Human milk oligosaccharides (HMOs) are the third most abundant solid components in human milk and are known to comprise more than 200 different kinds of oligosaccharides (Bode, 2015, Vandenplas et al., 2018). They are unique to human milk and have various biological functions in the nutrition and health of infants. The roles of HMOs in health promotion and immune system development are diverse and remain under investigation. They function as prebiotics to stimulate the growth of beneficial gut microorganisms (Barile and Rastall, 2013) and help enhance immunity by preventing the attachment of pathogens and inhibiting pathogenic infections in the intestinal mucosa (Kunz et al., 2000). Under these circumstances, HMOs are high-value-added substances recently developed for food ingredients, cosmetics (Liu et al., 2018), and pharmaceutical applications (Chin et al., 2017). Fucosylated oligosaccharides comprise approximately 30% of the HMOs, and 2'- fucosyllactose (2’-FL) is one of the most abundant (Eiwegger et al., 2004). Hence, large-scale and high-yield 2'-FL production is of interest.

To produce 2’-FL, the following methods are generally used: i) separation from human breast milk, ii) organic synthesis using chemical methods, and iii) in vitro enzymatic synthesis or in vivo microbial synthesis using glycosyltransferases (GTs) (Albermann et al., 2001). Isolating 2’-FL from human breast milk for large-scale production is impractical, whereas the complexity of organic chemical synthesis, involving several protection and deprotection steps, leads to low production titers and yields. Additionally, the in vitro bio-reaction systems (enzymatic methods) for producing 2’-FL should overcome the challenges to the high cost of supplying GDP-L-fucose and the issues with low solubility of α-1,2-fucosyltransferase (α1,2-FucT), which complicates the purification of the enzyme (Baumgartner et al., 2013, Han et al., 2012). Therefore, various strains, notably Escherichia coli, Saccharomyces cerevisiae, and Bacillus subtilis, have been developed to synthesize 2'-FL in vivo, facilitated by the enzyme α1,2-FucT. (Deng et al., 2019, Hollands et al., 2019, Lee et al., 2020, Liu et al., 2018, Yu et al., 2018b). This enzyme catalyzes the transfer of an L-fucose moiety from the donor substrate, GDP-L-fucose, to the acceptor sugar substrate, D-lactose, facilitating the production of 2'-FL (Parschat et al., 2020). Scheme 1.

Not only employing a different platform host strain but also various metabolic engineering strategies have been developed to increase the titer of 2'-FL in the cells (Chin et al., 2017, Ji et al., 2022, Lee et al., 2023, Park et al., 2022, Xu et al., 2021, Yu et al., 2018c, Zhang et al., 2022). In optimizing 2'-FL production, three main bottlenecks are often encountered: lactose utilization, the availability of GDP-L-fucose, and the solubility of α1,2-FucT. The uptake and utilization of lactose are pivotal since lactose acts as the acceptor substrate. High glucose and lactose concentrations in the medium can lead to catabolite repression, limiting lactose availability and, by extension, 2'-FL production. Strategies to overcome this include manipulating lactose flux to avoid repression, such as the inactivation or deletion of the lacZ gene in BL21 (DE3) strains (Chin et al., 2015, Huang et al., 2017, Li et al., 2021) and converting the native lac promoter in E. coli (Plac) to a Ptac promoter to enhance lactose uptake even in high glucose conditions (Park et al., 2022). For the donor substrate GDP-L-fucose, enhancing its intracellular pool is challenging due to its minimal natural production (Wan et al., 2020). It can be synthesized via the de novo and salvage pathways from D-glucose and L-fucose, respectively (Fig. 1) (Baumgartner et al., 2013). The most common approach is to overexpress the genes encoding enzymes related to the de novo pathway, such as manB, manC, gmd, and wcaG (Byun et al., 2007, Lee et al., 2009) and, for the salvage pathway, deleting fucI and fucK to conserve L-fucose, alongside adding L-fucose to the medium for strains overexpressing fkp to convert L-fucose into GDP-L-fucose (Liu et al., 2011, Zhai et al., 2015).

Despite such various engineering strategies to supply enough acceptor and donor substrates for 2’-FL synthesis, the industrial production of 2’-FL synthesis is often limited by the low solubility of α1,2-FucT and its low catalytic efficiency, i.e., kcat/Km (Yu et al., 2018a). Many investigations have been conducted to increase the solubility of α1,2-FucT using solubility-enhancing fusion tags (Albermann et al., 2001, Lee et al., 2023, Li et al., 2008a, Li et al., 2008b, Yi et al., 2005, Zhao et al., 2016, Zhong et al., 2021). Efforts to enhance α1,2-FucT solubility did not significantly boost 2’-FL production, possibly due to unaddressed rate-limiting steps and reduced enzyme activity from fusion tag alterations. Other strategies like growth optimization and molecular chaperone co-expression also failed to markedly improve 2’-FL yields, indicating solubility enhancement alone may not be sufficient (Hayhurst and Harris, 1999, Paraskevopoulou and Falcone, 2018).

In general, to enhance target protein stability and solubility itself, the following strategies are commonly used: i) replacement of non-conserved amino acids by conserved amino acids through consensus finder (Lee et al., 2023, Park et al., 2022), ii) exchanging wild type amino acids with the amino acids following α-helix rule and hydropathy contradiction rule (Shinoda et al., 2022), and iii) exchanging wild type amino acids with the amino acids using thermo-stability prediction by RISLNet (Upadhyay et al., 2019). To overcome this solubility issue in the case of α1,2-FucT (FutC) from Helicobacter pylori, we primarily identified specific individual sites increasing FutC solubility based upon α-helix rule and the hydropathy contradiction rule suggested by Asano group (Matsui et al., 2017), and subsequently performed their combinatorial site-saturation mutagenesis (SSM). First, α-helix rule was applied to identify aggregation hotspots in FutC, and SSM using degenerated codons was subjected to two or three hydrophobic amino acids present in hydrophilic surfaces on each α-helix. To perform high-throughput screening of enhanced solubility mutants from the saturated mutation library, the split-GFP system was used (Cabantous et al., 2005, Cabantous and Waldo, 2006).

Split-GFP consists of two fragments containing a GFP1-10 detector fragment and a 15-amino-acids GFP11 fragment (Cabantous and Waldo, 2006), which can spontaneously combine to form a complete GFP and emit fluorescence (Kakimoto et al., 2018). As the GFP11 fragment is very small, it does not significantly affect the solubility or folding of the tagged target protein. Therefore, when target proteins become more soluble and form correct protein folds, the GFP11-target protein conjugates the correct fluorescent GFP protein formation, whereas the GFP1-10 detector fragment does not fluoresce by itself. Each mutant library was screened using a fluorescence-activated cell sorter (FACS), and fluorescence spectrometry for the second screening. Through this approach, we discovered four mutation sites that could increase solubility and demonstrated that their combinations could further enhance the solubility of FutC. Throughout this study, we have proved that the solubility of FutC becomes a real rate-determining step for 2’-FL synthesis, so that enhancing its solubility indeed increased the 2’-FL production by 3.4-folds. The results of this study show the possibility of improving the production of 2’-FL in E. coli by solely increasing the solubility of FutC.