**3. Results**

#### *3.1. Adaptation of R. sphaeroides and R. capsulatus to Di*ff*erent Growth Conditions*

Based on the di fferent pigment content of the two *Rhodobacter* species, we hypothesized di fferences in their adaptation to phototrophic growth. When cultures of the two species were kept under microaerobic conditions in the dark, the growth behavior was nearly identical (doubling time td of 4 h ± 3 min for *R. capsulatus* and 4 h 10 min ± 15 min for *R. sphaeroides*, Figure 1A). a shift from high oxygen to phototrophic conditions with 60 W·m<sup>−</sup><sup>2</sup> white light revealed a remarkably faster adaption process for *R. capsulatus*, i.e., entering the exponential growth after the shift took ~4 h for *R. capsulatus* but ~21 h for *R. sphaeroides* (Figure 1B). As seen by the doubling time, *R. capsulatus* also grew faster in exponential phase under phototrophic conditions than *R. sphaeroides* (td of 3 h 35 min ± 4 min for *R. capsulatus* and 6 h 50 min ± 30 min for *R. sphaeroides*). This supported our hypothesis that the higher pigment content of *R. capsulatus* would allow a faster switch to occur to phototrophic growth. Interestingly, carotenoids (especially SE and SO) strongly contributed to the growth benefit of *R. capsulatus*, as shown by experiments with transposon mutants lacking these carotenoids (Figure S1).

 DHURELF SKRWRWURSKLF *R.c.* 6% *R.s.* **7LPHK**

**Figure 1.** Growth of *R. capsulatus* and *R. sphaeroides* under different oxygen and light conditions. Exponential phase cultures of *R. capsulatus* (*R.c.* SB1003) and *R. sphaeroides* (*R.s.* 2.4.1) were diluted to an OD660 of 0.2 at time point 0 h. The OD660 was plotted semi-logarithmically against the time. Data points represent the mean of biological triplicates and error bars depict the standard deviation (standard deviations might not be visible if they are too small). (**A**) Microaerobically growing cultures. (**B**) Aerobically growing cultures shifted to phototrophic growth. (**C**) Microaerobically growing cultures were exposed to photooxidative stress when an OD660 of ~0.4 was reached. Figure S2 shows growth under dark and light conditions with or without methylene blue.

We also asked the question of how a strong increase of photooxidative stress caused by 1O2 would affect growth of the two *Rhodobacter* species. To test this, pigmented cultures (after microaerobic cultivation) were cultivated under aerobic conditions in the dark and shifted to high light conditions (800 <sup>W</sup>·m<sup>−</sup>2) in the presence of methylene blue (0.2 μM) when an OD660 of 0.4 was reached. These conditions were previously shown to produce 1O2 and to induce a specific response in *R. sphaeroides* [27,28]. Importantly, neither methylene blue in the dark nor high light without methylene blue resulted in a strong growth retardation (Figure S2). After initiating photooxidative stress, *R. capsulatus* showed faster growth than *R. sphaeroides*, but slowed down earlier. As a consequence, both strains reached the same OD660after 12 h (Figure 1C).

Although a stronger pigmentation of *R. capsulatus* cultures under high oxygen tension compared to *R. sphaeroides* was obvious, we wanted to quantify the differences and also analyze the ratio of carotenoids and bacteriochlorophylls throughout growth at different conditions. While bacteriochlorophyll functions as a photosensitizer that promotes the production of 1O2, carotenoids can quench 1O2, and are thus part of the defense system against 1O2. In general, *R. capsulatus* showed a higher amount of bacteriochlorophyll *a* (Bchl *a*) and carotenoids, e.g., under microaerobic conditions (Figure S3) at an OD660 of ~0.4: *R. capsulatus* had ~3.0 μM per OD660 carotenoids and ~2.8 μM per OD660 Bchl *a*, whereas *R. sphaeroides* had ~0.7 μM per OD660 carotenoids and ~2.0 μM per OD660 Bchl *a*. After a shift to phototrophic growth, carotenoid and Bchl *a* levels steadily increased in *R. capsulatus,* reaching levels

that were much higher than under aerobic conditions (Figure 2A). Since *R. sphaeroides* stopped growing after this transition for nearly 13 h, the carotenoid and Bchl *a* levels stayed low during this time period and increased only slowly. Thirty-two hours after the shift, carotenoid and Bchl *a* levels were about 3.6-fold and 4.1-fold higher, respectively, in *R. capsulatus* (Figure 2A). After a shift from low to high oxygen tension and addition of methylene blue in the dark, the pigment level steadily dropped in both strains (Figure 2B). When illumination was started to generate 1O2, the Bchl *a* level dropped further. The carotenoid level in *R. capsulatus*, however, remained fairly constant (~1.4 μM per OD660), with a small peak (~1.6 μM per OD660) after four hours of stress. By contrast, *R. sphaeroides* showed a steady decline of carotenoids.

**Figure 2.** *R. capsulatus* has a stronger pigmentation and a higher carotenoid to bacteriochlorophyll *a* ratio than *R. sphaeroides*. Cultures in the exponential phase were diluted to an OD660 of 0.2 at time point 0 h. The content of carotenoids (Crt) and bacteriochlorophyll *a* (Bchl *a*) was normalized to the respective OD660-values and plotted against the time for (**A**) aerobically growing cultures shifted to phototrophic growth, and (**B**) microaerobically growing cultures shifted to aerobic dark conditions followed by exposure to photooxidative stress. The Crt:Bchl *a* ratio was plotted against the time for (**C**) *R. capsulatus* either continuously grown under microaerobic conditions or shifted to different oxygen and light conditions, and for (**D**) *R. capsulatus* and *R. sphaeroides* after a shift from microaerobic to aerobic dark conditions followed by photooxidative stress. Data points represent the mean of biological triplicates and error bars depict the standard deviation (standard deviations might not be visible if they are too small).

Not only the total amount of pigments may be important for adaptation of *Rhodobacter* species to changing conditions, but also the ratio of the 1O2-quenching carotenoids (Crt) to the 1O2-producing BChl *a*. In *R. capsulatus*, the Crt:BChl *a* ratio did not change much during continuous cultivation under microaerobic conditions (ratio of ~1.1) or after a shift from aerobic to phototrophic conditions (ratio of ~0.8; Figure 2C). However, a shift from microaerobic to aerobic conditions resulted in a strong increase of the Crt:Bchl *a* ratio from ~1.2 to ~3.0 in *R. capsulatus* (Figure 2C). Figure 2D compares the change in the Crt:Bchl *a* ratio between *R. sphaeroides* and *R. capsulatus* upon exposure to 1O2. The ratios remained fairly constant in both species during microaerobic growth and also after the shift to aerobic dark conditions in the presence of methylene blue. The ratios were ~1.1 in *R. capsulatus* and ~0.5 in *R. sphaeroides*. After the start of illumination and the production of 1O2, the Crt:Bchl *a* ratio increased in *R. capsulatus* from ~1.1 to nearly 3.0. In *R. sphaeroides* this ratio was below one under all conditions, and only increased from ~0.5 to ~0.8 after the initiation of photooxidative stress.

#### *3.2. Generation of ROS in R. sphaeroides and R. capsulatus upon Photooxidative Stress*

To see whether the different Crt:BChl *a* ratios observed in the two *Rhodobacter* species under photooxidative stress conditions (Figure 2D) would affect 1O2 levels, the fluorescent probe Singlet Oxygen Sensor Green (SOSG) was used for in vivo 1O2 measurements. Since white light itself can affect SOSG fluorescence, red light was used for illumination [62]. In cell-free reactions, the combination of oxygen, red light, and methylene blue caused enhanced SOSG fluorescence due to photosensitized formation of 1O2 (Figure S4). When pigmented cultures were incubated under aerobic conditions in the presence of methylene blue, the ratio of SOSG fluorescence between illuminated samples and dark controls was significantly higher for *R. sphaeroides* than for *R. capsulatus* (Figure 3A), indicating enhanced 1O2 levels in *R. sphaeroides*. In addition, general ROS formation was measured by applying 2,7-dihydrodichlorofluorescein diacetate (H2DCFDA) before and after starting the white light illumination of aerobic cultures in the presence of methylene blue. The fluorogenic probe H2DCFDA is mainly specific for hydrogen peroxide, peroxynitrite anions, and peroxyl radicals [63], ROS that are partly generated downstream of 1O2 [15,64]. An increase in DCF fluorescence indicates elevated ROS levels. We found significant differences in fluorescence levels between the two species before and 10 min after initiating photooxidative stress (Figure 3B). The relative fluorescence intensity indicated higher ROS levels in *R. sphaeroides* compared to *R. capsulatus* by a factor of 1.6 and 1.5 for 0 and 10 min of photooxidative stress, respectively (Figure 3B).

**Figure 3.** Determination of intracellular 1O2 and ROS levels in *R. capsulatus* and *R. sphaeroides* under photooxidative stress conditions. (**A**) The fluorogenic probe SOSG was used for 1O2 detection. Fluorescence intensities were normalized to BChl *a* levels. Ratios between illuminated samples (800 W·m<sup>−</sup><sup>2</sup> red light) and dark controls were calculated. (**B**) The fluorogenic probe H2DCFDA was used for ROS detection. Fluorescence intensities were normalized to the OD660 and displayed in arbitrary units (AU). Data points indicate individual measurements and bars represent the mean. Two-way ANOVA followed by Bonferroni posttest was used to compare results from *R. capsulatus* (*R.c.* SB1003) and *R. sphaeroides* (*R.s.* 2.4.1) (\* *p*-value < 0.05, \*\*\* *p*-value <0.001).

#### *3.3. The rpoE-chrR Locus of R. capsulatus Shows a Unique Genetic Context in Comparison to Other Bacteria within the Rhodobacteraceae*

Detailed work in *R. sphaeroides* identified the sigma factor RpoE as the master regulator of the response to 1O2 [1,65]. RpoE is primarily controlled by its cognate antisigma factor ChrR, but full activation of RpoE requires the RpoE regulon members RSP\_1090/91 (putative cyclopropane/cyclopropene fatty acid synthesis proteins) and the cyclopropane-fatty-acyl-phospholipid synthase CfaS [22,23]. In addition, DegS and RseP homologous proteases are involved in the degradation of ChrR [22]. In *R. sphaeroides*, the RSP\_1090/91 genes are located immediately upstream of the *rpoE-chrR* locus, a genetic arrangemen<sup>t</sup> that is conserved among many bacteria within the *Rhodobacteraceae* (e.g., *Roseobacter denitrificans*, *Dinoroseobacter shibae*, *Jannaschia rubra*, *Ruegeria litorea*, and *Oceanicola litoreus*). Interestingly, RpoE and ChrR homologs were not found in *R. capsulatus* using simple sequence alignment tools, but were revealed here by a structural homology search using Phyre<sup>2</sup> [56]. The RpoE proteins of *R. sphaeroides* and *R. capsulatus* only share 24% identity, but the confidence in the Phyre<sup>2</sup> structural homology analysis is high (99.9%). Importantly, the gene adjacent to the *rpoE* gene of *R. capsulatus* encodes a putative antisigma factor with 13% identity to ChrR (confidence of 99.4%). Although the identity values are relatively low, the high confidence in the Phyre<sup>2</sup> analysis strongly suggests that *R. capsulatus* has true RpoE (RCAP\_rcc00699) and ChrR (RCAP\_rcc00698) homologs. Moreover, and similar to *R. sphaeroides*, the *rpoE-chrR* locus of *R. capsulatus*is induced by 1O2, as revealed by RNA-seq (Figure 4 and Table 1). However, microsynteny analysis using Edgar 2.3 [57] showed that the genetic context of the *rpoE-chrR* locus in *R. capsulatus* is different from *R. sphaeroides* (Figure 4). The RpoE-dependent operon RSP\_1087-1091, which is located upstream of *rpoE-chrR* in *R. sphaeroides*, cannot be found next to *rpoE-chrR* in *R. capsulatus*. Moreover, respective homologs seem to be completely absent from the *R. capsulatus* genome (Table 1). Instead, the *rpoE-chrR* locus is located next to an operon, which encodes the glutathione peroxidase BsaA1, the cryptochrome/photolyase CryB, and two hypothetical proteins. All four genes are clearly induced by 1O2 (Figure 4). Interestingly, *cryB* belongs to the RpoHII regulon in *R. sphaeroides* [66]. Microsynteny analysis further revealed that *rpoE-chrR* is in close proximity to the photosynthetic gene cluster of *R. capsulatus* (*bch* and *puf* genes in Figure 4). This genetic arrangemen<sup>t</sup> cannot be found in closely related members within the *Rhodobacteraceae* (155 genomes analyzed in total), and it remains speculative whether this unique gene colocalization is purely coincidental or represents a strong functional relationship between photosynthesis and the response to 1O2.

**Figure 4.** Microsynteny analysis of the *rpoE-chrR* locus in *R. capsulatus* and *R. sphaeroides*. The two upper panels show normalized read count distributions of a representative RNA-seq experiment with *R. capsulatus* (*R.c.* SB1003) before (T0) and 10 min after the onset of photooxidative stress (T10). The lower panels show the microsynteny analysis of the *rpoE-chrR* locus in *R. capsulatus* and its homologous genes in *R. sphaeroides* (*R.s.* 2.4.1). Homologs are indicated by identical colors.


**Table 1.** Transcriptome changes upon photooxidative stress of the RpoE regulon in *R. sphaeroides* compared to homologs in *R. capsulatus*.

1 It is indicated whether log2 fold changes (log2 FC; stressed versus unstressed) were statistically significant (\* *p*-value < 0.05, \*\* *p*-value < 0.01, and \*\*\* *p*-value < 0.001). Data for *R. sphaeroides* were retrieved from Berghoff and colleagues [28]. 2 Homologs found either by pBLAST (30% protein identity) or Phyre2. 3 Annotated as *rpoHI* in public databases.

#### *3.4. Transcriptome Analysis of the Response to Singlet Oxygen in R. capsulatus*

To learn more about the response to 1O2 in *R. capsulatus*, RNA-seq was performed for samples collected before (T0) and 10 min after the onset of 1O2 stress (T10). The reproducibility of biological replicates is shown by correlation analysis with highly significant Pearson's *r*-values of ≥ 0.985 for all possible interreplicate comparisons (Figure S5). Principal component analysis (PCA) further revealed a clear separation between T0 and T10 samples along the first dimension (Figure S6A). In total, 3441 transcripts were quantified in the RNA-seq analysis. Four hundred and seventy-one transcripts were up- and 261 transcripts were down-regulated upon 1O2 stress (log2 fold change ≥ 1 or ≤ −1 and *p*-value < 0.05). To validate the RNA-seq approach for *R. capsulatus* and to further confirm the microarray results for *R. sphaeroides* [28], quantitative RT-PCR (qRT-PCR) was performed for selected genes. The qRT-PCR data were in good agreemen<sup>t</sup> with both RNA-seq and microarray results (Figure 5). However, for *cbbM*, *gltD*, *cysP*, *crtI*, and *cbiX*, transcript levels were only increased in *R. capsulatus*. In *R. sphaeroides*, three independent primer pairs were unable to detect *cbiX* transcripts.

**Figure 5.** Gene expression changes for selected genes upon photooxidative stress in *Rhodobacter*. Relative transcript levels (RTL) were calculated after the onset of photooxidative stress in comparison to a non-stressed control in *R. capsulatus* (*R.c.* SB1003) and *R. sphaeroides* (*R.s.* 2.4.1). RNA-seq (this study) and microarray data [28] are shown for comparison. For qRT-PCR (qRT), bars represent the mean of biological triplicates, and error bars depict the standard deviation. The *cbiX* transcript was not detected (nd) in *R. sphaeroides*.

From *R. sphaeroides*, it is known that 13 out of 15 genes of the RpoE regulon are induced by 1O2 (Table 1; [28]). In *R. capsulatus*, protein homologs were only found for seven of the 15 RpoE regulon members, which applies to RpoE, ChrR, photolyase PhrA (PhrB), sigma factor RpoHII, cytochrome c2 CycA (CycA1), a polyamine transporter subunit (PotH1), and cyclohydrolase FolE2 (RCAP\_rcc01493) (Table 1). It is worth noting that the aforementioned *R. capsulatus* sigma factors RpoHI and RpoHII are wrongly annotated in public databases: our analyses (see below) clearly show that RCAP\_rcc00458 (annotated as *rpoHI*) represents *rpoHII*, and vice versa, RCAP\_rcc02811 (annotated as *rpoHII*) corresponds to *rpoHI* from *R. sphaeroides.* Thus, we refer to RCAP\_rcc00458 as *rpoHII* and to RCAP\_rcc02811 as *rpoHI*. Among the seven homologs of the RpoE regulon, only *rpoE*, *chrR*, *phrB*, *rpoHII*, and *folE2* were induced upon 1O2 stress in *R. capsulatus*, indicating that similarities are probably limited to the most important features, which has also been observed for *Roseobacter denitrificans* [67]. One of the conserved 1O2-related features includes sigma factor RpoHII, which shares a partially overlapping regulon with heat-shock sigma factor RpoHI in *R. sphaeroides* [10,19,20]. In contrast to *rpoE-chrR*, the genetic context of both *rpoHI* and *rpoHII* is partly conserved between *R. sphaeroides* and *R. capsulatus* (Figure 6A,B). From *R. sphaeroides* it is known that RpoHII is more important for 1O2 stress resistance than RpoHI [20]; the same was observed here for *R. capsulatus* (Figure 6C).

To identify the most prominently enriched functional groups in *R. capsulatus*, we conducted a Gene Ontology (GO) term enrichment analysis using BiNGO in Cytoscape [58,59] for the 100 transcripts with the strongest increase. Transcripts contributing to oxidation–reduction processes were significantly increased and formed the largest group (24 transcripts; Figure 7). Furthermore, we could identify several stress-related functional groups. The first group comprises nine transcripts encoding proteins with a role in protein turnover and repair, which applies to the peptide methionine sulfoxide reductases MsrA1, MsrA2, MsrB1, and MsrB2, the chaperones GroS and GroL, and the peptidases/proteases Dcp, TldD, and RCAP\_rcc03333 (Figure 7). The second group comprises five transcripts encoding proteins with a known role in the (photo-) oxidative stress response, including RpoE and ChrR, RpoE-dependent GTP cyclohydrolase FolE2, peroxidase BsaA1, and glutathione-disulfide reductase Gor (Figure 7). The third and last group is formed by five transcripts encoding proteins with a function in DNA damage repair, which applies to photolyases PhrB and CryB, two components of the UvrABC complex, and the A/G-specific adenine glycosylase MutY (Figure 7).

**Figure 6.** Analysis of *rpoHI* and *rpoHII* in *R. capsulatus*. RNA-seq and microsynteny analysis for (**A**) *rpoHI* and (**B**) *rpoHII* in *R. capsulatus*. The two upper panels show read count distributions of a representative RNA-seq experiment with *R. capsulatus* (*R.c.* SB1003) before (T0) and 10 min after the onset of photooxidative stress (T10). The lower panels show the microsynteny analysis of the *rpoH* locus in *R. capsulatus* and its homologous genes in *R. sphaeroides* (*R.s.* 2.4.1). Homologs are indicated by identical colors. (**C**) Zone of inhibition assay showing the sensitivity of *R. capsulatus* strains to photooxidative stress. Data points indicate individual measurements and bars represent the mean. One-way ANOVA followed by Bonferroni posttest was used to compare results from different *R. capsulatus* strains (\* *p*-value < 0.05, \*\*\* *p*-value < 0.001).

**Figure 7.** Transcriptome analysis of *R. capsulatus* reveals functional groups with importance to photooxidative stress. Changes in transcript abundance after 10 min of photooxidative stress were determined by RNA-seq of biological triplicates. The volcano plot depicts log2 fold changes (10 min versus 0 min) of all quantified transcripts and the corresponding *p*-values (as negative log10). The horizontal dashed line indicates the cutoff for statistical significance (*p* < 0.05), and the vertical dashed lines indicate log2 fold changes ≤ −1 and ≥ 1. Significantly increased transcripts with a log2 fold change ≥ 1 are indicated as red open circles. Transcripts encoding proteins with known stress-related functions are highlighted in yellow. Boxes below the volcano plot depict functional groups as determined by GO term enrichment analysis of the 100 transcripts with the strongest increase. Log2 fold changes of ≥ 2 or ≥ 3 are indicated with + and ++, respectively.

#### *3.5. Similar Proteins Fulfill Important Functions in Response to Singlet Oxygen in R. sphaeroides and R. capsulatus*

It is known that changes on the RNA level are not necessarily reflected by changes in protein abundance. We therefore complemented our dataset by proteomic analysis of samples collected before (T0) and 90 min after the onset of 1O2 stress (T90). Correlation between the transcriptome (T10) and proteome (T90) was fairly low (Pearson's *r*-value = 0.49). Proteins were analyzed by LC-MS/MS and applied to a label-free quantification (LFQ) approach [68]. Pearson's *r*-values of ≥ 0.975 for interreplicate comparisons demonstrated high reproducibility of the LFQ approach (Figure S7). Principal component analysis (PCA) further revealed a clear separation between T0 and T90 samples along the first dimension (Figure S6B). LFQ intensities, reflecting protein abundance, were subsequently used to calculate fold changes between conditions. In total, 1507 proteins were quantified, revealing 46 increased and 35 decreased proteins upon 1O2 stress (log2 fold change ≥ 1 or ≤ −1 and *p*-value < 0.05). Increased proteins were subjected to GO term enrichment analysis using BiNGO in Cytoscape [58,59], and compared to proteome data from *R. sphaeroides* [28]. Seven homologous proteins were identified as increased in both organisms (Figure 8), including the three methionine sulfoxide reductases MsrA, MsrB1, and MsrB2, GTP cyclohydrolase FolE2, ATP-dependent protease ClpA, a putative protease (RCAP\_rcc03333/RSP\_1490), and an uncharacterized protein (RCAP\_rcc00543/RSP\_1760). All other proteins were only found to be increased in one of the two organisms (39 proteins in *R. capsulatus* and 43 proteins in *R. sphaeroides*), and only one functional group was clearly enriched in both organisms by means of GO terms, i.e., proteins with a function in oxidation–reduction processes. However, we identified several proteins with stress-related functions and grouped them accordingly (Figure 8). a prominent group relates to protein turnover and repair, including the aforementioned methionine sulfoxide reductases and several proteases, that are either increased in both organisms (ClpA and RCAP\_rcc03333/RSP\_1490) or only in one of the organisms (Lon, HslV, and ClpB in *R. capsulatus*; PqqL, MoxR, and ClpS in *R. sphaeroides*). Other proteins are directly related to the (photo-) oxidative stress response, like glutathione-disulfide reductase Gor and glutathione peroxidase BsaA1 in *R. capsulatus*, and a thioredoxin (RSP\_0725), a peroxiredoxin (RSP\_2973), and several RpoE regulon members in *R. sphaeroides* (Figure 8). a last group includes proteins involved in DNA damage repair, like components of the UvrABC complex and photolyase PhrB in *R. capsulatus*. Regarding other noteworthy increases, the protein CbiX caught our interest. As the terminal enzyme of the siroheme biosynthesis [69], CbiX does not belong to any of the described groups; however, its increase was one of the strongest. As *R. sphaeroides* lacked a comparable increase of the CbiX-homolog, this could hint at a core di fference in the photooxidative stress responses.

**Figure 8.** Functional characterization of proteins with increased abundance upon photooxidative stress. Changes in protein abundance after 90 min of photooxidative stress in *R. capsulatus* were determined by a label-free approach using LC-MS/MS of biological triplicates.

The volcano plot depicts log2 fold changes (90 min versus 0 min) of all quantified proteins and the corresponding *p*-values (as negative log10). The horizontal dashed line indicates the cuto ff for statistical significance (*p* < 0.05), and the vertical dashed lines indicate log2 fold changes ≤ −1 and ≥ 1. Significantly increased proteins with a log2 fold change ≥ 1 are highlighted. Proteome data were compared to *R. sphaeroides* [28] and increased proteins illustrated in an Euler diagram. Boxes below the volcano plot depict functional groups as determined by GO term enrichment analysis. Log2 fold changes of ≥ 2 or ≥ 3 are indicated with + and ++, respectively. Colors indicate whether proteins were found only in *R. capsulatus* (red), only in *R. sphaeroides* (black), or in both species (blue). See legend for details.
