The glutathione/metallothionein system challenges the design of efficient O₂-activating Cu-complexes

Abstract

Copper-complexes are of medicinal and biological interest, including as anticancer drugs designed to cleave intracellular biomolecules via O₂ activation. To exhibit such activity, the copper complex must be redox-active and resistant to dissociation. Metallothioneins (MTs) and glutathione (GSH) are abundant in the cytosol and nucleus. Because they also are thiol-rich reducing molecules with high Cu^I-affinity, they are potential competitors for copper ion bound in a copper drug. Here, we investigated a panel of Cu^I/ Cu^II-complexes often used as drugs, from various classes, with diverse coordination chemistries and redox potentials. We evaluated their catalytic activity in ascorbate oxidation based on redox-cycling between Cu^I and Cu^II, as well as their resistance to dissociation or inactivation under cytosolically-relevant concentrations of GSH and MT. Our findings show that O₂-activating Cu^I/ Cu^II-complexes for cytosolic/nuclear targets are generally not stable against the GSH/MT system, which creates a challenge for their future design.

Introduction

Copper is an essential element for most living organisms and plays a fundamental role in many biological processes. Its chemical activity as key enzymatic cofactor is central in redox-active copper-enzymes, like Cu/Zn-SOD or cytochrome c oxidase. To provide copper to these enzymes, a system of transmembrane transporters and chaperons is available. Intracellular cytosolic chaperons transport copper in its reduced Cu^I state.^[1,2] Copper is a powerful activator of O₂, which can lead to potentially deleterious reactive oxygen species (ROS).^[3] For this reason, its redox activity is tightly controlled, and normally restricted only to enzymes, while avoided during its transport. Hence, copper is classically tightly bound to transport proteins as loosely bound copper is much more prone to catalyze O₂-activation and ROS production.^[1]

Inorganic copper-complexes have been investigated for different applications, like artificial nucleases or proteases, and as anticancer, antimicrobial, antiviral and anti-inflammatory agents. ^[4,5] An often-supposed advantage of using bioactive copper-complexes is that fewer side-effects are expected, since copper is an essential element. For an intracellular activity, two cases are possible, either an exogenous copper-complex is applied and enters the cell, or an exogenous ligand is applied, which enters the cell and then binds endogenous copper. A very active field is the use of copper-complexes as artificial DNases in cancer research, since the seminal discovery that copper-phenanthroline complexes could induce O₂-dependent catalytic cleavage of DNA and RNA by attacking the sugar groups.^[6]

Since then, several other classes of ligands, among which the most prominent are (bis)-thiosemicarbazones, dithiocarbamates, cyclic amines etc, have been studied.^[4,7,8] The only relevant ligand in clinical use is bleomycin.^[9] Although mostly considered as an iron-based DNA cleavage agent, the involvement of copper-bleomycin in oxidative DNA cleavage has also been proposed. In most of these cases, the activity of the copper-complexes is based on the redox-cycling between Cu^I and Cu^II and activation of O₂ to form ROS. However, there is also interest in non-redox active copper-chelators in the cytosol, such as Cu^I-chelators for copper sensing or for copper sequestration (e.g. bathocuproines).^[10,11] A key aspect to consider is that copper-complexes with targets in the cytosol and nucleus have to compete with strong and abundant intracellular Cu^I-chelators such as glutathione (GSH) and metallothioneins (MTs), which could potentially impact their kinetic and thermodynamic stability.^[12] MTs are small (61–68 aa), cysteine-rich proteins which strongly bind Cu^I in thiolate clusters (reported logK of 19–21).^[13] Although they mostly bind Zn(II) under physiological conditions, some of them are well established Cu^I-chelators.^[14] MTs can form Cu^I or Cu^I₆- thiolate clusters in each of its two independent domains (N-terminal β-domain and C-terminal α-domain) depending on the Cu^I load, with the β-domain Cu^I₄-cluster formed first and possessing a higher Cu^I affinity. Interestingly, this Cu^I-thiolate cluster is stable in air and redox-silent.^[15] MTs are strong reducing agents, and hence able to reduce Cu^II to Cu^I and then bind it in the Cu^I-thiolate cluster with concomitant formation of intramolecular disulfide bonds.^[15,16] Another important compound is GSH, with concentrations ranging from 1–10 mM.^[17] GSH can bind Cu^I quite strongly (logK ~ 17) forming multinuclear clusters Cu^I_x-GSH_y, but under normal physiological conditions, a steady-state complex with Cu^I is unlikely.^[18] However, it can have an important impact as a Cu^II reducing agent.^[19] Indeed, our previous studies revealed that the GSH/MT system can efficiently abstract copper from Cu(II)-amyloid-β and Cu(II)-thiosemicarbazone complexes via Cu^II reduction and Cu^I-transfer to MT.^[13,20]

Ascorbate (AscH⁻) is another important intracellular reducing agent (in the cytosol and nucleus) with reported concentrations from 0.1 up to 5 mM.^[21,22] Noteworthy, it is one of the most competent endogenous reducing agents towards Cu^II and thus one of the most likely in performing the copper-based O₂ activation and ROS production.^[23,24] Consequently, it is the most used for in vitro studies of oxidative cleavage of biomolecules by copper-complexes (Figure 1).^[8]

Figure 1.

Schematic representation of the mechanism of copper-catalyzed ROS production with O₂ and AscH⁻. Copper undergoes redox cycling between Cu^II/Cu^I redox-states, through AscH⁻ oxidation to AscH^•−. O₂ is reduced by Cu^I to O₂^•−, H₂O₂ and HO^• via one electron events.

Based on the strong Cu^II-reducing and Cu^I-binding capabilities of the GSH/MT system, in our contribution, a panel of different copper-complexes (see Figure S1), representatives from different classes (bipyridines, (bis)-thiosemicarbazones, dithiocarbamates, tetraazamacrocycle, hydroxyquinoline, bathocuproine-based ligands), with different coordination chemistries and redox potentials, have been selected in order to evaluate their stability in cytosolic and nuclear relevant GSH and MT concentrations. Moreover, we correlated that to their catalytic activity in O₂ activation with AscH⁻ as reductant. Our findings show major challenges in developing/designing a redox active copper-complex for O₂ activation which is stable against the GSH/MT system.

Results and Discussion

First, the Cu^I/Cu^II catalytic redox-activity of all selected copper-complexes was determined, with AscH⁻ as reducing agent and O₂ as oxidant, since most of the redox-active copper-complexes are supposed to work via O₂ activation and hence production of ROS. The correlation between AscH⁻ oxidation and ROS production has been established in the past.^[23] Measurements were performed by monitoring the consumption of the substrate AscH⁻ at λ_max = 265 nm (ε = 14,500 M⁻¹ cm⁻¹).^[25] AscH⁻ oxidation was triggered with the preformed copper-complexes after 10 min, as indicated by the black arrow (Figure 2). For comparison, non-complexed copper in buffer was measured. Calculated initial molar AscH⁻ oxidation rates, i.e. r_obs (μM/min), are reported in Table 1.

Figure 2.

Time course of AscH⁻ oxidation monitored by absorbance spectroscopy at λ_max= 265 nm (representative data sets). The reaction of AscH⁻ oxidation was triggered with the addition of preformed Cu^II/Cu^I-complexes after 10 min (black arrow). Experimental conditions: preformed copper-complexes CuL_1/2 5 μM concentration, AscH⁻ 100 μM, in HEPES 50 mM, pH 7.4.

Table 1.

Table summarizing the i) t_1/2 (min or sec) values of Cu release from the selected Cu^II-complexes, calculated from the experimental kinetics of disappearance of the Cu^II or Cu^I CT bands, monitored by absorbance spectroscopy; ii) corresponding molar AscH⁻ oxidation rates (μM/min), and ii) redox potentials (mV).

Cu-complex	t1/2 transfer to Zn7MT-1 (with GSH)	robs AscH− oxidation (μM min−1)	Redox Potential (V) [NHE]
Background	-	0.11 ± 0.06	-
CuII	< 30 sec	9.5 ± 1.4	0.16 [28]
CuII-atsm	no transfer	0.05 ± 0.01	− 0.40 [29]
CuII-cyclam	no transfer	0.05 ± 0.02	− 0.74 (Epc) [30]
CuII-bleomycin	no transfer	0.06 ± 0.02	− 0.32 [31]
CuII-(CQ)2	< 30 sec	0.16 ± 0.04	− 0.36 [32]
CuII-(APDTC)2	~ 20 min	0.10 ± 0.06	− 0.14 [33]
CuII-gtsm	~ 50 min	0.49 ± 0.10	− 0.24 [29]
CuII-Dp44mT	~ 4 min	0.92 ± 0.11	− 0.21 [34]
CuII-(5,5’-DmBipy)2	< 30 sec	10.1 ± 1.0	0.12 [35]
CuII-(Phen)2	< 30 sec	12.4 ± 1.7	0.17 [35]
CuI-(BCS)2	< 30 sec	0.07 ± 0.01	0.62 [36]

Based on the rates of AscH⁻ oxidation by O₂, catalyzed by the copper-complexes, these were ranked in three arbitrary groups for ease of discussion (Figure 2 and S2), i.e., high efficiency copper-complexes (group 1: r_obs > 9 μM/min), low efficiency copper-complexes (group 2: 0.48 μM/min < r_obs < 1.02 μM/min) and inefficient copper-complexes (group 3: < 0.20 μM/min). Group 1 includes Cu^II-(Phen)₂, Cu^II-(5,5’-DmBipy)₂, which are very active in catalysing AscH⁻ oxidation, with values of oxidation rates of about the same order of copper in buffer. Cu^II-Dp44mT and Cu^II-gtsm are part of group 2, being able to oxidize AscH⁻ but slowly. Finally, Cu^I-(BCS)₂, Cu^II-(CQ)₂, Cu^II-(APDTC)₂ Cu^II-cyclam, Cu^II-atsm and Cu^II-bleomycin are included in group 3, with an order of magnitude lower initial rates compared to group 1, almost indistinguishable from the oxidation background reported in Table 1.

The redox-activity of the copper-complexes generally correlate with their respective redox-potentials, although these potentials were mostly obtained in organic solvents and in the absence of substrates. However, it is not known if these copper-complexes react via an inner- or outer-sphere mechanism, so no causal and quantitative relationship between AscH⁻ oxidation and redox potential can be made. Cu^II-(Phen)₂ and Cu^II-(5,5’-DmBipy)₂ are thermodynamically more favourable for Cu^II reduction to Cu^I (i.e. ~ 0.2 V vs NHE) in the presence of AscH⁻ (E° AscH⁻/AscH^•- = 0.28 V), compared to those of Cu^II-gtsm and Cu^II-Dp44mT (i.e. ~ - 0.2 V vs NHE). On the other hand, the very low catalytic redox-activity of the copper-complexes of group 3 relates to their either too negative (Cu^II-(CQ)₂, Cu^II-cyclam, Cu^II-bleomycin) (i.e. < - 0.3 V vs NHE), or too positive (Cu(I)-(BCS)₂) (i.e. > 0.6 V vs NHE) redox-potentials, which are not accessible for AscH⁻/AscH^•- redox-couple. Only for the inefficient Cu^II-(APDTC)₂ complex the reported redox potential in acetone was in the range of the low efficiency complexes (see Table 1).

The results were also confirmed changing the experimental protocol, i.e. triggering AscH⁻ oxidation adding either the ligand alone and then Cu^II or the opposite (Figure S3). To better control the timing of the different additions, the kinetics were slow down by lowering the concentrations of Cu^II and thus of the in-situ formed copper-complex (i.e., 1 μM Cu^II). These measurements confirmed what was previously observed with the preformed copper-complexes. The only differences were that group 1 complexes (i.e., Cu^II-(Phen)₂ and Cu^II-(5,5’-Dmbipy)₂) oxidized AscH⁻ much faster than copper in buffer and that Cyclam needed time to arrest AscH⁻ oxidation, in line with the slow kinetic of complexation of Cu^II.^[26]

Next, the stability of the copper-complexes in the presence of cytosolic and/or nuclear relevant concentrations of GSH and Zn-MT (i.e., 3 mM and 2.5 μM respectively) was investigated (corresponding to 4:1 mol/mol CuL_1/2:MT). The reactivity of the copper-complexes with GSH/Zn^II-MT was followed by absorbance spectroscopy, in order to estimate the t_1/2 of copper transfer to MT-1/−2. (Figure S4) For the reactive copper-complexes GSH alone was also tested. The transfer was then confirmed by luminescence spectroscopy at 77 K (results reported in Figure S5 for t = 4h and Figure S6 for t = 0h), as Cu^I binding to MT gives rise to a characteristic low-temperature luminescence spectrum due to the formation of its Cu^I₄-thiolate cluster which is characterized by two emissive bands, one at higher energy with cluster center (CC) origin at 425 nm (τ ~ 40 μs), and one at low energy mainly with charge transfer (CT) character with maximum between 560–595 nm (τ ~ 130 μs). The difference between the Cu^I₄-MT and Cu^I_x-GSH^y emission envelopes, relative intensity of the two emissive bands, and corresponding emission lifetimes under the condition used allowed the assignment of the predominant final complexes formed. Indeed, while mainly Cu^I₄-MT was formed after 4 hours incubation, depending on the copper-complex, contributions of other Cu^I₄-MT dependent complexes (such as ternary complexes of Cu^I₄-MT with GSH and/or ligands) were observed upon mixing.. ^[16,18]

Concerning the stability of the studied copper-complexes against copper-transfer to cytosolic relevant concentrations of GSH and Zn₇-MT, the following observations can be made: i) Cu^I-(BCS)₂ complex dissociates within mixing-time and Cu^I binds to MT. BCS and derivatives are some of the strongest Cu^I chelators used in biology. However, at 10 μM concentration, the complex is thermodynamically not stable enough to resist to Cu^I transfer to MT. A higher concentration of hundreds of μM BCS would be needed to compete for Cu^I with MT. However, such high concentrations are difficult to reach and less relevant from a drug point of view. ii) Cu^II-(Phen)₂ and Cu^II-(5,5’-DmBipy)₂ also dissociate within mixing-time. This is in line with their very fast reduction to Cu^I-(Phen)₂ and Cu^I-(5,5’-DmBipy)₂ and with the lower thermodynamic stability of the Cu(I)-complexes compared to Cu^I-MT complex (logβ₂ [Cu^I-(Phen)₂] = 15.8).^[27] Thus, Cu^II-(Phen)₂ and Cu^II-(5,5’-DmBipy)₂ are rapidly reduced and Cu^I is immediately transferred to MT. iii) Dissociation via reduction of Cu(II)-Dp44mT, Cu^II-gtsm and Cu(II)-(APDTC)₂ is slower compared to Cu^II-(Phen)₂ and Cu^II-(5,5’-DmBipy)_2, with t_1/2 of ~ 4, 50 and 20 min, respectively (Table 1). This is in line with the more negative redox potentials, thus with their slower reduction. iv) Cu^II-atsm, Cu^II-cyclam and Cu^II-bleomycin do not dissociate within 4h/5h and Cu is not transferred to MT. The stability of the complexes against GSH/Zn₇MT correlates with the very low negative redox potential of the complexes and hence with the fact that they are hardly reduced. v) Cu^II-(CQ)₂ was very inefficient in AscH⁻ oxidation but nevertheless Cu^I was transferred to MT, with t_1/2 of about < 30 sec. This can be explained based on the more labile 2:1 complex, i.e. one of the ligands is easier dissociated and reduction and dissociation is then favored.

Thus, we have evaluated the efficiency of various bioactive copper-complexes in catalyzing O₂ activation and their reactivity with cytosolic/nuclear relevant GSH/MT concentrations. The results are summarized in Table 1.

MTs are very strong Cu^I-chelators (logK of 19–21).^[16] Based on these values, a copper-complex must have a higher Cu^I-affinity than Cu^I₄-MT complex to be able to resist to copper abstraction by MT. To note, MTs can form a redox-inert Cu^I₄S_5–7 cluster in their N-terminal β-domain by either direct Cu^I binding or Cu^II to Cu^I reduction and concomitant formation of two disulfide bonds.^[15] Under the condition used, with an excess of GSH, we expect the formed Cu^I₄-cluster to be present in a fully reduced MT. Thus, in a cellular context even bathocuproine based Cu(I)-complexes, such as Cu^I-(BCS)₂, are not stable at relatively high concentration of 10 μM. All the other tested copper-complexes have a lower Cu(I)-affinity. Only Cu^II-(5,5’-DmBipy)₂ and Cu^II-(Phen)₂ have roughly similar thermodynamic stability constants for both Cu^I and Cu^II) redox-states (logβ₂ ~ 13–16).^[27] This implies that as soon as they are in the reduced Cu^I state, they are not thermodynamically stable against MT. Overall, this means that the reduction of Cu^II to Cu^I becomes the determining step towards copper-transfer to MT. Cu^II-(5,5’-DmBipy) ₂ and Cu^II-(Phen)₂ are very rapidly reduced by ascorbate, GSH or MT, and copper is fast transferred to MT. In contrast, Cu^II-atsm, -cyclam, and bleomycin are very inert against reduction, so they can resist and exist as Cu^II-complexes in the presence of GSH/MT.

However, the fact that a copper-complex is inert to reduction indicates that it cannot redox cycle rapidly between Cu^I/II redox states. This is a prerequisite for being an efficient O₂ activator and hence a ROS or oxidative cleavage catalyst. This connection can be illustrated by the catalytic activity of the copper-complexes in AscH⁻ oxidation. The most GSH/MT-resistant copper-complexes are also the least active, in agreement with the fact that Cu^II is very difficult to reduce. This correlates also with the low reduction potential reported for these complexes. On the other hand, the most catalytic active complexes like Cu^II-(Phen)₂ and Cu^II-(5,5’-DmBipy)₂ are also the ones which are very rapidly deactivated by Cu^I-transfer to MT.

Overall, these results indicate that in order to have a thermodynamically stable copper-complex against MT in a cytosolic/nuclear environment, it has either not to be reduced by the GSH/MT system or to have a very strong Cu^I-affinity with a logK of at least 20 (at pH ~ 7). Only rare examples are known, i.e. tetrathiomolybdate or phosphine sulfide-stabilized phosphine ligands.^[37,38] In both cases, they result in redox-inactive copper-complexes.

Here, the very different coordination chemistry preferred by copper in the two redox states, Cu^I and Cu^II), becomes important. With four coordination bonds, Cu^II prefers a square planar geometry, whereas Cu^I prefers a tetrahedral one. So, to stabilize Cu^II-complexes against reduction, a rigid square planar coordination is suited, as in the case of Cu^II-atsm or Cu^II-cyclam complexes. However, then the ligands are not adapted to bind Cu^I, and thus have a relatively low Cu^I-affinity compared to a strong competitor as MT. Hence, also in case of these type of ligands, as soon as Cu^II is reduced, CuI would be immediately lost into MT. As well as, a strong Cu^I-complex, like BCS, requires that a rigid tetrahedral geometry is imposed. As a consequence, the Cu^I-complex is very difficult to oxidize.

Hence an efficient redox-active and MT-resistant copper-complex would require a very high affinity for both Cu^I and Cu^II, with a logK higher than 20 for both redox states. For Cu^I, a logK higher than 20 is needed to resist to MT. On the other hand, a high activity in AscH⁻ oxidation, as observed in Cu^II-(5,5’-DmBipy)₂ and Cu^II-(Phen)₂, requires a roughly similar logK of Cu^I and Cu^II. This poses a huge design problem for a ligand.

So, one can ask the question how nature does it? To our best knowledge the only redox active copper-enzyme in the cytosol of mammals is Cu/Zn-SOD, catalyzing the superoxide-dismutation involving Cu^I/II) redox cycling. The reason is likely kinetics. Copper is buried into the active site of the enzyme and only a narrow channel lets superoxide approach copper. Moreover, during the maturation of SOD, the chaperon CCS delivers a copper ion into the active site and catalyzes the formation of an intra-subunit disulfide bond and the dimerization of two subunits (active enzyme).^[39] The reported Cu^I logK of SOD are around 15–16 ^[40,41], and hence are weaker than MTs. But the well folded enzyme keeps the copper inaccessible for MT or GSH. Such a buried and kinetically inert copper-environment is not easy to obtain with a much smaller inorganic complex.

Conclusion

In conclusion, the GSH/MT system, occurring in the cytosol and nucleus, is a powerful copper-abstractor. The high affinity of MT towards Cu^I, together with the high reducing power of GSH and MT is a threat for copper-complexes. It seems extremely difficult to obtain a highly redox-active copper-complex that resists to GSH/MT either thermodynamically or kinetically. This concerns all the copper-complexes that have a biomolecular target (like DNA, RNA or a protein) in the cytosol or nucleus and rely on a redox mechanism. But also, for ligands which are only supposed to bind copper in the cytosol/nucleus (like sensors, drugs etc.) GSH/MT is a severe competitor. It is of importance to be aware that MTs are predominantly present in cytosol and nucleus and indeed, very recently disulfiram was proposed to exert its anti-cancer activity as a copper-complex targeting a protein from the endoplasmic reticulum.^[42] Hence, the powerful GSH/MT system should be taken into account for all kind of complexes or ligands, such as drugs and sensors, that are based on copper-redox activity or are Cu^I-ligands in the cytosol/nucleus.

Ligands acronyms: Dp44mT (di-2-pyridylketone-4,4,-dimethyl-3-thiosemicarbazone), CQ (clioquinol), APDTC (ammonium pyrrolidinedithiocarbamate), BCS (bathocuproinedisulfonic acid disodium salt), gtsm (glyoxal-bis(N⁴-methyl-3-thiosemicarbazone), atsm (diacetylbis(4-methyl-3-thiosemicarbazone), cyclam (1,4,8,11-tetraazacyclotetradecane), 5,5^’-DmBipy (5,5’-dimethyl-2,2’-dipyridyl), Phen (1,10-phenantroline).