A basic/hydrophobic cleft of the T4 activator MotA interacts with the C-terminus of E. coli σ⁷⁰ to activate middle gene transcription

Summary

Transcriptional activation often employs a direct interaction between an activator and RNA polymerase. For activation of its middle genes, bacteriophage T4 appropriates E. coli RNA polymerase through the action of two phage-encoded proteins, MotA and AsiA. Alone, AsiA inhibits transcription from a large class of host promoters by structurally remodeling region 4 of σ⁷⁰, the primary specificity subunit of E. coli RNA polymerase. MotA interacts both with σ⁷⁰ region 4 and with a DNA element present in T4 middle promoters. AsiA-induced remodeling is proposed to make the far C-terminus of σ⁷⁰ region 4 accessible for MotA binding. Here, NMR chemical shift analysis indicates that MotA uses a “basic/hydrophobic” cleft to interact with the C-terminus of AsiA-remodeled σ⁷⁰, but MotA does not interact with AsiA itself. Mutations within this cleft, at residues K3, K28, and Q76, both impair the interaction of MotA with σ⁷⁰ region 4 and MotA-dependent activation. Furthermore, mutations at these residues greatly decrease phage viability. Most previously described activators that target σ⁷⁰ directly use acidic residues to engage a basic surface of region 4. Our work supports accumulated evidence indicating that “σ appropriation” by MotA and AsiA uses a fundamentally different mechanism to activate transcription.

Introduction

The activation of transcription at specific promoter sequences involves a delicate interplay between DNA-binding transcription factors and RNA polymerase (RNAP). In both eukaryotes and prokaryotes, these factors can work by directly contacting the subunits of polymerase or polymerase-associated factors (Browning & Busby, 2004, Chen & Hampsey, 2002). Bacterial RNAP is a multi-subunit enzyme consisting of a core of catalytic subunits (α₂, β, β′, ω) and a specificity factor, σ, that interacts with promoter elements to set the start site for transcription (Murakami & Darst, 2003). The primary σ factor needed for initiation of transcription from general housekeeping genes during exponential growth in E. coli is σ⁷⁰ (Gruber & Gross, 2003, Paget & Helmann, 2003). σ⁷⁰ recognizes conserved promoter elements through direct interactions with nucleotide base determinants. Regions 2 and 3, located within the central portion of the σ⁷⁰, interact with sequences at positions from −15 to −7 relative to the transcription start site, while the C-terminal portion of σ⁷⁰ (region 4, σ4) interacts with sequences in the −35 region of the promoter (reviewed in (Hook-Barnard & Hinton, 2007)). In addition, other residues within σ4 interact with the flap domain of the β subunit (β-flap), to position σ4 correctly relative to regions 2 and 3, allowing simultaneous binding of both the −10 and −35 elements (Kuznedelov et al., 2002).

Transcription initiation from sub-optimal canonical promoter sequences can require the interaction of a transcription factor with RNAP to enhance polymerase recruitment or to accelerate rate limiting steps in the formation of the transcriptionally competent RNAP/promoter complex (Browning & Busby, 2004, Lawson et al., 2004, Barnard et al., 2004). Bacterial class I and II activators stimulate transcription from less than ideal −35 and/or −10 promoter elements. Class I activators recognize DNA elements upstream of the conserved -35 promoter sequence and interact with RNAP through the α subunit, while Class II activators bind to a DNA target that overlaps the -35 element and interact with α and/or σ4. Another class of prokaryotic activators, exemplified by the MerR family, bind to the spacer region between the -10 and -35 elements. This binding then induces a distortion in the DNA, allowing RNAP to recognize a promoter with an unusually long spacer region (Hobman et al., 2005, Watanabe et al., 2008). However, despite the different locations for activator binding in these various systems, the overall result is to maintain the interaction of σ4 with the -35 DNA.

A fundamentally different form of transcriptional activation is used by bacteriophage T4, whereby σ⁷⁰ is appropriated by the virally-encoded proteins AsiA and MotA, for the recognition of the middle class of T4 promoters (reviewed in (Hinton et al., 2005). AsiA binds to σ4 (Severinova et al., 1996, Lambert et al., 2004, Stevens & Rhoton, 1975), inhibiting transcription from host promoters that require the −35 element for recognition and co-activating transcription, with MotA, from T4 middle promoters (Colland et al., 1998, Severinova et al., 1998, Ouhammouch et al., 1995, Stevens, 1976). The consequence of the AsiA interaction with σ4 is dramatic and results in a σ4 structure that differs substantially from the similar σ4 structures that have been observed using unbound σ^A of Thermus aquaticus (Campbell et al., 2002) or Thermatoga maritima (Lambert et al., 2004), RNAP of T. aquaticus (Murakami et al., 2002) or Thermus thermophilus (Vassylyev et al., 2002), or σ4 of E. coli when it is bound by the putative anti-sigma factor, Rsd (Patikoglou et al., 2007). In the AsiA/σ4 structure (Lambert et al., 2004), many of the σ4 residues that normally contact the -35 DNA or the β-flap now contact AsiA; biochemical analyses of σ⁷⁰ proteins containing specific mutations within σ4 have provided independent support for a number of these contacts (Baxter et al., 2006). Furthermore, the multiple interactions between AsiA and σ4 completely remodel σ4, ‘straightening’ the DNA-binding helix-turn-helix element (Lambert et al., 2004). As a consequence, σ4 loses its ability to interact with the β-flap portion of core and to recognize the −35 promoter element. The AsiA/σ4 interactions also indirectly affect the C-terminal fifth helix (H5), resulting in a more disordered region. The interaction of AsiA with σ⁷⁰ occurs before σ⁷⁰ associates with core RNAP (Hinton & Vuthoori, 2000); presumably the parts of σ4 needed for AsiA binding are inaccessible to AsiA once polymerase has formed.

The MotA protein activates transcription from phage middle promoters, which drive expression of genes through the middle phase of the T4 lifecycle. These promoters lack a -35 consensus DNA binding element, but replace this sequence with one centered at -30 (MotA box). MotA is a two-domain protein (Finnin et al., 1993) with a C-terminal domain (CTD) that binds the MotA box (Pande et al., 2002) and a N-terminal domain (NTD) that binds σ⁷⁰ (Gerber & Hinton, 1996, Pande et al., 2002) and is required to activate transcription (Gerber & Hinton, 1996, Finnin et al., 1997). These two functions would allow MotA to replace the canonical σ4/-35 interaction by acting as a bridge between σ⁷⁰ and viral promoters ((Pande et al., 2002); reviewed in (Hinton et al., 2005)). Assignment of these domain activities is supported by the demonstration that MotA with mutations in, or a deletion of, its NTD, binds DNA with an affinity similar to that of wild-type MotA, but fails to activate transcription (Gerber & Hinton, 1996, Pande et al., 2002, Finnin et al., 1997). Furthermore, 2-hybrid analyses have demonstrated that the MotA^NTD interacts with σ4, and in vitro transcription using σ⁷⁰ proteins with C-terminal deletions of 6 or 10 residues have indicated that a primary contact between MotA and the RNAP occurs through the far C-terminus of σ4 (Pande et al., 2002).

To better understand the mechanism of MotA activation, we have investigated the molecular determinants required for the MotA/σ4 interaction. Using NMR, molecular biological, biochemical, and genetic techniques, we have defined the interface of MotA that interacts with the far C-terminal region of σ4. Our data provide a model whereby three separate regions of MotA form a basic/hydrophobic cleft on one face of the protein to engage the C-terminus of σ4.

Results

MotA engages the C-terminal end of a remodeled σ4

In order to probe the interaction between MotA and the preformed AsiA/σ4 complex, NMR footprinting was used to determine which elements of each protein in the ternary complex were likely to be in contact with one another. The structure of AsiA bound to σ4 (residues 533 to 613) has previously been determined by NMR analysis and the ¹⁵N-¹H HSQC (HSQC hereafter) spectrum is known (Lambert et al., 2004). In the case of full length MotA, a standard set of two and three dimensional NMR experiments were collected in order to determine backbone chemical shifts (Fig.1). HSQC spectra were collected for ¹⁵N-AsiA/σ4/MotA, ¹⁵N-σ4/AsiA/MotA and ¹⁵N-MotA/σ4/AsiA (red in left panels A-C); they are shown, overlaid, with those for ¹⁵N-AsiA/σ4, ¹⁵N-σ4/AsiA and ¹⁵N-MotA respectively (black in left panels A-C). When plotting the chemical shift difference measured between the reference binary state or single MotA and the corresponding ternary complexes, with respect to the ¹⁵N-labeled protein residue number, the histograms on the right side of Fig. 1 are obtained.

Fig. 1. Mapping the σ4 and AsiA binding surfaces on MotA.

Full-length MotA was assembled into a ternary complex with σ4 and AsiA (see Experimental Procedures). Several different complexes were prepared in which either A) AsiA, B) σ4 or C & D) MotA was ¹⁵N-enriched in the complex and analyzed with a single ¹⁵N-¹H HSQC NMR spectrum (left panels A-D). On the right side of each panel, the chemical shift changes observed in the ternary complex compared to the corresponding binary complex (A-B) or free MotA (C-D) are shown. The amino acid position of the labeled protein is indicated on the horizontal axis of the histogram. Experiments were carried out at 36°C, panels A, B and C, with σ4 (σ⁷⁰ residues 533 to 613) or at 20°C, panel D, with σ4-H5, a peptide consisting of σ⁷⁰ residues 601-613. (While working at 20°C, the ternary complex spectra, viewed from the MotA side, had significantly broader peaks than those measured at 36°C and shown in Fig. 1C. This is illustrative of a weak binding affinity and a relatively fast exchange rate between the free and σ4-H5 bound states of MotA.) Grey bars in (B) represent σ4 peaks that have disappeared due to line broadening; residues without bars in (B) correspond to peaks that have either been broadened beyond detection or could not be identified. Missing peaks in the histograms A, C, and D correspond to near zero chemical shifts except for the following: C) MotA residues 30, 31 and 81 could not be assigned in the AsiA/σ4 bound state; D) MotA residues 81 and 82 could not be assigned in the σ4-H5 bound state. (Note that the Δ(ppm) scale in Fig. 1D differs from that of the other panels.)

There are minimal changes in chemical shift when the ternary complex is viewed from the perspective of AsiA (Fig. 1A), suggesting that AsiA and MotA do not interact in the complex. This is consistent with the fact that a variety of biochemical analyses have failed to demonstrate an interaction between MotA and AsiA (Hinton, unpublished). The line widths in the NMR spectra of the binary σ4/AsiA complexes (Fig. 1B black), here and previously (Lambert et al., 2004) have appeared broader than the expected 18 kDa molecular weight of σ4/AsiA, which may reflect the tendency of σ4 to self-aggregate. However, the addition of MotA to the binary complex results in a further, specific broadening of signals leading to their disappearance. When studying a complex in solution by NMR, the exchange rate (k_ex) between the free and bound states plays an important role. Indeed, if k_ex is slow, relative to the NMR timescale, two distinct signals will be observed: one for the free component and one for the complex. On the other hand, if k_ex is fast, a single resonance is observed at a population weighted average chemical shift. However, if k_ex is similar to the NMR timescale, the signal of interest will be smeared between the reference positions of the free and bound states, thereby broadening it beyond detection. This is precisely the case for the four peaks represented by grey bars in Fig. 1B. These vanishing signals correspond to residues within the extreme C-terminus of σ4, consistent with previous data suggesting an interaction between this segment of σ4 and MotA (Pande et al., 2002). From the perspective of MotA, the association with the σ4/AsiA complex results in significant chemical shift changes within the MotA^NTD, primarily located to two regions, residues 23-29 and 80-83 (Fig. 1C). No significant changes were observed for residues within the CTD of MotA (Supplemental Fig. 1).

To confirm that the C-terminus of σ4 forms the principal contact surface with MotA, a peptide of σ4 encompassing residues P601-D613, designated σ4-H5, was chemically synthesized and titrated into a solution of intact MotA (data not shown). The gradual shifting of MotA peaks with increasing amount of peptide added, indicates that the binding affinity for the peptide is weak and that the two components are in fast exchange on the NMR timescale. Chemical shift changes in MotA, induced by the peptide (Fig. 1D), occurred within similar regions to those observed to be perturbed by σ4/AsiA (Fig. 1C). However, in addition to the regions shifted using σ4/AsiA, significant changes were observed with σ4-H5 at the N-terminus of MotA (extending to I7) and in the region of Q76 to T84 (Fig. 1D and summarized in Fig. 2). When scrutinizing the histogram shown in Fig. 1C, it is possible to discern the aforementioned regions, so clearly defined in the presence of σ4-H5. These residues likely play a role in the ternary complex as well and might not have been considered, were it not for our ability to push the binding equilibrium to saturation, when using excess σ4-H5. Taken together, the NMR footprinting using σ4/AsiA or the σ4-H5 peptide suggests that three regions of MotA (V4-I7, L23-K29, and Q76-Y86) potentially interact with the far C-terminus of σ4.

Fig. 2. MotA interaction domain.

A) Clustal W (Thompson et al., 1994) alignment of MotA homologs from T4 like phages (Accession numbers, T4, NP_049873; RB69, NP_861954; PHG25, YP_656463; 44RR, NP_932590; PHG133, available at the T4-like Genome website, http://phage.bioc.tulane.edu/cgi-bin/gbrowse/PHG133?name=PHG133%3A1..159897;source=PHG133;width=800;label=Genes-T4BLAST%20Hits-ORFs-RB69BLAST%20Hits). Alpha helix and beta strand secondary structure elements are designated by blue rectangles and arrows, respectively. Assignments from the MotA^NTD solution structure (Li et al., 2001) are prefixed with an “N” and assignments from the MotA^CTD X-ray structure (Li et al., 2002) are prefixed with a “C”. Regions where MotA amino acid residues showed perturbed chemical shifts in the presence of σ4/AsiA or σ4-H5 (Fig. 1C and D) are indicated by yellow rectangles (The K3 residue was not observed in the NMR analysis but was included due to its proximity to a shifted region and its importance in biochemical and genetic assays. See Figs. 3 & 4). Dots indicate every tenth residue of T4 MotA. B) MotA residues with perturbed chemical shifts in the presence of σ4/AsiA or σ4-H5 (yellow in panel A) and other positions discussed in the text (grey) are mapped onto the solution structure of MotA^NTD (PDB ID 1I1S).

The MotA interaction face

MotA homologs have been found in a number of T4-related phages, and they contain regions of conservation/identity throughout both the activation (NTD) and DNA-binding (CTD) domains (Fig. 2A). In addition, although the structure of the intact MotA protein is not available, x-ray and solution structures for both the NTD (Finnin et al., 1997, Li et al., 2001) and CTD (Li et al., 2002, Finnin et al., 1994) domains have been obtained. The NMR solution structure of MotA^NTD ((Li et al., 2001); see Fig. 2B) is a monomer while it is observed as a dimer in the X-ray structure (Finnin et al., 1997). Biochemical evidence (Sharma et al., 1999, Cicero et al., 1998, Li et al., 2001) suggests that MotA binds to DNA as a monomer, suggesting that the MotA dimerization seen in the crystal is an artifact of crystal packing (Li et al., 2001).) Mapping of the regions of MotA that interact with σ4 or σ4-H5 onto the amino acid alignment of known motA homologs (Fig. 2A) and the MotA^NTD NMR structure (Fig. 2B) indicates that these regions correspond to portions of alpha helices Nα1, Nα2 and Nα5, which form a cleft on the same concave face of MotA^NTD.

To address the significance of the cleft, we over-expressed and purified MotA proteins containing single amino acid replacements (K3E, K28E, Q76R or S80R) introduced into each of the three putative interaction regions (Fig. 2). We expected these substitutions to alter the properties of the side chain sufficiently for the interaction between the MotA and σ4 proteins to be disrupted. Additionally, we introduced mutations in three other positions, D43K, K59D or D67K that showed little shift perturbation in the NMR footprints (Figs. 1 & 2). To avoid disrupting the hydrophobic core of the protein, all of the residues chosen for mutation were polar and were surface exposed in both the crystallographic and solution structures of MotA^NTD (Li et al., 2001, Finnin et al., 1997). Though residue K3 was not observed in the NMR analysis, it was chosen for mutation; it is both well conserved and close to V4 (Fig. 2).

MotA mutations at the MotA/σ4 interface destabilize the protein-protein interaction

Each of the MotA mutant proteins was assayed for its ability to bind σ4-H5 using fluorescence anisotropy with an N-terminal fluorescein-tagged σ4-H5 peptide (Fig. 3A). The affinity of MotA for σ4-H5 was reduced at least two-fold by the MotA mutants K3E, K28E, Q76R and S80R. For the K3E and K28E mutants, binding was so weak that only a lower limit to the binding constant could be estimated (data not shown). By contrast, the MotA D43K and K59E mutant proteins, whose mutations are predicted to be distant from the σ4-binding surface, showed no significant reduction in binding affinity, and MotA D67K showed a 1.5-fold decrease in binding affinity to the peptide.

Fig. 3. Functional analysis of the MotA/σ4 interface.

A) Fluorescence polarization assay of wild type or mutant full-length MotA interacting with fluorescein-tagged σ4-H5. Measured data is shown as colored points and the fit to the observed changes in polarization is shown as a solid line. The apparent equilibrium dissociation constant (in μM) for each of the indicated MotA species is wt, 15.9 +/- 1.3; K3E, >300; K28E, >300; Q76R, 39.1 +/- 5.0; S80R, 44.7 +/- 5.0; D43K, 12.0 +/-1.0; K59E, 10.0 +/- 1.0; D67K, 24.3 +/- 0.1. (For K3 and K28, only a lower limit to the dissociation could be determined as the peptide never reached saturation with these weakly binding species.) B) Bacterial two-hybrid assay of mutant MotA^NTD interacting with wild-type σ4. The histogram indicates β-galactosidase activity of MotA^NTD mutants relative to wild-type corrected for background (vector only) levels. The average β-galactosidase levels for wild-type MotA^NTD and vector only controls were 577 and 143 Miller units, respectively. Error bars represent one standard deviation computed from at least three independent experiments. The levels of β-galactosidase with the K3 and K28 mutants were consistently below background and were therefore set to zero. C) Single round transcription from the T4 middle promoter P_uvsX using wild type or mutant MotA. The autoradiogram image is from a representative experiment. All lanes were run on the same gel. Quantitative analysis of the observed level of transcript is shown as a histogram with error bars indicating one standard deviation from the mean from at least three independent measurements. D) Gel retardation of a 21 bp MotA box containing DNA in the presence of wild type or mutant MotA protein. Concentration of MotA protein in each reaction is noted below the lane. Multiple protein/DNA complexes are observed, indicated by the bracketed region. (See text for details.)

An E. coli 2-hybrid system (Dove & Hochschild, 2004) has previously been used to assay the interaction of MotA^NTD with σ4 (Pande et al., 2002). As an independent method for investigating the effect of MotA mutations on the interaction of MotA with σ4, the above mentioned mutations, as well as several individual alanine substitutions, were separately introduced into a fusion of the MotA^NTD (residues 1 to 98) with the λcI DNA binding domain and assayed for binding to σ4 (residues 528 to 613). Consistent with the anisotropy results, mutations at K3, K28, Q76 and S80 reduced the interaction at least 2-fold (Fig. 3B). Most notably, mutations of residues K3 and K28 reduced the interaction to background levels. In contrast, substitutions at positions D43 and K59, expected to be located distal from the MotA/σ4 interface, showed either marginal reduction or an increase in the interaction (Fig. 3B). As was seen with the fluorescence anisotropy analysis, substitution of D67 only modestly impaired MotA^NTD/σ4 interaction.

MotA mutations D30A, F31A or D30A/F31A have previously been implicated in the activation of T4 middle transcription (Finnin et al., 1997). MotA with D30A, F31A, or D30A/F31A substitutions was very deleterious for phage growth, and the MotA D30A/F31A double mutant was significantly impaired in multiple round in vitro transcription assays. We were unable to purify MotA mutant proteins with the D30A or F31A mutations to determine binding of these proteins to σ4-H5 or their activities in a single round transcription assay (see below) because these mutant proteins could not be overexpressed in a soluble form. However, we did not observe any chemical shift of MotA residues 30 and 31 when using the σ4-H5 peptide (Fig. 1D). In addition, only a modest effect on the interaction of MotA^NTD with σ4 was observed in 2-hybrid analyses for D30A and F31A. Even the double substitution (D30A/F31A) did not eliminate the MotA^NTD/σ4 interaction (Fig. 3B). Thus, it seems unlikely that positions D30 and F31 are required for the MotA interaction with the far C-terminal region of σ4.

MotA mutations at the MotA/σ4 interface reduce MotA-dependent transcription

Both fluorescence anisotropy and 2-hybrid analyses indicated that mutation of MotA residues K3, K28, Q76 or S80 greatly and D67 modestly impair the interaction between MotA and σ⁷⁰. However, each assay used only a fragment of σ⁷⁰, and lacked both the core subunits of RNAP and AsiA, which are always present during T4 infection. To determine the effect of each mutation in a more complete in vitro system in the context of full length σ⁷⁰, core RNAP and AsiA, MotA proteins with specific substitutions were assayed for their ability to activate transcription from the T4 middle promoter P_uvsX. Single round transcriptions, rather than multiple round, were chosen so that only transcription initiation events were monitored. Although P_uvsX lacks a canonical σ⁷⁰ −35 element, RNAP alone uses this promoter in vitro (Hinton et al., 1996) because P_uvsX contains other elements recognized by RNAP, a perfect σ⁷⁰ −10 element and a good UP element. Addition of AsiA eliminates this basal level transcription from P_uvsX, whereas addition of AsiA and wild type MotA gives a high level of activated transcription (Fig. 3C). Activated transcription was significantly reduced when using MotA with the substitutions of K3E, K28E, or Q76R. In contrast, MotA with the S80R substitution only slightly decreased transcription, and proteins with the substitutions D43K, K59E, or D67K, which were predicted to be distant from the face that binds to the C-terminal region of σ4, behaved like wild type. In the absence of AsiA, addition of wild type MotA to polymerase results in a slight increase in the level of P_uvsX transcription (Hinton, unpublished). The same effect was observed with the MotA mutant proteins (data not shown). Thus, the diminished activated transcription seen with the K3E, K28E, or Q76R protein is not because the presence of these proteins generally inhibits RNAP.

MotA^NTD mutations do not affect MotA box binding or protein integrity

MotA binds to a ds DNA oligomer containing a MotA box sequence (March-Amegadzie & Hinton, 1995, Sharma et al., 1999), and the C-terminal half of MotA by itself is capable of this binding (Pande et al., 2002). Thus, it seemed unlikely that the deleterious effect of the K3E, K28E, and Q76R mutations on activated transcription from P_uvsX was due to an impaired interaction with the DNA. To confirm this, we tested the interaction of the MotA mutant proteins with DNA containing a MotA box. Previous work has indicated that MotA does not bind the MotA box tightly; very high levels of MotA relative to the DNA are needed to observe a DNase I footprint (March-Amegadzie & Hinton, 1995). In addition, in gel retardation assays, multiple, slowly migrating protein/DNA complexes, which can appear as discrete or smeared species, are observed. Consequently, it is necessary to determine the K_d(app) from the loss of free DNA rather than from the bound species (Sharma et al., 1999). Previous work has reported a Kd_(app) for MotA binding to MotA box DNA of 100 to 200 nM (Cicero et al., 1998, Sharma et al., 1999). Under the conditions used here, we estimate a K_d(app) of ∼600 nM for wild type MotA binding to a 21 bp fragment containing a MotA box. Similar values were found for the MotA substitution mutants K3E (500 nM), K28E (500 nM), Q76R (600 nM), and S80R (600 nM), which were either severely or mildly defective for activation, or for K59E (500 nM), D67K (700nm) and D43K (500nm), which behaved like wild type (Fig. 3D). For D43K and D67K, the protein/DNA complexes appeared much less discrete, suggesting that these protein/DNA complexes were less stable than those formed with wild type MotA. In addition, other gel retardation assays using a different fragment resulted in somewhat higher K_d(app) for these mutant proteins relative to wild type MotA (data not shown). However, previous work has shown that the MotA/AsiA/polymerase complex at a T4 middle promoter is significantly more stable than the complex with MotA alone (March-Amegadzie & Hinton, 1995). Since the D67K and D43K mutant proteins are not impaired in the transcription assay (Fig. 3C), we assume that protein-protein interactions in the transcriptionally competent complex compensate for the less stable interactions between these proteins and the MotA box DNA.

To establish that the functional defect in the MotA mutants was not an artifact of the particular change introduced, each mutant was examined for structural integrity by NMR spectroscopy. None of the functionally defective mutants displayed a defect in protein folding, as evidenced by an almost complete superposition of the ¹⁵N-¹H HSQC spectra of the wild type with that of the K3, K28, Q76, or S80 mutant proteins (Fig. 4). Consequently, we conclude that the significant defects in activated transcription seen with the K3, K28, and Q76 MotA mutants are attributable to impaired activation function rather than defective protein folding or a decrease in DNA binding.

Fig. 4. MotA mutant proteins K3E, K28E, Q76R, and S80R are structurally similar to wild type MotA.

The ¹⁵N-¹H HSQC spectra of MotA-K3E, -K28E, -Q76R and -S80R (red) are shown overlayed with that of MotA-WT (black). The positions of the signals assigned to K28E, Q76R, and S80R are marked. The signal for K3E could not be assigned.

MotA mutations at the MotA/σ4 interface fail to support phage infection in vivo

Our in vitro results suggested that MotA residues K3, K28 and Q76 are vital for both an interaction with σ4 and activated transcription. To investigate whether these mutations were important for phage viability, we used an in vivo complementation assay. T4 motA^- mutants are unable to form plaques on E. coli TabG (Pulitzer et al., 1979), but phage growth can be complemented if the cell contains a plasmid expressing MotA [(Hinton, 1991) and Fig. 5]. Consequently, we plated dilutions of wild-type T4 or the T4 motA amber mutant, T4 amG1 (Mattson et al., 1978) on TabG cells containing plasmids encoding wild type or a mutant motA. As expected, wild-type T4 plat ing was unaffected by the plasmid-encoded genes (Fig. 5 right panels). In contrast, motA mutants at positions K3, K28, Q76, S80, or F31 failed to complement an amG1 infection, whereas alanine substitution mutants at positions D43, K59, or D67 plated with wild-type efficiency (Fig. 5, left panels). These results indicate that MotA residues K3, K28, and Q76, which are crucial for the interaction with σ4 and for transcription activation in vitro, and S80, which has a mild effect on in vitro transcription, are important for phage viability in TabG. These results also confirm the previous results indicating that MotA F31 is needed for growth in TabG (Finnin et al., 1997), although our results indicate that this residue is probably not involved in the direct interaction of MotA with σ4 (Figs. 1 and 3).

Fig. 5. Complementation of T4 motA^- infection by plasmid-encoded MotA mutants.

Lawns of E. coli TabG cells containing plasmids encoding the wild-type motA gene, motA mutants or the vector alone were infected with ∼10 to 10,000 pfu of T4 motA^- (left panels) or wild-type T4 (right panels) in 10-fold increments.

Discussion

Bacteriophage T4 employs an elegant strategy to shift expression of host genes to viral genes during infection (reviewed in (Hinton et al., 2005)). To this end the viral proteins AsiA and MotA work together to switch the specificity of RNAP from host promoters with −35 and −10 recognition elements to T4 middle promoters with a MotA box and −10 recognition elements. Activation of middle promoters occurs because MotA interacts with both σ4 and the MotA box, replacing the canonical σ4/-35 interaction. We have shown that MotA interacts with a peptide corresponding to the last 13 amino acids of σ4 in a similar manner to that seen with the σ4/AsiA complex (compare Fig. 1C vs 1D and Supplemental Fig. 1), supporting the model that the primary contact between MotA and σ⁷⁰ occurs at the far C-terminus of σ⁷⁰ (Pande et al., 2002, Hinton et al., 2005). We have also determined that a face of MotA^NTD, three non-contiguous regions that comprise a cleft on one side of the molecule (Fig. 2), is involved in the recognition of the σ⁷⁰ far C-terminus for activation from T4 middle promoters. Mutation of the K3, K28, or Q76 residues within this cleft reduces MotA-dependent transcription, interaction with σ⁷⁰ region 4, and phage viability (Figs. 3 and 5) supporting the conclusion that this face of MotA is the primary surface needed for activation.

Although several activators are known to interact with σ4, the MotA/σ4 interaction is unique. The salient features of the MotA activation cleft, clusters of basic and hydrophobic residues, are not shared by most of the other prokaryotic activation surfaces that have been described so far. λcI (Jain et al., 2004, Kuldell & Hochschild, 1994, Li et al., 1994, Nickels et al., 2002), CRP-AR3 (Rhodius & Busby, 2000), FNR (Lonetto et al., 1998), Ada at the alkA promoter (Landini & Busby, 1999), RhaS (Bhende & Egan, 2000), and RhaR (Wickstrum & Egan, 2004) all employ acidic residues to contact a basic patch in σ4 between residues 593 and 603 (Lonetto et al., 1998). This patch is contained within the σ⁷⁰ -35 recognition helix, just C-terminal of the residues that interact with the −35 element. Models of class II activation suggest that this patch is suitably positioned to contact activators which bind just upstream of promoters at a location centered at -41.5 (Lonetto et al., 1998, Rhodius & Busby, 2000, Campbell et al., 2002) In contrast, MotA binds the MotA box, centered at -30, just downstream of the -35 element and is not positioned to work with the class II activating patch of σ⁷⁰. Instead, MotA contacts the far C-terminal region of σ⁷⁰, which has been made available by the remodeling of AsiA. Our previous work has suggested that σ⁷⁰ residues L607, S609, F610, and/or L611 may be important for MotA activation (Baxter et al., 2006). If so, the basic/hydrophobic face of MotA could exploit these AsiA-liberated, hydrophobic residues.

MotA was previously proposed to use an acidic/hydrophobic surface consisting of residues D30, F31, E63, D67 and I70 to activate transcription (Finnin et al., 1997). This was an attractive hypothesis since many eukaryotic activators, which had been identified at this time, use an acidic/hydrophobic patch (Triezenberg, 1995). However, although MotA mutations at residues D30 and/or F31 were shown to exhibit deleterious effects on viral burst and the D30A/F31A double mutant was impaired for in vitro transcription (Fig. 5 and (Finnin et al., 1997)), neither of these residues show a significant chemical shift with σ4-H5 (Fig. 1D). Moreover, the D67K substitution has no effect on in vitro transcription from a MotA-dependent T4 middle promoter and D67A has no effect on phage viability (Figs. 3 and 5). However, our results show that MotA mutations of D30A, F31A, D30A/F31A, D67A, or D67K do have modest effects on the interaction of MotA with σ (Fig. 3). Thus, while the basic/hydrophobic patch of MotA defined by our analyses appears to be the primary target for activation and for interaction with the far C-terminus of σ⁷⁰, it seems likely that the surface of MotA that includes residues D30, F31, and D67 plays an as yet undefined role in phage viability and transcription. Further analyses will be needed to elucidate how this surface contributes to MotA activation.

Experimental Procedures

DNA

Plasmids were constructed using standard cloning procedures. Sequences of primers are available upon request. DNA sequence analyses (done by the Facility for Biotechnology Resources of the FDA or Genzyme, Inc.) were used to confirm the integrity of the various constructs throughout the cloned regions.

pNW129 contains an pACYC origin, the kan^r gene, the araC gene, and the arabinose-inducible P_BAD promoter upstream of a multiple cloning site and the rrnB transcription terminator. This plasmid was constructed by first isolating a PCR product using pBAD18 (Guzman et al., 1995), Pfu polymerase (Stratagene), an oligomer that annealed just upstream of the C-terminal end of the araC gene that introduced a BglII site, and an oligomer that annealed just downstream of the rrnB terminator that introduced a NsiI site. After digestion with BglII and NsiI, the product was ligated with pMS1005 (Sharma & Hinton, 1994) that had been digested with BamHI and PstI, cutting pMS1005 within its multiple cloning site. This construction essentially replaced the pBR322 ori and amp^r gene present in pBAD18 with the pACYC ori and kan^r gene in pMS1005. pNW143 was constructed by ligating the NsiI fragment from pMOT58 (Hinton, 1991), which contains the motA gene, into pNW129 that had been previously digested at the PstI within the multiple cloning site. pDKT90 (March-Amegadzie & Hinton, 1995), which contains the T4 middle promoter P_uvsX, was digested with BsaAI to generate a linear template for transcription.

pBRα-σ⁷⁰ (Dove et al., 2000) directs transcription of an α-σ chimera gene, composed of residues 1-248 of the α subunit of RNA polymerase fused in frame to residues 528-613 of σ⁷⁰, which is under the control of tandem promoters lpp and isopropyl-β-D-thiogalatoside (IPTG)-inducible lacUV5. pACλcI32 (Hu et al., 2000) encodes residues 1-236 of the bacteriophage λcI protein under the control of the IPTG-inducible lacUV5 promoter. pMWT expresses a cI-MotA^NTD protein in which the N-terminal domain of motA (encoding residues 1 to 98) is fused in frame to cI in pACλcI32. DNA corresponding to the first 98 codons of motA was amplified with an oligomer overlapping the first 8 codons that incorporates a NotI restriction site immediately upstream of the start codon and an oligomer that overlaps codons 91 to 98 and incorporates a BglII site downstream of codon 98. The amplification product was digested with NotI and BglII, ligated into similarly digested pACλcI32 and transformed into E. coli XL1-Blue (Stratagene).

Most of the plasmids containing motA with the desired mutation for use in complementation assays were derived from pNW143 by inverse PCR using mutagenic oligomers. Amplification products were then digested with DpnI to remove template DNA and transformed into E. coli XL1-Blue. To generate the plasmids containing the Q76R and S80R mutants, the mutant motA genes were amplified using PCR and the corresponding pET30b MotA mutant expression vectors (see below), digested with SfcI, and then ligated with pNW143 that had been digested with SfcI. For the plasmids needed in the 2-hybrid assays, the N-terminus of each mutant motA (codons 1 – 98) was amplified by PCR using either the pNW143 mutant plasmids or the pET30b MotA mutant derivatives as template and cloned into pACλcI32 as described for pMWT.

Buffers

Protein buffer contained 21 mM Tris-Cl (pH 7.9), 28 mM Tris-acetate (pH 7.9), 62 mM NaCl, 20% glycerol, 0.42 mM EDTA, 0.14 mM DTT, 0.003% Triton X-100, 62 mM immidazole, 103 mM potassium glutamate, 2.8 mM magnesium acetate, and 69 μg/ml BSA. DNA/NTP buffer contained 2.4 mM Tris-Cl (pH 7.9), 57 mM Tris-acetate (pH 7.9), 48 mM potassium phosphate (pH 6.5), 12% glycerol, 0.62 mM EDTA, 0.23 mM EGTA, 0.38 mM DTT, 214 mM potassium acetate, 5.7 mM magnesium acetate, 140 μg/ml BSA, 0.48 mM each ATP, GTP, and CTP, and 12 μM [α-³²P]UTP (5 × 10⁴ dpm/pmol). MotA buffer contained 200mM potassium phosphate (pH 6.5), 1mM DTT, 1mM benzamidine HCl, 1mM EDTA, 1mM EGTA, and 50% glycerol.

Expression and Purification of Proteins

The purification of σ⁷⁰ has been described (Gerber & Hinton, 1996). AsiA with a C-terminal his₆-tag (AsiA-his₆), for use in in vitro transcription reactions, was obtained as described (Pineda et al., 2004). E. coli RNA polymerase core was purchased from Epicentre Technologies. Full-length MotA was overexpressed in E. coli BL21-SI from the pET30b vector (Novagen) in the presence of 1 mM IPTG and 250 mM NaCl as previously described (Lambert et al., 2001). Cells were lysed by French Press in 50 mM Tris-HCl, 500 mM NaCl, 10 mM DTT, 10 mM Na₂EDTA, 10 mM benzamidine-HCl, insoluble material removed by centrifugation and nucleic acids precipitated by addition of polyethyleneimine to a final concentration of 0.7% (v/v). The protein fraction from the whole cell lysate was recovered from the polyethyleneimine solution by addition of ammonium sulfate to 85% saturation at 4 °C. Following resuspension and clarification of the ammonium sulfate precipitate in 50 mM Tris-HCl (pH 8), 1 mM benzamidine-HCl, 1 mM DTT, 0.1 M KCl, the protein solution was fractionated by phosphocellulose chromatography using a 1500 ml gradient from 100 mM to 900 mM KCl. Fractions containing MotA were purified to homogeneity by cation exchange chromatography on MonoS (G. E. Health Care) in 50 mM potassium phosphate (pH 6.5), 1 mM DTT, 1 mM benzamidine HCl, 1 mM EDTA, 1 mM EGTA using a 500 ml gradient from 75 mM to 500 mM NaCl. NMR samples were buffer exchanged into 200 mM potassium phosphate (pH 6.5), 1 mM DTT, 1 mM benzamidine HCl, 1 mM EDTA, 1 mM EGTA by dialysis. Preparation of AsiA and σ4 and formation of the AsiA/σ4 complex was performed as previously described (Lambert et al., 2001). The ternary complex of MotA with AsiA/σ4 was formed by mixing purified MotA and AsiA/σ4 complexes in a molar ratio of 1:1.2 in the above-mentioned NMR buffer.

Peptide Synthesis and Purification

Peptide (with and without a N-terminal Fluorescein 5-isothiocyanate) corresponding to residues 601-613 of σ4 (σ4-H5) was synthesized by standard solid-phase Fmoc chemistry, by the Rockefeller University Proteomics Resource Center, and purified by reverse phase HPLC using a semi-preparative Vydac C4 column. Purity and homogeneity of the peptide was validated by MALDI-TOF mass spectrometry

Two-hybrid assays

β-galactosidase assays were performed as described previously (Pal et al., 2003) using cultures of E. coli KS1 (Dove et al., 1997) containing the indicated plasmids except that cultures were grown for 3 hr in LB media supplemented with the appropriate antibiotics at the following concentrations: carbenicillin (50 ug/ml), chloramphenicol (25 ug/ml), and kanamycin (50 ug/ml). Relative β-galactosidase activities at a 100 μM IPTG concentration were calculated as follows: (Miller units with the MotA mutant – Miller units with the pACλcI32 control)/ (Miller units with pMWT – Miller units with the pACλcI32 control). β-galactosidase values represent the average of four assays.

In vitro transcription

Transcription reactions were assembled by adding a solution (2.9 μl) containing 0.2 pmol σ⁷⁰, 0.05 pmol core, and, as indicated, 3 pmol AsiA-his₆, in protein buffer to a solution (2.1 μl) containing 0.01 pmol linearized pDKT90 DNA and as indicated, 1 pmol of MotA or MotA mutant protein, in DNA/NTP buffer. (σ⁷⁰ was previously incubated with AsiA-his₆ at 37° C for 10 min to insure formation of the σ⁷⁰/AsiA-his₆complex and then incubated with core at 37° C for 15 min to form the AsiA-associated polymerase.) Transcription reactions were incubated at 37° C for 20 sec, rifampicin (0.5 μl of 300 ng/ml) was added, and the reactions were incubated for an additional 7 min before being collected on dry ice. Gel load solution (25 μl; 1 X TBE, 7 M urea, 0.1% bromophenol blue, 0.1% xylene cyanol FF) was added and the solution heated at 95° C for two minutes before electrophoresis on 4% polyacrylamide, 7 M urea denaturing gels run in 1/2 X TBE. After autoradiography, films were scanned using a Powerlook 2100XL densitometer and quantification was performed using Quantity One software from Bio-Rad, Inc.

DNA retardation assay

DNA binding reactions (10 μl) contained 0.04 pmol (1 × 10⁵ dpm) of a 5′-³²P labeled 21 bp fragment [5′ GAATTATTTGCTTTAGATTAC 3′] with a MotA box sequence (underlined), 40 mM Tris-acetate (pH 7.9), 20 mM Tris-Cl (pH 7.5), 20 mM potassium phosphate (pH 6.5), 5 % glycerol, 150 mM potassium glutamate, 50 mM NaCl, 4 mM magnesium acetate, 2 mM magnesium chloride, 0.2 mM EDTA, 0.1 mM EGTA, 0.2 mM DTT, 100 mg/ml BSA, and the indicated amount of wild type MotA or MotA mutant protein. Reactions were incubated at 37° C for 5 min and collected on ice. Reactions were then subjected to electrophoresis in a native, 12% acrylamide gel run in 1 X TBE at 4° C for 2.5 to 3 hr at a constant voltage of 450 volts. After autoradiography, films were scanned as described above. K_d(app) for wild type MotA was determined by generating a standard curve and then determining the amount of protein needed to bind 50% of the DNA as described (Sharma et al., 1999). K_d(app)'s for the mutant MotA proteins were then determined relative to the wild type.

Fluorescence Anisotropy

All fluorescence anisotropy measurements were carried out using an Olis RMF1000 Spectrofluoremeter equipped with a 450 watt xenon arc lamp and a jacketed cell holder for temperature control. Experiments were acquired at 20° C in 200 mM potassium phosphate (pH 6.5), 1 mM dithiothreitol, 1 mM benzamidine HCl, 1 mM EDTA, and 1 mM EGTA using equilibration times of 5 minutes for each data point sampled. Polarizer alignment was calibrated before each experiment with the aid of a 34% solution of Ludol in water. The scattered light was considered to be completely polarized when the anisotropy (A) values were ≥ 0.94. The fluorescein 5-isothiocyanate N-terminal tag attached to the H5 peptide was excited using a wavelength of 494 nm and fluorescence emission monitored at 520 nm.

A typical experiment was carried out using a 600 μM solution of peptide (150 μl) to which incremental amounts of protein was added. Protein increments were added, the solution mixed gently with a pipet and the anisotropy measured after equilibration. The data was analyzed by fitting the measured anisotropy assuming a two-state equilibrium:

$$L+R\leftrightarrow \text{LR}$$ (1)

Where the corresponding association constant (K_a) is given by

$$K_a=[\text{LR}]/([L][R])$$ (2)

L is the ligand, R is the receptor and LR the ligand-receptor complex. Considering that [L] = [L_T]-[LR] and [R] = [R_T]-[LR], ([L_T] and [R_T] are total ligand and receptor concentrations respectively), equation (2) can be expressed as,

$$K_a=[\text{LR}]/(([L_T]-[\text{LR}])([R_T]-[\text{LR}]))$$ (3)

Equation (3) can be rearranged to the quadratic equation 0 = ax² + bx + c, (a = K_a, b = -(1+K_a[L_T]+K_a[R_T]), c = K_a[L_T][R_T] and x = [LR]) and the solution divided by [R_T] to yield:

$$[\text{LR}]/[R_T]=(1+K_a[L_T]+K_a[R_T]-{(1+K_a[L_T]+K_a{[R_T]}^2-4K_a^2[R_T][L_T])}^{1/2})/(2K_a[R_T])$$ (4)

Plotting fractional saturation ([LR]/[R_T]), which is directly proportional to the measured fluorescence anisotropy, versus the concentration of added ligand, [L_T], a non-linear least-squares fit of the data can be performed and the dissociation constant (K_d=1/K_a) determined.

NMR Spectroscopy

NMR experiments were acquired at 20° or 36° C on a Bruker DMX500 or DMX600 equipped with a z-shielded gradient triple-resonance probe and a triple-resonance cryoprobe respectively. Backbone and sidechain assignments of MotA alone and in the ternary MotA/AsiA/σ4 complex were obtained using standard multi-nuclear, multi-dimensional NMR spectroscopy. The changes in N and HN chemical shifts which defined the ‘footprint’ were calculated according to the following equation: Δ(ppm)=[(N_bound-N_free)²+(H_Nbound-H_Nfree)²]^1/2.

Complementation

Wild-type T4D and T4D amG1(motA^-) (Mattson et al., 1978) were used to infect E. coli TabG cells containing plasmids with wild-type motA (pNW143), mutant motA genes derived from pNW143, or vector alone without motA (pNW129). Bacterial cells were grown to ∼2 × 10⁸ cells/ml in LB broth supplemented with 50 ug/ml kanamycin. ∼4 × 10⁷ cells were plated, infected with 5 μl of 10-fold serially diluted phage stocks containing 2 to 2000 pfu/μl and incubated overnight at 25° C. SDS-PAGE of protein extracts obtained after 3 hours of induction with 0.2% arabinose indicated that each of the mutant proteins were expressed at the same level as wild type.