Biochemical Journal

Research article

Mapping of a copper-binding site on the small CP12 chloroplastic protein of Chlamydomonas reinhardtii using top-down mass spectrometry and site-directed mutagenesis

Jenny Erales, Brigitte Gontero, Julian Whitelegge, Frédéric Halgand


CP12 is a small chloroplastic protein involved in the Calvin cycle that was shown to bind copper, a metal ion that is involved in the transition of CP12 from a reduced to an oxidized state. In order to describe CP12's copper-binding properties, copper-IMAC experiments and site-directed mutagenesis based on computational modelling, were coupled with top-down MS [electrospray-ionization MS and MS/MS (tandem MS)]. Immobilized-copper-ion-affinity-chromatographic experiments allowed the primary characterization of the effects of mutation on copper binding. Top-down MS/MS experiments carried out under non-denaturing conditions on wild-type and mutant CP12–Cu2+ complexes then allowed fragment ions specifically binding the copper ion to be determined. Comparison of MS/MS datasets defined three regions involved in metal ion binding: residues Asp16–Asp23, Asp38–Lys50 and Asp70–Glu76, with the two first regions containing selected residues for mutation. These data confirmed that copper ligands involved glutamic acid and aspartic residues, a situation that contrasts with that obtaining for typical protein copper chelators. We propose that copper might play a role in the regulation of the biological activity of CP12.

  • Calvin cycle
  • Chlamydomonas reinhardtii chloroplastic protein CP12
  • copper-binding site
  • redox transition
  • site-directed mutagenesis
  • top-down mass spectrometry (top-down MS)


CP12 is a small chloroplastic protein of about 80 residues that is involved in the Calvin cycle responsible for CO2 assimilation in photosynthetic organisms. It has been found in higher plants [1], green [2] and red [3] algae, in the cyanobacterium Synechocystis PCC6803 [2] and, recently, in a freshwater diatom, Asterionella formosa [4,5].

In higher plants and green algae, CP12 has four conserved cysteine residues that are believed to be functionally important and shares sequence similarities with a 21-amino-acid-long C-terminal sequence extension of the GAPDH (glyceraldehyde-3-phosphate dehydrogenase) B-subunit that contains the regulatory cysteine residues [6]. CP12 is involved in the regulation of GAPDHs that lack this C-terminal extension, such as the A4 GAPDH from cyanobacteria [7] and green algae [810]. CP12 forms a protein complex with PRK (phosphoribulokinase) and GAPDH. Putative N- and C-terminal peptide loops formed via disulfide bonds are involved in the interaction with PRK and GAPDH respectively [1,2]. The role of CP12 in the assembly process of the PRK–GAPDH–CP12 complex in the single-celled green alga Chlamydomonas reinhardtii [11,12] and in the higher plant Arabidopsis thaliana (thale cress) [13,14] has been extensively characterized. However, CP12 belongs to the IUP (intrinsically unstructured protein) family, members of which are involved in the protein–protein interactions that often play major roles in cellular functions [15]. This suggests that CP12 could be involved in protein recruiting in other metabolic pathways. Such a feature was recently reported with the demonstration that CP12 can bind to aldolase [16]. In addition, CP12 was found to bind a metal ion [17] in C. reinhardtii and, using non-covalent ESI (electrospray ionization)-MS experiments, the oxidized protein was found to bind Cu2+ specifically with a dissociation constant (Kd) of 26±1 µM. It was also shown that Cu2+ catalyses the re-formation of the disulfide bonds of the reduced CP12, leading to the recovery of fully oxidized CP12. That study also showed that CP12 has a high structural similarity to copper chaperones from A. thaliana, proteins that play different roles in copper homoeostasis [1820]. The copper-ion-binding site on CP12 was not identified, but it clearly does not involve cysteine residues.

Top-down MS is an emerging tool for obtaining structural information from intact proteins by fragmenting them by MS/MS (tandem MS) [21,22]. The use of top-down MS in proteomics has led to several studies reporting the characterization of new protein variants, isoforms and post-translational modifications [2325]. Such methodology was also used for the discovery of biomarkers [26], and was shown to be promising for the analysis of membrane proteins [2729]. Non-covalent protein assemblies were also shown to be accessible by top-down MS/MS experiments [30,31]. The impact of this new field of investigation is illustrated with the discovery of more extensive polymorphisms and structural complexity to proteins than had previously been envisaged [3234].

In the present study we used top-down MS coupled with site-directed mutagenesis and biochemical analysis to map the copper-binding site of C. reinhardtii CP12.



CdSO4, CuCl2 and FeCl2 were supplied by Merck. Ammonium bicarbonate was purchased from Sigma.

Mutation and purification of recombinant CP12

On the basis of previous MS experiments showing the removal of two protons on copper fixation to CP12 [17] and on the sequence alignment of CP12 from various species [3,35], site-directed-mutagenesis experiments were performed. The codon (GAG) for glutamate residues (Glu20, Glu41 and Glu42) in the Chlamydomonas CP12 coding sequence, in plasmid pGEMT (Promega), was mutated to alanine (GCG) or lysine (AAG) using the QuikChange™ site-directed-mutagenesis kit (Stratagene) and their incorporation was confirmed by Big-Dye DNA sequencing and ESI-MS (see below). Wild-type and mutant CP12 was expressed in Escherichia coli using a pET16b plasmid and purified by IMAC (immobilized-metal-ion affinity chromatography) as described in [11]. The N-terminal histidine tag sequence (HHHHHHHHHHSSGHIEGR), removed using Factor Xa (Merck), left an additional histidine residue (His1) and methionine residue (Met2) appended to the native CP12 N-terminus (SGQPA; Swiss-Prot accession no. A6Q0K5). CP12 sequences were numbered with these two additional residues.

CD studies on CP12 proteins

Spectra were recorded on a J-815 CD spectrometer (JASCO). All spectra are averages of two accumulations from 260 to 190 nm with a 10 nm/min scan rate at 23 °C. Prior to CD experiments, CP12 without the histidine tag was dialysed against phosphate buffer (10 mM Na2HPO4/NaH2PO4, pH 6.0). Measurements were performed using a 0.2-cm-pathlength quartz cuvette in a final volume of 350 μl and 10 μM CP12. For each sample, the fractional α-helical content was calculated using the assumption that, for 100% α-helix, [θ]222 (the mean residue ellipticity at 222 nm) is −36300×[1−(2.57/x)], where x is the number of residues in the protein [36].

Fluorescence studies

Fluorescence spectra of CP12 proteins were recorded on a Fluorolog-3 Jobin Yvon-Spex spectrometer (HORIBA Jobin Yvon SAS) and were recorded from 300 to 450 nm with an excitation wavelength of 290 nm at 23 °C. Measurements were performed with a scan speed of 1 nm/2 s. CP12 proteins were diluted to 1.8 μM in 50 mM Tris/HCl/100 mM NaCl, pH 8, in a 1-cm-pathlength quartz cuvette.

Cu-IMAC and dot-blot analyses of the copper-binding properties of wild-type CP12 and its mutants

IMAC columns (ZipTipMC; Millipore Corp.) were first loaded with copper using a 200 mM CuSO4 solution, except for the control experiment, where copper was replaced by 200 mM FeCl2. All steps were done according to the manufacturer's protocol. CP12 proteins were loaded on the Cu-IMAC column at 0.1 mg/ml. Screening of the differential protein behaviours on copper binding were revealed using dot-blot experiments performed at each washing and elution step. Strong metal binding affinity was systematically assessed by adding an EDTA elution step. Samples of the different eluted fractions of the IMAC column were spotted on to nitrocellulose membranes (Minifold I; dot/slot system SRC 96; Schleicher and Schüll). The nitrocellulose filters were then probed with a rabbit antiserum directed against recombinant C. reinhardtii CP12 (1:3000). Antibody binding was revealed with the ECL® (enhanced chemiluminescence) detection system (GE Healthcare), using donkey anti-rabbit IgG–horseradish peroxidase-linked antibody (1:10000).

Sample preparation for MS

Before MS experiments, CP12 was desalted using an ultra-free membrane filter with a molecular mass cut-off of 3 kDa (Millipore). The CP12 concentration was then determined with the Bradford assay, using BSA as standard [37]. For measurements under non-denaturing conditions, the protein was diluted to 5 μM in 25 mM ammonium bicarbonate, pH 6.8. Copper and cadmium were added at a fixed protein/metal ratio of 1:5 (mol/mol). The histidine tag was removed in order to avoid unspecific metal binding and to improve the fragmentation yield in MS/MS for maximal coverage of the sequence.


All samples were analysed using a LTQ-Orbitrap hybrid mass spectrometer (Thermo-Fisher Co.) fitted with a nanospray ion source. MS and MS/MS spectra were recorded in the FTICR-MS (Fourier-transform ion cyclotron resonance MS) mode at a resolution of 60000 using the normal scan rate mode at 3 microscans and a maximum injection time of 200 ms. Instrumental parameters were optimized in order to obtain the best sensitivity without disrupting non-covalent interactions (positive ion mode, centroid format; ion-spray voltage, 1.3 kV, capillary voltage, 150 V; temperature of the transfer capillary, 150 °C). The mass spectrometer was calibrated in the 500–2000 m/z range using the standard Ultramark™ solution provided by the manufacturer (Thermo-Fisher Co.). The RMS (root mean square) calculated on MS/MS fragments were typically below 2 p.p.m.. When necessary, the m/z range was extended to 4000 without re-calibrating the instrument. In that case the mass accuracy decreased a little, with an RMS below 5 p.p.m. All available collisional modes such as CID (collisionally induced dissociation), CAD (collisionally activated dissociation), PQD (pulsed-Q dissociation) and HCD (higher collisional dissociation) were tested. It appeared that standard CAD mode gave the best results and was finally used for all MS/MS experiments. For each sample, MS/MS experiments were performed on the main charge state (5+) and for the same deposited collisional energy (20 eV). When necessary, other charge states were fragmented to increase sequence coverage. All MS/MS spectra were recorded in the averaging mode to improve the signal-to-noise ratio. The reproducibility of these experiments was assessed. ProSight PTM software, developed by Professor Neil Kelleher's group at the Department of Chemistry, University of Illinois, Urbana–Champaign, IL, U.S.A., was used ( to data-mined MS/MS data with a 5 p.p.m. mass accuracy. Results gave excellent ‘crude Pscores’, assessing protein sequence identities, with an RMS of fragment masses between 0.97 and 1.5 p.p.m.


CP12 mutagenesis

Copper binding to CP12 promotes the removal of two protons, and the conserved residue His49 is not involved in this binding [17]. It was thus speculated that acidic residues may be involved in Cu2+ binding. From an alignment of CP12 from various species [3,34] three conserved glutamic acid residues (Glu20, Glu41 and Glu42) were identified as good candidates to bind copper ions (Figure 1). Consequently, these residues in C. reinhardtii CP12 were mutated to either alanine or lysine. Eight CP12 mutants were examined: E20A, E41A, E42A, E41A/E42A, E20K, E41K, E42K and the double mutant E41K/E42K. MS analysis of the purified wild-type and mutant CP12s showed that their experimental monoisotopic molecular masses matched their expected masses (see Supplementary Table S1a at Top-down MS/MS data were recorded for each free CP12 protein. All MS/MS data acquired for the wild-type CP12 and mutants unambiguously confirmed the expected sequences and location of mutations. CP12 protein sequences and corresponding b and y fragment ions for the free species are displayed in Supplementary Figure S1 at

Figure 1 (A) Ribbon and (B) surface structures of CP12 showing the location of the glutamic acid residues at positions 20, 41 and 42

Pymol software ( was used to visualize the CP12 protein structure. These analyses confirm the accessibility of selected residues (Glu20, and Glu41 and Glu42) in the highly conserved sequence (V40EESLAA46 in the one-letter amino acid code) to bind copper.

CD spectroscopy and fluorescence

The effect of the glutamate mutations on secondary structure was analysed by CD spectroscopy (Figure 2A). The similarities in the CD spectra for the wild-type and single-mutant CP12s indicated that they shared similar secondary structure, with only slight decreases in the α-helical content. This content was about 25% for E20A, 26% for E41A and 23% for E42A, whereas it was about 29% for the wild-type. In contrast, the decrease in the α-helix content in the E41A/E42A (17%) and E41K/E42K (19%) double mutants was more prominent, indicating a more substantial change in the global structure of these mutants. The influence of Cu2+ on the CD spectra of wild-type and mutant CP12s was also examined and it was shown that only the wild-type CP12 spectra, in which the α-helical content slightly increased to 32% (Figure 2A), was influenced by copper. No effect of copper was observed with any of the mutants (Table 1).

View this table:
Table 1 CD data for wild-type CP12 and its mutants recorded with increasing amounts of copper
Figure 2 CD and fluorescence spectra of CP12 and its mutants

(A) CD spectra of 10 μM wild-type CP12 and its mutants (curves 1–6) and of wild-type CP12 incubated with Cu2+(curves 7 and 8). Spectra were obtained in 10 mM phosphate buffer, pH 6.0. [θ] is mean residue ellipticity. (B) Fluorescence spectra of 1.8 μM wild-type CP12 and its mutants in 50 mM Tris/100 mM NaCl, pH 8 (λexcitation, 290 nm; λemission, 300–450 nm).

Fluorescence spectra obtained with wild-type and mutant CP12s revealed no difference between the E42A mutant and the wild-type CP12, whereas a quenching of tryptophan fluorescence was observed for all other mutants (Figure 2B). The result obtained for the E42A mutant was consistent with the fact that no hydrophobic contact could be contracted with this amino acid, since no apolar residue is in its vicinity. Quenching effects observed for other mutants were quantified and a scale was defined according to the effect of mutation on structure, with the following order: E41K/E42K>E20A=E41A/E42A>E41A>wild-type.

Effects of mutation on copper binding

The copper-chelation properties of wild-type CP12 and its mutants were tested using IMAC-Cu (an IMAC resin charged with Cu2+). As wild-type CP12 can bind copper, it bound to IMAC-Cu and could only be eluted with 0.3 M NH4OH or 1 mM EDTA. By contrast, mutants with impaired affinities for copper did not bind to IMAC-Cu, nor were they eluted during the washing steps. The affinity of the mutant CP12s was qualitatively assessed by monitoring the presence of CP12 after elution with NH4OH or EDTA (Table 2). On the basis of their affinity to IMAC-Cu, mutation of Glu20 had no influence on copper binding, whereas the mutation of Glu41 strongly affected copper binding and mutation of Glu42 into lysine only diminished copper binding. Wild-type CP12 did not bind to IMAC resin charged with Fe2+, indicating that the binding of CP12 to IMAC-Cu was copper-specific.

View this table:
Table 2 Summary of the effect of CP12 mutation on copper-binding properties observed in Cu-IMAC experiments

Symbols: (+), + and ++ reflect the proportion of CP12 detected at each step; –, CP12 not detected; NR (‘not retained’), elution of CP12 in the void volume of the column.

Assessment of copper binding

The molecular masses of wild-type CP12 and its mutants were experimentally measured after adding a 1:5 (mol/mol) protein/copper ratio. In each case, the mass spectra obtained displayed only the presence of one copper atom complexed to CP12. Calculated molecular masses showed that copper binds to all CP12 proteins (wild-type and mutants) with a mean monoisotopic mass increment of 60.91 Da (see Supplementary Table S1B). This confirms previous observations that the copper redox state binding is 2+ [38,39], and tetragonal co-ordination of Cu2+ is therefore proposed in accordance with [40]. The CP12–Cu complex is described by [M−2H+Cu+nH]n+, where M is the mass of the protein with subtraction of the mass of two hydrogen atoms linked to copper co-ordination, plus the addition of the mass of copper (explaining the mass increment of 60.91 observed) and n, the number of protons giving the charge state. Using this description the main charge state of the wild-type CP12 protein complexed with copper can be expressed as follows: Embedded Image with an expected monoisotopic m/z value at 1759.0023, in agreement with experimental value observed at 1759.0054 (1.73 p.p.m. difference). Deconvolution software was found to give the correct mass, but peak charge states were manually checked and found to be in agreement with the automated charge state assignments. A noticeable difference in the capacity of the E41K/E42K CP12 variant to bind copper ion was observed. Although the wild-type and other mutant CP12s fully bound Cu2+ within 2 min, incubation for 24 min was required to bind copper to ∼50% of the E41K/E42K CP12 double mutant. The decreased binding affinity might be attributed to a higher steric hindrance imparted by the length and protonation propensity of the lysine side chains (Figure 3).

Figure 3 Comparison among MS spectra of mutant CP12 proteins recorded under non-denaturing conditions without and with copper

(A) Free E20A mutant of CP12 (upper panel) and copper-complexed E20A (lower panel). (B) Free E41K/E42K double mutant of CP12 (upper panel) and copper-complexed E41K/E42K (lower panel). Peaks corresponding respectively to the free state and the E41K/E42K–Cu2+ complex state are labelled with X and ● respectively. z stands for the charge states of the peaks.

Studying protein CSD (charge state distribution)

Since it was demonstrated that variations of protein CSD could be related to structural properties, we used that feature to probe conformational changes linked to mutations and copper binding. In particular, we used Zav (mean charge state) and σ(i) (standard error on Zav) values that correspond to the calculated Zav and S.D. respectively [41]. Calculated Zav values and calculated standard errors [42] describing mutation effects and copper binding influence on structure are reported in Table 3. Results of CSD analysis indicate that there is no significant difference between wild-type CP12 and its mutants, except for the E41K/E42K double mutant that shows a slight effect. Similarly, comparison of data obtained for the free and copper-complexed states of each protein did not reveal important change. In these cases the slight difference observed for E20A and E42A mutants may be assigned to a gas-phase effect related to the strengthening of electrostatic bonds in a lower dielectric environment (vacuum).

View this table:
Table 3 Calculated Zav values between the free and copper-complexed states of each protein

‘Δ Mutation’ means the difference between the mean Zav of the wild-type and that of the mutant protein; ‘Δ Copper’ means the difference between the mean Zav of the free and copper-complexed state of each protein. Results are means±S.D. for four measurements.

MS/MS of Cu2+–CP12 complexes

Our approach was validated by recording MS/MS data for the CP12–Cd2+ complex (control experiment), Cd2+ being a metal ion that was shown to bind un-specifically to CP12 [17]. In that case, Cd2+ was observed as adduct on every fragment ion. Intensities of fragment ions bearing a Cd2+ adduct were minor and were in a mixture with the corresponding free fragment ions. In contrast, MS/MS recorded for CP12–Cu2+ complexes showed the presence of specific fragment ions bearing one copper ion, representing in that case the major species.

Extractions of MS/MS data were realized for spectra having the same acquisition time. ‘Artifactual’ fragment peaks with ProSight PTM were observed and were likely due to mis-assignments of peaks and/or isotopic profiles. Therefore, some peaks were manually removed from the peak lists and charts to clarify the Figures. Automated searches also led to the ‘recognition’ of peaks assigned to fragment ions located between the two cysteine residues linked through a disulfide bridge. Since the breakage of disulfide links is unexpected under low collisional energy mode or required mechanisms involving charge-directed neighbouring-group or non-mobile proton salt processes [43], we manually checked the presence of these ions. Scrutiny of MS/MS spectra finally proved the presence of such ions (b70 ions, which are labelled with an asterisk in sequence charts).

Two MS/MS spectra are shown in Figure 4. Less fragment ions were observed on MS/MS spectra recorded for the CP12–Cu2+ complexes compared with their respective free states (Figure 5A). This suggests that binding of copper triggered conformational changes leading to a more rigid structure. However, the yield of fragmentation differed from one protein to another, with the E41A mutant showing the strongest ‘stabilizing’ effect linked to copper binding.

Figure 4 MS/MS spectra of wild-type CP12 (A) and its E41K/E42K double mutant (B) without and with copper

(A) Wild-type CP12 without (upper panel) and with (lower panel) copper; (B) E41K/E42K mutant without (upper panel) and with (lower panel) copper. Fragment ions bearing a copper ion are labelled with an asterisk (*). z stands for the charge states of the peaks.

Figure 5 (A) Sequence charts for wild-type CP12 and its mutants and (B) location of copper-binding regions

(A) Sequence charts of wild-type CP12 and its mutants, showing fragment ions generated by top-down MS/MS experiments in the presence of copper. An illustration of the fragmentation nomenclature is given in the insert (top right) [50]. The bold angle brackets (⌉) represent fragment ions bearing a copper ion. The asterisks (*) denote b70 ions that correspond to disulfide bond breakage. These ions are only observed when CP12 is complexed with copper. The mutation positions are in bold and underlined. (B) Location on CP12 sequence of the three regions proposed to be involved in copper binding (region 1, Asp16–Asp23; region 2, Asp38–Lys50; region 3, Asp70–Glu76). The α-helices are depicted by grey boxes under the sequence, and positively and negatively charged residues are in bold. The residues that have been mutated in this study are indicated by arrows (↓).

Copper was not detected on fragment ions located on α-helix 2 (residues 30–56) until position 52 is reached (fragment b52), probably because crucial interactions between the two helices are maintained. Similarly, the proximity of the disulfide bridges seemed to be essential to maintain a minimal structure allowing the ligation of a copper ion, as suggested by the detection of y20 fragment ion. Three regions putatively involved in copper binding were proposed. Regions 1–3 spanned residues Asp16–Asp23 (fragments b16–b23), residues Asp38–Lys50 (fragments b38–b50) and residues Asp70–Glu76 (fragments b70–b76) respectively (see the locations in Figure 5B). Defining the first region (Asp16–Asp23 in ‘regular’ nomenclature), including the Glu20 residue, was straightforward, with the recurrent detection of b16, b22 and b23 ions bearing a copper ion for all CP12 species. Location of the second region (Asp38–Lys50) comes from the detection of overlapping b (b52–b62) and y (y25–y44) fragment ions bearing a copper ion. We narrowed region 2 to residues 38–50, because the upholding of CP12's secondary structure strongly depends on interactions between residues 11–16 of the N-terminus and residues 53–58 of the second α-helix [44]. Identification of the third putative region was based on the recurrent detection of y20 and y25 fragment ions, as well as the b70 ion observed for the MS/MS spectra of the E20A–Cu2+ and E42A–Cu2+ complexes. As mentioned above, since we proved the real presence of b70 ions in the MS/MS spectra and that disulfide breakage can only occur via a mechanism involving a fixed proton [43], the observation of b70 ions can only be explained by the presence of Cu2+ in a close proximity to the Cys68–Cys77 disulfide bridge.

A detailed description is proposed to explain the detection of the b70 ions only for the E20A and E42A MS/MS spectra. For E20A, mutation is proposed to promote a shift of the first α-helix that became closer to α-helix 2, owing to the smaller steric hindrance of the side chain. Bringing the two helices together can induce a movement of the copper ion toward the Cys68–Cys77 disulfide bridge, then inducing its breakage. This explanation is in concordance with the fluorescence data, which show that the E20A mutation induces an important quench of Trp37, suggesting rearrangements of the two α-helices. In the second case, mutant E42A is proposed to induce disulfide-bridge reduction in a similar manner by bringing together the C-terminal loop and α-helix 2, by reinforcing hydrophobic contacts with, for example, Phe67. This explanation appears to be in apparent discrepancy with fluorescence data indicating no change in the Trp37 fluorescence spectrum. However, the presence of a unique tryptophan residue in the CP12 sequence allows us to monitor structural changes only in its vicinity. The absence of the b70 ion in the MS/MS spectra of the E41A mutant complexed with copper could be explained by a strengthening of the hydrophobic contacts of the two α-helices. In that case the copper ion would not be directed toward the C-terminal disulfide bridge and no disulfide-bond breakage will occur. Finally, the lack of the b70 ion on the MS/MS spectra recorded for double mutants could be explained by any structural change pushing the copper ion remotely from the C-terminal disulfide bridge. All these results support the presence of copper in the C-terminal part of the protein. Finally, the model structure of CP12 protein shows that the loop made by the C-terminal disulfide bond contains four acidic residues (Asp70, Asp73 and Asp75, Glu76), in the close vicinity of residues Asp38 and Lys42, and thus may be involved in copper co-ordination. Region 3 was proposed to span residues Asp70–Glu76. Experimental values for fragments carrying a copper ion compared with theoretical ones are given in Supplementary Tables S2–S7 at


Although CP12 can bind copper ion [17], no information on the location of the binding site of this metal ion has so far been available, thereby impeding progress in understanding its function. After selecting residues presumably involved in copper binding, global structure of the mutants was checked using CD, fluorescence as well as MS. CD showed that the global structure of single mutants was not affected by mutations, whereas double mutants were shown to have a higher impact on CP12 tertiary and secondary structure. Fluorescence data indicated quenching effects of Trp37 for E20A, E41A, E41A/E42A and E41K/E42K mutants, results in agreement with those from CD experiments also demonstrating that single mutations had little influence (except for E20A) on CP12 structure, whereas double mutants were shown to significantly alter the protein conformation of CP12 (Figure 2). Analysis of Zav values deduced from CSD, and differences among these values for wild-type CP12 and its mutants were then used to confirm structural changes observed in CD and fluorescence experiments. Unfortunately, these values did not permit characterization of the structural effects of mutations or the impact of copper binding on CP12 structure. The global protonatable surfaces of mutants are similar to wild-type CP12 and the extent of structural changes linked to copper binding does not affect the global structure of the protein.

The effects of the glutamic acid mutations on copper binding by CP12 were also examined using Cu-IMAC, which indicated that Glu41 has a strong influence on copper binding (Table 2). Mutation of Glu20 had little or no influence on the avidity of CP12 for Cu2+, whereas the replacement of Glu42 by lysine impeded copper binding, possibly owing to a reverse charge effect or steric hindrance. An interesting feature of the Cu-IMAC binding assay is the likely requirement for the residues involved in the co-ordination with the immobilized Cu2+ to be appropriately exposed to the solvent (i.e. arranged on the protein surface). This may explain why the E41A mutant was shown by MS to bind copper while at the same time having decreased affinity for Cu-IMAC column.

To gain a better grasp of the structure–function relationships involved in the binding of CP12 to copper, top-down MS/MS experiments were performed on free and Cu-complexed states of each CP12 protein. Top-down MS/MS experiments not only allowed the protein sequence of the wild-type and mutants to be assessed, but also confirmed the location of the disulfide bridges.

ESI, and particularly nano-ESI, techniques produce ions with minimal ion fragmentations and internal rearrangements [38,45]. On the basis that no change was observed in Zav values, mutations seem not to affect the global desorption–ionization efficiency of CP12 mutants compared with the wild-type. Since we used identical MS/MS parameters and since control experiments conducted with Cd2+ confirmed that detection of fragment ions carrying a copper ion are significant, results deduced from top-down MS/MS were shown to directly reflect copper-binding effects in solution. The present results further support the involvement of aspartic acid and and glutamic acid residues in metal ion co-ordination with the location of three regions (region 1, Asp16–Asp23; region 2, Asp38–Lys50; and region 3, Asp70–Glu76) putatively involved in copper binding. These results are summarized in Figure 5(B). Detection of b70 ions, which were only found for the E20A– and E42A–Cu2+ complexes, demonstrates the proximity of the copper ion binding to the C-terminal disulfide bond. Such fragmentation, according to the literature, could only be related to mechanisms involving charge-directed neighbouring-group or non-mobile proton salt processes [43]. These mechanisms were confirmed by the detection of specific chemical losses (−44 Da and −45 Da) from the b or y fragment ions bearing a copper ion [43].

MS/MS data suggest that there is a relationship between the structural changes induced by the mutations and the impairment of copper binding. types of modifications involved in these structural changes include rupture of the electrostatic network or enhancement of hydrophobic interactions. This is illustrated with E20A and E41A mutants, which were shown to induce structural changes in the vicinity of Trp37 where stabilization of the hydrophobic core of the protein between the two α-helices is proposed. Scrutiny of the predicted CP12 structure showed that the two helices are maintained by a hydrophobic core, in concordance with the stabilization promoted by E41A mutation. This could also be the case for E42A mutant, but with a more moderated effect on structure. In that case, the rupture of electrostatic network involved in copper binding with the C-terminal loop is proposed to be compensated for by reinforcement of hydrophobic contacts with Phe67. This notion is supported by description of strong interactions between Phe67 and Ala45/Ala46 [44].

At this stage, one intriguing point remained to be explained. Whereas regions 2 and 3 are relatively close in the CP12 structure, allowing copper chelation, region 1 is remotely located in the CP12 structure. The detection of b16–b23 fragment ions bearing a copper ion means that any residue between positions 1 and 23 can fix the copper ion. Moreover, no candidate was found in the close vicinity of this region to allow copper binding. To bridge the gap between all our observations, we proposed that carbonyl groups of the α-helix peptidic backbone could be involved in the binding of the metal ion. Such a hypothesis can link the hydrophobic and electrostatic effects observed. Moreover, the implication of one of the acidic residues present inside the C-terminal disulfide loop, such as Asp70 or Asp73 or even the the C-terminal carboxylate group, in tetragonal co-ordination of copper, is not ruled out.

The C-terminal domain of the CP12 protein plays a major role in protein–protein interactions [1,46]. GAPDH and PRK interact with CP12 even in the absence of copper; thus binding of copper to CP12 is not necessary for the formation of the GAPDH–PRK–CP12 complex. Dissociation of this complex in vivo is triggered by a light-dependent ferredoxin–thioredoxin system and linked to the reduction of the two disulfide bridges of the CP12 protein [8,14,46]. In contrast, under dark conditions, where formation of this complex occurs, formation of disulfide bridges on CP12 could be triggered either by oxidized thioredoxin or by copper binding, thus leading to oxidized CP12. However, the role of the binding of a copper ion on the three regions defined in the present paper still remains to be elucidated, but could be linked to a specific feature of CP12. Indeed, this protein belongs to the IUP family [11], and a multitude of examples has demonstrated that the function of disordered proteins can be closely related to their metal-ion-binding properties. To cite but a few examples: prion protein can bind copper ion and, as a consequence, can play a role in amyloid formation [47]; the N-terminal domain of HIV integrase can bind Zn2+, leading to an increased activity [48]; and α-synuclein can bind many metal ions, thereby accelerating its fibrillation [49]. We show in the present study that binding of copper triggered conformational changes on CP12 leading to a more rigid structure. A decrease in plasticity and/or conformational changes upon metal-ion binding thus might represent a general mechanism for enhancing the target recognition of IUP and, in particular, of CP12.


This work was supported by the Ministère de la Recherche, France, and the Université Pierre et Marie Curie-Paris VI, Paris, France (Ph.D. funding for J. E.).


We thank Dr Jean-Yves Salpin (Laboratoire d'Analyse et Modélisation pour la Biologie et l'Environnement, Université d'Evry Val d'Essonne, Evry, France) and Dr Aram Nersissian (Department of Chemistry, Occidental College, Los Angeles, CA, U.S.A.) for helpful scientific discussions.

Abbreviations: CAD, collisionally activated dissociation; CID, collisionally induced dissociation; CSD, charge state distribution; ESI, electrospray ionization; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IMAC, immobilized-metal-ion affinity chromatography; IMAC-Cu, IMAC resin charged with Cu2+; IUP, intrinsically unstructured protein; MALDI-MS, matrix-assisted laser-desorption–ionization MS; MS/MS, tandem MS; PRK, phosphoribulokinase; RMS, root mean square; Zav, mean charge state; σ(i), standard error on Zav value


View Abstract