In this report, we show that zinc is required for endoplasmic reticulum function in Saccharomyces cerevisiae. Zinc deficiency in this yeast induces the unfolded protein response (UPR), a system normally activated by unfolded ER proteins. Msc2, a member of the cation diffusion facilitator (CDF) family of metal ion transporters, was previously implicated in zinc homeostasis. Our results indicate that Msc2 is one route of zinc entry into the ER. Msc2 localizes to the ER when expressed at normal levels. UPR induction in low zinc is exacerbated in an msc2 mutant. Genetic and biochemical evidence indicates that this UPR induction is due to genuine ER dysfunction. Notably, we found that ER-associated protein degradation is defective in zinc-limited msc2 mutants. We also show that the vacuolar CDF proteins Zrc1 and Cot1 are other pathways of ER zinc acquisition. Finally, zinc deficiency up-regulates the mammalian ER stress response indicating a conserved requirement for zinc in ER function among eukaryotes.
Zinc is an important cofactor for many proteins that reside in or pass through organelles of the secretory system. For example, packaging of insulin in secretory granules of pancreatic β-cells is dependent on Zn2+ (Huang and Arvan, 1995; Dodson and Steiner, 1998). Matrix metalloproteases are Zn2+-requiring enzymes that are secreted by cells to modify the ECM during development and tumor cell metastasis (Chang and Werb, 2001). These proteins acquire zinc at an early point in the secretory pathway (Pei and Weiss, 1995; Kang et al., 2002). Addition of phosphoethanolamine groups to glycosylphosphatidylinositol (GPI) anchors requires Zn2+-dependent enzymes in the ER (Galperin and Jedrzejas, 2001; Mann and Sevlever, 2001; Sevlever et al., 2001). As a last example, some protein chaperones homologous to Escherichia coli DnaJ, like the ER-localized Scj1 of Saccharomyces cerevisiae, likely require Zn2+ to function in protein folding and in the degradation of misfolded proteins by quality control mechanisms (Silberstein et al., 1998; Nishikawa et al., 2001; Tang and Wang, 2001; Linke et al., 2003). These many requirements for Zn2+ in the secretory pathway highlight the need for zinc transporters to deliver the metal to those organelles. Despite the importance of these various zinc-dependent processes in normal cell function and disease states such as cancer, the transport proteins responsible for zinc delivery have not yet been identified.
Members of the cation diffusion facilitator (CDF) family of transport proteins are likely candidates to transport Zn2+ into the secretory pathway. CDF proteins play conserved roles in transporting Zn2+ from the cytosol into organelles or out of the cell in many organisms (Nies and Silver, 1995; Paulsen and Saier, 1997; Gaither and Eide, 2001; Palmiter and Huang, 2004). In E. coli, for example, the ZitB CDF protein transports excess zinc from the cytosol to the extracellular environment (Grass et al., 2001). In mammals, ZnT-1 similarly transports excess cytosolic zinc, whereas ZnT-2 may detoxify the metal by sequestering it in the late endosome (Palmiter and Findley, 1995; Palmiter et al., 1996; Kobayashi et al., 1999). Three mammalian CDF proteins implicated in Zn2+ transport into the secretory pathway are ZnT-5, ZnT-6, and ZnT-7, each of which having been localized to the Golgi apparatus (Huang et al., 2002; Kambe et al., 2002; Kirschke and Huang, 2003). The functional roles of these proteins are not yet clear.
We have learned much about zinc homeostasis from studies of the yeast S. cerevisiae. In this yeast, zinc uptake is mediated by the high affinity Zrt1 transporter and the lower affinity Zrt2 and Fet4 proteins (Zhao and Eide, 1996a, 1996b; Waters and Eide, 2002). The genes encoding these transporters are controlled by the Zap1 transcriptional activator protein (Zhao and Eide, 1997). Zap1 is active in zinc-limited cells and its activity is repressed in zinc-replete cells. Zap1 binds to one or more zinc-responsive elements (ZREs) in the promoters of its target genes (Zhao et al., 1998).
Zinc storage and detoxification in S. cerevisiae is mediated by the vacuole. The Zrc1 and Cot1 proteins, both members of the CDF family, are responsible for zinc transport into the vacuole (Kamizono et al., 1989; MacDiarmid et al., 2000, 2002; Miyabe et al., 2001). Two other CDF proteins, Mmt1 and Mmt2, have been implicated in iron transport in the mitochondria (Li and Kaplan, 1997). The fifth yeast CDF member, and the subject of this report, is Msc2. Msc2 was first identified in a screen for mutations with altered frequencies of meiotic sister chromatid exchange (Thompson and Stahl, 1999). However, the link between DNA recombination and the MSC2 gene is unclear; the effects on recombination are allele specific and not observed in a full msc2 deletion mutant (Thompson and Stahl, 1999). A more recent study of Msc2 suggested that it played some role in zinc metabolism (Li and Kaplan, 2000). An msc2 mutant grew poorly on respired carbon sources at elevated temperatures and had an abnormally large cell size. Both of these phenotypes were suppressed by addition of excess zinc. In this report, we demonstrate that Msc2 is an ER membrane protein whose role is to maintain proper function of the ER. Our results indicate that key processes in the ER require Zn2+ and that Msc2 is involved in supplying Zn2+ to this compartment.
Zinc deficiency induces the unfolded protein response (UPR)
Genome-wide cDNA microarray analysis of gene expression in S. cerevisiae indicated that many genes (458) are induced by zinc deficiency (Lyons et al., 2000). This included 46 direct Zap1 targets and 412 genes up-regulated in a Zap1-independent fashion. Induction of this latter group probably reflects indirect effects of zinc deficiency on cell function. Among the genes indirectly affected by zinc were several involved in the UPR. The UPR is a response to the accumulation of unfolded proteins in the ER (Patil and Walter, 2001). Fig. 1 A summarizes the microarray results for eight genes of the UPR (Travers et al., 2000) that were up-regulated by twofold or more in zinc-deficient wild-type cells. The up-regulation of these genes is independent of Zap1; each showed similar induction by zinc deficiency in a zap1 mutant. The UPR genes induced by zinc deficiency include ERO1, EUG1, and PDI1 (encoding proteins involved in protein disulfide isomerization) and FKB2 (encoding peptidyl-prolyl cis-trans isomerase). In addition, JEM1, KAR2, LHS1, and SCJ1 were also up-regulated. These genes encode ER chaperones that facilitate the translocation of misfolded ER proteins back to the cytosol for degradation in a process known as ER-associated degradation (ERAD).
These data indicated that zinc deficiency up-regulates the UPR. To confirm this hypothesis, we used a UPR-responsive reporter gene to assess the effect of zinc deficiency on UPR induction. UPR target genes contain UPRE sequences in their promoters that confer this regulation. A UPRE-lacZ reporter was highly induced in wild-type cells grown in low zinc (Fig. 1 B, LZM + 0.3 μM Zn) relative to zinc-replete cells (Fig. 1 B, LZM + 1,000 μM Zn). These experiments were performed using LZM medium which is zinc-limiting due to addition of 1 mM EDTA. Adding as little as 10 μM ZnCl2 to LZM was sufficient to repress UPRE-lacZ expression. UPRE-lacZ induction in low zinc was also observed in zrt1 mutants defective for zinc uptake. Because Zrt1 is the major zinc uptake transporter in yeast, the zrt1 mutant is zinc deficient when grown in LZM supplemented with ≤300 μM ZnCl2 (Zhao and Eide, 1996a). The UPRE-lacZ reporter showed high levels of β-galactosidase activity in the zrt1 mutant over a wide range of zinc concentrations and was suppressed only by adding 1,000 μM zinc to the media (Fig. 1 B). A HIS4-lacZ control reporter, which is not responsive to either zinc or unfolded proteins, showed high expression in both the wild-type and the zrt1 mutant strains at all zinc concentrations (Fig. 1 C). Therefore, the effects seen with the UPRE-lacZ reporter are specific UPR effects and not general effects on β-galactosidase activity. The twofold increase in UPRE-lacZ expression observed between 30 and 100 μM zinc (Fig. 1 A) is similar to that seen with other promoters (Fig. 1 C, HIS4-lacZ) and probably reflects a general decrease in expression in severely zinc-deficient cells.
Msc2 localizes to the ER
Induction of the UPR in low zinc suggested that zinc transport into the lumen of the ER is required for ER function. A transporter protein possibly involved in delivering zinc into the ER is encoded by the MSC2 gene. Previous studies suggested that Msc2 was localized to the endoplasmic reticulum. However, these experiments were done under conditions where Msc2 was overexpressed. Because overexpression can result in protein mislocalization, the true intracellular location of Msc2 was unclear. Therefore, we determined the localization of Msc2 when expressed from its own promoter on a low copy plasmid. To aid detection of Msc2, we generated an MSC2 allele with three HA tags fused to its COOH terminus. Immunoblots detected only a single band near the predicted molecular mass of Msc2 (unpublished data). Moreover, the epitope-tagged protein complemented the temperature-sensitive growth defect phenotype of an msc2 mutant strain indicating that it is functional.
Attempts to determine the subcellular localization of Msc2 using immunofluorescence microscopy were inconclusive because the level of expression was too low. Therefore, we used sucrose gradient fractionation to assess the distribution of Msc2. Protein extracts of wild-type cells expressing HA-tagged Msc2 were separated on sucrose gradients. After centrifugation, fractions were collected and analyzed by immunoblotting. Previous reports have shown that the presence or absence of Mg2+ greatly alters the position of ER vesicles in the gradient (Roberg et al., 1997). Without added Mg2+, the ER colocalizes with the Golgi apparatus in the middle fractions of the gradient. However, in the presence of Mg2+, the ER localizes to the heavier fractions of the gradient, colocalizing with the plasma membrane. This shift to heavier fractions is likely due to ribosomes remaining associated with the ER when Mg2+ is present. If Msc2 localizes to the ER, we predicted that the protein would show this Mg2+-dependent shift in these gradients.
Kex2 is a Golgi marker protein and its localization, peaking around fraction 6 in the middle of the gradients (Fig. 2), was unaffected by Mg2+. A plasma membrane protein, Pma1, also was largely unaffected by Mg2+ levels with its peak localization in the heaviest fractions of both gradients. Dpm1 served as an ER marker. Both Msc2 and Dpm1 showed the diagnostic ER Mg2+ shift being found in the heavier fractions with Mg2+ and localizing in the middle fractions without Mg2+. Because Msc2 colocalizes with an ER marker protein and shows the characteristic ER Mg2+ shift, these studies strongly support the localization of Msc2 to that compartment. In the presence of Mg2+, some Msc2 was also found in lighter fractions that may correspond to the Golgi. Little Dpm1 was found in these lighter fractions.
Mutation of MSC2 alters zinc homeostasis
If Msc2 transports zinc into the ER lumen, we predicted that mutation of the MSC2 gene would alter homeostasis of cytosolic labile zinc. A useful bioassay of this labile zinc is the Zap1 transcription factor, which binds to ZREs in it target promoters. Zap1 is active in zinc-limited cells and repressed by high zinc. The zinc response curve of a ZRE-lacZ reporter in wild-type cells is shown in Fig. 3 A. If Msc2 transported zinc into the ER, we predicted that cytosolic zinc levels would increase in an msc2 mutant strain and repress Zap1 activity. Consistent with this prediction, expression of the ZRE-lacZ reporter was suppressed in msc2 cells grown in LZM + 0.3–10 μM ZnCl2 (Fig. 3 A). Mutation of MSC2 did not decrease expression of the HIS4-lacZ reporter indicating that the effects of the msc2 mutation were specific to the ZRE-lacZ reporter (Fig. 3 B). These results suggest that Msc2 transports zinc from the cytosol into the ER lumen.
Effects of msc2 mutation on UPR induction
If Msc2 transports zinc into the ER, we predicted that an msc2 mutant would increase induction of the UPR in low zinc relative to wild-type cells. This prediction was confirmed using the UPRE-lacZ reporter (Fig. 4 A). Although the wild-type strain showed some induction of the UPR in low zinc, this up-regulation was substantially increased in the msc2 mutant strain. These data suggest that a strain lacking MSC2 experiences greater ER stress when coupled with zinc deficiency. To determine the metal specificity of this effect, we measured UPRE-lacZ expression in an msc2 mutant under low zinc conditions when higher levels of zinc or other metals were added. Although as little as 10 μM zinc suppressed the UPR up-regulation, neither 10 nor 100 μM of Fe2+, Cu2+, Cd2+, Mn2+, Ni2+, or Co2+ greatly suppressed this up-regulation (Fig. 4 B). Therefore, UPR induction in an msc2 mutant is a zinc-specific effect.
UPRE-lacZ induction by low zinc and msc2 mutation requires a functioning UPR signaling pathway
UPR induction is achieved through a two-protein signaling pathway (for review see Patil and Walter, 2001). Ire1 is a transmembrane endo-RNase protein that senses unfolded proteins in the ER. Hac1 is the transcription factor that responds to that signal to increase expression of UPR target genes. When Ire1 is activated by unfolded ER proteins, it cleaves an intron out of the HAC1 mRNA. The spliced HAC1 mRNA is translated into protein (Hac1i), which then up-regulates UPR target genes. To verify that the induction of the UPR in an msc2 mutant is dependent on the Ire1–Hac1 signaling pathway, we constructed msc2 ire1 and msc2 hac1 double mutants and analyzed their UPRE-lacZ activity. DTT, a known inducer of the UPR because of its ability to disrupt disulfide bonds in ER proteins, did not cause UPRE-lacZ induction in ire1 and hac1 mutants in both the wild-type and the msc2 mutant backgrounds (unpublished data). These mutants were also impaired for induction of the UPR in low zinc (Fig. 5 A). These strains had no such effect on the control HIS4-lacZ reporter (not depicted). Therefore, the induction of the UPR in low zinc in an msc2 mutant requires the full UPR signaling pathway.
One can also envisage that an msc2 mutant may somehow hyperactivate Hac1i independently of Ire1. To test this hypothesis, we introduced an intron-less form of the HAC1 gene on a low copy plasmid (pHac1i) into both wild-type and msc2 mutant strains. This plasmid-encoded Hac1i does not rely on Ire1 for its activation, and therefore causes constitutive UPR induction. We predicted that if the msc2 mutation made endogenous Hac1i more active, it would have a similar effect on plasmid-encoded Hac1i. This was not the case. Introduction of pHAC1i into wild-type cells actually caused higher levels of UPRE-lacZ activity than was observed in the msc2 mutant (Fig. 5 B). Although the reason for the decreased expression in the mutant is not yet known, this result clearly demonstrates that the msc2 mutation does not increase Hac1i activity independently of the UPR signaling pathway.
Genetic and biochemical evidence for ER dysfunction in msc2 mutants
The requirement for an intact Ire1–Hac1 signaling pathway to induce the UPR in response to low zinc and loss of Msc2 function suggested that these factors perturbed ER function. Further evidence supporting this conclusion came from the observation that msc2 ire1 and msc2 hac1 mutants exhibit a synthetic lethal growth phenotype. msc2 single mutants grow poorly at 37°C on YP medium supplemented with glycerol and ethanol, two nonfermentable carbon sources, but show no such defect on media supplemented with glucose (YPD), a fermentable carbon source (Fig. 6). Similarly, neither ire1 nor hac1 mutants exhibit a growth defect on YPD at either 30 or 37°C. Combining these mutations, i.e., msc2 ire1 and msc2 hac1, resulted in a strong growth defect at 37°C. (The few colonies seen in the msc2 hac1 mutant at 37°C may be the result of spontaneously arising suppressor mutations. More than 99% of the inoculated msc2 hac1 cells failed to grow at the elevated temperature.) Introducing the MSC2 or HAC1 genes on low copy plasmids (pMSC2, pHAC1) back into the msc2 hac1 mutant restored growth at 37°C. Thus, an msc2 mutant requires both Ire1 and Hac1 to survive at higher temperatures.
To further test the hypothesis that msc2 mutation and low zinc causes ER dysfunction, we examined one aspect of ER function, degradation of unfolded proteins through the ERAD pathway. This particular aspect was addressed because Scj1, an ER lumenal chaperone required for ERAD function, has two predicted zinc binding sites and is likely to require zinc for its activity. The homologous E. coli DnaJ protein was shown to be zinc dependent (Tang and Wang, 2001; Linke et al., 2003). A useful assay of the ERAD system is to monitor the degradation of CPY*, a mutant carboxypeptidase Y protein that fails to fold properly and is degraded by ERAD (Finger et al., 1993). Wild-type and msc2 mutants were transformed with a plasmid expressing CPY* bearing an HA epitope tag (Ng et al., 2000). The cells were grown in high or low zinc and then treated with cycloheximide to block further protein synthesis. Aliquots were removed periodically over 60 min for protein extraction and immunoblotting. No difference was observed between wild-type and msc2 mutants in either growth condition (unpublished data).
The UPR system induces expression of Scj1 and a second, partially redundant chaperone called Jem1 (Fig. 1; Nishikawa and Endo, 1997). Although Scj1 is probably zinc dependent, Jem1 lacks the zinc-binding sites and is therefore likely to be zinc independent. Therefore, increased expression of the Jem1 chaperone may compensate for the loss of zinc-dependent Scj1 activity. To test this hypothesis, we examined CPY* degradation in jem1 mutant cells lacking the zinc-independent chaperone. CPY* turnover in a jem1 single mutant was rapid in both zinc-limited and replete cells (Fig. 7). CPY* degradation in zinc-replete jem1 msc2 mutants was indistinguishable from the jem1 single mutant. However, the jem1 msc2 mutant showed a marked defect in CPY* degradation in low zinc. These studies indicate that zinc deficiency impairs function of components, likely Scj1, of the ERAD system. Consistent with this hypothesis, CPY* degradation was defective in a jem1 scj1 double mutant at both zinc levels (Fig. 7).
Zrc1 and Cot1 also contribute to ER zinc
The ability of elevated zinc to suppress UPRE-lacZ induction in an msc2 mutant (Fig. 4 A) indicated that other transporters in addition to Msc2 may also provide zinc to the ER. Two possible candidates for this role are Zrc1 and Cot1, members of the CDF protein family that transport zinc into the vacuole. Although the steady-state location of these proteins is the vacuole membrane, they must transit through the ER en route to that compartment. Therefore, we tested the contribution of Zrc1 and Cot1 to ER function using the UPRE-lacZ reporter. Consistent with Zrc1 and/or Cot1 also contributing to ER zinc, induction of the UPRE-lacZ reporter was greatly increased in a zinc-deficient msc2 zrc1 cot1 triple mutant relative to an msc2 mutant or wild-type cells (Fig. 8 A). To determine if either or both Zrc1 and Cot1 were involved, we generated a series of single and double mutants and assayed their effects on UPRE-lacZ induction in low zinc and high zinc. In high zinc, all strains tested exhibited no UPRE-lacZ induction (Fig. 8 B). Under zinc-deficient conditions, zrc1, cot1, and zrc1 cot1 cells showed UPR induction indistinguishable from wild-type cells. Zinc-deficient msc2 mutant cells had increased UPRE-lacZ expression that was further increased in the msc2 zrc1 and msc2 cot1 mutants. Finally, UPRE-lacZ induction was highest in the msc2 zrc1 cot1 mutant. These results suggested that all three transporters contribute to ER function.
Zinc deficiency induces the UPR in mammalian cells
Finally, we addressed if the effects of zinc deficiency on the UPR was evolutionarily conserved. In mammals, ATF6 is a member of the basic-leucine zipper family of transcription factors that acts analogously to yeast Hac1 to control the UPR (for review see Harding et al., 2002). ATF6 contains a transmembrane domain and is localized to the ER membrane. Upon ER stress, ATF6 is proteolytically cleaved to release the NH2-terminal b-Zip–containing domain which then translocates into the nucleus to activate UPR gene expression. To address if zinc deficiency activates the UPR in mammalian cells, we examined the effects of zinc deficiency on expression of an ATF6-responsive reporter plasmid (p5xATF6GL3; Wang et al., 2000) transiently transfected into HeLa cells. p5xATF6GL3 contains five copies of the ATF6 binding site inserted upstream of the firefly luciferase gene. These cells were cotransfected with a lacZ gene driven by the SV40 promoter to normalize for differences in transfection efficiency. As shown in Fig. 9 A, treating HeLa cells bearing p5xATF6GL3 with tunicamycin, a glycosylation inhibitor and known inducer of ATF6 activity, resulted in a marked induction of luciferase expression. To induce zinc deficiency, cells were treated with 5 μM of the zinc chelator N,N,N′,N′-tetrakis-(2-pyridyl-methyl)ethylenediamine (TPEN). Cells were then harvested over a 12-h period and assayed for luciferase expression. After 5 h of TPEN treatment, a small increase in luciferase activity was observed (Fig. 9 B). After 12 h, luciferase activity was induced approximately fourfold. No loss of cell viability was detected over this 12-h period (unpublished data) indicating that the induction was not due to necrosis or apoptosis. To determine the concentration of TPEN optimal to triggering UPR induction, we assayed the effects of a range of TPEN concentrations after a 12-h treatment. TPEN concentrations of ≤ 3 μM had no effect on ATF6 activity whereas concentrations of 4 or 5 μM induced expression above basal levels (Fig. 9 C). Finally, we addressed if the effects of TPEN on UPR induction were due to zinc deficiency or a pharmacological effect of TPEN. Consistent with zinc deficiency being the cause, simultaneous incubation of cells with 5 μM TPEN and 5 μM ZnCl2 failed to induce ATF6 activity (Fig. 9 D). These results suggest that the requirement of the ER for zinc is evolutionarily conserved and that zinc deficiency induces the UPR in a wide range of organisms.
In a previous report, Li and Kaplan (2000) proposed that Msc2 transports zinc out of the nucleus into the cytosol. This hypothesis was based on the tentative localization of overexpressed Msc2 to the nuclear envelope (which is contiguous with the ER membrane) and the observation that the Zap1 transcription factor was completely repressed under their zinc-deficient conditions in an msc2 mutant. We questioned this hypothesis because the relatively large size of the nuclear pore is unlikely to limit diffusion of Zn2+ ions between the cytosol and the nucleus (Shulga et al., 2000). In this report, we propose the alternative model that zinc is required for ER function and Msc2 transports the metal into the ER lumen.
Several lines of evidence support the hypothesis that Msc2 is involved in zinc metabolism. First, Msc2 is a member of the CDF family of metal ion transporters. Studies of CDF proteins in many organisms have established that these proteins transport their metal ion substrate, usually Zn2+, from the cytosol to either outside of the cell or into organelles (Nies and Silver, 1995; Paulsen and Saier, 1997; Gaither and Eide, 2001; Palmiter and Huang, 2004). Thus, Msc2 is likely to be a zinc transporter whose topology is consistent with metal transport into the ER lumen. Second, mutations in msc2 alter zinc homeostasis. Although it was shown previously that the Zap1 transcription factor was completely inactive in a zinc-limited msc2 mutant (Li and Kaplan, 2000), our experiments further clarified these earlier results by showing that Zap1 can still respond to zinc deficiency in an msc2 strain. Nonetheless, the suppression of ZRE-lacZ expression caused by the msc2 mutation that we observed (Fig. 3) is consistent with msc2 mutants having elevated pools of labile cytosolic zinc. Third, mutations in MSC2 result in several phenotypes that are zinc suppressible. Li and Kaplan (2000) had shown that the growth defect of msc2 mutants on respired carbon sources at elevated temperatures is suppressed by zinc. Here, we document that zinc-deficient cells up-regulate their UPR system and this induction is further increased in an msc2 mutant. Like the temperature-sensitive growth phenotype, induction of the UPR is zinc suppressible. Finally, we present here genetic evidence that known zinc transporters, Zrc1 and Cot1, play redundant roles with Msc2. Their involvement in ER function indicates that Msc2 is also a zinc transporter.
Our results indicate that Msc2 delivers zinc to the lumen of the ER to maintain function of that compartment and, perhaps, other organelles of the secretory pathway. Although currently available methods do not allow us to assay lumenal ER zinc directly, several observations indirectly support this hypothesis. First, we found that Msc2 localizes to the ER when expressed at physiological levels from its own promoter. The presence of significant amounts of Msc2 protein in vesicles of lighter density (Fig. 2 A, +Mg2+) suggests that some protein may also be found in later compartments of the secretory pathway, perhaps the Golgi. Second, mutation of MSC2 activates the UPR system. Our results indicate that this induction requires the full UPR signaling pathway consistent with actual ER stress being responsible rather than downstream perturbations of, for example, Hac1i activity. Third, we found that msc2 mutants show synthetic growth defects when combined with mutations in either ire1 or hac1. This observation is especially intriguing in light of studies by Ng et al. (2000) in which mutations were identified that were synthetically lethal when combined with an ire1 mutation. This screen identified 16 different genes almost all of which were shown to be involved in ER functions such as glycosylation, GPI anchor synthesis, or ERAD function. Thus, mutations in genes affecting lumenal ER processing events are lethal when combined with ire1 mutations. Although msc2 mutants were not identified in this synthetic lethal screen, they too are synthetically lethal with an ire1 mutation when cells are grown at 37°C. Finally, we directly observed zinc-suppressible defects in one aspect of ER function, ERAD. We chose to examine this process because Scj1, a chaperone protein of the ER lumen that is required for ERAD, is likely to be zinc dependent. One caveat to this experiment is that the E3 ubiquitin-protein ligases required for proteasome degradation of unfolded proteins after their export from the ER, Hrd1/Der3, and Doa10/Ssm4, may also be zinc dependent (Bordallo and Wolf 1999; Swanson et al., 2001). Therefore, the defects in ERAD observed in msc2 mutants could be due to disruption of ligase function via alterations in cytosolic zinc homeostasis. This appears not to be the case; degradation of a model cytosolic substrate of Doa10, the Deg1-β-galactosidase protein (Swanson et al., 2001), was unaffected by either zinc limitation or mutation of msc2 (unpublished data). Compromised function of Hrd1/Der3 is still formally possible but less likely given the lack of effects on Doa10 function.
In addition to ERAD, it is also likely that other processes occurring in the ER are impaired by zinc deficiency. Several observations suggest that synthesis of GPI anchors may be disrupted by these perturbations. First, GPI anchor synthesis has been found to be zinc dependent both in vitro and in vivo (Mann and Sevlever, 2001; Sevlever et al., 2001). Second, the yeast MCD4, LAS21, and GPI13 genes encode related proteins required for GPI anchor synthesis (Gaynor et al., 1999; Tohe and Oguchi, 1999; Flury et al., 2000; Taron et al., 2000). These proteins all contain conserved domains similar to the zinc-binding sites of alkaline phosphatases (Galperin and Jedrzejas, 2001). Although the precise role of these proteins is still unclear, their importance in GPI anchor synthesis is well documented. MCD4 and GPI13 are essential genes whereas LAS21 is not. Temperature-sensitive mcd4 alleles and las21 mutants have a large cell morphology similar to that seen with msc2 (Mondesert et al., 1997; Li and Kaplan, 2000; Ni and Snyder, 2001). Suppressors of these mutations also link their activity to Msc2. The large cell phenotype of las21 is suppressed by overexpression of the HSP150 gene (Tohe and Oguchi, 1999). Hsp150 is a cell wall protein of unknown function. We have found HSP150 overexpression also suppresses the msc2 temperature-sensitive growth defect and large cell phenotype (unpublished data). Finally, MCD4 is a direct Zap1 target gene and is induced greater than or equal to fourfold by zinc deficiency (Lyons et al., 2000). These results suggest that GPI anchor synthesis is sensitive to zinc deficiency, and MCD4 is up-regulated to maintain sufficient activity in zinc-limited cells. Golgi function was unimpaired by mutation of the MSC2 gene; analysis of the kinetics of wild-type CPY processing indicated no differences between wild-type and mutant cells in high or low zinc (unpublished data).
In this work, we have identified three different transporters that likely contribute to ER zinc: Zrc1, Cot1, and Msc2. Of these three, Msc2 appears to play the predominant role because mutation of this gene alone had the strongest effect on UPR induction. Although Msc2 is resident in the ER membrane, Zrc1 and Cot1 are most abundant in the vacuolar membrane (Li and Kaplan, 1998; MacDiarmid et al., 2002). We can suggest two possible mechanisms to explain how Zrc1 and Cot1 could supply zinc to the ER. First, these proteins may mediate zinc transport soon after insertion into the ER membrane and before their transit to the vacuole. Alternatively, zinc may be transferred from the vacuole lumen to the ER by retrograde vesicular trafficking.
The ZRC1 gene is a Zap1 target and induced by zinc deficiency (Lyons et al., 2000; Miyabe et al., 2000). This was a surprising finding given the importance of this transporter in zinc storage and detoxification. Our previous results indicated that ZRC1 up-regulation in low zinc is required to tolerate “zinc shock” (MacDiarmid et al., 2003). Zinc shock occurs when zinc-limited cells, which express high levels of the Zrt1 zinc uptake transporter, are resupplied with zinc. The role of Zrc1 in supplying zinc to the ER is an additional reason why the ZRC1 gene may be induced in low zinc; i.e., to maintain ER zinc levels. However, given the relatively minor role Zrc1 plays in maintaining ER function (Fig. 8), our results argue that zinc shock tolerance is the major reason for the regulation of ZRC1 expression by Zap1.
We also predict that additional pathways contribute to ER zinc. UPR induction in the msc2 zrc1 cot1 triple mutant is still suppressible by adding 10 μM ZnCl2 to the medium. One possible route to bypass the loss of Msc2, Zrc1, and Cot1 activity is via fluid-phase endocytosis of zinc followed by its retrograde vesicular transport to the ER. Although formally possible, we do not favor this model based on our results. Specifically, although UPR induction in the msc2 zrc1 cot1 mutant is suppressed by 10 μM zinc, it requires 100-fold more zinc to suppress UPRE-lacZ activity in a zrt1 mutant (Fig. 1 A). Zrt1 transports zinc across the plasma membrane into the cytoplasm. Therefore, the much higher levels of zinc required to suppress UPRE-lacZ expression in a zrt1 mutant strongly argues that zinc must pass through the cytosol before entering the ER.
Msc2 is related to the Zhf protein of S. pombe (Borrelly et al., 2002; Clemens et al., 2002). Like Msc2, Zhf is a member of the CDF family of metal ion transporters. Furthermore, immunoelectron microscopy localized Zhf protein to the ER membrane. However, the phenotypic effects of zhf mutations argue that this protein plays a very different role in zinc metabolism. Although Msc2 is important for supplying zinc to the ER for organelle function, Zhf appears to be required for zinc storage and detoxification. For example, zhf mutations increase the sensitivity of cells to exogenous zinc indicating its role in detoxification. zhf mutants also have decreased zinc accumulation in cells indicating its role in zinc storage. For Msc2, we found no effect of msc2 mutations on zinc tolerance either in wild-type or zrc1 cot1 mutant cells that are greatly sensitized to exogenous zinc (unpublished data). Li and Kaplan (2000) showed that msc2 mutants actually hyperaccumulate zinc. Thus, the effects of zhf mutations in S. pombe are much more similar to mutations altering Zrc1 and Cot1 of S. cerevisiae. The ER of S. pombe may play a role in zinc storage and detoxification similar to that of the vacuole in S. cerevisiae.
Another protein related to Msc2 is mammalian ZnT-5. Although most members of the CDF family have only six transmembrane domains, Msc2 and ZnT-5 are predicted to have ∼15 transmembrane domains. For each, the conserved CDF region is found at the COOH terminus with several transmembrane domains attached to their NH2 termini. ZnT-5 was localized to the Golgi when expressed from the CMV promoter in HeLa cells and is widely expressed in mammalian tissues (Kambe et al., 2002). Expression was especially high in the β cells of the pancreas in which zinc is transported into secretory vesicles for the packaging of insulin. Finally, ZnT-5–dependent zinc transport activity could be observed in Golgi-derived vesicles in vitro. Given our results indicating that zinc transport into the secretory pathway is important for ER stress and UPR induction, we predict that ZnT-5 or related proteins carry out this important role in mammalian cells. Consistent with this hypothesis, we have found that expression of ZnT-5 in msc2 mutant yeast can suppress phenotypic defects of this mutant under certain conditions (unpublished data).
Materials And Methods
Yeast strains and growth conditions
Media used were YPD, SD, YPGE, and LZM as described previously (Gitan et al., 1998). Strains used are described in Table I. Yeast deletion mutants were obtained from Research Genetics. To generate several strains, the KanMX cassette with 500 bp flanking the gene to be mutated was amplified by PCR from the appropriate mutant. The PCR fragment was then transformed into the appropriate recipient strain. This method was used to generate strains CEY3-8, CEY27-29, CM158, CM160, CM162, and CM164. CM158 and CM164 were crossed to generate CEY25 and CM160 and CM162 were crossed to generate CEY26. CEY13 and CEY17 are haploid segregants of CEY25 and 26, respectively. CEY8 and DY150 were crossed to generate CEY24. CEY19-23 and CEY27-29 are haploid segregants of CEY24.
pDg2L (ZRE-lacZ; MacDiarmid et al., 2000), pHYC3 (HIS4-lacZ; Hinnebusch et al., 1985), pMCZ-Y (UPRE-lacZ; provided by A. Cooper, University of Missouri-Kansas City; Kansas City, MO; Kawahara et al., 1997) were described previously. pSN222 (Kex2-HA) was a gift of S. Nothwehr (University of Missouri). pRK315 (Pma1-HA) was a gift of R. Kölling (Heinrich-Heine-Universität, Düsseldorf, Germany). pDN436 (CPY*–HA) was a gift of D. Ng (Pennsylvania State University, University Park, PA). pGEV-TRP was cotransformed with pRK315 so that Pma1-HA expression levels could be controlled by β-estradiol (10−7M; Sigma-Aldrich) addition (Gao and Pinkham, 2000). pMCZ-YL was constructed by exchanging the URA3 gene on pMCZ-Y with the LEU2 gene by marker swapping (Cross, 1997). pDN436U was constructed by exchanging the LEU2 gene on pDN436 with the URA3 gene (Cross, 1997). To generate pMSC2, the MSC2 ORF plus ∼500 bp of flanking DNA was amplified by PCR from genomic DNA and inserted into EcoRI-, PstI-digested pFL38 by homologous recombination (Ma et al., 1987). Msc2 function was confirmed by complementing the msc2 mutant. To generate pMSC2HA, the MSC2 ORF plus ∼500 bp promoter and lacking the stop codon was PCR amplified from genomic DNA and inserted into YCpZRC1-HA (MacDiarmid et al., 2002) by homologous recombination. The resulting plasmid has the MSC2 promoter and ORF followed by three HA tags. The plasmid was confirmed by complementing the msc2 mutant. To generate pHAC1, the HAC1 ORF plus ∼1,300 bp upstream and ∼900 bp downstream flanking DNA were amplified from genomic DNA and inserted into SacI-, XmaI-digested pRS315 via homologous recombination. To generate pHAC1i, overlap PCR was used to generate the HAC1 ORF (plus flanking DNA as above) without the intron (Ho et al., 1989). This fragment was then inserted into pRS315 via homologous recombination. DNA sequencing confirmed the lack of the intron.
Yeast β-galactosidase assays
β-Galactosidase assays were performed on protein extracts (Ausubel et al., 1995) and specific activity was normalized to protein content.
Subcellular fractionation and immunoblotting
Subcellular fractionation was done largely as described by Roberg et al. (1997). Lysates were fractionated on 20–60% sucrose gradients prepared with 10 mM EDTA (−Mg) or with 2 mM MgSO4 (+Mg). Immunoblots were done by standard techniques (Harlow and Lane, 1988). Blots were visualized with ECL (Amersham Biosciences), and band quantitation was performed using NIH Image 1.61. Antibodies used were mouse anti-HA (12CA5, Roche), rabbit anti-HA (Sigma-Aldrich), mouse anti-Dpm1 (Molecular Probes), mouse anti-Pgk1 (Molecular Probes), goat anti–mouse HRP-conjugated secondary (Pierce Chemical Co.), and goat anti–rabbit HRP-conjugated secondary (Pierce Chemical Co.).
Assay of ERAD
Yeast were grown in 200-ml cultures to an OD600 = ∼0.5. Cells were harvested and resuspended in 100 ml of fresh media to a final OD600 = ∼1.0. Cells were grown 30 min, then cycloheximide (Sigma-Aldrich) was added to a final concentration of 100 μg/ml. 5 ml aliquots of cells were removed at each time point to tubes containing NaN3 to a final concentration of 10 mM. After the last time point, the cells were collected by centrifugation and washed once with cold buffer containing 10 mM NaN3, 1 mM EDTA. Cells were resuspended in 1 ml of the same buffer and transferred to microfuge tubes. Cells were pelleted and resuspended in 200 μl cold protein extraction buffer (10 mM Tris-Cl, pH 8, 25 mM ammonium acetate, 1 mM EDTA, 1 mM PMSF, 10% trichloroacetic acid [Sigma-Aldrich], yeast proteinase inhibitor cocktail [complete mini EDTA-free pellets; Roche]). An equal volume of glass beads was added and the tubes were vortexed five times for 1 min, with 1 min on ice between pulses. Lysates were transferred to fresh tubes. Another 500 μl of protein extraction buffer was added to the glass beads, vortexed 1 min, and then pooled with the previous lysates. Lysates were centrifuged at 14,000 g at 4°C for 10 min. The supernatant was removed and discarded. The pellets were resuspended in 120 μl buffer I (100 mM Tris base, 3% SDS, 1 mM PMSF), then boiled for 5 min. Insoluble debris was pelleted by centrifuging 5 min at 15,800 g. The supernatant was transferred to new tubes and protein concentrations were measured using the DC protein kit (Bio-Rad Laboratories). 10 μg of protein were loaded per lane when analyzed by immunoblotting.
Cell culture, transient transfection, mammalian plasmids, and assays
HeLa cells were cultured in DME (Invitrogen) plus 0.45% glucose under 5% CO2. All media contained 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, and 100 μM MEM nonessential amino acids (Invitrogen) supplemented with 10% FBS (Invitrogen). p5xATF6GL3 was a gift of R. Prywes (Columbia University, New York, NY), and contains five repeats of an ATF6 binding site cloned into a minimal promoter preceding the firefly luciferase gene (Wang et al., 2000). pSV-β-galactosidase control vector (Promega) was used as a control for transfection efficiency. HeLa cells (∼1.2 × 106) were seeded in 60-mm plates and transiently transfected using Lipofectamine 2000 (Invitrogen). Both p5xATF6GL3 and pSV-β-galactosidase plasmids were cotransfected in all experiments. Transfection efficiencies were typically 60%. 36–48 h after transfection, tunicamycin (Sigma-Aldrich), ZnCl2, or TPEN (Sigma-Aldrich) was added to the culture media at the indicated concentrations. After treatment, the cells were washed three times with cold PBS. To generate protein extracts and perform luciferase and β-galactosidase assays, the luciferase assay system with reporter lysis buffer (Promega) was used. Luciferase activity was normalized to β-galactosidase activity as an internal control.
The authors thank J. Kaplan (Table I, strains DY150 and DY150msc2; University of Utah, Salt Lake City, Utah), A. Cooper, S. Nothwehr, R. Prywes, and D. Ng for providing strains and/or plasmids and W. Folk for use of his luminometer.
This work was supported by National Institutes of Health grants GM56285 and GM69786. C.D. Ellis was supported by an MU Life Sciences Predoctoral Fellowship.
C.W. MacDiarmid's present address is R424 Research Building, Dept. of Biology, University of Missouri-St. Louis, One University Blvd., St. Louis, MO 63121-4499.
S. Clark's present address is F032A, 800 Hospital Dr., Dept. of Medicine-Endocrinology University of Missouri-Columbia, Columbia, MO 65211.
T. Lyons' present address is 404 Leigh Hall, Dept. of Chemistry, University of Florida, 127 Chemistry Research Building, P.O. Box 117200, Gainesville, FL 32611-7200.
Abbreviations used in this paper: CDF, cation diffusion facilitator; ERAD, ER-associated degradation; GPI, glycosylphosphatidylinositol; TPEN, N,N,N′,N′-tetrakis-(2-pyridyl-methyl)ethylenediamine; UPR, unfolded protein response; ZRE, zinc-responsive element.