Many studies now show that the cis, cis-1,3,5-triaminocyclohexane (tach) framework is well-suited to complexation of first-row dicationic transition-metals of ionic radius[1] 0.75 to 0.88 Å.[2–4] In studies of aqueous and biological coordination chemistry, we identified the N, N′, N″-2-pyridylmethyl pendant arm derivative, tachpyr, as a Zn(II)- and Fe(II)-complexing agent in vitro as well as a cytotoxic agent with potential application to cancer therapy.[5–7] Notably upon treatment of cells with tachpyr, ferritin synthesis is depressed, which is indicative of cellular iron deficiency.[6] In addition to a cytotoxic pathway involving sequestration and cellular deprivation of biometal ions by chelators, the induction of cellular oxidative damage via catalytic coordinated metal ion is also possible.[8] The cytotoxicity of tripodal aminopyridyl chelators based on tach and tren (tris(2-aminoethyl)amine) frameworks is quite sensitive to substituents on both the pyridyl rings and the framework amine nitrogens. Specifically, methylation at the pyridine-ring 6-position or the amine nitrogen completely suppresses aqueous metal ion-chelation and cytotoxicity, while methylation at the 3-ring position does not hinder metal chelation and accelerates cell death.[9–11]

When certain tachpyr-family chelators bind Fe(II) under aerobic conditions (or bind Fe(III) under N₂ or O₂, or when Fe(II)-tach-family complexes are exposed to oxidants), iron-mediated oxidative dehydrogenation occurs giving [Fe(tachpyr-monoimine)]²⁺ and [Fe(tachpyr-diimine)]²⁺ complexes, as shown in Scheme 1. For reasons that will be discussed in a separate paper, tachpyr-trisimine does not interact strongly with Fe(II) and the [Fe(tachpyr-trisimine)]2+ complexes do not form readily. This reaction is significant to chemistry and biology of these chelators in at least two ways. The redox process may generate reactive oxygen species that can cause cellular injury and death.[12] Second, π-backbonding is greater to the iminopyridyl group than to the aminopyridyl group, affording stronger Fe(II) binding.[13] Thus although Zn(II) competes favorably with Fe(II) for tachpyr under anaerobic conditions, under aerobic conditions tachpyr binds increasing amounts of iron relative to zinc, which we postulated is due to inertness of Fe(II)-tachpyr-mono- and bis-imine complexes relative to the reversibly-bound Zn(II) of the complex [Zn(tachpyr)]²⁺.[14] Studies of extracts of cells treated with tachpyr indicate that the mono- and bis-imine complexes are likely forms of sequestered iron or zinc in cells.[5]

Scheme 1 Fe(II) complexes studied herein.

It was therefore of great interest to pursue further substitutions (beyond methylation) at the 3-position of the tachpyr chelator, particularly to study oxidative dehydrogenation of Fe complexes with derivatized tachpyr. Herein, we report synthesis of the 3-methoxylated chelator, tach-3-MeOpyr, structural and solution properties of the Fe(II) complexes of tach-3-Mepyr, tach-3-MeOpyr and tach-6-Mepyr, the oxidative dehydrogenation reactivity of the Fe(II)-complexes of these three chelators relative to the parent tachpyr, and a comparison of cytotoxic action of the chelators toward cultured Hela cells.

The chelator¹ tach-3-MeOpyr was synthesized from the triamine and the appropriately substituted pyridine carboxaldehyde by first forming a tris-imine that was subsequently reduced to tris-amine. The ligand readily complexed Fe(II) as seen by reaction with a slight excess of FeCl₂•4H₂O in MeOH or H₂O under anaerobic conditions. The complex ^{crystallized as [Fe(tach-3-MeOpyr)](FeCl₄)•2MeOH2} when a greater excess of FeCl₂•4H₂O was employed. We prepared the ligands tach-3-Mepyr, tach-6-Mepyr and tachpyr plus their Fe(II) complexes 1, 3, and 4 as previously reported (Scheme 1).[9,15]

The ¹H NMR spectrum of [Fe(tach-3MeOpyr] ²⁺ indicates a diamagnetic, low-spin state of Fe(II), while the electronic spectrum is dominated by a metal-to-ligand charge-transfer (MLCT) band at 23000 cm⁻¹ (ε= 6500 M⁻¹ cm⁻¹) with a low-energy shoulder at 18800 cm⁻¹ (ε = 350 M⁻¹ cm⁻¹), quite similar to that of [Fe(tach-3-Mepyr)]²⁺ whose MLCT band appears at 22700 cm⁻¹ (ε = 7000 M⁻¹ cm⁻¹) and shoulder at 18200 cm⁻¹ (ε = 290 M⁻¹ cm⁻¹).[9] The low-energy shoulder has previously been assigned as the ¹A_1g → ¹T_1g transition.[16,17]

In order to investigate the process of oxidative dehydrogenation in the ring-substituted tach chelators, solutions of [Fe(tach-3-Mepyr)]²⁺ (1) or [Fe(tach-3-MeOpyr)]²⁺ (2) were exposed to air. This caused iron mediated oxidative dehydrogenation, with spectral changes matching those we reported for a solution of [Fe(tachpyr)]²⁺ (4).[11] For both 1 and 2, a new charge-transfer band having a high-energy shoulder appears at ca. 16250 cm⁻¹ and as the oxidation progresses, the MLCT band at ca. 23000 cm⁻¹ shifts toward higher energy while the band at ca. 16250 cm⁻¹ gains intensity as the solution changes from amber to green and ultimately blue, consistent with conversion to a mixture of two [Fe(tach-3-Rpyr-imino)]²⁺ species (Scheme 1). Because the monoimine complexes (left structure of Scheme 1) are oxidized by 2 H•, they are designated [Fe(tach-3-Rpyr-ox-2)]²⁺ and similarly the bisimine complexes (right structure of Scheme 1) are termed [Fe(tach-3-Rpyr-ox-4)]²⁺ (R = Me or MeO). As expected from the low-spin, diamagnetic NMR and in analogy to other chelated σ-donor, π-acceptor ligand Fe(II) complexes such as [Fe(phen)₃]²⁺, the tach-3-Mepyr and tach-3-MeOpyr ligands and their mono- and diimino derivatives interact strongly with Fe(II). Oxidative dehydrogenation reactivity is not observed for the high-spin complex [Fe(tach-6-Mepyr)]²⁺.

Comparisons of the rates of oxidative-dehydrogenation between 1, 2 and 4 were obtained using ¹H NMR and ESI-MS. Proton NMR proved useful for quantitation using the signals of the imino protons, –N=CH–, which appear as a singlet in the range of 9.2 to 9.6 ppm for the monoimino complexes derived from 1, 2 and 4 and as a doublet in the range of 8.8 to 9.2 ppm for the diimino complexes. Figure 1 shows the ¹H NMR of 1 in the useful region for quantitation of the amount of monoimine formed, where the aromatic protons are arbitrarily set to 100 H and the integral of the monoimine species at δ 9.59 is measured at 4 h and 24 h exposure to air. Similar data was obtained for 2 and 4 and all integrals are tabulated in Table 1. This data was supported by electrospray-mass spectroscopic studies of the oxidized mixtures, which are shown in the Supplementary Data along with the ¹H NMR spectra of 2 and 4. The qualitative order of oxidation rate of the tach-3-Rpyr family is tach-3-Mepyr > tachpyr > tach-3-MeOpyr (Table 1). Due to the similar chemical properties of the oxidized species, they could not be separated, and a rigorous kinetic analysis of the oxidation process was not successful, which was attributed to the overlap of spectral peaks from the concurrent conversions of [FeL]²⁺ → [Fe(L-ox-2)]²⁺ and [Fe(L-ox-2)]²⁺ → [Fe(L-ox-4)]²⁺, precluding accurate quantitation of reactants and products.

Figure 1 Proton NMR spectrum (D2O, 25 °C) of the progressing oxidation of [Fe(tach-3-Mepyr)]2+ (24 h, upper; 4 h, lower) showing the monoimine species as a singlet at δ 9.56 ppm and diimine as a doublet centered at δ 9.25 ppm. The integration (more ...)

Table 1 Ratio of the area under the monoimine peak of [Fe(L-ox-2)]2+ to the area under the aromatic peaks (1H NMR, D2O, 25 °C).

At present, an explanation for the above rate order is not apparent. Goto et. al. reported an effect of geometry upon oxidative dehydrogenation, whereby the mer- isomer of [Fe^II(ampy)₂(CN)₃]⁻ formed an Fe(II)-imine complex when oxidized with peroxodisulfate, but oxidation of the fac- isomer produced fac-[Fe^III(ampy)₂(CN)₃]. However in the present case, there is no significant difference in coordination geometry of 1 vs. 2, nor is there any substantial difference in ligand-field strength of the three ligands 1, 2 and 4, based on interpretation of the electronic spectra of the Ni(II) and Cu(II) complexes.[9][18] Although numerous examples of iron-mediated oxidative dehydrogenation are known, more study of the mechanistic aspects is needed.[19–22] By comparison greater progress has been made understanding the oxidative dehydrogenation of Ru and Os complexes, possibly because the reaction mediated by these metals is slower and easier to study.[13,23–27].

Although the chelator tach-6-Mepyr has been found not cytotoxic,[9] the possibility that its iron complex possesses an open coordination site was studied. It is known that open coordination sites in iron complexes are important to the redox activity of complexes such as [Fe^(II/III)EDTA]^2−/−.[28] The complex [Fe(tach-6-Mepyr)]²⁺ has been assigned as high-spin (HS) based on an MLCT band at 28800 cm⁻¹ (ε of this band is weak, 650 M⁻¹ cm⁻¹, as is characteristic of Fe(II)-HS) along with a weak d-d transition at 11700 cm⁻¹ (ε= 9.4 M⁻¹ cm⁻¹) in the UV-Vis spectrum and a measured magnetic moment μ_eff = 5.46 B.M. (DMSO solution, Evans’ method).[9] In view of this and the X-ray structural findings (vide infra), it was considered likely that 3 could contain an open or substitutable coordination site in solution. Therefore, ancillary ligands were added to a MeCN solution of 3 to probe the presence of such sites. With the addition of one molar equivalent of either a chloride or thiocyanate anion, a red shift of the charge-transfer band was observed. The shift is believed to be due to the formation of [Fe(tach-6-Mepyr)(Cl)]⁺ and [Fe(tach-6-Mepyr)(SCN)]⁺ respectively, as this is consistent with observations of Bernal and Toftlund who identified replacement of pyridine by water in another Fe^II complex of a ligand containing 6-methyl-2-aminomethylpyridyl donors, [Fe(N,N′-bis(6-methyl-2-pyridylmethyl)-N,N′-bis(2-pyridylmethyl)ethane-1,2-diamine)(H₂O)]²⁺.[29] In sum, although this experiment could not be conducted in aqueous medium due to the lability of [Fe(tach-6-Mepyr)]²⁺,[9], the reactivity of 3 in MeCN solvent is predicted by its structure.

To obtain further chelator structural parameters that may underlie optimal Fe(II) binding, the X-ray structures³ of the dications 1, 2 and 3 were determined as the salts [Fe(tach-3-Mepyr)]Cl₂•2MeOH, [Fe(tach-3-MeOpyr)](FeCl₄)•2MeOH, and [Fe(tach-6-Mepyr)](ClO₄)₂•H₂O•MeOH respectively (Figures 2 – 4). The average Fe-N_py and Fe-N_amine bond lengths for 1 and 2 are shown in Table 2. A comparison with other low-spin (LS), six-coordinate Fe(II) complexes that contain the 2-aminomethylpyridyl group such as [Fe(tptcn)]²⁺ and fac-[Fe(ampy)₃]²⁺ shows good agreement in the bond lengths (Table 1). As expected, the Fe-N_py bonds are generally shorter than Fe-N_amine bonds due to d_π → p_π* backbonding of Fe(II) to the ligand.[30] Both 1 and 2 possess a trigonally distorted octahedral coordination geometry, as indicated by their trigonal twist angles () of 54.9° and 54.6° respectively compared to a of 60° for a perfect octahedron. The average N_py–Fe–N_amine angles of the 5-membered chelate rings (α) of 83.6° in 1 and 84.0° in 2 are comparable to other LS aminomethylpyridyl iron complexes as seen in the top section of Table 2.

Figure 2 Structure of the complex cation of [Fe(tach-3-Mepyr)]Cl2 (ORTEP; 50% probability ellipsoids) showing the atom-numbering scheme.

Figure 4 Structure of the complex cation of [Fe(tach-6-Mepyr)](ClO4)2 (ORTEP; 50% probability ellipsoids) showing the atom-numbering scheme.

Table 2 Selected structural parameters of [Fe(tach-3-Mepyr)]2+ (1), [Fe(tach-3-MeOpyr)]2+ (2), [Fe(tach-6-Mepyr)]2+ (3), and reference compounds.

The average Fe-N_py and Fe-N_amine bond lengths for 3 are 2.279(3) and 2.173(3) Å respectively. The elongation of the Fe-N bonds in 3 compared to 1 and 2 is attributed to the HS nature of the complex where electrons now populate the e_g* orbitals and the six-coordinate ionic radius of Fe(II) is 0.92 Å. The steric interactions between the 6-methyl groups are most evident when examining the Fe-N_py bonds. For each of the two independent molecules in the asymmetric unit there is one Fe-N_py bond that is ~0.15 Å longer than the other two. The lengthening of one bond to compensate for steric interactions between 6-methyl groups was also reported in the Zn(II) and Mn(II) complexes of the same ligand, which are isostructural with [Fe(tach-6-Mepyr)](ClO₄)₂ (Table 1, bottom section).[9] Even with these steric interactions the complex is able to obtain a geometry close to octahedral with an average twist angle of 53.0°. The bite angle of 77.9° is consistent with the elongated Fe-N bonds and is in good agreement with other HS aminopyridyl iron complexes as seen in Table 2. Taken together the structural findings indicate that tach-3-Mepyr and tach-3-MeOpyr interact strongly with Fe(II) in an ideal geometry for this metal ion, whereas steric hindrance prevents a good interaction of tach-6-Mepyr with Fe(II).

The cytotoxicities of tach-3-Mepyr and tach-3-MeOpyr toward cultured Hela cancer cells⁴ were determined against a tachpyr control (Table 3). The chelator tach-6-Mepyr displayed no cytotoxicity and was not studied further. At 24 h tach-3-Mepyr is more toxic than tachpyr and much more toxic than tach-3-MeOpyr. However, the toxicity of tach-3-MeOpyr increases dramatically from 24 h to 72 h, as expressed by the value ΔIC₅₀ (24h – 72h). After 72 h, it only takes twice as much tach-3-MeOpyr as either tach-3-Mepyr or tachpyr to kill half the cells. This finding might be explained by the need for a greater degree of imine formation in order to retain bound iron. Thus, a chelator with slower iron-mediated oxidative dehydrogenation such as tach-3-MeOpyr does not convert to the inert [Fe(L-monoimine)]²⁺ form as quickly, so it takes longer to sequester iron inertly. An alternative explanation which cannot be distinguished at this time is that the oxidative dehydrogenation activity forms reactive oxygen species which cause cell death. It is also possible that tach-3-MeOpyr does not as readily diffuse to the regions of the cell where iron may be bound.

Table 3 IC50 values (μM) of tachpyr, tach-3-Mepyr, and tach-3-MeOpyr after 24, 48, and 72 h.

The chelators tach-3-Mepyr, tach-3-MeOpyr and tachpyr bind Fe(II) strongly and equally well based on ground-state structure, however their rates of iron-mediated oxidative dehydrogenation differ in the order, fast to slow, of tach-3-Mepyr > tachpyr > tach-3-MeOpyr. They are equally cytotoxic to cultured Hela cells, but their rates of cell-killing are also in the order fast to slow of tach-3-Mepyr > tachpyr > tach-3-MeOpyr. The underlying biological and chemical reasons for these relationships are not yet understood. Introduction of steric bulk produces an elongated Fe(II)-N_py bond in the structure of [Fe(tach-6-Mepyr)]²⁺ and accordingly this complex has a substitutable coordination site in MeCN solution.

We thank the NIH for the support of this work through Grant DK-57781-R01 to S. V. T. and through the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.

The ¹H NMR spectra of the oxidation process of 2 and 4 and ESI-MS spectra of 1 and 2 are contained in the supplementary data, which can be found in the online version, at _____. The CIF files for the structural analyses of [Fe(tach-3-Mepyr)]Cl₂•2MeOH, [Fe(tach-3-MeOpyr)](FeCl₄)•Et₂O and [Fe(tach-6-Mepyr)](ClO₄)₂•H₂O•MeOH havebeen deposited with the Cambridge Crystallographic Data Centre, CCDC nos. 621772, 620255, and 620253. Copies of this information may be obtained free of charge at www.ccdc.cam.ac.uk/conts/retrieving.html or from The Director, CCDC, 12 Union Road, Cambridge, CB2 1EZ, UK (fax: +44-1233-336033; e-mail deposit/at/ccdc.cam.ac.uk).