Vancomycin Derivatives That Inhibit Peptidoglycan Biosynthesis Without Binding d-Ala-d-Ala

See allHide authors and affiliations

Science  16 Apr 1999:
Vol. 284, Issue 5413, pp. 507-511
DOI: 10.1126/science.284.5413.507


Vancomycin is an important drug for the treatment of Gram-positive bacterial infections. Resistance to vancomycin has begun to appear, posing a serious public health threat. Vancomycin analogs containing modified carbohydrates are very active against resistant microorganisms. Results presented here show that these carbohydrate derivatives operate by a different mechanism than vancomycin; moreover, peptide binding is not required for activity. It is proposed that carbohydrate-modified vancomycin compounds are effective against resistant bacteria because they interact directly with bacterial proteins involved in the transglycosylation step of cell wall biosynthesis. These results suggest new strategies for designing glycopeptide antibiotics that overcome bacterial resistance.

Vancomycin is a glycopeptide antibiotic that kills bacterial cells by inhibiting peptidoglycan biosynthesis (1). It is the most important drug in current use for the treatment of Gram-positive bacterial infections, representing the final option for curing infections that are resistant to other antibiotics. The emergence of vancomycin-resistant bacterial strains is a very serious public health problem. Recently, a set of carbohydrate derivatives of vancomycin that are active against resistant bacterial strains was discovered (2). We now show that the modified carbohydrates alone are specific inhibitors of the transglycosylation step of peptidoglycan biosynthesis. This finding changes the picture for how modified glycopeptides kill resistant bacteria.

Vancomycin functions by binding to the terminald-Ala-d-Ala dipeptide of bacterial cell wall precursors (Fig. 1), thereby impeding further processing of these intermediates into peptidoglycan (3,4). The vancomycin complex involves a set of complementary hydrogen bonds between the peptide portion of vancomycin and the d-Ala-d-Ala dipeptide (4). Walsh and co-workers have shown that vancomycin resistance arises when bacteria acquire the ability to substituted-Ala-d-Ala withd-Ala-d-Lac (5). This structural change results in the loss of a critical hydrogen bond between the binding pocket of vancomycin and the peptide substrate. The binding affinity of vancomycin for the d-Ala-d-Lac substrate decreases by three orders of magnitude, with a concomitant loss of biological activity.

Figure 1

Proposed pathway for peptidoglycan synthesis in Gram-negative bacteria. Lipid II is assembled in the cytoplasm and then transported through the membrane where it is polymerized by transglycosylases to form immature peptidoglycan and then cross-linked by transpeptidases.

The complexity of the peptide portion of vancomycin makes it virtually impossible to reengineer the peptide backbone to include new contacts to the modified substrate (6). However, glycopeptides containing hydrophobic substituents on the vancosamine nitrogen (such as compounds 1 and 2, Fig. 2 and Table 1) are very active against vancomycin-resistant strains (2, 7). Although it is believed that these carbohydrate derivatives must bind the dipeptide termini of cell wall precursors to function (8–10), they do not show increased affinity ford-Ala-d-Lac (9); therefore, it is not obvious how they overcome resistance. It has been proposed that the hydrophobic substituents compensate for the decreased affinity ofd-Ala-d-Lac binding by facilitating dimerization (9, 11) and by anchoring the glycopeptides to the bacterial surface in close proximity to the cell wall precursors (9, 10). Although some elegant experiments support the hypothesis that membrane-anchoring and dimerization can significantly increased-Ala-d-Ala binding avidity (10,12), there is little evidence that these phenomena enhance binding to d-Ala-d-Lac sufficiently to explain the biological activity of the substituted glycopeptides against resistant bacteria.

Figure 2

Structures of vancomycin and compounds 1through 8.

Table 1

Twenty-two–hour minimum inhibitory concentrations (MICs) (μg/ml) for selected bacterial strains (Enterococcus faecium and Enterococcus faecalis). MICs were determined against strains grown in brain-heart infusion broth in a microdilution format according to NCCLS guidelines (23). The inoculum was 5 to 10 times higher than the recommended 3 × 105 to 7 × 105 colony-forming units per milliliter. The MIC is defined as the lowest antibiotic concentration that resulted in no visible growth after incubation at 35°C for 22 hours.

View this table:

We compared vancomycin and carbohydrate-modified glycopeptides1 and 2 in an assay designed to establish how different compounds inhibit peptidoglycan biosynthesis. Peptidoglycan biosynthesis takes place in three stages (Fig. 1) (1). The first two stages occur inside the bacterial cell and involve the assembly of a lipid-linked GlcNAc-MurNAc-pentapeptide (Lipid II). The third stage, which takes place on the exterior surface of the bacterial membrane, involves the polymerization of the GlcNAc-MurNAc disaccharide by transglycosylases and the cross-linking of the peptide side chains by transpeptidases. In the Gram-negative bacterium Escherichia coli, polymerization and cross-linking take place sequentially, producing immature (uncross-linked) peptidoglycan first and then mature (cross-linked) peptidoglycan (13). Because of the sequential nature of the polymerization and cross-linking reactions, it is possible to use permeabilizedE. coli strains to determine at which step peptidoglycan biosynthesis is blocked (14) simply by monitoring how much radioactivity is incorporated into Lipid II, immature, and mature peptidoglycan after treatment of the cells with [14C]UDP-GlcNAc and [14C]UDP-MurNAc-pentapeptide in the presence of increasing concentrations of an inhibitor. As shown in Fig. 3A, compounds such as ramoplanin that inhibit the formation of Lipid II cause a decrease in all the products (15); compounds such as bambermycin that inhibit transglycosylase activity cause a decrease in both immature and mature peptidoglycan (13); and compounds such as cefoxitin that inhibit transpeptidase activity cause a decrease in the formation of mature peptidoglycan (16).

Figure 3

(A) Inhibition patterns of three reference compounds with known mechanisms of action in the site of inhibition assay (13, 24, 25). (B) Inhibition patterns of vancomycin and compounds1, 2, 6, 7, and8 in the site of inhibition assay. PG, peptidoglycan.

Marked differences were observed in the inhibition patterns obtained for vancomycin and carbohydrate derivatives 1 and 2(Fig. 3B). Vancomycin causes a decrease in mature peptidoglycan and an increase in immature peptidoglycan, consistent with a mechanism of action in which the primary site of inhibition is the transpeptidation step. The carbohydrate derivatives, in contrast, cause a decrease in both immature and mature peptidoglycan, and an increase in the lipid intermediates, consistent with transglycosylase inhibition. Hence, the carbohydrate derivatives 1 and2 block a different step in peptidoglycan biosynthesis than does vancomycin.

We have considered two possible explanations for the above result. One explanation is that the hydrophobic substituent on the vancosamine sugar helps anchor the glycopeptide to the cell membrane, as proposed previously (8, 10). In vitro assays have shown that vancomycin itself can block either transglycosylation or transpeptidation depending on whether Lipid II or immature peptidoglycan is bound (3, 17). If 1and 2 are anchored to the cell membrane, they might preferentially bind membrane-bound Lipid II over immature peptidoglycan (Fig. 1), thus blocking transglycosylation more effectively than vancomycin itself, which binds to all cell wall precursors containing exposed d-Ala-d-Ala dipeptides. An alternative explanation for the change in the inhibition pattern is that the carbohydrate-modified glycopeptides interact with a target other thand-Ala-d-Ala, such as the enzymes involved in transglycosylation. If so, dipeptide binding might not be required for activity.

To evaluate the role of peptide binding in activity, we prepared des-leucyl vancomycin 3 (18) and its chlorobiphenyl derivative 4. Des-leucyl vancomycin3 does not bind d-Ala-d-Ala (19) and has no activity against either vancomycin-sensitive or -resistant strains (2); in contrast, chlorobiphenyl derivative 4 has good activity against both sensitive and resistant bacterial strains (Table 1). Derivative 4 is almost as active as 1 and 2 against resistant bacterial strains even though it is missing an important portion of the dipeptide-binding pocket. Although the lower minimum inhibitory concentrations (MICs) of 1 and 2 compared with4 in d-Ala-d-Ala–producing microorganisms indicate that dipeptide binding enhances antibiotic activity, it is not essential.

The preceding results imply that the modified disaccharide is a key determinant of biological activity, with a mechanism of action independent of peptide binding. To evaluate the activity of the carbohydrate portion alone, we synthesized disaccharide 5and the substituted analog 6 (20). Disaccharide5 was completely inactive, but 6 had antibiotic activity (Table 1). Indeed, derivative 6 was more than 10 times as effective as vancomycin itself against vancomycin-resistant strains. Moreover, 6 specifically inhibited the incorporation of radiolabeled precursors into peptidoglycan but not into DNA or protein, consistent with a mechanism of action involving inhibition of peptidoglycan biosynthesis (21). In addition, in the site of inhibition assay6 was found to inhibit the transglycosylation step (Fig. 3B). These results suggest a direct interaction between the substituted carbohydrates and proteins involved in transglycosylation.

On the basis of the results presented above, we propose that the modified vancomycin analogs 1 and 2 have a complex mechanism of action involving specific interactions with at least three targets: immature peptidoglycan, Lipid II, and proteins involved in transglycosylation. In vancomycin-sensitive strains, the aglycon binds the d-Ala-d-Ala dipeptides of Lipid II and immature peptidoglycan; the modified carbohydrates can also interact with one or more proteins involved in transglycosylation. In resistant strains where the aglycon does not effectively bindd-Ala-d-Lac, the modified carbohydrates can still inhibit bacterial transglycosylases.

The hypothesis that peptide binding is not required for activity against resistant bacterial strains suggests new strategies for designing better glycopeptide antibiotics. For example, instead of trying to increase binding to d- Ala-d-Lac, it might be simpler to optimize the carbohydrate portion of vancomycin for transglycosylase inhibition. We have recently developed chemistry for glycosylating the pseudoaglycon of vancomycin (20). This chemistry allows us to replace the vancosamine sugar of vancomycin with other sugars. Because we have no information about how specific the requirements are for carbohydrate recognition, we decided to first replace vancosamine with its des-methyl analog, daunosamine. After verifying that the des-methyl disaccharide 7 inhibits transglycosylase activity (Fig. 3B), which shows that the methyl group does not play an essential role in the recognition process, we attached daunosamine to the vancomycin pseudoaglycon to make 8. Although 8 differs from1 only in a single methyl group on the terminal sugar, it shows some notable differences in activity. The activity against methicillin-resistant staphylococcal strains is particularly good (Table 2). This is only the first attempt to explore unnatural sugars by chemical glycosylation, and it is quite likely that a wide-ranging investigation of different sugars will lead to more significant improvements across a range of bacterial strains. Because we have also developed chemistry to attach both sugars sequentially to the vancomycin aglycon (22), it is now possible to explore the effects of replacing either or both sugars.

Table 2

Twenty-two–hour MICs (μg/ml) of vancomycin and compounds 1 and 8 for selected staphylococcal strains. MSSA: methicillin-sensitive Staphylococcus; MRSA: methicillin-resistantStaphylococcus.

View this table:

The finding that peptide binding is not required for the biological activity of carbohydrate-modified glycopeptides should change the way scientists think about how to overcome vancomycin resistance. The challenge now is to determine the specific molecular targets of the substituted carbohydrates.

  • * To whom correspondence should be addressed. E-mail: dkahne{at}


View Abstract

Navigate This Article