Efflux Pump Overexpression in Multiple-Antibiotic-Resistant Mutants of Bacteroides fragilis

Abstract

Multidrug-resistant mutants of a wild-type Bacteroides fragilis strain (strain ADB77) and a quadruple resistance nodulation division family efflux pump deletion mutant (ADB77 ΔbmeB1 ΔbmeB3 ΔbmeB12 ΔbmeB15) were selected with antimicrobials. Ampicillin, doripenem, imipenem, levofloxacin, and metronidazole selected for mutants from both strains; cefoxitin selected for mutants from strain ADB77 only; and sodium dodecyl sulfate selected mutants from ADB77ΔbmeB1 ΔbmeB3 ΔbmeB12 ΔbmeB15 only. The mutants overexpressed one or more efflux pumps.

Bacteroides fragilis is a clinically significant bacterium well endowed with efflux pumps, including 16 efflux pumps (BmeB1 to Bme16) from the resistance nodulation division (RND) family. BmeB is the efflux pump component of the BmeABC tripartite efflux system, which comprises an efflux pump (BmeB), a membrane fusion protein (BmeA), and an outer membrane channel protein (BmeC) (7, 13). Expression of the efflux pumps of the RND family is prevalent in wild-type strains of B. fragilis, and these pumps transport several structurally unrelated antimicrobials (7, 13). Studies have demonstrated that antimicrobials can select for bacterial multidrug resistance (MDR) due to overexpression of the RND family of efflux pumps (3, 4, 5, 8). The aim of this study was to determine the potentials of various antimicrobials to select MDR mutants of B. fragilis that overexpress bmeB efflux pumps and to determine whether the absence of several bmeB genes would affect this selection.

The parent strains used in this study were B. fragilis ADB77 (Wadsworth Anaerobe Laboratory [WAL] strain 108), an isogen of B. fragilis 638R optimized for gene deletion, and its derivative, a quadruple bmeB deletion mutant, ADB77 ΔbmeB1 ΔbmeB3 ΔbmeB12 ΔbmeB15 (WAL 219) (7). Antimicrobial MICs for each strain were determined with the spiral gradient endpoint method (14). Mutants were selected from the strains described above by the spiral plater technique (9) in a single step on a gradient of increasing antimicrobial concentrations on brucella blood agar plates. Twenty-one antimicrobials (antibiotics, detergents, and dyes) were tested: ampicillin (AMP), cefoperazone, cefotaxime, ceftizoxime, cephalexin, chloramphenicol, ciprofloxacin, clindamycin, doripenem (DPM), faropenem (FPM), gatifloxacin, imipenem (IPM), levofloxacin (LVX), meropenem, metronidazole (MDZ), moxifloxacin, norfloxacin, tetracycline, ethidium bromide (ETBR), sodium dodecyl sulfate (SDS), and triclosan. The antimicrobials were deposited in a radially decreasing concentration gradient from the center to the outside of the plate by using a spiral plater (Autoplate 4000; Advanced Instruments, Norwood, MA). The strains were resuspended in brucella broth to an optical density equivalent to the 0.5 McFarland turbidity standard (10⁸ CFU/ml) and were inoculated along the antimicrobial concentration gradient. The assay plates were incubated for 12 to 24 h at 37°C under anaerobic conditions. The bacteria grew from the edge of the plate toward the center. The end radius (ER) where growth stopped was the MIC. Susceptibility studies were performed on at least three independent occasions. A fourfold or greater difference in susceptibility was considered significant. The effects of efflux pump inhibitors (EPIs) were determined by measuring the decrease in antimicrobial MICs after incorporation of 25 μg/ml carbonyl cyanide m-chlorophenylhydrazone (CCCP) or reserpine (RPN).

Incubation of the MIC assay plates for 48 to 60 h resulted in isolated mutant colonies growing as a tail beyond the MIC up the concentration gradient to the tail end radius (TER) (Fig. 1). Five of 21 antimicrobials, ampicillin, doripenem, imipenem, meteronidazole, and levofloxacin, selected mutants from both strains (strains 108-AMP, 108-DPM, 108-IPM, 108-MDZ, 108-LVX 219-AMP, 219-DPM, 219-IPM, 219-MDZ, and 219-LVX); cefoxitin (FOX) selected a mutant from the wild type only (strain 108-FOX), and SDS selected a mutant from the deletion mutant only (strain 219-SDS). All mutants were resistant to at least three antimicrobial classes and were therefore multidrug resistant (Table 1). The MDR antibiograms varied between the mutants and were stable even after repeated subculture on antibiotic-free medium (MIC data were reproducible between three independent experiments). The selection of MDR mutants by SDS was particularly intriguing, especially since SDS is a fairly ubiquitous component of normal household and cosmetic cleansers. A correlation between the use of antiseptics/disinfectants and antibiotic resistance (11) has been reported, and triclosan has been shown to select for MDR mutants in Stenotrophomonas maltophilia (10). However, to our knowledge this is the first study demonstrating the selection of MDR mutants by a detergent. Due to the sheer number of antimicrobials used and the mutants selected, only antibiotics that selected mutants were used as representatives to determine the effects of EPIs on MICs. Studies usually show two- to fourfold effects of EPIs on MICs (6). In the current study, both EPIs reduced the MICs of the antimicrobials tested (ampicillin, cefoxitin, doripenem, imipenem, levofloxacin, metronidazole, norfloxacin, and SDS; P < 0.05) by up to fourfold, suggesting that energy-dependent efflux was a major mechanism of resistance in the MDR mutants selected (Table 1).

FIG. 1.

Mutant selection with metronidazole by the spiral plater technique. (A) Locations 1 and 2, the wild-type strain (strain ADB77) that gave rise to mutants; locations 3 and 4, strain ADB77 ΔbmeB1 ΔbmeB3 ΔbmeB12 ΔbmeB15 that gave rise to mutants; location 5, metronidazole-resistant control strain; (B) enlargement of location 2. Colonies arose within the region (indicated by the white arrow, which also shows the increasing antibiotic concentration) between the endpoint radius (MIC) to the tail end radius (indicated by the two white lines).

TABLE 1.

Antimicrobial susceptibilities of selected mutants

Antimicrobiala	MIC (μg/ml)
	108	108-AMP	108-FOX	108-DPM	108-IMP	108-LVX	108-MDZ	219	219-AMP	219-DPM	219-IMP	219-LVX	219-MDZ	219-SDS
AMP	8	128	64	32	32	128	32	4	128	16	16	128	16	64
AMP + CCCPb		16	16	16	8	32	16		64	16	8	128	16	64
AMP + RPN		32	32	8	16	64	8		32	4	4	64	8	32
CFP	32	128	128	64	64	128	64	16	128	16	16	128	16	64
FOX	4	64	128	32	32	128	32	0.5	64	16	16	128	16	64
FOX + CCCP		32	16	32	16	64	32		64	16	16	128	16	16
FOX + RPN		32	64	16	16	64	16		16	4	4	32	4	64
ZOX	8	64	64	32	32	64	32	8	64	32	32	64	32	16
LEX	32	128	128	128	128	128	128	32	128	128	128	128	128	64
CIP	2	8	16	4	4	32	4	2	8	2	2	32	4	32
DPM	0.13	0.5	0.5	16	0.5	1	0.5	0.06	0.25	8	0.25	2	0.25	2
DPM + CCCP		0.5	0.5	8	0.5	0.5	0.5		0.25	8	0.25	2	0.25	2
DPM + RPN		0.25	0.5	4	0.25	0.25	0.25		0.13	4	0.13	0.25	0.13	0.5
FPM	0.5	8	8	8	8	8	8	0.25	4	4	4	4	4	8
GAT	0.5	8	8	8	8	8	8	0.5	4	4	4	4	4	8
IPM	0.13	0.5	0.5	0.5	32	2	0.5	0.03	0.25	0.25	32	8	0.25	64
IPM + CCCP		0.25	0.25	0.5	16	1	0.5		0.25	0.25	16	8	0.25	64
IPM + RPN		0.25	0.25	0.25	8	0.5	0.25		0.13	0.13	8	2	0.13	16
LVX	0.5	4	4	4	4	128	2	0.25	4	4	4	128	4	8
LVX + CCCP		2	2	4	2	64	2		4	4	4	128	4	8
LVX + RPN		2	4	2	2	32	1		2	2	2	16	2	2
MEM	0.13	0.5	0.5	0.5	0.5	4	0.5	0.06	0.25	0.25	0.25	2	0.25	4
MDZ	0.5	8	8	8	8	16	128	0.25	8	8	8	8	128	8
MDZ + CCCP		4	2	8	4	8	64		8	8	4	8	128	8
MDZ + RPN		2	4	4	4	4	32		2	2	2	2	16	2
MXF	0.5	2	2	2	2	8	2	0.5	2	2	2	8	2	4
NOR	8	32	32	32	32	64	32	16	32	32	32	64	32	128
NOR + CCCP		16	32	32	16	64	32		32	32	32	64	32	128
NOR + RPN		16	8	16	8	32	16		8	8	8	16	16	64
ETBR	128	>512	>512	>512	>512	>512	>512	64	>512	>512	>512	>512	>512	>512
SDS	512	>512	>512	>512	>512	>512	>512	256	>512	>512	>512	>512	>512	>512
SDS + CCCP		512	512	512	512	512	512		512	512	512	512	512	512
SDS + RPN		512	512	512	512	512	512		512	512	512	512	512	512
Triclosan	32	64	64	64	64	64	64	16	64	64	64	64	64	64

^aAMP, ampicillin; CFP, cefoperazone; FOX, cefoxitin; ZOX, ceftizoxime; LEX, cephalexin, CIP, ciprofloxacin; DPM, doripenem; FPM, faropenem; GAT, gatifloxacin; IPM, imipenem; LVX, levofloxacin; MEM, merpenem; MDZ, metronidazole; MXF, moxifloxacin; NOR, norfloxacin; ETBR, ethidium bromide; SDS, sodium dodecyl sulfate.

^bThe effects of CCCP and RPN on the MICs of antimicrobials for selected mutants were determined.

Total cellular RNA was isolated with an RNeasy Protect kit (QIAGEN, Valencia, CA). Gene expression was quantified by comparative quantitative real-time reverse transcription-PCR (RT-PCR) with a Quantitect SYBR green one-step RT-PCR kit (QIAGEN, Valencia, CA) and a SmartCycler instrument (Cepheid, Sunnyvale, CA). Primers amplified 130- to 170-bp products of the genes bmeB1 to bme16 (7). Final amounts of 250 ng of RNA were converted to cDNA, and the reactions were normalized with 16S rRNA. Expression was quantified by measuring changes in PCR cycle threshold (C_T) values of the parental versus mutant strains using the delta-delta C_T (ΔΔC_T) method (7, 12). Data from five experiments were analyzed by Student's t test, and a P value of ≤0.05 was considered significant. A twofold or greater difference in expression was considered different.

All mutants except for 219-DPM overexpressed one or more of the bmeB genes by greater than twofold (Table 2). Changes were detected in 11 of 16 bmeB genes (bmeB1, bmeB2, bmeB3, bmeB4, bmeB5, bmeB6, bmeB7, bmeB11, bmeB12, bmeB15, bmeB16). The order of the greatest fold overexpression was bmeB4 (14.9-fold), bmeB7 (14.0-fold), bmeB15 (9.1-fold), bmeB11 (7.9-fold), bmeB5 (6.3-fold), bmeB6 (6.2-fold), bmeB16 (4.3-fold), bmeB1 (4.2-fold), bmeB3 (3.7-fold), bmeB2 (3.2-fold), and bmeB12 (2.4-fold). The order of frequency of overexpression (the number of isolates in which the pump was overexpressed) was bmeB4 (six isolates); bmeB1 (five isolates); bmeB2, bmeB5, bmeB7, and bmeB11 (four isolates each); bmeB3, bmeB6, and bmeB12 (two isolates each); and bmeB15 and bmeB16 (one isolate each). A large number of publications on gene expression coupled with physiological data have shown that changes in gene expression as low as twofold can have a significant physiological impact, and real-time RT-PCR is a highly sensitive method which has been extensively used to study the differential expression of efflux pumps (2, 12). Therefore, the changes observed in this study were significant.

TABLE 2.

Efflux pump overexpression in mutants selected from strain 108 (ADB77) and its derivative bmeB quadruple deletion mutant 219 (ADB77 ΔbmeB1 ΔbmeB3 ΔbmeB12 ΔbmeB15)

Selecting agent (strain)	Avg no. of mutant colonies from five selections	Efflux pump(s) overexpressed (fold increase)a
Ampicillin
108	17	bmeB1 (3.5), bmeB3 (2.7),bmeB4 (9.4), bmeB5 (3.1), bmeB16 (4.3)
219	9	bmeB4 (10.2), bmeB5 (6.3)
Cefoxitin (108)	3	bmeB1 (2.5), bmeB3 (3.7), bmeB4 (3.1), bmeB15 (9.1)
Doripenem
108	30	bmeB1 (3.8), bmeB11 (4.4)
219	18	Other?
Imipenem
108	25	bmeB1 (3.2), bmeB2 (3.2), bmeB6 (5.1), bmeB7 (6.2), bmeB11 (5.5)
219	12	bmeB2 (4.3), bmeB6 (6.2), bmeB7 (6.9)
Levofloxacin
108	25	bmeB1 (4.2), bmeB4 (13.4), bmeB11 (7.9), bmeB12 (2.4)
219	12	bmeB4 (14.9)
Metronidazole
108	14	bmeB2 (2.4), bmeB5 (2.7), bmeB7 (14.0), bmeB11 (4.4)
219	13	bmeB2 (2.3), bmeB5 (2.8), bmeB7 (14.7)
SDS (219)	2	bmeB4 (6.3)

^aEfflux pump overexpression was assessed by comparative quantitative real-time RT-PCR, and the data shown are averages from five independent experiments.

No mutations in the gyrA quinolone resistance-determining region were observed in any of the mutants, including the mutants selected with levofloxacin. Mutants were also beta-lactamase negative, as determined by the nitrocefin test.

In summary, antimicrobials can select for MDR mutants that overexpress bmeB genes; overexpression of particular bmeB genes is associated with certain MDR antibiograms. When several bmeB genes are missing (i.e., in strain 219), the mutant selection potential of antimicrobials is unchanged, reduced, or increased. Mutant 219-DPM did not overexpress any bmeB pumps and yet had an MDR profile which was affected by EPIs. In addition, mutants 219-LVX and 219-SDS overexpressed bmeB4 only but had high-level MDR profiles which differed slightly. These data suggest that a non-RND efflux gene(s) is also overexpressed, especially when several RND pumps are missing.

In conclusion, the mutant selection window (MSW) hypothesis postulates that bacterial mutant subpopulations resistant to antimicrobials are selected predominantly when drug concentrations are above the MIC but below the mutant prevention concentration (MPC; i.e., the lowest concentration where further mutant selection is prevented) (1, 15). In this study, the ER represented the MIC and the TER represented the MPC, and mutants were selected as a tail in the window between these concentrations. Therefore, these data support the MSW hypothesis. A recent animal model study of infection demonstrated that levofloxacin-resistant mutants of Staphylococcus aureus were most readily selected when antimicrobial concentrations at the site of infection fluctuated within the MSW defined in vitro (J. Cui, Y. Liu, W. Tong, K. Drlica, and X. Zhao, Abstr. 45th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 431, 2005). Since several of the agents tested in our study are used in therapy for infections involving B. fragilis and our own studies have also demonstrated overexpression of bmeB efflux pump genes in clinical isolates of B. fragilis (7a), the data presented here could reflect a mechanism by which multidrug resistant isolates could arise during therapy.