Coactivation of the RpoS-Dependent proP P2 Promoter by Fis and Cyclic AMP Receptor Protein

Plasmids.

Plasmid pRJ4069 was used as a template for most of the single-round in vitro transcription assays (57). It contains a segment of the proP sequence from +109 to −200 with respect to the P2 transcription initiation site upstream of the rrnB terminators (T₁T₂) in the vector pKK223-3 (Pharmacia). The DNA template with the mutation that prevents CRP binding (pRJ4070) was derived from pRJ4069 (Fig. 1) (58). pRJ1677 (+5 insertion in Fis site II) was created from pRJ4069 by a two-step PCR method using an oligonucleotide ranging from −66 to −100 nucleotides that contained a 5-bp insertion between nucleotides −81 and −82 as shown in Fig. 1. For multiple-round transcription reactions, the plasmids used were pRJ4039 (wild type), pRJ4045 (mutations in Fis site II), and pRJ4051 (mutations in Fis site I) (57). All of these plasmids are proP-104Δ1–lacZ fusion derivatives of the lacZ protein fusion vector pRS414 (50). The sequence changes in these mutations are as noted in Fig. 1 and have been shown previously to strongly reduce or abolish Fis or CRP binding to their respective sites (56, 58).

The in vivo β-galactosidase assays were performed using two sets of plasmids containing CRP. pZYCRP was the parent of the activation region 1 (AR1) mutations, which were obtained from R. Ebright, Rutgers University. The AR2 mutation (H19YK101E) was derived from pDCRP⁺, which was provided by S. Busby, University of Birmingham.

Proteins.

Fis wild-type and mutant proteins were purified as described elsewhere (7, 22, 45). CRP was obtained from J. Krakow, Hunter College. The core RNA polymerase enzyme was isolated by the protocol of Burgess and Jendrisak (9) and Lowe et al. (36) as previously described (57). The ς³⁸ protein was overproduced from plasmid pETF and purified as described elsewhere (51). To purify the α-CTD, extracts containing overproduced levels of the α-CTD peptide (amino acids 245 to 329) were prepared from cells containing pEBT7-αCTD (20). DNA was removed by precipitation with polyethyleimine in the presence of 1 M NaCl, and the supernatant was subjected to a 50 to 80% ammonium sulfate fractionation. This material was loaded onto a phenyl-Sepharose (Pharmacia) column in 2.2 M ammonium sulfate–50 mM NaCl–20 mM Tris-HCl (pH 7.5)–10% glycerol–1 mM dithiothreitol–0.1 mM EDTA and washed extensively with the same buffer. Near-homogeneous preparations of the α-CTD peptide were obtained after elution with the same buffer minus ammonium sulfate.

In vitro transcription.

Single-round in vitro transcription reactions were performed as described elsewhere (38); 0.1 pmol of supercoiled pRJ4069 was used in all reactions unless otherwise noted. Reactions were typically carried out in 0.6 M potassium glutamate–0.1 mM cAMP with 1.6 pmol of Fis protein, 1.1 pmol of CRP, and 0.2 pmol of RNA polymerase (Eς³⁸) in a 50-μl reaction volume. Multiple-round transcription reactions were performed under the same conditions, and transcripts were visualized by primer extension assays as previously described (57). Fold activation was determined by quantification of the P2 transcript on a phosphorimager, normalizing each lane to the amount of rna1 transcript.

Determination of proP transcription in vivo.

For β-galactosidase assays, CRP mutants were transformed into strains RJ7024 (MG1655 ΔlacX74 Δcrp zhe::Tn10 λRJ4065 F′ lacI^qs A4 proAB⁺ fzz::Tn10dcam) and RJ7025 (MG1655 ΔlacX74 Δcrp zhe::Tn10 λRJ4066 F′ lacI^qs A4 proAB⁺ fzz::Tn10dcam). Both λRJ4065 and λRJ4066 are recombinants of λRS45 with plasmids pRJ4065 and pRJ4066, respectively (50, 58). Plasmid pRJ4065 contains a proP-104Δ1–lacZ gene fusion where the P1 promoter is inactivated by a mutation at −12. Plasmid pRJ4066 is identical to pRJ4065 except that it also carries a mutation that greatly reduces CRP binding as depicted in Fig. 1. Each strain was grown overnight, subcultured in Luria broth supplemented with ampicillin, and grown at 37°C for 4 h, where the peak of proP P2 expression occurs for the crp⁺ strain and the AR1 and AR2 mutants. β-Galactosidase activity was measured as described elsewhere (40). Each strain was grown in parallel from at least two separate transformants and assayed in duplicate. Primer extension assays of total mRNA isolated from RJ4446 (crp⁺) and RJ4445 (Δcrp zhe::Tn10) (MG1655 ΔlacX74) were performed as previously described (58).

DNase I footprint assay.

DNase I footprinting assays were performed as previously described (8). A uniquely 5′-end ³²P-labeled fragment from −190 to +103 (with respect to P2 transcriptional initiation) of proP was used; 1.1 pmol of CRP protein and 2.2 nmol of α-CTD were bound to the DNA fragment for 10 min under the same conditions used in the in vitro transcription reaction in a total volume of 20 μl. After binding, DNase I was added for 30 s at room temperature. The reaction was quenched, extracted with phenol-chloroform, and then subjected to ethanol precipitation and electrophoresis on a sequencing gel.

Fis and CRP activate transcription synergistically in vitro at proP P2.

A high-affinity CRP site is centered −121.5 nucleotides from the start of transcription at the proP P2 promoter (33, 58). To test the effect of CRP on proP P2 regulation, single-round in vitro transcription reactions were performed with CRP-cAMP and RNA polymerase complexed with ς³⁸ (Eς³⁸). As shown in Fig. 2, addition of CRP-cAMP had only a small effect, stimulating transcription about twofold over the basal level (no added activator). Fis typically activated transcription 20- to 30-fold. However, when both Fis and CRP-cAMP were added to the in vitro transcription reaction, the level of activated transcription ranged from 75- to 125-fold. This level of activation is greater than the multiplicative effect of each activator alone. Transcripts from each lane were normalized to the levels of rna1 from the vector promoter, which were unaffected by the addition of saturating amounts of CRP-cAMP or Fis. CRP coactivation was dependent on the addition of cAMP, as expected (data not shown). These data show that Fis and CRP activate transcription synergistically at proP P2.

The conditions that promote maximal coactivation by Fis and CRP in vitro were examined. Coactivation was strongly stimulated by a supercoiled template, as was observed with Fis alone (57) (data not shown). Figure 3 shows the effects on in vitro transcription of different concentrations of potassium glutamate in the presence of Fis and CRP, both as solitary activators and in combination. The weak activation by CRP-cAMP alone was constant at different concentrations of potassium glutamate, while the most efficient coactivation by CRP-cAMP plus Fis occurs at 0.6 M. Activation by Fis alone was greatly stimulated by the high concentrations of potassium glutamate, as was observed previously (57). Generally, the conditions which best promote activated transcription by Fis alone are the most favorable for coactivation with CRP.

Coactivation requires the CRP site at −121.5 and Fis site I but not Fis site II.

Fis site II, located at −81, has previously been shown to have little affect on Fis-dependent transcription at proP P2 (57). Since Fis site II is positioned between the essential Fis binding site I at −41 and the CRP binding site at −121.5, it seemed possible that it could influence coactivation with CRP. To test this hypothesis, transcription reactions on the wild-type proP P2 promoter and a template containing a mutation that prevents Fis from binding at site II (56) were compared (Fig. 4). The site II mutant template was able to support efficient synergistic activation with Fis and CRP, indicating that Fis binding to site II is not important for coactivation. As expected, Fis binding to site I was essential for coactivation of proP P2 (Fig. 4). We have shown previously that the site I mutation virtually eliminates binding whereas the site II mutation abolishes binding but creates a weak binding site centered 7 bp upstream of site II (56). The site II mutant was shown to have no effect on Fis activation (57). Likewise, when the CRP site was mutated to prevent binding (58), little coactivation was obtained in the presence of Fis (Fig. 5). Therefore, synergistic activation of proP P2 requires CRP binding at −121.5 and Fis binding at site I but not site II.

Coactivation is dependent on the phasing between CRP and Fis at site I.

To determine if the relative helical orientation between Fis and CRP was important for activation, a 5-bp insertion was created between nucleotides −81 and −82 in the Fis site II binding site which would shift the orientation of CRP by a half helical turn of DNA. As shown in Fig. 5, this change in the helical position of the CRP dimer abolished any contribution by CRP on proP P2 activation. As expected, Fis still activated transcription on this template in both the presence and absence of CRP. Thus, coactivation by Fis and CRP is absolutely dependent on the helical phasing between CRP and Fis bound at site I within the promoter.

CRP coactivates proP P2 in vivo.

Transcriptional activation by CRP in vivo was also examined. Figure 6A shows proP P2 transcripts visualized by primer extension of total mRNA isolated from wild-type and Δcrp cells. A sharp peak of P2 transcript at 3 h after subculture, typical of proP P2 transcription, was seen only in the wild-type cells, suggesting that CRP is important for optimal P2 expression. However, the growth defect inherent to the Δcrp strain raised the possibility that the CRP effect on P2 transcription might be amplified in this assay. For example, Fis and/or RpoS levels may be altered in the Δcrp relative to the crp⁺ cells. Therefore, β-galactosidase assays were also performed using two different proP-lacZ fusions carried on a λ prophage as reporters. Both reporters contain a P1(−12) mutation that abolishes transcriptional initiation from the P1 promoter of proP (39, 56). One construct has an otherwise wild-type proP promoter region, while the other carries a mutation that strongly reduces CRP binding to the promoter in vitro as well as in vivo (Fig. 1) (58). With these constructs, the effect of CRP on proP P2 transcription can be evaluated without using cells lacking CRP. Figure 6B shows that in the presence of wild-type CRP (pZYCRP⁺) and a functional CRP binding site, transcription at proP P2 was stimulated 2.5-fold in vivo when the cells were grown in Luria broth. The CRP site mutant had 126 U of β-galactosidase, while the natural promoter produced 391 U. Up to a sevenfold increase in β-galactosidase activity by CRP binding has been measured in M9 glycerol medium, although the overall level of expression was considerably lower (80 U of β-galactosidase). The stimulation by CRP in vivo was completely dependent on the presence of Fis; cells lacking Fis have <5 U of β-galactosidase (data not shown).

CRP activates proP P2 through AR1.

Two regions on the CRP protein, AR1 and AR2, have been found to be critical for transcriptional activation at other promoters (10). Depending on the position of the CRP binding site, either AR1 or both regions are required for efficient activation. CRP mutants containing substitutions in both AR1 and AR2 were tested for the ability to activate transcription in vivo at proP P2 by measuring β-galactosidase activity generated by the reporter constructs described above. Wild-type and mutant crp genes were supplied on a plasmid in a Δcrp strain. Figure 6B shows that the CRP mutants T158A and H159L (52), which contain mutations in AR1, had similar levels of activity regardless of the presence of the CRP binding site, indicating that they were unable to stimulate P2 transcriptional activation. The transcriptional activation was reduced to 106 and 82 U, respectively, from the 391 U of β-galactosidase produced by the wild-type control. Immunoblotting of cell extracts indicated that the CRP AR1 mutants did not alter the temporal expression of Fis (data not shown). In contrast to the AR1 mutants, CRP containing the substitutions H19Y K101E (47) located in AR2 had no detectable effect on P2 transcription (Fig. 6C). The twofold stimulation of transcription in the presence of a functional CRP binding site was maintained in the AR2-CRP mutant, although overall transcription in both reporters was elevated. The reason for these higher levels of transcription with the AR2 mutant is not known. Thus, AR1, but not AR2, is required for CRP activation of proP P2.

The α-CTD of RNA polymerase binds to the DNA adjacent to CRP.

Due to the remote position of the CRP binding site at proP, CRP is most likely contacting Eς³⁸ through the α-CTD of polymerase, which is attached to the rest of polymerase through a flexible linker (6, 29). As shown above, AR1 of CRP, a region that has been shown to contact α-CTD at other promoters (10), is necessary for CRP activation of P2. To determine if CRP is contacting the α-CTD at proP, we mapped the binding position on the proP promoter of a truncated α subunit containing only the last 85 amino acids comprising the C-terminal domain (20) by DNase I footprinting. The α-CTD peptide was used for these experiments since Eς³⁸ binds poorly to a linearized proP P2 template, consistent with the supercoiling requirement for transcription (57). The α-CTD peptide caused a modest reduction in DNase I cleavages throughout the lanes. However, in the presence of CRP, specific protection of DNase I cleavage was observed immediately adjacent to the downstream side of CRP at concentrations of α-CTD used to detect binding at the rrnB P1 promoter containing a strong UP element (6). This protection extends the CRP protected region at least 9 bp on the top strand (Fig. 7) and 7 bp on the bottom strand, as indicated in Fig. 1. The precise boundaries at the CRP and α-CTD junctions are not possible to determine because of the lack of DNase I reactivity within this A/T-rich segment. Since the α-CTD preferentially binds to A/T-rich regions in the minor groove (18), the binding site for the α-CTD may overlap the highly A/T-rich segment present at the downstream end of the CRP protected region. No CRP-dependent protection by the α-CTD was observed upstream of the CRP site. This footprinting data is consistent with CRP stimulating transcription by the promoter-proximal subunit of CRP directly contacting the α-CTD.

Coactivation by CRP and mutant Fis proteins.

We have previously shown that a small region on one subunit of Fis, the B-C loop, is required for activation of proP P2 by Fis alone. The important amino acids within this region include residues Gln 68, Arg 71, Gly 72, and Gln 74. Arginine 71 is believed to directly contact RNA polymerase because this residue strongly affects cooperative binding with Eς³⁸, and most substitutions at Arg 71 strongly reduce transcriptional activation (38). To determine the effect of this region on coactivation with CRP, in vitro transcription reactions were performed with CRP together with Fis mutants that by themselves are strongly defective in P2 activation. A gel of in vitro transcription reactions showing coactivation by CRP and representative Fis mutants containing substitutions at residue 71 and 72 is shown in Fig. 8. These Fis mutants promoted little to no activated transcription. However, when these mutants were combined with CRP, transcript levels were considerably increased. Coactivation ranged up to 30-fold, depending on the Fis mutant, even though these mutants only weakly potentiated P2 transcription on their own. The full level of activation that is seen with CRP and wild-type Fis is not achieved; however, it is evident that the transcriptionally impaired Fis mutants are still competent for coactivation with CRP.

The order of addition of activators is important for synergistic activation at proP P2.

In the previous transcription reactions, Fis and CRP were allowed to bind to the DNA prior to Eς³⁸ addition. To determine whether this order of addition was important for efficient activation, we incubated one activator plus Eς³⁸ with the plasmid template prior to addition of the second activator. Single-round transcription was then initiated by the addition of nucleoside triphosphates (NTPs) and heparin. In Fig. 9A, Fis plus Eς³⁸ were incubated with the DNA for 15 min followed by the addition of CRP-cAMP. This resulted in a 32-fold activation of transcription, which was somewhat greater than the level for Fis alone but considerably less than the 100-fold activation obtained by preincubating both activators with DNA prior to Eς³⁸ addition. Incubation for a longer period (30 min) prior to elongation did not improve the efficiency of coactivation. In Fig. 9B, CRP-cAMP plus Eς³⁸ were incubated with the template for 15 min before addition of Fis. Under these conditions, the degree of activation was similar to that obtained with Fis alone (20-fold). Again, longer incubations prior to elongation did not improve the efficiency of coactivation. We conclude that efficient coactivation by Fis and CRP requires that both activators be bound to the DNA prior to Eς³⁸.

Coactivation by CRP at proP P2.

To provide evidence that CRP binding at −121.5 directly contacts Eς³⁸ at proP P2, CRP mutants containing substitutions at residues that disrupt polymerase-CRP interactions at either class I or class II promoters were tested. The CRP AR1 mutants were found to strongly reduce activation, while the AR2 mutants had no effect on activation of transcription at proP P2. This is not surprising because CRP positioned upstream of the promoter usually does not mediate transcriptional activation through AR2. These results for CRP activation at proP through AR1 are similar to those found where CRP is acting as an upstream activator alone or in conjunction with either another molecule of CRP, FNR, or the λ cI protein bound in an analogous position to Fis site I at proP P2 (11, 30, 31). However, 121 bp upstream of the transcriptional start site is an unusually large distance for direct activation by CRP.

Because AR1 of CRP has been shown to stimulate transcription at other promoters by contacting the α-CTD of RNA polymerase, it is highly likely that CRP is contacting the α-CTD of Eς³⁸ at proP P2 (17, 27). In support of this, DNase I footprint analysis with purified α-CTD revealed binding of α-CTD to the DNA just downstream of the CRP binding site. This specific association of the α-CTD to the DNA is dependent on CRP binding, suggesting that the CRP protein, together with the sequence of the DNA on the downstream side, is directing the positioning of the α-CTD. Murakami et al. measured the position of both α-CTDs of RNA polymerase at promoters containing two CRP binding sites by affinity cleavage of α-CTD-EDTA · Fe conjugants (41). They found that the α-CTD was unable to contact the DNA when the binding site was at −113.5, even when another CRP dimer was bound at −41.5. Similar results have also been obtained by hydroxyl radical footprinting of CRP-RNA polymerase complexes (5). These findings differ from what we observe at proP P2, where the α-CTD contacts the CRP protein bound even further upstream at −121.5. While the localization of α-CTD by DNase I footprinting performed here is with the α-CTD not tethered to RNA polymerase, in contrast to the previously mentioned experiments, CRP bound at −121.5 nonetheless functions to activate transcription of proP P2 in a manner dependent on AR1. In addition to the sequence of the intervening DNA, an obvious difference between these situations is that at proP the Fis protein is bound at −41 instead of a second CRP molecule. However, it is not expected that Fis would induce a greater degree of DNA bending or a significantly different trajectory of the DNA from CRP bound at the same position (38, 45).

Coactivation by Fis at proP P2.

An essential activation region on Fis has previously been localized to the B-C loop region (38). In transcription reactions performed in vitro with Fis as the sole activator, B-C loop mutants stimulate little, if any, activated transcription. However, together with CRP, these mutants are able to synergistically activate transcription. Therefore, with CRP bound at the promoter, the dependence on the B-C loop of Fis appears to be alleviated. A competent B-C loop region is necessary to achieve the full level of transcription seen with CRP and wild-type Fis, implying that the B-C loop is still playing a role in coactivation. It is possible that CRP may stabilize the polymerase sufficiently to partially overcome a weakened interaction by the Fis B-C loop mutants with the α-CTD. In addition, it is possible, although we consider it unlikely, that DNA bending promoted by the B-C loop Fis mutants is sufficient to stimulate transcription when CRP is present.

An alternative explanation for coactivation between CRP and the Fis B-C loop mutants is that a second determinant other than the B-C loop region on the Fis dimer mediates the residual coactivation with CRP. Without a competent B-C loop region, this potential second transcriptional activation region is unable to stimulate significant transcription when Fis is the solitary activator (38). This postulated second patch has two possible targets. One is CRP protein bound at −121.5, but we have not observed any cooperative binding between Fis and CRP at proP that could support such a model. Another explanation is that a second patch on Fis could contact polymerase in a manner different from the previously discovered contact between the B-C loop of Fis and α-CTD of RNA polymerase. Because Fis binding site I overlaps the −35 sigma subunit recognition element, a Fis-ς³⁸ protein-protein contact is possible. Other transcriptional activators such as CRP, FNR, and the Mu Mor protein that bind to an analogous position have been shown to make multiple contacts with RNA polymerase (2, 4, 44, 53). Most of these activators contact the α-CTD through their upstream side and also make a contact on the downstream side with either the sigma subunit or the N-terminal domain of the alpha subunit. An α-CTD-independent stimulation of rrnB transcription has been noted by Bokal et al. (7). Muskhelishvili et al. have also reported cooperative DNA binding between Fis and ς⁷⁰ at the tyrT promoter (42).

A model of synergistic activation at proP P2.

In our model for synergistic activation of proP P2, Fis minimally contacts one α-CTD domain, and the second α-CTD domain is bound to the CRP subunit oriented proximal to the promoter (Fig. 10). The lack of coactivation by CRP on a DNA template with a 5-bp insertion is consistent with the requirement for looping of the intervening DNA. One notable aspect of the coactivation seen at proP is that it is most efficient when both activators are bound to the DNA prior to Eς³⁸. This observation implies that once a complex is formed between the first activator and Eς³⁸, the second activator is no longer fully capable of stimulating transcription. This could be due to the formation of one of a variety of specific protein-DNA architectures which limit accessibility to the second activator or, in the case of CRP alone, the targeting of Eς³⁸ to a nonproductive location. As mentioned previously, an alternative model is that Fis and CRP bound to their respective sites must first interact. In order to achieve maximal coactivation, this interaction may be required to colocalize the two activators close to the promoter via DNA looping prior to the binding of Eς³⁸.