Interplay among Membrane-Bound Lytic Transglycosylase D1, the CreBC Two-Component Regulatory System, the AmpNG-AmpD_I-NagZ-AmpR Regulatory Circuit, and L1/L2 β-Lactamase Expression in Stenotrophomonas maltophilia

Abstract

Lytic transglycosylases (LTs) are an important class of enzymes involved in peptidoglycan (PG) cleavage, with the concomitant formation of an intramolecular 1,6-anhydromuramoyl reaction product. There are six annotated LT genes in the Stenotrophomonas maltophilia genome, including genes for five membrane-bound LTs (mltA, mltB1, mltB2, mltD1, and mltD2) and a gene for soluble LT (slt). Six LTs of S. maltophilia KJ were systematically mutated, yielding the ΔmltA, ΔmltB1, ΔmltB2, ΔmltD1, ΔmltD2, and Δslt mutants. Inactivation of mltD1 conferred a phenotype of elevated uninduced β-lactamase activity. The underlying mechanism responsible for this phenotype was elucidated by the construction of several mutants and determination of β-lactamase activity. The expression of the genes assayed was assessed by quantitative reverse transcriptase PCR and a promoter transcription fusion assay. The results demonstrate that ΔmltD1 mutant-mediated L1/L2 β-lactamase expression involved the creBC two-component regulatory system (TCS) and the ampNG-ampD_I-nagZ-ampR regulatory circuit. The inactivation of mltD1 resulted in mltB1 and mltD2 upexpression in a creBC- and ampNG-dependent manner. The overexpressed MltB1 and MltD2 activity contributed to the expression of the L1/L2 β-lactamase genes via the ampNG-ampD_I-nagZ-ampR regulatory circuit. These findings reveal, for the first time, a linkage between LTs, the CreBC TCS, the ampNG-ampD_I-nagZ-ampR regulatory circuit, and L1/L2 β-lactamase expression in S. maltophilia.

INTRODUCTION

Peptidoglycan (PG) is an important component used to maintain the shape and structural integrity of bacteria. PG consists of repeated disaccharide subunits of N-acetylglucosamine and N-acetylmuramic acid with pentapeptides. These polysaccharide strands are cross-linked by peptide bridges into a rigid three-dimensional network (1). Enlargement and growth of the PG sacculus are achieved by a concerted action of both synthetic and lytic enzymes. Penicillin binding proteins (PBPs) and lytic transglycosylases (LTs) are such kinds of important enzymes involved in the coordinated balance. PBPs, which catalyze the transglycosylation and/or transpeptidation reaction, are responsible for peptidoglycan synthesis (2). LTs act on peptidoglycan by cleaving the glycosidic bond between N-acetylmuramoyl and N-acetylglucosaminyl residues, with the concomitant formation of a GlcNAc-1,6-anhydro-MurNAc peptide (3). In view of the vital role of PG for bacterial survival, bacteria possess an array of stress response systems to sense the defects of PG integrity, for example, the BlrAB system in Aeromonas hydrophila and the Rcs phosphorelay system in Escherichia coli (4, 5).

PG recycling has been shown to link to chromosomal β-lactamase induction in some ampR-ampC module-bearing Gram-negative bacteria, such as Enterobacter cloacae, Citrobacter freundii, and Pseudomonas aeruginosa (6). During the normal growth of members of the family Enterobacteriaceae, the PG fragments released are marked by a 1,6-anhydro ring structure of the terminal muramic acid due to the action of LTs (3). The GlcNAc-1,6-anhydro-MurNAc peptide is transported into the cytoplasm by AmpG permease (7). Within the cytoplasm, the GlcNAc-1,6-anhydro-MurNAc peptide is hydrolyzed by NagZ (β-N-acetylglucosaminidase), yielding 1,6-anhydro muropeptides, which function as the AmpR activator ligand (AL) for the derepression of the chromosomal ampC gene (8). The 1,6-anhydro muropeptides can be processed by AmpD (N-acetylanhydromuramyl-l-alanine amidase) and further recycled into UDP-MurNAc pentapeptide, which functions as the precursor of peptidoglycan synthesis, or the AmpR repressor ligand (RL), which represses chromosomal ampC expression (6). β-Lactam-mediated cellular stress or mutations in ampR itself are required to convert AmpR from its repressor to its activator form. In addition, the CreBC (BlrAB) two-component system (TCS) has also been reported to be associated with β-lactam resistance in P. aeruginosa (9). PBP 4 (encoded by dacB) inactivated by particular β-lactams (e.g., cefoxitin) or mutations in dacB itself increases the level of resistance to β-lactams in a CreBC (BlrAB)- and AmpR-dependent manner (9, 10).

Stenotrophomonas maltophilia is an aerobic Gram-negative bacillus and harbors two chromosomally encoded β-lactamase genes, L1 and L2 (11). Two different mechanisms of β-lactamase induction in S. maltophilia have been proposed. One is the nagZ-dependent mechanism; the other is the nagZ-independent mechanism (12, 13). The nagZ-dependent one, which has been shown to be involved in the uninduced β-lactamase overexpression of the ampD_I mutant (12), is basically similar to the mechanism of the known model of β-lactamase induction involving ampG-ampD-nagZ-ampR in the Enterobacteriaceae. Briefly, the degraded PG sacculus (primarily the GlcNAc-1,6-anhydro-MurNAc peptide) is transported by AmpN/AmpG permease from the periplasm into the cytoplasm (14). In the case of ampD_I inactivation, the transported degraded PG sacculus can largely be processed by β-N-acetylglucosaminidase (encoded by nagZ) to form the cytoplasmic activator ligand 1 (AL1; likely to be 1,6-anhydromuropeptides), which can induce β-lactamase expression in the presence of a functional AmpR protein. The nagZ-independent mechanism was first observed in an mrcA (predicted to encode PBP 1a) mutant (15). Distinct from the β-lactamase activity of the wild-type strain, the ΔmrcA mutant exhibits uninduced β-lactamase activity. nagZ inactivation of the ΔmrcA mutant does not compromise the uninduced β-lactamase activity, indicating that there can be a nagZ-independent activator ligand 2 (AL2) or another unidentified homologous NagZ enzyme responsible for uninduced β-lactamase expression in the ΔmrcA mutant (13). However, the involvement of a TCS in the β-lactamase expression of S. maltophilia has not been proposed.

S. maltophilia harbors six putative LT genes according to the sequenced S. maltophilia genomes (16). The impact of LTs on β-lactamase expression has not been described. In this study, we demonstrate that inactivation of membrane-bound lytic transglycosylase D1 (mltD1) triggers cross talk between the CreBC TCS, LTs, and the ampNG-ampD_I-nagZ-ampR regulatory circuit and then switches on L1/L2 β-lactamase expression.

MATERIALS AND METHODS

Bacterial strains, plasmids, and primers.

Table S1 in the supplemental material lists the strains, plasmids, and primers used in this study. S. maltophilia KJ harboring two active β-lactamases, L1 and L2, has been described previously (17). Tetracycline (30 μg/ml) was added to maintain the selection of cells carrying the pRK415 derivatives.

Construction of LT-knockout mutants.

The ΔmltA, ΔmltB1, ΔmltB2, ΔmltD1, ΔmltD2, and Δslt LT deletion mutants were constructed by double-crossover homologous recombination between the wild-type KJ chromosome and plasmids pΔMltA, pΔMltB1, pΔMltB2, pΔMltD1, pΔMltD2, and pΔSlt, respectively. The basic strategy for the construction of these plasmids was as follows: two DNA fragments targeting the 5′ terminus and the 3′ terminus of the mutated LT gene were obtained by PCR using the primer sets MltAN-F/MltAN-R and MltAC-F/MltAC-R for pΔMltA, MltB1N-F/MltB1N-R and MltB1C-F/MltB1C-R for pΔMltB1, MltB2N-F/MltB2N-R and MltB2C-F/MltB2C-R for pΔMltB2, MltD1N-F/MltD1N-R and MltD1C-F/MltD1C-R for pΔMltD1, MltD2N-F/MltD2N-R and MltD2C-F/MltD2C-R for pΔMltD2, and SltN-F/SltN-R and SltC-F/SltC-R for pΔSlt (see Table S1 in the supplemental material). These PCR amplicons (as shown in Fig. S1 in the supplemental material) were digested and subsequently cloned into pEX18Tc (18). Plasmid mobilization, transconjugant selection, and mutant confirmation were performed as described previously (12). The double mutant was sequentially constructed from the single mutant by use of the same procedure.

Determination of β-lactamase activity.

The induction of β-lactamase was studied using a published procedure (13) with nitrocefin (100 μM) as the substrate. Cefuroxime was used as the inducer and was added at 30 μg/ml. Enzyme activity was calculated by using a molar absorption coefficient for nitrocefin of 20,500 M⁻¹ cm⁻¹ at 486 nm. One unit of enzyme activity (U) was defined as the amount of enzyme that converts 1 nmol nitrocefin per minute. Specific activity (U/mg) was expressed as nanomoles of nitrocefin hydrolyzed per minute per milligram of protein. The protein content was determined by use of the Bio-Rad protein assay reagent, and bovine serum albumin was used as a standard. All experiments were performed in triplicate.

Construction of LT-overexpressing strains.

The full-length genes of six LTs were amplified by PCR and cloning into an expression plasmid, pRK415 (19), yielding pMltA, pMltB1, pMltB2, pMltD1, pMltD2, and pSlt. The PCR primer sets used for mltA, mltB1, mltB2, mltD1, mltD2, and slt were MltAN-F/MltAC-R, MltB1N-F/MltB1C-R, MltB2N-F/MltB2C-R, MltD1N-F/MltD1C-R, MltD2N-F/MltD2C-R, and SltN-F/SltC-R, respectively (see Table S1 in the supplemental material). The recombinant plasmids were transferred into wild-type strain KJ, obtaining the LT mltA⁺, mltB1⁺, mltB2⁺, mltD1⁺, mltD2⁺, and slt⁺ LT-overexpressing strains. The orientation of these cloned LT genes was the same as that of the resident lac promoter of pRK415. To avoid the possibility that PCR-introduced errors would affect LT activity, the PCR for each LT was performed at least three times to ensure the consistency of each LT gene sequence. The expression of the cloned gene was checked by quantitative reverse transcriptase PCR (qRT-PCR).

qRT-PCR.

Total cellular RNA and c-DNA preparation and the levels of transcription of the LT genes were determined by qRT-PCR as described previously (12). The primers used for qRT-PCR were MltAQ-F/MltAQ-R, MltB1Q-F/MltB1Q-R, MltB2Q-F/MltB2Q-R, MltD1Q-F/MltD1Q-R, MltD2Q-F/MltD2Q-R, SltQ-F/SltQ-R, and CreDQ-F/CreDQ-R for mltA, mltB1, mltB2, mltD1, mltD2, slt, and creD, respectively (see Table S1 in the supplemental material). The 16S rRNA gene was chosen for use for normalization of transcription levels. The relative quantities of mRNA from each gene of interest were determined by the comparative cycle threshold (C_T) method (20).

Construction of promoter-xylE transcriptional fusions.

A 227-bp fragment and a 578-bp DNA fragment upstream of the mltB1 and mltD2 genes were obtained by PCR amplification from the chromosome of S. maltophilia KJ using primer sets MltB1N-F/MltB1N-R and MltD2N-F/MltD2N-R, and the genes were cloned upstream of the promoterless xylE gene in plasmid pRKXylE (21), yielding plasmids pMltB1_xylE and pMltD2_xylE, respectively.

C23O activity assay.

Catechol-2,3-dioxygenase (C23O) activity was measured as described previously (22). One unit of enzyme activity (Uc) was defined as the amount of enzyme that converted 1 nmol substrate per minute. The specific activity was expressed as Uc/optical density at 450 nm (OD₄₅₀).

Antimicrobial susceptibility test.

The susceptibilities of the S. maltophilia strains to a number of β-lactams were tested following Clinical and Laboratory Standards Institute (CLSI) guidelines (23). The MIC was defined as the lowest concentration of antimicrobial that inhibited the visible growth of the bacteria.

Statistical analyses.

Data are presented as means ± standard deviations. Statistical significance was assessed by Student's t test.

RESULTS AND DISCUSSION

Multiple lytic transglycosylases are present in S. maltophilia.

The annotation of the S. maltophilia K279a genome indicates the presence of five putative mlt genes and a putative soluble LT (slt), including mltA (Smlt0155), two mltB genes (Smlt4052 and Smlt4650), two mltD genes (Smlt0994 and Smlt3434), and slt (Smlt4007) (see Table S2 in the supplemental material) (16). The genomic organization surrounding these putative LTs is pictured in Fig. S1 in the supplemental material. The predicted LTs of S. maltophilia shared 27 to 41% identity with those of E. coli (see Table S2 in the supplemental material). For convenience, the homologues of Smlt0155, Smlt4052, Smlt4650, Smlt0994, Smlt3434, and Smlt4007 in S. maltophilia KJ were named mltA, mltB1, mltB2, mltD1, mltD2, and slt, respectively.

LT-generated PG fragments are involved in β-lactamase expression in the E. coli in-trans recombinant ampC and ampR genes (24, 25). Therefore, reduced LT activity should mean less PG breakdown and lower levels of β-lactamase induction. However, it is possible that a loss of LT activity may have little effect on PG breakdown and β-lactamase induction because of the functional redundancy of these LTs. For this purpose, the uninduced and cefuroxime-induced β-lactamase activities between wild-type KJ and its derived LT deletion mutants (the ΔmltA, ΔmltB1, ΔmltB2, ΔmltD1, ΔmltD2, and Δslt mutants) were comparatively assessed. Inactivation of mltD1 resulted in an increase in uninduced β-lactamase activity (Fig. 1). The following study thus focused on elucidating the underlying mechanism of the ΔmltD1-mediated increase in uninduced β-lactamase activity. For the convenience of description, the term “ΔmltD1 mutant phenotype” used in this article represents the ΔmltD1-mediated increase in uninduced β-lactamase activity.

FIG 1.

Roles of lytic transglycosylases in β-lactamase expression by S. maltophilia. The strains assayed included the wild-type strain KJ and its derived LT deletion mutants (the ΔmltA, ΔmltB1, ΔmltB2, ΔmltD1, ΔmltD2, and Δslt mutants). Data are the means from three independent experiments. Error bars indicate the standard deviations for triplicate samples. *, P ≤ 0.01, calculated by Student's t test. (A) Uninduced β-lactamase activity. The overnight cultures assayed were inoculated into fresh LB with an initial OD₄₅₀ of 0.15. After 2 h of culture, the uninduced β-lactamase activity was determined. (B) Cefuroxime-mediated inducibility. The overnight cultures assayed were diluted to an OD₄₅₀ of 0.15 and subsequently grown at 37°C for 1 h. Induction was carried out using 30 μg/ml cefuroxime for 1 h. β-Lactamase activities were determined before and after induction.

Inactivation of mltD1 led to the upexpression of L1 and L2 β-lactamase genes.

Given the presence of two active β-lactamases, L1 and L2, in S. maltophilia KJ (17), we first assessed the involvement of L1 and L2 in the ΔmltD1 mutant phenotype. The L1 and L2 transcripts between the wild type and the ΔmltD1 mutant were quantitatively compared by qRT-PCR. Compared with their levels of expression in wild-type strain KJ, the L1 and L2 transcripts in the ΔmltD1 mutant had 13.5 ± 1.3-fold and 14.7 ± 1.5-fold increments, respectively (Fig. 2). As a result, mltD1 inactivation resulted in the coordinate overexpression of the L1 and L2 genes.

FIG 2.

Expression of lytic transglycosylase and β-lactamase genes by strain KJ and its derived strains determined by qRT-PCR. cDNA was prepared from the RNA of exponentially growing cells by RT-PCR and used as the template for qRT-PCR. The levels of expression of the target gene transcripts were normalized to the level of expression of the 16S rRNA gene using the ΔΔC_T method. Data are the means from three independent experiments.

Relationship between the ΔmltD1 mutant phenotype, the ampNG-ampD_I-nagZ-ampR regulatory circuit, and the CreBC TCS.

The ampNG-ampD_I-nagZ-ampR regulatory circuit is known to be responsible for β-lactamase expression in S. maltophilia (12–14, 22). In order to understand the role of the ampNG-ampD_I-nagZ-ampR regulatory circuit in the ΔmltD1 mutant phenotype, we constructed ampNG, ampR, and nagZ deletion mutants of ΔmltD1, yielding the ΔmltD1 ΔampNG, ΔmltD1 ΔampR, and ΔmltD1 ΔnagZ mutants, respectively. The uninduced β-lactamase activities of these mutants and their parent strain, the ΔmltD1 mutant, were determined. In the meantime, the β-lactamase activities of the KJΔNG, KJΔR, and KJΔZ mutants (consisting of wild-type strain KJ with the deletion of ampNG, ampR, and nagZ, respectively) reported in our previous publications (13, 14) were also included for comparison. Inactivation of ampNG and ampR decreased the uninduced β-lactamase activity of the ΔmltD1 mutant to the wild-type level (Fig. 3), indicating that the ΔmltD1 mutant phenotype relies on functional AmpN/AmpG permease and AmpR. nagZ inactivation partially reduced the uninduced β-lactamase activity of the ΔmltD1 mutant (Fig. 3).

FIG 3.

Uninduced β-lactamase activities of S. maltophilia KJ and its derived mutants. Bacteria that had been cultured overnight were inoculated into fresh LB with an initial OD₄₅₀ of 0.15 and cultured for 2 h, and the β-lactamase activities were determined. The data represent the means from three repetitions. Error bars indicate the standard deviations for triplicate samples. *, P ≤ 0.01, calculated by Student's t test. ΔmltD1ΔNG, ΔmltD1ΔR, ΔmltD1ΔZ, and ΔmltD1ΔBC, the ΔmltD1 ΔampNG, ΔmltD1 ΔampR, ΔmltD1 ΔnagZ, and ΔmltD1 ΔcreBC double mutants, respectively.

The CreBC TCS has been known to be involved in β-lactamase gene expression in Aeromonas spp. (5) and β-lactam resistance in P. aeruginosa (9). The homologues of creBC were identified within the S. maltophilia K279a genome by amino acid sequence alignment using P. aeruginosa CreBC as the query sequence. The homologues of CreBC in S. maltophilia K279a were identified to be Smlt1436 and Smlt1437, and their predicted amino acid sequences were found to have 50% and 48% identity to P. aeruginosa CreB and CreC, respectively, and 51% and 46% identity to Aeromonas BlrA and BlrB, respectively (see Fig. S2 in the supplemental material). To explore the relevance of the CreBC system to the ΔmltD1 mutant phenotype, we constructed the creBC deletion mutant of the ΔmltD1 strain, yielding the ΔmltD1 ΔcreBC strain. Inactivation of creBC in the ΔmltD1 mutant decreased the β-lactamase activity to nearly the wild-type level (Fig. 3).

Inactivation of mltD1 resulted in the upexpression of mltB1 and mltD2.

To verify the contribution of mltD1 inactivation to the β-lactamase activity increment, an mltD1 complemented strain was constructed by transferring plasmid pMltD1 harboring an intact mltD1 gene into the ΔmltD1 strain, yielding the ΔmltD1(pMltD1) mutant. The empty pRK415 plasmid was also transferred into the ΔmltD1 mutant, generating the ΔmltD1(pRK415) mutant as a control. The uninduced β-lactamase activities of strain KJ(pRK415) and the ΔmltD1(pRK415) and ΔmltD1(pMltD1) mutants were assessed. To our surprise, the uninduced β-lactamase activities of ΔmltD1(pRK415) and ΔmltD1(pMltD1) mutants were comparable and ca. 38-fold higher than the uninduced β-lactamase activity of KJ(pRK415). Therefore, complementation of the ΔmltD1 mutant with an intact mltD1 gene did not restore the uninduced β-lactamase activity of the ΔmltD1 mutant to the wild-type level.

Next, the mltD1 transcripts in strain KJ(pRK415) and the ΔmltD1(pRK415) and ΔmltD1 (pMltD1) mutants were checked by reverse transcriptase PCR (RT-PCR) and qRT-PCR. The mltD1 transcript was indeed undetectable in the ΔmltD1(pRK415) mutant; however, the level of the mltD1 transcript in the ΔmltD1(pMltD1) mutant was increased ca. 26-fold compared with that in wild-type strain KJ(pRK415) (Fig. 2). As a result, the situation in the ΔmltD1(pMltD1) mutant appeared to represent mltD1 overexpression. To verify the impact of LT overexpression on the uninduced β-lactamase activity, the uninduced β-lactamase activities of the six LT-overexpressing strains (the mltA⁺, mltB1⁺, mltB2⁺, mltD1⁺, mltD2⁺, and slt⁺ strains) were assessed. All LT-overexpressing strains tested except the mltD1⁺ strain had increased uninduced β-lactamase activity, and it was especially high for the mltB1⁺ strain (Fig. 4). Therefore, overexpression of mltD1 does not appear to be the key point for the increased uninduced β-lactamase activity observed in the ΔmltD1(pMltD1) mutant.

FIG 4.

Impact of LT overexpression on uninduced β-lactamase activities. The overnight cultures assayed were inoculated into fresh LB with an initial OD₄₅₀ of 0.15. After 3 h of culture, the uninduced β-lactamase activity was determined. The strains assayed included the wild-type KJ(pRK415) and its LT-overexpressing constructs derived from the wild type (the mltA⁺, mltB1⁺, mltB2⁺, mltD1⁺, mltD2⁺, and slt⁺ strains). Data are the means from three independent experiments. Error bars indicate the standard deviations for triplicate samples. *, P ≤ 0.01, calculated by Student's t test.

Since LTs have redundant functions (3) and overexpression of the genes for all LTs except mltD1 results in increased uninduced β-lactamase activity in S. maltophilia (Fig. 4), we considered whether mltD1 disruption results in altered expression of the remaining LT genes. The levels of expression of the six LT genes in strain KJ(pRK415) and the ΔmltD1(pRK415) mutant were compared by qRT-PCR. Most notably, the enhanced expression of mltB1 and mltD2 was observed in the ΔmltD1(pRK415) mutant (Fig. 2). To further confirm the correctness of the qRT-PCR results, the DNA segments comprising the mltB1 and mltD2 promoters were cloned into a modified pRKXylE reporter plasmid (21) to generate transcription fusion plasmids pMltB1_xylE and pMltD2_xylE, respectively. The effect of the ΔmltD1 deletion on mltB1 and mltD2 expression was further investigated by determining the C23O activity following the growth of wild-type KJ and the ΔmltD1 mutant carrying pMltB1_xylE or pMltD2_xylE. mltB1 and mltD2 were intrinsically expressed in wild-type KJ, and in response to mltD1 inactivation, their levels of expression were further increased (Fig. 5).

FIG 5.

Roles of creBC and ampNG in ΔmltD1-mediated mltB1 and mltD2 overexpression. The bacteria assayed were cultured overnight, inoculated into fresh LB with an initial OD₄₅₀ of 0.15, and cultured for 3 h, and the C23O activities were determined. The data represent the means from three repetitions. Error bars indicate the standard deviations for triplicate samples. *, P ≤ 0.01, calculated by Student's t test.

Given the evidence that mltD1 inactivation confers increased expression of the mltB1 and mltD2 genes (Fig. 5) and the uninduced β-lactamase activities of the mltB1⁺ and mltD2⁺ strains were higher than the β-lactamase activity of wild-type strain KJ (Fig. 4), we speculated that the mltB1 and mltD2 upexpression in the ΔmltD1 mutant may be involved in the ΔmltD1 mutant phenotype. Therefore, the mltB1 and mltD2 alleles were deleted from the ΔmltD1 mutant, yielding the ΔmltD1 ΔmltB1 and ΔmltD1 ΔmltD2 double mutants, respectively. Moreover, the ΔmltD1 ΔmltB1 ΔmltD2 triple mutant was also constructed. Figure 3 demonstrates that the uninduced β-lactamase activities expressed by the ΔmltD1 ΔmltB1 and ΔmltD1 ΔmltD2 mutants were lower than the β-lactamase activity expressed by the ΔmltD1 mutant but still higher than the β-lactamase activity expressed by wild-type strain KJ, supporting the suggestion that mltB1 and mltD2 overexpression in the ΔmltD1 mutant indeed contributes to the ΔmltD1 mutant phenotype. However, the ΔmltD1 ΔmltB1 ΔmltD2 triple mutant still displayed detectable uninduced β-lactamase activity (Fig. 3). As a result, the simultaneous inactivation of mltD1, mltB1, and mltD2 may actually upregulate the expression of the remaining LT genes. Thus, the levels of the mltA, mltB2, and slt transcripts in wild-type strain KJ and the ΔmltD1 ΔmltB1 ΔmltD2 triple mutant were determined by qRT-PCR. Compared to the levels of slt expression by wild-type KJ, the slt transcript of the ΔmltD1 ΔmltB1 ΔmltD2 triple mutant had a 4.85 ± 0.8-fold increment. Therefore, the overexpression of slt may be responsible for the uninduced β-lactamase expression of the ΔmltD1 ΔmltB1 ΔmltD2 triple mutant.

A more interesting observation was that overexpression of mltD1 in wild-type strain KJ and the ΔmltD1 mutant had different influences on the uninduced β-lactamase activity. KJ(pMltD1) displayed a basal level of uninduced β-lactamase activity (Fig. 4, mltD1⁺); however, the ΔmltD1(pMltD1) mutant had an elevated level of uninduced β-lactamase activity (471 ± 29 U/mg) comparable to that of the ΔmltD1(pRK415) mutant (360 ± 30 U/mg). For further elucidation, the L1/L2 and LT transcripts of KJ(pMltD1) and the ΔmltD1(pMltD1) mutant were evaluated by qRT-PCR. Compared with those of wild-type KJ(pRK415), a substantial increase in the L1 and L2 gene transcript levels was observed in the ΔmltD1(pMltD1) mutant but not in KJ(pMltD1) (Fig. 2), consistent with the results of β-lactamase activity determination. Furthermore, except for the mltD1 transcript level, the levels of all LT transcripts of KJ(pMltD1) were comparable to those of KJ(pRK415), indicating that the overexpression of mltD1 in wild-type KJ hardly affects the expression of the other LTs (Fig. 2). However, the levels of the mltB1 and mltD2 transcripts in the ΔmltD1(pMltD1) mutant were comparable to those of the ΔmltD1(pRK415) mutant (Fig. 2), indicating that overexpression of mltD1 in the ΔmltD1 mutant background barely blocks ΔmltD1-mediated mltB1 and mltD2 upexpression. This may be the reason why the complementation of mltD1 did not return the L1 and L2 gene transcripts and the uninduced β-lactamase activity of the ΔmltD1 mutant to the wild-type levels.

The ΔmltD1-mediated mltB1 and mltD2 upexpression is creBC and ampNG dependent.

Inactivation of creBC or ampNG led to the loss of the ΔmltD1 mutant phenotype (Fig. 3), emphasizing the importance of creBC and ampNG for the ΔmltD1 mutant phenotype. The effect of creBC and ampNG deletion on the ΔmltD1-mediated mltB1 and mltD2 upexpression was further assessed by a promoter transcription fusion assay. Compared with the activities in the wild type, the P_mltB1 and P_mltD2 activities in the ΔmltD1 mutant were increased and nearly reverted to the wild-type level once the ΔcreBC or ΔampNG deletion was introduced into the chromosome of the ΔmltD1 mutant (Fig. 5). Accordingly, the ΔmltD1-mediated mltB1 and mltD2 upexpression was processed by a creBC- and ampNG-dependent pathway. Because ΔmltD1-mediated mltB1 and mltD2 upexpression was creBC dependent, whether mltD1 inactivation activates the CreBC TCS is of great importance. creD expression is positively regulated by activated CreBC in E. coli and P. aeruginosa (9, 26). Therefore, the upexpression of creD is customarily used as an indicator for creBC TCS activation in E. coli and P. aeruginosa systems. To evaluate whether the mltD1 deletion leads to creBC activation in S. maltophilia, creD transcript levels in wild-type strain KJ and the ΔcreBC and ΔmltD1 mutants were assessed. creD was expressed in wild-type KJ, and its expression increased 3.83 ± 0.33- and 1.96 ± 0.24-fold in the ΔcreBC and ΔmltD1 mutants, respectively. Moya et al. have shown that deletion of creBC from P. aeruginosa PAO1 has little influence on creD expression (9). Therefore, the relationship between creD expression and creBC activation in S. maltophilia may be distinct from that in E. coli and P. aeruginosa systems. In fact, our unpublished data revealed that, in addition to the CreBC TCS, the promoter activity of P_creD was also regulated by other factors in S. maltophilia, which negates the possibility that creD expression can simply be linked to creBC activation. In view of the lack of a linkage of expression of a definite gene to the activation of CreBC in S. maltophilia so far, we cannot immediately conclude whether the mltD1 deletion directly activates CreBC TCS.

The impact of creBC, ampNG, and ampR on MltB1 and MltD2 overexpression-mediated uninduced β-lactamase activity was assessed. Strains KJ(pMltB1) and KJ(pMltD2) represent constructs overexpressing mltB1 and mltD2, respectively. The uninduced β-lactamase activities of KJΔBC(pMltB1) and KJΔBC(pMltD2) were comparable to those of KJ(pMltB1) and KJ(pMltD2), respectively (Fig. 6). Nevertheless, the uninduced β-lactamase activities were rarely detectable in strains KJΔNG(pMltB1), KJΔR(pMltB1), KJΔNG(pMltD2), and KJΔR(pMltD2) (Fig. 6). These results allow us to conclude that the mltB1 and mltD2 overexpression-mediated uninduced β-lactamase activities are dependent on functional AmpNG permease and AmpR systems but the creBC TCS is less relevant. In addition, the uninduced β-lactamase activity expressed in the mltB1-overexpressing strain was partially nagZ dependent; however, that in the mltD2-overexpressing strain was nagZ independent (Fig. 6).

FIG 6.

Roles of creBC, ampNG, ampR, and nagZ in mltB1 or mltD2 overexpression-mediated uninduced β-lactamase activities. The bacteria assayed were cultured overnight, inoculated into fresh LB with an initial OD₄₅₀ of 0.15, and cultured for 3 h, and the β-lactamase activities were determined. The data represent the means from three repetitions. Error bars indicate the standard deviations for triplicate samples. *, P ≤ 0.01, calculated by Student's t test.

Concluding remarks.

The impact of LTs on β-lactamase induction was studied in E. coli. Inactivation of LTs decreases the level of induced β-lactamase expression, and overexpression of LTs enhances β-lactamase induction (24, 25). The inhibition of LT activity has thus been proposed to be an alternative strategy for the control of β-lactam resistance caused by β-lactamase induction (24). However, in this study, the loss of mltD1 unexpectedly increased the level of uninduced β-lactamase expression, which is contrary to the previous perception. We have provided further evidence that mltB1 and mltD2 upexpression, rather than the mltD1 inactivation itself, provides the link between the ΔmltD1 deletion and uninduced β-lactamase expression in S. maltophilia and that the creBC TCS and ampNG-ampD_I-nagZ-ampR regulatory network are involved in this regulatory circuit. Figure 7 presents the relationship among the creBC, ampNG-ampD_I-nagZ-ampR regulatory network, mltB1 and mltD2 upexpression, and the uninduced L1 and L2 β-lactamase response driven by mltD1 inactivation. We propose that the disruption of mltD1 in some way upregulates the expression of mltB1 and mltD2 in a creBC- and ampNG-dependent way. It has been proposed that the murein turnover products can function as a signal to affect gene expression (27). The dependence of the ΔmltD1-mediated mltB1 and mltD2 upexpression on AmpNG strongly suggests that some degraded PG fragments which are transported from the periplasm to the cytoplasm by the AmpNG permease may be recognized by CreBC or by some members of the CreBC regulon and mediate mltB1 and mltD2 upexpression. The overexpressed MltB1 and MltD2 can further process peptidoglycan cleavage, and more PG fragments are transported into the cytoplasm by the AmpNG permease system and activate β-lactamase expression in an AmpR-dependent manner.

FIG 7.

Model of coordinated regulation of L1 and L2 expression by the creBC TCS, the ampNG-ampD_I-nagZ-ampR regulatory circuit, and lytic transglycosylase systems. Inactivation of mltD1 results in the upexpression of mltB1 and mltD2 in a creBC- and ampNG-dependent manner. The increased periplasmic MltB1 and MltD2 activity results in the accumulation of a variety of degraded murein fragments, which are then imported into the cytoplasm by the AmpNG permease system. The imported murein fragments are processed into the activator ligands (ALs) by either a NagZ-dependent or a NagZ-independent pathway. In the presence of ALs, the AmpR acts as an activator for the expression of L1 and L2 β-lactamases. OM, outer membrane; IM, inner membrane.

In our previous study, we have demonstrated that disruption of ampD_I or mrcA causes the coordinated hyperproduction of both the L1 and L2 β-lactamases in vitro (12, 15). As revealed in this article, mltD1 is the third identified gene whose inactivation confers a phenotype of β-lactamase hyperproduction in S. maltophilia. Talfan et al. have shown that the mutations in ampD_I or mrcA are linked to the phenotype of β-lactamase hyperproduction, as evidenced by the β-lactamase hyperproduction by mutants selected in the laboratory (28). They also noticed that some β-lactamase-hyperproducing mutants selected in the laboratory and clinical isolates harbored wild-type ampD_I and mrcA genes. Therefore, they proposed that at least one additional gene mutation in another gene, with the exception of ampD_I and mrcA, is involved in β-lactamase hyperproduction in S. maltophilia (28). On the basis of the findings presented in the present article, mltD1 is likely the candidate whose inactivation contributes to β-lactamase hyperproduction.

The conditions that directly perturb the PG balance have been noted to activate the Rcs phosphorelay-, CreBC-, and Cpx-mediated PG stress responses (4, 5, 9, 29). These conditions reported to trigger the PG stresses are generally associated with PBP inactivation, such as β-lactam treatment (4), PBP 4 inactivation of P. aeruginosa (9), and the simultaneous inactivation of PBP 4, PBP 5, PBP 7, and AmpH of E. coli (29). This study is the first to demonstrate that the inactivation of one LT triggers the upregulation of two others and that the creBC TCS is involved.