The potential for cobalt(III) complexes in medicine, as chaperones of bioactive ligands, and to target tumours through bioreductive activation, has been examined over the past 20 years. Despite this, chemical properties such as reduction potential and carrier ligands required for optimal tumour targeting and drug delivery have not been optimised. Here we review the chemistry of cobalt(III) drug design, and recent developments in the understanding of the cellular fate of these drugs.

We report a potential means of selectively delivering matrix metalloproteinase (MMP) inhibitors to target tumour sites by use of a bioreductively activated Co(III) carrier system. The carrier, comprising a Co(III) complex of the tripodal ligand tris(methylpyridyl)amine (tpa), was investigated with the antimetastatic MMP inhibitor marimastat (mmstH(2)). The X-ray crystal structure of [Co(mmst)(tpa)]ClO(4) x 4H(2)O was determined and two-dimensional NMR revealed the existence of two isomeric forms of the complex in solution. Electrochemical analysis showed that the reduction potential of the complex is suitable for it to be bioreductively activated at hypoxic tumour sites. In vitro assays confirmed the stability of the prodrug in solution prior to reduction and revealed very low cytotoxicity against A2780 cells. In vivo testing in mice showed a higher level of tumour-growth inhibition by the complex than by free marimastat. Both free marimastat and and its Co(III) complex increased metastasis in the model used, with the complex significantly more active.

Co(III) complexes of simple hydroxamic acids have been evaluated as models of hypoxia activated prodrugs containing MMP inhibitors. The complexes are based upon a proposed carrier system comprising the tripodal tetradentate ligand tris(2-methylpyridyl)amine (tpa) with the hydroxamate functionality occupying the remaining coordination sites of the Co centre. Acetohydroxamato (aha), propionhydroxamato (pha), and benzohydroxamato (bha) complexes were synthesised and characterised by single crystal X-ray diffraction. For aha and pha both the hydroxamato and hydroximato (deprotonated) forms were obtained and were readily interconverted by pH manipulation; for bha only the hydroximato complex was obtained as a stable species. Electrochemical analysis was used to probe the redox chemistry of the complexes and assess their ease of reduction. All of the complexes displayed irreversible reduction and had low cathodic peak potentials. This suggests that the Co-tpa carrier system would provide a suitably inert framework to deliver the drugs to target sites intact yet would release the ligands upon reduction to the more labile Co(II) oxidation state.

The development of complexes that allow the monitoring of the release and distribution of fluorescent models of anticancer drugs initially bound to cobalt(III) moieties is reported. Strong quenching of fluorescence upon ligation to cobalt(III) was observed for both the carboxylate- and the hydroximate-bound fluorophores as was the partial return of fluorescence following addition of ascorbate and cysteine. The extent of the increase in the fluorescence intensity observed following addition of these potential reductants is indicative of the fluorophore being displaced from the complex by the action of ascorbate or cysteine, by ligand exchange. The cellular distribution of the fluorescence revealed that coordination to cobalt can dramatically alter the subcellular distribution of a bound fluorophore. This work shows that fluorescence can be an effective means of monitoring these agents in cells, and of determining their sites of activation. The results also reveal that the cytotoxicity of such agents correlates with their uptake and distribution patterns and that these are influenced by the types of ligands attached to the complex.

BACKGROUND: Predictive preclinical models play an important role in the assessment of new treatment strategies and as avatar models for personalised medicine; however, reliable and timely model generation is challenging. We investigated the feasibility of establishing patient-derived xenograft (PDX) models of high-risk neuroblastoma from a range of tumour-bearing patient materials and assessed approaches to improve engraftment efficiency. METHODS: PDX model development was attempted in NSG mice by using tumour materials from 12 patients, including primary and metastatic solid tumour samples, bone marrow, pleural fluid and residual cells from cytogenetic analysis. Subcutaneous, intramuscular and orthotopic engraftment were directly compared for three patients. RESULTS: PDX models were established for 44% (4/9) of patients at diagnosis and 100% (5/5) at relapse. In one case, attempted engraftment from pleural fluid resulted in an EBV-associated atypical lymphoid proliferation. Xenogeneic graft versus host disease was observed with attempted engraftment from lymph node and bone marrow tumour samples but could be prevented by T-cell depletion. Orthotopic engraftment was more efficient than subcutaneous or intramuscular engraftment. CONCLUSIONS: High-risk neuroblastoma PDX models can be reliably established from diverse sample types. Orthotopic implantation allows more rapid model development, increasing the likelihood of developing an avatar model within a clinically useful timeframe.