Stable pantothenamide bioisosteres: novel antibiotics for Gram-positive bacteria

The emergence of multidrug resistant bacteria has prioritized the development of new antibiotics. N-substituted pantothenamides, analogs of the natural compound pantetheine, were reported to target bacterial coenzyme A biosynthesis, but these compounds have never reached the clinic due to their instability in biological fluids. Plasma-stable pantothenamide analogs could overcome these issues. We first synthesized a number of bioisosteres of the prototypic pantothenamide N7-Pan. A compound with an inverted amide bond (CXP18.6-012) was found to provide plasma-stability with minimal loss of activity compared to the parent compound N7-Pan. Next, we synthesized inverted pantothenamides with a large variety of side chains. Among these we identified a number of novel stable inverted pantothenamides with selective activity against Gram-positive bacteria such as staphylococci and streptococci, at low micromolar concentrations. These data provide future direction for the development of pantothenamides with clinical potential.


Introduction
There is an ongoing need for new drug scaffolds and targets to combat the increasing development of resistant bacterial strains. Methicillin-resistant and vancomycin-resistant Staphylococcus aureus (S. aureus) strains (MRSA and VRSA) have led to serious health problems in both hospital settings and the community, but resistance to antibiotics is certainly not limited to these examples. Pantothenate or pantothenic acid (vitamin B5) is required for coenzyme A (CoA) biosynthesis and is an essential and rate limiting nutrient for survival and/or growth of numerous bacteria, fungi and protozoa. A range of compounds analogous to pantothenate or to its cysteamine conjugate pantetheine (see Fig. 1a) have been reported that possess activity against bacteria, fungi and malaria parasites (for a review see ref. [1]). In 1970, N-substituted pantothenamides (in short: pantothenamides), were first reported to possess antibacterial activity in vitro (see Fig. 1b for the general pantothenamide structure) [2]. Pantothenamides are pantetheine analogs, of which the prototypic members N-pentylpantothenamide (N5-Pan) and N-heptylpantothenamide (N7-Pan) were found to be active against Gram-negative and Gram-positive bacteria [3]. During the last few decades, a number of additional pantothenamides with antibacterial activity have been synthesized [4][5][6][7][8] and their putative modes of action have been studied in detail [7][8][9][10][11]. Pantothenamides have been shown to serve as substrates or inhibitors (either competitive or allosteric) of pantothenate kinase (PanK), the first enzyme in the CoA biosynthesis pathway [4,5,7,8]. As a consequence, PanKcatalyzed pantothenate phosphorylation is partially or completely inhibited. In the case that pantothenamides serve as PanK substrates, competing with the natural substrate pantothenic acid, the resulting 4′-phosphopantothenamides may be further metabolized by the CoA biosynthetic machinery to yield analogs of CoA, as was shown for Escherichia coli (E. coli) [9,10]. Such CoA analogs were found to be incorporated in acyl carrier protein, thereby inhibiting its function in bacterial fatty acid biosynthesis, which requires the 4′-phosphopantetheine moiety of CoA to be active [12]. Whether the mechanisms that ultimately result in antimicrobial activity in the various target organisms (bacteria, fungi, protozoa) are the result of inhibition of CoA biosynthesis, fatty acid biosynthesis or another CoA-utilizing process, or a combination of the above, remains to be resolved for most organisms. The mechanism of action in S. aureus, however, has recently been investigated in detail indicating that PanK is the target enzyme [7].
Despite their potential use and selectivity for bacterial, fungal, and/or protozoan metabolic routes, no pantothenamide compound has ever made it to the clinic. It was recently shown that pantothenamides are not active as antimicrobials in the presence of serum, and that they were hydrolyzed by ubiquitous pantetheinases of the vanin family (see Fig. 1b) [13][14][15][16]. To address this, a series of vanin inhibitors based on a pantothenate scaffold were synthesized, which proved to inhibit serum vanin activity in the nanomolar range [17]. Combinations of these novel vanin inhibitors and prototypic pantothenamides like N5-Pan and N7-Pan exert antimicrobial activity in vitro, particularly against Gram-positive bacteria (S. aureus, Staphylococcus epidermidis (S. epidermidis), Streptococcus pneumonia (S. pneumoniae), and Streptococcus pyogenes (S. pyogenes)) even in the presence of serum [18]. These results indicate that pantothenamides, when protected against degradation by host vanins, are potentially useful antimicrobial agents [15,16].
As combinations of compounds are less desirable from a drug development perspective, chemical modification of pantothenamides to render them more stable would be preferable. Taking this approach, modified pantothenamides were generated that were active antimalarials or antibacterials in the presence of serum [19][20][21][22][23]. As the hydrolysis-susceptible amide bond was still present in these modifications or the potency was reduced significantly, we aimed to generate bioisosteric pantothenamides that would mimic the chemical space of the amide bond, but would lack the vanin sensitivity, without losing their potency. Here we describe such stable bioisosteric compounds as a starting point for lead optimization.

Chemistry of inverted pantothenamides
General procedure A To a solution of carboxylic acid B (0.5 mmol) in MeCN/ H 2 O (30:1, 4.3 ml) were added HOBt (0.6 mmol), NaHCO 3 (0.6 mmol), EDCI (0.6 mmol), and a solution of amine A (for the synthesis see ref 2, 0.6 mmol) in MeCN/H 2 O (0.7 ml). The progress of the reaction was monitored using LC-MS and upon completion, the reaction was quenched by the addition of saturated aqueous NH 4 Cl solution (15 ml) and the mixture was extracted twice using EtOAc (15 ml). The combined organic layers were dried over Na 2 SO 4 and filtered before concentration under reduced pressure. The residue was purified by flash column chromatography (DCM/MeOH = 98:2 → 80:20) to afford the product.
To a solution of amide D (0.15 g, 0.42 mmol) in MeCN (2.1 ml) was added an aqueous 0.2 M solution of HCl (2.12 ml, 0.42 mmol). The mixture was stirred at room temperature for 2 h. The mixture was added dropwise to a mixture of saturated aqueous NaHCO 3 (2 ml) and EtOAc (10 ml). The layers were separated and the aqueous phase was saturated with NaCl and then extracted with EtOAc (10 ml). The combined organic layers were dried over Na 2 SO 4 and filtered before concentration under reduced pressure to afford CXP14.18-037 (113 mg, 85%) as a pale yellow oil. 1
All strains, except mycobacteria, were grown overnight on Columbian blood agar plates (Thermo Scientific) at 37°C. Slow growing mycobacteria (SGM, M. avium and M. kansasii) were grown for 5-7 days in Middlebrook 7H9 Broth supplemented with 10% oleic acid, albumin, dextrose, and catalase (OADC) at 37°C. Rapid growing mycobacteria (RGM, M. abscessus) were grown for 3 days in Middlebrook 7H9 Broth at 30°C. S. pyogenes strains were incubated at 5% CO 2 , while all other strains grow in normal atmosphere. Liquid cultures of S. aureus, S. epidermidis, E. coli, and P.aeruginosa were grown in Mueller-Hinton Broth (BD Difco) at 37°C while shaking and S. pyogenes were grown statically in 5% CO 2 at 37°C in Todd Hewitt Broth (BD Bacto).
To test the MIC of compounds on S. aureus, S. epidermidis, S. pyogenes, E. coli, and P. aeruginosa overnight cultures were diluted 1:1000 in fresh media and 100 µl was added to 100 µl of serial diluted compound in 96-well plates. Plates were incubated at 37°C (S. pyogenes in 5% CO 2 , others in normal atmosphere) for 16 h and MICs were determined optically. The MIC was defined as the first well where no bacterial growth was observed. To test the MIC of compounds on M. avium, M. kansasii, and M. abscessus, a 0.5 McFarland suspension was made of the cultured strains during log phase growth. This suspension was then diluted to obtain an inoculum of 1 × 10 5 -1 × 10 6 CFU ml −1 in Cation Adjusted Mueller-Hinton broth (with 20% OADC for SGM) and 50 µl was added to 50 µl of serial diluted compound in 96-well plates. Plates were incubated at 37°C for 7 days for SGM and at 30°C for 3 days for RGM. MICs were determined optically. The MIC was defined as the first well where no bacterial growth was observed. The results are the median values of at least three independent assays.
Antimicrobial activity of N7-Pan and CXP18.6-012 in the presence of serum was performed as follows: to mimic human serum as much as possible, the highest concentration of serum (10%) was used that did not interfere with the optical readout. Complement inactivation of human serum (30 min at 56°C) was necessary to prevent antimicrobial activity of the complement system. Because heat inactivation of serum also led to decreased vanin activity, we supplemented the serum with fetal bovine serum (FBS) in a ratio of 4:1. The vanin activity in FBS is extremely high compared to human serum, leading to a total vanin activity in the assay that resembles full human serum. Furthermore, the standard MIC assay was performed as described above.

LC-MS analysis
To examine the stability of analogs in vitro, N5-Pan, its bioisostere CXP18.6-013, N7-Pan and its bioisostere CXP18.6-12 were incubated for 16 h at room temperature in 500 µM potassium buffer with and without 10% fetal bovine serum as a source of pantetheinase activity. Samples were taken and analyzed by LC-MS using a Shimadzu LC10ATvp HPLC coupled to a Shimadzu LCMS2010A mass spectrometer. The elution of the following molecules were shown: N5-Pan and CXP18

Cytotoxicity assay
Primary human keratinocytes obtained from biopsies of healthy volunteers were cultured following the Rheinwald-Green system at 37°C and 5% CO 2 [24]. Cells were grown until they reached confluency. Compounds were added to these cultures at a concentration of 100 µM and effect on growth and toxicity was determined microscopically after 24 h. Cytotoxicity was detected using the LDH cytotoxicity detection kit according to the manufacturer's protocol (Roche Applied Science, IN).

Synthesis of N7-pan bioisosteres
We investigated the structure-activity relationship of pantothenamide modifications by varying the N7-Pan molecule on three different positions, as indicated in Fig. 1b: the C2 linker between the two amides, the second amide bond and the right hand part of the molecule (further referred to as side chain). Table 1 lists the structural modifications in the C2 linker and amide bond of N7-Pan, and their effects on the minimal inhibitory concentration (MIC) values. It turned out that only a C2 linker was allowed. Substitution by an alkene or aromatic moiety was not allowed. The amide bond was probed by replacing with different bioisosteres, whilst keeping the C2 linker intact. Replacement by a keto, ester or a sulfonamide moiety was not allowed. Replacement by 5-membered ring amide bioisosteres was in general not successful, although in some cases minor activity was demonstrated. The only effective modification which yielded activities comparable to the parent compound was substitution of the amide bond by an inverted amide as shown in compound CXP18.6-012. Like N7-Pan, none of its bioisosteres showed activity against the Gram-negative bacteria that were tested (E. coli and P. aeruginosa) or the Mycobacterium species that were tested (M. avium, M. abscessus and M. kansasii). Some bioisosteres showed weak activities against S. epidermidis and S. pyogenes (MIC between 8 and 32 μg ml −1 ). Only bioisostere CXP18.6-012 showed activity against S. aureus, with a MIC value of 2 μg ml −1 , which is close to the potency of the parent compound N7-Pan. However, this modification decreased the potency towards S. epidermidis from 0.5 to 8 μg ml −1 , whereas the sensitivity towards S. pyogenes was decreased from 2 to 32 μg ml −1 . In addition, we synthesized the inverted amide bioisostere of another prototypical pantothenamide, N5-Pan, designated as CXP18.6-013 (see also Supplemental Table 1). This compound showed good activity towards S.pyogenes (2 μg ml −1 ) and weak activity towards S. epidermidis and S. aureus (resp. 16 and 32 μg ml −1 ). The experimental details on the synthesis of the compounds of Table 1 are given in online Supplemental File S1, with the exception of CXP18.6-012 which can be found in the Materials and Methods section, and N7-Pan, which has been described before [2].

Stability of inverted pantothenamides
We incubated the prototypical pantothenamides N5-Pan and N7-Pan as well as their inverted amide bioisosteres CXP18.6-012 and CXP18.6-013 in the presence and absence of fetal bovine serum, and analyzed the stability towards vanin-mediated hydrolysis, using LC-MS analysis. In Fig. 2a, c total degradation of N5-Pan and N7-Pan by serum is shown as the disappearance of the red peak of the parent compound, whereas the hydrolysis product pantothenate emerges (blue peak). Figures 2b, d show that the inverted pantothenamides CXP18.6-013 and CXP18.6-012 remain stable after overnight incubation in a serumcontaining buffer.

Activity of inverted pantothenamides under physiological conditions
We then investigated whether inverted pantothenamides would retain their ability to kill bacteria, even in the presence of serum. To test this, we performed MIC assays in Mueller-Hinton medium to which 10% serum was added. We tested both human serum and fetal bovine serum (FBS). Human serum was first decomplemented at 56°C, which also partly destroys its vanin activity. FBS, which has an extremely high vanin activity, was used without heat inactivation, which allowed testing of the compounds in the presence of a vanin activity which equals that of 100% human serum. Table 2 shows that addition of serum abolished the antibiotic properties of N7-Pan, whereas CXP18.6-012 remained active.

In vitro toxicology
Finally, we used cultured primary human keratinocytes to obtain limited in vitro toxicology data on inverted pantothenamides. Using LDH release as a measure of viability, we did not detect cytotoxicity of any of the compounds that showed antimicrobial activity up to concentrations of 100 μM (~30 μg ml −1 ).

Discussion
In the present study we describe a new class of stable pantothenamide bioisosteres that display antibiotic   Table 1. activity against staphylococci and streptococci. These compounds target a biochemical pathway that is distinct from currently marketed drugs, and their activity appears to be primarily against clinically relevant Gram-positive species. As there is a strong medical need for improved small-spectrum antibiotics, stable pantothenamides with improved potency may be moved forward as promising compounds for clinical use.
Previous studies have demonstrated that pantothenamide antibiotics based on the pantothenate/pantetheine scaffold are labile in biological fluids due to hydrolysis of the amide bond by serum vanins [13][14][15][16]. Our data reveal that inversion of the amide bond afforded a considerable increase of stability whilst largely preserving the biological activity [25]. A recent paper by Barnard et al. [20] reported that the inversion of the amide bond had a dramatic effect on antibiotic activity towards the S. aureus RN2440 strain. This is in contrast to our findings. We have tested multiple strains and species, and found similar MICs among strains. The basis for the discrepancy between these studies is presently unclear, but is currently actively being pursued in collaboration with Strauss and co-workers. Other structural modifications, reported for antimalarial pantothenamides, that increased the stability include introduction of a methyl group in the C2 linker [19] and variation of the linker length [8]. We further explored structural modifications at the side chain, which did not lead to improved potency compared to the parent compound. Remarkably, the taxon-specificity was clearly dependent on the structural modifications in the molecule. These were found to affect both the specificity at the genus and the species level for Gram-positive organisms studied here. We do not think that the pantothenamides as a class are selective for Gram-positive organisms. In the original paper by Clifton et al [2]. where N5-Pan and N7-Pan were first described, there was significant activity of N5-Pan against the Gram-negative organism E. coli. Here we have developed compounds based on N7-Pan, which appear to be more potent towards Gram-positive organisms. It has to be noted that we only tested a number of clinically relevant genera of Gram-positive and Gram-negative bacteria. A more extensive screen of bacterial species is required to establish the sensitivity of other Gram-negative organisms. The observed lack of sensitivity of E. coli might be explained by the fact that N7-Pan (and possibly also its derivatives) is a substrate for TolC-dependent efflux pumps, as reported by Zhang et al. [10].
Our present study suggests that pantothenamides may be further optimized to yield small-spectrum antibiotics that could target a pathogen without affecting the complete human microbiome. This would be a major advance to avoid the consequences of broad spectrum antibiotics such gut microbiota dysbiosis, thereby reducing antibiotic-associated complications such as diarrhea, Clostridium difficile infections or candidiasis. An example that such specificity is feasible is the development of Debio 1450 (now known as afabicin), an antibiotic candidate with a novel mechanism of action, currently in phase II clinical development for intravenous and oral treatment of staphylococcal infections [26]. It specifically targets Staphylococcus by inhibition of FabI, an enzyme essential for fatty acid synthesis in this genus. Debio 1450 shows a high potency against staphylococci, particularly MRSA, without crossresistance to other antibiotics, including other agents active against staphylococci. Interestingly, this indicates that bacterial fatty acid synthesis is a valid antibiotic target, at least for S. aureus, which has been questioned previously [27]. As CoA is also required for fatty acid synthesis, pantothenamides may also act on bacterial fatty acid synthesis. Addition of serum, which is a source of host fatty acids, to our MIC assays did not significantly inhibit the antibiotics effects. This indicates that exogenous available free fatty acids do not interfere with the mechanism of action.
The mechanism of action of the prototypic pantothenamides has been investigated previously and appears largely dependent on the PanK type of the organism. For E.coli, which has a PanK I type enzyme, the presumed mechanism of action of the pantothenamide N5-pan has been investigated in detail. The formation of coenzyme A antimetabolites [9] that interferes with fatty acid synthetic enzymes [11] and endogenous pantothenate biosynthesis [28] are two affected pathways that are interrupted. It was also reported that N5-Pan could inhibit the growth of S. pneumoniae and that this effect was probably due to fatty acid synthesis inhibition as it could be outcompeted with oleate [10]. The molecular mechanism by which pantothenamides inhibit growth of other Gram-positive bacteria is currently unknown, with an exception for S. aureus. In contrast to other gram-positive bacteria, S. aureus has a PanK II type enzyme that accepts pantothenamides as a substrate [8]. The co-crystal structures of SaPanK II and N7pan revealed that pantothenamides are phosphorylated but remain trapped in the active site [7]. Given the differential effects of many of our compounds towards S. aureus and S. pyogenes, it is likely that the modes of action are distinct.
Clearly, additional efforts are required to identify improved inhibitors that may progress to the clinic. So far we have relied on ligand-based design, but preferably we would direct future campaigns to structure-based design. We are currently investigating the mechanism of action of the pantothenamide bioisosteres.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.