Interplay between Amoxicillin Resistance and Osmotic Stress in Helicobacter pylori

ABSTRACT

Rising antibiotic resistance rates are a growing concern for all pathogens, including Helicobacter pylori. We previously examined the association of specific mutations in PBP1 with amoxicillin resistance and fitness in H. pylori and found that V374L and N562Y mutations were associated with resistance, but also resulted in fitness defects. Furthermore, we found that hyperosmotic stress differentially altered the fitness of strains bearing these mutations; survival of the V374L strain was decreased by hyperosmotic stress, but the N562Y strain showed increased cell survival relative to that of wild-type G27. The finding that amoxicillin-resistant strains show environmentally dictated changes in fitness suggests a previously unexplored interaction between amoxicillin resistance and osmotic stress in H. pylori. Here, we further characterized the interaction between osmotic stress and amoxicillin resistance. Wild-type and isogenic PBP1 mutant strains were exposed to amoxicillin, various osmotic stressors, or combined antibiotic and osmotic stress, and viability was monitored. While subinhibitory concentrations of NaCl did not affect H. pylori viability, the combination of NaCl and amoxicillin resulted in synergistic killing; this was true even for the antibiotic-resistant strains. Moreover, similar synergy was found with other beta-lactams, but not with antibiotics that did not target the cell wall. Similar synergistic killing was also demonstrated when KCl was utilized as the osmotic stressor. Conversely, osmolar equivalent concentrations of sucrose antagonized amoxicillin-mediated killing. Taken together, our results support a previously unrecognized interaction between amoxicillin resistance and osmotic stress in H. pylori. These findings have interesting implications for the effectiveness of antibiotic therapy for this pathogen.

IMPORTANCE Rising antibiotic resistance rates in H. pylori are associated with increased rates of treatment failure. Understanding how stressors impact antibiotic resistance may shed light on the development of future treatment strategies. Previous studies found that mutations in PBP1 that conferred resistance to amoxicillin were also associated with a decrease in bacterial fitness. The current study demonstrated that osmotic stress can enhance beta lactam-mediated killing of H. pylori. The source of osmotic stress was found to be important for these interactions. Given that relatively little is known about how H. pylori responds to osmotic stress, these findings fill important knowledge gaps on this topic and provide interesting implications for the effectiveness of antibiotic therapy for this pathogen.

INTRODUCTION

Helicobacter pylori is a Gram-negative, microaerophilic, capnophilic pathogen that colonizes the gastric mucosa of humans. Colonization is asymptomatic for the majority of individuals. However, colonized individuals show an increased risk for development of gastric diseases, such as gastritis, ulcers, and gastric cancer (1). Currently, the most common course of treatment for H. pylori infection is triple therapy (2) consisting of a proton pump inhibitor and two antibiotics, usually clarithromycin and metronidazole or amoxicillin (3, 4). As with all pathogens in the modern clinical setting, increased rates of antibiotic resistance have been reported for H. pylori worldwide (4, 5). Moreover, resistance has been linked to increased rates of treatment failure (6). Therefore, research on the mechanisms of antibiotic resistance is crucial; the obtained information may lead to new treatment strategies or to refinement of the current treatment strategies.

In H. pylori, amoxicillin resistance is mediated by the penicillin-binding proteins PBP1, PBP2, and PBP3 (7), although mutations in PBP1 appear to be the most associated with resistance (5, 8–10). As a high-molecular-weight PBP, PBP1 has both transglycosylase and transpeptidase activity; it elongates un-cross-linked glycan strands and catalyzes peptide cross-links between adjacent strands (11). Unlike other bacteria, H. pylori only has one copy of PBP1 (11, 12). PBP2 and PBP3 are low-molecular-weight PBPs with endopeptidase activity; they are involved in editing the shape of the bacterial cell (12). Depletion of PBP2 in H. pylori results in the loss of the characteristic spiral shape (13). It has been suggested that PBP3 plays a role in septum formation (14), as depletion of this protein results in filamentation (15). All three PBPs are essential in H. pylori (13, 15, 16).

In a previous study, we explored the contribution of specific mutations in PBP1 to amoxicillin resistance and strain fitness in H. pylori (17). We found that V374L and N562Y mutations in the transpeptidase domain were associated with amoxicillin resistance but also resulted in strain fitness defects when these strains were cultured in competition with the wild-type ancestral strain in liquid culture. Furthermore, hyperosmotic stress differentially altered the fitness of strains bearing these mutations; survival of the V374L strain was decreased relative to that of the wild-type by hyperosmotic stress, but the N562Y strain showed increased cell survival relative to that of wild-type G27 when grown in a 24-h liquid culture competition assay (17). Brightfield microscopic visualization revealed that osmotic stress exposure of the V374L strain resulted in an increase in the coccoid cell morphology, a likely indication of cell stress (18, 19). In contrast, under high osmotic stress the N562Y strain displayed hyperelongation but little coccoid cell formation relative to that of the control strain (17). These results indicate that a single point mutation in PBP1 can change how the cell responds to osmotic stress. PBP1b in Escherichia coli, which is a homolog of H. pylori PBP1, has also been shown to be important for survival under conditions of hyperosmotic stress (20). Thus, there may be previously unknown interactions between amoxicillin resistance and osmotic stress through PBP1.

There has been a great deal of research on the effects of elevated NaCl on H. pylori. Indeed, high levels of NaCl alter H. pylori growth (21) and the expression of genes encoding outer membrane proteins and cagA (22, 23). High NaCl also changes the protein profile of the membrane (24), and affects the prevalence of virulence factors in the exoproteome (25). Animal models suggest that a high-salt diet promotes H. pylori colonization (26), selects for strains that are more resistant to oxidative stress (27), and could potentially lead to increased inflammation (27) and carcinogenesis (28, 29). In humans, high dietary salt intake is associated with increased risk of gastric cancer (30). Given that an estimated 95% of gastric cancer cases are associated with H. pylori infection (31), the increased risk may be directly linked to NaCl-induced effects on H. pylori. Despite all of the research in this area, the molecular mechanisms by which H. pylori survives hyperosmotic stress remain unclear (32). The available evidence suggests that osmotic stress leads to upregulation of heat shock proteins (21, 33) and that Fur is required to survive NaCl stress (34). However, beyond that, H. pylori lacks many of the traditional components associated with tolerance to hyperosmotic stress (35, 36).

Herein, the interplay between amoxicillin resistance and osmotic stress was further characterized. Experiments using wild-type G27 and isogenic strains bearing point mutations in PBP1 revealed synergistic killing when amoxicillin and elevated levels of NaCl were combined; this was true even for amoxicillin-resistant strains bearing V374L or N562Y mutations. Moreover, similar synergy was found with other beta-lactams, but not with antibiotics that did not target the cell wall. Similar synergistic killing was also demonstrated when KCl was utilized as the osmotic stressor. Conversely, osmolar equivalent concentrations of sucrose antagonized amoxicillin-mediated killing. Microscopic examination showed that the combination of amoxicillin and salt stress altered H. pylori cell morphology, but sucrose rescued H. pylori from the damaging effects of amoxicillin. Taken together, our results support a previously unrecognized interaction between amoxicillin resistance and osmotic stress in H. pylori.

RESULTS

Combined amoxicillin and NaCl stresses synergistically decrease H. pylori viability.

It was previously shown that H. pylori mutant strains that are resistant to amoxicillin display altered fitness and liquid growth phenotypes in the presence of hyperosmotic stress induced by the addition of NaCl (17). These data suggest a previously uncharacterized interaction between amoxicillin resistance and osmotic stress in H. pylori. To study this phenomenon in more detail, the ability of H. pylori to grow on plates supplemented with increasing concentrations of NaCl was examined for wild-type G27, as well as for isogenic strains containing various point mutations in PBP1, namely, the amoxicillin-resistant mutants V374L and N562Y and the amoxicillin-susceptible mutants E406A, S414R, S417T, T593A, and A599G (17). The Columbia agar used in standard horse blood agar (HBA) plates contains 85 mM NaCl. To identify the threshold of H. pylori resistance to NaCl, the concentration of NaCl was increased by millimolar increments that were equivalent to the addition of 1 g of NaCl. Compared to that on HBA plates with no extra NaCl, the population of recoverable CFU remained relatively unchanged until the concentration of NaCl was increased to an additional 120 mM (Fig. 1A). At this concentration, a roughly 5-log drop in recoverable CFU was observed for G27, the V374L and N562Y mutants (Fig. 1A), and all of the PBP1 isogenic mutant strains (see Fig. S1A in the supplemental material). In contrast to what was previously observed in liquid culture competitions (17), the N562Y mutant strain also showed a small but statistically significant decrease in CFU on the plates containing an additional 103 mM NaCl, suggesting a slight sensitivity to this NaCl concentration during culture on plates. Although the recoverable CFU was unaltered, additional NaCl concentrations above 68 mM resulted in slower growth and smaller colonies for all strains compared to that on HBA plates (data not shown). At concentrations of an additional 103 mM and 120 mM, colonies took a minimum of 8 days to reach countable size.

FIG 1.

Effect of added NaCl on H. pylori and synergy with amoxicillin. (A) G27 and the isogenic amoxicillin-resistant PBP1 mutant strains (V374L and N562Y) were dilution plated onto horse blood agar (HBA) plates containing the indicated concentrations of NaCl; HBA naturally contains 85 mM NaCl (control) from the Columbia agar base. The added millimoles NaCl are shown after the plus sign (+), and the total amount of NaCl found within the plate is indicated in parenthesis. A one-way analysis of variance (ANOVA) of the log-transformed data with Dunnett’s correction was used to compare the various concentrations of salt to the control. (B) G27, V374L, and N562Y were dilution-plated onto HBA containing 0 μg/mL, 0.125 μg/mL, or 0.25 μg/mL amoxicillin, with concentrations of an additional 0 mM, 34 mM, or 86 mM NaCl. The limit of detection (LOD) was 500 cells. The geometric mean and geometric standard deviation were plotted from a minimum of three biologically independent experiments. The minimum bactericidal concentration (MBC) was defined as the point at which 99.9% of the population had been eliminated compared to the control. A two-way ANOVA of the log-transformed data with Tukey’s correction was used for statistical analysis; within-drug-concentration comparisons are displayed. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Based on these data, concentrations of an additional 34 mM or 86 mM NaCl were selected to represent intermediate and high osmotic stress, respectively; these concentrations were chosen since they did not cause loss of viability or excessive growth delays. To understand the interplay with antibiotic stress, amoxicillin concentrations of 0.125 μg/mL or 0.25 μg/mL were also chosen. These concentrations represent the resistance breakpoint and twice the breakpoint for amoxicillin for H. pylori (37), respectively. A breakpoint is a clinically chosen concentration of antibiotic that determines whether a strain of bacteria is considered susceptible or resistant to that antibiotic; if the MIC of a strain is above the breakpoint of an antibiotic, the strain is considered resistant to that antibiotic (38). The ability of H. pylori to grow on plates supplemented with these NaCl and amoxicillin concentrations was then tested for G27, V374L, and N562Y (Fig. 1B). When considering the wild-type strain, neither of the tested NaCl concentrations alone affected viability. Conversely, the presence of 0.125 μg/mL of amoxicillin resulted in a 4-log decrease in recoverable CFU, and of 0.25 μg/mL amoxicillin resulted in a decrease below the limit of detection (LOD) for G27. The addition of high osmotic stress (86 mM NaCl) to the 0.125 μg/mL amoxicillin-containing plate resulted in a statistically significant synergistic decrease in recoverable G27 cells; a 5-log decrease in recoverable CFU was now observed. To determine if the observed decreases were due to slowed growth or due to cell death, the plates were swabbed, and samples were then plated onto HBA plates containing no osmotic stressor or antibiotic; no colonies grew on these plates. The synergy between NaCl and amoxicillin stress was even more readily observable in the strains containing the amoxicillin-resistant PBP1 mutations (Fig. 1B); V374L and N562Y were minimally affected by the highest concentration of either NaCl or amoxicillin when tested alone. Conversely, when 0.25 μg/mL amoxicillin and an additional 86 mM NaCl were combined, a statistically significant and dramatic decrease in recoverable CFU was observed for each strain, namely, a 5-log decrease for V374L and a 3-log decrease for N562Y. It is of note that this synergistic interaction was also evident for the PBP1 mutations that were not associated with amoxicillin resistance (Fig. S1B). Taken together, these results suggest a strong interplay between osmotic stress and amoxicillin resistance in H. pylori and underscore the importance of the bacterial cell wall in resisting both stressors.

Aztreonam and mezlocillin combined with NaCl osmotic stress synergistically decrease H. pylori viability.

PBP1 physically interacts with other PBPs to form the divisome and elongasome (16). Thus, rather than being due to changes in the function of the transpeptidation domain of PBP1, the observed amoxicillin-resistance and amoxicillin/osmotic stress interaction phenotypes could potentially be due to altered physical interactions of PBP1 with PBP2 and/or PBP3. To test this possibility, beta-lactam antibiotics that show greater specificity for PBP2 or PBP3 than for PBP1 were examined. In H. pylori, PBP2 shows a greater binding affinity for aztreonam than that of other PBPs (14), and PBP3 shows a greater binding affinity for mezlocillin than that of other PBPs (14). Since neither aztreonam or mezlocillin are normally used to treat H. pylori infection, established breakpoints for these antibiotics were not available; thus, concentrations of each drug that showed a minor effect or a major effect were determined experimentally (Fig. S2A and B). For aztreonam, concentrations less than 8 μg/mL resulted in recoverable CFU, while concentrations greater than or equal to 8 μg/mL resulted in no recoverable CFU (Fig. S2A). For mezlocillin, concentrations less than 16 μg/mL resulted in minimal effects on CFU recovery, while 32 μg/mL resulted in no recoverable CFU (Fig. S2B). Notably, both the amoxicillin-resistant mutants, V374L and N562Y, and the moxicillin sensitive PBP1 mutants E406A, S414R, S417T, T593A, and A599G demonstrated sensitivity to aztreonam and mezlocillin comparable to the wild-type G27. These results suggest that the amoxicillin resistance displayed by V374L and N562Y is due to the point mutation within the transpeptidation site of PBP1 and not due to altered interactions between PBP1 and PBP2 or PBP3. Based on these results, 2 and 4 μg/mL of aztreonam and 8 and 16 μg/mL of mezlocillin were selected for further experimentation.

When grown on HBA plates that contained 2 μg/mL aztreonam, G27 demonstrated a statistically insignificant decrease in CFU compared to growth on nonsupplemented HBA (Fig. 2A). While 34 mM NaCl with 2 μg/mL aztreonam resulted in a further decrease in recoverable CFU, 2 μg/mL aztreonam with 86 mM NaCl resulted in no recoverable colonies. This synergy was even more striking when the concentration of aztreonam was increased to 4 μg/mL aztreonam. Alone, this concentration resulted in nearly a 1-log decrease in CFU compared to the control. When combined with 34 mM or 86 mM NaCl, no CFU were recovered. Similar trends were observed for all of the PBP1 mutant strains, regardless of whether they were amoxicillin resistant (V374L and N562Y) (Fig. 2A) or sensitive (E406A, S414R, S417T, T593A, and A599G) (Fig. S3A).

FIG 2.

Aztreonam and mezlocillin synergy with NaCl. G27 and the isogenic amoxicillin-resistant PBP1 mutant strains (V374L and N562Y) were dilution plated onto HBA plates containing an additional 0 mM, 34 mM, or 86 mM NaCl and the indicated concentrations of aztreonam (A) or mezlocillin (B). HBA naturally contains 85 mM NaCl (control) from the Columbia agar base. The added millimoles of NaCl are shown after the “+,” and the total NaCl found within the plate is indicated in parenthesis. Concentrations of 0 μg/mL, 2 μg/mL, and 4 μg/mL were used for aztreonam, and concentrations of 0 μg/mL, 8 μg/mL, and 16 μg/mL were used for mezlocillin. The geometric mean and geometric standard deviation were plotted from a minimum of three biologically independent experiments. The limit of detection (LOD) was 500 cells. The minimum bactericidal concentration (MBC) was defined as the point at which 99.9% of the population had been eliminated compared to the control. A two-way ANOVA of the log-transformed data with Tukey’s correction was used for statistical analysis; within-drug-concentration comparisons are displayed. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Exposure of G27 to 8 μg/mL mezlocillin resulted in a slight decrease in recoverable CFU (Fig. 2B). While the results of the addition of 34 mM NaCl were not significantly different compared to those with mezlocillin alone, the combination with 86 mM NaCl resulted in no recoverable CFU. Similar trends were also observed when the concentration of mezlocillin was increased to 16 μg/mL, although the results were less striking due to the decreases in CFU observed with the antibiotic alone. As with aztreonam, similar trends were observed for all of the PBP1 mutant strains, regardless of whether they were amoxicillin resistant (V374L and N562Y) (Fig. 2B) or sensitive (E406A, S414R, S417T, T593A, and A599G) (Fig. S3B). Combined, these data suggest that hyperosmotic stress shows synergistic interactions with numerous cell wall-targeting beta-lactam antibiotics, regardless of whether those antibiotics preferentially target PBP1, PBP2, or PBP3.

Non-beta-lactam antibiotics do not display synergy with NaCl.

To determine whether antibiotics that do not target the cell wall also demonstrate synergistic effects with NaCl, metronidazole, clarithromycin, and tetracycline were selected for further study. In addition to each being commonly used to treat H. pylori infection, these antibiotics each possess different modes of action. Metronidazole creates free radicals that damage the DNA of the cell (39, 40); clarithromycin binds to the 50S ribosomal subunit and blocks transpeptidation, elongation, and empty tRNA release (40); and tetracycline inhibits the formation of the codon-anticodon link in the 30S ribosomal subunit (40). None of these three antibiotics directly interferes with cell wall formation.

While CLSI and EUCAST breakpoints for these antibiotics are available for H. pylori, the specific sensitivity of strain G27 was examined to facilitate selection of suitable concentrations for further study. Agar dilution plating revealed that inhibition was observed at concentrations equal to or greater than 2 μg/mL for metronidazole, equal to or greater than 0.0156 μg/mL for clarithromycin, and equal to or greater than 4.5 μg/mL for tetracycline (Fig. S4). Based on these data, concentrations of each drug were selected to represent a minor and major antibiotic stress to the cells; concentrations of 1 and 2 μg/mL for metronidazole, 0.0117 and 0.0156 μg/mL for clarithromycin, and 4 and 8 μg/mL for tetracycline were selected for further study. Combining the indicated concentrations of each antibiotic with 34 or 86 mM NaCl resulted in no evidence of synergy for G27, V374L, or N562Y (Fig. 3A to C). Only the S414R and T593A strains demonstrated a small but statistically significant difference with metronidazole at 2 μg/mL and 86 mM NaCl; the magnitudes of the differences and sizes of the error bars, combined with the fact that none of the other PBP1 mutant strains demonstrated this phenotype, may indicate that these are spurious results with no biological significance. The amoxicillin-sensitive mutants did not demonstrate synergy with either clarithromycin or tetracycline (Fig. S5A to C). En masse, these results suggest that the observed synergistic effects of antibiotics and NaCl are specific to antibiotics that target aspects of cell wall assembly in H. pylori.

FIG 3.

Non-beta-lactam antibiotics do not display synergy with NaCl. G27 and the isogenic amoxicillin-resistant PBP1 mutant strains (V374L and N562Y) were dilution plated onto HBA plates containing an additional 0 mM, 34 mM, or 86 mM NaCl and 0 μg/mL, 1 μg/mL, or 2 μg/mL metronidazole (A); 0 μg/mL, 0.0117 μg/mL, or 0.0156 μg/mL clarithromycin (B); or 0 μg/mL, 4 μg/mL, or 8 μg/mL tetracycline (C). HBA naturally contains 85 mM NaCl (control) from the Columbia agar base. The added millimoles of NaCl are shown after the “+,” and the total NaCl found within the plate is indicated in parenthesis. The geometric mean and geometric standard deviation were plotted from a minimum of three biologically independent experiments. The limit of detection (LOD) was 500 cells. The minimum bactericidal concentration (MBC) was defined as the point at which 99.9% of the population had been eliminated compared to the control. A two-way ANOVA of the log-transformed data with Tukey’s correction was used for statistical analysis; within-drug-concentration comparisons revealed no significant differences.

Effects of other sources of osmotic stress on H. pylori and amoxicillin resistance.

When considering osmotic stress, it is important to consider that different solutes may result in different effects on the bacterial cell (41); for example, osmotic stress induced by NaCl, KCl, or sucrose elicited different long-term responses from a Staphylococcus aureus population, despite the use of equivalent osmolarities (42). To determine whether the synergistic activity observed with the addition of NaCl extended to other osmolytes, KCl and sucrose were also examined. To compare the various osmotic stressors, the osmolarity of both KCl and sucrose were calculated to be consistent with the osmolarity used for NaCl. Since both NaCl and KCl disassociate into chloride and their respective metal ions, each mole of salt produces two osmoles of osmotic pressure. Since sucrose does not disassociate, the molar concentration was doubled to match the osmolarity of NaCl and KCl. When examined at concentrations that were the osmolar equivalent to that used for NaCl (Fig. S6A and B), neither of these osmolytes created a dramatic decrease in CFU. A small but statistically significant decrease for most of the strains was observed when KCl was tested at 120 mM, but no change was evident at 240 mM sucrose. Analysis of higher concentrations revealed that a concentration of 154 mM KCl caused a 3-log decrease in CFU (Fig. S7A), while higher concentrations resulted in no recoverable CFU. For the examined concentrations of sucrose (Fig. S7B), small decreases were observed at 376 mM, and 411 mM resulted in complete loss of recoverable CFU. Based on these data, 34 and 86 mM KCl and 68 and 171 mM sucrose were chosen for further study. Similarly to NaCl (Fig. 1B), when amoxicillin was combined with KCl, synergy was observed and no CFU for G27 were recovered for either amoxicillin concentration plus 86 mM KCl (Fig. 4A). Additionally, both V374L and N562Y mutant strains also experienced multilog decreases in recoverable CFU (Fig. 4A). This phenotype also held true for the amoxicillin-susceptible PBP1 mutants (Fig. S8A).

FIG 4.

Amoxicillin synergy with KCl and sucrose. G27 and the isogenic amoxicillin-resistant PBP1 mutant strains (V374L and N562Y) were dilution plated onto HBA plates with concentrations of 0 μg/mL, 0.125 μg/mL, or 0.25 μg/mL amoxicillin and 0 mM, 34 mM, or 86 mM KCl (A) or 0 mM, 68 mM, or 171 mM sucrose (B). HBA naturally contains 85 mM NaCl (control) from the Columbia agar base. The added mM of KCl or sucrose are shown after the “+,” and the total osmolarity of NaCl and KCl or sucrose found within the plate is indicated in parentheses. The limit of detection (LOD) was 500 cells. The geometric mean and geometric standard deviation were plotted from three biologically independent experiments. The minimum bactericidal concentration (MBC) was defined as the point at which 99.9% of the population had been eliminated compared to the control. A two-way ANOVA of the log-transformed data with Tukey’s correction was used for statistical analysis; within-drug-concentration comparisons are displayed. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

In contrast to the above results for G27, when amoxicillin was combined with the osmolar equivalent of 68 or 171 mM sucrose, the bactericidal effect of amoxicillin was antagonized (Fig. 4B). While 0.25 μg/mL amoxicillin reduced the population of G27 to below the limit of detection, the addition of sucrose resulted in a dose-dependent increase in recoverable CFU that was statistically significant with 171 mM added sucrose. Similar phenotypes were observed for the PBP1 mutant strains, regardless of their amoxicillin resistance profile (Fig. 4B and Fig. S8B). Thus, in contrast to the results obtained with NaCl and KCl, osmolar equivalent concentrations of sucrose act as an osmoprotectant that blocks amoxicillin-mediated killing of H. pylori. En masse, these results indicate that different solutes may result in different effects on the bacterial cell and may suggest that the charge of the osmotic stress agent is important for the resulting interaction with cell wall-targeting antibiotics.

Visualization of G27 and N562Y cells exposed to amoxicillin and NaCl, KCl, or sucrose.

Prior work demonstrated that G27 cells that were exposed to elevated NaCl showed marked morphological changes that included a loss of spiral shape, elongation, and chain formation (17, 21). Given the synergistic or antagonistic effects on amoxicillin activity observed with the various osmolytes, the morphological changes resulting from each stress alone and in combination with the antibiotic were next examined. Wild-type G27 and N562Y were selected for study as representative amoxicillin-sensitive and -resistant strains, respectively. The strains were spread onto plates supplemented with 0 μg/mL or 0.25 μg/mL amoxicillin with or without 86 mM NaCl, 86 mM KCl, or 171 mM sucrose. Bacterial cells were collected from the plate surface after 2 days, fixed, Gram stained, and then examined microscopically. In the absence of amoxicillin, G27 displayed normal spiral morphology characteristic of H. pylori (Fig. 5, top left, G27 plain). However, the addition of either NaCl or KCl resulted in cell elongation and chain formation (Fig. 5, G27 plain +86 mM NaCl +86 mM KCl); the latter was most evident with KCl. In contrast, bacteria grown in the presence of sucrose showed a characteristic spiral morphology that was similar to that on plates with no osmotic stressor addition (Fig. 5, G27 plain +171 mM sucrose). In the absence of amoxicillin, the N562Y mutant strain displayed similar morphology to that of G27 for each of the osmolytes (Fig. 5, top second from right, N526Y plain). The addition of NaCl or KCl also induced cell elongation and chain formation for N562Y (Fig. 5, N526Y plain +86 mM NaCl or +86 mM KCl). When grown with sucrose (Fig. 5, N526Y plain +171 mM sucrose), N562Y demonstrated the characteristic spiral morphology analogous to growth without sucrose.

FIG 5.

Amoxicillin and osmotic stress induced changes in morphology for the wild-type and N562Y strains. Lawns of G27 and N562Y, representing amoxicillin-sensitive and -resistant strains, respectively, from optical density (OD)-controlled overnight cultures were grown on nonsupplemented (plain) HBA plates or HBA plates containing an additional 86 mM NaCl, 86 mM KCl, or 171 mM sucrose, with or without 0.25 μg/mL amoxicillin; plain HBA naturally contains 85 mM NaCl from the Columbia agar base. After 2 days of growth, the cells were fixed with 4% paraformaldehyde for 30 min and washed 2 or 3 times with phosphate-buffered saline (PBS). The cells were then resuspended in ultrapure water, affixed to glass slides by drying and then Gram stained. Color images were taken with an Olympus BX60 microscope and a Spot Insight camera. The images were captured using the SPOT 5.6 program and interpreted using ImageJ. The experiment was performed three times, and representative images are shown. Bar indicates 5 μm in length and is identical for each image.

When amoxicillin was added at 0.25 μg/mL (Fig. 5, top second from left, G27 0.25 μg/L), G27 demonstrated coccoid morphology, indicating that the cells were under considerable stress; under stress, cells of H. pylori transition from the classical S-shaped cells to a coccoid appearance (18). Furthermore, the coccoid cell phenotype is associated with cell death (43), but has also been proposed to represent a viable but not culturable state that can occur during times of stress (44, 45). The addition of NaCl or KCl did not change this phenotype (Fig. 5, G27 0.25 μg/mL +86 mM NaCl or +86 mM KCl). Sucrose addition resulted in phenotypic restoration to predominantly spiral morphology, even at twice the breakpoint of amoxicillin (Fig. 5, G27 0.25 μg/mL + 171 mM sucrose). The N562Y mutant strain was unaffected by 0.25 μg/mL (Fig. 5, N526Y 0.25 μg/mL); while a few coccoid cells were visible, the majority of the cells appeared spiral and healthy. Again, the addition of NaCl resulted in elongation of the cells (Fig. 5, N526Y 0.25 μg/mL +86 mM NaCl or +86 mM KCl). However, in contrast to the morphological changes seen with NaCl alone (Fig. 5, N526Y plain +86 mM NaCl), the amoxicillin/NaCl-exposed cells demonstrated a tapered phenotype; the cells appeared thicker at the center than they did at the end. In addition, although this assay is qualitative and not quantitative, staining of the cells appeared uneven; portions of the cells appeared darker than the rest. Similar results were also obtained when the N562Y cells were exposed to elevated levels of KCl plus 0.25 μg/mL amoxicillin (Fig. 5, N526Y 0.25 μg/mL +86 mM KCl). Finally, the addition of sucrose resulted in typical spiral morphology (Fig. 5, N526Y 0.25 μg/mL +171 mM sucrose). Similar, though less pronounced, results were also found with 0.125 μg/mL amoxicillin (Fig. S9). En masse, the microscopic data support the results obtained with the plating assays (Fig. 1 and 4) and further support the notion that various osmotic stressors can synergistically or antagonistically effect amoxicillin activity against H. pylori.

DISCUSSION

Here, the interaction between amoxicillin resistance and osmotic stress was characterized for H. pylori. It was found that the combination of NaCl plus amoxicillin resulted in synergistic killing of H. pylori (Fig. 1B). Furthermore, this synergy with NaCl extended to other beta-lactam antibiotics (Fig. 2), but not to antibiotics that do not target the cell wall (Fig. 3). The synergy extended to other salts (KCl) (Fig. 4A), but the addition of sucrose effectively antagonized amoxicillin-mediated killing of H. pylori (Fig. 4B). Finally, microscopic visualization of cells exposed to each stress—alone and in combination with amoxicillin—revealed morphological changes that were consistent with the data obtained in the plating assays (Fig. 5). Together, these results indicate that osmotic stress can alter H. pylori resistance to amoxicillin and other beta-lactam antibiotics.

It has been previously noted that elevated levels of NaCl alter antibiotic resistance in both Gram-negative (46–48) and Gram-positive (49, 50) bacteria. However, for these other microbes, the addition of NaCl actually increased antibiotic resistance; this is believed to be due to increased activity of efflux pumps that decreases the exposure of the cell to the drug (46, 49, 50). Interestingly, the activity of H. pylori efflux pumps is unaffected by NaCl levels (22, 51, 52), and efflux pump inhibitors do not affect amoxicillin resistance for strain 26695 (53). Thus, the paradigm established in other bacteria does not apply to H. pylori. In this same vein, while sucrose protected H. pylori from antibiotic-mediated killing (Fig. 4B), sucrose has actually been shown to potentiate the activity of antibiotics against Acinetobacter baumannii biofilms (54), and other sugars have been demonstrated to have synergy with aminoglycosides in E. coli (55). Thus, as with other studied systems (56), the unique physiology of H. pylori results in phenotypes that do not always follow those established in other prototypical bacteria.

Despite the fact that the osmolarity of the various stressors was the same, NaCl and KCl increased amoxicillin-mediated killing of H. pylori, while sucrose antagonized it. These opposite results indicate that the observed activities depend on the properties of the various osmolytes being examined. While the exact molecular mechanism at play here is not clear, it is likely that the charge of the ionic compounds comes into play. NaCl and KCl disassociate into chloride and their respective metal ions, but sucrose does not disassociate. Peptidoglycan is known to possesses a net negative charge, and the addition of NaCl can cause peptidoglycan to contract (57). This contraction appears to be independent of osmotic stress, since sucrose does not result in cell wall contraction even when added at the same osmolar concentration (57). If a similar phenomenon is occurring in H. pylori, then peptidoglycan contraction could cause steric hinderance of the PBPs, slowing the formation and repair of the peptidoglycan cross-links. Since reduced cross-linking between the glycan strands is associated with decreased mechanical stiffness of the cell wall and reduced survival upon exposure to hyperosmotic stress (58), the reduced activity of H. pylori PBP1 under these conditions would make the cell even more susceptible to amoxicillin, leading to cell lysis. While one could hypothesize that amoxicillin-resistant PBP1 mutant strains, which likely already have alterations in their peptidoglycan structure, might not be subject to the same effects, we found that the V374L and N562Y mutant strains were still susceptible to the effects of salt. It is important to note that NaCl-induced contraction of the cell wall was discovered in a Gram-positive bacterium (57), which possesses a more substantial layer of peptidoglycan compared to Gram-negative bacteria. While similar contraction phenotypes have been found in Gram-negative bacteria (59, 60), to the best of our knowledge, this has not been demonstrated to be charge related. Thus, future molecular and biochemical studies will need to examine the peptidoglycan cross-linking seen in the various amoxicillin-resistant and -sensitive strains and determine if the peptidoglycan does indeed contract in response to charged ionic compounds.

The in vitro results suggest that the in vivo amoxicillin resistance of H. pylori may be altered by the human diet. We previously (17) suggested that a form of coselection called cross-resistance (61) might be occurring. Resistance to amoxicillin varies geographically, with lower resistance in Europe and the Americas and higher amoxicillin resistance in parts of Asia (2, 62). Furthermore, the N562Y mutation is found in 63.1% of amoxicillin-resistant strains from Japan and South Korea (7, 9, 63–65) and in 14.3% of amoxicillin-resistant strains from Europe and the Americas (8, 66, 67). Given that diets in South Korea and Japan contain a higher salt content (68, 69) than those in the West (70, 71), we previously speculated that the increased osmotic stress from the diet was selecting for increased prevalence of the N562Y mutation (17). The data contained here further supports this hypothesis; while N562Y was affected by the synergy of NaCl and amoxicillin, the population did not die to the same extent as the wild type or other PBP1 mutant strains (Fig. 1B; see also Fig. S1B in the supplemental material) Thus, the greater selective pressure from amoxicillin/salt synergy could select for more amoxicillin resistance in H. pylori.

To the best of our knowledge, this is the first time that NaCl and KCl have been demonstrated to have a synergistic effect on amoxicillin-mediated bacterial killing. This difference from other bacteria may be due to the fact that many of the traditional components associated with tolerance to hyperosmotic stress are absent in H. pylori; the trehalose synthetic pathway (72), kdp or trk K⁺ transporters (36, 73), and RpoS (35) are all missing from the H. pylori genome. Depending on the strain, H. pylori may even lack glycine betaine/proline transporters (74). Thus, the lack of these components may be responsible for the effects seen herein. Furthermore, it is interesting to speculate that Campylobacter jejuni, which is genetically closely related to H. pylori, may be similarly affected by the combination of osmotic stress and cell wall-targeting antibiotics; C. jejuni also lacks many osmotic response factors (75).

Our results naturally lead to the question “could high concentrations of NaCl or KCl be combined with amoxicillin in order to treat H. pylori infection?” A direct combination of this type is highly unlikely for numerous reasons. For example, it is well established that high-salt conditions are associated with higher expression of virulence factors in H. pylori (22, 23, 25) and are associated with increased risk of gastric cancer (30) and other detrimental health effects (76). Furthermore, high osmotic stress from salt can increase antibiotic resistance in other bacteria (46, 49, 50). Thus, an H. pylori treatment that includes salt could result in an increase in antibiotic resistance for other bacteria in the gastrointestinal track. A therapeutic impact will likely only be feasible after we obtain a greater understanding of the mechanism of synergy between osmotic stress and beta-lactam antibiotics; it may then be possible to capitalize on this information to develop novel treatment strategies for this important pathogen.

As with all studies, it is worth noting that there are limitations to our work. First, while PBP2 and PBP3 have greater binding affinity than other PBPs for aztreonam and mezlocillin, respectively, the specificity of this binding is not absolute (14). Thus, aztreonam and mezlocillin may still have some effects on PBP1 that we are unable to distinguish. For this reason, we are not able to determine if the observed phenotypes are absolutely specific for PBP1 or also involve PBP2 and PBP3. Furthermore, while three different beta-lactam antibiotics were tested for synergy, additional beta-lactams and other cell wall-targeting antibiotics such as cephalosporins and bacitracin, could also be tested to determine the generalization of this phenomenon. Finally, in thinking about the applicability of our in vitro findings to an in vivo environment, it is worth noting that the length of exposure to the osmotic stress and antibiotics is likely dramatically longer on plates than it is in the human stomach. Thus, these results may not be translatable when tested in the animal models. Finally, the exact molecular mechanism responsible for the observed phenotypes remains to be determined. Despite these limitations, we believe that the results present interesting areas for future investigation that will provide further insight into stress adaptation and antibiotic resistance in H. pylori. Furthermore, it is interesting to speculate that once the mechanism of synergy between osmotic stress and beta-lactam antibiotics is understood, it may be possible to capitalize on this information to develop novel treatment strategies for this important pathogen.

MATERIALS AND METHODS

Bacterial strains and growth.

The strains and plasmids used in this study are listed in Table 1. Strains were grown on horse blood agar (HBA) composed of 4% Columbia agar (Neogen Corporation), 5% defibrinated horse blood (HemoStat Laboratories, Dixon, CA), 2 mg/mL β-cyclodextrin (Sigma), and an antibiotic-antifungal cocktail composed of 10 μg/mL vancomycin (Amresco), 5 μg/mL cefsulodin (Sigma), 2.5 U/mL polymyxin B (Sigma), 5 μg/mL trimethoprim (Sigma), and 8 μg/mL amphotericin B (Amresco), or in liquid brucella broth (BB) (Neogen Corporation) supplemented with 10% fetal bovine serum (FBS) (Gibco) and 10 μg/mL vancomycin. H. pylori cultures were grown at 37°C in gas evacuation jars under microaerobic conditions (5% O₂, 10% CO₂, and 85% N₂) generated with an Anoxomat gas evacuation and replacement system (Advanced Instruments, Inc.). Liquid cultures were grown with shaking at 110 rpm. Freezing medium consisting of brain heart infusion broth (BD Biosciences) containing 10% FBS and 20% glycerol (EMD Chemicals, Inc.) was used for the creation and storage of H. pylori −80°C stock cultures.

TABLE 1.

Strains used in this study

Lab designation	H. pylori strain description and notes	Reference(s)
DSM1	G27	77, 78
DSM1659	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-S417T	17
DSM1660	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-E406A	17
DSM1661	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-S414R	17
DSM1763	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-V374L	17
DSM1709	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-N562Y	17
DSM1716	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-T593A	17
DSM1717	G27 HP0203-HP0204::cat, chloramphenicol resistant, PBP1-A599G	17

Survival on various osmotic and/or antibiotic stressors.

The optical density at 600 nm (OD₆₀₀) of 20- to 24-h-old OD-controlled liquid cultures was determined and the volume of culture needed to reach a McFarland standard of 3 (OD₆₀₀ of 0.582) was added to 5 mL of BB plus10% FBS plus vancomycin. These cultures were then dilution plated onto HBA plates that contained additional osmotic stressors and/or antibiotics of interest: NaCl (EMD), KCl (EMD), sucrose (Sigma), amoxicillin (Sigma), aztreonam (Thermo Scientific), mezlocillin (Thermo Scientific), metronidazole (Fluka Analytical), clarithromycin (Sigma), and/or tetracycline (EMD). The utilized concentrations of the various supplements are provided in the figure legends. To compare the various osmotic stressors, the osmolarity of both KCl and sucrose were calculated to be consistent with the osmolarity used for NaCl. Briefly, since both NaCl and KCl disassociate into chloride and their respective metal ions, each mole of salt produces two osmoles of osmotic pressure. Since sucrose does not disassociate, the molar concentration was doubled to match the osmolarity of NaCl and KCl. The minimum bactericidal concentration (MBC) of the various stressors was defined as a 99.9% reduction in recoverable CFU compared to the average of the populations grown under the control condition. To determine if there were surviving H. pylori cells on plates that contained beta-lactam antibiotics plus NaCl, the surface of each HBA plate was swabbed with a sterile cotton swab, and samples were spread onto HBA plates containing no additional NaCl or antibiotic. Presented data represent a minimum number of three biological replicates.

Visual analysis of the response to osmotic and/or antibiotic stressors.

H. pylori images were captured as previously described (17) with modifications. Briefly, cultures from OD-controlled overnight liquid cultures were adjusted to a McFarland standard of 3. A 500-μL aliquot of the culture was then spread on the surface of HBA plates containing the following additional osmotic and/or antibiotic stressors: 86 mM NaCl, 86 mM KCl, or 171 mM sucrose and/or 0, 0.125, or 0.25 μg/mL amoxicillin. The lawns were allowed to grow for 48 h, after which the cells were swabbed from the plate surface and resuspended in 1 mL of 4% formaldehyde and incubated at room temperature for 30 min. The cells were then washed 3 times in PBS, and resuspended in 1 mL ultrapure water. A 20-μL aliquot of the resuspended cells was dried on a glass slide and then Gram stained. Pictures were taken at ×100 magnification with an Olympus BX60 microscope and a Spot Insight camera. The images were captured using the SPOT 5.6 program and interpreted using ImageJ. Three biological replicates of the experiment were performed; representative images are shown in Fig. 5.

Statistical analysis.

Statistical analyses were conducted with Prism version 6.0g (GraphPad Software, Inc.). For the NaCl, KCl, and sucrose data, one-way analysis of variance (ANOVA) of the log-transformed data with Dunnett’s correction was used to compare the various concentrations of osmolyte to the control. To determine the interactions between antibiotic and osmolyte, a two-way ANOVA of the log-transformed data with Tukey’s correction was used to make the within-salt-concentration and within-drug-concentration comparisons. Statistically significant within-drug comparisons are indicated within the graphs. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.