A new type of bacteriophytochrome acts in tandem with a classical bacteriophytochrome to control the antennae synthesis in Rhodopseudomonas palustris.

Phytochromes are chromoproteins found in plants and bacteria that switch between two photointerconvertible forms via the photoisomerization of their chromophore. These two forms, Pr and Pfr, absorb red and far-red light, respectively. We have characterized the biophysical and biochemical properties of two bacteriophytochromes, RpBphP2 and RpBphP3, from the photosynthetic bacterium Rhodopseudomonas palustris. Their genes are contiguous and localized near the pucBAd genes encoding the polypeptides of the light harvesting complexes LH4, whose synthesis depends on the light intensity. At variance with all (bacterio)phytochromes studied so far, the light-induced isomerization of the chromophore of RpBphP3 converts the Pr form to a form absorbing at shorter wavelength around 645 nm, designated as Pnr for near red. The quantum yield for the transformation of Pr into Pnr is about 6-fold smaller than for the reverse reaction. Both RpBphP2 and RpBphP3 autophosphorylate in their dark-adapted Pr forms and transfer their phosphate to a common response regulator Rpa3017. Under semiaerobic conditions, LH4 complexes replace specifically the LH2 complexes in wild-type cells illuminated by wavelengths comprised between 680 and 730 nm. In contrast, mutants deleted in each of these two bacteriophytochromes display no variation in the composition of their light harvesting complexes whatever the light intensity. From both the peculiar properties of these bacteriophytochromes and the phenotypes of their deletion mutants, we propose that they operate in tandem to control the synthesis of LH4 complexes by measuring the relative intensities of 645 and 710 nm lights.

Phytochromes are biliprotein photoreceptors originally discovered in plants (1,2) but only recently in bacteria (3)(4)(5)(6). They respond to red/ far-red light via a reversible shift from a red absorbing form (Pr) to a far-red absorbing form (Pfr). These light sensors regulate many aspects of photomorphogenesis in plants (1,2). In bacteria, they have been also described to play various roles, control of the light harvesting complexes in Fremyella diplosiphon (3), regulation of chalcone synthase in Rhodospirillum centenum (5), regulation of carotenoid synthesis in Deinococcus radiodurans (6), phototaxis in Synechocystis (7), and control of photosystem synthesis in Bradyrhizobium and Rhodopseudomonas (Rds.) 3 palustris (8). Based upon primary sequence alignment, phytochromes and bacteriophytochromes possess a similar protein organization with an N-terminal chromophore binding domain (CBD) and a C-terminal module involved in signal transduction and dimerization. The plant and cyanobacteria chromophores are 3E-phytochromobilin and 3Z-phycocyanobilin, respectively (1,9), whereas other proteobacteria use, as chromophore, biliverdin, the simplest linear tetrapyrrole synthesized from heme by a heme oxygenase (10). Like plant phytochromes, most bacteriophytochromes studied so far contain a twocomponent histidine kinase motif at their C termini. However some exceptions occur such as the bacteriophytochromes of the photosynthetic bacteria Bradyrhizobium (BrBphP) and Rps. palustris (denoted here RpBphP1), which do not possess a histidine kinase motif but an S-box domain that might be involved in protein-protein interactions (8). Photoconversion of these two bacteriophytochromes from their Pfr to Pr form triggers the synthesis of the entire photosynthetic apparatus and the associated bacteriochlorophyll and carotenoid molecules (8,11,12). This photosynthetic apparatus is composed of a photochemical reaction center and of light-harvesting (LH) complexes. The photosynthetic apparatus of Rps. palustris presents the peculiarity of possessing several peripheral LH complexes, encoded by distinct pucBA genes (13). The relative proportion of these LH complexes varies according to the light environment (14,15). They differ by their absorption properties and carotenoid content. Some of these complexes present two distinct optical transitions absorbing around 800 and 850 nm in the near infrared. They are named LH2 (or B800 -850 LH2) to differentiate them from the LH1 complexes, which present a single broad absorption band around 870 nm. A novel type of LH, recently characterized and designated LH4 (or B800 LH2), is expressed under low light intensity and presents a single absorption band in the near infrared centered at 800 nm (15).
The recent sequencing of the complete genome of Rps. palustris strain CGA009 revealed, in addition to RpBphP1, the unexpected presence of five other putative bacteriophytochrome genes scattered over the genome (16). This suggests that this bacterium has developed a sophisticated and complex network of photoreceptors for its adaptation to light environment. Four of these bacteriophytochromes genes, RpBphP1 (rpa1537), RpBphP2 (rpa3015), RpBphP3 (rpa3016), and RpBphP4 (rpa1490), are located close to photosynthesis genes. The RpBphP1 gene is found inside the photosynthesis gene cluster, and, as already stated, its product controls the synthesis of the RC-LH1 core complexes (8). The RpBphP4 gene is located close to the pucBAe genes encoding one of the LH2 complexes. Remarkably, the RpBphP2 and RpBphP3 genes are organized in tandem downstream the pucBAd encoding the apoproteins of the LH4 complexes. Two other genes coding for LH complexes (the pucBAc genes rpa3009 and rpa3010) are located upstream the pucBAd genes. However, one of them, pucAc, contains a frameshift mutation and has been annotated as a pseudogene (16).
To better understand the role and function of these bacteriophytochromes in the adaptation of Rps. palustris to changes in its light environment, we initiated their systematic study at the genetic and biochemical level. Of particular interest are the two bacteriophytochromes, RpBphP2 and RpBphP3. What is the significance of the tandem organization of the two bacteriophytochromes genes? Do these two bacteriophytochromes possess different properties? Are these two bacteriophytochromes involved in the regulation of the closely located pucBAd genes or in different regulatory processes? In this report, we show by combining genetics, biochemical and biophysical approaches that the bacteriophytochrome RpBphP3 possesses unusual photochemical properties and works in tandem with RpBphP2 to regulate the synthesis of the LH4 complexes.

MATERIALS AND METHODS
Bacterial Strains and Growth Conditions-Rps. palustris CEA001 strain (11) was grown under semiaerobic conditions for 72 h at 30°C on Petri dishes filled with 40% Sistrom-agar medium and sealed with tape. The cells were subjected to continuous illumination provided by a series of light-emitting diodes (LEDs) of different wavelengths between 590 and 875 nm. Each LED illuminated a 3.5 cm 2 area. The half-peak bandwidth was below 25 nm for all wavelengths. Irradiance was adjusted between 1 and 200 mol of photon/m 2 /s. Expression and Purification of Rpa3014, RpBphP2, RpBphP3, and Rpa3017 Proteins-The rpa3014, RpBphP2, RpBphP3, and rpa3017 genes were amplified by polymerase chain reaction from genomic DNA of Rps. palustris CEA001 using primers designed to add appropriate restriction sites for expression as His 6 -tagged versions in pBAD/HisB expression vector (Invitrogen). To reconstitute in vivo the RpBphP2 and RpBphP3 holobacteriophytochromes, the hmuO gene from Bradyrhizobium ORS278 was amplified by polymerase chain reaction using primers designed to add a ribosome binding site upstream the gene and restriction site for insertion in the previous constructions pBAD::RpBphP2 and pBAD::RpBphP3. The recombinant proteins were overexpressed in Escherichia coli LMG194 and purified as described previously (11).
Protein Kinase Assays-Protein kinase reaction was performed as described previously (9). For autophosphorylation experiments, the bacteriophytochromes RpBphP2 or RpBphP3 were either dark-adapted for more than 1 h or pre-illuminated with saturating red light (705 nm) at 30°C. The kinase reactions were initiated by adding [␥-32 P]ATP, and the resulting mixtures were subjected to a further 1-h darkness condition or illumination. At this stage, the reactions were stopped by addition of the buffer used for the SDS-PAGE assay and analyzed by gel electrophoresis. For the phosphotransfer experiments, the RpBphP2 and RpBphP3 were incubated in the presence of [␥-32 P]ATP during 1 h in the dark or under 705 nm light and for an additional 10 min with equal amount of Rpa3014 or Rpa3017 or a mixture of Rpa3014 and Rpa3017. 32 P-Labeled products were quantified by using a Typhoon Phosphor-Imager (Amersham Biosciences).
Construction of RpBphP2 and RpBphP3 Mutant Strains-For the construction of the RpBphP2 null mutant, a 1.4-kb PstI fragment inside the RpBphP2 gene was deleted and replaced by the lacZ-Km r cassette of pKOK5 (17). The RpBphP3 null mutant was obtained by the deletion of a 0.5-kb SalI fragment inside the RpBphP3 gene and replacement by the lacZ-Km r cassette. These constructions were introduced in the pJQ200 suicide vector (18) and delivered by conjugation into the Rps. palustris CEA001 strain as described (11). Double recombinants were selected on sucrose and confirmed by polymerase chain reaction.
Absorbance Spectra Measurements-Absorbance spectra of purified RpBphP2 and RpBphP3 bacteriophytochromes were recorded with a Cary 50 spectrophotometer. Bacteriophytochrome spectra were recorded either in the dark or under continuous illumination at 645, 705, or 750 nm provided by LEDs with an irradiance of 15 mol of photon/m 2 /s. Light-induced absorbance changes were performed with a laboratory-built spectrophotometer similar to the one developed by Joliot et al. (19). The absorption level is sampled using 2-s monochromatic flashes given from 1 ms to several seconds after actinic excitation. Actinic illumination was provided by a Xenon flash (2 s).
Fluorescence Measurements-Fluorescence measurements have been performed at room temperature using a Cary Eclipse spectrophotometer. To perform the photoinduced transition of RpBphP2 and RpB-phP3, excitation light was provided by 645-, 705-, or 750-nm LEDs with an irradiance of 15 mol of photon/m 2 /s. For the measurement of the excitation spectra (2-nm slit bandwidth), the fluorescence was detected at 730 nm (10-nm slit bandwidth). For the emission spectra (2-nm slit bandwidth), the excitation was set at 400 nm (10-nm slit bandwidth). Measurements of the variation of the fluorescence yield were performed at 400 nm for the excitation wavelength (10-nm slit bandwidth) and 730 nm for the emission wavelength (2-nm slit bandwidth).
Light-induced FTIR Difference Spectra Measurements-Light-induced FTIR difference spectra were recorded at 10°C at 4 cm Ϫ1 resolution, with a Bruker 66 SX spectrometer equipped with a KBr beam splitter and nitrogen-cooled MCT-A detector. The FTIR samples consisted in 10 l of a solution of bacteriophytochrome at a concentration of Ϸ200 M, in 50 mM Tris-HCl, pH 8, 100 mM KCl, 20 mM MgCl 2 , deposited between two calcium fluoride windows sealed with silicon grease to avoid sample dehydration. The samples were subjected to successive light cycles, applying 705 nm light to generate Pfr (or Pnr) and 750 or 645 nm light to restore the Pr form of RpBphP2 or RpBphP3, respectively.

Expression and Purification of Recombinant Holobacteriophytochromes-The RpBphP2
and RpBphP3 genes belong to the same putative operon, which includes three other putative response regulator genes (rpa3014, rpa3017, and rpa3018) (Fig. 1A). This operon is found upstream the pucBAd genes encoding the apoproteins of the light-harvesting complexes LH4. Sequence analysis reveals that the RpBphP2 and RpBphP3 proteins are highly homologous, with 54% of sequence identity and 76% similarity. In contrast, they present only 30 -35% sequence identity with the other bacteriophytochromes of Rps. palustris. These two bacteriophytochromes display the classical bacteriophytochrome architecture with an N-terminal CBD and a C-terminal histidine kinase domain (Fig. 1B). To prove that RpBphP2 and RpBphP3 encode functional bacteriophytochromes, they were co-expressed with the heme oxygenase gene (required for chromophore synthesis) in E. coli and subsequently purified. Gel electrophoresis and zinc fluorescence (not shown) demonstrated that the biliverdin molecule was covalently bound in both proteins as previously reported for bacteriophytochromes of various bacterial species (8,10,20). Site-directed mutagenesis (data not shown) indicate that the Cys residues of the N terminus, located at position 16 and 28 for RpBphP2 and RpBphP3, respectively, are necessary for attachment of chromophore and are likely to be the site of attachment based upon their position and the previous data from the bacteriophytochrome Agp1 of Agrobacterium (20).
Absorption and Photochemical Properties-Unexpectedly, the purified RpBphP2 and RpBphP3 bacteriophytochromes present slightly different colors ( Fig. 2A). The absorption spectra of dark-adapted samples of RpBphP2 and RpBphP3 are typical of the Pr form with a main band centered at 710 and 705 nm, respectively (Fig. 2, C and D). Both RpB-phP2 and RpBphP3 behave as usual bacteriophytochromes assuming a Pr ground state after dark adaptation. This behavior is at variance to that reported for the first bacteriophytochrome (RpBphP1) studied in Rps. palustris, which was shown to be stable under its Pfr form (8,11). Presence of two bacteriophytochromes which differ by their dark-adapted state has also been reported in Agrobacterium tumefaciens (21). In this bacterium, the ground state of the bacteriophytochrome AtBphP1 is the Pr state, whereas the Pfr form of AtBphP2 corresponds to its darkadapted state (21). Upon illumination with a 705-nm light of RpBphP2 (Fig. 2C), a typical Pr/Pfr transition is observed with the partial bleaching of the main absorption band and appearance of a broad band cen-  tered at 750 nm (Fig. 2C). Illumination with a 750-nm light restores the dark-adapted spectrum. Dark reversion from the Pfr to the Pr state is a slow process, which requires several tens of minutes (Fig. 3A). Unexpectedly, illumination with the 705-nm light on a dark-adapted sample of RpBphP3 (Fig. 2D) leads to a large decrease of the main absorption band at 705 nm, but no concomitant absorption band develops in the infrared region. In addition, the shoulder at 650 nm of the Pr form, which corresponds to the vibrational transition of the S1 state, is not bleached. On the contrary, a minor increase in absorption is observed in this wavelength region. This suggests that the 705-nm illumination has induced a new form characterized by an absorption band centered around 650 nm. In agreement with this hypothesis, a fast and near complete reversion to the dark-adapted state is observed when the 705-nm illumination is followed by excitation with a 645-nm light (Fig.  2D). In the absence of 645-nm illumination, reversion to the darkadapted state is very slow. Total dark reversion requires more than 2 h (Fig. 3B).
The near complete reversion of RpBphP3 to the dark-adapted state by the 645-nm light (Fig. 2D) implies a large difference in quantum yield for the two photoreversible reactions. This difference in quantum yield between the two forms of RpBphP3 is also evidenced by the measurement of the initial rate of Pr formation per absorbed photon upon illumination by a 645-nm light, which is ϳ6-fold faster than the one measured for the photoconversion under 705-nm illumination (Fig. 2B). This behavior is different from that observed for RpBphP2 where the initial rates of the phototransformation of Pr into Pfr and of Pfr into Pr are similar (Fig. 2B). Note that these initial rate measurements, determined in the first 2 s of illumination, are not affected by the dark reversion of RpBphP2 and RpBphP3, which takes several tens of minutes in both cases (Fig. 3). The ratio between these initial rates is therefore a good estimate of the ratio of the quantum efficiencies of the photoconversion of the two photoconvertible states.
To determine the absolute absorption spectra of the two photoconvertible forms of RpBphP2 and RpBphP3, we have used the following method. The steady-state spectrum S 1 () obtained under illumination at a given wavelength, 1 , is a mixture of the spectra S A () and S B () of the two pure photoconvertible forms, with relative weights ␣( 1 ) and [1 Ϫ ␣( 1 )], respectively, The steady-state condition implies where the s stand for the quantum efficiencies for the conversion into the other form.
From Equation 1, and from Equation 2, where ϭ A / B . Making ϭ 1 in Equation 3 and equating with Equation 4, one obtains the relation between ␣ and , If S A () (dark-adapted spectrum) and S 1 () (under steady-state illumination at ϭ 1 ) are known, then ␣( 1 ) is a unique function of the ratio of quantum efficiencies ϭ A / B . Thus, the determination of S B () requires one additional information that may be (i) an independent measurement of or (ii) the acquisition of a steady-state spectrum under actinic illumination at another wavelength 2 . The first method was applied to both RpBphP2 and RpBphP3 bacteriophytochromes. The measurement of Ϸ 1 for RpBphP2 (Fig. 2B), implies that the S 705 spectrum of Fig. 2C is contributed by ␣ (705) Ϸ 30% of the Pr form. The corresponding S B () spectrum, i.e. the absorption spectrum of the Pfr form is shown in Fig. 4A. For RpBphP3, using Ϸ 6 (Fig. 2B), we found  that the S 705 spectrum of Fig. 2D is contributed by ␣(705) Ϸ 50% of the Pr form. Using method (ii) for RpBphP2 with the spectra obtained under steady-state illumination at 1 ϭ 705 nm and 2 ϭ 645 nm (Fig. 2D), we confirmed a value 6 of for . The pure S B () spectrum of RpBphP3 shown in Fig. 4B.
The absorption spectra of the pure Pr and Pfr forms of RpBphP2 are typical of bacteriophytochromes described so far (8,10). As already mentioned, the dark-adapted state of RpBphP3 is similar in shape to the Pr state of RpBphP2, i.e. a classical Pr spectrum. On the other hand, the light-induced state of RpBphP3 absorbs maximally at 650 nm. The absorption spectrum of this state is similar in shape to the Pfr state of RpBphP2 but blue-shifted by 95 nm (Fig. 4B). Because of its peak location, we name this new spectral form Pnr for pigment absorbing in the near red.
In conclusion, RpBphP3 presents unusual photochemical properties in terms of both quantum yield for photoconversion and wavelength position of the Pnr state.
Chromophore Conformation and Transient States-The new optical properties of the short wavelength state of RpBphP3 poses the problem of the biliverdin conformation in this state and the dark-adapted state. High similarities can be found between the dark-adapted state of RpB-phP3 and the classical Pr state of RpBphP2 and accordingly between the classical Pfr state and the new Pnr state of RpBphP3. A first argument comes from the almost identical absorption spectra of the two darkadapted states of these bacteriophytochromes (Fig. 4). A second argument is the observation that the dark-adapted states of both RpBphP2 and RpBphP3 are fluorescent. The fluorescence excitation spectra of RpBphP2 and RpBphP3 closely match their respective absorption spectra, and the emission spectra are centered at 725 and 720 nm, respectively (Fig. 5, A and B). On the other hand, a large fluorescence decrease occurs upon 705 nm illumination in both cases. This is shown in Fig. 5, C and D where the fluorescence yield of both bacteriophytochromes have been measured in the dark or under 705-nm illumination. The decreases in fluorescence yield, ϳ65 and 45% for RpBphP2 and RpB-phP3, respectively (Fig. 5, C and D), correspond nicely to the amounts of Pfr and Pnr states formed under such illumination (Fig. 2). The nonfluorescent character of the new Pnr state of RpBphP3 is a typical prop-   erty of the "classical" Pfr form of phytochromes (22). A third argument comes from the comparison of the FTIR difference spectra recorded with both bacteriophytochromes. The Pfr Ϫ Pr FTIR difference spectrum recorded with RpBphP2 (Fig. 6A) presents a striking similarity to that observed for the plant phytochrome (23,24) for which ZZZ to ZZE isomerization in C 15 of phytochromobilin was demonstrated using NMR spectroscopy. The occurrence of common infrared bands at 1653/1643-1638, 1248, 1118, 960 -958/947 cm Ϫ1 and especially at 1595/1583 cm Ϫ1 in the Pnr-Pr and Pfr-Pr spectra recorded with RpB-phP3 or RpBphP2 strongly supports that Z to E isomerization also occurs around the C 15 ϭC 16 double bond for RpBphP3. Previous studies on phytochromes using chromophores chemically modified at ring D or 18 O-labeled at ring A have shown that the contributions of the carbonyl group of ring A lie above 1730 cm Ϫ1 , whereas those of ring D are at 1710 -1696 cm Ϫ1 (24). Thus we assign the upshift of 1735-1750 cm Ϫ1 upon Pr to Pnr transition to ring A carbonyl, whereas the large positive band at 1698 cm Ϫ1 is assigned to the carbonyl group of ring D in the Pnr form of RpBphP3 (Fig. 6). The last argument comes from the analysis of the flash-induced absorption changes following an exciting flash. Fig.  7A shows the light-induced absorbance changes detected at different times, between 1 and 100 ms, after flash excitation of a dark-adapted sample of RpBphP2. The light-induced difference spectrum detected 1 ms after the exciting flash presents an absorbance increase centered at 740 nm and a bleaching around 640 and 705 nm. When the light-induced absorbance changes are detected at longer times, a shift to longer wavelengths is observed for the positive changes together with an increase of the amplitude of the negative changes (Fig. 7A). These lightinduced changes can be interpreted as follows. Flash excitation induces the formation of a short-lived state absorbing around 740 nm, which is transformed with a half-time of 10 ms (Fig. 7A, inset) into the stable Pfr state absorbing around 760 nm. The properties of this transient state, in terms of half-time and absorption at shorter wavelength than the stable Pfr form, are similar to the lumi-R state observed for plant phytochrome (22,25). An equivalent transitory state is observed upon phototransformation from the Pr to the Pnr state in the case of RpBphP3 (Fig. 7B). This short lived state peaks around 730 nm and is converted into the stable Pnr form with a half-time of 10 ms. On the other hand, we have observed no transition state, in the ms to second time range, for both the flash-induced transformation of Pfr into Pr or Pnr into Pr for RpBphP2 and RpBphP3, respectively (data not shown). We take the similarity between the transient states measured in RpBphP2 and RpBphP3 as supplementary evidence that their dark-adapted states correspond to a Pr state. Collectively, the above results strongly support an identical conformational state of the chromophore for the new Pnr form of RpB-phP3 and the classical Pfr form of RpBphP2. We therefore propose that the new Pnr form of RpBphP3 is equivalent to the classical Pfr but absorbing at much shorter wavelengths.
Autophosphorylation and Phosphotransfer-Both RpBphP2 and RpBphP3 bacteriophytochromes possess a histidine kinase module at their C-terminal regions (Fig. 1B). We have determined the correlation between this kinase activity and their conformational states by incubation in the presence of [␥-32 P]ATP. Both bacteriophytochromes autophosphorylate in the dark, i.e. under their Pr forms (Fig. 8A). In agreement with His serving as phosphoacceptor site, the autophosphorylated RpBphP2 and RpBphP3 are acid labile and base stable (data not shown). Illumination of RpBphP2 with a 705-nm light strongly represses its kinase activity indicating that its Pfr form is not able to autophospho-  rylate. Excitation with a 705-nm light induces a 45% decrease of the level of phosphorylation in the case of RpBphP3. This value is very close to the amount of state Pnr formed under this light condition (Fig. 2D), indicating that the Pnr form of RpBphP3 displays probably no kinase activity.
We next investigated whether the Pr forms of RpBphP2 and RpBphP3 could transfer their phosphates to the response regulators Rpa3014 and Rpa3017, whose genes are located on both sides of RpBphP2 and RpB-phP3 (Fig. 1, A and B). A rapid and efficient phosphotransfer was observed from both bacteriophytochromes to Rpa3017 (Fig. 8, B-D), indicating that this response regulator is a common element of the signaling pathways initiated by RpBphP2 and RpBphP3. In contrast, although the prediction of a helix-turn-helix DNA binding domain in Rpa3014 suggests that this response regulator could be the last element of the transduction pathway of RpBphP2 and RpBphP3, no phosphotransfer was observed from these bacteriophytochromes to Rpa3014, even in the presence of Rpa3017 (Fig. 8, C and D). We cannot exclude at the moment that this lack of phosphotransfer is because of an incorrect folding of the recombinant Rpa3014 protein. Another explanation could be that additional partners are implicated in the phosphotransfer from RpBphP2 or RpBphP3 to Rpa3014, via Rpa3017. One of these partners could be the response regulator containing also a histidinekinase domain, encoded by the rpa3018 gene found just downstream of rpa3017 (Fig. 1, A and B). Experiments to complete the characterization of this complex light signaling pathway are in progress.
Action Spectrum of LH4 Synthesis and Phenotypes of RpBphP2 and RpBphP3 Deletion Mutants-To optimize their metabolism according to environmental changes, photosynthetic bacteria regulate the expression level and composition of their photosystems by means of highly sophisticated mechanisms. In the case of Rps. palustris, growth under low light intensity under anaerobic condition results in the enhancement of the synthesis of LH4 complexes encoded by the pucBAd genes and a decrease of LH2 complexes encoded by the other pucBA genes (14,15). The pucBAd genes are located upstream the RpBphP2 and RpBphP3 genes. Based on these genes arrangement and the ability of RpBphP2 and RpBphP3 to phosphorylate the same response regulator, one appealing hypothesis is to suppose that these two bacteriophytochromes are involved in the regulation of the LH4 complexes.
To test this hypothesis, Rps. palustris CEA001 was subjected to illumination at various wavelengths provided by LEDs emitting between 590 and 875 nm. The experiment was performed under semiaerobic condition because of the requirement for oxygen in the bacteriophytochrome chromophore synthesis. Fig. 9A shows the absorption spectra of cells grown under some selected wavelengths. Clearly, the relative amount between the LH4 complexes and the LH2 complexes depends highly upon the illumination wavelength. The ratio between the absorptions at 805 and 860 nm (A805/A860), which is a relative measure of the ratio between LH4 and LH2 complexes, is shown in Fig. 8B. The synthesis of the LH4 complexes, concomitant with a decrease in LH2 complexes, is specifically enhanced for wavelengths comprised between 680 and 730 nm with a maximum around 705-710 nm (Fig. 9B). This wavelength range corresponds precisely to illumination for which both RpB-phP2 and RpBphP3 are predominantly under their non-phosphorylated states, Pfr and Pnr forms (Fig. 8A). Indeed, on the one hand, illumination with wavelengths longer than 740 nm, not absorbed by the dark Pr forms of both RpBphP2 and RpBphP3, does not modify the phosphorylation state of these bacteriophytochromes, which both remain phosphorylated. On the other hand, excitation with wavelengths shorter than 670 nm converts RpBphP2 to its non-phosphorylated Pfr form but maintains RpBphP3 predominantly in its phosphorylated Pr state (Figs.  2 and 8).
Because the enhancement of LH4 complexes synthesis has been reported to occur only at very low light intensity under anaerobic conditions (15), we have followed the LH4 synthesis as a function of the intensity of 710 nm light. This enhancement was half-saturated for intensity of 5 mol of photon/m 2 /s and remained constant up to 200 mol of photon/m 2 /s (not shown). This unexpected behavior is probably because of the difference in growth conditions between the present work (semiaerobic) and the previous report (anaerobic) (15).
Because of the peculiar absorption and quantum yield properties of the RpBphP3 Pnr state, we expect that both bacteriophytochromes are in their dephosphorylated activating state only when the 710-nm light is in excess to 645-nm light. To test this possibility the synthesis of LH4 complexes has been measured as a function of the ratio between 645 and 710 nm light intensities. We indeed observed that increasing this ratio led to a large decrease in the synthesis of LH4 complexes. Typically, the ratio A805/A860 decreases from 2.0 to 1.2 when the ratio between 645 and 710 nm light intensities increases from 0.1 to 10 (not shown).
To definitively prove that the LH4 synthesis is regulated by the action of RpBphP2 and RpBphP3, we have examined the phenotype of mutants deleted in RpBphP2 or RpBphP3 genes. In contrast to the wild-type strain, no synthesis of LH4 complexes was observed for these mutants irrespective of the illumination intensity and wavelength (not shown).

DISCUSSION
Based on the results described above, we conclude that the lightinduced dephosphorylation of RpBphP2 and RpBphP3 is the first step of the signal transduction pathway of the LH4 synthesis. These two bacteriophytochromes act as the inputs to an "and gate" for the regulation of the LH4 synthesis as a function of the relative intensities of 645 and 710 nm light. We propose that measurement of this ratio is an indirect detection of the presence of phytoplankton. Indeed, in their natural environment, photosynthetic bacteria may develop below dense layers of phytoplankton. Because of the strong absorption of cyanobacteria and microalgae chromophores in the visible region up to 700 nm, the ratio of photons of 645 and 710 nm reaching the lower layer of photosynthetic bacteria is significantly affected by the presence of phytoplankton. In particular, photosynthetic bacteria receive only far-red light in these conditions. This significantly decreases their available light energy. To overcome this limitation in light supply, Rps. palustris activates the synthesis of LH4 complexes, via the action of RpBphP2 and RpBphP3 and decreases the synthesis of LH2 complexes. These changes in LH complexes significantly enhance the light capture efficiency of the bacteria. First, the infrared absorption of all the bacteriochlorophyll molecules of the LH4 complexes is centered around 800 nm, whereas