Unprecedented in Vitro Antitubercular Activitiy of Manganese(II) Complexes Containing 1,10-Phenanthroline and Dicarboxylate Ligands: Increased Activity, Superior Selectivity, and Lower Toxicity in Comparison to Their Copper(II) Analogs

Mycobacterium tuberculosis is the etiologic agent of tuberculosis. The demand for new chemotherapeutics with unique mechanisms of action to treat (multi)resistant strains is an urgent need. The objective of this work was to test the effect of manganese(II) and copper(II) phenanthroline/dicarboxylate complexes against M. tuberculosis. The water-soluble Mn(II) complexes, [Mn2(oda)(phen)4(H2O)2][Mn2(oda)(phen)4(oda)2]·4H2O (1) and {[Mn(3,6,9-tdda)(phen)2]·3H2O·EtOH}n (3) (odaH2 = octanedioic acid, phen = 1,10-phenanthroline, tddaH2 = 3,6,9-trioxaundecanedioic acid), and water-insoluble complexes, [Mn(ph)(phen)(H2O)2] (5), [Mn(ph)(phen)2(H2O)]·4H2O (6), [Mn2(isoph)2(phen)3]·4H2O (7), {[Mn(phen)2(H2O)2]}2(isoph)2(phen)·12H2O (8) and [Mn(tereph)(phen)2]·5H2O (9) (phH2 = phthalic acid, isophH2 = isophthalic acid, terephH2 = terephthalic acid), robustly inhibited the viability of M. tuberculosis strains, H37Rv and CDC1551. The water-soluble Cu(II) analog of (1), [Cu2(oda)(phen)4](ClO4)2·2.76H2O·EtOH (2), was significantly less effective against both strains. Whilst (3) retarded H37Rv growth much better than its soluble Cu(II) equivalent, {[Cu(3,6,9-tdda)(phen)2]·3H2O·EtOH}n (4), both were equally efficient against CDC1551. VERO and A549 mammalian cells were highly tolerant to the Mn(II) complexes, culminating in high selectivity index (SI) values. Significantly, in vivo studies using Galleria mellonella larvae indicated that the metal complexes were minimally toxic to the larvae. The Mn(II) complexes presented low MICs and high SI values (up to 1347), indicating their auspicious potential as novel antitubercular lead agents.

Keywords: Mycobacterium tuberculosis, manganese(II), 1, 10-phenanthroline, metal-based complex, antimicrobial agent, Galleria mellonella INTRODUCTION Mycobacterium tuberculosis is a pathogenic, acid-alcohol resistant bacillus and is responsible for the highly contagious and potentially fatal disease, tuberculosis (TB) (Ryan et al., 2014). The bacterium has an unusual, impermeable, waxy coating composed mainly of mycolic acids, a feature in part responsible for its inherent resistance to numerous drugs. The infected host is thought to contain populations of M. tuberculosis in cavitary lesions, closed caseous lesions, and within macrophages (Bennett, 1994). In cavities, the oxygen level is high, the medium is neutral or slightly alkaline and replication is relatively fast. With the other two populations, the oxygen concentration is lower, the medium is neutral or acidic and multiplication is slower. The disease most commonly involves the lungs and is readily spread via aerosol. However, the infection may also spread to distant sites, such as the brain, kidneys, spleen, liver, and bones (Krishnan et al., 2010). In 2016, there were 9.6 million new cases of TB and 1.5 million deaths (Who.int). Furthermore, there has been an alarming rise in the number of patients presenting with multidrug-resistant (MDR) TB, which is defined by resistance to the two front-line drugs, isoniazid (INH), and rifampicin, and extensively drug-resistant (XDR) TB, which additionally exhibits resistance to two of the most important second-line drug classes (fluoroquinolones and injectable agents). The World Health Organization estimated that ca. 4,80,000 people developed MDR-TB in the world in 2014, and that 9.7% of these cases had XDR-TB (Who.int). The treatment for MDR-TB and XDR-TB is costly, toxic, lengthy and less effective than the standard regime, which contributes to medical non-adherence and the emergence of totally drug-resistant strains (TDR-TB). Clearly, in order to effectively treat these highly resistant strains of M. tuberculosis there is an urgent need for new drug classes possessing novel mechanisms of action.
Throughout classical antiquity, empirical formulations comprising metal ions have been used for medicinal purposes. However, following the discovery of penicillin and other drugs of biological and synthetic organic origin the clinical use of metallo compounds declined markedly. But within the past 50 years there has been a renaissance in metal-based pharmaceuticals, driven mainly by problems of efficacy and resistance. Some examples of therapeutically important, metal-containing systemic drugs include platinum and arsenic for cancer treatment, samarium for metastatic tumor pain relief, gold as an anti-arthritic, bismuth as a gastrointestinal antimicrobial, antimony and arsenic as anti-parasitics, iron in cardiovascular disease, lithium for bipolar disorder, barium and gadolinium as diagnostic imaging agents, radioactive isotopes of gallium, indium and technetium in tomography, and radiopharmaceuticals containing strontium, yttrium, samarium, and radium (Gielen and Tiekink, 2005;Dabrowiak, 2009;Mjos and Orvig, 2014). Nanoparticulate silver and silver salts are also being applied topically as antibacterial agents in wound and burn treatments (Stobie et al., 2008;Landsdown, 2010).
There are numerous examples where transition metal complexes have been shown to inhibit the growth of M. tuberculosis in vitro. Integration of metal ions into the drug structure offers structural diversity, possible access to numerous oxidation states of the metal and the potential of enhancing the efficacy of an established organic drug through its coordination to the metal (Viganor et al., 2015). Metal complexes containing a variety of ligands, such as thiosemicarbazones, quinolones, amines, imines and phenanthrolines, have shown substantial growth inhibition of M. tuberculosis. Mechanistic studies have been conducted on metal ligated by the pro-drug INH and some of its derivatives. In particular, the octahedral Fe(II) complex trianion, [Fe(CN) 5 (INH)] 3− , which returned a MIC value of 0.43 µM (based on Na 3 [Fe(CN) 5 (INH)]·4H 2 O), has been scrutinized in detail (Oliveira et al., 2004). Studies have inferred that the mode of action of INH in blocking the synthesis of M. tuberculosis cell wall mycolic acids is linked to the in situ formation of coordination complexes with redox-active metal ions like Cu(II) and Fe(II) (Bernardes-Génisson et al., 2013 In 1969, Dwyer et al. (1969) published their comprehensive, landmark treatise on the in vitro and in vivo antibacterial activities of dicationic Mn(II), Fe(II), Co(II), Ni(II), Cu(II), Zn(II), Cd(II), and Ru(II) chelates containing 1,10-phenanthroline (phen), substituted phen (R-phen), 2.2'bipyridine (bipy), and substituted bipy (R-bipy) ligands. Against M. tuberculosis H37Rv the bipy complexes were considerably less potent than their phen analogs. Metal centers chelated by the 5-NO 2 -phen ligand showed the most potent antitubercular activity, with the substitutionally-inert Ru(II) tris chelate being 128-fold less active. Importantly, the bacilli did not develop significant resistance to the 5-NO 2 -phen complexes. However, treatment with the phen complexes did not increase the survival of M. tuberculosis-infected mice relative to the untreated rodent (Dwyer et al., 1969). The low in vivo activity was attributed to either poor bioavailability at doses safe for the host and/or a failure to access locations where the organism proliferates. More recently, Hoffman and coworkers (Hoffman et al., 2013) (II) and Cu(II) derivatives). Additionally, SI values (based on A549 cells) for the Co(II) complexes were much larger than those of their respective Cu(II) counterparts. The complexes resisted efflux mechanisms in mycobacteria and interfered with multiple biochemical pathways. Dholariya et al. (2013) reported the antitubercular activities of Cu(II) complexes ligated by dicoumarol (dicoum) derivatives and phen, formulated as [Cu(dicoum)(phen)(H 2 O)(OH)]·xH 2 O (Devereux et al., 2000). Against M. tuberculosis H37Rv, only complexes incorporating hydroxylated (-3-OH) and chlorinated (-4-Cl) dicoumarols showed appreciable activity (MIC 90 = 4.05 and 64.8 µM, respectively). Patel et al. (2012) tested an array of similar Cu(II) acyl coumarin/phen complexes which displayed only moderate anti-M. tuberculosis activity (MIC range 49.5->243 µM). Segura et al. (2014) , having MIC values of 11.0 and 14.2 µM (X = NO − 3 and CF 3 SO − 3 , respectively) against H37Rv. In the present study, we report the in vitro anti-M. tuberculosis activity of the water-soluble Mn(II) and Cu(II) phen/dicarboxylate complexes (Figures 1,2 ·3H 2 O·EtOH}n (3) and {[Cu(3,6,9tdda)(phen) 2 ]·3H 2 O·EtOH}n (4) (odaH 2 = octanedioic acid, tddaH 2 = 3,6,9-trioxaundecanedioic acid) (Figure 2 [Mn(tereph)(phen) 2 ]·5H 2 O (9) (phH 2 = phthalic acid, isophH 2 = isophthalic acid, terephH 2 = terephthalic acid) (Figure 2). In addition, we present toxicity profiling data for the complexes, obtained using mammalian VERO (normal kidney) and A549 (adenocarcinomic alveolar) epithelial cells in vitro and against Galleria mellonella larvae for the in vivo systemic toxicity study.

Synthesis of Complexes
All the chemicals were purchased from commercial sources and used without further purification. Complexes 1-9 (Devereux et al., 2000;Kellett et al., 2011;Gandra et al., 2017) were prepared as previously reported by our group.

In Vitro Screening Against M. tuberculosis H37Rv
The anti-mycobacterial activity of the complexes was determined by the resazurin microtiter assay method (Palomino et al., 2002). Stock solutions of the test complexes were prepared and diluted in Middlebrook 7H9 broth (Difco) supplemented with oleic acid, albumin, dextrose, and catalase (OADC enrichment -BBL/Becton-Dickinson), to obtain the final drug concentration range of 0.09-25 mg/L. INH was dissolved in distilled water and was used as control. A suspension of H37Rv cells was cultured in Middlebrook 7H9 broth supplemented with OADC and 0.05% Tween-80 until an OD 600 of ≈1.0. The culture was diluted to 5× 10 5 bacilli per mL and of 100 µL were added to each well of a 96well microplate together with equal volumes of the complexes. Samples were set up in triplicate. The plates were incubated for 7 days at 37 • C. Resazurin (solubilized in water) was added (30 µL of 0.01%). The fluorescence of the wells was read after 24 h with a Cytation 3 R . The MIC was defined as the lowest concentration resulting in 90% inhibition of growth of the bacterium.

In Vitro Screening Against M. tuberculosis CDC1551
A total of 10 5 bacilli (OD 600 = 0.5) were inoculated into separate tubes containing 1 mL of supplemented Middlebrook 7H9 broth lacking Tween. To these cultures increasing (2-fold) concentrations of test compounds were added and the tubes were left standing at 37 • C for 14 days. The MIC was defined as the lowest concentration failing to produce a visible pellet.

Mammalian Cell Cytotoxicity
A549 cytotoxicity was evaluated using the 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT; Sigma-Aldrich, USA) assay. A549 lung epithelial cells (10 4 ) were seeded into tissue culture plates (TPP, Switzerland) and cultured during 24 h in confluence at 37 • C in a 5% CO 2 atmosphere. The wells were then washed twice with DMEM to remove non-adherent cells and the test compounds were added in different concentrations (ranging from 0.0313 to 512 mg/L), followed by incubation the plates for 48 h under the same conditions mentioned above. Subsequently, the cellular viability was evaluated by adding the MTT reagent to each well and by incubating the plates for 3 h, allowing the viable cells containing active mitochondrial dehydrogenases to metabolize the tetrazolium salt into formazan. The formazan crystals were dissolved with DMSO (100 µL) and the absorbance was measured using a Thermomax Molecular Device microplate reader at 450 nm. In parallel, cytotoxicity was also performed on normal epithelial cells VERO (ATCC CCL-81) as previously described by Pavan et al. (2010). The cells were incubated at 37 • C in a humidified 5% CO 2 atmosphere in flasks with a surface area of 12.50 cm 2 in DMEM medium (10 mL) supplemented with 10% fetal bovine serum, gentamicin sulfate (50 mg/L) and amphotericin B (2 mg/L). The technique consists of collecting the cells using a solution of trypsin/EDTA, centrifugation (2,000 rpm for 5 min), counting the number of cells in a Neubauer chamber and then adjusting the concentration to 3.4 × 10 5 cells/mL in FIGURE 1 | Ligand structures: 1,10-phenanthroline (phen), octanedioic acid (odaH 2 ), 3,6,9-trioxaundecanedioic acid (tddaH 2 ), phthalic acid (phH 2 ), isophthalic acid (isophH 2 ), terephthalic acid (terephH 2 ). DMEM. Then, 200 µL of this suspension was deposited in each well of a 96-well microplate to obtain a concentration of 6.8 × 10 4 cells per well and incubated at 37 • C in an atmosphere of 5% CO 2 for 24 h to allow the cells to attach to the microplate. Dilutions on the test compounds were prepared to obtain concentrations from 500 to 1.95 mg/L. The dilutions were added to the cells after the medium and the non-adherent cells were removed. The cells were then incubated for an additional 24 h. The cytotoxicity of the compounds was determined by adding 30 µL of resazurin and reading on a Synergy H1 (BioTek R ) reader after 6 h of incubation using a microplate and excitation and emission filters at wavelengths of 530 nm and 590 nm, respectively. For both cellular systems, the 50% cytotoxic concentration (CC 50 ) was defined as the compound concentration which caused a 50% reduction in the number of viable cells. In addition, selectivity index (SI) is calculated as follows: CC 50 (mammalian cells)/MIC (M. tuberculosis cells).

FIGURE 2 |
In Vivo Cytotoxicity G. mellonella larvae in the 6th developmental stage were used to determine the in vivo cytotoxicity of the test complexes (Kellett et al., 2011;Desbois and Coote, 2012;McCann et al., 2012). Thirty healthy larvae, each weighing between 0.2 and 0.4 g and with no cuticle discolouration, were used for each experiment. Fresh solutions of the test complexes were prepared immediately prior to testing under sterile conditions. Test compounds (0.05 g) were dissolved in DMSO (1 mL) and added to sterile water (9 mL) to give a stock solution/suspension (5,000 µg/mL). Test solutions/suspensions were prepared from the corresponding stock solution using Millipore water only to dilute to the desired concentration and each compound was screened across the concentration range of 5,000-100 mg/L. Test solutions/suspensions (20 µL) were administered to the larvae by injection directly into the haemocoel through the last proleg. The base of the pro-leg can be opened by applying gentle pressure to the sides of the larvae and this aperture will reseal after removal of the syringe without leaving a scar. Larvae were placed in sterile Petri dishes and incubated at 30 • C for 72 h. The survival of the larvae was monitored every 24 h. Death was assessed by the lack of movement in response to stimulus together with discolouration of the cuticle. Three controls were employed in all assays. The first consisted of untouched larvae maintained at the same temperature as the test larvae. The second was larvae with the pro-leg pierced with an inoculation needle but no solution injected. The third control was larvae that were inoculated with 20 µL of sterile water.  (3), all of the metal complexes showed markedly increased inhibitory activity (up to 10-fold in some instances) against strain CDC1551 relative to H37Rv, and many had an MIC value less than that of INH (0.44 µM). Against CDC1551, {[Mn(phen) 2 (H 2 O) 2 ]} 2 (isoph) 2 (phen)·12H 2 O (8), 1 and [Mn 2 (isoph) 2 (phen) 3 ]·4H 2 O (7) were the most active (MIC range 0.12-0.18 µM), with an almost 3-fold increase in potency compared to INH. The Cu(II) counterparts of 1 and 3, i.e., [Cu 2 (oda)(phen) 4 ](ClO 4 ) 2 ·2.76H 2 O·EtOH (2) and {[Cu(3,6,9tdda)(phen) 2 ]·3H 2 O·EtOH}n (4), were considerably less active against H37Rv, but this disparity was noticeably smaller for the CDC1551 strain, possibly reflecting a degree of specificity by these d 9 metal complexes.

Activity Against Mammalian Vero and A549 Epithelial Cells
In vitro growth inhibitory data (IC 50 values) for the compounds against mammalian VERO and A549 epithelial cells are listed in Table 1 Activity Against G. mellonella Larvae G. mellonella larvae possess an immune system which is analogous to the human innate immune system and are now commonly employed as a convenient, inexpensive, and less ethically sensitive screening model to ascertain the in vivo systemic toxicity of potential drugs, the results of which are comparable to those derived from murine studies (Krishnan et al., 2010;Gandra et al., 2017). Larvae were dosed with varying concentrations of phen and the metal complexes and the percentage of dead larvae was recorded ( Table 2). At the highest administered concentration of 0.1 mg (333 mg/kg) of test compound per larvae, 10% of the larvae treated with the metal complexes survived, whilst all of the larvae injected at this concentration with phen died. The Mn(II) complexes, 1, 3, and 7, and the Cu(II) complex, 2, all showed a marked improvement in survival observed upon decreasing the inoculant concentration to 0.02 mg (67 mg/kg). There were no fatalities when 0.01 mg (33 mg/kg) of the test compounds were dispensed. When assessing the relative toxicity of the test complexes an important consideration is the number of phen ligands each complex contains. From Table 3 it is apparent that the toxicity order, when normalized to the number of phen ligands per complex, is essentially unaltered.

DISCUSSION
Although the two mycobacterial strains, H37Rv and CDC1551, are equally virulent (Manca et al., 1999) the latter strain is known to induce a more rapid and robust host response in a mouse-infected model. The seven Mn(II) complexes were more active against the H37Rv strain than the two Cu(II) samples, but this inequality was less pronounced against the CDC1551 strain, probably due to difference at media culture. Previous studies involving the fungal pathogen, Candida albicans, revealed that the dicarboxylic acids alone were not bioactive (Devereux et al., 2000). This finding, coupled with the fact that the current Mn(II) phen/dicarboxylate complexes are 30-328 times more active against H37Rv than MnCl 2 ·2H 2 O and also show 12-132 times greater activity than phen alone ( . Furthermore, the 3-fold increase in activity against CDC1551 over H37Rv for 1 elevated the SI values to 1017/1347. As the Cu(II) complexes, 2 and 4, were relatively more toxic than the Mn(II) complexes toward the two types of mammalian cells, this severely reduced the SI values of the Cu(II) complexes. The highly cytotoxic nature of 2 and 4 toward A549 cells parallels our previous findings for Cu(II)phen/diacid complexes against cancer cells (Kellett et al., 2011). The isophthalate/phen complexes, [Mn 2 (isoph) 2 (phen) 3 ]·4H 2 O (7) and {[Mn(phen) 2 (H 2 O) 2 ]} 2 (isoph) 2 (phen)·12H 2 O (8), showed highly favorable SI values for CDC1551 when referenced against VERO/A549 cells (>661/1347 and >240/1325, respectively), and this is mainly attributable to their very low MIC 100 values (<0.18, <0.12 µM) against the CDC1551 strain. Dwyer et al. (1969) reported that the Mn(II), Cu(II), Zn(II) and Cd(II) phen dicationic complexes, [M(phen) 2 ](CH 3 CO 2 ) 2 and [M(R-phen) 2 ](CH 3 CO 2 ) 2 , all had similar activities against H37Rv with MIC values ranging from approximately 30 µM for [M(phen) 2 ] 2+ to 0.1 µM for [M(5-NO 2 -phen) 2 ] 2+ . This uniformity in activity between Mn(II) and Cu(II) contrasts somewhat to our findings, which clearly show that in the case of the phen/oda and phen/tdda ligand combinations the Mn(II) complexes were 27-and 18-fold more active, respectively, against H37Rv than their Cu(II) analogs. Against CDC1551, the difference between the Mn(II) phen/oda and Cu(II) complexes was less marked (10-fold), whilst the two phen/tdda samples showed the same activity. It is primarily the high tolerance of the mammalian cells toward the Mn(II) complexes, in contrast to their Cu(II) analogs, that accounts for the remarkably high SI values of the Mn(II) complexes. Also of relevance to the present work is a recent publication (Oliveira et al., 2014)    Whilst relatively small quantities of transition metal ions (primarily Mn, Fe, Co, Ni, Cu, Zn) are essential micronutrients for sustaining microbial growth and homeostasis (Braymer and Giedroc, 2014;Neyrolles et al., 2015), exposure to high concentrations can be devastating as they can bind to and disable important biomolecules and/or promote oxidative stress through Fenton chemistry. It is important to note that Mn(II) and Cu(II) complexes are kinetically labile, meaning that they can rapidly exchange their original ligands (phen and dicarboxylate in the present cases) for other donor ligands present in a biological milieu which includes the bacterium itself. Thus, it is anticipated that a dynamic equilibrium is rapidly established between the original M(II)-phen/dicarboxylate and the newly formed various M(II)-bioligand complexes. phen and metal content of the complexes suggest that it is the complex as a whole, rather than the individual components of the complexes, that are responsible for the observed effects on the larvae.

CONCLUSION
In conclusion, Mn(II) phen/dicarboxylate complexes, which can be synthesized efficiently, utilizing economical and readily available starting materials, offer realistic promise as effective, selective and safe lead candidates in the search for new drugs for the treatment of TB.

AUTHOR CONTRIBUTIONS
PM, MM, MD, KK, CS, PK, AA, TM, AS, DC, and FP conceived and designed the study. PM, MM, AS, and FP analyzed the data and wrote the manuscript. All authors approved of the manuscript.

FUNDING
This work was supported by National Institute of Health (grant numbers R01AI083125 and R01HL106786) to PK and also by Brazilian agencies Fundacao de Amparo à Pesquisa no Estado do Rio de Janeiro (FAPERJ) and Estado de São Paulo (FAPESP) (grant 2013/14957-5), Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). PM would like to acknowledge the funding received through Dublin Institute of Technology's Arnold F. Graves Postdoctoral Fellowship scheme. Programa de Apoio ao Desenvolvimento Científico da Faculdade de Ciências Farmacêuticas da Unesp-PADC.