In Halobacterium salinarum mutants containing either of the light-driven ion pumps bacteriorhodopsin (H+) or halorhodopsin (Cl–) as their only retinal protein, a decrease of irradiance in the absence of respiration causes a phototactic response. The conversion of the causal event, a decrease of proton motive force across the cell membrane, into a reversal of flagellar motor rotational direction was expected to involve a transducer. Via deletion analysis of all 18 known and putative halobacterial transducer (htr) genes, we found that Htr14, a methylatable membrane-bound transducer lacking an extracellular domain, mediates the biological response, which includes adaptive methylation. Based on a minimal stimulus length of 200 ms and the determined cytoplasmic buffering capacity, we conclude that the change in the membrane potential (ΔΨ), and not that of the internal pH, is the signal-generating event. Htr14 was therefore renamed to Membrane potential change Transducer, or MpcT. It is the first transducer for which the causative stimulus could be narrowed to a change in ΔΨ, as opposed to a change in pH or cellular redox state.
The extremely halophilic archaeon Halobacterium salinarum thrives in hypersaline environments like solar salterns or salt lakes (Oren, 1994). These environments are characterized by intense illumination and a shortage of oxygen. In addition to aerobic respiration and arginine fermentation via the arginine deiminase pathway (Ruepp and Soppa, 1996), H. salinarum can also use light as a source of energy via the actions of two retinal-containing light-driven ion pumps. Bacteriorhodopsin (BR) exports H+ ions, thereby providing the cell with the capacity for phototrophic growth, and halorhodopsin (HR) imports Cl– ions, which helps to maintain the osmotic balance during growth (reviewed by Oesterhelt, 1998).
Halobacterial migration towards favourable environmental conditions occurs via changes in the switching probability of flagellar motor rotation, which is controlled by a signalling machinery resembling that of eubacteria like Escherichia coli (reviewed by Stock and Surette, 1996) and especially Bacillus subtilis (Bischoff and Ordal, 1992). Eubacterial intracellular signal integration and adaptation are mediated by operon-encoded Che proteins. The che operon of H. salinarum (Rudolph and Oesterhelt, 1996; Ng et al., 2000; D. Oesterhelt et al. unpublished; http://www.halolex.mpg.de) displays similarities to the arrangement of che genes in B. subtilis. In this bacterium (Rao et al., 2004), signals are relayed to the flagellar motor via phosphorylation of the response regulator CheY by the histidine kinase CheA. Upon stimulation, CheA is activated by a methyl-accepting chemotaxis protein (MCP), to which it is coupled via adaptor proteins CheW or CheV. The methyltransferase CheR and methylesterase CheB are involved in adaptive methylation/demethylation of certain glutamate residues of the MCPs, a process in which CheC and CheD play regulatory roles.
Halobacterial CheB contains a putative methylesterase domain (Rudolph et al., 1995). This finding is consistent with the observed release of volatile methyl groups in response to attractants and repellents, a release pattern also found in B. subtilis (Alam et al., 1989). Furthermore, deletion of cheY, cheA or cheB genes leads to a complete loss of taxis in H. salinarum (Rudolph and Oesterhelt, 1996).
Molecular components of the halobacterial signalling machinery, including receptors, binding proteins and halobacterial transducers (Htrs), are involved in sensing a variety of stimuli. A total of 18 Htr-encoding genes (htrs) have been identified in the halobacterial genome on the basis of high homology to the well conserved signalling domain of eubacterial MCPs (http://www.halolex.mpg.de;Zhang et al., 1996; Ng et al., 2000).
Taxis towards orange and away from UV light, and taxis away from blue light, are mediated by the photoreceptors sensory rhodopsin I and II (SRI and SRII), respectively, together with their associated transducers HtrI and HtrII (Htr1 and Htr2) (Yao and Spudich, 1992; Seidel et al., 1995; Zhang et al., 1996). For HtrII, a function in serine chemotaxis has also been reported (Hou et al, 1998). Photostimulation of SRI and SRII additionally causes the release of the switch factor fumarate, which restores motor switching in the straight-swimming mutant strain M415 (Marwan and Oesterhelt, 1991). Aerotaxis is mediated by membrane-bound HtrVIII (Htr8) (Brooun et al., 1998) and the soluble transducer HemAT (Htr10) (Hou et al., 2000).
The transducers BasT (Htr3) (Kokoeva and Oesterhelt, 2000) and CosT (Htr5) (Kokoeva et al., 2002) mediate taxis towards Leu, Ile, Val, Met and Cys, and to compatible solutes of the betaine family respectively. These processes involve the putative binding proteins BasB and CosB respectively (Kokoeva et al., 2002). Chemotaxis towards Arg is mediated by the cytoplasmic transducer Car (Htr11) (Storch et al., 1999). Responses to Asp, Glu and His were reported by Brooun et al. (1997) for a transducer from halobacterial strain Flx15 called HtrXI, which differs from Car (strain S9) at only nine residues.
The halobacterial photoresponse to changes in light intensity at high irradiance was proposed to be mediated by BR, based on the observation that cyanide, an inhibitor of the respiratory chain, or dicyclohexylcarbodiimide (DCCD), an inhibitor of the H+-ATPase, both enhanced the sensitivity of cells to a sudden decrease of green light intensity (Baryshev et al., 1981). BR, as well as the respiratory chain and the H+-ATPase, are generators of proton motive force (pmf) across the cell membrane. Inhibition of respiration or H+-ATPase activity leads to a decreased pmf in the dark, and therefore to a greater drop of pmf upon a decrease of irradiance. Oxygen-depleted Pho81-B4 cells, which contain BR as their only functional retinal protein, are sensitive to changes at high irradiance levels, whereas Flx15 cells containing only SRI and II are sensitive to changes at low irradiance levels. These observations demonstrate that the combination of the two systems enables wild-type cells to react not only to different wavelengths of light but also within a wide range of light intensities (Bibikov et al., 1991). No photoresponses are seen in cells lacking all retinal proteins.
The decrease of BR-mediated photosensitivity upon addition of the fermentative substrate arginine (also able to generate pmf), the effects of cyanide and DCCD, and comparable signal processing times and adaptation kinetics for SRI- and BR-mediated photosensing argue for an interaction of a putative ‘pmf-sensor’ with components of the halobacterial signal transduction chain (Bibikov et al., 1993). Grishanin et al. (1996) proposed that a change of the membrane potential (ΔΨ), rather than of the pH difference across the membrane (ΔpH), is essential for the communication of changes in illumination from BR to the taxis system (ΔΨ and ΔpH comprise the two constituents of the pmf). The underlying observation was that the addition of the hydrophobic cation tetraphenylphosphonium (TPP+), known for its ability to quench ΔΨ, decreased the sensitivity of cells to BR-mediated photosensing, whereas the addition of ammonium acetate, expected to decrease ΔpH, was without effect.
The indications that the halobacterial signal transduction machinery is involved in BR-dependent phototaxis prompted us to test whether any of the 18 Htrs mediates this process. We have shown that Htr14 is the transducer that is responsible for the response and have collected additional evidence that it is ΔΨ that is monitored. Htr14 is the first transducer for which changes in ΔΨ could be identified as the signal-generating stimulus. Htr14 was therefore renamed as the Membrane potential change Transducer, or MpcT.
Results and discussion
Construction of retinal-protein knockout strains
We chose strain OMI1 (M. Otsuka, unpublished), which lacks all four retinal proteins and is well characterized genetically, as the basis for all subsequent genetic manipulations (Table 1 and Supplementary material, Fig.S1). Complementation of OMI1 with the bop and hop loci to produce strains MKK101 and MKK102, respectively, restored the wild-type alleles at their original positions. MKK101 and MKK102 were generated using pUS-Mev-derived plasmids pMKK101 and pMKK100-derived pMKK102 respectively (Fig. 1). Southern blot or polymerase chain reaction (PCR) analysis of all strains confirmed the expected genotypes (Supplementary material, Fig.S1BE). OMI1, and all strains derived from it, also lack the genes for the transducers Htr1, Htr2 and Car (Htr11). The loss of the car gene probably happened during strain construction via a partial loss of DNA from the halobacterial plasmid encoding car.
. The exact DNA regions missing in the htr-deletion plasmids are listed in Fig. 2 and plasmid construction is described in Fig. 1. Information about all mentioned genes, their sequences and additional names is available at the database http://www.halolex.mpg.de (D. Oesterhelt et al. unpublished).
. OMI1 (M. Otsuka, unpubl.) was generated from strain S9 (Wagner et al., 1981) as described in Supplementary material, Fig. S1.
. In addition to the systematic designations (htr1–18), final names were assigned to these htrs and their gene products after the elucidation of a biological function, as described in the Introduction chapter.
. Nucleotide positions are given relative to the first base of the ORF (+1).
Like OMI1, except bop+– generated from OMI1 by restoration of the wild-type bop locus using plasmid pMKK101
Like OMI1, except hop+– generated from OMI1 by restoration of the wild-type hop locus using plasmid pMKK102
Like OMI1, except bop+ and Δhtr3c– generated from MKK101 by deletion of htr3 using plasmid pMKK103
Like OMI1, except bop+ and Δhtr4– generated from MKK101 by deletion of htr4 using plasmid pMKK104
Like OMI1, except bop+ and Δhtr5c– generated from MKK101 by deletion of htr5 using plasmid pMKK105
Like OMI1, except bop+ and Δhtr6 – generated from MKK101 by deletion of htr6 using plasmid pMKK106
Like OMI1, except bop+ and Δhtr7– generated from MKK101 by deletion of htr7 using plasmid pMKK107
Like OMI1, except bop+ and Δhtr8c– generated from MKK101 by deletion of htr8 using plasmid pMKK108
Like OMI1, except bop+ and Δhtr9– generated from MKK101 by deletion of htr9 using plasmid pMKK109
Like OMI1, except bop+ and Δhtr10c– generated from MKK101 by deletion of htr10 using plasmid pMKK110
Like OMI1, except bop+ and Δhtr12– generated from MKK101 by deletion of htr12 using plasmid pMKK112
Like OMI1, except bop+ and Δhtr13– generated from MKK101 by deletion of htr13 using plasmid pMKK113
Like OMI1, except bop+ and Δhtr14– generated from MKK101 by deletion of htr14 using plasmid pMKK114
Like OMI1, except bop+ and Δhtr15– generated from MKK101 by deletion of htr15 using plasmid pMKK115
Like OMI1, except bop+ and Δhtr16– generated from MKK101 by deletion of htr16 using plasmid pMKK116
Like OMI1, except bop+ and Δhtr17– generated from MKK101 by deletion of htr17 using plasmid pMKK117
Like OMI1, except bop+ and Δhtr18– generated from MKK101 by deletion of htr18 using plasmid pMKK118
Like OMI1, except bop+– generated from MKK114 by restoration of the wild-type htr14 locus using plasmid pMKK119
Like OMI1, except hop+ and Δhtr14– generated from MKK102 by deletion of htr14 using plasmid pMKK114
ΔcheB– generated from S9 by deletion of cheB (the deleted region comprises the DNA from position +19 to +1026 of the ORF starting with: ATGACA . . .)d
ΔcheR– generated from S9 by deletion of cheR (the deleted region comprises the DNA from position +91 to +807 of the ORF starting with: TTGACT . . .)d
Generation of htr-deletion strains
To determine if one of the Htrs is mediating BR-dependent phototaxis, htr-deletion strains were generated from parental strain MKK101 and subsequently subjected to phenotypic analysis. In each of these strains a different one of the remaining 15 Htrs was deleted. The deletion strategy is exemplified by the construction of deletion-fragment Δ14, which was used, contained in plasmid pMKK114 (Fig. 1), for the generation of htr14-deletion strain MKK114 (Fig. 2). Strain MKK120 was generated from strain MKK102 using the same plasmid. The deletion removes the entire open reading frame (ORF) for the respective Htr, either exactly or within a margin of only a few base pairs. Screening for the deletions was greatly facilitated by the use of red–blue clone selection (described in detail in Supplementary material, AppendixS1). In each case Southern blot analysis confirmed the expected genotype (shown for MKK114 in Fig. 2).
BR-dependent phototaxis is mediated by Htr14 and involves stimulus-induced methanol release
As previously shown by tracking of Pho81-B4 cells (Bibikov et al., 1993), the BR-dependent photoresponse is characterized by a significant increase in the number of reversing cells when orange light is switched off. The percentage of cells, which reverse their swimming direction within 4 s after a light-off stimulus, increases with increasing initial light intensity. With less than approximately 20% reversals in the absence of a stimulus, a maximum of approximately 85–95% reversals is reached with Pho81-B4 cells when the prestimulus irradiance exceeds 150–200 W m−2, values which we confirmed with MKK101 cells.
To identify the Htr involved in this response, computerized cell tracking experiments were performed (Fig. 3). We observed the characteristic photoresponse with MKK101 and all of the htr-deletion strains derived from it except MKK114 (Δhtr14). The photoresponse was restored in the unresponsive strain by complementation with htr14, i.e. in strain MKK119. The response was investigated under conditions of oxygen depletion, which leads to a low pmf in the dark. As the pmf is the driving force for flagellar rotation, the drop in pmf could be observed as a progressive immobilization of the cells during preincubation in the dark. Motility returned upon re-illumination because of the pmf-regenerating proton-transport activity of BR.
Addition of the respiratory-chain inhibitor cyanide to the cell suspension allowed for investigation of the BR-dependent photoresponse in the presence of oxygen. Cyanide-treated MKK101 cells that were not oxygen-depleted performed more photostimulated reversals as the concentration of cyanide increased. The cyanide effect was seen at concentrations as low as 0.1 mM, increased in a concentration-dependent manner up to approximately 2 mM, and then remained constant at least up to 10 mM. As expected, in the presence of 10 mM cyanide no photoresponse was seen with MKK114 but the response was restored in MKK119.
As an alternative approach to investigate BR-dependent photosensing by MKK101 cells, we employed an assay that measures the adaptive release of volatile radioactive methyl groups (as methanol) from transducers (Alam et al., 1989). After radioactive labelling and immobilization of the cells on a syringe-filter, a constant flow of chase buffer was applied, and the cells were subjected to orange light stimuli.
To demonstrate functional adaptation, l-leucine stimuli were used as a control. Without cyanide in the chase buffer, none of the tested strains showed a release of methanol upon light stimulation (data not shown), whereas l-leucine evoked methanol release in all experiments. When 10 mM cyanide was present, we could produce strong demethylation signals in MKK119 cells with orange light-on or -off stimuli (Fig. 4A). These signals were absent with MKK114, confirming the findings of the cell tracking experiments. For OMI1 cells, which lack all retinal proteins, no demethylation response could be observed upon photostimulation, although it was seen in response to l-leucine (data not shown).
While the cell tracking experiments demonstrate that Htr14 is the only transducer whose absence abolishes BR-dependent phototaxis, and therefore indicate that Htr14 mediates this process, the methanol-release experiments additionally showed that adaptive methylation is involved in Htr14-dependent sensing.
Htr14 is also responsible for HR-dependent photoresponses
Upon illumination, HR pumps chloride ions into anaerobic dark-adapted cells, thereby increasing the ΔΨ-component of the pmf as demonstrated by a transient (several minutes) generation of maximal (aerobic) cellular ATP level with concomitant acidification of the cytosol (Mukohata and Kaji, 1981). When tracking motile MKK102 cells in the presence of 10 mM cyanide, we were unable to detect swimming reversals upon light-off stimuli (data not shown), a result which contrasts with the observation of a reversal response in strain Pho81-HR, reported by Grishanin et al. (1996). Both strains contain HR as the only retinal protein, although in the case of Pho81-HR it was overproduced.
Figure 4B shows results of methanol-release experiments using MKK102 and an htr14-deletion strain derived from it (MKK120). Upon a light-off stimulus, methanol was released only from MKK102 cells, i.e. only when Htr14 was present. This result demonstrates that Htr14 also mediates HR-dependent photoresponses. The reason why MKK102 cells did not exhibit swimming reversals upon a decrease in irradiance is unclear.
Htr14 is bound to the plasma membrane and can exist in differently methylated forms
A visual inspection of the amino acid sequence of Htr14 identified a hydrophobic region, including residues 28–76, that lacks charged amino acids. This stretch could accomodate two transmembrane helices connected by a hairpin loop, as depicted in Fig. 5. To determine the cellular localization of Htr14, an antiserum (anti-Htr14D585) was generated against a synthetic peptide which contains a stretch of amino acids present in the C-terminal region of the protein (D585 to S606). The antiserum is highly specific as demonstrated by the complete absence of reactivity with proteins from the htr14-deletion strain MKK114 (data not shown). Immunoblots of the cytosolic and membrane protein fractions of strains MKK101, S9 and the cheB- and cheR-deletion strains WFS101 and WFS102, respectively, show that the majority of cellular Htr14 pelleted together with the membranes (Fig. 6). The small amount of apparently cytosolic Htr14 might result from incomplete pelleting of small membrane fragments. These results suggest that Htr14 is anchored to the membrane, probably via the two predicted N-terminal transmembrane helices.
To examine the extent of Htr14 methylation, the mobility of Htr14 in SDS-PAGE gels was investigated in strains MKK101, S9, WFS101 and WFS102 (Fig. 6). In strain WFS101 (ΔcheB) Htr14 should be fully methylated, because methyl groups can be added by the methyltransferase CheR but not removed by the methylesterase CheB, whereas in strain WFS102 (ΔcheR) Htr14 should be completely unmethylated.
The major Htr14 species in the cytosolic as well as in the membrane protein fractions of the investigated strains appear in three bands at apparent molecular weights (MW) of 122, 117 and 115 kDa. Of these three bands only the one at 122 kDa is present in strain WFS102. This band is therefore assigned to an unmethylated Htr14 species, consistent with the previous observation that decreasing degrees of methylation of E. coli transducers cause them to run with progressively higher apparent MW (Boyd and Simon, 1980). For strain WFS101 the main band runs at 115 kDa and is attributed to a highly methylated Htr14 species. It is not clear whether the faint band observed in the membrane protein fraction at an MW below 115 kDa represents another, differently methylated species of Htr14 or whether it results from an Htr14 species with identical methylation that runs differently because of methylation-independent conformational differences. For strains S9 and MKK101 a strong band is seen at an MW of 117 kDa and is attributed to a moderately methylated Htr14 species. Above this band another, weaker band can be seen at 122 kDa and points to the additional presence of an unmethylated Htr14 species in these strains.
A discrepancy between apparent (122 kDa for unmethylated Htr14) and calculated (65.6 kDa) MW has been noted previously for halobacterial proteins and is thought to be due to their acidic nature (Monstadt and Holldorf, 1991). The different electrophoretic mobilities of the described Htr14 species could be abolished by addition of 6 M urea to the sample buffer and gel, suggesting that conformational differences resulting from different transducer methylation disappear under these conditions (data not shown). The described pattern for the major Htr14 species in the investigated strains is repeated at higher molecular weights. This points to conformational differences between Htr14 species, which are visible under the used electrophoresis conditions but are not due to different methylation levels.
Recently obtained mass spectrometrical data demonstrated that Htr14 contains (at least) two methylation sites, one within the glutamate pair E310/E311 and one within E506/E507 (M.K. Koch et al., unpublished). The first pair is located within a sequence that is well conserved among the 18 Htrs (Fig. 5) and perfectly matches the consensus sequence for transducer methylation by CheR in E. coli (Nowlin et al., 1987): A/S-X-X-E-E*-X-A/S/T-A-S/T/A (with X symbolizing an arbitrary amino acid and the asterisk marking the methylation site). Perazzona and Spudich (1999) showed that mutations within this conserved halobacterial sequence, which convert E265/E266 of Htr1 and E513/E514 of Htr2 respectively, to alanine pairs, completely abolish methylation of these transducers. The sequence surrounding E506/E507 of Htr14 also matches the E. coli methylation consensus, except that the last residue is a glutamate.
It cannot be excluded that additional methylation sites exist in Htr14. However, the existence of three differently methylated Htr14 species, which appear as three different bands on the blot, could be explained by attributing these bands with increasing mobility to an unmethylated, singly methylated and doubly methylated Htr14 species.
The buffering capacity of the H. salinarum cytoplasm argues for a change in ΔΨ rather than in pH as the stimulus that is sensed via Htr14
To assess which of the two constituents of the pmf, ΔpH or ΔΨ, is actually sensed during BR- and HR-mediated photosensing, we compared the minimum stimulus length, required to elicit a reversal response, with the time that is necessary to generate significant changes in ΔΨ and in internal pH (pHi). When orange light of 390 W m−2 was switched off for 800 ms, 500 ms and 200 ms, the percentages of MKK101 cells that reversed within 4 s were determined to be 85%, 65% and 40%, compared to 20% in the absence of a stimulus. These data demonstrate that 200 ms of darkness is sufficient to elicit a reversal response in MKK101 cells, which is in accordance with previous results obtained with Pho81-B4 cells (Bibikov et al., 1993).
To estimate how rapidly pHi changes after the light is shut off, it is necessary to know the cytoplasmic buffering capacity at a given pHi and to compare it with the net rate of proton transport across the plasma membrane after stimulus application. We therefore determined the buffering capacities of halobacterial cell fractions at different pH values via titration with NaOH (Fig. 7). The pH values of MKK101 and S9 lysates were 6.6 and 6.8 respectively. Between pH 6.5 and pH 7.0, the buffering capacities of S9 and MKK101 cytosolic fractions were between 400 nmol and 250 nmol H+ per pH unit per mg protein. The protein concentrations of the cytosolic fractions of MKK101 and S9 cells were 0.30 ± 0.02 and 0.32 ± 0.01 mg per millilitre of cell suspension at an optical density of 1.0 at 600 nm (ODml) respectively, and cell lysates both contained 0.43 ± 0.01 mg per ODml.
With 1.4 × 109 cells per ODml, the number of protons that must be transported out of a cell with pHi values between 6.5 and 7.0 to increase its cytosolic pH by 0.1 was calculated to be between 5.1 × 106 and 3.1 × 106 for a 1 femtolitre cell. This number is in good agreement with the value of 8 × 106 protons calculated earlier for a 1.4 femtolitre cell of strain R1M1 at pH 7.0 (Michel and Oesterhelt, 1980).
The maximum number of H+ ions transported per second across the plasma membrane of MKK101 cells was determined by measuring the pH change in the external medium when switching off orange light of 80 W m−2. With a buffering capacity of the external medium of 0.150 ± 0.008 pH units per 100 nmol H+, and with 7 × 109 cells in the assay, we found a maximum transport rate of 48 000 ± 5000 H+ per second and cell. When using light intensities of or below 80 W m−2, the initial rates seen when switching off the light were directly proportional to the original light intensity and remained constant for 10–12 s before they began to decrease. The maximum observed change of external pH (0.020 ± 0.002 pH units) was reached after approximately 60 s and corresponded to approximately 1.1 × 106 protons per cell and a change in pHi of approximately 0.03 pH units. The initial transport rates allowed us to estimate the maximum change in pHi that can be expected when the light is switched off for 200 ms. Assuming a fivefold higher transport rate in the cell tracking experiments, when the prestimulus irradiance was 390 instead of 80 W m−2, the maximum number of H+ ions transported across the membrane within the first 200 ms would be 48 000, corresponding to a change in pHi of only 48 000/3.1 × 107 = 0.0015 pH units. Even a 10-fold greater change in pHi would not be expected to alter the protonation state of amino acid residues significantly. These considerations lead to the conclusion that the event that is sensed via Htr14 is not a change in pHi (ΔpHi) but rather a change in ΔΨ (ΔΔΨ).
As previously determined for halobacterial strain R1M1 (Michel and Oesterhelt, 1976), the ΔΨ of oxygen-depleted dark-adapted cells at an external pH between 6.0 and 6.8 can be increased from approximately −90 mV to approximately−130 mV upon illumination with 73 W m−2 for 15 min. The difference of 40 mV should therefore be a good estimate for the maximum initial drop of ΔΨ when light is switched off under the conditions of our experiments. A ΔΔΨ of 1 mV requires the net transport of 330 charges across the cell membrane for an average H. salinarum cell (Michel and Oesterhelt, 1980). A transport rate of 48 000 charges per second per cell upon switching off 80 W m−2 light, which represents a minimum rate for net charge transport in our cell tracking experiments, would thus correspond to a drop of 29 mV within 200 ms. We conclude that a ΔΔΨ that can be sensed by the cell can be generated within less than 100 ms and must be maintained for a few hundred milliseconds to be sensed.
Changes in irradiance also influence, via pmf changes, the cellular ATP level and electron flow through the respiratory chain. In principle, these factors could be alternatives to ΔΨ as the parameter monitored by the cells. A decrease in ATP level might cause decreased phosphorylation of one or more proteins which in their dephosphorylated state might induce a reversal response via the transducer. The decrease of cellular ATP concentration within the 200 ms stimulus time, however, would only be about 10 µM, corresponding to < 0.3% of the ATP level under the given conditions (Wagner et al., 1978). The changes in ATP levels that occur during stimulus application should therefore not be responsible for significant changes in the phosphorylation levels of putative signalling proteins.
Changes in electron flow through the respiratory chain would shift the redox state of chain components. Any of these, through direct or indirect interaction with Htr14, could cause the repellent response. Such a mechanism has been proposed for the E. coli aerotaxis transducer Aer, which is thought to sense the redox state of the electron transport system via an FAD cofactor non-covalently bound to its PAS domain (reviewed by Taylor et al., 1999). For Tsr, which can also mediate aerotactic responses in E. coli but is lacking a prosthetic group, Rebbapragada et al. (1997) proposed a pmf-sensing function. However, electron transport sensing could not be excluded under the respiration conditions prevailing in those experiments.
In halobacteria, light inhibition of respiration in the presence of oxygen is the consequence of a pmf increase (Oesterhelt and Krippahl, 1973). A decrease of oxygen concentration in the medium also causes chain components to assume a reduced state, although in the latter case the pmf falls rather than increases. The BR-dependent photoresponse, however, is observed in the presence of cyanide or in the absence of oxygen. As alternative terminal electron acceptors were absent in our experiments, both conditions should prevent electron flow through the respiratory chain, thereby eliminating the possibility of changes in the redox state of chain components. After excluding the generation of sufficient changes in pHi and cellular ATP concentration within the period of stimulus application, the only viable explanation is that Htr14 responds to changes in ΔΨ. Therefore, we renamed Htr14 as the Membrane potential change Transducer, or MpcT.
MpcT (Htr14) by itself probably serves as the ΔΨ sensor
The question remains whether MpcT detects ΔΔΨ directly or in conjunction with another protein or proteins. A known example of ΔΔΨ sensing is the voltage-gated K+ channel KvAP from Aeropyrum pernix for which a positively charged but otherwise very hydrophobic ‘sensor paddle’ domain was suggested to move perpendicular to the plane of the membrane upon a ΔΔΨ, thereby inducing a conformational change in the gating part of the channel (Jiang et al., 2003).
One might speculate that ΔΔΨ sensing via MpcT occurs by a similar mechanism, either involving a membrane-embedded segment of MpcT containing charged residues or via an interaction of MpcT with another protein containing such a segment. For the predicted transmembrane (TM) region (maybe in combination with some charged, also membrane-embedded residues from the N-terminal region) both options are possible, whereas the Pro-rich C-terminal region of MpcT would seem to require interactions with another protein or proteins. These two regions in MpcT are quite dissimilar in their primary sequence to the corresponding regions of other Htrs. Together with the linker region (between TM and putative coiled-coil region) they are proposed to be situated close to the membrane, as depicted in a hypothetical working model of an MpcT structure on the basis of a proposed structure of E. coli Tsr (Supplementary material, Fig.S2).
When the amino acid sequence of MpcT was compared with that of its putative orthologue from the halophilic archaeon Natronomonas pharaonis (M. Falb et al., unpublished), no significant sequence identity could be found for the C-terminal region downstream of position 552 of MpcT (Fig. 5). However, in both proteins this region contains a stretch of 33 residues with a proline content of 21% in MpcT and 18% in its putative orthologue, which is unusually high for Htrs. This might hint to a conserved function of this region, despite the poor sequence identity. The N-terminal region (up to position 166 of MpcT) is reasonably well conserved, with 43% identity, including all the prolyl residues. The 34% identity found for the intermediate region (positions 167–552 of MpcT) is within the range seen when the corresponding intermediate regions of any two halobacterial transducers are compared. In addition to the putative coiled-coil of methylatable and signalling domain (residues 245–552 of MpcT), this intermediate region consists of an additional stretch of residues. This stretch comprises putatively α-helical regions (as deduced from the presence of hydrophobic heptad sequences), interrupted by a region of unknown structure, that contains a sequence we termed GDL-motif (Fig. 5). A GDL-motif indicates the beginning of the ‘connector region’ within a ‘HAMP domain’ of a transducer, as described in detail in Supplementary material, AppendixS2.
A comparison of MpcT with proteins from H. salinarum strain R1 (D. Oesterhelt et al. unpublished; http://www.halolex.mpg.de) discovered protein OE2712R (the equivalent of protein Vng1193c from Halobacterium sp. NRC-1; Ng et al., 2000). OE2712R is 39% identical (from positions 25–180) to the mentioned N-terminal region of MpcT. The C-terminal region of OE2712R (from positions 183–392), however, is 32% identical to the putative histidine kinase domain of protein OE2961F (Vng1374g). His203 of OE2712R corresponds to the conserved phosphorylatable His405 of B. subtilis histidine kinase A (Wang et al., 2001). Assuming that the same signal is sensed by MpcT and OE2712R, these findings suggest that the N-terminal rather than the C-terminal regions of MpcT and OE2712R are responsible for the input of the ΔΔΨ signal. In the case of MpcT, the output is likely to be via an interaction of its cytoplasmic signalling domain with the associated histidine kinase CheA, which then relays the signal via the response regulator CheY to the flagellar motor as the final target. In the case of the putative sensor kinase OE2712R, the output might be a change in an intrinsic histidine kinase activity and involve transcriptional regulation via an unknown response regulator. None of the other Htrs show a comparable similarity between a transducer region and a region from a putative sensor kinase.
Htr1, Htr2, CosT and BasT are the Htrs for which interactions with a photoreceptor or binding protein have been reported. All corresponding gene pairs are co-transcribed, and they either overlap or, in the case of htr2, are separated by just one base (Supplementary material, Fig.S1B). An inspection of the flanking sequences of mpcT revealed an AT-rich region centred approximately 30 base pairs upstream of the coding region that could represent a promoter sequence. This finding, together with the relatively large distance of 81 base pairs between mpcT and its neighbouring upstream gene OE1537B1R (D. Oesterhelt et al. unpublished; http://www.halolex.mpg.de), strongly argues that mpcT is not co-transcribed with an upstream gene. The nearest downstream gene, OE1534F (VNG0354C), is located 13 bp downstream and is in the opposite orientation. A deletion of this gene in strain MKK101 had no detectable effect on BR-dependent phototaxis (data not shown), which excludes it as being a partner with MpcT in ΔΨ sensing.
Figure 8 summarizes our current view of the halobacterial signalling machinery. Although a role for MpcT as the direct sensor of changes in ΔΨ is supported by the absence of a co-transcribed receptor gene, it cannot yet be ruled out that one or more additional proteins are required to interact with MpcT in the process of ΔΔΨ sensing. However, contrary to signalling via Htr1 and Htr2, which in both cases requires a physical interaction with the corresponding photoreceptors, such an interaction of MpcT with BR or HR is not assumed. A ΔΔΨ, generated via BR or HR, should be detectable anywhere along the cell membrane, and does not necessarily require a protein–protein interaction between the sensor and the generator of ΔΔΨ. As photostimuli can be sensed in both combinations, BR/MpcT or HR/MpcT, a physical interaction would require that MpcT interacts alternatively with two different receptors, i.e. with BR as well as with HR. Taking into account the differences especially in the structures of the membrane-exposed surfaces, the tilt of the helices and the photocycles of HR and BR, this seems rather unlikely.
The elucidation of the molecular mechanism of ΔΔΨ sensing via MpcT will be the subject of further investigations.
Culture conditions, strains and strategy of strain generation
All strains were grown in complex medium in the dark as described (Oesterhelt and Krippahl, 1983) to an optical density at 600 nm (OD600) of 0.7–0.9, unless otherwise stated. See Supplementary material, AppendixS1 for details about the strains and the procedures used during their construction, i.e. PCR strategy, plasmid construction, red–blue selection of clones, and verification of strain genotypes.
Computerized analysis of swimming behaviour
Cellular motion was recorded under infrared observation light by collection of raw video data at a frequency of 10 frames per second using a frame grabber (Motion Analysis Corporation, Santa Rosa, CA). Microscopic set-up and motion analysis algorithms were essentially as described (Marwan and Oesterhelt, 1990). In all experiments, cells were tracked for 5 s, either under constant illumination with orange light or starting 1 s before application of a 3 s light-off stimulus. Only cells with a minimum swimming speed of 8 µm s−1 were included in the calculation of the percentage of cells reversing within the last 4 s of tracking.
After growing the cells under standard conditions, they were diluted to an OD600 of 0.3. In the case of oxygen depletion, basal salt medium (complex medium without peptone) buffered to pH 7.0 with 20 mM HEPES (BSH) was added. For experiments to test the effects of cyanide, cells were concentrated by centrifugation and resuspended in complex medium (pH 7.0) containing differing concentrations of KCN. Eight microlitres of diluted cell suspension were applied to a microscopic slide and sealed under a coverslip with a paraffin–vaseline mixture (2:1) to avoid evaporation and to limit oxygen diffusion to the cells. To decrease the number of cells sticking to the glass surface, slides and coverslips were thoroughly cleaned with ethanol, incubated in freshly prepared aqueous 0.5% pectin (Sigma) solution for 2 min, and blown dry with pressurized air.
For experiments carried out under conditions of oxygen depletion, cells were kept on the slide in the dark for more than 1 h, and tracking was initiated after re-illumination for at least 10 min. Light of 640 ± 80 nm at an intensity of 300 W m−2 in the plane of the cells was provided by a 100 W mercury lamp equipped with cut-off filters (Schott) KG1, W360 and OG570. For investigation of the cyanide effect, cells were applied to the slides and immediately illuminated with 580 ± 50 nm light at an intensity of 390 W m−2. Measurements started after 10 min of preillumination.
Determination of protein content and buffering capacity of H. salinarum cellular fractions
A 350 ml culture of strain S9 or MKK101 was grown at 37°C with swirling at 180 r.p.m. to an OD600 of 1.0 and 0.7 respectively. The cells were pelleted by centrifugation, washed three times with 40 ml of unbuffered basal salt medium without citrate (BSWC), and finally resuspended in 50% BSWC to an OD600 of 10. A lysate corresponding to 200 ODml of cells was generated by sonication (2 × 1.5 min, Branson sonifier 450, 1/2 inch disruptor horn, 30% duty cycle, level 7) and centrifuged for 20 min at 6300 g to remove unlysed cells (<1%). An aliquot of 4 ml (the total lysate fraction) was kept for determinations of protein concentration and buffering capacity. The remaining lysate was subjected to ultracentrifugation (60Ti-rotor, Beckman, 200 000 g, 60 min, 4°C). The cytosolic fraction was removed and the membrane pellet was washed with 50% BSWC, centrifuged under the same conditions, and resuspended in 0.9 ml of water. Protein concentrations of all fractions were determined by the BCA assay (Pierce).
To determine the buffering capacities of the different fractions first of all 40 ml of 50% BSWC were brought to pH 4.5 by addition of HCl and then titrated under a stream of nitrogen up to pH 9.3 with 0.01 N NaOH and back to pH 4.5 with 0.01 N HCl. In separate experiments, these titrations were repeated after addition of 2.5 ml of total lysate, 5 ml of cytosolic fraction, or 600 µl of membrane fraction, respectively, this time using 0.1 N NaOH and HCl. The buffering capacities at different pH values (in nmol H+ per pH unit) were calculated for each fraction as the difference in the slopes of the NaOH titration curves in the presence and absence of the respective fraction. From the volumes of the fractions and their protein concentrations, buffering capacities could be calculated as nmol H+ per pH unit per mg protein.
Antibody generation, cell fractionation and immunodetection
The polyclonal antiserum anti-Htr14D585 was generated against a synthetic peptide, comprising amino acids 585–606 of MpcT (Htr14) plus an N-terminally attached glycyl-cysteine. The peptide was coupled via the cysteine to maleimide-activated keyhole limpet hemocyanin, according to the manufacturer's protocol (Pierce). After the conjugate was purified via gel filtration as recommended, it was emulsified 1:1 with 0.5 ml of TiterMax Gold adjuvant (Sigma). Five hundred microlitres of the mixture were subcutaneously injected into a rabbit followed by two booster shots with 250 µl each after 28 and 78 days. The serum was collected on day 92 and after clotting and centrifugation was stored at −70°C and used without further purification.
Membrane and cytosolic fractions of halobacterial protein were prepared under high salt conditions (1.3 M) as follows: 450 ODml of cells from a 350 ml culture with an OD600 of 1.4 were spun down, and resuspended in 20 ml of 30% basal salt medium (30% complex medium without peptone) additionally containing 10 µg ml−1 PMSF, 20 mM HEPES pH 7.5 and 100 µg ml−1 DNase I. After sonication at 0–10°C (2 × 1 min, Branson sonifier 450, 1/2 inch disruptor horn, 30% duty cycle, level 6) the lysate was centrifuged at 6300 g and 20°C for 20 min to remove unlysed cells and allow for DNase I action. The membrane protein fraction was obtained as a pellet after centrifugation (60Ti-rotor, Beckman, 200 000 g, 90 min, 4°C). After removing the supernatant the pellet was resuspended in 20 ml of the mentioned buffer and centrifuged as before. It was finally resuspended in 1 ml of protein resuspension buffer (20 mM HEPES pH 7.5, containing 10 µg ml−1 of PMSF and 1 mM EDTA). To avoid contamination of the cytosolic protein fraction with the pellet only the upper 7 ml of the supernatant of the first ultracentrifugation were carefully collected, the rest was discarded. After protein precipitation by adding six volumes of acetone (−20°C) to the saved supernatant, followed by two washing steps with 40 ml of ice-cold 50% acetone to remove the salt, the precipitate was pelleted, dried and resuspended in 1 ml of protein resuspension buffer. Protein concentrations of membrane and cytosolic fractions were determined by the BCA assay (Pierce). Aliquots were stored at −70°C.
For immunodetection proteins were heated in SDS sample buffer [60 mM Tris/HCl pH 6.8, 1.5% (v/v) 2-mercaptoethanol, 10% (v/v) glycerol, 2.4% SDS, 0.006% bromophenol blue] at 40°C for 20 min, then separated on an 8% SDS-PAGE gel, and finally blotted onto nitrocellulose membrane (Schleicher and Schuell). Sample buffer and gel optionally also contained 6 M urea. Blots were incubated with anti-Htr14D585 at a dilution of 1:30 000 followed by treatment with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:10 000, Sigma) and developed with SuperSignal West Pico chemiluminescence substrate (Pierce) according to the manufacturers’ protocols.
In vivo labelling of halobacterial transducer proteins with l-[methyl-3H] methionine and analysis of the adaptive release of volatile methyl groups
The assay measuring the adaptive release of volatile methyl groups from halobacterial cells was essentially performed as described by Alam et al. (1989) with some modifications. Cells were grown to an OD600 of 0.8–0.9 as described (Oesterhelt and Krippahl, 1983). A total of 1.8 ODml of cells were centrifuged, washed twice and resuspended in BSH (pH 7.5). After 30 min incubation at 37°C in the presence of 130 µg ml−1 puromycin, the cells were radiolabelled with 30 µl (30 µCi, 375 pmol) of l-[methyl-3H] methionine (80 Ci mmol−1, Amersham Biosciences) and incubated at 37°C for additional 45 min. After centrifugation the cells were washed twice and resuspended in 1.5 ml of BSH followed by transfer to a 0.22 µm filter-unit. The chase buffer contained 0.1 mM l-methionine in BSH plus 10 mM KCN. It was pumped over the cells at a constant flow rate of 1.5 ml min−1 and fractions of 18 s were collected. Light stimuli of 120 W m−2 (measured behind the transparent part of the filter-unit) were applied by switching on or off a 450 W xenon lamp equipped with cut-off filters KG3 and OG570 (Schott). Leucine stimuli were applied by switching to chase buffer that additionally contained 10 mM l-leucine and back to chase buffer without leucine, by use of a three-way valve. Fractions of 450 µl were collected in Eppendorf tubes which were placed upright in vials containing 6 ml of scintillation liquid (EcoSzint Plus, Roth). All vials were tightly closed in uninterrupted succession at the end of fraction collection. Volatile radioactivity was determined after ≥ 48 h, when ≥ 70% had entered the scintillation liquid (as determined in a control experiment using known amounts of 14C-methanol).
Measurements of pH changes in cell suspensions upon changes in illumination
Measurements were basically performed as described (Wagner et al., 1978). In brief, MKK101 cells grown to an OD600 of 0.94 were washed several times and resuspended in 8 ml of BSWC to a final OD of 0.63. After incubation in a glass vessel at 35°C in the dark under a constant flush of nitrogen for 2 h, the cells were illuminated with light of 80 W m−2 (measured inside the vessel) or fractions thereof, passed through cut-off filters (Schott) KG1, W360 and OG570. Illumination periods of approximately 10 min were interrupted by equally long dark phases. Upon switching off the light, the initial rates of external pH decreases were measured. Finally, pH changes were calibrated with a standard solution of HCl.
We thank Douglas Griffith and Birgit Wiltschi for critically reading and discussing the manuscript and Wilfried Staudinger for his help in the generation of strains WFS101 and WFS102.
AppendixS1. Strains, strategy of strain generation and description of the red-blue clone selection method.
AppendixS2. Working model of an MpcT (Htr14) structure based on a proposed Tsr model, which is consistent with the EM data of overexpressed E. coli Tsr.
Fig.S1. (included in Appendix S1) A. Overview of the generated strains leading to retinal protein-deficient strain OMI1, and strains MKK101 and MKK102. B. Gene loci of the four halobacterial retinal protein genes (hop, bop, sopI and sopII). C. Southern blot analysis of the four retinal protein gene loci in strain OMI1. D. Southern blot analysis of hop-complemented strain MKK102 with DNA probe COREhop. E. PCR analysis of bop-complemented strain MKK101.
Fig.S2. (included in Appendix S2) A. Proposed structures of halobacterial MpcT (Htr14) and E. coli Tsr. B. Section of the suggested hexagonal array of Tsr molecules, viewed from the plane of the membrane into the cytosol.