Reactions of Antitumor Active Dirhodium(II) Tetraacetate Rh2(CH3COO)4 with Cysteine and Its Derivatives.

We have combined results from several spectroscopic techniques to investigate the aerobic reactions of Rh2(AcO)4 (AcO- = CH3COO-) with l-cysteine (H2Cys) and its derivatives d-penicillamine (3,3'-dimethylcysteine, H2Pen), with steric hindrance at the thiol group, and N-acetyl-l-cysteine (H2NAC), with its amino group blocked. Previous investigations have shown that antitumor active dirhodium(II) carboxylates may irreversibly inhibit enzymes containing a thiol group at or near their active sites. Also, cysteine, the only thiol-containing proteinogenic amino acid, interacts in vivo with this class of antitumor compounds, but structural information on the products of such reactions is lacking. In the present study, the reactions of Rh2(AcO)4 and H2L were carried out in aqueous solutions at the pH of mixing (acidic) and at physiological pH, using the different mole ratios 1:2, 1:4, and 1:6, which resulted in the same products in increasing yields. Electrospray ionization mass spectrometry (ESI-MS) indicates formation of dimeric [RhIII 2Pen4]2- or oligomeric {RhIII 2L4} n (L = Cys, NAC) complexes with bridging thiolate groups. Analyses of Rh K edge extended X-ray absorption fine structure (EXAFS) data reveal 3-4 Rh-S and 2-3 Rh-(N/O) bonds around six-coordinated Rh(III) ions at mean distances of 2.33 ± 0.02 and 2.09 ± 0.02 Å, respectively. In the N-acetyl-l-cysteine compound, the RhIII···RhIII distance 3.10 ± 0.02 Å obtained from the EXAFS spectrum supports trithiolate bridges between the Rh(III) ions, as was also found when using glutathione as ligand. In the cysteine and penicillamine complexes, double thiolate bridges join the Rh(III) ions, with the nonbridging Cys2- and Pen2- ligands in tridentate chelating (S,N,O) mode, which is consistent with the ΔδC = 7.3-8.4 ppm shift of the COO- signal in their carbon-13 cross polarization magic angle spinning (CPMAS) NMR spectra. For the penicillamine complex, the 2475.6 eV peak in its S K edge X-ray absorption near edge structure (XANES) spectrum shows partial oxidation, probably caused by peroxide generated from reduction of dissolved O2, of thiolato to sulfenato (S=O) groups, which were also identified by ESI-MS for all three {RhIII 2L4} n compounds.

Introduction

The antitumor activity of neutral dirhodium(II) complexes with bridging carboxylate groups in a cage structure, Rh2(μ-RCOO)4 (R = CH3, C2H5, and C3H7), was initially reported in 1972 by Bear and co-workers.[1−3] They observed that the survival time of mice bearing Ehrlich ascites or leukemia L1210 tumors increased when they were treated with dirhodium(II) carboxylates. The treatment was also effective against leukemia P388 and sarcoma 180 tumors. This class of compounds could inhibit in vivo DNA synthesis of Ehrlich ascites tumor cells, and in vitro DNA and RNA polymerase from Escherichia coli.[4−6] The metabolism of dirhodium(II) tetraacetate was explored by injecting 14C-labeled Rh2(AcO)4 in tumor-bearing Swiss mice, which exhaled 14CO2 within 2 h after injection, indicating breakdown of the cage structure into carboxylate and rhodium ions. The rhodium was mainly deposited in the liver, with only ∼5% excreted through urine within the first 24 h after administrating the drug.[7] A similar structural breakdown was observed when reacting dirhodium(II) tetraacetate (1, Rh2(AcO)4) with cysteine in the mole ratio 1:4 at pH = 7.5, which resulted in an insoluble Rh(III)–cysteine complex, protons, and free acetate groups.[7,8] In vitro studies showed that human serum albumin maintained the cage structure of Rh2(AcO)4 by binding to its axial positions via the N-imidazole moieties of its histidine residues.[9] Although enzymes containing thiol (−SH) groups at or near their active site were irreversibly inhibited by dirhodium(II) carboxylates, other enzymes without free −SH groups, or those containing thiol groups away from their active site, were not affected.[8,10] Inhibition was proposed to involve reactions of protein–thiol groups with dirhodium(II) carboxylates, leading to breakdown of their cage structure, oxidation of Rh(II), and tight binding of Rh(III) ions to the active site of those enzymes. The results were then correlated to the limited excretion of rhodium from mice treated with dirhodium(II) tetraacetate, indicating that rhodium probably binds to thiol-containing proteins at the cell membrane.[7,8] Bear et al. suggested that the antitumor efficiency of this class of dirhodium(II) carboxylate compounds could be enhanced by increasing their resistance toward reaction with −SH groups, for example, by blocking the axial positions (assumed as the initial −SH coordination sites) with stronger ligands such as polyadenylic acid.[11] They also proposed that as a detoxification step after administering a large dose of a dirhodium(II) carboxylate drug, treatment with a nontoxic thiol-containing molecule such as glutathione should be considered.[11]

In 1980, Pneumatikakis and Psaroulis reported the reactions of [Rh2(AcO)4(CH3OH)2] with cysteine, penicillamine, or cysteine methylester (1:4 mole ratio), which produced precipitates characterized as mononuclear paramagnetic square-planar Rh(II) complexes with (S,N)-coordinated ligands.[12]

Two decades later, the reactions of [Rh2(AcO)2(N–N)2(CH3CN)2]X2 {N–N = 1,10-phenantroline (phen) or 2,2′-bipyridine (bpy), 2; X = PF6 or BF4, respectively}, among other Rh complexes,[13] were investigated with 2-aminothiophenol (at mole ratios of 1:10 and 1:4) under aerobic conditions as models for how antitumor active dirhodium(II) bis-polypyridine compounds react with thiol-containing biomolecules. Reaction products included mononuclear [RhIII(S,N-C6H6NS)(S-C6H6NS)2(phen)] or binuclear [RhIII(μ-S,N-C6H6NS)(η1-S-C6H6NS)(bpy)]22+ (3a) with mononuclear [RhIII(S,N-C6H6NS)2(bpy)]+ (3b) cocrystallized in the same unit cell (Scheme ).[14] With excess sodium benzene-thiolate (Na+C6H5S–) in dry degassed methanol, the bipyridine complex 2 generated the dark red binuclear [RhII(μ-S-C6H5S)(η1-S-C6H5S)(bpy)]2·CH3OH crystalline complex (3′) with two bridging and two terminal thiolates.[15] This complex was considered as a model for a reactive intermediate that, in the presence of oxygen, could produce Rh(III) species that were similar to those formed with 2-aminothiophenol.[14] Dark blue precipitates were obtained by reacting 2 with cysteine and glutathione (at a mole ratio of 1:4) under anaerobic conditions, but these were not structurally characterized.[15]

Scheme 1

Structures of [RhII2(AcO)4] (1), [RhII2(AcO)2(bpy)2(CH3CN)2]2+ (2), and Its Reaction Products with Thiolates: [RhII(μ-S-C6H5S)(η1-S-C6H5S)(bpy)]2 (3′), [RhIII(μ-S,N-C6H6NS)(η1-S-C6H6NS)(bpy)]22+ (3a), and [RhIII(S,N-C6H6NS)2(bpy)]+ (3b) Reported in Refs (13−15)

Here, we report the results of our characterization of the reaction products from Rh2(AcO)4 (1) and the thiol-containing amino acids l-cysteine (H2Cys), d-penicillamine (3,3′-dimethylcysteine, H2Pen), and N-acetyl-l-cysteine (H2NAC), using electrospray ionization mass spectrometry (ESI-MS), solid state carbon-13 cross polarization magic angle spinning (CPMAS) NMR, and X-ray absorption spectroscopy (XAS). The choice of these biologically relevant ligands was based on their potential to form S,N-chelates (H2Cys and H2Pen), or to act solely as a monodentate thiol ligand (H2NAC). Moreover, penicillamine probes the influence of increased steric hindrance in the vicinity of the thiol group. The results reveal the nature and structure of the decomposition products that were previously reported but not identified,[7,8] and increase our understanding of the fate of dirhodium(II) carboxylates in vivo in the presence of thiol-containing proteins and enzymes.

Experimental Section

Sample Preparation

l-Cysteine, d-penicillamine, N-acetyl-l-cysteine, and sodium hydroxide were purchased from Sigma-Aldrich, and dirhodium(II) tetraacetate (pure; 47% Rh) was obtained from Pressure Chemical Company; all were used without further purification. Degassed water and D2O were prepared by bubbling argon through boiled distilled water or D2O for 3 h until they had cooled down to room temperature. Sephadex G-15 was obtained from VWR International.

A calibrated Thermo Scientific Orion Star semi-micro electrode was used for pH measurements. Thermogravimetry (TG), to measure water content, was carried out using a Netzsch STA 409 instrument, and a Johnson Matthey magnetic susceptibility balance was used to analyze the magnetic properties of the solid samples.

Solid {Rh2III(HCys)2(Cys)2·4H2O} (4)

To a completely dissolved solution of dirhodium(II) tetraacetate (0.1123 g, 0.254 mmol) in 40 mL of oxygen-free water, solid cysteine (0.1234 g, 1.018 mmol) was added under a stream of argon to prevent oxidation of the ligand to cystine. The solution color gradually changed from emerald green to orange-brown and a precipitate was formed (pH = 3.2). The mixture was stirred for 24 h at room temperature under aerobic conditions. The precipitate was filtered, washed with methanol and ether several times and dried under vacuum in a desiccator. Elemental analysis: calcd for {Rh2III(HCys)2(Cys)2·4H2O}, (Rh2C12H30O12S4N4): %C 19.05, %H 4.00, %N 7.41 (9.5% H2O); found: %C 19.30, %H 4.06, %N 6.86 (TG: 9.9% H2O); yield = 91%. Magnetic susceptibility measurements showed the solid bulk, which is insoluble in solvents, to be diamagnetic.

Aerobic Reaction of Rh2(AcO)4 with Thiol-Containing Ligands H2L (pH = 7.4)

An emerald green suspension of 0.100 g (0.226 mmol) of dirhodium(II) tetraacetate in 30 mL of degassed water was added dropwise to colorless solutions containing one of the following H2L ligands (0.909 mmol of H2Cys, or 0.905 mmol of H2Pen, or 0.906 mmol of H2NAC) dissolved in 90 mL of degassed water. The reactions were carried out under a stream of argon to prevent formation of disulfides by oxidation of the ligand −SH group. When adding 1.0 M NaOH dropwise to adjust the pH of the solutions to 7.4, their color turned to dark red/orange. The reaction mixtures were stirred at room temperature (RT) for 2 h under argon flow, and then exposed to air for another 46 h, while regularly adjusting the pH by adding 1.0 M NaOH. ESI-mass spectra were measured at different time intervals (30 min, 2 h, 24 h, and 48 h after pH adjustment). A rotary evaporator was used (at RT) to reduce the volume before passing the solution through a Sephadex G-15 size exclusion chromatography column with water as eluent. ESI-mass spectra were measured for different parts of the orange band, which was separated and concentrated. A portion of the concentrated solution was set aside for extended X-ray absorption fine structure (EXAFS) spectroscopic measurement, and the remaining part was evaporated to dryness and further dried in a desiccator under vacuum. Elemental analyses of the solid products from the three reactions are as follows:

Calcd for {Na2[Rh2(Cys)4]·5H2O} (Rh2C12H30N4O13S4Na2) (5): %C 17.61, %H 3.69, %N 6.85 (11.00% H2O); found: %C 17.93, %H 3.69, %N 6.82 (TG: 10.57% H2O); yield = 72%. Calcd for Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O (Rh2C20H45N4O14.5S4Na2) (6): %C 25.19, %H 4.76, %N 5.87 (8.50% H2O); found: %C 25.22, %H 4.72, %N 5.76 (TG: 8.12% H2O); yield = 67%. Calcd for {Na2[Rh2(NAC)4]·4.5H2O} (Rh2C20H37N4O16.5S4Na2) (7): %C 24.57, %H 3.81, %N 5.73 (8.29% H2O); found: %C 24.21, %H 3.41, %N 5.45 (TG: 7.99% H2O); yield = 76%.

These solid samples, which were all diamagnetic and water soluble (only partially soluble in dimethyl sulfoxide and insoluble in other common solvents), were used for 13C CPMAS NMR, ESI-MS, electron paramagnetic resonance (EPR), Rh K edge EXAFS, and S K edge X-ray absorption near-edge structure (XANES) spectroscopic measurements.

NMR Spectroscopy

Carbon-13 cross-polarization magic angle spinning (CPMAS) NMR spectra were measured for solid compounds at room temperature using a Bruker Avance 500 MHz spectrometer with high power proton decoupling, setting α-glycine carbonyl at 176.50 ppm as reference.[16] The ground solids 1, 4–7, and the H2L compounds used as ligands were packed separately into a 4 mm zirconia rotor spinning at an magic angle spinning rate of 12 kHz, collecting ∼100 scans with a 10 s recycle delay for each sample.

ESI-Mass Spectrometry

An Agilent 6520 Q-ToF instrument was used to measure ESI-mass spectra both in positive (+) and negative (−) ion modes. Voltages for capillary, skimmer, and fragmentor were set at 4000, 65, and 80 V, respectively. Solid samples were dissolved in water, and methanol was used as the mobile phase with a continuous injection flow rate of 0.2 mL min–1 and a drying gas flow rate of 7 L min–1 at 200 °C. For voltage variable measurements, the fragmentor voltage varied between 0 and 80 V. The nature of the mass ions and their corresponding m/z values were confirmed using the isotope distribution calculator from Agilent, and a high resolution calculation method chosen from Scientific Instrument Services (SIS).[17]

Electronic Spectroscopy

UV–vis absorption spectra were measured at RT using a Cary 300 UV–vis double-beam spectrophotometer. Samples were measured in quartz cells with 1 mm path-length, using water in the reference position.

X-ray Absorption Spectroscopy Data Collection

Rh K edge EXAFS spectra of the solid {Rh2III(Cys)2(HCys)2·4H2O} (4) precipitated at pH = 3.2, the concentrated solutions (pH = 7.4), and corresponding solid compounds {Na2[Rh2(Cys)4]·5H2O} (5), Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O (6), and {Na2[Rh2(NAC)4]·4.5H2O} (7) were measured at room temperature at BL 7–3 (500 mA) at the Stanford Synchrotron Radiation Lightsource (SSRL) operating under 3 GeV. Higher-order harmonics were rejected by detuning the Si(220) (ϕ = 0°) double-crystal monochromator to 50% of maximum intensity at the end of the Rh K edge scan range. The X-ray energy was internally calibrated by assigning the first inflection point of the absorption edge of a Rh foil placed between the ion chambers I1 and I2 to 23 219.80 eV. Three ion chambers (I0, I1, and I2) were filled with nitrogen (N2). Four to ten scans were collected for each sample in transmission mode and compared prior to averaging to ensure that no radiation damage occurred during measurement.

S K edge XANES spectra of the solid compounds 5–7 were measured at BL 4–3 (SSRL), equipped with a Si(111) double-crystal monochromator, a harmonic rejection mirror, and a Vortex fluorescence detector. The solid samples were finely ground, dusted on Mylar tape, and placed in a sample chamber filled with helium. Two scans collected in fluorescence mode were averaged. Energy calibration was achieved by setting the first peak maximum in the S XANES spectrum of Na2S2O3·5H2O to 2472.02 eV.

XAS Data Analysis

The WinXAS 3.1 program[18] was used for first order polynominal pre-edge background subtraction and normalization of the Rh K edge XAS spectra, followed by subtraction of a seven segment cubic spline in the post-edge region to extract the EXAFS oscillations. The first inflection point of the absorption edges varied over a narrow range (E0 = 23 226.3–23 226.7 eV). To simulate theoretical EXAFS oscillations, the atomic coordinates of ΔΛ-[Rh{Ir(aet)3}2]3+ (Haet = 2-aminoethanethiol) with the Cambridge Structural Database (CSD) code AQUVOX,[19] were used for the ATOMS and Feff 7.0 programs,[20,21] replacing Ir with Rh, as the ionic radii of Rh3+ and Ir3+ ions are very similar (0.665 and 0.68 Å, respectively).[22]

Least-squares curve-fitting of the theoretically simulated EXAFS oscillations to the experimental EXAFS spectra was performed over the k-range ∼2.8–18.2 Å–1. The wide k-range EXAFS data of good quality allowed simultaneous refinement of the coordination numbers (CN), bond distances (R), and the Debye–Waller factor parameters (σ2), unless otherwise stated. The ΔE0 was allowed to float as a common value for all scattering paths involved in the fitting. The amplitude reduction factor (S02) was set at 0.92, as obtained from the EXAFS data analysis of solid Rh2(AcO)4.[23] The accuracy of the bond distances, σ2 (mean square of the variation around the mean distance), and refined CN values is within ±0.02 Å, ±0.001 Å2, and ±10–15%, respectively. The residual () from the least-squares curve-fitting in k-space is defined aswhere yexp and ytheo denote experimental and theoretical data points, respectively.

Results

Dirhodium(II) tetraacetate, Rh2(AcO)4 (1), has limited solubility in water. When a completely dissolved aqueous solution of 1 (emerald green) is mixed with a fourfold amount of cysteine under aerobic conditions, a red-brown precipitate is formed after 5 min (pH = 3.2), whereas similar reactions with N-acetylcysteine or penicillamine result in clear emerald green solutions. When adjusting the pH to 7.4, the reactions proceed quickly with a color change to orange-red (see Experimental Section). To monitor the early stages of the reaction for the solutions containing N-acetylcysteine or penicillamine, we measured, over a period of time, their ESI-MS, UV–vis, and/or 1H NMR spectra at the pH of mixing (acidic) at which the reaction is relatively slow. The majority of these results are presented in Appendices 1 and 2 in the Supporting Information (Figures S-1–S-5, Tables S-1, and S-2), including the ESI-mass spectra of pure Rh2(AcO)4 in water (Figure S-6).

The UV–vis spectrum of a solution containing Rh2(AcO)4 and H2NAC (C = 1.0 mM, mole ratio of 1:4) at the pH of mixing (2.8) initially displayed a broad peak around 340 nm, which gradually red-shifted with time while gaining intensity, with its highest intensity at λmax = 367 nm observed after 48 h (see Figure , left). ESI-mass spectra of the same UV–vis samples showed peaks associated with [Rh2III(AcO)4(HNAC)2 ± H+]+/– (m/z = 764.90 and 766.91) and the corresponding Na+ mass ions (+m/z = 788.90 and 810.88) as well as intense peaks assigned to unreacted Rh2(AcO)4 (+m/z = 459.90, 464.85, and 505.88); see Figures (right) and S-3, and Tables and S-1. No mass peak was related to the mononuclear Rh(III) complex. When the pH of the solution mixture was adjusted to 7.4 after 48 h, the peak intensity at λmax = 367 nm substantially decreased within a few minutes. This peak does not appear in the UV–vis spectra when adjusting the reaction pH to 7.4 directly after mixing the reactants (see Figure S-7).

Figure 1

(Left) UV–vis spectroscopy of reaction progress for Rh2(AcO)4 with H2NAC (mole ratio of 1:4, C = 1.0 mM, pH of mixing = 2.8); peak positions are obtained from second derivatives. (Right) ESI-mass spectrum (+ ion mode) of the same solution measured after 48 h; see Table for peak assignments.

Table 1

Assignment of the ESI-MS Peaks Shown in Figure (Right)a

m/z (M)	isotopic pattern	assignment	m/z (M)	isotopic pattern	assignment
186.02	M + 1	[H2NAC + Na+]+	627.88	M + 1	[2RhII + 4AcO– + H2NAC + Na+]+
325.05	M + 1	[2H2NAC – 2H+ + H+]+	671.08	M + 1	[4H2NAC – 4H+ + Na+]+
347.03	M + 1	[2H2NAC – 2H+ + Na+]+	766.91	M + 1	[2RhIII + 4AcO– + 2HNAC– + H+]+
459.90	M + 1	[2RhII + 4AcO– + NH4+]+	788.90	M + 1	[2RhIII + 4AcO– + 2HNAC– + Na+]+
464.85	M + 1	[2RhII + 4AcO– + Na+]+	810.88	M + 1	[2RhIII + 4AcO– + 2HNAC– – H+ +2Na+]+
505.88	M + 1	[2RhII + 4AcO– + CH3CN + Na+]+

H2NAC = C5H9NO3S; 2H2NAC – 2H+ is the oxidized form of N-acetylcysteine with an S–S bond; AcO– = CH3COO–.

Reaction of Rh2(AcO)4 with Thiols at Physiological pH

The results presented below refer to the aerobic reactions of Rh2(AcO)4 with thiol-containing ligands (H2L) in the mole ratio of 1:4. Other mole ratios (1:2 and 1:6) led to the same products (5–7) in different yields (44–49 and 77–96% for mole ratios of 1:2 and 1:6, respectively); see Appendix 3 in the Supporting Information, where an image of the size exclusion chromatography purification column is shown for a solution containing Rh2(OAc)4 and H2NAC (mole ratio of 1:4).

ESI-Mass Spectrometry

The most intense peaks in the ESI-mass spectra (− ion mode, fragmentor voltage 80 V) of the products obtained from the reactions of 1 with cysteine (5), penicillamine (6), or N-acetylcysteine (7) at pH = 7.4 were associated with [RhIII2L4]2– ions that had satellite peaks with (−)m/z +8 and +16; see Figure and Table . In all mass ions, the oxidation state of the rhodium ions is +3, except for [RhII2(NAC)3]2– (−m/z = 344.43). On the basis of the observed isotopic patterns, tri-, tetra-, or hexanuclear mass ions were also detected, but no mononuclear rhodium mass ion was observed in the spectra shown in Figure . ESI-mass spectra of reaction mixtures with Rh2(AcO)4/H2L in a mole ratio of 1:2 showed the same mass ions, together with large peaks related to unreacted Rh2(AcO)4 (see Appendix 3 in the Supporting Information for H2Pen reaction).

Figure 2

ESI-mass spectra (− ion mode) of the products (in water) obtained from the reactions of Rh2(AcO)4 with cysteine (5, top), penicillamine (6, middle), and N-acetylcysteine (7, bottom) at a mole ratio of 1:4 at pH = 7.4 (fragmentor voltage 80 V). For peak assignments, see Table .

Table 2

Assignment of the Mass Ions Observed in the ESI-Mass Spectra (− Ion Mode) of Products 5–7 Shown in Figure a

m/z (M)	isotopic pattern	assignment	m/z (M)	isotopic pattern	assignment
Rh–Cysteine Product (5)
340.91	M + 0.5	[2RhIII + 4H2Cys – 8H+]2–	511.87	M + 0.5	[3RhIII + 6H2Cys – 11H+]2–
	M + 0.33	[3RhIII + 6H2Cys – 12H+]3–	519.87	M + 0.5	[3RhIII + 6H2Cys – 11H+ + O]2–
348.91	M + 0.5	[2RhIII + 4H2Cys – 8H+ + O]2–	682.83	M + 0.5	[4RhIII + 8H2Cys – 14H+]2–
356.91	M + 0.5	[2RhIII + 4H2Cys – 8H+ + 2O]2–		M + 1	[2RhIII + 4H2Cys – 7H+]−
409.42	M + 0.5	[2RhIII + 5H2Cys – 8H+ + O]2−	690.83	M + 0.5	[4RhIII + 8H2Cys – 14H+ + O]2–
417.42	M + 0.5	[2RhIII + 5H2Cys – 8H+ + 2O]2–	698.83	M + 0.5	[4RhIII + 8H2Cys – 14H+ + 2O]2–
454.89	M + 0.33	[4RhIII + 8H2Cys – 15H+]3–		M + 1	[2RhIII + 4H2Cys – 7H+ + O]−
Rh–Penicillamine Product (6)
339.44	M + 0.5	[2RhIII +3H2Pen + S2– – 6H+]2–	794.96	M + 0.5	[4RhIII + 8H2Pen – 14H+]2–
347.44	M + 0.5	[2RhIII + 3H2Pen + S2– – 6H+ + O]2–		M + 1	[2RhIII + 4H2Pen – 7H+]−
355.44	M + 0.5	[2RhIII + 3H2Pen + S2– – 6H+ + 2O]2–	810.95	M + 0.5	[4RhIII + 8H2Pen – 14H+ + O]2–
396.98	M + 0.5	[2RhIII + 4H2Pen – 8H+]2–		M + 1	[2RhIII + 4H2Pen – 7H+ + O]−
404.97	M + 0.5	[2RhIII + 4H2Pen – 8H+ + O]2–	816.94	M + 1	[2RhIII + 4H2Pen – 8H+ + Na+]−
412.97	M + 0.5	[2RhIII + 4H2Pen – 8H+ + 2O]2–	826.95	M + 1	[2RhIII + 4H2Pen – 7H+ + 2O]−
Rh–N-acetylcysteine Product (7)
323.04	M + 1	[2H2NAC – 3H+]−	637.91	M + 0.5	[3RhIII + 6H2NAC – 11H+]2–
344.43	M + 0.5	[2RhII + 3H2NAC – 6H+]2–		M + 0.25	[6RhIII + 12H2NAC – 22H+]4–
424.93	M + 0.5	[2RhIII + 4H2NAC – 8H+]2–	648.90	M + 0.5	[3RhIII + 6H2NAC – 12H+ + Na+]2–
	M + 0.33	[3RhIII + 6H2NAC – 12H+]3–		M + 0.25	[6RhIII + 12H2NAC – 24H+ + 2Na+]4–
432.93	M + 0.5	[2RhIII + 4H2NAC – 8H+ + O]2–	850.87	M + 1	[2RhIII + 4H2NAC – 7H+]−
506.45	M + 0.25	[2RhIII + 5H2NAC – 10H+]4–		M + 0.5	[4RhIII + 8H2NAC – 14H+]2–
512.57	M + 0.33	[4RhIII + 7H2NAC – 15H+]3–		M + 0.33	[6RhIII + 12H2NAC – 21H+]3–
556.39	M + 0.5	[3RhIII + 5H2NAC – 11H+]2–	861.87	M + 0.5	[4RhIII + 8H2NAC – 15H+ + Na+]2–
	M + 0.25	[6RhIII + 10H2NAC – 22H+]4–	872.86	M + 1	[2RhIII + 4H2NAC – 8H+ + Na+]−
566.92	M + 0.33	[4RhIII + 8H2NAC – 15H+]3–		M + 0.5	[4RhIII + 8H2NAC – 16H+ + 2Na+]2–
621.26	M + 0.33	[4RhIII + 9H2NAC – 15H+]3–		M + 0.33	[6RhIII + 12H2NAC −24H+ + 3Na+]3–

H2Cys = C3H7NO2S; H2Pen = C5H11NO2S; H2NAC = C5H9NO3S; 2H2NAC – 2H+ is the oxidized form of N-acetylcysteine with a S–S bond.

When setting the fragmentor voltage at 0 V, the main mass peaks observed for the Rh–penicillamine reaction product 6 were those of [RhIII2(Pen)4 + H+]− (−m/z = 794.96), [RhIII2(Pen)4 + H+ + O]− (−m/z = 810.95), and [RhIII2(Pen)4 + H+ + 2O]− (−m/z = 826.95), indicating that the dinuclear species dominate in 6; see Figure S-8. For the Rh–N-acetylcysteine reaction product 7, the majority of the mass peaks at 0 V were those of the tetranuclear RhIII4(NAC)8–10 species with an added O atom or H+/Na+ ions, with only a few peaks detected for the dinuclear RhIII2(NAC)4 species (see Figure S-9 and Table S-3).

The lowest fragmentor voltage that could generate mass peaks for the Rh–cysteine reaction product 5 was 10 V; however, no isotopic pattern could be observed (Figure S-10 and Table S-4). Similar to Rh–penicillamine 6, peaks were assigned to the dinuclear species [RhIII2(Cys)4 + H+]− (−m/z = 682.83), [RhIII2(Cys)4 + H+ + O]− (−m/z = 698.83), and [RhIII2(Cys)4 + H+ + 2O]− (−m/z = 714.81), assuming the charge −1. In addition, two less intense peaks were detected at (−)m/z = 803.84 and 819.85 for [RhIII2(Cys)5 + 3H+]− and [RhIII2(Cys)5 + 3H+ + O]− ions, respectively. Because the reaction of Rh2(AcO)4 with cysteine was carried out at the mole ratio 1:4, the presence of such a Rh/Cys 2:5 complex may indicate fragmentation of higher complexes, for example with a Rh/Cys 4:8 ratio, as observed in the ESI-mass spectrum measured at 80 V (see Table ).

In summary, analyses of the ESI-mass spectra show formation of a dinuclear [RhIII2(Pen)4]2– complex in 6, dinuclear [RhIII2(Cys)4]2– or tetranuclear {RhIII2(Cys)42–}2 complexes in 5, and oligomeric {RhIII2(NAC)42–} species in 7 (with 1–2 added O atoms), and exclude formation of mononuclear Rh complexes.

Sulfur K Edge XANES Spectroscopy

To examine whether the detected mass ions with 1–2 added O atoms correspond to sulfenato (S=O) or sulfinato (O=S=O) groups, S K edge XANES spectra of the Rh–cysteine (5), Rh–penicillamine (6), and Rh–N-acetylcysteine (7) solid reaction products were measured. The spectra and their corresponding smoothed second derivatives are shown in Figure . There are four distinct absorption features for 6, and only two absorption peaks at 2472.0 and ∼2474.0 eV for 5 and 7.

Figure 3

Sulfur K edge XANES spectra (left) and the corresponding smoothed second derivatives (right) of the solid products obtained from the reactions of Rh2(AcO)4 with cysteine (5), penicillamine (6), or N-acetylcysteine (7) at a mole ratio of 1:4 at pH = 7.4.

In the spectrum of the nickel(II) cysteinate complex with 3d8 low-spin configuration and square-planar coordination, K2[Ni(S,N-Cys)2], the distinct pre-edge feature observed at 2471.1 eV and the main peak at 2472.8 eV correspond to S 1s transitions to an unoccupied frontier orbital with S 3p and Ni 3d character, and to a σ*(C–S) orbital, respectively.[24] For the Rh–cysteine reaction product 5 with Rh(III) low-spin d6 configuration, it seems likely that the peaks at 2472.0 and 2474.0 eV correspond to the S(1s) → lowest unoccupied molecular orbital (LUMO) (with S(3p) and Rh(4d) character) and S(1s) → σ*(C–S) transitions, respectively. The additional feature appearing in the spectrum of the Rh–penicillamine reaction product 6 at 2475.6 eV (less pronounced or absent for 5 or 7) is probably due to a sulfenato (S=O) group bound to the Rh(III) ion, that is, a SS=O (1s) → σ* (C–SS=O) transition. The assignment is based on the peak at 2475.9 eV in the S XANES spectrum of methionine sulfoxide.[25] The low-energy shoulder at 2470.6 eV in the spectrum of 6 can then be tentatively assigned to SS=O (1s) → LUMO (with SS=O (3p) and Rh(4d) character).[24,26]

Rh K Edge EXAFS Spectroscopy

Because no single crystals could be prepared, we used Rh K edge X-ray absorption spectroscopy to determine the local structure, that is, nature/number of the nearest neighbors and bond distances, around the Rh(III) ions in the solid reaction products 5–7. The k3-weighted Rh K edge EXAFS spectra and the corresponding Fourier transforms (FTs) for the solids 5–7 are shown in Figure with the least-squares curve-fitting results presented in Table . For comparison, EXAFS data for a concentrated solution of the related Rh–glutathione compound 8, {Na2[RhIII2(HA)4]·7H2O} (H3A = glutathione), have also been included.[23] For all three H2L ligands, the EXAFS spectra collected from the solid Rh products (5–7) overlapped with those of the corresponding solution (Figure S-11), whereas the EXAFS spectra of 5–8 did not completely match (Figure S-12).

Figure 4

k3-Weighted Rh K edge EXAFS spectra (left) and the corresponding Fourier transforms (right) of the solid products obtained from the reactions of Rh2(AcO)4 with cysteine (5), penicillamine (6), and N-acetylcysteine (7) at a mole ratio of 1:4 at pH = 7.4, compared to those of a concentrated solution of the dinuclear Rh-glutathione reaction product (8) (see Table ).

Table 3

Least-Squares Curve-Fitting Results for EXAFS Spectra of the Solid Reaction Products 5–7 and a Concentrated Solution of the Rh–Glutathione Reaction Product 8a,b

	Rh–(N/O)			Rh–S			Rh···Rhc
sample	CN	R (Å)	σ2 (Å2)	CN	R (Å)	σ2 (Å2)	CN	R (Å)	σ2 (Å2)	ΔE0	d
5	2.0	2.10	0.0037	4.1	2.34	0.0044	2.0	3.44	0.0142	0.7	14.7
6	3.2	2.10	0.0039	3.2	2.315	0.0042	0.77	3.03 f	0.0078	–0.4	18.5
7	2 f	2.09	0.0032	4.6	2.32	0.0050	0.80	3.10	0.0047	–0.2	17.5
8	2.5	2.08	0.0034	4.1	2.33	0.0050	0.85	3.11	0.0046	1.2	17.6

See Figure .

Amplitude reduction factor (S02) = 0.92 fixed;[23]f = fixed value; estimated errors: R ± 0.02 Å; σ2 ± 0.001 Å2; CN ± 10–15%.

Uncertain values for 5 and 6 (see text).

= fitting residual (%).

The curve-fitting results in k-space revealed that the mean Rh–(N/O) and Rh–S distances vary over a narrow range of 2.08–2.10 and 2.315–2.34 Å, respectively, with reasonable σ2 values (0.003–0.005 Å2). The Rh···Rh scattering path included in the fitting models for the Rh–N-acetylcysteine reaction product 7 resulted in the Rh···Rh distance 3.10 ± 0.02 Å (σ2 = 0.0047 ± 0.001 Å2), which is similar to that of the Rh–glutathione reaction product 8.[23] For the Rh–cysteine (5) and Rh–penicillamine (6) reaction products, the introduction of a Rh···Rh scattering path to explain the small FT features at ∼3.0 and ∼2.8 Å, respectively, did not yield conclusive results for such a long and diffuse interaction (see Table S-5).

13C NMR Spectroscopy

To find out whether the carboxyl group in the thiol-containing ligands is bound to the Rh(III) ions, carbon-13 cross-polarization magic angle spinning (CPMAS) NMR spectra of the solids 4–7 were measured, and are compared with those of the pure ligands and Rh2(AcO)4 in Figure . We previously assigned the peaks for 0.1 M alkaline solutions of these ligands (calibrated relative to the CH3OH signal at 49.15 ppm).[27−29] The sets of doublets observed for solid d-penicillamine can be attributed to polymorphs, that is, two nonequivalent molecules in the crystal asymmetric unit,[30,31] although only the structure of crystalline racemic dl-penicillamine is known.[32] Similarly, the crystal structure has been reported for dihydrated Rh2(AcO)4,[33] but not for anhydrous Rh2(AcO)4. The broad peaks at δCOO = 191.7 and 193.3 ppm observed for Rh2(AcO)4 powder could be due to the presence of two different acetate groups, with O atoms of some acetate ligands in one molecule coordinated to the axial positions of another molecule, as in crystalline Rh2(CF3COO)4.[34,35]

Figure 5

Solid state 13C CPMAS NMR spectra obtained for Rh2(AcO)4 and its reaction products at a mole ratio of 1:4 with the ligands cysteine (4, pH = 3.2; 5, pH = 7.4), penicillamine (6), and N-acetylcysteine (7), and for the pure solid ligands.

The features in the 13C NMR spectra of the diamagnetic solids 4–7 are broad. However, an EPR spectrum of 7 showed only trace amounts of paramagnetic species, making this an unlikely cause of the broadness (see EPR Spectroscopy section in the Supporting Information).[36] Although only one set of broad 13C NMR features is observed for the N-acetylcysteinate compound 7, there are two sets of NMR signals for 4–6, probably due to two types of binding modes for the coordinated cysteinate and penicillaminate ligands. Also, different stereoisomers could lead to some splitting of the 13C NMR resonances into different sets, as for the [CoIII{Rh(N,S-Cys)3}2]3– complex with three possible isomers (ΔΔ, ΛΛ, ΔΛ) with slightly different chemical shifts.[37] However, the 13C NMR signals for 4–6 in Figure are too broad to recognize separate stereoisomers.

The 13C NMR resonances associated with the carboxyl groups in the H2L ligands (173.6–175.4 ppm) are generally deshielded in the spectra of their Rh(III) compounds (4–7). For the Rh–penicillamine reaction product 6 there are two NMR resonances at 176.1 and 181.4 ppm, with the latter ∼7.3 ppm deshielded relative to that of the pure ligand (average ∼ 174.1 ppm). Also, for the Δ-fac-H3[Rh(N,S-Cys)3] complex crystallized from a HCl(aq) solution, the chemical shift δC = 181.0 ppm has been reported,[37] with the deshielding probably due to weak forces between −COO– and the counter H+ ions.

Previously, we observed a similar shift of Δδ (13C) = 6.0 ppm for the carboxylate group in a Pb(II)–penicillamine solution (CPb(II) = 10 mM, CH = 20 mM, pH = 9.6), which was then attributed to the tridentate coordination of a (S,N,O)-Pen2– ligand to the Pb(II) ion.[27] Therefore, we assign the resonance at 181.4 ppm to a (S,N,O)-Pen2– ligand bound to a Rh(III) ion in 6. The other peak at 176.1 ppm is shifted only ∼2.0 ppm downfield relative to penicillamine, which indicates a penicillaminate ligand with a COO– group that is not bound to Rh(III). The small shift could be due to ionic interactions of the COO– group.

For the Rh–cysteine reaction product 4 that precipitated at pH = 3.2, there are three peaks in the carboxylate region at 174.0, 182.0, and 194.3 ppm. The latter peak was attributed to a protonated, noncoordinated carboxyl group (−COOH). This peak, which disappears in the Rh–cysteine reaction product 5 at pH = 7.4, is ∼1.0 ppm deshielded relative to that of Rh2(AcO)4 at δC = 193.3 ppm, and could not be related to unreacted dirhodium(II) tetraacetate. The peak at 182.0 ppm (182.3 ppm in 5), which is shifted downfield by ∼8.4 ppm relative to that of cysteine (173.6 ppm), could again be related to a tridentate (S,N,O)-Cys2– ligand bound to the Rh(III) ion. This peak has lower intensity relative to the corresponding resonance in the spectrum of the Rh–penicillamine reaction product 6 (181.4 ppm). The more intense peak at 174.0 ppm (176.3 ppm in 5) is probably related to a cysteinate ligand with an uncoordinated COO– group.

Discussion

Aerobic Reaction of 1 with Penicillamine

Within the first few hours after mixing the reagents Rh2(AcO)4 (1) and penicillamine at a mole ratio of 1:4 (initial pH 4.1), the species [Rh2II(AcO)4(H2Pen)1–2] (Scheme a) and [Rh2II(AcO)3(HPen)] (+ or −H+) with a Rh2(II, II) core could be detected with ESI-MS. Oxidation of the rhodium ions was observed when two deprotonated penicillamine ligands were present, as in the [Rh2III(AcO)4(HPen)2] complex (Scheme b). [Rh2III(Pen)4]2– was detected within the first 10 min of reaction (see Appendix 2 in the Supporting Information). The reaction occurs much faster at pH = 7.4.

Scheme 2

Proposed Pathway for Aerobic Reaction of Rh2(AcO)4 with Penicillamine: (a) [RhII2(AcO)4(H2Pen)], (b) [RhIII2(AcO)4(HPen)2], (c) H2[RhIII(Pen)4], and Its Oxidized Form (d) H2[Rh2(Pen)2(Pen(SO))2], All Detected by ESI-MS at Acidic pH

The reactions occur much faster at pH = 7.4, leading to compound 6, Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O, with a structure similar to (d).

The ESI-mass spectrum in the – ion mode of the Rh–penicillamine compound Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O (6) dissolved in water was dominated by mass peaks for [RhIII2Pen4 + H+]− (−m/z = 794.96) at fragmentor voltage = 0 V, and [RhIII2Pen4 ]2– (−m/z = 396.98) at 80 V, and their m/z +8 and +16 satellites. An absorption peak at 2475.6 eV in the S K edge XANES spectrum of 6 further revealed the presence of a sulfenato group (S=O) bound to Rh(III); see Results section.

Oxidation of metal-bound thiolates by peroxides to sulfenato groups has been previously reported for thiolato Co(III)[38,39] and thiolato Ru(II) polypyridyl complexes,[40] as well as for glutathione-bound Ru(II)-arene and organo-Ir(III) anticancer compounds.[41−43] Liu and Sadler proposed that hydrogen peroxide (detected by a H2O2 test stick), formed from O2(g), provides the sulfenato (S=O) oxygen atom,[42] as previously suggested by the Dunbar group.[15] Here, we also propose that oxidation of the Rh24+ core in 1 to Rh26+ in 6 occurs by reduction of oxygen (O2) to peroxide (eq ), which in turn may oxidize a coordinated penicillamine thiolato to a sulfenato (S=O) group (eq ). Formation of peroxide in the reaction mixture was confirmed by an iodine test (see Appendix 1 in the Supporting Information).[44]The structure of the Rh–penicillamine compound 6 was probed by Rh K edge EXAFS, complemented by 13C CPMAS NMR spectroscopy, which revealed 3.1 Rh–(N/O) and 3.2 Rh–S distances at 2.09 ± 0.02 and 2.32 ± 0.02 Å, respectively (Figure and Table ). The observation of two 13C NMR signals at 181.4 and 176.1 ppm for this solid (Figure ) provides evidence for two types of carboxylate groups, and thus supports two coordination modes for the Rh(III)-bound penicillamine ligands: tridentate (S,N,O-Pen) and bidentate (S,N-Pen); see Results section. Compound 6 is expected to have a similar structure as that shown in Scheme d.

Scheme displays our proposed pathway for the aerobic reaction between Rh2(AcO)4 and penicillamine (mole ratio of 1:4). In the proposed structure for Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O (6), the O atom is bound to the S atom of a tridentate (S,N,O-Pen) ligand because in Rh–S–C, the S atom, with bonds to two other atoms, is sterically less hindered for nucleophilic attack by peroxide than the bridging S atom in Rh–Sbridge–Rh with three bonds (Scheme ).[38,39]

In our CSD survey, only three crystalline dithiolato-bridged binuclear Rh(III) complexes were found, for which the average Rh···Rh distance is 3.54 ± 0.03 Å (see Table S-6a–c in the Supporting Information). In the binuclear [RhIII(μ-S,N-C6H6NS)(η1-S-C6H6NS)(bpy)]22+ complex with a similar (S,N)-donor ligand as that in the Rh–penicillamine compound 6, the Rh···Rh distance is 3.51(3) Å.[14] The contribution of Rh···Rh scattering to the overall Rh K edge EXAFS spectrum of 6 would be insignificant for a long nonbonded distance (with high σ2), and does not appear as a well-resolved peak in the EXAFS Fourier transform in Figure .

Aerobic Reaction of 1 with Cysteine

The aerobic reaction of Rh2(AcO)4 with cysteine at a mole ratio of 1:4 differs from that with penicillamine. The diamagnetic precipitate gradually formed at pH = 3.2 was separated and analyzed with the empirical formula {Rh2III(HCys)2(Cys)2·4H2O} (4). When raising the pH of the reaction mixture, the precipitate dissolved completely at pH = 3.9. The solid product obtained after initially adjusting the pH of the reaction mixture to 7.4 (see Experimental Section) empirically corresponds to the formula {Na2[RhIII2(Cys)4]·5H2O} (5).

The Rh K edge EXAFS spectrum of this solid was identical to that of 4 (Figure S-13), but had a different phase and amplitude when compared with that of the Rh–penicillamine compound 6 (see Figure S-12, top). Least-squares curve-fitting of the EXAFS oscillation for the solid 5 revealed contributions from 2.0 Rh–(N/O) and 4.1 Rh–S bond distances at 2.10 ± 0.02 and 2.34 ± 0.02 Å, respectively (see Figure and Table ).

The 13C CPMAS NMR spectrum of the solid 5 displays a main peak at 176.3 ppm and a shoulder at 182.3 ppm (Figure ) assigned to bidentate (S,N-Cys) and tridentate (S,N,O-Cys), respectively (see Results section). The different relative intensity of these two peaks, as compared with the 13C NMR spectrum of the Rh–penicillamine compound Na2[Rh2(Pen)2(Pen(SO))2]·4.5H2O (6), implies that the bidentate (S,N)-mode of the cysteinate ligands in 5 dominates.

The weak absorption at 2475.6 eV in the S K edge XANES spectrum of the Rh–cysteine compound 5 shows a significantly smaller amount of sulfenato (S=O) groups than that in 6 (Figure ). However, in the ESI-mass spectrum of 5 (dissolved in water), the m/z +8 and +16 satellites for the peaks at (−)m/z = 340.91 and 682.83 correspond to one or two additional oxygen atoms to the [Rh2(Cys)4]2– and [Rh2(Cys)4 + H+]− mass ions (Figure and Table ), respectively, indicating that ESI-MS is a more sensitive probe for detecting small amounts of additional O atoms. Even at the low fragmentor voltage of 10 V, intense m/z +8 and/or 16 peaks were identified for the mass ions [Rh2(Cys)4 + H+]− (−m/z = 682.83) and [Rh2(Cys)5 + 3H+]− (−m/z = 803.84); see Figure S-10 and Table S-4. It is unlikely that the added O atoms are generated from oxidation in the ESI-MS instrument because, for example, no m/z +8 and/or 16 satellites were observed for [RhIII2(AcO)4(HNAC)2 + Na+]+ (+m/z = 788.90) or N-acetylcysteine [H2NAC – H+]− (−m/z = 162.02) in Figures S-2 and S-6, respectively.

At the fragmentor voltage of 80 V, additional mass peaks corresponding to tri- and tetranuclear species appeared, such as: [Rh3(Cys)6]3– (−m/z = 340.91), [Rh3(Cys)6 + H+]2– (−m/z = 511.87), [Rh3(Cys)6 + H+ + O]2– (−m/z = 519.87), [Rh4(Cys)8 + H+]3– (−m/z = 454.89), and [Rh4(Cys)8 + 2H+ + O]2– (−m/z = 690.83). The ESI-MS analysis of the Rh–penicillamine reaction product 6 also shows minor amounts of similar tetramers {Na2[Rh2(Pen)4]} (n = 2), based on the isotopic patterns of the peaks (−)m/z = 794.96 and 810.95 for the [Rh4(Pen)8 + 2H+]2– and [Rh4(Pen)8 + 2H+ + O]2– ions, respectively (see Table ). Although some cluster formation could also be observed in the ESI-mass spectra of pure Rh2(AcO)4 in water (Figure S-6), the difference between the EXAFS spectra of 5 and 6 (Figure S-12) can only be explained by the presence of oligomeric species in 5.

On the basis of the combined results, we conclude that in the reaction product 5 of Rh2(AcO)4 and cysteine (pH = 7.4), the Rh(III) ions are connected via double thiolate bridges probably in at least up to tetrameric species, although some dimeric Na2[Rh2(Cys)4] complexes, similar to those found for the Rh–penicillamine compound 6, cannot be ruled out. Scheme displays our proposed structure for a tetramer {Na2[Rh2(Cys)4]} (n = 2) and its oxidized form, as identified by ESI-MS.

Scheme 3

Proposed Structures for a Tetrameric {Na2[Rh2(Cys)4]}2 Complex in Compound 5

The lower image is its oxidized form, detected by ESI-MS, with sulfenato (S=O) groups.

In a large oligomeric structure for 5, there are four Rh–S and two Rh–(N/O) bonds around the Rh(III) ions (except for the terminal units), as indicated by the EXAFS data analysis (Table ). Only the S atoms with two bonds in the terminal tridentate (S,N,O-Cys) ligands are expected to be oxidized by peroxide to sulfenato (S=O) groups because the bridging thiolate groups with three bonds are less susceptible to nucleophilic attack by peroxide.[38] The number of unbound COO– groups is considerably higher than that of the bound terminal COO–, which gives rise to a small shoulder at δ = 182.3 ppm in the 13C CPMAS NMR spectrum of 5 (Figure ). The close overlap of the EXAFS spectra (Figure S-13) indicates that the structures of the Rh–cysteine precipitate 4, {Rh2(Cys)2(HCys)2·4H2O} from acidic (pH = 3.2) media, and 5 are similar, apart from the protonated −COOH groups in 4, shown by the 13C NMR resonance at δ = 194.3 ppm (Figure ).

The chiral [Rh(S,N-Cys)3]3– complex readily functions as a S-donating ligand to form linear S-bridged trinuclear [M{Rh(S,N-Cys)3}2]3–/4– complexes with various metal ions (M = CrIII, CoII or III, and NiII) with MS6 coordination,[37,45,46] as well as in the T-cage S-bridged polynuclear complex [{Rh(S,N-Cys)3}4Zn4O]6–, which has been studied for decades.[47] Polynuclear S-bridged structures such as the tetranuclear one proposed in Scheme for the {[RhIII2(S,N-Cys)4]2–} (n = 2) complex are other examples of this type of multinuclear compound, formed in absence of a different metal ion (M).

Such oligomers could be formed with morphologically similar but not identical Rh(III) centers, for example, with different orientations/conformations of Cys2– or Pen2– rings. The variations in the local environment of the carbon atoms in these ligands would result in broad signals in the 13C NMR spectra in Figure .[48−52] Recently, a 1H NMR spectrum was reported for a D2O solution of S-methylcysteine coordinated to the axial position of Rh2(AcO)4 (mole ratio of 1:1); the broad signals observed in this spectrum were attributed to J(H,H) coupling.[53]

Aerobic Reaction of 1 with N-Acetylcysteine

Monitoring the slow reaction of Rh2(AcO)4 (1) with N-acetylcysteine (mole ratio of 1:4) at the pH of mixing (pH = 1.9) with ESI-MS revealed that species with one N-acetylcysteine ligand (in protonated or even fully deprotonated form) still keep the RhII–RhII core, as in [RhII2(AcO)4(H2NAC)] and [RhII2(AcO)3(HNAC)] (with ±H+ or +Na+) as well as [RhII2(AcO)2(NAC)]. Similar to the penicillamine reaction, oxidation to Rh(III) is observed in species with two deprotonated N-acetylcysteine ligands: [RhIII2(AcO)4(HNAC)2] and [RhIII2(AcO)3(HNAC)2]+ (see Appendix 1 in the Supporting Information).

Following the reaction in acidic solution with UV–vis spectroscopy showed after 48 h, an intense absorption peak at λmax = 367 nm (Figure , left), which is attributed to a S– → RhIII ligand-to-metal charge-transfer band of the [RhIII2(AcO)4(HNAC)2] complex, which is the main Rh(III) complex according to the ESI-mass spectrum (+m/z = 766.91, 788.90, and 810.88 for +H+/Na+ ions, Table and Figure right). The EXAFS spectrum of this complex could not be obtained due to the presence of unreacted Rh2(AcO)4 (see 1H NMR and ESI-mass spectra in Appendix 1).

The solid reaction product at pH = 7.4, purified by size exclusion chromatography, has the empirical formula {Na2[Rh2(NAC)4]·4.5H2O} (7), based on elemental analysis. The Rh K edge EXAFS spectra of 7 and of the corresponding Rh–cysteine reaction product 5 show main FT peaks of similar magnitude, but with an additional feature at ∼2.7 Å for 7, as also found for the Rh–glutathione reaction product 8 (Figure S-12).[23] Its EXAFS oscillation was best modeled with 2 Rh–O, 4.6 Rh–S, and 0.8 Rh···Rh interactions at mean distances of 2.09 ± 0.02, 2.32 ± 0.02, and 3.10 ± 0.02 Å, respectively, which are similar to those obtained for 8 (see Figure and Table ). The Rh–O scattering contribution is better resolved for 8, appearing as a shoulder in the main FT peak, probably due to less noise in its EXAFS oscillation. Simultaneous refinement of the Rh–O coordination number for 7 resulted in an unreasonably high value (5.7) for the Rh–S coordination number, and very small σ2 for Rh–O (see Table S-5). In the trimeric ΔΛ-[Rh{Ir(aet)3}2]3+ complex (CSD code: AQUVOX), the central Rh(III) ion is surrounded by six bridging thiolate groups with the average Rh–S distance 2.381 Å,[19] which is longer than the refined mean Rh–S distance in 7. Therefore, a model including two Rh–O and four Rh–S scattering paths seems reasonable. The RhIII···RhIII distance in 7 is comparable with the RhIII···IrIII distance of 3.0199(3) Å in the [Rh{Ir(aet)3}2]3+ complex with three bridging thiolate groups between the Rh(III) and Ir(III) ions in N3Ir(μ-S)3Rh(μ-S)3IrN3 coordination.[19]

The 13C CPMAS NMR spectrum of 7 only shows one set of broad peaks for the coordinated N-acetylcysteine, with the COO– group resonating at δ = 177.1 ppm, shifted downfield only by Δδ (13C) = 1.7 ppm relative to that of the pure ligand (Figure ), indicating that the carboxylate group is not directly bonded to the Rh(III) atoms in 7. The ESI-mass spectrum of this product in water at fragmentor voltage 0 V was dominated by mass peaks for tetranuclear RhIII4(NAC)8–10 (m/z = 850.87, 932.39, and 1024.88) or dinuclear RhIII2(NAC)4 species, and their m/z +8 satellites for one added O atom (see Figure S-9 and Table S-3).

The EXAFS results show that {Na2[Rh2(NAC)4]·4.5H2O} (7) contains triply S-bridged RhIII···RhIII units with a well-defined distance of 3.10 ± 0.02 Å that gives rise to an extended EXAFS oscillation (Figure ). A single or double thiolate bridge connecting those dimeric units into an oligomeric structure, as proposed in Scheme , would be consistent with the refined coordination number of 4.6 for the Rh–S scattering path. The scattering contribution from Rh(III) ions bridged by one or two thiolate groups (as in compound 5) at Rh···Rh distances >3.5 Å would be damped in the EXAFS oscillation.

Scheme 4

Proposed Oligomeric Structures for the {Na2[Rh2(NAC)4]·4.5H2O} (Compound 7)

An oligomeric structure for 7 would justify the Rh4(NAC)8–10 and Rh6(NAC)10–12 mass ions in the ESI-MS spectra (Tables and S-3). In such an oligomeric structure, only the terminal thiolate can be oxidized by peroxide to a sulfenato group. In the ESI-mass spectrum of 7, just one added O atom was observed as an m/z +8 satellite for the dimeric (−m/z = 866.87) or tetrameric (−m/z = 858.87) Rh–N-acetylcysteine species (see Tables and S-5), indicating that there are more bridging than terminal thiolate groups.

Conclusions

Aerobic reactions of Rh2(AcO)4 with the thiol-containing amino acid cysteine (H2Cys) and its derivatives penicillamine (H2Pen) and N-acetylcysteine (H2NAC) in different mole ratios (1:2, 1:4, and 1:6) at pH = 7.4 result in the formation of the diamagnetic compounds {Na2[RhIII2L4]·xH2O} (L = Cys and NAC; x = 4.5–5) and Na2[RhIII2(Pen)2(Pen(SO))2]·4.5H2O. The complexes were structurally characterized by combining the results from Rh K edge EXAFS, S K edge XANES, and solid state 13C CPMAS NMR spectroscopic techniques. In the binuclear penicillamine complex, the thiolate groups of two (S,N-Pen) ligands bridge the Rh(III) ions and the other two terminal ligands bind in tridentate mode (S,N,O-Pen); the bond to the COO– group was evidenced by a change of chemical shift ΔδC = 7.3 ppm relative to that of the pure ligand. Oxidation of Rh24+ to Rh26+ occurs through oxygen (O2) that is reduced to peroxide, which in turn can partially oxidize terminal thiolate (S–RhIII) to sulfenato (O=S–RhIII) groups. The presence of sulfenato groups in Na2[RhIII2(Pen)2(Pen(SO))2]·4.5H2O was confirmed by a peak at 2475.6 eV in its S K edge XANES spectrum.

The high Rh–S coordination number obtained from the EXAFS data analysis of the compound {Na2[RhIII2(Cys)4]·5H2O} is consistent with oligomeric species in which dithiolate bridges of chelating (S,N-Cys) ligands connect the Rh(III) ions. In comparison with the dimeric penicillamine complex, there are fewer terminal tridentate (S,N,O-Cys) ligands that can then be oxidized to sulfenato groups (O=S–RhIII), which could only be detected by ESI-MS.

In the compound {Na2[RhIII2(NAC)4]·4.5H2O}, N-acetylcysteine behaves similarly to glutathione and acts as a monodentate S-donor ligand, forming a triple (S-NAC) bridge between two Rh(III) ions with a well-defined Rh···Rh distance of 3.10 ± 0.02 Å, according to the Rh K edge EXAFS data analysis. Those dimeric units are probably linked together into oligomeric complexes by single or double thiolate bridges. The bridging thiolate groups, sterically hindered by bonds to three atoms, are not oxidized by peroxide, leaving only the terminal thiolates (S–RhIII) exposed to oxidation to sulfenato groups.

The results of this study reveal the structure of products that could form when antitumor active rhodium(II) tetraacetate, Rh2(AcO)4, interacts with thiol-containing proteins and peptides, whether acting like N-acetylcysteine or glutathione as a monothiolate entity, or as the (S,N-) chelate cysteine in an N-terminal site of a protein.