• Open Access

Life cycle analysis of the model organism Rhodopirellula baltica SH 1T by transcriptome studies

Authors

  • Patricia Wecker,

    1. Max Planck Institute for Marine Microbiology, Microbial Genomics Group, Celsiusstr. 1, 28359 Bremen, Germany.
    2. Jacobs University Bremen gGmbH, Campusring 1, 28759 Bremen, Germany.
    Search for more papers by this author
    • Both authors contributed equally to this paper.

  • Christine Klockow,

    1. Max Planck Institute for Marine Microbiology, Microbial Genomics Group, Celsiusstr. 1, 28359 Bremen, Germany.
    2. Jacobs University Bremen gGmbH, Campusring 1, 28759 Bremen, Germany.
    Search for more papers by this author
    • Both authors contributed equally to this paper.

  • Margarete Schüler,

    1. Max Planck Institute for Biochemistry, Department. Molecular Structural Biology, Am Klopferspitz 18, 82152 Martinsried, Germany.
    Search for more papers by this author
  • Jérôme Dabin,

    1. UPMC Univ Paris 6, UMR 7139 Végétaux marins et Biomolécules, Station Biologique, F 29682, Roscoff, Bretagne, France.
    2. CNRS, UMR 7139 Végétaux marins et Biomolécules, Station Biologique, F 29682, Roscoff, Bretagne, France.
    Search for more papers by this author
  • Gurvan Michel,

    1. UPMC Univ Paris 6, UMR 7139 Végétaux marins et Biomolécules, Station Biologique, F 29682, Roscoff, Bretagne, France.
    2. CNRS, UMR 7139 Végétaux marins et Biomolécules, Station Biologique, F 29682, Roscoff, Bretagne, France.
    Search for more papers by this author
  • Frank O. Glöckner

    Corresponding author
    1. Max Planck Institute for Marine Microbiology, Microbial Genomics Group, Celsiusstr. 1, 28359 Bremen, Germany.
    2. Jacobs University Bremen gGmbH, Campusring 1, 28759 Bremen, Germany.
    Search for more papers by this author

E-mail fog@mpi-bremen.de; Tel. (+49) 2028 970; Fax (+49) 2028 580.

Summary

The marine organism Rhodopirellula baltica is a representative of the globally distributed phylum Planctomycetes whose members exhibit an intriguing lifestyle and cell morphology. The analysis of R. baltica's genome has revealed many biotechnologically promising features including a set of unique sulfatases and C1-metabolism genes. Salt resistance and the potential for adhesion in the adult phase of the cell cycle were observed during cultivation. To promote the understanding of this model organism and to specify the functions of potentially useful genes, gene expression throughout a growth curve was monitored using a whole genome microarray approach. Transcriptional profiling suggests that a large number of hypothetical proteins are active within the cell cycle and in the formation of the different cell morphologies. Numerous genes with potential biotechnological applications were found to be differentially regulated, revealing further characteristics of their functions and regulation mechanisms. More specifically, the experiments shed light on the expression patterns of genes belonging to the organism's general stress response, those involved in the reorganization of its genome and those effecting morphological changes. These transcriptomic results contribute to a better understanding of thus far unknown molecular elements of cell biology. Further, they pave the way for the biotechnological exploitation of R. baltica's distinctive metabolic features as a step towards sourcing the phylum Planctomycetes at large.

Introduction

Rhodopirellula baltica (RB) SH 1T was isolated from the water column in the Kiel Fjord (Baltic Sea) and taxonomically grouped within the bacterial phylum Planctomycetes (Schlesner et al., 2004). Members of this group are abundant in aquatic habitats and are considered to play a significant role in carbon cycling (Glöckner et al., 2003). Planctomycetes share several unique properties, such as peptidoglycan-free proteinaceous cell walls (Liesack et al., 1986), intracellular compartmentalization (Lindsay et al., 2001) and a mode of reproduction via budding. The latter trait results in a life cycle comprised of motile and sessile morphotypes and resembles that of Caulobacter crescentus (Gade et al., 2005a). The annotation of the completely sequenced genome of R. baltica allowed an initial assessment of its genetic potential and revealed several interesting and surprising traits such as a high number of sulfatase genes, a fascinating set of carbohydrate-active enzymes [CAZymes (Cantarel et al., 2009), (Michel et al., 2006; Dabin et al., 2008)] and a conspicuous C1-metabolism pathway (Glöckner et al. 2003 and references therein). Moreover, several features were detected that promise a potential for biotechnological exploitation. For instance, the genome harbours enzymes for the synthesis of complex organic molecules with possible applications in the pharmaceutical field, such as a gene set encoding a polyketide synthase, and enzymes important for the production of natural products in the food or animal-feed industry, such as those active in vitamin and amino acid biosynthesis. The application of sulfatases of R. baltica for the stereoselective conversion of racemates has been discussed in Wagner and Horn (2006) and Gadler and colleagues (2006). Many of the genes are unique and may be employed to enhance the efficiency of microbial strains already in biotechnical use. For example, genes coding for antibiotic resistance might be used to immunize production strains against certain antibiotics, while genes conferring virulence resistance could improve the robustness of these strains. Rhodopirellula baltica is capable of a sessile lifestyle, a feature which is desirable for cell recovery during the production of secreted products in biotechnological processes. Salt resistance is desirable for production strains as industrial media or waste water often have high salt concentrations. The fact that about half of the genes in R. baltica currently lack an assigned function further demonstrates the huge, but so far hidden, genetic potential of this model organism.

During the life cycle of R. baltica, cells change their morphology, form swarmer cells to sessile cells with holdfast substances, produce secondary metabolites and experience different conditions such as nutrient excess and deprivation as well as high cell densities. The analysis of growth-dependent expression patterns during the life cycle of R. baltica enables a screening for genes for specific physiological functions, morphotype differentiation and cellular processes most probably typical for this phylum.

Results and discussion

Gene expression studies

Rhodopirellula baltica was grown in a defined mineral medium with glucose as a sole carbon source to study changes in gene expression that reflect the organism's life cycle phases and response to nutrient depletion. The growth of R. baltica cells could not be synchronized. Nevertheless, microscopic examination showed that the culture is dominated by swarmer and budding cells in the early exponential growth phase, shifting to single and budding cells as well as rosettes in the transition phase, while the stationary phase is dominated by rosette formations. These observations are in accordance with previous results (Gade et al., 2005a).

Global differential gene expression results

Cultures of R. baltica were harvested at the onset of the exponential phase (after 44 h of incubation), during the exponential phase (mid-log phase, 62 h), in the transition phase between exponential and stationary phase (late-log phase, 82 h), and in the early and late stationary phase (96 h and 240 h respectively). Figure 1 presents an overview of the system in which the different growth phases were compared. The total number of genes detected as differentially regulated for each of the growth stage investigated is shown in Table 1. A complete list of genes can be found in the Appendix S1. Samples taken from the early log-phase (44 h) and the exponential phase (62 h) were differentiated by the up- and downregulation of 149 genes, of which 56% have been annotated as hypothetical proteins. The comparison between the exponential (62 h) and transition phase (82 h) revealed the regulation of only 90 genes, 40% of which have unknown functionality. A pronounced change in the transcriptome, with the differential regulation of 235 genes (59% hypothetical proteins), was observed when the transition phase (82 h) was compared with the stationary phase (96 h). A comparison between cultures from 82 h with those of late stationary phase (240 h) revealed the most pronounced changes of the transcriptome in our study: 863 genes were differentially expressed (Fig. 2), of which 58% are annotated as hypothetical proteins.

Figure 1.

Schematic drawing of the life cycle of R. baltica, modified after (Gade et al., 2005a) and hybridization scheme: R. baltica cultures were harvested at five time points indicated by the red circles. cDNA samples were hybridized in two replicates. Dots and arrowheads represent samples labelled with Alexa 547 and Alexa 647 respectively.

Table 1.  Total number of regulated genes during each measured time point of the growth curve and the fraction of the regulated genes annotated as conserved and/or hypothetical.
Number of regulated genes62 h vs. 44 h82 h vs. 62 h96 h vs. 82 h240 h vs. 82 h
  1. (%) * relative to the total number of 7325 open reading frames (ORFs) annotated in R. baltica genome.

  2. (%) ** relative to the total number of regulated genes.

Total (%)*149 (2%)90 (1%)235 (3%)863 (12%)
Encoding hypothetical proteins (%)**84 (56%)40 (44%)139 (59%)499 (58%)
Figure 2.

Number of significantly up- and downregulated genes during the growth curve at the four time points which were compared. White means upregulated and grey downregulated.

The analysis of the exponential phase (a comparison of the early-log at 44 h and late-log at 82 h with mid-log at 62 h) showed that a maximum fraction of 2% of the total number of genes was regulated. This reflects the favourable nutritional conditions encountered throughout these growth stages (Gade et al., 2005a).

The assignment of differentially expressed genes to functional cluster of orthologous group (COG) classes offered a first impression of R. baltica's metabolic activity through the different growth stages (Table 2). Genes associated with the ‘metabolism of amino acids’[E] and ‘carbohydrates’[G], genes for ‘energy production and conservation’[C], as well as genes for ‘DNA replication and recombination’[L] were downregulated in mid-exponential phase (62 h) in comparison with the 44 h sample. The repression (62 h vs. 44 h) of genes coding for oxidases (RB4416 and RB5701), a peptidase [RB4269, which was also found regulated under stress conditions (Wecker et al., 2009)], a synthase (RB6443) as well as lipases and esterases (RB7072 and RB3413) suggests a lower metabolic activity in the cells most likely due to lower nutrient availability compared with the early exponential phase.

Table 2.  Number of regulated genes with an assigned COG-category compared with the whole genome functional classification assignment according to the NCBI database (cut-off e-value e-4).
COG-classTotal in genome62 h vs. 44 h82 h vs. 62 h96 h vs. 82 h240 h vs. 82 h
UpDownUpDownUpDownUpDown
  1. Rows: [C] Energy production and conversion, [D] Cell division and chromosome partitioning, [E] Amino acid transport and metabolism, [F] Nucleotide transport and metabolism, [G] Carbohydrate transport and metabolism, [H] Coenzyme metabolism, [I] Lipid metabolism, [J] Translation, ribosomal structure and biogenesis, [K] Transcription, [L] DNA replication, recombination and repair, [M] Cell envelope biogenesis, outer membrane, [N] Cell motility and secretion, [O] Posttranslational modification, protein turnover, chaperones, [P] Inorganic ion transport and metabolism, [Q] Secondary metabolites biosynthesis, transport and catabolism, [R] General function prediction only, [S] Function unknown, [T] Signal transduction mechanisms.

C17004015097
D1800000000
E21113211176
F6400001011
G17212204085
H117010010110
I7501000012
J14401004049
K156011120510
L176021000134
M14203102086
N10203001011
O106009060312
P234041131156
Q14102204084
R41808311321911
S1771220521110
T21000202066

Interestingly, genes of the ‘cell envelope biogenesis, outer membrane’[M] and ‘cell motility and secretion’[N] categories, such as genes coding for flagellin [RB4454; also downregulated under stress (Wecker et al., 2009)], bacterial type II and III secretion system protein (RB5174) and for diverse transferases (RB6305 and RB6624), were repressed at 62 h. Thus, when compared with the mid-exponential phase (62 h), it appears that the onset of the exponential phase (44 h) triggers a remodelling of the cell wall and morphology, perhaps explained by an enhancement of the swarmer cell production. This response is likely to be very expensive energetically, and would explain the higher expression level of genes from COG category C observed (Jolly et al., 2007).

Comparison of the transition phase and stationary growth phase (82 h) with the mid-exponential phase (62 h) revealed that more genes were regulated in the stationary phase than the mid-exponential. It could be speculated that genes necessary for exponential growth under favourable conditions were already expressed in early log and mid-log phase (44 h and 62 h). Still, several genes showed differential expression. For example, R. baltica increased the glutamate dehydrogenase (RB6930) level, thus confirming proteome experiments (Hieu et al., 2008). The enzyme is involved in the biosynthesis of arginine, glutamate and proline. The latter is one of the major components of the R. baltica's cell wall. Upregulation of glutamate dehydrogenase could indicate that the organism's cell wall composition has been adapted to more unfavourable conditions in response to the decreasing nutrient concentration.

The cell density reached its highest point during the transition phase of the R. baltica flask culture. Stressed cells need to adapt metabolically to the nutrient limitation in the culture, which is reflected by our results. Genes coding for glutathione peroxidase (RB2244), thioredoxin (RB12160, see also Hieu et al., 2008) and bacterioferritin comigratory protein (RB12362), universal stress protein (uspE, RB4742) and chaperones (e.g. RB8966) were induced and diverse dehydrogenases, hydrolases and reductases were up- and downregulated for metabolic adaptation and to prepare for long-term survival under unfavourable conditions. The regulation of RB2244, RB12362, RB8966 is in line with the proteome results of the life cycle (Gade et al., 2005a).

Earlier experiments have shown that R. baltica survives for at least 14 days in the stationary phase at 28°C (Gade et al., 2005a). We compared expression profiles of the early (96 h) and the late stationary phase (240 h) with that of the transition phase (82 h). The overlap of both sample sets (96 h vs. 82 h and 240 h vs. 82 h), visualized by a Venn diagram (Fig. 3), revealed 103 genes with similar regulation behaviour. Seventy-one out of the 103 genes are annotated as hypothetical proteins and most of the remaining 32 genes are stress-related, as expected.

Figure 3.

Venn diagrams showing the distribution of all regulated genes.

The general response of R. baltica cells to the stationary phase environment includes the induction of genes associated with energy production, amino acid biosynthesis, signal transduction, transcriptional regulation, stress response and protein folding. Genes coding for the enzymes of carbon metabolism, translation control, energy production and amino acid biosynthesis were repressed. Rhodopirellula baltica decreased the expression level of genes belonging to the ribosomal machinery (RB12197, RB2543, RB264, RB5801, RB5804, RB7022, RB7818, RB8253, RB8725 and RB9304), transcription regulation (RB10339, RB10458, RB11223, RB12372 and RB1392) and DNA replication and recombination (RB11863 and RB3281) in the late stationary phase, probably due to reduced growth activity. Additionally, the organism expressed many genes coding for transposases, integrases and recombinases (RB10096, RB11303, RB11750, RB1190, RB3144, RB4826, RB5887, RB7388, RB12239, RB2186, RB6736, RB9907, RB6167, RB7389 and RB934), suggesting that under stress conditions, or in the late stationary phase, genome rearrangements take place. Due to the scarcity of operon structures in the genome of R. baltica, the rearrangement might be necessary to enable an efficient transcription during stressful phases. Six out of the 15 genes (RB11750, RB12239, RB2186, RB9907, RB7389 and RB934) have been found active in response to non-physiological conditions of temperature and salinity (Wecker et al., 2009), which indicates that expression of these genes constitutes a response to general stress. The upregulation of genes for the phenylalanine, tyrosine and tryptophane biosynthesis (RB6822 and RB6147) is noteworthy. This is consistent with proteome data (Gade et al., 2005a), but the physiological meaning is currently unknown. Furthermore, genes involved in the biosynthesis of serine, threonine and glycine (RB6932) and of lysine (RB4382) were upregulated.

Reaction to oxygen depletion is shown as part of the stress response as well. It is known from Agrobacter tumefaciens and Rhodobacter sphaeroides that the lack of oxygen triggers the increased production of ubiquinone (Cluis et al., 2007). The same effect can be seen in our results with the induction of genes for the ubiquinone biosynthesis (RB2748, RB2749 and RB2750) in the stationary phase. Also, it seems that the composition of the cell wall was modified in response to changes in the physiological state of the R. baltica cell during the late stationary phase. The organism exported more polysaccharides, as shown by the enhanced formation of rosettes indicating production of a holdfast substance. Diverse genes belonging to the cell membrane class [M] were upregulated. Rhodopirellula baltica activated the production of many outer membrane transporters and biopolymers (RB12053, RB12055, RB9308, RB1120, RB754), as well as several transferases (RB12690, RB2507, RB2985, RB5846) for alteration of the lipopolysaccharides that belong to the cell membrane. RB12053, RB12055 and RB12690 showed also a response to stressful conditions in our earlier study (Wecker et al., 2009). In addition, we found 19 genes annotated as membrane proteins to be regulated, as well as 13 secreted proteins. Two genes coding for bacterial micro-compartment proteins (RB2585 and RB2586) were upregulated. These proteins supposedly form primitive organelles in a variety of bacteria (Kerfeld et al., 2005). Such findings may indicate that the cell wall composition was adapted to maintain the cell shape. It could be that after entering the stationary phase, R. baltica cells, similar to what was found for the marine bacterium ‘Candidatus Pelagibacter ubique’, undergo morphological changes (cell shape), enabling them to return to exponential growth after being transferred to fresh medium or to a nutrient-rich environment (Sowell et al., 2008).

Cell division and cell cycle

Little is known about the cell division and replication regulation in Planctomycetes. None of the 18 R. baltica genes assigned to the COG category ‘cell division and chromosome partitioning’[D] were found to be regulated during this study, nor were any of the unclassified COG homologues of genes that are generally known to be involved in the cell division. This suggests that in R. baltica a number of regulated genes with thus far unknown function must play a role in the regulation of replication and cell division. However, clues can be garnered from the fact that R. baltica exhibits a life cycle similar to C. crescentus. DnaK/DnaJ and GroES/GroEL are the best-characterized chaperone systems in bacteria and are induced in cell cycle regulation and under heat shock in C. crescentus (Susin et al., 2006). Those chaperones have an important role in chromosome replication and partition and/or cell division. The R. baltica genome harbours genes coding for groEL (RB10629 and RB8970), dnaK (RB5754 and RB5755) and dnaJ[RB8972 (Wecker et al., 2009)] in the genome. Of these, only groEL (RB10629) is upregulated at 82 h (transition phase) in the life cycle experiment, which could also indicate a general response to stress (Wecker et al., 2009). In addition, both groEL-genes showed a regulation during the proteome experiments (Gade et al., 2005a). Rhodopirellula baltica also possesses two genes encoding DnaA (RB11579 and RB1706), a protein which initiates not only the DNA replication, but also promotes the expression of the components necessary for successful chromosome duplication in C. crescentus. However, dnaA was not found to be regulated in our experiments.

In order to acquire more knowledge about genes possibly involved in changes of the morphotype throughout the cell cycle and the role of genes classified as hypothetical proteins, the gene expression data were clustered (see Experimental procedures). Clusters can be found in Table 3 and Appendix S2. The clusters 4, 12 and 17 group genes which are only regulated in the exponential phase where nutrient limitation or stress conditions are unlikely. Therefore, these genes can be supposed to have a function in the morphology differentiation to swarmer cells or in the budding process. Genes belonging to clusters 11, 13, 19, 20, 21, 22 and 23 are upregulated during the transition phase. Genes annotated as hypothetical proteins belonging to the early stationary phase (96 h) were grouped in clusters 18, 24 and 25. Cluster 18 and 24 consist of many genes annotated as hypothetical proteins, which might be involved in morphology changes, e.g. cell shape and wall. Finally, cluster 14 genes are upregulated in the exponential (62 h) and downregulated in the stationary (96 h) phase in comparison with the transition phase. This cluster contains one sulfatase gene and 14 genes encoding hypothetical proteins. It can be assumed that these have a function in the morphotype differentiation.

Table 3.  Overview of the k-means clustering of the different growth stages of R. baltica. Thumbnail image of

Lipids

In this study, delta-9 desaturase (RB4876), which belongs to the unsaturated fatty acid biosynthesis pathway, was upregulated in the transition phase (82 h). Furthermore, expression of malonyl-CoA acyl carrier protein transacylase (RB314), an enzyme involved in fatty acid biosynthesis, was upregulated in the late stationary phase (240 h). Results from proteome studies (Gade et al., 2005a) indicate that the biosynthesis of sulfated saturated fatty acids occurs specifically during the stationary phase, but this could not be reproduced on the transcriptome level. In our recent study (Wecker et al., 2009), RB314 (together with the acyl carrier protein, RB318) was also upregulated under stress-inducing conditions. The increased expression of these genes might be part of a general stress response, possibly to enhance the stability of the membrane to protect the cell.

Tetrahydromethanopterin-(H4MPT)-linked C1-compound metabolism

Rhodopirellula baltica has the genetic potential to degrade C1-compounds, and the functional annotation predicted all enzymes necessary for the oxidation of formaldehyde to formate (Glöckner et al., 2003). How this pathway benefits the organism in its environment is so far unknown. Out of the 19 proposed genes encoding enzymes of C1-metabolism (Woebken et al., 2007), four genes were regulated in our experiment: fae (RB9338), mtdA (EC 1.5.1.5, RB10295) and orf7 (RB10246) were induced in the late stationary phase (240 h) compared with the transition phase (82 h), whereas orfY (RB10299) encoding a potential ATP-dependent carboligase (COG2232) was induced in the exponential and repressed in the early stationary phase. Table 4 gives an overview on C1-genes that have, to our knowledge, been found expressed so far. Although in silico studies show the presence of the C1-pathway in other Planctomycetes (Chistoserdova et al., 2004; Woebken et al., 2007), the specific functions of the genes in Planctomycetes are still unknown. However, the expression of some genes observed in this study suggests that this pathway has physiological relevance in R. baltica and most likely in other Planctomycetes as well.

Table 4.  Overview of the expressed or regulated genes coding for the C1-transfer enzymes in R. baltica.
IDGene name or productTranscriptomeProteome
Current studyWoebken et al., 2007Gade et al., 2005b
RB6210Tetrahydromethanopterin biosynthesis protein (ORF21)  
RB6300pabA  
RB6759mch  
RB8806ptpS   
RB9338fae1.5 
RB9834fm(w)dA  
RB9836fm(w)dC  
RB10246ORF72.6 
RB10295mtdA1.7 
RB10299ORFY1.7/ 2.5/ −1.7/ 0  
RB11969Protein containing ATP-grasp fold, DUF201-type (ORF1)  

Potential biotechnologically relevant genes

Throughout the life cycle, biotechnologically interesting genes were found to be regulated. The microarray experiments provide valuable information under which conditions these genes are expressed and how they interact. With a doubling time of approximately 12 h and complex growth conditions, R. baltica has limitations as a production strain. Nevertheless, the respective genes could be cloned and heterologously expressed. Considered here are gene products relevant for (i) new biotechnological processes, (ii) the improvement of existing processes, (iii) processes which could help to shift classical chemical production towards enzyme catalysed systems and (iv) genes with potentially medical applications. An overview of the regulated genes organized according to their possible application is presented in Appendix S3.

Food industry and animal feeding form a globally important market and many of the processes existing today leave room for improvements. Almost 50 genes of R. baltica, which were found to be differentially expressed in the present study, have potential applications in this field. Among them are genes for vitamin and sugar synthesis or amino acid conversion and several of the organism's sulfatase genes. The efficient and inexpensive synthesis of amino acids and vitamins is increasingly important as feed additives to raise livestock for a growing world population and sulfatases can be used in the food industry as thickening agent, and to convert racemates to single stereoisomers that may play a role in the pharmaceutical field.

The 41 genes listed in the category ‘pharmaceutical & medical application’ have a potential to improve current biosynthesis methods for vitamins, antibacterial drugs or complex organic molecules synthesis, with applications in cancer research. The biotechnological production of certain amino acids is important for the pharmaceutical industry. An example would be the biosynthesis of tryptophan, which can be administered as a mild sleeping drug (Hartmann, 1977). An additional instance of a gene product that may be valuable for the development of new medical treatments is an FKBP-like isomerase found to be differentially expressed in R. baltica. FKBP isomerases are involved in the immunoregulation.

Other enzymes, such as polymerases, might improve methods in molecular biology and would therefore benefit medical research or analytical techniques. For example, nine polymerase-associated genes are regulated in this study.

In the emerging field of white biotechnology, R. baltica could offer genes to improve the robustness of existing production strains with respect to antibiotics and phage resistance, for the production of washing detergents, for the synthesis of natural substances or for waste water treatment. The ability of R. baltica to form a holdfast substance might prove valuable as well. In many biotechnology processes the biomass is harvested together with the desired product and thus lost, which leads to additional costs for replacing the production cells.

Hence, it is favourable to retain the biomass during harvesting secreted proteins or enzymes, for example, by employing attached cells.

Sulfatases

The genome of R. baltica harbours an exceptional number of 110 sulfatase genes. Based on sequence similarity, it has been proposed that they are involved in the utilization of carbon from complex sulfated heteropolysaccharides (Glöckner et al., 2003).

During transcriptional analysis of the different R. baltica growth stages, we found 12 sulfatases up- or downregulated including one putative choline sulfatase (RB2254), three different arylsulfatases (RB4815, RB684 and RB686), two N-acetylgalactosamine-6-sulfate sulfatase (RB13148, RB5195), one mucin-desulfating sulfatase and four other sulfatases (RB110, RB10612, RB5294 and RB5424) (Table 5). Organic sulfate is not likely to have caused the induction of the sulfatase genes observed in this study as the medium lacked sulfated carbon. Five of the sulfatase genes mentioned above (RB110, RB2254, RB4815, RB5294 and RB686) were also found upregulated when the expression profile of R. baltica cells grown on a solid surface was compared with that of cells grown in liquid culture (data not shown), and RB2254 was found in the extracellular proteome (data not shown). The observation that certain sulfatases were only expressed in specific growth phases suggests that degrading polymers is not the only role of sulfatase genes in R. baltica. Based on the differential expression pattern of these sulfatase genes we propose an involvement in structural remodelling during the morphological differentiation of R. baltica cells, such as the formation of the holdfast substance, which is needed to attach to surfaces, the production of slime polysaccharides (Kumar, 2007) or in the cell wall formation, which consists of a protein sacculus with disulfide bonds (Liesack et al., 1986).

Table 5.  Differentially expressed sulfatase genes of R. baltica during the growth curve plus additional regulation data.
IDProductAASignalpGene nameEC number62 vs. 44 h62 vs. 82 h96 vs. 82 h240 vs. 82 hRemarks
  1. W: transcriptome studies. cells attached to chemostat wall (unpublished data), M: metagenome studies (Woebken et al., 2007), P: extracellular proteome (Hieu et al., 2008), T: transcriptome data (Wecker et al., 2009).

RB10612Sulfatase family protein4920     1.9 
RB110Sulfatase family protein5820  −1.5   W
RB13148N-acetyl-galactoseamine-6-sulfate sulfatase10120.2   −1.5−2.91.5T
RB2254Choline sulfatase5410betC3.1.6.6   1.8W, P
RB4815Arylsulfatase A4911arsA3.1.6.8   1.9W
RB5195N-acetyl-galactoseamine-6-sulfate sulfatase5051arsb3.1.6.12   1.5 
RB5282Mucin-desulfating sulfatase5780.5betC3.1.6.6   −1.7 
RB5294Sulfatase5330     1.6W, T
RB5424Sulfatase5730  −1.8    
RB684Arylsulfatase precursor6530.8atsA3.1.6.1   −1.5T
RB686Arylsulfatase5491atsA3.1.6.1−1.7   W, M
RB7875Arylsulfatase A precursor4920.9arsA3.1.6.8   2.3 

Another hypothesis is that, when sensing decreasing glucose concentrations, the cells express sulfatases to prepare for the degradation of alternative carbon sources. More experimental work is needed to clarify the diverse roles of the unusually high number of sulfatase genes in R. baltica.

Carbohydrate-active enzymes

Carbohydrate-active enzymes are responsible for the synthesis (glycosyltransferases), the modification (carbohydrate esterases) and the degradation [glycoside hydrolases (GH) and polysaccharide lyases] of polysaccharides and glycoconjugates, which play an important role in, e.g. cell wall synthesis and carbon storage. The degradation of carbohydrates taps a special nutrient source for the microorganisms (Michel et al., 2006). As the predicted lifestyle of R. baltica in its natural environment involves the attachment to and degradation of marine snow, its CAZymes might have a function in both the formation of the hold-fast substance and the breakdown of marine snow particles. Similarly, the field of application in biotechnology comprises the degradation of complex substances as well as the synthesis of specific organic substances for pharmaceutical use or for food industry.

Throughout R. baltica's life cycle, four CAZymes were found to be regulated. RB8895 is a sialidase, which is upregulated during exponential growth and downregulated in the stationary phase. Sialic acids are particularly abundant in animals in extracellular glycoproteins and glycolipids. When present in bacteria, sialic acids are found mostly in capsular polysaccharides and lipopolysaccharides (Angata and Varki, 2002). However, the key enzymes for sialic acid biosynthesis, N-acetyl neuraminate synthase (NeuB, EC 2.5.1.56) and N-acylneuraminate cytidyltransferase (NeuA, EC 2.7.7.43) are absent in the genome of R. baltica. In contrast, R. baltica possesses a complete pathway for sialic acid catabolism with seven family GH33 sialidases (including RB8895), an N-acetylneuraminate lyase (RB3352, EC 4.1.3.3), which was found to be expressed on the proteome level (Gade et al., 2005c), and a N-acetyl-D-glucosamine 2 epimerase (RB3348, EC 5.1.3.8). In absence of endogenous sialic acids, the role of RB8895 in the cell cycle is rather intriguing. The other three regulated genes encoding CAZymes (RB548, RB4894, RB5196) all belong to the family GH13, but have distinct activities. These enzymes are induced in the early or late stationary phase. RB5196 is highly similar to the amylosucrase from Neisseria polysaccharea (52% sequence identity) (Büttcher et al., 1997). This enzyme is not a GH, but a transglycosidase which catalyses the synthesis of an alpha-(1,4)-linked glucan polymer from sucrose. Therefore an alpha-(1,4) glucan is likely produced during the R. baltica cell cycle. Such a marine amylosucrase has exciting biotechnological potential, as it uses sucrose as sugar donor instead of an expensive activated sugar, such as ADP- or UDP-glucose. RB548 and RB4894 are both related to glycogen metabolism, encoding a glycogen branching and debranching enzyme respectively. RB548 was shown to be active on the proteome level (Gade et al., 2005c). A systematic search of glycogen-related enzymes (Ball and Morell, 2003) demonstrated that R. baltica possesses a complete glycogen metabolism (Dabin et al. in preparation). Therefore, it can be assumed that R. baltica makes use of its intracellular carbon storage (glycogen) in order to produce other polysaccharides or glycoconjugates, for example, to generate the holdfast substance. RB548 was also found to be expressed in proteome studies of cells grown on agar plates (Gade et al., 2005a).

Hypothetical proteins co-regulated with genes of potential biotechnologically relevant applications

Given the limited functional knowledge on cellular processes in Planctomycetes, it was not surprising that a large number of genes found to be differentially expressed in this study encoded (conserved) hypothetical proteins. In order to get information on the potential function of those genes, we delineated groups of genes with similar expression patterns during culture growth. Information on the genes populating clusters, which comprise genes with possible functions in biotechnology, can be found in the Appendix S2. Table 3 shows the average expression pattern of the clusters, in which genes with potential biotechnological applications were found.

We first focused on genes with unknown functions, which were co-regulated with potentially biotechnologically relevant genes (Appendix S4). For example, cluster 16 (repression in stationary phase) contains genes upregulated in the stationary and late stationary phase, among them several sulfatases and genes involved in amino acid turn over, as well as almost 150 genes with unknown function. Cluster 3 (induction in late stationary phase) includes genes that play a potential role in the formation of the holdfast substance. So far, little is known about the holdfast substance of R. baltica and thus the 185 genes of unknown function could serve as a starting point to learn more about the organisms' ability to attach. Three sulfatases were also assigned to cluster 3 (induction in late stationary phase), emphasizing the proposal of an additional function which is connected with the cell structure formation.

Interestingly, four of the nine polymerase-associated genes in cluster 11 (repression in late stationary phase) were gathered together with 67 hypothetical proteins. The possibility that some of these genes have a similar function should be further investigated.

Several genes involved in amino acid or sugar metabolism in cluster 20 (induced in late stationary phase) are co-regulated with 103 hypothetical proteins. These hypothetical proteins may be candidates for the completion of the amino acid and sugar metabolism pathways of R. baltica. The co-regulated hypothetical genes are hence prime targets to detect and confirm alternative ways of biochemical conversions in the laboratory.

Conclusions

In this transcriptome study, gene regulation throughout the life cycle of R. baltica was examined in order to learn more about this model organism's behaviour as well as its regulation of genes with putative biotechnological application.

This study documents transcriptomic changes concomitant with morphological changes during the exponential growth of an R. baltica culture. The lack of nutrients and a high cell density in the stationary phase imposes the activation of general stress response genes, genes for the reorganization of the genome and the cell wall, and possibly also for the formation of the holdfast substance. The results suggest that a subset of the 110 sulfatases may be involved in the formation of planctomycete-specific morphological features (e.g. cell wall components or holdfast substance), rather than in the degradation of sulfated carbon compounds.

Proteins of unknown function constitute a major part of the regulated genes in all growth stages of R. baltica. Their correlation to potential biotechnologically relevant genes, their expression at specific growth stages and their correspondence with certain morphotypes provide some preliminary insights into their function. In the future, next generation sequencing technologies will be applied to confirm the results presented here and to complement the data with information about regulatory RNA.

Our results provide a substantial foundation for further studies on the transcriptomic and metabolomic elucidation of the unknown mechanisms behind the lifestyle, cell cycle and cell division of R. baltica. Furthermore, the insights they provide form a basis for more directed experiments.

Experimental procedures

Cultivation of Rhodopirellula baltica cells

Cells of R. baltica were grown in batch cultures (500 ml) with mineral medium containing 10 mM glucose as sole carbon and 1 mM ammonium chloride as nitrogen source (Rabus et al., 2002). The R. baltica cultures were incubated at 28°C on a rotary shaker. Two independent batch cultures (biological replicates) per time point were inoculated and were treated separately over the whole experimental processes.

Preparation of samples for hybridizations and microarray processing

Cultures were harvested independently by centrifugation (Beckman Coulter™ Avanti™ J-20XP, JA10 Rotor, 20 min, 6000 r.p.m., 4°C), after different incubation times (44, 62, 82, 96 and 240 h) for expression profiling and microscopic analysis. The culture broth was collected in 500 ml of tubes and swirled briefly in an ethanol/dry ice bath to rapidly cool down the cultures and ‘freeze’ the RNA profile. Subsequently, the broth was centrifuged at 6000 r.p.m. for 20 min at 4°C (Beckman Coulter™ Avanti™ J-20XP, JA10 Rotor). The pellets were re-suspended in 0.1 M Tris-HCl and then re-centrifuged to obtain cell pellets that were subsequently shock-frozen in liquid nitrogen and stored at −80°C. Total RNA isolation and cDNA synthesis were performed as previously described (Wecker et al., 2009). For hybridization at least 2 µg of Alexa 547 dye-labeled (PlatinumBright: Nucleic Acid Labelling Kit – 547 Red/Orange, Kreatech, Amsterdam, the Netherlands) and 2 µg of Alexa 647 dye-labelled total cDNA (PlatinumBright: Nucleic Acid Labelling Kit – 647 Infrared, Kreatech, Amsterdam, the Netherlands) were combined and taken up in a final volume of 100 µl of DIG Easy Hyb hybridization solution (Roche Diagnostics, Mannheim, Germany). After the blocking step, the sample solution was applied to the arrays, denatured at 95°C for 3 min and hybridized under stringent conditions at 52°C for over 12 h. After hybridization, slides were washed at room temperature in ULTRArray Low Stringency Wash Buffer (Ambion, Austin, USA) and dried by N2.

In two independent hybridizations cDNA samples were hybridized according to Fig. 1. Dots and arrowheads represent samples labelled with Alexa 547 and Alexa 647 respectively. The data shown are based on the analysis of the two technical and biological replicates performed for each of the conditions.

For the microarrays a set of oligonucleotides corresponding to the whole genome of R. baltica SH 1T (Pirellula AROS Version 1.0) was purchased from Operon (Cologne, Germany) and diluted to 20 µm concentration in Micro Spotting Solution Plus spotting buffer (Telechem Sunnyvale, CA, USA). Spotting was done in three replicates onto GAPS II aminosilane slides (Corning, Schiphol-Rijk, the Netherlands) using a SpotArray 24 spotting device (Perkin Elmer, Wellesley, MA, USA). Post-processing and blocking of the slides were done according to the manufacturer's instructions. Blocking, hybridization and washing were carried out in an automated hybridization station HS400 (Tecan, Crailsheim, Germany). A more detailed description of the array can be found at the NCBI's Gene Expression Omnibus database (GEO, 2009) under accession Platform number GPL7654 and in Wecker and colleagues (2009).

Signal detection and data analysis

Slides were scanned at a resolution of 5 µm using a ScanArray Express Microarray scanner (Perkin Elmer). The image analysis software provided with this scanner was used for automatic spot detection and signal quantification. Raw data were automatically processed using the microarray data analysis software tool MADA (MADA, 2010). Series of probe replicates were tested for outliers using the outlier test implemented in MADA. First of all, only signals were considered as positive if mean spot pixel intensity was higher than local mean background pixel intensity plus twice the standard deviation of the local background. Only those spot signals were further processed, for which all three replicates, which are present for each gene on the whole genome microarray, were considered as a valid, positive signal. An outlier test was performed for the three replicates. In this procedure, the standard deviation for all replicates is calculated. Next, one of the replicates is left out and the standard deviation is computed once more. In case the newly calculated standard deviation differs more than 50% from the former one, the replicate is flagged as an outlier and removed from the data set. This procedure is rerun for all replicates. Subsequently, the mean of the remaining replicates is computed. Expression is described through the ratio (R) and intensity (I), where R is the fluorescence log ratio of one experimental time point over the other [e.g. R = log2 (result of channel for sample 44 h/result of channel for sample 62 h)] and I is the log mean fluorescence intensity [e.g. I = log10 (result of channel 44 h × result of channel 62 h)].

LOWESS normalization was applied on an R-versus-I plot with a smoothing factor of 0.5. The expression data (ratio) of all hybridizations (array replicates) were averaged and combined to one expression data point (ratio).

An expression ratio greater than 1.5 or less than −1.5 was used as a threshold for significant changes. The threshold was determined in a self-self-hybridization experiment. Reference sample material was labelled with both dyes and hybridized onto the same array. All steps were carried out as described above. The resulting signals were analysed and the log ratio was calculated. In theory, all log ratios should be zero in this experimental set up. However, due to noise the log ratios usually differ from zero. The cut-off was determined in an R-versus-I plot (data not shown).

Cluster analysis and database

All genes of the genome and their gene expression values were clustered using the k-means clustering approach [Euclidean distance metric, k = 25 clusters and nine iterations (max. 500)] with the software tool Multi-experiment Viewer MeV Version 4.0.2 from the TM4 microarray software suite (Saeed et al., 2003). The genome of R. baltica was automatically re-annotated based on updated homology searches [June 2005 – MicHanThi (Quast, 2006)]. The updated annotation including all tool results are publicly available (Quast, 2009). JCoast (Richter et al., 2008) was used as a tool for the visualization, interpretation, statistic (COG-assignment) and comparison of genomic data stored in GenDB V2.2 (Meyer et al., 2003). Venn diagrams were generated by BioVenn (Hulsen et al., 2008).

A detailed description of the expression data is available from the GEO database under the accession Series number GSE19405.

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

We would like to thank Andreas Ellrott for excellent support concerning the microarray technology and Sebastian Rokitta for excellent technical assistance. The setup of the microarray and transcriptomic analysis of R. baltica was supported by the European Commission contract GOCE-CT-2004-505403 and the Max Planck Society.

Ancillary