First Step Towards the Humanisation of Protein <i>N</i> ‐Glycosylation in <i>Chlamydomonas reinhardtii</i> Through Heterologous Expression of α(1,2)‐Mannosidases

To meet the growing demand for recombinant therapeutic proteins, the development of new expression systems able to compete with the current mammalian cell lines such as CHO cells represents a challenge for the pharmaceutical industry. In this context, microalgae are attractive alternatives as they are photosynthetic unicellular organisms that do not require costly media for growing and are able to perform protein glycosylation (Mathieu-Rivet et al. 2014). Moreover, recent improvements allow reaching almost 50 mg/L of secreted recombinant proteins in the culture medium (Torres-Tiji et al. 2022). N-Glycosylation of recombinant therapeutic proteins is a key issue because N-glycans are required for the bioactivity of biologics. However, non-human glyco-epitopes introduced by the host expression system may induce adverse immune responses in humans (van Beers and Bardor 2012). Consequently, glycoengineering of the N-glycosylation pathway of the cell used as a bio-factory is a prerequisite for the production of biologics compatible with usage in humans. The green chlorophyte Chlamydomonas reinhardtii is one of the most studied microalgae, especially for biotechnology applications like the production of biologics (Leprovost et al. 2024). In contrast to most eukaryotes, C. reinhardtii synthesises in the endoplasmic reticulum (ER) a truncated glycan precursor Glc3Man5GlcNAc2 that lacks mannose (Man) extensions on α(1,6)-Man (Figure 1A) (Vanier et al. 2017). The resulting non-canonical Man5GlcNAc2 structure, annotated as M-M2-2 using the proglycan nomenclature (Altmann et al. 2024), is then further processed in the Golgi apparatus through xylosylation and fucosylation, as well as by O-methylation of Man and fucose residues (Mathieu-Rivet et al. 2013; Leprovost et al. 2024). Thus, mature N-glycans in C. reinhardtii largely differ from human ones that are composed of a core GlcNAc2Man3GlcNAc2 (Figure 1B) substituted by various glycan decorations (galactose linked in β(1,4) and eventually sialic acids in terminal position). Consequently, the humanisation of C. reinhardtii N-glycans would require the implementation of both knockout approaches to remove non-human epitopes and knock-in strategies to in vivo remodel the human N-glycan chassis. Recently, we demonstrated that the human erythropoietin (hEpo) expressed in C. reinhardtii is N-glycosylated with non-canonical Man5GlcNAc2 substituted by fucose and xylose epitopes (Leprovost et al. 2024). In this study, we also demonstrated that the removal of immunogenic xylose and fucose epitopes on biologics can be achieved through the expression of hEpo in mutants impaired for xylosyltransferase and fucosyltransferase activities. This knock-out strategy enabled the production of C. reinhardtii of a glycoengineered hEpo mainly carrying non-canonical Man5GlcNAc2 (Leprovost et al. 2024). As mentioned, the humanisation of C. reinhardtii N-glycans requires the synthesis of a GlcNAc2Man3GlcNAc2 chassis (Figure 1B). In plants and mammals, this N-glycan is synthesised through the action of N-acetylglucosaminyltransferase I (GnT I) on the terminal α(1,3)-Man of a canonical Man5GlcNAc2 ((M6M3)M (Altmann et al. 2024)) and then the trimming of the two Man extensions on the α(1,6)-Man by mannosidase II. Finally, by transferring a second GlcNAc residue on the new terminal α(1,6)-Man residue, GnT II completes the synthesis of the GlcNAc2Man3GlcNAc2 chassis (Figure 1B). Alternatively, transferases such as UDP-GlcNAc:α3-d-mannoside β1,2-N-acetylglucosaminyltransferases from the protozoan parasite Trypanosoma brucei (TbGnT) enable the transfer of GlcNAc on Man3GlcNAc2 to obtain in one step the GlcNAc2Man3GlcNAc2 chassis (Damerow et al. 2014). Given the non-canonical structure of Man5GlcNAc2 lacking additional Man residues linked to the α(1,6)-Man residues (Figure 1A) in C. reinhardtii, the expression of heterologous GnTI was demonstrated to be unsuccessful (Vanier et al. 2017). Such Man5GlcNAc2 structure was demonstrated to be sensitive to the commercially available α(1,2)-mannosidase from Aspergillus saitoi (AsMAN) since N-glycans from wild type (WT) glycoproteins treated with this enzyme exhibited an increase in Man3GlcNAc2 to the detriment of Man5GlcNAc2 (Figure S1, Vanier et al. 2017). Consequently, we have performed in this study the in vivo removal of α(1,2)-Man extensions from the non-canonical Man5GlcNAc2 through the expression in C. reinhardtii of AsMAN (Inoue et al. 1995) or α(1,2)-mannosidase from Arabidopsis (AtMNS1) as this last one was demonstrated to remove α(1,2)-Man from Man9GlcNAc2 to give rise to Man5GlcNAc2 in plants (Liebminger et al. 2009), with the objective to give rise to Man3GlcNAc2 corresponding to the substrate for T. brucei TbGnT. Indeed, the non-canonical Man5GlcNAc2 harboured by glycoproteins in C. reinhardtii results from the removal of terminal Glc residues by ER glucosidase I and II from the ER glycan precursor Glc3Man5GlcNAc2. Proteins then move to the Golgi apparatus for further modifications of Man5GlcNAc2. Therefore, the non-canonical N-linked oligomannoside resides in both the ER and the Golgi apparatus during the secretion of glycoproteins. To investigate in which cell compartment the trimming of the non-canonical N-linked Man5GlcNAc2 would be more efficient, we thus have considered the expression of the Golgi-resident Arabidopsis AtMNS1, as well as the targeting of A. saitoi AsMAN in either the ER or the Golgi apparatus. To target AsMAN in the ER, a sequence coding for the retention signal HDEL was fused at the 3′ end of AsMAN to enable its retrieval from the Golgi apparatus to the ER (AsMANHDEL) (Figure 1C). To address A. saitoi AsMAN in the Golgi apparatus, the endogenous signal peptide of AsMAN was replaced by the cytosol–transmembrane–stem (CTS) domain of Arabidopsis AtMNS1 (AtMNS1AsMAN) that was demonstrated to be responsible for the targeting of this α(1,2)-mannosidase in the cis-Golgi compartment in Nicotiana benthamiana leaf epidermal cells (Liebminger et al. 2009). C. reinhardtii UVM4 strain (WT) was transformed with either AtMNS1, AsMANHDEL or AtMNS1AsMAN. The expression of mannosidases was then confirmed in two positive clones per construct by Western blot (Figure 1D). As expected for a Golgi-located glycosidase, AtMNS1 was immunodetected in the microsomal fraction. By contrast, AsMANHDEL was detected in the soluble proteins fraction of C. reinhardtii, which is consistent with an ER location of the enzyme (Figure 1D). In contrast, the detection of AtMNS1AsMAN in the soluble fraction is questionable and suggests that the fusion of AsMAN with the CTS domain of AtMNS1 failed to anchor the recombinant mannosidase in the membrane of the Golgi apparatus of C. reinhardtii. The impact of α(1,2)-mannosidase expression was investigated on the population of PNGAse A–released N-glycans from proteins of WT and transformants. In the N-glycan profile from WT, the major ion at m/z 1377 was assigned to the non-canonical Man5GlcNAc2 as previously described (Figure 1E, Table S1, Figure S3). Additional N-glycans bearing mono- and di-xylosylated and minor O-methylated N-glycans were also observed as previously shown (Mathieu-Rivet et al. 2013). By contrast, expression of AtMNS1, AsMANHDEL or AtMNS1AsMAN led to a large increase in Man3GlcNAc2 (m/z 1053), to the detriment of Man5GlcNAc2. These results demonstrated that AtMNS1 and AsMAN are able to efficiently in vivo process C. reinhardtii oligomannosides into Man3GlcNAc2 regardless of the cellular location of the recombinant α(1,2)-mannosidase. It should be noted that the expression of α(1,2)-mannosidases only weakly affected the xylosylation levels of N-glycans suggesting that the trimming of Man residues from the non-canonical Man5GlcNAc2 does not affect C. reinhardtii xylosyltransferase activity. Here, we demonstrated that efficient knock-in strategies can be performed in C. reinhardtii for in vivo remodelling of the N-glycan profiles. This suggested that the targeting of glyco-enzymes occurs in microalgae as described for other eukaryotes. However, it should be noted that the expression of AtMNS1AsMAN performed the trimming of Man5GlcNAc2 in Man3GlcNAc2 (Figure 1) although this enzyme was not detected in the membrane fraction, raising the question of the Golgi retention functionality of AtMNS1 CTS in C. reinhardtii. This unexpected result highlights the necessity to better define retention sequences able to properly target glycosidases and glycosyltransferases in the specific compartments of the secretory system of microalgae for a future efficient glycoengineering of biologics. Finally, further studies have to be performed to select a C. reinhardtii strain accumulating Man3GlcNAc2 on their proteins by expressing AsMAN or AtMNS1 in xylosylation mutants. Complementation of this glycoengineered line with TbGnT would enable obtaining C. reinhardtii proteins with mammalian GlcNAc2Man3GlcNAc2 N-glycan chassis. P.L. and E.M.-R. designed the study. J.B. provided technical assistance. C.P. performed the experiments. P.L., M.B. and E.M.-R. prepared the manuscript. M.B. secured the funding. This work was supported by the University of Rouen Normandy. The authors thank Prof. Dr. Ralph Bock (Max Planck Institute of Molecular Plant Physiology, Potsdam) for providing the UVM4 strain. The authors thank Dr. Pierre-Louis Lucas for initiating this idea a few years back. This work was funded by the French government through the ANR agency under the action of the ‘Grand Défi Biomédicaments’, project PhaeomAbs (ANR-21-F2II-0005) and the project DAGENTA (ANR-21-ce20-0038). It was also partly supported by the ‘European Union's Horizon 2020 research and innovation programme under grant agreement 774078 (Pharma-Factory)’. The funding bodies were not involved in the design of the study, collection and interpretation of data nor in the writing of the manuscript. The data that support the findings of this study have not been shared in a public repository yet. Data S1: pbi70482-sup-0001-Supinfo1.docx. Figure S1: pbi70482-sup-0002-Supinfo2.pdf. Data S2: pbi70482-sup-0003-Supinfo3.docx. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.