Glycosylation is one of the most important post-translational modifications. Proteins can be decorated with glycans, which are linked either with asparagine residues via an N
-glycosidic linkage (N
-glycans) or with serine or threonine residues via an O
-glycosidic linkage (O
]. The majority of N
- and O
-glycans consists of six different monosaccharides, namely mannose, galactose, N
-acetylglucosamine (GlcNAc), N
-acetylgalactosamine (GalNAc), fucose and sialic acids. Xylose is a rare monosaccharide component of O
], and it is part of the linker between glycosaminoglycans and proteins [3
]. Furthermore, pentose, together with glucuronic acid, is one of the two monosaccharide components of a heteropolymer linked to α-dystroglycan in muscle cells [4
], where it mediates its interaction with the extracellular matrix component laminin. In N
-glycans xylose is only found in plants and helminth [5
]. There it is usually β1,2-linked to the central mannose residue of the N
For mammals, xylosylated N
-glycans are alien structures. Parenteral introduction of glycoproteins carrying this kind of N
-glycans into mammals therefore leads to strong immune reactions [5
]. For the production of recombinant therapeutic glycoproteins in plant cells immunogenicity of the N
-glycans is therefore an issue [7
]. To date, plant cell lines used for recombinant glycoprotein productions are engineered by knock-out of the respective β1,2-xylosyltransferase [8
]. However, the metabolic pathway for the synthesis of xylose and its activated nucleotide sugar UDP-xylose, respectively, is present also in mammalian cells [10
]. UDP-xylose is synthesized from UDP-glucose in two steps; first by oxidation of the C-6 atom, resulting in UDP-glucuronic acid. The same C-6 atom is then decarboxylated, and the pentose xylose is generated, which is bound to UDP in the furanose form. UDP-xylose is then transported to the Golgi apparatus, where it serves as substrate for xylosyltransferases.
Altered glycans may increase or modulate the immunogenicity with potential benefits for recombinant glycoprotein vaccines [11
]. Introduction of core β1,2-xylosylation is one promising approach [13
]. In this study we have used a protein of the human respiratory syncytial virus (RSV) as a model vaccine. RSV is the main cause of lower respiratory tract infections in infants and the elderly, and also affects high-risk adults [14
]. Although several strategies, including use of vaccines and therapeutic antibodies, have been tested for treatment of RSV infections in the last few decades, no highly efficient prevention therapy is presently available [16
]. RSV vaccines based on recombinant proteins could benefit from the two major surface proteins RSV-F (fusion protein) and RSV-G (glycoprotein). In this study we used a truncated form of RSV-F, which only contains the extracellular part. RSV-F has five N
-glycosylation sites, whereby two of them are released together with a 27 amino acid peptide by intracellular cleavage by the Golgi protease furin [17
]. Mature RSV-F therefore consists of two disulfide-bridged polypeptide chains with N
-glycans at the positions Asn-23, Asn-66 and Asn-497 [17
]. The recombinant RSV-F protein was produced in glycoengineered CHO DG44 cells, co-expressing the β1,2-xylosyltransferase (XylT) from Nicotiana tabacum
. We were able to show that RSV-F from glycoengineered cells contains xylosylated N
-glycans, and that this recombinant vaccine, compared to RSV-F with non-xylosylated N
-glycans, displays stronger immunogenicity in a human artificial lymph-node (HuALN) model.
2. Materials and Methods
CDC4 medium with 4.5 g/L glucose was obtained from ProBioGen AG, Berlin, Germany, Dulbecco’s phosphate buffered saline (DPBS) from PAN-Biotech GmbH (Aidenbach, Germany), adenovirus expression medium (AEM) from Life Technologies GmbH (Darmstadt, Germany), fetal calf serum (FCS) superior from Merck Millipore (Darmstadt, Germany). Unless otherwise stated, all chemicals were purchased from Carl Roth GmbH + Co. KG (Karlsruhe, Germany) or Sigma-Aldrich GmbH (Taufkirchen, Germany).
2.2. Cell Culture
Standard cultivation of all cells was performed in CDC4 medium supplemented with 6 mM l-glutamine, 50 ng/mL IGF and 100 µg/mL penicillin/streptomycin (100 U/mL) within a 8% CO2 atmosphere at 36.5 °C. Stably RSV-F-expressing CHO-DG44 cells and additional XylT expressing CHO-DG44 cells were cultivated under serum-free conditions in CDC4 medium supplemented with 6 mM l-glutamine, 1% penicillin/streptomycin (100 U/mL, 100 µg/mL) and insulin-like growth factor (50 ng/mL).
2.3. Construction of Expression Vectors
The RSV-F cDNA was constructed corresponding to Ternette et al. [20
], with the exception that the signal sequence was replaced by the mellitin signal sequence (according to Acc. No. P01501; MKFLVNVALVFMVVYISYIY). It is followed by the ectodomain of the RSV-F protein (amino acids 26 to 530; according to Acc. No. EF566942), a Factor Xa cleavage site (IEGR), and a GSGS linker fused to a 6xHis-tag (HHHHHH). The gene was codon optimized for Cricetulus griseus
and synthesized with flanking EcoRI and BamHI restriction by Gene Art (Regensburg, Germany). The gene was cloned into the EcoRV restriction site of the vector pBGGPEX1 (ProBioGen AG, Berlin, Germany) by EcoRI/BamHI digestion, and followed by DNA polymerase Klenow (Roche, Mannheim, Germany) treatment resulting in the pBGGPEX1-RSV-F vector.
The gene of Nicotiana tabacum XylT (Acc. No. AJ627182) was synthetized by Gene Art (Regensburg, Germany) with a codon-optimized sequence for Cricetulus griseus, flanked by BglII/EcoRI restriction sites. The gene was cloned into the expression vector EF2flag neo (ProBioGen) using the BglII/EcoRI restriction sites resulting in the vector EF2flag XylT. Plasmids were prepared by the QIAprep® Spin Miniprep kit (Qiagen, Hilden, Germany) and EndoFree® Plasmid Maxi kit (Qiagen).
2.4. Transfection of CHO Cells
The CHO-DG44 cells were transfected with pBGGPEX1-RSV-F vector by electroporation using the Neon® Transfection system (Thermo Fisher, Schwerte, Germany). Selection of transfected has been carried out for 17 days in serum-free C8862 medium supplemented with puromycin and methotrextate (MTX). The resulting CHO RSV-F cells were additionally transfected with the EF2flag XylT vector by lipofection using the Freestyle Max Reagent (Thermo Fisher) in Optipro medium (Thermo Fisher). CHO-F-XylT clones were selected with Neomycin (G418). The expression of the RSV-F protein has been verified by western blot analysis using a Penta-His HRP antibody (Qiagen, 1:2000). The expression of the XylT was detected on the transcription level by RT-PCR with specific primers (forward 5′-GAGAACCACCACGACAAC-3′, reverse 5′-CTGTTCCTCGTTGGACAG-3′). The resulting PCR product of 1077 bp was visualized by agarose gel electrophoresis.
2.5. Protein Production
Serum free fed-batch culture of transfected CHO cells producing modified RSV-F protein was carried out in 50 mL bioreactor tubes containing CD-C4 growth medium supplemented with l-glutamine. Bioreactor tubes were inoculated at a starting cell density of 4 × 105 cells/mL and incubated at 146 rpm, 37 °C, 8% CO2. Performance of the fed-batch cultures was monitored for 14 days and supernatant containing the RSV-F protein was harvested after declining of cell viability down to 75%. The supernatant was analyzed for RSV-F production by western blotting.
2.6. Purification of Proteins
RSV-F protein purification was performed by a one-step purification using a 5 mL Ni-NTA cartridge (Machery & Nagel, Düren, Germany). 250 mL of the sterile filtered supernatant from batch production was twice concentrated using an Amicon Filter device (300 mL) and diluted in binding/washing buffer (20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole, pH 7). After equilibration of the column with 5 volumes binding buffer the sample was loaded and the column washed with 10 volumes of binding/washing buffer. Protein was eluted in one step by 5 mL elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7), dialysed against PBS, sterile filtered and stored at 4 °C. Protein concentration was quantified by BCA protein assay for total quantification and RSV-F proteins were analyzed by western blotting.
2.7. High-pH Anion-Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) Monosaccharide Analysis
20 µg of purified proteins was hydrolyzed in 2 M trifluoroacetic acid (TFA) for 4 h at 100 °C. A blank sample was used as negative control. As internal standards 200 pmol 2-deoxy-D-ribose (DR), d-fructose (Fruc) and d-melibiose (Mel) (each from Sigma-Aldrich, Taufkirchen, Germany) were used. As external standards a set of monosaccharides, including 100 pmol L-fucose (Fuc), d-arabinose (Ara), d-galactosamine (GalN), d-galactose (Gal), d-glucosamine (GlcN), d-glucose (Glc), d-Xylose (Xyl) and d-Mannose (Man), was used and run prior to the protein samples. HPAEC-PAD was performed on an ICS-3000 Ion Chromatography System (Thermo Fisher) using a Dionex CarboPac® PA200 column. Monosaccharides were separated by isocratic 2.25 mM NaOH elution while post-column addition of 200 mM NaOH provided the conditions for pulsed amperometric detection.
2.8. Release and Separation of N-Linked Glycans
Tryptic digestion was performed twice (for 4 h at 37 °C and overnight, respectively) using 2.5 µg trypsin (Sigma-Aldrich) per 30 µg of glycoprotein. After trypsin inactivation, samples were treated with 0.5 U of N
-glycosidase F from Flavobacterium meningospeticum (Roche Diagnostics, Mannheim, Germany) and incubated at 37 °C. After 4 h additional enzyme was added, the sample was incubated overnight followed by inactivation (5 min, 95 °C). N
-Glycans were separated from the peptide fraction using reversed-phase C18 cartridges and a subsequent desalting step on graphitized cartridges (Grace Davison Discovery Sciences, Worms, Germany) as described before [21
]. Purified N
-glycans were lyophilized and stored at −20 °C.
2.9. Permethylation and Matrix-Assisted Laser Desorption/Ionization Time-of-Flight (MALDI-TOF) Mass Spectrometry
Prior to mass spectrometric (MS) analysis, N
-glycan samples were permethylated in order to stabilize sialic acids and to improve the efficiency of positive ion formation [22
]. The derivatization procedure followed standard protocols of the solid sodium hydroxide technique [22
] with minor modifications. Incubations were carried out under continuous shaking at room temperature. The iodomethane reaction was stopped by the addition of chloroform and subsequent washing steps with water until achieving a neutral pH of the aqueous phase.
For MALDI-TOF-MS analysis dried permethylated N
-glycans were dissolved in 75% (v/v) acetonitrile in water and mixed with super-dihydroxybenzoic acid (sDHB) matrix (Sigma-Aldrich). Recording of mass spectra on an Ultraflex III MALDI-TOF/TOF spectrometer (Bruker Daltonik, Bremen, Germany) and subsequent data processing was realized as reported previously [24
]. Structures were assigned to related peaks according to the Glycoworkbench 2.0 database, or constructed by the same software if not available. Fragmentation analysis was performed by the integrated Lift method of the mass spectrometer. Fragment sizes were compared to the theoretically determined size of the glycans using Glycoworkbench 2.0. Schematic representation of glycan structures are according to the symbol nomenclature of the Consortium for Functional Glycomics [25
]: green circle, mannose; yellow circle, galactose; blue square, GlcNAc; yellow square, GalNAc; red triangle, fucose; purple diamond, N
-acetylneuraminic acid; star, xylose.
2.10. Cytokine and Gene Expression Analysis
Stimulation experiments were performed in a HuALN (ProBioBen AG; [26
]) In brief, RSV-F and RSV-F Xyl+ were cultivated in the presence of CD14(-) cells and mature dendritic cells prior to addition of PBMC and matrix. HuALN reactors were run for 28 days and re-stimulation took place at days 7, 14 and 21. Cell culture supernatants were taken daily and analyzed with a bead-based multiplexed immune assay (Luminex®
technology, Thermo Fisher). A custom Bio-Plex®
Express Aassay (Bio-Rad, München, Germany) was used to quantify the six analytes IL-2, IL-4, IL-10, GM-CSF, IFN-γ and TNF-α. The assay was performed in duplicates as described by the manufacturer, but conducted automated on a pipetting robot (Freedom Evo 200; Tecan, Crailsheim, Germany). Quantification of samples based on a logistic 5-point regression method using standard curves with 8 point 4-fold dilution series with analyte-specific concentration ranges. Error limits of data more than twice as much as the detection limit were below 20%.
To investigate the first response of the immunologic reaction comprehensively, gene expression was analyzed with a Human Immunology v2 nCounter®
Gene Expression assay (NanoString®
Technologies, Seattle, WA, USA) covering 579 immunology related genes. Therefore, PBMC of healthy donors were stimulated for 48 h with antigen-specific mDC and the according proteins, or were left untreated (negative control). Total RNAs of cells were extracted with a High Pure RNA Isolation Kit (Roche Diagnostics). Concentration and purity of total RNAs were controlled with a Nanodrop 1000 (Thermo Fisher) using the spectral absorption quotients 260/230 and 260/280. RNA Quality Numbers (RQN) were determined on a Fragment Analyzer™
(Advanced Analytical Technologies, Heidelberg, Germany) with a DNF-472 High Sensitivity RNA Analysis Kit, using extended runtimes of 60 min per sample to be able to detect any remaining genomic DNA contaminations. The nCounter technology [28
] is a multiplexed method that quantifies mRNA on single molecule level by using fluorescent barcoding probes and is described in detail in [29
]. NanoString experiments were performed according to the manufacturer protocol. Due to slight fragmentation of RNA increased amounts of input material (150 ng) were used in the experiment. RNAs were quantified with Qubit®
RNA HS AssayKit (Thermo Fisher), immediately before the nCounter experiment was conducted. The nSolver™
Analysis Software 3.0 (NanoString®
) was used to perform data handling, including automated background subtraction, spike-in-control normalization and reference gene normalization. Furthermore, datasets from duplicates were grouped and fold change estimates were calculated by building ratios with the unstimulated negative control. Since analysis could be performed in duplicates only, somewhat higher p-values were accepted to keep data sets for broad immunologic parts descriptive.
In this study we successfully glyco-engineered CHO cells for the production of xylosylated N
-glycans. Nicotiana tabacum
XylT was functionally expressed in a mammalian cell system, where it was localized in the correct Golgi compartment, had access to endogenous UDP-xylose and could transfer xylose to the N
-glycan core of a recombinant, co-expressed viral glycoprotein. We were able to show that xylosylated N
-glycans had strong adjuvant effects in a well-established human organoid immune model, indicating the usefulness of this approach for in vivo models and for vaccination strategies in general. Furthermore, glyco-engineering of CHO cells by the methodology presented here could be extended to other glycan structures with antigenic and therefore adjuvant potency—for example, α1,3-linked core-fucose natively occurring in plant and insect N
], or the non-human sialic acid N
-glycolylneuraminic acid [35