Sigma factor σ54 has a distinct modus operandi for mediating the activation of bacterial RNA polymerase (RNAP) at promoter recognition motifs 12 and 24 bp upstream from transcription start sites. σ54 was thought to act as monomer in all transcription steps. However, we provide evidence that σ54 of Pseudomonas putida interacts stably with itself. The interface between monomers involves contacts in σ54 regions I and III, sequences that play key roles in the transcription activation of σ54–RNAP holoenzyme. These roles include interactions with activator proteins and the −12 and −24 motifs. In particular, we detected inter-monomer contacts between region I, and between region I and the C–terminal portion of region III. Our results suggest a new auto-antagonistic regulatory state of σ54.
Structured digital abstract
- σ54 and σ54 bind by molecular sieving (View Interaction: 1, 2)
- σ54 and σ54 bind by blue native page (View interaction)
- P. aeruginosa σ54 physically interacts with P. aeruginosa σ54 by two hybrid (View interaction)
- σ54 physically interacts with σ54 by two hybrid (View Interaction: 1, 2, 3)
- p53 physically interacts with SV40T by two hybrid (View interaction)
In bacteria, transcription is performed by a multi-subunit enzyme, RNA polymerase (RNAP), encompassing the enzymatic core, where polymerase activity is housed, and one of the σ factors that are involved in recognition of and the binding to specific promoter sequences, in the interaction with transcription activators, in promoter melting near the transcription start site, in the inhibition of non-specific initiation and in the regulation of early phases of transcription . σ factors have been classified into two major classes, σ70 and σ54. Most factors belong to the σ70 class [2-4], with σ54 being the only member of its class. σ54 shares little sequence similarity and no promoter specificity with the σ70 class. Large-scale comparative genomic analyses indicate that σ54 is widely distributed among bacteria, and strongly suggest that it is an important regulator of bacterial envelope structures, as it is associated with genes involved in transport and biosynthesis of the metabolic intermediates of exopolysaccharides, flagella, lipids, lipopolysaccharides, lipoproteins and peptidoglycan .
The σ54–RNAP binds to a promoter at consensus motifs −24(GG) and −12(TGC) upstream of the transcription start site, and forms a stable closed complex ; at a later stage, this isomerizes into an open complex by interaction with specialized ATP-hydrolyzing σ54 activators  that bind to enhancer-like elements known as upstream activating sequences.
The most studied σ54 is from Escherichia coli. Its primary sequence (477 amino acids) has been sub-divided into three regions based on function. Region I (amino acids 1–56) interacts with σ54 activators and the −12 promoter motif, where DNA melting starts [8-11]. Region II (amino acids 57–120) has no clear function; it is the least conserved and is sometimes absent in σ54 proteins from other species. Region III (amino acids 121–477) has functional modules that include determinants that are able to bind core RNAP (amino acids 120–215) and promoter DNA (amino acids 329–463). In particular, the C–terminal module contains a sub-region (amino acids 329–346) that may be cross-linked to DNA and a signature motif for σ54 proteins called an RpoN box (amino acids 454–463) [12, 13], which has been shown to contain a classical helix-turn-helix module that is able to bind to the −24 promoter elements [14, 15]. In σ54–RNAP, σ54 region I localizes in the channel between the β and β' subunits and prevents entry of single-stranded template DNA into the active site of RNAP [10, 11]. This accounts for the inability of the closed complex of σ54–RNAP to transcribe. Region I is rearranged following interaction with an ATP-hydrolyzing activator, and no longer blocks the active site of the enzyme. In addition, region I/activator interactions also induce the region III C–terminal DNA-binding domains of σ54 that interact with the −24 motif to move downstream, relocating the −12 motif closer to the RNAP site of template loading.
As described above, when docked in core RNAP and in the presence of an activator bound to upstream activating sequences, σ54 plays a positive role in transcription initiation. However, negative roles of σ54 have been also reported. In Pseudomonas aeruginosa, σ54 was shown to play a negative regulatory role in AlgU(σE)-dependent transcription of the PalgD promoter . In addition, when the transcriptome of Geobacter sulfurreducens over-expressing σ54 was analyzed , the expected general increase in expression of σ54-dependent genes was observed together with a certain degree of down-regulation of a number of genes that were predicted as being downstream of σ54-dependent promoters. These examples support the possibility that σ54 is involved in a complex regulatory network with a role that is not congruent with that described above for its association with RNAP.
Whatever the context, σ54 from E. coli and closely related bacteria is thought to act as monomer [18, 19]. However, in this study, we provide evidence that σ54 of Pseudomonas putida interacts with itself to form multimers. In addition, we characterized the contacts between monomers and found that they include regions I and III, which both play key roles in transcription activation. The equilibrium between monomer and multimer states may sustain novel regulatory activities of σ54.
Results and Discussion
Early in vitro evidence of multimer assembly of σ54
Purified σ54 of P. putida KT2440  was analyzed by blue native and SDS/PAGE. In each case, the protein was detected by both western blotting, using mouse antibodies raised against σ54, and Coomassie Brilliant Blue staining. As shown in Fig. 1A, σ54 migrated under denaturing conditions as single band, i.e. neither contaminant proteins nor degradation forms were detected. However, under native conditions, western blotting clearly shows that σ54 migrates as at least three distinct bands with different intensities (Fig. 1B). These results provide the first evidence of σ54 self-assembly. The monomer may correspond to the most abundant and fastest-migrating band in Fig. 1B. To obtain further insight into the size of σ54 higher-order aggregates, analytical gel filtration experiments were performed (Fig. 1C), and showed close correspondence with the migration pattern of σ54 in native gels (Fig. 1B). The gel-filtration elution profile of purified σ54 showed a shoulder that eluted earlier than the primary peak (1; Fig. 1C), corresponding to the monomer. Deconvolution of the elution profile described the shoulder as a secondary peak (2; Fig. 1C) of approximately 239 kDa, a molecular mass consistent with self-association of σ54 as tetramers. The elution profile also showed an additional peak (3; Fig. 1C), accounting for approximately 2% of the protein, whose molecular mass approximately corresponded to a σ54 16-mer. The monomer–multimer equilibrium observed in these in vitro assays may not represent physiologically relevant concentrations of σ54 multimers. However, the monomer–multimer equilibrium was evaluated in the presence of high levels of glycerol, a component that was supplied during purification to prevent protein precipitation. We suggest that glycerol may antagonize the tendency of σ54 to form higher-order aggregates, displacing the equilibrium towards monomers. We speculate that in vitro conditions closer to physiological ones may give rise to higher proportion of multimers at lower protein concentrations. Moreover, in vivo macromolecular crowding  is expected to further displace the equilibrium to give rise to physiologically relevant concentrations of σ54 multimers. Other approaches to assess multimeric state, such as dynamic light scattering and nano-electrospray ionization, were attempted. Unfortunately, all of them were hindered by the high glycerol concentration of the purified σ54 storage buffer.
Evidence of σ54 monomer–monomer interaction via yeast two-hybrid assay
Self-association of σ54 was evaluated in yeast two-hybrid experiments. A wild-type sequence of the rpoN gene from P. putida KT2440 was cloned in the yeast two-hybrid vectors pEG202 (‘bait’) and pJG4–5 (‘prey’), to give pEG202–rpoN and pJG4–5–rpoN, respectively. Then the yeast strain EGY48, containing the reporter plasmid pSH18–34, was transformed with various combinations of bait and prey plasmids: (a) pEG202 and pJG4–5, (b) pJG4–5 and pEG202–rpoN, (c) pEG202 and pJG4–5–rpoN, and (d) pEG202–rpoN and pJG4–5-rpoN. As shown in Fig. 2A, a strong σ54/σ54 interaction, comparable to that occurring in the control two-hybrid experiment between human p53 and simian virus 40T antigen (SV40T) [22, 23], was detectable. A slight activation of the reporter gene was apparent in the strain expressing only the σ54 hybrid with the LexA DNA-binding domain from pEG202-rpoN (Fig. 2A). This effect may resemble that described previously  for a fragment of E. coli σ54, encompassing amino acid residues 29–177, which is able to activate transcription in yeast as a mono-hybrid fused to the Gal4 DNA-binding domain.
To assess whether σ54 self-association activity was specific to P. putida or is shared with other bacterial species, we cloned the rpoN genes of P. aeruginosa and E. coli in the same yeast two-hybrid system. As shown in Fig. 2C, the P. aeruginosa σ54/σ54 interaction was as strong as for P. putida, suggesting that this feature may be conserved in the Pseudomonas genus. In contrast, an E. coli σ54/σ54 interaction was undetectable in this assay.
Mapping the domains involved in the P. putida σ54/σ54 interaction
Based on several functional studies [6, 25], the σ54 primary sequence has been divided into three regions: I, II and III (Fig. 2B). In region III, two sub-regions have been recognized as being involved in strong RNAP interactions and promoter DNA interactions, respectively. These sub-regions are referred to as IIIA and IIIB, respectively.
To identify which σ54 region(s) are involved in the σ54/σ54 interaction, we generated progressive deletions of regions I and II from the N–terminus and of sub-region IIIB from the C–terminus in pJG4–5–RpoN (Fig. 2B). These deletions were tested for interaction with full-length σ54. Region I or IIIB deletions did not affect the interaction with full-length σ54 (Fig. 2B). Similar results were obtained (data not shown) for deletion of both regions I and II (rpoNΔI–ΔII). However, simultaneous deletions of regions I and IIIB resulted in a loss of interaction with full-length σ54. The results indicated that the σ54/σ54 interaction simultaneously involves at least the two terminal regions I and IIIB, respectively. Taken together, these results suggested the possible interactions ‘a’ (I/I), ‘b’ (IIIB/IIIB) and ‘c’ (I/IIIB) between σ54 monomers (Fig. 3A).
To assess the interactions ‘a’, ‘b’ and ‘c’, and other potential interactions involving regions II and IIIA, all four σ54 regions were cloned singly into both pEG202 and pJG4–5. Each region was tested in the reporter strain EGY48 in combination with the others, itself and full-length σ54. The reciprocal match was performed for any heterogeneous combination. Negative controls, consisting of any prey vector carrying a single region combined with an empty bait vector and vice versa, were also performed. As shown in Fig. 3B, only regions I and IIIB strongly interacted with full-length σ54, respectively. This provides further confirmation of the involvement of regions I and IIIB in the σ54/σ54 interaction as previously suggested by the deletion experiments (Fig. 2B). Of the proposed interactions involving I and IIIB (Fig. 3A), we reliably detected interactions ‘a’ (I/I) and ‘c’ (I/IIIB). The reporter activity in the combination of pJG4–5–IIIA with pEG202–II (Fig. 3B) may suggest an interaction between regions IIIA and II. However, this may be an artifact as: (a) no reporter activity was observed for the reciprocal combination of pJG4–5–II with pEG202–IIIA, and (b) the interactions of both regions II and IIIA with full-length protein are negligible. We may reasonably speculate that fusion of region II with the LexA DNA-binding domain in pEG202–II generates a domain that interacts appreciably with IIIA. This artifactual domain may be responsible for the weak reporter activity observed in the strain that harbors pEG202–II in combination with pJG4–5–rpoN expressing the full-length protein. Together, these results suggest that interaction between σ54 monomers is mainly mediated by regions I and IIIB.
A model for σ54 self-association
We now propose a model of σ54 self-association in tetramers (Fig. 1C) that includes the monomer–monomer interactions I/I and I/IIIB found by yeast two-hybrid assay (Fig. 3B). In our model (Fig. 4), the σ54 tetramer is envisaged as a ‘dimer of dimers’. Both interactions I/I and I/IIIB may be involved in either dimer stabilization and dimer–dimer interaction. Fig. 4 illustrates the case of the I/IIIB interaction stabilizing the dimer while the I/I interaction holds the dimers together. The reverse scenario in which the I/I interaction and the I/IIIB interaction are involved in dimer stabilization and dimer–dimer interaction, respectively, is also possible.
We suggest that the σ54 multimeric state is not effective in transcription initiation activation due to an auto-antagonistic effect (Fig. S1). As the interfaces between monomers involve σ54 regions that are instrumental to transcription activation of the σ54–RNAP holoenzyme, the multimeric state may influence such activity. In addition, σ54 may be unable to form productive interactions for docking in core RNAP in a multimeric form. In fact, the electron density assigned to σ54 in the σ54–RNAP holoenzyme was only compatible with the monomer [11, 26]. Given this σ54 auto-antagonism, the equilibrium between monomer and multimer states may be a switch that promptly regulates the levels of effective σ54 monomers, i.e. dockable and active in core RNAP, when the overall amounts of σ54 do not vary significantly in different physiological states . Any perturbation of the monomer–multimer equilibrium may have a positive or negative influence on σ54-dependent transcription with no involvement of σ54 activators or any change in σ54 expression levels. For instance, a physiological state that favors σ54 auto-antagonism would result in down-regulation of σ54-dependent promoters due to decreased levels of effective σ54. Enhancement of σ54 auto-antagonism during the exponential growth phase may sustain the ‘exponential silencing’ of σ54-dependent promoters described in P. putida . A physiological change at the onset of the stationary phase may reverse the silencing of σ54-dependent transcription mediated by auto-antagonism. In addition, the regulation of the effective σ54 levels by σ54 auto-antagonism may be linked to the so-called sigma cycle, in which competition of various sigma factors for binding to core RNAP partly determines the promoter specificity of transcription initiation .
In summary, although the genetic, biochemical and structural basis of σ54-dependent transcription have been extensively studied , we now provide evidence for a new property and state of σ54. σ54 of P. putida associates with itself to form multimers. In this paper, we suggest the geometry of self-association of P. putida σ54. This includes contacts between the functional regions I and IIIB that play key roles in the mechanism for activating σ54–RNAP. These results suggest novel layers of σ54-mediated regulation that may be based on an auto-antagonistic mechanism. The property of self-association is also detectable for σ54 of P. aeruginosa. The very high σ54 sequence conservation suggests that the σ54 auto-antagonistic mechanism may exist throughout the species belonging to genus Pseudomonas, and may be a component in the regulatory potential underlying their ubiquity and metabolic versatility [30, 31]. However, σ54 self-association activity was undetectable in E. coli, and consequently we suggest that σ54 auto-antagonism is absent in this bacterium. This may reflect differences in niche and adaptability compared with Pseudomonas. Further analysis of this novel potential role in P. putida requires the analysis of σ54 variants that are unable to form multimers yet retain the ability for transcription activation of the σ54–RNAP holoenzyme. Transcriptome comparison of such variants with wild-type σ54 will provide information on genes whose expression is influenced by the equilibrium between monomer and multimer states.
Bacterial and yeast strains, and vectors
The P. putida strain KT2440  is a plasmid-free derivative of P. putida mt–2 , the original isolate of P. putida harboring the archetypal TOL plasmid pWW0 . E. coli strains DH5a and JM109 were used as host strains for cloning experiments . pJG4–5, pEG202 and pSH18–34 are specialized vectors of the LexA-based MATCHMAKER yeast two-hybrid system (Clontech, Mountain View, CA, USA). Saccharomyces cerevisiae strain EGY48 (Clontech) was used as the host in two-hybrid experiments. The sources of the P. aeruginosa and E. coli rpoN genes were strains PAO1  and MG1655 , respectively.
Yeast two-hybrid assays
σ54-based hybrids linked to B42 activation and LexA DNA-binding domains (Table S1) were generated by cloning, into pJG4–5 and pEG202 respectively, of DNA fragments generated by PCR amplification from genomic DNA using Pfu high-fidelity DNA polymerase (Fermentas, Thermo Fisher Scientific, Inc., Waltham, MA, USA) and oligonucleotides listed in Table S2. For cloning, plasmids pJG4–5 and pEG202 and the DNA fragments were digested with EcoRI and XhoI (New England Biolabs, Ipswich, MA, USA), and ligated using T4 DNA ligase (Takara, Otsu, Shiga, Japan). To test interaction, pJG4–5- and pEG202-based plasmids were co-transformed into the yeast strain EGY48 carrying the vector pSH18–34, which harbors the sequence of the reporter gene lacZ downstream of eight LexA operators, and maintained on selective SD medium lacking without histidine, uracil and tryptophan (SD–HUW) supplemented with 2% glucose. Yeast two-hybrid experiments were performed by spotting yeast suspensions on yeast two-hybrid plates (SD–HUW, 2% raffinose, 2% galactose), supplemented with 20 mg·mL−1 5–bromo-4–chloro-3–indolyl-β–d–galactopyranoside and 1× BU salts (7 g·L−1 Na2HPO4 heptahydrate and 3 g·L−1 NaH2PO4) as a buffering system. For β–galactosidase quantitative assays, yeast strains were pre-inoculated in SD medium with 2% glucose for 4–6 h, and then inoculated in SD medium with 2% raffinose and grown for 15–17 h. At a cell concentration of 0.5–1 × 107 cell·mL−1, cultures were induced by addition of 2% galactose and further incubated for 24 h. β–galactosidase activity was determined by the 4–methylumbelliferyl-β–d–galactopyranoside (MUG) assay . Briefly, 50 μL of induced cell cultures were added to 50 μL Z buffer (60 mm Na2HPO4, 40 mm NaH2PO4, 10 mm KCl, 1 mm MgSO4, 50 mm β–mercaptoethanol, pH 7) in special 96-well microtiter plates (Optilux black 96-well plates, with a clear flat bottom, BD Biosciences, Sparks, MD, USA). After addition of 25 μL of 1 mg·mL−1 MUG indimethylsulfoxide, the cell suspension was incubated at room temperature for 15 min with gentle shaking. To stop the reaction, 30 μL of 1 m Na2CO3 were added. Finally, the fluorescence (excitation 360 nm/emission 460 nm) was determined in a microplate fluorimeter (TECAN, Männedorf, Switzerland) together with the attenuance at 595 nm (D595 nm) to evaluate cell density. Calculations of arbitrary MUG units were performed using the formula: MUGunits = (Fex/em − Fex/em(blank))/(timemin × A595 nm).
Proteins and protein analyses
Antibodies against σ54 and purified σ54 of P. putida at a concentration of 10 μm in 52.5% glycerol, 5 mm Tris/HCl pH 8, 0.05 mm EDTA, 0.5 mm dithiothreitol and 150 mm NaCl were kindly provided by Victor de Lorenzo and Carlos Alvarez, Systems Biology Program, Centro Nacional de Biotecnología, Madrid, Spain. The detection of σ54-based hybrids expressed from pEG202 and pJG4–5 was performed by western blotting (see below) using antibodies against LexA and HA, respectively. For western blotting, protein were separated by SDS/PAGE [10% acrylamide/bisacrylamide 29 : 1, 0.1% SDS, in 1× Tris/glycine buffer (Bio–Rad, Hercules, CA, USA)]and transferred via a semi-dry apparatus (Fastblot B33; Biometra, Goettingen, Germany) onto nitrocellulose membranes (Protran; Whatman, GE Healthcare, Uppsala, Sweden) with fixed 1.66 mA·cm−2 for 1 h. The membranes were blocked with NaCl/Pi with 0.05% Tween–20 and 5% skimmed milk, and incubated with antibodies in NaCl/Pi with 0.05% Tween–20 and 0.5% skimmed milk. Band detection was performed by using secondary antibodies conjugated to horseradish peroxidase and standard enhanced chemilumescent substrates of horseradish peroxidase enzyme activity. For native protein gel electrophoresis, purified σ54 was run in an NativePAGE™ Novex® Bis-Tris gel system, scanned for detection of Coomassie Brilliant Blue staining, and then transferred onto nitrocellulose membranes for the western blotting assay with antibodies against σ54 as described above. Analytical gel-filtration experiments were performed in an ÄKTA Purifier–10 system (GE Healthcare, Uppsala, Sweden) with a Superdex 200 10/300 GL column (GE Healthcare, Uppsala Sweden). A 200 μL aliquot of purified σ54 was eluted using 5 mm Tris/HCl, pH 8.0, 10% glycerol, 0.05 mm EDTA, 150 mm NaCl and 1 mm dithiothreitol. A calibration curve for molecular mass determination was prepared by measuring the elution volumes of the standards albumin, aldolase and catalase (GE Healthcare, Uppsala, Sweden) (67, 158 and 232 kDa, respectively). The protein concentration was assessed on the basis of a theoretical extinction coefficient ε = 52 000 m−1·cm−1 and a σ54 molecular mass of 56.5 kDa. The amount of protein in each elution peak was determined by integrating the Gaussian curve used to fit each peak.
We are grateful to Victor de Lorenzo and Carlos Alvarez, Systems Biology Program, Centro Nacional de Biotecnología, Madrid, Spain, for the kind donation of antibodies against σ54 and purified σ54 of P. putida.