The alcR gene of Aspergillus nidulans, which encodes the specific transactivator of the ethanol utilization pathway, is positively autoregulated and carbon catabolite repressed. Regulation by these two circuits occurs at the transcriptional level via the binding of the two regulators, AlcR and CreA, to their cognate targets respectively. We demonstrate here that out of two clustered putative AlcR repeated consensus sequences, only the palindromic target is functional in vivo. Hence, it is solely responsible for the alcR positive autogenous activation loop. Transcript mapping of the alcR gene showed that transcription initiation can occur at 553 bp and at or near 86 bp upstream of the start codon. These transcription start sites yield a transcript of 3.0 kb, which appears only under induced growth conditions, and of 2.6 kb, which is present under both induced and non-induced growth conditions respectively. Nine CreA consensus sites are present in the alcR promoter but only two pairs of two sites are functional in vivo. One of them is located in close proximity to the AlcR functional target. Within this pair, both sites are necessary to mediate a partial repression of alcR transcription. Disruption of either site results in an overexpression of alcR due to the absence of direct competition between AlcR and CreA for the same DNA region. The second functional pair of CreA sites is located between the two transcription initiation sites. Disruption of either of the two sites results in a totally derepressed alcR transcription, showing that they work as a pair constituting the more efficient repression mechanism. Thus, CreA acts by two different mechanisms: by competing with AlcR for the same DNA region and by an efficient direct repression. The latter mechanism presumably interfers with the general transcriptional machinery.
The utilization of less preferable carbon sources in Aspergillus nidulans is controlled by two regulatory mechanisms (Arst and Scazzocchio, 1985): specific induction, which is mediated by an activator that belongs generally to the zinc binuclear cluster family, and carbon catabolite repression, which is controlled by the CreA repressor, a protein with two zinc fingers of the Cys2–His2 type, whose homologues have been found in other fungal species. This is the case for the ethanol oxidation pathway (alc), which is one of the best-studied catabolic systems in filamentous fungi (reviews Felenbok, 1991; Felenbok and Sealy Lewis, 1994; Felenbok and Kelly, 1996; and references therein).
AlcR is the specific activator required for ethanol oxidation in A. nidulans. In the presence of an external coinducer (like ethanol or ketones), AlcR activates the coordinated expression of the two structural genes alcA and aldA (encoding alcohol- and aldehyde-dehydrogenases respectively) (Lockington et al., 1987). Their expression is necessary and sufficient for ethanol utilization (Fillinger and Felenbok, 1996). The function of a number of other alc genes clustered with the alcR–alcA genes on chromosome VII, and subject to the same regulatory circuits, is still unknown (Fillinger and Felenbok, 1996).
The alcR gene itself is likewise subject to the forementioned regulatory circuits. It is controlled at the transcriptional level by AlcR via a positive feedback loop and by CreA, which represses directly alcR transcription (Lockington et al., 1987; Mathieu and Felenbok, 1994). AlcR is a zinc binuclear cluster protein harbouring two zinc atoms that are necessary for DNA binding (Sequeval and Felenbok, 1994; Ascone et al., 1997). It contains an asymmetric DNA-binding domain with an extended loop between the third and fourth cysteine, but lacks the conserved proline residue (Kulmburg et al., 1991; Ascone et al., 1997). AlcR occurs in solution as a monomer, able to bind in vitro to single DNA targets (Cerdan et al., 1997; Lenouvel et al., 1997; Nikolaev et al., 1999a). Moreover, an arginine residue located in the N-terminus, outside the DNA-binding domain is involved in DNA-binding and transcriptional activation (Nikolaev et al., 1999b). Finally, AlcR is the only representative of this class of transcriptional activators whose functional targets are organized as repeats of the same core consensus (5′-T/AGCGG-3′), in either orientations (inverted or direct repeats) and with a variable spacing (Kulmburg et al., 1992a, b; Fillinger, 1996; Lenouvel et al., 1997; Panozzo et al., 1997). Deletion analysis of the alcR gene promoter indicated that the upstream activating region was located within 645 bp upstream of the initiation codon, encompassing two AlcR repeated consensus sites (Kulmburg et al., 1992a). Previous studies have shown that the inverted repeat can efficiently bind in vitro to an AlcR–His-tagged protein, whereas the direct repeat only binds weakly (Lenouvel et al., 1997). Therefore, it is not clear whether this latter sequence is functional in vivo.
A number of putative CreA binding sites have been identified in the alcR promoter region. One of these, in the direct vicinity of the AlcR repeats was found to be functional in vivo (Kulmburg et al., 1993). After disruption of this CreA site, a partial derepression was observed and an overinduction of AlcR was monitored (Mathieu and Felenbok, 1994). This strongly suggested that additional CreA sites could be involved in further repression. Moreover, substantial overexpression of the alcR gene was observed under non-repressive conditions. These data are consistent with a competition between AlcR and CreA for the same alcR promoter region (Mathieu and Felenbok, 1994), which occurs under all physiological growth conditions (Fillinger et al., 1995).
Here, we report on regulatory features within the alcR promoter and more specifically on the interplay between the two regulatory circuits, i.e. specific induction and carbon catabolite repression. We have identified in vivo, the sole functional AlcR induction target and four functional CreA repression sites, and analysed their individual contribution to the transcription of the transacting alcR gene, as well as on that of the structural alcA gene. The positive activation feedback of the alcR gene is not necessary for basal level transcription yielding active AlcR protein under conditions of induction. CreA was found to act in two distinct fashions. Besides the competition between AlcR and CreA, shown previously for one CreA site (Mathieu and Felenbok, 1994), we show here that a second CreA site is involved to form a functional pair. In addition, there is an efficient direct repression mediated via another pair of CreA sites, located between the two transcription initiation sites of alcR. These sites enable a complete repression of alcR transcription.
Transcript mapping of the alcR gene
Before starting these studies, we needed to establish the transcription initiation site(s) of the alcR gene. Interestingly, the alcR mRNA size is different under induced (I) and non-induced (NI) growth conditions. Previous analyses have shown that without an externally added inducer, a mRNA of 2.6 kb could be detected whereas an additional transcript whose size was estimated to be 2.8 kb, was visible upon induction (Lockington et al., 1987; Mathieu and Felenbok, 1994). Previous experiments of transcript mapping (Felenbok et al., 1988) indicated a single transcription start point. These primer extension experiments were carried out utilizing mRNA isolated from induced biomass of an alcR multicopy transformant. Moreover, our current results show that the oligonucleotide we used complements a sequence upstream of the non-induced (NI) transcription start point, which therefore could not have been localized.
Transcription initiation points were now identified on mRNAs isolated under non-induced and induced growth conditions using appropriate primers and a 3′/5′ RACE system (see Experimental procedures). After cloning of the PCR products, two types of clones were obtained from induced biomass, harbouring inserts of ≈ 110 bp and 580 bp respectively. Non-induced biomass yielded only clones with the smaller insert. These results confirm the existence of two distinct mRNAs under various physiological growth conditions, arising from two different transcription start points.
Sequence analysis of three independent clones revealed that the smaller insert initiates at or near 86 bp upstream of the start codon, corresponding to the transcription initiation point used in the absence of an external inducer. The longer mRNA species initiates at 553 bp with regard to the ATG. No additional intron could be identified within this longer mRNA, which could give rise to a supplementary protein sequence. Thus, the coding sequences are the same for both messengers. The differences between the non-induced and induced transcription start points are: (i) the localization vis-à-vis the putative regulatory sites (Fig. 1); and (ii) the initiation codon usage. The start of the coding sequence corresponds to the first ATG in the case of the smaller mRNA which is the ninth ATG in the case of the longer mRNA. The correlation between the two alcR transcription initiation sites and the sizes of the two alcR mRNAs is now established. The shorter transcript observed under both induced and uninduced conditions is 2.6 ± 0.1 kb long, whereas the one observed only in induced conditions is 3.0 ± 0.1 kb (results not shown). These results are in good agreement with previous evaluations performed without an RNA molecular weight marker (Lockington et al., 1987; Mathieu and Felenbok, 1994).
The −86 proximal start point (ST2) is the only one used in the absence of an inducer, but it is also used in its presence and lies downstream of all identified putative regulatory sequences. The distal −553 start point (ST1), used only upon administration of an external inducer, is situated downstream of both AlcR putative targets and CreA-targets A1 and A2 (see Fig. 1). Considering transcriptional regulation, it was of particular interest to specify the role of both regulatory proteins and their respective targets for each transcription initiation point.
Molecular basis for alcR positive feedback control
Figure 1 shows the localization of the AlcR and CreA putative binding sites in the alcR promoter. The palindrome a (encompassing a1 and a2 inverted repeats) close to the direct repeat b (encompassing b1 and b2 direct repeats) with the same AlcR consensus core, 5′-T/AGCGG-3′, are located at −643 and −663bp, respectively, upstream of the alcR initiation codon (Kulmburg et al., 1992a).
The effects of the disruption of both repeated sites a and b were tested in monocopy transformants (Tmab), integrated at the unlinked uaZ locus of an alcR-deletion mutant (see Experimental procedures). Figure 2 shows that these mutations do not appear to impair growth on ethanol and allyl alcohol sensitivity tests are confirming these results. Northern blot analysis of the modified alcR genes and that of the structural gene alcA, was performed (Fig. 3A). In the transformant Tmab, low constitutive levels of alcR and alcA expression are observed. It is noticeable (Fig. 3A), that the band corresponding to the alcR transcript of high-molecular weight, which appears under induced growth conditions in the control and in the wild type, is not observed in Tmab where only the shorter transcript is present.
These results show that AlcR positive activation feedback control essentially acts via AlcR targets. Moreover this mechanism is indispensable for transcription initiation of the 3.0 kb transcript at −553.
The palindromic AlcR binding site is the sole target responsible for specific transcriptional induction of the alcR gene
The direct repeat consensus b, binds weakly in vitro to the AlcR (1:197) His-tagged protein (Lenouvel et al., 1997). However, we know from functional analyses of the alcA promoter that in vitro AlcR-binding to direct repeats depends upon the flanking regions of the consensus core (Panozzo et al., 1997; Nikolaev et al., 1999b). Therefore, it is possible that in the alcR promoter the direct repeat, could be functional in vivo.
Figure 3B shows that the disruption of the direct repeat in the transformant Tmb, does not have any effect on alcR transcription, and consequently on alcA expression. This result excludes a possible role for the direct repeated sequences in alcR induction and contrasts with the situation found in the alcA promoter (Panozzo et al., 1997).
However, when the two inverted sites a1–a2 are disrupted in the transformant Tma, a drastic decrease of the induced transcription of both the alcR and alcA genes is observed (Fig. 3B). This decrease is similar to that observed in the transformant Tmab, in which both repeats are disrupted, where the same pattern of transcription is observed (Fig. 3A).
We can conclude that the AlcR palindromic site is the sole AlcR functional target in the alcR promoter and is fully responsible for the positive autogenous regulation.
The three CreA consensus sites located upstream of the AlcR induction target are not functional in vivo.
Three consensus CreA sites, D, E and F are located at −1052, −962 and −905 bp, respectively, upstream of the alcR translation start point (see Fig. 1). These were able to bind in vitro a CreA fusion protein [GST–CreA (35–240) (49 kDa), Kulmburg et al., 1993]. A blunt deletion of these three sites [from SalI (1064) to NdeI (721)] was performed in order to determine their functionality in repressing alcR transcription.
Transcriptional analysis of alcR and alcA in this transformant shows that the transcriptional behaviour of the alc genes is exactly the same as that found in the wild type (Fig. 4). Therefore, putative CreA-binding sites D, E and F, although able to bind in vitro a GST–CreA protein, are not functional in alcR repression in vivo.
The two CreA sites located between the two alcR transcription start points are functional in vivo.
We examined the role of three putative CreA sites C1, C2, C3 in alcR repression. These are located between the two alcR transcription starts ST1 and ST2 utilized under induced and, induced and uninduced growth conditions respectively (Fig. 1). The effects of individual disruption of sites C1, C2, C3 were monitored by Northern blot analysis. In the three transformants TmC1, TmC2 and TmC3, the induced transcription is similar to the wild type, but derepression is complete (100%) only in the strains TmC1 and TmC3 (Fig. 4). Putative CreA site C2 is thus not functional in alcR repression. This result is in agreement with the data of Cubero and Scazzocchio (1994) showing that this site (CTGGAG) is not protected in DMS interference experiments. In transformants TmC1 and TmC3 under glucose repressed growth conditions, as expected whenever the steady-state amounts of alcR transcript are increased, a significant derepression of alcA is observed.
CreA sites C1 and C3 act as a pair, as disruption of either site results in an almost complete derepression.
A pair of functional CreA sites overlaps the AlcR induction target.
Two further CreA binding sites, A1, and A2 are located in the same region as the AlcR functional induction palindrome target a; CreA site A1 is situated 9 bp upstream of AlcR site a1, whereas putative CreA site A2 completely overlaps it (Fig. 1). Disruption of CreA site A1, previously performed in another background, resulted in a partial derepression and a very strong overexpression of AlcR (Mathieu and Felenbok, 1994). However, subsequent Southern blot analysis has shown that three copies of alcR were present in this particular transformant, explaining the very strong overexpression. We decided to disrupt in the same genetic background as all other strains utilized in this study (see Experimental procedures), CreA sites A1 and A2 both individually and simultaneously, while selecting for single-copy transformants. As shown in Fig. 5, disruption of CreA site A1 (TmA1) or CreA site A2 (TmA2), or both CreA sites A1 and A2 simultaneously (TmA1A2), results in a transcriptional overinduction (3.5- to 4-fold) and a similar partial derepression (45–60%) of alcR in all three cases. As expected from the competition model, the alcA transcription level is increased. The alcA derepression observed here is due to increased amounts of the AlcR protein present under both induced and glucose growth conditions, while alcA remains under its genuine repression.
These results show that CreA sites A1 and A2 are both functional in repressing alcR transcription and that they act as a pair. It is important to note that the repression conferred through the CreA couple A1A2 is only partial, whereas that mediated via the CreA pair C1C3, is virtually complete.
This work aimed to dissect specific and wide domain transcriptional regulatory mechanisms and to investigate their interaction in alcR gene expression with respect to their cis-acting elements.
We have elucidated the mechanism of positive autoactivation of the alcR gene. One palindromic site, target a, is involved in the control of the alcR induced transcription. This functional element is very similar to other symmetric elements found in other alc responsive promoters (Fillinger, 1996). Interestingly, the mutagenesis of one nucleotide in the spacer region of the AlcR palindromic site a in the CreA-site mutant TmA2, does not affect induction. We confirmed here in vivo, our previous in vitro data, showing that the composition of the spacer region between symmetric alcR sites is irrelevant for AlcR binding (Nikolaev et al., 1999b). The lack of autoregulation, which is observed when the palindromic site is mutagenized, still permits growth on ethanol. The role of this upstream activating sequence, responsible for the positive autofeedback, is to amplify an external induction signal and results in a relatively high level of alcR expression compared with those of some other responsive alc genes (Fillinger and Felenbok, 1996).
The utilization of the two transcription initiation sites in the alcR gene, separated by 467 bp, notably depends upon the presence of an externally applied coinducer. Without an external coinducer the proximal site is used which corresponds to a 2.6 kb long mRNA. The presence of an external coinducer leads in addition to synthesis of a second longer transcript (3.0 kb), which is, however, dispensable for induced transcription. The palindromic AlcR target a is necessary for induced initiation at both the distal and the proximal transcription start sites. The messengers share the same coding sequence, both therefore lead to a functional AlcR protein able to activate transcription of the alc structural genes. Differential transcription initiation opens the possibility for an additional regulatory checkpoint of AlcR expression, concerning translation.
The other regulatory circuit controlling alc gene expression, is carbon catabolite repression, mediated by the repressor CreA (Bailey and Arst, 1975; Dowzer and Kelly, 1991; Felenbok and Kelly, 1996). This repressor acts at two levels: (i) by repressing directly alcR expression and thereby that of alcR-responsive genes; and (ii) by repressing directly and independently most of the responsive genes, although not all. Among the nine CreA consensus sites present in the alcR promoter, two pairs are shown here to be functional in vivo. CreA site A1 is close to the AlcR palindromic target a whereas CreA site A2 overlaps it. We have previously shown that the CreA site A1 is involved only partially in glucose repression of the alcR gene (Mathieu and Felenbok, 1994). We show here that the second CreA site A2 is likewise involved in a similar fashion, while the two sites are working as a pair. The direct competition for the same alcR promoter region between the two regulators involves both CreA sites A1 and A2. A similar situation occurs in the alcA promoter in which two functional CreA sites eclipse one of the AlcR induction targets (Panozzo et al., 1998). In both cases, AlcR and CreA are exclusive antagonists in the control of the expression of these two alc genes.
The respective levels of the AlcR and the CreA proteins in the cell are also determinant. Mutagenesis of CreA sites A1 and A2 resulted in an overexpression of the alcR gene (Mathieu and Felenbok, 1994; and this work). As a consequence, the alcA structural gene is overinduced and partially derepressed. In the case of alcR, the positive feedback loop plays a key role in the regulatory process. However, repression mediated by CreA via these overlapping sites is only partial, which strongly indicates the involvement of other functional CreA sites.
Mutagenesis of the second pair of functional CreA sites, C1 and C3, does not change the induced steady-state level of alcR gene transcription and overexpression could not be observed. This indicates a direct mechanism of CreA repression irrespective of alcR activation, which does not occur via competition with AlcR. Mutagenesis of either of these two sites results in an almost total derepression of alcR transcription showing that these sites act also as a pair and are the most prominent for CreA repression of alcR. The localization of these sites between the two alcR transcription starts may be an important feature in the efficiency of the repression process, as this region could become inaccessible to the transcriptional machinery.
We clearly show here that CreA controls alcR expression by two different mechanisms. The first one involves a competition with the activator AlcR via the CreA sites overlapping the functional AlcR induction target, although this repression is only partial. However, it is expected to play a decisive role under physiological growth conditions in which an inducer of the alc regulon is present. The second repression mechanism, mediated via the CreA targets localized between the two transcription starts is drastic and presumably affects the action of the transcriptional machinery. It should be very efficient under physiological conditions in which a rich carbon source is present. Other molecular mechanisms, such as promoter bending by the influence of another protein and/or nucleosome positioning as shown in the nitrate utilization pathway (Muro-Pastor et al., 1999) and in 5′ region of the amdS gene (Narendja et al., 1999), might be involved in transcription initiation site selection. Moreover, the coexistence of two populations of alcR messengers opens the possibility that other regulatory circuits affect alcR expression, for example the level of translation initiation or elongation.
Strains, media and growth conditions
The A. nidulans strain used as the recipient in transformation was C1R7 (alcR7, yA2, paba A1 and uaZ11) (Nikolaev et al., 1999a). Media and supplements were as described by Cove (1966). Mycelia were grown submerged for 8 h at 37°C using 0.1% fructose as the sole carbon source. Induction was achieved by adding the gratuitous inducer ethyl methyl ketone at 50 mM and biomass was harvested after a further 2.5 h of incubation (induced conditions). For repressed conditions, 1% glucose was added simultaneously with the inducer.
Transcript start mapping
The 5′ end of the alcR mRNA was determined using the 5′/3′ RACE kit from Roche-Diagnostics according to the method described by Frohman et al. (1988). First-strand cDNA was synthesized from 2 µg of total RNA extracted from the wild-type strain grown under non-induced or induced conditions, respectively, using the alcR– specific primer, AlcRSP1, overlapping the unique intron in alcR. All subsequent steps were carried out according to the supplier’s protocol. Two PCR reactions were performed to amplify the 5′ RACE product prior to cloning. The alcR-specific nested primers, AlcRSP2 and AlcRSP3, are displayed below.
The NI RNA led to a small PCR product (≈110 bp in length) and the I RNA sample gave rise to two products, the same small one and a more abundantly present, longer fragment of about 580 bp. The isolated fragments were digested with SalI and cloned into pBluescriptSK+ linearized with SalI and EcoRV. Three independent clones of each product were subjected to sequence analysis using the commercial SK-primer, which allows you to read the complete cDNA insert. In the case of the smaller product, we considered the longest extension product, the others being up to nine bases smaller. The longer extension product mapped precisely to position −553. The oligonucleotides used are:
The palcR mutant plasmids harbouring the investigated mutations within the AlcR or the CreA binding sites in the alcR promoter were generated by oligonucleotide-directed mutagenesis (Kunkel et al., 1987). A uracil–bluescript–KS template, containing the appropriate DNA fragment, SalI to AhaII(1–1081) (Kulmburg et al., 1991) was mutagenized with oligonucleotides:
The double mutagenesis involved both oligonucleotides. All mutations were confirmed by DNA sequencing. The SalI–HindIII (1–844) restriction fragment was cloned into the alcR Bluescript plasmid (1–3916) (Felenbok et al., 1988). To select for transformants integrated at the ectopic uaZ locus, the filled-in ClaI–XhoI fragment of the uaZ gene was inserted into the filled-in EcoRI site of the aforementioned alcR plasmid. The resulting plasmids were used to transform the A. nidulans recipient strain C1R7 following Tilburn’s procedure (Tilburn et al., 1983). Construction of a plasmid bearing the mutated CreA site A1 and A2 was carried out in a similar way as described above for the AlcR site b1b2 and a1a2. The following oligonucleotides were used:
The oligonucleotides used for mutation of CreA sites C1, C2 and C3 were:
Cloning of the mutagenized fragment in alcR was in this instance achieved by substituting the HindIII–NcoI restriction fragment with the corresponding mutated sequence in the alcR Bluescript plasmid.
Isolation of RNA and quantitative analysis
Total RNA was isolated from A. nidulans as described by Lockington et al. (1987) and separated on glyoxal agarose gels according to Sambrook et al. (1989). The entire genes of alcR and alcA, cloned into Bluescript plasmids, were used as probes (Felenbok et al., 1988; Gwynne et al., 1987). The membranes were further hybridized with a probe for the A. nidulans actin gene (Fidel et al., 1988), as an internal control, to estimate the amount of specific mRNAs relative to that of actin mRNA. Autoradiographs were exposed for various time periods, to avoid saturation of the film. Densitometric scanning was performed with a system Biosoft-Orkis. The intensities of the signals were also quantified using a PhosphorImager (Molecular Dynamics). Experiments were repeated three times.
It should be mentioned that the quantification of mRNA should reflect a steady-state level. Formally, there is a possibility that changes in carbon source and therefore growth rate could affect e.g. mRNA stability. In order to avoid such effects the time period for induction and repression of alc gene expression was kept as short as possible.
This work was supported by grants from the Centre National de la Recherche Scientifique (UMR no. 8621), the Université Paris-Sud XI, and from the European Communities contract no. BIO4-CT96-0535. We thank Dr M. Blight for correcting the English and Dr M. Flipphi for helpful discussions and critical reading of the manuscript. The nucleotide sequence reported in this paper has been submitted to GenBank (EBI data bank with accession number G168009alcR).
†Present address: Unité de Physiologie Cellulaire, Département des Biotechnologies, Institut Pasteur, 28, rue du Docteur Roux, 75724 Paris Cedex 15, France.