Involvement of Mutation in ampD I, mrcA, and at Least One Additional Gene in β-Lactamase Hyperproduction in Stenotrophomonas maltophilia

Abstract

It has been reported that targeted disruption of ampD I or mrcA causes β-lactamase hyperproduction in Stenotrophomonas maltophilia. We show here that β-lactamase-hyperproducing laboratory selected mutants and clinical isolates can have wild-type ampD I and mrcA genes, implicating mutation of at least one additional gene in this phenotype. The involvement of mutations at multiple loci in the activation of β-lactamase production in S. maltophilia reveals that there are significant deviations from the enterobacterial paradigm of AmpR-mediated control of β-lactamase induction. We do show, however, that S. maltophilia ampD I can complement a mutation in Escherichia coli ampD. This suggests that an anhydromuropeptide degradation product of peptidoglycan is used to activate AmpR in S. maltophilia, as is also the case in enteric bacteria.

INTRODUCTION

Stenotrophomonas maltophilia is an important, emerging cause of soft tissue, ocular, bloodstream, and respiratory tract infections in severely ill and/or immunocompromised patients and is increasingly being found colonizing the lungs of patients with cystic fibrosis. Members of this species are intrinsically multidrug resistant, with resistance to most β-lactams and aminoglycosides being expressed by the majority of clinical isolates (1, 2). β-Lactam resistance is due to the production of two β-lactamase enzymes, L1 and L2, which are coordinately upregulated during β-lactam challenge via the AmpR transcriptional regulator (3).

AmpR-mediated control of β-lactamase production is common in Gram-negative bacteria. In the paradigm system from Citrobacter freundii, AmpR is stimulated to act as a transcriptional activator in the presence of certain peptidoglycan recycling products, the anhydro-N-acetylmuramyl peptides (AHM-peptides) (4). These AHM-peptides are normally kept at low cytoplasmic concentrations by the enzyme AmpD, but β-lactams cause a release of AHM-peptide precursors from peptidoglycan in such quantities that AmpD is overwhelmed and the cytoplasmic AHM-peptide concentration increases to a point where β-lactamase production is induced (5). Some β-lactams, e.g., ceftazidime, cefotaxime, and aztreonam, are poor inducers of AmpR regulated β-lactamase production; others, e.g., cefoxitin and imipenem, are strong inducers. It is not certain why different β-lactams induce with different potencies, but there is a correlation between β-lactamase induction strength and inhibition of the low-molecular-weight penicillin-binding proteins (PBPs), e.g., DacB (PBP4) (6). Its low potency for β-lactamase induction explains why ceftazidime can successfully be used to treat infections caused by bacteria having AmpR-regulated β-lactamases, e.g., C. freundii and Pseudomonas aeruginosa. Indeed, many S. maltophilia clinical isolates are susceptible to ceftazidime (7). However, as is the case for other Gram-negative species, ceftazidime-resistant mutants of S. maltophilia can readily be selected in vitro (3) and are commonly isolated from clinical specimens (8). In most bacteria carrying an AmpR regulated β-lactamase, e.g., C. freundii, ceftazidime resistance results from mutation to a β-lactamase-hyperproducing phenotype. The cause in almost every case is a mutation or insertion in ampD (9). Loss of AmpD blocks the breakdown of AHM-peptides, so their concentrations are very high even during growth in the absence of β-lactam challenge, activating AmpR and causing β-lactamase hyperproduction (5).

It is now clear that in P. aeruginosa, whereas inactivation of AmpD causes AmpR-dependent β-lactamase hyperproduction and ceftazidime resistance, an alternative route to this phenotype is inactivation of PBP4 (10). The reason for this is not entirely clear, but loss of PBP4 also causes β-lactamase hyperproduction in Aeromonas hydrophila, a bacterium that uses a two-component regulator, BlrAB, rather than an AmpR-like regulator to control β-lactamase production and, in this case, the reason is an overabundance of pentapeptide side chains in peptidoglycan (a substrate of PBP4), which appear to be the signaling molecule for BlrB (11). Loss of PBP4 also activates a BlrAB homologue in P. aeruginosa, known as CreBC, and this helps to increase β-lactam MICs through an unknown mechanism; CreBC does not control β-lactamase production.

The S. maltophilia genome carries two ampD homologues, named ampD I and ampD II (12), as do enteric bacteria, where they are named ampD and amiD; P. aeruginosa has three ampD homologues, named ampD h1, ampD h2, and ampD h3 (13). It has been shown that targeted disruption of ampD I in S. maltophilia causes coordinate hyperproduction of both the L1 and L2 β-lactamases in vitro, but disruption of ampD II does not (14). Targeted disruption of mrcA, predicted to encode the high-molecular-mass PBP1A also activates β-lactamase production in S. maltophilia. Disruption of PBP1A is not associated with β-lactamase hyperproduction in enteric bacteria, so this hints at differences in S. maltophilia from the paradigm β-lactamase induction system (15). The question remains: is disruption of ampD I or mrcA commonly seen in β-lactamase-hyperproducing S. maltophilia mutants and clinical isolates, or is it rare, implicating mutations elsewhere? We address these questions below.

MATERIALS AND METHODS

Bacterial isolates.

S. maltophilia clinical isolate K279a and mutant derivatives (see below) were used in the present study. K279a is the prototype S. maltophilia genome sequence strain. It was cultured from a blood sample taken from a bacteremic patient being treated at the Bristol Oncology Centre, Bristol, United Kingdom (16) and has been very well characterized in this laboratory and elsewhere (3, 8, 12, 16–20). It is representative of S. maltophilia phylogenetic group A, which comprises genetically homogeneous isolates representing ca. 50% of the total number of clinical S. maltophilia isolates worldwide and includes the S. maltophilia type strain (8). Other S. maltophilia clinical isolates used were from a worldwide collection previously described, and originally from the SENTRY antimicrobial surveillance program (21). The specific isolates used were those that are phylogenetically similar to K279a as previously defined (8). Escherichia coli K-12 mutant JW0106-1 (F⁻ Δ[araD-araB]567 ΔampD728::kan ΔlacZ4787[::rrnB-3] λ⁻ rph-1 Δ[rhaD-rhaB]568 hsdR514) was obtained from the E. coli Genetic Stock Center, being from the Keio Collection of E. coli mutants (22) and having ampD disrupted by the insertion of a kanamycin resistance cassette.

Bacterial growth media and antibiotic discs for susceptibility testing were obtained from Oxoid (Basingstoke, United Kingdom). Chemicals and antimicrobials were obtained from Sigma-Aldrich (Gillingham, United Kingdom) except for nitrocefin, which was from Becton Dickinson (Hoddesdon, United Kingdom). Oligonucleotide primers were obtained from Eurofins (Ebersberg, Germany).

Selection of K279a mutants with reduced susceptibility to ceftazidime and assay of β-lactamase.

It was observed that when performing disc susceptibility testing on S. maltophilia isolate K279a according to the BSAC guidelines but using an inoculum that was 100-fold higher than the recommended inoculum (23), individual colonies appeared within the zone of clearing around a number of β-lactam impregnated discs, as has previously been seen with Etest strips (24). These colonies were picked, and MICs of β-lactams against K279a and mutant derivatives were determined using the Clinical and Laboratory Standards Institute (CLSI) broth dilution methodology (25). Total β-lactamase production relative to total protein was assayed in cell extracts of bacteria grown in nutrient broth in the presence or absence of cefoxitin (100 mg/liter) or ceftazidime (25 mg/liter). Inducer was added when the optical density of the culture measured at 600 nm was 0.2 to 0.4, and the total β-lactamase activity in cell extracts was assayed after 2 h of growth in the presence of inducer using nitrocefin as a substrate. To do this, bacteria in 10 ml of culture were pelleted by centrifugation (4,000 × g, 10 min) and lysed by the addition of 200 μl of Bugbuster (Novagen, Darmstadt, Germany), followed by gentle rocking for 20 min. Cell debris was pelleted by centrifugation (13,000 rpm, 5 min in a benchtop centrifuge), and the β-lactamase activity was measured from 10 μl of supernatant in a 200-μl assay using nitrocefin (100 μM) dissolved in 60 mM Na₂HPO₄·7H₂O, 40 mM NaH₂PO₄·H₂O, 10 mM KCl, and 1 mM MgSO₄·7H₂O (pH 7.0) containing 100 μM ZnCl₂ as a substrate. Assays were performed in a 96-well plate with the absorbance at 482 nm (AU) being measured using a plate-reading spectrophotometer (Spectra Max 190; Molecular Devices, Wokingham, United Kingdom) every 5 s over a 5-min time period. Linear gradients (ΔAU/min) were extrapolated, and an extinction coefficient of 17,400 AU/M was used to calculate nitrocefin hydrolyzing activity. The total protein concentration in each cell extract was quantified using the Bio-Rad protein assay reagent (Bio-Rad, Hemel Hempstead, United Kingdom) according to the manufacturer's instructions and used to calculate relative specific enzyme activity in each cell extract.

PCR and sequencing of ampD I, ampR, and mrcA and attempted complementation of an E. coli ampD loss-of-function mutation.

RAPD [random(ly) amplified polymorphic DNA]-PCR was performed as previously reported (26). Standard PCR was performed using the general method described previously (16) and the ampD I- or ampR-specific primers published elsewhere (3, 14). The S. maltophilia K279a mrcA gene was identified from the genome sequence (12) and amplified using three sets of primers designed to generate three overlapping ∼900-bp PCR products. The primer sequences were as follows: mrcA1F, 5′-ATGACTCGACTCCGCCGC-3′; mrcA1R, 5′-CATTTCCTGGCGCACCAGC-3′; mrcA2F, 5′-CGTACGTGGCCGAGCTG-3′; mrcA2R, 5′-CGCTGATGTACTTGCGCGC-3′; mrcA3F, 5′-TACGCGCGCAAGTACAT-3′; and mrcA3R, 5′-TGAGAAGATATCGAACGCGG-3′. All PCR products were cleaned using a Qiagen PCR cleanup kit and sent for dideoxy sequencing by Eurofins (Acton, United Kingdom). For the E. coli ampD complementation assay, an approach similar to that reported previously (27) was used. S. maltophilia ampD I PCR products were ligated into the pBAD-TOPO expression vector (Invitrogen, Leek, Netherlands) according to the manufacturer's instructions, and recombinants were used to transform E. coli JW0106-1 (ΔampD::kan) to ampicillin resistance (50 mg/liter) using kanamycin (25 mg/liter) to select for the ampD mutation. PCR was performed to check the orientation of the insert in various recombinants using pBAD primers (Invitrogen) according to the manufacturer's instructions. The C. freundii ampC-ampR encoding plasmid pNU305, based on the p15a origin of replication, which is compatible with the pMB1 origin found in pBAD-TOPO (28) was then used to transform E. coli JW0106-1 or its pBAD::ampD I transformant to tetracycline resistance (10 mg/liter). Successful complementation of E. coli JW0106-1::pNU305—i.e., to increase the zone diameter around a 30-μg aztreonam disc without altering the zone diameter around a 30-μg cefoxitin disc—by a cloned S. maltophilia ampD I allele confirms it encodes a functional protein with the same activity as E. coli AmpD (27).

RESULTS AND DISCUSSION

Selection and characterization of ceftazidime-resistant S. maltophilia K279a mutants.

The MIC of ceftazidime against the S. maltophilia genome sequence strain K279a is 8 mg/liter (20) (Table 1), meaning that the strain is susceptible, according to S. maltophilia specific breakpoints from the CLSI (29). Ceftazidime susceptibility is also a property of a significant proportion of clinical S. maltophilia isolates (7). Using BSAC disc susceptibility testing methodology, the zone diameter across a ceftazidime (30 μg) disc was found to be 34 mm for K279a. Because S. maltophilia resistance breakpoints are not reported by BSAC for ceftazidime, this is not a suitable method for testing susceptibility, but its use does allow the selection of mutants that have reduced susceptibility by comparison with the parent strain by picking colonies that grow within the zone of clearing around an antibiotic disc. Using this approach, eight S. maltophilia K279a mutants were selected based on their reduced susceptibility to ceftazidime. These eight mutants were confirmed to be K279a derivatives using RAPD-PCR (data not shown). All of the mutants were ceftazidime resistant (MIC ≥ 32 mg/liter) according to CLSI breakpoints (29) with a marked elevation in MIC compared to the parent strain. In contrast, the cefoxitin MIC against K279a is 128 mg/liter, and MICs against the ceftazidime-resistant mutants are not more than 256 mg/liter (Table 1).

Table 1.

β-Lactamase production phenotypes of S. maltophilia clinical isolate K279a and mutant derivatives

Isolate or mutant	Mean β-lactamase activitya (no inducer) ± SD	Mean fold inductionb		MIC (mg/liter)
		Cefoxitin (100 mg/liter)	Ceftazidime (25 mg/liter)	Cefoxitin	Ceftazidime
K279a	0.04 ± 0.01	50.3	25.2	128	8
K CAZ4	0.05 ± 0.01	35.5	6.2	256	64
K CAZ5	0.93 ± 0.18	1.9	1.6	256	128
K CAZ10	1.14 ± 0.17	0.9	1.5	128	128
K CAZ13	0.05 ± 0.01	46.4	3.3	256	256
K CAZ14	1.18 ± 0.14	1.3	1.4	128	128
K CAZ16	0.03 ± 0.01	59.3	6.0	128	32
K CAZ18	0.03 ± 0.01	23.7	9.0	128	32
K CAZ19	1.18 ± 0.21	1.5	1.3	128	128

^aExpressed as nmol of nitrocefin hydrolyzed/min/μg of protein.

^bβ-Lactamase activity in induced cultures relative to noninduced cultures. The value reported is a mean. Variation was <10% (n = 4).

Four of the ceftazidime-resistant mutants (K CAZ5, K CAZ10, K CAZ14, and K CAZ19) hyperproduce β-lactamase in the absence of β-lactam challenge, and expression is not significantly further induced after challenge with cefoxitin or ceftazidime (Table 1). In contrast, the other four mutants express wild-type levels of β-lactamase in the absence of β-lactam challenge. It has been noted before that S. maltophilia mutants selected for reduced ceftazidime susceptibility do not always hyperproduce β-lactamase. The existence of an additional ceftazidime resistance mechanism, possibly involving reduced permeability and/or ceftazidime efflux, has been proposed to explain this finding (3).

After cefoxitin challenge (100 mg/liter, 2 h), the fold induction of β-lactamase in K279a was found to be very similar to the four ceftazidime-resistant mutants that do not hyperproduce β-lactamase. This reflects their similar cefoxitin MICs (Table 1). However, the ability of ceftazidime to induce β-lactamase production in these four mutants is considerably less than in K279a. This is particularly true for K CAZ13, against which ceftazidime has the highest MIC (Table 1). Because the induction potential for ceftazidime against K279a and these four mutants is roughly inversely proportional to its MIC, ceftazidime resistance in the mutants cannot be caused by a mutation that makes β-lactamase production hyperinducible. Instead, these data add further evidence for the existence of permeability- or efflux-mediated ceftazidime resistance in S. maltophilia, where ceftazidime added at 25 mg/liter in the growth medium achieves a lower periplasmic concentration in the mutants than it does in the wild-type strain because of this permeability/efflux effect, and reduced β-lactamase induction is seen as a result.

S. maltophilia AmpDI has the same biological function as E. coli AmpD.

The physiological role of the enterobacterial AmpD proteins is to remove the peptide side chain from AHM-peptides, which are derivatives of the main peptidoglycan recycling products, the N-acetylglucosaminyl-1,6-anhydro-N-acetylmuramyl-peptides (G-AHM-peptides) that are released from peptidoglycan during normal cell growth (5). Disruption of ampD I has been shown to cause β-lactamase hyperproduction in S. maltophilia in the laboratory (14). The loss of or damage to AmpD has also been implicated in β-lactamase hyperproduction in ceftazidime-resistant clinical isolates from a variety of bacteria, where the resultant build-up of AHM-tri- and AHM-penta-peptides activates the β-lactamase regulator AmpR (27).

In order to test whether S. maltophilia AmpDI has the same biological activity as E. coli AmpD, wild-type ampD I from K279a was ligated into the pBAD-TOPO vector in the appropriate orientation to allow arabinose inducible expression. The recombinant plasmid was used to transform the E. coli ampD mutant JW0106-1 (22). Next, plasmid pNU305, which carries the ampC-ampR region from C. freundii (28), was introduced, and BSAC disc susceptibility testing was carried out for cefoxitin and aztreonam in the presence of 0.1% (wt/vol) arabinose, which stimulates production of recombinant S. maltophilia AmpDI. The data obtained confirmed that wild-type ampD I from S. maltophilia K279a can complement an E. coli ampD mutation. A JW0106-1 recombinant carrying ampD I from K279a was more susceptible to the poor AmpC inducer, aztreonam (28-mm zone diameter) than a recombinant carrying the control plasmid (18-mm zone diameter), but both recombinants were equally resistant to the strong AmpC inducer, cefoxitin (6-mm zone diameter for both).

These data confirm that S. maltophilia AmpDI it is highly likely to have the same biological function as E. coli AmpD. Furthermore, because loss of AmpDI causes β-lactamase hyperproduction in S. maltophilia (14), S. maltophilia AmpR is probably activated by an AHM-peptide that accumulates upon deletion of ampD I. Another alternative would be that a G-AHM-peptide is the AmpR activatory ligand in S. maltophilia, and it is true that we have not confirmed the specificity of S. maltophilia AmpDI for AHM-peptides over G-AHM-peptides, a defining feature of enterobacterial AmpD proteins (30–32). However, the observation that deletion of nagZ in S. maltophilia abolishes β-lactamase hyperproduction caused by deletion of ampD I (33) provides strong evidence against the possibility that G-AHM-peptides are AmpR activatory ligands in S. maltophilia and strengthens the evidence in favor of AHM-peptides fulfilling this role.

S. maltophilia ampD I can remain intact in β-lactamase-hyperproducing mutants.

In order to assess whether ampD I mutation is involved in the β-lactamase-hyperproducing phenotype expressed by the four ceftazidime-resistant S. maltophilia mutants, K CAZ5, K CAZ10, K CAZ14, and K CAZ19 (Table 1), ampD I was sequenced from each. Only one mutant, K CAZ10, was found to have a mutation in ampD I: a deletion of 30 nucleotides starting at position 478 of the 567-nucleotide gene. This deletion is likely to have arisen following strand slippage during replication because the seven-nucleotide sequence immediately before the start and end of the deletion has an identical sequence, TGATCCG, creating a “hot-spot” for the deletion of this 30-bp region of ampD I. This pair of seven nucleotide hot-spot sequences is perfectly conserved in ampD I from the phylogenetic group A clinical isolate KJ (14), but there are multiple mutations in the equivalent sequence in ampD I from the phylogenetic group B clinical isolate D457 (34), meaning that the deletion is less likely to happen.

The effect of deleting these 30 nucleotides from ampD I on its protein product is to remove 10 amino acids, between positions 159 and 168 of the 188-amino-acid AmpDI. This deleted amino acid sequence, PTLMVARKRD, contains the RKxD quadrad, equivalent to positions 161 to 164 of C. freundii AmpD, which is critical for substrate binding and zinc coordination (30). Accordingly, we had strong reason to suspect that this S. maltophilia AmpDI mutant would be functionless, and this was confirmed experimentally. The ampD I gene from S. maltophilia K CAZ10 was unable to complement an E. coli ampD-negative phenotype, unlike the wild-type ampD I above. Zone diameters for aztreonam and cefoxitin were identical in a JW0106-1 recombinant carrying the cloned mutant ampD I allele and in a recombinant carrying the control plasmid.

Although we have not formally confirmed that this 30-nucleotide deletion in ampD I is the only mutation carried by the K CAZ10 chromosome, it is a single-step mutant, and disruption of ampD I is known to cause β-lactamase hyperproduction in S. maltophilia (14). Accordingly, this ampD I loss-of-function mutation is almost certainly the reason for β-lactamase hyperproduction in K CAZ10. However, the other three β-lactamase-hyperproducing mutants selected in the present study, which look phenotypically indistinguishable from K CAZ10 (Table 1), all have a wild-type ampD I sequence.

Involvement of the mrcA mutation in β-lactamase hyperproduction in S. maltophilia.

Targeted disruption of mrcA, predicted to encode PBP1A in S. maltophilia strain KJ, has also been reported to activate β-lactamase production in the absence of β-lactam challenge (15). Furthermore, activatory mutations in ampR, encoding the β-lactamase transcriptional regulator have been found in β-lactamase-hyperproducing mutants derived from K279a (3). Using PCR sequencing, all four β-lactamase-hyperproducing mutants were confirmed to have a wild-type ampR gene. The ampR PCR product generated includes the ampR-L2 intergenic region (3), so we can confirm that this is also wild type in all mutants. Furthermore, the ampD I loss-of-function mutant K CAZ10 was found to carry a wild-type mrcA gene. An mrcA mutation was found in K CAZ19, however: a G-to-T transversion at position 487 causing a nonsense mutation at codon 163. Again, while we cannot be certain that this is the only mutation in K CAZ19, the fact that it is a single step mutant and that the loss of mrcA is known to cause β-lactamase hyperproduction in S. maltophilia (15) makes it almost certain that this nonsense mutation in mrcA is the reason for β-lactamase hyperproduction in K CAZ19. In the remaining two β-lactamase-hyperproducing mutants, however, ampR, mrcA, and ampD I are all wild type, implicating the mutation of at least one additional locus in β-lactamase hyperproduction in S. maltophilia.

Evidence for the involvement of AmpDI loss of function in β-lactamase hyperproduction in S. maltophilia clinical isolates.

In a previous study, we identified S. maltophilia clinical isolates that hyperproduce both the L1 and the L2 β-lactamases. Of these, three were from S. maltophilia phylogenetic group A, as are K279a and ca. 50% of isolates from a worldwide collection (8). Table 2 shows the amount of β-lactamase activity produced by them in the absence of β-lactamase challenge, relative to K279a. To investigate the possible involvement of ampD I mutation or disruption in β-lactamase hyperproduction in these three clinical isolates, the gene was PCR sequenced. The PCR product obtained when using DNA from isolate 98 from Brazil as a template was far larger than the expected size. Sequencing revealed a region of additional DNA integrated into ampD I after nucleotide 279. This region was found to be 88% identical to the gene smlt0835, identified as being like the insertion sequence ISPSy9, which is present in multiple places on the S. maltophilia K279a genome (12). It is therefore likely that this insertion sequence has moved into ampD I in isolate 98, and this insertion clearly would destroy the ampD I gene. We conclude, therefore, that this insertion is the reason for β-lactamase hyperproduction in this isolate.

Table 2.

Basal β-lactamase production and ampD I/mrcA sequence polymorphisms in S. maltophilia clinical isolates

Clinical isolate	Mean β-lactamase activitya ± SD	Sequence polymorphism(s)b
		ampD I	mrcA
K279a	0.04 ± 0.01	None (by definition)	None (by definition)
49-6147	0.22 ± 0.03	Leu122Val*, Arg127Gln*	Met347Leu*, Ala363Ser, Asn431Ser*, Thr645Ala, Ala649Thr, Val758Ile*
3800	0.19 ± 0.04	Ile65Val, Arg127Gln*, Val182Met	Gln307Leu, Ser308Ala, Glu409Asp, Ile417Leu, Asn431Ser*, Leu604Met, Thr645Ala, Ala652Val, Val758Ile*
98	0.87 ± 0.09	Insertion of IS element at nucleotide position 279	Not determined

^aExpressed as nmol of nitrocefin hydrolyzed/min/μg of protein.

^bThat is, polymorphisms with respect to K279a that affect the protein sequence. *, This difference (with respect to the sequence in K279a) is also found in the functionally wild-type AmpDI or MrcA proteins from S. maltophilia isolates KJ and/or D457.

β-Lactamase-hyperproducing S. maltophilia clinical isolates can have a wild-type ampD I and a seemingly wild-type mrcA.

The ampD I PCR products from the other two β-lactamase-hyperproducing clinical isolates (Table 2) were of the expected size, and sequencing revealed no gross lesions, frameshift, or nonsense mutations. A theoretical translation was performed for each ampD I sequence, and the predicted amino acid sequences were compared to AmpDI sequences from S. maltophilia clinical isolates known to carry normally inducible L1 and L2 β-lactamases and so, by definition, isolates that must have functional AmpDI proteins. These were isolates KJ (14), K279a (3, 12), and D457 (8, 34). Clinical isolate 49-6147 from Venezuela and isolate 3800 from Turkey encode AmpDI proteins that, respectively, have two and three amino acid differences compared to K279a AmpDI (Table 2). The two differences seen in isolate 49-6147 AmpDI are shared by the wild-type AmpDI from isolate KJ, so we concluded that the β-lactamase-hyperproducing isolate 49-6147 encodes a phenotypically wild-type AmpDI protein.

Of the three differences from K279a AmpDI seen in AmpDI from isolate 3800, one is also found in the wild-type AmpDI protein from isolate KJ (Table 2). In contrast, the other two changes are not found in any known wild-type AmpDI. Cloning the ampD I allele from isolate 3800 confirmed its ability to complement the E. coli ampD loss-of-function mutation with effects seemingly identical to the wild-type K279a ampD I sequence (data not shown). Accordingly, it must be concluded that the ampD I sequence from the β-lactamase-hyperproducing isolate 3800 is, like that from isolate 49-6147, an allele that encodes a phenotypically wild-type AmpDI protein, and that the differences seen relative to other S. maltophilia clinical isolates are simply due to genetic drift in what is known to be a highly heterogeneous species (8). We can therefore conclude that the loss of ampD I is not the only way for S. maltophilia clinical isolates to acquire a β-lactamase-hyperproducing phenotype.

Having ruled out the involvement of loss-of-function mutations in ampD I as a cause of β-lactamase hyperproduction in S. maltophilia clinical isolates 49-6147 and 3800, we also amplified mrcA by PCR, sequenced the products, and subjected the DNA sequences to theoretical translation. MrcA is an 807-amino-acid protein, and so we were not surprised to see differences in these two MrcA sequences with respect to MrcA from isolate K279a, and Table 2 lists these differences. The MrcA protein from isolate 49-6147 is different from K279a MrcA at six amino acids, but three of these differences are shared with MrcA from isolates KJ and/or D457 and so must not abolish the function of the protein. Of the remaining three differences, two are in a region spanning positions 645 to 655, which is hypervariable among S. maltophilia isolates. Using a COBALT analysis (35), 6 of 10 positions in this region were found to be variable among the 10 S. maltophilia MrcA GenBank database entries analyzed (date of access, 9 May 2013). The remaining difference in 49-6147 MrcA relative to K279a MrcA, Ala363Ser, is located immediately after another region of high variability, between positions 331 and 362, where there are eight variable positions out of 32 among the 10 GenBank sequences analyzed (25%). This compares with only 23 positions of variation among these 10 isolates in the remaining 765 amino acids of MrcA (3%). Accordingly, we consider these two regions to be amenable to random genetic drift because of their relatively low importance and that the differences seen in 49-6147 MrcA are highly unlikely to affect its function.

A similar case can be made against a functional effect for all eight amino acid sequence differences seen in MrcA from isolate 3800 compared to K279a MrcA (Table 2). Two of these differences are shared by MrcA from isolates KJ and/or D457, where MrcA is known to be functional. Two other differences are in the hypervariable region between positions 645 and 655 described above, and the remaining four are differences shared by MrcA sequences from at least 2 of the 10 GenBank entries for S. maltophilia, most of which are environmental isolates where it is highly unlikely that β-lactamase hyperproduction has been selected. Accordingly, we conclude that both β-lactamase-hyperproducing clinical isolates have wild-type ampD I sequences and are highly likely also to have wild-type mrcA sequences. This is further evidence that mutation of at least one additional locus can cause β-lactamase hyperproduction in S. maltophilia.

Conclusions: evidence for differences in the β-lactamase induction pathways of S. maltophilia and enteric bacteria.

In the enterobacterial AmpC/AmpR paradigm, some β-lactams induce β-lactamase production poorly, whereas others are strong inducers. However, all β-lactams cause a similar release of G-AHM-peptides from peptidoglycan, so an increase in peptidoglycan breakdown is not the only cause of β-lactamase induction in these bacteria (36). The key to this difference in inducer strength is the effect of the inducer on the periplasmic G-AHM-tetra-/tripeptide to G-AHM-pentapeptide ratio. In E. cloacae, the weak β-lactamase inducer, cefotaxime, produces a 50:1 ratio of G-AHM-tetra/tripeptide to G-AHM-pentapeptide, whereas with the strong inducer, imipenem, the ratio is < 10:1 (36). This happens because strong inducers dramatically increase the amount of pentapeptide present in peptidoglycan, something that depends on the amount of dd-carboxypeptidase activity in the cell (6, 11). Such enzymes convert pentapeptides into tetrapeptides, so if they are inhibited, the ratio of G-AHM-tetra/tripeptide to G-AHM-pentapeptide decreases (11). Accordingly, in enteric bacteria, β-lactamase induction is caused when a β-lactam causes both an increase in the amount of G-AHM-peptide released and a decrease in the ratio of G-AHM-tetra/tripeptide to G-AHM-pentapeptide (6, 36).

In S. maltophilia, there does not seem to be any clear delineation between strong and weak inducers of β-lactamase production. For example, aztreonam, which is highly specific for PBP3 and so does not induce β-lactamase production in enteric bacteria (6), strongly induces β-lactamase production in S. maltophilia (33). Accordingly, it would seem that muropeptide chain length is not important for β-lactamase induction in this species and, if true, this would mean that any mutation or cellular stress that causes a release of G-AHM-peptide will potentially activate β-lactamase production; there is no requirement for the inhibition of dd-carboxypeptidase activity as well.

This fundamental difference between β-lactamase induction in S. maltophilia and enteric bacteria may well explain why S. maltophilia mutants selected for β-lactamase hyperproduction can have mutations at multiple different loci: ampD I, mrcA, and at least one other, as shown in the present study. This is not merely of academic interest because inhibitors are being designed to target β-lactamase induction mechanisms, and if they are too specific, they may not solve the problem of β-lactamase hyperproduction, or even β-lactamase induction in S. maltophilia, as they have been suggested to do in other species (37).