- Email: [email protected]
Preclinical characterization of anticancer gallium(III) complexes: Solubility, stability, lipophilicity and binding to serum proteins
JOURNAL OF
Inorganic Biochemistry Journal of Inorganic Biochemistry 100 (2006) 1819–1826 www.elsevier.com/locate/jinorgbio
Preclinical characterization of anticancer gallium(III) complexes: Solubility, stability, lipophilicity and binding to serum proteins Alexander V. Rudnev a, Lidia S. Foteeva a, Christian Kowol b, Roland Berger b, Michael A. Jakupec b, Vladimir B. Arion b, Andrei R. Timerbaev a,b,*, Bernhard K. Keppler a
b
Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Sciences, Kosygin St. 19, 119991 Moscow, Russia b Institute of Inorganic Chemistry, University of Vienna, Waehringer Str. 42, A-1090 Vienna, Austria Received 13 February 2006; received in revised form 11 July 2006; accepted 12 July 2006 Available online 25 July 2006
Abstract The discovery and development of gallium(III) complexes capable of inhibiting tumor growth is an emerging area of anticancer drug research. A range of novel gallium coordination compounds with established cytotoxic efficacy have been characterized in terms of desirable chemical and biochemical properties and compared with tris(8-quinolinolato)gallium(III) (KP46), a lead anticancer gallium-based candidate that successfully finished phase I clinical trials (under the name FFC11), showing activity against renal cell cancer. In view of probable oral administration, drug-like parameters, such as solubility in water, saline and 0.5% dimethyl sulfoxide, stability against hydrolysis, measured as the rate constant of hydrolytic degradation in water or physiological buffer using a capillary zone electrophoresis (CZE) assay, and the octanol–water partition coefficient (log P) providing a rational estimate of a drug’s lipophilicity, have been evaluated and compared. The differences in bioavailability characteristics between different complexes were discussed within the formalism of structure-activity relationships. The reactivity toward major serum transport proteins, albumin and transferrin, was also assayed in order to elucidate the drug’s distribution pathway after intestinal absorption. According to the values of apparent binding rate constants determined by CZE, both KP46 and bis(2-acetylpyridine-4,4-dimethyl-3-thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) (KP1089) bind to transferrin faster than to albumin. This implies that transferrin would rather mediate the accumulation of gallium antineoplastic agents in solid tumors. A tendency of being faster converted into the protein-bound form found for KP1089 (due possibly to non-covalent binding) seems complementary to its greater in vitro antiproliferative activity. 2006 Elsevier Inc. All rights reserved. Keywords: Anticancer drugs; Gallium(III) complexes; Drug-like properties; Serum protein-binding; Capillary electrophoresis
1. Introduction Complexation of gallium with organic ligands has been recognized as a promising strategy for creating tumorinhibiting therapeutics with a number of advantages over gallium salts regarding oral bioavailability [1–3]. Indeed, the evolved hydrolytic stability and membrane penetration ability render gallium complexes improved intestinal
*
Corresponding author. Tel.: +43 1 4277 52609; fax: +43 1 4277 9526. E-mail address: [email protected] (A.R. Timerbaev).
0162-0134/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2006.07.003
absorption functions compared to nitrate or chloride salt which leads to increased plasma concentrations of gallium. Owing to such benefits, as well as better antiproliferative properties, two oral compounds, tris(8-quinolinolato)gallium(III) (KP46) [4] and tris(3-hydroxy-2-methyl-4Hpyran-4-onato)gallium(III) [5] have been selected from a series of gallium complexes for clinical development. Likewise, bis(2-acetylpyridine-4,4-dimethyl-3-thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) (KP1089) was the first representative of the class of a-N-heterocyclic thiosemicarbazone complexes that has been assayed for antineoplastic activity in human tumor cell lines [6]. In
1820
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
vitro cytotoxic potency of KP1089 was found to be the highest among the gallium compounds tested. This finding provides evidence in favor of a modern metallodrug-developing concept [7] that implies combining a metal ion and a ligand which are both pharmaceutically active and aimed at attacking the same molecular target (most likely, ribonucleotide reductase in the case of gallium thiosemicarbazonates). Another complex with high gallium-induced anticancer activity in vitro, KP1438, is formed by the reaction of Ga(III) with a paulone derivative [8]. Still, gallium complexes are thought to be immature with regard to the number of compounds examined, especially if one takes into account the well-known activity of this metal against malignant tumors and a diversity of anticancer complexes of other metals such as platinum(II) or ruthenium(III). Furthermore, a consensus has not yet been reached on the structural effects of different organic ligands on the antineoplastic action of gallium. A successful lead-drug candidate selection program requires careful consideration of various chemical and biochemical assays in the identification of the most active, ‘hit’ compounds. The respective physicochemical characterization includes solubility, hydration, lipophilicity (log P), metabolic stability, metabolite profiles, protein-binding, and other testing. Conducted in parallel with in vitro antitumor inhibiting studies, such an early hit compound identification stage permits further selection of drug nominees with desirable absorption, distribution and metabolism properties. When properly designed it may bring about significant merits in shortening timelines and reducing costs between discovery and clinical development. The aim of the present study was the evaluation of the drug-like properties of a representative number of cytotoxic gallium complexes, with the objective of identification and optimization of drug candidates for further preclinical development. Both the investigational KP46 drug taken as a reference compound, several neutral and cationic thiosemicarbazonates with, respectively, 1:1 and 1:2 metal-toligand stoichiometry, including KP1089, and the gallium chelate with a N,N,O-binding site (KP1438) were selected for this systematic examination. Special emphasis was put on the comparative estimation of the reactivity toward serum proteins performing transport functions, as to the authors’ knowledge no relevant information exists in the literature [9]. For this purpose, as well as for the monitoring of hydrolytic conversion of some gallium complexes under scrutiny, capillary zone electrophoresis (CZE) was utilized as an up-and-coming tool for preclinical characterization of anticancer metallodrugs [10,11]. 2. Experimental section 2.1. Materials The structures of eight gallium complexes that were investigated in this work, along with their order-of-discovery names, are shown in Fig. 1. The compounds were syn-
thesized and characterized at the Institute of Inorganic Chemistry, University of Vienna as described elsewhere [6,8,12,13]. Human serum albumin (min. 96%; fraction V powder) and transferrin (min. 90%) were purchased from Sigma–Aldrich (Vienna, Austria) and Fluka (Buchs, Switzerland), respectively. Sodium dihydrogenphosphate, disodium hydrogenphosphate, sodium chloride, and sodium hydroxide, all of highest purity available, were obtained from Merck (Darmstadt, Germany). Dimethyl sulfoxide (DMSO) and n-octanol were the products of Fluka. All solutions for CZE were prepared using high-purity water obtained by distillation of deionized water. 2.2. Instrumentation For CZE experiments a CAPEL 105 instrument (Lumex, St. Petersburg, Russia) equipped with a variablewavelength UV detector was used. Data analysis was performed using a MultiChrom program (Ampersend, Moscow, Russia). A Polymicro Technologies (Phoenix, AZ, USA) fused-silica capillary’s dimension was 40 cm total length (30.5 cm to the detection window) and 75 lm I.D. For preparations of working solutions of proteins and sparingly soluble complexes, an ultrasonication bath (Sapphire, Moscow, Russia) was used. Mechanical shaking was done with a C2184 mechanical horizontal shaker (Russia). Incubation experiments were performed using a U-10 thermostat (Mechanik Pru¨fgera¨te, Medingen, Germany). 2.3. Procedures 2.3.1. Solubility Gallium complexes were shaken with a fixed volume of individual solvent [water, 0.9% (w/v) NaCl or 0.5% (v/v) DMSO] at room temperature for 2 h. The amounts of solid substances were taken so as to form the saturated solution. The supernatant was filtered through a 0.45-lm disposable membrane filter (Sartorius, Go¨ttingen, Germany) and the concentration of Ga in the resultant solution was determined (after an appropriate dilution) by inductively coupled plasma atomic emission spectroscopy (higher concentrations) or inductively coupled plasma mass spectrometry (lower contents) at the Institute of Microelectronics Technology and High-Purity Materials, Russian Academy of Sciences (Chernologolovka, Russia). The detection limits were 10–50 and 0.2–1 lg l1, respectively (depending on the dilution factor) and standard errors did not exceed 10%. 2.3.2. log P Weighted amounts of gallium complexes were partitioned between water and n-octanol for 2 h at room temperature by the shake flask method. Gallium concentrations in the aqueous phase before (C0) and after partitioning (Caq) were measured using the same analytical techniques, from which the partition coefficients were calculated as log P = log (C0 Caq)/Caq.
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
N
1821
N N
N
S
O
N
GaCl4
Ga
O
N
Ga N O
S N
N
N
KP46
N
KP1089 Br
H N N
NH N
O
Cl
Ga O N
Cl HN
Cl
N
Ga
N H
N
S N N
N
Br
KP1438
N
KP1492
N N
N S
N
N
S
N
Cl
PF6
Ga
Cl
Ga N
N N
S
N
N N
KP1495
N
KP1497
N
N
N
S
N
PF6
Ga N
N
N
S
N
N N
PF6
Ga
S
KP1500
N
N
N S
N
N N
N
KP1511 Fig. 1. Structures of gallium complexes.
N
1822
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
2.3.3. Capillary zone electrophoresis The capillary conditioning and cleaning protocols were as described in our previous work [14]. To improve the repeatability of peak-area signals used as response in kinetic experiments, the corrected peak-areas (A 0 ) were calculated by dividing the peak-area by the migration time. Sample introduction was performed by applying pressure injections at 10 or 20 mbar for 10 s. In all cases, the experiments were carried out at a constant temperature of 37 C using a liquid-thermostated capillary cassette. The applied voltage was 7 or 10 kV (for chloride-containing and -void electrolytes, respectively) at the injection end of the capillary. The wavelength of the UV detector was set to 200 nm. Electrolytes were passed through the membrane filter before use. 2.3.4. Hydrolysis rate The hydrolytic stability of gallium complexes in water and physiological buffer (10 mM NaH2PO4/Na2HPO4, 100 mM NaCl, pH 7.4) was evaluated by incubating a given solution at 37 C, taking aliquots continuously for CZE analysis, and monitoring a decrease of peak-area responses (A 0 ). Ten mM phosphate buffers (pH 7.4), containing no and 100 mM NaCl, were used as respective electrolytes. For KP46, the rate of aqueous hydrolysis was monitored after 1:1 dilution of its aqueous solution with 20 mM phosphate buffer (to give the phosphate concentration in the sample the same as in electrolyte). The kinetic series was repeated four times for each gallium complexhydrolysis medium combination to calculate the average rate constant (khyd) and the standard deviation. The halflife was calculated as s1/2 = ln 2/khyd. 2.3.5. Protein-binding kinetics All binding experiments were performed in 10 mM phosphate buffer, 100 mM NaCl, pH 7.4 as incubation solution, at 37 C and with 1 · 105 M KP46 or KP1089. The protein concentration in the reaction mixture was kept constant at 5 · 105 M. The kinetics of Ga-protein binding was assessed by recording the relative change in the protein signal due to adduct evolution (the detection method avail-
able cannot differentiate the adduct and protein peaks). When the interaction was fairly fast and hence the interval between two subsequent CZE measurements too short (to acquire a sufficient number of data points), the aliquots collected were placed in a refrigerator in order to cease the binding reaction. The apparent rate constant (k) was calculated by polynomial approximation of the A 0 -time plot (see Section 3.4), assuming that the binding reaction obeys the first-order kinetics and the following equation is valid: ln A 0 = kt. 3. Results 3.1. Solubility Yet before it is absorbed in the intestine and enters the bloodstream, an oral gallium complex is required to possess sufficient solubility in order to subsequently exhibit the cytotoxic activity. To underscore, almost half of new molecular entities synthesized by pharmaceutical companies have failed to be developed because of poor water solubility [15]. A summary of saturation solubility data is given in Table 1. It is important to note that the only value available in the literature for gallium complexes of interest, is water solubility of KP46, i.e., 22 mg l1 or 4.4 · 105 M [3]. A detailed analysis of solubility thresholds showed that all the complexes can be subdivided into two groups: those that are soluble in the micromolar concentration range and others, much less soluble compounds (6105 M). The first group includes gallium thiosemicarbazonates with 1:1 stoichiometry, KP1492 and KP1497, which are less hydrophobic molecules than the corresponding chelates with a metal-to-ligand ratio of 1:2, as well as KP1089 whose improved solubility is due to less hydrophobic counterion and 4N-substituents (two methyl groups). One should also be aware that regardless of the solvent nature, the 1:2 complex with GaCl 4 as counter-ion, namely KP1089, does rearrange in solution into the analogous 1:1 chelate, i.e., KP1492, and vice versa. For instance, according to NMR data [13] KP1089 exists in the aged solution as a ca. 1:2 equilibrium mixture of the original complex and
Table 1 Summary of drug-related physicochemical properties of gallium complexes Complexa
Solubility (mg l1 M1) Water
0.9% NaCl
0.5% DMSO
KP46 (502.18) KP1089 (723.87) KP1438 (993.80) KP1492 (361.93) KP1495 (711.35) KP1497 (387.97) KP1500 (739.40) KP1511 (709.37)
17.9/3.6 · 105 4.42b/6.1 · 103 5.0/5.0 · 106 2.84b/7.9 · 103 1.9/2.7 · 106 1.45b/3.7 · 103 2.4/3.3 · 106 12.6/1.8 · 105
1.0/2.0 · 106 2.47b/3.4 · 103 2.0/2.0 · 106 2.14b/5.9 · 103 3.7/5.2 · 106 0.52b/1.3 · 103 5.2/7.0 · 106 9.1/1.3 · 105
15.2/3.0 · 105 2.86b/3.9 · 103 0.2/1.6 · 107 – 2.0/2.9 · 106 1.40b/3.6 · 103 13.3/1.8 · 105 5.2/7.3 · 106
a b
Given in parentheses is the molecular mass in Da. In g l1.
log P
0.88 1.15 1.10 1.40 0.95 0.69 0.05 0.47
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
3.2. Octanol–water partition coefficients The efficient penetration through biomembranes is a prerequisite for a gallium complex (like every oral drug) on the way from intestine to bloodstream and then to the tumor cell. The ability of a potential drug candidate to cross these barriers until it binds to the target and induces the desired response is most often described by the octanol–water partition coefficient. Shown in Table 1 are log P values collected to estimate the comparative distribution of gallium complexes between water and n-octanol as the solubilizing medium mimicking the interior part of biological membranes. Although characteristic of only the intact complexes (not their metabolites or protein adducts), such lipophilicity thresholds allow the bioavailability estimation for the compounds of interest being applied as anticancer drugs. Indeed, KP1438, KP1495, and KP46 do prefer a non-aqueous over an aqueous environment, behaving as essentially non-polar compounds. KP1500, KP1487, KP1089 and KP1492 all manifest a tendency of being predominately partitioned into water and thereby could hardly be expected at high concentration across the cell membrane. The membrane permeation potency of KP1511 can be characterized as intermediate one. Notably, no definite relationship between the structure and lipophilicity properties was observed for a given selection of compounds. Also to be pointed out are quite different log P values of closely related KP1497 and KP1511 compounds.
3.3. Hydrolytic stability As soon as a gallium complex (generally speaking, any metal complex) encounters body-fluid circumstances, it would inevitably be the subject of hydrolysis. It is this unwanted process that posed clinical limitations on gallium salts and gave rise to the development of tumor-inhibiting gallium complexes. Therefore, assessing survival rates is a basic requirement in systematic metallodrug discovery process, and CZE offers a rapid, cost-efficient, highthroughput, and reliable screening method for such measurements. This was demonstrated in the monitoring of hydrolysis kinetics for several platinum(II) [16,17] and ruthenium(III) [14,18] coordination compounds. Typical electropherograms recorded for KP46 and KP1089 are shown in Fig. 2. As follows from the data of Table 2, both
0.08
Absorbance (mAU)
KP1492. Therefore, the respective solubility figures represent rather a mixture of two compounds, and the corresponding rearrangement is partly responsible for better solubility of KP1089 compared to other, more stable 1:2 thiosemicarbazonates. The cationic thiosemicarbazonates with the hexafluorophosphate counter-ion (KP1495, KP1500, and KP1511) display considerably lower solubility. This is also because their derivatives with piperidinyl and pyrrolidinyl groups at 4N-position appear to be less solubilized in essentially aqueous media. In this regard, the analysis of solid-state crystal structures of compounds is anticipated for the presence of intermolecular interaction. What is likewise noteworthy is that compounds KP1511 and KP1495, which have different substituents at N4 (in terms of thiosemicarbazide derivatives nomenclature), demonstrate such dissimilar solubilities (and also lipophilicities; see below). The lowest solubility of KP1438 among the complexes studied can be explained taking into account a more developed structure of respective ligands, containing several hydrophobic aromatic entities (see Fig. 1). Comparison of solubility figures obtained for different biologically relevant media revealed that the gallium complexes were generally more soluble in water than in the saline and 0.5% DMSO, whereas the relative solubility in the latter two solutions varied from one complex to another without distinctive relation to its structural type.
1823
0.05
0
1
2
3
Time (min) Fig. 2. Electropherograms of aqueous solutions of: (a) KP46 and (b) KP1089. Samples: (a) 1.8 · 105 M KP46 and (b) 1 · 104 M KP1089. Carrier electrolyte: 10 mM phosphate buffer, pH 7.4. Voltage: 10 kV (current, 20–21 lA). Sample introduction: (a) 200 mbar s and (b) 100 mbar s. Other CZE conditions, see Section 2. GaLþ 2 ; (2) GaCl4 [for L assignment, see Fig. 1].
Table 2 Hydrolysis rate constants and half lives (n = 4) Complex
KP46 KP1089
Watera
Physiological buffer
khyd ± SD (·105, s1)
s1/2(h)
khyd ± SD (·105, s1)
s1/2(h)
4.8 ± 0.9 4.3 ± 0.5
4.0 4.5
1.4 ± 0.5 3.7 ± 0.9
14.2 5.5
a pH of aqueous solutions of KP46 and KP1089 was 3.8 and 6.6, respectively.
1824
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
gallium complexes investigated are fairly stable, retaining their integrity at a level of 50% for at least several hours. In view of the afore-mentioned rearrangement of KP1089 that proceeds immediately after dissolution, the corresponding k values represent rather overall stability than stability against hydrolysis. At any rate, the decomposition of complexes measured under conditions simulating an extracellular solution environment (pH 7.4) is slower than in water (ca. 3 times for KP46). Also worthwhile of special noting is at average lower hydrolysis rate of KP46. (Our attempts to assess another complex, KP1438, were unsuccessful because of its extremely low solubility.) 3.4. Binding kinetics to albumin and transferrin In the bloodstream, gallium complexes may form (stable) adducts with plasma proteins, and such interactions would also affect the bioavailability and the pharmacokinetic profiles of these compounds. Good tolerance to proteinaceous samples and short analyses time make the CZE technique very convenient for kinetic characterization of platinum group metallodrugs participating in protein-binding processes [14,19,20]. As can be seen in Table 3, there are notable differences in the binding rates of KP46 and KP1089 expressed as the acceptably accurate values of k, which were by a factor of 3–4 higher for the latter complex. Furthermore, there was also some dissimilarity in the reaction kinetics toward different proteins. The fact that transferrin binding is kinetically more favorable, also appearing Table 3 Reactivity of gallium complexes toward different proteins (n = 3–5) Complex
KP46 KP1089
Albumin
Transferrin
k ± SD (·105, s1)
k ± SD (·105, s1)
3.3 ± 0.1 11.2 ± 2.1
5.0 ± 0.8 21.1 ± 3.4
Corrected peak area
4
1 3
2
2
1
0
10
100
150
200
250
300
Time (min) Fig. 3. Binding curves exemplifying the rate of KP46 interaction with: (1) albumin and (2) transferrin. Incubation and CZE conditions, see Section 2.
from a steeper course of the corresponding binding curve shown for KP46 in Fig. 3, does seem apparent accounting for the gallium cation resemblance to iron(III). 4. Discussion The core properties required to estimate absorption, distribution, and transport of metal complexes in the body are solubility, lipophilicity, stability, and affinity toward transport proteins. Determinations of these properties are of crucial importance in anticancer metallodrug research as they help to select a lead candidate for further preclinical investigation and to design more active and/or less toxic compounds. Most of the gallium complexes which are considered, in a comparative manner, in this study are deficient in such screening assays, and this seems to be a bottleneck in their discovery process. An appropriate hydrophilic/lipophilic balance is the drug’s attribute that affords its sufficient plasma concentration after oral administration. From the analysis of solubility and log P data (see Table 1), one can conclude that KP46 possesses the most striking features with respect to its bioconcentration and membrane permeability. On the other hand, KP1438 would unlikely be sufficiently bioavailable to express the in vivo cytotoxicity (its in vitro tumorinhibiting activity is prominent at a 1 · 104 M level [8]) because of formulation difficulties related to solubilization. In addition, the complex’s lipophilicity might be too high to permit its passage through the cell membrane. Among gallium thiosemicarbazonates, KP1089, KP1492, KP1497, and KP1500 are marginally promising as valid oral drug nominees, oppositely, because of poor lipophilicity. As such they will not be very bioavailable after oral administration, and hence are to be applied via intravenous route. Such a conclusion seems especially important for KP1089 which is currently undergoing the toxicity and in vivo anticancer activity testing. Both N-pyrrolidinyl-substituted complexes (KP1495 and KP1511) display on contrary reasonable lipophilicity properties. This suggests that provided that their solubility is sufficient to exert antiproliferative effects [13], these Ga compounds can be recommended for moving into the next hit compound identification stage. As predicted, complexation with strong, chelating ligands renders gallium(III) high stability against hydrolysis. Furthermore, one can anticipate that other possible metabolic processes do not also affect much the integrity of oral gallium drugs on the way to cancer cells. In this context, the examined Ga complexes (see Table 2) do appear advantageous over one of the lead Ru-based anticancer agents, indazolium [trans-tetrachlorobis(1Hindazole)ruthenate(III)] (KP1019), exhibiting more than an order of magnitude stronger tendency to biotransformation via hydrolysis in physiological solution (s1/2 16 min [14]). The role and proper functions of protein-binding for gallium antineoplastic agents remain unclear. However, in view of the fact that interactions with serum proteins
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
determine differences in efficacy, activity and toxicity, as well as the overall distribution and excretion, of other metal complexes [9], it is reasonable to propose that the mode of action of gallium-based drugs would be highly dependent on their protein-binding affinity. Our findings pointed to substantial differences in reactivity toward albumin and transferrin exhibited by KP46 and KP1089 (see Table 3). The higher rate of protein adducts formation observed for KP1089 is probably associated with the cationic nature of the interacting gallium functionality [both proteins of interest bear a negative net charge at the extracellular (and CZE electrolyte) pH]. It can be predicted that the gallium-protein binding will proceed in a high-speed way for other cationic chelates as well. However, as justly pointed out an anonymous reviewer, other transporters, e.g., multidrug resistance (MDR1), would likely dominate pharmacokinetics and ultimately oral bioavailability of cationic gallium(III) complexes due to P-glycoprotein expression on surface of intestinal epithelial cells. The analysis of the literature data on apparent rate constants for other metal-based drugs [14,19] demonstrates that the gallium complexes have much in common regarding their reactivity toward serum proteins. Indeed, the kinetics of KP46 interaction with albumin is similar to that of another neutral drug, cisplatin (k = 3.3 · 105 and 5.5 · 105 s1, respectively). On the other hand, KP1089 rather resembles one more (but negatively) charged complex, KP1019: 1.1 · 104 and 3.3 · 104 s1, respectively. Such similarities also provide proofs of the possible contribution of non-covalent interactions in the binding process. Yet more remarkable looks the superior binding rate of KP46 and KP1089 to transferrin compared to albumin, thus reflecting certain likenesses in protein-binding behavior of gallium(III) and iron(III). This observation supports the potential function of gallium as an inhibitor of iron-dependent bioligands. It is the competitive interaction of gallium with ribonucleotide reductase, the iron-containing enzyme, which is thought to be the main target for Ga-based chemotherapeutics in proliferating tumor cells. In conclusion, having information on the characterization of novel metal coordination compounds in terms of relevant chemical and biochemical properties, preferably determined using rapid and inexpensive analytical methods, would lend to a better and more detailed understanding of chemotherapeutic significance of given drug candidates. In our opinion, such assaying should be implemented before (or conducted in parallel with) much more expensive and labor intensive in vitro and in vivo cytotoxicity studies, the positive results of which to say do not necessarily imply that a would be drug, being administered, finds its gateway into cells. 5. Abbreviations A0 CZE DMSO
corrected peak-area capillary zone electrophoresis dimethyl sulfoxide
k khyd KP46 KP1019 KP1089
KP1438
KP1492
KP1495
KP1497
KP1500
KP1511
log P s1/2
1825
protein-binding rate constant hydrolysis rate constant tris(8-quinolinolato)gallium(III) trans-tetrachlorobis(1Hindazole)ruthenate(III) bis(2-acetylpyridine-4,4-dimethyl-3thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) gallium(III) complex of N-(9-bromo-7, 12-dihydroindolo[3,2-d][1]benzazepin6(5H)-yliden)-N 0 -(2-hydroxybenzyliden)-azine 2-acetylpyridine-4,4-dimethyl-3thiosemicarbazonato-N,N,Sdichlorogallium(III) bis(2-acetylpyrazine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate 2-acetylpyridine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,Sdichlorogallium(III) bis(2-acetylpyrazine-N-piperidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate bis(2-acetylpyridine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate octanol–water partition coefficient half-life
Acknowledgements The authors acknowledge with gratitude Dr. Vasilii Karandashev, Institute of Microelectronics Technology and High-Purity Materials, for supervising analytical measurements, and Dr. Olga Semenova (University of Vienna) for performing kinetic calculations. Dr. Peter Unfried and Anatoly Dobrov are thanked for preparing certain gallium complexes. This paper is a contribution to the project P18123-N11 financed by the Austrian Science Foundation (FWF). Financial support from the Faustus Translational Drug Development is also gratefully acknowledged. References [1] M. Galanski, V.B. Arion, M.A. Jakupec, B.K. Keppler, Curr. Pharm. Des. 9 (2003) 2078–2089. [2] M.A. Jakupec, B.K. Keppler, in: A. Siegel, H. Siegel (Eds.), Metal Ions in Biological Systems, vol. 42, Dekker, New York, 2004, pp. 425–462. [3] M.A. Jakupec, B.K. Keppler, Curr. Top. Med. Chem. 4 (2004) 1575– 1583. [4] M. Thiel, T. Schilling, D.C. Gey, R. Ziegler, P. Collery, B.K. Keppler, in: H.H. Fiebig, A.M. Burger (Eds.), Relevance of Tumor Models for Anticancer Drug Development, in: W. Queisser, W. Scheithauer (Eds.), Contributions to Oncology, vol. 54, Karger, Basel, 1999, pp. 439–443. [5] L.R. Bernstein, T. Tanner, C. Godfrey, B. Noll, Metal-Based Drugs 7 (2000) 33–47. [6] V.B. Arion, M.A. Jakupec, M. Galanski, P. Unfried, B.K. Keppler, J. Inorg. Biochem. 91 (2002) 298–305.
1826
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
[7] K.H. Thompson, C. Orvig, Dalton Trans. (2006) 761–764. [8] A. Dobrov, V.B. Arion, N. Kandler, W. Ginzinger, M.A. Jakupec, A. Rufin´ska, N. Graf von Keyserlingk, M. Galanski, C. Kowol, B.K. Keppler, Inorg. Chem. 45 (2006) 1945–1950. [9] A.R. Timerbaev, C.G. Hartinger, S.S. Aleksenko, B.K. Keppler, Chem. Rev. 106 (2006) 2224–2248. [10] A.R. Timerbaev, A. Ku¨ng, B.K. Keppler, J. Chromatogr. A 945 (2002) 25–44. [11] C.G. Hartinger, A.R. Timerbaev, B.K. Keppler, Electrophoresis 24 (2003) 2023–2037. [12] P. Collery, J.L. Domingo, B.K. Keppler, Anticancer Res. 16 (1996) 687–691. [13] C. Kowol, R. Berger, R. Eichinger, A. Roller, M.A. Jakupec, V.B. Arion, B.K. Keppler, J. Med. Chem., submitted for publication.
[14] A.R. Timerbaev, A.V. Rudnev, O. Semenova, C.G. Hartinger, B.K. Keppler, Anal. Biochem. 341 (2005) 326–333. [15] P.F. Dimond, Gen. Eng. News 25 (2005) 21–26. [16] A. Zenker, M. Galanski, T.L. Bereuter, B.K. Keppler, W. Lindner, J. Chromatogr. B 745 (2000) 211–219. [17] Z. Huang, A.R. Timerbaev, B.K. Keppler, T. Hirokawa, J. Chromatogr. A 1106 (2006) 75–79. [18] A. Ku¨ng, T. Pieper, R. Wissiack, E. Rosenberg, B.K. Keppler, J. Biol. Inorg. Chem. 6 (2001) 292–299. [19] A.R. Timerbaev, S.S. Aleksenko, K. Polec-Pawlak, R. Ruzik, O. Semenova, C.G. Hartinger, S. Oszwaldowski, M. Galanski, M. Jarosz, B.K. Keppler, Electrophoresis 25 (2004) 1988–1995. [20] A. Warnecke, I. Fichtner, D. Garmann, U. Jaehde, F. Kratz, Bioconjug. Chem. 15 (2004) 1349–1359.
Inorganic Biochemistry Journal of Inorganic Biochemistry 100 (2006) 1819–1826 www.elsevier.com/locate/jinorgbio
Preclinical characterization of anticancer gallium(III) complexes: Solubility, stability, lipophilicity and binding to serum proteins Alexander V. Rudnev a, Lidia S. Foteeva a, Christian Kowol b, Roland Berger b, Michael A. Jakupec b, Vladimir B. Arion b, Andrei R. Timerbaev a,b,*, Bernhard K. Keppler a
b
Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Sciences, Kosygin St. 19, 119991 Moscow, Russia b Institute of Inorganic Chemistry, University of Vienna, Waehringer Str. 42, A-1090 Vienna, Austria Received 13 February 2006; received in revised form 11 July 2006; accepted 12 July 2006 Available online 25 July 2006
Abstract The discovery and development of gallium(III) complexes capable of inhibiting tumor growth is an emerging area of anticancer drug research. A range of novel gallium coordination compounds with established cytotoxic efficacy have been characterized in terms of desirable chemical and biochemical properties and compared with tris(8-quinolinolato)gallium(III) (KP46), a lead anticancer gallium-based candidate that successfully finished phase I clinical trials (under the name FFC11), showing activity against renal cell cancer. In view of probable oral administration, drug-like parameters, such as solubility in water, saline and 0.5% dimethyl sulfoxide, stability against hydrolysis, measured as the rate constant of hydrolytic degradation in water or physiological buffer using a capillary zone electrophoresis (CZE) assay, and the octanol–water partition coefficient (log P) providing a rational estimate of a drug’s lipophilicity, have been evaluated and compared. The differences in bioavailability characteristics between different complexes were discussed within the formalism of structure-activity relationships. The reactivity toward major serum transport proteins, albumin and transferrin, was also assayed in order to elucidate the drug’s distribution pathway after intestinal absorption. According to the values of apparent binding rate constants determined by CZE, both KP46 and bis(2-acetylpyridine-4,4-dimethyl-3-thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) (KP1089) bind to transferrin faster than to albumin. This implies that transferrin would rather mediate the accumulation of gallium antineoplastic agents in solid tumors. A tendency of being faster converted into the protein-bound form found for KP1089 (due possibly to non-covalent binding) seems complementary to its greater in vitro antiproliferative activity. 2006 Elsevier Inc. All rights reserved. Keywords: Anticancer drugs; Gallium(III) complexes; Drug-like properties; Serum protein-binding; Capillary electrophoresis
1. Introduction Complexation of gallium with organic ligands has been recognized as a promising strategy for creating tumorinhibiting therapeutics with a number of advantages over gallium salts regarding oral bioavailability [1–3]. Indeed, the evolved hydrolytic stability and membrane penetration ability render gallium complexes improved intestinal
*
Corresponding author. Tel.: +43 1 4277 52609; fax: +43 1 4277 9526. E-mail address: [email protected] (A.R. Timerbaev).
0162-0134/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2006.07.003
absorption functions compared to nitrate or chloride salt which leads to increased plasma concentrations of gallium. Owing to such benefits, as well as better antiproliferative properties, two oral compounds, tris(8-quinolinolato)gallium(III) (KP46) [4] and tris(3-hydroxy-2-methyl-4Hpyran-4-onato)gallium(III) [5] have been selected from a series of gallium complexes for clinical development. Likewise, bis(2-acetylpyridine-4,4-dimethyl-3-thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) (KP1089) was the first representative of the class of a-N-heterocyclic thiosemicarbazone complexes that has been assayed for antineoplastic activity in human tumor cell lines [6]. In
1820
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
vitro cytotoxic potency of KP1089 was found to be the highest among the gallium compounds tested. This finding provides evidence in favor of a modern metallodrug-developing concept [7] that implies combining a metal ion and a ligand which are both pharmaceutically active and aimed at attacking the same molecular target (most likely, ribonucleotide reductase in the case of gallium thiosemicarbazonates). Another complex with high gallium-induced anticancer activity in vitro, KP1438, is formed by the reaction of Ga(III) with a paulone derivative [8]. Still, gallium complexes are thought to be immature with regard to the number of compounds examined, especially if one takes into account the well-known activity of this metal against malignant tumors and a diversity of anticancer complexes of other metals such as platinum(II) or ruthenium(III). Furthermore, a consensus has not yet been reached on the structural effects of different organic ligands on the antineoplastic action of gallium. A successful lead-drug candidate selection program requires careful consideration of various chemical and biochemical assays in the identification of the most active, ‘hit’ compounds. The respective physicochemical characterization includes solubility, hydration, lipophilicity (log P), metabolic stability, metabolite profiles, protein-binding, and other testing. Conducted in parallel with in vitro antitumor inhibiting studies, such an early hit compound identification stage permits further selection of drug nominees with desirable absorption, distribution and metabolism properties. When properly designed it may bring about significant merits in shortening timelines and reducing costs between discovery and clinical development. The aim of the present study was the evaluation of the drug-like properties of a representative number of cytotoxic gallium complexes, with the objective of identification and optimization of drug candidates for further preclinical development. Both the investigational KP46 drug taken as a reference compound, several neutral and cationic thiosemicarbazonates with, respectively, 1:1 and 1:2 metal-toligand stoichiometry, including KP1089, and the gallium chelate with a N,N,O-binding site (KP1438) were selected for this systematic examination. Special emphasis was put on the comparative estimation of the reactivity toward serum proteins performing transport functions, as to the authors’ knowledge no relevant information exists in the literature [9]. For this purpose, as well as for the monitoring of hydrolytic conversion of some gallium complexes under scrutiny, capillary zone electrophoresis (CZE) was utilized as an up-and-coming tool for preclinical characterization of anticancer metallodrugs [10,11]. 2. Experimental section 2.1. Materials The structures of eight gallium complexes that were investigated in this work, along with their order-of-discovery names, are shown in Fig. 1. The compounds were syn-
thesized and characterized at the Institute of Inorganic Chemistry, University of Vienna as described elsewhere [6,8,12,13]. Human serum albumin (min. 96%; fraction V powder) and transferrin (min. 90%) were purchased from Sigma–Aldrich (Vienna, Austria) and Fluka (Buchs, Switzerland), respectively. Sodium dihydrogenphosphate, disodium hydrogenphosphate, sodium chloride, and sodium hydroxide, all of highest purity available, were obtained from Merck (Darmstadt, Germany). Dimethyl sulfoxide (DMSO) and n-octanol were the products of Fluka. All solutions for CZE were prepared using high-purity water obtained by distillation of deionized water. 2.2. Instrumentation For CZE experiments a CAPEL 105 instrument (Lumex, St. Petersburg, Russia) equipped with a variablewavelength UV detector was used. Data analysis was performed using a MultiChrom program (Ampersend, Moscow, Russia). A Polymicro Technologies (Phoenix, AZ, USA) fused-silica capillary’s dimension was 40 cm total length (30.5 cm to the detection window) and 75 lm I.D. For preparations of working solutions of proteins and sparingly soluble complexes, an ultrasonication bath (Sapphire, Moscow, Russia) was used. Mechanical shaking was done with a C2184 mechanical horizontal shaker (Russia). Incubation experiments were performed using a U-10 thermostat (Mechanik Pru¨fgera¨te, Medingen, Germany). 2.3. Procedures 2.3.1. Solubility Gallium complexes were shaken with a fixed volume of individual solvent [water, 0.9% (w/v) NaCl or 0.5% (v/v) DMSO] at room temperature for 2 h. The amounts of solid substances were taken so as to form the saturated solution. The supernatant was filtered through a 0.45-lm disposable membrane filter (Sartorius, Go¨ttingen, Germany) and the concentration of Ga in the resultant solution was determined (after an appropriate dilution) by inductively coupled plasma atomic emission spectroscopy (higher concentrations) or inductively coupled plasma mass spectrometry (lower contents) at the Institute of Microelectronics Technology and High-Purity Materials, Russian Academy of Sciences (Chernologolovka, Russia). The detection limits were 10–50 and 0.2–1 lg l1, respectively (depending on the dilution factor) and standard errors did not exceed 10%. 2.3.2. log P Weighted amounts of gallium complexes were partitioned between water and n-octanol for 2 h at room temperature by the shake flask method. Gallium concentrations in the aqueous phase before (C0) and after partitioning (Caq) were measured using the same analytical techniques, from which the partition coefficients were calculated as log P = log (C0 Caq)/Caq.
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
N
1821
N N
N
S
O
N
GaCl4
Ga
O
N
Ga N O
S N
N
N
KP46
N
KP1089 Br
H N N
NH N
O
Cl
Ga O N
Cl HN
Cl
N
Ga
N H
N
S N N
N
Br
KP1438
N
KP1492
N N
N S
N
N
S
N
Cl
PF6
Ga
Cl
Ga N
N N
S
N
N N
KP1495
N
KP1497
N
N
N
S
N
PF6
Ga N
N
N
S
N
N N
PF6
Ga
S
KP1500
N
N
N S
N
N N
N
KP1511 Fig. 1. Structures of gallium complexes.
N
1822
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
2.3.3. Capillary zone electrophoresis The capillary conditioning and cleaning protocols were as described in our previous work [14]. To improve the repeatability of peak-area signals used as response in kinetic experiments, the corrected peak-areas (A 0 ) were calculated by dividing the peak-area by the migration time. Sample introduction was performed by applying pressure injections at 10 or 20 mbar for 10 s. In all cases, the experiments were carried out at a constant temperature of 37 C using a liquid-thermostated capillary cassette. The applied voltage was 7 or 10 kV (for chloride-containing and -void electrolytes, respectively) at the injection end of the capillary. The wavelength of the UV detector was set to 200 nm. Electrolytes were passed through the membrane filter before use. 2.3.4. Hydrolysis rate The hydrolytic stability of gallium complexes in water and physiological buffer (10 mM NaH2PO4/Na2HPO4, 100 mM NaCl, pH 7.4) was evaluated by incubating a given solution at 37 C, taking aliquots continuously for CZE analysis, and monitoring a decrease of peak-area responses (A 0 ). Ten mM phosphate buffers (pH 7.4), containing no and 100 mM NaCl, were used as respective electrolytes. For KP46, the rate of aqueous hydrolysis was monitored after 1:1 dilution of its aqueous solution with 20 mM phosphate buffer (to give the phosphate concentration in the sample the same as in electrolyte). The kinetic series was repeated four times for each gallium complexhydrolysis medium combination to calculate the average rate constant (khyd) and the standard deviation. The halflife was calculated as s1/2 = ln 2/khyd. 2.3.5. Protein-binding kinetics All binding experiments were performed in 10 mM phosphate buffer, 100 mM NaCl, pH 7.4 as incubation solution, at 37 C and with 1 · 105 M KP46 or KP1089. The protein concentration in the reaction mixture was kept constant at 5 · 105 M. The kinetics of Ga-protein binding was assessed by recording the relative change in the protein signal due to adduct evolution (the detection method avail-
able cannot differentiate the adduct and protein peaks). When the interaction was fairly fast and hence the interval between two subsequent CZE measurements too short (to acquire a sufficient number of data points), the aliquots collected were placed in a refrigerator in order to cease the binding reaction. The apparent rate constant (k) was calculated by polynomial approximation of the A 0 -time plot (see Section 3.4), assuming that the binding reaction obeys the first-order kinetics and the following equation is valid: ln A 0 = kt. 3. Results 3.1. Solubility Yet before it is absorbed in the intestine and enters the bloodstream, an oral gallium complex is required to possess sufficient solubility in order to subsequently exhibit the cytotoxic activity. To underscore, almost half of new molecular entities synthesized by pharmaceutical companies have failed to be developed because of poor water solubility [15]. A summary of saturation solubility data is given in Table 1. It is important to note that the only value available in the literature for gallium complexes of interest, is water solubility of KP46, i.e., 22 mg l1 or 4.4 · 105 M [3]. A detailed analysis of solubility thresholds showed that all the complexes can be subdivided into two groups: those that are soluble in the micromolar concentration range and others, much less soluble compounds (6105 M). The first group includes gallium thiosemicarbazonates with 1:1 stoichiometry, KP1492 and KP1497, which are less hydrophobic molecules than the corresponding chelates with a metal-to-ligand ratio of 1:2, as well as KP1089 whose improved solubility is due to less hydrophobic counterion and 4N-substituents (two methyl groups). One should also be aware that regardless of the solvent nature, the 1:2 complex with GaCl 4 as counter-ion, namely KP1089, does rearrange in solution into the analogous 1:1 chelate, i.e., KP1492, and vice versa. For instance, according to NMR data [13] KP1089 exists in the aged solution as a ca. 1:2 equilibrium mixture of the original complex and
Table 1 Summary of drug-related physicochemical properties of gallium complexes Complexa
Solubility (mg l1 M1) Water
0.9% NaCl
0.5% DMSO
KP46 (502.18) KP1089 (723.87) KP1438 (993.80) KP1492 (361.93) KP1495 (711.35) KP1497 (387.97) KP1500 (739.40) KP1511 (709.37)
17.9/3.6 · 105 4.42b/6.1 · 103 5.0/5.0 · 106 2.84b/7.9 · 103 1.9/2.7 · 106 1.45b/3.7 · 103 2.4/3.3 · 106 12.6/1.8 · 105
1.0/2.0 · 106 2.47b/3.4 · 103 2.0/2.0 · 106 2.14b/5.9 · 103 3.7/5.2 · 106 0.52b/1.3 · 103 5.2/7.0 · 106 9.1/1.3 · 105
15.2/3.0 · 105 2.86b/3.9 · 103 0.2/1.6 · 107 – 2.0/2.9 · 106 1.40b/3.6 · 103 13.3/1.8 · 105 5.2/7.3 · 106
a b
Given in parentheses is the molecular mass in Da. In g l1.
log P
0.88 1.15 1.10 1.40 0.95 0.69 0.05 0.47
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
3.2. Octanol–water partition coefficients The efficient penetration through biomembranes is a prerequisite for a gallium complex (like every oral drug) on the way from intestine to bloodstream and then to the tumor cell. The ability of a potential drug candidate to cross these barriers until it binds to the target and induces the desired response is most often described by the octanol–water partition coefficient. Shown in Table 1 are log P values collected to estimate the comparative distribution of gallium complexes between water and n-octanol as the solubilizing medium mimicking the interior part of biological membranes. Although characteristic of only the intact complexes (not their metabolites or protein adducts), such lipophilicity thresholds allow the bioavailability estimation for the compounds of interest being applied as anticancer drugs. Indeed, KP1438, KP1495, and KP46 do prefer a non-aqueous over an aqueous environment, behaving as essentially non-polar compounds. KP1500, KP1487, KP1089 and KP1492 all manifest a tendency of being predominately partitioned into water and thereby could hardly be expected at high concentration across the cell membrane. The membrane permeation potency of KP1511 can be characterized as intermediate one. Notably, no definite relationship between the structure and lipophilicity properties was observed for a given selection of compounds. Also to be pointed out are quite different log P values of closely related KP1497 and KP1511 compounds.
3.3. Hydrolytic stability As soon as a gallium complex (generally speaking, any metal complex) encounters body-fluid circumstances, it would inevitably be the subject of hydrolysis. It is this unwanted process that posed clinical limitations on gallium salts and gave rise to the development of tumor-inhibiting gallium complexes. Therefore, assessing survival rates is a basic requirement in systematic metallodrug discovery process, and CZE offers a rapid, cost-efficient, highthroughput, and reliable screening method for such measurements. This was demonstrated in the monitoring of hydrolysis kinetics for several platinum(II) [16,17] and ruthenium(III) [14,18] coordination compounds. Typical electropherograms recorded for KP46 and KP1089 are shown in Fig. 2. As follows from the data of Table 2, both
0.08
Absorbance (mAU)
KP1492. Therefore, the respective solubility figures represent rather a mixture of two compounds, and the corresponding rearrangement is partly responsible for better solubility of KP1089 compared to other, more stable 1:2 thiosemicarbazonates. The cationic thiosemicarbazonates with the hexafluorophosphate counter-ion (KP1495, KP1500, and KP1511) display considerably lower solubility. This is also because their derivatives with piperidinyl and pyrrolidinyl groups at 4N-position appear to be less solubilized in essentially aqueous media. In this regard, the analysis of solid-state crystal structures of compounds is anticipated for the presence of intermolecular interaction. What is likewise noteworthy is that compounds KP1511 and KP1495, which have different substituents at N4 (in terms of thiosemicarbazide derivatives nomenclature), demonstrate such dissimilar solubilities (and also lipophilicities; see below). The lowest solubility of KP1438 among the complexes studied can be explained taking into account a more developed structure of respective ligands, containing several hydrophobic aromatic entities (see Fig. 1). Comparison of solubility figures obtained for different biologically relevant media revealed that the gallium complexes were generally more soluble in water than in the saline and 0.5% DMSO, whereas the relative solubility in the latter two solutions varied from one complex to another without distinctive relation to its structural type.
1823
0.05
0
1
2
3
Time (min) Fig. 2. Electropherograms of aqueous solutions of: (a) KP46 and (b) KP1089. Samples: (a) 1.8 · 105 M KP46 and (b) 1 · 104 M KP1089. Carrier electrolyte: 10 mM phosphate buffer, pH 7.4. Voltage: 10 kV (current, 20–21 lA). Sample introduction: (a) 200 mbar s and (b) 100 mbar s. Other CZE conditions, see Section 2. GaLþ 2 ; (2) GaCl4 [for L assignment, see Fig. 1].
Table 2 Hydrolysis rate constants and half lives (n = 4) Complex
KP46 KP1089
Watera
Physiological buffer
khyd ± SD (·105, s1)
s1/2(h)
khyd ± SD (·105, s1)
s1/2(h)
4.8 ± 0.9 4.3 ± 0.5
4.0 4.5
1.4 ± 0.5 3.7 ± 0.9
14.2 5.5
a pH of aqueous solutions of KP46 and KP1089 was 3.8 and 6.6, respectively.
1824
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
gallium complexes investigated are fairly stable, retaining their integrity at a level of 50% for at least several hours. In view of the afore-mentioned rearrangement of KP1089 that proceeds immediately after dissolution, the corresponding k values represent rather overall stability than stability against hydrolysis. At any rate, the decomposition of complexes measured under conditions simulating an extracellular solution environment (pH 7.4) is slower than in water (ca. 3 times for KP46). Also worthwhile of special noting is at average lower hydrolysis rate of KP46. (Our attempts to assess another complex, KP1438, were unsuccessful because of its extremely low solubility.) 3.4. Binding kinetics to albumin and transferrin In the bloodstream, gallium complexes may form (stable) adducts with plasma proteins, and such interactions would also affect the bioavailability and the pharmacokinetic profiles of these compounds. Good tolerance to proteinaceous samples and short analyses time make the CZE technique very convenient for kinetic characterization of platinum group metallodrugs participating in protein-binding processes [14,19,20]. As can be seen in Table 3, there are notable differences in the binding rates of KP46 and KP1089 expressed as the acceptably accurate values of k, which were by a factor of 3–4 higher for the latter complex. Furthermore, there was also some dissimilarity in the reaction kinetics toward different proteins. The fact that transferrin binding is kinetically more favorable, also appearing Table 3 Reactivity of gallium complexes toward different proteins (n = 3–5) Complex
KP46 KP1089
Albumin
Transferrin
k ± SD (·105, s1)
k ± SD (·105, s1)
3.3 ± 0.1 11.2 ± 2.1
5.0 ± 0.8 21.1 ± 3.4
Corrected peak area
4
1 3
2
2
1
0
10
100
150
200
250
300
Time (min) Fig. 3. Binding curves exemplifying the rate of KP46 interaction with: (1) albumin and (2) transferrin. Incubation and CZE conditions, see Section 2.
from a steeper course of the corresponding binding curve shown for KP46 in Fig. 3, does seem apparent accounting for the gallium cation resemblance to iron(III). 4. Discussion The core properties required to estimate absorption, distribution, and transport of metal complexes in the body are solubility, lipophilicity, stability, and affinity toward transport proteins. Determinations of these properties are of crucial importance in anticancer metallodrug research as they help to select a lead candidate for further preclinical investigation and to design more active and/or less toxic compounds. Most of the gallium complexes which are considered, in a comparative manner, in this study are deficient in such screening assays, and this seems to be a bottleneck in their discovery process. An appropriate hydrophilic/lipophilic balance is the drug’s attribute that affords its sufficient plasma concentration after oral administration. From the analysis of solubility and log P data (see Table 1), one can conclude that KP46 possesses the most striking features with respect to its bioconcentration and membrane permeability. On the other hand, KP1438 would unlikely be sufficiently bioavailable to express the in vivo cytotoxicity (its in vitro tumorinhibiting activity is prominent at a 1 · 104 M level [8]) because of formulation difficulties related to solubilization. In addition, the complex’s lipophilicity might be too high to permit its passage through the cell membrane. Among gallium thiosemicarbazonates, KP1089, KP1492, KP1497, and KP1500 are marginally promising as valid oral drug nominees, oppositely, because of poor lipophilicity. As such they will not be very bioavailable after oral administration, and hence are to be applied via intravenous route. Such a conclusion seems especially important for KP1089 which is currently undergoing the toxicity and in vivo anticancer activity testing. Both N-pyrrolidinyl-substituted complexes (KP1495 and KP1511) display on contrary reasonable lipophilicity properties. This suggests that provided that their solubility is sufficient to exert antiproliferative effects [13], these Ga compounds can be recommended for moving into the next hit compound identification stage. As predicted, complexation with strong, chelating ligands renders gallium(III) high stability against hydrolysis. Furthermore, one can anticipate that other possible metabolic processes do not also affect much the integrity of oral gallium drugs on the way to cancer cells. In this context, the examined Ga complexes (see Table 2) do appear advantageous over one of the lead Ru-based anticancer agents, indazolium [trans-tetrachlorobis(1Hindazole)ruthenate(III)] (KP1019), exhibiting more than an order of magnitude stronger tendency to biotransformation via hydrolysis in physiological solution (s1/2 16 min [14]). The role and proper functions of protein-binding for gallium antineoplastic agents remain unclear. However, in view of the fact that interactions with serum proteins
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
determine differences in efficacy, activity and toxicity, as well as the overall distribution and excretion, of other metal complexes [9], it is reasonable to propose that the mode of action of gallium-based drugs would be highly dependent on their protein-binding affinity. Our findings pointed to substantial differences in reactivity toward albumin and transferrin exhibited by KP46 and KP1089 (see Table 3). The higher rate of protein adducts formation observed for KP1089 is probably associated with the cationic nature of the interacting gallium functionality [both proteins of interest bear a negative net charge at the extracellular (and CZE electrolyte) pH]. It can be predicted that the gallium-protein binding will proceed in a high-speed way for other cationic chelates as well. However, as justly pointed out an anonymous reviewer, other transporters, e.g., multidrug resistance (MDR1), would likely dominate pharmacokinetics and ultimately oral bioavailability of cationic gallium(III) complexes due to P-glycoprotein expression on surface of intestinal epithelial cells. The analysis of the literature data on apparent rate constants for other metal-based drugs [14,19] demonstrates that the gallium complexes have much in common regarding their reactivity toward serum proteins. Indeed, the kinetics of KP46 interaction with albumin is similar to that of another neutral drug, cisplatin (k = 3.3 · 105 and 5.5 · 105 s1, respectively). On the other hand, KP1089 rather resembles one more (but negatively) charged complex, KP1019: 1.1 · 104 and 3.3 · 104 s1, respectively. Such similarities also provide proofs of the possible contribution of non-covalent interactions in the binding process. Yet more remarkable looks the superior binding rate of KP46 and KP1089 to transferrin compared to albumin, thus reflecting certain likenesses in protein-binding behavior of gallium(III) and iron(III). This observation supports the potential function of gallium as an inhibitor of iron-dependent bioligands. It is the competitive interaction of gallium with ribonucleotide reductase, the iron-containing enzyme, which is thought to be the main target for Ga-based chemotherapeutics in proliferating tumor cells. In conclusion, having information on the characterization of novel metal coordination compounds in terms of relevant chemical and biochemical properties, preferably determined using rapid and inexpensive analytical methods, would lend to a better and more detailed understanding of chemotherapeutic significance of given drug candidates. In our opinion, such assaying should be implemented before (or conducted in parallel with) much more expensive and labor intensive in vitro and in vivo cytotoxicity studies, the positive results of which to say do not necessarily imply that a would be drug, being administered, finds its gateway into cells. 5. Abbreviations A0 CZE DMSO
corrected peak-area capillary zone electrophoresis dimethyl sulfoxide
k khyd KP46 KP1019 KP1089
KP1438
KP1492
KP1495
KP1497
KP1500
KP1511
log P s1/2
1825
protein-binding rate constant hydrolysis rate constant tris(8-quinolinolato)gallium(III) trans-tetrachlorobis(1Hindazole)ruthenate(III) bis(2-acetylpyridine-4,4-dimethyl-3thiosemicarbazonato-N,N,S)gallium(III) tetrachlorogallate(III) gallium(III) complex of N-(9-bromo-7, 12-dihydroindolo[3,2-d][1]benzazepin6(5H)-yliden)-N 0 -(2-hydroxybenzyliden)-azine 2-acetylpyridine-4,4-dimethyl-3thiosemicarbazonato-N,N,Sdichlorogallium(III) bis(2-acetylpyrazine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate 2-acetylpyridine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,Sdichlorogallium(III) bis(2-acetylpyrazine-N-piperidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate bis(2-acetylpyridine-N-pyrrolidinyl-3thiosemicarbazonato-N,N,S)gallium(III) hexafluorophosphate octanol–water partition coefficient half-life
Acknowledgements The authors acknowledge with gratitude Dr. Vasilii Karandashev, Institute of Microelectronics Technology and High-Purity Materials, for supervising analytical measurements, and Dr. Olga Semenova (University of Vienna) for performing kinetic calculations. Dr. Peter Unfried and Anatoly Dobrov are thanked for preparing certain gallium complexes. This paper is a contribution to the project P18123-N11 financed by the Austrian Science Foundation (FWF). Financial support from the Faustus Translational Drug Development is also gratefully acknowledged. References [1] M. Galanski, V.B. Arion, M.A. Jakupec, B.K. Keppler, Curr. Pharm. Des. 9 (2003) 2078–2089. [2] M.A. Jakupec, B.K. Keppler, in: A. Siegel, H. Siegel (Eds.), Metal Ions in Biological Systems, vol. 42, Dekker, New York, 2004, pp. 425–462. [3] M.A. Jakupec, B.K. Keppler, Curr. Top. Med. Chem. 4 (2004) 1575– 1583. [4] M. Thiel, T. Schilling, D.C. Gey, R. Ziegler, P. Collery, B.K. Keppler, in: H.H. Fiebig, A.M. Burger (Eds.), Relevance of Tumor Models for Anticancer Drug Development, in: W. Queisser, W. Scheithauer (Eds.), Contributions to Oncology, vol. 54, Karger, Basel, 1999, pp. 439–443. [5] L.R. Bernstein, T. Tanner, C. Godfrey, B. Noll, Metal-Based Drugs 7 (2000) 33–47. [6] V.B. Arion, M.A. Jakupec, M. Galanski, P. Unfried, B.K. Keppler, J. Inorg. Biochem. 91 (2002) 298–305.
1826
A.V. Rudnev et al. / Journal of Inorganic Biochemistry 100 (2006) 1819–1826
[7] K.H. Thompson, C. Orvig, Dalton Trans. (2006) 761–764. [8] A. Dobrov, V.B. Arion, N. Kandler, W. Ginzinger, M.A. Jakupec, A. Rufin´ska, N. Graf von Keyserlingk, M. Galanski, C. Kowol, B.K. Keppler, Inorg. Chem. 45 (2006) 1945–1950. [9] A.R. Timerbaev, C.G. Hartinger, S.S. Aleksenko, B.K. Keppler, Chem. Rev. 106 (2006) 2224–2248. [10] A.R. Timerbaev, A. Ku¨ng, B.K. Keppler, J. Chromatogr. A 945 (2002) 25–44. [11] C.G. Hartinger, A.R. Timerbaev, B.K. Keppler, Electrophoresis 24 (2003) 2023–2037. [12] P. Collery, J.L. Domingo, B.K. Keppler, Anticancer Res. 16 (1996) 687–691. [13] C. Kowol, R. Berger, R. Eichinger, A. Roller, M.A. Jakupec, V.B. Arion, B.K. Keppler, J. Med. Chem., submitted for publication.
[14] A.R. Timerbaev, A.V. Rudnev, O. Semenova, C.G. Hartinger, B.K. Keppler, Anal. Biochem. 341 (2005) 326–333. [15] P.F. Dimond, Gen. Eng. News 25 (2005) 21–26. [16] A. Zenker, M. Galanski, T.L. Bereuter, B.K. Keppler, W. Lindner, J. Chromatogr. B 745 (2000) 211–219. [17] Z. Huang, A.R. Timerbaev, B.K. Keppler, T. Hirokawa, J. Chromatogr. A 1106 (2006) 75–79. [18] A. Ku¨ng, T. Pieper, R. Wissiack, E. Rosenberg, B.K. Keppler, J. Biol. Inorg. Chem. 6 (2001) 292–299. [19] A.R. Timerbaev, S.S. Aleksenko, K. Polec-Pawlak, R. Ruzik, O. Semenova, C.G. Hartinger, S. Oszwaldowski, M. Galanski, M. Jarosz, B.K. Keppler, Electrophoresis 25 (2004) 1988–1995. [20] A. Warnecke, I. Fichtner, D. Garmann, U. Jaehde, F. Kratz, Bioconjug. Chem. 15 (2004) 1349–1359.