Structural Characterization of Cu(I)/Zn(II)-metallothionein-3 by Ion Mobility Mass Spectrometry and Top-Down Mass Spectrometry

Mammalian zinc metallothionein-3 (Zn7MT3) plays an important role in protecting against copper toxicity by scavenging free Cu(II) ions or removing Cu(II) bound to β-amyloid and α-synuclein. While previous studies reported that Zn7MT3 reacts with Cu(II) ions to form Cu(I)4Zn(II)4MT3ox containing two disulfides (ox), the precise localization of the metal ions and disulfides remained unclear. Here, we undertook comprehensive structural characterization of the metal-protein complexes formed by the reaction between Zn7MT3 and Cu(II) ions using native ion mobility mass spectrometry (IM-MS). The complex formation mechanism was found to involve the disassembly of Zn3S9 and Zn4S11 clusters from Zn7MT3 and reassembly into Cu(I)xZn(II)yMT3ox complexes rather than simply Zn(II)-to-Cu(I) exchange. At neutral pH, the β-domain was shown to be capable of binding up to six Cu(I) ions to form Cu(I)6Zn(II)4MT3ox, although the most predominant species was the Cu(I)4Zn(II)4MT3ox complex. Under acidic conditions, four Zn(II) ions dissociate, but the Cu(I)4-thiolate cluster remains stable, highlighting the MT3 role as a Cu(II) scavenger even at lower than the cytosolic pH. IM-derived collision cross sections (CCS) reveal that Cu(I)-to-Zn(II) swap in Zn7MT3 with concomitant disulfide formation induces structural compaction and a decrease in conformational heterogeneity. Collision-induced unfolding (CIU) experiments estimated that the native-like folded Cu(I)4Zn(II)4MT3ox conformation is more stable than Zn7MT3. Native top-down MS demonstrated that the Cu(I) ions are exclusively bound to the β-domain in the Cu(I)4Zn(II)4MT3ox complex as well as the two disulfides, serving as a steric constraint for the Cu(I)4-thiolate cluster. In conclusion, this study enhances our comprehension of the structure, stability, and dynamics of Cu(I)xZn(II)yMT3ox complexes

M ammalian metallothioneins (MTs) are small (∼6−7 kDa) proteins that are rich in cysteine residues and play a crucial role in the metabolism of zinc and copper (Zn(II) and Cu(I)). 1−3 MTs also act as a detoxification system by binding potentially harmful metal ions such as Cd(II), Pb(II), Ag(I), or Hg(II). 4 There are a dozen different MT proteins (MT1− MT4 isoforms and multiple subisoforms), each with distinct metal-binding properties and tissue localization. 5 MT1 and MT2 have widespread tissue expression, while MT3 and MT4 are specifically found in the central nervous system (CNS) and stratified epithelial tissue, respectively. 6,7 Initially identified as a neuronal growth-inhibitory factor in 1991, MT3 interacts with brain proteins and plays a role in regulating Zn(II) trafficking and controlling neurodegeneration in the CNS. 7,8 MT3 acts as a Zn(II) sensor in the synaptic cleft, regulating synaptic exocytosis and preventing neurodegenerative disorders by swapping Zn(II) for excess copper (Cu(II)/Cu(I)) in the brain. 9 So far, there has been only one X-ray structure solved for the hepatic rat mix Cd 5 Zn 2 MT2 species. 10 The protein folds into a unique dumbbell-shaped structure, with the α-domain forming a Cd 4 Cys 11 cluster and the β-domain containing the Cd 1 Zn 2 Cys 9 cluster. 10 In general, when divalent metal ions (M) are present, the 20 cysteine residues (21 in the case of MT1b) form metal-binding clusters, with the N-terminal βdomain forming an M 3 Cys 9 and the C-terminal α-domain forming an M 4 Cys 11 . 5 However, MT3 has several structural differences that may contribute to its functional differences from those of other MT isoforms. It is the only mammalian MT isoform with two prolines in its β-domain and an acidic hexapeptide loop in its α-domain. 11 This hexapeptide loop was found to be responsible for the low stability of the metal cluster in the α-domain, likely due to the increased solvent-exposed surface area of the polypeptide. 12−14 Additionally, unlike most MTs, MT3 expression is not induced by metal ions, suggesting a different biological role compared to other MT isoforms. 15,16 Due to its unique structural features, such as the presence of the prolines in the conserved T 5 CPCP 9 motif in the β-domain, MT3 displays stronger copper binding character than MT1 or MT2. As a result, it is purified from mammalian brains as a heterometallic complex, Cu(I) 4 Zn(II) 3−4 MT3. 17−20 In vitro studies have shown that Zn 7 MT3 can scavenge free Cu(II) ions or remove Cu(II) bound to amyloid β peptides and αsynuclein, thereby eliminating harmful redox chemistry. 21−25 This highlights the potential role of MT3 in controlling low levels of free Cu(II) in the brain and mitigating the effects of oxidative stress, which is implicated in the pathogenesis of several neurodegenerative disorders. 26 The mechanism of the reaction involves a Zn(II)-to-Cu(I) exchange, reducing Cu(II) to Cu(I) through the formation of two intramolecular disulfide bonds (indicated here as "ox") (Cu(I) 4 Zn(II) 4 MT3 ox ); however, their localization has not yet been thoroughly investigated. 27 The results of biophysical characterization of the Cu(I) 4 Zn(II) 3−4 MT3 ox complex showed that copper forms a Cu(I) 4 -thiolate cluster, likely located in the β-domain, which has been shown using isolated domains. 27−29 The close proximity of the Cu(I) ions in the cluster, with distances less than 2.8 Å, allows for d 10 -d 10 orbital overlap and provides exceptional stability and resistance to oxidation in air-exposed conditions. 28 Mass spectrometry techniques, including nanoelectrospray mass spectrometry (nano ESI-MS) and ion mobility-mass spectrometry (IM-MS), have been effectively used to analyze heterogeneous protein systems, enabling the characterization of their conformation and dynamics. 30−35 Meloni and Vasǎḱ measured the masses of the complexes formed upon the titration of Zn 7 MT3 with Cu(II) ions, finding that the Cu(I) 4 Zn(II) 4 MT3 ox should contain two disulfides. 27 In the same work, the Cu(I) 4 -thiolate cluster was localized based on isolated domains obtained through subtilisin digestion and measured by low-temperature luminescence. Recently, Fan and Russell utilized IM-MS to analyze the Cu(I)-MT2 complexes that were formed upon the addition of Cu(I) (not Cu(II)) to metal-free MT2 (apo-MT2). 35 Their findings revealed the existence of several species including αCu 4 MT2, βCu 6 MT2, and βCu 6 αCu 4 MT2. These results are in line with those from the Stillman group, who studied Cu(I) binding to apoMT1a and showed that the α-domain binds four Cu(I) ions, while the β-domain binds six Cu(I), forming copper-thiolate clusters. 36,37 In their recent research, they showed that Cu(I) (not Cu(II)) binds to Zn 7 MT3 forming multiple species, with no preference for Cu 4 Zn 4 MT3. 38 In another study, Melenbacher et al. studied Cu(I) binding to Zn 7 MT1a instead of apoMT1a. 39 Their results suggested the formation of two main complexes, βZn 1 Cu 5 αZn 4 MT1a and βCu 6 αZn 4 MT1a. Furthermore, in the presence of excess glutathione (GSH), Austin and co-workers reported, based on ITC analysis, that Zn 7 MT2 and Zn 7 MT3 can form Cu(I) 4 -thiolate clusters, resulting in βCu 4 αCu 4 MT2 ox and βCu 4 αCu 4 MT3 ox . 40 Here, using native ion mobility mass spectrometry, we characterized the conformation and dynamics of the metalprotein complexes formed by the reaction between Zn 7 MT3 red (where "red" denotes reduced cysteine residues in the protein) and Cu(II) ions. Our results showed the formation of Cu(I) 4 Zn(II) 4 MT3 ox , which contains two disulfides and has four Cu(I) ions bound to the β-domain in the full-length protein. IM-MS provides information about charge state distribution (CSD) and collision cross section (CCS). 41 Extended conformations with larger CCS experience more collisions with the buffer gas, resulting in a longer drift time or reduced mobility. The complex was found to be smaller and less flexible than Zn 7 MT3 red , as indicated by a lower and narrower distribution of traveling-wave-derived collision cross section ( TW CCS N2 ) values. Gas-phase activation of protein ions via collisional activation (CA) can be used to probe subtle structural differences between similar conformations and study protein ion stability and dynamics. 42−44 Recording IM-MS under different CA conditions showed that Cu(I) 4 Zn-(II) 4 MT3 ox was more compact and stable compared to Zn 7 MT3 red . The formation of Cu(I) 4 MT3 ox (full-length protein) upon acidification suggests that MT3 is an effective Cu(II) scavenger even at low pH. Native top-down collisioninduced dissociation (CID) experiments reveal that while Zn(II) ions were partially distributed in both αand β-domains in Zn 7 MT3 red , the Cu(I) ions were found exclusively in the βdomain in Cu(I) 4 Zn(II) 4 MT3 ox.

■ EXPERIMENTAL SECTION
We overexpressed and purified MT3 (Addgene plasmid ID 105710) in a bacterial system as described in the Supporting Information. UV−vis spectra were recorded on a JASCO V-750 spectrophotometer at 25°C in 25°C, in a 1 cm quartz cuvette as described in the Supporting Information.
MS and IM-MS experiments were carried out on a Synapt XS HDMS instrument equipped with nanoelectrospray ionization (Waters Corporation, Manchester, UK). 3−10 μL of sample (10−20 μM in 200 mM ammonium acetate) was loaded into borosilicate glass capillaries (O.D. 1.2 mm, I.D. 0.9 mm, World Precision Instruments, Stevenage, UK) produced in-house using a Flaming/Brown P-1000 micropipette puller (Sutter Instrument Co., Novato, CA, USA), and ions were produced by applying a positive potential of 0.9−1.4 kV via a platinum wire (Goodfellow). Two sets of TWIMS parameters of traveling wave (TW) velocity and height were used, 300 ms −1 and 20 V and 480 ms −1 and 20 V. Collision-induced unfolding (CIU) experiments were performed by increasing trap collision energies (0−60 V range) of quadrupole-selected ions and recording ion arrival time distributions. We used ubiquitin (bovine), cytochrome C (equine heart), and βlactoglobulin (bovine milk) to calibrate the TW device. 45,46 Native top-down collision-induced dissociation (CID) mass spectrometry experiments were performed by applying 20−60 V trap collision energies of quadrupole-selected ions with argon as the collision gas. Calculation of the survival yield curves 47,48 and data analysis information 49 can be found in the Supporting Information.

pubs.acs.org/ac Article
To perform top-down electron transfer dissociation (ETD) mass spectrometry experiments, the sample was introduced using ESI via a syringe with a 3 μL·min −1 flow rate and a spray voltage of 2.5 kV. The glow discharge was tuned to obtain an ETD reagent (1,4-dicyanobenzene) current of ∼1e6 counts/s for charge reduction. The anions were accumulated in the trap collision cell for 100 ms using a refill interval of 1 s. The reaction was started by lowering the wave height from 1.5 to 0.2−0.3 V, using a wave velocity of 300 ms −1 . Data were analyzed by means MassLynx v4.2 (Waters Corp., UK), ORIGAMI, and custom scripts in Python 3.5 (available in https://github.com/ManuelPerisDiaz/Cu-Zn-MT3). 46

■ RESULTS AND DISCUSSION
To monitor the reaction between Zn 7 MT3 red and Cu(II) (CuCl 2 ) and the assembly of a Cu(I) 4 -thiolate cluster, we utilized isotopically resolved native mass spectrometry (MS) data. The nESI mass spectra of Zn 7 MT3 red sprayed with 200 mM AmAc (pH 6.8) showed a charge state distribution (CSD) covering three charge states, 3 ≤ z ≤ 5 ( Figure S1). Upon incubating the Zn 7 MT3 red protein with 1 CuCl 2 mol equivalent (eq), we observed the rapid and cooperative formation of the Cu(I) 4 Zn(II) 4 MT3 ox complex along with other metal-MT3 species ( Figure S1). As the molar eq of CuCl 2 was increased, new signals appeared in the mass spectrum, which temporarily overlapped with previous signals until the previous ones eventually disappeared. The assignment of metal stoichiometries in systems containing both Cu(I) and Zn(II) ions presents a challenge, as the two metals differ only by an average mass of 1.8 Da, and their isotope patterns overlap. To determine the stoichiometry accurately, multiple theoretical protein isotopic distributions were generated, and the one with the best fit was selected (Table S1). Figure S2 depicts the computational workflow for generating multiple isotopic patterns, fitting them to the experimental data, and scoring the results. Accurate mass measurements and fitting of the isotopic distributions indicated that Cu(II) is reduced to Cu(I) by oxidizing four cysteine residues from the protein and displaces Zn(II) from Zn 7 MT3 red to form Cu(I) 4 Zn-(II) 4 MT3 ox with two intramolecular disulfide bonds ( Figure  1A,B, Table S1). The transition from a tetrahedral Zn(II) environment to diagonal/trigonal Cu(I) coordination occurred without proton dissociation or association. All Cys residues in Zn 7 MT3 red were bound to seven Zn(II) and were deprotonated, and since we did not observe a change in the number of protons during the displacement of Zn(II) by Cu(I) to form the Cu(I) 4 Zn(II) 4 MT3 ox , we infer that the noncoordinating Cu(I) residues in the Cu(I) 4 Zn(II) 4 MT3 ox complex have formed disulfides ( Figure 1B).
We also observed all series from tetra-to heptametallic complexes, which suggest that the Cu(I)-to-Zn(II) swap proceeds by first Zn(II) cluster disassembly (Zn 3 S 9 and Zn 4 S 11 ) and then reassembly following a probability distribution ( Figure 1B). We do not expect that these partially metalloaded MT3 species are a result of gas-phase dissociation, as we could maintain seven intact Zn(II) bonds bound to Zn 7 MT3 red ( Figure 1A). Other higher Cu(I)-loaded states were also present, in particular, Cu(I) 5 Zn(II) 4 MT3 ox and Cu(I) 6 Zn(II) 4 MT3 ox (Table S1). To shed more light on this process, we monitored the reaction of Zn 7 MT3 red with Cu(II) by UV−vis spectroscopy ( Figure S3). Metal binding or dissociation can be studied by observing the ligand-to-metal charge transfer (LMCT) transition in the middle-to-far UV range. 27,28 Analysis of the absorption spectrum demonstrates two major LMCT bands centered at 215 and 255 nm, the CysS-Zn(II) LMCT and the CysS-Cu(I) LMCT, respectively. Varying intensities of the LMCT bands result in an isosbestic point at 230 nm ( Figure S3). The intensity of the higher energy band (CysS-Zn(II) LMCT) linearly decreases with the addition of successive portions of Cu(II), with a simultaneous linear increase in the intensity of the S−Cu(I) charge transfer band at 255 nm. This trend continues until ∼8 Cu(II) eq are added when the final complex form is formed. This is very consistent with previous spectroscopic results and indicates that Zn 7 MT3 red is able to bind and reduce Cu(II) to Cu(I) with concomitant Zn(II) displacement and disulfide formation ( Figure S3). 27,28,50 To confirm the presence of two disulfides in the Cu(I) 4 Zn-(II) 4 MT3 ox complex, we measured the native and denaturing mass spectra of Zn 7 MT3 red and Zn 7 MT3 red after the addition of 2 and 4 CuCl 2 molar eq (Figure 2A,B). We observed that apoMT3 derived from native Zn 7 MT3 could be fit to a reduced state ("red"), and the disulfide formation is a function of CuCl 2 molar eq ( Figure 2C). Upon acidification of the Simulations of theoretical isotopic patterns for individual proteins were plotted as stem plots. The molecular formulas can be found in Table S1. "red" and "ox" subscripts refer to reduced and oxidized (two intramolecular disulfides) MT3 proteins. Analytical Chemistry pubs.acs.org/ac Article native complex obtained after incubation of Zn 7 MT3 red with expected 1:4 stoichiometric CuCl 2 molar eq, only the oxidized form of apoMT3, apoMT3 ox , with two disulfides, is visible ( Figure 2C). Interestingly, the Cu(I) 4 MT3 ox intermediate formed predominates with only a 1:4 Zn 7 MT3:CuCl 2 stoichiometry ( Figure 2B). This shows that the Cu(I) 4 -thiolate cluster is greatly stabilized by the presence of disulfides, likely acting as a steric constraint, preventing the escape of Cu(I) ions. The fitting of the isotopic distributions can be found in Figure S4A,B and Table S1. The disulfides from the Cu(I) 4 Zn(II) 4 MT3 ox complex were reduced in the presence of 1 mM TCEP, as observed by the mass shift ( Figure S4C). We then examined the thermodynamics of Zn(II) dissociation in the Cu(I) 4 Zn(II) 4 MT3 ox complex by gradually increasing the concentration of EDTA added ( Figure S5). We observed a sequential Zn(II) dissociation from the Cu(I) 4 Zn-(II) 4 MT3 ox complex that leads to the Cu(I) 4 MT3 ox intermediate. The Cu(I)-thiolate cluster was then disrupted, resulting in the formation of apoMT3 ox with Cu(I) 1 MT3 ox as an intermediate. Such a mechanism slightly differs from that in the previous experiment based on cysteine residue protonation ( Figure 2). With all this body of work, we may conclude that the oxidized β-domain in full-length MT3 can accommodate up to six Cu(I) ions ( Figure 1B), but the product with four Cu(I) seems to be preferred, as it forms a well-defined Cu(I)thiolate cluster.
Native and Activated Ion Mobility-Mass Spectrometry to Study Conformational Properties and Stability of Cu(I)/Zn(II)-MT3 ox Species. Following the stoichiometry and thermodynamic characterization of the Cu(II) reaction with Zn 7 MT3 red , we then attempted to investigate the conformational properties of the Cu(I) x Zn(II) y MT3 ox complexes by native IM-MS. The estimated TW CCS N2 values are presented in Table S2 and Figure S6. The Zn 7 MT3 red 5+ ions exhibit a broad TW CCS N2 distribution, centered at ∼1069 Å 2 , while the TW CCS N2 for Zn 7 MT3 red 4+ ions reveal a single compact conformation at 1006 Å 2 ( Figure 3A). The swap from the Zn(II) 3 -to Cu(I) 4 -thiolate cluster induces a structural change that leads to several complexes, including Cu(I) 4 Zn(II) 3 MT3 ox species with lower TW CCS N2 values centered at ∼1020 and 992 Å 2 for 5+ and 4+ ions.
Since the ionic radius of Cu(I) is comparable to that of Zn(II) ions, the presence of two disulfides in the protein together with the Cu(I) 4 -thiolate cluster reduces both its size and flexibility, resulting in narrower TW CCS N2 values. Saturating the protein with one Zn(II), forms the Cu(I) 4 Zn-(II) 4 MT3 ox species and results in a more compact conformer, with TW CCS N2 values of ∼990 and 980 Å 2 for 5+ and 4+ ions, respectively. This species represents the maximum number of Zn(II) that can bind in the presence of Cu(I). The addition of one more Cu(II) and formation of the Cu(I) 5 Zn(II) 4 MT3 ox species require the protein to open up its structure to accommodate the additional metal ion, as reflected in a slight increase in TW CCS N2 to ∼1020 and 998 Å 2 for 5+ and 4+ ions, respectively ( Figure 3A). On the other hand, one might expect an increase in TW CCS N2 for Cu(I) 4 MT3 ox compared to Cu(I) 4 Zn(II) 4 MT3 ox , as the dissociation of four Zn(II) should result in larger degrees of freedom. However, the cysteines that were bound to Zn(II) formed disulfides, as estimated isotopic pattern fitting, which in turn results in a compact conformation. This is demonstrated by the TW CCS N2 values, which are centered at ∼1000 and 977 Å 2 for 5+ and 4+ ions, respectively (Table S1). To conclude, looking at the apex of the TW CCS N2 distributions for 4+ ions seems to be useful in accurately discerning structural changes related to the stoichiometry and composition of metal ions and/or disulfides. On the other hand, the TW CCS N2 width for the 5+ ions is more sensitive to metal/disulfide-imposed degrees of freedom.
To examine the stability and dynamics of the Cu(I) 4 Zn-(II) 4 MT3 ox species formed upon Cu(I)-to-Zn(II) swap and the formation of a Cu(I) 4 -thiolate cluster, we performed IM-MS on mass-selected 5+ and 4+ ions under different collisional  Table S1. The m/z region corresponds to 5+ ions in all cases. "red" and "ox" subscripts refer to reduced and oxidized (two intramolecular disulfide) MT3 proteins. Note that all complexes of copper are Cu(I), and Zn(II), as discussed in the text.  Table S2.

Analytical Chemistry pubs.acs.org/ac
Article activation conditions ( Figure 3B−D). At low collision energy (CE), Zn 7 MT3 red 5+ ions exhibit a single compact conformation with TW CCS N2 ∼1000 Å 2 , but upon activation, some of the ions shift to extended conformations with TW CCS N2 ∼ 1300 Å 2 . A stable 60:40 ratio of compact to extended conformations was observed across a range of CE. This stability was not due to an oxidation-induced effect that trapped the compact conformer, preventing unfolding ( Figure S7). The Cu(I) 4 Zn-(II) 4 MT3 ox 5+ ions exhibit a more compact and stable conformation than Zn 7 MT3 red 5+ ions, with only partial activation upon increasing CE. The Cu(I) 4 Zn(II) 4 MT3 ox 4+ ions remained very stable and did not undergo ion activation, supporting the different information provided by the 4+ and 5+ charge states.

Native Top-Down Mass Spectrometry of Cu(I)/Zn(II)-MT3 ox Species.
A native top-down MS approach was employed to further characterize the metal-protein complexes. The Zn 7 MT3 red 5+ and Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions were quadrupole-selected and subjected to collision-induced dissociation (CID) ( Figure 4A,B). The CID spectra contain two m/z regions that provide different information. A 250−1250 m/z region contains fragment ions, and a 1250−1550 m/z region contains ions corresponding to the precursor and to the precursor with the loss of some metal ions. The latter region reveals interesting clues about the thermodynamic stability of metal-thiolate bonds. In Figure 4C,D, we observed how increasing the collisional activation of both Zn 7 MT3 red 5+ and Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions yielded the dissociation of four Zn(II) ions, albeit at different activation energies.
To estimate the different relative ion stabilities of the metalprotein complexes, we calculated survival yield (SY) plots of each precursor ion ( Figure 4E). SY is defined as the fraction of ions that did not fragment under particular activation energy, here normalized to center-of-mass energies (E com ). 47 Fitting the data to a sigmoid curve and the derived midpoint, known as E 50 , provides a quantitative estimate of the ion stability. In agreement with CIU experiments, the Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions were more stable than Zn 7 MT3 red 5+ ions (1.04 vs 0.95 eV). We observed a cooperative metal ion dissociation, whereby four metal ions dissociate without the formation of any intermediate upon CID (Figure 4C,D). This mechanism resembles the one obtained when the native metal-protein complex was disrupted by acidification, where the Cu(I) 4 Zn-(II) 4 MT3 ox 5+ ions yielded Cu(I) 4 MT3 ox 5+ (Figure 2). CID promoted dissociation of entropically favored Zn(II) ions, while the enthalpically and entropically favored Cu(I) remained bound to the protein. 40 Not only does the nature of the metal ion dictate the CID-induced metal dissociation mechanism but also the thermodynamics of each protein domain. Isothermal titration calorimetry (ITC) studies have shown that Zn(II) binding to MT3 is enthalpically disfavored and entropically driven. 40 The increase in entropy is attributed to Zn(II) and protein desolvation; cysteine deprotonation overcomes the conformational protein entropy induced by metal binding. In light of our results, the four metal ions dissociate from the α-domain, as there is a lower protein conformational entropic penalty than in the β-domain ( Figure  4F).
Dissociation of four Zn(II) ions from Zn 7 MT3 red 5+ and Cu(I) 4 Zn(II) 4 MT3 ox 5+ resulted in Zn 3 MT3 red 5+ and Cu-(I) 4 MT3 ox 5+ , and this was followed by covalent bond fragmentation ( Figure 4A,B). As a control that the structural changes are imposed by metal binding and not by the protein fold, we performed CID experiments on metal-free apoM-T3 red 5+ . We observed a CID spectrum with extensive metal-free y-fragment ions where the charge is retained by the α-domain C-terminus in both activated Cu(I) 4 Zn(II) 4 MT3 ox 5+ and Zn 7 MT3 red 5+ ions ( Figure 5A). For Zn 7 MT3 red 5+ , the larger α-domain y-fragment ion corresponds to y10, evidencing that upon Zn(II) dissociation by CID, the remaining three Zn(II) ions are partially redistributed in both the αand β-domains (Table S4). However, we cannot exclude Zn(II)-induced migration to "non-native" sites as a result of the Zn(II) release by CID. In the case of Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions, y-fragment ions appeared, ranging from y2 to y30, clearly supporting our conjecture that the four Cu(I) are bound exclusively to the βdomain ( Figure 5A, Table S3). For apoMT3 red 5+ , y-fragment ions ranging from y7 to y30 were also found ( Figure 5B, Table  S5). To reinforce our results, we sprayed the Cu(I) 4 Zn-(II) 4 MT3 ox sample under denaturing conditions where the four Zn(II) dissociated but the Cu(I) remained bound to MT3 forming Cu(I) 4 MT3 ox . The Cu(I) 4 MT3 ox 5+ ions were then mass-selected and subjected to collisional activation ( Figure  S8A).
We identified characteristic fragment ions and the fragmentation pattern for Cu(I) 4 Zn(II) 4 MT3 ox 5+ that clearly confirm our hypothesis. We obtained an almost complete series of y-fragment ions spanning the C-terminal α-domain, except for the range between y20 and y30. Since the α-domain consists of a total of 30 amino acids, our conclusion that the four Cu(I) ions are not bound in the α-domain is reasonable. Otherwise, fragmentation would not be observed because the presence of the metal would provide protection against fragmentation. We rule out the possibility of a Cu(I) ion binding in the 20−30 amino acid region, as a similar fragmentation gap was found for apoMT3, indicating that the absence of fragmentation is not attributed to metal binding . Survival yield plots for quadrupole-selected Zn 7 MT3 red 5+ (green) and for Cu 4 Zn 4 MT3 ox 5+ (red) (E). Schematic representation of the metal ion dissociation mechanism inferred from the CID experiments (F). The proteins (10 μM) were sprayed in 200 mM ammonium acetate (pH 6.8), and the activation was performed in the trap cell. "red" and "ox" subscripts refer to reduced and oxidized (two intramolecular disulfides) MT3 proteins. Note that all complexes of copper are Cu(I), and Zn(II), as discussed in the text. Analytical Chemistry pubs.acs.org/ac Article but rather to some protein folding effect. We found three characteristic y30 fragment ions that differed in the number of disulfides ( Figures 5C, S9). This can be understood as the protein region having a Zn(II) ion bound, and once it dissociates, it is prone to oxidation. As the dissociation of Zn(II) upon ion activation of Cu(I) 4 Zn(II) 4 MT3 ox 5+ is not "complete", we expected to find some Zn(II)-bound fragment ions. Indeed, we observed the same y30 ion carrying a single Zn(II) ion. To further support our findings, we also observed the metal-free y25 3+ and y30 3+ ions, but the Zn 1 y30 3+ disappeared for Cu(I) 4 MT3 ox 5+ ions ( Figure S8B). Together, this indicates that the four Zn(II) were bound to the α-domain and the four Cu(I) to the β-domain. In an attempt to further localize the Cu(I) binding sites, we employed top-down electron transfer dissociation (ETD) ( Figure S10A). The results showed that no fragmentation occurred via nondissociative electron transfer dissociation (ETnoD), 51 and additional CID supplementation in the transfer collision cell does not help in obtaining ETD fragmentation ( Figure S10B).
Examining the b-type fragment ions corroborated our findings that the α-domain is metal-free and the four Cu(I) are bound in the β-domain. The most abundant β-domain bfragment ions correspond to b4 to b6. In the absence of metal ions, we observed the most abundant β-domain b-fragment ions correspond to b4 to b6 but also visible are b14 and b15 ions ( Figure 5B). Upon Cu(I) binding, b14 and b15 ions disappeared, suggesting that binding of the four Cu(I) takes place in the β-domain protecting toward CID dissociation. Visible ions ranging from b4 to b6 appeared, indicating that the first cysteine residue, Cys6, weakly interacts with Cu(I) ions. For Zn 7 MT3 red 5+ , we also identified the b13 fragment, stressing that not whole βor α-domain binds three Zn(II) ions, and these are redistributed in both αand β-domains ( Figure 5A).

■ CONCLUSIONS
Metallothionein-3 (MT3) is a major player in the regulation of copper and zinc levels in the central nervous system. 52,53 Unlike other metallothionein isoforms, MT3 is specifically expressed in brain tissue and has distinctive structural characteristics, such as a conserved T 5 CPCP 9 motif, a Glu23, a Gly24, and an acidic insert in the α-domain (E 55 AAEAE 60 ). 6,7,11 Examining metal-binding properties of engineered MT3 variants showed that the T 5 CPCP 9 motif and the acidic insert in the α-domain are critical in modulating the Cu(I)/Zn(II) exchange rate. 20 In vitro studies have revealed that MT3 has the ability to exchange Zn(II) with Cu(II) that is bound to cellular ligands, such as amyloid β and α-synuclein, which highlights its protective role. 21−25 An interesting observation is that MT3 can be extracted from mammalian brains in the form of an air-stable complex, Cu(I) 4 Zn-(II) 3−4 MT3. 18,19 A significant body of spectroscopic research has investigated the reaction between Zn 7 MT3 and Cu(II), which has been found to result in the binding and reduction of Cu(II) to Cu(I), dissociation of Zn(II), and formation of two intramolecular disulfides. [27][28][29]50 A detailed analysis of the reaction between Zn 7 MT3 with Cu(II) and Cu(I) using electronic absorption and low-temperature luminescence has revealed that the presence of O 2 does not affect the nature of the products formed and that direct binding of Cu(I) to Zn 7 MT3 does not result in disulfide formation. 28 Despite the extensive efforts made, it is important to note that the spectroscopic signal obtained represents an average response from the dynamic equilibria between multiple Cu(I)/Zn(II) MT3 species. To gain more precise information about the isolated Cu(I) 4 Zn(II) 3−4 MT3 ox complex, this research utilized high-resolution ion-mobility mass spectrometry and various mass spectrometry-based approaches.
Our study utilized native mass spectrometry (MS) to observe the reaction between Zn 7 MT3 and Cu(II) and uncover the resulting interactions. Through fitting isotopically simulated distributions, we were able to assign molecular formulas to each metal-protein complex. Our results show that the β-domain can accommodate up to six Cu(I) ions and form a Cu(I) 6 Zn(II) 4 MT3 ox complex. Additionally, we observed the formation of the Cu(I) 4 Zn(II) 4 MT3 ox complex as well as partially metal-loaded species, including Cu(I) 4 MT3 ox and Cu(I) 4 Zn(II) 1−4 MT3 ox . Our findings indicate that these species are not a result of gas-phase dissociation but rather the reaction mechanism proceeds through the disassembly of Zn(II) clusters, followed by reassembly into multiple metal-MT3 complexes. The presence of two disulfides in the βdomain of the Cu(I) 4 Zn(II) 4 MT3 ox complex was confirmed (A) and for apoMT3 red 5+ and Cu 4 Zn 4 MT3 ox 5+ (B). Fitting of the yfragment ion data to theoretical isotopic distributions (C). Simulations of theoretical isotopic patterns were plotted as stem plots. Identified fragment ions (20 ppm) are colored according to bions (red) or y-ions (blue), while the experimental mass spectrum is shown in gray. The asterisk denotes a water loss. All of the fragment ions matched can be found in Tables S3−S5. "red" and "ox" subscripts refer to reduced and oxidized (two intramolecular disulfides) MT3 proteins. through mass spectrometry analysis under both native and denaturing conditions and by utilizing the reducing agent TCEP.
We then investigated the thermodynamics of dissociation of Zn(II) ions from the Cu(I) 4 Zn(II) 4 MT3 ox complex by gradually increasing the addition of EDTA and through acidification. The metal chelation mechanism occurs in a stepwise sequential manner with EDTA binding to one metal ion at a time, Conversely, acidification demonstrates high cooperativity, leading to the dissociation of all four Zn(II) ions from the Cu(I) 4 Zn(II) 4 MT3 ox complex, forming the Cu-(I) 4 MT3 ox intermediate. The cooperativity arises from the enhancement of proton acceptance by neighboring residues after protonation of one or several cysteine residues, leading to a change in protein conformation. Our findings indicate that, at pH ≈ 4, the Cu(I) 4 Zn(II) 4 MT3 ox complex undergoes a series of protonation events, resulting in a stable Cu(I) 4 MT3 ox complex state. Such pH corresponds to pK a ′ values of the thiols in high-affinity zinc sites in numerous proteins. 54,55 This implies that MT3 can serve as an effective copper scavenger under lower pH conditions, as found in certain cellular compartments or physiological states. 56,57 For instance, in vitro studies show that MTs are less susceptible to lysosomal proteolysis when loaded with metals. 58 The Cu(I) 4 -thiolate cluster is greatly stabilized, possibly due to the disulfides that constrain the Cu(I) ions within the protein. In addition, our metal chelation experiments suggest that upon exposure to cellular apoproteins, the heterogenuous Cu(I) 4 Zn(II) 4 MT3 ox species may exchange Zn(II) ions with zinc proteins. 59 Next, we used native ion mobility-mass spectrometry (IM-MS) to study the conformational characteristics of the Cu(I) x Zn(II) y MT3 ox complexes observed when Zn 7 MT3 red reacts with Cu(II). Our results show that the exchange of three Zn(II) for four Cu(II) in Zn 7 MT3 red with subsequent disulfide formation resulting in Cu(I) 4 Zn(II) 4 MT3 ox species is accompanied by a reduction in size. This can be attributed to the stronger Cu(I)-thiolate bonds and their more covalent character than in the case of Zn(II)-thiolate as well as the presence of two disulfides in the Cu(I) 4 -thiolate cluster, which both reduce the size and restrict the flexibility of the cluster, resulting in a narrower distribution of traveling wave (TW)derived collision cross section ( TW CCS N2 ) values. Additionally, to gain insight into the stability and dynamics of the formed Cu(I) x Zn(II) y MT3 ox complexes, we carried out collisioninduced unfolding (CIU) experiments. Our results indicate that the Cu(I) 4 Zn(II) 4 MT3 ox exhibits a more compact and stable conformation compared to the Zn 7 MT3 red .
In order to further investigate the stability of the Cu(I) 4 Zn(II) 4 MT3 ox complex, we performed collision-induced dissociation (CID) experiments and electron transfer dissociation (ETD). The collisional activation of both Zn 7 MT3 red 5+ and Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions resulted in the dissociation of four Zn(II) ions without the formation of intermediates. The relative stability of the metal-protein complexes was estimated by calculating survival yield plots, which showed that Cu(I) 4 Zn(II) 4 MT3 ox 5+ was more stable than the Zn 7 MT3 red 5+ ions. The dissociation of Zn(II) ions is attributed to both the nature of the metal ion and the thermodynamics of the protein domain. Isothermal titration calorimetry studies have shown that Zn(II) binding to MT3 is not enthalpically favored and is entropically driven, and the four metal ions dissociate from the α-domain due to a lower protein conformational entropic penalty than in the β-domain. 40 The dissociation of Zn(II) ions from Zn 7 MT3 red 5+ and Cu(I) 4 Zn(II) 4 MT3 ox 5+ ions was followed by covalent bond fragmentation. Experiments on metal-free apoMT3 red 5+ showed that the structural changes were due to metal binding rather than the protein fold. The results indicated that while Zn(II) ions are partially distributed in both the αand β-domains, Cu(I) ions are bound exclusively to the β-domain. The analysis of b-type fragment ions supports the conclusion that the binding of Cu(I) ions in the β-domain protects against CID dissociation. On the other hand, the results for Zn 7 MT3 showed that not all of the βdomain binds Zn(II) ions, and these ions are redistributed in both αand β-domains.
Thorough using high-resolution mass spectrometry approaches, our study sheds light on the unique function of MT3 among other isoforms by demonstrating the binding of four Zn(II) ions in the α-domain and four Cu(I) ions in the βdomain in the full-length MT3, which not only increases the protein's metal-binding capacity but also allows it to serve as a Zn(II) delivery system to maintain essential processes while serving as a scavenger for free or protein-bound Cu(II) ions. In the presence of free Cu(II) or Cu(II) donors, Zn(II) ions of Zn 7 MT3 may be exchanged by this metal, protecting the cell from its toxic effects through Cu(II) reduction to Cu(I) and partial β-domain oxidation. Moreover, the stability of this complex at a pH lower than that of cytosolic makes MT3 an ideal acceptor for copper from degraded copper proteins in lysosomes, enabling it to act as a scavenger in cellular compartments with lower pH or in certain pathological states. Despite Zn(II) dissociating easily, Cu(I) remains strongly bound to the protein, providing protection against oxidative damage. Overall, our findings provide a new understanding of the structure, stability, and dynamics of Cu(I)/Zn(II)-MT3, underscoring its crucial role in regulating zinc and copper levels in the human body.