Structural and Functional Studies of the Amino Terminus of Yeast Metallothionein*

Purified yeast copper-metallothionein lacks 8 amino-terminal residues that are predicted from the DNA sequence of its gene. The removed sequence is unusual for metallothionein in its high content of hydrophobic and aromatic residues and its similarity to mitochondrial leader sequences. To study the significance of this amino-terminal cleavage, several muta- tions were introduced into the metallothionein coding gene, CUPl. One mutant, which deletes amino acid residues 2-8, had a minor effect on the ability of the molecule to confer copper resistance to yeast but did not affect CUPl gene regulation. A second mutation, which changes two amino acids adjacent to the cleav- age site, blocked removal of the extension peptide but had no effect on copper detoxification or gene regula- tion. Immunofluorescence studies showed that both the wild-type and these two mutant proteins are predomi- nantly cytoplasmic with no evidence for mitochondrial localization. The cleavage site mutation allowed isola- tion and structural characterization of a full length metallothionein polypeptide. The copper content and luminescent properties of this molecule were identical to those of the truncated wild-type protein indicating a homologous cluster structure. Moreover, the amino-terminal peptide was selectively removed by various endopeptidases and an exopeptidase suggesting that it does not participate in the tertiary fold. These results argue that the amino-terminal peptide is not required against prote- olysis. In this study 15 pg of protein were reconstituted anaerobically with increasing mol eq of Cu(1). The metal-protein mixture was neutralized to pH 7 with Hepes-C1 and subsequently incubated with 0.5 pg of proteinase K for 1.5 h at 37 "C. The fluorescamine reactivity was monitored for each sample (8).

Purified yeast copper-metallothionein lacks 8 amino-terminal residues that are predicted from the DNA sequence of its gene. The removed sequence is unusual for metallothionein in its high content of hydrophobic and aromatic residues and its similarity to mitochondrial leader sequences. To study the significance of this amino-terminal cleavage, several mutations were introduced into the metallothionein coding gene, CUPl. One mutant, which deletes amino acid residues 2-8, had a minor effect on the ability of the molecule to confer copper resistance to yeast but did not affect CUPl gene regulation. A second mutation, which changes two amino acids adjacent to the cleavage site, blocked removal of the extension peptide but had no effect on copper detoxification or gene regulation. Immunofluorescence studies showed that both the wild-type and these two mutant proteins are predominantly cytoplasmic with no evidence for mitochondrial localization. The cleavage site mutation allowed isolation and structural characterization of a full length metallothionein polypeptide. The copper content and luminescent properties of this molecule were identical to those of the truncated wild-type protein indicating a homologous cluster structure. Moreover, the aminoterminal peptide was selectively removed by various endopeptidases and an exopeptidase suggesting that it does not participate in the tertiary fold. These results argue that the amino-terminal peptide is not required for either the structural integrity or biological function of yeast metallothionein.
The CUPl locus of the yeast Saccharomyces cereuisiae encodes a small, cysteine-rich copper-binding peptide that belongs to the metallothionein super-family of proteins (1). Genetic studies have shown that the CUPl gene product plays two related functional roles in yeast: at high copper concentrations it protects the organism against the harmful effects of this ion, and at low physiological concentrations of copper it represses basal transcription from the CUPl promoter (2)(3)(4). Purification of the product of the CUPl locus revealed that the amino acid sequence was precisely that predicted by the DNA sequence of the gene except that the first eight amino acids were lacking. Even when the protein is purified in the presence of proteolytic inhibitors or from a pep4 yeast * The costa of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by United States Public Health Service National Research Science Award 1 F32 GM10958-01 from the National Institute of General Medical Sciences. ** Acknowledges the support of Grants ES 03817 and ES00147 from the National Institutes of Health. strain deficient in vacuolar proteases, the amino terminus starts at Gln' of the coding sequence.
The isolated 53-residue yeast metallothionein bound 8 mol of eq Cu(1) or 4 mol of eq Cd(I1) via the 12 cysteinyl thiolates in the molecule (5). This differs from mammalian metallothionein which binds either 12 Cu(1) or 7 Cd(I1) (or related metal ions) per molecule with 20 cysteinyl thiolates serving as ligands (6)(7)(8). The structure of rat Cd, Zn-metallothionein, deduced by x-ray crystallography, consists of two domains with each polypeptide segment wrapped around a separate metal-thiolate cluster (9). The clusters replace the usual hydrophobic core in typical globular proteins. While there is a paucity of structural data on the yeast metallothionein, by analogy one would predict a similar structural motif of a polypeptide monolayer enfolding a Cu-thiolate polynuclear cluster.
The removal of the amino-terminal 8-residue segment of yeast metallothionein is highly unusual in that no other metallothionein from fungi, invertebrates, or vertebrates is known to undergo amino-terminal cleavage. Furthermore, the composition of the amino-terminal peptide is not characteristic of metallothionein in that it is highly hydrophobic and includes two phenylalanines, an aromatic residue not found in any other metallothionein. It is possible that the cleavage of yeast metallothionein is an artifact of isolation resulting from exposure to proteases. For example, the amino-terminal Met from aldolase is known to be susceptible to removal by an endogenous protease during purification (10). Alternatively, cleavage of the amino-terminal segment might be a physiological processing event such as occurs during zymogen activation, signal peptide cleavage for secretory molecules, and removal of initiator methionine.
In order to understand the significance of the amino-terminal cleavage event, we prepared four metallothionein mutants by site-specific mutagenesis of the CUPl gene within the amino-terminal region. Two of these mutations were characterized in detail. The first of these is a protein in which the NH2-terminal methionine is fused directly to Gln', producing a yeast strain in which the intervening amino acids are never present in the cell. The second is a protein containing a double codon substitution which blocks cleavage and allows the full length 61-residue peptide to be isolated from yeast cells. We report here the biological activity of these two mutant metallothioneins as well as the isolation and properties of the 61-residue polypeptide.

EXPERIMENTAL PROCEDURES'
Portions of this paper (including "Experimental Procedures" and Tables V and VI) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-3944, cite the authors, and include a check or money order for $2.40 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.  Table V in Miniprint section.

RESULTS
Generation of Mutant Metallothioneins-In order to study the function of the amino terminus of yeast metallothionein, four different mutations were introduced into the CUPl gene by oligonucleotide-directed mutagenesis ( Table I). The first is a mutation, designated by A N , which joins the NH2-terminal methionine to Gln' creating a deletion of seven amino acids in the NH2 terminus of the protein. The second mutation, denoted the Barn mutant because the base changes which lead to the amino acid substitutions create a BarnHI site in the CUPl gene, is a double amino acid substitution which changes Asn' to Arg and Phes to Ile. A third mutation, the Sal mutant, is a double amino acid substitution converting Phe2 to Ser and Ser3 to Thr. The fourth mutant, denoted Sal Barn, is a quadruple amino acid substitution which combines the substitutions present in the Barn and Sal mutants. Fragments of DNA containing these mutations in the CUPl gene were subcloned into both high copy (336/Bam, 336/AN, 3361 Sal, 336/Sal Barn) and low copy number yeast vectors (SC/ Barn, SCIAN, SCISal, SCISal Barn). As controls, the wildtype CUPl gene was also inserted into high and low copy number vectors (YEP 36 and SCIWT). These plasmids were transformed into a cup1 A yeast strain in which the chromosomal CUPl gene was deleted by single-step gene replacement (4). Hence, these strains produce metallothionein exclusively from the cloned, plasmid-borne wild-type or mutated CUPl genes.
To assess the ability of several of these molecules to direct metallothionein synthesis, yeast cells containing the 336/ Barn, 336/AN and YEp 36 plasmids were grown in the presence or absence of copper and labeled with [%]cysteine. Cell lysates were prepared and analyzed by gel electrophoresis. Fig. 1 shows that all of the plasmids directed the synthesis of approximately equivalent amounts of MT polypeptide in response to copper. The mobility of the Barn mutant protein was shifted up the gel as would be expected if the mutation blocked the amino-terminal cleavage reaction (amino acidsequencing results are described below). The NH,-terminal deletion mutation, on the other hand, had a very small shift in mobility as compared to the wild-type protein as would be expected since this mutant protein is presumably the 53residue polypeptide with an added NH,-terminal methionine.
Biological Activity of the Mutants-As a first test of the biological activity of the mutant proteins, strains carrying high and low copy number plasmids were streaked onto a series of plates containing various concentrations of copper.
As summarized in Table I, none of the mutations had an effect on copper detoxification when tested on high copy number vectors (see Table V for details). However, with low copy number vectors the AN mutant did show slightly decreased resistance compared to wild-type. It is not yet clear whether this is a direct effect of the ability of this protein to bind copper or an indirect effect on the synthesis or stability of the molecule.
Previous results showed that cuplA strains have a high basal level of transcription from an episomal CUPl promoter (4). To determine the effect of the mutant proteins on regulation of the CUPl gene, RNA was isolated from yeast strains transformed with plasmids SC/Barn and SC/AN as well as strains containing the wild-type gene on a single copy plasmid or chromosomally. The results of a gel blot analysis with a CUPl probe are shown in Fig. 2 and demonstrate that both of these mutants are wild-type with respect to the size and copper inducibility of the CUPl mRNA produced. Analysis of RNA from SC/Sal and SCISal Barn transformants demonstrated that these mutants are also wild-type with respect to regulation of the CUP1 gene (data not shown). Thus, substitutions of four of the eight amino acids in this NH2terminal segment, or the deletion of 7 of these residues, have no observable effect on the biological activity of the CUPl protein.
Since the amino terminus of the CUPl protein did not appear to be absolutely needed for regulation, we next performed immunofluorescent studies to determine if the AN or Barn mutations affected the intracellular location of metallothionein. Yeast cells were prepared from strains transformed with high copy number vectors containing the wildtype CUPl gene or the two mutant genes. Indirect immunofluorescence studies were done using affinity purified IgG from a polyclonal antibody raised against the CUPl protein in rabbits and then reacting with fluorescein conjugated goat anti-rabbit IgG. As shown in Fig. 3, no fluorescence was observed in control incubations with preimmune IgG or with immune IgG and cells prepared in the absence of copper. However, cells prepared in the presence of copper showed a bright fluorescence when incubated with the antimetallothionein immune IgG. Both the wild-type CUPl protein and the two mutant proteins appear to be localized in the cytoplasm. The cells were also stained with 4',6-diamidio-2-phenylindole (DAPI)

50.L4 SCIAN
Analysis of C U P l mRNA in the cuplA strain transformed with plasmids SCIAN, SCIBam, and SC/WT or strain 50.L4. which carries one copy of the wild-type CUPl gene in its normal chromosomal location. Cultures used to prepare the RNA were untreated (-) or incubated in the presence of 75 pm CuSO, (+). Densitometric analysis of the basal levels of CUP1 mRNA indicated that expression of SC/Barn was approximately 2.5-fold higher than the wild-type and SC/AN approximately 2-fold higher. This blot was subsequently washed and reprobed with a 3ZP-labeled EcoRI-Hind111 fragment containing the yeast CYCZ-coding sequence, allowing the quantitation of an internal standard. Analysis of this message revealed that approximately 1.5-fold more RNA was present in the SC/AN and SCIBarn uninduced lanes than in the SC/WT lane. Therefore, the basal levels of these messages are very similar to the wild-type.
of the DAPI-staining patterns with the fluorescein-staining patterns shows that the CUP1 protein is distinctly nonnuclear in the majority of cells. This cytoplasmic location of the wild-type and mutant CUPl proteins was confirmed by fractionation studies done on [35S]cysteine labeled cell lysates (data not shown). Therefore, the inability of yeast to process the NH2-terminal sequence, or the complete absence of the NH2-terminal sequence, does not seem to affect the intracellular location of metallothionein.
Characterization of the Full Length Peptide-Since the double amino acid substitution appeared to block processing, it afforded a unique opportunity to isolate a full length species of metallothionein. This molecule was used to study the contribution of the amino-terminal peptide to the structure of the molecule. The mutant metallothionein eluted from gel filtration on Sephadex G-50 with a distribution coefficient of 0.41, unlike the value of 0.48 for the wild-type molecule. Thus, the molecule has a larger Stoke's radius, consistent with a larger molecular weight. Amino acid analysis of several samples revealed the presence of the residues expected in the aminoterminal extension although in somewhat variable proportions (see Miniprint Table VI for details). This variation was clarified when the amino acid sequence was obtained by Edman degradation (Table 11). The initiator Met at the amino acid terminus was not acetylated and thereby permitted cycles of sequencing. From the yield of different amino acids in each cycle of Edman degradation, it was clear that multiple sequences were present. Although the major sequence initiated at Met, the relative amount of other amino termini was greater in the metallothionein preparation with low amounts of Phe. This is most likely due to varying degrees of proteolytic processing at the amino terminus during isolation.
The copper content of the mutant protein from two preparations was 7.8 and 7.4 mol eq. The binding stoichiometry in seven preparations from wild-type yeast was 7.6 A 0.3 Cu mol eq. Reconstitution studies also suggested that the 61-residue mutant polypeptide bound close to 8 g atoms of Cu per mol of protein. This was apparent in the proteolytic protection assay in which approximately 8 mol eq yielded maximal luminescence in the region of 500-600 nm, the quantum yield protection against proteolysis by proteinase K (Fig. 4). We of which depends on the accessibility of the cluster to the have previously shown that susceptibility toward proteolysis solvent (21,22). Interaction of a copper complex with solvent correlates inversely with the metal content of the protein provides a radiationless mode of energy decay from an atomic since saturation of binding sites confers complete protection. excited state (23). The luminescence spectra of Cu-metallo-Cu-metallothioneins are known to exhibit metal-dependent thionein from the mutant and wild-type yeast exhibited a

Amino acid sequence of 61 -residue Cu-metaUothwnein preparations from 336/Bam mutant Data show pmol recovered at each cycle, uncorrected. The diagonal lines represent assigned sequences in the two preparations of the protein. Since Ile and Glu are each present at two distinct positions in the sequence, the values for these residues at a given cycle are listed for only one position, the other position is shown with a parenthesis.
Cvcle Sequence corrected emission maximum at 609 nm with a half-band width of 37 nm and were of similar quantum yields.

Met-Phe-Ser-Glu-Leu-Ile -Arg-Ile -Gln-Am-Glu-Gly-His
A hydropathic profile of the 61-mer metallothionein sequence revealed the protein to be quite hydrophilic except for the amino-terminal segment (Fig. 5). The ambiguity in assigning a hydropathic value for cysteinyl thiolates involved in metal ligation skews certain segments of the curve in positive values in that the value assigned cysteine is positive by virtue of its typical occurrence in internal disulfide bonds. The hydrophilic nature of the sequence found in the wild-type molecule is expected assuming the protein is related structurally to the mammalian metallothionein in which the conformation is a polypeptide monolayer enfolding the internal metal-thiolate cluster. It is unclear how the hydrophobic peptide segment at the amino terminus of the mutant protein would be accommodated in such a structure.
In an attempt to determine the effect of the amino-terminal extension peptide on protein stability, we explored the susceptibility of this peptide segment to proteolysis. Metallothionein from wild-type yeast is resistant toward proteases in both the native and reconstituted states. The only residues affected by certain proteases are the amino-terminal Gln and the carboxyl-terminal Lys. If the amino-terminal extension peptide participated in the tertiary fold it might likewise be inaccessible to proteases. Treatment of the mutant 61-residue metallothionein with a variety of endoproteases led to removal of the amino-terminal peptide (Table 111). This was observed in incubations of the native Cu-mutant protein as well as reconstituted Cu'-mutant metallothionein with trypsin, proteinase K, and subtilisin. The amino terminus exposed with subtilisin and proteinase K was predominantly Gln9 whereas Ile' was the apparent terminus in the trypsinized sample. The Lys content in the subtilisin-treated sample was low, suggesting that an internal cleavage liberated 1 or 2 lysyl residues. This was confirmed by sequence analysis of this sample in which significant amounts of Ser and Glu were present in the first cycle of Edman degradation in addition to the expected Gln. Presumably, the peptides are held together by metal ligation since the subtilisin-treated sample eluted in the normal monomeric position by gel permeation HPLC analysis. Chromatography of the 61-residue Cu-protein incubated with proteinase K revealed a shift in the distribution coefficient from 0.41 to 0.47, a value identical to that seen with the 53-    Table VI) residue wild-type metallothionein. Similar results were obtained when the protein reconstituted with 4 mol eq of Cd(I1) was digested with proteinase K. The 61-residue mutant metallothionein was also incubated with aminopeptidase M to determine whether the polypeptide was a substrate for an exopeptidase. Both native Cwmetallothionein and protein reconstituted with 8 mol eq of Cu(1) or 4 mol eq of Cd(I1) were processed by aminopeptidase M to approximately the same extent as the apoprotein itself (Table  IV). The susceptibility of the amino-terminal extension to both endopeptidases and exopeptidases suggests that it does not participate in the tertiary fold of the molecule.

DISCUSSION
The DNA-encoding yeast metallothionein was mutagenized to produce four different mutant proteins: two double amino acid substitutions (the Barn and Sal mutants), a quadruple amino acid substitution (the Sal Barn mutant), and a deletion of seven amino acids at the NH2 terminus of the molecule (the AN mutant). These proteins were studied as to their ability to impart copper resistance to yeast cells and their ability to regulate the CUP1 gene in a normal fashion. The Barn and AN mutant proteins were also studied as to their intracellular location. The Barn mutant allowed a 61-residue species of metallothioncin to be isolated from yeast cells and therefore became the focus of detailed biochemical studies designed to probe the structure of the NH, terminus of the protein.
The amino acid substitution mutant proteins were completely wild-type with respect to their ability to detoxify copper. The NH2-terminal deletion protein also gave wildtype protection when assayed on a multicopy number plasmid although somewhat decreased protection was observed with a single copy plasmid. It is not yet clear whether this slight effect reflects decreased copper binding or decreased gene expression due to changes in transcription, translation, or mRNA stability. All mutant genes also exhibited normal patterns of copper-inducible transcription. These results show that the amino-terminal peptide plays little if any role in the two known biological functions of yeast metallothionein, namely copper detoxification and feedback regulation of CUP1 transcription. The possibXty that this sequence somehow plays a role at the DNA level remains to be tested in strains in which the mutant is integrated into the chromosome. In addition, indirect immunofluorescence showed that metallothionein in yeast cells, as in all other species studied to date, is localized in the cytoplasm. This cytosolic intracellular localization was not altered by the presence or absence of the amino-terminal peptide.
The 61-residue peptide produced by the Barn mutant was purified from yeast cells and subjected to sequence analysis. In four different mutant protein preparations, some heterogeneity was observed at the amino terminus. Although the major sequence invariably initiated at the Met', secondary sequences were observed to a variable degree in the samples. Proteolytic processing had clearly occurred during isolation. The protein prepared from wild-type S. cereukciue lacks the first eight amino acids predicted from the nucleotide sequence of the gene with no evidence for amino-terminal heterogeneity (5). The properties of the 61-residue molecule were not different from the truncated wild-type protein. Specifically, both metalloproteins bound eight Cu ions in a complex exhibiting similar luminescence. Since the luminescent quantum yield of Cu-thiolates is influenced by solvent accessibility (21, 22), similar results with the two metallothioneins implies that electronic transitions of the Cu(1) cluster in the two proteks are equally shielded from solvent. It appears, therefore, that the cluster structures must be related.
The 8-residue peptide at the amino terminus of the mutant molecule did not appear to participate in the tertiary fold. Those residues were accessible to endoproteases whereas the wild-type 53-residue protein was largely resistant to proteolytic digestions (5). The amino-terminal residues are proteolytically cleaved in native mutant Ch-metallothionein as well as in Cu-and Cd-reconstituted forms. The exopeptidase, aminopeptidase M, trims the amino terminus to a similar extent in both the apo-and metallo-states of the mutant protein. Despite the hydrophobic nature of the amino-terminal extension peptide, it appeared to be solvent accessible. This is consistent with the predicted structure of the molecule based on the known conformation of mammalian metallothionein in which the molecular interior consists of a polar metal-thiolate cluster (9).
A number of other proteins possess amino termini that are disordered and therefore do not contribute to the conformational stability. This is apparent in crystallography when a terminal region has no electron density. Although most examples of this flexibility are restricted to 1-3 residues (25-27), a few cases of longer regions of disorder are known. The

Amino Terminus
of Yeast Metallothionein first 9 residues in pyruvate kinase display no electron density and therefore are presumed to be highly flexible (28). Likewise, the conformation of the first 18 residues in glutathione reductase is not defined (29). The amino-terminal peptide may be either cleaved during isolation or in vivo by one of the numerous proteases in yeast.
In addition to the well-known five proteases found in yeast vacuoles, numerous proteases in other cellular compartments have been identified (30-32). The pep4 mutant in S. cerevisiue reduces all vacuolar hydrolase activities by failing to process precursor forms of the enzymes (30). The ability to isolate the truncated 53-residue protein from the pep4 mutant implies that the vacuolar hydrolases are not involved in the trimming of the metallothionein sequence (5). Rather, a nonvacuolar endoprotease is implicated by the ability to isolate the full length 61-residue protein with the Arg7-Ile8 mutation. Minor sequences of lengths shorter than the complete 61 residues presumably result from the activity of one of the numerous peptidases in yeast. Unlike most soluble eukaryotic proteins, the initiator methionine at the amino terminus of the mutant metallothionein is not acetylated. A second common amino-terminal-processing reaction is the excision of terminal methionine residues. The specificity of amino-terminal acetyltransferase appears to be dependent on the adjacent 3 residues and methionine aminopeptidase on the penultimate residue (33). The lack of processing reactions by those enzymes suggests that the terminal sequence Met-Phe-Ser in the mutant protein is not a substrate for either enzyme.
One curiosity of the extension peptide is the similarity to peptide segments that serve as targeting signals for mitochondrial proteins. One common feature of those sequences is the presence of apolar, hydroxyl, basic, and carboxyl-amide residues (34). Carboxylates are not found in the targeting sequences. There are similarities in the sequences of the extension peptide of metallothionein and the presequence of certain mitochondrial proteins. The corresponding sequence in yeast cytochrome c l is Met-Phe-Ser-Asn-Leu-Ser- (34). Not only is the sequence similar but the length of the metallothionein extension peptide is nearly appropriate. Only 9 residues of the presequences of either 3-aminolevulinate synthase or the 70 kDa outer mitochondrial membrane protein appear to be critical for targeting molecules to the mitochondrion (35,36). Twelve residues from the presequence of subunit IV of cytochrome c oxidase are sufficient for targeting the molecule to the mitochondrion (37). Even though yeast metallothionein does not appear to be localized within the mitochondria in either the wild-type or mutant strains, the sequence similarities raise the possibility that the protein once was located within mitochondria. A mutation of residue 4 in the ancestral molecule yielding Glu may have precluded mitochondrial targeting. We plan to explore this possibility with site-specific mutagenesis experiments.