Synthesis and In Vitro Biological Evaluation of Quinolinyl Pyrimidines Targeting Type II NADH-Dehydrogenase (NDH-2)

Type II NADH dehydrogenase (NDH-2) is an essential component of electron transfer in many microbial pathogens but has remained largely unexplored as a potential drug target. Previously, quinolinyl pyrimidines were shown to inhibit Mycobacterium tuberculosis NDH-2, as well as the growth of the bacteria [ShirudeP. S.; ACS Med. Chem. Lett.2012, 3, 736−74024900541]. Here, we synthesized a number of novel quinolinyl pyrimidines and investigated their properties. In terms of inhibition of the NDH-2 enzymes from M. tuberculosis and Mycobacterium smegmatis, the best compounds were of similar potency to previously reported inhibitors of the same class (half-maximal inhibitory concentration (IC50) values in the low-μM range). However, a number of the compounds had much better activity against Gram-negative pathogens, with minimum inhibitory concentrations (MICs) as low as 2 μg/mL. Multivariate analyses (partial least-squares (PLS) and principle component analysis (PCA)) showed that overall ligand charge was one of the most important factors in determining antibacterial activity, with patterns that varied depending on the particular bacterial species. In some cases (e.g., mycobacteria), there was a clear correlation between the IC50 values and the observed MICs, while in other instances, no such correlation was evident. When tested against a panel of protozoan parasites, the compounds failed to show activity that was not linked to cytotoxicity. Further, a strong correlation between hydrophobicity (estimated as clog P) and cytotoxicity was revealed; more hydrophobic analogues were more cytotoxic. By contrast, antibacterial MIC values and cytotoxicity were not well correlated, suggesting that the quinolinyl pyrimidines can be optimized further as antimicrobial agents.

I nfectious diseases have an enormous impact on the health and welfare of the global population. Recent trends of increasing resistance to existing antimicrobial drugs are therefore alarming, and new drugs are urgently needed to replace older ones as they lose their potency. At the moment, however, not enough compounds are entering development pipelines to ensure that new treatment options will become available in the next decade. In addition, most drugs currently used share similar mechanisms of action, inhibiting bacterial cell wall biosynthesis or affecting protein synthesis on ribosomes, leaving them vulnerable to further issues of resistance. The recent discovery of bedaquiline and other inhibitors targeting bacterial energy-generating systems opened up new avenues for the development of effective treatment strategies. 1 In our work, we make extensive use of a same-target-otherpathogen (STOP) strategy, where studies of homologous enzymes from different pathogens give added value to the information gained and compounds synthesized. Where feasible, structural information is obtained by X-ray crystallog-raphy and used together with biochemical and medicinal chemistry studies to achieve better inhibition of the target enzymes. A priority is to test for action against relevant pathogens very early on, as this is a major stumbling block in finding new antibiotics. An effective enzyme inhibitor is not necessarily active against any given pathogen (because of problems with uptake, degradation, efflux, etc.), so it is essential to broadly profile compounds against multiple pathogens, to find those that are susceptible. As selectivity of action is vital, cytotoxicity and other (physicochemical) properties are also used to help identify potential problems in vivo.
The pathogens covered in this work cause a variety of important diseases. Even now, in the midst of the COVID-19 pandemic, they are still huge killers and, in fact, account for a large proportion of deaths nominally attributed to the coronavirus. 2 Tuberculosis killed about 1.4 million people in 2019. 3 Approximately one-third of the world's population is infected with Mycobacterium tuberculosis, although most of those infected do not develop the active form of the disease. Co-infection by M. tuberculosis and HIV is particularly deadly, and combined with drug resistance, treatment of either infection becomes greatly complicated. Malaria (resulting from infection with one of several Plasmodium species) is also an immense problem, especially in the developing world. There are currently about 229 million new cases each year, and 409 000 deaths, mostly among African children (www.who. int). With increasing drug resistance, and climate change, the developed world is expected to become increasingly affected. Additional pathogens have been designated as ESKAPE bacteria (Enterococcus faecium, Staphylococcus aureus (causing MRSA), Klebsiella species, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter sp.). 4 These largely Gramnegative organisms cause most nosocomial (hospital-acquired) infections, resulting in many deaths (about 25 000 in Europe alone), much suffering, and massive economic loss (about 1.5 billion euros) each year. 5 As the ESKAPE bacteria are rapidly becoming resistant to existing drugs, there is an urgent need for new treatment options.
All of these pathogens have an essential type II NADHdehydrogenase (NDH-2, EC 1.6.5.3), a monotopic membrane protein 6,7 that takes part in electron transport. A representative electron acceptor is menadione (2-methyl-1,4-naphthoquinone); however, the most common electron carriers are menaquinone derivatives, including ubiquinones, which can also be reduced. 8 In some pathogens, NDH-2 is the only dehydrogenase in the respiratory chain, while in others, multiple NADH dehydrogenases are known. In M. tuberculosis, NDH-2, encoded by the ndh gene, is the most favored NADH dehydrogenase of the three enzymes present; it is essential for growth and persistence. 9 Although the specifics vary with the particular pathogen and its metabolic state, compounds that target NDH-2 are of interest as new options for therapy, perhaps in combination with compounds that block other pathways. The absence of mammalian homologues further adds to NDH-2's potential interest.
Phenothiazines, including chlorpromazine and triflupromazine, are known inhibitors of the NDH-2 of M. tuberculosis (MtNDH-2) with in vitro and in vivo activity. 10−12 However, the doses required for antimycobacterial activity are much higher than those eliciting CNS effects, 13 so they are unsuitable for the treatment of tuberculosis. Furthermore, more recent work suggests that phenothiazines do not inhibit membranebound NDH-2 but instead function by disrupting pH gradients across bacterial membranes. 14 NDH-2 is also known to be sensitive to flavones, 12,15 but their low potency (half-maximal ACS Infectious Diseases pubs.acs.org/journal/aidcbc Article inhibitory concentrations (IC 50 s) of the order of 750 μM) does not suggest that these are useful starting points for improved compounds. More recently, polymyxin B was identified as an inhibitor of NDH-2, 16 but this is not its primary mode of action, and such a complex molecule does not offer helpful ideas for the development of small-molecule drugs. A breakthrough was provided when Shirude et al. 17 discovered that quinolinyl pyrimidines such as 1 (Figure 1) are potent inhibitors of MtNDH-2 (IC 50 s as low as 40 nM) with good in vitro antimycobacterial activity (minimum inhibitory concentrations (MICs) as low as 0.8 μM). NDH-2 has also been shown to be targeted by 2-phenylquinolones such as CK-2-68, RYL-552 and RYL-552S, 18 and MTC420 19 (see Figure 1). In the present study, we synthesized 26 novel quinolinyl pyrimidines, which were evaluated against mycobacterial NDH-2 enzymes, as well as a number of bacterial and protozoan pathogens. The strategy for choosing compounds to be synthesized can best be described as exploratory. The primary goal was to determine whether optimization would be feasible, and we therefore made a diverse set of compounds to study the effects on various properties. Tests of cytotoxicity and whole genome sequencing of resistant mutants provided additional information to guide future work.
Knowing from previous reports 17 that the chosen class of compounds was plagued by poor solubility, we replaced the mtolyl group of the reported inhibitor 1 with more hydrophilic groups. Furthermore, we postulated that solubility would be increased by incorporating functionalities that are charged at physiological pH, e.g., amino groups. We saw that amines and alcohols could serve as nucleophiles to substitute the 6-chloro on the pyrimidine ring. To explore the chemical space around the 6-position of the pyrimidine, a diverse set of secondary amines and primary alcohols was selected. Secondary amines were heated with compound 7 and N,Ndiisopropylethylamine using microwave irradiation 21 at 150°C for 90 min, or until all starting material had been consumed, to yield final products 8a−l. Compound 8h was obtained after Boc-deprotection from compound 8g. Alkoxy-substituted final compounds were furnished from compound 7 by heating in neat alcohol with potassium hydroxide. This approach afforded compounds 9a−c as the trifluoroacetic acid (TFA) salts in 13− 63% yield after semipreparative high-performance liquid chromatography (HPLC) purification.
To prepare compounds lacking the 4-quinoline amine group, a different quinoline intermediate was prepared. A microwaveheated Suzuki coupling reaction 22 with 2-chloro-6-nitroquinoline (16) and 4-fluorophenylboronic acid (17) yielded 2-(4fluorophenyl)-6-nitroquinoline (18). After reduction of the nitro group and subsequent substitution of the 6-amino group (19), 6-chloro-N 4 -(2-(4-fluorophenyl)quinolin-6-yl)pyrimidine-2,4-diamine (20) was obtained. Substitution with piperazine produced the final compound (21). To prepare the analogue lacking both amines, compound 19 was reacted with   Biology and SAR. Inhibition of Mycobacterial NDH-2s. Inhibition of two mycobacterial enzymes by quinolinyl pyrimidines is reported in Table 1. One represents MtNDH-2 expressed in Mycobacterium smegmatis, while the other is M. smegmatis NDH-2 (MsNDH-2) expressed in Escherichia coli. Both enzymes contain a C-terminal His-tag. The MsNDH-2 IC 50 s give much the same picture as those for MtNDH-2, despite being on average about 4-fold higher, suggesting that this is a reasonable alternative for ranking compounds when expression in M. smegmatis is not a good option. The two sequences are, however, only ∼80% identical, which could account for some of the differences.
Compound 1 had been reported earlier to inhibit MtNDH-2 with an IC 50 of 96 nM; 17 under our standardized assay conditions, the IC 50 was 26 nM. When the m-tolyl group of 1 was replaced with the more hydrophilic piperazine, morpholine, and 4-methyl-piperazine (8a−c), inhibition was significantly poorer. Substitution of piperazine and piperidine with some bulkier groups (8d−f) also appeared to be unfavorable, although bulky Boc-protected 8g and 8i were relatively good inhibitors. The corresponding deprotected amines (8h and 8j) had IC 50 s approximately 10-fold higher than their Bocprotected counterparts, suggesting that hydrophobic interactions may be required in that part of the structure or that hydrophilic interactions are not desirable. The introduction of a small hydrophilic chain in 8l and 8m gave IC 50 s of ∼10 μM for both enzymes. The additional changes in 8n−p gave only modest improvement of inhibition. When a small alkoxy group was instead introduced on the pyrimidine ring (9a), a lower IC 50 was observed. When the ethoxy group in 9a was replaced with 2-hydroxylethoxy (9b) or (1-methylpiperidine-3-yl)methoxy (9c), inhibition was roughly 3-fold weaker. Exchanging the pyrimidine ring for a pyrimidone (12) did not change potency significantly, which shows that changes in this part of the structure are possible. Removal of the 2pyrimidine amine (15) gave a similar IC 50 value to that of the corresponding compound 8a, suggesting that this functionality is not essential for activity on the enzyme. Additionally, removal of the 4-quinoline amine (21), or both the 4-quinoline and 2-pyrimidine amines (23), resulted in activity on the MsNDH-2 enzyme equivalent to that of the corresponding compound 8a (MtNDH-2 was not tested in this case).
A number of X-ray structures have been solved for NDH-2like enzymes in complex with different kinds of inhibitors (summarized in Petri et al. 23 ); none involves a quinolinyl pyrimidine. The most similar protein is the enzyme from Caldalkalibacillus thermarum, 23 with 31% amino acid sequence identity to MtNDH-2, bound to 2-heptyl-4-hydroxyquinoline-N-oxide ( Figure 1). Docking of the present compounds based on this structure is problematic because the inhibitors are not very similar, and the sequence identity is not very high, particularly in the C-terminal regions relevant for binding. However, if the mode of docking the quinolinyl pyrimidines into MtNDH-2 proposed by Petri et al. is correct, a C-terminal hydrophobic groove would be interacting with the phenyl pyrimidine group of our present compounds. The essential primary amine on the pyrimidine 17 would then be interacting with some MtNDH-2 group within this groove. The 4fluorophenyl group would be placed in a hydrophobic pocket near the fused-ring system of the flavin adenine dinucleotide (FAD); the essential primary amino group of the quinoline 17 could, for instance, form a hydrogen bond with the carbonyl group at C4 of the flavin. Clearly, more X-ray structures will be required to understand inhibitor binding in the various enzymes.
MIC Evaluation on ESKAPE Bacteria and M. tuberculosis. The new compounds were evaluated for antibacterial activity against a panel of ESKAPE bacteria, as well as M. tuberculosis (using a resazurin reduction microplate assay (REMA) 24 ), see Table 1. The quinolinyl pyrimidine inhibitor 1, reported earlier as an inhibitor of M. tuberculosis growth, was essentially inactive on Gram-negative bacteria with the exception of the efflux-defective (ΔtolC) and drug-hypersensitive (lpxC) E. coli strains. However, the MIC for this compound on the mycobacterium was found to be in line with the value previously reported. 17 Although the IC 50 s for MtNDH-2 generally correlate well with MICs on M. tuberculosis (see Figure 2), a few exceptions were noted (8a, 8j, 9c, and 15). Compounds 8a, 8d, and 9c have very similar IC 50 s (approx. 1.5 μg/mL), but widely varying MICs. The same is true for 8c, 8j, and 15. Note that 15 is an analogue of 8a, as indeed are 21 and 23. In terms of MIC, removing both exocyclic amines as in 23 was preferable; in terms of IC 50 , the reverse was the case. The differences in MIC could arise from variations in the ability to penetrate the notably thick and impermeable mycobacterial cell wall.
To gain a better understanding of how structural changes modulate the MIC values in general, and the relative MIC values for various strains, a partial least-squares (PLS) analysis was performed including also a few standard molecular descriptors (see the Methods for details). This analysis revealed that the overall ligand charge is one of the most important factors affecting the MICs. Figure 3 shows a principal component analysis based on MIC data only, highlighting how the overall charge affects the antibacterial patterns. Thus, good MICs on wild-type P. aeruginosa are dependent on the presence of a charged group on the pyrimidine (e.g., 8a, 8h,and 8j) whereas MICs on A. baumannii and M. tuberculosis are improved by neutral groups. Moreover, replacing the m-tolyl group of 1 with both smaller (−Cl and −OEt, 7 and 9a, respectively) and larger (Boc-protected amines, e.g., 8g and 8i) noncharged substituents was linked to retained activity against M. tuberculosis and A. baumannii (see Table 1).
Interestingly, compounds containing a primary (8h and 8j) or secondary amine (8a) inhibited both P. aeruginosa wild-type and mutated strains at a similar level. By contrast, when a less basic tertiary amine (e.g., 8c−e) or a neutral substituent (e.g., 8b and 9b) is introduced, the P. aeruginosa wild-type activity is lost (MIC ≥ 64 μg/mL). This suggests that a positive charge in this part of the structure is essential for activity on wild-type P. aeruginosa, although the role of this amine is as yet unclear.
For some of the strains tested (E. coli D22, A. baumannii, S. aureus, and M. tuberculosis) correlations between pIC 50 and pMIC values (r between 0.48 and 0.69) were observed. For the bacteria least susceptible to the tested class of compounds (E. coli, P. aeruginosa, and Klebsiella pneumoniae wild types), such correlations were not seen (r between −0.30 and 0.03). The tested compounds generally show little inhibition of wild-type E. coli growth, although several did inhibit the ΔtolC and lpxC strains. Together, these observations suggest that this class of compounds needs optimization in terms of efflux and permeability.
Evaluation of Cytotoxicity. Cytotoxicity was assessed against MRC-5 SV2 (human lung fibroblast) cells as well as a human hepatoblastoma cell line, HepG2 (Table 2). Where both cell lines were tested, the trends in toxicity correlate well; although MRC-5 IC 50 s were always higher than HepG2 TD 50 s, compounds that were toxic to HepG2 cells were also toxic to MCR-5 cells. Compound 9a stands alone as an example where toxicity against MRC-5 cells is greater. In general, the cytotoxicity of the various compounds was striking, and furthermore, inhibitors were relatively cytotoxic at concentrations similar to the corresponding IC 50 values. However, there was no clear correlation between high cytotoxicity and the IC 50 values, suggesting that these two properties could be optimized independently.
Conversely, the SAR reveals that there is a strong positive correlation between hydrophobicity (estimated as clog P) and cytotoxicity ( Figure 4). This trend is particularly clear for the more hydrophobic Boc-protected 8g and 8i, which have considerably lower CC 50 on MRC-5 cells than the deprotected amines 8h and 8j. Furthermore, comparing 9a with 9b (ethyl vs hydroxyethyl), the 1 log reduction of clog P results in more than 5-fold reduction of cytotoxicity. Similarly, the compounds with the lowest clog Ps, 6 and 12, have the lowest toxicity. The less-substituted pyrimidone 12 stands out as an acceptable inhibitor of MtNDH-2 with reasonable activity against M. tuberculosis, although its effects against the ESKAPE pathogens are more modest.
We also sought to determine whether the anilinic amines of the quinoline and pyrimidine scaffolds were giving rise to cytotoxicity. By comparing the substitution patterns of 8a, 15, 21, and 23, and considering the clog P values, it can be concluded that the pyrimidine and quinoline NH 2 groups are not the key drivers for cytotoxicity. By means of PLS, a model for the cytotoxicity was generated, indicating a positive correlation between cytotoxicity and molecular weight (MW), and a negative correlation to descriptors contributing to the decrease of the hydrophobicity (see Figure S1).
Evaluation on Parasites. Comparison of the cytotoxicity results and the effects on in vitro growth of the multidrugresistant P. falciparum K1 strain, Trypanosoma brucei brucei Squib 427, Trypanosoma cruzi Tulahuen LacZ (clone C4), and Leishmania infantum MHOM/MA(BE)/67 strains (Table S1) suggested only nonspecific parasite growth inhibition (maximum concentration tested was 64 μM).
Selection and Sequence Identification of Resistant Mutants of ESKAPE Pathogens. Resistant mutants were raised against 8a, 8g, 8j, and 9a. The results obtained when these were analyzed by whole-genome sequencing are summarized in Table 3. The best of the new NDH-2 inhibitors, 8g, does not give rise to mutations in the ndh gene that codes for NDH-2, instead producing primarily mutations in ackA (coding for acetate kinase). The links between this enzyme and NDH-2 are not clear. The poorer inhibitors 8a and 8j give rise to mutations in members of a two-component signaling system, pmrA (basR) and pmrB (basS), as was reported earlier for mutations that cause resistance to polymyxin B, a known NDH-2 inhibitor that also has several other biological effects. 25 The moderate inhibitor 9a gave rise to yet another pattern of mutations. It is known that polymyxin B has a number of different effects on Gram-negative cells, including disrupting the outer membrane. More work will be required to determine whether quinolinyl pyrimidines have other effects, in addition to inhibiting NDH-2.

■ CONCLUSIONS
Twenty-six novel quinolinyl pyrimidines were designed, synthesized, and evaluated as NDH-2 inhibitors, as well as for their effects on bacteria and parasitic protozoa, and cytotoxicity. Most of the compounds were cytotoxic. Those that were not were also relatively poor inhibitors of NDH-2. However, it is worth stressing that cytotoxicity and inhibition are not correlated, and so the properties can be optimized separately. We conclude that, although the enzyme can probably be targeted successfully, this particular series of compounds may not offer the best prospects for systemic drug treatment. However, similar problems for polymyxin B have meant that its primary approved use in Europe is as a highly effective topical drug, and this avenue could be explored for the quinolinyl pyrimidines, as well.     (6) (90 mg, 0.36 mmol) was dissolved in N-methyl-2-pyrrolidinone (1 mL) and added to the above reaction followed by the addition of HCl (4 M in dioxane, 0.36 mL, 1.42 mmol). The mixture was heated at 80°C for 10 h where upon a solid was formed. The reaction mixture was allowed to cool, and then methanol (10 mL) was added. The resulting solid was collected by filtration and washed with methanol. The crude was purified by preparative HPLC (MeCN/H 2 O (0.1% TFA)). All pure fractions were pooled, pH was set to basic, and the product was extracted with ethyl acetate. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated to yield a yellow solid (60 mg, 39%). 1  4-Chloro-2-(4-fluorophenyl)-6-nitroquinoline (4). 2-Amino-5-nitro-benzoic acid (1.00 g, 5.49 mmol) was suspended in phosphorous oxychloride (3.0 mL, 33 mol) at room temperature. 4-Fluoroacetophenone (0.670 mL, 5.49 mmol) was added dropwise, and the resulting mixture was heated at 90°C for 5 h. After completion of the reaction, most of the POCl 3 was evaporated under reduced pressure and the resulting residue was then poured into a mixture of ice, 25% ammonium hydroxide, and chloroform (10:4:4 w/v). The reaction mass was stirred at room temperature overnight. The chloroform layer was then separated and washed with brine, dried over Na 2 SO 4 , filtered, and evaporated under vacuum to give the crude product as a yellow powder. The crude product 2-(4-Fluorophenyl)quinoline-4,6-diamine (6). Sodium azide (2.15 g, 33.0 mmol) was added in one step to a solution of 4-chloro-2-(4-fluorophenyl)-6-nitroquinoline (4) (1.00 g, 3.30 mmol) in N-methylpyrrolidone (10 mL) in a roundbottom flask fitted with a condenser. The resulting suspension mixture was stirred and warmed to 60°C for 18 h. At the end of this time, the mixture was cooled to room temperature and poured into a mixture of cold water and 50 mL of ethyl acetate. pH was adjusted to 9 with ammonia (aq), and the layers were separated. The aqueous layer was extracted with ethyl acetate (2 × 50 mL) and the combined organic layers were washed with brine and dried over MgSO 4 and concentrated under reduced pressure to give dark semiviscous crude product (solidified over 2 h at room temp). The crude product was used in the next step without further purification. Crude 6nitro-4-azido-2-(3-fluorophenyl)-quinoline (5) (1.00 g, 3.23 mmol) obtained from the previous step was dissolved in ethyl acetate (50 mL) and ethanol (25 mL). The stirred mixture was heated to reflux, and SnCl 2 ·H 2 O (4.38 g, 19.4 mmol) was cautiously added in portions over 10 min. The reaction mixture was heated for an additional 2 h and then cooled to room temperature. Water (100 mL) was added to the reaction mixture, and pH was adjusted cautiously to 9, after which the solution was filtered. Solids were washed with a minimum amount of ethyl acetate. The filtrates were combined, and the aqueous layer was extracted with ethyl acetate. The combined organic layers were washed with brine, dried over MgSO 4 , and filtered before evaporation to give the crude product as a black thick mass. The crude product was chromatographed on a silica gel column, using 10% methanol in ethyl acetate as an eluent. Column purified compound was triturated with petroleum ether to give 4,6-diaminoquinoline (6) as a dark tan powder (826 mg, 75%). 1  N 6 -(2-Amino-6-chloropyrimidin-4-yl)-2-(4-fluorophenyl)quinoline-4,6-diamine (7). Compound 6 (0.62 g, 2.45 mmol) was suspended in N-methylpyrrolidone (8 mL) and 2-amino-4,6-dichloropyrimidine was added to the above dark solution in one portion followed by the addition of HCl in dioxane (3.5 mL, 4.0 M solution, 14 mmol). The reaction mixture was heated in a heating block at 100°C overnight. After this time, LCMS showed full conversion of starting material. The reaction mixture was made basic with saturated NaHCO 3 and extracted with ethyl acetate (3 × 30 mL). The combined organic layers were washed with water and finally with brine. The organic layer was dried over Na 2 SO 4 , filtered, and concentrated to obtain a crude product. The crude material was purified by silica gel chromatography using a gradient elution with (0.5−5%) methanol in ethyl acetate to obtain the pure product as a tan solid (960 mg, 76%). 1  Method A: General Procedure for the Synthesis of Compounds 8a−p. Amine (0.27 mmol, 2.5 equiv) and N,Ndiisopropylethylamine (0.21 mmol, 1.9 equiv) were added to a solution of pyrimidine chloride (7) (40 mg, 0.11 mmol, 1.0 equiv) in absolute ethanol (0.5 mL). The reaction was heated under microwave irradiation at 150°C for 90 min, or until all starting material had been consumed. Ethanol was removed under reduced pressure, and the crude material was purified by preparative HPLC (MeCN/H 2 O (0.1% TFA)). The combined pure fractions were freeze-dried to yield the pure compound as the TFA-salts unless otherwise stated. N 6 -(2-Amino-6-(piperazin-1-yl)pyrimidin-4-yl)-2-(4fluorophenyl)quinoline-4,6-diamine (8a). Prepared from piperazine (27 mg, 0.32 mmol) and 7 (48 mg, 0.13 mmol) using method A. The crude product was purified by preparative HPLC (MeCN/H 2 O (0.1% TFA)). All pure fractions were pooled and pH was set to basic, and the product was extracted with ethyl acetate. The combined organic layers were dried over Na 2 SO 4 , filtered, and concentrated to yield 8a as a yellow solid (23 mg, 42%). 1   tert-Butyl ((1-(2-Amino-6-((4-amino-2-(4-fluorophenyl)quinolin-6-yl)amino)pyrimidin-4-yl)piperidin-4-yl)methyl)carbamate (8g). Prepared from tert-butyl (piperidin-4ylmethyl)carbamate (45 mg, 0.21 mmol) and 7 (40 mg, 0.11 mmol) using method A to yield 8g as a yellow solid (47 mg, 57%). 1     (1-(2-Amino-6-((4-amino-2-(4-fluorophenyl)quinolin-6yl)amino)pyrimidin-4-yl)piperidin-4-yl)methanol (8n). Prepared from 4-piperidylmethanol (24.2 mg, 0.210 mmol) and 7 (40 mg, 0.105 mmol) using method A and heating for 3.0 h to yield 8n as a yellow solid (40 mg, 55%). 1 Compound 7 (40 mg, 0.11 mmol) and KOH (30 mg, 0.54 mmol) were suspended in absolute ethanol (0.5 mL). The reaction was heated under microwave irradiation at 150°C for 30 min. Ethanol was removed under reduced pressure and the crude was purified by preparative HPLC. The combined pure fractions were freezedried to yield the pure compound 9a as a yellow solid (8 mg, 13%). 1  2-Amino-6-chloropyrimidin-4(3H)-one (11). Prepared following a literature procedure 26 to yield 11 as a white solid (178 mg, 90%). 1 H NMR (DMSO-d 6 ) δ 7.13 (br s, 2H), 5.58 (s, 1H).
NDH-2 oxidizes NADH in the presence of hydrogen acceptor quinones, e.g., menadione, allowing the reaction to be followed spectrophotometrically as a decrease in absorbance at 340 nm. The linear changes observed with an enzyme concentration of 1.25 nM for MtNDH-2 and 50 nM for MsNDH-2, corresponding to substrate reduction, were monitored at 22°C for 0.5 and 2 h, respectively. For inhibition tests, the enzyme was preincubated with the compound for 10 min at 22°C, after which the reaction was started by adding 50 μM menadione. Reaction rates were measured as a function of inhibitor concentration (highest concentration 20 or 100 μM, depending on compound solubility), and half-maximal inhibitory concentration (IC 50 ) values were determined by a nonlinear regression analysis of the sigmoidal dose−response curves in GraphPad Prism (GraphPad Software, Inc., La Jolla, CA). Polymyxin B was used as a positive control in the assay. The measured IC 50 s, 2.5 μg/mL for MsNDH-2 and 0.41 μg/mL for MtNDH-2, were similar to the value published earlier, 1.6 μg/mL for MsNDH-2. 16 Evaluation of Minimum Inhibitory Concentration (MIC) on ESKAPE Pathogens. Compound prepared in MHII medium was dispensed into a 96-well round-bottom microtiter plate to give final assay concentrations from 128 μg/mL down to 0.25 μg/mL (twofold dilution series in 10 wells, plus two control wells: medium control with no bacteria or compound, and growth control with bacteria added but no compound). Bacteria prepared from fresh colonies (grown on nonselective agar, incubation 18−24 h at 35 ± 2°C) were suspended in saline to 0.5 McFarland (≅1.5 × 10 8 CFU/mL). This bacterial suspension (50 μL) was transferred to 10 mL of MHII broth to give a final bacterial concentration ≅5 × 10 5 CFU/mL (acceptable range (3−7 × 10 5 ) CFU/mL). The 50 μL bacterial suspension was pipetted into each well (except medium control well, where 50 μL of MHII was pipetted). The final volume in each well was 100 μL. Plates were covered and incubated without shaking for 16−20 h at 35 ± 2°C. MIC was read visually, as complete inhibition of growth by the unaided eye, using the medium-only wells as the control.
Evaluation of Minimum Inhibitory Concentration (MIC) on M. tuberculosis by Resazurin Reduction Microplate Assay (REMA). Twofold serial dilutions of each test compound were prepared in 96-well plates. Frozen aliquots of replicating tubercle bacilli (reference strain H37Rv) were thawed and diluted to an optical density at 600 nm (OD600) of 0.0001 (3 × 10 4 cells/mL) and added to the plates to obtain a total volume of 100 μL. Plates were incubated for 6 days at 37°C before the addition of resazurin (0.025% [w/v] to 1/10 of well volume). After overnight incubation, the fluorescence of the resazurin metabolite, resorufin, was determined (by excitation at 560 nm and emission at 590 nm, measured using a TECAN infinite M200 microplate reader). The MIC was defined visually as the last concentration preventing resazurin turnover from blue to pink and confirmed by the level of fluorescence measured by the microplate reader. MIC was determined using GraphPad Prism version 7.0 software (GraphPad Software, Inc., La Jolla, CA). The experiment was performed twice, and all of the compounds were tested in duplicate. Rifampicin (MIC = 0.0003 μM) and bedaquiline (MIC = 0.37 μM) were used as reference antitubercular agents.
Evaluation of the Cytotoxicity of the HepG2 Cell Line. HepG2 cells were cultured in FBS-free medium and were then seeded into 96-well plates in 200 μL of complete culture medium at a final concentration of 5 × 10 4 cells/well and treated by rifampicin, or by compounds (at a range of concentrations) for 72 h. Redox status was estimated using resazurin reduction (0.025% [w/v] to 1/10 of well volume). The resazurin assay was performed with a fluorometric method according to the procedure described previously for REMA assay. Three hours before the end of incubation, 10 μL of resazurin/well were added, yielding a final concentration of 10% resazurin. Plates were returned to the incubator, and the fluorescence was read after 6 h. The plates were exposed to an excitation wavelength of 535 nm, and emission at 580 nm was recorded on a TECAN infinite M200 microplate reader. The percent viability was expressed as fluorescence emitted by treated cells compared to control (medium or vehicle only). Compounds were run once in the primary screen; repeated evaluation is typically performed only if relevant potency and selectivity are shown. For these compounds, retesting was not done in view of the nonspecificity, and so error estimates cannot be provided.
Evaluation of Cytotoxicity on the MRC-5 Cell Line. MRC-5 SV2 cells were cultured in Earl's MEM + 5% FCSi. Assays were performed in 96-well microtiter plates, each well containing about 10 4 cells/well. After 3 days of incubation with or without compounds, cell viability was assessed fluorimetrically after the addition of resazurin (λ ex 550 nm, λ em 590 nm). The results were expressed as % reduction in cell growth/viability compared to untreated control wells and CC 50 was determined. Testing/retesting strategy was as described above for HepG2 cells.
Resistant Mutants and Whole Genome Sequencing. Mutants of E. coli ΔtolC resistant to compounds 8a and 8g, wild-type P. aeruginosa resistant to compound 8j, and A. baumannii resistant to compound 9a were selected by serial passage of independent lineages in Luria broth (LB) at increasing concentrations of compound (up to 4× initial MIC). At the end-point of selection, a single clone was isolated from each resistant culture for sequence analysis, by streaking for single colonies on LB agar and measuring MIC against the selective compound. Mutations were identified by wholegenome sequencing of independently selected mutants with increased MIC values. Genomic DNA for whole-genome sequencing was prepared using the MasterPure DNA Purification Kit (Epicentre, Illumina, Inc., Madison, Wisconsin). Final DNA was resuspended in EB buffer. Genomic DNA concentrations were measured in a Qubit 2.0 Fluorometer (Invitrogen via Thermo Fisher Scientific). DNA was diluted to 0.2 ng/mL in water (Sigma-Aldrich, Sweden), and the samples were prepared for whole genome sequencing according to the Nextera XT DNA Library Preparation Guide (Illumina, Inc., Madison, Wisconsin). After the polymerase chain reaction (PCR) clean-up step, the samples were validated for DNA fragment size distribution using the Agilent High Sensitivity D1000 ScreenTape System (Agilent Technologies, Santa Clara, California). Sequencing was performed using a MiSeq desktop sequencer, according to the manufacturer's instructions (Illumina, Inc., Madison, Wisconsin). The sequencing data were aligned and analyzed in CLC Genomics Workbench version 8.0.3 (CLCbio, Qiagen, Denmark).
Multivariate Modeling. The multivariate modeling was performed in Simca, version 13.0.0.0, Umetrics AB (www. umetrics.com). The analyses (principle component analysis (PCA) and PLS) were made using negative logarithms of the MIC, TD99, TD50, and CC50 values; all variables were scaled to unit variance and centered. As MIC values in Table 1 were greater than the maximum concentration tested, twice this value is used in the analyses, to provide more guidance for the PCA/PLS analysis. The following molecular descriptors were used for the PLS modeling: molecular weight (MW), polar surface area (PSA), octanol/water partition coefficient (log P), number of hydrogen-bond donors and acceptors of the pyrimidine substituent (#HBD and #HBA), and the total charge of the compound, assuming that all aliphatic amines are protonated as well as all 4-amino quinolones (charge). The PSA and log P values were calculated using Instant JChem, version 17.1.9.0. ■ ASSOCIATED CONTENT * sı Supporting Information