Differential growth of Escherichia coli and Roseobacter strains in response to varying salt concentrations in the culture medium

It was reported that several antibiotics, including tetracycline and gentamicin, can be affected in their chemistry by high salt concentrations as found in MB [27]. For example, the aminoglycoside kanamycin chelates Cu²⁺[28] and tetracycline forms complexes with divalent cations such as Mg²⁺, Fe²⁺ and Ca²⁺. These interactions have no significant impact on the stability of tetracycline, but decrease the membrane permeability of a cell and therefore the bioavailability of this antibiotic [27, 29–31]. Up to ten times higher concentrations of gentamicin, carbenicillin, chloramphenicol and tetracycline were required for Roseobacter growth inhibition in MB medium (data not shown) compared to hMB with lower sea salt concentrations (see below). Control experiments with E. coli showed that all used antibiotics were active over the whole incubation time in hMB at chosen conditions (data not shown). Consequently, hMB medium was used for further investigations.

Screening of Roseobacter clade bacteria for antibiotic susceptibility

All tested species showed different susceptibilities to the antibiotics (Table 2). As expected, the seven D. shibae strains followed the same trend, with slight variations. They were all resistant to the β-lactam antibiotics ampicillin and carbenicillin. The level of tolerance to ampicillin was up to 500 μg/ml. The Phaeobacter strains, R. denitrificans and R. litoralis also showed resistance to ampicillin, whereas, in contrast to D. shibae, they were sensitive to carbenicillin. Initially, we hypothesised, that the unexpected high ampicillin tolerance might occur due to instability of this antibiotic. It has been reported that ampicillin lost 28% of activity after 24 h at room temperature [30]. However, control experiments with the E. coli strain DH5α revealed complete activity of ampicillin even after incubation for five days at 30°C (data not shown).

Analysing the annotated genomes of the strains by BLAST search and functional predictions (for details see Methods section), we identified genes encoding for β-lactamases, indicating that they are widespread over the Roseobacter clade. They were also found in R. denitrificans, R. litoralis, P. gallaeciensis, O. indolifex and D. shibae. For the latter strain, three β-lactamases encoding genes were identified [using ROSY; [12]]. Thus, the inactivation of the antibiotics via degradation by β-lactamases seems to be an intrinsic resistance mechanism.

Susceptibility of the Roseobacter strains differed towards the four tested aminoglycosides. R. denitrificans showed no susceptibility to all tested aminoglycosides. In contrast R. litoralis and P. gallaeciensis were sensitive to this group of antibiotics. Growth of P. inhibens was inhibited by high concentrations of kanamycin, but the bacterium reacted very sensitive to spectinomycin and gentamicin. The D. shibae strains were resistant to kanamycin, but relatively sensitive to the three other aminoglycosides. O. indolifex was susceptible to all aminoglycosides.

The resistance to the aminoglycoside gentamicin was already reported by Shiba [1991] as one of the characteristic properties of R. denitrificans. The corresponding genome exhibits a gene encoding for a putative aminoglycoside phosphotransferase, a type of enzyme inactivating aminoglycosides via modification [using IMG; [35], and ROSY; [12]]. Putative aminoglycoside phosphotransferase genes were also found in the D. shibae genome and those of other Roseobacter clade species. Most of these strains contain two putative genes, except for D. shibae DFL12^T which exhibits 10 genes [using IMG; [35]]. This observation might explain the high kanamycin tolerance of D. shibae.

The MICs for tetracycline and chloramphenicol were in a range of 10 - 50 μg/ml and 10 - 30 μg/ml, respectively. None of the tested species showed resistance to these two antibiotics. In summary, we identified at least three antibiotics for every strain which are suitable as selective makers for use in molecular biology and genetic protocols. In the following experiments we used twice the amount of the MIC for the selection of plasmid-containing strains and for the maintenance of the plasmids within the Roseobacter strains. Several groups reported that the MICs of bacteria grown in liquid cultures can be lower than for the same bacteria grown on agar plates as biofilms [36, 37]. Control experiments demonstrated that only plasmid-containing cells survived twice of the MIC via expression of the plasmid-encoded resistance gene. Also in case of differences between MICs determined in static liquid culture and in aerated liquid cultures, the use of twice of the MIC ensured selection of plasmid-containing cells.

Roseobacter clade bacteria are resistant to common chemical transformation approaches

First, chemical transformation methods [38] were tested for the transformation of the various Roseobacter strains. Chemo-competent cells were prepared with CaCl₂ and furthermore with RbCl₂. Plasmid-DNA transfer experiments were carried out by mixing bacteria with 50 ng plasmid-DNA (pBBR1MCS), followed by a 30 min-incubation on ice and a subsequent 2 min heat shock at 42°C similar to the standard procedure for E. coli[38]. Transformation of Roseobacter strains led to no transformants, either with CaCl₂-competent or with RbCl₂-competent cells. No transformants were observed for any of the 12 tested strains. Similar observations were made for Rhodobacter strains, which are close relatives of the Roseobacter strains [39]. Only one successful approach was described for R. sphaeroides in 1982 [16]. Initial experiments using the published method did not lead to transformants of Roseobacter clade bacteria.

Transformation of Roseobacter clade bacteria via electroporation

Since common chemical transformation methods as described by Sambroock et al. [1989] did not lead to successful DNA transfer in Roseobacter clade bacteria (see above), the electroporation method was tested. Electroporation was performed following the protocol of Miller and Belas [2006]. This method was successfully used for other members of the Roseobacter clade as Silicibacter sp. [19, 20] and S. pomeroyi[22]. Salt-free cell suspensions were prepared by washing with 10% (v/v) glycerol in ultra-pure water. We tested the washing procedure with increasing numbers of separate washing steps. A minimal of 5 washing steps was necessary to effectively remove the salts from culture media, verified by a resulting optimal pulse length of 4.9 - 5.0 ms. The competent cells were subsequently frozen in liquid nitrogen and stored at -80°C. Under these conditions cells can be stored for about 3 weeks, except of R. denitrificans, which was viable only for a maximum of 1 week. We used 25 ng and 50 ng plasmid-DNA (pBBR1MCS), both resulting in similar transformation rates. Different pulse intensities were tested (1.5 - 3.0 kV). An intensity of 2.5 kV revealed the best results and was used for further experiments. The electroporation method was successful for all tested strains, although transformation rates differed between them. A maximum of 1 × 10³ cfu/μg plasmid-DNA were observed for P. inhibens and R. litoralis. Slightly higher efficiencies of 1 × 10⁴ cfu/μg plasmid-DNA were observed for D. shibae and R. denitrificans. Good efficiencies were observed for P. gallaeciensis with 1 × 10⁵ cfu/μg plasmid- DNA and O. indolifex with an efficiency of 1 × 10⁷ cfu/μg plasmid-DNA. Recently, an optimized electroporation method was described for the Gram-negative P. aeruginosa resulting in transformation efficiencies ranging from 10⁷ to 10¹¹ cfu/μg plasmid-DNA [40]. These results are comparable with the efficiencies obtained in O. indolifex, indicating that our protocol is sufficient for the members of the Roseobacter clade. Although the transformation efficiencies are much less for most of the tested Roseobacter strains, this technique can be used as a fast and easy method to transfer plasmids into Roseobacter cells.

Efficient conjugal transformation of Roseobacter clade bacteria

Biparental mating using E. coli S17-1 as donor strain was described for plasmid transfer into S. pomeroyi and Sulfitobacter before [21, 23]. Thereby, the use of spontaneous emerged antibiotic-resistant mutants of the recipient strains is one of the principles used to counter-select against the E. coli donor strain after conjugation [e.g. [23, 41]]. It is well known that such mutations may also cause indirect pleiotropic effects that might influence the general physiology of the target strain. Changes in growth behaviour, uracil sensitivity and bacteriophage sensitivities were reported for spontaneous rifampicin-resistant mutants [42, 43]. A second approach utilises auxotrophic donor strains. Here, we used E. coli ST18 as donor strain for the conjugation procedure, which is a hemA mutant of E. coli S17 λ-pir [26]. This strain cannot synthesize the general tetrapyrrole precursor aminolevulinic acid (ALA). Hence, to complement the lethal mutation ALA has to be added to the medium for growth. Consequently, for the selection of plasmid-containing Roseobacter recipients after conjugation hMB agar plates without ALA were used to inhibit growth of the E. coli donor cells. Several conditions of the conjugation procedure were varied including medium composition and conjugation time (for details see Methods section). We found no obvious differences between 24 h- and 48 h-conjugation time (data not shown). Even conjugation times below 24 h might be sufficient for the fast growing Phaeobacter strains and O. indolifex.

Several plasmids were tested for transfer via conjugation. These plasmids were successfully used for homologous expression of genes to complement gene knockouts in trans in other Gram-negative bacteria before. The IncP-plasmids pFLP2, pLAFR3 and pUCP20T were not transferable or not stable in the tested Roseobacter strains (see below). In contrast, the IncQ-plasmids pRSF1010, pMMB67EH and the tested pBBR1MCS derivates were transferable. They were recovered from exconjugants by plasmid-DNA preparation and subsequently visualized via gel electrophoresis.

Plasmid Stability

There is only one report about homologous gene expression in Roseobacter clade bacteria using the vector pRK415 [21]. This vector was widely used for a broad range of Gram-negative species, including R. sphaeroides [e.g. [44, 45]]. However, the small numbers of restriction enzyme sites available for cloning and the use of tetracycline as selective marker represent major drawbacks for its use. Therefore, we tested the stability of several mobilizable plasmids of different sizes, carrying different antibiotic resistance genes and origins of replication that belong to distinct incompatibility groups (IncP, IncQ). The maintenance of the plasmids was analysed by spreading cells, which were grown over 10 passages until stationary phase in MB without antibiotics, on hMB agar plates in the presence and absence of antibiotics. Moreover, we tested the cells for the presence of the plasmid by plasmid preparation and visualisation via gel electrophoresis. A reproducible and stable transformation of the Roseobacter cells was only obtained with pBBR1MCS derivates. This broad-host-range vector contains the origin of replication of pBBR1 from Bordetella bronchiseptica. It has a wide compatibility to IncQ, IncP, IncW, ColE1 and p15A ori plasmids [46, 47]. The IncQ containing plasmids pRSF1010 and pMMB67EH were also transferable into the Roseobacter bacteria, except for the Phaeobacter strains. But in contrast to pRSF1010, pMMB67EH was not stable and got lost after 1 - 2 passages even in the presence of selection pressure. Interestingly, the IncP plasmids pLAFR3, pUCP20T and pFLP2, which are suitable for many other Gram-negative bacteria [48–50], were not transferable or not stable in the tested Roseobacter strains.

The members of the Roseobacter clade contain up to 13 natural plasmids in a size range of 4.3 - 821.7 kb [4]. For example, D. shibae DFL12^T type strain contains five plasmids with a size of 72 to 190 kb [51]. Three of the five plasmids harbor a repABC-type replicon, one contains a repA- and one a repB-type replicon [51]. The stability of different plasmids within one cell depends mainly on their incompatibility groups, which are based on the nature of genetic elements involved in plasmid replication or partitioning [15]. Incompatibility is thereby a manifestation of relatedness of these elements, meaning that plasmids with closely related replication origins are incompatible and therefore not stable within one cell [15]. The replicons of the IncP plasmids seem to be closely related to the natural plasmids of the Roseobacter bacteria, resulting in the observed instability.

Moreover, at least four of the five plasmids of D. shibae contain additional systems for plasmid maintenance. These are composed of two small genes, encoding a stable toxin as well as a less stable antitoxin [51]. The antitoxin must be continually produced to prevent the long-living toxin from killing the cell. Otherwise the toxin induces cell death once the plasmid gets lost during cell division [51, 52]. Such toxin/antitoxin systems are characteristic for low copy plasmids and provide plasmid specific differences between various vectors and therefore sustain their compatibility and plasmid replacement protection [53].

Reporter gene system

Reporter genes are commonly used for the analysis of promoter activities and transcriptional regulation events. A system using lacZ reporter gene fusions was recently described for Sulfitobacter[23]. Using the novel broad-host-range expression vector pRhokHi-2FbFP [54] based on the conjugally transferable and stably maintained plasmid pBBR1MCS, a novel expression system was established. The vector pRhokHi-2 has been designed for constitutive expression of target genes in the phototrophic bacterium R. capsulatus. For the analysis of vector-mediated gene expression a pRhokHi derivative was used harbouring a reporter gene encoding the oxygen-independent, flavin mononucleotide-based fluorescent protein FbFP from Bacillus subtilis as reporter [55]. Thus, it is applicable under both aerobic and anaerobic conditions [55]. For many natural habitats oxygen limiting conditions are of central importance. However, the commonly employed reporters including β-galactosidase, luciferase or green fluorescent protein (GFP) require oxygen for dye development, bioluminescence and fluorescence [55]. For a proof of principle the fbFP gene was cloned under the control of the constitutive promoter of the aminoglycoside phosphotransferase II (aphII) gene in a pBBR1MCS derivate [55]. The plasmid was introduced into the Roseobacter strains by conjugation and fluorescence measurement were made of the fbFP expressing recipients in comparison to the wildtype strains. Clear emission signals were observed in the range of 15 RFU for O. indolifex to 72 RFU for P. gallaeciensis at 520 nm upon excitation with blue-light at 450 nm (Figure 1). The altered range of fluorescence might be explained by different copy numbers of the plasmid, different codon usages or different promoter utilisation by the tested strains. The stability and the evenly distribution of pRhokHi-2FbFP within the populations was verified by fluorescence microscopy (data not shown). The observations indicated that FbFP can be used for in vivo fluorescence measurements in various Roseobacter strains.

Gene deletion mutants of D. shibae DFL12^T

The d issimilative n itrate r espiration regulator Dnr is a global regulator for anaerobic growth under denitrifying conditions in pseudomonads [56–58]. Six homologous genes were identified in the genome of D. shibae. For the construction of a gene deletion mutant of one of these dnr genes (Dshi_3189) the vector pEX18Δdnr::Gm^r was constructed (Figure 2A). Replacement of the dnr gene with the gentamicin cassette was varified via PCR (Figure 2B) resulting in a 2.12 kb fragment for the wildtype corresponding to the 711 bp dnr gene, the 652 bp upstream region and the 758 bp downstream region of dnr. For the deletion mutants a 3.21 kb fragment was observed containing the up- and downstream regions and the 1.8 kb gentamicin cassette.

Growth behaviour analysis of D. shibae DFL12^T under anaerobic conditions with nitrate as electron acceptor clearly showed that D. shibae was able to grow by denitrification (Figure 2C). This is of special interest, since D. shibae was previously described as strict aerobic bacterium [25]. The recently sequenced and annotated genome on D. shibae DFL12^T recovered clusters of genes necessary for anaerobic metabolism [51].

The comparison of the D. shibae wildtype to the obtained dnr- mutants revealed a significant reduction of anaerobic nitrate respiratory growth of the tested mutants (Figure 2C), demonstrating the influence of the regulator Dnr on the growth under denitrifying conditions. The presence of six dnr genes indicated a fine-tuned regulation of this metabolic pathway. This was confirmed by the minor growth reduction of the dnr mutants.

Bacterial strains, plasmids and growth conditions

Determination of the minimal inhibitory concentration

For the determination of minimal inhibitory concentrations (MIC) 5 ml hMB was supplemented with freshly prepared antibiotic solutions from 0 - 500 μg/ml in 5 μg steps. Test tubes were inoculated with cultures of the respective bacterial strains grown for 24 h in MB to an OD₅₇₈ of 0.05. Subsequently, bacterial growth was checked by OD₅₇₈ measurements after incubation for 3 and 5 days at 30°C without shaking. The MIC is defined as the lowest concentration of a tested antibiotic, which inhibits the growth of bacteria. All experiments were repeated three times in duplicate. The used antibiotics were obtained from manufactures as followed: ampicillin (Roth, Karlsruhe, Germany), carbenicillin disodium salt (Gerbu Biotechnik GmbH, Gaiberg, Germany), chloramphenicol (Roth, Karlsruhe, Germany), gentamicin sulphate (Roth, Karlsruhe, Germany), kanamycin sulfate (Gerbu Biotechnik GmbH, Gaiberg, Germany), spectinomycin dichloride pentahydrate (Sigma-Aldrich, Munich, Germany), streptomycin sulphate (United States Biochemical Corp., Cleveland, USA), tetracycline hydrochloride (United States Biochemical Corp., Cleveland, USA). For selection of plasmid-containing Roseobacter recipients on agar plates after conjugation the twofold concentration of the MIC of the respective antibiotic in hMB was used.

Preparation of chemically competent cells for the transfer of plasmid-DNA into Roseobacter strains

Chemo-competent cells were prepared as described by Sambrook et al. [1989]. To prepare CaCl₂- competent cells, the Roseobacter strains were cultivated in MB at 30°C and 200 rpm up to an OD₅₇₈ of 0.7. Ten ml of the culture were centrifuged for 15 min at 3,200 × g and 4°C. The bacterial pellet was resuspended in 2 ml cold 10% (v/v) glycerol with 100 mM CaCl₂ in ultra-pure water and centrifuged for 2 min at 8,000 × g and 4°C. Afterwards, the cells were resuspended in 100 μl cold 10% (v/v) glycerol with 100 mM CaCl₂ in ultra-pure water and incubated on ice for 1 h. Subsequently, 200 μl aliquots were frozen in liquid nitrogen and stored at -80°C.

To prepare RbCl₂-competent cells, the Roseobacter strains were cultivated in 20 ml MB supplemented with 400 μl of a stock solution containing 500 mM MgCl₂ and 500 mM MgSO₄ at 30°C and 200 rpm up to an OD₅₇₈ of 0.7. Four ml of the culture were centrifuged for 2 min at 8,000 × g and 4°C. Cells were resuspended in 2 ml ice cold transformation buffer (100 mM CaCl₂, 50 mM RbCl₂, 40 mM MnCl₂) and incubated on ice for 30 min, followed by a centrifugation step for 2 min at 8,000 × g and 4°C. Finally, cells were resuspended in 200 μl transformation buffer. The chemo-competent cells were stored on ice until they were used or frozen at -80°C in 20% (v/v) glycerol.

For the transformation, 200 μl of chemo-competent cells (CaCl₂- or RbCl₂-competent) were gently mixed with 50 ng plasmid-DNA and incubated for 30 min on ice. After a heat shock for 2 min at 42°C, 800 μl MB medium was added and the bacteria were incubated for 3 h at 30°C for the expression of the antibiotic resistance marker encoded by the plasmid. Afterwards the cells were sedimented by centrifugation for 2 min at 8,000 × g and 4°C and the supernatant was decanted. Cells were resuspended in the residual liquid and spread on hMB agar supplemented with the appropriate antibiotic.

Transfer of plasmid-DNA into Roseobacter strains by electroporation

Electro-competent cells were prepared as described previously by Miller and Belas [2006] with slight modifications. Therefore, cells were grown in MB medium at 30°C and 200 rpm to an OD₅₇₈ of 0.5. Ten ml culture was centrifuged for 15 min at 3,200 × g. Sedimented cells were washed 5 times with 10 ml cold 10% (v/v) glycerol in ultra-pure water. Then, the cell pellet was resuspended in 400 μl 10% (v/v) glycerol and 50 μl aliquots were frozen in liquid nitrogen and stored at -80°C. For electroporation, 25 - 50 ng plasmid-DNA were added to 50 μl competent cells in an ice cold 2 mm pulser cuvette (Bio-Rad, Munich, Germany). The mixture was treated in a Bio Rad gene pulser II with field strength of 1.5 - 3.0 kV, resistance of 200 Ω and capacitance of 25 μF. After electroporation the cells were transferred to 1 ml cold MB media and incubated overnight at room temperature with shaking at 300 rpm to allow the expression of antibiotic resistance genes. To investigate electroporation efficiency, cells were serially diluted in 1.7% (w/v) sea salt solution and plated on hMB agar plates with the appropriate antibiotic concentration and incubated for 2 days (Phaeobacter strains and O. indolifex) or 4 days (Roseobacter strains and D. shibae) at 30°C. Subsequently, colony forming units (cfu) were determined.

Conjugal transfer of plasmid-DNA from E. coli into Roseobacter strains

The conjugation procedure was modificated for Roseobacter strains from the protocol of Thoma and Schobert [2009]. The recipient Roseobacter strains were cultivated for 18 h in MB-Medium. The donor E. coli strain ST18 was grown in LB-medium supplemented with 50 μg/ml ALA (Sigma-Aldrich, Munich, Germany) up to the logarithmic phase (OD₅₇₈ = 0.5 - 0.6). Both cultures were mixed in a donor:recipient ratio of 1:1; 2:1; 5:1 or 10:1 according to the optical density (OD₅₇₈) of the cultures. Cells were sedimented by centrifugation for 2 min at 8,000 × g at 20°C, resuspended in the residual liquid and used to inoculate hMB agar, LB+hs agar and hLB+hs agar respectively, all supplemented with 50 mg/ml ALA, in form of a spot. The plates were incubated at 30°C for 24 h and 48 h. Subsequently, cells were scraped from the plate and resuspended in 1 ml MB by vigorous shaking. Disruption of cell aggregates was confirmed via microscopic inspection of the resulting single cells. A dilution series in 1.7% (w/v) sea salt solution was prepared and plated on hMB with the appropriate antibiotic concentration to determine the number of transconjugants per ml. Since the plates did not contain ALA the auxotrophic donor E. coli strain was not able to grow. In parallel, transconjugants were also plated on hMB without antibiotics to determine the number of viable cells per ml. The efficiency of conjugation was defined as the ratio between conjugants:viable cells. Moreover, a minimum of 10 exconjugants were tested for the presence of the plasmid by plasmid-DNA isolation and gel electrophoresis.

Isolation of plasmid-DNA from the Roseobacter strains

Plasmids used in this study were isolated using the mini plasmid isolation kit from Qiagen (Qiagen, Hilden, Germany) following the manufacturers' instructions for Gram-negative bacteria.

Genome analysis and bioinformatics approach

For genome analysis of the Roseobacter strains the databases of the Joint Genome Institute http://www.jgi.doe.gov [35] and the Roseobacter database http://rosy.tu-bs.de/ [12] were used. Open reading frames were identified using BLASTX analysis with a cutoff E value of 1e^-5. β-lactamase and aminoglycoside resistance genes from P. aeruginosa and E. coli were used for the study. Moreover, Pfam [59], TIGRfam [60] and COG [61] predictions were used to identify functional homologues.

The genome of D. shibae DFL12^T was sequenced at the Joint Genome Institute (JGI) Production Genomics Facility. The sequences, comprising a chromosome and 5 plasmids, can be accessed using the GenBank accession numbers NC_009952, NC_009955, NC_009956, NC_009957, NC_009958 and NC_009959. Manual curation and reannotation of the genome was carried out using the Integrated Microbial Genomes Expert Review System (img/er http://imgweb.jgi-psf.org) [51] and the Artemis software package http://www.sanger.ac.uk/Software/Artemis/v9.

The complete and annotated genome sequences of Roseobacter denitrificans strain OCh 114 and its associated plasmids have been deposited in the DDBJ/EMBL/GenBank database under accession numbers CP000362, CP000464, CP000465, CP000466, and CP000467. Initial annotation data were built using the Annotation Engine at The Institute for Genomic Research http://www.tigr.org/edutraining/training/annotation_engine.shtml, followed by comprehensive manual inspection and editing of each feature by using Manatee http://manatee.sourceforge.net/.

Fluorescence imaging of reporter gene carrying cells

Some of the Roseobacter clade members were fluorescence labelled using the FMN-based fluorescence protein FbFP [55] (available as evoglow-Bs1 from Evocatal GmbH, Düsseldorf, Germany). Therefore, the fluorescence reporter gene was constitutively expressed using the broad-host-range expression vector pRhokHi-2-FbFP [54]. Fluorescence microscopy was used for in vivo fluorescence imaging of living cells. An aliquot of the microbial cell culture was placed on a microscope slide and illuminated with light of the wavelength 455-495 nm. Fluorescence emission of single cells was analyzed in the ranges of 500-550 nm using a Zeiss Fluorescence Microscope (Axiovert 200 M) at a magnification of 600-fold. Documentation was carried out using the camera AxioCam (Carl Zeiss, Jena, Germany) and the software AxioVision Rel 4.5 (Carl Zeiss, Jena, Germany).

Quantification of reporter gene activity

Liquid cultures (MB, 48 h, 30°C, 200 rpm) of the pRhokHi-2-FbFP containing Roseobacter strains were diluted in MB to an OD₅₇₈ of 0.7 and excited with light of the wavelength of 450 nm. Fluorescence emission was detected at 475 - 550 nm. Cultures of the wildtype strains served as negative control. For quantification of the fluorescence the luminescence spectrometer LS 50 B (Perkin Elmer, Waltham, Massachusetts, USA) was used.

Construction of D. shibae DFL12^T dnr (Dshi_3189) deletion mutants

To obtain gene deletion mutants from D. shibae DFL12^T, the well-established suicide vector for Gram-negative bacteria pEX18Ap was used [48]. To construct the gene deletion vector pEX18Δdnr::Gm^r, the Sac I-digested Ω-gentamicin resistance cassette of pPS858 [48] was cloned between two PCR fragments of the dnr gene (Dshi_3189) in the multiple cloning site of pEX18Ap. The two PCR fragments contained DNA homologous to upstream and downstream regions of the Dshi_3189 gene. A 652-bp fragment containing the upstream promoter region of Dshi_3189 was amplified using primer oPT19 (5'-GGGGTACC AATGCCATGACCT ACTTC-3'), which contains a Kpn I restriction site at the 5' end, and oPT20 (5'-CGAGCTC CGCATGAACGAGTCATCTT-3'), containing a Sac I site (both restriction sites underlined). The primers oPT21 (5'-CGAGCTC AGCAGAACCATGCGGAGAT-3'), containing a Sac I site, and oPT22 (5'-CCCAAGCTT TCACCAGCGGGCTTTTC-3'), which contains a Hind III site (both restriction sites underlined), amplified 758 bp of the corresponding downstream region of Dshi_3198. The suicide vector pEX18Δdnr::Gm^r was used to replace the Dshi_3198 gene with the Ω-gentamicin cassette. To confirm homologous recombination PCR analysis was performed. Furthermore, the growth behaviour of the resulting mutants was analysed under anaerobic conditions with nitrate as electron acceptor, as outlined before [57].