Introduction

Bacteria can deal with unfavorable conditions in multiple ways, where the time and extent of the reaction is crucial for survival. One of the global and far-reaching mechanisms of adaptation to environmental conditions is the stringent response, where unusual nucleotides (guanosine tetra- and pentaphosphate, ppGpp, and pppGpp, collectively named (p)ppGpp), herald starvation, and various physicochemical stress (Cashel et al. 1996; Potrykus and Cashel 2008 and refs therein). These alarmones are produced promptly after the onset of stress, directly and indirectly affect major cellular processes such as adaptation to adverse conditions, sporulation, biofilm formation, quorum sensing and virulence (Jain et al. 2006; Potrykus and Cashel 2008; Dalebroux et al. 2010). (p)ppGpp has been identified in all free-living eubacteria tested so far (Potrykus and Cashel 2008; Atkinson et al. 2011) and chloroplast-bearing plants (Braeken et al. 2006). The enzymes responsible for its synthesis and degradation could be one dual-function or separate proteins, coded by rsh or rel (synthesis) and spo (hydrolysis) genes, respectively. In model bacterium, Escherichia coli and some of β- and γ-proteobacteria, two enzymes are responsible for (p)ppGpp accumulation: the monofunctional synthetase I encoded by relA gene and the bifunctional synthetase SpoT which have also (p)ppGpp hydrolase activity. RelA protein, associated with ribosomes, responds to amino acids starvation, while SpoT synthetase activity is induced by limitation of other nutrients or by stresses (Potrykus and Cashel 2008; Dalebroux et al. 2010). In addition, some short enzymes with only synthesis domains were identified in some bacterial species. This suggests a possible generalized ppGpp role in all living organisms. (p)ppGpp has a major impact on cellular metabolism and gene expression which are altered in order to survive and to use limited resources in the economical way, e.g., the expression of the growth-related genes is inhibited, while those coding for proteins connected with survival and efficient nutrient utilization are induced (Cashel et al. 1996; Magnusson et al. 2005; Potrykus and Cashel 2008). However, despite almost 50 years of study, many aspects of ppGpp-mediated regulation remain obscure or have been solved only recently e.g., the ppGpp influence on transcription or DNA replication (for review, see Potrykus and Cashel 2008). Most of the knowledge of the mechanisms of bacterial stress response comes from the studies on the model bacteria, Escherichia coli that can switch between free-living and intestine environment, soil bacterium Pseudomonas putida, or Gram-positive Bacillus subtilis.

The marine environments present an exceptionally severe challenge for free-living bacteria because of the varying nutrient supplies and other important factors. Thus, it can be hypothesized that the marine microorganisms are particularly well-adjusted to the altering conditions. In fact, marine bacteria are particularly well-adapted to an environment with limited resources; it has been documented that they can stop and resume their biological activities faster than bacteria that thrive in less restrictive environments (Amy et al. 1983; Kurath and Morita 1983). Understanding the physiological mechanisms of these adjustments is therefore crucial to reveal the molecular mechanisms underlying bacterial stress responses and ultimately leading to survival. However, the information on their stringent response is quite limited with few publications describing the stringent response of a single species, Vibrio sp. S14 identified later as V. angustum which can synthesize ppGpp during amino acid and carbon starvation (Flärdh et al. 1992, 1994; Östling et al. 1996). The presence of (p)ppGpp-metabolizing enzymes and the relevant genes was shown in the phylogenetic analyses of the genomes from microorganisms associated with marine environment (Guzow-Krzemińska et al. 2015). However, the mechanisms and the kinetics of the induction of the stringent response in marine bacteria have not been studied so far. Thus, we employed a series of various stresses to compare the stringent response of representative marine bacteria to the known model of E. coli. Importantly, the study on the stringent response in other than model bacteria is often more difficult because of the growth conditions and requirements for guanosine nucleotide labeling (low phosphate medium) and for stress introduction. For the quantification of bacterial alarmones, HPLC method has been employed (Varik et al. 2017); however, this method is relatively complex and has some serious limitations. Thus, it is important to prove that the standard procedure established already 50 years ago (Cashel et al. 1969) can be employed with some culturing modifications to environmental bacteria.

Material and methods

Bacterial strains

Bacterial strains used in this work are listed in Table 1.

Table 1 Bacterial strains used in this study
Full size table

Media and growth conditions

Bacteria were cultured in liquid LB broth (Sigma), Marine broth (Difco), MOPS minimal (morpholinepropanesulfonic acid) medium (for E. coli, S. baltica, A. johnsonii) (Mechold et al. 2013), or Tris-minimal medium (V. harveyi) (Kaempfer and Magasanik 1967), in shake flasks with agitation (150 rotation per min).

The carbon and nitrogen limitation stress were introduced by pelleting the bacterial cultures and resuspending in relevant minimal (MOPS or Tris) medium depleted of carbon or nitrogen source. The amino acid starvation was achieved by adding serine hydroxamate (SHX) to the cultures at 1.5 mg/ml. The heat stress was introduced by transferring bacterial cultures to water baths with the elevated temperature: 37 °C for S. baltica and A. johnsonii and 43 °C for E. coli and V. harveyi. The relevant temperatures were established by increasing culture temperature until the significant growth inhibition was observed (Table 1S). The osmotic stress was achieved by adding LiCl to the growth-inhibiting concentrations, established for each strain − 0.6 M for E. coli, 0.2 M for S. baltica, and 0.1 M for A. johnsonii and V. harveyi. The DNA damage stress was introduced by adding EMS (ethyl methanesulfonate) to the final concentration of 28 mM. The inhibition of DNA synthesis was started by adding mitomycin C to the final concentration of 1 mg/ml.

Analysis of (p)ppGpp synthesis

(p)ppGpp levels were measured as previously described, with minor modifications (Mechold et al. 2013). Briefly, bacteria were grown overnight in minimal MOPS (containing 5 mM KH2PO4) or Tris medium (containing 1 mM K2HPO4) and then washed and resuspended in low-phosphate (0.4 mM KH2PO4 or K2HPO4, respectively) MOPS or Tris-labeling medium at A600 of 0.02. Cultures were grown until they reached A600 of 0.2 and diluted again (1:10) in the same medium. [32P] orthophosphoric acid was added to 5,550,000 Bq/ml, and then bacteria were cultured for at least 2 generations before the first sample was taken. The stress conditions were implemented, and then at indicated time points, samples (50 μl) were taken and then extracted with ice-cold formic acid (13 M) by three cycles of freeze-thaw. Samples were centrifuged (5000 x g, 4 °C, 5 min), and nucleotides present in the supernatant were separated by thin-layer chromatography (TLC) on polyethylenimine (PEI) cellulose TLC plates in 1.5 M potassium phosphate buffer and analyzed in a phosphorimager (Typhoon; GE Healthcare). The spots corresponding to ppGpp and pppGpp were identified according to previously reported characteristics (Mechold and Malke 1997; Mechold et al. 2013).

Statistical analysis

Data were analyzed using the GraphPad Prism and PQStat softwares. Student’s t-test or one-way ANOVA, followed by Kołmogorow-Smirnow or Shapiro-Wilk tests, respectively, was used to determine statistical significance of the obtained results. The equality of variances was checked using the Fisher-Snedecor and Brown-Forsythe tests. To check the differences between the groups, a post hoc test by Tukey was performed. Difference was considered significant at p < 0.05.

Results and discussion

The choice of bacterial strains and stress conditions

To ensure survival, bacteria as unicellular organisms need to react promptly and precisely to the alteration of environmental factors. For bacteria living in the marine habitats, these changes involve both physicochemical and nutrient conditions. To address the question about the adaptive regulation upon stress, we chose to study the reaction to a variety of stresses: carbon and nitrogen source limitation, amino acid depletion by modifying the medium composition, DNA damage, temperature, and osmotic challenge by varying growth parameters. Nutrients are usually quite sparse in most of cold-water marine habitats; however, the availability of them may rapidly increase especially in the enclosed ecosystems such as the Baltic Sea due to floods and river-mediated depositions from agricultural areas. Moreover, the Baltic Sea is exposed to recurrent seasonal variations of salt and anthropogenic pollutions, including various chemical mutagens. Bacterial strains used in this work are listed in Table 1. The choice of bacterial species for this study was based upon several factors: the representation in the natural population of the marine environment, in particular, the Baltic Sea and the possibility to study the selected species under defined laboratory conditions. The bacterial isolates from the Baltic Sea from the collection of Department of Molecular Biology, University of Gdańsk and results of the analysis by Moskot et al. (2012) supported this choice. As a control, Escherichia coli was used due to several reasons: this bacterium serves as a model in molecular biology, and its stringent response was extensively studied; moreover, E. coli belongs to proteobacteria as the marine representatives. The strain of Shewanella baltica M1 was isolated from the Baltic Sea, and its full genome was sequenced in our lab (Karczewska-Golec et al. 2016). S. baltica as other species from Shewanella genus plays an important role in the turnover of the organic substances as one of the most important denitrifying strains in the redox interface (Ziemke et al. 1997). Acinetobacter johnsonii is a Gram-negative bacterium with broad range of habitats, quite common in the low-salinity water (Guardabassi et al. 1999; Doughari et al. 2011). Its high flexibility in terms of environments suggest the efficient way of dealing with stresses, e.g., by the stringent response. Vibrio harveyi is motile marine bacterium, free-living and facultative pathogen that can grow in the wide range of temperatures; however, its efficient growth requires NaCl (Farmer III et al. 2005; Owens and Busico-Salcedo 2006). All three bacterial species have a genetic basis for the stringent response that is homologs of rsh genes encoding two paralogue enzymes, similarly to E. coli. The comparison of enzymes amino acid sequences of S. baltica revealed 60% identity for RelA homolog, Sbal_3146, and 70% for SpoT homolog, Sbal_0353 with E. coli counterpart (Figs. 1–2 S). A. johnsonii enzymes (RZ95_RS13970 and RZ95_RS01580) revealed quite low level of identity with E. coli RelA and SpoT proteins (39 and 34 %, respectively); however, the identity at the nucleotide level was considerably higher (63 and 69%, respectively) (Figs. 3–4 S). V. harveyi homologs bear 65% identity to RelA (VIBHAR_03528) and 75% to SpoT (VIBHAR_00627) (Figs. 5–6 S). Numerous species from Vibrio genus have additional protein involved in (p)ppGpp synthesis, RelV. Analysis of V. harveyi genome sequence revealed presence of relV-related gene, VC1224. The protein hypothetically encoded by this gene, VIBHAR_RS09915, shows 71% identity with RelV of V. cholerae (Fig. 7 S). All analyses were performed using BLAST. These analyses confirmed that all studied bacteria have an ability to produce (p)ppGpp upon various stresses; thus we proceeded to address the question whether the reaction of marine bacteria to adverse conditions show their specific adaptations to unfavorable environment.

The stringent response induction during nutrient limitation stresses

The typical marine habitat does not provide an abundance of nutrients albeit bacteria employ various methods of utilizing unusual energy and compounds sources. The most common is a deficit of carbon and nitrogen sources. First, we assessed the growth in defined minimal MOPS (morpholinepropanesulfonic acid) medium containing 5 mM KH2PO4 (for E. coli, S. baltica, A. johnsonii) (Mechold et al. 2013) or Tris minimal medium containing 1 mM K2HPO4, (V. harveyi) (Kaempfer and Magasanik 1967), at 30 °C in shake flasks with agitation (150 rotation per min). The generation times, which turned out to be 30 min for E. coli, 60 min for S. baltica and A. johnsonii, and 90 min for V. harveyi, have been assessed in the experiments preceding labeling assays. For efficient labeling of nucleotides, bacteria have to be cultured in limiting phosphate medium, containing 0.4 mM KH2PO4 or K2HPO4, respectively. To check whether these media do not provoke stress in bacteria, we evaluated the (p)ppGpp levels during prolonged cultures up to 210 min. The alarmones synthesis could be observed from 180 min, likely due to the exhausting of available nutrients (data not shown). Thus, the maximal time of experiment was established as 120 min. To provide the limiting growth conditions, we depleted the medium of C and N sources and continued cultures for additional 120 min taking samples every 30 min. Interestingly, only V. harveyi accumulated ppGpp during carbon starvation (Fig. 1A), while nitrogen limitation resulted in the induction of the stringent response in E. coli and S. baltica (Fig. 1B). The amino acid starvation was achieved by adding the toxic amino acid analog serine hydroxamate (SHX) to the cultures at 1.5 mg/ml, and the (p)ppGpp synthesis was observed in all bacterial species (Fig. 2), with the nearly maximum accumulation at 30 min. Notably, the presence of guanosine pentaphosphate was noted only for E. coli and in lesser extent in S. baltica. Apparently, from tested nutrient deficiencies, the amino acid limitation constitutes the strongest inducing factor for stringent response. The Rel-mediated reaction to the presence of uncharged tRNA bound to ribosomes triggers promptly (p)ppGpp synthesis in order to stop energy-consuming processes such as stable RNA production. The marine bacteria response to amino acid starvation show that their Rel homologs sense this limitation and react similarly to E. coli RelA. In contrary, carbon or nitrogen starvation induce only moderate (p)ppGpp accumulation limited to some bacterial strains. This may suggest that marine bacteria are relatively resistant to these limitations which may be a part of their adaptation to the frequent famine periods. Interestingly, it was reported that Vibrio sp. responded to the nitrogen limitation with growth inhibition only after 30 h (Nyström et al. 1992). Just in opposite, the carbon limitation was the factor that induced (p)ppGpp accumulation only in V. harveyi. These observations point to varying sensitivity of marine bacteria to nutrient deficiency.

Fig. 1
figure 1

The stringent response induction by the carbon (A) and nitrogen (B) limitation. The upper panels present an example of the thin-layer chromatogram with ATP, GTP, and ppGpp positions indicated. At the time 0, bacteria were transferred to the media depleted of C and N source. The lower panels show kinetics of relative (p)ppGpp synthesis in V. harveyi (magenta squares) (A), E. coli (blue circles) and S. baltica (red triangles) (B) based on the densitometrical analysis. The relative (p)ppGpp level was calculated as a ratio of (p)ppGpp amount to the sum of GTP and (p)ppGpp (set as 100 %). The results are from at least three independent experiments. The statistical significance of differences in (p)ppGpp amount compared with time 0 was determined by t-test (*p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001) as indicated above or below plots.

Full size image
Fig. 2
figure 2

The stringent response induction by the amino acid starvation. The left panel presents an example of the thin-layer chromatogram with GTP and ppGpp positions indicated. SHX (1.5 mg/ml) was added in the time 0. The right panel shows kinetics of relative (p)ppGpp synthesis in E. coli (blue circles) and S. baltica (red triangles), A. johnsonii (green inverted triangles), and V. harveyi (magenta squares) based on the densitometrical analysis. The relative (p)ppGpp level was calculated as a ratio of (p)ppGpp amount to the sum of GTP and (p)ppGpp (set as 100%). The results are from at least three independent experiments. The statistical significance of differences in (p)ppGpp amount compared with time 0 was determined by t-test (*p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001) as indicated above or below plots.

Full size image

The response of marine bacteria to the DNA damaging agents

The free-living bacteria in their environment encounter numerous factors that could either block DNA replication or damage genetic material such as mutagenic pollutants or UV irradiation. The direct effect is usually induction of the SOS response; however, the stringent response could also be involved in overcoming the DNA damage. ppGpp was suggested to facilitate DNA repair by exposing DNA breaks while bound to RNA polymerase paused at the damage site (McGlynn and Lloyd 2000). We employed mitomycin C (in the final concentration of 1 mg/ml) as DNA damaging agent which may subsequently interfere with DNA replication and observed prompt reaction by ppGpp accumulation to this stressor in S. baltica and A. johnsonii, while E. coli and V. harveyi exhibited more delayed yet significant response in further time points (Fig. 3A). Another DNA damaging agent, EMS (ethyl methanesulfonate), at the final concentration of 28 mM, provided most pronounced response in V. harveyi; nevertheless the ppGpp accumulation was delayed with the highest levels at 120 min after implementation of the stress condition (Fig. 3B). Our results suggest that the stringent response mediates the reaction to DNA damage stress in marine bacteria, facilitating their survival. However, the ppGpp accumulation may not be a first line of defense in the environmental induced DNA damage agents.

Fig. 3
figure 3

The stringent response induction by the DNA damaging agents, mitomycin C (A), and EMS (B). The upper panels present an example of the thin-layer chromatogram with GTP and ppGpp positions indicated. The mitomycin C (1 mg/ml) (A) or EMS (28 mM) (B) was added at the time 0. The lower panels show kinetics of relative (p)ppGpp synthesis in E. coli (blue circles), S. baltica (red triangles), A. johnsonii (green inverted triangles), and V. harveyi (magenta squares) based on the densitometrical analysis. The relative (p)ppGpp level was calculated as a ratio of (p)ppGpp amount to the sum of GTP and (p)ppGpp (set as 100%). The results are from at least three independent experiments. The statistical significance of differences in (p)ppGpp amount compared with time 0 was determined by t-test (*p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001) as indicated above or below plots.

Full size image

The stringent response induction by physicochemical stresses

Aquatic environment is a subject to alteration in basic characteristics such as salinity or temperature. The temperature changes not only seasonally but also depending on the water depth. The marine bacteria could be sensitive to both elevated and reduced water temperature, while high temperature can be more detrimental. The heat stress was introduced by transferring bacterial cultures to water baths with the elevated temperature: 37 °C for S. baltica and A. johnsonii and 43 °C for E. coli and V. harveyi. We observed that all analyzed bacteria responded by increased (p)ppGpp synthesis already after 30 min of stress implementation with the highest response of S. baltica (Fig. 4A). This observation could be explained by relatively low temperatures of the Baltic Sea which is the most common habitat where this bacterium was identified.

Fig. 4
figure 4

The stringent response induction by the temperature (A) and osmotic (B) stress. The upper panels present an example of the thin-layer chromatogram with GTP and ppGpp positions indicated. At the time 0, bacteria were transferred to the elevated temperature (43 °C for E. coli and V. harveyi and 37 °C for S baltica and A. johnsonii) (A), or LiCl (0.6 M for E. coli and 0.1 M for A. johnsonii) was added. The lower panels show kinetic of relative (p)ppGpp synthesis in E. coli (blue circles) and S. baltica (red triangles), A. johnsonii (green inverted triangles) and V. harveyi (magenta squares) based on the densitometrical analysis. The relative (p)ppGpp level was calculated as a ratio of (p)ppGpp amount to the sum of GTP and (p)ppGpp (set as 100%). The results are from at least three independent experiments. The statistical significance of differences in (p)ppGpp amount compared with time 0 was determined by t-test (*p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001) as indicated above or below plots.

Full size image

The salinity level in marine environment may fluctuate due to seasonal variations, water inflows, and currents. Bacteria need to adapt to survive these changes. In the laboratory conditions, we implemented increasing LiCl concentrations, established for each strain − 0.6 M for E. coli, 0.2 M for S. baltica, and 0.1 M for A. johnsonii and V. harveyi as a concentration causing significant growth inhibition. The choice of this salt was accounted for technical problems with thin-layer chromatography when NaCl was used. LiCl was reported to be employed as osmotic stress factor in the concentrations usually lower than NaCl (Serrano 1996). The introduction of the growth-inhibiting concentrations of LiCl resulted in the significant (p)pGpp synthesis in E. coli and moderate in A. johnsonii (Fig. 4B) with no response in other strains. The strong response to the osmotic stress in E. coli may be attributed to the typical habitat of these bacteria, where the conditions could change from stable habitat of intestines of homeothermic animals to the free-living status (e.g., water, liquid wastes). The marine bacteria can apparently cope with changes in salinity in the way not involving the stringent response.

Conclusions

The studied representative marine bacteria species were shown to be able to induce the stringent response upon the onset of stress and limitation conditions. The adjustment of growth and labeling conditions allowed for application of the standard method of nucleotide visualization to assess the time and level of the (p)ppGpp accumulation upon stress for environmental bacteria. Using this method, it was possible to quantitatively evaluate the accumulation of the stringent response alarmones in comparison to the control bacterium, E. coli. We showed that the extent of (p)ppGpp synthesis and time of the response vary depending of the stress conditions (summarized in Fig. 5). The highest (p)pGpp synthesis was stimulated by amino acid starvation for all tested bacteria, while nitrogen and carbon limitation did not efficiently induce the stringent response effector accumulation. From the physicochemical stresses, temperature stress was the most universal in inducing (p)pGpp synthesis, while only E. coli and A. johnsonii responded to osmotic stress. The DNA damage triggered (p)pGpp synthesis; however, the total alarmone production was moderate. The most sensitive to most of tested stress conditions proved to be A. johnsonii, while S. baltica did not respond to some conditions. The level of the stringent response may correspond with the ability of bacteria to efficiently confront adverse conditions common in marine environment.

Fig. 5
figure 5

The summary of (p)ppGpp synthesis upon stringent response induction by various stressors. The stress conditions are indicated below graph, with the data from 60 min time point after onset of stress for SHX and 120 min for other stresses. The (p)pGpp production in E. coli (blue) and S. baltica (red), A. johnsonii (green), and V. harveyi (magenta) was calculated as a ratio of (p)ppGpp amount to the sum of GTP and (p)ppGpp (set as 100%). The statistical significance of differences between respective strains compared to E. coli as a control was determined by t-test or one-way ANOVA (*p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001; ns nonsignificant) as indicated above columns (colors correspond to the column representing relevant strain compared to E. coli).

Full size image