ATP, GTP, UTP, CTP, CP, creatine kinase, E. coli total tRNA mixture, and restriction enzymes were purchased from Roche Applied Science (Indianapolis, IN, United States). _L-[U-¹⁴C] leucine (11.9 GBq/mmol) was obtained from Amersham Biosciences (Uppsala, Sweden). 1-Step Ultra TMB-ELISA solution and HisProbe-HRP Conjugate were purchased from Thermo Fisher Scientific (Waltham, MA, United States). High-fidelity VELOCITY DNA polymerase was obtained from Bioline (London, United Kingdom). The hemA gene of Rhodobacter capsulatus (ATCC 11166) encoding 5-aminolevulinic acid synthase (ALAS) was synthesized by Twist Bioscience (San Francisco, CA, United States) after optimizing the codons for expression in E. coli. PCR primers listed in Supplementary Table S1 were synthesized by Macrogen (Seoul, Korea). E. coli S12 extracts were prepared from the BL21 Star (DE3) strain as described previously (Park et al., 2017). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO, United States) and used without further purification.

The hrp gene was cloned into the pET21a vector using NdeI/SalI restriction sites to generate the pET21a-HRP construct. The template DNA for CFPS was prepared by PCR using a forward primer for the 15 base pairs upstream of the T7 promoter (T7Pro-15UP), and a reverse primer for the T7 terminator (T7Ter). The PCR mixture consisted of 1 ng/ µl template, 200 nM of each (forward and reverse) primer, 0.02 U/µl DNA polymerase, 1 mM each of dNTP, and 1 × Hi-Fi buffer (Bioline). Thermal cycling involved heating at 95°C for 2 min, followed by 28 cycles at 95°C for 30 s, 55°C for 1 min, 72°C for 2 min, and a final extension at 72°C for 2 min. PCR products were purified using a PCR Clean-up kit (Promega, Madison, WI, United States) prior to use for CFPS. For randomization of the +2 and +3 codons, the hrp gene was amplified using a forward primer containing degenerated +2 and +3 codons (+2/+3MutHRP, Supplementary Table S1), re-cloned into the pET21a plasmid, and transformed into competent E. coli DH5α cells. After overnight culturing of the transformed E. coli DH5α cells on Luria-Bertani (LB) agar plates, individual colonies were picked for colony PCR to prepare template DNA for CFPS.

The standard reaction mixture for cell-free synthesis of HRP and ubiquitin-fused HRP (ubi-HRP) consisted of 57 mM HEPES-KOH (pH 8.2); 1.2 mM ATP; 0.85 mM each of CTP, GTP, and UTP; 0.17 mg/ml E. coli total tRNA mixture (from strain MRE600); 0.64 mM cAMP; 90 mM potassium glutamate; 80 mM ammonium acetate; 12 mM magnesium acetate; 34 μg/ml _L-5-formyl-5,6,7,8-tetrahydrofolic acid; 2 mM each of the 20 essential amino acids; 10 μM _L-[U-¹⁴C]leucine (11.9 GBq/mmol); 2% (w/v) polyethylene glycol-8000; 67 mM creatine phosphate; 3.2 µg/ml creatine kinase; 2 mM calcium acetate; 67.7 µM hemin; 5 mM reduced glutathione (GSH); 5 mM oxidized glutathione (GSSG); 13.3 ng/µl template DNA; and 26% (v/v) S12 extract. A mixture of S12 extract and ubiquitin hydrolase (UBP1)-enriched S12 extract (3:1 v/v) was used in experiments involving in situ cleavage of ubiquitin during the synthesis of ubi-HRP. Cell-free synthesis reactions were performed for 3 h at different temperatures.

The amount of cell-free synthesized protein was determined by measuring trichloroacetic acid (TCA)-precipitated radioactivity of the reaction mixture as described previously (Byun et al., 2019). The amount of soluble protein was determined by measuring the TCA-precipitated radioactivity of the supernatant after centrifugation (10,000 × g, 10 min) of the reaction mixture. For western blot analysis to determine the size of synthesized protein, poly-histidine tag was added to the C-terminus of HRP. HisProbe-HRP was used for direct detection of poly-histidine tagged proteins following the manufacturer’s instruction. Enzymatic activity of HRP was determined using 1-Step Ultra TMB-ELISA solution, which contains 3,3’,5,5’-tetramethylbenzidine (TMB) and hydrogen peroxide. After diluting the completed reaction mixture with 10 volumes of phosphate-buffered saline (PBS), 2 µl of the diluted solution was mixed with 50 µl TMB substrate solution. After a 3 min incubation at room temperature, 50 µl of 1 M sulfuric acid was added to quench the chromogenic reaction, and the absorbance was measured at 450 nm using a Clariostar microplate reader (BMG Labtech, Ortenberg, Germany).

The hemA gene was cloned into the pET28a plasmid using the NcoI/XhoI restriction sites to generate the pET28a-ALAS construct. For preparation of ALAS, E. coli strain BL21 Star (DE3) cells transformed with the pET28a-ALAS plasmid were cultured in 200 ml LB medium supplemented with kanamycin (25 mg/L). Isopropyl β-D-1-thiogalactopyranoside (1 mM) was added to induce the expression of ALAS after cells had grown to an absorbance at 600 nm (OD₆₀₀) of 0.6. Cells were harvested at OD₆₀₀ = 4, and disrupted with a French Press cell at 12,000 psi after being resuspended in lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 8.0). The supernatant of the lysate resulting from centrifugation (15,000 × g for 30 min at 4°C) was loaded onto a 5 ml Ni-NTA agarose bead column (Qiagen, Hilden, Germany). After washing four times with 10 ml washing buffer (50 mM NaH₂PO₄, 300 mM NaCl, 30 mM imidazole, pH 8.0), ALAS was eluted with 2 ml elution buffer (50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole, pH 8.0). The eluate was dialyzed against 2 L PBS using Spectra/Por 2 dialysis tubing (molecular weight cut-off: 12–14 kDa) according to the manufacturer’s instructions (Repligen, Waltham, MA, United States). The purity of ALAS was determined by 16% tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by Coomassie Blue staining (Supplementary Figure S3). The concentration of purified ALAS was measured by Bradford assay (Bradford, 1976).

Peroxidases catalyze redox reactions on a wide range of substrates, and have potential industrial and biotechnological applications (Regalado et al., 2004; Lopes et al., 2014; de Oliveira et al., 2021). HRP C1A (hereafter referred to as HRP), the major enzyme among the 19 isoenzymes produced in horseradish root (Krainer and Glieder, 2015), is one of the most extensively studied peroxidases. Comprising 308 amino acids, this enzyme contains a hemin prosthetic group and two Ca²⁺ ions. It also requires four sets of properly formed disulfide bridges for full biological activity (Veitch, 2004). When expressed in E. coli cells, recombinant HRP forms inclusion bodies, and requires time- and labor-intensive refolding steps to recover the functional enzyme (Chaunhan and Kang, 2018). Similarly, incubation of template DNA in the standard reaction mixture produced ∼0.4 μM (16 μg/ml) of inactive protein (Figure 1). Taking advantage of the open nature of CFPS, we attempted to adjust the molecular environment of the translation reaction to achieve expression of functional HRP. Assuming that inefficient formation of disulfide bonds was one of the factors preventing the synthesis of functional HRP, we replaced the S12 extract with one enriched with DsbA and DsbC (Schlapschy et al., 2006). As expected, this led to the generation of significantly increased peroxidase activity in the reaction mixture (Supplementary Figure S1A). It was also found that expression of functional HRP was heavily dependent on the reaction temperature of cell-free synthesis: optimizing the reaction temperature markedly enhanced HRP activity (Supplementary Figure S1B). Although a lower temperature generally reduces the overall yield of protein synthesis, HRP activity in the reaction mixture increased 3-fold when the reaction temperature was shifted from 30°C to 20 °C. The concentrations of calcium, hemin, and glutathione were subsequently optimized to maximize the yield of functional HRP (Supplementary Figures S1C−E). The optimized reaction conditions (S12 extract enriched with DsbA/C, 20 °C reaction temperature, 0.5 mM GSH, 0.5 mM GSSG, 16.7 mM hemin, 1.0 mM calcium acetate) were similar to those reported previously by Zhu et al. (2015). Under these conditions, ∼57% synthesized HRP was partitioned in the soluble fraction and exhibited 4-fold higher peroxidase activity than enzyme produced under standard reaction conditions (Figure 1).

After optimizing the physicochemical reaction parameters, we attempted to enhance the production of HRP by engineering the nucleotide sequence of the template DNA. It has been reported that the efficiency of protein translational is strongly influenced by the nucleotide sequences of the initial codons downstream of the start codon. For example, when nucleotides were randomly changed at the +2 and +3 codons of human erythropoietin (hEPO), several variant genes exhibited expression yield increases by as much as 7-fold (Ahn et al., 2008). Based on this, we randomized the +2 and +3 codons of the HRP gene and measured the relative expression levels of 500 variant clones. As expected, the variant HRP genes exhibited a wide distribution of expression levels, some of which were substantially higher than the parental sequence (Figure 2A). Interestingly, subsequent enzyme activity assays revealed that expression levels of some variants did not always correlate with enzymatic activity. For example, while the 1F7 variant exhibited more than a 2-fold increase in HRP production, the peroxidase activity of this variant was less than 40% that of the parental gene. In total, 15 out of 500 variant clones showed enhanced expression levels, but only two exhibited enzyme activity similar to that of the parental gene (Figure 2B). Sequencing of the variant genes revealed that it is critical for variant enzymes to encode leucine in the +3 codon for HRP activity. The three clones exhibiting significant activity were found to have synonymous +3 leucine codons, and all other clones had substituted codons at the same position (Figure 2C). Moreover, in a separate experiment in which the +3 position was replaced with the other 19 essential amino acids, even conservative replacements (glycine, alanine, and valine) failed to support HRP activity (Supplementary Figure S2A). In addition to the +3 position, the identity of the codon at the +2 position also appeared to affect HRP activity because the three clones in which the +2 codon was changed to lysine or proline from glutamine exhibited comparable or lower activity, although the expression levels were enhanced by almost 2-fold. Because these N-terminal residues are not in close proximity to the cofactor pockets or active site (Supplementary Figures S2B,C), it seems reasonable to infer that the leucine residue at this position plays an important role in forming the active structure of HRP (Yates et al., 2014; Kelly et al., 2015).

The results described above indicate that changing the initial codons is not a valid option for enhancing the expression of functional HRP. We therefore took an alternative approach by introducing an N-terminal fusion partner. N-terminal fusion partners are often used to boost the translation efficiency and increase the solubility of recombinant proteins (Ahn et al., 2011). However, we were concerned that the presence of a fusion partner might affect the enzyme activity of HRP, which is sensitive to the amino acid residue at the N-terminus. Previously, we demonstrated that a ubiquitin sequence can be used as an effective N-terminal fusion partner to enhance the translational efficiency and solubility of recombinant proteins that are otherwise difficult to produce in significant yields (Kasi et al., 2017). One of the advantages of using ubiquitin is that it has a cognate protease, ubiquitin hydrolase (UBP1), which removes ubiquitin without leaving additional amino acid residues on the cargo protein (Figure 3A). Therefore, by using a UBP1-enriched extract, a CFPS reaction can produce a target protein with its native N-terminal amino acid sequence from a template DNA fused to the ubiquitin gene. This approach was successfully employed to enhance the cell-free synthesis of HRP. In the presence of the N-terminal ubiquitin sequence, the expression level of HRP was increased more than 2-fold (Figure 3B). Western blotting analysis revealed that most of the synthesized protein was the correct size for HRP, indicating successful co-translational cleavage of ubiquitin by the UBP1 enzyme (Figure 3C). Therefore, in subsequent experiments, the ubiquitin-fused HRP (ubi-HRP) construct was used in combination with UBP1-enriched S12 extract to achieve enhanced production of HRP with its native N-terminal amino acid sequence.

Because the cell-free synthesis reaction applied in this study employed a crude cell extract, it is reasonable to assume that most of the metabolic enzymes of E. coli are retained in the reaction mixture. Taking advantage of this, we attempted to provide the CFPS system with an in situ supply of the heme prosthetic group. In E. coli, cells use the C5 pathway for heme synthesis, and the generation of 5-aminolevulinic acid (5-ALA) is tightly regulated via inhibition of glutamyl-tRNA reductase by heme (Woodard and Dailey, 1995) (Figure 4A). Therefore, to create a sufficient pool of 5-ALA from glucose, we activated the alternative C4 pathway by introducing the ALAS from Rhodobacter capsulatus (R. capsulatus). In the C4 pathway, 5-ALA is generated from succinyl-CoA and glycine. Since the Krebs cycle is operational in E. coli extracts, we predicted that feeding glucose to the cell-free synthesis system would generate succinyl-CoA, which can be condensed with glycine in the reaction mixture to supply 5-ALA for subsequent heme synthesis (Kwon et al., 2013). As expected, cell-free expression of the ubi-HRP gene in the reaction mixture supplemented with 250 µg/ml ALAS, 180 mM glucose, and 2.5 mM glycine resulted in the production of functional HRP (Figure 4B and Supplementary Figure S4).