It has been a whirlwind week for the GlycoUniverse team. Fresh off a captivating (and summa cum laude) thesis defense in Berlin, our newly minted PhD Yan-Ting Kuo is celebrating yet another milestone: his work on the automated synthesis of sialylated human milk oligosaccharides (HMOs) is now live on Nature Communications! It is a proud moment for everyone involved, but the Glyconeer® machine didn't start humming overnight. To understand the breakthrough, we look back at the challenges that paved the way.
The Biological "Outer Shell"
Sialic acids sit at the outermost ends of many glycans, where they mediate critical recognition events involving cells, pathogens, immune receptors, and proteins. Their biological importance is especially visible in HMOs, which contribute to infant nutrition, immune development, and host-microbiome interactions. Yet sialic acids remain among the most difficult motifs to install chemically, particularly in an automated solid-phase setting. Our recently published work, “Synthesis of Sialylated Human Milk Oligosaccharides by Automated Glycan Assembly,” addresses this challenge. While natural isolation provides only tiny quantities and enzymatic methods can be restricted by scope, Automated Glycan Assembly (AGA) offers a faster, modular route to structurally defined, homogeneous glycans.
The Challenge: When Solution-Phase Logic Fails
The working principle of AGA is simple: glycans are built stepwise on a solid support where monosaccharide building blocks are added through repeated cycles of deprotection, glycosylation, washing, and capping. However, sialic acid is uniquely uncooperative. Its tertiary anomeric center, electron-withdrawing carboxylate, lack of neighboring-group participation, and tendency toward elimination make both reactivity and stereocontrol challenging. Previous AGA approaches often avoided direct sialylation by using preformed sialylated disaccharide building blocks. This works, but each linkage and branching pattern requires a customized donor. We wanted to go further: Could sialic acid be installed directly during AGA using a monosaccharide building block? The use of macrobicyclic sialic acid donors, developed by Ando et. al (Science, 2019) was evaluated. These donors contain a tether between the C-1 carboxylate and N-5 position, enforcing a conformation that favors α-selective attack while suppressing elimination. However, we soon learned that "what works in solution stays in solution." On-resin, these donors behaved quite differently, forcing us back to the drawing board for optimization. Solid-phase chemistry introduces a unique, and often fickle, set of variables: resin swelling, diffusion rates, local concentrations, and microenvironment effects. These factors can paradoxically alter reaction outcomes, turning a successful solution-phase method into a complex solid-phase puzzle.
Figure 1. AGA on the Glyconeer® synthesizer make production of complex human milk oligosaccharides quick and simple.
The Phd Marathon: 3 Years of Grinding, 6 Months of Sprinting
If you look at the final paper, you see a library of nine complex, highly pure sialylated HMOs. What the Supporting Information doesn't explicitly show is the sheer tenacity required to get there. For three years, Yan-Ting worked as a "team of one," meticulously troubleshooting every variable of the optimization phase. He navigated the frustrations of activator stability and donor delivery modes that often seemed to defy logic. However, once the "code" for automated sialylation was finally cracked, the results were impressive and satisfying: The entire library of nine complex HMOs was completed in 6 months. In traditional carbohydrate chemistry, producing this many defined, sialylated structures would typically require a dedicated team of expert chemists and many years of arduous manual synthesis.
AGA as a Discovery Platform
Much of the project therefore involved systematic adaptation. We discovered that simple model systems did not always predict the behavior of larger glycans. For example, a galactose acceptor that worked in disaccharide model became significantly less reactive when embedded in an LNnT-type tetrasaccharide. A lesson is learnt: “remote” structural features are rarely truly remote. Protecting groups, neighboring branches, and even a fucose residue several bonds away could influence sialylation efficiency. This is where AGA became more than a production tool; it became a discovery platform. Because the workflow is modular, we could rapidly test building blocks, protecting-group patterns, and coupling sequences. These findings provide practical design rules for future synthesis.
DSLNF II Synthesis and a Clean Purification Method
Using this optimized approach, Yan-Ting synthesized nine sialylated HMOs, including α(2,3)- and α(2,6)-sialylated structures. The “final boss” was DSLNF II, a formidable branched heptasaccharide. Earlier synthetic routes failed because first installation of fucose and sialic acid blocked later galactosylation, or because alternative sequences caused fragmentation. Success required a late-stage bis-sialylation step, and a carefully designed orthogonal protecting-group logic.
Even after assembly, the final deprotection was a hurdle. These glycans are delicate; both sialic acid and fucose linkages are labile, and the macrobicyclic donor introduces protecting groups that must be removed cleanly. We developed a specialized global deprotection sequence (Zn/Cu-mediated reduction, followed by acetylation, mild saponification, and hydrogenolysis) to deliver analytically pure glycans. The final compounds also carry an aminopentyl spacer, making them suitable for glycan array printing or bioconjugation to protein carriers.
Figure 2. Comparison of common deprotection sequence (Condition A) and our improved method (Condition B).
There are still limitations. The macrobicyclic donor is used in excess, and some α(2,3)-sialylated targets remain lower-yielding than α(2,6)-linked analogues. However, excess donor can be recovered and recycled automatically, and the insights from this study provide a foundation for further optimization.
Today Access Shapes Future Discovery
Sialylated glycans are important class of molecules beyond HMOs and infant nutrition: They are central to immunology, infection biology, neurobiology, and cancer research. Yet many biological questions require defined panels of linkage isomers, branched structures, analogues, and immobilization-ready compounds. Automated synthesis is especially powerful when the aim is not just one molecule, but a family of related structures.
Behind this paper is a simple conviction: Access shapes discovery. When complex glycans are difficult to obtain, biological hypotheses remain untested. By making sialylated HMOs more accessible through automation, we hope to move glycoscience away from isolated heroic syntheses and toward systematic exploration of biologically relevant glycan libraries.
Funding & Acknowledgements
This work was made possible through close collaboration with Prof. Peter H. Seeberger’s group at the Max Planck Institute of Colloids and Interfaces in Potsdam and was supported by the GlyTunes consortium (EU’s Horizon MSCA grant agreement No. 956758).
About GlycoUniverse
GlycoUniverse is an automation and biotechnology company based in Potsdam, Germany. Founded in 2013 as a spin-off from the Max Planck Institute, GlycoUniverse translates innovations in automated carbohydrate chemistry into practical tools for glycan research. Its core platform, the Glyconeer®, is the world’s first commercial automated oligosaccharide synthesizer. Beyond the instrument, GlycoUniverse provides custom synthesis of defined carbohydrates and glycoconjugates, glycan building blocks, glycan standards, glycan array printing, and glycan analysis. GlycoUniverse is also active as an industrial partner in several MSCA Horizon consortia, including ImmunoShape, GlycoVax, GlyTunes, AciNetwork, and the upcoming GlycoAxis consortium.