The combination of chemical synthesis and enzymatic modification, known as chemoenzymatic glycosylation, provides the structural versatility of chemical synthesis and the stereoselectivity of enzymatic synthesis. It allows the site-specific addition of defined unnatural glycans to proteins in aqueous environments and under mild reaction conditions. It also circumvents the protecting group strategy associated with fully chemical synthesis and expands on the substrate scope available from biological hosts to yield defined glycoforms that can be customized for their desired clinical activity, regulatory approval, and commercial manufacturing.
Chemical synthesis of activated sugars followed by enzymatic glycosylation (chemoenzymatic glycosylation) is a convergent approach where unnatural high-energy sugar donors such as oxazolines, fluorides, or azide-tagged sugars are chemically prepared, and then presented to mutant glycosynthases or endoglycosidases capable of transferring an entire oligosaccharide from these donors onto a protein acceptor, in one enzymatic step. Enzymatic reactivity installs the glycosidic linkage with regio- and stereoselectivity native to the enzyme, while the chemical synthesis allows for unnatural monosaccharides, click handles, or photo-cleavable capping groups that are unavailable from native biosynthesis.
Fig. 1 Parts, systems, and applications of synthetic glycobiology.1,5
Chemical-enzymatic chemoenzymatic synthesis combines the strengths of both parent fields while avoiding their weaknesses. Chemical synthesis provides access to any desired structure, but traditionally results in anomeric mixtures that require separation and contain protecting-group chemical waste; it is also typically not performed under aqueous conditions, as strong acid catalysts irreversibly denature proteins. In contrast, enzymatic glycosyltransfer reactions are isoform-specific and retain the protein's native chemical ligation-compatible glycans in aqueous solution. However, they are limited to incorporating only those sugars that cells naturally process. Chemoselective glycosyltransferases solve both of these issues by moving the chemical synthesis step off-protein (avoiding acid denaturation) to create complex, unnatural, or isotopically labeled sugars, which can then be chemoselectively transferred onto proteins as oxazolines by engineered EndoS mutants. Not only does this prevent the need for protecting groups and chromatographic separation of anomeric mixtures, but donors can now be engineered to require inexpensive chemical activation and be recyclable or used in catalytic quantities, significantly reducing material costs. This approach also works with proteins that are sensitive to acid or have structures that are shielded from enzyme access. Since glycan assembly occurs off-protein, Lewis-acid activation is not needed, and the enzymatic transfer is performed at neutral pH at room temperature, preventing denaturation of disulfide bonds. Lastly, in regulated industries, workup and analysis for the enzymatic deglycosylation portion of the synthesis is considered routine for biologics production, while synthesis of the chemical donor would be considered chemical synthesis of an upstream raw material, rather than forcing organizations to push an entire CMC effort through one group.
Table 1 Strategic benefits of the chemoenzymatic synergy
| Limitation vector | Pure chemical route | Pure enzymatic route | Chemoenzymatic solution | Value inflection |
| Anomeric control | Statistical mixture | Hard-wired selectivity | Synthetic block + enzymatic lock | Single-isomer guarantee |
| Donor scope | Unlimited, costly | Natural only | Designer donor, catalytic transfer | Functional tags enabled |
| Protein stability | Acidic/corrosive conditions | Aqueous compatibility | Off-protein synthesis, mild coupling | Fold preservation |
| Scale-up burden | Chromatographic armies | Enzyme production cost | Convergent assembly, fewer steps | Timeline compression |
| Regulatory optics | ICH M7 scrutiny | Bioprocess familiarity | Split raw material/bioprocess | Audit modularity |
Central themes in chemoenzymatic glycosylation are founded on principles of building block assembly. This employs a modular synthetic approach whereby chemically constructed donors with orthogonal functionalities are enzymatically coupled to acceptor molecules (aglycones) with defined stereochemistry. Such a strategy has allowed glycoscience to disconnect its synthetic ambitions from biological limitations without sacrificing reactions performed in water under mild conditions.
To make sure each chemical synthesis of glycan building blocks contains all of the features described above, we have developed a general approach. First, all hydroxyl groups except the anomeric center are strategically protected to prevent side-chain involvement during regioselective activation (protecting-group choreography). The anomeric center is then installed as a high-energy leaving group (oxazoline, fluoride, or trichloroacetimidate) in the presence of reagents orthogonal to acid-labile protecting groups that may remain on the scaffold. At this point non-natural modifications can be introduced; azido/alkynyl groups for click chemistry can be added to the C-6 position through nucleophilic displacement of a primary triflate and photo-cleavable groups such as o-nitrobenzyl ethers can be added to any hydroxyl groups whose modification we want to trigger later. Large oligosaccharide scaffolds can be assembled first as large protected blocks and convergently joined in a late stage coupling to minimize synthetic tree size. Enzymatic-release protecting groups are selected that are compatible with glycosyltransferases (acetates, levulinoyl esters or silyl ethers) but which can be cleaved by orthogonal conditions that do not affect the oxazoline donor (hydrazinolysis for levulinates or fluoride for silyls). At this point isotopic labels can be added such as deuterium from deuterated acetyl anhydride or 13C-labels from 13C-labeled cyanide during cyanohydrin formation. This can serve as an internal standard for mechanistic enzymology experiments. Finally, donor stability can be improved by converting unstable phosphates to phosphoramidates or adding trifluoroacetyl groups that slow-down hydrolysis during storage.
The assembled glycan stub is further extended enzymatically, starting with acceptor priming, the addition of a single GlcNAc or glucose onto the protein using bacterial orthologues of N-glycosyltransferases or engineered endoglycosidases. This stub then acts as a platform for further modification using glycosynthase mutants (catalytically dead mutants with the active-site carboxylate replaced with alanine) of bacterial oligosaccharyltransferases, which accept synthetic oxazoline donors and perform the reverse reaction to add on pre-assembled oligosaccharide blocks onto the reducing-end GlcNAc without trimming the glycan. Extension continues by incubation with the appropriate mix of transferases: e.g. β1,4-galactosyltransferase to add LacNAc repeats followed by α2,3- or α2,6-sialyltransferase to add sialic acid residues and optionally α1,3-fucosyltransferase to generate sialyl-Lewis (X) structures. Intermediates can be analyzed by electrospray ionisation mass spectrometry to ensure complete conversion before moving on to the next enzyme. One-pot multi-enzyme (OPME) systems can shorten the enzymatic assembly lines by co-delivering multiple transferases and regenerating nucleotide-sugars on-line using pyrophosphorylases and kinases; such reactions have been scaled up to produce glycans at gram-scale without intermediate purification. Transferase substrate promiscuity can also be used to enzymatically extend otherwise sterically hindered core glycans; bacterial β1,4-galactosyltransferases accept a wider range of substrates than mammalian enzymes enabling the addition of LacNAc repeats onto core M2 O-mannose glycans. The assembled glycoform is then released from the enzymatic module and subjected to glycopeptide mapping and arrays of linkage-specific exoglycosidases to confirm assembly. Any glycoform that fails to match the expected glucose-unit map at this stage can be returned for polishing.
Chemical glycosylation and enzymatic glycosylation individually have their advantages and disadvantages. Glycosylation chemoselectivity can be traded off against reagent or reaction harshness in chemical methods, while enzymatic methods are limited by strict substrate specificity. Chemoenzymatic glycosylation aims to provide the best of both methods. Coupling chemical carbohydrate synthesis with enzymatic processing allows for access to glycan structures that may not be possible from one method alone. Chemoselectivity and regioselectivity provided by enzymatic processing can allow for the placement of unnatural modifications, click chemistry handles, and isotopic labels on sugars without the need for extensive protecting group strategies or sacrificing a mixture of anomers that can increase the difficulty of purification.
Orthogonality between glycan logic and catalysis also allows control over structural diversity and rigidity. During chemical preparation, protecting-group strategy can be used to block all sugars' hydroxyl groups except for the anomeric carbon. This allows one to introduce unnatural functional groups at other positions for later orthogonal modification (such as azides at C-6 for subsequent click chemistry, photo-cleavable ethers for photoswitchable proteins, or fluorinated building blocks for NMR spectroscopy), which would not be possible with enzymatic manipulation. Furthermore, anomeric activation with a high-energy leaving group such as an oxazoline or fluoride allows for coupling with an engineered glycosynthase that does not require nucleophilic side reactions, offering better stereocontrol over the glycosidic linkage than chemically glycosylated proteins. Having decoupled synthetic logic in the glycan building block from that in the protein acceptor, one can use any method to prepare the former (such as purification techniques used in organic synthesis), and does not need to denature the protein with acidic conditions required by Lewis-acid catalysis. Furthermore, since individual sugar building blocks can be pre-assembled into large oligosaccharide clusters that are protected as one entity, conjugation becomes a one-step, convergent process, minimizing the number of linear steps needed. Use of bacterial transferases also allows unnatural branching (due to some bacteria having greater tolerance for bulky groups at C-6). Finally, given the wide variety of glycan donors that can be screened with each enzyme–acceptor pair, glycoengineering allows much quicker production of variant libraries than metabolic engineering of cell lines (often days versus months).
One advantage of the chemoenzymatic approach is glycan homogeneity. The cell's heterogeneity is "wiped clean" by enzymes and replaced with one defined glycoform which can be measured. Typical IgG Fc glycans made by standard cell culture processes contain mixtures of galactosylation, sialylation and bisection degrees of freedom. Chemoenzymatic remodeling strategies first reduce this diversity to a homogeneous GlcNAc stub using glycosidases and then attach synthetic oligosaccharide building blocks all at once, achieving homogeneity often greater than 95%. This allows precise control of drug-to-antibody ratios for conjugates where the engineered glycan acts as the point of attachment. An azido-galactose handle can be installed site-specifically and then clicked to make ADCs where each antibody contains two drug molecules with minimal low or high drug species. In addition, because remodeling is done in aqueous enzymatic conditions, the disulfides remain intact so the antibody retains its antigen-binding function while the effector functions are reprofiled. Regulatory agencies welcome glycan homogeneity because fewer impurities need to be characterized. Instead of monitoring several glycoforms and potential loss of handle variants during the conjugation process, the chemistry, manufacturing and controls (CMC) package will primarily consist of the desired product and a small number of well-defined process-related impurities. The process is easily standardized between lots since glyco-replacement enzymes (the catalysts) and donor cells (the reagents) are well-characterized. Once an enzyme and donor lot are qualified, follow-on production campaigns will have chromatograms with a cosine correlation of >0.95, narrowing batch variation down to analytical variance.
Chemoenzymatic glycosylation consists of two discrete modular steps: In one module, the glycan donor is chemically synthesized and activated, complete with orthogonal handles. In a second module, purified glycosynthases or endoglycosidases retransfer these preassembled scaffolds onto protein acceptors in aqueous solution under mild conditions. This approach separates glycan complexity from enzyme reactivity, enabling individual optimization, orthogonal analysis, and scaling of each independent step unlike either fully chemical or fully enzymatic methodologies.
The preparation of glycans for glycosynthase reactions takes place before enzyme incubation and involves some strategic considerations. First, protecting-group strategies allow all functional groups but the anomeric centre to be chemically modified, preventing side reactions and allowing regioselective activation of donor substrate (usually an anomeric fluoride or oxazoline activated leaving group). All other hydroxyl groups can be protected in such a way that they can later be selectively removed in the presence of the activated donor. At this stage, chemoselective modifications can be introduced: azides for click chemistry at position C-6 (generated through nucleophilic substitution with the corresponding primary triflate), photocleavable groups such as o-nitrobenzyl ethers tethered to reactive hydroxyl groups, fluorinated monosaccharides for detection by 19F NMR, and any other non-natural modification not found in biosynthetic metabolites. Glycan synthesis is made scalable by judicious choice of protecting groups that allow convergent synthesis. Large oligosaccharide portions are synthesized separately and joined through one glycosylation step at the end of the synthesis, rather than synthesising the glycan linearly, which would require many glycosylation steps for a decasaccharide. Choice of protecting groups can also be used to ensure synthetic glycans are tolerated by glycosyltransferases. Leaving groups such as acetates, levulinoyl esters, or silyl ethers can be tolerated by glycosyltransferases and subsequently removed by selective hydrazinolysis (for levulinates) or fluoride treatment (for silyl ethers) to regenerate the enzyme labile leaving group (usually benzyl). Glycans can be isotopically labelled with deuterated acetyl anhydride or 13C cyanide during cyanohydrin formation for use as internal standards in mechanistic enzymology. Synthetic stability of glycan donors can be improved by converting the naturally occurring unstable phosphate leaving group to a more stable phosphoramidate or by converting it to a trifluoroacetyl group, both of which can later be converted back to phosphate by the enzyme.
Transferase enzymes are typically used to elongate pre-established glycan stubs in sequential fashion. Donor activated sugars are added onto proteins one at a time by glycosyltransferases. A GlcNAc or glucose stub is first placed on the protein via bacterial N-glycosyltransferases or engineered endoglycosidases (also known as "acceptor priming"). Glycosynthases, mutants of glycosidases in which the active site carboxylate is mutated to alanine, then take over using synthetic oxazoline donors. Glycosynthases reverse the hydrolysis reaction by transferring pre-formed oligosaccharides onto the reducing-end GlcNAc of the primer without destroying the product via hydrolysis. Enzymes can then be added sequentially to elongate the oligosaccharide (e.g. β1,4-galactosyltransferase then α2,6-sialyltransferase) (each reaction verified by positive ion mode electrospray ionisation mass spectrometry prior to next enzyme addition). One-pot, multi-enzyme approaches shorten the enzyme chain by including multiple glycosyltransferases in the same vessel along with enzymes to regenerate nucleotide-sugars (pyrophosphorylases and kinases). These pathways can produce glycoprotein products on the gram scale without purification of intermediates. Addition of sugars onto sterically-confined cores can also be achieved by leveraging the acceptor substrate promiscuity of bacterial glycosyltransferases. For example, bacterial β1,4-galactosyltransferases have greater acceptor flexibility than mammalian enzymes and can be used to add LacNAc onto core M2 O-mannose glycans. Completed glycoforms can be analyzed by glycopeptide mapping and linkage specific exoglycosidase arrays to confirm successful glycan remodeling. If the glucose-unit composition is off-by-one or more than expected, the glycoprotein will go back into the polishing module for another cycle.
Chemoenzymatic glycosylation has emerged as a powerful enabling platform with applications ranging from basic glycobiology to translational medicine. Linking synthetic carbohydrate building blocks to enzymatic transfer, homogeneous unnatural glycoforms can be bestowed upon protein targets to serve as molecular probes, reference materials, or next-generation therapeutics. It shines brightest where heterogeneity introduced by cellular methods obfuscates structure–function analysis or where regulatory bodies require single-species active pharmaceutical ingredients (APIs).
Fig. 2 A schematic of N-glycosylation patterns found in (A) humans, (B) established biofactories including yeasts, plants, CHO cells, and (C) specific N-glycosylation patterns in microalgae.2,5
Synthesis of glycoproteins of research grade, homogenous to a single glycoform for which detailed characterization has been documented, represents the first field where chemoenzymatic remodeling becomes essential. Endogenous glycoproteins purified from serum or cell culture media are often plagued by macro- and micro-heterogeneity that complicates binding assays, crystallization efforts, lectin mapping etc.; reducing these glycans down to a homogeneous GlcNAc stub with ENGases and then appending a chemically synthesized oligosaccharide in trans glycosylates the protein with a carbohydrate fingerprint that can be reduced to a single peak in LC–MS. Such homogeneous glycoproteins can be used as standards to calibrate glyco-analytics, validate intact-mass workflows and produce isotopologues for use as internal standards in quantitative glycoproteomics. High-avidity probes for lectin screening can be generated in the same manner from multivalent glyco-polymers and glyco-cyclodextrins. Incorporation of photo-cleavable groups or click chemistry handles in a site-selective manner also permits temporal control of glycan presentation during live-cell imaging experiments. Traceability of each batch to a defined synthetic donor lot and enzyme 'passport' dramatically improves inter-laboratory consistency and prevents historical drift, which is a problem that cannot be overcome with naturally derived glycans.
The most impactful translational applications are therapeutic in nature. Monoclonal antibodies with engineered Fc glycans through Endo-S mutants are now available as uniform non-fucosylated or hyper-sialylated isoforms with 'locked' ADCC or anti-inflammatory activities, respectively; these glycoproteins circumvent decades-long cell-line knockout efforts necessary to diminish fucosylation. Applications outside cancer therapy have been pursued using this same remodeling strategy. Erythropoietin can be modified to carry extra sialylated antennae that increase in vivo half-life without escalating protein dosage, and lysosomal enzymes can have mannose-6-phosphate added at defined positions to direct cellular uptake. Chemically synthesized glycans are foundational to an emerging class of vaccines as well. Pathogen O-antigens are chemically synthesized in lipid-linked form and conjugated en masse to carrier proteins using relaxed-specificity oligosaccharyltransferases; the resulting monospecies glycoconjugates reliably elicit protective immune responses and are now entering clinical trials. With the glycoform encoded by a synthetic donor substrate instead of the host cell, modifications can be made quickly by exchanging sialylation patterns or incorporating novel sugars without altering the fermentation workflow.
Table 2 Therapeutic modalities advanced by chemoenzymatic glycoengineering
| Therapeutic format | Glycan function | Benefit |
| Afucosylated mAb | Enhanced ADCC | Tumor cell lysis |
| Hyper-sialylated enzyme | Extended half-life | Reduced dosing frequency |
| Man-6-P enzyme | Lysosomal targeting | Improved efficacy in LSD |
| Glycoconjugate vaccine | Defined epitope density | Consistent immunogenicity |
Chemoenzymatic glycosylation is not without its own technical challenges despite offering single isomer selectivity and structural flexibility. Successfully coupling an enzyme with its rather limited substrate scope to a chemically altered donor molecule followed by optimization of that well-balanced relationship at a GMP-compliant scale can contain hurdles including catalyst availability, reaction optimization, and extensive analytical characterization to ensure each glycan is placed correctly and exclusively.
Engineering your enzyme of choice starts with an obvious contradiction: highly selective catalysts allow only a small chemical space. Glycosynthase mutants engineered to accept synthetic oxazolines may reject donors with large C6 tags. These deletions often require cycles of active-site remodeling spanning months of protein engineering effort. When an enzyme candidate is finally obtained, its host of expression must be considered; truncated mammalian transferases often mis-fold when overproduced in bacteria while many full-length, membrane bound forms require detergent micelles, which can denature activity or cause micelle-to-micelle variation. Optimization of the reaction buffer must consider pH, ionic strength, co-solvent effects on enzyme stability, donor solubility, protein acceptor stability—a balancing act where a change of 0.2 pH units could flip regioselectivity. Multi-enzyme sequential cascades present another wrinkle; intermediate glycoforms must stay soluble long enough to pass-the-torch without being subject to product inhibition of upstream enzymes. Inherent promiscuity of enzymes must also be considered; for example, a β1,4-galactosyltransferase will begin to sialylate unless CMP-sialic acid is aggressively depleted causing minute micro-heterogeneity that will be mistaken for "enzyme lot drift". Immobilzation allows reusability but can obscure the active site with spacers that decrease kcat and change previously optimized kinetic parameters determined in solution.
Validation needs to prove that the purified glycoprotein is not a distribution but a single molecule species. Initial confirmation by intact-mass does not assure this as the resolution limits of typical instruments will often allow isobaric glycoforms (species that differ by a single glycosyl linkage position) to co-elute. Glycopeptide mapping (MS/MS after enzymatic proteolysis) is therefore essential to validate occupancy at the correct site and linkage. Sequential digestion with exoglycosidases allows linkage determination, but between lot variability of exoglycosidases can alter integer values of glucose units. For best practice a full suite of standards should be run alongside an experimental digested glycoform to account for this variability. Assays for chemical handles such as azides or photocleavable groups cannot be determined by lectin blot technologies and therefore assay developers must internally validate custom MS/MS ion transitions to quantify the attached chemical handle. At pg/mL levels any residual protecting-group-related impurities or metals displaced from the immobilized metal during loading must also be validated to ppm levels per ICH Q3D requirements. Elemental analysis via ion-coupled plasma MS is outside the scope of most biologics facilities. Degradation studies need to demonstrate clearance of the enzyme catalyst. In most cases a suitable ELISA will not exist so a surrogate peptide from the catalyst is spiked into the protein and measured by LC–MS.
Chemoenzymatic glycosylation becomes favored when a project encounters the "precision frontier" where cell-based systems produce mixtures, chemical synthesis threatens protein integrity, and regulatory bodies demand one defined glycoform. Decision points usually occur after repeated failure to reach sub-percent glycoform overlap, the need for novel sugars no host can produce, or timelines that cannot accommodate iterative cell-line engineering.
Applications requiring heterogeneous glyco-patterns (bispecifics with afucosylated Fc for improved ADCC and hyper-sialylated Fab to increase serum half-life, for example) are impossible to fulfill with homogeneous Golgi-dependent processes. Chemoenzymatic capabilities enable site-specific conjugation: Fc trimming to GlcNAc remnant followed by GlcNAc-based build of defined biantennary complex type; Fab extension with α2,6-sialic acid derived from an orthogonal transferase, entirely in one reaction vessel. ADCs needing click-able glycans for site-specific drug attachment also leverage mutant transferases capable of utilizing azido-GalNAc donors; providing DARs determined by glycan numbers instead of random lysine.attachments. Maternal vaccines engineered for α2,6-exclusive sialylation to eliminate pro-inflammatory signals, or biosimilars that need to match an originator glycoform profile within ±2 % relative abundances are further examples where this platform becomes necessary.
Table 3 Comparison of precision requirement
| Precision Requirement | Cell-Based Limitation | Hybrid Solution |
| Asymmetric Fc/Fab | Single Golgi pool | Site-selective rebuild |
| Click-handle ADC | No azido-metabolism | Mutant GalT accepts azido |
| Maternal sialylation | Mixed α2,3/α2,6 | ST6Gal1-only extension |
| Biosimilar overlay | Ensemble drift | Locked single peak |
Virtual companies combine capabilities for synthesis (organic and enzymatic), biology, characterization, and regulatory document preparation. The combined efforts prevent disconnects seen when outsourcing different steps to different providers. Chemists supply azido donor sugar-building blocks, photocleavable fucose donors, or alkyne-pkg GalNAc donors optimized for GMP manufacture. Synthetic chemistry ensures orthogonal protecting-group strategies to facilitate late-stage incorporation of these unnatural sugar donors. Meanwhile, enzymologists branch-evolve glycosynthases to utilize unnatural donors, with the goal of both teams converging on the desired glycoform. Intact-glycoprotein LC-MS, glycopeptide mapping LC-MS, and glycoprotein ion-mobility MS are developed downstream, with structural interpretation supporting development of bioinformatics tools that construct on-resin glucose-unit dictionaries and cosine similarity passports useful for IND inclusion. Regulatory personnel will have elements of the CMC ready for review (addressing impurity disposition, etc., as well as donor specific details) that are compliant with ICH Q11 model requirements. The expectation is that when the sponsor receives their vial of material, they also have a template batch record and risk assessment prepared for their review. Additional knowledge from optimizing formulations to process failures such as crystallization or lyophilization failures are looped back so that the next company doesn't need to relearn the lessons from these experiences.
Chemoenzymatic glycosylation combines the flexibility of chemical synthesis with the specificity of enzymatic modification, making it particularly powerful for complex or highly defined glycan designs. When projects require both structural precision and synthetic versatility, specialized glycosylation services enable the effective integration of chemical and enzymatic workflows.
Chemoenzymatic and custom glycosylation services are designed for projects that demand precise control over glycan structure beyond the capabilities of purely enzymatic or purely chemical methods. By leveraging chemically synthesized glycan building blocks together with enzyme-mediated transfer, extension, or remodeling, these services enable the generation of glycoproteins and glycoconjugates with well-defined and reproducible glycan architectures. Such approaches are especially valuable for complex branched glycans, non-native structures, and applications where subtle glycan differences have significant functional consequences.
Successful chemoenzymatic glycosylation relies on access to high-quality, structurally defined glycan substrates. Glycan synthesis and modification services provide custom-designed glycans and tailored functionalization options that serve as reliable inputs for downstream chemoenzymatic workflows. By supplying well-characterized glycan building blocks and post-synthetic modifications, these services support consistent enzyme performance and enhance overall control in hybrid glycosylation strategies.
References
