Preparation and Anti-Tumour Activity of Some Arylbismuth(III) Oxine Complexes

New arylbismuth(lll) oxinates, PhBi(MeOx)2, (p-MeC6H4)Bi(Ox)2, (p-MeC6H4)Bi(MeOx)2, (p-ClC6H4)Bi(Ox)2, and (p-ClC6H4)Bi(MeOx)2 (Ox− = quinolin-8-olate and MeOx−=2-methylquinolin-8-olate) have been prepared by reaction of the appropriate diarylbismuth chlorides with Na(Ox) or Na(MeOx) in the presence of 15-crown-5. An X-ray crystallographic study has shown PhBi(MeOx)2 to be a five coordinate monomer with distorted square pyramidal stereochemistry. Chelating MeOx ligands have a cisoid arrangement in the square plane and the phenyl group is apical. The lattice is stabilised by significant π-π interactions between centrosymmetric molecules. A range of these complexes has been shown to have high in vitro biological activity (comparable with or better than cisplatin) against L1210 leukaemia, the corresponding cisplatin resistant line, and a human ovarian cell line, SKOV-3. However, initial in vivo testing against a solid mouse plasmacytoma (PC6) and P388 leukaemia has not revealed significant activity.


Introduction
It has been shown recently that metathetical reactions of Ph2BiX and PhBiX2 (X CI, Br, or I) with sodium quinolin-8-olate (NaOx) are complicated by rearrangement reactions and by incomplete elimination of sodium halides [1]. Typically bimetallics of composition PhBi(Ox)2(NaX)n were isolated together with varying amounts of Bi(Ox)3. However, halide-free PhBi(Ox)2 was isolated from the reaction of Ph2BiCI with Na(Ox) in the presence of 15-crown-5 and from reaction of PhBil2 with Bu4N(Ox). The low solubility of PhBi(Ox)2 suggested an associated structure but no clear structural conclusions could be drawn from spectroscopic data.
We now report the synthesis of a range of pure organobismuth quinolin-8-olate and 2-methylquinolin-8-olate complexes, RBi(Ox)2 and RBi(MeOx)2, the X-ray crystal structure of PhBi(MeOx)2 and a preliminary survey of the anti-cancer activity of these quinolin-8-olate complexes together with some related reference compounds.
There has been sporadic interest in antitumour properties of bismuth compounds [2][3][4][5][6][7] the most successful of which have been methylbismuth(lll) thiolates. In addition, it has been shown that 8-quinolinol has antitumour activity [8,9]. Bismuth compounds are much more widely used in the treatment of gastrointestinal disorders [7] and of peptic ulcers [10]. The low toxicity of most bismuth compounds, especially relative to that of many other heavy metal compounds, makes them attractive candidates for drug usage [7]. It has been observed, however, that regular ingestion of high doses of bismuth salts (5 15 g/day by many French citizens) led to ca 300 cases of bismuth(Ill) intoxication with some deaths in 1976 [11]. Experimental Instruments and Procedures Details have been described previously [1]. Reagents Sodium salts of quinolin-8-ol and 2-methylquinolin-8-ol were prepared as described previously [1]. The following diarylbismuth chlorides were prepared by reaction of stoichiometric amounts of BiCl3 and Ar3Bi [12]: Ph2BiCl, m.p. 185 (4-CH3C6H4)2BiCI, m.p. 182C (dec.)lit. [14] 181C; (4-CIC6H4)2BiCI, 1H nmr (300 MHz) 6 (4-CH3C6H4)2BiCI + Na(Ox) A solution of Na(Ox) (0.89 g, 5.3 mmol) with 15-crown-5 ether (2.34 g, 2.1 ml, 10.6 mmol) in ethanol (25 ml) was added to a suspension of (4-MeC6H4)2BiCI (2.27 g, 5.3 mmol) in ethanol (20 ml)o The reaction mixture became a clear yellow colour before a yellow precipitate appeared. After stirring for 2 h, the reaction mixture was filtered and the resulting yellow solid was washed with ethanol (3 x 5 ml) and light petroleum (3 x 10 ml)o This solid was further stirred for h in ether (50 ml) to fully remove the (4-MeC6H4)3Bi which was also formed, then refiltered and rewashed with ether (50 ml), ethanol (2 x 20 ml), light petroleum (2 x 20 ml) and dried to yield (4-methylphenyl)bis(quinolin-8-olato)bismuth(lll) (1.31 g, 78%). The 1H nmr spectrum of the product taken a week later showed that the intensity of the resonances assigned to ethanol had decreased to give a mole ratio of 10" (Found: C, 49.9; H, 3 (solid state) 400 nm. The solvent was removed from the filtrate to yield a yellow solid. This was extracted with water (6 x 10 ml) and ether (6 x 10 ml)o NaCI was isolated from the aqueous Yields are calculated on the basis of reaction Prime numbers refer to atoms in the aryl group attached directly to bismuth extract (0.28 g, 90% by AgNO3 titration). The ether washings were combined and evaporated to yield (4-CH3C6H4)3Bi (1.22 g, 95%).

In Vitro Growth Inhibition and In Vivo Antitumour Testing
The methods for testing the compounds for growth inhibition against L1210 or the cisplatin resistant derivative L1210/DDP mouse leukaemia cells in vitro and P388 mouse leukaemia in vivo have been described previously [15]. Briefly, for the cell culture testing, cells were exposed to the drug at several concentrations for 48 h, after which they were counted using a Coulter counter (Model ZM) and compared to control cells grown in the presence of only the vehicle. The IC50, or concentration causing 50% inhibition of cell growth, was determined from the curve of percentage growth versus drug concentration. The SKOV-3 human ovarian carcinoma cell line was maintained in o-minimum essential medium plus 15% foetal calf serum. For these growth inhibition studies, 5x103 exponentially growing cells in 100 #1 medium were allowed to adhere in 96-well culture plates for 12 to 16 h at 37C in a humidified incubator gassed with 10% CO2/90% air. Drugs were dissolved in Me2SO and diluted in medium to 10 concentrations over a 4-log range, and 100 #1 of each drug solution was added to 5 wells. Cells were incubated for a further 72 h, after which viable cells were measured using the sulforhodamine B (SRB) assay [16] that measures cellular protein content. Briefly, cells were fixed with trichloroacetic acid and stained with SRB. Unbound dye was removed by washing with acetic acid, protein-bound dye was solubilised with Tris base, and the optical density was read at 550 nm using an automatic plate reader. The percentage cell growth inhibition was calculated as described above. All in vitro tests were done in duplicate, with repeats if greater than a 20% difference, and the results are reported as the mean.
All animal protocols were approved by the Institutional Animal Experimentation and Ethics Committee. For the animal studies with mouse leukaemia, DBA/2 mice received 106 P388 cells intraperitoneally (i.p.), and drugs were injected i.p. on days 1,5, and 9. Although increased lifespan measured in days was the endpoint, animals were sacrificed prior to death, when movement was restricted by ascites. Controls received vehicle only, and %T/C was calculated as the ratio of survival time of treated animals over control animals, where compounds with %T/C greater that 125% were considered to have some activity.
For the in vivo antitumour activity against PC-6 plasmacytoma, female Balb/c mice (10-15 weeks old) were maintained in controlled atmospheric conditions and fed standard mouse chow and water ad lib.. The compounds were suspended by sonication in peanut oil. The murine PC6 plasmacytoma (obtained from L. Kelland, Institute of Cancer Research, Sutton, U K) was inoculated as mm cubes subcutaneously on the flanks of the mice, and approximately 20 days later, mice with tumours were randomised into groups of 5 to 10 animals, which received either nothing (no-drug control), or an intraperitoneal injection of peanut oil at 10 ml/kg (vehicle control), cisplatin in saline at 6 mg/kg (positive control), or the test compounds at the maximum tolerated dose of 30 mg/kg. Eight to ten days later, mice were sacrificed and the tumours were dissected and weighed. The results were expressed as %T/C mean tumour weight of the treated animals over mean tumour weight of the vehicle control group, where values less than 75% are considered to be worth further investigation. Crystallography Intensity data for a yellow crystal (0.21 x 0.21 x 0.21 mm) were measured at room temperature on a Rigaku AFC6R diffractometer fitted with MoKo radiation (graphite monochromator, X 0.71073 A) using the e):2e scan technique so that 0max was 28.0. NO decomposition of the crystal occurred during the data collection and the data set was corrected for Lorentz and polarization effects [17], and for absorption employing an empirical procedure [18]. A total of 5905 data (5601 unique) were collected and of these, 3349 that satisfied the/_> 3.0(/) criterion were used in the subsequent analysis.  Table and (2) a where Beq 8=2/3(U1 l(aa*) 2 + U22(bb*) 2 + U33(cc*) 2 + 2U12aa*bb*cosy + 2U13aa*cc*cos13 + 2 U23bb*cc*coso)

Results and Discussion Preparation and characterisation of arylbismuth oxinates
A range of new arylbismuth(lll) oxinates, ArBi(Ox)2 (Ar p-MeC6H4 or p-CI6H4) and ArBi(MeOx)2 (Ar Ph, p-Me6H4, or p-CIC6H4) have been prepared by reaction of diarylbismuth(lll) chlorides with Na(Ox) or Na(MeOx)in ethanol in the presence of 15-crown-5. 2Ar2BiCI + 2Na(Ox or MeOx) --, ArBi(Ox or MeOx)2 + Ar3Bi + 2NaCI The crown ether prevented retention of sodium chloride-by the arylbismuth oxinate [1]. In representative cases, the amount of sodium chloride produced was shown to be ca. 90% as was the yield of Ar3Bi. Reaction (1) rather than reaction of ArBiCI2 with Na(Ox or MeOx) was chosen to avoid formation of Bi(Ox)3 or Bi(MeOx)3 [1], which cannot be separated from ArBi(Ox or MeOx)2 owing to their mutual low solubilities. The Ar3Bi compounds can be recycled by reaction with BiCl3 to give the reactants, Ar2BiCI. Most complexes were analytically pure except RBi(Ox)2 (R p-MeC6H4 or p-CIC6H4) where use of the crown ether makes it unlikely that the low carbon analyses were due to sodium halide retention. Satisfactory 1H and 13C NMR spectra and UV/Vis spectra (showing chelation of Ox and MeOx) [21,22] were obtained for all complexes including RBi(Ox)2 (R p-MeC6H4 or p-CIC6H4). A COSY 1H NMR spectrum of PhBi(MeOx)2 showed that H2,6 of the phenyl group was at lower frequency than H4 of MeOx by contrast with the reported spectrum [1] of PhBi(Ox)2.
Molecular structure of PhBi(MeOx)2 The molecular structure of PhBi(MeOx)2 is illustrated in Figure and selected interatomic parameters are listed in Table 2. This is the first structurally characterised bismuth(Ill) oxine complex. Only the bismuth(V) derivatives, Ph3Bi(Ox or MeOx)CI have been analysed crystallographically previously [21,23]. The bismuth atom is five coordinate, with distorted square pyramidal stereochemistry. The square plane is defined by a N202 donor set derived from two chelating MeOx anions, and the bismuth atom lies 0.0565(2) A above the least-squares plane through the N202 atoms (mean deviation" 0.014 A). The square plane is significantly distorted towards a trapezoidal geometry owing to the presence of disparate BiO and BiN distances. This disparity is manifested in the wide N--BimN angle of 145.0(2) and may arise in order to minimise intramolecular repulsions between the methyl groups. The phenyl group is almost symmetrically inclined with respect to the BiN202 plane, forming a dihedral angle of 95.6 with it.  a %T/C mean tumour weight of drug treated animal/mean tumour weight of controls, where values less than 75% are considered to indicate activity Disappointingly, the compounds are inactive. This may be attributable to delivery problems as the complexes were completely insoluble in the delivery medium. The bimetallic [KPhBi(Ox)3] was also tested against P388 leukaemia (Table 5), with the compound delivered intraperitoneally as a solution in Me2SO. Three separate concentrations were examined including the maximum tolerated dose but no activity was detected. These results contrast the reported in vivo activity of 6mercaptopurinebismuth(lll) complexes [4], the methylbismuth(lll) thiolates [7], and, more closely relevant, Ph2BiO2CR (R Et, iPr or Ph) [6], for which activity against P388 leukaemia (T/C 147-168% for doses of 50-200 mg/kg) was recorded [6]. Of relevance to the latter observation is that diphenylbismuth(lll) carboxylates have been shown to be highly sensitive to hydrolytic cleavage of aryl groups [30]. For PhBi(Ox or MeOx)2, the contrast between the excellent activity in vitro and the lack of activity in vivo is dramatic, and points to delivery problems in animal models. This has not been overcome by dissolution of the low solubility drugs in Me2SO, where precipitation on injection into aqueous biological fluids may deactivate the compounds.  a %T/C was calculated as the ratio of survival time of treated animals over control animals, where compounds with %T/C greater that 125% are considered to have some activity b positive control