Inferring the genetic network of m-xylene metabolism through expression profiling of the xyl genes of Pseudomonas putida mt-2



A subgenomic array of structural and regulatory genes of the TOL plasmid pWW0 of Pseudomonas putida mt-2 has been constructed to sort out the interplay between m-xylene catabolism and the environmental stress brought about by this aromatic chemical. To this end, xyl sequences were spotted along with groups of selected P. putida genes, the transcription of which become descriptors of distinct physiological conditions. The expression of the TOL pathway in response to pathway substrates was thus profiled, uncovering a regulatory network that overcomes and expands the predictions made by projecting known data from individual promoters. First, post-transcriptional checks appear to mitigate the burden caused by non-productive induction of the TOL operons. Second, the fate of different segments of the polycistronic mRNAs from the upper and lower TOL operons varies depending on the metabolism of their inducers. Finally, m-xylene triggers a noticeable heat shock, the onset of which does interfere with optimal expression of catabolic genes. These results reveal a degree of regulatory partnership between TOL plasmid-encoded functions and host physiology that go beyond transcription initiation control.


Many Gram-negative soil bacteria can employ toxic aromatic compounds (which are otherwise environmental pollutants) as the only carbon and energy sources. Typically, when Pseudomonas putida mt-2 faces m-xylene in the medium, a paradoxical scenario occurs in which this aromatic compound is sensed both as a growth substrate to metabolize (Assinder and Williams, 1990) and as an environmental stressor to endure (Ramos et al., 2001; 2002). The transcriptional choices between activating metabolic genes and setting off responses to stress form a network of regulatory switches which, given the redundancy and robustness of the regulatory circuits involved, cannot be entirely unravelled through standard genetic analysis (Cases and de Lorenzo, 2005). P. putida mt-2 metabolizes m-xylene by virtue of a complex enzymatic machinery encoded by the TOL plasmid pWW0 (Assinder and Williams, 1990; Marques et al., 1999a; Ruiz et al., 2004). Such a machinery includes functions for the stepwise oxidation of one of the methyl groups of m-xylene into m-toluate (encoded by the so-called upper operon; Fig. 1), followed by the dioxygenation of this carboxylic acid, and the meta-cleavage of the resulting methylcatechol and subsequent steps down to Krebs’ cycle intermediates (determined by the lower TOL pathway; Fig. 1). Expression of each of the gene clusters is submitted to an intricate interplay between plasmid-encoded regulators and host factors (Ramos et al., 1997), which is sketched in Fig. 1. In the presence of pathway substrates (i.e. m-xylene), the upper operon is transcribed from a σ54-dependent promoter (Pu), which is activated by the enhancer-binding protein XylR. The same XylR/m-xylene complex increases expression of the activator of the lower operon (XylS), that is transcribed from a second, semi-constitutive σ54-dependent promoter Ps1 (Perez-Martin and de Lorenzo, 1995) overlapping a weak σ70 site (Ps2; Gallegos et al., 1996a). Activation of the lower operon from Pm (Fig. 1) needs not only XylS and m-toluate, but also the participation of either the heat-shock sigma σH or the stationary stress σS of the host (Marques et al., 1999b). This is because the RNA polymerase (RNAP) bound to the housekeeping σ70 is unable to start transcription in Pm. Furthermore, overproduction of XylS can activate Pm in the absence of lower pathway effectors (Inouye et al., 1987; Kessler et al., 1994). Finally, XylR expression appears to be autoregulated, as the UAS of the σ54Ps promoter overlap the Pr promoter of xylR (Bertoni et al., 1998; Marques et al., 1998).

Figure 1.

Organization of the m-xylene biodegradation pathway born by the TOL plasmid pWW0. The top of the figure sketches the reactions implicated in metabolism of this aromatic compound, including the stepwise oxidation of one methyl group of the substrate to an alcohol and eventually to a carboxylic acid, yielding m-toluate through the action of the enzymes encoded by the upper TOL pathway. m-toluate is then cis-dioxygenated to yield 3-methylcatechol, which is cleaved in meta and finally channelled into the Krebs cycle by means of the products of the lower or meta operon. The upper operon is transcribed from the σ54 promoter Pu upon activation by the cognate regulator of the pathway (XylR) bound to specific effectors. These include the substrate of the pathway (m-xylene) as well as the two first metabolic intermediates: 3-methylbenzylalcohol and 3-methylbenzylaldehyde. The lower operon is transcribed from the Pm promoter, which is activated by the m-toluate responsive activator XylS. Finally, xylS and xylR are transcribed from the divergent and overlapping promoters Ps and Pr respectively. The regulation of the latter is connected, because the Ps promoter is activated by XylR, which also binds and downregulates its own Pr promoter. The bottom of the figure is a blow-up of the regions of interest (not to scale) in the TOL plasmid pWW0, in which the upper operon is not adjacent to that of the lower pathway. Relative orientations of each gene and their approximate sizes are indicated along with some enzymatic activities encoded by each DNA segment. The genes corresponding to gross multienzymatic complexes/steps are grouped with the same colour code.

Apart of the pathway-specific control imposed by substrates, expression of the xyl genes is also subject to various global regulation mechanisms including: (i) catabolite repression mediated by intermediates of the Entner–Doudoroff route (Velazquez et al., 2004), (ii) growth-phase control caused by IHF binding to Pu (Valls et al., 2002) and (iii) perhaps also by sigma factor competition in stationary phase (Carmona et al., 2004). The current view of the overall regulation of the xyl genes of the TOL plasmid (Fig. 1) is built from the patching of observations from individual promoters, often away from their native context. The functioning of the system has not to date been visualized in its entirety. Furthermore, induction of the xyl genes and products has been typically measured during saturation by volatile upper pathway effectors. Finally, the deleterious effects of exposure to m-xylene and their influence on the levels of the TOL products have been basically overlooked.

In this work, we have employed DNA array technology to examine simultaneously expression of each of the blocks of xyl genes of pWW0 when P. putida mt-2 faces subsaturating concentrations of m-xylene in the medium. Such an expression was visualized on the background of the physiological stress brought about by this aromatic compound. Our results reveal the link of m-xylene metabolism with heat-shock response and suggest that mRNA stability of xyl genes is influenced by intermediates of m-xylene degradation.


Rationale of the subgenomic TOL stress DNA chip

The question that we have addressed in this work is not so much the transcriptome of P. putida mt-2 in the presence of m-xylene (P. Dominguez-Cuevas et al., in preparation), but the regulation of the xyl genes encoded by the TOL plasmid pWW0 with a tool tailored for monitoring the expression of the whole genetic complement of the system and its interplay with major physiological conditions. To this end, the subgenomic DNA microarray described in Experimental procedures and in the supplementary material available at No. E-MEXP-305) consist of a matrix of 88 polymerase chain reaction (PCR) products generated by amplification of xyl genes of pWW0 along with groups of cistrons known to be descriptors of the main classes of environmental stresses, i.e. oxidative stress, osmotic stress, starvation of key nutrients (iron, nitrogen, carbon, sulphur), heat shock, stringent response, SOS response to DNA damage and nucleoid-associated proteins, as well as a repertoire of the sequences for housekeeping and alternative sigma factors. The concept behind the choice was to ensure that exposure to m-xylene under our assay conditions affected exclusively or predominantly expression of cognate metabolic genes rather than triggering a severe response to the toxicity of the pathway effectors (Ramos et al., 2002) that could cause misinterpretation of the expression of the xyl genes. As shown below, this simple notion allowed us to unveil features of the regulatory network that were invisible to the genetic approaches which have been largely employed so far to tackle the same problem.

Expression profile of the xyl genes of plasmid pWW0 in the presence of m-xylene

Figure 2 A shows the gross transcription pattern of the xyl genes when P. putida mt-2 cells were submitted to subsaturating concentrations of m-xylene for a short (15 min) or a long (3 h) period. To do this, cells were grown aerobically in 10 mM succinate as the only C source up to an absorbance at 600 nm (A600) of 0.5. Cultures were then split and either incubated in the same conditions or exposed to the vapours of a 1:5 dilution of m-xylene in dibutylphthalate. The slow release of m-xylene from this mix prevented the noticeable slowdown of growth that occurs when cells are abruptly exposed to saturating vapours of pure m-xylene (Garmendia and de Lorenzo, 2000). Comparisons were thus made between cells pre-grown in succinate and then exposed to m-xylene rather than permanently grown in m-xylene versus permanently grown in succinate. This allowed a faithful comparison of the sole effects of the aromatic substrate on expression of the xyl genes rather than setting a major physiological switch caused by growth on one substrate versus the other. In addition, succinate does not cause catabolic repression or any physiological downregulation on the xyl genes (Holtel et al., 1994; Cases et al., 1999), thus ensuring that the effects of m-xylene can be traced exclusively to this aromatic and not to any side-effect of the choice between the two C sources.

Figure 2.

Response of P. putida mt-2 to TOL pathway substrates.
A. Induction of the xyl genes of the TOL plasmid. Cells were treated as explained in the text and the relative expression profile of the upper and lower pathway genes in cells exposed to m-xylene for the time indicated versus untreated samples was determined. Each bar of the plot represents the mean value of 10 spots in each of the tailored DNA chip analysed in three to four separate experiments with dye-switched samples (see Experimental procedures and, Accession No. E-MEXP-305), and normalized using as invariable references the signals from genes atkA, dbhA, phaF, gst, lexA, rpoN and rpoD. The thick arrows on top of the plot indicate the colinearity of the signals with the order of the upper and lower xyl genes of the TOL plasmid.
B. Specificity of the response of xyl genes to m-xylene. The plot represents the signals of xyl spots as a percentage of the intensity of the reference, invariable rpoD gene in P. putida KT2440 (blue dots) or P. putida mt-2 (pink dots) exposed or not to m-xylene. Note that All xyl spots are virtually below the detection threshold in the case of TOL-less P. putida KT2440 strain, thereby ensuring the reliability of the signals for the TOL-plus strain P. putida mt-2.
C. Induction of the TOL meta pathway by m-toluate. P. putida mt-2 cells were grown up to an A600 of 0.5, added with 5 mM of 3-methylbenzoate (m-toluate) and collected for analysis 15 min later. The relative expression profile of the lower pathway genes in treated versus untreated samples is shown.

The raw results of all the hybridizations, along with supplementary material on the primers employed to generate the PCR products and thorough details on experimental conditions are available on line under the MIAME guidelines at the ArrayExpress Database ( with Accession No. E-MEXP-305. The first salient feature of the results shown in Fig. 2A is that the whole cluster of the upper xyl operon is quickly and strongly induced upon exposure to m-xylene. As early as 15 min after the culture confronts the slow release of m-xylene, all upper xyl genes become noticeably expressed. This is remarkable, as contact of the volatile aromatic compound with the P. putida cells under the assay conditions has to go through three phase-partition events. The same strong and early induction is true for the xylS gene which, like the upper operon, is also controlled by the m-xylene responsive regulator XylR (Fig. 1). On the contrary, the bulk of the lower pathway operon seems to be silent after 15 min except for some minor signals from the leading cistrons of the operon. The situation changes after a prolonged exposure to m-xylene (Fig. 2A). At that time, expression of xylS remains high and the transcripts from the upper xyl cluster grow to be stronger, with their relative signals reaching similar levels

However, the most evident change after 3 h of contact of P. putida mt-2 with m-xylene is the expression of the entire complement of lower operon xyl genes, although the relative levels of each sequence segment did vary. Perusal of the data on the meta operon in Fig. 1A suggest that the full-size mRNA has at least three distinct segments in terms of their relative abundance in vivo. The first would encompass the leading xylXYZL genes, which encode the enzymes for conversion of m-toluate to 3-methyl catechol (Fig. 1). A second segment could start by the xylT/xylE region and terminate by xylJ, the third fragment including the rest of the meta cistrons. It is noteworthy that an earlier study on mRNA stability of the xyl genes (Marques et al., 1993) has suggested the presence of specific endonuclease cleavage sites within the same regions of the mRNA where we have detected the boundaries of the various RNA portions relative to their actual abundance (see Discussion). Figure 2B shows that all the variations observed were due to genuine changes in the relative levels of xyl transcripts and not to cross-hybridization with chromosomal genes. This is an important control, as mRNA extracted from pWW0-less cells did not cause any significant variation on the xyl spots of the DNA chip whether or not the cultures had been induced with m-xylene (Fig. 2B).

Interplay between induction of heat shock by m-xylene and expression of xyl genes

Figure 3 shows the rest of the signals detected in the DNA array from P. putida mt-2 cells exposed to subsaturating m-xylene vapours. Basically, none of the physiological descriptors spotted on the chip was indicative of a major stress condition that could prevail over expression of the xyl genes. This was not unexpected, as the growth rate of the induced cultures was indistinguishable from the non-exposed counterparts. However, the group of genes that pinpoint a heat-shock response (groEL, groES, ftsH, dnaK, clpP, clpX) became consistently induced in the presence of the aromatic substrate. Whether such a stress is beneficial or detrimental for expression of the xyl genes is paradoxical. On the one hand, expression of the meta pathway during exponential growth needs XylS in concert with the form of RNAP associated to the heat-shock sigma factor σH holoenzyme (Marques et al., 1999b). On the other hand, activation of σH would compete with other sigmas in stationary phase for accessing the limiting pool of core RNAP (Jishage et al., 2002; Laurie et al., 2003), predicting a decrease in expression of σ54 promoters such as Pu and Ps. To sort out such opposite effects, we re-examined expression of both the upper and lower TOL operons under conditions of an artificially provoked heat shock.

Figure 3.

Expression profiling of stress descriptors in cells exposed to m-xylene. Cells were treated as explained in the legend to Fig. 2A and the relative expression profile of the lower pathway genes following 3 h of exposure to m-xylene was determined. The colour code gathers different types of genes that are indicative of the distinct classes of stresses indicated at the bottom of the plot (see Experimental procedures and, Accession No. E-MEXP-305).

Figure 4 shows the expression profile of the xyl genes of P. putida mt-2 at 42°C versus 30°C in the presence of m-xylene (Fig. 4A), along with those of the DNA chip that are indicative of metabolic stress (Fig. 4B). That a heat shock occurred shortly after the temperature shift was indicated by the immediate induction of virtually all cognate descriptors (dnaK, groES, groEL, dnaJ, ftsH). None of the other stress markers were significantly affected. Both the upper and the lower TOL pathways were consistently less expressed at the higher temperature. It is thus likely that heat shock is a setback – rather than a helpful signal for the full expression of the pathway. Similarly, an induction of the lower pathway enzyme catechol 2,3 dioxigenase (C2,3O, encoded by xylE) was not detected when cells were co-exposed to m-xylene and a range of ethanol concentrations 0–4% (not shown). These results were consistent with the notion that promoters influenced by sigma factor competition in stationary phase (typically, σ54 promoters; Jishage et al., 2002; Laurie et al., 2003), to which class Pu and Ps belong, are downregulated by the onset of a heat-shock response. Such a setback seems to override over any positive effect on Pm (Marques et al., 1999b).

Figure 4.

Effect of heat shock on expression of the xyl genes of P. putida mt-2. Samples for hybridization were prepared with RNA of cells pre-grown at 30°C in the presence of m-xylene and then subjected to 42°C for 15 min.
A. Comparative levels of xyl transcripts in the upper and the lower pathway operons at 42°C versus 30°C.
B. Expression profiling of stress descriptors in cells exposed to 42°C versus 30°C.

Expression of the TOL meta pathway and metabolism of m-xylene

The delay in expression of the meta pathway in respect to that of the upper pathway (Fig. 2A) is informative, as it suggests that increased production of XylS caused by the activation of Ps by the XylR/m-xylene complex is not sufficient for a significant induction of the lower operon. This is unexpected, as overexpression of XylS can induce transcription of Pm even in the absence of cognate substrate (Kessler et al., 1994; Gallegos et al., 1996a,b). We thus examined such an induction with an alternative method, namely monitoring directly the levels of the product of the key gene of the meta pathway xylE (encoding C2,3O) with (i) a monoclonal anti-BphC antibody, able to specifically recognize C2,3O and (ii) direct measure of C2,3O activity. Figure 5A shows that C2,3O levels become detectable in P. putida mt-2 cells only after a considerable time of exposure to m-xylene. It therefore seems that a certain intracellular threshold concentration of the XylS effector m-toluate originated by the metabolism of m-xylene is required for a major expression of the meta operon genes. To further examine this hypothesis, we reasoned that artificially overcoming the first and more energetically difficult m-xylene monooxygenation step of the TOL route (Brinkmann and Reineke, 1992; Buhler et al., 2000; Haro and de Lorenzo, 2001) should result in a higher accumulation of m-toluate and thus in a better induction of the meta operon. To this end, we compared the expression levels of meta operon genes in response to m-xylene with those in response to 3-methylbenzylalcohol (3MBA). To this end a P. putida mt-2 culture pre-grown in succinate was split and either exposed to vapours of m-xylene as before or added with 3 mM 3MBA. The latter is the product of the early oxidation of one of the methyl groups of m-xylene by the TOL-encoded monooxygenase (encoded by xylMA; Fig. 1), and it is more water-soluble. These two circumstances are predicted to improve the metabolic flow of the upper pathway to produce m-toluate. Following 15 min, RNA from each m-xylene-exposed or 3MBA-amended culture was labelled and hybridized to the tailored chips, with the results shown in Fig. 6A. While the same levels of induction for upper pathway genes were observed in both conditions, 3MBA consistently triggered a higher expression of the meta operon genes. The level of xylS transcript remained the same with both inducers, thereby suggesting that the only significant change was an increased intracellular supply of m-toluate available for activating XylS.

Figure 5.

Monitoring intracellular levels of xyl products in P. putida mt-2. Protein preparations from each culture were separated by denaturing SDS-PAGE, transferred to a PVDF membrane and probed with antibodies against the products indicated.
A. Accumulation of XylE in cells of P. putida mt-2 exposed to m-xylene at different times. The control lane 7 was loaded with cells of E. coli CC118 transformed with the xylE+ plasmid pXYLE10.
B. Detection of XylS in various bacterial hosts. Lane 1: control E. coli CC118 transformed with the xylS+ plasmid pVLT43. Lane 2, control P. putida KT2440. Lanes 3–5: P. putida mt-2 exposed for 3 h to the inducers indicated in each case.
C. Detection of XylE in cells of P. putida mt-2 exposed for a short (15 min) or a long (3 h) period to dibutylphathalate (DBP), m-xylene or o-xylene. Lane 1 was loaded with a control sample of strain P. putida KT2440, which is devoid of the TOL plasmid. The relative C2,3O activity of the cultures adjusted to an A600 of 1.1 after 3 h of exposure to the inducers was as follows: with DBP, 21 arbitrary units (AU); with o-xylene, 27 AU; with m-xylene, 150 AU.

Figure 6.

Differential effect of 3-methylbenzylalcohol (3MBA) versus m-xylene on P. putida mt-2. Cultures were grown up to an A600 of 0.5, and either added with 3 mM 3MBA or exposed to m-xylene vapours for 15 min and the relative expression profile was determined.
A. Expression profiling of the xyl genes of the TOL plasmid.
B. Relative activity of stress descriptors.

Examination of Fig. 6B revealed also a noticeable change in expression of some of the stress descriptor genes brought about by exposure to 3MBA – as compared with m-xylene. First, there was an increase in some of the indicators of the heat-shock response (GroEL and DnaJ). This is not altogether unexpected, as the direct addition of 3 mM 3MBA to the culture, rather than exposure to vapours, may ease the access to the cell membrane, resulting in a more intense stress. The one significant difference between the two conditions (3MBA versus m-xylene) is the clearly higher levels of two descriptors of higher nitrogen demand (gltB, glnA). The significance of this change is not trivial, although it might be related to the channelling of 3MBA biodegradation intermediates into N-containing metabolites. The fur gene (related to iron regulation) seemed also to be enhanced with 3MBA, conceivably reflecting a higher demand of the metal for the assembly of iron-containing oxidases and oxygenases.

The lower TOL loop: induction of the meta operon by m-toluate

To further examine whether induction of the meta TOL operon by m-xylene needs the build-up of meta pathway inducers, we examined the response of P. putida mt-2 to direct addition of m-toluate. This allows the induction of the lower TOL operon separately from the upper pathway and independent of its accumulation as a downstream product of m-xylene metabolism. Figure 2C shows the result of comparing P. putida mt-2 cells pre-grown in minimal medium and succinate, 15 min after addition of 5 mM m-toluate. This aromatic compound is water-soluble and quickly diffuses through the cell membrane. Access to the cognate regulator is thus expected to take much shorter and gain access to XylS at concentrations quite above the transient accumulation that could result from metabolism of m-xylene. Despite this, the overall expression profile was somewhat modest as compared with the eventual induction of the lower operon observed with m-xylene (Fig. 2A) or 3MBA (Fig. 6A). The only major difference between the two scenarios is the relative intracellular levels of XylS as, unlike m-xylene, addition of m-toluate to the culture does not seem to raise the concentration of the regulator above a very low, barely detectable level (Fig. 5B). However, it is a fact that P. putida mt-2 can grow well on m-toluate as sole C and energy source (Assinder and Williams, 1990), so even such a relatively low transcriptional activity should suffice for a vigorous, unrestrained growth on this substrate. A second aspect of the induction of P. putida mt-2 by m-toluate is the lack of any significant initiation of a heat-shock response – or, for the same thing, any significant physiological stress (not shown). This is in contrast with the consistent manifestation of heat-shock descriptors when cells are exposed to upper pathway inducers (m-xylene, 3MBA; Figs 3 and 6B). As the Pm promoter appears to be transcribed with forms of RNAP bearing alternative sigma factors such as σH (Marques et al., 1999b), it is possible that the dearth of the corresponding stress signal contributes also to the low activity of the meta pathway caused by m-toluate only.

The upper TOL loop: induction of xyl genes with o-xylene

To study the degree of synchronization of the onsets of both the upper and the lower TOL pathways, we measured the response of P. putida mt-2 to subsaturating vapours of o-xylene. While this compound is the closest chemical relative of m-xylene as well as a strong effector of XylR (Abril et al., 1989), the leading enzyme of the upper pathway (xylene monooxygenase, encoded by xylMA) is altogether unable to use o-xylene as a substrate (Buhler et al., 2000). This results in a gratuitous transcriptional activation of Pu and a metabolically non-productive expression of the upper pathway. On this background, we re-examined the expression of the gross complement of both TOL operons when cells were exposed to o-xylene under conditions similar to those employed before for m-xylene.

Figure 7 shows that expression of the leading gene of the upper pathway (xylU, downstream of the Pu promoter) reached a remarkable expression level within the range – or even superior to that of the authentic pathway inducer m-xylene (see Fig. 2A). However the intensity of the downstream upper operon segments appeared to decline with the distance from Pu. This was in contrast with the uniformity of expression of the upper TOL genes found upon induction to m-xylene (Fig. 2A). A second feature of this experiment (Fig. 7) was the virtual lack of induction of the meta operon genes, despite the noticeable increase in the transcript of the cognate regulator, xylS. The lack of expression of the meta pathway operon was confirmed, as above, by examining the levels of the product of the xylE gene (C2,3O) with a monoclonal antibody as well as by the measure of the actual activity levels of the enzyme (Fig. 5C). In cells exposed to o-xylene, these were below the detection limits of both procedures – in contrast with the presence of C2,3O in cells induced with m-xylene. The differences in induction of the TOL genes caused by either m-xylene or o-xylene cannot be attributed to an indirect effect of the physiological stress that accompanies exposure to volatile solvents, as the expression profile of the stress descriptors of the DNA chip remains unaffected between the two inducers (not shown).

Figure 7.

Response of P. putida mt-2 to o-xylene. Induction of the xyl genes of the TOL plasmid. Cells were exposed to o-xylene for 3 h and the relative expression profile of the upper and lower pathway genes versus that of untreated samples was determined.

These results make perfect physiological sense, as complete expression of the 22 TOL-encoded proteins necessary for channelling m-xylene into the central metabolism would become a burden if they were deceived into full production by a non-substrate. However, as all the factors required for transcription initiation of the corresponding promoters are in place, it is likely that there are other post-transcriptional checkpoints that may prevent transcription of full-length mRNAs (and thus the assembly of the encoded enzymatic complexes) unless the right substrates and metabolic intermediates happen to be also present. One plausible prospect to explain this could be that under non-productive conditions XylS is transcribed but not translated. To check this possibility, we employed an anti-XylS serum to visualize the relative levels of the protein under various growth settings. As shown in Fig. 5B, despite some differences in intracellular concentration, the XylS protein is clearly produced at considerable levels when cells are exposed to either m-xylene or o-xylene. This rules out the lack of XylS as the reason for the very limited activation of the lower pathway by o-xylene and suggests, instead, other mechanisms perhaps related to controlling mRNA stability (see below).


This work provides for the first time a general view of the gross expression profile the two biodegradative operons encoded by the archetypal catabolic pWW0 plasmid of P. putida mt-2. Most studies on the regulation of this model for metabolism of recalcitrant aromatics have concentrated so far on the regulation of transcription initiation of the individual promoters of the TOL genes. To this end, the Pm, Pu, Pr and Ps promoters (Fig. 1) have been studied in vivo with gene fusion technology and primer extension analysis and, where possible, in vitro as well using pure or semi-pure proteins and regulatory factors (Ruiz et al., 2004). Some features of the regulation have also been revealed by data on the meta operon mRNA stability (Marques et al., 1993) and the determination of some enzymatic activities encoded by the TOL system (Duetz et al., 1994; 1996). However, the number of promoters, factors and effectors involved in such regulation has prevented so far the visualization of the complete system functioning under various growth conditions. This is of essence, as the biological functionality of a regulatory network is not only based on the whole of on/off switches and rheostatic responses of the promoters involved (Cases and de Lorenzo, 2005), but also on the downstream expression checkpoints. The TOL system offers a suitable experimental model to this end, as the performance of the regulatory network and its implantation within the general physiology of cells is directly connected to the survival of P. putida mt-2 in sites polluted with given aromatic compounds (Ramos et al., 1991; Sarand et al., 2000). This study intends to contribute to sort out such a multifaceted issue.

The first aspect of the work to be considered is that the tailored DNA chip employed says little on the detailed transcription initiation from the separate promoters. Instead, it provides a gross picture of the production of transcripts under various conditions – along with alerts on general types of physiological stresses. Hybridization conditions were such that only major responses were programmed to show up in the chip. While high-resolution profiles of the responses of P. putida to diverse conditions are likely to result from the use of genome-wide oligonucleotide arrays (Yuste et al., 2005), the tailored chip employed in this work clearly shows that transcription initiation at the TOL promoters – as known from previous work – may not translate automatically into an even expression of the corresponding transcripts. One significant clue on this subject is the induction of the upper TOL pathway genes by various inducers. As shown in Figs 2A, 6A and 7, there was a rapid and intense response to any of the three XylR effectors tested: (i) the first substrate of the pathway (m-xylene), (ii) the first intermediate following the restrictive monooxygenation bottleneck (3MBA) and (iii) a gratuitous, non-metabolizable effector (o-xylene). But the overall pattern of expression of the upper genes did change with each inducer. As revealed by the evenness of the signals from the various DNA segments spotted on the DNA chip, m-xylene caused a smooth expression of the upper TOL transcripts throughout the whole extent of the corresponding sequence (Fig. 2A). The same was basically true for the metabolizable inducer 3MBA (Fig. 6A). However, when P. putida mt-2 was exposed to o-xylene (Fig. 7), we noticed that the high level of the first gene of the upper operon was followed by a progressive reduction of the downstream sequences. This cannot be attributed to a side-effect of each of the inducers on cell physiology, as the gross stress descriptors behaved almost identically in all cases. Instead, comparison of the data of Fig. 2A versus those of Fig. 7 suggests that whether the inducer is or is not metabolized makes a difference in the relative levels of the upper operon mRNA segments downstream of the transcription initiation site.

One unanticipated side result involved the xylR transcript in cells induced by any of the upper pathway inducers. As the XylR protein represses its own promoter (Bertoni et al., 1998; Marques et al., 1998), XylR effectors are predicted to inhibit the build-up of xylR transcripts. Instead, we observed that such levels were maintained or even increased when cells faced m-xylene, 3MBA or o-xylene (Figs 2A, 6A and 7). This suggests that, under induction conditions, the XylR protein might be subject to a very rapid turnover that limits autorepression. In any case the unusual signal of xylR transcripts cannot be traced to cross-hybridization of the printed sequence on the chip with any of the 22 genes encoding prokaryotic enhancer-binding proteins (EBPs) in the P. putida genome (Cases et al., 2003). When the same experiments were made with a pWW0-free P. putida counterpart (Fig. 2B), we could not find any signal matching the xylR spot under our stringent hybridization conditions. Perhaps this paradox will have a satisfactory explanation when higher resolution arrays become available.

That the gross expression of TOL genes is connected to metabolism of their substrates and not only to straight promoter induction by their respective effectors was further exposed by the data on the mRNA levels of the meta operon. Figure 2A shows that unlike the upper pathway, the lower operon is not induced instantly after cells encounter m-xylene, and it takes some time to see any significant build-up of meta xyl transcripts. In contrast, the meta operon is rapidly induced in cells grown in a medium amended with three mM 3MBA (Fig. 6A). This more soluble compound results naturally from the first monooxygenation of m-xylene, which is the bottleneck of the whole process (Brinkmann and Reineke, 1992; Buhler et al., 2000; Haro and de Lorenzo, 2001). It is thus likely that 3MBA addition to the culture immediately results in enough intracellular concentrations of m-toluate for inducing the lower pathway. In contrast, non-metabolizable o-xylene kept expression of the lower pathway at low levels even 3 h after exposure to its vapours. Interestingly, in all three induction conditions (m-xylene, 3MBA and o-xylene), the levels of xylS transcript were increased (Fig. 6A) and XylS protein was produced at considerable concentrations (Fig. 5B). But such levels were insufficient to induce the meta operon in the absence of pathway substrates. Once more this indicated that, regardless of the mechanisms that operate at the level of transcription initiation, post-transcriptional checks are likely to mitigate the burden caused by non-productive induction of the TOL operons. While DNA chip technology says little on the mechanisms behind such an effect, possibilities include the requirement of a dedicated sigma factor for transcription of Pm (out of the 24 available in the genome of P. putida; Martinez-Bueno et al., 2002) or the dearth of full-length mRNA due to early termination or even metabolite-mediated riboswitches (Nahvi et al., 2002; Lai, 2003; Winkler et al., 2004). That the mRNA of the meta operon is subject to some sort of post-transcriptional control is consistent with earlier observations on the presence of cleavage sites around the xylT/xylE and xylJ sequences of the transcripts (Marques et al., 1993). The signals encompassing these sites in the chips are indeed less intense, even at full induction (Fig. 2A and C), suggesting that they possibly mark the borders of the various RNA segments in respect to their relative abundance.

When the meta pathway was separately induced with a high concentration of m-toluate (Fig. 2C), there was an instant induction of the cognate genes, although not to the extent detected when the whole TOL pathway was stimulated with the metabolizable upper pathway effectors m-xylene or 3MBA (Figs 2A and 6A). This is not surprising, as the levels of XylS were very low in cells exposed to m-toluate (Fig. 5B). Moreover, while the heat-shock factor σH seems to be favourite partner of XylS for activation of Pm (Marques et al., 1999b) m-toluate did not trigger any significant response of this sort under our assay conditions. It would hence seem that the regulation of the lower operon is optimized to deal with the products of the upper route and not for keeping a separate metabolic block for m-toluate only. This is consistent with the observation that rpoN mutants of P. putida mt-2, which lack σ54 and are thus altogether unable to express the upper pathway, grow very poorly in m-toluate (Kohler et al., 1989).

A sidelight worthy of note is the part of heat shock in expression of the TOL genes. Upper pathway inducers do trigger a moderate heat shock (Figs 3 and 6B), a logical consequence of being apolar solvents (Ramos et al., 2001; 2002). As σH is one sigma that transcribes Pm in Escherichia coli (Marques et al., 1999b), it would make sense that such a stress has been evolutionarily recruited as a useful signal for integrating expression of the TOL genes into cell physiology (Rojo and Dinamarca, 2004; Cases and de Lorenzo, 2005). On the other hand, there is a competition between sigma factors for the core RNAP at the stationary phase, which is mediated by ppGpp (Jishage et al., 2002; Laurie et al., 2003) and possibly by factor DksA as well (Hirsch and Elliott, 2002; Paul et al., 2004; Perron et al., 2005). The onset of an active σH brought about by heat shock would lessen the pool of RNAP available for σ54 binding and thus expression of the upper genes (Fig. 1). Expression of XylS might be less affected by heat shock, as Ps (Fig. 1A) is in fact a combination of one σ54 promoter with an overlapping σ70 counterpart (Gallegos et al., 1996a). In any case, the outcome of such a paradox is evident in the data of Fig. 4A: in the context of the whole TOL plasmid, heat shock is a problem – not an asset for expression of the biodegradation ability. Similar results became evident when heat shock was provoked with ethanol without altering the temperature (not shown). The effect of heat shock and further environmental stresses on expression of catabolic genes of the TOL plasmid and other degradative operons of soil bacteria (Denef et al., 2004) has a wide biotechnological implication that will surely be the subject of future work.

Experimental procedures

Strains and growth conditions

Pseudomonas putida mt-2 is the original isolate of P. putida bearing the archetypical TOL plasmid pWW0 (Greated et al., 2002). The plasmidless derivative of P. putida mt-2 is strain P. putida KT2440 (Regenhardt et al., 2002). Bacteria were cultured aerobically at 30°C in M9 mineral medium (Miller, 1972) supplemented with 10 mM succinate and spiked with 0.05% of the non-ionic detergent Triton X-100 to avoid clumping of the cells – otherwise Triton X-100 had no effect on growth. For induction experiments, overnight cultures were diluted 100-fold in fresh medium and grown with vigorous shaking until an A600 of 0.5 was reached. Where indicated samples were then exposed to subsaturating vapours of m-xylene or o-xylene in airtight flasks. To this end, the release of the inducer was slowed down by generating vapours of the xylenes from a 1/5 dilution in dibutylphthalate. This solvent is a non-inducer of XylR with an extremely low vapour pressure (∼0.00006 mmHg) that rules out any interference with the action of the inducers on the cells. Alternatively, once the cultures had reached an A600 of 0.5, soluble inducers 3MBA or m-toluate were added, where indicated, at concentrations of 3 mM and 5 mM, respectively, and incubated in the same airtight flasks. Chemicals used for induction experiments were purchased from Aldrich, Fluka or Merck and were always of superior analytical purity (>99%). For induction of heat shock, P. putida mt-2 strain was pre-grown at 30°C in the presence of m-xylene/dibutylphathalate vapours up to an A600 of 0.5 and split in two cultures. One of them was heated to 42°C for 15 min while the other was kept on growing at 30°C for the same period of time before collection and RNA extraction (see below).

Protein techniques

SDS-PAGE was performed by standard protocols using the miniprotean system (Bio-Rad). Whole-cell extracts were prepared by harvesting the cells (10000 g, 5 min) from 1 to 10 ml of cultures and by resuspending the cell pellet in 100 µl of 10 mM Tris-HCl pH 7.5. Next 100 µl of 2X SDS loading buffer [120 mM Tris-HCl (pH 6.8), 2% v/v 2-mercaptoethanol, 2% w/v SDS, 10% v/v glycerol, 0.01% w/v bromophenol blue] was added. Samples were boiled (10 min) and centrifuged (14000 g, 10 min). Before electrophoresis crude extract volumes were adjusted in order to load the equivalent of 108 colony-forming units (cfu) per lane in a 10 µl sample. Proteins were separated by SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane (Inmobilon P, Millipore) using a semi-dry electrophoresis transfer apparatus (Bio-Rad). Membranes were blocked for 1 h at room temperature (or 16 h at 4°C) using A buffer [PBS 1× (0.8 mM Na2HPO4, 0.15 mM KH2PO4, 0.3 mM KCl, 13.7 mM NaCl; pH 7.0), 5% skimmed milk, 1% BSA, 0.1% Tween 20]. For detecting XylS, membranes were probed with a 1:5000 dilution of a rabbit anti-XylS serum in 30 ml of A buffer. The corresponding band was developed with an anti-rabbit antibody coupled to horseradish peroxidase and reaction with a chemiluminescent substrate (Roche). The samples were then recorded with an X-ray film (X-OMAT; Kodak) or in a Chemi-Doc (Bio-Rad) luminometer. E. coli CC118 bearing the xylS overproduction plasmid pVLT43 (de Lorenzo et al., 1993) was used as a positive control. For detection of XylE, a similar procedure was followed, except that a monoclonal anti-C2,3O antibody (anti-BphC, kindly provided by M. Tesar, GBF Braunschweig) was employed instead of a polyclonal antiserum. Also, E. coli CC118 with the xylE+ plasmid pXYLE10 (Stein, 1992) was used as a positive control. In this case, XylE was detected with a peroxidase-labelled goat anti-mouse antibody. Alternatively, C2,3O activity was semi-quantitatively detected by adding 15 µl of a solution of 1% catechol to 2 ml of samples of the cultures of the tested strains. After 1 min at room temperature, the samples were spun down for 2 min at 14.000 r.p.m. and the absorbance of the supernatant was determined at 375 nm. Concentration of the semi-aldehyde resulting from the reaction was estimated using an extinction coefficient (ɛ) of 46 000 M cm−1 (McPhee et al., 2003).

DNA microarray design and construction

The set of 88 P. putida mt-2 genes selected for amplification included both TOL genes (xylUWCMABN, xylXYLTEFJQKI, xylRS) and cistrons that are descriptors of various kinds of environmental stresses and distinct growth conditions: sigma factors (rpoD, rpoN, rpoH, algU, mucA, pvdS, fliA, rpoS), oxidative stress (sodA, sodB, catA, zwf-1), osmotic stress (beta, betB, trkA, prop, proV, proX), iron stress (pigA, fur, pvdE), nitrogen stress (ntrC, gltB, glnA), sulphur stress (ssuD, ahpC, cysK), stringent response (spoT, sspA), temperature shock (groEL, groES, ftsH, dnaK, dnaJ, clpP, clpX, fimC, cspA), carbon metabolism (ptsN, ptsO, phaC1, phaC2, phaZ, armG, cstA), DNA damage (recA, recN, lexA1) and miscellaneous markers (hupA, hupB, hupN, lrp, gacA, cti, msrA). Their DNA sequences were retrieved from the TIGR database ( and separately amplified with the PCR using the sets of primers listed in the supplementary material on line deposited in No. E-MEXP-305). Primers were designed such that the amplified DNA covered > 90% of the coding sequence gene under scrutiny. PCR products were checked by agarose gel electrophoresis in a ready-to-run mini-system (Amersham Biosciences) and purified by using a 96-well PCR purification kit (TeleChem International, Sunnyvale, CA). Spotting was carried out in 10-times replicates on Epoxy slides (Telechem International) with a MicroGrid II arrayer (BioRobotics, UK) at 45–50% humidity. Slides were subsequently dried and processed following supplier recommendations.

RNA extraction, labelling and chip hybridization

8 ml of each of the P. putida cultures under examination were spun down in the cold for 30 s at 14.000 r.p.m. and the pellet was rapidly frozen in liquid nitrogen or carbonic ice. Cells were then thawed in 4 ml of TRIZOL (Gibco), resuspended, added with 1 ml of chloroform and vortexed three times for 30 s with alternating 30 s incubations on ice. The RNA-containing water phase was then separated from the organic phase by centrifugation at 14.000 r.p.m., 5°C in a phase-lock eppendorf tube (Eppendorf). The solution with the RNA was passed to a clean tube, added with 1 volume of pure isopropanol pre-cooled at −20°C, mixed and left to precipitate at −80°C for 30 min. Tubes were centrifuged in the cold for 30 min at 14.000 r.p.m. and the RNA sediment was let to dissolve in diethylpyrocarbonate (DEPC)-treated water for 24 h at 4°C. Samples were then added with 20 U ml−1 DNase I (Ambion) to eliminate any traces of DNA. RNA was then re-purified by acid-phenol extraction, re-precipitated with 2 v of absolute ethanol and resuspended in DEPC-treated water. Total RNA amount and its quality were checked by spectrometry with a BioPhotometer (Eppendorf) and a Bioanalyzer 2100 (Agilent Technologies). Twenty microlitres of each RNA sample were retro-transcribed to cDNA by reaction with reverse transcriptase primed with random hexamers (SuperScript II, Amersham) in the presence of 5-(3-aminoallyl)-2′-deoxyuridine 5′-triphosphate as described by Xiang et al. (2002). Each of the reaction products were purified and labelled separately with fluorophores Cy3 or Cy5 (CyTM3 o CyTM5 monofunctional reactive dye, Amersham). For preparation of probes, each of the two reciprocally labelled samples was mixed with the same amount of its corresponding partner. Before use, slides were first washed at room temperature for 2 min in 2× SSC + 0.1% sarcosyl, and then in 2× SSC only. Slides were subsequently boiled for 100°C in dH2O, cooled 10 s at room temperature and fixed by plunging for 2 min in ice-cold 100% ethanol. Five microlitres of each probe mix were laid on the DNA chips, covered with hybridization lids (HybriSlipTM Grace BioLabs) and cased in ArrayItTM (Telechem ♯AHC) hybridization cassettes. These were first heated in a boiling water bath for 2 min in order to denature the probes, and then let to hybridize for 6 h at 58°C. The unbound dyes were then removed by two washings of 5 min at room temperature with 2× SSC + 0.1% sarcosyl, one 1 min washing in 0.2× SSC only and two light rinses in H2O, after which they were dried with a mild centrifugation.

Scanning of DNA slides and data analysis

Slides were scanned for Cy3- and Cy5-specific fluorescence in a 418 Array Scanner (Affymetrix), signals from each array being measured twice in order to generate foci with enough fluorescence intensity for as many spots as possible. Images were overlapped and normalized with the GenePix Pro 4.1 software (Axon, California USA), by imposing a value of 1 to median of fluorescence ratios (MFR) of reference genes atkA, dbhA, phaF, gst, lexA, rpoN and rpoD. MFR was chosen instead of average values as the median is less sensitive to outliers (although no major differences were detected after ruling out unreliable spots). Data were exported to an Excel sheet where unreliable spots data were excluded if: (i) the total intensity of the spot minus background intensity was lower than 1000, (ii) the intensity of the spot in any channel (Cy3 or Cy5) reached the saturation limit (65535 arbitrary fluorescence units) or (iii) the spot was cathegorized by the GenePix software as ‘non-reliable’ based on roundness or homogeneity. Median and standard deviation of the MFR were calculated with the 10 replicates of each gene (excluding those mentioned above) for every hybridization experiment. Data sets where the number of valid replicates was lower than 5 or the standard deviation higher than 15% were ruled out of the subsequent analysis. Average of Cy5/Cy3 MFRs of replicates (duplicates of the same RNA samples), repetitions (same comparison but different RNA samples) and dye switches (same samples but switching the fluorophores) were calculated and represented as statistical estimators of the differential expression of each gene. Raw experimental results were formatted according to the standards specified by the Microarray Gene Expression Data Society (MGED Society) and can be consulted in under the Accession No. E-MEXP-305.


Authors are indebted to E. Dominguez for the anti-XylS serum and to M. Tesar for the anti-C2,30 (BphC) monoclonal antibody. The help of S. Fraile and L.A. Fernandez is also gratefully acknowledged, as well as inspiring discussions with I. Cases, M. Valls and V. Shingler. This work was supported in part by EU grants of the fifth and the sixth Framework Programme.