The PhaD regulator controls the simultaneous expression of the pha genes involved in polyhydroxyalkanoate metabolism and turnover in Pseudomonas putida KT2442

Authors

  • Laura Isabel De Eugenio,

    1. Environmental Biology Department, Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu, 9, 28040 Madrid, Spain.
    Search for more papers by this author
    • Authors contribute equally to this work.

  • Beatriz Galán,

    1. Environmental Biology Department, Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu, 9, 28040 Madrid, Spain.
    Search for more papers by this author
    • Authors contribute equally to this work.

  • Isabel F. Escapa,

    1. Environmental Biology Department, Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu, 9, 28040 Madrid, Spain.
    Search for more papers by this author
  • Beatriz Maestro,

    1. Instituto de Biología Molecular y Celular, Universidad Miguel Hernández, Av. Universidad, s/n. 03202 Elche, Spain.
    Search for more papers by this author
    • Present address: Instituto Universitario de Electroquímica. Universidad de Alicante. 03080- Alicante, Spain.

  • Jesús M. Sanz,

    1. Instituto de Biología Molecular y Celular, Universidad Miguel Hernández, Av. Universidad, s/n. 03202 Elche, Spain.
    Search for more papers by this author
  • José Luis García,

    1. Environmental Biology Department, Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu, 9, 28040 Madrid, Spain.
    Search for more papers by this author
  • María A. Prieto

    Corresponding author
    1. Environmental Biology Department, Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu, 9, 28040 Madrid, Spain.
    Search for more papers by this author

E-mail auxi@cib.csic.es; Tel. (+34) 918 373 112; Fax (+34) 915 360 432.

Summary

The promoters of the pha gene cluster encoding the enzymes involved in the metabolism of polyhydroxyalkanoates (PHAs) in the model strain Pseudomonas putida KT2442 have been identified and compared. The pha locus is composed by five functional promoters upstream the phaC1, phaZ, phaC2, phaF and phaI genes (PC1, PZ, PC2, PF and PI respectively). PC1 and PI are the most active promoters of the pha cluster allowing the transcription of phaC1ZC2D and phaIF operons. All promoters with the sole exception of PF are carbon source-dependent. Their transcription profiles explain the simultaneous production of PHA depolymerase and synthases to maintain the metabolic balance and PHA turnover. Mutagenesis analyses demonstrated that PhaD, a TetR-like transcriptional regulator, behaves as a carbon source-dependent activator of the pha cluster. The phaD gene is mainly transcribed as part of the phaC1ZC2D transcription unit and controls its own transcription and that of phaIF operon. The ability of PhaD to bind the PC1 and PI promoters was analysed by gel retardation and DNase I footprinting assays, demonstrating that PhaD interacts with a region of 25 bp at PC1 promoter (named OPRc1) and a 29 bp region at PI promoter (named OPRi). These operators contain a single binding site formed by two inverted half sites of 6 bp separated by 8 bp which overlap the corresponding promoter boxes. The 3D model structure of PhaD activator predicts that the true effector might be a CoA-intermediate of fatty acid β-oxidation.

Introduction

The environmental pollution problems caused by the use of conventional plastics has generated a huge interest in the study of sustainable processes to generate these frequently used products from raw materials of agricultural or urban origins (Gavrilescu and Chisti, 2005). Some of the polyesters most seriously being considered as alternative plastics are the polyhydroxyalkanoates (PHAs). These are biodegradable polymers naturally produced by bacteria as carbon storage granules from renewable resources like glucose, fructose or fatty acids that form part of the vegetal oils (Madison and Huisman, 1999; Dias et al., 2006; Prieto et al., 2007). Medium-chain-length PHAs (mcl-PHA), containing 6–14 carbon atoms per monomer, are mainly produced by Pseudomonas species (Luengo et al., 2003; Prieto et al., 2007). Although bacterial fermentation and physicochemical characterization of the polymer have been studied extensively during the past few decades, knowledge on the molecular mechanisms regulating its synthesis and degradation is relatively limited (Prieto et al., 1999; Hoffmann and Rehm, 2004; 2005; Sierro, 2005; Sandoval et al., 2007). This is in part due to the complexity of PHA metabolism, which implies an extremely intricate regulatory system. Thus, PHA synthesis in Pseudomonads comprises: (i) central pathways, such as β-oxidation pathway and fatty acid de novo synthesis to convert fatty acid or carbohydrate intermediates, respectively, into different (R)-3-hydroxyalkanoyl-CoAs; and (ii) a specific or peripheral pathway encoded by the pha cluster including the genes encoding a depolymerase (PhaZ) (de Eugenio et al., 2007; 2008; 2010) and two synthases (PhaC1 and PhaC2) (Huisman et al., 1991), which coordinately with (iii) a granule-associated acyl-CoA-synthetase (Acs1) (Ruth et al., 2008; Ren et al., 2009), direct the carbon flux of these central metabolites towards PHA accumulation or hydrolysis as an active response to the carbon andenergy cellular demand (de Eugenio et al., 2010; Ren et al., 2009). It has been recently demonstrated in Pseudomonas putida KT2442 that PHA metabolism is an ongoing cycle where synthesis and degradation of the polyesters are simultaneously active facilitating the turnover of the polymer (de Eugenio et al., 2010; Ren et al., 2009). Besides the PHA related enzymes, formation and maintenance of PHA granules in the cytoplasm need the involvement of the major granule-associated proteins named phasins that play structural and regulatory roles (Prieto et al., 1999).

The pha cluster is very well conserved among mcl-PHA producer strains (de Eugenio et al., 2007; Prieto et al., 2007) (Fig. 1). In addition to the two synthases and depolymerase coding genes, the cluster is composed by the phaD gene encoding a TetR-like transcriptional regulator (PhaD protein) (Klinke et al., 2000; Ramos et al., 2005) and the phasin-encoding phaF and phaI genes, which are transcribed divergently to the other pha genes (Prieto et al., 1999; Moldes et al., 2004; Sandoval et al., 2007). PhaD plays an important role in mcl-PHA biosynthesis in P. putida GPo1 and its mutation affects polymer accumulation in several ways (Klinke et al., 2000). In this sense, the mcl-PHA production was reduced in a phaD- mutant and the number of PHA granules increased while size concomitantly decreased. Those effects have been ascribed to an altered expression of phasins (Klinke et al., 2000; Sierro, 2005). Therefore, it has been suggested that PhaD is a transcriptional regulator which drives the expression of the PhaI and PhaF proteins (Sandoval et al., 2007).

Figure 1.

Genetic organization of pha cluster, promoter sequences and intergenic regions in P. putida KT2442. The open arrows indicate the directions of gene transcription. PC1, PZ, PC2, PF and PI are active promoter regions. The nucleotide sequences of the phaD-phaF intergenic region, 266 bp from PC1 promoter and 187 bp from PI promoter are shown. The −35 and −10 boxes (blue) of the PC1 and PI promoters and their transcription start sites (+1) (pink) are indicated. The DNA sequences corresponding to PhaD operators are underlined, with the inverted regions indicated in solid boxes. The REP sequences found between phaD and phaF genes in P. putida KT2442 are indicated in dashed boxes. Stop codons are marked in red. Translational start codons are indicated in bold letter.

In this work, we have compared the activity of the promoter regions of the pha genes in the model strain P. putida KT2442. The role of PhaD as an activator of the pha cluster was demonstrated by mutagenesis analyses. Furthermore, the interaction of PhaD with their cognate promoters was studied by using gel retardation and DNase I footprinting analyses. Finally, the 3D structure of PhaD was modelled allowing us to predict that this protein might be activated by metabolic intermediates of the β-oxidation pathway of PHA precursors.

Results

Identification of promoter regions in the pha gene cluster of P. putida KT2442: carbon source dependency

The recent finding of a PHA turnover cycle due to the simultaneous activation of PHA synthase and depolymerase during growth of P. putida KT2442 (de Eugenio et al., 2010) suggested that the expression of both genes might be coordinately regulated. To demonstrate this hypothesis we firstly identified the promoter regions driving the expression of the pha genes in P. putida KT2442. We constructed six lacZ promoter fusions with the upstream region of each pha gene (named PC1, PZ, PC2, PD, PF and PI promoter regions, Fig. 1), which were transferred into the chromosome of the wild-type strain by using the mini-transposon technique (a detailed description of DNA fragment sizes, primers used and plasmid constructions is shown in Table 1). The six isolated P. putida strains KTpC1, KTpZ, KTpC2, KTpD, KTpF and KTpI, carrying each lacZ fusion (Table 1), were cultured under PHA production conditions. The β-galactosidase activities were monitored in these strains at middle exponential phase (8 h) and at stationary phase (24 h) (Table 2). The main conclusion that can be drawn from these results was that under PHA production conditions functioned as promoter regions showing different expression levels. Conversely, no reporter activity was detected by this method when the strain P. putida KTpD (PD::lacZ) was analysed. The highest level of reporter expression was detected in the strain P. putida KTpI carrying the PI::lacZ fusion, which was more than 50-fold higher to that observed in KTpF and KTpC1 strains equipped with the PF and PC1 reporter fusions respectively. Expression levels 5–10-fold lower than those found in KTpF and KTpC1 strains, were detected in cells carrying the PZ::lacZ and PC2::lacZ fusions. Our results also indicate that the expression level driven by each promoter, with the sole exception of PF, was dependent of the carbon source supplied into the media (Table 2). The induction rate observed in the expression levels driven by the promoters PC1, PZ, PC2 and PI when cells were cultured in octanoate versus glucose as preferred and poor PHA precursors, respectively, ranged from 1.5- to 2.5-fold higher (Table 2).

Table 1.  Construction of lacZ reporter P. putida KT2442 strains. Region amplified, fragment size, primers used, plasmid construction and resulting reporter strain.
Promoter regionFragment size (bp)PrimersPrimers nucleotide sequencepUJ9 derivativepUTminiTn5-Km derivativeP. putida strain
PC1266pC15′TTTGAATTCGGCCTGCGGGGTTTAGAGpUJC1pUTC1KTpC1 (PC1::lacZ)
pC13′CGGGGATCCATCTACGACGCTCCGTTGT
PC2242pC25′TTTGAATTCCCCGTTGATCCCGpUJC2pUTC2KTpC2 (PC2::lacZ)
pC23′CGCGGATCCATGGCAACACTCCCTCGTC
PZ239pZ5′TTTGAATTCGACCCGGTGGCCTGGCpUJZpUTZKTpZ (PZ::lacZ)
pZ3′CGCGGATCCATGCACGTGACTCTTG
PD737pD5′CCGGAATTCGCCACGTATCTGGTCAGCTTGCpUJDpUTDKTpD (PD::lacZ)
pD3′CGCGGATCCATCCAGTCAGCAGCTCATCGG
PF219pF5′CCGGAATTCCCAGCTTGACGAAGTCGGTGApUJFpUTFKTpF (PF::lacZ)
pF3′CGCGGATCCATCCTGCTCTCCTTATGGTTTGTG
PI441pI5′CCGGAATTCGCCAGAAAATGCCTGAGAAGCTCpUJIpUTIKTpI (PI::lacZ)
pI3′CGCGGATCCATGCTGTGTACCTCATGCTC
Table 2.  β-Galactosidase activities of lacZ reporter P. putida KT2442 strains, PHA content and biomass concentrations.
Carbon sourceP. putida strainTime (h)PHA content (%CDW)Biomass (g l−1)U βgal
  1. ND, not detected.

Sodium octanoateKTpI833.31.127 974 ± 362
KTpI2453.91.5424 356 ± 684
KTpF832.71.23181.9 ± 5.3
KTpF2454.91.55380.8 ± 2.3
KTpD830.00.99ND
KTpD2460.01.43ND
KTpC1831.41.09204.7 ± 2.9
KTpC12463.81.56434.4 ± 4.6
KTpC2836.51.05ND
KTpC22461.51.5657.1 ± 1.8
KTpZ829.90.94ND
KTpZ2455.21.3474.5 ± 2.2
GlucoseKTpI8< 50.62581.9 ± 18.6
KTpI2429.10.9516 484 ± 438
KTpF8< 50.61380.3 ± 6.1
KTpF2425.50.93499.4 ± 12.2
KTpD8< 50.67ND
KTpD2427.80.90ND
KTpC18< 50.6361.23 ± 3.5
KTpC12420.50.95172.4 ± 1.0
KTpC28< 50.68ND
KTpC22428.00.9547.7 ± 1.0
KTpZ8< 50.67ND
KTpZ2424.20.8837.20 ± 2.6

Comparative quantification of pha gene transcription rate by real-time RT-PCR

Our results demonstrated that, with the exception of the putative phaD regulatory gene, the rest of the genes are preceded by a functional promoter region, and consequently, they might be transcribed independently responding to different physiological and PHA metabolic conditions. However, the absence of putative transcription terminators downstream the phaC1, phaZ, phaC2 and phaD genes (Fig. 1) or between the phasin-coding genes phaI and phaF, indicated that the co-existence of a variety of polycistronic transcription units (operons) cannot be excluded. Consequently and to ensure the detection of all putative pha transcription units, the expression profile of the pha genes was comparatively monitored by real-time RT-PCR throughout the growth curve in the wild-type strain cultured in octanoate or glucose as preferred and poor PHA precursors respectively (see Experimental procedures for details). In agreement with the experiments performed with the lacZ reporter strains (Table 2), the transcription rate of all pha genes in the wild-type strain was optimal when the cells were cultured in the presence of octanoate, reaching in all cases a maximum level at exponential phase (7 h) (Table 3 and Fig. S1, see Supporting information). The transcription levels of the phaF and phaI genes were more that sevenfold higher than those of phaC1 gene. These results are in agreement with the fact that phasins are the major proteins associated to the PHA granule (Moldes et al., 2004).

Table 3.  Quantification of the expression rates of the pha genes by real-time RT-PCR assays.
P. putida strainGeneTranscription level (cDNA ng × 103)a
OctanoateGlucose
  • a. 

    Samples were taken 7 h after inoculation.

  • b. 

    Not detected.

KT2442phaC12 654 ± 742162 ± 57
phaZ996 ± 31430.8 ± 13.6
phaC2639 ± 21969.9 ± 17.8
phaD167 ± 4733.0 ± 16.0
phaF21 089 ± 1 5371 288 ± 308
phaI20 578 ± 4 882879 ± 94
KT42C1phaC1NDbNDb
phaZ5.4 ± 1.86.2 ± 5.4
phaC214.7 ± 5.619.3 ± 5.2
phaD9.2 ± 4.28 ± 6.7
phaF869 ± 601505 ± 34
phaI94.3 ± 12.143.4 ± 12.5
KT42DphaC1527 ± 17886.3 ± 33.6
phaZ115 ± 6438.7 ± 34.4
phaC245 ± 24.442.5 ± 15.1
phaD729 ± 127310 ± 106
phaF444 ± 35312 ± 91
phaINDbNDb

PhaD activator role on the PC1 promoter

Promoter activity quantification data derived from the lacZ reporter strains showed that PC1 is 10-fold more active than PZ and PC2 (Table 2). However, the results obtained by the relative quantification of the phaC1, phaZ and phaC2 mRNAs (Table 3 and Fig. S1) showed lower transcription rate differences suggesting that the expression of phaC1, phaZ and phaC2 genes could be partially driven by the PC1 promoter directing the transcription of a polycistronic unit. Furthermore, the transcription rates of phaC1, phaZ, phaC2 and phaD were inversely proportional to the distance from the PC1 promoter region suggesting as well that they were co-transcribed as an operon. Accordingly, the absence of promoter activity in the upstream region of the phaD gene suggested that the expression of this gene is controlled by an upstream promoter. To check this possibility, the expression of the pha genes was analysed in the P. putida KT42C1 (PhaC1-) and KT42D (PhaD-) mutant strains (Table 3 and Fig. S1). When phaC1 was disrupted, the transcription levels of the other pha genes were severely affected (Table 3 and Fig. S1). Similar polar effects on the expression rate of pha genes were observed when phaD was interrupted (Table 3 and Fig. S1). The reduced transcription rate observed for the phaC1, phaI and phaF genes in the PhaD mutant can only be explained if PhaD protein works as an activator of the cognate promoters. With the sole exception of phaI gene, only basal expression levels were detected in the genes phaC1, phaZ, phaC2 and phaF, in the KT42D (PhaD-) mutant strain, suggesting that they are likely driven by the internal promoters PZ, PC2 and PF and a basal activity of the PC1 promoter.

These results clearly demonstrated that the integrity of the DNA region comprising phaC1, phaZ, phaC2 and phaD is crucial for the efficient transcription of the whole pha gene cluster when cells were grown in octanoate as carbon and energy sources, suggesting the existence of the phaC1ZC2D operon where PhaD acts as an activator of the pha cluster. The expression level of phaC1ZC2D operon detected when the wild-type cells were cultured in glucose did not differ significantly from those detected in the KT42D (PhaD-) mutant strain growing in octanoate, indicating that PhaD controls the carbon source dependence of the transcription profile of this operon and that the system might be induced either by octanoate or some octanoate-derived metabolite.

The PHA content data did proportionally reflect the lack of induction in the presence of octanoate. Then, PHA accumulation from octanoate was significantly affected but not abolished in the KT42D (PhaD-) mutant strain after 24 h of culture [18 ± 2 PHA as % of cell dry weight (CDW)] when compared with the wild-type strain (70 ± 2 PHA as % of CDW). Thus, the existence of alternative weak promoters (non-PhaD-induced PC1, PZ and PC2) ensures the accumulation of the polymer under non-favourable conditions. In this sense, the KT42C1 (PhaC1-) mutant is able to produce 0.5 ± 0.1 PHA as % of CDW (de Eugenio et al., 2010), very likely due to a basal activity of the PhaC2 polymerase.

PhaD activates the PI promoter

The PhaD protein of P. putida GPo1 and P. putida U has been proposed to act as a transcriptional regulator belonging to the TetR family of regulators, which activates the transcription of the phaI gene (Klinke et al., 2000; Ramos et al., 2005; Sandoval et al., 2007). The experiments shown in Table 3 and Fig. S1 demonstrated that phaI mRNA is completely absent in the KT42D (PhaD-) mutant strain and only a very weak expression was detected for the phaF gene (Fig. S1). In fact, the PhaI and PhaF phasins were not detected in the PHA granules when the KT42D (PhaD-) cells were cultured under PHA production conditions in the presence of octanoate (Fig. 2). The pIZD plasmid expressing PhaD was able to restore the capacity of KT42D (PhaD-) mutant to accumulate PHA (65 PHA as % CDW) at a similar ratio than the wild-type strain. As expected, the PHA granules of P. putida KT42D (pIZD) contained PhaI and PhaF as the major granule-associated proteins.

Figure 2.

Comparative analyses of granule-associated proteins from P. putida KT2442 and KT42D strains. Cells were grown under PHA production conditions for 24 h, and granules were isolated using a glycerol gradient centrifugation as described (de Eugenio et al., 2007). Suspensions of granules were normalized at 150 optical units (OD650). Samples of purified granules were mixed 1:1 (v/v) with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer, and the bound proteins were separated by SDS-PAGE as described (Laemmli, 1970).

Taking together the RT-PCR results (Table 3 and Fig. S1), which showed constitutive levels of phaF mRNA in the absence (P. putida KT42D) or low levels (P. putida KT42C1) of PhaD protein, our findings confirm the crucial role of PhaD in the expression of the phasin-coding genes, and suggest the existence of a phaIF operon driven by the PI promoter as well as a PhaD-independent expression of phaF driven by the PF promoter. These results agree with those reported in P. putida GPo1, where a weak constitutive PF promoter was proposed for the permanent production of the PhaF protein (Prieto et al., 1999).

Characterization of the PhaD-operator regions at PC1 and PI promoters

To demonstrate the direct implication of PhaD in the transcription driven by PC1 and PI promoters and the PhaD-independent PF activity, we analysed the ability of PhaD to bind the corresponding intergenic regions by gel retardation assays [electrophoretic mobility shift assay (EMSA)] using cell-free extracts from Escherichia coli DH10B (pIZD) and PC1, PI and PF DNA fragments as probes respectively (Table 1, Fig. S2). While PhaD was able to retard the migration of both PC1, and PI probes in a protein concentration-dependent manner, the PF probe did not, suggesting that PhaD is able to bind to PC1 and PI promoters even in the absence of any inducer. Moreover, experiments with unrelated DNA from salmon sperm demonstrated that the PhaD binding to the PI promoter region was highly specific (Fig. S2). On the other hand, the lack of PhaD binding to the PF promoter region confirmed the existence of two transcription units in this region. Finally, to identify the inducer of PhaD protein, EMSAs were performed in the presence of PHA latex (2–20 µg, prepared as described in Schirmer and Jendrossek, 1994) and octanoate (0.1–1.0 mM) (see Experimental procedures). However, these compounds did not change the retardation pattern of the DNA probes (data not shown).

To precisely localize the PhaD operators and transcription start sites in the PC1 and PI promoters, primer extension and DNase I footprinting analyses were performed using total RNA isolated from P. putida KT2442 and the PC1 and PI probes respectively (Figs 1 and 3). The PI promoter transcription initiation site was mapped 54 nucleotides upstream of the ATG translation initiation codon of the phaI gene, showing two −10 (TACCTT) and −35 (GTGAGG) putative boxes, which differ in three nucleotides from the consensus RpoD-dependent promoters (Fig. 1). A major transcription initiation site was located in the PC1 promoter 224 nucleotides upstream of the ATG translation initiation codon of the phaC1 gene showing the −10 (TAAATT) and −35 (TGCGGG) putative boxes (Fig. 1).

Figure 3.

Characterization of PI and PC1 promoters.
A and B. DNase I footprinting analyses of PhaD interaction with the PI and PC1 promoter regions respectively. The reaction mixture was treated as described in Experimental procedures, using the 5′-end-labelled non-coding strand and the 5′-end-labelled coding strand of the PI and PC1 regions respectively. The A+G sequencing ladder is indicated with an asterisk.
A. Lanes 1 and 2: naked DNA; lanes 3 and 4 contain 0.6 and 6 µg of total protein of PhaD-free extracts obtained from E. coli cells bearing control plasmid pIZ1016 respectively, and lanes 5–10 contain 0.6, 1.2, 1.5, 2, 3 and 6 µg of total protein of PhaD+ extracts,
B. Lane 1, naked DNA; lanes 2–7 contain 0.6, 1.2, 1.5, 2, 3 and 6 µg of total protein of PhaD+ extracts and lanes 8 and 9 contain 0.6 and 6 µg of total protein of PhaD-free extracts obtained from cells bearing control plasmid pIZ1016 respectively.
C. Sequences of the PhaD-binding sites at PI and PC1 promoters. The palindromic sequences are indicated by inverted black arrows.

DNase I footprinting experiments revealed that PhaD protects a region of 25 bp at PC1 promoter (named OPRc1) and a 29 bp region at PI promoter (named OPRi) (Fig. 3). Both protected sequences contains a single binding site formed by two inverted half sites of 6 bp separated by 8 bp (Fig. 3C). Interestingly, OPRc1 and OPRi operators overlap the promoter boxes in their cognate promoter regions suggesting a similar regulatory mechanism of the PhaD protein.

Homology modelling of PhaD

PhaD controls the carbon source dependence of the transcription profile of the pha cluster and this regulatory system seems to be induced by a metabolite derived from fatty acid β-oxidation pathway. To get insight into the structural basis supporting this hypothesis, we modelled the 3D structure of the PhaD protein using the 3D structures of similar regulators belonging to the TetR family as templates (i.e. Schumacher et al., 2001; 2002; Rajan et al., 2006; Premkumar et al., 2007). All these regulators show α-helical dimeric structures containing an N-terminal helix–turn–helix (HTH) DNA-binding domain, plus a C-terminal domain involved in dimerization and effector binding. Among the available templates, the structure of the EthR repressor from Mycobacterium tuberculosis complexed with hexadecyloctanoate (Frénois et al., 2004) was especially attractive due to the similarity of this ligand with some of the putative PhaD effectors like hydroxyoctanoyl-CoA (HOCoA). The modelled PhaD structure predicts the existence of a dimer, being each monomer composed of two domains built up from nine α-helices (α1-9), where the α2 and α3 helices constitute a HTH motif in the N-terminal domain (Fig. 4). Helices α8 and α9 would be involved in dimerization through the configuration of a four-helix bundle. The most noteworthy feature of this model is the presence of a long, deep crevice surrounded by helices α5–α8 that spans the whole C-terminal domain (Fig. 4 and Fig. S3) and that is used in the case of EthR to bind its hexadecyloctanoate ligand. We observed that an elongated HOCoA molecule could also fit inside the crevice without sterical impediments and establish a number of stabilizing interactions with several amino acid side-chains (Fig. S3).

Figure 4.

Homology-based 3D model of dimeric PhaD. Three-dimensional model based on the structure of EthR from M. tuberculosis. The protein is depicted in cartoon representation, and the HOCoA molecule is shown in spacefill format. Numbered α-helices are also shown. Oval and arrow highlight the 3-hydroxyoctanoyl moiety in HOCoA.

Discussion

To date, the transcriptional regulation of pha genes of Pseudomonas species has been scarcely studied and only incomplete and disperse information is available for some of the best-known PHA producer strains (reviewed in Prieto et al., 2007). In this work, we have identified the activity of the promoter regions of the pha genes in the model strain P. putida KT2442 and compared the activity of each individual promoter with the total mRNA generated for each pha gene. These experiments have allowed us to definitively determine that the pha locus in P. putida KT2442 is composed by at least five functional promoters upstream the phaC1, phaZ, phaC2, phaF and phaI genes (PC1, PZ, PC2, PF and PI respectively), and that PC1 and PI are the most active promoters in the pha cluster, allowing the transcription of phaC1ZC2D and phaIF operons. This transcription profile correlates with the simultaneous production of PHA depolymerase and synthases to maintain the metabolic balance and PHA turnover (de Eugenio et al., 2010). All promoters with the sole exception of PF are carbon source-dependent. Octanoate, a PHA precursor carbon source, increased the transcription driven by these dependent promoters, while less preferred PHA precursor carbon sources, like glucose, caused lower levels of gene transcription. It is interesting to notice that the presence of weak promoters might allow the expression of specific pha genes under diverse physiological circumstances (see below). The most active promoter is PI driving the expression of the phaF and phaI genes, which agrees with the major content of phasins in the PHA granules (Prieto et al., 1999; Moldes et al., 2004).

The phaD gene encoding a TetR-like transcriptional activator is mainly transcribed as part of the phaC1ZC2D transcription unit, although it could be also expressed from PZ or PC2 promoters which can drive the alternative transcription units phaZC2D and phaC2D respectively. The results shown in Table 3 and Fig. S1 demonstrated that the carbon source-dependent induction of the pha genes is controlled by PhaD that is acting as an activator. However, the influence of the carbon source in the transcription profile of the phaC1 and phaZ mRNAs was still detectable in the absence of PhaD mutant strain (KT42D) suggesting that additional factors might control the expression of the pha cluster (Table 3 and Fig. S1). In this sense, the PhaF phasin and some global transcriptional factors like the GacS/GacA system have been demonstrated to be involved in the regulation of pha gene expression (Prieto et al., 1999; Castaneda et al., 2000; Kessler and Witholt, 2001; Hoffmann and Rehm, 2004; Sandoval et al., 2007).

Pseudomonas putida KT2442 is able to accumulate PHA from glucose although at minor extend than from fatty acids (Huijberts et al., 1992) (Table 2) due to the presence of the specific PhaG transacylase that converts the (R)-3-hydroxyacyl-acyl carrier protein [(R)-3-hydroxyacyl-ACP)] intermediates of the fatty acid biosynthesis route into the substrate of the PHA synthase, i.e. the (R)-3-hydroxyacyl-CoA (Rehm et al., 1998). The lack of pha transcriptional activation when the strain grows in glucose suggests that the actual PhaD inducer might be an intermediate of the fatty acid β-oxidation pathway, which is only barely produced when cells are cultured in this carbon source. Nevertheless, in these circumstances the basal activities of the pha promoters still allow P. putida KT2442 to accumulate PHA.

Although the pha gene organization and gene products are highly conserved in the P. putida mcl-PHA producer strains, nucleotide sequences of pha cluster intergenic regions differ considerably among these strains (Kessler and Witholt, 2001; Solaiman et al., 2008). In fact, very recent studies about the pha clusters from the phylogenetically related strains Pseudomonas corrugata and Pseudomonas mediterranea show vastly differences in the pha intergenic regions (Solaiman et al., 2008). Consequently, it would be not surprising that transcriptional factors might trigger different effects in different PHA producer strains. Unlike in other bacteria, the pha locus in P. putida U seems to be integrated by six different functional units (phaC1, phaZ, phaC2, phaD, phaI and phaF) (Sandoval et al., 2007). In P. putida GPo1, two promoters have been reported upstream of the phaC1 gene (PC1 region), showing transcriptional start sites at 198 and 112 bp upstream of the phaC1 ribosomal binding site, respectively, being the later RpoN dependent (Huisman, 1991). On the other hand, there are at least two promoters involved in the expression of the pha genes in Pseudomonas aeruginosa that are located upstream of the phaC1 gene and resemble the consensus sequences for RpoN and RpoD-dependent promoters (Timm and Steinbüchel, 1992). We have shown that the PC1 transcription start site in P. putida KT2442 lays at 224 bp upstream the ATG codon.

The mRNA analyses carried out in the GPo1 strain showed that all the transcripts were shorter than 3 kb, suggesting that phaC1, phaZ, phaC2 and phaD do not form part of the same transcription unit, or that an mRNA processing event is taking place (Prieto et al., 1999). The existence of phaC1 and phaC1Z transcription units was demonstrated in GPo1 as also reported for the homologous system of P. aeruginosa (Prieto et al., 1999; 2007). Thus, in P. putida GPo1 the transcription should partially stop at the end of the phaC1 gene and partially continue to the end of phaZ. This assumption is in agreement with the presence in this strain of a structure analogous to the enterobacteriaceae repetitive extragenic palindromic (REP) sequences located downstream of the phaZ gene, which might function as transcription terminator (Huisman et al., 1991; Aranda-Olmedo et al., 2002). Interestingly, there is not a REP downstream phaZ in P. putida KT2442, supporting the existence of the mRNA phaC1ZC2D transcribed from PC1 promoter in this strain. An additional role of REP sequences would be to allow the binding of DNA gyrase to relax DNA when excessive positive supercoiling is generated, especially between two convergent genes simultaneously transcribed (Liu and Wang, 1987; Wu et al., 1988; Yang and Ames, 1988). In this sense, inverted REP sequences were found also downstream of phaD in P. putida GPo1 and KT2442 (Fig. 1), although the nucleotide sequence in this region is not identical in both strains (data not shown). Thus, REP sequences found between phaD and phaF genes could favour simultaneous transcription of pha genes, enabling the relaxation of supercoiled DNA in the intergenic region.

The different expression profiles of the transcription units phaF and phaIF were observed firstly in P. putida GPo1 (Prieto et al., 1999) suggesting the possible presence of two promoters, one located upstream of the phaI gene and the other located upstream of phaF. A different expression of phaF and phaIF transcription units was also observed for P. putida KT2440 and P. aeruginosa (Hoffmann and Rehm, 2004; 2005; Rehm, 2006). In this study, we have demonstrated experimentally the existence and different expression profiles of these transcription units showing different activity in terms of transcription intensity and PhaD dependence.

The transcription regulators of the TetR family whose functions have been described so far for 85 out of more than 2300 proteins are usually repressors which are inactivated by small ligands such as antibiotics, drugs, catabolites, osmoprotectants and quorum-sensing autoinducers (Ramos et al., 2005). One exception is the DhaS regulator, an activator of the dha operon in Lacotococcus lactis (Christen et al., 2006). Similarly to PhaD (Figs 1 and 3), DhaS binds to an operator sequence partially overlapping the −35 promoter box (Christen et al., 2006), whereas TetR-like repressors bind to inverted repeats that overlap or are located downstream of the −10 promoter box (Grkovic et al., 1998). Few other TetR-type proteins have been suggested to activate gene expression; however, their mechanisms have not been described so far (Alatoom et al., 2007; Chatterjee et al., 2007; Audra et al., 2008; Hirano et al., 2008).

According to our results, binding to operator DNA sequences is a necessary, albeit not sufficient, requirement for PhaD to become an activator of pha gene expressions. In this sense, our data also demonstrate the need for PHA precursors (i.e. octanoate), or related intermediates of the fatty acid β-oxidation pathway for the complete transcription of the pha genes (Table 3 and Fig. S1). The 3D model of the PhaD structure that we have proposed appears to be compatible with the binding of such metabolic intermediates (Fig. 4). The most noteworthy feature of this model is the presence of a long deep crevice that may accommodate an extended HOCoA molecule without relevant steric hindrances (Fig. 4). Whereas the EthR effector-binding site is mainly hydrophobic to accommodate its hexadecyloctanoate ligand, many polar residues are found in the PhaD effector-pocket (Fig. S3), in accordance with the presence of a more polar ligand such as a CoA-intermediate of fatty acid β-oxidation.

Summarizing, our results have provided the first exhaustive experimental analysis of the complex transcription system that allows the simultaneous production of the PHA-related proteins to maintain the metabolic balance and polymer turnover in P. putida KT2442. In addition, we have demonstrated for the first time that PhaD regulator is directly involved in the transcriptional control of the pha cluster. Moreover, we have proved that PhaD is one of the few activators of the TetR-like family of regulators. Although the pha cluster is induced by octanoic acid, the 3D structural model of PhaD allowed us to propose that a CoA-derivative could be the true inducer. These findings allow us to conclude that PhaD represents a paradigmatic example within the whole scenario of bacterial regulators.

Experimental procedures

Bacterial strains, media and growth conditions

Pseudomonas putida KT2442 (Franklin et al., 1981) is a derivative strain of the parental strain KT2440 whose complete nucleotide sequence is accessible in the data bank (Nelson et al., 2002). Pseudomonas putida KT42C1 (de Eugenio et al., 2010) is a PhaC1 mutant strain of P. putida KT2442 constructed via disruption of phaC1 gene by insertion of a minitransposon containing kanamycin-resistance gene (de Lorenzo et al., 1990; Herrero et al., 1990). Unless otherwise stated, Escherichia coli and P. putida strains were grown in Luria–Bertani (LB) medium (Sambrook and Russell, 2001) at 37°C and 30°C respectively. The appropriate selection antibiotics, ampicillin (100 µg ml−1), gentamicin (10 µg ml−1), chloramphenicol (34 µg ml−1) or kanamycin (50 µg ml−1) were added when needed. For poly(hydroxyoctanoate-co-hydroxyhexanoate) [P(HO-co-HH)] production, P. putida strains were grown as previously described (de Eugenio et al., 2010).

DNA and RNA manipulations

DNA manipulations and other Molecular Biology techniques were essentially performed as described previously (Sambrook and Russell, 2001). Plasmid transference to the target Pseudomonas strains was done by the filter-mating technique (Herrero et al., 1990). DNA fragments were purified by standard procedures using Gene Clean (Bio 101, Vista, CA). For PCR amplification, we used two units of AmpliTaq DNA polymerase (PerkinElmer Life Sciences), 10 ngr of DNA template, 1 µg of each deoxynucleoside triphosphate and 2.5 mM MgCl2 in the buffer recommended by the manufacturer. Conditions for amplification were chosen according to the G + C content of the corresponding oligonucleotides. Nucleotide sequences were determined directly with the same oligonucleotides used for cloning. Standard protocols of the manufacturer for Taq DNA polymerase-initiated cycle sequencing reactions with fluorescently labelled dideoxynucleotide terminators (Applied Biosystems) were used in an ABI Prism 3730 DNA Sequencer.

Construction of PhaD mutant

Pseudomonas putida KT42D strain was constructed by disruption of phaD gene using pK18mob plasmid (Schafer et al., 1994). An internal EcoRI/BamHI fragment of 384 bp of the gene was cloned in the polylinker of pK18mob after PCR amplification with primers D5mut (5′-CCGGAATTCCAACGAACTCGGCATCAGC-3′) and D3mut (5′-CGGGATCCGATCTGCTCGACCAGTTGCCC-3′). The resulting construction, pK18mob-mutd, was introduced into P. putida KT2442 by triparental mating using the strains E. coli DH10B (Invitrogen) and E. coli HB101 (pRK600) (de Lorenzo et al., 1990) as donor and helper strains respectively. Conjugants were selected in M63 0.2% citrate plus kanamycin. Disruption of phaD gene in the strain KT42D was confirmed by PCR sequencing analyses.

Plasmid constructions

The phaD coding sequence was amplified by PCR using oligonucleotides phaD5 (5′-CCCAAGGCTTATGAAAACCCGCGATCGTATCC-3′) (the start codon is indicated in bold and an engineered HindIII site is underlined) and phaD3 (5′-CTAGTCTAGACTACCCCTCCAGGTACTTCACTGCC-3′) (XbaI site is underlined) and KT2442 genome as DNA template. The amplified DNA fragment was digested with HindIII and XbaI and then inserted into pIZ1016 vector (Moreno-Ruiz et al., 2003). The resulting recombinant plasmid was transformed into E. coli DH10B, P. putida KT2442 and P. putida KT42D.

Construction of lacZ reporter P. putida KT2442 strains

Upstream regions of each pha genes of 219–737 bp flanked by engineered EcoRI and BamHI sites were fused to the reporter gene lacZ of E. coli using pUJ9 plasmid (de Lorenzo et al., 1990) (Table 1). Reporter fusions were inserted into the chromosome of the target strains by the pUT-Km miniTn5 delivery system by the filter-mating technique (Herrero et al., 1990), which allows the generation of reporter strains carrying translational fusions with the lacZ gene stably inserted into their chromosome. The selection of each reporter strain was made among three different candidates with similar expression levels to avoid promoter-unrelated lacZ expression. E. coli CC118λpir and E. coli HB101 (pRK600) were used as donor and helper strains, respectively, during the parental mating as previously described (de Lorenzo et al., 1990). Table 1 shows the details (region amplified, fragment size, primers used, plasmid construction and resulting reporter strain) of each construction. Correct insertion was tested by PCR technique. Isolated transconjugant strains grown at an OD600 of 0.2 were cultured in 0.1 N M63 medium plus 15 mM octanoic acid or 20 mM glucose. Three strains of each fusion selected did not show differences in terms of PHA content and growth profile on LB and 0.1 N M63 medium plus 15 mM octanoic acid or 20 mM glucose when compared with the wild-type strain.

Real-time PCR assay

Total RNA was extracted from P. putida KT2442 grown in 0.1 N M63 medium plus 15 mM octanoic acid or 20 mM glucose. For RNA purifications, 500 ml flasks containing 200 ml of culture medium were inoculated with overnight grown cells to reach an optical density (OD600) of 0.3, and incubated at 30°C with shaking. Aliquots of 5 ml of cells were harvested throughout the growth curve (0, 3.5, 7 and 24 h) and stored at −20°C. Pellets were thawed and cells lysed in TE buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA) containing 5 mg of lysozyme per ml by a series of freeze/thaw cycles. RNA was extracted using the RNeasy mini kit (Qiagen), including a DNase I treatment according to the manufacturer's instructions, precipitated with ethanol, washed and resuspended in 40 µl of RNase-free water. The concentration and purity of the RNA samples were measured by using a ND1000 spectrophotometer (Nanodrop Technologies). Synthesis of total cDNA was carried out with 20 µl reverse transcription reactions containing 1 µg of RNA, 0.5 mM dNTPs, 200 U of SuperScript II Reverse Transcriptase (Invitrogen) and 5 µM of random hexamers as primers, in the buffer recommended by the manufacturer. Samples were initially heated at 65°C for 5 min and then incubated at 42°C for 1 h, terminated by incubation at 70°C for 15 min. The cDNA obtained was purified using Geneclean Turbo kit (MP Biomedicals) and the concentration was measured using a ND100 Spectrophotometer (Nanodrop Technologies). For the analysis of the transcripts levels target cDNAs (0.5 and 5 ng) and reference samples were amplified three times in separate PCR with 0.2 µM each of target primers by using the iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad). Target primers were: primers C1RT5′ (5′-CTGGGCACCAGCGAAGGCG-3′) and C1RT3′ (5′-GTAATCGACAGCACCGCGTC-3′) for phaC1; C2RT5′ (5′-GCGGCGTGGCTCACCTG-3′) and C2RT3′ (5′-GAAGCTGTTGGTCGCGCTG-3′) for phaC2; ZRT5′ (5′-GAAGTCATCGCCTTTGATGTGCC-3′) and ZRT3′ (5′-ATCATCCACAGCACCTTGGGCTTG-3′) for phaZ; D-RTf (5′-CATCAGCCCAGGCAACCTGTAC-3′) and D-RTr (5′-GCGCTCGACGATCAAGTGCAG-3′) for phaD; F-RTf (5′-GTCATGTTTAGACGGAATACCCAG-3′) and F-RTr (5′-GCGGCCAACCACCAGCTTG-3′) for phaF; I-RTf (5′-GCACCGGTCAGCTTCTCGATC-3′) and I-RTr (5′-GGAGCGAACTTGAAGAAGCC-3′) for phaI. Real-time PCR was performed using SYBR Green technology in an ABI Prism 7000 Sequence Detection System (Applied Biosystems). Samples were initially denatured by heating at 95°C for 4 min, followed by 30 cycles of amplification (95°C, 1 min; test annealing temperature, 65°C, 1 min; elongation and signal acquisition, 72°C, 30 s). For quantification of the fluorescence values, a calibration curve was made using dilution series from 5.10−7 to 5 ng of P. putida KT2442 genomic DNA sample.

Gel retardation assays

PC1, PF and PI DNA fragments (Table 1) used as probes were labelled at their 5′-end with phage T4 polynucleotide kinase and [α-32P]-ATP (3000 Ci mmol−1) (Amersham Biosciences). The DNA probes were purified by the QIAquick nucleotide removal kit (Qiagen). Reactions with increasing concentrations of crude extracts (0.08–3.5 µg of total protein) were performed as described elsewhere (Manso et al., 2009). The gels were dried onto Whatman 3MM paper and exposed to Hyperfilm MP (Amersham Biosciences).

Mapping transcription start sites

Pseudomonas putida KTpC1 and P. putida KTpI cells were grown in 0.1 N M63 medium with 15 mM octanoate during 7 h until the cultures reached an OD600 of about 2. Total RNA was isolated using the RNA/DNA Midi kit (Qiagen) according to the instructions of the supplier. Primer extension reactions were carried out with the avian myeloblastosis virus reverse transcriptase as described previously (Prieto and García, 1997), using LAC57 primer (5′-CGATTAAGTTGGGTAACGCCAGGG-3′). To determine the length of the primer extension products, sequencing reactions of plasmids pUJC1 and pUJI (Table 1) were carried out with the same primer by using the T7 sequencing kit and [α-32P]-dATP (Amersham Biosciences) as indicated by the supplier. Products were analysed on 6% polyacrylamide-urea gels. The gels were dried onto Whatman 3MM paper and exposed to Hyperfilm MP (Amersham Biosciences).

DNase I protection experiments

For DNase I footprinting experiments, the DNA fragments PC1, and PI were amplified by PCR with the target primers (Table 1) using 10 ng of plasmids pUJC1 and pUJI respectively (Table 1) as templates. Fragments were labelled using a combination of one unlabelled primer and a second primer end-labelled with phage T4 polynucleotide kinase [α-32P]-ATP (111 TBq mmol−1). Then the PCR fragment was purified using the High Pure PCR Product Purification kit from Boehringer Mannheim according to the manufacturer's instructions. Reactions with increasing concentrations of crude extracts (0.6–6 µg of total protein) were performed as described before (Galán et al., 2001). Protected bands were identified by comparison with the migration of the same fragment treated for the A+G sequencing reaction (Maxam and Gilbert, 1977).

Analytical procedures

Cell densities, expressed in grams of CDW per litre, were determined gravimetrically as described previously (de Eugenio et al., 2010).

For PHA content determination, about 5–10 mg of lyophilized cells was analysed following the procedure reported before (Lageveen et al., 1988). The P(HO-co-HH) was isolated from P. putida KT2442 cells as described in Lageveen and colleagues (1988).

Assay of β-galactosidase activity with whole P. putida cells was performed as described previously (Miller, 1972). It is worth to notice that the P. putida strains produced PHA at different rates under the assayed growth conditions using octanoate or glucose as carbon sources. Then, differences in the PHA cell content affect considerably to the turbidity measurement (de Eugenio et al., 2010) and consequently, the classical Miller method is not suitable for the determination of the β-galactosidase in a comparative and quantitative assay. β-Galactosidase activity assays in P. putida strains were performed as described previously, by considering one unit of β-galactosidase activity in P. putida strains is defined as 1 µmol of o-nitrophenyl-β-d-galactopyranoside hydrolysed per mg of residual biomass per min. Residual biomass is defined as PHA-free CDW (Prieto et al., 1999).

Homology modelling of PhaD

The 3D structure of PhaD was modelled using Swiss PDB Viewer 3.7 (Guex and Peitsch, 1997). The template used was the EthR regulator from Mycobacterium tuberculosis complexed with hexadecyloctanoate (PDB code 1U9N) (Frénois et al., 2004). Primary structure alignments were performed by the CLUSTALW utilities contained in the Network Protein Sequence Analysis server at http://npsa-pbil.ibcp.fr (Combet et al., 2000). In order to model the protein–ligand interaction, a PDB file of 3-hydroxyoctanoyl-coenzyme A was generated using the ChemOffice 8.0 utilities (CambridgeSoft) and manually docked onto the PhaD model in order to get the maximum overlap with the template ligand hexadecyloctanoate, using Swiss PDB viewer. Raw structures obtained from fitting were subsequently refined by steepest descent energy minimization and checked for packing errors and optimal bonding. Figures were rendered using the software package PyMol (Delano Scientific LLC).

Acknowledgements

We thank Dr E. Díaz for helpful discussions. The technical works of A. Valencia and V. Morales are greatly appreciated. This work was supported by grants from the Comunidad Autónoma de Madrid (P-AMB-259-0505), the Ministry of Science and Innovation (BIO2007-67304-C02, CSD2007-00005) and by European Union Grants (GEN 2006-27750-C5-3-E and NMP2-CT-2007-026515).

Ancillary