Expression of the pilin genes pilAB is affected by PilRS and regulated in a population density-dependent manner

The co-transcribed genes pilA and pilB are crucial for the formation of type IV pili [18]. Upstream of pilAB, the two component regulatory system PilSR is encoded [13], with PilR being a transcriptional regulator that likely activates pilAB expression [20] upon phosphorylation by PilS or another sensor kinase. Examination of the nucleotide sequence upstream of pilA [18] revealed a putative promoter sequence for an alternative sigma factor, σ⁵⁴ (RpoN1 or RpoN2). To determine the transcriptional start site of pilAB, primer extension analysis was performed (Figure S1A). The signals indicated a transcriptional start site at the G12 downstream of the putative RpoN-dependent promoter, confirming the utilization of this promoter sequence.

In order to analyze the expression level of pilAB by transcriptional reporter gene fusions independent of plasmid copy numbers, we constructed a chromosomal pilAB::uidA fusion strain (BH72::pJBLP14) (Table S1). In the wild type in complex medium (VM-ethanol), the expression increased significantly (2.6-fold, P<0.0001) at high population density (Figure 1A, first and last time point). In addition to cell density, conditions of carbon starvation affected pilAB::uidA expression. GUS activity of cells from a pre-culture in complex medium on ethanol (VM) did not change significantly after transfer to synthetic medium with or without carbon source (Figure S1B); however after one hour of aerobic incubation, expression increased almost 2-fold (1.85, P<0.0001) without C-source, but did not change when a carbon source was present (potassium malate).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Population density-dependent expression of pilAB of Azoarcus sp.

(A) Population density-dependent induction of pilAB gene expression in Azoarcus sp. BH72 wild type background and BHΔpilS mutant background. Cultures were grown in liquid aerobic culture (VM-Ethanol), and samples were taken at certain time points to measure β-glucuronidase activity (pilAB::uidA-fusion, left axis). The optical densities at these time points are shown in the inlay. The results are representative of three independent experiments. Error bars indicate standard deviations. Increase of the expression levels at exponential in comparison to stationary growth phase (last time point) was significant for wild type (BH72::pJBLP14, black triangles, P<0.001), and highly significant for pilS mutant cells (BHDpilS::pJBLP14, black squares, P<0.0001, unpaired t-test). (B) PilA protein abundance in Azoarcus sp. BH72 wild type and mutant background. Stationary phase cultures (OD 2.5) of Azoarcus wild type (wt) and ΔpilS mutant cells were compared. Western blot of whole-cell protein extracts with antiserum against PilA. Equal amounts of protein were loaded (8 µg).

https://doi.org/10.1371/journal.pone.0030421.g001

To elucidate whether the sensor kinase PilS affected the expression pattern, an in-frame pilS deletion was constructed that was not likely to cause polar effects (Supporting Information); this mutant Azoarcus sp. BHΔpilS::pJBLP14 carried the same chromosomal pilAB::uidA fusion as above. The GUS activity was not abolished in the pilS deletion mutant, in contrast the expression increased more strongly at high cell densities, 10- to 11-fold (P<0.0001) (Figure 1A, first and last time point). A similar response was observed for GFP fluorescence as obtained by experiments with gfp reporter strains (data not shown). This suggested that PilS had a strong negative effect on the pilAB expression under these conditions. The result was confirmed by Western blot analysis; immunodetection of PilA showed a strong increase in whole-cell pilin abundance in the ΔpilS mutant in comparison to wild type in the stationary phase (Figure 1B).

In contrast, the induction of pilAB expression upon carbon starvation was abolished in the pilS deletion (Figure S1B). Thus, PilS appeared to be involved in sensing the absence of carbon sources in the medium, but seemed to act negatively on density-dependent pilAB induction.

To analyze the role of PilS further, a point mutant BHpilSM was generated in which the conserved histidine residue in the HisKA-domain that was putatively required for phosphorylation was exchanged against arginine (His314Arg) to prevent phosphorylation. Autophosphorylation of an overexpressed truncated PilS variant lacking the transmembrane helices occurred in vitro in presence of radiolabeled ATP, while the point mutant indeed failed to be phosphorylated (not shown). The Azoarcus mutant BHpilSM carrying the chromosomally integrated pilAB::uidA fusion (pJBLP14) showed an approximately 4-fold decrease in expression in comparison to wild type at early exponential phase (1230±320 Miller units or 4615±520 Miller units, respectively, P<0,0001, unpaired t-Test), a ratio which was retained also in stationary growth phase. This suggested that PilS might act as bifunctional sensor kinase acting negatively on PilR phosphorylation and pilAB expression when not phosphorylated.

Population density-dependent induction suggested that pilin gene expression in Azoarcus sp. strain BH72 might be under the control of “quorum sensing”-like mechanisms. Therefore, we tested whether cell-free supernatants of stationary phase cultures (conditioned supernatants) could induce pilin gene expression in exponentially grown test cells at low cell densities. To exclude putative effects of carbon starvation, ethanol was added to the supernatants. An elevated expression was not observed in the wild type background, where pilAB expression was not significantly induced (1.2-fold). However, in the ΔpilS background the expression of pilAB::uidA increased: after one hour of incubation with conditioned supernatant, the GUS activity was almost unchanged (1.3 fold±0.22), whereas an increase was detected after two hours (1.6 fold±0.06). After four hours, the gene expression of pilAB was induced 2.4-fold (±0.32; P<0.0001) compared to the negative control with medium. Further incubation with conditioned supernatant did not lead to increased GUS activity (data not shown). Although the induction values were relatively low, this observation suggested that the conditioned culture supernatant from Azoarcus sp. BH72 had an impact on the pilAB gene expression within few hours, likely due to unknown signal molecule(s) or metabolites accumulating in the bacterial conditioned culture supernatant.

pilAB expression is regulated via molecules different from AHL-based or well-known signaling systems

Based on the results above we developed a supernatant bioassay for Azoarcus sp. BH72. Briefly, an aliquot of cells of the reporter strain BHΔpilS::pJBLP14 growing exponentially on liquid VM-ethanol medium was transferred to either fresh VM-ethanol medium, or to conditioned supernatant produced by Azoarcus wild type. Conditioned supernatants were supplemented with 3 ml ethanol per liter to overcome possible carbon source depletion during previous growth. The induction of pilAB expression was not altered when ethanol was not supplemented (data not shown). After four hours of aerobic incubation, pilAB::uidA expression of washed cells was quantified in GUS assays. This experimental setup was used for mutant studies below.

The bioinformatic analysis of the Azoarcus sp. BH72 genome revealed no evidence for genes encoding proteins for known quorum sensing systems, such as homologues of LuxIR/LasIR or LuxS [13]. Therefore, the genome was screened for predicted protein domains that are relevant in known quorum sensing systems. The proteins with best matches were chosen for further analysis, albeit E-values were rather poor. The conserved hypothetical protein Azo3178 carried an autoinducer synthethase (LasI) [21] domain (Pfam PF00765, E-value 9.40e⁻⁰³); the conserved hypothetical protein Azo1746 a peptidase C39 motif (Pfam PF03412, E-value 5.10e⁻⁰¹) typical for ABC-type bacteriocine exporters responsible for the export of signaling peptides [22]; proteins Azo0390 and Azo3379 contained a lactamase B domain (for Azo0390 E-value 2.00e⁻²⁹, for Azo3379 E-value 6.40e⁻²⁴) putatively involved in quinolone (PQS) signaling [23]. All four genes were separately inactivated in strain BH72 by directed plasmid insertional mutagenesis (see Table S1), and the mutants analyzed for production of a pilAB-inducing conditioned supernatant. The pilAB::uidA expression in Azoarcus supernatant bioassays was still significantly induced (P<0.01) by supernatants of all mutants (Table S2), indicating that autoinducer production was not affected by any of the mutations in the candidate proteins.

To further exclude the influence of AHL on pilAB induction, conditioned culture supernatant was extracted with dichloromethane, and the extracts were tested in an AHL plate detection assay and in a supernatant bioassay. The AHL-producing strain Chromobacterium violaceum CV017 was used as a positive control. For the AHL detection assay three different AHL monitor strains were used: the GFP-based sensor strains described above, or C. violaceum CV026 as a reporter for short chain AHL (C4–C8) and cyclic dipeptides. Only dichloromethane extracts of the positive control strain, Rhizobium sp. NGR234 producing AHL (3-oxo-C8-HSL) [24], yielded a strong signal for short chain AHLs in both reporter strains. As expected, the extracts of Azoarcus yielded no signals (data not shown).

To test whether the putative signaling molecule is hydrophobic, conditioned supernatants were extracted with dichloromethane and subjected to the Azoarcus supernatant bioassay. The inducing effect of the untreated conditioned supernatants, the supernatants after extraction with dichloromethane, and the dichloromethane extract were tested for induction of pilAB::uidA expression in comparison to fresh medium. Inducing activity was not found in the dichloromethane extract, but remained in the conditioned culture supernatant (Table S2). Thus, in contrast to AHLs and most other known autoinducers, the molecule(s) was apparently not sufficiently hydrophobic to be extracted by the solvent. As gene induction was also retained when the conditioned supernatant was passed through an ultrafiltration membrane (size exclusion 1000 Da), the inducing factor appeared to be a small hydrophilic molecule (hydrophilic signal factor, HSF).

Transcriptomic studies reveal that around 8.0% of all Azoarcus genes are differentially regulated in conditioned supernatant

We designed genome-scale experiments to identify additional genes that might be regulated in a manner similar to pilAB, of which some might show higher induction values. Characterization of such genes would be instrumental for further genetic or biochemical studies on a putative cell signaling system in Azoarcus sp. strain BH72. A whole genome microarray approach was carried out to compare gene expression under standard growth conditions in comparison to conditioned culture supernatant. Briefly, the change of gene expression in wild type cells was monitored after one hour and four hours incubation of cells at low population density in conditioned supernatants obtained from Azoarcus wild type. Genes were considered as being differentially expressed if the differential expression level between the tested conditions was at least 1.8 fold and the P-value≤0.05, as for many genes significant values could be obtained already at this low induction value. Expression of 7.7% of all Azoarcus sp. BH72 genes was found to be influenced by the conditioned culture supernatant containing the unknown molecule(s). This indicated that density-dependent regulation may be an important, global gene regulation process in this grass endophyte. Which portion of the differentially regulated genes was affected by a putative cell-signaling system or by side effects, as the conditioned culture supernatants were obtained from Azoarcus cultures from the stationary growth phase which might contain metabolites or antibiotics, is not clear at this stage. From all regulated genes, the mRNA abundance of 157 genes (3.9%) was increased, whereas 150 genes (3.8%) showed decreased mRNA abundance. Out of the up-regulated genes detected by microarray analysis, 39 genes were found to be activated after one as well as four hours of incubation of Azoarcus with conditioned culture supernatant, while 36 or 82 genes were up-regulated only after one or four hours, respectively. From all down-regulated genes, 35 genes showed repression after one and four hours of application, while 25 or 90 were down-regulated only after one or four hours, respectively. The detailed results of all differentially regulated genes detected by the transcriptional profiling are listed in Table S3.

It would be expected that nearly all genes in the same operon show differential gene expression in a microarray approach. Indeed, for a number of gene clusters differential regulation was detected for several genes, four typical clusters being depicted in Figure S2. The atp-operon (C), coding for the subunits of ATP synthase, as well as the nuo-cluster (A) encoding the NADH-ubiquione oxidoreductase chains were repressed at both time points. The nap-genes (D), coding for the subunits of the periplasmic nitrate reductase complex, were activated, although cells were cultivated under aerobic conditions with vigorous shaking, so that secondary effects of oxygen supply are not to be expected. Several genes that were down-regulated encoded ribosomal proteins and proteins involved in general translation processes. Those genes have different localization sites in the Azoarcus genome, and only one large cluster (B) is shown. This cluster was delimited with three putative terminator structures, leading to four probable transcripts from which only three were differentially regulated, spanning genes from azo3390 to azo3422, from azo3425 to azo3428 as well as from azo3429 to azo3431.

Comparative proteomic studies support the global gene expression approach

To identify putative targets of density-dependent regulation also at protein level, the proteome of Azoarcus sp. BH72 grown in conditioned supernatant for four hours versus early exponential growth was compared. Two-dimensional gel electrophoresis was performed, and the protein patterns were compared with the Image master software package for 2D gel electrophoresis. Changes in protein spot intensity were determined from mean intensity values of four parallel gels each and only changes in intensity of at least 2.5 fold were taken into consideration. These experiments showed that 18.0% of all detected protein spots in the 2D-gels were differentially regulated; around 14.0% showed decreased intensities (49 protein spots) whereas 4.0% (12 protein spots) appeared to be present with higher intensities upon incubation in conditioned supernatant. Forty-four protein spots could be identified by mass spectrometry (Figure 2). Detailed protein parameters such as Gravy values, subcellular localization as well as occurrence of signal peptides and transmembrane domains and mass spectrometry data can be found in Table S4. In general, the proteomic approach supported the microarray data (see Table S3 and S4). Unfortunately, PilA could not be detected on the 2D protein gels, but with a theoretical pI of 9.7 and a mass of only 6403 Da PilA would not be expected in the analytical window employed [20].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Protein pattern of Azoarcus sp. BH72 grown aerobically under standard conditions.

Encircled spots have been identified by MALDI-TOF-MS. Proteins marked with blue circles showed increased intensity in cultures grown in conditioned supernatant, red circles indicate proteins with decreased levels.

https://doi.org/10.1371/journal.pone.0030421.g002

The combination of transcriptomic and proteomic approaches revealed that exposure of exponentially growing Azoarcus sp. BH72 to conditioned culture supernatant had a strong impact on gene expression and protein synthesis. Altogether 440 genes or proteins of Azoarcus sp. BH72 were found to respond in expression or abundance, detailed results of all regulated genes and proteins are depicted in Figure S3. Several genes that were regulated in a cell density-dependent manner could be shown to display differences in the level of the corresponding proteins as well (Figure S3). For eight of the proteins that showed lower levels in the 2D-gels (encoded by azo0086, azo0156, azo0718, azo0754, azo1062, azo1280, azo2396, azo3419) in samples with conditioned supernatant, the mRNA levels were reduced as well. However, some discrepancies between the two approaches appeared: The genes azo2062 and azo3896 (only at 1 h) were found to be up-regulated in the transcriptome approach, whereas the corresponding proteins showed decreased levels in conditioned supernatants in the protein gels. Discrepancies might be caused by protein binding predicted for the protein domain FKBP_C in Azo2062, or to the timing of analysis (4 h for proteome).

Quantitative PCR studies validated data obtained by microarray

Selected results obtained by the microarray approach were further examined by real-time RT-PCR. The expression of seven genes that showed differential expression in conditioned supernatant compared to exponentially growing cells was analyzed with 16S rRNA as it is often chosen as a reference [25], [26], [27], [28]. The data of this real-time PCR analyses (Table 1) corroborated the microarray results, that the genes azo0156, coding for the δ-subunit of ATPase, and azo3412, encoding the ribosomal protein L22P, were indeed down-regulated in conditioned supernatant with the factors −2.8±0.8 and −2.4±1.3, respectively. Moreover, the genes azo0673 (11.5±4.6), azo3674 (62.4±33.5), azo3868 (44.9±6.2) as well as azo3874 (40.2±2.5) were found to be up-regulated under the mentioned growth conditions. Only gene azo3294 (1.0±0.1) could not be shown to be differentially expressed by real-time PCR studies, whereas its expression was highly up-regulated and well detectable in the microarray approach.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Differential gene expression of Azoarcus sp. BH72 upon incubation in conditioned supernatant as detected by real-time PCR and microarray transcriptome studies.

https://doi.org/10.1371/journal.pone.0030421.t001

The induction factors obtained by microarray and real-time experiments showed some quantitative differences. This observation was not unexpected since a bias towards underestimating the magnitude of mRNA change has previously been described for oligonucleotide microarray data. Moreover, the fold changes of highly expressed genes and genes expressed at low levels are difficult to compare with the different methods applied [29], [30]. The direction of regulation mostly coincided between the microarray and quantitative RT-PCR approach. Accordingly, the microarray approach used here was suitable for monitoring gene expression changes in Azoarcus sp. BH72.

Cellular processes such as type IV pili formation and regulation, nutrient transport and energy metabolism are affected by incubation in conditioned supernatant

To gain better insights into the functions of genes and proteins differentially regulated, COG-categories according to Tatusov et al. [31] can be used for assignment of gene products (Figure 3). Several cellular processes were affected by incubation in conditioned supernatant in the grass endophyte. In general, energy production and conversion (C) as well as translation, ribosomal structure and biogenesis (J) were repressed. In contrast, genes for a variety of hypothetical proteins (R, S and no COG) and signal transduction mechanisms (T) appeared to be activated after incubation with conditioned culture supernatants. They are summarized in Table S5; particular processes are given in more detail below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Distribution of differentially regulated proteins and genes of Azoarcus sp. BH72 according to COG categories.

Differentially regulated proteins detected by two-dimensional gel electrophoresis are shown with grey bars and mRNAs identified by microarray with black bars. Activation in conditioned supernatant is indicated with “up” and repression with “down”. A: RNA processing and modification, B: Chromatin structure and dynamics, C: Energy production and conversion, D: Cell cycle control, mitosis and meiosis, E: Amino acid transport and metabolism, F: Nucleotide transport and metabolism, G: Carbohydrate transport and metabolism, H: Coenzyme transport and metabolism, I: Lipid transport and metabolism, J: Translation, K: Transcription, L: Replication, recombination and repair, M: Cell wall/membrane biogenesis, N: Cell motility, 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, U: Intracellular trafficking and secretion, no: not in COG. Absolute protein counts are shown.

https://doi.org/10.1371/journal.pone.0030421.g003

Many proteins that are involved in translation processes were differentially expressed, among them the bacterial translation initiation factor 3 (infC), elongation factors such as Efp, Tsf, TufA and TufB. Fourty-eight out of 53 genes coding for ribosomal proteins were found to be repressed according to the Azoarcus sp. BH72 microarray. Concordantly, the inhibitory influence of the conditioned culture supernatant was obvious for energy metabolic processes as for example all subunits of ATP synthase were down-regulated. Furthermore, almost all factors from the Nuo-cluster responsible for the formation of the NADH dehydrogenase complex in the respiratory chain were found to be repressed, and expression of genes for five electron transfer flavoproteins (EtfA1, EtfB1, EtfB2, EtfB3 and Etf1) were also found to be negatively affected. However, genes for specialized RNA-polymerase sigma factors, sigma-38 (RpoS) and sigma-24 (AlgU), were around 2.0-fold up-regulated; genes for regulators related to metal and iron uptake, such as the nickel responsive regulator (nikR), the phosphate regulon sensor proteins PhoR as well as the phosphate uptake regulator PhoU, and the ferric uptake regulator (azo0644) were around 2- to 3-fold up-regulated according to the microarray approach. Also other genes related to iron were differentially expressed: the genes for bacterioferritins Bfr1 and Bfr2 and the bacterioferritin-associated ferredoxin Bfd were activated, and three TonB-dependent receptor genes were differentially expressed (azo2156, azo2396 and azo3023).

Related to the first known target pilA, eight genes localized in the Azoarcus sp. pil-clusters were found to be activated, encoding the putative type IV pili biogenesis proteins Azo1608, PilY1A and PilW, the prepilin like proteins Azo2180 and PilV, the type IV pilus assembly protein PilX, as well as the twitching motility protein PilU1. As expected, pilA was also detected among the regulated genes in the microarray with a fold expression of 2.9 after four hours incubation with conditioned culture supernatant.

Striking was the high amount (70) of genes affected in expression that encoded proteins belonging to the group of (conserved) hypothetical proteins or proteins that were poorly characterized. Some of those proteins were highly regulated, such as the conserved hypothetical membrane protein Azo2876 and the hypothetical secreted protein Azo0456 that showed 6.8-fold as well as 13.1-fold gene expression after four hours of incubation with conditioned culture supernatant, respectively. Moreover, the gene expression of azo1684 was 10.9-fold up-regulated according to the microarray approach. The gene product shows a domain that is required for attachment to host cells in Agrobacterium tumefaciens and is therefore of high potential interest in the plant-associated bacterium Azoarcus sp. BH72.

Among 16 other (conserved) hypothetical secreted proteins whose gene expression was activated, Azo0275, Azo0347 and Azo3784 showed only poor sequence similarities to known bacterial proteins.

The differential expression of four additional genes validated in response to conditioned supernatant

The transcriptome profiling now led to the detection of new conditioned supernatant- regulated target genes, in addition to pilAB, allowing analysis of possible similarities of regulation patterns for different genes. Several up-regulated genes encoding proteins of different functions were selected. Up-regulated was the expression of the gene azo3874, encoding a conserved hypothetical secreted protein for which the Glo_EDI_BRP_like domain suggests functions related to glyoxylase I (catalyzing the glutathione-dependent inactivation of toxic methylglyoxal) or to antibiotic resistance proteins. Moreover, the genes azo2876, coding for a conserved hypothetical membrane protein, azo1684, encoding a conserved hypothetical protein, and azo1544, coding for a regulatory GGDEF/EAL/PAC/PAS domain-containing protein, were selected. For more detailed expression studies applying transcriptional reporter gene fusions, the plasmid integration mutants Azoarcus sp. strains BHazo1544, BHazo1684, BHazo2876 as well as BHazo3874 were constructed. These strains carry a transcriptional gfp::uidA-fusion to the target genes, in strains BHazo1544 and BHazo3874 the target genes are additionally inactivated. Upon incubation with conditioned supernatant from Azoarcus wild type, azo1544, azo1684, azo2876 and azo3874 gene expression was significantly induced after four hours compared to the negative control with VM-ethanol medium (Figure 4B, C, D, E and F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Expression of transcriptional gene fusions in conditioned supernatant.

(A–F) Gene expression determined by ß-glucuronidase activities from respective transcriptional reporter gene fusions with uidA, in VM-ethanol medium. (A) Induction of pilAB gene expression in BHΔpilS::pJBLP14 by supernatants of Azoarcus sp. BH72, Azoarcus communis, Chromobacterium violaceum, Pseudomonas stutzeri and Azospirillum brasilense obtained by supernatant bioassays after four hours of incubation. (B, C, D, E) Induction of azo1544 (B), azo1684 (C), azo2876 (D) and azo3874 (E) gene expression in the respective strains by supernatants of Azoarcus sp. BH72, Azoarcus communis and Azospirillum brasilense obtained by supernatant bioassays after four hours of incubation. (F) Fold changes of induction for all tested genes (pilAB, azo1544, azo1684, azo2876 and azo3874) in comparison to each other. For all experiments, fresh medium was used instead of supernatant as negative control, and the values were set to one for calculation of fold changes. Standard deviation was calculated from at least three independent experiments. Stars indicate significance (at least P<0.05) as determined by unpaired t-test analyses.

https://doi.org/10.1371/journal.pone.0030421.g004

The supernatant bioassays were routinely carried out with conditioned culture supernatants obtained from stationary cultures of Azoarcus sp. BH72 grown in VM-ethanol medium. However, this complex medium containing yeast extract and peptone might handicap analytical analyses for identification of the unknown signaling molecule(s). Therefore, growth of Azoarcus sp. BH72 was tested in the minimal medium SM-ethanol. In this medium the generation time of the wild type and the mutant BHazo3874 was higher, however, cultures reached the stationary phase after overnight incubation at 37°C. The generation time of Azoarcus sp. wild type in VM-ethanol medium was 1.5±0.03 hours compared to 2.2±0.02 hours in minimal medium, whereas strain BHazo3874 showed a generation time of 2.2±0.15 hours in SM-ethanol and 1.6±0.16 hours in complex medium. Bioassays with conditioned culture supernatants harvested from cultures grown in the two different media were performed with the reporter strains Azoarcus sp. BHΔpilS::pJBLP14 and BHazo3874 (data not shown). Expression of both genes was induced in presence of conditioned culture supernatant obtained from Azoarcus sp. cultures grown in complex VM-ethanol medium as well as grown in synthetic SM-ethanol medium.

The signaling molecule(s) in conditioned supernatant is widespread among bacteria

The inducing ability of conditioned cell-free culture supernatants from different bacterial species and genera was tested using the Azoarcus sp. BHΔpilS::pJBLP14 reporter strain system. The soil-borne species Azoarcus evansii did not produce active supernatant (data not shown), whereas supernatants of the Kallar grass-associated strain Azoarcus communis induced the pilAB::uidA gene expression 2.0-fold (5663±727 Miller Units, P 0.0071; Figure 4A). Other tested plant-associated or species of Proteobacteria did not lead to significant induction, such as Azospirillum lipoferum, the endophytes Azonexus fungiphilus and Azovibrio restrictus, and Azotobacter vinelandii (data not shown). The phytopathogenic bacterium Chromobacterium violaceum even led to significant inhibition of expression, presumably by antagonistic effects (Figure 4A), whereas supernatants from P. syringae and X. oryzae did not lead to a significant change in pilAB gene expression (data not shown). Surprisingly, the incubation with supernatants from Pseudomonas stutzeri strain DSM4166 isolated from the rhizosphere of Sorghum mutans led to induction of the Azoarcus sp. pilAB gene expression (1.8-fold±0.62, 5451±1377 Miller Units, P 0.0171; Figure 4A). The incubation of the reporter strain Azoarcus sp. BHΔpilS::pJBLP14 with supernatants from Azospirillum brasilense also enhanced the pilAB gene expression 2.1-fold (7194±1275 Miller Units, P 0.0004; Figure 4A). These observations indicate that a similar inducing molecule might be synthesized in several plant-associated bacteria from Alpha- and Gammaproteobacteria.

Gene induction by heterologous supernatant was also observed for the newly discovered target genes azo1544, azo1684 and azo2876, although only slight induction could be shown for azo1544. The incubation of Azoarcus sp. strains BHazo2876 with conditioned supernatants of Azoarcus communis or Azospirillum brasilense showed a fold change of 1.9 or 7.3, respectively (76±13 or 319±127 Miller Units, P 0.0017 and P 0.0199; Figure 4D, F), whereas the expression of the gene was 11.8-fold (P<0.0001) increased in conditioned supernatants of strain BH72. The gene expression of azo1684 was also significantly enhanced (Figure 4C, F) after incubation with supernatants from Azoarcus wild-type (5.9±2.45, 379±122 Miller Units, P 0.0022), A. communis (1.5±0.15, 168±36 Miller Units, P 0.0376) as well as A. brasilense (1.9±0.28, 319±127 Miller Units, P 0.0075).

Gene azo2876 up-regulated in conditioned supernatant is expressed in rice roots

As pilA is involved in root colonization, also some newly identified target genes that are up-regulated in conditioned supernatant might be expressed in association with the plant. We selected strain BHazo2876 for a plant experiment, as this strain showed relatively high induction levels in GUS assays (Figure 4) which might allow differentiating induced from uninduced state by fluorescence. Single cell fluorescence can only be detected in Azoarcus sp. when gfp is expressed at relatively high levels [32]. Indeed weak (Figure 5 F) or strong fluorescence (Figure 5 G) could be detected in pure cultures only in conditioned supernatant or stationary growth phase, respectively, but not in exponentially growing cells (Figure 5 E). Strong bacterial fluorescence was visible at emergence points of lateral roots (Figure 5 D) and inside epidermal cells (Figure 5 H) of roots of rice seedlings inoculated with the reporter strain BHazo2876, indicating that this gene is highly expressed during infection of rice roots, as well.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Expression of a transcriptional azo2876::gfp fusion in pure culture or during interaction with rice roots.

Phase contrast (A–C) and corresponding fluorescence micrographs (E–G), of strain Azoarcus sp. BHazo2876 expressing a transcriptional 2876::gfp fusion in pure culture, or in infected rice roots (fluorescence micrographs D, H). Cells grown in VM-ethanol medium at exponential (A, E) or stationary growth phase (C, G), or in conditioned supernatant for 4 h (B, F); all fluorescence images taken with the same setting of the video camera. (D, H) Roots of rice seedlings 13 d after inoculation; bacterial GFP fluorescence at emergence points of lateral roots (D, with close-up in right corner, and in epidermal root cell (H). Bars correspond to 7 µm (A–C, E–G), 10 µm (D) and 20 µM (H).

https://doi.org/10.1371/journal.pone.0030421.g005

Bacterial strains and growth conditions

In general, Azoarcus sp. strains (Table S1) were grown aerobically under standard growth conditions in VM-ethanol medium (per L 0.4 g KH₂PO₄, 0.6 g K₂HPO₄, 1.1 g NaCl, 0.5 g NH₄Cl, 0.2 g MgSO₄, 26 mg CaCl₂, 10 mg MnSO₄, 2 mg Na₂MoO₄, 66 mg Fe(III)-EDTA, 1 g yeast extract, 3 g bacto peptone, 6 mL ethanol, pH 6.8) at 37°C with constant shaking at 180–200 rpm. For bacterial cultivation the Erlenmeyer flasks were only filled up to 1/10 of the total volume to assure sufficient aeration.

For production of conditioned supernatants the following strains were used: Azoarcus evansii KB740, Azospirillum lipoferum Sp59b, Azospirillum brasilense Sp7, Azoarcus communis SWub3, Chromobacterium violaceum ATCC31532 (30°C), Xanthomonas oryzae PXO99 (30°C), Azotobacter vinelandii MV531 (30°C), Pseudomonas syringae DC3000 (30°C), Azonexus fungiphilus BS5-8, Azovibrio restrictus S5b2 and Pseudomonas stutzeri DSM4166 (30°C), which were all grown on VM-ethanol medium at 37°C if not mentioned otherwise.

For AHL detection assays Escherichia coli and Chromobacterium violaceum were grown aerobically on Luria-Bertani-medium (LB) at 37°C or 30°C [50], and Pseudomonas putida on modified LB medium containing 4 g NaCl per L at 30°C [51]. Rhizobium sp. NGR234 was grown on TY medium [52] at 30°C. Antibiotics used for E. coli or Azoarcus strains were tetracycline (12.5 µg per mL), chloramphenicol (12.5 µg per mL), ampicillin (150 or 30 µg per mL), kanamycin (30 µg per mL or for C. violaceum 25 µg per mL). Gentamicin (25 µg per mL) was used in the cultivation of Pseudomonas putida.

For the proteomic approach pre-cultures of Azoarcus sp. BH72 were grown in 100 mL VM-ethanol medium until the early exponential growth phase (OD_{578 nm} of 0.3). For growth under “quorum sensing”-like conditions, these cultures were incubated for four hours with four volumes (400 mL) of conditioned culture supernatant from Azoarcus sp. BH72 wild type in a 5 L Erlenmeyer flask. As controls, Azoarcus sp. BH72 was grown in 500 mL VM-ethanol medium until the exponential growth phase. Cells from three independent cultures were harvested by centrifugation (4°C, 15 min, 6,300 g), pellets were resuspended in cold PBS (14 mM NaCl, 0.27 mM KCl, 1.65 mM Na₂HPO₄, 0.15 mM KH₂PO₄, pH 7.0), centrifuged again and stored at −80°C until protein extraction. For collection of conditioned culture supernatants, containing the unknown signaling molecule, strain BH72 was grown in VM-ethanol medium at 37°C. After 24 h pre-cultures were diluted 1∶6 with fresh medium and incubated for another 24 h. The optical density at 578 nm was measured and only supernatants from cultures that reached the stationary growth phase (OD₅₇₈>1) were harvested by centrifugation with 12,857 g for 20 min at room temperature, the supernatants were supplemented with 3 mL ethanol per L and subjected to supernatant bioassays without filtration.

For the comparative transcriptomic study, Azoarcus sp. BH72 was grown until the early exponential growth phase and under “quorum-sensing”-like conditions (see above) for one or four hours, respectively. In three biological independent replicates for each condition tested, the cells from two cultures were harvested quickly by centrifugation (room temperature, 5 min, 6800 g), pellets were suspended in PBS, pooled, centrifuged again and stored at −80°C until RNA isolation.

Construction of mutants and reporter strains

The plasmids used in this study are listed in Table S1. They were constructed as follows: The transcriptional fusions of reporter genes with pilAB were generated by cloning an 1.8 kb uidA gene, coding for the reporter enzyme β-glucuronidase, or 0.7 kb gfp gene, encoding the green fluorescent protein, into the MfeI site of the plasmid pJBLP1, resulting in pJBLP14 (carrying pilAB::uidA) [20] or pJBLP1gfp (carrying pilAB::gfp).

Plasmid integration mutagenesis was used for inactivation of the genes azo0390, azo1746, azo3178 and azo3379. Truncated gene fragments (500–600 bp) starting close to the 5′ end but lacking the start codon and carrying a stop codon instead, were amplified by PCR. PCR products were cloned into the pPCR-Script™ AmpSK(+) vector and subcloned into the conjugative vector pK18mobsacB with HindIII-XbaI. The final constructs were conjugated into Azoarcus by triparental mating, and the correct integrations were confirmed by Southern blot analysis.

To study the gene expression of azo1544, azo1684, azo2876 and azo3874, a transcriptional reporter gene fusion with uidA encoding the ß-glucuronidase and in tandem gfp encoding green fluorescent protein was generated. Insertional mutants were constructed by integrating plasmids pK18GGST-1544, pK18GGST-1684pro, pK18GGST-2876pro and pK18GGSTazo3874, carrying a fusion of the promoterless reporter genes gfp and uidA to the respective Azoarcus gene, by homologous recombination into the Azoarcus sp. BH72 chromosome. This cloning strategy also inactivated genes azo1544 and azo3874.

For deletion of the pilS gene, a 0.9 kb BsaBI-NruI fragment of pJBLP231 was excised, yielding an in-frame deletion of pilS (pJBLP2311) consisting of amino acid 107–407. The 1.8 kb SmaI fragment of pJBLP23 was exchanged by the mutagenized 0.9 kb SmaI fragment of pJBLP2311, yielding pJBLP232. The 3.2 kb insert of pJBLP232 was subcloned into the EcoRI-HindIII restriction site of pK18mobsacB for sucrose selection [53]. The final construct pJBLP234 was conjugated into Azoarcus sp. by triparental mating. Transconjugants selected for single recombination with kanamycin were further selected on 6% sucrose media for plasmid deletion. PCR and Southern blot analysis confirmed the chromosomal deletion of pilS. For point mutation of the conserved histidine residue putatively required for phosphorylation of PilS, in a short 0.6 kb PstI-fragment of pilS (pUC-SC27) this histidine codon was exchanged to an arginine codon (CGC) (pUC-SC27M), and the resulting fragment enlarged by successive addition of 5′- and 3′- fragments of pilS (pUC-SC27M+3′+5′). To allow sucrose selection, the mutagenized EcoRI-HindIII fragment was subcloned into pK18mobSacB, and a double recombinant was selected after transfer into strain BH72 (BHpilSM). For reporter gene studies, the pilAB::uidA fusion was integrated into the chromosome (BHpilSM::pJBLP14).

Supernatant bioassays and determination of β-glucuronidase activity

The impact of conditioned culture supernatants from Azoarcus sp. and other species (see above) on the gene expression of pilAB, azo1544, azo1684, azo2876 and azo3874 were monitored by reporter protein (β-glucuronidase) studies using a transcriptional pilAB::uidA (Azoarcus sp. BHΔpilS::pJBLP14), azo1544::uidA (A. sp BHazo1544), azo1684::uidA (A. sp. BHazo1684), azo2876::uidA (A. sp. BHazo2876), or azo3874::uidA (A. sp. BHazo3874) fusion, respectively. These reporter strains (Table S1) were routinely grown in VM-ethanol medium at 37°C. When the cells had reached the early exponential growth phase, they were diluted with four volumes of conditioned culture supernatant and afterwards further incubated for one, two, three or four hours, respectively. As negative controls, cultures were incubated with fresh medium instead of supernatant. The conditioned culture supernatants were either harvested from wild type cultures grown in complex medium (VM-ethanol, see above) or minimal medium (SM-ethanol, consisting of modified VM-ethanol medium containing (per L) only 0.1 g NaCl, 0.1 g yeast extract, and no Bacto peptone. The production of conditioned culture supernatant was performed in 250 or 300 ml Erlenmeyer flasks with a total volume of 30 mL. The subsequent supernatant bioassays were carried out in 100 mL Erlenmeyer flasks with a final volume of 10 mL.

β-Glucuronidase activity was assayed according to [54] with modifications, in a buffer containing 60 mM Na₂HPO₄ and 40 mM NaH₂PO₄ (pH 7), 1 mM EDTA, 14 mM β-mercaptoethanol, 0.05% SDS and 2 mM p-nitrophenyl β-D-glucuronide. Reactions in 0.65 mL volumes at 37°C were stopped by adding 0.2 mL 2.5 M 2-amino-2-methylpropandiol, and the absorbance of p-nitrophenol was measured at 420 nm. β-Glucuronidase activity was calculated in Miller Units defined by: (A_{420 nm}×1000)/(t (min)×OD_{600 nm}). The GraphPad InStat software package (GraphPad software, San Diego, CA) was used for statistical analysis carried out by unpaired or paired T-tests, respectively.

AHL detection assays based on sensor strains and AHL extraction

For cross-streak experiments, the test strain Azoarcus sp. BH72 or Rhizobium sp. NGR234 as a positive control were streaked close to the GFP(ASV)-based AHL sensor strains E. coli JM105 (pJBA89) or P. putida IsoF (pKR-C12) on VM-ethanol plates to form a T. After 24 h of incubation at 30 or 37°C, the green fluorescence of the AHL sensor strains were visualized by excitation with blue light, and the results were documented with a Hamamatsu Color Chilled 3CCD camera mounted on a binocular (Olympus SZX12). For the AHL plate detection assay, single colonies of AHL monitor strains P. putida IsoF (pKR-C12) or E. coli JM105 (pJBA89) were resuspended in 0.9% sodium chloride and plated onto LB agar plates. If C. violaceum CV026 was used, 100 µL of the reporter culture was mixed with 3 mL of LB soft agar and poured on LB agar plates. The test samples were filled up in holes punched into the agar. The plates were incubated at 30/37°C for 24 h. For analysis either the GFP fluorescence was quantified in a Fluoroimager Typhoon 8600 (GE Healthcare, Freiburg, Germany), or by visual inspection for violacein production. Putative AHL molecules were extracted with one volume dichlormethane from culture supernatants (grown in VM-ethanol medium to stationary phase). The extracts were evaporated at 40°C to dryness. Residues were dissolved in 1 mL acetonitrile following volume reduction to 0.1 mL by rotational evaporation. The liquid phase and the extracts were then applied to the AHL plate detection assay and/or to the quorum sensing bioassay (described above). For size fractionation, conditioned supernatant was first passed through an ultrafiltration membrane YM10, then through YM1 (Amicon, Bedford, USA).

Infection of Oryza sativa seedlings with Azoarcus sp. BHazo2876

Rice seedlings (Oryza sativa ssp. japonica cv. Nipponbare) were surface-sterilized, germinated and inoculated with Azoarcus sp. BHazo2876 as described [15], [55] with the following modifications: Germination occurred for 3 days at 30°C in the dark, followed by 2 days at 30°C, 15 kLux light intensity, 80% humidity and 14 /10 hours day/night intervals. Plant medium [55] contained 20 mg neutralized DL-malic acid per L, and seedlings were incubated at 30°C, 15 kLux light intensity, 80% humidity and 14/10 hours day/night intervals for 13 days.

Fluorescence microscopic analysis of infected rice roots

13 days after inoculation roots were washed in 0.9% NaCl solution to remove remaining sand, separated from the shoot and rice grain. Roots were incubated for 30 min at 4°C in phosphate buffered saline to allow oxidation of GFP and placed on glass slides in 10% phosphate buffered saline, 90% glycerol and 0.25% of the antioxidant 1,4-Diazabicyclo(2,2,2)octan (DABCO) (pH 8.6 adjusted with HCl/ NaOH). For microscopic analysis an Axioplan 2 fluorescence microscope from Zeiss (Jena, Germany), equipped with a C5180 camera from Hamamatsu Photonics K.K., (Hamamatsu, Japan) was used [56].

PAGE and Western blot analysis

Sodium dodecyl sulfate (SDS)-soluble cellular proteins were extracted [15] from equal amounts of cells from liquid cultures, and subjected to Tris-Tricine polyacrylamide gel electrophoresis (PAGE) followed by Western blot analysis with anti-PilA rabbit polyclonal antibodies (1∶5000) [18].

Two-dimensional gel electrophoresis

Protein extraction, two-dimensional gel electrophoresis, and gel analysis was performed as described in [20] with 500 µg of protein extracted from Azoarcus sp. BH72. Only those proteins that showed at least 2.5-fold change in level between the two tested conditions were considered to be affected by conditioned culture supernatant.

Protein identification by MALDI-TOF-MS

Identification of proteins by MALDI-TOF-MS/MS was performed as described in Hauberg et al. 2010 [20].

RNA isolation and primer extension

For the “hot phenol” extraction method [57], bacterial cells from a 50 ml culture were resuspended in a 1∶1 mixture of prewarmed phenol-chloroform (pH 4.7) and 50 mM sodium acetate/10 mM EDTA/1% SDS (pH 5.1). Cells were incubated for 5 min at 65°C followed by 10 min incubation on ice. For phase separation the mixture was centrifuged for 10 min at 12°C and 8000 g, and the upper phase was mixed with the same volume of phenol-chloroform (pH 4.7). This phenol extraction was repeated three times followed by extraction with chloroform-isoamylalcohol (24∶1). RNA precipitation was performed with the same volume of isopropanol for 45 min on ice. The RNA was pelleted by centrifugation (12900× g, 10 min, 4°C) and washed with 70% ethanol. After drying RNA was dissolved in 1× RNAsecure (Applied Biosystems), and contaminating DNA was removed from RNA preparations by using the RNeasy Mini Kit (Qiagen) according to manufacturer's instructions. Finally, RNA was frozen in liquid nitrogen and kept at −80°C until further processing. Contaminating DNA was removed from RNA preparations by DNase I using Qiagen columns (RNeasy Mini Kit, Qiagen, Hilden, Germany) according to manufacturer's instructions.

Primer extension analysis was performed by standard procedure [58]. The primer (5′-GCAGTTTCTTCATTTCAATTCTCC-3′) annealed in pilA, 42 bp downstream of the predicted transcriptional start. It was end-labeled by ³⁵S-α-ATP, mixed with 30 µg RNA, and extended using avian-myleoblastosis-virus reverse transcriptase (Roche, Mannheim, Germany). The cDNA was purified via extraction with 50% phenol, 49% chloroform and 1% isoamyl alcohol followed by ethanol precipitation. A DNA sequencing reaction [59] with the same oligonucleotide primer was run on the denaturing gel next to the cDNA and evaluated by autoradiography.

Microarray processing and data analyses

20 µg of total RNA from one biological sample was reverse transcribed with BioScript RT in respective reaction buffer (Bioline) with amino modified random hexamers for 90 min at 42°C. For two-color labeling, the aminoallyl-labeled first strand cDNA from the two different RNA samples was coupled to fluorescent Cy3-NHS or Cy5-NHS esters (GE Healthcare), respectively. DNA contamination was excluded by PCR, and the quality of fluorescently labeled cDNA was checked on an agarose gel and with a Typhoon scanner (GE Healthcare, Fairfield, CT; USA) at 800 pmt, medium sensitivity and with the respective filters for Cy3 (green 532 nm laser, 555 BP20) and Cy5 (red 633 nm laser, 670 BP30) detection.

For construction of the oligonucleotide microarray, a collection of gene-specific 70mer oligonucleotide probes (designed and synthesized by Eurofins MWG Operon) for the 3,992 predicted protein-coding genes from Azoarcus sp. BH72 was spotted on epoxysilane-coated Nexterion Slide E (Schott) (CeBiTec, University Bielefeld). The array consisted of 17,280 single spots, spotted in rows of 20×18 per grid, which were arranged in 4×12 grids. Each gene was spotted in quadruplicates; five positive controls were spotted in additional replicates (azo1072 (30S ribosomal protein S1), azo1081 (50S ribosomal protein L35), azo2104 (30S ribosomal protein S15), azo2837 (probable glyceraldehyde 3-phosphate dehydrogenase) and azo2898 (30S ribosomal protein S16). Negative buffer controls, as well as negative controls consisting of a rice-gene targeted oligonucleotide (Os05g0438800 similar to actin1) and randomized negative control oligonucleotides (human H2NC00001, H2NC00002, H2NC00003, H2NC00004, H2NC00005, H2NC00006; mouse M2NC000001, M2NC000005, M2NC000006, M2NC000008, M2NC000009, M2NC000010, M2NC000012; customized by Eurofins) which do not match to the genome of Azoarcus BH72 were also included. Prior to hybridization, samples were denatured at 65°C for 8 min. Hybridization was performed at 42°C for 16 h in Easy hybridization solution (Roche) supplemented with 5 µg salmon sperm DNA in a final volume of 60 µL under a cover slip in a hybridization chamber in a water bath. Afterwards the slides were washed once in 2× SSC/0.2% SDS for 5 min at 42°C, twice in 0.2× SSC/0.1% SDS for 1 min each at room temperature, twice in 0.2× SSC for 1 min each at room temperature and for 1 min on 0.05× SSC at 21°C. The glass slides were dried by centrifugation (7 min, 650 g, room temperature), and the Cy3- and Cy5-fluorescence was scanned with the GenePix™ Scanner 4000A (Molecular Devices) with a pixel size of 10 µm.

Image analysis was performed with the GenePix 4.1 program, and subsequent analyses were carried out with the open-source software TM4 [60] (http://www.tm4.org/). Before intensity values that were measured with GenePix could be compared, LOWESS normalization with MIDAS v2.19 was performed. The normalized data for each spot were aligned to the corresponding Azoarcus sp. strain BH72 gene name. Three independent experiments including dye-swap were performed and average expression folds were obtained from the replicates. A one-tailed paired t-test was performed with Bonferroni correction, and only genes that showed an expression of at least 1.8 fold and a P-value≤0.05 were regarded as being differentially expressed.

The microarray data are MIAME compliant and have been deposited in the GEO database (http://www.ncbi.nlm.nih.gov/geo) under accession no. GSE25939, GPL11300.

Real-time PCR

Synthesis of cDNA was achieved by using gene specific reverse primers for genes azo0156, azo0673, azo3294, azo3412, azo3674, azo3868, azo3874 and 16S rRNA with the Verso 2-Step QRT-PCR Kit from ABgene. 30 ng of total RNA was reverse transcribed in a total volume of 20 µL for 30 min at 42°C, followed by a denaturation step for 2 min at 95°C. 0.5 µL to 1.5 µL of the cDNA were used for the quantitative PCR step with 1×2-Step QPCR Mix (ABgene) and 0.5× SYBR green I dye (Molecular Probes) in a total volume of 25 µL. The QPCR was carried out in the Chromo 4 PTC-200 real-time PCR cycler (MJ Research) with the following program: Initial denaturation and enzyme activation for 15 min at 95°C linked to a loop of 40 cycles each having a denaturation step for 15 sec at 95°C, 30 sec at 60°C for annealing followed by an elongation step for 30 sec at 72°C. The PCR was followed by a melting curve from 60°C to 99°C with 0.5°C steps for 10 sec. The 2^−ΔΔC_T method was applied for data analyses, and 16SrRNA was used as a reference with the following primers: 16SRTfor: CTTGACATGCCTGGAACCTT, 16SRTrev: ATGACGT GTGAAGCCCTACC, azo0156for: ATCAACGATCCCAAGCTTTC, azo0156rev: CG TGTTCGTTCTTCAGAGCA, azo0673for: TCAGGAGGTGGGCAACTG, azo0673rev: ACAAGAACCGCCGTCCAC, azo3294for: CACGCAAAGATGATCAGGAA, azo3294rev:TGATCTACACCCTGCTGCTG, azo3412for: GAAACGCTTGAGGGTA GTGC, azo3412rev: GCTGAACATTCTGGCCTTCT, azo3674for: AGTTCAAGGCC AAGGTGCT, azo3674rev: CGTAACGGAGTTTTCGAAGC, azo3868for: CACTCGC AGTGCCTGTACTC, azo3868rev: CCCTCGAAGTAGGACATCCA, azo3874for: CCTTCAAGTTCGAGGACGAC, azo3874rev: ACGTAGAAGGCCAGGTGATG.