Total Syntheses of Bulgecins A, B, and C and Their Bactericidal Potentiation of the β-Lactam Antibiotics.

The bulgecins are iminosaccharide secondary metabolites of the Gram-negative bacterium Paraburkholderia acidophila and inhibitors of lytic transglycosylases of bacterial cell-wall biosynthesis and remodeling. The activities of the bulgecins are intimately intertwined with the mechanism of a cobiosynthesized β-lactam antibiotic. β-Lactams inhibit the penicillin-binding proteins, enzymes also critical to cell-wall biosynthesis. The simultaneous loss of the lytic transglycosylase (by bulgecin) and penicillin-binding protein (by β-lactams) activities results in deformation of the septal cell wall, observed microscopically as a bulge preceding bacterial cell lysis. We describe a practical synthesis of the three naturally occurring bulgecin iminosaccharides and their mechanistic evaluation in a series of microbiological studies. These studies identify potentiation by the bulgecin at subminimum inhibitory concentrations of the β-lactam against three pathogenic Gram-negative bacteria and establish for the first time that this potentiation results in a significant increase in the bactericidal efficacy of a clinical β-lactam.

The appearance of extensively drug-resistant bacterial infections in both the hospital and the community has sustained debate as to whether the era of successful monochemotherapy of bacterial infections is fading. An alternative approach, with precedent in cancer chemotherapy, is the combination of complementary agents. The combination of an antibiotic with another potentially synergistic antibiotic, or with a compound that would potentiate the activity of the antibiotic, is attractive in the light of the dearth of clinical options. While such combinations can have extraordinary medical value (as exemplified by the combination of β-lactams with β-lactamase inhibitors),[1−4] every aspect to the selection and implementation of successful combination therapy is a scientific challenge.[5−10] With respect to this challenge, useful guidance may be provided by Nature.

Shinagawa and co-workers,[11,12] of the antibiotic discovery group at Takeda, reported the isolation of bulgecins A–C (1–3 of Figure ) in 1982. Each is a simple variation on a glycosylated iminosaccharide, and each is biosynthesized by a Gram-negative bacterium (annotated then as Pseudomonas mesoacidophilia but recently reassigned as Paraburkholderia acidophila).[13,14] Additional studies with this bacterium by Asai and co-workers led to the isolation of a monobactam-class β-lactam antibiotic, sulfazecin (4).[15,16] The combination of bulgecin (itself bereft of antibacterial activity) with sulfazecin significantly improved the latter’s antibacterial activity toward other Gram-negative bacteria.[17,18] All doubts that this potentiation was fortuitous were dispelled by the observation of an intimate connection of the genes encoding the biosynthetic enzymes for sulfazecin and bulgecin.[14,19,20] Hence, Nature selected coproduction of an antibacterial agent and a potentiator within the same producer organism.

Figure 1

Chemical structures of the bulgecins and of sulfazecin. Bulgecins are glycopeptides comprised of a common β-sulfo-GlcNAc linked to a disubstituted l-proline. Bulgecin A is the most abundant of the bulgecin natural products. Sulfazecin is a monobactam β-lactam antibiotic that is structurally distinguished by the sulfamate functional group.

The morphological effect resulting from the complementarity of these two agents is a characteristic midcell bulge (hence “bulgecin”) that precedes lysis of the bacterium. A biochemical basis for this effect was provided by Templin et al.,[21] who confirmed the original observations of Imada et al.[11] that the bulgecin effect extended to other β-lactam antibiotics and determined that the bulge was a result of bulgecin inhibition of a periplasmic enzyme, the Slt lytic transglycosylase. As optimal potentiation with bulgecin was found with β-lactams that inactivated the penicillin-binding proteins (PBPs) of septal cell-wall formation, the hypothesis was presented that the simultaneous loss of the catalytic activity of these PBPs by β-lactams and of the lytic transglycosylase(s) (LTs) by bulgecin results in a structurally defective cell-wall septum. This hypothesis has experimental support.[22]

The lytic transglycosylases of Gram-negative bacteria share key attributes with the PBPs. Both are found as families of enzymes within Gram-negative bacteria. For example, the Gram-negative pathogen Pseudomonas aeruginosa has eight different PBPs and 11 distinct LTs. Within each family, there is a redundancy of function but an inability to tolerate overall loss of all function. In comparison to the PBPs (where there is a credible hypothesis for the function of each family member), the LT family is both more structurally and more functionally diverse. At present within the LT family, there are few guiding hypotheses as to the function of its individual members.[23,24] The understanding of LT function is (even 25 years after their discovery) a frontier at the nexus of bacterial shape, bacterial cell-wall biosynthesis, and antibiotic mechanism.

Given our ambitions to critically explore this nexus,[24−28] access to bulgecins (which have proven invaluable for LT study)[29−32] was required and we opted for total synthesis. We report here practical syntheses of bulgecins A, B, and C. Bulgecin A was accomplished in 34 steps and 3% overall yield from d-serine as the starting material. Furthermore, we revisit the biological activities of these compounds by modern methodology to disclose the unique potentiating activity of bulgecins.

Results and Discussion

Synthesis of the Bulgecins

Bulgecin A (1) is a glycopeptide comprised of three segments: a sulfo-N-acetyl glucosamine (sulfo-GlcNAc) saccharide, a 3R-hydroxy-4S-(hydroxymethyl)-l-proline (numbering from Cα) core (named bulgecinine), and a taurine in amide linkage to the bugecinine core. Bulgecin A is the most abundant of the three bulgecins found as natural products (Figure : bulgecin B has a β-alanine amide replacing the taurine; bulgecin C lacks amide functionalization of its proline carboxylate: structures 2 and 3, respectively). While several syntheses of bulgecinine were reported in the decades following the discovery of the bulgecins, only two syntheses of bulgecins have been reported. The first synthesis reported was that of bulgecin A by a group from Osaka University (without experimental details).[33] Their synthesis was followed by the synthesis (with experimental details) of bulgecin C by Barrett and Pilipauskas.[34,35] The key decision points for our own synthetic plan (as guided by the useful disclosures in these previous syntheses) were the selection of a concise and high-yielding preparation of a protected bulgecinine and the choice of a protected and activated GlcNAc glycosyl donor for the β-selective glycosylation. In both respects, our own choices diverged from those of Wakamiya et al.[33,36] and Barrett and Pilipauskas.[34,35]

For the synthesis of protected bulgecinines, we were drawn to a classical method for the stereoselective synthesis of substituted pyrrolidines, that of an intramolecular aminomercuration of a γ-alkene to a carbamate-protected amine, followed by the retentive O2-induced oxidative cleavage of the intermediate carbon–mercury bond to give a secondary alcohol.[37] The value of this methodology for the synthesis of bulgecinine was established by Khalaf and Datta[38] and confirmed by Wang et al.[39] In this route, the single stereocenter of the d-serine starting material sets the absolute stereochemistry of the two additional stereogenic carbons of the pyrrolidine. We therefore envisioned 5 as the key intermediate, disassembled retrosynthetically into the known GlcNAc donor 6 and the protected bulgecinine 7, which in turn would be synthesized from d-serine (9) through Wang’s trisubstituted pyrrolidine 8. The synthetic plan is shown in Scheme .

Scheme 1

Synthetic Plan for Bulgecin A (1)

The GlcNAc donor 6 was prepared using a reported procedure.[40] Allyl carbamate 10 (Scheme ) was prepared as described by Wang et al.[39] Its transformation to the protected pyrrolidine 8 (with confirmation of its structure by X-ray crystallographic analysis; see box in Scheme and Figure S1), by successive aminomercuration and oxidative demercuration with oxygen gas, used a modified adaptation of Wang’s procedure. Selective oxidation of the primary alcohol of 8 to the carboxylic acid used the sequential combination of Dess-Martin and Pinnick oxidations, with the resulting carboxylic acid transformed to the methyl ester (11) with TMSCHN2. Zemplen-type removal of the O-acetyl group afforded the key protected bulgecinine derivative 7.

Scheme 2

Synthesis of 5 from d-Serine (9)

Reagents and conditions: (a) Hg(OAc)2, MeCN, 0 °C, 1.5 h; EtOAc, brine, 2 h; (b) NaBH4, O2, DMF, 3 h; (c) DMP, DCM, 0 °C to rt, O/N; (d) 2 M 2-methyl-2-butene in THF, t-BuOH, NaClO2–NaH2PO4 in H2O, 3 h; (e) 2 M TMSCHN2 in hexane, toluene, MeOH, 2 h; (f) K2CO3, MeOH, 1.5 h, 96%; (g) 6, TMSOTf, MS4A, DCM, 0 °C, 3 h; (h) BnBr, Ag2CO3, toluene, O/N; (i) Ac2O, pyridine, 0 °C to rt, O/N; (j) Hg(OAc)2, MeCN, 0 °C to rt, O/N; EtOAc, brine, 1.5 h; (k) NaBH4, O2, DMF, 3 h; (l) DMP, DCM, 0 °C to rt, 1.5 h; (m) 2 M 2-methyl-2-butene in THF, t-BuOH, NaClO2–NaH2PO4 in H2O, 1 h; (n) 2 M TMSCHN2 in hexane, toluene, MeOH, 1.5 h; (o) K2CO3, MeOH, 1 h; (p) 6, TMSOTf, MS4A, DCM, 0 °C, 3 h; (q) 1 M TBAF in THF, AcOH, O/N; (r) BTCA, TfOH, DCM, O/N; (s) K2CO3, MeOH, 3 h.

Glycosylation of 7 with GlcNAc donor 6 using TMSOTf activation gave a complex mixture of byproducts. Presuming that the TBS protecting group is not preserved under the acidic conditions, we decided to use the O-benzyl bulgecinine derivative 15 as the acceptor (Scheme ). We derived the necessary O-benzylated allylcarbamate 13 by selective benzylation of the primary alcohol, followed by acetylation of diol 12. Aminomercuration and successive oxidative demercuration gave 14. The yield (72%) was comparable to the yield reported previously.[39] Oxidation and protecting group manipulation of 14 (as described for 7) gave 15. To confirm that the stereochemistry of 15 is identical to that of 7, we also synthesized 15 from 11 (desilylation with TBAF, followed by O-benzylation using benzyl trichloroacetimidate with TfOH activation; third row of Scheme ). In contrast to the glycosylation of 7, glycosylation of 15 was a clean reaction that proceeded with satisfactory yield (71%).

The reductive dechlorination (Bu3SnH, AIBN) of the nonparticipating N-trichloroacetyl group of 5 was uneventful, giving 18 in a 91% yield (Scheme ). Mindful of the possibility of Cα epimerization of the bulgecinine core, milder saponification conditions (compared to previous reports)[35] were used (aq KOH in iPrOH). The resulting carboxylic acid 19 was converted to O-succinimidyl active ester 20, and its secondary alcohol was sulfated to afford 21. The overall yield for these four steps from 5 was excellent (67%). The remaining two steps provided a challenge. Acyl transfer to taurine occurred in moderate yield (52%), in part due to the difficult isolation and purification of 22.

Scheme 3

Synthesis of Bulgecin A (1)

Reagents and conditions: (a) Bu3SnH, AIBN, toluene, rt to 90 °C, O/N; (b) 4 M KOH aq., i-PrOH, 3 d; (c) NHS, DIC, DCM, 0 °C to rt, O/N; (d) SO3·py., DMF, 2 h; (e) taurine, TEA, H2O, MeCN, O/N; (f) HCO2NH4, Pd black, MeOH, O/N.

In both previous syntheses of bulgecins, difficulties were encountered with the hydrogenolysis required for the final deprotections.[33−35] We confirmed this difficulty. In our hands as well, transfer-hydrogenation conditions, as identified by Barrett and Pilipauskas,[34,35] were the superior choice. However, while this procedure in our hands achieved full deprotection to bulgecin A (1), the sample had 15% of a byproduct that could not be removed without substantial loss of product. On the basis of an observation that ammonium formate was a superior reducing agent compared to formic acid under transfer-hydrogenation conditions (increased yield of O-benzyl deprotection and reduced impurity levels),[41] we finally obtained pure bulgecin A (1) in excellent yield (89%) for this last step. We further observed that the use of Na+-form cation-exchange resin in the reaction workup was superior to the use of H+-form strongly acidic cation-exchange resin. This latter resin also gave an impurity. These observations suggest that bulgecin A is not stable in the presence of strong acid. Bulgecin A was obtained from d-serine as the starting material in 34 synthetic steps and with a 3% overall yield. Syntheses of bulgecins B and C were accomplished from the intermediates used in the synthesis of bulgecin A (Scheme ).

Scheme 4

Synthesis of Bulgecin B (2) and C (3)

Reagents and conditions: (a) β-alanine, TEA, H2O, DCM, O/N; (b) HCO2NH4, Pd black, MeOH, O/N; (c) SO3·py., DMF, 3 h; (d) HCO2NH4, Pd black, MeOH, O/N.

Coupling of 21 with β-alanine gave precursor 23, which was converted quantitatively into bulgecin B (2). Compound 24 was synthesized by O-sulfation of 19. Transfer hydrogenation using ammonium formate gave bulgecin C (3) in good yield (79%).

Bulgecin A Potentiation with a Clinical β-Lactam Antibiotic As Assessed by Bacterial Growth Curves

We assessed bulgecins A, B, and C in microbiological assays. We performed growth-curve assays[14,31] for P. aeruginosa. These experiments were conducted in the presence of 2- to 8-fold below minimal-inhibitory concentration (sub-MIC) of the β-lactam antibiotic ceftazidime, in the absence of ceftazidime, in the presence of bulgecin (A, B, or C), and in the presence of bulgecin (A, B, or C) and sub-MIC of ceftazidime (Figure S2). Each synthetic bulgecin analog demonstrated comparable ability to potentiate ceftazidime against P. aeruginosa. For this reason, we continued our additional analyses with bulgecin A.

Next, we performed growth-curve assays for the four Gram-negative members of the ESKAPE panel of bacterial pathogens (“KAPE”: , , , and species). The ESKAPE panel members comprise the most problematic bacterial pathogens.[42−45]Figure documents the potentiation of the ceftazidime activity for P. aeruginosa, Enterobacter aerogenes, and A. baumannii. The results for K. pneumoniae indicated no potentiation (data not shown), possibly implying lack of penetration of bulgecin A into the periplasmic space of this organism. The bacteria grew in the absence of the antibiotic, in the presence of sub-MIC of ceftazidime, or in the presence of bulgecin A by itself. However, potentiation by bulgecin A was seen at sub-MIC of ceftazidime for these three bacteria (Figure A). In the case of P. aeruginosa, bulgecin A caused rapid cell lysis once bacteria reached the mid-logarithmic phase of growth (approximately OD600 = 0.5) in the presence of ceftazidime. In contrast, bulgecin A and sub-MIC of ceftazidime caused sustained and early inhibition of growth of E. aerogenes and A. baumannii.

Figure 2

Bacterial growth-curve assays and bacterial bulge-formation as visualized by microscopy. (a) Growth-curve assays of P. aeruginosa, E. aerogenes, and A. baumannii. The results for no antibiotic (yellow curve), 50 μg/mL bulgecin A (green curve), ceftazidime (CAZ; blue curve), and 50 μg/mL bulgecin A + CAZ (red curve) are shown. The sub-MIC concentrations of CAZ used for P. aeruginosa, E. aerogenes, and A. baumannii were 0.28, 8, and 2 μg/mL, respectively. (b) Confocal microscopy demonstrating that GFP-labeled P. aeruginosa grown in the presence of sub-MIC of CAZ exhibits elongation without bulges (on the left) and those grown in the presence of CAZ and bulgecin A exhibit elongation with bulges (on the right). The boxed area is expanded to highlight the bulges. A 10 μm scale bar is given in the top left corner. (c) In cell-wall homeostasis, lipid II is polymerized to the nascent peptidoglycan chain, comprised of repeats of the N-acetylglucosamine (NAG, light-green hexagons)–N-acetylmuramylpentapeptide (NAM, dark-green hexagons) disaccharide, by the transglycosylase (TG). The transpeptidase (TP) cross-links the nascent peptidoglycan to the growing edge of the cell wall. (d) β-Lactam antibiotics inhibit the TP activity, resulting in the accumulation of non-cross-linked nascent peptidoglycan, which serves as substrate for the lytic transglycosylase Slt in P. aeruginosa. Bulgecin A inhibits the Slt activity. Dual inhibition of Slt and TP results in the formation of the bulge, leading to the breached structural integrity of the cell wall.[22]

We visualized these events using a PAO1 strain of P. aeruginosa containing a mini-Tn7 chromosomal, constitutive green fluorescent protein (GFP)-expressing gene insertion, which fluoresces upon excitation at 488 nm with light emission at 525 nm. We performed swarm-motility assays (as described previously).[46,47] Bulgecin A was combined with P. aeruginosa planktonic culture and spot inoculated at the plate center, and ceftazidime was placed at a second site 20 mm away. These placements allowed bacteria to be exposed to bulgecin A prior to swarming toward ceftazidime. A control experiment omitted bulgecin A. Bacteria were imaged at the swarm-colony edge. We saw the elongated phenotype in both cases due to bacterial exposure to sub-MIC of ceftazidime.[48,49] However, in the presence of bulgecin A, we saw extensive and periodic bulge formation in elongated bacteria. The bacteria in this group also demonstrated an increase in girth. In the control experiment in the absence of bulgecin A, these bulges were absent (Figure B). The formation of bulges was documented previously by Imada and co-workers by scanning electron microscopy and is the origin of the name of the natural product.[11,18] The bulges are at the sites of cell division (midcell), whereby failure of the organism to divide in the presence of ceftazidime is coupled with the formation of the bulges (Figure B). We also showed that cells grown in the presence of bulgecin A, but without antibiotic, displayed no alteration in the phenotype (not shown). This observation is likely due to the multiplicity of LTs, 11 are known in P. aeruginosa, with overlapping redundant activities.[27] Not all of the 11 LTs are expected to be inhibited by bulgecin A.[50]

Bulgecin A potentiation of a clinical β-lactam antibiotic was visualized by microscopy. A distinct and reproducible feature of P. aeruginosa growth in the presence bulgecin A and sub-MIC of ceftazidime is that the rate of growth was largely unperturbed until the midlog phase, when cell lysis occurs. The reason for this lag, not seen with the other two organisms, is not known. We analyzed this event using the swarm assays, with one exception. The bacteria grown to midlog phase were  spot inoculated at one position on an agar plate, and a mixture of bulgecin A and ceftazidime was placed at a second position on the plate at a distance of 20 mm. This permutation was intended to mimic the effect of an encounter of bacteria with two drugs in combination (bulgecin A and ceftazidime). At the conclusion of the experiment, propidium iodide (λexci 561 nm, λemi 595 nm) was imaged to visualize DNA liberated by cell lysis.[51] We imaged the swarm-colony edge (white arrow, top left panel) for each case (Figure A) at 14 and 22 h, after inoculation. In the case of control bacteria and that of bacteria in the presence of bulgecin A alone, we saw no effect. In the presence of ceftazidime alone, we saw an elongated cell phenotype after 14 h (Figure A). Cell lysis was not detected. We began to see cell lysis by 22 h (Figure B) in the presence of ceftazidime alone. In contrast, when bulgecin A and ceftazidime were present at 14 h, the elongated-bulged bacteria had already begun to lyse (Figure A, bottom right corner). Significant cell death at 22 h is seen only with the combination of the two (Figure B, bottom right corner).

Figure 3

Bulgecin potentiation of the bactericidal activity of ceftazidime. GFP-labeled P. aeruginosa was  spot inoculated at the center of a 100 mm swarm plate and 350 μg of bulgecin A and/or 5 μg of CAZ were/was spotted at the red dot. Bacteria were imaged at (a) 14 h and (b) 22 h postinoculation. The fluorescent images (100× magnification) were captured at the  swarm-colony edge. As a representative example, an arrow marks the imaging site for the top left plate. Red fluorescence indicates cell lysis. A 10 μm scale bar is given in the left-most panel.

Conclusions

The prescient discovery of Imada et al.[11] of the bulgecins was made in an era when clinical options for treatment of infections were not as limited as the present. Some infections by Gram-negative bacteria may be treated currently with only a single antibiotic, and certain infections could be fatal by more than 50%, such that a return to the preantibiotic era has become a subject of discussion.[52−55] We revisited bulgecins with this clinical backdrop. Our practical syntheses of the bulgecins and our microbiological proof that their potentiation is bactericidal at sub-MIC of ceftazidime, a β-lactam used clinically to treat Gram-negative ESKAPE infections, merit a fresh look at bulgecins. We hasten to add that the mixture of bulgecin A and sub-MIC of ceftazidime results in early onset of bactericidal activity, which is critical for rapid reduction of bacterial load in an infection. Bulgecins as potentiators of antibacterial activity hold great promise. This study is not a culmination for the bulgecins but emphatic evidence supporting their further mechanistic study.