The A12.2 Subunit Is an Intrinsic Destabilizer of the RNA Polymerase I Elongation Complex

Abstract

Despite sharing a highly conserved core architecture with their prokaryotic counterparts, eukaryotic multisubunit RNA polymerases (Pols) have undergone structural divergence and biological specialization. Interesting examples of structural divergence are the A12.2 and C11 subunits of Pols I and III, respectively. Whereas all known cellular Pols possess cognate protein factors that stimulate cleavage of the nascent RNA, Pols I and III have incorporated their cleavage factors as bona fide subunits. Although it is not yet clear why these polymerases have incorporated their cleavage factors as subunits, a picture is emerging that identifies roles for these subunits beyond providing nucleolytic activity. Specifically, it appears that both A12.2 and C11 are required for efficient termination of transcription by Pols I and III. Given that termination involves destabilization of the elongation complex (EC), we tested whether A12.2 influences stability of the Pol I EC. Using, to our knowledge, a novel assay to measure EC dissociation kinetics, we have determined that A12.2 is an intrinsic destabilizer of the Pol I EC. In addition, the salt concentration dependence of Pol I EC dissociation kinetics suggests that A12.2 alters electrostatic interactions within the EC. Importantly, these data present a mechanistic basis for the requirement of A12.2 in Pol I termination. Combined with recent work demonstrating the direct involvement of A12.2 in Pol I nucleotide incorporation, this study further supports the concept that A12.2 cannot be viewed solely as a cleavage factor.

Introduction

Whereas prokaryotic and eukaryotic multisubunit RNA polymerases (Pols) share a highly conserved core architecture, eukaryotic Pols have undergone structural divergence and biological specialization (1). Of specific interest is the relationship between Pols and their cognate cleavage factors, which are proteins that bind the polymerase and stimulate cleavage of the nascent RNA transcript. Despite significant diversity in their amino acid sequence, cleavage factors all appear to contain a structural element that enters and modulates the polymerase active site (2, 3). Importantly, eukaryotic Pols I and III have incorporated their cleavage factors as subunits known as “A12.2” and “C11,” respectively (3). Defining the functional contribution of these subunits to polymerase activities will provide a more detailed understanding of the specialized roles for eukaryotic RNA polymerases and potentially the selective pressures that drove their divergence.

We recently published a characterization of the nucleotide incorporation properties of a Pol I isoform that lacks the A12.2 subunit (ΔA12) (4). In that work we determined that, in addition to its role in Pol I-catalyzed RNA cleavage (5), A12.2 is directly involved in Pol I-catalyzed nucleotide incorporation. Combined with earlier studies that revealed roles for A12.2 in Pol I termination (6, 7), our recent work has painted a picture in which A12.2, and possibly other cleavage factors, have important roles that extend beyond RNA cleavage.

Here, we characterized the influence of A12.2 on the stability of the transcription elongation complex (EC). Although it was demonstrated that A12.2 is required for efficient termination of transcription by Pol I, the mechanism underlying this observation was not clear (7). Given that destabilization of the Pol I EC is required for termination (8), we asked whether A12.2 directly contributes to EC stability. By characterizing the kinetic stability of ECs formed by wild-type (WT) and ΔA12 Pol I, using, to our knowledge, a novel coupled EC dissociation assay, we have determined that A12.2 is an intrinsic destabilizer of the Pol I EC. Further, by measuring the salt-concentration dependence of WT and ΔA12 EC stability, we have determined that A12.2 modulates important electrostatic interactions within the EC. These data provide a mechanistic basis for the observed requirement of A12.2 in Pol I termination and add to the growing list of roles for A12.2 in Pol I transcription.

Methods

Proteins

ΔA12, and WT Pol I were purified from Saccharomyces cerevisiae as previously described (4, 9).

RNase A (catalog No. LS002132; Worthington Biochemical, Lakewood, NJ) was dissolved in 40 mM KCl, 40 mM TrisOAc pH 7.9 at 25°C and dialyzed into the same buffer with 20% glycerol in three 1:500 exchanges at 4°C using 3500–5000 MWCO cellulose ester dialysis tubing (Spectra/Por; Spectrum Labs, Los Angeles, CA). The concentration of the dialyzed protein was measured by spectroscopic assay in denaturing protein measurement buffer (6 M guanidinium hydrochloride, 0.05 M K₂PO₄ pH 6.48 at 25°C) (10). Absorbance was measured between 360 and 280 nm, and absorbance values at 280 nm were corrected for absorbance at 320 nm (absorbance due to scattering effects) by subtraction. An extinction coefficient of 8640 M⁻¹ cm⁻¹ was used to determine the concentration of RNase A.

EC dissociation assay

Elongation complexes were prepared in reaction buffer (40 mM KCl, 40 mM TrisOAc pH 7.9 at 25°C, 0.2 mg∙mL⁻¹ BSA) at ambient temperature as previously described (11), with the minor change that all reactions lacked DTT, which was withheld due to the requirement of disulfide bonds in RNase A (12). Therefore, in this work, all reactions lack DTT. All subsequent reaction steps were carried out at 25°C maintained using a Cool-Hotter Dry Bath Incubator (Major Science, Saratoga, CA).

As discussed in the text, it was found that under the reaction conditions previously reported by us (4, 11) ECs are very stable (lifetimes ≫ 24 h at 25°C, 40 mM KCl). To bring the EC dissociation kinetics into a measurable range we employed a salt-concentration jump strategy, where EC dissociation is initiated by rapidly increasing [KCl]. In this experimental strategy, ECs were prepared using our previously reported methods (11). ECs were transferred from ambient temperature to 25°C at the beginning of the 10-min EC labeling step. To initiate EC dissociation, elongation complexes were mixed 1:1 with a solution of 10 μM RNase A containing varying concentrations of KCl in reaction buffer. RNase A solution was incubated for ∼10 min at 25°C before mixing with the EC solution. Throughout the EC dissociation reaction, samples were incubated at 25°C. Time points were collected continuously by removing 5 μL aliquots of EC/RNase A mix and pipetting directly into 25 μL of gel loading buffer (90% formamide, 25 mM EDTA-Na₃, 0.025 mg/mL bromophenol blue). A “t = 0” point for each reaction was collected by directly mixing a 2.5 μL aliquot of labeled EC mixture into 25 μL loading buffer that contained 2.5 μL RNase A mixture. All mixing events were performed by manual pipetting.

RNase A-catalyzed RNA cleavage reactions were performed as described above except that ECs were denatured by incubating at ∼95°C for 5 min followed by a 5-min incubation at 25°C before mixing with RNase A. Samples were collected as a function of time, as described immediately above.

To separate reactant 10-mer RNA and product 7-mer RNA, samples were incubated at ∼80°C for 5 min and subjected to denaturing polyacrylamide gel electrophoresis using 28% (19:1) acrylamide/bis-acrylamide, 7 M urea, 1× TBE gels. Gels were exposed to phosphor-imager screens and quantified using a Typhoon scanner (GE Healthcare Life Sciences, Marlborough, MA) (11).

Data analysis

Exposed gels for all time courses were quantified according to Eq. 1,

$$Fra{c}_i\left(t\right)=\frac{\frac{\left[RN{A}_i\right]\left(t\right)}{\sum _i\left[RN{A}_i\right]\left(t\right)}-\frac{\left[RN{A}_i\right]\left(0\right)}{\sum _i\left[RN{A}_i\right]\left(0\right)}}{1-\frac{\left[RN{A}_i\right]\left(0\right)}{\sum _i\left[RN{A}_i\right]\left(0\right)}},$$ (1)

in which Frac_i(t) represents the fraction of RNA in the form i at time t and [RNA_i](t) is the phosphor-imager counts of the RNA representative of the RNA form i at time t. Time courses were collected in duplicate or triplicate and averaged. Error bars in Figs. 2 and 3 display the SD about the average.

Figure 2.

WT and ΔA12 EC dissociation time courses. Time courses were collected at 0.29 M KCl using the salt-jump strategy (see text). Circles represent the average of duplicate or triplicate measurements and error bars display the SD about the average. The WT EC dissociation time course is in blue and the ΔA12 EC dissociation time course is in red. To see this figure in color, go online.

Figure 3.

EC dissociation and RNase A-catalyzed RNA cleavage time courses collected as a function of [KCl]. (A) Native WT and ΔA12 EC dissociation time courses were fit to the reaction model given by Eq. 9, as well as RNase A-catalyzed RNA cleavage time courses fit to Eq. 9. (B) Native WT and ΔA12 EC dissociation time courses fit to the reaction model as given by Eq. 13, as well as RNase A-catalyzed RNA cleavage time courses fit to Eq. 13. Circles in all panels represent the average of at least two measurements of the fraction of RNA in the 7-mer (cleaved) state and error bars represent the SD about the average. Solid lines represent simultaneous pairwise (EC dissociation and RNase A-catalyzed RNA cleavage) fits to Eq. 9 (A) or Eq. 13 (B), as described in the text and Methods.

Due to the coupled nature of our EC dissociation assay, the observed time courses are inherently a convolution of the reaction of interest (EC dissociation) and the coupling reaction (RNase A-catalyzed RNA cleavage). By using the fitting approach described below and the RNase A-catalyzed RNA cleavage time courses collected at each [KCl], we were able to specifically extract EC dissociation kinetics.

From the fits in Fig. 3, it is clear that all WT and ΔA12 data sets, except WT 0.44 M KCl and WT 0.54 M KCl, cannot be described by the reaction model given by Eq. 9. Based on an F-test comparing fits of the WT 0.54 M KCl data set to the reaction models given by Eqs. 9 and 13, it was determined that Eq. 13 provides a significantly better fit. Therefore, all data sets with the exception of WT 0.44 M KCl were fit to the reaction model given by Eq. 13.

Each EC dissociation time course was simultaneously fit with its cognate (i.e., same [KCl]) RNase A-catalyzed RNA cleavage time course to the R₇(t) solutions of Eq. 2, where the dot over S denotes the time derivative:

$$\dot{\mathbf{S}}=\mathbf{K}\mathbf{S},$$ (2)

$$\mathbf{S}=\left\{\begin{array}{c}{\left[Pol•R_{10}\right]}_f\left(t\right)\\ {\left[Pol•R_{10}\right]}_s\left(t\right)\\ Pol\left(t\right)\\ R_{10}\left(t\right)\\ R_7\left(t\right)\end{array}\right\},$$ (3)

$$\mathbf{K}=\left\{\begin{array}{ccccc}-k_f& 0& 0& 0& 0\\ 0& -k_s& 0& 0& 0\\ k_f& k_s& 0& 0& 0\\ k_f& k_s& 0& -k_{\text{RNAse}}& 0\\ 0& 0& 0& k_{\text{RNAse}}& 0\end{array}\right\}.$$ (4)

Solutions of Eq. 2 were obtained by using Eq. 5:

$$s_i\left(t\right)=\sum _{j=1}{c}_j{\mathbf{V}}_{ij}{e}^{{\lambda }_{j}t},$$ (5)

in which s_i represents the ith element of S, c_j are constants, V is a matrix with columns that are the eigenvectors of K, and λ_j is the jth eigenvalue of K. The eigenvectors and eigenvalues of K were obtained using the MATLAB “eig” function (The MathWorks, Natick, MA). The c values were obtained by evaluating Eq. 5 at t = 0 using Eq. 6,

$${\mathbf{S}}^{\circ }=\mathbf{S}\left(0\right)=\left\{\begin{array}{c}{\left[Pol•R_{10}\right]}_f^{\circ }\\ {\left[Pol•R_{10}\right]}_s^{\circ }\\ 0\\ R_{10}^{\circ }\\ 0\end{array}\right\},$$ (6)

which gives the initial values of each element of S. Solutions of Eq. 5 for R₇(t) obtained in this manner can then be evaluated at any set of initial ${\left[Pol•R_{10}\right]}_f$, ${\left[Pol•R_{10}\right]}_s$, and $R_{10}$ values. This allows the fraction of total elongation complex in the ${\left[Pol•R_{10}\right]}_f$ or ${\left[Pol•R_{10}\right]}_s$ form to be floated in a weighted nonlinear least squares fit. Further, R₇(t) obtained in this way can be used to simultaneously fit EC dissociation time courses, in which ${\left[Pol•R_{10}\right]}_f$ and ${\left[Pol•R_{10}\right]}_s$ assume finite values and $R_{10}^{\circ }$ = 0, and RNase A-catalyzed RNA cleavage time courses, in which ${\left[Pol•R_{10}\right]}_f$ = ${\left[Pol•R_{10}\right]}_s$ = 0 and $R_{10}^{\circ }$ = 1. In fitting time courses, both R₇(t) solutions (corresponding to EC dissociation time courses and RNase A-catalyzed RNA cleavage time courses) were scaled by constants to account for each time course not reaching a value of 1. For R₇(t) solutions corresponding to EC dissociation time courses (i.e., $R_{10}^{\circ }$ = 0), the total EC population was assigned a value of 1 and ${\left[Pol•R_{10}\right]}_f$ and ${\left[Pol•R_{10}\right]}_s$ were calculated according to Eqs. 7 and 8, respectively:

$${\left[Pol•R_{10}\right]}_f^{\circ }=\alpha ,$$ (7)

$${\left[Pol•R_{10}\right]}_s^{\circ }=1-\alpha ,$$ (8)

where the value α was floated in the fitting routine and reports on the fraction of the initial EC population in the ${\left[Pol•R_{10}\right]}_f$ state.

WT and ΔA12 time courses collected at each [KCl] were fit simultaneously with the RNase A-catalyzed RNA cleavage time course collected at the same [KCl]. The residuals in the fit were weighted by the SD obtained from duplicate or triplicate measurements of each time point. Fitting was accomplished using custom-written MATLAB scripts and the MATLAB program “Lsqnonlin”. Confidence intervals on the fitted parameter values were calculated using a grid searching approach previously described by us (11).

Fig. 4 B displays the log₁₀ values of k_f and k_s plotted as a function of log₁₀([KCl]). Log₁₀ transforms were taken for rate constant and [KCl] values in s⁻¹ and M, respectively. These log₁₀ × log₁₀ data were subjected to unweighted fits to lines using Kaleidagraph. The error bars displayed in Fig. 4 C correspond to mean ± SE on the fitted slope provided by Kaleidagraph.

Figure 4.

[KCl] dependence of EC dissociation kinetics. (A) α-values reporting on the fraction of the EC population in the fast dissociating state (see Eq. 13), from fits shown in Fig. 3B, plotted as a function of [KCl]. (B) log₁₀(k_f) and log₁₀(k_s) obtained from fits shown in Fig. 3B plotted as a function of log₁₀([KCl]); rate constants and [KCl] are reported in units of s⁻¹ and M, respectively. Solid lines represent fits to lines. In (A) and (B), circles represent best-fit values and error bars represent the 68% confidence interval obtained from grid search analysis (see Methods). (C) Slopes of fitted lines in (B) are given as (n). Error bars represent ± SE of fitted slopes (see Methods).

Results

Developing an assay for EC stability

To measure WT and ΔA12 Pol I EC stability, we developed a ribonuclease(RNase)-coupled assay to measure dissociation kinetics of the nascent RNA from the EC. Fig. 1 A displays a cartoon schematic of the RNase-coupled EC dissociation assay. This assay relies on the ability of the polymerase to protect the nascent RNA from RNase-catalyzed cleavage in the context of an EC. ECs are reconstituted as described previously (11). This is performed in three steps. In step 1, the RNA component of the EC is labeled by Pol I-catalyzed incorporation of a single ³²P-labeled cytosine monophosphate. In step 2, the labeling reaction is quenched by the addition of EDTA and unlabeled cytosine triphosphate. (Some important points must be noted regarding steps 1 and 2: each RNA is singly labeled on its 3′-terminus; addition of unlabeled CTP and EDTA to stop the labeling reaction ensures that the amount of labeled RNA does not change during the EC dissociation reaction; and requiring the polymerase to label the RNA ensures that the assay is only sensitive to active ECs (11).) In step 3, the labeled EC mixture is combined with an excess (with respect to polymerase and nucleic acids) of ribonuclease A (RNase A). Inclusion of RNase A couples EC dissociation to RNA cleavage and therefore the fraction of cleaved RNA reports on the fraction of the EC population that has dissociated.

Figure 1.

Development of an RNase-coupled EC dissociation assay. (A) Schematic of the RNase-coupled assay. See Methods for detailed description of the assay. (B) Sequence of RNA and site of RNase A-catalyzed cleavage is shown. Red C represents cytosine monophosphate incorporated by Pol I during the labeling reaction. (C) Denaturing PAGE separation of WT EC dissociation time course collected at 0.29 M KCl using the salt-jump strategy (see text). (D) Same as in (C), except ECs were heated 5 min at 95°C and cooled to 25°C before the salt-jump. To see this figure in color, go online.

RNase A was chosen as a coupling enzyme based on its substrate specificity and divalent cation-independent reaction mechanism. Given the sequence of the RNA substrate used to reconstitute ECs, it is anticipated that reaction of the RNA with RNase A will specifically produce unlabeled 3-mer and radiolabeled 7-mer cleavage products (Fig. 1 B) (12). RNase A does not require a divalent cation and therefore should remain active in the presence of EDTA, which is used in our EC reconstitution protocol (12).

Fig. 1, C and D, displays representative denaturing PAGE analyses of two EC dissociation time courses. The leftmost lane in each gel image represents “t = 0” material prepared by mixing an aliquot of labeled EC solution with RNase A solution that has been premixed, and therefore inactivated, with quench solution. Each subsequent lane moving from left to right represents a time point collected after combining labeled ECs with RNase A. Time points were collected continuously by mixing an aliquot of the EC–RNase A mixture with quench solution (see Methods). Fig. 1 C displays a WT EC dissociation reaction carried out as schematized in Fig. 1 A. Fig. 1 D displays a WT EC dissociation reaction identical to that depicted in Fig. 1 C except that the EC was heat-denatured before adding RNase A to fully dissociate 100% of the ECs. In each reaction, 10-mer RNA is converted to 7-mer as time progresses. By comparing the RNA cleavage kinetics between the two reactions, it is clear that an intact EC protects the RNA from RNase A-catalyzed cleavage whereas a heat-denatured EC does not protect the RNA from cleavage. Thus, this approach is suitable for measuring EC dissociation kinetics. Furthermore, these raw data demonstrate that WT Pol I forms long-lived ECs.

Defining WT and ΔA12 EC stability

Initial measurements of WT and ΔA12 EC dissociation reactions using the RNase-coupled assay described above, collected under our standard, low-salt nucleotide incorporation reaction conditions (11), revealed that the dissociation rate constants of both WT and ΔA12 ECs were effectively zero (data not shown). These slow EC dissociation kinetics observed under our standard reaction conditions prohibited quantitative comparison between the WT and ΔA12 EC dissociation reactions. To increase the EC dissociation rate constants to a level amenable to quantitative measurement, we pursued a salt-jump strategy in which [KCl] was rapidly increased concurrently with the addition of RNase. The increase in [KCl] effectively initiates the EC dissociation reaction, which is coupled to RNase A-catalyzed cleavage of the RNA as described above. Notably, Record and co-workers have successfully employed a [KCl]-jump strategy to characterize the kinetics of Escherichia coli RNA polymerase open complex formation and disassembly (13). Fig. 2 displays a plot of the fraction of RNA in the 7-mer (RNase-cleaved) state as a function of time from WT and ΔA12 EC dissociation reactions collected at 0.29 M KCl using the salt-jump strategy. The data in Fig. 2 indicate that even at a relatively high salt concentration, both WT and ΔA12 form stable ECs. However, it is clear that WT dissociates much faster than ΔA12. The time courses in Fig. 2 reveal a striking difference in the kinetic stability of WT and ΔA12 ECs, and indicate that loss of the A12.2 subunit modifies protein-protein and/or protein-nucleic acid contacts within the Pol I EC that influence kinetic stability of the EC.

To probe the importance of electrostatic interactions in the maintenance of EC kinetic stability and to test if these interactions are modulated by A12.2, we measured EC dissociation kinetics of WT and ΔA12 at three additional [KCl]. In addition, to ensure that observed EC dissociation kinetics do not become rate-limited by the coupled RNase A reaction, we measured RNase A-catalyzed cleavage of RNA derived from heat-denatured ECs over the same [KCl] range. These experiments were performed using the salt-jump strategy described above.

WT and ΔA12 EC dissociation time courses and RNase A-catalyzed RNA cleavage time courses collected at all [KCl] are plotted in Fig. 3. Consistent with the time courses presented in Fig. 2, a striking difference is revealed in the EC dissociation kinetics between WT and ΔA12 at all [KCl]. Qualitatively, these data demonstrate a large contribution of the A12.2 subunit to destabilizing transcription elongation complexes formed by Pol I.

Quantitative analysis of WT and ΔA12 EC stability

Whereas the data in Fig. 3 A clearly demonstrate large differences in the stabilities of WT and ΔA12 ECs, the RNase A-catalyzed RNA cleavage time courses reveal that the coupling reaction relied on by our assay is sensitive to [KCl]. To deconvolute the observed kinetics into the contributions from EC dissociation and RNase A-catalyzed RNA cleavage, we simultaneously fit the EC dissociation and RNase A-catalyzed RNA cleavage time courses at each [KCl] using the approach described below.

The simplest reaction model that accounts for the design of our coupled assay is given by Eq. 9:

$$\begin{array}{c}Pol\\ +\\ Pol•R_{10}\stackrel{k_{\mathit{EC}}}{\to }{R}_{10}\stackrel{k_{\mathit{RNAse}}}{\to }{R}_7+R_3,\end{array}$$ (9)

where $Pol•R_{10}$ denotes the EC bound to the 10-mer RNA starting material; Pol and R₁₀ denote polymerase and 10-mer RNA, respectively, which have dissociated from the EC; and R₇ and R₃ denote 7-mer and 3-mer RNA produced by RNase A-catalyzed cleavage of the 10-mer RNA, respectively. Recall that as a result of our RNA labeling scheme, only the 10-mer and 7-mer RNA species are visible to us. The EC dissociation kinetics and RNase A-catalyzed RNA cleavage kinetics are described by the two observed rate constants k_EC and k_RNAse, respectively. The ordinary differential equations defining $Pol•R_{10}$, R₁₀, and R₇ from Eq. 9 are given by Eqs. 10, 11, and 12.

$$\frac{dPol•R_{10}\left(t\right)}{dt}=-\frac{dPol\left(t\right)}{dt}=-k_{\text{EC}}Pol•R_{10}\left(t\right),$$ (10)

$$\frac{d{R}_{10}\left(t\right)}{dt}=k_{\text{EC}}Pol•R_{10}\left(t\right)-k_{\text{RNAse}}{R}_{10}\left(t\right),$$ (11)

$$\frac{d{R}_7\left(t\right)}{dt}=k_{\text{RNAse}}{R}_{10}\left(t\right).$$ (12)

To fit the time courses displayed in Fig. 3 A, Eqs. 10, 11, and 12 were solved to obtain R₇(t) using two different sets of initial conditions (ICs):

$$\text{IC}1=\left\{Pol•R_{10}\left(0\right)=1,Pol\left(0\right)=R_{10}\left(0\right)=R_7\left(0\right)=0\right\};$$

$$\text{IC}2=\left\{R_{10}\left(0\right)=1,Pol•R_{10}\left(0\right)=Pol\left(0\right)=R_7\left(0\right)=0\right\}.$$

IC1 corresponds to the experiments using WT or ΔA12 ECs, whereas IC2 corresponds to the experiments using heat-denatured ECs. This strategy was used to perform simultaneous pairwise fits of EC dissociation and RNase A-catalyzed RNA cleavage time courses at each [KCl]. The results of these fits are displayed as solid lines in Fig. 3 A. The systematic deviations observed between the data and best-fit lines in Fig. 3 A clearly reveal that the reaction model given by Eq. 9 cannot adequately describe the time-course data. This result indicates that there are processes that occur sequentially or concurrently with the two steps described by Eq. 9.

With the goal of providing a phenomenological description of the EC dissociation kinetics, we have chosen to use the above described simultaneous pairwise fitting strategy to fit the data using the reaction model given by Eq. 13 (see Methods for a detailed description of the Eq. 13 fitting procedure):

$$\begin{array}{ccc}{\left\{Pol•R_{10}\right\}}_f& & \\ & \searrow k_f& \\ & & Pol+R_{10}\stackrel{k_{\mathit{RNAse}}}{\to }Pol+R_7\\ & \nearrow k_s& \\ {\left\{Pol•R_{10}\right\}}_s& & \end{array}.$$ (13)

The reaction model given by Eq. 13 is identical to that given by Eq. 9 except that the EC population is divided into two subpopulations, ${\left\{Pol•R_{10}\right\}}_f$ and ${\left\{Pol•R_{10}\right\}}_s$, with fast and slow dissociation kinetics, defined by the rate constants k_f and k_s, respectively. The logic for using Eq. 13 as a model to fit the data is as follows: The RNase A-catalyzed RNA cleavage time courses are well described by Eq. 9 (Fig. 3 A) and are well described by single exponential functions (data not shown). These data indicate that multiple steps are not needed to describe the conversion of dissociated RNA to cleaved RNA in our coupled assay. Further, we have previously determined that, under our reaction conditions, WT and ΔA12 Pol I EC populations are composed of at least two predominant components with distinct behaviors (4, 11).

The results of simultaneous pairwise fitting of the EC dissociation time courses and the RNase A-catalyzed RNA cleavage time courses to the reaction model given by Eq. 13 are displayed in Fig. 3 B. The quality of these fits indicate that the data are well described by Eq. 13 and indicate that this fitting strategy can be used to extract quantitative values describing WT and ΔA12 EC dissociation kinetics.

It is important to note that in fitting the WT EC dissociation time course collected at 0.44 M KCl to Eq. 13, constraint was lost on k_f. For this reason, the WT EC dissociation time course, and the cognate RNase A-catalyzed RNA cleavage time course, collected at 0.44 M KCl were fit to the reaction model given by Eq. 9 to obtain a single observed EC dissociation rate constant that we are reporting as k_s.

Fig. 4 A shows the α-values obtained from the fits presented in Fig. 3 B, reporting on the fraction of the EC population in ${\left\{Pol•R_{10}\right\}}_f$, plotted as a function of [KCl]. In Fig. 4 B the observed rate constants obtained from the fits presented in Fig. 3 B are displayed as a function of [KCl]. To visualize the observed rate constant values, which span multiple orders of magnitude, the observed rate constants (in units of s⁻¹) were log₁₀-transformed and are plotted as a function of log₁₀-transformed [KCl] (in units of M).

The data in Fig. 4 A indicate that there are differences in the relative representation of ${\left\{Pol•R_{10}\right\}}_f$ and ${\left\{Pol•R_{10}\right\}}_s$ isoforms between WT and ΔA12 EC populations. First, although there is considerable scatter in the limited data set, there is no trend in WT α-values across the [KCl] range. This is in contrast to the ΔA12 EC population, which displays a small but clear increase in α-values as [KCl] is increased. Finally, all WT α-values are significantly higher than ΔA12 α-values across the [KCl] range, indicating that WT ECs are composed of a larger portion of fast dissociating complexes as compared to ΔA12 ECs.

It is immediately clear from Fig. 4 B that there is a large difference in the EC dissociation kinetics between WT and ΔA12. Specifically, the slower dissociating EC component (${\left\{Pol•R_{10}\right\}}_s$ in Eq. 13, dissociating with observed rate constant k_s) displays a difference in kinetic stability of over two orders of magnitude between WT and ΔA12 across the experimental [KCl] range (compare red circles to black circles). In addition to the differences in kinetic stability observed at each [KCl], the data in Fig. 4 B indicate that the EC dissociation kinetics of both EC populations (${\left\{Pol•R_{10}\right\}}_f$ and ${\left\{Pol•R_{10}\right\}}_s$) formed by WT and ΔA12 each display unique dependencies on [KCl]. For example, compare in Fig. 4 B the WT k_s values observed across the assayed [KCl] range (red circles) to the ΔA12 k_s values observed across the same [KCl] range (black circles). The WT k_s values span nearly two orders of magnitude whereas the ΔA12 k_s values are virtually invariant. To compare how the k_obs values obtained from each enzyme vary as a function of [KCl], the log₁₀(k_obs) versus log₁₀([KCl]) data were fit to lines (Fig. 4 B, solid lines) and the slopes from these fits are presented in the form of a bar graph in Fig. 4 C.

In Fig. 4 C it is observed that the dependence of both k_f and k_s on [KCl] differ between WT and ΔA12. Slopes of the log₁₀(k_obs) versus log₁₀([KCl]) data sets differ significantly for k_f and k_s between the two enzymes, with WT displaying a steeper [KCl]-dependence in each parameter. Finally, Fig. 4 C demonstrates that for both WT and ΔA12, k_s displays a steeper [KCl] dependence than k_f.

The data in Fig. 4, B and C, demonstrate that A12.2 acts as an intrinsic destabilizer of the Pol I EC. Consistent with our previous work, ECs reconstituted using both WT and ΔA12 Pol I are composed of two dominant populations with distinct biochemical attributes (4, 11). Removal of A12.2 from Pol I results in significant stabilization of each EC subpopulation. Further, stability of ΔA12 ECs displays a markedly different salt concentration dependence in comparison to WT EC stability. Importantly, this change in salt sensitivity is strong evidence that A12.2 exerts its destabilization function, at least in part, by modulating key electrostatic interactions within the EC. Although identification of the specific interactions that give rise to A12.2-modulated EC stability remains an open question, the data presented here provide a plausible mechanism for the observed role for A12.2 in Pol I termination. Specifically, A12.2 is required to physically destabilize the Pol I EC to enable efficient termination of transcription and clearance from the ribosomal DNA. This observation is consistent with previous studies performed in both S. cerevisiae and Schizosaccharomyces pombe that implicate A12.2 in transcription termination by Pol I (6, 7).

Discussion

It has long been recognized that cellular Pols are capable of forming long-lived ECs (14). Further, modulation of [KCl] has been employed extensively to characterize factors that affect RNA polymerase-nucleic acid interactions as well as to specifically identify and isolate properly formed ECs (13, 14, 15, 16, 17, 18). These previous studies demonstrated the utility of direct reconstitution of ECs using scaffold template-based approaches similar to those used here in defining specific factors the affect EC stability (15, 16). To characterize EC stability, a variety of methods to assay the state of the EC have been employed including monitoring the release of nucleic acids from immobilized ECs, separation of polymerase and nucleic acids from dissociated ECs by size exclusion chromatography, and monitoring extension competency of a given RNA species upon addition of NTP substrates (14, 15, 16). Here we have employed a strategy that does not require EC immobilization, separation of protein and nucleic acid components, or discontinuous assay of RNA extension. The coupled EC dissociation assay described in this work permits time-resolved simultaneous quantification of EC-bound and EC-dissociated RNA and enables collection of quantitative EC dissociation time courses over a range of solution conditions, including high [KCl]. Using this approach we have determined that the A12.2 subunit is an intrinsic destabilizer of the Pol I EC. Although it may seem counterintuitive that a subunit of Pol I would have roles in destabilizing the EC, our current findings are consistent with recent structural studies of the Pol I EC as well as with what is known regarding involvement of A12.2 in Pol I termination (7, 19, 20).

Ribosomal DNA genes (rDNA) are present in 150–200 tandemly arrayed copies in S. cerevisiae (yeast) (8). In yeast, each rDNA repeat can be divided into two main regions: Beginning at the Pol I promoter and moving 5′ to 3′ along the nontranscribed strand, the first region corresponds to the Pol I-transcribed 35S precursor rDNA and the second region corresponds to the Pol III-transcribed 5S rDNA (8). Importantly, the Pol I and Pol III promotors are on opposite DNA strands, resulting in convergent transcription and the possibility of collisions between the two transcription complexes (8). Due to the high levels of Pol I transcription and the genomic arrangement of the rDNA, efficient Pol I and III termination is critical to avoid polymerase collisions. Although the details of the Pol I termination mechanism are not fully described, it is accepted that termination involves polymerase pausing followed by or concomitant with protein factor-dependent destabilization of the EC (6, 7, 8, 21, 22). Therefore, factors that influence EC stability should alter Pol I termination efficiency. Importantly, the data presented in this work clearly demonstrate that the A12.2 subunit destabilizes the EC and provide a mechanistic explanation for the involvement of A12.2 in Pol I termination (7). However, these data do not exclude potential additional roles for DNA binding proteins like Reb1 or Nsi1 in the natural termination mechanism (6, 21, 22).

Whereas all known cellular Pols possess cognate cleavage factors, it appears that Pols I and III have incorporated their cleavage factors as bona fide subunits (A12.2 and C11 in Pols I and III, respectively) (3). Although it is not clear why Pols I and III have evolved such tight interactions with their cleavage factors, these data indicate that at least one role for this interaction in the Pol I system is destabilization of the EC. Importantly, it has been demonstrated that C11 is involved in Pol III termination (23, 24, 25, 26). Consistent with C11 providing intrinsic EC destabilization, it was demonstrated in vitro that the termination defects of a Pol III isoform lacking C11 could largely be rescued by increasing the number of U-A basepairs in the EC RNA-DNA hybrid (24). It is established that the EC RNA-DNA hybrid is a critical determinant of EC stability (15, 16) and therefore it appears that in the context of Pol III termination, EC destabilization afforded by C11 can at least partially be replaced by EC destabilization resulting from a weak RNA-DNA hybrid. Studies of the Pol III/C11 system similar to those presented here would provide a direct test of C11’s role in intrinsic EC destabilization.

In addition to the observed effects of A12.2 on Pol I termination in vivo, roles for A12.2 in the modulation of Pol I EC stability are supported by recent structural data. Cryo-electron microscopy-derived structures of Pol I reveal that the Pol I DNA binding cleft undergoes progressive closure in the transition from the monomeric nucleic acid-free apo state to the nucleic acid-bound open complex and elongation complex states (20). These structures indicate that closure of the DNA binding cleft is accompanied by and requires displacement of the C terminus of A12.2 from the Pol I secondary channel and active site. From these structural data, it is anticipated that closure of the Pol I DNA binding cleft has major impacts on the stability of the RNA-DNA hybrid and the EC as a whole (19, 20).

In reconciling these structural findings with our current results, the following simple model for the involvement of A12.2 in the regulation of Pol I EC stability emerges: The Pol I DNA binding cleft lies in a dynamic equilibrium populated by multiple states ranging from the fully open state observed in the apo Pol I structure to the fully closed state observed in the Pol I EC structure. The macroscopic observables studied here, the EC dissociation rate constants, are determined by the populations and stabilities of these various EC states. Insertion of the C-terminal domain of A12.2 into the Pol I secondary channel and active site is incompatible with full closure of the DNA binding cleft and achievement of what are presumably the most stable EC states. Thus, A12.2 is linked to the cleft opening/closing equilibrium and this linkage enables A12.2 to modulate EC stability. We hypothesize that complete removal of A12.2 results in a DNA binding cleft conformational distribution that is highly skewed toward closed states, causing stabilization of the EC.

Our data reveal that both WT and ΔA12 EC dissociation kinetics are well described by the presence of two polymerase populations. Given the above discussion, the slower dissociating population presumably reports on DNA binding-cleft-closed states. Assignment of this slow dissociating population to DNA binding-cleft-closed states is consistent with the increased [KCl] sensitivity of k_s as compared to k_f for each polymerase (see Fig. 4 C). Specifically, the Pol I structures imply enhancement of protein-nucleic acid interactions in going from open DNA binding cleft to closed DNA binding cleft conformations (19, 20). Therefore, more protein-nucleic acid ionic bonds must be broken when an EC dissociates from a closed DNA binding cleft conformation as compared to dissociation from an open DNA binding cleft conformation. The relative [KCl] sensitivities of k_s and k_f indicate that more ions are taken up from solution during the process reported on by k_s as compared to that reported on by k_f, consistent with k_s reporting on dissociation of the EC from DNA binding-cleft-closed conformations.

Finally, the slow dissociating ΔA12 EC population displays a significantly reduced salt sensitivity as compared to its WT counterpart. These data indicate that formation of stabilized, presumably DNA binding-cleft-closed, ΔA12 EC conformations rely less on protein-nucleic acid interactions than in the WT enzyme. This idea is consistent with the above discussion of modulation of the DNA binding cleft equilibrium by A12.2. Closure of the ΔA12 DNA binding cleft results predominantly from the skewed DNA binding cleft conformational distribution, whereas WT DNA binding cleft closure requires more extensive protein-nucleic acid contacts.

It is now clear from both structural and functional studies that the A12.2 subunit has critical roles in multiple steps of the Pol I transcription cycle (4, 8, 19, 20, 27). This work adds to the growing list of roles for A12.2 in Pol I activity and, given the conservation between Pol - cleavage factor interactions, should stimulate further study of the role(s) for cleavage factors in RNA polymerase activity. In particular, it will be interesting to see if the C11 subunit of Pol III serves a similar role in EC stability as that observed for A12.2.

Author Contributions

F.D.A. designed research, analyzed data, and wrote the manuscript. C.E.S. designed research, performed research, analyzed data, and wrote the manuscript. A.L.L. designed research, analyzed data, and wrote the manuscript. D.A.S. designed research, analyzed data, and wrote the manuscript.

Editor: Karin Musier-Forsyth.