Research articleAstrobiology as a framework for investigating antibiotic susceptibility: a study of Halomonas hydrothermalis
Abstract
Physical and chemical boundaries for microbial multiplication on Earth are strongly influenced by interactions between environmental extremes. However, little is known about how interactions between multiple stress parameters affect the sensitivity of microorganisms to antibiotics. Here, we assessed how 12 distinct permutations of salinity, availability of an essential nutrient (iron) and atmospheric composition (aerobic or microaerobic) affect the susceptibility of a polyextremotolerant bacterium, Halomonas hydrothermalis, to ampicillin, kanamycin and ofloxacin. While salinity had a significant impact on sensitivity to all three antibiotics (as shown by turbidimetric analyses), the nature of this impact was modified by iron availability and the ambient gas composition, with differing effects observed for each compound. These two parameters were found to be of particular importance when considered in combination and, in the case of ampicillin, had a stronger combined influence on antibiotic tolerance than salinity. Our data show how investigating microbial responses to multiple extremes, which are more representative of natural habitats than single extremes, can improve our understanding of the effects of antimicrobial compounds and suggest how studies of habitability, motivated by the desire to map the limits of life, can be used to systematically assess the effectiveness of antibiotics.
1. Introduction
Most extreme ecosystems on Earth are characterized by combinations of several physical and chemical parameters that can limit microbial proliferation. By doing so, these parameters define the ‘habitable’ boundary space for life (i.e. the collective set of conditions able to sustain the activity of at least a single organism) [1]. Attempts to identify physico-chemical limits for life have often focused on individual parameters at a time (such as temperature or water activity). However, their net impacts on the biological permissiveness of extreme environments have attracted increasing interest as a topic in astrobiological research [1–6]. The combined effects of multiple extremes on selected microbial taxa have now been investigated on the level of both individual ecosystems and the Earth as a whole [7–12], as well as with reference to simulated Martian brines [13]. One of the ways in which this topic has been approached has involved culture-based experiments where a model strain(s) is exposed to several distinct permutations of environmental conditions, yielding information on growth ranges, rates and other properties, such as growth yields and cell viability [7–9,11].
Despite recent advances in our understanding of how multiple stress parameters influence the physical and chemical boundary space for life, microbial responses to several combinations of environmental extremes are yet to be explored. Antibiotic resistance within natural ecosystems and within humans, for example, has rarely been investigated from this viewpoint, despite being of major environmental and medical significance [14,15]. This is in direct contrast to fields like food microbiology, where growth and survival responses to several stress parameters have been intensively researched (e.g. [16–19]). Thus, a promising approach to understanding antibiotic resistance would be to systematically investigate life under multiple extremes and determine the effects of these extremes on susceptibility to selected antibiotics. Ultimately, such research might even allow us to more effectively use antibiotics by determining physical and chemical conditions that enhance their effectiveness.
Additionally to their medical uses, antibiotics can have a major influence on microbial multiplication and survival in the environment [15]. They have an ancient evolutionary origin and are distributed throughout several extreme habitats, including oligotrophic caves, permafrost, deep ocean waters and hypersaline sediments [20–24]. However, in comparison with clinical and biotechnological research, relatively little is known about the impacts of antibiotics on the habitability of natural ecosystems [15,22,25,26]. This is especially true for extreme environments, despite these habitats having already been recognized as important reservoirs of novel antimicrobial strategies and metabolic products [27–29], and certain antibiotic-resistant taxa having been proposed as extremophiles in their own right [30].
To date, research into the interplay between environmental conditions and antibiotic sensitivity has focused on individual parameters at a time, as well as a limited selection of known pathogens [31–35]. The impacts of antibiotic compounds on microorganisms are concentration-dependent [25,26,33,36], but they have also been found to be affected by several other parameters. For example, the availabilities of essential nutrients (e.g. iron) and oxygen can individually modify cellular sensitivity to antibiotics [32,33,35,37–40], which is of importance because life on Earth is mostly located within nutrient- and/or oxygen-poor ecosystems [41,42]. Other parameters, including high or low temperatures, osmotic and pH stress, as well as reactive oxygen species (ROS), are additionally known to affect antibiotic susceptibility [31,34,36]. Pseudomonas aeruginosa has been found to display increased tolerance of antibiotics under nutrient and oxygen starvation [32,38], and a positive correlation between salinity (NaCl) and antibiotic resistance has also been observed for several moderately halophilic strains [31]. Stressors such as supra-optimal temperatures, on the other hand, can decrease antibiotic tolerance in a number of taxa [36]. However, often the interactions between antibiotic exposure and other stress parameters are highly heterogeneous, with varying impacts on sensitivity having been reported depending on the strains and compounds in question [35,39].
To improve our knowledge of how physico-chemical extremes and antibiotic resistance interactively influence the habitability of natural environments, experiments are required where natural isolates are exposed to several antibiotics under diverse environmental conditions. Here, we used a series of growth assays to explore the susceptibility of the polyextremotolerant marine bacterium Halomonas hydrothermalis [43] to three antibiotics (ampicillin, kanamycin and ofloxacin) under several permutations of NaCl concentration, iron (Fe3+) availability and aerobic or microaerobic atmospheric composition. The growth and survival responses of H. hydrothermalis to these and other conditions have previously been characterized in the absence of antibiotics [9,11], providing a foundation for our study. Ampicillin, kanamycin and ofloxacin were examined owing to their different modes of action [35,44]. While ampicillin is a β-lactam antibiotic that prevents cell wall synthesis through inhibited transpeptidase activity, kanamycin is an aminoglycoside compound which interferes with protein synthesis by interacting with the 30S ribosomal subunit. Ofloxacin is a synthetic fluoroquinolone antibiotic that prevents cell division and repair by inhibiting DNA gyrase and topoisomerase IV. The data show how detailed studies of life under multiple-stress conditions can be used to advance our understanding of microbial ecophysiology and the effects of antimicrobial compounds within natural environments.
2. Material and methods
2.1. Bacterial strain
Halomonas hydrothermalis DSM 15725T [43] was obtained from DSMZ (Braunschweig, Germany) and was cultured using marine agar or marine broth (DSMZ medium no. 514), as described in [11]. The strain is Gram-negative and exhibits cell division at temperatures of 2–40°C (optimal growth reported at 30°C), total salinities of 0.5–22% (w/v) (optimal range of 4–7% w/v) and pH values of 5–12 (optimal range of 7–8) [43].
2.2. Antibiotics and minimal inhibitory concentration determination
Ampicillin (sodium salt), kanamycin (sulfate) and ofloxacin were obtained from Sigma-Aldrich (UK). Antibiotic stocks were created using saline water (1%, 3.5% or 11.8% w/v NaCl) and by filtering the solution through membrane filters with a pore size of 0.2 µm (Millipore, USA). The stocks were stored at −20°C and thawed immediately before use.
To determine the lowest antibiotic concentration that completely inhibited cell division (minimal inhibitory concentration, MIC) in H. hydrothermalis, a starter culture was incubated overnight at 30°C, using antibiotic-free marine broth (3.5% w/v NaCl, pH 7.5). The starter culture was diluted to give a cell density equivalent to an optical density at 600 nm (OD600) of 0.3, measured using a volume of 2.5 ml (Helios Gamma spectrophotometer; Thermo Spectronic, UK). MIC assays were initiated by adding starter culture to antibiotic-spiked marine broth (3.5% w/v NaCl, pH 7.5) at a ratio of 1 : 100 (v/v), using culture volumes of 300 µl in 96-well microplates covered with gas-permeable optical seals (4titude Ltd, Dorking, UK). Cultures (n = 3 for each culture condition) were incubated at 30°C for 72 h, using the following concentrations of ampicillin and kanamycin: 7.8, 15.6, 31.2, 62.5, 125, 250, 500 and 1000 µg ml−1 (concentrations based on [31]). For ofloxacin, a 10× dilution of this concentration range was used. The cultures were incubated inside a Synergy 2 microplate reader (BioTek Instruments Inc., Winooski, VT, USA), using continuous shaking (1080 r.p.m.). MICs were determined using OD600 measurements performed by the microplate reader, with both positive (antibiotic-free) and negative (medium-only) controls included (n = 3) [45].
2.3. Multiple-parameter assays
Following determination of MICs (§2.2), further assays were established to characterize the impacts of multiple environmental parameters on antibiotic sensitivity in H. hydrothermalis. The experimental design was based on [11], involving 12 permutations of salinity (1%, 3.5% or 11.8% w/v NaCl), Fe3+ availability (iron-rich or iron-starved) and (aerobic or microaerobic) atmospheric composition for each antibiotic type. Briefly, iron-depleted media were prepared by excluding Fe(III) citrate and adding 100 µM of diethylenetriaminepentaacetic acid (Sigma-Aldrich, UK; also see [35]). The iron-rich and iron-depleted media contained approximately 0.8 mg l−1 and approximately 0.1 mg l−1 iron, respectively, as determined by inductively coupled plasma–optical emission spectroscopy (Perkin Elmer Optima 5300 DV) using a VI CertiPUR multi-element standard solution (Merck, Germany). Cultures (n = 3 for each culture condition) were established using total volumes of 100 µl in sealed 96-well microplates covered with gas-permeable seals (§2.2). Cultures under microaerobic conditions (5% (v/v) O2, 10% (v/v) CO2 and 85% (v/v) N2) were incubated inside 2.5 l jars containing an Oxoid CampyGen envelope (Oxoid Ltd, Basingstoke, UK) [46], using continuous shaking (100 r.p.m.). For each antibiotic type, the highest antimicrobial concentration that permitted multiplication of H. hydrothermalis was used.
All cultures were incubated at 30°C, with cell density (OD600) monitored for 96 h using the microplate reader (§2.2). Aerobic cultures were incubated inside the reader and were continuously shaken (1080 r.p.m.), with medium-only controls (n = 2) included for each culture condition. Upon completion of the assays, pH values were measured for each well to determine whether antibiotic addition affected culture pH (values for antibiotic-free cultures in [11]).
2.4. Data analysis
Maximal OD600 values for each antibiotic type were compared using three-way analyses of variance (ANOVA), with salinity (1%, 3.5% or 11.8% (w/v) NaCl), Fe3+ availability (Fe-rich or Fe-depleted) and atmospheric composition (aerobic or microaerobic) as the fixed factors. The data were expressed as ratios (% of mean calculated for antibiotic-free controls; control values (n = 5) based on [11]). Linear model assumptions were evaluated by visual inspection of diagnostic plots and Levene's tests for the homogeneity of variance. While the three-way models constructed using raw data showed no significant deviations from variance homogeneity (p > 0.05), Box–Cox transformations [47] were also performed to meet model assumptions for each antibiotic type (λ = 0.61, 0.42 and 0.47 for ampicillin, kanamycin and ofloxacin, respectively). Non-significant terms (p > 0.05) were sequentially removed from the models, until simplified models including only significant (or nearly significant, i.e. p < 0.075) terms were obtained, unless higher-level terms were significant. Following ANOVAs, post hoc Tukey's honestly significant difference (HSD) tests were also performed for each antibiotic type. All analyses were completed using R v. 3.2.1 [48]. Levene's tests and transformations were completed using the libraries ‘car’ [49] and ‘MASS’ [50], respectively. Tukey's HSD tests were completed using the library ‘agricolae’ [51].
Additionally to comparisons of maximal OD600 values, specific growth rates (h–1; designated by µ) were calculated as previously described [9,11,52] and expressed as ratios of antibiotic-free controls. Owing to low OD600 values measured for several culture conditions involving antibiotic exposure, growth rates could be reliably determined only for a subset of the data and were compared qualitatively (§3.2). Times required to reach maximal OD600 values were determined for those culture conditions for which specific growth rates were estimated and were also compared qualitatively. Focusing the statistical analysis on maximal OD600 values enabled ANOVAs based on a balanced design. Additionally, in contrast with measurements performed at a specific sampling interval, maximal OD600 data account for variation in H. hydrothermalis growth trends (such as shifts in the duration of the lag phase) across the multiple culture conditions that were considered.
3. Results
3.1. Determination of minimal inhibitory concentrations
When cultured using aerobic, iron-rich media with an NaCl concentration of 3.5% (w/v), H. hydrothermalis exhibited cell division at the highest concentration of ampicillin used in this study (1000 µg ml−1). Concentrations of kanamycin and ofloxacin required for complete inhibition of growth (based on OD600 data) were 125 µg ml−1 and 6.25 µg ml−1, respectively. As such, 1000 µg ml−1 ampicillin, 62.5 µg ml−1 kanamycin and 3.12 µg ml−1 ofloxacin were used to establish growth assays involving multiple-parameter exposure. No antibiotic-induced shifts in medium pH were observed (for pH values of antibiotic-free controls, see [11]).
3.2. Responses to antibiotics under multiple environmental conditions
3.2.1. Ampicillin
Under aerobic conditions involving ampicillin exposure, H. hydrothermalis DSM 15725T was able to multiply at NaCl concentrations of 1%, 3.5% and 11.8% (w/v) in both iron-rich and iron-depleted media (figure 1a–c). Under a microaerobic (low O2 and high CO2) atmosphere, this ability was strongly suppressed by iron starvation (figure 2a–c). When these conditions were further coupled with sub- or supra-optimal salinities, growth was low or negligible (mean OD600 values of less than 0.05; figure 2a–c).
Figure 1. Growth curves of H. hydrothermalis incubated at 30°C under aerobic conditions. Data are shown for antibiotic-free (Ctrl; n = 5) and antibiotic-spiked (n = 3) variants of marine broth, with cells grown under iron-rich (Fe+) or iron-starved (Fe−) conditions. Ampicillin-, kanamycin- and ofloxacin-spiked media at (a,d,g) 1%, (b,e,h) 3.5% and (c,f,i) 11.8% (w/v) NaCl. Data are presented as means and standard errors of the means (s.e.). Figure 2. Growth curves of H. hydrothermalis incubated at 30°C under microaerobic conditions. Data are shown for antibiotic-free (Ctrl; n = 5) and antibiotic-spiked (n = 3) variants of marine broth, with cells grown under iron-rich (Fe+) or iron-starved (Fe−) conditions. Ampicillin-, kanamycin- and ofloxacin-spiked media at (a,d,g) 1%, (b,e,h) 3.5% and (c,f,i) 11.8% (w/v) NaCl. Data are presented as means and standard errors of the means (s.e.).
Increased sensitivity of H. hydrothermalis to ampicillin under a combination of microaerobic and iron-depleted conditions was further evidenced by a comparison of maximal OD600 values, expressed as ratios of antibiotic-free controls (three-way ANOVA followed by Tukey's HSD test; table 1 and figure 3a). Significant effects of salinity, iron availability and atmospheric composition on ampicillin sensitivity were observed individually and in combination (table 1), with differences being most pronounced between iron-depleted and microaerobic (‘Fe−/Aer−’) conditions, and other culture conditions (figure 3a). Although the impacts of ampicillin on maximal OD600 values did not differ between low and intermediate (or low and high) salinities under microaerobic and iron-depleted conditions (Tukey's HSD test, p > 0.05), Fe−/Aer− cultures incubated using 11.8% (w/v) NaCl showed an approximately ninefold increase in susceptibility compared with those at 3.5% (w/v) (mean maximal OD600 fractions of 4 ± 1% s.e. versus 35 ± 10% s.e., respectively; figure 3a).
Figure 3. Maximal OD600 values of H. hydrothermalis cultures (n = 3), expressed as ratios (%) of antibiotic-free controls. The cultures were incubated at 30°C and exposed to variants of marine broth spiked with (a) ampicillin, (b) kanamycin or (c) ofloxacin under iron-rich (Fe+) or iron-starved (Fe−) conditions, in the presence of an aerobic (Aer+) or microaerobic (Aer−) atmosphere. Data are presented as untransformed means and standard errors of the means (s.e.). Different letters above the bars indicate significant differences between samples exposed to a given antibiotic (p < 0.05, Tukey's HSD test following three-way ANOVA; see table 1 for ANOVA results). (Online version in colour.)
| antibiotic | effect | d.f. | residual d.f. | F-value | p-value |
|---|---|---|---|---|---|
| ampicillin | salinity (Sal) | 2 | 24 | 5.92 | 0.008 |
| Fe3+ availability (Fe) | 1 | 84.34 | <0.001 | ||
| atmospheric composition (Atm) | 1 | 32.05 | <0.001 | ||
| Sal × Fe | 2 | 7.63 | 0.003 | ||
| Sal × Atm | 2 | 2.287 | 0.123 | ||
| Fe × Atm | 1 | 119.37 | <0.001 | ||
| Sal × Fe × Atm | 2 | 3.43 | 0.024 | ||
| kanamycin | Sal | 2 | 28 | 222.31 | <0.001 |
| Fe | 1 | 3.64 | 0.067 | ||
| Sal × Fe | 2 | 2.86 | 0.074 | ||
| Fe × Atm | 2 | 3.87 | 0.033 | ||
| ofloxacin | Sal | 2 | 30 | 20.05 | <0.001 |
| Fe × Atm | 3 | 5.01 | 0.006 |
Owing to low maximal OD600 values measured for ampicillin-exposed cultures under a combination of Fe−/Aer− conditions and sub- or supra-optimal salinity, specific growth rates and times required to reach maximal OD600 values were not calculated for these treatments. With reference to the remaining culture conditions, mean growth rates ranged from 44% to 90% relative to ampicillin-free controls (figure 4a). Differences in the growth rates were in general agreement with those observed for maximal OD600 values (compare figures 3a and 4a). Ampicillin exposure also prolonged the mean durations required to reach maximal OD600 values, with 2- to 2.9-fold delays and up to 1.4-fold delays occurring under aerobic and microaerobic conditions, respectively (table 2).
Figure 4. Specific growth rates (h−1; designated by the letter μ) of H. hydrothermalis, expressed as ratios (%) of antibiotic-free controls. The cultures were incubated at 30°C and exposed to variants of marine broth spiked with (a) ampicillin, (b) kanamycin or (c) ofloxacin under iron-rich (Fe+) or iron-starved (Fe−) conditions, in the presence of an aerobic (Aer+) or microaerobic (Aer−) atmosphere. Growth rates were calculated from growth curves as described in the Material and methods. Data are presented as untransformed means and standard errors of the means (s.e.). Culture conditions for which growth rate fractions were not calculated due to low OD600 values are indicated as ‘n.d.’ (not determined). (Online version in colour.)
| antibiotic | NaCl (% w/v) | Fe availability/gas composition | time to max OD600 (h) | fold change relative to antibiotic-free controla |
|---|---|---|---|---|
| ampicillin | 1 | Fe+/Aer+ | 33 ± 2 | 2.5 ± 0.2 |
| Fe−/Aer+ | 59 ± 6 | 2.4 ± 0.3 | ||
| Fe+/Aer− | 66 ± 7 | 1.4 ± 0.2 | ||
| Fe−/Aer− | n.d. | n.d. | ||
| 3.5 | Fe+/Aer+ | 37 | 2.9 | |
| Fe−/Aer+ | 47 ± 2 | 2.5 ± 0.1 | ||
| Fe+/Aer− | 54 ± 13 | 1.1 ± 0.3 | ||
| Fe−/Aer− | 96 | no change | ||
| 11.8 | Fe+/Aer+ | 63 ± 2 | 2.0 ± 0.1 | |
| Fe−/Aer+ | 91 ± 4 | 2.5 ± 0.1 | ||
| Fe+/Aer− | 92 ± 5 | 1.3 ± 0.1 | ||
| Fe−/Aer− | n.d. | n.d. | ||
| kanamycin | 1 | all permutations | n.d. | n.d. |
| 3.5 | Fe+/Aer+ | 55 | 4.2 | |
| Fe−/Aer+ | 96 | 5.1 | ||
| Fe+/Aer− | 88 ± 5 | 1.8 ± 0.1 | ||
| Fe−/Aer− | 96 | no change | ||
| 11.8 | Fe+/Aer+ | 30 | no change | |
| Fe−/Aer+ | 48 | 1.3 | ||
| Fe+/Aer− | 52 ± 2 | 0.7 | ||
| Fe−/Aer− | 84 | 0.9 | ||
| ofloxacin | 1 | Fe+/Aer+ | n.d. | n.d. |
| Fe−/Aer+ | 79 | 3.2 | ||
| Fe+/Aer− | 84 | 1.8 | ||
| Fe−/Aer− | n.d. | n.d. | ||
| 3.5 | Fe+/Aer+ | 55 ± 18 | 4.2 ± 1.4 | |
| Fe−/Aer+ | 75 ± 6 | 3.9 ± 0.3 | ||
| Fe+/Aer− | 80 ± 5 | 1.7 ± 0.1 | ||
| Fe−/Aer− | 96 | no change | ||
| 11.8 | Fe+/Aer+ | 43 ± 11 | 1.4 ± 0.4 | |
| Fe−/Aer+ | 51 ± 5 | 1.4 ± 0.1 | ||
| Fe+/Aer− | 52 ± 2 | 0.7 | ||
| Fe−/Aer− | 96 | no change |
3.2.2. Kanamycin
The sensitivity of H. hydrothermalis to kanamycin was strongly affected by salinity, with antibiotic tolerance being positively correlated with NaCl concentration under both aerobic and microaerobic conditions (figures 1d–f and 2d–f). This was confirmed by comparisons of maximal OD600 values expressed as ratios of antibiotic-free controls (three-way ANOVA and a Tukey's HSD test; table 1 and figure 3b). Although turbidimetric assays indicated complete or near-complete suppression of cell division by kanamycin at 1% (w/v) NaCl, the cultures showed only minimal growth inhibition at 11.8% (w/v) NaCl (mean maximal OD600 fractions of 82–99%; figure 3b).
Comparing maximal OD600 values indicated only a nearly significant overall impact of iron availability on H. hydrothermalis kanamycin tolerance, but there was a significant interaction between iron availability and atmospheric composition (table 1). A post hoc Tukey's HSD test showed this to be specific to an NaCl concentration of 3.5% (w/v), with iron starvation-enhanced sensitivity occurring only under an aerobic atmosphere (mean maximal OD600 fractions of 64 ± 1% s.e. and 21 ± 13% s.e. under iron-rich and iron-depleted conditions, respectively; figure 3b).
Similar to the data shown in figure 3b, the ability of kanamycin to suppress H. hydrothermalis specific growth rates was more pronounced at 3.5% (w/v) than at 11.8% (w/v) NaCl (figure 4b). Growth rates and times to maximal OD600 values were not calculated for cultures incubated at 1% (w/v) NaCl due to low OD600 values. While maximal OD600 values did not differ between any of the cultures incubated at 11.8% (w/v) NaCl (figure 3b), kanamycin-induced reductions in growth rates were varied at this salinity (figure 4b). In particular, under microaerobic conditions, the mean growth rate of cultures incubated in iron-rich media was 62% (±2% s.e.) of that determined in the absence of kanamycin, while the corresponding value for iron-starved cultures was 115% (±9% s.e.). Increases in durations required to reach maximal OD600 values were also strongly affected by salinity, with greater delays (relative to controls) measured at 3.5% (w/v) NaCl than at 11.8% (w/v) (table 2).
3.2.3. Ofloxacin
Similar to assays involving ampicillin or kanamycin, salinity had a significant impact on the susceptibility of H. hydrothermalis to ofloxacin (three-way ANOVA, table 1; figures 1g–i and 2g–i). Tolerance appeared to increase with salinity, with a Tukey's HSD test showing significant differences between two of the culture conditions at 1% and 11.8% (w/v) NaCl (figure 3c). While responses to ofloxacin were varied at 1% and 3.5% (w/v) NaCl and there were no significant effects of iron availability or atmospheric composition on maximal OD600 values (expressed as ratios of antibiotic-free controls), there was a significant interaction between these factors (three-way ANOVA; table 1). Intriguingly, the OD600 data were indicative of an iron starvation-induced suppression in ofloxacin sensitivity under aerobic conditions, with this pattern being reversed under a microaerobic atmosphere (figure 3c). However, most of the between-sample differences were not statistically significant (p > 0.05, Tukey's HSD test) and only limited confidence could be placed on this finding.
In agreement with the maximal OD600 data (figure 3c), the specific growth rates of ofloxacin-exposed H. hydrothermalis cultures were slightly higher at 11.8% (w/v) NaCl than at lower salinities (rates expressed as ratios of antibiotic-free controls; figure 4c). This pattern was also observed for durations required to reach maximal OD600 values (table 2).
4. Discussion
Antibiotic resistance has become a problem of global concern [14]. Despite an abundance of data on the effects of antimicrobial compounds on different organisms, we still lack a good understanding of how multiple environmental factors interact to influence microbial susceptibility to antibiotics within natural ecosystems. Astrobiologists are interested in understanding the habitability of extreme environments on Earth and other planetary bodies, and thus their research motivates them to study the limits of life under single and multiple extremes [1–6]. This approach might be applied to investigate the efficacy of antibiotic compounds within natural extreme environments and in a medical context. Ultimately, it might lead to means to enhance antibiotic effectiveness through providing an empirical framework for determining how microorganisms respond to these compounds under diverse environmental conditions. Motivated by an interest in this topic, we studied the response of a polyextremotolerant bacterium, H. hydrothermalis DSM 15725T [43], to multiple extremes found in the natural environment. We carried out a series of laboratory-based growth assays to assess the individual and combined impacts of salinity, iron (Fe3+) availability and ambient gas composition (aerobic versus microaerobic) on the resistance of this strain to ampicillin, kanamycin and ofloxacin.
The cultures used in our study were able to multiply over a broad range of culture conditions when exposed to a high concentration (1000 µg ml−1) of ampicillin, despite a reduction in growth rates and increased durations of time required to reach maximal OD600 values, consistent with the resistance of several Halomonas spp. (including H. hydrothermalis VITP09) to this antibiotic [31,53]. This ability, however, was strongly suppressed when the strain was incubated under a combination of iron-deprived and microaerobic conditions at 1% or 11.8% (w/v) NaCl. A single previous study has investigated bacterial sensitivity to antibiotics under a combination of iron- and oxygen-limited conditions [37], but potential interactions between iron and oxygen availability were not systematically examined. Our findings show that, for ampicillin-exposed cultures, these factors exerted a greater impact on growth than salinity. Thus, the compounded impacts of multiple extremes on antibiotic resistance can exceed those exerted by individual stress parameters. We also observed interactive effects of iron availability and atmospheric composition when H. hydrothermalis was exposed to kanamycin or ofloxacin.
For all three antibiotics, differences in maximal OD600 values were accompanied by broadly corresponding differences in specific growth rates. Comparable patterns in durations required to reach maximal OD600 values were additionally observed, with greater delays (relative to antibiotic-free controls) occurring at lower salinities and under aerobic conditions. While the former affirms the increased sensitivity of the cultures to antibiotics at low salinity [31], the latter is likely to reflect resource limitation encountered by H. hydrothermalis under microaerobic conditions, which may have limited the relative differences observed between antibiotic-exposed and antibiotic-free cultures. Taken together, these data demonstrate that iron availability and atmospheric composition modulate the sensitivity of H. hydrothermalis to all three antibiotics by affecting several growth properties and not merely maximal cell densities. This is of importance because rates of cell division have previously been shown to correlate with the susceptibility of Burkholderia cepacea to ciprofloxacin and tobramycin, with positive and negative relationships between the growth rate and bacterial survival observed under oxygen-limited and oxygen-rich conditions, respectively [37]. Our study focused on measurements of properties other than survival. The available data, however, suggest that the relationship between H. hydrothermalis growth rates and the susceptibility of this strain to antibiotics is more complex than that reported for B. cepacea [37], with little evidence for an obvious overall difference between growth rate fractions measured under aerobic versus microaerobic conditions. Indeed, the growth rates calculated for antibiotic-exposed cultures showed pronounced variation, and in a limited number of cases were similar to the antibiotic-free controls or even exceeded them (figure 4; conditions with growth rate fractions exceeding 100%). Where differences between culture conditions were observed, these appeared more closely linked to interactions between the ambient gas composition and iron availability, as opposed to either of these parameters alone.
Further to the results discussed above, our experiments showed the influences of iron availability and atmospheric composition on antibiotic sensitivity (as well as their interplay with salinity) to be compound-specific. For example, while ampicillin-exposed cultures exhibited an iron starvation-mediated increase in sensitivity under microaerobic conditions, iron availability only affected the susceptibility of H. hydrothermalis to kanamycin under an oxygen-rich atmosphere. Moreover, iron availability and atmospheric composition only had an interactive influence on kanamycin tolerance at an NaCl concentration of 3.5% (w/v), with the data suggesting that increased tolerance of kanamycin at 11.8% (w/v) NaCl can counteract the impact of iron starvation observed under lower salinity. These types of complex interactions between stress parameters are likely to be important for understanding the distribution and activities of microorganisms within diverse habitats, and illustrate the role that antibiotic compounds are likely to play in this respect within extreme environments in particular.
Our experiments were aimed at characterizing phenotype-level responses of H. hydrothermalis to antibiotics under multiple-parameter conditions. While the overall physiological basis of the compounded impacts reported in this study has not been investigated, they are known to involve diverse mechanisms [31,32,34,35,37,39,40]. The surface charge and membrane permeability of bacterial cells, for example, have strong influences on antibiotic resistance and are affected by both salinity and nutrient availability [35,54,55]. Moreover, salt-inducible drug efflux pumps are known to play an important role in antibiotic resistance [56,57], and it is possible that such a system could be partially responsible for the observed patterns of antibiotic tolerance in H. hydrothermalis. Iron is also known to directly modulate the activity of several antibiotics, for example, via the formation of iron–antibiotic complexes which can lead to the production of ROS (reviewed in [35]). However, we found no evidence for increased sensitivity to ampicillin, kanamycin or ofloxacin under a combination of aerobic and iron-rich conditions in particular. Indeed, as combined impacts of iron availability and atmospheric composition on the temperature and salt tolerance of H. hydrothermalis have also recently been reported [11], the net impact(s) of these parameters on antibiotic resistance are likely to be attributable to biological (e.g. metabolic) processes [33,34,40]. A recent transcriptomics study suggested that overlapping stress responses (e.g. responses to hypoxia and osmotic stress) play a key role in determining antibiotic tolerance in P. aeruginosa [40]. A key aim for future research, therefore, will be to determine whether the patterns of antibiotic resistance reported in our study are attributable to a single general mechanism, or whether multiple specific responses are involved. As part of this, the potential de novo emergence of antibiotic resistance would also need to be accounted for, particularly given the late (more than 60 h) onset of H. hydrothermalis cell division under certain culture conditions (e.g. figure 1g).
To fully interpret the results discussed in our study, it is also important to consider the ecology of H. hydrothermalis and how the findings relate to wider research into how multiple-parameter conditions constrain microbial proliferation in the environment. The strain used in this work was originally isolated from low-temperature hydrothermal fluid at a depth of 2580 m in the South Pacific Ocean [43]. While deep-sea habitats are characterized by multiple physical and chemical extremes [9], little information exists on concentrations of antibiotic compounds within these environments (including the compounds used in our study). Deep-sea microorganisms are, however, known to produce antibiotics [58,59] and the occurrence of antibiotic resistance enzymes in these environments has been reported [60]. Steep spatial gradients in environmental conditions are common in hydrothermal ecosystems [61], and it is possible that these can also influence antibiotic susceptibility, which in turn could have implications for competitive interactions within these environments. The impacts of antibiotics and environmental stress parameters on microbial proliferation could therefore parallel those described for other antimicrobials (such as ethanol and glycerol), which can profoundly influence the ability of specific taxa to colonize and dominate natural habitats [62–65]. Complex impacts of multiple-parameter conditions on microbial cell division are also well documented in the field of food microbiology [16–19]. Indeed, sublethal food preservation methods (including exposure to osmotic stress) have previously been shown to alter antibiotic susceptibility in food-borne pathogens including Escherichia coli, Salmonella enterica and Staphylococcus aureus [36]. Our results suggest that such modification of antibiotic resistance is likely to be common within diverse natural ecosystems as well, including extreme environments that define the physical and chemical boundaries of the biosphere [1–6,12].
5. Conclusion
Research into the habitability of extreme environments has often focused on single parameters including, among others, access to liquid water, the availability of energy, temperature, salinity, pH, hydrostatic pressure and ionizing radiation [66–70]. A limited number of studies have also sought to determine how combinations of multiple extremes influence physical and chemical boundaries for microbial multiplication and survival [2–5,10,12,13]. One application of this research is to understand how complex interactions between environmental extremes affect microbial susceptibility to antibiotic compounds and, conversely, how these compounds may alter the responses of organisms to extreme conditions.
In this study, we examined how 12 distinct permutations of salinity, iron (Fe3+) availability and ambient gas composition (aerobic versus microaerobic) influence phenotypic resistance in H. hydrothermalis to ampicillin, kanamycin and ofloxacin. Our results show that some combinations of extremes enhanced susceptibility to antibiotics while others reduced it, and that these effects are compound-specific. They also show that unravelling these effects can only be achieved via multi-factorial experiments involving several combinations of environmental conditions. Although extremes of salinity are not of obvious medical relevance and we did not study a pathogen, iron and oxygen limitation are encountered within a medical setting (e.g. [37]). The results we obtained demonstrate the principle that astrobiological research can provide new data on the efficacy of antibiotics, including those environmental conditions which are most likely to enhance their effectiveness. To further improve our knowledge on this topic, antibiotic susceptibility assays could be complemented by modelling-based approaches, including response surface analysis for determining responses to multiple environmental parameters [16] and estimating growth/no-growth boundaries for selected model strains [17]. Such approaches are already common in the field of food microbiology and could help establish a more predictive understanding of how complex combinations of extreme conditions influence microbial sensitivity to antibiotics.
Finally, the findings have relevance to understanding the impacts and ecological roles of antibiotic compounds within the environment. Given the ubiquitous distribution of antibiotics and antimicrobial resistance [14,20–24], these compounds could even be hypothesized to play a role in determining boundaries for life on a global scale.
Authors' contributions
J.P.H. and C.S.C. designed the study; J.P.H. and R.A. analysed the data; J.P.H. performed the experiments and wrote the initial draft of the manuscript; all authors contributed to and approved the final version of the manuscript.
Competing interests
We declare we have no competing interests.
Funding
This work was supported by the UK Science and Technology Facilities Council (STFC) (Consolidated grant no. ST/1001964/1 and Consortium grant no. ST/M001261/1).
Acknowledgements
We thank the editor, three anonymous referees and Dr Stefano Romano for their valuable comments on the manuscript.
