Moderate-Level Resistance to Glycopeptide LY333328 Mediated by Genes of the vanA and vanB Clusters in Enterococci

Abstract

Three of five natural plasmids carrying a wild-type vanA gene cluster did not confer LY333328 glycopeptide resistance on Enterococcus faecalis JH2-2 (MIC = 2 μg/ml). The two remaining plasmids conferred resistance to the drug (MIC, 8 μg/ml). The vanB gene cluster did not confer resistance to LY333328, since this antibiotic was not an inducer. Mutations in the vanS_B sensor gene that allowed induction by teicoplanin or constitutive expression of the vanB cluster led to LY333328 resistance (MIC, 8 to 16 μg/ml). Overproduction of the VanH, VanA, and VanX proteins for d-alanyl-d-lactate (d-Ala-d-Lac) synthesis and d-Ala-d-Ala hydrolysis was sufficient for resistance to LY333328 (MIC, 16 μg/ml). Mutations in the host d-Ala:d-Ala ligase contributed to LY333328 resistance in certain VanA- and VanB-type strains, but the MICs of the antibiotic did not exceed 16 μg/ml. Addition of d-2-hydroxybutyrate in the culture medium of mutants that did not produce the VanH d-lactate dehydrogenase led to incorporation of this d-2-hydroxy acid at the C-terminal ends of the peptidoglycan precursors and to LY333328 resistance (MIC, 64 μg/ml). The vanZ gene of the vanA cluster conferred resistance to LY333328 (MIC, 8 μg/ml) by an unknown mechanism. These data indicate that VanA- and VanB-type enterococci may acquire moderate-level resistance to LY333328 (MIC ≤ 16 μg/ml) in a single step by various mechanisms.

Acquired resistance to glycopeptides in enterococci is due to production of peptidoglycan precursors ending in the depsipeptide d-alanyl-d-lactate (d-Ala-d-Lac) instead of the dipeptide d-Ala-d-Ala in susceptible bacteria (11, 35). The substitution prevents the formation of complexes between glycopeptides and peptidoglycan precursors that are responsible for the inhibition of cell wall synthesis (14). Resistance by this mechanism is mediated by two types of gene clusters, vanA and vanB, which encode related proteins (21). Gene clusters related to vanA confer high-level resistance to vancomycin and teicoplanin (3). In contrast, enterococci belonging to the vanB hybridization class remain susceptible to teicoplanin because the VanS_B sensor kinase does not trigger induction of the resistance genes in response to this antibiotic (13, 21). Coproduction of peptidoglycan precursors ending in d-Ala and d-Lac leads to low-level vancomycin resistance in certain VanB-type strains that express the resistance genes at a low level (7).

LY333328 is a semisynthetic N-alkylated glycopeptide active on VanA- and VanB-type enterococci (16, 28). The mechanisms of action of LY333328 and the structurally related LY191145 are believed to be identical (1, 2). These two glycopeptides differ in that LY333328 has a chlorobiphenyl N-alkyl substitution on the vancosamine sugar, whereas LY191145 has a chlorophenyl moiety (19). In solution, the affinities of LY191145 and vancomycin for peptidoglycan precursor analogues with a C-terminal d-Lac residue are similarly low (2). However, the interaction of LY191145 with d-Lac-ending precursors may be significantly increased in vivo, since the drug binds to the membrane and dimerizes (2, 15).

This study was undertaken to assess the risk of emergence of resistance to LY333328 in VanA- and VanB-type enterococci. The activity of LY333328 against a collection of glycopeptide-resistant strains constructed in vitro that overproduce the resistance proteins or harbor mutations in the vanS_B sensor and host d-Ala:d-Ala ligase genes was studied.

MATERIALS AND METHODS

Strains and growth conditions.

The origins and properties of the strains are described in Table 1. Enterococcus faecalis JH2-2 and BM4110 are derivatives of strain JH2 that are resistant to fusidic acid and rifampin or to streptomycin, respectively. Strains were grown in brain heart infusion (BHI) broth and agar (Difco Laboratories, Detroit, Mich.) at 37°C. Vancomycin (10 μg/ml) was added to the medium for growth of glycopeptide-dependent strains. The MICs of LY333328 (Eli Lilly & Co., Saint-Cloud, France), vancomycin (Eli Lilly & Co.), and teicoplanin (Hoechst-Marion-Roussel, Levallois-Perret, France) were determined by the method of Steers et al. (33) with 10⁵ CFU per spot on BHI agar after 24 h of incubation at 37°C. Since there are as yet no formally recognized breakpoints for LY333328, resistance is defined in this work as an increase in the MIC of the glycopeptide in comparison to that for the glycopeptide-susceptible host E. faecalis JH2-2. Glycolic acid, d,l-lactate lithium salt, d,l-2-hydroxybutyrate sodium salt, and d,l-2-hydroxyvalerate sodium salt (Sigma, Saint-Quentin Falavier, France) were added to the medium as 1 M neutralized (pH 7.0) solutions. Economic constraints imposed the use of d,l-2-hydroxy acids, although the VanA ligase uses exclusively d-2-hydroxy acids as substrates (18). For selection of spontaneous mutants, strains were grown overnight in broth and concentrated by centrifugation, and ca. 5 × 10⁹ CFU was plated on agar containing appropriate antibiotics.

TABLE 1.

MICs of LY333328, vancomycin, and teicoplanin against various VanA- and VanB-type strains

Strain and characteristics	MIC (μg/ml) of:			Source or reference
	LYa	VMb	TEc
E. faecalis JH2-2	2	2	1	25
E. faecalis BM4110	1	2	0.5	20
Group A1: E. faecium clinical isolates harboring Tn1546 or related elements on natural plasmids
BM4147(pIP816)	2	1,024	512	26
HM1071(pIP837)	4	1,024	512	5
HM1072(pIP838)	2	1,024	512	5
HM1073(pIP839)	4	512	64	5
HM1074(pIP840)	4	1,024	512	5
Group A2: E. faecalis harboring Tn1546 or related elements on natural plasmids
BM4110/pIP816-1	8	1,024	512	5
JH2-2/pIP837	8	512	64	5
JH2-2/pIP838	2	256	64	5
JH2-2/pIP839	2	64	8	5
JH2-2/pIP840	2	256	64	5
Group A3: E. faecalis JH2-2 with vanRSHAX of Tn1546 at various copy numbers per chromosome
BM4299 (1 copy)	2	4	1	7
BM4300 (1 copy)	2	8	1	7
BM4301 (2 copies)	2	32	2	7
BM4302 (5 copies)	4	256	4	7
JH2-2/pAT80 (20 copies)	16	2,048	128	10
Group A4: Glycopeptide-dependent derivative of E. faecalis JH2-2/pIP837
JH2-2/pIP837	8	512	64	5
VmD TeD derivative of JH2-2/pIP837d	16	2,048	256	This study
Group A5: E. faecalis JH2-2 harboring combinations of genes from Tn1546 in multicopy plasmid pAT29
pAT80 (vanRSHAX)	16	2,048	128	10
pAT382 (vanRSHAXY)	8	1,024	32	8
pAT398 (vanRSHAXYZ)	8	1,024	128	6
pAT393 (P2vanY)	2	2	1	8
pAT396 (P2vanZ)	8	2	8	6
Group B1: E. faecalis BM4110 harboring transposon Tn1547 on pIP964 and mutations in ddl and vanSBe
BM4305 VanB (i)	1	1,024	0.5	30
BM4320 VmD Ddl(S319-I) (i)	NA	2,048	NA	13
BM4322 VmD TeD Ddl(S319-I) VanSB(A30-G) (i)	16	2,048	2,048	13
BM4344 Vmr Ter VanSB(A30-G) (i)	2	1,024	1,024	13
BM4346 Vmr Ter VanSB(D168-Y) (i)	16	1,024	2,048	13
BM4347 Vmr Ter VanSB(T237-K) (c)	8	1,024	2,048	13
BM4348 VmHet TeHet VanSB(Y426-ter) (i)	8	512	2,048	13
BM4323 Vmr Ter Ddl(S319-I) VanSB(Y426-ter) (c)	16	1,024	2,048	13

^aLY, LY333328; NA, not applicable, since the strain did not grow in the absence of vancomycin. 

^bVM, vancomycin. 

^cTE, teicoplanin. 

^dD, dependent. 

^ec, constitutive; ^Het, heterogeneous; i, inducible; ter, termination. 

Analysis of peptidoglycan precursors.

Extraction and analysis of peptidoglycan precursors was performed essentially as described previously (27). Enterococci were grown in BHI broth (120 ml) supplemented with 0.5% yeast extract, vancomycin, and d,l-2-hydroxy acids at the concentrations specified in Table 2. At mid-exponential phase (optical density at 600 nm = 1), ramoplanin (3 μg/ml) was added, and incubation was continued for 20 min. Bacteria were harvested by centrifugation at 12,000 × g for 2 min at 4°C, resuspended in water, and treated with 7% trichloroacetic acid for 15 min at 0°C in a final volume of 2 ml. The extract was centrifuged (at 48,000 × g for 1 min at 4°C), and the supernatant fraction containing the cytoplasmic peptidoglycan precursors was collected. Trichloroacetic acid was neutralized by the addition of solid sodium bicarbonate, and salt was removed from the extract by gel filtration of 0.5-ml samples on a Sephadex G-10 column (28 by 1 cm). The eluate was monitored at 252 nm, and cell wall precursors eluted immediately after the void volume of the column. Twenty-microliter samples of the fraction containing the cell wall precursors were analyzed by high-performance liquid chromatography on a C₁₈ reverse-phase column with 0.05 M ammonium acetate (pH 5.3) as the eluant at a flow rate of 0.8 ml per min and with the application of two consecutive methanol gradients in the same buffer (0 to 2.5% between 5 and 45 min; 2.5 to 7.5% between 45 and 60 min). Elution was continued for a further 15 min. The elution times for the precursors were as follows: for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys (UDP-MurNAc-tripeptide), 6 min; for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys-d-Ala-d-Ala (UDP-MurNAc-tetra-d-Ala), 22 min; for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys-d-Ala-d-Lac (UDP-MurNAc-tetra-d-Lac), 39 min; for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys-d-Ala-glycolic acid (UDP-MurNAc-tetra-Gly a.), 28 min; for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys-d-Ala-d-2-hydroxybutyrate (UDP-MurNAc-tetra-d-HBut), 57 min; and for UDP-N-acetylmuramyl-l-Ala-γ-d-Glu-l-Lys-d-Ala-d-2-hydroxyvalerate (UDP-MurNAc-tetra-d-HVal), 68 min. The relative proportions of these peptidoglycan precursors were determined from the integrated peak areas, and the results were expressed as percentages. UDP-MurNAc-tetrapeptide was not detected in significant amounts, since the strains analyzed did not produce the VanY d,d-carboxypeptidase.

TABLE 2.

Cytoplasmic cell wall precursors of E. faecalis JH2-2 and derivatives

E. faecalis strainc	Hydroxy acid in culture medium (concn, mM)d	% Cytoplasmic peptidoglycan precursora						Glycopeptide MIC (μg/ml)b
		Tripeptide	Pentapeptide	Pentadepsipeptide ending in:				LY	VM	TE
				Lac	HBut	HVal	Glycolic acid
JH2-2	No addition	<2	100	<2	<2	<2	<2	2	2	1
JH2-2/pAT83	No addition	8	19	73	<2	<2	<2	2	4	1
	d,l-Lactate (50)	<2	<2	100	<2	<2	<2	8	2,048	4
	d,l-2-Hydroxybutyrate (50)	<2	<2	4	96	<2	<2	8	512	2
	d,l-2-Hydroxyvalerate (10)	<2	6	58	<2	36	<2	2	16	1
	Glycolic acid (50)	<2	18	33	<2	<2	49	1	4	1
JH2-2/pAT83 m1e	No addition	68	<2	32	<2	<2	<2	4	2,048	4
	d,l-Lactate (50)	41	<2	59	<2	<2	<2	32	4,096	128
	d,l-2-Hydroxybutyrate (50)	44	<2	4	52	<2	<2	64	>4,096	1,024
	d,l-2-Hydroxyvalerate (10)	52	<2	33	<2	15	<2	16	2,048	256
JH2-2/pAT83 m2	No addition	12	<2	88	<2	<2	<2	4	32	1
	d,l-Lactate (50)	<2	<2	100	<2	<2	<2	16	4,096	64
	d,l-2-Hydroxybutyrate (50)	<2	<2	23	77	<2	<2	64	>4,096	256
	d,l-2-Hydroxyvalerate (10)	<2	<2	87	<2	13	<2	16	4,096	32
	Glycolic acid (50)	<2	<2	46	<2	<2	54	8	2,048	4

^aTripeptide, UDP-MurNAc-tripeptide; pentapeptide, UDP-MurNAc-tetra-d-Ala; pentadepsipeptide, UDP-MurNAc-tetra-d-2-hydroxy acid; Lac, d-lactate; HBut, d-2-hydroxybutyrate; HVal, d-2-hydroxyvalerate. 

^bLY, LY333328; TE, teicoplanin; VM, vancomycin. 

^cPlasmid pAT83 harbors the vanRSHAX gene cluster with vanH inactivated by insertion of the aphA-1 kanamycin resistance gene. Spontaneous mutants of E. faecalis JH2-2/pAT83 were selected on medium containing 1 mM d,l-hydroxyvalerate and 8 μg of vancomycin/ml (mutant m1) or 50 mM glycolic acid and 64 μg of vancomycin/ml (mutant m2). 

^dFor analysis of peptidoglycan precursors, the medium also contained 4 μg of vancomycin/ml, except for JH2-2/pAT83 grown in the absence of d,l-2-hydroxy acid (vancomycin, 0.5 μg/ml) and JH2-2 (no vancomycin added). 

^eGlycolic acid was not tested because JH2-2/pAT83 m1 did not grow in 24 h in the presence of this compound at 50 mM. 

RESULTS AND DISCUSSION

Activities of glycopeptides against wild VanA-type strains.

The five clinical isolates of Enterococcus faecium harboring wild-type vanA gene clusters on natural plasmids were susceptible to LY333328 (MIC ≤ 4 μg/ml) (group A1 in Table 1). Three of the five corresponding plasmids did not confer resistance to LY333328 on E. faecalis JH2-2 (group A2). The MIC of LY333328 against strains harboring the two remaining plasmids was 8 μg/ml. Thus, the glycopeptide remained active or partially active in spite of high-level resistance to vancomycin and teicoplanin as reported elsewhere (16, 28). It is worth noting that the MICs of LY333328 for JH2-2, BM4281, and BM4316 are lower in broth (32).

Effect of coproduction of d-Ala- and d-Lac-ending peptidoglycan precursors on the activities of glycopeptides.

We previously showed that constitutive high-level production of the VanH dehydrogenase, VanA d-Ala:d-Lac ligase, and VanX d,d-dipeptidase encoded by the vanRSHAX gene cluster present on a multicopy plasmid was necessary and sufficient for high-level vancomycin and teicoplanin resistance (JH2-2/pAT80; group A3 in Table 1) (7). In this strain, production of peptidoglycan precursors ending in d-Ala was not detectable (≤2%) (7). Vancomycin and teicoplanin remained partially active against strains harboring one to five chromosomal copies of the vanRSHAX gene cluster, since these strains coproduced d-Ala- and d-Lac-ending precursors (7). The increase in the copy number of the resistance genes (from 1 to 5) was associated with an increase in the relative proportion of d-Lac-ending precursors (from 51 to 96%) (7). The vancomycin resistance level increased as a function of increased replacement of d-Ala by d-Lac at the C-terminal ends of peptidoglycan precursors (7). Resistance to teicoplanin required more complete elimination of d-Ala-ending precursors (7).

Qualitatively, the results obtained for teicoplanin and LY333328 were similar, since increases in the MICs of the drugs were detectable only for strains harboring 5 or 20 copies of the vanRSHAX gene cluster per chromosome (group A3 in Table 1). Quantitatively, high-level expression of the resistance genes in JH2-2/pAT80 conferred lower-level resistance to LY333328 than to teicoplanin.

Although incorporation of d-Ala-d-Ala into the peptidoglycan precursors of JH2-2/pAT80 is not detectable (7), this observation does not exclude the possibility that binding of LY333328 to minute amounts of d-Ala-ending precursors could be responsible for the residual activity. In order to explore this possibility, we screened for mutants with increased resistance to glycopeptides that could potentially have reduced d-Ala:d-Ala ligase activity (13, 22, 31, 34). Mutants derived from JH2-2/pAT80 were not obtained on agar containing 16 μg of LY333328/ml. A vancomycin-dependent derivative of JH2-2/pIP837 harboring a wild-type vanA gene cluster on a natural plasmid was obtained on agar containing 1,024 μg of vancomycin/ml (group A4 in Table 1). The MICs of vancomycin, teicoplanin, and LY333328 for the mutant were increased two- to fourfold over those against the parental strain.

Taken together, these data indicate that extensive replacement of d-Ala-ending precursors by d-Lac-ending precursors can be responsible for an eightfold increase in the MIC of LY333328 (from 2 to 16 μg/ml). No further increase in the level of LY333328 resistance was obtained by this mechanism. This observation is in agreement with the proposal that binding of LY333328 to d-Lac-ending precursors could be responsible for the inhibition of peptidoglycan synthesis (1, 2).

Role of the VanY and VanZ accessory proteins in LY333328 resistance.

Previous analyses indicated that the VanY d,d-carboxypeptidase and the VanZ protein contribute to vancomycin and teicoplanin resistance in a host that coproduces peptidoglycan precursors ending in d-Ala and d-Lac (4, 6, 7). Introduction of vanY or of both vanY and vanZ in pAT80 (vanRSHAX) did not result in an increase in the level of LY333328 resistance (group A5 in Table 1).

Expression of vanZ alone under the control of the heterologous P₂ promoter led to four- and eightfold increases in the MICs of LY333328 and teicoplanin, respectively (group A5 in Table 1). These results indicate that resistance to LY333328 can be due to two mechanisms, as previously found for teicoplanin (6, 7). The first involves synthesis of d-Ala-d-Lac by the VanH dehydrogenase and the VanA ligase and elimination of d-Ala-ending precursors by the VanX and VanY d,d-peptidases. The second mechanism does not involve an alteration of the assembly pathway of UDP-MurNAc-tetra-d-Ala (6) and requires only the VanZ protein. Independently, the two mechanisms conferred similar levels of resistance to LY333328. Combination of both mechanisms did not increase the MIC of the glycopeptide.

Activity of LY333328 against a VanB-type E. faecalis strain and teicoplanin-resistant or glycopeptide-dependent derivatives.

E. faecalis BM4305 harbors transposon Tn1547 on plasmid pIP964. The wild-type copy of the vanB cluster located in Tn1547 did not confer resistance to LY333328 (group B1 in Table 1). A vancomycin-dependent mutant of BM4305 with the ddl(S₃₁₉-I) mutation in the host d-Ala:d-Ala ligase gene did not grow in the presence of LY333328 or teicoplanin, indicating that these glycopeptides did not induce expression of the vanB d-Ala:d-Lac ligase gene (see reference 13 for a discussion). The double mutant BM4322 [vanS_B(A₃₀-G) ddl(S₃₁₉-I)] grew in the presence of vancomycin, teicoplanin, and LY333328. Thus, the A₃₀-G substitution in VanS_B allowed induction by teicoplanin and LY333328. The vanS_B(A₃₀-G) mutation alone led only to a twofold increase in the MIC of LY333328 in comparison to wild-type BM4305 (VanB). In contrast, the vanS_B(D₁₆₈-Y) and vanS_B(T₂₃₇-K) mutations were sufficient for higher-level resistance to LY333328 (MIC, 16 and 8 μg/ml, respectively). These mutations were previously shown to lead to inducible [vanS_B(D₁₆₈-Y)] or constitutive [vanS_B(T₂₃₇-K)] expression of resistance to vancomycin and teicoplanin (13). Likewise, the vanS_B(Y₄₂₆-ter) mutation responsible for the emergence of heterogeneous resistance to vancomycin and teicoplanin conferred resistance to LY333328 (MIC = 8 μg/ml). Combination of the vanS_B(Y₄₂₆-ter) and ddl(S₃₁₉-I) mutations in BM4323 led to a further twofold increase in the MIC of LY333328. These results indicate that VanB-type strains may simultaneously acquire resistance to teicoplanin and LY333328 in a single step. The emergence of resistance to these glycopeptides can be due to three alterations of the VanS_B sensor that lead to inducible, constitutive, or heterogeneous expression of the resistance genes. The MIC of LY333328 did not exceed 16 μg/ml for any of the mutants studied, even though certain mutants harbored combinations of mutations in vanS_B and ddl. Quantitatively, the level of resistance against LY333328 was significantly lower than that against teicoplanin.

Incorporation of various d-2-hydroxy acids at the C-terminal extremity of peptidoglycan precursors of JH2-2/pAT83.

Insertional inactivation of the vanH dehydrogenase gene of pAT80 (vanRSHAX) was reported to lead to glycopeptide susceptibility (10). The addition of d-Lac in the culture medium restores the glycopeptide resistance of the resulting vanH null mutant JH2-2/pAT83 (vanRSHΩaphA-1AX) (9). This indicates that d-Lac penetrates into the cytoplasm and is incorporated in the depsipeptide d-Ala-d-Lac by the VanA ligase (9). The addition of various d-2-hydroxy acids was tested in this system to determine whether incorporation of homologues of d-Lac into the peptidoglycan precursors could confer high-level resistance to LY333328 (Table 2). The d-2-hydroxy acids had various side chains, including a hydrogen atom (−H) for glycolic acid, a methyl group (−CH₃) for d-lactate, an ethyl group (−CH₂-CH₃) for d-2-hydroxybutyrate, and an isopropyl group [−CH(CH₃)₂] for d-2-hydroxyvalerate. These d-2-hydroxy acids were previously shown to be substrates in vitro for the VanA ligase, with the exception of glycolic acid, which was not tested (18). Replacement of d-Lac by the d-2-hydroxy acids could potentially reduce the binding of glycopeptides to peptidoglycan precursors by steric hindrance (d-2-hydroxybutyrate and d-2-hydroxyvalerate) or by suppressing the hydrophobic interaction with the methyl side chain of d-Lac (glycolic acid) (17, 18, 29).

Incorporation of d-Lac into the precursors of JH2-2/pAT83 was detected even in the absence of supplementation of the culture medium (Table 2). However, d-Lac was not present in sufficient amounts to compete efficiently with d-Ala, thus leading to the coproduction of d-Ala- and d-Lac-ending precursors. The MIC of vancomycin was increased only twofold in comparison to that for JH2-2 (Table 2). Supplementation of the culture medium of JH2-2/pAT83 with d-Lac and d-2-hydroxybutyrate resulted in high-level resistance to vancomycin and moderate-level resistance to teicoplanin and LY333328. Peptidoglycan precursors ending in d-Ala were not synthesized at a detectable level (<2%). The d-2-hydroxybutyrate added to the culture medium was efficiently used by VanA, since 96% of the peptidoglycan precursors contained this d-2-hydroxy acid. Incorporation of d-2-hydroxyvalerate and glycolic acid was less efficient (36 and 49% of the precursors, respectively). Supplementation of the medium with d-2-hydroxyvalerate reduced but did not abolish the incorporation of d-Ala into the precursors (from 19 to 6%). This was associated with low-level resistance to vancomycin and susceptibility to teicoplanin and LY333328. Similar amounts of peptidoglycan precursors ending in d-Ala were detected in the presence (18%) or absence (19%) of glycolic acid. Accordingly, supplementation of the culture medium with this acid did not lead to glycopeptide resistance.

Selection of m1 and m2 mutants.

The d-2-hydroxy acids added to the culture medium of JH2-2/pAT83 competed with the endogenous sources of d-Ala and d-Lac, leading to coproduction of various peptidoglycan precursors (Table 2). In order to improve incorporation of the d-2-hydroxy acids, mutants of E. faecalis JH2-2/pAT83 were selected on agar containing 1 mM d,l-hydroxyvalerate and 8 μg of vancomycin/ml (mutant m1) or 50 mM glycolic acid and 64 μg of vancomycin/ml (mutant m2). Incorporation of d-Ala in the peptidoglycan precursors of mutants m1 and m2 was not detectable even in the absence of supplementation (Table 2). As expected, these mutants expressed vancomycin resistance in the medium devoid of d-2-hydroxy acid. The mutations in m1 and m2 did not allow increased incorporation of d-2-hydroxyvalerate, d-2-hydroxybutyrate, or glycolic acid to the detriment of d-Lac. Further attempts to obtain mutants that expressed resistance only in media containing d-2-hydroxybutyrate, d-2-hydroxyvalerate, or glycolic acid were unsuccessful (data not shown). Thus, mutations that result specifically in increased incorporation of d-2-hydroxy acids added to the culture medium were not obtained.

Characteristics of mutants m1 and m2.

In the absence of supplementation of the culture medium, JH2-2/pAT83 accumulated low amounts of UDP-MurNAc-tripeptide (Table 2). This observation suggests that the combined pool of d-Ala-d-Ala and d-Ala-d-Lac was limiting, probably because the dipeptide was hydrolyzed by the VanX d,d-dipeptidase and the concentration of d-Lac was too low for efficient synthesis of the depsipeptide. In agreement with this notion, the tripeptide was not detected if the medium was supplemented with the d-2-hydroxy acids. Synthesis of d-Ala-d-Ala may be impaired in mutant m2; this would account for the absence of precursors ending in d-Ala and the accumulation of the tripeptide only in the absence of supplementation. A distinct mutation was present in m1, since large amounts of the tripeptide were detected under all conditions.

Supplementation of the culture medium with d-2-hydroxybutyrate, d-2-hydroxyvalerate, or glycolic acid led to the incorporation of these d-2-hydroxy acids in the peptidoglycan precursors of mutants m1 and m2 and to increased resistance to vancomycin, teicoplanin, and LY333328 (Table 2). The levels of resistance to teicoplanin and LY333328 were higher for production of precursors ending in d-2-hydroxybutyrate than for production of the natural precursors terminating in d-Lac. Of note, the MIC of LY333328 obtained for mutants m1 and m2 grown in the presence of d-2-hydroxybutyrate corresponds to the highest MIC obtained in this study (64 μg/ml). These observations suggest that the binding of LY333328 to peptidoglycan precursors is reduced by steric hindrance if the methyl side chain of d-Lac is replaced by an ethyl group. Whether residual activity of LY333328 is actually due to binding of the drug to peptidoglycan precursors is not known. As discussed by Ge et al. (23), experiments support the hypothesis that membrane anchoring and dimerization can significantly increase d-Ala-d-Ala binding avidity, but there is little evidence that these features enhance binding to d-Ala-d-Lac sufficiently to explain the antibacterial activity of the substituted glycopeptide against resistant bacteria. These authors showed that vancomycin derivatives carrying a chlorobiphenyl substituent, such as LY333328, remain active even in the absence of an intact peptidoglycan precursor binding pocket. Moreover, the chlorobiphenyl-substituted carbohydrate moiety of the molecule was sufficient to inhibit transglycosylation and bacterial growth, albeit at high concentrations (on the order of 100 μg/ml). The finding that the peptide binding pocket is not essential for biological activity suggests that substituted glycopeptides have an additional mode of action. Our results indicate that the activity of LY333328 against VanA-type enterococci is partly due to binding to residual d-Ala-ending precursors and suggest that this glycopeptide might interact with d-Lac-ending precursors, since replacement of d-Lac by d-2-hydroxybutyrate increased the level of resistance. LY333328 might inhibit the growth of enterococci producing d-2-hydroxybutyrate-ending precursors by a different, unknown mechanism that appears to require relatively high concentrations of the drug (64 μg/ml [Table 2]), in agreement with the data presented by Ge et al. (23).

The VanA ligase has a broad substrate specificity (18). Detection of peptidoglycan precursors ending in d-2-hydroxybutyrate, d-2-hydroxyvalerate, and glycolic acid (Table 2) indicates that various depsipeptides synthesized by VanA can be added to UDP-MurNAc-tripeptide. Furthermore, the increase in glycopeptide resistance associated with production of these precursors strongly suggests that they can be used for peptidoglycan synthesis. Glycopeptide-resistant enterococci produce peptidoglycan precursors ending in d-Lac or d-Ser (17, 18). Further evolution of these resistance pathways leading to incorporation of other substituents in the C-terminal positions of peptidoglycan precursors could occur if the bacteria gain the capacity to synthesize these compounds.

Conclusions.

In agreement with previous studies (16, 28), acquisition of wild-type vanA and vanB gene clusters by E. faecalis had little or no effect on the in vitro activity of LY333328 (Table 1). However, VanA- and VanB-type enterococci may acquire resistance to this glycopeptide by various mechanisms. Production of precursors ending in d-Lac was sufficient for resistance to this glycopeptide if precursors terminating in d-Ala were completely eliminated (Table 1). This could be achieved by increased expression of the resistance genes or reduced production of d-Ala-d-Ala by the host ligase. Expression of the vanZ gene may also confer LY333328 resistance. Finally, cross-resistance to teicoplanin and LY333328 was acquired following various mutations in the vanS_B sensor gene of the vanB cluster. These observations indicate that emergence of LY333328 resistance should be anticipated, since mutations in ddl and in vanS_B are known to be selected under treatment and are present in natural populations of VanA- and VanB-type enterococci (12, 24, 31, 34). However, the levels of LY333328 resistance obtained by these mechanisms were moderate (MIC ≤ 16 μg/ml), probably because replacement of the C-terminal d-Lac residue of peptidoglycan precursors by another d-2-hydroxy acid is required for higher-level resistance to this glycopeptide. A high dosage of LY333328 or the combination of this glycopeptide with an aminoglycoside might be able to prevent the emergence of LY333328 resistance, since the LY333328-gentamicin combination is synergistic against vancomycin-resistant E. faecium (36) and the teicoplanin-gentamicin combination prevents the emergence of teicoplanin resistance in VanB-type enterococci (12).