3′-Sialyllactose (3′-SL, Neu5Acα2,3Galβ1,4Glc), is an abundant and important sialylated human milk oligosaccharides (HMOs), with Neu5Ac linked to galactose of lactose by α2,3-glycoside bond [1]. 3′-SL accounts for approximately 2.05 % of total HMOs in molar ratio [2]. Specifically, in colostrum (0–5 d) and mature milk (15–90 d), the mean concentrations of 3′-SL are both 0.19 g/L, while in transitional (6–14 d) and late milk (> 90 d), the values are both 0.13 g/L [2]. 3′-SL exhibits promising prebiotic effects [3] and several other health effects, such as antiadhesive, antibacterial, and antiviral activity [[4], [5], [6]], necrotizing enterocolitis prevention [7], immunomodulatory [8,9], intestinal epithelial cell response regulation [10], and brain development promotion and cognition enhancement [11]. Collectively, 3′-SL has various beneficial effects, and it has received widespread attention.

Biotechnological approaches to both 3′-SL and 6′-SL have been systematically and elaborately discussed in our recent review [1]. Among these approaches, sialyltransferases (SiaTs) are the most important and pivotal element for the high-level synthesis of 3′-SL. Schelch et al. summarized bacterial SiaTs, their structure and functional properties, and their use in enzymatic cascades to synthesize sialo-oligosaccharides [12]. Priem et al. constructed a biosynthetic route of 3’-SL by heterologously overexpressing nst and N. meningitidis CMP-Neu5Ac synthetase-encoding gene in engineered Escherichia coli JM107 [13]. Another alternative synthetic pathway for 3’-SL biosynthesis was constructed by heterologously overexpressing nst and Campylobacter jejuni gene cluster (neuBCA), which encodes UDP-GlcNAc 2-epimerase (NeuC), Neu5Ac synthase (NeuB), and CMP-Neu5Ac synthetase (NeuA), respectively, in engineered E. coli DH1 [14].

The gene nst is involved in lipooligosaccharide biosynthesis in N. meningitidis and was first cloned in E. coli by Gilbert et al. [15]. In 1997, Gilbert et al. purified recombinant NST, tested its enzyme activity, characterized its enzymatic properties, and confirmed the effective trans-sialylation activity toward lactose generating 3′-SL [16]. 3′-SL was efficiently enzymatically produced from the substrates including sialic acid, lactose, phosphoenolpyruvate, and catalytic amounts of ATP and CMP, by NST and N. meningitidis CMP-Neu5Ac [17]. Although NST has been studied extensively, its crystal structure has not yet been elucidated. But the crystal structure of N. meningitides 126E SiaT (126NST) has been resolved (PDB code 2YK4) [18]. Sequence alignment indicated that the amino acid sequence of 126NST had six differences from the NST sequence (E40D, R102W, S129A, G168I, T242A, and K273N). Small differences on amino acids led to a high-quality template model for potential molecular modification of NST.

Recently, a series of genetically engineered strains for biosynthesis of 3′-SL were developed. In 2022, Zhang et al. introduced Campylobacter jejuni neuBCA genes involved in CMP-Neu5Ac biosynthesis, along with several genes encoding α2,3-SiaTs, such as NST, Pasteurella multocida Pm70 α2,3-SiaT (Pm0188), Vibrio sp. JT-FAJ-16 α2,3-SiaT (Vs16), and Photobacterium sp. JT-ISH-224 α2,3-SiaT (Ps224) into EZAK (E. coli BL21(DE3)ΔlacZΔnanAΔnanT), respectively. The final strain ES2 yielded 23.1 g/L 3′-SL, extracellularly [19]. Similarly, in 2024, Li et al. introduced a precursor CMP-Neu5Ac synthesis pathway and high-performance α2,3-SiaT genes. They also optimized the expression of glmS-glmM-glmU, inactivated competitive pathway genes and enhanced catalytic performance of the rate-limiting enzyme α2,3-SiaT by RBS screening and protein tag cloning. The final strain yielded total 44.2 g/L 3′-SL in a 3 L bioreactor [20].

Obviously, the majority of research efforts are directed toward enhancing the production of 3′-SL through the modulation of metabolic gene expression levels, the identification of efficient SiaT, and the inhibition of competing pathways. However, to achieve large-scale production of pure 3′-SL, downstream purification must adhere to more stringent requirements beyond efficient in vivo synthesis rates and yields. Another challenge identified in this study in the bioproduction of 3′-SL is the residue of multiple unknown by-products accompanying the production process, which may cause difficulties during the subsequent downstream separation and purification processes.

This study aims to mitigate the formation of byproducts and enhance the synthesis of 3′-SL through the application of protein engineering and metabolic engineering methods, thereby averting cost escalation resulting from excessive accumulation of byproducts. Initially, we compared the fermentation performance of four strains, each expressing different exogenous α2,3-SiaTs, NST, Vs16, Ps224 and BtST1 (from Bibersteinia trehalosi USDA-ARS-USMARC-192). Among these α2,3-SiaTs, NST-based strain produced the least byproducts was selected for subsequent molecular modifications. Based on the crystal structure of 126NST, computer-assisted modification with the help of the Hot-Spot Wizard 3.1 server were designed. Saturation mutagenesis and multipoint mutagenesis from beneficial residues also were experimentally validated. The best mutant was integrated into the genome to improve sialylation. Finally, 5-L bioreactor fermentations were conducted to demonstrate the large-scale biosynthesis of the beneficial variant strains. The final strain EZAKH41 improved the titer of 3′-SL to 32.1 g/L, extracellularly. Overall, we have effectively reduced the production of by-products and attained elevated levels of 3′-SL.