Molecular Basis of pH-Dependent Activity in T7 RNA Polymerase Revealed by Constant pH Replica Exchange Molecular Dynamics

Abstract

T7 RNA polymerase (T7 RNAP) catalyzes in vitro transcription (IVT) for mRNA vaccines and therapeutics, yet the molecular basis of its pH-dependent activity remains unclear. Here we employ constant pH replica exchange molecular dynamics (pH-REMD) spanning pH 4.75–9.15 (2.88 μs total sampling) to characterize pH-dependent conformational dynamics of the T7 RNAP initiation complex. Our simulations reveal that His811, adjacent to catalytic Asp812, functions as the primary active site pH sensor (pK _a 5.9). Protonation of His811 at acidic pH creates an electrostatic trap that restricts conformational flexibility required for catalytic turnover, while its neutral state at optimal pH (∼8.0) permits dynamic motions essential for nucleotide incorporation. Two critical structural parameters, C-terminal Ala883 to active site distance and specificity loop to β-hairpin spacing, exhibit opposing pH-dependencies that converge at optimal pH to establish productive catalytic geometry. The Y639-RNA 3′-end distance analysis shows that optimal pH maintains the functionally open conformation required for binding by the incoming nucleotide. In contrast, acidic and basic pH conditions have shorter Y639-RNA 3′-end distances, promoting collapsed geometries that sterically occlude binding of the incoming nucleotide. Energy decomposition analysis reveals that C-terminal repositioning is driven by a pH-dependent switch in the electrostatic coupling hierarchy from protonated His545 and His790 at acidic pH to catalytic metal ions at optimal pH. Glu148 (pK _a 5.3) emerges as a structural pH sensor stabilizing the specificity loop–subdomain H interaction critical for elongation complex formation. Aggregate charge analysis reveals that excessive positive charge at acidic pH (+6 at pH 4) impedes promoter escape, while reduced charge at optimal pH (+2 at pH 8) balances binding and release. These insights establish important electrostatic and dynamic features that promote effective catalysis and can guide optimization of IVT conditions for the manufacturing of mRNA therapeutics and inform engineering of pH-stable T7 RNAP variants.

Introduction

T7 RNA polymerase (T7 RNAP) is a single-subunit enzyme encoded by bacteriophage T7 that has become indispensable in molecular biology and biotechnology. Unlike multisubunit bacterial RNA polymerases, T7 RNAP functions autonomously without auxiliary transcription factors, exhibits high specificity for its cognate promoter, and produces long transcripts with remarkable efficiency. , These properties have made T7 RNAP the enzyme of choice for in vitro transcription (IVT), heterologous gene expression systems, and emerging applications in synthetic biology and mRNA therapeutics. ,

The critical importance of T7 RNAP has been dramatically underscored by the COVID-19 pandemic, which propelled mRNA vaccine technology to the forefront of modern medicine. T7 RNAP serves as the catalytic engine for industrial IVT, the cell-free enzymatic process used to synthesize the mRNA component in all regulatory-approved mRNA vaccines, including those developed by Pfizer-BioNTech and Moderna. , The enzyme’s ability to produce large quantities of high-fidelity RNA transcripts, combined with its capacity to incorporate modified nucleotides such as N1-methylpseudouridine that reduce immunogenicity and enhance stability, proved instrumental in enabling rapid vaccine production at unprecedented speed and scale. , The mRNA platform technology is now advancing development of hundreds of vaccines and therapeutics against a wide range of diseases, including infectious diseases, cancers, immune disorders, rare diseases, and cardiovascular diseases. ,

From an economic and manufacturing perspective, the T7 RNAP-catalyzed process represents a substantial component of mRNA production costs, accounting for approximately 34% of the overall raw material cost for mRNA manufacturing. During the pandemic, commercial suppliers were unable to meet the increased global demand for this critical enzyme, highlighting vulnerabilities in the vaccine production supply chain and emphasizing the need for optimized enzyme production and deeper mechanistic understanding. , As the mRNA vaccine field expands into personalized cancer immunotherapy, with over 500 clinical trials − currently underway and the first commercial approvals anticipated by 2029, the demand for high-quality T7 RNAP is projected to increase substantially. , Understanding the molecular determinants of T7 RNAP activity and stability therefore has direct implications for improving vaccine manufacturing efficiency and reducing production costs.

The 98 kDa T7 RNAP comprises an N-terminal domain (residues 1–325) responsible for promoter recognition and a C-terminal polymerase domain (residues 326–883) that adopts the characteristic right-hand fold common to single-subunit polymerases. The polymerase domain is further organized into thumb (residues 326–411), palm (residues 412–553 and 785–879), and fingers (residues 554–784) subdomains, with the catalytic center residing in a deep cleft formed by these structural elements. Two magnesium ions coordinated by conserved aspartate residues (Asp537 and Asp812) mediate the phosphoryl transfer reaction essential for RNA synthesis. ,

Several structural elements important for T7 RNAP function have been identified through crystallographic and biochemical studies. The AT-rich recognition loop (residues 93–101) contacts the upstream promoter region, while an intercalating β-hairpin (residues 232–242) separates the template and nontemplate DNA strands during transcription initiation. The specificity loop (residues 740–769) in the fingers subdomain provides sequence-specific promoter recognition, with Asn748 making a critical major groove contact. The C-terminal carboxylate group of Ala883 has been proposed to coordinate active site Mg²⁺ ions; deletion or modification of the C-terminal carboxylate group abolishes Mg²⁺-dependent transcription, establishing this residue as a structural requirement for catalytic metal coordination. , These structural elements function as an integrated system. The specificity loop, AT-rich recognition loop, and intercalating β-hairpin cooperate during initiation to recognize and melt the promoter, while the transition to elongation requires coordinated repositioning of the N-terminal domain and specificity loop to establish the RNA exit tunnel.

The catalytic activity of T7 RNAP exhibits pronounced pH dependence, with optimal activity observed in the range of pH 8–9 and substantially diminished activity below pH 6 or above pH 10. , Despite the well-documented pH-activity profile, the molecular basis for this behavior remains incompletely understood. A previous computational study employing conventional molecular dynamics simulations at fixed protonation states (pH 5, 7, and 7.9) suggested that pH-dependent conformational changes, particularly in the positioning of the C-terminal region relative to the catalytic center, may contribute to the observed activity differences. However, the use of static protonation states and the limited number of examined pH conditions precluded a detailed mechanistic understanding.

Recent computational advances have provided unprecedented mechanistic insights into T7 RNAP function during transcription elongation. Extensive all-atom molecular dynamics simulations combined with Markov state modeling revealed that translocation proceeds via a Brownian ratchet mechanism, with the O-helix and Y-helix in the fingers domain playing key roles in facilitating nucleic acid movement. Complementary studies employing umbrella sampling calculations have elucidated the thermodynamic and kinetic determinants of nucleotide selectivity during the preinsertion to insertion transition, establishing the molecular basis for T7 RNAP fidelity control. These computational investigations, focused primarily on the elongation phase at near-neutral pH, have established T7 RNAP as a tractable model system for understanding single-subunit polymerase mechanisms. However, the pH-dependent regulation of the initiation complexcritical for the transition to productive elongation in biotechnological applicationshas not been systematically investigated using methods that capture dynamic protonation equilibria. Proper binding of the promoter to start the initiation phase is important for product yield and proper dynamic processes at the end of the initiation phase (as the system transitions to the elongation phase) are necessary for product purity to avoid production of toxic impurities. This study focuses on how pH regulates this processing in the initiation phase.

The active site of T7 RNAP contains several titratable residues whose protonation states could influence catalytic efficiency. , His811, located adjacent to the catalytic Asp812, is of particular interest given that histidine side chains typically titrate in the physiologically relevant pH range (pK _a ∼ 6–7). The protonation state and tautomeric form of active site histidines have been shown to modulate enzyme function in numerous systems, yet this aspect has not been systematically investigated for T7 RNAP. ,

Constant pH molecular dynamics (CpHMD) methods have emerged as powerful tools for investigating pH-dependent biomolecular behavior by allowing protonation states to fluctuate dynamically during simulation. When combined with replica exchange approaches, these methods enable efficient sampling across broad pH ranges while capturing coupled conformational and protonation state changes. Such approaches are particularly valuable for enzymes like T7 RNAP where multiple titratable residues may collectively influence structure and function.

In this study, we employ constant pH replica exchange molecular dynamics (pH-REMD) to investigate the pH-dependent conformational dynamics of T7 RNAP in complex with its promoter DNA and nascent RNA transcript. Using 18 replicas spanning pH 4.75 to 9.15 and a total simulation time of 2.88 μs, we provide an unprecedented view of how pH modulates the structure, flexibility, and functional competence of this important enzyme. Our results reveal that His811 protonation and tautomerism serve as key determinants of C-terminal positioning, we identify distinct pH-dependent flexibility profiles in promoter-interacting regions, and we establish a comprehensive mechanistic framework for understanding the pH-activity relationship of T7 RNAP. These insights have direct relevance for optimizing IVT reaction conditions in mRNA vaccine manufacturing, where pH control is essential for maximizing transcript yield, purity and quality and may further inform rational engineering strategies to develop T7 RNAP variants with enhanced performance characteristics for next-generation mRNA therapeutics.

Methods

System Preparation

The initial structure of T7 RNA polymerase (T7 RNAP) in complex with promoter DNA and a three-nucleotide RNA transcript was obtained from the Protein Data Bank (PDB ID: 1QLN), which represents the initiation complex reported at 2.4 Å resolution. The initiating nucleotide in the original structure is a GTP with a 5′-triphosphate group; the triphosphate moiety was removed and replaced with a 5′-hydroxyl group while retaining the original crystallographic coordinates of the ribose and guanine base, yielding a 5′-GGG-3′ RNA transcript compatible with standard force field parameters. Missing loop regions in the protein were modeled using the SWISS-MODEL server to generate a complete structure. Two catalytic magnesium ions were positioned in the active site (Figure ) based on their coordinates in the T7 RNAP elongation complex structure (PDB ID: 1S76), justified by the high structural conservation of the active site between initiation and elongation states.

1.

Crystal structure of T7 RNAP in initiation complex with DNA promoter and RNA transcript (PDB: 1QLN ), including the key functional regions highlighted: specificity loop (residues 740–769, magenta); intercalating β-hairpin (residues 232–242, cyan); O-helix (residues 624–639, blue); subdomain H (residues 151–190, tan); AT-rich loop (residues 91–98, red); and catalytic Mg ions (silver spheres).

Force Field Parameters

The protein was parametrized using the constant pH force field implemented in AMBER, , which enables discrete protonation state changes during simulation. The following residue types were designated as titratable: histidine (HIS), aspartate (ASP), glutamate (GLU), lysine (LYS), tyrosine (TYR), and cysteine (CYS). RNA nucleotides were described using the OL3 force field, while DNA was parametrized with the OL24 force field, representing current state-of-the-art nucleic acid parameters. Magnesium ions were modeled using the 12–6–4 Lennard-Jones-type parameters optimized for the OPC water model (frcmod.ionslm_1264_opc). The system was solvated with OPC water molecules in a truncated octahedral box with a minimum buffer distance of 10 Å from the solute to the box edge. Sodium and chloride ions were added to neutralize the system and achieve a physiological salt concentration of 0.15 M. The final system contained 99,487 atoms.

Energy Minimization and Equilibration

The solvated system was subjected to energy minimization for 5000 steps using a combination of steepest descent and conjugate gradient algorithms to remove steric clashes and optimize geometry. The system was then heated from 0 to 310 K over 500 ps under NVT conditions with a time step of 1 fs, employing harmonic restraints on all heavy atoms of the solute. Temperature was maintained using the Langevin thermostat with a collision frequency of 2.0 ps^–1.

Equilibration was performed in four successive stages under NPT conditions at 310 K and 1 bar, with the Monte Carlo barostat used for pressure control. Harmonic positional restraints on all heavy atoms of the protein, nucleic acids, and magnesium ions were progressively reduced across the four stages (5.0, 2.5, 1.0, and 0.5 kcal mol^–1 Å^–2, respectively), with each stage consisting of 1 ns of simulation (4 ns total equilibration). This gradual release of restraints allowed the system to relax while maintaining structural integrity. Throughout minimization, heating, and all equilibration stages, the constant pH MD (cpHMD) routine was disabled, so all titratable residues retained their initial protonation states; cpHMD sampling of protonation states was activated only at the start of the production runs.

Constant pH Replica Exchange Molecular Dynamics

Production simulations employed the discrete constant pH molecular dynamics (CpHMD) method with pH-based replica exchange, as implemented in AMBER with GPU acceleration. A total of 18 replicas were simulated spanning pH values from 4.75 to 9.15. The specific pH values for each replica were: 4.75, 4.95, 5.15, 5.35, 5.55, 5.80, 6.05, 6.30, 6.55, 6.85, 7.15, 7.45, 7.75, 8.05, 8.35, 8.65, 8.90, and 9.15.

Simulations were conducted under NVT conditions at 310 K using a time step of 2 fs. All bonds involving hydrogen atoms were constrained using the SHAKE algorithm. Long-range electrostatic interactions were treated using the particle mesh Ewald (PME) method with a real-space cutoff of 8 Å. Protonation state changes were attempted every 2,000 MD steps (4 ps) using Monte Carlo sampling in implicit solvent, with explicit solvent relaxation performed for 100 steps following each protonation state change. Replica exchanges between adjacent pH values were attempted every 2000 MD steps (4 ps), coinciding with protonation state sampling. Each replica was simulated for 160 ns, yielding a cumulative sampling time of 2.88 μs across all pH conditions.

Exchange acceptance ratios between adjacent pH replicas ranged from 19.5% to 35.0%, with a mean acceptance rate of 28.4% (Table S1), indicating adequate sampling across pH space and efficient mixing between replicas. These acceptance rates fall within the optimal range of 20–40% recommended for replica exchange simulations.

A zipped file with details about how to run the cpHMD is included in the Supporting Information.

Analysis

Protonation State Analysis

Protonation state populations were calculated using the cphstats tool in AMBER. The pK _a values were determined by fitting eq to the pH-dependent deprotonated fractions obtained from simulations across multiple pH conditions (pH 4.75–9.15). Eq is taken from eq (4) of Swails, York and Roitberg and is a form of the Hill equation used commonly in biochemistry, e.g. ,

$$f_{\mathrm{d}\mathrm{eprot}}=\frac{1}{1+{10}^{n(\mathrm{p}{K}_{\mathrm{a}}-\mathrm{pH})}}$$ 1

Here f _deprot is the fraction deprotonated, pK _a is the acid dissociation constant, and n is the Hill coefficient. A Hill coefficient of 1 corresponds to ideal Henderson–Hasselbalch behavior, while values significantly different from 1 indicate cooperative effects or coupling between protonation and conformational dynamics, with n > 1 indicating positive cooperativity and a steeper binding curve and n < 1 indicating negative cooperativity and a flatter binding curve. Fitting was performed using nonlinear least-squares regression. Goodness of fit was assessed using the coefficient of determination (R²). Convergence of protonation states was verified by monitoring cumulative fraction protonated over the trajectory for each pH condition.

Structural Analysis

Root mean square deviation (RMSD) and root-mean-square fluctuation (RMSF) were calculated using the cpptraj module in AMBER. RMSD was computed for backbone atoms and RMSF for Cα atoms. For both analyses, plots for each pH are presented alongside the pH 8.05 replica to facilitate comparison with the optimal activity condition. Key distances, including the separation between the C-terminal Ala883 and the catalytic residue Asp537, were monitored throughout the simulations.

Regional Flexibility Analysis

RMSF profiles were examined for functionally important regions: the AT-rich recognition loop (residues 91–98), intercalating β-hairpin (residues 231–245), specificity loop (residues 740–769), and extended foot module including the C-terminal region (residues 835–883).

Free Energy Landscape Analysis

To visualize the conformational preferences of T7 RNAP as a function of pH, we constructed two-dimensional free energy landscapes (FELs) using two functionally important structural parameters as reaction coordinates: (d ₁) the distance between the centers of mass of the specificity loop (residues 740–769) and the intercalating β-hairpin (residues 232–242), which reports on promoter channel geometry; and (d ₂) the distance between the Cβ atoms of Ala883 and Asp537, which reflects C-terminal positioning relative to the active site. Together, d₁ and d₂ capture the two structural requirements for productive catalysis: proper promoter engagement during initiation and correct positioning of the C-terminal carboxylate for catalytic Mg²⁺ coordination, making their joint distribution a sensitive reporter of the pH-dependent functional competence of T7 RNAP. These distances are used to analyze the coordinated motions of regions shown in Figure .

For each pH condition, the two distances were calculated for all frames of the production trajectory. The free energy surface was computed from the joint probability distribution P(d ₁, d ₂) of the two distances above, using

$$F(d_1,d_2)=-k_{\mathrm{B}}T\ln P(d_1,d_2)$$ 2

where k_B is the Boltzmann constant and T = 310 K. The probability distribution was estimated using two-dimensional histogramming with 80 bins in each dimension, spanning the range of observed distances. Conformational basins were identified as contiguous regions of the free energy surface surrounding each local minimum. Basin centers were defined as the point of lowest F; representative structures (N = 5 per basin) were selected as the trajectory frames with coordinates closest to the basin center in the two-dimensional distance space.

Energy Decomposition Analysis (EDA) and Electrostatic Coupling

Residue–residue interaction energies were computed using a custom Python implementation (cpHMD-EDA) that extends the AMBER-EDA framework. For each saved snapshot, the pairwise non-bonded interaction energy between two residues was decomposed into electrostatic (Coulomb potential) and van der Waals (12–6 Lennard-Jones) contributions using AMBER force-field parameters read directly from the topology file, summed over all atom pairs without distance truncation. At each snapshot, the instantaneous protonation state of every titratable residue was read from the corresponding cpHMD state-trajectory file, and partial charges on hydrogen atoms that do not exist in that microstate were set to zero (e.g., HE2 in HID, NZ hydrogen atom in neutral lysine), so that the computed energies reflect the actual ensemble of microstates sampled at each pH. Minimum-image distances were evaluated under full triclinic periodic boundary conditions. The values in Tables and are mean values ± standard deviations across all analyzed frames at each pH, under the AMBER point-charge convention, so that positive E _elec indicates net-repulsive, and negative E _elec net-attractive, interactions. The cpHMD-EDA code is available from the authors upon request.

1. Top 10 Electrostatic Coupling Partners of His811 at Three pH Values (E _elec ± std, kcal/mol) .

	pH 4.95		pH 8.05		pH 9.15
rank	residue	E elec	residue	E elec	residue	E elec
1	Mg2+	111.0 ± 12.4	Mg2+	84.6 ± 11.8	Mg2+	72.9 ± 6.3
2	Arg425	48.4 ± 15.3	Arg425	33.1 ± 14.0	Arg425	36.3 ± 10.3
3	Asp787	–45.1 ± 18.4	Asp787	–26.1 ± 12.0	Asp787	–28.0 ± 9.4
4	G3-RNA (3′-terminus)	–35.3 ± 18.8	Asp421	–20.3 ± 6.7	Asp421	–21.6 ± 4.3
5	Lys441	33.8 ± 10.7	Arg298	19.4 ± 4.9	Arg298	19.9 ± 3.5
6	Asp421	–30.9 ± 8.2	Asp537	–18.9 ± 4.8	DT-DNA	–18.8 ± 3.4
7	G-RNA899	–27.9 ± 9.8	G-RNA899	–18.0 ± 5.3	DC-DNA	–17.7 ± 3.0
8	Arg298	26.7 ± 6.4	G3-RNA (3′-terminus)	–17.7 ± 8.6	Arg792	17.0 ± 3.8
9	Asp537	–25.2 ± 5.3	Asp438	–17.7 ± 7.8	Asp537	–16.4 ± 2.3
10	Arg423	24.4 ± 6.4	DT-DNA	–17.4 ± 4.3	DC-DNA	–16.2 ± 2.3

^aPairwise electrostatic interaction energies between His811 and all other residues were calculated using microstate-aware charges and the AMBER Coulomb potential. Negative values indicate attractive interactions; positive values indicate repulsive interactions.

2. pK _a Values of Titratable Residues within 5 Å of DNA Promoter .

residue	type	pK a	notes
Lys 93	Basic (+)	>9.15	Locked protonated across sampled range
Lys 95	Basic (+)	>9.15	Locked protonated
Lys 98	Basic (+)	>9.15	Locked protonated
Lys 206	Basic (+)	8.0	R 2 = 0.99
Glu 207	Acidic (−)	5.4	R 2 = 0.99
His 211	Basic (+)	8.8	R 2 = 1.00
Asp 240	Acidic (−)	<4.75	Locked deprotonated
Glu 242	Acidic (−)	9.1	R 2 = 0.97
His 300	Basic (+)	8.3	R 2 = 0.94
Asp 421	Acidic (−)	<4.75	Locked deprotonated
Tyr 427	Polar	>9.15	Locked protonated (neutral)
Tyr 639	Polar	>9.15	Locked protonated (neutral)
Lys 642	Basic (+)	9.7	R 2 = 0.93
Glu 652	Acidic (−)	5.2	R 2 = 1.00
Tyr 739	Polar	>9.15	Locked protonated (neutral)
Asp 766	Acidic (−)	<4.75	Locked deprotonated
His 772	Basic (+)	7.2	R 2 = 1.00
Lys 773	Basic (+)	>9.15	Locked protonated
His 784	Basic (+)	3.7	R 2 = 0.95

^a*Net charge at any pH can be estimated using the Henderson–Hasselbalch equation. At pH 4, most histidines and lysines are protonated (+1), while most acidic residues are neutral (0). At pH 8, most histidines are neutral (0), lysines remain charged (+1), and acidic residues are deprotonated (−1).

All molecular visualizations were generated using PyMOL and VMD. Statistical analyses and plotting were performed using Python with the NumPy, SciPy, and Matplotlib libraries.

Results

Structural Stability Across pH Conditions

All 18 replicas maintained a stable overall fold throughout the 160 ns simulations, with backbone RMSD values plateauing by ∼50 ns and equilibrated values ranging from 4–6 Å across the full pH range (Figures S1–S2). Acidic conditions (pH 4.75–5.80) produced the highest RMSD values and the broadest distributions, consistent with experimental circular dichroism data showing no gross secondary structure loss between pH 5 and 7.9, yet indicating enhanced conformational sampling at lower and higher pHs compared to 8.05. These findings are in agreement with those of Borkotoky et al., who reported backbone RMSD values of ∼5 Å at optimal pH and ∼6 Å at pH 5 using conventional MD with fixed protonation states; our CpHMD approach extends those observations by providing continuous pH sampling across 18 conditions. Regional flexibility analysis (Figures S3–S5) revealed that the AT-rich recognition loop (residues 91–98) and intercalating β-hairpin (residues 232–242) are the most pH-sensitive structural elements; both are destabilized at acidic pH (RMSF elevated by 0.3–0.6 Å relative to the pH 8.05 reference) and at highly basic pH (≥8.35), while exhibiting minimal fluctuation near the activity optimum. The specificity loop (residues 740–769) remains largely pH-insensitive across pH 5.80–8.65, consistent with stabilization by direct DNA contacts. Together, these observations establish that the T7 RNAP initiation complex undergoes meaningful pH-dependent dynamic changes without global unfolding, motivating detailed analysis of the specific structural and electrostatic mechanisms described below.

Protonation State Analysis and pH-Dependent Activity

To further elucidate the molecular basis of the pH dependency of T7 RNA polymerase activity, we performed titration analysis on key residues including the catalytic aspartates (Asp537 and Asp812), catalytically significant residues His811, Lys472, Lys631, and Tyr639, and the structurally important Glu148 (Figures and S6). Asp537 and Asp812 remain persistently deprotonated (fractional deprotonation value of 1.0) across the entire pH range simulated (pH 4.75–9.15), consistent with their essential role in coordinating the catalytic Mg²⁺ ions. Lys472, which promotes pyrophosphate release during catalysis, is predicted to have a high pK _a (≫9), suggesting it remains mostly positively charged throughout the physiological pH range. Lys631 exhibits a predicted pK _a of 9.6 and is primarily positively charged at optimum pH 8 to stabilize the triphosphate moiety of incoming nucleotides. The titration curve reveals that K631 begins significant deprotonation above pH 8.5, with approximately 50% deprotonation at pH 9.57 and near-complete deprotonation by pH 11. The pH-dependent titration of Lys631 represents one important regulatory mechanism for T7 RNA polymerase catalytic activity. Site-directed mutagenesis studies have demonstrated that the positive charge of K631 is essential for efficient catalysis, with charge-neutralizing mutations (K631M, K631G, K631L) reducing activity by 98–99% while maintaining DNA binding capability, indicating K631’s primary role in phosphodiester bond formation rather than substrate recognition. , At alkaline pH values approaching and exceeding the pK _a of 9.57, progressive deprotonation of K631 would effectively mimic these charge-neutralizing mutations, providing one component of a molecular explanation for the observed pH-dependent activity profile of T7 RNAP. Tyr639 remains protonated across the simulated pH range (pK _a > 12), maintaining the hydroxyl group necessary for ribose 2’–OH discrimination during substrate selection.

2.

Predicted Titration Profiles for T7 RNAP Active Site Residues. R ² and n represent the goodness of fit to the titration model and the Hill coefficient, respectively; n values significantly below 1.0 indicate anticooperative behavior within the crowded active site environment.

In contrast, His811 exhibits a calculated pK _a of 5.9 (R ² = 0.996), identifying it as the primary pH-sensitive ionizable residue in the active site. At the enzyme’s optimal pH 8.0, His811 is predominantly neutral (>90% deprotonated), consistent with mutational studies showing that the neutral glutamine mimic (H811Q) retains partial catalytic activity. As pH decreases below 6.0, His811 becomes increasingly protonated, acquiring a positive charge that fundamentally alters active site electrostatics. His811, located within the highly conserved palm subdomain motif, has long been recognized as an important component of the T7 RNAP catalytic machinery. Early biochemical investigations demonstrated that the H811Q substitution, despite glutamine serving as a near-isosteric neutral mimic of deprotonated histidine, results in some kinetic impairments: a 3- to 5-fold reduction in k _cat and a 2- to 3-fold increase in K _M for incoming nucleotides. These findings indicate that the chemical identity of the histidine imidazole ring, rather than merely its steric properties, contributes significantly to catalytic function. His811 is thought to modulate the nucleophilicity of the adjacent catalytic residue Asp812, thereby fine-tuning the rate of phosphodiester bond formation.

Beyond its catalytic role, His811 contributes significantly to initiation complex stability. Mutations at this position lead to increased abortive cycling, where the enzyme repeatedly synthesizes and releases short (2–14 nt) RNA transcripts rather than transitioning to productive elongation. This suggests His811 plays a dual role: facilitating active site chemistry while providing structural stabilization of the nascent RNA-DNA heteroduplex during the DNA scrunching process.

To identify residues involved in coordinating the catalytic metal ions and to understand their pH-dependent reorganization, we performed energy decomposition analysis (EDA) on His811. For each trajectory frame, pairwise electrostatic interaction energies between His811 and all other residues were calculated using microstate-aware charges (adjusted for instantaneous protonation states) and the AMBER Coulomb potential. Ensemble-averaged coupling energies were computed at pH 4.95, 8.05, and 9.15, and the top ten coupling partners at each pH were identified (Table and Figure a).

3.

(a) Electrostatic coupling network around His811. His811 (magenta, sticks) is shown together with its top coupling partners from Table . (b) pH-Sensitive E148-Met750 Anchor in the T7 RNAP Elongation Complex (PDB 1S76 ). Structure of T7 RNAP elongation complex showing Glu148 (orange, stick representation) in close contact (∼3.2 Å) with Met750 (green, stick representation) at the tip of the specificity loop (green ribbon).

At optimal pH 8.05, where His811 is predominantly neutral, the dominant coupling partners are the catalytic Mg²⁺ ion (84.6 ± 11.8 kcal/mol) and structural residues Arg425 and Asp787, reflecting an active site organized for efficient metal-mediated catalysis. As pH decreases to 4.95 and His811 becomes protonated, coupling to RNA nucleotides G3-RNA­(3′-terminal) and G-RNA strengthens markedly (−35.3 ± 18.8 and −27.9 ± 9.8 kcal/mol respectively), consistent with stronger electrostatic attraction between the positively charged imidazolium ion and the negatively polarized RNA backbone. Conversely, at basic pH 9.15, RNA nucleotides disappear entirely from the top coupling partners, suggesting conformational heterogeneity or active site destabilization at elevated pH. The strengthened His811–RNA electrostatic coupling at acidic pH, while reflecting stronger attraction between the protonated imidazolium and the RNA backbone, paradoxically impairs catalytic function. The protonated histidine forms a persistent electrostatic interaction with the RNA, restricting the conformational flexibility required for translocation to the next template position or release of abortive transcripts. This “electrostatic trap” constitutes one molecular-level explanation for the diminished polymerase activity observed at pH < 6.

While the catalytic aspartate residues and His811 govern active site chemistry, Glu148 emerges as a significant structural pH sensor with a calculated pK _a of 5.25 ± 0.01 (R ² = 0.997, n = 0.71). This elevated pK _a (compared to ∼4.2 for glutamate in solution) positions Glu148 as a pH-responsive regulator in the physiologically relevant range. Located next to subdomain H, Glu148 plays a conformational-state-dependent role. In the initiation complex, Glu148 is remote (∼23 Å) from the specificity loop, but structural analysis of the elongation complex reveals that the dramatic N-terminal domain reorganization brings Glu148 OE2 into close contact (∼3.2 Å) with the sulfur atom (SD) of Met750 at the tip of the specificity loop (Figure b).

At physiological pH (7–8), where Glu148 is predominantly deprotonated (>85%), the negatively charged carboxylate forms a stabilizing charge-induced dipole interaction with the polarizable thioether sulfur atom of Met750. This interaction, reinforced by longer-range electrostatic coupling through Arg155, anchors the specificity loop in the conformation required to form the RNA exit tunnel, a structural feature essential for processive elongation. At acidic pH (<5.25), protonation of Glu148 neutralizes the carboxylate, weakening this interaction and destabilizing the elongation complex. This mechanism rationalizes the phenotype of the E148A variant, which can initiate transcription but fails to progress beyond ∼ 5 nt transcripts; the variant permanently lacks the pH-sensitive carboxylate anchor required to stabilize the conformational transition from initiation to elongation.

DNA promoter binding and release are governed by electrostatic interactions between the protein and the polyanionic DNA backbone. To assess how pH modulates these interactions, we identified all titratable residues within 5 Å of the DNA promoter and calculated the net charge of this protein–DNA interface as a function of pH (Table and Figure ).

4.

Structural and pH-dependent electrostatic features of the T7 RNA polymerase–promoter DNA binding interface. (a) Surface representation of titratable residues within 5 Å of the promoter DNA. Catalytic Mg²⁺ ions are shown as magenta spheres. (b) Aggregate Electrostatic Charge of the Promoter-Binding Interface as a Function of pH. Gray region represents the simulation window. The dramatic decrease in positive charge from pH 4 to pH 8 modulates DNA-binding affinity, with excessive positive charge at acidic pH preventing efficient promoter clearance and transition to elongation.

The aggregate charge of the promoter-proximal residues decreases dramatically from approximately +6 at pH 4.0 to +2 at pH 8.0 (Figure ). At acidic pH, the DNA-T7 RNAP binding interface is highly positively charged due to protonation of histidine residues (His211, His300, His772, His784) and neutralization of acidic residues (Glu207, Asp421, Asp766). This results in strong electrostatic attraction to the negatively charged DNA backbone, which may prevent efficient promoter clearance, a critical step in the transition from initiation to elongation. Above pH 8, the interface charge becomes negative (Figure ), generating electrostatic repulsion with the negatively charged DNA phosphate backbone and thereby disfavoring promoter binding at alkaline pH values. In T7 RNAP, productive elongation requires promoter release following synthesis of an ∼8-nucleotide transcript, allowing the enzyme to escape the promoter and enter the processive elongation phase. The excessive positive charge at acidic pH would increase the electrostatic binding energy, potentially trapping the enzyme on the promoter and promoting abortive cycling rather than promoter escape. Conversely, at optimal pH ∼ 8.0, the reduced net positive charge (+2) permits a balanced interaction: strong enough for specific promoter recognition during initiation, yet weak enough to allow efficient promoter release and transition to elongation.

Electrostatic Coupling Analysis of C-Terminal Ala883

Ensemble-averaged coupling energies were computed at pH 4.95, 8.05, and 9.15, and the top five coupling partners of Ala883 at each pH were identified (Table ). The coupling network (Figure ) undergoes dramatic pH-dependent reorganization. At pH 4.95, protonated histidines (His545 and His790, pK _a ∼ 6) dominate the coupling hierarchy, with interaction energies (−59.2 and −54.5 kcal/mol) exceeding those of the catalytic metals. As pH increases to 8.05, histidine deprotonation causes their coupling to weaken by ∼50%, while metal ion coupling intensifies (MG1: −69.3 kcal/mol, 20% increase). At pH 9.15, metal coordination strengthens further (MG1: −84.4 kcal/mol, 46% increase from pH 4.95), driven by accumulating negative charge as additional residues deprotonate. Notably, the weakening of histidine coupling at elevated pH reflects not only charge neutralization upon deprotonation but also a simultaneous increase in spatial separation.

3. Top Five Electrostatic Coupling Partners for C-Terminal Ala883 with Ensemble-Averaged Coupling Energies, Which Are Reported in kcal/mol.

	pH 4.95		pH 8.05		pH 9.15
rank	residue	energy	residue	energy	residue	energy
1	His545	–59.2	MG1	–69.3	MG1	–84.4
2	MG1	–57.8	MG2	–68.4	MG2	–71.6
3	His790	–54.5	Arg829	–33.0	Arg829	–28.8
4	MG2	–54.4	His545	–27.2	His790	–28.3
5	Arg829	–32.9	His790	–27.1	Arg425	–26.9

5.

Coupling network of Ala 883 and its position with respect to catalytic Asp 537 (PDB ID: 1QLN ).

Violin plots of the Ala883­(OXT)–His545­(ND1) and Ala883­(OXT)–His790­(ND1) distances across all pH replicas (Figure S8) reveal that both residues transition from close, tightly distributed contacts at acidic pH (mean ∼ 5.4 Å at pH 4.8) to more distant conformations at alkaline pH (mean ∼ 8.5 and ∼8.0 Å for His545 and His790, respectively, at pH 8). This structural reorganization coincides with histidine deprotonation near pH 6, indicating that loss of the positive charge on the imidazolium ring eliminates the electrostatic attraction that maintained close contact with the C-terminal carboxylate group. The broader distance distributions observed above pH 8 further indicate increased conformational heterogeneity in the active site at alkaline pH. The optimal pH 8 represents a balance where histidines are deprotonated, allowing strong metal coordination to support the two-metal-ion catalytic mechanism.

Together, these observations reveal that pH modulates T7 RNAP activity through complementary mechanisms affecting both catalysis and promoter dynamics.

pH-Dependent Positioning of the C-Terminal Carboxylate

The C-terminal carboxylate group of Ala883 is positioned near the catalytic center of T7 RNAP and has been implicated to participate in metal ion-dependent catalysis. Previous structural comparisons between apo-T7RNAP and the T7RNAP-lysozyme inhibited complex suggested that displacement of this C-terminal end away from the active site contributes to enzyme inactivation. , To investigate the pH-dependence of C-terminal positioning, we monitored the distance between the Cβ atom of Ala883 and the Cβ atom of the catalytic residue Asp537 throughout our simulations.

The C-terminal to active site distance exhibits a striking pH-dependent behavior (Figure ). At acidic pH (4.75–5.5), the mean distance is substantially elevated, ∼9 Å at pHs 4.75–5.5 (Figure B). As pH increases toward the optimal range, the distance decreases progressively, reaching approximately 8.5–8.0 Å between pH 6.05 and 7.75, and stabilizing around 7.5 Å at the optimal pH of 8.05. At more basic pH values above 8, the distance continues to decrease, dropping from 7.5 Å to less than 7 Å at pH 9.15.

6.

Distribution (A) and mean (B) distance between the C-terminal Ala883 and the catalytic Asp537 as functions of pH.

The distribution analysis reveals additional insights into the conformational heterogeneity at each pH (Figure A). At acidic pH, the violin plots display broad distributions centered around 9–10 Å with substantial density extending up to 12–14 Å, indicating frequent excursions of the C-terminus away from the active site. The distributions narrow progressively as pH is increased, with the tightest distributions observed in the pH 6.5–7.15 range. As pH increases, the sampling of shorter distances becomes more probable.

These results are consistent with the findings of Borkotoky et al., who reported that the C-terminal end flips away from the catalytic center at pH 5, with an average distance exceeding 10 Å compared to approximately 7 Å at optimal pH. Our continuous pH sampling reveals that this transition follows a monotonic curve with an apparent transition around pH 6.0–7.75. Since the critical role of the C-terminal carboxylate is specifically in Mg²⁺-dependent metal coordination, the progressive decrease in this distance at basic pH is unlikely to account for the reduced enzymatic activity observed above pH 8.5, suggesting that other structural factorssuch as perturbations in promoter channel geometryare responsible for the loss of activity at high pH.

pH-Dependent Spacing of Promoter Recognition Elements

The specificity loop (residues 740–769) and the intercalating β-hairpin (residues 232–242) are two critical structural elements involved in promoter DNA recognition and melting. The specificity loop makes sequence-specific contacts with the promoter, while the β-hairpin inserts between the template and nontemplate strands to initiate duplex opening. The spatial relationship between these elements defines the geometry of the promoter binding channel. To assess pH effects on promoter recognition competency, we monitored the distance between the centers of mass of these two structural elements.

Remarkably, the average spacing between the specificity loop and β-hairpin exhibits an opposite, albeit weak, pH-dependence compared to the C-terminal positioning (Figure ). At acidic pH (4.75), the mean distance is minimal at 14.5 ± 0.51 Å (Figure B). The distance then increases gradually, reaching 15.1 ± 0.75 Å at the optimal pH of 8.05. Above pH 8.0, the distance increases more steeply, reaching 15.3 ± 0.75 Å at pH 8.35 and continuing to rise to 15.6 ± 0.62 Å at pH 9.15.

7.

Distance between the centers of mass of the specificity loop (residues 740–769) and the intercalating β-hairpin (residues 232–242).

The distribution analysis provides more significant insights into the conformational dynamics of the promoter binding channel (Figure A). At acidic pH (4.75–5.15), the distributions show a prominent lower tail extending to ∼13 Å, indicating transient compression of the promoter channel. The distributions become progressively narrower as pH rises to the 5.55–7.45 range, suggesting a more well-defined channel geometry. At basic pH values above 8.0, the distributions shifts toward higher values and broadened again, with upper tails extending to 18–19 Å at pH 8.65–9.15.

These findings complement and extend the observations of Borkotoky et al., who reported a similar trend in their apoenzyme simulations. The opposing pH-dependencies of the two distancesC-terminal to active site decreasing while promoter channel spacing increases with pHreveal a coordinated conformational response across the enzyme. At acidic pH 4.75, the C-terminus is displaced (∼9 Å, Figure B) while the promoter channel is compressed (14.5 Å, Figure B), simultaneously impairing both metal coordination and promoter accommodation. At optimal pH 8.05, the C-terminal distance reaches its functional value (7.5 Å, Figure B) while the channel stabilizes at a geometry conducive to promoter engagement (15.1 Å, Figure B), representing the balanced configuration that facilitates productive transcription. At basic pH 9.15, the channel becomes wider (15.6 Å, Figure B) potentially weakening promoter binding, while the C-terminus approaches the active site more closely (∼6.5 Å, Figure B). The convergence of both structural parameters toward their functional optima exclusively at pH 8.05 provides a structural explanation for the bell-shaped pH-activity profile of T7 RNAP.

Free Energy Landscape Analysis

To visualize the conformational landscape of T7 RNAP as a function of pH, we constructed two-dimensional free energy surfaces using the two functionally important distances, C-terminal (A883) to active site (D537) and specificity loop to β-hairpin, as reaction coordinates. This approach provides an intuitive representation of the conformational states sampled at each pH condition.

The free energy landscapes (FELs) reveal distinct conformational basins at different pH values (Figure ). At acidic pH 4.95, the energy minimum is centered at approximately 14.5 Å for the SPL−β-hairpin distance and above 10 Å for the A883–D537 distance, representing a state with a compressed promoter channel and displaced C-terminus. The landscape displays a relatively compact, roughly circular basin with well-defined boundaries, indicating limited conformational sampling around this suboptimal state. At optimal pH 8.05, the free energy minimum shifts to approximately 15 Å for the SPL−β-hairpin distance and ∼7 Å for the A883–D537 distance. Notably, the landscape at optimal pH exhibits a broader, more diffuse basin extending toward higher SPL−β-hairpin distances (up to 18 Å). This expanded accessible conformational space at optimal pH may reflect the dynamic flexibility required for the catalytic cycle, including promoter binding, DNA melting, and nucleotide incorporation. At basic pH 9.15, two energy minima are observed: one at approximately 15.5 Å for the SPL−β-hairpin distance with A883–D537 distance of ∼6.5 Å, and a second, narrow minimum at ∼16.5 Å SPL−β-hairpin with A883–D537 distance of ∼9 Å (Figure ). These two states represent a heterogeneous C-terminal population sampling both tightly docked (∼6.5 Å) and partially displaced (∼9 Å) conformations within an expanded promoter channel. The landscape shows an asymmetric basin extending toward higher SPL−β-hairpin values (up to 18 Å), consistent with the widened promoter channel observed at elevated pH. The coexistence of these two C-terminal states at basic pH, one competent (∼6.5 Å) and one displaced (∼9 Å) for Mg²⁺ coordination, suggests that the reduced enzymatic activity at high pH reflects in part a decreased probability of productive metal coordination, as the enzyme samples a partially inactive C-terminal conformation with increased frequency.

8.

2D free energy surfaces are projected onto two important reaction coordinates: the distance between the specificity loop SPL and the intercalating β-hairpin, and the distance between the C-terminal Ala883 and the catalytic Asp537.

The diagonal progression of the energy minima across pH conditions, from upper-left (acidic: high A883–D537, low SPL−β-hairpin) to lower-right (basic: low A883–D537, high SPL−β-hairpin), visually demonstrates the coordinated, opposing conformational changes in these two structural parameters. The optimal pH 8.05 basin occupies an intermediate position, representing a balanced conformational state where both the C-terminal positioning and promoter channel geometry are conducive to efficient catalysis. This free energy landscape analysis provides a unified framework for understanding how pH modulates the conformational ensemble of T7 RNAP and thereby controls enzymatic activity.

pH-Dependent Y639 Positioning

To assess the translocation state of T7 RNAP under different pH conditions, we measured the distance between Y639 and the RNA 3′-end nucleotide in representative structures (N = 5) from the Basin of each FEL (Figure ). Since a basin represents the conformational state optimized for both C-terminal positioning (A883–D537 distance) and promoter channel geometry (SPL−β-hairpin distance), structures from this basin reflect the catalytically relevant enzyme conformation at each pH. The Y639-RNA 3′-end distance (measured between Y639 CZ and RNA base C2) is 6.4 ± 0.88 Å at pH 8.05. This distance decreases to 5.2 ± 0.85 Å at pH 4.95 and further to 4.1 ± 0.83 Å at pH 9.15 (Table ).

9.

Positioning of RNA 3′-terminal to Tyr 639 and Mg ion at different pH values (A) pH 4.95, (B) pH 8.05, (C) pH 9.15 based on averaged structures from the respective trajectories. (D) Functional NTP binding geometry in T7 RNAP, based on the crystal structure of the substrate insertion complex (PDB ID: 1S76 ) showing the incoming NTP analog (3′-deoxyATP, 3DA) positioned at the N site between the RNA 3′-terminal base and the next template base.

4. Y639-RNA 3′-End Distances at Different pH Conditions in Representative Structures (N = 5).

pH	mean distance (Å)	Std (Å)
4.95	5.2	0.85
8.05	6.4	0.88
9.15	4.1	0.83

A critical geometric constraint for nucleotide addition is that Y639 must be positioned sufficiently far from the RNA 3′-terminus to permit the incoming NTP access to the N site. The crystal structure of the T7 RNAP substrate insertion complex (PDB: 1S76 ), which contains the incoming NTP analog 3′-deoxyATP (3DA) bound at the N site, shows Y639 positioned approximately 8 Å from the RNA 3′-terminal base at the P site (Figure ). This spacing represents the functional geometry required to accommodate the incoming nucleotide between Y639 and the primer terminus.

10.

Summary of the pH-sensing network in the T7 RNA polymerase initiation complex. Titratable residues identified throughout this work are shown as side-chain sticks on the transparent polymerase cartoon, with the promoter DNA in orange. Interface residues (Table ) are blue; functionally highlighted residues are color-coded by role: His811 (magenta), Ala883 network including His790 and His545 (yellow), Asp537 (red), Glu148 (salmon), Tyr639 (cyan), Lys631 (purple). The specificity loop (green) and intercalating β-hairpin (firebrick) are shown as ribbons. Catalytic Mg²⁺ ions are green spheres.

At pH 8.05, the mean Y639-RNA 3′-end distance was 6.4 Å. This distance decreased to 5.2 Å at pH 4.95 and to 4.1 Å at pH 9.15 (Table and Figure ). The Y639-RNA 3′-end distance of 6.4 Å at pH 8.05 is closer to the ∼8 Å spacing observed in the substrate-bound (PDB ID 1S76) crystal structure, indicating that our simulations at optimal pH capture the functional open conformation required for NTP binding.

Importantly, our pH 8.05 result (6.4 Å) significantly exceeds the Y639-RNA 3′-end distance observed in the 1MSW crystal structure of the elongation complex (4.9 Å, measured from Y639-CZ to RNA base C2). The 1MSW structure, determined in the absence of incoming NTP substrate, shows Y639 collapsed toward the P site with the template base proximal to Y639 (template T) at position (i) pairing with the already-transcribed RNA 3′-terminus (A). This compressed geometry would sterically occlude NTP access to the N site. In contrast, the 1S76 substrate insertion complex shows the functional ∼ 8 Å spacing when NTP is bound, our pH 8.05 simulations show the closest distance to it. This suggests that the 1MSW crystal structure captured a closed, substrate-free conformation, while our MD simulations at optimal pH reveal the functional open state that is competent for NTP binding and required for processive elongation.

The correlation between the Y639-RNA distance and catalytic activity across pH conditions indicates that proper Y639 positioning is pH-dependent and essential for maintaining the open active site conformation required for NTP substrate access. The reduced distances at pH 4.95 (5.2 Å) and pH 9.15 (4.1 Å) show incomplete opening or premature closure of the active site, which would restrict NTP entry and impair catalytic turnover. These results demonstrate that pH modulates the Y639-RNA 3′-end distance in a manner consistent with the known catalytic activity profile, with the functional ∼8 Å separation occurring at optimal pH and reduced, nonfunctional distances at both acidic and basic conditions.

Conclusions

This study provides the first comprehensive view of pH-dependent conformational dynamics in T7 RNA polymerase using constant pH replica exchange molecular dynamics simulations spanning 18 pH conditions from 4.75 to 9.15. Our results reveal a sophisticated pH-sensing network wherein multiple titratable residues collectively modulate enzyme structure, dynamics, and function across the physiological pH range.

We identified His811 (palm subdomain next to catalytic Asp812, pK _a 5.9) as the primary active site pH sensor whose protonation state directly controls catalytic competence. At acidic pH, protonated His811 forms persistent electrostatic interactions with the RNA 3′-terminal nucleotide, as reflected by the strong coupling energies to G3-RNA­(3′-terminal) (−35.30 ± 18.79 kcal/mol) and G-RNA (−27.93 ± 9.76 kcal/mol), creating an electrostatic trap that impairs the conformational flexibility required for nucleotide incorporation and transcript elongation. At optimal pH 8.0, deprotonation of His811 weakens these RNA coupling energies substantially, while coupling to the catalytic Mg²⁺ ion dominates (84.63 ± 11.82 kcal/mol), establishing the active site geometry necessary for efficient catalysis. This finding reconciles previously reported biochemical data showing that H811Q mutations severely impair catalytic efficiency despite glutamine serving as a neutral mimicthe imidazole ring’s pH-responsive chemistry, not merely its size, is essential for efficient function.

Our results show that pH modulates two critical structural parameters with opposing dependencies: the C-terminal Ala883 to catalytic Asp537 distance decreases with increasing pH (∼9 Å at pH 4.75 → 7.5 Å at pH 8.05 → less than 7 Å at pH 9.15), while the specificity loop to β-hairpin spacing increases. Energy decomposition analysis reveals that this C-terminal repositioning is driven by a pH-dependent switch in the electrostatic coupling hierarchy: at acidic pH, protonated His545 and His790 dominate the coupling network with interaction energies exceeding those of the catalytic metals, whereas histidine deprotonation near pH 6 weakens these interactions by ∼50% and redirects C-terminal coupling toward the catalytic metal ions (MG1: −57.8 → −84.4 kcal/mol). These coordinated conformational changes converge at optimal pH to establish the balanced geometry required for both productive promoter engagement and efficient catalysis. At acidic pH, the displaced C-terminus and compressed promoter channel both compromise function, while at highly basic pH, excessive C-terminal proximity and widened channel geometry combine to reduce efficiency.

The Y639-RNA 3′-end distance analysis revealed that optimal pH maintains the functional open conformation (6.4 ± 0.88 Å) required for NTP substrate access, approaching the ∼8 Å spacing observed in the substrate-bound 1S76 crystal structure. In contrast, acidic (5.2 Å) and basic pH (4.1 Å) conditions promote collapsed geometries that sterically occlude binding of the incoming nucleotide, providing a direct structural explanation for reduced activity at pH extremes.

We identified Glu148 (pK _a 5.3) as an important structural pH sensor whose deprotonated carboxylate stabilizes the Met750-subdomain H interaction essential for elongation complex formation. This explains why E148A mutants initiate transcription but fail to progress beyond ∼5 nucleotide transcripts; they lack the pH-sensitive anchor required for the dramatic 70 Å N-terminal domain repositioning. Similarly, aggregate charge analysis (Figure ) revealed that the promoter-binding interface transitions from +6 charge at pH 4 to +2 at pH 8, providing a molecular explanation for how excessive positive charge at acidic pH creates an electrostatic trap that prevents promoter escape and promotes abortive cycling.

From a practical perspective, our findings provide molecular-level guidance for optimizing in vitro transcription conditions in mRNA vaccine manufacturing. The convergence of multiple structural and dynamic parameters at ≈pH 8.0 explains why this narrow range maximizes transcript yield and quality. As mRNA vaccine technology continues to expand into personalized cancer immunotherapy and rare disease treatment, such structure-guided enzyme engineering will become increasingly valuable for ensuring reliable, cost-effective production of these life-saving therapeutics. More broadly, this work demonstrates the power of constant pH molecular dynamics combined with multiscale analysis (structural dynamics, free energy landscapes, and protonation equilibria) to elucidate pH-dependent enzyme regulation. The principle that optimal enzymatic function requires convergence of multiple structural parameters, rather than optimization of any single geometric feature, may extend to other pH-sensitive enzymes where coordinated conformational transitions are essential for catalytic cycling.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsphyschemau.6c00063.

• Instructions for input and execution (ZIP)

• RMSD of backbone atoms as functions of simulation time at each pH value (Figure S1); distribution (A) and mean (B) RMSD values as functions of pH (Figure S2); RSMF of AT-rich region at the specified pH and the optimum pH (Figure S3); RMSF of Intercalating β-hairpin region at the specified pH and the optimum pH (Figure S4); RMSF of the specificity loop at the specified pH and the optimum pH (Figure S5) replica exchange acceptance rates for constant pH REMD simulations (Table S1); convergence of protonation state sampling for titratable residues across pH replicas (Figure S6); distance between specificity loop (residues 740–769) and the intercalating β-hairpin (residues 232–242) as functions of simulation time at each pH (Figure S7) the distance between (A) Ala883 (OXT)His 545 (ND1) and (B) Ala883 (OXT)His 790 (ND1) (Figure S8); cumulative variance of the Cα PCA modes at three representative pH replicas Justification of d1 and d2 as collective slow coordinates (Figure S9); absolute Pearson correlation |r| between d1 and d2 and top 20 PCA modes (Figure S10); time evolution of d1and d2 Convergence of the population landscapes (Figure S11) (PDF)

The authors declare no competing financial interest.