In bacterial habitats, the ability to follow spatial gradients of environmental factors that affect growth and survival can be largely advantageous. The bacterial strategy for unidirectional chemotactic movement in gradients of typical attractants or repellents, such as nutrients or toxins, is well understood. Optimal levels of other factors, however, may be found at intermediate points of a gradient and thus require a bidirectional tactic movement towards the optimum. Here we investigate the chemotactic response of Escherichia coli to pH as an example of such bidirectional taxis. We confirm that E. coli uses chemotaxis to avoid both extremes of low and high pH and demonstrate that the sign of the response is inverted from base-seeking to acid-seeking at a well-defined value of pH. Such inversion is enabled by opposing pH sensing by the two major chemoreceptors, Tar and Tsr, such that the relative strength of the response is modulated by adaptive receptor methylation. We further demonstrate that the inversion point of the pH response can be adjusted in response to changes in the cell density.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Bacterial chemotaxis is well known to allow bacteria to navigate in gradients of physiologically relevant chemical ligands, such as nutrients or poisons, to find an optimal niche for growth. The overall chemotaxis strategy and molecular mechanisms that underlie unidirectional movement along gradients of positive (attractant) or negative (repellent) chemical stimuli have been extensively investigated (Wadhams and Armitage, 2004; Vladimirov and Sourjik, 2009; Sourjik and Wingreen, 2012). Chemical stimuli are sensed and processed by allosteric complexes of ligand-specific transmembrane receptors, a histidine kinase CheA and an adaptor protein CheW (Sourjik, 2004; Hazelbauer et al., 2008). Escherichia coli has five homologous receptors that act cooperatively in mixed sensory complexes, with Tar and Tsr being the most abundant (also called major) receptors. Activation or inactivation of the sensory complexes controls the rotational direction of the flagellar motor. Increased binding of attractant to the periplasmic domain of receptors inhibits autophosphorylation activity of CheA, which leads to decreased phosphorylation of the response regulator CheY and thus promotes lengthened runs towards high attractant concentrations. Increased binding of repellent induces an opposite effect, activating CheA and causing shorter runs up the repellent gradient. The chemotaxis pathway also includes a CheY-phosphatase CheZ, and an adaptation system consisting of a methyltransferase CheR and a methylesterase CheB, which add and remove methyl groups on four specific glutamyl residues in the cytoplasmic part of the receptors respectively. Methylation increases receptor activity and reduces its sensitivity to attractants, thereby offsetting the effects of attractant stimulation. It also provides a short-term memory for temporal comparisons of attractant concentrations by cells swimming in a gradient (Berg and Brown, 1972; Macnab and Koshland, 1972; Vladimirov and Sourjik, 2009; Sourjik and Wingreen, 2012).
A unidirectional mode of chemotaxis, from low to high attractant concentrations and from high to low repellent concentrations, is physiologically meaningful in the case of nutrient or toxin gradients. However, the chemotaxis machinery also enables bacteria to follow gradients of other environmental factors for which the physiological optimum may correspond to an intermediate rather than to the highest or lowest level in a gradient. An example of such response is pH taxis, which E. coli cells use to avoid both extremely acidic and extremely alkaline environments (Tso and Adler, 1974). This behaviour is likely to rely on opposing pH sensing by the two major receptors such that an increase in pH elicits an attractant response via Tsr but a repellent response via Tar (Slonczewski et al., 1982; Krikos et al., 1985; Khan et al., 1995). However, it remains unclear how this interplay of opposing responses allows cells to determine the preferred pH and whether the preferred pH may be subject to regulation.
Here we investigated intracellular processing of pH signals by the chemotaxis pathway, showing how the ‘push–pull’ mode of opposing pH sensing by Tar and Tsr allows cells to invert their response at a particular level of stimulation, and therefore to have a preferred pH value at which they accumulate in a gradient. Our analysis suggests that such inversion requires receptor-specific modulation of the response strength by the ambient pH, which is mediated by receptor methylation. We further demonstrate that the exact point of inversion is influenced by growth-dependent changes in the relative expression levels of Tar and Tsr and that the pH response depends on adaptation to chemoattractants. We propose that such a push–pull mode of sensing may be generally applicable and evolutionarily beneficial, allowing cells to flexibly adjust their environmental preferences.
Tar and Tsr mediate opposite pH responses
To better understand how pH stimuli are processed in E. coli chemotaxis, we first characterized responses mediated by each of the major receptors, Tsr or Tar, to step-like changes in external pH. The intracellular response of the chemotaxis pathway was measured using a reporter that relies on fluorescence resonance energy transfer (FRET) between CheY-YFP and CheZ-CFP (Sourjik and Berg, 2002a; Sourjik et al., 2007). Because the interaction of these two proteins requires CheY phosphorylation, higher activity of CheA kinase results in the increased energy transfer from the excited CheZ-CFP to CheY-YFP and as a consequence to higher ratio of the YFP fluorescence to CFP fluorescence. Consistent with previous observations (Krikos et al., 1985; Khan et al., 1995), when cells expressing Tar were pre-adapted at neutral pH of 7.0 and subsequently stimulated with steps of increasing or decreasing pH, they showed an attractant response (i.e. decrease in the kinase activity) to low and a repellent response (i.e. increase in the kinase activity) to high pH (Fig. 1A). An opposite response was observed for cells expressing Tsr (Fig. 1B). Both receptors were highly sensitive to small changes in pH, with a nearly saturating response being observed already at ΔpH of ∼0.2 (Fig. S1A and B). Similar to the adaptive response observed for chemical ligands, the initial change in the kinase activity upon a change in pH was followed by a slower recovery and the subsequent return to pH 7.0 elicited an opposite response. In addition to eliciting the specific chemotactic response, changes in pH had a weak direct effect on the fluorescence of YFP and CFP, which resulted in a modest decrease in the baseline ratio of YFP/CFP fluorescence at the higher pH (Figs 1A and B and S1D).
The observed adaptation to the pH stimuli resulted from changes in receptor methylation, which increased upon positive stimulation and decreased upon negative stimulation for both Tar and Tsr (Fig. 1D and E). As a consequence, Tar methylation decreased at higher pH (Fig. 1D), whereas Tsr methylation increased at higher pH (Fig. 1E), in agreement with a previous study (Slonczewski et al., 1982). Consistent with the role of the methylation system in adaptation to pH stimuli, no recovery of kinase activity was observed in cheR cheB cells (Fig. S2).
Although Tsr and Tar are able to sense both external (periplasmic) and internal (cytoplasmic) pH, changes in external pH are believed to be primarily sensed by the periplasmic domain of receptors (Krikos et al., 1985). In agreement with that prediction, the FRET response mediated by a hybrid Tsar receptor, which has the periplasmic domain of Tsr and the signalling domain of Tar, resembles the response of Tsr (Fig. S1C).
Response dependence on ambient pH and receptor methylation
We further characterized the pH response over a broad range of ambient pH values. In these experiments, pH was increased or decreased in a series of steps while allowing cells to adapt to current ambient pH before being stimulated with a next step. This type of measurement was previously used to determine the dynamic range of response for a number of chemoeffectors (Neumann et al., 2010). We observed that while the Tar-mediated pH response remained relatively unchanged over the tested range of pH, the Tsr-mediated response decreased and eventually ceased at high pH (Fig. 2A and B). We speculate that this effect is at least partly mediated by the increased methylation of Tsr upon adaptation to high pH (Fig. 1B). The dependence of the pH response on methylation could be directly demonstrated for both Tsr and Tar by adapting cells to the respective ligands, serine and MeAsp (Fig. 2C). As expected, receptor methylation desensitized or even inverted the pH response, with the effects for Tsr being more pronounced than for Tar.
Regulation of pH response in wild-type cells
The chemotactic response of wild-type E. coli cells that express a full set of receptors was consistent with the integration of opposing Tsr- and Tar-mediated responses. When pre-adapted at pH 7.0, wild-type cells showed a repellent response to decrease in pH and an attractant response to increase in pH (Figs 1C and S3), suggesting that at this ambient pH the response is dominated by Tsr. However, in this case the response depended on the ambient pH even more strongly than it did for individual receptors (Fig. 3A). Whereas an attractant response of gradually decreasing magnitude was observed to steps of increasing pH up to pH 7.5, no response was observed for the pH increase from 7.5 to 8.0. At even higher pH, the response inverted and turned repellent-like, becoming larger the farther away from the inversion point of 8.0. A mirror-image response of opposing sign was observed for a sequence of steps of decreasing pH, confirming that the inversion point of the response was between 7.5 and 8.0. As in the strains expressing individual receptors, the response of wild-type cells to pH changes was adaptive (Fig. 1C) and led to a weak but reproducible change in the methylation patterns of both receptors such that cells adapted to lower pH showed decreased methylation of Tsr, but increased methylation of Tar (Fig. 1F).
When compared with the pH preference of individual major receptors (Fig. 2), the observed dependence of the sign of the response on the ambient pH suggests that Tar dominates the response at high pH and Tsr dominates at low and neutral pH. This switch in dominance may be related to the inverse correlation between the strength of the pH response and receptor methylation, whereby a higher methylation level of Tar at low pH and of Tsr at high pH weakens – or even inverts – the response of the respective receptor. Consistent with such selective inactivation of receptors by methylation, we observed that the pH preference of the wild-type cells strongly depends on whether the cells have been adapted to attractants sensed by Tar or Tsr (Fig. 3B). The response became more Tsr-like upon adaptation to the Tar ligand MeAsp and more Tar-like upon adaptation to the Tsr ligand serine. This inactivation by methylation was particularly pronounced at low pH for Tsr and at high pH for Tar, consistent with dominance of the respective receptor in that region of pH.
Cell-density dependence of the pH preference point
The relative levels of receptors were previously shown to depend on the density of the cell culture (Salman and Libchaber, 2007; Kalinin et al., 2010), and we speculated that this change might affect the pH response. Indeed, the ratio of Tar to Tsr increased at high density of cells grown in tryptone broth (TB; Figs 4A and B and S4). The resulting larger contribution of Tar to pH sensing apparently shifted the inversion point of pH response to lower values (Fig. 4C). Interestingly, the observed acidophilic shift in the pH preference apparently correlated with increased alkalinity of the TB medium at high cell density (Fig. 4D), indicating that cells might purposely change their pH preference to avoid regions of high cell density. The change in pH of the medium was not, however, the cause of the change in tactic pH preference, as cells that were grown in pH-buffered TB and therefore did not experience an increase in pH still showed an increase in the Tar/Tsr ratio at high cell density (Figs 4B and D and S4).
Finding optimal conditions for cell growth and development is a challenge faced by all organisms, most of which solve it by active search and biased movement towards more favourable regions. In bacteria, this task is solved by chemotaxis, which allows bacteria to navigate in gradients of nutrients or toxins but also to follow gradients of such general physiologically relevant environmental stimuli as pH or temperature (Tso and Adler, 1974; Maeda et al., 1976) and of developmental signals such as autoinducers involved in quorum sensing (Hegde et al., 2011). However, accumulation at high concentrations of nutrients or avoidance of toxins can be achieved by a unidirectional movement in a gradient, whereas finding the optimal levels of such factors as pH requires accumulation at an intermediate point in a gradient and thus requires more sophisticated navigation mechanisms. Here we show that E. coli can specifically regulate not only the magnitude but also the sign of the tactic response to the changes in external pH depending on the value of ambient pH, inverting their response from alkaliphilic to acidophilic between pH 7.5 and 8.0. For cells in a gradient, such inversion creates a point for bidirectional accumulation. The preferred pH corresponds to the value of external pH (7.6) at which the values of external and internal pH are exactly matched (Slonczewski et al., 1981). The task of response inversion is solved by a specific push–pull mechanism, whereby the opposing and tunable responses mediated by the two major receptors Tar and Tsr allow cells to adjust their pH preference. We further propose that the switch from dominance of Tsr at low pH to the dominance of Tar at high pH is enhanced by the methylation dependence of the pH response mediated by each receptor (Fig. 5) so that the response is weaker at higher levels of methylation. Because an increase in pH is sensed as an attractant stimulus by Tsr but as a repellent stimulus by Tar (Krikos et al., 1985; Khan et al., 1995), higher ambient pH leads to an increased methylation of the former and a decreased methylation of the latter. As a result, the Tsr-mediated attractant response weakens as the Tar-mediated repellent response becomes stronger. Notably, the observed differential methylation of Tar and Tsr confirms recent analyses showing that the activity states of these two receptors are not perfectly coupled within the receptor lattice (Hansen et al., 2010; Lan et al., 2011) and demonstrates the benefits of such partial decoupling for the regulation of the tactic responses.
The preference for a particular value of external pH may need to be regulated to match the metabolic state of the cell. We observed that the inversion point of the pH taxis can indeed be tuned by culture density. At high density, cells grown in TB medium shift their preference to lower pH. This effect can be explained by the increasing ratio of Tar to Tsr with higher cell density (Salman and Libchaber, 2007; Kalinin et al., 2010), making Tar the dominant receptor. This preference for lower pH correlates with the increasing pH of the TB growth medium at high culture density. Under these conditions, the increased preference for acidic pH may benefit cells by allowing them to steer away from regions of high cell density where the nutrients are likely to be depleted. However, the change of the ratio of Tar to Tsr expression – and consequently in the pH preference – was not caused by the increased pH because the relative levels of receptor expression also changed in pH-buffered TB medium.
A similar push–pull mechanism may regulate the tactic preference in a number of other instances where it is advantageous for bacteria to accumulate at a well-defined intermediate stimulus levels. Several examples of bidirectional tactic accumulation towards an optimum have been described in E. coli and other bacteria, including redox taxis or aerotaxis (Alexandre, 2010; Schweinitzer and Josenhans, 2010), thermotaxis (Maeda et al., 1976) and magnetotaxis (Jogler and Schuler, 2009). A variant of the push–pull mechanism is likely to operate in these cases, and E. coli aerotaxis is indeed mediated by two receptors, Tsr and Aer, and results in preferred accumulation of the wild-type bacteria at a defined level of oxygen (Rebbapragada et al., 1997).
Strains and plasmids
All strains and plasmids used in this work are listed in Table S1. FRET pair CheY-YFP and CheZ-CFP was expressed from pVS88, a bi-cistronic construct (Sourjik and Berg, 2004). pVS1252, which expresses the Tsr–Tar chimera (Tsar), was constructed by replacing the periplasmic sensory domain of Tar with the sensory domain of Tsr in pHP2 (Pham and Parkinson, 2011) using NdeI and NotI restriction sites. The construct was then recloned into a salicylate-inducible plasmid pKG116 (Buron-Barral et al., 2006). Strain VS104 [Δ(cheY cheZ)] (Sourjik and Berg, 2002b) transformed with pVS88 was used as the wild type for FRET analysis. Receptorless strain VS181 [Δ(cheY cheZ) Δ(tar tsr tap trg aer)] (Sourjik and Berg, 2004) was transformed with pVS88 and a plasmid expressing either Tar (pVS1092), Tsr (pVS160) or Tsar (pVS1252).
For FRET experiments, cells were grown in TB (1% tryptone and 0.5% NaCl) supplemented with appropriate antibiotics (100 μg ml−1 ampicillin; 17 μg ml−1 chloramphenicol) overnight at 30°C, diluted 1:100 in 10 ml of fresh medium containing antibiotics and 50 μM isopropyl-β-D-thiogalactopyranoside (IPTG) to induce expression of the FRET pair, and grown to OD600 of 0.6 (unless specified otherwise) at 34°C and 275 r.p.m. Where required, receptor expression was induced by salicylate (0.7 μM for pVS160 and pVS1252, 2 μM for pVS1092).
Fluorescence resonance energy transfer measurements were performed as described before (Sourjik and Berg, 2002b; Sourjik et al., 2007; Neumann et al., 2010) on custom-modified Zeiss Axiovert 200 or Axio Imager.Z1 fluorescence microscopes. Cells were harvested by centrifugation (3200 g for 5 min), washed once with 10 ml of tethering buffer (10 mM KPO4, 0.1 mM EDTA, 1 μM methionine, 10 mM lactic acid, pH 7.0), resuspended in 9 ml of tethering buffer and stored at 8°C. For FRET experiments, cells were attached to a poly-lysine-coated coverslip and placed into a flow-chamber that was maintained under a constant flow (0.5 ml min−1) of tethering buffer using a syringe pump. Cells were allowed to adapt in the tethering buffer for at least 10 min and subsequently stimulated with specified levels of chemoeffectors or with different pH values in tethering buffer. The sample was excited at 436/20 nm by 120W EXFO X-Cite® 120 lamp (Axio Imager.Z1) or by 75W Xenon lamp (Axiovert 200) that were attenuated 500-fold or 550-fold, respectively, with neutral density filters. Fluorescence of a monolayer of 300–500 cells was continuously recorded in the cyan and yellow channels using photon counters with a 1.0 s integration time.
Levels and methylation of Tar and Tsr were analysed using immunoblotting with polyclonal αTar antibody raised against the conserved cytoplasmic region of Tar. This antibody recognizes Tar and Tsr with similar efficiency (Neumann et al., 2010). Cells were prepared as described above for FRET experiments, resuspended in 1× Laemmli buffer, lysed by incubation at 95°C for 5 min and subjected to electrophoretic separation using 8% or 10% SDS-PAGE. Protein samples were transferred to a 0.2 μm pore-size Hybond ECL nitrocellulose membrane using semi-dry blotter (Bio-Rad) at 2.5 mA cm−2 for 30 min. Receptors were detected with αTar at a 1:5000 dilution and IRDye®800-conjugated secondary antibody (Rockland) at a 1:5000 dilution. Signals from the membrane were detected using an Odyssey® imager (LI-COR). Receptor quantification was performed as described before (Neumann et al., 2010) using ImageJ software (http://rsbweb.nih.gov/ij/download.html), taking into account the 1.39-fold difference in the relative antibody sensitivity for Tar relative to Tsr (Neumann et al., 2010).
We thank Ned Wingreen, Yuhai Tu, Bo Hu and Matthias Mayer for discussions, and Michael Manson and Abiola Pollard for helpful comments on the manuscript. This work was supported by the China Scholarship Council fellowship for Y. Y. and by grant SO 421/11-1 from the Deutsche Forschungsgemeinschaft to V. S. Authors declare no conflict of interest.