Steric Hindrance Regulation of the Pseudomonas aeruginosa Amidase Operon*

Expression of the amidase operon ofPseudomonas aeruginosa is controlled by AmiC, the ligand sensor and negative regulator, and AmiR the transcription antitermination factor activator. We have titrated out AmiC repression activity in vivo by increased AmiR production intrans and shown AmiC regulation of the antitermination activity of AmiR by a steric hindrance mechanism. In the presence of the co-repressor butyramide we have isolated a stable AmiC·AmiR complex. Addition of the inducing ligand acetamide to the complex trips the molecular switch, causing complex dissociation and release of AmiR. The AmiC·AmiR butyramide complex exhibits acetamide-dependent, sequence-specific RNA binding activity and a K d of 1.0 nm has been calculated for the AmiR·RNA interaction. The results show that amidase operon expression is controlled by a novel type of signal transduction system in which activity of a site-specific RNA binding activator is regulated via a sequestration mechanism.

Expression of the amidase operon of Pseudomonas aeruginosa is controlled by AmiC, the ligand sensor and negative regulator, and AmiR the transcription antitermination factor activator. We have titrated out AmiC repression activity in vivo by increased AmiR production in trans and shown AmiC regulation of the antitermination activity of AmiR by a steric hindrance mechanism. In the presence of the co-repressor butyramide we have isolated a stable AmiC⅐AmiR complex. Addition of the inducing ligand acetamide to the complex trips the molecular switch, causing complex dissociation and release of AmiR. The AmiC⅐AmiR butyramide complex exhibits acetamide-dependent, sequence-specific RNA binding activity and a K d of 1.0 nM has been calculated for the AmiR⅐RNA interaction. The results show that amidase operon expression is controlled by a novel type of signal transduction system in which activity of a sitespecific RNA binding activator is regulated via a sequestration mechanism.
Pseudomonas aeruginosa strain PAC1 grows on short chain length aliphatic amides by virtue of a chromosomal aliphatic amidase operon (1)(2)(3)(4)(5). Amidase activity is inducible by acetamide, propionamide, and lactamide, although only acetamide and propionamide are substrates for the enzyme. Butyramide is hydrolyzed very poorly by the enzyme and acts as a corepressor of amidase expression (6). Amidase operon expression is regulated by a signal transduction system involving AmiC and AmiR (see Fig. 1, A and B). AmiC negatively regulates operon expression, and disruption of the amiC open reading frame leads to constitutive amidase synthesis (4). AmiC functions as the ligand sensor conferring inducibility and amide sensitivity on the system and binds both inducing and non-inducing amides (4,7,8). AmiR, the response regulator, positively controls amidase expression, and disruption of the amiR gene leads to an amidase-negative phenotype. The protein functions as a transcription antitermination factor that mediates the extension of a short, constitutively synthesized, leader transcript through a rho-independent transcription terminator and into the operon (see Fig. 1, A and B) (9,10).
Overexpression of amiC in the wild type strain PAC1 leads to a non-inducible phenotype, and overexpression of amiR leads to a semiconstitutive phenotype (4,7). These results suggested that the mechanism of AmiC inhibition of amidase expression involved a direct protein-protein interaction with AmiR (4,8). Thus AmiC and AmiR form a signal transduction autogenous control circuit, regulating their own and amidase expression in response to inducing amides (10).
The amidase operon is expressed from a strong constitutive Escherichia coli-like 70 promoter sequence, followed by a short leader open reading frame of no apparent function, within which lies a rho-independent transcription termination sequence (T1). Downstream lies the amidase S-D 1 sequence and amiE gene (Fig. 1C). The antitermination regulatory system was defined by the analysis of a 10-bp T1 terminator deletion, which led to amide-independent, AmiR-independent, constitutive amidase expression (9) and by transcript analysis, which showed constitutive production of the leader mRNA (10). Direct AmiR/leader mRNA interaction has been demonstrated in vivo by AmiR titration with excess leader transcript and the analysis of constructed leader region mutants. This study defined two short regions of RNA of different sequence within the leader transcript independently needed for full antitermination activity in vivo (11), indicating that the interactions between AmiR and the leader transcript probably involve the correct three-dimensional presentation of these two protein recognition sequences.
The structure of AmiC-acetamide has been determined and consists of a single polypeptide chain folded into two domains (12). Each domain consists of a ␤-␣-␤ topology linked by a hinge region and closed down with a trapped acetamide molecule within the ligand binding cleft, which lies at the interface of the two domains. The overall fold is very similar to that of members of the small molecule binding protein family, particularly to LivJ, the periplasmically located leucine/isoleucine/valine binding protein from E. coli. Attempts to overproduce and isolate AmiR have been unsuccessful, because the protein forms insoluble aggregates. The structure of the AmiC⅐AmiRbutyramide complex described in this paper has recently been determined by x-ray crystallography and comprises an intimate AmiR dimer bound to two AmiC monomers (13). Each AmiC monomer interacts simultaneously with both molecules in the dimer and makes identical contacts to the top surface formed by the two AmiR N-terminal domains. Comparison of AmiR with known structures revealed a significant similarity between the AmiR N-terminal domain and the response regulator receiver domain family of bacterial two-component signal transduction systems (13). However, despite the structural homology of AmiR with this large group of regulators, there is no conservation of the key residues, suggesting that AmiR is the first member of the family not regulated by phosphorylation. The AmiR C-terminal domain consists of a long ␣-helix terminating in a three-helix bundle. Analysis of AmiR mutants deleted for C-terminal regions strongly implicated this domain in RNA binding, because the mutants are non-functional with respect to antitermination in vivo (13).
In this paper we report the results of in vivo and in vitro studies to investigate the mechanism of AmiC regulation of AmiR antitermination activity, the isolation of an AmiC⅐AmiRbutyramide complex, and its use in showing acetamide-dependent, sequence-specific RNA binding.
Amidase Assays-Amidase activity in intact cells was measured as described previously (8). One unit represents 1 mol of acetohydroxamate formed per minute. Specific activities presented are units per milligram of bacterial cells (14).
AmiR Quantitation-Total equalized and washed cell pellets were dissolved in loading buffer, and the proteins were resolved by SDS-PAGE (15). Proteins were electrophoretically transferred to nitrocellu-lose. AmiR quantitation used a polyclonal rabbit anti-maltose binding protein-AmiR serum (11) and horseradish peroxidase-linked goat antirabbit IgG for ECL Western blotting detection (Amersham Pharmacia Biotech). Fluorescence was detected by x-ray film, captured with a UVItech photodocumentation system and analyzed with UVIDoc and UVIBand software (UVI Cambridge, United Kingdom).
Purification of the AmiC⅐AmiR-butyramide Complex-The complex was isolated from PAC452, pRAN1 cells grown, harvested, and broken as described previously using buffers containing butyramide (11). The cleared lysate was ammonium sulfate-fractionated, and the 40 -50% portion cut was resuspended in Buffer A (20 mM Tris, pH 9.0, 1 mM EDTA, 1 mM dithiothreitol, 23 mM butyramide). 2 ml of extract (90 mg/ml) was subjected to preparative gel filtration (Superdex 200, Hi-Load 16/60) and eluted with Buffer A ϩ 150 mM NaCl. 1-ml fractions were collected, and those shown to contain most of the AmiC⅐AmiR complex by SDS-PAGE were pooled for anion-exchange chromatography. Pooled samples were fractionated using a HiLoad Q-Sepharose FF 26/10 column pre-equilibrated with Buffer A, and protein was eluted with a linear gradient of 0.15-0.8 M NaCl. 5.0-ml fractions were collected and analyzed by SDS-PAGE. Fractions containing most of the AmiC⅐AmiR complex were pooled and made up to 1.2 M ammonium sulfate for hydrophobic interaction chromatography using a 20-ml phenyl-Sepharose HP column. The column was pre-equilibrated with Buffer A containing 1.0 M ammonium sulfate, and protein was eluted with Buffer A containing decreasing ammonium sulfate concentrations. 5.0-ml fractions were collected, and those containing most of the AmiC⅐AmiR complex were pooled. The protein was concentrated to 5 mg/ml (PM30 Diaflo Ultrafilter, Amicon) and subject to analytical gel filtration using a Superose 12 HR 10/30 column. Fractions identified as containing the AmiC⅐AmiR complex were pooled and concentrated to 16 mg/ml.
AmiC⅐AmiR Complex Assays-For gel filtration studies, pooled and concentrated AmiC⅐AmiR-butyramide complex samples were analyzed using a Superose 12 HR 10/30 column (Amersham Pharmacia Biotech). Samples were initially in Buffer A (containing 23 mM butyramide and 0.6 M NaCl). Acetamide was added, where necessary, and samples were incubated at 25°C for 30 min. Fractions (0.2 ml) were collected after elution with Buffer A ϩ 0.6 M NaCl containing either 23 mM butyramide or appropriate amounts of acetamide and trichloroacetic acid-precipitated. The precipitate was dissolved, and proteins were resolved by SDS-PAGE and quantitated using the UVItech photodocumentation system and analysis with UVIDoc and UVIBand software (UVI Cambridge). The Superose 12 column was calibrated using gel filtration standards (Amersham Pharmacia Biotech).
AmiC⅐AmiR Complex RNA Interactions-The labeled RNA and AmiC⅐AmiR-butyramide complex were mixed in 100 mM NaCl, 1 mM MgCl 2 , and acetamide (136 mM) if necessary. Reactants were incubated at 4°C for 20 min and then diluted into 20 mM Tris (pH 8.2), 80 mM NaCl, 50 mM KCl, 0.25 mM MgCl 2 , and acetamide (34 mM), if needed, and then incubated at 4°C for a further 20 min. After addition of 50% glycerol, 1% bromphenol blue, samples were run on a 5% non-denaturing PAGE. The dried-down gels were exposed, and reaction products were quantitated using a FujiFilm phosphorimager.

Measurement of AmiC:AmiR
Stoichiometry-Previous studies using cloned amiC and amiR genes in both E. coli and P. aeruginosa showed that overexpression of amiR led to a semiconstitutive phenotype and overexpression of amiC to a non-inducible phenotype (4, 7). However, a normal amide-inducible phenotype was found with a regulated expression vector (pMW21) in which the two regulatory genes were present together in their normal orientation (8). It thus appeared that AmiC:AmiR stoichiometry was important for the maintenance of the inducible phenotype. This has been investigated by measurement of the amounts of AmiC and AmiR in total extracts of PAC452, PAC452,pMMB66EH, and PAC452,pRAN1. The host strain carries a chromosomal deletion of the amidase operon, and pRAN1 carries the polymerase chain reactioncloned wild type genes, including S-D sequences, downstream from the pMMB66EH-regulated tac promoter. Plasmid pRAN1 shows IPTG-dependent lactamide-inducible, butyramide-repressible amidase expression in strain PAC327, an amiC,amiR double mutant host clearly showing the production of a normal AmiC:AmiR ratio from pRAN1 (not presented). After growth in medium containing IPTG, total cell pellets were solubilized, and the proteins were separated by SDS-PAGE (Fig. 2). The induced whole cell extract of PAC452,pRAN1 (lane 6) showed the presence of a major new band corresponding to AmiC and a fainter AmiR band neither of which are present in any other lane. Amounts of AmiC and AmiR were quantitated by scanning densitometry. The values obtained were standardized to molarities by application of a correction factor of 0.892 to the AmiR value obtained. This was obtained by scanning densitometry of a Coomassie Blue-stained SDS-PAGE gel containing a washed and solubilized AmiC⅐AmiR-butyramide crystal containing 2xAmiC/2xAmiR (13). In stained SDS-PAGE AmiR, overstains with respect to its M r and AmiC and AmiR molarities are obtained by multiplication of AmiR values by 0.892. In Fig. 2 (lane 6) AmiC and AmiR together represent 9% of the total stained material in a molar ratio of 3.3:1. There is thus an excess of the AmiC-negative regulator within the cells.
AmiC Titration by AmiR and the Steric Hindrance Regulatory Mechanism-To investigate the effect of AmiR overproduction, amidase activity was measured and AmiR quantitated in non-induced and induced PAC1, PAC452, and in PAC1,pSW40 (Table I). PAC1 the wild type strain showed a non-induced background level of amidase activity (specific activity ϭ 0.16) and a 40-fold S-lactamide-induced level of activity (7.07). Plasmid pSW40 is an IPTG-inducible amiR expression vector, and LacI regulation of the vector tac promoter is apparently leaky, because in the absence of IPTG with PAC1, pSW40 sufficient AmiR is produced to titrate out the excess AmiC and generate 12.9% (specific activity ϭ 1.31) of the fully induced level of amidase expression. Addition of IPTG up to 0.1 mM showed a complete titration of AmiC repression activity and full operon expression (specific activity ϭ 10.16) in the absence of inducing ligand (Table I). Experiments with added IPTG up to 0.150 mM showed no further increase in amidase specific activity (data not presented). Overproduction of AmiR in trans will lead to both increased amidase synthesis and additionally increased AmiC and AmiR production in the 3.3:1 ratio measured above. At any level of ligand-free amidase operon expression, sufficient additional AmiR must be produced from pSW40 to titrate out the molar excess of AmiC. Total AmiR was quantitated by ECL Western blotting in one experiment, and the values of the amidase specific activity and total AmiR from Run 8 are shown in Table I. The plot of total AmiR against amidase specific activity (Fig. 3) showed a direct linear relationship between the two values consistent with a steric hindrance mechanism for the regulation of the antitermination activity of AmiR by AmiC.
Isolation of the AmiC⅐AmiR-butyramide Complex-A corollary of the steric hindrance mechanism is that, for cells growing under amide non-inducing or -repressing growth conditions, an AmiC⅐AmiR complex should exist in which the antitermination activity of AmiR is occluded by AmiC. Previous investigations using PAC1 showed that, during amidase induction studies in which activity was required for cell growth, as the cells entered stationary phase a second phase of enzyme induction occurred despite the absence of inducing amides in the growth medium at this time (6). To overcome this potential problem strain PAC452,pRAN1 was grown in the presence of IPTG and butyramide for the isolation of an AmiC⅐AmiR complex. The major purification problem was the contamination of extracts by excess AmiC, which was removed during the gel filtration stages. The final product ran as a single peak in gel filtration chromatography (Fig. 4A) with a calculated M r of 130,000 and an organization of 2xAmiC/2xAmiR from SDS-PAGE (Fig. 2, lane  8). The complex in the presence of butyramide was stable between pH 5.5 and 9.0 and was equally stable in the presence of up to 1 M NaCl as determined by gel filtration chromatography (not presented).
Analysis of the AmiC⅐AmiR Regulatory Complex-The K d of acetamide for AmiC was determined previously at 3.7 M and that for butyramide at 0.31 mM (8). Thus addition of equimolar acetamide to the AmiC⅐AmiR-butyramide complex would be expected to displace the butyramide from the AmiC ligandbinding site. It had been shown previously using a semipurified AmiC⅐AmiR-butyramide extract that addition of acetamide caused partial complex dissociation (11). To quantitate this effect AmiC⅐AmiR-butyramide samples were incubated with  mM, and 34 mM). Changes were monitored by high resolution gel filtration chromatography (Fig. 4), and the fractions collected were analyzed by SDS-PAGE and Western blot (not presented). The initial complex sample ran as a single discrete peak containing AmiC and AmiR, which represented 0% dissociation (Fig. 4A). As the acetamide concentration was increased the main peak decreased and was replaced by a high M r AmiR shoulder and a lower M r AmiC peak. Fractions from the 34 mM acetamide run (Fig. 4F)  complex against acetamide concentration shows that at 2.5 M acetamide the complex is 50% dissociated (Fig. 5). This value is close to the K d of acetamide for AmiC (3.7 M). Thus at an acetamide concentration at which 50% of the AmiC ligand binding sites are occupied, the complex is 50% dissociated into AmiC-acetamide and free AmiR. The AmiC and AmiR interactions involved in complex dissociation probably do not represent a true equilibrium situation, and thus it is not possible to calculate a K d for the interaction. However, it is apparent that acetamide addition to the AmiC⅐AmiR-butyramide complex causes dissociation and the release of AmiR. The released AmiR appears to form a series of high molecular weight soluble aggregates shown as a broad shoulder on the elution profile. The monomer M r of AmiR is 21,775 (3), and the position of the AmiR shoulder seen on the gel filtration profile shows the protein present as aggregates of M r 130,000 and greater. Thus, AmiR must be present as at least a hexamer and other higher oligomeric forms. It is difficult at this stage to establish whether these forms are inactive aggregates due to the inherent properties of the protein or whether they represent a biologically relevant organization. The partially overlapping AmiC-acetamide peak seen under 100% dissociation conditions (Fig. 4F) appears to comprise a mixture of AmiC-acetamide dimers (M r 85,600) and trimers (M r 128,600).
In Vivo States of the Regulatory Switch-The isolation of an AmiC⅐AmiR-butyramide complex represents the repressed configuration of the switch regulating amidase operon expression. In vivo the normal states of the switch are predicted to be that present under non-inducing growth conditions, i.e. small amounts of ligand-free AmiC⅐AmiR complex, and under inducing growth conditions, i.e. large amounts of AmiC-acetamide plus AmiR bound to the RNA leader sequence. The AmiC⅐AmiR-butyramide complex was subject to extensive equilibrium dialysis against amide-free buffer to reduce the butyramide to Ͻ40 pM and analyzed by high resolution gel filtration chromatography. The results (not presented) show the ligandfree AmiC⅐AmiR as a stable soluble trapped complex migrating as a single species of M r 130,000 identical to the AmiC⅐AmiRbutyramide complex (Fig. 4A).
RNA Binding Studies-The role of AmiR as an antitermination factor was established by showing in vivo titration of the protein by competing leader transcripts produced in trans, and the analysis of leader transcript mutants in vivo in which the antitermination reaction failed to work (11). The proposed mechanism for the reaction is that of free AmiR interacting with the leader transcript, initially with the RNA sequences identified by the mutagenesis to generate an RNA/protein configuration in which the formation of the leader terminator stem/loop structure is precluded, allowing RNA polymerase to continue transcription into the operon. Only with the isolation of the AmiC⅐AmiR-butyramide complex have we obtained AmiR in a soluble form. Thus, to investigate the AmiR⅐RNA interaction in vitro we have used the AmiC⅐AmiR-butyramide complex and labeled leader RNA transcripts with analysis of the reaction products by non-denaturing PAGE.
Initial studies used a fixed amount of continuously labeled gel-purified run-off transcript and increasing complex concentrations, which showed acetamide-dependent gel retardation (not presented). To investigate the interaction, fixed concentrations of complex were used with increasing amounts of transcript and the values of bound and unbound RNA were quantitated (Fig. 6). The autoradiograph shows the transcript (16 nM) (lane 1), transcript plus protein complex (2.7 M AmiR dimers) plus acetamide (lane 2), and reactions containing increasing labeled transcript (lanes 3-10). The transcript migrates as a major single band with a minor slower migrating uncharacterized form (lane 1). The transcript plus complex shows the migration of the RNA is retarded (lane 2). Lanes 6 -10 show the presence of the unbound RNA. The levels of bound and unbound labeled transcript were quantitated (not shown). The curve of labeled transcript added versus labeled transcript bound displayed negative deviation from ideal behavior. Values obtained were used to generate a Scatchard plot, which shows non-linear characteristics (Fig. 6B). The calculated K d for the binding of leader transcript to AmiR was 1.0 nM, and the value of indicates a stoichiometry between AmiR dimers and RNA of 152:1. A second attempt to determine the K d of the AmiR⅐RNA interaction used fixed amounts of labeled transcript (13 nM) incubated with increasing concentrations of the AmiC⅐AmiR-butyramide complex (0.19 -2.7 M AmiR dimers) plus acetamide (not presented). The proportions of bound and unbound transcript were again quantitated and are shown in Fig. 7. The AmiR concentration at which 50% of the transcript is bound was 172 nM, which would correspond to the K d if the AmiR dimer/RNA stoichiometry was 1:1. Taking into account the calculated stoichiometry of 152:1, the K d value here for the AmiR dimer/RNA interaction is 1.1 nM and close to the value obtained previously.
In all of these experiments there was an apparent large molar excess of AmiR dimers to RNA required to bandshift the labeled transcript. The most probable explanation for these results is that, once the reaction is set up, the AmiC⅐AmiR complex completely dissociates and much of the AmiR released aggregates into a form that is unable to interact normally with the transcript. Reaction conditions used for these studies are the same as those used for the analysis of the AmiC⅐AmiR complex in which the complex was completely dissociated.
AmiR⅐RNA Specificity Studies-To show the specificity of the AmiR⅐RNA interaction, reactions were carried out with the inclusion of either cold pMW42 transcripts or unlabeled yeast tRNA (Fig. 8). Fig. 8A shows the result of the cold specific competition carried out with 8 nM labeled transcript, 2.7 M AmiR dimers, and from 13 to 250 nM cold transcript. Lane 1 contains labeled transcript (47 nM), lane 2 the complete reaction in the absence of cold competitor, and lanes 3-9 the reaction in the presence of competitor. Competition is seen in lane 6 (63 nM competitor) onwards. Similar studies were carried out using cold yeast tRNA (Fig. 8B) yeast tRNA (lanes 3-10). In these studies there was no evidence of competition, however, competition was observed with 24 M yeast RNA (not shown). These results showed that the RNA binding capacity of the AmiC⅐AmiR complex is sequence-specific. DISCUSSION Previous investigations of the regulatory system of the P. aeruginosa amidase operon indicated that a direct protein/ protein interaction was responsible for the AmiC control of AmiR antitermination activity (4,7,8,11). We have now shown that the ligand sensor is normally present in molar excess compared with the response regulator and that the stoichiometry must be maintained to preserve normal inducibility. Overexpression of amiR showed titration out of AmiC repressor activity, and measurement of AmiR showed a linear relationship between amidase activity and total AmiR consistent with a steric hindrance regulatory mechanism. For this type of control system, regulation must be flexible enough to sense appropriate levels of inducing ligand for induction to occur and capable of producing relatively high levels of the activator component for mRNA binding. The flexibility is maintained by the fixed 3.3:1 AmiC:AmiR stoichiometry, and the AmiR production capability is maintained by having the regulatory genes expressed from the main operon promoter. In the wild type non-induced culture, slight leakiness of the T1 terminator ensures the production of residual amidase to prevent unnecessary induction and guarantees the presence of low levels of the regulators.
The two amidase regulator genes are linked such that the stop codon of amiC and the start codon of amiR overlap (AUGA), and they lie within a polycistronic transcript during both non-inducing and inducing growth conditions. We propose that a translational coupling system (16) operates such that amiR is only translated by reinitiation after translation of amiC and the stoichiometry is thus determined by the reinitiation frequency. The S-D sequence of amiC shows much higher homology to the 3Ј-end of the 16 S rRNA than that of the amiR S-D sequence, and RNA secondary structure predictions of the respective translation initiation regions show potential secondary structure formation (Fig. 9). However, the amiC leader secondary structure is relatively unstable (Ϫ4.4 kcal/ mol) and the S-D sequence would be ribosome-accessible, whereas the potential amiR leader secondary structure would be much more stable (Ϫ13.8 kcal/mol) and the S-D sequence would be sequestered into a secondary structure stem/loop formation. It is thus an attractive hypothesis that amiR is only translated by reinitiation after translation of amiC, and the cell uses this mechanism to maintain the AmiC:AmiR ratio. Similar translational regulation of proteins that interact via a steric hindrance mechanism has been investigated in other systems. In the nifLA operon of Klebsiella pneumoniae, which controls expression of the nif nitrogen fixation regulon, translational coupling has been shown to operate to ensure an excess of the NifL repressor over the NifA transcription activator (17). With the bacteriophage T4 clamp loader genes, translational coupling has been shown to be the major factor used to maintain the stoichiometry of the subunits (18).
A prediction of the steric hindrance regulatory mechanism was that under non-inducing or repressing growth conditions AmiC and AmiR should exist as a complex within which the AmiR RNA binding activity is occluded. Thus, three states of the molecular switch may be envisaged: under non-inducing growth conditions-small amounts of ligand-free AmiC⅐AmiR complex; under inducing growth conditions-larger quantities of AmiC-acetamide with the released AmiR bound to the leader transcript; and under repressing growth conditions-the AmiC⅐AmiR-butyramide complex. We have examined the repressed and induced states and have shown that the ligand-free AmiC⅐AmiR complex is a stable entity. It is predicted that this ligand-free complex is a structural homolog of the butyramide complex with water molecules occupying the ligand binding site.
Previously identified problems with endogenous inducers led us to use the addition of the co-repressor butyramide for the successful isolation of the complex. This is the first occasion on which AmiR has been isolated in a soluble form useful for further investigations.
A steric hindrance mechanism additionally presupposes that addition of inducing amides would either change the complex structure to expose the AmiR RNA binding motif or cause dissociation. In vivo the AmiR-dependent antitermination reaction functions in a normal but unregulated way in the absence of AmiC (4,7,8). Analysis of the AmiC⅐AmiR complex upon addition of acetamide showed 50% dissociation at an acetamide concentration of 2.5 M close to the K d of acetamide for AmiC (3.7 M). Thus, at an acetamide concentration in which 50% of the AmiC ligand binding sites are occupied, the AmiC⅐AmiR complex is 50% dissociated.
To investigate the AmiR⅐RNA interaction, we have used the AmiC⅐AmiR-butyramide complex in the presence of acetamide and continuously labeled RNA transcripts. In preliminary experiments, which showed ligand-dependent gel retardation, there was an obvious molar discrepancy between AmiR dimers and RNA, and in all cases there was a requirement for an excess of dimers for gel retardation. K d determinations of the AmiR⅐RNA interaction used two approaches. The results using varying transcript concentrations showed negative deviation from ideal behavior in the transcript-bound versus transcriptadded plot indicative of subpopulations of AmiR in the binding assays, active AmiR, and partially or inactive aggregated AmiR. The Scatchard plot obtained is non-linear, and the observed K d for the binding of leader transcript to the site of highest affinity, presumed AmiR dimers, was 1.0 nM. The observed stoichiometry of 152:1 AmiR dimer:RNA is the excess described previously. The K d determination by varying the AmiR dimer concentration was of a similar value (1.1 nM) assuming a 152:1 stoichiometry.
The specificity of the AmiR⅐RNA interaction was studied by cold competition. Use of yeast tRNA showed competition only at a 616 molar excess, and the use of cold specific competitor RNA showed AmiR sequence-specific RNA binding activity.
K d values for RNA/protein interactions vary in different biological systems. The Trp RNA-binding attenuation protein-Trp/leader RNA interaction involved in the regulation of tryptophan biosynthesis in several Bacillus species has a K d of Ϸ 0.1 nM, showing high affinity (19), at least partly due to the repetitive nature of the interaction between the 11-subunit TRAP-Trp-mer and the GAG repeats in the RNA (20). Other characterized protein/RNA interactions include that of BIV Tat-TAR with a K d Ϸ 0.75 nM (21,22) and in other bacterial catabolic antitermination systems such as SacY where a much lower K d value (1.4 mM) has been reported (23). Deletion analysis of AmiR has implicated the C-terminal three-helix bundle as the RNA binding domain (13), and it is expected that deter- mination of the AmiR⅐RNA complex structure will provide an explanation of the apparent high AmiR⅐RNA K d value.
Biological use of steric hindrance (sequestration) for regulatory systems occurs rarely and would appear to be old in evolutionary terms. Only the NifLA nitrogen fixation regulatory systems of the free living diazotrophs K. pneumoniae and Azotobacter vinelandii appear to be functionally related. In these cases in which the NifL sensor responds to both fixed nitrogen and the external oxygen concentration to inhibit activity of the response regulator NifA, only the mechanism is common with the amidase system (23)(24)(25). NifA is a member of the NtrC response regulator family with an N-terminal regulatory domain, a central catalytic (ATPase) domain, and a C-terminal DNA binding domain, which activates transcription from the nif regulon 54 promoters (26). The domain structures of both NifL and NifA have been determined, and they are unlike the regulators of amidase expression (27). The mechanism(s) by which NifL regulates the activation activity of NifA has not been fully characterized. However, in A. vinelandii the Nterminal response regulator receiver domain of NifA and the NifL C-terminal histidine kinase homolog transmitter domain potentiate the ADP-dependent formation of the NifL⅐NifA complex, and the C-terminal domain of NifL has additionally been shown to interact with the catalytic domain of NifA (28,29).
The amidase signal transduction system thus contains protein folds utilized in other control networks but found here in novel combinations, with an unusual regulatory mechanism. The negative regulator AmiC has the small molecule binding protein fold comprising two domains, a hinge region and a central cleft for ligand binding. In the absence of ligand or presence of the co-repressor, AmiC adopts an open configuration and forms a complex with AmiR within which the biological activity of the response regulator is occluded. In the presence of inducing ligands, AmiC adopts the closed-down form in which the interactions with AmiR are broken and the regulator is released. AmiR, the antitermination factor, contains the classic N-terminal response regulator receiver domain and a C-terminal, three-helix bundle, site-specific RNA binding domain. Our current model from site-directed mutagenesis of AmiR suggests that the role of AmiC is to prevent the oligomerization of AmiR into a form functional in RNA binding.
These results show a steric hindrance-regulated signal transduction system used for controlling expression of the P. aeruginosa amidase operon. The mechanism is distinct from the classic two-component systems found in the majority of bacterial regulatory networks, and currently AmiR appears to be a novel site-specific RNA binding protein.