**4. Discussion**

A photosynthetic lifestyle allows organisms to use light as an energy source for growth and proliferation. However, this benefit comes at a price, that is, the risk of 1O2 generation by energy transfer from (bacterio-) chlorophyll to molecular oxygen (3O2) within photosynthetic complexes. It is suspected that purple bacteria from the genus *Rhodobacter* address this problem by avoiding strong pigmentation under high light and/or high oxygen conditions, a response that is mainly regulated by light- and oxygen-sensing proteins [2,3]. Despite regulation of pigmentation in response to oxygen tension and light, *Rhodobacter* species cannot completely avoid 1O2 stress in their natural aquatic environments. They have therefore evolved strategies to counteract 1O2 and to deal with the resulting damages [1,65]. *R. sphaeroides* is a well-studied model organism with regard to the photooxidative stress response [70], and was compared here to *R. capsulatus* using transcriptomics and proteomics. Intriguingly, the two *Rhodobacter* species elicit very similar responses in the light of functional categories. Many proteins with an increased abundance upon 1O2 exposure are involved in oxidation–reduction processes and DNA damage repair (Figure 8). Furthermore, several methionine sulfoxide reductases (MsrA and MsrB orthologs) and a variety of proteases were increased, which was also confirmed on the transcript level (Figure 7). Since proteins are the main targets of 1O2 [15], we conclude that the repair of proteins and the removal of damaged proteins is essential to survive this particular stress, and that efficient protein maintenance likely represents a key feature of the response to 1O2 in many organisms. In *R. sphaeroides*, the 1O2 stress response is mainly controlled by the alternative sigma factors RpoE and RpoHII [10,18–21]; we have reason to believe that this also holds true for *R. capsulatus*. Firstly, both *rpoE* and *rpoHII* have elevated transcript levels upon 1O2 exposure (Table 1), and secondly, an *rpoHII* deletion strain is more sensitive to 1O2 than the *R. capsulatus* wild type (Figure 6). Hence, the basic regulatory principles of the photooxidative stress response might be very similar in both *Rhodobacter* species.

Despite the aforementioned similarities, there are remarkable di fferences between *R. sphaeroides* and *R. capsulatus*. Even though both species fall into the same clade within a phylogenetic tree based on a core genome of 580 orthologous proteins [71], synteny analysis revealed that the genetic organization of orthologous genes on the chromosome is considerably di fferent, with the exception of, e.g., the photosynthetic gene cluster (Figure S8). This pronounced genomic rearrangemen<sup>t</sup> is not observed between di fferent *R. sphaeroides* strains (e.g., strains 2.4.1, KD131, and WS8N), but between di fferent genera within the *Rhodobacteraceae* (e.g., between *Rhodobacter* and *Roseobacter*). Hence, our analyses sugges<sup>t</sup> that *R. capsulatus* and *R. sphaeroides* are, from a genomic point of view, more distantly related than expected from their common lifestyle and the arrangemen<sup>t</sup> of the photosynthetic gene cluster [71]. Another remarkable finding concerns the genetic context of the *rpoE-chrR* locus in *R. capsulatus*, which is in close proximity to the photosynthetic gene cluster (Figure 4). This genetic arrangemen<sup>t</sup> was not found in other *Rhodobacteraceae* species, and can therefore be considered as unique. It is worth noting that RpoE from *R. capsulatus* shares higher identity with the extra-cytoplasmic function sigma factor SigK from *Mycobacterium tuberculosis* (34%), compared to only 24% identity with *R. sphaeroides* RpoE. The same is true for *R. capsulatus* ChrR, which shares 25% identity with *M. tuberculosis* SigK antisigma factor RskA, but only 13% with *R. sphaeroides* ChrR. Phylogenetic trees further support the special position of the *R. capsulatus* RpoE and ChrR proteins within the *Rhodobacteraceae*family (data not shown). These findings raise the question of whether the *rpoE-chrR* locus in *R. capsulatus* and *R. sphaeroides*

originated from a common ancestor and was then intensely remodeled in *R. capsulatus*, or whether *R. capsulatus* has received a *sigK-rskA*-like locus from another bacterial lineage (e.g., Gram-positives like *M. tuberculosis*) via horizontal gene transfer.

*R. capsulatus* not only displays a unique colocalization of the photosynthetic gene cluster near the *rpoE-chrR* locus; it also differs to *R. sphaeroides* on the physiological level with regard to its pigmentation. a higher pigmentation of *R. capsulatus* under high oxygen conditions obviously allows much faster adaptation to occur to phototrophic conditions compared to *R. sphaeroides* (Figure 1B). Despite this higher pigmentation, *R. capsulatus* has no disadvantage when exposed to sudden photooxidative stress (Figure 1C). Low expression of BChl *a* and carotenoid biosynthesis genes under aerobic conditions is reflected by low BChl *a* and carotenoid levels in both *R. capsulatus* and *R. sphaeroides* (Figure 2A). Upon photooxidative stress, Bchl *a* and carotenoid levels decreased in *R. sphaeroides*. By contrast, in *R. capsulatus,* only BChl *a* levels declined, while carotenoid levels remained fairly constant (Figure 2B), resulting in an increasing Crt:BChl *a* ratio in the course of 1O2 exposure (Figure 2D). These differences in Crt:BChl *a* ratio may be the reason for lower 1O2 levels in *R. capsulatus* under photooxodative stress considering the 1O2 quenching ability of carotenoids (Figure 3A). Increased carotenoid biosynthesis under conditions of photooxidative stress is a strategy that is used by some microorganisms, including the deltaproteobacterium *Myxococcus xanthus* and the yeas<sup>t</sup> *Pha*ffi*a rhodozyma*, in order to avoid extensive cellular damages by direct quenching [72,73]. Accumulation of carotenoids upon 1O2 exposure was, however, not observed in *R. sphaeroides* [27], and omics data even revealed declining transcript and protein levels of genes from the photosynthetic gene cluster, including genes for carotenoid biosynthesis [28]. Interestingly, RNA-seq revealed that three genes within the photosynthetic gene cluster had increased transcript levels in *R. capsulatus*, which applies to *crtI* (log2 fold change of ~2.3), *crtB* (log2 fold change of ~2.0), and *tspO* (log2 fold change of ~1.3). Induction of *crtI* was validated by qRT-PCR (Figure 5). The three genes form an operon between the carotenoid biosynthesis genes *crtA* and *crtC*. TspO is an outer membrane protein, which controls efflux of porphyrin intermediates, and thereby negatively modulates expression of photosynthesis genes, likely through AppA [74]. The increased expression of *tspO* might enhance porphyrin efflux under photooxidative stress conditions, limiting both expression of photosynthesis genes and accumulation of potential porphyrin-derived photosensitizers, which may support adaptation to this particular stress. CrtB is a phytoene synthase and CrtI is a phytoene desaturase, which catalyze the reaction of precursors to phytoene (CrtB) and from phytoene to zeta-carotene and neurosporene (CrtI). The carotenoids spheroidene (SE) and spheroidenone (SO) are then synthesized in subsequent steps from neurosporene. Even though a switch from SE to SO (catalyzed by CrtA) might be an important adaption to photooxidative stress, we could not detect a strong increase for *crtA* mRNA in the RNA-seq data. It is conceivable that enhanced CrtB and CrtI levels are needed to provide sufficient amounts of neurosporene as a precursor for SE and SO biosynthesis to maintain carotenoid levels in *R. capsulatus* (Figure 2B). By contrast, in *R. sphaeroides crtI* and *crtB* transcript levels do not increase upon photooxidative stress (Figure 5) [28], which even coincides with a decrease in carotenoid levels (Figure 2B). Obviously, both *Rhodobacter* species use different strategies to adapt to high light regimes. Since *R. capsulatus* initially grows better than *R. sphaeroides* both after a shift to phototrophic conditions (Figure 1B) and upon photooxidative stress (Figure 1C), elevated carotenoid levels are expected to be advantageous to photosynthetic bacteria in rapidly changing environments.

Besides the protective function of carotenoids, *cbiX* might also play a more crucial role in defending against photooxidative stress in *R. capsulatus* than in *R. sphaeroides*. Our omics data show that *cbiX* is strongly increased on both transcript and protein level in *R. capsulatus* (log2 fold change of ~3). The RSP\_1566 protein of *R. sphaeroides* shows 47% identity to CbiX, but neither protein nor transcript levels change significantly in response to 1O2 [28], and *cbiX* transcripts were not even detected by qRT-PCR (Figure 5). The *cbiX* gene is annotated as a cobaltochelatase which incorporates cobalt into sirohydrochlorin to form cosirohydrochlorin, an early precursor of vitamin B12 [75]. B12 has special functions in the formation of the photosynthetic apparatus in *Rhodobacter*, i.e., it is required for the

conversion of protoporphyrin IX to Mg-protoporphyrin monomethyl ester [76]. Furthermore, it is needed by the antirepressor AerR to efficiently bind to repressor CrtJ, thereby inducing *bch* gene expression in *R. capsulatus* [77,78]. Theoretically, increased CbiX levels favor B12 biosynthesis and, consequently, derepression of *bch* genes via AerR. However, increased *bch* expression would then increase photooxidative stress, but this was not observed. a structural homology search for *R. capsulatus* CbiX using Phyre<sup>2</sup> suggests that it might rather act as a ferrochelatase involved in the biosynthesis of siroheme. This assumption is supported by a study by Bali and colleagues [69], showing that CbiX functionally replaces characteristic siroheme biosynthesis enzymes, which are missing in alphaproteobacteria, including *R. sphaeroides*. As siroheme is a cofactor of nitrite and sulfite reductases, an increased siroheme production might also explain the high accumulation of the mRNA for sulfite reductase CysI (log2 fold change of ~3.2). Increased sulfur assimilation might help to counteract the 1O2-caused depletion of glutathione and abundant damages on sulfur-containing amino acids. More importantly, however, the insertion of Fe2<sup>+</sup> into sirohydrochlorin reduces the concentration of free iron, which would otherwise be available for both the Fenton reaction and the formation of the photosensitizer protoporphyrin IX. Hence, additional stress is prevented by changing the flux through tetrapyrrole pathways [79–81]. As a conclusion, the induction of both the *crtIB-tspO* operon and *cbiX* might represent a successful strategy to respond to photooxidative stress, which was specifically invented in *R. capsulatus* to support adaptation. The question of whether other phototrophs have evolved similar strategies will be an exciting subject for future studies.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2607/8/2/283/s1, Figure S1: Analysis of *Rhodobacter* transposon mutants (*R.c.* Tn5 and *R.s.* Tn5) lacking important carotenoids, Figure S2: Growth of *R. capsulatus* and *R. sphaeroides* with and without methylene blue in the dark or under high light conditions, Figure S3: Pigmentation of *R. capsulatus* and *R. sphaeroides* under microaerobic conditions, Figure S4: SOSG fluorescence in cell-free reactions, Figure S5: Correlation analysis of RNA-seq replicates, Figure S6: PCA of RNA-seq and LC-MS/MS replicate samples, Figure S7: Correlation analysis of LC-MS/MS replicates, Figure S8: Synteny plot comparing *Rhodobacter* genomes, Table S1: Strains used in this study, Table S2: Transcriptome data for *Rhodobacter* upon photooxidative stress, Table S3: Primers used for qRT-PCR in this study, Table S4: LC-MS/MS data for *R. capsulatus* upon photooxidative stress.

**Author Contributions:** Conceptualization, G.K.; methodology, M.K.L., A.M.N., A.K., M.B., B.A.B. and G.K.; software, M.B.; validation, M.K.L.; formal analysis, M.K.L., A.K., M.B. and B.A.B.; investigation, M.K.L., A.M.N., M.V., A.K.; data curation, M.K.L., A.K. and M.B.; writing—original draft preparation, M.K.L., B.A.B. and G.K.; writing—review and editing, M.K.L., B.A.B. and G.K.; visualization, M.K.L. and B.A.B.; supervision, B.A.B. and G.K.; project administration, G.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank Petra Dersch (University of Münster, Germany) for generous support with RNA-seq and corrections on the manuscript, and Andrew Lang (Memorial University of Newfoundland, Canada) for providing strains.

**Conflicts of Interest:** The authors declare no conflict of interest.
