Nội dung

Coordination of three and four Cu(I) to the a- and b-domain of vertebrate Zn-metallothionein-1, respectively, induces significant structural changes Benedikt Dolderer1, Hartmut Echner1, Alexander Beck1, Hans-Jürgen Hartmann1, Ulrich Weser1, Claudio Luchinat2 and Cristina Del Bianco2 1 Interfakultäres Institut für Biochemie, University of Tübingen, Germany 2 Magnetic Resonance Center, University of Florence, Sesto Fiorentino, Italy Keywords copper; domain; metallothionein; protein structure; 2D NMR Correspondence U. Weser, Anorganische Biochemie, Interfakultäres Institut für Biochemie, University of Tübingen, Hoppe-Seyler-Str. 4, D-72076 Tübingen, Germany Fax ⁄ Tel: +49 7071295564 E-mail: ulrich.weser@uni-tuebingen.de (Received 17 January 2007, revised 28 February 2007, accepted 5 March 2007) doi:10.1111/j.1742-4658.2007.05770.x Vertebrate metallothioneins are found to contain Zn(II) and variable amounts of Cu(I), in vivo, and are believed to be important for d10-metal control. To date, structural information is available for the Zn(II) and Cd(II) forms, but not for the Cu(I) or mixed metal forms. Cu(I) binding to metallothionein-1 has been investigated by circular dichroism, luminescence and 1H NMR using two synthetic fragments representing the a- and the b-domain. The 1H NMR data and thus the structures of Zn4a metallothionein (MT)-1 and Zn3bMT-1 were essentially the same as those already published for the corresponding domains of native Cd7MT-1. Cu(I) titration of the Zn(II)-reconstituted domains provided clear evidence of stable polypeptide folds of the three Cu(I)-containing a- and the four Cu(I)-containing b-domains. The solution structures of these two species are grossly different from the structures of the starting Zn(II) complexes. Further addition of Cu(I) to the two single domains led to the loss of defined domain structures. Upon mixing of the separately prepared aqueous three and four Cu(I) loaded a- and b-domains, no interaction was seen between the two species. There was neither any indication for a net transfer of Cu(I) between the two domains nor for the formation of one large single Cu(I) cluster involving both domains. The first member of the metallothionein (MT) family was isolated in 1957 [1]. Since then, a large number of proteins have been described featuring common characteristics. They include ubiquitous small cysteine-rich proteins (50–70 amino acids) that are able to bind many d10 metal ions [2]. A wealth of different biological functions has been proposed and continues to be discovered. Obviously, MTs play important roles in minimizing the uncontrolled reactions of heavy metal ions like cadmium and the homeostasis of essential metal ions including copper(I) and zinc(II) ions [2,3]. They are known to successfully cope with oxidative stress and ionizing radiation [4,5]. Other functions may be associated with the occurrence of tissue-specific isoforms, such as the brain-specific isoform MT-3, which acts as neuronal growth inhibitory factor [6,7]. Both mammalian MT-1 and MT-2 are composed of the N-terminal b- and the C-terminal a-domain. They are predominantly isolated containing zinc or cadmium exclusively bound to cysteine thiolates. The nine cysteine residues of the b-domain accommodate a metal (M)(II)3S9 cluster, while 11 cysteine residues contribute to the formation of a M(II)4S11 cluster in the a-domain [8]. However, under certain physiological conditions, e.g. when isolated from fetal liver, mammalian MT-1 Abbreviations M, metal; MT, metallothionein. FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS 2349 Murine Cu(I)a- and Cu(I)b-MT-1 domains B. Dolderer et al. and MT-2 are also found to be enriched with Cu(I) [9]. For other members of the MT family, different metal cluster architectures were reported. The previously mentioned MT-3, which is also a two-domain protein, for example, is composed of a Cu(I)4S9 cluster in the N-terminal b-domain and a Zn(II)4S11 cluster in the C-terminal a-domain [10,11]. Examples for solely Cu(I) binding MTs are Cu(I)8 thionein from Saccharomyces cerevisiae and Cu(I)6 thionein from Neurospora crassa [12–14]. Differently from other described MTs, these two fungal proteins consist only of a single domain harbouring homometallic Cu(I) thiolate clusters [13,14]. The three-dimensional structure of Cd5Zn2MT-2, isolated from rat liver after cadmium supplementation, was determined using both NMR and X-ray crystallography [15]. The entire protein is dumbbellshaped and contains two independent domains. The polypeptide backbone wraps around the metal thiolate core forming the scaffold of the two domains. All metal ions are tetrahedrally surrounded by four thiolate sulphur atoms. In the N-terminal b-domain, the three metal ions and the three bridging thiolate sulphurs are ordered to form a distorted chair. The C-terminal a-domain is characterized by an adamantane-like four-metal cluster. Solution structures of 113 Cd-substituted Cd7MT-2 from rabbit, rat and human are available and revealed structural identity with the structure of Cd5Zn2MT-2 [8]. Comparative NMR studies provided evidence that Zn(II) can isomorphically replace Cd(II) in MT-2 [16]. This result was corroborated by NMR studies on cobalt substituted MTs, as cobalt was often used as a zinc analogue in structural investigations [17–19]. The solution structure of murine 113Cd7MT-1 showed high similarity with rat liver MT-2. Its b-domain, however, turned out to be more flexible than in the latter protein, exhibiting enhanced cadmium–cadmium exchange rates [20]. The structural and spectroscopic data available on Cd(II)-substituted human MT-3 indicated the formation of two metal thiolate clusters, similar to those found in MT-1 and MT-2. Further investigation of that protein pointed towards a highly dynamic structure [8]. Recently, a high-resolution solution structure of the C-terminal a-domain has become available. The data revealed a tertiary fold very similar to that of MT-1 and MT-2, except for a loop that contains an acidic insertion that is highly conserved in these isoforms. The structure of the b-domain has escaped its experimental solution, as no characteristic signals attributable to its residues were observed using NMR. On the basis of homology modelling, a backbone 2350 arrangement virtually identical to the corresponding domains in MT-1 and MT-2 was suggested [21]. Despite the large number of structural data available for the MT family, only the structures of two MTs containing Cu(I) were known until now. One of them is the aforementioned yeast MT whose structure was successfully determined using both 2D NMR and X-ray diffraction [22–24]. This protein forms one single domain that harbours eight Cu(I) ions. Six of them are coordinated by three thiolate sulphur atoms, whereas a linear binding mode was observed for the remaining two. The solution structure of N. crassa MT backbone, in which, like yeast, the MT solely binds Cu(I), represents the second known structure of a copper thionein [25]. Its polypeptide chain wraps around the copper sulphur cluster in a left-handed form in the N-terminal half of the molecule and in a right-handed form in the C-terminal half. Due to the lack of copper isotopes with NMR-active spin ½, no metal–cysteine restraints were available to solve the positions of Cu(I) within the N. crassa MT polypeptide fold. At present, the structural information on Cu(I)-loaded forms of mammalian MTs is rather limited. In vitro, Cu(I) titrations of isolated MT-2 and its separate domains demonstrated that up to six Cu(I) ions can bind to each domain [26]. In another extensive titration study, it was postulated that zinc was required for the in vivo and in vitro folding of the two domains of copper MTs [27]. Replacement of Zn(II) by Cu(I) led to the proposal of the formation of Cu,Zn-MT intermediates and that, during the last steps of copper titration, the two domains merge into one. However, earlier Cu(I) titration studies of rat liver MT clearly showed that the two domains remained separated [26]. Additionally, the cooperative formation of (Cu3Zn2)a(Cu4Zn1)bMT)1 upon addition of Cu(I) to (Zn4)a(Cu4Zn1)bMT)1 indicated that the preference of Cu(I) for binding to the b-domain is only partial and not absolute, as widely accepted until now [27]. It was of interest to shed some light on the changes of the molecular architecture of the two domains of vertebrate MT when Cu(I) is added to them. For this task, the synthetic murine aMT-1 and bMT-1 domains were prepared for subsequent Cu(I) titrations. Employing NMR, we obtained an interesting and unexpected picture of the Cu(I) binding to the two single domains. Results and Discussion Cu(I) titration of Zn4aMT-1 and Zn3bMT-1 As both the structure of native Zn7MT-1 was known, and several Cu(I) binding stoichiometries for its two FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS B. Dolderer et al. Murine Cu(I)a- and Cu(I)b-MT-1 domains domains had been proposed, it was of interest to shed some more light upon their reactivity towards the presence of Cu(I). To this end, a Cu(I) titration study of the two separated domains was performed employing the combined detection of luminescence, circular dichroism and 1H NMR. Solid-phase peptide synthesis was successfully employed to prepare the independent a- (residues 31–61) and b-domains (residues 1–30) of murine MT-1. Either domain was fully reconstituted under anaerobic conditions with Zn(II) to yield Zn4aMT-1 and Zn3bMT-1. For each Cu(I) titration step, a new sample was prepared in order to minimize the risk of oxidation during sample manipulation and measurement. The Zn4aMT-1 and Zn3bMT-1 derivatives were separately titrated with Cu(I) under a nitrogen atmosphere to yield Cu(I)–polypeptide stoichiometries from zero to six. The sample solution contained 20% (v ⁄ v) acetonitrile, as the presence of soft ligands prevents Cu(I) from disproportionation to Cu(II) and Cu(0). CD and luminescence emission was measured in order to assess the sample preparation quality and to compare the obtained results with those previously published [26,27]. These physicochemical properties are exclusively attributable to the metal-thiolate chromophores that have been proven to be essential for the proper polypeptide folding in other MTs [2]. The overall shape of the CD spectra was essentially the same as reported before (Fig. 1). During the titration of the adomain, two positive dichroic bands developed at 248 and 300 nm, respectively, and one negative band at 275 nm. The addition of Cu(I) to Zn3bMT-1 shifted the positive dichroic band at 248 to 260 nm. A second positive band at 300 nm, that was not present in the spectrum of Zn3bMT-1, appeared on addition of Cu(I). As in the case of the CD spectra, the results of luminescence emission were comparable to earlier studies (Fig. 2). An almost linear increase of intensity was observed until the addition of the third and fourth Cu(I) ions to the a- and b-domain, respectively. Further Cu(I) addition led to a much more pronounced increase of intensity in both species. Two-dimensional 1H-1H NOESY spectra of each sample were acquired at 700 MHz (Figs 3 and 4). The spectrum of Zn4aMT-1, corresponding to the starting point of the aMT-1 titration, was consistent with a well-folded polypeptide. Spin systems of the amide protons spread from 6.8 to 9.2 p.p.m. Upon addition of the first equivalent Cu(I), the spin systems of the starting point remained preserved, but additional new spin systems started to appear. In the spectra of the samples containing two, three and four equivalents of Cu(I), these new spin systems were prevalent with the most and strongest signals observed for the three Cu(I)-containing sample. On further additions of Cu(I), the signals faded away such that the spectra of the six and seven Cu(I) titration steps were devoid of cross-peaks (not shown). For the b-domain similar results were obtained with the difference that the first addition of Cu(I) led only to the reduction of signals in the NOESY spectrum and that new spin systems appeared only after the second equivalent Cu(I) was added. The spectra of the samples containing three, four and five equivalents A B 20 10 0 -10 Fig. 1. CD spectra of Zn4aMT (A) and Zn3bMT (B) along the titration with Cu(I). Samples containing 35 lM of the respective domains dissolved in 15% (v ⁄ v) acetonitrile, 20 mM sodium phosphate buffer pH 7.6 were prepared under nitrogen containing <1 p.p.m. O2. -20 Zn4- -MT Zn3- -MT + 1 eq. Cu(I) + 2 eq. Cu(I) + 3 eq. Cu(I) + 4 eq. Cu(I) + 5 eq. Cu(I) + 6 eq. Cu(I) + 1 eq. Cu(I) + 2 eq. Cu(I) + 3 eq. Cu(I) + 4 eq. Cu(I) + 5 eq. Cu(I) + 6 eq. Cu(I) -30 250 300 / nm FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS 350 400 250 300 350 400 / nm 2351 Murine Cu(I)a- and Cu(I)b-MT-1 domains A B 6 equivalents Cu(I) 3000 6 equivalents Cu(I) 2000 1000 0 relative intensity 6000 4000 relative intensity 50 40 7000 5000 relative intensity 60 B. Dolderer et al. 5000 4000 3000 2000 1000 0 equivalents Cu(I) 0 0 1 2 3 4 5 6 mole equiv. of Cu(I) 0 1 2 3 4 5 6 mole equiv. of Cu(I) 0 equivalents Cu(I) 30 20 10 0 500 550 600 650 700 500 550 / nm Cu(I) contained the same new spin systems. The most and strongest NOEs were observed in the spectrum of the four Cu(I)-containing sample. As with the a-domain, progressive disappearance of NOEs without reappearance of any new signals was the result of Cu(I) to polypeptide stoichiometries higher than five. Taken together, the initial additions of Cu(I) to each domain caused the disappearance of a large set of NOESY cross-peaks and the parallel appearance of another set of cross-peaks, until a clean 2D spectrum belonging to a single species was obtained. Judging from the highest number of NOEs and the strongest signals in the respective NOESY spectra, the recovery of a single species was completed after the addition of three Cu(I) equivalents to Zn4aMT-1 and of four Cu(I) equivalents to Zn3bMT-1. This result was surprising insofar as structurally defined Cu(I)-containing species were identified with such unexpectedly low stoichiometries of Cu(I) to polypeptide. Several different Cu(I) binding stoichiometries had been proposed for the two domains, among which Cu3aMT-1 and Cu4bMT-1 had mostly been considered to be transient intermediates in the pathways describing the formation of the fully loaded domains [27–30]. Cu6aMT-1 and Cu6bMT-1 had been the most prominent among the candidates for the fully Cu(I) loaded domains [26]. In the present titration study, however, the distinct structures disappear upon addition of more than three and four Cu(I) equivalents without any sign of newly forming defined structures. We can only speculate what happens at this stage of titration. One possible explanation for the disappearance of NOESY signals at high Cu(I) concentration might be that two or more Cu(I) 2352 600 / nm 650 700 Fig. 2. Luminescence emission spectra of Zn4aMT (A) and Zn3bMT (B) along the titration with Cu(I). Samples containing 14 lM of the respective domains dissolved in 15% (v ⁄ v) acetonitrile, 20 mM sodium phosphate buffer, pH 7.6, were prepared under nitrogen containing <1 p.p.m. O2. Spectra were recorded at 25 C using slits of 15 and 20 mm for excitation and emission monochromators, respectively. Excitation was at k ¼ 300 nm. The insets show the emission intensities plotted against the respective polypeptide stoichiometries. binding modes coexist in an intermediate exchange regime, such that signals are exchange broadened and become invisible. Of course, there is still the possibility that the separated domains are simply incapable of binding more than three and four Cu(I) without aggregating and denaturing, whereas in the native MT-1, the presence of the other domain would help to accommodate additional ions. We do not believe that this is very likely, however, because of the similarity of the Zn4aMT-1 and Zn3bMT-1 structures with the structures of the domains of the intact protein (see below). [Correction added after publication 30 March 2007: in the preceding sentence the first term, Zn3aMT-1 was corrected to Zn4aMT-1]. The biophysical similarities of intact protein and separated domains would also argue against this proposal [26]. Luminescence titration series also provide noteworthy information. Luminescence intensities increased almost linearly until Cu(I) stoichiometries of three and four for the a- and b-domain, respectively. At this point, the curves were bent and further Cu(I) equivalents caused a much stronger, but also linear increase of intensity. As luminescence intensity is a measure of how the Cu-thiolate luminophore is shielded from solvent quenching, the titrations indicate that the shielding of the metal-thiolate cluster in the newly identified structures is not optimal compared with the situation with higher Cu(I):polypeptide stoichiometries. An explanation for this and the loss of structural information might be the formation of polymolecular structurally undefined aggregates of native MT or its single domains when they are overloaded with Cu(I) in the presence of unphysiologically high Cu(I) concentra- FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS B. Dolderer et al. Murine Cu(I)a- and Cu(I)b-MT-1 domains Fig. 3. Upper-left parts of the 1H-1H NOESY spectra of Zn4aMT (A), Zn4aMT +1 Cu(I) (B), Zn4aMT +2 Cu(I) (C), Zn4aMT +3 Cu(I) (D), Zn4aMT +4 Cu(I) (E) and Zn4aMT +5 Cu(I) (F). All samples contained 1 mM polypeptide dissolved in 15 mM acetate-d3, 15% acetonitrile-d3, 10% D2O, 50 mM potassium phosphate, pH 6.5 and were prepared under a nitrogen atmosphere containing less than 1 p.p.m. O2. Measurements were performed at 283 K on a Bruker AVANCE 700 spectrometer operating at 700.13 MHz using 600 ms mixing time. tions. Unlike the observed distinct stoichiometries of three and four Cu(I) leading to a sharp rise of the luminescence, only a very small dependency was seen in the circular dichroic measurements. This was also shown earlier by Bofill et al. [27], although CD spectrometry is obviously not sensitive enough to detect comparable significant changes as deduced from luminescence data. 1 H NMR and solution structures of Zn4aMT-1 and Zn3bMT-1 From previous different studies on vertebrate Zn(II)and Cd(II)-containing MTs, it was already known that the two domains form independently from each other and do not interact with each other. Therefore, it was suggested that the two single domains possess very FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS 2353 Murine Cu(I)a- and Cu(I)b-MT-1 domains B. Dolderer et al. Fig. 4. Upper-left parts of the 1H-1H-NOESY spectra of Zn3bMT (A), Zn3bMT +1 Cu(I) (B), Zn3bMT +2 Cu(I) (C), Zn3bMT +3 Cu(I) (D), Zn3bMT +4 Cu(I) (E), and Zn3bMT +5 Cu(I) (F). Sample and measurement conditions were the same as described in Fig. 3. similar structures, if not identical, to their structure in the intact protein. Analysis of the NOESY and TOCSY (not shown) spectra of the two domains permitted the full sequence-specific assignments, the identification of the spin systems and the assignment of 398 and 252 of the NOESY cross-peaks of the a- and b-domain, respectively. The comparison of the chemical shifts with those reported for the cadmium derivative revealed very close similarities (supplementary Tables S1 and S2). In the spectrum of the a-domain, the resonances were shifted marginally by some hundredths of 2354 a p.p.m. The differences observed for the b-domain were more pronounced, with some deviations of up to 0.2 p.p.m., which is probably due to an increased flexibility in this domain. The spin system patterns reported for the published cadmium protein, however, were preserved in both domains. Most importantly, 22 of the 28 long-range NOEs that were reported for Cd7MT-1 were also found in the spectra of the zinccontaining a- and b-domain (Table 1). It should be noted that three of the six missing long-range NOEs were assigned to residues of the linker region between FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS B. Dolderer et al. Murine Cu(I)a- and Cu(I)b-MT-1 domains Table 1. Comparison of the long-range NOEs (dij j > I +4) of Zn4aMT and Zn3bMT with those observed for Cd7MT [20]. Presence (+) or absence (–) of NOEs is indicated. b-Domain Proton 1 Asn 4 (a) Asn 4 (a) Asn 4 (b) Cys 5 (a) Cys 5 (b) Asn 23 (b) a-Domain Lys 1 (NH) Lys 1 (a) Ser 2 (b) Ser 2 (HN) Cys 3 (HN) Cys 3 (a) Cys 3 (b) Ser 5 (b) Cys 6 (a) Cys 6 (a) Cys 6 (a) Cys 6 (b) Cys 6 (b) Cys 6 (a) Cys 6 (b) Cys 6 (b) Cys 6 (b) Lys 13 (b) Lys 13 (c) Lys 13 (c) Cys 14 (NH) Val 19 (c) Proton 2 Lys 22 (b) Asn 23 (NH) Asn 23 (NH) Cys 21 (b) Cys 21 (b) Cys 29 (a) + + + + + – Val 9 (c) Val 9 (c) Val 9 (b) Val 9 (c) Val 9 (c) Cys 18 (b) Cys 18 (HN) Asp 25 (HN) Asp 25 (HN) Asp 25 (a) Lys 26 (HN) Lys 26 (HN) Lys 26 (a) Cys 27 (HN) Cys 27 (HN) Cys 27 (a) Cys 27 (b) Val 19 (c) Val 19 (c) Cys 30 (a) Val 19 (c) Cys 19 (b) – + + – + + + – + – + + + + + – + + + + + + the structure calculation process. Using these connectivities together with the data for Zn4aMT-1 resulted in a target function of 0.52 ± 0.13 Å2 and rmsd values of 1.04 ± 0.12 Å and 1.51 ± 0.18 Å for a new structure family. Both mean structures were essentially the same, with rmsd values of 0.79 Å and 1.19 Å for the backbone and heavy atoms, respectively. Thus, the addition of metal–sulphur connectivities to the structure calculation of Zn4aMT-1 resulted in a betterresolved structural family but did not change the overall protein fold. Figure 5 shows the superposition of the structural family obtained with these additional connectivities and the mean structure of the previously published Cd4aMT-1, showing that they possess the same structure, within the experimental uncertainty. A separate structure determination of Zn3bMT-1 on the basis of the present NMR data was not attempted, as only a small number of NOEs and only four longrange NOEs were found in its NOESY spectrum. Only with the help of the metal–sulphur connectivities discovered using the cadmium-containing derivative could a structure determination have been possible. However, preserved chemical shifts and spin system patterns indicate an identical structure for Zn3bMT-1 as for Cd3bMT-1. As anticipated, the structures of the separated Zn(II)-containing domains are indistinguishable from those of Cd7MT-1 and, knowing the appearance of the starting points, it was of interest to know to what extent they would change in the presence of copper(I). 1 the two domains. Therefore, the lack of the second domain seems to lead to increased flexibility at either end that would build up the linker region in native MT-1. The fact that the majority of the long-range NOEs observed in Cd7MT-1 is still present in the single zinc-containing domains suggests that the global structures of the two domains are preserved, regardless of whether zinc or cadmium is bound to them and also regardless of the existence of the second domain. The assigned peaks of the a-domain were integrated and their consistency with the published solution structure was checked using the program dyana and the published solution structure of the respective cadmium-containing domain of the intact protein as a starting point. The resulting structure family had a target function of 0.15 ± 0.11 Å2 and rmsd values of 1.51 ± 0.27 Å and 2.22 ± 0.30 Å with respect to the mean structure for the backbone and all heavy atoms, respectively. In the previous study on the Cd7 derivative, metal–sulphur connectivities were obtained using the NMR-active cadmium isotope 113Cd and added to H NMR and solution structures of ZnxCu3aMT-1 and ZnyCu4bMT-1 The NOESY spectra of the ZnxCu3aMT-1 and ZnyCu4bMT-1 domains are markedly different from the starting Zn4aMT-1 and Zn3bMT-1 derivatives, pointing to a different arrangement of the polypeptide chains, which is probably needed to accommodate the resulting Cu(I)- or mixed metal–sulphur clusters (Fig. 6). From a cursory inspection of the superimposed spectra, it has already become clear that the addition of Cu(I) not only leads to a completely different pattern of spin systems, but also to a significantly higher number of NOEs. Therefore, we expected the structures of the newly identified Cu(I)-containing domains to be more rigid and distinct from their Zn(II)-containing forms. The spectra of both the only Zn(II)-containing b-domain and its Cu(I)4 derivative seem to be of lower quality with large areas of broad overlapping peaks. This behaviour might be due to higher flexibility within the b-domain which has been reported already before [17–20]. TOCSY spectra (not FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS 2355 Murine Cu(I)a- and Cu(I)b-MT-1 domains B. Dolderer et al. Fig. 5. Superposition of the present family of 30 structures of Zn4aMT (blue) with the published average structure of Cd4aMT (red). The coordinates of the Cd4aMT structure were extracted from the Brookhaven protein data bank (1DFS). In the last run of structure calculations for Zn4aMT 398 upper distance limits (upl) obtained from the assignment of its NOESY spectrum and the metal-sulphur connectivities reported by Zangger et al. [20] for Cd4aMT were used as input for the program DYANA. Twenty out of 200 structures were combined into a structure family with a target function of 0.52 ± 0.13 Å2 and RMSD values of 1.04 ± 0.12 Å and 1.51 ± 0.18 Å for the backbone and all heavy atoms, respectively. Fig. 6. 1H-1H-NOESY spectra of Zn4aMT (red) superimposed with that of ZnxCu3aMT (green) (A) and of Zn3bMT (red) superimposed with that of ZnyCu4bMT (green) (B). Sample and measurement conditions were the same as described in Fig. 3. shown) were recorded for the ZnxCu3aMT-1 and ZnyCu4bMT-1 derivatives and used, together with the NOESY spectra, to obtain their sequence specific assignments (supplementary Tables S1 and S2). The resonances of the new Cu(I)-containing species differed mostly by several tenths of a p.p.m. from those of their starting points, thereby confirming the observations mentioned above. In the NOESY spectrum of ZnxCu3aMT-1 502, NOEs were assigned, integrated and converted into distance constraints. In the last dyana run, a set of 200 structures was generated out of which the 20 best were combined to a structure family (Fig. 7). The target function was 0.43 ± 0.10 Å2, and the rmsd values 2356 were 0.70 ± 0.12 Å and 1.03 ± 0.14 Å for the polypeptide backbone and heavy atoms, respectively. Likewise, 500 NOEs of the ZnyCu4bMT-1 spectrum were used to derive a structure family for this species (Fig. 8). In this case, a target function of 0.19 ± 0.02 Å2 and rmsd values of 0.49 ± 0.21 Å and 0.72 ± 0.21 Å for the backbone and heavy atoms were obtained. As with other known structures of MTs, those presented here also lack typical secondary structure elements. The structure of ZnxCu3aMT-1 is roughly a two-level structure (Fig. 7) where the segment 5–10 of the a-domain polypeptide backbone forms the first level. The stretch 10–16 links the first with the second FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS B. Dolderer et al. Murine Cu(I)a- and Cu(I)b-MT-1 domains Fig. 7. Stereoview of the 20-structure family of ZnxCu3aMT. Polypeptide backbone bonds are shown in grey, cysteinyl side chain bonds in blue and sulphur atoms as yellow spheres. 502 NOEs were converted into upl and were used as input for the structure calculation program DYANA. Twenty out of 200 structures were combined into a structure family with a target function of 0.43 ± 0.10 Å2 and RMSD values of 0.70 ± 0.12 Å and 1.03 ± 0.14 Å for the backbone and all heavy atoms, respectively. Fig. 8. Stereoview on the family of 20 best structures of ZnyCu4bMT. Polypeptide backbone bonds are shown in grey, cysteinyl side chain bonds in blue and sulphur atoms as yellow spheres. 500 NOEs were converted into upl and were used as input for the structure calculation program DYANA. Twenty out of 200 structures were combined into a structure family with a target function of 0.19 ± 0.02 Å2 and rmsd values of 0.49 ± 0.21 Å and 0.72 ± 0.21 Å for the backbone and all heavy atoms, respectively. level, which is constituted by the second half of the polypeptide chain. The region between residue 20 and 26 includes a loop that is more disordered than the rest of this domain, shielding the rear of its core. The positions of several cysteine sulphur atoms are not very well defined (Fig. 7). Nevertheless, the 11 cysteines seem to form a somewhat broader cavity than in its Cd4aMT-1 counterpart, with the subgroup of Cys3, 11 and 18 being rather isolated from the other eight cysteines. The polypeptide backbone of the b-domain wraps around its core in a right-handed fashion (Fig. 8). The central part of its polypeptide chain, residues 8–24, limits an almost elliptical planar structure. At the point where the two ends of the polypeptide chain would meet, they leave the central plane and continue in opposite directions. Again, outside the uncertainty in the positions of some of the sulphur atoms, the cavity encased by the cysteines is somewhat broader than its Cd3bMT-1 counterpart, although in this case the nine cysteines still point to a unique core. The superposition of the newly identified Cu(I)-containing domains onto the mean structures of their Cd(II)-coordinating forms highlights the striking structural differences that are caused only by the binding of different metal ions to the two domains of MT (Fig. 9). When the structures are superimposed throughout the full length of their polypeptide chains only very faint similarities of some parts of the polypeptide backbone folds could be observed. Separate superpositions of stretches 3–12, 11–20 and 18–28 were also attempted. For both domains, the first and third stretches gave very poor overlap, while the central part showed a more pronounced similarity. This could indicate that each domain adapts to host the additional copper(I) ions by opening up and rearranging its N- and C-terminal parts, minimizing the structural perturbation of its central part. The arrangement of the cysteine sulphur donor atoms within the two Cu(I)-containing domains is also shown in Figs 7 and 8. Although, from the present data, it is neither possible to guess the number of FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS 2357 Murine Cu(I)a- and Cu(I)b-MT-1 domains B. Dolderer et al. Fig. 9. (A) Stereoview of the superposition of the Cd4aMT-1 mean structure (red) to the structure family of ZnxCu3aMT-1 (blue). (B) Stereoview of the superposition of the Cd3bMT-1 mean structure (red) to the structure family of ZnyCu4bMT-1 (blue). residual Zn(II) ions in each structure nor the overall topology of the clusters, it appears that all cysteine residues point towards the inside of the respective domain, as expected if all of them are to be involved in metal coordination. In turn, the somewhat broader cavities encased by the cysteines are consistent with the increased number of metals in each domain. At this point it was still an open question as to whether or not the single species ZnxCu3aMT-1 and ZnyCu4bMT-1 would be stable when the other Cu(I)containing domain was also present in solution. To this end, both Cu(I)-containing domains were prepared as for the titration experiment mentioned above, combined in equal amounts at a final concentration of 1 mm each and incubated for >48 h before the measurement of their 1H NMR spectra. The observed NOESY and TOCSY spectra (not shown) consisted of the sum of the respective spectra of the single species. The spectral resonances were assigned to all the protons present in the two domains and are listed in supplementary Tables S1 and S2. This result indicates that the two domains are stable and independent from each other. Cu(I) is not transferred between the two domains to form new species with higher and lower Cu(I):poly2358 peptide stoichiometries. As no additional NOEs and ⁄ or changes of the spectral resonances were observed in the NOESY spectrum of the mixture, an interaction of the two single domains and the formation of one single Cu(I)-containing domain with the involvement of both domains could be excluded for the present Zn(x+y)Cu7MT-1 stoichiometry. Conclusion The Cu(I) titration of the independent Zn(II)-loaded domains of mouse MT-1 revealed Cu(I) stoichiometries of three and four for the a- and b-domain, respectively. The presence of Cu(I) led to dramatic conformational changes of both polypeptide folds. Cu(I) stoichiometries of up to six Cu(I) ions each led to the progressive disappearance of the altered structures. [Correction added after publication 30 March 2007: in the preceding sentence, disappearing of the affered structurer, was corrected to disappearance of the altered structures]. Unfortunately, due to the lack of metal Æ sulphur constraints, the Cu(I) positions within the resolved polypeptide folds remained unclear. Therefore, crystallization of the newly identified Cu(I)-containing species FEBS Journal 274 (2007) 2349–2362 ª 2007 The Authors Journal compilation ª 2007 FEBS