Human milk oligosaccharides are among the most structurally intricate carbohydrates in biology, and the subset bearing terminal sialic acids is especially valuable for studying infant nutrition, immune development, and host–pathogen recognition. A new study published in Nature Communications reports a general strategy for assembling sialylated human milk oligosaccharides (HMOs) by automated glycan assembly, removing a long-standing bottleneck in chemical sialylation on solid support. The work makes collections of pure, biologically relevant sialylated glycans accessible on practical timescales, with direct implications for glycobiology research and for how the field sources these challenging molecules.

Why Pure Sialylated Glycans Are So Elusive

Sialic Acid caps the non-reducing ends of many glycans on glycoproteins, glycolipids, and glycoRNAs, and it mediates processes from cell differentiation and immune regulation to pathogen adhesion. The most common member, N-acetylneuraminic acid (Neu5Ac), can be further modified by acetylation, methylation, or sulfation, generating structural diversity that cells exploit for signaling. Human milk oligosaccharides—the third most abundant solid component of breast milk—include roughly one sialylated structure for every five of the more than 200 known HMOs. Access to pure molecules is essential for defining their biological functions, yet natural sources provide only minute quantities that are painstaking to isolate individually.

Solution-phase routes to sialylated oligosaccharides—purely chemical, enzymatic, and chemoenzymatic—each carry limitations. Chemical synthesis is flexible but labor-intensive because every step demands purification. Enzymatic and chemoenzymatic routes deliver complete stereoselectivity yet depend on glycosyltransferases whose substrate scope is constrained. The result is that most sialylated HMO research to date has rested on simple, commercially available trisaccharides such as 3′-sialyllactose and 6′-sialyllactose, leaving the more complex branched structures understudied.

Automated Glycan Assembly Shrinks the Timeline

Automated glycan assembly (AGA) offers a different paradigm: Monosaccharide Building Blocks are coupled iteratively on a solid support, so intermediate purifications are bypassed and assembly is driven by machine. Despite these gains, integrating sialylation into AGA has remained stubbornly difficult. The sialic acid anomeric center is tertiary and bears an electron-withdrawing carboxylate at C-1, a combination that favors elimination over substitution; the anomeric effect and the absence of a participating group at C-3 further undermine α-selectivity, so on-resin attempts routinely suffered from low yields and 2,3-elimination side products.

Macrobicyclic Sialic Acids Solve the Selectivity Problem

Prior AGA strategies sidestepped the problem in different ways. One approach embedded sialylation in pre-formed disaccharide building blocks, which worked but required a custom disaccharide for every target because natural sialylated glycans differ in linkage and branching. A second used cyclic oxazolidinone-protected sialic acid donors that improved α-selectivity, yet performed well only on mono- or disaccharide acceptors. The new work instead turns to macrobicyclic sialic acid donors, in which the C-1 carboxylate is tethered to the N-5 carbamate through a finely tuned aliphatic linker. This constraint forces incoming acceptors to attack from the α-face, simultaneously improving α-selectivity and suppressing the anti-Bredt elimination pathway. A Troc-like group built into the tether raises donor reactivity and provides a handle for later functionalization at N-5.

A Portfolio of Nine Sialylated HMOs

Fig. 1 Three strategic routes for assembling sialylated glycans on solid support highlight the macrobicyclic building-block approach used in this work. (Kuo, et al. 2026)

With efficient on-resin conditions in hand, the team explored scope by targeting nine sialylated HMOs built on the lacto-N-tetraose (LNT) and lacto-N-neotetraose (LNnT) cores. These targets were chosen to probe α(2,6)- and α(2,3)-sialylation at terminal and internal positions and to combine fucosylation with sialylation. Oligosaccharide Synthesis on this scale was supported by a set of sixteen monosaccharide building blocks prepared on a multi-gram scale, and the compounds included biologically notable structures such as the influenza decoy LSTc, the necrotizing enterocolitis–preventing DSLNT, and the cancer-associated DSLNF II.

Fig. 2 Automated assembly of nine sialylated human milk oligosaccharides and their crude purification profiles. (Kuo, et al. 2026)

Optimization revealed that the activation solvent matters acutely: a dichloromethane/dioxane mixture for the NIS/TfOH activator gave the highest conversion, and delivering the full ten equivalents of sialyl donor in a single portion outperformed staged additions by maximizing local donor concentration at the resin. Two consecutive coupling cycles of the lead donor furnished a sialylated disaccharide in 67% isolated yield. Even the more hindered α(2,3)-sialylation—historically the tougher linkage—was brought to 35% yield by temporarily protecting both the C-3 and C-4 hydroxyls of the galactose acceptor, which relieves steric crowding at the sialylation site.

Remote Protecting Groups Reshape Reactivity

A striking finding is that protecting groups on one sugar can reach across the molecule to govern sialylation elsewhere. When fucose was installed on the internal glucosamine, the acid-sensitive O-fucosidic bond proved vulnerable: the final α(2,6)-sialylation step, run under NIS/TfOH with an activated sialyl donor, cleaved the linkage and fragmented the product. Switching the fucose building block to one bearing benzoyl esters preserved the bond and delivered the protected hexasaccharide in 22% yield. More broadly, a benzoyl ester at the C-3 position of the internal glucosamine drove complete terminal α(2,6)-sialylation in just two cycles, whereas a benzyl ether needed extra cycles—evidence that an electron-withdrawing group (or fucose itself) tunes the conformation or electron density of the galactose terminus. Notably, fucose accelerated only α(2,6)-sialylation; α(2,3)-installation still required four cycles. Glycan Profiling of the crude mixtures, together with Oligosaccharide Analysis by HPLC, let the authors map these long-range effects rapidly across many sequences.

DSLNF II: The Branched Stress Test

Fig. 3 Competing synthetic routes to the branched heptasaccharide DSLNF II by automated glycan assembly. (Kuo, et al. 2026)

The heptasaccharide DSLNF II—a branched LNT core bearing one α(1,4)-linked fucose and two sialic acids—has long been a formidable target: enzymatic routes failed, and only a lengthy total synthesis had reached it. On the AGA platform, early sialylation and fucosylation blocked the subsequent branch galactose, and late-stage fucosylation unexpectedly cleaved the glucosamine linkage. Success came only from a late-stage bis-sialylation design using a glucosamine building block whose hydroxyls could be unmasked independently, with a galactosyl phosphate installed before fucosylation and di-sialylation at both the internal glucosamine and the terminal galactose. The internal glucosamine emerged as the hardest site to sialylate, and forcing it with extra cycles triggered decomposition—yet the route delivered the target heptasaccharide and demonstrated that AGA can build even structures that resist enzymatic assembly.

A Kinder Deprotection for Acid-Labile Sugars

The protected oligosaccharides still required gentle global deprotection that spares the labile sialyl and fucosyl linkages and the base-sensitive tether. A base-first sequence recovered only 27% of product, with higher-mass byproducts pointing to incomplete deprotection and base-promoted peeling. The improved route instead opens with a reductive step using a Zn–Cu couple in acetic acid, which cleanly cleaves the N-5 Troc-like carbamate and reduces the tethered group to an acetamido moiety, sidestepping the solubility and reduction problems of hydrogenolysis. Acetylation, mild saponification, and a final hydrogenolysis then removed the remaining benzyl and Cbz groups, delivering the pentasaccharide in 64% over four steps; α(2,6)-sialylated products overall gave 56–64% recovery, while α(2,3)-bearing products returned 23–40%. Every glycan carried a 5-aminopentyl spacer at the reducing end, providing a bioconjugation handle for glycan arrays and immunogenicity studies.

Implications for Glycobiology and Glycan Supply

This study resolves a central challenge in automated glycan assembly—reliable on-resin sialylation—and does so with a broadly applicable design rather than case-by-case workarounds. Homogeneous sialylated HMOs can now be assembled rapidly and modularly, enabling structure–activity studies that were previously impractical and supplying the pure standards needed to interpret complex biological glycan mixtures. For groups studying sialic acid biology, host–microbe interactions, or tumor-associated carbohydrate antigens, the availability of defined sialylated structures lowers the barrier to mechanistic work. At CD BioGlyco, Custom Glycosylation Service workflows and related synthesis and analysis offerings help researchers obtain the tailored glycans and the characterization data that turn such foundational chemistry into reproducible experiments. Glycomics and glycobiology as a whole benefit when pure, well-defined sialylated glycans move from scarce curiosities to routine research reagents.

Related Services & Products

• Glycan Release Service
• Glycan Separation and Purification Service
• Glycan Derivation Service
• Glycomic Characterization Service
• Glycomic Quantitation
• N-Glycan Profiling Service
• O-Glycan Profiling Service
• Sialic Acid Analysis Service
• Glycoprotein Enrichment
• Custom Glycoprotein Synthesis

Reference

1. Kuo, Y.T., et al. (2026). Synthesis of sialylated human milk oligosaccharides by automated glycan assembly. Nature Communications. DOI: 10.1038/s41467-026-73028-w.