Portal de Periódicos da CAPES

Protein–Protein Contacts that Activate and Repress Prokaryotic Transcription

1998; Cell Press; Volume: 92; Issue: 5 Linguagem: Inglês

10.1016/s0092-8674(00)81126-5

1097-4172

Ann Hochschild, Simon L. Dove,

Bacteriophages and microbial interactions

Many prokaryotic activators and repressors contact RNA polymerase (RNAP) directly, and this minreview focuses on recent work that illuminates the mechanistic consequences of these protein–protein contacts. An important implication of this work is that the promoter is a critical determinant that dictates whether a given protein–protein contact will have an effect on transcription, what the effect will be (activation or repression), and which specific step(s) in the initiation process will be targeted. In vitro studies have led to the following general picture of transcription initiation. RNAP first recognizes and binds to double-stranded promoter DNA, forming a complex that is referred to as the "closed" complex. This must then isomerize to form a transcriptionally active open complex in which the DNA strands are locally melted. At this stage RNAP can direct the synthesis of short abortive products, but for full-length transcripts to be generated, RNAP must also escape from the promoter; this involves breaking contacts that stabilize the open complex. Thus, activators and repressors can, in theory, affect closed complex formation, open complex formation, or promoter clearance. The structures of a typical E. coli promoter and RNAP are outlined in Figure 1A. The promoter is recognized by RNAP holoenzyme, which consists of the enzymatic core (α2ββ′) and a σ factor that directs binding to a specific promoter class. Many transcriptional activators in E. coli bind to specific sites located upstream of the transcription start point. Furthermore, most of the activators that have been examined so far appear to interact with either the α subunit (its C-terminal domain [CTD], in particular) or the σ70 subunit (the most commonly used σ factor), both of which can bind DNA. A DNA-bound activator that contacts the α-CTD is likely to function, at least in part, by stabilizing the binding of RNAP to the promoter, thus facilitating closed complex formation. A classic example is provided by the cyclic AMP receptor protein (CRP) working at the lac promoter (p lac). Here, CRP binds to a site centered 61.5 bp upstream of the transcriptional start site and interacts specifically with the α-CTD, apparently stabilizing its association with the DNA between the CRP-binding site and the promoter −35 region (Figure 1B) (2Busby S Ebright R.H Cell. 1994; 79: 743-746Abstract Full Text PDF PubMed Scopus (345) Google Scholar). Although this activation targets a subunit of RNAP that interacts directly with the DNA, in principle, any sufficiently strong protein–protein interaction between a DNA-bound activator and any subunit of RNAP should stabilize the binding of RNAP to the promoter. That such a protein–protein contact itself is sufficient to activate transcription was demonstrated with an artificial activator (5Dove S.L Joung J.K Hochschild A Nature. 1997; 386: 627-630Crossref PubMed Scopus (221) Google Scholar; see also4Dove S.L Hochschild A Genes Dev., in press. 1998; Google Scholar). Specifically, the α-CTD was replaced by a heterologous protein domain (X) with no determinants for binding to DNA (see Figure 1C). Contact between a DNA-bound protein domain (Y) and the protein domain fused to α (X) resulted in transcriptional activation (Figure 1C) (presumably by facilitating closed complex formation) and the strength of the engineered protein–protein interaction correlated with the magnitude of the activation (5Dove S.L Joung J.K Hochschild A Nature. 1997; 386: 627-630Crossref PubMed Scopus (221) Google Scholar). The implication that any subunit of RNAP can, in principle, serve as an activation target, has been confirmed by the demonstration that protein domains fused to the omega subunit of RNAP, a small RNAP-associated protein of unknown function, can also mediate the effects of artificial activators (4Dove S.L Hochschild A Genes Dev., in press. 1998; Google Scholar). These findings with artificial activators that make arbitrary contacts with RNAP suggest that natural activators that contact the σ subunit might also function by stabilizing the binding of RNAP to the promoter. For illustration we will consider the cI protein of bacteriophage λ (λcI), which stimulates transcription of its own gene from promoter PRM when bound at a site centered 42 bp upstream of the transcription start point. This activation apparently depends on a specific, genetically defined interaction between λcI and the domain of σ that binds the promoter −35 region, and molecular modeling suggests that the λcI activating region is closely juxtaposed to this domain of σ (see2Busby S Ebright R.H Cell. 1994; 79: 743-746Abstract Full Text PDF PubMed Scopus (345) Google Scholarreferences therein). Nevertheless, kinetic studies have indicated that λcI has no effect on the stabilization of the closed complex at PRM, but rather increases the rate at which this complex is converted to the transcriptionally active open complex (i.e., the isomerization step). Although the kinetic data have previously been taken to imply that the mechanisms of action of λcI and CRP working at PRM and p lac, respectively, are fundamentally different, it has recently been proposed (16Ptashne M Gann A Nature. 1997; 386: 569-577Crossref PubMed Scopus (901) Google Scholar; R. Ebright, personal communication) that both work by stabilizing binding of RNAP to these promoters. Accordingly, the kinetic data are explained by postulating that in the closed complex the activating region of λcI and the complementary surface on RNAP are not properly aligned, but that this alignment occurs during the isomerization process, thus stabilizing a transition state between the closed and open complexes. This view is supported by several observations. First, λcI and RNAP bind cooperatively to specific DNA templates in vitro under conditions that permit the detection of open, but not closed, promoter complexes (as do CRP and RNAP at p lac). This cooperative binding is not detected with mutant forms of the activators that are specifically defective for transcriptional activation. Second, a recent study with a mutant form of σ has revealed that λcI can stimulate transcription from PRM by stabilizing the closed complex (9Li M McClure W.R Susskind M.M Proc. Natl. Acad. Sci. USA. 1997; 94: 3691-3696Crossref PubMed Scopus (43) Google Scholar). This activation depends on the same essential residue on λcI as is required for activation with wild-type RNAP. The simplest view consistent with these findings is that, depending on the molecular details of the interacting surfaces, λcI interacts favorably with σ either during the isomerization step, or initially when the closed complex is formed, in both cases stabilizing interaction of RNAP with the promoter. An attractive possibility, based on the proximity of the activating region of λcI to the promoter −35 region recognition motif of σ, is that λcI stabilizes the association of this domain of σ with the −35 region. The available evidence does not however exclude more complex mechanisms involving, for example, the induction of an activator-specific conformational change in σ that facilitates the isomerization reaction, or even a repulsive interaction that drives isomerization. The activities of p lac and λPRM, and presumably many regulated promoters in E. coli, are limited, at least in part, by promoter occupancy (see16Ptashne M Gann A Nature. 1997; 386: 569-577Crossref PubMed Scopus (901) Google Scholar). Thus, in vivo evidence indicates that both p lac and λPRM can be activated by contacts between DNA-bound proteins and protein domains fused to RNAP. Furthermore, in vitro, transcription from such promoters can be stimulated by the use of high concentrations of RNAP (i.e., concentrations higher than those required in the presence of activator). These two properties can be used as criteria to classify promoters that are similarly limited by promoter occupancy. This leaves open the question of how any particular activator might work at a promoter from this class, but suggests a plausible general mechanism; activation may be achieved by increasing the probability that a transcriptionally active complex will be formed before RNAP has a chance to dissociate from the promoter. In contrast, the glnA promoter (p glnA) in Salmonella is an example of an activatable promoter whose activity is not limited by promoter occupancy (15North A.K Kustu S J. Mol. Biol. 1997; 267: 17-36Crossref PubMed Scopus (51) Google Scholar). p glnA is recognized by a form of RNAP in which σ70 is replaced by an alternative sigma factor (σ54). In the absence of the activator NTRC, this σ54-dependent promoter binds RNAP in a stable, but transcriptionally inactive, closed complex. NTRC binds to enhancer-like sequences and catalyzes the isomerization of this closed complex to a transcriptionally active open complex, in an ATP-dependent reaction. Because the activity of p glnA is not limited by promoter occupancy, it is unaffected by high concentrations of RNAP in vitro and should not be stimulated by arbitrary protein–protein interactions in vivo. Studies on the regulation of T4 phage late gene expression have led to the identification of a DNA-tracking protein, the Gp45 sliding clamp, that activates transcription. The activator, when loaded onto an enhancer, tracks along the DNA toward the promoter by one-dimensional diffusion and activates transcription of the phage's late genes by E. coli RNAP core in complex with the phage-encoded late-gene specific σ factor, Gp55 (Figure 1D). The sliding clamp activator affects late gene transcription through specific interactions with both Gp55 and a phage-encoded coactivator, Gp33, and these two interactions contribute synergistically to the activation of transcription. The observation that the phage-encoded sigma factor can cotrack along the DNA with the sliding clamp suggests that the Gp45–Gp55 complex may function in part by capturing RNAP core and delivering it to the promoter, thus increasing the rate of promoter binding. The sliding clamp, which is a component of stable open transcription complexes, might also stabilize these complexes on the DNA. Also, the clamp might facilitate promoter melting, perhaps stabilizing a relevant transition state through its contacts with Gp55 and Gp33 (20Sanders G.M Kassavetis G.A Geiduschek E.P EMBO J. 1997; 16: 3124-3132Crossref PubMed Scopus (33) Google Scholarreferences therein). A recent study has demonstrated that a phage-encoded transcriptional activator works without binding to DNA (11Miller A Wood D Ebright R.H Rothman-Denes L.B Science. 1997; 275: 1655-1657Crossref PubMed Scopus (59) Google Scholar). The bacteriophage N4 single-stranded DNA binding protein (N4 SSB) specifically activates transcription of the phage's late genes by E. coli RNAP holoenzyme (containing σ70), in this role binding neither single-stranded nor double-stranded DNA. This activation is mediated through a specific contact with the β′ subunit of RNAP and has been proposed to involve a postrecruitment step in the initiation process (11Miller A Wood D Ebright R.H Rothman-Denes L.B Science. 1997; 275: 1655-1657Crossref PubMed Scopus (59) Google Scholar). It should be noted, however, that activators that contact RNAP but are not associated with the DNA cannot always be presumed to act on preformed complexes; they could, in principle, enhance the binding of RNAP to particular promoters. Transcription can be activated by several DNA-bound regulators each of which interacts directly with RNAP (see2Busby S Ebright R.H Cell. 1994; 79: 743-746Abstract Full Text PDF PubMed Scopus (345) Google Scholar). In principle, two or more DNA-bound regulators can affect different kinetic steps or act simultaneously at a single step. Recent studies of CRP (14Niu W Kim Y Tau G Heyduk T Ebright R.H Cell. 1996; 87: 1123-1134Abstract Full Text Full Text PDF PubMed Scopus (223) Google Scholar, 17Rhodius V.A West D.M Webster C.L Busby S.J.W Savery N.J Nucleic Acids Res. 1997; 25: 326-332Crossref PubMed Scopus (70) Google Scholar) and the bacteriophage Mu Mor protein (1Artsimovitch I Murakami K Ishihama A Howe M J. Biol. Chem. 1996; 271: 32343-32348Crossref PubMed Scopus (43) Google Scholar) have revealed that a single DNA-bound activator can also interact with two different targets on RNAP. The analysis of CRP has further shown that the two interactions affect separate kinetic steps (14Niu W Kim Y Tau G Heyduk T Ebright R.H Cell. 1996; 87: 1123-1134Abstract Full Text Full Text PDF PubMed Scopus (223) Google Scholar). Although genetic analysis performed with p lac initially led to the identification of a single activating region on CRP (now called AR1), subsequent studies performed with a different promoter resulted in the identification of a second activating region (called AR2). The location of the CRP-binding site(s) varies from one CRP-regulated promoter to another and the promoters have been classified accordingly. At class I promoters (such as p lac), the CRP-binding site is centered at least 61.5 bp upstream of the transcription start point, whereas at class II promoters a single CRP-binding site is centered at position −41.5, overlapping the −35 region. When bound at a class II promoter, CRP utilizes both AR1 and AR2 to contact the α-CTD and the α-NTD, respectively (Figure 1E), thereby stimulating both closed complex formation (with AR1) and the rate of open complex formation (with AR2). Thus, the mechanism of activation by CRP, and presumably other activators, depends on the architecture of the target promoter and, in principle, also on the step or steps that are rate-limiting for that promoter. In fact, not all class II promoters are equally dependent on AR1 of CRP for transcriptional activation, and these differences were found to depend on the sequence of the −35 region (17Rhodius V.A West D.M Webster C.L Busby S.J.W Savery N.J Nucleic Acids Res. 1997; 25: 326-332Crossref PubMed Scopus (70) Google Scholar), suggesting that they reflect differences in the kinetic parameters of the promoters. An artificial activator has also been described that can make two contacts with RNAP. In this example, the λcI protein uses its natural activating region to contact the σ subunit of RNAP and an artificial activating region (within domain Y) to contact an α chimera (see above) (Figure 1F), and the effect of the two contacts on transcription is synergistic (5Dove S.L Joung J.K Hochschild A Nature. 1997; 386: 627-630Crossref PubMed Scopus (221) Google Scholar). This artificial form of activation has not yet been examined kinetically, but it is likely that the engineered contact stabilizes the closed complex, whereas the natural contact presumably accelerates the rate of open complex formation (see above). Although the α, σ, β′, and β subunits of RNAP have all been implicated as natural activation targets (2Busby S Ebright R.H Cell. 1994; 79: 743-746Abstract Full Text PDF PubMed Scopus (345) Google Scholar, 8Lee J.H Hoover T.J Proc. Natl. Acad. Sci. USA. 1995; 92: 9702-9706Crossref PubMed Scopus (84) Google Scholar, 11Miller A Wood D Ebright R.H Rothman-Denes L.B Science. 1997; 275: 1655-1657Crossref PubMed Scopus (59) Google Scholar), a survey of natural activators suggests that the α-CTD and the σ subunit, are preferred targets. Why might this be? Firstly, the α-CTD is particularly accessible to activators bound at a variety of positions upstream of the promoter. It is flexibly tethered to the α-NTD, and hence to the body of RNAP, and is known to be able to contact the DNA over a range of positions extending from just upstream of the promoter −35 region to as far as 90–100 bp upstream of the transcription start point (13Murakami K Owens J.T Belyaeva T.A Meares C.F Busby S.J.W Ishihama A Proc. Natl. Acad. Sci. USA. 1997; 94: 11274-11278Crossref PubMed Scopus (77) Google Scholar), thus minimizing geometric constraints. Secondly, since the α-CTD has the potential to bind the DNA (see2Busby S Ebright R.H Cell. 1994; 79: 743-746Abstract Full Text PDF PubMed Scopus (345) Google Scholar, 6Gaal T Ross W Blatter E.E Tang H Jia X Krishnan V.V Assa-Munt N Ebright R.H Gourse R.L Genes Dev. 1996; 10: 16-26Crossref PubMed Scopus (170) Google Scholar), a relatively weak protein–protein interaction can mediate a significant amount of stimulation through the stabilization of suboptimal contacts between the α-CTD and the DNA. Accordingly, the energetic requirements for a DNA-bound activator that contacts the α-CTD would be expected to be smaller than for an artificial activator that contacts an α-chimera bearing a protein domain with no determinants for DNA binding. Finally, the α subunit is present in two copies, and the two α-CTDs can bind to the DNA independently of one another (13Murakami K Owens J.T Belyaeva T.A Meares C.F Busby S.J.W Ishihama A Proc. Natl. Acad. Sci. USA. 1997; 94: 11274-11278Crossref PubMed Scopus (77) Google Scholar), thus providing two independent potential regulatory targets. The σ subunit, which makes direct contacts with both the −10 and −35 regions of the promoter, is evidently also accessible to activators bound upstream of the promoter, in this case in the immediate vicinity of the −35 region. Presumably, this arrangement positions the activator so that it can, in principle, stabilize the binding of the relevant domain of σ to the −35 region, which would again permit a relatively weak protein–protein interaction to mediate a significant amount of stimulation through the stabilization of suboptimal contacts between an RNAP subdomain and the DNA. Thus, it is likely that both the α-CTD and the σ subunit are preferred targets for DNA-bound activators partly because relatively weak protein–protein interactions can be used to stimulate transcription, a strategy that ensures that a given activator interact with its target only when that activator is appropriately positioned on the DNA in the vicinity of the promoter to be regulated. In particular, such an activator, when present at physiological concentrations, would not bind to RNAP in solution, and would not therefore compete with other DNA-bound activators for access to the relevant target surface on RNAP (a phenomenon that has been called squelching [see16Ptashne M Gann A Nature. 1997; 386: 569-577Crossref PubMed Scopus (901) Google Scholar). An increasing body of evidence suggests that protein–protein contact between a DNA-bound regulator and RNAP can repress as well as activate transcription (see, for example,3Choy H.E Park S.W Aki T Parrack P Fujita N Ishihama A Adhya S EMBO J. 1995; 14: 4523-4529Crossref PubMed Scopus (93) Google Scholar). Recent studies of the bacteriophage φ29 p4 protein provide a particularly striking example. In this case, precisely the same protein–protein contact between the regulator and RNAP can lead either to repression or to activation depending on the characteristics of the target promoter (12Monsalve M Calles B Mencia M Salas M Rojo F Mol. Cell. 1997; 1: 99-107Abstract Full Text Full Text PDF PubMed Scopus (54) Google Scholar). When activating transcription, protein p4 binds upstream of the A3 promoter and interacts with the α-CTD, thereby stabilizing the closed complex. Surprisingly, protein p4 also interacts specifically with the α-CTD to repress transcription from the A2c promoter, and this interaction depends on the same residues of p4 (located near its C terminus) as are required for activation. However, in this case, although the formation of open complexes is not inhibited, the interaction impedes promoter clearance, i.e., results in a reduction in the number of full-length transcripts that are generated. What then accounts for the repressive effect of this interaction at the A2c promoter and the stimulatory effect of the same interaction at the A3 promoter? Recent findings have shown that it is the sequence of the promoter (specifically, the degree of similarity of the −35 region sequence to the consensus) that determines the regulatory outcome (12Monsalve M Calles B Mencia M Salas M Rojo F Mol. Cell. 1997; 1: 99-107Abstract Full Text Full Text PDF PubMed Scopus (54) Google Scholar). The A2c promoter bears a near consensus −35 element, whereas the A3 promoter lacks a recognizable −35 box; introduction of a consensus −35 box upstream of the A3 −10 box caused a switch in the activity of protein p4, which now repressed transcription from the modified promoter. Conversely, elimination of the −35 element within the A2c promoter caused p4 to activate transcription from the modified promoter. These findings indicate that the same protein–protein interaction can stabilize closed complex formation at a promoter that is inefficiently recognized by RNAP or impede promoter clearance at a promoter that is efficiently recognized, the latter effect presumably occurring because overstabilization of the open complex traps RNAP at the promoter. The Gal repressor has also been shown to work either as a repressor or a weak activator when bound at a single site located upstream of the overlapping gal promoters. In this case, although both effects require the α-CTD of RNAP, the regulatory outcome is apparently not determined by the kinetic parameters of the promoter, but rather by the angular alignment of the repressor-binding site with respect to the promoter (3Choy H.E Park S.W Aki T Parrack P Fujita N Ishihama A Adhya S EMBO J. 1995; 14: 4523-4529Crossref PubMed Scopus (93) Google Scholar). Recently the structures of two of the DNA-binding domains of RNAP, the α-CTD (7Jeon Y.H Negishi T Shirakawa M Yamazaki T Fujita N Ishihama A Kyogoku Y Science. 1995; 270: 1495-1497Crossref PubMed Scopus (151) Google Scholar, 6Gaal T Ross W Blatter E.E Tang H Jia X Krishnan V.V Assa-Munt N Ebright R.H Gourse R.L Genes Dev. 1996; 10: 16-26Crossref PubMed Scopus (170) Google Scholar), and a portion of σ70 containing the −10 region recognition motif (10Malhotra A Severinova E Darst S.A Cell. 1996; 87: 127-136Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar), have been determined. Such advances in the understanding of RNAP structure should facilitate elucidation of the more complex activation and repression mechanisms. In the case of σ70, this structural information in conjunction with the finding that σ plays an important role in directing and stabilizing promoter melting is likely to shed light on the mechanism of action of at least some activators that mediate their effects through σ. Since −10 region recognition involves base-specific contacts between the σ subunit and the melted nontemplate strand (18Roberts C.W Roberts J.W Cell. 1996; 86: 495-501Abstract Full Text Full Text PDF PubMed Scopus (119) Google Scholar), it is possible that regulators that interact with σ may, in some cases, stabilize a conformation that favors the formation of these contacts, rather than merely stabilizing the binding of the −35 region recognition domain. Unfortunately, there is as of yet no high resolution structural information about the portion of σ that binds the promoter −35 region, the apparent target of a number of activators that bind in the immediate vicinity of the −35 box. Whether or not the effects of such activator–σ interactions can be transmitted through the structure of σ to the −10 region recognition motif remains to be learned. As the structural analysis of RNAP proceeds, it will be particularly informative to study complexes containing a DNA-bound regulator together with a relevant portion of RNAP. Finally, structural information about the catalytic subunits of RNAP are likely to enhance our understanding of how some activators that contact these subunits work.