- Email: [email protected]
Proton-linked bi- and tri-metallic gold cyanide complexes observed by ESI-MS spectrometry
Available online at www.sciencedirect.com JOURNAL OF
Inorganic Biochemistry Journal of Inorganic Biochemistry 102 (2008) 576–583 www.elsevier.com/locate/jinorgbio
Proton-linked bi- and tri-metallic gold cyanide complexes observed by ESI-MS spectrometry Philip M. Yangyuoru, James W. Webb, C.F. Shaw III * Department of Chemistry, Illinois State University, Normal, IL 61791-4160, United States Received 29 June 2007; received in revised form 15 October 2007; accepted 26 October 2007 Available online 22 November 2007
Abstract Electrospray ionization spectra of potential cyanide-containing gold-drug metabolites revealed additional, weak, unanticipated peaks at approximately twice the mass of the gold(I) and gold(III) cyanide complexes. The exact masses correspond to proton-linked bimetallic complexes, [H{Au(CN)m}2], (m = 2,4). Further investigation revealed a total of 12 examples, including trimetallic complexes, [H2{Au(CN)m}3]; mixed species with two complexes, [H{Au(CN)2}{Au(CN)4}]; and thiolato species, [H{(RS)Au(CN)3}2]. trans[AuX2(CN)2Cl2] and trans-[AuX2(CN)2Br2] generated 35Cl/37Cl and 79Br/81Br isotopic patterns for the protonated bi- and tri-metallic analogues which were in good agreement with the presence of four or six halide ligands, respectively. Concentration-dependent studies demonstrated that the signals are independent of the solution concentrations of mono-metallic precursors, suggesting formation in the gas phase during or following droplet desolvation. Ó 2007 Elsevier Inc. All rights reserved. Keywords: Electrospray ionization mass spectrometry; Gold(I); Gold(III); Cyanide; Proton-linked
1. Introduction Gold-based anti-arthritic drugs have been used for nearly eight decades [1–3] and are considered to be disease-modifying drugs, as opposed to palliative treatments. One of the known metabolites is aurocyanide ([Au(CN)2]), which has been detected in blood and urine of patients using three different drugs: myochrysine (gold sodium thiomalate), solgonal (gold thioglucose) and auranofin (2,3,4,6-tetra-O-acetyl-1-thio-b-D-glucopyranosato-S-(triethylphosphine)gold(I)) [4,5]. Evidence for the conversion of gold(I) to gold(III) has been reported in mice [6] and men [7]. The enzyme myleoperoxidase has been implicated in both processes [8–10] and this has led to the hypothesis that auricyanide ([Au(CN)4]) might also be a metabolite [10]. A variety of gold(III) complexes show very promising anti-tumor activity [11], but are subject to poten-
*
Corresponding author. Tel.: +1 309 438 7207; fax: +1 209 438 5538. E-mail address: [email protected] (C.F. Shaw III).
0162-0134/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2007.10.020
tial reduction to gold(I), giving impetus to studies of the mechanisms of the biological reduction mechanisms. While studying the reduction of auricyanide by thiols, mass spectra revealed negative-ion peaks resulting from complexes containing two or three gold ions, which could not be accounted for by a ligand exchange of thiolate for cyanide (P.M. Yangyuoru, J.W. Webb, C.F. Shaw III, this issue). The masses of these peaks correspond to formation of bi- and tri-metallic complexes, [Hn{Au(CN)m}n+1]. It was therefore interesting to investigate these species further because such proton-linked clusters have not been documented yet in the solution chemistry of gold nor, to our knowledge, in ESI-MS spectra of other metal cyanides. 2. Experimental Reagents were obtained as follows: reduced glutathione from Sigma–Aldrich; cysteine, Sigma Chemical Company; cysteine methyl ester (Lot # A016240201), N-acetylcysteine (Lot # A018477901) and formic acid (Lot # A019673801) from Acros Organics; HPLC-grade acetonitrile (ACN,
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Lot # 054754) from Fisher Chemicals Company. K[Au(CN)2Cl2] was prepared by the method of Canumalla et al. [10], kmax = 217, 287 nm (lit 217, 287 nm), m/z = 319.0, 321,0, 323.0; mCN = 2175 cm1. K[Au(CN)2Br2] was a generous gift of Dr. Anna Cannumalla. 2.1. ESI-mass spectrometry ESI-mass spectra were recorded on a Hewlett Packard (Agilent) LC–MS (Series 1100 LC–MSD with a UV–vis diode array detector) set in the negative ion mode with 40 V fragmentation voltage, 3500 V capillary voltage; 10 L/min N2 flow at 25 psi; ionization temperature 250 °C. Samples (10.0 lL of 0.1–100 mM concentration of gold complex) were injected into the ESI chamber. The flow rate was set at 0.400 mL/min. Mobile phase solvents were typically (A) 0.1% formic acid in nanopure water and (B) 0.1% formic acid in acetonitrile (ACN). Unless otherwise indicated, a mobile phase composition of 50% A and 50% B was used throughout this research. Solvent composition and spray parameters were fixed for most of the analyses. Using the auto sampler, the delay time between mixing and the first spectrum was typically 3 min. 2.2. [{Au(CN)m}n+1Hn] Solutions typically containing 2.00 mM [Au(CN)m] (m = 2, 4) in nanopure water were analyzed by ESI mass spectrometry over the range 100–1000 m/z. To study the possibility of the formation of mixed clusters, solutions containing both 1.00 mM [Au(CN)4] and 1.00 mM [Au(CN)2] in nanopure water were analyzed by ESI-MS. Apart from nanopure water, studies were also conducted in acetonitrile + 0.1% HCOOH solvent medium. 2.3. [Au(CN)m] and [H3O+] concentration-dependence studies Varying concentrations (1.00, 2.00, 4.00, 5.00, 50.0 and 100 mM) of only [Au(CN)4] or equimolar mixtures of [Au(CN)4] and [Au(CN)2] in nanopure water were analyzed. Studies were done in the acetonitrile + 0.1% HCOOH medium and compared to identical studies in nanopure water to determine whether the formation of the protonated clusters is affected by the solvent and the [H3O+] concentration. 2.4. [Hn{Au(CN)2X2}n+1] Aqueous solutions ranging from 2.3 to 22.8 mM of K[Au(CN)2Br2] and K[Au(CN)2Cl2] were examined. The chloride complex was used to study the effect of the fragmentation voltage (0–120 V in 20 V steps) on the intensity of the bi- and tri-metallic species and the extent of reductive elimination of X2 from the anionic complexes.
577
2.5. [Hn{Au(SR)(CN)3}n+1] An extensive library of ESI-MS spectra obtained during examination of the auricyanide reduction (P.M. Yangyuoru, M.S. Thesis, Illinois State University, 2007) was scanned for bi- and tri-metallic thiolate-substituted complexes. Only bi-metallic complexes with RS = cysteine or O-methyl-cysteine were noted. 3. Results While studying the reduction of auricyanide by thiols, Eq. (1): [Au(CN)4 ] + 2G SH ! [Au(CN)2 ] + GSSG2 + 2HCN ð1Þ several ESI mass spectra revealed peaks of previously unrecognized complexes containing two or three gold ions. Fig. 1, for example, exhibits peaks corresponding to the reactants (GSH, m/z = 306.0 and [Au(CN)4], m/z = 301.0), two intermediates ([Au(SGH)(CN)3], m/z = 581.0 and [Au(SCy-Gly)(CN)3], m/z = 452.0) and a trace of the product ([Au(CN)2], m/z = 249.0). There is an additional peak at m/z = 603.0, approximately twice the mass of aurocyanide, but a dimer would have a mass of 602.0 and a net 2 charge, and would therefore have an m/z value of 301.0, indistinguishable from the monomeric complex. Nor did this peak correspond to any of the possible reactants or intermediates containing a thiolate ligand in the complex. The observed m/z of 603.0 can be accounted for by a proton linking two [Au(CN)4] ions to form a bimetallic cluster with a net 1 charge, as detected by ESI-MS. The following protonation reaction, Eq. (2), was therefore suggested to generate the cluster [H{Au(CN)4}2]: 2[Au(CN)4 ] + Hþ ! [HfAu(CN)4 g2 ]
ð2Þ
If the postulated complex and its formation are correctly defined, it should form independently of the glutathionederived species involved in the overall auricyanide reduction reaction, Eq (1). ESI mass spectra of [Au(CN)4] obtained before the reduction process also exhibited, upon reexamination, minor peaks corresponding to the bi-metallic complex, [{Au(CN)4}2H] (Fig. 2). Two additional proton-linked clusters were observed in ESI mass spectra of pure K[Au(CN)2] in nanopure water (Fig. 3). In addition to [Au(CN)2] (m/z = 249.0), a bimetallic gold cluster, [H{Au(CN)2}2] (m/z = 498.9), analogous to the bimetallic gold(III) species, was present. In addition, a peak at m/z = 749.2 resulting from the formation of [H2{Au(CN)2}3] was observed. The protonation reactions for [Au(CN)2] (Eqs. (3) and (4)) correspond to peaks observed at m/z = 498.9 and 749.2, respectively:
578
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Fig. 1. Negative-ion electrospray mass spectrum of a reaction mixture initiated with 0.100 mM [Au(CN)4] and 0.200 mM glutathione (GSH) in nanopure water and recorded after mixing plus auto sampling time of 3 min. The m/z values are rounded to whole numbers in the figures, but given to tenths of a mass unit in the text and typically agree with calculated values ±0.1 unit.
Fig. 2. Mass spectrum of 1.00 mM K[Au(CN)4] in nanopure water. The intensity scale was expanded to maximize the low-intensity peaks. The average intensity of [Au(CN)4] peak was about 250,000 counts.
2½AuðCNÞ2 þ Hþ ! ½HfAuðCNÞ2 g2
þ
ð3Þ
3½AuðCNÞ2 þ 2H ! ½H2 fAuðCNÞ2 g3
ð4Þ
To our knowledge, the formation of protonated bi- and trimetallic gold clusters of the form [{Au(CN)x}n+1Hn] has not been reported to date. Therefore, additional examina-
tion of these species was of interest. Because self-complexation of either [Au(CN)4] or [Au(CN)2] occurs independently, leading to formation of bi- and tri-metallic complexes, one would predict that mixed clusters of gold(I) and gold(III) should form in solutions containing both [Au(CN)4] and [Au(CN)2]. The homoleptic gold(I) and
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
579
Fig. 3. Mass spectrum of 1.00 mM K[Au(CN)2] in nanopure water. The intensity scale was expanded to maximize the low-intensity peaks. The average intensity of [Au(CN)2] peak was about 300,000 counts.
gold(III) cyanides were mixed and examined by ESI-MS in two different solvent systems: nanopure water (not shown) and acetonitrile + 0.1% HCOOH (Fig. 4). In addition to the complexes [{Au(CN)2}2H], [{Au(CN)2}3H2], and [{Au(CN)4}2H], a mixed bi-metallic complex (m/z = 551.0) was formed from the two monomeric complexes and a proton: [Au(CN)2 ] + [Au(CN)4 ] + Hþ ! [HAu(CN)2 Au(CN)4 ] ð5Þ
The potassium linked-complexes of auricyanide (Eq. (6)) were also observed at m/z = 641.1 (Figs. 2 and 4). These
potassium complexes are analogous to the protonated complexes and have the same 1 charge as the protonated complexes. 2[Au(CN)4 ] + Kþ ! [KAu(CN)42 ]
ð6Þ
Potassiated and sodiated complexes of peptides and proteins are commonly encountered in positive-ion ESI mass spectra [12]. Although K[Au(CN)2] was used in these studies, no analogous potassium-linked aurocyanide clusters were formed. Two examples of bimetallic clusters of tricyano(thiolatoS)gold(III) complexes were observed during reduction
Fig. 4. Mass spectrum of 1.00 mM [Au(CN)4] and 1.00 mM [Au(CN)2] in acetonitrile + 0.1% HCOOH. The intensity scale was expanded to emphasize the low-intensity peaks. The noise of the spectrum is due to the expansion of the less intense peaks. The average intensities of the [Au(CN)4] and [Au(CN)2] peaks were 200,000 and 122,000 counts, respectively.
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583 10000 8000
Intensity
reactions (analogous to Eq. (1)) in which cysteine (Fig. 5) or O-methylcysteine (not shown) replaced glutathione. The mass spectrum in Fig. 5, taken one hour after mixing cysteine and auricyanide, exhibited a peak at m/z = 791.6, which corresponds to [H{Au(SCy)(CN)3}2]. To further explore the phenomenon, solutions of the dihalodicyanogold(III) complexes of chloride and bromide were examined. The mass spectra for K[Au(CN)2Cl2] showed three parent ion peaks for [Au(CN)2Cl2] at m/ z = 319.0, 321.0 and 323.0, consistent with the 35Cl (0.76 abundance) and 37Cl (0.24 abundance) isotope pattern predicted for two chloride ligands: 0.57:0.37:0.06. Two peaks assignable as proton-linked cluster species were observed, Fig. 6. An isotopic pattern at m/z = 638.7, 640.8, 642.8 and 644.8 arises from the four chlorides present in the bimetallic complex [H{Au(CN)2Cl2}2], Fig. 6 (upper), and corresponds to the intensities (0.31:0.39:0.19:0.04) predicted for the most intense peaks. As expected, the m/z = 640.8 peak is most intense of the four, but the remaining 6 weaker peaks are not observed. A weaker isotopic pattern at m/z = 958.7, 960.7 962.7, 964.7 and 965.7, Fig. 6 (lower), corresponds to the five most intense peaks with abundances of 0.17: 0.33:0.27:0.11:0.03 of the 11 expected for the presence of six chloride ions. The principal isotopic peaks for these proton-linked bi- and tri-metallic species are separated by the mass difference (2.0) of the chlorine isotopes, as expected, for monoanions. The exact agreement of masses, and reasonable agreement with the predicted abundances provide strong confirmation that the species being observed correspond to the protonated bi- and tri-metallic species: [H{Au(CN)2Cl2}] and [H2{Au(CN)2Cl2}3]. In some runs, there also a weak peak at m/z = 249.0, which results from gas phase decomposition [Au(CN)2Cl2] by reductive elimination of chlorine at elevated temperatures in the electrospray chamber:
-
[H{Au(CN)2Cl2}2]
6000 4000 2000 0 625
630
635
640
645
650
10000 8000
Intensity
580
-
[H2{Au(CN)2Cl2}3]
6000 4000 2000 0 950
955
960
965
970
m/z Fig. 6. Expanded scale spectrum of 22 mM aqueous K[Au(CN)2Cl2] in the m/z ranges for (upper) the protonated bimetallic complex; and (lower) the protonated trimetallic complex.
[Au(CN)2 Cl2 ] ! Cl2 + [Au(CN)2 ]
ð7Þ
The absorbance peaks characteristic of [Au(CN)2] at 204, 211, 230 and 239 nm were absent from the UV–vis spectrum of the K[Au(CN)2Cl2] sample used. The intensity of the aurocyanide peak increased and decreased with changes in the fragmentation voltage (0–70 V), indicating that the aurocyanide formed during the mass spectral analysis. Additionally, a single peak at m/z = 499.0 resulting from the formation of [H{Au(CN)2}2], and an isotopic pattern at m/z = 568.9, 570.9 and 572.9 demonstrating the formation of the mixed bimetallic complex [H{Au(CN)2Cl2}{Au(CN)2}] were observed.
Fig. 5. Mass spectrum of a 1.00 mM [Au(CN)4] + 1.00 mM CySH reaction mixture in nanopure water measured after 1 h.
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Similar monometallic and bimetallic complexes were observed in the spectra of K[Au(CN)2Br2] (not shown). The isotopic abundances of bromine, 79Br (0.507) and 81 Br (0.493), are sufficiently close to being equal that the patterns closely approximate binomial distributions for any number of bromine ligands. The parent ion, [Au(CN)2Br2] was a very strong ‘‘triplet” (m/z = 407.8, 409.8, 411.8). The bimetallic species, [H{Au(CN)2Br2}2] exhibited the expected five peaks (m/z = 814.6, 816.6, 818.6, 820.6, 822.6). For the trimetallic species [H2{Au(CN)2Br2}3], a weak but clearly interpretable isotope pattern corresponding to the five strongest bands (m/z = 1224.4, 1226.4, 1228.4, 1230.4, 1232.4) of the predicted seven-peak pattern were present, providing the third example of a trimetallic species in this study. In these spectra, signals for [H{Au(CN)2Br2}{Au(CN)2}] (m/z = 656.8, 658.8, 660.8), [H{Au(CN)2}2] (m/z = 499.0) were also clearly identified. Table 1 summarizes the twelve protonated bi- and trimetallic complexes observed in solutions of pure compounds or during the reduction of auricyanide by thiols. In every case the agreement between the mass of the mono-metallic species and the formulation as a protonlinked bimetallic or trimetallic species was within ±0.1 mass unit. Two possible origins of the proton-linked bi- and trimetallic cyanogold species were considered. They might arise as the result of equilibria in solution, or they may form in the ESI ionization chamber during desolvation of the analyte droplets. Cyanide complexes of gold and many other metal ions are highly water soluble, implying strong hydration. H-bonds to cyanide are a significant source of solvation energy contributing to the high solubilities of cyanide complexes, and are also frequently generated in crystal structures by waters of crystallization or other Hbond donors: e.g., [Au(CN)4] [13,14], [Ni(CN)5]3 [15], [HO-Cr(cyclam)-NC-Cr(CN)5] [16] [16-pyrimidinium crown-4][Au(CN)2]4 6.5H2O [17]. In some cases, cyanide ligands of low-spin ferrous complexes (which are d6 and hence kinetically inert) can be protonated in solution, e.g., [Fe(R2PCH2CH2PR2)2H(CNH)]+ [18] and [Fe(CO)(CN)4(CNH)]2 [19]. Metal–cyanide complexes
form ion pairs with cations in aqueous solution: e.g., a series of {[cation]+[Fe(CN)5(NO)]2} complexes, where cation+ = H+, Li+, Na+, K+, Rb+, Cs+ (the first being another protonated cyanide as in the previous examples) have been observed by ESI-MS and shown by 13C NMR spectroscopy to involve direct cyanide–cation interactions [20]. Finally, numerous linearly-bridged cyanides in bimetallic complexes also demonstrate the nucleophilicity of coordinated cyanide ligands: e.g., [(cyclen)Cu-(l-NC)Au(CN)]+ [21], [l-CN-(AuC9H11)2] [22], trans-[(HO)Cr(cyclam)-(l-NC)-Cr(CN)5] [16]. These selected examples provide four clear lines of evidence for the nucleophilicity and basicity (in both the Lewis and Bronsted-Lowry senses) of cyanide nitrogen atoms. If the protonated bi- and tri-metallic gold complexes form in solution, their concentrations should be dependent on the concentration of the mono-metallic species as well as the pH of the solution, according the equilibrium reaction: 2[Au(CN) ] + Hþ ½HfAuðCNÞ g ð8Þ 4
4 2
Therefore, concentration-dependent studies of the formation of the protonated bi- and tri-metallic gold cyanides were conducted using ESI mass spectrometry. The expected concentration dependence for formation of the bimetallic clusters from the homoleptic gold cyanides in solution can be derived for the postulated equilibrium (Eq. (8)) by solving the Keq expression (Eq. (9)) for the concentration of the bimetallic species, as shown in Eq. (10):
½HfAuðCNÞx g2
þ
¼ ðKeq ½H Þ
m/z
[H{Au(CN)4}2] [H{Au(CN)2}2] [{Au(CN)4}H{Au(CN)2}] [H{CySAu(CN)3}2] [H{O-MeCySAu(CN)3}2] [H{Au(CN)2Cl2}2] [H{Au(CN)2Cl2}{Au(CN)2}] [H{Au(CN)2Br2}2] [H{Au(CN)2Br2}{Au(CN)4}]
603.0 499.0 551.0 791.6 819.2 640.8 568.9 818.6 656.8
Trimetallic species [H2{Au(CN)2}3]
m/z 749.2
ð9Þ
2 ½AuðCNÞx
ð10Þ
Eq. (10) predicts an increase in the protonated bimetallic auricyanide species, [H{Au(CN)4}2], according to the square of the [Au(CN)4] concentration, Fig. 7. Analysis of 0.100–100 mM [Au(CN)4] solutions revealed only marginal changes in the concentrations of [{Au(CN)4}2H], erratically increasing up to a maximum at 20.0 mM [Au(CN)4] and then decreasing. When plotted on the 600000
Intensity
Bimetallic species
2
Keq ¼ ½HfAuðCNÞx g2 =f½AuðCNÞx ½Hþ g
500000 Table 1 m/z valuesa of protonated bi- and tri-metallic gold cyanide complexes
581
Measured Intensity
400000
Expected Intensity
300000 200000 100000 0 0
[H2{Au(CN)2Cl2}3]
960.7
1
3
2
Au(CN)4-,
[H2{Au(CN)2Br2}3]
1228.4
a Values for the most intense peak of the chlorine and bromine complexes are given.
4
5
mM
Fig. 7. Intensities of [H{Au(CN)4}2] predicted for a solution-phase equilibrium over the concentration range 1–5 mM [Au(CN)4] (- - -) vs the observed intensities (—). The nearly constant signal does not reflect the anticipated 25-fold increase of the protonated bimetallic species over that range.
582
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
same scale as the predicted behavior, the experimental intensities are essentially constant, Fig. 7 Similarly, when equimolar mixtures of [Au(CN)4] and [Au(CN)2] were examined in either nanopure water or acetonitrile +0.1% HCOOH, the peak intensities for the bi-metallic clusters were not significantly dependent on the concentration product {[Au(CN)2]] * [Au(CN)4]]} applicable to the mixtures. Nor did the pH of the solvent medium affect the extent of the protonated-cluster formation observed in acetonitrile +0.10% formic acid, compared to similar concentrations in nanopure water. The lower dielectric constant of the acetonitrile and the increased hydrogen ion concentration should favor the formation of the bimetallic species, which was not observed. This pH-independence is also inconsistent with a preexisting equilibrium of the bimetallic complexes, and suggests that they form during or subsequent to the droplet desolvation in the gas phase. The observation of the symmetric complex [H{Au(CN)2}2] and the mixed bimetallic complex [H{Au(CN)2Cl2}{Au(CN)2}] in the spectra of K[Au(CN)2Cl2] provides further evidence that the bimetallic species arise during the mass spectral analyses. The aurocyanide ion observed in the mass spectra of this compound is known (from infrared analysis) to be absent from the solution, but forms by reductive elimination of chlorine at higher temperatures during (and possibly subsequent to) desolvation. Therefore, the two bimetallic species which contain aurocyanide must also be formed subsequent to injection of the sample, eliminating the existence of the clusters as an equilibrium phenomenon in solution and strongly supporting the conclusion that these bi- and tri-gold clusters are formed by a novel process during desolvation of the droplets or in the gas phase. 4. Discussion What bonding forces cause the proton and two cyanide complexes to associate with one another in the gas phase? The most likely explanations are hydrogen bonding [23] and/or aurophilicity [24,25]. Aurophilic interactions are correlation effects between two or more gold atoms. The range of energies for aurophilic interactions, 7–12 kcal/ mol [24,25] falls within the typical range for hydrogen bonds 2–20 kcal/mol [26]. They give rise to aurated nonmetal complexes such as [N(AuPPh3)4]+, extremely stable gold cluster compounds, and gold–gold interactions in crystal structures [24]. These interactions, however, are strongest between d10 gold(I) centers, weak between gold(I) and gold(III) and negligible between d8 gold(III) centers [24]. On the other hand, intramolecular relativistic forces shorten the Au–Au bond distances in the Au2X6 dimers (X = Cl, I) [27,28]. It is possible, then, that gold(I) interactions are important in addition to hydrogen bonding for aurocyanide, but are less important contributors to stabilization of the mixed oxidation state complexes or dihalodicyano complexes, which have been observed here.
Cyanides are good hydrogen bond acceptors. HCN in the solid state is hydrogen bonded to itself to form linear head-to-tail chains HACBNH ACBNHACBNHACBN ˚ . The mN H stretching with an NH N distance of 3.2 A 1 frequency, 3132 cm is significantly decreased from that of individual molecules in the gas phase, 3312 cm1 [29– 31]. Auricyanide crystallizes as H[Au(CN)4]2H2O [13] and as K[Au(CN)4] H2O [14]. Crystal structures of both have been reported. The ‘‘protonated” salt is better formu lated as ½H5 Oþ 2 ½AuðCNÞ4 . In the structure each of the four terminal OH groups of the cation is hydrogen bonded to a cyanide from a different auricyanide ion, and each cyanide nitrogen of a given anion is hydrogen bonded to four distinct H5 Oþ 2 cations, forming an extended three dimensional network. The hydrogen bonds are classical, nearly ˚ ) aligned with the Au–C„N linear OH N bonds (2.75 A axis. In contrast, the potassium salt contains a single water of hydration. It forms two crystallographically independent hydrogen bonds to cyanide nitrogens on different aurocyanide ions. One H-bond is somewhat bent ˚ ) and aligned slightly off (\OH N = 155.8° and 2.94 A axis to the Au C„N. The second H-bond to an independent auricyanide molecule is nearly linear and longer ˚ ), but has an angle 120° (\OH N = 155.8° and 3.14 A to the Au–C„N axis. Many other metal cyanide salts crystallize with hydrogen bonded waters of hydration as discussed above. Thus, the formation of symmetrical hydrogen bonds AAuACBNHNBCAAuA provides a reasonable explanation and would be common to the gold(I) and gold(III) species. Recently, Roscioli et al. described analogous gas-phase cations formed by hydrogen bonding of a proton to two oxygens of alcohols and ethers, R1R2O H OR3R4, where R1, R3 = H, CH3, or C2H5, and R3, R4 = CH3 or C2H5 [23]. If aurophilic effects are also present, the proximity of the gold atoms to one another would place severe restraints on the hydrogen bonds, which could not be colinear with the Au–C„N axes, but would have to be directed at the N atoms of the cyanides at oblique angles. Models for such non traditional H-bonds are found in the structures of K[Au(CN)4] H2O [14] discussed above, [HO-Cr(cyclam)NC-Cr(CN)5] [16] and especially H4Fe(CN)6 [32]. In the latter, intermolecular hydrogen bonds between the ferrocyanide ions exhibit oblique angles between the N H H axis and each Fe–C„N axis [32]. The observation that the intensities of the bimetallic and trimetallic peaks decrease as the ionizing voltage increases from 0 to 120 V supports the suggestion that these species are formed by a weak interaction such as the hydrogen bond, aurophilic interactions or relativistic effects. As is the case for the gas-phase proton-linked alcohols and ethers [23], the existence of these new proton-linked
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
bimetallic species poses interesting theoretical questions regarding the bonding forces that stabilize them. Acknowledgements C.F.S. thanks the ISU College of Arts and Sciences for a Research Enhancement Award in 2006. We thank Dr. Anna Canumalla and Ms. Jin Hwang for the synthesis of K[AuBr2(CN)2] and K[AuCl2(CN)2], respectively. References [1] C.F. Shaw III, Chem. Rev. 99 (1999) 2589–2600. [2] C.F. Shaw III, in: H. Schmidbaur (Ed.), Gold: progress in the chemistry, Biochemistry and Technology, J. Wiley & Sons, Chichester UK, 1999, pp. 259–308. [3] L. Messori, G. Marcon, Met. Ions Biol. Syst. 41 (2004) 279–304. [4] R.C. Elder, Z. Zhao, Y. Zhang, J.G. Dorsey, E.V. Hess, K. Tepperman, J. Rheumatol. 20 (1993) 268–272. [5] Z. Zhao, W.B. Jones, K. Tepperman, J.G. Dorsey, R.C. Elder, J. Pharm. Biomed. Anal. 10 (1992) 279–287. [6] D. Schumann, M. Kubicka-Muranyi, J. Mirtcheva, J. Gunther, P. Kind, E. Gleichmann, J. Immunol. 145 (1990) 2132–2139. [7] J. Verwilghen, G.H. Kingsley, L. Gambling, G.S. Panayi, Arthritis Rheum. 35 (1992) 1413–1418. [8] G.G. Graham, A.J. Kettle, Biochem. Pharmacol. 56 (1998) 307–312. [9] C.F. Shaw III, S. Schraa, E. Gleichmann, Y.P. Grover, L. Dunemann, A. Jagarlamudi, Metal-Based Drugs 1 (1994) 351–362. [10] A.J. Canumalla, N. Al-Zamil, M. Philips, A.A. Isab, C.F. Shaw III, J. Inorg. Biochem. 85 (2001) 67–76. [11] L. Messori, G. Marcon, Met. Ions Biol. Syst. 42 (2004) 385–425. [12] A. Mun˜oz, F. Laib, D.H. Petering, C.F. Shaw III, J. Biol. Inorg. Chem. 4 (1999) 495–507.
583
[13] R.A. Penneman, R.R. Ryan, Acta Cryst. B28 (1972) 1629–1632. [14] C. Bertinotti, A. Bertinotti, Acta Cryst. B26 (1970) 422–428. [15] K.N. Raymond, P.W.J. Corfield, J.A. Ibers, Inorg. Chem. 7 (1968) 1362–1372. [16] P.A. Albore´s, L.D. Slep, T. Weyhermu¨ller, E. Rentschler, L.M. Baroldo, J. Chem. Soc. Dalton Trans. (2006) 948–954. [17] R.E. Cramer, D.W. Smith, W. VanDoorne, Inorg. Chem. 37 (1998) 5895–5901. [18] P.I. Amrhein, S. Drouin, C.E. Forde, A.J. Lough, R.H. Morris, Chem. Commun. (1996) 1665–1666. [19] S.M. Contakes, S.C.N. Hsu, T.B. Rauchfuss, S.R. Wilson, Inorg. Chem. 41 (2002) 1640–1648. [20] G.A. Lawrance, M. Maeder, Y.M. Neuhold, K. Szacilowski, A. Barbieri, S. Stasicka, J. Chem. Soc. Dalton Trans. (2002) 3649– 3655. [21] W.-F. Yueng, W.-T. Wong, J.-L. Zuo, T.-C. Lau, Chem. Commun. (2000) 629–631. [22] R. Uso´n, A. Laguna, E.J. Ferna´ndez, M.E. Ruiz-Romero, P.G. Jones, J. Lautner, J. Chem. Soc. Dalton Trans. (1989) 2127– 2131. [23] J.R. Roscioli, L.R. McCunn, MS. Johnson, Science 316 (2007) 249– 254. [24] H. Schmidbaur, Gold Bull. 33 (2000) 3–10. [25] P. Pyykko¨, Angew. Chem., Int. Ed. 43 (2004) 4412–4456. [26] J.E. Huheey, E.A. Keiter, R.L. Keiter, Inorganic Chemistry, fourth ed., Harper Collins, NY, 1993. [27] T. So¨hnel, R. Brown, L. Kloo, P. Schwerdtfeger, Chem. Eur. J. 7 (2001) 3167–3173. [28] M. Hargittai, A. Schulz, B. Re´ffy, M. Kolontis, J. Am. Chem. Soc. 123 (2001) 1449–1458. [29] W.C. Hamilton, J.A. Ibers, Hydrogen Bonding in Solids, W.A. Benjamin, Inc., 1968, p. 219. [30] W.J. Dulmage, W.N. Lipscomb, Acta Cryst. 4 (1951) 330–334. [31] R.E. Hoffman, D.F. Hornig, J. Chem. Phys. 17 (1949) 1163. [32] M. Pierrot, R. Kern, Acta Cryst. B25 (1968) 1685–1691.
Inorganic Biochemistry Journal of Inorganic Biochemistry 102 (2008) 576–583 www.elsevier.com/locate/jinorgbio
Proton-linked bi- and tri-metallic gold cyanide complexes observed by ESI-MS spectrometry Philip M. Yangyuoru, James W. Webb, C.F. Shaw III * Department of Chemistry, Illinois State University, Normal, IL 61791-4160, United States Received 29 June 2007; received in revised form 15 October 2007; accepted 26 October 2007 Available online 22 November 2007
Abstract Electrospray ionization spectra of potential cyanide-containing gold-drug metabolites revealed additional, weak, unanticipated peaks at approximately twice the mass of the gold(I) and gold(III) cyanide complexes. The exact masses correspond to proton-linked bimetallic complexes, [H{Au(CN)m}2], (m = 2,4). Further investigation revealed a total of 12 examples, including trimetallic complexes, [H2{Au(CN)m}3]; mixed species with two complexes, [H{Au(CN)2}{Au(CN)4}]; and thiolato species, [H{(RS)Au(CN)3}2]. trans[AuX2(CN)2Cl2] and trans-[AuX2(CN)2Br2] generated 35Cl/37Cl and 79Br/81Br isotopic patterns for the protonated bi- and tri-metallic analogues which were in good agreement with the presence of four or six halide ligands, respectively. Concentration-dependent studies demonstrated that the signals are independent of the solution concentrations of mono-metallic precursors, suggesting formation in the gas phase during or following droplet desolvation. Ó 2007 Elsevier Inc. All rights reserved. Keywords: Electrospray ionization mass spectrometry; Gold(I); Gold(III); Cyanide; Proton-linked
1. Introduction Gold-based anti-arthritic drugs have been used for nearly eight decades [1–3] and are considered to be disease-modifying drugs, as opposed to palliative treatments. One of the known metabolites is aurocyanide ([Au(CN)2]), which has been detected in blood and urine of patients using three different drugs: myochrysine (gold sodium thiomalate), solgonal (gold thioglucose) and auranofin (2,3,4,6-tetra-O-acetyl-1-thio-b-D-glucopyranosato-S-(triethylphosphine)gold(I)) [4,5]. Evidence for the conversion of gold(I) to gold(III) has been reported in mice [6] and men [7]. The enzyme myleoperoxidase has been implicated in both processes [8–10] and this has led to the hypothesis that auricyanide ([Au(CN)4]) might also be a metabolite [10]. A variety of gold(III) complexes show very promising anti-tumor activity [11], but are subject to poten-
*
Corresponding author. Tel.: +1 309 438 7207; fax: +1 209 438 5538. E-mail address: [email protected] (C.F. Shaw III).
0162-0134/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2007.10.020
tial reduction to gold(I), giving impetus to studies of the mechanisms of the biological reduction mechanisms. While studying the reduction of auricyanide by thiols, mass spectra revealed negative-ion peaks resulting from complexes containing two or three gold ions, which could not be accounted for by a ligand exchange of thiolate for cyanide (P.M. Yangyuoru, J.W. Webb, C.F. Shaw III, this issue). The masses of these peaks correspond to formation of bi- and tri-metallic complexes, [Hn{Au(CN)m}n+1]. It was therefore interesting to investigate these species further because such proton-linked clusters have not been documented yet in the solution chemistry of gold nor, to our knowledge, in ESI-MS spectra of other metal cyanides. 2. Experimental Reagents were obtained as follows: reduced glutathione from Sigma–Aldrich; cysteine, Sigma Chemical Company; cysteine methyl ester (Lot # A016240201), N-acetylcysteine (Lot # A018477901) and formic acid (Lot # A019673801) from Acros Organics; HPLC-grade acetonitrile (ACN,
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Lot # 054754) from Fisher Chemicals Company. K[Au(CN)2Cl2] was prepared by the method of Canumalla et al. [10], kmax = 217, 287 nm (lit 217, 287 nm), m/z = 319.0, 321,0, 323.0; mCN = 2175 cm1. K[Au(CN)2Br2] was a generous gift of Dr. Anna Cannumalla. 2.1. ESI-mass spectrometry ESI-mass spectra were recorded on a Hewlett Packard (Agilent) LC–MS (Series 1100 LC–MSD with a UV–vis diode array detector) set in the negative ion mode with 40 V fragmentation voltage, 3500 V capillary voltage; 10 L/min N2 flow at 25 psi; ionization temperature 250 °C. Samples (10.0 lL of 0.1–100 mM concentration of gold complex) were injected into the ESI chamber. The flow rate was set at 0.400 mL/min. Mobile phase solvents were typically (A) 0.1% formic acid in nanopure water and (B) 0.1% formic acid in acetonitrile (ACN). Unless otherwise indicated, a mobile phase composition of 50% A and 50% B was used throughout this research. Solvent composition and spray parameters were fixed for most of the analyses. Using the auto sampler, the delay time between mixing and the first spectrum was typically 3 min. 2.2. [{Au(CN)m}n+1Hn] Solutions typically containing 2.00 mM [Au(CN)m] (m = 2, 4) in nanopure water were analyzed by ESI mass spectrometry over the range 100–1000 m/z. To study the possibility of the formation of mixed clusters, solutions containing both 1.00 mM [Au(CN)4] and 1.00 mM [Au(CN)2] in nanopure water were analyzed by ESI-MS. Apart from nanopure water, studies were also conducted in acetonitrile + 0.1% HCOOH solvent medium. 2.3. [Au(CN)m] and [H3O+] concentration-dependence studies Varying concentrations (1.00, 2.00, 4.00, 5.00, 50.0 and 100 mM) of only [Au(CN)4] or equimolar mixtures of [Au(CN)4] and [Au(CN)2] in nanopure water were analyzed. Studies were done in the acetonitrile + 0.1% HCOOH medium and compared to identical studies in nanopure water to determine whether the formation of the protonated clusters is affected by the solvent and the [H3O+] concentration. 2.4. [Hn{Au(CN)2X2}n+1] Aqueous solutions ranging from 2.3 to 22.8 mM of K[Au(CN)2Br2] and K[Au(CN)2Cl2] were examined. The chloride complex was used to study the effect of the fragmentation voltage (0–120 V in 20 V steps) on the intensity of the bi- and tri-metallic species and the extent of reductive elimination of X2 from the anionic complexes.
577
2.5. [Hn{Au(SR)(CN)3}n+1] An extensive library of ESI-MS spectra obtained during examination of the auricyanide reduction (P.M. Yangyuoru, M.S. Thesis, Illinois State University, 2007) was scanned for bi- and tri-metallic thiolate-substituted complexes. Only bi-metallic complexes with RS = cysteine or O-methyl-cysteine were noted. 3. Results While studying the reduction of auricyanide by thiols, Eq. (1): [Au(CN)4 ] + 2G SH ! [Au(CN)2 ] + GSSG2 + 2HCN ð1Þ several ESI mass spectra revealed peaks of previously unrecognized complexes containing two or three gold ions. Fig. 1, for example, exhibits peaks corresponding to the reactants (GSH, m/z = 306.0 and [Au(CN)4], m/z = 301.0), two intermediates ([Au(SGH)(CN)3], m/z = 581.0 and [Au(SCy-Gly)(CN)3], m/z = 452.0) and a trace of the product ([Au(CN)2], m/z = 249.0). There is an additional peak at m/z = 603.0, approximately twice the mass of aurocyanide, but a dimer would have a mass of 602.0 and a net 2 charge, and would therefore have an m/z value of 301.0, indistinguishable from the monomeric complex. Nor did this peak correspond to any of the possible reactants or intermediates containing a thiolate ligand in the complex. The observed m/z of 603.0 can be accounted for by a proton linking two [Au(CN)4] ions to form a bimetallic cluster with a net 1 charge, as detected by ESI-MS. The following protonation reaction, Eq. (2), was therefore suggested to generate the cluster [H{Au(CN)4}2]: 2[Au(CN)4 ] + Hþ ! [HfAu(CN)4 g2 ]
ð2Þ
If the postulated complex and its formation are correctly defined, it should form independently of the glutathionederived species involved in the overall auricyanide reduction reaction, Eq (1). ESI mass spectra of [Au(CN)4] obtained before the reduction process also exhibited, upon reexamination, minor peaks corresponding to the bi-metallic complex, [{Au(CN)4}2H] (Fig. 2). Two additional proton-linked clusters were observed in ESI mass spectra of pure K[Au(CN)2] in nanopure water (Fig. 3). In addition to [Au(CN)2] (m/z = 249.0), a bimetallic gold cluster, [H{Au(CN)2}2] (m/z = 498.9), analogous to the bimetallic gold(III) species, was present. In addition, a peak at m/z = 749.2 resulting from the formation of [H2{Au(CN)2}3] was observed. The protonation reactions for [Au(CN)2] (Eqs. (3) and (4)) correspond to peaks observed at m/z = 498.9 and 749.2, respectively:
578
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Fig. 1. Negative-ion electrospray mass spectrum of a reaction mixture initiated with 0.100 mM [Au(CN)4] and 0.200 mM glutathione (GSH) in nanopure water and recorded after mixing plus auto sampling time of 3 min. The m/z values are rounded to whole numbers in the figures, but given to tenths of a mass unit in the text and typically agree with calculated values ±0.1 unit.
Fig. 2. Mass spectrum of 1.00 mM K[Au(CN)4] in nanopure water. The intensity scale was expanded to maximize the low-intensity peaks. The average intensity of [Au(CN)4] peak was about 250,000 counts.
2½AuðCNÞ2 þ Hþ ! ½HfAuðCNÞ2 g2
þ
ð3Þ
3½AuðCNÞ2 þ 2H ! ½H2 fAuðCNÞ2 g3
ð4Þ
To our knowledge, the formation of protonated bi- and trimetallic gold clusters of the form [{Au(CN)x}n+1Hn] has not been reported to date. Therefore, additional examina-
tion of these species was of interest. Because self-complexation of either [Au(CN)4] or [Au(CN)2] occurs independently, leading to formation of bi- and tri-metallic complexes, one would predict that mixed clusters of gold(I) and gold(III) should form in solutions containing both [Au(CN)4] and [Au(CN)2]. The homoleptic gold(I) and
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
579
Fig. 3. Mass spectrum of 1.00 mM K[Au(CN)2] in nanopure water. The intensity scale was expanded to maximize the low-intensity peaks. The average intensity of [Au(CN)2] peak was about 300,000 counts.
gold(III) cyanides were mixed and examined by ESI-MS in two different solvent systems: nanopure water (not shown) and acetonitrile + 0.1% HCOOH (Fig. 4). In addition to the complexes [{Au(CN)2}2H], [{Au(CN)2}3H2], and [{Au(CN)4}2H], a mixed bi-metallic complex (m/z = 551.0) was formed from the two monomeric complexes and a proton: [Au(CN)2 ] + [Au(CN)4 ] + Hþ ! [HAu(CN)2 Au(CN)4 ] ð5Þ
The potassium linked-complexes of auricyanide (Eq. (6)) were also observed at m/z = 641.1 (Figs. 2 and 4). These
potassium complexes are analogous to the protonated complexes and have the same 1 charge as the protonated complexes. 2[Au(CN)4 ] + Kþ ! [KAu(CN)42 ]
ð6Þ
Potassiated and sodiated complexes of peptides and proteins are commonly encountered in positive-ion ESI mass spectra [12]. Although K[Au(CN)2] was used in these studies, no analogous potassium-linked aurocyanide clusters were formed. Two examples of bimetallic clusters of tricyano(thiolatoS)gold(III) complexes were observed during reduction
Fig. 4. Mass spectrum of 1.00 mM [Au(CN)4] and 1.00 mM [Au(CN)2] in acetonitrile + 0.1% HCOOH. The intensity scale was expanded to emphasize the low-intensity peaks. The noise of the spectrum is due to the expansion of the less intense peaks. The average intensities of the [Au(CN)4] and [Au(CN)2] peaks were 200,000 and 122,000 counts, respectively.
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583 10000 8000
Intensity
reactions (analogous to Eq. (1)) in which cysteine (Fig. 5) or O-methylcysteine (not shown) replaced glutathione. The mass spectrum in Fig. 5, taken one hour after mixing cysteine and auricyanide, exhibited a peak at m/z = 791.6, which corresponds to [H{Au(SCy)(CN)3}2]. To further explore the phenomenon, solutions of the dihalodicyanogold(III) complexes of chloride and bromide were examined. The mass spectra for K[Au(CN)2Cl2] showed three parent ion peaks for [Au(CN)2Cl2] at m/ z = 319.0, 321.0 and 323.0, consistent with the 35Cl (0.76 abundance) and 37Cl (0.24 abundance) isotope pattern predicted for two chloride ligands: 0.57:0.37:0.06. Two peaks assignable as proton-linked cluster species were observed, Fig. 6. An isotopic pattern at m/z = 638.7, 640.8, 642.8 and 644.8 arises from the four chlorides present in the bimetallic complex [H{Au(CN)2Cl2}2], Fig. 6 (upper), and corresponds to the intensities (0.31:0.39:0.19:0.04) predicted for the most intense peaks. As expected, the m/z = 640.8 peak is most intense of the four, but the remaining 6 weaker peaks are not observed. A weaker isotopic pattern at m/z = 958.7, 960.7 962.7, 964.7 and 965.7, Fig. 6 (lower), corresponds to the five most intense peaks with abundances of 0.17: 0.33:0.27:0.11:0.03 of the 11 expected for the presence of six chloride ions. The principal isotopic peaks for these proton-linked bi- and tri-metallic species are separated by the mass difference (2.0) of the chlorine isotopes, as expected, for monoanions. The exact agreement of masses, and reasonable agreement with the predicted abundances provide strong confirmation that the species being observed correspond to the protonated bi- and tri-metallic species: [H{Au(CN)2Cl2}] and [H2{Au(CN)2Cl2}3]. In some runs, there also a weak peak at m/z = 249.0, which results from gas phase decomposition [Au(CN)2Cl2] by reductive elimination of chlorine at elevated temperatures in the electrospray chamber:
-
[H{Au(CN)2Cl2}2]
6000 4000 2000 0 625
630
635
640
645
650
10000 8000
Intensity
580
-
[H2{Au(CN)2Cl2}3]
6000 4000 2000 0 950
955
960
965
970
m/z Fig. 6. Expanded scale spectrum of 22 mM aqueous K[Au(CN)2Cl2] in the m/z ranges for (upper) the protonated bimetallic complex; and (lower) the protonated trimetallic complex.
[Au(CN)2 Cl2 ] ! Cl2 + [Au(CN)2 ]
ð7Þ
The absorbance peaks characteristic of [Au(CN)2] at 204, 211, 230 and 239 nm were absent from the UV–vis spectrum of the K[Au(CN)2Cl2] sample used. The intensity of the aurocyanide peak increased and decreased with changes in the fragmentation voltage (0–70 V), indicating that the aurocyanide formed during the mass spectral analysis. Additionally, a single peak at m/z = 499.0 resulting from the formation of [H{Au(CN)2}2], and an isotopic pattern at m/z = 568.9, 570.9 and 572.9 demonstrating the formation of the mixed bimetallic complex [H{Au(CN)2Cl2}{Au(CN)2}] were observed.
Fig. 5. Mass spectrum of a 1.00 mM [Au(CN)4] + 1.00 mM CySH reaction mixture in nanopure water measured after 1 h.
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
Similar monometallic and bimetallic complexes were observed in the spectra of K[Au(CN)2Br2] (not shown). The isotopic abundances of bromine, 79Br (0.507) and 81 Br (0.493), are sufficiently close to being equal that the patterns closely approximate binomial distributions for any number of bromine ligands. The parent ion, [Au(CN)2Br2] was a very strong ‘‘triplet” (m/z = 407.8, 409.8, 411.8). The bimetallic species, [H{Au(CN)2Br2}2] exhibited the expected five peaks (m/z = 814.6, 816.6, 818.6, 820.6, 822.6). For the trimetallic species [H2{Au(CN)2Br2}3], a weak but clearly interpretable isotope pattern corresponding to the five strongest bands (m/z = 1224.4, 1226.4, 1228.4, 1230.4, 1232.4) of the predicted seven-peak pattern were present, providing the third example of a trimetallic species in this study. In these spectra, signals for [H{Au(CN)2Br2}{Au(CN)2}] (m/z = 656.8, 658.8, 660.8), [H{Au(CN)2}2] (m/z = 499.0) were also clearly identified. Table 1 summarizes the twelve protonated bi- and trimetallic complexes observed in solutions of pure compounds or during the reduction of auricyanide by thiols. In every case the agreement between the mass of the mono-metallic species and the formulation as a protonlinked bimetallic or trimetallic species was within ±0.1 mass unit. Two possible origins of the proton-linked bi- and trimetallic cyanogold species were considered. They might arise as the result of equilibria in solution, or they may form in the ESI ionization chamber during desolvation of the analyte droplets. Cyanide complexes of gold and many other metal ions are highly water soluble, implying strong hydration. H-bonds to cyanide are a significant source of solvation energy contributing to the high solubilities of cyanide complexes, and are also frequently generated in crystal structures by waters of crystallization or other Hbond donors: e.g., [Au(CN)4] [13,14], [Ni(CN)5]3 [15], [HO-Cr(cyclam)-NC-Cr(CN)5] [16] [16-pyrimidinium crown-4][Au(CN)2]4 6.5H2O [17]. In some cases, cyanide ligands of low-spin ferrous complexes (which are d6 and hence kinetically inert) can be protonated in solution, e.g., [Fe(R2PCH2CH2PR2)2H(CNH)]+ [18] and [Fe(CO)(CN)4(CNH)]2 [19]. Metal–cyanide complexes
form ion pairs with cations in aqueous solution: e.g., a series of {[cation]+[Fe(CN)5(NO)]2} complexes, where cation+ = H+, Li+, Na+, K+, Rb+, Cs+ (the first being another protonated cyanide as in the previous examples) have been observed by ESI-MS and shown by 13C NMR spectroscopy to involve direct cyanide–cation interactions [20]. Finally, numerous linearly-bridged cyanides in bimetallic complexes also demonstrate the nucleophilicity of coordinated cyanide ligands: e.g., [(cyclen)Cu-(l-NC)Au(CN)]+ [21], [l-CN-(AuC9H11)2] [22], trans-[(HO)Cr(cyclam)-(l-NC)-Cr(CN)5] [16]. These selected examples provide four clear lines of evidence for the nucleophilicity and basicity (in both the Lewis and Bronsted-Lowry senses) of cyanide nitrogen atoms. If the protonated bi- and tri-metallic gold complexes form in solution, their concentrations should be dependent on the concentration of the mono-metallic species as well as the pH of the solution, according the equilibrium reaction: 2[Au(CN) ] + Hþ ½HfAuðCNÞ g ð8Þ 4
4 2
Therefore, concentration-dependent studies of the formation of the protonated bi- and tri-metallic gold cyanides were conducted using ESI mass spectrometry. The expected concentration dependence for formation of the bimetallic clusters from the homoleptic gold cyanides in solution can be derived for the postulated equilibrium (Eq. (8)) by solving the Keq expression (Eq. (9)) for the concentration of the bimetallic species, as shown in Eq. (10):
½HfAuðCNÞx g2
þ
¼ ðKeq ½H Þ
m/z
[H{Au(CN)4}2] [H{Au(CN)2}2] [{Au(CN)4}H{Au(CN)2}] [H{CySAu(CN)3}2] [H{O-MeCySAu(CN)3}2] [H{Au(CN)2Cl2}2] [H{Au(CN)2Cl2}{Au(CN)2}] [H{Au(CN)2Br2}2] [H{Au(CN)2Br2}{Au(CN)4}]
603.0 499.0 551.0 791.6 819.2 640.8 568.9 818.6 656.8
Trimetallic species [H2{Au(CN)2}3]
m/z 749.2
ð9Þ
2 ½AuðCNÞx
ð10Þ
Eq. (10) predicts an increase in the protonated bimetallic auricyanide species, [H{Au(CN)4}2], according to the square of the [Au(CN)4] concentration, Fig. 7. Analysis of 0.100–100 mM [Au(CN)4] solutions revealed only marginal changes in the concentrations of [{Au(CN)4}2H], erratically increasing up to a maximum at 20.0 mM [Au(CN)4] and then decreasing. When plotted on the 600000
Intensity
Bimetallic species
2
Keq ¼ ½HfAuðCNÞx g2 =f½AuðCNÞx ½Hþ g
500000 Table 1 m/z valuesa of protonated bi- and tri-metallic gold cyanide complexes
581
Measured Intensity
400000
Expected Intensity
300000 200000 100000 0 0
[H2{Au(CN)2Cl2}3]
960.7
1
3
2
Au(CN)4-,
[H2{Au(CN)2Br2}3]
1228.4
a Values for the most intense peak of the chlorine and bromine complexes are given.
4
5
mM
Fig. 7. Intensities of [H{Au(CN)4}2] predicted for a solution-phase equilibrium over the concentration range 1–5 mM [Au(CN)4] (- - -) vs the observed intensities (—). The nearly constant signal does not reflect the anticipated 25-fold increase of the protonated bimetallic species over that range.
582
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
same scale as the predicted behavior, the experimental intensities are essentially constant, Fig. 7 Similarly, when equimolar mixtures of [Au(CN)4] and [Au(CN)2] were examined in either nanopure water or acetonitrile +0.1% HCOOH, the peak intensities for the bi-metallic clusters were not significantly dependent on the concentration product {[Au(CN)2]] * [Au(CN)4]]} applicable to the mixtures. Nor did the pH of the solvent medium affect the extent of the protonated-cluster formation observed in acetonitrile +0.10% formic acid, compared to similar concentrations in nanopure water. The lower dielectric constant of the acetonitrile and the increased hydrogen ion concentration should favor the formation of the bimetallic species, which was not observed. This pH-independence is also inconsistent with a preexisting equilibrium of the bimetallic complexes, and suggests that they form during or subsequent to the droplet desolvation in the gas phase. The observation of the symmetric complex [H{Au(CN)2}2] and the mixed bimetallic complex [H{Au(CN)2Cl2}{Au(CN)2}] in the spectra of K[Au(CN)2Cl2] provides further evidence that the bimetallic species arise during the mass spectral analyses. The aurocyanide ion observed in the mass spectra of this compound is known (from infrared analysis) to be absent from the solution, but forms by reductive elimination of chlorine at higher temperatures during (and possibly subsequent to) desolvation. Therefore, the two bimetallic species which contain aurocyanide must also be formed subsequent to injection of the sample, eliminating the existence of the clusters as an equilibrium phenomenon in solution and strongly supporting the conclusion that these bi- and tri-gold clusters are formed by a novel process during desolvation of the droplets or in the gas phase. 4. Discussion What bonding forces cause the proton and two cyanide complexes to associate with one another in the gas phase? The most likely explanations are hydrogen bonding [23] and/or aurophilicity [24,25]. Aurophilic interactions are correlation effects between two or more gold atoms. The range of energies for aurophilic interactions, 7–12 kcal/ mol [24,25] falls within the typical range for hydrogen bonds 2–20 kcal/mol [26]. They give rise to aurated nonmetal complexes such as [N(AuPPh3)4]+, extremely stable gold cluster compounds, and gold–gold interactions in crystal structures [24]. These interactions, however, are strongest between d10 gold(I) centers, weak between gold(I) and gold(III) and negligible between d8 gold(III) centers [24]. On the other hand, intramolecular relativistic forces shorten the Au–Au bond distances in the Au2X6 dimers (X = Cl, I) [27,28]. It is possible, then, that gold(I) interactions are important in addition to hydrogen bonding for aurocyanide, but are less important contributors to stabilization of the mixed oxidation state complexes or dihalodicyano complexes, which have been observed here.
Cyanides are good hydrogen bond acceptors. HCN in the solid state is hydrogen bonded to itself to form linear head-to-tail chains HACBNH ACBNHACBNHACBN ˚ . The mN H stretching with an NH N distance of 3.2 A 1 frequency, 3132 cm is significantly decreased from that of individual molecules in the gas phase, 3312 cm1 [29– 31]. Auricyanide crystallizes as H[Au(CN)4]2H2O [13] and as K[Au(CN)4] H2O [14]. Crystal structures of both have been reported. The ‘‘protonated” salt is better formu lated as ½H5 Oþ 2 ½AuðCNÞ4 . In the structure each of the four terminal OH groups of the cation is hydrogen bonded to a cyanide from a different auricyanide ion, and each cyanide nitrogen of a given anion is hydrogen bonded to four distinct H5 Oþ 2 cations, forming an extended three dimensional network. The hydrogen bonds are classical, nearly ˚ ) aligned with the Au–C„N linear OH N bonds (2.75 A axis. In contrast, the potassium salt contains a single water of hydration. It forms two crystallographically independent hydrogen bonds to cyanide nitrogens on different aurocyanide ions. One H-bond is somewhat bent ˚ ) and aligned slightly off (\OH N = 155.8° and 2.94 A axis to the Au C„N. The second H-bond to an independent auricyanide molecule is nearly linear and longer ˚ ), but has an angle 120° (\OH N = 155.8° and 3.14 A to the Au–C„N axis. Many other metal cyanide salts crystallize with hydrogen bonded waters of hydration as discussed above. Thus, the formation of symmetrical hydrogen bonds AAuACBNHNBCAAuA provides a reasonable explanation and would be common to the gold(I) and gold(III) species. Recently, Roscioli et al. described analogous gas-phase cations formed by hydrogen bonding of a proton to two oxygens of alcohols and ethers, R1R2O H OR3R4, where R1, R3 = H, CH3, or C2H5, and R3, R4 = CH3 or C2H5 [23]. If aurophilic effects are also present, the proximity of the gold atoms to one another would place severe restraints on the hydrogen bonds, which could not be colinear with the Au–C„N axes, but would have to be directed at the N atoms of the cyanides at oblique angles. Models for such non traditional H-bonds are found in the structures of K[Au(CN)4] H2O [14] discussed above, [HO-Cr(cyclam)NC-Cr(CN)5] [16] and especially H4Fe(CN)6 [32]. In the latter, intermolecular hydrogen bonds between the ferrocyanide ions exhibit oblique angles between the N H H axis and each Fe–C„N axis [32]. The observation that the intensities of the bimetallic and trimetallic peaks decrease as the ionizing voltage increases from 0 to 120 V supports the suggestion that these species are formed by a weak interaction such as the hydrogen bond, aurophilic interactions or relativistic effects. As is the case for the gas-phase proton-linked alcohols and ethers [23], the existence of these new proton-linked
P.M. Yangyuoru et al. / Journal of Inorganic Biochemistry 102 (2008) 576–583
bimetallic species poses interesting theoretical questions regarding the bonding forces that stabilize them. Acknowledgements C.F.S. thanks the ISU College of Arts and Sciences for a Research Enhancement Award in 2006. We thank Dr. Anna Canumalla and Ms. Jin Hwang for the synthesis of K[AuBr2(CN)2] and K[AuCl2(CN)2], respectively. References [1] C.F. Shaw III, Chem. Rev. 99 (1999) 2589–2600. [2] C.F. Shaw III, in: H. Schmidbaur (Ed.), Gold: progress in the chemistry, Biochemistry and Technology, J. Wiley & Sons, Chichester UK, 1999, pp. 259–308. [3] L. Messori, G. Marcon, Met. Ions Biol. Syst. 41 (2004) 279–304. [4] R.C. Elder, Z. Zhao, Y. Zhang, J.G. Dorsey, E.V. Hess, K. Tepperman, J. Rheumatol. 20 (1993) 268–272. [5] Z. Zhao, W.B. Jones, K. Tepperman, J.G. Dorsey, R.C. Elder, J. Pharm. Biomed. Anal. 10 (1992) 279–287. [6] D. Schumann, M. Kubicka-Muranyi, J. Mirtcheva, J. Gunther, P. Kind, E. Gleichmann, J. Immunol. 145 (1990) 2132–2139. [7] J. Verwilghen, G.H. Kingsley, L. Gambling, G.S. Panayi, Arthritis Rheum. 35 (1992) 1413–1418. [8] G.G. Graham, A.J. Kettle, Biochem. Pharmacol. 56 (1998) 307–312. [9] C.F. Shaw III, S. Schraa, E. Gleichmann, Y.P. Grover, L. Dunemann, A. Jagarlamudi, Metal-Based Drugs 1 (1994) 351–362. [10] A.J. Canumalla, N. Al-Zamil, M. Philips, A.A. Isab, C.F. Shaw III, J. Inorg. Biochem. 85 (2001) 67–76. [11] L. Messori, G. Marcon, Met. Ions Biol. Syst. 42 (2004) 385–425. [12] A. Mun˜oz, F. Laib, D.H. Petering, C.F. Shaw III, J. Biol. Inorg. Chem. 4 (1999) 495–507.
583
[13] R.A. Penneman, R.R. Ryan, Acta Cryst. B28 (1972) 1629–1632. [14] C. Bertinotti, A. Bertinotti, Acta Cryst. B26 (1970) 422–428. [15] K.N. Raymond, P.W.J. Corfield, J.A. Ibers, Inorg. Chem. 7 (1968) 1362–1372. [16] P.A. Albore´s, L.D. Slep, T. Weyhermu¨ller, E. Rentschler, L.M. Baroldo, J. Chem. Soc. Dalton Trans. (2006) 948–954. [17] R.E. Cramer, D.W. Smith, W. VanDoorne, Inorg. Chem. 37 (1998) 5895–5901. [18] P.I. Amrhein, S. Drouin, C.E. Forde, A.J. Lough, R.H. Morris, Chem. Commun. (1996) 1665–1666. [19] S.M. Contakes, S.C.N. Hsu, T.B. Rauchfuss, S.R. Wilson, Inorg. Chem. 41 (2002) 1640–1648. [20] G.A. Lawrance, M. Maeder, Y.M. Neuhold, K. Szacilowski, A. Barbieri, S. Stasicka, J. Chem. Soc. Dalton Trans. (2002) 3649– 3655. [21] W.-F. Yueng, W.-T. Wong, J.-L. Zuo, T.-C. Lau, Chem. Commun. (2000) 629–631. [22] R. Uso´n, A. Laguna, E.J. Ferna´ndez, M.E. Ruiz-Romero, P.G. Jones, J. Lautner, J. Chem. Soc. Dalton Trans. (1989) 2127– 2131. [23] J.R. Roscioli, L.R. McCunn, MS. Johnson, Science 316 (2007) 249– 254. [24] H. Schmidbaur, Gold Bull. 33 (2000) 3–10. [25] P. Pyykko¨, Angew. Chem., Int. Ed. 43 (2004) 4412–4456. [26] J.E. Huheey, E.A. Keiter, R.L. Keiter, Inorganic Chemistry, fourth ed., Harper Collins, NY, 1993. [27] T. So¨hnel, R. Brown, L. Kloo, P. Schwerdtfeger, Chem. Eur. J. 7 (2001) 3167–3173. [28] M. Hargittai, A. Schulz, B. Re´ffy, M. Kolontis, J. Am. Chem. Soc. 123 (2001) 1449–1458. [29] W.C. Hamilton, J.A. Ibers, Hydrogen Bonding in Solids, W.A. Benjamin, Inc., 1968, p. 219. [30] W.J. Dulmage, W.N. Lipscomb, Acta Cryst. 4 (1951) 330–334. [31] R.E. Hoffman, D.F. Hornig, J. Chem. Phys. 17 (1949) 1163. [32] M. Pierrot, R. Kern, Acta Cryst. B25 (1968) 1685–1691.