Isolation of mutants unable to catabolize tyramine
The degradative route required for the aerobic assimilation of Tyn was analysed by bacterial mutagenesis with the transposon Tn5. This approach allowed the isolation of a series of P. putida U mutants that, when cultured in a chemically defined media (MM), were unable to assimilate Tyn as the sole carbon source. The 12 mutants were classified in three different groups. Class 1 consisted of three mutants unable to catabolize Tyn but which grew well when cultured in MM containing 4HPA or 3,4HPA. The two mutants of class 2 were unable to catabolize Tyn and 4HPA but could grow in MM supplied with 3,4HPA. The seven mutants of class 3 were unable to grow in MM containing Tyn, 4HPA or 3,4HPA as the sole carbon source (Fig. 3). Furthermore, when the catabolism of DA was examined, class 2 mutants grew efficiently in a chemically defined medium containing DA as the sole carbon source, whereas class 1 and class 3 mutants did not (Fig. 3). These data suggest that the degradation of Tyn and DA involves several common steps that constitute a convergence pathway for the assimilation of different structurally related compounds: the amines Tyn and DA, the aromatic acids 4HPA and 3,4HPA as well as all molecules (e.g. the important neurotransmitter DOPA) that generate many of these acids as catabolic intermediates. Moreover, the fact that class 2 mutants were able to degrade DA and 3,4HPA but were unable to assimilate 4HPA and Tyn suggests that the mutated gene is the one that normally encodes the enzyme responsible for the hydroxylation of 4HPA to 3,4HPA (Fig. 1). These data revealed that the catabolic pathways involved in the degradation of Tyn, DA, 4HPA and 3,4HPA form a complex catabolic unit, henceforth referred to as the 3,4HPA catabolon, that contains 3,4HPA as the central intermediate (Fig. 1) and includes a convergence pathway for the assimilation of different, but structurally related compounds.
Figure 3. Growth (A540) of P. putida U (wild type) and representative mutants belonging to class 1, class 2 or class 3 when cultured in a chemically defined medium (MM) containing 2-phenylethylamine (▴), tyramine (●), dopamine (□), 4HPA (○) or 3,4HPA (▪). All of the mutants were able to catabolize histamine. The absorbance value measured in the cultures of class 3 mutants when grown in MM + DA or 3,4HPA is due to the colour resulting from the oxidation of these compounds.
Download figure to PowerPoint
Furthermore, the fact that all of the mutants efficiently assimilated 2-phenylethylamine and histamine suggests that neither the deamination step nor the following catabolic steps involved in the assimilation of these two amines are related with those required for the degradation of Tyn and DA.
Genetic organization and functional analysis of the Tyn pathway in P. putida U
Identification of the Tn5 insertion point in the three classes of mutants unable to catabolize Tyn revealed that in each one the transposon disrupted a distinct ORF within a 25 132 bp DNA fragment. This fragment contains 21 genes organized in consecutive divergent operons (Fig. 4). Thus, in the three mutants belonging to class 1 (unable to catabolize Tyn and DA but able to grow in MM containing 4HPA or 3,4HPA, Fig. 3), the Tn5 transposon knocked out the tynA gene (in two different positions) whereas in the third mutant it was inserted in the tynB gene (Fig. 4). In the seven class 3 mutants (unable to grow in MM containing Tyn, DA, 4HPA or 3,4HPA as sole carbon sources, Fig. 3) the transposon was inserted in genes belonging to the hpa cluster (Fig. 4), a locus that contains the 13 genes required for 4HPA assimilation in P. putida U and in other bacteria (Arunachalam et al., 1992; Olivera et al., 1994; Prieto et al., 1996). Finally, in the two class 2 mutants (unable to catabolize Tyn and 4-HPA but able to grow in MM supplied with DA and 3,4HPA) the transposon had integrated in the genes hpaB and hpaC (Fig. 4), which encode two different subunits of the hydroxylase required for the synthesis of 3,4HPA from HPA in other bacteria (Arunachalam et al., 1992; Prieto et al., 1996).
Figure 4. Schematic representation of the tyn and hpa clusters, encoding all of the enzymes required for the transformation of tyramine, dopamine, 4HPA and 3,4HPA into general metabolites (pyruvic and succinic semialdehyde). Different genetic constructions (pKtyn-1, pKtyn-2, pKtyn-4 and pKtyn-6) used for transforming E. coli and other bacterial strains are also indicated. The insertion point of trasposon Tn5 in the different mutants is indicated.
Download figure to PowerPoint
Sequence analysis of the genes flanking those interrupted by the transposon in the different mutants revealed that the genetic information required for the transformation of Tyn, DA, 4HPA and 3,4HPA into succinic semialdehyde and pyruvic acid in P. putida U is grouped together in a DNA fragment (Fig. 4) that is lacking in other closely related species, such as P. putida KT2440, which, accordingly, is unable to catabolize these aromatic compounds. Further studies (see below) revealed that the genes tynABFECGRD, spanning 12 338 bp, are needed for the degradation of Tyn and DA to 4HPA and 3,4HPA, respectively, whereas hpaRBCIHXFDEG2G1AY, spanning 12 722 bp, encodes enzymes in the pathway that transforms 4HPA into general catabolites.
TynA. The tynA gene (Table 1 and Fig. 4), interrupted in P. putidaΔtynA::Tn5 mutants 1 and 2, encodes a protein (TynA) that shows high per cent identity with putative FAD oxidoreductases involved in deamination reactions and identified in numerous species of microorganisms. However, surprisingly, the per cent identity between TynA and tyramine oxidases described in Micrococcus luteus, Klebsiella aerogenes and Salmonella tiphymurium (Okamura et al., 1976; Murooka and Harada, 1981; Roh et al., 2000) was very low (lower than 18% identity), in spite of the fact that the enzymes of all these species contain a well-conserved sequence, identified as a FAD-binding domain, at their amino-termini (Wierenga et al., 1986). Furthermore, the C-terminal pentapeptide SGGCY, involved in the covalent binding of FAD in the aminooxidases of different organisms (human, bovine, Aspergillus niger and Micrococcus rubens) (Hiro et al., 1996), was not found in the TynA sequence.
Table 1. Main characteristics of the tyn genes.
|Gene||Gene product||Size (aa)||Mw||Closest identities|
|tynA||A Subunit of tyramine oxidase||431||46.2 kDa||97% FAD-dependent oxidoreductase (Pseudomonas putida GB-1, ABY99409.1). 77% oxidoreductase (Erwinia tasmaniensis, CAO98358.1). 71% FAD-dependent oxidoreductase (Sphingomonas wittichii RW1, ABQ67955.1). 71% FAD-dependent oxidoreductase (Marinomonas sp. MWYL1, ABR69396.1)|
|tynB||B Subunit of tyramine oxidase||379||41.19 kDa||97% ornithine cyclodeaminase (P. putida GB-1, ABY99407.1). 81% predicted ornithine cyclodeaminase (E. tasmaniensis, CAO98360.1). 79% ornithine cyclodeaminase/mu-crystallin (Marinomonas sp. MWYL1, ABR69398.1). 78% FAD-dependent oxidoreductase (S. wittichii RW1, ABQ67953.1)|
|tynF||Efflux transporter||405||41.5 kDa||95% Bcr/CflA subfamily drug resistance transporter (P. putida GB-1, ABY99406.1). 57% multidrug transport protein (MFS superfamily) (Pseudomonas entomophila L48, CAK15856.1). 51% putative drug resistance transporter (Bradyrhizobium sp. BTAi1, ABQ36731.1)|
|tynC||4-Hydroxyphenylacetaldehyde dehidrogenase||495||52.38 kDa||96% aldehyde dehydrogenase (P. putida GB-1, ABY99404.1). 87% aldehyde dehydrogenase (P. entomophila L48, CAK15874.1). 63% phenylacetaldehyde dehydrogenase (Burkholderia oklahomensis EO147, ZP_02360828)|
|tynE||Putative 4-hydroxyphenyl acetaldehyde dehydrogenase cooperative protein||414||44.74 kDa||97% hypothetical protein PputGB1_3514 (P. putida GB-1, ABY99405.1). 80% hypothetical protein PSEEN3112 (P. entomophila L48 CAK15879.1). 57% hypothetical protein (P. putida U, ABR57206.1)|
|tynG||Amino acid permease||499||53 kDa||99% amino acid permease-associated region (P. putida GB-1, ABY99403.1). 93% amino acid permease (P. entomophila L48, CAK15871.1). 76% amino acid permease family protein CC0426, putative (Pseudomonas fluorescens Pf-5, AAY92485.1)|
|tynR||Transcriptional activator tyn||313||34.45 kDa||97% transcriptional regulator, AraC family (P. putida GB-1, ABY99401.1). 78% transcriptional regulator, AraC family (P. entomophila L48, CAK15869.1). 54% transcriptional regulator, AraC family (P. fluorescens Pf-5, AAY92483.1)|
|tynD||Putative tyramine oxidase||432||48 kDa||94% FAD-dependent oxidoreductase (P. putida GB-1, ABY99399.1). 90% oxidoreductase; putative aromatic-ring hydroxylase (P. entomophila L48, CAK15867.1). 65% tyramine oxidase (Klebsiella aerogenes, BAD88607.1)|
In order to establish whether the inability of these two mutants to grow in MM containing Tyn was not due to a polar effect caused by the presence of Tn5 but instead reflected the absence of TynA, the tynA gene was polymerase chain reaction (PCR) amplified, cloned into the plasmid pBBR1MCS-3 (Tcr) (hereon abbreviated as pMC), and the resulting construct used to transform class 1 mutants 1 and 2 (P. putidaΔtynA::Tn5). The recombinant strains (P. putidaΔtynA::Tn5 pMCtynA) were able to assimilate Tyn whereas those transformed with the plasmid pMC without insert (P. putidaΔtynA::Tn5 pMC) could not (Fig. S1A). Thus, in silico and genetic analyses confirmed that TynA is a FAD oxidoreductase directly involved in the deamination of Tyn and that expression of pMCtynA restores the capacity to assimilate Tyn to class 1 mutants carrying a non-functional tynA gene.
TynB. In the third mutant belonging to class 1, the transposon had affected a gene located upstream of tynA (see Fig. 4). Analysis in silico revealed that TynB (Table 1) is a Rossmann folded FAD/NAD(P) amino oxidase. Sequence analysis showed that TynB is very similar to the ornithine cyclodeaminases of other microbes (Costillow and Laycock, 1971; Sans et al., 1987), suggesting that, like these other enzymes, it could be involved in the formation of the imine intermediate, required for the oxidation to 4-hydroxyphenylacetaldehyde. Similar results were obtained when used a mutant obtained by homologous recombination of a internal fragment of the gene cloned into pK18::mob (P. putidaΔtynB::pK18::mob, see Experimental procedures and Table S1). The recombinant strain was also unable to assimilate Tyn (Fig. S1B), showing that TynB is directly involved in Tyn catabolism. However, it can be argued that tynA and tynB belong to a single operon (tynAB), and thus, the presence of either the transposon or plasmid pK18::mob in tynB could have prevented the assimilation of Tyn by preventing the expression of tynA. To clarify whether tynB is essential for tyramine degradation, tynA, tynB and tynAB were cloned into plasmid pMC, and used to transform P. putidaΔtynB::Tn5. The recombinant strains P. putidaΔtynB::Tn5 pMCtynA and P. putidaΔtynB::Tn5 pMCtynB were still unable to degrade Tyn whereas in strain P. putidaΔtynB::Tn5 pMCtynAB this catabolic function was fully restored (Fig. S1C). Similar results were obtained when the mutant P. putidaΔtynB::pK18::mob was transformed with pMCtynB or pMCtynAB. These results suggest that tynA and tynB transcribe together as an operon (tynAB) and that both enzymes (TynA and TynB) are required for Tyn catabolism. The fact that these enzymes showed high similarities with other proteins involved in the oxidative deamination of several organic compounds (see genome projects and Costillow and Laycock, 1971; Sans et al., 1987) suggests that TynAB constitutes a functional complex that catalyses the synthesis of 4-hydroxyphenylacetaldehyde from Tyn in P. putida U. Multienzymatic complexes such as the TynAB amino oxidase are not unusual in microbes, and similar complexes have been described for the degradation of primary amines (both aliphatic and aromatic) (Arias et al., 2008).
Since among the three mutants blocked in the catabolism of Tyn only tynA and tynB were affected by the transposon, it could be argued that this genetic information is enough to convert Tyn to 4HPA. This hypothesis was tested using the genetic construct pMCtynAB to transform Escherichia coli W14, a strain that contains the catabolic pathway required for the assimilation of 4HPA (the hpa cluster) (Prieto et al., 1996) but which is unable to grow in chemically defined medium containing Tyn as the sole carbon source. The results showed that the recombinant strain E. coli W14 pMCtynAB (see Table S1) was still unable to degrade Tyn (Fig. 5), suggesting that other genes, probably those linked to tynAB, are also required for its catabolism. Based on this line of reasoning, the genes located upstream of tynA and tynB were analysed.
Figure 5. Growth (A540) of E. coli W14 pMC (○), E. coli W14 pK18::mob (●), E. coli pMCtynAB (▴), E. coli W14 pKtyn-2 (▪), E. coli W14 pKtyn-2 pMCtynR (▵) and E. coli W14 pKtyn-4 (□) when cultured in MM containing tyramine as the sole carbon source (5 mM) at 30°C and 37°C. Tyramine consumption when pKtyn-2 pMCtynR (red bars) and E. coli W14 pKtyn-4 (blue bars) were cultured in MM with Tyn (5 mM).
Download figure to PowerPoint
TynF. A different ORF, tynF, was identified upstream of tynB (Fig. 4, Table 1). Sequence analysis revealed that TynF shows homology with proteins belonging to the MFS superfamily, which include the more common membrane transporters (Pao et al., 1998). Furthermore, structural studies showed that TynF, like other members of the MFS superfamily, contains 12 putative transmembrane regions, suggesting that it is involved in a transport mechanism of Tyn or a derivative. To establish its functional role, tynF was inactivated using a genetic construct containing a PCR-amplified internal fragment of the gene cloned into the plasmid pK18::mob. The recombinant strain P. putidaΔtynF::pK18::mob (Table S1) was able to grow in MM containing Tyn as the sole carbon source, and did so at the same rate that the wild-type (P. putida U). Thus, either TynF is not needed for Tyn degradation or the protein is not essential for its assimilation, implying that the enzymatic function can be replaced by other cellular enzymes. Since most TynF-related proteins are transporters commonly involved in drug efflux (Pao et al., 1998), it could be argued that TynF has a similar function. Pseudomonas putida U, like other phylogenetically related species, contains redundant efflux systems assuring the prevention of metabolic damage caused by the accumulation of intermediates that are toxic or otherwise detrimental to this species. Thus, in the absence of TynF, P. putida U could use, for example, the efflux system involved in protecting against catabolites generated from 2-phenylethylamine, another aromatic amine that can be efficiently used as carbon and nitrogen source by this bacterium (Arias et al., 2008).
The fact that disruption of tynF did not affect Tyn assimilation by P. putidaΔtynF::pK18::mob also indicates that tynAB was expressed under these conditions.
TynC and TynE. The gene tynC encodes a putative aldehyde dehydrogenase (ALDH) that is similar to members of the NAD-dependent ALDH family (see Table 1). TynC contains an Aldedh conserved domain (amino acids 26–491) present in dehydrogenases with aldehyde substrates as well as a NAD/NADP-binding motif (GIGREFG, positions 472–478) corresponding to ALDHs that catalyse their respective reactions irreversibly (Arias et al., 2008). Furthermore, other motifs present in bacterial ALDHs were also identified. One of them, LELGGKNAG, was localized to amino acid positions 267–274 and another, FLHSGQICAAGE, to amino acids 295–306. These sequences contain the glutamic acid and cysteine residues present at the active site of these enzymes (Weretilnyk and Hanson, 1990).
To determine whether TynC is involved in Tyn degradation, its encoding gene, tynC, was disrupted using a single recombination procedure in which an internal fragment of the gene was cloned into the suicide plasmid pK18::mob. The recombinant strain containing the disrupted tynC (P. putidaΔtynC::pK18::mob, see Table S1) was able to grow in MM containing Tyn although with a delayed growth rate. This effect was not observed when the carbon source was 4HPA (Fig. S2). The fact that, despite the inactivation of tynC, this mutant was still able to catabolize Tyn suggests that TynC was functionally replaced, although with a lower efficiency, by another ALDH (e.g. PeaE or PedI) involved in the oxidation of phenylacetaldehyde generated from 2-phenylethylamine or 2-phenylethanol in P. putida U (Arias et al., 2008). Moreover, the delayed growth rate (Fig. S2) suggests either that 4-hydroxyphenylacetaldehyde is not used by these ALDHs at the same rate as their natural substrates, or that the genes encoding these alternative enzymes belong to other clusters and their expression was not optimal.
Downstream of tynC, another gene, tynE, was identified. The encoded protein (Table 1) shows a high percentage of identity with the products of genes usually adjacent to those encoding ALDHs. This protein contains an Aldedh consensus sequence GXGXXXG (GNGVADG in TynF, amino acids 135–141) involved in NAD(P) binding. These results suggest that TynE could have a cooperative function during catalysis, perhaps facilitating the binding of NAD(P) to the ALDH. Bearing in mind that the oxidative degradation of many aromatic compounds requires the participation of multienzymatic complexes (Diaz et al., 2001; Benedetti et al., 2007; Arias et al., 2008), it could be argued that TynC and TynE could constitute an oxidative complex involved in the oxidation of the 4-hydroxyphenylacetaldehyde generated by the deaminating complex TynAB, but deeper studies should be performed in order to clarify this point.
Although the mutant disrupted in tynE (P. putidaΔtynE::pK18::mob, see Table S1) was able to catabolize Tyn, conserved location of the gene (always adjacent to those encoding ALDHs), the presence of sequences typical of ALDHs, as well as the possible participation of TynE-like proteins in the degradation of different primary amines (2-phenylethylamine, propyl-, butyl- and pentylamine) (Arias et al., 2008) suggest that it could be involved in Tyn catabolism. However, it is not likely to be essential since, as noted above, it could be functionally replaced by other enzymes.
Two transcription attenuators-like loops are located upstream of tynC and downstream of tynE (see Fig. 4), suggesting that: (i) both genes constitutes an operon and (ii) TynC and TynE could constitute a single functional unit.
To define the minimal amount of genetic information required to catalyse the oxidative deamination of Tyn, a construct containing all of the genes (tynABFEC) was generated (pKtyn-2, see Fig. 4), cloned into plasmid pK18::mob and used to transform E. coli W14. The recombinant strain E. coli W14 pKtyn-2 (see Table S1) was unable to grow in chemically defined medium containing Tyn as the sole carbon source, suggesting that, at least in this bacterium, additional gene(s) is/are needed to transform this aromatic amine into 4HPA, a compound that can be efficiently catabolized by this strain (Fig. 5). For this reason, other ORFs belonging to the same piece of DNA were analysed (Fig. 4).
TynG. Upstream of tynC and divergently transcribed, a gene denominated tynG could be observed (Fig. 4, see Table 1). The encoded protein has a high identity with different active transporters. In silico analysis showed that TynG potentially contains 12 transmembrane regions (as reported in other bacterial permeases involved in amino acid uptake) that extend throughout the protein (amino acids 16–466) (Vandenbol et al., 1989).
To analyse the function of TynG, an internal fragment of tynG was used to disrupt the wild-type gene in P. putida U. The recombinant strain (P. putidaΔtynG::pJQ200KS, see Table S1) retained the ability to catabolize Tyn since it grew as the wild type in MM containing this amine as the sole carbon source. Thus, either TynG is not involved in Tyn catabolism or it is not essential because it could be functionally replaced by other permeases, which are located at different sites in the bacterial chromosome. This finding in turn implies that inside the tyn cluster there are two different ORFs (tynF and tynG) likely encoding transport proteins. A double mutant (P. putidaΔtynF::pK18::mobΔtynG::pJQ200KS, see Table S1) was able to grow in MM with Tyn as the sole carbon source, suggesting that, at least in P. putida U, neither TynF nor TynG is essential for Tyn catabolism.
TynR. The tynR (Table 1, Fig. 4) gene was identified downstream of tynG. TynR shows high homology with transcriptional activators belonging to the AraC family (Gallegos et al., 1997). TynR contains a DNA-binding region, characteristic of the proteins belonging to this family, located between positions 227 and 300.
To study the implication of this regulator in Tyn catabolism, tynR was mutated using a internal fragment of the gene cloned into plasmid pK18::mob. The recombinant strain (P. putidaΔtynR::pK18::mob, see Table S1) was unable to grow in MM containing Tyn as the sole carbon source, whereas it did grow well in the same medium containing 4HPA (see Fig. S2). These results revealed that TynR is essential for Tyn catabolism but does not participate in the degradation of 4HPA.
The expression TynR-dependent genes was studied by transforming the mutant P. putidaΔtynR::pK18::mob with constructs inserted into the plasmid pMC. The constructs consisted of the genes tynA (pMCtynA), tynB (pMCtynB), tynAB (pMCtynAB), and tynC and tynE (pMCtynCE). The recombinant strains (P. putidaΔtynR::pK18::mob pMCtynA; P. putidaΔtynR::pK18::mob pMCtynB; P. putidaΔtynR::pK18::mob pMCtynAB; and P. putidaΔtynR::pK18::mob pMCtynCE, see Table S1), which expressed the genes tynA, tynB or tynC under the control of the plasmid promoter, were unable to grow in MM + Tyn, suggesting that TynR is needed for the expression of genes belonging to the tyn cluster (or at least for the coordinate expression of tynA, tynB, tynC and tynE). To prove this assumption, the construct pKtyn-2, containing the genes tynABFEC (Fig. 4), was used to transform E. coli W14. The recombinant strain, E. coli W14 pKtyn-2, was unable to grow in MM + Tyn either at 30°C or at 37°C. However, in trans expression in this bacterium of a plasmid carrying tynR conferred on this strain (E. coli pKtyn-2 pMCtynR) the ability to assimilate Tyn as the sole carbon source at either temperature (Fig. 5), although at 37°C the growth rate was greatly delayed (this effect is analysed below). Thus, we can conclude that tynR is essential in P. putida U and in recombinant E. coli W14 and encodes a protein needed for the degradation of this aromatic amine. Furthermore, when E. coli W14 was transformed with a single construct (pKtyn-1) containing the genes tynABFECGR inserted into the plasmid pK18::mob (Fig. 4), the resulting strain (E. coli W14 pKtyn-1, Table S1) grew efficiently in MM + Tyn both at 30°C and at 37°C, although, as noted above, growth at 37°C was considerably delayed (Fig. 5). Concluding, all these results suggest that TynR may act as the transcriptional activator needed for the expression of tyn genes.
TynD. Another ORF, tynD (Table 1), was identified downstream of and divergently transcribed to tynR, and adjacent to the cluster hpa (containing all of the genes involved in the catabolism of 4HPA, Fig. 4). TynD showed high homology with FAD oxidoreductases showing a βαβ consensus sequence present in the amino-terminus that seems to be associated with a FAD-binding domain (Wierenga et al., 1986). Furthermore, a comparison of TynD and TynA showed that despite their low percentage of identity (17%) both proteins have maintained the structural domain characterizing enzymes containing FAD as a prosthetic group (Dailey and Dailey, 1998; Roh et al., 2000).
In order to establish the functional role of TynD, its gene, tynD, was disrupted using a pJQ200KS construction containing an internal fragment of the gene (Table S1). The resulting mutant (P. putidaΔtynD pJQ200KS) grew well in MM + Tyn, suggesting that tynD, in contrast to tynA and tynB, is not essential for Tyn catabolism in P. putida U. These results agree well with those obtained in E. coli W14 transformed with the construct pKtyn-1 (containing the genes tynABFECGR), since, as indicated above, the recombinant strain (E. coli W14 pKtyn-1) was able to catabolize Tyn. It therefore appears that pKtyn-1 contained all the genetic information needed to transform Tyn into 4HPA. Moreover, when E. coli W14 pKtyn-1 was cultured in MM + Tyn at 37°C, bacterial growth was considerably delayed compared with cultures grown at 30°C (Fig. 5). It could be argued that this difference was due to the fact that these genes come from P. putida U, a bacterium with an optimal growth temperature of 30°C and which does not grow, or does so very poorly, at 37°C. However, when E. coli W14 was transformed with pKtyn-4 (containing all the tyn genes, Fig. 4), the recombinant strain (E. coli W14 pKtyn-4) grew well in MM + Tyn at 30°C and at 37°C (Fig. 5), suggesting that TynD is only required for growth at 37°C. As in preceding analyses it was concluded that while TynD is not needed for Tyn degradation in P. putida U, this enzyme can functionally replace tyramine deaminase (TynA) in E. coli. To test this hypothesis, E. coli W14 was transformed with a pK18::mob construct (pKtyn-6) carrying the genes tynBFECGR. The recombinant strain (E. coli W14 pKtyn-6), unable to catabolize Tyn (since it lacked the essential tynA), was transformed with plasmid pMC carrying the gene tynD (pMCtynD). This new recombinant strain (E. coli W14 pKtyn-6 pMCtynD, see Table S1) which expressed the protein TynD in trans, was also unable to grow in MM + Tyn, neither at 30°C nor at 37°C. Therefore, TynD, reportedly a tyramine oxidase in other microbes (Okamura et al., 1976; Murooka and Harada, 1981), could not functionally replace TynA. However, when E. coli W14 pKtyn-6 was transformed with the plasmid pMCtynA, the recombinant E. coli strain W14 pKtyn-6 pMCtynA grew efficiently at 30°C but very poorly at 37°C (data not shown). Taken together, the data suggest a model in which the tyramine deaminase of P. putida U could be organized as an enzymatic complex TynABD, with TynAB able to catabolize the Tyn deamination at 30°C but only very slowly at 37°C. At the higher temperature, a third protein, TynD, is needed to assure the efficiency of the catalytic process.
These above results demonstrate that this catabolic pathway can be successfully transferred to other microbes, as was achieved by transforming E. coli W14 with the genetic constructs pKtyn-1 (pK18::mobtynABFECGR) and pKtyn-4 (pK18::mobtynABFECGRD) from P. putida U (Figs 4 and 5).
Design of a new catabolic pathway by metabolic engineering
As indicated above, many different microbes are known to accumulate tyramine in their growth medium by means of a decarboxylating activity that transforms l-Tyr into tyramine (Rice et al., 1976; Ten Brink et al., 1990), which, due to the absence of a specific catabolic pathway, cannot be further catabolized. In contrast, P. putida U degrades l-Tyr through the homogentisate pathway (Arias-Barrau et al., 2004; 2005) since it lacks the l-Tyr decarboxylase needed to generate Tyn. In order to enlarge the catabolic potential of P. putida U, the tyrosine decarboxylase gene of Lactococcus lactis IPLA 655 (Fernández et al., 2004), tdcA, was cloned as described in Experimental procedures. This gene was PCR amplified, cloned into the plasmids pGEM®-T Easy and pMC. The resulting construct, pMCtdcA, was used to transform P. putida U as well as mutants deficient in genes encoding the homogentisate pathway (hmgABC), or those encoding the p-hydroxyphenylpyruvate dioxygenase (hpd) and homogentisate pathways (strains P. putidaΔhmgABC and P. putidaΔhpdΔhmgABC respectively). Whereas neither strain was able to grow in MM +l-Tyr (5 mM), the recombinants P. putidaΔhmgABC pMCtdcA and P. putidaΔhpdΔhmgABC pMCtdcA (Table S1) grew well in this medium (Fig. 6), suggesting that tdcA expression enables P. putida U to degrade l-Tyr via tyramine, thus enlarging the catabolic potential of this bacterium and increasing the number of enzymatic steps in the 3,4HPA catabolic shunt (hereon referred to as the 3,4HPA catabolon).
Figure 6. A. Growth (A540) of E. coli W14 pMC (▪), E. coli W14 pMCtdcA (▵), E. coli W14 pKtyn-4 (▴) and E. coli W14 pKtyn-4 pMCtdcA (●) when cultured in MM containing l-Tyr (5 mM) as the sole carbon source. B. Growth (A540) of P. putida U Δhpd ΔhmgABC pMC (□), P. putida U ΔhmgABC (▪), P. putidaΔhpdΔhmgABC pMCtdcA (○) and P. putidaΔhmgABC pMCtdcA (●) when cultured in MM containing l-Tyr (5 mM) as the sole carbon source.
Download figure to PowerPoint
In other experiments, strain E. coli W14, which is unable to catabolize l-Tyr because it lacks the homogentisate pathway or to degrade tyramine due to the absence of the tyn cluster (Arias-Barrau et al., 2004; 2005; and above), was transformed with two genetic constructs, one containing all of the tyn genes (pKtyn-4, Fig. 4) and the other tdcA from L. lactis. The recombinant strain, E. coli W14 pKtyn-4 pMCtdcA, was able to grow in MM containing l-Tyr (5 mM) as the sole carbon source, whereas a strain lacking tyn, tdcA or both (E. coli W14 pMC) could not (Fig. 6). These data showed that metabolic engineering is a very useful strategy to enlarge the degradative potential of many microbial strains, and a powerful tool for designing new microbes with important biotechnological applications, such as amine elimination, detoxification, biotransformation or the catalysis of other chemical processes of interest.
Genetic organization and functional analysis of the Hpa pathway in P. putida U
The hpa cluster is made up of the genes hpaRBCIHXFDEG2G1AY (12 722 bp) and encodes proteins required for the catabolism of 4HPA. It is organized in consecutive operons (hpaBC, hpaG1G2EDF, hpaX, hpaHI) and three additional independent units (hpaA, hpaR and hpaY) (Fig. 4). These genes were found to be highly similar to those involved in the degradation of 4HPA, as reported by other authors (Arunachalam et al., 1992; Olivera et al., 1994; Prieto et al., 1996; Diaz et al., 2001). The functions of the Hpa proteins are described below.
The proteins HpaB and HpaC (Table 2) are the two subunits of an enzymatic complex that hydroxylates 4HPA to 3,4HPA. This functional complex (4HPA hydroxylase) is associated with a NADH-dependent FMN reductase (HpaC) and a monooxygenase (HpaB) that uses reduced FMN and O2 to catalyse the monooxygenation of 4HPA (Fig. 1). Although both enzymes are quite similar to those reported in other microbes (Thotsaporn et al., 2004), HpaB and HpaC from P. putida U have a very low percentage of identity (< 17%) with the ring hydroxylases described in E. coli (Prieto et al., 1996), Klebsiella pneumoniae (Gibello et al., 1997) and Bacillus thermoglucosidasius (Duffner et al., 2000). Whereas HpaBC from E. coli hydroxylates both 4HPA and 3HPA to 3,4HPA, HpaBC from P. putida U is the enzymatic complex responsible of 4HPA hydroxylation for its catabolism. In this strain, a different system catalyses the hydroxylation of 3HPA to 2,5HPA (homogentisic acid), which is catabolized through a different pathway (Arias-Barrau et al., 2005). Thus, the two P. putida U mutants with transposon-interrupted hpaB and hpaC genes (Fig. 3) were unable to degrade tyramine and 4HPA but grew well in medium containing 3,4HPA, dopamine (which leads to 3,4HPA by deamination) or 3HPA (which requires a different hydroxylase for catabolism through the homogentisate pathway). Furthermore, the accumulation of 4HPA by these two mutants when cultured in MM + Tyn and PA (Fig. S3) unequivocally showed that the amine is converted into 4HPA.
Table 2. Main characteristics of the hpa genes.
|Gene||Gene product||Size (aa)||Mw||Closest identities|
|hpaB||Oxygenase component of 4-hydroxyphenylacetic acid hydroxylase||389||42.8 kDa||78% acyl-CoA dehydrogenase (Azotobacter vinelandii AvOP, EAM07573.1). 71% p-hydroxyphenylacetate hydroxylase C2:oxygenase component (Acinetobacter baumannii, AAS75429.1)|
|hpaC||Reductase component of 4-hydroxyphenylacetic acid hydroxylase||309||34.3 kDa||62% Flavin reductase-like (A. vinelandii AvOP, EAM07574.1). 57% p-hydroxyphenylacetate hydroxylase C1:reductase component (A. baumannii, AAS75430.1)|
|hpaD||Homoprotocatechuate 2,3-dioxygenase||307||33.7 kDa||83% homoprotocatechuate 2,3-dioxygenase (Pseudomonas aeruginosa PAO1, AAG07511.1). 71% homoprotocatechuate 2,3-dioxygenase (Bordetella bronchiseptica RB50, CAE31237.1)|
|hpaE||5-Carboxy-2-hydroxymuconate semialdehyde dehydrogenase||486||52.9 kDa||90% 5-carboxy-2-hydroxymuconate semialdehyde dehydrogenase (P. aeruginosa PAO1, AAG07510.1). 84% HpaE (Pasteurella multocida ssp. multocida str. Pm70, AAK03614.1)|
|hpaF||5-Carboxymethyl-2-hydroxymuconate isomerase||134||14.5 kDa||53% 5-carboxymethyl-2-hydroxymuconate delta-isomerase (B. bronchiseptica RB50, CAE31238.1). 42% 5-carboxymethyl-2-hydroxymuconate isomerase (Polaromonas sp. JS666, EAM36812.1)|
|hpaG1||5-Oxo-pent-3-ene-1,2,5-tricarboxilic acid decarboxilase/2-hidroxi-hept-2,4-diene-1,7-dioico acid isomerase||219||23.4 kDa||73% 2-keto-4-pentenoate hydratase/2-oxohepta-3-ene-1,7-dioic acid hydratase (P. aeruginosa UCBPP-PA14, 00137573.1). 44% 5-oxo-1,2,5-tricarboxilic-3-penten acid decarboxilase/isomerase (Escherichia coli, CAA86040.1)|
|hpaG2||5-Oxo-pent-3-ene-1,2,5-tricarboxilic acid decarboxilase/2-hidroxi-hept-2,4-diene-1,7-dioico acid isomerase||250||27.9 kDa||73% 2-keto-4-pentenoate hydratase/2-oxohepta-3-ene-1,7-dioic acid hydratase (P. aeruginosa UCBPP-PA14, 00137573.1). 44% 5-oxo-1,2,5-tricarboxilic-3-penten acid decarboxilase/isomerase HpaG (E. coli, CAA86040.1)|
|hpaH||2-Oxo-hept-3-ene-1,7-dioate hydratase||267||29.2 kDa||85% 2-oxo-hept-3-ene-1,7-dioate hydratase (P. aeruginosa PAO1, AAG07514.1). 73% 2-oxo-hept-4-ene-1,7-dioate hydratase (E. coli, AAB91474.1)|
|hpaI||2,4-Dihydroxyhept-2-ene-1,7-dioic acid aldolase||267||28.3 kDa||79% 2,4-dihydroxyhept-2-ene-1,7-dioic acid aldolase (P. aeruginosa UCBPP-PA14, 00137596.1). 71% 2,4-dihydroxyhept-2-ene-1,7-dioic acid aldolase (Deinococcus radiodurans, AAF12475.1)|
|hpaX||4-Hydroxyphenylacetic acid transporter||435||47 kDa||74% probable MFS transporter (P. aeruginosa PAO1, AAG07513.1). 48% putative permease transmembrane protein (Burkholderia pseudomallei K96243, CAH35014.1)|
|hpaA||4-Hydroxyphenylacetic acid operon regulatory protein||301||34.5 kDa||39% 4-hydroxyphenylacetate 3-monooxygenase operon regulatory protein (Salmonella enterica ssp. enterica serovar Paratyphi A str. ATCC9150, AAV77662.1). 39% regulator of the 4HPA-hydroxylase operon (E. coli, CAA86047.1)|
|hpaR||3,4-Dihydroxyphenylacetic acid operon regulatory protein||140||16.1 kDa||50% 4-hydroxyphenylacetate catabolism (Klebsiella pneumoniae ssp. pneumoniae MGH 78578, CAY49503.1). 49% 4-hydroxyphenylacetate degradative operon repressor (E. coli UMN026, CAR16087.1)|
|hpaY||3,4-Dihydroxyphenylacetic acid operon regulatory protein||140||16 kDa||52% homoprotocatechuate degradation operon regulator, HpaR (S. enterica ssp. enterica serovar Newport str. SL317, EDX48740.1). 50% 4-hydroxyphenylacetate degradative operon repressor (E. coli UMN026, CAR16087.1)|
HpaD (Table 2) is a member of the group III extradiol dioxygenases, which use ortho-dihydroxylated compounds (in this case 3,4HPA, Vetting et al., 2004; Groce and Lipscomb, 2005) as substrates along with O2 to generate the reaction product 5-carboxymethyl-2-hydroxymuconic semialdehyde (CHMS, Fig. 1). HpaD has a well-conserved extradiolic dioxygenase catalytic domain extending between amino acids 6 and 281. It also contains two His residues (His-12, His-57) and an Asp residue (Asp-258) that together constitute the Fe coordination sphere present in all class III oxygenases (Vetting et al., 2004; Groce and Lipscomb, 2005). Furthermore, a His residue present in the catalytic site is also found in HpaD (His-186) from P. putida U.
The fact that the two mutants carrying an interrupted hpaD gene were able to accumulate 3,4HPA when cultured in MM + Tyn (or DA, or 4HPA) and PA (to support bacterial growth) (Fig. S3) indicates that, as expected, HpaD catalyses cleavage of the aromatic ring in 3,4HPA.
HpaE (Table 2) catalyses the oxidation of CHMS to 5-carboxymethyl-2-hydroxymuconic acid (CHM, Fig. 1). It is very similar to the CHMS dehydrogenases described in other microbes (Prieto et al., 1996) and contains a well-conserved NAD-dependent ALDH domains (Adedh, amino acids 5–472), a NAD binding sequence (GYGATAG, positions 199–206), and many other sequences typical of bacterial ALDHs. One such sequence (FSLNGERCTAGS, amino acids 269–280) contains the glutamic acid and cysteine residues belonging to the active site and which are therefore involved in the catalytic process.
HpaF (Table 2) is an isomerase that catalyses the conversion of CHM into 5-oxo-pent-3-en-1,2,5-tricarboxylic acid (OPET, Fig. 1). It also has a CHM isomerase conserved domain between amino acids 2 and 105. The enzyme's catalytic mechanism involves a Pro residue (Subramanya et al., 1996) located at the amino-terminus, while in the P. putida U enzyme this residue is contained in the sequence MPHLV.
HpaG1 and HpaG2 (Table 2) are closely related to different OPET decarboxylases. HpaG1 and HpaG2 contain conserved sequences (positions 11–17 and 43–243) present in all proteins belonging to the fumarylacetoacetate hydrolase family, which also includes OPET decarboxylases and HHDD isomerases using Mn2+ as a cofactor (Prieto et al., 1996; Diaz et al., 2001). Both proteins have a 44% identity with HpcE from E. coli C. The latter enzyme is responsible for the decarboxylation of OPET to 2-oxo-hept-3-en-1,7-dioic acid (OHED) via 2-hydroxy-hept-2,4-diene-1,7-dioic acid (HHDD). HpaG from E. coli owns two different activities (decarboxylase and isomerase) distributed into two different domains (one amino-terminal and the other in the carboxy end) highly similar in sequence (Roper and Cooper, 1993). This, together with the existence of hpaG homologues repeated in tandem in different microbes (Yersinia pestis, K. pneumoniae and P. putida U), suggests that these genes arose through a duplication event, and that, over the course of evolution, they fused in some microbes (e.g. E. coli; Prieto et al., 1996; Diaz et al., 2001), thus generating a single gene which encodes two enzymatic activities. A comparative analysis of HpaG1 and HpaG2 with the amino-terminal and carboxy end of HpcE from E. coli C strongly reinforces this hypothesis.
HpaH (Table 2) catalyses the incorporation of a molecule of water into OHED, thus generating 2,4-dihydroxy-hept-2-ene-1,7-dioic acid (HHED, Fig. 1). This protein presents well-conserved sequences (GHKIG, amino acid residues 59–63; PDYG, 78–81; PRVEVE, 103–108; and DTISDNA, 159–165) found in OHED hydratases and related proteins. Furthermore, some of the Asp residues present in the DTISDNA sequence seem to be involved in cofactor (Mn2+) binding.
The gene hpaI is located between hpaC and hpaH (see Fig. 4, Table 2). HpaI shows a high percentage of homology with HHED aldolases (Stringfellow et al., 1995; Prieto et al., 1996). One such enzyme is the last in the 4HPA meta degradation pathway and catalyses the aldolic breakdown of HHED, releasing succinic semialdehyde and pyruvate (Fig. 1).
HpaX (Table 2) is quite similar to transmembrane transporters (the previously described MFS superfamily, amino acids 32–394), suggesting the involvement of HpaX in the uptake of 4HPA. In one of the class 3 mutants, the transposon was inserted in the hpaX gene. This strain was able to grow in MM + 4HPA and Tyn albeit very slowly (Fig. S4), suggesting that the function of HpaX is related to the catabolism of these compounds but is not essential as in the mutant bacterium it was probably replaced by the activity of other enzymes (Table 2). When this mutant was cultured in MM + Tyn (5 mM), 4HPA was detected in the culture (Fig. S4), which suggested that deamination of Tyn to generate 4HPA could occur within the periplasmic space, as has been described in other bacteria (Roh et al., 2000). The 4HPA formed in the periplasm would then be taken up by the bacteria via HpaX or, in its absence, as in this mutant, via an unknown uptake mechanism or perhaps passively.
The gene hpaA is found upstream of hpaG1 (Fig. 4). The encoded protein shows strong homology with regulators belonging to the above-described AraC family (Table 2, Gallegos et al., 1997; Roh et al., 2000) as well as with a transcriptional activator that modulates expression of the hpaBC operon (containing the genes encoding the two subunits of the enzyme involved in the hydroxylation of 4HPA to 3,4HPA) in E. coli K12 (Prieto et al., 1996; Diaz et al., 2001). In HpaA of P. putida U, the characteristic sequence presents on AraC members, which constitutes a DNA-binding domain which interacts with RNA polymerase to enhance transcription, is located between amino acids 198 and 296 (Gallegos et al., 1997).
HpaR is encoded by hpaR, located between tynD (belonging to the Tyn pathway) and hpaB (involved in the degradation of 4HPA) (Fig. 4). This protein is similar to a repressor present in E. coli, K. pneumoniae, as well as Y. pestis and involved in controlling expression of the 3,4HPA operón (Table 2). Moreover, hpaR handicapped mutants of E. coli constitutively expressed genes comprising the hpaIHXFDEG2G1 cluster (Prieto et al., 1996; Diaz et al., 2001). Structural analyses revealed that HpaR has a HTH motif (amino acids 5–137) involved in DNA binding and conserved in all regulators belonging to the MarR family (Alekshun and Levy, 1999). It is interesting to note that sequencing of the region adjacent to hpaA in P. putida U led to the detection of a second hpaR (hpaY) which, although not identical to the one located upstream of hpaB, has strong sequence identity with it (73%), suggesting a recombination mechanism between two similar genes rather than a duplication event, in which case the two genes would be identical or almost identical. From a functional viewpoint, these two hpaR genes (see Fig. 4), which have not been described in other microorganisms, might assure a stronger control of the expression of genes involved in 3,4HPA catabolism, such that these genes are not expressed until the concentration of inducer (3,4HPA) is very high, at least to compensate the induction of all these proteins.
Comparative analysis of the tyn and hpa loci in different microbes
Comparative analysis of the genetic organization of the tyn and hpa clusters in different microbes is summarized in Fig. 7 and Fig. S5 respectively. Based on a comparison between the sequences of genes belonging to the tyn cluster and those reported in the Genome Projects, it can be concluded that all of the genes are conserved only in P. putida GB-1, whereas some of them are lacking in other microbes (Fig. 7). Specifically, in all the clusters analysed (with the exception of P. putida GB-1) tynE and tynF were absent while tynG was present only in Erwinia tasmaniensis Et1/99 and Marinomonas sp. MWYL1. Furthermore, in all the microorganisms analysed, tynR was adjacent to tynAB but in the opposite orientation. However, in Pseudomonas tynR was separated from tynAB by four additional ORFs (tynFECG) some of which encode proteins potentially involved in the oxidation of 4-hydroxyphenylacetaldehyde. This arrangement suggests that in pseudomonad species the Tyn pathway arose from what originally was a widely conserved cluster (tynRAB) in which a fragment was inserted.
Figure 7. Comparative analysis of the tyn clusters found in different microbes. Arrows with the same colour correspond to genes encoding proteins with a similar function. Colour combinations indicate the organization of these clusters in relation to those reported in P. putida U. The sequences of the different genes were obtained from the Genome Projects and carried the following accession numbers: P. putida GB-1, NC010322.1; Erwinia tasmaniensis Et1/99, NC010694.1; Marinomonas sp. MWYL1, NC009654.1; Sphingomonas wittichii RW1, NC009511.1 and Vibrio vulnificus YJ016, NW001999242.1.
Download figure to PowerPoint
In most microbes, the gene encoding the aldehyde dehydrogenase, tynC, is far removed from tynA and tynB, suggesting that its incorporation into the tyn cluster occurred only in Pseudomonas species. Similarly, tynD, encoding a putative additional tyramine deaminase activeat 37°C, is only found in the tyn clusters of the two strains of P. putida (Fig. 7).
An analysis of the genetic organization of hpa loci (Diaz et al., 2001) (Fig. S5) showed that, analogous to the tyn cluster, the hpa genes were arranged identically in P. putida U and P. putida GB-1 whereas differences were found in other microbes. In the latter, the hpa genes are usually located within a single cluster (P. putida, E. coli, Salmonella typhimurium and Y. pestis), but in K. pneumoniae, Pseudomonas aeruginosa and Pseudomonas entomophila some of them are found in different regions of the bacterial genome. Hence, in K. pneumoniae the hpa genes are organized in two widely separated clusters, hpaR and hpaG1G2EDFHIXABC; in P. aeruginosa there are three closely spaced clusters (hpaBC, hpaG1G2EDFXHI and hpaA); and in P. entomophila all of the genes (with the exception of hpaX) are grouped together. Moreover, in E. coli and S. typhimurium the genes hpaG1 and hpaG2 apparently fused, generating a single gene, hpaG. A characteristic common to all the clusters in Pseudomonas is that the genes encoding regulators (hpaA and hpaR) are distant from those whose expression is predicted to be regulated (Fig. S5). Two additional differences were observed when the hpa clusters from P. putida U and P. putida GB-1 were compared with the hpa clusters of other bacteria. In the P. putida clusters there is an extra copy of hpaR (hpaY, flanking the tyn cluster), suggesting that the hpa genes were acquired in a recombination event involving homologous sequences. Additionally, in both clusters (and in P. entomophila) a tetR gene, which does not seem to be involved in the regulation of any of the tyn or hpa genes, is present between hpaB and hpaY genes.
Cloning of the tyn and hpa cluster in a single vector: transference to other microbes
Once the genes belonging to the tyn and hpa cluster were identified and the functions of the encoding proteins established, we approached the construction of genetic constructions containing the tyn or tyn + hpa clusters (Fig. 4), with the goal of transforming different microbes and thus equip them with the ability to degrade Tyn, DA, 4HPA and/or 3,4HPA. When E. coli W14, a mutant of E. coli W unable to degrade PA and 2-phenylethylamine but containing the HPA pathway, was transformed with the pKtyn-4 construct, the recombinant strain was able to grow in a chemically defined medium containing Tyn as the sole carbon source (see above and Fig. 5), in contrast to E. coli W14 pK18::mob. Similar results were recorded when these strains were cultured in MM + DA (5 mM). These results confirm that the pKtyn-4 construct contains all the genes required to transform Tyn and DA into 4HPA and 3,4HPA respectively.
Although these results are technologically interesting, the complete degradation of tyramine and dopamine requires both the Tyn and the HPA pathway, in such a way that microbes lacking the hpa cluster are unable to assimilate these two aromatic amines. To solve this limitation, we cloned the tyn and hpa clusters from P. putida U in a single plasmid, following the protocol described in Fig. 8. This construct was used to transform P. putida KT2440, a strain closely related to P. putida U but lacking the Tyn and HPA pathways and therefore unable to degrade HPA, Tyn or DA. When the recombinant strain P. putida KT2440 pK18::mobtynhpa was cultured in MM containing Tyn (or DA) as the sole carbon source, it grew efficiently (Fig. 9), whereas recombinant P. putida KT2440 pK18::mob was unable to degrade Tyn or 4HPA. These data showed that pK18::mobtynhpa contains the genetic information required for cellular assimilation of Tyn, DA, 4HPA and 3,4HPA. Importantly, this is the first description of a genetic element that can be transferred to other bacteria, conferring upon them the ability to degrade all the above-mentioned compounds. Furthermore, pK18::mobtynhpa, containing the hpa and the tyn clusters, can be easily released from the pK18::mob plasmid (after XbaI/HindIII digestion), and then cloned in other vectors and used to transform different microbes, thus enlarging the possibilities to design novel bacterial strains with high biotechnological potential (i.e. lactic acid bacteria) (Arcos et al., 2008). These strains could be used to eliminate tyramine or dopamine from different media or, even, from fermented foods.
Figure 8. Schematic representations of the genetic constructs (pK18::mobtynhpa) containing the genetic information required to assimilate tyramine, dopamine, 4HPA and 3,4HPA.
Download figure to PowerPoint
Figure 9. Growth (A540) of P. putida KT2440 pK18::mob (▪, □) and P. putida KT2440 pK18::mobtynhpa (○, ●) when these strains were cultured in MM + 4HPA (5 mM) (▪, ●) or tyramine (5 mM) (□, ○). Similar results were obtained when dopamine (5 mM) was the carbon source.
Download figure to PowerPoint