A single flavoprotein, AppA, integrates both redox and light signals in Rhodobacter sphaeroides

Authors

  • Stephan Braatsch,

    1. Institut für Mikrobiologie und Molekularbiologie, Universität Giessen, Heinrich-Buff-Ring 26–32, D-35392 Giessen, Germany.
    Search for more papers by this author
  • Mark Gomelsky,

    1. Department of Molecular Biology, University of Wyoming, Laramie, WY 82071-3944, USA.
    Search for more papers by this author
  • Silke Kuphal,

    1. Institut für Mikrobiologie und Molekularbiologie, Universität Giessen, Heinrich-Buff-Ring 26–32, D-35392 Giessen, Germany.
    Search for more papers by this author
  • Gabriele Klug

    Corresponding author
    1. Institut für Mikrobiologie und Molekularbiologie, Universität Giessen, Heinrich-Buff-Ring 26–32, D-35392 Giessen, Germany.
    • *For correspondence. E-mail Gabriele.Klug@mikro.bio.uni-giessen.de; Tel. (+49) 641 993 5542; Fax (+49) 64 993 5549

    Search for more papers by this author

Summary

Anoxygenic photosynthetic proteobacteria exhibit various light responses, including changing levels of expression of photosynthesis genes. However, the underlying mechanisms are largely unknown. We show that expression of the puf and puc operons encoding structural proteins of the photosynthetic complexes is strongly repressed by blue light under semi-aerobic growth in Rhodobacter sphaeroides but not in the related species Rhodobacter capsulatus. At very low oxygen tension, puf and puc expression is independent of blue light in both species. Photosynthetic electron transport does not mediate the blue light repression, implying the existence of specific photoreceptors. Here, we show that the flavoprotein AppA is likely to act as the photoreceptor for blue light-dependent repression during continuous illumination. The FAD cofactor of AppA is essential for the blue light-dependent sensory transduction of this response. AppA, which is present in R. sphaeroides but not in R. capsulatus, is known to participate in the redox-dependent control of photosynthesis gene expression. Thus, AppA is the first example of a protein with dual sensing capabilities that integrates both redox and light signals.

Introduction

The most evolutionarily ancient genes involved in photosynthetic pigment production are present in anoxygenic photosynthetic proteobacteria of the family Rhodospirillaceae (Xiong et al., 1998; 2000). This suggests that the ancestors of Rhodospirillaceae were the first bacteria to use sunlight as a source of energy. However, beyond photosynthesis, little is known about the mechanisms of photoreception and light-dependent signal transduction in these bacteria. It was discovered only a few years ago that photoreceptors similar to well-characterized plant phytochromes have originated from bacteria. The bacterial phytochromes have now been identified in cyanobacteria (Hughes et al., 1997; Yeh et al., 1997), Deinococcus and Pseudomonas (Davis et al., 1999), yet their physiological function is not understood. The photoactive yellow protein PYP was identified in several bacterial species and shown to undergo a blue light-driven photocycle (Hoff et al., 1994). It was suggested to function as the receptor for phototaxis in Ectothiorhodospira (Sprenger et al., 1993). Some representatives of Rhodospirillaceae contain pyp gene homologues, e.g. certain species of R. sphaeroides and R. capsulatus; however, their roles have not yet been assigned. In Rhodospirillum centenum, the protein Ppr was identified that shows similarities to both PYP and phytochromes (Jiang et al., 1999). The Ppr protein affects light-dependent expression of a gene for chalcon synthesis but does not affect phototaxis. A role in the light-regulated formation of photosynthetic complexes could not be assigned to any of these light receptors to date. Indeed, our knowledge of sensing and transduction of light signals in bacteria remains poor. Anoxygenic photosynthetic proteobacteria of the genus Rhodobacter are extremely metabolically versatile. They can gain energy by aerobic respiration, anaerobic respiration, anoxygenic photosynthesis or fermentation. Rhodobacter forms a photosynthetic apparatus at low oxygen tension, i.e. under semi-aerobic conditions. Under high oxygen tension, production of the photosynthetic complexes is completely shut off in R. sphaeroides, whereas it persists at low levels in R. capsulatus. A drop in oxygen tension below a certain threshold level significantly increases formation of the photosynthetic complexes in both species even in the absence of light. The photosynthetic apparatus of these bacteria consists of two light-harvesting complexes (LH I and LH II), which channel energy to the photosynthetic reaction centre (RC), where charge separation and electron transport is initiated. Bacteriochlorophyll and carotenoids are the major photosynthetic pigments. The genes encoding pigment-binding proteins of the RC and the LH I are part of the puf operon, whereas the genes encoding pigment-binding proteins of the LH II complex are encoded by the puc operon. The oxygen-dependent regulation of photosynthesis genes has been studied extensively in Rhodobacter (reviewed by Zeilstra-Ryalls et al., 1998; Bauer et al., 1999; Gregor and Klug, 1999; Oh and Kaplan, 2000). A central role for oxygen-dependent gene expression could be assigned to the two-component system RegB/RegA in R. capsulatus (Sganga et al., 1992; Mosley et al., 1994) and its counterpart PrrA/PrrB in R. sphaeroides (Eraso and Kaplan, 1995). The sensor kinase PrrB does not sense oxygen concentration directly but, rather, a redox signal generated by the electron flow through the cbb(3) oxidase of R. sphaeroides (Oh and Kaplan, 2000). Dependent on this redox signal, PrrB/RegB undergoes autophosphorylation and transfers the phospho group to the PrrA/RegA protein (Inoue et al., 1995; Oh and Kaplan, 2000). The phosphorylated DNA-binding protein PrrA/RegA activates the transcription of several photosynthesis genes including those of the puf and puc operons (Eraso and Kaplan, 1996; Du et al., 1998; Bowman et al., 1999). Redox regulation occurs not only by activation of photosynthesis genes at low oxygen tension but also by gene repression at high oxygen tension. The CrtJ protein of R. capsulatus and its counterpart in R. sphaeroides, PpsR, repress the expression of the puc operon at high oxygen tension (Ponnampalam et al., 1995). For CrtJ, redox-dependent DNA binding was demonstrated (Ponnampalam and Bauer, 1997). Another redox regulator detected in R. sphaeroides is the AppA protein, which activates puc expression at high oxygen tension, presumably by releasing the effect of the PpsR repressor protein (Gomelsky and Kaplan, 1995; 1997). In addition to the redox sensor RegB/PrrB, other redox-active proteins affect the expression of photosynthesis genes in R. sphaeroides. The FnrL protein regulates puc expression, probably by binding to the promoter region (Zeilstra-Ryalls and Kaplan, 1998), whereas thioredoxin affects puf and puc expression by an as yet unidentified mechanism (Pasternak et al., 1999). Although many factors involved in the oxygen/redox control of photosynthesis genes have been identified in Rhodobacter, little is known about light-dependent signalling. Light is believed to be harmful to semi-aerobically grown Rhodobacter cells because of the formation of reactive oxygen species (ROS) during the simultaneous presence of the photosynthetic pigments, light and oxygen. In semi-aerobic cultures of R. sphaeroides exposed to light, puf and puc expression is repressed, and formation of the photosynthetic complexes is decreased (Shimada et al., 1992). Shimada et al. (1992) showed that blue light is most potent in repressing puf and puc expression in R. sphaeroides. Buggy et al. (1994) described the involvement of the R. capsulatus hvrA gene during low light activation of puf transcription. A homologous gene in R. sphaeroides, the spb gene, was found to repress puf expression at elevated photon fluence rates (Shimada et al., 1996). A more recent analysis revealed that HvrA and Spb are proteins of the H-NS family of DNA-binding proteins (Bertin et al., 1999), suggesting that they are unspecific regulators of various different genes. For HvrA, involvement in the regulation of nitrogen fixation genes has been demonstrated (Kern et al., 1998). We have studied and compared the blue light-dependent expression of puf and puc operons in the two related species R. sphaeroides and R. capsulatus. We find that responses to blue light in these species differ and depend on the concentration of oxygen. The R. sphaeroides flavoprotein AppA acts as a mediator of blue light-dependent puf and puc repression and is likely to be the primary photoreceptor. As AppA is known to be part of the redox-dependent signal chain, this protein would thereby be capable of sensing two different signals, i.e. blue light and a redox signal. The FAD cofactor of AppA is essential for the maintenance of blue light-dependent gene repression, but not for redox sensing.

Results

Semi-aerobically grown cultures of R. sphaeroides and R. capsulatus show different responses to blue light

It has been reported by Shimada et al. (1992) that the expression of puf and puc operons is repressed by light in semi-aerobically grown R. sphaeroides (98 ± 25 μM dissolved oxygen). Repression was maximal during illumination with blue light (400–500 nm). We tested whether the same was true for the related bacterium R. capsulatus. The cultures of R. capsulatus were grown in the dark under semi-aerobic conditions (104 ± 24 μM dissolved oxygen) and then exposed to blue light at a fluence rate of 20 μmol m−2 s−1. The levels of puf and puc mRNA in these cultures remained unchanged compared with the dark control (Fig. 1). The doubling times of the cultures, 130–150 min, did not change during the course of the experiment (data not shown). These results are in contrast to the observations described for R. sphaeroides (Shimada et al., 1992). In order to test whether the difference in the responses to blue light between R. capsulatus and R. sphaeroides resulted from differences in the experimental conditions, we repeated the experiment with R. sphaeroides. We observed a blue light-dependent decrease in puf and puc mRNA levels similar to that described by Shimada et al. (1992) (Fig. 1). After 60 min of illumination with blue light, the level of puf mRNA decreased to 35% of the level present in the dark control, whereas the puc mRNA was no longer detectable. In contrast, puf and puc mRNA levels in the control cultures kept in the dark showed little variation over the time course of the experiment. Thus, our results demonstrate that R. capsulatus and R. sphaeroides differ in blue light-dependent regulation of gene expression.

Figure 1.

Northern blot analysis of RNA isolated from R. sphaeroides 2.4.1 or R. capsulatus SB1003. Cells grown at 104 ± 24 μM dissolved oxygen were shifted from the dark into blue light or kept in the dark. Total RNA was isolated at the indicated time points. puc -, puf - and rRNA-specific DNA probes were used for hybridization.

The blue light repression of puf and puc expression in R. sphaeroides depends on the oxygen tension

As shown above, blue light strongly represses puf and puc expression when R. sphaeroides is grown in the presence of ≈ 100 μM dissolved oxygen. To see whether the blue light response is affected by the oxygen concentration, we repeated the experiment at much lower oxygen tension. The oxygen tension dropped from 10– 15 μM to <3 μM during the first 45 min of illumination. In contrast to the repression observed in semi-aerobically grown cultures, no light-mediated repression was observed at this very low oxygen tension (Fig. 2). An increase in puf and puc mRNA levels was observed in both species. This increase in mRNA levels resulted from the decrease in oxygen tension. The low oxygen tension is expected to increase puf and puc expression, and apparently overpowers the effect of blue light in R. sphaeroides. As the blue light-dependent repression of puf and puc in R. sphaeroides requires a certain minimum oxygen tension, we considered the possibility that R. capsulatus might require an oxygen tension higher than 100 μM to manifest this response. We found, however, that blue light did not affect puf and puc expression in R. capsulatus even when grown at high oxygen tension, i.e. 200 μM dissolved oxygen (not shown). Thus, blue light indeed differentially affects puf and puc expression in R. sphaeroides and R. capsulatus. We grew R. sphaeroides or R. capsulatus anaerobically with DMSO as terminal electron acceptor. When the cultures were illuminated with blue light, an increase in puf and puc mRNA levels (factors 2.5–6) was observed (data not shown). The growth rates did not change significantly after illumination, implying that no significant gain in energy by photosynthesis occurred. Under these growth conditions, the formation of photosynthetic complexes is induced. We conclude that multiple mechanisms are involved in response to blue light under different physiological conditions.

Figure 2.

Northern blot analysis of RNA isolated from R. sphaeroides 2.4.1. Aerobically grown cells were shifted from the dark into blue light or kept in the dark. The oxygen tension decreased from 10– 15 μM dissolved oxygen to <3.4 μM during the first 45 min of the time course. Total RNA was isolated at the indicated time points. puc- , puf - and rRNA-specific DNA probes were used for hybridization.

Involvement of a specific light receptor(s) in blue light-dependent repression of puf and puc operons in R. sphaeroides

For R. sphaeroides, evidence was presented that electron flow through the components of the respiratory chain converts the oxygen signal into an intracellular redox signal (Oh and Kaplan, 2000). It was suggested that oxygen and light signals could act on the transcriptional repressor PpsR via the redox state of the quinone pool (Oh and Kaplan, 2001). In order to test for the involvement of the photosynthetic electron transport in blue light repression of photosynthesis genes, we investigated the R. sphaeroides reaction centre mutant PUHA1. Here, the puhA gene for the non-pigment-binding subunit of the reaction centre has been insertionally inactivated, preventing the assembly of functional reaction centres: thus, no photosynthetic electron transport takes place (Sockett et al., 1989). We found that PUHA1 showed the same light-dependent repression of the puf and puc operons as the wild-type strain (Fig. 3), implying that photosynthetic electron transport is not involved in transmission of the blue light signal. This role must be played by one or more specific blue light receptor(s). We exposed semi-aerobically growing R. sphaeroides cultures to blue light of lower fluence rates for 90 min in order to determine the minimum fluence rate for the blue light-dependent puf and puc repression. Fluence rates below or equal to 8 μmol m−2 s−1 had no significant effect on the puf and puc expression compared with the cultures kept in the dark (not shown). We conclude that the fluence rate of 20 μmol m−2 s−1 is close to the minimum required for this blue light response.

Figure 3.

Northern blot analysis of RNA isolated from R. sphaeroides PUHA1 mutant that does not assemble functional reaction centres. Experimental conditions are the same as in the legend to Fig. 1 .

Maintenance of blue light-dependent repression during illumination of R. sphaeroides is mediated by the FAD cofactor of the AppA protein

The most common blue light receptors are the flavoproteins, cryptochromes and phototropins of plants and animals and the PYP-type proteins of bacteria. The 2.4.1 strain of R. sphaeroides used in this study does not contain a pyp gene as evidenced by (i) the absence of pyp in the genome sequence of this strain (http://genome.ornl.gov/microbial/rsph); and (ii) negative hybridization results with the heterologous pyp probe (M. Gomelsky and S. Kaplan, unpublished data). Action spectra for the blue light-dependent puf and puc expression in R. sphaeroides determined by Shimada et al. (1992) revealed maximal repression around 420 nm. This maximal repression is consistent with photoreception via flavin. We decided to test whether the flavoprotein AppA has a role in this process. AppA binds a FAD cofactor at a flavin-binding domain of a new type (Gomelsky and Kaplan, 1998) and is known to be involved in redox-dependent regulation of the puc operon in the absence of light (Gomelsky and Kaplan, 1995; 1997). It is required for activation of puc transcription at low oxygen tension, most probably by antagonizing the repressing effect of the DNA-binding protein PpsR (Gomelsky and Kaplan, 1995; 1997). Our choice of AppA as a putative mediator of blue light repression in R. sphaeroides was supported by the fact that R. capsulatus, which is devoid of the blue light effect, does not contain an AppA homologue. This is based on the lack of an appA homologue in the genome sequence of R. capsulatus (http://wit.integratedgenomics. com/IGwit/) and by experimental evidence (C. Bauer, Department of Biology, Indiana University, Bloomington, IN, USA, personal communication). We first tested the effect of blue light on puf and puc expression in the AppA null mutant, APP11 (Gomelsky and Kaplan, 1995). As anticipated, the puc mRNA level in strain APP11 grown semi-aerobically in the dark was significantly lower than in the wild type. This is in agreement with the view that AppA is required to release the repressing effect by PpsR. Upon illumination with blue light, no significant decrease in the very low puc mRNA levels was observed (not shown), suggesting that, in the absence of AppA, the puc operon is repressed independently of blue light. Strain APP11 also showed considerably reduced puf mRNA levels in the dark compared with the wild type. These low puf mRNA levels were further reduced by about 50% in the presence of blue light. Under all conditions tested, the low level of puf mRNA in APP11 is similar to the level of puf mRNA in the wild-type strain after illumination (Fig. 4). Thus, the high levels of puf and puc mRNA of wild-type cells in the dark and their strong decrease after blue light illumination depends on the presence of AppA. To reveal the possible light-dependent involvement of AppA, we focused on the role of its FAD cofactor in blue light repression. In order to test for the role of the FAD cofactor of AppA, we compared puf and puc expression in the AppA null strain, APP11, expressing from a plasmid either a full-length AppA protein (plasmid p484-Nco5) or mutant AppA proteins that do not bind FAD (plasmids p484-Nco5-F and p484-Nco5Δ). The appA genes present in plasmids p484-Nco5 and p484-Nco5-F are identical except for the two point mutations in the gene on plasmid p484-Nco5-F. These mutations result in the substitution of two neighbouring amino acids of AppA, i.e. (51, 52) TG to AA. Plasmid p484-Nco5Δ expresses an AppA that contains an in frame deletion of the 185 N-terminal amino acids comprising the FAD-binding domain (Gomelsky and Kaplan, 1998). Both mutant AppA proteins are incapable of FAD binding, as revealed by spectral analysis of the proteins after overexpression and purification from Escherichia coli (Gomelsky and Kaplan, 1998; unpublished data). Under semi-aerobic conditions in the dark, the puf and puc mRNA levels in the APP11 strains expressing the flavin-deficient AppA proteins (plasmids p484-Nco5-F or p484-Nco5Δ) were almost identical to the levels observed in APP11 expressing the intact AppA protein (plasmid p484-Nco5). Upon illumination with blue light, the puf and puc mRNA levels in all strains decreased to similar values (p484-Nco5-F, Fig. 5; p484-Nco5Δ, data not shown). This indicates that (i) the flavin-deficient mutant AppA proteins are virtually intact regarding redox regulation of gene expression in the dark; and (ii) the FAD cofactor bound to AppA is not involved at the onset of blue light repression. We noticed, however, that although the puf and puc mRNA levels in the strain that expressed the full-length AppA from plasmid p484-Nco5 remained low during prolonged exposure to blue light, the mRNA levels recovered significantly in the strains that expressed either of the flavin-deficient AppA proteins (p484-Nco5-F, Fig. 5; p484-Nco5Δ, data not shown). In order to confirm our data further, the measurements were repeated for the time points 0 and 135 min to obtain average values. As an average percentage of puf inhibition (calculated as described in the legend to Fig. 5; all mRNA quantifications were normalized to the rRNA level; arithmetic mean ± sample standard deviation given) from five independent experiments, we determined 58.8 ± 8.3% for the control strain but only 15.4 ± 5.4% for APP11(p484-Nco5-F). We also compared the mRNA levels after 135 min of illumination with the mRNA levels at time 0 before illumination. The puf mRNA levels decreased to 42.7 ± 7.5% of the level at t = 0 in strain App11(p484-Nco5) but only to 98 ± 3.03% in strain App11(p484-Nco5-F). As an average percentage for puc inhibition, we determined 73.2 ± 12.1% for control strain App11(p484-Nco5) but only 27.8 ± 8.4% for strain App11(p484-Nco5-F). During 135 min of illumination, the puc mRNA levels decreased to 35.7 ± 15.5% of the level at time point t = 0 in the control strain, but only to 95.5 ± 4.9% in the mutant strain that expresses the FAD-deficient AppA protein. These data suggest that the FAD cofactor of AppA is required to maintain maximal repression of puf and puc operons during continuous blue light illumination. Therefore, AppA responds to two distinct stimuli through different sensing mechanisms: a redox signal, which is sensed independently of FAD, and blue light, which is sensed via FAD. The light-dependent increase in puf and puc expression after blue light illumination of anaerobically grown R. sphaeroides is not mediated by the FAD group of AppA, as it is also observed in the mutant APP11(p484-Nco5Δ), in which the N-terminal FAD-binding domain of AppA is deleted (Fig. 6). Both strains showed about 3.5- to fourfold increase in puf mRNA levels and six- to sevenfold increase in puc mRNA levels after 180 min of illumination with blue light. We conclude that this bacterium has evolved multiple blue light-dependent signal chains.

Figure 4.

Relative radioactivity of Northern blot signals of the 0.5 kb puf mRNA from strain R. sphaeroides 2.4.1 (open circles) or APP11 (closed circles). Experimental conditions for the light shift and RNA detection are the same as in the legend to Fig. 1 . The intensities of the signals were quantified by phosphoimaging, and the phosphostimulated fluorescence was plotted as relative radioactivity.

Figure 5.

Northern blot analysis of RNA isolated from R. sphaeroides APP11(p484-Nco5-F) and R. sphaeroides APP11(p484-Nco5). Experimental conditions are the same as in Fig. 1 . The intensities of the signals were quantified by phosphoimaging, and the intensities of the mRNA bands was normalized to the intensity of the rRNA. Percentage inhibition of puf and puc mRNA levels was plotted [inhibition as a percentage = 100 × (1–mRNA level in light-irradiated cells/mRNA level in dark cells)]. Values for APP11(p484-Nco5) are represented by open circles, values for APP11(p484-Nco5-F) by closed circles. The higher appA gene dosage in the AppA null strain expressing the full-length AppA protein from a plasmid is responsible for higher puc mRNA levels compared with the wild-type strain 2.4.1 ( Gomelsky and Kaplan, 1998 ).

Figure 6.

Relative increase in puf (circles) and puc (diamonds) mRNA levels in strains R. sphaeroides APP11(p484-Nco5Δ), indicated by broken lines, and R. sphaeroides APP11(p484-Nco5), indicated by solid lines, grown anaerobically in the presence of DMSO, as determined by Northern blot analysis. Light conditions are the same as in Fig. 1 . The intensities of the signals were quantified by phosphoimaging. The ratio of mRNA levels measured in the light and in the dark control was calculated after normalizing to the rRNA signals. The relative increase in this ratio is plotted as a percentage. The graph represents the result from one single experiment, but the result was confirmed in repeated experiments.

Discussion

Light-dependent responses have been studied extensively in higher organisms, and a number of photoreceptors and proteins involved in light signal transduction have been identified and characterized. However, our knowledge of photoreceptors and light-dependent signal transduction chains in prokaryotes and their evolution is limited. The discovery of phytochromes in bacteria (Hughes et al., 1997; Yeh et al., 1997; Davis et al., 1999; Jiang et al., 1999) proved that this photoreceptor evolved much earlier than previously assumed. We initiated a systematic study of the light-dependent processes in representatives of the genus Rhodobacter, the ancestors of which are likely to be the first organisms to convert light energy into ATP by anoxygenic photosynthesis (Xiong et al., 2000).

Responses to blue light in Rhodobacter

The data presented in this study suggest that responses of the related Rhodobacter species to blue light are (i) different depending on oxygen tension; and (ii) species specific. Under very low oxygen tension, blue light has no effect on the expression of photosynthesis operons in R. sphaeroides and R. capsulatus, whereas it leads to increased expression under anaerobic conditions in both species. During semi-aerobic growth, blue light represses the puf and puc operons in R. sphaeroides but has no effect in R. capsulatus. This indicates that multiple blue light-dependent signal transduction chains are involved. We focused on the blue light effect under semi-aerobic conditions. Below, we present possible explanations for the observed difference in species behaviour under these conditions. The process of anoxygenic photosynthesis developed under the reducing atmosphere of the primeval Earth and is designed to operate under anaerobic conditions. In the presence of oxygen and light, the photosynthetic complexes generate harmful ROS. Under high oxygen tension, the production of photosynthetic complexes is strictly repressed in R. sphaeroides. In R. capsulatus, however, low amounts of photosynthetic complexes are produced even at high oxygen tension, despite the risk of generating ROS. Such incomplete repression of photosynthetic complexes could be beneficial, as it allows for a faster adaptation to photosynthetic energy conversion if oxygen tension suddenly decreases and limits energy generation by respiration. It is conceivable that R. capsulatus has developed more efficient systems for protection against ROS compared with R. sphaeroides. The latter might compensate for the lack of efficient protection against ROS by a combination of two strategies: (i) tighter oxygen-dependent repression of photosynthesis genes; and/or (ii) an additional repression system that operates when light and oxygen are present together. This hypothesis needs to be tested in the future.

AppA as a mediator of the blue light effect and putative blue light receptor in R. sphaeroides

Action spectra for the blue light-dependent puf and puc repression in semi-aerobically growing R. sphaeroides (Shimada et al., 1992) indicate that this response may be mediated by a flavin. Although flavoproteins are known to function as blue light receptors in plants and animals, to our knowledge, no bacterial flavoprotein has been implicated in blue light sensing. It was suggested that photosynthetic electron flow mediates some of the light responses in Rhodobacter, which are further mediated by redox reaction of factors acting downstream (Gomelsky et al., 1999; Oh and Kaplan, 2001). For example, photosynthetic electron transport is required for phototaxis in R. sphaeroides (Grishanin et al., 1997), but the blue light-dependent phototaxis response in this bacterium also involves an additional, unidentified photosensor (Kort et al., 2000). Our data exclude a role for photosynthetic electron transport in blue light-dependent repression of photosynthesis genes in this organism. We distinguished two phases in blue light-dependent puf and puc repression: (i) onset of repression directly after exposure to blue light; and (ii) maintenance of repression upon continuous illumination. The photoreceptor responsible for the onset of repression remains to be identified. Other molecules with absorbance maxima around 420 nm are bacteriochlorophyll precursors such as protoporphyrin IX, Mg-protoporphyrin IX and protochlorophyllide a. Negative phototaxis in E. coli is mediated by protoporphyrin IX and depends on the presence of oxygen (Yang et al., 1995). The fact that the action spectra (Shimada et al., 1992) also revealed puf and puc repression at longer wavelengths indicates that additional photoreceptors might be involved. We have revealed that the flavoprotein AppA, which is present in R. sphaeroides but not in R. capsulatus, is involved in mediating the blue light effect in the former species. More precisely, the FAD group of AppA is required for the maintenance of puf and puc repression during illumination with blue light. Below, we present arguments that FAD bound to AppA could serve as a primary blue light receptor. The minimum fluence rate for the blue light-dependent puf and puc repression in R. sphaeroides is in the range of 8–20 μmol m−2 s−1, which is well within the range of blue light responses in plants. Higher fluence rates of blue light are required for avoidance reactions such as hmg1 gene repression in Arabidopsis (Learned, 1996), chloroplast movement in Arabidopsis, which is mediated by the flavin-containing phototropin NPL1(PHOT2) (Jarillo et al., 2001; Kagawa et al., 2001), and phototrophic curvature of Arabidopsis hypocotyls, which is mediated by PHOT1 and 2 (Sakai et al., 2001). Apparently, other flavin-containing receptors of blue light show similar sensitivity to blue light as the reaction mediated by the FAD cofactor of AppA. From the physiological point of view, it is not surprising that the blue light repression of the puf and puc operons in R. sphaeroides occurs at moderately high fluence rates. Such repression would only be required at dangerously high fluence rates, i.e. those at which harmful concentrations of ROS appear. At this time, we have no direct evidence that AppA is the primary photoreceptor. We have, however, been unable to find any molecule acting upstream of AppA (unpublished data).

AppA as an integrator of redox and blue light responses in R. sphaeroides

AppA is unique in mediating light and redox signals. The role of AppA in oxygen-dependent regulation is demonstrated by the fact that the AppA null mutant has low puf and puc expression under high oxygen tension and is impaired in the ability to increase expression after a drop in oxygen tension (Gomelsky and Kaplan, 1995). Because AppA can sense both a redox signal and blue light, it seems ideally suited to control the blue light-dependent repression of puf and puc operons in the presence of oxygen. The signal transduction scheme con-necting AppA to puf expression remains to be elucidated. The effect of AppA on puc expression is mediated by the transcriptional repressor PpsR, which represses puc (and photopigment genes) under high oxygen tension. AppA antagonizes (directly or indirectly) the repressor activity of PpsR (Gomelsky and Kaplan, 1997); thus, it functions as an antirepressor of PpsR. Based on the proposed role of AppA as a PpsR antirepressor, we offer the following model for the combined effects of blue light and oxygen on AppA (Fig. 7). At high oxygen tension, the AppA holoprotein exists in a hypothetical state I, in which it is unable to release the repression of photosynthesis genes by PpsR. A decrease in oxygen tension to semi-aerobic conditions, in the absence of light, results in the transition of AppA into state II through an as yet unknown redox-dependent reaction. This transition apparently does not involve the FAD cofactor of AppA. AppA in state II antagonizes PpsR and, therefore, PpsR-mediated repression of the puf and puc operons is released. When blue light is present, it interacts with FAD, facilitating reversion of AppA from state II to state I, thus leading to a tighter repression of the puf and puc operons. If oxygen tension is very low, the redox signal results in the transition of AppA into state III, which, like state II allows puf and puc expression. However, under these conditions, blue light can no longer revert AppA into state I. Thus, puf and puc expression at low oxygen tension is independent of light. The exact nature of states I, II and III are unknown. The dual sensitivity of AppA would make it well suited to integrate light- and oxygen-dependent signals to guarantee repression of the puf and puc genes under semi-aerobic conditions in the light. To our knowledge, AppA would thereby be the first example of a protein involved in both redox- and light-dependent sensory transduction. The AppA protein does not appear to contain a PAS domain, a highly conserved structure seen in diverse proteins, often binding flavin cofactors and involved in sensing stimuli such as light, oxygen or voltage (Taylor and Zhulin, 1999). Further studies on the light- and redox-dependent transitions of AppA and on the nature of the signal transmitted from AppA to PpsR are required to understand the further functioning of this unusual molecule. It will be of special interest to see whether proteins with similar dual sensory capacities are present in higher organisms or whether the putative simultaneous transduction of redox and light signals is rather an ancient feature that evolved later into specialized redox- or light-dependent pathways.

Figure 7.

Hypothetical model for the function of AppA as an integrator of redox and light signals. The AppA protein exists in a hypothetical state I under high oxygen tension (about 200 μM dissolved oxygen). State I results in repression of puf and puc transcription independently of light. Under medium oxygen tension (about 100 μM dissolved oxygen), AppA undergoes a transition into state II, which allows the expression of photosynthesis genes but can be reverted into state I by a light signal. Under very low oxygen tension (<10 μM dissolved oxygen), AppA exists in state III, which allows the expression of photosynthesis genes but, in contrast to state II, cannot be reverted into state I by a light signal. The influence of AppA on DNA binding of transcription repressor PpsR is not necessarily by direct interaction as implied in the model but could also occur by indirect mechanisms.

Experimental procedures

Bacterial strains and growth conditions

The following strains were used in this study: R. capsulatus SB1003, R. sphaeroides 2.4.1, R. sphaeroides 2.4.1 PUHA1, which is devoid of the photosynthetic electron transport (Sockett et al., 1989), and APP11, the appA deletion mutant of 2.4.1 (Gomelsky and Kaplan, 1995). Plasmid p484-Nco5 carries the full-length appA gene (Gomelsky and Kaplan, 1995). Plasmid p484-Nco5Δ carries the appA gene that contains an in frame deletion corresponding to the FAD-binding domain of the protein (Gomelsky and Kaplan, 1998). Plasmid p484-Nco5-F carries the appA gene containing point mutations that result in the substitution of two amino acids of AppA, i.e. Thr Gly (51, 52) → Ala Ala. The strains were cultivated at 32°C in a malate-minimal salt medium (Drews, 1983). The overnight cultures used for the experiments contained ≈ 3–10 μM dissolved oxygen. For semi-aerobic growth, oxygen concentration in the culture flasks was 104 ± 24 μM throughout the experiment. The cultures were adapted to these conditions for one doubling time before illumination. Oxygen tension was monitored with a Pt/Ag-electrode (microoxygen sensor 501, Ums) and adjusted by varying the rotation speed of the shaker. For studying the blue light effect at low oxygen concentration, the experiments were started with cultures that had an optical density of about 0.3 (660 nm) and contained 10–15 μM dissolved oxygen. Owing to increasing cell number, the concentration of dissolved oxygen dropped to <3 μM during the first 45 min of incubation. Under these conditions, oxygen is still dissolved at the liquid–air surface of the gently agitated culture flasks, but the bacteria immediately use up the dissolved oxygen for respiration. For anaerobic growth, 60 mM dimethyl sulphoxide (DMSO) was added to the medium as terminal electron acceptor, and cells were grown at 32°C in the dark in screw-cap flasks filled to the top. When required, antibiotics were added to the growth medium at the following concentrations: tetracycline 1.5 μg ml−1; kanamycin 25 μg ml−1; trimethoprim 50 μg ml−1. In the presence of light, no tetracycline was used.

Blue light experiments

The light source was the lamp of a slide projector (250 W, 24 V; General Electric). The light passed through a heat-absorbing filter (KG1; Schott) and a bandpass filter (BG12; Schott). The BG12 filter transmits blue light maximally at 400 nm; no light is transmitted above 500 nm. The irradiance produced by this light source measured at the culture level (Light Meter Li-189; Li-Cor) was 20 μmol m−2 s−1. For light shift experiments, cultures grown in the dark to an optical density at 660 nm of 0.1–0.3 were transferred into the light at time zero. Flasks containing cultures grown in the dark were wrapped in aluminium foil. The variation in light fluence rates was achieved using a threshold box containing an array of mirrors. Each mirror is passed by approximately half the light flow and reflects the other half which is used to illuminate a culture. Using this experimental set-up, a number of cultures can be investigated in parallel in otherwise identical conditions.

RNA isolation and quantification

Samples from R. capsulatus and R. sphaeroides cultures were taken at the indicated time points, and total RNA was isolated as described previously (Nieuwlandt et al., 1995). Total RNA (7.5–10 μg per lane) was run on a 1% (w/v) agarose 2.2 M formaldehyde gel (Sambrook and Russell, 2001) and transferred to nylon membrane (Pall Biodyne B) by vacuum pressure blotting (vacuum blotting; Amersham Pharmacia Biotech) according to the manufacturer's recommendations. puc- and puf-specific DNA fragments from R. capsulatus and R. sphaeroides were radiolabelled with [α-32P]-dCTP using nick translation (nick translation kit; Amersham Pharmacia Biotech). The oligonucleotide used for hybridization with processed Rhodobacter 23S rRNA was 5′-CTTAGATGTTTCAGTTCCC-3′ corresponding to the 23S rDNA positions 187–205 (E. coli numbering). Aliquots of 20 pmol of oligonucleotide were labelled for 30–60 min at 37°C with 20 μCi of [γ-32P]-ATP using polynucleotide kinase. The DNA fragments used for hybridizations were purified on microcolumns (Probe Quant G-50; Amersham Pharmacia Biotech). Aliquots of 2 × 106 c.p.m. were used per hybridization reaction. The signals were quantified using a Phosphoimaging system (Molecular Imager FX; Bio-Rad) and the appropriate software (QUANTITYONE; Bio-Rad).

Acknowledgements

We thank A. Batschauer and P. Galland for making equipment for light experiments available to us, P. Galland for many valuable discussions, and J. Hughes for correcting the manuscript. M.G. is grateful to S. Kaplan for support and encouragement of the earlier work on AppA. This work was supported by grants from Fonds der Chemischen Industrie and BMBF to S.B. and G.K., DFG (KL 563/15-1) to G.K., and from NIH NCRR (P20RR15640-01) to M.G.

Ancillary