Heterologous expression of the N-acetylglucosaminyltransferase I dictates a reinvestigation of the N-glycosylation pathway in Chlamydomonas reinhardtii

Eukaryotic N-glycosylation pathways are dependent of N-acetylglucosaminyltransferase I (GnTI), a key glycosyltransferase opening the door to the formation of complex-type N-glycans by transferring a N-acetylglucosamine residue onto the Man5GlcNAc2 intermediate. In contrast, glycans N-linked to Chlamydomonas reinhardtii proteins arise from a GnTI-independent Golgi processing of oligomannosides giving rise to Man5GlcNAc2 substituted eventually with one or two xylose(s). Here, complementation of C. reinhardtii with heterologous GnTI was investigated by expression of GnTI cDNAs originated from Arabidopsis and the diatom Phaeodactylum tricornutum. No modification of the N-glycans was observed in the GnTI transformed cells. Consequently, the structure of the Man5GlcNAc2 synthesized by C. reinhardtii was reinvestigated. Mass spectrometry analyses combined with enzyme sequencing showed that C. reinhardtii proteins carry linear Man5GlcNAc2 instead of the branched structure usually found in eukaryotes. Moreover, characterization of the lipid-linked oligosaccharide precursor demonstrated that C. reinhardtii exhibit a Glc3Man5GlcNAc2 dolichol pyrophosphate precursor. We propose that this precursor is then trimmed into a linear Man5GlcNAc2 that is not substrate for GnTI. Furthermore, cells expressing GnTI exhibited an altered phenotype with large vacuoles, increase of ROS production and accumulation of starch granules, suggesting the activation of stress responses likely due to the perturbation of the Golgi apparatus.


Results
The commonly used cw92 laboratory strain was transformed with the codon optimized sequences encoding for the catalytically active GnTI from Arabidopsis (AtGnTI) or from the diatom Phaeodactylum tricornutum (PtGnTI). Both have been shown to be able to process N-linked glycans. AtGnTI (At4g38240) encodes for a Golgi enzyme that is responsible for the transfer of a terminal GlcNAc residue onto Man 5 GlcNAc 2 N-glycan 18,19 . PtGnTI has been shown to restore in vivo the biosynthesis of complex-type N-glycans in CHO Lec1 mutant that lacks endogenous GnTI activity 17 . These sequences fused to a tag sequence encoding for a V5-epitope were used for nuclear expression in C. reinhardtii [20][21][22] . RT-PCR analyses show that both plant and diatom GnTI were expressed in C. reinhardtii transformed lines ( Fig. 1 and Fig. S1). Further experiments were focused on two AtGnTI and four PtGnTI expressing lines, all of which exhibit the highest transcription levels (Fig. 1a). Furthermore, using anti-V5  (1) or PtGnTI (2) specific primers. Actin (lower panel) was used as an RT-PCR control. (b) Immunodetection of recombinant GnTI in the microsomal fraction isolated from cw92 cells and AtGnTI#3 lines respectively. The immunodetection was performed using an anti-V5 antibody as a primary antibody. A protein extract from CHO cells expressing PtGnTI-V5 (+) was used as a positive control 17 . Full images of the agarose gel and the Western blot are presented in Figs S1 and S2. epitope antibodies, a signal around 56 kDa was immunodetected on a western blot in the microsomal fraction isolated from the transgenic lines but not in that of the non-transformed cells ( Fig. 1b and Fig. S2).
The phenotypic analysis of C. reinhardtii transformed cell lines revealed that the expression of AtGnTI or PtGnTI was correlated to a modification of the cell size. Indeed, measurements of longitudinal cell diameters showed that cells expressing AtGnTI or PtGnTI were statistically enlarged as compared to the cw92 cells and cells transformed with the empty vector (Fig. 2a). Despite the size difference, the growth rates of transformed cells (e) ROS levels in transformed cell lines determined through the oxidation measurement of CMH spin probe by electron paramagnetic resonance spectroscopy. The ROS level (arbitrary units/12 × 10 4 cells hour −1 ) in each transformed cell line was normalized against ROS level measured in cw92 cells. After normalization, a statistical test was performed between GnTI expressing cell lines and cells transformed with the empty vector using Ordinary One-Way ANOVA with n = 3 and p-value fixed at 0.05.
were similar to those of cw92 cells indicating that the swelling observed in transformed lines did not alter cell growth (Fig. 2b).
To examine whether the ultrastructure of swollen cells is altered, cells were high pressure frozen and analyzed by thin-section electron microscopy. To further characterize the cell swelling, the ultrastructure of the transformed lines was analyzed by Transmission Electron Microscopy (TEM). As illustrated in Fig. 2c,d, GnTI expressing cells exhibited a large number of starch granules. In addition, large vesicles appeared in the GnT I expressing lines as compared to the cw92 and cells transformed with the empty vector. In order to check if the increase in size and the starch accumulation were associated to a stress phenotype in transformed cells, we measured the production of reactive oxygen species (ROS) in the different lines using electron paramagnetic resonance (EPR) spectroscopy. Figure 2e shows that ROS levels were significantly higher in the cells expressing AtGnTI or PtGnTI than in the cw92 cells, except for PtGnTI#1 line, in which ROS content was not significantly increased.
Protein N-glycosylation in cw92 and transformed cells were then investigated by mass spectrometry profiling. N-linked glycans were released from total protein extracts using PNGase F, labeled with 2-aminobenzamide (2AB) and analyzed by liquid chromatography coupled to electrospray ionization mass spectrometer (LC-ESI-MS). Although the expression of GnTI induced altered phenotypes in C. reinhardtii, the comparison of the N-glycan structures between cw92 cells and the transformed lines did not reveal any modification of the N-glycan profiles (Fig. S3). A specific search for mono-and discharged ions corresponding to N-glycans exhibiting extra terminal GlcNAc residue was unsuccessful. This suggested that the expression of AtGnTI or PtGnTI did not affect, in a detectable manner, the N-glycosylation of endogenous proteins in C reinhardtii.
Different scenarios may explain the inability of exogenous GnTI to affect the protein N-glycan profiles in C. reinhardtii. This may result from a very weak protein expression of active GnTI in the Golgi apparatus or its mislocalization in this organelle. Absence of the appropriate substrates in the Golgi apparatus could also explain the GnTI inactivity. This concerns the nucleotide-sugar (UDP-GlcNAc) or the oligomannoside Man 5 GlcNAc 2 N-linked to secreted proteins. Man 5 GlcNAc 2 oligomannoside was previously identified in C. reinhardtii cw92 proteins and was assigned to a branched Man 5 GlcNAc 2 by analogy with data published regarding the Golgi N-glycan processing in eukaryotes 15 . The non-effect of GnTI on the N-glycan profiles from C. reinhardtii raised questions regarding the structure of this Man 5 GlcNAc 2 , which therefore needs to be reinvestigated. Here, we made use of Ion Mobility Spectrometry-Mass Spectrometry (IMS-MS), a reliable analytical method which allows the separation of isomers including oligosaccharide isomers [23][24][25] . In this analysis, the ion mobility of Man 5 GlcNAc 2 coupled to 2AB oligosaccharide (sodium adduct) prepared from C. reinhardtii proteins was compared to the ion mobility of branched sodiated 2AB labeled Man 5 GlcNAc 2 obtained from the bovine ribonuclease B. As illustrated in Fig. 3, Man 5 GlcNAc 2 -2AB from C. reinhardtii and from bovine ribonuclease B exhibited different ion mobilities (drift time of 10.12 ms and 11.28 ms, respectively) which suggested that they possess distinct structures. This data indicated that Man 5 GlcNAc 2 in C. reinhardtii is different from the branched Man 5 GlcNAc 2 that results from the trimming of oligomannosides Man [8][9] GlcNAc 2 in the ER and the Golgi apparatus in plants and mammals.
The structure of C. reinhardtii Man 5 GlcNAc 2 isomer was further investigated using electrospray ionization-multistage tandem mass spectrometry (ESI-MS n ) (with n = 2, n = 3 and n = 4) taking advantage of  (Table 1). Similar fragmentation patterns were observed for permethylated Man 5 GlcNAc 2 -2AB prepared from cw92 cells and AtGnTI cells (Fig. S4). In contrast, the fragmentation pattern deduced from the MS 4 analysis of the m/z 882 for Man 5 GlcNAc 2 -2AB isolated from bovine ribonuclease B revealed the specific fragmentations m/z 882 → m/z 1302 → m/z 866 → m/z 648, as it is expected for the branched isomer of Man 5 GlcNAc 2 -2AB ( Fig. 4b) 26 (Table 1). In conclusion, the comparison of ESI-MS n fragmentation patterns suggested that C. reinhardtii proteins carry linear Man 5 GlcNAc 2 ( Fig. 5d) instead of the conventional branched isomer (Fig. 5c).
According to the N-glycan biosynthesis pathway described in land plants and mammals, a linear Man 5 GlcNAc 2 is meant to exhibit terminal α(1,2/6)-Man residues instead of α(1,3/6)-Man. For further confirmation of the C. reinhardtii Man 5 GlcNAc 2 topology, the N-glycan mixture was submitted to exoglycosidase degradations using either Jack bean mannosidase, a non-specific α-mannosidase and Aspergillus saitoi mannosidase, an exoglycosidase specific for α(1,2)-mannose residues 27 . Man 5 GlcNAc 2 -2AB from C. reinhardtii was efficiently converted into ManGlcNAc 2 -2AB by Jack bean mannosidase after removal of the four α-mannose units. In contrast, Man 3 GlcNAc 2 -2AB was the major end product after treatment with Aspergillus saitoi α(1,2)-mannosidase that results from the removal of two terminal α(1,2)-mannose residues from the linear trimannoside arm (Fig. S5). Taken together, both mass spectrometry and enzyme sequencing analyses demonstrated that C. reinhardtii glycoproteins carry a non-canonical linear Man 5 GlcNAc 2 , as depicted in Fig. 5d.
The occurrence of a linear Man 5 GlcNAc 2 oligosaccharide onto C. reinhardtii gycoproteins may result either from the trimming of Man 8-9 GlcNAc 2 in the Golgi apparatus by the action of α-mannosidases or from a truncated biosynthesis of the LLO in the ER (Fig. 5). The latter hypothesis was investigated by isolation and analysis of C. reinhardtii LLO. LLO were isolated from cw92 microsomal preparations of C. reinhardtii using a methanol/ chloroform extraction procedure according to a protocol adapted from 27,28 . The oligosaccharide was hydrolyzed from the PP-dolichol anchor by mild acidic cleavage and then permethylated as previously reported 27,29 . A predominant ion at m/z 2192.7 was observed using MALDI-TOF-MS and could be assigned to a permethylated oligosaccharide containing 8 hexoses and 2 HexNAc residues (Hex 8 HexNAc 2 ) (Fig. S6). A minor ion at m/z 1987.6 could be assigned to Hex 7 HexNAc 2 (Fig. S6). The sequence of this oligosaccharide was analyzed by ESI-MS n and compared to the one isolated from LLO of the YG170 (alg3) yeast mutant (Fig. 6). Indeed, this strain lacks α1,3-mannosyltransferase ALG3 activity and accumulates Glc 3 Man 5 GlcNAc 2 LLO in the ER 30,31 . Both ESI-MS n data exhibited similar fragmentation patterns that are consistent with a linear arrangement in the oligosaccharide. Taken together, these analyses demonstrated that C. reinhardtii accumulates a predominant linear truncated Glc 3 Man 5 GlcNAc 2 LLO in the ER (Fig. 5b).

Discussion
In previous work, we characterized N-glycan structures linked to endogenous proteins in C. reinhardtii and showed they were mostly oligomannosides and for 30% novel mature structures containing xylose residues and . These structures resulted from a Golgi GnTI-independent processing of oligomannosides as the bioinformatics analysis of the genome has revealed that it lacks GnTI. Furthermore, no N-glycan harboring terminal GlcNAc residues has been identified in the whole glycan population 11,15 . This situation contrasts with most eukaryotic organisms in which the GnTI is a key Golgi enzyme in the N-glycosylation pathway opening the door to the biosynthesis of structurally diverse mature N-glycans 12 . In this canonical pathway, the addition of a first GlcNAc by GnTI is required for the sequential activity of a large repertoire of other specific transferases giving rise to complex N-glycans involved in numerous biological processes, such as intracellular communication and signaling 13,14 . Inactivation of GnTI in those organisms induces strong developmental phenotypes. For instance, GnTI-null embryos of mouse die at about 10 days after fertilization indicating that mature N-glycans are required for morphogenesis in mammals 32,33 . Also, inactivation of GnTI reduces the viability in worm and fly 34,35 . In plants, GnT I mutants exhibit a stress phenotype, thus suggesting a role for mature N-glycans in specific physiological processes 36,37 . In rice, gntI mutants, impaired in the N-glycans maturation showed severe developmental defects, resulting in early lethality, associated to reduced sensitivity to cytokinin 36 .
GnTI genes are predicted in different microalgae genomes. In the diatom P. tricornutum, PtGnTI was demonstrated to encode for an active glycosyltransferase 17 . Therefore, the capacity of C. reinhardtii to express GnTI from Arabidopsis and from P. tricornutum was analyzed to investigate whether C. reinhardtii N-glycan biosynthesis can be complemented with this key transferase and shift into a GnTI -dependent pathway. Moreover, in a biotech context, the production, in microalgae, of recombinant glycoproteins for therapeutic applications will require the engineering of their endogenous N-glycosylation pathway for the production of biopharmaceuticals exhibiting human-compatible N-glycans. Therefore, implementation of a GnTI -dependent pathway is a prerequisite for any production of glycosylated biopharmaceuticals in C. reinhardtii.
Transgenic lines expressing the Arabidopsis or diatom GnTI were obtained and GnTI protein was immunodetected as expected in the microsomal fraction. However, further experiments are required to confirm the GnTI localization within the Golgi apparatus. Mass spectrometry analyses of N-glycan profiles from proteins secreted in transgenic lines did not show any modification of the N-glycan population by comparison with the cw92 cells.
Among the different scenarios that may explain this result, the absence of UDP-GlcNAc nucleotide sugar in the Golgi apparatus was first considered. The transport of the cytosolic nucleotide sugars across the Golgi membrane is performed by Nucleotide Sugar Transporters (NSTs). Searching for NST orthologues in the C. reinhardtii genome allowed the identification of 23 candidate genes of which the deduced amino-acid sequences harbor the characteristic Triose Phosphate Translocator (TPT) domain (Pfam 03151) present in NSTs (Mathieu-Rivet et al., in press). However, as most of the characterized NSTs has been shown to transport at least two distinct substrates, determination of their specificity for nucleotide sugars based on sequence homologies with other Golgi NSTs remains difficult without additional biochemical evidence 38 . Therefore, further experimental work is needed to determine whether a specific Golgi UDP-GlcNAc transporter exists in C. reinhardtii. In relation to this, we cannot ruled out that the cytosolic abundance of UDP-GlcNAc may not be sufficient to supply the Golgi apparatus which would consequently limit GnTI in C. reinhardtii. Inability of GnTI to affect the endogenous N-glycan population in C. reinhardtii transformed lines may also result from the absence of the appropriate glycan substrate. Man 5 GlcNAc 2 oligomannoside was previously identified in C. reinhardtii 15 and considered as being the branched isomer substrate for GnTI by analogy with mammals and plants N-glycan pathways. Its structure was reinvestigated by mass spectrometry and enzyme sequencing and compared to a branched Man 5 GlcNAc 2 standard from mammals 26 . Ion mobilities determined by IMS-MS and fragmentation patterns resulting from ESI-MS n clearly show that C. reinhardtii and mammalian oligomannosides differ in their shape and sequence. Moreover, the fragmentation pattern determined by MS n and enzyme digestion with an α(1,2)-mannosidase are consistent with the presence of a linear Man 5 GlcNAc 2 on C. reinhardtii proteins as depicted in Fig. 5d. This oligosaccharide may result from a truncated LLO biosynthesis occurring in the ER (Fig. 5). Indeed, ALG3, ALG9 and ALG12 candidates are not predicted in the C. reinhardtii genome 15,39,40 . These ER enzymes are involved in the completion of the biosynthesis of the LLO precursor Man 9 GlcNAc 2 -PP-Dol prior to its glucosylation (Fig. 5a). As a consequence, the absence of ALG3, ALG9 and ALG12 activities results in the secretion, from ER, of proteins carrying a linear Man 5 GlcNAc 2 instead of Man 8-9 GlcNAc 2 (Fig. 5b). In contrast, in land plants and mammals, branched Man 5 GlcNAc 2 is obtained by the trimming of mannose residues of Man 8-9 GlcNAc 2 oligomannosides by Golgi α-mannosidases (Fig. 5a). However, linear Man 5 GlcNAc 2 oligomannoside could also result from the trimming of Man 8-9 GlcNAc 2 by Golgi α-mannosidases predicted in C. reinhardtii genome 11, 15 but having different glycan specificities as compared to those of homologous enzymes involved in mammalian and plant N-glycan pathways. To discriminate between the two possibilities, C. reinhardtii LLO was isolated, characterized by mass spectrometry analysis and compared with the one extracted from the ALG3 deficient yeast mutant 31 . These analyses demonstrated that this microalga accumulates a linear Glc 3 Man 5 GlcNAc 2 ( Fig. 6 and Fig. S6). Such a truncated ER pathway has been already characterized in some unicellular organisms such as the coccidian parasites Toxoplasma and Cryptosporidium [41][42][43][44][45] . In conclusion, data obtained in this study revealed a truncated ER N-glycan pathway in C. reinhardtii and required the reevaluation of the previously published N-glycan pathway 15 . In this reassessed N-glycan processing, Man 5 GlcNAc 2 results from the deglucosylation of Glc 3 Man 5 GlcNAc 2 precursor in the ER and its methylation and xylosylation in the Golgi apparatus (Fig. 5). Location of xylose residues on Man 5 GlcNAc 2 was reinvestigated by ESI-MS n (Fig. S7). This analysis confirmed the presence of the first xylose residue onto the β-Man 15 . In addition, the second xylose could be positioned either  Fig. 5.
Surprisingly, cells expressing GnTI exhibited an altered phenotype although no modification of the N-glycans has been observed. The presence of large vacuoles as well as the increase of ROS production observed in transformed cell lines suggested that transformation with GnTI induced the activation of stress responses such as autophagy and oxidative stress 46,47 . Previously, Pérez-Martin and collaborators showed that ER stress caused by tunicamycin or DTT triggers autophagy 48 . It can be hypothesized that AtGnTI or PtGnTI may localize in the ER of C. reinhardtii. Indeed, a previous study published by Schoberer and coworkers in 2009 demonstrated that overexpressed GnTI in tobacco recruit the ER protein Sar1p resulting in a GnTI-Sar1p association in the ER membranes 49 .
In addition, chlamydomonas cells expressing AtGnTI or PtGnTI accumulate starch. In BY2 cells and chlamydomonas noctigama, Hummel and collaborators 50 reported in 2010 that the disassembly of the Golgi apparatus using the secretion inhibitor Brefeldin A (BFA) caused the accumulation of plastid starch. This starch increase was also observed when the CopII-mediated ER to Golgi transport was inhibited in tobacco plants expressing a dominant negative version of the small GTPase Sar1p 51 . Moreover, it was proposed that disruption of the Golgi apparatus resulting in the loss of secretory activity may cause the redirection of free carbohydrates to the plastids where they would be converted into starch 51 .
Therefore, we postulate here that the heterologous expression of GnTI may affect Golgi organization or the endosomal system, which would consequently disturb the biosynthesis of glycans or glycoconjugates and the cell biology of C. reinhardtii. Additional work and phenotypic studies of transformants expressing a non-functional GnTI would allow the confirmation of this hypothesis.

Methods
Strains and growth conditions. CC-503 cw92 strain, later called cw92 cells, was obtained from the Chlamydomonas Culture Collection at Duke University (Durham, NC, USA) and grown in batch cultures at 25 °C, illuminated with 150 μmol m −2 s −1 using TAP medium redirection of free carbohydrates to the plastids where they would be converted into starch 52 .
Vector construction and transformation. The sequences encoding for AtGnTI (At4g38240) and PtGnTI (gi: 307604450) fused to the V5 epitope 17 were codon optimized and synthesized by Sloning Biotechnology. Optimized sequences were cloned as a XbaI/NdeI fragment in the pSL18 vector 53 . pSL18 (empty vector), pSL18-AtGnTI or pSL18-PtGnTI construct was introduced using the glass beads method 54 . To detect the transgene in genomic DNA, screening was performed by PCR using V5-reverse GGAGTCCAGGCCCAGCAGG and with AtGnTI-forward GCCCTAAGTGGCCAAGGC or PtGnT1-forward CCAGTCCAAGTGGCCGGGC as primers.
RT-PCR analysis. Total RNA were isolated from fresh cell pellets (1.5 × 10 6 cells), using TRIzol (Invitrogen). gDNA contamination was removed by a Turbo DNase treatment (Thermo Fisher Scientific). Reverse transcription was performed on 2 µg of RNA using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). The transcription level of AtGnTI and PtGnTI transgenes was analyzed by PCR using the V5-reverse primer, with either AtGnTI-forward or PtGnT1-forward primers. Each PCR contained 2 µL of diluted cDNA (1/10), 0.8 mM of each primer, 0.25 mM dNTPs and 0.6 U GoTaq polymerase (Promega) and was performed in the reaction buffer provided by the manufacturer, according the following program: 95 °C for 5 min, 35 cycles of 95 °C for 30 s, 62 °C for 30 s, 72 °C for 30 s and a final step of 72 °C for 5 min. The expression of the actin gene was monitored as a control using CrActin-forward CGCTGGAGAAGACCTACGAG and CrActin-reverse GGAGTTGAAGGTGGTGTCGT as primers.
Microsomal preparations and Western-blot analysis. 1.4 × 10 7 of C. reinhardtii cells were collected (2,500 g for 5 min) and washed with a 20 mM potassium phosphate buffer at pH 7.4. All the following preparation steps were carried out at 4 °C. The cell pellet was broken with 2 mL of protease inhibitor cocktail 25X (Roche) dissolved in 10 mM potassium phosphate buffer (pH 7.4) using the FastPrep-24 ™ 5G 15 . Samples were then spun (300 g for 3 min) in order to remove intact cells and debris. The supernatant was collected and centrifuged (20,000 g for 30 min) in order to eliminate pigments and chloroplasts. Finally, the supernatant was ultracentrifuged (100,000 g for 1 h) to pellet the microsomal fraction. Immunodetection using anti-V5 antibodies has been performed as described in ref. N-glycans were analyzed with the QTOF analyser. Full autoMS scans from 290 to 1,700 m/z and autoMS/MS from 59 to 1,700 m/z were recorded. In every cycle, a maximum of 5 precursors sorted by charge state (1+ and 2+ preferred) were isolated and fragmented in the collision cell with fixed collision cell energy at 15 eV. Scan speed raise based on precursor abundance (target 5,000 counts/spectrum) and precursors sorted only by abundance. Active exclusion of precursors was enabled and the threshold for precursor selection was set to 1,000 counts. The ESI acquisition parameters in positive mode were: capillary voltage; 1.9 kV, drying gas temperature; 250 °C, gas flow (air); 11 L min −1 ; fragmentor voltage, 360 V; Skimmer1 voltage, 65 V and OctopoleRFPeak voltage, 750 V. Electron microscopy. High pressure freezing was performed with the freezer HPM100 Leica-microsystems.

IM-MS analyses.
Prior to freezing, 72 h old C. reinhardtii cells were treated at room temperature 1 hour with 100 mM mannitol as cryoprotectant diluted in fresh culture medium. Pre-treated C. reinhardtii cells were transferred into an aluminium cryocapsule covered by soy lecithin dissolved at 100 mM in chloroform. Excess medium was absorbed by filter paper. After fixation on the loading device, samples were frozen according to a maximum cooling rate of 20,000 °C s −1 and a pressure of 2,100 bars. Samples were transferred to a freeze substitution automate (AFS2, Leica) pre-cooled to −110 °C. Samples were substituted in anhydrous acetone with 0.5% uranyl acetate and 0.5% osmium tetroxide at −90 °C for 72 h. The temperature was gradually raised to −60 °C using a gradient of +2 °C h −1 and stabilized during 12 h, then gradually raised to −30 °C by using the same gradient during 12 h and gradually raised again to +4 °C. Then, samples were rinsed twice at room temperature with anhydrous acetone. Infiltration was then processed in acetone-Spurr's resin (  Reactive oxygen species measurement. Reactive oxygen species (ROS) production was evaluated by electron paramagnetic resonance (EPR) spectroscopy. Cells were incubated at room temperature in the dark for 60 min in Krebs-HEPES buffer containing 5 µM diethyl dithiocarbamate, 25 µM deferoxamine, and the spin probe 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethyl pyrrolidine hydrochloride (CMH; 500 µM; Noxygen, Elzach, Germany). Spectra of the oxidized product of CMH (CM.) were recorded from frozen samples with a X-band spectrometer (MS-400; Magnetech, Berlin, Germany) with the following acquisition parameters: microwave power, 1 mW; modulation amplitude, 5 G; sweep time, 60 s; and 1 scans. After correction of the baseline, the total amplitude of the signal was measured and expressed in arbitrary units produced per 12 × 10 4 cells for 60 min. After EPR analyses, total chlorophyll was quantified for each sample according to ref. 57. The quantity ScieNtiFic RepoRTS | 7: 10156 | DOI:10.1038/s41598-017-10698-z of total chlorophyll was used to normalize the signal intensity evaluated by EPR for each sample. The results were statistically analyzed using GraphPad Prism ® software. The ROS level in each transformed cell line was normalized against ROS level measured in cw92 cells. To determine the significant level, a statistical test was performed between the GnTI transformed cell lines and the lines transformed with the empty vector using Ordinary One-Way ANOVA with n = 3 and p-value fixed at 0.05.
LLO preparation. The LLO extraction was performed on microsomal fractions from C. reinhardtii according to a protocol adapted from 27,28 . Released oligosaccharides were then lyophilized and resuspended in 500 µL of water and purified on a carbograph column (Hypersep Hypercarb, Thermo Scientific) using the manufacturer's instructions.