Fluoroquinolone–metal complexes: A route to counteract bacterial resistance?

Journal of Inorganic Biochemistry 138 (2014) 129–143

Contents lists available at ScienceDirect

Journal of Inorganic Biochemistry journal homepage: www.elsevier.com/locate/jinorgbio

Fluoroquinolone–metal complexes: A route to counteract bacterial resistance? Maria J. Feio a, Isabel Sousa a, Mariana Ferreira a, Luís Cunha-Silva a, Raúl G. Saraiva a, Carla Queirós a, José G. Alexandre a, Vasco Claro a, Adélia Mendes a, Rosa Ortiz b, Sandra Lopes a, Ana Luísa Amaral a, João Lino a, Patrícia Fernandes a, Ana João Silva a, Lisete Moutinho a, Baltazar de Castro a, Eulália Pereira a, Lourdes Perelló b, Paula Gameiro a,⁎ a b

REQUIMTE, Departamento de Química e Bioquímica, Faculdade de Ciências da Universidade do Porto, Rua do Campo Alegre, 4169-007 Porto, Portugal Departamento de Química Inorgánica, Facultad de Farmácia, Universidad de Valencia, Avda. Vicent Andrés Estellés S/N, 46100 Burjassot, Valencia, Spain

a r t i c l e

i n f o

Article history: Received 7 March 2014 Received in revised form 16 May 2014 Accepted 20 May 2014 Available online 27 May 2014 Keywords: Fluoroquinolones Metalloantibiotics Bacterial resistance Solution equilibria

a b s t r a c t Microbial resistance to antibiotics is one of the biggest public health threats of the modern world. Antibiotic resistance is an area of much clinical relevance and therefore research that has the potential to identify agents that may circumvent it or treat resistant infections is paramount. Solution behavior of various fluoroquinolone (FQ) complexes with copper(II) in the presence and absence of 1,10-phenanthroline (phen) was studied in aqueous solution, by potentiometry and/or spectrophotometry, and are herein described. The results obtained showed that under physiological conditions (micromolar concentration range and pH 7.4) only copper(II):FQ:phen ternary complexes are stable. Hence, these complexes were synthesised and characterised by means of UV–visible and IR spectroscopy, elemental analysis and single-crystal X-ray diffraction. In these complexes, the FQ acts as a bidentate ligand that coordinates the metal cation through the carbonyl and carboxyl oxygen atoms and phen coordinates through two Natoms forming the equatorial plane of a distorted square-pyramidal geometry. The fifth position of the penta-coordinated Cu(II) centre is generally occupied axially by an oxygen atom from a water molecule or from a nitrate ion. Minimum inhibitory concentration (MIC) determinations of the complexes and comparison with free FQ in various E. coli strains indicate that the Cu-complexes are as efficient antimicrobials as the free antibiotic. Moreover, results strongly suggest that the cell intake route of both species is different supporting, therefore, the complexes' suitability as candidates for further biological testing in FQ-resistant microorganisms. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Quinolones are a family of widely prescribed synthetic antibacterial agents due to their broad spectrum of activity and safety profile. Their mechanism of action relies on the inhibition of the enzymes responsible for DNA replication—topoisomerase II (or DNA gyrase) and topoisomerase IV [1,2]. Due to the limited activity of the first quinolones (e.g. nalidixic acid), structural changes to the basic nucleus were introduced to broaden their antibacterial spectrum of activity, namely the introduction of a fluorine atom in the 6-position of the basic quinolone ring, giving rise to fluoroquinolones (FQs) (Fig. 1) [3]. Certain quinolone features remain constant throughout the class of antimicrobials and are believed to be required for antibacterial activity. However various chemical substitutions have taken place over the decades in an attempt to broaden the spectrum of activity and potency of the FQ, leading to their classification in terms of generations. A bicyclic aromatic core, the basic quinolone structure, is formed by a pyridone ⁎ Corresponding author. Tel.: +351 220402589; fax: +351 220402659. E-mail address: [email protected] (P. Gameiro).

http://dx.doi.org/10.1016/j.jinorgbio.2014.05.007 0162-0134/© 2014 Elsevier Inc. All rights reserved.

ring (shown on the right in Fig. 1) fused to another aromatic ring which contains a carbon at the 8-position in most cases (X8 = C), but that can also contain a nitrogen or even a fluorine, as in the case of sparfloxacin (spx) and lomefloxacin (lmx) (Fig. 2). Both the carboxylic acid at the 3-position and the ketone at C-4 are generally required for antimicrobial activity, as is the R1-substituted nitrogen at the 1-position. In most modern agents, a 5- or 6-membered cyclic diamine R7 is present, attached through one of its nitrogens to the 7-position. Most FQ are unsubstituted at C-5 (R5 = H) but some recent compounds have small substituents such as methyl or ethyl at this site. Substitutions at C-2 are usually deleterious, nevertheless exceptions have been described (e.g. prulifloxacin has a C-2 substituent that forms a ring with R1) [3]. Microbial resistance to antibiotics is one of the biggest public health threats of the modern world [4]. The emergence and dissemination of antimicrobial resistance is a complex problem driven by many interconnected factors: the overuse of antimicrobials, their misuse due to lack of access to appropriate treatment and their underuse due to lack of financial support to complete treatment courses. On the basis of the antibiotic resistance phenomena are the bacterial adaptations which occur at an impressive rate and reduce antimicrobial efficiency. Limiting the

130

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

R5 5 6

4

3

1

2

7 7

R

8

X8

R2

R1 Fig. 1. Chemical structure of the FQ basic nucleus. Detailed information on the substituents R1-R7 and X8 can be found in the text.

than the binary ones, suggesting stabilisation due to an intra-molecular interaction between the ligands. The distribution diagrams indicate that only the copper(II) ternary species are stable at physiological concentrations and pH. Hence, the synthesis, characterisation and single-crystal X-ray diffraction (XRD) structure of several copper(II) complexes of FQs and N-donor heterocyclic ligand phen were undertaken and are herein described. These ternary complexes exhibit five-coordinated centers {CuN2O3} with a slightly distorted square-pyramidal geometry that includes two oxygen atoms from the FQ and two nitrogen atoms from the phen ligand in the equatorial plane. The axial position is generally occupied either by the oxygen from a water molecule or from a nitrate ion. Overall data on the antibacterial activity of these compounds and intake route are also reported and discussed. Obtained data suggest that, in many cases, there is a different translocation route for the free FQ and its ternary copper complex which make the copper(II) ternary complexes particularly promising as a route to overcome bacterial resistance to FQs. 2. Experimental

permeability of the cell to drugs is a known mechanism of antibiotic resistance, by down-regulating or altering a porin required for cell entry [5]. Point mutations that compromise the porin affinity for the antibiotic or expression of a different porin that blocks the antimicrobial translocation are observed manifestations of bacterial response to antibiotic therapy [6]. According to their chemical properties, some antibiotics, such as macrolides and other lipophilic drugs, do not seem to require protein channels, crossing the lipid bilayer directly [7]. Other antibiotics however require porins to enter the bacterial cell: influx mechanisms of β-lactams and FQ, for example, are regularly described as being dependent of the porins OmpF and OmpC [8,9]. Various strategies have been suggested to control the growing problem of microbial resistance to available antibiotics however resistance to antimicrobials is a result of an intrinsic property of the pathogens, their astonishing adaptability. Thus, resistance is currently and will continue to be a problem, and safe and effective new antimicrobials are needed now and will continue to be needed in the future. There has been an increasing menace of bacterial resistance to quinolones [5] and the concept that metal complexes as novel derivatives of FQs could be an alternative to conventional drugs has been pushed forward [10,11]. Numerous studies regarding the interaction between quinolones and metal cations have been reported and reviewed in the literature ([12–39] and references therein). In particular, the study of quinolone-copper and quinolone-copper-1,10-phenanthroline complexes has become an increasingly important field since they seem to exhibit high affinity towards DNA binding as well as nuclease activity towards plasmid, genomic and internucleosomal DNA ([22,40–48] and references therein). Copper is a physiologically relevant metal ion that plays an important role in many biological processes both as an essential trace metal and as a constituent of various exogenously administered compounds in humans [49]. Since the discovery, in 1979 [50], that copper ions were capable of DNA cleavage when complexed to 1,10-phenanthroline (phen, a nitrogen donor heterocyclic ligand), this element's biological activity has been the subject of numerous studies [10,51]. Although the interaction of various metal ions has been thoroughly studied either as binary or ternary complexes with FQs, to date, little has been reported about the solution behavior of these complexes. In this work we review the outcome of experiments conducted in our laboratory regarding the solution behavior of various FQs with Zn2+, Ni2+, Co2 + and Cu2 + in the presence and absence of phen. New data is combined and compared with previously published data [52–55] and provide a consistent view of divalent metal ion complex behavior across the FQ family. In all cases, the values obtained for the stability constants of the binary and ternary divalent metal ion complexes are very high and clearly show that the ternary complexes are more stable

2.1. Materials Ofloxacin (ofx), norfloxacin (nfx), ciprofloxacin (cpx), enrofloxacin (erx), sparfloxacin (spx), levofloxacin (lvx) and lomefloxacin (lmx) were purchased from Sigma-Aldrich. Moxifloxacin (mxfx) was a gift from Bayer®. 1,10-phenantroline (phen), Cu(NO3)2.3H2O and NaOH were purchased from Merck (pro analysis grade). All other reagents were of analytical grade and used with no further purification. All drug solutions were prepared in aqueous 10 mM HEPES (4-(2-hydroxyethyl)-1piperazineethanesulfonic acid) buffer solution pH 7.4 (ionic strength (I) 0.1 M adjusted with NaCl). 2.2. Culture media and bacterial strains Iso-Sensitest broth was obtained from Oxoid (Basingstoke, UK). Antimicrobial susceptibility was assessed for one of the Clinical and Laboratory Standards Institute (CLSI) reference strain, Escherichia (E.) coli ATCC 25922, for E. coli BE BL21(DE3) and E. coli BL21(DE3)omp8, a mutant strain devoid of porins (ΔlamB ompF:Tn5 ΔompA ΔompC) [56]. Some formulations were also tested against a collection of E. coli derived from K12: JF568, the parental strain [57,58] and E. coli JF701 and JF703, strains derived from JF568 but devoid of major porins OmpC and OmpF, respectively [58]. Additionally, two other strains were used: E. coli W3110 [58] tested as a control and W3110 ΔFΔC, a derivative devoid of both porins OmpC and OmpF (H. Wiengart, unpublished results). 2.3. Potentiometric titrations All potentiometric measurements were carried out with a Crison 2002 pH meter and a Crison 2031 burette controlled by a microcomputer. The electrode assembly consisted of an Orion 900029 doublejunction AgCl/Ag reference electrode, and a Russell SWL07 glass electrode as indicator. System calibration was performed by the Gran method in terms of hydrogen ion concentration [60], by titrating solutions of strong acid with strong base. A calibration was performed before each run to determine stability constants which also provided a check to the electrode behavior. All titrations were carried out under argon atmosphere in a thermostat-controlled double walled glass cell with the temperature set at 25.0 (±0.1 °C) and the ionic strength adjusted to 0.1 M with NaCl. 2.3.1. Potentiometric determination of stability constants Stock solutions of mxfx and phen (1.0 × 10− 2 M) were prepared in water (I = 0.1 M NaCl). The concentration of FQ was measured

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

131

Second-generation FQ

cpx

nfx

erx

ofx lmx Third-generation FQ

lvx

spx

Fourth-generation FQ

mxfx

Fig. 2. Structure of the various FQs used in this study. Second generation FQs: norfloxacin (nfx), 1-ethyl-6-fluoro-4-oxo-7-piperazin-1-yl-1H-quinoline-3-carboxylic acid, C16H18FN3O3, MW 319.331 g mol−1; ciprofloxacin (cpx), 1-cyclopropyl-6-fluoro-1,4-dihydro-4-oxo-7-(1-piperazinyl)-3-quinolinecarboxylic acid, C17H18FN3O3, MW 331.4 g mol−1; enrofloxacin (erx) 1-cyclopropyl-7-(4-ethyl-1-piperazinyl)-6-fluoro-1,4dihydro-4-oxo-3-quinolone carboxylic acid, C19H22FN3O3, MW 359.39 g mol−1; ofloxacin (ofx), (RS)-7-fluoro-2-methyl-6(4-methylpiperazin-1-yl)-10-oxo-4-oxa-1-azatricyclo [7.3.1.05,13] trideca-5(13),6,8,11-tetraene-11-carboxylic acid, C18H20FN3O4, MW 361.368 g mol−1and lomefloxacin (lmx), (RS)-1ethyl-6,8-difluoro-7-(3-methylpiperazin-1-yl)-4-oxo-quinolone-3-carboxylic acid, C17H19F2N3O3, MW 351.35 g mol−1. Third generation FQs: levofloxacin (lvx) (2S)-7-fluoro-2-methyl6-(4-methylpiperazin-1-yl)-10-oxo-4-oxa-1-azatricyclo[7.3.1.05,13]trideca-5,7,9(13),11-tetraene-11-carboxylic acid, C18H20FN3O4, MW 361.368 g mol−1 and sparfloxacin (spx), 5-amino-1cyclopropyl-7-[(3R, 5S) 3, 5-dimethylpiperazin-1-yl]-6, 8-difluoro-4-oxo-quinoline-3-carboxylic acid, C19H22F2N4O3, MW 392.41 g mol−1. Fourth generation FQ: moxifloxacin (mxfx), 1cyclopropyl-7-[(1S,6S)-2,8-diazabicyclo[4.3.0]non-8-yl]-6-fluoro-8-methoxy-4-oxo-quinoline-3-carboxylic acid, C21H24FN3O4, MW 401.431 g mol−1.

by checking the compliance of the absorbance of the isosbestic point with the Beer-Lambert law. Aqueous copper(II) nitrate solution (0.01 M) was standardised with EDTA 0.1 M (Titriplex III). For the determination of the acid dissociation constants, an aqueous solution (1-4 × 10−3 M) of the protonated ligand was titrated with NaOH (ca. 0.02 M; I = 0.1 M NaCl; 25 °C) under an argon atmosphere. For the determination of the association constants between the FQ and

phen, an aqueous solution of HCl (1-2 × 10−3 M; I = 0.1 M NaCl; 25 °C), in the presence of both ligands (1-2 × 10− 3 M) was titrated with ~ 0.03 M NaOH, under an argon atmosphere. The stability constants of the binary and ternary complexes were determined by titrating aqueous HCl (1-2.5 × 10 − 3 M; I = 0.1 M NaCl; 25 °C), in the presence of Cu(NO3)2 (1-2.5 × 10−3 M) and of the ligands (1-8 × 10−3 M), with ~ 0.03 NaOH, under an argon atmosphere. Each

132

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

titration was repeated four times in order to check the reproducibility of the data. The equilibrium constants defined by Eqs. (1) and (2) pM þ qHL þ rH þ sA⇌Mp Lq Hr As h βpqrs ¼

M p L q H r As

ð1Þ

i

½M p ½HLq ½Hr ½As

ð2Þ

(where M is metal, HL is the FQ in the zwitterionic form [61], H is the proton and A is phen) were refined by least-squares calculation using the computer program Hyperquad 2008 [62,63] (available at http://www.hyperquad.co.uk) taking into account the presence of the hydroxide species of copper and the autoprotolysis of water. This same procedure had already been used for ofx and nfx [53], erx [54], lvx [55] and lmx [52]. Speciation diagrams were constructed using the application HySS [64] (freeware available at http://www.hyperquad. co.uk).

Elemental analysis of the synthesised ternary complexes showed: Cu-nfx-phen (C28H27CuFN6O7: Calculated: C, 52.38; H, 4.24; N, 13.09. Found: C, 51.71; H, 4.21; N, 12.94. Yield 81% (MW = 642.09 g mol−1); Cu-ofx-phen (C30H33CuFN6O10: Calculated: C, 50.03; H, 4.62; N, 11.67. Found: C, 49.95; H, 4.58; N, 11.75. Yield 69% (MW = 720.16 g mol−1); Cu-spx-phen (C31H37CuF2N7O10: Calculated: C, 48.40; H, 4.85; N, 12.75. Found: C, 47.99; H, 4.83; N, 12.62. Yield 83% (MW = 769.21 g mol−1); Cu-mxfx-phen (C31 H37CuF2 N7 O 10 : Calculated: C, 50.35; H, 5.12; N, 10.68. Found: C, 50.64; H, 5.37; N, 10.50. Yield 61% (MW = 787.25 g mol− 1). Although some of the elemental analysis values for “C” were somewhat unsatisfactory, the X-ray diffraction data definitively support the formula and structure described in the Results and discussion section. Stock solutions of the FQ, Cu(II) and phen were also prepared in HEPES buffer 10 × 10−3 M (pH = 7.4; I = 0.10 M NaCl). Complex formation in solution was achieved by mixing these stock solutions in the proportion 2:1:0 or 1:1:1 (FQ:Cu(II):phen), for binary and ternary complexes, respectively. 2.6. Physical measurements

2.4. Spectrophotometric determination of stability constants Both cpx [65] and spx [66] are relatively insoluble in aqueous solution and the concentrations required for the potentiometric studies (1 × 10−3 M) cannot be reliably attained. Hence, for these two FQs, stability constants were determined spectrophotometrically, a method that yields reliable and comparable results using lower concentrations of drug, as previously reported [53]. Absorption spectra of cpx and spx were recorded in a UNICAM UV-300 spectrophotometer equipped with a constant-temperature cell holder. Spectra were recorded at 25.0 °C in 1 cm quartz cuvettes with a slit width of 2 nm in the wavelength range of 230-400 nm. Spectrophotometric pH titrations were performed in stock solutions of the same metal:ligand molar ratio as previously used in potentiometry —1:1 and 1:2 (1.25 × 10−5-2.5 × 10−5 M) for the binary species and 1:1:1 (2.5 × 10−5 M) for the ternary species. Aliquots of strong acid or base were added to adjust pH to the desired value. pH measurements and system calibration were performed as described previously for potentiometry. This same procedure has already been described for ofx and nfx [53]. 2.5. Preparation of Cu(II) binary and ternary complexes Cu(Hcpx)2Cl2·7H2O complex was synthesised by adding to 50 mL of a solution containing 0.35 g (1.0 × 10− 3 mol) of cpx.HCl in HCl 0.1 M, 0.9 g (17 × 10− 3 mol) of NaCl and 0.085 g (0.5 × 10− 3 mol) of CuCl2·2H2O. A 1 M NaOH solution was added drop wise until pH 5 and the mixture was stirred for 10 min. The reaction mixture was transferred to an ice bath and the crystals obtained by filtration were kept in a desiccator until dry. Elemental analysis for C34H50Cl2CuF2N6O13: Calculated: C, 44.23; H, 5.46; N, 9.10. Found: C, 44.03; H, 5.48; N, 8.91, Yield 58% (MW = 923.24 g mol−1). [Cu(erx)2]Cl and [Cu(erx)(phen)]Cl2 were synthesised as described previously [54]. [Cu(cpx)(phen)](NO3)·4H2O was synthesised as described in [44]. [Cu(lvx)(phen)(H2O)](NO3)·2H2O as described in [55] and [Cu(lmx)(phen)(NO3)]·5H2O as described in [52]. The synthesis of the spx, ofx, nfx and mxfx ternary complexes was performed based on a published procedure [45] with minor alterations. Briefly, to a water/ethanol (1:1) solution containing 0.5 × 10− 3 mol of FQ and NaOH 1 M (0.5 mL for all quinolones with the exception of mxfx which was 1.0 mL), 0.5 × 10− 3 mol of phen were added. To this mixture (under continuous stirring) Cu(NO3)2·3H2O (0.5 × 10 − 3 mol), previously dissolved in water, was added. The greenish/blue solutions (final volume of 25 mL) were concentrated on a rotovapor until precipitation started, and then left to stand at room temperature.

Infrared spectra, measured in the 4000–400 cm−1 range, were recorded in transmission mode on a Perkin-Elmer Spectrum BX spectrophotometer (Waltham, MA, USA) using KBr pellets or in ATR-mode, coupled to a PIKE Technologies GladiATR accessory (diamond crystal). Absorption spectra were recorded in a Varian Cary Elipse 50 Bio and Shimadzu 3600 UV–vis-NIR spectrophotometer equipped with a Peltier temperature controller, in 1 cm quartz cuvettes with a slit width of 2 nm. 2.7. Single-crystal X-ray diffraction Crystalline materials of Cu-nfx-phen and Cu-ofx-phen were manually harvested, immersed in highly viscous inert oil, and a single-crystal for each sample suitable for XRD analysis was mounted on respective CryoLoops [67]. Data were collected on a Bruker X8 Kappa APEX II charge-coupled device (CCD) area-detector diffractometer (Mo Kα graphite-monochromated radiation, λ = 0.71073 Å) controlled by the APEX2 software package [68], and equipped with an Oxford Cryosystems Series 700 cryostream monitored remotely using the software interface Cryopad to set the temperature at 150 or 180 K [69]. The images were processed using the software package SAINT+ [70] and the absorption correction was carried out by the multi-scan semi-empirical method implemented in SADABS [71]. The structures were solved by the direct methods implemented in SHELXS-97 [72,73] allowing the direct location of most of the heaviest atoms. All the remaining non-H atoms were located from difference Fourier maps calculated from successive full-matrix least squares refinement cycles on F2 using SHELXL-97 [73,74]. All non-hydrogen atoms were successfully refined using anisotropic displacement parameters. H-atoms attached to carbon were located at their geometrical positions using appropriate HFIX instructions in SHELXL (43 for the aromatic, 23 for the -CH2- and 137 for the -CH3 groups) and included in subsequent refinement cycles in riding-motion approximation with isotropic thermal displacements parameters (Uiso) fixed at 1.2 or 1.5 × Ueq of the parent C-atom. Additionally, H-atoms associated with the coordinated and uncoordinated water molecules were markedly visible in the difference Fourier maps, and included in subsequent refinement stages with the O–H and H · · · H distances restrained to 0.90(2) and 1.47(2) Å, respectively, and using a riding-motion approximation with an isotropic thermal displacement parameter fixed at 1.5 × Ueq of the respective O-atom. Information about the crystallographic data collection and structure refinement details is summarised in Table 1. Crystallographic data (excluding structure factors) for the structures reported have been deposited with the Cambridge Crystallographic Data Centre (CCDC) as

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

except for mxfx in which the basic group is a diazabicyclonyl ring (Fig. 2). At low pH, both groups are protonated (H2L) and, at high pH, none is protonated (L) [76,77]. The carboxylic group is normally a stronger acid than the distal amino group, therefore the neutral form is rearranged spontaneously to the zwitterion and there is extensive literature showing that the values for pKa1 and pKa2 are approximately 5.56.3 and 7.6-9.3, respectively [3,31,52–55,76–78]. The pKa1 value for the FQ is higher than is generally observed for carboxylic acids, a characteristic that has been associated with an intramolecular hydrogen-bond formation between the carboxylic acid and the neighbouring keto group which results in the stabilisation of the protonated species [33]. As can be seen in Table 2, the values obtained for pKa1 (log β0110) are very similar amongst the FQ studied, in the range of 5.75 to 6.25 and in excellent agreement with previously reported values [3,31,52–55, 76–78]. In contrast, the value of pKa2 (−log β01–10) is very sensitive to the substitutions on the piperazine ring or the substitution of this ring by a diazabicyclonyl ring (mxfx). It is interesting to note that a change as small as the replacement of a proton by an ethyl group, results in a change of 1.2 log units between the pKa2 of cpx and erx.

Table 1 Crystal and structure refinement data for Cu-nfx-phen and Cu-ofx-phen.

Formula Molecular weight Crystal description Crystal size/mm Crystal system Space group a/Å b/Å c/Å α/° β/° γ/° Volume/Å3 Z ρcalculated/g cm−3 F(000) μ/mm−1 θ range/° Index ranges

Reflections collected Independent reflections Final R indices [I N 2σ(I)] Final R indices (all data) Largest diff. peak and hole/e Å3

Cu-nfx-phen

Cu-ofx-phen

C56H66Cu2F2N12O20 1392.29 Green prism 0.26 × 0.05 × 0.07 Triclinic P-1 10.5243(2) 11.5733(3) 13.0166(3) 77.7990(10)° 84.7350(10)° 71.8820(10) 1472.21(6) 1 1.570 722 0.817 3.68–26.73 −13 ≤ h ≤ 13 −14 ≤ k ≤ 14 −16 ≤ l ≤ 16 29649 6179 (Rint = 0.0309) R1 = 0.0339 wR2 = 0.0864 R1 = 0.0416 wR2 = 0.0908 0.796 and −0.345

C30H33CuFN6O10 720.16 Green prism 0.30 × 0.18 × 0.10 Triclinic P-1 10.2708(3) 12.7337(5) 12.8304(5) 73.440(2) 85.530(2) 71.037(2) 1520.91(10) 2 1.573 746 0.794 3.78–27.48 −13 ≤ h ≤ 13 −16 ≤ k ≤ 16 −16 ≤ l ≤ 16 47594 6973 (Rint = 0.0316) R1 = 0.0375 wR2 = 0.0998 R1 = 0.0436 wR2 = 0.1041 1.110 and −0.557

133

3.1.1. Binary systems For the phen/FQ system the data obtained were treated in Hyperquad 2008/HypSpec [62,63] (available for download at http://www.hyperquad. co.uk), assuming the model (charge omitted for simplicity):

supplementary publication numbers: CCDC-990233 (Cu-nfx-phen), and 990231 (Cu-ofx-phen). Copies of these data can be obtained free of charge on application to CCDC, 12 Union Road, Cambridge CB2 2EZ, UK; FAX: (+44) 1223 336033, or online via www.ccdc.cam.ac.uk/data_ request/cif or by emailing [email protected]. 2.8. Antimicrobial susceptibility testing MICs were determined in Iso-Sensitest broth following a standard micro-dilution technique [75]. Stock solutions of the antimicrobials were prepared in 10 × 10− 3 M HEPES buffer pH 7.4 (I = 0.10 M NaCl), and stored, protected from light, at 4 °C. Immediately before each MIC determination the antibacterial agent solutions were sterilised by membrane filtration and diluted as required in sterile buffer of the same composition. Each assay was repeated at least six times with each antimicrobial agent and independent starting inocula. The bactericidal effect of copper(II) as it's pentahydrate chloride salt and phen were also tested as controls. 3. Results and discussion 3.1. Solution behavior: potentiometric studies FQs have two relevant ionisable functional groups, a carboxylic acid group in the 3-position and a basic piperazinyl group in the 7-position,

HL þ A⇌HLA

β0101

ð3Þ

HL þ A þ H⇌HLHA

β0111

ð4Þ

HL þ A⇌LA þ H

β01−11

ð5Þ

where, as previously stated, HL is the FQ in the zwitterionic form [61], H is the proton and A is phen. The phen acidity constant determined was 4.95 ± 0.02, in agreement with previously published data [52,54, 55,78] and the values of the binary metal(II)/phen (MA) complex used are depicted in Table 3 (obtained from the IUPAC Stability Constants Database). The values of the equilibrium constants (Table 2) obtained for the HLHA species (log β0111) show that they are very similar for most of the FQ/phen adducts. The same cannot be immediately said for the LA species (log β01–11), nevertheless, as these values are pKa2 dependent, data correction (log LA) allows us to conclude that the values are also very similar. These results are in the expected range considering that FQ/phen adducts occur normally by charge transfer, aromatic ring stacking and/or hydrogen-bond formation [79–81]. The data obtained from the titration curves of the M(II)/FQ binary system, for all the M:L molar ratios, was treated with the programs Hyperquad 2008/HypSpec [62,63] (available for download at http://www.hyperquad.co.uk) assuming the model (charge omitted for simplicity): M þ HL⇌MHL

β1100

ð6Þ

Table 2 Logarithm of protonation constants of the FQs described in this work and stability constants (log βpqrs) of its phenanthroline complexes at 25 °C and I = 0.1 M (NaCl). See equilibria in main text for βpqrs cross referencing. ofx [53] H2L L HLHA HLA LA a

log β0110 log β01–10 log β0111 log β0101 log β01–11 log LAa

6.10 −8.60 9.41 3.92 −3.35 5.25

logLA = log β01–11–log β01–10.

± ± ± ± ± ±

nfx [53] 0.02 0.01 0.08 0.08 0.07 0.07

6.25 −8.44 9.61 3.91 −4.50 3.94

± ± ± ± ± ±

0.01 0.01 0.06 0.07 0.06 0.06

cpx

erx [54]

spx

lvx [55]

6.15 ± 0.01 −8.95 ± 0.03 9.81 ± 0.06 – −5.35 ± 0.08 3.60 ± 0.08

6.17 ± 0.01 −7.72 ± 0.01 9.34 ± 0.06 2.94 ± 0.08 – –

6.13 ± 0.15 −7.43 ± 0.04 – 4.16 ± 0.04 −3.26 ± 0.09 4.17 ± 0.09

6.02 −8.15 8.85 2.91 −4.22 3.93

± ± ± ± ± ±

0.02 0.04 0.01 0.02 0.02 0.02

lmx [52]

mxfx

5.75 ± 0.05 −8.65 ± 0.13 8.54 ± 0.06 2.30 ± 0.09 – –

6.23 −9.53 9.29 3.58 −5.15 4.38

± ± ± ± ± ±

0.01 0.01 0.06 0.10 0.08 0.08

134

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

Table 3 Stability constants (log βpqrs) of M(II):phen complexes at 25 °C and I = 0.10 M (NaCl). Data obtained from the IUPAC Stability Constants Database.

MA MA2

log β1001 log β1002

Cu(II)

Ni(II)

Co(II)

Zn(II)

9.24 16.01

8.65 16.70

7.08 13.72

6.40 12.20

M þ 2HL⇌MðHLÞ2

β1200

ð7Þ

M þ HL⇌ML þ H

β11−10

ð8Þ

M þ 2HL⇌ML2 þ 2H

β12−20

ð9Þ

The results obtained show that all the equilibria described by Eqs. (6) to (9) can be present, and that the values obtained for their constants are similar to those found in the literature for the stability constants of metal ions with FQs [18,33,46,47,52–55,76,82] (Tables 4 to 6). The formation of Cu2 + complexes have been studied for all FQ. For ofx, nfx, and spx only the stability constants described by Eqs. (6) and (7) could be determined. For cpx the stability constants for the equilibria described by Eqs. (6), (7) and (8) were determined and for the other FQ the stability constants for the four equilibria described could be determined (Table 4). For most of the FQ used the stability constants for the equilibria described by Eqs. (6) to (8) with Ni(II) were determined, with the exception of the formation of lvx and lmx Ni2+-complexes that haven't been studied. For cpx the determination of the equilibrium described by Eq. (7) was not possible but the stability constant for Eq. (9) was obtained; this was also the case for erx. For mxfx, however, only the stability constants for the equilibria described by Eqs. (6) and (7) were obtained. Furthermore, as expected, the values obtained were generally smaller than those found for copper(II) (Table 5). For Co2 + and Zn2 + (Table 6) the only stability constants that could be determined were for ofx, nfx and erx. Studies were also performed for cpx and spx, but no interaction could be quantified under the experimental conditions used. Given these results and the low values obtained for the stability constants determined, no experiments were performed for the remaining FQ. The stability constants determined for the different metal ions with the FQ studied followed, in all cases, the Irving-Williams series and the values obtained for the copper(II) complexes are much higher than those found for the other metal ions. The calculation of the stepwise stability constants is also in agreement with what is statistically expected for this kind of complexes when no significant structural changes occur, i.e., K1 N K2 [83,84]. For all FQs studied, with the exception of cpx, speciation of the Cu2+ complexes as a function of pH (for 1 × 10− 3 M, the concentration range used for the potentiometric titrations), M:L molar ratios of 1:1 and 1:2, shows the predominance of the binary species M(HL), M(HL)2 and/or ML, ML2 (depending on the FQ) in a wide pH range (4.5-9.0). Furthermore, the predominant species near physiological pH

are M(HL) and M(HL)2 for M:L molar ratios of 1:1 and M(HL)2 for M:L molar ratios of 1:2. Speciation diagrams representative of this general species distribution have already been published for ofx and nfx [53], erx [54], lvx [55] and lmx [52]. Cpx displayed a slightly different behaviour evidenced by the speciation diagram shown in Fig. 3 where it can be seen that, unlike the other FQs studied, around physiological pH there is a considerable proportion (circa 35%) of free, un-complexed antimicrobial (HL). For Ni2+, the speciation diagrams are very similar to those of Cu2+ for most of the FQs. However, and under the same conditions, for Co2 + and Zn2+, a mixture of complexed and non-complexed species is observed for all the FQs studied. Fig. 4 provides a representative example of the species distribution for nfx that can be applied to all other FQs studied (except for cpx) and that is in agreement with the data already published for erx [54]. Cpx continues to display the same unusual behaviour already described for the Cu(II)-complexes, showing a large proportion of non-complexed species around physiological pH in the presence of Ni(II). 3.1.2. Ternary systems The stability constants for the ternary systems, M(II):FQ:phen were obtained by the program Hyperquad2008 [62,63] (available for download at http://www.hyperquad.co.uk) assuming the model (charge omitted for simplicity): M þ HL þ A⇌MHLA

β1101

M þ HL þ A⇌MLA þ H

ð10Þ

β11−11

ð11Þ

as well as all the equilibria already described for the binary systems, the acidity constants of the ligands and the protolysis of the divalent metal cations. For Cu2 + the stability constant of the formation of ternary complexes described by Eqs. (10) and (11) were obtained for all FQs (Table 4). For Ni2+ these same constants were determined for all the FQs except lvx and lmx, and as expected, the values were much smaller than those found for Cu2+, although higher than those found for the binary complexes (Table 5). For Co2+ and Zn2+ the only stability constants that could be determined were for ofx, nfx and erx and the same trends as previously described for the binary complexes could be observed (Table 6). For cpx and spx, no interaction could be quantified under the experimental conditions used and given these results, no experiments were conducted for lvx, lmx and mxfx. These results were expectable as they are in good agreement with data obtained from solution studies of some of the FQ-ternary complexes that were carried out to verify the tendency of their formation. Complex formation can be attributed to the combined existence of five- and six-membered rings, the formation of stacking interactions and the formation of π-back bonding, a synergy favouring ternary complex formation [79–81]. The results obtained showed that the most important contribution derives from the metal-ligand π-back bonding from the heteroaromatic N-base (i.e. metal d-electrons are pushed into the vacant π* of phen) present in addition to the metal-ligand σ-donation. Due to this effect, MA2+

Table 4 Association constants (log βpqrs) of Cu(II) complexes determined at 25 °C and I = 0.1 M (NaCl). See equilibria in main text for βpqrs cross referencing. ofx [53]

nfx [53]

cpx

erx [54]

6.24 ± 0.04 – 11.20 ± 0.04 – 16.69 ± 0.03 9.21 ± 0.04

6.95 ± 0.03 – 12.70 ± 0.04 – 17.56 ± 0.05 9.55 ± 0.04

5.48 ± 0.03 −2.49 ± 0.05 – −6.13 ± 0.04 17.96 ± 0.05 8.42 ± 0.04

6.39 −0.42 10.97 −3.32 16.54 9.21

spx

lvx [55]

6.09 ± 0.08 – 12.58 ± 0.08 – 16.46 ± 0.09 8.93 ± 0.09

6.22 −1.02 10.92 −4.52 16.53 9.30

lmx [52]

mxfx

Cu2+ MHL ML M(HL)2 ML2 M(HL)A MLA

log β1100 log β11–10 log β1200 log β12–20 log β1101 log β11–11

± ± ± ± ± ±

0.03 0.04 0.04 0.06 0.05 0.04

± ± ± ± ± ±

0.04 0.08 0.05 0.07 0.04 0.09

6.04 −0.98 10.51 −5.84 16.35 8.45

± ± ± ± ± ±

0.01 0.03 0.15 0.15 0.07 0.05

6.01 −1.32 12.24 −4.89 15.95 7.84

± ± ± ± ± ±

0.04 0.08 0.07 0.04 0.04 0.09

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

135

Table 5 Association constants (log βpqrs) of Ni(II) complexes determined at 25 °C and I = 0.1 M (NaCl). See equilibria in main text for βpqrs cross referencing. ofx [53]

nfx [53]

cpx

erx [54]

4.82 ± 0.03 −2.69 ± 0.06 8.40 ± 0.03 – 11.76 ± 0.06 4.38 ± 0.09

4.49 ± 0.03 −3.51 ± 0.06 8.40 ± 0.03 – 12.47 ± 0.06 4.56 ± 0.09

4.46 ± 0.03 −1.52 ± 0.06 – −2.72 ± 0.04 15.31 ± 0.06 4.38 ± 0.09

4.60 −2.48 8.99 −5.74 14.80 7.75

spx

mxfx

3.92 ± 0.07 −2.6 ± 0.1 9.66 ± 0.03 – – 5.54 ± 0.06

4.68 ± 0.04 0.67 ± 0.08 8.67 ± 0.04 3.11 ± 0.05 13.32 ± 0.09 –

Ni2+ MHL ML M(HL)2 ML2 M(HL)A MLA

log β1100 log β11–10 log β1200 log β12–20 log β1101 log β11–11

remains about as electrophilic as M2+(aq), favouring the formation of the ternary complexes, a phenomenon that in the particular cases of Cu2+ and Ni2+ complexes must prevail over the remaining stabilizing factors given that these metal ions are better π-donors than Co2+ and Zn2+ [54,79]. The speciation as a function of pH, for M:HL:A molar ratios 1:1:1 (in the concentration range used for the constant determinations), shows the predominance of the ternary species described by Eqs. (10) and (11) within the whole pH range for Cu2+ and Ni2+ ions, but not for Co2 + and Zn2 + where, again, a mixture of complexed and uncomplexed species is observed (Fig. 5).

3.1.3. Solution behavior at biological concentrations Considering the possibility of using the FQ-complexes as metalloantibiotics, distribution diagrams were constructed in a micromolar concentration range, similar to that used for the antimicrobial testing of FQ in E. coli strains (Fig. 6). The speciation diagrams for cpx shown are in agreement with previously published data on erx [54], lvx [55] and lmx [52] and are representative of the FQs studied. They clearly show that, under a biologically relevant concentration and pH ranges, only the copper(II) ternary complexes are present. All of the ternary complexes of Ni(II), Co(II) and Zn(II) appear dissociated into the independent ligands, leaving the active FQ free in solution. This same behavior is also displayed by all the binary complexes studied. Many other authors have studied the formation, physicochemical characteristics and biological activity of FQ-metal complexes [10, 12–38,40–42,44–48,82,85,86] however, to the best of our knowledge, no other research group has published data on the solution behavior of this complexes. Similar results have already been described by us for some FQs [52,54,55] and the conclusions drawn then can now be extended to all FQs studied in this work, a representative number of members of this family of antimicrobials. All of our data indicate that only copper(II) ternary complexes are stable enough to lead to potentially useful metalloantibiotics.

± ± ± ± ± ±

0.03 0.06 0.03 0.05 0.06 0.09

3.2. Spectroscopic characterisation of the Cu(II) binary and ternary FQ complexes The binary and ternary complexes synthesised according to the Experimental section were characterised as follows.

3.2.1. UV–visible spectroscopy The electronic spectra of all FQ, phen and all binary and ternary copper complexes were obtained in 10 × 10−3 M HEPES buffer, pH 7.4 (I = 0.10 M NaCl) or water. The UV–vis characteristic bands of the FQs can be seen in Table 7 along with the band shift observed upon binding. In all cases, spectra of the FQs presented the characteristic absorption band around 270-290 nm and one or two smaller bands around 300-320 nm, all of which are in good agreement with spectra previously described [13,35,42,52–55,87–89]. The spectra obtained for the metal complexes exhibited a slight bathochromic shift of the maximum wavelength compared to the spectra of free FQ and phen which indicate complex formation in agreement with previous observations [13,40–42,52]. A characteristic lower intensity band ≈320-360 nm (or ≈420-430 nm in the case of erx) could also be observed in the complex spectra which have previously been assigned to a ligand-to-metal charge transfer (LMCT) for the quinolone ligand [42,52,54,55]. In all cases, the Cu(II)complexes showed a band centred between 620 and 660 nm which is characteristic of square pyramidal geometry in Cu(II)-complexes [87]. UV–vis spectroscopic study of the complexes formed in solution when the components were mixed in stoichiometric proportions showed similar results further confirming the coordination to the metal divalent cation and the tetragonal geometry of the complexes by exhibiting the characteristic bands of O-coordination to the divalent metal cations [83,84]. The spectroscopic characterisation of the ternary complexes, obtained by mixing the components in stoichiometric proportion, also clearly showed that Ni2+ and Co2+ complexes in aqueous solution are

Table 6 Association constants (log βpqrs) of Co(II) and Zn(II) complexes determined at 25 °C and I = 0.1 M (NaCl). See equilibria in main text for βpqrs cross referencing.

Co2+ MHL ML M(HL)2 ML2 M(HL)A MLA Zn2+ MHL ML M(HL)2 ML2 M(HL)A MLA

ofx [53]

nfx [53]

cpx

erx [54]

log β1100 log β11–10 log β1200 log β12–20 log β1101 log β11–11

4.40 ± 0.07 −3.04 ± 0.03 7.82 ± 0.05 −6.62 ± 0.04 11.08 ± 0.05 –

4.37 ± 0.07 −3.59 ± 0.03 7.85 ± 0.05 – 12.64 ± 0.11 3.97 ± 0.07

4.84 ± 0.07 – – −5.20 ± 0.04 – –

4.40 −2.81 7.42 −7.34 11.22 4.07

± ± ± ± ± ±

0.07 0.03 0.05 0.04 0.05 0.02

log β1100 log β11–10 log β1200 log β12–20 log β1101 log β11–11

4.07 ± 0.02 −3.37 ± 0.07 7.28 ± 0.04 – – –

4.07 ± 0.02 −4.34 ± 0.07 7.54 ± 0.04 – – –

– – – – – –

4.22 −2.84 7.50 −6.57 10.30 2.99

± ± ± ± ± ±

0.02 0.07 0.04 0.03 0.04 0.06

136

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

100

100

(A) 80

H2L

CuL2 % formation relative to HL

% formation relative to HL

80

H2 L 60

CuHL

40

HL

Ni(HL)2 60

40

NiHL

NiL HL

20

20

CuL

L

0

0 3

5

7

9

3

5

pH

7

9

pH

Fig. 3. Species distribution as a function of pH of 1:2 Cu(II):cpx complex, calculated from the stability constants listed in Table 4. Concentration (10−4 M) equivalent to the maximum solubility range.

100

(B) 80

% formation relative to HL

octahedral. For the Ni(II) complex, analysis of the spectrum shows the characteristic features of nickel octahedral complexes with two overlapping bands between 612 and 850 nm and a band at 1050 nm, all assigned to d-d transitions and typical of a N/O coordination sphere (spectrum not shown). The analysis of the spectrum for Co2 + is also typical of an octahedral geometry for a N/O coordination sphere, with d-d transitions at wavelengths near 480 and 1050 nm [84]. For Cu2+, the band that appears at 633 nm complies with the expected energy range for the Cu(II) d-d transitions for a similar geometry and a coordination sphere N/O. The stability constants determined for the ternary complexes also follow the Irving–Williams series, which provides additional independent experimental evidence for similar coordination geometry for all the complexes.

H2 L

60

CoL Co(HL)2 40

HL

L

CoHL 20

3.2.2. Infrared analysis The analysis of the FT-IR spectra of both ligands and complex provided information on the coordination mode between the ligands and the metal ion. The infrared spectra of FQs are complex due to the presence of the numerous functional groups in the molecules, therefore their interpretation is based on the most typical vibrations [33], which for the FQs are found between 1800 and 1300 cm−1. All FQs showed the characteristic absorptions band for the carboxylic and the pyridone stretch ν(C_O) at 1740-1715 (circa 1700 cm−1 in mxfx's case) and 1620-1640 cm−1, respectively (Table 7) [90]. Upon binding, ν(C_O)p is shifted towards ≈1630 cm−1 (with the exception of mxfx that has an hypsochromic shift towards 1616 cm−1), the carboxylic stretch disappears and is replaced by two strong and characteristic bands assigned as asymmetric and symmetric vibrations (around 1530-1620 cm−1 and 1310-1450 cm−1, respectively). This indicates the deprotonation of the carboxylate moiety and the probable coordination to the metal ion and is in agreement with data previously published [13,18,30,39,41–45,50,52,54,55,86]. These spectral changes, Δν ≈ 200 cm− 1, indicate a monodentate coordination mode of the carboxylate group [91,92]. The overall changes suggest that the FQs are coordinated to the metal via the carbonyl and one of the oxygen atoms from the carboxylate group. For all complexes, split bands between 3500 and 3000 cm−1 can be attributed to the O-H and N-H stretching vibrations of water molecules

0 3

5

7

9

pH Fig. 4. Species distribution as a function of pH of 1:2 Ni(II):nfx (A) and Co(II):nfx (B) complexes, calculated from the stability constants listed in Tables 5 and 6. Diagram (B) is also representative of species distribution for Zn(II) complexes. Concentrations in the range used for the potentiometric determination of the stability constants (10−3 M).

and piperazinyl moiety, respectively and the strong band at 1384 cm−1 also corresponds to a ν(NO3) vibration confirming the presence of a free nitrate group (not for erx complexes since the counter ion is, in these cases, Cl−). 3.2.3. Crystal Structure description The synthesis of the ternary complexes with the second-generation of FQs nfx and ofx was carried out as described in the Experimental section and further investigated by crystallographic methods. Crystals of Cu-nfx-phen and Cu-ofx-phen were prepared and isolated by controlled evaporation of water–ethanol mixtures, and the respective

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

137

100

100

(A)

CuHLA

(A) CuHLA % formation relative to HL

% formation relative to HL

H2L

80

80

60

40

20

60

40

CuLA

20

CuLA

HL

H2L 0

0 3

5

7

3

9

5

100

9

100

(B) 80

60

CoHLA

40

20

HL

80

CoL2

H2L

(B)

H2L

% formation relative to HL

% formation relative to HL

7

pH

pH

L

60

40

NiHLA

20

HL

NiL2

CoHL Co(HL)2 0

0 3

5

7

9

pH

3

5

7

9

pH

Fig. 5. Species distribution as a function of pH of 1:1:1 Cu(II):cpx:phen (A), also representative of species distribution for Ni(II) and Co(II):ofx:phen (B), also representative of species distribution for Zn(II) complexes, calculated from the stability constants listed in Tables 4 and 6. Concentrations in the range used for the potentiometric determination of the stability constants or in the case of cpx, equivalent to the maximum solubility range.

Fig. 6. Species distribution as a function of pH of 1:1:1 Cu(II):cpx:phen (A) and Ni(II):cpx: phen (B). Diagram (B) is also representative of species distribution of Co(II) and Zn(II) complexes, calculated from the stability constants listed in Tables 4 to 6. Concentrations in the range used for MIC determinations (10−6 M).

structures determined by single-crystal XRD. The crystallographic data (collected or used from the literature) confirms unequivocally the formation of cationic Cu(II) ternary complexes with phen and the distinct FQs: [Cu(nfx)(phen)(H2O)](NO3)∙3H2O (Cu-nfx-phen) which only differs in the hydration water molecules from the previously published complex [45], [Cu(cpx)(phen)](NO3)∙4H2O (Cu-cpxphen) [44], [Cu(erx)(phen)(H2O)](NO3)∙2H2O (Cu-erx-phen) [93], [Cu(ofx)(phen)(H2O)](NO3)∙2H2O (Cu-ofx-phen) and [Cu(lmx)(phen) (NO3)]∙5H2O (Cu-lmx-phen). The molecular structures of these five Cu(II) cationic complexes are depicted in Fig. 7, and reveal similar coordination features. All the Cu(II)-centres are five-coordinated with a geometry that resembles a square pyramid slightly distorted. In the Cu-nfx-phen, Cu-erx-phen and Cu-ofx-phen the FQ molecules behave

as bidentate deprotonated ligands, coordinating to the Cu(II) by the pyridone oxygen and one carboxylate oxygen. The coordination centre is completed by the two nitrogen atoms of the phen ligand and one water molecule, {CuN2O3}. The two oxygen atoms of the FQ ligand and two nitrogen atoms of phen form the basal (equatorial) plane of the square pyramid while the oxygen atom (O1W) of the water molecule is positioned axially. As is typical in complexes structurally analogous, the Cu–O and Cu–N distances in the basal plane are found in the ranges of 1.8-2.0 Å, while the bond Cu–O(water) is considerably longer [around 2.2 to 2.3 Å]. Furthermore, the cis and trans N/O–Cu1–O/N angles are found in the 80-105° and 165-170° ranges, respectively [44,45,87,93]. In the Cu-lmx-phen complex the features of the Cu(II)-coordination centre are identical to the described previously

138

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

Table 7 Characteristic absorption bands of UV–vis and IR spectra of the FQ and metal-complexes. Antimicrobial agent

ofx [Cu(ofx)(phen)H2O]NO3.2H2O nfx [Cu(nfx)(phen)H2O]NO3 cpx Cu(Hcpx)2Cl2 · 7H2O [Cu(cpx)(phen)](NO3).4H2O erx [54] [Cu(erx)2]Cl [54] [Cu(erx)(phen)]Cl2 [54] spx [Cu(spx)(phen)H2O]NO3.3H2O lvx [55] [Cu(lvx)(phen)(H2O)]NO3.2H2O [55] lmx [52] [Cu(lmx)(phen)(NO3)]∙5H2O [52] mxfx [Cu(mxfx)(phen)]NO3. 4.5H2O (a)

UV–vis

IR

λmax (nm)

ν(C_O)p

ν(CO2)asym

ν(CO2)sym

Δ(a)

1622 1624 1624 1631 1622 1634 1628 1628 1634 1632 1641 1636 1621 1589 1612 1616 1622 1616

1714 1613 1734 1588 1700 1530 1610 1740 1606 1594 1717 1576 1725 1628 1722 1582 1704 1576

1714 1384 1734 1384 1700 1310 1472 1740 1383 1382 1717 1384 1725 1384 1722 1458 1704 1454

– 229 – 204 – 220 138 – 223 212 – 192 – 244 – 124 – 122

292 274 276 274 276 273 274 271 266 266 298 274 292 274 282 274 292 272

332 294 316 318 318 323 320 324/336 319 319 368 294 328 294 324 288 336 294

– ≈318–368 – ≈318–368 – ≈320–340 ≈318–368 – ≈420–430 ≈420–430 – ≈318–368 – 320–340 – 320–340 – ≈318–368

– 636 – 636 – – 636 – 640 – 656 – 638 – 642 – 634

Δ = ν(CO2)asym − ν(CO2)sym.

however the axial position of the square pyramid is occupied by a nitrate anions instead of the water molecule [52]. In the Cu-cpx-phen complex, the cpx molecule behaves as a tridentate deprotonated bridging ligand, coordinating to one Cu(II) centres through the pyridone oxygen and one carboxylate oxygen and to a neighbour

Cu(II) centres through one nitrogen of the terminal piperazine ring. The coordination sphere is completed by the two nitrogen atoms of the phen yielding a {CuN3O2} coordination sphere (Figs. 7B and 8). The crystallographic independent Cu(II) centre is ultimately penta-coordinated revealing a square pyramid geometry slightly distorted, with the two

Fig. 7. Schematic representation of the crystal structures of the cationic complexes in the ternary compounds with second-generation FQs: (A) [Cu(nfx)(phen)(H2O)](NO3)∙3H2O (Cu-nfxphen), (B) [Cu(cpx)(phen)](NO3)∙4H2O (Cu-cpx-phen) [51], (C) [Cu(erx)(phen)(H2O)](NO3)∙2H2O (Cu-erx-phen) [107], (D) [Cu(ofx)(phen)(H2O)](NO3)∙2H2O (Cu-ofx-phen) and (E) [Cu(lmx)(phen)(NO3)]∙5H2O (Cu-lmx-phen) [61]. Hydrogen atoms were omitted for clarity.

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

139

features of the Cu-lvx-phen complex, in particular the geometric characteristics of the Cu(II) coordination centre, including bond distances and angles, and the crystal packing arrangement are comparable to those previously reported for the ternary Cu complexes with the remaining second/third-generation FQs (Cu-nfx-phen, Cu-cpx-phen, Cu-erx-phen and Cu-ofx-phen). 3.3. Antimicrobial activity

Fig. 8. Schematic representation of 1D coordination polymer (chain) [Cu(cpx)(phen)]+, formed in the crystal structure of Cu-cpx-phen [51]. Hydrogen atoms were omitted for clarity purposes.

oxygen atoms of the cpx ligand and two nitrogen atoms of phen in the equatorial positions and the nitrogen atom of the piperazine ring of a crystallographic equivalent cpx molecule in the axial position. This coordination motif lead to the formation of a cationic 1D coordination polymer chain (Fig. 8), [Cu(cpx)(phen)]+, in contrast with all the remaining structures that revealed discrete mononuclear complexes. The crystal packing of mononuclear cationic complexes (in the Cu-nfx-phen, Cu-erx-phen and Cu-ofx-phen structures) and of the cationic chains (in Cu-cpx-phen) is mediated by a considerable number of C–H∙∙∙O and weak C-H∙∙∙F hydrogen bonds and π · · · π stacking interactions involving the phen rings of neighbouring complexes. These structural arrangements originate voids which accommodate the NO− 3 counter-ions and crystallisation water molecules. Additionally, these nitrate ions and water molecules are crucial in the stabilisation of the extended crystalline structures, since they engage in a series of strong and highly directional O–H · · · O and O–H · · · N hydrogen bonding interactions with the coordinated water molecules, the oxygen atoms of the pyridone and carboxylate groups, and the nitrogen atoms of the FQs ligands. The synthesis of ternary complexes with some other third- and fourth-generation FQs was also investigated for single-crystal XDR however, only the Cu-lvx-phen complex produced crystalline material suitable for crystallographic analysis [55]. Despite numerous experimental attempts it was not possible to isolate adequate crystals of Cu-spx-phen or Cu-mxfx-phen. Furthermore, cautious search in the literature and in the Cambridge structural database (CSD, Version 5.34) [94,95] did not revealed any crystal structure for these two complexes. The crystal structure of Cu-lvx-phen, recently reported by our group, confirms the formation of a ternary complex, [Cu(lvx) (phen)(H2O)](NO3)∙2H2O (Fig. 9) [55]. The main geometrical structural

Fig. 9. Schematic representation of the crystal structure of the cationic complex [Cu(lvx)(phen)(H2O)](NO3)∙2H2O (Cu-lvx-phen). Hydrogen atoms were omitted for clarity purposes. For more details about the structure see reference [64].

The antimicrobial efficacies of the FQs used in this comparative study and their derivatives were tested against Gram-negative model strains: E. coli ATCC 25922, a standard reference strain; E. coli BE Bl21(DE3); E. coli BE Bl21(DE3)omp8, derived from the previous one but devoid of the major outer-membrane porins; and also, in some cases, against a collection of E. coli strains derived from the K12-strain: JF568, JF701 and JF703, the latter two being porin-deficient mutants (OmpC− and OmpF−, respectively). Additionally, two other strains were used, E. coli W3110 [59] and derivative W3110 ΔFΔC, that is devoid of both porins OmpC and OmpF. The porin-deficient mutants were used to assess whether the cellular uptake of the antimicrobial agents was porin dependent and to compare the intake route of the free FQ with that of their metal-derivatives. The MIC values determined for the FQs against the reference strain following the CLSI methodology are presented in Table 8 in μM and are, in all cases, within the quality control range described in the CLSI Performance Standards [96], which validates the methodology followed. A curious fact that arose from the conversion of the acceptable limits for the MIC values from the commonly used units of μg/mL to M was that, the common misconception that activity increased as the FQ generations evolved is not strictly correct. As can be seen from the MIC of the various FQs assayed in Table 8, their molar efficacies are roughly in the same range, regardless of the generation. This highlights the care that has to be taken when comparing the efficacy of different antimicrobial formulations, especially the ones belonging to different chemical families, as the expression of the activity in μg/mL does not take into account the molecular weight of the individual active ingredients and can be misleading. All MICs determined were 4-20 fold higher for the reference strain than for E. coli BE strain BL21(DE3) whose MICs are within the nM range. E. coli K-12 and E. coli B are the ancestors of the majority of the E. coli laboratory strains, which gave rise to two distinct lineages. Although similar in most of the genomic traits, these two lineages present some relevant differences, namely in what antibiotic susceptibility is concerned, due to the extensive manipulation they were subjected to [97]. Our data highlights some of these E. coli lineage differences and the importance of the detailed description of the organisms chosen for antimicrobial susceptibility testing as even small phylogenetic differences can account for very diverse antibiotic susceptibility profiles. Table 8 also shows that the JF/W3110 collection of E. coli strains which share with E. coli ATCC 25922 the common K-12 ancestor have MIC values in the same order of magnitude as the ATCC reference strain. Given that FQs act on intracellular targets, considerable debate over the entry-route of FQs into the cell has been maintained over the years [6,7,51]. The consensus seems to be that this family of antimicrobials enters the Gram-negative cell through a combination of diffusion through the outer membrane (OM) lipidic layer, controlled diffusion through OM porins and through the interface porin-bilayer. The contribution of each of these routes of FQ intake may depend on specific FQ properties, such as relative hydrophobicity or size. Based on our microbiological results, with the exception of spx and mxfx, all other FQs seemed to rely on the presence of porins for their antimicrobial activity. For all the FQs assayed but spx and mxfx, the MICs obtained for E. coli strains with a double deletion of the porins were invariably higher than those obtained for the parental strain, showing that in the absence of both channel-proteins to facilitate the FQ transport, cells became

140

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

Table 8 MIC results obtained for the different FQ in E. coli strains. The values presented are the mean value calculated for all reproducible results obtained, expressed in μM, accompanied by the values of SD (when dispersion could be determined). Antimicrobial agent

ATCC 25922

ofx [53] nfx [53] cpx erx [54] spx lvx [55] lmx [52] mxfx

0.0415 0.094 0.020 ± 0.005 0.033 0.012 ± 0.003 0.05 ± 0.02 0.16 ± 0.06 0.098

a b c

μg/mL [97]

μMa

0.015–0.12 0.03–0.12 0.004–0.015 – 0.004–0.015 0.008–0.06 0.03–0.12 0.008–0.06

0.0415–0.33 0.094–0.38 0.012–0.045 – 0.010–0.038 0.022–0.17 0.08–0.31 0.020–0.15

Bl21(DE3)

Bl21(DE3)omp8

JF568b

JF701b

JF703b

W3110c

W3110 ΔFΔC

0.0104 0.023 0.0078 0.0040 0.0025 – – 0.0045

0.0415 0.19 0.022 ± 0.007 0.0088 ± 0.0022 – – – 0.0032 ± 0.0005

– – 0.038 ± 0.002 – 0.0622 0.09 0.50 ± 0.07 0.22 ± 0.02

– – 0.03 ± 0.01 – 0.0155 0.09 0.48 0.11 ± 0.04

– – 0.15 – 0.0622 ± 0.020 0.36 0.51 ± 0.10 0.21 ± 0.07

– – – – – – 0.31 ± 0.10 –

– – – – – – 0.53 –

± 0.0026 ± 0.0012 ± 0.0012

± 0.0002

Original acceptable limits for MIC values converted to μM for comparative purposes. E. coli K12 obtained from E. coli Genetic Stock Centre, Yale University, USA. Strain obtained from the School of Engineering and Science, Jacobs University Bremen, Germany.

considerably less susceptible to treatment (≈2 to 72 times higher MIC). This phenomenon may signify a real dependence on at least one of the porins in E. coli FQ influx, but can also be interpreted as the consequence of a steric difficulty in permeation resulting from membrane rearrangement in the absence of its most abundant porins or even the need for the interface between the protein and the lipid bilayer for the antibiotic to enter the cell. As for the role of the individual porins, namely OmpF and OmpC, our data showed that both cpx and lvx required the presence of OmpF for influx as evidenced by the MIC increase in the absence of this porin (Table 8). For the other FQs neither OmpF nor OmpC seemed to play a crucial role in the intracellular transport and the knockout OmpC mutant even showed decreased MIC values relative to the parental OmpC+ OmpF+ strain in the cases of spx and mxfx. These observations can be associated with structural and chemical alterations of the lipidic components of the OM bilayer as a result of the lack of expression of the porin. Surprisingly, this effect was not observed in the absence of OmpF even though it is the most abundant of the OM channel proteins. Another explanation has to do with differential regulation: in the absence of one of the porins, the other is up-regulated as a compensation mechanism. This could also account for the observed effect, as the absence of OmpC leading to a greater availability of OmpF channels would explain how spx and mxfx can translocate more favorably when OmpC is not expressed. Given the complexity of the porin expression regulation process, it is conceivable that knock-out porin mutants are also able to overcome their physiological handicap by expressing an alternative porin which could explain some of our results. In fact, Prilipov et al. [56] have already described some quiescent porins in E. coli B, such as OmpN, which is described as functionally similar to OmpC but of almost negligible constitutive expression levels, and NmpC, whose expression is allegedly impaired by an insertion sequence, though mechanisms of excision have already been reported [98]. Overall, the results obtained point to differences in the entry pathway for the different FQ depending on their physicochemical properties and constitute a good platform for further biophysical studies in order to ascertain the nature of the specific FQ–porin interactions. 3.3.1. Binary systems The MIC values determined can be seen in Table 9. MIC values obtained for CuCl2·5H2O solutions using the same methodology in all strains were around 104-fold higher than those obtained for the antimicrobial agents tested, i.e. 492 × 10−6 M. These results are in accordance with previous reports of Cu(II) E. coli toxicity levels, in the mM range [40,99,100]. Hence, in the event of antibiotic-complex dissociation in the test solutions, any bactericidal effects attributed to the metal cation can be disregarded. In light of the results previously described for the solution studies, only the cpx and erx [54] Cu(II)-FQ complexes were synthesised.

As proof of concept, the antimicrobial activity of solutions of these complexes were compared with the activity of solutions obtained by mixing the components in stoichiometric proportion and the results (Table 9) are similar for all the strains tested. Equivalent results had already been described [52,54,55] for ternary complexes hence, for the remaining FQs the results herein described for the binary complexes were obtained using solutions of Cu(II) and FQ in a 1:2 molar ratio. With the exception of nfx, comparison of the data present in Tables 8 and 9 clearly shows that the MICs of the binary systems are, within experimental error, half those determined for the free antibiotic. Taking into account the stability constants obtained for the copper:FQ systems and the concentration range determined for the MICs, at pH ~ 7.4 the predominant species are free FQ in the zwitterionic (HL) and/or anionic (L) forms (see detailed explanation in Section 3.1.1.). Hence, for a given concentration of binary complex solution, the concentration of the free ligand is approximately twice of that intended for the complex therefore, the MICs obtained are approximately half of those determined for the free antibiotic as stated and do not reflect a higher bactericidal effect but merely the dissociation of the complex. These overall results presented agree with the ones already reported for erx [54], lvx [55] and lmx [52]. In the cases where the determinations were conducted, the MIC of the binary complexes for the porin-deficient strains followed the same dependence profile as the free antimicrobial agents, further supporting the evidence for dissociation. Antimicrobial testing of similar complexes although already reported [40,42,43] does not render itself to direct comparison because results were obtained with different strains and followed a different methodology. Nevertheless, it is worth pointing out that the MIC values obtained for our E. coli BL21(DE3) collection of strains are in the range of nM whereas the strains used in the articles above mentioned have MICs in the μM range. Obviously, a 1000-fold difference will justify a different behavior and distribution of the chemical species in solution. Moreover,

Table 9 MIC results obtained for the different FQ-binary complexes in E. coli strains. The values presented are the mean value calculated for all reproducible results obtained, expressed in μM, accompanied by the vales of SD (when dispersion could be determined). Antimicrobial agenta

ATCC 25922

Bl21(DE3)

Bl21(DE3) omp8

Cu(II):ofx solution (1:2) [53] Cu(II):nfx solution (1:2) [53] Cu(Hcpx)2Cl2 · 7H2O Cu(II):cpx solution (1:2) [Cu(erx)2]Cl [53] Cu(II):erx solution (1:2) [54] Cu(II):spx solution (1:2) Cu(II):mxfx solution (1:2)

0.019 0.17 0.016 ± 0.006 0.013 0.010 ± 0.003 0.011 ± 0.002 0.0070 ± 0.0026 –

0.0048 0.043 0.0017 0.0014 0.0023 0.0024 0.0014 0.0030

0.019 0.17 – 0.08 ± 0.02 – – – 0.0010

a

± ± ± ± ± ±

0.0006 0.0005 0.0007 0.0009 0.0005 0.0010

MIC value obtained for CuCl2.5H2O was 492 μM for all strains.

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

careful analysis of the data previously published [40–43] and conversion of the MIC values presented to a molarity scale (the only unequivocal way to understand the molecular basis of the efficacy of compounds with different MW) showed that, despite the higher efficiency described for Cu(II) complexes against the strains tested, all the other binary complexes (namely for Co(II), Ni(II) and Zn(II)) have MICs of exactly half the magnitude of the free antibiotic which clearly indicates complex dissociation and is in full agreement with the data herein described. Our overall data, combining the solution with the microbiological studies (both with some of the synthesised binary complexes and with the equivalent solution mixtures), indicate that binary Cu(II)-FQ complexes are not stable enough in the concentration range suitable for antimicrobial activity and were therefore not considered worth pursuing in the search for metalloantibiotics. 3.3.2. Ternary systems The MICs obtained for the ternary Cu(II):FQ:phen complexes can be seen in Table 10. MIC values obtained for phen following the same methodology were 22.3 × 10−6 M for all strains, results which are in accordance with previous reports for E. coli toxicity levels [85,101]. This MIC along with the one determined for the Cu(II) salt described in the previous section reassured us that, in the event of complex dissociation, the concentration of these species in solution is below the toxicity threshold and therefore will not be responsible for the antimicrobial effect measured. Comparison of the data obtained for the solutions prepared by dissolution of the synthesised complexes and by the mixing of the three components in stoichiometric proportions (1:1:1) showed once more that there is no difference in their microbial efficacy (within experimental error) and that, effectively, the mixing method to prepare the metallocomplexes is valid for screening purposes as previously suggested [52,54,55]. All of the ternary complexes tested exhibited MICs that were comparable to the ones obtained for the free FQs both in the case of the E. coli ATCC 25992 and E. coli BL21(DE3). The new data described here are in good agreement with the results previously reported for ofx and nfx [53], erx [54], lvx [55] and lmx [52] and support the suggestion that these ternary complexes might be valuable antimicrobial agents, especially in the case of antibiotic resistant species. More diversified results were however obtained with porindeficient strains. There seemed to be a decrease in susceptibility in the absence of porins as shown by the increase in the MIC for E. coli BL21(DE3)omp8 and the JF collection for all Cu(II):phen ternary complexes, with the exception of the mxfx and lmx ternary complexes. This behavior is equivalent to the one seen for the free FQs, however

141

the responses to the individual porin-deficiencies were a little different. Spx which influx seemed to be marginally affected by the absence of OmpC in its free form, in the complexed form appears to be unaffected by the absence of either porin. As previously reported, the tests conducted clearly showed that the ternary erx complex has an antimicrobial effect comparable to that of free erx but results from investigations on the intake pathway of the antimicrobial formulations (data not included) seem to indicate a different translocation route for the free erx and its ternary copper complex [54]. Lmx-Cu(II) ternary complex's biological activity was tested and compared with that of the free lmx and despite some changes in presence/absence of the major OmpF and OmpC porins, the values of MIC were very similar for the two compounds which shows that complexation does not inhibit the activity of lmx. However, the results also suggest that lmx and its ternary complex may have different influx routes, which could prove to be an asset in the fight against microorganisms with permeability-related resistance phenotypes [52]. Moreover, the spectroscopic study of lmx previously conducted [52] confirmed its photo-instability in solution but the study of the ternary complex gave preliminary indications of a substantially improved photostability. The formation of less reactive species and thus a reduction of the photosensitivity attributed to lmx make this lmx-derivative a viable antimicrobial alternative with fewer side effects. Results obtained for lmx also clearly showed that the ternary complex has an antimicrobial effect comparable to that of the free lvx and that there is a different translocation route for the free lvx and its ternary copper complex [55]. Another factor that can contribute to the suitability of the FQ metalcomplexes as metalloantibiotics relates to their solubility. As previously stated, at low pH, both the amine and the carboxylic acid of the FQ are protonated, giving the molecule an overall positive charge. Conversely, at high pH the amine is in the free base form, while the carboxyl group exists as the carboxylate anion, providing a net negative charge. Because of this, quinolones tend to be more soluble in water at acidic and basic pH, with minimum solubility at neutral (physiological) pH values [102]. The crystal packing of quinolones, in which the aromatic nuclei are stacked [103], also contributes to lower aqueous solubility and extremely low solubility has been measured for some quinolone agents [104]. It has been found however that many complexes are actually more soluble than the parent quinolones, leading to the suggestion that the decrease in lipophilicity that occurs on metal complexation may be an important factor that leads to reduced bioavailability [105,106] but can also be exploited to clinical advantage as the increased hydrophilicity might favour translocation through porins. In short, despite the need for further in vitro and in vivo studies, the [Cu(FQ)(phen)] complexes appear as suitable candidates for metalloantibiotic testing. Obviously, more advanced studies should

Table 10 MIC results obtained for the different FQ-ternary complexes in E. coli strains. The values presented are the mean value calculated for all reproducible results obtained, expressed in μM, accompanied by the vales of SD (when dispersion could be determined). Antimicrobial agenta

ATCC 25922

Bl21(DE3)

Bl21(DE3)omp8

JF568

JF701

JF703

Cu(II):ofx:phen solution (1:1:1) [53] [Cu(ofx)(phen)H2O]NO3.2H2O Cu(II):nfx:phen (1:1:1) [53] [Cu(nfx)(phen)H2O]NO3 [Cu(cpx)(phen)](NO3).4H2O Cu(II):cpx:phen solution (1:1:1) [Cu(erx)(phen)]Cl2 [54] Cu(II):erx:phen solution (1:1:1) [54] Cu(II):spx:phen solution (1:1:1) [Cu(spx)(phen)H2O]NO3.3H2O [Cu(lvx)(phen)(H2O)]NO3.2H2O [55] Cu(II):lvx:phen solution (1:1:1) [55] [Cu(lmx)(phen)(NO3)]∙5H2O [52] Cu(II): mxfx:phen solution (1:1:1) [Cu(mxfx)(phen)]NO3. 4.5H2O

0.050 0.083 0.107 0.185 0.026 0.026 0.027 0.025 0.020 0.018 0.07 0.07 – – 0.018

0.012 0.019 ± 0.008 0.027 0.047 ± 0.015 0.0028 ± 0.0014 0.0034 ± 0.0011 0.0047 ± 0.0017 0.0041 ± 0.0014 0.0031 ± 0.0012 0.006 ± 0.003 – – – 0.0046 ± 0.0005 0.006 ± 0.002

0.050 0.08 ± 0.02 0.21 0.43 ± 0.02 – 0.04 ± 0.02 0.013 ± 0.002 0.014 ± 0.003 – 0.05 ± 0.002 – – – 0.0017 ± 0.0006 0.008 ± 0.003

– 0.17 ± 0.06 – 0.25 ± 0.07 0.028 ± 0.003 0.026 ± 0.005 – – 0.117 ± 0.020 0.056 ± 0.004 0.08 ± 0.02 0.13 ± 0.03 0.46 ± 0.11 – 0.09 ± 0.02

– 0.21 ± 0.01 – 0.29 ± 0.06 0.025 ± 0.005 0.028 ± 0.006 – – 0.124 0.10 ± 0.04 0.12 ± 0.03 0.11 ± 0.04 0.55 ± 0.14 – 0.06 ± 0.03

– 0.6 ± 0.2 – 1.5 ± 0.4 0.12 ± 0.04 0.09 ± 0.06 – – 0.124 0.15 ± 0.07 0.19 ± 0.05 0.31 ± 0.08 0.59 ± 0.19 – 0.18 ± 0.03

a

± 0.023 ± 0.094

± ± ± ±

0.003 0.007 0.004 0.006

± 0.005

MIC values obtained for CuCl2.5H2O and 1,10-phenanthroline were 492 μM and 22.3 μM, respectively, for all strains.

142

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143

be undertaken using resistant strains exhibiting FQ-resistance phenotypes and experimental conditions need to better mimic the physiological environment to take account of the enormous complexity of the eukaryotic host's biological fluids. Several of their components (i.e. proteins, lipids or small organic molecules) might compete for the metal coordination sites jeopardizing both the complex stability and their ability to reach the pathogen's cellular target. 4. Overview and conclusions The determination of the solution behavior of the metal ion (Cu2+, Ni2+ Co2+ and Zn2+) complexes with various FQ in the presence and absence of phen (both the results herein described and our previous solution studies published) showed that, although all metal ions are able of forming binary and ternary complexes with FQ, only the Cu(II) complexes are stable enough in solution under biologically relevant conditions. All other complexes, under physiological conditions (pH ≈ 7.4 and concentration in the μM range), are dissociated into free FQ, metal ion and phenanthroline (if present). In light of these results, the synthesis of some of the copper binary and ternary complexes was performed. Complexes were characterised by elemental analysis, UV-vis and FT-IR spectroscopies. The structure of the crystals prepared was obtained by XRD and all results showed that the ternary complexes formed exhibit a five-coordinated centre {CuN2O3} with a slightly distorted square-pyramidal geometry that includes two oxygen atoms from the FQ and two nitrogen atoms from the phen ligand in the equatorial plane and, one oxygen atom from a water molecule or a nitrate anion generally occupying the axial position. Biological tests conducted with solutions of this complexes and solutions with the complex components mixed in equivalent proportions (Cu(II):FQ 1:2 and Cu(II):FQ:phen 1:1:1) showed no significant difference in efficacy and demonstrate that is possible to screen the antimicrobial activity of a metalloantibiotic prior to the synthesis and characterisation of the pure complex by using an appropriate mixture of the complex components. The results obtained for various E. coli strains clearly showed that the ternary complexes have an antimicrobial effect comparable to that of the free FQ and showed that in most cases there is a different translocation route for the free antimicrobial and its ternary copper complex. Our overall observations support the ternary complexes stability in the presence of the model target microorganisms and their suitability as candidates to further biological testing using MDR-species exhibiting a FQ-resistance phenotype. In this context, we feel that solution studies are paramount to the understanding of the behavior of metalloantibiotics since, despite their limitations, they are a better representation of an in vivo situation. Not many research teams have focused their efforts in this area and therefore, most of the published work relates biological activity of metalloantibiotics with their properties in the solid state, leading to extrapolations and, in some cases, erroneous conclusions. The FQs have become an increasingly popular class of antibiotics for use in a variety of infections. As previously stated, newer drugs in this family have been developed with a broader spectrum of activity including better coverage of gram-positive organisms and, in some cases, even anaerobes. However, toxicity has been associated with some of these newer agents. For instance, grepafloxacin has been withdrawn from the market by the manufacturer because of adverse cardiac events [107], spx was withdrawn from the US market in February 2001 primarily due to lack of sales but also due to a controversial safety profile, trovafloxacin was withdrawn from the European markets because of the risk of hepatic toxicity [108] and gatifloxacin was withdrawn because of an increased frequency of hypoglycemia and hyperglycemia compared to other marketed FQs [109]. Our data, namely regarding lmx, suggests a significant increase in the stability of the [Cu(lmx)(phen)(NO3)] complex over that of the free lmx, associated with decreased photodecomposition and therefore possibly leading to a reduction in toxicity [52]. The formation of less reactive species and thus a reduction of the photosensitivity

attributed to lmx make this lmx-derivative a viable antimicrobial alternative with fewer side effects a phenomenon that is also worth investigating for some of the never FQ agents with an arguable safety profile. Moreover, the development of new drugs has stagnated in recent years with very few new antimicrobial agents entering human trials and even fewer being approved for use in humans. In particular, no new classes of antibiotics to treat Gram-negative bacilli were proposed for more than 40 years (since the introduction of FQs). The challenges posed to overcoming antibiotic resistance by developing new drugs are mostly due to reduced investment from the pharmaceutical industry. Discovery and development of antibiotics has become scientifically more complex, more expensive, and more time consuming over time; antibiotics represent a poor return on investment relative to other classes of drugs and lastly because, over the past decade, the pathways to antibiotic approval through the U.S. FDA and European Medicines Agency have become confusing, generally inapplicable and of questionable relevance to patients and clinicians. Our data suggesting the potential of FQ metal complexes as antimicrobial agents proposes the use of new formulations, and not necessarily new compounds which could be a means to introduce faster-track therapies with reduced trial periods and lesser regulation hoops to jump through. In sum, [Cu(FQ)(phen)] complexes appear as suitable candidates for more advanced metalloantibiotic testing, potentially exhibiting efficacy, stability and toxicity advantages over their respective free FQs. Acknowledgements This work was funded by FEDER funds through the Programa Operacional Factores de Competitividade—COMPETE, the Quadro de Referência Estratégico Nacional—QREN and by national funds throught Fundação para a Ciência e a Tecnologia (FCT, Portugal) through EUMRTN-CT-2005-019335 (Translocation), PTDC-SAU-FAR/111414/2009 and Pest-C/EQB/LA0006/2011 projects. The authors are also grateful for specific funding toward the purchase of the single-crystal X-ray diffractometer. IS thanks FCT for the PhD scholarship SFRH/BD/47486/ 2008. MJF and LCS are in debt to Programa Ciência 2008 (Programa Operacional Potencial Humano) for their financial support. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jinorgbio.2014.05.007. These data include MOL files and InChiKeys of the most important compounds described in this article. References [1] V.T. Andriole (Ed.), The Quinolones, Academic Press, San Diego, 2000. [2] P. Ball, in: V.T. Andriole (Ed.), The Quinolones, Academic Press, San Diego, 2000, pp. 1–33. [3] K.E. Brighty, T.D. Gootz, in: V.T. Andriole (Ed.), The Quinolones, Academic Press, San Diego, 2000, pp. 34–98. [4] WHO, World Health, Organization, Global Strategy for Containment of Antimicrobial Resistance, 2001. [5] F.C. Tenover, Am. J. Infect. Control 34 (2006) S3–S10. [6] J.-M. Pagès, C.E. James, M. Winterhalter, Nat. Rev. Microbiol. 6 (2008) 893–903. [7] A.H. Delcour, Biochim. Biophys. Acta 1794 (2009) 808–816. [8] F. Yoshimura, H. Nikaido, Antimicrob. Agents Chemother. 27 (1985) 84–92. [9] K. Mahendran, M. Kreir, H. Weingart, N. Fertig, M. Winterhalter, J. Biomol. Screen. 15 (2010) 302–307. [10] A. Serafin, A. Stańczak, Russ. J. Coord. Chem. 35 (2009) 81–95. [11] M. Rizzotto, in: V. Bobbarala (Ed.), A Search for Antibacterial Agents, InTech, 2012, http://dx.doi.org/10.5772/45651. [12] P. Drevenšek, J. Košmrlj, G. Giester, T. Skauge, E. Sletten, K. Sepčić, I. Turel, J. Inorg. Biochem. 100 (2006) 1755–1763. [13] E.K. Efthimiadou, Y. Sanakis, M. Katsarou, C.P. Raptopoulou, A. Karaliota, N. Katsaros, G. Psomas, Polyhedron 27 (2008) 1729–1738. [14] E.K. Efthimiadou, A. Karaliota, G. Psomas, Bioorg. Med. Chem. Lett. 18 (2008) 4033–4037. [15] E.K. Efthimiadou, A. Karaliota, G. Psomas, J. Inorg. Biochem. 104 (2010) 455–466. [16] H.F.A. el-Halim, G.G. Mohamed, M.M. el-Dessouky, W.H. Mahmoud, Spectrochim. Acta A 82 (2011) 8–19.

M.J. Feio et al. / Journal of Inorganic Biochemistry 138 (2014) 129–143 [17] N. Jiménez-Garrido, L. Perelló, R. Ortiz, G. Alzuet, M. González-Alvarez, E. Cantón, M. Liu-González, S. García-Granda, M. Pérez-Priede, J. Inorg. Biochem. 99 (2005) 677–689. [18] M.P. López-Gresa, R. Ortiz, L. Perelló, J. Latorre, M. Liu-González, S. García-Granda, M. Pérez-Priede, E. Cantón, J. Inorg. Biochem. 92 (2002) 65–74. [19] B. Macías, M.V. Villa, I. Rubio, A. Castiñeiras, J. Borrás, J. Inorg. Biochem. 84 (2001) 163–170. [20] B. Macías, M.V. Villa, M. Sastre, A. Castiñeiras, J. Borrás, J. Pharm. Sci. 91 (2002) 2416–2423. [21] D. Perez-Guaita, S. Boudesocque, S. Sayen, E. Guillon, Polyhedron 30 (2011) 438–443. [22] M. Ruiz, R. Ortiz, L. Perelló, J. Latorre, J. Server-Carrió, J. Inorg. Biochem. 65 (1997) 87–96. [23] S.A. Sadeek, J. Mol. Struct. 753 (2005) 1–12. [24] S.A. Sadeek, W.H. El-Shwiniy, J. Mol. Struct. 981 (2010) 130–138. [25] S. Sagdinc, S. Bayari, J. Mol. Struct. 691 (2004) 107–113. [26] K.C. Skyrianou, E.K. Efthimiadou, V. Psycharis, A. Terzis, D.P. Kessissoglou, G. Psomas, J. Inorg. Biochem. 103 (2009) 1617–1625. [27] K.C. Skyrianou, F. Perdih, I. Turel, D.P. Kessissoglou, G. Psomas, J. Inorg. Biochem. 104 (2010) 161–170. [28] K.C. Skyrianou, F. Perdih, I. Turel, D.P. Kessissoglou, G. Psomas, J. Inorg. Biochem. 104 (2010) 740–749. [29] K.C. Skyrianou, F. Perdih, A.N. Papadopoulos, I. Turel, D.P. Kessissoglou, G. Psomas, J. Inorg. Biochem. 105 (2011) 1273–1285. [30] K.C. Skyrianou, V. Psycharis, C.P. Raptopoulou, D.P. Kessissoglou, G. Psomas, J. Inorg. Biochem. 105 (2011) 63–74. [31] I. Turel, N. Bukovec, E. Farkas, Polyhedron 15 (1996) 269–275. [32] I. Turel, I. Leban, G. Klintschar, N. Bukovec, S. Zalar, J. Inorg. Biochem. 66 (1997) 77–82. [33] I. Turel, Coord. Chem. Rev. 232 (2002) 27–47. [34] N. Sultana, M.S. Arayne, S.B.S. Rizvi, U. Haroon, M.A. Mesaik, Med. Chem. Res. 22 (2013) 1371–1377. [35] S.A. Sadeek, W.H. El-Shwiniy, M.S. El-Attar, Spectrochim. Acta A 84 (2011) 99–110. [36] F. Gao, P. Yang, J. Xie, H. Wang, J. Inorg. Biochem. 60 (1995) 61–67. [37] V. Uivarosi, Molecules 18 (2013) 11153–11197. [38] I. Turel, A. Šonc, M. Zupančič, K. Sepčić, T. Turk, Met.-Based Drugs 7 (2000) 101–104. [39] J.R. Anacona, C. Toledo, Trans. Met. Chem. 26 (2001) 228–231. [40] P. Drevenšek, T. Zupančič, B. Pihlar, R. Jerala, U. Kolitsch, A. Plader, I. Turel, J. Inorg. Biochem. 99 (2005) 432–442. [41] E.K. Efthimiadou, M.E. Katsarou, A. Karaliota, G. Psomas, J. Inorg. Biochem. 102 (2008) 910–920. [42] E.K. Efthimiadou, Y. Sanakis, M. Katsarou, C.P. Raptopoulou, A. Karaliota, N. Katsaros, G. Psomas, J. Inorg. Biochem. 100 (2006) 1378–1388. [43] Ε.Κ. Efthimiadou, Υ. Sanakis, Κ. Raptopoulou, Α. Karaliota, N. Katsaros, G. Psomas, Bioorg. Med. Chem. Lett. 14 (2006) 3864–3867. [44] J. Hernández-Gil, L. Perelló, R. Ortiz, G. Alzuet, M. González-Álvarez, M. LiuGonzález, Polyhedron 28 (2009) 138–144. [45] M. Ruiz, R. Ortiz, L. Perelló, G. Alzuet, M. González-Alvarez, M. Liu-González, F. Sanz-Ruiz, J. Inorg. Biochem. 101 (2007) 831–840. [46] S.C. Wallis, B.G. Charles, L.R. Gahan, L.J. Filippich, M.G. Bredhauer, P.A. Duckworth, J. Pharm. Sci. 85 (1996) 803–809. [47] S.C. Wallis, L.R. Gahan, B.G. Charles, T.W. Hambley, P.A. Duckworth, J. Inorg. Biochem. 62 (1996) 1–16. [48] P. Zivec, F. Perdih, G. Psomas, J. Inorg. Biochem. 117 (2012) 35–47. [49] R.R. Crichton, Biological Inorganic Chemistry: An Introduction, Elsevier, Amsterdam, 2008. [50] D.S. Sigman, D.M. Graham, V. D'Aurora, A.M. Stern, J. Biol. Chem. 254 (1979) 12269–12272. [51] B.S. Shehom, J. Pharm. Edu. Res. 1 (2010) 1–20. [52] P. Fernandes, I. Sousa, L. Cunha-Silva, M. Ferreira, B. de Castro, E. Pereira, M.J. Feio, P. Gameiro, J. Inorg. Biochem. 131 (2013) 21–29. [53] P. Gameiro, C. Rodrigues, T. Baptista, I. Sousa, B. de Castro, Int. J. Pharm. 334 (2007) 129–136. [54] R. Saraiva, S. Lopes, M. Ferreira, F. Novais, E. Pereira, M.J. Feio, P. Gameiro, J. Inorg. Biochem. 104 (2010) 843–850. [55] I. Sousa, V. Claro, J.L. Pereira, A.L. Amaral, L. Cunha-Silva, B. de Castro, M.J. Feio, E. Pereira, P. Gameiro, J. Inorg. Biochem. 110 (2012) 64–71. [56] A. Prilipov, P. Phale, R. Koebnik, C. Widmer, J. Rosenbusch, J. Bacteriol. 180 (1998) 3388–3392. [57] T.J. Chai, J. Foulds, J. Bacteriol. 130 (1977) 781–786. [58] J. Foulds, T.J. Chai, J. Bacteriol. 133 (1978) 1478–1483. [59] B.J. Bachmann, Bacteriol. Rev. 36 (1972) 525–557. [60] G. Gran, Analyst 77 (1952) 661–671. [61] J. Inczéby, T. Lengyel, A.M. Ure, Compendium of Analytical Nomenclature, IUPAC, 1997.

143

[62] P. Gans, A. Sabatini, A. Vacca, Talanta 43 (1996) 1739–1753. [63] P. Gans, A. Sabatini, A. Vacca, Ann. Chim. 89 (1999) 45–49. [64] L. Alderighi, P. Gans, A. Ienco, D. Peters, A. Sabatini, A. Vacca, Coord. Chem. Rev. 184 (1999) 311–318. [65] A.I. Caco, F. Varanda, M.J. Pratas de Melo, A.M.A. Dias, R. Dohrn, I.M. Marrucho, Ind. Eng. Chem. Res. 47 (2008) 8083–8089. [66] G. Montay, J. Antimicrob. Chemother. 37 (Suppl. A) (1996) 27–39. [67] T. Kottke, D. Stalke, J. Appl. Crystallogr. 26 (1993) 615–619. [68] APEX2, Data Collection Software Version 2.1–RC13, Bruker AXS, Delft, The Netherlands, 2006. [69] Cryopad, Remote monitoring and control, Version 1.451, Oxford Cryosystems, Oxford, United Kingdom, 2006. [70] SAINT+, Data Integration Engine v. 7.23a ©, Bruker AXS, Madison, Wisconsin, USA, 1997–2005. [71] G.M. Sheldrick, SADABS v. 2.01, Bruker/Siemens Area Detector Absorption Correction Program, Bruker AXS, Madison, Wisconsin, USA, 1998. [72] G.M. Sheldrick, SHELXS-97, Program for Crystal Structure Solution, University of Göttingen, 1997. [73] G.M. Sheldrick, Acta Crystallogr. A 64 (2008) 112–122. [74] G.M. Sheldrick, SHELXL-97, Program for Crystal Structure Refinement, University of Göttingen, 1997. [75] CLSI, Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically: approved standard, CLSI document M07–A8, Eighth Edition, 2009. [76] Y. Kawai, K. Matsubayashi, H. Hakusui, Chem. Pharm. Bull. 44 (1996) 1425–1430. [77] D.L. Ross, C.M. Riley, J. Pharm. Biomed. Anal. 12 (1994) 1325–1331. [78] F. Quentel, M. L'Her, J. Courtot-Coupez, Anal. Chim. Acta. 97 (1978) 373–383. [79] A. Fernández-Botello, A. Holý, V. Moreno, H. Sigel, J. Inorg. Biochem. 98 (2004) 2114–2124. [80] M.S. Luth, L.E. Kapinos, B. Song, B. Lippert, H. Sigel, J. Chem. Soc. Dalton (1999) 357–365. [81] H. Sigel, J. Inorg. Nucl. Chem. 37 (1975) 507–509. [82] Y. Li, Y. Chai, R. Yuan, W. Liang, Russ. J. Inorg. Chem. 53 (2008) 704–706. [83] P. Atkins, T. Overton, J. Rourke, M. Weller, F. Armstrong (Eds.), Shriver & Atkins: Inorganic Chemistry, Oxford University Press, Oxford, 2006. [84] A.B.P. Lever, Inorganic Electronic Spectroscopy, Elsevier, Amsterdam, 1984. [85] M.E. Katsarou, E.K. Efthimiadou, G. Psomas, A. Karaliota, D. Vourloumis, J. Med. Chem. 51 (2008) 470–478. [86] M. Ruiz, L. Perelló, J. Server-Carrió, R. Ortiz, S. García-Granda, M. Diaz, E. Cantón, J. Inorg. Biochem. 69 (1998) 231–239. [87] C.-Y. Chen, Q.-Z. Chen, X.-F. Wang, M.-S. Liu, Y.-F. Chen, Transit. Met. Chem. 34 (2009) 757–763. [88] M. Lizondo, M. Pons, M. Gallardo, J. Estelrich, J. Pharm. Biomed. Anal. 15 (1997) 1845–1849. [89] H.-R. Park, T.H. Kim, K.-M. Bark, Eur. J. Med. Chem. 37 (2002). [90] U. Neugebauer, A. Szeghalmi, M. Schmitt, W. Kiefer, J. Popp, U. Holzgrabe, Spectrochim. Acta A 61 (2005) 1505–1517. [91] G.B. Deacon, R.J. Phillips, Coord. Chem. Rev. 33 (1980) 227–250. [92] K. Nakamoto, Infrared Spectra of Inorganic and Coordination Compounds, Wiley, New York, 1963. [93] J. Recillas-Mota, M. Flores-Alamo, R. Moreno-Esparza, J. García-Mora, Acta Crystallogr. E 63 (2007) (M3030-U1551). [94] F.H. Allen, Acta Crystallogr. B 58 (2002) 380–388. [95] F.H. Allen, W.D.S. Motherwell, Acta Crystallogr. B 58 (2002) 407–422. [96] CLSI, Performance standards for antimicrobial disk susceptibility tests: approved standard, CLSI document M100–S20, Tenth Edition, 2010. [97] F.W. Studier, P. Daegelen, R.E. Lenski, S. Maslov, J.F. Kim, J. Mol. Biol. 394 (2009) 653–680. [98] A.J. Blasband, W.R. Marcotte, C.A. Schnaitman, J. Biol. Chem. 261 (1986) 12723–12732. [99] B. Grey, T.R. Steck, Appl. Environ. Microbiol. 67 (2001) 5325–5327. [100] S.D. Gupta, B.T.O. Lee, J. Camakaris, H.C. Wu, J. Bacteriol. 177 (1995) 4207–4215. [101] A.M. Zoroddu, S. Zanetti, R. Pogni, R. Basosi, J. Inorg. Biochem. 63 (1996) 291–300. [102] C.M. Riley, C.G. Kindberg, V.J. Stella, in: P.B. Fernandes (Ed.), International Telesymposium on Quinolones, J. R. Prous Science Publishers, Barcelona, 1989, pp. 21–36. [103] L.L. Shen, D.T.W. Chu, Curr. Pharm. Des. 2 (1996) 195–208. [104] T. Rosen, D.T.W. Chu, I.M. Lico, P.B. Fernandes, K. Marsh, L. Shen, V.G. Cepa, A.G. Pernet, J. Med. Chem. 31 (1988) 1598–1611. [105] D.L. Ross, C.M. Riley, Int. J. Pharm. 87 (1992) 203–213. [106] D.L. Ross, S.K. Elkinton, S.R. Knaub, C.M. Riley, Int. J. Pharm. 93 (1993) 131–138. [107] WHO, Drug Inf. 13 (4) (1999). [108] European Agency for the Evaluation of Medicinal Products, Public statement on trovafloxacin/alatrofloxacin: recommendation to suspend the marketing authorisation in the European Union. London (1st October 1999), 1999. [109] US Federal Register 73, No. 175, September 9 2008. 52357.