Assay and heterologous expression in Pichia pastoris of plant cell wall type-II membrane anchored glycosyltransferases
- First Online:
- Cite this article as:
- Petersen, B.L., Egelund, J., Damager, I. et al. Glycoconj J (2009) 26: 1235. doi:10.1007/s10719-009-9242-0
- 485 Views
Two Arabidopsis xylosyltransferases, designated RGXT1 and RGXT2, were recently expressed in Baculovirus transfected insect cells and by use of the free sugar assay shown to catalyse transfer of D-xylose from UDP-α-D-xylose to L-fucose and derivatives hereof. We have now examined expression of RGXT1 and RGXT2 in Pichia pastoris and compared the two expression systems. Pichia transformants, expressing soluble, secreted forms of RGXT1 and RGXT2 with an N- or C-terminal Flag-tag, accumulated recombinant, hyper-glycosylated proteins at levels between 6 and 16 mg protein • L-1 in the media fractions. When incubated with 0.5 M L-fucose and UDP-D-xylose all four RGXT1 and RGXT2 variants catalyzed transfer of D-xylose onto L-fucose with estimated turnover numbers between 0.15 and 0.3 sec-1, thus demonstrating that a free C-terminus is not required for activity. N- and O-glycanase treatment resulted in deglycosylation of all four proteins, and this caused a loss of xylosyltransferase activity for the C-terminally but not the N-terminally Flag-tagged proteins. The RGXT1 and RGXT2 proteins displayed an absolute requirement for Mn2+ and were active over a broad pH range. Simple dialysis of media fractions or purification on phenyl Sepharose columns increased enzyme activities 2-8 fold enabling direct verification of the product formed in crude assay mixtures using electrospray ionization mass spectrometry. Pichia expressed and dialysed RGXT variants yielded activities within the range 0.011 to 0.013 U (1 U = 1 nmol conversion of substrate • min-1 • µl medium-1) similar to those of RGXT1 and RGXT2 expressed in Baculovirus transfected insect Sf9 cells. In summary, the data presented suggest that Pichia is an attractive host candidate for expression of plant glycosyltransferases.
KeywordsPlant cell wallGlycosyltransferaseFree sugar assayProtein expressionPichia pastoris
Carbohydrate Active EnZyme
counts per minute
electrospray ionization mass spectrometry
N-terminally Flag-tagged soluble Rhamnogalacturonan Xylosyltransferase 1, or 2 or 3
Rhamnogalacturonan Xylosyltransferase 1 or 2 soluble with a C-terminal Flag-Tag
trans membrane spanning domain
A main challenge in the post genomic era is the assignment of function to open reading frames with unknown function. The plant cell wall (CW) consists of numerous complex carbohydrate polymers and glycoproteins. Although the sugar composition of the plant CW components is largely resolved, less is known with respect to the corresponding biosynthetic machinery. The Carbohydrate Active EnZyme (CAZy) database [1–3], (http://afmb.cnrsmrs.fr/CAZY/index.html) currently lists 448 sequences for Arabidopsis thaliana as proven or putative glycosyltransferases (GTs). CAZy classifies GTs into 91 families, 40 of which are represented in Arabidopsis. So far, thirteen non-cellulosic/callosic CW GT activities have been successfully heterologously expressed and biochemically characterized (Cf. reviews by e.g. [4–7]). The GTs responsible for the synthesis of cellulose and callose are multi membrane spanning glucan synthases located at the plasma membrane. The cellulose synthases belong to family GT2, which also comprises the cellulose synthase like genes, Csl. Four genes from this group have been shown by heterologous expression and biochemical assay to encode β-1,4-mannan synthases [8, 9]. One gene product was shown to possess β-1,4-glucan synthase activity in raw cell extracts  and two barley CslF genes, were shown to confer production of mixed linkage (1-3;1-4)-β-D-glucan polymers when stably expressed in A. thaliana .
The GTs responsible for the synthesis of the CW polymers hemicellulose and pectin, and of the various CW-glycoproteins are thought primarily to be Golgi localized, where the majority of these GTs adopt the type II structure comprising a short N-terminal cytoplasmic tail followed by a single trans membrane spanning domain (TMD), and facing the lumen a stem region of varying length, followed by the large C-terminal globular domain, possessing the catalytic activity.
Heterologously expressed non-cellulosic/callosic plant CW biosynthetic GTs with proven biochemical activity
CAZY GT fam
Portion of GT expressed
GalT of fenugreek
FL and S (secreted)
AtXT1 & -2
Pichia pastoris, Drosophila S2 and Baculo virus Sf21 cells
ManS of Gua and AtCslA2,7,9
soybean somatic embryos and Drosophila (S2) cells
mixed linkage glucan syntase OsCSLF2 and OsCSLF4
Mixed linkage glucan
AtRGXT1, -2 & -3
Baculo virus Sf9 cells, Pichia pastoris
XyGS of Tropaeolum majus and AtCSLC4
Expression systems that rely on higher plant cells may overcome some of these problems. The GT can be targeted in its full length, membrane-anchored form to the Golgi vesicles. This approach largely eliminates problems with incorrect folding and inadequate posttranslational modifications, and the assay may rely on required but unidentified endogenous factors and even utilize acceptor molecules present in the GT containing microsomal fraction. Background activities may be an issue in a higher plant system, and preparing stable transformants is a slow process, often too slow to be practical for screening purposes. A transient expression system using Agrobacterium mediated transient transformation of N. benthamiana  shows some promise of solving the low through-put issue with stable transformants (JK Jensen, BL Petersen, N Geshi, P Ulvskov, HV Scheller, unpublished results).
Expression in prokaryotes, having no support for posttranslational processing, represents the other extreme. The relative high number of successfully expressed GTs of CAZy GT family 1 in E. coli  may primarily be ascribed to the fact that this GT family incorporates soluble enzymes acting on relatively low molecular mass acceptor substrates. Eukaryotic hosts are usually considered for human and other mammalian gene products, see e.g. review , and none have emerged as the ideal system, and likewise for higher plant gene products, all of the relatively few expression systems used at present are far from guaranteeing production of correctly folded and processed gene-products.
In recent years, the methylotropic yeast P. pastoris has become increasingly popular as expression host (for a comprehensive introduction to expression of heterologous proteins in P. pastoris see reviews [25, 26] and list of proteins expressed in P. pastoris (http://faculty.kgi.edu/cregg/index.htm)). In the present study we have expressed two A. thaliana α-1,3-xylosyltransferases, RGXT1 and RGXT2, implicated in the xylosylation of the internal fucose moiety of the A-chain of pectic rhamnogalacturonan II (RG II) , in P. pastoris and made a first characterization of the enzymes. Using the P. pastoris system, we have addressed a number of problems to consider with any eukaryotic expression system: post translational modification (especially glycosylation), activity in relation to the introduction and placement of tags, and the occurrence of interfering substances in the medium, a problem that applies to many liquid culture-based systems. On a µl to µl medium basis the P. pastoris system provides similar activity yields as compared to the baculovirus insect cell system, thus making the P. pastoris system a highly competitive system to be used e.g. in large-scale gene discovery efforts.
2 Materials and methods
2.1 Prediction servers
Prediction of N-glycosylation sites and TMDs in RGXT1 (At4g01770), RGXT2 (At4g01750), and RGXT3 (At1g56550), were done using the NetNGlyc (using 0.5 as threshold) and TMHHM servers at Center for Biological Sequence Analysis (CBS, http://www.cbs.dtu.dk).
2.2 Cloning of the soluble parts of RGXT1 and RGXT2 in to the pPiczαA vector
where capital letters encode the Flag-Tag (’MDYKDDDDK’, Invitrogen), the underlined sequences denote PmlI, NotI and EcoRI restriction sites and s and F denote the soluble part of the protein and Flag-tag, respectively (as read from the N-terminus to the C-terminus of the fusion proteins). Initially, F-sRGXT2 and sRGXT2-F, containing an N-terminal and a C-terminal Flag-tag, respectively, were PCR amplified, subcloned, sequenced and finally cloned into the pPiczαA vector (Invitrogen) using the 5’ and 3’ restriction sites, EcoRI and NotI, respectively, introduced by the primerpair PF-sRGXT2-5 & PF-sRGXT2-3 or the primer-pair PsRGXT2-F-5 & PsRGXT2-F-3, producing the constructs F-sRGXT2-pPiczαA and sRGXT2-F-pPiczαA, respectively. Subsequently, F-sRGXT1-pPiczαA and sRGXT1-F-pPiczαA (sRGXT1 with either an N-terminal or C-terminal fused Flag-tag) were made by replacing the PmlI-NotI (5’-3’) insert or the EcoRI-PmlI (5’-3’) insert of F-sRGXT2-pPiczαA and sRGXT2-F-pPiczαA, respectively. PCR was performed in 50 μl reaction volumes using the Expand High Fidelity system (Beohringer Ingelheim, Copenhagen, Denmark) with the cycle parameters: 3 min 97°C (Denaturation), 30–35 cycles: 94°C for 30 s, 50°C for 30 s and 68°C for 1’ followed by 12 min at 72°C. All PCR amplifications were cloned into the pCR®2.1 vector using the TOPO-TA cloning kit (Invitrogen) and the authenticity of the inserts was verified by sequencing, before the final cloning into the end vector (pPiczαA with/without an N-terminal or C-terminal Flag- tag).
Cloning and expression of F-sRGXT3 are described in .
2.3 Scoring, growth and induction of the highest expressed P. pastoris transformants
Linearized DNA of the constructs F-sRGXT1-pPiczαA, sRGXT1-F-pPiczαA, F-sRGXT2-pPiczαA and sRGXT2-F-pPiczαA were transformed into the P. pastoris KM71H strain by electroporation, plated on yeast extract peptone dextrose sorbitol plates containing zeocin (100 μg • mL-1), and incubated for 2–3 days at 30°C as described in Invitrogen Life technologies: The Pichia pastoris Expression system. In order to score the highest expressing transformants, 10 transformants of each construct were grown in 5 ml cultures and expressed as described below for two days at 30°C, where after the medium fraction was subjected to western blot analysis.
In expression studies, 10 mL of buffered complex glycerol medium supplemented with 100 μg • mL-1 zeocin were inoculated with the highest expressed transformant of each construct and grown in an incubator at 28°C, 280–300 rpm, ON until the OD600 reached a value between 2 and 6. Cells were harvested by centrifugation and re-suspended in buffered 50 mL complex methanol medium to an OD600 of 1.0 in 250-mL Erlenmeyer flasks. Incubation was continued for a total of 120 h at 20°C, 280–300 RPM, and at 24 h intervals, methanol was added to 0.5% (v/v) final concentration. At each time point 1 ml fractions were collected and the media fractions were recovered by centrifugation, frozen in liquid N2 and stored at -80°C. At day five (120 h of induction) cells were pelleted by centrifugation and the supernatant (medium fraction) was collected for further processing or analysis (see below). The presence of 1 mM DTT in xylosyltransferase assays (see below) resulted in slightly but reproducibly reduced activities (data not shown). More than one freeze thaw cycle resulted in detectable loss of activity.
2.4 SDS-PAGE, immunoblotting analysis, and quantifications of protein levels
Proteins were dissolved in NuPAGE Antioxidant and NuPAGE Loading Sample Buffer (LDS) from Invitrogen (1 : 1), boiled for 5–10 min, spun down and separated on SDS–containing 4–12% polyacrylamide Tricine gels (Invitrogen) using the MOPS-SDS system (Invitrogen) as running buffer. Electrophoresis was performed with the settings: 200 V, 150 mA, 200 W, 1 h. Molecular markers were the SeeBlue Plus2 Pre-Stained Standard (Invitrogen) and MagicMark XP Western Standard (Invitrogen). Proteins were transferred to activated (99.9% EtOH) polyvinylidene difluoride (PVDF) membranes (0.2 μm pore size, Invitrogen) using semi-dry electro blotting (settings: 200 V, 150 mA, 200 W, 1 h). Following transfer the PVDF membranes were washed 2 × ddH2O, 2–3’, blocked in 1 × TBS 3% skimmed milk powder (SKP), 30’, washed in 1 × Tris Buffered Saline (1 × TBS), pH 8.0, probed with mouse Anti Flag M2 monoclonal Antibody (Sigma Aldrich, Denmark) in a 1:1000 dilution in 1 X TBS 3% SKP overnight at 4°C under mild shaking, washed in 1 × TBS, probed with rabbit anti mouse peroxidase-conjugated secondary antibody (DAKO, Copenhagen, Denmark) in a 1:1,000 dilution in 1 X TBS 3% SKP and washed 8 × 3’ in 1 × TBS with 0.05% Tween20 and finally placed in 1 × TBS. Chemi-luminiscence was monitored using the Super Signal Enhanced Chemical Luminiscence (ECL, Copenhagen, Denmark), which was visualized on a BioSpectrum (UVP BioImaging Systems, Upland, California, USA) and semi quantified using the LabWorks program. Amount of Flag-tagged GTs were semi-quantified by comparison to standard curves of dilution series of 5 ng, 10 ng, 20 ng and 30 ng Amino-terminal Bovine Serum Albumin Protein Met-FLAG (BAP) (468 a.a., 49.4 kDa) from Sigma. Expression levels of secreted T7-tagged RGXT1 and RGXT2 (39.9 and 40.7 kDa, respectively) in the baculovirus system were estimated by co-electrophoresis and blotting of a dilution series of purified T7 tagged Muc1 3 1/2 tandem repeat (10.2 kDa). Immuno detection of the T7-epitope was carried out using a mouse monoclonal antibody directed against the T7-tag conjugated to horseradish peroxidase (Cat. No. 69048-3, Novagen, Merck KGaA, Darmstadt, Germany) in a 1:1000 dilution. An SDS-PAGE gel including both the BAP and the Muc1-standard series was run and Coomassie stained to verify the estimated amounts. Total protein in the media fractions was estimated using the Bradford Reagent (Sigma).
2.5 Purification of expressed proteins
In order to identify the optimal (NH4)2SO4 concentration for binding to phenyl Sepharose, 3 × 5 ml F-sRGXT2 containing media were incubated with 25%, 50% 75% saturated (NH4)2SO4, respectively, for 1–2 h on ice. The mixtures were spun down (5000 × g, 4°C, 10 min) and the three supernatants were applied on columns containing 300 μl CL-4B phenyl Sepharose® (GE Healthcare, Buckinghamshire, UK). Each column was washed with five column volumes of 50% (NH4)2SO4 and eluted with 3 × 1.5 ml 2.5 mM ammonium formate, pH 7.6. Eluates and precipitates were subjected to SDS-PAGE and analyzed by immunoblotting. From this experiment it was evident that F-sRGXT2 precipitates at (NH4)2SO4 saturation higher than 50% and lower than 75% and that the phenyl Sepharose binds F-sRGXT2 in 50% saturated (NH4)2SO4. In scaled up experiments 16 ml GT containing medium and 16 ml 100% (NH4)2SO4 were mixed and applied to 2 ml phenyl Sepharose, which was eluated in 2 × 2 ml 2.5 mM ammonium formate, pH 7.5. Crude media fractions (5 – 10 ml) and phenyl Sepharose eluates were dialyzed in 5 L 25 mM ammonium formate, pH 7.5, using Medicell MWCO 12-14000 Da dialysis bags (KEBO Lab) for 3 h at 4°C, and continued overnight with 5 L of fresh dialysis buffer.
Purified and dialysed proteins (9.3 µg F-sRGXT1, 7.1 µg sRGXT1-F, 10.6 µg F-sRGXT2 and 6.5 µg sRGXT2-F) were mixed with 20 µl 5 × incubation buffer (0.25 M sodium phosphate, pH 7), 2 µl Peptide:N-glycanase F [5 U • ml-1] and 2 µl O-glycanase [1.25 U • ml-1] in total volumes of 100 µl according to Glyko Reactionlab AS (Lynge, Denmark). Two aliquots of 33 µl of each mixture were incubated 20 h at 25°C and 37°C, respectively. Aliquots (10 µl) of each sample were used in the xylosyltransferase assay using 0.5 M L-fucose as acceptor. Aliquots (2 µl) of the remaining samples were subjected to immunoblotting as described above.
2.7 The xylosyltransferase assay
Standard xylosyltransferase assay was carried out according to the general free sugar assay  in 50 µl reaction mixtures containing 25 mM ammonium formate, pH 7.5, 10 mM MnCl2, 0.75 μl UDP-α-[14C]-D-xylose (9.8 GBq/mmol, 264 mCi/mmol, 10 nCi/μl, 0.75 μl ~ 12-14000 CPM) (NEN, Boston MA, USA), 100 µM UDP-α-D-xylose (CarboSource (Complex Carbohydrate Research Center, Athens, Georgia, USA)), 0.5 M L-fucose and 5 µl, except when otherwise stated, of the enzyme in question, which were incubated for 1 h at 30°C. Unincorporated UDP-α-[14C]-D-xylose and UDP-α-D-xylose were removed by passing the reaction mixture through a Dowex-1 anion exchanger (Sigma-Aldrich Denmark A/S) and the radioactivity in the flow through was determined by scintillation counting.
2.8 Electrospray ionization mass spectrometry of assay product
ESI-MS data were obtained by use of a ThermoFinnigan connected to an AXP-MS, Dionex. The mass spectrometer was run in positive mode with 0.1% HCOOH in 50 mM NaCl solution. Samples were prepared as described under xylosyltransferase assay, but without use of UDP-α-[14C]-D-xylose and omitting the anion exchange step.
2.9 Mn2+ and pH dependency
Ammonium formate stock solutions (250 mM) with pHs in the range from pH 3 to 11 were prepared and standard xylosyltransferase assays were incubated at 30°C for 60 min. The pH of each assay was monitored in parallel by probing the assay mixtures on pH strips directly after incubation and confirmed to be identical to the pH of the stock solutions.
3.1 Detection of heterologously expressed and secreted protein in P. pastoris
3.2 Expression levels and estimated turnover rates
The four soluble RGXT1 and RGXT2 variants produced high MW smears centred around 70 to 100 kDa, when subjected to immunoblotting analysis (Fig. 2) suggesting that the secreted proteins were hyper-glycosylated. The proportion of recombinant protein, as determined by immunoblotting, to total protein of the media fractions were in the range of 35–55% for F-sRGXT1, sRGXT1-F, F-sRGXT2 and sRGXT2-F, and more than 70% in case of F-sRGXT3 (data not shown).
3.3 Optimization of activity
The authenticity of the formed disaccharide product was verified using electrospray ionization mass spectrometry (ESI-MS). However, substantiation of disaccharide product formed in xylosyltransferase assays containing the standard amounts of free sugar (0.5 M L-fucose) by ESI-MS was only feasible when the enzymes had either been subjected to simple dialysis or purification on a phenyl Sepharose column (Fig. 3b). Replacing L-fucose with the better acceptor substrate methyl α-L-fucoside , enabled lowering the acceptor-concentration 10 or 100 fold in the assay, thus improving the signal to noise ratio significantly (Fig. 3c).
3.4 Deglycosylation of hyper-glycosylated proteins
3.5 Divalent cations are required for enzymatic activity
3.6 The four xylosyltransferase variants are active over a broad pH spectrum
3.7 Comparison of RGXT1 and RGXT2 expressed in insect cells versus in P. pastoris
Based upon semi-quantitative western analysis and coomassie stained SDS-page gels with T7-tagged reference protein, T7-tagged RGXT1 and RGXT2 were estimated to be expressed in baculovirus transfected insect sf9 cells at levels within the range of 5–9 mg protein • L-1 medium (data not shown). On a unit to unit basis (1 U = 1 nmol substrate conversion • min-1 • μl medium-1), the four proteins expressed in P. pastoris and dialysed yielded similar activities (within the range 0.011 to 0.013 U) to RGXT1 and RGXT2 expressed in baculovirus transfected insect sf9 cells.
Dialysis of sf9 derived media fractions containing expressed RGXT1 and RGXT2 resulted in slightly reduced activities.
More than 200 type II GTs are believed to participate in the synthesis of the unique and complex plant CW, and although heterologous expression is a key element in the assignment of function, successful heterologous expression and demonstration of enzymatic activity of plant CW GTs are rare events (Cf. e.g. recent 11th Cell Wall Meeting, Physiol Plant. 130). One prerequisite for advancing this field is identification and development of suitable expression hosts and schemes. In the present study, we investigated P. pastoris as host system for expression of two retaining plant CW xylosyltransferases, RGXT1 and RGXT2, and using this host system we have made a first basic characterization of the enzymes and provide a simple straightforward framework for how expression and elucidation of enzymatic activity may be addressed in the attempt to assign function to a given GT.
By introducing subtle modifications in the commercially available vector pPiczαA, constructs enabling expression of secreted N- and C-terminal Flag-tagged fusion-proteins were prepared. In the course of a 5 day induction period at 20°C, both N- and C-terminal Flag-tagged RGXT1 and RGXT2 were shown to accumulate in ample amounts with somewhat constant catalytic turnover numbers in the range of 0.15 to 0.3 • sec-1, thus suggesting that the enzymes were relatively stable in the media fractions under the conditions used for expression. The relatively low turnover rates may result from sub-optimal assay conditions of the free sugar assay and inhibitory substances in untreated media fractions. While enzymatic activity of the four RGXT1 and RGXT2 variants was found to increase 2-3 fold upon simple dialysis of the media, phenyl Sepharose mediated purification of F-sRGXT3 resulted in a ca. 8 fold increase in activity when compared on a μl to μl basis. Importantly, these simple extra rough purification steps may determine whether margin activities in sub-optimal general assays, such as the free-sugar assay, are detected. Simple dialysis and phenyl sepharose mediated purification of the media fractions were also found to enable direct verification of the product formed in crude assay mixtures using ESI-MS, thus adding a confirmatory layer to the free sugar assay.
For many mammalian GTs a native C-terminus has been shown to be absolutely required for activity [31–33], a requirement that was confirmed for the Golgi localized β-1,2-XylT from A. thaliana, which is involved in protein N-glycosylation . Here we demonstrate that addition of the apparent non-disruptive Flag-tag to the C-terminus, possessing the catalytic domain, of the soluble part of RGXT1 and RGXT2 does not appear to influence enzyme activity.
The migration patterns of RGXT1, RGXT2 and RGXT3 on denaturing SDS-PAGE gels, suggest that the enzymes are hyper-glycosylated, and that the predicted number of N-glycosylation sites in the proteins correlate with the degree of glycosylation. In order to investigate whether the enzymatic activity might be hampered by the hyper-glycosylation, the enzymes were subjected to enzyme mediated deglycosylation. Loss of activity following deglycosylation of Pichia-expressed enzymes is not unusual (see e.g. ). The enzymatic activity after deglycosylation was almost retained for N-terminal tagged RGXT1 and RGXT2, however, demonstrating that activity can be retained and that glycosylation of the two enzymes is not needed for activity in vitro.
Previously, the A. thaliana RGXT1 and RGXT2 were predicted to adopt the GT-A fold . The dependence of Mn2+ or, less efficiently, Mg2+ is common for GTs especially of the GT-A (SpsA) superfamily [29, 36]. The absolute requirement of Mn2+ with a somewhat broad optimum in the range between 2 and 20 mM was demonstrated for all of the enzymes tested, thus in magnitudes exceeding the concentration of Mn2+ in the Golgi vesicles, which is expected to be in the sub millimolar range. A similar range was observed for another member of the GT77-family, the Dictyostelium cytosolic UDP-Galactose:Fucoside α-1,3-Galactosyltransferase .
Expression of RGXT1 and RGXT2 in P. pastoris and in another commonly used system the baculovirus sf9 cells, yielded similar activities with estimated turnover numbers within the same range thus also in this respect making P. pastoris along with the overall expression and initial characterisation scheme presented here, a strong competitive host candidate for expression of plant GTs.
Dorthe Christiansen, Annette Hansen and Hanne Hansen are thanked for skilful technical assistance. This work was supported by the Danish National Research Foundation (Peter Ulvskov, Bent Larsen Petersen, Henrik Vibe Scheller), The Ministry of Science, Technology and Innovation (I.D. Grant No. 23-04-0238), the Carlsberg Foundation (Jack Egelund) and by contract MRTN-CT-2004-512265 'Wallnet' of the European Union FP6 Program (Kirsten Faber).
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.