In search of higher energy: metabolism-dependent behaviour in bacteria


  • Barry L. Taylor,

    1. Department of Microbiology and Molecular Genetics, School of Medicine, Loma Linda University, Loma Linda, CA 92350, USA.,
    2. Center for Molecular Biology and Gene Therapy, School of Medicine, Loma Linda University, Loma Linda, CA 92350, USA.
    Search for more papers by this author
  • Igor B. Zhulin

    1. Department of Microbiology and Molecular Genetics, School of Medicine, Loma Linda University, Loma Linda, CA 92350, USA.,
    Search for more papers by this author

Barry L. Taylor. E-mail; Tel. (909) 824 4480; Fax (909) 824 4035.


Bacteria use different strategies to navigate to niches where environmental factors are favourable for growth. Chemotaxis is a behavioural response mediated by specific receptors that sense the concentration of chemicals in the environment. Recently, a new type of sensor has been described in Escherichia coli that responds to changes in cellular energy (redox) levels. This sensor, Aer, guides the bacteria to environments that support maximal energy levels in the cells. A variety of stimuli, such as oxygen, alternative electron acceptors, light, redox carriers that interact with the electron transport system and metabolized carbon sources, effect changes in the cellular energy (redox) levels. These changes are detected by Aer and by the serine chemotaxis receptor Tsr and are transduced into signals that elicit appropriate behavioural responses. Diverse environmental signals from Aer and chemotaxis receptors converge and integrate at the level of the CheA histidine kinase. Energy sensing is widespread in bacteria, and it is now evident that a variety of signal transduction strategies are used for the metabolism-dependent behaviours. The occurrence of putative energy-sensing domains in proteins from cells ranging from Archaea to humans indicates the importance of this function for all living systems.


Motile microorganisms actively seek nutrient-rich microenvironments, moving to another site after they deplete the surrounding nutrients. Escherichia coli uses two signal transduction strategies to migrate to preferred microenvironments. In metabolism-independent chemotaxis, a change in the concentration of molecules binding to transmembrane receptors constitutes the signal that elicits a behavioural response. In metabolism-dependent behaviours, the signalling molecules must be metabolized to produce a signal. The term ‘chemotaxis’ usually refers to metabolism-independent behaviour, and we use ‘energy taxis’ to denote metabolism-dependent behaviour. Recent studies on metabolism-dependent behaviour in E. coli confirmed that internal transducers sense the change in cellular energy that results from metabolism of the signalling molecules. Energy taxes include aerotaxis (the response to oxygen) (Taylor, 1983a), electron acceptor taxis (Taylor et al., 1979), taxis to rapidly metabolized substrates (Zhulin et al., 1997a; Jeziore-Sassoon et al., 1998), redox taxis (Bespalov et al., 1996) and phototaxis (the response to light) (Armitage et al., 1985; Spudich, 1997).

Metabolism-independent chemotaxis was demonstrated by Adler (1969) and has been extensively investigated since then. The transduction pathway has been elucidated and these findings have been reviewed recently (for example, see Parkinson, 1993; Stock and Surette, 1996). This review describes briefly the chemotaxis pathway but focuses on progress in identifying the transduction strategy for the various energy taxes.

Chemotaxis in E. coli: sensing and signalling without metabolic changes

Evidence that chemotaxis in E. coli is independent of metabolism came from biochemical and genetic investigations (Adler, 1969). Galactose is a chemoattractant for E. coli whether or not the strain metabolizes galactose. Non-metabolizable analogues of attractants can also be attractants. Specific mutations in a chemotaxis receptor impair chemotaxis to attractants sensed by the receptor but do not affect metabolism of the attractants. Some attractants bind to the binding protein of an ABC-type transport system, but the pathways for chemotaxis and transport diverge, depending on whether the binding protein subsequently attaches to a chemotaxis receptor or a transport protein.

There are four homologous chemotaxis receptors in E. coli expressed by the tsr, tar, trg and tap genes (Stock and Surette, 1996). Each receptor has two transmembrane sequences that demarcate a periplasmic sensing domain and a cytoplasmic signalling domain. The dimeric active form of the periplasmic domain of the Tar receptor is a 4-helix bundle that binds aspartate asymmetrically (Milburn et al., 1991). Linear displacement of the α-4 helix by aspartate induces a conformational change in a highly conserved signalling domain in the interior of the cell (Fig. 1). The K1 and R1 coiled-coil methylation regions flank the highly conserved domain. Adaptation to aspartate occurs when methylation of the K1 and R1 sites by the CheR methyltransferase restores signalling by the receptor to the pre-stimulus state (Stock and Surette, 1996). Methyl groups can be removed by the CheB methylesterase. The homologous Tsr, Trg and Tap receptors are believed to have a signal transduction mechanism that is similar to Tar.

Figure 1.

. Metabolism-independent chemotaxis in E. coli to removal of aspartate. Signal transduction events in excitation include dissociation of aspartate from the receptor, autophosphorylation of CheA, phosphorylation of CheY and binding of phosphorylated CheY to the flagellar motor. CheY is dephosphorylated by the CheZ phosphatase. The adaptation pathway includes phosphorylation of the CheB methylesterase by CheA and demethylation of the receptor by CheB. Addition of an attractant promotes dephosphorylation of CheA, CheY and CheB. Asp, aspartate; A, CheA; B, CheB; P, phosphate; R, CheR; W, CheW; Y, CheY; Z, CheZ.

Enteric bacteria swim by rotating helical flagella. They randomly change direction by briefly reversing the direction of rotation of the flagellar motors. The signalling pathway from the chemotaxis receptor to the switch that controls the flagellar motor is a histidine kinase/response regulator phosphorylation cascade (Fig. 1) that is homologous to the other two component regulatory systems in bacteria. The CheA histidine kinase is autophosphorylated by ATP, followed by transfer of the phosphate to the cognate response regulator CheY. The binding of phospho-CheY to the switch on the flagellar motor is the signal for reversal of the motor. CheA and a docking protein CheW form a complex with the chemotaxis receptor, enabling the receptor to control the autophosphorylation of CheA, and thereby the phosphorylation of CheY. Adaptation to a decrease in attractant concentration occurs when cheA phosphorylates the methylesterase CheB, stimulating demethylation of the receptor (Stock and Surette, 1996).

A special case of chemotaxis is the response to sugars transported by the phosphoenolpyruvate–phosphotransferase system. Transport, but not metabolism of the sugar attractant is required for the behavioural response. During transport, the sugar is phosphorylated by a phosphate group donated by phosphoenolpyruvate and relayed by the HPr and Enzyme I proteins to the sugar-specific Enzyme II transport protein. Enzyme I interacts with CheA and the phosphorylation state of Enzyme I regulates autophosphorylation of CheA (Lux et al., 1995).

Aerotaxis requires metabolism of oxygen

Aerotaxis was discovered in bacteria more than a hundred years ago (reviewed by Taylor, 1983a,b). Motile bacteria migrate to a microenvironment where the oxygen concentration is optimal for growth. They are repelled by oxygen concentrations that are higher or lower than the preferred concentration. The preferred oxygen concentration has been determined for several bacterial species. The obligate aerobe Bacillus subtilis seeks oxygen concentrations in the range of 200 μM (Wong et al., 1995), the facultative anaerobe E. coli in the range of 50 μM (V. A. Bespalov, I. B. Zhulin and B. L. Taylor, unpublished), the microaerophil Azospirillum brasilense in the range of 4 μM (Zhulin et al., 1996) and the aerotolerant anaerobe Desufovibrio vulgaris in the range of 0.4 μM (Johnson et al., 1997). When it is known, the preferred oxygen concentration in aerotaxis correlates with the oxygen affinity of the major terminal oxidoreductase. Aerotactic behaviour requires a functional electron transport system. Strains that do not reduce oxygen, such as a cyo cyd mutant of E. coli, lack aerotaxis (Rebbapragada et al., 1997). Although reduction of oxygen is required for aerotaxis, oxygen is not the primary signal. Experimental evidence has suggested that the signal for aerotaxis is a change in electron transport and proton motive force during oxygen reduction (Taylor, 1983a,b). Where the relationship between oxygen concentration and proton motive force was investigated, e.g. A. brasilense (Zhulin et al., 1997a) and E. coli (V. A. Bespalov, I. B. Zhulin and B. L. Taylor, unpublished), the bacteria generate maximal proton motive force at the preferred oxygen concentration. Thus, aerotaxis navigates bacteria to the optimal oxygen concentration for energy production and growth. The receptor that senses changes in electron transport/proton motive force has remained unidentified until recently.

Phototaxis: responses to photosynthetic electron transport or light-driven proton pumping

Phototaxis in photosynthetic bacteria also requires metabolic changes. Studies of Rhodobacter sphaeroides (Armitage et al., 1985) and Rhodospirillum centenum (Romagnoli et al., 1997) demonstrated that the electron transport pathway, shared by respiratory and photosynthetic systems, controls both aerotaxis and phototaxis. It is difficult to determine whether a change in electron transport or a change in proton motive force is the signal controlling these behavioural responses. Electron transport is tightly coupled to proton translocation across the membrane. If electron transport increases, the proton motive force also increases until it reaches a maximum. However, this is true only for the branches of the electron transport system that are coupled with proton pumping. When electrons are diverted to non-coupled branches, proton motive force may decrease upon an increase in electron transport. Therefore, the relationship between electron transport and proton motive force is dependent on a particular electron transport system and may be unique for a given bacterial species. Recent findings indicate that a step down in the rate of electron transport (redox), rather than a change in the proton motive force, is the primary photosensory signal in R. sphaeroides (Grishanin et al., 1997). Findings in E. coli are consistent with proton motive force as the primary signal for aerotaxis (Shioi and Taylor, 1984).

The archaeon Halobacterium salinarum has sensory rhodopsins as specialized receptors for phototaxis. Excitation of the photoreceptors by blue light generates repellent signals and orange light generates an attractant signal. As a result, the cells move away from harmful ultraviolet light at the surface of the brine pools in which they live and congregate at depths where orange light provides optimal stimulation of energy producing bacteriorhodopsin. The mechanism of sensory transduction for this unique metabolism-independent photobehaviour was reviewed recently (Spudich, 1997) and will not be discussed further. A H. salinarum strain lacking sensory rhodopsins showed phototaxis that was mediated by proton-pumping bacteriorhodopsin (Yan et al., 1992; Bibikov et al., 1993). The signal for taxis was a light-dependent change in the electrical component of the proton motive force (Grishanin et al., 1996). Evidence is consistent with membrane potential as a common signal for both aerotaxis and the secondary phototaxis in H. salinarum (Bibikov and Skulachev, 1989; Lindbeck et al., 1995), guiding the bacteria to an environment that optimizes energy production.

Behavioural responses to perturbation of electron transport

Evidence that a change in electron transport/proton motive force is the signal for aerotaxis and metabolism-dependent phototaxis suggested that other environmental signals that alter electron transport would also elicit behavioural responses. Electron transport is limited by the availability of an electron acceptor, the supply of electron-donating carbon sources or by diversion of the electrons from the respiratory pathway. Behavioural responses to environmental stimuli that act at each of these regulatory sites have been identified.

Electron acceptor taxis

For maximal energy production, oxygen is the preferable electron acceptor, however, a majority of bacteria can use alternative electron acceptors. For example, E. coli uses nitrate, trimethylamine oxide and fumarate for respiration under anaerobic conditions. Alternative electron acceptors elicit an aerotaxis-like behaviour that requires a functional electron transport system, an appropriate terminal reductase and reduction of the acceptor (Taylor et al., 1979; Shioi et al., 1988). As a result, under anaerobic conditions, E. coli and the purple non-sulphur bacterium Rhodobacter sphaeroides (Gauden and Armitage, 1995) move to the environments that have an optimum concentration of alternative electron acceptors for growth. Light, oxygen and alternative electron acceptors compete for the behavioural response, as would be expected if these responses use electron transport as a common signal transduction mechanism (Shioi et al., 1988; Gauden and Armitage, 1995). Such responses are likely to be widespread in bacteria.

Redox taxis

In a redox gradient, A. brasilense, a microaerophilic nitrogen-fixing bacterium, and E. coli both swim to sites that have a preferred redox potential (Grishanin et al., 1991; Bespalov et al., 1996). These were the first demonstrations of redox taxis. In E. coli, artificial quinone analogues and other redox carriers elicit behavioural responses that depend on the redox potential (electron affinity) of the molecules and the permeability to the cytoplasmic membrane. Both chemicals donating electrons to and accepting from intermediate components of the electron transport system cause taxis, but only if they are reduced or oxidized by a component of the electron transport system (Bespalov et al., 1996). Therefore, redox taxis is metabolism dependent. All changes in electron transport upon stimulation with redox molecules are coincident with a correspondent change in the proton motive force (Bespalov et al., 1996).

Taxis to metabolized substrates

In R. sphaeroides, only chemicals that are transported into the cells and metabolized are attractants (Jeziore-Sassoon et al., 1998). Even for fructose, which is transported via a phosphotransferase system, it is metabolism, not the mode of transport, that elicits the behavioural response. Electron transport (redox) sensing not only mediates behavioural responses to light and oxygen in R. sphaeroides but is also involved in sensing chemical gradients (Armitage and Schmitt, 1997). Metabolism of a carbon source provides reducing equivalents for the electron transport system, and the cells may respond to a step down in electron transport (redox). Sinorhizobium meliloti appears to have a similar mechanism for sensing chemical attractants (Armitage and Schmitt, 1997), and the gliding bacterium Myxococcus xanthus may use soluble cytoplasmic receptors to sense metabolic changes (Zusman and McBride, 1991). However, chemical sensing in E. coli is almost exclusively metabolism independent, using chemoreceptors (Fig. 1) or the phosphotransferase pathway. It is only recently that taxis to glycerol was demonstrated to be the first example of metabolism-dependent energy taxis to a carbon source in E. coli (Zhulin et al., 1997a).

Sensing and signalling in aerotaxis

The mechanism for signal transduction in aerotaxis has been extensively investigated in E. coli. Aerotaxis is observed in the absence of chemotaxis receptors but does require the cheA, cheW and cheY genes (Rowsell et al., 1995). The Aer protein, recently identified in E. coli as the transducer for aerotaxis and other energy taxis responses (Bibikov et al., 1997; Rebbapragada et al., 1997), has a carboxyl domain that is homologous to those of the chemotaxis receptors (Fig. 2). Overexpression of Aer in a strain lacking all chemotaxis receptors imparts some clockwise rotation to the flagella (Bibikov et al., 1997). Thus, the signalling domain of Aer is likely to interact with the CheA and CheW proteins regulating the phosphorylation cascade that transmits the chemotaxis signal from the receptor to the flagellar motors. The sensing domain of Aer (Fig. 2) is homologous to the redox sensing domain of the NifL protein of Azotobacter vinelandii and some other oxygen (redox) sensors (Bibikov et al., 1997; Rebbapragada et al., 1997). The conserved residues constitute a PAS domain, which is also present in a large family of sensor proteins from all kingdoms, including Archaea and Eukarya (Zhulin et al., 1997b). Initially associated with protein–protein interactions, light reception, light regulation and circadian clock proteins, PAS domains are also proposed to be involved in the sensing of oxygen and redox potential in organisms from bacteria to humans (Zhulin et al., 1997b).

Figure 2.

. Comparison of the domain structure of Aer and other bacterial sensors. The chemoreceptors Tar, Tsr and Aer contain K1 and R1 methylation regions and a highly conserved signalling domain (HCD). A PAS domain consisting of S1 and S2 boxes (Zhulin et al., 1997b) is shared by Aer (P50466), the E. coli aerobic respiration control sensor kinase, ArcB (P22763), the Bradyrhizobium japonicum oxygen sensor kinase, FixL (P23222), the Azotobacter vinelandii redox sensor, NifL (P30663), the B. subtilis sporulation kinase, KinC (P39764). The numbers in parentheses indicate sequence accession numbers in the SwissProt database. SD, the periplasmic sensing domain in bacterial chemoreceptors. Transmitter (TD) domain (ID 94, 96, 129 in the ProDom protein domain database) is shown for sensory kinases. Black rectangles denote known or putative transmembrane regions.

Only one hydrophobic sequence is predicted in Aer. This sequence is flanked by three positively charged residues at the amino end and an amphipathic sequence at the carboxyl end that typically anchor hydrophobic sequences to the cytoplasmic surface of the membrane. It is possible that the hydrophobic sequence in Aer anchors two cytoplasmic domains to the membrane, constituting an intracellular sensor (Fig. 3).

Figure 3.

. A proposed mechanism of redox (energy) sensing in E. coli. See text for details.

The aerotactic responses of null aer strains are approximately half those in wild-type cells. If the serine chemoreceptor Tsr is deleted in addition to Aer, the cells show no responses to oxygen. The aerotaxis responses are restored in the aer tsr double mutant by expression of Aer or Tsr from a plasmid (Rebbapragada et al., 1997). Thus, Aer and Tsr are independent transducers for aerotaxis. Aerotaxis in tsr strains is near normal, explaining why a role of Tsr in aerotaxis was not detected previously.

Aer has non-covalently bound FAD as a cofactor (Bibikov et al., 1997) and probably senses redox changes in the electron transport system (Bespalov et al., 1996; Rebbapragada et al., 1997). Tsr has no known prosthetic group and may not respond to redox changes. However, a functional electron transport system is required for Tsr-transduced aerotaxis. The protein senses both periplasmic pH and cytoplasmic pH (Slonczewski et al., 1982; Krikos et al., 1985), and it is possible that Tsr responds to changes in the proton motive force in transducing the aerotaxis signal. This would be consistent with previously demonstrated aerotaxis responses that are dependent on the proton motive force (Shioi and Taylor, 1984).

A large family of putative transducers for behavioural responses has been described recently in H. salinarium (Rudolph et al., 1996; Zhang et al., 1996), including a novel aerotaxis transducer HtrVIII (Brooun et al., 1998). The N-terminal domain of the 642 residue HtrVIII contains six transmembrane segments that exhibit homology to the haem-binding domains of the mitochondrial cytochrome c oxidase. The carboxyl domain is similar to E. coli chemotaxis receptors. A strain deleted for the htrVIII gene lacks aerotaxis, whereas a strain overproducing the protein exhibits stronger aerotaxis. HtrVIII was shown to be responsible for demethylation (Brooun et al., 1998) that was previously observed during adaptation in aerotaxis of H. salinarum (Lindbeck et al., 1995).

Candidates for the energy-sensing receptors in other bacteria and archaea include PAS domain-containing chemotaxis transducers identified in a genome of Archaeoglobus fulgidis (I. B. Zhulin, unpublished), the TlpA protein in R. sphaeroides and S. meliloti (Armitage and Schmitt, 1997) and the DcrA protein in Desulfovibrio vulgaris, which has a putative haem-containing sensing domain and a signalling domain that is homologous to those of E. coli chemotaxis receptors (Fu et al., 1994). However, the dcrA null mutant has a normal aerotactic response, suggesting that either DcrA is not an aerotaxis receptor or that there is a second receptor that supports aerotaxis in dcrA mutants (B. L. Taylor, I. B. Zhulin, M. S. Johnson, R. Fu and G. Voordouw, unpublished).

The concept of energy taxis: something old and something new

Energy taxis is an old concept, elaborated by Clayton, for which various transduction mechanisms have been suggested (see Taylor, 1983b and references therein). Menten in 1948 proposed that the photophobic behaviour of Rhodospirillum rubrum resulted from an abrupt decrease in photosynthesis. Clayton in 1953 demonstrated an inter-relationship between aerotaxis and phototaxis and recognized that R. rubrum responded to a decrease in the rate of metabolism. Links refined the concept proposing that ATP was the critical metabolite that was sensed. Later, aerotaxis was demonstrated in an E. coli unc mutant in which oxygen increased the proton motive force but not ATP. That excluded ATP as the signal for aerotaxis in E. coli and the proton motive force was proposed as the parameter sensed in energy taxis (Taylor, 1983a,b). As discussed above, Armitage (Grishanin et al., 1997) recently showed that in R. sphaeroides taxis can be mediated by the electron transport system, independent of the proton motive force. But the possibility remains that either electron transport or proton motive force can mediate energy taxis.

As predicted, the E. coli aerotaxis transducers Aer and Tsr sense other stimuli that modulate electron transport and proton motive force (Rebbapragada et al., 1997), such as redox chemicals and rapidly metabolized carbon sources. Any environmental conditions that result in an appropriate step-up or step-down in energy levels will elicit a behavioural response. It follows that additional energy taxes transduced by Aer and Tsr remain to be identified.

A possible signal transduction model for aerotaxis and related energy taxes is shown in Fig. 3. Aer, a homodimer (M. S. Johnson and B. L. Taylor, unpublished), has two cytoplasmic domains. A FAD molecule associated with the PAS domain in the amino terminus is proposed to be in contact with a complex in the electron transport system. Changes in the flux of electrons through the electron transport system are reflected in changes in the redox state of the FAD in Aer. A resulting conformational change transmits a signal to the carboxyl domain, which controls autophosphorylation of CheA. It is possible that the PAS domain alters protein–protein interactions between the amino and carboxyl domains of Aer, as this is a role of PAS domains in other signalling proteins. Tsr has no redox prosthetic group, and the model shows Tsr directly sensing the proton motive force, but it remains to determine whether Tsr senses electron transport or the proton motive force. The carboxyl domain of Tsr regulates CheA phosphorylation in aerotaxis, as it does in chemotaxis. The convergence of the Tsr and Aer pathways for energy taxis integrates the signals in controlling motor reversal and swimming direction.

Selective advantage for cells that have both chemotaxis and energy taxis

Predictions that energy taxis is a residual primitive behaviour that preceded chemotaxis were not validated by comparison of the amino acid sequence of the E. coli Aer energy transducer and known chemotaxis receptors (H. Le Moual, personal communication; I. B. Zhulin and B. L. Taylor, unpublished). Whatever the origin of energy taxis, this complex behaviour would not have survived in E. coli unless it conferred a selective advantage on the organism. We present the following scenario to illustrate why it is an advantage for E. coli to have both energy taxis and chemotaxis.

In a freshwater pond, the degradation of a particle of dead cells releases amino acids that are strong chemoattractants for bacteria. Guided by chemotaxis, E. coli bacteria swim up the gradient of amino acids and cluster around the particle, where they rapidly oxidize the amino acids. The density of bacteria in the cluster may grow to reach 109 cells per ml and be visible to the naked eye (Taylor, 1983a). At this density, the bacteria rapidly deplete the oxygen surrounding the particle. Unless the bacteria are within 1 mm of the surface, diffusion of oxygen across the air–water interface will be inadequate to maintain aerobic metabolism. Without an aerotactic response, the bacteria clustered around the particle would become trapped in an anaerobic, growth-limiting environment. Aerotaxis stimulates the cells to swim away from the hypoxic environment, even though they are moving away from the highest concentrations of chemoattractants.

If the bacteria were solely dependent on energy taxis, they could be attracted preferentially to sugars and organic acids and become deprived of an adequate source of nitrogen. Through chemotaxis, E. coli preferentially seek amino acids and could be attracted to constituents that aid cell synthesis, but are not metabolized to produce energy. Under conditions that support vigorous growth, chemotaxis appears to be the primary determinant of behaviour in E. coli. In other bacterial species, in which energy taxis is a dominant behaviour, a different scenario has been proposed for the survival value of energy taxis (Armitage and Schmitt, 1997).

Energy taxis may be the bacterial equivalent of the adrenaline-mediated flight response in humans. When bacteria sense a decrease in intracellular energy, energy taxis overrides chemotaxis, giving the bacteria a chance to escape to an environment that supports optimal energy levels. This is a versatile flight response; any conditions that impair cellular energy can elicit a response. The importance of energy sensing is supported by the biological occurrence of putative redox-sensing PAS domains throughout all kingdoms.


We are grateful to M. Alam, J. P. Armitage, H. Le Moual and R. Schmitt for providing preprints and unpublished results and to M. S. Johnson for helpful discussions.The work from the authors' laboratories was supported in part by grants from the National Institute of General Medical Sciences (GM29481) and Loma Linda University (to B.L.T.), and from the United States Department of Agriculture/NRICGP (96–35305–3795) (to I.B.Z.).