status of the LTK Ϫ cells does not affect their ability to proliferate in zinc-deficient media. These findings also demon- strate the reproducibility of the selection process for cells that grow in a low zinc medium.

DISCUSSION
Upon encountering conditions of extreme zinc deficiency, mouse fibroblasts undergo a series of changes that ultimately allow them to survive deprivation of this essential nutrient. One dramatic difference observed between the cells that have been subjected to prolonged zinc deficiency (LZA-LTK Ϫ ) and the parental LTK Ϫ cells is a large increase in basal MT mRNA levels. The opposite was anticipated, because MT gene transcription is normally induced by high, not low, concentrations of heavy metals and because the only known role attributed to MT is in metal ion detoxification. Furthermore, high levels of the cysteine-rich MT protein would be expected to compete with cellular zinc-binding molecules for the limited available supply of the metal. Clearly, MT is not simply sequestering or irreversibly tying up zinc.
Three key biochemical observations establish that MT protein participates in a zinc-scavenging system in the LZA-LTK Ϫ cell line. First, LZA-LTK Ϫ cells grown at the highest [Zn] media tested accumulate more than 6 times the amount of intracellular zinc than their LTK Ϫ counterparts (8. Recent genetic studies have provided complementary evidence that MT is able to exert a cytoprotective role in organisms exposed to zinc deficiency. Transgenic mouse embryos that express multiple copies of the MT gene demonstrated a greater resistance to developmental defects caused by zinc deficiency than did wild-type embryos (55). Furthermore, mouse strains in which the MT-I and MT-II alleles were disrupted demonstrated a much greater sensitivity to the phenotypic effects of zinc deficiency, such as malformed kidney structures, than their wild-type counterparts (56). These authors suggested that MT-bound zinc may serve as a reservoir that can be accessed and utilized at times of zinc deficiency. However, we propose that MT also plays a role in scavenging zinc from extracellular sources when intracellular supplies of this metal are low. Indeed, MT may bind zinc entering the cell and subsequently chaperone the metal to specific storage sites or proteins in a manner that is analogous to the protein metallochaperone, Atx1, which delivers copper to specific intracellular vesicle proteins (57). It is clear that zinc-laden MT can readily donate its metal cargo to restore the function of inactive apoenzymes in vitro, including alcohol dehydrogenase, aldolase, thermolysin, alkaline phosphatase, carbonic anhydrase, TFIIIA, and Sp1 (22, 58 -60). To date, however, there is no evidence for direct MT interaction with specific partners. This leaves open the question of a metallochaperone function for MT. and L (open triangles) cell cultures were established from frozen ATCC stocks in unmanipulated 10% calf serum/Dulbecco's modified Eagle's medium prior to exposure to the zinc-deficient media. During the course of the experiment, the cultures were passaged every fourth day. Periodically, the cells were harvested, counted, and measured for intracellular metal content by ICP-AES. S.D. was calculated from n ϭ 3 plates at each time point.
Amplification of the MT gene locus in the LZA-LTK Ϫ cell line explains, at least in part, the high levels of MT mRNA. Cell lines or organisms that have increased copy number of specific genes can be selected by environmental stress (61). Numerous studies have demonstrated that amplification of the MT gene locus in mammalian and yeast cells can be selected by continuous exposure to high concentrations of cadmium, copper, or metal-based drugs such as cisplatin (27,45,47,62). In these cases, cells containing the amplified genes produce elevated levels of MT protein compared with those that contain only a single copy of the gene. Under conditions of exposure to high levels of heavy metals, abundant MT provides a cytoprotective role by sequestering the potentially harmful ions from the rest of the cell. We propose that elevated levels of MT can also provide a zinc-scavenging mechanism and thus a selection for amplification of the MT gene in cells exposed to zinc-deficient conditions. Yet, mechanisms in addition to gene amplification cannot be ruled out to explain the high levels of MT mRNA in response to the zinc-deficient conditions. For instance, it has been recently demonstrated that interleukin-1 ␣-induced MT-I gene expression is markedly enhanced in rats fed zinc-restrictive diets (15). Additional processes may also participate in elevation of MT mRNA levels in the zinc-deprived cell lines.
A Model for the Zinc-scavenging Role of MT in Conditions of Extreme Zinc Deprivation-Regardless of the mechanism by which the LZA-LTK Ϫ cells accumulate high levels of MT mRNA, an increase in the steady state concentration of apo-MT is expected. The apoprotein is predicted to be rapidly proteolyzed in vivo when [Zn] media levels are insufficient to saturate the MT protein pool. Yet even at low [Zn] media , small "steadystate" levels of apo-MT may persist and bind zinc as it is transported in from the environment, trapping the ion within the cell. Since MT is being continuously produced from the large, constitutive MT mRNA pool in the LZA-LTK Ϫ cells, enough molecules of apo-MT are present at any given time in the cell to retain additional zinc as it becomes available (Fig.  10). Although the assiduous production and degradation of MT may seem like a futile and energetically unfavorable cycle, it apparently provides a mechanism by which the LZA-LTK Ϫ cells can efficiently scavenge zinc from the environment, thus permitting survival under conditions in which the availability of the ion is severely restricted.
This amplification of the MT gene locus is the result of a long term process that requires many cell doublings. Increased MT levels may not be utilized by a cell to deal with short term metal ion deficiencies. Indeed, no appreciable increase in MT mRNA levels was noted in the parental LTK Ϫ cells after transfer into zinc-deficient media, even over a period of several days. Since BALB/c 3T3 cells were unable to tolerate even one passage in the zinc-deficient medium without entering quiescence (19), the LTK Ϫ cell line may have characteristics in addition to the ability to amplify the MT gene locus that allow survival and that are not shared by all immortalized mouse cell lines. One difference between the BALB/c 3T3 and LTK Ϫ cell lines is that the latter lacks functional thymidine kinase, an enzyme that has previously been reported to be extremely sensitive to zincdeficient conditions (53,54). However, both L and L-M(TK Ϫ ) cell lines had similar intracellular zinc profiles when exposed to zinc-deficient conditions, suggesting that the ability to utilize functional thymidine kinase does not affect the ability of these two cell lines to proliferate in zinc-deficient media. Consistent with the requirement of events in addition to MT amplification, the constitutive overexpression of either MT-I or MT-III was insufficient for long term survival of baby hamster kidney cells, a line that also lacks thymidine kinase, under zinc-restrictive conditions (63). These results lead us to conclude that high levels of MT expression alone are not responsible for the robustness of the adapted L cells in response to zinc limitation.
Thus, in addition to increasing the steady-state concentration of apo-MT, it is likely that the LZA-LTK Ϫ cell line upregulates other components of the zinc homeostasis machinery (Fig. 10) in response to the zinc-deficient conditions. For example, a family of homologous mammalian proteins, designated as ZnT-1, ZnT-2, ZnT-3, and ZnT-4, thought to encode zinc transport proteins, has recently been identified (24, 64 -66). ZnT-1, the first member isolated, appears to be responsible for pumping excess zinc out (efflux) of the cell. ZnT-2 encodes a protein that sequesters zinc into intracellular storage compartments. However, none of these proteins appears to play an important role in shuttling zinc into the cell. Eide and co-workers (67,68) have recently identified two genes in Saccharomyces cerevisiae, designated as ZRT1 and ZRT2 (zinc-regulated transporters), which appear to encode zinc transporters of a high affinity and FIG. 10. Model of zinc ion homeostasis in the LZA-LTK ؊ cell line. Zinc ions are represented by the shaded circles, and apo-MT was drawn to resemble a cylinder on a revolver. The packed bundles represent transporter proteins that have been identified in mammalian or yeast cells (24,64,67,68). Black boxes represent putative interactions of zinc-loaded MT that may donate its cargo to nascent, zinc-requiring polypeptides or to cellular compartments responsible for the storage of loosely complexed zinc. See text for the discussion of the interactions between the components. low affinity uptake system, respectively, the former of which is up-regulated under conditions of zinc deficiency.
Zinc is an essential structural component in a wide variety of proteins, and therefore the maintenance of appropriate intracellular zinc levels is critical for cell survival. Although the exact function of the metallothionein proteins remains elusive (69), one proposed role for the protein is to maintain cellular homeostasis by metal ion detoxification, because transcription of MT can be induced by high levels of heavy metals. We have shown that the amplification of the MT gene locus can occur in response to conditions of extreme zinc deprivation. This finding represents the first example of increased MT expression in response to low concentrations of metals and implies that MT can act as a survival factor by assisting cells in sequestering limited supplies of available zinc from the environment. The derivation and characterization of this cell line should also be useful in identifying other components of the zinc homeostasis system.