Genetic evidence for interdomain regulation of the phenol-responsive final sigma54-dependent activator DmpR.

The σ54-dependent DmpR activator regulates transcription of the dmp operon that encodes the enzymes for catabolism of (methyl)phenols. DmpR is expressed constitutively, but its transcriptional promoting activity is controlled positively in direct response to the presence of aromatic pathway substrates (effectors). DmpR has a distinct domain structure with the amino-terminal A-domain controlling the specificity of activation of the regulator by aromatic effectors (signal reception), a central C-domain mediating an ATPase activity essential for transcriptional activation, and a carboxyl-terminal D-domain involved in DNA binding. Deletion of the A-domain has been shown previously to result in an effector-independent transcriptional activator with constitutive ATPase activity. These results, in conjunction with the location of mutations within the A- and C-domains which exhibit an effector-independent (semiconstitutive) property, have led to a working model in which the A-domain serves to mask the ATPase and transcriptional promoting activity of the C-domain in the absence of effectors. To investigate the mechanism by which the A-domain exerts its repressive effect, we developed a genetic system to select positively for intramolecular second site revertants of DmpR. The results demonstrate (i) that mutations within the A-domain can suppress the semiconstitutive activity of C-domain located mutations and vice versa; (ii) that the C-domain located mutations do not influence the intrinsic ATPase and transcriptional promoting property of the C-domain in the absence of the A-domain; and (iii) that semiconstitutive mutations of the A- and C-domain have an additive effect. Taken together these results support a model in which the A-domain represses the function(s) of the C-domain by direct interactions between residues of the two domains.

The pVI150 plasmid-encoded dmp system of Pseudomonas sp. strain CF600 confers the ability to utilize phenol, monomethylated phenols, and 3,4-dimethylphenol as sole carbon and energy source. The dmp system is composed of the closely linked but divergently transcribed dmpR regulatory gene and the 15-gene dmp operon that encodes the catabolic enzymes required for conversion of substrates to central metabolites (1,2). Transcription of the dmp operon from the operon promoter Po is regulated positively by DmpR, resulting in expression of the specialized catabolic enzymes only in the presence of pathway substrates or structural analogs (3,4).
DmpR belongs to the prokaryotic enhancer-binding family of 54 -dependent regulators. These activators function to control transcription positively from Ϫ12,Ϫ24 promoters that are recognized by RNA polymerase utilizing the alternative sigma factor, 54 , encoded by rpoN or its analogs (for review, see Refs. 5 and 6). Close physical contact between the regulators bound to their enhancer-like sequences and the cognate promoterbound 54 -RNA polymerase is believed to be brought about by a common mechanism involving looping out of the intervening DNA. In some 54 -dependent systems, including the dmp system (7), this process has been suggested to be assisted by binding of the DNA-bending protein integration host factor, whereas in others, binding of the HU protein or intrinsic bends have been implicated (for review, see Refs. 8 and 9).
Members of the 54 -dependent regulator family have distinct domains that mediate specific function(s) and exhibit varying degrees of homology (for review, see Ref. 5). The amino-terminal signal reception A-domains are joined to the central activation C-domains by means of short flexible B-domain (Qlinker). A region of variable length separates the C-domains from the conserved carboxyl-terminal D-domains that contain a helix-turn-helix DNA binding motif analogous to those found in a number of transcriptional regulators (see Fig. 1A; for review, see Ref. 10). The highly conserved central C-domains of the regulators contain a nucleotide binding motif and are also believed to encompass the region involved in direct interaction with 54 -RNA polymerase. The C-domain mediates ATP binding and hydrolysis essential for transcriptional activation of and open complex formation by 54 -RNA polymerase (11,12). In the case of NtrC, this domain has also been shown to mediate oligomerization required prior to ATPase activity and thus transcriptional activation (13,14), a process that is facilitated by binding to its cognate enhancer sequence (15). The mechanism underlying the coupling of ATP hydrolysis and transcriptional activation is not yet understood fully but may involve a process by which ATP hydrolysis allows the regulator to interact successfully with 54 and thereby relieve the repressive effect of 54 on the ability of 54 -RNA polymerase to form open transcriptional complexes (16).
The A-domain signal reception module of the regulators is the least conserved domain among the different members of the family. Different mechanistic subgroups, which reflect the mode of activation, have been identified (for review, see Ref. 10). Many members of the family, including two archetypal members, NtrC and DctD, are part of so-called two-component regulatory systems. The activities of these regulators are controlled by the phosphorylation status of a conserved Asp residue of the A-domain which is modulated by a sensor histidinekinase in response to an appropriate environmental signal. DmpR, however, belongs to a different mechanistic subgroup that also includes XylR (17,18) and FhlA (19), which responds directly to small effector molecules (for review, see Ref. 10).
The transcriptional promoting activity of DmpR is activated in the presence of the dmp pathway substrates and some, but not all, structural analogs (4). Chimeric proteins (4) and isolation of effector specificity mutants (20,21) have been used to show that the specificity of activation of DmpR by its aromatic effectors resides within its A-domain. Direct interaction of DmpR with its effector molecule allows expression of its otherwise repressed ATPase activity (21). Moreover, deletion of the signal reception amino-terminal A-domain of DmpR results in an effector-independent regulator with full in vivo constitutive transcriptional promoting activity and in vitro ATPase activity (21). Thus, effector-mediated derepression of the C-domain ATPase and transcriptional promoting activity can be mimicked by deletion of the A-domain.
Within DmpR, single amino acid changes in the A-domain (e.g. E135D, E135A), the central activation C-domain (e.g. V276A, V276G), and the short flexible B-linker domain (e.g. L219P), lead to a semiconstitutive (sc) activity with varying degrees of transcriptional promoting ability in the absence of effector (21). These mutations, therefore, partially mimic the activated state of the protein. The location of the mutations within the A-, B-, and C-domains suggests that the A-domain may serve to repress the function(s) of the C-domain by A/C interdomain interactions that are tethered via the normally flexible B-domain. Here, using a genetic system designed to select intramolecular second site suppressor mutations of DmpR-sc derivatives, we investigate the mechanism by which A-domain repression of the C-domain is achieved. Genetic data are presented supporting the model described above, i.e. that repression is mediated via direct interactions between residues of the A-and C-domains rather than by steric hindrance.

Construction of SacB Selection Strain and Isolation of Second Site
Mutations-A promoterless sacB gene of Bacillus subtilus (22) was generated as a BamHI to NdeI fragment as follows. First, the ribosome binding site and the 5Ј portion of the gene to an internal EcoRI site were amplified by polymerase chain reaction and sequenced to ensure that no mutations were introduced; the complete sacB was subsequently reconstituted in its native configuration by cloning of an EcoRI-NdeI fragment spanning the remainder of the gene. The promoterless sacB gene was then cloned under the control of the dmp operon promoter region, Po (base pairs Ϫ422 to ϩ2 relative to the start of transcription (3)), in a modified pBluescript (Strategene) derivative that contained an NdeI and two NotI sites in the polylinker. The resulting fusion was cloned as an NotI fragment into the defective transposon, mini-Tn5 Km2, carried on a suicide vector (23). The Po-sacB fusion of the resulting plasmid, pVI468, was inserted into the chromosome of Pseudomonas putida KT2440 (24) to generate KT2440::Po-sacB, as described previously (25). The broad host range plasmid pVI401 harbors the dmpR gene expressed from its native promoter (20). Semiconstitutive mutant derivatives of DmpR, derived from pVI401, DmpR-V276A (pVI444), DmpR-E135A (pVI447), and DmpR-E135D (pVI448), and the effector specificity mutant DmpR-E135K (pVI428) have been described previously (20,21). Plasmids expressing dmpR-sc derivatives were introduced into KT2440::Po-sacB to generate strains that will lyse in the presence of 5% sucrose (26) due to the mutant DmpR-mediated expression of sacB from the Po promoter in the absence of effectors. To select second site mutations that suppressed the semiconstitutive activity of DmpR derivatives, the cultures were patched onto Luria broth containing 5% sucrose and 4 mM of the mutagen ethyl methanesulfonate. Cultures were left to lyse for 2-3 days and survivors purified. Plasmid DNA from pools of survivors were introduced into the KT2440::Po-luxAB reporter strain and screened for the desired phenotype of low expression in the absence of effector. Single base pair changes were identified in the coding region of DmpR using custom designed oligonucleotides. A subfragment, spanning the identified second site mutation in each case, was then used to replace the equivalent region in pVI401 (to generate the mutation in isolation) and into the original selection plasmid by standard techniques (27). Both strands of the replaced region were sequenced to ensure that the identified mu-tation was the sole mutation responsible for the phenotype. The resulting derivatives were designated DmpR-E135A/L83P (pVI469), -E135D/ F93L (pVI470), -E135D/K188E (pVI471), -E135D/P297R (pVI472), -V276A/W193R (pVI473), -L83P (pVI474), -F93L (pVI475), -K188E (pVI476), -W193R (pVI477), and -P297R (pVI478). Site-specific mutations were introduced into dmpR by overlapping extension using the polymerase chain reaction (28). As above, subfragments of dmpR harboring each of the mutations were exchanged in pVI401, sequenced, and the resulting derivatives designated DmpR-P297K (pVI479), -P297G (pVI480), and -P297E (pVI481). Mutations were combined in a similar manner to generate DmpR-E135D/P297K (pVI482), -E135D/P297G (pVI483), -E135D/P297E (pVI484), -E135K/P297R (pVI485), -E135A/ P297R (pVI486), -E135D/V276A (pVI487), -V276A/L83P (pVI488), -V276A/F93L (pVI489), and -V276A/K188E (pVI490).
Luciferase Reporter Strains and Assays-Construction of the luciferase reporter strain P. putida KT2440::Po-luxAB has been described previously (20). Plasmids expressing dmpR derivatives were introduced into KT2440::Po-luxAB and cultured at 30°C in Luria broth (27). Prior to assay determinations of luciferase activity, cells were grown overnight in Luria broth containing appropriate antibiotics for strain and resident plasmid selection. Cultures were diluted and grown to A 600 ϭ 2.5. Samples were either left unsupplemented or supplemented with 2 mM 2-methylphenol and incubated for a further 3 h. When dmpR derivatives were expressed from the Ptac promoter, as opposed to their native constitutive promoter, isopropyl 1-thio-␤-D-galactopyranoside was added to a final concentration of 0.5 mM. Luciferase activity assays of the luxAB gene product were performed on whole cells using a 1:2,000 dilution of the decanal substrate as described (7).
Western Blot Analysis-Proteins were separated by 11% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose filters, and Western blot analyzed using anti-DmpR serum as described previously (21). Anti-DmpR decorated bands were revealed using Amersham's chemiluminescence reagents as directed by the supplier. Differences in expression levels were assessed by comparison with dilution series of the sample.
Affinity Purification of ⌬A2-DmpR-Flag Derivatives and ATPase Assays-The plasmids expressing ⌬A2-DmpR-Flag derivatives from the T7 promoter were introduced into Escherichia coli BL21(DE3) plysS (30) and the resulting strains grown in Luria broth containing appropriate antibiotics and as described previously (21). Crude extracts were made by sonication of cells in lysis buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 2 mM dithiothreitol) in the presence of the protease inhibitors (complete 228, Boehringer Mannheim), followed by centrifugation at 30,000 ϫ g. Flag epitope-tagged proteins were immunoprecipitated from the crude extract using Flag M2 affinity gel (International Biotechnologies, Inc.), and the bead-bound proteins were washed extensively as described previously (21). The protein-bound beads were resuspended subsequently as a slurry in assay buffer (35 mM Tris acetate, pH 7.9, 5 mM magnesium acetate, 70 mM potassium acetate, 20 mM ammonium acetate, 1 mM dithiothreitol). The concentration of proteins bound to beads was estimated by SDS-polyacrylamide gel electrophoretic comparison of serial dilutions of each sample with standards of known concentration. ATPase assays were performed essentially as in Ref. 11 with modification of the method as described in Ref. 21. In brief, reactions were performed at 30°C in a total of 60 l of 1 ϫ assay buffer (see above) containing 1 l of bead-bound protein. Radiolabeled [␥-32 P]ATP (Amersham) was prediluted (1:9) with cold ATP (Pharmacia Biotech Inc.), and addition of this solution to a final concentration of 3 mM was used to initiate the reaction. Aliquots were removed at the times indicated, adjusted to contain 0.1% SDS, and stored at Ϫ80°C prior to analysis. 1-l aliquots were spotted onto polyethyleneimine cellulose plates (Merck), dried, and the free phosphate separated from ATP by ascending chromatography in 0.75 M potassium hydrogen phosphate buffer, pH 3.5. Radioactivity was quantified using a Molecular Dynamics PhosphorImager. The value for phosphate released was calculated as a percentage of the total amount of radioactivity in each 1-l sample.

Isolation of Second Site Suppressor Mutations of Semiconstitutive DmpR Derivatives-
The simplest interpretation of the semiconstitutive activity mediated by A-and C-domain located mutations, and the proline substitution in the flexible linker that separates them, is that in each case the mutation leads to a weakening of a putative A/C-domain interactions. If this is indeed the case then the model would predict that compensatory mutations in either domain should be selectable. To test this prediction, we constructed a positive selection system to isolate second site revertants. The genetic selection system developed utilizes the conditional lethal effects of the sacB gene product in the presence of sucrose (26). A promoterless sacB gene was constructed and placed under the control of the dmp operon promoter Po and subsequently inserted into the chromosome of a Pseudomonas host to generate KT2440::Po-sacB as described under "Materials and Methods." Trans activation of the Po promoter by plasmid-encoded semiconstitutive derivatives of DmpR (E135A, E135D, or V276A), active even in the absence of aromatic effectors, completes the system to give a strain that will lyse in the presence of sucrose. Positive selection of second site revertants was achieved by isolation of mutants that survived on sucrose-containing plates, i.e. those that inefficiently or no longer activate Po and thus do not express SacB. Five DmpR mutants that had this property, but which were still capable of activation in the presence of aromatic compounds, were analyzed further to locate the mutated residue(s) that mediated the suppression.
The results, summarized in Fig. 1A, revealed two cases of interdomain suppression, where the second mutation resides in a domain distinct from the original mutation. Starting with the A-domain mutant DmpR-E135D, the suppressor mutation P297R was selected. The C-domain mutant DmpR-V276A was used to select the A-domain located second site mutation W193R. In addition to the two cases of interdomain suppression, three examples of intra-A-domain second site revertants were isolated: the semiconstitutive activity of DmpR-E135D was found to be suppressed by either F93L or K188E, whereas that of DmpR-E135A was suppressed by L83P.
Transcriptional Activation by DmpR Derivatives-To quantify the transcriptional response mediated by DmpR derivatives, we used the previously constructed luciferase reporter strain KT2440::Po-luxAB. Plasmids expressing dmpR derivatives harboring the original semiconstitutive mutations, the genetically selected double mutations, or the second site mutations alone were introduced into KT2440::Po-luxAB, and the mutant DmpR-mediated transcriptional response from Po was monitored in the absence and presence of its effector 2-methylphenol. The transcriptional response and protein expression levels are compared with those of wild type DmpR in Fig. 1, B and C. DmpR-E135D, the interdomain suppressed mutant E135D/P297R, and the derivative with the suppressor mutation P297R alone are all expressed at similar levels (Fig. 1C,  lanes 2-4). The P297R mutation clearly suppresses the semiconstitutive activity of DmpR-E135D; the transcriptional response in the absence of effector in E135D/P297R is reduced 4.4-fold compared with -E135D, whereas the effector-activated level is only affected marginally, being reduced by 1.3-fold. The P297R mutation in isolation is indistinguishable from wild type. In the second example of interdomain suppression, DmpR-V276A/W193R, the second site mutation influences the expression level (Fig. 1C, lanes 5-7). The protein level of V276A/W193R is about 50% that of the original V276A mutant, and this level is reflected in the 2-fold reduction in effectoractivated transcription. However, since the transcriptional activation of V276A/W193R in the absence of effector is reduced by at least 3.4-fold relative to that of V276A, it appears that W193R genuinely suppresses the semiconstitutive activity of V276A rather than just bringing the level and thus activity of the protein below threshold levels. It is notable that the W193R substitution alone results in an even lower expression level, i.e. less stable polypeptide, than the W193R/V276A mutant. This observation provides additional support for the interdependence of this pair of mutations.
The intra-A-domain E135D/F93L mutant is expressed at the same level as derivatives harboring each mutation in isolation (Fig. 1C, lanes 2, 11, and 12). The F93L mutation once again clearly suppresses the semiconstitutive activity of E135D with a reduction of 12.8-fold in the absence effectors. In contrast to P297R, which has wild type effector-activated activity, the F93L mutation alone has a 2.1-fold reduced effector-activated activity. In the second example of intradomain suppression, DmpR-E135D/K188E and the K188E mutant are expressed at 50 -75% of the level of E135D and are phenotypically indistinguishable. The K188E mutation reduces the transcriptional response of E135D by 11.3-fold and 2.4-fold in the absence and presence of effector, respectively. Thus, K188E also suppresses the semiconstitutive activity of E135D. In E135A/L83P, the final example of intradomain suppression, the second site mutation L83P causes a 2-fold reduction in the protein level of E135A/L83P but does not influence the expression of the pro- tein harboring the mutation alone (Fig. 1C, lanes 8 -10). E135A/L83P compared with E135A has a 5.2-and 3.2-fold reduced transcriptional response in the absence and presence of effector, respectively. Thus, despite differences in protein levels of three derivatives harboring two substitutions, in all five cases the transcriptional promoting ability in the absence of effector is affected more severely than that in the presence of effector. Therefore we conclude that all five cases are genuine second site suppressors but that the degree with which they suppress the semiconstitutive activities of their respective primary mutation varies between 3.4-and 12.8-fold.

C-domain Located Mutations Do Not Influence the Intrinsic ATPase and Transcriptional Promoting Activities of DmpR-
The identification of two interdomain suppressor mutations, E135D/P297R in which the C-domain located P297R mutation suppressed the semiconstitutive activity of the A-domain located E135D mutation, and V276A/W193R in which the A-domain located mutation suppressed the semiconstitutive activity of the C-domain located V276A mutation, strongly suggests that interdomain interactions mediate A-domain repression of the transcriptional promoting ability of DmpR. However, as outlined in the Introduction, once activated, the transcriptional promoting properties of the regulator are mediated by the C-domain. Therefore, it is possible that each C-domain located mutation may in itself influence the ability of the mutant regulator to promote transcription. To test this possibility we made use of the fact that deletion of the A-domain of DmpR results in a protein, ⌬A2-DmpR, that is fully active in intact cells in terms of transcriptional activation and in in vitro ATPase activity (21). Thus, deletion of the A-domain of DmpR allows assessment of the transcriptional promoting property in the absence of A-domain-mediated effector activation and A/Cdomain interactions. Plasmids expressing ⌬A2-DmpR derivatives harboring the C-domain located mutations involved in the second site revertants (V276A and P297R) were constructed and introduced into the luciferase reporter strain. Transcription activation by mutant ⌬A2-DmpR derivatives was compared with that of wild type and a negative control protein, an inactive derivative harboring a mutation within the ATP binding site (G268S). As shown in Fig. 2A, the two C-domain mutant derivatives are expressed and mediated transcription at levels comparable to wild type in this reporter system.
To analyze the in vitro ATPase activity of ⌬A2-DmpR derivatives we employed the eight-amino acid Flag epitope (29). In-frame carboxyl-terminal fusion of this tag to wild type DmpR does not influence the specificity or transcriptional response of the protein and allows rapid purification from contaminating ATPase activity (21). ⌬A2-DmpR-Flag derivatives were constructed and expressed at high levels using a promoter from phage T7 (see "Materials and Methods"). Monoclonal Flag M2 antibodies coupled to beads were used to affinity purify the individual ⌬A2-DmpR-Flag derivatives and the resulting preparations used in in vitro ATPase activity assays. The results shown in Fig. 2B demonstrate that neither the V276A nor the P297R mutation altered the ATPase activity of the protein in vitro.
The Role of Residue Charge in Second Site Suppression-Both interdomain second site mutations, P297R and W193R, involve a residue charge change. This observation led us to investigate the role of charge in mediating the observed suppression of the semiconstitutive mutant activity. Since the P297R mutation did not influence the protein expression level, either alone or in combination with E135D which it suppressed (Fig. 1C, lanes 3 and 4), this residue was chosen for further analysis. Mutagenesis of codon 297 of DmpR to introduce pos-itive, negative, or neutral amino acids was performed, and the effects of these mutations in isolation or in combination with differentially charged residues at position 135 were monitored in the luciferase reporter system as described above. All mutants used in the analysis are expressed at comparable levels (data not shown), and the results are summarized in Table I. With the exception of P297E, which introduces a negative charge, none of the mutants of P297 influenced the regulatormediated response in isolation. The P297E mutation slightly increased the level of transcription in the absence of effectors. The P297K mutation, which like P297R has a positive charge, similarly also suppressed the semiconstitutive activity of E135D, whereas the neutral exchange P279G and the negatively charged mutation P297E could not. Therefore, we conclude that the charge of the residue at position 279 plays a major role in suppression of the E135D.
The activity mediated by changes in the charged residue at position 135 does not follow a charge-related pattern; substitution of the negatively charged E135 by the alternative negatively charge D residue or neutral A residue results in a semiconstitutive activity, whereas substitution with a positively charged K residue results in an effector specificity mutant which, unlike the wild type, can be activated by the effectors 4-ethylphenol and 2,4-dimethylphenol (20, 21; see Table I). The P297R mutation, which was originally isolated on the basis of suppression of the semiconstitutive activity of E135D, can also suppress the same activity of E135A (Table I) but did not influence the novel effector specificity of E135K (data not shown). These results suggest that the P297R mutation is an independent mutation that compensates for a weakening in the A/C-domain interaction in E135D and E135A, rather than residue P297R interacting directly with corresponding residue at position 135. That this is also the case for A-domain located second site suppressors is suggested below.

The Semiconstitutive and Suppressor Mutations of DmpR
Modulate Independent Interactions-The simplest interpretation of the data described above is that mutations leading to semiconstitutive activity in the absence of effectors weaken interactions between the A-and C-domains, whereas second site suppressor mutations mediate compensatory tightening interactions. If this is indeed the case then the model would predict that combining mutations of different domains that mediate semiconstitutive activity should result in an additive effect. To test this prediction we combined the A-domain E135D and the C-domain V276A mutations and compared the activity of the double substituted DmpR derivative with that of each mutant alone. As shown in Fig. 3A, the prediction holds true; the semiconstitutive activities of V276A (16% of effector-activated) and E135D (42% of effector-activated) are additive in V276A/E135D (63% of effector-activated). These data also suggest that E135D and V276A mutations are likely to be independent mutations that affect different A/C-domain interactions.
Additive effects of different second site suppressor mutations could not be tested since each mutation brought the activity in the absence of effectors so close to basal levels (Fig. 1B). However, a second prediction from the model outlined above could be tested; namely, that if each second site mutation mediated a tightening of the A/C-domain interaction, then each mutation should also be able to suppress the activity of semiconstitutive mutations other than the one on the basis of which it was isolated. To test this prediction the three A-domain located intradomain suppressors, L83P, F93L, and K188E, isolated on the basis of E135D/A semiconstitutive mutants, were combined with the semiconstitutive C-domain mutant V276A. The data summarized in Fig. 3B demonstrate that all three mutants fully suppress the semiconstitutive activity of V276A and have an influence on effector-mediated activation similar to that with their original semiconstitutive selection mutant (see Fig.  1B).

DISCUSSION
Repression appears to be a common mechanism by which the intrinsic activity of different families of prokaryotic transcriptional activators is controlled. For example, truncated derivatives of FixJ (31), Spo0A (32), AraC (33), and LuxR (34) are active in the absence of their normal activating signal. Similarly, the DNA binding properties of 70 and related factors are also under repression control (36). In this work we have investigated the mechanism by which the transcriptional promoting ability of DmpR is regulated in response to the presence of aromatic compounds. As outlined in the Introduction, previous work has demonstrated that the activity of DmpR is regulated by direct interaction with its aromatic effectors leading to expression of its C-domain-mediated ATPase and transcriptional promoting activity (21). Deletion of the amino-terminal A-domain results in a protein that no longer requires effectors to achieve the fully active state. This finding suggested that the A-domain, in the absence of effectors, represses the activity of the C-domain (21), but did not provide any information on how the repression may be mediated. Two possible mechanisms could be envisaged: (i) steric hindrance of the C-domain by the A-domain or (ii) direct A/C interdomain interactions via specific pairs of amino acid residues. Here we used a genetic approach to investigate which mechanism operates. Using a positive genetic selection system we successfully isolated both interand intradomain suppressors of DmpR derivatives which mimic the activated state of the regulator in the absence of effector (Fig. 1A). The fact that interdomain suppressors could be isolated, i.e. that mutations of the A-domain of DmpR could compensate for the effect of a mutation in the C-domain and vice versa (Figs. 1A and 3B) strongly supports the suggestion that specific residues of the two domains interact to mediate repression of C-domain function(s). In this model the original semiconstitutively active mutants would cause a weakening of the A/C-domain interaction, whereas the suppressor mutation would provide a compensatory tightening of the interaction. This model does not necessarily imply that the second site mutation has to influence the same A/C residue/residue interaction as the semiconstitutive mutation that it was selected against (see Fig. 3). It could still be argued, however, that the steric hindrance model cannot be excluded since the original mutations could cause distortions of the overall shape of one domain, which could possibly be compensated for by accommo-  dating distortions of the reciprocal domain. To distinguish between these alternative possibilities we constructed novel combinations of the genetically selected mutants. The A/C-domain interaction model would predict that a suppressor (tightened interaction) of one semiconstitutively active mutant (weakened interaction) would also be able to suppress the same activity of an independent semiconstitutively active mutant, a situation that would not be predicted for the steric hindrance model. All three intra-A-domain suppressors (L83P, F93L, and K188E) were found to also be capable of suppressing the C-domain located semiconstitutive mutant (V276A), once again supporting the A/C-domain interaction model.
Like DmpR, truncated derivatives of the highly related aromatic responsive regulator XylR are constitutively active in vivo (36). By coexpressing different domains of XylR, it has been shown recently that the A-domain of XylR can specifically inhibit the constitutive activity of A-domain-deleted derivatives of XylR and DmpR (37). Although not responsive to aromatic effectors, the inhibition again supports a model in which interdomain interactions mediate repression. This observation, in conjunction with the finding that swapping the A-domain of DmpR for that of the XylR results in a protein with fully regulated XylR effector specificity (4), suggests that the interactions mediating the repression and thus activity are conserved in the two proteins. Sequential deletion into the A-domain of XylR has pinpointed the C-domain proximal portion as the region responsible for interdomain repression (37). Therefore, the DmpR A-domain suppressors (L83P, F93L, K188E, and W193R) of the C-domain semiconstitutive mutant (V276A) may reflect interactions that can compensate for, but are not necessarily normally involved in, the interdomain repression.
In addition to DmpR and XylR, regulatory domain deletions of a number of other 54 -dependent regulators have been shown to be constitutively active, including DctD (38) and LevR (39), whose activities are normally controlled by phosphorylation, and NifA and NifA-like regulators (40,41), whose activities are modulated by protein:protein interactions (for review, see Ref. 10). However, this is not the case for 54 -dependent NtrC. Deletion of the A-domain of NtrC results in an inactive protein, whereas the same deletion of a partially constitutive mutant retains the same level of activity (14). Hence, the phosphorylation of the A-domain of NtrC seems to serve a genuine stimulatory function. Despite this difference, interdomain repression may be the key control system. Recently, on the basis of chimeric proteins, the central C-domain of NtrC has been postulated to repress the formation of A-domain dimers in the absence of phosphorylation (42). Promotion or inhibition of A-domain dimers may serve as a role model for the future challenge of elucidating how aromatic effectors control the A/Cdomain repression of DmpR.