Substitutions in the Escherichia coli RNA polymerase inhibitor T7 Gp2 that allow inhibition of transcription when the primary interaction interface between Gp2 and RNA polymerase becomes compromised

The Escherichia coli-infecting bacteriophage T7 encodes a 7 kDa protein, called Gp2, which is a potent inhibitor of the host RNA polymerase (RNAp). Gp2 is essential for T7 phage development. The interaction site for Gp2 on the E. coli RNAp is the β′ jaw domain, which is part of the DNA binding channel. The binding of Gp2 to the β′ jaw antagonizes several steps associated with interactions between the RNAp and promoter DNA, leading to inhibition of transcription at the open promoter complex formation step. In the structure of the complex formed between Gp2 and a fragment of the β′ jaw, amino acid residues in the β3 strand of Gp2 contribute to the primary interaction interface with the β′ jaw. The 7009 E. coli strain is resistant to T7 because it carries a charge reversal point mutation in the β′ jaw that prevents Gp2 binding. However, a T7 phage encoding a mutant form of Gp2, called Gp2β, which carries triple amino acid substitutions E24K, F27Y and R56C, can productively infect this strain. By studying the molecular basis of inhibition of RNAp from the 7009 strain by Gp2β, we provide several lines of evidence that the E24K and F27Y substitutions facilitate an interaction with RNAp when the primary interaction interface with the β′ jaw is compromised. The proposed additional interaction interface between RNAp and Gp2 may contribute to the multipronged mechanism of transcription inhibition by Gp2.


INTRODUCTION
Central to the regulation of bacterial transcription is the RNA polymerase (RNAp), the enzyme responsible for the synthesis of all RNA in bacteria. The bacterial RNAp is composed of a multisubunit catalytic core (subunit composition a 2 bb9v; abbreviated E) and a sigma (s) factor subunit that determines the specificity of the RNAppromoter interaction. In Escherichia coli, the s 70 -containing RNAp (Es 70 ) is responsible for the transcription of most genes during exponential growth (Haugen et al., 2008). Transcription by Es 70 begins with the formation of the initial Es 70 -promoter complex known as the closed promoter complex (RPc). The RPc is often very unstable and isomerizes, via several intermediates, to the transcriptionally proficient open promoter complex (RPo). In the RPo,~15 base pairs of promoter DNA around the transcription start site are melted to form the 'transcription bubble' and the template DNA strand is positioned at the active centre of the RNAp. The catalytic b and b9 subunits of the RNAp define the main DNA binding channel (DBC), and the active centre where RNA synthesis takes place is located deep within the DBC. The double-stranded DNA immediately downstream of the active centre (dwDNA) interacts with a segment of the DBC called the downstream DBC (dwDBC), and this interaction is essential for the formation and stability of the RPo (Saecker et al., 2011).
The RPo formation also represents an important regulatory step during transcription initiation, and thus the activity of the bacterial RNAp is controlled by an array of transcription regulatory factors, which repress, stimulate or modulate its ability to form the RPo (Browning & Busby, 2004). Most bacteriophages (phages) rely on the RNAp of their host bacteria for expression of their genes during the infection process. Therefore, RPo formation by the bacterial RNAp is also subjected to regulation by phageencoded proteins through covalent RNAp modifications and modulation by RNAp-binding proteins (Nechaev & Severinov, 2003). T7 phage encodes a 7 kDa protein, called Gp2, which binds to a domain in the b9 subunit, called the 'jaw' domain, and potently inhibits RPo formation by the E. coli RNAp (Nechaev & Severinov, 1999). The b9 jaw domain is an integral component of the dwDBC and plays a major role during RPo formation and stable maintenance of the RPo (Ederth et al., 2002(Ederth et al., , 2006. The biological role of T7 Gp2 is to ensure efficient and coordinated transcription of phage genes by the T7 RNAp without interference from bacterial RNAp (Savalia et al., 2010). In the solution structure of the complex of Gp2 with a fragment of the b9 jaw domain (E. coli RNAp b9 residues 1153-1213), two invariant arginine residues (R56 and R58; R56 and R58 are invariant in 25 and 23, respectively, out of 25 Gp2-like proteins in the EBI database at the time of writing) located in the b3 strand of Gp2 are in close proximity to E1158 and E1188 in the b9 jaw, and thereby contribute to significant ionic interactions across the interface (Fig. 1a, b) (James et al., 2012). Alanine or charge reversal substitutions at either R56 or R58 prevent Gp2 from binding to the RNAp (Cámara et al., 2010). Conversely, RNAp containing a lysine substitution at either E1158 (BR3 E. coli) or E1188 (7009 E. coli) is resistant to inhibition by Gp2 (Nechaev & Severinov, 1999). Thus, the primary interaction interface between Gp2 and the RNAp involves R56 and R58 in the b3 strand of Gp2 and E1158 and E1188 in the b9 jaw domain of the RNAp.
Wild-type T7 is unable to productively infect E. coli 7009 (Chamberlin, 1974;Nechaev & Severinov, 1999;Schmitt et al., 1987). A mutant phage, called T7b, that successfully infects the 7009 E. coli strain (which has a mutant RNAp containing the E1188K substitution in the b9 jaw domain), encodes a triple mutant form of Gp2 (harbouring substitutions E24K, F27Y and R56C; hereafter called Gp2 b ) (Chamberlin, 1974;Schmitt et al., 1987). The molecular basis by which E24K, F27Y and R56C substitutions in Gp2 suppress the E1188K mutation in the b9 jaw domain, which prevents wild-type Gp2 from binding to the RNAp, is not known. In the context of the Gp2-b9 jaw domain fragment structure, the E24K and F27Y substitutions are located in the middle and close to the end, respectively, of the loop connecting the b1 and b2 strands in Gp2, i.e. at the opposite side to the b3 strand, which forms the interface with the b9 jaw domain and carries R56 (Fig. 1a). In this study, we established a simple in vivo bacterial growth attenuation assay to assess the activity of recombinant Gp2 on E. coli RNAp in the absence of T7 infection, to study the molecular basis by which Gp2 b inhibits wild-type, 7009 and BR3 E. coli RNAp. The results strongly suggest that the E24K and F27Y mutations facilitate the interaction between Gp2 b and RNAp when the primary interaction interface between Gp2 and RNAp is compromised. Experiments addressing the loop interconnecting b1 and b2 strands in Gp2 homologues from two T7-like phages that infect Citrobacter rodentium further corroborate our results.

METHODS
Bacterial strains and plasmids. Table S1 (available with the online version of this paper) lists the bacterial strains and plasmids used in this study. Plasmid pCS1 : Gp2 and derivatives were constructed by digesting pSW33 : Gp2 (and derivatives) with XbaI and HindIII, and the 0.3 kb fragment containing gene 2 was cloned into the same sites in pBAD18 (Invitrogen). The coding sequences of Gp2 homologues CR44b protein 13 (CR44b_13) and CR8 protein 16 (CR8_16) were PCR-amplified from phage genomic DNA (kindly provided by Ana Louisa Toribio and Gordan Dougan, The Wellcome Trust Sanger Institute, Cambridge, UK) with primers containing restriction sites for NdeI and BamHI and cloned into the same sites in plasmid pET33b+ (Novagen) to generate pAS33 : CR44b_13 and pAS33 : CR8_16, respectively. Similarly, plasmids pAS1 : CR44b_13 and pAS1 : CR8_16 were constructed using the procedure described for pCS1 : Gp2. Mutant forms of T7 Gp2 and CR44b_13 were constructed using the QuikChange Site-Directed PCR Mutagenesis kit (Stratagene) using pCS1 : Gp2, pSW33 : Gp2 or pAS1 : CR44b_13 as a template. T7 Gp2 was overexpressed and purified as described previously; CR44b_13 and CR8_16 were overexpressed and purified coli RNAp b9 jaw domain fragment (residues 1153-1213) complex. Gp2 is shown in cyan and the b9 jaw domain fragment is shown in green. The positions of the amino acids E24, F27 and R56 in T7 Gp2 are indicated along with the E1158 and E1188 residues in the b9 jaw domain. (b) Alignment of amino acid sequences from known Gp2-like proteins. The sequences are displayed in single-letter amino acid code, with the length of the sequence indicated on the far right. The localization of the b-strands and a-helix based on the structure of T7 Gp2 is indicated. The highly conserved R56 and R58 residues are highlighted in blue. The positions of amino acids residues E24 and F27 in T7 Gp2 are underlined and in bold type. The arginine residue, corresponding to T7 Gp2 amino acid residue S23 in Gp2-like proteins in phages CR8, CR44b, K1F and EcoDS1, is also underlined in bold type. (c) Alignment of the b9 jaw domain amino acid residues 1153-1213 from a representative set of host RNAps, with those infected by T7-like phages encoding Gp2-like proteins indicated by asterisks. Amino acid residues corresponding to E1158 and E1188 of the E. coli RNAp are shown in bold type. (d) As in (a), except that the structural models of Gp2-like proteins CR44b_13 (orange) and CR8_16 (pink) are superimposed and amino acid residue R15 is indicated. The structural models of the Gp2 homologues were calculated with SWISS-MODEL using NMR structures of T7 Gp2 as a template (Arnold et al., 2006;Cá mara et al., 2010).
Inhibition of E. coli RNA polymerase by T7 Gp2 exactly as T7 Gp2 (Cámara et al., 2010). Wild-type and mutant forms of the E. coli core RNAp were overexpressed from pVS10 and purified by affinity chromatography exactly as described previously (Cámara et al., 2010). The E1188K and E1158K mutant E. coli RNAps were constructed by using the QuikChange Site-Directed PCR Mutagenesis kit using pVS10 as the template (Belogurov et al., 2007). Purified proteins were then dialysed into storage buffer [10 mM Tris/HCl (pH 8.0), 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 50 % (v/v) glycerol] and aliquots were stored at 220 uC (short-term) and 280 uC (long-term). The sequences of primers used for the cloning and mutagenesis in this study are available from the authors upon request.
Bacterial growth attenuation assays. Seed cultures were grown at 37 uC, shaking at 700 r.p.m. for 6-7 h in a THERMOstar (BMG Labtech) plate incubator by directly inoculating a colony from a freshly transformed Luria agar plate into 200 ml of Luria broth (LB) medium (Difco) containing 100 mg ampicillin ml 21 and 0.5 % (w/v) glucose (to prevent leaky expression of Gp2 from the P BAD promoter) into a 96well microtitre plate (Sterilin). The experimental growth curves were also performed in 96-well microtitre plates in a POLARstar Omega multiwell plate reader (BMG Labtech). The seed cultures were diluted 1 : 100 in a final volume of 200 ml of fresh LB medium containing 100 mg ampicillin ml 21 and incubated at 30 uC, shaking at 500 r.p.m. The expression of Gp2 was induced at OD 600 of~0.2-0.25 by adding 0.04 % (w/v) arabinose for MG1655, JE1134 and BR3 cultures, and 0.4 % (w/v) arabinose for 7009 cultures. The doubling time (Dt) was calculated from the gradient of the growth curve typically between the OD 600 corresponding to induction time and OD 600 0.7-0.8. The R 2 values in each case were .0.98. At least three biological and technical replicates were performed for each growth curve.
Quantitative Western blotting. E. coli MG1655 cultures were grown in conditions as described above. At 2 h post induction (see text for details), cell samples were taken and the total amount of Gp2 molecules per cell was determined by Western blotting using standard protocols. E. coli whole-cell extract (corresponding to 2.5610 7 cells) was analysed by SDS-PAGE on gels that were calibrated with known amounts of purified Gp2. Proteins were transferred to a Hybond-ECL nitrocellulose membrane using a Trans-Blot Semi-Dry transfer cell (Bio-Rad). To detect Gp2, anti-His monoclonal antibodies conjugated with horseradish peroxidase (HRP) (Sigma) in combination with the ECL SuperSignal West Femto Chemiluminescent Substrate kit (Pierce) were used. Digital images of the blots were obtained using an LAS-3000 Fuji Imager, and signal quantification and calculations were performed exactly as described by Piper et al. (2009). Briefly, the amount of Gp2 (ng) (Fig. 2d, table, column 2) at each time point post induction was estimated from the calibration curve. The number of molecules of Gp2 per cell (Fig. 2d, table, column 3) was determined by first calculating the total number of moles of Gp2 [mass (g)/M r (10 000 Da)], multiplying by Avogadro's constant (6.022610 23 mol 21 ) and then by dividing the total number of Gp2 molecules by the number of cells in the sample (i.e. 2.5610 7 cells).
In vitro transcription assays. These were performed exactly as previously described (Cámara et al., 2010).

Insights from multiple protein sequence alignment of T7 Gp2 homologues
To determine how amino acids at positions corresponding to T7 Gp2 E24, F27 and R56 compare with corresponding residues in homologous proteins encoded by related phages, we conducted a BLAST search using standard search parameters and T7 Gp2 as a query sequence. The multiple protein sequence alignment of known Gp2 homologues (in the EBI database, July 2012) was done using COBAT ( Fig.  1b) (Papadopoulos & Agarwala, 2007). We also built an alignment of the T7 Gp2 target (the b9 jaw fragment region, E. coli b9 amino acids 1153-1213) with corresponding sequences from several bacterial species (Fig. 1c). As expected, the amino acid residues R56 and R58 in T7 Gp2 that are essential for the binding to the E. coli b9 jaw are identical in 25 and 23, respectively, out of 25 known Gp2 homologues (Fig. 1b). Similarly, amino acid residues E1158 and E1188 display a high degree of conservation only in bacteria infected by phages encoding Gp2 homologues (Fig. 1c). Overall, this observation is consistent with previous structural and biochemical analyses which show that the primary interaction interface between T7 Gp2 and the E. coli RNAp involves R56 and R58 in strand b3 of Gp2 and E1158 and E1188 in the b9 jaw domain.
A phenylalanine residue at a position corresponding to F27 of T7 Gp2 is conserved in 18 out of 24 Gp2 homologues, whereas amino acids at a position corresponding to E24 in T7 Gp2 display a significant degree of variation (Fig. 1b). Interestingly, we note that phages CR8, CR44b, K1F and EcoDS1 encode Gp2-like proteins that contain a positively charged amino acid (R15) at the position that corresponds to T7 Gp2 S23, a residue immediately adjacent to E24 that is changed for lysine in Gp2 b (Fig. 1b, underlined). From structural models generated for Gp2 homologues from phages CR8 and CR44b and based on the solution structure of the T7 Gp2-b9 jaw domain fragment complex, it is evident that R15 in CR44b_13 and CR8_16 (and by extension also in Gp2 homologues encoded by K1F and EcoDS1) is located in the loop interconnecting b1 and b2 strands of the proteins (Fig. 1d). Thus, it is possible that Gp2 homologues from CR8, CR44b, K1F and EcoDS1 phages, like T7 Gp2 b , can also inhibit 7009 and/or BR3 RNAp. If true, this would indicate that Gp2 homologues encoded by wild-type CR8, CR44b, K1F and EcoDS1 phages interact with RNAp like the mutant Gp2 encoded by T7b.
A simple in vivo assay measures the activity of Gp2 in the absence of T7 phage infection To experimentally address the observations derived from multiple sequence alignment and to understand the molecular basis by which the triple amino acid substitutions in Gp2 b suppress the E1188K mutation in the b9 jaw domain, we developed a simple in vivo assay to determine and compare the ability of wild-type and mutant Gp2 proteins to inhibit E. coli RNAp in the absence of T7 infection. In this assay, gene 2 (which encodes Gp2) is placed under the control of an arabinose-inducible araBAD promoter (P BAD ) in plasmid pCS1 : Gp2 and introduced Inhibition of E. coli RNA polymerase by T7 Gp2 into E. coli strain MG1655. As shown in Fig. 2(a), induction of Gp2 expression upon addition of 0.04 % (w/ v) arabinose (at t50) results in efficient attenuation of bacterial growth. The attenuation of bacterial growth is specific, since it is not observed in wild-type cells transformed with the pCS1 : Gp2 R56E plasmid, which encodes a functionally defective Gp2 mutant (Cámara et al., 2010) (Fig. 2a), or in E. coli strain 7009 transformed with pCS1 : Gp2 (Fig. 2b). Control experiments established that attenuation of bacterial growth was not caused by a loss of selection due to the inhibition of transcription of the b-lactamase gene from pCS1, since attenuation of bacterial growth upon arabinose induction was observed in the absence of ampicillin (Fig. 2a, b). Additional control experiments, shown in Fig. 2(c), established that recombinant Gp2, when expressed in E. coli, acts like a bacteriostatic agent: growth-attenuated E. coli MG1655 cells obtained 2 h (i.e. at t52) after Gp2 induction resumed growth when inoculated into fresh growth medium after removal of the inducer, and stopped growing upon reinduction with arabinose at t56 h (Fig. 2c). Further analysis of whole-cell extracts from cells obtained at t52 h after Gp2 induction revealed that the total number of Gp2 molecules per E. coli cell is~5000 molecules (Fig. 2d). Since the total number of RNAp molecules in an exponentially growing E. coli cell is estimated to be~2000-2500 molecules (Ishihama, 1999), it seems that the total number of Gp2 molecules per E. coli cell exceeds that of the total number of RNAp molecules per E. coli cell by at least twofold under conditions where attenuation of bacterial growth is observed. In summary, the in vivo bacterial growth attenuation assay established here can be used to determine and compare the activity of Gp2 mutants to inhibit the E. coli RNAp in vivo in the absence of T7 phage infection.

E24K and F27Y substitutions facilitate Gp2 interaction with RNAp when the primary interaction interface is compromised
It is not known which one of the three changes (or a combination thereof) restores the binding of Gp2 b to the 7009 RNAp. Therefore, we constructed single, double and triple mutant versions of Gp2 based on Gp2 b and tested their ability to inhibit the wild-type, 7009 and BR3 E. coli RNAp using appropriate strains and the in vivo bacterial growth attenuation assay described above (Fig. 2). Consistent with previous results from alanine-scanning mutagenesis (Cámara et al., 2010), single or double amino acid substitutions at E24 (to K) and/or F27 (to Y) attenuated the growth of E. coli MG1655 to the same degree, as did the wild-type Gp2. On the other hand, expression of Gp2 R56C did not significantly affect the growth rate (expressed as Dt in Table 1) as compared with cells harbouring the empty plasmid vector (pBAD18), indicating that the R56C substitution strongly affected the interaction with the wild-type RNAp. However, the Dt of cells expressing double mutant Gp2 variants containing the R56C mutation in combination with either the E24K or the F27Y mutation (i.e. Gp2 E24K/R56C and Gp2 F27Y/R56C ) was detectably increased (by~4.1-and~1.8-fold, respectively) compared with the growth rate of cells harbouring the empty plasmid vector (Fig. 3a, Table 1). Although compared with wildtype Gp2, Gp2 E24K , Gp2 F27Y and Gp2 E24K/F27Y mutants, the Gp2 E24K/R56C and Gp2 F27Y/R56C mutants attenuated the growth of MG1655 E. coli cells relatively weakly (Fig. 3a,  Inhibition of E. coli RNA polymerase by T7 Gp2 Table 1), the effect was highly reproducible. When present together, the E24K and F27Y mutations further improved the ability of Gp2 R56C (i.e. Gp2 b ) to attenuate wild-type E. coli growth (Fig. 3a), and the Dt of cells expressing Gp2 b was increased~5.3-fold compared with that of cells harbouring the empty plasmid vector (Table 1). In summary, it appears that mutations in amino acids residues in and surrounding the loop interconnecting b1 and b2 strands in Gp2 (i.e. E24K and F27Y) can compensate for mutations in Gp2 (i.e. R56C) that compromise (but not abolish; see below) the interaction interface between Gp2 and the b9 jaw domain. Based on Dt values calculated for MG1655 E. coli cells after induction of Gp2 E24K/R56C , Gp2 F27Y/R56C and Gp2 b (Table 1), it seems that the charge reversal substitution at E24 (i.e. E24K) in the loop interconnecting the b1 and b2 strands in Gp2 rather than the F27Y mutation at the beginning of the b2 strand is more important in enabling the interaction between Gp2 and the RNAp when the primary interaction interface with the b9 jaw is compromised by the R56C substitution. However, since a glutamate substitution at position R58 in the context of Gp2 b effectively abolishes the ability of Gp2 b to attenuate the growth of wild-type E. coli, it would seem that an interaction interface between the b3 strand and b9 jaw domain still persists in the Gp2 b -RNAp complex (Fig. 3a). Experiments with the 7009 and BR3 E. coli strains, which encode Gp2-resistant forms of RNAp (see Introduction), further corroborated this view: even though the Dt of 7009 and BR3 E. coli cells, when compared with the MG1655 E. coli cells, is relatively unaffected by overproduction of wild-type Gp2, Gp2 b increases the Dt of 7009 and BR3 E. coli cells by~2.6and 2.2-fold, respectively, compared with corresponding cells expressing wild-type Gp2 (Fig. 3b, c, Table 1).
Interestingly, in the context of the BR3 E. coli cells, expression of Gp2 E24K/F27Y attenuates growth as efficiently as that of Gp2 b (Fig. 3c, Table 1). In both cases, the R58E substitution abolishes the ability of Gp2 b to attenuate cell growth (Fig. 3b, c). Overall, the results strongly suggest that the E24K and F27Y substitutions facilitate an interaction between Gp2 and RNAp when the primary interaction interface in the b3 strand is compromised.
T7 Gp2 amino acids 14-59 are sufficient to inhibit the E. coli RNAp It is evident from the alignment (Fig. 1b) that the highest degree of sequence conservation occurs between residues  corresponding to T7 Gp2 amino acids 14-59. Therefore, it is possible that amino acids outside this region are dispensable for activity. We used the in vivo bacterial growth attenuation assay to determine whether the truncated form of Gp2 (Gp2 14-59 ) is functionally active. Results shown in Fig. 4(a) and Table 2 reveal that expression of Gp2 14-59 detectably attenuates MG1655 E. coli cell growth, though not as efficiently as full-length Gp2 expression. The Dt of MG1655 cells expressing Gp2 14-59 is increased~4.9-fold when compared with MG1655 E. coli cells harbouring the empty plasmid vector (Table 2). Since further N-and C-terminal truncations (e.g. Gp2 16-59 , Gp2 14-58 and Gp2 14-55 ) resulted in proteins unable to affect cell growth rate (data not shown), the results suggest that a minimal functional region of T7 Gp2 comprises amino acids 14-59. The attenuation of cell growth by Gp2 14-59 is specific, since no growth attenuation is detected in the context of the R56C mutation (Fig. 4b). However, the E24K and F27Y mutations clearly improve the efficiency by which Gp2 14-59 attenuates MG1655 cell growth (Fig. 4c). The Dt of MG1655 E. coli cells expressing Gp2 14-59 carrying the E24K/F27Y double substitution is increased~6.9-fold compared with that of cells harbouring an empty plasmid vector (Table 2). This result indicates that E24K and F27Y substitutions improve the binding of Gp2 to the RNAp even if the primary interaction interface with the b9 jaw domain is intact but the binding affinity is attenuated by N-and C-terminal deletion of evolutionarily variable segments.
Gp2 homologues from phages that infect C. rodentium inhibit E. coli RNAp even if the interaction with the b § jaw is compromised To obtain further experimental evidence that E24K and F27Y substitutions in Gp2 facilitate the interaction of Gp2 with RNAp when the primary interaction interface with the b9 jaw domain is compromised, we conducted experiments with Gp2 homologues, proteins 13 and 16 from phages CR44b and CR8 (CR44b_13 and CR8_16), which infect C. rodentium, a model animal bacterial pathogen used to study human infections by enteropathogenic and enterohaemorrhagic E. coli. Initially, we wanted to determine whether recombinant CR44b_13 and CR8_16 can complement a T7 mutant phage (T72am64) harbouring a mutation in gene 2 (Burck & Miller, 1978 Next, we conducted a growth attenuation assay to establish whether CR44b_13 and CR8_ 16, like T7 Gp2, bind to the b9 jaw domain of the E. coli RNAp. As shown in Fig. 5(a), induction of CR44b_13 and CR8_16 expression with 0.04 % (w/v) arabinose from pAS1 : CR44b_13 and pAS1 : CR8_16, respectively, in wild-type E. coli MG1655, resulted in efficient growth attenuation. However, no growth attenuation of E. coli strain JE1134, which encodes a b9 subunit with residues 1149-1190 deleted (Fig. 5b), was observed. We next determined whether CR44b_13 and CR8_16 can inhibit E. coli RNAp harbouring either the E1158K (i.e. found in BR3 E. coli strain) or E1188K (i.e. found in 7009 E. coli strain) mutations, as suggested by sequence analysis (above). As shown in Fig. 5(c), whereas the growth of E. coli strain 7009 was not detectably attenuated by either CR44b_13 or CR8_16, the growth of E. coli strain BR3 was efficiently attenuated by both phage proteins (Fig. 5d). In vitro transcription assays, which report the ability of T7 Gp2 CR44b_13 and CR8_16 proteins to inhibit the synthesis of the tetranucleotide RNA product (ApApUpU) from the lacUV5 promoter by the RNAp containing s 70 , with purified wild-type, BR3 and 7009 RNAp and T7 Gp2, confirmed the in vivo results (Fig. 5e).
If a charge reversal substitution (E24K) in T7 Gp2 b indeed facilitates binding to the RNAp when the primary contact with the b9 jaw domain is weakened (Fig. 3), then R15 of CR44b_13 should be important for attenuation of BR3 E. coli growth. Results shown in Fig. 5(f) confirm that this is indeed the case: the mutant variant of CR44b_13 harbouring the R15E mutation fails to attenuate the growth of E. coli strain BR3 upon induction. However, the R15E mutation has no detectable effect on the ability of CR44b_13 Gp2 to attenuate the growth of wild-type E. coli.
Overall, results with Gp2-like proteins encoded by C. rodentium phages provide further support for the view that the molecular basis by which Gp2 b is able to inhibit the 7009 RNAp is that E24K and F27Y substitutions facilitate an interaction between Gp2 and RNAp when the primary interaction interface is compromised.

DISCUSSION
T7 Gp2 binds tightly to and potently inhibits E. coli RNAp. Previous biochemical and structural studies have shown that amino acid residues in the b3 strand of Gp2 contribute to the primary interaction interface with the b9 jaw domain Inhibition of E. coli RNA polymerase by T7 Gp2 of the RNAp. Results from the current study show that substitutions in amino acid residues of Gp2 located in the region surrounding and including the loop interconnecting the b1 and b2 strands, and, therefore, located on the opposite side to the b3 strand, can compensate for amino acid substitutions in Gp2 and/or the b9 jaw domain that compromise the primary interaction interface between Gp2 and the RNAp. This seems to constitute the molecular basis by which T7b (encoding Gp2 b harbouring the E24K, F27Y and R56C substitutions) can successfully infect the 7009 E. coli strain (harbouring the E1188K mutation) that is resistant to infection by wild-type T7 phage.    Inhibition of E. coli RNA polymerase by T7 Gp2