Different Antioxidant Efficacy of Two MnII-Containing Superoxide Anion Scavengers on Hypoxia/Reoxygenation-Exposed Cardiac Muscle Cells

Oxidative stress due to excess superoxide anion (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\bf{O}}}_{{\bf{2}}}^{{\boldsymbol{\cdot }}{\boldsymbol{-}}}$$\end{document}O2⋅−) produced by dysfunctional mitochondria is a key pathogenic event of aging and ischemia-reperfusion diseases. Here, a new \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\bf{O}}}_{{\bf{2}}}^{{\boldsymbol{\cdot }}{\boldsymbol{-}}}$$\end{document}O2⋅−-scavenging MnII complex with a new polyamino-polycarboxylate macrocycle (4,10-dimethyl-1,4,7,10-tetraazacyclododecane-1,7-diacetate) containing 2 quinoline units (MnQ2), designed to improve complex stability and cell permeability, was compared to parental MnII complex with methyls replacing quinolines (MnM2). MnQ2 was more stable than MnM2 (log K = 19.56(8) vs. 14.73(2) for the equilibrium Mn2+ + L2−, where L = Q2 and M2) due to the involvement of quinoline in metal binding and to the hydrophobic features of the ligand which improve metal desolvation upon complexation. As oxidative stress model, H9c2 rat cardiomyoblasts were subjected to hypoxia-reoxygenation. MnQ2 and MnM2 (10 μmol L−1) were added at reoxygenation for 1 or 2 h. The more lipophilic MnQ2 showed more rapid cell and mitochondrial penetration than MnM2. Both MnQ2 and MnM2 abated endogenous ROS and mitochondrial \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\bf{O}}}_{{\bf{2}}}^{{\boldsymbol{\cdot }}{\boldsymbol{-}}}$$\end{document}O2⋅−, decreased cell lipid peroxidation, reduced mitochondrial dysfunction, in terms of efficiency of the respiratory chain and preservation of membrane potential (Δψ) and permeability, decreased the activation of pro-apoptotic caspases 9 and 3, and increased cell viability. Of note, MnQ2 was more effective than MnM2 to exert cytoprotective anti-oxidant effects in the short term. Compounds with redox-inert ZnII replacing the functional MnII were ineffective. This study provides clues which further our understanding of the structure-activity relationships of MnII-chelates and suggests that MnII-polyamino-polycarboxylate macrocycles could be developed as new anti-oxidant drugs.

www.nature.com/scientificreports www.nature.com/scientificreports/ NMR, due to the paramagnetic nature of this ion. Both elemental analysis and mass spectrometry measurements accounted for the formation of Mn II and Zn II complex with metal to ligand 1:1 stoichiometry. The NMR and mass spectra are reported in the Supplementary Information 1 (Figs S1-S30).
Metal binding by H 2 Q2 and H 2 M2 and stability of their complexes. Both these organic scaffolds contain a tetra-amine macrocyclic ring and two appended carboxylic groups, making them potential chelating agents for both transition, alkali and alkaline metal ions. In addition, H 2 Q2 contains two quinoline moieties. Quinoline is also able to coordinate transition metals, including Mn II , although its binding ability is far lower than aliphatic amine groups. At the same time, this heterocycle shows poor affinity for alkali and alkaline earth metals, which result essentially not bound at least in aqueous solution 24 . On this ground, we first investigated Mn II coordination in aqueous solution by means of potentiometric titrations. This study requires preliminary determination of the protonation constants of the ligands. These values were in the range generally observed for polyamino-polycarboxylic acids 20 and are reported in Table 1, together with the complexes formed by H 2 Q2 in solution and their formation constants and those formed by H 2 M2, determined in a previous study 17 .
Analysis of the data in Table 1 reveals that Mn II gives stable 1:1 complexes in aqueous solutions with the fully deprotonated species of both ligands. In this respect, the MnQ2 complex is markedly more stable than MnM2, the stability constant of the MnQ2 complex being 4.83 log. units higher than that of MnM2 (log K = 19.56 (8) vs 14.73 (2) for the MnM2 and MnQ2 complexes, respectively: Table 1). In the latter, both the carboxylate groups are bound to the metal, a structural feature often found in Mn II complexes with polyamine macrocycles functionalized with acetate pendant arms 17,[25][26][27][28][29] . In fact, the amine function and anionic carboxylate groups normally    www.nature.com/scientificreports www.nature.com/scientificreports/ form stable complexes with transition metals, including Mn II , and, at the same time, binding to the metals of the N-CH 2 -COO − coordination bite affords a pentadentate chelate ring, which increases the stability of the complexes with first-tow transition metal cations. However, the higher stability observed for MnQ2 would suggest that, besides the carboxylate groups, in MnQ2 the quinoline nitrogen donors could be involved in metal coordination. On the other hand, the heterocyclic nitrogen possesses a lower binding ability than aliphatic amine and carboxylate groups, due to the electron-poor nature of the quinoline 24 . Attempts to obtain crystals of MnQ2 suitable for X-ray single-crystal diffraction analysis failed, thus precluding determination of the coordination sphere of the metal cation in the solid complex. To verify the possible involvement of quinoline in metal coordination we performed UV-vis, 1 H NMR, e.p.r. (in solution) and IR (in the solid state) measurements on the Mn II complex. UV-vis spectra of H 2 Q2 at pH 7.4 in the presence of increasing amounts of Mn II (Fig. 3a) clearly shows that Mn 2+ addition to a solution of H 2 Q2 induces an increase of the typical structured absorption band of quinoline between 310 and 320 nm. The absorbance monitored at λ 316 nm (Fig. 3b) linearly increases until a 1:1 metal-to-ligand molar ratio is reached, while the addition of more than 1 eq. of Mn II does not lead to any significant spectral changes. These data confirm the formation of a stable 1:1 complex in aqueous medium, and suggest that one or both quinoline nitrogens could be involved in metal binding (Fig. 3c). The 1 H NMR spectrum of H 2 Q2 at pH 2 shows a single set of signal in the aromatic region for the quinoline hydrogens (Fig. S15, Supplementary Information 1). In the aliphatic region, the signals are somewhat broaden, making more difficult their attribution. This feature is often found in tetraamine macrocycles, including H 2 M2 30 , and can be related to presence in solution of conformers, which differ in the conformation of the ethylenic chains, with interconversion times of the same order to that of the NMR scale. At pH 7.4, the aromatic region spectrum is composed by two different sub-spectra, which can be attributed to the presence of two different conformers, one of which www.nature.com/scientificreports www.nature.com/scientificreports/ present in very minor percentage, slowly exchanging on the NMR time scale, while the signals in the aliphatic region are broader. 1 H NMR signals of Mn II (S = 5/2) compounds are expected to give rise to signals experiencing large contact shift contributions and severe paramagnetic broadening, probably beyond detection. Accordingly, addition of Mn II to a solution of the ligand at pH 7.4 induces progressive line broadening upon increasing Mn II concentration in the spectra recorded at 400 Mhz (Fig. S31, Supplementary Information 1). This phenomenon includes both the aliphatic and the aromatic part of the spectrum, the signal broadening effects being somewhat more evident in the aliphatic region, and leads to the complete disappearance of the less intense subspectrum even in the presence of 0.4 eq. of Mn II . These results might be suggestive of the presence of dynamic equilibrium between the Mn II complex and the free ligand. To corroborate this suggestion, the spectra were also recorded at higher field (900 MHz), in order to slow down the exchange regime (Fig. S32, Supplementary Information 1). As expected, the observed line broadening during the titration with Mn II is reduced. In conclusion, the observed spectral changes in the presence of increasing amounts of Mn II support the presence of a complex where not only the tetraamine-dicarboxylate scaffold, but also the quinoline moieties interact with the metal ion.
NMR measurements were also used to analyze Zn II coordination (Fig. S33, Supplementary Information 1). Differently from Mn II , the addition of Zn II in solution only induced changes in the aliphatic section of the 1 H NMR spectrum. Despite the fluxional aspect of the aliphatic signals at pH 7.4, a slight, but not negligible, downfield shift of the CH 2 signals upon metal coordination could be observed. Conversely, the aromatic part of the spectrum is almost not affected by metal coordination. This titration could not be followed by 13 C NMR, due to poor solubility of the free ligand at neutral pH (see the Methods section). However, the 13 C NMR spectrum of ZnQ2 (Fig. S9, Supplementary Information 1) shows two different sets of nine signals for the quinoline carbons in the aromatic region, one of which with weaker signal intensities, in keeping with the presence of two conformers slowly exchanging on the NMR time scale, as already observed in the 1 H NMR spectra at pH 7.4. Interestingly, the signal of the carboxylate group occurs at 179.79 ppm, as normally found in Zn II -bound carboxylate groups 31 . The same signal is shifted at 168.24 ppm in the free ligand at acidic pH values (Fig. S6) and at 164.12 in the case of sodium acetate in water 32 . These results suggest that in the Zn II complex the metal is firmly bound by the carboxylate groups, while quinoline is not involved in the coordination. The different role of quinoline in Mn II and Zn II coordination reflects the ability of the former to expand its coordination environment achieving coordination number greater than six 26,27 . This hypothesis is confirmed by the EPR spectra recorded on the MnM2 and MnQ2 complexes in frozen solution (Fig. S34, Supplementary Information 1). Both spectra are characterized by a signal consisting of six sharp lines, centered at field corresponding at g = 2.00, superimposed on a broader envelope with partially resolved structure (Fig. S34, Supplementary Information 1). The six sharp lines are due to the 55 Mn-hyperfine splitting associated with the central m S = |−1/2〉 → |+1/2〉 transition, while the structures on the broader features are attributed to transitions between higher m S levels due to the moderate zero field splitting (ZFS) interactions, as expected for Mn II . However, the different resonant fields of the broader features for MnM2 and MnQ2 are indicative of different ZFS interactions in the two systems. In particular, the larger extension of the spectrum of MnQ2 lead us to suppose a ZFS larger than for MnM2. These differences in the ZFS interactions can be traced back to a different coordination environment for Mn II in the two complexes. The possible involvement of quinoline in Mn II coordination was also investigated by recording IR spectra on 4, H 2 Q2 and MnQ2 solid compounds. In particular we paid attention to a typical set of three signals, attributed to C-C and C-N stretching vibrations, easily recognizable in the IR spectra of quinoline compounds between 1530 and 1650 cm −1 32 . The same set of three signals is also observed in the case of the simple pyridine heterocycle and its shift toward higher wavelength number has been considered diagnostic for coordination of the nitrogen donor to transition metals, including Mn  . In the parent compound 4, not containing carboxylate units, these bands, indicated with a, b and c in Supplementary Information 1, Fig. S35a, are positioned at 1643, 1603 and 1541 cm −1 , respectively. In H 2 Q2 two bands attributable to the quinoline ring can be easily recognized at 1599 (b band) and 1539 cm −1 (c band), while the a signal originally at 1643 cm −1 is superimposed with the C=O stretching at 1626 cm −1 ( Supplementary Information 1, Fig. S35b). In the MnQ2 spectrum, two bands (a and c in Supplementary Information 1, Fig. S35c) of the quinoline moieties are positioned at higher wavelength numbers (1661 and 1554 cm −1 , respectively) than in H 2 Q2, while the third band b (at 1589 in H 2 Q2) is superimposed with the C=O stretching signal at 1602 cm −1 . The observed shift toward higher wavelength number can be indicative of the interaction of the heteroaromatic nitrogen(s) with Mn II .
All together, these results suggest that one or both quinoline moieties are involved in metal coordination (Fig. 3c). The observed spectroscopic changes, in particular in the NMR spectra, suggest a weaker interaction of the quinoline nitrogen(s) with respect to that of the amine groups of the macrocycle and the carboxylate groups, in keeping with the lower binding ability of quinoline for metal cations. However, this interaction would justify, at least in part, the higher stability observed for its Mn II complex with respect to MnM2. At the same time, the increased hydrophobic characteristics of H 2 Q2 can also contribute to complex stability. In fact, binding of the metal implies its envelopment in a hydrophobic pocket and hence its complete desolvation, with a stabilizing energetic effect due to the increased translational entropy.
Both the MnM2 and MnQ2 complexes bind acidic protons to give protonated species of the type [Mn(H 2 L)] 2+ and [Mn(HL)] + in aqueous solution (L = M2 or Q2, see Table 1). In MnM2 complex, both carboxylate groups are coordinated to the metal 17 and, similarly to most complexes of polyamine-polycarboxylate ligands with transition metals, protonation occurs at the carboxylate groups, leading to their detachment from the metal. In principle, in MnQ2, proton binding can occur at the anionic carboxylate functions or the quinoline nitrogens. However, the protonation constants of MnM2 and MnQ2 complexes are quite similar and, at the same time, methylen-carboxylate groups are more basic than quinolines. These observations suggest that protonation takes place on the carboxylate groups in MnQ2 as well.
Polyamine-polycaboxylate ligands can form complexes even with 'hard' metal cations, including K I , Na I , Ca II and Mg II , which are abundant in the cellular and extracellular environments. Therefore, we analyzed the (2019) 9:10320 | https://doi.org/10.1038/s41598-019-46476-2 www.nature.com/scientificreports www.nature.com/scientificreports/ coordination of these metals with H 2 Q2 by potentiometric titrations: the results are reported in Table 1, together with those previously found with H 2 M2 17 . The stability of the K I complexes with both ligands is too low to be confidently determined (log K < 2), while Na I , Ca II and Mg II formed markedly less stable complexes than Mn II . At a lesser extent than Mn II , the Na I , Ca II and Mg II complexes with H 2 Q2 are somewhat more stable than those with H 2 M2. Interestingly, the difference in stability between H 2 Q2 and H 2 M2 complexes with these metals is remarkably lower than that observed for Mn II , reflecting the poor affinity of quinoline toward alkali and alkaline earth metal cations. The observed higher stability of the H 2 Q2 complexes with Na I , Ca II and Mg II is likely due to a stronger desolvation of these metal cations upon binding to the hydrophobic scaffold rather than to the interaction with the heteroaromatic nitrogens of quinolines. As a result, H 2 Q2 is more selective for Mn II over Na I , Ca II and Mg II than H 2 M2. To assess whether Na I , Ca II and Mg II could compete with Mn II in the complexation process, we considered competitive systems containing each ligand and Mn II at very low concentrations (1 μM) in the presence of excess of Na I (165 mM), Ca II (2.5 mM) and Mg II (1.2 mM), roughly reproducing the cellular concentrations of these metals. We then calculated the overall percentages of the different metal cations complexed with H 2 Q2 over a wide pH range 36,37 , which are shown in the competition diagram in Fig. 4a. The diagram obtained for H 2 M2 is also reported (Fig. 4b). Mn II remained completely bound to the two scaffolds while Ca II , Mg II and Na I appeared unbound. This finding indicates that MnM2 and MnQ2 are substantially unaffected by other metal ion competitors in the medium and no detectable Mn II release occurs due to displacement by other metals. Mn II release occurs in low percentage (less than 10%) only below pH 4.8 and 6 in the case of H 2 Q2 and H 2 M2, respectively, due to extensive protonation of the ligands at acidic pH values. H 2 Q2 is more resistant to dissociation in acidic media than H 2 M2, a relevant characteristic if taking into account that some functional cell compartments, such as the mitochondrial outer chamber, have acidic features.
The stability of MnM2 and MnQ2 was also higher than other Mn II complexes with endogenous ligands present in the cellular environment, such as ATP (log K = 4.66 for the equilibrium Mn II + ATP = [Mn II (ATP)]), glutathione (log K = 2.7 for Mn II + GSH = [Mn II (GSH)]) and carboxylate anions (log K = 3.79, 1.68 and 0.92 for Mn II complexation by citrate, maleate and L-lactate anions) 24 , thus excluding substantial de-metallation of these scaffolds by biological chelators.
We also determined the stability of the complexes formed by H 2 M2 and H 2 Q2 with Zn II . Like Mn II , Zn II is a bivalent first-row transition metal cation sharing several chemical features with Mn II . However, Zn II ion is extremely stable to reduction, while oxidation states higher than II are not accessible. These characteristics make In keeping with these findings, the degree of cell lipid peroxidation, a by-product of excess ROS, assessed by the red-to-green fluorescence shift of BODIPY-581/591-C 11 lysochrome, was markedly enhanced upon H + R www.nature.com/scientificreports www.nature.com/scientificreports/ and significantly reduced by MnQ2 and MnM2 added at reoxygenation, the former compound being slightly more active than the latter one in the short term (1 h) (Fig. 6c).
In all the above experiments, ZnQ2 and ZnM2 added in the place of their Mn-containing counterparts had no antioxidant effects (Fig. 7). Representative confocal images and fluorescence-activated cell sorting (FACS) analysis of these experiments at the 1-h time point are shown in Fig. 10 and Supplementary Information 3, respectively.

Preservation of impaired cell mitochondrial activity by MnQ2 and MnM2. Both MnQ2 and
MnM2, added at reoxygenation (10 μmol/L), improved the assayed markers of mitochondrial activity which were compromised by H + R-induced oxidative stress, albeit with different timing. In detail, the efficiency of the respiratory chain, measured as fluorescence of reduced resazurin (RSZ) dye (Fig. 8a) and the mitochondrial membrane potential (Δψ), evaluated through the inlet of the fluorochrome tetramethylrhodamine methyl ester perchlorate (TMRM) (Fig. 8b), increased significantly in H9c2 cells treated with MnQ2 and MnM2 in comparison with the untreated controls. Conversely, opening of transition pores (mPTP), an index of mitochondrial dysfunction and early apoptosis measured through the extinction of calcein fluorescence, decreased significantly upon addition of MnQ2 and MnM2 (Fig. 8c). Similarly to what observed in the above experiments on oxidative stress, MnQ2 was more effective than MnM2 in preserving mitochondrial function in the short term (1 h), while at the longest time (2 h) the two compounds showed similar potency. Replacement of the redox-active compounds with ZnQ2 and ZnM2 resulted in loss of protection against H + R-induced mitochondrial dysfunction ( Protection from H + R-induced cell death by MnQ2 and MnM2. Since mitochondrial impairment triggers the intrinsic pathway of apoptosis operated by the caspase 9 -caspase 3 cascade 38 , we next investigated whether the protection afforded by MnQ2 and MnM2 on mitochondrial oxidative dysfunction reflected in improved survival of H9c2 cells subjected to H + R. Compared with the untreated controls, the cells added with MnQ2 and MnM2 at reoxygenation showed a decrease in pro-apoptotic activation of caspases 9 and 3, whereas caspase 8, involved in the extrinsic apoptotic pathway, was substantially unchanged (Fig. 9a-c). Consistent with these findings, evaluation of the percentage of dead cells in the cultures showed a significant rise in the untreated controls, which was blunted by MnQ2 and MnM2 (Fig. 9d). Again, MnQ2 was more effective than MnM2 to preserve H9c2 cell viability in the short term (1 h). ZnQ2 and ZnM2 substituted for the redox-active compounds showed no cytoprotective effects (Fig. 9). Representative confocal images and FACS analysis of these experiments at the 1-h time point are shown in Fig. 10 and Supplementary Information 3, respectively.
Modulation of MAPK activation by MnQ2 and MnM2. Among MAPKs, extracellular signal-regulated kinases (ERK) p38 and JNK are activated upon harmful stimuli, including oxidative stress 39 . In the present experiment, H + R induced a significant increase in p38 and JNK phosphorylation, an index of enzyme activation, while this phenomenon was reduced by 1-h incubation with MnQ2 and, at a lesser extent, MnM2 (Fig. 11). On the other hand, phosphorylation/activation of ERK1/2, which can operate cytoprotective and anti-apoptotic mechanisms in response to oxidative stress 39,40 , was impaired by H + R and up-regulated by 1-h incubation with MnQ2 and MnM2. In both instances, ZnQ2 and ZnM2 were ineffective. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
The present study indicates that the pharmacokinetic and pharmacological characteristics, in terms of intracellular permeation and consequent ⋅− O 2 -scavenging and anti-oxidant efficacy of the studied redox-active Mn II complexes can be improved by appropriate modulation of the lipophilic and structural characteristics of their  www.nature.com/scientificreports www.nature.com/scientificreports/ of different functional moieties on selected nitrogen atoms within a tetraamine scaffold. Of note, the strategy to increase lipophilicity by targeted chemical substitutions has also been adopted to improve the bioavailability of Mn-porphyrins, another class of redox-active drugs 16 . In both MnM2 and MnQ2 simultaneous binding of Mn II to the tetraamine macrocycle and to the 2 carboxylate appendices makes them thermodynamically stable and highly resistant to Mn II release. At physiological pH, Mn II forms a highly stable complex with the polyamine-polycarboxylate scaffold, markedly more stable than that with other metal cations present in the cellular environment, such as Ca II , Mg II and Na I . Moreover, the neutral charge of these compounds facilitates trans-membrane passage and intracellular localization. The presence of two quinoline or methyl groups appended to amine groups in 4, 10 position prevents the nitrogen atoms to interact with water molecules through hydrogen bonds. The insertion of two quinoline units not only increases the hydrophobic features of the resulting polyamine ligand, but also markedly enhances the stability of the MnQ2 complex. (log K = 19.56(8) and 14.73(2) for the equilibrium Mn 2+ + L 2− = [MnL], with L = Q2 and M2, respectively). In fact, in MnQ2 the heteroaromatic nitrogens of quinoline can also participate in Mn II coordination, reinforcing the overall metal-to-ligand interaction and increasing complex stability. Although the present results are not sufficient to discern whether one or both quinoline nitrogen are involved in Mn II coordination, the metal ion is coordinated not only by the tetraamine macrocycle and the carboxylate groups, as in the case of the MnM2 complex, but also by at least one quinoline nitrogen, in agreement with the ability of Mn II to achieve coordination number greater than 6. In the meantime, the two quinolines increase the lipophilic features of the complex and, overall, create a hydrophobic environment for the metal cation, leading to enhanced desolvation of the metal and increased translational entropy, further stabilizing the Mn II complex. Conversely, the scarce binding affinity of the quinoline nitrogen for alkali and www.nature.com/scientificreports www.nature.com/scientificreports/ alkaline-earths leads to a binding ability for these metal cations that was only slightly higher than that of the H 2 M2 ligand. As a result, H 2 Q2 shows a higher selectivity for Mn II than for other metal ions largely present in the cellular environment, such as K I , Ma I , Ca II and Mg II . These characteristics prevent both trans-metallation reactions, due to complexation of the ligand to other metals in the cell, and de-metallation due to Mn II complexation by cellular chelating agents. In fact, MnQ2 and MnM2 did not release Mn II even in the presence of large excess of Ca II and other metal ions, or broad variations of pH. Both Mn II complexes are stable in the alkaline region and at neutral pH. Metal release is observed at acidic pH values, where extensive ligand protonation compete with metal complexation. Mn II release occurs only below pH 6 for MnM2 and pH 5 for the more stable MnQ2 (in the latter case the release of Mn II is about 10% at pH 4.8 and 50% at pH 3.8). Both complexes are completely dissociated at more acidic pH (below pH 4 for MnM2 and pH 3 for MnQ2). This characteristic can be important in light of the fact that in active mitochondria the outer chamber has acidic features. Of note, the absence of negative charge and increased lipophilicity have been shown to facilitate accumulation of Mn complexes in mitochondria 15,22 . These reports and the present findings suggest that MnQ2 and, at a lesser extent, MnM2 can readily enter the cells and directly scavenge ⋅− O 2 at its mitochondrial generation sites. The molecular mechanism of ⋅− O 2 scavenging of MnQ2 and MnM2 likely consists in a catalytic cycle involving: (i) oxidation of Mn II to Mn III by ⋅− O 2 , (ii) reduction of Mn III complex by another ⋅− O 2 to form the initial Mn II compound 17,20 . To confirm these assumptions, the ⋅− O 2 -scavenging effects of MnQ2 and MnM2 were completely lost when redox-inert Zn II was substituted for Mn II . As shown by time-course inhibition of cytochrome c reduction, the ⋅− O 2 -scavenging ability of MnQ2 was lower than MnM2 (v max : 0.000460 vs. 0.000457 absorbance arbitrary units s −1 ) and k cat evaluation (2.5 vs. 4.5 × 10 6 M −1 s −1 ). This is likely due to the steric hindrance that quinolines www.nature.com/scientificreports www.nature.com/scientificreports/ impose towards the approach of ⋅− O 2 to the Mn center. On the other hand, in a cellular model, this limitation of MnQ2 appears to be well balanced by its increased lipophilicity which improves biodistribution.
There is a general trend of pharmaceutical research towards the tuning-up of existing medicinal ingredients, by targeted chemical modifications or proper formulations, in order to improve their pharmacokinetic properties, selectivity of action and overall efficacy. Concerning anti-oxidant drugs, their actual efficacy largely depends on Figure 11. FACS analysis of immunofluorescent expression of phosphorylated (p−) JNK (a) p38 (b) and ERK1/2 (c), evaluated at 1 h and expressed as percent changes of the untreated controls. The oxidative stressrelated MAPKs p-JNK and p-p38 were increased by H + R and significantly decreased by MnQ2 and MnM2. Conversely, the cytoprotective pERK was decreased by H + R and significantly increased by MnQ2 and MnM2, the former compound being significantly more effective than MnM2. ***p < 0.001 vs. controls; + p < 0.05, ++ p < 0.01 vs. H + R; (one-way ANOVA); # p < 0.05 vs. MnM2 (Student's t test).
www.nature.com/scientificreports www.nature.com/scientificreports/ the ability to reach the intracellular sites of ROS generation, mainly mitochondria, and keep them under the toxicity threshold, mimicking and reinforcing the natural anti-oxidant systems 41 . The present cell culture system is a good model to study and compare the anti-oxidant properties of MnQ2 and MnM2, since the mechanism of cellular damage induced by H + R involves primarily a dysregulated mitochondrial generation of ⋅− O 2 42,43 , similarly to aging [5][6][7] . Under the reported experimental conditions, the redox-active compounds MnQ2 and MnM2, added at reoxygenation at micromolar concentrations, effectively abated intracellular ROS and mitochondrial ⋅− O 2 generation, reduced mitochondrial dysfunction and activation of the caspases of the intrinsic apoptotic pathway, and increased cell viability. Of note, the more lipophilic the molecule, the faster its effects: MnQ2 was able to afford significant protection 1 h after administration to the cells. During prolonged incubation time (2 h), the anti-oxidant efficacy of both MnQ2 and MnM2 became similar, along with the progressive increase in MnM2 intracellular levels approaching those of MnQ2. In keeping with these findings, previous studies have shown that lipophilicity of different Mn-containing compounds can be a good predictor of their ability to enter the cells 44 .
The present findings on MAPK phosphorylation are consistent with activation of an intracellular response to oxidative stress. As expected on the basis of on their antioxidant properties, MnQ2 and MnM2 (1 h) were able to reduce p38 and JNK phosphorylation increased by H + R. Conversely, both compounds significantly increased phosphorylation of cytoprotective, anti-apoptotic ERK1/2. This suggests that MnQ2 and MnM2, besides ⋅− O 2 scavenging, could also activate this intrinsic cell defense mechanism. These findings fit well with a previous study showing similar effects of a Mn-porphyrin, belonging to a similar class of redox-active drugs, on mammary carcinoma cells subjected to radiation-induced oxidative stress 45 .
Modulation by phosphorylation of key cellular pathways involving transcription factors, such as Nrf2 and NF-kB, has recently emerged as a parallel mechanism of action of Mn-porphyrins, adjunctive to or even prevailing on canonical ROS scavenging. In particular, Mn-porphyrins have been shown to exhibit protective effects by upregulation of Nrf2 and inhibition of NF-kB and MAPK 15,16,45 . Nrf2 is known to upregulate endogenous antioxidant enzymes, including MnSOD, catalase, peroxiredoxins, glutathione peroxidase, etc., which in turn can reduce the levels of ⋅− O 2 and H 2 O 2 . On the other hand, inhibition of NF-kB results in suppression of NADPH oxidases and reduction of RS levels 15,16 . Whether MnQ2 and MnM2, similarly to Mn-porphyrins, may modulate Nrf2 and NF-kB to exert their antioxidant cytoprotective effects cannot be inferred by the present data. Of note, MnQ2 and MnM2 contain Mn II while Mn-porphyrins contain Mn III , and it cannot be ruled out that the above properties may be related, at least in part, to the oxidation state of the Mn center. However, this is a tempting working hypothesis for further studies aimed at clarifying the spectrum of biological effects of these new redox-active compounds.
In conclusion, this study indicates that the insertion of hydrophobic groups with metal-coordinating ability within a polyamine-polycarboxylate scaffold can improve its pharmacological properties. Both MnQ2 and MnM2 behave as efficient ⋅− O 2 scavengers and may represent a promising new class of redox-active drugs. Of note, in respect to MnM2, MnQ2 has a greater tendency to readily enter the cells owing to its enhanced lipophilic features, is less susceptible to transmetallation reactions and is more resistant to Mn II decomplexation at acidic pH values, as those occurring in the mitochondrial compartment. These characteristics make MnQ2 able to efficiently reduce oxidative cell injury mediated by mitochondrial generation of ⋅− O 2 , a key pathogenic event of ischemia-reperfusion damage 42,43 and aging [5][6][7] . These findings can improve the understanding of the structure-activity relationships of Mn II -polyamine-polycarboxylates as redox-active drugs. In perspective of possible drug development, the chemical features of MnQ2 and MnM2 suggest that these molecules can withstand inactivation by oxidative stress conditions, at variance with synthetic or extractive SOD 46 , whose pharmaceutical use is also limited by poor stability in water and intracellular penetration, short half-life and immunogenicity concerns 47,48 . Further investigations on organ and animal models of endogenous oxidative stress are required to validate the actual therapeutic potential of MnQ2 and MnM2. In this context, transplant medicine may represent a promising field: hypothetically, these compounds could be added to the incubation fluid of explanted organs to extend their viability and reduce ischemia-reperfusion injury upon re-implantation. Of note, suitability of Mn-based redox-active molecules as a promising new class of anti-oxidant drugs is indicated by the fact that the Mn-containing compounds BMX-001 and GC4419 (formerly M40403) are being tested in clinical trials to reduce the adverse effects of radiotherapy-induced oxidative stress in cancer patients 49 .

Methods
Reagents. MnM2 was synthesized in our laboratory from 4,10-dimethyl-1,4,7,10-tetraazacyclododecane-1,7-diacetate, as previously described 17 , on kind permission of the patent owner (General Project Ltd., Montespertoli, Italy). The aqueous solutions used in complex synthesis and in potentiometric measurements were deoxygenated by bubbling N 2 to prevent possible oxidation of Mn II . All new compounds were characterized by elemental analysis and mass spectroscopy. Elemental analyses (C, H, N content) were performed with Perkin-Elmer 2400 CHN elemental analyzer. C, H, N and Mn percentages were within ±0.2% of theoretical value, in keeping with a >95% purity for all compounds). Determination of the Mn content in the complexes was performed using a Varian 720-ES inductively coupled plasma atomic emission spectrometer (ICP-AES). The biological samples were treated as previously reported 17 to obtain suitable solutions for the ICP-AES determination of Mn II content. IR spectra were collected by a IRAffinity-1S Shimadzu instrument. UV-vis spectra were recorded on a Perkin-Elmer lambda 25 spectrometer. II complexes 4a-8a-bis(methylen-2-quilolyn)dodecahydro  -2a,4a,6a,8a-

1,7-bis(methylen-2-quinolyl)-1,4,7,10-tetraazacyclododecane three-hydrobromide salt (4).
1.28 g (2.33 mmol) of 3·2Cl were dissolved in 25 mL of N 2 H 4 and 5 mL of ethanol and the mixture was stirred at 110 °C for 6 hours. After cooling at room temperature, the solvent was evaporated under reduced pressure affording a yellow solid deposit, which was dissolved in NaOH 15 M aqueous solution (10 mL). The resulting solution was extracted with chloroform (4 × 25 mL). The organic layers were collected, dried with Na 2 SO 4 and the solvent was finally removed under vacuum. The crude product was dissolved in ethanol (20 mL). Then, 48% HBr (1 mL) was added dropwise to the resulting solution, affording the three-hydrobromide salt of 4 (4·3HBr·2H 2 O as a yellow solid. Yield 1.33 g (78%) Anal. calcd. for C 28 -2-quinolyl)-1,4,7,10-tetraazacyclododecane-1,7-diacetic acid (5). Synthesis of this compound was performed with a similar procedure to that reported previously 17 . A solution of chloroacetic acid (3.03 g, 32.1 mmol) in water (10 mL) was added to a solution of 4 (0.97 g, 2.14 mmol) in water (10 mL), adjusting pH to 9.5 by using a 5 M NaOH solution. Temperature was kept at 70 °C for 18 h. The pH of the solution was maintained at 9.5 with drops of 5 M NaOH. The solution was then dried by evaporation, the crude residue dissolved in water (20 mL), and the pH of the solution adjusted to 7 by addition of 5 M HCl. The solution was then passed through a cation-exchange column (Amberlite IR 120; acidic form, bed volume 60 cm 3 ), which was eluted with water (500 mL), then with 0.5 M NH 3 (600 mL) and finally with water (700 mL). Each fraction was vacuum-evaporated and analysed by 1 H NMR. The fractions containing the desired product were collected together and the resulting solution was vacuum evaporated, yielding the ligands as ammonium salt. This compound was dissolved in 10 mL water and passed through a column filled with anion-exchange resin (Amberlite IRA 900, alkaline form, bed volume 50 mL), eluted with water (350 mL), then with 5 M HCl (350 mL) and finally with 0.1 M HCl (300 mL). Each fraction was vacuum-evaporated and analyzed by 1 H NMR. The fractions containing the desired product were combined and dried under vacuum. The resulting yellow solid was recrystallized from ethanol/H 2 O mixture to afford compound 5 as anhydrous tetrahydrochloride salt (5·4HCl·2H 2 O). Yield 0.53 g (33%). Anal. calcd. for C 32     NMR experiments. All 1 H and 13 C spectra were recorded on a Bruker Avance III 400 MHz instrument. For H 2 Q2, the 1 H spectrum was collected at pH 2 and pH 7.4, while the 13 C spectrum was recorded only at pH 2. In fact, at pH 7.4 this ligand shows limited solubility in water, even at 100 °C, which precludes recording of the 13 C spectrum. The 1 H NMR spectra of H 2 Q 2 in the presence of increasing amounts of Mn II (as MnSO 4 salt) were collected at 400 MHz and at 900 MHz on a Bruker NEO 900-MHz equipped with 5 mm CP TCI 1 H/ 13 C/ 15 N z grd probe. In this experiment, a de-areated MnSO 4 solution (0.3 M) in D 2 O was added to a de-areated boiling solution of H 2 Q 2 in D 2 O (1 · 10 −2 M), phosphate-buffered at pH 7.2. After the addition, the solution was kept at 100 °C for 4 h before collecting the spectrum. The 1 H NMR spectra of H 2 Q 2 in the presence of increasing amounts of Zn II (as ZnCl 2 salt) were collected at 400 MHz. This experiment was performed in the same conditions described for Mn II complexation.
ESI-MS spectroscopy. ESI-MS spectra were recorded on a LTQ Orbitrap high-resolution mass spectrometer. In a typical experiment, stock solutions of the samples (10 −3 M) were prepared in methanol (compound 3) or water (compounds 4, 5, ZnQ2 and MnQ2). After a 50-fold dilution in MeOH, high resolution ESI mass spectra were recorded by direct introduction at 5 μL min −1 flow rate in an LTQ-orbitrap high-resolution mass spectrometer (Thermo, San Jose, CA, USA), equipped with a conventional ESI source. The working conditions were: spray voltage 5 kV, capillary voltage 39 V, capillary temperature 280 °C, tube lens voltage 130 V. The sheath and the auxiliary gases were set, respectively, at 10 and 5 (arbitrary units).
Potentiometric measurements. All pH measurements (pH = −log [H + ]) employed for the determination of ligand protonation and metal complex stability constants were carried out in 0.10 M NMe 4 Cl aqueous solution at 298.1 ± 0.1 K by conventional titration experiments under inert atmosphere. The equipment and procedure used were as previously described 50 . The standard potential E° and the ionic product of water (pK w = 13.83(1) at 298.1 ± 0.1 K in 0.10 M NMe 4 Cl) were determined by Gran's method 51 . At least three measurements (with about 100 data points for each) were performed for each system in the pH ranges 2-10.5. In all experiments, ligand concentration [L] was about 0.5 × 10 −3 M. In the complexation experiments the metal ion concentration was changed from 0.5:1 to 1.5:1. The computer program HYPERQUAD 37,52 was used to calculate both protonation and stability constants from potentiometric data. Distribution diagrams and competition plots were calculate by using the Hyss program 37 . Preliminary experiments were performed to assess the toxicity of MnQ2 and ZnQ2, which were added to H9c2 cell cultures (5 × 10 4 /well in 24-well plates) at a 10-fold higher concentration (100 μmol L −1 ) than that used for the experiments (10 μmol L −1 ) for 24 h. Under these conditions, cell viability assayed by the 3-(4,5-dim ethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) test was not impaired in comparison with the untreated cultures (data not shown). Lack of toxicity of MnM2 and ZnM2 were demonstrated previously with the same method 18 .

Cell permeation.
To investigate the kinetics of cell permeation of MnQ2 and MnM2, H9c2 cells were seeded in 24-well plates (5 × 10 4 cells/well) and allowed to adhere. The medium was replaced with phosphate-buffered saline, either alone (controls) or added with MnQ2 or MnM2 (10 μmol L −1 ), and the cells were incubated for 30 min, 1 h or 2 h. Then, the medium was removed and the cells were thoroughly washed with phosphate-buffered saline. Finally, the cells were detached in distilled water with a cell scraper and subjected to a first centrifugation at 1200 g for 5 min to separate the nuclei, the pellet was discarded and the supernatant was centrifuged again at 10,000 g for 10 min to separate a pelleted fraction containing mitochondria and a supernatant fraction corresponding to cytosol 53 . In both fractions, Mn, assumed as indicator for MnQ2 and MnM2 permeation, was measured by ICP-AES. Raw Mn values were normalized to cellular proteins, evaluated by the micro-bicinchoninic acid (BCA) method, and expressed as ng μg −1 of proteins. These experiments were performed in triplicate.

Cellular oxidative stress.
A cellular model of oxidative stress was used to investigate the SOD-mimetic properties of MnQ2 and MnM2. Briefly, H9c2 cells were subjected to hypoxia and reoxygenation (H + R) as previously described 18 . The cells were incubated in Dulbecco modified Eagle's medium (DMEM) with no serum or glucose and placed in a hypoxic chamber saturated with a 0.1% O 2 , 5% CO 2 , 95% N 2 gaseous mix, humidified and warmed at 37 °C, for 7 h. At end hypoxia, the cells were incubated in normoxic conditions in glucose-containing, serum-free DMEM. Cells were treated with either MnQ2 or MnM2 at concentration of 10 μmol L −1 , based on our previous in vitro studies 18,19 , added to the medium at the time of reoxygenation concurrently with the climax of ROS generation 42,43 and maintained for 1 or 2 h As controls for the specific capability of MnQ2 and MnM2 to suppress oxidative stress by ⋅− O 2 dismutation, some cell viability experiments were performed using the inactive congeners ZnQ2 and ZnM2 at the same concentration (10 μmol L −1 ) added at reoxygenation. www.nature.com/scientificreports www.nature.com/scientificreports/ Mitochondrial number. Preliminary experiments were performed to evaluate the amount of mitochondria in control and treated cells to exclude that possible changes in mitochondrial function could be due to uneven mitochondrial numbers. Briefly, H9c2 cells seeded on glass coverslips were loaded with MitoTracker Deep Red 633 mitochondrial fluorescent dye (0.5 μM, Life Technologies, Carlsbad, CA, USA) dissolved in 0.1% DMSO and Pluronic acid F-127 (0.01% w/v), added to the culture medium for 20 min at 37 °C. Cells were fixed in 2% buffered paraformaldehyde for 10 min at room temperature and red fluorescence was analyzed using a Leica TCS SP5 confocal scanning microscope (Leica, Mannheim, Germany) equipped with an argon laser source (excitation λ 633 nm) and a x63 oil immersion objective. Mitochondrial amount was also measured by flow cytometry. Single-cell suspensions were incubated with MitoTracker Deep Red 633 (200 nM) for 20 min at 37 °C and immediately analysed with a FACSCanto flow cytometer (Becton-Dickinson, San Jose, CA). Data were analyzed using FACSDiva software (Becton-Dickinson). Since both methods detected no substantial differences among the different experimental groups, the results of the subsequent experiments to assess mitochondrial function were deemed reliable. Mitochondrial activity. This was measured using the fluorometric resazurin reduction method (CellTiter-Blue, Promega Corp.). Resazurin is a redox dye commonly used as an indicator of chemical cytotoxicity in cultured cells. The assay is based on the ability of viable, metabolically active cells to reduce resazurin to resorufin and dihydroresorufin, proportionally to their number. Conversion occurs intracellularly and is facilitated by mitochondrial, microsomal and cytosolic oxido-reductases. Resorufin produced by resazurin bioreduction was measured fluorometrically (excitation λ 571 nm). Resazurin is non-toxic to cells and stable in culture medium, allowing continuous measurement of cell proliferation in vitro as either a kinetic or endpoint assay.

Intracellular ROS and mitochondrial
Mitochondrial membrane potential (Δψ). Mitochondrial membrane potential was assessed using tetramethylrhodamine methyl ester perchlorate (TMRM), a lipophilic potentiometric fluorescent dye whose accumulation in mitochondria is directly related to mitochondrial potential Δψ, as described 18 . For confocal microscope analysis, cells were cultured on glass coverslips and loaded for 20 min at 37 °C with TMRM, dissolved in 0.1% DMSO to a 100 nM final concentration in the culture medium. The cells were fixed in 2% buffered paraformaldehyde for 10 min at room temperature and the TMRM fluorescence analyzed under a confocal Leica TCS SP5 scanning microscope equipped with a helium-neon laser source (excitation λ 543 nm) and a x63 oil immersion objective. Mitochondrial membrane potential was also quantified by flow cytometry, as described 18,55 . Single-cell suspensions were washed twice with phosphate-buffered saline (PBS) and incubated for 20 min at 37 °C in the dark with TMRM dissolved in DMEM (100 nM). The cells were then washed, resuspended in PBS and analyzed using a FACSCanto flow cytometer (Becton-Dickinson).
Mitochondrial permeability transition pore opening (mPTP). Mitochondrial permeability, an index of mitochondrial dysfunction and early apoptosis, was measured by calcein fluorescence, as described 18,56 . The fluorescent probe calcein-AM readily enters into cells and emits fluorescence upon de-esterification. Co-loading of cells with cobalt chloride, which cannot cross the mitochondrial membranes in living cells, quenches calcein fluorescence in the whole cell except mitochondria. During induction of mPTP, cobalt can enter mitochondria and quenches calcein fluorescence, whose decrease can be taken as a measure of the extent of mPTP induction. H9c2 cells grown on glass coverslips were loaded with calcein-AM (3 µM) and cobalt chloride (1 mM) added to the culture medium for 20 min at 37 °C. The cells were then washed in PBS, fixed in 2% buffered paraformaldehyde for 10 min at room temperature and analyzed by a Leica TCS SP5 confocal laser scanning microscope equipped with an argon laser source (excitation λ 488 nm) and a x63 oil immersion objective. Mitochondrial permeability was also monitored by flow cytometry: single-cell suspensions were incubated with calcein-AM (3 µM) and cobalt chloride (1 mM) for 20 min at 37 °C, washed twice with PBS and analyzed using a FACSCanto flow cytometer (Becton-Dickinson).
Evaluation of lipid peroxidation. Lipid peroxidation was investigated by confocal scanning microscopy using BODIPY 581/591 C11 (Life Technologies, Carlsbad, CA, USA), a fluorescent probe that is intrinsically lipophilic and thus mimics the properties of natural lipids, as described 54,57 . BODIPY 581/591 C11 acts as a fluorescent lipid peroxidation reporter that shifts its fluorescence from red to green in the presence of oxidizing agents. Briefly, cells were cultured on glass coverslips and loaded with BODIPY, dissolved in 0.1% DMSO (2.5 mM final concentration), added to the cell culture media for 15 min at 37 °C. The cells were fixed in 2.0% buffered paraformaldehyde for 10 min at room temperature and the BODIPY fluorescence analyzed (excitation λ 581 nm) using a confocal Leica TCS SP5 scanning microscope equipped with an argon laser source for fluorescence measurements. A series of optical sections (1024 × 1024 pixels) 1.0 μm in thickness was taken through the www.nature.com/scientificreports www.nature.com/scientificreports/ cell depth at intervals of 0.5 μm using a Leica Plan Apo 63X oil immersion objective and then projected as a single composite image by superimposition. Moreover, lipid peroxidation was quantified by flow cytometry. Single-cell suspensions were washed twice with PBS and incubated, in the dark, for 30 min at 37 °C with BODIPY 581/591 (2.5 mM) in DMEM. After labeling, cells were washed and resuspended in PBS and analyzed using a FACSCanto flow cytometer (Becton-Dickinson, San Jose, CA). Assessment of caspase activity. Since mitochondrial dysfunction is a well-known trigger of apoptosis, we next investigated the activation of pro-apoptotic initiator caspases 8 (extrinsic pathway) and 9 (intrinsic pathway), and effector caspase 3 and the possible influence of H + R, as described 18 . Briefly, MnQ2 and MnM2.
H9c2 cells seeded on glass coverslips were incubated with FAM-FLICA ™ Caspase assay kit (Immunochemistry Technologies, Bloomington, MN, USA) for 30 min, following the manufacturer's instructions. After incubation, the cells were thoroughly washed and fixed in 2% buffered paraformaldehyde for 10 min at room temperature. Fluorescence was detected by a confocal Leica TCS SP5 scanning microscope equipped with an argon laser source (excitation λ 488 nm) and a x63 oil immersion objective. Caspase activity was also quantified by flow cytometry, as previously reported 18,58 single-cell suspensions were incubated with FAM-FLICA ™ for 30 min at 37 °C, washed twice with PBS and analyzed using a FACSCanto flow cytometer (Becton-Dickinson).
Cell death assay. Lactate dehydrogenase (LDH) activity, accounting for cell death, was assessed spectrophometrically in the culture medium and in adherent cells (in order to obtain total LDH content) using the LDH assay kit (Roche Diagnostics, Mannheim, Germany). LDH release was calculated as a percentage of total LDH content.
Statistical analysis. The reported data are expressed as the mean ± SEM of at least 3 independent experiments. Statistical comparison of differences between groups was carried out using one-way analysis of variance (ANOVA) followed by Student-Newman-Keuls multiple comparison test. Comparisons between MnQ2 and MnM2 were performed with Student's t test for unpaired values. A p value ≤ 0.05 was considered significant. Calculations were done using GraphPad Prism 5.0 statistical program (GraphPad Software, San Diego, CA, USA).