Detailed Structural and Spectroscopic Elucidation of Ferrocenium Coupled N-heterocyclic Carbene Gold(I) Complexes

Abstract

Unambiguous assignment of redox sites on ferrocene coupled N-heterocyclic carbene gold(I) complexes [(Fc-NHC)₂Au(I)]⁺ is critical to gain a greater mechanistic understanding of their activity in a cellular environment. Such information can be garnered with isolation and detailed characterization of the oxidized version of [(Fc-NHC)₂Au(I)]⁺. Herein we disclose a study that unambiguously illustrates redox events pertaining to [(Fc-NHC)₂Au(I)]⁺ that stem exclusively from ferrocene sties. This work also describes novel synthetic methodologies for isolating ferrocenium coupled N-heterocyclic carbene gold(I) complexes.

Graphical Abstract

Unambiguous assignment of redox sites on ferrocene coupled N-heterocyclic carbene gold(I) anticancer therapeutic agents is described.

Introduction

Cisplatin (Figure 1A) and its clinically approved derivatives are the most widely known metal-containing anticancer drugs used to treat various cancers, including head and neck, bladder, breast, cervical, and ovarian.^{1, 2} While effective as an anticancer drug, patients often experience adverse events, including 1) platinum resistance associated with DNA damage resistance mechanisms and 2) platinum reactivity as a coordination complex that is partly responsible for the distressing side effects with nephrotoxicity often being dose-limiting.^{3, 4} As a consequence of this, organometallic compounds with novel mechanisms of action have been explored to maximize efficacy and minimize the adverse side effects.^{5, 6} Drawing inspiration from the success of platinum-based drugs, several metal-containing compounds have been identified as a potential cancer therapeutic agents.^6–18

Figure 1:

Metal containing drugs in clinics.

In particular, metallocenes have gained interest due to their enhanced stability in biological environments relative to coordination complexes such as cisplatin.¹⁹ Of all metallocenes, ferrocene derivatives were the first to exhibit therapeutic properties, and as a result, ferrocene derivatives have been extensively studied by the medicinal chemistry community.¹⁹ For example, ferroquine (Figure 1B), containing a ferrocene unit, continued to phase II clinical trials.¹⁹ With this initial advancement, several other ferrocene containing compounds have been developed and tried as potential therapeutic agents, especially as potential anticancer agents.²⁰

Gold has long been known for having medicinal properties. Gold(I) salts such as aurothioglucose (Figure 2A, Solganol®),²¹ disodium aurothiomalate (Figure 2B, Myocrysin®),²¹ sodium gold(I) 4-amino-2-mercaptobezoic acid (Figure 2C, Krysolgan®²¹ and (2,3,4,6-tetra-O-acetyl-1-thio-D-glucopyranosyl) gold(I) triethylphosphine (Figure 2D, Auranofin®)²¹ are approved and clinically used in the treatment of rheumatoid arthritis.²² Due to the success of platinum anticancer drugs and the success of gold complexes in medicine throughout history, gold complexes have been explored as potential anticancer agents.^{16, 23–36} Both gold (I) and gold (III) complexes have been shown to inhibit cancer cell growth by inhibiting the antioxidant enzyme thioredoxin reductase (TrxR).³⁷ Recently, our group reported several gold(I) complexes (Figure 3A and 3B) containing redox-active motifs (e.g., ferrocene, naphthoquinone) as a means of targeting the antioxidant pathway via 1) protein inhibition of TrxR and 2) accentuation of reactive oxygen species (ROS) via the redox cycling center (Figure 4).^{29, 33, 38}

Figure 2:

Molecular structure of gold(I) based compounds in therapeutics, aurothioglucose (2A; Sloganol®), disodium aurothiomalate (2B; Myochrisin®), sodium gold(I) 4-amino-2-mercaptobezoic acid (2C; Krysolgan®), and (2,3,4,6-tetra-O-acetyl-1-thio-D-glucopyranosyl) gold(I) triethylphosphine (2D; Auranofin®).

Figure 3:

Previously reported ferrocene coupled N-heterocyclic carbene gold(I) complexes from our group.

Figure 4:

The plausible mechanistic pathway of ferrocene coupled N-heterocyclic carbene gold(I) complexes killings cancer cells.

It is thus proposed that the combination of these two systems overwhelms the cellular redox regulation system and triggers apoptosis in cancer cells (Figure 4). Moreover, we discovered that the potency of gold(I) complexes containing ferrocene coupled NHCs is greater than the sum of its parts. In simple terms, its antiproliferative activity is greater than ferrocene and gold(I) units when used in combination as individual parts (i.e., ferrocene + Au-NHC).²⁹ Cell proliferation studies also supported the notion that the potency of these drugs is directly proportional to the number of ferrocenes attached to the NHC unit.²⁹ To gain a greater mechanistic understanding of these potential therapeutic agents in a cellular environment, an unambiguous assignment of redox sites must be established. Unfortunately, it is impossible to generate this data using electrochemical measurements, but rather it requires isolation and detailed characterization of oxidized gold(I) complexes containing ferrocene coupled NHCs. To facilitate this, herein, we disclose single-crystal X-ray diffraction studies and Mössbauer measurements that unambiguously illustrate redox events pertaining to gold(I) complexes (Figure 3A and 3B) stem exclusively from ferrocene sties.

Results and discussion

Synthesis and Characterization

As previously reported, cell proliferation studies of dual targeting compounds [2][Cl] and [5][Cl] (Figure 3A and 3B) containing gold(I) and ferrocene as a single unit increased anticancer cytotoxicity that was proportional to the increase in ferrocene units per complex.²⁹ It is proposed that gold(I) binds to TrxR and suppresses its antioxidant activities while ferrocene participates in generating non-specific ROS. This dual process perturbs ROS levels in the cellular environment to trigger apoptosis in cancer cells. However, the specific site that participates in ROS generation (redox activity) in these compounds remains unknown. Moreover, it is proposed that ferrocene groups are oxidized to the ferrocenium species (Figure 4) during the ROS generation cycle. To confirm this, we set out to isolate chemically oxidized versions of [Fc-NHC)₂Au(I)]⁺ complexes, [3][BF₄]₃ and [6][BF₄]₅ (Scheme 1 and 2). To avoid unnecessary complications with the halide counterions in [2][X] and [5][X] (X = Cl or I) complexes during the redox processes, we choose to work with inert BF₄^– counter ions.

Scheme 1:

Preparation of oxidized gold(I) complexes containing ferrocene coupled NHC units. a = Na(N(SiMe₃)₂, toluene, (C₄H₈S)AuCl; b = 2.3 equiv. AgBF₄, CH₂Cl₂.

Scheme 2:

Preparation of oxidized gold(I) complexes containing ferrocene coupled NHC units. a = Na(N(SiMe₃)₂, toluene, (C₄H₈S)AuCl; b = 5 equiv. AgBF₄, CH₂Cl₂.

The imidazolium salts [1][BF₄] and [4][BF₄] were readily prepared by treating [1-mesityl(3-ferrocenylmethyl)imidazolium)][iodide]³⁹ or [1,3-di(ferrocenylmethyl) imidazolium)][iodide]⁴⁰ with triethyloxonium tetrafluoroborate, respectively. The key difference between [1][BF₄] and [4][BF₄] is the number of ferrocene units; [1][BF₄] contains a single ferrocene unit, whereas [4][BF₄] contain two ferrocene units. The yield for compound [1][BF₄] was over 90%, while compound [4][BF₄] had only around a 70% recovery. Imidazolium salts [1][BF₄] and [4][BF₄] were then treated with sodium bis(trimethylsilyl)amide and the resulting solution was then treated with chloro(tetrahydrothiophene)gold(I) to afford compounds [2][BF₄] and [5][BF₄], respectively. Compound [2][BF₄] had a much higher yield at 74% compared to [5][BF₄] with a yield of 33%. Also, compound [5][BF₄] had limited solubility in common organic solvents, while compound [2][BF₄] was readily soluble in CH₂Cl₂ and tetrahydrofuran (THF). Finally, the compound [2][BF₄] and [5][BF₄] were subjected to chemical oxidation in CH₂Cl₂ using Ag[BF₄]. However, there was a slight variation in their reaction parameters. The compound [2][BF₄] only required 2.3 equiv. of Ag[BF₄] and an hour for oxidation to go to completion. While for [5][BF₄], it took 18 hours and 5 equiv. of Ag[BF₄] to go to completion. After passing through a celite plug and removing the volatiles yielded [3][BF₄]₃ in 89.3% yield. Due to limited solubility in CH₂Cl₂ [6][BF₄]₅ was re-dissolved in acetonitrile and filtered through celite to obtain compound [6][BF₄]₅ in 65.3% yield. Longer hours ensured complete oxidation of [6][BF₄] in CH₂Cl₂; moreover, conducting the reaction in CH₃CN did not yield the desired oxidized product.

The synthesized compounds [1-6][BF₄]_n were analyzed using various analytical techniques, including NMR spectroscopy and single-crystal X-ray diffraction studies. ¹H and ¹³C NMR spectra for compounds [1][BF₄], [2][BF₄], [4][BF₄] and [5][BF₄] were recorded in CDCl₃ and the key values are summarized in Table 1. Proton NMR spectral analyses of [1][BF₄] and [4][BF₄] in CDCl₃ displayed a characteristic peak at 8.70 ppm and 8.93 ppm, respectively, consistent with the presence of –NC(H)N– imidazolium proton. Compound [2][BF₄] displayed diagnostic proton NMR signals consistent with the presence of [2][BF₄] in the solution. For example, in ¹H NMR (CDCl₃), mesityl-CH₃ hydrogen signals (ortho-CH₃), corresponding to 12 hydrogen atoms, were observed at 1.80 ppm, while mesityl-CH₃ hydrogen signals (para-CH₃), corresponding to 6 hydrogen atoms, were observed at 2.39 ppm. In the ¹³C NMR spectrum (CDCl₃), a diagnostic chemical shift corresponding to C_carbene–Au–C_carbene was observed at 183.19 ppm. This value is consistent with other reported [(NHC)–Au–(NHC)]⁺ complexes, such as bis(1,3-dimesitylimidazol-2-ylidene)-gold(I) tetrafluoroborate (185.1 ppm, CDCl₃),⁴¹ bis(1,3-dimethylimidazol-2-ylidene)-gold(I) bromide (183.3 ppm, (CD₃)₂SO),⁴² and bis(1,3-dicyclohexylimidazol-2-ylidene)-gold(I) chloride (180.4 ppm, (CD₃)₂SO).⁴² While a diagnostic chemical shift corresponding to C_carbene–Au–C_carbene for [5][BF₄] was observed at 181.72 ppm. This confirms the presence of [(NHC)–Au–(NHC)]⁺ core in the solution for complexes [2][BF₄] and [5][BF₄]. Since [3][BF₄]₃ and [6][BF₄]₅ are paramagnetic, NMR data were not generated for these samples.

Table 1:

¹H and ¹³C NMR chemical shifts for the metal complexes.

Compounds	–NC(H)N– (δ, ppm)	–N(C)N– (δ, ppm)
[1][BF4]	8.77	–
[2][BF4]	–	183.19
[4][BF4]	8.93	–
[5][BF4]	–	181.72

Crystallography

Next, we investigated the solid-state structure of these complexes ([2][BF₄], [3][BF₄]₃, and [5][BF₄] using single-crystal X-ray crystallography. Unfortunately, we could not obtain suitable single crystals for [6][BF₄]₅ for diffraction analysis. Prior reports from our group elucidate the solid-state characterization of [2][Cl], and [5][Cl], all with halide counter ions, this report details solid-state characterization of [2][BF₄], [3][BF₄], and [5][BF₄]₃. Golden yellow crystals of [2][BF₄] were obtained by slow diffusion of pentane into a saturated dichloromethane solution. While forest green crystals of [3][BF₄]₃ were collected by slow diffusion of diethyl ether into a saturated 1,2-dichloroethane solution. Golden orange crystals of [5][BF₄] were collected by slow diffusion of pentane into a saturated 1,2-dichloroethane solution. The crystallographic parameters are displayed in Table 2. The molecular structures of [2][BF₄], [3][BF₄]₃, and [5][BF₄] are presented in Figures 5.

Table 2.

Selected Bond Length (Å) and Angles (deg) for complexes [2][BF₄], [3][BF₄]₃, and [5][BF₄]^a,b.

	[2][BF4]	[3][BF4]	[5][BF4]	[Fc][BF4]
Ccarbene−N	1.344[6]	1.349[6]	1.351[3]	-
Ccarbene−Au	2.015[5]	2.014[5]	2.115[3]	-
N—C=C—N	1.330[8]	1.341[9]	1.347[4]	-
CCH2−N	1.496[7]	1.456[7]	1.476[3]	-
CCH2−CCpc	1.482[8]	1.498[8]	1.485[4]	-
Fe−CCp	2.013[7]	2.079[6]	2.036[3]	2.095[5]
N−Ccarbene−N	105.0(4)	105.6[4]	105.1[2]
C−Au−C	175.8(2)	175.2(2)	177.2[2]
N-CCH2-Ccp	111.7[4]	112.5[4]	111.1[2]
CTCp-CTCp	3.262	3.401	3.289	3.421
Fe-CTCpd	1.631	1.711	1.644	1.711

^aAverage values are reported when two or more chemically identical interatomic distances or angles are present.

^bUncertainties were computed according to Taylor, J. R. An Introduction to Error Analysis, 2^nd ed.; University Science Books: Sausalito, CA, 1997; pp 73–77. The generated uncertainty numbers are represented within [ ]. Cp = cyclopentadienly; CT = centroid.

^cC_cp represents a carbon atom part of cyclopentadienyl ring that is attached the iron atom.

^dCT_cp represents centroid defined by carbon atoms that constitutes cyclopentadienyl ring.

Figure 5:

Molecular structure of [2][BF₄], [3][BF₄]₃, and [5][BF₄].. Thermal ellipsoid plots are drawn at 50% probability level, hydrogen atoms are omitted for clarity.

Solid-state structural analysis of [2][BF₄] and [5][BF₄] reveal that they crystallize in monoclinic space group P21/c and C2/c, respectively, with Z = 4. A linear geometry was observed with C_carbene–Au–C_carbene bond angle of 175.78(17) Å for [2][BF₄] and 177.24(13) Å [5][BF₄], respectively. The average Au–C_carbene bonds distances in [2][BF₄] (2.015[5] Å) and [5][BF₄] (2.015[3] Å) were in good accordance with prior reported [(NHC)–Au–(NHC)]⁺ core.³³ The two imidazolium rings that are connected to the gold(I) atom is twisted at an angle of 64.45° for [2][BF₄] and 83.79° [5][BF₄] to avoid steric crowing exhibited by mesityl and ferrocene groups. X-ray analysis of the ferrocenium complex reveals that [3][BF₄]₃ crystallize in the monoclinic space group P2₁/c. Due to the non-availability of solid-state structure for [6][BF₄]₅, all bond parameter comparisons will be made using [3][BF₄]₃ to confirm the site of oxidation in these complexes. The ferrocenium core in [3][BF₄]₃ exhibited an average Fe–C_cp distance of 2.079[6] Å, which is similar to the bond length observed for [FeCp₂][BF₄] (2.095[5] Å) complex (Table 2).⁴³ The average C–C (1.398 Å) bond lengths observed in [3][BF₄]₃ were in good accordance with the reported ferrocenium species (1.417[7]).⁴³ Analysis of cyclopentadienyl rings reveals that they are eclipsed in both the ferrocene and ferrocenium structures. The most common indicator to differentiate the presence of ferrocene (Fe(II)) or ferrocenium (Fe(III)) is the Fe-CT_cp (CT = centroid defined by five carbon atoms of the cyclopentadienyl ring) distance. Ferrocenium has one less electron than ferrocene as the electron is removed from the bonding orbital molecular orbital (Figure 6). The removal of an electron from the bonding orbital results in the expansion or lengthening of the Fe-CT_cp distance. Seemingly counterintuitive, removing an electron from the weak bonding orbital weakens the bond between iron and the carbon atoms of the cyclopentadienyl carbons. Hence a longer Fe-CT_cp distance is observed in the ferrocenium species. The average Fe-CT_cp (CT = centroid) distance for [2][BF₄] was found to be 1.619 Å, while 1.711 Å was observed for [3][BF₄]₃. This value matches perfectly with the reported ferrocenium species (1.711 Å). The longer Fe-CT_cp distance is consistent with the fact that the site of oxidation is the iron atom. Moreover, all the other bond lengths and angles are intact, and no significant change was observed; hence it was concluded that the site of oxidation is iron.

Figure 6:

Occupancy of iron valence orbitals for ferrocene and ferrocenium.

Electrochemistry

To ascertain the electronic properties of [2][BF₄], electrochemical measurements were performed in CH₂Cl₂ with [N(nBu)₄][PF₆] as the electrolyte. Cyclic voltammogram of [2][BF₄] and [5][BF₄] are presented in Figure 7. Peak potentials were obtained via differential pulse voltammetry measurements, and decamethylferrocene was used as an internal standard. The peak potentials were adjusted to SCE (Standard Calomel) electrode. The half-wave potential (E_1/2) of [2][BF₄] was found to be at 0.60 V and ascribed to iron-centered (Fe²⁺ to Fe³⁺) reversible oxidation. No gold oxidation was observed. Similar behaviour was observed for [5][BF₄] at 0.61 V. This experiment again indicates the first site of oxidation in [2][BF₄] and [5][BF₄] is ferrocene.

Figure 7:

Cyclic voltammogram of [2][BF₄] (top) and [5][BF₄] (bottom) in DCM with [N(ⁿBu₄)][PF₆] as electrolyte.

Ultraviolet-visible Spectroscopy

To further corroborate the site of oxidation, ultraviolet-visible absorption measurements were carried out in CH₂Cl₂ for [2][BF₄] and [3][BF₄]₃. The UV-vis absorption spectra of compounds [2][BF₄] and [3][BF₄]₃ are presented in Figure 8. The most prominent spectral feature of [2][BF₄] is the presence of a forbidden d-d transition that is localized on the Fe(II) atom at 433 nm, and that is observable in most ferrocene-containing species.³⁹ Compound [3][BF₄]₃ displayed a prominent peak at 630 nm corresponding to allowed ligand to metal charge transfer (LMCT) transition indicating the presence of ferrocenium ions. Again, this confirms the site of first oxidation in [2][BF₄] is the iron (Fe(II)→Fe(III)). The detailed analysis of UV-vis absorption spectra of [6][BF₄]₅ also revealed the site of first oxidation in [5][BF₄] is the iron (Figure S12).

Figure 8:

Electronic absorption spectra of [2][BF₄] (black) and [3][BF₄]₃ (red).

Mössbauer Spectroscopy

To further probe the oxidation of ferrocene unit, [2][BF₄], [3][BF₄]₃, [5][BF₄], and [6][BF₄]₃ were all independently analyzed using Mössbauer spectroscopy. As shown in Figure 9 and 10, complexes [2][BF₄] and [5][BF₄] were characterized by an isomer shift (δ, mm/s) at 0.56 and 0.58, respectively. Also, a large quadrupole splitting (ΔE_Q, mm/s) at 2.36 and 2.33, respectively, is consistent with the low-spin Fe(II) d⁶ atom of ferrocene. The oxidized product [3][BF₄]₃ and [6][BF₄]₅ exhibits an isomer shift (δ, mm/s) at 0.59 and 0.21, respectively. But the quadrupole splitting (ΔE_Q, mm/s) collapses and appears as a single peak. In a nutshell, Mössbauer measurements suggest that the above-observed features are characteristic of a Fe(III) d⁵ (ferrocenium) atom.

Figure 9:

Mössbauer spectra of [2][BF₄] (blue, top) and [3][BF₄]₃ (red, bottom) recorded at 4 K. The open circles represent the experimental data and the solid line corresponds to spectral fits.

Figure 10:

Zero-field Mössbauer spectra of [5][BF₄] (blue, top) and [6][BF₄]₅ (red, bottom) recorded at 4 K. The open circles represent the experimental data and the solid line corresponds to spectral fits.

Experimental

Synthetic procedures

[1-Mesityl(3-ferrocenylmethyl)imidazolium)][tetrafluoroborate], [1][BF₄].

To a 40 mL vial containing a stir bar, [1-mesityl(3-ferrocenylmethyl)imidazolium)][iodide] (1.7028 g, 3.32 mmol) was added followed by the addition of triethyloxonium tetrafluoroborate (0.6425 g, 3.38 mmol). Dry dichloromethane (8 mL) was added to the vial and the homogenous mixture was stirred overnight (18 hrs). The reaction mixture was then filtered through celite into another 40 mL vial and the resulting mixture was evaporated in vacuo to a minimal amount of solvent. The solution triturated with diethyl ether (30 mL) and then subjected to a series of washes (3 × 5 mL diethyl ether). A yellow solid was obtained after trituration and washing. Yield: 90.1 % (1.4147g). ¹H NMR (δ, CDCl₃, 300 MHz): 1.97 (s, 6H, Mes), 2.30 (s, 3H, Mes), 4.24–4.23 (m, 7H, Fc), 4.49–4.48 (m, 2H, Fc), 5.44 (s, 2H, CH2), 6.95 (s, 2H, Mes), 7.10 (s, 1H, FcNCH), 7.56 (s, 1H, MesNCH), 8.77 (s, 1H, NCHN). ¹³C NMR (δ, CDCl3, 75 MHz): 17.25, 21.07, 50.25, 69.30, 69.57, 69.83, 78.67, 122.47, 123.17, 129.80, 130.64, 134.32, 135.98, 141.29. ESI-MS. Calcd for C₂₃H₂₅FeN₂: m/z 385.1362. Observed: m/z 385.1355. Anal. Calcd. for: C₂₃H₂₅FeN₂BF₄. (H₂O)0.5: C, 57.42; H, 5.45; N, 5.82; Found: C, 57.32; H, 5.03; N, 5.60.

[Bis-(1-mesityl(3-ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tetrafluoroborate], [2][BF₄].

To a 40 mL vial containing a stir bar, [1-mesityl(3-ferrocenylmethyl)imidazolium))] [tetrafluoroborate] (0.5056 g, 1.071 mmol), and NaHMDS (0.2042 g, 1.113 mmol) were added followed by the addition of dry toluene (10 mL). The resulting mixture was stirred for an hour. (C₄H₂Si)AuCl (0.1622 g, 0.4630 mmol) was then added to the mixture and then stirred for another hour. The volatiles were then evaporated off in vacuo and dry dichloromethane (10 mL) was added to the resulting solids and the solution was then filtered through celite. The resulting homogenous solution was triturated with diethyl ether (25 mL) and washed (3 × 5 mL diethyl ether). A dark yellow solid was obtained. Yield: 74.4% (0.3624 g). ¹H NMR (δ, CDCl₃, 300 MHz): 1.80 (s, 12H, Mes), 2.39 (s, 6H, Mes), 4.19–4.18 (m, 4H, Fc) 4.22 (s, 9H, Fc), 4.26–4.25 (m, 4H, Fc), 5.08 (s, 4H, CH2), 6.84–6.83 (m, 2H, FcNCH), 6.93 (s, 4H, Mes), and 7.51–7.50 (m, 2H, MesNCH). ¹³C NMR (δ, CDCl3, 75 MHz): 17.62, 21.22, 50.68, 68.61, 68.85, 68.99, 82.53, 122.36, 129.19, 134.78, 134.85, 139.35, and 183.19. ESI-MS. Calcd for C₄₆H₄₈Fe₂N₄Au: m/z 965.2238. Observed: m/z 965.2214. Anal. Calcd for C₄₆H₄₈Fe₂N₄AuBF₄: C, 52.50; H, 4.60; N, 5.32; Found: C, 50.38; H, 4.12; N, 5.04.

[Bis-(1-mesityl(3-ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tri-tetrafluoroborate], [3][BF₄]₃.

To a 20 mL vial containing a stir bar, [Bis-(1-mesityl(3-ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tetrafluoroborate] (0.0524 g, 0.04979 mmol) and silver tetrafluoroborate (0.0224 g, 0.1151 mmol) were added followed by the addition of dry dichloromethane (3 mL). The resulting mixture was stirred for 2 hours. The reaction mixture was celite filtered and the resulting solution was evaporated in vacuo to minimal amount of dichloromethane and triturated with diethyl ether. Product was green solid. Yield: 89.3% (0.0584 g). Elemental analysis was used to confirm the identity of compound. ESI-MS. Calcd for C₄₆H₄₈Fe₂N₄Au: m/z 965.2238. Observed: m/z 965.2219. C₄₆H₄₈Fe₂N₄AuB₃F₁₂: C, 45.07; H, 3.95; N, 4.57; Found: C, 44.98; H, 3.99; N, 4.49.

[1,3-di(ferrocenylmethyl)imidazolium)][tetrafluoroborate], [4][BF₄].

To a 40 mL vial containing a stir bar, [1,3-di(ferrocenylmethyl)imidazolium)][iodide] (1.2526 g, 2.11 mmol) was added followed by the addition of triethyloxonium tetrafluoroborate (0.4522 g, 2.38 mmol). Dry CH₂Cl₂ (8 mL) was added to the mixture, and the solution was stirred overnight. The reaction mixture was then filtered through celite into another 40 mL vial, and the resulting mixture was evaporated in vacuo to a minimal amount of solvent. The solution was triturated with methyl tert-butyl ether (30 mL). The solids were dried, dissolved in methanol (1 mL), and then triturated with diethyl ether (30 mL). A yellow-green solid was obtained. Yield: 72.29 % (0.8413 g). ¹H NMR (δ, CDCl₃, 300 MHz): 4.25 (s, 14H, FcH), 4.38 (s, 4H), 5.12 (s, 4H, FcCH2), 7.02 (s, 2H, NCHCHN), 8.93 (s, 1H, NCHN). ¹³C NMR (δ, CDCl3, 75 MHz): 49.90, 69.44, 69.70, 70.08, 78.50, 120.86, 134.88. ESI-MS. Calcd for C₂₅H₂₅Fe₂N₂: m/z 465.0712. Observed: m/z 465.0709. Calcd. for: C₂₅H₂₅Fe₂N₂BF₄: C, 54.40, H, 4.57, N, 5.08; Found: C, 54.39, H, 4.78, N, 4.99.

[Bis-(1,3-di(ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tetrafluoroborate], [5][BF₄].

To a 40 mL vial containing a stir bar, [1,3-di(ferrocenylmethyl)imidazolium)][tetrafluoroborate] (0.2276 g, 0.4123 mmol) and NaHMDS (85.7 mg, 0.4661 mmol) were added followed by the addition of dry toluene (10 mL). The resulting mixture was then stirred for an hour. (C₄H₂Si)AuCl (62.3 mg, 0.194 mmol) and the solution was stirred for another hour. Volatiles were evaporated off in vacuo and the solids were dissolved in dry DCM (5 mL) and the solution was filtered through celite. The resulting homogenous solution was triturated with diethyl ether (25 mL) and washed with diethyl ether (3×5 mL). Yield: 32.9 % (164.6 mg). 1H NMR (δ, d-DMSO, 300 MHz): 4.21 (s, 28H, Fc), 4.42–4.40 (m, 8H, Fc), 5.18 (s, 8H, CH₂), 7.63 (s, 4H, FcNCH). ¹³C NMR (δ, CDCl3, 75 MHz): 51.04, 68.80, 68.92, 69.01, 81.78, 121.27, 181.72. ESI-MS. Calcd for C₅₀H₄₈Fe₄N₄Au: m/z 1125.0937. Observed: m/z 1125.0938. Anal. Calcd. for: C₅₀H₄₈AuFe₄N₄BF₄: C, 49.55, H, 3.99, N, 4.62; Found: C, 49.67, H, 4.25, N, 4.47.

[Bis-(1,3-di(ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tri-tetrafluoroborate], [6][BF₄]₅.

To a 20 mL vial containing a stir bar, [Bis-(1,3-di(ferrocenylmethyl)imidazole-2-ylidene)-gold(I)][tetrafluoroborate] (0.1978 g, 0.1632 mmol) and silver tetrafluoroborate (0.158 g, 0.816 mmol) were added followed by the addition of dry DCM (10 mL). The reaction mixture was stirred for 18 hours. The volatiles in the solution were removed in vacuo. Dry acetonitrile (5 mL) was added to the reaction mixture and stirred for 30 minutes. The reaction solution was filtered through celite. The acetonitrile was removed in vacuo and the product afforded was a dark green solid. Yield: 65.3 % (174.6 mg). HRMS (ESI) for [C₅₀H₄₈AuFe₄N₄]³⁺ [M]³⁺ Calcd. 1125.0915 Found. 1125.0928. Anal. Calcd. for: C₅₀H₄₈AuFe₄N₄B₅F₂₀(AgBF₄): C, 34.24, H, 2.76, N, 3.19; Found: C, 33.15, H, 3,02, N, 5.12.

Materials and methods

All synthetic manipulations were done in nitrogen-filled glove box (Inert Glove box system) unless otherwise noted. All glassware were oven-dried at 110 ^oC for 12 hours prior to use. 1-(ferrocenylmethyl)-3-mesitylimidazolium iodide,⁴⁴ [1, 3-bis (ferrocenylmethyl) imidazolium)][iodide],⁴⁰ and (C₄H₂S)AuCl were all prepared according to known literature procedures. Other reagents used such as sodium bis(trimethylsilyl)amide [((CH₃Si)₂N]Na (NaHMDS) and triethyloxonium tetrafluoroborate [(CH₃CH₂)₃O]BF₄ were purchased from commercial sources and used without further purification. Deuterated solvent d-chloroform (CDCl₃ 99.8%) was purchased from Acros Organics and dried using molecular sieves prior to the data collection. Dry solvents used were either purchased nitrogen flushed or dried using Inert (Inert Technologies) solvent system; prior to use, all solvents underwent freeze-pump-thaw cycles. These solvents include diethyl ether, toluene, dichloromethane, hexanes, and acetonitrile. The PureLab HE 2 GB Glovebox by Inert Technologies was used to carry out and work up reactions that require a nitrogen atmosphere. Glasswares used were first oven-dried at 115^oC for 12 hours before use.

Instrumentation

Ultraviolet-visible spectra were acquired at 25^°C using a Varian Cary 50 Bio UV-vis spectrophotometer with molar absorptivity recorded in M⁻¹ cm^-1. ¹H and ¹³C NMR spectra were collected using a Bruker 300 MHz NMR spectrometer. Residual solvents were used as references for internal standard for spectra, for ¹H NMR: CDCl₃, 7.26 ppm; for ¹³C NMR: CDCl₃, 77.16 ppm. Hertz (Hz) is used to show coupling constants. Midwest Micro-lab, LLC in Indianapolis, IN, performed all elemental analyses for the new compounds. Using a CHI620E electrochemical analyzer, measurements were obtained using a silver wire reference electrode, a platinum working electrode, and a platinum auxiliary electrode under a nitrogen atmosphere in a small sealed electrochemical cell. Electrochemical measurements were performed using 1.0 mM solutions of analyte in dry DCM with 0.1 M [N(nBu)₄][PF₆] as the electrolyte. Decamethylferrocene (Fc*) was used as the internal standard. 50 mV pulse amplitudes and 4 mV data intervals were used to collected differential pulse voltammetry measurements. Peak potentials listed were established by cyclic voltammetry (CV) at 100 mVs⁻¹ scan rates and the values were adjusted using decamethylferrocene potentials and reference to a saturated calomel electrode (SCE). High-resolution mass spectra were obtained using Q Exactive Focus Hybrid Quadrapole-Orbitrap Mass Spectrometer instrument and are reported as m/z. For single-crystal X-ray diffraction analysis, a suitable crystal with appropriate dimensions was mounted on a loop with Paratone-N oil and transferred to the goniostat bathed in a cold nitrogen stream. Intensity data were collected at 170 K on a standard Bruker DUO APEX-II equipped with a CCD detector with MoKα radiation (λ=0.71073Å) and a graphite-monochromator. Mössbauer spectra were collected at 4 K using a SEE Co. Mössbauer spectrometer with a Janis cryostat. The isomer shift is reported relative to that of α-Fe at room temperature. All spectral simulations were performed using the WMOSS software.⁴⁵

Conclusions

In conclusion, we have illustrated the novel synthesis of ferrocenium containing NHC-gold(I) complexes. This strategy would allow easy access to other ferrocenium-containing gold complexes. Hence holds promise for the exploration of a library of gold(I) complexes containing ferrocenium coupled NHCs towards toxicity studies in cancer lines. Collectively, UV-vis, Mössbauer, and X-ray diffraction measurements confirm that the first site of oxidation in [2][BF₄] and [5][BF₄] is ferrocene. This work also highlights that a significant increase in Fe-C_Cp bond length for oxidized [3][BF₄]₃ relative to its reduced form ( [2][BF₄]), which confirms the site of oxidation is ferrocene. This unambiguous assignment of the redox site may establish that the potential site of oxidation for these complexes in the biological media may very well be ferrocene. This study further corroborates the rationale that ferrocene to ferrocenium shuttle may be a possible pathway for ROS generation in the cell lines. Therefore, this finding is helpful in developing tailor-made cancer drugs containing ferrocene moiety that mainly targets ROS systems. Furthermore, the synthetic strategies reported herein will help develop similar oxidized complexes.