Protein-induced fit: the CRP activator protein changes sequence-specific DNA recognition by the CytR repressor, a highly flexible LacI member
1997; Springer Nature; Volume: 16; Issue: 8 Linguagem: Inglês
10.1093/emboj/16.8.2108
ISSN1460-2075
Autores Tópico(s)DNA and Nucleic Acid Chemistry
ResumoArticle15 April 1997free access Protein-induced fit: the CRP activator protein changes sequence-specific DNA recognition by the CytR repressor, a highly flexible LacI member Henrik Pedersen Henrik Pedersen Department of Molecular Biology, Odense University, Campusvej 55, DK-5230 Odense M, Denmark Search for more papers by this author Poul Valentin-Hansen Corresponding Author Poul Valentin-Hansen Department of Molecular Biology, Odense University, Campusvej 55, DK-5230 Odense M, Denmark Search for more papers by this author Henrik Pedersen Henrik Pedersen Department of Molecular Biology, Odense University, Campusvej 55, DK-5230 Odense M, Denmark Search for more papers by this author Poul Valentin-Hansen Corresponding Author Poul Valentin-Hansen Department of Molecular Biology, Odense University, Campusvej 55, DK-5230 Odense M, Denmark Search for more papers by this author Author Information Henrik Pedersen1,2 and Poul Valentin-Hansen 1 1Department of Molecular Biology, Odense University, Campusvej 55, DK-5230 Odense M, Denmark 2Department of Chemistry, University of California, Berkeley, Berkeley, CA, 94720 USA The EMBO Journal (1997)16:2108-2118https://doi.org/10.1093/emboj/16.8.2108 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The CytR repressor and the cAMP receptor protein (CRP) bind cooperatively to several promoters in Escherichia coli to repress transcription initiation. The synergistic binding is mediated by protein–protein interactions between the two regulators. Here, in vitro selection experiments have been used to examine the DNA-binding characteristics of CytR, by itself and when co-binding with cAMP–CRP. We show that the optimal CytR-binding site consists of two octamer repeats, in direct or inverted orientation, and separated by 2 bp. However, when co-binding with cAMP–CRP, CytR instead recognizes inverted repeats separated by 10–13 bp, or direct repeats separated by 1 bp. The configurations of the latter set of operators correlate well with the configurations of natural CytR targets. Thus, cAMP–CRP induces conformational changes in CytR so that the repressor fits the natural targets. Most strikingly, CytR can adopt widely different conformations that are equally favored energetically for complex formation with cAMP–CRP. We propose that this structural adaptability is essential for CytR repression of promoters with diverse architectures. We discuss these novel concepts in the context of the CRP/CytR regulatory system, as well as the structural and functional implications for multiprotein–DNA complex formation in general. Introduction The selective regulation of cellular processes such as site-specific recombination, transcription and DNA replication depends upon the recognition of specific DNA sites by DNA-binding proteins. In the simplest cases, the DNA-binding domain of the protein carries all the information necessary to specify its site of action. There are, however, many systems in which additional factors are required to specify the exact binding site of a protein. The complexity of these systems is often very high, and the molecular mechanisms that provide affinity and specificity remain elusive. In principle, a DNA-binding protein may improve the DNA-binding specificity of another protein (i) by providing additional contacts through its surface, (ii) by creating a DNA conformation that is better recognized by the partner protein, or (iii) by inducing a conformational change in the partner protein that promotes its interaction with the operator. Here, we have sought the mechanism by which sequence-specific DNA recognition is achieved in a relatively simple prokaryotic system. In this system, the global activator protein cAMP receptor protein (CRP; also referred to as the catabolite gene activator protein, CAP), guides another DNA-binding protein, the CytR repressor, to a number of binding sites that share minimal sequence homology. The CytR repressor belongs to the LacI family of regulators, and possesses, like the other members, an N-terminal helix–turn–helix (HTH) DNA-binding motif (Valentin-Hansen et al., 1986; Weickert and Adhya, 1992). However, unlike a typical bacterial repressor, CytR binds with only modest affinity to its operators and cannot repress its cognate promoters independently in vivo. This deficiency is overcome by interaction with DNA-bound cAMP–CRP complexes. Thus, repression involves the formation of nucleoprotein complexes held together by multiple protein–DNA and protein–protein interactions (for review, see Valentin-Hansen et al., 1996). The synergy in the system is prominent; depending on which promoter is examined, the cAMP–CRP activator complex strengthens binding of CytR from 100- to several thousand-fold in vitro (Gerlach et al., 1991; Pedersen et al., 1991, 1992, 1995; Holst et al., 1992). At physiological CytR concentrations, the CytR–DNA interaction is absolutely required for formation of repression complexes. However, when overexpressed, a mutant CytR protein lacking its DNA-binding domain can repress the deoP2 promoter in vivo (Søgaard-Andersen and Valentin-Hansen, 1993). Thus, CytR's site of action can be specified solely by the contact with CRP. The apparently low sequence homology among the natural CytR-binding sites has, therefore, been explained by the fact that interactions between neighboring proteins on the DNA could provide the repression complex with adequate specificity, even without very specific CytR–DNA interactions. Recently, the DNA recognition specificities of several transcription factors including HSF, SRF, MyoD, E2A, Oct1 and GCN4 have been analyzed by in vitro binding site selection experiments (Blackwell and Weintraub, 1990; Mavrothalassitis et al., 1990; Pollock and Treisman, 1990; Verrijzer et al., 1992; Kroeger and Morimoto, 1994). We have taken this approach one step further, to define the optimal binding site of the CytR repressor, alone as well as in the presence of its helper protein, cAMP–CRP. We find that CytR by itself preferentially binds two octamer repeats, in direct or inverted orientation, and separated by 2 bp. Surprisingly, these DNA configurations are not optimal for co-binding with cAMP–CRP. Rather, cAMP–CRP stabilizes alternative DNA-binding modes of CytR. In one mode, CytR recognizes inverted octamer repeats separated by 10–13 bp; in another the repressor binds direct octamer repeats separated by 1 bp. Thus, CytR is a very adaptable DNA-binding protein that retains a high degree of flexibility even in the presence of its co-repressor. Results Selection strategies To define the optimal binding site for the CytR repressor, we used a PCR-based binding site selection assay (see Materials and methods). The starting material for the experiments was a population of ∼1011 DNA fragments carrying deoP2 promoter sequences in which the central 27 bp, containing the CytR-binding site, had been randomized. In a first set of experiments, the pool of DNA fragments was incubated with CytR, and an electrophoretic mobility shift assay was performed. DNA was isolated from the band containing CytR–DNA complexes and amplified by error-prone PCR, in order to increase the DNA sequence diversity before the next round of selection. After several rounds, the isolated DNA fragments were cloned, sequenced and characterized by mobility shift and footprinting assays. To investigate whether cAMP–CRP has any effect on the DNA recognition specificity of CytR, we repeated the selection/amplification experiments in the presence of both CytR and cAMP–CRP. Isolation of the optimal CytR-binding site The course of the selection for CytR operators (in the absence of cAMP–CRP) was evaluated by mobility shift assays (data not shown). The analyses revealed that after three rounds of selection, the enriched pool of DNA sequences has a higher affinity for CytR than deoP2 wild-type (wt) fragments, and that near-optimal binding sites for the CytR repressor are obtained after 4–5 rounds of selection. Hence, fragments obtained after 6–8 and 12 rounds were cloned and sequenced (Figure 1). All the sequences show homology to the octamer motifs 5′-AATGT/CAAC-3′ and 5′-GTTGCATT-3′. We have termed these left (L) and right (R) half-sites, respectively. Based on the orientation and number of octamer repeats, the selected fragments were divided into four groups (Figure 1). Figure 1.Isolation of CytR-binding sites. The deoP2-wt sequence is shown at the top; the originally randomized region is underlined. Sequences obtained from one experiment were divided into four groups, and gaps (−) introduced to display homologies. Bases are colored according to the consensus of Group A1: green color indicates the four bases that are identical in both halves of the consensus (--TGCA--); yellow and blue indicate bases specific for left (AA--T-AC) and right (GT---TT) half-operators, respectively. Numbers to the left of the sequences indicate the number of selection rounds performed (e.g. ‘12’), followed by a serial number (e.g. ‘05’). ‘A’ indicates that the selection was performed in the absence of cAMP–CRP. PCR amplification primers (deoprim1 and deoprim2) anneal immediately upstream and downstream of the presented sequences, respectively. Two copies of the A12-14 sequence was recovered; otherwise, all sequences were different. Download figure Download PowerPoint Figure 2.Isolation of CytR-binding sites that support CRP2–CytR–DNA complex formation. The deoP2-wt sequence is shown at the top. The CRP2- and CRP1-binding sites are centered at −93.5 and −40.5, respectively, relative to the transcription initiation site. cAMP–CRP complexes are hatched; CytR is in black. The CRP2–CytR–DNA repression complex covers ∼80 bp on one face of the DNA helix, and is held together by CRP–DNA, CytR–DNA and CRP–CytR interactions (Pedersen et al., 1991; Søgaard-Andersen et al., 1991a, b; Rasmussen et al., 1993). The originally randomized portion is indicated by thick underlining; the flanking half-site recognition motifs of the CRP targets are indicated by thin underlining (the centers of CRP2 and CRP1 are 3 bp upstream and downstream, respectively, of the sequences shown). Sequences from one selection experiment were divided into three groups, and gaps (−) introduced to show homologies. Bases are colored according to the consensus of Group C1: green indicates consensus bases that are identical in the two half-sites (–TGCA–); yellow and blue indicate bases specific for left (AC----AC or TG----AC) and right (GT-A--TT) half-operators, respectively. The PCR amplification primers (deoprim1 and deoprim2) anneal upstream of and including the tcgca sequence, and immediately downstream from the shown sequence, respectively. Numbers to the left of the sequence indicate the number of selection rounds (e.g. ‘13’) followed by a serial number (e.g. ‘01’). ‘C’ indicates that this selection was done in the presence of cAMP–CRP. Two copies each of C13-32 and C13-03 were recovered; the remaining sequences are different. Download figure Download PowerPoint Of 46 recovered fragments, two were identical. The majority (33 fragments) contain two inverted octamer repeats, separated by 2 bp (Group A1). Additionally, seven sequences contain inverted repeats separated by 1 bp; most of these isolates were obtained after a few (6–8) rounds of selection. Three sequences contain a direct repeat arrangement of two octamer motifs (Group A2), and one sequence contains three repeats (Group A3). Finally, two sequences (obtained after six or seven rounds of selection) carry only one repeat (Group A4). The consensus sequence of group A1 is a near perfect palindrome, 5′-AATGT/CAAC-GC-GTTGCATT-3′. There seems to be a preference for asymmetry. First, only one entirely symmetric 18 bp sequence (AATGTAAC-GC-GTTACATT) was obtained, while the consensus sequence (AATGTAAC-GC-GTTGCATT) was found in eight copies. Second, the L half-sites are more diverse than the R half-sites among A1 sequences. Finally, the octamer boxes in most of the fragments reside in the right end of the randomized 27 bp region. Presumably, the two thymidines at the 3′ end of the recognition motif are provided by the constant DNA region next to the randomized portion (see Figure 1). This may have biased the selection in favor of R versus L direct repeats. The affinity of CytR for individual DNA sequences was determined by mobility shift assays (Figure 3). Isolates obtained after 12 rounds of selection bind with almost indistinguishable affinities to CytR, and have 30-fold higher affinity for the repressor than the deoP2-wt promoter. The sequence A08-21 (L–1–R) with a spacing of 1 bp between the octamer motifs has a 2-fold lower affinity for CytR than the L–2–R or R–2–R sequences; DNA fragments containing a single octamer repeat bind CytR half as efficiently as the deoP2-wt promoter. Finally, A12-29 has an affinity for CytR similar to that of other sequences obtained after 12 selection rounds, despite the presence of an extra octamer repeat in this sequence. Figure 3.Apparent CytR affinity (1/KD) for the selected DNA. The apparent affinity of CytR for selected DNA sequences and deoP2-wt, in the absence (black bars) or presence (gray bars) of 6×10−9 M CRP, was determined by the mobility shift assay (see Materials and methods). The apparent dissociation constant, KD, was taken as the CytR concentration that binds 50% of the DNA fragments. CytR was in >100-fold excess to the binding site (cAMP–CRP absent); in the presence of cAMP–CRP, both proteins were in at least 5-fold excess to the binding sites. The cAMP concentration was 50 μM whenever CRP was employed. Note that previously published protein–DNA affinities were determined by footprinting; the mobility shift assay gives considerably higher affinities for the interaction of CytR with DNA (Pedersen et al., 1991, 1992, 1995). Download figure Download PowerPoint To gain insight into the qualitative interactions of CytR with DNA, we performed DNase I and dimethyl sulfate (DMS) footprinting at repressor concentrations that saturate the binding site. The DNase I footprints serve to delineate the binding sites; DMS footprinting identifies purines in close contact with protein. The salient features of the footprints of sequences A12-09 and A12-44, representing the inverted (L–2–R) and direct (R–2–R) repeat configuration, are as follows (Figure 4). The DNase I footprint covers ∼20 bp at both sequences, and, as expected, CytR interacts specifically with both operator half-sites. Interaction of CytR with the octamer motifs AATGTAAC and GTTGCATT invariably protects the central guanine from DMS methylation, consistent with the well conserved G at this position (Figure 1). Methylation of the adenine at position 6 is only observed in the L half-site of A12-09. Figure 4.DNase I and DMS footprints of CytR and cAMP–CRP. (A) Footprinting of A12-09, C13-20, C09-20 and C09-17. The left and right halves of the panels show DNase I and DMS footprints, respectively. (*) denotes guanines protected from DMS methylation by CytR; DNase I enhancements in the CytR-binding region are indicated by (+). Final CRP concentrations: lanes 1–3, no CRP added; lanes 4–6, 5×10−8 M CRP. Final CytR concentrations: lanes 1, no CytR; lanes 2, 5×10−8 M CytR; lanes 3, 10−8 M CytR; lanes 4, no CytR; lanes 5, 5×10−9 M CytR; lanes 6, 5×10−10 M CytR. cAMP was added to 50 μM in reactions containing CRP. The initially randomized region (27 bp), and the CRP1 and CRP2 targets are indicated to the left of each panel. (B) Schematic representation of DNase I and DMS footprints of sequences isolated in the absence of cAMP–CRP (A12-09 and A12-44) or in its presence (C13-20, C13-40, C09-20, C13-26, C13-05 and C09–17). Circles and squares indicate positions affected by independent CytR binding and by formation of the CRP2–CytR–DNA complex, respectively. Empty/filled symbols indicate weak/strong effects. Nucleotides protected from DNase I cleavage by CytR, or by formation of the CRP2–CytR–DNA complex, are shown by thin and thick underlining, respectively. Coloring is as in Figure 1 (A-sequences) or Figure 2 (C-sequences). Download figure Download PowerPoint Isolation of DNA sequences that support CRP2–CytR–DNA complex formation We next performed the selection in the presence of both CytR and cAMP–CRP. As the two protein species bind cooperatively to deoP2, even when the entire sequence between the two CRP targets has been randomized (Søgaard-Andersen and Valentin-Hansen, 1993), we were able to obtain a discrete band containing the quaternary CRP2–CytR–DNA complex (Figure 2), in the first round of selection. Analytical mobility shift assays showed that after eight rounds of selection the DNA pool forms the CRP2–CytR–DNA complex more efficiently than the deoP2-wt promoter (data not shown). Sequencing revealed 34 unique DNA fragments out of a total of 36 recovered. Many different CytR-binding site configurations seem to allow cooperative interaction of CytR with cAMP–CRP. We have divided them into three groups (Figure 2). The largest group, consisting of 23 sequences (Group C1), contains near perfect palindromes with the consensus 5′-AC/TGTGCAAC-Nx-GTTG/ACATT-3′, x = 10, 11, 12 or 13. The 5′ end of the motif is either AC or TG, suggesting that these nucleotide steps share a common feature of importance for repression complex formation at C1 sequences. Six sequences have two octamer boxes separated by 1 bp, in either R–1–R or L–1–L direct repeat configuration (Group C2). The thymidine at the 3′ end of R half-sites seems to be provided by the constant region next to the originally randomized region, and may have biased the selection in favor of C1 sequences. No everted (R–L) sequences were isolated. Finally, five DNA isolates contain three repeats, in various orientations (Group C3). The two outermost octamer motifs of this group are inverted repeats separated by 10–14 bp; moreover, three of the C3 sequences bear two octamer boxes in a direct repeat arrangement, and separated by 1 bp. Thus, group C3 sequences show homology to both C1 and C2 sequences. There seem to be rather strict rules for the position of the L and R CytR half-sites relative to the flanking CRP sites. For C1 sequences, the separation between the TCGCA motif of CRP2 and the L repeat is almost exclusively 5 or 6 bp, and the spacing between the R repeat and TGTGA of CRP1 is 4 bp. Correspondingly, for C2 sequences, the R–1–R and L–1–L repeats are separated from the CRP2 and CRP1 targets by 14/6 and 6/14 bp, respectively. Thus, the C13-26 sequence with the L–1–L arrangement is basically the inversion of the R–1–R sequences. The independent affinity of CytR for the C-sequences is considerably lower than for the A-sequences (2- to 5-fold lower; Figure 3, black bars). Independent binding of CytR to the selected sequences thus produces a hierarchy of affinities, A-operators>C-operators>deoP2-wt-operator. In the presence of cAMP–CRP, CytR binds the C1, C2 and C3 sequences with very similar affinity (Figure 3, hatched bars). However, the cooperativity exhibited by the C-fragments (∼20- to 50-fold) is less than that of the deoP2-wt fragment (∼160-fold). As a result, the cAMP–CRP-dependent affinity of CytR for the C-sequences is only 2-fold higher than for deoP2-wt. Thus, the high independent affinity of CytR for the C-sequences, relative to deoP2-wt, has not resulted in a corresponding increase in cAMP–CRP-dependent affinity. We could not find A-sequences for which it would make sense to test for cooperative binding, since most of the A-fragments contain mutations in the CRP targets, or because the binding sites are too close to allow simultaneous binding of cAMP–CRP and CytR. However, C09–14 from the cAMP–CRP-dependent selection (Figure 2, last sequence in Group C1) contains two inverted repeats separated by 2 bp, like sequences from the independent selection (Group A1). The cooperative binding of CytR and cAMP–CRP to this sequence is very inefficient, even though the independent CytR affinity is similar to that of other C-sequences (Figure 3). This implies that it is the configuration of the CytR half-sites and the position of these half-sites relative to the flanking CRP targets, and not the strength of the CytR–DNA interaction, that are important for cooperative binding of CytR and cAMP–CRP. Two sequences from each of groups C1, C2 and C3 were footprinted by DNase I and DMS in the presence of CytR and cAMP–CRP (Figure 4). As observed for the A-sequences, CytR protects the central guanine of the octamer motifs from DMS methylation, and only L half-sites exhibit increased DMS reactivity of the adenine at position 6. Notably, the footprint patterns in the CytR region are relatively independent of the presence of cAMP–CRP. For example, in the DNase I footprints, the C1 sequences (C13-20 and C13-40) exhibit a 3 bp unprotected region in the middle of the CytR operator, and C09-20 (R–1–R) and C13-26 (L–1–L) contain unprotected regions to the left or right of the operator, regardless of the presence of cAMP–CRP. These results imply that a CytR protein bound to C-sequences does not change shape upon addition of cAMP–CRP. It thus appears that selection in the presence of cAMP–CRP has produced DNA targets to which CytR binds in a conformation that is designated to interaction with cAMP–CRP. Independent binding of CytR to isolates containing three octamer boxes does not result in the simple pattern described above. The DNase I footprints are extended, and cover all three repeats; correspondingly, all three repeats exhibit DMS protection. This is presumably caused by simultaneous binding of two CytR molecules, or by a mixed population of complexes in which one CytR binds either two of the three repeats. The combined footprints on these isolates, however, resemble those of group C1 and C2 sequences: Addition of cAMP–CRP to the C09-17 sequence creates a C09-20 (R–1–R)-like footprint, and the C13-05 footprint resembles those of C13-40 (L-11-R) and C13-20 (L-13-R). Discussion Previous studies have revealed that protein–DNA and protein–protein interactions, as well as protein-induced DNA-bending, cooperate in an organized manner to form repression complexes at CytR-regulated promoters (Søgaard-Andersen et al., 1991a; Pedersen et al., 1992; Søgaard-Andersen and Valentin-Hansen, 1993). In these complexes, the DNA bends strongly around cAMP–CRP, bringing the DNA-bound regulators into close proximity (Søgaard-Andersen et al., 1991b; Crothers and Steitz, 1992). We show here that besides acting as an architectural element, and providing contacts through its surface, CRP also alters the DNA-binding mode of the CytR repressor. Furthermore, our results reveal that widely different conformations of the CytR repressor can cooperate with cAMP–CRP to form nucleoprotein complexes of equal stability. Below we discuss these novel aspects and their implications for CRP/CytR combinatorial regulation and multiprotein–DNA complex formation in general. cAMP–CRP changes the DNA-binding mode of CytR The present work allows us to define an 8 bp half-operator consensus (ATTGT/CAAC) for CytR. In the absence of cAMP–CRP, the CytR repressor binds two such octamer motifs in inverted or direct repeat arrangement, preferably separated by 2 bp (Figure 6A). The CytR half-operator consensus for cooperative binding with cAMP–CRP is slightly different. The differences are at the left edges of the CytR operator, and might facilitate wrapping of the DNA helix around the proteins in the combined complex. Alternatively, since the strong CRP2 target is expected to span ∼30 bp (Liu-Johnson et al., 1986), the sequence at the edges of the CytR operator may optimize both cAMP–CRP and CytR interactions with DNA. Nevertheless, the absence of major changes to the half-operator consensus indicate that cAMP–CRP only minimally interferes with the structures of the individual DNA-binding domains of CytR. cAMP–CRP does, however, induce drastic changes in CytR's quaternary structure. Thus, cAMP–CRP preferentially stabilizes a set of CytR conformations that fit operators composed of inverted repeats with wide spacing (10–13 bp), or direct repeats separated by 1 bp. As illustrated in Figure 6B and C, these DNA arrangements are expected to bind either a roughly symmetrical conformation of the CytR dimer in which the DNA-binding domains are held in an inverted orientation with their recognition centers approximately two DNA helical turns apart, or an asymmetric conformation in which the DNA-binding motifs are in direct orientation, with their centers approximately one helical turn apart. Generality of the isolated DNA sequences One complication in devising a general selection scheme for cAMP–CRP-dependent CytR-binding sites was that cooperative binding of cAMP–CRP and CytR had been observed on DNA templates with 52–54 bp separations between the centers of the two CRP targets (Søgaard-Andersen et al., 1990; Pedersen et al., 1991; Holst et al., 1992). Thus, variation in rotational and translational separation of the flanking CRP sites could potentially affect CytR's DNA-binding mode, and, consequently, conclusions derived in the context of one promoter might not necessarily apply to others. This potential problem was circumvented by employing error-prone PCR during each selection/amplification round. In good agreement with results obtained for natural promoters, the selected C-sequences have CRP–CRP distances of 51–54 bp (Figure 2). Therefore, it is plausible that the conclusions established from the biochemical experiments with deoP2 apply in vivo as well, and should be applicable to other promoters of the CytR regulon. In fact, natural CytR-binding sites exhibit a striking homology to the cAMP–CRP-dependent binding sites (C-sequences) identified in the present work (Figure 5). The CytR operator of the cdd promoter contains R–1–R direct repeats, and the repeats are separated by 5 and 14 bp, respectively, from the flanking CRP targets. This is very similar to the arrangement of group C2 sequences (Figure 2). Also, the CytR-binding site in the cytR promoter (L–1–L) matches the C2 sequences. nupG is a representative of group C1 sequences, and the arrangement of the three CytR repeats in the divergent cytX-rot promoter region corresponds to the group C3 sequences. Finally, the CytR operator site of the wild-type deoP2 promoter deviates somewhat from the consensus sequences. Thus, the CytR half-operators in deoP2 are immediately adjacent, and their orientation is not obvious. However, the relative positions of the CytR- and CRP-binding sites resemble those of group C2 sequences. Taken together, the configurations of the natural CytR operators and the selected C-sequences are very similar. Thus, the heterogeneous nature of natural CytR operators, which appeared at first rather puzzling, can now be understood in detail. In this regard, we note that combination of many sub-optimal interactions seems to be inherent in the design of gene regulatory systems composed of multiple factors. The CRP/CytR regulatory system obviously contains many adjustable parameters; for a given natural promoter, only a subset of these is optimized, thus preserving a dynamic regulatory circuit. Figure 5.Natural CytR-binding sites. The sequences of five CytR-regulated promoters are compared with the cAMP–CRP-dependent CytR consensus determined in this report. Coloring is as in Figure 2. One half-site recognition sequence in each of CRP2 and CRP1 is underlined; the CRP consensus sequence is TGTGA-N6-TCACA (de Crombrugghe et al., 1984; Ebright et al., 1989). The centers of the flanking CRP targets are 3 bp upstream and downstream from the shown sequence. CRP- and CytR-binding sites were determined by deoP2 (Valentin-Hansen, 1982; Pedersen et al., 1991; Rasmussen et al., 1993), cdd (Holst et al., 1992), cytR (Pedersen et al., 1992); nupG (Pedersen et al., 1995); and cytX-rot (Nørregaard-Madsen et al., 1994). Download figure Download PowerPoint Figure 6.cAMP–CRP-induced conformational changes of the CytR repressor. (A) In the absence of cAMP–CRP, the CytR repressor preferentially binds octamer repeats in direct or inverted orientation, separated by 2 bp. In the presence of cAMP–CRP, CytR recognizes (B) inverted octamer repeats separated by 10–13 bp or (C) direct repeats separated by 1 bp. Download figure Download PowerPoint Structural implications: the CytR repressor protein CytR is a dimer in solution and when bound to the deoP2 promoter (H-H.Kristensen et al., 1996). On the basis of the characteristics of the selected operators, the footprinting patterns and the migration rate of the CytR–DNA and CRP2–CytR–DNA complexes in the gel mobility shift experiments (this study; data not shown), we conclude that CytR also binds as a dimer to the selected A- and C-sequences. CytR belongs to the LacI repressor family and exhibits extensive amino acid sequence homology to several other members (e.g. PurR, LacI and GalR; Weickert and Adhya, 1992). Based on the PurR–DNA co-crystal structure (Schumacher et al., 1994), and three crystal structures of LacI (as free protein and in complex with operator DNA or inducer; Lewis et al., 1996), CytR is expected to consist of an N-terminal DNA-binding domain of ∼60 amino acids, connected via a hinge region to a ∼270 amino acid C-terminal domain that mediates dimerization and ligand binding. The ability of CytR to contact operators with widely different half-site spacings reveals a rotational and translational flexibility th
Referência(s)