Compound Efflux in Helicobacter pylori

Abstract

Susceptibility testing with a variety of structurally unrelated compounds showed that hefC in Helicobacter pylori is involved in multidrug efflux. This efflux was shown to depend on the proton motive force, as demonstrated by ethidium bromide accumulation experiments. Thus, H. pylori contains an active multidrug efflux mechanism.

Efflux of compounds is a phenomenon commonly observed in bacteria (2, 10). Through this process, organisms are protected from possible toxic effects of metabolite accumulation or external compounds, and compound efflux results in a decreased susceptibility for a variety of antibiotics. Efflux can be mediated through specific pumps, as is often seen with gram-positive organisms, or through pumps that transfer a broad range of substrates (including antibiotics, detergents, and dyes), as is frequently the case with gram-negative bacteria (7). There are several families of these so-called multidrug efflux pumps. One of these is the resistance-nodulation-division (RND) family, whose members play a role in the decreased susceptibility to structurally diverse antibacterial compounds in different organisms (4, 5, 7, 8, 11, 12). The only known exception seems to be the RND family of efflux pumps in Helicobacter pylori, for which no role in antibacterial compound efflux could be identified. Genetic inactivation of three putative RND pumps identified on the basis of sequence homology in this organism did not result in an increased susceptibility for 19 antibacterial agents tested, many of which are prone to efflux in other organisms (1). In the present study, we have reexamined this anomaly by testing susceptibilities for additional antibacterial agents and using a susceptibility method that is more sensitive than the disk diffusion method used previously.

(Part of these studies was presented at the 2004 ASM General Meeting, New Orleans, La.).

Susceptibilities were determined for a wild-type strain of H. pylori (ARHp80) and its three isogenic mutants that had the putative RND cytoplasmic membrane protein, HefC (HP0607, JHP0554), HefF (HP0969, JHP0903), or HefI (HP1329, JHP1249), genetically inactivated (1). Broth microdilution MICs were determined in 96-well plates using twofold serial dilutions of compound. Cells (∼10⁶ CFU/ml starting inoculum in 150 μl total volume) were suspended in bisulfiteless Brucella broth (Difco) that was supplemented with 0.1% lactic acid (Sigma) and 5% fetal calf serum (Biowhittaker). The plates were incubated at 37°C under 5% O₂, 10% CO₂, and 85% N₂ atmosphere for 48 h, and subsequently the optical density was read at 640 nm (Titertek Multiskan Ascent plate reader). The MICs were defined as the lowest concentration that showed at least a 70% decrease in transmission. The susceptibilities of the hefF and hefI mutants did not differ significantly (more than fourfold) from the isogenic parental strain for the 20 compounds tested. (Table 1). This result was not unexpected for the hefI mutant, as it was demonstrated earlier that this gene is not or poorly expressed in vitro (1). The similar susceptibilities for the parental strain and the hefF mutant indicate that hefF does not play a role in efflux of these compounds under these conditions. In contrast, 9 of the 20 compounds tested showed at least an eightfold lower MIC in the hefC mutant, and the extent could be as high as 32-fold (penicillin G) or 64-fold (novobiocin) (Table 1). Decreased MICs were seen for compounds belonging to different structural classes, consistent with the hypothesis that the product of hefC participates in multidrug efflux.

TABLE 1.

MICs determined for wild-type strain ARHp80 and three isogenic mutants^a

Drug	MIC (μg/ml)
	ARHp80	hefC	hefF	hefI
Amoxicillin	0.008	0.008	0.015	0.008
Ampicillin	0.015	0.008	0.03	0.015
Penicillin G	0.002	0.00006	0.004	0.002
Piperacillin	0.125	0.008	0.5	0.125
Cefaclor	0.5	0.25	1	1
Ceftazidime	0.5	0.5	4	1
Cefotaxime	0.125	0.015	0.5	0.25
Ceftriaxone	0.125	0.008	0.5	0.125
Aztreonam	4	1	4	4
Ciprofloxacin	0.25	0.25	0.25	0.25
Nalidixic acid	32	32	64	64
Novobiocin	2	0.03	4	8
Clarithromycin	0.008	0.002	0.008	0.015
Clindamycin	1	0.125	2	8
Erythromycin	0.25	0.015	0.25	0.25
Chloramphenicol	4	2	4	4
Linezolid	8	8	8	16
Tetracycline	0.125	0.015	0.125	0.125
Gentamycin	1	0.5	2	2
Ethidium bromide	8	0.5	8	8

^aRepeat MICs were generally within a twofold dilution range. Values with a greater-than-fourfold decrease compared to ARHp80 are shown in boldface.

These results appear to be in disagreement with a previous study that failed to establish a role for hefC in efflux (1). However, an unfortunate choice of compounds and a methodology that may not be sensitive enough to detect relatively small differences in susceptibilities between the mutant and parental strains may have impacted that study. This assertion is based on the reexamination of 12 compounds from that study. Nine of these (amoxicillin, cefaclor, ceftazidime, aztreonam, ciprofloxacin, nalidixic acid, clarithromycin, chloramphenicol, and gentamicin) were not subject to efflux, while three (clindamycin, tetracycline, and cefotaxime) were found to be prone to HefABC-mediated efflux, with all three showing an 8- to 16-fold decrease in MIC for the hefC mutant. It is possible that this magnitude might not be easily detected when using the disk diffusion susceptibility testing employed in the previous study (1).

The energy needed for compound efflux by members of the RND family is derived from the proton-motive force (7, 9). To determine whether HefABC-mediated efflux depended on this, the accumulation of ethidium bromide was measured according to published procedures (3), with some modifications. Ethidium bromide was used because it is actively effluxed by H. pylori as evidenced by the 16-fold reduction in MIC for the hefC mutant compared to the parental strain (Table 1), and its accumulation in cells can be easily detected using fluorescence detection. Exponentially grown cells were resuspended in prewarmed Brucella broth (37°C) to an (OD₆₀₀) equivalent to 5.0, and ethidium bromide was added to a final concentration of 20 μg/ml. Fluorescence was measured using excitation at 544 nm and emission at 590 nm (Fmax 96-well plate reader; Molecular Devices). Carbonyl cyanide m-chlorophenylhydrazone (CCCP; 200 μM final concentration; Sigma) was added to collapse the proton-motive force. An increased rate of accumulation of ethidium bromide was observed in the hefC mutant compared to the parental strain before the addition of CCCP (Fig. 1), as is expected for strains with a reduced efflux capability. Similar differences have been reported with other members of the RND family of efflux pumps (3, 6). The slow and gradual rather than complete lack of accumulation of ethidium bromide in the wild-type strain is likely caused by the inability of H. pylori to maintain an optimal proton motif force under the experimental (aerobic) conditions used, which are unfavorable for this organism. Even less favorable conditions disallowing H. pylori to maintain a proton motif force, such as incubation at room temperature rather than at 37°C or incubation in a phosphate buffer (50 mM, pH 7) rather than growth medium, resulted in an accumulation of ethidium bromide similar to one seen with the hefC mutant (data not shown).

FIG. 1.

Accumulation of ethidium bromide as measured by an increase in relative fluorescence units (RFU) in the wild-type strain H. pylori ARHp80 (circles) and its hefC mutant (triangles). CCCP was added only to ARHp80 (closed circles) at time indicated with the arrow. Linear regression analysis (solid lines) calculated slopes of 0.027 (r² = 0.96) and 0.043 (r² = 0.99) for the accumulation in ARHp80 and hefC, respectively. The experiment is a representative of four individual determinations.

Accumulation of ethidium bromide increased rapidly to the levels of the hefC mutant for the wild-type strain upon addition of the uncoupler CCCP (Fig. 1). Again, these data are in agreement with studies using other organisms (3, 6) and can be interpreted as a reduction in compound efflux due to the loss of the proton-motive force driving efflux. Taken altogether, the results indicate that hefC is involved in energy-dependent multidrug efflux in H. pylori. Based on the 27% sequence homology of hefC to Escherichia coli acrB as well as its genomic arrangement compared to other components of the efflux system, it is postulated that HefC acts as the inner membrane component of a three-component efflux system common for RND systems.

In conclusion, this study shows that, similar to other gram-negative organisms, H. pylori contains an active multidrug efflux mechanism and therefore compound efflux needs to be taken into account when determining resistance mechanisms in this organism.