Two distinct crt gene clusters for two different functional classes of carotenoid in Bradyrhizobium

Aerobic photosynthetic bacteria possess the unusual characteristic of producing different classes of carotenoids. In this study, we demonstrate the presence of two distinct crt gene clusters involved in the synthesis of spirilloxanthin and canthaxanthin in a Bradyrhizobium strain. Each cluster contains the genes crtE , crtB and crtI leading to the common precursor lycopene. We show that spirilloxanthin is associated with the photosynthetic complexes while canthaxanthin protects the bacteria from oxidative stress. Only the spirilloxanthin crt genes are regulated by light via the control of a bacteriophytochrome. Despite this difference in regulation, the biosyntheses of both carotenoids are strongly interconnected at the level of the common precursors. Phylogenetic analysis suggests that the canthaxanthin crt gene cluster has been acquired by a lateral gene transfer. This acquisition may constitute a major selective advantage for this class of bacteria, able to photosynthesize only under condition where harmful reactive oxygen species are generated. Bacterial strains and growth conditions. Bradyrhizobium sp. strain ORS278 (wild type strain), and isogenic mutants were grown in a modified YM-agar medium with addition of appropriated antibiotic when required (24). All the strains were cultured for 7 days in sealed Petri dishes at 35°C in either complete darkness or different continuous illumination provided by light emitting diodes (LEDs) of different wavelength between 590 and 870 nm with an irradiance of 6.6 µmoles photons/m2/s. Escherichia coli was grown in Luria-Bertani (LB) medium supplemented with the appropriated antibiotics. spirilloxanthin standard. The data represent the mean of three independent cultures. The mutants harboring the various lacZ-crt fusions were grown under continuous illumination with low irradiance of different wavelengths as previously described (25). After growth, the cells under the illuminated area were re-suspended in 3 ml of water and ß-galactosidase activity was measured as previously described (24).


INTRODUCTION
Carotenoids comprise a large class of pigments which are widely distributed in living organisms. They are synthesized by all photosynthetic organisms from bacteria to plants where they play at least three essential functions (1). First, they act as accessory light-harvesting pigments by absorbing light in the 450-570 nm region. Second, they are important for the assembly and stability of some of these light-harvesting complexes. Finally, they operate as photo-protectors by directly quenching both triplet excited (bacterio)chlorophylls and singlet oxygen. Carotenoids are also synthesized by a wide variety of non-photosynthetic bacteria (2). Less is known about their precise function in these bacteria but it is well accepted that their strong antioxidant character may protect the organisms against (photo)oxidative damage.
A remarkable feature of aerobic phototrophic bacteria, besides their ability to photosynthesize only under aerobic condition, is their carotenoid composition. Indeed, most strains synthesize, in addition to the carotenoids involved in photosynthesis such as spirilloxanthin, a large amount of unusual carotenoid molecules (3). The most striking complexity is observed for Erythrobacter species such as E. longus or E. ramosum, which have been reported to produce about twenty different carotenoids (4, 5). Another example comes from various strains of photosynthetic Bradyrhizobia, symbionts of Aeschynomene (6), which synthesize, in addition to spirilloxanthin, large amounts of canthaxanthin of unknown function (7, 8).
All carotenoids are synthesized from geranylgeranyl pyrophosphate. This compound is formed by the enzyme geranylgeranyl pyrophosphate synthase (CrtE) which catalyzed the condensation of farnesyl pyrophospahate with an isopentyl pyrophosphate moiety. The second step catalyzed by phytoene synthase (CrtB) is the formation of phytoene from the head-tohead condensation of two molecules of geranylgeranyl pyrophosphate. Subsequent dehydrogenations catalyzed by the phytoene desaturase (CrtI) convert the phytoene to neurosporene in 3 desaturation steps or to lycopene in 4 steps. After the action of these three enzymes (CrtE, CrtB and CrtI), the biosynthetic pathways diverge depending on the species leading to the accumulation of various different carotenoids. The synthesis of canthaxanthin from lycopene necessitates two enzymes: CrtY which catalyses cyclisation of lycopene leading to ß-carotene and CrtW which oxygenates ß-carotene to form canthaxanthin ( Fig.   1A), (9). The sequence of the reactions from lycopene to spirilloxanthin includes the successive reactions of hydratation, desaturation and methylation catalyzed respectively by the three enzymes CrtC, CrtD and CrtF (Fig. 1A), (10,11). These reactions are performed first on one half of the molecule and then on the other half.
The genes encoding many carotenoid biosynthetic enzymes (crt genes) have been characterized in plants and in various bacteria (12,13). In bacteria, they are always found clustered, except in the cyanobacteria. In purple photosynthetic bacteria, the genes involved in carotenoid biosynthesis are localized within the photosynthesis gene cluster (PGC)-a 45 kb DNA region which contains the essential genes involved in the synthesis of the photosynthetic apparatus (14,15,16,17). In aerobic photosynthetic bacteria, a crt gene cluster has been characterized for the Bradyrhizobium ORS278 strain (18). This cluster contains the five crt genes, crtE, crtY, crtI, crtB and crtW, necessary for the canthaxanthin biosynthesis. Neither photosynthesis genes nor specific genes of spirilloxanthin biosynthesis from lycopene (crtC, crtD or crtF) have been identified in this cluster.
Light stimulation of carotenoid biosynthesis has been reported in numerous organisms including plants, fungi and bacteria (19). In higher plants, regulation of carotenoid biosynthesis occurs at the level of phytoene synthase expression (20). This expression is controlled by a phytochrome, a plant photoreceptor that mediates response to red and far-red light through photoconversion between two stable conformations, a red-absorbing form (Pr) and a far-red absorbing form (Pfr). Biochemical and genetic studies have recently demonstrated the occurrence of phytochrome-like proteins in photosynthetic and nonphotosynthetic bacteria (21,22). Such a (bacterio)phytochrome appears to control the synthesis of the carotenoid deinoxanthin in the non-photosynthetic bacteria Deinococus radiodurans (23).
In this report, we describe the characterization of a second crt gene cluster, in the Bradyrhizobium ORS278 strain, coding the enzymes of spirilloxanthin synthesis. This second crt gene cluster contains all the genes necessary for the synthesis of spirilloxanthin from farnesyl pyrophosphate. Biochemical analysis and phenotypes of mutants deleted in specific genes of canthaxanthin and spirilloxanthin synthesis allow us to establish the involvement of spirilloxanthin in the photosynthesis activity and the protective role of canthaxanthin in response of the bacteria to oxidative stress. We also demonstrate that the spirilloxanthin crt genes are specifically regulated by light via the control of a bacteriophytochrome. These results provide the first demonstration of two independent and differently regulated crt gene clusters in a living organism.

EXPERIMENTAL PROCEDURES
5 by guest on March 22, 2020 http://www.jbc.org/ Downloaded from Bacterial strains and growth conditions. Bradyrhizobium sp. strain ORS278 (wild type strain), and isogenic mutants were grown in a modified YM-agar medium with addition of appropriated antibiotic when required (24). All the strains were cultured for 7 days in sealed Petri dishes at 35°C in either complete darkness or different continuous illumination provided by light emitting diodes (LEDs) of different wavelength between 590 and 870 nm with an irradiance of 6.6 µmoles photons/m 2 /s. Escherichia coli was grown in Luria-Bertani (LB) medium supplemented with the appropriated antibiotics.

Pigments analysis.
Cells grown at the surface of the Petri dishes in the dark or under different light conditions were re-suspended in 6 ml of water + NaCl 9 g/l and spun down for 10 min at 4,000 g. The pellets were extracted 3 times in the dark with 1 ml of cold acetone/methanol (7/2, vol/vol).
The carotenoids in the pooled extracts were analysis by HPLC using an ALLIANCE Waters 2690 Channel. The conditions were: 5 µm Hypersil C 18 column (250 by 4.6 mm, Alltech France), acetonitrile/methanol/isopropanol (40/50/10, vol/vol/vol) as eluent, flow rate 0.8 ml/min. The eluted fractions were monitored using a Waters 996 photodiode array detector, scanning from 270 to 600 nm every 2s. Carotenoids were identified by their retention times and by comparison of the spectral features with those of pure compounds or with reported data. The amount of canthaxanthin was determined from the area of the peak detected at 480 nm using a calibration curve obtained with a canthaxanthin standard kindly provided by Aventis (France). The amount of spirilloxanthin was estimated from the area of the peak detected at 494 nm using the canthaxanthin correlation coefficient due to the lack of spirilloxanthin standard. The data represent the mean of three independent cultures.

Light action spectrum on crt genes expression
The mutants harboring the various lacZ-crt fusions were grown under continuous illumination with low irradiance of different wavelengths as previously described (25). After growth, the cells under the illuminated area were re-suspended in 3 ml of water and ßgalactosidase activity was measured as previously described (24).

Construction of crt mutant strains.
Constructions of mutants deleted in the bacteriophytochrome (278BrbphP) and in the transcriptional factor PpsR (278ppsR) have been previously described (25 derivatives obtained, which encoded a counter selective sacB marker, were transformed into E. coli S17-1 for mobilization into ORS278 as previously described (24). Double recombinants were selected on sucrose and the insertion was confirmed by PCR.

Preparation of membranes and RC-LH1 complexes. A 200 ml culture of Bradyrhizobium
(WT or mutant) was collected and resuspended in 10 ml of Tris-HCl buffer (50 mM, pH 8).
The cells were disrupted by three passages through a French Press at 50 MPa. The suspension was centrifuged for 10 min at 4,000 g to remove the unbroken cells and cells debris. The supernatant was loaded on a discontinuous sucrose gradient (0.6-1.2 M sucrose; Tris-HCl buffer 50 mM, pH 8) and centrifuged at 255,000 g for 90 min. The membranes localized at the interface of the two sucrose layers constitute the chromatophores fraction while the pellet contains the cytoplasmic membranes. Each fraction is diluted in 25 ml Tris-HCl buffer (50 mM, pH 8), spun down at 255,000 g for 90 min to removed the sucrose and then resuspended in Tris-HCl buffer (10 mM, pH 8).
RC-LH1 complexes were isolated by addition of 1.5% LDAO to purified chromatophores or cytoplasmic membranes whose optical density was adjusted to 5 OD/cm at 870 nm. After an incubation of 15 min at room temperature in the dark, the membrane suspension was loaded on discontinuous sucrose gradient ( particles were performed using a mono-Q column coupled to a FPLC (Pharmacia) and submitted to a NaCl gradient.

Absorption and fluorescence spectroscopy
Absorption spectra and light-induced absorption changes in intact cells were measured as previously described (24). Fluorescence excitation and emission spectra were recorded on a Spex Fluorolog 3 spectrofluorimeter (Jobin Yvon). For excitation spectra, the excitation slits were 5 nm and the emission was measured at 870 nm (on the blue side of the emission spectrum where the instrument sensitivity was highest) with 15 nm slits. For emission spectra excitation was through 10 nm slits and emission measured through 7 nm slits. Emission spectra were corrected for the wavelength dependence of the instrument response and excitation spectra were corrected for variations in excitation intensity. For all fluorescence spectra the detector was protected from scattered excitation by a Wratten 88A gelatin filter.

Isolation and characterization of the spirilloxanthin biosynthesis genes.
In all photosynthetic bacteria studied so far, the carotenoid biosynthesis genes have been found linked to the photosynthesis gene cluster (PGC). We have previously isolated, from a genomic DNA library of the ORS278 strain, a cosmid (pSTM1)  Interestingly the genes crtE, crtI, and crtB that encode enzymes of lycopene biosynthesis, a

Regulation of canthaxanthin and spirilloxanthin synthesis by light.
We have previously shown that the photosynthetic activity in Bradyrhizobium is stimulated DNase I digestion by addition of PpsR. We therefore conclude that only the expression of the spirilloxanthin crt genes are under the control of PpsR and BrbphP. There is a clear discrepancy between this conclusion and the observed enhancement of the production of canthaxanthin by far-red light (Fig. 5A) and the carotenoid content of the BrbphP and ppsR null mutants (Fig. 5C). To resolve this apparent contradiction, we hypothesize that the up regulation of spirilloxanthin crt genes by the phytochrome leads to the over production of some common precursors to the biosynthesis pathway of canthaxanthin and spirilloxanthin.
For example lycopene produced by the CrtE.s, CrtB.s and CrtI.s enzymes could be partially re-routed by the CrtY and CrtW enzymes to form canthaxanthin. To test this hypothesis we have measured the carotenoid content in mutants deleted in one of the genes of the enzymes of lycopene synthesis in each carotenoid pathway (278crtE.c and 278crtI.s). These two mutants produce both canthaxanthin and spirilloxanthin demonstrating a crosstalk between the two carotenoid synthesis pathways ( Fig. 5E and F). Other proofs of this link are the lower level of canthaxanthin produced by the 278crtI.s mutant (Fig. 5F) and the strong enhancement by far-red light of canthaxanthin production, especially in the 278crtCD mutant. These different observations imply that crtE.s, crtI.s or crtB.s genes can complement respectively for the deletion of the crtE.c, crtI.c or crtB.c genes and vice-versa. We therefore conclude that the biosynthesis of spirilloxanthin and canthaxanthin are strongly connected at the level of lycopene or one of its precursors.

DISCUSSION
Aerobic phototrophic bacteria present a surprising complex carotenoid composition. In the present study, our goal was to clarify this complexity using the bacterial model Bradyrhizobium ORS278, which synthesizes two major carotenoids, the linear spirilloxanthin and the bicyclic canthaxanthin. Combining biophysical, biochemical and genetic approaches, we determined the function of each of these carotenoids, characterized their biosynthesis genes and described their regulation. We demonstrate the presence of two distinct carotenoids gene clusters, involved in the biosynthesis of spirilloxanthin and of canthaxanthin, respectively. One striking result is the presence in each of these clusters of the three genes, crtE, crtB, crtI, implicated in the synthesis of the precursor lycopene, common to the two biosynthetic pathways. Altogether these results, discussed in more detail below, give the first indications to understanding of the presence of different carotenoids in a photosynthetic bacterium.

Functions of spirilloxanthin and canthaxanthin
Biochemical analysis and phenotypes of mutants clearly demonstrate that the spirilloxanthin molecules are the only carotenoid associated with the photosynthetic apparatus in Bradyrhizobium ORS278. Comparison between the excitation spectrum and the absorption spectrum leads to the conclusion that only about 30% of the light energy absorbed by the spirilloxanthin molecules are transferred to the bacteriochlorophyll molecules. This efficiency is typical to spirilloxanthin molecules associated with LH1, as already observed in the case of Rhodospirillum rubrum for example (31). When the synthesis of spirilloxanthin is blocked at the CrtC and CrtD enzymes level, spirilloxanthin is replaced by lycopene in the LH1 complexes and probably in the RC. Lycopene has also been shown recently to be an integral part of the LH2 complexes in a mutant of Rhodobacter sphaeroides in which the native 3-step phytoene desaturase (CrtI) was replaced with the 4-step enzyme from Erwinia herbicola (32).
However, to our knowledge, this is the first example of the presence of lycopene in LH1 complexes. Although the energy transfer between lycopene and bacteriochlorophylls is less efficient than the one measured for spirilloxanthin, lycopene and spirilloxanthin present a similar arrangement and conformational state as shown by linear dichroism measurements on intact cells oriented in polyacrylamide gels (data not shown). Furthermore, the amount of photosynthetic apparatus per cell is reduced by a factor of 3 to 5 in the 278crtCD mutant.
Therefore, our results not only demonstrate that the spirilloxanthin is the major carotenoid associated with the photosystem in Bradyrhizobium but also that this molecule plays an important role in the structural stabilization of the bacteriochlorophyll LH1 complexes. In contrast, canthaxanthin appears to be localized predominantly in the cytoplasmic part of the membrane and to protect the cells against oxidative stress in agreement with different in vitro studies (33, 34).

Regulation of carotenoids biosynthesis
In this study, we clearly show that the canthaxanthin and spirilloxanthin crt gene The molecular mechanism by which BrbphP activates the expression of spirilloxanthin crt genes and anti-represses the action of PpsR remains unknown. This mechanism may be close to the dual mechanism of action between AppA and PpsR recently described in R.
sphaeroides (35). Like BrbphP, Appa is a light photoreceptor, which activates the expression of photosynthesis genes including crt genes. It has been shown that Appa antagonizes the repressive effect of PpsR by forming a blue light sensitive and redox dependent AppA-PpsR complex. By analogy, one can speculate BrbphP and PpsR form a light-dependent complex.
Although the canthaxanthin crt genes cluster is not regulated via BrbphP and PpsR, we observe an increase in production of canthaxanthin caused by far red light in the WT strain.
We propose that a part of the common intermediates synthesized by the CrtE.s, CrtI.s, and CrtB.s enzymes could be diverted towards canthaxanthin production. The production of both canthaxanthin and spirilloxanthin by the crtE.c and crtI.s minus mutants is in agreement with such a hypothesis. Despite this interconnection between the two biosynthesis pathways, the amount of spirilloxanthin does not exceed 0.2 mg/g of dry cells even in the mutant (

Origin of carotenoid gene clusters
A recent analysis of the diversity of photosynthetic Bradyrhizobia shows a monophyletic origin of the strains producing canthaxanthin (38). It is therefore very tempting to suggest that an ancestral photosynthetic Bradyrhizobium acquired, by lateral gene transfer, the canthaxanthin crt gene cluster. Phylogenetic analysis based on CrtI (Fig. 6) or CrtB sequences (data not shown) provide strong support for this hypothesis. Indeed, the trees obtained clearly show the presence of two distinct groups which are not related to the taxonomical position of the various species but to the function or the nature of their carotenoid (cyclic or non cyclic).   C. HPLC separation of the carotenoids of WT strain and the mutants 278crtCD and 278crtY.
Peaks correspond to : 1, trans-canthaxanthin; 2, cis-canthaxanthin isomers; 3, spirilloxanthin.  Exponentially growing cells (WT, 278crtCD, 278crtY) were diluted and plated onto YM agar medium containing various concentration of Methyl Viologen. Plates were incubated at 37°C and, after one week, the CFU were counted. F. Production of canthaxanthin in WT strain (ORS278) and the 278crtE.c, 278crtY, 278crtI.s, and 278crtCD mutants after growth under complete darkness or far-red light (770 nm). This tree has been constructed by using Neighbour-joining method (40