The σ-factor FliA, ppGpp and DksA coordinate transcriptional control of the aer2 gene of Pseudomonas putida

Authors


E-mail victoria.shingler@molbiol.umu.se; Tel. (+46) 90 785 2534; Fax (+46) 90 771420.

Summary

Here the σ-factor requirement for transcription of three similar, but differentially regulated, aer genes of Pseudomonas putida KT2440 is investigated. Previous work has shown that the three Aer proteins, like chemoreceptors, colocalize to a single pole in a CheA-dependent manner. Lack of Aer2 – the most abundant of these three proteins – mediates defects in metabolism-dependent taxis and aerotaxis, while lack of Aer1 or Aer3 has no apparent phenotype. We show, using wild-type and mutant P. putida derivatives combined with P. putida reconstituted FliA- (σ28) and σ70-dependent in vitro transcription assays, that transcription of aer2 is coupled to motility through the flagella σ-factor FliA, while σ70 is responsible for transcription of aer1 and aer3. By comparing activities of the wild-type and mutant forms of the aer2 promoter, we present evidence (i) that transcription from FliA-dependent Paer2 is enhanced by changes towards the Escherichia coli consensus for FliA promoters rather than towards that of P. putida, (ii) that the nature of the AT-rich upstream region is important for both output and σ70 discrimination of this promoter, and (iii) that Paer2 output is directly stimulated by the bacterial alarmone ppGpp and its cofactor DksA.

Introduction

Soil- and water-dwelling Pseudomonas putida strains are metabolically versatile, being able to use a wide range of organic compounds. Their innately broad metabolic capacity is frequently augmented by plasmids that encode the ability to utilize aromatic pollutants, such as phenolics, as sole carbon and energy sources (Shingler, 2004). Efficient utilization of these carbon and energy sources in the environment requires both catabolic proficiency and the ability to detect and move towards them (Paul et al., 2006). For the latter, P. putida KT2440 uses flagella-mediated taxis responses. These responses are mediated via 27 potential receptors, three of which (Aer1 to Aer3) are similar to the Escherichia coli Aer receptor (Sarand et al., 2008). Aer2, the most abundantly expressed of these three polar localized receptors, mediates directional responses through oxygen gradients (aerotaxis) and metabolism-dependent taxis towards readily metabolized compounds such as citrate, glycerol and phenolics (Sarand et al., 2008). Unlike conventional chemoreceptors that specifically bind ligands to propagate taxis responses, energy- and metabolism-dependent receptors do not sense specific compounds per se. Rather, these receptors use membrane-coupled PAS domains and bound cofactors to detect change in cellular energy that results from metabolism of the compounds to guide the bacteria to energetically favourable environments (reviewed in Alexandre and Zhulin, 2001; Taylor, 2007).

Pseudomonas putida KT2440 possesses polar bundles of six or more flagella. Biosynthesis of the flagella motility apparatus is temporally controlled and coupled to checkpoints during the flagella assembly pathway. The hierarchical expression of flagella genes can be achieved by diverse mechanisms, but usually involve a master regulator that initiates the cascade, and coupling of late flagella gene expression to completion of the hook basal body structure. Secretion of the anti-σ-factor FlgM through the hook basal body structure releases the FliA σ-factor (also known as σ28 and sigD) for association with core RNA polymerase (RNAP) to form the holoenzyme required for transcription of genes needed for completion of the flagella filament (reviewed in Chevance and Hughes, 2008; Smith and Hoover, 2009).

FliA is but one of 24 exchangeable σ-factors of P. putida KT2440 (Martinez-Bueno et al., 2002) that confer promoter-recognition specificity to the multisubunit catalytic core RNAP (α2ββ'ω). In actively growing cells, most transcription is mediated through a housekeeping σ-factor (σ70 in E. coli and P. putida) that directs RNAP to the bulk of promoters within the genome. The activities of many alternative σ-factors are, however, triggered by environmental cues, specific stress conditions, or growth and developmental transitions to control specific subsets of genes to aid survival (reviewed in Helmann, 2002; Paget and Helmann, 2003). Different subregions of the σ-factors provide promoter specify to the holoenzyme RNAPs through recognition of discrete signature motifs of bacterial promoters, namely the −35 box (σ-domain 4), the −10 box (σ-domain 2) and the extended −10 motifs (σ-domain 3) (Gruber and Gross, 2003; Paget and Helmann, 2003). In the case of housekeeping σ70, an additional promoter interaction with the −10 downstream DNA is provided by σ-region 1.2 (Haugen et al., 2006). Promoter affinity afforded by the σ-factor in the context of RNAP holoenzymes can also be augmented through interaction of the α-subunit of the holoenzyme with AT-rich UP-elements located upstream of some bacterial promoters (Gourse et al., 2000, and references therein).

In this study we aimed to identify the σ-factor(s) required for transcription of the three differentially regulated aer genes of P. putida KT2440 (Sarand et al., 2008). In particular, we were interested in examining factors that impact output from the promoter responsible for transcription of the aer2 gene and thus ultimately the ability of P. putida KT2440 to mediate aero- and metabolism-dependent taxis. Many of the nutritional and stress signals that soil bacteria are exposed to elicit production of the bacterial alarmone guanosine bispyrophosphate (ppGpp) that directly targets RNAP to alter its properties (reviewed in Potrykus and Cashel, 2008). The ppGpp alarmone frequently acts together with the RNAP-binding protein DksA to inhibit or stimulate transcription from kinetically susceptible σ70- and σE-dependent promoters (e.g. Paul et al., 2004; Paul et al., 2005; Aberg et al., 2008; Costanzo et al., 2008; Johansson et al., 2008). Here we present evidence that the Paer2 promoter is dependent on FliA-RNAP and an AT-rich upstream region, and that this FliA promoter joins the rank of promoters whose activity can be directly regulated by ppGpp and DksA.

Results

FliA-RNAP controls transcription of aer2 in vivo

The genome sequence of P. putida KT2440 (http://www.ncbi.nlm.nih.gov/genomes, Nelson et al., 2002) revealed that the region immediately upstream of the aer2 (PP_2111) gene contains a sequence with high homology to known consensuses for the flagella σ-factor FliA (Fig. 1A). This, together with the recent finding from a transcriptomics analysis that aer2 is downregulated in P. putida KT2440 lacking FliA (σ28) (Rodriguez-Herva et al., 2009), strongly suggested that aer2 is transcribed by FliA-RNAP. To verify this, we generated monocopy aer2 transcriptional reporter strains using wild-type P. putida KT2440 and its FliA null counterpart, P. putida KT2440 fliA::Km, in which the fliA gene is disrupted by a kanamycin resistance gene cassette (Rodriguez-Herva et al., 2009). These reporter strains were generated by recombination and contain the luxAB luciferase genes of Vibrio harveyi fused immediately downstream of the aer2 gene on the host chromosome. As anticipated, transcriptional analysis revealed that lack of FliA essentially abolishes transcription of aer2 throughout growth on rich media (Fig. 1B, compare open squares and circles). A C-terminal His-tagged version of P. putida FliA, expressed from the isopropyl β-D-thiogalactopyranoside (IPTG)-inducible lacIQ/Ptac promoter of a 16- to 20-copy-number vector (pVI983), both restored transcription of aer2 in the FliA null mutant to above that observed in the wild-type strain (Fig. 1B, compare open squares and triangles), and complemented the inability of the FliA null mutant to produce polar flagella bundles that were indistinguishable in size or number from those of the wild-type (Fig. 1C). The transcriptional response from the Paer2 promoter in the FliA null complemented strain was always higher than that in the wild type but fluctuated across the growth curve (Fig. 1B, open triangles). We attribute these fluctuations to the incorrect temporal expression of excess FliA–His from the Ptac promoter, which would be anticipated to have complex repercussions through σ-factor competition for limiting core RNAP (Bernardo et al., 2006, and references therein).

Figure 1.

Transcription of aer2 is dependent on the flagella σ-factor FliA (σ28).
A. Nucleotide sequence of the 100 bp region upstream of the aer2 ATG initiation codon (bold italic), with the AT-rich Paer2 promoter upstream region highlighted in grey. The DNA regions showing similarity to FliA consensus sequences are underlined with the 11 out of 13 residues identical to the P. putida consensus shown in bold. Alignment of Paer2 with consensus FliA promoter sequences can be found in Fig. 3A.
B. In vivo transcription of aer2–luxAB monocopy P. putida KT2440 transcription reporter strains cultured in rich (LB) medium. Growth of all strains was indistinguishable from average shown (filled squares). Luciferase activity profiles (open symbols) of wild-type KT2440 (squares), the KT2440 FliA null mutant (circles), and the KT2440 FliA null mutant harbouring the pVI983 FliA–His expression plasmid (triangles) induced by addition of 0.1 mM IPTG in the growth medium are shown. Data are the average of duplicate determinations with standard error. Specific activities are given as relative luciferase units (RLU) per OD600.
C. Electron micrographs of strains as under (B), grown overnight on rich LB plates with 0.1 mM IPTG. Examination of > 50 individual cells from each culture revealed no difference between the number or location of flagella of the wild-type and its FliA null mutant counterpart harbouring the pVI983 FliA–His expression plasmid.

σ70-RNAP, rather than FliA-RNAP, controls transcription of aer1 and aer3 in vivo

Because P. putida encodes three highly homologous but differentially regulated Aer-like receptors (Sarand et al., 2008), we were interested in determining if aer1 (PP_2257) and aer3 (PP_4521) were likewise controlled by FliA. The aer1 gene is the first gene in a bi-cistronic operon (Sarand et al., 2008), while aer3 is the middle gene of a tri-cistronic operon with PP_4522 upstream and PP_4520 encoded downstream (data not shown). Analysis using monocopy aer1 and aer3 luciferase reporter strains, analogous to those described above for aer2, revealed no differences in their transcription in the wild-type compared with the FliA null P. putida KT2440 strain (data not shown). In addition to FliA and housekeeping σ70, the P. putida KT2440 genome encodes another 22 alternative σ-factors, of which 13 extracytoplasmic σ-factors (ECF 2, 4–9 and 14–19) appear dedicated to iron acquisition (Martinez-Bueno et al., 2002). Therefore, to further investigate the σ-factor(s) responsible for transcription from the promoters of aer1 and aer3, we used P. putida KT2440 derivatives with individual insertions within 12 σ-encoding genes; namely within three representatives of the iron-acquisition dedicated σ-factors (ECF 2, 4 and 6), and within the other dispensable σ-factors of the KT2440 genome: rpoN54N), rpoS38S), rpoH38H), ECF 1, 3 and ECF 10–13. Plasmid-based transcriptional reporters bearing the promoter regions of aer1 and aer3 (DNA spanning 300 bp upstream of the ATG start of the corresponding operons) controlling transcription of the luxAB genes were generated (pVI987 and pVI900 respectively). These transcriptional reporter plasmids were introduced in the KT2440 σ-factor mutants and luciferase activity was monitored throughout the growth curve on rich media. In no case was a decrease in transcription from the aer1 or aer3 operon reporter plasmids detected (data not shown). These findings suggested to us that, although these promoter regions only contain sequences with relatively poor homology to the −35 and −10 consensus for σ70, the essential σ70 factor is likely responsible for control of transcription of the aer1 and aer3 operons. The importance of σ70 for transcription of these two operons was verified by introducing the aer1 (pVI987) and aer3 (pVI900) transcriptional reporter plasmids into a P. putida KT2440 strain that overexpresses σ70 by virtue of a duplicate copy of the rpoD gene under the control of the arabinose inducible araC/PBAD promoter. Arabinose induction to elevate σ70 levels increased output from both reporter plasmids 1.5- to 2.5-fold during all phases of growth (Fig. S1). Taken together, the results suggest that utilization of FliA for transcription of aer2, and σ70 for transcription of the aer1 and aer3 operons, contributes to their differential regulation in vivo.

P. putida FliA-RNAP transcribes aer2 in vitro

To enable analysis of transcription from the aer2 promoter in vitro, the C-terminal His-tagged FliA used for in vivo complementation of the FliA null mutant was purified. The resulting FliA protein was used in conjunction with P. putida KT2440-derived core RNAP to reconstitute P. putida FliA-RNAP holoenzyme, and supercoiled pVI1010 carrying the Paer2 promoter region (spanning 300 bp upstream of the aer2 initiation codon) in front of a strong transcriptional terminator. Multiple-round in vitro transcriptional assays with increasing concentrations of FliA, but constant levels of core RNAP and the Paer2 DNA template, showed that P. putida FliA-RNAP can recognize and promote transcription from the aer2 promoter in a concentration-dependent manner in vitro (Fig. 2A, filled squares). As a specificity control, the Paer2 DNA template was also used in a similar titration with P. putidaσ70. However, σ70-RNAP was only capable of eliciting low levels of transcription from the Paer2 promoter of the plasmid (Fig. 2A, open circles). For comparison, we also performed similar titrations with pVI948 that carries the σ70-dependent Pr promoter that in its native context drives transcription of the P. putida-derived dmpR gene (Johansson et al., 2008). As anticipated, while σ70-RNAP elicited dose-dependent transcription from Pr, FliA-RNAP was incapable of promoting transcription from this promoter (Fig. 2B). Consistent with in vivo findings, in vitro transcription from analogous DNA templates carrying the aer1 and aer3 operon promoter regions (pVI000 and pVI1005 respectively) produced transcripts with σ70-RNAP but not with FliA-RNAP in vitro (data not shown). These results clearly demonstrate the predominant recognition and specificity of the Paer2 promoter for FliA-RNAP.

Figure 2.

In vitro transcription of Paer2 is dependent on the flagella σ-factor FliA (σ28).
A. Multiple-round in vitro transcription by FliA-RNAP (filled squares) and σ70-RNAP (open circles) with the Paer2 DNA template, pVI1010, which carries 300 bp upstream of aer2 initiation codon. The reactions contained 10 nM template, 10 nM core RNAP and increasing amounts of corresponding σ-factor FliA–His or native σ70. The autoradiograms show transcripts from Paer2 with each holoenzyme.
B. As for (A), but with the DNA template pVI948 that carries the σ70 Pr promoter.

The P. putida FliA consensus promoter sequence exhibits suboptimal output in vitro

The Paer2 promoter bears strong similarity to the consensus sequence for P. putida FliA (Rodriguez-Herva et al., 2009) and somewhat less similarity to the FliA consensus sequence derived from E. coli (Ide et al., 1999; Koo et al., 2009), which differ as shown in the comparison in Fig. 3A. To determine if P. putida FliA-RNAP can efficiently recognize both these consensus sequences, in vitro transcription plasmids bearing either the P. putida (pVI1014) or the E. coli (pVI1015) consensus sequences were generated within the context of Paer2. Only sequences that differed in the −35 and −10 were mutated, leaving the intervening, upstream and downstream regions like those of the Paer2 promoter plasmid pVI1013. Transcription using these DNA templates was then compared with that from the native Paer2 promoter template in multiple-round in vitro transcription assays (Fig. 3B). The P. putida consensus and the aer2 promoter templates only differ at two positions in the −35 region and they also mediated production of approximately the same amount of transcripts. The E. coli consensus deviates more: by five bases in the −35 region and one in the −10 region. Surprisingly, reconstituted P. putida FliA-RNAP generated ∼2.5-fold higher levels of transcripts with the E. coli consensus template than with the aer2 promoter or P. putida consensus templates (Fig. 3B). Thus, the identified P. putida FliA consensus appears suboptimal for recognition and transcription by P. putida FliA-RNAP. As expanded upon in the discussion, this may reflect a trade-off to accommodate wide-ranging regulation by FliA promoters by P. putida.

Figure 3.

Transcription of P. putida and E. coli consensus sequences by P. putida FliA-RNAP.
A. Comparison of the DNA sequence of the Paer2 promoter with the consensus sequences derived for P. putida (PP cons.) and E. coli (EC cons.) FliA-dependent promoters. Differences between the sequences, as discussed in the text, are highlighted by boxes; while grey shading indicates the universally conserved extended −10 motif (Koo et al., 2009).
B. Multiple-round in vitro transcription assays with 10 nM P. putida core RNAP, 40 nM FliA–His and 10 nM plasmid template: 1, Paer2 (pVI1013); 2, P. putida consensus within the context of the Paer2 (pVI1014); 3, E. coli consensus within the context of Paer2 (pVI1015). Data are the average of two independent experiments ± standard error, normalized by setting transcription of pVI1013 as 1.

An AT-rich upstream region of Paer2 is involved in both output and specificity for P. putida FliA-RNAP

AT-rich UP-like elements that are recognized by the α-subunits of RNAP have previously been shown to enhance transcription by RNAP utilizing FliA homologues in different bacteria, including P. putida (Rodriguez-Herva et al., 2009), Bacillus subtilis (Fredrick et al., 1995) and Chlamydia trachomatis (Shen et al., 2006), and the consensus for UP-elements has been determined for E. coli RNAP (Estrem et al., 1998). Rodriguez-Herva and colleagues (2009) suggested that an AT-rich upstream region of FliA promoters is required for transcription by FliA-RNAP in P. putida. The sequence upstream of the aer2 promoter is AT-rich and has some similarities to the UP-consensus (Fig. 4A). Therefore, it was of interest to determine the role of this region in performance of the Paer2 promoter both in vivo and in vitro.

Figure 4.

The Paer2 AT-rich UP-element is important for output and discrimination as a FliA-dependent promoter.
A. Schematic representation of the aer2 upstream region and the extent of the DNA in Paer2–luxAB transcriptional reporters (used in B) and in vitro transcription templates (used in C). The DNA sequence of the AT-rich consensus UP-element sequence in pVI998 and pVI1016 is highlighted in grey above the sequence it replaces in the native Paer2 counterparts pVI996 and pVI1012.
B. Growth (closed symbols) and luciferase activities (open symbols) of LB cultured KT2440 carrying transcriptional Paer2–luxAB reporter plasmids: pVI997 (circles), pVI996 (up-triangles), pVI995 (down-triangles), pVI998 (squares). The data are averages of duplicate determinations from each of two independent experiments with standard error. Growth of all strains was indistinguishable from the average shown (filled squares).
C. Multiple-round in vitro transcription assays with 10 nM P. putida core RNAP and 40 nM FliA–His (open bars), or 25 nM P. putida core RNAP and 40 nM σ70 (hatched bars), with 10 nM plasmid DNA template carrying the corresponding regions as used in the transcriptional reporter plasmids in (B). Namely: 1, the Paer2 promoter only (pVI1013; circles in B); 2, the Paer2 promoter and native AT-rich upstream region (pVI1012; up-triangles in B); 3, the Paer2 promoter, native AT-rich upstream region and additional upstream DNA (pVI1011; down-triangles in B); 4, the Paer2 promoter with its native AT-rich upstream region replaced by the AT-rich UP-element (pVI1016; squares in B). The data are the averages of two independent experiments with standard error.

For the in vivo analysis, four Paer2–luxAB transcriptional reporter plasmids were generated. The extent and nature of the Paer2 promoter region present in these constructs is schematically illustrated in Fig. 4A. Plasmid pVI995 and pVI996 both carry the entire AT-rich region and the Paer2 promoter region and both are equivalently transcribed when introduced in P. putida KT2440 (Fig. 4B, open triangles). Plasmid pVI997 lacks the AT-rich region but retains the Paer2 promoter; however, like the null mutant of FliA, deletion of the AT-rich region essentially abolished transcription from Paer2 in vivo (Fig. 4B, open circles). In contrast, plasmid pVI998 – in which the native AT-rich region of Paer2 is replaced by an AT-rich consensus UP-element – was highly transcribed. Enhanced transcription due to the consensus AT-rich UP-element was most notable in the exponential phase and resulted in approximately eightfold higher transcription in P. putida KT2440 during this phase of growth (Fig. 4B, open squares). These results demonstrate that the AT-rich promoter upstream region of Paer2 is essential for transcription of Paer2 in P. putida KT2440, and suggest that it likely serves as an UP-element for FliA-RNAP.

To allow analysis with a purified system, the same Paer2 promoter regions used in the transcriptional reporter plasmids were also incorporated into in vitro transcription plasmids and used as templates for in vitro transcription assays. As shown in Fig. 4C (open bars), the same trends as observed in vivo were also observed in vitro. However, the differential performance of the four promoter templates was muted as compared with the in vivo result. In particular, transcripts were readily detected with the template lacking the AT-rich region, and the presence of the AT-rich UP-consensus rather than the native AT-rich region of Paer2 only resulted in marginally higher levels of transcripts. Muted differences are frequently observed in vitro where promoter activities are analysed in isolation, as compared with the in vivo situation where promoter performance occurs against a background of competition by other promoters for the cognate holoenzymes. Nevertheless, the notably poor enhancing effect of the consensus UP-element sequence in vitro prompted us to determine the consequences of the changes to the Paer2 promoter on specificity. Hence, we also tested the four templates with σ70-RNAP. As shown in Fig. 4C (hatched bars), possession of the consensus AT-rich UP-element changed the specificity of the Paer2 from an almost completely FliA-dependent promoter to a dual specificity promoter that can be recognized and transcribed by holoenzymes utilizing either FliA or σ70. These results suggest that the great stimulatory effect of this region in vivo is, at least in part, due to cross-recognition and transcription of the Paer2 promoter by σ70-RNAP.

The global regulators ppGpp and DksA influence motility and transcription of aer2

Global transcriptional profiling has highlighted a role for ppGpp and DksA in control of motility and taxis related genes in E. coli (Aberg et al., 2009). Lack of either ppGpp and/or DksA produces notable effects on motility and on transcription of the E. coli aerotaxis receptor gene aer (Aberg et al., 2009). Transcription of one of the two aer genes that mediate aerotaxis in Pseudomonas aeruginosa requires the FNR homologue ANR (Hong et al., 2004). Because Aer2 is the aerotaxis receptor as well as metabolism-dependent taxis receptor in P. putida KT2440 (Sarand et al., 2008), these observations prompted us to test for potential control of aer2 transcription by ppGpp, DksA and ANR. As an initial approach, the motility of P. putida strains individually lacking each of the three global regulators was compared with that of the wild type and the Aer2 null mutant strain on soft (0.3%) agar plates containing M9-minimal medium with citrate or glucose as sole carbon source – two carbon sources to which the Aer2 null mutant exhibits defective metabolism-dependent taxis (Sarand et al., 2008). The absence of ANR had little or no effect, while the lack of ppGpp had a detrimental effect comparable to the loss of Aer2, and the absence of DksA had an even more severe effect (Fig. 5A). Because ppGpp, DksA and ANR are all global regulators, pleiotropic effects through multiple promoters likely underscore the severity of the phenotype observed in this assay.

Figure 5.

Effects by global regulators on motility and output from the Paer2 promoter.
A. Motility of P. putida KT2440 and counterparts lacking ppGpp, DksA, ANR or Aer2 after overnight growth on soft agar minimal medium containing citrate as the carbon source. Similar results were seen when glucose was used as the carbon source (data not shown).
B. Relative luciferase activities from the Paer2–luxAB reporter plasmid pVI995 when resident in LB-cultured KT2440 (white bars) or counterparts lacking ppGpp (ppGpp0, hatched bars), or DksA (dksA::Tc, grey bars). The results are the average ± standard error of determinations made from two independent cultures grown to exponential phase (OD600∼0.4) or stationary phase (OD600∼3.5). Data were normalized by setting values for wild-type KT2440 as 1.
C. Single-round in vitro transcription reactions with 10 nM core RNAP, 40 nM FliA, increasing concentrations of DksA, and 10 nM plasmid pVI011 that carries the same Paer2 promoter region DNA as the reporter plasmid used in (B).
D. Single-round in vitro transcription reactions as in (C), but with either increasing concentrations of ppGpp alone (open circles), or with 0.5 µM DksA and increasing concentrations of ppGpp (closed squares). The data in (C) and (D) are the average of two independent experiments with standard error and are given in arbitrary units with that in the absence of any additions of ppGpp or DksA set as 1. Inserts show autoradiograms of Paer2-derived transcripts from one of the two experiments.

Next we used the Paer2–luxAB transcriptional reporter plasmid pVI995 to determine the impact of the three global regulators on the performance of the Paer2 promoter. As anticipated, transcription from Paer2 in P. putida KT2440 lacking ANR was indistinguishable from that in the wild-type strain during growth on rich media (data not shown). However, lack of either ppGpp or DksA resulted in approximately twofold reduced transcription from Paer2 in the post-exponential phase of growth (Fig. 5B). It has previously been experimentally demonstrated that in P. putida KT2440, as in E. coli, ppGpp is rapidly synthesized at the exponential-to-stationary phase transition, while DksA levels remain relatively constant over the entire growth curve under the rich media culture condition used here (Sze and Shingler, 1999; Bernardo et al., 2009). Thus, these results suggest that DksA, directly or indirectly, primarily mediates its effects on Paer2 promoter output through co-action with ppGpp.

DksA can function independently of ppGpp to affect transcription through altering binding of RNAP to promoters (Aberg et al. 2008 and references therein), and also sensitizing RNAP to the effects of ppGpp to account for their concerted effects on transcription (Paul et al., 2004; 2005). To determine if ppGpp and/or DksA can directly modulate transcription from the FliA-dependent Paer2 promoter, single-round in vitro transcriptions were performed under conditions that have previously been demonstrated to recapitulate the known negative, positive and co-regulatory effects of ppGpp and DksA on σ70-dependent transcription in vitro (Johansson et al., 2008; Bernardo et al., 2009). Increasing concentrations of either DksA alone (Fig. 5C) or ppGpp alone (Fig. 5D, open circles) directly stimulated transcription from Paer2 by FliA-RNAP. However, the greatest stimulation (3.5- to 4-fold) was observed when both ppGpp and DksA were present in the reaction mixes (Fig. 5D, filled squares). In this case, the presence of DksA allowed concentrations of ppGpp that elicited little stimulation on their own (e.g. 200 and 400 µM) to promote near maximal stimulation under the conditions used. The results above suggest that, like ppGpp/DksA co-stimulated σ70 promoters, DksA sensitizes FliA-RNAP to ppGpp which functions to lower the transition energy required for formation of a rate-limiting intermediate on the route to open-complex formation (Paul et al., 2005). To our knowledge, this is the first demonstration that FliA-dependent promoters can be directly regulated by ppGpp and DksA. The stimulatory effect of ppGpp and DksA on Paer2 output in vitro suggests that the approximately twofold decreased transcription seen in stationary-phase P. putida KT2440 derivatives lacking either of these regulatory molecules is due to the absence of their direct effects on Paer2.

Discussion

Pseudomonas putida KT2440 encodes an aerotaxis and metabolism-dependent taxis receptor called Aer2 and two other similar proteins, Aer1 and Aer3, which colocalize to one pole in a similar manner as the sensory chemotaxis receptors of E. coli. We found that transcriptional control of these three proteins is mediated by different σ-factors, with that of aer2 being driven through the flagella σ-factor FliA while transcription of aer1 and aer3 are controlled via σ70 (Figs 1 and 2, and Fig. S1). Thus, since FliA-RNAP is involved in coordinate expression of the flagella genes, FliA-mediated transcription of Aer2 also coordinates production of this taxis-coupled receptor with the motility apparatus of the bacterium. Aer2 is the most abundantly expressed of the three Aer proteins under the conditions used in this study and, unlike the defects seen upon loss of Aer2, lack of Aer1 or Aer3 mediates no apparent phenotype under the conditions tested (Sarand et al., 2008). However, it is likely that all three Aer-like proteins are involved in some form of energy sensing and their relative abundance may be modulated by other growth conditions (Sarand et al., 2008) and/or the activities of associated regulatory genes, although this remains to be experimentally verified. Differential regulation involving the use of alternative σ-factors has previously been found for the two aerotaxis aer genes of P. aeruginosa. In P. aeruginosa transcription of one of the aer genes is completely dependent on the alternative stationary/stress σ-factors σS, while activity of the other is controlled through ANR, a homologue of E. coli oxygen sensor FNR that functions as a switch between aerobic and anaerobic growth (Hong et al., 2004; 2005).

The σ28 promoter consensus sequence has been determined for several different bacterial species, including E. coli (Ide et al., 1999; Yu et al., 2006; Zhao et al., 2007), B. subtilis (Gilman et al., 1984), C. trachomatis (Shen et al., 2006; Yu et al., 2006) and P. putida (Rodriguez-Herva et al., 2009). Recently, it has been shown using E. coli FliA (σ28) that the unusually long −10 regions of these consensuses are a composite of a core −10 region and an extended −10 motif which are recognized by universally conserved residues of σ-domain 3 and region 2.4 of the σ28-subunit (Koo et al., 2009). The P. putida consensus (Fig. 3A), while retaining the conserved extended −10 motif, is unusual in possessing a C, rather than an A, residue at the penultimate position of the −35 consensus (TCAAGT). This C residue is found in Paer2 and all but one of the other seven P. putida sequences used to derive the consensus (Rodriguez-Herva et al., 2009). The Paer2 promoter is close to the consensus for P. putida FliA and changes of two −35 residues to produce an exact match had little if any effect on transcription mediated by P. putida FliA-RNAP. However, changes made to Paer2 to match the E. coli consensus resulted in more than 2.5-fold enhanced transcription with the reconstituted P. putida holoenzyme (Fig. 3B). This unexpected finding is consistent with mutational analysis of FliA-dependent transcription of the hctB promoter in which substitution of the penultimate −35 region A residue to a C decreases transcription by both C. trachomatis and E. coli holoenzymes (Yu et al., 2006). Suboptimal performance of P. putida FliA-RNAP on the Paer2 promoter and the P. putida consensus promoter in comparison with the E. coli counterpart suggests that the majority of FliA-dependent transcription in P. putida is below the maximum that could potentially be achieved from FliA promoters. The P. putida consensus was derived from analysis of 18 promoter regions that control 23 genes observed to be downregulated in a transcriptomics analysis of the P. putida KT2440 FliA null mutant (Rodriguez-Herva et al., 2009). Of these, only eight possessed FliA promoter sequences and thus appeared to be directly controlled by FliA-RNAP. While the majority of P. putida genes controlled by FliA are related to taxis and motility, others are potentially involved in amino acid utilization, stress responses, regulation, or have unknown function (Rodriguez-Herva et al., 2009). Pseudomonas putida is not unique in utilizing FliA to directly and/or indirectly control genes other than those involved in flagella biosynthesis and taxis responses. For example, FliA homologues controls production of a lipase and a protease in Xenorhabdus nematophilia, and are involved in stress responses, sporulation and developmental processes in other bacteria (Josenhans et al., 2002; Yu and Tan, 2003; Serizawa et al., 2004; Park and Forst, 2006). Thus, it is possible that P. putida has evolved suboptimal FliA promoter sequences as part of a trade-off to balance FliA promoter output to multiple traits of the cell to allow competitive success. In this respect it is interesting that screening for lower death frequencies in the absence of stress resulted in the isolation of an E. coli FliA null mutant that out-competed its wild-type counterpart (Fontaine et al., 2008).

In addition to recognition of the −35 region and the composite extended −10/core −10 motif by FliA (Koo et al., 2009), a third promoter element – the so-called UP-element – influences the output from FliA promoters. However, it is the highly conserved α-subunits of the core RNAP (Murakami et al., 1996), rather than the σ-factor, that binds the UP-element. Analogous to the functioning of UP-elements located at −61 to −41 upstream of σ70 promoters (Ross et al., 1993; Gourse et al., 2000), AT-rich regions upstream of FliA promoters increases the promoter affinity of FliA-RNAP in B. subtilis (Fredrick et al., 1995) and C. trachomatis (Shen et al., 2006), and are associated with higher FliA promoter output in P. putida (Rodriguez-Herva et al., 2009). Although AT-rich upstream regions are associated with FliA promoters in P. putida, no common consensus could be identified (Rodriguez-Herva et al., 2009). Consistent with serving as an UP-element, we found that lack of the AT-rich region of the Paer2 region essentially abolishes detectable transcription from Paer2 in vivo and decreased promoter output in vitro (Fig. 4). Strikingly, however, we also found that replacement of the native AT-rich region by a consensus UP-element (Estrem et al., 1998) greatly increased transcription from this promoter in vivo, and also rendered the normally FliA-specific Paer2 promoter susceptible to recognition and transcription by σ70-RNAP in vitro. This finding suggests that recognition of the AT-rich region by RNAP does more than simply compensates for suboptimal binding through DNA elements recognized by the σ-factor, it can also be involved in maintaining exclusivity of the promoter for FliA-RNAP. What properties of FliA promoter architecture counteract the more promiscuous promoter recognition and transcriptional properties of σ70-RNAP are unknown, but could be related to the use of G and/or C residues at the upstream end of the core −10 motif, which are never found in this position in σ70 promoters (Koo et al., 2009). The AT-rich regions of FliA promoters such as Paer2 may have evolved to play two roles, namely to provide sufficient affinity for FliA-RNAP and to prevent consequent potential promiscuous transcription by σ70-RNAP. Two non-mutually exclusive mechanisms to achieve the latter include not exceeding a critical limit of affinity so as to overcome other limitations of σ70-RNAP activity at the promoter, or to create promoter–RNAP complexes that disfavour σ70-RNAP activity but do not adversely affect that of FliA-RNAP.

The Paer2 promoter controls the production of an aero- and metabolism-dependent taxis receptor. The Aer2 receptor is probably important for the bacterium to be able to locate environments where sufficient energy sources are available and oxygen levels are appropriate for their catabolism (Sarand et al., 2008) – a process that would aid avoidance of nutritional stress. By analogy, production of the bacterial alarmone ppGpp, which immediately precedes nutrient depletion and thus heralds the onset of metabolic stress, allows the bacterium to preload itself with factors to prepare to counteract hard times ahead (Gourse et al., 2006; Potrykus and Cashel, 2008). It is therefore perhaps not surprising that we found that ppGpp in conjunction with DksA directly stimulates transcription from the aer2 promoter (Fig. 5). Superimposing ppGpp/DksA stimulation on the FliA-dependent Paer2 promoter creates a scenario where the presence of the receptor is both coordinated with production of the motility apparatus and ensured when condition cue the need to relocate to a more energetically favourable environment.

Experimental procedures

Bacterial strains, growth conditions and general procedure

Escherichia coli and P. putida were grown at 37°C and 30°C respectively. Prototrophic E. coli DH5 (Hanahan, 1985) was used for construction and maintenance of all but R6K-based suicide plasmids, which were maintained and conjugated into P. putida strains from the replication-permissive but auxotrophic S17λpir E. coli host (de Lorenzo and Timmis, 1994). Pseudomonas putida KT2440 and previously constructed derivatives incapable of synthesizing ppGpp or defective in production of individual proteins are listed in Table 1. Luria–Bertani/Lennox (LB) (AppliChem GmbH) was used as rich medium and was supplemented with carbenicillin (Cb, 100 µg ml−1 for E. coli strains and 1 mg ml−1 for P. putida strains), kanamycin (Km, 50 µg ml−1 for both E. coli and P. putida strains), tellurite (Tel, 40 µg ml−1 for P. putida strains), tetracycline (Tc, 5 µg ml−1 for E. coli strains and 8 µg ml−1 for P. putida strains), IPTG (0.1 or 0.5 mM) and/or L-arabinose (0.2%) were added to the media as required. Motility assays were performed on minimal M9-salts soft-agar plates containing 0.3% agar and citrate (25 mM) or glucose (0.2%) as carbon sources as previously described (Sarand et al., 2008).

Table 1.  Parental P. putida KT2440-based strains.
StrainRelevant propertiesReferences/source
  1. Gm, gentamicin; Km, kanamycin; Tc, tetracycline; Tel, tellurite.

KT2440Prototrophic, plasmid cured strainFranklin et al. (1981)
KT2440-AKT2440 ANR null; anr::Tc insertion mutantUgidos et al. (2008)
KT2440-fliA::KmKT2440 FliA null; fliA::Km insertion mutantRodriguez-Herva et al. (2009)
KT2440-dmpR-telKT2440 tagged with miniTn5-Tel carrying dmpRSze et al. (2002)
PP1:dmpRKT2440 ppGpp0; ΔrelA::Km ΔspoT::Gm derivative of KT2440-dmpR-telSze et al. (2002)
PP2:dmpRKT2440 DksA null; dksA::Tc derivative of KT2440-dmpR-telJohansson et al. (2008)
PP2062KT2440 carrying araC/PBAD-rpoD linked to a TcR markerLaboratory collection

DNA manipulations

Plasmids were constructed by standard techniques. The fidelity of all polymerase chain reaction (PCR)-derived or double-stranded linker DNA was confirmed by DNA sequencing. Luciferase transcriptional reporter plasmids, based on the CbR promoter probe vector pVI928 (Johansson et al., 2008), and in vitro transcription plasmid generated using pTE103 (Elliott and Geiduschek, 1984) are listed in Table 2. Oligonucleotides used to generate plasmid inserts can be found in Table S1.

Table 2.  Key reporter and in vitro transcription plasmids.
PlasmidRelevant propertiesReferences/source
  • a. 

    Fragments were cloned using the XhoI or SmaI (blunt end), and the BglII (BamHI-compatible ends) sites of the vector.

  • b. 

    Fragments were cloned using EcoRI, SmaI and BamHI sites unless otherwise stated.

  • B 

    Indicates where DNA ends were pre-treated with Klenow to generate blunt ends.

Luciferase transcriptional reporter plasmids based on CbR pVI928a
pVI987aer1 operon-luxAB; XhoI to BglII fragment generated using primers 2364 and 2423
pVI990aer3 operon-luxAB; XhoI to BamHI fragment generated using primers 2368 and 2369
pVI995aer2–luxAB fusion carrying Paer2, its native AT-rich region and additional upstream DNA; EcoRIB to BamHI fragment from pVI1011
pVI996aer2–luxAB fusion carrying Paer2 and its native AT-rich region; XhoI/BamHI linker 2372/2373
pVI997aer2–luxAB fusion carrying Paer2 only; XhoI/BamHI linker 2374/2375
pVI998aer2–luxAB fusion carrying Paer2 with an AT-rich UP-element consensus; XhoI/BamHI linker 2370/2371
In vitro transcription plasmids based on CbR pTE103b
pVI948σ70-dependent Pr promoter (Johansson et al., 2008)
pVI1000aer1 operon upstream region; XhoIB to BglII fragment from pVI987
pVI1005aer3 operon upstream region; EcoRI to StuI fragment from pVI990 between the EcoRI and HindII sites
pVI1010Paer2 promoter with its native AT-rich region and additional upstream DNA; EcoRI to BamHI fragment generated using primers 2286 and 2287
pVI1011Paer2 promoter with its native AT-rich region and additional upstream DNA; EcoRI to BamHI fragment generated using primers 2286 and 2332
pVI1012Paer2 and its native AT-rich region; EcoRI/BamHI linker 2333/2334
pVI1013Paer2 only; EcoRI/BamHI linker 2335/2336
pVI1014P. putida FliA consensus promoter sequence within the context of Paer2; EcoRI/BamHI linker 2503/2504
pVI1015E. coli FliA consensus promoter sequence within the context of Paer2; EcoRI/BamHI linker 2505/2506
pVI1016Paer2 with the AT-rich UP-element consensus; SmaI/BamHI linker 2370/2371 between the EcoRIB and BamHI sites of the vector

P. putida KT2440 derivatives lacking σ-factors

Pseudomonas putida strains with insertions within individual σ-factor encoding genes were constructed as previously described for inactivation of σ54 (Johansson et al., 2008). Internal portions of the individual genes were PCR amplified as BglII to SpeI fragments and cloned between the corresponding sites of the kanamycin-resistant R6K-based suicide plasmids pDM4-Km (Sarand et al., 2008), to generate the plasmids listed in Table S2. To create insertions, single-site recombination into the chromosome of P. putida KT2440:dmpR-Tel was performed by conjugation from E. coli S17λpir. Tellurite and kanamycin were used as counter selection for the donor and recipient parental strains. Since the suicide plasmids carry only internal portions of the target genes, the resulting strains contain two inactive truncated copies of the gene separated by plasmid DNA. Diagnostic PCR of strains using primers homologous to regions upstream of the gene fragments on the suicide plasmids and to the plasmid encoded kanamycin resistance gene was used to confirm correct recombination into the target gene. Plasmid-encoded kanamycin resistance selection was used to ensure maintenance of the integrated plasmids.

Luciferase reporter strains

Tetracycline-resistant monocopy reporter strains with the luxAB genes fused immediately downstream of aer1, aer2 and aer3 were constructed by single-site recombination as previously described for otherwise identical kanamycin-resistant counterparts (Sarand et al., 2008). In brief, BglII to SpeI fragments spanning the 3′-ends to the last codons of the aer genes were excised from pVI805-807 and cloned between the corresponding sites of the TcR R6K-based suicide plasmid pTc805L (Sarand et al., 2008). These plasmids, carrying aer1-luxAB (pV984), aer2–luxAB (pVI985) or the aer3-luxAB (pVI986) fusions, were then conjugated from E. coli S17λpir to P. putida KT2440 derivatives. Prototrophy and tetracycline were used as counter-selection for the donor and recipient parental strains. Diagnostic PCR of the resulting strains using primers homologous to regions upstream of the gene fragments of the suicide plasmids and to the luxAB genes were used to confirm correct recombination into the target gene. Luciferase transcriptional reporter plasmids, constructed by placing promoter regions upstream of the promoterless luxAB genes on the CbR broad-host-range promoter probe vector pVI928 (Table 2), were introduced into P. putida strains by electroporation.

Luciferase assay

Strains were grown overnight at 30°C in rich LB medium and then diluted and cultured under the same conditions. To make sure that the growth was balanced, cultures were grown to exponential phase and then diluted again before starting the measurements. Light emission was measured from 100 µl whole cells after addition of decanal (1:2000 dilution) using a PhL Luminometer (Aureon Biosystems) or Infinite M200 (TECAN).

Transmission electron microscopy

Bacteria, grown overnight on LB-agar plates containing 0.1 mM IPTG, were resuspended in Tris pH 7.4 (10 mM) and MgCl2 (10 mM). Aliquots (4–5 µl) were placed on formvar-coated Cu-grid (Embra grids, TAAB Laboratories Equipment) and allowed to settle for approximately 3 min prior to removal of surplus liquid and addition of 1% sodiumsilicotungstate pH 7.4 (TAAB Laboratories Equipment). After 10 s incubation, excess liquid was removed and the bacteria analysed and visualized with a JEM 1230 transmission electron microscope (Jeol).

FliA–His expression plasmids

Plasmid pVI982, carrying the fliA gene under the control of PT7 promoter, was constructed by PCR amplification of the fliA coding region (PP_4341) as an NdeI to BamHI fragment and cloning between the corresponding sites of the CbR pET-based expression vector pET3H-His (Johansson et al., 2008). The resulting plasmid expresses P. putida FliA with a GSHHHHHH extension on its C-terminal end. The NdeI to HindIII fragment spanning the fliA–His tag fusion was excised from pVI982 and cloned between these sites of the CbR broad-host-range expression pVI520 (O'Neill et al., 1998). The resulting plasmid, pVI983, expresses the fliA–His tag fusion from the IPTG inducible lacIQ/Ptac expression system of the vector.

Purified reagents for in vitro assays

Nucleotides and [α-32P]-UTP were purchased from Roche Molecular Biochemicals and from Perkin Elmer, respectively, while ppGpp was synthesized and purified as previously described (Cashel, 1974). Purification of P. putida-derived native σ70, native core RNAP and N-terminal His-tagged DksA (His-DksA) has previously been described (Johansson et al., 2008; Bernardo et al., 2009). Pseudomonas putida FliA–His was purified from E. coli BL21(DE3)pLysS (Invitrogen) carrying pVI982 essentially as described for DmpR-His (Wikström et al., 2001), using successive affinity purification through HisPure cobalt spin columns (Pierce) and Ni2+-chelating resin (Qiagen). The resulting FliA–His preparation was stored at −80°C in buffer (50 mM Tris-HCl pH 8.0, 500 mM NaCl, 2 mM EDTA, 2 mM DTT) containing 50% glycerol. Protein concentration was determined using a BSATM Protein Assay Kit (Pierce) with bovine serum albumin (BSA) as a standard.

In vitro transcription assays

Transcription assays were performed at 30°C essentially as described previously (Bernardo et al., 2006) using supercoiled pTE103-based plasmids as DNA templates (Table 2). Assays of a final volume of 20 µl were performed in T-buffer (35 mM Tris-Ac pH 7.9, 70 mM KAc, 5 mM MgAc2, 20 mM NH4Ac, 1 mM DTT and 0.275 mg ml−1 BSA). Core RNAP and σ-factors were pre-mixed and incubated for at least 5 min to allow holoenzyme formation. When used, ppGpp, DksA and/or DksA storage buffer were added to holoenzyme mixes and incubated for 5 min prior to addition of template DNA. Reactions were incubated for 20 min to allow open-complex formation. For multiple-round assays, transcription was initiated in the presence of anti-RNase (Ambion) by the addition of NTPs (final concentration: ATP, 500 µM; GTP and CTP, 200 µM each; UTP, 80 µM and [α-32P]-UTP, 5 µCi at > 3000 Ci mmol−1) and the reactions incubated for 10 min prior to addition of heparin (0.1 mg ml−1) to prevent re-initiation. After a further 8 min at 30°C, the reactions were terminated by adding 5 µl of a stop/load mix [150 mM EDTA, 1 M NaCl, 14 M urea, 3% glycerol, 0.075% (w/v) xylene cyanol, 0.075% (w/v) bromophenol blue]. For single-round assays, heparin was added at the same time as the NTPs and reactions were included for 10 min prior to termination of the reactions. Transcripts were analysed on 7 M urea/5% (w/v) polyacrylamide sequencing gels and quantified using a Molecular Dynamics Phosphoimager.

Acknowledgements

We are grateful to Juan-Luis Ramos and co-workers for providing the P. putida KT2440 fliA::Km mutant and information prior to publication, and Fernando Rojo for the P. putida KT2440 ANR mutant. We also wish to thank Lenore Johansson (Umeå University Electron Microscopy Facility) for help with obtaining the images, Kristina Ygberg for participation in construction of the P. putida KT2440 σ-factor mutant strains, and Linda Holmfeld for the rpoD overexpression strain. This work was supported by the grants from J.C. Kempes foundation and the Swedish Research Council (VR).

Ancillary