The AlkS regulator, encoded by the alkS gene of the Pseudomonas oleovorans OCT plasmid, activates the expression of a set of enzymes that allow assimilation of alkanes. We show that the AlkS protein regulates, both negatively and positively, the expression of its own gene. In the absence of alkanes, alkS is expressed from promoter PalkS1, which is recognized by σS-RNA polymerase, and whose activity is very low in the exponential phase of growth and considerably higher in stationary phase. AlkS was found to downregulate this promoter, limiting expression of alkS in stationary phase when alkanes were absent. In the presence of alkanes, AlkS repressed PalkS1 more strongly and simultaneously activated a second promoter for alkS, named PalkS2, located 38 bp downstream from PalkS1. Activation of PalkS2 allowed efficient transcription of alkS when alkanes were present. Transcription from PalkS2 was modulated by catabolite repression when cells were provided with a preferred carbon source. We propose that the expression of alkS is regulated by a positive feedback mechanism, which leads to a rapid increase in alkS transcription when alkanes are present. This mechanism should allow a rapid induction of the pathway, as well as a fast switch-off when alkanes are depleted. An improved model for the regulation of the pathway is proposed.
Pseudomonas oleovorans GPo1 harbours a large plasmid, named OCT, encoding the enzymes required to oxidize medium chain length n-alkanes (C6 to C12) to the corresponding terminal acyl-CoA derivatives, which then enter the β-oxidation cycle (reviewed by van Beilen et al., 1994; Fig. 1). The genes coding for these enzymes, which comprise the so-called alk pathway, are grouped in two clusters, alkBFGHJKL and alkST (Fig. 1). The alkS gene encodes a transcriptional regulator, which, in the presence of alkanes, activates the PalkB promoter, from which the alkBFGHJKL operon is expressed (Kok et al., 1989a; Panke et al., 1999). Expression of the PalkB promoter is also modulated by catabolite repression depending on the carbon source being used (Yuste et al., 1998; Staijen et al., 1999). When cells are grown in rich Luria–Bertani (LB) medium, repression is believed to occur by a mechanism that impedes AlkS to activate transcription (Yuste et al., 1998). The alkS gene is under the influence of a promoter recognized by a form of RNA polymerase (RNAP) bound to the alternative sigma factor σS (Canosa et al., 1999), a sigma factor that directs the expression of many stationary phase and stress-induced genes (reviewed by Loewen and Hengge-Aronis, 1994; Hengge-Aronis, 1996). The activity of this promoter is very low during the exponential phase of growth, but increases considerably when cells enter the stationary phase (Canosa et al., 1999).
This model for the regulation of the alk pathway poses the question as to why alkS is expressed from a promoter that is more active in stationary phase than in exponential phase, when the amounts of AlkS protein present in exponential phase are enough to achieve full induction of the pathway. As transcriptional activators frequently repress expression of their own gene to limit their concentration in the cell, we considered that AlkS may downregulate the activity of the σS-dependent promoter in stationary phase. Previous analyses of the expression of alkS have been performed with a strain that contained a transcriptional fusion of the alkS promoter region to the lacZ gene, but lacked the alkS gene, and therefore could not show the influence of AlkS on the expression of its own gene. We have now studied this issue in detail, finding that the AlkS protein regulates both negatively and positively the expression of its own gene. Autoactivation occurs from a previously unknown promoter, generating a positive feedback loop that is modulated by catabolite repression. The results presented here extend our understanding of the regulation of the alk pathway and allow us to propose a more complete model, which explains issues that were formerly poorly understood.
AlkS represses PalkS1 and activates a second promoter for alkS
To determine whether the AlkS protein can regulate the expression of its own gene, we analysed the effect of introducing increasing copies of the alkS gene into the reporter strain PS16, which lacks alkS but contains in the chromosome a transcriptional fusion of the alkS promoter region to the lacZ reporter gene (positions − 344 to + 53 relative to the start site of PalkS1, the previously mapped σS-dependent promoter; Yuste et al., 1998; Canosa et al., 1999). Strain PS16 derives from Pseudomonas putida KT2442, in which the alk genes are correctly expressed, maintaining all known regulatory features (Yuste et al., 1998; Canosa et al., 1999). In fact, P. putida and the strain from which the OCT plasmid was originally isolated, P. oleovorans, are closely related species (Moore et al., 1996). The alkS gene was introduced into strain PS16 either with the help of a mini-Tn5 transposon, which allows the insertion of a single copy of alkS in the chromosome, or using the broad-host-range, high-copy-number plasmid pHCS1, which contains alkS (Yuste et al., 1998). Cells containing the PalkS–lacZ fusion and either no alkS (strain PS16), a single copy of alkS (strain PS16S1) or multiple copies of alkS (strain PS16/pHCS1) were grown in LB medium in the absence or presence of the non-metabolizable inducer dicyclopropylketone (DCPK), a water-soluble compound that mimics the effect of alkanes (Grund et al., 1975). Samples were obtained at different growth stages and processed to obtain total RNA. Transcripts arising from the alkS promoter located upstream from the lacZ reporter gene were analysed by S1 nuclease protection assays, using as probe a single-stranded DNA (ssDNA) whose labelled 5′ end hybridized to the start of the lacZ gene. In this way, only the transcripts arising from the PalkS–lacZ fusion were detected, whereas those arising from the promoter of the introduced alkS gene were not. As reported previously (Canosa et al., 1999), in the absence of alkS, expression of the PalkS–lacZ fusion was very low during exponential growth and increased considerably as cells entered the stationary phase, because expression occurs from a σS-dependent promoter (Fig. 2). Transcription from this promoter was very similar in both the absence or the presence of inducer. A weak additional signal, however, could be detected in stationary phase cells, suggesting the presence of a transcript 38 bp shorter than the former one. In the case of cells containing a single copy of alkS (strain PS16S1), the signal corresponding to the transcript arising from the σS-dependent promoter was considerably weaker in the absence of inducer (about 10-fold lower, as deduced from samples run in parallel in the same gel; not shown), and essentially undetectable when the inducer was present. At the same time, the inducer led to a considerable increase in the signal corresponding to the transcript 38 bp shorter than that arising from the σS-dependent promoter (Fig. 2). This new band was detectable only in stationary phase cells. When multiple copies of alkS and the inducer were present (strain PS16/pHCS1), the abundance of this transcript increased significantly, particularly in exponential phase. The transcript appeared as a set of four or five bands, each one being 1 bp shorter than the preceding one, suggesting heterogeneity in its 5′ end (this can be better appreciated in Fig. 6). A similar heterogeneity has been observed in several prokaryotic promoters whose start sites include a tract of three or more thymines (Harley et al., 1988; Guo and Roberts, 1990; Jacques and Susskind, 1990; Gülland and Hillen, 1992; Xiong and Reznikoff, 1993; Jin, 1994). At these promoters, the initial ribonucleotide pAAA or pUAAA slips against the template TTT, beginning a slippage cycle in which the transcript acquires a number of extra untemplated adenines at the 5′ end. Finally, an additional weak band about 15 bp shorter appeared in some assays; this band was not reproducible and was not considered further.
The above results suggest that AlkS can repress the σS-dependent promoter in both the absence and the presence of inducer and that, when the inducer is present, it activates a new promoter located 38 bp downstream from the σS-dependent promoter. As shown in Fig. 2, we have named the σS-dependent promoter PalkS1 (or P1), and the new AlkS-activated promoter PalkS2 (or P2). Repression of PalkS1 and the simultaneous activation of PalkS2 in the presence of inducers could easily be explained if we assume that AlkS binds to PalkS1, a site at which the regulator would be adequately positioned to activate PalkS2 (see Fig. 2B). Interestingly, comparison of promoters PalkS2 and PalkB, both of which are activated by AlkS, reveals the presence of a highly homologous inverted repeat immediately upstream from the − 35 region (Fig. 2B). Accurate comparison of promoters PalkS2 and PalkB requires knowledge of the precise start site of PalkB, but the reported transcription initiation site for PalkB had been deduced by S1 protection assays using a set of end-labelled restriction fragments as size markers, which does not allow for precise mapping (Kok et al., 1989a). Using a DNA sequence ladder as size marker, we have now found that the start site for PalkB is 5 bp further upstream than previously proposed (see Figs 3 and 2B). Considering this new start site, the mentioned inverted repeat is centred at position − 41.5 at PalkB and at position − 42.5 at PalkS2, relative to their respective start sites (Fig. 2B). To investigate whether this inverted repeat could be the target for AlkS, we generated a series of PalkS2 derivatives in which the sequences upstream from position − 36 were deleted stepwise (see Fig. 4). The resulting promoters were fused to lacZ, introduced into a P. putida KT2442 derivative containing the alkS gene (strain PS1), and the ability of AlkS to activate transcription from each of them was analysed. As shown in Fig. 4, sequences upstream from position − 59 were dispensable for AlkS activation of promoter PalkS2. It should be noted, however, that deletion of sequences between positions − 120 and − 59 increased the basal expression levels of PalkS2 in the absence of inducer, although the presence of the inducer stimulated transcription significantly. Extending the deletion up to position − 42, which eliminates the left arm of the repeat, or to position − 36, which affects the right arm as well, rendered promoters that could not be activated by AlkS. Therefore, PalkS2 activation seems to require only the DNA region comprising the inverted repeat centred at position − 42.5.
It is worth noting that, when alkS was in monocopy, AlkS-dependent expression of promoter PalkS2 was detected only after cells entered the stationary phase of growth (A600 of 2 or 3). This behaviour is similar to that of promoter PalkS1, whose expression is much higher in stationary phase because it depends on σS-RNAP (Canosa et al., 1999). However, in the presence of multiple copies of alkS, activation of PalkS2 by AlkS was efficient both in the exponential and in the stationary phases of growth. This is reminiscent of the behaviour of the PalkB promoter in cells grown in LB medium and an inducer, a situation in which PalkB activation is silenced by catabolic repression during exponential growth unless AlkS is present in high amounts (Yuste et al., 1998). We therefore investigated whether the preferential expression of PalkS2 in stationary phase was caused by a dependence on σS or by a catabolite repression effect of the LB medium used to grow the cells.
Promoter PalkS2 is expressed in cells lacking σS-RNAP
The possible influence of σS on the expression of PalkS2 was analysed in strain PSPS1, a rpoS-deficient derivative of P. putida KT2442 that contains the alkS gene inserted into the chromosome. Analysis of the transcripts arising from the alkS promoters in this strain showed that, in the presence of inducer, PalkS2 was active despite the absence of σS (Fig. 5). As described previously (Canosa et al., 1999), expression from PalkS1 was undetectable. This result is consistent with the absence in PalkS2 of a sequence at the −10 region resembling the consensus characteristic of σS-dependent promoter (CTATACT; Espinosa-Urgel et al., 1996), a sequence that is present in PalkS1 (Fig. 2B).
Activation of promoter PalkS2 is modulated by catabolite repression
When cells are grown in a minimal salts medium, AlkS-dependent expression of the promoter for the alkBFGHJKL operon, PalkB, is three- to fourfold lower when lactate or succinate are used as a carbon source than when citrate is used as carbon source (Yuste et al., 1998; Staijen et al., 1999). Expression decreases more than 30-fold when cells are grown in rich LB medium (Yuste et al., 1998). Repression in LB medium is relieved when cells enter the stationary phase or when multiple copies of alkS are present, which suggests that it is mediated by a factor that interferes with AlkS function (Yuste et al., 1998). As promoter PalkS2 is also activated by AlkS, we analysed whether growth of cells at the expense of different carbon sources had any effect on PalkS2 expression. For this purpose, strain PBS4 was used, which is a derivative of KT2442 harbouring the alkS gene and a PalkB–lacZ fusion inserted into the chromosome, and is the strain used to analyse catabolite repression on the PalkB promoter (Yuste et al., 1998). Transcription from PalkS2 was monitored by S1 nuclease protection assays using a probe whose labelled 5′ end hybridized to the start of the alkS gene. As shown in Fig. 6, when cells were grown in a minimal salts medium containing citrate as carbon source, promoter PalkS2 was efficiently activated soon after addition of the inducer, and its activity was high in both the exponential and the stationary phases of growth. As occurred at PalkB (Yuste et al., 1998), the use of lactate as a carbon source led to a delay in the start of PalkS2 expression after addition of the inducer. This delay was much more pronounced when cells were grown in LB medium, in which PalkS2 expression was restricted to stationary phase. When a multicopy plasmid containing alkS was introduced into strain PBS4, the repressing effect on PalkS2 in exponential phase was significantly relieved (Fig. 6). The expression pattern of PalkS2 in the different growth media used is very similar to that described previously for the PalkB promoter (Yuste et al., 1998), indicating that both promoters are subject to a similar modulation by catabolic repression. On the other hand, expression of promoter PalkS1 was similar under all growth conditions tested (not shown).
The results reported here show that alkS is expressed from two promoters, PalkS1 and PalkS2, which are both regulated by AlkS. This allows us to propose a modified model for the regulation of the alk pathway, which is summarized in Fig. 7. In the absence of alkanes, alkS is expressed primarily from promoter PalkS1, which depends on σS-RNAP and provides very low expression levels during exponential growth and higher levels in stationary phase (Canosa et al., 1999). The AlkS protein, however, downregulates expression of PalkS1, maintaining a low activity of the promoter in stationary phase. Expression of alkS from a σS-dependent promoter in the absence of alkanes ensures the presence of very low levels of AlkS when cells grow exponentially at the expense of carbon sources other than alkanes. It seems clear that alkanes are not preferred carbon sources, and cells avoid efficient expression of alkS when other carbon sources are present. Entry into stationary phase would lead to a higher expression of alkS from the σS-dependent promoter PalkS1, allowing the cell to be ready to sense alkanes efficiently when these eventually become available. In the presence of alkanes, the AlkS regulator switches on the expression of the alkBFGHJKL operon by activating promoter PalkB (Kok et al., 1989a), and stimulates promoter PalkS2 as well (this work). AlkS activation of promoter PalkS2, despite being paralleled by increased repression of promoter PalkS1, would lead to a rapid increase in the amounts of AlkS protein.
The AlkS protein is a member of the LuxR family of regulators (De Schrijver and De Mot, 1999; Panke et al., 1999). Many of these proteins activate transcription by binding to a 20 bp inverted repeat located just upstream of the − 35 region of the cognate promoter (reviewed by Fuqua et al., 1996), although the target is more complex for some members of the family (Vidal-Ingigliardi et al., 1991). Interestingly, comparison of promoters PalkB and PalkS2, both of which are activated by AlkS, shows the presence of a highly homologous 20 bp inverted repeat centred at positions − 41.5 and − 42.5 respectively. A deletion analysis showed that elimination of the DNA sequences upstream from position − 59 relative to the PalkS2 start site, which conserves the inverted repeat but eliminates sequences immediately upstream from it, does not impair activation of PalkS2 by AlkS. Deeper deletions affecting the inverted repeat resulted in promoter derivatives that could not be stimulated by AlkS. Therefore, our data suggest that AlkS activates PalkS2 by binding upstream from the promoter, presumably recognizing the inverted repeat centred at − 42.5. The proposed target for AlkS at PalkS2 overlaps the −10 region of PalkS1, so that binding of AlkS to it would simultaneously activate PalkS2 (if alkanes are present) and repress PalkS1. AlkS can repress PalkS1 in the absence of alkanes, which suggests that it can bind DNA in the absence of inducers. Repression of PalkS1 was more efficient under induced conditions, perhaps because of a higher occupancy of its target as a result of the increase in AlkS concentrations, or because the AlkS protein has higher affinity for DNA when bound to the inducer.
Activation of PalkS2 upon addition of the inducer was fast when cells grew at the expense of citrate, but was delayed when lactate was provided as the carbon source and was severely repressed when a complete medium such as LB was used. Repression in LB medium was relieved when cells reached the stationary phase of growth, or when multiple copies of alkS were present, which has been shown to increase the amounts of AlkS protein in the cell (Yuste et al., 1998). This indicates that activation of promoter PalkS2 is modulated by catabolic repression in a way that is similar to that observed at the PalkB promoter (Yuste et al., 1998; Staijen et al., 1999).
Therefore, unless a preferred carbon source is present in the environment, the availability of alkanes would generate a fast induction of the pathway through a positive feedback mechanism that is governed by the AlkS activator. Autoamplification of AlkS levels in the presence of alkanes presumably provides a fast increase in AlkS concentrations, thereby allowing for a rapid synthesis of the enzymes required to assimilate alkanes. Autoamplification is a frequent strategy for regulating stress responses or developmental programmes, such as the expression of virulence factors (Stachel and Zambryski, 1986; Roy et al., 1990; Scarlato et al., 1990; Soncini et al., 1995) or of genes involved in sporulation (Strauch et al., 1993) or competence (van Sinderen and Venema, 1994). Probably, autoactivation in these systems serves to commit the cell to a given response once the appropriate signal is sensed, ensuring that the response is not initiated inappropriately and allowing for a rapid induction once the decision is taken. Activation of catabolic pathways should not need such a sophisticated control and, to our knowledge, has not been described before for a biodegradation pathway. Cascade amplification systems, however, are sometimes found to connect different segments of a pathway that can operate either independently or in a co-ordinated fashion; the best studied example is that of the P. putida TOL pathway for toluene and benzoate (reviewed by Ramos et al., 1997). A possible advantage of a positive feedback mechanism in the alk pathway is that it provides a fast induction, as well as a fast switch-off when alkanes are consumed. A rapid switch-off may be particularly important. It is known that the first enzyme of the pathway, the alkane hydroxylase, is induced to very high levels in the presence of alkanes, which has been shown to be detrimental to cell physiology (Chen et al., 1996). This enzyme is composed of a membrane-bound hydroxylase and two soluble proteins, rubredoxin and rubredoxin reductase, which act as electron carriers between NADH and the hydroxylase (Peterson et al., 1966). The alkB and alkG genes, specifying the hydroxylase and the rubredoxin respectively, are expressed from the PalkB promoter (see Fig. 1; Kok et al., 1989a,b). Rubredoxin reductase, encoded by the alkT gene, is located immediately downstream from the alkS gene, which led to the proposal that it could be transcribed from the same promoter as alkS (Eggink et al., 1988; 1990). We have evidence showing that the expression of alkT is induced in the presence of alkanes, although it is unknown whether transcription occurs from promoter PalkS2 or from another uncharacterized promoter, also activated by AlkS. As the role of AlkT protein is to transfer electrons from NADH to the rest of the alkane hydroxylase system (Peterson et al., 1966), a fast switch-off of alkT expression may provide a very efficient way to stop the drain of electrons into alkane hydroxylase once the enzyme is not needed further, therefore saving the metabolically expensive NADH for other purposes.
Bacterial strains and plasmids
All P. putida strains used derive ultimately from strain KT2442 (Franklin et al., 1981), a derivative of P. putida mt-2. P. putida PS16 (Canosa et al., 1999) contains a PalkS–lacZ transcriptional fusion inserted into the chromosome (the alkS promoter region spanned positions − 344 to + 53 relative to PalkS1 start site). P. putida PS16S1 was obtained by insertion of a copy of the alkS gene into the chromosome of strain PS16 with the help of the suicide donor plasmid pSS1. This plasmid derives from the delivery vector mini-Tn5-Sm (de Lorenzo and Timmis, 1994) by insertion at its NotI site of a NotI DNA fragment containing the alkS gene (including 627 nucleotides upstream from the initiation codon and 387 nucleotides downstream from the stop codon), obtained in turn from plasmid pUJS1 (Yuste et al., 1998). P. putida PSPS1 (Canosa et al., 1999) contains the PalkB–lacZ fusion and the alkS gene inserted into the chromosome, as well as an insertion mutation in the rpoS gene. P. putida PBS4 (Yuste et al., 1998) contains a PalkB–lacZ transcriptional fusion and the alkS gene inserted into the chromosome.
Deletion derivatives of promoter PalkS2 were obtained by polymerase chain reaction (PCR), using as forward primers a set of oligonucleotides whose 5′ ends contained an EcoRI site followed by sequences complementary to PalkS2 nucleotides downstream from positions −180, −120, −80, −59, −42 or −36 relative to the transcription start site. As reverse primer, an oligonucleotide was used that included a BamHI site at its 5′ end followed by PalkS2 sequences + 15 to − 2. The amplified DNA fragments were digested with EcoRI and BamHI, and fused to the lacZ reporter gene by cloning them between the EcoRI and BamHI sites of plasmid pUJ8 (de Lorenzo and Timmis, 1994), obtaining plasmids pUJS1, pUJS2, pUJS3, pUJS4, PUJS5 and pUJS6 (deletions up to position − 180, − 120, − 80, − 59, − 42 and − 36 respectively). After verification of the sequence, the transcriptional fusions were excised from these plasmids as a NotI segment and cloned into the NotI site of the delivery vector mini-Tn5 Km (de Lorenzo and Timmis, 1994). The resulting plasmids (pUTK1, pUTK2, pUTK3, pUTK4, pUTK5 and pUTK6 respectively) were used to insert the transcriptional fusions into the chromosome of P. putida PS1, which contains the alkS gene in its chromosome, obtaining strains PS1S1, PS1S2, PS1S3, PS1S4, PS1S5 and PS1S6 respectively. P. putida strain PS1 was obtained by insertion of a copy of alkS into the chromosome of strain KT2442, using the mini-Tn5 delivery plasmid pTLS1, which has been described previously (Yuste et al., 1998).
Plasmid pHCS1 (Yuste et al., 1998) contains the alkS gene inserted into the broad-host-range plasmid pKT231. Plasmids pUJPS16 and pTS1 have been described before (Yuste et al., 1998). Plasmids pHCS1 and pSS1, as well as the mini-Tn5 derivatives, were mobilized from Escherichia coli to P. putida by conjugation, as described previously (Yuste et al., 1998).
Media and culture conditions
Cells were grown at 30°C in rich LB medium or in minimal salts M9 medium (Sambrook et al., 1989), the latter supplemented with citrate or lactate (30 mM) as carbon source and with trace elements (Bauchop and Eldsen, 1960). To induce expression of PalkB or PalkS2 promoters, the non-metabolizable inducer DCPK was added up to 0.05% (v/v) when cultures reached an absorbance (A600) of 0.08.
S1 nuclease protection assays
Total RNA was isolated from bacterial cultures as has been described previously (Monsalve et al., 1995). S1 nuclease reactions were performed as has been described previously (Ausubel et al., 1989), using 25 μg of total RNA and an excess of a 5′ end-labelled single-stranded DNA (ssDNA) hybridizing to the 5′ region of the mRNA. This allows the detection of the transcription start sites as well as the amounts of transcripts generated. The ssDNA probe was generated using linear PCR as has been described previously (Yuste et al., 1998), employing as substrates plasmids pUJPS16 (contains the PalkS–lacZ fusion; Yuste et al., 1998) or pTS1 (contains alkS and 627 nucleotides upstream from the alkS translation start site; Yuste et al., 1998). The primers used were oligonucleotides lac1 (5′-TCAATCGCTGCTTTTACCAG) and alkS3 (5′-CCTTAGCGACCGGGAAATC) respectively. Before use as templates for the amplification reactions, plasmids were cut with NotI (pUJPS16) or HindIII (pTS1), whose targets are located more than 400 nucleotides upstream from the PalkS start site.
Assays for β-galactosidase
Overnight cultures were diluted to a final density of about 0.04 in minimal salts M9 medium supplemented with citrate as carbon source. Cultures were grown in duplicate at 30°C and, when an optical density of 0.08 was reached (at 600 nm), expression of promoter PalkS2 was induced in one of the samples by the addition of the non-metabolizable inducer DCPK up to 0.05% (v/v); the other sample remained as a non-induced control. At different cell densities, aliquots were taken, and β-galactosidase activity was measured as has been described previously (Miller, 1972).
We are grateful to José Pérez-Martı´n for helpful discussions and critical reading of the manuscript. This work was supported by grants 07M/0720/1997 from Comunidad Autónoma de Madrid and BIO97-0645-C02-01 from Comisión Interministerial de Ciencia y Tecnología.