d-(+)-Pantothenate (DPA, also known as vitamin B5) is an essential nutrient for normal growth of living organisms, involving in the biosynthesis of coenzyme A (CoA) and acyl carrier protein (ACP) (Li et al., 2022, Tigu et al., 2018), which play important roles in fatty acids metabolism. Therefore, DPA is widely used as food and feed additive as well as cosmetic and pharmaceutical ingredient (Wang et al., 2022). Currently, the industrialized production of DPA is mainly through the condensation reaction of d-(−)-pantolactone (DPL) and β-alanine. d-(−)-Pantolactone, also called (R)-pantolactone, is a key chiral intermediate for the synthesis of DPA. To date, considerable efforts have been devoted to the racemic resolution of D,L-PL for producing DPL, as exemplified by the chemical resolution method, lactonase-catalyzed enantioselective hydrolysis resolution method and as well as, oxidoreductase-catalyzed enantioselective dehydrogenation and reduction resolution route B (Zhu et al., 2022). Admittedly, these routes achieved decent DPL productivity with high enantiomeric purity. However, chemical synthetic route was restricted to the use of dangerous reagents, which does not match the requirements of “green chemistry”. d-Lactonohydrolase-catalyzed route only possess 50% maximum theoretical yield (or racemization and reuse of l-(+)-pantolactone (LPL) could increase the yield), and laborious and time-consuming separation of d-(−)-pantoic acid, acidification and re-esterification steps were required, which are unfavorable for industrialized production of DPL. Using oxidoreductase resolution involves two approximative routes (Fig. 1): (A) the dehydrogenation intermediate of LPL is asymmetrically reduced to DPL; (B) the dehydrogenation intermediate ketopantoyl lactone (KPL) spontaneously hydrolyzed to ketopantoic acid, followed by asymmetric reduction, acidification and re-esterification. Obviously, route A has fewer steps and more simple process. Although enzymatic resolution based on route A has important application prospect due to the advantages of being environmentally friendly and 100% theoretical yield, it is still hindered in oxidation of LPL in the first step catalyzed by membrane-bound dehydrogenase. Moreover, compared with the thorough study for the second reduction of KPL to DPL catalyzed by conjugated polyketone reductase (Cheng et al., 2019, Wang et al., 2019, Zhao et al., 2017), extracellular application of dehydrogenases involved in the first step is insufficient due to limited enzyme resources. Hence, mining novel enzymes or modifying the already obtained enzymes enable the dehydrogenation of LPL and exploring their potentials in bioconversion is crucial for DPL synthesis from the racemic mixture.

l-Pantolactone dehydrogenase (LPLDH) (EC 1.1.99.27) mediates stereospecific dehydrogenation of LPL using unknown electron acceptor (Kataoka et al., 1992, Si et al., 2012), belonging to FMN-dependent α-hydroxy acid oxidase/dehydrogenase family (Kean and Karplus, 2019, Liu et al., 2018), which are widely found in eukaryotes and prokaryotes. The members of this family share a high sequence similarity (35%-40%) and their protein expression can vary from cytoplasm to cytomembrane to organelle (Gillet et al., 2016, Xu and Mitra, 1999). The catalytic reaction of the α-hydroxy acid oxidase/dehydrogenase family follows a “ping-pong” bi-bi mechanism comprising two half-reactions (Hiraka et al., 2020, Romero and Gadda, 2014). It includes that substrate α-hydroxy acids are oxidized to oxo-acids with flavin reduction in the reductive half-reaction, and the electron acceptor acquires the electron transfer from the reduced flavin regenerating the oxidized flavin in the oxidative half-reaction. To date, only two representative LPLDHs from Nocardia asteroids (Kataoka et al., 1992) and Rhodococcus erythropolis (Si et al., 2012) have been studied. LPLDH from N. asteroids was merely purified and characterized from the wild-type strain. Overexpression of LPLDH from R. erythropolis in E. coli was verified to generate a large amount of inactive proteins. Therefore, efficient expression of highly active LPLDHs in heterologous hosts such as E. coli is required.

It is known that LPLDH is a membrane-bound enzyme, but the location and distribution of its heterologous expression are blurred. To efficiently express LPLDH in E. coli, it requires intensive insight of membrane protein (MP) expression. MPs, as the main carriers of biomembrane functions, are crucial for cell proliferation and differentiation, energy conversion, signal recognition and transduction, and material transportation (Birch et al., 2018, Hegde and Keenan, 2022, Shimizu et al., 2018). E. coli is typically used host for heterologous production of proteins from both eukaryotes and prokaryotes origin (Dilworth et al., 2018). Adversely, heterologous production of MPs is generally difficult, mainly due to (1) only little MPs can be correctly folded and bind to membrane, (2) overexpressed MPs have a strong propensity to aggregate and form inclusion bodies (IBs) and it is impossible to predict whether a specific MP is ultimately on the cytoplasmic membrane or in IBs (Drew et al., 2001), and (3) overexpression of MPs cause cytotoxicity, in turn, resulting in low biomass and yield (Satheeshkumar et al., 2018). Nevertheless, producing recombinant proteins in soluble form and averting forming biologically inactive IBs are highly desirable. A membrane-bound mandelate dehydrogenase from Pseudomonas putida was rendered soluble in E. coli host by replacing its membrane-anchoring region with an analogous segment in soluble glycolate oxidase (Sukumar et al., 2018, Xu and Mitra, 1999). The extensive formation of IBs is another challenge for overexpressing MPs. Misfolding/aggregation of membrane proteins produces IBs as a result of saturation of Sec translocon, which is a protein-conducting channel in membrane protein biogenesis process mediated by the signal recognition particle/Sec translocon/YidC pathway (Hegde and Keenan, 2022, Klepsch et al., 2011, Zhang and Miller, 2012). A variety of strategies were adopted to minimize IBs expression by tailoring culture conditions, modifying proteins of interest, altering expression vectors, engineering host, and so on (Bhatwa et al., 2021, Kaur et al., 2018, Schlegel et al., 2010). Therefore, in order to efficiently produce LPLDH, the issues on insolubility and IBs of heterologous expression in E. coli need to be further investigated.

In this present study, a panel of candidates for l-pantolactone dehydrogenase were heterologously expressed in E. coli. The heterologous expression localization and distribution of this membrane-bound protein were investigated for explicating low solubility. Then, combinatorial co-expression with molecular chaperone plasmid pGro7 and low-temperature induction strategy remitted IBs formation. The soluble RhoLPLDH was isolated from chaperone proteins through Ni-chelating affinity chromatography. The catalytic activity of RhoLPLDH toward the substrate LPL was further reinforced by directed evolution focusing on substrate docking pocket and hydrophilicity-substitution towards residues of loop 4. A highly efficient conversion for 1 M LPL to KPL with 97.8 g L−1 d−1 space-time yield was accomplished using the evolved mutant E. coli BL21 (DE3)/pGro7/RhoLPLDHL254I/V241I/I156L/F224Q/N164K (CM5) as biocatalyst. Finally, the strain CM5 combined with a strain coexpressing conjugated polyketone reductase and glucose dehydrogenase for one-pot synthesis of DPL in the deracemization of D,L-PL obtaining 91.2% yield. This study paved a foundation for the up-scale production of DPL.