Diverse Arrangement of Photosynthetic Gene Clusters in Aerobic Anoxygenic Phototrophic Bacteria

Background Aerobic anoxygenic photototrophic (AAP) bacteria represent an important group of marine microorganisms inhabiting the euphotic zone of the ocean. They harvest light using bacteriochlorophyll (BChl) a and are thought to be important players in carbon cycling in the ocean. Methodology/Principal Findings Aerobic anoxygenic phototrophic (AAP) bacteria represent an important part of marine microbial communities. Their photosynthetic apparatus is encoded by a number of genes organized in a so-called photosynthetic gene cluster (PGC). In this study, the organization of PGCs was analyzed in ten AAP species belonging to the orders Rhodobacterales, Sphingomonadales and the NOR5/OM60 clade. Sphingomonadales contained comparatively smaller PGCs with an approximately size of 39 kb whereas the average size of PGCs in Rhodobacterales and NOR5/OM60 clade was about 45 kb. The distribution of four arrangements, based on the permutation and combination of the two conserved regions bchFNBHLM-LhaA-puhABC and crtF-bchCXYZ, does not correspond to the phylogenetic affiliation of individual AAP bacterial species. While PGCs of all analyzed species contained the same set of genes for bacteriochlorophyll synthesis and assembly of photosynthetic centers, they differed largely in the carotenoid biosynthetic genes. Spheroidenone, spirilloxanthin, and zeaxanthin biosynthetic pathways were found in each clade respectively. All of the carotenoid biosynthetic genes were found in the PGCs of Rhodobacterales, however Sphingomonadales and NOR5/OM60 strains contained some of the carotenoid biosynthetic pathway genes outside of the PGC. Conclusions/Significance Our investigations shed light on the evolution and functional implications in PGCs of marine aerobic anoxygenic phototrophs, and support the notion that AAP are a heterogenous physiological group phylogenetically scattered among Proteobacteria.


Introduction
Aerobic anoxygenic photototrophic (AAP) bacteria represent an important group of marine microorganisms inhabiting the euphotic zone of the ocean. They harvest light using bacteriochlorophyll (BChl) a and various carotenoids serving as auxiliary pigments. These phototrophic microorganisms are thought to be important players in oceanic carbon cycling [1][2][3]. Cultureindependent studies have shown that marine AAP bacterial communities are mostly represented by Alphaand Gammaproteobacteria [4,5]. Most cultured marine Alphaproteobacterial AAPs belong to Roseobacter clade and the order Sphingomonadales, which includes members of the genera Erythrobacter and Citromicrobium [6][7][8]. AAP bacterial isolates related to Gammaproteobacteria belong to the clade NOR5/OM60 which contains Congregibacter litoralis KT71 [9,10] and strain HTCC2080 [11].
Compared to the oxygenic phototrophs, the anoxygenic species contain a relatively simple photosynthetic apparatus, which consists of a reaction center surrounded by one to three types of antenna complexes [12]. Both aerobic and anaerobic anoxygenic phototrophs have most of the photosynthetic genes organized in a so-called photosynthesis gene cluster (PGC) [13]. The PGC contains genes for the photosynthetic reaction center, light harvesting complexes, BChl and carotenoid biosynthesis, as well as some regulatory factors. Despite the fact that the basic set of genes in PGC is conserved, the gene organization of operons in PGC largely varies among different AAP bacterial lineages. Two conserved subclusters, crt-bchCXYZ-puf (about 10 kb) and bchFNBHLM-IhaA-puh (about 12-15 kb) were identified in PGCs of different phototrophic Proteobacteria [14][15][16]. The orientation of the genes in each subcluster was the same, although the gene order could vary slightly (e.g. pufBA and pufLM). Interestingly, the regulatory elements such as the transcriptional regulator ppsR gene were conserved as well, suggesting that the operons in the PGCs are co-expressed. The organization of puf (photosynthetic unit forming, approximately 3 kb) operon varies among different AAP bacterial species. The presence/absence of pufC and pufQ, as well as various gene orders of puf genes, were observed [5,[15][16][17]. Further investigation indicated that such gene organization is crucial for environmental adaptation [14]. Figure 1. Phylogenetic analysis of pufM gene sequences from GenBank database. Symbols ''w'' represents the pufM sequences from whole genome sequence. The whole PGC's of the ten strains highlighted in boxes were also analyzed (Fig. 2). Bootstrap percentages from both neighbor joining (above nodes) and maximum parsimony (below nodes) are shown. Scale bar represents 10% nucleotide substitution percentage. doi:10.1371/journal.pone.0025050.g001 Despite its diversity, complexity and functional importance for AAP bacteria, a detailed investigation of the gene and operon arrangement of PGC has not been performed in their entirety. In this study, we analyzed the structure and arrangement of PGC in the AAP bacterial genomes available to date, with the aim of addressing the frequency of homologous gene recombination as well as the differences in carotenoid gene composition and biosynthetic pathways.

The structure and arrangement of PGC
The PGCs have a mosaic structure and consist of five main sets of genes: bch genes encoding enzymes of BChla biosynthetic pathways, puf operons encoding proteins forming the reaction centers, puh operons involved in the RC assembly, crt genes responsible for biosynthesis of carotenoids and various regulatory genes. A core set of 27 genes were identified, which were present in all analyzed PGCs (Fig. S1). Most of them came from the BChl a biosynthetic pathway. The genes bchBCDFGHILMNOPXYZ and ascF, with exception of 8-vinyl reductase, represent the complete biosynthetic pathway from protoporphyrin XI to BChl a. In contrast, there are only two genes involved in carotenoid synthesis which are common for all PGCs. Other shared core genes encode proteins pufABLM and assembly factors puhABCE and lhaA of the bacterial photosynthetic units.
More complete PGC structures are observed in AAP of Roseobacter clade compared to Sphingomonadales or NOR5/ OM60 clades. The majority of Roseobacter-related species contained all the puf genes organized in pufQBALMC operon, which is involved in the assembly of the photosynthetic units. The only exception was L. vestfoldensis SKA53, in which some photosynthetic genes are located outside the PGC and spread throughout the genome. Previously it was reported that the PGC in Rsb. litoralis Och 149 is located on a linear plasmid, with two RPA genes between bchFNBHLM-LhaA-puh and crtF-bchCXYZ-puf, which act as a centromere-like anchor when plasmids replicate [19,23].
The PGC organization in Erythrobacter sp. NAP1 and Citromicrobium sp. JL354 (order Sphingomonadales) is almost identical in terms of gene arrangement and composition. When compared to Roseobacters, this group contains less carotenoid genes and no light-harvesting 2 (LH2) genes. The presence of a smaller number of photosynthetic genes in Sphingomonadales is consistent with the smaller size of their PGCs (Table 1).
Similarly, PGCs of two NOR5/OM60 strains have very comparable gene composition and organization. It contains less transcriptional regulators compared to the other groups. Conversely, a BLUF (blue light using flavin adenine dinucleotide sensors) was usually observed in upstream regions of PGCs of NOR5/OM60 clade [9].
Two conserved gene arrangements are found in all analyzed PGCs: bchFNBHLM-LhaA-puhABC and crtF-bchCXYZ (Fig. 2). According to their direction and order, the ten PGCs can be divided into three groups: Type I (forward bchFNBHLM-LhaA-puh plus forward crtF-bchCXYZ-puf) includes Rsb. denitrificans OCh 114 and Rsb. litoralis Och 149. Cb. litoralis KT71, Gammaproteobacterium HTCC2080 and Roseovarius sp. 217 belong to type II (forward bchFNBHLM-LhaA-puh plus reverse crtF-bchCXYZ-puf), and the last five organisms form type III (forward crtF-bchCXYZ-puf plus forward bchFNBHLM-LhaA-puh). The last possible arrangement (type IV, reverse bchFNBHLM-LhaA-puh plus forward crtF-bchCXYZ-puf) has not been yet found in AAP (or AAP candidates) bacterial genomes (Fig. S1), however it is present in the purple non-sulfur anaerobic bacteria Rba. sphaeroides and Rba. capsulatus (Fig. 2 and Fig. S2). The distribution of PGC types does not correspond to their phylogenetic affiliation. For example, the Roseobacter clade shows all three PGC arrangement types observed in AAP genomes. This suggests that complex operon recombination in PGC occurred after phylogenetic divergence of AAP bacterial genera.
There are four conserved regions in PGCs for BChl a expressing of AAP bacteria: bchFNBHLM, bchCXYZ, bchIDO and bchOP. Gene bchEJ, which exists in most Rhodobacter, was found in Cb. litoralis KT71 (Fig. 2). There are carotenoid genes between bchCXYZ and bchIDO, except in D. shibae DLF 12 and Jannaschia sp. CCS1. The region between bchOP and bchFNBHLM is of variable sequences in different AAP bacteria clades. In Roseobacter clade and Sphingomonadales, there are two regulators (ppsR and ppaA) which are sensitive to light intensity and oxygen concentration [24]. In NOR5/OM60 clade, a crtJ gene was found, which controls aerobic repression of BChl, carotenoid, and LH2 gene expression [25,26].
Four structural types of puf gene organization were observed in the ten PGCs: pufQBALMC, pufQBALM, pufBALM and pufLMCBA. Unlike the purple non-sulfur species Rba. sphaeroides and Rba. capsulatus, all the AAP strains studied lack the pufX gene in the PGC. The pufQ gene, is absent in the puf operon of NOR5/OM60 and Sphingomonadales clades. In addition, Sphingomonadales and L. vestfoldensis SKA53 do not have a pufC gene. The gene encoding 1deoxy-D-xylulose-5-phosphate synthase (DXPS) is always located downstream of puf genes in the Roseobacter clade. DXPS is part of a mevalonate-independent pathway for isopentenylpyrophosphate (iPP) biosynthesis, a precursor for carotenoid and bacteriochlorophyll biosynthesis [27]. Interestingly, a switch of order in the puf gene cluster is observed in NOR5/OM60 clade (pufLMC-BA) compared to the other two AAP clades (pufBA-LMC).

The composition and organization of carotenoid genes in PGC
The main difference among analyzed PGCs was found in the genes encoding the carotenoid biosynthetic pathway. The standard set of crt genes identified in Rba. capsulatus contains crtAIBKCDEFJ ( Table 2). A slightly reduced set of genes (crtAIBCDEF) was also found in some Roseobacter species (Table 2). However, the organization of the crt operon in Roseobacter clade is most variable among PGCs (Fig. 2). The almost complete structure crtAIBK-hyp-crtCDEF is present in the genera Roseobacter and Dinoroseobacter (Fig. 2), while in D. shibae, crtA and crtIBK are separated. Homologous recombination occurred between crtAIB  and crtCDEF in Jannaschia sp. CCS1. Comparably, crtICDEF and crtCDF are missing in NOR5/OM60 clade and order Sphingomonadales, respectively. The re-arrangement of crt genes may result from events of gene duplication and loss, accounting for the absence of crtA gene in Sphingomonadales and NOR5 clade (Table 2), and duplication of some of the crt genes, such as the crtE and crtIB found outside the PGC in Bradyrhizobium sp. ORS278 [31].

The biosynthetic pathway for carotenoids in AAP bacterial strains
A typical feature of AAP bacteria is their pigmentation due to abundant carotenoids, which spans from yellow/orange to brown or from pink/red to purple. While some of the carotenoids serve as harvesting pigments, most of them do not participate in the light harvesting likely having a photoprotection function [32,33]. As suggested earlier, spheroidenone is the main light harvesting carotenoid in Roseobacters [34][35][36] (Table S1). Spheroidenone is also produced by anaerobic purple non-sulfur photoautotrophic organisms such as Rba. sphaeroides or Rhodovulum. marinum when grown under aerobic conditions [37,38]. This is consistent with the closer phylogenetic relationship of these two organisms to Roseobacter related photoheterotrophic species (Fig. 1 and Fig.  S2). This indicates the presence of the same carotenoid biosynthetic pathway in all Rhodobacterales. The central biosynthetic pathway for carotenoids in the Roseobacter clade is the spheroidene pathway (Fig. 3, Table S1), and all the necessary genes (crtAIBCDF) for it are located in the PGCs (Fig. 2 and Table 2).
In most studied Erythrobacter species, erythroxanthin sulfate was shown to be the main carotenoid [7,39] (Table S1), however, it does not participate in the photosynthetic processes [39]. Light is harvested by other pigments such as bacteriorubixanthinal, zeaxanthin and b-carotene [39]. The main carotenoid identified in Citromicrobium sp. JL354 was nostoxanthin (Table S1). We assume that both species share similar carotenoid biosynthetic pathways (Fig. 3). First, b-carotene is produced from lycopene by the action of lycopene cyclase (crtY gene product). Zeaxanthin is obtained by two step hydroxylation of b-carotene catalyzed by b-carotene hydroxylase (crtZ gene product). Interestingly, the key genes (crtY and crtZ) for zeaxanthin pathway are not organized in the PGCs, but are spread throughout the chromosome (Table 2). Zeaxanthin is then a starting intermediade for synthesis of both nostoxanthin (in genus Citromicrobium) and erythroxanthin (in genus Erythrobacter) (Fig. 3).
The major carotenoids in Cb. litoralis KT71 is spirilloxanthin, the same as in Rhodospirillum rubrum DSM 467 T [10] (Table S1). There are two possible options for spirilloxanthin biosynthesis: typical-spirilloxanthin biosynthestic pathway and unusual-spirilloxanthin pathway (Fig. 3). Interestingly, the gene crtD was found to be out of PGC in Cb. litoralis KT71 ( Table 2), indicating that Cb. litoralis KT71 might use the shorter unusual-spirilloxanthin pathway.
In summary, this study showed that most of the photosynthetic genes in AAP species were organized in the PGC. Two conserved regions bchFNBHLM-LhaA-puhABC and crtF-bchCXYZ, were identified in all studied PGC. Based on their orientation we can divide the studied strains into four different groups. The composition of bch, puf and puh genes in the analyzed PGCs was relatively similar, and the main difference was found among crt genes. Such variability was mainly connected with different carotenoid biosynthetic pathways present in AAP groups: spheroidenone biosynthetic pathway in Roseobacters, zeaxanthin pathway in Sphingomonadales and spirilloxanthin pathway in gammaproteobacterial NOR5/OM60 clade. Our investigation shed light on the evolution and functional implications of PGCs of marine aerobic anoxygenic phototrophs.
For comparison two anaerobic anoxygenic phototroph Rhodobacter sphaeroides strain 2.4.1 (NC_007493) and Rhodobacter capsulatus SB 1003 (NC_014034) also were included in the analysis. Another three green sulfur bacteria genome information used to outgroup of phylogenetic tree, and their Genbank accession numbers are Chloroflexus aggregans DSM 9485 (NC_011831), Chloroflexus aurantiacus J-10-fl (NC_010175) and Roseiflexus castenholzii DSM 13941 (NC_009767). In some cases the automatic gene annotation was corrected manually.
Nearly complete pufM (.900 bps) genes and 27 core proteins in PGCs were used to construct phylogenetic trees [17]. Both pufM gene sequences collected from NCBI database were aligned using Clustal X and phylogenetic trees were constructed using the neighbour-joining and maximum-parsimony algorithms of MEGA software 3.0 [40]. The phylogenetic trees were supported by bootstrap for resampling test with 1000 replicates.