Characterization of the AmpC β-Lactamase from Burkholderia multivorans

Burkholderia multivorans is a member of the Burkholderia cepacia complex, a group of >20 related species of nosocomial pathogens that commonly infect individuals suffering from cystic fibrosis. β-Lactam antibiotics are recommended as therapy for infections due to B.

ABSTRACT

Burkholderia multivorans is a member of the Burkholderia cepacia complex, a group of >20 related species of nosocomial pathogens that commonly infect individuals suffering from cystic fibrosis. β-Lactam antibiotics are recommended as therapy for infections due to B. multivorans, which possesses two β-lactamase genes, bla_penA and bla_AmpC. PenA is a carbapenemase with a substrate profile similar to that of the Klebsiella pneumoniae carbapenemase (KPC); in addition, expression of PenA is inducible by β-lactams in B. multivorans. Here, we characterize AmpC from B. multivorans ATCC 17616. AmpC possesses only 38 to 46% protein identity with non-Burkholderia AmpC proteins (e.g., PDC-1 and CMY-2). Among 49 clinical isolates of B. multivorans, we identified 27 different AmpC variants. Some variants possessed single amino acid substitutions within critical active-site motifs (Ω loop and R2 loop). Purified AmpC1 demonstrated minimal measurable catalytic activity toward β-lactams (i.e., nitrocefin and cephalothin). Moreover, avibactam was a poor inhibitor of AmpC1 (K_{i app} > 600 μM), and acyl-enzyme complex formation with AmpC1 was slow, likely due to lack of productive interactions with active-site residues. Interestingly, immunoblotting using a polyclonal anti-AmpC antibody revealed that protein expression of AmpC1 was inducible in B. multivorans ATCC 17616 after growth in subinhibitory concentrations of imipenem (1 μg/ml). AmpC is a unique inducible class C cephalosporinase that may play an ancillary role in B. multivorans compared to PenA, which is the dominant β-lactamase in B. multivorans ATCC 17616.

INTRODUCTION

Burkholderia multivorans is a member of a larger group of pathogens, the Burkholderia cepacia complex (BCC), that includes >20 different species (1, 2). BCC bacteria can cause infections (e.g., pneumonia) in immunocompromised persons or in other individuals, such as those with cystic fibrosis (3–7). Currently, β-lactam antibiotics are recommended as a treatment option for infections due to BCC bacteria (8, 9). However, BCC bacteria are inherently resistant to multiple antibiotics, including β-lactams, in which the mechanistic basis for resistance is correlated with the expression of β-lactamases (10–13).

β-Lactamases are clustered into four-different classes based on structural similarities (classes A, B, C, and D) (14). Class A, C, and D β-lactamases use a serine as the nucleophile, while class B β-lactamases are metalloenzymes, using a Zn²⁺ ion(s) for catalysis. Analysis of the B. multivorans ATCC 17616 genome revealed the presence of two β-lactamases: PenA, a class A β-lactamase, and AmpC, a class C β-lactamase (Table 1) (15). A LysR-type transcriptional regulator (PenR_A) is upstream of bla_PenA and is divergently transcribed. Expression of bla_PenA is regulated by PenR_A through a system analogous to AmpC/AmpR regulatory pathways present in Enterobacteriaceae and Pseudomonas aeruginosa (13, 16).

TABLE 1.

Annotated β-lactamases in the B. multivorans ATCC 17616 genome

No.	Locus tag; locus; chromosome	Class	Assessment
1	BMUL_RS18715; NC_010086; 2	A	PenA; possesses a signal peptide, SXXK, SDN, and KTG motifs
2	BMUL_RS20495; NC_010086; 2	A	Possesses a signal peptide but lacks SXXK, SDN, and KTG motifs
3	BMUL_RS30175; NC_010087; 3	C	AmpC1; possesses a signal peptide, SXXK, YSN, and KTG motifs

PenA was previously characterized and found to possess a very broad hydrolytic profile that includes penicillins, cephalosporins, carbapenems, and the monobactam aztreonam, as well as the β-lactamase inhibitors clavulanic acid, sulbactam, and tazobactam (17). In addition, PenA demonstrated significant sequence heterogeneity within clinical isolates of B. multivorans (18). Here, we analyze the second β-lactamase present in B. multivorans ATCC 17616, AmpC1. Similarly to PenA, AmpC also displays sequence diversity within clinical isolates of B. multivorans. Conversely, AmpC1 from B. multivorans ATCC 17616 possesses a narrow spectrum, and the contribution of AmpC1 to β-lactam resistance appears to be negligible, yet expression of AmpC1 is regulated by β-lactam inducers (e.g., imipenem).

RESULTS AND DISCUSSION

The amino acid sequence of AmpC is diverse compared to those of other AmpC β-lactamases, but the structure is similar.

AmpC possesses the classic motifs (SXXK, YSN, Ω loop, KTG, and R2 loop) found in other AmpC proteins (Fig. 1A). However, AmpC is most similar to other Burkholderia AmpC proteins (75% identity and 80% homology to Burkholderia cenocepacia AmpC [BcAmpC]) (Fig. 1B). Comparisons to prevalent chromosomal (Pseudomonas-derived cephalosporinase 1 [PDC-1], Acinetobacter-derived cephalosporinase 1 [ADC-1], P99 [Enterobacter cloacae AmpC], and Escherichia coli AmpC) and plasmidic (CMY-2 and FOX-4) AmpC proteins reveal lower identity (38 to 46%) and homology (58 to 61%) (Fig. 1B).

FIG 1.

Comparison of AmpC1 from B. multivorans (AmpC1_Bm) with AmpC from B. cenocepacia J2315 (AmpC_Bc) and other prevalent class C β-lactamases (PDC_1 and E. coli AmpC [EcAmpC]). (A) Amino acid sequence alignments using Clustal Ω. Black shading indicates identity, dark-gray shading is similarity, and light-gray and white shading are differences. AmpC motifs (SXXK, YSN, and KTG) are in red. (B) Percent identity and homology between prevalent AmpC proteins and AmpC1.

Two homology models of AmpC1 were generated using the B. cenocepacia AmpC crystal structure (Protein Data Bank [PDB] accession no. 5E2G). AmpC1 maintains the expected class C β-lactamase structure; however, the positioning of Q120 in the loop (residues L117 to H126) was nearly 5 Å away from the nucleophilic S64 unlike other AmpC proteins (i.e., CMY-2) (Fig. 2A, purple). The 500-ps molecular dynamics simulation (MDS) of the generated models revealed that the loop region is highly mobile, shifting 10 Å (Fig. 2A, yellow). The second homology model of AmpC1 shows that the loop between residues L117 and H126 was found in a different position (Fig. 2A, magenta). Further refinement of this loop region in AmpC1 was conducted in order to find the most favorable conformation. The loop's energy was optimized using the ab initio loop prediction algorithm LOOPER. The most energetically favorable conformation of the loop was most similar to model 2 (Fig. 2A, cyan). Efforts are under way to determine the crystal structure of AmpC1 to provide additional information on the positioning of the loop.

FIG 2.

(A) Homology models of AmpC1 combined with a loop refinement protocol suggest high flexibility of the L117-to-H126 loop (model 1, purple; model 2, cyan; multiple possible conformations from loop refinement, yellow; best energetically favorable conformation, magenta). The Q120 region has the largest movement (up to 10 Å), possibly changing the shape of the active-site entrance. (B) The amino acids that are replaced in the 27 AmpC variants (Table 2) are highlighted in green. Natural variants of AmpC have amino acid substitutions in two active-site motifs, the Ω loop (blue) (residues D193, R196, and T205) and the R2 loop (orange) (position K290); other active-site regions are the SXXK (with the nucleophilic S64), YSN, and KTG motifs (pink).

AmpC of B. multivorans exhibits considerable sequence heterogeneity.

To determine the sequence diversity of AmpC in B. multivorans, we analyzed the primary amino acid sequence of AmpC proteins from 49 different clinical isolates, and 27 novel AmpC variants were identified (Table 2). The variants had 1 to as many as 13 different amino acid substitutions. The locations of these amino acid substitutions were mapped to the AmpC1 homology model (Fig. 2B). The amino acid substitutions included those in the Ω loop (D193A, R196S, and T205A), as well as the R2 loop (K290T) (Fig. 2B). Amino acid substitutions in the Ω loop and the R2 loop of other AmpC proteins were shown to confer extended-spectrum AmpC phenotypes (e.g., the ability to hydrolyze cefotaxime and ceftazidime) (19).

TABLE 2.

Single amino acid substitutions found in AmpC proteins from 46 clinical isolates of B. multivorans^a

Strain(s)	Amino acid substitution(s), insertion(s), or deletion(s)	No. of differences	Refseq protein accession no.	AmpC designation
ATCC 17616, AU21747, AU17545, AU19729, AU10398, AU19518, AU14371, AU28069, AU25543, AU30760, AU20929		0	WP_012218336.1	AmpC1
AU19564	L104P	1	WP_105845787.1	AmpC2
AU13919, AU14328	T229V, G241A, R309P	3	WP_105776316.1	AmpC3
AU10897	G98A, T229V, K290T, A363E	4	WP_105758864.1	AmpC4
AU17534	D76A, T229A, G241A, K290T	4	WP_105783420.1	AmpC5
AU23995	A129V, R139H, T229A, I244T, A363E	5	WP_105810139.1	AmpC6
AU4507	A74V, R176Q, T229A, G241A, A363E	5	WP_006416937.1	AmpC7
AU24277	P18S, T205A, T229A, I244T, A363E	5	WP_105835089.1	AmpC8
AU12481	A137T, T205A, T229A, I244T, R261Q	5	WP_105789398.1	AmpC9
AU29198	TL11 M, T229A, D230E, K290T, P304S	5	WP_105844745.1	AmpC10
AU23668	L104V, A124E, V143 M, T229A, G362D, A363E	6	WP_105813500.1	AmpC11
AU14364	TL16S, T4A, T229A, G241A, G362D, A363E	6	WP_105765510.1	AmpC12
AU11233	TL11S, I11F, I33V, E79Q, M184L, T229A, I244T	7	WP_105770301.1	AmpC13
AU26250, AU15954, AU22892	TL11 M, V89I, R196S, T229V, I244T, K290T, A363E	7	WP_105780250.1	AmpC14
AU15814	TL16S, T4A, T111A, T229A, G241A, G362D, A363E	7	WP_105764661.1	AmpC15
AU30441	H7R, I8V, S47A, A141V, T229A, G241A, Q249R, A363D	8	WP_105811053.1	AmpC16
AU30050	S47A, A141V, D193A, T205A, T229A, G241A, Q249R, A363D	8	WP_105793891.1	AmpC17
AU21015	I8V, P18S, A165S, D190N, T205A, T229V, I245T, R309P	8	WP_107998461.1	AmpC18
AU14786, AU11772, AU23919	A17V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, A363D	9	WP_105767194.1	AmpC19
AU11358	TL22S, H7R, I8V, S47A, A141V, T205A, T229A, G241A, Q249R, A363D	10	WP_105775307.1	AmpC20
AU22436	H7R, I8V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, A363D	10	WP_006405341.1	AmpC21
AU17135, AU10047, AU28442	H7R, I8V, S47A, A141V, D193A, T229A, G241A, Q249R, P281S, I352 M, A363E	11	WP_105752596.1	AmpC22
AU16734	H7R, I8V, S47A, A141V, D193A, T229A, G241A, Q249R, P281S, L305R, A363E	11	WP_105792096.1	AmpC23
AU23690, AU11204	TL22S, H7R, I8V, S47A, A141V, D193A, T229A, G241A, Q249R, P281S, A363E	11	WP_105761708.1	AmpC24
AU19659	H7R, I8V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, P281S, L305R, A363E	12	WP_105807900.1	AmpC25
AU30438	TL22S, H7R, I8V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, V278I, A363E	12	WP_105847138.1	AmpC26
AU25057	TL22S, H7R, I8V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, P281S, A363E	12	WP_105838144.1	AmpC27
AU27706, AU23365	TL22S, H7R, I8V, I11V, S47A, A141V, D193A, T205A, T229A, G241A, Q249R, A279E, A363E	13	WP_069221591.1	AmpC28
AU18096, AU21596, AU10086	No class C β-lactamase found

^aThe AmpC sequences were compared to that of AmpC1 from B. multivorans ATCC 17616.

AmpC1 possesses a narrow spectrum for β-lactams.

To assess the contribution of AmpC1 to β-lactam resistance, the bla_AmpC1 gene was cloned into pBC SK(+) and expressed in E. coli DH10B for susceptibility testing and was cloned into the pGEX-6p-2 plasmid and expressed in E. coli Rosetta-gami 2 DE3 for protein purification and kinetics. In addition, the bla_AmpC1 gene was knocked out of B. multivorans ATCC 17616 using allelic replacement followed by subsequent excision via recombination. The bla_AmpC1 knockout strain was verified by PCR, and the phenotypic verification is presented in Fig. S1 in the supplemental material. An analytic isoelectric focusing (aIEF) gel overlaid with nitrocefin revealed the absence of AmpC1 activity in the knockout compared to the parental strain.

Susceptibility testing was conducted via disk diffusion and Etest using a panel of select β-lactams (e.g., ampicillin, cephalothin, ceftriaxone, cefotaxime, ceftazidime, and imipenem) (see Fig. S2A in the supplemental material). E. coli producing bla_AmpC1 was highly susceptible to all the agents tested, with only minor nonsusceptibility observed for cephalothin, with an intermediate zone of 17 mm (see Fig. S2B in the supplemental material). Subsequently, E. coli bacteria producing bla_AmpC1, along with B. multivorans ATCC 17616 Δbla_AmpC1, were subjected to MIC determination via agar dilution. Control strains, including B. multivorans ATCC 17616 and E. coli DH10B pBC SK(+) with bla_PenA, as well as an empty vector, were also tested. B. multivorans ATCC 17616 was highly resistant to all of the β-lactams (i.e., ampicillin, cephalothin, and imipenem) tested, and susceptibility to ampicillin was restored with avibactam, as expected from our previous work (Table 3) (20).

TABLE 3.

Susceptibility testing results

Strain	MIC (μg/ml)
	Ampicillin	Ampicillin-avibactama	Cephalothin	Imipenem
B. multivorans
ATCC 17616	1,024	4	≥4,096	64
ATCC 17616 ΔblaAmpC1	1,024	4	≥4,096	64
E. coli
DH10B pBC SK(+)	2	1	2	1
DH10B pBC SK(+) blaPenA	4,096	4	4,096	2
DH10B pBC SK(+) blaAmpC1	8	4	32	1

^aAmpicillin was varied, and avibactam was held at a constant concentration of 4 μg/ml. CLSI interpretive indices were as follows: Enterobacteriaceae (ampicillin, ≥32 μg/ml, resistant; cephalothin, ≥32 μg/ml, resistant; imipenem, ≥4 μg/ml, resistant; and imipenem, 2 μg/ml, intermediate), nonfermenting Gram-negative bacteria (penicillins, ≥128 μg/ml, resistant; cephalosporins, ≥32 to 64 μg/ml, resistant; and carbapenems, ≥8 to 16 μg/ml, resistant).

The MICs of ampicillin, cephalothin, and imipenem for B. multivorans ATCC 17616 Δbla_AmpC1 were equivalent to those for the parent strain (Table 3). Moreover, E. coli with bla_AmpC1 was susceptible to ampicillin (MIC = 8 μg/ml) and imipenem (MIC = 1 μg/ml) but demonstrated resistance to cephalothin (MIC = 32 μg/ml) (Table 3). The observations with AmpC1 are in sharp contrast to those for E. coli producing bla_PenA. The MICs for E. coli carrying bla_PenA more closely mimicked those of the parent strain, B. multivorans ATCC 17616, with nonsusceptibility to all the agents tested and susceptibility restored with the addition of avibactam (Table 3). The susceptibility testing results suggest that bla_PenA contributes significantly to β-lactam resistance in B. multivorans, while AmpC1 plays a minor role.

AmpC proteins typically demonstrate catalytic efficiencies near the diffusion limit (10⁹ M⁻¹ s⁻¹) for select cephalosporins (e.g., cephalothin) (21). Avibactam, a novel β-lactamase inhibitor of the diazabicyclooctane family, inactivates class A, class C, and some class D β-lactamases and, in combination with ceftazidime, restores susceptibility to ceftazidime against B. multivorans (20). Steady-state kinetics were conducted with a panel of β-lactams to evaluate the hydrolytic profile of purified AmpC1, as well as with the β-lactamase inhibitor, avibactam, to assess the ability of avibactam to inactivate AmpC1.

The purified AmpC1 β-lactamase was able to appreciably hydrolyze only cephalothin and nitrocefin; all other β-lactam substrates were not measurably hydrolyzed (see Fig. S2C in the supplemental material). AmpC1 showed high K_m values for nitrocefin (176 ± 36 μM) and cephalothin (392 ± 88 μM) (Table 4). However, the k_cat value for nitrocefin was considerable, resulting in a catalytic efficiency of 1.4 ± 0.3 μM⁻¹ s⁻¹, which is only ∼2-fold lower than that for PenA (Table 4). The catalytic efficiency for cephalothin was ∼30-fold lower for AmpC1 than for PenA (Table 4). AmpC1 is able to hydrolyze select cephalosporins, but not nearly as well as other AmpC proteins (19). In B. multivorans, PenA is a more efficient β-lactamase than AmpC.

TABLE 4.

Steady-state kinetic parameters for PenA and AmpC1 with selected β-lactams^a and avibactam^b

Compound	β-Lactamase	Km or Ki app (μM)	kcat (s−1)	kcat/Km (μM−1 s−1)
Cephalothin	PenAc	71 ± 7	221 ± 21	3.1 ± 0.3
	AmpC1	392 ± 88	43 ± 5	0.11 ± 0.02
Nitrocefin	PenAc	142 ± 14	460 ± 46	3.2 ± 0.3
	AmpC1	176 ± 36	251 ± 25	1.4 ± 0.3
Avibactam	PenAd	0.5 ± 0.1	NAe	NA
	AmpC1	>600	NA	NA

^aK_m and k_cat were determined under pseudo-1st-order conditions, and a nonlinear least-square fit of the data to the Henri-Michaelis-Menten equation determined the kinetic parameters. Each experiment was completed in triplicate, and the error values represent the standard errors of the mean.

^bK_{i app} was determined using a direct-competition assay under steady-state conditions. Each experiment was completed in triplicate, and the error values represent the standard errors of the mean.

^cData were previously published (17).

^dData were previously published (20).

^eNA, not applicable.

Avibactam is a potent inhibitor of PenA (apparent K_i [K_{i app}] = 500 nM) (Table 4) (20). Conversely, the K_{i app} endpoint (>600 μM) for AmpC1 could not be determined due to the poor inhibitory activity of avibactam against AmpC1 (Table 4).

Avibactam is able to form an acyl complex with AmpC1.

To determine whether avibactam is able to acylate AmpC1, electrospray ionization-mass spectrometry (ESI-MS) was conducted on AmpC1 incubated at 1:1 and 1:10 AmpC1/avibactam ratios with reactions terminated at 15 min and 24 h. The calculated molecular mass for AmpC1 is 39,748.61 Da, and the theoretical molecular mass for avibactam is 265 Da. The measured masses of apo-AmpC1 were 39,748 ± 5 Da and 39,747 ± 5 Da after 15-min and 24-h incubations (Fig. 3A and B). After a 15-min incubation with avibactam at a 1:1 ratio, only apoenzyme (39,748 Da) was observed; however, at a 1:10 ratio, small amounts of acyl-AmpC1 (40,014 ± 5 Da and 39,933 ± 5 Da) were observed (Fig. 3A and B). The 40,014- ± 5-Da peak corresponds to the addition of the unmodified avibactam (+265 Da) to AmpC1. The 39,933- ± 5-Da peak is also an acyl adduct; however, the mass is decreased by 80 Da, suggestive of the desulfated form of avibactam bound to AmpC1 (22). At 24 h, the acyl complex of AmpC1-avibactam (40,012 ± 5 Da) predominates at both 1:1 and 1:10 ratios (Fig. 3A and B). The desulfated avibactam-acyl complex (39,932 ± 5 Da) is observed in both; in addition, a new minor peak of 39,729 ± 5 Da was observed at the 24-h time point. The 39,729- ± 5-Da peak differs from that of apo-AmpC1 (39,748 Da) by −19 Da, suggestive of loss of a water molecule. Further analysis is under way to determine the significance of this peak. Avibactam is able to acylate AmpC1; however, the kinetic and ESI-MS data suggest that the rate is very low. Conversely, PenA forms an acyl-enzyme complex with avibactam within 15 s of incubation, as detected via ESI-MS (20).

FIG 3.

(A) Mass spectrometry of AmpC1 apoenzyme after 15 min (shaded) and 24 h (nonshaded) of incubation. Shown is mass spectrometry of the AmpC1-avibactam acyl-enzyme complex after 15 min and 24 h of incubation at 1:1 and 1:10 enzyme/avibactam ratios. (B) Summary of mass spectrometry data, including description of peaks formed.

Avibactam forms unfavorable complexes with AmpC1.

Using the homology model of AmpC1 with the more energetically favorable conformation of the L117-to-H126 loop, avibactam was docked into the active site. The active-site pocket of AmpC1 is a deep, narrow groove, resulting in avibactam clashing with some of the side chains during entry into the active site. In the Michaelis-Menten complex of avibactam in AmpC1, the carbonyl of avibactam is positioned in the oxyanion hole (i.e., backbone amides of S64 and S318). However, the carbamate moiety of avibactam is within hydrogen-bonding distance of active-site residues S64, K67, and N152, which decreases the probability of acylation (Fig. 4). The active-site pocket is constraining, preventing productive interactions. The observations of molecular modeling support the high K_{i app} of avibactam for AmpC1.

FIG 4.

Michaelis-Menten complex of AmpC1 (gray) and avibactam (cyan) revealing nonproductive interactions of avibactam with active-site residues. Potential hydrogen bonding interactions are indicated by dashed green lines.

The morphology of B. multivorans Δbla_AmpC1 is similar to that of wild-type cells. To assess if knocking out bla_AmpC1 alters the overall cell shape, B. multivorans ATCC 17616 and the Δbla_AmpC1 strains were stained using BacLight LIVE/DEAD staining and viewed using fluorescence microscopy. The morphology of the Δbla_AmpC1 strain was very similar to that of the wild type; thus, lack of AmpC1 does not affect the cell's shape (see Fig. S1B in the supplemental material).

AmpC1 and PenA expression is induced by imipenem.

Previously, bla gene expression in B. multivorans ATCC 17616 was shown to be inducible by β-lactams (13). However, an E. coli background was used, only bla_PenA induction was evaluated, and induction was measured via nitrocefinase activity. Here, the induction of bla_AmpC1 and bla_PenA expression was assessed via immunoblotting using B. multivorans ATCC 17616 grown in lysogeny broth (LB) in the presence of a β-lactam inducer (i.e., 1 μg/ml imipenem) for different amounts of time (5, 10, 15, 30, 45, 60 75, 90, 105, and 120 min). In the absence of imipenem, PenA protein could not be detected, while a very minor amount of AmpC1 was observed at 75, 90, and 105 min (Fig. 5). The addition of imipenem to the medium resulted in robust expression of AmpC1 and PenA that was time dependent (Fig. 5). The amount of AmpC1 protein produced was greatest between 30 and 90 min and began to decrease by 120 min (Fig. 5). With PenA, at 30 min, the PenA protein level was high, and it remained high up to 120 min (Fig. 5). The presence of 1 μg/ml imipenem in the medium did not alter the growth of B. multivorans during the course of the experiment. Expression of bla_AmpC1 and bla_PenA is induced by imipenem, supporting a role for these proteins in β-lactam resistance and/or cell wall recycling and metabolism.

FIG 5.

Immunoblotting reveals that protein expression of AmpC1 and PenA is inducible in B. multivorans ATCC 17616 after growth in subinhibitory concentrations of imipenem for different amounts of time. The anti-RecA antibody was used as a loading control.

Conclusions.

In this study, AmpC1 was found to be a weak β-lactamase. Knocking out bla_AmpC1 did not impact the susceptibility of B. multivorans; however, bla_AmpC1 expression was induced by imipenem. So, what is the true role of AmpC in B. multivorans? Additional investigations are under way to further assess the contribution of AmpC in B. multivorans. The catalytic efficiency of one of the 27 variants of AmpC identified here against β-lactams may be greater than that of AmpC1, thus providing a larger contribution to resistance. Alternatively, AmpC may play some role in peptidoglycan metabolism. For example, AmpH class C low-molecular-weight penicillin-binding proteins (PBPs) characterized in E. coli possess a structure similar to that of AmpC and have bifunctional dd-carboxypeptidase and dd-endopeptidase activities (23, 24). In summary, AmpC from B. multivorans is a unique inducible class C cephalosporinase that may play a subsidiary role compared to PenA, which is the dominant β-lactamase.

MATERIALS AND METHODS

Protein analysis.

AmpC sequences were analyzed by BLASTP and Bl2Seq from the National Center for Biotechnology Information website. Multiple-sequence alignments were generated using CLUSTAL Ω from the European Bioinformatics Institute (EMBL-EBI) and Discovery Studio (25–27).

Whole-genome sequencing (WGS) of B. multivorans.

Genomic DNA was purified from the clinical B. multivorans isolates using a MasterPure Gram-positive DNA purification kit (Epicentre Inc., Madison, WI) as recommended by the manufacturer. The genomes of 49 B. multivorans isolates were sequenced at the J. Craig Venter Institute, Rockville, MD (JCVI) with Illumina NextSeq (two 150-bp reads of sequence data). Paired-end libraries were constructed using Illumina Nextera XT kits. Sequence reads were generated with a target average read depth of ∼100-fold coverage. Sequence reads for each isolate were assembled individually using SPAdes (28) and annotated using the National Center for Biotechnology Information (NCBI) Prokaryotic Genome Annotation Pipeline (PGAP) (29). Raw DNA sequence reads were submitted to the NCBI Sequence Read Archive (SRA). A bla_AmpC gene could not be found in the AU21596, AU18096, and AU10086 genomes.

Bacterial strains and plasmids.

The bla_AmpC1 gene was synthesized (Celtek Bioscience, LLC) using the deposited nucleotide sequence from B. multivorans ATCC 17616 and cloned into pET24a(+) using NdeI and XhoI restriction enzymes. The bla_AmpC1 gene was subcloned from pET24a(+) to include the ribosomal binding site (RBS) of pET24a(+), using XbaI and XhoI restriction sites; ligated into pBC SK(+); and electroporated in E. coli DH10B for susceptibility testing. The bla_AmpC1 gene was also cloned into pGEX-6P-2 (minus nucleotides 1 to 75, encoding the signal peptide) using BamHI and EcoRI restriction sites and transformed into E. coli Rosetta-gami 2 DE3 cells for protein expression. The B. multivorans clinical isolates used in this study were obtained from the Burkholderia cepacia Research Laboratory and Repository (BcRLR) strain collection (20).

The construction of Δbla_AmpC1 was conducted using previously described methods (30). Briefly, nucleotide sequences flanking bla_AmpC1 (630 bp upstream and 432 bp downstream) were synthesized as one fragment by Celtek Bioscience, LLC. An NdeI restriction site was added to the pGPI-SceI vector using a Quikchange XL kit via site-directed mutagenesis. The bla_AmpC1 flanking sequence was cloned into the pGPI-SceI vector using XbaI and NdeI restriction sites and transformed into E. coli PIR1 cells (Invitrogen). E. coli PIR1 containing pGPI-SceI with the flanking sequences was used as the donor strain, and B. multivorans ATCC 17616 was used as the recipient strain. The “helper strain” was E. coli DH5α(pRK2013). Triparental mating was conducted according to previously published methods (30), and the mixed culture was plated in a single spot on a superoptimal broth (SOB) agar plate containing a nitrocellulose filter and incubated accordingly. The following day, the filter was washed in 1× sterile phosphate-buffered saline (PBS) in order to extract the entire spot of culture and plated onto LB agar with polymyxin B and trimethoprim. The resistant colonies were patched onto phenol red glucose agar plates (5 g peptone, 2.5 g NaCl, 1.8 ml 0.5% phenol red, 7.5 g agar, 50 ml 50% glucose, 450 ml H₂O), and those that did not ferment glucose were subjected to PCR to further confirm the first crossover event. The second crossover was performed as described above, with a different donor strain, E. coli DH5α(pDAI-SceI-SacB), using the first-crossover B. multivorans as the recipient and E. coli DH5α(pRK2013) as the helper. Agar plates containing sucrose were used to cure the pDAI-SceI-SacB plasmid as previously described (30). The resulting knockout strain was confirmed by PCR using primers that lie upstream and downstream of bla_AmpC; thus, the final PCR product of the bla_AmpC knockout was smaller than the wild type. In addition, internal bla_AmpC primers were also used to verify the lack of a PCR product with the knockout and the presence of a PCR product with the wild type. All the PCR products were sequenced for verification.

Analytical isoelectric focusing.

B. multivorans ATCC 17616 and the Δbla_AmpC1 strain were grown in LB to an optical density at 600 nm (OD₆₀₀) of 0.6. A final concentration of 1 μg/ml of imipenem was added to each culture to induce the expression of β-lactamases, and the cultures were grown for 1 h. Cells were pelleted by centrifugation at 12,000 rpm for 10 min. Crude extracts were prepared and loaded onto a FocusGel 3-10 24S (Serva Electrophoresis GmbH) with a pH gradient from 3 to 10 and electrophoresed using a Multiphor II apparatus. The gels were focused at 4°C at 8 W for 120 min. The detection of β-lactamases was performed using an overlay of 2 mM nitrocefin, a chromogenic β-lactam substrate that turns from orange to red upon hydrolysis by a β-lactamase. Purified PenA or AmpC1 β-lactamases were used as controls.

Fluorescence microscopy.

B. multivorans ATCC 17616 and Δbla_AmpC1 were grown to an OD₆₀₀ of 0.6 in LB. The cells were pelleted and washed twice in 0.9% NaCl to remove the medium. The cells were resuspended to an OD₆₀₀ of 1.25 in 1 ml of 0.9% NaCl and stained for 15 min in the dark using LIVE/DEAD BacLight bacterial viability kits (Molecular Probes; L7012) with equal ratios of SYTO9 (3 μl) and propidium iodide (3 μl) stains, according to the manufacturer's instructions. Three microliters of each sample was placed onto a glass slide with a coverslip. Samples were viewed on a Zeiss Axio Imager M1 microscope with a metal halide arc lamp using the green fluorescent protein (GFP) and DSRED filters, and images were obtained at ×400 and ×1,000 (oil immersion) magnification.

Antibiotic susceptibility.

Mueller-Hinton (M-H) agar dilution MICs, according to the Clinical and Laboratory Standards Institute (CLSI) (31, 32), were used to phenotypically characterize strains, as previously described (32).

Protein expression and purification of AmpC1.

E. coli Rosetta 2 DE3 or Origami 2 DE3 cells producing the AmpC1 β-lactamase from the pGEX-6P-2 plasmid were grown, lysed, and processed as previously described (33, 34). The recombinant AmpC1 possesses a cleavable N-terminal glutathione S-transferase (GST) tag, and thus, the crude extracts were run on a GSTrap FF column (GE LifeSciences). The GST tag was cleaved using PreScission protease as previously described (33, 34). The purity of the fractions was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Gels were stained with Coomassie brilliant blue R250, and the protein's mass was verified via ESI-MS. Protein concentrations were determined by measuring absorbance at a wavelength of 280 nm (λ₂₈₀) and using the protein's extinction coefficient (Δε) (49,850 M⁻¹ cm⁻¹ at 280 nm), which was obtained using the ProtParam tool available through the ExPASy Bioinformatics Resource Portal (31). To maintain full activity, AmpC1 was stored in 10 mM PBS, pH 7.4, with 25% glycerol at −20°C. Three milligrams of purified AmpC1 was sent to New England Peptide for antibody production in rabbits.

Kinetics.

Steady-state kinetic parameters were determined using an Agilent 8453 diode array spectrophotometer, as previously described (32, 35). Enzfitter (Biosoft) was used to fit the data to the Henri-Michaelis-Menten equation and to obtain the kinetic parameters, V_max and K_m, for the β-lactam substrates, nitrocefin and cephalothin, as follows: v = (V_max × [S])/(K_m + [S]), where v indicates velocity and [S] indicates concentration of substrate.

Using a direct-competition assay under steady-state conditions, the K_{i app} of avibactam was determined for AmpC1 as previously described (32, 35). The data were analyzed according to the following equation to account for the affinity of nitrocefin ([S], ncf) for the lactamase: K_{i app} (corrected) = K_{i app} (observed)/(1 + [S]/K_m ncf).

ESI-MS.

ESI-MS was conducted on a Waters Synapt G2-Si as previously described (20). Purified AmpC1 was incubated with either a 1:1 or 1:10 enzyme/inhibitor ratio with avibactam. Reactions were terminated by the addition of 0.1% formic acid and acetonitrile.

Molecular modeling.

The AmpC1 homology model was generated using the SWISS-MODEL repository with the B. cenocepacia AmpC crystal structure (PDB accession no. 5E2G) as the template (identity, 298/389 [77%]; similarity, 320/389 [82%]) (36–38). The local quality of the AmpC1 model was estimated using the raw QMEAN scoring function and was found to be 0.7 (39–41). The majority of the residues in AmpC1 had a QMEAN score of 0.8 (1 is maximum for the best model). However, the score for the L117-to-H126 loop was 0.3, indicating that the positioning of the loop was not reliable. The L117-to-H126 loop sits at the entrance to the AmpC1 active site; thus, the model required further refinement. A second homology model generated via the SWISS-MODEL repository had an overall QMEAN score for the protein of 0.73, and the scores for the residues in the L117-to-H126 loop were between 0.6 and 0.8. Around amino acid 300, the QMEAN score was near 0.3; however, the overall quality of the second model, especially for the residues near the active site, was much better. A multistep CHARMm protocol using an ab initio loop prediction algorithm, LOOPER, in BIOVIA Discovery Studio 2017 (DS2017) R2 (Accelrys, Inc. San Diego, CA) was used to generate a set of low-energy conformations for the specified loop region (L117 to H126), and they were scored based on their CHARMm energy. The best-scoring conformation was very similar to that of the second model. Thus, the second AmpC1 model used to construct Michaelis-Menten complexes with avibactam as previously described, using BIOVIA DS2017 molecular-modeling software (32). Intact and acylated avibactams were constructed using Fragment Builder tools and minimized using a Standard Dynamics Cascade protocol of DS2017. The avibactam was automatically docked into the active site of AmpC1 using the CDOCKER module of DS2017 (42).

Cloning, protein expression, and purification of recA.

The recA gene was amplified from genomic DNA isolated from B. multivorans ATCC 17616. The PCR product obtained was cloned into the pCR-XL-TOPO vector (Invitrogen) and electroporated into E. coli TOP10. The pGEX-6P-2 empty vector (allowing for the addition of a cleavable N-terminal GST tag) and pCR-XL-TOPO recA were digested using BamHI and XhoI restriction enzymes. The digested products, corresponding to pGEX-6P-2 and recA, were ligated overnight at 22°C, and the ligation reaction product was electroporated into E. coli DH10B. The pGEX-6P-2 recA plasmid was verified by DNA sequencing using MCLab. For protein expression, the pGEX-6P-2 recA plasmid was transformed into E. coli Rosetta 2 DE3 cells (Novagen). The RecA protein purification was conducted as described above for AmpC. The protein concentrations were determined by measuring the absorbance at λ₂₈₀ and using the protein's extinction coefficient (Δε, 14,440 M⁻¹ cm⁻¹ for RecA at λ₂₈₀). The theoretical mass of RecA (37,765 Da), as well as the extinction coefficient, was obtained using the ProtParam tool on the ExPASy Bioinformatics Resource Portal (31). Three milligrams of purified RecA was sent to New England Peptide for antibody production in rabbits.

Immunoblotting and induction of β-lactamase expression.

To measure β-lactamase induction, immunoblotting was conducted using the polyclonal AmpC1, PenA peptide (18), and RecA antibodies. B. multivorans ATCC 17616 was grown in LB to log phase at an OD₆₀₀ between 0.6 and 0.7. B. multivorans ATCC 17616 was treated with 1 μg/ml of imipenem to induce expression of bla_penA and bla_AmpC1, and aliquots of cells were collected after 5, 10, 15, 30, 45, 60, 75, 90, 105, and 120 min of induction. B. multivorans cells not induced with imipenem at the above-mentioned time points were also collected. Subsequently, the cells were pelleted and lysed using stringent periplasmic fractionation to prepare lysates, as previously described (43). These crude extracts were subjected to SDS-PAGE and transferred to polyvinylidene difluoride membranes. The membranes were blocked in 5% nonfat dry milk in 20 mM Tris-Cl with 150 mM NaCl, pH 7.4 (TBS), for 1 h and probed in 5% nonfat dry milk in TBS with 1 μg/ml of polyclonal AmpC, PenA peptide, or RecA antibodies. The membranes were washed five times for 10 min each time with TBS with 0.05% Tween 20 (TBST), and for protein detection, blots were incubated for 1 h in 1:5,000 dilutions of anti-rabbit secondary horseradish peroxidase (HRP)-conjugated antibodies in 5% nonfat dry milk in TBS. The immunoblots were washed five times for 10 min each time with TBST and developed using an ECL-Plus developing kit (GE Healthcare Life Sciences) or SuperSignal West Femto chemiluminescent substrate (Thermo Scientific) according to the manufacturers' instructions. Fotodyne Luminary/FX was used to capture images.

Accession number(s).

Annotated genomes were deposited in the GenBank WGS repository and can be obtained within BioProject PRJNA434393.