Biguanide Iridium(III) Complexes with Potent Antimicrobial Activity.

We have synthesized novel organoiridium(III) antimicrobial complexes containing a chelated biguanide, including the antidiabetic drug metformin. These 16- and 18-electron complexes were characterized by NMR, ESI-MS, elemental analysis, and X-ray crystallography. Several of these complexes exhibit potent activity against Gram-negative bacteria and Gram-positive bacteria (including methicillin-resistant Staphylococcus aureus (MRSA)) and high antifungal potency toward C. albicans and C. neoformans, with minimum inhibitory concentrations (MICs) in the nanomolar range. Importantly, the complexes exhibit low cytotoxicity toward mammalian cells, indicating high selectivity. They are highly stable in broth medium, with a low tendency to generate resistance mutations. On coadministration, they can restore the activity of vancomycin against vancomycin-resistant Enterococci (VRE). Also the complexes can disrupt and eradicate bacteria in mature biofilms. Investigations of reactions with biomolecules suggest that these organometallic complexes deliver active biguanides into microorganisms, whereas the biguanides themselves are inactive when administered alone.


INTRODUCTION
Infectious diseases caused by drug resistant bacteria are currently the second main cause of death worldwide and the third leading cause of death in developed countries. 1 Fungal infections are also a human health threat, 2 their clinical treatment presents profound challenges. 3 Grampositive and Gram-negative bacteria have cell envelopes which guard against changes in osmotic pressure, chemical or enzymatic lysis and mechanical damage, and can survive under extreme conditions. 4 The cell wall of Gram-positive bacteria comprises a thick layer of peptidoglycan and the additional outer membrane of Gram-negative bacteria is populated with lipopolysaccharides, both of which features can protect bacteria from antibiotics. 5,6 Drug resistance was originally found in hospitals where most antibiotics are used, e.g.
sulfonamide-resistant Streptoccoccus pyogenes emerged in the 1930s and Staphylococcus aureus showed resistance to penicillin shortly after it was introduced in the 1940s. 7 Multidrug resistant bacteria, including the notorious Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acetinobacter baumanii, Pseudomonas aeruginosa, and Enterobacteriaceae species, abbreviated as 'ESKAPE', are now a major threat to human health and cause of bacterial infectious diseases with high mortality. [8][9][10] Therefore, novel, effective, and safe antibiotics are urgently needed. 11 Organometallic half-sandwich complexes provide a highly versatile platform for drug design. 12 The antiproliferative and antimicrobial activities of organometallic complexes can be finetuned by choice of the π-bonded arene or cyclopentadienyl ligand, the metal itself and its oxidation state, and by the other monodentate or chelating ligands. 13 We focus here on the third-row transition metal ion Ir III , which with its low-spin 5d 6 outer shell electronic configuration can be relatively inert and therefore likely to reach drug target sites with at least some of its initial ligands still bound. 14 So far, there are relatively few reports on the antimicrobial properties of organometallic iridium complexes. 14-16 Here we have introduced biguanide ligands into organo-iridium cyclopentadienyl complexes.
Biguanides are an important class of compounds that have wide pharmaceutical applications.
One of the best known biguanide derivatives is the drug metformin (Metf), which has been used to treat type II diabetes for over 60 years. Other derivatives like phenformin, buformin, 1-phenylbiguanide and chlorophenylbiguanide are reported to exhibit antimicrobial and antiviral activity (Chart 1). 17

Synthesis and Characterization
Organometallic Ir III complexes 1-14 were synthesised following a reported general procedure, 27 involving reaction of the appropriate chlorido-bridged Ir III dimer and biguanide ligands in anhydrous methanol, to which triethylamine was added, followed by heating at shown in Figure 1. Complex 1 adopts a pseudo-octahedral structure with Ir III bound to a η 5 -Cp* ring, a chelated neutral metformin and chloride as ligands to form an 18e 1+ cation with 'piano-stool' geometry and chloride as the counter anion. In contrast, iridium in complex 4 is also a 1+ cation but bound only to a η 5 -Cp Xbiph ring and deprotonated N,N-bound phenylbiguanide, giving a 16e species, with chloride as counter anion.
The asymmetric unit of complex 4 contains two crystallographically independent but chemically identical complexes, two chloride counter ions and a small amount of electron density modelled as a partially occupied methanol (40% occupancy). The Ir-N bond lengths of complex 1 are slightly longer than in complex 4,

Relative Hydrophobicity
The relative hydrophobicities of complexes 1-14 were determined by RP-HPLC using a reverse-phase C18 column. To ensure solubility of the Ir III complexes, MeOH/H2O, 1:9 v/v was used with NaCl (50 mM) present to suppress the hydrolysis. The HPLC eluents were also prepared with 50 mM NaCl ( Figure S1 in the Supporting Information). The resulting retention times are shown in Antifungal activity against Candida albicans and Cryptococcus neoformans was also studied.
First, the antimicrobial activity of selected biguanide chelating ligands L1-6 and L10 against E. coli, K. pneumoniae, P. aeruginosa, A. baumannii and MRSA, and fungi was determined.
None of these ligands exhibited activity, with MICs > 32 µg/mL.
The more hydrophilic Ir III complexes 1 and 2 are inactive against the majority of the pathogens studied, with MICs over 32 µg/mL (Figure 2), but the activity increased with increase in hydrophobicity (longer HPLC retention times), which is obvious seen from complex 3, with MICs in the range 2->32 µg/mL, probably due to the increased uptake of the complexes within the membranes of the bacteria. 28 This trend is also apparent from the antibacterial activity of complexes 1-4 against MRSA ( Figure S2 in the Supporting Information).
Among the Gram-negative bacteria, complexes 4-7 exhibit the highest potency against A.
baumannii (MICs, 4 μg/mL (5.4-5.8 μM), an important nosocomial non-motile aerobic bacterial pathogen 29 ) and E. coli (MICs 4-8 μg/mL (5.4-11.2 μM)) ( Figure 2 and Table S4 in the Supporting Information). These complexes had moderate potency (MICs 16-32 μg/mL (21.6-44 μM)) towards K. pneumoniae, a cause of various nosocomial infections, e. g. urinary tract, pneumonia, and intra-abdominal infections (Figure 2). 30 However, all biguanide complexes have little activity towards P. aeruginosa (MICs above 32 µg/mL), probably due to the poor membrane permeability (only ca. 8% that of E. coli) and very effective efflux system, which makes P. aeruginosa intrinsically resistant to many antibiotics ( Table S4 in the   Supporting Information). 31 In order to study the effect of halido ligand on the antimicrobial activity, the Cl in complex 7 were substituted by Br and I to obtain complexes 8 and 9, respectively. To increase the hydrophobicity and potentially enhance uptake of the complexes, a sulfonyl group with an aromatic substituent was introduced onto the terminal nitrogen of the biguanide ligand, to obtain complexes 10-14 (Chart 2). Interestingly, complexes bromido 8 and iodido 9 showed higher antibacterial activity against K. pneumoniae compared to chloride complex 7, but were less potent towards E. coli and A. baumannii (Figure 2). By introducing the sulfonyl substituents, the potency of complexes 10-14 decreased dramatically, with MICs all above 32 μg/mL (Figure 2).
The antifungal activity of complexes 1-14 was screened towards C. albicans, a common fungus in humans which can cause superficial mycoses, invasive mucosal infections, and disseminated systemic disease, 32,33 and C. neoformans, an opportunistic yeast that can cause meningitis. 34 Interestingly, complexes 4-9 exhibited excellent antifungal activity against these fungi (MICs  show potent antibacterial activity against E. faecalis (MICs, 0.5-1 μg/mL (0.58-1.45 μM), We next investigated whether generation of reactive oxygen species (ROS) could be a key process in the activity of the complexes. The MICs of complexes 4-10 were assessed under strict anaerobic conditions (generated with Oxoid AnaeroGen 2.5L sachets in a plastic container) for bacteria that lack superoxide dismutase (SOD) and so cannot quench the high levels of superoxide, 42 S. aureus (ATCC 29213) and S. pyogenes (ATCC 151112). As can be seen from Table S5, there were no significant changes in MICs of complexes 4-10 compared with aerobic conditions, which suggests that ROS generation is not a key process in their activity.
We also determined the cytopathic effect of a representative set of complexes 4-7, 11, 12 and 14 on HaCaT human keratinocyte cells (an immortalized, non-tumorigenic cell line, Table   2). 43

Long-term Antibacterial Investigation
The stability of

Resistance Evolution
To investigate the rate of generation of bacterial resistance towards these novel Ir III biguanide complexes, we determined the mutation rate of standard strain S. aureus exposed to complexes aureus remained unchanged, suggesting that Gram-positive bacteria do not rapidly evolve resistance to these Ir III biguanide complexes.

Kinetics of Growth Inhibition
The kinetics of growth inhibition by complex 7 was studied in three different S. aureus cultures of densities 10 5 , 10 7 and 10 8 CFU/mL (Figure 3a-

Synergistic Effects on Antibiotic Resistance
The intrinsic and acquired resistance of pathogens towards antibiotics has become a major problem. 46 Some newer antibiotics have little effect on highly resistant microorganisms. 47  The MICs for co-administration of complexes 4, 7 and 10 and the clinical drugs cefoxitin and vancomycin towards MRSA and VRE, respectively, were determined (  Table 3).
In further experiments, we examined the reverse synergistic effect upon co-incubation of vancomycin (at the concentrations of 0.25, 4 and 2 μg/mL) with complexes 4, 7 and 10.
However, no growth inhibitory towards VRE below 0.5× MIC complex concentrations was observed after 24 h, at 37 o C.

Anti-biofilm Study
Biofilms are integrations of microorganism communities with extracellular polymeric substances which consist mainly of a variety of bio-polymers. 48 The slow growth rate or low metabolism of organisms in biofilms makes the bacteria difficult to eradicate, and thus bacteria in biofilm are more tolerable to antibiotics. 48,49 Biguanide derivatives, both polymers 50 and low weight molecules 21 are reported as biofilm disruptors. We studied the ability of complexes 4-9 (at 100, 50, 30, 20, 10, 5 and 2 μg/mL) to kill S. aureus in biofilm model of soft-tissue biofilm infection. The logarithms of the numbers of bacterial colonies are listed in Table S8. A twofactor analysis of variance was carried out to compare the effectiveness of complexes with negative controls. It is evident from Figure 4, that after treatment of mature biofilms with Ir III biguanide complexes, at the concentrations of 100 and 50 μg/mL, at least a 3 log difference from the negative control is observed, suggesting that at such complex concentrations, over 99.9% of S. aureus are killed. Anti-biofilm efficacy at complex concentrations of 30 and 20 μg/mL decreased significantly, but there was still a reduction of 1.5 and 1 log, respectively, compared to negative control, which indicates that Ir III biguanide complexes can eradicate over 90% biofilm cells at equipotent 10× to 15× MIC concentrations. A decrease of cell viability was still observable, but very limited at lower complex concentrations (10, 5 and 2 μg/mL).
Overall, there was a significant effect of each concentration on the number of CFU and this differed between the complexes (all p-values <0.01, see ANOVA Table S9 in the Supporting

Induced Permeability Changes in Bacterial Cell Walls
In order to gain insight into the mechanism of the potent bactericidal activity of these complexes against Gram-positive bacteria, we evaluated the permeability change of bacterial implies that the bacterial cell membranes are intact and no leakage of nucleobases occurs.
Next, we investigated the change in morphology of cell walls by transmission electron microscopy (TEM). It is apparent from Figure 5d, that complex 7 did not break cell walls at equipotent concentrations of 10× MBC and 50× MBC. This is consistent with the confocal microscopy observations, which indicates that these biguanide Ir complexes are less likely to target bacterial cell walls.

Interaction with Nucleobases and Amino-Acids
To provide preliminary indications of possible target sites for these potent antimicrobial biguanide complexes, we investigated reactions of some active complexes with nucleobases and amino acids.

Antimicrobial Activity
The prevalence of antibiotic resistance towards the traditional clinical drugs has stimulated the development of more novel and potent antibiotics, especially the metal-based antimicrobial agents over the last decade. 5 A broad spectrum of organometallic complexes (e.g. Pt, 53 62 In the present work, Ir III -biguanide complexes 4-9 showed higher antibacterial activity against S. aureus and MRSA, with MICs as low as 1 µg/mL, more potent than the reported Ir III complexes. 59 Interestingly, bromo complex 8 and iodido complex 9 show higher potency in cytotoxicity and haemolytic activity than chlorido complex 7, probably because complexes 8 and 9 are more inert and more stabilized from aquation; the complexes may remain intact before approaching intracellular targets, and this is in agreement with previous observations. 66,67 However, complexes 8 and 9 have similar antimicrobial activity (MIC), which is indistinctively different from complex 7, suggesting that these complexes may have a different mode of action in killing microorganisms and human mammalian cells.

Potential Targets and Mode of Action
These organometallic biguanide complexes contain a π-bonded Cp X ligand which occupies 3 coordination sites and a chelated biguanide, ligand exchange at the 6 th coordination site is facile. In fact some complexes exist as either 16e or 18e species (Chart 2). The binding of the 6 th ligand is therefore quite weak; no binding to the DNA nucleobase guanine (as 9-EtG or 5'-GMP) was detected, nor to the amino acids histidine, tryptophan and leucine. However, the thiol amino acid L-Cys reacted readily and displaced the biguanide ligand (Figures S3 and S4 in the Supporting Information).
L-Cysteine is an important biosynthetic material for thiol-containing proteins and enzymes in cells, and is the major thiol donor for many intracellular cofactors, e.g. GSH (in eukaryotes and Agents which disrupt the cell wall or membrane, or interfere with essential enzymes, are often bactericidal, whereas agents which only inhibit protein synthesis tend to be bacteriostatic. 70,71 In the present study, the biguanide ligands alone are inactive towards all the microorganisms The purity of the synthesized materials has been determined to be ≥95% by elemental analysis, 1 H and 13 C NMR, high resolution MS, and RP-HPLC. (

Biofilm Cultivation and Antibiotic Treatment
Biofilms were prepared according to a reported literature procedure, with modifications. 82 Generally, bacteria strain S. aureus (ATCC 29213) was cultured in synthetic wound fluid (SWF, consisting of 50% fetal bovine serum and 50% autoclaved peptone water, v/v) at 37 o C on an orbital shaker. In a sterile falcon tube, polymerized rat tail collagen matrix was prepared and kept on an ice bath. Typically, 10 mL collagen matrix, 2 mL collagen stock solution (10 mg/mL), was mixed with 6 mL SWF, 1 mL 0.1%, v/v acetic acid and 1 mL 0. We used R to fit an ANOVA on log-transformed data to determine the effect of complex identity and concentration (and their interaction) on numbers of viable bacteria ( Table S10 in the Supporting Information). 82 We then used the lsmeans package to perform a one-sided test of whether each concentration of each complex led to a reduction in numbers of viable bacteria, 84 compared with the mean number of bacteria recovered from 9 replica untreated cultures (which was 3.03e+9). A Tukey correction for multiple comparisons was used.

Live/Dead Cell Assessment by PI Staining
1 × 10 8 CFU/mL of S. aureus cells were seeded in 50-mL Falcon tubes and exposed to two

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: xxxx Crystallographic data (Table S1-3), antimicrobial activity and RP-HPLC retention times (Table S4), antibacterial activity under aerobic and anaerobic conditions (Table S5), selectivity factors and stability testing (Tables S6 and S7), antibacterial activity of complexes and clinical drugs ( Table S8), effect of biofilm disruption data (Tables S9 and S10), MBC/MIC ratios (Table S11); and relative hydrophobicity measurements (Figure S1), correlation of retention times with MICs ( Figure S2), 1 H NMR spectra of L-Cys reactions (Figures S3 and S4) and      Table S4.