Toward stable genetic engineering of human O-glycosylation in plants

Abstract

Glycosylation is the most abundant and complex posttranslational modification to be considered for recombinant production of therapeutic proteins. Mucin-type (N-acetylgalactosamine [GalNAc]-type) O-glycosylation is found in eumetazoan cells but absent in plants and yeast, making these cell types an obvious choice for de novo engineering of this O-glycosylation pathway. We previously showed that transient implementation of O-glycosylation capacity in plants requires introduction of the synthesis of the donor substrate UDP-GalNAc and one or more polypeptide GalNAc-transferases for incorporating GalNAc residues into proteins. Here, we have stably engineered O-glycosylation capacity in two plant cell systems, soil-grown Arabidopsis (Arabidopsis thaliana) and tobacco (Nicotiana tabacum) Bright Yellow-2 suspension culture cells. Efficient GalNAc O-glycosylation of two stably coexpressed substrate O-glycoproteins was obtained, but a high degree of proline hydroxylation and hydroxyproline-linked arabinosides, on a mucin (MUC1)-derived substrate, was also observed. Addition of the prolyl 4-hydroxylase inhibitor 2,2-dipyridyl, however, effectively suppressed proline hydroxylation and arabinosylation of MUC1 in Bright Yellow-2 cells. In summary, stably engineered mammalian type O-glycosylation was established in transgenic plants, demonstrating that plants may serve as host cells for the production of recombinant O-glycoproteins. However, the present stable implementation further strengthens the notion that elimination of endogenous posttranslational modifications may be needed for the production of protein therapeutics.

Figure 1.

Construct designs for engineering O-glycosylation. A, Depiction of O-glycosylation machinery construct designs. Designs included Flag-tagged cytoplasm-targeted P. aeruginosa C4-epimerase WbpP (CytoEpi) and Golgi-targeted polypeptide GalNAc-T2 (T2) expressed as a single polycistronic protein interspaced by the 2A self-cleaving sequence (cleavage site indicated by the arrow; T2-2A-CytoEpi) or as separate transcriptional units (CytoEpi-T2). B, Depiction of the O-glycosylation reporter constructs used: MUC1-3.5TR (MUC1) with and without a C-terminal fusion to YFP (MUC1-YFP); MUC1 embedded into GFP [GF(MUC1)P]; Full-coding IFNα2B cytokine with a C-terminal T7 and SP₁₀ glycomodule (IFN-SP₁₀). All reporter constructs included N- or C-terminal His₆ tags and N-terminal signal peptide sequences (SS). C, Depiction of a single combined construct with reporter MUC1-YFP and the 2A-linked O-glycosylation machinery under the control of two promoters (MUC1-YFP-T2-2A-CytoEpi).

Figure 2.

Implementation of O-glycosylation in Arabidopsis and tobacco BY-2 suspension cells. A, SDS-PAGE western-blot analysis of four Arabidopsis lines expressing either MUC1 (lanes 1 and 2) or MUC1-YFP (lanes 4 and 5) alone. The contrast of the image was adjusted to visualize very weak bands for MUC1 (lanes 1 and 2). Lane 3 shows results for the wild type (WT). B, SDS-PAGE western-blot analysis of two tobacco BY-2 suspension cell lines expressing MUC1-YFP alone (lane 1) and MUC1-YFP together with the 2A-linked O-glycosylation machinery T2-2A-CytoEpi (lane 2). The absence (−) or presence (+) of O-glycosylation machinery is indicated above the lanes. C, An Arabidopsis line transgenic for T7-tagged full-coding secreted IFNα2B expressed alone (−) or coexpressed with the O-glycosylation machinery CytoEpi-T2 (+). The glycosylation of His tag-purified IFNα2B was detected by VVA lectin-blot analysis. Total protein extracts from transgenic Arabidopsis leaves or BY-2 cell callus were loaded and blots reacted with MUC1-specific MAbs 5E10 and 5E5, where 5E5 is specific for GalNAc-glycosylated MUC1 (Tn-MUC1) and does not react with nonglycosylated MUC1. Approximately 30 μg of total protein was loaded in each lane.

Figure 3.

MUC1-YFP is expressed and glycosylated in BY-2 cells, but MUC1 is degraded in the medium. A, UV image of fractions (E1–E8) from HIC of MUC1-YFP secreted from BY-2 cells coexpressing T2-2A-CytoEpi. B, SDS-PAGE Coomassie blue-stained analysis of fractions. C, Western blot with anti-GFP antibody. D, Western blot with anti-MUC1 (5E10). Ten microliters of 2-mL eluate fractions was applied in each lane. [See online article for color version of this figure.]

Figure 4.

O-Glycosylation and embedding in GFP stabilize MUC1 in BY-2 cell culture. A, Construct design for embedding MUC1-2.5TR into GFP [GF(MUC1)P], and corresponding barrel structure showing the loop into which MUC1-2.5 TR flanked with His₈ and C-Myc tags was inserted after Asp-196 in the loop between the β-strands (blue) located opposite the N and C termini. This figure was adapted from Kobayashi et al. (2008). B, Fluorescence microscopy of a stable BY-2 transgenic line coexpressing GF(MUC1)P with the O-glycosylation machinery CytoEpi-T2 (left panel), and SDS-PAGE Coomassie blue staining of nickel chromatography-purified secreted GalNAc-glycosylated GF(MUC1)P [Tn-GF(MUC1)P] of this line (right panel). C, Analysis of the degradation of MUC1-YFP and GF(MUC1)P in BY-2 cell culture medium by SDS-PAGE western blotting: (1) MUC1-YFP (lanes 1–10) and GalNAc-glycosylated MUC1-YFP (Tn-MUC1-YFP; lanes 11–20) transiently produced in leaves of N. benthamiana plants; (2) intracellularly embedded GF(MUC1)P (lanes 21–30) purified from a stably transformed BY-2 cell line expressing GF(MUC1)P alone; (3) intracellularly (lanes 31–40) and extracellularly (lanes 41–50) embedded (Tn)-GF(MUC1)P purified from a transgenic BY-2 cell line additionally coexpressing CytoEpi-T2. The isolated proteins were added to 7-d-old unboiled (−) or boiled (+) wild-type BY-2 medium fractions, which were further incubated for up to 24 h under the same conditions. Approximately 5 μg of purified proteins was added to 1 mL of BY-2 medium fractions, of which approximately 15 μL was loaded and blots probed with anti-MUC1 MAb 5E10. The corresponding Tn-MUC1-specific western blot using MAb 5E5 is presented in Supplemental Figure S3. [See online article for color version of this figure.]

Figure 5.

Inhibition of Pro hydroxylation of MUC1 in BY-2 cells. A, MALDI-TOF-MS analysis of endoproteinase Asp-released MUC1 tandem repeats (DTRPAPGSTAPPAHGVTSAP; indicated by arrows) from Tn-GF(MUC1)P expressed in BY-2 cells coexpressing CytoEpi-T2. The analysis revealed the presence of 3 mol of attached GalNAc (i.e. complete GalNAc-T2-mediated GalNAc occupancy of sites), with varying Pro hydroxylations of up to 4 mol per tandem repeat. Hydroxylated MUC1 peptides were further modified by the attachment of mainly three, and less frequently of six, Ara residues. B, MALDI-TOF-MS analysis of MUC1 tandem repeats (indicated by arrows) of GF(MUC1)P coexpressed with CytoEpi-T2 in BY-2 cells grown in the presence of a P4H inhibitor. Growth medium was exchanged at day 7 with fresh medium including 200 µm 2,2-DP, followed by incubation in similar growth conditions. [See online article for color version of this figure.]

Figure 6.

MUC1-YFP and GF(MUC1)P expressed with or without O-glycosylation machinery in Arabidopsis. A, SDS-PAGE analysis of a transgenic Arabidopsis line expressing a combined construct (MUC1-YFP-T2-2A-CytoEpi) comprising both the reporter MUC1-YFP and the O-glycosylation machinery T2-2A-CytoEpi. Shown are MUC1-specific and Tn-MUC1-specific western-blot analysis of total leaf extracts (1) and MUC1-YFP purified from leaves (2) and roots (3) by HIC as visualized by Coomassie blue-stained SDS-PAGE gels and MUC1-specific MAb 5E10 western blots of the eluates. B, His tag purification of embedded GF(MUC1)P coexpressed with the O-glycosylation machinery CytoEpi-T2 as visualized by Coomassie blue-stained SDS-PAGE gels and MUC1-specific MAb 5E10 western blot. C and D, MALDI-TOF-MS analysis of MUC1 tandem repeats (DTRPAPGSTAPPAHGVTSAP; indicated by arrows) from His tag-purified Asp-N-digested MUC1-YFP (C) and GF(MUC1)P (D) coexpressed with the O-glycosylation machineries T2-2A-CytoEpi and CytoEpi-T2, respectively. [See online article for color version of this figure.]