A Comprehensive Review of Fucosylated Human Milk Oligosaccharides: Biological Functions, Production Technologies, and Safety Regulation With Commercial Translation in Infant Nutrition

ABSTRACT

Human milk oligosaccharides (HMOs) are important functional bioactive components of human milk. Fucosylated HMOs (FHMOs) represent a major subclass, accounting for approximately 35%–50% of total HMOs, and have attracted increasing attention in infant nutrition and health. Current evidence indicates that FHMOs play important roles in shaping the infant gut microbiota, modulating host–microbe interactions, supporting intestinal barrier function, and maintaining immune and metabolic homeostasis. However, their commercialization still faces several challenges, including high production costs, difficulties in the scalable manufacture of structurally complex members, uneven functional evidence across individual compounds, incomplete safety and regulatory evaluation frameworks, and the need to further validate processing stability and application performance in final products. From an industrial perspective, microbial fermentation has become the main production route for 2′‐fucosyllactose and 3‐fucosyllactose, whereas the large‐scale production of more structurally complex or low‐abundance FHMOs remains technically and economically challenging. Regulatory approvals in major markets indicate acceptable safety at authorized use levels, but uncertainties remain regarding long‐term intake, vulnerable infant populations, matrix‐dependent stability, and the combined use of multiple HMOs. This review summarizes recent advances in the structural characteristics, biological functions, production technologies, safety assessment, regulatory status, and commercial applications of FHMOs, with emphasis on 2′‐fucosyllactose, 3‐fucosyllactose, and selected structurally related members. It also discusses commercialization challenges and future research priorities to support their responsible application in infant nutrition.

1. Introduction

Human milk is a complex natural nutritional matrix and the ideal source of nutrition for newborns (Kalluri et al. 2025; Moore et al. 2025; Zhernakova et al. 2025). It is widely regarded as the gold standard for infant nutrition during the first 2 years of life (Geniselli da Silva et al. 2025). Human milk oligosaccharides (HMOs) are the third most abundant solid component in human milk after lactose and lipids, and they exert diverse biological effects that are essential for infant growth, development, and physiological health (Del Compare et al. 2025; Song et al. 2026; J. Xu et al. 2025; J. Yu et al. 2025). Based on structural differences, HMOs can be broadly classified into three groups: fucosylated HMOs (FHMOs), neutral non‐FHMOs, and acidic sialylated HMOs (J. Xu et al. 2025). Among them, FHMOs represent a major subclass, accounting for approximately 35%–50% of total HMOs, although their abundance and composition vary markedly among mothers (Y. Zhu et al. 2026). Maternal secretor and Lewis phenotypes, determined mainly by FUT2 and FUT3 activity, are major determinants of these profiles. Secretor‐positive milk generally contains abundant α1,2‐FHMOs, such as 2′‐FL and LNFP I, whereas these structures are absent or markedly reduced in nonsecretor milk; Lewis status further affects α1,3/4‐fucosylated structures, including LNFP II (J. Xu et al. 2025). Because these compounds are resistant to digestion in the upper gastrointestinal tract, they can reach the colon, where they participate in host–microbe interactions, selectively modulate microbial communities, and interfere with pathogen adhesion by mimicking host glycans (Q. Chen, Ma, et al. 2024; Yao et al. 2024). Together, these microbiota‐modulating and anti‐adhesive properties may contribute to infant physiological health, although the strength of evidence differs among individual FHMOs and health outcomes (Masi and Stewart 2022; R. P. Singh et al. 2022; Spicer et al. 2022). Because FHMOs constitute a major HMO subclass and have been associated with diverse biological activities, they have become important targets for the development of human milk‐inspired infant formulas and other early‐life nutritional products (W. Li, Wang, et al. 2021; Mills et al. 2023).

Although observational, clinical, and preclinical evidence suggests that selected FHMOs may exert prebiotic effects and have commercial potential, the magnitude and consistency of these effects vary with the individual compound, dose, baseline microbiota, and study model, and several issues remain to be clarified before their use as supplements in infant formulas for this sensitive population (Del Compare et al. 2025; L. Zhu, Li, et al. 2023). First, the quality and quantity of available evidence are uneven across different compounds. 2′‐Fucosyllactose (2′‐FL) and 3‐fucosyllactose (3‐FL) have been the most extensively studied, whereas more structurally complex compounds, such as lacto‐N‐fucopentaoses (LNFPs) and lacto‐N‐difucohexaoses (LNDFHs), remain insufficiently investigated (Urashima et al. 2025; J. Xu et al. 2025). Second, many current mechanistic findings are derived from in vitro systems or animal models and should therefore be interpreted with caution when extrapolated to infants (Q. Chen et al. 2025; Wallingford et al. 2022; J. Xu et al. 2025). On the one hand, several physiological systems in humans, including the nervous and immune systems, are still immature and highly sensitive during infancy (Del Compare et al. 2025; X. Wang, Li, et al. 2024). FHMOs may act through multiple regulatory axes, but their activation thresholds, dose‒response relationships, specific pathways, key regulatory factors, and patterns of action under different physiological and pathological conditions have not yet been systematically investigated (Song et al. 2026; J. Xu et al. 2025). On the other hand, as nutritional components that are not directly absorbed or metabolized by the human host, the biological effects of FHMOs largely depend on host–microbiota interactions (Q. Chen, Ma, et al. 2024; Y. Chen, Chen, et al. 2024; Y. Chen, Wen, et al. 2024). However, the infant gut microbiota is still undergoing dynamic succession, and its composition and developmental status can differ markedly among infant populations, such as preterm and full‐term infants, as well as low birth weight and macrosomic infants (Pang et al. 2025; Sindi et al. 2025; Whittaker et al. 2025; N. Wu, Wang, et al. 2025). These differences may lead to substantially different beneficial outcomes after the interaction between FHMOs and the gut microbiota (Hu et al. 2023). These uncertainties highlight the need for long‐term, well‐controlled human studies and standardized endpoints relevant to infant nutrition.

Beyond biological activity, the translation of FHMOs into foods depends on scalable manufacturing, consistent product purity and performance in complex food matrices. Chemical and enzymatic synthesis provide structural precision, whereas synthetic biology‐enabled microbial fermentation is currently the predominant scalable route for abundant compounds such as 2′‐FL and 3‐FL. More complex FHMOs remain constrained by pathway efficiency, by‐product formation, downstream purification, and analytical validation, while plant‐based systems are still exploratory (Mills et al. 2023; Zheng et al. 2022; Y. Zhu, Cao, 2022; Y. Zhu et al. 2026). Regulatory translation is likewise ingredient‐, process‐, use‐, and jurisdiction‐specific. Safety evaluation must consider product identity and purity, the production organism and manufacturing process, intended exposure, target population, and stability in the final food matrix. Commercial availability should therefore not be interpreted as equivalent to human milk or as class‐wide approval of FHMOs (Donovan et al. 2023; Wichmann et al. 2026).

Recent reviews have provided valuable summaries of the structural diversity, biological functions, biosynthetic strategies, and industrial development of FHMOs (M. Li, Yao, et al. 2026; Y. Yang et al. 2025). Nevertheless, the rapid expansion of functional, manufacturing, safety, and application studies has created a need for a critical synthesis that evaluates not only what has been reported, but also the strength, attribution, consistency, and translational relevance of the available evidence. Accordingly, the present review provides a comprehensive and multidimensional evaluation of FHMOs, including 2′‐FL and 3‐FL, from biological evidence to responsible application in infant nutrition. First, it distinguishes mechanistic observations from preclinical findings, human observational associations, clinical safety and tolerance data and demonstrates clinical benefits, while separating evidence obtained using purified individual FHMOs from that derived from HMO mixtures, pooled human‐milk HMOs, synbiotics, or multicomponent formulas. Second, it links structure‐specific biological evidence with manufacturing readiness by evaluating production performance together with enzyme and strain characteristics, product selectivity, impurity profiles, downstream purification, and scale‐up constraints. Third, it considers safety and regulatory conclusions at the level of defined preparations, production organisms, manufacturing processes, specifications, intended uses, and target populations, while distinguishing regulatory authorization from evidence supporting clinical efficacy. Finally, it extends the analysis to processing stability and product quality in infant‐formula matrices, including thermal treatment, spray drying, storage, moisture sensitivity, and potential Maillard reaction‐related changes. By integrating these dimensions, this review aims to distinguish well‐supported conclusions from emerging hypotheses, identify compound‐ and application‐specific evidence gaps, and provide a scientific framework for future research, product development, safety assessment, and regulatory decision‐making. Figure 1 summarizes the scope of this review.

FIGURE 1.

Overview of the biological functions, production technologies, safety assessment, and commercial applications of FHMOs in infant nutrition.

2. Structural Diversity of FHMOs

HMOs are structurally diverse unconjugated glycans composed mainly of five monosaccharide units: d‐glucose, d‐galactose, N‐acetyl‐d‐glucosamine, l‐fucose, and N‐acetylneuraminic acid (J. Xu et al. 2025). Most HMOs contain lactose at the reducing end and are extended by Type I or Type II N‐acetyllactosamine units, followed by varying degrees of fucosylation and/or sialylation (J. Xu et al. 2025). Although HMOs have traditionally been defined as oligosaccharides with a limited degree of polymerization, advances in mass spectrometry, chromatography, and glycomics have substantially expanded our understanding of their structural diversity (Navarro‐Esteve et al. 2025; J. Tan et al. 2022). To date, more than 200 distinct HMO structures have been reported, but only a relatively small proportion have been subjected to detailed quantitative analysis, biological evaluation, or food application development (J. Xu et al. 2025; J. Yu et al. 2025). On an approximate compositional basis, neutral fucosylated, neutral non‐fucosylated, and acidic/sialylated HMOs account for approximately 35%–50%, 42%–55%, and 12%–14% of total HMOs, respectively, although these ranges vary with maternal secretor and Lewis status, lactation stage, population, and analytical method.

FHMOs are characterized by the presence of one or more l‐fucose residues linked through α1‐2, α1‐3, or α1‐4 glycosidic bonds. These linkage patterns generate considerable structural diversity and are closely associated with maternal secretor and Lewis blood group status (Song et al. 2026; J. Xu et al. 2025). Among them, 2′‐FL and 3‐FL are the most representative and extensively studied trisaccharide isomers. Both are composed of fucose, galactose, and glucose and account for approximately 31% (Bych et al. 2019) and 5% (Huang et al. 2024) of total HMOs, respectively. In 2′‐FL, the fucose residue is attached to the galactose moiety of lactose through an α1‐2 glycosidic bond, whereas in 3‐FL, fucose is attached to the glucose moiety through an α1‐3 glycosidic bond (J. Xu et al. 2025). This positional difference is chemically subtle but biologically important, as it may affect molecular conformation, recognition by microbial enzymes or transporters, resistance to degradation, and interactions with host or pathogen lectins (Del Compare et al. 2025; Song et al. 2026; Tonon et al. 2024). Therefore, in infant formula design, 3‐FL should be regarded as a structurally related but functionally distinct compound rather than a simple substitute for 2′‐FL.

More structurally complex FHMOs arise from the combination of Type I LNT or Type II LNnT cores with linkage‐specific fucosylation. LNFP I and LNFP II are Type I structures carrying α1,2‐linked fucose on the terminal galactose and α1,4‐linked fucose on GlcNAc, respectively, whereas LNFP III contains α1,3‐linked fucose on GlcNAc of a Type II core. LNFP V and LNFP VI carry α1,3‐linked fucose on the reducing‐end glucose of Type I and Type II cores, respectively. The addition of a second fucose generates LNDFHs: LNDFH I combines α1,2‐fucosylation of the terminal galactose with α1,4‐fucosylation of GlcNAc, whereas LNDFH II combines α1,4‐fucosylation of GlcNAc with α1,3‐fucosylation of the reducing‐end glucose. The DFLNH a‐c structures are branched, difucosylated higher HMOs that differ in branch topology and the positions of the two fucose residues. These compounds are included in Figure 2 to illustrate how core type, branching, fucose number, and linkage position generate distinct FHMO isomers; their inclusion does not imply equivalent abundance, strength of biological evidence, or manufacturing maturity. Indeed, their biological functions remain much less characterized than those of 2′‐FL and 3‐FL (Song et al. 2026; J. Xu et al. 2025).

FIGURE 2.

Compositional features, monosaccharide building blocks, and representative structural diversity of HMOs. (A) Major nutritional components of human milk, the proportion of HMOs among total nutrients, major structural classes of HMOs, and the distribution of representative HMO species. (B) The five principal monosaccharide building blocks of HMOs and their symbolic representations, including d‐glucose, d‐galactose, N‐acetylglucosamine, l‐fucose, and N‐acetylneuraminic acid. (C) Representative structures of FHMOs: 2′‐fucosyllactose (2′‐FL), 3‐fucosyllactose (3‐FL), lacto‐N‐fucopentaose I (LNFPH‐I), LNFP‐II, LNFP‐III, LNFP‐V, LNFP‐VI, lacto‐N‐difucosylhexose I (LNDFH‐I), LNDFH‐II, difucosyllacto‐N‐hexaose a (DFLNHa), difucosyllacto‐N‐hexaose b (DFLNHb), and difucosyllacto‐N‐hexaose c (DFLNHc). (D) Representative structures of neutral non‐FHMOs: lacto‐N‐triaose II (LNT2), lacto‐N‐tetraose (LNT), lacto‐N‐neotetraose (LNnT), lacto‐N‐neohexaose (LNnH), lacto‐N‐hexaose (LNH), and lacto‐N‐neooctaose (LNnO). (E) Representative structures of acidic sialylated HMOs: 3′‐sialyllactose (3′‐SL), 6′‐sialyllactose (6′‐SL), sialyllacto‐N‐tetrasaccharide a (LST a), LST b, LST c, and disialyl‐lacto‐N‐tetraose (DSLNT).

From the perspective of food science and production, the number and position of fucose residues have practical implications beyond structural diversity and biological differences (J. Xu et al. 2025). They influence the choice of chemical, enzymatic, or microbial synthesis strategies; the requirement for specific fucosyltransferases or nucleotide‐sugar precursors; the complexity of downstream purification; and the analytical methods needed for structural identification and purity confirmation (Haselberger et al. 2023; Navarro‐Esteve et al. 2025; J. Tan et al. 2022). For example, isomeric structures such as 2′‐FL and 3‐FL require reliable chromatographic or mass spectrometric discrimination, whereas more highly substituted FHMOs require more advanced structural characterization (Haselberger et al. 2023; Navarro‐Esteve et al. 2025; J. Tan et al. 2022). These analytical requirements are essential for regulatory approval, quality control, and reliable comparison across biological studies.

Overall, FHMOs should not be regarded as a homogeneous class with a single function. Their biological and technological properties are influenced by multiple factors, including glycosidic linkages, backbone structure, degree of fucosylation, abundance in human milk, microbial accessibility, and feasibility of industrial production (J. Xu et al. 2025; J. Yu et al. 2025). Therefore, a structure‐based framework is needed to evaluate their potential applications in infant nutrition (Urashima et al. 2025; J. Xu et al. 2025). Such a framework can help distinguish well‐characterized ingredients, such as 2′‐FL and 3‐FL, from emerging candidates, such as LNFPs and LNDFHs, and guide future research on safety, efficacy, and formulation design (Del Compare et al. 2025; Koirala et al. 2024). With continued advances in analytical technologies, more low‐abundance but functionally important FHMOs are expected to be identified. Elucidating their structure‒function relationships will support the development of precision nutrition strategies and open new avenues for improving infant health (Song et al. 2026; J. Xu et al. 2025; Y. Zhu et al. 2026).

3. Diversified Biological Functions

Most orally ingested FHMOs resist digestion in the upper gastrointestinal tract and reach the colon, where they can be utilized by specific microorganisms or participate in microbial cross‐feeding networks. Their fermentation may generate acetate, lactate, other short‐chain fatty acids, and fucose‐derived 1,2‐propanediol (Song et al. 2026; J. Xu et al. 2025). However, the available biological evidence is highly uneven across individual compounds. Most studies have examined 2′‐FL, fewer have investigated 3‐FL, and evidence for LNFPs, LNDFHs, and other structurally complex FHMOs remains limited (Song et al. 2026; J. Xu et al. 2025). The results obtained using pooled human‐milk HMOs or multicomponent formulas are not attributed to an individual FHMO unless supported by compound‐specific evidence. Current evidence suggests that 2′‐FL, and in some cases 3‐FL, may influence gut microbial metabolism, epithelial barrier‐related pathways, and selected immune or neurodevelopment‐related endpoints (Q. Chen, Ma, et al. 2024; Y. Chen, Wen, et al. 2024; Song et al. 2026; J. Xu et al. 2025). Nevertheless, evidence concerning necrotizing enterocolitis, neurodevelopment, systemic immune responses, and allergy‐related outcomes is derived predominantly from mechanistic, preclinical, or observational studies (Shenhav et al. 2024; J. Xu et al. 2025; Zhernakova et al. 2025).

3.1. Modulation of Gut Microbiota

The development of the early‐life gut microbiota is influenced by delivery mode, gestational age, antibiotic exposure, feeding pattern, maternal microbiota, and environmental factors (Cani 2018). During breastfeeding, HMOs exert selective ecological pressure and support the establishment of an infant‐adapted microbial community, particularly one enriched in bifidobacteria (de Muinck and Trosvik 2018). FHMOs contribute to this process by serving as substrates for specific microbial taxa and by participating in cross‐feeding networks (Guo et al. 2026; Shan et al. 2025). The strongest evidence currently comes from 2′‐FL and 3‐FL. In vitro fermentation systems, infant gut simulation models, and animal studies have shown that, under specific conditions, 2′‐FL has been associated with changes in microbial community composition, including increases in selected HMO‐utilizing or infant‐associated taxa and decreases in taxa associated with dysbiosis in the corresponding study models (Feng et al. 2025; Kong et al. 2021; S. Zhang, Chen, 2023). However, the direction and magnitude of these effects vary across studies because of differences in inoculum source, host species, diet, intervention duration, intake level, sequencing depth, and analytical methods (Y. Luo, Xia, et al. 2025; Sato et al. 2025). Therefore, although 2′‐FL is generally considered beneficial for the gut microbiota, its effects should not be regarded as universally applicable or independent of ecological context (Mills et al. 2023). Compared with 2′‐FL, evidence on the effects of 3‐FL on the gut microbiota is more model‐dependent (Hao et al. 2025; Y.‐J. Kim et al. 2023). In pooled infant fecal fermentation experiments, 3‐FL has been associated with temporal changes in Actinobacteria and Bacteroidetes and with altered relative abundances of taxa such as Bacteroides and Enterococcus (Kong et al. 2021). These observations should be regarded as exploratory because pooling fecal inocula may obscure interindividual differences in microbial composition and HMO utilization. Neonatal piglet studies have also reported changes in Blautia, Prevotella, Eubacterium, and other anaerobic taxa following 3‐FL supplementation (Pitt et al. 2024). However, differences between piglet and infant microbiota limit direct extrapolation. Therefore, the currently reported taxonomic changes should not be considered a characteristic or reproducible microbial signature of 3‐FL.

These findings suggest that 3‐FL may influence microbial communities through mechanisms different from those of 2′‐FL, but the current evidence remains insufficient to define reproducible microbial signatures. A key limitation is that taxonomic changes are often used as surrogate indicators of functional effects. Increases in selected infant‐associated or HMO‐utilizing taxa, including bifidobacteria and lactobacilli, may be biologically relevant; however, their functional interpretation depends on species‐ or strain‐level identity and cannot be inferred from genus‐level abundance alone (Fu et al. 2024; Y. Luo, Li, et al. 2025). Functional interpretation requires supporting evidence from metabolite profiling, carbohydrate utilization pathways, short‐chain fatty acid production, barrier‐related markers, immune endpoints, and well‐designed human studies (S. Zhang, Chen, et al. 2023). Table 1 summarizes reported microbiota‐related findings for defined FHMO and FHMO‐containing combinations, organized according to intervention type, study model, and level of translational relevance.

TABLE 1.

Effects of defined FHMO and FHMO‐containing combinations on gut microbiota across study models and levels of translational relevance.

Intervention	Study model and evidence level	Reported microbiota‐related findings	Other reported outcomes	Evidence interpretation and attribution	References
(1) Defined individual fucosylated‐HMO interventions
2′‐FL	Host‐free human colonic model inoculated with representative infant gut bacteria	Altered the abundance of representative infant gut bacteria; increased Parabacteroides distasonis and Lactobacillus acidophilus, while reducing Clostridium perfringens in specific colon regions	Changed free fatty acid profiles and extracellular metabolites, including glutathione and serotonin‐related metabolic features	2′‐FL can reshape infant‐type microbial communities and microbial metabolic activity in vitro, but responses may depend on the initial microbial community	S. Zhang, Chen, et al. (2023)
3‐FL	In vitro fermentation using pooled fecal microbiota from 12‐week‐old infants	Delayed fermentation; enriched Bacteroides and Enterococcus; microbiota response differed from GOS/inulin and LNT2	Gradual production of acetic acid and lactic acid; fermentation digesta increased Lactobacillus plantarum adhesion to Caco‐2 cells at later time points	3‐FL fermentation is structure‐dependent and may contribute to age‐related microbial maturation and cross‐feeding networks	Kong et al. (2021)
3‐FL	21‐day neonatal farm piglet safety model	Did not markedly alter overall alpha diversity or global microbiota composition; bifidobacteria were absent in this piglet model; changes included enrichment of Blautia, Prevotellamassilia timonensis, and Eubacterium coprostanoligens	No adverse effects on growth, clinical observations, or toxicological endpoints	3‐FL appears safe in a neonatal piglet model, but microbiota outcomes should be interpreted cautiously because piglet microbiota differ from infant microbiota	Pitt et al. (2024)
2′‐FL and 3‐FL	Adult SHIME model using healthy adult fecal microbiota	Both 2′‐FL and 3‐FL altered adult gut microbial composition	Increased acetate, propionate, and butyrate production, with different timing of butyrate generation between 2′‐FL and 3‐FL	Fucosyllactoses can modulate microbiota and SCFA production beyond infancy, but findings from adult models should not be directly extrapolated to infants	Sato et al. (2025)
(2) FHMO‐containing combinations and synbiotic interventions
2′‐FL + Limosilactobacillus reuteri	Double‐blind randomized controlled trial in healthy term infants	Microbiota pattern shifted toward that of breastfed infants; lower abundance of opportunistic pathogenic bacteria during early infancy; bifidobacteria abundance tended to approach the breastfed reference group	Supported age‐appropriate growth and good gastrointestinal tolerance	2′‐FL may provide incremental microbiota‐modulating effects when added to probiotic‐containing infant formula	Alliet et al. (2022)
2′‐FL + Bifidobacterium longum subsp. infantis	Early‐life ceftriaxone‐exposed BALB/c mouse model	Restored antibiotic‐disrupted gut microbiota structure; reduced opportunistic pathogens such as Enterococcus and Staphylococcus; improved microbial diversity	Increased short‐chain fatty acids, improved crypt depth, and enhanced IgG and secretory IgA production	HMOs‐based synbiotics may help repair early‐life antibiotic‐induced dysbiosis and support intestinal and immune development	Y. Luo, Li, et al. (2025) and Y. Luo, Xia, et al. (2025)
2′‐FL + Bifidobacterium bifidum	Constipated mouse model	Increased beneficial taxa, including Akkermansia, Bifidobacterium, and Parabacteroides	Improved defecation parameters, reduced inflammatory cytokines, strengthened tight junction proteins, and upregulated butanoate metabolism	2′‐FL‐based synbiotics may regulate gut microbiota and intestinal transcriptional function in gastrointestinal disorders	Shan et al. (2025)
2′‐FL + 6′‐SL	Healthy BALB/c mouse model	Increased microbial diversity; enriched Bacteroides and unclassified Clostridia_UCG_014; reduced potentially unfavorable taxa such as Ruminococcus	Increased SCFA production and enhanced immune‐related indicators, including lymphocyte proliferation and NK cell activity	Combining fucosylated and sialylated HMOs may produce stronger microbiota‐mediated immunomodulatory effects than 2′‐FL alone	Feng et al. (2025)

From the perspective of infant formula development, responses to FHMO supplementation may vary depending on baseline microbiota, mode of administration, antibiotic exposure, gestational age, feeding pattern, and maternal secretor status (Alliet et al. 2022; Chatchatee et al. 2022; Donovan et al. 2023; Estorninos et al. 2022; Jochum et al. 2023; G. Luo et al. 2023; Neumer et al. 2021; Zhernakova et al. 2025). A single fucosylated oligosaccharide is unlikely to fully reproduce the microbial ecology associated with human milk (Q. Chen, Ma, et al. 2024; Del Compare et al. 2025). Future studies should move beyond single‐compound evaluation and focus on human‐milk‐inspired formulation strategies that integrate defined combinations of FHMOs with other bioactive milk components. Particular attention should be given to next‐generation HMO‐based synbiotics, in which specific FHMOs are paired with infant‐adapted or rationally selected probiotic strains to reconstruct microbiota‐mediated benefits more closely resembling those of human milk.

3.2. Intestinal Barrier Support

The intestinal barrier is a multilayered defense system composed of epithelial cells and tight junctions, mucus and antimicrobial factors, immune components, resident microbiota, and microbial metabolites such as short‐chain fatty acids (Chelakkot et al. 2018; Y.‐J. Kim et al. 2023; Neurath et al. 2025). During early life, this barrier is still developing and is sensitive to nutritional, microbial, and inflammatory stimuli (X. Lin, Lin, et al. 2025; Martínez‐Augustin et al. 2025). FHMOs have therefore attracted interest as dietary components that may support barrier maturation and intestinal homeostasis (Y.‐J. Kim et al. 2023; S. Lee, Goodson, et al. 2021; Y. Luo, Zhang, et al. 2023). Mechanistic studies suggest that FHMOs may influence epithelial integrity through several pathways (Neurath et al. 2025; C.‐C. Sun et al. 2025). In vitro and animal studies have reported that 2′‐FL and 3‐FL can modulate transepithelial electrical resistance, reduce permeability markers, and affect tight junction proteins such as ZO‐1, occludin, and claudins (Y.‐J. Kim et al. 2023; S. Lee, Goodson, et al. 2021). These findings support a potential role in maintaining epithelial integrity under inflammatory or stress conditions. Nevertheless, many studies use supraphysiological concentrations or disease‐like challenge systems; therefore, tight junction‐related outcomes should be regarded as mechanistic indicators rather than direct evidence of clinical benefit in healthy infants.

FHMOs may also support barrier function indirectly through microbiota‐mediated mechanisms (Y.‐J. Kim et al. 2023; S. Lee, Goodson, et al. 2021; Y. Luo, Zhang, et al. 2023). By shaping microbial communities and fermentation activity, they can influence metabolites such as acetate, propionate, and butyrate, which are involved in epithelial energy metabolism, mucus production, immune signaling, and luminal pH regulation (Gu et al. 2024; Y. Luo, Zhang, et al. 2023). However, causal links among HMO intake, microbial shifts, metabolite production, and barrier improvement remain difficult to establish without integrated multiomics, microbial depletion or rescue experiments, and controlled human studies. Additional experimental evidence indicates that FHMOs may modulate inflammation‐ and oxidative stress‐related pathways, including TLR4/NF‐κB, MAPK, and Nrf2‐associated responses (Duan et al. 2026; Y.‐J. Kim et al. 2023; C.‐C. Sun et al. 2025). These findings are relevant because excessive inflammation and oxidative stress can compromise epithelial integrity. However, most studies have been conducted in colitis models, LPS‐challenged cells, high‐fat diet models, or other stress systems that do not fully represent healthy term infants (Duan et al. 2026; Y.‐J. Kim et al. 2023; S. Lee, Goodson, et al. 2021). Thus, these data are best interpreted as evidence of protective potential under disturbed intestinal conditions rather than proof of routine barrier enhancement.

Current evidence supports a plausible role for FHMOs in intestinal barrier‐related functions, especially through epithelial, microbial, metabolic, and immune‐associated pathways. Future studies should use physiologically relevant doses, infant‐derived microbiota models, standardized barrier endpoints, and well‐designed clinical trials to determine whether experimental findings translate into meaningful nutritional benefits.

3.3. Relevance to Intestinal Injury Risk in Vulnerable Infants

Necrotizing enterocolitis is a severe inflammatory intestinal disorder that primarily affects preterm and very low birth weight infants (X. Wang, Li, et al. 2024). Its development is associated with intestinal immaturity, impaired barrier function, dysregulated host–microbe interactions, ischemia‒reperfusion injury, and excessive inflammatory signaling (Autran et al. 2016; Good et al. 2016; Wejryd et al. 2018). Because breastfeeding is associated with a lower risk of necrotizing enterocolitis than formula feeding, human milk components, including HMOs, have been investigated as potential contributors to intestinal protection in vulnerable infants (Autran et al. 2016; Sodhi et al. 2020). Evidence obtained using defined individual HMOs should be distinguished from that obtained using pooled HMO preparations. Studies in which 2′‐FL was administered as a defined intervention have reported reduced intestinal injury scores, inflammatory signaling, epithelial apoptosis, and histological damage in neonatal rodent or piglet models (Autran et al. 2016; Good et al. 2016; Sodhi et al. 2020; C. Wang et al. 2019). These findings support the protective potential of 2′‐FL under experimental NEC‐like conditions but do not establish clinical prevention of NEC in infants.

In contrast, several studies used HMOs isolated from pooled human milk rather than a purified individual compound (B. Li et al. 2020; Sodhi et al. 2020; C. Wang et al. 2019, 2020). These studies reported effects on epithelial differentiation, crypt cell proliferation, mucin expression, and intestinal injury. Such findings support the biological activity of the tested HMO mixture, but they cannot identify fucosylation or any individual fucosylated component as the determinant of the observed protection. Evidence from pooled HMO preparations should therefore not be attributed specifically to 2′‐FL, 3‐FL, or the broader class of FHMOs. Compared to 2′‐FL and 3‐FL, the evidence for other FHMOs is more limited. Some experimental studies suggest that both 2′‐FL and 3‐FL may attenuate injury markers, reduce pro‐inflammatory cytokine expression, and support mucus‐related responses such as MUC2 expression in neonatal mouse or epithelial cell models (Y.‐J. Kim et al. 2023; R. Y. Wu et al. 2022). However, direct comparative evidence remains sparse, and the relative effectiveness of 2′‐FL, 3‐FL, and more complex fucosylated structures cannot yet be ranked with confidence. Human observational studies provide another line of evidence, reporting associations between lower diversity or lower abundance of specific FHMOs in milk from mothers of preterm or extremely low birth weight infants and increased necrotizing enterocolitis risk (Pang et al. 2025; Wejryd et al. 2018; Whittaker et al. 2025; N. Wu, Wang, et al. 2025; Zhernakova et al. 2025). These findings support a possible role of the broader HMO profile in intestinal resilience, but they cannot establish causality and may be confounded by maternal secretor status, lactation stage, milk composition, feeding practices, clinical care, antibiotic exposure, and infant maturity (Kalluri et al. 2025; W. J. Li, Gao, et al. 2025; Song et al. 2026; J. Xu et al. 2025; Zhernakova et al. 2025). As summarized in Table 2, the strength of compound attribution depends on the intervention examined. Findings obtained using defined individual HMOs can be assigned to the tested compound within the limitations of the experimental model, whereas effects observed with pooled human‐milk HMO preparations or broader HMO profiles cannot be attributed specifically to their fucosylated components.

TABLE 2.

Compound‐ and preparation‐specific evidence for HMOs in NEC‐related and experimental intestinal injury models.

HMO intervention or preparation	Study model	Main protective or regulatory outcomes	Proposed mechanisms	Evidence interpretation and structural attribution	References
(1) Defined individual fucosylated HMO interventions
2′‐FL (defined individual HMO)	Neonatal rat and C57BL/6 mouse experimental NEC models	Reduced intestinal pathology and NEC severity; decreased pro‐inflammatory markers; preserved small‐intestinal mucosal architecture	Attenuation of intestinal injury and inflammatory signaling; improved mesenteric perfusion through eNOS upregulation	Compound‐specific preclinical evidence for 2′‐FL; does not establish clinical NEC prevention in infants or a class‐wide effect of FHMOs	Autran et al. (2016); Good et al. (2016)
2′‐FL and 3‐FL (evaluated separately)	IL‐6‐induced Caco‐2 barrier dysfunction and DSS‐induced mouse colitis	Improved epithelial barrier function and tight‐junction expression; reduced inflammatory cytokines and colitis severity	Reinforcement of epithelial integrity, suppression of inflammatory responses, and modulation of the gut microbiota	Defined‐compound evidence in non‐NEC injury models; findings cannot be directly extrapolated to neonatal NEC or clinical effects in infants	Y.‐J. Kim et al. (2023)
(2) Comparative structure‐specific studies using defined HMOs
2′‐FL and 6′‐SL (evaluated individually and, where applicable, together)	Newborn mouse NEC, premature piglet, IEC‐6, mouse enteroids, and human intestinal explants	Reduced experimental NEC severity, apoptosis, inflammation, weight loss, and histological injury; attenuated inflammatory responses in human intestinal tissues	Inhibition of TLR4/NF‐κB signaling; possible interaction with the TLR4–MD2 complex	Activity of both fucosylated 2′‐FL and nonfucosylated 6′‐SL argues against fucosylation alone as the common determinant of protection	Sodhi et al. (2020)
2′‐FL, 3‐FL, 6′‐SL, LNT, and LNnT (evaluated individually)	Caco‐2Bbe1 cells and murine experimental NEC model	Produced distinct epithelial transcriptomic responses and variable protection against barrier dysfunction, inflammation, and experimental NEC	HMO‐dependent regulation of epithelial signaling, mucin‐related responses, and intestinal morphology	Direct comparison demonstrates structure‐dependent activity; protective responses are not restricted to FHMOs	R. Y. Wu et al. (2022)
(3) Human milk‐derived HMO mixtures and observational HMO profiles
Pooled human milk‐derived HMO preparations containing fucosylated and nonfucosylated structures	Neonatal mouse NEC, intestinal organoids, Caco‐2 cells, and hypoxic intestinal injury models	Reduced inflammatory signaling and tissue injury; promoted epithelial proliferation, differentiation, and maturation; decreased apoptosis	Suppression of TLR4/NF‐κB and stress‐associated pathways; enhanced epithelial turnover and differentiation; activation of EGFR signaling	Preparation‐level evidence only; effects cannot be attributed specifically to fucosylated constituents or to any single HMO	B. Li et al. (2020); C. Wang et al. (2019, 2020)
Broader human‐milk HMO profile, including fucosylated structures	Observational study of exclusively breastfed extremely low birth weight infants	Lower HMO diversity and lower LNDFH I concentrations were associated with NEC cases	Potential contribution of overall HMO composition, maternal phenotype, and correlated variation among multiple HMOs to intestinal resilience	Observational association; cannot establish causality or identify a single active HMO or structural motif	Wejryd et al. (2018)

3.4. Gut–Brain Axis and Neurodevelopment

Most mechanistic evidence concerning the gut–brain axis has been obtained using 2′‐FL in rodent models. These studies have reported changes in microbial composition, tryptophan‐ and neurotransmitter‐related metabolism, hippocampal signaling, long‐term potentiation, and learning‐related behavior (L. Hou et al. 2021; W. J. Li, Gao, et al. 2025; Xing et al. 2026; Zhernakova et al. 2025). Antibiotic depletion and vagal interruption experiments further suggest that microbiota‐mediated and vagal pathways may contribute to some of these responses. However, these findings are model‐dependent, and results from adult or disease‐associated neurological models are regarded as hypothesis‐generating rather than direct evidence for neurodevelopmental benefits in infants (Berger et al. 2022; L. Zhu et al. 2025). Both microbial metabolites and intact HMOs may contribute to extraintestinal signaling. Although most ingested HMOs remain within the gastrointestinal tract, small proportions of intact HMOs, including 2′‐FL, have been detected in infant plasma and urine, indicating limited systemic absorption (Ke et al. 2024; Okburan and Kiziler 2023; Oliveros et al. 2016; Vázquez et al. 2015; J. Xu et al. 2025; L. Zhu et al. 2025). Therefore, systemic effects cannot be assumed to depend exclusively on microbial metabolites. Nevertheless, the circulating concentrations, tissue distribution, ability to reach the developing nervous system, and biological relevance of intact HMOs remain insufficiently established.

Human evidence is currently limited mainly to observational associations. Maternal milk 2′‐FL concentrations or broader HMO profiles have been associated with selected measures of infant brain organization or later cognitive development (Jia et al. 2025; S. Lee, Goodson, et al. 2021; Scavone et al. 2016; Q. Xie et al. 2022). However, residual confounding by maternal phenotype, breastfeeding exposure, socioeconomic factors, and other milk components cannot be excluded. Controlled infant intervention studies isolating the effects of 2′‐FL and incorporating validated neurodevelopmental endpoints remain insufficient. Moreover, evidence for 3‐FL, LNFPs, LNDFHs, and other complex FHMOs is minimal. Consequently, the current data support biological plausibility but not established neurodevelopmental benefits of FHMOs as a class.

3.5. Immune Maturation and Homeostasis

Early‐life immune development is shaped by interactions among diet, gut microbiota, epithelial barrier function, and mucosal immune cells (Burris et al. 2021; Koren et al. 2024; S. Li, Guo, et al. 2025; Rajani et al. 2018). FHMOs have been investigated as nutritional factors that may influence this process through microbial colonization, microbial metabolites, epithelial signaling, and host immune responses (Y. Chen, Wen, et al. 2024; Koren et al. 2024; Y. Luo, Zhang, et al. 2023; R. P. Singh et al. 2022). However, the available evidence predominantly concerns 2′‐FL, whereas evidence for 3‐FL and more complex FHMOs remains limited. Their potential role in infant nutrition should therefore be framed as modulation of immune maturation and homeostasis rather than treatment of inflammatory or immune‐mediated diseases. A plausible indirect mechanism involves microbiota‐mediated regulation of mucosal immunity (T. Liu et al. 2020; Schneider et al. 2022). By supporting selected infant‐associated microbial taxa and cross‐feeding networks, 2′‐FL may alter the production of short‐chain fatty acids and other microbial metabolites. These metabolites may, in turn, affect regulatory T‐cell differentiation, epithelial barrier function, cytokine balance, and secretory immunoglobulin A production (Y. Chen, Wen, et al. 2024; J.‐Y. Kim et al. 2024; J. Li, Wei, et al. 2024; Moon et al. 2024; Yan et al. 2024). Experimental studies have reported that 2′‐FL supplementation is associated with changes in mucosal immune markers, including secretory immunoglobulin A, interleukin‐10, and markers related to Type 2 immune responses in selected animal models (T. Liu et al. 2020; Y. Luo, Li et al. 2025; Y. Luo, Zhang, et al. 2023). However, these responses are context‐dependent and model‐dependent and should not be assumed to occur uniformly across infants or across different FHMOs.

Evidence from challenged‐cell and animal models further suggests that 2′‐FL can influence inflammatory signaling under experimentally disturbed conditions (de Melo et al. 2025; Neurath et al. 2025). Reductions in interleukin‐6, tumor necrosis factor‐α, interleukin‐1β, or related signaling activity have been reported in association with TLR4/NF‐κB‐, MAPK‐, JAK–STAT‐, and STAT3‐related pathways (J.‐Y. Kim et al. 2024; J. Li, Wei, et al. 2024; Na et al. 2026). Most of this evidence, however, has been obtained from LPS‐challenged cells, pathogen‐exposed animals, antibiotic‐disrupted microbiota, or chemically induced colitis models. These findings demonstrate changes in selected inflammatory and immune biomarkers under challenge conditions but do not establish generalized immune support or clinically meaningful anti‐inflammatory effects in healthy infants.

The combination of 2′‐FL with probiotic strains has also attracted attention. Studies involving 2′‐FL and infant‐associated bifidobacteria have reported changes in mucosal immune markers, microbial metabolites, and inflammatory responses following early‐life disturbances such as antibiotic exposure (Y. Luo, Li, et al. 2025; Y. Luo, Zhang, et al. 2023). Because these interventions contain both an HMO and a selected microbial strain, the observed outcomes cannot be attributed to 2′‐FL alone. They are likely to depend on the probiotic strain, HMO dose, baseline microbiota, administration schedule, and formulation characteristics. Consequently, the safety, compatibility, stability, and reproducibility of each synbiotic preparation require separate evaluation before application in infant formula.

It is important to distinguish immune maturation from immune stimulation. A desirable nutritional outcome in early life is balanced immune development, including mucosal defense, dietary tolerance, controlled inflammatory responses, and epithelial integrity (Koren et al. 2024; S. Li, Guo, et al. 2025; W. J. Li, Gao, et al. 2025). Biomarkers such as secretory immunoglobulin A, regulatory cytokines, Th1/Th2 balance, Treg‐related markers, and inflammatory mediators provide mechanistic information but do not automatically demonstrate clinical benefit (Y. Luo, Zhang, et al. 2023; Na et al. 2026; Nyangahu et al. 2023; Y. Wang, Yue, et al. 2026; W. Zhang et al. 2019). Overall, current evidence supports the biological plausibility that 2′‐FL may influence selected pathways involved in immune homeostasis. Nevertheless, stronger evidence from well‐controlled human studies is required before these findings can support compound‐specific or formula‐related immune benefit claims, and they should not be generalized to the broader class of FHMOs (Chouraqui 2020; Gold et al. 2022; Sekerel et al. 2021).

3.6. Allergy‐Related and Tolerance Development

Allergic diseases in early life arise from complex interactions among genetic predisposition, epithelial barrier integrity, microbial colonization, immune maturation, environmental exposure, and dietary factors (Burris et al. 2021; den Elzen et al. 2026; Rajani et al. 2018). In food‐allergy challenge models, 2′‐FL and 3‐FL have been associated with changes in allergen‐specific immunoglobulin E, Th1/Th2‐associated cytokines, regulatory T‐cell‐related markers, mucus and tight junction‐related responses, and gut microbial composition (Castillo‐Courtade et al. 2015; Kou et al. 2023; A. Li, Li, et al. 2021; T. Liu et al. 2020; S. Wu, Chen, et al. 2025). These findings provide mechanistic evidence that the tested compounds may influence pathways involved in allergic sensitization or mucosal tolerance. However, changes in these biomarkers under experimental challenge conditions do not prevent clinically diagnosed food allergies in infants.

Airway‐allergy models provide additional but less directly translatable evidence. Studies using house dust mite‐, papain‐, or fungal allergen‐induced airway inflammation have reported reductions in selected Type 2 immune markers, epithelial alarmins, eosinophil‐associated responses, or innate lymphoid cell Type 2 activity following interventions containing FHMOs (Bozorgmehr et al. 2023; X. Han et al. 2024; Zuurveld et al. 2025). Some studies have proposed that these responses involve gut microbiota‐derived short‐chain fatty acids and receptors such as GPR43 (Bozorgmehr et al. 2023; X. Han et al. 2024). Nevertheless, these results are specific to the tested compounds or preparations and the corresponding allergen challenge models. Airway‐allergy models are relatively distant from the intended nutritional use of FHMOs in infant formula and should therefore be regarded as hypothesis‐generating rather than evidence of allergy prevention in infants (Chouraqui 2020; Sekerel et al. 2021).

Some infant studies using formulas containing 2′‐FL have reported acceptable tolerance or changes in gastrointestinal and allergy‐related symptoms in selected populations, including infants with suspected cow's milk protein allergy or feeding intolerance (Gold et al. 2022; Ramirez‐Farias et al. 2024). These findings are relevant to product development but cannot readily be attributed to 2′‐FL alone because the tested formulas may also differ in protein source, degree of protein hydrolysis, fat blend, additional prebiotics or probiotics, and micronutrient composition (Gold et al. 2022; Ramirez‐Farias et al. 2021). Short intervention periods, small sample sizes, absence of appropriate randomized controls, and reliance on symptom‐based endpoints further limit the strength of the conclusions (Ramirez‐Farias et al. 2021, 2024). These studies therefore provide information primarily about the safety and tolerance of the complete formula preparation rather than demonstrating an allergy‐related benefit of 2′‐FL as an isolated ingredient.

Allergy‐related immune modulation should be distinguished from clinically established allergy prevention (Chouraqui 2020). Infant formula‐related claims would require adequately powered randomized controlled trials with prospectively defined, clinician‐assessed allergy and tolerance outcomes, sufficient follow‐up, appropriate comparison formulas, and careful control of relevant confounders (Sekerel et al. 2021). Priority outcomes should include clinically confirmed food allergy, standardized measures of gastrointestinal and feeding tolerance, growth, and adverse events. Immunological, microbiota, and metabolite measurements may provide valuable secondary mechanistic information but should not substitute for clinical outcomes. Current evidence remains concentrated on 2′‐FL and, to a lesser extent, 3‐FL, whereas evidence for LNFPs, LNDFHs, and other complex FHMOs is insufficient. Accordingly, the existing allergy‐related findings should be considered preliminary and compound‐specific and should not be generalized to the broader FHMO family (Versluis et al. 2024; S. Wu, Chen, et al. 2025; Zuurveld et al. 2025).

4. Synthesis Strategies

The commercial application of FHMOs depends not only on structural design and biological functionality but also on the ability to manufacture food‐grade ingredients with high purity, consistent quality, acceptable cost, and regulatory‐compliant safety profiles (Del Compare et al. 2025; J. Xu et al. 2025; J. Yu et al. 2025). Several production routes have been explored, including chemical synthesis, enzymatic synthesis, microbial fermentation, and plant‐based biomanufacturing (Pooladian et al. 2025; Y. Zhang et al. 2026; Zhao et al. 2025). Each platform has distinct advantages and limitations in terms of regioselectivity, stereochemical control, productivity, sustainability, downstream purification, impurity profile, and scalability. Therefore, the suitability of a production strategy should be evaluated according to its intended use, target compound, required purity, production scale, environmental footprint, and compatibility with infant nutrition regulations.

Chemical synthesis offers precise structural control and remains valuable for preparing reference standards, rare structures, and structurally defined glycans for mechanistic studies (S. Wang et al. 2022). However, its use in large‐scale food ingredient production is constrained by multistep protection‐deprotection procedures, organic solvent use, purification burden, stereochemical side products, and environmental impact (S. Wang et al. 2022). Enzymatic synthesis provides improved regioselectivity and milder reaction conditions, but donor cost, enzyme stability, competing hydrolysis, and productivity remain major limitations (Niharika et al. 2025). Microbial fermentation has become the dominant industrial platform for commercially available FHMOs, especially 2′‐FL and 3‐FL, because intracellular sugar nucleotide generation and glycosyltransferase activity can be integrated in scalable cell factories. Nevertheless, microbial production still requires careful control of host‐derived impurities, metabolic by‐products, endotoxin risk when using Gram‐negative hosts, downstream purification, and batch‐to‐batch consistency (J. Wang, Lao, et al. 2026; L. Yang et al. 2026; Y. Zhang et al. 2026). Plant‐based systems and artificial intelligence (AI)‐driven biomanufacturing are emerging approaches that may improve sustainability and design efficiency, but their application to food‐grade FHMOs remains at an early stage (Barnum et al. 2024; Q. Chen et al. 2025). Figures 3 and 4 summarize the technical principles and historical development of FHMO production, respectively. To facilitate cross‐platform comparison, Table 3 summarizes representative studies according to the target FHMO, host or catalyst and key enzyme source, core synthesis or engineering strategy, maximum reported scale and performance, and major translational limitations.

FIGURE 3.

Overview of chemical, enzymatic, microbial, and plant‐based production platforms for fucosylated human milk oligosaccharides, together with computational and synthetic biology tools for enzyme, pathway, and process optimization.

FIGURE 4.

Technical principles, key milestones, and strategic evolution of FHMO production.

TABLE 3.

Comparative assessment of representative production platforms for fucosylated human milk oligosaccharides.

Production platform	Target FHMOs	Host/catalyst and key enzyme source	Core synthesis/engineering strategy	Maximum reported scale and performance	Main translational limitations	References
Chemical synthesis	2′‐FL	Protected fucosyl donor and lactose acceptor	Convergent glycosylation and global deprotection; process redesigned to require only one chromatographic purification	Kilogram scale; 2′‐FL isolated on the kilogram scale; 19.8% over glycosylation/deprotection from a 5.0‐kg intermediate	Protecting‐group burden; high solvent and reagent demand; chromatography and waste generation; lower overall yield than fermentation	Agoston et al. (2019)
Chemical synthesis	3‐FL; LNFP V	Protected lactose‐, fucose‐, and LNT‐derived building blocks	Regioselective chemical glycosylation followed by staged deprotection; first reported chemical route to LNFP V	Laboratory/preparative scale; 3‐FL final deprotection, 73%; LNFP V key coupling, 42%, and final deprotection, 56% (mg scale)	Multiple protected intermediates and purification steps; coupling side reactions; limited scale and unfavorable material intensity for complex FHMOs	Pooladian et al. (2025)
Cell‐free enzymatic synthesis	2′‐FL	Recombinant Escherichia coli Gmd/WcaG; Helicobacter pylori NCTC 364 α1,2‐FucT	Two‐stage cascade: GDP‐l‐Fuc generation from GDP‐d‐Man, followed by α1,2‐fucosylation of lactose	18 mg isolated 2′‐FL; 65% isolated yield; GDP‐l‐Fuc generation, 78%	Cost and preparation of nucleotide‐sugar donor; low soluble FucT expression without fusion/low‐temperature expression; chromatography; mg scale	Albermann et al. (2001)
Cell‐free enzymatic synthesis	3‐FL; LNFP III; LNFP VI (LNnFP V); LNnDFH II and higher FHMOs	C‐terminally truncated H. pylori α1,3‐FucT (Hp3FT), expressed in E. coli BL21(DE3)	Truncation‐assisted soluble expression; one‐pot multienzyme GDP‐l‐Fuc supply; acceptor choice and reaction order used to control regioselectivity and fucosylation degree	Milligram scale; representative isolated yields: 3‐FL, 88%; LNFP III, 80%; LNnDFH II, 68%; other products, 80%–94%	Nucleotide‐sugar cost; mg scale; marked acceptor dependence (activity toward lactose is much lower than toward LacNAc); regioisomer control and purification remain critical	Bai et al. (2019)
Cell‐free enzymatic synthesis	2′‐FL	α‐l‐Fucosidase OUC‐Jdch16 from marine Flavobacterium algicola 12076	Transfucosylation from activated pNP‐α‐l‐Fuc to lactose using a recombinant glycosidase	Laboratory scale; donor‐based conversion, 84.82% at 48 h and 92.15% at 120 h	Activated donor requirement; long reaction time; competing hydrolysis and declining enzyme activity; performance not demonstrated at process scale	Zhou et al. (2022)
Cell‐free enzymatic synthesis	Branched FHMOs and related complex HMO library	Human GCNT2 plus six bacterial glycosyltransferases; enzymes produced recombinantly	Seven modular one‐pot multienzyme systems generate sugar nucleotides in situ and assemble linear or asymmetrically branched backbones before terminal fucosylation/sialylation	Preparative library synthesis at up to 100 mM reaction steps; 31 branched and 13 linear HMOs were prepared	Sequential modules, multiple enzymes and repeated purification increase cost and operational complexity; suited mainly to standards/library generation rather than bulk ingredients	Y. Li, Li, et al. (2024)
Microbial fermentation	2′‐FL	Plasmid‐, antibiotic‐, and inducer‐free E. coli; engineered WPfutC α1,2‐FucT (ΔC12)	Modular precursor optimization, rational FucT engineering, and a self‐assembled ManC–Ndk complex to improve GDP‐l‐Fuc supply	Shake flask, 25.6 g/L; fed‐batch bioreactor, 108.3 g/L and 1.18 g/L/h (reactor volume not reported in the manuscript)	E. coli‐derived host impurities/endotoxin and downstream polishing remain relevant; long‐term genetic robustness and performance at manufacturing scale require confirmation	B. Li, Liu, et al. (2026)
Microbial fermentation	2′‐FL	Engineered Bacillus subtilis; Neisseria meningitidis β1,4‐GalT and H. pylori α1,2‐FucT	De novo intracellular lactose and GDP‐l‐Fuc synthesis from glucose; lactose‐responsive LacI‐Q291F circuit dynamically balances precursor modules	Shake flask, 9.67 ± 0.65 g/L; 3‐L fed batch, 30.1 g/L, and 0.15 g/g glucose	Lower titer than leading E. coli systems; complex circuit and precursor balance; demonstrated only at 3 L; industrial genetic stability remains unverified	Q. Zhang, Xu, et al. (2025)
Microbial fermentation	2′‐FL	Genome‐engineered Pichia pastoris; tagged and semirationally engineered FucT2	CRISPR‐assisted genomic integration; optimization of GDP‐l‐Fuc, GTP/NADPH, lactose transport, and FucT2 expression modules	3‐L fed‐batch fermentation, 22.35 g/L	Lower titer than leading bacterial systems; extensive multigene engineering; 3‐L evidence only; yeast‐specific impurity and downstream profiles require process‐specific evaluation	Fang, Gao, Mund, et al. (2025); l Fang, Gao, Yu, et al. (2025)
Microbial fermentation	2′‐FL; 3‐FL	Engineered Saccharomyces cerevisiae W303‐1a; Bacillus cereus α1,2‐FucT and Neobacillus cucumis α1,3‐FucT	Pheromone‐mediated quorum‐sensing circuit dynamically regulates GDP‐l‐Fuc supply and fucosyllactose synthesis	5 L fed batch: 32.05 g/L 2′‐FL and 20.91 g/L 3‐FL; lactose conversions, 0.80 and 0.64 mol/mol, respectively	Lactose feed is needed; substantial intracellular retention for 3‐FL; lower productivity than leading E. coli systems; scale‐dependent circuit robustness is unresolved	M. Xu et al. (2023)
Microbial fermentation	3‐FL	Plasmid‐free E. coli BL21(DE3); chromosomally integrated N. cucumis α1,3‐FucT (Fut3Bc)	Multiple chromosomal FucT copies, promoter replacement for GDP‐l‐Fuc genes, and deletion of competing lactose/colanic acid nodes	5 L fed batch, 28.13 g/L at 60 h; 0.54 g/L/h; 0.95 mol/mol lactose	E. coli‐derived impurities/endotoxin; only 5 L validation; oxygen transfer, feeding, and downstream purification remain scale‐up constraints	Y. Chen, Chen, et al. (2024); Q. Chen, Ma, et al. (2024); Y. Chen, Wen, et al. (2024)
Microbial fermentation	DFL	Plasmid‐ and inducer‐free E. coli MG1655; engineered H. pylori α1,3‐FucT and α1,2‐FucT fusion	FucT screening, systematic pathway balancing, lactose‐operon engineering, and iMARS‐guided fusion‐enzyme design to convert the 2′‐FL intermediate	5 L fed batch, 74.0 g/L; 0.96 g/L/h	Sequential fucosylation requires tight control of 2′‐FL intermediate and regioisomer/byproduct profiles; E. coli purification and manufacturing‐scale validation remain necessary	Z. Liu et al. (2026)
Microbial fermentation	LNFP I	E. coli W3110S; Neisseria polysaccharea LgtA, Chromobacterium violaceum β1,3‐GalT, and engineered Francisella α1,2‐FucT	Stepwise LNT‐pathway construction combined with directed evolution for selective terminal α1,2‐fucosylation and suppression of 2′‐FL	3 L fed batch, 19.6 g/L LNFP I; 65% lactose‐to‐LNFP I conversion; 2′‐FL, 0.04 g/L	Plasmid/inducer dependence; poor FucT solubility; residual LNT and unidentified byproduct; 3 L validation only	Endo et al. (2024)
Microbial fermentation	LNFP I	Three‐plasmid E. coli system; Sulfuriflexus mobilis α1,2‐FucT (Smob)	RIDD–RIAD self‐assembling enzyme complex spatially couples LNT formation and fucosylation; plasmid copy numbers balance pathway modules	Shake flask, 2.37 g/L; 5 L fed batch, 11.6 g/L in 34 h, 0.34 g/L/h, and 86.7% LNT conversion	Three plasmids plus antibiotic/inducer dependence; lower titer than genome‐integrated systems; transfer to a stable production chassis remains necessary	J. Wang, Lao, et al. (2026)
Microbial fermentation	LNFP I	Plasmid‐, antibiotic‐, and inducer‐free genome‐integrated E. coli MG1655; Thermoanaerobacterium α1,2‐FucT	Chromosomal integration and modular optimization of LNT, GDP‐L‐Fuc, UDP‐Gal, cofactor, and efflux pathways; two‐stage temperature control	5 L fed batch: 37.44 g/L with two‐stage temperature control; 31.42 g/L at 37°C, 0.83 g/L/h and 90.73% LNT conversion	Substantial LNT/LNTri II precursor accumulation under the highest‐titer condition; temperature‐sensitive process control; 5 L evidence and E. coli downstream requirements	L. Yang et al. (2026)
Plant molecular farming	2′‐FL; LNFP I; LDFT; LNDFH I and related HMO mixtures	Transiently or stably engineered Nicotiana benthamiana via Agrobacterium; bacterial glycosyltransferase modules	Photosynthetic production with subcellularly targeted HMO pathways and optional reinforcement of plant GDP‐l‐Fuc supply	Transient LNFP I, 1075.03 µg/g dry weight; stable lines averaged 6.88 µg/g LNFP I and 130.35 µg/g 2′‐FL	Low stable accumulation and heterogeneous product profile; plant‐specific glycosylation/background glycans; extraction, purification, containment, and food regulatory pathways remain unproven	Barnum et al. (2024)

Note: Reported titer, isolated yield, conversion, productivity, and plant dry‐weight accumulation were measured under nonuniform experimental conditions and should not be directly ranked. “Maximum reported scale” denotes the largest scale described in the cited study, not commercial manufacture. The limitations column is a process‐level critical assessment based on the reported configuration. High‐throughput screening, molecular docking, molecular dynamics, biosensors, CRISPRi, and synthetic regulatory circuits are treated as enabling computational or synthetic biology tools, not as stand‐alone AI production platforms.

4.1. Chemical Synthesis

Chemical synthesis has long served as an important strategy for preparing structurally defined FHMOs, particularly analytical standards and rare glycans required for structural confirmation and mechanistic studies (Agoston et al. 2019; Pooladian et al. 2025). In 1981, Abbas et al. (1981) successfully synthesized 2′‐FL by chemical synthesis, which is the earliest known report of the successful synthesis of this compound. Its main advantage lies in precise control over glycosidic linkages, stereochemistry, and protecting‐group‐defined regioselectivity (Bandara et al. 2019; S. Wang et al. 2022). For structurally simple targets such as 2′‐FL and 3‐FL, stepwise glycosylation strategies can provide well‐defined products when carefully optimized (Agoston et al. 2019). For more complex fucosylated structures, convergent assembly and orthogonal one‐pot approaches can reduce the number of purification steps and improve synthetic efficiency, although their practical applicability depends strongly on donor reactivity, protecting‐group compatibility, and stereochemical control (Arboe Jennum et al. 2014; He et al. 2021; Pooladian et al. 2025).

Recent studies have demonstrated both the potential and limitations of chemical synthesis. Kilogram‐scale preparation of 2′‐FL has been achieved by optimizing crystalline intermediates and minimizing chromatographic purification, indicating that chemical synthesis can be adapted for selected targets under carefully controlled conditions (Agoston et al. 2019). Convergent fragment coupling has also enabled access to more complex FHMOs, such as LNPF‐related structures (Pooladian et al. 2025), while automated glycan assembly has expanded the feasibility of producing libraries of fucosylated glycans for structure‒function studies (Guberman et al. 2019; M.‐H. Lin, Kuo, et al. 2025). However, these advances do not necessarily mean that chemical synthesis is the most suitable route for routine food‐grade production, particularly when large volumes, low cost, and strict impurity control are needed.

The main constraints of chemical synthesis are associated with regioselective fucosylation, repeated protection and deprotection, solvent and reagent use, stereochemical by‐products, and downstream purification (Arboe Jennum et al. 2014; S. Wang et al. 2022). These limitations become more pronounced as glycan complexity increases (Das et al. 2025; Y. Singh et al. 2023). For infant nutrition applications, residual solvents, catalysts, protecting‐group‐derived impurities, and structurally related isomers must be carefully monitored and removed, particularly when multistep chemical routes are considered for ingredient production (Agoston et al. 2019). In addition, the environmental burden of multistep synthesis may conflict with the demand for greener and more sustainable ingredient manufacturing (Arboe Jennum et al. 2014; He et al. 2021; S. Wang et al. 2022). Therefore, chemical synthesis is best viewed as a high‐precision tool for standards, rare structures, and early‐stage validation, while large‐scale food ingredient production often requires additional evaluation of cost, purification burden, environmental footprint, and regulatory‐grade quality (Chouraqui 2020).

Future development of chemical synthesis for FHMOs should focus on reducing synthetic steps, improving atom economy, replacing hazardous solvents and reagents, increasing crystallization‐based purification, and integrating enzymatic steps where they provide superior selectivity (Agoston et al. 2019; Arboe Jennum et al. 2014; He et al. 2021). In this context, chemoenzymatic synthesis may serve as a bridge between the structural precision of chemistry and the regioselectivity of biocatalysis, particularly for complex or low‐abundance FHMOs that are not yet efficiently produced by microbial fermentation.

4.2. Enzymatic Synthesis

Enzymatic synthesis uses glycosyltransferases or glycosidases to construct fucosylated glycosidic linkages under relatively mild conditions. Compared with purely chemical synthesis, enzymatic approaches offer high regioselectivity and stereoselectivity, reduced dependence on extensive protecting‐group chemistry, and better compatibility with aqueous reaction systems (Albermann et al. 2001; Bai et al. 2019; Zhou et al. 2022). These characteristics make enzymatic synthesis particularly attractive for preparing structurally defined FHMOs and for expanding access to complex or rare structures (Bai et al. 2019; C. Li, Cao, et al. 2024).

Glycosyltransferase‐based synthesis is the most selective enzymatic route. α1‐2‐ and α1‐3‐fucosyltransferases can transfer fucose from GDP‐l‐fucose to lactose or extended oligosaccharide acceptors, enabling the synthesis of 2′‐FL, 3‐FL, and more complex fucosylated structures (Albermann et al. 2001; Bai et al. 2019). Recent studies using bacterial or recombinant fucosyltransferases have shown that enzyme specificity, acceptor preference, donor‐to‐acceptor ratio, and reaction sequence can strongly influence product distribution (Bai et al. 2019). Modular multienzyme strategies, including in situ nucleotide‐sugar regeneration, have further expanded the structural scope of enzymatic synthesis and are valuable for producing compound libraries for structure‒function studies (L. Li et al. 2017; Y. Li, Li, et al. 2024). However, the high cost of GDP‐l‐fucose and the need for efficient cofactor or sugar nucleotide recycling remain central barriers to large‐scale application (Jegal et al. 2025; L. Li et al. 2017).

These Leloir glycosyltransferases transfer l‐fucose from GDP‐l‐fucose, but enzymes from different sources preferentially fucosylate lactose or extended oligosaccharide acceptors, thereby determining the resulting product and isomer profiles. For example, screening of nine candidate enzymes identified Bf13FT from Bacteroides fragilis NCTC 9343 as the most suitable α1,3/4‐fucosyltransferase for regioselective LNFP V biosynthesis, demonstrating that these enzymes are not interchangeable across FHMO targets (N. Wang, Zhu, et al. 2024). Their heterologous application is frequently constrained by insufficient soluble expression, substrate affinity, catalytic efficiency, and stability. Recent engineering has therefore shifted toward site‐saturation mutagenesis and structure‐guided combinatorial design. The FucTa‐Y218K variant increased enzyme activity by 3.48‐fold and enabled 91.6% lactose conversion, whereas FutM2‐Q126A improved soluble expression and substrate affinity and supported 52.1 g/L 3‐FL production in a 5 L fermenter (Du et al. 2024; Y. Xie et al. 2024). More recently, structure‐guided iterative mutagenesis of FutA M32 generated variants with 28%–84% higher specific activity and increases in melting temperature of 0.22°C–4.87°C; the combined mutant also exhibited improved kinetic performance toward lactose and GDP‐l‐fucose (M. Li, Jia, et al. 2025). Future enzyme selection should therefore consider acceptor‐specific kinetics, regioisomer formation, soluble expression, stability, and performance under process‐relevant conditions rather than catalytic activity alone.

Glycosidase‐mediated transglycosylation offers an alternative route that avoids direct reliance on nucleotide sugar donors. α‐l‐Fucosidases can transfer fucosyl residues from activated donors to lactose or other acceptors, and enzyme discovery and engineering have improved thermostability, donor conversion, and substrate tolerance (Murata et al. 1999; Zhao et al. 2025; Zhou et al. 2022). Nevertheless, this strategy is frequently limited by competing hydrolysis, low product yield, and the cost or suitability of activated donors such as p‐nitrophenyl fucosides (Murata et al. 1999; Zeuner et al. 2018). These constraints currently limit its competitiveness for commodity‐scale production, although it may remain useful for selected synthetic targets, enzyme engineering studies, or integrated chemoenzymatic workflows (Zeuner et al. 2018; Zhao et al. 2025).

For food‐grade production, the key challenges for enzymatic synthesis are not only catalytic selectivity but also donor economy, enzyme reuse, productivity, and downstream purification (Albermann et al. 2001; Jegal et al. 2025; L. Li et al. 2017). Process improvements such as enzyme immobilization, continuous flow reactors, whole‐cell biocatalysis, low‐cost donor regeneration, and protein engineering may improve industrial feasibility (Jegal et al. 2025; L. Li et al. 2017; G. Luo et al. 2021; Zhao et al. 2025). In addition, enzymatic processes must meet safety and regulatory expectations related to enzyme sources, residual proteins, reaction impurities, and process‐derived contaminants. As a result, enzymatic and chemoenzymatic synthesis should be regarded as complementary to microbial fermentation: highly useful for rare, complex, or structurally diverse FHMOs but less clearly advantageous for large‐volume production of established ingredients such as 2′‐FL when efficient fermentation platforms are available (Jegal et al. 2025; Y. Li, Li, et al. 2024).

4.3. Microbial Cell Factories

Microbial fermentation has become the most important platform for the commercial production of FHMOs, particularly 2′‐FL and 3‐FL (Bych et al. 2019; Y. Chen, Zhu, et al. 2024; W. Li, Zhu, et al. 2021). In engineered microbial hosts, fucosyltransferases and sugar nucleotide biosynthetic pathways are introduced or optimized to convert lactose and simple carbon sources into target FHMOs (G. Chen et al. 2022; R. Chen et al. 2023b; W. Li, Zhu, et al. 2021). Compared with chemical and enzymatic synthesis, microbial cell factories offer integrated sugar nucleotide generation, lower raw material cost, reduced need for isolated enzymes or nucleotide sugar donors, and better scalability (Bych et al. 2019; J. W. Lee, Kwak, et al. 2021; W. Li, Zhu, et al. 2021).

Two major metabolic strategies are commonly used to supply GDP‐l‐fucose: the de novo pathway and the salvage pathway (W. Li, Zhu, et al. 2021; Liang et al. 2024). The de novo pathway converts endogenous GDP‐d‐mannose into GDP‐l‐fucose through engineered biosynthetic enzymes and is attractive because it can use inexpensive carbon sources (R. Chen et al. 2023b; M. Li et al. 2022). Its performance depends on precursor supply, cofactor balance, competing pathways, lactose uptake, fucosyltransferase activity, and by‐product control (Y. Chen et al. 2023a; Y. Zhu, Chen, et al. 2023). Metabolic engineering strategies such as deletion of lactose‐degrading enzymes, reduction of GDP‐l‐fucose‐consuming pathways, optimization of fucosyltransferase expression, enhancement of NADPH and GTP availability, and balancing of pathway modules have been shown to substantially increase 2′‐FL titers (M. Li et al. 2022; L. Lin et al. 2022; X. Sun et al. 2023). The salvage pathway uses exogenous l‐fucose to generate GDP‐l‐fucose and can simplify pathway design, but its economic feasibility depends strongly on l‐fucose cost, uptake efficiency, and conversion yield (N. R. Han et al. 2022; J. W. Lee, Kwak, et al. 2021; Liang et al. 2024).

Host selection is a critical factor for food‐grade production. Escherichia coli is widely used because of its genetic tractability, high productivity, and mature fermentation technology (Bych et al. 2019; R. Chen et al. 2023b; W. Li, Zhu, et al. 2021). However, the use of Gram‐negative hosts requires rigorous control of endotoxins, host cell proteins, DNA residues, and fermentation‐derived impurities, especially for infant nutrition products (J. W. Lee et al. 2020; Pitt et al. 2019). Alternative hosts such as Bacillus subtilis, Priestia megaterium, Saccharomyces cerevisiae, Yarrowia lipolytica, or Pichia pastoris may offer advantages related to food safety perception, secretion capacity, renewable‐substrate utilization, or regulatory acceptance, but they may also have lower productivity, more complex metabolic engineering requirements, or greater metabolic burden (Ji et al. 2022; J. W. Lee et al. 2020; Y. Li, Wang, et al. 2025; Park et al. 2025; Y. Zhang, Liu, et al. 2025). Therefore, host choice should be evaluated not only by titer but also by safety profile, purification feasibility, strain stability, scalability, and regulatory compatibility.

Recent synthetic biology strategies have extended FHMO cell factory design beyond static gene overexpression toward modular, dynamic, spatial, and genome‐integrated control. For pathway optimization, thermosensor‐coupled CRISPR interference and lactose‐responsive genetic circuits have been used to coordinate GDP‐l‐fucose supply, lactose availability, competing pathways, and the transition from biomass accumulation to 2′‐FL synthesis (W. Yu et al. 2022; Q. Zhang, Xu, et al. 2025). The spatial organization of pathway enzymes represents another recent advance. Synthetic membraneless organelles have been constructed to colocalize enzymes involved in 3‐FL biosynthesis, supporting a titer of 10.8 g/L in a 3‐L fermenter (Huang et al. 2024). More recently, a plasmid‐, antibiotic‐, and inducer‐free E. coli platform combined modular precursor optimization, rational engineering of α1,2‐fucosyltransferase, and artificial multienzyme complexes, achieving 108.3 g/L 2′‐FL in fed‐batch fermentation (B. Li, Liu, et al. 2026). CRISPR‐assisted pathway construction and subsequent optimization of precursor, cofactor, and fucosyltransferase modules have also enabled de novo 2′‐FL production in P. pastoris (Fang, Gao, Yu, et al. 2025). For more complex FHMOs, recent plasmid‐ and inducer‐free DFL production has integrated α1,3‐fucosyltransferase selection, systematic pathway balancing, and rational fusion‐enzyme engineering to reduce accumulation of the 2′‐FL intermediate (Z. Liu et al. 2026). Nevertheless, FHMO‐specific applications of genome‐scale metabolic models and data‐trained AI tools remain limited, indicating that these approaches are emerging opportunities rather than established production technologies. Nevertheless, high‐titer laboratory fermentation does not automatically translate into industrial robustness. Important scale‐up issues include oxygen transfer, substrate feeding, lactose solubility and uptake, osmotic stress, acetate or other by‐product accumulation, plasmid stability, genetic containment, downstream resin load, and product crystallization or drying behavior (Bych et al. 2019; Leng et al. 2026; X. Sun et al. 2023).

From the perspective of infant formula development, microbial fermentation‐based production of FHMOs should be evaluated within a comprehensive process‐quality framework. Key indicators include product identity, isomeric purity, residual lactose and monosaccharides, host‐derived impurities, endotoxin level, residual DNA and protein, heavy metals, solvent residues from purification, batch‐to‐batch consistency, and stability during downstream processing (Bych et al. 2019; Pitt et al. 2019). For 2′‐FL and 3‐FL, microbial fermentation is already the most commercially relevant route (Bych et al. 2019; Y. Chen, Zhu, et al. 2024; Y. Zhu, Chen, et al. 2023). For more complex FHMOs, however, additional challenges remain, including pathway length, competing glycosylation reactions, transport of larger acceptors, reduced yields, and more difficult purification (Hu et al. 2022; C. Li, Li, et al. 2023; J. Wang et al. 2025; L. Yang et al. 2026). For a detailed product‐level comparison, recent enzymatic and microbial biosynthesis studies of individual FHMOs have been comprehensively tabulated by M. Li, Yao, et al. (2026), including enzyme source, host strain, pathway design, genetic modification strategy, substrates, conversion yield, productivity, and fermentation titer.

4.4. Plant Factory

Plant factories have emerged as promising biomanufacturing platforms for the heterologous production of medium‐ and high‐value complex natural products (Bally et al. 2018). Their core strategy is to introduce biosynthesis‐related gene clusters or key enzyme‐encoding genes into plant hosts and exploit endogenous nucleotide sugar pools and protein expression systems to support pathway reconstruction (Golubova et al. 2024). Nicotiana benthamiana is widely used for transient expression because of its high susceptibility to Agrobacterium tumefaciens‐mediated infiltration, which enables rapid and high‐level expression of exogenous genes (Bally et al. 2018; Delatte et al. 2018; Nester 2015). This platform has been successfully applied to the production of pharmaceutical proteins (Bally et al. 2018), vaccines (Martin et al. 2024), terpenoids (Cankar et al. 2014; Delatte et al. 2018), alkaloids (Dudley et al. 2022; Nett et al. 2020), cannabinoids (Berman et al. 2023), triterpenes (Bibik et al. 2022; Reed et al. 2017), taxane intermediates (Y. Zhang, Wiese, et al. 2023), and other structurally complex natural products (Irmisch et al. 2020; Kruse et al. 2023; Pan et al. 2021; Selma et al. 2022; Takahashi et al. 2020), demonstrating its capacity for multigene pathway reconstruction and glycosylated metabolite biosynthesis. These advances suggest that plant hosts, particularly N. benthamiana, may also be suitable for producing FHMOs. Plants offer several potential advantages, including scalable cultivation, short growth cycles, mature transformation systems, and endogenous nucleotide sugar metabolism. In principle, plant cells can provide key sugar donors, such as UDP‐galactose, UDP‐N‐acetylglucosamine, GDP‐fucose, and CMP‐sialic acid, thereby supporting diverse HMO glycosylation reactions. Once fucosyltransferases and upstream HMO core pathway enzymes are introduced, plant cells can potentially synthesize and accumulate target FHMOs in leaf tissues.

Plant‐based HMO production has been demonstrated by transiently expressing bacterial glycosyltransferases in the cytoplasm of N. benthamiana. Co‐expression of core glycosyltransferases enabled the synthesis of lactose, LNT, LNnT, and higher oligosaccharide isomers, while introduction of α1,2‐fucosyltransferase Te2FT generated 2′‐FL, LNFP I, and other FHMO isomers. Reconstruction of CMP‐Neu5Ac biosynthesis further allowed the production of acidic HMOs, including 3′‐SL, 6′‐SL, sialyllacto‐N‐tetraose c (LSTc), and sialyllacto‐N‐tetraose d (LSTd). Enhanced GDP‐fucose biosynthesis increased LNFP I accumulation, indicating that donor supply is a key limitation. Plant‐derived HMO mixtures promoted Bifidobacterium longum subsp. infantis growth, and techno‐economic analysis suggested potential competitiveness under optimized biorefinery scenarios (Barnum et al. 2024).

Compared with microbial fermentation, plant factories may offer advantages such as low feedstock cost, reduced requirements for sterile fermentation equipment, scalable biomass production, and a lower risk of endotoxin contamination (Barnum et al. 2024; Q. Chen et al. 2025). However, this technology remains at an early stage for HMO production. Current challenges include low product accumulation, metabolic complexity, unexpected endogenous glycosylation or acylation, product heterogeneity, limited multigene expression stability, transgene silencing, and uncertainty in large‐scale field performance (Bally et al. 2018; Barnum et al. 2024; Golubova et al. 2024; Nester 2015). Moreover, although techno‐economic models indicate potential cost advantages, commercial feasibility still requires validation through stable high‐yielding lines, optimized downstream purification, and regulatory assessment (Barnum et al. 2024; Q. Chen et al. 2025). Overall, the plant factory has demonstrated technical feasibility for the biosynthesis of FHMOs and may become a green and sustainable complementary production route. Future efforts should focus on reconstructing more FHMO pathways through multigene co‐expression, enhancing nucleotide sugar supply, and using genome editing to eliminate competing endogenous glycosyltransferase activities. These strategies may enable the low‐cost production of high‐purity FHMOs for infant formula and functional nutrition applications.

4.5. Computational and Synthetic Biology Tools and Emerging AI/ML Opportunities

In recent years, the rapid development of AI has provided new strategies for overcoming current bottlenecks in HMO production. Unlike conventional optimization approaches that largely rely on empirical trial and error, AI‐driven strategies integrating data‐driven high‐throughput screening, structural prediction, and dynamic regulatory modeling are shifting FHMO production from trial‐and‐error development toward predictive design (S. Lin et al. 2026). Several recent studies have provided methodological foundations for this transition (Endo et al. 2024; Fang, Gao, Mund, et al. 2025; Y. M. Tan et al. 2019; M. Xu et al. 2023; W. Yu et al. 2022; Q. Zhang, Xu, et al. 2025). A fluorescence‐activated cell sorting‐based single‐cell ultrahigh‐throughput screening platform was developed for the directed evolution of α1,3‐fucosyltransferase, enabling the screening of more than 10⁷ mutants per hour and yielding variants with markedly improved catalytic efficiency for 3‐FL synthesis. This study demonstrates that high‐throughput screening can accelerate the discovery of fucosyltransferase variants suitable for FHMO biosynthesis (Y. M. Tan et al. 2019). In silico structural modeling, docking simulation, and enzyme engineering were further combined to redesign α1,2‐fucosyltransferase, thereby reducing the undesired formation of 2′‐FL during LNFP I production while maintaining efficient pentasaccharide synthesis (Endo et al. 2024). Similarly, consensus‐guided engineering, directed evolution, and molecular dynamics simulation were integrated to improve the thermostability of α‐l‐fucosidase, providing a rational framework for engineering enzymes used in 2′‐FL synthesis (Zhao et al. 2025). It should be noted that high‐throughput screening, molecular docking, molecular dynamics simulations, and synthetic regulatory circuits are experimental, computational, or synthetic biology tools but do not necessarily constitute AI. Genuine AI‐assisted approaches could use data‐trained models to predict enzyme activity, substrate specificity, pathway behavior, or fermentation performance and thereby prioritize experimental designs within iterative design‐build‐test‐learn cycles. However, their application to FHMO production remains limited by the scarcity and heterogeneity of datasets across enzyme variants, host strains, cultivation conditions, and analytical methods. At present, AI should therefore be viewed as a complementary tool for guiding experimental optimization rather than as an established FHMO production technology.

In addition, AI can support techno‐economic analysis by integrating production cost, yield, downstream purification difficulty, and minimum selling price models, thereby helping to compare the economic competitiveness of plant factory and microbial fermentation platforms for HMO production. Rather than replacing existing synthetic strategies, AI should be viewed as an enabling tool for optimizing current production routes. By coordinating enzyme properties, metabolic networks, chassis‐specific features, and process parameters within a unified digital framework, AI‐assisted strategies may facilitate the rational design of efficient, green, and scalable FHMO biosynthetic platforms. However, their practical value depends on the quality, scale, and standardization of experimental datasets, which are currently fragmented across hosts, enzymes, media, analytical methods, and reporting formats. Moreover, predicted pathway performance does not guarantee food‐grade quality, regulatory acceptability, or robust scale‐up. Therefore, AI should complement, rather than replace, experimental validation, safety assessment, and quality‐control testing.

5. Safety Regulation and Commercial Translation

As naturally occurring components of human milk, FHMOs require a safety and regulatory framework distinct from that applied to conventional nutrients when they are intended for use in infant formula (Casavale et al. 2019; Donovan et al. 2023; Moore et al. 2025; Morissette et al. 2023; Wichmann et al. 2026). This requirement arises not only because the target population is infants but also because commercial products must be obtained through specific manufacturing processes, including microbial fermentation, enzymatic synthesis, or other controlled biomanufacturing routes (EFSA 2011; EFSA et al. 2025; EFSA Panel on Additives and Products or Substances used in Animal Feed [FEEDAP] et al. 2018). Therefore, safety assessment should consider not only potential risks introduced during production, including host cell residues such as E. coli endotoxins, process‐related by‐products, residual organic solvents, and nonnatural isomers but also storage safety and stability during the shelf life of finished products, which may differ from those of conventional food ingredients (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2021, 2022a, 2022b). At present, regulatory decisions on FHMOs in most countries and regions are generally based on a weight‐of‐evidence approach. Key elements include ingredient identification and specifications, description of the manufacturing process, compositional analysis, stability data, dietary exposure assessment, toxicological studies, genotoxicity evaluation, subchronic feeding studies, allergenicity considerations, and clinical tolerance data (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2023a, 2023b; Morissette et al. 2023).

However, global regulatory frameworks are not yet fully harmonized. In the United States, some HMO ingredients have been evaluated through the Generally Recognized as Safe (GRAS) notification pathway, and the Food and Drug Administration (FDA) may issue no‐objection letters for specific intended uses and use levels (Morissette et al. 2023; Wichmann et al. 2026). In the European Union, HMOs are regulated as novel foods and, once authorized, are included in the Union list with defined conditions of use, labeling requirements, and specifications (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2021, 2022a, 2022b, 2022c, 2023b, 2023c, 2023d). In China, 2′‐FL has been approved as a nutrient fortifier for specified categories of infant and young child foods (National Health Commission of the People's Republic of China 2023; H. Hou et al. 2024; Wichmann et al. 2026). These regional differences largely reflect differences among competent authorities and policy frameworks in food additive regulations, infant nutrition standards, and approval procedures for novel ingredients. They are also closely related to the ingredient source, manufacturing method, purity specifications, intended food categories, and maximum permitted use levels (Casavale et al. 2019; Donovan et al. 2023; Morissette et al. 2023; Nowak‐Wegrzyn et al. 2019; Wichmann et al. 2026). Consequently, safety conclusions, manufacturing processes, or purification strategies established for a given FHMO preparation in one country or region cannot be automatically extrapolated to other regulatory contexts.

At present, there is no globally unified framework governing the approval status, permitted types, applicable age groups, maximum use levels, or labeling requirements of FHMOs (Donovan et al. 2023; Morissette et al. 2023; Wichmann et al. 2026). This lack of harmonization may, to some extent, affect the global commercialization of FHMOs. Therefore, a systematic review of regulatory policies and approval progress for FHMOs in major countries and regions, together with an analysis of the core elements of safety assessment and the common challenges and regional differences across regulatory frameworks, would provide important scientific support for their global application, regulatory decision‐making, and industrial deployment in infant nutrition.

5.1. Preparation‐Specific Regulatory Pathways and Global Authorization Status

The commercialization of FHMOs largely depends on regulatory recognition of the safety and compliance of specific preparations. Regulatory status should not be regarded as an inherent property of an HMO compound name or of HMOs as a general ingredient category. Instead, each regulatory decision applies to a defined preparation characterized by its production organism, manufacturing process, compositional specifications, purity and impurity profile, intended food category, target population, maximum use level, and conditions of use. Accordingly, the term “authorized” is used in this section only in relation to the specific preparation and regulatory jurisdiction evaluated. Among FHMOs, 2′‐FL is the most extensively reviewed and widely commercialized compound worldwide. In the United States, specific 2′‐FL preparations have been evaluated through individual GRAS notices. Subsequently, several companies, including Glycom A/S, DuPont, and Glycosyn, submitted separate GRAS notices for 2′‐FL preparations produced by fermentation using different genetically engineered strains, and the FDA issued “no questions” letters for the respective notices under their specified manufacturing conditions, intended uses, and use levels (Morissette et al. 2023; Y. Zhu et al. 2020; Y. Y. Zhu, Wan, et al. 2022). In the European Union, specific 2′‐FL preparations are regulated as novel foods under Regulation (EU) 2015/2283 and are included in the Union list established by Commission Implementing Regulation (EU) 2017/2470. The corresponding entries define the production process or microorganism, compositional specifications, authorized food categories, maximum use levels, and labeling requirements (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2022a, 2022c, 2023b, 2023c, 2023d). In China, 2′‐FL was initially authorized as a nutritional fortifier in October 2023 for defined infant and young‐child food categories, subject to specified production‐source, quality, and use‐level requirements (National Health Commission of the People's Republic of China 2023). Australia, New Zealand, Japan, South Korea, and other countries or regions have also evaluated individual HMO preparations through their respective regulatory pathways. Figure 5 summarizes the global regulatory landscape of HMOs in infant formula; however, the information should be interpreted as time‐, jurisdiction‐, preparation‐, and use‐specific rather than as a general authorization of an HMO compound category.

FIGURE 5.

Global regulatory pathways and authorization status of FHMOs in infant formula.

However, not all professional organizations have adopted a fully positive view toward the commercial use of formulas supplemented with manufactured FHMO preparations. In a 2022 position statement, the Nutrition Committee of the German Society for Pediatric and Adolescent Medicine (DGKJ) stated that current evidence is insufficient to demonstrate clinically relevant health benefits of infant formulas supplemented with synthetic HMOs and therefore did not recommend the preferential use of such products (Bührer et al. 2022). The committee emphasized that the oligosaccharide profile of human milk is highly complex and varies substantially among individuals, whereas current formulas contain only a limited number of simple short‐chain HMOs and therefore cannot reproduce the true complexity of HMOs. The committee also strongly criticized the use of scientific terms such as “human milk oligosaccharides” in product advertising, arguing that such claims may imply an unfounded similarity between formula and human milk, mislead consumers, and undermine breastfeeding promotion. It therefore called on manufacturers to discontinue inappropriate marketing practices and urged regulatory authorities to strengthen supervision to ensure that infant formula advertising complies with relevant laws and regulations.

Despite expanding regulatory acceptance of specific preparations, professional and scientific organizations have emphasized the need for cautious interpretation of marketing claims for formulas supplemented with synthetic or nature‐identical HMOs. Such caution does not necessarily imply that an authorized preparation is unsafe under its defined conditions of use; rather, it reflects the distinction between a preparation‐specific safety authorization and evidence for particular functional benefits. For infant formula, claims related to immune protection, allergy prevention, cognitive development, or disease‐risk reduction require a higher level of clinical evidence than regulatory safety authorization alone. Therefore, the authorization of a defined HMO preparation should be discussed separately from efficacy claims and marketing communication.

5.2. Preparation‐ and Process‐Specific Safety Assessment and Exposure Framework

Safety assessment of a FHMO preparation should integrate ingredient identity, compositional specifications, production organism, manufacturing process, impurity profile, toxicological data, clinical tolerance, and realistic dietary exposure (Donovan et al. 2023; Morissette et al. 2023; Wichmann et al. 2026). Preclinical studies on 2′‐FL and 3‐FL have generally shown no genotoxicity or treatment‐related adverse effects at high tested doses in standard toxicological models. Such studies provide important support for regulatory evaluation by establishing no‐observed adverse‐effect levels and margins of exposure (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2021, 2022a, 2022b, 2022c, 2023a, 2023b, 2023c, 2023d). However, toxicological safety alone is not sufficient for infant formula application. Because the target population is physiologically sensitive, clinical tolerance, growth adequacy, stool characteristics, gastrointestinal symptoms, and adverse event profiles should also be considered (Alliet et al. 2022; Donovan et al. 2023; Jochum et al. 2023).

The available safety evidence is strongest for 2′‐FL, which has been evaluated in multiple toxicological and clinical contexts (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2022a, 2022c, 2023b, 2023c, 2023d; Neumer et al. 2021; Nowak‐Wegrzyn et al. 2019; Wichmann et al. 2026). By comparison, evidence for 3‐FL, structurally more complex FHMOs, and multi‐HMO preparations remains less extensive and should not be inferred directly from the evidence available for 2′‐FL (EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2021, 2022b, 2023a; H. Hou et al. 2024; Z. Li, Zhu, et al. 2023; Pitt et al. 2024). Nevertheless, safety conclusions are preparation‐specific. Differences in production host, purification method, residual impurities, isomeric profile, and compositional specifications may influence regulatory evaluation, even when the target oligosaccharide has the same chemical name (EFSA et al. 2025; Morissette et al. 2023; Wichmann et al. 2026). Consequently, safety conclusions obtained for one preparation should not be automatically extrapolated to another preparation containing the same nominal HMO.

Clinical tolerance studies are particularly important for infant formula applications. Relevant endpoints include formula intake, growth parameters, stool frequency and consistency, gastrointestinal tolerance, regurgitation, crying or fussiness, infection‐related adverse events, and standard biochemical or hematological markers when appropriate (Donovan et al. 2023; Morissette et al. 2023). Existing infant studies generally suggest that defined formula products containing selected HMO preparations at the investigated use levels are well tolerated, but many studies are limited by short duration, specific product formulations, relatively small sample sizes, or focus on healthy term infants (Alliet et al. 2022; Jochum et al. 2023; Nowak‐Wegrzyn et al. 2019). Data for preterm infants, low birth weight infants, infants with medical conditions, and infants receiving multiple HMO combinations remain comparatively limited (Donovan et al. 2023; Morissette et al. 2023; Wichmann et al. 2026). These results should therefore be interpreted as formulation‐, dose‐, duration‐, and population‐specific rather than as evidence of class‐wide clinical tolerance.

Exposure assessment should consider not only the concentration of added HMOs in formula but also total daily intake, age‐specific formula consumption, body weight, background exposure from human milk or mixed feeding, and potential combined exposure from multiple fortified foods (Donovan et al. 2023; H. Hou et al. 2024; G. Luo, Zhu, et al. 2023; Wichmann et al. 2026). This is particularly relevant as the market moves from single‐HMO supplementation toward blends containing 2′‐FL, 3‐FL, LNT, LNnT, 3′‐SL,6′‐SL, and other emerging HMOs (H. Hou et al. 2024; Wichmann et al. 2026). Although such blends may better approximate the compositional complexity of human milk, they also require evaluation of cumulative exposure, gastrointestinal tolerance, and possible interactions among ingredients (Donovan et al. 2023). Exposure estimates should therefore be derived for the defined preparation, intended food categories, target age group, proposed maximum use level, and anticipated consumption pattern rather than from the HMO name alone.

In addition, under conventional food safety regulatory frameworks, FHMOs produced by engineered microbial fermentation require process‐specific regulatory consideration, as the production strains are generally constructed through the introduction of heterologous genes and deliberate genetic modifications. These genetic modifications enable the host microorganisms to synthesize target HMOs, but also require characterization of potential process‐related hazards, including the presence of virulence determinants or transferable antimicrobial‐resistance genes in the production strain and the possible carry‐over of viable cells or recombinant DNA into the final preparation (EFSA 2011; EFSA et al. 2025; EFSA Panel on Additives and Products or Substances used in Animal Feed [FEEDAP] et al. 2018). Although the final products obtained by engineered microbial fermentation may be structurally identical to natural FHMOs in human milk, regulatory authorities must still conduct additional assessments of both the safety of the genetically modified production strain and the potential risks introduced during the manufacturing process (EFSA et al. 2025; EFSA Panel on Nutrition, Novel Foods and Food Allergens [NDA] et al. 2022b, 2023b). Therefore, safety evaluation of such products cannot rely solely on structural equivalence to naturally occurring human milk components but should be based on a process‐oriented regulatory framework focused on risk control during production (EFSA et al. 2025; Wichmann et al. 2026).

Specifically, depending on the jurisdiction, product category, and characteristics of the defined preparation, regulatory dossier requirements for FHMOs produced by engineered microbial fermentation may include, but are not limited to, the following aspects. First, at the production strain level, regulatory authorities may require whole‐genome sequencing of genetically modified microorganisms to systematically assess their genetic safety. This approach aims to identify potential virulence factor genes and mobile antibiotic resistance genes at the genomic level, ensuring that the production strain does not possess pathogenic potential or the ability to disseminate antimicrobial resistance (EFSA 2011; EFSA et al. 2025; EFSA Panel on Additives and Products or Substances used in Animal Feed [FEEDAP] et al. 2018). Second, at the product classification level, if the engineered strain is used only as a processing aid and adequate purification ensures that the final product contains neither viable cells nor recombinant DNA, the FHMO produced by fermentation may not be classified as a genetically modified food and may therefore be exempt from relevant GMO food regulations, depending on the regulatory jurisdiction and product category (EFSA et al. 2025; Wichmann et al. 2026). Finally, at the product purity level, the absence of viable genetically modified production cells should be demonstrated using suitably sensitive culture‐based methods, including conditions that permit the recovery of injured cells. In addition, residual DNA derived from the production strain should remain below relevant detection or specification thresholds; otherwise, it may require further risk assessment for consumers and the environment (EFSA et al. 2025; EFSA Panel on Additives and Products or Substances used in Animal Feed [FEEDAP] et al. 2018).

Overall, current toxicological and clinical evidence supports the safety of selected FHMO preparations under their defined and authorized conditions of use (Alliet et al. 2022; Donovan et al. 2023; Jochum et al. 2023). However, future safety evaluations should move beyond single‐ingredient assessments and address multi‐HMO blends, vulnerable infant populations, longer intervention periods, post‐market surveillance, and matrix‐dependent stability (Donovan et al. 2023; Morissette et al. 2023; Wichmann et al. 2026). This broader safety framework will be essential for responsible commercial expansion.

5.3. Dose‒Response Evidence and Use‐Level Considerations in Infant Formula

Building on the preparation‐ and process‐specific safety framework described in Section 5.2, this subsection focuses on quantitative dose‒response evidence, margins of exposure, clinical tolerance at defined intake levels, and the derivation of appropriate use levels for infant formula. Although food safety authorities worldwide have gradually authorized specific FHMO preparations, such as 2′‐FL, for use in infant formula, their optimal supplementation levels and long‐term oral safety still need to be clarified through systematic preclinical and clinical studies (Ramirez‐Farias et al. 2021; Song et al. 2026; Xing et al. 2026; J. Xu et al. 2025). Most available clinical studies have compared one supplemented formula with a control rather than evaluating multiple graded doses. Consequently, minimum effective levels, dose‐dependent functional responses, and upper intake boundaries cannot yet be defined consistently across compounds or preparations. Claims that a low dose is ineffective or that a higher dose disrupts the infant gut microbiota should therefore be regarded as hypotheses requiring direct dose‒response evidence rather than as established effects (Donovan et al. 2023; H. Hou et al. 2024; Jochum et al. 2023; G. Luo, Zhu, et al. 2023). Therefore, systematically evaluating defined FHMO preparations at different doses and establishing the no‐observed‐adverse‐effect level (NOAEL) and margin of exposure (MoE) are essential for evidence‐based use‐level determination.

Preclinical toxicological studies have provided important evidence for the safety of FHMOs, particularly 2′‐FL and 3‐FL (Coulet et al. 2014; Parschat et al. 2020; Phipps et al. 2022; Pitt et al. 2019, 2024; van Berlo et al. 2018). Synthetic, nature‐identical 2′‐FL has been evaluated using Ames tests, mouse lymphoma assays, and a 90‐day oral gavage study in juvenile rats, with no evidence of genotoxicity or treatment‐related adverse effects at doses up to 5000 mg/kg bw/day, which was established as the NOAEL (Coulet et al. 2014). Biotechnologically produced 2′‐FL has also been assessed in a 13‐week oral toxicity study, in which dietary exposure up to 10% 2′‐FL, corresponding to approximately 7.25 and 7.76 g/kg bw/day in male and female juvenile rats, respectively, caused no adverse effects (van Berlo et al. 2018). Similarly, a mixture of five HMOs, including 2′‐FL, 3‐FL, LNT, 3′‐SL, and 6′‐SL, was not genotoxic and produced no adverse effects in rats, with a NOAEL of 10% in the diet (Parschat et al. 2020). Because this intervention was a five‐HMO mixture, its NOAEL applies to the tested preparation and cannot be attributed independently to either 2′‐FL or 3‐FL. For 3‐FL, biotechnologically produced 3‐FL has been evaluated using acute oral toxicity, in vitro and in vivo genotoxicity assays, and a subchronic rat feeding study. The results showed no biologically relevant adverse effects at dietary levels up to 10%, corresponding to average daily intakes of approximately 5.98 and 7.27 g/kg bw/day in male and female rats, respectively (Pitt et al. 2019). The absence of genotoxicity and subchronic toxicity has also been confirmed for human‐identical 3‐FL in neonatal rats, with a NOAEL of 4000 mg/kg bw/day, the highest dose tested (Phipps et al. 2022). In addition, a 21‐day safety study in neonatal farm piglets found that 3‐FL at approximately 246–494 mg/kg/day had no adverse effects on growth, feed intake, clinical observations, or pathological endpoints, providing preparation‐specific evidence of tolerance within the tested dose range (Pitt et al. 2024). However, numerical NOAELs obtained across these studies should not be compared directly without considering differences in the test preparation, species, developmental stage, administration route, exposure duration, and toxicological design.

Human clinical studies have also supported the safety and tolerability of defined HMO‐containing interventions at the investigated intake levels (Alliet et al. 2022; Elison et al. 2016; Lasekan et al. 2022; Parschat et al. 2021; Puccio et al. 2017; Schonknecht et al. 2023). A randomized, double‐blind, placebo‐controlled trial in 100 healthy adults showed that oral supplementation with 2′‐FL and/or LNnT at doses up to 20 g/day for 2 weeks was safe, well tolerated, and selectively increased the relative abundance of Bifidobacterium (Elison et al. 2016). However, adult tolerance at this intake cannot be used directly to define an appropriate dose for infants because of differences in body weight, gastrointestinal development, background diet, and exposure duration. In infants, formula supplemented with 1.0 g/L 2′‐FL and 0.5 g/L LNnT supported normal growth and was well tolerated, with secondary outcomes suggesting reduced parent‐reported respiratory morbidity and medication use (Puccio et al. 2017). These findings apply to the combined formulation and do not isolate the contribution of 2′‐FL. A probiotic‐containing infant formula supplemented with 1.0 g/L 2′‐FL further supported age‐appropriate growth, showed gastrointestinal tolerance comparable to the control formula, and shifted the gut microbiota composition toward that of breastfed infants (Alliet et al. 2022). Because this intervention combined 2′‐FL with a probiotic, its microbiota‐related outcomes cannot be attributed specifically to 2′‐FL. Clinical evidence has also been extended to HMO mixtures. A 5HMO mixture containing 2′‐FL, 3‐FL, LNT, 3′‐SL, and 6′‐SL was evaluated in healthy term infants over 16 weeks, with no differences in weight, length, or head circumference gain compared with the control formula group; the HMO‐supplemented group also showed softer stools and higher stool frequency, more closely resembling breastfed infants (Parschat et al. 2021). Milk‐based formula supplemented with the same five HMOs similarly supported normal growth and showed good gastrointestinal tolerance in healthy term infants (Lasekan et al. 2022). These studies support the tolerance of the tested multi‐HMO preparations but do not establish compound‐specific dose‒response relationships for their fucosylated components. A systematic review concluded that manufactured HMOs tested in clinical studies were consistently safe and well tolerated, while infant studies generally showed shifts in stool characteristics, gut microbiota, and immune‐related markers toward patterns observed in breastfed infants (Schonknecht et al. 2023). Nevertheless, these findings remain specific to the preparations, doses, formulations, populations, and intervention periods included in the reviewed studies.

5.4. Processing Stability and Safety of FHMOs Added to Infant Formula

As functional nutritional fortifiers in infant formula, the actual content and biological efficacy of FHMOs in final products are closely associated with their stability during processing, storage, and shelf life (Donovan et al. 2023; G. Luo, Zhu, et al. 2023; Wichmann et al. 2026). Unlike naturally occurring HMOs in human milk, FHMOs added to formula must undergo multiple processing steps, including ingredient blending, homogenization, heat treatment such as pasteurization or ultrahigh‐temperature sterilization, spray drying, and subsequent storage (Morissette et al. 2023; Wichmann et al. 2026). These conditions may induce thermal degradation, moisture absorption, caking, crystallization, morphological transitions, or the formation of heat‐induced by‐products, such as Maillard reaction products (MRPs), thereby affecting their prebiotic efficacy in the infant gut (J. Ahn, Choi, et al. 2025; Marousez et al. 2022). Therefore, evaluating the processing stability of FHMOs throughout infant formula production, identifying key processing parameters that influence their retention, and characterizing potential by‐products and their safety are essential for optimizing manufacturing conditions, defining storage requirements, and ensuring the functional consistency and safety of the final products.

5.4.1. Thermal Stability and Hygroscopic Properties of FHMOs

The thermal stability and moisture‐related physical stability of FHMOs are critical determinants of their processability, powder handling, and shelf‐life performance in infant formula systems. Among FHMOs, 2′‐FL has been the most extensively investigated. Differential scanning calorimetry and thermogravimetric analysis showed that crystalline 2′‐FL undergoes dehydration at approximately 143.4°C and melting at approximately 230.6°C, whereas amorphous 2′‐FL exhibits glass transition, crystallization, and subsequent melting at approximately 127.6°C, 192.8°C, and 230.6°C, respectively (S. Ahn and Chung 2023). Compared with lactose, 2′‐FL shows higher glass‐transition and crystallization temperatures and a slightly higher thermal decomposition temperature, indicating its relatively greater resistance to thermally induced molecular mobility and degradation during processing.

However, the stability of 2′‐FL is strongly dependent on its physical state and environmental water activity. Water acts as an efficient plasticizer for amorphous 2′‐FL, reducing its glass‐transition temperature from 127.6°C to 36.5°C as water activity increases from 0 to 0.53, thereby increasing molecular mobility and promoting structural collapse, stickiness, caking, and moisture‐induced crystallization during storage (S. Ahn and Chung 2023). Further moisture sorption analysis demonstrated that crystalline 2′‐FL is more susceptible to moisture uptake than crystalline α‐lactose monohydrate, with a lower deliquescence point and markedly higher water solubility, suggesting a greater tendency toward surface dissolution and capillary condensation under humid conditions (S. Ahn, Chung, et al. 2025). For amorphous 2′‐FL, moisture‐induced crystallization occurs at higher water activity than that of amorphous lactose, which has been attributed to the higher molar mass and lower molecular mobility of 2′‐FL. Moreover, the sticky point temperature of amorphous 2′‐FL decreases substantially with increasing water activity, confirming that humidity control is essential to prevent powder agglomeration and maintain flowability during storage and handling.

Overall, available evidence suggests that FHMOs such as 2′‐FL are sufficiently heat‐stable to tolerate common thermal operations used in infant formula manufacture, including pasteurization, concentration, and spray drying. Nevertheless, their post‐processing stability is highly dependent on moisture control. Therefore, low‐water‐activity processing environments, moisture‐barrier packaging, and appropriate control of powder crystallinity should be considered key strategies for preserving the physicochemical quality and functional integrity of 2′‐FL‐fortified infant formula products.

5.4.2. Maillard Reactions Involving FHMOs and the Safety of Their Reaction Products

During the manufacture and storage of infant formula powders, FHMOs may participate in Maillard reactions because they contain a reducing end that can react with free amino groups in proteins or amino acids (J. Ahn, Choi, et al. 2025; G. Luo, Zhu, et al. 2023; Marousez et al. 2022; Peled et al. 2024; Tu et al. 2022). This reaction may generate a broad spectrum of MRPs, including early glycation products, α‐dicarbonyl intermediates, fluorescent compounds, brown pigments, volatile compounds, and advanced glycation end‐products (AGEs), thereby influencing the nutritional quality, sensory attributes, functionality, and safety of the final products (J. Ahn, Choi, et al. 2025; Marousez et al. 2022; Peled et al. 2024; Tu et al. 2022; J. Xu et al. 2024).

Among FHMOs, 2′‐FL reacts with α‐lactalbumin under dry heating, forming Maillard conjugates with decreased free amino groups and high‐molecular‐weight products. Conjugation preserves the α‐LA secondary structure and markedly enhances antioxidant activity, increasing DPPH radical scavenging and FRAP reducing power (Tu et al. 2020). Maillard conjugation between 2′‐FL and β‐lactoglobulin under accelerated dry heating reduced free amino groups, promoted high‐molecular‐weight conjugate formation, and significantly decreased β‐LG antigenicity, probably by masking lysine‐containing epitopes. Meanwhile, the conjugates showed enhanced DPPH radical‐scavenging activity and FRAP reducing capacity (Tu et al. 2022). Small‐molecule Maillard intermediates are also important for product safety. Compared with monosaccharides, 2′‐FL produced lower levels of α‐dicarbonyl compounds and fewer pyrazines, suggesting reduced formation of reactive intermediates. Its volatile profile was mainly dominated by furans and furan derivatives (J. Ahn, Choi, et al. 2025). Maillard reactions of 2′‐FL may reduce lysine availability, lowering protein nutritional quality, and generate AGEs such as furosine, N ^ε‐carboxymethyllysine (CML), and N ^ε‐carboxyethyllysine (CEL). Their biological effects in early life remain unclear. High hydrostatic pressure (HHP) preserves HMOs and limits MRP formation, highlighting the need to control heat‐induced glycation in infant nutrition (Marousez et al. 2022). On the other hand, selected 2′‐FL‐derived conjugates may provide functional advantages, including enhanced antioxidant activity, reduced β‐LG antigenicity, and even beneficial modulation of gut microbiota when 2′‐FL is conjugated with lactoferrin hydrolysates (Peled et al. 2024).

Overall, current evidence suggests that 2′‐FL‐mediated Maillard reactions should be regarded as both a potential quality risk and a possible functional modification pathway. Current studies use simplified, short‐term models that do not replicate commercial formula conditions. Future work should quantify MRPs and AGEs in real formulas and assess digestibility, bioavailability, and the metabolic/toxicological impact of markers such as CML and CEL in infants.

5.4.3. Effects of Thermal and Nonthermal Processing on FHMOs

Conventional thermal processing, such as Holder pasteurization (HoP), is effective for microbial inactivation but may promote protein denaturation and MRP formation in milk systems (Donovan et al. 2023; G. Luo, Zhu, et al. 2023). In this context, nonthermal technologies, particularly HHP, have attracted increasing attention as alternative processing strategies because they can improve microbiological safety while better preserving heat‐sensitive bioactive components. High‐pressure processing and HoP have been compared for their effects on 22 HMOs, including several FHMOs such as 2′‐FL, 3‐FL, LNFP I, II, III, and V, LDFT, LNDFH, monofucosyllacto‐N‐hexaose III (MFLNH III), and difucosyllacto‐N‐hexaose a (DFLNHa). Both HHP and HoP preserved HMOs without significantly altering their concentrations. However, HoP markedly increased MRPs and AGEs, including furosine, lactuloselysine, CML, and CEL, potentially reducing lysine bioavailability. In contrast, HHP limited these modifications, offering a nonthermal strategy to maintain biochemical quality in human milk or HMO‐fortified dairy matrices while minimizing potentially undesirable glycation compounds, which is particularly relevant for infant nutrition (Marousez et al. 2022).

6. Conclusions and Future Prospects

FHMOs should not be interpreted as a functionally interchangeable group. The most developed evidence base remains concentrated on 2′‐FL, for which preparation‐specific safety and short‐term tolerance at authorized uses are reasonably well supported. Its effects on microbial utilization and pathogen‒ interactions are also supported by convergent mechanistic and preclinical evidence, although demonstrated clinical benefits remain dependent on the dose, study population, endpoint, background formula, and baseline microbiota. The evidence for 3‐FL is less extensive, while that for DFL, LNFP isomers, LNDFHs, and other structurally complex FHMOs remains preliminary. Many proposed effects involving necrotizing enterocolitis, allergy, immune regulation, and neurodevelopment are derived primarily from cell experiments, animal models, observational associations, or interventions using HMO mixtures and multicomponent formulas. These findings cannot yet be attributed to individual FHMOs or interpreted as evidence of disease prevention. Moreover, the limited systemic absorption of intact HMOs indicates that extraintestinal effects may involve both microbiota‐dependent and microbiota‐independent pathways. Future health claims should therefore remain aligned with the identity of the compound and the level of evidence supporting each outcome.

The technological maturity of FHMO production is similarly product‐dependent. Microbial production of 2′‐FL has reached high titers and substantial commercial maturity, whereas 3‐FL, DFL, LNFPs, and LNDFHs generally require more complex control of acceptor synthesis, fucosyltransferase specificity, precursor supply, intermediate accumulation, and isomeric purity. Enzymatic and chemoenzymatic systems remain complementary options for structurally complex or low‐volume products because of their selectivity, although donor cost, enzyme reuse, and volumetric productivity remain important limitations. Future cell factory development should integrate genome‐stable pathway construction, dynamic and spatial regulation, enzyme engineering, and validated genome‐scale or data‐trained predictive models. However, production platforms should not be evaluated by fermentation titer alone. Product purity, regioisomer distribution, carbon conversion, genetic stability, downstream recovery, impurity burden, water and energy demand, and batch‐to‐batch consistency are equally important indicators of industrial readiness.

The next stage of FHMO research requires a compound‐resolved and translation‐oriented framework. Biological studies should clearly distinguish purified individual FHMOs from HMO mixtures, synbiotics, and multicomponent formulas and should account for dose, infant age, gestational status, maternal secretor and Lewis phenotype, baseline microbiota, and dietary context. Controlled clinical studies should prioritize predefined, clinically relevant outcomes and include long‐term follow‐up and vulnerable populations only after appropriate safety evidence has been established. Safety and regulatory assessment must remain linked to a defined preparation, production organism, manufacturing process, specification, exposure level, and intended use rather than to an “approved HMO” category. Processing studies should also quantify degradation, isomerization, moisture uptake, MRPs, and nutrient interactions in real infant formula matrices. Rational FHMO blends may ultimately provide greater functional breadth than single compounds, but their development should be based on component‐specific attribution, interaction testing, exposure assessment, and clinical validation rather than compositional resemblance to human milk alone. The decisive benchmark for future translation will therefore be the establishment of a reproducible chain of evidence connecting molecular structure, manufacturing quality, matrix stability, biological mechanism, and clinically meaningful benefit.

Author Contributions

Jinghong Xu: writing – original draft, validation, project administration, methodology, investigation, formal analysis, data curation, conceptualization. Bowen Li: writing – review and editing, project administration, funding acquisition. Huajian Xu: writing – review and editing, conceptualization. Qing Zhang: writing – review and editing, data curation. Zheng Li: writing – review and editing, investigation. Xiaocui Zhong: writing – review and editing, validation. Dongling Qiao: writing – review and editing, conceptualization. Ting‐Li Han: writing – review and editing, validation. Binjia Zhang: writing – review and editing, formal analysis, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data will be made available on request.