Gallic acid (3,4,5-trihydroxybenzoic acid, GA) is widely distributed in nature, being a major phenolic pollutant and a commonly used antioxidant and building-block for drug development. We have characterized the first complete cluster (gal genes) responsible for growth in GA in a derivative of the model bacterium Pseudomonas putida KT2440. GalT mediates specific GA uptake and chemotaxis, and highlights the critical role of GA transport in bacterial adaptation to GA consumption. The proposed GA degradation via the central intermediate 4-oxalomesaconic acid (OMA) was revisited and all enzymes involved have been identified. Thus, GalD is the prototype of a new subfamily of isomerases that catalyses a biochemical step that remained unknown, i.e. the tautomerization of the OMAketo generated by the GalA dioxygenase to OMAenol. GalB is the founding member of a new family of zinc-containing hydratases that converts OMAenol into 4-carboxy-4-hydroxy-2-oxoadipic acid (CHA). galC encodes the aldolase catalysing CHA cleavage to pyruvic and oxaloacetic acids. The presence of homologous gal clusters outside the Pseudomonas genus sheds light on the evolution and ecology of the gal genes in GA degraders. The gal genes were used for expanding the metabolic abilities of heterologous hosts towards GA degradation, and for engineering a GA cellular biosensor.
Gallic acid (3,4,5-trihydroxybenzoic acid, GA) is an common phenolic compound of plant origin that can be found in the free state or in the form of esters (e.g. tannins) or ethers (e.g. syringic acid and other lignin constituents), being a major pollutant present in the wastewaters generated in the boiling cork process and in food manufacturing industries. GA and its derivatives are used in industry as antioxidants, and they are of interest also in drug development and as antimicrobial agents (Ow and Stupans, 2003). Although GA is widely distributed in nature, it is easily oxidized at neutral or alkaline pH becoming a difficult carbon source for bacterial growth. In fact, only bacteria of the genus Pseudomonas have been reported to utilize free GA as sole carbon and energy source under aerobic conditions (Beveridge and Hugo, 1964; Chowdhury et al., 2004). Moreover, the oxidation of GA also hinders the study of its degradation under laboratory conditions. The aerobic metabolism of GA usually starts with a direct ring-cleavage reaction and formation of the central intermediate 4-oxalomesaconic acid (OMA), which then undergoes hydration to 4-carboxy-4-hydroxy-2-oxoadipic acid (CHA) and aldolic cleavage to oxaloacetic and pyruvic acids (Fig. 1A). This pathway has been studied in some Pseudomonas and Sphingomonas strains that, surprisingly, cannot use GA as sole carbon source but are able to metabolize the GA generated when they grow at the expense of syringic acid (Tack et al., 1972a,b; Sparnins and Dagley, 1975; Kersten et al., 1982; Kasai et al., 2005). In Sphingomonas paucimobilis SYK-6, the lig genes involved in the meta-cleavage pathway for protocatechuic acid degradation via OMA are also involved in the lower pathway for syringic acid metabolism (Kasai et al., 2005). On the contrary, in most Pseudomonas strains protocatechuic acid is mineralized via ortho-cleavage throughout the β-ketoadipic acid pathway (Harwood and Parales, 1996) and therefore they have evolved a devoted pathway for GA degradation via OMA most of whose genetic determinants are still unknown.
Pseudomonas putida KT2440 is a paradigmatic bacterium renowned for its ability to degrade a wide range of aromatic compounds (Nelson et al., 2002; Jiménez et al., 2010), and it can be also adapted to degrade GA as sole carbon and energy source (Nogales et al., 2005). In this work, we have identified and characterized in P. putida KT2440 the first complete gene cluster (gal genes) responsible for bacterial growth using GA as sole carbon and energy source. The GA uptake and the biochemistry of its degradation via OMA have been revisited, new enzymes were unravelled, and some evolutionary considerations regarding the gal cluster are also discussed.
Results and discussion
Transport drives bacterial adaptation to GA consumption
The in silico analysis of the P. putida KT2440 genome revealed a 8 kb gal cluster (Fig. 1C), which included the galA gene encoding an extradiol gallate dioxygenase essential for GA metabolism (Nogales et al., 2005), a putative CHA aldolase encoding gene (galC), two additional genes of unknown function (galB and galD), a putative transport system encoded by galT* and galP, and a regulatory gene (galR) encoding a peculiar LysR-type transcriptional regulator with two DNA-binding domains (Table 1). We have previously described that a GA-adapted KT2440 strain, hereafter referred as P. putida KTGAL (Table S1), is able to use GA as sole carbon and energy source (Nogales et al., 2005). Interestingly, the nucleotide sequence of the gal cluster from P. putida KTGAL was identical to that of the wild-type strain unable to grow in GA, with the sole exception of a deletion of the cytosine located at position 2867088 of the KT2440 genome that caused the suppression of a frameshift mutation in galT* and allowed the expression of a GalT protein which presented the 11th and 12th transmembrane segments and the conserved motives that characterize aromatic acid/H+ symporter permeases (AAHS) from the MFS superfamily of transporters (Fig. S1) (Ditty and Harwood, 2002). To confirm that the molecular basis of the genetic adaptation of P. putida KTGAL to grow in GA was a single point mutation in the genome that restored a functional galT gene, GA uptake was monitored in the KT2440 and KTGAL strains. Whereas the parental KT2440 strain was unable to transport GA, the adapted strain showed a GA inducible uptake rate (Fig. 2A), indicating the existence of an efficient GA transport system in P. putida KTGAL. Moreover, a disruption of the galT gene in P. putida KTGALdgalT (Table S1) did not allow growth of the corresponding mutant strain in GA (data not shown). As expected, the KT2440 strain expressing the functional galT gene from plasmid pIZGalT recovered the ability to transport GA (Fig. 2A) and use GA as sole carbon source (Fig. 1B), demonstrating that deficient transport explains the lack of growth of P. putida KT2440 in GA.
Table 1. gal genes and their products in P. putida KT2440.
Distance to next gene (bp)
Gene product (aa/kDa)
Function of gene product
Related gene products
Accession No. or PDB code
. Only the first 368 amino acids of GalT* have been compared with AAHS transporters. This percentage of identity is observed along the whole amino acid sequence (449 residues) of a functional GalT transporter (Accession No. FN669140) encoded by galT in P. putida KTGAL.
The P. putida KTGAL cells expressing the galT gene from the multicopy plasmid pIZGalT showed an increased uptake not only of GA but also of its dihydroxylated analogue protocatechuate (Fig. 2A). Thus, GalT is a high-affinity permease that specifically recognizes GA [Vmax 8.4 nmol min−1 (mg of protein)−1, Km 4.3 µM] and, with less efficiency, protocatechuate [Vmax 3.2 nmol min−1 (mg of protein)−1, Km 9.1 µM] (Fig. 2B), but it does not transport other dihydroxylated acids, like homogentisate and caffeate, monohydroxylated acids, like 4-hydroxybenzoate, and some other GA analogues such as methyl gallate, syringate and pyrogallol. GA uptake was inhibited by agents that dissipate the electrochemical gradient across cell membranes (Fig. 2C), indicating that GalT is energized by the proton motive force. An unusual feature of some AAHS permeases is that they also mediate chemotaxis to the cognate aromatic compounds (Ditty and Harwood, 2002; Parales et al., 2008). The chemotactic response of P. putida KT2440 towards a range of metabolites can be analysed using agarose plug assays (Lacal et al., 2010). Whereas P. putida KTGAL showed a GA-inducible chemotaxis to GA, the P. putida KT2440 strain did not, suggesting that the GalT permease also mediates this chemotactic response (Fig. 2D). A P. putida KTGALdgalB mutant strain unable to degrade GA (see below) still showed a chemotactic response towards this aromatic acid (Fig. 2D), indicating that chemotaxis to GA does not require metabolism. The substrate specificity and phylogenetic analysis reveal that GalT is the first GA permease described in any organism and the founding member of a new group (TC 2.A.1.15.10) closely related to that of the archetypical 4-hydroxybenzoate/protocatechuate PcaK transporter (Ditty and Harwood, 2002; Parales et al., 2008).
The galP gene (Fig. 1C) encodes a protein that shows significant similarity to outer membrane porins of the OprD family (Table 1) (Tamber et al., 2006). Whereas disruption of galP in P. putida KTGALdgalP did not affect growth of the mutant strain in GA, the uptake of 10 µM GA [4.2 nmol min−1 (mg of protein)−1] decreased to half of the value obtained with the KTGAL strain [7.4 nmol min−1 (mg of protein)−1]. These data suggest that GalP plays an important physiological role in the uptake of low GA concentrations that may exist in the natural habitats of P. putida.
The gal cluster is responsible for GA catabolism in P. putida and heterologous hosts
To check whether other gal genes were also involved in GA catabolism, P. putida KTGALdgalA, dgalB, dgalC and dgalR knockout mutants were constructed (Table S1), and all of them were unable to use GA as sole carbon source (data not shown), suggesting their participation in the degradation of this aromatic compound. Growth in GA could be restored when the P. putida knockout mutants were complemented with a broad-host-range plasmid (pGAL) harbouring a DNA cassette containing the complete gal cluster (Table S1). Moreover, the pGAL plasmid also conferred the ability to use GA to other bacteria unable to degrade this aromatic compound, such as Pseudomonas sp. MT14 (Williams and Worsey, 1976) or Escherichia coli W (Davis and Mingioli, 1950), both of which lack the gal genes (Fig. 1B). Therefore, these results demonstrate that the gal cluster identified in P. putida constitutes the first complete set of genes reported so far in any organism that is responsible for the use of GA as carbon source.
Transcriptional organization of the gal cluster
An in silico analysis of the gene arrangement and length of the intergenic regions within the gal cluster suggest the existence of three putative transcriptional units, i.e. galBCD, galR and galTAP (Fig. S2A). The transcriptional organization of the gal cluster was determined by reverse transcription-PCR (RT-PCR) gene expression experiments with RNAs obtained from P. putida KTGAL cells grown in GA and primer pairs that were complementary to neighbouring genes for amplification of the four intergenic regions in the putative galBCD and galTAP operons (Fig. S2A). Since RT-PCR fragments of the expected sizes were obtained for genes galB-galC, galC-galD, galT-galA and galA-galP (Fig. S2B), these results suggest that the galBCD and galTAP genes are co-transcribed. To confirm the organization of the galBCD and galTAP genes as two transcriptional units, we checked whether there are polar effects derived from the insertional disruption of the galB and galT genes in P. putida KTGALdgalB and P. putida KTGALdgalT mutant strains respectively. RT-PCR experiments revealed that the insertional disruption of the galB and galT genes cause polar effects avoiding the expression of the galD and galP genes, respectively (Fig. S2C), thus confirming the existence of the galBCD and galTAP operons within the gal cluster. Moreover, since no RT-PCR fragments were obtained when the parental and mutant strains were grown in citrate in the absence of GA (Fig. S2B and C), the galBCD and galTAP operons were shown to be GA inducible.
Characterization of the GA ring-cleavage product
GA ring cleavage is the first enzymatic step in the proposed GA degradation pathway (Tack et al., 1972a) (Fig. 1A), and the galA and desB genes encoding a specific extradiol gallate dioxygenase have been characterized in P. putida KT2440 (Tack et al., 1972a; Nogales et al., 2005) and S. paucimobilis SYK-6 (Kasai et al., 2005) respectively. The product of the gallate dioxygenase-mediated reaction has been reported to be a mixture of the OMAketo and the two OMAenol isomers (Kersten et al., 1982; Hara et al., 2000; Kasai et al., 2005; Nogales et al., 2005) (Fig. 1A), although OMAenol was suggested to be the product of GA ring cleavage by purified protocatechuate 4,5-dioxygenase (Tack et al., 1972a). To determine unequivocally the chemical structure of the product generated by purified GalA acting on GA, NMR spectroscopy was performed. Since a single opening product whose 1H MNR spectra was consistent with the keto form rather than with the enol forms of OMA was detected (Fig. 3), we can conclude that OMAketo constitutes the real GA ring-cleavage product and therefore the controversy about the nature of the product of the GalA-mediated reaction has been finally clarified (Fig. 1A).
GalD catalyses a new biochemical step in the GA degradation pathway
The GA ring-cleavage product was proposed to be the substrate of the hydratase that catalyses the second enzymatic step in GA degradation (Tack et al., 1972a) (Fig. 1A). However, we were unable to detect any hydratase activity in P. putida KTGAL extracts when using as substrate a fresh solution of the GA ring-cleavage product. This result suggested the existence of an additional enzymatic step that converts OMAketo into the substrate of the hydratase. Although this conversion can occur spontaneously at a low rate, it becomes significantly increased when P. putida KTGAL extracts, but not P. putida KTGALdgalD extracts, were added to the reaction mixture (Fig. 4A), thus suggesting a GalD-mediated reaction. To confirm this assumption, the galD gene was overexpressed in E. coli BL21 (DE3) cells containing plasmid pETGalDc (Table S1 and Fig. S3). The spectral changes of the OMAketo after addition of overproduced GalD protein (Fig. 4B) were similar to those reported in keto–enolic tautomeries (Harayama et al., 1989; Whitman et al., 1991), suggesting that GalD was involved in a keto–enol isomerization of OMA. To elucidate the enzymatic reaction catalysed by GalD, 1H NMR spectra of the OMAketo after treatment with the GalD protein were analysed. The appearance of a new olefinic signal at a δ value of 6.73 ppm together with the absence of the signal at δ 7.38 ppm, which characterizes one of the olefinic protons of (Z)-OMAenol, is consistent with the enolization of the OMAketo to (E)-OMAenol form (Fig. 3). These results demonstrate that GalD is a tautomerase catalysing a new biochemical step, which remained unknown in the proposed GA degradation pathway (Fig. 1A). According to the data presented in Fig. 4A, the specific activity of GalD in the crude extract of P. putida KTGAL is 2.2 µmol min−1 (mg of protein)−1.
Disruption of the galD gene does not prevent growth of the P. putida KTGALdgalD strain in 5 mM GA, but the mutant strain grew poorly (A600 < 0.1) at GA concentrations lower than 2 mM. This observation revealed the importance of a functional galD gene for an efficient GA degradation, and suggests that the spontaneous isomerization of OMA and/or the unspecific activity of additional isomerases might replace the role of GalD when the mutant cells use high GA concentrations. It is worth noting that galD orthologues, e.g. orf1 and fldA genes found in the lig and fld clusters for protocatechuate mineralization via OMA in Sphingomonas strains (Table 1), were proposed to be involved in membrane-associated transport (Wattiau et al., 2001; Hara et al., 2003). In light of the data presented here, Orf1 and FldA can be tentatively reassigned as OMA tautomerases involved also in a previously unnoticed biochemical step of the well-studied protocatechuate 4,5-meta-cleavage pathway.
GalD shows moderate sequence identity with two isomerases that share a common structural fold (Table 1) (Alhapel et al., 2006; Garvey et al., 2007). Homology modelling of the three-dimensional structure of GalD reveals that the protein is composed of two structural domains that exhibit the same topology, i.e. a central α-helix surrounded by mainly antiparallel β-strands to form a pronounced β-barrel and an additional α-helix packed against the outside of β-barrel (Fig. S4). The cleft between the two domains contains the proposed active site (Fig. S4), as deduced by its similarity with the architecture of the active sites of PrpF and Mii isomerases (Garvey et al., 2007). Although the predicted overall fold of GalD with two domains that might evolve from a common ancestor via a duplication event has been also observed in other members of the diaminopimelate epimerase family (Pillai et al., 2006; Velarde et al., 2009), this fold has not been observed so far in other isomerases, e.g. XylH or HpcD/HpaF, involved in the catabolism of aromatic compounds (Roper et al., 1994; Subramanya et al., 1996; Taylor et al., 1998). Therefore, GalD constitutes the prototype of a new subfamily of isomerases that are involved in the metabolism of tricarboxylic acids, e.g. OMA, generated during the degradation of some aromatic compounds.
GalB is the founding member of a new family of zinc-containing hydratases
Cell-free extracts of GA-grown P. putida KTGAL exhibited OMA hydratase activity on OMAenol [1.0 µmol min−1 (mg of protein)−1], but this activity was not detected when OMAketo was used as substrate, suggesting that hydration of OMAenol was the third step in the GA degradation pathway (Fig. 1A). Although amino acid sequence analysis of the gal gene products did not reveal the presence of any putative hydratase (Table 1), the P. putida KTGALdgalB mutant strain lacked OMA hydratase activity (data not shown), pointing out that the galB gene might encode such activity. This was confirmed when a significant OMA hydratase activity was observed in extracts of E. coli BL21 (DE3) (pETGalB) cells (Table S1) overproducing the GalB protein (Fig. 5A). Since EDTA inhibited the GalB hydratase activity of cell extracts in a concentration-dependent manner (Fig. 6A), the presence of divalent cations appears to be essential for catalysis. EDTA-inactivated GalB could not be reactivated by the addition of metals, pointing that loss of the metal causes an irreversible inhibition of the protein. Cell extracts containing GalB showed a significant loss of OMA hydratase activity when diluted in the absence of metals (Fig. 6B), but the addition of Zn2+, and to a minor extent Co2+, reduced the enzyme inactivation (Fig. 6C). These findings suggested that GalB is a Zn2+-dependent enzyme and, accordingly, a functional GalB hydratase was purified by adding Zn2+ to the purification buffer. The GalB protein was purified from E. coli BL21 (DE3) (pETGalB) cell extracts and analysed by SDS-PAGE, showing a band whose apparent molecular mass (27 kDa) corresponded with that predicted for the galB gene product (27.4 kDa) (Fig. 5A and B). Analyses by inductively coupled plasma emission spectroscopy (ICP) of the metal content of purified GalB protein revealed 1 mol of this metal per mol of protein. Purified GalB was active in the pH range of 6.0–9.0, with maximum activity at pH 7.0. The optimal temperature was around 30°C and the Km and Vmax values for OMAenol were 85 µM and 76 µmol min−1 (mg of protein)−1 respectively. Amino acid modification studies with specific inhibitors of His, Cys, Ser and Asp/Glu residues revealed that only diethyl pyrocarbonate significantly reduced the hydratase activity to 15% of that of the untreated enzyme, suggesting the existence of His residue(s) in the active site of GalB. The GalB protein has a corrected sedimentation coefficient of 7.0 ± 0.1 S and a molar mass of 151.3 ± 12 kDa, which are compatible with the main OMA hydratase holoenzyme species being a globular homohexamer (Fig. 5C). 1H NMR analysis of the reaction product of OMAenol with the GalB hydratase revealed new aliphatic signals at δ values of 2.41 and 2.49 ppm (coupling constant 14.8 Hz), consistent with the outcome of two AB systems for the CH2 groups present in CHA (Fig. 3).
Although other OMA hydratases involved in the protocatechuate 4,5-cleavage degradation pathway have been previously characterized, e.g. LigJ (Hara et al., 2000) and ProH (Li et al., 2007), GalB does not show amino acid sequence identity with these enzymes. Moreover, these LigJ-type enzymes are homodimers whose activity is inhibited by Cys-specific reagents but not by chelators (Hara et al., 2000; Li et al., 2007). Nevertheless, the three-dimensional structure of the LigJ-type OMA hydratase from Rhodopseudomonas palustris (PDB 2GWG) revealed the existence of one Zn ion per subunit, with a His residue located near the Zn ion that could be involved in enzyme catalysis and with a Cys residue located on a flexible loop that lies just above the Zn ion so as to protect it from chelators and accounting for the enzyme inactivation by Cys-specific reagents (Li et al., 2007). Therefore, GalB constitutes the first member of a new family of OMA hydratases that have a different evolutionary origin and are predicted to use a different catalytic mechanism for the hydration of OMA than the LigJ-type enzymes. Interestingly, when the galB gene was transferred to S. paucimobilis DLJ, a derivative of S. paucimobilis SYK-6 unable to grow in syringic or vanillic acids since it lacks the ligJ gene encoding the corresponding OMA hydratase (Hara et al., 2000), the recombinant strain acquired the ability to grow in both aromatic compounds as sole carbon source (Fig. S5A and B). These experiments demonstrate that GalB is able to replace the function of LigJ, thus confirming its physiological role as an OMA hydratase and suggesting that the real substrate of the equivalent hydratases acting in the protocatechuate meta-cleavage pathway is also the OMAenol rather than the previously suggested OMAketo form (Hara et al., 2000).
Although the amino acid sequence of GalB does not show significant similarity in its entire length to any protein of known function (Table 1), a detailed analysis restricted to its N-terminal region revealed the presence of a sequence that resembles the conserved motif (V X P/A H P D D) found in the active site of deacetylases from the Pig-L family of mononuclear zinc-hydrolases (Fig. S6). The known three-dimensional structure of some of these mononuclear zinc-hydrolases revealed that the His and the second Asp residue of the conserved motif, together with a second His residue spaced approximately 100 residues from the former, constitute the Zn2+ binding site, acting the first Asp probably as general acid-base catalyst in the reaction (Maynes et al., 2003; McCarthy et al., 2004; Hernick and Fierke, 2005). Using the known three-dimensional structure of the MshB deacetylase (Maynes et al., 2003), a homology modelling for the active site of GalB was accomplished. The modelled active site of GalB fits the geometry of the Zn2+ binding site of MshB, being H14, D17 and H127 the proposed metal co-ordination residues. However, the catalytic Asp residue conserved in all deacetylases of the Pig-L family (D15 in MshB) is replaced by an Ala residue (A16 in GalB) in GalB-like proteins, and a conserved motif of GalB-like proteins (V/I S A H S/A A D) different from that of the classical Pig-L deacetylases can be proposed (Fig. S6). Whether the substitution of Asp by Ala in the active site of GalB might reflect that this protein is a hydratase rather than a hydrolase is a tempting hypothesis that requires further confirmation. All these data support that GalB is the prototype of a new family of Zn2+-containing hydratases that are evolutionary related to the Pig-L family of zinc-hydrolases rather than to other previously characterized hydratases involved in the catabolism of aromatic compounds (Pollard and Bugg, 1998; Li et al., 2007).
The galC gene encodes a CHA aldolase
Although the enzyme that catalyses the aldolic cleavage of CHA to produce pyruvic plus oxaloacetic acids (Fig. 1A) had been biochemically characterized many years ago (Tack et al., 1972b; Sparnins and Dagley, 1975), the cognate gene has remained unknown in P. putida. Here we show that the galC gene overexpressed in E. coli BL21 (DE3) cells harbouring plasmid pETGalC (Table S1) encodes a 25.1 kDa protein (Fig. S7A) that exhibits a significant CHA aldolase activity [24.5 µmol min−1 (mg of protein)−1] in cell extracts in the presence of Mg2+. Moreover, whereas extracts of P. putida KTGAL strain showed CHA aldolase activity [2.1 µmol min−1 (mg of protein)−1], this activity could not be detected in P. putida KTGALdgalC, and this strain was unable to grow in GA. These data indicate that the galC gene is essential for GA degradation encoding a type II (metal ion-dependent) CHA aldolase whose function cannot be replaced by that of other aldolases present in P. putida. GalC shows high sequence similarity to LigK and ProA, CHA aldolases from S. paucimobilis (Hara et al., 2003) and Pseudomonas straminea (Maruyama et al., 2001) respectively (Table 1).
Interestingly, GalC shows a significant amino acid sequence identity with proteins of the RraA family, e.g. the Yer010c protein from Saccharomyces cerevisiae (Table 1), that were originally misannotated as S-adenosylmethionine-dependent methyltransferases but whose three-dimensional structure resembles the phosphohistidine domains of phosphotransfer systems (Leulliot et al., 2005). Based on the known three-dimensional structure of the Yer010c protein (Leulliot et al., 2005), the predicted structure of the GalC monomer fits a α−β−β−α sandwich (Fig. S7B), being the last α5 and α6 helices likely involved in oligomerization of the protein and accounting for the hexameric quaternary conformation reported in native CHA aldolases from different bacteria (Tack et al., 1972b; Maruyama, 1990; Hara et al., 2003). A multiple sequence alignment among GalC, other CHA aldolases and proteins of the RraA family reveals highly conserved residues, including R123 and D124 in GalC (Fig. S7C), that have been postulated to be involved in a phosphotransfer reaction (Leulliot et al., 2005). Residues C95 and C159 in GalC are also conserved in other CHA aldolases (Fig. S7C), and they could be involved in catalysis and account for the previously described N-ethylmaleimide-dependent inhibition of the aldolase activity (Hara et al., 2003). The enzyme structure proposed here for GalC provides a new framework to advance in the study of the molecular mechanism of the CHA aldolases, and will help to elucidate the phosphate-dependent activation observed with some CHA aldolases (Maruyama, 1990; Hara et al., 2003), which might constitute an additional level of post-translational control of the cognate degradation pathway.
Unveiling gal clusters in bacterial genomes
In silico searches predicted the existence of gal clusters in the genomes of several Proteobacteria (Fig. 1C), most of which are able to colonize the plant rhizosphere. The overall organization of gal genes present in most γ-Proteobacteria is very similar, i.e. a galA-galT(galP) module, encoding GA uptake and ring cleavage, separated of the galBCD module, encoding the OMA degradation enzymes, by a regulatory galR gene. The similar organization and high gene identity may reflect a common evolutionary origin for the gal clusters in γ-Proteobacteria. Within the Pseudomonas genus, the gal genes are present only in P. putida strains, e.g. KT2440, F1 or W619, where they show > 95% sequence identity. This observation is in agreement with the GA-degradation abilities detected in this work with different P. putida strains, e.g. GPo1, U, F1 (data not shown), and with those previously reported in P. putida X.1 (Beveridge and Hugo, 1964) and in closely related Pseudomonas species (Chowdhury et al., 2004). The presence of a transposase-enconding gene (PP_2522) at the 5′ end of the galT gene, as well as the observation that the average GC content of the gal cluster (65.1%, Table 1) is higher than that of the whole genome (61.5%), and the fact that this cluster is located at the right end of genomic island 46 of strain KT2440 (Weinel et al., 2002), suggest that the gal genes constitute an evolutionary acquisition, via horizontal gene transfer, by P. putida to increase its metabolic proficiency as a saprophytic omnivore. In β- and α-Proteobacteria, some of which as Burkholderia sp. 383 were confirmed here as efficient GA degraders, the gene arrangement of the gal clusters shows more heterogeneity, and reveals that a ligJ-like gene replaces the galB gene in the genome of these α in β-Proteobacteria (Fig. 1C). This replacement is in agreement with our previous finding that although possessing different structures, both types of OMA hydratases and thus, their encoding genes, are interchangeable (Fig. S5A and B), and suggests that the ligJ-like genes might have been recruited from protocatechuate 4,5-meta-cleavage clusters.
Interestingly, all predicted gal clusters harbour the galA-galT module, which reinforces the key role of an active GA uptake for the consumption of this aromatic compound. The proof of concept of this assumption was to observe that whereas a derivative of S. paucimobilis SYK-6 was unable to use GA as sole carbon source, this strain harbouring the galT gene from P. putida KTGAL acquired the ability to grow in GA-containing minimal medium (Fig. S5C). Since we have found in several bacterial genomes gene clusters containing galRBC orthologues that could be involved in the metabolism of tricarboxylic acids analogous to OMA, it is tempting to speculate that the association of the latter to the galA-galT module might have evolved into a functional gal cluster. The galD and galP are the only gal genes that were not strictly essential for the catabolism of GA in P. putida (see above), and they are lacking in some gal clusters (Fig. 1C), which suggests that their function in GA degradation can be replaced by the unspecific activity of other genes present in the corresponding bacterial genomes.
The characterization of the gal cluster reveals the existence of a new and paradigmatic aromatic catabolic pathway that surprisingly remained unknown in one of the best-studied model bacteria, P. putida KT2440. The gal genes from P. putida became useful to identify homologous gene clusters annotated as of unknown function in bacteria whose ability to degrade GA was unexplored so far, and thus expanding outside the Pseudomonas genus our current view on aerobic GA degraders. The GalT protein mediates specific GA uptake and chemotaxis in P. putida KTGAL, and highlights the critical role of transporters in the adaptation of bacteria to aromatic carbon sources, such as GA, that can easily oxidize and become unavailable for the cells. Remarkably, proteins as the new OMA tautomerase (GalD) and the OMA hydratase (GalB) constitute the founding members of new families of enzymes, and they allowed to revisit the proposed GA biochemical pathway, as well as to assign new functions to some of the gene products involved in the well-studied protocatechuate 4,5-meta-cleavage pathway. The gal cluster is also a useful biotechnological tool for expanding the metabolic abilities of bacteria towards GA degradation, and it has been used also for engineering the first GA cellular biosensor able to detect GA concentrations as low as 0.5–1 µM (Fig. S8).
Bacterial strains, plasmids and growth conditions
The bacterial strains, plasmids and oligonucleotides used in this study are listed in Table S1. Bacteria were grown in LB medium (Sambrook and Russell, 2001) or MC minimal medium (Nogales et al., 2005) at 37°C (E. coli) or 30°C (Pseudomonas, Burkholderia, Sphingomonas). When used as carbon sources, 0.2% citrate, and 5 mM syringate or vanillate, were added to the MC medium. Antibiotics were used at the following concentrations: ampicillin (100 µg ml−1), kanamycin (50 µg ml−1), gentamicin (50 µg ml−1), streptomycin (50 µg ml−1).
Pseudomonas putida KT2440 was adapted to use 5 mM GA as sole carbon source by using MC minimal medium (pH 6.5) containing 2 mM l-cysteine as reducing agent. The slightly acidic pH and low amount of Fe2+ in the MC medium, as well as the presence of l-cysteine avoid the rapid oxidation of GA and therefore the formation of a black polyphenol that cannot be used by the cells. After a 24 h incubation of P. putida KT2440 cells in GA-containing MC medium, poor growth and a black coloration of the medium, due to GA oxidation, was observed. The culture was then centrifuged, and the cells were inoculated again into fresh medium. After an additional 24 h incubation, a reasonable cell growth and a slightly brown coloration of the medium was observed. This culture was then centrifuged, and the cells inoculated again into fresh medium. After an additional 12 h incubation, a significant cell growth (A600 of about 0.9) and the lack of brown coloration in the culture medium was observed. These GA-adapted cells were named P. putida KTGAL.
Molecular biology techniques
Standard molecular biology techniques were performed as previously described (Sambrook and Russell, 2001). PCR products were purified with the High Pure plasmid isolation kit (Roche Applied Science). DNA fragments were purified with Gene-Clean Turbo (Q-BIOgene). Genomic DNA from P. putida KT2440 was isolated with GenomicPrep Cells and Tissue DNA Isolation kit (Amersham Biosciences). All cloned inserts and DNA fragments were confirmed by DNA sequencing with fluorescently labelled dideoxynucleotide terminators (Sanger et al., 1977) and AmpliTaq FS DNA polymerase (Applied Biosystems) in an ABI Prism 377 automated DNA sequencer (Applied Biosystems). Transformation of E. coli cells was carried out by using the RbCl method (Sambrook and Russell, 2001) or by electroporation (Gene Pulser, Bio-Rad). Oligonucleotides were purchased from Sigma. The proteins were analysed by SDS-PAGE and Coomassie-stained according to standard protocols (Sambrook and Russell, 2001). The protein concentration was determined by the method of Bradford (1976) using BSA as the standard.
Construction of P. putida KTGALdgal mutant strains
To construct the P. putida KTGALdgalB, KTGALdgalC, KTGALdgalD, KTGALdgalP, KTGALdgalR and KTGALdgalT mutant strains, an internal fragment (300–700 bp) of the target gal gene was PCR-amplified and cloned into the polylinker of pK18mob (Table S1), a mobilizable plasmid that does not replicate in Pseudomonas. To transfer the corresponding pK18gal plasmids into P. putida KTGAL, triparental filter matings were performed as previously described (de Lorenzo and Timmis, 1994) using E. coli DH10B (pK18gal) as donor strain, E. coli HB101 (pRK600) as helper strain and P. putida KTGAL as recipient strain. P. putida KTGAL exconjugants harbouring the gal-disrupted gene were isolated on MC minimal medium plates containing citrate (which selected for the Pseudomonas recipient cells) and kanamycin (which selected for the insertion of the suicide vector) after incubation at 30°C for 16 h (Nogales et al., 2005). All mutant strains were analysed by PCR to confirm the disruption of the target gene.
Recombinant plasmids constructions
To construct the pGAL plasmid, a 9.6 kb EcoRI fragment containing the entire gal cluster from PPZIA58, a Lawrist 7-derived cosmid that contains a 41 kb DNA fragment from position 2.854.165 to 2.895.166 of the P. putida KT2440 chromosome (Table S1), was cloned into the EcoRI-digested broad-host-range plasmid pBBR1MCS-5 (Table S1), generating the pGALKT plasmid (14.4 kb). The AclI/XhoI fragment containing part of galA and the entire galT* gene from pGALKT plasmid was replaced by the homologous AclI/XhoI fragment PCR-amplified from P. putida KTGAL, generating the pGAL plasmid (13.6 kb) that expresses the complete gal cluster from P. putida KTGAL (Table S1). The pIZGalT plasmid expressing the galT gene under the control of the LacIq/Ptac promoter was constructed by cloning first the 2.5 kb NdeI/blunt PCR-amplified galT gene from the KTGAL strain into the NdeI/EcoRV double-digested pET29-a(+) plasmid, producing plasmid pETGalT (Table S1). An XbaI/SacI fragment containing the galT gene was then subcloned from pETGalT into pIZ1016 giving rise to plasmid pIZGalT (Table S1). The pETGalB plasmid expressing the galB gene under the PT7 promoter was constructed by cloning a 0.7 kb NdeI/SacI PCR-amplified galB gene into pET29-a(+). The pETGalDc and pETGalC plasmids expressing the galD and galC genes under control of the PT7 promoter were constructed by cloning the 1.2 kb NdeI/SalI and the 0.7 kb NdeI/SacI PCR-amplified galD and galC genes into pET29-a(+) respectively (Table S1). Plasmid pIZGalB expresses the galB gene under the control of the LacIq/Ptac promoter, and it was generated by subcloning the 0.7 kb XbaI/SacI galB gene from plasmid pETGalB into the XbaI/SacI double-digested pIZ1016 plasmid (Table S1). The pIZGalBT plasmid, which expresses both the galB and galT genes under the LacIq/Ptac promoter, was generated by cloning a 0.7 kb EcoRI/blunt PCR-amplified galB gene into the EcoRI/SmaI double-digested pIZGalT plasmid (Table S1).
For RT-PCR experiments, cultures of P. putida grown in MC medium with GA or citrate as carbon source were collected at an A600 of 0.5. Total RNAs were obtained with the RNeasy kit (Qiagen). Any contamination by DNA was eliminated by the use of a DNase treatment and removal kit (Ambion). One microgram of purified total RNA was used to prepare cDNA by the use of 200 U of SuperScript reverse transcriptase (Invitrogen), 0.5 mM dNTPs and 0.5 µM of the cognate oligonucleotide. The cDNA obtained was amplified with 1 U of AmpliTaq DNA polymerase (Biotools) and 0.5 µM of the corresponding primer pairs (Table S1). Control reactions in which reverse transcriptase was omitted from the reaction mixture ensured that DNA products resulted from the amplification of cDNA rather than from DNA contamination.
Phenolic compounds transport assays
Measurement of transport of GA and other phenolic compounds was accomplished by using a laccase-based electrochemical biosensor (Gamella et al., 2006). Briefly, by coupling the decrease of an electrochemical signal to a proportional decrease in the content of a phenolic compound in solution, a real-time estimation of cell-mediated transport of such particular compound can be monitored. The transport assay (5 ml) contains different initial concentrations of the target aromatic compound in 100 mM sodium phosphate buffer, pH 6.0. P. putida cells grown until mid-exponential phase were harvested, washed twice and added to the transport mixture. The decrease of the electrochemical signal, which corresponds with a proportional decrease in the aromatic compound concentration, was monitored for 5 min at 25°C. To check the substrate range of the GalT transporter, the transport assays were carried out in the presence of 50 µM of different phenolic compounds, i.e. GA, methylgallate, syringate, protocatechuate, 4-hydroxybenzoate, pyrogallol, homogentisate, caffeate and catechol.
GA chemotaxis assays
Chemotaxis to GA was measured by using the agarose plug assay. Agarose plug assays were carried out as previously described (Parales et al., 2000) with slight modifications. Plugs contained 2% low-melting agarose in MC medium (Nogales et al., 2005) harbouring 1 mM GA. A drop (10 µl) of the melted agarose mixture was placed on a microscope slide, and a coverslip supported by two plastic strips was then placed on top to form a chamber. P. putida cells were grown in 0.2% citrate-containing MC medium in the absence or presence of 2 mM GA until the cultures reached mid-exponential phase, and then flooded into the chamber to surround the agarose plug. The chemotactic response was monitored under the phase-contrast microscope 5 min after addition of cells. Control plugs contained no attractant, and no response was seen.
P. putida cell extract preparations
Pseudomonas putida cells were grown until the cultures (0.2 l) reached an absorbance at 600 nm of 0.5. Cells were harvested at 4°C and resuspended in 10 ml of extraction buffer: (i) 50 mM HEPES, pH 7.0, 10% glycerol (w/v), 0.2 mM ZnSO4 and 1 mM DTT for OMA hydratase, and (ii) 50 mM sodium phosphate, pH 7.0 for OMA isomerase and CHA aldolase. Cells were then disrupted by two consecutive passages through a French press (Aminco Corp.) operated at a pressure of 20 000 p.s.i. The cell lysates (crude extracts) were centrifuged at 13 000 g for 30 min at 4°C. The clear supernatant fluid was carefully decanted and used as the soluble fraction of the crude extract.
Overproduction of GalB, GalC and GalD proteins
Escherichia coli BL21(DE3) cells harbouring plasmids pETGalDc, pETGalB or pETGalC were grown in LB medium at 37°C until the cultures (0.2 l) reached an absorbance at 600 nm of 0.6. Overexpression of the cloned genes was then induced for 3 h by the addition of 0.1 mM isopropyl-1-thio-β-d-galactopyranoside. Cells were harvested at 4°C and resuspended in 10 ml of extraction buffer (see above). Cells were then disrupted by two consecutive passages through a French press (Aminco Corp.) operated at a pressure of 20 000 p.s.i. The cell lysates (crude extracts) were centrifuged at 13 000 g for 30 min at 4°C. The clear supernatant fluid was carefully decanted and used as the soluble fraction of the crude extract.
Purification of the GalA dioxygenase and the GalB hydratase
The overproduction and purification of GalA was as previously described (Nogales et al., 2005). The purification of the OMA hydratase (GalB) from E. coli BL21(DE3) (pETGalB) was performed at 4°C by a three-step procedure in the presence of 0.2 mM ZnSO4. The E. coli BL21(DE3) (pETGalB) soluble extract (20 ml, 230 mg of protein) was ultracentrifuged at 80 000 g for 60 min (step 1, ultracentrifugation fraction). NaCl was added to the supernatant to a final concentration of 0.05 M and loaded onto a DEAE-cellulose column (Sigma-Aldrich) previously equilibrated with extraction buffer (see above) containing 0.05 M NaCl. Proteins were eluted from the column with a linear gradient (0.05 to 0.5 M) of NaCl in 0.1 l of extraction buffer. The OMA hydratase GalB was eluted at approximately 0.3 M NaCl (step 2, DEAE cellulose fraction). The fractions containing OMA hydratase activity were pooled, NaCl was added to a final concentration of 0.5 M, and the solution was loaded onto a phenyl-Sepharose CL-4B column (Amersham Biosciences) previously equilibrated with extraction buffer containing 0.5 M NaCl. Proteins were eluted from the column with a linear gradient (0.5 to 0 M) of NaCl. Finally, the fractions containing OMA hydratase activity, which eluted around 0.1 M NaCl, were pooled, desalted in extraction buffer and stored at −20°C.
Analytical ultracentrifugation analyses
Sedimentation velocity experiments of purified OMA hydratase (GalB) solutions in extraction buffer [50 mM HEPES, pH 7.0, 10% (w/v) glycerol, 0.2 mM ZnSO4 and 1 mM DTT] were carried out at 40 000 r.p.m. and 20°C in an XL-A analytical ultracentrifuge (Beckman-Coulter) equipped with UV-visible absorbance optics, using an An60Ti rotor and 12 mm double-sector centrepieces of Eponcharcoal. Absorbance scans were measured at 280 nm. Sedimentation coefficient distributions, c(s), were calculated by least-squares boundary modelling of sedimentation velocity data using the program sedfit (Schuck, 2000). These coefficients were corrected to standard conditions (water and 20°C) to obtain the corresponding s20,w values using the sednterp program (Minton, 1994). The latter program was also used to calculate the partial specific volume of the protein from its amino acid composition. Short-column (80 µl) sedimentation equilibrium experiments were performed at two successive speeds (10 000 and 13 000 r.p.m.) on parallel protein samples. Afterward, baselines were measured at high speed (40 000 r.p.m.). The weight-average molar mass of the protein was calculated by fitting the experimental gradients to the equation that describes the radial concentration distribution of a solute at sedimentation equilibrium, as implemented in the eqassoc program (Minton, 1994).
Synthesis of OMAketo, OMAenol and CHA substrates
Since GA (Sigma-Aldrich) is the only commercially available substrate of the GA degradation pathway, enzymatic synthesis was necessary for the production of the rest of the GA degradation pathway intermediates. OMAketo (200 µM) was freshly prepared routinely by incubation for 5 min at 25°C of 200 µM GA with 5 µg of purified GalA in 50 mM sodium phosphate buffer, pH 7.0. OMAenol (200 µM) was prepared by incubation of the fresh OMAketo form preparation for 5 min at 25°C with 5 µg of E. coli BL21 (DE3) (pETGalDc) extracts. Since the ε265 for OMA calculated previously did not differentiate between the keto and enol forms (Maruyama, 1985; Hara et al., 2000), we have determined in this work an ε265 = 2.000 M−1 cm−1 for OMAenol. CHA (200 µM) was prepared by incubating 200 µM OMAenol with 5 µg of purified OMA hydratase for 5 min at 25°C in 50 mM sodium phosphate buffer, pH 7.0.
Gallic acid dioxygenase (GalA) was assayed as previously described (Nogales et al., 2005). The OMA tautomerase (GalD) and OMA hydratase (GalB) activities were assayed spectrophotometrically at 30°C by measuring the increase of A265 (ε = 2.000 M−1 cm−1) due to the conversion of OMAketo to OMAenol, and the decrease of A265 due to the disappearance of OMAenol, respectively, in 50 mM sodium phosphate buffer, pH 7.0. The CHA aldolase (GalC) activity was assayed at 30°C by monitoring the decrease in A340 derived from NADH oxidation (ε = 6.6 × 103 M−1 cm−1) in a coupled assay containing 200 µM CHA, 140 µM NADH, 30 U lactate dehydrogenase (Roche), MgCl2 1 mM, and crude extract from E. coli BL21 (DE3) (pETGalC) (1 µg) or P. putida KTGAL (5 µg), in 50 mM sodium phosphate buffer, pH 7.0, as previously described (Maruyama, 1990). The β-galactosidase activity of permeabilized P. putida KTGAL (pIZPb) cells was monitored as previously described (Miller, 1972).
Chemical modifications of GalB by site-specific reagents
Stock solutions of the reagents were made up fresh daily. The cysteine-specific reagents, N-ethylmaleimide and iodoacetamide, were prepared at a concentration of 0.2 M by dissolving in water. The glutamic/aspartic acid-specific reagent 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) was prepared at a concentration of 0.2 M by dissolving in water. The histidine-specific reagent diethyl pyrocarbonate (DEPC) was diluted to 0.5 M in acetonitrile. The serine-specific reagent phenylmethylsulphonyl fluoride (PMSF) was diluted to 0.1 M in isopropyl alcohol. Inactivation of purified GalB enzyme was performed by incubation for 30 min (7 min when DEPC was used) at 25°C in 50 mM sodium phosphate buffer, pH 7.0 containing the inhibitor, i.e. 1 mM N-ethylmaleimide, 0.5 mM PMSF, 5 mM iodoacetamide, 1 mM EDC or 0.5 mM DEPC. The remaining OMA hydratase activity was assayed and compared with that of the untreated enzyme.
Metabolite and Zn2+ detection
Metabolite detection was carried out by 1H MNR and samples were prepared in 10% D2O (pH 7.0). The spectra were acquired at 298 K on an AVANCE 500 MHz spectrometer, equipped with a broad-band z-gradient probe (Bruker). The Watergate module was employed to suppress the residual water resonance. The 1H NMR chemical shifts are given using trimethylsilyl propionate as a reference (0 ppm). The Zn2+ content in purified GalB was determined by ICP using an ICP-OES Optima 2000DV equipment (PerkinElmer).
Three-dimensional modelling of GalB, GalC and GalD
The three-dimensional models for GalB, GalC and GalD from P. putida KT2440 were built by using the alignment mode at the SWISS-MODEL Protein Homology-Modeling Server (http://swissmodel.expasy.org/) (Arnold et al., 2006). The crystal structures of the MshB deacetylase from Mycobacterium tuberculosis H37Rv (PDB 1Q7T), the Yer010c protein from S. cerevisiae (PDB 2C5Q) and the PrpF 2-methylcitrate cis-trans-isomerase from Shewanella oneidensis (PDB 2PVZ) were used as templates for GalB, GalC and GalD modelling respectively. The figure management was performed by using the PyMOL viewer program (DeLano, 2002).
We thank A. Valencia for technical assistance; C. Alfonso and G. Rivas for the ultracentrifugation analyses; S.L. Secugen for DNA sequencing; D. Stjepandic and J.D. Hoheisel for the gift of the P. putida cosmid library; and E. Masai and E. Mahenthiralingam for strains S. paucimobilis DLJ and Burkholderia sp. 383 respectively. This work was supported by Grants GEN2001-4698-C05-02, GEN2006-27750-C5-3-E, CSD2007-00005, BIO2009-10438. J.N. was the recipient of an I3P predoctoral fellowship from the CSIC.