Introduction

Cadmium is a metal that is well known for being toxic to organisms, in general, and to plants, in particular, where it causes severe metabolic malfunctions leading to intense chlorosis and growth impairment. Consequently, plants have developed efficient defence systems against cadmium toxicity, which mainly consist of chelating polypeptides that immobilize metal ions inside the cell. Two types of plant (including algae) metal-chelating peptides have been reported, enzymatically synthesized phytochelatins (PCs) and gene-encoded metallothioneins (MTs) [1]. Plant PCs, such as yeast cadystins, are polymers of glutamic acid–Cys γ-dipeptide linked to a terminal glycine residue, the number of units in the polymer ranging from 5 to 17 [2]. They bind CdII through metal–thiolate bonds, forming CdII PC complexes of variable size [3]. These complexes typically include acid-labile sulfide ligands, also in a variable metal to sulfide to peptide ratio, which contribute to the formation of semicrystal particles known as crystallites, analogous to those extensively studied in yeasts [4]. Arabidopsis mutants lacking PC synthase have a definite cadmium-sensitive phenotype [5], and cadmium tolerance has been related to CdII PC accumulation in tobacco [6], tomatoes [7] and maize [8].

PCs were considered the only metal-defence mechanism in plants until 1980, when an MT-like peptide was first isolated in copper-treated Agrostis (redtop bent) roots [9], more than 20 years after the discovery of MT in animals. MTs are ubiquitous, small, Cys-rich proteins that chelate heavy-metal ions through metal–thiolate bonds. Currently, MTs have been extensively identified as a multigenic family in angiosperms (A. thaliana as a model [1]), in gymnosperms [10] and in algae (Fucus) [11], constituting family 15 of the global MT Kägi classification [12]. Plant MTs are considerably longer than their animal counterparts owing to the exclusive presence of a 30–50-residue-long, Cys-devoid region, between the N- and C-terminal Cys-rich domains (four to eight Cys each). Specifically, the distribution of the Cys residues and the length of the spacer region have been used to further classify plant MTs into four subtypes [1, 13]. Although plant MTs have been extensively related to housekeeping functions in physiological zinc and copper metabolism [14, 15] and in reactive oxygen species scavenging [1, 16, 17], early studies report that plant MT synthesis also responds to cadmium induction [17, 18]. Confirmation of the putative cadmium detoxification role of plant MTs was primarily achieved by yeast complementation studies [19]. More recently, it has been directly shown in plant cells that MTs mediate resistance and tolerance to cadmium [20, 21]. Strikingly, very little is known about the CdII-MT complexes that are formed upon plant MT synthesis in response to cadmium, mainly owing to the high level of proteolysis associated with native protein purification. Consequently and in comparison with the structural knowledge of animal MT complexes [22], there is an appalling lack of data about the interaction between heavy-metal ions and plant MTs. Unfortunately, many initial efforts in recombinant (Escherichia coli) plant MT synthesis did not make it possible to overcome this drawback [23, 24], and MT complexes were directly characterized as fusion proteins [25], which is of dubious biological relevance. Fortunately, this scenario is beginning to change, and the characterization of recombinant Triticum aestivum (wheat, [26]) and Musa acuminata (banana, [27]) MT complexes was recently reported.

Some time ago, we adapted our glutathione S-transferase based MT expression system in E. coli, which we have fully validated for animal MTs [28, 29], to obtain highly homogeneous preparations of undigested, full-length metal complexes from a typical dicot angiosperm MT (Quercus suber MT, QsMT) [30]. QsMT is a type 2 plant MT isolated in our laboratory from a cork oak (Q. suber) phellem complementary DNA (cDNA) library of oxidative stress induced genes. Protein expression and purification from E. coli cells grown in the presence of zinc, cadmium or copper enabled us to determine the ZnII-, CdII- and CuI-binding properties of the full-size peptide. Our results showed that when expressed in the presence of cadmium, recombinant QsMT (rQsMT) binds a surprisingly high content of CdII in comparison with ZnII, confirming that this protein could play an important role in the heavy-metal detoxification of plants [30]. Interestingly, our results also suggested the presence of acid-labile sulfide ligands in the CdII-rQsMT complexes, concordantly with the sulfide anions mediating the formation of the CdII PC crystallites. This was the first time that the participation of sulfide ligands was suspected in any MT preparation, and led to the significant discovery that, although at different ratio, all MT recombinant complexes with divalent metal ions may contain these nonproteic ligands [31]. To gain some insight into the metal cluster structure and protein folding of plant MTs we then engineered three QsMT-derived peptides: the N-terminal Cys-rich domain (N25), the C-terminal Cys-rich domain (C18) and a chimera where both Cys-rich domains were linked by a four-Gly bridge (N25-C18) instead of the original linker region of 39 amino acids (Scheme 1). Expression of these constructs in the presence of copper or zinc allowed us to analyse the binding properties for these metal ions, and to propose a protein folding model in a single metal cluster formed by the interaction of both Cys-rich domains where the linker domain, though not participating in metal coordination, is important for the stability and function of the protein [32].

Scheme 1
scheme 1

Amino acid sequences of the wild-type recombinant Quercus suber metallothionein (rQsMT) and of the three deletion mutants as constructed in [32]: N25, the rQsMT N-terminal region, containing the first eight Cys; C18, the QsMT C-terminal region, containing six Cys; and N25-C18, the fusion of N25 and C18 through a flexible bridge of four Gly (box), thus devoid of the spacer region of QsMT. Additional Gly and Ser are present in the N-terminus of the four peptides owing to the recombinant synthesis strategy [28]

Full size image

In the current study, we applied the same rationale to analyse the CdII-binding features of rQsMT at stoichiometric and conformational levels. This is especially interesting owing to the participation of sulfide ions in CdII-rQsMT. Thus, the four QsMT peptides (wild-type rQsMT, N25-C18, N25 and C18) were purified from E. coli cells grown in the presence of cadmium, and their in vivo CdII-binding ability was determined through the chemical, spectroscopic and spectrometric characterization of the corresponding clusters. Then, the in vitro ZnII/CdII replacement reactions were studied, as well as the in vitro acid denaturation and sulfide renaturation reactions. Finally, to test the capacity of the four peptides for preventing cadmium deleterious effects in yeast cells, a functional approximation was performed through yeast complementation assays. All the results enable us to suggest, for the first time, a metal-binding and folding model for a typical plant CdII-MT complex.

Materials and methods

Recombinant synthesis and purification of the ZnII and CdII complexes of wild-type QsMT, N25-C18, N25 and C18

Isolation of the QsMT cDNA, construction of the N25, C18 and N25-C18 coding sequences, and cloning into the pGEX expression vector have been previously described [30, 32]. E. coli BL21 cells transformed with the respective recombinant plasmids pGEX-QsMT, pGEX-N25, pGEX-C18 and pGEX-N25-C18 were grown in the presence of 300 μM ZnCl2 or CdCl2 and hence used for recombinant syntheses. Expression and purification were performed as reported in [31], so all MT complexes were recovered in 50 mM tris(hydroxymethyl)aminomethane hydrochloride pH 7.0 solution, and were kept at −70 °C until used.

Chemical, spectroscopic and spectrometric characterization of the metal peptide complexes

Following the procedures already described by our group [31, 32], acid inductively coupled plasma atomic emission spectroscopy (ICP-AES) and amino acid analysis were used to determine the protein concentration of the different ZnII- or CdII-containing preparations. Their metal-to-protein ratios were also deduced from the acid ICP-AES measurements and their mean sulfide-to-protein contents were estimated by gas chromatography–flame photometric detection (GC-FPD) [31]. The use of Na2SO4 as an ICP-AES standard for the Cys- and Met-derived sulfur quantification in MTs was validated by Bongers et al. [33]. However, as we reported in [31], Na2SO4 cannot be used as a standard for sulfide sulfur determinations as both types of sulfur enter into the plasma phase differently. Consequently, the S2−-to-protein ratios cannot be obtained by direct subtraction of the acid from the conventional ICP-AES values, although both types of data are included in Tables 1 and 2. A Polyscan 61 E (Thermo Jarrell Ash) spectropolarimeter and an Alpha Plus amino acid autoanalyser (Pharmacia LKB Biotechnology) were respectively used for the ICP-AES measurements and amino acid analysis. An HP 5890 series II gas chromatograph coupled to an FPD80 CE detector (Thermo Finnigan) was employed for the GC-FPD sulfide quantifications.

Table 1 Analytical characterization of the recombinant ZnII complexes of Quercus suber metallothionein (QsMT) and the three QsMT-derived peptides (N25-C18, N25 and C18)
Full size table
Table 2 Analytical characterization of the recombinant CdII complexes of QsMT and the three QsMT-derived peptides (N25-C18, N25 and C18)
Full size table

The in vitro CdII-binding analyses were performed by CdII titration of the ZnII peptides as described elsewhere [28], and were monitored spectroscopically and spectropolarimetrically. Electronic absorption measurements were performed using an HP-8453 diode array UV–vis spectrophotometer. A JASCO spectropolarimeter (J-715) interfaced to a computer (GRAMS/AI 7.02 software) was used for circular dichroism (CD) determinations. All manipulations involving metal ion and protein solutions were performed under an argon atmosphere, and titrations were carried out at least in duplicate to ensure reproducibility. The pH for all experiments remained constant throughout, without further addition of buffers, and the temperature was kept constant at 25 °C by means of a Peltier PTC-351S apparatus.

The molecular mass of the metal peptide species was determined by electrospray ionization (ESI) mass spectrometry (MS) performed either with a Fisons Platform II instrument (VG Biotech) controlled by MassLynx software following the same conditions previously described [32] or with an Ultima Micromass quadrupole time of flight (QTOF) instrument (ESI-QTOF), also controlled by MassLynx software and calibrated with NaI (0.2 g NaI dissolved in 100 ml of a 1:1 H2O/2-propanol mixture). In the ESI-TOF analysis of the metallopeptides, 5 μl of the sample was injected at 40 μl/min under the following conditions: source temperature, 150 °C; desolvation temperature, 250 °C; capillary counter electrode voltage, 3.0 kV; cone potential, 80 V. Spectra were collected throughout an m/z range from 950 to 2,150 at a rate of 2 s per scan with an interscan delay of 0.1 s. The liquid carrier was a 10:90 mixture of acetonitrile and 5 mM ammonium acetate, pH 7. For analysis of the apo form, the samples were demetalated by acidification with HCl at pH 1.5 and MS measurements were carried out as explained for the holo forms, except that the liquid carrier was a 10:90 mixture of methanol and ammonium formate/ammonia at pH 1.5. In all cases, the theoretical molecular masses were calculated according to [32] except for the sulfide-containing species, where two additional protons were added per sulfide anion.

Demetalation and reconstitution of the CdII complexes of rQsMT and the three derived peptides

Two different strategies were used for the demetalation of the MT complexes in this work: acidification and EDTA treatment. For acidification, and according to equivalent experiments with CdII PC complexes [34], the four CdII peptide preparations were acidified from neutral pH to a pH lower than 1 with 1−10−3 M HCl depending on the stage of the titration, and were reneutralized afterwards to pH 7.0 with 1−10−3 M NaOH, also depending on the stage of the titration. After reneutralization, several molar equivalents of a standard solution of Na2S prepared as described in [31] were added. All the CD and UV–vis changes experienced by the samples during these pH variations and sulfide additions were recorded and corrected for dilution effects. When possible, ESI-MS analyses of the intermediate and resulting final solutions were also performed.

According to procedures reported in the literature [35], a 16 μM solution of CdII-rQsMT at pH 7.5 was treated with 10–50 mM EDTA, depending on the stage of the titration, and the spectropolarimetric changes were recorded.

Yeast functional complementation assays

Following the details reported in [32], two MT-deficient, copper-sensitive Saccharomyces cerevisiae strains were used, cup1 S: DTY3 (MATa, leu2-3, 112his3 Δ 1, trp1-1, ura3-50, gal1 CUP1 S), harbouring only one copy of the CUP1 MT gene; and cup1 Δ: DTY4 (same with cup1::URA3), thus with no copy of CUP1 [36]. The growth of yeast cells transformed with the plasmids p424-QsMT, p424-N25-C18, p424-N25 or p424-C18, constructed as previously described [30, 32, 37], was assayed in culture media supplemented with or without CdCl2 (1.5, 2.5 or 3.5 μM for the plate).

Results and discussion

The metal complexes rendered by the three QsMT-derived peptides (N25-C18, N25 and C18) when biosynthesized in zinc- or cadmium-enriched media were analysed and characterized by spectroscopic and spectrometric methods, and the data were compared with those of the full-length rQsMT [30, 32]. Independently of the metal ions supplemented in the media, acidification to pH 1.0 of each recombinant peptide yielded single apo forms whose molecular masses were in accordance with the values calculated from their amino acid composition [32], this confirming their identity and integrity. None of the CD spectra of the four demetalated peptides exhibited absorptions in the 220–400-nm range, which is especially significant in the case of apo-QsMT, as this indicates that the aromatic residues of the spacer region are CD-silent. However, the UV–vis spectrum of apo-QsMT showed absorptions in the range 260–280 nm (Fig. 3j) attributable to the two Phe residues of the spacer.

The ZnII-binding features of rQsMT and derived peptides: a deeper insight

The in vitro ZnII/CdII replacement studies of the four recombinant peptides required biosynthesis and analytical characterization of the corresponding ZnII complexes, previously characterized in [32]. However, our current knowledge of the presence of sulfide ligands in the recombinant MT species [31] together with the use of ESI-QTOF allowed refining of our previous data [32], particularly their metal and S2− contents (Table 1). The present results revealed that ZnII-C18 was the only case where S2− ligands were not detected by GC-FPD. In contrast, rQsMT, N25-C18 and N25 gave rise to minor S2−-containing species. It should be noted that, as already reported in [31], we have found that GC-FPD always overestimates the S2− content of the MT samples. Consequently, there is a discordance between the S2−/protein quantification achieved by GC-FPD and the stoichiometries and relative abundances of the MT species detected by ESI-MS.

Analysis of the CdII-binding features of rQsMT

Multiple recombinant syntheses of the full-length QsMT in cadmium-rich medium yielded three kinds of in vivo preparations, namely rQsMT types 1, 2 and 3, which could not be related to specific culture conditions. According to the MS data shown in Table 2, CdII-rQsMT(1) and CdII-rQsMT(2) showed identical speciation: Cd6S4 and Cd7S4 as the most abundant species, while CdII-rQsMT(3) yielded major Cd5 and minor Cd6S4 complexes. Therefore, the unknown ligand of [30] could be readily identified as four S2− anions. In any case, none of the CdII-rQsMT complexes were either isostoichiometric or isostructural to their ZnII-rQsMT counterpart. Although all CdII-rQsMT types consisted of CdII homometallic samples of close speciation, their CD fingerprints were markedly dissimilar (Fig. 1a). Interestingly, these CD features were exchangeable by in vitro acidification or demetalation treatments, as explained further below.

Fig. 1
figure 1

Comparison of the normalized circular dichroism (CD) spectra of the following recombinant metallothionein (MT) preparations: a CdII-rQsMT(1) (dotted line), CdII-rQsMT(2) (dashed line), CdII-rQsMT(3) (solid line) b CdII-N25-C18 (solid line), CdII-N25 (dashed line), CdII-C18 (dotted line). rQsMT recombinant Quercus suber MT

Full size image

Overall analysis of the in vitro ZnII/CdII replacement in ZnII-rQsMT led us to propose Cd4-QsMT (major), Cd5-QsMT, and several minor S2−-containing complexes of close metal stoichiometry, as the final products of this reaction, even in the presence of excess CdII (whole spectroscopic and spectrometric data included in Fig. 2, Table S1). These species were similar to those of the CdII-rQsMT(3) preparation, but were absent in CdII-rQsMT(1) and CdII-rQsMT(2) samples (Table 2). Notably, none of the CD fingerprints of the three CdII-rQsMT types were reproduced during the CdII titration (Figs. 1a, 2a, b). As the main difference between the in vivo and in vitro CdII-binding abilities of QsMT was the presence of S2−-rich Cd6S4-rQsMT complexes (in the in vivo samples), we extended the CdII titration by gradually adding Na2S after 7 equiv CdII had been added to ZnII-rQsMT. This gave rise to an absorption increase in the 260–320-nm region (Fig. 2c, d, g, h, k, l), in accordance with the S2− anions being incorporated into the CdII complexes [34]. The final CD fingerprints (Fig. 2d) clearly evolved towards the envelopes recorded for CdII-rQsMT (Fig. 1a). Despite the drawbacks associated with generation of Na+ adducts, ESI-MS analysis of the final samples revealed the presence of S2−-containing species with higher nuclearity than those present before Na2S addition (i.e. Cd7S9-QsMT and Cd6S6-QsMT, Table S1). All these data reinforce the hypothesis of S2− as a determinant of (1) the nuclearity of CdII-rQsMT complexes and (2) the differences between the biosynthesized CdII-rQsMT samples and the in vitro constituted CdII-QsMT complexes.

Fig. 2
figure 2

CD (a–d), UV–vis (e–h) and UV–vis difference (i–l) spectra corresponding to the titration of a 20 μM solution of Zn-rQsMT with Cd(II) at pH 7.0 followed by the addition of several equivalents of Na2S. Arrows show the evolution of the spectra when the indicated number of Cd(II) or S2− equivalents were added

Full size image

Finally, acidification/reneutralization of the three CdII-rQsMT preparation types shed light on their different nature. Acidification of CdII-rQsMT(3) from pH 7 to 4.5 caused a decrease in the intensity of the CD shoulder at approximately 250 nm, to give rise to a CD profile very similar to that of CdII-rQsMT(1) (whose CD spectrum remains invariable between pH 7.0 and 4.3, Fig. S1), with one intermediate step corresponding to the CD of CdII-rQsMT(2) (Fig. 3a). Thus, at pH 4.5 any of the three CdII-rQsMT types exhibited the same CD spectrum [i.e. that of CdII-rQsMT(1)], which remained unaltered between pH 4.5 and 3.5 (Fig. 3b), to render at pH 1 the typical CD and UV spectra of an apo-MT with aromatic amino acids (Fig. 3c, d, i, j). The H2S odour was perceptible during the three acidifications, confirming the acid-labile character of the S2− ligands of the original complexes. Reneutralization up to pH 7 of the resulting S2−-devoid samples (Fig. 3e, k, q) gave rise to Cd4-QsMT (major) and Cd5-QsMT (minor) species (ESI-MS data not shown) whose CD envelopes evidently did not reproduce those of any of the CdII-rQsMT preparations. As before, addition of Na2S to these solutions caused dramatic changes to their spectroscopic features (Fig. 3f, l, r) and rendered S2−-containing complexes (Cd8S6 and Cd7S2) whose CD fingerprint resembled that of CdII-rQsMT(3) (Fig. 4a). Interestingly, during the demetalation of CdII-rQsMT(1) by EDTA (Fig. 4b, c), the addition of the first EDTA equivalent (Fig. 4b) increased chirality at approximately 280 nm, while not altering that at approximately 250 nm. Afterwards, increasing molar ratios of EDTA led to samples showing CD spectra similar to that of CdII-rQsMT(3) (Fig. 4c).

Fig. 3
figure 3

CD (a–f), UV–vis (g–l) and UV–vis difference (m–r) spectra corresponding to the acidification (a–d, g–j, m–p) and reneutralization (e, k, q) of a 20 μM solution of CdII-rQsMT(3); and addition of several Na2S equivalents (f, l, r) to the final reneutralized solution. Arrows show the evolution of the spectra during acidification and reneutralization processes. Curves in f, l and r correspond to the reneutralized CdII-QsMT solution of e (solid line) and those recorded after addition of 1 equiv (dashed line) and 4 equiv (dotted line) S2− to the former

Full size image
Fig. 4
figure 4

a Comparison of the CD spectra of CdII-rQsMT(3) (solid black line), the acidified/reneutralized CdII-QsMT sample (solid grey line) and with addition of 1 equiv (dashed line) and 4 equiv (dotted line) S2− to the previous sample. b, c CD spectra corresponding to the addition of the first EDTA equivalent and of 1, 1.5, 2, 3, 9 and 30 equiv EDTA to a 16 μM CdII-rQsMT(1) sample. The dotted line in c corresponds to the CD spectrum of the CdII-rQsMT(3) preparation. Arrows show the evolution of the spectra during the demetalation process

Full size image

The comprehensive consideration of all these data suggests that the heterogeneity of the CdII-rQsMT samples (types 1–3) was due to two main factors: (1) the already mentioned relative abundance of S2−-containing complexes in the sample and (2) the putative participation of the His of the spacer in cadmium coordination, as suggested by the following hints. First, preliminary Raman data revealed the presence of metal–His bonds in the CdII-rQsMT complexes, and their absence in ZnII-rQsMT [38]. Second, literature data suggest that CD shoulders at approximately 250 nm can be attributed to CdII–His coordination [39, 40], which we have corroborated with studies in mammalian MT1 mutants [41] and chicken MT [42]. And third, a differential His participation in CdII binding would be consistent with the initially different CD spectra of the three types of CdII-rQsMT converging to an identical CD fingerprint at pH 4.5, after His protonation. Finally, the EDTA-induced CdII displacement from the rQsMT complexes could cause conformational rearrangements allowing His participation in CdII coordination. This hypothesis is highly consistent with the fact that the lower the CdII content, the higher the chirality at approximately 250 nm [cf. rQsMT(3), Table 2, Fig. 1a].

Thus, all our data are in concordance with rQsMT(3) being mainly composed of S2−-devoid Cd5 complexes where His may participate in cadmium coordination; rQsMT(1) containing S2−-rich Cd6 and Cd7 complexes with no His participation; and rQsMT(2) being a mixture of rQsMT(1) and rQsMT(3). Therefore, when CdII-rQsMT is synthesized in E. coli, and depending on the folding that the protein adopts when binding the CdII ions, the His residue of the spacer may or may not participate in CdII binding, with this determining the stoichiometry and the conformation of the final complexes.

Analysis of the CdII-binding features of N25-C18

N25-C18 synthesized in cadmium-rich medium yielded homometallic CdII complexes, among which Cd6S4-N25-C18 and Cd5-N25-C18 were the most abundant species (Table 2). Thus, the CdII-N25-C18 complexes were neither isostoichiometric nor isostructural to their ZnII-N25-C18 counterparts, as neither were the rQsMT species. CdII-N25-C18 showed a characteristic CD spectrum composed of two Gaussian bands centred at approximately 250 (CdII thiolate) and 280 nm (CdII sulfide) chromophores (Fig. 1b), which was clearly different from those of the diverse CdII-rQsMT types (Fig. 1).

Titration of Zn4-N25-C18 with CdII (full data in Fig. S2, Table S2) rendered Cd5-N25-C18 as the major species even for an excess of CdII and resulted in a different CD spectrum from that of the in vivo sample. Surprisingly, although blueshifted it resembled that of CdII-rQsMT(1) (Fig. 5a). Addition of Na2S after the final titration step further increased this resemblance, with a clear indication of S2− CdII coordination (Fig. 5). These results are fully consistent with the previous hypothesis about the CdII-coordinating behaviour of His in rQsMT. Hence, N25-C18 that is devoid of this residue can reproduce the features of CdII-rQsMT(1), where we presume no CdII–His contributions, and never those of rQsMT(3).

Fig. 5
figure 5

a Comparison between the CD spectra of CdII-rQsMT(1) (solid line) and those recorded after the addition of 1, 2, 5, 7 and 10 equiv Na2S at the end of the titration of ZnII-N25-C18 with Cd(II), i.e. after 10 equiv CdII. b UV–vis spectra recorded during the addition of 1–10 equiv CdII to Zn4-N25-C18, leading to the formation of Cd5-N25-C18, followed by the addition of 1, 2, 5, 7 and 10 equiv Na2S. The difference in CD intensities between CdII-rQsMT(1) and the final CdII-N25-C18 sample and the deviations of the baseline of the UV–vis spectra are due to the turbulence of the final stages of Na2S additions, caused by precipitation of the excess CdII as CdS

Full size image

It is especially worth noting that titration of ZnII-N25-C18 with CdII and S2− yielded complexes more similar to CdII-rQsMT than to CdII-N25-C18. However, a process of acidification/reneutralization/S2− addition of the biosynthesized CdII-N25-C18 ended up with a CD fingerprint very similar to the initial one (Fig. S3). The interpretation of the spectroscopic data of these reactions was more straightforward than for CdII-rQsMT owing to the absence of His and Phe in the N25-C18 polypeptide. Thus, it could be deduced that acidification of Cd6S4-N25-C18 from pH 7 to 4.5 promoted an important structural rearrangement. The 250-nm Gaussian band became a derivative-shaped band at the same wavelength, so some CdII thiolate chromophores could be lost, while S2− would remain bound to the CdII ions. We cannot discard the migration of some thiolate-bound CdII to S2−-rich environments, as suggested by the UV–vis difference spectra in Fig. S3c. It was not until pH between 4 and 2 that CD absorptions at approximately 280 nm—together with those remaining at approximately 250 nm—disappeared, to generate a characteristic apo-MT spectrum. At this point, a strong H2S odour was perceptible and the CdII ions released to the solution visibly precipitated as CdS. In spite of the turbidity of the sample, it was reneutralized to pH 7. Reincorporation of the CdII ions to N25-C18 gave rise to an intense and very wide CD signal centred at approximately 260 nm and that was very different from that of the initial in vivo CdII-N25-C18 (Fig. 6a), as expected from the loss of most of the S2− ligands. This new CD fingerprint could be interpreted as being composed of one absorption centred at about 250 nm—attributable to CdII(SCys)4— and other absorptions in the 270–320-nm range—due to CdII-S2− if it is assumed that some CdS particles became trapped by some Cys residues. Although this may appear speculative, it is consistent with the observations that (1) the envelope of the CD spectrum of the reneutralized sample perfectly matched that recorded for the addition of 3 equiv CdII to Zn4-N25-C18 (Fig. 6a), a preparation that contained one S2− per MT (Table 1), and (2) that the tail of the CD absorptions extending until 300 nm could only be attributed to CdII–S2− chromophores. Subsequent Na2S addition to the reneutralized sample caused dramatic changes in the CD spectra already from the first step (Fig. 6b) to practically reproduce, for 3–4 equiv S2− added, the spectrum of the initial in vivo CdII-N25-C18. This final CD profile was not too different from that achieved by CdII-rQsMT(1) after a similar acidification/reneutralization/S2− addition process (Fig. 6b), which indicates that both polypeptides can, depending on the conditions, show similar CdII-binding behaviour when His does not contribute to CdII coordination.

Fig. 6
figure 6

a Comparison between the normalized CD spectra of recombinant CdII-N25-C18 (solid line), ZnII-N25-C18 after the addition of 3 equiv CdII (dotted line) and the reneutralized CdII-N25-C18 sample (dashed line). b Comparison between the CD spectra of recombinant CdII-N25-C18 (solid line), the reneutralized CdII-N25-C18 sample after the addition of 4 equiv S2− (dashed line) and the reneutralized CdII-QsMT sample after the addition of 1 equiv S2− (dotted line)

Full size image

Analysis of the CdII-binding features of the separate N25 and C18 peptides

The syntheses of the separate N25 and C18 peptides in cadmium-rich media yielded dimeric CdII homometallic complexes (Table 2). Cd7S4-(N25)2 and Cd6-(N25)2 were the main species of a CdII-N25 preparation exhibiting high sulfide content (2.8 S2− per peptide). Conversely, CdII-C18 was mainly composed of major Cd4-(C18)2 and minor Cd5-(C18)2 complexes (Fig. 7), in concordance with the very low S2− content detected by GC-FPD (0.5 S2− per peptide). The CD spectra of these samples (Fig. 1b) also reflected their differential S2− content, since CdII-N25, unlike CdII-C18, gave rise to CD absorptions at approximately 280 nm.

Fig. 7
figure 7

Electrospray ionization (ESI) mass spectrometry (MS) spectrum of the recombinant CdII-C18 preparation with indication of the theoretical molecular weights of the CdII-C18 species and the MS peaks expected for each charge state. The assessment of the presence of C18 dimers was made on the basis of a deconvolution method [32] allowing us to identify two types of ESI-MS peaks corresponding to dimeric forms: (1) peaks that match the m/z charge states of a (Cd n -MT)2 form (z being an odd value); (2) peaks that only match the molecular weight of two peptide chains binding an odd number of CdII ions. Furthermore, some peaks could be either interpreted as corresponding to a monomer of m/z or to a dimer of 2m/2z ratio

Full size image

The in vitro ZnII/CdII replacement followed by S2− addition, and the acidification/reneutralization/S2− addition studies were also undertaken for the separate N25 and C18 peptides (results summarized in Fig. 8, and full data included in Figs. S4–S7, Tables S3, S4). In vivo CdII-N25 aggregates could not be reproduced in vitro by any of the methods assayed. Starting from major monomeric Zn2-N25 species with a very low S2− content, the dimeric Cd7S4-(N25)2 complexes could hardly be obtained, considering that species with a maximum of three CdII ions were obtained at the end of the titration. Remarkably, acidification and reneutralization of in vivo Cd7S4-(N25)2 did not lead to the original complexes. However, the addition of Na2S either at the end of the CdII titration or after reneutralization gave rise to CD envelopes that practically coincided with that obtained for the acidification at pH 4 of in vivo CdII-N25 (Fig. 8a). This suggests that N25 is unable to achieve in vitro the same folding as in in vivo conditions, which basically implies dimerization and participation of S2− ligands.

Fig. 8
figure 8

a Comparison between the normalized CD spectra of recombinant CdII-N25 (solid black line), ZnII-N25 after the addition of 9 equiv CdII (solid grey line), ZnII-N25 after the addition of 10 equiv CdII and 4 equiv S2− (dotted grey line), the CdII-N25 sample acidified to pH 4 (dashed black line) and the reneutralized CdII-N25 sample after the addition of 1 equiv S2− (dotted black line). b Comparison between the CD spectra of recombinant CdII-C18 (solid black line), the reneutralized CdII-C18 sample after the addition of 2 equiv S2− (dashed line) and the ZnII-C18 sample after the addition of 2 equiv CdII (dotted line)

Full size image

In a completely different scenario, the monomeric Zn2-C18 complexes, where S2− was not detected by GC-FPD, easily rendered, after addition of 2 equiv CdII, Cd4-(C18)2 dimers that exactly reproduced the CD fingerprint of the in vivo CdII-C18 preparation (Fig. 8b). As acidification and reneutralization of the in vivo CdII-C18 sample required just a small amount of Na2S to regenerate the initial CD envelope, it is sensible to assume the presence of minute amounts of S2− in ZnII-C18, enough to yield in vitro the S2−-containing CdII-C18 complexes.

Finally, the CdII titration of an equimolar mixture of ZnII-N25 and CdII-C18, hereafter referred to as cotitration, was performed to analyse possible interactions between the separate N25 and C18 peptides (Fig. 9, Table S5). The CD spectrum of the initial mixture perfectly matched the sum of the spectra of both separated ZnII complexes (Fig. 10a), which suggests that they do not interact in solution. As this CD fingerprint was clearly different from that of ZnII-N25-C18, a dumbbell fold for this chimeric ZnII peptide could already be ruled out. The ZnII-N25 plus ZnII-C18 mixture (Fig. 9) saturated for 7 equiv Cd(II) added, yielding a CD fingerprint (Fig. 10b) practically coincident with the sum of the final spectra of the separate CdII titrations of ZnII-N25 and ZnII-C18. In these reactions, major Cd3-N25 and Cd4-(C18)2 complexes were respectively formed; the same species that were detected as major products of the cotitration (ESI-MS analysis in Table S5). Thus, both separate peptides behaved equally when titrated alone or in each other’s presence: N25 evolving from monomeric Zn2 to monomeric Cd3 species, and C18 from monomeric Zn2 to dimeric Cd4 complexes. The difference between the final CD fingerprint of the cotitration and those of the in vivo CdII-N25-C18 (Fig. 10b) or those reached at the end of the ZnII-N25-C18 titration with CdII (Fig. S2) implies a dependent behaviour of both regions in the N25-C18 polypeptide when coordinating CdII. It is worth noting that heterodimers (N25/C18) were also detected at the end of the cotitration, although as minor species (Table S5), which is highly significant in order to support a hairpin model for CdII-N25-C18 (discussed later). The small amount of heterodimers is consistent with the low peptide concentration at which the cotitration was performed to allow monitoring by CD, and with the fact that N25 seems not to require interaction with other peptides (same N25 or C18) to form CdII complexes in solution.

Fig. 9
figure 9

CD (a–c), UV–vis (d–f) and UV–vis difference (g–i) spectra corresponding to the titration of a solution of 20 μM ZnII-N25 and 20 μM ZnII-C18 with CdII at pH 7.0 followed by the addition of 1 and 2 equiv Na2S. Arrows show the evolution of the spectra when the indicated number of CdII or S2− equivalents were added

Full size image
Fig. 10
figure 10

a Comparison between the CD spectra of the mixture of equimolar amounts of ZnII-N25 and ZnII-C18 (solid black line) with the sum of the CD spectra of ZnII-N25 and ZnII-C18 (solid grey line). The CD spectra of ZnII-N25 (dotted line), ZnII-C18 (dashed grey line) and ZnII-N25-C18 (dashed black line) are also included. b Comparison between the CD spectra of the mixture obtained in the cotitration of ZnII-N25 and ZnII-C18 after the addition of 7 equiv CdII (solid black line) and the sum of the final CD spectra of the separate titrations of ZnII-N25 and ZnII-C18 with CdII (solid grey line). The CD spectrum of recombinant CdII-N25-C18 (dotted line) is also included

Full size image

QsMT cadmium detoxification capacity in yeast

To test whether the QsMT-derived peptides provided protection against cadmium toxicity, and to what extent, N25, C18 or N25-C18 were expressed in CUP1-deficient yeast cells (cup1 Δ). Cells transformed with the nonrecombinant plasmid or cells synthesizing the full-length QsMT were used as negative and positive controls, respectively. In the absence of supplemented cadmium, all the strains yielded colonies of similar size (Fig. 11a), this showing that the presence of the QsMT peptides had no inherent effect on growth. Phenotype recovery was then evaluated in terms of capacity for growing in the presence of cadmium. Control p424 cup1 Δ cells were sensitive to CdII concentrations as low as 1.5 μM, whereas the same cells synthesizing QsMT were able to grow at a similar rate as cup1 S at 1.5 μM CdII, which is definitely better than the parental strain at 2.5 and 3.5 μM CdII (Fig. 11b). This is highly consistent with the copper thionein character of the endogenous yeast MT, since we have shown that the single copy of CUP1 present in the cup1 S strain is able to exhibit fairly normal growth under copper stress [32]. Cells synthesizing the QsMT-derived peptides exhibited a markedly reduced growth rate in relation to cells synthesizing QsMT. The higher the CdII concentration, the greater the disparity in growth rate between the pQsMT-transformed strain and the other three strains. This result is especially significant for N25-C18, which has the same number of Cys as QsMT and which yields aggregates of equivalent CdII and S2− content (Table 2), with the only difference being the lack of the spacer region. This finding fully corroborates the same behaviour we previously reported for copper stress [32]. The observation that the plant MT spacer is crucial for its in vivo metal detoxification function was already reported in [19] after comparison of the cadmium tolerance exhibited by yeast cells expressing different Arabidopsis MTs: MT1, an isoform naturally devoid of a spacer region, and MT2, an isoform with the typical plant MT sequence. The differences were then attributed either to the presence of the central domain or to the different arrangement of the Cys residues between both Arabidopsis MTs, but this second possibility can be now fully ruled out.

Fig. 11
figure 11

Yeast functional complementation assays. The cup1 S strain presents only one copy of the CUP1 gene, coding for an MT, while the cup1 Δ strain includes no copy of this gene. cup1 Δ cells have been transformed with the plasmid p424 without insertion, or with the constructions p424-QsMT, p424-N25-C18, p424-N25 or p424-C18. For the metal tolerance tests, transformed cup1 Δ cells were initially grown in selective SC-Trp-Ura medium and cup1 S strain in SC medium, both at 30 °C and 220 rpm to an optical density at 600 nm of 0.5–0.7. Cultures were then tenfold serially diluted three times, and 3 ml of each final sample was spotted on SC medium plates, supplemented or not supplemented with cadmium. Plates were incubated for 3 days at 30 °C and photographed. a Control SC medium without cadmium, to asses the viability of all the transformants. b The same medium supplemented with 1.5, 2.5 or 3.5 mM CdCl2. The first column of each assay corresponds to the original culture, and each of the subsequent columns to its sequential tenfold dilution, as explained before

Full size image

Conclusion

A comprehensive evaluation of all the data gathered provides a first approach to the structure/function relationship in a typical type 2 plant MT (QsMT) when coordinating CdII, which completes our previous studies of the ZnII- and CuI-binding abilities of this same MT [30, 32]. For the sake of clarity, a synopsis of the stoichiometric and spectroscopic results is included (Scheme 2) with indication of the precise figure and/or table where they are shown.

Scheme 2
scheme 2

The proposed composition and fold of the ZnII and CdII complexes of recombinant a QsMT and N25-C18 and b N25 and C18. In vivo indicates a complex directly purified from recombinant synthesis, and in vitro refers to complexes obtained by the in vitro reaction indicated. For the sake of clarity the sulfide anions are shown reduced. When possible, the interrelationship between species has also been shown, as well as all the figures and tables from which the results have been drawn. The symbols used are detailed in the inset to b. H + sample acidification, OH − sample reneutralization, S 2− sulfide addition, grey species of uncertain metal-to-protein stoichiometry, dashed arrows equivalences deduced from similar CD spectra

Full size image

All our current results are in full concordance with our previous assumption of rQsMT folding into a hairpin structure upon ZnII coordination, enclosing four ZnII ions and a low number of S2− ligands, with no hint of participation of either the spacer region or, consequently, its His residue [32]. A hairpin model can now be also proposed for the in vivo folded CdII-rQsMT complexes. The ready dissimilarity between the rQsMT CdII-binding capacity and that deduced from the addition of those of N25 and C18 clearly rules out domain independence, thus discarding a dumbbell-like fold. Furthermore, the major species in the CdII-N25 and CdII-C18 preparations were dimeric CdII complexes, which further supports a hairpin model for CdII-rQsMT, as a dumbbell fold would rely on the ability of each Cys-rich region to fold into a monomeric metal complex. Consistently with the stoichiometric data, the CD analyses clearly reveal that the sum of CdII-N25 and CdII-C18 spectra is far from reproducing the CD fingerprint of any of the CdII-rQsMT types.

The unexpected recovery of distinct CdII-rQsMT types (Table 2) can be fully explained by assuming two alternative global conformations for the same hairpin fold. Hence, CdII-rQsMT(1) would be mainly composed of complexes containing six Cd(II) ions, with participation of four S2− ligands but with no indication of the spacer His residue contribution. Conversely, CdII-rQsMT(3) would mainly consist of complexes of lower CdII content, (five CdII) and devoid of S2− ligands, in which there are indications of CdII-His coordination, and therefore of the contribution of the spacer to the cluster architecture. This hypothesis is fully supported by the in vitro interconvertibility between both CdII-rQsMT types, inducible by slight acidification or demetalation treatments (Scheme 2a). Furthermore, it is also consistent with the facts that (1) complexes with at most five CdII, analogous to those of CdII-rQsMT(3), are obtained from the low-S2−-containing Zn4-rQsMT species by in vitro ZnII/CdII replacement and (2) the subsequent addition of S2− to the end of this titration renders species similar to those of CdII-rQsMT(1).

Following a similar reasoning to that used for rQsMT, our current data also suggest a hairpin model when N25-C18 binds ZnII or CdII in vivo, although we rated this possibility as second best for ZnII-N25-C18 in previous studies [32]. CdII-N25-C18 shares metal and sulfide content with CdII-rQsMT, but exhibits chirooptical properties different from those of all CdII-rQsMT types, which suggests that even if the spacer does not contribute to metal coordination, its presence determines some structural features that lead to different CD fingerprints for CdII-rQsMT(1) and CdII-N25-C18. This is exactly the same situation we observed for the ZnII-binding features of these two polypeptides [32]. The lack of S2− anions in the ZnII-N25-C18 preparations also caused the CdII-N25-C18 complexes obtained from in vitro replacement to clearly differ from the in vivo recovered species, but again the spectroscopic features of both CdII-N25-C18 samples could be mutually reproduced after the Acidification/reneutralization/sulfide-addition processes, this highlighting their close relationship (Scheme 2a).

Finally, the analysis of the separate N25 and C18 peptides provided further evidence for a hairpin folding model of CdII-rQsMT (Scheme 2b). C18, the smallest, six-Cys domain, yielded in vivo major monomeric complexes containing two ZnII, but CdII coordination induced its dimerization both in vivo and in vitro, rendering Cd4-(C18)2 dimers. This tendency was already observed in ZnII coordination by the presence of minor Zn5-(C18)2 forms, containing small amounts of S2− (Table 1). Therefore, and probably owing to the cadmium ionic radius, formation of dimeric CdII-C18 complexes is favoured. The behaviour of N25 was more complex than that of C18, most likely because its greater length and higher Cys content enable it to alternate between monomers and dimers when binding CdII. In vivo, N25 basically folds into Zn2 and Zn3 monomers, although as for C18, the presence of minor S2−-containing species already evidences a dimerization tendency. But, unlike C18, N25 gives completely different results for in vivo and in vitro CdII binding. Hence, in vivo, S2−-containing Cd7S4-(N25)2 or S2−-devoid Cd6-(N25)2 dimers were recovered. This was in major concordance with the results for rQsMT and N25-C18, the two additional Cys in the N25 dimer easily accounting for the extra CdII bound (Cd7S4 and Cd6 for dimeric N25, compared with Cd6S4 and Cd5 for rQsMT and N25-C18). But in vitro, the ZnII-N25 monomers evolve to Cd3-N25 monomers, with once again subsequent S2− addition at the end of this titration bringing the CD features of the sample close to those of the in vivo complexes. Therefore, the overall results of N25 analysis pointed to the tendency for dimerization being directly related to the availability of S2− ligands, both conditions concomitantly enhancing the metal content of the clusters. If S2− is absent or scarce, the dimeric CdII complexes are always a minor species. The result of the cotitration experiment further corroborates this hypothesis, both peptides behaving independently in low-sulfide conditions. In conclusion, if the tendency of both peptides is to dimerize when binding CdII in vivo, the most likely scenario is that N25-C18 and rQsMT fold into hairpin structures, to fully accomplish this requirement.

In summary, to our knowledge, this is the first characterization of type 2 plant MT CdII-binding behaviour, including a molecular dissection of its functional regions. Other studies were carried out with undigested fusion constructs and/or with other types of plant/algae MTs [23–27, 43]. We have shown that CdII-rQsMT most probably adopts a hairpin structure that increases its metal-binding capacity with the aid of S2− ligands. The S2−-devoid complexes always exhibit a lower CdII content and the data suggest that in this case the His residue from the spacer region most likely contributes to CdII coordination. The major participation of S2− ligands in CdII-rQsMT complexes accounts for two uncommon features among MTs: the recovery of nonisostoichiometric ZnII and CdII complexes, and the tendency for dimerization of the separate Cys-rich domains to enhance CdII coordination. Globally, all these attributes recall those of the well-known plant PCs, revealing similar molecular strategies of both Cys-rich polypeptides for CdII coordination.