Metabolic reconstruction of aromatic compounds degradation from the genome of the amazing pollutant-degrading bacterium Cupriavidus necator JMP134
Laboratorio de Microbiología, Departamento de Genética Molecular y Microbiología, Center for Advanced Studies in Ecology and Biodiversity, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Millennium Nucleus on Microbial Ecology and Environmental Microbiology and Biotechnology, Santiago, Chile
Laboratorio de Microbiología, Departamento de Genética Molecular y Microbiología, Center for Advanced Studies in Ecology and Biodiversity, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Millennium Nucleus on Microbial Ecology and Environmental Microbiology and Biotechnology, Santiago, Chile
Laboratorio de Microbiología, Departamento de Genética Molecular y Microbiología, Center for Advanced Studies in Ecology and Biodiversity, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Millennium Nucleus on Microbial Ecology and Environmental Microbiology and Biotechnology, Santiago, Chile
Correspondence: Bernardo González, Laboratorio de Microbiología. Departamento de Genética Molecular y Microbiología, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Casilla 114-D, Santiago, Chile. Tel.: 56 2 6862845; fax: 56 2 6862185; e-mail: email@example.com
Cupriavidus necator JMP134 is a model for chloroaromatics biodegradation, capable of mineralizing 2,4-D, halobenzoates, chlorophenols and nitrophenols, among other aromatic compounds. We performed the metabolic reconstruction of aromatics degradation, linking the catabolic abilities predicted in silico from the complete genome sequence with the range of compounds that support growth of this bacterium. Of the 140 aromatic compounds tested, 60 serve as a sole carbon and energy source for this strain, strongly correlating with those catabolic abilities predicted from genomic data. Almost all the main ring-cleavage pathways for aromatic compounds are found in C. necator: the β-ketoadipate pathway, with its catechol, chlorocatechol, methylcatechol and protocatechuate ortho ring-cleavage branches; the (methyl)catechol meta ring-cleavage pathway; the gentisate pathway; the homogentisate pathway; the 2,3-dihydroxyphenylpropionate pathway; the (chloro)hydroxyquinol pathway; the (amino)hydroquinone pathway; the phenylacetyl-CoA pathway; the 2-aminobenzoyl-CoA pathway; the benzoyl-CoA pathway and the 3-hydroxyanthranilate pathway. A broad spectrum of peripheral reactions channel substituted aromatics into these ring cleavage pathways. Gene redundancy seems to play a significant role in the catabolic potential of this bacterium. The literature on the biochemistry and genetics of aromatic compounds degradation is reviewed based on the genomic data. The findings on aromatic compounds biodegradation in C. necator reviewed here can easily be extrapolated to other environmentally relevant bacteria, whose genomes also possess a significant proportion of catabolic genes.
Cupriavidus necator JMP134 (ex Alcaligenes eutrophus; ex Ralstonia eutropha; ex Wautersia eutropha) was isolated from an Australian soil by its ability to grow on 2,4-dichlorophenoxyacetate (2,4-D) (Pemberton et al., 1979; Don & Pemberton, 1981). Early studies also showed that this strain grows on 4-methyl-2-chlorophenoxyacetate (MCPA) and 3-chlorobenzoate (3-CB) (Pemberton et al., 1979; Don & Pemberton, 1981), and is resistant to mercurial compounds. The determinants for 2,4-D and 3-CB degradation and for mercury resistance are encoded in the catabolic plasmid pJP4 (Don & Pemberton, 1981). The presence of the tfd genes on this plasmid, which encode the ortho ring-cleavage pathway for chlorocatechol intermediates, facilitated the use of this bacterium as a model for chloroaromatic degradation. A number of articles have dealt with the presence of tfd and tfd-like sequences both in pristine and in polluted environments, and tfd genes have also been used to track the presence and transference of the pJP4 plasmid and related plasmids in the environment. These topics have been covered in a couple of reviews (Top et al., 1998, 2002).
In recent years, the discovery of a second pJP4-encoded tfd genes cluster has led to a revision of the role of the tfd-encoded functions (Leveau et al., 1999; Laemmli et al., 2000, 2004; Perez-Pantoja et al., 2000; Plumeier et al., 2002; Schlomann, 2002). In addition, increasing evidence indicates that C. necator JMP134 grows on several other pollutants and natural compounds using chromosomally encoded functions (Schlomann et al., 1990b; Ecker et al., 1992; Schenzle et al., 1997; Padilla et al., 2000; Louie et al., 2002). The recent availability of the complete genome sequence of several environmentally relevant bacteria such as Pseudomonas putida KT2440 (Nelson et al., 2002); Burkholderia xenovorans LB400 (Chain et al., 2006); Rhodococcus sp. RHA1 (McLeod et al., 2006), and the denitrifying bacterium ‘Aromatoleum aromaticum’ sp. strain EbN1 (Rabus et al., 2005), has revealed the presence of a significant number of genes encoding determinants for aromatic compounds degradation. This fact suggests that different bacteria are potentially able to degrade several types of both natural and man-made aromatic compounds. The genome sequence of C. necator also reveals the presence of a significant (more than 5% of all the putative genes) proportion of aromatics degradation genes. We selected this versatile pollutant-degrading bacterium as a model of aromatic compounds degraders, and performed the metabolic reconstruction of aromatic compounds degradation. In this review, we analyze the main features of the catabolism of aromatic compounds in C. necator JMP134 within the context of the abundant literature on this topic. Special attention is given to those aspects that would explain the impressive catabolic versatility of this and other bacteria.
Aromatic growth substrates for C. necator JMP134
Out of 140 aromatic substrates tested, 60 can be used by C. necator JMP134 as the sole carbon and energy source; these include c. 40 compounds that have not been reported previously as growth substrates. The growth supporting aromatic compounds are (brackets indicate the section where the corresponding degradation pathway is discussed): benzoate, benzaldehyde, benzyl alcohol, phenylglyoxylate (benzoylformate), benzyl acetate, benzylamine (‘The cat and ben genes’, benzoate also in ‘The aerobic benzoyl-CoA pathway’); 4-hydroxybenzoate, 3,4-dihydroxybenzoate (protocatechuate), chlorogenate, quinate (‘The pob and pca genes’); phenylpropionate, cinnamate, 4-hydroxyphenylpropionate, 4-hydroxycinnamate (coumarate), 3,4-dihydroxyphenylpropionate, 3,4-dihydroxycinnamate (caffeate), ferulate (‘Peripheral pathways that channel phenylpropenoid and phenylpropanoid compounds to the β-ketoadipate pathway’, and some of these compounds in ‘The 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway’); 3-hydroxyphenylpropionate (‘The 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway’); 2-hydroxy (salicylate), 3-hydroxybenzoate, 3-hydroxybenzyl alcohol, 2,5-dihydroxybenzoate (gentisate), ethylsalicylate (‘Degradation of salicylate and 3-hydroxybenzoate: the gentisate pathway’); phenol, 2-, 3-, and 4-methylphenol, 2,3- and 3,4-dimethylphenol, 4-ethylphenol, 2-methylphenoxyacetate, benzene, toluene [‘The catabolic pathways for benzene, toluene and (methyl)phenols’]; phenylacetate, phenylacetaldehyde, phenylethylamine, phenylpyruvate, 4-phenylbutyrate, 5-phenylvalerate, 6-phenylhexanoate (‘The phenylacetyl-CoA ring-cleavage pathway’); 2-, 3-, and 4-hydroxyphenylacetate, 4-hydroxyphenylpyruvate, phenylalanine, tyrosine (‘The homogentisate ring-cleavage pathway’); tryptophan, 2-aminobenzoate (anthranilate) [‘The 2-aminobenzoyl-CoA pathway’]; 3-hydroxyanthranilate (‘The 3-hydroxyanthranilate pathway’); 3-nitrophenol, 2-chloro-5-nitrophenol (‘Catabolic pathways for nitrophenols’); hydroquinone, 2,4,6-trichlorophenol [‘The (chloro)hydroxyquinol ring-cleavage pathway’]; 4-fluorobenzoate (‘Fluorobenzoate catabolism’); 3-CB, 2,4-D and MCPA (‘Catabolic pathway for mono- and dichlorinated compounds: the tfd genes’). In addition, strain JMP134 grows on 2,6-dinitrophenol and 3,5-dihydroxybenzoate, although for these compounds no degradation pathway could be identified. Cell yields range from 0.9 to 5.7 mg cells mmol−1 of carbon. Chlorinated compounds, aromatic aldehydes and ferulate produce lower growth yields. Eighty compounds failed to support the growth of this strain: 2- and 4-chlorobenzoate, 3,5-dichlorobenzoate, 3-chloro-4-hydroxybenzoate, 2- and 4-chlorophenoxyacetate, 3-(2,4-dichlorophenoxy)propionate, 4-(2,4-dichlorophenoxy)butyrate, 2-, 3- and 4-chlorophenylacetate, 3-chloro-4-hydroxyphenylacetate, tropate, 2- and 3-phenoxypropionate, 4-phenoxybutyrate, 2-hydroxyphenylpropionate, 2-hydroxycinnamate, 2-phenylpropionate, 2-, 3- and 4-methoxybenzoate, 2,6-, 2,3- and 2,4-dihydroxybenzoate, 3- and 4-methylbenzoate, 3- and 4-phenoxybenzoate, 2- and 4-hydroxybenzylalcohol, 3,5-dihydroxybenzylalcohol, phthalate, 1,2-, 1,3- and 1,4-dimethylbenzene, ethylbenzene, chlorobenzene, biphenyl, coniferyl alcohol, vanillyl alcohol, 4-hydroxy-3-methoxybenzoate (vanillate), 3-hydroxy-4-methoxybenzoate (isovanillate), isovanillyl alcohol, vanillaldehyde, 3-hydroxy-4-methoxycinnamate, 5-chlorovanillate, 2-, 3- and 4-chlorophenol, 2,3-, 2,5-, 2,6-, 3,4- and 3,5-dichlorophenol, 2,3,4-, 2,3,5- and 2,4,5-trichlorophenol, pentachlorophenol, 2,4- and 3,5-dimethylphenol, 4-propylphenol, resorcinol, mandelate, 4-hydroxymandelate, 3-, 4- and 5-chlorosalicylate, 3,5-dichlorosalicylate, 5-methoxysalicylate, syringate, 4-hydroxyacetophenone, acetophenone, 2-, 3- and 4-nitrobenzoate, 2- and 4-nitrophenol, 4-sulfophenol, 4-sulfobenzoate.
The ortho ring-cleavage pathways for benzoate and p-hydroxybenzoate (the β-ketoadipate central pathway)
The β-ketoadipate pathway is widely distributed among soil bacteria because it plays a central role in the degradation of naturally occurring aromatic compounds (Harwood & Parales, 1996). Some environmental pollutants are also degraded through this pathway. The central reactions of the different branches of the β-ketoadipate pathway in C. necator JMP134 are shown in Fig. 1. The catechol branch, encoded by cat genes, converts the catechol generated from benzoate (through the action of the ben gene products), phenol and some lignin monomers, into β-ketoadipate. The protocatechuate branch, encoded by pca genes, converts the protocatechuate derived from 4-hydroxybenzoate (through the action of the pob gene products) and numerous lignin monomers, into β-ketoadipate. Two additional steps accomplish the conversion of β-ketoadipate into the Krebs cycle intermediates: succinyl-CoA and acetyl-CoA (Harwood & Parales, 1996). Biochemical studies and amino acid sequence data indicate that the pathway enzymes are highly conserved among the phylogenetically diverse organisms that possess this pathway. Despite this biochemical conservation, studies of a limited number of soil bacteria demonstrate a remarkable diversity of this pathway in terms of gene organization, type of inducers and regulation mechanism (Harwood & Parales, 1996).
The cat and ben genes
The ben and cat gene products of C. necator JMP134 are highly similar, in their amino acid sequence, to proteins of the catechol branch of the β-ketoadipate pathway that has been characterized in other bacteria, mainly Acinetobacter, Pseudomonas, and Burkholderia (Table 1). The gene encoding CatA1, which starts the catechol branch of the β-ketoadipate pathway (Fig. 1), clusters together with the benABCD genes (small chromosome [C2] in Fig. 2), which are responsible for funneling benzoate into this pathway. This putative operon includes the catR1 gene; this gene encodes a putative LysR-type regulatory protein. A gene putatively encoding a muconate cycloisomerase, catB1, is localized close to the ben and catA1 genes, but is not part of this putative operon (Fig. 2, C2). A similar organization of the β-ketoadipate pathway genes has been observed previously in Burkholderia sp. strain TH2 (Suzuki et al., 2002), Burkholderia sp. NK8 (Francisco et al., 2001), and C. necator 335 (accession number AF042281, I.S. Hinner et al., unpublished data). However, unlike those strains where a gene encoding muconolactone isomerase is positioned directly downstream of the muconate cycloisomerase encoding gene, no additional β-ketoadipate pathway genes are localized downstream of catB1 in strain JMP134 (Fig. 2, C2). The catC gene is clearly located in a second operon, which also encodes a muconate cycloisomerase and an enol lactone hydrolase, namely enzymes for the conversion of muconate to β-ketoadipate (large chromosome [C1] in Fig. 2). This gene organization contrasts the one previously observed in Pseudomonas strains, where operons encoding cat genes do not comprise a catD gene; it also supports reports suggesting that catechol and protocatechuate branches of the β-ketoadipate pathway in Cupriavidus strains converge at the stage of β-ketoadipate rather than β-ketoadipate enol lactone, as is the case in Pseudomonas strains (Harwood & Parales, 1996; Jimenez et al., 2002). Genes encoding β-ketoadipyl-CoA transferase and thiolase are not comprised in the ben/cat genes operons (see next section), as opposed to Acinetobacter, where a full set of genes for the transformation of muconate into succinyl-CoA and acetyl-CoA is present in the cat operons (Harwood & Parales, 1996).
Table 1. Genes encoding the β-ketoadipate pathway, the methylmuconolactone pathway and peripheral reactions
Both catA1benABCD and catB2CXD gene clusters are preceded by putative catR genes that encode LysR-type regulatory proteins. An interesting point for further investigation would be whether both catR genes are required for full induction of the benzoate degradation pathway. Benzoate and muconate have been reported as inducers of the benzoate pathway in other bacteria, through the activity of LysR-type regulatory proteins (Harwood & Parales, 1996; McFall et al., 1998; Bundy et al., 2002). It could be speculated that CatR1 gene product is responsive to benzoate and muconate, whereas the CatR2 gene product responds predominantly to the muconate generated by the CatA1 gene product, which indicates a sequential and modular induction of the cat genes, similar to that observed in Acinetobacter baylyi ADP1 (Ezezika et al., 2006). Further evidence for a modular gene organization is the fact that a second catechol-1,2-dioxygenase encoding gene, catA2 (Fig. 2, C1, and pathway in Fig. 3), is localized in a cluster of genes encoding a multicomponent phenol hydroxylase [see ‘The catabolic pathways for benzene, toluene and (methyl)phenols’]; this enzyme would probably channel at least some phenol when added as growth substrate, into the β-ketoadipate pathway (Pieper et al., 1989). CatA1 and CatA2 gene products from C. necator cluster together in the dendrogram of the intradiol 1,2-dioxygenases (Fig. 4).
The pob and pca genes
The pca genes that encode enzymes of the protocatechuate branch of the β-ketoadipate pathway in C. necator JMP134 are organized in two clusters: pobAR-pcaLBGHQK and pcaIJF (C2 in Fig. 2), unlike gene organizations described previously. The pcaIJF genes encode the enzymes required for the conversion of β-ketoadipate to the Krebs cycle's intermediates; these catabolic steps are common to both branches of the β-ketoadipate pathway (Fig. 1, Table 1). In several Proteobacteria, β-ketoadipate is the inducer of pcaIJF genes by activation of PcaR/PcaQ, transcriptional regulators of the IclR-type family (Harwood & Parales, 1996). However, pcaR/pcaQ genes are not found in the vicinity of the pcaIJF genes in C. necator JMP134. A gene encoding a putative LysR-type regulator homologous to the pcaR gene of Pseudomonas strains (about a 40% aa identity) is located 9 kb away from the pcaK gene in C. necator JMP134, and could be the regulatory gene involved in pcaIJF gene induction. It is not uncommon that transcriptional regulators control the expression of distal genes. For example, PcaR gene products in P. putida are encoded at a distance from the target pcaHG genes (Harwood & Parales, 1996). Alternatively, the fact that in strain JMP134 the pcaIJF genes are not linked to the pca gene cluster would indicate that these gene functions are involved in other CoA transferase activities, such as those reported in the degradation of straight-chain dicarboxylic acids (Parke et al., 2001); therefore, they would be controlled by different regulatory proteins and/or additional inducers when nonaromatic substrates are metabolized through the activities of the PcaIJF gene products. The pca gene organization in C. necator JMP134 is different from that found in Ralstonia solanacearum (two clusters), P. putida (four clusters) and A. baylyi. ADP1 (one supraoperonic cluster). There exists, then, a great diversity in the genetic organization of this pathway in Proteobacteria (Harwood & Parales, 1996). Furthermore, gene order does not appear to be maintained within the clusters, except in cases where genes may coevolve because they encode subunits of a single enzyme and are cotranscribed; e.g., pcaGH genes encoding protocatechuate dioxygenase and pcaIJ genes encoding β-ketoadipate succinyl-CoA transferase (Harwood & Parales, 1996). A striking aspect of the pca genes in C. necator JMP134 is that pcaC and pcaD genes – encoding γ-carboxymuconolactone decarboxylase and β-ketoadipate enol-lactone hydrolase, respectively, two enzymes that perform successive steps in the protocatechuate degradation branch (Fig. 1) – are fused in a unique pcaL gene (C2 in Fig. 2). Sequence analysis of pcaL gene reveals that the N-terminal two thirds of the protein are homologous to the enol-lactone hydrolases, whereas the C-terminal third is homologous to the decarboxylases (Table 1). Such a gene fusion is not observed in R. solanacearum, P. putida and Bradyrhizobium japonicum, where pcaD and pcaC genes are located together (Lorite et al., 1998; Jimenez et al., 2002; Salanoubat et al., 2002), in what seems to be a previous stage of the gene fusion in C. necator JMP134. A similar gene fusion of pcaD and pcaC genes has been described previously in Rhodococcus opacus 1CP (Eulberg et al., 1998), in Streptomyces sp. strain 2065 (Iwagami et al., 2000) and, very recently, in Acinetobacter baumannii DU202 (Park et al., 2006). Gene databases also show the presence of similar gene fusions in unrelated bacteria such as Caulobacter, Nocardioides and Mycobacterium (data not shown). Sequence comparison of pcaL genes indicate that the gene fusions in C. necator and R. opacus took place separately, and are not due to a horizontal gene transfer from Gram-positive bacteria to C. necator, because each catalytic domain in the fused PcaL gene product of C. necator has a much higher identity with the proteobacterial PcaC and PcaD counterparts than with the PcaL gene product from R. opacus (Table 1). The fact that these gene fusions are present in distantly related bacterial groups strongly suggests a biochemical advantage of these fused gene products.
Another striking aspect of the protocatechuate branch genes in C. necator JMP134 is the presence of two genes, pobA and pobB that putatively encode p-hydroxybenzoate hydroxylases. The position of both gene products in the dendrogram of FAD-dependent hydroxylases is shown in Fig. 5. The pobA and pobB genes in C. necator JMP134 have a 59% aa identity, indicating a rather far evolutionary origin. However, both gene products contain the sequence Gly–X–Gly–X–X–Gly (residues 9–14), which is characteristic of flavoproteins. These glycine residues have been claimed to play an important structural role (Hofsteenge et al., 1980). The first case in which two genes encode a p-hydroxybenzoate hydroxylase has been reported in Pseudomonas fluorescens, in which an isoenzyme gene was cloned and showed to express half of the total p-hydroxybenzoate hydroxylase activity (Shuman & Dix, 1993). However, this isoenzyme was not expressed during growth of P. fluorescens on 4-hydroxybenzoate. It could be of interest to investigate whether pobA and pobB genes are differential or simultaneously expressed in C. necator JMP134.
Two putative regulatory proteins are encoded in the pobAR-pcaLBGHQK gene cluster: (1) PobR, a XylS/AraC family regulator that might activate the expression of pobA gene in response to 4-hydroxybenzoate, as described in Azotobacter chroococcum ATCC 9043 (Quinn et al., 2001) and in P. putida WCS358 (Bertani et al., 2001); (2) pcaQ (C2 in Fig. 2), a LysR-type regulator homologous to the PcaQ regulator from Agrobacterium tumefaciens (Parke, 1996), that might control the expression of the pcaHGBL genes required to transform protocatechuate into β-ketoadipate (Fig. 1).
Peripheral pathways that channel phenylpropenoid and phenylpropanoid compounds to the β-ketoadipate pathway
Phenylpropenoid compounds are structural components of plant polymers, such as lignin and suberin, and constitute a common carbon source for plant-associated microorganisms. The ability to grow on phenylpropenoid compounds is widely distributed in bacteria. Among phenylpropenoid compounds, the largest group corresponds to hydroxycinnamates (i.e. ferulate, coumarate, caffeate and others). The hca genes, encompassing the hcaABCDEFG and hcaKR gene clusters, are responsible for the degradation of coumarate, caffeate and ferulate in A. baylyi ADP1 (Smith et al., 2003). The hcaG gene encodes a chlorogenate esterase that hydrolyzes the ester bond of chlorogenate, an abundant hydroxycinnamic compound, further producing quinate and caffeate (Fig. 1). Enzymes with a relatively broad substrate specificity – encoded by the hcaABC genes – carry out key steps in the dissimilation of coumarate, caffeate and ferulate (Fig. 1). HcaC, a hydroxycinnamoyl-CoA ligase, activates hydroxycinnamates to their thioester derivatives; HcaA, a bifunctional hydratase/lyase, converts the thioester derivative into an aldehyde intermediate; and HcaB, an aldehyde dehydrogenase, transforms the aldehyde to 4-hydroxybenzoate, protocatechuate and vanillate, respectively. The latter substrates are further degraded through the protocatechuate branch of the β-ketoadipate pathway (Fig. 1). The hcaRGKXABC gene cluster identified in C. necator (C2 in Fig. 2), has a relatively high identity to the Acinetobacter counterparts (Table 1). In addition to the hcaGABC genes, hcaK, a gene that encodes a putative transporter for hydroxycinnamate compounds, and hcaX, a gene that encodes a putative porin of unknown function, were also found in the hca cluster of C. necator. A putative repressor-encoding gene, hcaR, is located divergently from the hca operon. The hcaR gene is homologous to the A. baylyi ADP1 hcaR gene and related to the MarR-like family of transcriptional repressors (Table 1). By analogy with the hca gene cluster of A. baylyi ADP1 (Parke & Ornston, 2003), the inducers of the expression of hcaGKXABC genes in C. necator JMP134 may be the hydroxycinnamoyl-CoA thioesters.
It should be noted that ferulate allowed the growth of C. necator, although with a very low yield, but vanillate, an intermediate in the ferulate dissimilation pathway encoded by the hca genes, is not a growth substrate. This may be explained by the absence of demethylases in C. necator. In fact, none of the tested methylated compounds (vanillate, isovanillate, vanillin, vanillyl alcohol, isovanillyl alcohol, 2-, 3-, and 4-methoxybenzoates and 5-methoxysalicylate) allowed the growth of C. necator. In Pseudomonas and Acinetobacter species, a vanillate demethylase encoded by vanAB gene (Priefert et al., 1997; Segura et al., 1999) channels vanillate to protocatechuate. A genomic search for aromatic demethylase genes in C. necator renders only ORFs with a low identity with the vanAB genes of Pseudomonas and Acinetobacter strains. We hypothesize that C. necator JMP134 is unable to metabolize methoxylated aromatic compounds because it lacks the needed demethylase enzymes. The hca encoded functions convert ferulate to vanillate and acetyl-CoA. Acetyl-CoA formation would explain the weak growth and very low yield observed when ferulate is used as the growth substrate for C. necator. Phenylpropanoid compounds, i.e. saturated derivatives of hydroxycinnamates, such as 4-hydroxyphenylpropionate and 3,4-dihydroxyphenylpropionate serve as growth substrates for C. necator. In A. baylyi ADP1, it has been proposed that HcaD is a FAD-dependent acyl-CoA dehydrogenase which oxidizes the saturated propionyl-CoA side chain of the hydroxyphenylpropanoyl thioesters produced by HcaC (Smith et al., 2003), to form hydroxycinnamoyl-CoA thioesters; these are channeled by HcaA and HcaB gene products to the protocatechuate branch of the β-ketoadipate pathway (Fig. 1). In C. necator, a genomic search for homologues to the hcaD gene only renders ORFs with a very low identity with the A. baylyi ADP1 gene. It remains to be studied whether hydroxyphenylpropanoyl thioesters are transformed by a different acyl-CoA dehydrogenase in C. necator, to be further metabolized by the Hca gene products or are catabolized by a hca genes independent pathway.
Degradation of salicylate and 3-hydroxybenzoate: the gentisate pathway
In addition to 4-hydroxybenzoate, 2-hydroxybenzoate (salicylate) and 3-hydroxybenzoate also support the growth of C. necator JMP134. It has been well documented that, in Pseudomonas and Acinetobacter species, salicylate is oxidatively decarboxylated to produce catechol by salicylate-1-hydroxylase, a flavoprotein monooxygenase (Fig. 5) (You et al., 1991; Lee et al., 1996; Bosch et al., 1999; Jones et al., 2000), or by a three-component salicylate-1-hydroxylase in Sphingomonas species (Fig. 6) (Demaneche et al., 2004; Cho et al., 2005). In Burkholderia– a genus closely related to Cupriavidus– the conversion of salicylate into catechol has been also demonstrated (Hamzah & Al-Baharna, 1994); the hydroxylase has been purified (Ramsay et al., 1992) and the gene has been cloned, although not sequenced (Kim & Tu, 1989). The search in the genome of C. necator only showed two ORFs with low identity (c. a 30% aa identity) with salicylate 1-hydroxylase genes from gammaproteobacterial Pseudomonas and Acinetobacter strains. Furthermore, most salicylate-1-hydroxylases described so far are active with 4-chloro- and 5-chlorosalicylate which, in combination with a functional chlorocatechol ortho ring-cleavage pathway, such as that encoded in the plasmid pJP4 (see ‘Catabolic pathway for mono- and dichlorinated compounds: the tfd genes’), would allow degradation of chlorosalicylates. The fact that a salicylate 1-hydroxylase is absent in C. necator JMP134 is supported by the failure of the strain to grow on chlorosalicylates. An alternative route of salicylate degradation, via gentisate as an intermediate, is initiated by a salicylate-5-hydroxylase, as has been recently described in Ralstonia sp. U2 (Zhou et al., 2002). In contrast to salicylate-1-hydroxylases, salicylate-5-hydroxylase is a multicomponent enzyme consisting of an oxygenase, comprising the NagG and H subunits and an electron transport chain, comprising NagAa ferredoxin reductase, and NagAb ferredoxin. A putative salicylate-5-hydroxylase gene cluster comprising the genes hybRBCDA was identified (Table 2 and Fig. 2, C2) in the genome of strain JMP134. The putative LysR-type transcriptional regulator encoding hybR gene, and genes coding for the large (hybB) and small (hybC) subunits of the oxygenase as well as the ferredoxin encoding gene (hybD), show a significant similarity with the nagGHAb genes of Ralstonia sp. U2 and the hybBCD genes of Pseudomonas aeruginosa JB2 (Table 2). The hybBCD genes encode the second confirmed example of salicylate-5-hydroxylase activity carried out by a multicomponent oxygenase (Hickey et al., 2001). In contrast, the HybA gene product in strain JMP134 does not show any similarity to the NagAa protein from Ralstonia sp. U2 or to the HybA protein from P. aeruginosa JB2, which indicates a different evolutionary origin. However, the HybA gene product from strain JMP134 shows a similarity to MocF, the ferredoxin reductase of rhizopine demethylase in Rhizobium leguminosarum bv. viciae strain 1a (Bahar et al., 1998) and to AndAa, the ferredoxin reductase of a three-component anthranilate 1,2-dioxygenase in Burkholderia cepacia DBO1 (Chang et al., 2003). A further biochemical topic yet to be investigated is whether HybA and NagAa gene products are interchangeable in supplying the ferredoxin reductase function to the NagGHAb and HybBCD proteins, respectively.
Table 2. Genes encoding gentisate and homogentisate pathways and peripheral reactions
A second gene cluster, ReutB3775-B3778, with homology to salicylate-5-hydroxylase components has been found in C. necator (Fig. 6). However, its identity with nag genes of Ralstonia sp. U2 or with the hyb genes found in P. aeruginosa JB2 (Hickey et al., 2001), is much lower (c. 40–45% aa identity) than that of the hyb genes cluster found in strain JMP134, which ranges between a 56% and a 78% aa identity (Table 2). The ReutB3775-B3778 gene cluster in strain JMP134 also shows a significant similarity with the ant gene cluster, which encodes anthranilate 1,2-dioxygenase of B. cepacia DBO1 (c. 35–45% aa identity) and with the three-component salicylate-1-hydroxylases (c. 35–50% aa identity), described in Sphingobium sp. P2 (Pinyakong et al., 2003). The actual role of this gene cluster in strain JMP134 remains to be elucidated.
In Proteobacteria, the following processes for the dissimilation of 3-hydroxybenzoate have been described: (1) in Pseudomonas alcaligenes (Poh & Bayly, 1980), Delftia acidovorans (Harpel & Lipscomb, 1990), B. cepacia (Wang et al., 1987), Klebsiella pneumoniae (Jones & Cooper, 1990; Suarez et al., 1995) and Salmonella typhimurium (Goetz & Harmuth, 1992), 3-hydroxybenzoate is degraded through gentisate by the activity of a 3-hydroxybenzoate-6-hydroxylase (Fig. 7); (2) in Comamonas testosteroni (Michalover et al., 1973; Hiromoto et al., 2006) and Bacillus sp. (Mashetty et al., 1996), the same compound is degraded through protocatechuate by a 3-hydroxybenzoate-4-hydroxylase. No ORFs closely related to the 3-hydroxybenzoate-4-hydroxylase gene (mobA) from C. testosteroni KH122-3S were found in the genome of strain JMP134 (Fig. 5). On the other hand, in the genome of this strain, two ORFs (mhbM1 and mhbM2) were identified with a significant (85% aa) identity with the 3-hydroxybenzoate-6-hydroxylase gene sequences from P. alcaligenes NCIB 9867 (xlnD) (Gao et al., 2005) and from K. pneumoniae M5a1 (mhbM) (Liu et al., 2005) (Table 2). The genes that encode the 3-hydroxybenzoate-6-hydroxylase in strain JMP134 (mhbM1 and mhbM2) cluster in the same group as those encoding the salicylate-1-hydroxylases, but in different branches in the dendrogram of FAD-dependent monooxygenases (Fig. 5). Both ORFs are located in putative gene clusters that encode enzymes involved in gentisate degradation (Fig. 2, C2, Table 2), thus supporting the possibility that C. necator JMP134 degrades 3-hydroxybenzoate through the gentisate pathway, as was suggested in an early report for C. necator 335 (Johnson & Stanier, 1971).
The gentisate pathway (Fig. 7) is initiated by a gentisate-1,2-dioxygenase, which cleaves the aromatic ring between the carboxyl and the vicinal hydroxyl group to form maleylpyruvate. The latter compound can be converted into central metabolites of the Krebs cycle either by cleavage to pyruvate and maleate – performed by a maleylpyruvate hydrolase in a glutathione-independent way (Crawford & Frick, 1977; Bayly et al., 1980) – or by isomerization to fumarylpyruvate, via a glutathione-dependent maleylpyruvate isomerase (Zhou et al., 2001) (Fig. 7). In strain JMP134, both mhb gene clusters (Fig. 2, Table 2) contain putative mhbI and mhbH genes that are homologous to genes encoding maleylpyruvate isomerase (nagL) and fumarylpyruvate hydrolase (nagK) in Ralstonia sp. U2 (Zhou et al., 2001), respectively. This strongly suggests that gentisate is metabolized by a glutathione-dependent pathway in C. necator JMP134. The reason for the presence of two putative gene clusters encoding the enzymes required for the degradation of 3-hydroxybenzoate is not clear. The identity between homologous gene products from both clusters is very high (from a 78% aa identity in MhbI1/MhbI2 gene products to a 95% aa identity in MhbH1/MhbH2 gene products) and the gene order, mhbDHIM, is similar in both clusters (Fig. 2). However, the genomic context for each cluster is different. The first mhb gene cluster is located downstream of a LysR-type regulator gene, mhbR, and includes a mhbT gene that encodes a putative 3-hydroxybenzoate transporter (Fig. 2). The second cluster (Fig. 2) is associated at both ends with genes related to ATP-binding cassette (ABC)-type transporters, and is close to genes putatively encoding a FAD-dependent hydroxylase – moderately related to 3-hydroxyphenylpropionate hydroxylase – and an extradiol dioxygenase. It should be noted that two sets of isofunctional enzymes for the gentisate pathway, including 3-hydroxybenzoate-6-hydroxylases, have been described in P. alcaligenes NCIB 9867; one set is constitutively expressed, whereas the other set is strictly inducible by gentisate (Poh & Bayly, 1980). A similar situation may take place in strain JMP134. The mhbR gene is located divergently from the first mhb gene cluster and could be responsible for the induction of this or of both mhb gene clusters. The second mhb gene cluster is not associated to a regulatory gene and could be constitutively expressed, as has been shown in P. alcaligenes NCIB 9867 (Poh & Bayly, 1980).
The catabolic pathways for benzene, toluene, and (methyl)phenols
Conversion of benzene, toluene and (methyl)phenols into catechol
Cupriavidus necator JMP134 is able to grow on toluene, benzene, phenol, 2-, 3- and 4-methylphenols, 4-ethylphenol, 2,3- and 3,4-dimethylphenol. Some of these catabolic abilities have been described previously and are in agreement with our results (Pieper et al., 1995; Lang, 1996). Benzene, toluene, ethylbenzene and xylenes, commonly referred to as BTEX, are important nonoxygenated aromatic pollutants often found as mixtures at contaminated sites where fuels, solvents, or chemicals that are confirmed or suspected carcinogens, even at very low concentrations, have been spilled (Dean, 1985).
The utilization profile of methylphenols (2-, 3-, 4-methylphenol, 2,3- and 3,4-dimethylphenol) in C. necator JMP134 is similar to that of strains harboring a multicomponent phenol hydroxylase coupled with a catechol meta ring-cleavage pathway (see next section), as has been described for Pseudomonas sp. CF600 (Shingler et al., 1989). The (dimethyl)phenol hydroxylases catalyze the conversion into the corresponding catechols (Shingler et al., 1989). Multicomponent phenol hydroxylases belong to an evolutionary related family of soluble diiron hydroxylases, including enzymes involved in monooxygenation of inactive compounds such as methane and toluene, that lack an electron donating hydroxyl group (Leahy et al., 2003). The enzyme complexes consist of: an electron transport system comprising a reductase (and in some cases a ferredoxin); a catalytic effector protein containing neither organic cofactors nor metal ions, that is assumed to assemble an active oxygenase; a terminal hydroxylase with a (αβγ)2 quaternary structure; and a diiron center contained in each α-subunit (Leahy et al., 2003). Recently, these monooxygenases have been classified, according to their α-subunits, into four phylogenetic groups (Leahy et al., 2003): the soluble methane monooxygenases, the alkene monooxygenase of Rhodococcus corallinus B-276, the phenol hydroxylases and the four-component alkene/aromatic monooxygenases. Remarkably, two phenol hydroxylase encoding genes are found in the genome of C. necator JMP134. Genes that encode one phenol hydroxylase (phl1) (C1 in Fig. 3, Table 3), are associated with the catA2 gene, which encodes catechol-1,2-dioxygenase (see ‘The cat and ben genes’). This organization is highly similar to that of the mop genes operon of Acinetobacter calcoaceticus NCIB8250 (Ehrt et al., 1995). Therefore, it can be suggested that the catechol produced by the phl1 phenol hydroxylase gene cluster is metabolized through the ortho ring-cleavage pathway (Fig. 1), as has been described for phenol degradation in A. calcoaceticus NCIB8250. A second phenol hydroxylase gene cluster, phl2, is encoded by genes located downstream of the genes that encode a catechol meta ring-cleavage pathway (Table 3, C2 in Fig. 2), which also suggests a functional association. A growth rate-dependent expression of phenol assimilation pathways has been reported in C. necator JMP134 growing in continuous culture (Müller & Babel, 1996). At low growth rates, the ortho ring-cleavage pathway is almost exclusively expressed, but at high growth rates both ring-cleavage pathways are equally expressed. Induction of catechol-1,2-dioxygenase and catechol-2,3-dioxygenase is also detected in phenol-grown cells in batch cultures (Pieper et al., 1989; Kim & Harker, 1997); this indicates that C. necator JMP134 uses both phenol hydroxylases and catechol ring-cleavage pathways to grow on phenol, and perhaps on some methylphenols (Fig. 3). Analysis of the phl2 phenol hydroxylase showed the highest identities (83–96% aa, depending of the subunit, Table 3), with phenol hydroxylases of B. cepacia JS150 and Ralstonia pickettii PKO1, whose genes are associated with those of the meta ring-cleavage pathway. The phl1 phenol hydroxylase is highly identical to the corresponding enzyme of Wautersia numadzuensis TE26 (Table 3), and has lower identities (45–75% aa, depending on the subunit) with phenol hydroxylases of Ralstonia strains E2 (Hino et al., 1998) and KN1, (Nakamura et al., 2000), of the Comamonas strains TA441 (Arai et al., 1998) and R5 (Teramoto et al., 1999), and of the alkylphenol-degrading strains Pseudomonas sp. KL28 (Jeong et al., 2003) and P. putida MT4 (Takeo et al., 2006). Remarkably, the phl1 phenol hydroxylase of strain JMP134, and that of W. numadzuensis TE26, are the only ones whose genes are associated with a catechol-1,2-dioxygenase gene (note that the catechol-1,2-dioxygenase gene of strain TE26 is wrongly classified as a catechol-2,3-dioxygenase in the Genbank accession AB177762).
Table 3. Genes encoding catabolic pathways for benzene, toluene, (methyl)phenols, 3-hydroxyphenylpropionate and 3-hydroxyanthranilate
Multicomponent phenol hydroxylases display a significant oxidizing activity with trichloroethylene and have attracted considerable attention for their potential applications in bioremediation of this important pollutant. Also, the strain JMP134 exhibits significant trichloroethylene-oxidizing activity (Kim et al., 1996). It has been found that trichloroethylene degradation by phenol-degrading bacteria can be classified into three distinct kinetic groups: low-Ks (the half-saturation constant in Haldane's equation), moderate-Ks and high-Ks, which strictly correspond to three phylogenetic groups that are defined according to the phenol hydroxylase α-subunit sequence alignment (Futamata et al., 2001a). Consequently, both phenol hydroxylase α-subunits (PhlN) from C. necator JMP134 (Fig. 8a) fall into the group with a low Ks constant for trichloroethylene, which is considered the most useful group for an effective bioremediation of trichloroethylene-contaminated groundwater (Futamata et al., 2001b).
Another phenol hydroxylase gene cluster associated with a catechol-1,2-dioxygenase gene has been described in A. calcoaceticus NCIB8250 (Ehrt et al., 1995). However, the amino acid identities of the phenol hydroxylase multicomponent of C. necator JMP134 with the phenol hydroxylase of A. calcoaceticus are lower than the identities with the meta ring-cleavage-associated phenol hydroxylases indicated above. This suggests that the clustering of phenol hydroxylase operons with catechol-1,2-dioxygenase genes occurred independently and after the divergence of both phenol hydroxylases. In addition, the amino acid identity between the catechol-1,2-dioxygenases from both clusters is not particularly high (54%), as compared with the amino acid identities (c. 70%) of the CatA2 gene product and the catechol-1,2-dioxygenases that are not directly related to phenol hydroxylase gene clusters (Fig. 4).
Based on the overall sequence identity and gene order, it is reasonable to propose that PhlN1/PhlN2 (Fig. 8a), PhlL1/PhlL2, and PhlO1/PhlO2 gene products correspond to the α, β, and γ subunits of the putative (αβγ)2 hexameric nonheme diiron monooxygenases, respectively. Both putative α subunits have the highly conserved motif Asp–Glu–X–Arg–His, which occurs twice in the PhlN1/PhlN2 gene products at positions 148/139 and 243/234, respectively. These ligands form the dinuclear iron-binding site in the large subunits of this family of monooxygenases (Fox et al., 1988). In addition, the spacing of 94 aa between these motifs is conserved in both PhlN gene products.
PhlP1/PhlP2 gene products show identity with a very large group of iron–sulfur flavoproteins that transfer electrons from reduced pyridine nucleotides to a terminal electron acceptor via a flavin and [2Fe–S] center (Mason & Cammack, 1992). PhlM1/PhlM2 gene products are homologous to the small polypeptides that are believed to play a role in regulating monooxygenase activity, namely the effector proteins. The only member of this group of proteins that has been studied in detail is the DmpM protein of Pseudomonas sp. CF600 (Qian et al., 1997). DmpM protein binds to the DmpNLO hydroxylase, but does not participate directly in redox reactions. Rather, its role seems to be to increase the steady state turnover rates and the product yields from the phenol hydroxylase, possibly by controlling the entrance of substrate and exit of products (Qian et al., 1997). PhlM1/PhlM2 gene products from C. necator JMP134 possess the conserved amino acids Glu56/Glu54 and Gly59/Gly57, respectively, which were identified as conserved components of the functionally important helix 2 of the DmpM protein (Qian et al., 1997). However, Leu56 of the DmpM protein is replaced by Met58 in the PhlM1 gene product and by Thr56 in the PhlM2 gene product, which indicates that the conservation of this residue is not required.
The PhlK1/PhlK2 gene products are unique, because they share significant similarity only with a small group of polypeptides that are exclusively found as components of phenol hydroxylases, and not as components of other multicomponent diiron monooxygenases. The role of the PhlK1/PhlK2 gene products can be inferred from studies on the DmpK protein from Pseudomonas sp. CF600 (Powlowski et al., 1997). The DmpK protein binds to both DmpN and DmpL proteins and plays an essential role in the assembly of the active form of the oxygenase – possibly by posttranslational insertion of iron into the subunit – because the DmpK protein can catalyze in vitro reactivation of the inactive enzyme in the presence of iron.
A multicomponent monooxygenase that belongs to an alkene/aromatic monooxygenase subfamily and putatively encodes a toluene/benzene/xylene monooxygenase (tbc genes, C2 in Fig. 2) is located divergently from the phl2 gene cluster in C. necator. The analysis of the deduced protein products of this tbcABCDEF gene cluster revealed a strong similarity to the nearly identical tbc2 gene cluster of Burkholderia sp. JS150 and to the tbu gene cluster of R. pickettii PKO1 (76–95% aa, Table 3). The TbcA/TbuA1/Tbc2A gene products cluster in a separate branch in the dendrogram of α subunits of the toluene/benzene monooxygenases (Fig. 8c), and have been involved in the hydroxylation of inactivated aromatic compounds like benzene, toluene and o-xylene. Three other strains have a similar arrangement of phenol hydroxylase and toluene/benzene/xylene monooxygenase encoding genes: Burkholderia sp. JS150 (Kahng et al., 2001), R. pickettii PKO1 (Byrne et al., 1995), and Pseudomonas stutzeri OX1 (Bertoni et al., 1998). In the latter strain, as has been recently proposed, the coupling of the two enzymatic systems optimizes the use of nonhydroxylated aromatic molecules by the draining effect of the phenol hydroxylase on the product(s) of the oxidation catalyzed by toluene monooxygenase; this avoids phenol accumulation (Cafaro et al., 2004).
The toluene/benzene monooxygenase gene cluster of strain JMP134 has been cloned previously, partially sequenced, and expressed (Kim et al., 1996). The plasmid pYK3021, which contains the cloned toluene/benzene monooxygenase gene cluster, exhibited phenol hydroxylase activity. However, it is not possible to establish if this activity corresponds to the toluene/benzene monooxygenase [as has been shown for the toluene/o-xylene monooxygenase of P. stutzeri OX1 (Bertoni et al., 1998)] or to the presence of the phl2 gene cluster that encodes the meta ring-cleavage pathway associated phenol hydroxylase, in the same cloned 9.1-kb fragment. This is why the genes encoding the toluene/benzene/xylene monooxygenase in strain JMP134 were originally labeled as phl (GenBank accession AF065891) (Kim et al., 1996). Given the genomic information presented here and the high identity with the toluene/benzene/xylene monooxygenases from Burkholderia sp. JS150 and P. stutzeri OX1 (Table 3), we consider the phl denomination incorrect, and have changed it to tbc genes, leaving the former phl denomination for the phenol hydroxylase gene clusters of C. necator JMP134 (Table 3).
The enzymes of the toluene/benzene/xylene monooxygenase family consist of four dissociable components, three of which constitute a short electron transfer chain with an oxidoreductase, a ferredoxin, and a terminal hydroxylase. Based on the overall sequence and gene order similarity, it is reasonable to consider the TbcF gene product as the oxidoreductase, the TbcC gene product as the ferredoxin, and the TbcAEB gene products as the terminal hydroxylase, in the electron transfer chain of the Tbc monooxygenase (Table 3). The TbcF gene product from strain JMP134 is similar to the proteins that comprise the very large family of iron–sulfur flavoproteins; these function as oxidoreductases for most mono- and dioxygenase systems. Supporting this assumption is the presence, in the TbcF gene product, of the conserved Cys (Cys37, -42, -45 and -77) and Gly (Gly40 and -52) residues that are required for the coordination of the two iron atoms of the [2Fe–2S] cluster (Mason & Cammack, 1992). The TbcC gene product is highly homologous to the ferredoxin component of other toluene/benzene/xylene monooxygenases. This group of ferredoxins is related to a large family of Rieske-type ferredoxins that function as soluble electron carriers for a variety of bacterial oxygenases. The TbcC gene product contains the metal-binding motif Cys–X–His–X15–21–Cys–X2–His at positions 44–66, which is characteristic of all Rieske-type proteins (Mason & Cammack, 1992). A high degree of similarity with the components of the terminal hydroxylases of multicomponent monooxygenases suggests that the TbcAEB gene products correspond to the (αβγ)2 dimeric nonheme iron monooxygenase. In agreement with this, in the TbcA gene product, the putative α-subunit of the monooxygenase, there are two copies of the motif (Asp/Glu)–X26–30–Asp–Glu–X–Arg–His at positions 104–137 and 197–234; these are the ligands for the diiron center in the active sites of the enzymes of this family (Fox et al., 1994).
The TbcD gene product appears homologous to the small polypeptides that might play a role in regulating the monooxygenase activity. Among the toluene/benzene/xylene monooxygenases, the TmoD protein from Pseudomonas mendocina KR1 has been studied in detail (Pikus et al., 1996). The TmoD protein, which appears to be a substoichiometric constituent of the TmoAEB hydroxylase, can mildly stimulate the rate of toluene hydroxylation when added to purified hydroxylase. Moreover, the TbcD gene product from strain JMP134 has, at positions Glu67, Leu70 and Gly71, the amino acids identified as conserved components of the functionally important helix 2 of the DmpM protein, which provides a similar regulatory function for the phenol hydroxylase of Pseudomonas sp. CF600 (Qian et al., 1997). Therefore, it is reasonable to conclude that the TbcD gene product might function as a regulatory component of the Tbc monooxygenase complex.
The regulation of the expression of the putative phenol hydroxylases and the toluene/benzene/xylene monooxygenase of strain JMP134 seems to be complex. The ortho ring-cleavage pathway associated with the Phl1 phenol hydroxylase has a regulatory gene encoded divergently from the putative phl1 genes operon (C1 in Fig. 2). This is the classical organization in most multicomponent phenol hydroxylases. The product of this gene, PhlR1, belongs to the aromatic-responsive σ54-dependent family of regulators that includes the well-characterized DmpR transcriptional activator involved in the regulation of phenol degradation in Pseudomonas sp. CF600 (Shingler et al., 1993; Shingler & Moore, 1994). On the other hand, three putative regulatory genes are located next to the tbc genes cluster: tbcR, phlR2 and phlS (C2 in Fig. 2). The tbcR and phlR2 genes encode regulators that also belong to the aromatic-responsive σ54-dependent family. Both PhlR1 and PhlR2 gene products have the highest identity with the PoxR and AphR proteins, which are positive regulators of phenol degradative operons from W. numadzuensis TE26, and C. testosteroni TA441, respectively (Table 3). These regulators are responsive to phenol or alkylphenols, because members of this family exhibit broad effector-response profiles (Shingler & Moore, 1994). On the other hand, the TbcR gene product, putatively involved in the regulation of the tbc gene cluster, has a high amino acid identity with the TbuT protein (88%); TbuT is the activator of the toluene monooxygenase tbu genes operon of R. pickettii PKO1, which is activated not only by aromatic effectors as benzene, toluene or ethylbenzene, but also by trichloroethylene (Kahng et al., 2000). The PhlS gene product, the third putative regulator encoded in the region of the phl2 and tbc gene clusters in C. necator, has significant identity with regulators belonging to the GntR family of transcriptional repressors, like the aphS or phcR genes from C. testosteroni strains TA441 (Arai et al., 1999b) and R5 (Teramoto et al., 2001), respectively. In the absence of the genuine substrate, these regulators repressed the gratuitous expression of phenol-metabolizing enzymes. The presence of several regulators in C. necator, putatively involved in phenol degradation, suggests a complex regulatory system that comprises cross-regulation, regulatory cascades, competition for binding sites and regulatory hierarchy. Finally, between the tbcR and phlR2 genes, it was possible to identify the tbcX gene (Table 3), whose gene product is almost identical (99% in amino acid) with the TbuX protein from R. pickettii PKO1 (Kahng et al., 2000), an outer membrane protein that plays an important role in the catabolism of toluene.
The catechol meta ring-cleavage pathway
The biochemical route of the meta ring-cleavage pathway for the degradation of catechol produced by phenol hydroxylation is illustrated in Fig. 3. The putative functions of the phl genes encoded enzymes are shown in Table 3. The critical ring-opening step of the meta ring-cleavage pathway is typically catalyzed by type I catechol-2,3-dioxygenase enzymes, the so called vicinal oxygen chelate family enzymes (Vaillancourt et al., 2004), that usually contain nonheme Fe+2 at the active site. Inactivation during turnover of para-substituted catechols (Cerdan et al., 1995) and 3-chlorocatechol has been described for members of this family. The phlB gene-encoded enzyme shares the highest identity with catechol-2,3-dioxygenases of the subfamily 1.2.C of extradiol dioxygenases (Fig. 9a) in the classification system proposed by Eltis and Bolin (Eltis & Bolin, 1996). Two genes, named mpcI and mpcII– that encode enzymes with activity against catechol derivatives – had been cloned previously from C. necator JMP222, a pJP4 cured derivative of C. necator JMP134 (Kabisch & Fortnagel, 1990a, b). The Km values for catechol of these two enzymes are in the millimolar range; this is unexpected for catechol-2,3-dioxygenases and indicates that catechol is not the native substrate. None of these two mpc genes correspond to the phlB gene, the catechol-2,3-dioxygenase encoding gene involved in phenol degradation. The genome analysis established that the mpcI gene should be designated mhpB (Fig. 9b), given that the gene product does not belong to the vicinal oxygen chelate superfamily, but shows a high similarity to 2,3-dihydroxyphenylpropionate-1,2-dioxygenases (Bugg, 1993) (see ‘The 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway), which makes it part of the type II extradiol dioxygenases that exhibit only poor activity with catechol (Spence et al., 1996).
The mpcII (ReutB4677 mhpB) gene – together with two other extradiol dioxygenase encoding genes found in the genome of C. necator JMP134 (ReutA1133, and ReutC6234, Fig. 9c) – is related to the bphCII gene of Sphingomonas xenophaga BN6 and to the bphC2 and bphC3 genes of Rhodococcus globerulus P6, which encode a new dimeric type of extradiol dioxygenase with 2,3-dihydroxybiphenyl dioxygenase activity (Heiss et al., 1997). The important feature of this group of enzymes is their small subunit molecular weight; only half as much as other types of extradiol dioxygenases. Phylogenetic analyzes indicated that the ancestral type I extradiol dioxygenase was, like the small size subunit enzymes, a one-domain enzyme, and that two-domain enzymes arose from a single duplication event (Eltis & Bolin, 1996). Unfortunately, no physiological function has been assigned to the BphCII gene product in the naphthalenesulfonate-degrading strain BN6 that clarify the role of this mpcII-like class of extradiol dioxygenases (Fig. 9c). In R. globerulus P6, the BphC2 gene product was constitutively expressed, and supposedly supported the degradation of PCBs (McKay et al., 2003).
A new gene, mhqB, related to mpcII-like dioxygenases was recently described in Burkholderia sp. NF100 and putatively considered to encode a extradiol dioxygenase involved in methylhydroquinone catabolism. Such protein showed an extradiol ring-cleavage activity toward 3-methylcatechol and a lower activity with catechol (Tago et al., 2005), suggesting possible physiological substrates of these enzymes in C. necator JMP134. Several mutagenesis attempts to obtain catechol-2,3-dioxygenase mutants of C. necator JMP134, which should show impaired growth on methylphenols, have been unsuccessful (D. Pieper, unpublished data); this suggests that if the phlB gene is inactivated, redundant functions supply the catechol-2,3-dioxygenase activity. The mpcII-like group of extradiol dioxygenases is the most likely candidate to undertake this function, as these enzymes usually exhibit some activity against 3-methylcatechol and very poor activity with catechol (Asturias & Timmis, 1993; McKay et al., 2003).
In the meta ring-cleavage pathway operon of strain JMP134 (C2 in Fig. 2), the phlB gene is preceded by the phlQ gene that encodes a putative ferredoxin with a high identity with the cbzT gene product from P. putida GJ31; this strain has a chlorocatechol-2,3-dioxygenase, the CbzE protein, which is exceptionally resistant to inactivation by 3-chlorocatechol (Kaschabek et al., 1998). These small auxiliary ferredoxin proteins – whose genes are frequently encoded adjacently to the catechol-2,3-dioxygenase genes in the meta-pathway operons – have a reactivating function, as has also been shown for the XylT protein in toluene catabolism, for the NahT protein in naphthalene catabolism and for the PhhQ and DmpQ proteins in methylphenol catabolism (Hugo et al., 2000). Similarly, the CbzT protein is able to reactivate the CbzE protein in vitro, through reduction of the iron atom, when the enzyme had been fully inactivated by 4-methylcatechol (Tropel et al., 2002). Loss of the activity of the DmpQ or XylT proteins, respectively, results in strains that are unable to grow on compounds that are catabolized through p-substituted methylcatechols (i.e. 4-methyl and 3,4-dimethylcatechol) (Polissi & Harayama, 1993; Powlowski & Shingler, 1994). However, these strains can still grow on compounds that are catabolized through catechol or 3-methylcatechol. Therefore, it can be proposed that the ability of strain JMP134 to metabolize 4-methylphenol and 3,4-dimethylphenol might depend on the regeneration of an active catechol-2,3-dioxygenase through the activity of the PhlQ gene product.
An ORF of unknown function, phlX, which encodes a relatively hydrophobic protein, is located downstream of the phlB gene. Similar genes have been found in other gene clusters encoding the meta ring-cleavage pathway, such as the phnX gene of Burkholderia sp. RP007, the orfY gene of C. testosteroni TA441, the cbzX gene of P. putida GJ31 and the nahX gene of plasmid NAH7 from P. putida G7 (Grimm & Harwood, 1999; Laurie & Lloyd-Jones, 1999; Mars et al., 1999; Arai et al., 2000). The role of these phlX-like genes has not been determined, but an orfY mutant of C. testosteroni TA441 grew poorly on phenol and accumulated the yellow 2-hydroxymuconic semialdehyde intermediate in the medium; this suggests that it could be involved in additional catabolism of 2-hydroxymuconic semialdehyde (Arai et al., 2000). The next two genes in the meta ring-cleavage pathway operon, phlC and phlD, encode 2-hydroxymuconic semialdehyde dehydrogenase (HMSD) and hydrolase (HMSH) and show a high identity with homologous gene products of the meta ring-cleavage pathway operon in Pseudomonas sp. CF600 (Table 3). Both enzymes use ring-cleavage products as substrates: 2-hydroxymuconic semialdehyde from catechol, 5-methyl-2-hydroxymuconic semialdehyde from 4-methylcatechol and 2-hydroxy-6-oxo-2,4-heptadienoate from 3-methylcatechol. However, 2-hydroxymuconic semialdehyde is only a poor substrate for HMSH, and it is preferentially degraded via the oxalocrotonate branch of the meta ring-cleavage pathway (Harayama et al., 1987). Because the ring-cleavage product of 3-methylcatechol is a ketone, rather than an aldehyde, it cannot be further oxidized by the HMSD and must be metabolized via the hydrolytic route (Fig. 3). Correspondingly, it has been shown that mutants of the HMSH encoding gene (dmpD) in Pseudomonas sp. CF600 still grew on phenol or 4-methylphenol but failed to grow on phenols that are channeled through 3-methylsubstituted catechols (Powlowski & Shingler, 1994). On the other hand, Pseudomonas sp. CF600 with a deletion in the HMSD gene (dmpC) or in either of the genes encoding the other two enzymes of the 4-oxalocrotonate branch (the dmpI and dmpH genes) resulted in strains that grew on 2-methyl-, 3-methyl- and 3,4-dimethylphenol, but not on phenol or 4-methylphenol. These results indicate that despite the potential use of either branch for the metabolism of the ring-cleavage products of catechol and 4-methylcatechol, these compounds are preferentially metabolized by the HMSD rather than by the hydrolase of the meta ring-cleavage pathway (Powlowski & Shingler, 1994). The same situation may take place with the different phenols that are growth substrates for C. necator. The last three genes identified in the meta ring-cleavage pathway operon in strain JMP134 are phlE, phlH and phlI; these encode: (1) 4-oxalocrotonate isomerase (4OI), which catalyzes the isomerization of the enol form of 4-oxalocrotonate to its keto-form; (2) 4-oxalocrotonate decarboxylase (4OD), which catalyzes the formation of 2-oxopent-4-dienoate, the common intermediate of the hydrolytic branch and the 4-oxoalocrotonate branch of the meta ring-cleavage pathway, and (3) 2-oxopent-4-dienoate hydratase (Fig. 3). All these genes in strain JMP134 show the highest identities with the homologous gene products of C. testosteroni TA441 (Arai et al., 2000). Gene order in the phl genes of the meta ring-cleavage pathway in C. necator JMP134 is clearly different from any other reported phl gene cluster. Moreover, it is the only example in which genes encoding the 4-hydroxy-2-ketovalerate aldolase (HOA, the phlG gene) and the aldehyde dehydrogenase (acylating, the phlF gene) – that catalyze the final steps of the meta ring-cleavage pathway to generate the end-products pyruvate and acetyl-CoA, respectively (Powlowski et al., 1993) – are separated from the rest of the other meta ring-cleavage pathway genes (C2 in Fig. 2). These genes show a high identity with the xylQ and xylK genes, respectively, from the pWW0 plasmid of P. putida, where they are encoded with the rest of the xyl genes (Assinder & Williams, 1990; Aemprapa & Williams, 1998).
The methylcatechol ortho ring-cleavage pathway
The degradation of methyl-substituted catechols as intermediates in the degradation of methylaromatics usually proceeds through a catechol meta ring-cleavage pathway. The metabolism of methylcatechols via the ortho ring-cleavage pathway results in the formation of methyl-substituted 4-carboxymethylbut-2-en-4-olides (methylmuconolactones) as dead-end products (Catelani et al., 1971; Knackmuss et al., 1976). In the transformation of 4-methylcatechol, 4-methylmuconolactone (4-ML) is formed, which cannot be processed by enzymes of the β-ketoadipate pathway as no proton is available to be abstracted by the muconolactone isomerase. In C. necator JMP134, however, a different ortho ring-cleavage pathway for the degradation of 4-methylcatechol has been described (Pieper et al., 1985) (Fig. 3). A key enzyme of this new pathway in C. necator JMP134 (Pieper et al., 1985, 1990) was initially characterized as a 4-ML methylisomerase capable of converting 4-ML to 3-methylmuconolactone (3-ML). This enzyme's function may be to compensate for the initial ‘incorrect’ cycloisomerization of 3-methylmuconate. 3-ML is further metabolized via 4-methyl-β-ketoadipate, and hence, probably, by analogous reactions to those of the classical β-ketoadipate pathway (Fig. 3).
A 3-kb mml gene cluster, harboring the gene that encodes 4-ML methylisomerase, was cloned and sequenced (Erb et al., 1998), and additional genes were identified (C1 in Fig. 2, Table 2). The first gene in the mml cluster, mmlH, encodes a putative transporter protein for 4-ML and exhibits a sequence homology to other members of the major superfamily of transmembrane facilitators, showing the common structural motif of 12 transmembrane-spanning α-helical segments, and the key amino acid motif that is characteristic of this superfamily (Erb et al., 1998). The second gene, mmlI, encodes the 4-ML methylisomerase and, given the novelty of the reaction, no sequence homologies were found. Finally, the mmlJ gene encodes a methylmuconolactone isomerase (Prucha et al., 1997) with significant identity to the muconolactone isomerases of Pseudomonas and Acinetobacter strains (Houghton et al., 1995) and of strain JMP134, which is involved in the degradation of catechol via the β-ketoadipate pathway. The methylmuconolactone isomerase encoded by the mmlJ gene is supposed to transform 3-ML produced from 4ML, by 4-ML methylisomerase, to 4-methyl-β-ketoadipate enol-lactone. Further metabolism of this intermediate and the possible formation of 4-methyl-β-ketoadipate are far from being understood.
By genomic analysis, it was possible to identify the genetic context of the mml genes in strain JMP134. Two genes, mmlFG, putatively encoding both subunits of a β-ketoadipate CoA transferase were found upstream to the mmlH gene (C1 Fig. 2, Table 1). Both genes show a high identity with the homologous genes (pcaIJ and catIJ genes) in the Pseudomonas and Acinetobacter strains (Shanley et al., 1994; Gobel et al., 2002), suggesting a possible role in the further metabolism of 4-methyl-β-ketoadipate enol-lactone. A putative LysR-type regulator encoding gene (mmlR) was found directly upstream of mmlFG, which indicates that mmlFG and mmlHIJ genes are part of one transcriptional unit (C2 in Fig. 2. Table 1). It should be noted that no gene that putatively encodes an isoenzyme of β-ketoadipate enol-lactone hydrolase was found in the mml gene cluster. This point supports the idea that 4-methyl-β-ketoadipate enol-lactone is not further metabolized through a classical β-ketoadipate pathway. Recently, a gene cluster was found on the megaplasmid pHG1 of C. necator H16 (Schwartz et al., 2003) with the same order of the mml genes in strain JMP134, and with amino acid identity levels of 80–90% (Table 1). This mml gene cluster also includes the genes that putatively encode the β-ketoadipate-CoA transferase and the LysR-type regulator.
Degradation of C6-C2 and C6-C3 compounds
The phenylacetyl-CoA ring-cleavage pathway
Although phenylacetate is a common source of carbon and energy for a wide variety of microorganisms, knowledge on the bacterial catabolism of this natural aromatic compound is still fragmentary, and details on the enzymatic mechanisms and the nature of intermediates are scarce. The general pathway for aerobic phenylacetate metabolism has initially been characterized in Gammaproteobacteria, Escherichia coli (Ferrandez et al., 1998), P. putida (Olivera et al., 1998) and the betaproteobacterium Azoarcus evansii (Mohamed Mel et al., 2002; Rost et al., 2002) (Fig. 10). This pathway does not follow the conventional route for the aerobic degradation of aromatics. In E. coli, there are 14 paa genes that encode for phenylacetate degradation, organized in three transcriptional units; two of them, paaZ and paaABCDEFGHIJK, encode the catabolic genes; the third, paaXY, contains the paaX regulatory gene (Ferrandez et al., 1998). In a study of paa mutants of E. coli K12, a phenylacetate catabolic pathway has been proposed (Ismail et al., 2003). The initial step of the pathway involves the activation of phenylacetate into phenylacetyl-CoA by a phenylacetate-coenzyme A ligase, encoded by the paaK gene (Ferrandez et al., 1998). The respective enzymes have also been identified in P. putida (Olivera et al., 1998) and A. evansii (El-Said Mohamed, 2000). The phenylacetyl-CoA is attacked by a ring-oxygenase/reductase (the PaaABCDE gene products), generating a hydroxylated and reduced derivative of phenylacetyl-CoA, which is not reoxidized to a dihydroxylated aromatic intermediate as in other known aromatic pathways (Fig. 10). It has been proposed that this nonaromatic intermediate CoA ester is further metabolized in a complex reaction sequence comprising enoyl-CoA isomerization/hydration, nonoxygenolytic ring opening and dehydrogenation, which is catalyzed by the PaaG and PaaZ gene products. The resulting aliphatic CoA dicarboxylate compound is further catabolized by a β-oxidation-like pathway via β-ketoadipyl-CoA into succinyl-CoA and acetyl-CoA, which appears to be catalyzed by the PaaF, PaaJ and PaaH gene products (Ismail et al., 2003).
A search in the genome of strain JMP134 showed 19 genes putatively involved in phenylacetate catabolism (Table 4), organized in three clusters (C1 and C2 in Fig. 2). Two genes were found, paaK1 and paaK2, which putatively encode phenylacetate CoA-ligase, both showing over a 70% aa identity with the PaaK gene product of A. evansii KB740 (Table 4). The amino acid sequence identity between both PaaK gene products is also over 70%, but the genetic context is completely different (Fig. 2). The analysis of the sequence of both paaK gene products revealed the presence of three conserved motifs for AMP and substrate binding in acyl-adenylate-forming enzymes (Ferrandez et al., 1998). The motif I comprises residues 97/96SSGTTGKPTV106/105 in the PaaK1/PaaK2 gene products, matching the AMP-binding site consensus sequence T(SG)-S(G)-G-(ST)-T(SE)-G(S)-X-P(M)-K-G(LAF) (predominant aa are underlined and alternatives are in parentheses) (Ferrandez et al., 1998). The sequences 239/238DIYGLSE245/244 and 305/304YRTRD309/308 in the PaaK1/PaaK2 gene products from C. necator JMP134, which are conserved in all putative or bona fide phenylacetate CoA-ligases, correspond to motifs II and III, respectively, and they may contribute to the substrate-binding sites (Ferrandez et al., 1998). It should be noted that two functional, almost identical copies of genes that encode phenylacetate-CoA ligases, each one located in a different genetic context, have also been reported in the styrene-degrading Pseudomonas sp. strain Y2 (Alonso et al., 2003); however, the physiological meaning of the existence of two phenylacetate-CoA ligases in this strain is not clear. Two gene clusters, paaA1B1C1D1E1 and paaA2B2C2D2E2, that putatively encode ring-oxygenase/reductase multicomponent proteins are also found in C. necator JMP134 (Table 4, and Fig. 2). The amino acid identity between the gene products of both gene clusters range from 45% to 65%. However, the PaaA1 and PaaA2 subunits of the multicomponent oxygenase map far apart in the corresponding dendrogram (Fig. 8b). Both gene clusters show the highest identities with different homologous clusters (Table 4); paaA1B1C1D1E1 is closer to the paaGHIJK gene cluster of Pseudomonas sp. strain Y2 (Alonso et al., 2003), whereas the paaA2B2C2D2E2 is closer to the paaABCDE (pacEFGHI) gene cluster of A. evansii KB740 (Mohamed Mel et al., 2002; Rost et al., 2002). It has been proposed that the paaABCDE genes encode a five-subunit oxygenase enzyme complex, using phenylacetyl-CoA as substrate (Ferrandez et al., 1998; Diaz et al., 2001; Mohamed Mel et al., 2002; Ismail et al., 2003; Fernández et al., 2006). PaaACD may function as a terminal oxygenase, with PaaA as the large α subunit containing the dinuclear iron-binding site (Ferrandez et al., 1998). The small protein PaaB may be the dissociable activator protein required for an optimal turnover of the oxygenase component (Ferrandez et al., 1998). Finally, the similarity of the PaaE protein to class IA-like reductases – members of the ferredoxin-NADP+ reductase family – indicate that it could function as a reductase delivering electrons from NAD(P)H to the terminal PaaACD oxygenase (Ferrandez et al., 1998). The phenylacetyl-CoA oxygenase constitutes the first reported multicomponent oxygenase acting on a CoA-activated aromatic compound (Fernández et al., 2006). Interestingly, although all bacterial multicomponent oxygenases described so far are monooxygenases, the product of the reaction catalyzed by the phenylacetyl-CoA oxygenase is a dihydrodiol, and therefore this enzyme could be a hydroxylating dioxygenase rather than a monooxygenase (Fernández et al., 2006). In C. necator JMP134, the presence of two gene clusters putatively encoding phenylacetyl-CoA multicomponent oxygenases has been reported. This is the first reported case for double dosage in these genes and its role is unclear. On the other hand, only one copy of the paaZ gene that putatively encodes the proposed ring-opening enzyme is found in the genome of strain JMP134 (C2 in Fig. 2). The PaaZ gene product from C. necator JMP134 shows only a 23% aa identity with the E. coli PaaZ. The E. coli protein has an N-terminal region (residues 1–527) that contains all the conserved motifs that characterize the aldehyde dehydrogenase superfamily, and a C-terminal domain with some similarity to enoyl-CoA hydratases (Diaz et al., 2001). Therefore, it has been proposed that the E. coli PaaZ is a fused, bifunctional protein with the enoyl-CoA hydratase-like C-terminal domain involved in the ring-cleavage of the phenylacetate intermediate, because enoyl-CoA hydratases have been linked with the ring-cleavage in the anaerobic benzoate degradation pathway (Diaz et al., 2001). The PaaZ gene product of C. necator JMP134 is a shorter polypeptide (554 aa) than the E. coli PaaZ protein (681 aa), and it lacks the enoyl-CoA hydratase-like C-terminal domain, which suggests that this paaZ gene is in a prefusion state. Close relatives of the C. necator JMP134 PaaZ gene product are the protein PaaZ (PacL) of A. evansii KB740 (Mohamed Mel et al., 2002; Rost et al., 2002) and the protein PaaN2 of Pseudomonas sp. Y2 (Alonso et al., 2003). In these bacteria, the ring-cleavage could not be undertaken by PaaZ, because the enoyl-CoA hydratase-like domain is also absent; therefore, PaaZ may participate in the conversion of the aldehyde, produced by the ring opening, into a carboxylic acid, as has been recently proposed in E. coli (Ismail et al., 2003). Given the similarity of the PaaG protein with some members of the enoyl-CoA hydratase/isomerase family, it has been proposed that, in E. coli, the ring opening may be preceded by a reversible PaaG-catalyzed Δ3, Δ2 isomerization of double bonds and/or by the addition of water in the putative cis-dihydrodiol derivative of phenylacetyl-CoA (Ismail et al., 2003). In addition, given that C–C-cleaving enoyl-CoA hydratases have been described, it has been proposed that PaaG may play a role in C–C cleavage (Ismail et al., 2003). Therefore, it is interesting to speculate that in C. necator JMP134 and A. evansii KB740 strains that possess an ‘aldehyde dehydrogenase only’ version of PaaZ, PaaG could be directly involved in the ring opening of the putative cis-dihydrodiol derivative of phenylacetyl-CoA (Fig. 10). The paaG gene of C. necator JMP134 is located close to the paaF gene, another putative enoyl-CoA hydratase encoding gene, paaI, that encodes a protein of unknown function and the paaK2 gene, that encodes a putative phenylacetyl-CoA ligase (Fig. 2). The amino acid identity with the E. coli paaG gene product is moderate (58%), and similar to the identity showed with the P. putida U paaB gene product (Table 4). In the mutational analysis of paa genes, performed on E. coli, in addition to paaZ and paaG, three additional genes proved to be essential for the utilization of phenylacetate as the carbon source: paaF, paaH and paaJ (Ismail et al., 2003), all of them putatively involved in the final steps in phenylacetate degradation (Fig. 10). The PaaH protein of E. coli is similar to 3-hydroxyacyl-CoA dehydrogenases, which suggests that it catalyzes the dehydrogenation of 3-hydroxyadipyl-CoA to produce β-ketoadipyl-CoA. On the other hand, the PaaF protein has a sequence similarity to proteins of the enoyl-CoA hydratase (isomerase) family (crotonase family), which may have cis-Δ3-trans-Δ2-enoyl-CoA isomerase activity, in addition to the enoyl-CoA hydratase activity (Ismail et al., 2003). Supposedly, in the absence of the PaaF and PaaH proteins, the catabolism of phenylacetate ends at the level of 3-hydroxyadipyl-CoA. Homologous genes to paaF and paaH are found in the genome of strain JMP134. The paaF gene is located close to the paaG gene and the paaH gene is found just next to the second copy of the paaG gene, paaG2 (Table 4). In E. coli, the β-ketoadipyl-CoA intermediate produced by the action of the PaaH protein could be thiolytically cleaved by the PaaJ/PaaE protein (which would be similar to the β-ketoadipyl-CoA thiolases involved in the lower part of the β-ketoadipate pathway) in order to produce acetyl-CoA and succinyl-CoA (Ismail et al., 2003; Nogales et al., 2007). It should be emphasized that no gene homologous to paaJ/paaE is located in the paa gene clusters of C. necator JMP134 (Fig. 2). Therefore, it is reasonable to assume that for the phenylacetate catabolism in this strain, the enzyme that should be encoded by the paaJ/paaE gene can be recruited from other catabolic pathways such as the β-ketoadipate pathway; specifically, the pcaF gene encoding a β-ketoadipyl-CoA thiolase (Fig. 10).
Table 4. Genes encoding pathways for phenylacetate, benzoate, and anthranilate proceeding through aryl-CoA intermediates
The regulation of the paa gene cluster in E. coli is controlled by two elements: (1) PaaX, a transcriptional repressor which contains a stretch of 25 residues with similarity to the HTH motif of transcriptional regulators of the GntR family, but constitutes a different family; and (2) phenylacetyl-CoA, which specifically inhibits the binding of PaaX to the repression-binding sites in the promoter sequences of the paa cluster (Ferrandez et al., 2000). A PaaX-binding sequence is located divergently to the paaA1B1C1D1E1ZK1PL gene cluster in C. necator JMP134, which suggests that this paa gene cluster is regulated in a similar way to that of E. coli. It should be noted that downstream of the paaA2B2C2D2E2 genes in strain JMP134, a putative TetR-family transcriptional regulator gene (paaR) is encoded, whose gene product shows a 44% aa identity with ORF3 of the phenylacetate catabolism gene cluster of A. evansii KB740 (Rost et al., 2002); this indicates that this putative gene could also be involved in the regulation of this paa gene cluster.
In C. necator JMP134, the degradation of several aromatic compounds can be assumed to proceed through the phenylacetyl-CoA pathway. The genes that encode a periplasmic aromatic amine dehydrogenase (aau genes) were identified (Table 4; Fig. 2). These could be responsible for the catabolism of phenylethylamine into phenylacetaldehyde, which could then be transformed into phenylacetate by a phenylacetaldehyde dehydrogenase (Pad) (Fig. 10) (Chistoserdov, 2001). The degradation of phenylalkanoates that contain an even number of carbon atoms as phenylbutyrate and phenylhexanoate, is assumed to be accomplished by a β-oxidation complex which catalyzes the formation of phenylacetyl-CoA, as has been shown in P. putida (Olivera et al., 2001), but the presence and identity of these genes in C. necator JMP134 could not be determined. On the other hand, the metabolism of phenylpyruvate in Azospirillum brasilense (Somers et al., 2005) and Saccharomyces cerevisiae (Vuralhan et al., 2003) has been proposed to occur through decarboxylation into phenylacetate. In C. necator JMP134, however, putative genes for phenylpyruvate decarboxylase were not identified, which suggests the existence of a different pathway or a different kind of decarboxylase-encoding gene.
The homogentisate ring-cleavage pathway
The homogentisate ring-cleavage pathway is widespread in eukaryotic and prokaryotic cells, because it is the central route for phenylalanine and tyrosine catabolism. Phenylalanine is converted into tyrosine by a pterin- and metal-dependent phenylalanine hydroxylase (PhhA) with an auxiliary carbinolamine dehydratase (PhhB) that is responsible for the regeneration of the pterin cofactor. A tyrosine aminotransferase (TyrB) transforms tyrosine into 4-hydroxyphenylpyruvate, which is further converted into homogentisate by a 4-hydroxyphenylpyruvate dioxygenase (Hpd) (Fig. 7). Homologous genes for the phenylalanine and tyrosine peripheral pathways are present in the genome of strain JMP134 (Table 2). The phhA and phhB genes are clustered (C1 in Fig. 2), and show a 46% and a 30% aa identity, respectively, with the orthologous genes described in P. aeruginosa (Zhao et al., 1994). However, a higher identity of the PhhA gene product was observed with the phenylalanine hydroxylase of the betaproteobacterium Chromobacterium violaceum (Onishi et al., 1991) (Table 2). It should be noted that the PhhA protein of P. aeruginosa binds iron at the active site, like all mammalian aromatic amino acid hydroxylases (Zhao et al., 1994), whereas the C. violaceum PhhA protein is unique in its use of copper as a cofactor (Onishi et al., 1991), which suggests that the C. necator JMP134 PhhA gene product may also be a copper-containing enzyme. The following steps in the phenylalanine pathway in strain JMP134 are probably carried out by the products of the tyrB and hpd genes, which form a gene cluster together with aroP, that encode a general aromatic amino acid permease (C2 in Fig. 2. Table 2). This is different from most other proteobacteria, in which the hpd gene is not linked to other genes involved in phenylalanine and tyrosine degradation.
The homogentisate central pathway (Fig. 7) includes a homogentisate dioxygenase (HmgA) that opens the aromatic ring of homogentisate producing maleylacetoacetate which, in some bacteria, is directly hydrolyzed, yielding maleate and acetoacetate (Crawford, 1976); nevertheless, in most bacteria it is isomerized into fumarylacetoacetate (Chapman & Dagley, 1962). This isomerization is either catalyzed by a GSH-independent maleylacetoacetate isomerase, as in most Gram-positive bacteria (Hagedorn & Chapman, 1985), or by GSH-dependent enzymes (hmgC gene products), as has been reported in the Gram-negative strains D. acidovorans (Hareland et al., 1975), B. cepacia (Hamzah & Al-Baharna, 2001), and A. evansii KB740 (Mohamed Mel et al., 2002). Finally, fumarylacetoacetate is hydrolyzed by a specific hydrolase (HmgB) forming fumarate and acetoacetate (Fig. 7). Genes encoding enzymes of the homogentisate pathway are found scattered in the genome of strain JMP134 (Fig. 2, Table 2). A cluster comprising putative hmgA and hmgB1 genes is encoded in the small chromosome (Fig. 2), but a putative hmgC gene is located in a different region of the genome (C1 in Fig. 2). The nonlinkage of the hmgAB1 and hmgC genes has also been observed in R. solanacearum, Bordetella bronchiseptica, B. japonicum, Silicibacter pomeroyi and Pseudomonas syringae (Arias-Barrau et al., 2004). However, a second copy of a putative hmgB gene is found clustered to the hmgC gene (C1 in Fig. 2, Table 2). Both hmgB genes of C. necator JMP134 are equally related to the hmgB gene of P. putida U (Table 2), but the putative hmgB2 gene has only a 45% aa identity with hmgB1.
All three isomers of hydroxyphenylacetate are growth substrates for C. necator JMP134. Unfortunately, the information on the genes and biochemical functions involved in the catabolism of hydroxyphenylacetates in bacteria is too limited to search for the presence of such pathways in C. necator JMP134. However, it is possible to predict that, in this strain, at least 3- and 4-hydroxyphenylacetate would be catabolized by the homogentisate ring-cleavage pathway. 4-Hydroxyphenylacetate degradation has been reported to take place via homoprotocatechuate and a subsequent meta ring-cleavage pathway in the Gammaproteobacteria, E. coli and P. putida (Arunachalam et al., 1992; Olivera et al., 1994; Diaz et al., 2001). A genomic search for hpaBC genes that encode a two component 4-hydroxyphenylacetate-3-hydroxylase, and for an hpaD gene, that encodes a homoprotocatechuate-2,3-dioxygenase, does not show significant matches in the genome of strain JMP134; this indicates the absence of a meta ring-cleavage pathway for the degradation of homoprotocatechuate. A different 4-hydroxyphenylacetate degradation pathway has been reported in the Betaproteobacteria, D. acidovorans (Hareland et al., 1975), B. cepacia (Hamzah & Al-Baharna, 2001), and A. evansii KB740 (Mohamed Mel et al., 2002); it involves hydroxylation of the aromatic ring at C-1 with a concomitant migration of the carboxymethyl side chain to C-2 (the ‘NIH’ shift reaction), catalyzed by a NADH-dependent 4-hydroxyphenylacetate-1-hydroxylase and yielding homogentisate (Fig. 7). Therefore, it is likely that 4-hydroxyphenylacetate is metabolized by the homogentisate pathway in C. necator.
The degradation of 3-hydroxyphenylacetate has not been thoroughly studied. In E. coli, it has been reported that, like 4-hydroxyphenylacetate, this compound is degraded through the homoprotocatechuate pathway. However, in Flavobacterium species, a FAD-dependent 3-hydroxyphenylacetate-6-hydroxylase, that produces homogentisate, has been described and an N-terminal amino acid sequence has been determined (Van Berkel & Van Den Tweel, 1991). The use of this N-terminal sequence as an ‘in silico’ probe did not show any putative ORF in the C. necator JMP134 genome. It has been shown recently that P. putida U also metabolizes 3-hydroxyphenylacetate through the homogentisate pathway (Arias-Barrau et al., 2004), and genes mhaA and mhaB have been identified. They encode a 3-hydroxyphenylacetate-6-hydroxylase, a novel two-component flavoprotein aromatic hydroxylase. A mhaC gene that encodes a 3-hydroxyphenylacetate permease has also been identified (Arias-Barrau et al., 2004). In C. necator JMP134, a mhaA gene has been found that shows a 43% aa identity with the mhaA gene product from P. putida U (C2 in Fig. 2; Fig. 5, Table 2), but a mhaB homologous gene is absent.
The catabolism of 2-hydroxyphenylacetate has also been scarcely studied in bacteria. In early reports, 2-hydroxyphenylacetate was proposed as an intermediate in the phenylacetyl-CoA pathway, but recent studies discard this possibility in E. coli (Ismail et al., 2003). On the other hand, in P. fluorescens ST, it has been proposed that 2-hydroxyphenylacetate could be metabolized via homogentisate, based on the detection of a 2-hydroxyphenylacetate-5-hydroxylase activity that transforms 2-hydroxyphenylacetate into homogentisate in bacteria (Baggi et al., 1983), as has been reported for fungi (Mingot et al., 1999). This transformation is analogous to that catalyzed by salicylate-5-hydroxylase to produce gentisate in Ralstonia sp. U2 (Zhou et al., 2002). In strain JMP134, there are two putative gene clusters that would encode this enzyme (see ‘Degradation of salicylate and 3-hydroxybenzoate: the gentisate pathway’). One of these putative gene clusters has a much lower identity with the Ralstonia sp. U2 genes. Therefore, it is possible to speculate that this lower identity gene cluster encodes a 2-hydroxyphenylacetate-5-hydroxylase responsible for 2-hydroxyphenylacetate catabolism in C. necator JMP134.
The 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway
The aerobic degradation of 3-hydroxyphenylpropionate (3-HPP) and 3-hydroxycinnamate is commonly started by a monooxygenase whose activity generates 2,3-dihydroxyphenylpropionate or 2,3-dihydroxycinnamate as central intermediates; these are further degraded via a meta ring-cleavage hydrolytic pathway (Diaz et al., 2001), that has been well described in E. coli (Ferrandez et al., 1997; Diaz et al., 1998, 2001; Torres et al., 2003). The 3-HPP and 3-hydroxycinnamate degradation pathway in E. coli is encoded by the mhp genes cluster. The MhpA monooxygenase transforms 3-HPP or 3-hydroxycinnamate into 2,3-dihydroxyphenylpropionate or 2,3-dihydroxycinnamate, respectively, which are then further converted into succinate (or fumarate, in the case of 3-hydroxycinnamate degradation), pyruvate, and acetyl-CoA, through the action of an extradiol dioxygenase (MhpB), a hydrolase (MhpC), a hydratase (MhpD), an aldolase (MhpE) and an acetaldehyde dehydrogenase (MhpF) (Fig. 3). A similar pathway has been described in C. testosteroni TA441, whose mhp gene cluster resembles that of E. coli (Arai et al., 1999b). In addition, a hpp gene cluster responsible for the partial catabolism of 3-HPP has been described in R. globerulus PWD1 (Barnes et al., 1997), but with a different gene organization, and a low sequence similarity with the mhp gene clusters of Gram-negative bacteria (Diaz et al., 2001). Cupriavidus necator JMP134 is able to grow on phenylpropionate, cinnamate, 3-HPP and 3-hydroxycinnamate, and a genomic search rendered a gene cluster resembling that of E. coli, but lacking the mhpE and mhpF genes, and with a slightly different gene organization (C1 in Fig. 2). The mhpE and mhpF genes encode the functions of the last two steps in the pathway: 4-hydroxy-2-ketovalerate aldolase and acetaldehyde dehydrogenase, respectively; these are shared with the catechol meta ring-cleavage pathway, which are encoded by the phlG and phlF genes in C. necator (Fig. 3). It is noteworthy that, in C. necator JMP134, phlG and phlF genes, thus named because of their higher identity to the gene products involved in the catechol meta ring-cleavage pathway (Table 3), are located apart from both gene clusters that encode meta ring-cleavage pathways (Fig. 2), unlike in gene organizations in other bacteria. In C. testosteroni TA441, in which both pathways have been described, there are two pairs of phlG and phlF genes located in the corresponding gene clusters (Arai et al., 1999b, 2000).
The C. necator JMP134 mhp genes have a moderate (50–70%) amino acid identity with the E. coli and C. testosteroni mhp genes (Table 3). The mhpA gene of strain JMP134 clusters together with other mhpA genes of Gram-negative bacteria and the hppA of R. globerulus PWD1, in the dendrogram of FAD-dependent hydroxylases (Fig. 5). It should be noted that a second putative FAD-dependent hydroxylase gene product (ReutB5808) of C. necator JMP134 groups with MhpA proteins (Fig. 5). However, it is unknown whether this gene product acts as a second 3-hydroxyphenylpropionate hydroxylase.
Cupriavidus necator JMP134 does not grow on 2-hydroxyphenylpropionate (melilotate) or 2-hydroxycinnamate, which could be explained by the inability of the MhpA protein to hydroxylate 2-hydroxyphenylpropionate to produce 2,3-dihydroxyphenylpropionate, as reported on the MhpA protein of E. coli (Burlingame & Chapman, 1983) and on the HppA protein of R. globerulus PWD1 (Barnes et al., 1997). In contrast, the strain Rhodococcus sp. V49 – which is able to grow on 2-hydroxyphenylpropionate using a 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway – possesses a 2-hydroxyphenylpropionate hydroxylase encoded by the ohpB gene, that also belongs to the flavin-type aromatic hydroxylases family (Powell & Archer, 1998); nevertheless, its gene is more closely related to the monocomponent phenol hydroxylase encoding pheA gene of Pseudomonas sp. EST1001 (Nurk et al., 1991) than to the mhpA genes. No close relative of ohpB genes was found in the genome of strain JMP134, in concordance with the inability of this strain to grow on 2-hydroxyphenylpropionate or 2-hydroxycinnamate.
In C. necator JMP134, the 2,3-dihydroxyphenylpropionate-1,2-dioxygenase encoded by the mhpB gene, clusters with the type II extradiol dioxygenases (Fig. 9b) (Vaillancourt et al., 2004). The mhpB gene corresponds to the mpcI gene formerly described in C. necator JMP222 as a catechol-2,3-dioxygenase (Kabisch & Fortnagel, 1990a) (see ‘The catechol meta ring-cleavage pathway’). Later studies have shown that MpcI and MhpB proteins from E. coli are structurally and functionally related, and they have been proposed to constitute the type II family of extradiol dioxygenases (Spence et al., 1996). MpcI (MhpB) of C. necator JMP134 showed a broad specificity toward three-substituted catechols; propionate was found to be the optimum side chain (Spence et al., 1996). 2,3-Dihydroxycinnamate was as good a substrate as 2,3-dihydrophenylpropionate; this indicates that the enzyme is able to bind the alkyl side chain in a transoid conformation. Two other putative gene products in strain JMP134 (ReutB5775 and ReutB4784) have been assigned to type II extradiol dioxygenases, but they are distantly related to the MhpB gene product (Fig. 9b), and cluster with the extradiol dioxygenases involved in the degradation of syringate in Sphingomonas paucimobilis SYK-6 (Peng et al., 1998; Kasai et al., 2004).
The next step in this pathway is the hydrolytic cleavage of the extradiol ring fission product of 2,3-dihydrophenylpropionate; the resulting products are succinate (or fumarate from 2,3-dihydroxycinnamate) and 2-hydroxy-penta-2,4-dienoate (Fig. 3). This step is catalyzed by the 2-hydroxy-6-keto-nona-2,4-diene-1,9-dienoate hydrolase encoded by the mhpC gene. All other reported hydrolases that act on the ring-cleavage product of 2,3-dihydrophenylpropionate and 2,3-dihydroxycinnamate – i.e. MhpC from E. coli, MhpC from C. testosteroni TA441, HppC from R. globerulus PWD1, and OhpC from Rhodococcus sp. V49 – appear to be highly specific for the cleavage products (Barnes et al., 1997; Lam & Bugg, 1997; Powell & Archer, 1998; Arai et al., 1999a), so the same substrate specificity is expected for the MhpC gene product of C. necator. The hydrolase most closely related to MhpC of C. necator JMP134 is MhpC from E. coli. In contrast, the MhpD gene product from C. necator, responsible for the conversion of 2-hydroxy-penta-2,4-dienoate into 4-hydroxy-2-ketopentanoate, is most closely related to the MhpD protein of C. testosteroni TA441 (Table 3). The MhpD gene product is also homologous to the 2-hydroxypent-2,4-dienoate hydratase of the catechol meta ring-cleavage pathway in strain JMP134 (PhlE, 40% aa identity). A putative 3-hydroxyphenylpropionate transporter is encoded by the mhpT gene as part of the mhp genes cluster in C. necator; it shows a high amino acid identity with the MhpT protein of E. coli (Diaz et al., 2001), a member of the aromatic : H+ symporter family of transport proteins. Finally, a mhpR gene, divergently located in the C. necator mhp genes cluster (C1 in Fig. 2, Table 3), putatively encodes an IclR-type transcriptional regulator, which is homologous to the mhpR gene of E. coli (Torres et al., 2003). Although IclR-type regulators are generally transcriptional repressors, those which control catabolic pathways have always been described as activators (Tropel & van der Meer, 2004), including the MhpR protein of E. coli, which, in the presence of 3-HPP, activates the expression of mhp genes by binding to an operator region located upstream of the promoter (Torres et al., 2003).
Other catabolic pathways for aromatic compounds
The aerobic benzoyl-CoA pathway
A novel aerobic pathway for benzoate degradation has recently been described in A. evansii KB740. In this pathway (Fig. 10), benzoate is first converted, by an AMP forming benzoate-CoA ligase, into benzoyl-CoA. Benzoyl-CoA is hydroxylated and reduced at positions 2 and 3. The reaction is catalyzed by benzoyl-CoA oxygenase/reductase, a two component benzoyl-CoA dioxygenase, which is very dissimilar to other known oxygenase systems (Zaar et al., 2004), and it is composed by two proteins: an iron–sulfur-flavoprotein reductase (BoxA) and an oxygenase (BoxB). The dihydrodiol is the substrate for ring fission catalyzed by dihydrodiol lyase (BoxC) (Gescher et al., 2005). This homodimeric enzyme does not require oxygen and catalyzes the transformation to 3,4-dehydroadipyl-CoA semialdehyde. The latter intermediate is subsequently oxidized by 3,4-dehydroadipyl-CoA semialdehyde dehydrogenase (BoxD) to 3,4-dehydroadipyl-CoA (Gescher et al., 2006). The further metabolism is thought to lead to β-ketoadipyl-CoA, which is finally cleaved into succinyl-CoA and acetyl-CoA (Zaar et al., 2001). A cluster of fifteen genes that putatively encode this new benzoate pathway has been recently identified in the chromosome of A. evansii (Gescher et al., 2002). A gene cluster that putatively encodes a similar aerobic CoA-dependent pathway for benzoate degradation is also found in C. necator JMP134 (C1 in Fig. 2). This gene cluster comprises almost all the genes that are apparently essential for the benzoate catabolism in A. evansii (Table 4), including the benzoate CoA ligase (BzdA), the NADPH- and oxygen-dependent benzoyl-CoA oxygenase/reductase (BoxAB), the dihydrodiol lyase – involved in hydrolytic ring-cleavage (BoxC) – and a lactonase (ORF2), that putatively hydrolyzes the 3,6-lactone of the β-hydroxyadipyl-CoA formed in the last steps of the pathway (Fig. 10). Nevertheless, an 3,4-dehydroadipyl-CoA semialdehyde dehydrogenase (BoxD), which is involved in the oxidation of the aldehyde group in the 3,4-dehydroadipyl-CoA semialdehyde (Gescher et al., 2006), is absent. This enzyme activity can, however, be recruited from the dehydrogenase gene pool in C. necator JMP134. The C. necator's box gene cluster, like A. evansii's, includes an ABC transporter, putatively responsible for an effective benzoate uptake. However, whereas in A. evansii (Gescher et al., 2002) this ABC transporter comprises an ATP-binding membrane-spanning protein encoded by a single ORF (ORF4), C. necator contains two separate ORFs: one of these encodes the N-terminal domain (ORF4n) and the other the C-terminal domain (ORF4c) (not shown). A putative regulator, the BzdR gene product, with a high identity with the product of the bzdR gene of A. evansii CIB (Lopez Barragan et al., 2004) and with ORF10 of A. evansii KB740 (Gescher et al., 2002) is also encoded by the C. necator box gene cluster (Table 4). This regulator is a two-domain protein with an N-terminal domain that is similar to the regulatory proteins of the HTH family, and a C-terminus related to shikimate kinase I of E. coli. This suggests that the activity of this protein could be modulated by ATP-dependent phosphorylation in response to benzoate (Gescher et al., 2002).
The role of the box gene cluster in C. necator JMP134 is intriguing, because it is well established that strain JMP134 and related C. necator strains degrade benzoate through the β-ketoadipate pathway (Johnson & Stanier, 1971; Sauret-Ignazi et al., 1996; Ampe et al., 1997). The presence of two homologous copies of a gene cluster that encode for the enzymes of benzoate degradation via CoA activation, related to the box gene cluster of A. evansii, has been recently demonstrated in B. xenovorans LB400 (Gescher et al., 2002; Denef et al., 2004). Upregulation of one copy of these two gene clusters was found in cells grown on biphenyl, a compound that is degraded through benzoate, whereas no induction of these gene clusters was observed in benzoate grown cells (Denef et al., 2004). However, B. xenovorans LB400 mutants, which are defective in benzoate degradation via the β-ketoadipate pathway, were still able to grow on benzoate, recruiting both gene clusters for benzoate degradation via benzoyl-CoA (Denef et al., 2006). A similar situation may take place in C. necator JMP134, if the box gene cluster – putatively used for the degradation of the benzoyl-CoA produced from peripheral pathways – can be expressed when a high turnover of benzoate is required (because the β-ketoadipate pathway is defective) or if reduced oxygen tension is present, as has been suggested for B. xenovorans LB400 (Denef et al., 2006).
Pathways for amino- and nitroaromatic compounds
The 2-aminobenzoyl-CoA pathway
A new pathway for the metabolism of 2-aminobenzoate has been recently described in A. evansii KB740 (Schuhle et al., 2001). This not yet completely elucidated pathway (Fig. 10), similar to the benzoate pathway encoded by the box genes, begins with a 2-aminobenzoate-CoA ligase activity forming 2-aminobenzoyl-CoA; then, a 2-aminobenzoyl-CoA monooxygenase/reductase (ACMR) forms 2-amino-5-oxo-cyclohex-1-ene-1-carbonyl-CoA, and proceeds with β-oxidation (Schuhle et al., 2001). Two similar copies of the abm gene cluster – including the genes encoding 2-aminobenzoate-CoA ligase, the ACMR enzyme and three enzymes of a β-oxidation pathway – have been reported in A. evansii KB740 (Schuhle et al., 2001), and are coordinately expressed during the aerobic growth of this bacterium on 2-aminobenzoate. In C. necator, a gene cluster is found that encodes almost all the homologues of the A. evansii abm genes (C1 in Fig. 2, Table 4), with the exception of the abmF and abmH genes, which putatively encode a MarR-like regulator and a substrate-binding protein of an ABC transporter, respectively (Schuhle et al., 2001). In addition, a gene (abmX) that putatively encodes a thioesterase – with similarity to one found in the box gene cluster of A. evansii (see previous section) – is also present in the abm gene cluster of C. necator (Table 4). This putative thioesterase can undertake the final steps of the β-oxidation pathway in C. necator, involved in 2-aminobenzoate catabolism. It is worth noting that in the closely related betaproteobacterium B. cepacia DBO1 (Chang et al., 2003), and in the two Gammaproteobacteria Pseudomonas resinovorans (Urata et al., 2004) and A. baylyi ADP1 (Eby et al., 2001), a ‘classical anthranilate pathway’ has been described; in it, anthranilate is converted into catechol by anthranilate-1,2-dioxygenase and metabolized through the classical β-ketoadipate pathway. A genomic search in strain JMP134 does not yield any similar ORF. In addition, an early study in the closely related strain C. necator 335 showed no induction of the enzymes of the β-ketoadipate pathway in tryptophan-grown cells (Johnson & Stanier, 1971), which excludes the possibility that 2-aminobenzoate is metabolized by this pathway. In C. metallidurans CH34, a three-step pathway of aerobic l-tryptophan degradation to anthranilate has been described recently (Kurnasov et al., 2003). The kynBAU operon encoding three required enzymes for this pathway: tryptophan 2,3-dioxygenase (gene kynA), kynurenine formamidase (gene kynB), and kynureninase (gene kynU), is found in C. necator JMP134 (Table 4). Taken together, these clues indicate that the kyn and abm gene clusters in C. necator are responsible for tryptophan catabolism through a 2-aminobenzoyl-CoA pathway.
The 3-hydroxyanthranilate pathway
A pathway for the degradation of 2-nitrobenzoate has been recently described in P. fluorescens KU-7 (Hasegawa et al., 2000; Muraki et al., 2003). This pathway is peculiar, because it proceeds through the meta ring-cleavage of 3-hydroxyanthranilate, an aromatic intermediate which has been described before in the kynurenine pathway of eukaryotic organisms, but not in bacteria (Kucharczyk et al., 1998). Further catabolism of 3-hydroxyanthranilate is undertaken by a 3-hydroxyanthranilate-3,4-dioxygenase (NbaC), which cleaves the aromatic ring into 2-amino-3-carboxymuconate-6-semialdehyde, while the 2-amino-3-carboxymuconate-6-semialdehyde decarboxylase (NbaD) catalyzes the decarboxylation of the latter compound into 2-aminomuconate-6-semialdehyde (Fig. 3). The subsequent action of 2-aminomuconate-6-semialdehyde dehydrogenase (NbaE), 2-aminomuconate deaminase (NbaF), 4-oxalocrotonate decarboxylase (NbaG), 2-oxopent-4-dienoate hydratase (NbaH), 4-hydroxy-2-oxovalerate aldolase (NbaI) and acylating aldehyde dehydrogenase (NbaJ), finally produce pyruvate and acetyl-CoA. The nba genes responsible for the 3-hydroxyanthranilate meta ring-cleavage pathway in P. fluorescens KU-7 have been sequenced (Muraki et al., 2003), and a gene cluster closely resembling that of P. fluorescens KU-7 has been found in C. necator (C2 in Fig. 2). However, C. necator JMP134 is unable to grow on 2-nitrobenzoate and the corresponding genes for 3-hydroxyanthranilate meta ring-cleavage pathway have been designated as haa genes (3-hydroxyanthranilic acid). The haa gene cluster of C. necator shows amino acid sequence identities ranging from 64% to 72% (Table 3), excluding the LysR-type regulator haaR gene, which shows a lower identity with nbaR. The haa gene cluster encodes all the genes that are essential for the catabolism of 3-hydroxyanthranilate, with the exception of nbaI and nbaJ genes. However, these two functions are shared with the catechol- and the 2,3-dihydroxyphenylpropionate meta ring-cleavage pathways of C. necator (see ‘The catechol meta ring-cleavage pathway’ and ‘The 2,3-dihydroxyphenylpropionate meta ring-cleavage pathway’) and can be recruited from the above mentioned PhlG and PhlF gene products.
Catabolic pathways for nitrophenols
Three nitrophenols support the growth of C. necator JMP134: 2,6-dinitrophenol (Ecker et al., 1992), 3-nitrophenol (Schenzle et al., 1997) and 2-chloro-5-nitrophenol (Schenzle et al., 1999a). The initial degradation pathway for 3-nitrophenol consists of the transformation into 3-hydroxylaminophenol and then into aminohydroquinone; these are catalyzed by a 3-nitrophenol nitroreductase (MnpA) and a 3-hydroxylaminophenol mutase (Fig. 11), respectively (Schenzle et al., 1997, 1999b). The 3-hydroxylaminophenol mutase was identified as a glutamine syntethase (Table 5) (Schenzle et al., 1999b). The initial degradation of 2-chloro-5-nitrophenol is analogous to that of 3-nitrophenol, and results in the formation of 2-amino-5-chlorohydroquinone (Schenzle et al., 1999a). The removal of the chlorine group in the latter intermediate is reductively mediated and produces aminohydroquinone, a common intermediate. Further degradation of aminohydroquinone (G. Zylstra, pers. commun.) (Fig. 11), proceeds through the ring-cleavage between the adjacent hydroxyl and amino groups and is catalyzed by an aminohydroquinone dioxygenase (MnpC); then, the amide group is removed by an amidase (MnpD) to form maleylacetate, which is transformed into β-ketoadipate by a maleylacetate reductase (MAR) (MnpE). It is worth mentioning that, of the six MAR genes present in C. necator (Fig. 12), mnpE gene is the less related one (see next sections). A gene cluster, mnpCARED (megaplasmid pJPL in Fig. 2), that putatively encodes the enzymes that catalyze the conversion of 3-nitrophenol into β-ketoadipate is found in strain JMP134. The amino acid sequences of the Mnp proteins of C. necator have low identity levels with known homologues (Table 5). Interestingly, putative gene sequences are found at the flanks of the mnpCARED gene cluster, that encode the dehalogenating activity in 2-chloro-5-nitrophenol degradation (mnpF); and an aminobenzoquinone reductase activity that forms aminohydroquinone (mnpG).
Table 5. Genes encoding pathways for (chloro)hydroxyquinol, (amino)hydroquinone and methylhydroquinone and peripheral reactions
With the exception of 2,4,6-trichlorophenol (2,4,6-TCP), and 2,4-dichlorophenol, C. necator is unable to use chlorinated phenols as carbon sources (Clement et al., 1995). Whereas 2,4-dichlorophenol is metabolized via 3,5-dichlorocatechol (see ‘Tfd functions: biochemistry and genetics’), 2,4,6-TCP, – like 2,4,5-trichlorophenol (2,4,5-TCP) in other bacteria – is metabolized through the so-called (chloro)hydroxyquinol pathway (Fig. 11) (Kasberg et al., 1995; Zaborina et al., 1995). The first step, catalyzed by the 2,4,6-TCP monooxygenase (TcpA), is the oxidative conversion of 2,4,6-TCP into 2,6-dichlorobenzoquinone, followed by a hydrolytic dechlorination that produces 6-chloro-2-hydroxybenzoquinone (Xun & Webster, 2004). 6-chloro-2-hydroxybenzoquinone is either chemically or enzymatically reduced (probably by TcpB) to 6-chloro-2-hydroxybenzoquinol (Fig. 11). Contrary to previous assumptions, 2,6-dichlorobenzoquinol is not an intermediate of 2,4,6-TCP degradation in strain JMP134 (Xun & Webster, 2004). 6-Chloro-2-hydroxybenzoquinol is transformed into 2-chloromaleylacetate by 1,2-hydroxyquinol dioxygenase (TcpC), and then converted into β-ketoadipate by MAR (TcpD) (Louie et al., 2002; Matus et al., 2003). This pathway is different from that reported for 2,4,5-TCP degradation in B. cepacia AC1100, which starts with the oxidation of 2,4,5-TCP to 2,5-dichlorobenzoquinone, and its further transformation into 5-chloro-2-hydroxybenzoquinol (Kasberg et al., 1995). The latter compound is dechlorinated to hydroxybenzoquinone and then reduced to hydroxybenzoquinol by a quinone reductase. Thus, all chloride substituents are removed from the aromatic ring before its cleavage. Whereas the hydroxyquinol-1,2-dioxygenase of strain AC1100 is unable to use 5-chloro- or 6-chlorohydroxyquinol (Kasberg et al., 1995), hydroxyquinol-1,2-dioxygenases of 2,4,6-TCP-degrading strains usually use chlorohydroxyquinol and hydroxyquinol as substrates, although they vary in their substrate preferences (Kasberg et al., 1995; Latus et al., 1995; Zaborina et al., 1995; Hatta et al., 1999). (Chloro)hydroxyquinol dioxygenases form a distinct group in the dendrogram of intradiol-1,2-dioxygenases (Fig. 4), that also includes the enzyme from Arthrobacter sp. strain BA-5-17 (Murakami et al., 1999).
In strain JMP134, enzymes for the metabolism of 2,4,6-TCP are encoded by the tcpRXABCYD gene cluster (C1 in Fig. 2) (Matus et al., 2003). Sequence analysis indicates that tcpA and tcpB genes encode a FADH2 utilizing monooxygenase – with a 65% aa sequence identity with the TftD protein from B. cepacia AC1100 – and a flavin reductase, respectively (Table 5). However, the analysis of mutants of strain JMP134 that are defective in different tcp genes, showed that tcpB is not required for the conversion of 2,4,6-TCP (Louie et al., 2002; Sanchez & Gonzalez, 2007). It has been shown very recently that TcpB has activity for quinone reduction with FMN or FAD as the cofactor, and NADH as the reductant (Belchik & Xun, 2008). Sequence comparison with the tftC gene – which encodes a flavin reductase in B. cepacia AC1100 – strongly suggests that, as in 2,4,5-TCP degradation, a NADH : FAD oxidoreductase is involved in the initial monooxygenation and that in strain JMP134 this function is carried out by the TcpX gene product (Fig. 11, Table 5) (Matus et al., 2003; Sanchez & Gonzalez, 2007). In vitro assays of coupling of TcpX and TcpA demonstrated that TcpX provided FADH2 for TcpA catalysis (Belchik & Xun, 2008).
4-Fluorobenzoate degradation is common to most Cupriavidus strains (Schlomann et al., 1990b). The degradation of 4-fluorobenzoate does not require the enzymes specialized in halocatechol degradation (see next section), as evidenced by the fact that C. necator JMP222 – a derivative of strain JMP134 cured of plasmid pJP4 – grows on 4-fluorobenzoate (Schlomann et al., 1990b). The degradation pathway for this halobenzoate, studied in the strain C. necator 335, is started by the transformation of 4-fluorobenzoate into 4-fluorocatechol, performed by benzoate dioxygenase and benzoate dihydrodiol dehydrogenase [see ‘The ortho ring-cleavage pathways for benzoate and p-hydroxybenzoate (the β-ketoadipate central pathway)]. In contrast to chlorinated derivatives (see next section), 4-fluorocatechol is a good substrate for proteobacterial catechol-1,2-dioxygenases [see ‘The ortho ring-cleavage pathways for benzoate and p-hydroxybenzoate (the β-ketoadipate central pathway)] which produce 3-fluoromuconate. Cycloisomerization of this ring-cleavage product results in the formation of 4-fluoromuconolactone (Schlomann et al., 1990a) (see next section). In both Cupriavidus strains, JMP134 and 335, a trans-dienelactone hydrolase that produces maleylacetate has been reported to be induced during growth in 4-fluorobenzoate (Schlomann et al., 1990b); this enzyme is supposed to transform 4-fluoromuconolactone into maleylacetate (Nikodem et al., 2003). The pathway is completed with the transformation of maleylacetate into β-ketoadipate by MAR. Recent evidence indicates that in strain 335, a MAR encoding gene, macA, is found in a gene cluster along with macB gene, which hypothetically encodes a membrane transport protein, and macR gene – which encodes a putative regulator (Seibert et al., 2004). The fact that strain JMP222 is able to grow on 4-fluorobenzoate is indicative that plasmid-encoded MAR would be functional for maleylacetate turnover, but not essential. Any of the chromosome-encoded MAR would assume the main role in 4-fluorobenzoate degradation, being tcpD, hqxD and hqoD genes more closely related to macA gene than mnpE gene (Fig. 12), and probably are functionally redundant. It should be noted that no ORF similar to the macR gene from strain 335 is found in the neighborhood of the tcpD, hqxD or hqoD genes, or elsewhere in the genome of strain JMP134.
Catabolic pathway for mono- and dichlorinated compounds: the tfd genes
Tfd functions: biochemistry and genetics
The catabolic pathway for 2,4-D has been thoroughly studied in strain JMP134 (Fig. 13). This pathway is encoded by tfd (two, four-dichlorophenoxyacetate) genes (Fig. 13b), which are located in the pJP4 plasmid, and initiated by a 2,4-D/α-ketoglutarate dioxygenase (Fukumori & Hausinger, 1993a, b). The same pathway has been reported in various 2,4-D-degrading isolates (Beadle & Smith, 1982; Chaudhry & Huang, 1988; Bhat et al., 1994; Maltseva et al., 1996; Poh et al., 2002; Thiel et al., 2005); however, in some strains, 2,4-D degradation is started by monooxygenases, labeled CadAB, that are related to the TftAB protein involved in the degradation of 2,4,5-trichlorophenoxyacetic acid (Kitagawa et al., 2002). Both catabolic activities form 2,4-dichlorophenol as intermediate. In the pathway initiated by the TfdA protein, α-ketoglutarate is transformed into succinate and CO2, and 2,4-D is converted into glyoxylate and 2,4-dichlorophenol. The enzyme of strain JMP134 uses several other phenoxyacetates as substrates: MCPA, phenoxyacetate, 2-methylphenoxyacetate, 4-methylphenoxyacetate and 2-chlorophenoxyacetate (Pieper et al., 1988); however, not all of these compounds support the growth of the wild type strain (Pieper et al., 1988, 1989). The inability to use these phenoxyacetates as growth substrates is not due to a restricted specificity of the TfdA protein but to regulatory constraints, as mutants that constitutively express tfd genes could grow on all these compounds (Fig. 3) (Pieper et al., 1989). Even 2-naphthoxyacetate, benzofuran-2-carboxylate or 2,4-dichlorocinnamate are substrates for the TfdA protein of strain JMP134 (Dunning Hotopp & Hausinger, 2001). Based on the ability of the TfdA protein to metabolize chlorinated cinnamic acids, it has been proposed that tfdA-like sequences present in 2,4-D-nondegrading bacteria may metabolize substituted cinnamic acids (Dunning Hotopp & Hausinger, 2001).
The next step in 2,4-D degradation is the transformation of 2,4-dichlorophenol into 3,5-dichlorocatechol (Fig. 13a), which is catalyzed by the 2,4-dichlorophenol hydroxylase, a single component flavoprotein monooxygenase (TfdB). In pJP4, there are two genes that encode chlorophenol hydroxylases (Fig. 13b), named tfdBI and tfdBII. The enzyme activity previously purified from C. necator JMP134 grown on 2,4-D, corresponds to TfdBI (Liu & Chapman, 1984; Farhana & New, 1997), but the substrate profile of TfdBII is similar (Ledger et al., 2006): both enzymes use 2,4-dichlorophenol, 4-methyl-2-chlorophenol and 2- and 4-monosubtituted phenols as substrates, and exhibit only poor activity with phenol (Liu & Chapman, 1984; Farhana & New, 1997; Ledger et al., 2006). Moreover, both gene sequences map in the same branch of the dendrogram for FAD-dependent hydroxylases (Fig. 5).
TfdA and TfdB proteins transform 2,4-D, via 2,4-dichlorophenol, into 3,5-dichlorocatechol (3,5-DCC), a central intermediate in chloroaromatic metabolism. The reactions that produce the corresponding chlorocatechols during growth on 3-CB are carried out by (1) the benzoate dioxygenase, which introduces two oxygen atoms into the benzoate molecule to produce the benzoate-cis-1,2-dihydrodiol derivative, and (2) the benzoate dihydrodiol dehydrogenase, which restores aromaticity by forming the catechol (Fig. 13a) (Pieper et al., 1993). These enzymes are recruited from the benzoate degradation pathway and exhibit some activity with 3-chlorobenzoate, but not with 4-chlorobenzoate [see ‘The ortho ring-cleavage pathways for benzoate and p-hydroxybenzoate (the β-ketoadipate central pathway)’]. It is interesting to note that these enzymes produce 70% of 3-chlorocatechol (3-CC) and 30% of 4-chlorocatechol (4-CC) from 3-CB (Pieper et al., 1993).
The chlorocatechols produced during growth on 2,4-D or 3-CB are metabolized through the chlorocatechol ortho ring-cleavage pathway (Fig. 13a). This pathway has been reported in a number of chloroaromatic-degrading bacteria (Reineke, 1998; Timmis & Pieper, 1999; Pieper & Reineke, 2000). In C. necator JMP134, like in other bacteria, the chlorocatechol ortho ring-cleavage pathway is initiated by chlorocatechol-1,2-dioxygenase (TfdC), which produces 2,4-dichloromuconate, 3-chloromuconate and 2-chloromuconate from 3,5-DCC, 4-CC, and 3-CC, respectively (Fig. 13a). Two genes that encode TfdC (Fig. 13b) are present in strain JMP134, and both TfdCI and TfdCII proteins show a similar substrate profile (Perez-Pantoja et al., 2000; Plumeier et al., 2002). However, under the same growth conditions, the activity of TfdCI enzyme is about two to three times higher than that of TfdCII enzyme. Whether this is due to higher enzyme amounts or to differences in their specific activity remains to be elucidated. It should also be mentioned that chlorocatechol-1,2-dioxygenases usually exhibit a broad substrate range, whereas catechol-1,2-dioxygenase, only transforms 4-chlorocatechol at significant rates (Pieper et al., 1993).
Chloromuconates formed by intradiol ring-cleavage of chlorocatechol are substrates for chloromuconate cycloisomerase (TfdD), which converts them into the corresponding dienelactones (Vollmer & Schlomann, 1995). Dechlorination was shown to be enzyme catalyzed, as muconate cycloisomerases form 2- and 5-chloromuconolactone from 2-chloromuconate and are not able to dechlorinate (Vollmer et al., 1994). In contrast, chloromuconate cycloisomerases form trans-dienelactone through a specific cycloisomerization to 5-chloromuconolactone and the rotation of the catalytic center of the lactone ring that allows proton abstraction and thus dehalogenation (Schell et al., 1999). In the case of 3-chloromuconate cycloisomerization by muconate cycloisomerases, a highly unstable intermediate, 4-chloromuconolactone, is first formed; then dechlorination produces protoanemonin. Only chloromuconate cycloisomerases form cis-dienelactone, probably via an enol-enolate intermediate (Pieper & Reineke, 2004).
As in the case of tfdC genes, there are also two copies of tfdD genes (Fig. 13b). The TfdDI enzyme produces dienelactones from 2-chloro-, 3-chloro- and 2,4-dichloromuconate at significant rates; this substrate profile is shared by other chloromuconate cycloisomerases from Gram-negative strains (Vollmer et al., 1999). In contrast, the TfdDII enzyme exhibits high activity against 3-chloromuconate (forming cis-dienelactone) and poor activity against 2-chloromuconate. Moreover, this enzyme's activity results in an equilibrium between 2-chloromuconate and 5-chloro- and 2-chloromuconolactone, and it is very inefficient in catalyzing dehalogenation to form trans-dienelactone; therefore, it differs from all (chloro)muconate cycloisomerases described so far (Plumeier et al., 2002).
Dienelactones are converted into maleylacetates by dienelactone hydrolase, and MAR catalyzes the reduction of the double bond of the maleylacetate to form β-ketoadipate, the common metabolite of the catechol and the chlorocatechol ortho ring-cleavage pathways (Kaschabek & Reineke, 1992). There are two tfdE genes (Fig. 13b); however, the activity levels of the TfdEI protein, with cis-dienelactone as a substrate, are significantly higher than those of the TfdEII protein (Plumeier et al., 2002). Maleylacetates with chlorine substituents in the 2-position, such as 2-chloromaleylacetate (formed from 3,5-dichlorocatechol as intermediate of 2,4-D metabolism), are reduced by MAR in a first step to yield maleylacetate. Obviously, the enzymatic attack on the C2-carbon results in an intermediate that spontaneously eliminates chloride. The tfd gene clusters encode two tfdF genes (Fig. 13b), which means that the genome of C. necator has six MAR encoding genes (see previous sections and Fig. 12). The presence of such level of gene redundancy is intriguing; it is of key importance to study the expression of these genes in C. necator cells exposed to compounds whose degradation pathway requires MAR activity. Further degradation of the β-ketoadipate formed by MAR proceeds as in the catechol ring-cleavage pathway, using pcaIJF gene encoded functions (see ‘The pob and pca genes’); however, it is also possible that the related functions encoded in the mml genes (see ‘The methylcatechol ortho ring-cleavage pathway’) also play a role in chlorocatechol degradation.
The overall organization of tfd genes, which are located in a 22-kb region of the pJP4 plasmid and have two copies of the chlorocatechol ortho ring-cleavage pathway genes (Fig. 13), is unique among the 2,4-D-degrading strains, although D. acidovorans P4a has also been reported to comprise two tfd gene clusters (Hoffmann et al., 2003). The tfd-I gene cluster of strain JMP134 encodes one putative LysR-type regulator (the TfdT gene product), which is interrupted in its carboxyl end by an ISJP4 insertion sequence (Leveau & van der Meer, 1997). The tfd-II cluster encodes the same functions as the tfd-I cluster; however, the tfdDII and tfdCII genes are in a different order. Moreover, this cluster comprises the tfdK gene, which encodes a membrane protein involved in 2,4-D transport (Leveau et al., 1998) (Fig. 13a). The tfdA gene is found close to this gene cluster together with two inverted, perfect copies of the LysR-type regulatory genes: tfdR and tfdS (Fig. 13b). The gene organization of the tfd-I gene cluster is similar to that observed for the clc genes of P. putida AC25 (Ghosal & You, 1988, 1989) and Pseudomonas knackmussii B13 (Frantz et al., 1987), the tcb genes of Pseudomonas sp. P51 (van der Meer et al., 1991), the cbn genes of C. necator NH9 (Ogawa & Miyashita, 1999), and the tfd genes of Burkholderia sp. NK8 (Liu et al., 2001) and D. acidovorans P4a (Hoffmann et al., 2003). All these gene clusters – except for that of Burkholderia sp. NK8 and of the strain JMP134 – are characterized by the presence of an ORF of unknown function between the genes that encode chloromuconate cycloisomerase and dienelactone hydrolase. In strain JMP134, the tfd-II gene cluster organization is similar to the tfdRCEBKA gene order of the 2,4-D-degrading strains Variovorax paradoxus TV1 (AB028643), B. cepacia 2a (Poh et al., 2002), D. acidovorans P4a (Hoffmann et al., 2003) and Achromobacter xylosooxidans ssp. denitrificans EST4002 (Vedler et al., 2004). However, the tfd-II cluster of strain JMP134 is the only one that comprises the tfdD and tfdF genes. No clear trend in the evolutionary relatedness of the tfd counterparts is observed in strain JMP134. Whereas the tfdB genes map relatively close in the corresponding dendrogram (Fig. 5), the tfdC genes (Fig. 4), and specially the tfdF genes, are clearly part of unrelated branches (Fig. 12). This suggests that these two clusters are not the product of a recent gene duplication event. It is possible that the presence of redundant gene clusters in strain JMP134 may be the effect of a lateral acquisition of catabolic genes (Trefault et al., 2004).
What would be the role of these two gene clusters in strain JMP134? Although each tfd gene cluster is enough to allow growth on 3-CB (Perez-Pantoja et al., 2000; Plumeier et al., 2002), some reports suggest that the catabolic functions encoded by the tfd-II gene cluster are not required (Don et al., 1985; Laemmli et al., 2004). However, the presence of two apparently redundant gene clusters would be important for the ‘fine tuning’ of the expression of catabolic functions. On the one hand, the number of copies of these tfd gene clusters is important for the catabolic performance of these strains (Klemba et al., 2000; Trefault et al., 2002), and for the adequate response to the toxicity of catabolic intermediates, e.g. chlorocatechols (Perez-Pantoja et al., 2003). On the other hand, the presence of two tfdB encoded functions would be important to prevent the accumulation of the toxic intermediate 2,4-dichlorophenol (Ledger et al., 2006).
Regulation of the tfd genes
One of the most interesting aspects of the tfd catabolic genes, their regulation, is still not well understood. Its study has been affected by the recent realization that tfd gene organization is complex and most gene functions are redundant. Available evidence indicates that all the tfd genes are expressed during growth on 2,4-D (Leveau et al., 1999; Laemmli et al., 2004). However, it should be emphasized that studies with 3-CB or other substrates that are metabolized through the chlorocatechol ring-cleavage pathway are required, because it is highly possible that the tfd genes are expressed in a growth-substrate depending manner. Early reports indicate that two regulatory genes control the expression of the tfd genes: tfdS, described as an activator of the expression of the tfdA gene (You & Ghosal, 1995) and a repressor of the tfdB gene (Kaphammer & Olsen, 1990), and tfdR, reported to regulate the tfd-I gene cluster (Harker et al., 1989). After the work of Matrubutham & Harker (1994), which showed that the tfdS and tfdR genes were identical, earlier reports needed to be revised. The use of constructs with different copy numbers and different hosts have been proposed to explain these early results (Leveau & van der Meer, 1996). The role of the tfdR gene as a regulatory element of the tfd-I gene cluster has been nicely shown by J.R. van der Meer's lab (Leveau & van der Meer, 1996), which has provided evidence that the tfdR gene can replace the ISJP4-interrupted tfdT gene. The role of tfdR gene as a master regulatory gene for tfd gene expression is one of the few aspects that are clear. Three intergenic regions –tfdT-tfdCI, (tfdT/C), tfdR-tfdDII (tfdR/D) and tfdA-tfdS (tfdA/S) (Fig. 13b) – share significant levels of nucleotide sequence identity (40–60%) with the intergenic regions involved in (chloro)catechol catabolism that are regulated by other LysR-type transcriptional activators (Matrubutham & Harker, 1994; McFall et al., 1998). Evidence has been found for in vitro binding of the TfdR protein to the tfdT/C region (Matrubutham & Harker, 1994), which coincides with its role as activator of the tfd-I genes cluster. Albeit indirect, additional, support for the regulatory role of the TfdR protein on both tfd gene clusters is provided by the fact that the tfd-I and tfd-II modules, cloned separately and under the control of the tfdR gene, express all the corresponding Tfd enzymes (Perez-Pantoja et al., 2000). Despite this evidence, the role of the TfdR protein in the expression of the tfdA gene is still unclear. Preliminary evidence indicates, however, that the TfdR protein is able to bind to the tfdA/S intergenic region (N. Trefault, L. Guzmán, M. Manzano, D.H. Pieper & B. González, unpublished data), but further investigation is clearly required.
Is there any differential role for the TfdS and the TfdR proteins? Although the idea that two identical genes would produce proteins with different activities is bizarre, the effect of their respective positions along with very limiting amounts of a regulatory protein may produce a differential effect on gene expression of the immediately adjacent intergenic regions (tfdA/S for the TfdS protein and tfdR/D for the TfdR protein). tfdS and tfdR-inactivated pJP4 derivatives may help to explore this point. In this respect, it should be noted that a pJP4 derivative lacking the tfdS-tfdR region (pYG1010) is able to support growth on 3-CB, but not on 2,4-D (You & Ghosal, 1995). This could lead to the conclusion that tfdS/tfdR gene regulation is not necessary for the expression of tfd genes during growth on 3-CB (You & Ghosal, 1995), but a cross-talk effect mediated by catR chromosomal functions, such as those present in C. necator [see ‘The ortho ring-cleavage pathways for benzoate and p-hydroxybenzoate (the β-ketoadipate central pathway)’] has also been suggested as an explanation (Leveau & van der Meer, 1996). Although plausible, such cross-talk needs to be demonstrated. In addition, a possible participation of the truncated TfdT protein cannot be fully discarded, because the binding of the TfdT protein to the tfdT/C intergenic region has been demonstrated (Leveau & van der Meer, 1996). It is important to note that the tfdT gene of strain JMP134 has a significant level of identity with the tfdT gene of the tfd gene cluster of Burkholderia sp. NK8 (Liu et al., 2001). This TfdT protein responds to chloromuconates, chlorocatechols and even chlorobenzoates, an inducer profile which is very different from other LysR-type regulators involved in (chloro)catechol catabolism. This makes possible that the TfdT protein from strain JMP134 may play an unexpected role in regulation of tfd gene expression.
Which are the inducers involved in the LysR-type gene mediated regulation of the tfd genes? By analogy with the catechol pathway, the corresponding (chloro)muconates may be the inducers for the chlorocatechol ortho ring-cleavage pathway. This has been clearly shown by Chakrabarty's lab for clc genes and cbn genes (McFall et al., 1997c; Ogawa et al., 1999). Their studies have proven that ClcR can bind two sequences in the clcA upstream region, and that this binding is modified by the presence of 2-chloromuconate and muconate. This shift in binding of the ClcR protein to clcA DNA is necessary for transcriptional activation. Such studies have also shown that the CatR protein interacts in a slightly different way than the CIcR protein with the catA promoter region (it can bind to three sequences, in a dimer/tetramer fashion). The CatR protein only interacts with muconate (McFall et al., 1997b) and the CatR-mediated transcriptional activation is not repressed by fumarate (McFall et al., 1997a), as has been reported for the ClcR protein.
Little is known about the tfd genes system. A genetic approach was used in order to identify the inducer of tfd gene expression (Filer & Harker, 1997). Because only mutants that were blocked in the tfdDI and tfdEI genes showed an increased level of induction in the presence of 2,4-D – whereas mutants blocked in tfdA, tfdBI or tfdCI genes did not – 2,4-dichloromuconate was suggested as the inducer. Considering the presence of a second tfd gene module, these observations are difficult to explain, inasmuch as only mutants blocked in the tfd-I genes were used in this work. In fact, the TfdCII protein (Plumeier et al., 2002) should allow further metabolism of 3,5-dichlorocatechol in the tfdCI mutant. However, as the relative contribution of the first three enzymes from the tfd-I gene cluster is two to five times higher than those from the tfd-II gene cluster (Perez-Pantoja et al., 2000), the overall balance may favor the functions encoded by tfd-I gene cluster. This explanation may also be useful to understand why transposon mutagenesis of tfd-I genes produced an accumulation of catabolic intermediates (Don et al., 1985). In this context, the possibility that the TfdR protein interacts with 2,4-dichloromuconate to control the tfdA gene expression is interesting, but has not yet been proved.
The genetic background of C. necator JMP134 is clearly important for an adequate tfd gene expression. Interestingly, 26 out of 28 different betaproteobacterial strains harboring pJP4 grow efficiently on 2,4-D, whereas 17 out of 20 different alpha- or gammaproteobacterial strains harboring pJP4 do not (D. Pérez-Pantoja, B. González, unpublished data). This observation suggests that tfd gene expression clearly requires chromosomally encoded functions that are usually present in Betaproteobacteria, but absent in alpha- or gammaproteobacterial strains.
The pJP4 plasmid
The broad-host, conjugative, IncPβ plasmid pJP4, has recently been sequenced (Trefault et al., 2004). The tfd genes are flanked by two IS1071 elements in the well-conserved IncPβ backbone. IS1071 elements have been associated with several gene clusters that encode the degradation of anthropogenic compounds (Sota et al., 2006). ISJP4 elements (Leveau & van der Meer, 1997) and a complex transposon, Tn5504 (Trefault et al., 2004), are also present in this catabolic plasmid. There are about 20 other ORFs in the non IncPβ backbone region of the plasmid, but none of them is directly involved in aromatic compound degradation (Trefault et al., 2004).
Several genetic changes in the pJP4 plasmid have been reported; for example, 40 kb deletions, including the tfd genes region, have been observed after transposon mutagenesis with Tn1771 (Don et al., 1985). Tn10-stimulated pJP4 deletions in a noncatabolic region have been also described (Clement et al., 2000). Early work described large genetic changes in the pJP4 plasmid, after transfer of the pJP4 plasmid to P. putida and a selection for growth on 3-CB (Ghosal et al., 1985). Electron microscopy, Southern analyzes and the restriction enzyme profile of the obtained pJP4 plasmid derivative (pYG2) clearly suggested that a c. 15 kb deletion, along with c. 25 kb duplication had taken place. About 15 years later, another pJP4 plasmid derivative, pJP4-F3 – obtained by subculturing C. necator JMP134 in liquid cultures containing 3-CB – appeared to undergo the same rearrangement as the pYG2 plasmid. The molecular characterization of pJP4-F3 plasmid showed (Clement et al., 2001) that the 15-kb deletion is flanked by the tfdS/tfdR region and the IS1071 insertion sequence. This implies a loss of the tfdA gene and explains the inability to grow on 2,4-D that is observed in strains harboring pYG2 or pJP4-F3 plasmids. The duplication produced a 21-kb inverted repeat that included the tfdS/R region and all the tfd genes, except tfdA. A plausible model for this pJP4 plasmid rearrangement has been proposed (Clement et al., 2001): it consists of an intermolecular, double crossover, homologous recombination. Some pJP4 plasmid features predicted by this model, as the presence of two sequences related to the IS1071, were later confirmed by the complete sequencing of the plasmid (Trefault et al., 2004). The main features of this rearrangement on pJP4 plasmid, at the molecular and plasmid population level, have been recently studied using a multiple-PCR approach (Larrain-Linton et al., 2006). It is clear that, in wild-type populations, the pJP4 plasmid form is the most abundant, but that at least two recombinant forms – pJP4-F3 and pJP-FM – are also detectable (around 1% of plasmid population). Successive transfers of the wild type in 3-CB strongly select cells harboring the pJP4-F3 plasmid form, which is the derivative with a higher tfd gene dosage (Clement et al., 2001), but this enrichment can be reversed after transfers in 2,4-D. In this context, it is worth mentioning that a C. necator JMP134 derivative with the integrated, one copy per cell, pJP4 plasmid, does not grow on 3-CB (Trefault et al., 2002). Given that the pJP4 plasmid forms appear in about five copies per cell (Trefault et al., 2002), the presence of only one copy can be detrimental for growth on chloroaromatic compounds. Two is a threshold number of copies that allows the clc element to support growth on chlorobenzene (Ravatn et al., 1998), which suggests that small differences in gene dosage may have a critical effect on growth properties. As indicated in ‘Regulation of the tfd genes’, the avoidance of chlorocatechol and chlorophenol toxicity requires an adequate balance between the producing and the transforming activities of chlorocatechol and chlorophenol (Perez-Pantoja et al., 2003; Ledger et al., 2006).
A model for the pJP4 plasmid evolution has recently been proposed (Trefault et al., 2004); it suggests the simultaneous acquisition of both tfd gene clusters in the IncPβ backbone of the pJP4 precursor (Trefault et al., 2004). Several lines of evidence support this possibility. Among them, the fact that only simultaneous acquisition of both tfd gene clusters allows the bacterium to grow efficiently on 3-CB or 2,4-D, avoiding toxicity problems. The unique tfd gene organization in pJP4 may have arisen from the complementation between the two tfd gene modules and the concerted regulation commanded by the activity of the TfdR protein.
Cupriavidus necator possesses 11 of the 12 main routes for aromatic degradation reported in Proteobacteria; the only one absent is the homoprotocatechuate pathway. Functional redundancy seems to be a key feature in the ability of strain JMP134 to degrade a significant number of different aromatic compounds. Redundant functions were observed in the catechol, protocatechuate, salicylate and phenylacetyl-CoA pathways; in the degradation of benzoate and chloroaromatic compounds; in some of the 4-hydroxybenzoate and (methyl)phenols peripheral reactions; and in the presence of several meta ring-cleavage enzymes and other oxygenases, maleylacetate reductases and regulatory proteins. Interestingly, the genome of C. necator encodes more than 70 oxygenases (Table 6) that belong to the main oxygenase groups reported for aromatic compound catabolism. The systematic metabolic reconstruction work reviewed here could only assign functions to half of these oxygenases. Whether the unknown oxygenase functions are involved in aromatic compounds degradation remains to be studied.
Table 6. Oxygenases related to catabolism of aromatic compounds encoded in the genome of Cupriavidus necator JMP134
Is the high catabolic versatility of strain JMP134 common to other soil bacteria? Genome-wide studies performed on P. putida KT2440 (Nelson et al., 2002); B. xenovorans LB400 (Chain et al., 2006), Rhodococcus sp. strain RHA1 (McLeod et al., 2006), and ‘A. aromaticum’ sp. EbN1 (Rabus et al., 2005), show a similar level of catabolic versatility; this clearly suggests that this kind of bacteria may be more common in nature than previously expected. Are most of these catabolic gene clusters evolutionary remnants, or an adaptative improvement in the constant competition for limited carbon sources in natural habitats? A plausible assumption would be that bacteria, such as strain JMP134, were naturally exposed to mixtures of different aromatic compounds (in low amounts), that could be used as carbon sources; for example, during microbial degradation of lignin, exposure to plant exudates or in environments polluted with aromatic compound mixtures. If such was the case, would all the catabolic genes be coordinately expressed? In other words, are aromatic compounds hierarchically degraded? Are there catabolic misroutings when mixtures of aromatic compounds, metabolized by different pathways, are being used as carbon sources? Studies of the gene expression profile using catabolic DNA microarrays will clearly help to answer such questions. Obviously, classical biochemical work on key enzymes would be needed to fully understand the catabolism of aromatic compound mixtures. The analysis of protein expression profiles would also help to compare the relative contribution of different gene clusters during growth on mixtures of aromatic compounds.
This work has been funded by the FONDECYT grants 1030493, and 1070343, the ‘Millennium Institute for Fundamental and Applied Biology’, the ‘Millennium Nucleus in Microbial Ecology and Environmental Microbiology and Biotechnology’ and the ICA4-CT-2002-10011 (ACCESS) Contract of the European Union. Joint Genome Institute work was performed under the auspices of the United States Department of Energy's Office of Science, Biological and Environmental Research Program and by the University of California, Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48, Lawrence Berkeley National Laboratory under contract No. DE-AC02-05CH11231 and Los Alamos National Laboratory under contract No. DE-AC02-06NA25396. D.P.-P. is a CONICYT-DAAD Ph.D. fellow. R.D.I. is a MECESUP Ph.D. fellow.