OUP user menu

Initial characterization of a bolA homologue from Pseudomonas fluorescens indicates different roles for BolA-like proteins in P. fluorescens and Escherichia coli

Birgit Koch, Ole Nybroe
DOI: http://dx.doi.org/10.1111/j.1574-6968.2006.00359.x 48-56 First published online: 1 September 2006


The RpoS-regulated bolA gene in Escherichia coli is important for the decrease in cell size during stationary phase or sudden carbon starvation. A Pseudomonas fluorescens strain mutated in a gene with homology to bolA reduced its cell size upon carbon starvation, and RpoS had little effect on bolA expression. The mutant grew slower than the wild-type strain in minimal medium with l-serine as the sole nitrogen source, while growth rates were similar on a mixture of l-serine and l-cysteine. Reverse transcriptase polymerase chain reaction analysis indicated that the bolA homologue is the second gene in an operon where the two next ORFs encode putative proteins with homology to sulphurtransferases and protein disulphide isomerases. Complementation of the mutant phenotypes was only obtained by plasmids encoding BolA as well as the above two proteins. Growth phenotypes and gene homologies suggest that BolA-like proteins have different functions in E. coli and Pseudomonas.

  • bolA
  • Pseudomonas
  • sulphur metabolism


Members of the BolA protein family are widespread (Huynen et al, 2005), and genome analyses (Tatusov et al, 2001) have shown that several Proteobacteria, including Escherichia coli and Pseudomonas, have two genes encoding BolA-type proteins. However, functional studies have been carried out only for the E. coli BolA protein.

In E. coli, the bolA gene is under control of the stationary phase σ factor RpoS (Lange & Hengge-Aronis, 1991). The BolA protein is essential for a shift in cell morphology upon entry to stationary phase in minimal medium or in response to sudden carbon starvation. Furthermore, E. coli bolA is induced by several stresses, including low pH, high salinity and high temperature (Santos et al, 1999), and plays a role in E. coli biofilm formation (Vieira et al, 2004). Overexpression of bolA in E. coli leads to increased transcription of the cell wall synthetic genes dacA (PBP5), dacC (PBP6) and ampC (Santos et al, 2002). The BolA effect on cell morphology depends on PBP5 and PBP6, indicating a regulatory function, and BolA proteins have been suggested to be DNA-binding proteins serving as transcriptional regulators (Santos et al, 2002; Kasai et al, 2004). Further, the three-dimensional structures of BolA-like proteins from Mus musculus and Xanthomonas campestris pv. campestris show similarities to nucleic acid-binding proteins (Kasai et al, 2004; Chin et al, 2005).

According to Huynen (2005), the BolA protein from M. musculus is structurally most closely related to OsmC, an enzyme that reduces organic peroxides. Recently, interactions have been reported between BolA proteins of Drosophila melanogaster and Saccharomyces cerevisiae and mono-thiol glutaredoxins, a class of enzymes involved in thiol-disulphide metabolism and required for maintaining a reduced intracellular environment (Ito et al, 2001; Giot et al, 2003). Interestingly, genes encoding mono-thiol glutaredoxins are situated next to bolA genes in some bacteria and it has been suggested that BolA can serve as a reductase interacting with glutaredoxin (Huynen et al, 2005). In this paper, we report an initial characterization of a bolA homologue in Pseudomonas fluorescens.

Materials and methods

Bacterial strains and growth conditions

The bacterial strains and plasmids used in this study are described in Table 1. Escherichia coli was cultured in Luria Broth (LB; Sambrook et al, 1989) at 37°C. Pseudomonas fluorescens strains were cultured at 28°C in LB or at 20°C in Davis minimal medium (DMM; Difco Laboratories, Detroit, MI) containing 0.4% glucose and micronutrients. To obtain carbon-starved cells, cultures in DMM were harvested in the exponential phase, washed twice in DMM without carbon sources and suspended in this medium at the original cell density. Growth on various amino acids, serving as the only nitrogen source, was determined in DMM where the nitrogen sources were replaced by 20mM of l-serine, d-serine, l-cysteine, l-valine, l-glycine, l-threonine, l-aspartic acid, l-isoleucine, l-methionine, l-leucine or homocysteine (l-methionine and homocysteine did not support growth). Twenty millimoles pyruvate or 0.1mM α-ketobutyrate was added in combination with 20mM l-serine. Cultures were inoculated to an initial OD600nm of 0.01–0.05 with overnight cultures grown in DMM. All experiments were repeated at least twice. Carbon source utilization profiles were determined using the 95 different C sources of the Biolog GN system (Biolog Incet al, Hayward, CA). Antibiotics were used at the following concentrations (in μgmL−1): ampicillin, 100; kanamycin, 25; and gentamicin, 10.

View this table:
Table 1

Plasmids and strains used in this study

Plasmids and strainsRelevant characteristicsSource or reference
PRL1063Tn5 delivery plasmidWolk (1991)
pML107Translational fusion broad-range-host vector. GmRLabes (1990)
pUX-BF13R6K replicon-based helper plasmid, providing the Tn7 transposition function in trans. AmpR, mob+Bao (1991)
pUCP24E. coliPseudomonas shuttle vector. GmRWest (1994)
pBluescript SKCloning vector. AmpRStratagene
pBK-miniTn7-ΩGmpUC19-based delivery plasmid for miniTn7-ΩGm. AmpR, GmR, mob+.Koch (2001)
p11D1Contains Tn5 and flanking chromosomal DNA from DF57-11D1. KmRThis study
pBKP32.3kb XbaI–EcoRI fragment from p11D1 in pBluescript. AmpRThis study
pBMB36.1kb SalI–EcoRI fragment from p11D1 in pBluescript. AmpRThis study
pML107Pst2.0kb PstI fragment from pBKP3 in pML107. GmRThis study
pCB10.8kb NotI fragment from pBMB3 in pBK-miniTn7-ΩGm. GmRThis study
pUCP24bol10.5kb PCR amplified HindIII fragment in pUCP24 under control of the rep promoter. GmRThis study
pUCP24bol30.5kb PCR amplified HindIII fragment in pUCP24 under control of the lac promoter. GmRThis study
pUCP24Pst12.9kb PstI fragment from pBMB3 in pUCP24. GmRThis study
pUCP24Komp12.6kb HindIII–EcoRI fragment from pBMB3 in pUCP24This study
pUCP24bolKomp2.30.5kb PCR amplified HindIII fragment in pUCP24PstIThis study
E. coli
XL1-BluerecA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [FproAB, lacIqZDM15,Tn10 (tetR)]Stratagene
DF57Rhizosphere isolateSørensen (1992)
DF57-11D1Derivative of DF57, bolA::Tn5.Kragelund (1995)
Pf-5Rhizosphere isolateHowell & Stipanovic (1979)
JL3985Derivative of Pf-5, rpoS::Tn5Sarniguet (1995)
DF57GmDF57, GmR tagged with pBK-miniTn7-ΩGm. GmRThis study
DF57-11D1GmDF57-11D1, GmR tagged with pBK-miniTn7-ΩGm. GmR, KmRThis study
DF57-ABCpCB1 integrated into the chromosome of DF57. GmRThis study

DNA manipulations

Plasmid constructions were made in E. coli strain XL1-Blue (Stratagene Cloning Systems, Heidelberg, Germany) using standard E. coli techniques (Sambrook et al, 1989). The generation of Tn5-luxAB-tagged mutants of P. fluorescens strain DF57 has been described elsewhere (Kragelund et al, 1995). To clone the Tn5-tagged gene in DF57-11D1, chromosomal DNA was cut with EcoRI, ligated and electroporated into E. coli XL-1 Blue. The plasmid p11D1 containing Tn5 and flanking Pseudomonas DNA is able to replicate in E. coli because Tn5 on pRL1063 (Wolk et al, 1991) used for the construction of DF57-11D1 carries oriV.

To generate plasmids for complementation studies, ORF2 (bolA; see Fig. 1) was amplified from chromosomal DNA of strain DF57 using Vent DNA polymerase (New England Biolabs, Beverly, MA) and the primers: 5′-TCT AAC AAA AAA CAG GGG G-3′ and 5′-TTC CCT CAA TCA TCG CTG-3′. All primers used in this study were delivered by TAG Copenhagen (Copenhagen, Denmark). The PCR fragment was cut with HindIII, yielding a 512bp fragment, which was cloned into the HindIII site of pUCP24 in both orientations, giving pUCP24bol1 and pUCP24bol3. To construct a plasmid expressing ORF3 and ORF4, a 2594bp HindIII/EcoRI fragment (see Fig. 1) was cloned from pBMB3 (see Table 1) into pUCP24 cut with HindIII and EcoRI, yielding pKomp1. To construct a plasmid expressing BolA, ORF3 and ORF4, a 2991bp PstI fragment was initially cloned from pBMB3 (see Table 1) into pUCP24, yielding pUCP24PstI. The HindIII-digested 512bp PCR fragment carrying bolA (see above) was then inserted into the HindIII-digested pUCP24PstI, yielding the plasmid pUCP24Komp2.3.

Figure 1

Gene organization and RT-PCR analysis. (a) Proposed operon structure in Pseudomonas aeruginosa based on data obtained from http://www.pseudomonas.com/. (b) Gene organization in P. fluorescens DF57 including restriction sites used for constructions of mutants and plasmids. The amino acid identity between proteins encoded by the P. aeruginosa genes and by the corresponding genes in P. fluorescens DF57 is shown. [c(1)] Amplification products expected from overlapping RT-PCR analysis of P. fluorescens DF57 total RNA. The vertical bars indicate location of primers with reference to panel (b), and the lengths of the expected PCR products are indicated. [c(2)] Agarose gel analysis of the PCR products. A, B, C and D refer to the PCR products shown in c(1). Lane 0: Molecular weight standard. All lanes labelled 1: Positive control reaction obtained with chromosomal DNA. All lanes labelled 2: Negative control reactions obtained with RNA extracts not subjected to reverse transcription. All lanes labelled 3: RT-PCR reactions obtained with RNA extracts.

For construction of an orf5 mutant, an 825bp NotI fragment from pBMB3 (encoding the first 788bp of orf5) was cloned into the NotI site in pBKminiTn7ΩGm (Koch et al, 2001), yielding the plasmid pCB1, which was electroporated into DF57 in the absence of the helper plasmid to eliminate transposition. PCR analysis of the gentamicin-resistant transformant showed that pCB1 was inserted into orf5 on the DF57 chromosome by homologous recombination. The mutant lacks the major part of the putative ATP-binding cassette (ABC) transporter encoded by orf5, and is therefore considered a knockout mutant. Strains DF57 and DF57-11D1 were labelled with gentamicin resistance at a neutral chromosomal site using the mini-Tn7 system (Koch et al, 2001) and used as control strains when examining phenotypes of the orf5-mutant.

To construct a bolA-lacZ reporter plasmid, a 1977bp PstI fragment carrying the promoter region, orf1 and a fragment of bolA encoding the N-terminal 33 amino acids was fused with lacZ in the translational lac-fusion broad-range-host vector pML107, yielding the plasmid pML107Pst (see Table 1). pML107Pst was inserted into P. fluorescens DF57, P. fluorescens Pf5 and an RpoS mutant strain of P. fluorescens Pf5, JL3985 (Sarniguet et al, 1995), by electroporation. As a negative control, pML107 was inserted into the three strains.

DNA sequencing and sequence analysis

DNA sequencing was performed using a Vistra DNA sequencer 725 and the Thermo Sequenase premixed cycle sequencing kit (Amersham Biosciences, Piscataway, NJ) as recommended by the manufacturer or carried out by GATC Biotech (Konstanz, Germany). Both DNA strands were sequenced. A combination of primer walking and sequencing (with standard primers) of various subclones of p11D1 constructed in pBluescript was used. The sequence data obtained were analysed using the University of Wisconsin Genetics Computer Group package, version 10.2 (Devereux et al, 1984), basic local alignment search tool (blast) (NCBI) (Altschul & Lipman, 1990), Pfam (Sanger Institute) (Bateman et al, 2002) and COGnitor (NCBI) (Tatusov et al, 2001). The EMBL accession number for the sequence of ORF1, ORF2, ORF3, ORF4 and the partial sequence of ORF5 is AJ243174.

The P. fluorescens Pf5 genome data (Paulsen et al, 2005) were obtained from GenBank (accession number CP00076). For the published genomes of P. putida KT2440 (Nelson et al, 2002) and P. syringae pv. tomato DC3000 (Buell et al, 2003), genome sequence data were obtained from http://www.tigr.org. The P. aeruginosa PA01 genome sequence data (Stover et al, 2000) were obtained from http://www.pseudomonas.com.

Transcriptional analysis by reverse transcriptase PCR (RT-PCR)

Total RNA was isolated from exponentially growing cultures of DF57 using the Qiagen RNeasy kit as recommended by the manufacturer. Total RNA was treated with 1U RNase-free DNase I (Amersham Bioscience) in 1 × reaction buffer (40mM Tris/HCl pH 7.5; 6mM MgCl2) and in the presence of 2UμL−1 ANTI-RNase (Ambion, Austin, TX) for 15min at 37°C. After digestion, the DNase was inactivated by the addition of EDTA to 6mM, followed by incubation at 65°C for 10min.

The RNA was reverse-transcribed in a final volume of 20μL containing 0.1μM primer, 20U ANTI-RNase, 0.5mM each of dATP, dCTP, dGTP and dTTP, 10mM dithiothreitol and 200U Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA) in a reaction buffer consisting of 50mM Tris/HCl, 75mM KCl and 3mM MgCl2. Two microliters of RNA was added to 18μL of the above mix and incubated for 50min at 42°C. The reaction was inactivated by heating at 70°C for 15min. Sequences of the primers used for reverse transcription and PCR are listed under supplementary materials. PCR was carried out with a 10-fold dilution of reverse transcriptase reaction as a template using Dynazyme as recommended by the manufacturer (Finnzymes Oy, Espoo, Finland). The PCR reactions were initially heat-denatured for 3min at 94°C, followed by 40 cycles of 52°C for 1min, 72°C for 1min 72 and 94°C for 1min, and finally 52°C for 1min and 72°C for 10min. PCR products were analysed directly on agarose gels.

β-Galactosidase assay

β-Galactosidase measurements were performed as described by Miller (1972). Exponentially growing cultures in LB were obtained by inoculation to an initial OD600nm of 0.05 with overnight cultures grown in the same medium. Each β-galactosidase measurement was based on, and results were calculated from, four replicates.

Biofilm assay

Biofilm formation by P. fluorescens DF57 and DF57-11D1 in microtitre wells (Cellstar polystyrene plates, Greiner Bio-One Gmbh, Frickenhausen, Germany) was quantified essentially according to the procedure described by O'Toole & Kolter (1998). In brief, overnight cultures in DMM or 10% tryptose soy broth (TSB, Difco) were adjusted to an OD600nm of 0.1 and incubated in microtitre wells at 28°C for up to 8h. The liquid phase was removed to determine OD600nm and the wells were rinsed twice with phosphate-buffered saline (PBS). After staining with crystal violet (1% in water) for 15min, the stained biofilm was rinsed four times with PBS and extracted with 96% ethanol. The OD600nm of the stained extract was then determined. Two independent experiments with each strain inoculated into 24 wells were conducted.

Microscopic observations

Samples of bacterial cultures were directly applied to microscope slides and observed by phase-contrast microscopy with a Zeiss Axioplan microscope at 1000-fold magnification.

Results and discussion

Sequence analysis

Pseudomonas fluorescens strain DF57-11D1 was isolated after Tn5 mutagenesis (Kragelund et al, 1995). Sequencing showed that Tn5 was inserted into a putative gene encoding a 99 amino acid ORF (see Fig. 1) with homology to BolA proteins. Groups of Proteins (COGs) represent an attempt to classify proteins encoded in complete genomes (Tatusov et al, 2001). Escherichia coli possesses two proteins belonging to the BolA protein group COG0271 (see http://www.ncbi.nlm.nih.gov/COG/old/palox.cgi?seq=BolA) i.e. BolA and the hypothetical protein YrbA. In P. aeruginosa, PA4451 shows the highest amino acid identity to E. coli YrbA (39%) and has been annotated YrbA, while PA0857 shows the highest amino acid identity to E. coli BolA (48%) and has been annotated BolA (see http://www.pseudomonas.com/AnnotationByPAU.asp?PA=PA0857). The Tn5-tagged gene in DF57-11D1 encodes a protein with 73% identity to PA0857 from P. aeruginosa, 28% identity to PA4451 from P. aeruginosa, 46% identity to E. coli BolA and 37% identity to E. coli YrbA. In accordance with the P. aeruginosa annotation we refer to this gene as bolA.

While the E. coli bolA gene is monocistronic (Aldea et al, 1989), the sequence of the region surrounding bolA in P. fluorescens DF57 shows that it is situated as the second gene in a larger cluster. Genome sequence information for P. fluorescens Pf5, P. aeruginosa PAO1, P. putida KT2440 and P. syringae pv. tomato DC3000 (Stover et al, 2000; Nelson et al, 2002; Buell et al, 2003; Paulsen et al, 2005) shows that the bolA gene is situated as the second gene in a row of seven genes with high homology between the species as shown for P. aeruginosa in Fig. 1.

The first gene in the sequenced cluster encodes a putative protein (ORF1; see Fig. 1) with an unknown function. A blast search reveals that a protein with 65–70% amino acid identity is present in several Pseudomonas strains and in Azotobacter vinelandii. No proteins from other bacteria show more than 29% amino acid identity.

Using the COGnitor (http://www.ncbi.nlm.nih.gov/COG/), ORF3 could be placed in COG1054, ‘uncharacterized enzymes related to sulphurtransferases’, a group of proteins already including PA0858 from P. aeruginosa. A rhodanese (thiosulphate-cyanide sulphurtransferase) domain could be identified in ORF3 by Pfam, a protein family database designed to identify structural domains in proteins (Bateman et al, 2002).

ORF4 (Fig. 1) corresponds to PA0859 in P. aeruginosa, which is a protein with no known function. ORF4 is related to COG3531 ‘predicted protein-disulphide isomerases’ and possesses a DsbA-like thioredoxin domain. Protein disulphide isomerases, thioredoxins and glutaredoxins all catalyse thiol-disulphide interchanges in other proteins and all belong to the thioredoxin superfamily of oxidoreductases (Hu et al, 1997; Fernandes & Holmgren, 2004).

Only 802bp of the gene encoding ORF5 with homology to PA0860 (1790bp) from P. aeruginosa has been sequenced (Fig. 1). PA0860 encodes an ABC-transporter carrying both an ABC-ATPase domain and a transmembrane domain, which is indicative of an ABC-transporter system involved in export (Holland & Blight, 1999).

To determine whether ORFs1–5 were part of the same operon, an overlapping RT-PCR analysis was carried out on total RNA from strain DF57. Primers for the analysis (see supplementary material) annealed within ORF1 and bolA, bolA and ORF3, ORF3 and ORF4 and ORF4 and ORF5, respectively, allowing the formation of PCR products A–D (Fig. 1b). The RT-PCR analysis generated the above four PCR products, showing that the five ORFs in the sequenced region are cotranscribed. Operon predictions for P. aeruginosa PA01, based on the method of Ermolaeva (2001), are available through http://www.tigr.org/tigrscripts/operons/operons.cgi. These predictions indicate that PA0856 to PA0862 (se Fig. 1) are situated in an operon in P. aeruginosa PA01. The two last genes in this putative operon are not sequenced in DF57. Both genes encode hypothetical proteins with no known function.

Cell morphology and growth phase-dependent bolA expression

Originally, the E. coli bolA gene was described as a gene that caused round cell morphology when overexpressed (Aldea et al, 1989). Later, it has been shown that E. coli bolA mutant cells fail to reduce their length when grown in minimal medium to the stationary phase or when subjected to sudden carbon starvation (Santos et al, 2002). To gain an understanding of the function of BolA in P. fluorescens, we initially compared the morphology of strains DF57 and DF57-11D1 during sudden carbon starvation after growth in minimal medium, and observed a comparable shift in cell morphology to small and round cells (Fig. 2). We then determined that the cell morphologies of DF57(pUCP24bol1) and DF57(pUCP24bol3), which overexpresses the Pseudomonas bolA gene, were similar to the wild type under these conditions (data not shown).

Figure 2

Cell morphology of Pseudomonas fluorescens during exponential growth in LB and after a shift to a glucose-free medium as observed by phase-contrast microscopy. (a) Growing cells of P. fluorescens DF57. (b) Cells of P. fluorescens DF57 after 24h of carbon starvation. (c) Growing cells of P. fluorescens DF57-11D1. (d) Cells of P. fluorescens DF57-11D1 after 24h of carbon starvation.

In E. coli, a bolA1p::lacZ fusion showed a 12-fold RpoS-dependent induction during the transition to the stationary phase of growth in rich medium (Lange & Hengge-Aronis, 1991). To study the expression of bolA in P. fluorescens, the bolA–lacZ reporter plasmid pML107Pst was inserted into P. fluorescens strains DF57, Pf5 and JL3985, which is an RpoS mutant of Pf5. As shown in Fig. 3, β-galactosidase activity of DF57(pML107Pst) and Pf5(pML107Pst) only showed a two- to threefold induction upon entry into the stationary phase in rich medium, and β-galactosidase activity from JL3985(pML107Pst) was not induced during late-exponential growth. The signals from the negative controls were below 20 Miller units. In conclusion, RpoS seems to have limited influence on bolA expression in P. fluorescens under conditions where RpoS influences bolA expression in E. coli. For comparison, microarray analyses of E. coli K-12 show a 6.4 fold RpoS-dependent upregulation of bolA upon entry into the stationary phase in rich medium (Patten et al, 2004), while comparable analyses for P. aeruginosa show that neither the bolA gene (PA0857) nor the yrbA gene are regulated by RpoS under these growth conditions (Schuster et al, 2004). Hence, data on expression and cell morphology indicate that BolA plays different roles in Pseudomonas and E. coli.

Figure 3

Expression of an ORF2-lacZ protein fusion during growth in LB. Miller units are shown with filled symbols; OD600nm is shown with open symbols. (a) Pseudomonas fluorescens Pf5(pML107Pst). (b) Pseudomonas fluorescens JL3985(pML107Pst). (c) Pseudomonas fluorescens DF57(pML107Pst). Data are mean values from an experiment performed in triplicate. Standard deviations are shown as bars.

Growth phenotypes, biofilm formation and medium-dependent bolA expression

In search for a phenotype of the DF57-11D1 mutant, biofilm formation was determined after incubation for 8h in minimal medium (DMM) or rich medium (10% TSB). In both media, DF57 and DF57-11D1 formed comparable biofilms that were thicker in minimal medium (OD600nm about 0.90) than in rich medium (OD600nm about 0.3). This is in contrast to the situation found in E. coli, where a bolA mutant forms less biofilm than the wild type in minimal medium (Vieira et al, 2004).

We also looked for differences in carbon source utilization profiles by the Biolog GN system. The results indicated a deficient utilization of l-serine by DF57-11D1 (data not shown). To substantiate this result, DF57 and DF57-11D1 were grown in Davies minimal medium (DMM) with l-serine as the only C-source, but neither strain was able to grow in this medium. However, in DMM without the normal mineral N-source (DMM-N) but supplemented with l-serine as the only N-source, DF57-11D1 showed a lower growth rate than the wild type although the same final population size in the stationary phase was obtained (Fig. 4a). This growth deficiency was specific for l-serine (data not shown). DF57-11D1 and the wild type also showed similar growth rates in DMM (Kragelund et al, 1995).

Figure 4

(a) Growth on l-serine and l-cysteine as nitrogen sources for Pseudomonas fluorescens DF57 (open symbols) and DF57-11D1 (closed symbols). Growth was determined as OD600nm in DMM without a mineral nitrogen source but supplemented with 20mM l-serine (○, •), 20mM l-glycine (□, ▪) or 20mM l-serine and 20mM l-cysteine (▵, ▴). (b) Growth on a combination of 20mM l-serine and 20mM l-glycine for DF57(pUCP24) and DF57-11D1(pUCP24) (○, •), and for DF57(pKomp2.3) and DF57-11D1(pKomp2.3) (□, ▪). Data are mean values from an experiment performed in triplicate. Standard deviations are shown as bars.

The possible fates for l-serine in P. aeruginosa include deamination to form pyruvate or conversion to l-glycine and l-cysteine (http://www.genome.jp/kegg/pathway.html). We therefore tested whether any of these compounds affected the growth phenotype of DF57-11D1 in DMM-N with l-serine. We found that addition of l-cysteine to this medium led to comparable growth rates of DF57-11D1 and the wild type (Fig. 4a), while DF57-11D1 showed a poorer growth than the wild type in DMM-N with l-serine plus l-glycine (Fig. 4b) and with l-serine plus pyruvate (data not shown).

Expression studies showed that addition of l-serine to DMM did not induce β-galactosidase expression from the bolA-lacZ reporter plasmid pML107Pst, and β-galactosidase expression was similar for cells grown with l-serine and l-glycine as the sole nitrogen source (data not shown).

Pathways where l-serine could be involved in l-cysteine synthesis have not been characterized in P. fluorescens, but in P. aeruginosal-serine can participate in l-cysteine synthesis either by direct sulph-hydrylation or by reverse transsulphuration (Vermeij & Kertesz, 1999), whereby l-cysteine and α-ketobutyrate are formed. High intracellular levels of α-ketobutyrate can be toxic (Danchin et al, 1984; LaRossa et al, 1987), making it important for the cells to maintain an optimal level of this metabolite. When DF57 and DF57-11D1 were grown in DMM-N with l-serine plus 0.1mM α-ketobutyrate, a larger difference in the growth rate between DF57 and DF57-11D1 was observed than when the two strains were grown in DMM-N with l-serine (compare Fig. 4a and Fig. 5a). This effect was specific for growth on l-serine as the only nitrogen source as no or only minor differences in the growth rates of DF57 and DF57-11D1 were observed in media with α-ketobutyrate plus l-cysteine, l-threonine or a mineral nitrogen source (data not shown).

Figure 5

(a) Growth on l-serine in the presence of α-ketobutyrate for Pseudomonas fluorescens DF57 and DF57-11D1 (○, •). Growth was determined as OD600nm in DMM without a mineral nitrogen source but supplemented with 20mM l-serine and 0.1mM α-ketobutyrate. (b) Growth, determined as OD600nm in DMM without a mineral nitrogen source but supplemented with 20mM l-serine for DF57Gm (○), DF57-ABC (□) and DF57-11D1Gm (•). Data are mean values from an experiment performed in triplicate. Standard deviations are shown as bars.

The growth phenotypes of DF57-11D1 suggest that the bolA gene might be involved in a pathway that includes a conversion of l-serine to l-cysteine. The pathway further appears to generate α-ketobutyrate, or to be involved in the ability of the cells to maintain homeostasis of this metabolite. The observed phenotypes could be due to the lack of a functional BolA protein or it could be due to polar effects on downstream genes in the operon. To discriminate between these possibilities, we constructed four plasmids for complementation: pUCP24bol1 and pUCP24bol3 expressing bolA, pKomp1 expressing ORF3 and ORF4 and pKomp2.3 expressing bolA, ORF3 and ORF4. In DMM-N with l-serine plus α-ketobutyrate, pKomp2.3 partly complemented the growth phenotype of DF57-11D1 (Fig. 4b), while complementation was not achieved by any of the other plasmids (data not shown). A knock-out mutant of ORF5 encoding an ABC-transporter grew as the wild type in this medium (Fig. 5b).

Mono-thiol glutaredoxin genes are situated next to bolA genes in the bacteria Synechococcus elongantus, Rhizobium loti and Leptospira interrogans (Huynen et al, 2005). This resembles the situation in P. fluorescens as ORF4 is predicted to be a protein disulphide isomerase with a thioredoxin domain. Protein disulphide isomerases, thioredoxins and glutaredoxins all catalyse thiol-disulphide interchanges in other proteins and all belong to the thioredoxin superfamily of oxidoreductases (Hu et al, 1997; Fernandes & Holmgren, 2004). In P. fluorescens DF57, however, the gene next to bolA encodes a sulphurtransferase homologue, further supporting a function of the operon in sulphur metabolism. Our complementation experiments indicate that BolA and the putative sulphurtransferase, ORF3, as well as the protein disulphide isomerase, ORF4, function in concert. This is in line with results of yeast-2-hybrid experiments in S. cerevisiae and D. melanogaster, which have detected interactions between homologues of the bacterial mono-thiol glutaredoxins and BolA homologues (Ito et al, 2001; Giot et al, 2003; Huynen et al, 2005).

In conclusion, our data on cell morphology, biofilm formation and bolA expression show distinct differences between the role of BolA in P. fluorescens and E. coli. The growth phenotypes and protein homologies suggest a putative role for bolA and other genes in the operon in metabolism of sulphur-containing amino acids in Pseudomonas. These observations will make it possible to focus further work on the specific role of this operon.

Supporting Information

Fig. S1. Primers used for PCR analysis (Word document).

Supporting info item


This work was supported by the Danish Agricultural and Veterinary Research Council, Grant 9313839 and Grant 9702796. We thank Joyce Loper for providing the P. fluorescens RpoS mutant JL3985 and Herbert P. Schweizer for sending us the plasmid pUCP24.


View Abstract