Network of nicotinamide mononucleotide biosynthesis: substrate tracking, rate-limiting enzyme, and regulatory strategy

Abstract

Nicotinamide mononucleotide (NMN) is a nucleotide of significant biological importance, found abundantly in various foods such as meat, fruits, and vegetables. Recently, its potential effects in delaying aging have attracted considerable attention. Although chemical synthesis methods are commonly employed, they do not align with green production standards. In contrast, the biosynthesis of NMN is both safer and more environmentally sustainable. In this review, we established a novel “substrate-pathway-enzymology” framework to analyze the research on NMN biosynthesis. First, we systematically trace four substrates (nicotinamide ribose, nicotinamide, niacin, and nicotinamide adenine dinucleotide) and their respective metabolic routes. Then, we thoroughly investigate key enzymes through structural biology and protein engineering approaches, and converge the fragmented research findings across pathways to construct a comprehensive NMN biosynthesis network, revealing intricate metabolic regulations and pathway interactions. Through comparative analysis, the most promising biosynthetic pathway and prospects are discussed. Additionally, this review also provides original perspectives for NMN industrial development.

Graphical abstract

Highlights

• •The biosynthesis of nicotinamide mononucleotide (NMN) has profound implications for human health.
• •NMN biosynthesis involves multiple pathways.
• •Genetic engineering and protein engineering enhance NMN biosynthetic efficiency.
• •Researchers have a great interest in NMN production due to its broad market prospects.

1. Introduction

Nicotinamide mononucleotide (NMN) is a precursor to nicotinamide adenine dinucleotide (NAD⁺) and represents a naturally occurring mononucleotide compound recognized for its significant anti-aging effects in rodent models [1,2]. The reduction in NAD⁺ levels concomitant with the aging process is considered a major pathogenic mechanism underlying age-related disorders [3]. A recent study indicates that NMN synthesized by eNAMPT can be converted into NAD⁺ intracellularly, which can extend the average remaining lifespan by up to 2.3-fold in aged mouse models [4]. NMN has shown efficacy in improving a diverse range of cardiovascular conditions, such as vascular aging [[5], [6], [7]], ischemic injury [8,9], as well as metabolic disorders like diabetes mellitus and obesity [10,11]. Furthermore, it shows potential therapeutic benefits in neurological disorders, including Alzheimer's disease [[12], [13], [14]], Parkinson's disease [15], and intracerebral hemorrhage [16]. Additionally, NMN has been shown to mitigate aging-associated complications, such as osteoporosis [17,18], chronic inflammation [[19], [20], [21]], and age-related pigmentation changes [22]. Recent studies have underscored the capacity of NMN to enhance maternal oocyte quality [23,24], suppress tumor growth [25,26], and exert therapeutic benefits for diverse ocular ailments [[27], [28], [29]]. Nevertheless, the over-accumulation of NMN might potentially lead to certain adverse consequences, including hepatic pressure, cancer growth and axon degeneration [10,30]. In recent years, NMN-related products have found applications across diverse fields. In biomedical applications, the combination of NMN with novel hydrogel materials enables targeted therapy and sustained release for wound healing and bone regeneration [31,32]. When integrated with microneedle technology, it achieves efficient transdermal delivery and precise modulation of cutaneous NAD⁺ levels, providing a novel biomedical therapeutic product for anti-aging and metabolic disorders [33]. To amplify synergistic anti-aging effects, NMN is often formulated with resveratrol, glutathione, and other active compounds in nutraceutical products [34]. For transdermal delivery, encapsulation in liposomes and nanospheres dramatically improves skin permeability, facilitating its use in anti-wrinkle and reparative skincare products [22,35,36]. In comparison with NAD⁺, NMN is considered a more economically viable option [37]. Hence, advancing the synthesis of NMN presents promising economic benefits and developmental prospects for the future.

NMN synthesis methods are primarily classified into chemical and biological approaches. To enhance the efficiency of NMN chemical synthesis, Migaud modified the synthesis pathway originally developed by Tanimori. This pathway utilizes ethyl nicotinate and tetra acetyl ribose as precursor compounds, followed by a series of steps, including TMSOTf condensation, deacetylation, phosphorylation with phosphorus oxychloride/trimethyl phosphate, and ammonolysis [[38], [39], [40]]. However, chemical synthesis relies on the use of organic solvents, which fails to fulfill the criteria for green production. In contrast, the biosynthesis of NMN has a more favorable market prospect owing to its environmental friendliness, safety, and efficiency [41]. As a result, research efforts focused on the biosynthesis of NMN are attracting growing attention [42]. Enzymatic catalysis is one of the principal methods for NMN biosynthesis [43], which can be categorized into four distinct pathways (Fig. 1). This review creates a “substrate-pathway-enzymology” framework for NMN biosynthesis study. It traces four substrates (NAM, NR, NA, NAD⁺) and their metabolic routes and examines key enzymes via structural biology and protein engineering. Through the construction of a comprehensive NMN biosynthesis network, this review explores the most promising pathways and prospects and provides novel insights into NMN industrial development.

Fig. 1.

The four main synthetic pathways of NMN. Referring to relevant references [[44], [45], [46], [47], [48], [49]], NMN biosynthetic pathways were mapped with four main routes. MNA: N-methyl nicotinamide; NNMT: nicotinamide N-methyltransferase; NAM: nicotinamide; PRPP: phosphoribosyl pyrophosphate; PRS: phosphoric pyrophosphate synthetase; NAMPT: nicotinamide phosphoribosyltransferase; NR: nicotinamide riboside; NRK: nicotinamide riboside kinase; NA: niacin; NaPRT: nicotinic acid phosphoribosyltransferase; NaMN: niacin mononucleotide; NadE: NMN synthetase; NAR: nicotinic acid RNA; NaAD: nicotinic acid adenine dinucleotide; NADPP: NAD pyrophosphatase; nadV: nicotinamide phosphoribosyltransferases; ushA: UDP-sugar hydrolase gene; deoD: purine nucleoside phosphorylase gene; pncB: niacin phosphoribosyl-transferase gene; nudC: NADH pyrophosphatase gene; mazG: nucleoside triphosphate pyrophosphohydrolase gene; nadD: nicotinic acid mononucleotide adenosine transferase gene; nadR: transcriptional regulator; pncC: nicotinamide nucleotide amidase gene; nadE: NAD⁺ synthase genes; nudE: NADH hydrolase gene.

2. The biosynthesis of NMN from NAM

2.1. Advances in the biosynthesis of NMN from NAM

The biocatalytic pathway for synthesizing NMN from NAM was first proposed by Preiss and Handler [50]. Specifically, NAMPT catalyzes the reaction between one molecule of NAM and one molecule of PRPP, yielding one molecule of NMN and one molecule of pyrophosphate. In addition, inhibiting the activity of nicotinamide N-methyltransferase can increase NAM levels [51]. Several NNMT inhibitors, such as GYZ-319, cyclic peptides, and compound 17u, have been developed, providing a theoretical basis for enhancing NAM availability [[52], [53], [54]]. The kinetic parameters of NAMPT are shown in Table 1. Research primarily focuses on screening NAMPT across various species, including Haemophilus ducreyi [55], Bacillus velezensis [49], Meiothermus ruber [56], Chitinophaga pinensis [57], Arcobacter species phage KVP40 [58], Francisella tularensis [59], and Vibrio nigripulchritudo [60]. Moreover, Asp. KVP40-NAMPT has a relatively ideal specific activity among studied species. Hd-NAMPT shows a high K_cat/K_m of (481.67 ± 11.67) mM⁻¹s⁻¹ for PRPP, and Mr-NAMPT has a high K_cat/K_m of (3.60 ± 0.14)x10³ mM⁻¹s⁻¹ for NAM. This implies that NAMPTs from different species may vary in substrate affinity and catalytic efficiency.

Table 1.

Kinetic parameters of NAMPT.

NAMPT	Spe. Act. (U/mg)	Km, PRPP (mM)	Km, NAM (mM)	Kcat/Km, PRPP (mM−1s−1)	Kcat/Km, NAM (mM−1s−1)	Ref
Hd-NAMPT	/	/	/	481.67 ± 11.67	/	[61]
Mr-NAMPT	3.20 ± 0.11	/	(0.39 ± 0.07) × 10−3	/	(3.60 ± 0.14) × 103	[56]
Mr-NAMPT	0.038 (37 °C)	/	/	/	/	[62]
Asp. KVP40-NAMPT	2.890 × 103	/	/	/	/	[58]
Cp-NAMPT	0.89 ± 0.08	4.45 ± 0.06	5.22 ± 0.21	1.21	0.61	[63]
Cp-NAMPT-Y15S	1.40 ± 0.05	2.01 ± 0.07	4.92 ± 0.23	3.53	0.64	[63]
Cp-NAMPT	/	4.45 ± 0.06	5.22 ± 0.21	1.21	0.61	[64]
Cp-NAMPT-Y13G/Y15S/F76P	/	6.59 ± 0.08	2.52 ± 0.04	1.48	2.41	[64]

To obtain a higher yield of NMN, researchers typically adopt the strategy of multi-enzyme cascade reaction, which has show great potential in NMN synthesis, with advantages of biodegradability, high selectivity, and compatibility [61]. By eliminating numerous purification steps, this approach improves the synthesis efficiency and reduces waste production [65,66]. Meanwhile, the simultaneous co-immobilization of multiple enzymes has proven effective in enhancing overall performance [[67], [68], [69]]. For instance, PRPP and NAMPT were cross-linked with glutaraldehyde by using an amino-type resin carrier [70]. NAMPT and phosphoribosyl pyrophosphate synthetase (PRS) were immobilized simultaneously with an epoxy-type resin (LX3000) [71]. Furthermore, microspheres and resins were utilized to immobilize the mutated PRS and NAMPT [72]. Additionally, researchers have also immobilized enzymes on various other carriers, including silica [73], magnetite [74], and activated carbon [75]. Emerging materials like the composite of polyacrylic acid, porous carbon nanotubes (CNTs) and mesoporous magnetic silica microspheres (mSiO2@SiO2@Fe3O4) have also been applied [76]. The advantages of enzyme immobilization in terms of recycling and reuse contribute to its practical application and advancements in industrial production [77].

2.2. Biosynthesis of precursor PRPP and research progress

PRPP is a vital precursor for NMN synthesis via the NAM pathway [78]. To address the challenges associated with PRPP production, recent studies have proposed various strategies utilizing more accessible and cost-effective substrates, including saccharide and nucleotide groups [61,[79], [80], [81], [82], [83]]. These approaches present feasible pathways for PRPP production, as illustrated in Fig. 2.

Fig. 2.

Other NMN biosynthetic pathways based on precursor PRPP. Referring to relevant references [61,[79], [80], [81], [82], [83]], the synthesis process of PRPP was sorted out into multiple alternative pathways. Xyl: xylose; XR: xylose reductase; XDH: xylitol dehydrogenase; XI: xylose isomerase; Xul: xylulose; Xu5P: xylulose-5-phosphoric acid; XK: xylulokinase; PPI: pentose phosphate isomerase; Glc: glucose; GK: glucokinase; G6P: d-glucose-6-phosphate; G6PD: glucose-6-phosphate dehydrogenase; 6PGL: glucose-delta-lactone 6-phosphate; PGLS: 6-phosphogluconate lactonase; 6 PG: 6-phosphogluconic acid; 6PGD: 6-phosphogluconate dehydrogenase; Ru5P: ribulose-5-phosphate; rpiA/B: ribose-5-phosphate isomerase gene; Ara: arabinose; AI: arabinose isomerase; Ru: ribulose; RuK: ribulokinase phosphate; R5P: ribose-5-phosphate; RK: ribokinase; RPPK: phosphoribosyl pyrophosphorylase; Ado: adenosine; ADK: adenosine kinase; APRT: adenine nucleotidyl transferase; GMP: guanosine monophosphate; IMP: inosine monophosphate; AMPase: AMP nucleosidase; IMPase: IMP nucleosidase; HPRT: hypoxanthine phosphoribosyltransferase; zwf: glucose-6-phosphate dehydrogenase gene; pgl: 6-phosphogluconate lactonase gene; gnd: 6-phosphogluconate dehydrogenase gene.

2.2.1. Saccharides as starting substrates to synthesize PRPP

The enzymatic using saccharides as starting substrates are complex due to the cascade process involving multiple enzymes and substrates. For instance, when converting ribose, ribose-5-phosphate (R5P), NAM, and polyphosphates into NMN, the participation of key enzymes including NAMPT, PRS, ribokinase, and polyphosphate kinase (PPK) is required [84]. Additionally, a NAMPT mutant derived from M. ruber DSM1279, in conjunction with externally added PRS and ribokinase, facilitates the conversion of ribose, NAM, and ATP, achieving a conversion rate of up to 100 % [71]. Similarly, using R5P, ATP, and NAM as raw materials, immobilized cells containing PRS and NAMPT have successfully synthesized NMN at a concentration of 13.3 g/L [70]. Most current studies mainly use glucose as a substrate to obtain PRPP through the pentose phosphate (PP) pathway and regulate the genes in this pathway [57,60,[85], [86], [87], [88], [89]]. In the latest research, by adding the over-expressed xylose transporter XlyFGH, the coupled feeding with xylose and glucose as dual carbon sources in E. coli has been achieved [60]. Notably, this feeding method results in nearly double the NMN yield compared to the single-carbon-source feeding method, revealing the potential of coupled feeding. In term of substrate transport studies for NMN production, the NAM transporter NiaP from Burkholderia cenocacia and the NMN transporter PnuC from Bacillus megaterium exhibit optimal transport efficiency [57,81]. Meanwhile, in mammals, the NMN-specific transporter, Slc12a8, whose mRNA expression increases with the consumption of NAD⁺, promotes the biosynthesis of NAD⁺ [90]. These findings underscore the significance of identifying, evaluating, and applying transporters. Furthermore, an efficient enzymatic approach for the single-step biological production of NMN, which employs eight crucial enzymes, has been demonstrated. This method utilizes cost-effective and renewable resources, such as starch and NAM [62], thereby offering the potential for economical and high-quality industrial NMN production.

2.2.2. Nucleotide as starting substrates to synthesize PRPP

The main nucleotide substrates for PRPP synthesis include adenosine, AMP, IMP, and GMP (Fig. 2). In a certain study, using adenosine and NAM as substrates, through the purification of adenosine kinase, adenine nucleotide transferase, and NAMPT, NMN was synthesized with a concentration of 19.67 g/L [91]. Researchers utilized two distinct multi-enzyme complexes for NMN synthesis when AMP served as the core substrate. One contains AMPase, phosphoribosyl pyrophosphorylase and NAMPT; the other contains adenine nucleotide transferase and NAMPT [61]. Additionally, in a study that utilized IMP as a substrate, purified hypoxanthine phosphoribosyltransferase and NAMPT were employed, along with xanthine oxidase to degrade hypoxanthine produced during the reaction, resulting in a crude NMN product [83]. Alternatively, NMN can also be synthesized through a one-pot reaction catalyzed by IMP nucleotidase, PRPP synthetase, and NAMPT, with IMP, ATP and NAM as substrates [82]. However, the higher costs of raw materials associated with substrates such as AMP, IMP, and GMP, in comparison to adenosine, limit their practical application in large-scale production.

2.2.3. Other supply strategies to synthesize PRPP

One study was designed to augment the expression of zwf and gnd in a mutated strain of Escherichia coli to activate the PP pathway [92]. This investigation demonstrated that the upregulation of zwf remarkably enhanced both the PP and Entner-Doudoroff (ED) pathways, while concurrently suppressing the Embden-Meyerhof-Parnas (EMP) pathway. This led to reduction in the production of pyruvate and an augmentation of the flux of d-glucose-6-phosphate (G6P). Moreover, the overexpression of gnd facilitated an increase in carbon flux through the PP pathway, leading to enhanced NADPH accumulation. Furthermore, the co-expression of zwf and gnd synergistically directed more carbon sources toward the PP and ED pathways without adversely affecting cell growth. Subsequently, researchers utilized glucose as the primary substrate for PRPP synthesis. They passively expressed seven endogenous genes (zwf, pgi, pgl, gnd, prs, rpiA, rpiB) in E. coli, effectively converting it into PRPP via the PP pathway [49]. Although the passive expression of zwf and gnd increased the carbon flux in the PP and ED pathways, the yield of NMN did not increase significantly. This observation suggests that alterations in the balance of cofactors consumed and produced during the process may influence the experimental outcomes. Further refinement and exploration of the regulatory mechanisms, as well as the repression of key factors in the pathway, are warranted in future studies.

2.3. Key enzymes in the biosynthesis of NMN from NAM

2.3.1. Structure and protein engineering of NAMPT and PRS

The NAMPT protein, consisting of 491 amino acids and having a molecular weight of approximately 52 kDa [93,94], typically functions as a symmetrical homodimer, forming two active sites at opposite ends of the dimer [95,96]. Its crystal structure reveals that NAMPT functions as a type II phosphoribosyltransferase in its dimeric form [[97], [98], [99]]. Research has elucidated the dimeric crystal structure of murine NAMPT and determined the crystal structure of the NAMPT-NMN complex (Fig. 3A), providing insights into how NAMPT interacts with its substrate to produce NMN [99]. A distinguishing feature of NAMPT is the hydrogen bond between Asp219 and the amide group, which designates NAM as a specific substrate. His247 plays a crucial role in enzyme activity, assisted by the conserved aspartic acids Asp313 and Asp279, which impart a negative charge to His247, promoting its phosphorylation. This phosphorylation breaks the pyrophosphate bond of PRPP, forming a positively charged intermediate that facilitates the conversion of NAM to NMN. Subsequent investigations using Homo sapiens NAMPT as a model have further explored these molecular mechanisms [43], underscoring the importance of conserved residues such as His247 and Asp313 in the NMN binding pocket and influence on NAMPT's catalytic efficiency.

Fig. 3.

Crystal structures of key enzymes related to NMN biosynthesis from NAM and NR. (A) Structure diagram of mouse NAMPT and NMN complex (PDB: 2H3D) [99]. The salmon color stick part is NMN, and the green, blue and red part is adjacent amino acid residues of NMN (ASP219, ASP313, HIS247, ASP279, ARG311, ARG392, GLY355). (B) Xcc NAMPT and NMN complex structure diagram (PDB: 7YQQ) [100]. The salmon color stick part is NMN, and the cyan, blue, and red part is the adjacent amino acid residue of NMN (GLY364, ARG373, ARG180, PHE177, ARG293, HIS229, ASP335, ASP12). (C) Diagram of the crystal structure of PRS (PDB: 1DKU). [101]. The orange part is the flexible loop, and the cyan, blue, and red are some residue sites (ARG101, HIS135). (D) Crystal structure diagram of the complex of human NRK1 and NMN (PDB: 2QG6) [102]. NMN in salmon color stick, and amino acid residues around NMN are represented by green, blue, and red colors (AGR128, AGR129, ASP56, AGR132, ASP36, LYS16). All protein structures were drawn by PyMoL.

Recently, an intriguing piece of research has explicated that NA is capable of acting as a substitute substrate for NAM within the catalytic procedure of NAMPT, yet this occurs only under the condition that His247 is phosphorylated (pHis247) [103]. Additionally, pHis247 optimizes the positioning of PRPP, stabilizing it to facilitate catalysis. One study investigated the crystal structure of NAMPT from Xanthomonas campestris pv. campestris (Xcc), complexed with its product NMN (Fig. 3B) [100]. Its activity is higher than Homo NAMPT. Moreover, the ribose-phosphate moiety of NMN forms a hydrogen bond with Arg373 and increases contact density with surrounding residues (e.g., Asp12, Phe177, Arg180, Arg293, Gly364) through van der Waals interactions. The study also emphasized the facilitative role of Asp355 and phosphorylated His229 in NMN synthesis and introduced the H229A mutation. To date, there has been notable progress in the protein engineering of NAMPT and PRS. Researchers have performed site-directed mutagenesis to develop a highly catalytic NAMPT mutant, utilizing NAMPT from M. ruber DSM1279 and Luteibacter sp. as the parental strain. The resulting mutants, E231Q/D298A/D338E/D337E and R189H/S232T/R392K, demonstrated specific activities that were 6.9 times and 9.5 times higher than that of the wild type, respectively [104,105]. Furthermore, a study introduced single and multiple point mutations at positions 314, 315, 417, 419, 450, and 452 in the sequence, resulting in a NAMPT mutant with a specific activity that is 5.3 times higher than that of the wild type [106]. Additionally, a previous study demonstrated that the Y15S single-point mutation significantly improved NAMPT's thermostability (extending t_1/2 by 150.15 h) and PRPP catalytic efficiency (1.92-fold increase in K_cat/K_m) by optimizing hydrogen-bonding networks [63]. However, this single mutation failed to address the substrate channel bottleneck under high-concentration conditions. Building on this foundation, researchers advanced a dual-channel engineering strategy by combining Y15S with channel-widening mutations (Y13G/F76P), which boosted K_cat/K_m for NAM by 3.95-fold, while Y15S preserved PRPP-binding stability. This synergistic design enabled NAMPT-Y13G/Y15S/F76P to achieve 99.8 % conversion (19.94 g/L NMN), marking a pivotal leap from localized optimization to systematic enzyme engineering [64]. Moreover, recent structural analysis of Hd-NadV also shows it regulates NMN synthesis via dual NAM binding sites (active and allosteric) [107]. Conserved residues like D222 and R311 in the active site ensure substrate specificity, while the allosteric site's hydrophobic pocket (H194/I363) likely boosts catalytic efficiency through conformational changes. This discovery provides well-defined targets for Hd-NadV's protein engineering.

PRS is widely distributed across various organisms and utilizes ATP and R5P as substrates for the catalysis of PRPP synthesis. The PRS family is classified into three distinct classes [[108], [109], [110]], with recent studies primarily focusing on type I PRSs in biosynthetic research. A prior investigation revealed a flexible loop comprised of residues Tyr97 to Thr113 between the two domains of B. subtilis PRS1 in its crystal structure (Fig. 3C) [101]. This loop undergoes conformational changes during catalysis and contains conserved residues (Arg101, His135) that are crucial for PRPP synthesis. Marinescu et al. successfully synthesized NMN by enhancing the activity of a PRS enzyme [55]. They achieved this by introducing a specific mutation L135I in PRS from Bacillus amyloliquefaciens, in combination with NAMPT in E. coli. This mutation effectively eliminated the inhibitory feedback mechanism of PRPP synthase, leading to an increase in PRPP production [111]. Furthermore, the combined mutations of PRS and NAMPT show promise for enzyme research and the application of the NMN synthesis pathway. Researchers performed site-specific mutations on PRS and NAMPT and analyzed them using molecular docking technology [63]. The co-catalysis of the advantageous mutants PRS-H150Q and NAMPT-Y15S led to an NMN concentration of 8.1 g/L. Moreover, it is worth noting that the K_cat/K_m value of the PRS-H150Q (11.33 mM⁻¹ s⁻¹) toward ATP was increased by 3.26-fold compared with the wild type, and when R5P was used as the variable substrate, the K_m value of the PRS-H150Q (3.46 mM) was decreased by 20.8 % compared with that of the wild type, and the interaction between the PRS-H150Q and the byproducts (ATP, ADP, PPi, AMP) was weakened. These findings provide valuable insights and guidance for the protein engineering of key enzymes in the NAM pathway for NMN synthesis.

3. The biosynthesis of NMN from NR

3.1. Advances in the biosynthesis of NMN from NR

The enzymatic conversion of NR into NMN through NRK enzyme was first documented in 2004 [112]. Besides utilizing NR as a substrate, NR can also be synthesized through a series of enzymatic reactions, which generally take glycoside compounds as substrates. Consequently, there are roughly four pathways for the biosynthesis of NMN from NR (Fig. 4) [61,[113], [114], [115]]. Recent research on NRK has increasingly focused on heterologous expression and species screening. Thermothielavioides terrestris NRRL 8126 [116], Homo [117], Kluyveromyces marxianus [105], and Saccharomyces cerevisiae has been successfully expressed [118]. The kinetic parameters of NRK are shown in Table 2. In particular, the specific activities of Sc-NRK1 and Klm-NRK enzymes are 7.9 U/mg and 2252.59 U/mg respectively [118,119]. Klm-NRK and Homo-NRK2 also demonstrate favorable binding affinities and catalytic efficiencies towards NR and ATP substrates [120]. Additionally, a separate study utilized a bioassay system with an auxotrophic yeast that requires NR to isolate strains capable of producing NMN and NR [42]. However, the yields of strains used for naturally expressing products are usually low, which limits their industrial applicability.

Fig. 4.

Other synthetic pathways based on NR. Referring to relevant references [61,[113], [114], [115]]. PPM: pentose phosphate mutase; R1P: ribose-1-phosphate; Ado: adenosine; PNP: purine nucleoside phosphorylase; Gua: guanosine; Uri: uracil; PyNP: pyrimidine nucleoside phosphorylase; UPP: uridine phosphatase.

Table 2.

Kinetic parameters of NRK.

NRK	Spe. Act. (U/mg)	Km, NR (mM)	Km, ATP (mM)	Kcat/Km, NR (mM−1s−1)	Kcat/Km, ATP (mM−1s−1)	Ref
Sc-NRK1	2252.59	18.69	54.27	/	/	[118]
Sc-NRK1	0.641	/	/	/	/	[121]
Klm-NRK	0.270	/	/	/	/	[105]
Klm-NRK-D45E/D58Q/R61K/Y64W	1.725	/	/	/	/	[105]
Klm-NRK	7.9 ± 0.42	(4.5 ± 1.1) × 10−2	(7 ± 0.6) × 10−2	84.4	57.4	[119]
Homo-NRK	1.75	/	/	/	/	[117]
Homo-NRK2	0.064	3.6 × 10−3	7.5 × 10−3	4.725 × 103	2.286 × 103	[118]
NRK-D45T/I88R/E189A	22.95 ± 0.31	(40 ± 0.36) × 10−3	/	(62 ± 0.59) × 103	/	[122]

To strengthen the enzyme activity and stability of NRK, researchers have adopted various immobilization methods. For instance, the soluble protein expression of NRK from S. cerevisiae was achieved in BL21, and the stability of this enzyme was further improved through immobilization using a 6-HIS tag [118]. Meanwhile, the flocculin system in S. cerevisiae was also applied to the cell surface engineering of NRK [118]. In addition, researchers utilized the agglutinin system in yeast cell expression engineering to fuse the C-terminus of Homo NRK2 with the binding subunit Aga2p, realizing its efficient display on the cell surface [120]. The yeast surface display engineering is a mature technology that eliminates the need for additional protein purification and immobilization steps, thus simplifying the operation [[123], [124], [125]]. Integrating traditional immobilization enzyme techniques with the yeast surface display system is expected to open up promising directions for the follow-up research on NRK.

Additionally, to reduce feedstock cost, researchers have explored alternative routes for NR synthesis. In one study, with d-ribose, NAM, and ATP as substrates, the production of NMN was successfully achieved through a multi-enzyme cascade reaction involving ribokinase, pentose phosphate mutase, and NRK [113]. A mutant uridine phosphatase was discovered in E. coli K12, and this enzyme could catalyze the conversion of NAM and ribose-1-phosphate into NR [115]. Moreover, when adenosine was used as a substrate, through the catalysis action of purine nucleoside phosphorylase from Bos taurus, NRK from Haemophilus influenzae ATCC14592, and PPK, NMN could be produced at a high titer of 75.15 g/L [114]. Additionally, methods for increasing the yield of NR with uridine as a substrate were also explored. This process employed a cascade of enzymes, including pyrimidine nucleoside phosphorylase, purine nucleoside phosphorylase, NRK, and PPK2 [61]. Notably, pnuC, an efficient NR transmembrane transporter, can enhance NR intracellular uptake, thus showing great potential in the whole-cell NR-to-NMN synthesis pathway [126]. Furthermore, research has indicated that NMN can be converted extracellularly to NR through a CD73-dependent pathway by ecto-5′-nucleotidase [127]. This pathway also provides ideas for the mutual conversion mechanism between NMN and NR. Researchers have also optimized the HPLC method under this pathway, providing a reference for subsequent product analysis [128].

3.2. Structure and protein engineering of NRK

NRK, a member of the nucleoside monophosphate kinase superfamily, plays a pivotal role in the synthesis of NMN [102]. Previously, research has been investigated the crystal structure and molecular mechanisms of NRK1, with particular emphasis on its catalytic processes and the relationship with related products (Fig. 3D) [102,129]. In NRK1, Asp56 and Arg129 are critical for recognizing the ribose hydroxyl group of NR, while Arg128 and Arg132 can bind to ATP. Additionally, Asp36, Asp56, and Lys16 are proposed to be active sites. Researchers often use mutation and protein engineering to enhance the catalytic activity of NRK. For instance, a quadruple mutation (D45E/D58Q/R161K/Y164W) in an NRK mutant derived from K. marxianus exhibited a 6.4-fold increase in activity compared to the wild-type [105]. There is also a study in which 15 mutations were introduced into the wild-type Homo NRK, enhancing the stability and catalytic activity of the mutant enzyme [117]. Furthermore, by analyzing the amino acid sequence and structure of NRK1, researchers carried out rational design and site-directed mutagenesis, introducing two mutations (T47V and N62D) into NRK1 [130]. Compared with the wild type, the catalytic activity of the T47V mutant increased by 130 %, the catalytic efficiency increased by 100 %, and the stability increased by 20 %, highlighting the importance of rational amino acid mutations near the active site. Currently, the semi-rational mutagenesis strategy of docking combination simulation (DCS) was developed. Using DCS, the mutant NRK-D45T/I88R/E189A was obtained. Its enzyme activity (22.95 U/mg) increased by 1.9-fold, and both stability and substrate affinity were notably enhanced. This shows that applying computer networks for protein structure prediction and virtual mutation is crucial in protein engineering. It boosts accuracy and saves time for wet experiments [122].

4. The biosynthesis of NMN from NA

4.1. Advances in the biosynthesis of NMN from NA

In the synthesis of NMN from NA (Fig. 1), NaMN is involved as a key intermediate, and the enzyme involved in NMN synthesis is NadE. It is worth mentioning that Heuser et al. significantly enhanced intracellular NAD⁺ levels in E. coli by expressing pncB and nadE [131]. Subsequent studies focused on screening NadE from F. tularensis to facilitate the amination of NaMN for NMN production [132]. Black et al. further found that by overexpressing Ft. NadE and knocking out the pncC and nadR genes, the intracellular NMN levels in E. coli was increased 1000-fold [59]. In parallel, research successfully improved NaMN biosynthesis by expressing pncB [133]. Researchers also heterologously express Ft. NadE in the L. lactis NZ9000 strain and knocked out the nadR gene using CRISPR/Cas9, increasing the NMN accumulation by 61 % [46], highlighting the potential of heterologous NadE expression in L. lactis NZ9000. Future investigations should screen the enzymes in this pathway for species and strain expression systems and actively explore modification strategies for NMN-related metabolic pathways.

4.2. Structure and protein engineering of NadE and NaPRT

NaPRT is the initial enzyme that catalyzes the synthesis of NAMN and PPi from NA and PRPP, respectively. Initially referred to as NaMN pyrophosphatase. NaPRT plays a critical role in increasing NAD⁺ levels and serves as the rate-limiting enzyme in the third step of the Preiss–Handler pathway for NAD⁺ biosynthesis. Homo NaPRT is predicted to have a molecular weight of 58 kDa, with a molecular mass of approximately 87,000 Da after gel filtration, and forms a dimeric crystal structure [134]. A study that mutated the conserved residues R318, Y21, and H213 in NaPRT to reduce its catalytic activity highlighted these amino acids as essential for catalytic function [134]. Subsequent crystal structure analysis indicated that R171, K396, and S214 likely play crucial roles in substrate recognition, while R318 was the main contributor to the catalytic action of this enzyme (Fig. 5A) [135]. The NaMN synthetic pathway is catalyzed by NadE from Ft. novicida U112. Following this, the crystal structure of Ft. NadE was determined (Fig. 5B), revealing that it exists as a homodimer [132]. This enzyme is specific for NaMN and requires NH₃ for the production of NMN. The study identified key residues in the catalytic active site of Ft. NadE, including E147, D202, K171, D41, T142, and S193, with Gln149 playing a significant role in substrate binding channel for NaMN.

Fig. 5.

Crystal structures of key enzymes related to NMN biosynthesis from NA and NAD⁺. (A) NAPRT crystal structure (PDB: 4YUB) [135]. The purple, blue, and red sticks are part of the related amino acid residues (SER214, ARG171, HIS213, ASN196, TYR21, ARG318). (B) Compound crystal of Ft. NadE and AMP (PDB: 3FIU) [132]. AMP in deep blue stick, some amino acid residues around AMP in pink, red, and blue sticks (SER193, ASP41, THR142, ASP202, GLU147, TYR149). (C) Crystal structure of NADPP and NAD complex (PDB: 5IW4) [136]. The purple stick is NAD, and the amino acid residues around NAD are represented by light-purple, blue, and red sticks (TRP194, TYR124, PHE160, GLU174, GLU219, GLU178). (D) Crystal structure of NADPP and NMN complex (PDB: 5IW5) [136]. NMN in salmon color stick and the amino acid residues around NMN are represented by blue-green, red, and blue sticks (SER199, PRO193, VAL240, TRP194, THR239, ARG69). All protein structures were drawn by PyMoL.

5. The biosynthesis of NMN from NAD⁺

5.1. Advances in the biosynthesis of NMN from NAD⁺

The pathway for NMN synthesis from NAD⁺ involves the enzymatic action of NAD pyrophosphatase (NADPP), which catalyzes the hydrolysis of NAD⁺ to produce NMN and AMP [137]. NADPP, a member of the nucleoside diphosphate hydrolases family, specifically cleaves the pyrophosphate bond in NAD⁺ and is widely distributed across plant, animal, and microbial cells. This enzymatic pathway was one of the earliest methods utilized for the synthesis of NMN [138]. However, there are limited reports on NMN synthesis using NAD⁺ as a substrate, likely due to the instability and high cost associated with NAD⁺, which make it unsuitable for industrial-scale NMN production.

Current research on this pathway primarily focuses on NAD⁺ metabolism, involving enzymes such as nudE [139,140], nudC [136], mazG [141], as well as enzymes participating in the process of NAD⁺ being decomposed into NAM, including SIRTs, PARPs, NAD(H) glycosidases, and cADPR synthase CD38 [[142], [143], [144], [145]]. Previous studies have employed enzymes like NudC and NudE to convert NAD⁺ into NMN [139,140]. Enzymes with similar functions in the catabolism of NAD(P)H, such as mazG and ushA, also possess relatively high NAD⁺ hydrolase activities [141,146]. Furthermore, the deletion of ushA could stabilize the NAD⁺ levels [146], while the deletion of nudC and nudE would attenuate the catabolic pathway of NAD⁺ [147]. Therefore, enhancing the activities of nudC and nudE can promote the conversion of NAD⁺ into NMN and increase the accumulation of NMN. Moreover, NAD-consuming enzymes are of crucial importance in the catabolism of NAD⁺, including cADPR synthases CD38 and CD157 [145], PARPs [142,143], SIRTs, and NAD(H) glycohydrolases [148,149], which cleave NAD(H) into NAM [144]. Consequently, conducting engineering research on these related enzymes to reduce the conversion pathways of NAD⁺ to substances other than NMN might yield valuable insights for further research in this field.

5.2. Structure and protein engineering of NADPP

Höfer et al. elucidated the crystal structure of NAD pyrophosphatase (NADPP, also referred to as NudC) homodimer in E. coli complexed with NAD and NMN (Fig. 5C/D) [136]. This study revealed the molecular mechanism underlying the enzyme's interaction with NAD and NMN. Specifically, Trp194 was found to engage in distinct interactions with both the substrate NAD⁺ and the product NMN. It imperfectly stacks with the nicotinamide moiety upon binding to NAD⁺ and interacts with the ribose ring when bound to NMN. The inactivation of the W194A mutant further confirmed the significance of this residue in these interactions. Furthermore, Arg69 was shown to be the main cause for the change in orientation of the bound NMN product compared to the NAD substrate. These findings provide a theoretical foundation for future investigations aimed at engineering the catalytic properties of NADPP.

6. The regulatory of ATP in the biosynthesis of NMN

ATP plays a pivotal role in driving various metabolic pathways [150]. Strategies to enhance ATP availability mainly cover methods such as metabolic engineering and manipulation of energy substrates. One common method is the fused expression of enzymes involved in ATP generation and cycling [61,119], along with key enzymes in the tricarboxylic acid cycle, such as pyruvate kinase and phosphoglycerate kinase [151]. NAMPT can undergo phosphorylation via ATP hydrolysis, which activates its catalysis activity and generates a by-product ADP [95,152,153]. However, in the process of synthesizing NMN by enzymatic coupling reaction with AMP as the substrate, when the optimal enzymatic coupling catalytic ratio of AMPase/RPPK/NAMPT changes from 1:6:8 to 1:6:10, the yield of NMN does not increase but decreases. This shows that the excessive addition of NAMPT enzyme will lead to the excessive accumulation of ADP [61]. Moreover, ADP is also an allosteric inhibitor of RPPK, thus inhibiting the catalytic efficiency of the enzymes throughout the process [78]. To address this issue, researchers introduced polyphosphate kinase 2 from Sinorhizobium meliloti to facilitate the cycling of ADP back to ATP. As a result, the ADP content in the reaction decreased by nearly 1.5-fold, and the yield of NMN increased by 16.7 % [61]. Moreover, the PPK2 from Lampropedia hyalina DSM 16112 and Sulfurovum lithotrophicum exhibited specific activities of 172.3 U/mg and 282.9 U/mg respectively [154], species Sinorhizobium meliloti 02148 and Deinococcus radiodurans have also been reported to be used in the reaction of NMN synthesis [64,155]. This provides references for screening species for co-expression with PPK2 in the ATP cycling system. Similarly, by simultaneously expressing NRK and acetate kinase (ACK) in the reaction of NR to NMN synthesis, researchers found that ACK, with its exceedingly high specific activity of 1876 U/mg, enabled ATP to be recycled 144.8 times in the system. Thus, ACK demonstrated great promise in terms of ATP regeneration [119].

In addition to directly regulating ATP, indirect regulation can be achieved by manipulating the precursor substances or transformation products of ATP. Previous studies have demonstrated that PPi, a byproduct of NAMPT catalytic reaction, can enhance ATP hydrolysis, thereby increasing substrate affinity and improving the catalytic efficiency of NAMPT [152]. Metabolites such as ADP and AMP have also been shown to influence ATP levels [156]. The ATP can be indirectly regulated by modifying the AMP pool through the salvage pathway involving the amn, ado1, and add genes [48]. In one study, the ado1 gene was integrated into E. coli while the amn gene was knocked out for the synthesis of NMN, resulting in a marked increase in intracellular ATP levels [49]. In addition, regulating the ATP level of the host strain is also a promising optimization strategy. CRISPR interference (CRISPRi) gene editing, with its precise gene suppression effect, is increasingly being used in the field of biosynthesis. Shen et al. utilized CRISPRi to selectively target genes that consume NADPH and ATP in E. coli, significantly increasing the yield of the target product 4-hydroxyphenylacetic acid without affecting the normal growth of the strain [157], highlighting the importance of inhibiting non-essential enzyme genes for increasing the yield of biological products. Moreover, the introduction of the SOMACA method has enhanced the rapid assembly of crRNAs, improved the efficiency of CRISPR-Cpf1 gene editing in B. subtilis [158], simplified the editing process, and enhanced the utility of CRISPRi in biosynthesis. Future research could focus on using CRISPRi gene editing to manipulate ATP level, which might change the biosynthetic pathway of NMN, particularly in enhancing the synthesis of NMN from the NR and NAM pathways.

7. Future and prospects

7.1. Prospects for potential pathways

To meet the demands of industrial production, researchers have been constantly and actively exploring a simple, efficient, and cost-effective biosynthetic method for NMN. Firstly, an analysis was carried out on the advantages and disadvantages of the four main synthetic pathways of NMN (Table 3), and various pathways for NMN synthesis in current investigation were compared, comprehensively presenting the corresponding research results (Table 4). Apparently, the pathways for synthesizing NMN from NA and NAD⁺ have received less attention and have relatively lower yields (Table 4). In contrast, research on synthesizing NMN via the NAM pathway is more extensive. Especially when NAM and carbon source are used as substrates, a multi-enzyme “one-pot” synthesis method can obtain a relatively high yield of NMN. However, the involvement of multiple enzymes complicates NMN's subsequent purification. The synthesis of NMN through NR pathway, featuring one-step synthesis, a single enzyme, and low-cost substrates, shows an interesting contrast. Despite relatively limited research, it offers relatively high yields. Thus, we believe that this pathway holds great potential in NMN production.

Table 3.

Advantages and disadvantages of the four NMN synthesis pathways.

Routes	Advantages	Disadvantages	Optimization schemes
Biosynthesis of NMN from NAM	NAM is economical Nampt is more studied	Poor PRPP stability Higher cost Difficult to purify	1) Study PRPP synthesis pathway 2) Mutate NAMPT via protein engineering 3) Modify metabolic pathways to improve NMN in expression strain
Biosynthesis of NMN from NR	Simple Reaction NR is cheaper Easy to purify NMN	NRKs are less studied ATP is more expensive	1) Explore species with higher NRKs activity 2) Mutate NRK via protein engineering 3) Promote screening and testing of ATP coupling enzymes or regulate ATP levels in expression strains
Biosynthesis of NMN from NA	Simple Reaction NA is cheaper	NadE is less studied NaMN's high price Difficult to purify	1) Screen for NadE-related enzymes with higher activity 2) Modify pathways to reduce NMN catabolism
Biosynthesis of NMN from NAD+	Single substrate Simple Reaction Easy to purify NMN	NAD+ is less stable and high price.	1) Modify NAD+ metabolic pathway to enhance NMN production

Table 4.

Summary of the method and yield of NMN synthesis through different substrates and enzymes.

Substrate	Enzyme	Year	Strategy	NMN Production	Reference
NAM Glucose Xlyose	NAMPT (V. nigripulchritudo)	2024	Co-expression of NAMPT, prs, zwf, gnd, glk (encoding glucokinase) Addition of PnuC transporter Deletion of pncC, nadR, pncR, ushA	46.66 g/L	[60]
NAM Lactose	NAMPT (H. ducreyi)	2018	Co-expression of NAMPT, PRS-L135l	0.015 g/L	[55]
NAM AMP R5P PRPP	NAMPT (H. ducreyi)	2022	Co-expression using multiple enzymes (AMPase, RPPK, PPK2) with reaction optimization	3.000 g/L	[61]
NAM Ribose R5P ATP	NAMPT (H. ducreyi)	2018	Co-expression of NAMPT, PRS-N120S, RK, PPK, PPase	1.370 g/L	[84]
NAM Glucose	NAMPT (H. ducreyi)	2023	Addition of PnuC, NiaP transporters Co-expression of zwf, pgl, gnd, ripA/B, prs	0.602 g/L	[81]
NAM AMP	NAMPT (H. ducreyi)	2020	Co-expression of AK, APRT, NAMPT	19.670 g/L	[91]
NAM IMP ATP	NAMPT (Shewanella oneidensis MR-1, H. ducreyi)	2022	Co-expression of NAMPT, IMPase, PRS-L135l, XOD, CAT, PPase	11.120 g/L	[82]
NAM PRPP	NAMPT (M. ruber)	2021	Single-enzyme catalysis. Optimization of expression conditions	0.030 g/L	[56]
NAM Ribose ATP	NAMPT (M. ruber DSM1279)	2021	Co-expression of NAMPT-E231Q/D298A/D338E/D377E, PRS, RK	8.300 g/L	[71]
NAM IMP PPi	NAMPT (M. ruber DSM1279)	2021	Co-expression of NAMPT-E231Q/D298A/D338E/D377E, XOD, HPRT	14.640 g/L	[83]
NAM starch	NAMPT (M. ruber)	2023	Co-expressed αGP,PGM,G6PDH,6PGDH,RPI,PRS,PK,ADK,PPK	CVR = 87.8 %	[62]
NAM Glucose	NAMPT (B. velezensis)	2021	Co-expression of NAMPT, PRPPs, G6PD, PGLS, PGD, RpiA, YgcS, ado1 Deletion of pncC, nadR, amn	0.496 g/L	[49]
NAM Glucose Ribose	NAMPT (Homo) PRPS1, PRPS2 (Homo)	2025	Co-expression of NAMPT, PRS1, PRS2, Utilization of a cell-free enzyme system	0.501 g/L	[87,159]
NAM Glucose PRPP	NAMPT (C. pinensis)	2021	Co-expression of NAMPT, PRPPs, G6PD, PGLS, RpiA. Screening for NiaP (B. cenocepacia), PunC (B.mycoides)	6.790 g/L	[57]
NAM PRPP	NAMPT (C. pinensis)	2024	Co-expression of NAMPT-Y13G/Y15S/F76P, PPK	19.940 g/L	[64]
NAM Ribose	NAMPT (C. pinensis)	2024	Co-expression of NAMPT-Y15S, PRS-H150Q, RK,PPK,PPase	8.100 g/L	[63]
NAM Ribose ATP	NAMPT (C. pinensis)	2024	Co-expression of NAMPT, PRS, RK,PPase and double PPK2	CVR = 97.5 %	[155]
NAM Glucose	NAMPT (C. pinensis)	2024	Co-expression of NAMPT, PRS Addition of NiaP, PnuC transporters	1.005 g/L	[88]
NAM Glucose	NAMPT (Vibrio bacteriophage KVP40)	2022	Co-expression of NAMPT, PRS-L135l Addition of NiaP, PnuC transporters Deletion of pncC, purR, nadR, ushA	16.200 g/L CVR = 97 %	[45]
NAM	NAMPT (Vibrio bacteriophage KVP40)	2025	Co-expression of NAMPT, PRS-D128A, zwf, gnd Addition of PnuC transporter Deletion of pncC, nadR, ushA, pfkA, talA Addition of ML-guided cofactor engineering	20.13 g/L	[89]
NAM	NadV, NadE (F. tularensis)	2020	NadV is co-expressed with NadE enzyme, knockdown of pncC, nadR	0.360 g/L	[59]
NA NAM Asp	NadV, NadE (F. tularensis)	2024	Co-expression of NadV, NadE NadàE, mazG, pncB Addition of PnuC and NiaP transporters Deletion of pncC Addition of ML-guided cofactor engineering	1.026 g/L	[86]
NA NAM	NadE (F. tularensis) NadV (Ralstonia solanacearum)	2020	NadE and NadV co-expression, knockdown of pncC	0.501 g/L	[160]
NAM R5P ATP	NAMPT PRS	2018	Immobilization of NAMPT, PRS and realization of whole-cell catalysis	13.300 g/L	[70]
NAM R5P ATP PRPP	NAMPT (Homo) PRPS1, PRPS2 (Homo)	2021	Co-expression of NAMPT, PRPS1, PRPS2	/	[159]
NR ATP	NRK	2024	Co-expression of NRK-D45T/I88R/E189A, PPK2	15.160 g/L	[122]
NR ATP	NRK (S. cerevisiae)	2023	Immobilized enzyme with 6-HIS tag	activity of fixed enzyme was 7 times higher than the wild type	[118]
NR ATP	NRK (S. cerevisiae)	2021	NRK expressed in tandem with FLO1 (S. cerevisiae) on the surface of S. cerevisiae cell walls	16.440 g/L CVR = 98 %	[121]
NAM Pi ATP Guanosine	NRK (S. cerevisiae S288C)	2020	Expression strain E.coli K12 co-expresses NRK, UPP, PNP	0.769 g/L	[115]
NR ATP	NRK (Kl. marxianus)	2019	Expression of NRK-D45E/D58Q/R161K	CVR > 90 %	[105]
NR ATP	NRK (Kl. marxianus)	2022	Co-expression of NRK, ACK (Geobacillus stearothermophilus)	93.5 g/L	[119]
NR ATP	NRK (Homo)	2020	Co-expression of NRK, PPK2 (E. coli)	CVR > 99 %	[117]
NR ATP	NRK2 (Homo)	2022	Expression of NRK2 fused to Aga2p on the surface of S. cerevisiae cell walls	12.600 g/L CVR = 98.2 %	[120]
NAM Ribose ATP	NRK (Homo)	2021	Co-expression of NRK, RK, PPM	RVR > 90 %	[113]
NR ATP	NRK1 (Homo)	2023	Catalyzed by the single enzyme NRK-T47V	4.130 g/L	[130]
NR ATP	NRK (T. terrestris NRRL 8126)	2020	monoenzyme catalysis	CVR = 93 %	[116]
NAM Uri Pi ATP	NadE (EC:2.7.1.22)	2022	Co-expressed with PyNP, PNP, PPK2	0.040 g/L	[61]
Ethanol NaMN	NadE (F. tularensis)	2024	Co-expressed with Aro3/4, PhzE, PhzD, DhbX, NbaC, NadC, NadE	0.980 g/L	[88]
NaMN NH3	NadE (F. tularensis)	2009	monoenzyme catalysis	/	[132]
Glucose	NadE (Bacillus subtilis)	2024	Co-expression of NadE, PncB, PnuC	3.398 g/L	[161]
NAM	NadE (F. tularensis)	2023	Knockdown of nadR, expressed in L. lactis NZ9000	0.765 g/L	[46]
NAD+	NADPP	1988	monoenzyme catalysis	/	[137]

CVR: conversion rate; RVR: recovery rate.

Currently, the enzymatic conversion of NR to NMN faces four key limitations in NMN applications: 1) over-reliance on complex multi-enzyme systems documented in patents. 2) insufficient exploration of single-enzyme (NRK-catalyzed) pathways. 3) The application of chassis strains is relatively monotonous. 4) The supply issue of ATP. To optimize this bioconversion process for industrial-scale production, five strategic research directions should be prioritized: 1) Enzyme discovery: screen more promising NRK species to expand the NRK gene pool. 2) Protein engineering: Leverage AI and machine learning for rationally designing the NRK enzyme, enhancing targeted mutation precision and protein engineering efficiency. 3) Optimization of cost: research the feeding systems in industrial fermenters, such as exploring strains that can naturally synthesize NR or ATP to reduce industrial feeding costs. 4) ATP optimization: identify more highly active ATP-recycling enzymes and screen master cell bank genes sensitive to ATP-level regulation. 5) Discovery of chassis strains: conduct a comprehensive exploration and practical application of the genera used to construct chassis cells, so as to continuously enrich the application database of chassis strains. These optimizations will improve the efficiency and scalability of NR to NMN production, and promote the industrialization of this approach.

7.2. Prospects for future industrialization

This review constructs a “primary substrates-precursor tracking-research progress-key enzyme analysis” framework to summarize NMN biosynthesis advances across multiple pathways. It: 1) Traces NMN precursors' (NAM, NR, NA, NAD⁺) metabolic traits and bioproduction uses from substrate utilization; 2) Focuses on key enzymes' (NAMPT, PRS, NRK, NAPRT, NADPP) catalytic mechanisms and engineering strategies; 3) Highlights host metabolic network optimization cases like ATP regeneration. Despite much current research, industrial NMN synthesis still faces challenges: 1) Synthesis efficiency due to pathway competition and metabolic burden; 2) High-cost precursor reliance; 3) Compliance risks from microbial host endotoxin residues. Addressing these bottlenecks requires multidimensional strategy integration, spanning from metabolic flux optimization to innovative host system engineering.

Recent studies focused on the synergistic activation of multiple NMN-related pathways through co-substrate strategies and metabolic engineering. In E. coli, simultaneous NAM, aspartate, NA, and PRPP supplementation was shown to concurrently activate the Preiss-Handler, salvage, and de novo biosynthesis pathways [86]. Similarly, Pichia pastoris demonstrated enhanced NMN production when ethanol, glucose, and NAM were co-supplied, achieving dual activation of de novo and salvage pathways [88]. Another innovative strategy involves overexpressing the xylose transporter to enable coupled xylose-glucose utilization for NMN biosynthesis [60], effectively linking substrate co-utilization with pathway optimization. These integrated approaches maximize the inherent metabolic potential of host cells through pathway complementarity and substrate channeling with synergistic yield improvement. Moreover, the Yeast-E. coli co-culture system increased the production of NMN by 2.7 times through the sharing of ATP, demonstrating the industrial value of cross-species metabolic complementarity in cost control [87]. In addition, the integration of machine learning (ML)-guided optimization with quorum sensing (QS)-controlled cofactor engineering dynamically balanced intracellular NADPH/NADH levels has significantly mitigated the problems of limited NMN production and reduced strain viability during high-density fermentation caused by elevated cofactor levels and verified the feasibility of industrial-scale expansion [89]. The deep integration of multi-pathway activation, co-substrate optimization, and cross-species metabolic interactions, combined with QS-ML dynamic cofactor regulation, holds promise for overcoming NMN biosynthesis bottlenecks while maintaining strain viability, potentially advancing industrial-scale fermentation process innovation. Additionally, although E. coli remains the dominant host for NMN production, its endotoxin content restricts food-grade applications [162]. Alternative hosts like yeast (GRAS organism) have been explored [88,123], yet industrial scalability is hindered by prolonged fermentation, strain instability, and low secretion efficiency. Notably, Bacillus subtilis, a food-safe chassis already adopted in food biotechnology has demonstrated direct NMN synthesis capability [161], positioning a promising candidate for industrial-scale NMN production.

Moreover, emerging advances in rapid enzyme prototyping emphasize deeper integration of computational and experimental approaches. For instance, the DCS semi-rational design strategy synergizes molecular docking (AutoDock Vina), structure-guided consensus mutagenesis (Rosetta Design), molecular dynamics simulations (GROMACS), and machine learning tools (AlphaFold) to precisely screen key mutation sites, achieving a 75 % effective mutation rate and establishing a streamlined “compute-predict-validate” framework for industrial enzyme development [122]. Notably, CaverDock simulations revealed channel energy barrier, which were mitigated by the Cp-NAMPT-Y13G/F76P mutation. This dual-channel engineering Molecular increased channel cross-sectional area by 38 %, improved K_cat/K_m by 2.3-fold, and enhanced PPitolerance by 29.4 %, demonstrating a closed-loop design paradigm combining simulation, structural engineering, and multiscale validation [64]. Look forward, the filed should prioritize a tripartite combined strategy of “dynamic conformational analysis-intelligent design-automated validation”: Molecular simulations will guide structural optimization (e.g., substrate channel expansion), AI algorithms will accelerate mutant design, and automated platforms (e.g., microfluidic) will enable rapid high-throughput screening. Future efforts should focus on simplifying workflows to bridge lab-to-industry translation while exploring non-natural catalytic functions in artificial enzymes, transcending the natural evolutionary constraints. Overall, Integrating these research efforts will help to develop high-quality NMN supplements for daily use and ultimately benefit human health.

CRediT authorship contribution statement

Puyue Zhang: Writing – review & editing. Ziru Ye: Formal analysis. Zhong Tian: Supervision. Qing Liu: Supervision. Yong Huang: Conceptualization.

Data availability

No data was used for the research described in the article.

Funding

This work was supposed by the Talent Introduction Program at Chengdu University of Traditional Chinese Medicine (030041227).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.