On the Mechanism by which Alkaline pH Prevents Expression of an Acid-Expressed Gene

Abstract

Previous work has shown that zinc finger transcription factor PacC mediates the regulation of gene expression by ambient pH in the fungus Aspergillus nidulans. This regulation ensures that the syntheses of molecules functioning in the external environment, such as permeases, secreted enzymes, and exported metabolites, are tailored to the pH of the growth environment. A direct role for PacC in activating the expression of an alkaline-expressed gene has previously been demonstrated, but the mechanism by which alkaline ambient pH prevents the expression of any eukaryotic acid-expressed gene has never been reported. Here we show that a double PacC binding site in the promoter of the acid-expressed gabA gene, encoding γ-aminobutyrate (GABA) permease, overlaps the binding site for the transcriptional activator IntA, which mediates ω-amino acid induction. Using bacterially expressed fusion proteins, we have shown that PacC competes with IntA for DNA binding in vitro at this site. Thus, PacC repression of GABA permease synthesis is direct and occurs by blocking induction. A swap of IntA sites between promoters for gabA and amdS, a gene not subject to pH regulation, makes gabA expression pH independent and amdS acid expressed.

A form of gene regulation common in the microbial world, enabling organisms to adapt to differing environments, is control of gene expression by ambient pH. Such regulation tailors the syntheses of molecules operating outside the protection of the organism's internal pH homeostatic system, such as permeases, secreted enzymes, and exported metabolites, to the pH of the growth environment. Thus, for example, the syntheses of molecules which are effective only at acidic pH can be restricted to acidic environments. In the fungus Aspergillus nidulans, the zinc finger-containing transcriptional regulator PacC mediates such pH regulation (17, 18, 34). In response to a signal transduced by the products of the six pal genes (3, 8, 13, 14, 24, 28, 29) at alkaline ambient pH, the 674-residue full-length form of PacC is proteolyzed, yielding the functional form, containing the ∼249 N-terminal residues, which facilitates the expression of genes expressed at alkaline ambient pH and prevents the expression of genes expressed under acidic growth conditions (27, 30). Interactions involving three regions of PacC are responsible for maintaining the full-length form in a protease-inaccessible conformation in the absence of signal transduction (16a). Loss-of-function mutations in the pacC gene and the pH-signaling pal genes have an acidity-mimicking phenotype, whereas gain-of-function mutations, designated pacC^c, obviate the need for pH signal transduction and have an alkalinity-mimicking phenotype (27, 30, 34). pacC is itself an alkali-expressed gene (30, 34).

PacC is able to bind DNA, the consensus site being 5′-GCCARG (18, 27, 30, 34). A direct role for PacC in activating the expression of the alkali-expressed ipnA gene, encoding isopenicillin N synthase, has been demonstrated (16). The mechanism by which expression of a eukaryotic acid-expressed gene is prevented at alkaline pH has, however, never been investigated. An intriguing aspect is that PacC consensus sites tend to be less frequent in promoters of acid-expressed genes than in those of alkali-expressed genes (20, 23, 32, 34).

The acid-expressed gabA gene (2, 6, 20), encoding γ-aminobutyrate (GABA) permease, is subject to four known forms of regulation: ω-amino acid induction mediated by the zinc binuclear cluster transcription factor IntA (AmdR) (1, 2, 12), nitrogen metabolite repression mediated by the GATA transcription factor AreA (4, 21, 36), carbon catabolite repression mediated by the zinc finger transcription factor CreA (5, 15, 22), and ambient pH regulation mediated by PacC. Here we show that a double PacC binding site overlaps an IntA binding site and that a PacC fusion protein competes with an IntA fusion protein for DNA binding in vitro at this site. Thus repression of GABA permease synthesis at alkaline pH occurs through the prevention of induction. We further show that swapping a hexadeca- or heptadecanucleotide encompassing the IntA sites in the gabA and amdS (encoding acetamidase) promoters renders gabA expression pH independent and converts amdS into an acid-expressed gene.

MATERIALS AND METHODS

A. nidulans strains, growth conditions, phenotype analysis, and genetic techniques.

The relevant genotypes of the A. nidulans strains used are described in the text and figures and included otherwise previously described standard markers (2, 8, 9, 30, 34). Standard media and phenotype testing and genetic procedures were used (references 8, 9, and 11 and references therein). Standard transformation procedures (34, 35) were used.

Protein extraction, EMSA, probe construction, and DNase I footprinting.

A. nidulans extracts were prepared as described previously (30) with the following modifications. Mycelia were grown in Aspergillus complete medium (11) containing 3% (wt/vol) sucrose (in place of glucose) rather than in penicillin production broth. The extraction buffer contained (final concentration) 4 mM Pefablock (Boehringer) instead of phenylmethylsulfonyl fluoride. Protein extraction was done by disrupting ∼200 mg of wet mycelia in a 2-ml tube with 0.5 ml of 1-mm beads using an Anachem Fastprep apparatus (FP-120) with six 20-s pulses at power 6.0. Cell debris was pelleted by centrifugation at 11,600 × g for 30 min at 4°C. The supernatant, whose protein concentration was determined by the Bradford (7) method, was used for electrophoretic mobility shift assays (EMSA) and footprinting.

Expression and purification protocols for glutathione S-transferase (GST)- and His-tagged PacC fusion proteins have been described previously (18, 34), as have polyclonal antibodies raised against GST and GST::PacC fusion proteins (30). Standard PCR techniques were used to make the GST::IntA(2–186) construct. Exon 2 of intA was amplified from A. nidulans genomic DNA using oligonucleotides intA1 and intA2 (Table 1) and cloned as a BamHI-EcoRI fragment into pD1 (for His tagging). Exon 1 of intA was added by annealing oligonucleotides intA3 and intA4 and introducing exon 1 into the BamHI site of the His-tagging construct using the compatible 5′ overhang sequences. The resulting construct, encoding residues 2 to 383 of IntA, was used as template for PCR with oligonucleotides intA5 and intA6, which contain BamHI and EcoRI sites, respectively. This PCR product was cloned into pGEX-2T for GST::IntA(2–186) expression, and the resulting construct was verified by sequencing. [The His-tagged IntA(2–383) was not sufficiently soluble for use.] DNA binding assays using GST::IntA(2–186) followed the procedures used for GST::PacC (18, 34), except that 1 μM ZnCl₂ was added to the reaction mixtures and poly(dI-dC) was reduced to 1 μg per reaction.

TABLE 1.

Oligonucleotides used in this work (other than those shown in figures)

Oligonucleotide	Sequence (5′→3′)
lac1	AGCGGATCCCTGGCCGTCGTTTTACAACG
lac2	CGCGATCGGCATAACCACCACGC
gbATG	GCGGATCCCGCTGGAATGTCTTTCATGG
gbUP	GCTCTAGAGCGCCTCGAGAGCG
gbA	GCGGATCCCACAGTTACGGAAGTACACC
gbB	CCGCTCGAGATAAGCGGCGCATTGC
gbC	GCGGATCCGCCGCTTATCCCATGG
gbD	CCGCTCGAGCCGGAGTGGAGACTTG
gbE	GCGGATCCCAAGTCTCCACTCCG
gbF	CCGCTCGAGACAAGAGAGCGTTGGG
gbG	GCGGATCCCCAACGCTCTCTTGTG
gbH	CCGCTCGAGCCAGAAGTAGGGCCAC
AMDATG	GCGGATCCGGCGCGCTTATCAGCG
AMDSPRO1	TGAAGGTCGGATGTACG
AMDSPRO2	CTCTGAAAGGATCCCCG
CCAAT-1	CGTCCACCACACAGCCAATCAGCATTGCTT
CCAAT-2	CGAAGCAATGCTGATTGGCTGTGTGGTGGA
intA1	GCGGATCCGCTGCCTGCGTCCAC
intA2	GCGAATTCGCAAACTGGATTACGGTGC
intA3	GATCGTCCACAGCGCATCCGACGAACCTTGCACCCTCAGGAAATC
intA4	GATCGATTTCCTGAGGGTGCAAGGTTCGTCGGATGCGCTGTGGAC
intA5	CGGGATCCTCCACAGCGCATCCG
intA6	CGGAATTCGCAAGCTTCCCCGTCAATCAG

EMSAs using 4 or 8% (wt/vol) polyacrylamide gels followed previously described procedures (30, 34). gabA promoter fragment probes were amplified by PCR using primers containing an XbaI or XhoI site that was end filled using Klenow fragment after digestion. Oligonucleotides gbA, gbC, gbE, and gbG contain an XbaI site, and gbB, gbD, gbF, and gbH contain a XhoI site (Table 1). PCR amplifications using the corresponding named primers gave fragments AB, CD, EF, and GH, respectively, which were cloned into Bluescript [pBS SK(+)] using the XbaI and XhoI sites and sequenced to confirm the absence of mutations. Fragments were excised from the plasmids by digestion with XbaI and XhoI, labeled with [α-³²P]dCTP to a final specific activity of ∼20 kcpm/ng, and used at 1 ng per EMSA. Binding reactions using crude extracts and GST fusion proteins were performed as described in references 30 and 34, respectively. The use of antibodies in EMSAs has been described previously (30). Synthetic probes were constructed and labeled as described previously (34).

DNase I footprinting with both strands of probe CD was as described previously (34). A 1-μg amount of GST::PacC(69–168*) or 15 μg of protein in crude extracts was used.

Promoter constructs with lacZ translational fusion.

pBSΔ2 (31) was prepared, using standard PCR techniques, for replacement of ipnA sequences by gabA or amdS promoter fragments digested with XbaI plus BamHI or SpeI plus BamHI, respectively. Oligonucleotides lac1 and lac2 (Table 1) were used to amplify the region including codons 8 to 280 of lacZ using pBSΔ2argB⁻ as template. The PCR fragment was digested with BamHI (for which a site is located in lac1) plus ClaI and cloned into pBSΔ2argB⁻ digested with BamHI plus ClaI to give pBSΔlacZargB⁻, which lacks ipnA sequences and the first 7 codons of lacZ. These argB⁻ plasmids contain a truncated argB mutant allele that gives rise to a functional argB⁺ allele and consequent arginine prototrophy when transformed into an argB2 strain only if integration occurs homologously in the argB gene. The recipient strain had the genotype yA2 argB2 pantoB100. Truncated mutant and wild-type promoters were fused to lacZ using standard PCR techniques with oligonucleotides gbATG for gabA and AMDATG for amdS. Both contain a BamHI site allowing in-frame fusion of the first 6 codons of gabA and the first 14 codons of amdS to codon 8 of lacZ (see Fig. 8). Successive truncations of the gabA promoter were constructed by PCR using plasmid pM18 (20) and oligonucleotides gbATG plus gbUP, gbA, gbC, gbE, or gbG. PCR fragments were digested with XbaI plus BamHI and ligated to pBSΔlacZargB⁻ digested with SpeI plus BamHI.

FIG. 8.

Interchange of IntA binding sites between the gabA and amdS promoters. (A) Comparison of consensus amdS and gabA IntA binding sites. Vertical lines indicate identities in the amdS and gabA promoters. (B) Diagram of amdS::lacZ translational fusion, with altered base pairs giving the gabA IntA binding site shown in capital letters and sequences of oligonucleotides used for PCR underlined (see Materials and Methods). (C) Diagram of the gabA::lacZ translational fusion, with altered base pairs giving the amdS IntA binding site shown in capital letters and sequences of oligonucleotides used for PCR underlined.

The amdS promoter was amplified by PCR with genomic DNA of the recipient strain as template and primers AMDSPRO1 and AMDSPRO2. The fragment was digested with SpeI and BamHI to give a fragment from positions −1008 to +127 (relative to translational initiation) of amdS, cloned into pBS SK(+), and sequenced to confirm the absence of PCR-generated mutations. This pBS amdS promoter clone was amplified by PCR using AMDATG and reverse oligonucleotide from pBS SK(+), giving a fragment which was digested with SpeI and BamHI (for which a site is present in AMDATG) and cloned into pBS SK(+). The recombinant plasmid was digested with SpeI and BssHII (for which a site at +35 is adjacent to the BamHI site in AMDATG), giving an amdS fragment which was substituted for the original amdS SpeI-BssHII fragment. The resulting plasmid was digested with SpeI and BamHI, and the amdS fragment was cloned into pBSΔlacZargB⁻.

Point mutant gabA promoters were obtained by PCR using promoter A as a template and external primers gbA and gbATG with internal mutant primers gb1e and gb2e (for single mutation A₄→T), gb1b and gb2b (for double mutation G_iv→T A₄→T), or gb1c and gb2c (for triple mutation G_iv→T G₁→T A₄→T). The sequences of these internal oligonucleotide primers are shown in Fig. 5C.

FIG. 5.

Effects of mutations in the PacC consensus and nonconsensus sites of the gαβ synthetic probe on in vitro binding properties. (A) EMSA using the GST::PacC(69–168*) fusion protein. (B) EMSA using extracts of a wild-type strain grown at acidic, neutral, or alkaline pH, a neutral-pH-grown pacC null mutant, and a neutral-pH-grown pacC^c14 strain. f.p., free probe. (C) gαβ probe and positions of mutations. The PacC consensus site is underlined, and the nonconsensus site has dotted underlining. (D) Competition experiment using a neutral-pH-grown pacC⁺ strain extract and unlabeled wild-type or mutant forms of probe gαβ. We used 30, 150, or 300 ng of cold competitor and 0.3 ng of labeled gαβ per reaction.

The strategy for exchanging IntA binding sites in the gabA and amdS promoters is shown in Fig. 8. PCR amplifications used gbA and gbATG together with the oligonucleotides shown in Fig. 8C to construct a gabA promoter with an amdS IntA site. Reverse and AMDATG primers were used with the oligonucleotides shown in Fig. 8B to construct an amdS promoter with a gabA IntA/PacC site.

All promoters were sequenced to ensure the absence of unintended mutations, and the presence of a single copy of the construct integrated at argB was confirmed by Southern blotting as described previously (31).

β-Galactosidase assays.

Mycelia were grown from inocula of 1 × 10⁶ to 2 × 10⁶ conidiospores/ml in appropriately supplemented shaken liquid minimal medium (11) adjusted to pH 8.0 with 25 mM Tris HCl, to pH 6.5 with 25 mM 2-(N-morpholino)ethanesulfonic acid, or to pH 4.0 with 25 mM citric acid. The final pH in each case was within 0.5 U of the initial pH. Unless otherwise noted, 1% (wt/vol) d-glucose was the carbon source (carbon catabolite-repressing conditions). For carbon catabolite-derepressing conditions, 0.1% (wt/vol) d-fructose served as the carbon source. Ammonium [as the (+)-tartrate] at 10 mM served as nitrogen source for nitrogen metabolite-repressing conditions, and 100 mg of uric acid per liter was used for nitrogen metabolite-derepressing conditions. For induction, 5 mM β-alanine was used. After 16 h of growth at 30°C, mycelia were harvested, washed with distilled water, dried, and frozen in liquid nitrogen. Protein extraction followed the protocol described above but using the buffer described previously (31). β-Galactosidase activities were determined as described previously (31). Values given are the average of at least three independent experiments for each transformant, and standard errors are indicated.

RESULTS AND DISCUSSION

Localizing the ambient pH regulatory region in the gabA promoter.

The positions of consensus binding sites for AreA, CreA, and PacC and the only near-consensus binding site for IntA within 1,347 bp of the gabA initiation codon are indicated in Fig. 1A. Most of these sites are clustered between −150 and −450 (relative to the initiation codon). Deletion analysis to determine which of these sites are physiologically important for gabA regulation was performed using transformants in which a single copy of a construct containing a gabA translational fusion with the Escherichia coli lacZ gene had been targeted to the argB locus and assaying reporter β-galactosidase activity (Fig. 1A). Four truncated promoters, designated A, C, E, and G, were compared with a promoter containing 1,347 bp upstream of the coding region (Fig. 1). Promoters A, C, and E, containing ≥274 bp upstream of the coding region, showed relatively normal responses to nitrogen metabolite repression (Fig. 1B) and carbon catabolite repression (Fig. 1C). Only promoters A and C, containing ≥494 bp upstream of the coding region, responded to ω-amino acid induction and ambient pH regulation (Fig. 1B). This localizes a region required for induction and pH regulation to between coordinates −274 and −494, within which lie the sole PacC consensus and IntA near-consensus binding sites (Fig. 1A).

FIG. 1.

Functional analysis of the gabA promoter. (A) Diagram of the gabA promoter and the translational fusion to the E. coli lacZ gene. Consensus binding sequences for CreA, AreA, and PacC are indicated, along with a near-consensus binding sequence for IntA. Truncated forms of the promoter, with distances in base pairs to the initiation codon, are shown. (B) β-Galactosidase activities for each promoter form normalized against promoter A activity (since most further work was done with promoter A) under induced, nitrogen metabolite-repressed, carbon catabolite-repressed, neutral pH growth conditions. N-R indicates nitrogen metabolite-repressing conditions; N-D indicates nitrogen-derepressing conditions; +I/−I indicates inducing or noninducing conditions. All cultures were grown under carbon catabolite-repressing conditions. (C) Comparison of promoter activities (normalized against promoter A activity) under carbon catabolite-derepressing (0.1% d-fructose) and -repressing (1% d-glucose) conditions. Cultures were grown under inducing, nitrogen metabolite-repressing, neutral-pH conditions.

In vitro binding studies using gabA promoter fragments.

Prior to analyzing the gabA promoter for PacC binding sites, we wished to establish whether PacC can bind sequences other than the previously determined GCCARG consensus. Using bacterially expressed PacC DNA binding domain (DBD) oligohistidine- or GST-tagged fusion proteins, a random oligonucleotide selection experiment failed to identify any sequence beyond the known consensus, even with large amounts of protein (800 ng per reaction) in three selection cycles (data not shown). EMSA using four probes covering the region involved in gabA regulation were consistent with this result (Fig. 2A). Probes AB, EF, and GH have no detectable PacC binding activity using cell extracts and 50 to 100 times less affinity for the GST::PacC(69–168*) fusion protein than does probe CD containing the pH regulatory region (data not shown).

FIG. 2.

Gel shift analysis of probe CD containing the ambient pH regulatory region. (A) Locations of the four overlapping probes with coordinates relative to the initiation codon assayed for binding properties. (B) Gel shift using probe CD and extracts of strains carrying various mutations affecting pH regulation or the GST::PacC(69–168*) fusion protein. (C) The two retardation complexes formed using extracts from pacC^c strains. (D) Competition experiments using pacC^c14 extract, 1 ng of probe CD per reaction, and the amounts shown of the ipnA2 synthetic site (30) or the 32-bp CCAAT site from the pacA promoter (32) formed by annealing oligonucleotides CCAAT-1 and CCAAT-2 (Table 1) and end filling. (E) Gel shifts using CD and extracts of a wild-type strain grown under different pH conditions and a pacC^c14 strain and supershifts using antibodies against the PacC DBD and extracts of the two strains grown at neutral pH. f.p., free probe.

Probe CD, covering −260 to −494, strongly binds both PacC from cell extracts and the PacC DBD-containing GST::PacC(69–168*) fusion protein (Fig. 2B). A single major retardation complex is seen with extracts of a neutral-pH-grown wild type strain (Fig. 2B). This complex does not form with extracts of a null pacC deletion (34) strain, and its level is greatly reduced if extracts of a palH17 strain, defective in pH signaling, are used. The partial-loss-of-function palH45 mutation (29) and the palI30 mutation (13), which also has a less extreme acidity-mimicking phenotype than palH17, only slightly reduce the amount of this complex. The fusion protein forms two complexes with probe CD (Fig. 2B), as do extracts of strains carrying various pacC^c mutations, whose elevated PacC levels (27, 30) are probably relevant to the formation of the second complex (Fig. 2C). The full-length pacC translation products for pacC^c14, pacC^c50, pacC^c200, and pacC^c63 contain 488, 262, 575, and 674 residues, respectively. Thus, the equivalent mobilities of both complexes for all four strains in Fig. 2C establish beyond doubt that neither complex contains the full-length form of PacC. The mobility of the faster-moving complex corresponds to that of the wild-type complex. Formation of both complexes with pacC^c14 extracts can be competed by a 31-mer double-stranded oligonucleotide containing the ipnA2 high-affinity PacC binding site but not by a 32-mer fragment from the pacA promoter (32) which contains a CCAAT sequence (Fig. 2D). The presence of the sequence CAAAT in probe CD (see below) prompted the latter competition experiment. Consistent with the presence of processed PacC in the complexes, the amount of wild-type complex increased upon raising the growth pH and PacC DBD antibodies supershifted both pacC⁺ and pacC^c14 complexes (Fig. 2E).

After confirming that the retardation complexes observed with probe CD involved the PacC consensus site region by using an EMSA with the 32-mer probe (gαβ) shown in Fig. 3 (data not shown), we proceeded to DNase I footprinting. A protected window was seen with extracts of wild-type and pacC^c14 strains as well as with the fusion protein (Fig. 3). The GST fusion protein protected a 23-nucleotide region on the gabA coding strand and a 26-nucleotide region on the complementary strand. pacC^c14 extract alters the cutting pattern, suggesting that very high levels of PacC affect the structure of the probe. Adjacent to the PacC consensus site is a divergently orientated hexanucleotide differing from the PacC consensus only by replacement of A₄ by G.

FIG. 3.

DNase I footprinting using probe CD, extracts from neutral-pH-grown pacC⁺ and pacC^c14 strains, and the GST::PacC(69–168*) fusion protein. G+A indicates the Maxam and Gilbert (26) depurination reaction used to map the protected sequences. The synthetic gαβ probe, shown between the gels for the two strands, covers the protected region. The PacC consensus site is in bold and underlined, and a nonconsensus site is underlined with dots. Both gels are presented in the 5′-to-3′ orientation. The intensities of the numbered bands in each strand in tracks containing PacC⁺ are less than 60% of those of the bands labeled A, B and C, chosen for their similarity between protein-lacking (− Protein) and PacC⁺-containing tracks, as determined by phosphorimaging with ImageQuant software.

In vivo analysis of the consensus PacC binding site.

Promoter C (Fig. 1A) contains all of the sequences necessary for pH regulation of gabA, but A was chosen as the standard promoter since the additional 180 bp of upstream sequence would help to ensure the absence of an effect of vector sequences on reporter gene expression after integration at the argB locus. Promoter A shows all known forms of gabA regulation.

The A₄→T change abolishes PacC binding to the high-affinity ipnA2 single site and the function of PacC sites in the ipnA promoter (16, 18, 34). In the gabA PacC consensus site, however, the A₄→T substitution does not significantly affect reporter gene expression in a pacC⁺ strain under neutral growth conditions (Fig. 4, compare lanes 2 and 6). Although the activity of the mutant promoter is no longer reduced by the weakly alkalinity-mimicking mutation pacC^c50 (compare lanes 6 and 7), it is markedly reduced by the strongly alkalinity-mimicking pacC^c14 mutation (compare lanes 8 and 9 with lane 6). In addition to showing that mutating the PacC consensus site with A₄→T only partially alleviates PacC repression, this experiment shows that mutation of a PacC-binding site alters the response to a mutant pacC allele, consistent with a direct effect of PacC binding on gabA expression.

FIG. 4.

Effect of the A₄→T mutation in the PacC consensus site on gabA promoter A (gabA^p::lacZ) activity in pacC⁺, pacC^c14, and pacC^c50 strains. Two progeny are shown for the pacC^c14 background. Strains were grown under inducing, nitrogen metabolite-repressing, carbon catabolite-repressing conditions in media buffered at neutral (lanes 2 to 9) or alkaline (lane 1) pH. β-Galactosidase activities are expressed as percentages of the neutral-pH-grown gabA^p::lacZ pacC⁺ strain activity.

Construction of a triple mutation preventing PacC binding.

The A₄→T substitution reduces the binding of the PacC fusion protein to the 32-mer gαβ probe at least fivefold but does not abolish it (Fig. 5A). Reduced but significant PacC binding to gαβ (A₄→T) is also seen using cell extracts (Fig. 5B). Although this mutant probe forms non-PacC complexes of similar mobility to the PacC complex, the continued predominance of the PacC-containing complex is shown by the increasing prominence of the main band as the growth pH is increased for the pacC⁺ strain and by its even greater prominence using pacC^c14 extract (Fig. 5B), as well as by a supershift when PacC DBD antibodies are added to the reaction mixture (data not shown).

Since the A₄→T substitution in the PacC consensus site neither abolishes PacC regulation of the gabA promoter (Fig. 4) nor abolishes PacC binding (Fig. 5A and 5B) and as PacC protects the near consensus PacC site from DNase I digestion (Fig. 3), a G_iv→T (using Roman numerals to number near-consensus-site positions), substitution was introduced to give a doubly mutant gαβ probe and a G₁→T substitution was also introduced to give a triply mutant gαβ probe (Fig. 5C). Only the triply mutant gαβ failed to compete significantly with the wild-type probe (Fig. 5D). When this triply mutant gαβ was used as probe, no retardation complexes were observed using a variety of extracts, including all of those used for the experiment in Fig. 5B (data not shown).

It is worth noting that the continued functionality of the A₄→T singly mutated PacC site shows that PacC can repress in the absence of a consensus binding site when two near-consensus sites are immediately adjacent.

PacC and IntA target sequences overlap.

The consensus IntA binding site is TTCGGCGN₇CCAAT (reference 12 and references therein). The binding of IntA apparently requires the binding of the hap gene product complex to the CCAAT component (33). There are no IntA consensus sites in the 1,347 bp upstream of the gabA coding region, and the only near-consensus site overlaps the double PacC binding site (Fig. 6A). It differs in having a C rather than a T at the first position and CAAAT rather than CCAAT. Although the A₄→T substitution in the PacC consensus sequence would not be expected to affect IntA binding, both the G_iv→T and G₁→T changes would.

FIG. 6.

Effects of mutations in the PacC consensus and nonconsensus sites on promoter A activity. (A) IntA consensus binding site (12) and gabA IntA near-consensus site with mutational changes shown. PacC site underlining is as in Fig. 5. Noncritical nucleotides in the IntA consensus site are italicized, and deviations from the IntA consensus are shown in lowercase type. (B) Relative β-galactosidase activities of wild-type and mutant forms of promoter A in intA⁺ (where no intA allele is indicated), intA⁻101 (loss-of-function), and intA^c2 (constitutive) backgrounds. Cultures were grown under inducing (except where otherwise noted), nitrogen metabolite-derepressing, carbon catabolite-repressing, alkaline conditions.

Data in Fig. 6B confirm these expectations. Using mutant versions of promoter A in single-copy argB-targeted transformants, the A₄→T substitution elevates both ω-amino acid-induced and noninduced levels of reporter gene expression. However, the additional G_iv→T or G_iv→T plus G₁→T considerably reduces both induced and noninduced levels. Because β-alanine is synthesized as a precursor for coenzyme A, the “noninduced” levels are likely to be, in effect, partially induced. An additional endogenous coinducer is likely to be GABA resulting from glutamate decarboxylation, since at least two other ascomycetes, Saccharomyces cerevisiae (see database entries Z48639.1 and U51031.1) and Neurospora crassa (19), have glutamate decarboxylases. In the presence of the intA⁻101 loss-of-function mutation, the G_iv→T ± G₁→T substitutions have little or no effect. In the presence of the intA^c2 constitutive mutation, which greatly elevates activity levels with the wild-type and A₄→T promoters, G_iv→T ± G₁→T produce a drastic reduction in expression (Fig. 6B, compare lanes 12 and 16 with lane 8). These data establish a physiological role for the IntA near-consensus site in IntA-mediated induction of GABA permease synthesis and strongly support the hypothesis that PacC repression of GABA permease synthesis occurs through prevention of induction, a consequence of the overlap of PacC and IntA target sites.

This situation is reminiscent of that in the promoter of alcR, which encodes the transcriptional regulator of the A. nidulans alcohol regulon, where there is binding antagonism between overlapping CreA and AlcR sites so that carbon catabolite repression interferes with induction (25).

A PacC fusion protein competes with an IntA fusion protein for DNA binding in vitro in the gabA promoter.

DNA binding by a bacterially expressed IntA fusion protein has never been reported. We encountered severe solubility problems in attempting to obtain an IntA fusion protein expressed in E. coli and had to try a number of constructs before obtaining soluble GST::IntA(2–186). An additional problem is that large amounts of this protein are necessary to detect DNA binding (Fig. 7), possibly reflecting a requirement for CCAAT binding complex AnCF (33) for efficient DNA binding. The identity of the GST::IntA-containing complex was confirmed by the ability of anti-GST serum to supershift it (data not shown). Competition for IntA binding by the PacC protein is apparent with both the wild-type probe and the A₄→T mutant probe (Fig. 7). The ability of increasing amounts of PacC to increase the levels of PacC-containing complexes at the expense of IntA-containing complexes and the absence of a supershifted complex indicate that competition rather than simultaneous binding of both proteins is involved. The A₄→T mutation does not reduce IntA binding, consistent with the fact that it abolishes neither IntA-mediated ω-amino acid induction nor the effect of the intA^c2 constitutive mutation (Fig. 6B). Data in Fig. 7 suggest that PacC occupancy of the PacC near-consensus site is crucial to the competition, consistent with the overlap between this site and the IntA consensus site (Fig. 6A) and with the correlation between reduced gabA expression (Fig. 4) and elevated PacC levels (27, 30) (Fig. 2B and D) in pacC^c mutant strains.

FIG. 7.

Oligohistidine-tagged PacC protein competes with GST::IntA for binding to gαβ wild-type and A₄→T mutant probes. GST::IntA(2–186) was added at 900 ng per reaction, whereas His₍₁₀₎::PacC(69–168) was added at 12.5, 25, 50, 100, or 200 ng (wild-type probe) or 50 or 100 ng (mutant probe) per reaction. The binding-site orientation is that shown in Fig. 6A. This is the only experiment shown in which an 8% (wt/vol) polyacrylamide gel was used.

Site swapping: eliminating pH regulation from gabA expression and installing it in amdS expression.

IntA binding sites in the gabA and amdS promoters are compared in Fig. 8A. Figures 8B and C show the sequence alterations for swapping IntA sites between the two promoters for translational fusion genes. The wild-type amdS promoter containing 1,008 bp upstream of the coding region (10) contains all necessary regulatory elements for amdS expression and shows normal responses to ω-amino acid induction, nitrogen metabolite repression, and carbon catabolite repression (data not shown). The wild-type gabA promoter is the 674-bp promoter A (Fig. 1). The wild-type amdS promoter does not respond to pH regulation of transcript levels (data not shown) or reporter gene expression (Fig. 9A), whereas the wild-type gabA promoter does (20) (Fig. 1B and 9B). With the change of 10 bp in the amdS promoter (Fig. 8B), pH regulation is introduced and the amdS promoter becomes acid expressed (Fig. 9A). A change of 9 bp in the IntA site of the gabA promoter (Fig. 8C) virtually abolishes pH regulation (Fig. 9B). Both IntA site-swapped promoters retain responses to ω-amino acid induction, nitrogen metabolite repression, and carbon catabolite repression (data not shown). Thus, the 19 bp of the IntA target site contain the entire region responsible for repression of GABA permease synthesis at alkaline pH, with the overlap of PacC and IntA sites enabling PacC to prevent IntA-mediated induction.

FIG. 9.

Effects of IntA site exchange on regulation of amdS and gabA promoters. (A) Effects of exchanging the gabA IntA/PacC site for the amdS IntA site in the amdS promoter. (B) Effect of exchanging the amdS IntA site for the gabA IntA/PacC site in the gabA promoter. For both panels, β-galactosidase activities are expressed relative to the activity of the wild-type (wt) promoter under alkaline growth conditions and all cultures were grown under inducing, nitrogen metabolite-repressing, carbon catabolite-repressing conditions.

In conclusion, we show here for the first time that PacC acts as a genuine repressor for an acid-expressed gene through preventing the binding of a positively acting transcription factor. Since we have previously shown that PacC acts as a transcriptional activator for alkali-expressed genes (16), this work establishes the dual role of PacC as activator and repressor in pH regulation.