An Ion Mobility-Mass Spectrometry Study of Copper-Metallothionein-2A: Binding Sites and Stabilities of Cu-MT and Mixed Metal Cu-Ag and Cu-Cd Complexes

Abstract

The presence of Cu, a highly redox active metal, is known to damage DNA as well as other cellular components, but the adverse effects of cellular Cu can be mitigated by metallothioneins (MT), small cysteine rich proteins that are known to bind to a broad range of metal ions. While metal ion binding has been shown to involve the cysteine thiol groups, the specific ion binding sites are controversial as are the overall structure and stability of the Cu-MT complexes. Here, we report results obtained using nano-electrospray ionization mass spectrometry and ion mobility-mass spectrometry for several Cu-MT complexes and compare our results with those previously reported for Ag-MT complexes. The data include determination of the stoichiometries of the complex (Cu_i-MT, i = 1 -19), and Cu⁺ ion binding sites for complexes where i = 4, 6, and 10 using bottom-up and top-down proteomics. The results show that Cu⁺ ions first bind to the β-domain to form Cu₄MT then Cu₆MT, followed by addition of four Cu⁺ ions to the α-domain to form a Cu₁₀-MT complex. Stabilities of the Cu_i-MT (i = 4, 6 and 10) obtained using collision-induced unfolding (CIU) are reported and compared with previously reported CIU data for Ag-MT complexes. We also compare CIU data for mixed metal complexes (Cu_iAg_j-MT, where i + j = 4 and 6 and Cu_iCd_j, where i + j = 4 and 7). Lastly, higher order Cu_i-MT complexes, where i = 11-19, were also detected at higher concentrations of Cu⁺ ions, and the metalated product distributions observed are compared to previously reported results for Cu-MT-1A (Scheller et al., Metallomics 2017, 9, 447-462).

Graphical Abstract

Cu(I) binds to human metallothionein MT2 to form stable Cu₆- and Cu₁₀-MT complexes that exhibit high gas-phase stabilities.

Introduction

Metallothioneins (MTs) exhibit a high degree of conformational heterogeneity and are even described as intrinsically disordered proteins (IDPs), but it is generally agreed that the structure(s) of MTs are determined by the number and types of metals they coordinate. Chen et al. reported results from ion mobility-mass spectrometry and molecular dynamics studies that suggest that the apo-MTs are better described as having a high degree of conformational heterogeneity.^{1, 2} In fact, metalated MT analogues appear to form more ordered structures, but even the highly metalated MT analogues are difficult to characterize using traditional techniques.^{1, 3, 4} In part, these difficulties are due to low abundances of apo-, partially-, and fully-metalated MT products that present high degrees of conformational heterogeneity unfavorable for growth of single crystals required for X-ray crystallography diffraction. Also, conformational dynamics limit the utility of NMR spectroscopy for determination of the detailed structure information, including determination of metal binding sites of MTs.^{3, 4} Fenselau and co-workers first reported electrospray ionization-mass spectrometry (ESI-MS) of MTs,^{5, 6} and Stillman and co-workers have taken a more holistic approach, comparing results obtained using ESI-MS for studies of reactivities and kinetics of various metalated MTs, including the pH-dependent folding and unfolding behaviors by cysteine modification, and optical spectroscopy.^7-9 Blindaeur and co-workers^10-12 reported evidence that apo- and partially-metalated MTs are able to unfold, which increases the complexity of traditional characterization methods. While ESI-MS is able to determine the binding stoichiometry of MT with different metals,^{1, 2, 13-16} binding site assignments and structural stability changes are more difficult to assign. The utility of ion mobility combined with mass spectrometry (IM-MS) for separation of gas phase ions by difference in mass and charge^15-17 provides increased peak capacity¹⁸ and more detailed structure analysis.^{15, 16} In addition, ion-neutral collision cross sections (CCS) obtained by using ion mobility-mass spectrometry and collision-induced unfolding (CIU) add additional dimensions for studies of metalated MTs, including gas phase stabilities of partially- and fully-metalated MT ions.^18-21 Recent studies by Scheller et al.^{22, 23} illustrate how integrated approaches, ESI-MS and optical spectroscopy, provide complementary information for Cu(I) metallothionein-1 complexes. Peris-Diaz et al.²⁴ further illustrate the utility of integrating mass spectrometry, label-free proteomics and molecular dynamics simulations to unravel complexities of the Zn_4-6MT complexes.

In our previous studies of MTs^{1, 2, 13, 14} we have developed integrated IM-MS-based approaches to investigate metal binding (Zn, Cd, and Ag) of metallothionein-2A focusing on the sites of metal ion binding, metal binding induced conformational changes as well as changes in the stabilities of complexes. For example, we showed that the metalation product Cd₄-MT has all four Cd ions bound to the α-domain,^{1, 2} whereas Ag₄-MT complex has all four Ag⁺ ions bound in the β domain.¹⁴ We also showed that in the mixed metal Ag₄Cd₄-MT complexes these domain specific metal binding preferences are retained. The collision-induced unfolding (CIU)¹⁴ studies of the Ag-, Cd- and Ag/Cd- metalated products show that the stabilities of the gas phase ions increase with a higher degree of metalation, higher collision energies are required to unfold, illustrating that the structures of the metalated MTs are more ordered.

Here, we employ these same approaches to characterize the Cu-MT-2A system.^{1, 2, 13, 14, 25} Cu is a physiological important trace metal, and copper toxicity is often associated with genetic defects of copper metabolism.²⁶ Specifically, Cu-metalated metallothionein is implicated in different types of human diseases, most notably Wilson’s disease, a genetic disease caused by accumulation of copper in brain, liver, and cornea.²⁷ Cu-MT is also implicated in neuropathology of Alzheimer’s disease (AD),^28-30 where MTs maintain metal homeostasis and prevent the formation of Cu-Aβ complexes.^31-34 MT-3 is also thought to antagonize neurotoxic and neurotrophic effects of Aβ complexes though exchange of Zn₇-MT-3 and Cu-Aβ.^35-39 Pountney et al.⁴⁰ reported that two Cu₄ clusters are formed in both the α- and β-domains of MT-2A and these clusters can be expanded to Cu₆ as well as Cu₈ to Cu₁₂ complexes. Scheller et al.^{22, 23} recently reported results for Cu⁺ MT-1A and Melenbacher et al.⁴¹ reported the binding behavior of Cu⁺ with recombinant human apo-MT-1A and showed that Cu⁺ ions form metal-cysteine clusters in both domains. Although the sequences and structures of the different MTs are overlapping, the metal binding behavior could be different. Artells et al.⁴² provided evidence of different metal ions binding of MT1 and MT2, which respectively correspond to their physiological roles. In the studies described below we show that Cu₄Cys₆ and Cu₆Cys₉ clusters are formed in the α and β domain, respectively. The collision-induced unfolding behavior of Cu-MT species follow a general pattern similar to that of the Cd- and Ag-MT complexes, specifically the higher degree of metalation stabilizes the gas phase structures, though differences are observed for each of metal-MT complexes.

Experimental

Copper acetate, diethylenetriaminepentaacetic acid (DTPA), dithiothreitol (DTT), tris (2-carboxyethyl) phosphine hydrochloride (TCEP), and N-ethylmaleimide (NEM) were purchased from Sigma-Aldrich (St. Louis, MO). Proteomics grade trypsin used for digestion was purchased from Thermo Fisher Scientific (Waltham, MA). Deionized water (18.2 MΩ) was obtained from a Milli-Q water apparatus (Millipore, Billerica, MA). Micro Bio-Spin 6 columns for buffer exchange were purchased from Bio-Rad Laboratories, Inc. (Hercules, CA). The procedure for MT2A expression and purification are described in our previous publication.¹⁴

Titration of Cu²⁺ solutions was performed by sequential addition of 1 mM copper (II) acetate solution to 50 μL of 15 μM MT solution including 50 mM ammonium acetate, 1 mM TCEP and 10% (by volume) of methanol. With the presence of TCEP, Cu(II) is reduced to Cu(I),⁴³ and cysteines can also reduce Cu(II) to Cu(I) to form Cu(I)-Cys complexes.⁴⁴ In MT systems it is also reported that Zn₇-MT scavenges Cu²⁺ and reduces Cu²⁺ to Cu⁺,^{44, 45} so the Cu in this solution system can be regarded as Cu(I). The solution was incubated for 30 min at room T to reach equilibrium. After the 30-min incubation, the solution was analyzed by nESI-MS on a Waters Synapt-G2 HDMS instrument. Covalent labelling experiments were conducted by adding excess concentration (10 mM) of N-ethylmaleimide (NEM) to Cu-MT solutions. nESI-MS spectra were acquired after incubation for 30 min.

The Cu binding sites of the alkylated Cu₆-MT and Cu₁₀-MT were investigated using top-down proteomics experiments described previously.^{1, 14} The bottom-up proteomics experiments were performed using standard tryptic digestion protocols.¹⁵ Individual solutions of Cu₆-MT and Cu₁₀-MT were prepared by careful addition of 6 or 10 equivalents of Cu to solutions of MT. The resulting Cu₆-MT solution contained small amounts of Cu₄-MT and Cu₁₀-MT (see Figure 1), thus low abundance tryptic fragments from these complexes might complicate assignment of the tryptic fragments. Tryptic digestion of Cu₆-MT and Cu₁₀-MT was performed by addition of trypsin (10 μg/mL) to the protein-containing solution in a 1:50 mass ratio at 37 °C and incubated overnight. After quenching with 0.1% formic acid, the solution was concentrated by a Ziptip concentrator before mass spectral analysis.

Figure 1.

(A) ESI-MS spectra for the titration of MT with Cu(I). Ions corresponding to [Cu-MT]⁴⁺ species are much more abundant than are [Cu-MT]⁵⁺ species. This is just the opposite of the mass spectra obtained for titration of Zn-, Cd- and Ag-MT complexes. (B) Apparent Cu-binding constants for Cu₄-MT to Cu₁₉-MT. The binding constants are obtained using the methods published in reference 1. The molar fractions of apo and Cu-MT are plotted in Figure S1. (C) Mass spectra showing the products formed following addition of 6 and 14 equivalents of NEM to a solution containing Cu₆-MT and Cu₁₀-MT. Note that after the addition of NEM, the 5⁺ ions are more abundant than the NEM 4⁺ ions.

Collision-induced unfolding (CIU) experiments were performed as described previously,^{13, 14} and the CIU heat maps were generated using the CIUSuite2 developed by Ruotolo and co-workers^{20, 21, 46} The collision cross section profiles (see Figure S9) were acquired by incrementally increasing the trap collision energy in 5 V increments from 5 V to 50 V, which corresponds to lab-frame collision energies of 25 eV for the 5⁺ ions.

Mixed-metal complexes including Cu/Ag- and Cu/Cd-MTs were synthesized by mixing the metal solutions in the same concentration (1 mM each to make a 0.5 mM and 0.5 mM system) before addition to the apo-MT. The mass spectra of 3 and 5 equivalents per MT molecule (45 and 75 μM for each metal, respectively) were obtained to show the effect of increasing metal concentration. CIU experiments of mixed-metal products were performed using the same method above.

Results and Discussion

Titration of Human MT-2A with Cu⁺ Ions

Titration of solutions containing MT-2A (15 μM) by sequential addition of Cu²⁺ (1 mM copper acetate) yields a distribution of metalated Cu_i-MT⁴⁺ (i = 4 – 19) and Cu_i-MT⁵⁺ (i = 4, 6, 10) complexes; under the experimental conditions, the bound Cu ions of the Cu-MT complex are formally Cu(I). In solutions containing TCEP, Cu(II) is reduced to Cu(I), and cysteines can also reduce Cu(II) to Cu(I) to form Cu(I)-Cys complexes.^{43-45, 47} The Cu_i-MT⁵⁺ complexes are formed in lower abundance relative to that for Cu_i-MT⁴⁺, which is very different from that for Zn-MT, Cd-MT and Ag-MT, where the abundances of the higher charge state (5⁺) products are preferred (Figure 1A). The differences in charge states for Cu-MT and similar complexes of Zn, Cd and Ag suggest that the solvent accessible surface areas (SASA) and conformations are different. Changes in SASA and/or conformation may reflect changes in intramolecular interactions and/or changes in hydration of the side chains of polar residues. The results for the different charge states do not show significant difference, other than ion abundances. At no point during the titration are signals for Cu_i-MT (i = 1, 2, 3) observed, similar behavior was observed for Ag⁺ titration of MT.¹⁴ For both Ag⁺ and Cu⁺ this behavior is attributed to the high concentration of the thiol ligands (20 per MT molecules) and high level of cooperativity for formation of Cu₆-MT. Apparent Cu-binding constants (Ka) for each metal ion binding (i = 4 - 19) are shown in Figure 1B. Note that similar cooperativity is observed for formation of Cu₆- and Cu₁₀-MT as evidenced by the absence or low abundances of Cu₅- Cu₇-, Cu₈- and Cu₉-MT products. At high concentrations of Cu (>1 equivalent of Cu per -SH group) higher metalated products are observed up to Cu₁₉-MT. Similar higher order Cu-MT complexes were reported by Scheller et al. in their study of Cu-MT-1A,^{22, 23} and similar Ag-MT clusters were observed by Dong et al.¹⁴ The Cu_i-MT complexes (i = 13, 17, and19) appear in higher abundances, but at high concentrations of Cu (over 20 equivalents) the abundances of these products decrease, suggesting larger complexes are not formed or possibly that they precipitate from solution.

Figure 1C contains mass spectra following addition of N-ethyl maleimide (NEM) to solutions where the abundances of Cu₆-MT and Cu₁₀-MT reach a maximum, i.e. addition of 6 and 14 equivalents of NEM, respectively. First, addition of NEM produces higher abundances of Cu₆NEM₁₁⁵⁺ and Cu₁₀NEM₅⁵⁺ compared to the 4⁺ ion complexes, which suggests a change in the SASA and conformation of the complexes. As noted above, these SASA and/or conformation upon addition of NEM (addition of a non-polar site) can alter intramolecular interactions as well as solvation of the Cu-MT complex. NEM covalently binds to metal-free thiol groups and these products are then analyzed using both top-down and bottom-up (tryptic digest) proteomic approaches. Interrogating the covalently labeled complexes using both top-down and bottom-up proteomic approaches provide complementary information regarding the Cu⁺ ion binding sites.^{1, 2, 13, 14, 25} Top-down proteomics reports on ions that are formed in solution and then sampled by fragmentation of the gas-phase ions, whereas bottom-up proteomics samples ions formed by tryptic digestion (in solution) followed by MS analysis of the tryptic peptide fragments. The mass spectra obtained following addition of NEM to solutions containing Cu₆- and Cu₁₀-MT yield abundant signals for Cu₆NEM₁₁-MT and Cu₁₀NEM₅-MT, respectively (Figure 1C). Formation of Cu₆NEM₁₁-MT⁴⁺ and Cu₆NEM₁₁-MT⁵⁺ is interpreted as evidence that 11 of the 20 cysteines are labeled by NEM, while the remaining 9 are coordinated to 6 Cu ions. Likewise, the detection of Cu₁₀NEM₅-MT⁴⁺ and Cu₁₀NEM₅-MT⁵⁺ is evidence that 15 cysteines are coordinated by Cu and 5 are labeled by NEM (see Figure 2 for structure illustrations). In our previous MT studies we have observed that NEM is capable of producing products best described as metal ion displacement from cysteines that are weakly interacting with metal ions.¹⁴ Thus we also interrogate the Cu-MT complexes using top-down proteomic approaches in order to minimize errors in the bottom-up approaches. The assignments for NEM binding sites for Cu₆NEM₁₁-MT and Cu₁₀NEM₅-MT were first investigated^{1, 14} using a two-dimensional top-down proteomics sequencing approach developed by Zinnel et al.²⁵ Tandem mass spectra are obtained by mass selecting Cu₆NEM₁₁-MT ions, followed by collision-induced dissociation (CID, see Experimental section) and analysis of the CID fragment ions. Because the CID spectra are rather congested, owing to the large numbers of CID fragment ions, and further complicated by the formation of fragment having different charge states, the CID fragment ion spectra were recorded using IM-MS, which disperses ions on the basis of charge and m/z (see Figure S4).²⁵

Figure 2.

Proposed binding domains of Cu(I) for the Cu₆NEM₁₁-MT and Cu₁₀NEM₅-MT complexes. The placement of the Cu atoms (brown) is not intended to show specific binding interactions, they only represent domains in which Cu(I) ions are located. The GS tag (labeled blue) is not removed by TEV cleavage, and Gly-Ser tags are not included in numbering of the amino acid (single letter notation).

The extracted CID spectra of Cu₆-MT (Figure 3A) contains a series of metal-free b₇, b₈ and b₉ ions that are interpreted as evidence that Cys5 and Cys7 are either not involved in metal ion binding or are weakly binding. It is interesting that neither Cys5 or Cys7 are labeled by NEM or coordinated by Cu, but we cannot rule out the possibility that Cu is weakly interacting with these cysteines; similar behavior was noted previously for Ag-MT.¹⁴ Similarly, the fragment ions that contain Cys33 – Ala61 of the α domain do not contain Cu ions, and the N₈y25 fragment ion contains 8 free cysteines of the α domain from Cys33 to Ala61. Lastly, the fragment ion N₁₁y30 further confirms that all 11 cysteines labeled by NEM are bound to Cys located in the α domain. A similar strategy was used to probe the metal ion binding sites of Cu₁₀NEM₅-MT, but this proved to be less informative because a limited number of low abundance fragment ions that correspond to bond cleavage reactions near the N- and C-terminus of the MT sequence, viz. b_4, y₇ and y₁₀ (see Figure S5). These data are consistent with other data for the Cu₁₀NEM₅-MT complex being a stable Cu₁₀ cluster (vide infra). Scheller et al.^{22, 23} attribute the Cu₁₀-MT-1 to a similar composition of Cu₆β-Cu₄α-MT-1.

Figure 3.

(A) Collision-induced dissociation (CID) mass spectra of [Cu₆NEM₁₁-MT]⁵⁺, and (B) labelled CID fragment ions and Figure S3 contains the top-down fragment ion spectrum obtained using 2 dimension IM-MS data acquisition. The fragment ions assigned in Figure 3(A) are labeled in panel B with all b ions labeled as red, y ions containing NEM labels are labeled as blue, and y ions containing coppers are labeled as brown. As noted above the Gly-Ser tag marked in blue are not included in numbering of the tryptic cleavage products.

Figure S6 contains mass spectra and list of peaks corresponding to fragments containing Cu ions and/or NEM labels for trypsin digested Cu₆NEM₁₁-MT and Cu₁₀NEM₅-MT. Metal-free fragment ions of Cu₆NEM₁₁-MT correspond to Ser32-Lys43 that contains 5 NEM labels and Cys44-Ala61 with 6 NEM labels, which accounts for the 11 NEM labels, and strong evidence that the 6 Cu ions are located in the β domain and which contains 8 possible Cu⁺ ions binding sites, e.g., Cys 5, 7, 13, 15, 19, 21, 24, 27, between GS-Met1 and Lys31. This assignment is supported by the fact that the fragment ions GS-Met1-Lys30 and GS-Met1-Lys31 do not contain NEM labels. Collectively, these data are interpreted as evidence for formation of a Cu₆Cys₉ cluster in the β domain.

The mass spectrum of digested Cu₁₀NEM₅-MT contains a small number of low abundance product ions that limit our abilities to definitively assign the metal ion binding sites. Nonetheless, a fragment ion corresponding to GS-Met1-Lys43 (m/z = 1364.36 (4⁺)) containing 6 Cu ions (out of the 10) and 5 NEM labels is consistent with the assignment of 6 Cu bound in the β domain. The fragment ion corresponding to Ser32-Lys43 (893.34(3+)) contains 5 NEM labels and no Cu ions, evidence that this region of the protein is metal free in both Cu₆ and Cu₁₀-MT. The fragment ion corresponding to Cys44-Ala61 (m/z 666.91 (3⁺)) contains 4 Cu ions and is consistent with the presence of a Cu₄Cys₆ cluster in the α domain of Cu₁₀NEM₅-MT. Definitive evidence for these assignments are the detected fragment ions corresponding to Cys44-Ala61 with 4 Cu and 2 NEM (m/z 750.98 (³⁺)) and Cys44-Ala61 with 6 NEMs (m/z 834.66 (³⁺)). It appears that the Cu ions bound in the α domain may be labile and subject to displacement by NEM. We previously showed evidence that Cd ions can be displaced by NEM, and we have previously noted that Cd₇-MT-2A complex was also resistant to tryptic digestion.^{1, 2}

Mixed Metal MT Clusters: Competitive binding of Cu, Ag and Cd.

Chen et al.^{1, 2} and Dong et al.^{13, 14} previously reported domain specific binding for Cd-MT and Ag-MT; Cd²⁺ preferentially binds to the α domain, whereas Ag⁺ prefers to bind to the β domain, and these same preferred binding sites were found for Cd/Ag mixed metal systems.¹⁴ Dong et al.¹⁴ reported that Ag/Cd mixed metal complexes (Ag₄β-Cd₄α-MT and Ag₅β-Cd₄α-MT) are formed independent of the order in which the metal addition occurs, and they found no evidence for metal swapping between domains. Here, we investigate metal binding preferences and dynamics using mixed-metal titration of MT in solutions of Cu/Cd and Cu/Ag (Figure 4A and 4B). From prior studies we know that Cd-MT complexes are Cd₃β-Cd₄α-MT,^{1, 2} and we find similar stoichiometries for the Cu/Cd system. Solutions that contain low concentrations of Cu and Cd ions favor formation of M₄-MT product ions, e.g., Cu₃Cd₁, Cu₂Cd₂ and Cu₁Cd₃, and higher metal ion concentrations produce M₇-MT complexes, e.g., Cu₄Cd₃, Cu₃Cd₄, Cu₅Cd₂. We also detect low abundance product ions M₁₁-MT (Cu_x-Cd_y, x = 6, 7, 8 and y = 3, 4, 5). The Cu/Cd system yields products including Cu₃Cd₄ and Cu₂Cd₅ whose number of Cd equal or beyond the stable domain specific Cd₄. We interpret these results as evidence that Cu is competitive with Cd in forming these cooperative products, even the preferred domain of Cu is β and the preferred domain of Cd is α domain. Similar behavior was not observed for the Ag/Cd system.¹⁴ For comparison (see Figure S8), Cu was added to an equilibrated Cd₄-MT solution. At lower concentrations, Cd₄ was detected as the major product, along with some Cu-displaced peaks detected at low abundances. For solutions containing excess Cu, more Cu-displaced products were observed. This result reveals that Cu binding is competitive with Cd when adding the mixture to apo-MT, but the affinity of Cu binding to the α domain does not displace the stable Cd₄ without an excess concentration. These results agree with Cd binding in the α domain as being more stable, which is also in agreement with results obtained by Li et al. who reported Cu displacement of Cd₇ in the β domain followed by displacement in the α domain at high concentration of Cu (up to 12 equivalents).⁴⁸

Figure 4.

Mass spectra of (A) Cu/Cd and (B) Cu/Ag-MT mixed metal complexes. Data are shown for metal ion concentration (per MT) of 3 equivalents (45 μM) and 5 equivalents (75 μM), respectively. The mixed-metal products are assigned based on the numbers of each metal ion. In some cases the product ions overlap, but high-resolution mass analysis allows for accurate assignments, a specific example for partial overlapping of Cu₁Ag₃-MT with the low abundance Cu₆-MT signals is shown in Figure S7.

Mixed metal Cu/Ag complexes having stoichiometries of M₄ and M₆, are formed upon mixing solutions containing equimolar concentrations of each metal ion. At low concentrations of metal ions, Cu₄Ag_o, Cu₃Ag₁, Cu₂Ag₂, Cu₁Ag₃ and Ag₄ are the major products, and at higher concentrations low abundances of M₆ Cu_iAg_j (i = 2, 3, 4 and j = 2, 3, 4) and M₁₀ products, Cu_iAg_j (i = 6, 7 ,8 and j = 2, 3, 4) are formed. It appears that both Cu and Ag are bound in the β domain, forming a stable M₄ complex, but at higher concentrations an M₆ complex is favored. As metal concentration is increased, the numbers of Ag remain smaller than 4 in the M₁₀ products, while the numbers of Cu increase to greater than 4, suggesting that excess Cu⁺ ions are bound to the α domain to form the Cu₄ α structure, and this result suggests that Ag only cooperatively binds to the β domain.

Comparison of stabilities of Cu_i-MT and Ag_i-MT complexes (i = 4, 6, 10) and mixed metal Cu/Cd and Cu/Ag clusters as determined by collision-induced unfolding (CIU)

Collision cross section (CCS) profiles of metalated (Cd, Ag and Cu) MT are similar, which suggests that the overall shapes of these complexes are similar, but the broad CCS profiles are indicators of some degree of conformational heterogeneity. CCS information can be augmented by collision-induced unfolding (CIU), which can be used to monitor changes in CCS, e.g., conformational changes, of the gas-phase ionic complex induced changes in the internal energy of the ion,^{46, 49} an approach that is analogous to solution phase melting.⁴⁹ For comparisons, CIU heat maps for Ag-MT and Cu-MT are shown in Figure 5B, and CIU heat maps for mixed metal complexes (Cu/Cd and Cu/Ag) are shown in Figure 5C. CCS profiles used to generate these heat maps are shown in Figure S9, and those for Ag-MT complexes are shown in the original paper (Dong et al.).¹⁴ Prior to collisional activation the CCS profiles for Cu₄-MT span a range of CCS from 750 – 800 Ų, and the CCS profile contains low abundance signals corresponding to CCS of ~875 and 950 Ų. The CCS peak widths and the appearance of multiple peaks suggest that the ion population is highly heterogeneous in terms of conformation. At higher collision voltage (~50 V) the abundances of the 850 to 950 Ų peaks increase, and at even higher energies Cu₄-MT states with CCSs of ~1000 Ų are detected. The peak profile for the ion population with average CCS of 1000 Ų shows peaks are not resolved; the asymmetrical broadening in the CCS profiles, suggests a conformationally heterogeneous population of ions. The CCS for the initial and final states for Cu₆-MT are similar to those for Cu₄-MT; however, they differ in terms of the initial peak widths for the lowest energy states and the higher threshold energies for the first unfolding transition, which suggests lower degree of conformational heterogeneity and higher stability. The CCS for the intermediate state(s) (~875 Ų) is slightly smaller and narrower than that of the intermediate state of Cu₄-MT complex. In addition, at collision energies of ~175 eV another intermediate state with slightly larger CCS appears. It appears that this state is formed by rearrangement of some members of the higher CCS conformers of the extended Cu₆-MT complexes. The final state for Cu₆-MT complex appears at ~75 eV collision energies, and the narrow CCS profile is consistent with an ion population having a lower degree of conformational heterogeneity than that for Cu₄-MT.

Figure 5.

(A-C) CIU heat maps comparing Cui-MT and Agi-MT (taken from reference 14), where i = 4, 6, 10. Metal binding to β and α domains are very specific, both Cu and Ag preferentially bind to the β. The M4β and M6β complexes have ordered configurations and the α domain is more flexible, owing to weaker interactions with the cysteine core. The M6β-M4α have two ordered domains as evidenced by the two populations of ions having CCS of ~900 Å2 and ~950 Å2, that upon activation can reorganize to change the overall shape of the complex (see Figure S8).¹⁴ (B) and (C) contain CIU heatmaps for mixed metal-MT5+ complexes of Cu/Cd-MTs and Cu/Ag-MT, respectively. The heat maps for Cu3Cd1 and Cu2Cd2 are interpreted as evidence for two distinct unfolded conformer populations. These intermediate conformers (CCS ~890 Å2) are formed at similar collision energies; however, the Cu2Cd2 intermediate complex appears to be more stable than the Cd3Cd1 complex, e.g., similar CCS profiles are detected across a range (870 – 910 Å2) of collision energies. For both complexes the CCS (~950 Å2) of the final product(s) are similar. The heat maps for Cu3Cd4 and Cu2Cd5 suggest these ions have similar stabilities, and the CCS of the final products of CIU are quite similar. It is interesting, and possibly surprising, that the heat maps for Cu/Ag complexes are essentially identical, both in terms of the number of CCS families and the collision energies at which each domain is populated. Close inspection of the CCS profiles (Figure S9) reveals some differences in the abundances of the intermediate and final states for Cu3Ag3 and Cu2Ag4; however, the origins of these differences are not fully understood.

CIU of the Cu₁₀-MT complex does not show evidence for unfolding below ~125 eV, and the unfolded state is stable up to ~ 240 eV, consistent with a more rigid and compact conformation. Note also that the unfolded state(s) has a CCS centered at about 900 Ų, which is similar to the intermediate states of Cu₄-MT and Cu₆-MT. The smaller CCS is also evidence of the stabilization.

Note that the CCS profiles for both Cu₄-MT and Ag₄-MT are quite broad (see CCS profiles in Figure S8) and the first CIU transitions occur at different collision energies. Moreover, the CCS of the first unfolded states (~875 Ų) Cu₄-MT and Cu₆-MT are similar, and both have Cu bound in the β domain. Based on these data we propose two possible arrangements for Cu₄ binding in Cu₄-MT, with Cu₄ located in the β domain but we cannot rule out the possibility of a less favored product having Cu₄ in the α domain. The slightly larger CCS for the unfolded Cu₄-MT corresponds to the unfolded β domain while the α domain is compact. The observed changes in CCS may also be an effect of weaker interaction of Cu in the β domain with Cys5 and Cys7. The CCS of Cu₆-MT has a more narrow, conformationally homogenous, unfolded state than that of Cu₄-MT, and CIU occurs at slightly higher energies to form products having CCS centered at ~980 Ų, which is consistent with a compact conformation in the β domain and an extended α domain.

CIU experiments were also performed on selected mixed metal complexes; the choices for which data to present was based on signal abundances, e.g., low abundance ions yielded poor data quality (Figure 5C, 5D). The heatmaps for Cu₃Cd₁ and Cu₂Cd₂-MT show significant differences in that the CIU transitions for the intermediate state at ~900 Å are shifted to higher collisional energy in Cu₂Cd₂-MT than Cu₃Cd₁ and for the final state at ~1000 Å, whereas the CIU transitions for Cu₃Cd₄ and Cu₂Cd₅-MT are almost identical. The products of the Cu/Ag system containing similar numbers of metal ions undergo similar unfolding transitions and similar CCS, showing that the products in the Cu/Ag system are more alike than those in the Cu/Cd system. Moreover, it is interesting to note the 2 unfolded states for the 6-metal Cu/Ag system, whose behavior is more similar to Ag₆, while the middle one at ~900 Å is closer to the value of the attributed Cu₆-MT “refolded” state. These results suggest a high degree of structural conformational heterogeneity, possibly owing to different binding geometries for Ag₄ and Ag₆ clusters in the β domain as was previously noted for 2 Ag ions that appear to be weakly bound to Cys5 and Cys7 of the β domain.¹⁴

Conclusions

Nano-ESI-MS combined with ion mobility spectrometry is used to determine the stoichiometry, metal ion binding sites and stabilities of Cu-metallothionein-2A complexes. Titration of metallothionein 2A by Cu(I) yields clusters of the general type of MT-Cu_i, i = 4 – 19, with Cu₄, Cu₆ and Cu₁₀ being most abundant. While Cu₄-MT is formed early in the titration, its low abundance relative to that of Cu₆ is an indicator of favorable cooperativity for the formation of Cu₆-MT. Likewise, the low abundances of Cu₈-MT and Cu₉-MT reflect favorable cooperativity for formation of Cu₁₀MT. While Cu₁₄-MT is formed first during the titration, the Cu₁₃-MT cluster becomes the dominant product at higher Cu concentrations and longer reaction times. These higher-order Cu-MTs could be formed by addition of Cu⁺ to the Cys of the α domain (Cys33-Cys41) as well as restructuring the Cu binding in the Cu₆-MT Cu₁₀-MT complexes. We speculate that Cu₁₃-MT could be formed by addition of three Cu ions to the α domain of Cu₁₀-MT followed by addition of four Cu ions to form the Cu₁₇-MT complex. Attempts to grow crystals for further crystallographic analysis have not been successful; such difficulties have been noted for other metal-MT complexes. We speculate that conformational heterogeneity of the complexes is the underlying reason; note that in all cases the CCS determined by ion mobility spectrometry are quite broad, as expected for conformationally heterogeneous populations of complexes.

Top-down and bottom-up proteomics analysis provides strong evidence that Cu ions bind to both domains of the MT, with Cu₆ binding to the β domain more stable than Cu₄ binding in the α domain. Based on results from top-down proteomics, the overall structure of Cu₁₀-MT is assigned as Cu₆β+Cu₄α; this structure is also consistent with bottom-up proteomics data. Notably, some of the products obtained from trypsin digest suggest that NEM can partially displace Cu from the α domain, providing further evidence of greater stability for Cu binding in the β domain. Note that the Cu₆Cys₉ cluster occupies all cysteines in the β domain; CID analysis of the Cu₆-MT complex shows that Cys5 and Cys7 are either not involved in Cu binding or are only weakly binding. We previously reported similar behavior for Ag-MT complexes. Overall, the results for Cu-MT-2A are in general agreement with results for Cu-MT-1A reported by Scheller et al.^{22, 23}

Collision-induced unfolding (CIU) of Cu-MT complexes provide further evidence for increased stabilities of the Cu-MT complexes, but not necessarily diminished conformational heterogeneities. The relative stabilities of the complexes increase as the numbers of bound metal ions increase, viz. threshold energies for CIU increase in the order Cu₁₀-MT > Cu₆-MT > Cu₄-MT. The CIU profiles for Cu-MTs differ from those for Ag-MT that have similar numbers of metal ions, viz. Cu₄-MT shows greater instability and heterogeneity than Ag₄, but Cu₁₀MT appears to be more stable than Ag₁₀. We infer that smaller CCS (870 Ų) detected at collision energies > 125 eV reveal formation of a refolded state and suggests reorganization of the Cu binding. The CIU heat map for Cu₆-MT is very different from that observed for Ag₆-MT, both in terms of threshold energies for changes in CCS and the abundances of the intermediate states. The temperature-induced unfolding (TIU) of the Cu-MTs as measured by variable-temperature ESI (Figure S10) reveal differences in the stabilities of solution and gas phase Cu₁₀-MT complexes, which suggests that water may play an important role in stabilizing these complexes. Further studies on these effects are currently underway.

Cu shows a greater diversity for forming mixed-metal clusters with Cd and Ag. For example, the domain specific binding of Cd and Ag is retained in the mixed metal complexes, but this behavior is not found for Cu mixed metal complexes. The numbers of Cu/Cd and Cu/Ag follow specific patterns that are similar to the single metal results (Cd₄, Cd₇, Ag₄, Cu₆, Cu₁₀). Although assignment of binding sites is not possible, it appears that Cu binding with Ag or Cd retains the domain preference and cooperativity of Ag or Cd. Previous studies presented evidence that Zn binds to MT with the least domain specificity, and Cu cooperatively binds to MT with both Cd and Ag while mixing Cd²⁺ and Ag⁺ can only form the domain specific products as considered in the Ag-MT paper.¹⁴ The CIU heatmap features of these species raise more questions about the metal-metal interactions and potential intramolecular migration in the MT structure, as the change of a single metal in the composition of Cu/Cd-MT significantly affects the unfolding behavior.

Overall, integrated analytical methods play increasingly important roles in many areas of biology, biochemistry and biophysics, and MS-based approaches are becoming increasingly important among these methods. Although many questions remain unanswered, the combination of mass spectrometry and ion mobility provides abundant structural information of the intrinsically disordered protein metallothionein, including the binding behavior of different metals and also the properties of metalated MTs. Information derived from MS-based approaches can serve as the basis for further studies on MT about their physiological functions and pathways. The MS-IMS approaches do not address other important issues, including the actual binding geometry of metal to thiolates. For example, Cu⁺ and Ag⁺ adapt to one or both of the diagonal and trigonal binding modes,^50-53 which would explain different observations for Cu- and Ag-MT. However, the reported metal-cysteine clusters of Ag and Cu-MTs are M₄Cys₆ and M₆Cys₉ that both of them are easier to propose the geometry when binding trigonally. So other reasons should be also considered, including that the ionic radius of Cu⁺ is smaller than Ag⁺, which could be a reason to allow a cluster with more cysteines. Also the cooperativity associated with possible Cu binding to the Cu-cysteine cluster remains unclear, and similar level of cooperativity is observed for Ag.