Introduction

Top Abstract Introduction Results Discussion Materials & Methods References

Recent findings in both eukaryotes and prokaryotes indicate that arbitrary protein-protein contacts can trigger gene activation provided one of the protein partners is tethered to the DNA and the other is a component (or is tethered to a component) of RNA polymerase (RNAP) (Barberis et al. 1995; Chatterjee and Struhl 1995; Klages and Strubin 1995; Xiao et al. 1995; Apone et al. 1996; Farrell et al. 1996; Dove et al. 1997; Gaudreau et al. 1997; Gonzalez-Couto et al. 1997; Lee and Struhl 1997; for review, see Ptashne and Gann 1997). Experiments in yeast have shown further that direct fusion of a DNA-binding domain to a component of the RNAP II holoenzyme can activate transcription from a promoter bearing a recognition site for the DNA-binding domain (Barberis et al. 1995; Farrell et al. 1996; Gaudreau et al. 1997; for review, see Ptashne and Gann 1997), but analogous experiments have not been performed previously in bacteria.

RNAP in Escherichia coli consists of an enzymatic core composed of subunits , , and in the stoichiometry ₂, and one of several alternative  factors that confer on the enzyme the ability to recognize specific promoters (Burgess 1976; Hellman and Chamberlin 1988). An additional protein, omega (), has been called a subunit of RNAP on the basis of its copurification with RNAP core and holoenzyme in near stoichiometric amounts (Burgess 1969). The function of  is unknown and, unlike the other subunits,  is not required for transcription either in vitro (Heil and Zillig 1970) or in vivo (Gentry and Burgess 1989; Gentry et al. 1991). Cells deleted for the gene encoding  (rpoZ) have no discernible mutant phenotypes (Gentry and Burgess 1989; Gentry et al. 1991).

Many natural activators in bacteria bind the DNA near the promoters they regulate and interact directly with one or more subunits of RNAP (Busby and Ebright 1994). The best known target of these interactions is the  subunit of RNAP (Ishihama 1992; Russo and Silhavy 1992; Ebright and Busby 1995; Niu et al. 1996). Some activators, however, interact with the  subunit (Hochschild 1994; Kuldell and Hochschild 1994; Li et al. 1994; Artsimovitch et al. 1996; Gerber and Hinton 1996), and evidence suggests that the  subunit may also serve as an activation target in at least one case (Lee and Hoover 1995). Finally the subunit has been identified as the target of action of an activator that functions without binding to the DNA (Miller et al. 1997). In contrast, the  subunit has not been implicated in activation to date.

In a previous study we fused a heterologous protein domain to the  subunit of E. coli RNAP and demonstrated that interaction between a DNA-bound protein and the heterologous protein domain tethered to  activated transcription from a test promoter (Dove et al. 1997). The magnitude of the activation correlated with the strength of the protein-protein interaction, the interaction presumably functioning to stabilize the binding of RNAP to the promoter. These findings suggest that contact between a DNA-bound protein and any subunit of RNAP could, in principle, activate transcription.

Here we show that covalent linkage of a DNA-binding protein to the  subunit can activate transcription from a test promoter bearing a recognition site for the DNA-binding domain, and further that this covalent linkage can be replaced by a protein-protein bridge. These results support the hypothesis that any subunit of RNAP can serve as an activation target, and provide evidence that the  protein is associated with RNAP holoenzyme in vivo and that it is accessible at the surface of the enzyme complex.

	Results

Top Abstract Introduction Results Discussion Materials & Methods References

The  subunit can activate transcription from a test promoter when fused to a DNA-binding protein

To determine whether the  subunit of RNAP can mediate transcriptional activation when tethered to the DNA upstream of a promoter, we fused the  protein to the repressor (cI) protein of bacteriophage  (see Fig. 1B). The cI protein is a two-domain protein that binds DNA as a dimer; the amino-terminal domain (NTD) is the DNA-binding domain, whereas the carboxy-terminal domain (CTD) mediates dimer formation (and higher order oligomerization) (Sauer et al. 1990).

View larger version (24K): [in this window] [in a new window]   Figure 1.     (A) Basal transcription by E. coli RNA polymerase. A promoter comprised of a 10 and a 35 element is depicted bound by RNAP with subunit composition 2. The  subunit (shaded) is depicted as interacting with the subunit as a monomer (see Gentry and Burgess 1993). (B) Fusion of the  subunit of RNAP to the carboxy terminus of cI permits interaction of a modified polymerase with a  operator. The artificial promoter derivative plac OR2-62 is shown; it bears the  operator OR2 centered 62 bp upstream of the transcriptional start site of the lac promoter.

We fused the entire  protein (residues 1-90) to the carboxyl terminus of the cI protein through a small alanine linker (see Materials and Methods). We placed the gene encoding this fusion protein downstream of an inducible promoter on a plasmid vector, thus generating plasmid pBRcI-. We introduced pBRcI- into strain KS1Z, which harbors on its chromosome a lac promoter derivative (termed placO_R2-62) bearing a single  operator centered 62 bp upstream of the transcription start point. Note that cI, which activates transcription from the P_RM promoter when bound at a site centered at position 42, cannot activate transcription from placO_R2-62 (Dove et al. 1997) because the  operator is positioned too far from the promoter. In addition, KS1Z bears a deletion of the chromosomal locus encoding the  subunit. Unlike cI, the cI- fusion protein stimulated transcription ~70-fold, as measured by -galactosidase assay (Fig. 2A). Primer extension analysis confirmed that the fusion protein stimulated the production of correctly initiated transcripts (Fig. 2B). A similar experiment performed with strain KS1 revealed that the cI- fusion protein was unable to stimulate transcription from placO_R2-62 in the presence of endogenous  protein encoded by the chromosomal rpoZ gene (Fig. 2A).

View larger version (19K): [in this window] [in a new window]   Figure 2.     Transcriptional activation by cI- fusion protein. (A) Effect of cI- on transcription in vivo from plac OR2-62. KS1Z cells harboring plasmids encoding either cI- (), cI(S45A)- (), cI () or no cI (cI) (), and KS1 cells (containing wild-type ) () harboring a plasmid encoding cI-, were grown in the presence of the indicated concentrations of IPTG and assayed for -galactosidase activity. (B) Primer extension analysis of transcripts produced from placOR2-62 in the presence of cI-. Total RNA was isolated from KS1Z cells grown in the presence of 20 µM IPTG and harboring plasmids encoding the indicated proteins. Primer extension analysis was done by using a primer complementary to the lacZ transcript produced by the placOR2-62 promoter. Primer extension products produced by correctly initiated plac OR2-62 transcripts are indicated by +1 and excess unincorporated primer is shown.

Transcriptional activation by the cI- fusion protein is dependent on its ability to bind DNA

To demonstrate that the stimulation of transcription from placO_R2-62 in KS1Z by cI- depends on the ability of the fusion protein to bind to the  operator of placO_R2-62, we introduced a single amino acid substitution (S45A) into the cI moiety of the fusion protein that results in a severe reduction in operator binding (Hochschild and Ptashne 1986). This mutant version of the cI- fusion protein failed to stimulate transcription from placO_R2-62 (Fig. 2A).

We confirmed that the cI(S45A)- fusion protein is specifically defective for operator binding by measuring its binding to a consensus  operator using an in vivo repression assay (data not shown). Western blot analysis confirmed that the cI- and the cI (S45A)- fusion proteins were present in the cell in comparable amounts (data not shown).

Interaction between a DNA-bound domain of Gal4 and a domain of Gal11^P fused to either the  or  subunit of RNAP results in transcriptional activation from a test promoter

We then replaced the covalent interaction between the cI protein and the  subunit of RNAP with a protein-protein contact. For this purpose we took advantage of a pair of protein domains originally shown to interact in yeast cells. Transcriptional activation in yeast can be triggered by an apparently fortuitous interaction between the dimerization region of the yeast transcriptional activator Gal4 and a mutant form of the Gal11 protein (Himmelfarb et al. 1990; Barberis et al. 1995), which despite its name, is a component of the RNAP II holoenzyme and is required for full transcription of many genes (Kim et al. 1994; Barberis et al. 1995; Hengartner et al. 1995). Ordinarily, in yeast, the dimerization region of Gal4 does not mediate transcriptional activation when connected to its own or another DNA-binding domain. However, in the presence of a Gal11 mutant (called Gal11^P for potentiator) bearing a single amino acid substitution at position 342, the Gal4 dimerization region functions as a powerful activating region; this activation results from a specific interaction between the Gal4 dimerization region and the portion of Gal11^P bearing the amino acid substitution (Barberis et al. 1995; Farrell et al. 1996).

To establish that this protein-protein interaction can also trigger gene activation in E. coli, we first tested the abilities of the relevant portions of Gal11^P and Gal4 (Farrell et al. 1996) to mediate transcriptional activation when fused to the  subunit of RNAP and to the cI protein, respectively. To do this, we proceeded as we had done previously (Dove et al. 1997), taking advantage of the domain structure of , which initiates the assembly of RNAP by forming a dimer. The -NTD is responsible for the assembly reaction (Hayward et al. 1991; Igarashi et al. 1991), and the -CTD, which is connected to the -NTD by a flexible linker region (Blatter et al. 1994; Jeon et al. 1997), can bind DNA (Ross et al. 1993; Blatter et al. 1994) and is the natural target for many transcriptional activators (Ishihama 1992; Ebright and Busby 1995). We reasoned that if we replaced the -CTD with an appropriate domain of Gal11^P, the resulting -Gal11^P chimera would display a target that could be contacted by an appropriately positioned cI-Gal4 dimer (Fig. 3). Therefore, we created two chimeric genes, one encoding the -NTD and linker connected to residues 263-352 of Gal11^P, and the other encoding full-length cI (residues 1-236) connected to the dimerization region of Gal4 (residues 58-97) (see Materials and Methods). We then tested the ability of the cI-Gal4 fusion protein to activate transcription from placO_R2-62 in cells containing the -Gal11^P fusion protein (as well as wild type  encoded by the chromosomal rpoA gene). Figure 4A shows that the cI-Gal4 fusion protein activated transcription ~45-fold in KS1 cells containing the -Gal11^P fusion protein but not in control cells containing an otherwise identical fusion protein bearing the wild type form of Gal11. Primer extension analysis confirmed that the fusion protein stimulated the production of correctly initiated transcripts (Fig. 4B). We also found that a different cI-Gal4 fusion protein comprising only the NTD and linker region of cI (residues 1-132) fused to Gal4(58-97) activated transcription in KS1 cells harboring the -Gal11^P fusion protein by a factor of approximately eight (data not shown).

View larger version (22K): [in this window] [in a new window]   Figure 3.     (A) Basal transcription by E. coli RNA polymerase drawn to illustrate the structure of the  subunit. A promoter comprised of a 10 and a 35 element is depicted together with the subunit composition of RNAP holoenzyme. The  subunits of RNAP consist of two independently folded domains; an NTD and a CTD connected by a flexible linker. (B) Replacement of RNAP -CTD by Gal11P (residues 263-352) permits interaction with the Gal4 dimerization domain of a cI-Gal4 fusion protein. The diagram depicts the test promoter placOR2-62, which bears the  operator OR2 centered 62 bps upstream of the transcriptional start site of the lac promoter.

View larger version (17K): [in this window] [in a new window]   Figure 4.   Transcriptional activation by cI-Gal4 fusion protein in presence of -Gal11P fusion protein. (A) Effect of cI-Gal4 on transcription in vivo from plac OR2-62 in the presence of -Gal11P fusion protein. KS1Z cells harboring plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for -galactosidase activity. () cI-Gal4 + -Gal11P; () cI-Gal4 + -Gal11WT; () cI + -Gal11P; () cI + -Gal11WT. (B) Primer extension analysis of transcripts produced from plac OR2-62 in the presence of cI-Gal4 and -Gal11P. Total RNA was isolated from KS1Z cells grown in the presence of 20 µM IPTG and harboring plasmids encoding the indicated proteins. Primer extension analysis was done by using a primer complementary to the lacZ transcript produced by the plac OR2-62 promoter. Primer extension products produced by correctly initiated plac OR2-62 transcripts are indicated by +1 and excess unincorporated primer is shown.

We reasoned that if we fused the appropriate domain of Gal11^P to the  subunit of RNAP the resulting -Gal11^P chimera would, like the -Gal11^P chimera, display a target that could be contacted by an appropriately positioned cI-Gal4 dimer (Fig. 5A). Therefore, we constructed a chimeric gene encoding the  subunit connected to residues 263-352 of Gal11^P (see Materials and Methods). Figure 5B shows that the cI-Gal4 fusion protein (comprising residues 1-236 of cI fused to residues 58-97 of Gal4) activated transcription ~20-fold in KS1Z cells containing the -Gal11^P fusion protein, but not in control cells containing the -Gal11^WT fusion protein. Primer extension analysis confirmed that the fusion protein stimulated the production of correctly initiated transcripts (Fig. 5C). In contrast to the cI- fusion protein, the cI-Gal4 fusion protein activated transcription in KS1 cells containing both wild-type  and the -Gal11^P chimera (data not shown), indicating that in this case endogenous  does not compete effectively with the -Gal11^P chimera for binding to RNAP (see Discussion).

View larger version (26K): [in this window] [in a new window]   Figure 5.     Transcriptional activation by cI-Gal4 fusion protein in presence of -Gal11P fusion protein. (A) Fusion of Gal11P (residues 263-352) to the  subunit of RNA polymerase permits interaction with the Gal4 dimerization domain of a cI-Gal4 fusion protein. The diagram depicts the test promoter plac OR2-62, which bears the  operator OR2 centered 62 bp upstream of the transcriptional start site of the lac promoter. (B) Effect of cI-Gal4 on transcription in vivo from plac OR2-62 in the presence of -Gal11P fusion protein. KS1Z cells harboring plasmids expressing the indicated proteins were grown in the presence of different concentrations of IPTG and assayed for -galactosidase activity. () cI-Gal4 + -Gal11P; () cI-Gal4 + -Gal11WT; () cI + -Gal11P; () cI + -Gal11WT. (C) Primer extension analysis of transcripts produced from plac OR2-62 in the presence of cI-Gal4 and -Gal11P. Total RNA was isolated from KS1Z cells grown in the presence of 50 µM IPTG and harboring plasmids encoding the indicated proteins. Primer extension analysis was done by using a primer complementary to the lacZ transcript produced by the plac OR2-62 promoter. Primer extension products produced by correctly initiated plac OR2-62 transcripts are indicated by +1 and excess unincorporated primer is shown.

	Discussion

Top Abstract Introduction Results Discussion Materials & Methods References

Transcriptional activation in E. coli by tethering a subunit of RNAP to DNA

We have demonstrated that covalent linkage of a DNA-binding protein (cI) to a component of RNAP () results in transcriptional activation from a promoter bearing a recognition site for the DNA-binding protein (a  operator). This form of activation resembles the natural activation that occurs when the RNAP  subunit interacts with a DNA sequence element termed the UP element that is found upstream of the 35 hexamer of rRNA promoters (as well as some other promoters) (Ross et al. 1993; Gaal et al. 1996). As mentioned above, the  subunit has two domains, and its CTD is a sequence-specific DNA-binding domain (Blatter et al. 1994; Gaal et al. 1996). The interaction of the -CTD with naturally occurring UP elements has been reported to increase transcription by as much as 30-fold, and this increase reflects an increase in the initial binding of RNAP to the promoter and possibly a subsequent step in the initiation process (Ross et al. 1993; Rao et al. 1994).

Experiments performed in eukaryotic cells have also shown that transcription can be activated by the direct fusion of DNA-binding domains to various components of the transcriptional machinery. For example, fusion of the E. coli LexA repressor, a sequence-specific DNA-binding protein, to the wild-type Gal11 protein creates a powerful transcriptional activator in yeast that works on promoters bearing LexA-binding sites, and this activation depends on the portion of Gal11 that mediates its association with the RNAP II holoenzyme (Himmelfarb et al. 1990; Barberis et al. 1995; see also Farrell et al. 1996; Gaudreau et al. 1997). Similarly, direct or indirect fusion of a DNA-binding domain to the yeast TATA-binding protein (TBP) creates a transcriptional activator that works on promoters bearing a recognition site for the DNA-binding domain upstream of the TATA element, and this activation depends on the ability of TBP to interact with the TATA element (Chatterjee and Struhl 1995; Klages and Strubin 1995; Xiao et al. 1995). Thus, recruitment of the polymerase II holoenzyme and TBP to promoters in yeast cells, as well as RNAP holoenzyme to bacterial promoters can be a rate-limiting step in transcription initiation in vivo.

Transcriptional activation by arbitrary protein-protein interactions

In a previous study we showed that contact between a DNA-bound protein and a heterologous protein domain fused to the -NTD can activate transcription in E. coli (Dove et al. 1997). We have now generalized this finding by showing that an interacting pair of protein fragments that triggers gene activation in yeast also triggers gene activation in E. coli when one of the pair is fused to a DNA-binding protein and the other is fused either to the -NTD or to the  protein. Specifically, we fused the dimerization domain of Gal4 to the cI protein and the relevant fragment of the Gal11^P protein either to the -NTD or to  and demonstrated that the cI-Gal4 fusion protein stimulated transcription from an appropriately designed test promoter in cells containing either the -Gal11^P or the -Gal11^P fusion protein. These findings provide support for the hypothesis that contact between a DNA-bound protein and any component of RNAP can activate transcription in E. coli (Fig. 6). In particular, they indicate that the  subunit is not unique in its ability to mediate the effects of artificial activators.

View larger version (14K): [in this window] [in a new window]   Figure 6.     Any subunit of RNAP can serve as an activation target. An activator bound upstream of a promoter is depicted, and the possible interactions with any of the subunits of RNAP are illustrated with arrows. The interactions mediated by the artificial activators described herein are indicated by solid arrows; additional interactions are indicated by broken arrows. The  and  subunits are known targets of natural DNA-bound activators (for review, see Busby and Ebright 1994), the  subunit is suggested as a target for at least one DNA-bound activator (Lee and Hoover 1995), and the subunit is the target of an activator that works without binding DNA (Miller et al. 1997).

We note that the cI-Gal4 fusion protein stimulated transcription more strongly with the -Gal11^P than with the -Gal11^P fusion protein (45- vs. 20-fold). We do not know the reason for this difference, but we speculate that it may reflect a difference in the number of Gal11^P moieties displayed by the RNAP holoenzyme in the two cases. Whereas the  subunit is presumed to associate with RNAP as a monomer, the  subunit assembles as a dimer. Therefore, at least a fraction of the polymerase molecules assembled in the presence of chromosomally encoded wild-type  and an excess of the -Gal11^P fusion protein should display two Gal11^P moieties, each of which might be able to interact with a Gal4 moiety displayed on the DNA-bound cI dimer. It is also possible that the observed difference in activities (45-fold for the -Gal11^P fusion protein and 20-fold for the -Gal11^P fusion protein) reflects a difference in the fraction of RNAP molecules in the cell that contain the fusion protein. Further experiments will be required to test these possibilities.

The  protein is a subunit of RNA polymerase in vivo

Our demonstration that the -Gal11^P fusion protein can function as an activation target provides strong evidence that the protein is associated with RNAP in growing cells. This demonstration taken together with the finding that the cI- fusion protein is a powerful activator of transcription indicates, moreover, that  is accessible at the surface of RNAP. Therefore, our results raise the possibility that  might serve as a target for natural DNA-bound activators, as well.

In our experiments with the cI- fusion protein (in which the amino terminus of  is fused to the carboxyl terminus of cI), we observed transcriptional activation only in a strain deleted for the chromosomal rpoZ gene (encoding ). This suggests first that the cI- fusion protein associates with the same surface of RNAP as native  protein, and second, that the native  protein associates preferentially. In contrast, the -Gal11^P fusion protein (in which the amino terminus of  is free) mediated transcriptional activation by the cI-Gal4 fusion protein in both the absence and presence of chromosomally encoded  protein. This suggests that this fusion protein competes effectively with native  protein for association with RNAP. We suggest a possible explanation: The amino-terminal portion of the  protein mediates its association with RNAP with the consequence that fusion of another protein at the amino terminus weakens the association.

In vitro cross-linking experiments have suggested that  binds to the subunit of RNAP (Gentry and Burgess 1993). The biological activity of our cI- fusion protein should permit genetic identification of not only the residues on  that mediate its association with RNAP, but the interacting residues on (or any other subunit of RNAP) as well.

Practical implications for prokaryotic two-hybrid and one-hybrid systems

Our previous demonstration that contact between a DNA-bound protein and a protein domain fused to the  subunit of RNAP can activate transcription suggested the possibility of establishing a transcription-based two-hybrid assay for detecting protein-protein interactions in E. coli (Dove et al. 1997). Here we have demonstrated the feasibility of this approach by showing that two polypeptides known to interact in yeast and in vitro (Farrell et al. 1996) can activate transcription in E. coli when one of the pair is fused to a subunit of RNAP (either  or ) and the other is fused to a DNA-binding protein (cI).

Previous studies have demonstrated that heterologous protein domains that mediate dimer formation can functionally substitute for the cI-CTD when fused to the cI-NTD, resulting in biologically active fusion proteins that bind efficiently to  operators (for review, see Hu 1995). In designing the cI-Gal4 fusion protein, we sought to compare the effects of fusing the Gal4 moiety to the end of the cI linker (at residue 132) and to the end of full-length cI (at residue 236). Although both of the resulting fusion proteins bound to  operators and stimulated transcription in the presence of the -Gal11^P fusion protein (Fig. 4; data not shown), the full-length cI-Gal4 fusion protein was more active, presumably because the cI-CTD mediates more efficient dimerization than the Gal4 moiety. The finding that a heterologous protein domain can be fused to the carboxyl terminus of cI without interfering with cI-CTD-mediated dimerization implies that heterologous protein domains can be tethered to the DNA through the cI protein regardless of whether or not they have the potential to dimerize.

The utility of this strategy was confirmed by our construction of a biologically active cI- fusion protein. In turn, the ability of this fusion protein to activate transcription from a promoter bearing a  operator demonstrates the feasibility of establishing a so-called "one-hybrid" assay in E. coli to detect specific protein-DNA interactions (Li and Herskowitz 1993; Wang and Reed 1993; Inouye et al. 1994).

Mechanistic implications

Together our findings with both the  and the  fusion proteins suggest that depending on the nature of the promoter, contact between a DNA-bound protein and any accessible surface of RNAP can activate transcription, presumably by stabilizing the binding of RNAP to the promoter. We suspect, however, that this form of activation would not work at all promoters. In particular, if the activity of a given promoter is not limited by its ability to stably bind RNAP in vivo (for example, see Morett and Buck 1989; Hidalgo and Demple 1997; for reviews, see also Kustu et al. 1991; Summers 1992; Gralla 1996; Ptashne and Gann 1997), then we hypothesize that the artificial activators we have designed would be ineffective. In this regard, it might be possible to use the cI- fusion protein as a tool for classifying promoters in vivo.

	Materials and methods

Top Abstract Introduction Results Discussion Materials & Methods References

Media, growth conditions, and genetic techniques

Bacteria were cultured routinely in L broth or on L-agar plates (Miller 1972). Where needed, antibiotics were used at the following concentrations: carbenicillin (50 µg/ml), chloramphenicol (25 µg/ml), kanamycin (50 µg/ml), and tetracycline (20 µg/ml). Transductions were performed using generalized transducing phage P1vir as described previously (Sternberg and Maurer 1991).

Bacterial strains

E. coli strain XL1-blue (Stratagene) was used routinely as a cloning vehicle for plasmid constructions. The E. coli strain KS1 has been described previously (Dove et al. 1997) and harbors the artificial promoter derivative placO_R2-62 consisting of the  operator O_R2 centered 62 bp upstream of the transcriptional start site of the lac promoter. This promoter and the linked lacZ gene are present on a imm21 prophage. The E. coli strain KS1Z was created by P1-mediated transduction of the spoS3::cat mutation (an  null allele) from CF2790 (Xiao et al. 1991) into recipient strain KS1.

DNA manipulation and oligonucleotides

Standard molecular biology techniques (Sambrook et al. 1989) were used for cloning, DNA purification, and analysis. The PCR was performed using Expand (Boehringer Mannheim) and restriction enzymes were obtained from New England Biolabs. DNA was sequenced by the dideoxy method using Sequenase (U.S. Biochemical).

Oligonucleotides used to make the different plasmids were purchased from Operon Technologies, Inc., and were as follows: OL.2 (5-CAGTGATTCTGCATTCTGGCTTGAG-3); OL.3 (5-GCGGATCCTAGGTCAAAATAATCCTGTTAA-3); OL.6 (5-CAGACGTTTGGCGAATCAAGGCTAGAAAGACTGG-3); OL.7 (5-TAGCCTTGATTCGCCAAACGTCTCTTCAGG-3); OL.13 (5-CTGCTGTTGAGGCTCTGGTTTCTCTTCTTTCAC-3); OL.15 (5-GAGAAACCAGAGCCTCAACAGCAGCAAATGCAACC-3); OL.18 (5-AGCGGATCCTCACAAAGCTTGGATTTTTCTCAGG-3); OL.R1 (5-GTGCCGGTTCTACCC-3); OL.32 (5-TAGGATCCGGCGCGCCTAAGATCTTGCGGCCGCGCCAAACGTCTCTTCAGGCCACTG-3); OL.39 (5-ATATGCGGCCGCACGCGTAACTGTTCAGGACG-3); OL.40 (5-ATATGTCGACTTAACGACGACCTTCAGCAAT-3); OL.41 (5-AAAGTTCCATATGGCACGCGTAACTGTTCAGG-3); OL.42 (5-TATATGCGGCCGCACGACGACCTTCAGCAATAGCG-3); OL.43 (5-ATATGCGGCCGCACCTCAACAGCAGCAAATGCAACC-3); OL.44 (5-ATATGTCGACTCACAAAGCTTGGATTTTTCTCAGG-3); OL.54 (5-ATATATCATATGAGCACAAAAAAGAAACC-3); OL.55 (5-TTCTCTGGCGATTGAAGGGC-3).

Plasmids

Plasmids used in this study are listed in Table 1. All inserts in plasmids that were generated by the PCR were subsequently sequenced to confirm that no errors had been introduced as a result of the PCR process.

Plasmid pLX20 is a derivative of pBR322 (Bolivar et al. 1977), confers Ap^R, and bears the cI gene under the control of the lacUV5 promoter (F. Whipple, unpubl.).

Plasmid pBRcI- is a derivative of pLX20, confers Ap^R, and encodes residues 1-236 of cI fused to two Ala residues, which in turn are fused to residues 1-90 of the  subunit of E. coli RNAP. The expression of the cI- fusion protein is under the control of the lacUV5 promoter. The primary sequence of the cI- fusion protein junction from cI residue 235 inclusive is PheGlyAlaAlaAlaArg, where the underlined residues are the first and second, respectively, in the primary sequence of . Note that the initiating Met of  was not included in the fusion protein, as it is not present in the mature protein (Gentry and Burgess 1986), and therefore, the Ala residue that follows the initiating Met is classified here as residue one. pBRcI- was constructed by replacing the HindIII-SalI fragment from pLX20 with two fragments of DNA. One fragment (comprising a segment of the cI gene) was a HindIII-NotI-digested PCR product that was made using primers OL.2 and OL.32 with pACcI as template. This fragment contains a NotI site at the 3 end of cI. The second fragment (comprising the  gene) was a NotI-SalI-digested PCR product that was made using primers OL.39 and OL.40 with pEC-1 as template.

Plasmid pBRcI(S45A)- is identical to pBRcI- except that the cI moiety of the fusion protein harbors the S45A mutation. pBRcI(S45A)- was constructed by replacing the NdeI-NsiI fragment from pBRcI- with an NdeI-NsiI-digested PCR product that was made using primers OL.54 and OL.55 with pLCF3 as template. pBRcI is the same as pLR1cI (Whipple et al. 1994).

Plasmid pACLGF2 is a derivative of pACcI, confers Cml^R, and encodes residues 1-236 of cI fused to residues 58-97 of Gal4 under the control of the lacUV5 promoter. pACLGF2 was made by replacing the HindIII-BstYI fragment from pACcI with a HindIII-BamHI digested PCR product made using primers OL.2 and OL.3. The PCR product comprised a fragment of the 3 end of the cI gene fused directly to the coding sequence of Gal4 (residues 58-97), and two PCR products (made using primers OL.2 and OL.7 with pACcI as template and primers OL.6 and OL.3 with pNS113 as template, respectively) served as template for its generation.

Plasmid pACTcLGF2 is a derivative of pACLGF2 that confers Tc^R and like pACLGF2 encodes residues 1-236 of cI fused to residues 58-97 of Gal4 under the control of the lacUV5 promoter. pACTcLGF2 was made by replacing the HindIII-EcoRI fragment from pACcI with the EcoRI-BstYI fragment from pBR322 (encoding the Tc^R gene) and the appropriate BstYI-HindIII fragment from pACLGF2.

Plasmid pBR-Gal11^WT is a derivative of pBR, confers Ap^R, and encodes residues 1-248 of the  subunit of E. coli RNAP fused to residues 263-352 of wild-type Gal11 under the control of tandem lpp and lacUV5 promoters. pBR-Gal11^WT was made by replacing the EcoRI-BamHI fragment from pBR with an EcoRI-BamHI-digested PCR product made using primers OL.R1 and OL.18. The PCR product comprised a fragment of the 3 end of the cI gene fused directly to the coding sequence of Gal11 (residues 263-352), and two PCR products (made using primers OL.R1 and OL.13 with pBR as template and primers OL.15 and OL.18 with pSO23 as template, respectively) served as template for its generation. pBR-Gal11^P was similarly made using pSO32 instead of pSO23 as template.

Plasmid pBR-Gal11^WT confers Ap^R and encodes residues 1-90 of the  subunit of E. coli RNAP fused to three Ala residues, which in turn are fused to residues 263-352 of wild-type Gal11. The expression of the -Gal11^WT fusion protein is under the control of the lacUV5 promoter. pBR-Gal11^WT was made by replacing the NdeI-SalI fragment from pLX20 with two fragments of DNA. One fragment (comprising a segment of the  gene) was an NdeI-NotI-digested PCR product that was made using primers OL.41 and OL.42 with pE3C-1 as template. This introduces an NdeI site at the start and a NotI site at the 3 end of . The second fragment (comprising the Gal11^WT coding sequence from residues 263-352) was a NotI-SalI-digested PCR product that was made using primers OL.43 and OL.44 with pSO23 as template. pBR-Gal11^P was similarly made using pSO32 instead of pSO23 as template.

-Galactosidase assays

SDS-CHCl₃ permeabilized cells were assayed for -galactosidase activity essentially as described (Miller 1972). Assays were performed at least three times in duplicate on separate occasions, with similar results. Values are the averages from one experiment and duplicate measurements differed by <10%.

Primer extension analysis

RNA isolation, primer labeling, primer extension assays, and transcriptional start site identification were as described previously (Dove et al. 1997).