QsIA disrupts LasR dimerization in antiactivation of bacterial quorum sensing

Significance

Quorum sensing is a bacterial cell–cell communication system that is activated when the concentration of quorum sensing signal (autoinducer) reaches a threshold. In Pseudomonas aeruginosa, an opportunistic human pathogen, the quorum sensing threshold and response are defined by a novel antiactivator, QslA, which binds to the transcription factor LasR and prevents it from binding to its target DNA. However, how QslA binds to LasR and negatively regulates quorum sensing is poorly understood. Here we show that QsIA binds LasR to disrupt its dimerization, thereby inhibiting the DNA binding of LasR and shutting down transcription. Our findings reveal the molecular basis of a unique QsIA-mediated LasR inactivation and add an example to understand the antiactivation mechanism in bacterial quorum sensing.

Abstract

The human pathogen Pseudomonas aeruginosa coordinates the expression of virulence factors by using quorum sensing (QS), a signaling cascade triggered by the QS signal molecule and its receptor, a member of the LuxR family of QS transcriptional factors (LasR). The QS threshold and response in P. aeruginosa is defined by a QS LasR-specific antiactivator (QslA), which binds to LasR and prevents it from binding to its target promoter. However, how QslA binds to LasR and regulates its DNA binding activity in QS remains elusive. Here we report the crystal structure of QslA in complex with the N-terminal ligand binding domain of LasR. QsIA exists as a functional dimer to interact with the LasR ligand binding domain. Further analysis shows that QsIA binding occupies the LasR dimerization interface and consequently disrupts LasR dimerization, thereby preventing LasR from binding to its target DNA and disturbing normal QS. Our findings provide a structural model for understanding the QslA-mediated antiactivation mechanism in QS through protein–protein interaction.

Quorum sensing (QS) is a bacterial cell–cell communication system that uses small diffusible molecules as signals that are also known as autoinducers (1, 2). Bacteria can release, detect, and respond to those signal molecules, as a measure of their own population density, to synchronize gene expression and coordinate biological functions, such as virulence, motility, biofilm formation, symbiosis, luminescence, antibiotic production, and plasmid transfer. A range of Gram-negative bacterial species, including several human and plant pathogens, use acylated homoserine lactones (AHLs) as QS signal molecules (3–5). AHLs are one family of the best-characterized cell–cell communication signals, which are synthesized by LuxI-type synthases and detected by LuxR-type regulators, which serve as the signal receptors. When AHLs reach a threshold concentration, the LuxR–AHL complex, as a dimer, binds to conserved palindromic sequences of the quorum-controlled promoters and activates the expression of QS-dependent genes, including the genes encoding AHL synthase and LuxR-type regulator, to generate a positive feedback.

Pseudomonas aeruginosa is a ubiquitous pathogen capable of causing serious and often life-threatening infections in cystic fibrosis patients and immunocompromised individuals. Some strains also infect plants and animals, and such infection is mediated by many QS-regulated virulence factors, such as pyocyanin, protease, elastase, exotoxin, and biofilms (6). It has been known that QS plays an important role in the regulation of virulence factor production and biofilm formation in P. aeruginosa (7–10). P. aeruginosa possesses at least two well-defined, interrelated QS systems, las and rhl. The las system consists of the transcriptional activator LasR, and a QS signal synthase LasI that directs the biosynthesis of 3-oxo-C12-HSL; the rhl system consists of the transcriptional activator RhlR, and an enzyme RhlI that is responsible for the biosynthesis of C4-HSL. The two QS systems are organized in a hierarchical manner such that the las system exerts transcriptional control over both rhlR and rhlI (11). LasR, as a key player in the circuit, requires 3-oxo-C12-HSL for its proper folding to control the whole QS systems (12).

LasR shows remarkable structural and functional similarity to the QS regulators TraR of the plant pathogen Agrobacterium tumefaciens and SdiA of Escherichia coli (13–15). AHLs bind to a conserved binding site in both LasR and TraR. The activity of LasR is negatively regulated by RsaL, which competes against LasR for binding to its DNA-binding sites (16). Like LasR, TraR exists as a dimer to bind AHL and activate QS in A. tumefaciens. TraR activity is negatively modulated by antiactivator TraM, which forms a stable antiactivation complex that prevents TraR from recognizing its target DNA-binding sites (17–21). We recently identified a unique antiactivator, QslA, in P. aeruginosa and showed that QsIA inhibits QS and virulence factor production through interacting with the QS signal receptor LasR and preventing LasR from binding to its target DNA sequence (22). Therefore, the QS threshold and response in P. aeruginosa are defined by QslA. However, how QslA interacts with LasR is unknown. It is thus of interest to understand the mechanism of QsIA–LasR interaction to inactive bacterial QS signaling system.

In this study we report the crystal structure of QsIA in complex with the LasR ligand-binding domain (LBD) and demonstrate that QsIA exists as a tightly associated dimer directly interacting with the LasR LBD. QsIA disrupts the LasR dimer through occupying the LasR dimerization interface. As a result, the LasR C-terminal DNA binding domain (DBD) dissociates from DNA promoters. Our results reveal a unique QsIA-mediated LasR inactivation mechanism in bacterial quorum sensing.

Results

Structure Determination.

To investigate the mechanism of how QslA interacts with LasR, we determined the crystal structures of full-length QslA in complex with the LasR LBD (amino acids 1–170). The complex of QsIA and LasR LBD was crystallized in space group P2₁2₁2, and its crystal structure was determined at a resolution of 2.3 Å by molecular replacement method combined with the single-wavelength anomalous dispersion phasing method. Statistics of the structure determination and refinement are summarized in Table 1.

Table 1.

Data collection and refinement statistics

Parameter	Native	Peak (Ta)
Data collection statistics
Space group	P21212	P21212
a/b/c (Å)	163.5, 185.9, 56.1	163.6, 185.3, 56.0
α/β/γ (°)	90.0, 90.0, 90.0	90.0, 90.0, 90.0
Wavelength (Å)	0.9793	1.2548
Resolution limit (Å)	2.4	3.0
Observed reflections (n)	252,559	121,142
Unique reflections (n)	64,498	33,978
Completeness (%)*	96.5 (78.0)	97.0 (89.3)
Rmerge (%)*	6.7 (46.4)	14.3 (68.1)
<I/σ(I)>*	14.0 (2.1)	6.2 (1.6)
Refinement statistics
Resolution range (Å)	48–2.4
No. of molecules/ASU	4 LasR and 8 QslA
Rwork/Rfree (%)†	23.0/27.4
No. of atoms
Protein/ligand/water	10,788/84/286
Mean B value
Protein/ligand/water	53.0/40.0/52.9
rmsd
Bond length (Å)/Bond angle (°)	0.01/1.325
Ramachandran plot (%)‡	85.6/12.9/1.5/0

^*Values in the highest resolution shell are shown in parentheses.

^†R_work = Σ||Fobs| − |Fcalc||/Σ|Fobs|. R_free is calculated identically, with 5% of randomly chosen reflections omitted from the refinement.

^‡Fractions of residues in most favored/allowed/generously allowed/disallowed regions of the Ramachandran plot were calculated according to PROCHECK.

Overall Structure.

In the asymmetric unit (ASU), there are four copies of the QslA-LasR LBD complex, with each copy containing one autoinducer (3-oxo-C12-HSL)-bound LasR LBD and two QslA molecules (Fig. S1). The two QslA molecules form a tight dimer to interact with one autoinducer-bound LasR molecule, giving rise to a 2:1 binding stoichiometry. Each pair of four LasR LBD molecules is related by a twofold noncrystallography symmetry axis, whereas no interdimer interaction was observed between the QsIA dimmers. To confirm the 2:1 binding stoichiometry in the crystal, analytical ultracentrifugation was used to examine the association state between QslA and LasR LBD in solution. Sedimentation velocity analysis gave an average molecular weight of 48.7 kDa (Fig. 1A and Table S1), which is close to the theoretical molecular weight of 46.4 kDa for two QslA molecues binding to one LasR LBD. Intriguingly, analytical gel filtration analysis showed that QslA alone appears to form a tetramer in solution (Fig. S2). Because four copies of the QslA–LasR LBD complex are similar to one another (pairwise rmsd of 0.5 Å) in the ASU, only the copy containing chains E and F of QslA and chain A of LasR is used for subsequent analysis.

Fig. 1.

Overall structure of the QslA–LasR LBD complex. (A) Two QslA molecules bind to one LasR LBD in solution, giving rise to a 2:1 binding stoichiometry. Sedimentation velocity analysis of the QslA–LasR LBD complex in solution at two concentrations (0.4 mg/mL and 0.8 mg/mL) was carried out and fitted according to the c(M) and c(S) size-distribution functions. The c(S) and c(M) distribution profile at both concentrations is similar, and a representative profile is shown at a concentration of 0.4 mg/mL. (B) Structure of the QslA–LasR LBD complex. The LasR LBD subunit (chain A) is shown in green, whereas the dimeric QslA molecules are shown in light blue (chain E) and in cyan (chain F). The autoinducer 3-oxo-C12-HSL is shown as a stick model.

As shown in Fig. 1B, LasR LBD folds into an α-β-α sandwich with a central five-stranded antiparallel β-sheet surrounded by three α-helices at both sides. 3-oxo-C12-HSL binds to the same antiparallel β-sheet as reported previously (13). QslA is composed of four α-helices with a short two-stranded antiparallel β-sheet following the N-terminal α1 helix. Dali server analysis (23) showed that no obvious QslA homologs were identified in Protein Data Bank, suggesting that the structure of QslA may represent a unique fold (Fig. S3).

The Interaction of QslA with LasR Is Crucial for Its Anti-LasR Activity.

The dimeric QslA interacts with the LasR LBD to generate two interfaces, I and II. In interface I (Fig. 2A), QslA mainly uses α1, β2, and the loop regions of α1-β1 and β2-α2 to interact with α1, α2, and α6 of LasR LBD through a combination of hydrogen bonding and hydrophobic interactions with a buried surface area of ∼1,510 Ų. In total, seven hydrogen bonds are formed, which involves residues His19, Tyr27, Ser44, and Leu46 of QslA and residues Asp29, Gly31, Ser146, and Thr150 of LasR. Besides these hydrogen-bonding interactions, the interaction of QslA with LasR LBD is strengthened by extensive hydrophobic interaction networks, which involve Leu3, Val4, Phe7, Leu30, Phe143, Val147, Leu148, Pro149, Trp152, and Met153 from LasR, and Val20, Ile23, Ile24, Tyr27, Leu30, Leu45, Leu46, Trp47, Pro48, Val49, and Met89 from QslA.

Fig. 2.

Interfaces in the QslA–LasR LBD complex. (A) Interface I between LasR LBD (chain A) and QslA (chain E). (B) Interface II between LasR LBD (chain A) and QslA (chain F). (C) Dimerization interface between QslA (chain E) and QslA (chain F). Residues involved in the interfaces are shown in stick model.

In interface II, QslA uses α3 from another subunit to interact with α6 of the LasR LBD through predominately hydrophobic interactions and buries a solvent-accessible area of ∼1,040 Ų (Fig. 2B). A hydrophobic cluster is identified, contributed by residues Trp75, Ala79, and Phe80 of QslA and residues Val83, Leu84, Pro85, Leu148, Pro149, and Trp152 of LasR. An additional interaction is provided by a salt bridge formed between Asp83 of QslA and Lys155 of LasR LBD.

To study the role of QslA–LasR LBD interface, we constructed a series of QslA Ala point mutants. These mutants, predicted to weaken the interaction of QslA with LasR, include the residues involved in hydrogen bonding (Ser44 and Leu46), ionic interaction (Asp83), and hydrophobic interactions (Leu45, Trp47, Pro48, Phe80, and Met89). A deletion mutant of QslA amino acids 1–27, covering the residues involved in hydrogen bonding (His19 and Tyr27) and hydrophobic interactions (Val20, Ile23, and Ile24), was also constructed. As shown in Fig. 3A, deletion of QslA amino acids 1–27 and Ala mutations of Leu45, Leu46, Trp47, and Asp83 almost completely abolish its anti-LasR ability with regard to the regulation of lasI gene promoter activity and pyocyanin production by LasR, whereas other mutants have moderate effect on reducing its anti-LasR activity. Consistent with these observations, bacterial two-hybrid assays showed that mutations of Leu45, Leu46, and Trp47 to Ala and deletion of residues 1–27 in QslA abrogate its association with LasR (Fig. 4A). Isothermal titration calorimetry analysis showed that QslA mutants L45A, W47A, and amino acids 1–27 deletion had no interactions with LasR-LBD, whereas L46A mutant showed fourfold lower binding affinity compared with wild-type QslA (Fig. S4). Together, these results demonstrated that the interaction of QslA with LasR is critical for its anti-LasR activity.

Fig. 3.

Mutational analysis of QslA on its interaction with LasR in P. aeruginosa. (A) Effect of wild-type QslA and its mutants on lasI promoter activity, which is regulated by active LasR. (B) Effect of wild-type QslA and its mutants on pyocyanin production, which responds to LasR activity. P. aeruginosa PAO1 carrying a pME-PlasI-lacZ reporter was introduced either in an empty vector pDSK or wild-type qslA or its mutant as indicated. The data are the mean of three repeat measurements.

Fig. 4.

Bacterial two-hybrid assays. (A) Bacterial two-hybrid analysis of the predicted residues of QslA involved in the interaction with LasR. (B) Bacterial two-hybrid analysis of the predicted residues involved in QslA dimerization. Bacteria were transformed with pairwise combination of pTRG-qslA or pTRG-lasR along with pBT containing wild-type qslA or its mutations. Positive control (pTRG-Gal11P + pBT-LGF2) was indicated. These bacteria were grown on nonselective (Left) or selective (Right) (with 5 mM 3-AT) plates. Bacterial growth in the selective plate indicates a positive interaction. (C) Stable accumulation of wild-type QslA and its point mutations. Wild-type and mutated QslA were tagged with the Flag epitope at their C terminals, and the protein expression was analyzed by Western blotting using anti-Flag antibody.

QslA Forms a Functional Dimer.

The dimerization of QslA buries a solvent-accessible area of ∼1,850 Ų mainly mediated by helices α3 and α4 from each monomer through a network of hydrogen-bonding and hydrophobic interactions (Fig. 2C). Specifically, the main-chain amide group of Leu74 (α3) from one monomer is hydrogen-bonded to the OE1 group of Gln103 (α4) from another monomer, whereas Arg82 (α3) from one monomer forms a salt bridge with Glu95 (α4) from another monomer. The hydrophobic interactions involve residues Trp65 and Leu66 (α2), Leu74, Trp75, and Phe78 (α3), and Leu99, Leu102, Leu106, and Val107 (α4) from each subunit.

To examine the important role of these residues in QsIA dimerization and in anti-LasR activity, several single point mutations were created. Bioassay results (Fig. 3B) showed that substitution of Leu74, Trp75, Arg82, Glu95 with alanine in QslA completely abolished its anti-LasR activity and eliminated LasR-regulated phenotype; mutation in Gln103 also showed a reduction in anti-LasR activity. Further mutational analysis by using bacterial two-hybrid assay confirmed the roles of the predicted residues in the QsIA dimerization. As shown in Fig. 4B, the E. coli strain coexpressing the wild-type QslA and its mutation either in Leu74, Trp75, Arg82, or Glu95 could not grow in the selective medium, whereas both wild-type QslA showed a good growth, indicating that mutations in these residues were unable to interact with wild-type QslA and perturb dimerization in the two-hybrid assay. These results indicate that α3 and α4 play a critical role in QsIA dimerization and that QslA functions as a dimer to exert its anti-LasR activity.

The Mechanism of QsIA-Mediated Antiactivation of LasR.

Previous studies indicate that the binding of the autoinducer 3-oxo-C12-HSL stabilizes LasR and promotes its dimerization, thereby enabling the resultant homodimeic LasR-autoinducer complex to bind to its target DNA promoter and activate gene transcription (6, 13, 22, 24, 25). We have shown previously that QslA can associate with LasR and thus disturb its target DNA-binding ability (22). However, how the interaction of QslA with LasR leads to its anti-LasR activity is unknown. Structural comparison of LasR LBD in complex with 3-oxo-C12-HSL with our QsIA-LasR LBD complex shows that QsIA binding to LasR LBD disrupts the LasR dimerization by occupying the LasR dimerization interface (Fig. 5). Current structures of LuxR-type regulators show that their dimerization is mainly mediated by their LBDs, whereas their DBDs have a minor role for their dimerization (15, 26–28). The LBD dimerization of LuxR-type regulators may help to position the DBD dimer in a correct conformation to recognize the target DNA sequence. Thus, disruption of the LasR LBD dimer by QslA might result in the DBD dimer adopting an unfavorable conformation, thereby affecting its ability to recognize its target DNA sequence efficiently (Fig. S5).

Fig. 5.

Disruption of LasR dimerization by the QslA dimer. Superposition of LasR (chain A) of the QslA–LasR LBD complex with one subunit (chain F, not shown for simplicity) of the LasR LBD symmetrical dimer (Protein Data Bank code 2UV0, chains F and H). Color scheme: QslA (chain E) in light blue, QslA (chain F) in cyan, LasR (chain A from QslA-LasR LBD complex) in gray, LasR (chain H from 2UV0) in wheat. Surface of LasR (chain A) involved in interaction with the other LasR subunit from 2UV0 or QslA (chain F) is shown in light green, whereas surface of LasR (chain A) associated with QslA (chain E) is in pink.

QslA also dissociates DNA from a preformed LasR–DNA complex (22). Insight into this mechanism comes from molecular modeling (SI Materials and Methods) of the full-length LasR with bound DNA based on the structures of the TraR–DNA complex (15, 26) and QscR (28) and the structure of QscR (27). The resultant LasR–DNA model shows a symmetric architecture. Upon binding to this LasR–DNA complex, QslA would disrupt the LasR dimer by occupying the same surface region used for LasR dimerization, therefore causing the dissociation of DNA from LasR and shutting down transcription (Fig. S5).

Discussion

QsIA is a unique antiactivator that prevents transcription factor LasR binding to target DNA. Antiactivators from several systems seem to occupy sites on the transcription factors that would otherwise coordinate specific base contacts on the DNA, thereby precluding or inhibiting binding of the transcription factor to its target elements (29–31). However, unlike those antiactivators, QsIA uses a unique mechanism by which QsIA interacts with LasR LBD and hence disrupts the LasR dimerization, which is essential for LasR–DNA interaction.

TraM, an antiactivator of TraR, uses a markedly different mechanism to inhibit the DNA binding activity of TraR. Unlike QslA, which binds to the LasR LBD, TraM binds between the LBD and DBD of the TraR dimer such that the two TraR DBDs are separated from each other, thus preventing them simultaneously binding to TraR box DNA (32–34). In addition to the protein antiactivators identified in inhibition of QS, small molecules of synthesized AHL analogs have been examined extensively for any potential effect of antagonism of QS (35–37). Through structure–function studies of natural and synthetic ligands, it was found that chlorolactone compound, as a strong potent antagonist, occupies the natural ligand (C6-HSL) binding pocket of CviR from Chromobacterium violaceum and results in an opening of approximately 20° between the two DBDs compared with that of cognate C6-HSL bound structure (27). This opening widens the two DBDs from their optimal distance for DNA binding and thus disfavors the activation of their target genes (27).

In summary, our finding adds an example to understand the antiactivation mechanism in bacterial QS. Current structure–function studies of antiactivators in bacterial QS unravel three different ways against LuxR type regulators, including disruption of LBD dimerization interface by QslA in P. aeruginosa, occupation of AHL binding pocket by its analogs in C. violaceum, and association with the DBDs by TraM in Rhizobium species. Importantly, all these interactions position the DBDs into unfavorable orientations to recognize their target DNA sequence. These findings shed light on how the bacterial QS can be inhibited and may provide an alternative strategy of easing infectious disease and antibiotics resistance. Moreover, these findings further our knowledge of how transcriptional factors can be modulated.

Materials and Methods

Protein Expression and Purification.

The full-length LasR (239 residues) was insoluble when expressed in E. coli, either in the presence or absence of 3-oxo-C₁₂-HSL. A construct spanning LasR residues 1–170, the LBD, was soluble in the presence of 3-oxo-C₁₂-HSL. For coexpression of LasR and QsIA in E. coli, LasR LBD and full-length QsIA were cloned to pETDuet vector with an N-terminal His-tag added to LasR LBD. Bacteria were grown at 37 °C to midlog phase (OD₆₀₀ 0.6–0.8), and expression was induced with 0.15 mM isopropyl β-d-1-thiogalactopyranoside. After induction, bacteria were grown at 18 °C for 12 h and then harvested by centrifugation (6,328 × g, 20 min, room temperature). Bacteria were lysed by sonication in PBS, and the lysate was centrifuged (17,418 × g, 20 min, 4 °C). The supernatant was applied to a Ni²⁺-nitrilotriacetic agarose column and was washed with three column volumes of washing buffer [0.8 M NaCl, 20 mM Hepes (pH 8.0)], and the QslA–LasR LBD complex was eluted from the column with elution buffer [20 mM Hepes (pH 8.0), 0.3 M imidazole, and 0.5M NaCl]. The QslA–LasR LBD complex was further purified on a Superdex-75 gel filtration column equilibrated in 20 mM Hepes (pH 8.0) and 0.2 M NaCl. The protein was dialyzed in 20 mM Hepes (pH 8.0) and 50 mM NaCl and concentrated by ultrafiltration to 15 mg/mL for crystallization.

Crystallization and Structure Determination,

Crystals of the QsIA–LasR LBD complex were grown at 18 °C by hanging drop vapor diffusion method. One microliter of the QslA–LasR LBD complex was mixed with 1 μL of 20% (wt/vol) polyethylene glycol 1,000 (PEG1K), 0.2 M MgCl₂, 0.1 M NaCl, and 50 mM sodium cacodylate (pH 6.5). Crystal morphology was improved by microseeding. Heavy atom derivative crystals were obtained by soaking the crystals in 2 mM tantalum bromide, 20% (wt/vol) PEG1K, 0.2 M MgCl₂, 0.1 M NaCl, and 50 mM sodium cacodylate (pH 6.5) for 2 h at room temperature. Native and derivative crystals were cryoprotected in the mother liquor supplemented with 15% (vol/vol) glycerol before freezing in liquid nitrogen. Diffraction data of native and derivative crystals at the peak edge of tantalum were collected at beamline 14-4, European Synchrotron Radiation Facility (ESRF), Grenoble. Optimal diffraction volumes were defined by diffraction cartography (38) as implemented in the workflow interface of the beamline control GUI MXCuBE (39, 40). Integration, scaling, and merging of intensities were carried out using Mosflm and Scala from the CCP4 program suite (41). Initially, molecular replacement using PHENIX (42) was used to find four clear solutions using a single LasR LBD as a search model (Protein Data Bank code 2UV0, chain F). From the maps of 2Fo-Fc and Fo-Fc, the density of QslA near each copy of LasR LBD was partially visible. The density of QslA was improved by phase combination of model phase and experimental phase from the tantalum derivative. After density modification, QslA C-terminal α-helical backbones were built automatically using the PHENIX autobuild module, whereas the N-terminal backbones of QslA were manually built in COOT (43).

The model was then refined against the high-resolution native data set. Simulated annealing with torsion angle dynamics was carried out from 3,000 K using the slow-cool protocol of CNS (44, 45). Ten cycles of maximum likelihood restrained refinement were subsequently carried out using REFMAC in the CCP4 program suite (41), each cycle being followed by manual rebuilding into σ_A-weighted 2mFo-DFc and mFo-DFc maps using COOT (43). Waters were added in the later stages of the refinement using PHENIX with default parameters (3σ peak height in mFo-DFc maps), followed by inspection of maps. Structure validation was performed using PROCHECK (46). The final model contains four LasR LBD molecules (chains A, B, C, and D) and eight QslA molecules (chains E, F, G, H, I, J, K, and L). Some residues are disordered, which include residues 1, 168–170 for LasR LBD (chains A, B, C, and D) and residues 1–16 (chains E, H, I, and L) and residues 1–28 (chains F, G, J, and K) for QslA. Structural figures were prepared in Pymol (www.pymol.org).

β-Galactosidase Assay and Pyocyanin Measurement.

For measurement of β-galactosidase activity, P. aeruginosa PlasI-lacZ reporter strain was transformed with qslA or its mutants in pDSK vector. The cultured bacterial cells were collected and measured for β-galactosidase activity. The experiment was repeated three times and each repeat with two duplicates. Results were given as Miller units of β-galactosidase activity per OD₆₀₀. Pyocyanin was assayed from bacterial supernatant according to the method of Essar et al. (47).

Bacterial Two-Hybrid Analysis.

The BacterioMatch II Two-Hybrid System (Stratagene) was used. qslA and its variants was fused separately to λcI, which encodes a bacteriophage λ repressor protein in the vector pBT, and the coding sequences of LasR LBD was fused to the N-terminal domain of the α-subunit of RNA polymerase in the vector pTRG. The resultant constructs, including positive (pTRG-Gal11P and pBT-LGF2) and several negative control plasmids, were cotransformed into E. coli strain XL1-Blue RF’ Kan. These strains were assayed for their growth on M9+ His-dropout medium (Clontech) with or without 5 mM of 3-amino-1,2,4-triazole (3-AT) according to the manufacturer’s instructions.

Protein Data Bank Deposition.

The coordinates and diffraction data have been deposited into the Protein Data Bank with accession code 4NG2.