Influence of Flexible “ω” on the Activity of E. coli RNA Polymerase: A Thermodynamic Analysis

Abstract

The Escherichia coli RNA polymerase (RNAP) is a multisubunit protein complex containing the smallest subunit, ω. Despite the evolutionary conservation of ω and its role in assembly of RNAP, E. coli mutants lacking rpoZ (codes for ω) are viable due to the association of RNAP with the global chaperone protein GroEL. With an aim to get better insight into the structure and functional role of ω, we isolated a dominant negative mutant of ω (ω₆), which is predominantly α-helical, in contrast to largely unstructured native ω, and then studied its assembly with reconstituted core1 (α₂ββ′) by a biophysical approach. The mutant showed higher binding affinity compared to native ω. We observed that the interaction between core1 and ω₆ is driven by highly negative enthalpy and a small but unfavorable negative entropy term. Extensive structural alteration in ω₆ makes it more rigid, the plasticity of the interacting domain formed by ω₆ and core1 is compromised, which may be responsible for the entropic cost. Such tight binding of the structured mutant (ω₆) affects initiation of transcription. However, once preinitiated, the complex elongates the RNA chain efficiently. The initiation of transcription requires recognition of appropriate σ-factors by the core enzyme (core2: α₂ββ′ω). We found that the altered core enzyme (α₂ββ′ω₆) with mutant ω showed a decrease in binding affinity to the σ-factors (σ⁷⁰, σ³² and σ³⁸) compared to that of the core enzyme containing native ω. In the absence of unstructured ω, the association of σ-factors to the core is less efficient, suggesting that the flexible native ω plays a direct role in σ-factor recruitment.

Introduction

Protein-protein interactions play a vital role in a wide range of biological processes (1, 2, 3). Intersubunit interaction in a multisubunit protein complex is fundamental to the basic function of the protein. The prokaryotic RNA polymerase (RNAP) core enzyme, which is a multisubunit complex and performs gene transcription, is composed of five subunits: two α subunits, one β subunit, one β′ subunit, and one ω subunit. The core enzyme couples with one of the σ-subunits, which promotes transcription initiation at the appropriate start sites, and mostly uncouples upon escape of the promoter (4, 5).

Understanding the roles of individual subunits of RNAP is a prerequisite for following the complex mechanism of transcription. The functional characterization of RNAP has assigned specific roles to the different subunits, except for the smallest subunit, ω, whose roles in transcription still remain elusive (6). It was observed that in bacteria, removal of the ω-subunit does not impair the transcription process (7). Moreover, relative to the wild-type strain, the ω knockout strain in Escherichia coli has been found to have slightly retarded growth but appears morphologically the same, implying that ω may not be an essential subunit of RNAP (8, 9). However, the crystal structure of Thermus aquaticus RNAP indicated that the spatial disposition of ω in the holoenzyme is like a chaperone (10). This hypothesis was reinforced by subsequent work (11, 12) in which reconstitution experiments demonstrated that ω is indeed like a chaperone for the largest subunit, β′, during the assembly of RNAP. However, upon deletion of ω, other proteins pertaining to the same function get associated with RNAP (8). Thus, it is difficult to ascribe the importance of ω in RNAP or to determine its fundamental role. With the aim of assigning a functional role to ω, we created a dominant negative strain of E. coli in which the mutant protein (ω₆), carrying an N60D mutation, is predominantly α-helical, in contrast to native ω, which is largely unstructured and exhibits enhanced affinity for β′ (13).

The unstructured ω falls into the category of intrinsically disordered proteins (IDPs). Recent studies highlight several instances of the importance of intrinsic disorder in proteins. The high intramolecular flexibility and conformational plasticity of IDPs allows them to bind to their target with high specificity and modulate their function (14, 15, 16). As ω₆ has rigidity and strongly binds to the β′ subunit, we envisaged that perhaps the flexibility in native ω is required for the proper assembly and mobility necessary for catalysis. This turns out to be important, as the β′-ω interacting domain is the target for several modulators, such as ppGpp (17, 18). Thus, it is important to elucidate the thermodynamics of the interaction of structured and unstructured ω with core1 (α₂ββ′).

In addition, we wanted to find out the impact of this inherent flexibility on the functioning of the enzyme. Previously, we established that the mutant ω₆ remained strongly bound to the enzyme and the initiation of transcription was abrogated. If provided with preinitiated complex, RNAP containing ω₆ elongates the RNA chain similarly to the wild-type enzyme (13). The initiation of transcription in bacteria requires assembly of a σ−factor (σ) with the core RNAP (α₂ββ′ω (core2)) to form the RNAP holoenzyme (Eσ), which in turn recognizes promoter sequences. Switching among the σ-factors that bind to core RNAP reprograms the cell’s transcriptional networks, redistributing RNAP to promoters that are utilized under alternative conditions (4, 19, 20). These σ-factors dissociate from the RNAP core enzyme after transcription initiation. Thus, it is likely that the flexibility in ω may have an influence on the association of σ-factors to the core enzyme. To address this question, we studied the interaction of σ-factors with the reconstituted core2 (α₂ββ′ω_variant) to follow whether the structural disorder in ω has a role in fine-tuning transcription.

In this work, we purified the wild-type ω and its dominant negative variant, ω₆ (13, 21), and then studied their assembly with reconstituted RNAP (core1: α₂ββ′). Subsequently the interaction of σ-factors with reconstituted RNAP (core2: α₂ββ′ω; mutated core2: α₂ββ′ω₆) was also analyzed. Our results indicate that ω is involved in maintaining the structural integrity of RNAP, as well as in σ-factor recruitment, and that the intrinsic disorder that is present in ω plays a crucial role in the proper functioning of the enzyme.

Materials and Methods

Protein purification and reconstitution of E. coli RNAP from recombinant subunits

The E. coli RNAP α-subunit (C-terminally His-tagged, α-His) was overproduced in BL21(DE3) using pETα plasmids and isolated under native conditions by Ni-NTA affinity chromatography (22). The other core subunits, β, β′, ω, and the ω-mutant (ω₆), were overproduced using the plasmids pGETB, pRW308, pET-ω, and pRPZM6 (11, 13), respectively. These subunits were untagged and were purified from inclusion bodies as reported. The core enzyme was reconstituted by assembling the subunits in appropriate molar ratios, as described earlier (23). After reconstitution, the core enzyme was found to be active. The expression and purification of all the σ-subunits of E. coli RNAP was carried out based on a previously published protocol (24). The purity of the resulting proteins was confirmed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (Fig. S1). The protein concentration was estimated using the Bradford assay with bovine serum albumin as a standard.

The Langmuir-Blodgett technique: Immobilization of core1 and ω at the Langmuir monolayer

To form an Ni-arachidate monolayer on a Langmuir Blodgett (LB) trough (Nima Technology, Coventry, United Kingdom), 25 μL of arachidic acid (CH₃(CH₂)₁₈COOH, purity 99%; Sigma-Aldrich, St. Louis, MO) solution (1 mg/mL) in chloroform (HPLC grade; Merck, Rahway, NJ) was spread on 200 mL of 10⁻⁴ M NiSO₄ solution in water (Milli-Q water (Millipore, Billerica, MA), resistivity 18.2 Ω·cm). The pH of the subphase was maintained at 7.4 with 2 mM Tris-HCl buffer and 10 mM NaCl. At this pH, the dissociation of arachidic acid will be complete, and it will form an Ni-arachidate template by interacting with the Ni⁺² present in the subphase (25). Previously, we followed the sequential assembly of the multisubunit E. coli RNAP at the interface and also established the chaperone-like role of ω in the assembly pathway (22). In this study, we reconstituted the core enzyme (core1: α₂ββ′) in the absence of ω, and that core1 enzyme was introduced into the aqueous subphase when we were sure that the NiA monolayer had reached equilibrium (∼30 min). (NiA monolayer usually takes 30 min to reach equilibrium). The final concentration of the protein was 6 nM. Subsequently, ω subunits (2–45 nM) were added to the α₂ββ′ monolayer and the product, α₂ββ′ω (core2), was found to be active with full complementation of all subunits. To calculate the change in area per molecule due to addition of the protein, we recorded pressure-area (P-A) isotherms 1 h after injecting the ω-subunit. A similar methodology was followed to study the interaction of the σ-subunit with the different core enzyme. In such cases, the core enzyme was attached through the His-tagged α-subunit and different concentrations of σ were subsequently added (24). All the measurements were carried out at a barrier speed of 25 cm²/min. We calculated the area per molecule values at a surface pressure of 12 mN/m when the monolayer was in its condensed state. A thermostat bath attached to the trough was used to maintain the temperature of the trough at (T ± 0.1)°C.

Isothermal titration calorimetry

Isothermal titration calorimetry (ITC) experiments were conducted in a MicroCal VP-ITC unit (MicroCal, Northampton, MA). The calorimeter was periodically calibrated and verified with dilution experiments described by the manufacturer so that the mean energy per injection was <1.30 μcal with a standard deviation of <0.015 μcal. The protocol involved injection of aliquots of degassed protein (ω) solution from the syringe of the unit into the sample chamber containing the enzyme core1. Control experiments were performed by injecting identical volumes of protein (ω) solution into the buffer. Each injection generated a heat spike, the intensity of which diminished as the binding progressed and remained constant as the saturation point was reached. The area under each heat-burst spike was determined by integration, using the Origin software to yield the measure of the heat associated with the injection. The heat generated in the control experiments was subtracted from the heat of the core1-ω reaction to give the heat of the binding. The results were plotted as a function of the ω/core1 molar ratio, fitted with one set of model binding sites, and analyzed using the software to give the binding affinity (K_A), the stoichiometry (N), and the molar enthalpy change of binding (ΔH) (26).

Surface plasmon resonance studies

We performed surface plasmon resonance (SPR) experiments with a Biacore 3000 instrument to calculate the affinity between ω_variants and core1 using a CM5 chip (GE Healthcare, Little Chalfont, United Kingdom). The chip was activated using a 1:1 mixture of NHS/EDC at a flow rate of 5 μL/min to give a reactive succinimide ester. The changes in the refractive index at the surface of the sensor chip were monitored. These changes are generally assumed to be proportional to the mass of the molecules bound to the chip and are recorded in response units (RUs) (27). ω_variants were diluted to a concentration of ∼600–1200 pM in 10 mM sodium acetate buffer (pH 3.0) and injected over the activated flow cells at a flow rate of 2 μL/min until a change of 1000–1400 RU was obtained. Channel 1 was left underivatized to correct for nonspecific binding. The excess active groups were blocked using ethanolamine. The channels were washed repeatedly in HEPES-buffered saline running buffer (10 mM HEPES (pH 7.4), 150 mM NaCl, 3 mM EDTA, and 0.005% polysorbate-20) at a flow rate of 20 μL/min until a stable baseline was achieved. To follow the kinetics of the interaction, we allowed 50 μL of core RNAP at the desired concentration in running buffer to flow over all the channels at 30 μL/min. The chip surface was regenerated with 10–20 μL of 10 mM NaOH at a flow rate of 50 μL/min and then washed with running buffer until the baseline returned to zero. A similar methodology was followed to analyze the interaction of the σ-factors with mutated and wild-type reconstituted core enzyme (core2). Here, the pH of the sodium acetate buffer was maintained at 4.0. All the sensorgrams were analyzed through BIAevaluation software, version 3.1.

Results

We employed different biophysical techniques to characterize the differences between the interactions of structured and unstructured ω with core1. As the initiation of transcription is imperfect in the presence of the structured mutant of ω, we also studied the interactions of the wild-type core2 (α₂ββ′ω) and the mutated core2 (α₂ββ′ω₆) with the σ-factors to evaluate whether the primary step of transcription initiation, i.e., the association of the σ-factors, is dependent on the flexibility of ω.

Understanding the protein-protein interaction at the NiA monolayer by the LB technique

In this study, we used the LB technique (28) to probe the molecular interaction and binding between the core1 and ω-variants (native and mutant) based on surface-tension measurements. This technique is mainly useful for surface-active biomolecules that can readily adsorb on a surface and modify surface properties such as surface tension (29). However, the use of this technique to follow the behavior of a non-surface-active biomolecule is limited. Arachidic acid is a fatty acid (surface active) that can spread as an isotropic lamellar structure in aqueous solution, and because of the presence of an ionizable carboxyl group, it can interact with the divalent cation Ni²⁺ to form a condensed NiA monolayer at pH 7.4. Our lab has already established the capability of this monolayer to quantitatively characterize protein-protein and protein-DNA interactions (24, 30). In most cases, we utilized the unique ability of hexahistidine tags at the terminal of the protein (in this case, the α-subunit of RNAP) to complex with Ni²⁺ ions. Regioselective immobilization of core1 (the core enzyme without ω) at the air-water interface guaranteed sufficient mobility for the enzyme that the interaction interfaces with the other binding partners (proteins and DNA that can bind to RNAP) remain exposed.

With the aim of following the macromolecular interaction at the NiA monolayer, we first optimized the ionic strength of the medium. Previously, our lab had shown that increasing the concentration of Na⁺ in the subphase beyond 25 mM leads to commencement of the liquid phase, because replacement of the bivalent Ni²⁺ ion with monovalent Na⁺ lowers the condensation effect of the former (25). More importantly, we also require a sufficient amount of Ni²⁺ ion in the monolayer to capture and align core1. In this study, we optimized the salt concentration to 10 mM and this concentration of NaCl has no effect on P-A isotherms.

The stability of the NiA monolayer was checked by the absence of hysteresis (data not shown). Sufficient time was given for the area per molecule value for NiA-core1 to reach equilibrium. This experiment confirmed that the observed changes after injecting different ω-mutants were solely due to the interaction with the core1 enzyme, and the results eliminate the possibility of any contribution from excess core1 at the monolayer. In our previous attempt, the subunits present in the LB monolayer were probed with Western blot analysis as well as by transcription assay (22).

Interaction of ω-variants with the NiA-core1 monolayer

To evaluate the binding affinities between core1 and different ω-variants, we performed saturation binding experiments (30) where the binding of an increasing concentration of ω-protein is measured at equilibrium and analyzed to calculate the equilibrium association constant, K_A. In brief, after the NiA-core1 monolayer reached equilibrium, we injected ω-subunit in increasing concentrations until saturation was achieved. We recorded the P-A isotherm for each step. The fractional saturation, ν, i.e., the ratio of the concentration of bound ligand to the total available binding site is given by the equation

$$\upsilon =\frac{\left[\text{ML}\right]}{\left[{\text{M}}_{\text{o}}\right]},$$ (1)

where [M], [L], and [ML] are the concentrations of the unbound core1, unbound ligand (ω), and core1 bound to the ligand, respectively. [M_o] is the total concentration of the core enzyme. The concentration of the unbound ligand, [L], is given by

$$\left[\mathrm{L}\right]=\left[{\mathrm{L}}_{\mathrm{o}}\right]-n\left[{\mathrm{M}}_{\mathrm{o}}\right],$$ (2)

where [L_o] and [M_o] are the total concentrations of ω and core enzyme, respectively. n is the number of binding sites per protein molecule (RNAP).

We calculated the fractional saturation (ν) from the P-A isotherm curve using the equation

$$\nu =\frac{A_t-A_0}{A_{\text{max}}-A_0}.$$ (3)

A₀ is the area per molecule of only NiA-core1, A_max is the area per molecule of NiA-core1-ω at saturation, and A_t is the area per molecule of the NiA-core1-ω monolayer at any intermediate concentration of ω.

The binding isotherm was constructed by plotting ν/L versus ν, and the data were analyzed by the binding model of McGhee and von Hippel (31) as per the equation

$$\frac{\nu }{\mathrm{L}}={\mathit{K}}_{\text{A}}\left(1-n\nu \right){\left[\frac{\left(1-n\nu \right)}{\left\{1-\left(n-1\right)\nu \right\}}\right]}^{\left(n-1\right)},$$ (4)

where K_A is the intrinsic association constant of binding to an isolated site and n is the number of binding sites per protein molecule. All the binding data were analyzed using the Origin 7.0 software (Origin Laboratories, Northampton, MA), which determines the best-fit parameters of K_A and n to Eq. 4. The binding affinity for the interaction of structured ω₆ with core1 was calculated in a similar manner. Fig. 1 a shows the P-A isotherms for the NiA-His-core1-ω₆ monolayer when different concentrations of ω₆ were injected relative to the fixed concentration of His-core1, and the respective ν/L-versus-ν plot for the interaction is presented in Fig. 1 b. The corresponding curve for native ω is presented in Fig. S2. The association constant values are shown in Table 1. From the data, it is clear that the relative binding affinity of core1 with ω is much lower than that measured with ω₆ in the presence of 10 mM NaCl at 25°C. We have already mentioned in the Introduction that the mutant is predominantly α-helical, and our findings indicate that the binding affinity increases with the gain in helicity in the ω-structure.

Figure 1.

(a) Representative curves for P-A isotherms of NiA (dotted line) and NiA-core1 (core1:6 nM) (dashed line) with different amounts of ω₆ (solid lines) in molar ratios of (left to right) 1:1, 1:1.5, 1:2, 1:3, 1:4, 1:5, 1:6, and 1:7.5. (b) Scatchard plot for the interaction of core1 with mutant ω₆. The data were analyzed by the binding model of McGhee and von Hippel (31).

Table 1.

Macroscopic Equilibrium Association Constant Calculated for the Interaction of Core1 with Native and Mutant ω

Langmuir-Blodgett		SPR
KA × 10−7 (M−1)		KA × 10−6 (M−1)
ω	ω6	ω	ω6
0.56 ± 0.02	3.63 ± 0.11	0.278 ± 0.05	1.86 ± 0.35

Evaluation of the binding affinity using the SPR method

The above interaction studies were further extended by using a different nonhomogeneous technique, SPR, which measures a change in the refractive index of the medium in close vicinity to a metal surface and is used to monitor the binding of analyte molecules to immobilized receptor molecules (32). The data obtained from SPR studies were fitted to a simple 1:1 Langmuir binding model (33). In Fig. 2 a, the association between the mutant ω and core1 is shown by an increase in the RU values, whereas the dissociation of these two species is indicated by a decrease in the same trace. The values of the equilibrium association constants are tabulated in Table 1, and the corresponding kinetic parameters are depicted in Table S2. Compared to ω, ω₆ was shown to have a higher k_a value. For the dissociation rate constants, compared to ω₆, ω had a higher k_d value, and that is reflected in the equilibrium association constant, K_A, indicating preferential binding of ω₆ to core1. Here, the binding parameters are evaluated in the presence of 150 mM salt. The monomeric state of the analyte was confirmed by gel filtration chromatography (Fig. S6). When the temperature of the system was kept constant, the relative binding affinities were found to follow the same trend as observed by the LB technique, with almost seven times greater binding affinity for ω₆, as reflected in the comparative bar graph (Fig. 2 b).

Figure 2.

(a) Sensorgram for the interaction of core1 with ω₆. (b) Comparative binding affinity for the interaction of ω-variants with core1, obtained from SPR.

Temperature sensitivity of the interaction

Next, we wanted to evaluate the thermodynamics of the interaction mentioned above using the LB technique. The saturation binding experiments were performed at four different temperatures: 288.15 K, 293.15 K, 298.15 K, and 303.15 K. In LB analysis, there is a restriction in temperature variation to avoid surface evaporation. It was observed that the equilibrium binding constant decreases with the increase in temperature. Although the value of the binding affinity changes with temperature, the trend in the binding affinity remains the same throughout the temperature range (288.15–303.15 K), i.e., the affinity of native ω is less compared to that of ω₆ at all temperatures. We have constructed a van’t Hoff plot using the equation

$$-\text{ln}{K}_{\text{A}}=\frac{\Delta H}{RT}-\frac{\Delta S}{R}.$$ (5)

The plot of lnK_A versus 1/T is a straight line with slope −ΔH/R and y-intercept ΔS/R, from which the van’t Hoff enthalpy and entropy changes were calculated, as presented in Table 2. The respective van’t Hoff plots are shown in Fig. 3. The free energy for each interaction is less than zero, suggesting that the reaction is spontaneous. The interaction is exothermic with an unfavorable negative entropy change, implying extensive conformational alteration associated with the binding process.

Table 2.

Thermodynamic Parameters for the Interaction of Core1 with Native and Mutant ω Obtained from van’t Hoff Analysis of Langmuir Data

	ω	ω6
ΔH (kcal/mol)	−13.71 ± 1.1	−19.51 ± 2.5
ΔS (cal/mol/K)	−15.14 ± 1.0	−31.28 ± 1.6

Figure 3.

van’t Hoff plots for the interaction of (a) ω and (b) ω₆ with core1.

Elucidation of the thermodynamic parameters from ITC

ITC is a sensitive tool for monitoring macromolecular interactions and provides key insights into the molecular forces that drive complex formation (34, 35). In Fig. 4, the upper plot shows the representative primary data from calorimetric titration of the ω-mutant into a solution of core1 at 298.15 K. Each heat-burst curve in Fig. 4 corresponds to that of a single injection. These injection heats were corrected by subtracting the corresponding dilution heats derived from the injection of identical amounts of ω-mutants into the buffer alone. In the bottom plot in Fig. 4, the resulting corrected heats are plotted against the molar ratio of reactants. The ITC profile for the interaction of native ω with core1 is presented in Fig. S3. The data points reflect the experimental points, and the solid lines represent the calculated best fit to the data for the one-site model. The binding was characterized by exothermic heats in each case. The ITC data were fitted to a single-site model, as the integrated heat data showed only one binding event. The results of the ITC experiments are presented in Table 3. The affinity was almost six times higher for ω₆ compared to that of native ω. The salt concentration of the medium was kept at 150 mM in this case. The molar ratio of binding (N) is close to 1. The binding of both ω and ω₆ was driven largely by negative enthalpy and an unfavorable negative entropy change. The binding-affinity values observed from ITC are slightly different compared to that obtained from LB and SPR studies, but the trends in the binding-affinity values obtained from all the techniques are the same. Possible reasons for the observed discrepancy in the binding-affinity values obtained from different techniques are described in a later section.

Figure 4.

ITC profile for the interaction of core1 (10 μM) with ω_6. (Upper) Each heat-burst curve is the result of a 10 μL injection of ω₆ into the core1 solution. The molar ratio of the concentration of protein in the cell and syringe is 1:10. (Lower) The data points (solid squares) reflect the experimental injection heat, whereas the solid line represents the calculated fit of the data.

Table 3.

Thermodynamic Parameters for the Interaction of Wild-Type and Structured Mutant of Omega with Core1 Obtained from ITC Experiments

	ω	ω6
KA × 10−5 (M−1)	2.95 ± 0.40	11.84 ± 1.10
N	1.1 ± 0.02	0.81 ± 0.01
ΔH (kcal/mol)	−13.86 ± 0.4	−17.87 ± 0.518
ΔS (cal/mol/K)	−21.45	−32.16

Influence of ω in the recruitment of σ-factors as followed using the LB technique

The initiation of transcription in bacteria requires the association of a given σ-factor with the core enzyme, which confers the ability to bind to distinct subsets of promoter sequences (20, 36, 37, 38). Previously, our lab evaluated the thermodynamic parameters associated with core RNAP-σ interaction as a function of ionic strength and temperature (24). As mutated core2 is defective in transcription initiation (13), we designed experiments to follow the role of ω, and, more importantly, the presence of intrinsic disorder in the structure of ω, in the core RNAP-σ recognition. We reconstituted RNAP with native ω as well as mutant ω (ω₆) and studied the association of σ-factors with these reconstituted core enzymes in 10 mM NaCl at 25°C using the LB technique. The data obtained from the interaction studies are presented in Table 4. RNAP reconstituted with native ω showed the highest binding affinity to primary σ-factor σ⁷⁰, and the trend in binding affinity follows the order σ³⁸ ≈ σ³² < σ⁷⁰. In the presence of the highly structured mutant, i.e., ω₆, the binding affinity to all the σ-factors decreases. The influence of disordered ω on the binding of σ³² and σ³⁸ to the core enzyme is less pronounced compared to the binding of σ⁷⁰. This is probably because σ⁷⁰ was found to be more susceptible to changes in different conditions compared to the other σ-factors (39).

Table 4.

Macroscopic Equilibrium Association Constant Calculated for the Interaction of Core2-α₂ββ′ω_variants with Sigma Factors Obtained Using the LB Technique

		σ70	σ38	σ32
KA × 10−7 (M−1)	RNAP (α2ββ′ω)	10.96 ± 1.11	4.21 ± 0.26	4.94 ± 0.25
	RNAP (α2ββ′ω6)	0.972 ± 0.04	1.07 ± 0.05	1.71 ± 0.06

Evaluation of the macroscopic association constant using SPR

The binding of σ-factors to core RNAP was also evaluated by SPR. Here, the association rate constants (k_a) are higher for the interaction of σ-factors with core2 than for interaction with mutated core2 (Table S3). However, the difference between the dissociation rate constants (k_d) is not as prominent. These differences are reflected in the overall binding affinity value. Thus, in the presence of unstructured ω, the core enzyme shows the highest binding affinity to primary σ-factor σ⁷⁰. Fig. 5 a represents the sensorgram for the association of σ⁷⁰ with mutated core2. A comparative bar chart for the binding affinity values is presented in Fig. 5 b. The sensorgram for the association of σ³² and σ³⁸ with mutated core2 is presented in Fig. S4. The association was reduced for all the σ-factors with the incorporation of structured mutant in the core enzyme and follows the same trend as observed using the other techniques. There is an almost 10-fold decrease in the binding affinity for σ⁷⁰ and a threefold decrease for σ³⁸ and σ³². The values of the association constants are presented in Table 5. The results indicate that the loose association of the disordered ω in the catalytic core of RNAP favors the recruitment of the σ-factors to form the holoenzyme.

Figure 5.

(a) SPR sensorgram for the interaction of σ⁷⁰ with core RNAP(ω₆). (b) Comparative bar graph for the association of σ subunits with core2 containing native and mutated ω.

Table 5.

Macroscopic Equilibrium Association Constant for the Interaction of Core2 with Sigma Subunits Obtained Using the SPR Technique

		σ70	σ38	σ32
KA × 10−5 (M−1)	RNAP (α2ββ′ω)	37.9 ± 3.6	18.4 ± 0.60	22.5 ± 1.08
	RNAP (α2ββ′ω6)	3.59 ± 0.13	5.81 ± 1.01	6.12 ± 0.45

Thermodynamic characterization of the RNAP-σ interaction using ITC

We also employed ITC to elucidate the binding affinity and thermodynamic profile of the interaction of the three σ-factors (σ³², σ³⁸, and σ⁷⁰) with both the reconstituted native and the mutated core enzyme. The experiments were conducted in the presence of 150 mM NaCl at 25°C, contrary to the salt concentration used during surface tension measurements.

The binding affinity value and the thermodynamic parameters associated with the interaction are shown in Table 6. The binding affinity with the core enzyme containing ω follows the order σ³⁸ ≈ σ³² < σ⁷⁰. The σ-recognition is stronger with core2 in comparison to mutated core2. In all cases, the interaction is accompanied by negative enthalpy and an unfavorable negative entropy term (Table 6) indicating that the interaction may involve large-scale conformational change of the protein. It is also evident that the magnitude of the negative entropy for the interaction with core2 and mutated core2 is highest for σ⁷⁰ as compared to σ³² and σ³⁸. This is further elaborated in the Discussion. A typical ITC profile for the interaction of mutated core RNAP (core2) with σ³² and a comparative bar graph for all the interactions are presented in Fig. 6. ITC profiles for the interaction of mutated core2 with σ³⁸ and σ⁷⁰ are presented in Fig. S5

Table 6.

Thermodynamic Parameters for the Interaction of Core2 with Different Sigma-Factors Obtained from ITC Experiments

		N	KA × 10−5 (M−1)	ΔH (kcal/mol)	ΔS (cal/mol/K)
RNAP (α2ββ′ω)	σ70	0.80 ± 0.01	9.32 ± 1.61	−132.0 ± 4.02	−417.0
	σ38	1.06 ± 0.03	4.60 ± 0.62	−96.68 ± 6.03	−298.3
	σ32	0.76 ± 0.04	4.65 ± 0.92	−91.81 ± 6.74	−282.2
RNAP (α2ββ′ω6)	σ70	1.08 ± 0.09	1.03 ± 0.20	−141.7 ± 18.8	−452.2
	σ38	1.05 ± 0.06	1.47 ± 0.18	−96.90 ± 8.21	−301.4
	σ32	1.06 ± 0.03	2.01 ± 0.15	−114.7 ± 5.57	−349.0

Figure 6.

(a) ITC profile for the interaction of σ³² with RNAP(ω₆). Each heat-burst curve is the result of a 10 μL injection of σ³² into the mutated core2 solution. The molar ratio of the concentration of protein in cell and syringe is 1:12.5. The data points (solid squares) reflect the experimental injection heat, whereas the solid line represents the calculated fit of the data. (b) Comparative bar graph for the association of σ-subunits with core RNAP containing wild-type and mutated ω obtained from the ITC experiment.

Discussion

Most proteins need to adopt a defined three- dimensional structure to carry out their function. However the last decade has witnessed a growing recognition that a large fraction of the genome of any organism encodes proteins that do not adopt a defined three dimensional structure but are nevertheless important for cellular function (14, 15, 16). This group of proteins is referred to as IDPs. It appears that the smallest subunit of RNAP, ω, falls into the category of IDPs. With an aim to understand whether there is any correlation between the extent of interaction of core1 and the flexibility in ω, we have carried out the binding study with wild type as well as with a structured mutant. The detailed characterization of the mutant has been reported in our previous study (13). Strikingly, we have found that ω₆ interacts with RNAP with significantly higher affinity compared to that of native ω. Thermodynamics of the recognition suggests that the interaction is driven by highly negative enthalpy and a small but unfavorable negative entropy term. The key thermodynamic driving force for the binding reaction is generally a favorable enthalpy contribution. For protein–protein interaction the entropy of binding obtained from the calorimetric techniques is comprised of contribution from different factors (40, 41)

$$\Delta {\mathit{G}}_{\text{bind}}=\Delta H_{\text{bind}}-T\Delta S_{\text{bind}}=\Delta H_{\text{bind}}-T\left(\Delta S_{\text{RNAP}}+\Delta S_{\omega }+\Delta S_{\text{sol}}\right),$$

where, ΔG_bind, ΔH_bind, ΔS_bind are the changes in free energy, enthalpy and entropy due to binding. ΔS_RNAP, ΔS_ω, and ΔS_sol are the contribution associated with the core1, ω, and the solvent, respectively, to the entropy of binding. Generally, the entropy contribution from the solvent i.e., $\Delta S_{\text{sol}}$ is less compared to the other factors. The entropic loss (for the ligand, ΔS_ω) for the intrinsically unstructured protein should be more negative due to disorder-to-order (in case of ω) transition compared to the structured protein (ω₆). But here, the overall entropy loss for the interaction of mutants (ω₆) with core1 was found to be more compared to the flexible ω. This can only be explained if we assume that the large structural alteration in ω₆ enhances its affinity and affects the conformation of the target core1 to a great extent. It is well known that β′ F (bridge)-helix and β′ G-loop comprise the mobile element of the active center of RNAP and they work in together to ensure proper placement of NTPs at the active site and incorporation into RNA (42). We hypothesize that the tight binding of the structured mutant affects the flexibility of the mobile element in active center of RNAP needed for transcription initiation and that is reflected in the entropy value. This finding indicates that the intrinsic disorder present in ω has important role in maintaining proper activity of the RNAP catalytic center.

Sigma factors play a crucial role during promoter selection and initiation of transcription upon binding to core RNAP (43, 44). All the σ-factors used in our study belong to σ⁷⁰ family of proteins because of their significant homology in domain organization and amino acid sequence (20, 45). They contain four conserved domains (region 1.1, σ₂, σ₃ and σ₄). σ₂ makes primary contact with coiled-coil domain of β′ using region 2.2 (46). The σ₃ and σ₄ interact with the active site of β subunit and flap region, respectively. The binding of σ⁷⁰ to the catalytic core brings about conformational changes in both the proteins (43, 47). Within the core some regions move and some disordered region become ordered resulting in large scale conformational change. The conformational flexibility of both the partner molecules is restricted due to binding and that accounts for the observed negative entropy of the interaction (Table 6).The energy required for this conformational changes must be supplied from the energy released during electrostatic interaction between the charged protein lobes. So, the overall change in enthalpy is basically the excess energy released after allowing for the conformational changes of σ-factors. The lesser the value of the observed enthalpy, the greater will be the distortion involved for σ-factors to undergo molecular interaction with RNAP. We observed that smaller amount of enthalpy is released for wild type and mutated core2 interaction with σ³², σ³⁸ at 25°C as compared to the same for the interaction of core RNAP to σ⁷⁰. So, we can conclude that the alternative σ-factors may have undergone greater structural distortion to interact with core RNAP as compared to that with σ⁷⁰. However, the trend in the values of the thermodynamic parameters remains almost similar for the σ-interaction with mutated core2 with slightly more negative entropy change accompanied by reduced binding affinity. This suggests that the binding of σ factors to the core2 is greatly influenced by the presence of intrinsically disordered ω. As the initial contact between σ and the core enzyme are made through σ2.2 and β′, it is necessary that β′ should be in proper conformation for sigma recognition. In the scenario of loss of flexibility of ω, the conformational diversity of the interacting domain in RNAP may be compromised leading to reduced binding of the mutated core2 enzyme with sigma factors. We are tempted to conclude that the difference in strength of interaction is reflected during the initiation of transcription. However, once σ is released, elongation can continue in normal rate. The effect is more prominent with σ⁷⁰ compared to other σ-factors. The σ-70 family tree depicts a gradual increase in the distance of σ³² and σ³⁸ from σ⁷⁰, indicating a divergence of amino acid sequence from that of principal σ-factor (4) and that is reflected in the observed difference in binding affinity. This finding is supported by the fact that RNAP purified from ω-null strain showed altered composition containing lesser amount of σ⁷⁰ (48, 49, 50, 51) and thus highlighting the role of flexibility of this smallest subunit in σ-factors recruitment.

Discrepancy among the binding affinities obtained from ITC, SPR, and LB techniques

We note that there is a fundamental difference between the modes of detection of the three procedures, as manifested in the respective quantitative evaluation. Both LB and SPR are nonhomogenous techniques that function on the principle of partitioning of reactants between the bulk phase and the immobilized phase. The advantage of these two methods in comparison to other conventional biophysical techniques lies in the fact that they allow us to monitor the mechanistic pathway governing the crucial biological reaction while mimicking the crowded cellular milieu. In the LB technique, the protein molecule is immobilized at the air-water interface but site-directed immobilization assured that the binding interfaces remained unaffected, and we evaluated the binding parameters only after equilibrium was reached. However, the problem with the LB technique is the ionic strength of the medium, which is much lower than that of the physiological condition. We can overcome the problem of ionic strength in SPR. For SPR, we have an air-solid interface, with the analyte immobilized on the solid support. However, here the interactions are followed in real time, and mass transport affects the observed rate constant. The heterogeneity of covalent linkage formation in SPR may also compromise the availability of the interacting domain. In ITC, successive additions of one protein from the syringe to a solution of another protein in a reaction cell leads to a specific amount of protein-protein complexes dictated by their binding affinity monitored by the heat release. Here, we can maintain the ionic strength (150 mM) of the medium close to the physiological condition, which is not possible using the LB technique. As a result, the differences in the binding affinity values obtained from ITC and SPR are smaller than that obtained using the LB technique. However, cellular crowding cannot be modeled in ITC. Every technique has its distinct advantages and disadvantages. It should be emphasized that we are interested in discovering the role of intrinsic disorder in the ω-subunit in RNAP, which guides the interrogation of different σ-factors at different points of time, so all of the techniques are useful here.

Conclusion

Specific association between proteins is fundamental to all aspects of biology. In this study, we tried to decipher the physiological role of ω as a part of RNAP by analyzing the protein-protein interaction using biophysical approaches. Any alteration that affects the conformation of ω has a critical effect on RNAP activity, as is evident from the findings that the structured mutant protein exerted enhanced binding affinity for the enzyme, probably by affecting the flexibility of the RNAP catalytic center. This tight binding affects the initiation of transcription and the first step of initiation, i.e., the binding of the σ-factors to the catalytic core. Overall, the results presented here indicate that flexible ω plays a significant role in σ-factor recruitment.

Author Contributions

D.B., N.B., and D.C. designed the experiment and conceptualized the problem; D.B. and N.B. carried out the experiments; D.B. and D.C. interpreted the results; and all three authors wrote the manuscript.

Editor: Elizabeth Rhoades.