Domain Nature of Metallothionein*

Metallothionein purified from the livers of rats injected with CdCl3 was cleaved by proteolysis into a 32-residue polypeptide that contained 4 bound Cd ions. Appearance of this fragment designated alpha requires prior treatment of metallothionein with EDTA to remove the Zn ions and destabilize the 3-metal cysteine cluster in the other domain. The half-molecule domain was not efficiently produced by proteolysis of native metallothionein. The Cd4-alpha fragment is asymmetric in shape, as is the parent molecule. NH2-terminal sequence analysis revealed that the alpha fragment starts at Lys 30. Since the same amino acids are released from the COOH terminus of intact thionein and the alpha fragment by carboxypeptidase Y, the alpha domain generated by digestion with subtilisin therefore comprises residues 30 through 61. The amino acid composition of the alpha polypeptide is consistent with the structure of the 4-metal cysteine cluster proposed by Otvos and Armitage ((1980) Proc. Natl. Acad. Sci. U. S. A. 77, 7094-7098). Metallothionein appears to consist of a 3-metal cysteine domain in the NH2-terminal half of the thionein molecule and the 4-metal cysteine domain in the COOH-terminal half.

Metallothionein purified from the livers of rats injected with CdClz was cleaved by proteolysis into a 32residue polypeptide that contained 4 bound Cd ions. Appearance of this fragment designated a requires prior treatment of metallothionein with EDTA to remove the Zn ions and destabilize the 3-metal cysteine cluster in the other domain. The half-molecule domain was not efficiently produced by proteolysis of native metallothionein. The C&-a fragment is asymmetric in shape, as is the parent molecule. NHz-terminal sequence analysis revealed that the a fragment starts at Lys 30. Since the same amino acids are released from the COOH terminus of intact thionein and the a fragment by carboxypeptidase Y, the a domain generated by digestion with subtilisin therefore comprises residues 30 through 61. The amino acid composition of the a polypeptide is consistent with the structure of the 4metal cysteine cluster proposed by Otvos

and Armitage ((1980) Pmc. Natl. A c d Sci. U. S. A. 77, 7094-7098). Metallothionein appears to consist of a 3-metal cysteine domain in the NHz-terminal half of the thionein molecule and the 4-metal cysteine domain in the COOHterminal half.
Metallothionein is a 6800-dalton protein that contains an unusually high amount of cysteine (33 mol %) and is capable of binding a variety of metal ions including Cd, Zn, Cu, Co, Ag, and Hg (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). The physiological function of the protein is unknown. The number of binding sites for Zn or Cd is 7 metal ions/molecule (4)(5)(6). The protein has been isolated from various eukaryotic cells and is inducible by administration of metal ions (6)(7)(8)(9)(10)(11)(12), dexamethasone (13), food restriction (14), and stress (15). The induction occurs by an enhanced production of thionein mRNA (16,17). Two polymorphic variants of M-Th' are induced, with the isoforms differing in 7 positions in the 61-residue polypeptide (6). The variants will be referred to as isoforms I and 11. There has been much interest in deducing the nature of metal chelation in M-Th. Attempts to elucidate the structure of the native molecule by x-ray diffraction have failed due to a n inability to obtain crystals. However, the structure appears to be a defined, compact conformation lacking a helices or p sheets (18). It i s well established that chelation involves cysteines in the ligand field (2,8,(18)(19)(20). Recently, it was reported in a series of papers that the 7 metal ions in Cd7-Th * This work was supported by Grant AM 29207-05. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
f To whom correspondence should be addressed. The abbreviations used are: M-Th, metallothionein, implying that the metals bound are CdsZna-Th Cdb-Th, zinc-free Cd-thionein; I and 11, the two isoforms of metallothionein; a[, the fragment of isoform I containing the 4-metal cysteine cluster. appear to be bound in two separate polynuclear metal cysteine clusters (21)(22)(23). The most convincing evidence of metal cysteine clusters in Cd7-Th is from the work of Otvos and Armitage using 'I3Cd NMR spectroscopy (21). They observed that the '13Cd resonances are split into multiplets by '13Cd-113Cd spin coupling. Selective homonuclear decoupling of the multiplets permitted the identification of the '13Cd ions that were spin-coupled within the same polynuclear cluster (21). Analysis of the data revealed that the resonances arose from two distinct clusters of metal ions. One cluster contained 3 Cd ions arranged such that each ion was linked to the other two, presumably through cysteine thiolate bridges. The second cluster thus contained the remaining 4 Cd ions.
Since various M-Th sequenced to date show marked homology in the positions of cysteines (6), it is likely that the two-cluster arrangement of the bound metal ions is a common structural feature in all M-Th. We wanted to know whether the two clusters were formed by independent segments of the thionein polypeptide, and if so, whether the two domains could be separated. In the present study we report the resolution of the 4-metal cysteine cluster from M-Th and its characterization.

MATERIALS AND METHODS
Sephadex gels and DEAE-cellulose (DE-52) were obtained from Pharmacia and Whatman, respectively. Ninhydrin and amino acid analyzer buffers were from Beckman, and methyl-Cellosolve was from Fisher. The various proteases and EDTA were purchased from Sigma. The subtilisin used was subtilisin Carlsberg from Bacillus subtilis. Fluorescamine was obtained from Pierce. Male Sprague-Dawley rats weighing between 250 to 300 g were purchased from Simmonsen and were fed commercial rat chow ad libitum.
Metallothionein was purified from the livers of rats given repeated subcutaneous injections of CdC12 (2.5 mg of metal/kg, body weight) as described previously (7, 24). The isoforms of Cd,Zn-thionein separated by ion exchange chromatography were desalted, concentrated by lyophilization, and stored at -80 "C. Homogeneity was ascertained by polyacrylamide gel electrophoresis, the absorption ratio of Azm/ AZ8,,, and amino acid analysis. These samples represented the starting material for the proteolysis studies.
Amino acid analysis performed on a Beckman 120 C analyzer was used to assess purity and in all cases to quantitate protein concentrations. Cysteine was determined as cysteic acid after performic acid oxidation (25). Cysteines in the thionein samples were also modified by carboxymethylation using iodoacetate (26). Metal analysis was carried out on a Perkin-Elmer Model 305A spectrometer. Absorption spectra were recorded on a Cary 118 spectrometer. Fluorescence measurements were made on a Perkin-Elmer 650-10s spectrophotometer of samples treated with fluorescamine (27). Samples in 1.8 ml of 0.2 M sodium borate, pH 8.5, were mixed with 0.2 ml of fluorescamine (0.5 mg/ml) and the fluorescence emission at 475 nm was measured with excitation at 390 nm. Polyacrylamide gel electrophoresis was performed at pH 8.9 in 7.58 gels according to the procedure of Davis (28). Analytical gel filtration was done on a Sephadex G-75 (superfine) column (2.5 X 110 cm) in 10 m~ potassium phosphate, pH 7.8, at 25 "C. Protein standards and metallothionein samples were chromatographed in varying combinations. Reference proteins used to calibrate the column included albumin, Cu,Zn-superoxide dismutase, carbonic anhydrase, myoglobin, ribonuclease, cytochrome c, and pancreatic trypsin inhibitor. Blue dextran and NaCl were added to mark M is the molecular weight, B is the partial specific volume, and 6 is the degree of solvation (assumed 0.2 g of H*O/g of protein). Frictional ratios were also calculated from gel filtration data using the relationship (30): is the frictional ratio for reference proteins and was assumed to be 1.2 Me,, is the experimentally derived apparent molecular weight from gel filtration, and Mp is the actual molecular weight of the protein from sequence data.
NHr-terminal sequence analysis of the performic acid-oxidized thionein proteolytic fragment was carried out on a Beckman 89OB Sequencer using a 0.1 M Quadrol program (12078). Polybrene was included in each sequence run, and five pretreatment cycles were run in the presence of the dipeptide lysyl-glycine prior to loading thionein (31). Fractions recovered from each cycle were automatically converted to phenylthiohydantoins with a Sequemat P-6 converter. Phenylthiohydantoins were analyzed by high pressure liquid chromatography using a Waters SISP-79OA instrument. In standard runs using sperm whale myoglobin in the absence of polybrene, fmt step yields averaged about 60% with repetitive yields around 93%.

RESULTS
Incubation of native rat liver M-Th (1 to 5 mg) containing 4.7 Cd ions and 2 Zn ions/molecule with 1 m~ EDTA in 10 m~ potassium phosphate, pH 7.8, at 25 "C for 1.5 h leads to a selective loss of thionein-bound Zn (2,32,33). Upon removal of the Zn-EDTA by Sephadex G-25 column chromatography, the resulting M-Th retained the 5 g atoms of Cd/mol of protein, and contained less than 0.1 g atom of Zn/mol. With higher concentrations of M-Th (10 to 15 mg), the EDTA concentration was increased up to 4 mM to remove the zinc. Incubation of the EDTA-stripped M-Th isoform 11, designated Cds-ThrI, with various proteases including elastase, subtilisin, proteinase K, and thermolysin at a protein to protease ratio of 201 at 37 "C in 10 mM Tris-C1, pH 8.6, for 16 h resulted in discrete cleavage products (Fig. 1). The additional bands observed in the Cds-Th blank (lane B ) appeared to represent aggregates. Trypsin, chymotrypsin, and papain did not generate fragments under similar conditions. Incubation of native CdsZn2-Th with the mentioned proteases under the same conditions for up to 48 h did not result in any apparent proteolysis. In repeated proteolysis experiments with Cds-Th, subtilisin appeared to be the most effective protease in generating a single fragment of thionein, and therefore was the enzyme used in the subsequent studies.
Chromatography of the subtilisin-digested Cd5-ThI on Sephadex G-75 yielded two main Cd-containing fractions des-  (Table I) Table I. Although similar a fragments can be obtained from both isoforms of M-Th, the efficiency of a formation by subtilisin treatment is greater with isoform I compared to 11. Retreatment of Sephadex G-75 pool 1 with EDTA and subsequently redigestion of the desalted material with subtilisin again resulted in incomplete conversion of M-Th to a with both isoforms.
The a fragments from the two isoforms purified through Sephadex G-75 and desalted on Sephadex G-25 were electrophoresed on 7.5% polyacrylamide gels. Fig. 3 shows the Coomassie R-250-stained protein bands of the a fragments and their parent isoforms on the nonnaturing gels. The a fragments from both isoforms were applied to DEAE-cellulose columns equilibrated in 10 mM Tris-C1, pH 8.6, and both eluted in the column wash. No further degree of purification was observed after ion exchange chromatography as evidenced by nondenaturing polyacrylamide gel electrophoresis or amino acid analysis.
After the initial characterization of the a fragment, the A B C D conditions of proteolysis were investigated to determine what was required to generate a from Cds-Th. Digestion of Cds-Th with subtilisin at a 501 protein to protease ratio at 4 "C for 16 h yielded a fragments that exhibited properties identical with those of the a molecules generated by incubation at 37 "C and at higher concentrations of subtilisin. The proteolytic conditions may still be excessive, but the a molecule is extremely resistant to proteolysis. The ultraviolet absorption spectra of a1 and native M-Thl are featureless, but characteristic of M-Th. The AZW ,,,,,/AM , , , , , ratio for a was about 14, compared to over 2C for native Cds-Znz-Th. At 250 nm the absorbance of a 1% solution of a is about 160 cm" compared to 102 cm" for native CdaZn2-Th. The a fragment was characterized to determine the arrangement of the a polypeptide in the primary structure of M-Th. Edman degradations of the performic acid-oxidized derivative of a1 revealed that the NHz terminus was Lys 30 based on the known sequence of mouse M-Th (Table 11). The first step yield as quantitated by integration of the high pressure liquid chromatography peak was only 18%, but sequence runs of CNBr-treated isoforms of thionein invariably showed first step yields of about 20% (33). The sequence of a was carried out in duplicate and the results unambiguously show a starting at residue 30.
The a fragments and intact thionein were incubated with carboxypeptidase Y to determine if subtilisin cleaved any residues from the carboxyl-terminal end. The carboxymethylated samples were incubated at 37 "C in 0.1 M pyridine acetate, pH 5.5, at a protein to carboxypeptidase Y weight ratio of 7:l. As can be seen in Fig. 4, both a1 and thioneinl gave a time-dependent release of Ala, carboxymethylcysteine, and Ser (i.e. SerCysAla-COOH) consistent with the known COOH-terminal sequence of the thioneins characterized to date (6). No release with carboxypeptidase Y was observed if the thionein samples were performic acid-oxidized instead of carboxymethylated. Control incubations of only carboxypeptidase Y or only thionein did not reveal any significant concentration of free amino acids. from isoforms I1 and 1,~respectively.
Since the various metallothioneins sequenced to date show Domain Nature of Metallothionein marked homology (6), alignment of the a polypeptide in the known sequence of mouse thioneinl (32) is justified. The a fragment generated by subtilisin cleavage represents the polypeptide portion of thionein from residues 30 through 61. In Table I, the compositions of the a samples of the two isoforms are shown, with the compositions of the comparable mouse liver thioneinI polypeptide calculated from the known sequence. The correlation in the amino acid composition is excellent.
Although the calculated molecular weight of Cd,-a from sequence data is 3617, the apparent molecular weight from gel fitration is 6500, suggesting that the molecule exhibits asymmetry in shape. The frictional ratio ( f / f m i n ) of C&-a calculated according to Equation 2 under "Materials and Methods" is 1.45, whereas the ratio for native M-Th is 1.4. The frictional ratio ( f/fo) calculated from Stokes radii (R/Ro) for native M-Th was 1.3 (33). It therefore appeared that the asymmetry in metallothionein was inherent in the individual domains and did not arise primarily from the linkage of two globular domains.
Since a represents the COOH-terminal half of the thionein molecule and binds 4 metal ions, the question arises as to the fate of the NHz-terminal portion of the molecule. After proteolysis of Cds-Th with subtilisin, the elution fractions from Sephadex G-75 were monitored with the fluorescamine reaction to quantitate amino groups. As can be seen in Fig. 2, a major fluorescence peak emerged in the column internal elution volume (VI). The only source of amino groups was necessarily from thionein, since proteolysis and column elution were performed in potassium phosphate buffers. Since a was a homogeneous fragment, the NHz-terminal polypeptide segment was presumably degraded into small peptides. Quantitation of the relative fluorescence in the a and internal volume (VI) fractions suggested that approximately 4 to 5 cleavages occurred in the NHz-terminal 20-amino acid segment. The presence of amino acids in the VI fraction was confirmed by amino acid analysis.
If a represents a domain of M-Th containing the 4-metal cluster, then the Cd ions should be tightly complexed as in the native metalloprotein. We compared the Cd-binding properties of aI to those of the intact M-Th molecules by studying the pH-dependent displacement of Cd and the reactivity of the samples toward EDTA. Since the extinction at 250 nm was proportional to the bound Cd content (2), the absorbance at 250 nm was monitored as a function of pH. Fig. 5 shows that the pH-dependent loss of absorbance of the a1 fragment was coincident to the curve for Cd,Zn-Thl, suggesting that similar mercaptide chelation exists for both. The reactivity of the metallothionein samples toward EDTA was monitored by following the absorbance at 254 nm with time for incubations of M-Th and aI in the presence of 1 mM EDTA (Fig. 6) loss of absorbance, i.e. decrease in Cd content, was more pronounced in a1 compared to M-Thr. The time required for the absorbance to be reduced by 50% was twice as long for M-Thr and M-ThII compared to ar.

DISCUSSION
Metallothionein was cleaved by proteolysis into a 32-residue polypeptide domain that binds 4 Cd ions. The domain, designated a, was generated with a variety of proteases, but subtilisin has been most extensively used. The a fragment exhibits properties similar to native M-Th in its ultraviolet absorption spectra, asymmetry in conformation, and pH-dependent release of Cd ions. Although the hydrogen ion displacement of Cd ions in Cd-a and Cd,Zn-Th is similar, the reactivity of these samples toward EDTA differs. The slower reactivity of native M-Th with EDTA suggests that the juxtaposition of the two domains exerts a stabilizing effect on each other. We plan to study the actual metal-binding affinities of a and M-Th by microcalorimetry.
The a domain is efficiently produced only if native M-Th is initially treated with EDTA. Presumably the EDTA incubation removes the two Zn ions that appear to bind to the 3metal cluster (21). Loss of the Zn ions appears to destabilize the structure of the 3-metal cluster, resulting in a random coil conformation in the NH2-terminal portion of the molecule. It is well established that removal of metal ions from M-Th results in a random coil configuration of the polypeptide (18). Proteolysis with subtilisin cleaved the NHZ-terminal 29-residue polypeptide in numerous places, yielding small peptides and Q that was then separated from the peptides by gel filtration. Proteolysis of native M-Th (not EDTA treated) under rigorous conditions produced only insignificant amounts of a. Therefore, under the conditions employed, isolation of the intact 3-metal cluster was not possible. Complete conversion of M-Th to a fragment was not possible with repeated digestions of either isoform, suggesting that microheterogeneity may exist in the sequence of the susceptible hinge.
Proteolysis with elastase or proteinase K yielded additional charge variants of a (Fig. 1).
Otvos and Armitage (21) proposed from '13Cd NMR studies that M-Th contains two polynuclear metal cysteine clusters. They postulated structures for each cluster that would satisfy their '13Cd-"3Cd spin coupling data. The 4-metal ion center contained 11 cysteines with 6 thiolate bridges connecting adjacent metal ions, and the 3-metal ion center contained 9 cysteines and 3 thiolate bridges. The 4-metal cysteine center designated cluster A by Otvos and Armitage (21) is consistent with the composition of our a fragment. Our data, therefore, supports the postulate that M-Th contains two separate metal cysteine centers. The 4-metal center (cluster A) is in the a domain and that constitutes the COOH-terminal half of the molecule. Our data on the structure of metallothionein is consistent with that depicted in Fig. 7.
Characterization of a M-Thr genomic clone isolated by partial restriction nuclease digestion of mouse myeloma DNA showed that the thionein gene contained two introns (16). One intron-exon junction occurs in the coding region for Ser 32. Thus, exon 3 encodes the COOH-terminal29 amino acids that fold into the a domain. Whereas the location of this intron supports the idea that exons represent functional units in a protein, the location of the other intron in the polypeptide segment encompassing the 3-metal center raises doubts about this hypothesis. The polypeptide segment from Thr 27 to Ser 32 may be a partially exposed hinge peptide linking the two metal-binding domains.
The isolation of the half-thionein molecule will reduce the complexities in structural studies, since the structure of a single metal cysteine cluster can now be independently investigated. Elucidation of the crystal structure of Q may provide insights into the physiological role of the protein and would confirm cysteinate bridging in proteins. We are trying to successfully separate the two metal clusters from native M-Th in order to study the metal ion distribution between the two centers. In Cu-Th, 10 g atoms of Cu binds/mol of protein, unlike the 7 metal ions bound in Cd,Zn-Th. It is conceivable that different conformations of the clusters exist to accommodate varying metal ions, and the alterations in conformation may influence function.