Targeting DXP synthase in human pathogens: enzyme inhibition and antimicrobial activity of butylacetylphosphonate.

The unique methylerythritol phosphate pathway for isoprenoid biosynthesis is essential in most bacterial pathogens. The first enzyme in this pathway, 1-deoxy-D-xylulose 5-phosphate (DXP) synthase, catalyzes a distinct thiamin diphosphate (ThDP)-dependent reaction to form DXP from D-glyceraldehyde 3-phosphate (D-GAP) and pyruvate and represents a potential anti-infective drug target. We have previously demonstrated that the unnatural bisubstrate analog, butylacetylphosphonate (BAP), exhibits selective inhibition of Escherichia coli DXP synthase over mammalian ThDP-dependent enzymes. Here, we report the selective inhibition by BAP against recombinant DXP synthase homologs from Mycobacterium tuberculosis, Yersinia pestis and Salmonella enterica. We also demonstrate antimicrobial activity of BAP against both Gram-negative and Gram-positive strains (including E. coli, S. enterica and Bacillus anthracis), and several clinically isolated pathogens. Our results suggest a mechanism of action involving inhibition of DXP synthase and show that BAP acts synergistically with established antimicrobial agents, highlighting a potential strategy to combat emerging resistance in bacterial pathogens.

INTRODUCTION

The ever present need for discovery of new antibiotics demands concerted efforts to identify novel molecular scaffolds and develop underexplored targets, and ostensibly one will perpetuate the other. A relatively new set of potential antibacterial targets comprises the methylerythritol phosphate (MEP) pathway (Figure 1) to the essential isoprenoid precursors, isopentenyl diphosphate (IDP) and dimethylallyl diphosphate (DMADP) (. Orthogonal to the mevalonate pathway to IDP and DMADP in humans, the MEP pathway is widespread in bacterial pathogens and offers opportunities for development of non-toxic, broad spectrum antibacterial agents.

Figure 1

DXP synthase catalyzes the first step in IDP/DMADP biosynthesis and represents a branch point in bacterial metabolism.

An underexplored target within this pathway is 1-deoxy-D-xylulose 5-phosphate (DXP) synthase, which catalyzes the first step to form DXP from D-glyceraldehyde 3 phosphate (D-GAP) and pyruvate in a thiamin diphosphate-dependent manner (. In many bacterial pathogens, this step is also a branch point in metabolism; DXP is a precursor to pyridoxal phosphate (PLP) biosynthesis and, interestingly, also a precursor to its own essential cofactor, thiamin diphosphate (ThDP, Figure 1). Despite its importance in bacterial metabolism, DXP synthase is underdeveloped as a drug target. Ostensibly, this is largely due to perceived challenges in achieving selective inhibition of DXP synthase over mammalian ThDP-dependent enzymes, such as the E1 subunit of pyruvate dehydrogenase (E1 PDH) and transketolase (TK), and few reports describe inhibitors of this enzyme (. However, DXP synthase appears unique amongst ThDP-dependent enzymes. The Deinococcus radiodurans DXP synthase crystal structure shows a distinctive domain arrangement, ( and the active site is large compared to other ThDP-dependent enzymes (. The unique requirement for ternary complex formation in DXP synthase catalysis ( also sets it apart from other ThDP-dependent enzymes, suggesting that selective inhibition should be possible. Indeed, our previous work ( has demonstrated that selective inhibition of E. coli DXP synthase is achievable with unnatural bisubstrate analogs including butylacetylphosphonate (BAP) and benzylacetylphosphonate (BnAP), both bearing an acetylphosphonate mimic of pyruvate ( and a sterically demanding unnatural acceptor substrate. Presumably, sterically demanding acetylphosphonates bind in the large DXP synthase active site and react with bound ThDP, in a manner similar to methylacetylphosphonate (, to form a stable phosphonolactyl-thiamin diphosphate (PLThDP) intermediate, effectively trapping the enzyme at a pre-decarboxylation complex (Figure 2).

Figure 2

Proposed mechanism of selective inhibition of DXP synthase by butylacetylphosphonate.

Here, we characterize the inhibitory activity of butylacetylphosphonate (BAP), ( against DXP synthase from M. tuberculosis, Y. pestis, and S. enterica serovar Typhi to establish comparable inhibitory activity against this enzyme across these pathogens. Further, we have evaluated the antimicrobial activity of BAP against several bacterial pathogens. Our results indicate that BAP possesses modest antibacterial activity via a mechanism that appears to involve inhibition of DXP synthase, and drug synergism is evident with BAP-ampicillin and BAP-fosmidomycin combinations in E. coli. These results suggest that selective inhibitors of DXP synthase could be effective in antibiotic combinations.

MATERIALS AND METHODS

General Methods

Unless otherwise noted, all reagents were obtained from commercial sources. Spectrophotometric analyses were performed on a Beckman DU800 UV/Visible spectrophotometer. Primers were purchased from Integrated DNA Technologies. E. coli wild-type DXP synthase and the variant E370A DXP synthase were purified as described previously (. E. coli MEP synthase (IspC) was also purified as reported previously (. All microbial manipulation of pathogenic bacteria was conducted in a certified biosafety level 2 (BSL-2) laboratory with all associated safety protocols. S. enterica serovar Typhimurium LT2, B. anthracis Sterne, and all clinically isolated strains are maintained from an in-house bacterial strain library (NAU).

Cloning, overproduction, and purification of M. tuberculosis DXP synthase (Dxs1)

Rv2682c was PCR amplified from H37Rv genomic DNA and ligated into the pET28a+ vector (Novagen), as previously reported (. dxs1-pET28a+ was then transformed into ArcticExpress competent cells (Agilent Technologies). Cells harboring the over-expression plasmid were grown in LB broth at 30°C with shaking at 220 rpm until OD600 = 0.8, then cooled for 30 minutes on ice. After cooling, cultures were induced with 0.1 mM ispropylthiogalactoside (IPTG), and growth was continued with shaking at 13°C for 24 hours. Cells were harvested by centrifugation at 8 000 rpm for 10 min, and the cell pellet was stored at −20°C. The frozen cell pellet was suspended in lysis buffer (3 mL per gram of cells) containing 50 mM Tris HCl (pH =8), 5 mM MgCl2, 10% glycerol v/v, 5 mM mM β-mercaptoethanol (BME), 1 mM PMSF, 1 mM ThDP, 1x protease inhibitor cocktail, 100 mM NaCl, and 0.01% Igepal-CA630. Cells were lysed by sonication and centrifuged for 45 minutes at 18 000 rpm to pellet cell debris. The supernatant was incubated with nickel-nitrilotriacetic (NTA) acid resin in 20 mM imidazole at 4°C for 2 hours, and Dxs1 was eluted from the resin over a stepwise gradient of 5 to 500 mM imidazole. Fractions containing Dxs1 (as determined by 10% SDS-PAGE) were combined and subjected to dialysis overnight at 4°C against 1 liter of 50 mM Tris (pH 8.0), 10 mM MgCl2, 10% glycerol, 1 mM ThDP, and 100 mM NaCl. A second dialysis was carried out against 1 liter of 50 mM Tris (pH 8.0), 10 mM MgCl2, 10% glycerol, 1 mM ThDP, and 100 mM NaCl and 1 mM for an additional 4 h. After dialysis, protein concentration was determined using the Bio-Rad Protein Assay with bovine serum albumin as a standard (yield 1.0 mg/liter of culture). Protein was flash frozen in liquid nitrogen and stored at −80 °C.

Cloning, overproduction, and purification of Yersinia pestis, and Salmonella enterica serovar Typhi DXP synthase

The Y. pestis and S. enterica serovar Typhi DXP synthase genes were cloned into the pMCSG28 vector using ligation independent cloning (LIC) methods as previously described, ( and the resulting plasmids were transformed into Single Step KRX competent cells (Promega). Cells harboring the overexpression plasmid were grown in TB broth containing 100 μg/mL ampicillin, 34 μg/mL chloramphenicol, and 0.1% L(+)arabinose. After growth reached OD600 = 0.6 at 37°C, the cultures were cooled to room temperature and induced with 0.5 mM IPTG and 0.25% rhamnose, and shaking was continued for 16 hours. Cells were harvested, lysed, and purified using the same method as described for M. tuberculosis Dxs1.

DXP synthase kinetic analysis

Michaelis Menten kinetic analyses were carried out as previously described ( with both the natural substrates (pyruvate and D-GAP) and inhibitor, n class="Chemical">butylacetylphosphonate (BAP). BAP was synthesized as reported previously (.

Antimicrobial susceptibility studies

Using aseptic techniques, three to five isolated colonies were picked from a plate containing ATCC MG1655 E coli and Bacillus subtilis 168 were inoculated into 5 mL of cation adjusted Mueller Hinton Broth (CAMHB, Sigma) at 37°C. Incubation was continued until turbidity matched McFarland (MF) turbidity standard 0.5 (~OD600 = 0.10) (. Colony counts were checked after16 hours at 37°C for consistency between experiments. The standardized inoculums (MF = 0.5) contained approximately 1–2 × 108 CFU/mL. The final concentration in a well (or culture tube) was 5 × 105 CFU/mL (. For these studies, the standardized inoculum (MF = 0.5) was diluted 1:100 in CAMHB and 0.25 mL of this adjusted inoculum was added to each tube containing 0.25 mL antimicrobial agent(s) for a final volume of 0.5 mL. These tubes were then incubated at 37°C for 16–18 hours, with shaking. Fractional growth was determined relative to the no drug control, and averaged values were used. The minimum inhibitory concentration (MIC) was determined to be the concentration at which growth was ≦15% of the no drug growth controls (determined by OD600). The concentration of BAP evaluated for these studies ranged from 0 – 5 mM (0 – 1 000 μg/mL). ATCC 70084 Mycobacterium smegmatis was evaluated in a similar manner outlined above, but using enriched Middlebrook 7H9 broth. A 24 hour culture (~ OD600 = 0.4) was diluted to OD600 = 0.1 and subcultured 1:100 for 1 × 106 colony forming units. The concentra3on of BAP was varied 0 310 mM (0 – 2 000 μg/mL). The E. coli glycerol 3-phosphate transporter knockout strain (JW2234-2) was evaluated along with the parent strain (BW25113) using the above method (Yale, Coli Genetic Stock Center, CGSC). Susceptibility studies were performed for S. enterica serovar Typhimurium LT2, B. anthracis Sterne, and all clinical isolates essentially as outlined above with the exception that the assays were conducted in 96-well plates with final volume of 0.2 mL, and concentrations of BAP ranged from 0 – 20 mM (0 – 4 000 μg/mL). All antimicrobial susceptibility assays were performed in triplicate.

E. coli growth rescue studies

For the growth rescue experiments, 96-well plates were used with a final culture volume of 0.2 mL. The standardized inoculum (MF = 0.5) was prepared in the same manner as previously stated. These plates were then incubated in a SpectramaxPlus384 plate reader at 37°C for 16 hours, with shaking at 220 rpm every 30 minutes for 15 minutes. Fractional growth was then determined relative to the no BAP control, and averaged fractional growth was plotted with standard experimental error as a function of BAP concentration. For the 1-deoxy-D-xylulose (DX) rescue experiments, a range of 0 – 1.0 mM DX was evaluated in the presence of 0 – 5.0 mM BAP in CAMHB media. For the thiamin rescue experiments, a range of 0 – 1.0 mM thiamin was evaluated in the presence of 0 – 0.66 mM BAP in M9 minimal media (no thiamin). DX and thiamin rescue experiments were performed in quadruplicate. DXP synthase overexpression rescue experiments were carried out in LB broth, and the following cell types were treated with 0–5.0 mM BAP: BL21 (no vector), pET37b-BL21, wt E. coli dxs-pET37b/BL21, and E370A E. coli dxs-pET37b/BL21. Experiments were performed in triplicate.

DXP synthase overexpression was confirmed by 10% SDS-PAGE, and incorporation of the octa-histidine tag was verified using the InVision His-tag In-gel Stain (Invitrogen) (Figure S2). Site directed mutagenesis of the E370A residue was carried out as previously reported ( using the QuikChange sitedirected mutagenesis kit (Stratagene), and the following primers (Integrated DNA technologies): 5′ CGA CGT GGC AAT TGC CGC GCA ACA GCG GGT GAC CTT TGC 3′ and 5′ GCA AAG GTC ACC GCG TGT TGC GCG GCA ATT GCC ACG TCG 3′.

Checkerboard antibiotic combination studies

For the checkerboard antibiotic combination studies in E. coli, culture tubes were used with a final culture volume of 0.5 mL, prepared as described above. The concentration of ampicillin evaluated for these studies ranged from 0, 0.13, up to 8 μg/mL, tetracycline ranged from 0, 0.125, up to 4 μg/mL, and fosmidomycin ranged from 0, 0.19, up to 6 μg/mL.

For each antibiotic combination, the fractional inhibitory concentration (FIC) index was determined to predict drug synergism (.

The FIC index was calculated using equation 1 (. MICBAP and MICX are the lowest concentrations of BAP or drug X (X = ampicillin, fosmidomycin, or tetracycline) showing ≤ 15% growth. The FICBAP was calculated as the [BAP in the presence of drug X] for a well showing ≤ 15% growth, divided by MICBAP. FICX was calculated as the [drug X in the presence of BAP] in the same well, divided by MICX. The FIC index is the sum of FICBAP and FICX. For each combination the FICI values were used to indicate drug synergism (x < 0.5), additivity (0.5 < x < 1.0), indifference (1.0 < x < 2.0), or antagonism (x > 2). Fractional growth was determined relative to the no drug growth control and average values were used. The BAP-ampicillin and BAP-tetracycline checkerboard assays were performed in duplicate. The BAP-fosmidomycin checkerboard assay was performed in quadruplicate.

RESULTS

BAP selectively inhibits DXP synthase from Mycobacterium tuberculosis, Yersinia pestis, and Salmonella enterica serovar Typhi

Our previous work has established alkyl acetylphosphonates as selective inhibitors of E. coli DXP synthase (, suggesting that it is possible to achieve selective inhibition of this ThDP-dependent enzyme on the basis of differences in substrate usage and mechanism (. Here, in vitro inhibition kinetics were performed to characterize BAP inhibitory activity against DXP synthase from M. tuberculosis, S. enterica, and Y. pestis (Table 1) to determine the feasibility of this approach across pathogens. Both S. enterica and Y. pestis DXP synthase have similar primary amino acid sequence identity with the E. coli enzyme (~90%) (Table S1). Mycobacterium tuberculosis DXP synthase (Dxs1) is the least similar to each of the three (~40%). All DXP synthase enzymes were kinetically characterized using the enzyme coupled assays as previously described ( and the results are shown in Table 1. The Km values obtained for Dxs1 from M. tuberculosis in this study (K pyruvate = 81 ± 3 and K D-GAP = 16.6 ± 0.9) (Figure S2) are comparable to those previously reported (. K pyruvate and K D-GAP determined for DXP synthase from Y. pestis (K pyruvate = 51 ± 6 and K D-GAP = 26 ± 3) and S. enterica (Kpyruvate = 76 ± 19 and the K D-GAP = 12 ± 1) (Figure S2) are also comparable to kinetic constants measured for other DXP synthase enzymes (.

Table 1

Kinetic analysis of DXP synthase from Y. pestis, M. tuberculosis, and S. enterica DXP synthase.

species	substrate	Km (μM)	kcat (min−1)	Ki BAP (μM)
Y. pestis	pyruvate	51 ± 6	112 ± 7	7.5 ± 0.9
	D-GAP	26 ± 3
M. tuberculosis	pyruvate	81 ± 3	48 ± 1	4 ± 2
	D-GAP	17 ± 1
S. enterica	pyruvate	76 ± 19	140 ± 20	8.4 ± 0.4
	D-GAP	12 ± 1

The inhibitory activity of BAP was characterized against each enzyme using the enzyme coupled assay as previously described ( (Figure S3), and Ki values are summarized in Table 1. In all cases, low micromolar inhibitory activity of BAP is observed that is comparable to the inhibitory activity of BAP against E. coli DXP synthase (. As expected, the mode of inhibition is competitive with respect to pyruvate and noncompetitive with respect to D-GAP (for M. tuberculosis DXP synthase, ; for Y. pestis DXP synthase, ; for S. enterica DXP synthase, ).

BAP exhibits weak antimicrobial activity

As a starting point, BAP was evaluated for antimicrobial activity against the Escherichia coli ATCC strain MG 1655 using the macrodilution method (. BAP exhibits minimum inhibitory concentrations (MIC = [BAP] causing growth inhibition) of ~1 000 μg/mL in cation-adjusted Mueller Hinton Broth and a significantly lower MIC of 122 μg/mL in M9 minimal media (Table 2). BAP was evaluated against several clinically isolated pathogens (Table S2), and BAP exhibits an MIC of 4 000 μg/mL against Salmonella enterica serovar Typhimurium and Micrococcus sp., 1 000 μg/mL against Bacillus anthracis Sterne, and 4 000 μg/mL against Pseudomonas aeruginosa in cation adjusted Mueller Hinton Broth.

Table 2

Antimicrobial activity of BAP against pathogenic bacteria and FIC index values for antibiotics in combination with BAP.

organism	MIC BAP (μg/mL)	combination	FIC index
Escherichia coli MG1655	1000	BAP-fosmidomycin	0.25
		BAP-ampicillin	0.31
		BAP-tetracycline	0.53

We considered the possibility that the weak antimicrobial activity of BAP is attributed to poor cellular uptake. Small phosphorylated (or phosphonylated) molecules, including the phosphonate fosmidomycin, are actively transported via the glycerol 3-phosphate transporter (glpT) (. BAP was evaluated for antimicrobial activity against a glpT knockout strain (JW2234-2) and its parent strain (BW25513) (Yale, Coli Genetic Stock Center) to determine whether this is a possible uptake mechanism. However, while fosmidomcyin is inactive against the glpT knockout strain, BAP exhibits comparable antimicrobial activity against the glpT knockout and parent strains (Figure S4). This result indicates the glycerol 3-phosphate transporter is not likely a mechanism of entry for BAP.

DXP synthase is an intracellular target of BAP

Experiments to rescue the growth inhibitory effects of BAP against n class="Species">E. coli MG1655 were performed to provide evidence for DXP synthase as an intracellular target.

Rescue by downstream metabolites

Previous reports demonstrate labeling of downstream MEP pathway intermediates in E. coli in feeding experiments using labeled 1-deoxy-D-xylulose (DX), suggesting DX is cell permeable and undergoes intracellular phosphorylation at some level to produce DXP (. Here, partial rescue of BAP inhibitory activity is observed when the growth medium is supplemented with DX (50% reduction in growth inhibition by BAP in the presence of DX) (Figure 3A). As DXP synthase is a branch point in bacterial metabolism, additional biosynthetic pathways including thiamin diphosphate and pyridoxal phosphate biosynthesis ( should be affected by inhibition of DXP synthase. Further, inhibition of ThDP biosynthesis might potentiate the loss of DXP synthase activity as this cofactor is also required for DXP synthase-catalysis. Here, complete rescue of growth inhibitory effects of 31.25 μg/mL BAP (165 μM) is observed in the presence of thiamin (Figure 3B). E. coli growth rates in the presence or absence of thiamin are similar (Figure 3C), suggesting intracellular thiamin concentration is sufficient to support normal growth under conditions where ThDP biosynthesis is not inhibited. While growth of BAP treated cells is restored by 16 hours in the presence of thiamin, there is a clear delay in growth under these conditions. This may reflect the time required for intracellular phosphorylation of thiamin under growth inhibitory conditions. Under conditions of higher BAP concentration, thiamin supplementation does not rescue growth by 16 hours (Figure 3C).

Figure 3

Butylacetylphosphonate (BAP) inhibition of E. coli MG1655 growth is rescued by 1-deoxy-D-xylulose (DX), thiamin (Th) and E. coli DXP synthase (Dxs) overexpression. A) Fractional growth at 12.5 hours in the presence and absence of BAP and DX. Mean and standard error of four replicates are shown. *p = 0.022 (unpaired t test comparing DX treated and untreated cultures at 0.66 mM BAP). B) Fractional growth at 16 hours in the presence and absence of BAP and thiamin. Mean and standard error of four replicates is shown. ***p = 0.0000258 (unpaired t test comparing thiamin treated and untreated cultures at 165 μM BAP). C) Growth curve over 16 hours, in the absence or presence of BAP and thiamin. D) Fractional growth at 16 hours determined for E. coli BL21 cells only, cells transformed with empty pET37b vector, vector expressing recombinant E. coli DXP synthase protein (dxs-pET37b), or vector expressing the catalytically inactive ( E370A variant (E370A dxs-pET37b). Mean and standard error of four replicates is shown.

Rescue by DXP synthase overexpression

Overexpression of DXP synthase should rescue growth of n class="Chemical">BAP-treated E. coli ( if BAP acts by inhibition of DXP synthase. Indeed, increasing intracellular levels of DXP synthase (Figure S1) results in rescue of bacterial growth (Figure 3D) compared to E. coli harboring empty vector or the catalytically inactive E370A mutant (.

Growth inhibition is significantly enhanced when BAP is administered in antibiotic combinations

The weak antimicrobial activity of BAP on its own prompted a study to determine the synergistic effects of BAP in combination with several established antibiotics. As a starting point, we have investigated synergistic effects of BAP in combination with fosmidomycin, ampicillin, and tetracycline using the checkerboard method (. Fosmidomycin potently inhibits the second enzyme in the MEP pathway, MEP synthase (IspC) (Figure 1) (. Given the established synergy that can take place between agents that target the same metabolic pathway, and in particular the synergy observed between fosmidomycin and inhibitors of isoprenoid biosynthesis enzyme farnesylpyrophosphate synthase (FPPS) (, we hypothesized that BAP could synergize with fosmidomycin. Likewise, fosmidomycin has been shown to synergize with cell wall biosynthesis inhibitors, including ampicillin, to inhibit bacterial cell growth (. Those early observations established precedence for drug synergisms between agents targeting isoprenoid biosynthesis and cell wall biosynthesis (which relies upon the prenylated Lipid II) and suggest synergy between BAP and cell wall biosynthesis inhibitors is also possible. Finally, we considered the possibility that synergy could exist between BAP and the bacterial translation inhibitor tetracycline, on the basis that geranylation of tRNA is proposed to play a role to modulate codon recognition during translation (, thus drawing a connection between isoprenoid biosynthesis and protein synthesis.

Here, the fractional inhibitory concentration (FIC) index was determined for BAP-fosmidomycin, BAP-ampicillin, and BAP-tetracycline combinations against E. coli MG1655 to predict drug synergisms (. Based on this analysis, the antimicrobial activity of BAP appears to be enhanced in combination with fosmidomycin (FIC index = 0.25, Table 2) or ampicillin (FIC index = 0.31, Table 2). Analysis by isobolograms suggests pronounced synergy in the BAP-fosmidomycin combination against E. coli MG1655 (Figure S5) (. However, examination of the BAP-ampicillin combination by isobologram indicates synergy is less pronounced, suggesting this combination may display additive effects. The FIC index for BAP in combination with tetracycline is 0.53, indicating an additive relationship (Table 2).

DISCUSSION

As the isoprenoid biosynthetic pathway is absent in mammals, but essential in many human pathogens, each of the seven enzymes in the pathway represents a prospective antibiotic target. Selective inhibition of DXP synthase could reduce flux through multiple metabolic pathways, and this represents a potential advantage of targeting the first step of the MEP pathway. Here we have evaluated the inhibition and antimicrobial activity of butylacetylphosphonate (BAP), a DXP synthase inhibitor developed in our lab to selectively target the uniquely large active site of this enzyme (.

BAP displays low micromolar inhibitory activity against DXP synthase from M. tuberculosis, S. enterica, and Y. pestis, and this is consistent with the idea that BAP targets a mechanism common to all pathogenic DXP synthase enzymes, but distinct from mammalian ThDP-dependent enzymes. Not surprisingly, BAP acts as a competitive inhibitor with respect to pyruvate in all cases, indicating that the acetylphosphonate moiety of BAP acts as a pyruvate mimic in this unnatural bisubstrate analog.

BAP appears to exert antimicrobial activity against E. coli MG1655 through a mechanism that involves DXP synthase inhibition, as suggested by the observed rescue of its inhibitory effects in the presence of DX or thiamin, or under conditions of DXP synthase overexpression. Partial rescue is observed using nonphosphorylated 1-deoxy-D-xylulose (DX), implying DXP synthase as a target. More pronounced rescue is observed in the presence of thiamin or DXP synthase overexpression. In addition to rescuing the thiamin pathway and other thiamin-dependent cellular processes under these conditions, it is possible that ThDP generated under thiamin rescue conditions regenerates active DX synthase by competing with butylphosphonolactyl-thiamin disphosphate (butyl PLThDP) at the ThDP binding site. Alternatively, increasing intracellular ThDP levels could activate an apo-DXP synthase pool that presumably arises under conditions of DXP synthase inhibition and subsequent cofactor depletion. Not surprisingly, overexpression of active DXP synthase in E. coli also results in pronounced rescue of BAP growth inhibitory activity. The observation that rescue is less pronounced under all of these conditions with increasing BAP is consistent with DXP synthase inhibition coupled with other nonspecific antimicrobial effects of BAP at high concentrations.

The reason for the observed weak antimicrobial activity of BAP on its own against the pathogens tested here is not entirely clear, although poor cell permeability is conceivable. Recent studies by Circello, et al. ( reveal the natural product dehydrophos as a prodrug of the structurally related methyl acetylphosphonates (MAP), a mimic of pyruvate and known inhibitor of the ThDP-dependent enzymes that catalyze decarboxylation of pyruvate (. This phosphonotripeptide natural product is imported via oligopeptide permeases and cleaved by intracellular peptidases to release MAP. By extension, the acetylphosphonate scaffold of BAP may hamper efficient uptake in the absence of such a delivery mechanism. It is intriguing to contemplate a similar delivery strategy for enhancing cellular uptake of BAP and analogs, although it is also possible that incorporation of more hydrophobic substituents may enhance permeability.

Despite the weak antimicrobial effects of BAP on its own, BAP-fosmidomycin and BAP-ampicillin combinations are synergistic against E. coli MG1655. The synergistic BAP-fosmidomycin combination was anticipated, given that drug synergism in the MEP pathway has been previously reported with FPPS inhibitors and fosmidomycin (. Our previous work suggesting MEP activates and sustains IspF activity in a possible feed-forward regulatory mechanism in bacterial isoprenoid biosynthesis ( could provide additional basis for this synergistic combination. It is conceivable that inhibition of upstream enzymes DXP synthase and IspC by BAP and fosmidomycin, respectively, not only reduces flux through the pathway at these points, but also effectively expedites the loss of IspF activity by rapidly depleting MEP levels. There is established precedence for drug synergisms between inhibitors of isoprenoid biosynthesis and cell wall biosynthesis (, given the requirement for isoprenoid precursors in Lipid II biosynthesis (, and so the synergy observed with the BAP-ampicillin combination in this study was also anticipated. It is also plausible that ampicillin compromises cell wall integrity to permit more efficient uptake of BAP. These findings are significant given the necessity of combination therapies to prevent emergence of antimicrobial resistance (. Although inhibition of DXP synthase ought to simultaneously reduce flux through multiple essential pathways, the emergence of mechanisms to circumvent such an intervention in bacteria is inevitable (.

This work highlights unnatural bisubstrate analogs as selective inhibitors of this unique ThDP-dependent enzyme and promising antimicrobial agents as components of synergistic antimicrobial combinations. While the acetylphosphonates were developed over 30 years ago as ketoacid mimics, and are powerful tools to study ThDP-dependent enzymes (, in the case of dehydrophos, it is clear that Nature has already conceived of such an antimicrobial strategy to inhibit essential ThDP-dependent processes. Development of acetylphosphonates to increase selectivity, potency, and permeability is perhaps a new chapter in medicinal chemistry on natural product scaffolds.