Plasmids, bacterial strains, and oligonucleotide primers used in this study are listed in Table 1. R. sphaeroides 2.4.1 was grown aerobically in Sistrom’s minimal medium at 30˚C [18]. Photosynthetic cultures were grown under low light

intensity (3 watts/m²) or medium light intensity (15 watts/m²). In addition, the R. sphaeroides culture was also grown under an anaerobic and dark condition with DMSO (60 μmol/L). All Escherichia coli strains were grown at 37˚C in Luria-Bertani (LB) medium. Kanamycin (Km) was used to select for the presence of pLO1 plasmids in E. coli and R. sphaeroides, at a concentration of 50 μg/mL. For molecular analyses, standard methods were used for plasmid preparation, restriction endonuclease digestion, ligation, polymerase chain reaction, and DNA sequencing.

A ctrA null allele (an in-frame deletion) was constructed using the gene knockout method [19]. The in-frame deletion allele (ΔctrA allele) was constructed in vitro following two steps. A fragment containing 787-bp upstream of ctrA to 21-bp of the 5’ ctrA coding sequence was amplified using ctrA-upstream F1 (a XbaI site was introduced at the 5’ end of this primer) and ctrA-upstream R2 primers. Another fragment containing the 803-bp downstream of ctrA to 21-bp of the 3’ ctrA coding sequence was amplified using ctrA-downstream F3 and ctrA-downstream R4 primers (a XbaI site was introduced at the 5’ end of this primer). ctrA-upstream R2 and ctrA-downstream F3 primers were designed to be complementary with each other. These two PCR fragments were sequenced to confirm their sequences and then used in a second round of PCR as template DNA, using the ctrA-upstream F1 (forward primer) and the ctrA-upstream R2 (reverse primer). As a result, the complementary sequences at the ends of the PCR fragments produced a fusion ctrA deletion fragment (ΔctrA allele), which contained only a 14 amino acid coding sequence corresponding to the first seven and the last seven codons of the ctrA gene.

The suicide vector, pLO1 [20] , was linearized with XbaI and then ligated to the XbaI-digested ΔctrA allele. pLO1 is an E. coli plasmid containing the Km^r gene and the Bacillus subtilis sacB gene [21]. As a suicide vector, pLO1 cannot autonomously replicate in R. sphaeroides, but can replicate in E. coli [20]. E. coli DH5α cells were then transformed with the resulting recombinant plasmid, pLO1ΔctrA. The deletion was confirmed by sequencing the entire insert of the recombinant plasmid using flanking sequencing primers, ctrA-sequencing F1 and ctrA-sequencing R2.

The E. coli S17-1 cells were transformed with the pLO1ΔctrA plasmid, since S17-1 was used for transferring the pLO1ΔctrA plasmid to R. sphaeroides by conjugation. The plasmid was mobilized into R. sphaeroides using the biparental mating method [19] [22] [23]. E. coli S17-1 donor cells with recombinant plasmids (pLO1ΔctrA) were grown in LB mediums with kanamycin to mid-log phase; then 200 μl cultures were spun down. Pellets were washed with LB medium for five times to get rid of kanamycin. On the other hand, R. sphaeroides recipient cells were also grown to mid-log phase and a one-milliliter culture was then pelleted. Both pellets were then mixed in 50 μl LB medium and spotted onto solid LB medium for overnight incubation at 30˚C allowing plasmids to transfer to R. sphaeroides by conjugation.

After overnight incubation, cells on spotting plates were collected and washed with SIS medium 5 times to maximally avoid LB medium to prevent E. coli contamination in the next step. Ten independent heterogenotes of R. sphaeroides, which resulted from a single recombination event, were selected as Km^r colonies on SIS agar plates with kanamycin. Plates were incubated for four days until pink colonies were visible. These Km^r colonies were picked and grown in liquid culture without antibiotics for several times, and then plated out on LB agar plates containing 15% sucrose, which selected for the double crossover event leading to the replacement of the wild type ctrA gene by the deletion allele [24]. The sacB gene on the plasmid encodes levansucrase that is secreted in culture medium under induced conditions in the presence of sucrose, which causes cell-lysis or inhibition of cell growth [24]. Plates were incubated at 30˚C for several days until pink colonies appeared. Multiple colonies from each plate were patched on both SIS plates and SIS plus kanamycin replica plates. Plates were incubated for at least 5 days. Only km^s colonies were selected from SIS replica plates and were grown under both with and without kanamycin to confirm their kanamycin sensitivities. Eventually, colony PCR was performed on 10 kanamycin-sensitive colonies to screen for the loss of the ctrA gene. Putative colonies with loss of ctrA were then grown in SIS liquid cultures, and genomic DNA was then extracted. Gene replacement was confirmed by amplifying the fragment containing the ctrA deletion by performing PCR using genomic DNA previously extracted from putative mutant strains and subsequent sequencing of the PCR product.

Also, ctrA-null strain was complemented with pSHSU0003 recombinant plasmid, pRK415 containing ctrA wild-type gene with native promoter. A 931 base pair fragment containing the wild type ctrA was amplified using ctrA2F and ctrA2R primers. EcoRI and HndIII sires were introduced in ctrA 2F and ctrA 2R primer, respectively. PCR product was digested with EcoRI and HindIII and then cloned into EcoRI/HindII sites of pRK415. The recombinant plasmid, pRK415ctrA was mobilized in to R. sphaeroides ΔctrA strain for complementation study.

Total RNA was isolated from R. sphaeroides’ cells grown in lag, mid log, and stationary phases using a Qiagen RNeasy Midi Kit. All RNA samples were checked for concentration and DNA contamination by measuring absorbance at OD₂₆₀ and OD₂₈₀. A Power Sybr^@ Green RNA-to-C_T^TM 1-step Kit (Applied Biosystems) was used for qPCR. Three sets of internal primers as listed in Table 2

^aΔΔC_T = ΔC_T − ΔC_{T. rpoZ}. The average of expressions of genes derived from wild type cultures with OD₆₀₀ = 0.5 were set as the reference line for expression changes.

were used for measuring ctrA mRNA expression levels to ensure integrity of the full length mRNA. The rpoZ gene, which encodes the ω subunit of RNA polymerase, was used to normalize mRNA expression values for ctrA. The Applied Biosystem StepOne Real-time PCR System (Life Technologies, Gaithersburg, MD) was used to monitor amplifications to quantify mRNA expressions. The 2^-ΔΔC_T method was used to analyze relative gene expression [25]. Statistical analysis was performed using Minitab 15 statistical software (Minitab, Minneapolis, MN).

Cell densities of both wild-type and ctrA-null mutant cells were measured by optical densities at 600 nm at different time points to plot growth curves for both wild type and mutant strains. Microscopic examination was performed by fixing cells onto the slides using 20 μL of poly-L-lysine (Sigma-Aldrich P4832), and then examined under 1000x magnification. Swimming motility was assayed on swim plates, which contained SIS medium with 100 μM succinate as the minimal carbon source and solidified with 0.25% (w/v) agar. Swarming motility was tested on swarm plates containing regular SIS medium solidified with 0.25% (w/v) agar. R. sphaeroides cells (5 μl) were placed on swim plates or stabbed into semi-solid swarm plates and incubated at 30˚C aerobically in the dark for several days as previously described [26]. Swarming was also determined by stabbing cells with a straightened toothpick into test tubes with SIS medium solidified with 0.25% (w/v) agar as previously described [27]. Test tubes were incubated with constant illumination of medium light (15 watt/m²) for at least 4 - 5 days. Alternatively, 3 μl samples from cultures grown to stationary phase were placed on the surface of swarm plates and incubated at 30˚C aerobically in the dark [28]. Ring formations were monitored after several days.

Protein sequences of CtrA homologs from 40 different Proteobacteria species were obtained from NCBI. Protein sequences were aligned using MUSCLE [29] and phylogenetic analyses were performed using PhyML [30] , under the Jones-Taylor-Thornton (JTT) model [31] with an estimated transition/transversion ratio and estimated proportion of invariable sites. The number of bootstrap replications performed was 300. The workflow was carried out in Phylogeny.fr (Dereeper et al., 2008). The phylogenetic tree output was treated as unrooted.

The MEME program [32] , a model based on statistical expectation maximization, was used to identify the CtrA binding sites within the promoter regions or upstream regions of genes identified and confirmed as being part of the CtrA regulons in C. crescentus [5]. These predicted consensus sequence that binds CtrA in C. crescentus has been identified as TTAA-N₇-TTAAC (gapped) and TTAACCAT (ungapped) [5] , and these motifs were then compared with the experimentally validated CtrA binding motifs; the best match was subsequently used to scan the noncoding sequences in both chromosomes of R. sphaeroides for CtrA-binding motifs using the MAST program [33]. All DNA sequences and their corresponding protein sequences as well as upstream and downstream sequences for species examined in this study were obtained from the NCBI Database (http://www.ncbi.nlm.nih.gov/).

The Cluster of Orthologous Groups [34] [35] , a classification in the NCBI database provides a tool in examining gene roles. There are four major COG functions, which include 1) Information storage and Processing, 2) Cellular Processes, 3) Metabolism, 4) Poorly Characterized functions. These major groupings were further classified into 25 sub-groups (detailed in Figure 1 legend). COG functions for C. crescentus, R. capsulatus, and R. sphaeroides ctrA-regulated genes were subsequently analyzed for differences and similarities in ctrA control across these species. The ctrA-regulated genes ascertained in C.

crescentus and R. capsulatus were garnered from previous studies [5] [12] , although these studies had different methodologies in comparison to the present one in determining the association of genes to ctrA.

A protein homology search was performed using a gapped BLASTP [36] Genes with ctrA binding sites in R. sphaeroides were compared against the genomes of R. capsulatus and C. crescentus. The matches were then analyzed to see whether the ctrA-regulated genes of R. sphaeroides matched to the ctrA-regulated genes of those organisms.

The ctrA-null strain of R. sphaeroides was kanamycin-sensitive; this sensitivity of the mutant strains was checked repeatedly on both SIS plates and in SIS liquid culture with kanamycin. Km^s strains selected could be either ctrA-null mutants or wild type. To select for mutants, colony PCR was performed to distinguish mutants from wild type, as shown in Figure 2. When the primer ctrA-upstream F1 located on upstream of the ctrA gene and primer ctrA-qPCR-1 R2 located inside the deleted ctrA coding region was used to amplify the genomic DNA of mutants and wild type cells, it was predicted that wild type genomic DNA should amplify a 918-bp PCR product while the genomic DNA from mutant strains should not amplify any PCR product since ctrA-qPCR R2 primer binding site was missing in mutant strains. As shown in Figure 2(a), no PCR product was detected for the mutant strain, while the wild type produced a ~1000 bp PCR product. This result was consistent with the prediction that wild type ctrA would be replaced by the deleted ctrA allele. Genomic DNA of the false positive strains derived from sucrose plates produced the same PCR product as wild type genomic DNA. Alternatively, using sequencing primers F1 and R2 that are located on 181-bp upstream and 164-bp downstream regions, respectively, a much smaller fragment (387 bp) was predicted to be amplified using mutant genomic DNA as template, compared with a 1059-bp PCR product amplified using wild type DNA as shown in Figure 2(b). The sequencing of PCR products using this set of primers further confirmed the deletion of ctrA in this mutant strain. On the other hand, both fragments were amplified in the Km^r strains since the wild-type ctrA and ctrA-null alleles were present in the chromosome. Sequencing the ctrA target from the null strain confirmed its integrity and this strain was used for further analysis.

Growth curves for both wild type and ctrA-null mutant strains under both aerobic and photosynthetic conditions (10 watts/m²) revealed no significant differences as shown in Figure 3. In addition, wild type and ctrA-null mutant strains of R. sphaeroides grown under these conditions displayed no differences in colony morphology or pigmentation characteristics examined under light microscopy as shown in Figure 4. Notably though, although both the wild type and the ctrA-null mutant strains displayed slow growth under the anaerobic-dark-DMSO condition, the ctrA-null mutant grew noticeably slower (approximately 1.5 times slower) and its doubling time was noticeably increased as well compared to the wild-type strain (data not shown).

These findings demonstrate that CtrA is not essential for cell growth and cell cycle regulation in R. sphaeroides. Although these findings seem similar to the observations previously reported in R. capsulatus, a bacterium closely related to R. sphaeroides, it seems that ctrA may play a more important role in growth under certain conditions (such as the anaerobic-dark-DMSO condition) in R. sphaeroides that was not seen in R. capsulatus. However, the role of ctrA still does not seem to be “essential” for cell cycle regulation in R. sphaeroides. These results support the hypothesis that symmetrically dividing species in the α-Proteobacteria group undergo cell division using a different yet unknown regulatory system from asymmetrically dividing species such as C. crescentus [37] [38].

The loss of ctrA has been shown to impair motility in C. crescentus as this gene positively regulates the expression of genes involved in flagellar assembly [2]. Additionally, ctrA has been shown to be possibly involved in motility and flagellar functions in R. capsulatus via motility assays [15] and via microarray expression analysis [12]. Since R. sphaeroides is closely related to R. capsulatus, it was originally hypothesized that ctrA may play an essential role in flagellar motility in R. sphaeroides as well.

R. sphaeroides has two distinct flagellar systems and its motility is controlled by approximately 30 proteins encoded by two different gene clusters (fla1 and fla2) on the primary chromosome. In general, polar flagellum in R. sphaeroides has been shown to be functional in liquid medium and motility can be observed in denser environments through the production of a large number of lateral flagella [27]. More specifically, the first flagellar system (fla1) of R. sphaeroides produces polar flagellum for swimming under both aerobic and photosynthetic growth conditions, while the expression of the second set of flagellar genes (fla2) seems to be positively regulated under anaerobic growth conditions and produces lateral flagella for swarming motility. The fla2 gene cluster is native to R. sphaeroides and does not code for the entire flagellum without the cooperation of some gene products from the fla1 gene cluster [27].

The results of this study revealed no significant change in swimming and swarming motility due to loss of ctrA in R. sphaeroides. As shown in Figure 5(a), both wild type and mutant cells indicated very weak motilities during aerobic growth (dark) with a very small ring formation. No significant difference was observed. Furthermore, very weak phototaxis may have occurred in both strains when the illumination was provided from only one side while holding other side in the dark (Figure 5(b)). Results shown in soft agar test tubes were consistent with results on swarm plate as shown in Figure 6.

The wild type R. sphaeroides 2.4.1 has poor motility compared to another R. sphaeroides strain, WS8, which is usually used as a model strain for flagellar motility and chemotaxis studies [27] [39]. As such, even though no significant change of motility was observed due to loss of CtrA in R. sphaeroides 2.4.1, it cannot be definitively concluded that CtrA is not involved in flagellar motility. For instance, it is quite possible that the genes from the fla2 gene cluster compensate for the impairment of the gene regulation in fla1 genes that could be responsible for flagellum formation. Moreover, since the promoter of the fla1 operon

(between RSP_1312 and RSP_1313) contains a half-CtrA binding site, expressions of thesegenes could be modulated by a CtrA regulator. In short, the native fla2 genes could have allowed for normal or near-normal flagellum formation in R. sphaeroides 2.4.1 and therefore no defect in the motility phenotype was observed. Thus, CtrA could be involved in yet another unknown pathway in R. sphaeroides especially since the CtrA proteins have diverged across α-Proteobacteria species.

The mRNA expression levels of ctrA during different growth phases, normalized with the expression of a reference gene (rpoZ), are shown in Table 2. There were slight differences in ctrA expressions related to different growth stages; however, differences of relative expression levels were not statistically significant (one-way ANOVA, p = 0.344), which was similar to the results previously shown in R. capsulatus [12]. The expression pattern demonstrated that ctrA is expressed independently of cell densities in the batch culture. The effect of the deletion of ctrA was further confirmed on a transcriptional level. The mutant strain yielded 10,000-fold (virtually undetectable) lower level mRNA product compared to wild type cells. The decrease in relative mRNA expression levels was statistically significant (paired t-test, p = 0.004; Table 2). Moreover, the melt curves show a single sharp peak generated by all three sets of primers specific to three different regions of the ctrA gene. The methods indicate high target specificity where the primers are likely binding to single locations in the genome and producing uniform products in wild type cells and mutant cells. Complementation analysis revealed ~30% lower expression of ctrA gene in the complemented strains compared to the expression level in the wild type (data not shown). The initial over-expression of CtrA from plasmids may have possibly repressed its own expression indicative of the autorepression of the ctrA gene.

Sequence alignments revealed that the CtrA protein is highly conserved among α-Proteobacteria. CtrA in R. sphaeroides shares 92.4% and 72.2% amino acid identities with CtrA in R. capsulatus and C. crescentus, respectively. All the CtrA homologs among the species have a conserved residue for a putative phosphorylation site (aspartic acid) at position 51. A phylogenetic tree of CtrA homologs for 40 α-Proteobacteria species is shown in Figure 7 and reveals that the CtrA homolog of R. sphaeroides is most closely related to that of R. capsulatus.

The non-essential function of CtrA has been observed in R. capsulatus, Rugeria sp. TM1040, and now in R. sphaeroides. Even though CtrA is highly conserved among species of the subgroup, the phylogenetic analysis demonstrated that these three species where CtrA exhibits non-essential functions are a very closely related clade as shown in Figure 7. Furthermore, these three species are separated from other species like C. crescentus, which exhibits asymmetric cell division. These results are in line with previous findings [37].

A schematic diagram shown in Figure 8 demonstrates a variety of upstream flanking regions among α-Proteobacteria species. Species undergoing asymmetric cell division have significantly longer flanking upstream regions (197 bp to 550 bp), indicating a larger and more complex promoter region, while in species such as R. sphaeroides, R. capsulatus and Rugeria sp. TM1040 that divide symmetrically, relatively smaller promoter regions were found (95 bp to 125 bp). Furthermore, putative CtrA binding sites were found in the upstream flanking region of ctrA in R. sphaeroides, suggesting the possibility that CtrA is auto-regulated, similar to the findings in C. crescentus [8].

As previously described, the DNA consensus sequence that binds CtrA in C. crescentus has been identified as TTAA-N₇-TTAAC (gapped) and TTAACCAT

(ungapped) [5]. Using upstream sequences of 53 genes that were previously identified as being directly controlled by CtrA [5] , the MEME program generated more restrictive DNA binding sequences for CtrA (as referred to above). Using the binding motif generated and validated by the MEME program, potential CtrA binding sites were identified by scanning the consensus binding motif against the R. sphaeroides genome and the results are shown in supplementary Table 1. A total of 127 genes distributed on both chromosomes of R. sphaeroides were identified to have putative CtrA binding sites in their promoter regions. These genes are involved in diverse functions such as in energy production, iron utilization, transport function, and transcriptional regulation.

CtrA binding site search analysis in R. sphaeroides also revealed conservation of the CtrA regulatory circuit compared with C. crescentus. For example, putative binding sites were found in both ctrA and cckA, indicating the possibility of the autoregulation of CtrA and direct control of CckA. The major difference between CtrA of R. capsulatus and C. crescentus was the lack of CtrA control on genes involving cell division, DNA methylation, and energy production in R. capsulatus [12]. However, potential CtrA binding sites were found in promoter regions of genes displaying functions in the above pathways, such as NADH dehydrogenase gene clusters and SAM-dependent methyltransferase. Notably, not all intergenic regions with binding motifs bind with CtrA in vivo [5].

Protein homology searches revealed that only a small number of ctrA-regulated genes are common across C. crescentus, R. capsulatus, and R. sphaeroides. Table 3 details the findings of these comparisons. More specifically, for R. sphaeroides ctrA-regulated genes, less than 20% of them had matches to C. crescentus and R. capsulatus ctrA-regulated genes. When filtered for an e-value < 10⁻⁵, only ~6% of the R. sphaeroides genes retained their matches to R. capsulatus and C. crescentus ctrA-regulated genes. The findings were similar in a comparison from R. capsulatus to C. crescentus as well. These results suggest a significant variability in ctrA gene control in these organisms and could partly explain the differences in the functions and essentiality of ctrA across these species.

Figure 1 displays the COG group classification of ctrA-regulated genes in C. crescentus, R. capsulatus, and R. sphaeroides (data for which is available in Additional Files 2, 3, and 4). Putative CtrA-regulated genes in R. capsulatus [12] and C. crescentus [5] were taken from data collected in previous studies, although it must be noted that these studies had different methodologies in ascertaining how genes were regulated or controlled by ctrA in comparison to the present one. The number of ctrA-regulated genes utilized for analysis for R. sphaeroides, C. crescentus, and R. capsulatus are 127, 116, and 143 genes, respectively, and data concerning the functions of genes can be found in Additional Files 2, 3, and 4, respectively. As evidenced by Figure 1, there is a wide variability of the cluster of gene functions that are ctrA-regulated across the organisms but no specific pattern of control is apparent among the three, which is expected since the function of ctrA seems to differ among them. In C. crescentus, where ctrA is essential, there are three COG groups with a greater proportion of genes when compared to R. capsulatus or R. sphaeroides: D (cell division and chromosome partitioning), M (cell envelope biogenesis, outer membrane), and T (signal transduction mechanisms). Actually, R. capsulatus and R. sphaeroides possessed almost no genes in COG group D, which constitutes genes directly

participating in the cell cycle. However, outside of these three COG groups, there was no other pattern that the ctrA in C. crescentus seemed to exert that illustrated a drastic difference in cellular regulation compared to R. capsulatus and R. sphaeroides. Thus, the essentiality of CtrA in C. crescentus may depend more on selective control that it exerts on a few genes and pathways that are not controlled by CtrA in R. capsulatus or R. sphaeroides. This is logical as all three seem to possess extensive and broad pathway associations across a variety of classifications.