Glycosylation is a key protein post-translational modification that affects protein folding, stability, activity and transport, and is an important quality attribute of protein drugs.
A common glycoengineering method is to genetically modify the host cell's native glycosylation pathways, but this method is time-consuming and may interfere with normal cell growth and lead to a heterogeneous glycoprofile. Chemoenzymatic in vitro Glycoengineering can generate site-specific and homogeneous sugar structures, but the steps are cumbersome and accompanied by various by-products. Although the in vitro enzymatic reaction is simple, the competitive reactions of multiple enzymes lead to insufficient homogeneity.
On February 6, 2024, the team of Cleo Kontoravdi and Karen M. Polizzi from Imperial College London published an article titled "Immobilized enzyme cascade for targeted glycosylation" in Nature Chemical Biology. The authors developed an artificial Golgi reaction platform (sequential glycosylation reactions for tailored sugar structures (SUGAR-TARGET)) composed of multiple immobilized enzymes. The spatiotemporal separation of immobilized enzymes solves the existing problem of chaotic multi-enzyme catalysis and achieves customizable and controllable homogeneous N-linked glycosylation. The authors selected a human-like glycosylation pathway encoded by four enzymes (GnTI, ManII, GalT, SiaT) to demonstrate the potential of this platform to precisely control the sequential glycosylation of proteins of microbial and mammalian origin.
The authors expressed three Glycosyltransferases (GnTI, GalT and SiaT) in the pathway in E. coli. By fusing maltose-binding protein (MBP) at the N-terminus to improve the soluble expression of the protein in E. coli, and fusing AviTag with a two-residue glycine-serine linker at the C-terminus, the co-expressed biotin ligase can specifically bind free biotin to AviTag. Biotin forms a complex with streptavidin, allowing for a one-step immobilization/purification process. But attempts to express and purify ManII in E. coli were unsuccessful. Therefore, the authors produced and purified ManII from insect cells and chemically biotinylated it in vitro.
The single-step conversion rates of GnTI, ManII, GalT and SiaT to free glycans are all >95%, but GalT also has high activity towards GM5 and GM4, so the one-pot method will have substrate competition and cross-reactions, resulting in uneven glycoforms in the end. By immobilizing GnTI, ManII, GalT and SiaT on silica beads using the above method, the authors successfully constructed an immobilized enzyme cascade system (SUGAR-TARGET). The spatiotemporal separation of several enzymes avoids cross-reaction, and the conversion rate at each step is above 95%. This immobilized cascade platform can efficiently generate homogeneous glycoforms, and the target glycoforms can be customized by adjusting the catalytic sequence.
Fig.1 Strategy and design of artificial Golgi reactions (SUGAR-TARGET) for targeted glycosylation. (Makrydaki, et al., 2024)
In the one-step immobilization/purification process, cultured E. coli cells are collected, lysed by sonication, and then centrifuged to separate soluble enzymes and desalted. The desalted solution is mixed with streptavidin particles to specifically bind biotinylating enzyme. Finally, the immobilized enzyme is collected by centrifugation. When BirA is co-expressed and free streptavidin (Stv) is added, the protein electrophoresis of biotinylated GnTI and GalT will shift upward, indicating that biotinylated GnTI and GalT successfully bind to the streptavidin. The only impurity observed in the results of the one-step immobilization/purification was BirA that bound non-specifically to the immobilization material, possibly forming a complex with AviTag.
Fig.2 Experimental process for in vivo biotinylation and one-step immobilization/purification of GnTI and GalT. (Makrydaki, et al., 2024)
After success with free glycans, the authors applied this cascade to the mFc crystallizable domain of an antibody produced by Pichia pastoris and the non-enzymatic protein saposin B produced by HEK293 cells lacking GnTI activity. Both cells primarily produce M5 structures, which are the preferred substrates of GnTI. The mFc was first purified through a nickel column before being applied to the immobilization cascade. During this process, the author replaced the silica beads used for immobilization with magnetic beads to facilitate handling. The conversion rate of each step of mFc's glycosylation cascade reaches more than 95%, confirming that the spatiotemporal separation strategy solves the substrate competition of the enzyme and efficiently generates a homogeneous glycoform. Similarly, this cascade has a single-step conversion rate of >95% for saposin B and generates a homogeneous glycoform. These findings confirm that this platform can be used to reconstruct the N-linked glycosylation pathway of mammalian protein drugs produced by microbial cells.
The authors further applied this platform to increase the degree of galactosylation of full-length antibodies to improve antibody functionality. Galactose has anti-inflammatory properties and increases affinity to receptors. The authors tested the extent to which immobilized GalT increases galactosylation of three igGs from different sources. After treatment with immobilized GalT, the galactosylation degree of igG from three different sources was improved to varying degrees.
In order to further study the specificity and rate-limiting step of the enzyme, the authors treated G0/G0F and different intermediates with immobilized GalT. After reacting with G0/G0F for 15 minutes, only G1’/G1’F and the final product G2/G2F were detected, and the conversion rate of G1F by immobilized GalT was much higher than that of G1’F. Therefore, the author believes that it is difficult for GalT to catalyze G1’\G1’F, and the first generation of G1’\G1’F will limit subsequent reactions.
Next, the authors evaluated the reusability of immobilized GalT using CHO h-IgG as a substrate. After four rounds of reactions, with a total reaction time of more than 80 h, the immobilized GalT still retained more than 70% of its activity.
The authors constructed a targeted glycosylation platform containing four immobilized enzymes for targeted and sequential glycosylation reactions in an in vitro environment. The spatiotemporal separation of immobilized enzymes can solve the homogeneous glycoform caused by the enzyme's substrate competition and strictly control the glycoform. The in vivo biotinylation and site-specific immobilization of glycosyltransferase methods applied on this platform successfully achieved one-step immobilization/purification of GnTI, GalT, and SiaT.
The authors successfully demonstrated the development potential of the targeted glycosylation platform in the field of in vitro glycosylation. The targeted glycosylation platform can be used as a complement to natural protein post-translational modifications to achieve customization of N-linked glycosylation pathways, generate expected homogeneous glycoform, and be used in the development and research of protein drugs such as Vaccines.
Reference
