Chemo-Enzymatic Synthesis of Purine 3′-O-Amino-Deoxyribonucleoside Triphosphates

Abstract

3′-O-amino-deoxyribonucleoside triphosphates (3′-ONH₂-dNTPs) are crucial reversible terminators for template-independent DNA synthesis. Owing to the structural complexity of purine bases, here we propose a chemo-enzymatic process for the synthesis of purine-type 3′-ONH₂-dNTPs. First, the aminoalkoxyl group was introduced into commercially available dA and dG via a chemical catalytic process, affording yields of 24.9% and 27.3%, respectively. Subsequently, to establish a green and scalable enzymatic cascade for phosphorylation of the substrates, we engineered two key phosphorylation enzymes (Dm-dNK and MrPPK2), improving the rate-limiting monophosphorylation of 3′-ONH₂-dG from 26% to 97.8% and achieving 59.9% conversion to the corresponding triphosphate. Lastly, the produced nucleotides exhibit performance equivalent to commercial counterparts in terminal deoxynucleotidyl transferase (TdT)-mediated DNA synthesis, supporting the construction of 120-nt strands with ∼99.7% stepwise accuracy. This work establishes an efficient and generalizable catalytic platform for producing modified dNTPs, advancing the practical application of enzymatic DNA synthesis technologies.

1. Introduction

As a key enabling technology for molecular biology, genomics , and biomanufacturing, − enzymatic DNA synthesis plays an increasingly indispensable role in synthetic biology and biomedical research. , In enzymatic DNA synthesis, 3′-O-blocked deoxyribonucleoside triphosphates are widely employed to ensure the precise, single-nucleotide incorporation per cycle. , Among various blocked dNTP analogs, 3′-O-amino-modified deoxyribonucleoside triphosphates (3′-ONH₂-dNTPs) are widely utilized due to their minimal steric hindrance, which facilitates enzyme recognition, and their inherent structural stability. Currently, the phosphate groups of 3′-ONH₂-dNTPs are synthesized through chemical methods, with a yield generally ranging from 50% to 70%, , which requires selective protection and deprotection of the hydroxyl groups on nucleosides, leading to intricate synthetic steps and high purification costs. Furthermore, the extensive use of toxic reagents, such as phosphoryl chloride and triphenylphosphine, poses risks to human health and environment. Consequently, chemical triphosphate synthesis has become increasingly inadequate to meet the demands for low-cost and environmentally friendly production, which mandates the exploration of alternative synthetic approaches.

Notably, the chemical synthesis of purine-type 3′-ONH₂-dNTPs (3′-ONH₂-dATP and 3′-ONH₂-dGTP) is considerably more challenging than that of pyrimidine-type 3′-ONH₂-dNTPs (3′-ONH₂-dCTP and 3′-ONH₂-dTTP), which has become a key bottleneck limiting the large-scale preparation of 3′-ONH₂-dNTPs for enzymatic DNA synthesis. First, the purine base structures of dATP and dGTP are more complex than the pyrimidine bases of dCTP and dTTP, leading to reduced regioselectivity in phosphorylation reactions and a higher propensity for side reactions. Second, the amino and carbonyl groups on the purine ring exhibit higher reactivity, necessitating more stringent conditions for selective protection and deprotection during synthesis, which further increases the complexity of the reaction process; Lastly, the solubility of purine-type 3′-ONH₂-dNTP intermediates in reaction solvents is lower than that of pyrimidine-type analogs, resulting in lower reaction efficiency and even lower yields. ,

Enzymatic catalysis, with its advantages of high regioselectivity, strong stereospecificity, and mild reaction conditions, presents a promising solution for the phosphorylation of 3′-ONH₂-dNTPs. At present, the nucleotide phosphorylation pathways mainly include the following three categories: 1) The traditional chemical synthesis method; 2) The conventional enzymatic method that relies on ATP or PEG and other high-energy phosphate donors; 3) The emerging enzymatic method that uses inorganic phosphates instead of high-energy phosphate donors. The synthesis process based on inorganic phosphates relies on two sequential and tightly coupled phosphorylation reactions, which sequentially transfer phosphate groups to the deoxyribonucleoside substrate, ultimately forming the final triphosphate product. The first phosphorylation step is catalyzed by deoxyribonucleoside kinase, which specifically transfers a phosphate group from a phosphate donor to the 5′-hydroxyl group of deoxyribonucleosides, converting them into their corresponding monophosphate forms. Notably, the deoxyribonucleoside kinase from Drosophila melanogaster (Dm-dNK) possesses a relatively broad substrate spectrum, which can efficiently phosphorylate natural deoxyribonucleosides and various nucleoside analogs to their monophosphate products, including prodrugs − and modified nucleosides. , The second phosphorylation step is mediated by polyphosphate kinase, which further phosphorylates the monophosphorylated products to their triphosphate forms through two consecutive phosphate transfer reactions. Specifically, the polyphosphate kinase 2 from Meiothermus ruber (MrPPK2) is a Class III PPK2, which can utilize inorganic polyphosphate (PolyP) as a phosphate donor to catalyze the conversion of nucleoside monophosphates to their di- and triphosphate forms. Importantly, MrPPK2 exhibits activity not only toward natural nucleotides but also toward a range of nucleoside analogs, including 8-Bromo-AMP, Etheno-AMP and 6-Mercapto-PMP.

To achieve the enzymatic synthesis of 3′-ONH₂-dNTPs, we established an enzymatic cascade reaction using these key enzymes. Through the design and engineering of core enzymes and the introduction of a nucleoside triphosphate regeneration system, we successfully accomplished the synthesis of purine-type 3′-ONH₂-dNTPs. The products demonstrated excellent performance in TdT-mediated enzymatic DNA synthesis, enabling the construction of single-stranded DNA up to 120 nucleotides in length. This study provides a green, efficient, and scalable enzymatic synthesis strategy, thereby supporting the advancement of industrial-scale applications for enzymatic DNA synthesis technology.

2. Experimental Section

2.1. General Information

Reagents were purchased from Aladdin (Shanghai, China), OriLeaf (Shanghai, China), Merck (Darmstadt, Germany), Thermo Fisher (Waltham, USA) and Solarbio (Beijing, China), all of analytical grade or higher. Tryptone and yeast extract were purchased from OXOID (Shanghai, China). ATP was obtained from OriLeaf (Shanghai, China). 3′-ONH₂-dATP was provided by Synthgene (Suzhou, China). The 20% urea-PAGE gel solution was purchased from Coolaber (Beijing, China). 96-well plates were purchased from Corning (USA), and EchoMS-compatible 384-well plates were purchased from SCIEX (Singapore). Terminal deoxynucleotidyl transferase ZaTdT was obtained in our laboratory.

2.2. Site-Directed Saturation Mutagenesis and Library Screening

Site-directed saturation mutagenesis was performed on Dm-dNK and MrPPK2 using the degenerate codon BVT (B: cytosine/guanine/thymine; V: adenine/cytosine/guanine; T: thymine), which can encode nine amino acids including histidine, proline, arginine, aspartic acid, alanine, glycine, tyrosine, serine, and cysteine. The primers used in this study are listed in Table S2 and Table S3. For PCR amplification, the reaction system was 30 μL, consisting of 15 μL of PrimeStar DNA polymerase mix (Takara, China), 1 μL of 5′- and 3′-primers, 1 μL of DNA template, and 12 μL of sterile water. The PCR program was as follows: initial denaturation at 98 °C for 3 min, followed by 30 cycles of denaturation at 98 °C for 30 s, annealing at 58 °C for 20 s, and extension at 72 °C for 1.5 min, with a final extension at 72 °C for 5 min. The PCR products were digested with the restriction enzyme Dpn I at 37 °C for 1 h to remove the methylated DNA template. Subsequently, the products were ligated using the OK Clon DNA Ligation Kit III (Accurate Biology, China). The ligation products were transformed into E. coli BL21 (DE3) host cells (TransGen Biotech, China).

Single colonies were randomly selected and inoculated into 96-well plates containing 150 μL of 2YT medium supplemented with 80 μg/mL kanamycin. After overnight incubation at 37 °C with shaking at 700 rpm, the cultures were expanded into 96-deep-well plates containing 1 mL of 2YT medium supplemented with 80 μg/mL kanamycin for the expression of the target protein. The cell pellets were harvested by centrifugation at 3500 rpm for 5 min and lysed by the resuspending in 100 μL B-PER Bacterial Protein Extraction Reagent (Thermo Scientific, USA), followed by 15 min at 37 °C. Subsequently, 30 μL of magnetic beads with nickel filler (Beaver, China) was added to capture the protein.

100 μL of the reaction system (50 mM Tris-HCl buffer (pH 7.5), 3 mM 3′-ONH₂-dG, 1 mM ATP, 10 mM MgCl₂, 3 mM sodium hexametaphosphate, and 0.2 mg/mL SmPPK2) was directly added to the Dm-dNK-related immobilized protein. And the plates were further incubated overnight at 37 °C with shaking at 500 rpm. The high-throughput mutant reaction products were inactivated at 100 °C, diluted 5-fold, and centrifuged at 4000 rpm for 30 min. Then, 50 μL of the resulting supernatants were transferred to 384-well plates. The 384-well plates were subjected to centrifugation at 4000 rpm for 10 min. Following this, the samples were analyzed by using the EchoMS detection system. Promising candidates were selected and further confirmed by UPLC.

100 μL of the reaction system (the reaction product mixture from the first stage of 3′-ONH₂-dG to 3′-ONH₂-dGMP conversion and 3 mM sodium hexametaphosphate) was directly added to the MrPPK2-related immobilized protein. And the plates were further incubated at 37 °C with shaking at 500 rpm for 3 h. After centrifuging the reaction products at 4000 rpm for 5 min, 2 μL of the supernatant was transferred to a 98 μL PolyA tailing mix reaction system (6 μM template strand, 1× CoCl₂, 1× Tris-HCl, 0.1% Triton, ddH₂O, and 0.5 mg/mL ZaTdT). The reaction was conducted at 30 °C with shaking at 500 rpm for 10 min. Following inactivation at 100 °C, the reaction products were analyzed by gel electrophoresis on a 20% acrylamide gel and 8 M urea, and the results were processed with Tanon GIS software (Tanon, China). Promising candidates were selected and further confirmed by UPLC.

2.3. Expression of Target Genes and Purification of Proteins

The genes of Dm-dNK (with SUMO solubilization tag), MrPPK2 (with MBP solubilization tag), and SmPPK2 (with MBP solubilization tag) were synthesized by Zhonghe Gene Technology Co., Ltd. (Tianjin, China), each with an N-terminal His₆ tag, and subsequently cloned into the plasmid pET-28a­(+). The resulting recombinant expression vectors were transformed into E. coli BL21 (DE3) cells for the expression of Dm-dNK, MrPPK2, and SmPPK2 proteins, respectively. After transformation, the recombinant cells were incubated in 5 mL of 2YT medium supplemented with 80 μg/mL kanamycin at 37 °C and 220 rpm for 10 h and then scaled up to 800 mL of 2YT medium supplemented with 80 μg/mL kanamycin. When the OD₆₀₀ reached 0.6–0.8, IPTG was added to a final concentration of 0.5 mM, followed by induction of target gene expression via incubation at 16 °C with shaking at 180 rpm for 16–20 h.

The cells were collected by centrifugation at 8000 rpm for 5 min and then resuspended in 30 mL buffer (50 mM KH₂PO₄, 100 mM NaCl, pH 7.5). The resultant cell pellets were lysed by using a high-pressure homogenizer (JNBIO, China), and removed the cell debris by centrifugation at 10000 rpm for 30 min at 4 °C. The clarified lysate was loaded onto a nickel affinity column (GenScript, USA), and prewashed with a buffer (pH 7.5) containing 50 mM imidazole, 50 mM KH₂PO₄, and 100 mM NaCl. The target protein was eluted with an elution buffer (pH 7.5) containing 200 mM imidazole, 50 mM KH₂PO₄, and 100 mM NaCl. The target protein was concentrated to a final volume of 500 μL using Amicon Ultra centrifugal filter devices (Millipore, USA, cutoff 30 kDa) at 3500 rpm for 30 min, with the process repeated three times. The concentration of the purified protein was determined using a BCA Protein Assay Reagent Kit (Pierce, USA). And the protein was analyzed by SmartPAGE Precast Protein Gel Plus (Smart-Lifesciences, China). The protein was frozen with 10% glycerol and stored at −80 °C for further use.

2.4. Catalytic Function Verification

The first-stage reaction involves the Dm-dNK-catalyzed conversion of 3′-ONH₂-dA or 3′-ONH₂-dG to 3′-ONH₂-dAMP or 3′-ONH₂-dGMP. The reaction system consists of: 50 mM Tris-HCl buffer (pH 7.5), 3 mM 3′-ONH₂-dA or 3′-ONH₂-dG, 1 mM 3′-ONH₂-dATP or ATP, 10 mM MgCl₂, 3 mM sodium hexametaphosphate, 0.5 mg/mL Dm-dNK, and 0.2 mg/mL SmPPK2. The reaction was incubated at 37 °C overnight. Following completion, the reaction product was inactivated at 100 °C, filtered through a 0.22 μm filter membrane, and analyzed by UPLC.

The second-stage reaction involves the MrPPK2-catalyzed conversion of 3′-ONH₂-dAMP or 3′-ONH₂-dGMP to 3′-ONH₂-dATP or 3′-ONH₂-dGTP. The reaction system consists of: the product mixture from the first-stage reaction, 3 mM sodium hexametaphosphate, and 0.8 mg/mL MrPPK2. The reaction was incubated at 37 °C for 3 h. Following completion, the reaction product was inactivated at 100 °C, filtered through a 0.22 μm filter membrane, and analyzed by UPLC.

2.5. Enzyme Kinetics

Steady-state kinetic parameters of wild-type Dm-dNK and its M88G mutant were determined by measuring initial reaction velocities (v) at 3′-ONH₂-dG concentrations ranging from 0.1 to 12 mM. Reactions were initiated by the addition of 3′-ONH₂-dG to preincubated reaction mixtures consisting of 50 mM Tris-HCl (pH 7.5), 3 mM ATP, and 10 mM MgCl₂. The final enzyme concentration was set to 1 mg/mL for wild-type Dm-dNK and 0.5 mg/mL for the M88G variant.

For the single phosphorylation reaction converting 3′-ONH₂-dGMP to 3′-ONH₂-dGDP, steady-state kinetic parameters of wild-type MrPPK2 and its D127G mutant were determined by measuring initial reaction velocities (v) at 3′-ONH₂-dGMP concentrations ranging from 0.25 to 9 mM. Reactions were initiated by the addition of 3′-ONH₂-dGMP to preincubated reaction mixtures consisting of 50 mM Tris-HCl (pH 7.5), 3 mM sodium hexametaphosphate, and 10 mM MgCl₂. Both wild-type MrPPK2 and D127G mutant were used at a fixed final concentration of 0.2 mg/mL.

For the double phosphorylation reaction converting 3′-ONH₂-dGMP to 3′-ONH₂-dGTP, steady-state kinetic parameters of wild-type MrPPK2 and its D127G mutant were determined by measuring initial reaction velocities (v) at 3′-ONH₂-dGMP concentrations ranging from 0.25 to 6 mM. Reactions were initiated by the addition of 3′-ONH₂-dGMP to preincubated reaction mixtures consisting of 50 mM Tris-HCl (pH 7.5), 3 mM sodium hexametaphosphate, and 10 mM MgCl₂. Both wild-type MrPPK2 and D127G mutant were used at a fixed final concentration of 0.3 mg/mL.

All initial reaction velocities were quantified via UPLC analysis. The obtained v data were plotted against the corresponding substrate concentrations [S]. Kinetic parameters were calculated by fitting the experimental data to the Michaelis–Menten equation through nonlinear regression analysis using GraphPad Prism 6 (GraphPad Software, USA). All results are presented as the mean ± standard deviation (SD) of three independent biological replicates.

2.6. UPLC Analysis

UPLC analysis was conducted on an LC5190 low-pressure ultrahigh-performance liquid chromatography (UPLC) system (FULI, China) equipped with a Clarity 5 μm Oligo-RP C18 (ODS) column (250 × 4.6 mm, Phenomenex, USA). Mobile phase A was an aqueous solution containing 10 mM tetrabutylammonium hydroxide (TBAOH) and 10 mM KH₂PO₄, while mobile phase B was a methanol solution containing 10 mM tetrabutylammonium hydroxide (TBAOH). Isocratic elution was employed at a flow rate of 1 mL/min, with the UV detection wavelength was set to 254 nm, the column temperature was maintained at 30 °C, and the injection volume was 10 μL.

2.7. EchoMS Analysis

An EchoMS system (SCIEX, Canada) equipped with Triple quadrupole 6500+ was employed for the analysis. A carrier liquid consisted of 50% MeOH/ACN (v/v) supplemented with 2 mM NH₄F was used with a constant flow rate of 400 μL/min. For each sample, the injection volume was 2.5 nL, and the additional delay time was 2 s. The MS conditions were as follows: ESI source; multiple reaction monitoring (MRM) detection mode; temperature at 300 °C; negative ion voltage at 4500 V, ion source gas 1 at 90 psi, ion source gas 2 at 45 psi, curtain gas at 25 psi, and collision-activated dissociation gas at 9 psi. The dwell time was 95 ms. The declustering potential (DP) and collision energy (CE) for MRM transitions of each metabolite were −130 V and −23 V. Following sample batch acquisition, data was processed using SCIEX OS software.

2.8. LC-MS Analysis

An ACQUITY Premier system (Waters, USA) equipped with a Triple TOF 6600 mass spectrometer (SCIEX, USA) was used for the LC–MS analysis. The samples were separated using an ACQUITY UPLC BEH Amide column (100 × 2.1 mm, 1.7 μm; Waters, USA). Mobile phase A was an aqueous solution containing 25 mM ammonium acetate and 10 mM aqueous ammonia, while mobile phase B was acetonitrile. LC method was: 0–2 min, 95% B; 2–18 min, 95–65% B; 18–20 min, 65–40% B; 20–22 min, 40% B; 22–22.1 min, 40–95% B; 22.1–30 min, 95% B. A flow rate of 0.3 mL/min was employed. The MS parameters were as follows: ESI source; negative ion mode; ion voltage, 4500 V; declustering potential, −80 V; source temperature, 550 °C; curtain gas, 35 psi; nebulizer gas, 55 psi; heater gas, 55 psi. The mass ranges were m/z 63–1000.

2.9. NMR Analysis

All NMR spectroscopic measurements were performed on a Bruker Avance NEO 400 MHz spectrometer (Bruker, Germany) equipped with a iProbe BBFO 5 mm BBF/H (air) probe at ambient temperature. Samples were prepared at appropriate concentrations in deuterium oxide, and chemical shifts were referenced to the residual proton signal of the solvent.

Routine ¹H NMR spectra were acquired with the zg30 pulse sequence. The spectral width was set to 20 ppm, and the relaxation delay D1 was 1 s. The number of scans was 16, with 65536 data points collected and an acquisition time of 4.00 s.

Proton-decoupled ³¹P NMR spectra were recorded using the zgpg30 pulse sequence combined with composite pulse decoupling for broadband proton decoupling. The resonance was centered at −50 ppm, and the spectral width was 400 ppm. The relaxation delay D1 was set to 2 s, and the number of scans was 256.

All acquired data were processed and analyzed using Bruker TopSpin or MestReNova software.

2.10. Model Setup

The Dm-dNK structure was generated using AlphaFold3. The resulting structure was then used for molecular docking with AutoDock Vina to generate the Dm-dNK/3′-ONH₂-dG complex. For MrPPK2, the initial structure was obtained from the crystal structure of the MrPPK2/AMP complex (PDB ID: 5LCD, chain C). The coordinates of AMP and Mg²⁺ ions from the crystal structure were retained. The MrPPK2/3′-ONH₂-dGMP complex was then constructed by modifying AMP. All titratable residues (His, Asp, and Glu) were assigned protonation states based on pK _a values calculated using PROPKA3, , with careful visual inspection of the local hydrogen bond network. All histidine residues were set to their neutral state, and all acidic residues were set to their deprotonated state. The mutant M88G of Dm-dNK and the mutant D127G of MrPPK2 were generated by in silico site-directed mutagenesis using PyMOL based on the corresponding wild-type (WT) structures. The protein and substrate were described using the AMBER ff19SB and GAFF2 force fields, respectively. − Partial charges for the substrate were derived using the AM1-BCC method. Mg²⁺ ions were parametrized using the frcmod.ions234lm_1264_tip3p force field. The system was solvated in a TIP3P water box extending 12.0 Å from the solute boundary, and chloride ions (Cl^–) were added to neutralize the system. And the volume changes of the binding pocket were calculated using the POVME software.

2.11. Molecular Dynamics (MD) Simulations

After parametrization and system construction, each system underwent a two-stage energy minimization procedure to remove unfavorable steric contacts. In the first stage, the protein, ligands, and Mg²⁺ ions were restrained with a force constant of 50 kcal mol^–1 Å^–2, while solvent molecules and Cl^– ions were relaxed using 2500 steps of steepest descent followed by 2500 steps of conjugate gradient minimization. In the second stage, the ligand and Mg²⁺ ions were further restrained with a reduced force constant of 10 kcal mol^–1 Å^–2, and the system was minimized using the same parameters as in the first stage. Subsequently, the systems were gradually heated from 0 to 300 K over 300 ps under the NVT ensemble with a 2 fs time step, while positional restraints of 10 kcal mol^–1 Å^–2 were applied to the protein complexes. This was followed by a 200 ps equilibration under the NPT ensemble at 300 K, during which the positional restraints on the protein complexes were reduced to 5 kcal mol^–1 Å^–2. To further relax the protein side chains, an additional 200 ps NPT equilibration was performed for each system, with positional restraints of 10 kcal mol^–1 Å^–2 applied to the backbone atoms of the ligand and Mg²⁺ ions. Temperature was controlled using a Langevin thermostat with a collision frequency of 2 ps^–1, while pressure was maintained at 1 atm using a Berendsen barostat with a relaxation time of 1 ps. Subsequently, for the MrPPK2/3′-ONH₂-dGMP complex, 20 ns molecular dynamics simulations were performed under distance restraint conditions. The restraints were implemented using the AMBER DISANG file in combination with the nmropt = 1 option. The distance between the ligand and Mg²⁺ ions was constrained within a range of 2.0–3.0 Å, with a restraint force constant of 10.0 kcal mol^–1 Å^–2. Finally, 200 ns production molecular dynamics simulations were conducted in the NPT ensemble for each system, and trajectory frames were saved every 200 ps. All covalent bonds involving hydrogen atoms were constrained using the SHAKE algorithm, and a cutoff distance of 10 Å was applied for both van der Waals and short-range electrostatic interactions. Long-range electrostatic interactions were treated using the particle mesh Ewald (PME) method. Trajectory analyses were performed using the CPPTRAJ module in AmberTools. All molecular dynamics simulations were carried out using the GPU version of AMBER.

2.12. Scaled-Up Synthesis of 3′-ONH₂-dATP and 3′-ONH₂-dGTP

The enzymatic phosphorylation reaction was carried out in a 30 mL reaction mixture containing 50 mM Tris-HCl buffer (pH 7.5), 3 mM 3′-ONH₂-dA or 3′-ONH₂-dG, 1 mM 3′-ONH₂-dATP or ATP, 10 mM MgCl₂, 3 mM sodium hexametaphosphate, 0.5 mg/mL Dm-dNK, and 0.2 mg/mL SmPPK2. After incubating the mixture at 37 °C overnight, 3 mM sodium hexametaphosphate and 0.8 mg/mL MrPPK2 were added, and the reaction was continued at 37 °C for 3 h. Following completion, the reaction mixture was inactivated at 100 °C and filtered through a 0.22 μm filter membrane. The filtrate was diluted 1:1 and loaded onto a lexCap Smac Q 40 anion exchange column (Smart-Lifesciences, China). Ion exchange chromatography was initiated with 10 mM Tris-HCl (pH 7.2). After complete elution of impurities, the target products were eluted by adjusting the proportion of the elution buffer (10 mM Tris-HCl + 1 mM NaCl, pH 7.2). Fractions containing the target product were pooled and lyophilized at low temperature using a rotary evaporator. The isolated target products were analyzed by UPLC and LC-MS.

2.13. Application of 3′-ONH₂-dATP and 3′-ONH₂-dGTP Coupled with TdT

The reaction system for incorporation of a single 3′-ONH₂-modified nucleotide consisted of 1 μM oligonucleotide, 0.1 mM 3′-ONH₂-dATP or 3′-ONH₂-dGTP, 0.25 mM CoCl₂, 100 mM NaCl, 50 mM phosphate buffer (pH 6.8), and 0.5 mg/mL ZaTdT. After incubation at 40 °C for 1 min, the reaction was terminated by heating at 95 °C. The reaction products were analyzed by gel electrophoresis on a 20% acrylamide gel and 8 M urea, and the results were processed with Tanon GIS software (Tanon, China).

The 120-nt enzymatic DNA synthesis reaction was performed using a synthesizer (Model: TIESyno-96) and kit (Cat: 800201–00401) manufactured by Zhonghe Gene Technology Co., Ltd. (Tianjin, China), with the detailed procedures carried out in accordance with the manufacturer’s instructions.

3. Results and Discussion

3.1. Design of the Synthetic Route for 3′-ONH₂-dNTPs

Owing to the more complex base structures, greater number of reactive hydrogen-bonding sites, and more stringent reaction requirements, our research primarily focuses on the synthesis of purine-type 3′-ONH₂-dNTPs via integrating chemical and enzymatic catalysts together. , The process includes three steps: initially, aminooxy was added into commercially available dA and dG to generate 3′-ONH₂-dA and 3′-ONH₂-dG via chemical methods; second, monophosphorylation reaction was catalyzed by deoxyribonucleoside kinase (dNK) to yield 3′-ONH₂-dAMP and 3′-ONH₂-dGMP; last, two-step phosphorylation was catalyzed by polyphosphate kinase (PPK) to produce 3′-ONH₂-dATP and 3′-ONH₂-dGTP (Figure a).

1.

Synthesis of 3′-O-NH₂-dATP and 3′-O-NH₂-dGTP. (a) Technical route for enzymatic synthesis of 3′-O-NH₂-dATP and 3′-O-NH₂-dGTP. (b) Mass spectrum of 3′-ONH₂-dA sample. (c) Mass spectrum of 3′-ONH₂-dG sample.

Based on the previous reports, , dA and dG were used as starting materials for the synthesis of 3′-ONH₂-dA and 3′-ONH₂-dG at first. The synthesis proceeded through a sequential route: 5′-hydroxyl protection with DMT-Cl, Mitsunobu reaction to install the 3′-aminooxy moiety, and stepwise removal of protecting groups. Throughout the synthesis, reactions were monitored by thin-layer chromatography (TLC) or liquid chromatography–mass spectrometry (LC-MS), and all intermediates were rigorously purified via liquid–liquid extraction, normal-phase silica gel column chromatography (eluent: dichloromethane/methanol = 10:1–30:1), or methyl tert-butyl ether (MTBE) trituration (see Supporting Information and Figures S1–S2 for detailed procedures). Isolated yields of the final target products were determined quantitatively via UPLC analysis, affording 2.64 g of 3′-ONH₂-dA (24.9%) and 2.88 g of 3′-ONH₂-dG (27.3%). The chemical structures of both products were verified by LC-MS in negative ion mode, where the deprotonated molecular ion [M–H]⁻ was detected (Figure b and Figure c). For 3′-ONH₂dA (C₁₀H₁₄N₆O₃), the measured exact mass was 265.1052 Da (theoretical 265.1060 Da), corresponding to a relative error of 3.02 ppm (Figure b). For 3′-ONH₂dG (C₁₀H₁₄N₆O₄), the measured exact mass was 281.1003 Da (theoretical 281.1012 Da), with a mass error of 3.20 ppm (Figure c). These results confirm that we have successfully synthesized 3′-ONH₂-dA and 3′-ONH₂-dG, which can serve as substrates for subsequent phosphorylation.

3.2. Engineering Dm-dNK for the Synthesis of Purine Monophosphates

Dm-dNK has been widely reported to efficiently catalyze the four natural deoxynucleosides as well as a variety of nucleoside analogs. − Therefore, we tried to utilize this enzyme to phosphorylate 3′-ONH₂-dA and 3′-ONH₂-dG. The results demonstrated that at a substrate concentration of 3 mM, the wild-type Dm-dNK achieved a conversion rate of up to 99% for 3′-ONH₂-dA, whereas the conversion of 3′-ONH₂-dG was only about 26% (Figure S3). Molecular docking of both 3′-ONH₂-dG and natural dG into the active cavity of Dm-dNK revealed that the 3′- ONH₂ modification increases steric bulk, which would require the enzyme to form a larger active pocket (Figure a).

2.

Engineering and semirational designs of Dm-dNK for phosphorylation of 3′-ONH₂-dG. (a) Comparison of the docking poses of dG (yellow sticks) and 3′-ONH₂-dG (red sticks) in the wild-type Dm-dNK. (b) Residues within a 4 Å radius of 3′-ONH₂-dG in the wild-type Dm-dNK. (c) Phosphorylation conversion rate of 3′-ONH₂-dG by Dm-dNK single-point mutants and combinatorial mutants. All reactions were performed in triplicate (n = 3). Data are presented as the mean ± standard deviation (SD) of three independent experiments, where error bars represent the SD of repeated measurements and the centers of the error bars represent the mean of repeated measurements. (d) The pocket volumes of Dm-dNK and its mutant M88G. Cyan spheres represent the volumes of the binding-pockets. (e) Root mean square fluctuation (RMSF) profiles of the protein backbone for the M88G mutant and Dm-dNK during molecular dynamics simulations.

Owing to steric hindrance originating from the modified substrate, wild-type Dm-dNK exhibits poor catalytic activity toward 3′-ONH₂-dG. Therefore, it is required to engineer Dm-dNK to enhance its catalytic performance. Complete random mutagenesis suffers from substantial drawbacks including oversize mutant libraries, low screening efficiency and high experimental randomness. Although molecular dynamics (MD) simulations can assist in predicting mutation effects, each mutant requires at least three replicate simulations to ensure the reliability of screening results and obtain the optimal variant. This approach not only consumes enormous computational resources but also greatly prolongs the research period due to the subsequent experimental validation of numerous mutants. Accordingly, a semirational design strategy integrating structural simulation analysis and site-directed mutagenesis was employed herein to precisely pinpoint key amino acid residues within the active cavity responsible for substrate binding and catalytic turnover. As shown in Figure b, the 4Å-radius active cavity region of Dm-dNK is composed of several key amino acid residues, including I29, K33, E52, V54, W57, L66, Y70, F80, Q81, V84, M88, R105, A110, F114, E172, and V175. Site-saturation mutagenesis targeting these residues was performed using the degenerate codon BVT (B: C/G/T; V: A/C/G; T: T), generating a mutant library of 432 variants (16 sites × 27 variants per site) for screening.

Preliminary high-throughput screening based on the Echo-MS method indicated that beneficial mutations were primarily localized at residues L66, V84, M88, A110, E172, and V175 (Figure S4). Subsequent confirmatory screening by UPLC identified 10 single-point mutants (L66Y, V84A, V84S, M88A, M88G, M88E, M88H, A110G, E172D and V175C) exhibited significantly enhanced activity (Figure c). Notably, the substrate conversion rates for mutants M88G, M88A and A110G reached 97.8%, 86.2% and 90.7%, respectively, representing more than 3-fold improvement over the wild-type enzyme. To explore synergistic effects of positive mutations, M88G was used as the template for further directed evolution. Combinatorial active-site saturation testing/iterative saturation mutagenesis (CAST/ISM) with residues L66, V84, A110, E172, and V175 to generate double mutants. , However, the catalytic activity of these combinatorial mutants was either comparable to or lower than that of M88G alone (Figure c). Therefore, M88G was selected for the subsequent monophosphorylation to 3′-ONH₂-dGMP (Figure S5). A comparison of the kinetic parameters between the M88G mutant and wild-type Dm-dNK further confirmed that the M88G mutant exhibited remarkably improved catalytic efficiency toward 3′-ONH₂-dG (Table and Figure S6).

1. Kinetic Parameters (K _m, k _cat, k _cat/K _m) of Dm-dNK and Its M88G Mutant toward 3′-ONH₂-dG.

Source	Variant	Substrate	K m [mM]	k cat [min–1]	k cat/K m [M–1·s–1]
Dm	WT	3′-ONH2-dG	6.844	0.223	1.955 × 103
Dm	M88G	3′-ONH2-dG	2.738	0.994	2.178 × 104

We observed that the mutants M88G, M88A, and A110G, which substitute the wild-type residues with amino acids bearing smaller side chains (alanine or glycine), effectively increase the volume of the active pocket, thereby better accommodating the 3′-ONH₂dG (Figure d and Figure S7). Specifically, the M88G mutation increases the active site cavity volume from 132 ų in the wild-type enzyme to 195 ų (Figure d). The enhanced catalytic efficiency of these mutant enzymes confirms the critical role of these residues in processing modified nucleotides, which is consistent with previous studies. ,

Furthermore, we performed 200 ns molecular dynamics (MD) simulations of Dm-dNK and its optimal M88G mutant in complex with the substrate 3′-ONH₂-dG to investigate the effect of this mutation on protein dynamics and substrate-binding stability. The systems were analyzed using RMSD, RMSF, hydrogen bonding, and π-π interaction analyses. The results show that, compared with the wild-type enzyme, the M88G mutant exhibits lower and more stable RMSD values for both the protein backbone and the substrate (Figure S8), indicating reduced overall conformational fluctuations and a more stable substrate-binding state. Meanwhile, RMSF analysis reveals a slight increase in local flexibility around residue 88 in the M88G mutant (Figure e). This behavior can be attributed to the substitution of methionine with glycine, which significantly reduces side-chain bulkiness, thereby decreasing steric hindrance in the active site and increasing local conformational freedom. Importantly, although local flexibility around residue 88 increases in the M88G mutant, the overall structural stability of the enzyme is maintained, as reflected by the reduced backbone RMSD values. These results suggest that the mutation achieves a favorable balance between enhanced local conformational flexibility and global structural stability.

Hydrogen bond analysis further indicates that the M88G mutation significantly enhances hydrogen-bond interactions between the enzyme and the substrate. In the mutant system (Figure S9a), the substrate forms more stable and persistent hydrogen bonds with key residues, including Glu52, Glu171, Glu172, and Gln81. In particular, the hydrogen bond between the substrate and Glu52 persists throughout the entire simulation, while the interaction with Glu171 becomes stable after approximately 20 ns and persists thereafter. In contrast, these interactions in the wild-type system are intermittent or unstable (Figure S9b), indicating that the M88G mutation contributes to the formation of a more stable hydrogen-bonding interaction network.

To further evaluate aromatic interactions within the active site, centroid distances between the substrate pyrimidine ring and aromatic residues were analyzed. In the M88G mutant (Figure S10a), the centroid distance between the substrate and Phe80 is predominantly distributed in the 5–6 Å range, with reduced fluctuations and a more concentrated probability density, indicating the formation of a relatively stable but weak π–π stacking interaction. Nevertheless, its stability is improved compared with the wild-type system. In addition, the centroid distance between the substrate and Phe114 is mainly distributed within 3–5 Å (Figure S10b), also showing reduced fluctuations and a more concentrated distribution, suggesting a stronger and more stable π-π stacking interaction. In contrast, the wild-type system exhibits a broader distribution (approximately 3–6 Å) with a clear multimodal profile, indicating weaker and less stable π-π interactions.

Overall, the M88G mutation enhances substrate-binding stability by reducing steric hindrance at position 88 and increasing local conformational flexibility, thereby improving the dynamic regulation of the active site and promoting the formation of more stable hydrogen-bonding and π-π stacking interactions. This ultimately leads to a significantly more stable substrate-binding conformation.

3.3. Engineering MrPPK2 for the Synthesis of Purine Triphosphates

MrPPK2 has been reported to be capable of using polyphosphoric acid as a phosphate donor to catalyze the further conversion of nucleoside monophosphates and their analogs into nucleoside triphosphates. Therefore, we sought to exploit this enzyme for the phosphorylation of 3′-ONH₂-dAMP and 3′-ONH₂-dGMP, which were obtained from our preceding experiments. The results demonstrated that MrPPK2 can only phosphorylate 3′-ONH₂-dAMP with a conversion of 66%, while exhibiting negligible catalytic activity toward 3′-ONH₂-dGMP (Figure S11). Hence, following the same structure-guided semirational design strategy, we aimed to improve the catalytic activity of MrPPK2 toward 3′-ONH₂-dGMP by altering the amino acid residues in the enzyme’s catalytic active pocket. As shown in Figure a, the 4Å-radius active site region of MrPPK2 is composed of several key amino acid residues, including F91, G92, V93, P94, E98, L99, Y103, R106, V107, R122, S123, E126, D127, and V131. Site-directed saturation mutagenesis was performed on these residues using the degenerate codon BVT, generating a mutant library of 378 variants (14 sites × 27 variants per site) for screening.

3.

Engineering and semirational designs of MrPPK2 for phosphorylation of 3′-ONH₂-dGMP. (a) Residues within a 4 Å radius of 3′-ONH₂-dGMP in the wild-type MrPPK2. (b) Phosphorylation conversion rate of 3′-ONH₂-dGMP by MrPPK2 single-point mutants. All reactions were performed in triplicate (n = 3). Data are presented as the mean ± standard deviation (SD) of three independent experiments, where error bars represent the SD of repeated measurements and the centers of the error bars represent the mean of repeated measurements. (c) The binding-pocket volumes of MrPPK2 and its mutant D127G, which are 574.375 and 586.875 ų. (d) Root mean square fluctuation (RMSF) profiles of the protein backbone for the D127G mutant and MrPPK2 during molecular dynamics simulations.

The results showed that only a small subset of mutants exhibited catalytic activity toward 3′-ONH₂-dGMP, and these positive mutants were localized at the sites of V93, E126, and D127 (Figure S12). Further UPLC analysis confirmed that 7 mutants including V93S, V93Y, E126G, D127G, D127A, D127S, and D127P exhibited catalytic activity (Figure b). Among them, the mutants V93S, V93Y, and E126G showed poor catalytic activity toward 3′-ONH₂-dGMP, with substrate conversion rates of only 1.5%, 6.3%, and 1.1%, respectively. However, the four mutants localized at the D127 siteD127G, D127A, D127S, and D127Pexhibited better catalytic activity, with substrate conversion rate of 59.9%, 54.8%, 52.0%, and 46.1%, respectively. Therefore, D127G was selected for the subsequent phosphorylation to 3′-ONH₂-dGTP (Figure S13). A comparison of the kinetic parameters between the D127G mutant and wild-type MrPPK2 further confirmed that both the monophosphorylation and pyrophosphorylation efficiencies of the D127G mutant toward 3′-ONH₂-dGMP were significantly improved (Table and Figures S14–S15).

2. Kinetic Parameters (K _m, k _cat, k _cat/K _m) of MrPPK2 and Its D127G Mutant toward 3′-ONH₂-dGMP .

Source	Variant	Substrate	K m [mM]	k cat [min–1]	k cat/K m [M–1·s–1]
Mr	WT	3′-ONH2-dGMP	ND	ND	ND
			ND	ND	ND
Mr	D127G	3′-ONH2-dGMP	4.938	143.34	1.74 × 106
			1.347	2.605	1.16 × 105

^aND, not determined; WT, wildtype.

^bDenotes the monophosphorylation of 3′-ONH₂-dGMP to form 3′-ONH₂-dGDP.

^cDenotes the pyrophosphorylation of 3′-ONH₂-dGMP to form 3′-ONH₂-dGTP.

We observed that compared with the wild-type, the D127G, D127A and D127S mutants all have truncated side chains. This structural modification facilitates the expansion of the catalytic pocket, allowing it to accommodate the bulkier 3′-ONH₂dGMP substrate relative to the natural substrate, thereby enhancing catalytic efficiency (Figure c and Figure S16).

Furthermore, we performed 200 ns MD simulations of MrPPK2 and its optimal D127G mutant in complex with the substrate 3′-ONH₂-dGMP to investigate the effect of this mutation on protein dynamics and substrate-binding stability. The systems were analyzed using RMSD, RMSF, and hydrogen-bond interaction analyses. The results show that, compared with the wild-type enzyme, the D127G mutant exhibits lower and more stable RMSD values for both the protein backbone and the substrate (Figure S17), indicating enhanced overall structural stability and a more persistent substrate-binding state. Meanwhile, RMSF analysis reveals a decrease in local flexibility around residue 127 in the D127G mutant (Figure d), suggesting reduced positional fluctuations and improved local conformational stability. This behavior may be attributed to the removal of the charged side chain of aspartate upon mutation, which also reduces steric hindrance. As a result, local spatial and interactional perturbations within the active site are weakened, allowing the substrate interaction network to reorganize into a more stable conformational state.

Hydrogen bond analysis further indicates that the D127G mutation significantly enhances hydrogen-bond interactions between the enzyme and the substrate. In the mutant system (Figure S18a), a stable and persistent hydrogen bond is formed between the substrate and Arg182, whereas this interaction is lost after approximately 140 ns in the wild-type system (Figure S18b). In addition, enhanced hydrogen-bond interactions are observed between the substrate and Lys191, Lys188, and Glu187. These results indicate that the D127G mutation promotes the formation of a more stable hydrogen-bonding interaction network.

Overall, the D127G mutation enhances substrate-binding stability by reshaping the interaction network within the active site.

3.4. Large-Scale Production and Applications of Modified Purine Nucleotides

We established a larger-scale synthesis of 3′-ONH₂-dATP and 3′-ONH₂-dGTP using a cascade system composed of Dm-dNK and MrPPK2. Specifically, wild-type Dm-dNK and wild-type MrPPK2 were used for the cascade phosphorylation of 3′-ONH₂-dA to 3′-ONH₂-dATP; while the Dm-dNK mutant M88G and MrPPK2 mutant D127G were employed for the cascade phosphorylation of 3′-ONH₂-dG to 3′-ONH₂-dGTP. Since 3′-ONH₂-dATP and ATP serve as phosphate donors for Dm-dNK catalyzed conversion of 3′-ONH₂-dA/dG to 3′-ONH₂-dAMP/dGMP, we introduced SmPPK2 to establish a regeneration cycle of 3′-ONH₂-dATP and ATP (Figure S19). This design reduces the initial input of 3′-ONH₂-dATP and ATP in the first-stage reaction, which not only lowers reaction costs but also simplifies subsequent separation and purification by minimizing excess nucleotide impurities. The enzymatic phosphorylation reaction was carried out in a 30 mL mixture containing the substrate at a concentration of 3 mM. Upon completion of the reaction, the synthesized 3′-ONH₂-dATP and 3′-ONH₂-dGTP were purified by ion-exchange chromatography and confirmed using LC-MS, UPLC and NMR spectroscopy (Figure a, Figure b and Figures S20–S21). Finally, the isolated yield of 3′-ONH₂-dATP was 30.0% with a purity of 97%, and the isolated yield of 3′-ONH₂-dGTP was 36.3% with a purity of 85.1% (Figure S22).

4.

Application of the isolated products 3′-ONH₂-dATP and 3′-ONH₂-dGTP. (a) Mass spectra of 3′-ONH₂-dATP sample. (b) Mass spectra of 3′-ONH₂-dGTP sample. (c) Denaturing PAGE analysis of ZaTdT catalyzing reactions with 16 initiator strands, commercial 3′-ONH₂-dATP (Control) and self-prepared 3′-ONH₂-dATP (Sample) were used as substrates, respectively. (d) Denaturing PAGE analysis of ZaTdT catalyzing reactions with 16 initiator strands, commercial 3′-ONH₂-dGTP (Control) and self-prepared 3′-ONH₂-dGTP (Sample) were used as substrates, respectively. (e) Single-step synthesis accuracy of 120-nt DNA by TdT with commercial 3′-ONH₂-dATP/dGTP (Control) and self-prepared 3′-ONH₂-dATP/dGTP (Sample) as substrates.

To evaluate the application potential of the purified products, we compared their performance with commercial 3′-ONH₂-dATP and 3′-ONH₂-dGTP in TdT-mediated enzymatic DNA synthesis. Using 16 distinct nucleotide strands (Table S1), we tested single-nucleotide incorporation via ZaTdT. Polyacrylamide gel electrophoresis (PAGE) analysis showed that self-prepared 3′-ONH₂-dATP and 3′-ONH₂-dGTP exhibited comparable incorporation efficiency to their commercially available counterparts (Figure c and Figure d). Furthermore, we performed 120-nt enzymatic DNA synthesis: the control group used four commercial 3′-ONH₂-dNTPs, while the sample group replaced commercial 3′-ONH₂-dATP and 3′-ONH₂-dGTP with our self-prepared products. As shown in Figure e, the single-step synthesis accuracy of the sample group reached 99.70%, with no significant differences compared to the control group (99.69%). Collectively, our work demonstrates that the 3′-ONH₂-dATP/dGTP generated through enzymatic catalysis are viable replacements for chemical substrates, validating the scalable enzymatic cascade as a feasible new pathway to synthesize reagents for enzymatic DNA synthesis.

4. Discussion

This study establishes an efficient and green biocatalytic cascade for synthesizing 3′-ONH₂-dATP and 3′-ONH₂-dGTP, critical substrates for template-independent enzymatic DNA synthesis. Prior to this work, the synthesis of 3′-ONH₂-dNTPs was overwhelmingly dominated by chemical phosphorylation routes, and no chemo-enzymatic strategy for 3′-ONH₂-dNTPs has been reported to date. This work represents the first successful development of a scalable chemo-enzymatic platform for purine-type 3′-ONH₂-dNTPs, filling a key technical gap in the field.

By addressing the inherent substrate specificity limitations of wild-type enzymes through structure-guided engineering, we developed a scalable enzymatic alternative to traditional chemical phosphorylation methods. Traditional chemical phosphorylation of 3′-modified purine nucleosides typically relies on iterative protection/deprotection procedures, strict regioselective control, and hazardous reagents such as phosphoryl chloride and triphenylphosphine under harsh conditions. This approach not only generates large amounts of phosphorus-containing and organic waste (approximately hundreds of grams of waste per gram of product), but also suffers from poor regioselectivity (≤70%) and low overall yield (≤50%). In contrast, our chemo-enzymatic route combines mild chemical aminooxylation with a one-pot enzymatic cascade reaction, eliminating redundant protection/deprotection steps and the use of toxic phosphate donors. Wild-type enzymes exhibited poor activity toward 3′-ONH₂-dG (monophosphorylation yield ∼26%). In contrast, our engineered Dm-dNK (M88G) improved the yield to 97.8%, and the engineered MrPPK2 (D127G) achieved a 59.9% yield of 3′-ONH₂-dGTP. By comparison, chemical phosphorylation of purine 3′-ONH₂-dNTPs only affords yields of 50% or lower, due to poor regioselectivity that triggers extensive side reactions.

The core achievement lies in the successful redesign the active cavity of enzyme. For Dm-dNK, the activity toward the more sterically demanding substrate 3′-ONH₂-dG was poor. Residue M88 was identified as a critical constraint. The M88G mutation, by removing the bulky side chain, expanded the active site cavity volume from 132 ų to 195 ų, thereby enhancing the conversion rate from 26% to 97.8%. Similarly, the wild-type MrPPK2 was essentially inactive toward 3′-ONH₂-dGMP. Introducing the D127G mutation restored activity (59.9% conversion) by mitigating steric hindrance within the active site. These results underscore the critical role of active site compactness in governing substrate scope and provide a clear engineering blueprint for adapting kinases to bulkier nucleotide analogs.

Building on these engineered biocatalysts, we constructed an optimized three-enzyme cascade involving Dm-dNK, MrPPK2, and an ATP regeneration module catalyzed by SmPPK2. This system achieved the efficient one-pot conversion of 3′-ONH₂-dA and 3′-ONH₂-dG to their corresponding triphosphates. The scalability of this route was demonstrated at a 30 mL preparative scale, yielding isolated 3′-ONH₂-dGTP and 3′-ONH₂-dATP with purities of 85.1% and 97%, respectively. This cascade highlights the advantages of enzymatic synthesis, including milder conditions, fewer protection/deprotection steps, and a reduced environmental footprint compared to organic phosphitylation chemistry.

Most importantly, the practical utility of the enzymatically synthesized nucleotides was rigorously confirmed. In TdT-catalyzed single-nucleotide incorporation assays across 16 different initiator strands, the performance of our 3′-ONH₂-dATP and 3′-ONH₂-dGTP was indistinguishable from that of commercially available chemicals. Furthermore, in a demanding 120-nt DNA synthesis experiment, the use of self-prepared substrates achieved a per-step accuracy of 99.70%, statistically identical to the control (99.69%). This functional validation confirms that the enzymatic products are fully competent for high-fidelity enzymatic DNA synthesis.

5. Conclusions

In conclusion, this work provides a robust, sustainable, and generalizable platform for the synthesis of modified dNTPs. The integration of protein engineering, cascade design, and ATP regeneration offers a powerful alternative to conventional chemical methods. The principles established hereparticularly active site enlargement to accommodate steric modificationsare readily applicable to the production of other valuable nucleotide analogs, thereby supporting the advancing frontiers of synthetic biology, DNA data storage, and therapeutic oligonucleotide manufacturing.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/cbe.6c00032.

• Chemical modification methods for 3′-ONH₂-dA and 3′-ONH₂-dG, UPLC chromatograms, experimental validation of single-site saturation mutants of Dm-dNK and MrPPK2, kinetic comparisons between wild-type enzymes and their mutants, active-site pocket volume analyses of Dm-dNK and MrPPK2 variants, molecular dynamics (MD) simulation results, NMR spectra, SDS-PAGE analysis of purified proteins, and sequences of oligonucleotide strands and primers (PDF)

H.J. conceived and designed the study; Q.H. carried out most experiments, with technical assistance from X.L., J.S., and L.L.; H.J. and X.L. supervised the work; S.C. contributed to mutant library screening and UPLC analysis. J.Y., S.Z., and J.C. contributed to the analysis of the enzyme’s catalytic mechanism. C.L. and X.Q. performed the experiments of enzymatic DNA synthesis. Z.Z. performed the LC-MS analysis. Q.H. and X.L. interpreted the data. Q.H., X.L., L.L., and H.J. wrote the manuscript. All authors contributed to the writing of the manuscript, reviewed, and approved the final version of the manuscript.

The authors declare no competing financial interest.

Published as part of Chem & Bio Engineering special issue “Synthetic Biology: Enabling Technologies and Practical Applications”.