A Library of Cobalt (III) Schiff Base Complexes for pH-Triggered Protein Inhibition
Cobalt(III) Schiff base (Co(III)-sb) complexes inhibit histidine-containing proteins by interfering with structural or active sites. Through inhibition of proteins relevant to diseases processes such as cancer metastasis, Co(III)-sb complexes can be developed for therapeutic approaches. The Meade lab is developing pH-responsive, self-reporting Co(III)-sb complexes. Here, a library of Co(III)-sb complexes with modifications to the equatorial ligand was developed in an effort to enhance stability while retaining pH lability. Electron-withdrawing (chloride and phenyl) substituents were added to the parent Co(III)-sb complexes, termed Co-Acacen, to form Co-ClAcacen, Co-Benacen, and Co-ClBenacen. These complexes were compared to the parent compound through fluorescence studies. Although differences were seen between the compounds, solubility issues prevented a deeper analysis of the results. The complexes with phenyl substituents had a decreased pH-dependent lability while the complexes with chloride substituents exhibited greater stability under neutral conditions.
Following the discovery of cisplatin as an anti-cancer drug in the late 20th century, metal complexes have been gaining popularity as therapeutics. The Meade lab has extensively investigated cobalt(III) Schiff base complexes (Co(III)-sb) as inhibitors of proteins involved in disease progression (1-6). The complex contains two labile axial ligands and a tetradentate ligand, acetylacetonatoethylenediimine (acacen), in the equatorial plain (Figure 1). Co-sb complexes bind to histidine (His) residues through a dissociative exchange mechanism of the labile axial ligands (Figure 1) (1-5, 7). His-coordination disrupts protein structure and inhibits function (Figure 2).
Recent research has shown that Co(III)-sb complexes can inhibit two proteins involved in cancer metastasis: 1) Snail zinc finger transcription factors and 2) matrix metalloproteinases. Cancer metastasis is responsible for >90% of cancer-related deaths, making these proteins highly attractive drug targets (2, 5-6, 8). Snail proteins regulate a pathway known as the epithelial-to-mesenchymal transition (EMT) whereby cancer cells lose their adhesion properties and adopt migratory and invasive behavior (9). Matrix metalloproteinases (MMPs) are endopeptidases that cleave extracellular matrix proteins, facilitating cancer cell invasion into the bloodstream (10). Both of these proteins contain zinc-dependent sites that utilize histidine residues to coordinate to the zinc(II) ion. In the case of Snail proteins, the zinc(II) stabilizes the structure required for gene regulation, and Co(III)-sb coordination displaces the Zn(II) and disrupts this structure (Figure 2) (4). In the case of MMPs, zinc(II) facilitates their catalytic function, and Co(III)-sb coordination to the active site blocks enzyme activity (11). Targeted inhibition can be achieved by incorporating protein-specific molecules to the acacen equatorial ligand (6).
The goal of this research is to increase control of protein inhibition by developing a class of self-reporting, pH-responsive Co-sb complexes using a prodrug strategy. Prodrugs are inactive forms of a drug that activate in the presence of an internal trigger (such as a change in light, temperature, or chemical composition) (12-13). Prodrugs increase the efficacy and minimize undesirable side effects of parent drugs by reducing the necessary dosage and ensuring that the drug activates in the locale of its target (12-13). The use of pH as a prodrug trigger is applicable in two scenarios: 1) the low pH environment of tumor microenvironments for selective activation in cancer tissues and 2) the low pH of endosomes for intracellular activation to reduce off-target binding with extracellular and plasma proteins, a common challenge with metal-based therapies (14).
To achieve pH-dependent protein inhibition, axial ligands that are inert under neutral pH but dissociate under decreased pH are incorporated into the Co(III)-sb complex. Dissociation exposes coordination sites for His binding within the target proteins. The first generation activatable cobalt complex, termed Co-Acacen (Figure 5), contained axial ligands that included an imidazole moiety and a coumarin fluorophore (C3Im) (Figure 3). pH-dependent dissociation was conferred by the imidadazole moiety of C3Im. The imidazoles exhibited tight coordination to the cobalt(III) center under neutral aqueous conditions, rendering the complex inactive. In low pH environment, protonation of the imidazole nitrogens reduced binding affinity of the axial ligands, promoting ligand exchange with solvent (water) ligands. The coumarin fluorophore served to report on ligand dissociation. When coordinated to the Co(III) center, C3Im showed reduced fluorescence due to quenching by the cobalt(III) center. Fluorescence was restored upon ligand dissociation. Thus, fluorescence intensity could be used to monitor ligand dissociation and activation of the complex (Figure 4).
While ligand dissociation of Co-Acacen exhibited the desired pH-dependent dissociation, in the presence of a competing imidazole molecule (resembling a His residue) under neutral pH, substitution of the axial ligands in Co-Acacen by the competing imidazole was observed to occur in part. Consequently, in its unmodified form, Co-Acacen would be expected to prematurely activate at neutral pH in biological environments by coordinating to non-target molecules containing His residues. Thus, efforts must be directed to enhance the stability of the complex in neutral pH.
The acacen ligand in Co-Acacen contains proton (hydrogen) and methyl substituents. These substituents can be altered to influence the electropositivity of the Co(III) center and therefore axial ligand stability (15). We hypothesized that altering the substituents of the acacen ligand may confer the desired stability to the “inactive” state of the Co(III)-sb complexes (16).
The electron-withdrawing substituents are expected to decrease the electron density of the electropositive cobalt to tighten coordination of the electron-donating axial ligands. This has been demonstrated in computational modeling of Co(III)-sb complexes (15). To this end, electron-withdrawing groups, chloride and phenyl, were chosen as potential stability-enhancing substituents. Three new cobalt complexes derived from the parent Co-Acacen were synthesized with these substituents: Co-ClAcacen, Co-Benacen, and Co-ClBenacen (Figure 5). Successful synthesis and purification were verified by nuclear magnetic resonance (NMR spectroscopy), and characterized both for their pH-dependent lability and stability in neutral conditions.
Materials and Methods
Synthesis of Equatorial Ligands (Acacen, ClAcacen, Benacen, ClBenacen)
These syntheses were done according to literature procedures in references 17-21.
To synthesize Acacen, 2,4-pentadione was cooled to 0°C in 30% v/v ethanol (EtOH). Ethylenediamine was added dropwise while stirring. The solution was brought to room temperature, washed with cold ether, and filtered to produce a white powder (yield=16.6%).
To synthesize the equatorial ligand of Co-ClAcacen, ClAcacen, Acacen was combined with N-chlorosuccinimide (NClS) in dichloromethane (DCM) and stirred on ice for 20 min. The reaction mixture was concentrated by rotary evaporation, recrystallized in water, and filtered to obtain an off-white powder (yield=30.2%).
To synthesize the equatorial ligand of Co-Benacen, Benacen, 1-phenyl-1,3-butanedione was dissolved in EtOH and cooled to 0°C. Ethylenediamine in EtOH was added dropwise while stirring. The solution was warmed to room temperature and dried by rotary evaporation. Product was purified via recrystallization in hot tolune to yield white and shiny crystals (yield=77.0%).
To synthesize the equatorial ligand of Co-ClBenacen, ClBenacen, Benacen was combined with NClS in DCM and stirred on ice for 20 min. The mixture was dried by rotary evaporation and recrystallized in acetone to yield a pastel yellow powder (yield=67.6%).
Synthesis of Axial Ligand, C3Im
Coumarin-3-carboxylic acid, dicyclohexylcarbodiimide (DCC), and N-hydroxysuccinimide (NHS) were dissolved in dry dimethylformamide (DMF) and stirred for 2 h under N2. Histamine in DMF/H2O was added and the reaction was left stirring overnight under N2. The resulting solution was dried by rotary evaporation, filtered and purified by high-performance liquid chromatography (HPLC). After collection and evaporation of the HPLC fractions, the product was resuspended in water and lyophilized to yield a white flocculent powder. (yield=84.0%).
Synthesis of Cobalt Complexes
C3Im, CoBr2*XH2O, and the corresponding equatorial ligand (Acacen, ClAcacen, Benacen, or ClBenacen) were dissolved in methanol (MeOH). N,N-Diisopropylethylamine (DIEA) was added and the reaction was left stirring overnight open to air. The solution was concentrated by rotary evaporation and the product purified by ether precipitation from methanol and ethanol. (Co-Acacen: brown powder, yield=33.3%; Co-ClAcacen: light brown powder, yield=53.9%; Co-Benacen: orange-brown powder, yield=62.1%; Co-ClBenacen: light brown powder, yield=60.5%).
Characterization of pH-dependent Lability
The pH-dependent lability of the complexes in phosphate buffer (PB, 20 mM, pH 8.0) was assessed by monitoring fluorescence with HCl titration. 5 mM solutions of the cobalt complexes were prepared in MeOH. 10μL of each MeOH stock was added to 490μL PB and 0, 3, 7, 7.25, 7.5, 7.75, 8 or 10 μL of 1M HCl were added. The solutions were incubated at 37°C for 1 h and fluorescence intensities were measured using a Synergy 4 High Throughput Fluorimeter with excitation at 334nm and emission at 410nm. The pH of each sample was measured with a Symphony pH probe. To determine if fluorescence intensity changes could be primarily attributed to decrease in pH and not an increase in the ionic strength of the solution, fluorescence intensities were measured using similar conditions, but titrating NaCl in the place of HCl.
Stability of the complexes at neutral pH in 100% MeOH in the presence of a competing imidazole ligand was assessed with fluorescence intensity measurements. 5mM solutions of the complexes were prepared in MeOH. 0, 0.5, 1, or 4 equivalents of a competing ligand (4MeIm) were added to 10μL of the methanolic cobalt solutions. The resultant samples were diluted to a volume of 500μL with MeOH and incubated at 37°C for 1 h. Fluorescence intensity measurements were subsequently acquired as described for the pH-dependent lability studies on Synergy 4 High Throughput Fluorimeter.
Results and Discussion
The pH-dependent lability of the cobalt complexes was investigated by evaluating fluorescence intensity with HCl titrations (Figure 6). Despite altering the substituents on the equatorial ligand, the pH-dependent lability of the axial ligands is retained for the three new complexes. This suggests that changing the electron density does not affect the mechanism by which the C3Im axial ligand can dissociate, namely through protonation of the imidazole moiety. The midpoints of the titration curves (Fluorescence Intensity vs. pH) represent the pH when half of the axial imidazoles have been protonated and presumably dissociated from the cobalt complex. For an effective prodrug, an optimal midpoint would reside in a biologically relevant pH range (between a pH of 4 and 6). The Co-Acacen and Co-ClAcacen exhibit similar midpoints at pH ~4.5. The midpoint for the Co-Benacen complex is shifted to slightly more acidic values (pH 3-4). However, at the pH range tested, a plateau at low pH was not observed, complicating determination of exact midpoint. Additionally, it should be noted that the Co-Benacen complex exhibited slight insolubility in the various solvent combinations tested. These solubility issues were exacerbated with the Co-ClBenacen complex. This may explain the low fluorescence intensity values observed for the complex. The solubility issue precludes determination of the midpoint for this complex under the experimental conditions discussed above. Nonetheless, in comparing the behavior of Co-Acacen, Co-Clacacen, and Co-Benacen, these results suggest that introduction of a phenyl group may affect the pH range at which axial ligand lability is observed whereas introduction of a chloride group elicits minimal effect to this parameter. The complexes exhibited no change in fluorescence with NaCl titrations, indicating that the observed effects result from changes in pH rather than ionic strength.
Given the retained pH-dependent behavior of the complexes, the stability of the complexes at neutral pH in the presence of a competing imidazole ligand was evaluated through titration of 4MeIm, a small molecule model of His residues. Increase in fluorescence intensity with 4MeIm titrations corresponds to stability of the complex (higher increase in fluorescence relative to the control with no 4MeIm corresponds to lower stability). At pH 7.4, the presence of the electron-withdrawing substituents appears to alter stability of the cobalt complexes to ligand exchange (Figure 7). In comparing the complexes, Co-Acacen exhibited greatest reactivity to 4MeIm competition (fluorescence intensity increased by a factor of 3.7), and thus the lowest stability. Co-ClAcacen and Co-ClBenacen exhibited the lowest increases in fluorescence (a factor of 2.4), and thus highest stabilities; interestingly, despite the presence of a phenyl group on only one of the complexes, the behavior of these two complexes was very similar. Co-Benacen fluorescence increased by a factor of 3.0, suggesting that its stability is higher than Co-Acacen but lower than Co-ClAcacen and Co-ClBenacen.
These results imply that the addition of a chloride substituent induced greater effect on the stability of the complexes than addition of a phenyl group. Previous computational studies have shown that substitutions made at the ortho/para position have a greater effect on the thermodynamic and kinetic properties of the molecule, properties that ultimately affect both stability and lability (15). However, in these studies, the chloride groups are incorporated meta to both oxygen and nitrogen while the phenyl group is ortho to the oxygen and para to the nitrogen. Thus, the experimental data deviates from the results predicted by computation if the position of the substituent were the only influencing factor to stability. Our data suggest that electronic differences between the substituents may affect stability to a greater degree than the positions of the substituents of the equatorial backbone.
Conclusions and Future Work
The studies described sought to develop a next generation activatable cobalt complex through improving stability at a neutral pH while retaining pH-dependent lability at a biologically relevant range. Three new complexes were synthesized with varying electron-withdrawing substituents. The complexes differed in their stability and lability from the parent complex. Interestingly, introduction of phenyl groups showed greater effect on the pH-dependent lability of the complex than chloride substituents, whereas chloride substituents conferred greater stability to the complex in the presence of a competing imidazole than the phenyl group at neutral pH.
A major limitation of this study was the solubility of the complexes. The Co-ClBenacen in particular was partially insoluble in both phosphate buffer and MeOH, while the Co-Benacen and Co-ClAcacen were slightly insoluble in phosphate buffer. Therefore, although interesting differences were observed between the complexes, the results may be influenced by differences of solubility. One means by which this challenge was circumvented was by performing the stability studies in MeOH. However, the behavior of the complexes in methanol serves only as proof-of-concept and cannot be directly applied to a biological system. Nevertheless, these results illustrate design principles that can be applied to new generations of activatable cobalt complexes. Future work will explore different fluorophores for the axial ligands and more soluble electron-withdrawing substituents to the equatorial ligand to render the systems biologically relevant.
Another approach to improve stability of Co(III) Schiff base prodrugs that is currently under investigation is focusing on altering the axial rather than equatorial ligand. A strategy is being employed in which a bridging linker is incorporated between two imidazoles to produce a bidentate axial ligand for the complexes. Due to entropic effects, the complex shows enhanced stability compared to the parent Co-Acacen. This approach will be developed further, and it may be combined with alterations to the equatorial ligand (guided by these studies) to produce cobalt complexes with high stability that are activated under disease-relevant external triggers.