Semirational engineering of Cytophaga hutchinsonii polyphosphate kinase for developing a cost-effective, robust, and efficient adenosine 5′-triphosphate regeneration system

ABSTRACT

Adenosine 5′-triphosphate (ATP) is a substrate for the biological synthesis of various valuable compounds, such as pharmaceuticals and agrochemicals. To make these biocatalytic processes more economically valuable, several ATP regeneration systems comprising enzymes and phosphate donors were developed. Due to their ability to synthesize ATP from both 5′-diphosphate and adenosine 5′-monophosphate, polyphosphate kinase 2 class III (PPK2-III) enzymes such as Cytophaga hutchinsonii PPK (ChPPK) have attracted considerable attention in the construction of ATP regeneration systems. Furthermore, PPK2-III enzymes employ inorganic polyphosphate, which is a low-cost and stable substrate, as the phosphate donor. To facilitate the application of the PPK2-III ATP regeneration system in biocatalytic processes depending on ATP, a semirational strategy coupled with a high-throughput screening assay based on a fluorescent sensor was performed to engineer ChPPK. Saturation mutagenesis was performed on 16 critical residues of ChPPK. After screening approximately 4,800 colonies, nine single amino acid variants exhibited more than threefold increased activities than wild-type ChPPK. Beneficial mutations were subsequently recombined, resulting in a quadruple variant (ChPPK/A79G/S106C/I108F/L285P), compared with wild-type ChPPK, displaying 18.8-fold enhanced activity and better stability at an elevated temperature and under acidic conditions. The introduction of this variant ChPPK/A79G/S106C/I108F/L285P in the biological synthesis of nicotinamide mononucleotide catalyzed by phosphoribosyl pyrophosphate synthetase and nicotinamide phosphoribosyltransferase increased the yield by approximately threefold compared with wild-type ChPPK. Therefore, the ATP regeneration system described may have implications for bioconversions requiring ATP.

IMPORTANCE

The adenosine 5′-triphosphate (ATP) regeneration system can significantly reduce the cost of many biocatalytic processes. Numerous studies have endeavored to utilize the ATP regeneration system based on Cytophaga hutchinsonii PPK (ChPPK). However, the wild-type ChPPK enzyme possesses limitations such as low enzymatic activity, poor stability, and limited substrate tolerance, impeding its application in catalytic reactions. To enhance the performance of ChPPK, we employed a semi-rational design approach to obtain the variant ChPPK/A79G/S106C/I108F/L285P. The enzymatic kinetic parameters and the catalytic performance in the synthesis of nicotinamide mononucleotide demonstrated that the variant ChPPK/A79G/S106C/I108F/L285P exhibited superior enzymatic properties than the wild-type enzyme. All data indicated that our engineered ATP regeneration system holds inherent potential for implementation in biocatalytic processes.

INTRODUCTION

Adenosine 5′-triphosphate (ATP) is a critical molecule in living organisms. In addition to providing energy for many biological processes in cells (1), other important biochemical functions of ATP include intracellular signaling pathway transduction (2), DNA and RNA syntheses (3, 4), activation of amino acids during peptide synthesis (5), and extracellular signaling (6). ATP is also an important raw material for many biological processes that produce useful compounds. For example, the activities of numerous potential biocatalysts, including ligases, kinases, and synthetases, rely on ATP (7 – 9). However, the high cost of ATP makes many biocatalytic reactions that consume ATP economically unfavorable. Implementation of ATP regeneration systems in these bioconversion processes may significantly reduce catalytic costs (10). In addition, ATP regeneration systems promote the progress of reactions by continuously removing the reaction product of 5′-diphosphate (ADP) or adenosine 5′-monophosphate (AMP) from equilibrium and maintaining constant ATP concentrations. The majority of ATP regeneration systems comprise phosphate donors and enzymes that catalyze transphosphorylation between the ADP and the phosphate donors. Enzymes that can be used to construct such systems include acetate kinase, pyruvate kinase, and creatine kinase, which utilize acetyl phosphate, phosphoenolpyruvate, and phosphocreatine, respectively, as phosphate donors (11). However, ATP regeneration systems starting from AMP are highly desirable for the biosynthesis of certain valuable products, e.g., oxytetracycline, 1,6-hexamethylenediamine, and β-carboline amide (12 – 14). To regenerate ATP from AMP, enzymes that synthesize ADP by transferring a phosphate group to AMP and their corresponding phosphate donors were introduced into ATP regeneration systems starting from ADP (14). It is worth noting that the introduction of multiple enzymes and phosphate donors could make bioconversion processes more complex.

Polyphosphate kinases (PPKs) catalyze the transfer of phosphate groups between nucleotides and inorganic polyphosphate (polyP) (15). PPKs can be classified into two types based on their substrate specificities and enzymatic properties. PPK1 enzymes prefer to synthesize longer polyP chains by utilizing ATP as the phosphate group donor. In contrast, PPK2 enzymes tend to degrade polyP to generate ATP or ADP. PPK2 can be further classified into three types: PPK2-I, PPK2-II, and PPK2-III. While the first two types can convert AMP and ADP into ADP and ATP, respectively, PPK2-III can generate ATP from both ADP and AMP through two phosphorylation steps or pyrophosphorylation (16, 17). Therefore, PPK2-III enzymes can be used to construct cyclic systems that generate ATP from AMP. According to the enzyme kinetics parameters of PPK2-III, the main challenge in industrial catalysis for the PPK2-III-based ATP regeneration system is to achieve a satisfactory ATP generation rate, which requires improving the enzyme’s activity and stability (15).

Comprehensive phylogenetic analysis has identified thousands of PPK2-III enzymes, but only a few, such as PPK2-III enzymes derived from Meiothermus ruber, Cytophaga hutchinsonii, Arthrobacter aurescens, and Deinococcus radiodurans, have been experimentally characterized (15). Among these PPK2-III enzymes, Cytophaga hutchinsonii PPK (ChPPK) exhibited the highest activity for converting ADP to ATP (7, 12, 15, 18). Very recently, another PPK2-III enzyme derived from an unclassified Erysipelotrichaceae bacterium (PPK12) displayed a high ATP formation rate and enhanced stability (19). In the present study, we first compared the activities of ChPPK and PPK12. Semirational engineering of ChPPK was then performed to enhance the performance of the ATP regeneration system. Although a smaller and smarter library was required for the semirational approach to engineering an enzyme (20), developing a high-throughput screening method can significantly accelerate screening (21). Given that ATP is the product of the reaction catalyzed by ChPPK, several ATP detection methods (22 – 24) can theoretically be developed into high-throughput screening assays for determining the activity of ChPPK. For example, the firefly luciferase-catalyzed bioluminescence reaction was widely used for detecting ATP levels in enzymatic processes (25). However, the luciferase’s bioluminescent output was determined by not only the ATP level but also the substrate levels of luciferin and oxygen (26). Very recently, a GFP-based sensor (iATPSnFR) was described for imaging ATP in cells (27). The structure of iATPSnFR was represented by inserting the circularly permuted superfolder green fluorescent protein between the two α-helices of the ε subunit of an F₀F₁- ATP synthase from Bacillus PS3 (Fig. 1). The ε subunit in iATPSnFR is supposed to undergo a large conformational change into a folded form upon binding ATP, leading to the enhancement of the fluorescence value of iATPSnFR. In contrast to firefly luciferase, iATPSnFR responds to ATP without consuming either ATP or exogenous substrates. Therefore, iATPSnFR was employed to develop a high-throughput screening assay for comparing ChPPK variants in the present study (Fig. 1).

FIG 1.

The principle of the high-throughput screen method for comparing the activities of ChPPK variants. The fluorescence values of iATPSnFR responded to the levels of ATP, which were determined by the activities of ChPPK variants.

Due to its antiaging effect (28), nicotinamide mononucleotide (NMN) has a large market potential as a health supplement. The current synthetic methods for NMN include chemical synthesis, fermentation, and enzymatic synthesis (29). The enzymatic synthesis of NMN has advantages such as high product enantiomeric purity and ease of downstream separation and purification. To reduce the cost and increase the yield of enzymatic NMN production, the efficacy of the ATP regeneration system comprising the best ChPPK variant was evaluated in the synthesis of NMN catalyzed by the phosphoribosyl pyrophosphate synthetase variant derived from Bacillus amyloliquefaciens (BaPRS) and nicotinamide phosphoribosyltransferase derived from Haemophilus ducreyi (HdNampt) (30, 31).

RESULTS

The activity of ChPPK toward AMP is higher than that of PPK12

Both polyPs with a defined chain length and mixtures of polyPs with different chain lengths have been used as phosphate donors for PPK2-III enzymes (32 – 34). The price of mixtures of polyPs with different chain lengths is usually lower than that of polyPs with a defined chain length. In this study, polyphosphoric acid (PPA, a mixture of polyPs with 75% of the chain lengths ranging from 2 to 14 phosphate units) was used as the substrate for the PPK2-III enzymes. The efficacies of the PPA-driven transformation of AMP to ATP catalyzed by ChPPK and PPK12 were compared. Figure S1A shows that both ChPPK and PPK12 were efficiently expressed in soluble form by Escherichia coli (E. coli). The purified ChPPK generated higher levels of ATP than the purified PPK12 at all time points tested (Fig. S1B; Fig. 2A). The enzyme activity of purified ChPPK in catalyzing AMP to ATP was 4.8 times higher than that of PPK12 (Fig. 2B). In our results, the highest ATP conversion rates from AMP for ChPPK and PPK12 were only 37% and 26%, respectively. However, some studies have shown that PPK2-III enzymes can convert more than 70% of AMP to ATP (17, 35). This is mainly because only 5 µg/mL of enzymes was used in our study, which is one or two orders of magnitude lower than the enzyme concentrations used in the studies mentioned. The final ATP yield of ChPPK was higher than that of PPK12 (Fig. 2A). In agreement with our data, another report also showed that different PPKs produce significantly different final ATP yields even under the same substrates (19). However, these results cannot be explained by certain PPK enzymes changing the thermodynamic equilibrium; the possible reason for this is that a relatively small amount of enzymes were used, and enzymes can gradually lose activity due to collateral damage from the reaction they catalyze before the product yield reaches the kinetic equilibrium concentration (36). In addition, compared to PPK12, ADP was rapidly generated by ChPPK at early time points (Fig. 2C). The consumption rate of AMP at all tested time points was also faster in ChPPK (Fig. 2D). Based on these data, ChPPK was further investigated in the following study.

FIG 2.

ChPPK exhibited higher ATP regeneration activity from AMP than PPK12 in the presence of PPA. The courses of ATP generation (A), ADP generation (C), and AMP consumption (D) by ChPPK and PPK12 in the presence of PPA. Purified enzymes (5 µg/mL) of ChPPK or PPK12 were incubated with 1.6 g/L PPA and 2.25 mM AMP for the indicated periods. (B) The activities of purified ChPPK and purified PPK12. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute.

Identifying critical residues for the activity of ChPPK by alanine scanning mutagenesis

The crystal structure of ChPPK in complex with substrates, Mg²⁺, and products revealed that conserved residues 75–81 (Walker A loop) coordinate polyP and Mg²⁺, and conserved residues 131–136 (Walker B loop) contribute to the binding of the nucleotide phosphate group. Based on our docking experiment (Fig. 3A), the majority of residues within 5 Å and a subset of residues within 12 Å of polyP₅ and AMP in the catalytic pocket of ChPPK were subjected to alanine scanning mutagenesis. High-performance liquid chromatography (HPLC) analysis of the activities of purified enzymes indicated that most of the single amino acid variants showed similar or lower activities against AMP-producing ATP compared to wild-type ChPPK (Fig. S2; Fig. 3B). Alanine substitutions at D82, R133, E137, and R208 completely inactivated ChPPK. However, alanine substitutions at S106, I108, or S111 significantly increased the activity of ChPPK. Based on the results of alanine scanning mutagenesis, D77, G80, K81, D82, F102, K103, P105, S106, I108, S111, R117, R133, E137, and R208 were critical residues for ChPPK.

FIG 3.

The alanine scanning mutagenesis of critical residues in the ChPPK. (A) The crystal structure of ChPPK in complex with polyP₅ and AMP. The residues selected for alanine scanning mutagenesis, polyP₅ and AMP, were highlighted in the close-up stereo view of the catalytic pocket. (B) The relative activities of purified wild-type ChPPK and its variants. Enzymatic activity assays were performed for the purified enzymes (5 µg/mL) of ChPPK or its variants in the presence of 1.6 g/L PPA and 2.25 mM AMP. The ATP concentration after incubation was measured by HPLC. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute.

Nine ChPPK variants with enhanced activities are isolated using saturation mutagenesis

The 14 critical residues of ChPPK identified by alanine scanning mutagenesis were further subjected to saturation mutagenesis. The conserved residue A79 located in the Walker A loop was also included for saturation mutagenesis, although alanine substitution has not been performed in this residue. Furthermore, the structure of ChPPK contains a C-terminal extension (Fig. 3A). It has been shown that deletion of this C-terminal extension in ChPPK facilitates polyP_12-13-driven phosphorylation of AMP to ATP (37). However, our data indicated that no significant improvement was observed for the truncated ChPPK when PPA was used as the phosphate donor (Fig. S3A), revealing that the effect of the C-terminal extension on the activity of ChPPK was associated with the choice of polyP. To further investigate the role of the C-terminal extension in PPA-driven phosphorylation, residue L285 was also subjected to saturation mutagenesis.

Three hundred colonies from each site for saturation mutagenesis were picked for analysis. To efficiently screen these colonies, a high-throughput assay based on the fluorescent sensor iATPSnFR was developed. The fluorescence values of iATPSnFR were correlated with the concentrations of ATP from 0 to 0.5 mM but did not respond to the equivalent concentrations of AMP (Fig. 4A), indicating that the high-throughput assay based on iATPSnFR was feasible for determining the ATP synthesized by ChPPK. After screening of approximately 4,800 colonies by this high-throughput assay, nine variants with 0.5-fold improvements in relative fluorescence value compared to that of wild-type ChPPK were retained. HPLC analysis of the purified enzymes derived from these nine variants showed that their activities increased more than threefold compared with that of the wild-type ChPPK (Fig. S3B; Fig. 4B). The best one among these nine variants contained the mutation A79G, whose activity was 8.1-fold higher than the wild-type enzyme.

FIG 4.

Identifying ChPPK variants with improved activities by high-throughput screening. (A) The output of the high-throughput assay was correlated with the concentration of ATP from 0 to 0.5 mM. BL21-iATPSnFR cell lysate supernatant (5.6 mg/mL) was treated with a series of concentrations of ATP, ADP, or AMP. (B) The activities of the purified wild-type ChPPK and ChPPK variants. Enzymatic activity assays were performed for purified enzymes (5 µg/mL) of wild-type ChPPK or ChPPK variants in the presence of 1.6 g/L PPA and 2.25 mM AMP. The ATP concentration after incubation was measured by HPLC. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute.

Generating the best variant through the combination of beneficial mutations

The combination of beneficial mutations is likely to further improve the performance of an enzyme. However, some beneficial mutations may exert a negative effect when combined with others (38). We next constructed and screened a combinatorial library of sites A79, S106, I108, and S111, which are located close to the substrate binding pocket of the ChPPK. After screening 1,000 colonies by the high-throughput screen, variant ChPPK/A79G/I108F and variant ChPPK/A79G/S106C/I108F exhibited the highest improvements in relative fluorescence value. The activities of the purified variant ChPPK/A79G/I108F and variant ChPPK/A79G/S106C/I108F increased 13.9- and 14.2-fold compared to wild-type ChPPK, respectively (Fig. S4A; Fig. 5). The introduction of the beneficial mutation S111A in the variant ChPPK/A79G/S106C/I108F significantly decreased the activity. Contrary to the mutation S111A, the introduction of the mutation L285P in the variant ChPPK/A79G/S106C/I108F resulted in the best variant in the present study (variant ChPPK/A79G/S106C/I108F/L285P), whose activity was 1.3-fold higher than that of the variant ChPPK/A79G/S106C/I108F. Importantly, the expression level of variant ChPPK/A79G/S106C/I108F was similar to that of wild-type ChPPK in E. coli (Fig. S4B).

FIG 5.

Identification of the best variant ChPPK/A79G/S106C/I108F. The relative activities of the purified wild-type ChPPK and ChPPK variants containing multiple beneficial mutations. Enzymatic activity assays were performed for purified enzymes (5 µg/mL) of wild-type ChPPK or ChPPK variants in the presence of 1.6 g/L PPA and 2.25 mM AMP. The ATP concentration after incubation was measured by HPLC. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute.

Characterization of the variant ChPPK/A79G/S106C/I108F/L285P

The initial reaction rates of wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P at different temperatures, pH values, and substrate concentrations were investigated using purified enzymes (Fig. 6). Although 37°C was the optimal reaction temperature for both the wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P, the variant ChPPK/A79G/S106C/I108F/L285P exhibited significantly higher relative initial reaction rates at temperatures between 42 and 50°C (Fig. 6A). For example, the variant ChPPK/A79G/S106C/I108F/L285P retained 80% of its maximum initial reaction rate at 45°C, while wild-type ChPPK lost 98% of its maximum initial reaction rate under this condition. From 25 to 37°C, the initial reaction rate of the wild-type ChPPK increased following a typical exponential pattern. However, the initial reaction rate of the ChPPK/A79G/S106C/I108F/L285P displayed a linear increase within this range. This result indicated that the activity of ChPPK/A79G/S106C/I108F/L285P from 25 to 37°C was a result of a combination of increased reaction rate and partial denaturation or alteration of its active site. The initial reaction rates of the wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P reached their maximum value at pH 7.5 (Fig. 6B). However, the variant ChPPK/A79G/S106C/I108F/L285P had higher relative initial reaction rates under acidic conditions. Both the wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P reached the highest relative initial reaction rates at 2.0–2.5 mM AMP (Fig. 6C). The optimum PPA concentration for wild-type ChPPK was 1.6 g/L, whereas the variant ChPPK/A79G/S106C/I108F/L285P exhibited the highest relative initial reaction rates in the presence of 2.24 g/L PPA, revealing that the variant ChPPK/A79G/S106C/I108F/L285P had superior PPA tolerance (Fig. 6D).

FIG 6.

The effects of temperature (A), pH (B), AMP concentration (C), and PPA concentration (D) on the initial reaction rates of purified wild-type ChPPK and the variant ChPPK/A79G/S106C/I108F/L285P. Four different buffers, 50 mM citric sodium citrate buffer (pH 5–6), potassium phosphate buffer (pH 6–8), borate-boric acid buffer (pH 8–9), and glycine-sodium hydroxide buffer (pH 9–10), were used in B. Reactions were performed using a purified enzyme (5 µg/mL) of wild-type ChPPK or ChPPK variant. In (A), (B), (C), and (D), the concentrations of 2.25 mM AMP and 1.6 g/L PPA were used in the reaction, unless as a variable substrate concentration.

Given that the variant ChPPK/A79G/S106C/I108F/L285P exhibited higher relative initial reaction rates at 45°C and under acidic conditions, the next step was to compare its stability with that of wild-type ChPPK under these conditions. After 4 h of incubation at 45°C, the variant ChPPK/A79G/S106C/I108F/L285P retained 41% relative activity, while wild-type ChPPK was completely inactivated (Fig. 7A). The variant ChPPK/A79G/S106C/I108F/L285P also exhibited higher activities than wild-type ChPPK after 6 h of incubation at pHs 6.0 and 6.5 (Fig. 7B).

FIG 7.

The variant ChPPK/A79G/S106C/I108F/L285P exhibited improved thermal (A) and pH stability (B) compared to the wild-type ChPPK. For determining thermal stability, 5 µg/mL purified enzymes were incubated at 45°C for various periods. After incubation, an enzymatic activity assay was performed at 37°C. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute. The activity of unheated enzymes was taken as 100%. For determining pH stability, 50 µg/mL purified enzymes were preincubated in 50 mM potassium phosphate buffer (pH 6.0–7.0) at 4°C for 6 h. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute. The activity of enzymes preincubated in 50 mM potassium phosphate buffer (pH 7.5) was taken as 100%.

The kinetic parameters of wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P were determined (Fig. S5; Table 1). When AMP was used as a variable substrate, the Km values for wild-type and mutated enzymes catalyzing AMP to ADP were similar, but the catalytic efficiency of the mutated enzyme increased by 77%. When ADP served as a substrate for the synthesis of ATP, the Kprime value of the mutated enzyme was approximately half that of the wild-type enzyme. The Hill slope values for the wild-type and mutated enzymes in relation to ADP indicated that when one ADP molecule was bound to either the dimeric wild-type enzyme or the dimeric mutant enzyme, it enhanced their affinity for the second ADP molecule (Fig. S5B). Recent literature results show that PPK can catalyze the direct conversion of AMP to ATP by pyrophosphorylation (17). Furthermore, our results showed that high concentrations of AMP inhibited the activity of ChPPK (Fig. 6C). We therefore used the substrate inhibition equation to calculate the kinetics of this reaction, recognizing that some AMP may regenerate ATP through ADP. Regardless of whether AMP or PPA was used as the variable substrate, the variant ChPPK/A79G/S106C/I108F/L285P exhibited k _cat values more than eightfold higher than those of wild-type ChPPK when catalyzing AMP to ATP. Both the wild-type ChPPK and the variant ChPPK/A79G/S106C/I108F/L285P exhibited a rate of degradation of ATP to ADP or AMP that was one order of magnitude slower than the rate of ATP synthesis from these substrates. The variant ChPPK/A79G/S106C/I108F/L285P exhibited an approximately twofold increase in the efficacy of degrading ADP to AMP compared with the wild-type enzyme. Considering that the most significant change in k _cat values for the interconversion of ATP, ADP, and AMP occurs in the reaction of AMP to ATP, it is likely that the mechanism by which the mutant variant enhances ATP regeneration is by increasing the efficiency of the pyrophosphorylation of AMP to ATP.

TABLE 1.

Kinetic parameters of wild-type ChPPK and variant ChPPK/A79G/S106C/I108F/L285P

Variable substrates	Constant substrates	Measured products	PPKs	Km or Kprime (mM or g/L)	k cat (s−1)	Ki (mM or g/L)	k cat/Km or k cat/Kprime (S−1 M−1)
AMP	PPA	ADP	Wild-type ChPPK a	1.63 ± 0.35 d , f	9.19 ± 1.07	10.3 ± 2.6 d	5.64
			Variant ChPPK/A79G/S106C/I108F/L285P a	1.45 ± 0.26 d , f	16.3 ± 1.5	13.3 ± 3.0 d	11.2
ADP	PPA	ATP	Wild-type ChPPK b	0.509 ± 0.071 d , g	4.40 ± 0.09	NA h	8.64
			Variant ChPPK/A79G/S106C/I108F/L285P b	0.259 ± 0.050 d , g	3.10 ± 0.05	NA h	12.0
AMP	PPA	ATP	Wild-type ChPPK a	2.26 ± 0.63 d , f	4.78 ± 0.94	2.40 ± 0.67 d	2.12
			Variant ChPPK/A79G/S106C/I108F/L285P a	1.02 ± 0.13 d , f	41.7 ± 2.6	9.36 ± 1.36 d	40.9
PPA	AMP	ATP	Wild-type ChPPK a	1.96 ± 0.85 e , f	4.35 ± 1.27	2.34 ± 0.98 e	NA h
			Variant ChPPK/A79G/S106C/I108F/L285P a	0.800 ± 0.121 e , f	38.6 ± 2.2	22.1 ± 4.7 e	NA h
ATP	PPA	ADP	Wild-type ChPPK c	3.65 ± 0.67 d , f	0.720 ± 0.529	NA h	0.197
			Variant ChPPK/A79G/S106C/I108F/L285P c	1.36 ± 0.19 d , f	0.460 ± 0.016	NA h	0.388
ATP	PPA	AMP	Wild-type ChPPK c	4.40 ± 0.62 d , f	0.130 ± 0.008	NA h	0.0295
			Variant ChPPK/A79G/S106C/I108F/L285P c	0.440 ± 0.084 d , f	0.0557 ± 0.0112	NA h	0.136
ADP	PPA	AMP	Wild-type ChPPK b	1.75 ± 0.23 d , g	1.09 ± 0.03	NA h	0.623
			Variant ChPPK/A79G/S106C/I108F/L285P b	2.75 ± 0.34 d , g	2.07 ± 0.04	NA h	0.753

^a The data were fitted to the substrate inhibition equation. The equation is given by: V = Vmax × [S] / (Km + [S] × (1 + [S] / Ki)), where V is the reaction rate, Vmax indicates the maximum enzyme activity, [S] is the substrate concentration, Km indicates the Michaelis constant, and Ki indicates the dissociation constant.

^b The data were fitted to the allosteric sigmoidal equation. The equation is given by: V = Vmax × [S]^h / (Kprime + [S]^h), where V is the reaction rate, Vmax indicates the maximum enzyme activity, [S] is the substrate concentration, h indicates the Hill slope, and Kprime is related to the Michaelis constant.

^c The data were fitted to the Michaelis–Menten equation. The equation is given by: V = Vmax × [S] / (Km + [S]), where V is the reaction rate, Vmax indicates the maximum reaction rate when the enzyme is fully saturated with substrate, [S] is the substrate concentration, and Km indicates the Michaelis constant.

^d mM.

^e g/L.

^f Km.

^g Kprime.

^h Not available.

Implementation of the variant ChPPK/A79G/S106C/I108F/L285P in the biological synthesis of NMN

To evaluate the efficiency of the variant ChPPK/A79G/S106C/I108F/L285P in bioconversions, it was applied for regenerating ATP in the synthesis of NMN (Fig. 8A). In the absence of the ATP regeneration system, the NMN yield was 0.11 mM when 0.2 mM ATP was used (Fig. 8B). Increasing the ATP concentration to 20 mM only slightly enhanced the NMN yield to 0.40 mM, with over 95% of the ATP remaining unconverted after 3.5 h (Fig. 8C). It has been reported that PRS can be potently inhibited by its byproduct AMP (39); therefore, increasing ATP alone cannot efficiently promote the yield of NMN. The introduction of the ATP regeneration system comprising PPA and 8 mg/mL wild-type ChPPK cell lysate (corresponding to an enzyme activity concentration of 19.2 U/mL) in the reaction containing 0.2 mM ATP significantly enhanced the NMN yield to 6.6 mM, indicating that implementing the ATP regeneration system in NMN synthesis not only decreased ATP consumption but also avoided the inhibitory effect of the byproduct AMP on BaPRS. However, AMP still accumulated with increasing time (Fig. 8D). When the ATP regeneration system comprising PPA and the 8 mg/mL variant ChPPK/A79G/S106C/I108F/L285P cell lysate (corresponding to an enzyme activity concentration of 360.8 U/mL) was used in the reaction containing 0.2 mM ATP, the NMN yield reached 22.3 mM. Thus, ATP regeneration efficiency was critical for the synthesis of NMN.

FIG 8.

The effect of introducing an ATP regeneration system on NMN yield. (A) The scheme of the synthesis route for NMN from ribose-5-phosphate (R5P) via coupling BaPRS and HdNAMPT. The AMP was a byproduct of R5P phosphorylation. The ATP regeneration system was applied for the recycling of AMP back into ATP. The NMN generation (B), ATP consumption (C), and AMP generation (D) in the absence and in the presence of an ATP regeneration system containing wild-type ChPPK or the variant ChPPK/A79G/S106C/I108F/L285P at indicated time points. In B, C, and D, the reactions without an ATP regeneration system (the black and blue lines) consisted of 0.2 or 20.0 mM ATP, 30 mM R5P, 50 mM nicotinamide (NAM), 8 mg/mL BL21-BaPRS cell lysate supernatant, and 12 mg/mL BL21-HdNampt cell lysate supernatant. The reactions with an ATP regeneration system (the red and green lines) consisted of 0.2 mM ATP, 30 mM R5P, 50 mM NAM, 4.8 g/L PPA, 8 mg/mL BL21-BaPRS cell lysate supernatant, 12 mg/mL BL21-HdNampt cell lysate supernatant, and 8 mg/mL cell lysate supernatant of wild-type BL21-ChPPK or BL21-ChPPK/A79G/S106C/I108F/L285P.

DISCUSSION

The best variant, ChPPK/A79G/S106C/I108F/L285P, exhibited improved activity and thermostability compared with wild-type ChPPK. To determine possible reasons for the improvements, molecular modeling and docking experiments were performed. Cross-sectional analysis of ChPPK demonstrated that an internal substrate channel connecting AMP and polyP binding sites was located near catalytic D77 (37). As shown in Fig. 9A, Ala79 in wild-type ChPPK lay adjacent to the internal substrate channel, and the pink zone represented where one of the phosphate oxygens of polyP was in contact with the side chain of A79. The ChPPK A79G mutation led to the loss of a methyl group in the side chain of residue 79 (Fig. 9B), which enlarged the internal substrate channel at the pink zone. Figure 10A shows that the crystal structure of ChPPK contained a large, shallow cavity below a lid comprising three α-helices (α13, α14, and α15). Two conserved sequences, motifs Walker A (residues 75–81) and Walker B (residues 132–134), are located in this shallow cavity. S106 and I108 are two residues in the α6 helix, which are located adjacent to the AMP binding site in the shallow cavity (Fig. 10A and B). The S106C mutation replaced the hydroxyl group with the sulfhydryl group in the side chain (Fig. 10B). As the angle of the sulfhydryl hydrogen of the side chain in the variant was opposite to the hydroxyl hydrogen of that in wild-type ChPPK (Fig. 10C and D), the original hydrogen bond between the hydroxyl hydrogen in residue 106 and nitrogen in residue 109 was disrupted. The reduced number of hydrogen bonds in the α6 helix may release its rigidity and improve its flexibility, which may facilitate AMP and ChPPK docking. Residue 108 was located on the outside of the α6 helix on the surface of the protein (Fig. 10B). The replacement of a hydrophobic amino acid with an aromatic amino acid by the I108F mutation may improve the hydrophilization of the surface group of the protein. Furthermore, the mutation changed the side chain of residue 108 from a straight chain to a stable benzene ring (Fig. 10C). Therefore, the I108F mutation may influence the stability of ChPPK. It has been reported that a C-terminal extension of ChPPK comprising residues 285–305 may interact with the lid of another monomer in ChPPK dimers and hinder the ability of substrates to enter the active site of that monomer (37). While the side chain of L285 in the wild-type ChPPK was horizontal outward and very flexible, the L285P mutation imposed rigid constraints for this region and may consequently reduce the interaction between the C-terminal extension and the lid of another monomer (Fig. 10E). Therefore, L285P may improve the activity of ChPPK by facilitating substrate entry into the active site.

FIG 9.

Conformational changes of the substrate channel surface with and without the A79G mutation. (A) A cross-sectional view of the substrate channel in the wild-type ChPPK. (B) A cross-sectional view of the mutant substrate channel in the ChPPK with the A79G mutation. The blue zone represented the location of residue 79. The pink zone represented the interaction between one of the phosphate oxygens of polyP₅ and the side chain of residue 79.

FIG 10.

The effects of S106C, I108F, and L285P mutations on the structure of ChPPK. (A) The overall structure of ChPPK. The lid domain, Walker A motif, and Walker B motif were indicated with grayish blue, turquoise, and pink, respectively. (B) Schematic diagram of the α6 helix with or without S106C and I108F mutations. The hydrogen bonding network in the α6 helix of the wild-type ChPPK (C) and ChPPK with S106C and I108F mutations (D). (E) A schematic diagram of the α18 helix with or without S106C and I108F mutations. The wild-type residues were indicated in yellow, and the mutated residues were indicated in blue in B, C, D, and E.

A cost-effective, robust, and efficient ATP regeneration system is highly desirable for biocatalytic processes. The availability of phosphorus donors largely determines the cost of ATP regeneration systems. The phosphorus donor PPA used in this study can be easily prepared by heating a mixture of phosphoric acid and phosphorus pentoxide to 200°C, followed by cooling (40). Biocatalysts sometimes require extended reaction times, may be conducted at elevated temperatures, and involve the addition of small amounts of organic solvents to enhance substrate solubility. All of these factors place high demands on enzyme stability. However, Tavanti et al. showed that a lyophilized cell extract of E. coli expressing wild-type ChPPK displays poor stability, as it almost lost the ability to regenerate ATP after 2 h of preincubation at 30°C (19). In our experiment, the purified ChPPK enzyme exhibited a significant decrease in activity after 2 h of incubation at 45°C but still retained 39.5% residual activity. This inconsistency may be attributed to the choice of phosphate donor or the potential impact of the lyophilization process on the enzyme structure. Importantly, the variant ChPPK/A79G/S106C/I108F/L285P, which exhibits higher stability toward temperature and acidic pH conditions, can be utilized to construct a more robust ATP regeneration system, allowing for increased flexibility in catalytic process design. Due to the use of different polyPs with varying chain lengths as phosphate donors in different studies, directly comparing the efficiencies of different PPK2-III enzymes using reported enzyme kinetic parameters may not be appropriate. Despite this, it can still be concluded that most PPK2-III enzymes exhibit higher efficiency in catalyzing the conversion of AMP to ADP (with k _cat values ranging from 7.1 to 94.5 s⁻¹), while the efficiency of converting ADP to ATP is lower (with k _cat values ranging from 0.3 to 9.4 s⁻¹) (15). If the substrate inhibition equation was used to fit the catalysis of AMP to ATP by variant ChPPK/A79G/S106C/I108F/L285P in the presence of PPA, its k _cat value (41.7 s⁻¹) was apparently higher than that of any reported PPK2-III enzymes converting ADP to ATP. Therefore, if any of the PPK-III enzymes reported thus far use ADP as an intermediate product to convert AMP into ATP, their efficiency should be lower than that of the variant ChPPK/A79G/S106C/I108F/L285P. Of course, it should also be noted that it is currently unclear to what extent other PPK2-III enzymes utilize ADP as an intermediate product to synthesize ATP.

As a sensitive ATP sensor, iATPSnFR was mainly applied for imaging ATP in the intracellular space (41 – 43). Our data demonstrated that iATPSnFR can be used to determine the activity of kinases, including PPK. The classic method for measuring kinase activity, which is based on luciferase, requires enzymes, substrates, and oxygen, resulting in a complex system (44). In contrast, iATPSnFR-based ATP measurement only requires a single protein, which simplifies the detection process and reduces measurement costs. The implementation of iATPSnFR for developing kinase activity measurements also has limitations, with the main issue being its saturation concentration for ATP at 0.5 mM. While this is sufficient for detecting physiological concentrations of ATP in vivo, it may be too low for industrial catalytic processes catalyzed by kinase. For our high-throughput screening assay, designed to detect the activity of the PPK variant with multiple beneficial mutations, it was necessary to dilute the reaction mixture before measuring ATP levels using iATPSnFR.

Given its potential antiaging benefits, NMN and its synthesis have gained considerable attention in recent years (28). In agreement with our data, Ngivprom et al. showed that the yield of NMN through the enzymatic route can be limited by insufficient ATP regeneration from wild-type ChPPK (18). Figure 8C shows that the variant ChPPK/A79G/S106C/I108F/L285P consistently provided ATP throughout the entire catalytic process. It is worth noting that even with the introduction of the variant ChPPK/A79G/S106C/I108F/L285P, the conversion rate of the substrate R5P to NMN is only 74%. The cell lysate, which contains both recombinant proteins and host proteins, was used to catalyze the production of NMN in our study. NMN is an important intermediate in cellular metabolism. Within the host, various enzymes are present to facilitate the conversion of NMN into other substances, such as NMN amidohydrolase and NMN adenylyltransferase (45). Therefore, in our reaction, the NMN synthesized by recombinant proteins may be consumed by host proteins, thereby decreasing the conversion rate of R5P. Huang et al. also observed the influence of host proteins in their experiments on whole-cell synthesis of NMN (31). They increased the NMN yield by 5.3 times through the deletion of multiple genes involved in NMN metabolism.

MATERIALS AND METHODS

Strains, plasmids, enzymes, and reagents

The E. coli BL21 (DE3) was used as an expression host (Thermo Fisher Scientific, Carlsbad, USA). The pET-28a vector was purchased from Novagen (Darmstadt, Germany). The PPA (Cat No. P102919-500g), R5P, nicotinamide (NAM), and NMN were purchased from Aladdin (Shanghai, China). All other chemicals used were analytical reagent grade. The sequences of ChPPK, PPK12, BaPRS harboring an L135I mutation, HdNampt, and iATPSnFR were synthesized by Beijing Qingke Biocompany (Beijing, China). A 6His tag was fused at the C-terminal of ChPPK and PPK12.

Construction of E. coli expression vectors

The linearized pET-28a vector was obtained from pET-28a by reverse PCR using primers pET-28a-F and pET-28a-R (all primers used for the construction of expression plasmids are listed in Table 2). The sequences of ChPPK, PPK12, BaPRS, HdNampt, and iATPSnFR were amplified from the synthesized product using corresponding primer pairs and inserted into a linearized pET-28a vector to give pET-ChPPK, pET-PPK12, pET-BaPRS, pET-HdNampt, and pET-iATPSnFR, respectively, by recombination using the ClonExpress II One Step Cloning Kit according to the manufacturer. The constructed plasmids were transformed into E. coli BL21(DE3) by the calcium phosphate transformation method to give strains BL21-ChPPK, BL21-PPK12, BL21-BaPRS, BL21-HdNampt, and BL21-iATPSnFR.

TABLE 2.

Primers used for the construction of plasmids

Primer	Oligonucleotide sequences (5′–3′)	Amplified fragment
pET-28a-F	CTCGAGCACCACCACCACCACCACTGA	Linearized pET-28a
pET-28a-R	GCCCATGGTATATCTCCTTCTTAAAGT
ChPPK-F	TACCATGGGCATGGCAACCGATTTT	ChPPK gene
ChPPK-R	TGGTGCTCGAGTTAGTGGTGATGATG
PPK12-F	TACCATGGGCATGATCAACATTTACAAG	PPK12 gene
PPK12-R	TGGTGCTCGAGTCAGTGGTGGTGGTG
BaPRS-F	TACCATGGGCATGTCTAACGAATA	BaPRS
BaPRS-R	TGGTGCTCGAGTTAGCTGAATAAATA
HdNampt-F	TACCATGGGCATGATGGATAACCTGCTG	HdNampt gene
HdNampt-R	TGGTGCTCGAGTTAATGGTGATGAT
iATPSnFR-F	TACCATGGGCATGAAAACCATCCAT	iATPSnFR gene
iATPSnFR-R	TGGTGCTCGAGTTATTTCATTTCTGC

Molecular modeling and docking experiment

The 3D structure of ChPPK was taken from the Protein Data Bank (PDB code 6ANG). As the PPA is a mixture of linear phosphoric acids with different chain lengths, polyP₅ was used to represent the PPA in the docking experiment. The ligands AMP and polyP₅ were docked into the binding pocket of the ChPPK with default parameters implemented in AutoDock vina 1.1.2. PyMOL was used for visualization and analysis.

Alanine scanning mutagenesis, site-directed mutagenesis, and site-directed saturation mutagenesis of the ChPPK

The alanine scanning mutagenesis, site-directed mutagenesis, and site-directed saturation mutagenesis were performed by the Quick-Change site-directed mutagenesis kit (Agilent Technologies, Santa Clara, USA) using specific primers (Table 3) or degenerate primers (Table 4). The CASTER 2.0 tool (46) was used to design degenerate primers. The PCR reaction mixture contained 25 µL 2× Phanta Max Buffer, 0.4 µM of each primer, approximately 10 ng of DNA template, and water to 50 µL. The amplification reaction cycles were as follows: an initial denaturation step at 95°C for 1 min, followed by 20 cycles of amplification (denaturation at 95°C for 10 s, annealing at 55°C for 30 s, and extension at 72°C for 5 min), and a final extension at 72°C for 5 min. The PCR reaction mixture was treated with Dpn I to digest the remaining template and then transformed into E. coli BL21 (DE3). Positive colonies were selected on LB agar plates containing 50 µg/mL kanamycin. For alanine scanning mutagenesis and site-directed mutagenesis, the genes of the variants were Sanger-sequenced to verify that only the desired mutations were introduced.

TABLE 3.

Specific primers used for introducing site-directed mutations in ChPPK

Primer	Oligonucleotide sequences (5′−3′)	Mutation
D77A-F	TCAGGCAATGGCTGCAGCAGGTAAAGAT	D77A
D77A-R	ACCTGCTGCAGCCATTGCCTGAAAAACA
G80A-F	ATGCAGCAGCTAAAGATGGTACCGTT	G80A
G80A-R	ACCATCTTTAGCTGCTGCATCCATTG
K81A-F	AGCAGGTGCAGATGGTACCGTTAAACATAT	K81A
K81A-R	TACCATCTGCACCTGCTGCATCCATTG
D82A-F	AGGTAAAGCTGGTACCGTTAAACAT	D82A
D82A-R	ACGGTACCAGCTTTACCTGCTGCA
F102A-F	TGACCAGCGCTAAAGTTCCGTCCAAAAT	F102A
F102A-R	ACGGAACTTTAGCGCTGGTCACTTTAACA
K103A-F	ACCAGCTTTGCAGTTCCGTCCAAAATT	K103A
K103A-R	ACGGAACTGCAAAGCTGGTCACTTT
V104A-F	AGCTTTAAAGCTCCGTCCAAAATTGAAC	V104A
V104A-R	TTGGACGGAGCTTTAAAGCTGGTCA
P105A-F	TTTAAAGTTGCGTCCAAAATTGAACTGAG	P105A
P105A-R	ATTTTGGACGCAACTTTAAAGCTGG
S106A-F	AAGTTCCGGCCAAAATTGAACTGAGT	S106A
S106A-R	TCAATTTTGGCCGGAACTTTAAAGCTGG
I108A-F	TCCGTCCAAAGCTGAACTGAGTCATGATTA	I108A
I108A-R	ACTCAGTTCAGCTTTGGACGGAACTT
S111A-F	TTGAACTGGCTCATGATTATCTGTGGC	S111A
S111A-R	TAATCATGAGCCAGTTCAATTTTGGACG
R117A-F	TATCTGTGGGCTCATTATGTGGCAC	R117A
R117A-R	ACATAATGAGCCCACAGATAATCATGA
R133A-F	TTTTTAACGCTAGCCATTATGAAAATGTGC	R133A
R133A-R	AATGGCTAGCGTTAAAAATACCAATTTCG
E137A-F	AGCCATTATGCAAATGTGCTGGTTAC	E137A
E137A-R	AGCACATTTGCATAATGGCTACGGTTAAAAA
N138A-F	ATTATGAAGCTGTGCTGGTTACCCGT	N138A
N138A-R	ACCAGCACAGCTTCATAATGGCTAC
V141A-F	TGTGCTGGCTACCCGTGTACATCC	V141A
V141A-R	TACACGGGTAGCCAGCACATTTTCATA
R208A-F	TTATTGAAGCTATCGAACTGGATACCAA	R208A
R208A-R	AGTTCGATAGCTTCAATAAAACGCTT
R285STOP-F	ACCGTGAGCTAGGAACAGAAAGCGG	R285STOP
R285STOP-R	TTCTGTTCCTAGCTCACGGTCGGAAA

TABLE 4.

Degenerate primers used for site-directed saturation mutagenesis of the ChPPK

Primer	Oligonucleotide sequences (5′−3′) a , b
A79-S-F	TGGATGCANNKGGTAAAGATGGTAC
A79-S-R	TCTTTACCMNNTGCATCCATTGCC
G80-S-F	ATGCAGCANNKAAAGATGGTACCGTT
G80-S-R	ACCATCTTTMNNTGCTGCATCCATTG
K81-S-F	AGCAGGTNNKGATGGTACCGTTAAACATAT
K81-S-R	TACCATCMNNACCTGCTGCATCCATTG
F102-S-F	TGACCAGCNNKAAAGTTCCGTCCAAAAT
F102-S-R	ACGGAACTTT MNNGCTGGTCACTTTAACA
K103-S-F	ACCAGCTTTNNKGTTCCGTCCAAAATT
K103-S-R	ACGGAACMNNAAAGCTGGTCACTTT
P105-S-F	TTTAAAGTTNNKTCCAAAATTGAACTGAG
P105-S-R	ATTTTGGAMNNAACTTTAAAGCTGG
S106-S-F	AGTTCCGNNKAAAATTGAACTGAGT
S106-S-R	TCAATTTTMNNCGGAACTTTAAAGCTGG
I108-S-F	CGTCCAAANNKGAACTGAGTCATGA
I108-S-R	TCAGTTCMNNTTTGGACGGAACTTTAAAG
S111-S-F	TTGAACTGNNKCATGATTATCTGTGGC
S111-S-R	TAATCATGMNNCAGTTCAATTTTGGACG
R117-S-F	TATCTGTGGNNKCATTATGTGGCAC
R117-S-R	ACATAATGMNNCCACAGATAATCATGA
L285-S-F	ACCGTGAGCNNKGAACAGAAAGCGG
L285-S-R	TTCTGTTCMNNGCTCACGGTCGGAAA

^a Bold-marked bases in the primers indicate positions that contain the degenerate codon.

^b N = A, T, G, C (equimolar amounts); K = G, T (equimolar amounts); M = A, C (equimolar amounts).

Constructing the library comprising combinations of beneficial mutations

Beneficial mutations at positions A79, S106, I108, and S111 of the ChPPK gene were selected for constructing the combinatorial library. A fragment corresponding to position 219–346 within the ChPPK gene was amplified using primers containing degenerate codons located at positions 235–237, 316–318, 322–324, and 331–333 within the ChPPK gene (Fig. S6; Table 5). Primers PPK-219–346-F, PPK-219–346-R1, and PPK-219–346-R2 were mixed at a ratio of 2:1:1 in the reaction mixture of PCR. The linearized pET-ChPPK vector was obtained from pET-ChPPK by reverse PCR using primers pET-ChPPK-F and RP-ChPPK-R. The ChPPK gene fragment was inserted into the linearized pET-ChPPK vector to give pET-ChPPK-4mut. The pET-ChPPK-4mut was transformed into E. coli BL21 (DE3), and positive colonies were picked on LB agar plates containing 50 µg/mL kanamycin.

TABLE 5.

Degenerate primers used for constructing the library comprising combinations of beneficial mutations

Primer	Oligonucleotide sequences (5′−3′) a , b
pET-ChPPK-F	CATGATTATCTGTGGCGTCATTATGTG
pET-ChPPK-R	TGCATCCATTGCCTGAAAAACAATCAG
PPK-219–346-F	TCAGGCAATGGATGCAG SAGGTAAAGATGGTA
PPK-219–346-R1	ACAGATAATCATGCT YCAGTTCAWWTTTASACGGAAC
PPK-219–346-R2	ACAGATAATCATGTG MCAGTTCAWWTTTASACGGAAC

^a Bold-marked bases in the primers indicate positions that contain the degenerate codon.

^b S = G, C (equimolar amounts); Y = C, T (equimolar amounts); W = A, T; M = A, C (equimolar amounts).

Expression of recombinant protein using a shake flask

E. coli strains were cultured in LB medium with 50 µg/mL kanamycin at 37°C. When the optical density at 600 nm (OD600) reached 0.7, cells were induced with 0.1 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG) at 28°C. After 12 h of induction, 0.2 g potassium phosphate saline buffer (PBS, containing 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄ at pH 7.2) washed cells were resuspended in 5 mL cold PBS and disrupted on ice using the Xinzhi ultrasonic homogenizers (Model JY92-IIN, Ningbo, China) for 20 min with a sequence of 1 s pulse on and 2 s pulse off (40% amplitude, 20–25 kHz, and 50 W). The supernatant and precipitate were separated by centrifugation at 13,800 × g at 4°C for 20 min. The precipitate was then dissolved in a 5 mL SDS-PAGE loading buffer. The concentration of cell lysate supernatant obtained in this manner was defined as 40 mg/mL, corresponding to the wet cell weight.

Purification of the ChPPK and its variant

Five milliliters of the aforementioned cell lysate supernatant (40 mg/mL) was diluted into 40 mL PBS and then loaded onto a 10 mL Ni²⁺ affinity column (HisTrap HP, GE Healthcare, Waukesha, USA) preequilibrated with binding buffer (20 mM potassium phosphate, 500 mM sodium chloride, pH 7.2). The column was then washed with 100 mL wash buffer (20 mM potassium phosphate, 500 mM sodium chloride, 50 mM imidazole, pH 7.2) to eliminate non-specifically bound proteins. The ChPPK or its variant was eluted with elution buffer (20 mM potassium phosphate, 500 mM sodium chloride, 250 mM imidazole, pH 7.2). The elution fractions were diafiltrated with 20 mM potassium phosphate (pH 7.2) and analyzed by SDS-PAGE. The concentration of enzyme in the retained solution was measured using a BCA protein kit (Biyuntian, Beijing, China).

Culture conditions for 96 deep-well plates

Variants were cultured in 96 deep-well plates containing 1 mL of LB medium with 50 µg/mL kanamycin. The E. coli containing the pET-ChPPK and pET-28a was used as the positive control and negative control, respectively. The plates were incubated at 37°C for 10 h with shaking. Two hundred microliters of broth from each well was transferred to new 96-well microplates and preserved at 4°C. The gene expression in the original plates was induced at 28°C by adding 200 µL of LB medium containing 0.5 mM IPTG. After 16 h of induction, cells were harvested by centrifugation (1,500 × g for 15 min at 4°C).

High-throughput screening assay for detecting the ATP level

After IPTG induction for 16 h, all cells harvested in each well of 96 deep-well plates were resuspended with 100 µL of 0.1% Triton X-100 at 37°C for 20 min with shaking (180 rpm). All cells in each well were then harvested by centrifugation (1,500 × g for 15 min) and resuspended with 200 µL potassium phosphate buffer (50 mM, pH 7.5) containing 3.2 g/L PPA, 2.25 mM AMP, and 10 mM MgCl₂. The 96 deep-well plates were incubated at 37°C for 30 min with shaking (180 rpm) and subsequently centrifuged at 1,500 × g for 10 min. Thirty microliters of supernatants (for screening variants with single point mutations) or 30 µL of the supernatant diluted 10-fold (for screening variants with multiple mutations) in each well was transferred into 96-well black plates and mixed with 70 µL of BL21-iATPSnFR cell lysate supernatant (8 mg/mL). Fluorescence values were detected immediately using an excitation wavelength of 485 nm and an emission wavelength of 515 nm in a spectrometer. The relative fluorescence value for the wild-type ChPPK or a variant was calculated by subtracting the fluorescence value of E. coli BL21(DE3) cells from its fluorescence value.

Enzymatic activity assay of cell lysate supernatants and purified PPK2-III enzymes

The cell lysate supernatant (0.4 mg, diluted from the aforementioned mother liquor of cell lysate supernatant) or 0.05 mg purified enzyme, 1.6 g/L PPA, 2.25 mM AMP, and 10 mM MgCl₂ were mixed in 10 mL of potassium phosphate buffer (50 mM, pH 7.5). The reaction mixture was incubated at 37°C for 1 min and quenched by adding 10 mL of 0.2 M H₃PO₄. The levels of ATP, ADP, and AMP in the quenched reaction mixture were analyzed by HPLC using an XBridge C18 column (C18, 5 µm, 4.6 × 250 mm; Waters, California, USA). The HPLC was carried out at a follow rate of 1 mL/min with UV detection at 254 nm. The mobile phase was potassium phosphate buffer (50 mM, pH 7.5), and the injection volume was 20 µL. One unit of enzyme activity was given as the amount of enzyme that synthesized 1 µM ATP from AMP per minute during the initial minute.

Determining kinetic parameters of the wild-type and mutated ChPPK

To determine the kinetic parameters of wild-type and mutated ChPPK toward AMP, ADP, or ATP, initial rate measurements were conducted. The reaction mixture consisted of increasing concentrations of AMP (0.02–15.00 mM), ADP (0.12–10.00 mM), or ATP (0.25–11.0 mM), along with 1.6 g/L PPA in a potassium phosphate buffer (50 mM, pH 7.5), resulting in a total reaction volume of 0.5 mL. The kinetic parameters of wild-type and mutated ChPPK toward the PPA were determined similarly, except increasing concentrations of PPA (0.1–15.0 g/L) and 2.25 mM AMP were used. The reaction mixture was incubated at 37°C for 1 min and quenched by adding 0.5 mL of 0.2 M H₃PO₄. The concentration of ATP, ADP, or AMP after the reaction was analyzed by HPLC. Data were fitted to the substrate inhibition equation, the allosteric sigmoidal equation, or the Michaelis-Menten equation. Results were presented as mean ± SEM.

Determining thermal and pH stability of wild-type and mutated ChPPK

To determine the thermal stability, purified enzymes were incubated in 50 mM potassium phosphate buffer (pH 7.5) at 45°C for various periods, and residual activity was then measured as previously described. The activity of unheated enzymes was taken as 100%.

To determine the pH stability, 50 µg/mL purified enzymes were preincubated in 50 mM potassium phosphate buffer (pH 6.0–7.0) at 4°C for 6 h. The residual activity was then measured as previously described. The activity of enzymes preincubated in 50 mM potassium phosphate buffer (pH 7.5) at 4°C for 6 h was taken as 100%.

Implementation of the ChPPK in the biological synthesis of NMN

The biological synthesis of NMN from R5P was catalyzed by a biocatalytic cascade. In the first step, phosphoribosyl pyrophosphate (PRPP) is formed from R5P after receiving two phosphate groups catalyzed by the BaPRS, and the phosphate donor ATP is converted into AMP. In the second step, NMN is converted from PRPP and NAM by the HdNampt. The initial reaction was performed in a 1 mL mixture containing potassium phosphate buffer (50 mM, pH 7.2), 0.2 or 20.0 mM ATP, 30 mM R5P, 50 mM NAM, 10 mM MgCl₂, 8 mg/mL BL21-BaPRS cell lysate supernatant, and 12 mg/mL BL21-HdNampt cell lysate supernatant. When the ChPPK-based ATP regeneration system was introduced in the NMN synthesis, the 1 mL reaction mixture contained potassium phosphate buffer (50 mM, pH 7.2), 0.2 mM ATP, 30 mM R5P, 50 mM NAM, 10 mM MgCL₂, 4.8 g/L PPA, 8 mg/mL BL21-BaPRS cell lysate supernatant, 12 mg/mL BL21-HdNampt cell lysate supernatant, and 8 mg/mL cell lysate supernatant of wild-type BL21-ChPPK or BL21-ChPPK/A79G/S106C/I108F/L285P. The activities of 8 mg/mL cell lysate supernatant of wild-type BL21-ChPPK and BL21-ChPPK/A79G/S106C/I108F/L285P under standard activity assay conditions were 19.2 and 360.8 U/mL, respectively. The NMN formation was determined at 30, 90, 150, and 210 min by the aforementioned HPLC condition for analyzing the ATP.

Statistical analysis

The one-way analysis of variance test was applied for statistical analysis using the GraphPad Prism software v7.00. Unless otherwise specified, results were presented as mean ± SD (n = 3). Statistical significance was defined as *P < 0.05 and **P < 0.01.

DATA AVAILABILITY

The GenBank accession numbers for ChPPK, PPK12, BaPRS, HdNampt, and iATPSnFR were OM001107, OM001108, HQ636460, MW759281, and OM001110, respectively.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01106-23.

Supplemental file 1. aem.01106-23-s0001.pdf.

Fig. S1 to S6

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.