|Home | About | Journals | Submit | Contact Us | Français|
The lysogeny promoting protein CII from bacteriophage 186 is a potent transcriptional activator, capable of mediating at least a 400-fold increase in transcription over basal activity. Despite being functionally similar to its counterpart in phage λ, it shows no homology at the level of protein sequence and does not belong to any known family of transcriptional activators. It also has the unusual property of binding DNA half-sites that are separated by 20 base pairs, center to center. Here we investigate the structural and functional properties of CII using a combination of genetics, in vitro assays, and mutational analysis. We find that 186 CII possesses two functional domains, with an independent activation epitope in each. 186 CII owes its potent activity to activation mechanisms that are dependent on both the σ70 and α C-terminal domain (αCTD) components of RNA polymerase, contacting different functional domains. We also present evidence that like λ CII, 186 CII is proteolytically degraded in vivo, but unlike λ CII, 186 CII proteolysis results in a specific, transcriptionally inactive, degradation product with altered self-association properties.
Temperate bacteriophage have proven to be fruitful systems for the study of prokaryotic transcription, gene regulatory networks, and biological decision making. Many seminal contributions to these fields have been based on studies using the model system of the λ coliphage (1). However, investigation of λ only gives us insight into one instance of a solution to the various biological problems that a bacteriophage must solve. One way of working toward an understanding of the general principles governing these problems and their solutions is through the study of functionally similar, yet evolutionarily divergent systems.
One such system is the temperate coliphage 186, a member of the P2-like family, which has a similar network topology to that of λ, but no detectable sequence homology (2). Here we examine the CII protein of 186, which like λ CII, is a potent transcriptional activator and is essential for the establishment of the lysogenic life cycle of the phage (3,–5). 186 CII promotes establishment of lysogeny by activation of the pE promoter, which is one of two promoters that drive expression of the maintenance repressor CI (Fig. 1). Activation of pE is dependent only on CII and RNAP, and is necessary for establishment of lysogeny, but not its maintenance (6). CII protein itself is produced upon infection via the early lytic transcript under the control of pR. This results in an apparent paradox in which expression of the lytic transcript promotes lysogeny via the activity of CII. The switch region of λ yields a similar situation, which is resolved by the fact that λ CII is rapidly degraded by the host protease FtsH, such that λ CII only accumulates to the level necessary for establishment of lysogeny in a small fraction of infections (7,–10). Whether proteolysis of 186 CII plays a similar role in the life cycle decision of bacteriophage 186 remains an open question.
The CII protein is 169 amino acids in length, with a molecular mass of 18.7 kDa and a predicted helix-turn-helix motif in the N-terminal region (4, 11). The helix-turn-helix motif is presumed to be responsible for binding of CII to two inverted repeat 7-mer half-sites of the sequence ATGTTTG. 186 CII binds as a pre-formed dimer to the half-sites, the centers of which are 20 bp apart (Fig. 1) and upon binding induces moderate DNA bending of 40–45° (6). Binding of CII and its ability to activate the pE promoter is sensitive both to small changes in the spacing of the half-sites as well as larger changes in spacing, which preserve helical phasing, suggesting that the geometry of the dimer is relatively rigid (6).
The unusual arrangement of the CII binding sites at pE is not only distinct from that of λ CII, which recognizes direct repeats in close proximity to each other, but is highly unusual in the broader context of prokaryotic transcriptional activators. Few examples of activators with similarly wide spacing between half-sites are known, which poses a question as to whether CII activates transcription via novel mechanisms. The mechanism of CII activity is made even more interesting by its potency; 186 CII is capable of ~400-fold activation of pE (4). Previous work has shown that 186 CII acts via recruitment of RNAP to the promoter, but the way in which this is achieved remains unknown (6).
Here we demonstrate that like λ CII, the CII protein of 186 is rapidly degraded in vivo. We investigate the domain organization and the transcriptional activation mechanism of CII using genetic screens to isolate various classes of mutations, and then characterizing how these mutations exert their effects using genetic and biochemical methods. We find that proteolytic degradation of CII compromises the multimerization and thus indirectly the DNA binding of CII by removing part of a C-terminal self-association domain. We show that transcriptional activation by CII occurs via two independent mechanisms. The first is dependent upon the C-terminal domain of the RNAP2 α subunit (αCTD) and is conferred by an epitope in the C-terminal domain of CII. Second, our results suggest a contact between the N-terminal domain of CII and region 4 of σ70.
Given the prevalence and importance of various truncations of CII in this study, for the sake of clarity we will hereafter refer to wild-type CII as CII169, denoting its length of 169 amino acids, and refer to other truncations using the same convention.
A summary of all oligonucleotides used in this study is provided in Table 1. Oligonucleotides were purchased from Geneworks (Adelaide, Australia).
A summary of all plasmids (12) used in this study is provided in Table 2. Reporters were constructed using a plasmid integration system developed from the CRIM plasmids (13, 14), incorporating promoter fragments between KpnI and XbaI restriction sites, followed by an RNaseIII cleavage site (15) and a lacO2− lacZ reporter gene (16). The pE (−120 to +115) reporter fragment was PCR amplified using primers 168 and 169, and the pIT3CL-pE-LacZ was integrated into the host chromosome at the λ attB site. The same pE promoter fragment and cloning strategy was used to construct the matching pBC1-pE in vitro transcription template plasmid. The pCIIR reporter pIT3CH-pCIIR-LacZ was integrated into the host chromosome at the HK022 attB site. The pCIIR sequence was CTGGTACCATGTTTGATTTTCATATTACCAAACATTGAATGTTTGATTATCATTTATCCAAACATTGAGAACTTCTAGA (CII half-sites marked in bold, −35 hexamer in underline, −10 in double underline). PCR was used to screen for correct single-copy integrants.
pZS45 follows the modular cloning strategy and nomenclature of the pZ vector system (17), with the small XhoI/AvrII fragment of pZE15 (18) inserted between those sites. pZS45-CII169 was constructed by PCR amplification of the RBS and cII ORF from pKES1 using primers 164 and T7term, and insertion between the HindIII and BamHI sites of pZS45. pET15b-CII169 was constructed by ligating the NdeI/BamHI insert of pKES1 into the pET15b vector (Novagen). Plasmid pZE1-pE-CI-CTD encodes the pE promoter (−133 to +45, amplified by PCR with primers 735 and 736) between the AatII and XhoI sites, the pET RBS from pZS45-CII169 between the HindIII/NdeI sites, and the 186 CI residues 83–192 between the NdeI/BamHI sites. pHTf1-rpoA(235), encoding α235, was created by introduction of a stop codon after residue 235 by QuikChange mutagenesis of pHTf1-rpoA with primers 1239 and 1240.
Strains used for reporter assays and the genetic screen were all derivatives of Escherichia coli strain BW25113 (F-Δ(araD-araB)567ΔlacZ4787(::rrnB-3)λ-rph-1Δ(rhaD-rhaB)568 hsdR514; CGSC# 7636) (19), with the exception of PN376, which is a derivative of E. coli strain MC1061 (F-Δ(araA-leu)7697[araD139]B/r, Δ(codB-lacI)3, galK16, galE15(GalS) λ-e14-mcrA0 relA1 rpsL150(strR) spoT1 mcrB1 hsdR2; CGSC number 6649) and has been previously described as MC1061.5(λPN68) (4). Strain IM18, used for the genetic screen, carried pUHA1, pZE1-pE-CI-CTD, an integrated pCIIR reporter, and a lysogen of 186 cIts that was verified to kill its host by temperature induction. Reporter strains IM13 and IM26 carried pUHA1 and integrated pITCH-pCIIR-lacZ and pIT3CL-pE-lacZ reporters, respectively. Overexpression of 186 CII and derivatives for the purposes of protein purification was carried out using HMS174(λDE3) (F-, λ-, recA1, IN(rrnD-rrnE)1, rph-1, rpoB331(rifR), hsdR19; CGSC number 6576).
The target plasmid pZS45-CII169 was mutagenized using the GPS-LS transposon-based linker scanning mutagenesis kit (New England Biolabs). The terminated transposition reaction was transformed into electrocompetent DH5α cells and plated onto selective media. Insertions were identified by colony PCR using the T7prom primer, with the supplied N and S transprimers. The protocol was completed by PmeI digest and re-ligation of plasmids carrying insertions, resulting in a 15-bp scar at the insertion site. Truncations generated by this method encode a C-terminal valine residue following the last wild-type CII residue.
CII145 was amplified from pZS45-CII145 by high fidelity PCR using primers 289 and RSP (see Table 1 for sequences). The resulting PCR product was used as template for error-prone PCR with Taq DNA polymerase (New England Biolabs) and MgCl2 supplemented to a total concentration of 6 mm. The product of error-prone PCR was subcloned back into pZS45-CII145 via BamHI and XhoI restriction sites and transformed into screening strain IM18 by electroporation. Transformants were plated onto TB agar supplemented with 100 μg ml−1 of ampicillin, 50 μg ml−1 of kanamycin, 50 μg ml−1 of spectinomycin, 40 μg ml−1 of 5-bromo-4-chloro-indolyl-β-d-galactopyranoside (X-gal), 200 μm ITPG and incubated at 34 °C for 24 h, followed by 4 °C for 2–3 days to allow color development, before selection of healthy white colonies.
Kinetic LacZ assays in a 96-well microtiter plate were performed as previously described (20).
BW25113 or PN376 cells harboring pZS45-CII169 or pZS45-CII145 were grown in LB supplemented with 100 μm IPTG at 37 °C to mid-log phase. Cells were lysed using B-PER (Pierce) and Benzonase (0.25 units μl−1) (Novagen) and cell extracts were run on NuPAGE BisTris (12 or 4–12%) or Novex 12.5% Tricine gels (Invitrogen) in 1× NuPAGE MES SDS Running Buffer or Novex Tricine SDS Running Buffer, respectively (Invitrogen). Gels were blotted using an iBlot and Gel Transfer Stacks PVDF (Invitrogen) or the Novex wet transfer apparatus onto Hybond-LFP PVDF membrane (GE Healthcare). Membranes were blocked using 5% BSA. CII and λ CI primary detection was with rabbit antisera (IMVS Veterinary Services). Secondary detection used goat anti-rabbit IgG Cy5-labeled ECL plex secondary antibody (GE Healthcare). Membranes were scanned using a Typhoon Trio (GE Healthcare) and images analyzed by ImageQuant (GE Healthcare).
BW25113 cells harboring pZS45-CII169 or pZS45-CII145 were grown in LB supplemented with 50 μg ml−1 of spectinomycin and 100 μm IPTG at 37 °C to mid-log phase. Ten minutes prior to the assay start 50 mm MgSO4 was added to prevent chloramphenicol-induced cell lysis (21). Chloramphenicol (final concentration 100 μg ml−1) was added to inhibit translation, 1 min prior to the assay start. A reference sample was taken at time 0, and at various time points afterward, as indicated. Samples were taken by mixing 1 ml of culture with 5 ml of ice-cold phosphate-buffered saline (PBS). Samples were analyzed by Western blot as described above. Statistical analysis used Prism (GraphPad), fitting a simple one-phase exponential decay, with constraints Y0 = 100 and plateau = 0.
HMS174(λDE3) cells (22) carrying pLysS and pET15b-CII169 or pET15b-CII145 were diluted 1/100 from an overnight culture and grown at 37 °C to an A600 of 0.6–0.8 before addition of IPTG to 0.2 mm. Cultures were incubated a further 3–4 h before harvesting cells by centrifugation at 6,000 × g for 15 min. Cell pellets were stored at −20 °C for up to 1 month. For purification, cell pellets were resuspended in 30 ml of TEG buffer (25 mm Tris, 1 mm EDTA, 10% glycerol, 150 mm NaCl, pH 7.5) supplemented with 1 mm PMSF and 5 mm imidazole, and lysed mechanically using a Microfluidizer (Microfluidics). Lysate was cleared by centrifugation (25,000 × g, 20 min at 4 °C) before purification by nickel-affinity chromatography using washes with TEG buffer supplemented with 20 mm imidazole and 100 mm imidazole, before elution with TEG + 250 mm imidazole, 500 mm NaCl. Protein purity was verified by SDS-PAGE, and purified protein was dialyzed against two changes of buffer (25 mm Tris, 5 mm EDTA, 10% glycerol, 150 mm NaCl, pH 7.5) and stored at −80 °C.
Expression, purification, and reconstitution of RNAP was carried out as described elsewhere (23), using denaturing conditions for the preparation of α329 and α235.
In vitro transcription assays were carried out based on the method previously described (24). RNAP (100 nm final concentration) and CII169 or CII145 at concentrations ranging from 64 pm to 1 μm were incubated with supercoiled pBC1-pE template (2 nm) for 30 min at 37 °C in transcription buffer (20 mm Tris acetate (pH 8.0), 3 mm magnesium acetate, 200 mm potassium glutamate, 1 mm DDT, 5% glycerol, and 0.3 units μl−1 of RNase inhibitor) to allow open complex formation. Transcription was initiated by addition of rNTP/heparin solution (final concentrations of 0.5 mg ml−1 of heparin, 0.2 mm rATP, rCTP, and rGTP each, 0.02 mm rUTP, 0.5 mCi ml−1 of [α-32P]UTP), and transcription was allowed to proceed for 10 min at 37 °C before termination by addition of an equal volume of loading buffer (90% formamide, 10 mm EDTA). 8-μl samples were analyzed by electrophoresis on a 6% polyacrylamide gel containing 8 m urea. Gels were dried and exposed to a PhosphorImager screen overnight. Images were analyzed using ImageJ software (25).
BW25113 cells harboring the low copy pZS45-CII169 and pUHA-1 plasmids were grown at 37 °C to A600 0.8 in 2 liters of LB media containing spectinomycin (50 μg/ml), kanamycin (50 μg/ml), and IPTG (100 μm), harvested by centrifugation, resuspended in TEG150, lysed, and clarified (as above). Ammonium sulfate precipitation, dialysis, and heparin affinity chromatography were carried out as previously described (4), except that stepwise elution using TEG buffer containing 50 mm NaCl (TEG50) and 100 mm NaCl (TEG100) was employed. CII-containing fractions were identified by Western blot analysis. CII degradation product, but not detectable CII169, was identified in TEG50 elution fractions, which were pooled, dialyzed into water, lyophilized, and stored at −20 °C.
Mass spectrometry analysis was conducted by the Adelaide Proteomics Centre. Lyophilized protein was separated by SDS-PAGE, the region around 15.5 kDa was excised, washed in 500 μl of 100 mm ammonium bicarbonate, destained, reduced with dithiothreitol, and alkylated with iodoacetamide before being treated overnight with 200 ng of Asp-N (Promega, catalog number V1621) in 10 mm ammonium bicarbonate containing 10% acetonitrile. The reaction was halted with 30 μl of 1% formic acid, sonicated for 15 min, and buffer was removed. Two further trifluoroacetic acid/acetonitrile extractions were performed. Extracts were pooled and separated on a HPLC system (Thermo Scientific) using a separation column (Thermo Scientific) (Acclaim PepMap RSLC, 75 μm diameter × 15 cm length) and a trapping column (Thermo Scientific) (Acclaim PepMap100, 75 μm diameter × 2 cm length). The HPLC system was coupled to a LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Berman, Germany). The 6 most intense peptide ions from each scan with charge states ≥2 and minimum signal intensity of 1000 were sequentially isolated and fragmented in the high-pressure linear ion trap by low-energy CID.
Size exclusion chromatography was carried out using a SuperdexTM 200 10/300 GL column (GE Healthcare) equilibrated in PBS. MALS analysis used in-line detection with Wyatt MiniDawn Treos and Opti-lab rEx instruments, and analysis was conducted using Astra version 5 software (Wyatt Technology). Samples were 2 mg/ml, using injection volumes of 100–300 μl.
Structure predictions for the N-terminal domain generated 48,000 decoys using the abrelax protocol of Rosetta 3.1 (26) followed by clustering of the 500 lowest energy structures to identify favorable models and assess convergence. Docking of the N-terminal domain to DNA used the HADDOCK webserver (27, 28). UCSF Chimera (29) was used for structure depiction, analysis, and preparation of figures.
We used transposon-mediated linker insertion mutagenesis to investigate the functional organization of the CII protein. Linker insertion mutagenesis creates 15-bp insertions at random locations within the DNA sequence of interest, in this case a plasmid (pZS45-CII169) encoding CII169. Due to the sequence of this insert and the fact that it inserts independently of the reading frame, it can give rise to mutants possessing either a 5-amino acid translation insertion (“insertion mutants”), or a premature stop codon (“truncation mutants”). Insertion mutations in regions of structural or functional importance (such as solvent-excluded sites, sites of intermolecular interaction, and some regions of secondary structure) will result in substantially reduced function of the protein. In contrast, insertion mutations that retain substantial activity can be expected to lie in solvent-exposed regions, without direct functional importance or significant secondary structure. Thus previous work using similar techniques have shown that the activity of insertion mutants is correlated with insertions that lie within inter-domain linkers, or solvent-exposed loops (30,–33).
We isolated and sequenced 26 independent mutants via this method, comprising 10 truncation mutants, and 16 insertion mutants (Fig. 2A). Only four of these mutants retained at least 50% wild-type activity, assayed using a pE lacZ in vivo reporter. One of these encoded an insertion after residue Leu-87, whereas two independent insertions after nearby residue Cys-81 retain moderate activity, suggesting that this region is an inter-domain linker. This is in accordance with bioinformatic predictions that CII169 is composed of two structural domains with a boundary near residue Cys-81 using the Ginzu algorithm (34).
The other high activity mutants lie near the C terminus of the protein, insertions after Ser-153 and Phe-166, and a truncation following Ser-160. It is perhaps unsurprising that these mutations suggest that this C-terminal region is unimportant for structure and function, and can be discarded or disrupted without losing function. Unexpectedly, however, two of these were gain of function mutations, causing higher expression of the pE reporter. Observing that loss or interruption of C-terminal residues increased CII in vivo activity, we assayed the activity of a series of truncation mutants between 141 and 166 amino acids in length (Fig. 2B). Higher activity was correlated with shorter CII protein length, until activity dropped sharply with truncation to 141 amino acids, likely due to loss of structure and function of the C-terminal domain.
We also observed that truncations on the C-terminal side of the putative domain boundary around amino acid 81 retained low level activity (5–10% of wild-type), whereas those lying on the N-terminal side were inactive (<2% of wild-type activity) (Fig. 2C). These observations are consistent with the prediction of an N-terminal helix-turn-helix DNA binding domain (4, 11). The low activity of C-terminal truncations at or before residue 141 could indicate a loss of an independent DNA-binding function, loss of transcriptional activation, or a loss of self-association, leading to reduced DNA binding due to loss of cooperativity.
Given its functional similarity to λ CII, we hypothesized that 186 CII may also be proteolytically degraded in E. coli. The C terminally truncated gain of function mutations (Fig. 2, B and C) could then be explained by loss or disruption of the proteolysis signal. Examining the in vivo stability of CII169 by Western blot following addition of a translation inhibitor reveals a calculated half-life of 2.6 min (2.3 ≤ t½ ≤ 3.0 min, 95% CI)(Fig. 3A), which is similar to the half-life of λ CII (7, 8, 35). Western blot analysis also revealed that a specific degradation product is evident for CII169 (Fig. 3A, inset), which is suggestive of proteolytic degradation of CII169 by a site-specific protease.
Consistent with a loss of proteolytic degradation in the CII145 mutant, we observed higher in vivo activity, and saturation of activity at lower levels of induction of CII145, compared with CII169 (Fig. 3B). Indeed, CII145 has a measured half-life of ~38 min (31 ≤ t½ ≤ 49 min, 95% CI)(Fig. 3A), shows higher equilibrium cellular concentrations as detected by Western blot, and shows no evidence of any specific degradation product (Fig. 3A, inset). In vitro transcription assays using purified CII145 and CII169 with N-terminal His6 tags show no statistically significant difference in the specific activities of the two proteins (Fig. 3C), demonstrating that higher in vivo activity is due only to increased cellular concentrations, not a change in the way CII145 activates transcription.
We reasoned that the C-terminal portion of CII169 is likely responsible for recruitment of a protease. To test our hypothesis, we created a translational fusion of the proteolytically stable protein λ CI (36), with 186 CII residues 146–169 (λCI-fusion), and compared its stability to wild-type λ CI using translation stop experiments and Western blot analysis probing for λ CI (Fig. 3D). As expected, no degradation of wild-type λ CI is evident within the duration of the experiment (29 min). In contrast, the λ CI fusion protein is substantially degraded within 15 min, and almost completely depleted by 29 min. Thus we conclude that the last 24 amino acids of 186 CII169 constitute a degradation signal, which is sufficient to destabilize an otherwise stable protein. Interestingly, no specific degradation product is evident for the λ CI fusion protein.
To complement our structural insights, we set out to characterize which regions of 186 CII are responsible for its activity, and investigate the mechanism of transcriptional activation by CII via a mutational screen for loss of function (Fig. 4A). To avoid confounding our analysis with gain of function mutations that increase CII169 stability, we conducted our genetic screen using the stabilized CII145. This approach also allowed higher cellular concentrations to be achieved, leading to a greater measurement window in our selection assays. Because the specific activities of CII145 and CII169 are equal (Fig. 3C), we could be confident that the activation mechanisms are shared between the two forms. A simple genetic screen for mutations that abrogate CII transcriptional activity would yield many mutations that cause loss of function via loss of structure and/or DNA binding, rather than loss of transcriptional activation. We reasoned that a dual selection screen, in which we select for loss of pE transcriptional activity, but against loss of DNA binding, would yield greater insight.
To achieve this, we designed a synthetic promoter in which two pairs of CII recognition half-sites straddled the −10 and −35 σ70 binding sites, anticipating that binding of CII at this promoter would compete with RNA polymerase and cause repression of promoter activity (Fig. 4B). Degenerate bases in the −35 site generated a library of promoters with varying activities. Repression of the promoter should occur by competition, so we preferred promoters with a weaker affinity for σ70, expecting they would be more effectively repressed. The basal activities of 52 members of the library were assayed on a multiple copy plasmid. Of these, the 7 promoters with lowest activity were integrated into the host chromosome in single copy and assayed in the presence of IPTG-inducible CII169, supplied from pZS45-CII169 to test for repression by CII169. All 7 promoters demonstrated dose-dependent repression by CII169 (Fig. 4B). A weak negative correlation was observed between the extent of repression by CII169 and the basal activity of the promoters, consistent with our expectations. Promoter clone 19 (hereafter referred to as pCIIR) was selected for use in our genetic screen and sequenced (see “Experimental Procedures” for full sequence). Our assays show that pCIIR was repressed by almost 80% by CII145, but only around 25% by CII169, indicating a substantial measurement window for detection of mutants with compromised DNA binding (Fig. 4C).
Selection for reduced pE activity was achieved by constructing a gene circuit in which strong transcription from pE brings about host cell death via induction of the lytic life cycle of a 186 cIts prophage. At permissive temperatures (≤34 °C) the lysogen is maintained by the CIts repressor. If induced by increased temperature or other means, the lytic phase of the prophage is induced, causing host cell death or severe growth impairment. Previous work (37, 38) has shown that expression of the 186 CI C-terminal domain (CI-CTD) to titrate away functional CIts at permissive temperatures can be used to induce the prophage and thus induce cell killing. In this case, we have expressed CI-CTD from the pE promoter on the plasmid pZE1-pE-CI-CTD to select against pE activation.
Selection for activation mutants was combined with our synthetic pCIIR lacZ reporter construct. Thus, isolation of surviving colonies exhibiting low β-galactosidase activity in a strain with a chromosomal copy of pCIIR driving expression of lacZ enabled screening for CII145 mutants with low transcriptional activity but near wild-type DNA binding.
From three independent rounds of error-prone PCR mutagenesis and selection, 73 isolates of ~1000 colonies were identified for more detailed analysis. The CII-encoding plasmids were purified from these isolates and transformed into two separate reporter strains (IM13 and IM26) for quantitative analysis of pE and pCIIR activities, respectively, by LacZ assay. Isolates exhibiting pCIIR activities less than 8.5 units at 0.2 mm IPTG were sequenced, yielding 12 unique mutations at 7 different amino acid positions (Fig. 4D). The chosen cut-off corresponds to 1.65 S.D. above the mean of pCIIR activity for wild-type CII145, and 95% confidence that repression is equal to or better than wild-type. The mutations form two clusters; one in the N-terminal DNA-binding region, and a second in the C-terminal region (Fig. 4E). Mutants at two positions in particular, Glu-46 and Arg-115, were isolated repeatedly and from independent mutagenesis reactions, suggesting that these may be critical residues from the two clusters. Examining this possibility, we used site-directed mutagenesis to construct the charge reversal R115E mutant, which was shown to have a greater defect in activation than other substitutions at this position (Fig. 4D). We combined R115E with mutations R17L, E46G, or E46K in the N-terminal domain cluster that had a large effect on pE activity. Fig. 4D summarizes pE and pCIIR activities for all mutants, both random and site-directed. Notably, combining the R115E and E46K mutations to create a CII145E46K/R115E double mutant reduced activity on the pE reporter ~4-fold, relative to the E46K single mutant (p < 0.05; unpaired t test). The CII145E46K/R115E exhibits no statistically significant activity on the pE reporter relative to an empty vector control, without significantly compromising DNA binding as assessed by pCIIR activity. Thus we can conclude that these two mutations are sufficient to disrupt any and all mechanisms by which CII activates transcription.
Having established that Glu-46 and Arg-115 are critical to transcriptional activation by CII145, we set out to determine how these residues contribute to the function of CII145. Due to the proximity of the CII binding site to the expected location of the pE −35 promoter element (Fig. 1), we hypothesized that CII145 contacts the σ70 subunit of RNA polymerase. Based on the mechanisms by which other factors promote transcription, we also hypothesized that CII145 may contact the αCTD of RNA polymerase. Mutants E46K and R115E were selected for this investigation on the basis that they exhibit the greatest defects at the two critical sites. The strong R17L mutant was also included because it lies some distance in the primary sequence from the key N-terminal site Glu-46, and so could form part of either the Arg-115 or Glu-46 epitope dependent upon the tertiary structure of CII.
Potential contacts between CII and αCTD were investigated by expression of full-length α (α329) or α235, a truncation of α at residue 235, which deletes the αCTD. It is important to note that because the chromosomal copy of the rpoA gene encoding the α subunit is essential, wild-type α is expressed in addition to the plasmid-encoded variant of α in all assays. With this experimental design, the pool of RNAPs in the cell comprises a mixture of those with 0, one or two plasmid-derived α subunits. Nonetheless, if expression of α235 reduces transcriptional activity relative to an α329 control, this suggests that αCTD is involved in activation at that promoter, making it a simple, useful screen for evidence of αCTD involvement.
Such a decrease in pE activity is evident for CII145 mutants R17L and E46K, but not R115E or wild-type CII145 (Fig. 5A), strongly suggesting that αCTD contributes to activation by the R17L and E46K mutants. No statistically significant decrease is evident for wild-type or R115E CII145, meaning that αCTD may or may not contribute to the activities of these variants, because we do not know the magnitude of the effect we should see in the context of potential redundancy of activation mechanisms, and the confounding effects of the chromosomal (wild-type) copy of rpoA. Given the role of αCTD in activation by the R17L and E46K mutations, it seems unlikely that αCTD is not involved in activation by wild-type CII145. However, the results offer the possibility that the R115E mutation abrogates the role of αCTD, warranting further investigation.
To examine the role of αCTD in wild-type and R115E CII145 more thoroughly, we used in vitro transcription (Fig. 5B). pBC2-pE was supplied as supercoiled template for in vitro transcription using either wild-type or R115E CII145, and RNA polymerase reconstituted with either α329 or α235. The plasmid template produces multiple transcripts, including one from pE as well as the RNA1 transcript, which is independent of CII and serves as an internal control. This analysis shows that transcription from pE using RNAP incorporating α329 is considerably lower in the presence of CII145 R115E than wild-type CII145 (Fig. 5B, lanes 3 and 5), consistent with the in vivo data. Importantly, however, there is no discernable difference in transcription between wild-type and R115E CII145 variants when transcription is driven by α235-containing RNAP (Fig. 5B, lanes 4 and 6). We also observe that although deletion of the αCTD results in a large decrease in transcription in the presence of wild-type CII145 (Fig. 5B, lanes 3 and 4), there is no similar reduction in transcription in the presence of R115E CII145 (Fig. 5B, lanes 5 and 6). This shows that αCTD-dependent activation of pE is completely lost in the R115E mutant. We conclude that CII145 activates transcription via the αCTD, likely due to a direct contact between αCTD and CII145. We cannot formally exclude the possibility that rather than disrupting a direct contact, the R115E mutation indirectly inhibits the role of the αCTD by, for instance, altering the CII145 structure or the conformation of the promoter DNA. However, our finding that CII145 R115E does not significantly alter DNA binding, as measured by pCIIR activity (Fig. 4D), suggests that substantial changes to the CII145 structure and DNA binding are unlikely. In addition, our in vitro transcription experiments demonstrate that CII145 possesses a second, αCTD-independent mechanism, as both wild-type and R115E CII145 retain partial activity in assays using RNAP that lacks any αCTD (Fig. 5B, lanes 4 and 6).
We used genetic analyses to investigate the possibility of contact between CII and σ70. We looked for contacts with domain 4 of σ70, specifically residues 591–600, which have previously been shown to contact other transcriptional activators (39,–41). For each selected mutant of CII, we expressed σ70 mutants of charged residues in this region and analyzed pE activity, looking for reduced wild-type CII145 activity, and genetic suppression of the transcriptional defect of our CII145 mutants (Fig. 5C). None of the tested σ70 variants resulted in a clear reduction in reporter expression due to CII145. However, the K593A variant of σ70 partially suppresses the activation defects of the R17L and E46K mutants, but not the R115E mutant of CII145 (Fig. 5C). In light of this result, we also tested the charge reversal K593E σ70 mutant. Expression of K593E σ70 yielded almost complete restoration of the activity of E46K and R17L CII145, but not the R115E variant (Fig. 5C). Genetic suppression of the R17L and E46K defects by changes in σ70 suggest that CII145 activates via a σ70-dependent mechanism, and that Lys-593 or nearby residues are important to that mechanism. However, it must be noted that as a corollary we would normally expect, but do not observe, a reduction in wild-type or R115E activity associated with the expression of Lys-593 σ70 variants. This discrepancy could be explained either by some form of redundancy in the mechanism, or by the mechanism relying on σ70 residues that were not investigated here (see ”Discussion“ for a more detailed examination).
To further characterize the structure and function of CII, we sought to identify the sequence of the CII169 proteolytic cleavage product. We have observed that over-expression of CII169 results in further accumulation of full-length protein, but not the degradation product, presumably due to saturation of the responsible protease (8). Thus, isolating the degradation product for direct characterization was problematic. Nonetheless, we were able to employ two approaches to this problem. Initially, we used Western blot analysis of our linker insertion mutants to deduce the location of proteolytic cleavage. CII169 variants containing a 5-amino acid insertion exhibit a higher apparent molecular weight after SDS-PAGE and Western blot analysis. Whether the same is true or not of the degradation product allows us to deduce whether the insertion lies within the degradation product, or is removed by proteolysis (Fig. 6A). We observe that insertions near the N terminus, such as that after residue 4 (4ins) are evident in the degradation product, whereas the converse is true of an insertion near the C terminus after residue 166 (166ins), demonstrating that proteolysis is occurring at the C-terminal end of CII169. The presence of the insertion in the degradation product is evident up to and including an insertion after residue 131, allowing us to conclude that the degradation product comprises at least the first 132 residues of CII169. We also observe that the degradation product has an apparent molecular weight lower than that of the CII142 truncation mutant. Thus we deduce that the C-terminal end of the degradation product lies between residues 132 and 142.
To locate the proteolytic cleavage site more precisely, we used mass spectrometry. Briefly, CII169 was expressed from the low copy number plasmid pZS45-CII169, enriched by heparin affinity chromatography (4), and separated by SDS-PAGE. A faint band as visualized by Coomassie stain was observed migrating at a position consistent with the degradation product, which was excised and treated with Asp-N protease. The resulting peptide fragments were identified by LC-MS/MS (Fig. 6B). 63% sequence coverage was achieved. Asp-N protease has a high specificity for cleavage on the N-terminal side of aspartate and, at a lower rate, glutamate residues (42). In vivo proteolysis of CII169 is expected to yield peptide fragments with C-terminal sequences that do not conform to this pattern, because there are no aspartate or glutamate residues within or adjacent to the region of proteolysis (residues 132–142). Twelve such peptide fragments were observed, and are highlighted by green underlining in Fig. 6B. Fifty percent of these share a common N terminus, with some variation in the precise location of the C terminus. The longest of these end at Leu-135, which is within the region already identified as the C terminus of the degradation product. We interpret this collection of peptides as indicating that the CII169 degradation product that is detected in cellular extracts comprises the first 135 residues of CII169 (i.e. it is equivalent to a CII135 truncation), or a mixture of CII133, CII134, and CII135 forms, and that this form undergoes further proteolytic degradation from the C terminus. Cloning and expression of a CII135 truncation demonstrated that this truncation is found at the same apparent molecular weight as the degradation product after SDS-PAGE (Fig. 7A).
Having determined that the in vivo degradation product of CII169 corresponds to a CII135 truncation, we sought to characterize this truncation. Given the relative abundance of the degradation product in vivo, the activity and function of the CII135 form has important implications for the physiological role of 186 CII, but equally could provide further insights into the structure and function of the full-length CII169 protein. Reporter assays using pCIIR show that CII135 exhibits significantly reduced DNA binding, despite expressing at higher cellular concentrations relative to both CII169 and CII145 (Fig. 7A). Correspondingly, in vivo pE reporter assays show that CII135 exhibits some 97% less activity at maximum induction than CII169 (Fig. 7B). Nonetheless, some residual activity is still clearly detectable at higher levels of protein expression (Fig. 7B, inset), consistent with other truncations within the putative self-association domain isolated by linker insertion mutagenesis.
Thus, we expect that the primary mechanism by which truncations within the C-terminal domain lose activity is by loss of self-association and cooperativity in DNA binding, or by direct loss of an independent DNA-binding function. Addressing this question, we mutated residues Val-36 and Gln-37 of CII169. These amino acids are at positions 12 and 13 of the helix-turn-helix according to the positioning scheme adopted by the motif prediction algorithm (11), which contributes strongly to sequence specificity (43, 44), but tolerate a diversity of amino acids. In an effort to minimize the chances of causing structural defects, we mutated Val-36 and Gln-37 to other amino acids that are commonly found at positions 12 and 13 of helix-turn-helix proteins, resulting in the CII169V36E/Q37S mutant. This mutant does not exhibit any detectable activity at pE (Fig. 7B), and does not repress pCIIR, despite being observable in cellular extracts by Western blot (Fig. 7A). These results suggest that there is no significant DNA-binding function encoded by regions of the protein outside of the helix-turn-helix motif.
Examining the loss of self-association hypothesis directly, we used size-exclusion chromatography coupled with multiangle light scattering analysis (SEC-MALS) to determine the native, solution state molecular weights of purified His-tagged CII135, CII145, and CII169 (Fig. 7C). The observed peak molecular masses of CII169 (86.0 ± 3.4 kDa; 95% CI) and CII145 (82.9 ± 9.9 kDa; 95% CI) are consistent with a tetrameric native form. In contrast, the observed peak molecular mass of CII135 was 35.5 ± 2.8 kDa (95% CI), consistent with the theoretical dimer mass of 34.9 kDa. Thus we conclude that the CII135 truncation impairs CII self-association, presumably by disruption of the C-terminal domain.
Prokaryotic transcriptional activators employ a variety of mechanisms to increase transcription from their target promoters. Understanding a variety of activators is important to detecting patterns in how or why different activation mechanisms are employed, in terms of the constraints and features of their action. That is, it helps us to gain insights into what, if any, evolutionary pressures have steered organisms toward particular solutions to the problem of transcriptional control. The relationship between the structure and function of these activators is a key component of understanding the underlying mechanisms.
In this study, we find evidence that the DNA-binding activity of 186 CII is conferred by the N-terminal region, whereas the C-terminal region is responsible for CII self-association. We have not obtained direct evidence that both regions fold independently, so denote these as functional rather than structural domains. Attempts to express the His-tagged N-terminal domain (residues 1–81) and C-terminal domain (residues 82–145) independently have been unsuccessful. It is unclear whether this is due to the choice of domain boundary, an effect of the His6 affinity tag, or some level of structural dependence between the domains. Our data concerning CII135 suggest that the N-terminal domain at least is substantially structurally independent, but the independence of the C-terminal domain is less clear.
Our earlier work investigating CII self-association by analytical centrifugation has indicated a CII to DNA stoichiometric ratio of 2.5 under conditions of CII excess and saturated DNA binding sites (6). That study concluded that CII binds DNA as a pre-formed dimer, but that a weaker equilibrium with the formation of a CII tetramer also exists. Our SEC-MALS data are consistent with this model of self-association, because our experiments were necessarily conducted at higher protein concentrations, leading the tetramer to be the predominant species observed. The observation that CII135 can form only a dimer, but has significantly compromised DNA-binding activity suggests that there may be two independent dimerization interfaces, yielding two distinct dimer species: one that is competent for cooperative DNA binding, and one that is not. In light of this, one may speculate that the “dimer-tetramer” equilibrium is in fact not dependent upon prior formation of the DNA-binding dimer, but instead represents a second, independent mode of self-association in which either monomers or dimers may participate. Our in vitro transcription assays (Fig. 3C) and in vitro DNA binding experiments (6) indicate that both CII DNA binding (as a dimer) and pE activity occur at nanomolar concentrations of CII. We observe CII tetramers by SEC-MALS at substantially higher concentrations of CII (~100 μm), suggesting that the CII DNA-binding dimer is the transcriptionally relevant form.
We have also revealed two distinct epitopes responsible for activation by CII, one in each of the domains. The first, centered on residue Arg-115 of CII in the CTD, clearly functions dependently on one or more RNAP αCTDs. The pattern of activity of mutations at this position, in which several different changes are observed to confer a moderate defect, whereas the charge reversal mutation R115E confers a larger defect, are consistent with a direct interaction with a residue of opposite charge on αCTD. We cannot determine from our analysis whether αCTD-dependent activation by CII involves one or two αCTDs, but our observation that a single amino acid change for R115E destroys all αCTD-dependent activity does suggest that there are no functionally distinct roles for the two αCTDs of RNAP. That is, if two αCTDs are involved, they likely either cooperate in a single contribution to activity, or each contribute independently but symmetrically, interacting with the same epitope on different monomers of CII.
Conversely, activation by the second epitope, centered on residue Glu-46 of CII in the NTD, is αCTD-independent. The location of the promoter proximal binding site, overlapping the −35 element is similar to many activators that interact with σ70 (45). We have identified mutations in σ70 K593A and K593E that partially compensate for the activation defect of CII mutants E46K and R17L, suggesting that this epitope does indeed interact with σ70. However, the Lys-593 σ70 variants do not reduce pE activation by wild-type CII145. We can identify two possible explanations for these observations, the first being redundancy. If, for instance, CII Glu-46 can contact two alternate residues on σ70, then mutation of either one of the σ70 residues alone may not produce a defect, whereas mutation of Glu-46 would produce a strong defect, as observed. Under this model, the behavior of R17L CII145 would be explained by local structural rearrangements impacting the CII E46–σ70 Lys-593 interaction. The alternate explanation is that mutation at Glu-46 or Arg-17 rearrange the CII–σ70 interface such that Lys-593 becomes inhibitory to the interaction. We would then classify the effect of mutations in σ70 as relief of the inhibition caused by Lys-593. Because Lys-593 σ70 variants do not improve pE activation by wild-type or R115E CII145, the inhibitory role of σ70 Lys-593 must be reliant on prior rearrangement of the interface by the CII Glu-46 or Arg-17 mutations. According to this model, the key residue(s) on the σ70 side of the wild-type interface remain unidentified, because our analysis was restricted to a specific region of σ70. We favor the first model for its relative simplicity, but cannot fundamentally distinguish between them given our observations. Importantly, however, both models rely on close intermolecular contact between CII and σ70, giving a strong indication that the Glu-46 activation epitope of CII acts via a σ70-dependent mechanism.
De novo and homology structure prediction approaches yield models of the N-terminal domain with moderate levels of confidence, but significantly, these independent approaches produce very similar overall folds, with a 3.90-Å backbone root mean square deviation, excluding terminal regions without secondary structure (residues 8–74). The Rosetta de novo model predicts a solvent-exposed helix-turn-helix, positioned ready for DNA binding (Fig. 8A). Importantly, it predicts that Arg-17 and Glu-46 lie adjacent to each other in the three-dimensional fold, consistent with our experimental data showing that mutations at these residues form part of the same activation epitope. With the helix-turn-helix docked into the major groove, this lies on one side of the protein, positioned in a way that makes contact with σ70 plausible (46)(Fig. 8B).
With these models of self-association and activation in mind, it is worth re-examining prior evidence of the structural arrangement of CII and RNAP on the pE promoter. DNase I footprinting experiments using CII169 alone, or CII169 together with RNAP are instructive (4, 6). Both half-sites and the region between them are protected in the presence of CII169 alone, with the exception of a cleavage enhancement midway between the half-sites. In the presence of RNAP, two changes are evident. First, there is protection of the region downstream of the CII half-sites, extending to around +20 relative to the transcription start site. This is consistent with recruitment of σ70 and the RNAP core enzyme to the promoter by the N-terminal helix-turn-helix domain bound at the promoter proximal CII half-site. The second change is a shift in the enhancement between the CII half-sites. The current study suggests an αCTD-binding epitope in the CII CTD, which dimerizes to link the two half-site bound CII NTDs, and so the CTD must lie physically between the NTDs. Combining these observations, we arrive at a model of the promoter architecture in which CII stabilizes one or both αCTDs on the DNA region between the CII half-sites (Fig. 8C). To our knowledge, this arrangement of factors and interactions at a promoter has not been previously observed.
Our finding that CII is rapidly degraded in vivo extends our knowledge of the functional similarity between 186 CII and λ CII. In both cases, degradation of the protein is promoted by a C-terminal tail (47) and results in similarly short half-lives of around 2 min (8, 35). Given the role of the host protease FtsH in degradation of λ CII, and the importance of λ CII proteolysis to the frequency of lysogeny of λ phage (9), it would be interesting to investigate whether these are also conserved properties in 186 CII. FtsH-mediated proteolysis does not typically yield observable reaction intermediates (48), in contrast to our observation of a specific degradation product for 186 CII. Thus we speculate that 186 CII may be degraded by an alternative protease, offering the potential to explore whether the choice of protease used for degrading CII is important, as opposed to the simple fact of degradation itself. The mechanisms by which λ CII and 186 CII activate transcription also extend the functional, but not structural homology of their host phage, because both are thought to stabilize the αCTD and σ70 but in quite different ways. λ CII binds as a tetramer to direct repeat half-sites separated by only a single turn of the DNA helix, overlapping the σ70 binding site (49). λ CII is thought not to contact σ70 directly, but rather contacts αCTD, stabilizing it in a position conducive to σ70–αCTD contacts (50, 51). According to this model, λ CII recruits RNAP directly via the αCTD and indirectly via a resulting αCTD–σ70 interaction. This functional conservation likely stems from a physiological requirement for the functionally analogous CII proteins to be potent activators.
It remains unclear whether the particular arrangements of factors at the promoter confer specific advantages or properties on the promoters and what those advantages might be. One possibility is that αCTD binding between the CII half-sites on pE might result in greater activation due to the αCTD contact, by allowing αCTD to contact the promoter in close proximity to σ70. In the case of αCTD activation via an UP element contact, the strength of activation is inversely related to the distance of the UP element from σ70 (52), presumably due to the lower entropic cost of a minimally extended linker. Our model of pE activation by CII suggests that it is plausible that both αCTDs directly contact CII and DNA. In contrast, whereas both αCTDs are known to be involved with cAMP receptor protein-mediated activation, structures of the cAMP receptor protein-αCTD-DNA complex show that only one αCTD makes direct contact with cAMP receptor protein (53). Thus the unusual pE promoter architecture may serve to achieve greater αCTD-mediated activation through independent stabilization of both αCTDs on the DNA (Fig. 8C), without the need for sequential binding events and without compromising contacts with σ70.
Another possible reason for the adoption of this promoter architecture derives from the biological importance of rapid and effective proteolytic degradation of CII in the bacteriophage. Helix-turn-helix proteins bound as dimers in adjacent major grooves are in close physical proximity, which can lead to some cooperativity in DNA binding that is independent of a separate self-association domain, as in the case of λ CI (54). We have shown that proteolysis of CII is dependent on its C-terminal tail, and yields a specific, inactive product comprised of the first 135 residues. Based on the results presented here, the degradation product has lost one self-association interface, and this almost completely destroys its DNA-binding activity. By separating in space the half-sites, and thus DNA binding domains of CII, there is no possibility of residual interaction and cooperativity in the intact NTDs. We could expect then that this would lead to a more sensitive response of pE activity to proteolytic cleavage of CII. Encoding an activation epitope within the CII CTD, which becomes disrupted by degradation, would further enhance this sensitivity, which may be the key functional feature of this promoter architecture.
We thank Steven Busby, Richard Ebright, Seth Darst, Rachel Schubert, and eResearchSA for experimental tools; Daan van der Neut for preliminary data; and Julian Pietsch, Rachel Schubert, and Barry Egan for discussions.
*This work was supported by Australian Research Council Grants DP0665185 and DP110100824.
2The abbreviations used are: