Promoter Interference in a Bacteriophage Lambda Control Region: Effects of a Range of Interpromoter Distances

Open complex formation at the p_RM promoter was monitored with an electrophoretic mobility shift assay carried out as described by Mita et al. (21). Approximately 1 to 2 nM ³²P-labeled promoter DNA was incubated at 37°C with RNAP (activity, 50% ± 10% [mean ± standard deviation]), at a concentration of active enzyme of 100 nM, in 20 μl of HEPES buffer (30 mM HEPES [pH 7.6], 100 mM KCl, 10 mM MgCl₂, 1 mM dithiothreitol) containing 50 μg of bovine serum albumin per ml. After the addition of 1 μl of a 1-mg/ml solution of heparin to inactivate free RNAP as well as closed complexes and incubation for an additional minute at 37°C, 2 μl of a loading solution (30% glycerol, 0.25% bromophenol blue, 0.25% xylenecyanolphenolfluorine) was added to each reaction mixture prior to loading onto a 4% polyacrylamide gel (29:1 acrylamide-bisacrylamide). The gels were run in TAE buffer (0.04 M Tris-acetate, 0.001 M EDTA) at 6 V/cm for 1.5 h and then exposed to X-ray film to detect the radioactive bands. Two complex bands were observed. On the basis of actual footprinting of the complexes (21), we were able to determine that the faster-moving band represented DNA with an open complex at p_R only and the slower one represented a complex of DNA and RNAP bound in open complexes at both p_R and p_RM (see also Results). Open complexes at both promoters are very stable (reference 28 and our unpublished results); thus, no significant dissociation or redistribution of RNAP is expected to occur during electrophoresis of the complexes.

Full saturation of the p_R promoter occurs before our first time point (taken at 2 min) and probably within seconds (21, 28). Next, the much slower process of open complex formation at p_RM takes place. Our measurements follow the rate of conversion of DNA with one open complex (at p_R) to that with two open complexes (at p_R and p_RM) and thus the rate of open complex formation at p_RM. A comparison of the pseudo-first-order rate constants (k_obs) for the binding of RNAP to the p_RM promoter in the context of the different deletions is graphically shown in Fig. 2a, and the values for k_obs for each promoter deletion mutant are given in Table 1. The D10 construct is seen to be unique in the rate with which p_RMcan form an open complex with RNAP, which was enhanced greater than twofold on this construct. The rate of open complex formation at p_RM was slowest for the D8 construct.

The ability of RNAP to form open complexes at p_RM for each of the constructs was also determined with a single-round runoff transcription assay. Approximately 5 nM promoter was incubated with 50 nM RNAP in HEPES buffer for either 5 or 30 min, followed by a 1-min incubation with heparin (50 μg/ml). To allow RNA synthesis, ATP, CTP, and GTP were added to 200 μM and UTP (including [³²P]UTP) was added to 2 μM. After 10 min, UTP was added to 500 μM and the reaction mixture was incubated for an additional 5 min to ensure complete elongation of all transcripts. Finally, the products were separated on a denaturing gel. Bands apparent after exposure of the gel to Kodak Biomax film were scanned, and the intensities were normalized to the sum of the intensities of the p_RM and p_Rbands.

The amount of runoff product made in this assay is a reflection of the number of open complexes formed during the incubation of RNAP and the promoter, prior to the addition of heparin. The results of these experiments are shown in Fig. 2b. For all constructs, the p_RM promoter was found to be competent to initiate RNA synthesis (results not shown). Relative to the other deletion mutants, again a sharp increase is seen in the amount of RNA synthesized from the p_RM promoter on the D10 template. In this assay, but not the gel mobility shift experiments, the D6 and D7 constructs also show elevated levels of RNA synthesis, albeit not quite as high as that for D10. We do not understand the underlying cause of this difference between the two assays for these two constructs.

The results described above, as well as those from our previous studies (21, 37), indicate that utilization of the p_RM promoter on the construct with the 10-bp deletion was significantly increased in comparison to that on constructs with the wild-type or other spacings between the p_RM and p_R promoters. To better understand the effect of the 10-bp deletion on open complex formation at p_RM, we determined the dependence ofk_obs on RNAP concentration for two promoter mutants, D10 and p_R⁻/p_RM. The rates of open complex formation were determined for each concentration of RNAP as described above. The dependence ofk_obs on the concentration of RNAP is shown in Fig. 3; the data were fit as described in the figure legend to obtain the values of the association constant for RNAP binding to the promoter in a closed complex (K_B) and the first-order rate constant for the conversion of the closed to the open promoter complex (k_f). The values of K_B(7 × 10⁷ ± 3 × 10⁷M⁻¹) and k_f (0.13 ± 0.02 min⁻¹) for the p_R⁻/p_RM construct determined here were similar to those previously reported (16). The fact that mainly k_f is increased when p_R is inactivated (12, 16) indicates that RNAP binding to p_RM is not affected but is rather a subsequent step on the pathway to formation of an open complex. For p_RM on D10, similar values forK_B (6 × 10⁷ ± 3 × 10⁷ M⁻¹) and k_f(0.24 ± 0.04 min⁻¹) are obtained, indicating that on this template the formation of an open complex at p_RM takes place as if the p_R promoter were not occupied. We routinely observe a slightly greater rate of open complex formation at p_RM in the D10 than in the p_R⁻ context (reflected here by a twofold-greater k_f) (see also references 21 and 37). However, since the effect is quite small, we have not attempted to characterize it further.

In Fig. 4 we present a model, refined from Tang et al. (37), that takes our results into account and also draws upon recent insights into the interaction of the α subunit of RNAP with upstream DNA sequences. As first shown for therrnB P1 promoter, the alpha C-terminal domain (α-CTD) binds sequence specifically to an A+T-rich region located between −40 and −60 (the UP element), thereby greatly activating RNA synthesis in vivo and the rate of open complex formation in vitro (30). However, at other promoters, there is also evidence for interactions of the α-CTD with other DNA sequences in upstream regions at similar locations, both in the presence and absence of activator protein (7, 10, 13, 29, 37). The extent of activation that can result from such interactions has not been systematically studied. Based on results with RNAP deleted for the α-CTD, we estimate that at the p_R and p_RM promoters, the interactions with upstream DNA stimulate open complex formation two- to threefold (37). Without the ability of interaction with upstream sequences, RNAP always exhibited a low level of activity at p_RM, even when the p_R promoter was inactivated or the template used bore the D10 template (37). These results provide a strong indication that on the D10 template the RNAP at p_RM was able to engage in upstream interactions even in the presence of another RNAP at p_R.

In the model presented in Fig. 4, on the template with the wild-type spacing between the two promoters, the interference of p_R-bound RNAP with open complex formation at p_RM is exerted via obstruction of interactions between the α-CTD and DNA in the −40 to −60 region of p_RM. This obstruction would be relieved for the D10 construct. Here the 10-bp deletion between the −35 regions of p_RM and p_Rmakes the spacer DNA between the −10 and −35 regions of the p_R promoter coincident with bp −44 to −60 with respect to p_RM. We envisage that the α-CTD of the RNAP at p_RM would be able to reach over the RNAP at p_Rand contact this region. Few if any contacts have been demonstrated between promoter-bound RNAP and the spacer DNA (1, 35), so that the α-CTD of the RNAP at p_RM may well be able to interact with the spacer DNA of p_R, even when both promoters are occupied. The steep dependence of promoter activity and the rate of open complex formation on the interpromoter distance may reflect several factors. For shorter deletions (longer interpromoter distances), less of the spacer DNA but more of the −10 region of p_R is at −40 to −60 with respect to p_RM, leading to obstruction akin to that mentioned above for the wild-type spacer. Conversely, the longer deletions D11 and D12 (with shorter interpromoter distances) would keep the entire spacer DNA within the −40 to −60 region, but steric clashes between the two RNAPs would then become prohibitively severe.

We show that the putative upstream interactions lead to an increase ink_f, which is in agreement not only with the mode of p_RM activation obtained when the p_R promoter is inactivated (12, 16) but also with that observed when p_RM is provided with a genuine UP element (36, 37). Thus, the model is consistent with the available experimental evidence indicating that upstream interactions of the α-CTD facilitate a step subsequent to the initial binding of RNAP to the promoter.

Divergent promoters are fairly common in E. coli as well. In a 1988 review (3), many instances of divergently transcribed promoters in a back-to-back orientation (i.e., directing the synthesis of nonoverlapping transcripts) similar to that of p_R and p_RM of phage lambda were recognized. For our current analysis, we focused on promoter pairs that had start sites separated by 120 pdb or fewer. We chose this distance as an upper limit based on the observed 60-bp upstream extension in DNA interactions at promoters containing upstream elements (30). Thus, it is likely that start site separations beyond this distance will allow unimpeded interactions of RNAP at either promoter. Our search of the database RegulonDB (31) for known E. coli promoters satisfying the above criteria identified 13 promoter pairs, five of which were also represented in the earlier compilation (3) (Table 2). Three cases for which the separation between the start sites is in the range of 71 to 83 pdb investigated here were identified. Interestingly, all three have a separation of 78 pdb, similar to that for the D5 deletion (this work and reference 21), where the interference was found to be rather pronounced. The regulatory significance of a separation by this distance has yet to be investigated.

We thank Alberto Santos for updating the literature references presented in Table 2.

This research was supported by grant GM 31808 from the National Institutes of Health (to P.L.H.). The core facility at Case Western Reserve University (oligonucleotide synthesis) is supported by U.S. Public Health Service grant P30CA43703.