Comparative analysis of the site-specific N-glycosylation of human lactoferrin produced in maize and tobacco plants Be ´ ne ´ dicte Samyn-Petit 1 , Jean-Pierre Wajda Dubos 2 , Fre ´ de ´ ric Chirat 1 , Bernadette Coddeville 1 , Gre ´ gory Demaizieres 2 , Sybille Farrer 2 , Marie-Christine Slomianny 1 , Manfred Theisen 2 and Philippe Delannoy 1 1 Unite ´ de Glycobiologie Structurale et Fonctionnelle, UMR CNRS 8576, Laboratoire de Chimie Biologique, Universite ´ des Sciences et Technologies de Lille, Villeneuve d’Ascq, France; 2 Meristem Therapeutics, Clermont-Ferrand, France We have compared the site-by-site N-glycosylation status of human lactoferrin (Lf) produced in maize, a monocotyle- don, and in tobacco, used as a model dicotyledon. Maize and tobacco plants were stably transformed and recombinant Lf was purified from both seeds and leaves. N-glycopeptides were generated by trypsin digestion of recombinant Lf and purified by reverse-phase HPLC. The N-glycosylation pat- tern of each site was determined by mass spectrometry. Our results indicated that the N-glycosylation patterns of recombinant Lf produced in maize and tobacco share common structural features. In particular, both N-glycosy- lation sites of each recombinant Lf are mainly substituted by typical plant paucimannose-type N-glycans, with b1,2-xy- lose and a1,3-linked fucose at the proximal N-acetylgluco- samine. However, tobacco Lf shows a significant amount of processed N-glycans with one or two b1,2GlcNAc linked to the trimannose core, which are weakly expressed in maize Lf. Finally, no Lewis a epitope was observed on tobacco Lf. Keywords: glycosylation; N-glycopeptides; maize; tobacco; human lactoferrin. Several expression systems including bacteria, yeast, fungi, insect and mammalian cells, or transgenic animals are used to produce recombinant human proteins. This last decade, much attention has been paid to the plant expression systems in order to express mammalian proteins. By using strong promoters, high levels of expression can be achieved and production costs are relatively low [1]. In addition, plant expression systems are much less likely to harbor human pathogens than mammalian expression systems. This is a great advantage of the plant system for the production of therapeutic proteins such as vaccines and antibodies. Direct oral administration of plant material containing recombin- ant therapeutic molecules has been investigated for delivery of antigens and antibodies for active or passive immuniza- tion [2,3]. High-level production of recombinant human milk proteins in rice is also investigated as an addition to infant formula and baby foods [4]. Plant biologists have been able to express recombinant proteins in various plants including mono- and dicotyl- edons. Moreover, it is possible to direct the expression to specific parts of the plant, such as fruits, seeds, leaves and tubers. Several examples have shown that plants allow the production of complex human proteins that appear to have biological properties and activities similar to those of the native proteins, such as human collagens [5], human growth hormone [6] and antibodies [7,8]. Most therapeutic proteins are glycoproteins and glyco- sylation is often essential for the stability, the solubility, a proper folding and biological activity. In plants, even if the first steps of N-glycosylation that take place in the endoplasmic reticulum are identical to other eukaryotic cells, the Golgi processing of N-glycan chains displays some major differences compared to that of mammalian cells [9,10]. High-mannose-type N-glycans of plants are similar to those found in other eukaryotes. However, N-glycans found in plants are mostly of the paucimannose- type (Man 3 GlcNAc 2 -based structure), even if complex-type N-glycans with a Lewis a terminal sequence (Galb1– 3[Fuca1–4]GlcNAc-R) have been reported [11]. First described in sycamore [12,13], the Lewis a epitope is widespread among plants, but several examples have underlined the lack of such complex N-glycans in a number of mono- and dicotyledon species [14,15]. These findings indicate that plants do not exhibit the same potential of N-glycosylation, according to the level of expression of several key enzymes involved in the initiation (i.e. b1,2-N-acetylglucosaminyltransferases I and II) and the elongation (i.e. galactosyltransferases, fucosyltrans- ferases) of antennae of complex N-glycans, but also of the b-hexosaminidase, which governs the paucimannose- type N-glycan pathway [16]. Different plant species share similarities in their N-glycosylation, as the absence of N-acetylneuraminic acid residues in the terminal position Correspondence to P. Delannoy, Unite ´ de Glycobiologie Structurale et Fonctionnelle, UMR CNRS no 8576, Laboratoire de Chimie Biologique, Universite ´ des Sciences et Technologies de Lille, F-59655 Villeneuve d’Ascq, France. Fax: + 33 320 43 65 55, Tel.: + 33 320 43 69 23, E-mail: Philippe.Delannoy@univ-lille1.fr Abbreviations: Lf, lactoferrin; mLf, maize recombinant lactoferrin; tLf, tobacco recombinant lactoferrin. (Received 20 March 2003, revised 2 June 2003, accepted 5 June 2003) Eur. J. Biochem. 270, 3235–3242 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03706.x of the antennae, and the presence of a bisecting b1,2-xylose, and of an a1,3-fucose residue instead of a1,6-fucose, linked to the proximal N-acetylglucosamine. In a previous paper, we described the potential of maize glycosylation, a monocotyledon expression system, by using human lactoferrin (Lf) as a model glycoprotein that was expressed in the endosperm of seeds [17]. The molecular structure of human Lf has been studied in detail. This 80 kDa glycoprotein contains three potential N-glycosyla- tion sites located at Asn138, Asn479 and Asn624, respect- ively. The two first N-glycosylation sites are substituted by complex-type N-glycans whereas the third one (Asn624) is mostly unglycosylated [18, 19]. Human Lf plays a central role in numerous biological processes [20]. Among them, the antibacterial and anti-inflammatory activities of human Lf have led to its large-scale production by recombinant methods to supplement infant foods. In this paper, we report the site-by-site analysis of the maize recombinant lactoferrin (mLf) in comparison with the lactoferrin pro- duced in tobacco (tLf), used as a model for dicotyledons. For that purpose, the recombinant Lf, purified from both expression systems, was digested by trypsin after reduction and alkylation. Peptides were fractionated by RP-HPLC and analysed by MALDI-TOF. Glycopeptides and the corresponding peptides, generated from the glycopeptides by N-glycosidase A treatment, were also analysed by MALDI-TOF and ES-MS. Materials and methods Materials Sequencing-grade modified trypsin was from Promega (Zu ¨ rich, Switzerland). HPLC analyses were carried out on a Spectra Physics apparatus equipped with a semiprepar- ative Vydac C 18 ultrasphere (9.4 · 250 mm; 5 lm) column. Recombinant peptide-N-glycosidase F (PNGase F) from Escherichia coli and peptide-N-glycosidase A (PNGase A) from almonds were purchased from Roche Molecular Biochemicals (Meylan, France). All other reagents were of highest quality available. Isolation of hLf cDNA and vector construction hLf cDNA was according to Salmon et al. [21], and expression vectors containing the lactoferrin sequence, fused to the sporamin signal peptide from sweet potato for secretion, were obtained for maize as described in [17] and for tobacco as described in [21]. Transformation, production and purification of maize Lf and tobacco Lf As described previously, three successive generations of transgenic corn seeds were produced in a greenhouse (T1 to T3 generations) using self-pollinations and cross-pollina- tions with an untransformed elite inbred maize variety [17]. To obtain a greater quantity of raw material for extraction and large scale batch purification of mLf, a field trial was performed in the south of France throughout summer 1998 on a 0.45 ha plot of land. T3 seeds were sown by the end of May 1998 and T3 transgenic plants were crossed with the same elite inbred maize variety as mentioned above. Mature T4 seeds were then harvested in October 1998, with 32% humidity. They were dried at low temperature, cleaned to eliminate the refuse of the ears and bad grains, and stored in big bags. Maize Lf was extracted and purified from T4 corn seeds as described previously [17]. For tobacco, plant transformations were carried out according to Salmon et al. [21]. For extraction and purifi- cation of tLf, fresh tobacco leaves were harvested from the greenhouse and ground in liquid nitrogen. The raw material was treated and Lf was purified by the same protocol as for maize, with the following modifications. The ratio of biomass to extraction buffer volume was 1/4 and the maceration time was 2 h. Reduction, alkylation and tryptic proteolysis hLf, mLf and tLf (60 nmol of each) were solubilized in 6 M guanidinium chloride at a final concentration of 5 mgÆmL )1 , reduced and carboxamidomethylated as described previ- ously [22]. After extensive dialysis of denaturated proteins against Tris/HCl buffer (100 m M , pH 8.0), sequencing- grade modified trypsin was added to a final enzyme-to- substrate ratio of 1/100 (w/w) and incubated 16 h at 37 °C. Tryptic digestions were stopped by storing the hydrolysates at )20 °C. HPLC analysis of tryptic digests Peptides and glycopeptides generated by tryptic digestions were separated by RP-HPLC with a semipreparative Vydac C 18 ultrasphere (9.4 · 250 mm; 5 lm) column and eluted with a linear gradient of 0–80% acetonitrile containing 0.1% (v/v) trifluoroacetic acid for 90 min at a flow rate of 2mLÆmin )1 . Elution was monitored at 214 nm and peaks were collected, lyophilized and stored at )20 °C. Peptides and/or glycopeptides (2 nmol) were spotted on a silica gel 60 aluminium sheet (Merck, Germany) and revealed by using 0.2% (w/v) orcinol in a 60% (v/v) sulphuric acid solution. Enzymatic deglycosylation of glycopeptides The N-linked oligosaccharides from hLf glycopeptides (50 pmol) were enzymatically released with 0.25 U PNGase F in ammonium bicarbonate buffer (20 m M , pH 8.0) whereas those of mLf and tLf were released with 0.0125 mU PNGase A in sodium acetate buffer (100 m M , pH 5.1). After overnight incubation at 37 °C, peptides were desalted by C 18 phase Sep-Pak cartridges (Waters, MA, USA) and eluted with 80% acetonitrile containing 0.1% trifluoroacetic acid. After lyophilization, peptides (10 pmol) were analysed by MALDI-TOF mass spectrometry. Mass spectrometry analyses of peptides and glycopeptides MALDI-TOF. MALDI-TOF mass spectra were acquired on a Voyager Elite (DE-STR) linear or reflectron mass spectrometer (Perspective Biosystems, Framingham, MA, USA) equipped with a pulsed nitrogen laser (337 nm) and a gridless delayed extraction ion source. Samples were 3236 B. Samyn-Petit et al.(Eur. J. Biochem. 270) Ó FEBS 2003 analysed in delayed extraction mode using an accelerating voltage of 20 kV, a pulse delay time of 200 ns and a grid voltage of 66%. Detector bias gating was used to reduce the ion current below masses of 500 Da. Samples were prepared by mixing directly on the target 1 lL of peptide or glycopeptide solution (10–50 pmol) with 1 lL of 2,5-dihydroxybenzoic acid matrix solution (10 mgÆmL )1 in CH 3 OH/H 2 O, 70 : 30, v/v). The samples were allowed to dry for about 5 min at room temperature. Between 150 and 200 scans were averaged for every spectrum shown. ES-MS and CID-MS-MS. Mass spectra were acquired on Micromass Quattro II triple quadripole mass spectrometer operating with an API ion source in positive ion electrospray mode. Glycopeptide samples were diluted in CH 3 CN/H 2 O (50/50, v/v), 0.2% (v/v) formic acid, to a final concentration of about 15 pmolÆlL )1 and infused at 8 lLÆmin )1 .Mass spectra were acquired by scanning MS1 with appropriate mass range, while MS-MS analyses were performed by transmitting the appropriate precursor ion from MS1 to the collision cell. The collision gas used was argon at a pressure of 4.9 · 10 )3 mbar with an appropriate collision energy (25–50 eV). Product ions were scanned with MS2. Peptide sequencing. Nano-electrospray mass spectrometric analyses were performed using a QSTAR Pulsar quadru- pole time-of-flight (Q-TOF) mass spectrometer (AB/MDS Sciex, Toronto, Canada) equipped with a nano-electrospray ion source (Protana, Odense, Denmark). Peptides dissolved in MeOH/H 2 O (50/50, v/v), 0.1% (v/v) formic acid at a concentration of 10 pmolÆlL )1 were sprayed from gold- coated Ômedium lengthÕ borosilicate capillaries (Protana, Odense, Denmark). A potential of ± 800 V was applied to the capillary tip. The declustering potential varied between ± 60 V and ± 110 V and the focusing potential was set at )100 V. The molecular ions were selected in the quadrupole analyser and partially fragmented in the hexapole collision cell, with the pressure of collision gas (N 2 )5.3· 10 )5 Torr. The collision energy was varied between 40 and 110 eV depending on the sample. QSTAR spectra were acquired by accumulation of 10 MCA scans over the m/z range 700–1000 Da and 900– 2000 Da for MS analyses, and over m/z 150–1000 for MS-MS analyses. Signal detection was performed with a multichannel plate detector and time to digital conversion. Resolution was measured as the full width at half maximum and was 7000 in the used mass range. This was measured for both MS and MS-MS modes. All signals were mono- isotopically resolved and TOF calibration was performed with a solution of 4 pmolÆlL )1 of taurocholic acid in acetonitrile/H 2 O (50/50, v/v), 2 m M ammonium acetate. Results Purification of N-glycopeptides from trypsin digestion of natural and recombinant lactoferrins The hydrolysates obtained after tryptic digestion of reduced, alkylated hLf, mLf and tLf were fractionated by RP-HPLC and eluted with a linear gradient of 0–80% acetonitrile in 0.1% (v/v) trifluoroacetic acid. The elution profiles are shown in Fig. 1. Even if slight differences can be observed, the three elution profiles were very similar, indicating that the trypsin cleavage sites were identical between natural and both recombinant Lfs. The different fractions were collected and analysed by MALDI-TOF. The glycopeptide-contain- ing fractions were also confirmed by orcinol staining. In each case, we have identified two main glycopeptide fractions, eluted at 47 min for fraction 1 and at 56 min for fraction 2, which correspond to the glycosylation sites Asn479 and Asn138, respectively. These fractions were named H 1 and H 2 ,M 1 and M 2 ,T 1 and T 2 for hLf, mLf and tLf, respectively, as indicated in Fig. 1. In addition, we have also identified a peptide fraction named H 3 ,M 3 and T 3 , which corresponds to the unglycosylated Asn624 site. Structural analysis of N-glycopeptides Glycopeptide-containing fractions were analysed by MALDI-TOF before and after N-glycanase treatments. Fig. 1. Fractionation of tryptic digests of natural and recombinant lactoferrins by RP-HPLC. Trypsin digests of hLf (A), mLf (B) and tLf (C) were fractionated on a Vidac C 18 ultrasphere column and eluted by a linear gradient (0–80%) of acetonitrile containing 0.1% (v/v) tri- fluoroacetic acid. Peptides were detected at 214 nm. The glycopeptide- containing fractions are indicated. Ó FEBS 2003 N-glycosylation of maize and tobacco lactoferrin (Eur. J. Biochem. 270) 3237 MALDI mass spectra of the glycopeptide fractions reveal the heterogeneity of these fractions suggesting several glycoforms and/or peptidic backbone mixtures (Fig. 2). To identify the glycopeptides, PNGase treatment was carried out (Fig. 3). The MALDI mass spectra obtained after deglycosylation of the fractions H 1 ,M 1 and T 1 show the disappearance of peaks between 3800 and 4600 Da for hLf, and between 3000 and 3700 Da for the two recombin- ant lactoferrins. In contrast, these spectra reveal the appearance of peaks exhibiting [M + H] + ions at m/z 2049, 2053, 2097 and 2154. The peaks at 2053 and 2097 Da correspond to the peptide TAGWNIPMGLLFNQTGSCK (467–485) (Asn479 peptide), which has been identified as the second N-glycosylation site in hLf, with an oxidized methionine (expected average mass 2054.37) or the expected carboxamidomethylated cysteine (expected average mass 2095.42), respectively. The ion at m/z 2049 corresponds to the Asn479 peptide with a carboxamidomethylated cysteine and an oxidized methionine, which has lost the methylsul- foxide moiety [23] (expected average mass 2047.37). The peak at 2154 Da could correspond to this peptide with an extra carboxamidomethylated amino acid, that sequencing trials did not allow us to locate either by mass spectrometry or by Edman degradation. Mass spectra obtained for the N-glycosylation site Asn479 of natural (H 1 ) and recombinant lactoferrins (M 1 and T 1 ) are presented in Fig. 2. Concerning H 1 ,themass spectrum displays five main glycopeptides exhibiting [M + H] + ions at m/z 3921.16, 4066.70, 4212.43, 4358.24 and 4503.76 that are consistent with oligosaccharide struc- tures Hex5(dHex)HexNAc 4 ,NeuAcHex 5 HexNAc 4 ,Neu- AcHex 5 (dHex)HexNAc 4 ,NeuAcHex 5 (dHex 2 )HexNAc 4 and NeuAc 2 Hex 5 (dHex)HexNAc 4 (Hex, hexose; dHex, deoxyhexose) linked to the Asn479 peptide m/z 2154 (Fig. 2A). Three other minor ions at m/z 3863.84, 4008.99 and 4154.98 are also detected, which correspond to the oligosaccharide structures Hex5(dHex)HexNAc 4 ,NeuAc- Hex 5 HexNAc 4 ,NeuAcHex 5 (dHex)HexNAc 4 linked to the Asn479 peptide m/z 2097. As shown in Fig. 2B, MALDI mass measurements of M 1 indicate one major peak exhibiting [M + H] + ion at m/z 3220.95 that is consistent with the oligosaccharide structure Hex3(dHex)(Pen)Hex- NAc 2 (Pen, pentose) and three minor glycopeptides exhi- biting [M + H] + ions at m/z 3059.02, 3423.88 and 3626.77 consistent with the structures Hex2(dHex)(Pen)HexNAc 2 , Hex3(dHex)(Pen)HexNAc 3 and Hex3(dHex)(Pen)Hex- NAc 4 , respectively, all structures being linked to the Asn479 peptide. Glycopeptide T 1 MALDI-MS analysis Fig. 2. MALDI-TOF mass spectra of hLf, mLf and tLf Asn479-glycopeptides. After HPLC fractionation, glycopeptides (25–50 pmol) were analysed by MALDI-TOF using 2,5-dihydroxybenzoic acid as matrix. (A) The hLf spectrum was recorded in positive ion linear mode, while the mLf (B) and tLf (C) spectra were recorded in positive ion reflective mode. d, Mannose; j, N–acetylglucosamine; s,galactose; , a1,3-fucose; , a1,6-fucose; n, sialic acid; , xylose 5. 3238 B. Samyn-Petit et al.(Eur. J. Biochem. 270) Ó FEBS 2003 (Fig. 2C) displays three major glycopeptides at 3222.62, 3425.58 and 3628.60 Da consistent with oligosaccharide structures Hex3(dHex)(Pen)HexNAc 2 , Hex3(dHex)(Pen) HexNAc 3 and Hex3(dHex)(Pen)HexNAc 4 , respectively, probably linked to the oxidized methionine Asn479 site. The [M + H] + ions at m/z 3265.04, 3468.08 and 3672.15 correspond to these glycoforms linked to the carbamido- methylated cysteine Asn479 site. Concerning the glycopeptidic fractions H 2 ,M 2 and T 2 , we used the same strategy of analysis by MALDI-MS (data not shown). Spectra obtained after deglycosylation of H 2 ,M 2 and T 2 reveal one major peak corresponding to a [M + H] + ion at m/z 3232.05, 3232.69 and 3232.45, respectively. This peak at 3232 Da corresponds exactly to the first N-glycosylation site TAGWNVPIGTLRPFL NWTGPPEPIEAAVAR(123–152) (Asn138). The sequence of this peptide was also verified by mass spectrometry. MALDI-MS and ES-MS analysis of H 2 allowed us to detect three glycopeptide peaks represented by [M + H] + ions at m/z of 5291, 5437 and 5582 consistent with NeuAc- Hex 5 (dHex)HexNAc 4 ,NeuAcHex 5 (dHex 2 )HexNAc 4 and NeuAc 2 Hex 5 (dHex)HexNAc 4 linked to the Asn138 peptide. MALDI mass measurements of M 2 indicate two glyco- peptidic [M + H] + ions at m/z 4239.38 and 4403.41 matching with Hex2(dHex)(Pen)HexNAc 2 -Asn138peptide- and Hex3(dHex)(Pen)HexNAc 2 -Asn138 peptide- struc- tures. Glycopeptide fraction T 2 MALDI spectrum displays three major ions at 4403.39, 4606.32 and 4809.47 corres- ponding, respectively, with Hex3(dHex)(Pen)HexNAc 2 - Asn138 peptide- , Hex3(dHex)(Pen)HexNAc 3 -Asn138 peptide- and Hex3(dHex)(Pen)HexNAc 4 -Asn138 peptide- glycopeptidic structures. Analysis of the Asn624 site The MALDI-TOF analysis of the different peptide fractions collected after RP-HPLC fractionation of tryptic hydroly- sates allowed us to detect the potential glycosylation site Asn624. A peptide fraction (H 3 ,M 3 and T 3 )elutedatthe same elution time (24 min) was shown to correspond to the unglycosylated peptide site NGSDCPDK(624–631). How- ever, we were not able to identify any glycosylated form of this peptide. The MALDI spectra obtained for the three lactoferrins were very similar and the spectrum obtained for Fig. 3. MALDI-TOF mass spectra of hLf, mLf and tLf Asn479-glycopeptides after PNGase treatment. N-glycopeptides from hLf and recombinant lactoferrins were deglyco- sylated with PNGase F and PNGase A, respectively. Peptides were then desalted on C 18 phase Sep-Pak cartridges and analysed by MALDI mass spectrometry using 2,5-dihy- droxybenzoic acid as matrix. (A) The hLf spectrum was recorded in positive ion linear mode; mLf (B) and tLf (C) spectra were recorded in positive ion reflective mode. The masses of Na adducts are indicated in smaller font. Ó FEBS 2003 N-glycosylation of maize and tobacco lactoferrin (Eur. J. Biochem. 270) 3239 mLfisshowninFig.4A.Twopeaksatm/z of 893.26 and 915.26 were assigned to [M + H] + and [M + Na] + ions of the unglycosylated Asn624 site, respectively. Peptide sequence was analysed by nano-electrospray, by selecting the dicharged ion at m/z 447.10 that generated by fragmen- tation five [M + H] + ions at m/z 778.31, 721.21, 634.22, 519.20 and 359.18 (Fig. 4B). The mass increments between these peaks, i.e. 57, 87, 115 and 160 Da, correspond exactly to the masses of glycine, serine, aspartic acid and carb- oxamidomethylated cysteine, respectively, amino acid sequence GSDC, that corresponds to the glycosylation site Asn624. Discussion The present paper reports for the first time the site-by-site N-glycosylation pattern of recombinant human lactoferrin expressed in two different plant expression systems: the endosperm of maize seeds, a monocotyledon expression system allowing full-scale commercial production, and tobacco leaves used as a model of a dicotyledon plant. Human Lf is a convenient model to analyse the details of the glycosylation potential of plant expression systems because data are available on the glycosylation of native Lf and of recombinant Lf produced in other systems including mammalian cells [24], lepidopteran cells [25], and transgenic mice [26]. N-glycosylation of milk derived human lacto- ferrin has been extensively studied, showing that hLf contains two N-acetyllactosamine-type N-glycans, more or less fucosylated and sialylated. Moreover, a third N-glycosylation site (Asn624) is located in the C-terminal part of the glycoprotein but is mostly unglycosylated [18,19]. Human lactoferrin is also an interesting model because it is a natural defence iron-binding protein that has been found to possess antibacterial, antifungal, antiviral, antineoplastic and anti-inflammatory activity and is considered as a novel therapeutic with broad spectrum potential [27]. The relative proportion of glycans, estimated from the MALDI-TOF spectra of both N-glycopeptides (Asn138 and Asn479) of natural and recombinant Lf, are summar- ized in Table 1. As observed in the natural Lf, Asn138 and Asn479 sites in the recombinant proteins are substituted by Fig. 4. Mass spectrometry analysis of mLf glycosylation site Asn624. (A) MALDI-TOF mass spectrum of mLf peptide fraction M 3 .The peptide fraction M 3 has been analysed by MALDI-TOF using 2,5-dihydroxybenzoic acid as matrix and spectrum was recorded in positive ion reflective mode. (B) Sequencing by nano-electrospray mass spectrometry. Peptides from fraction M 3 were dissolved in MeOH/ H 2 O and analysed with a QSTAR quadrupole time-of-flight mass spectrometer. The molecular ion at m/z 447.10 was selected and frag- mented to determine the amino acid sequence of the corresponding peptide. The deduced peptidic sequence is indicated on the spectrum. Table 1. Relative amounts of N-glycans detected onto hLf, mLf and tLf N-glycosylation sites Asn138 and Asn479. 3240 B. Samyn-Petit et al.(Eur. J. Biochem. 270) Ó FEBS 2003 mature N-glycans. Moreover, the third site is unglycosyl- ated. As in human [17], both N-glycosylation sites of tLf and mLf are N-glycosylated by similar structures. N-gly- cans found on both sites of mLf and tLf are mostly of the paucimannose-type, substituted by a bisecting b1,2-xylose and a1,3-fucose residue linked to the proximal GlcNAc (compound 7). However, the N-glycan structures of tLf contain a remarkably higher level of terminal GlcNAc than the corresponding structures isolated from mLf. Significant amounts of compounds 8 (GlcNAc 1 XylFucMan 3 Glc- NAc 2 )and9(GlcNAc 2 XylFucMan 3 GlcNAc 2 )wereiden- tified in the tLf spectrum, whereas these glycans were virtually absent in the spectrum of mLf (Fig. 2 and Table 1). These results clearly indicated that the first steps of N-glycosylation are similar in plants and humans and that the observed differences only arise from the specificity of the Golgi plant glycosyltransferases and from post- Golgi degradations of the matured plant N-glycans. In parallel, no complex-type N-glycans with Lewis a terminal sequence have been found either in mLf or in tLf. The lack of complex type structures with Lewis a determinants has also been reported for other monocotyledons and dicotyl- edons endogenous glycoproteins [13] and several studies of the N-glycosylation of tobacco recombinant glycoproteins have also shown the absence of such complex-type N-glycans [28–30]. The transfers of bisecting b1,2-xylose and a1,3-linked core fucose require the presence of at least one terminal GlcNAc [31]. As the N-glycans identified in both plants contain both epitopes, the higher proportion of GlcNAc- containing glycans in tLf mainly reflects differences in N-acetylglucosaminidase activities that govern the biosyn- thesis of paucimannose-type glycans, after maturation of the N-glycans in the Golgi compartment. These changes in glycosylation pattern could be also related to a difference in glycosylation in seeds and leaves, a different subcellular localization and/or to a different developmental stage of the plants. Indeed, Elbers et al.[28] have recently shown that the developmental stage of tobacco leaves influences the N-glycosylation of transgenic IgG, with a higher proportion of GlcNAc-containing glycans in older leaves compared to younger ones. The differences in glycosylation patterns of plant and mammalian cells can represent a limitation for the produc- tion of some recombinant therapeutic glycoproteins of mammalian origin in transgenic plants, and efforts are underway to obtain the in planta conversion of N-glycans to a human-compatible type. Recently, tobacco cells transformed with human b1,4-galactosyltransferase were used to evaluate the possibility to galactosylate foreign glycoproteins such as horseradish peroxidase [32] or mouse antibody [33]. For example, coexpression of human b1,4- galactosyltransferase and heavy and light chains of mouse antibody results in the synthesis in tobacco plants, of a recombinant antibody that exhibits 30% of galactosylated N-glycans [33]. Even if the terminal GlcNAc content in N-glycans of maize origin appears to be low and that outcrossing of transgenic maize could not be excluded, the industrial advantages of maize seeds as a production system for recombinant proteins, compared to tobacco leaves, such as the absence of toxic compounds, the possibility of low cost storage of biomass and the ease of extracting protein from grains [34] has led us to initiate the engineering of maize N-glycosylation. Acknowledgements This work was supported in part by the University of Sciences and Technologies of Lille, by a grant (Saut Technologique) by the French Research Ministry and a grant CIFRE of the French ANRT to B. Samyn-Petit. We thank our colleagues in Plant Production and the Pilot Unit of Meristem-Therapeutics with help in growing and extracting the lactoferrin plants. References 1. Doran, P.M. (2000) Foreign protein production in plant tissues cultures. Curr. Opin. Biotechnol. 11, 199–204. 2. Daniell, H., Streatfield, S.J. & Wycoff, K. (2001) Medical mole- cular farming: production of antibodies, biopharmaceuticals and edible vaccines in plants. Trends Plant Sci. 6, 219–226. 3. Tacket, C.O., Mason, H.S., Losonsky, G., Estes, M.K., Levine, M.M. & Arntzen, C.J. (2000) Human immune responses to a novel Norwalk virus vaccine delivered in transgenic potatoes. J. Infect. Dis. 182, 302–305. 4. Lo ¨ nnerdal, B. (2002) Expression of Human Milk Proteins in Plants. J. Am. Coll. Nutr. 21, 218S–221S. 5. Merle, C., Perret, S., Lacour, T., Jonval, V., Hudaverdian, S., Garrone, R., Ruggiero, F. & Theisen, M. (2002) Hydroxylated human homotrimeric collagen I in Agrobacterium tumefaciens– mediated transient expression and in transgenic tobacco plants. FEBS Lett. 515, 114–118. 6. Staub, J.M., Garcia, B., Graves, J., Hajdukiewicz, P.T., Hunter, P., Nehra, N., Paradkar, V., Schlittler, M., Carroll, J.A., Spatola, L., Ward, D.Ye, G. & Russell, D.A. (2000) High-yield production of a human therapeutic protein in tobacco chloroplasts. Nat. Biotechnol. 18, 333–338. 7. Peeters, K. De Wilde, C. De Jaeger, G. Angenon, G. & Depicker, A. (2001) Production of antibodies and antibody fragments in plants. Vaccine 19, 2756–2761. 8. Sto ¨ ger, E. Vaquero, C. Torres, E. Sack, M. Nicholson, L. Drossard, J. Williams, S. Keen, Perrin, D. & Christou, Y. (2000) Cereal crops as viable production and storage systems for pharmaceutical scFv antibodies. Plant Mol. Biol. 42, 583–590. 9. Lerouge, P., Cabanes-Macheteau, M., Rayon, C., Fitchette-Laine ´ , A.C., Gomord, V. & Faye, L. (1998) N-glycoprotein biosynthesis in plants: recent developments and future trends. Plant. Mol. Biol. 38, 31–48. 10. Jenkis, N., Parekh, R.B. & James, D.C. (1996) Getting the gly- cosylation right. implications for the biotechnology industry. Nat. Biotechnol. 14, 975–981. 11. Staudacher, E., Altmann, F., Wilson, I.B. & Marz, L. (1999) Fucose in N-glycans: from plant to man. Biochim. Biophys. Acta 1473, 216–236. 12. Fitchette-Laine, A.C., Gomord, V., Cabanes-Macheteau, M., Michalski, J.C., Saint Macary, M., Foucher, B., Cavelier, B., Hawes, C., Lerouge, P. & Faye, L. (1997) N-glycans harboring the Lewis a epitope are expressed at the surface of plant cells. Plant J. 12, 1411–1417. 13. Wilson, I.B., Zeleny, R., Kolarich, D., Staudacher, E., Stroop, C.J., Kamerling, J.P. & Altmann, F. (2001) Analysis of Asn-linked glycans from vegeTable foodstuffs. widespread occurrence of Lewis a, core a1,3-linked fucose and xylose substitutions. Glyco- biology 11, 261–274. 14. Yang, B.Y., Gray, J.S.S. & Montgomery, R. (1996) The glycans of horseradish peroxydase. Carbohydr. Res. 287, 203–212. Ó FEBS 2003 N-glycosylation of maize and tobacco lactoferrin (Eur. J. Biochem. 270) 3241 15. Makino, Y., Shimazaki, A., Omichi, K., Odani, S. & Hase, S. (2000) Processing pathway deduced from the structures of N-glycans in Carica papaya. J. Biochem. 127, 1121–1126. 16. Altmann, F., Schwihla, H., Staudacher, E., Glo ¨ ss, J. & Ma ¨ rz, L. (1995) Insect cells contain an unusual, membrane-bound b-N-acetylglucosaminidase probably involved in the processing of protein N-glycans. J. Biol. Chem. 270, 17344–17349. 17. Samyn-Petit, B., Gruber, V., Flahaut, C., Wajda-Dubos, J.P., Farrer, S., Pons, A., Desmaizieres, G., Slomianny, M.C., Theisen, M. & Delannoy, P. (2001) N-glycosylation potential of maize: the humanlactoferrinusedasamodel.Glycoconjugate J. 18, 519–527. 18. Spik, G., Strecker, G., Fournet, B., Bouquelet, S. & Montreuil, J. (1982) Primary structure of the glycans from human lacto- transferrin. Eur. J. Biochem. 121, 413–419. 19. Van Berkel, P.H.C., Van Veen, H.A., Geerts, M.E.J., De Boer, H.A. & Nuijens, J.H. (1996) Heterogeneity in utilization of N-glycosylation sites Asn 624 and Asn 138 in human lactoferrin: a study with glycosylation-site mutants. Biochem. J. 319, 117–122. 20. Brock, J.H. (2002) The physiology of lactoferrin. Biochem. Cell. Biol. 80, 1–6. 21. Salmon, V., Legrand, D., Slomianny, M C., El Yazidi, I., Spik, G., Gruber, V., Bournat, P., Olagnier, B., Mison, D., Theisen, M. &Me ´ rot, B. (1998) Production of human lactoferrin in transgenic tobacco plants. Protein Express. Purif. 13, 127–135. 22. Flahaut, C., Capon, C., Balduyck, M., Ricart, G., Sautiere, P. & Mizon, J. (1998) Glycosylation pattern of human inter-a-inhibitor heavy chains. Biochem. J. 333, 749–756. 23. Qin, J., Fenyo, D., Zhao, Y., Hall, W.W., Chao, D.M., Wilson, C.J., Young, R.A. & Chait, B.T. (1997) A strategy for rapid, high- confidence protein identification. Anal. Chem. 69, 3995–4001. 24. Legrand, D., Salmon, V., Coddeville, B., Benaissa, M., Plancke, Y. & Spik, G. (1995) Structural determination of two N-linked glycans isolated from recombinant human lactoferrin expressed in BHK cells. FEBS Lett. 365, 57–60. 25. Salmon, V., Legrand, D., Georges, B., Slomianny, M C., Coddeville, B. & Spik, G. (1997) Characterization of human lactoferrin produced in the baculovirus expression system. Protein Expr. Purif. 9, 203–210. 26. Nuijens, J.H., van Berkel, P.H., Geerts, M.E., Hartevelt, P.P., de Boer, H.A., van Veen, H.A. & Pieper, F.R. (1997) Characteriza- tion of recombinant human lactoferrin secreted in milk of trans- genic mice. J. Biol. Chem. 272, 8802–8807. 27. Weinberg, E.D. (2001) Human lactoferrin: a novel therapeutic with broad spectrum potential. J. Pharm. Pharmacol. 53, 1303– 1310. 28. Elbers, I.J., Stoopen, G.M., Bakker, H., Stevens, L.H., Bardor, M., Molthoff, J.W., Jordi, W.J., Bosch, D. & Lommen, A. (2001) Influence of growth conditions and developmental stage on N-glycan heterogeneity of transgenic immunoglobulin G and endogenous proteins in tobacco leaves. Plant Physiol. 126, 1314– 1322. 29. Cabanes-Macheteau, M., Fitchette-Laine, A.C., Loutelier-Bour- his,C.,Lange,C.,Vine,N.D.,Ma,J.K.,Lerouge,P.&Faye,L. (1999) N-Glycosylation of a mouse IgG expressed in transgenic tobacco plants. Glycobiology 9, 365–372. 30. Dirnberger, D., Steinkellner, H., Abdennebi, L., Remy, J.J. & van de Wiel, D. (2001) Secretion of biologically active glycoforms of bovine follicle stimulating hormone in plants. Eur. J. Biochem. 268, 4570–4579. 31. Strasser, R., Mucha, J., Mach, L., Altmann, F., Wilson, I.B., Glossl, J. & Steinkellner, H. (2000) Molecular cloning and func- tional expression of b1,2-xylosyltransferase cDNA from Arabid- opsis thaliana. FEBS Lett. 472, 105–108. 32. Fujiyama, K., Palacpac, N.Q., Sakai, H., Kimura, Y., Shinmyo, A., Yoshida, T. & Seki, T. (2001) In vivo conversion of a glycan to human compatible type by transformed tobacco cells. Biochem. Biophys. Res. Commun. 289, 553–557. 33. Bakker, H., Bardor, M., Molthoff, J.W., Gomord, V., Elbers, I., Stevens,L.H.,Jordi,W.,Lommen,A.,Faye,L.,Lerouge,P.& Bosch, D. (2001) Galactose-extended glycans of antibodies produced by transgenic plants. Proc. Natl Acad. Sci. USA 98, 2899–2904. 34. Mison, D. & Curling, J. (2000) The industrial production cost of recombinant therapeutic proteins in transgenic corn. Biopharma- cology 13, 48–54. 3242 B. Samyn-Petit et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Comparative analysis of the site-specific N-glycosylation of human lactoferrin produced in maize and tobacco plants Be ´ ne ´ dicte Samyn-Petit 1 , Jean-Pierre. France; 2 Meristem Therapeutics, Clermont-Ferrand, France We have compared the site-by-site N-glycosylation status of human lactoferrin (Lf) produced in maize, a monocotyle- don, and in tobacco, used. species [14,15]. These findings indicate that plants do not exhibit the same potential of N-glycosylation, according to the level of expression of several key enzymes involved in the initiation (i.e.