PEDS Advance Access published online on November 22, 2007
Protein Engineering Design and Selection, doi:10.1093/protein/gzm062
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Extracellular self-assembly of virus-like particles from secreted recombinant polyoma virus major coat protein
MRC Clinical Sciences Centre, Imperial College London, Hammersmith Hospital, Du Cane Road, London W12 0NN, UK
3 To whom correspondence should be addressed.E-mail: j.krauzewicz{at}imperial.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Mouse polyoma virus major coat protein (VP1) expressed from a recombinant baculovirus is efficiently transported to infected cell nuclei and assembles into protein nanospheres morphologically similar to natural capsids. The nanospheres readily combine with plasmid DNA to form a hybrid gene therapy agent known as virus-like particles (VLPs). To facilitate large-scale production of VLPs free from cellular contaminants, the use of stable Drosophila cell lines expressing either wild-type protein, or VP1 tagged with a secretion signal for targeting to the extracellular medium, was investigated. Both wild-type and tagged VP1 expressed at 2–4 mg VP1/litre of culture. As expected, the wild-type protein self-assembled into VLPs. The tagged VP1 was efficiently secreted to the extracellular medium but was also glycosylated, unlike wild-type VP1. Despite this fact, a small fraction of the recombinant secreted protein assembled into VLP-like structures that had altered disulphide bonding, but were still biologically active. These results demonstrate the considerable tolerance in the nanosphere assembly to structural (i.e. aberrant glycosylation) and environmental (i.e. extracellular medium vs. nuclear milieu) changes. Thus, with modifications to improve nanosphere assembly, the secretion method could be adapted to large-scale preparation of VLPs, providing significant advantages over current methods of production of the vector.
Keywords: gene therapy/polyomavirus/virus-like particles/VP1
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
In recent years, the use of virus-like particles (VLPs), nanoparticles formed from self assembling coat proteins from a number of icosahedral viruses, has been widely explored for clinical applications, ranging from vaccine production to gene therapy (Garcea and Gissmann, 2004
). The major coat protein of mouse polyoma virus, VP1, is well characterised in this regard. VP1 is a 45 kDa polypeptide that readily assembles into higher order arrays and has multiple post-translational modifications, including phosphorylation and acetylation (Bolen et al., 1981), as well as sulphation, hydroxyprolination and methylation (Ludlow and Consigli, 1987; Ludlow and Consigli, 1989; Burton and Consigli, 1996). These modifications are not necessary for capsid assembly, since non-phosphorylated bacterially expressed VP1 can be induced to assemble into 50 nm particles (Salunke et al., 1986), but a role in mature virion assembly/infectivity has been postulated (Li and Garcea, 1994). When expressed in the recombinant baculovirus system, VP1 assembles into pentamers in the cytoplasm, localises to the nucleus and self-assembles into 50 nm diameter nanospheres (Montross et al., 1991; Forstová et al., 1993). The nanospheres morphologically resemble empty capsids formed in wild-type virus infections of mouse cells and can be purified to near homogeneity on CsCl gradients. When mixed with plasmid DNA (pDNA) each nanosphere stably associates with a single molecule of DNA and the complexes, termed VLPs, can be used to mediate heterologous gene expression (Forstova et al., 1995; 
tokrová et al., 1999). Delivery of the pDNA occurs along pathways used by the natural virus and expression levels of transgenes encoded by the pDNA are comparable to those achieved with typical transfection reagents, but are derived from very few copies of DNA per cell (Krauzewicz et al., 2000; Bishop et al., 2006). Thus, these particles show promise as an alternative solution for delivering transgenes that could be useful for gene therapy. Although the recombinant baculovirus system produces large quantities of protein, it requires continuous production of virus and, as the VP1 targets to the nucleus, several steps of purification are needed to remove cellular components. Furthermore, both ‘empty’ particles and those packaged with random fragments of cellular and baculoviral DNA are produced in similar proportions and a final CsCl gradient is necessary to separate out the second high molecular weight assembly, that differs from ‘empty’ VLPs only by density. These ‘full’ VLPs are of concern as they represent a safety risk for gene therapy applications by transferring unknown genes or gene fragments. Thus, finding alternative methods for producing VLPs that avoid formation of ‘full’ particles is important to developing papova VLPs for clinical use.
Polyoma VLPs can also be produced from recombinant bacterially expressed protein and VLPs from this source are being investigated for use as gene transfer agents (May et al., 2002; Gleiter and Lilie, 2003). However, since the multiple post-translational modifications reported for VP1 are not reproduced in prokaryotic systems (Leavitt et al., 1985), an alternative eukaryotic source using the Drosophila expression system which is more suitable than the baculovirus system for bulking up production of recombinant protein was explored.
The Drosophila system involves the production of recombinant proteins from stable insect cell lines under the control of inducible promoters. It has been shown to be capable of generating up to 22 mg protein/l cell culture (Bunch et al., 1988; Johanson et al., 1995). Therefore, recombinant VP1 production was assessed using this system, and, to prevent nuclear assembly and contamination with cellular DNA, the gene was also fused to a cleavable secretion tag to export it to the extracellular medium. The wild-type recombinant protein was expressed but at a much lower level than was produced from recombinant baculoviruses and only a small proportion assembled into nanospheres in the nuclei of the expressing cells. When fused to the secretion tag, the recombinant VP1 was successfully exported to the extracellular medium, but was glycosylated. Glycosylation most likely results in aberrant disulphide bonding in VP1 pentamers and the loss of the 86 kDa VP1 dimer. Despite these facts, some secreted VP1 (0.3% of total) was assembled into VLP-like structures that were able to mediate gene transfer.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Materials
All chemicals and enzymes used were purchased from Sigma-Aldrich Company Ltd, BDH Laboratory Supplies Ltd. or Invitrogen BV unless otherwise stated. Trypsin was obtained from Promega Corporation and PGNase F from New England Biolabs. Antibodies used were mouse monoclonal antibodies raised against VP1 (pyVP1-A; Forstová et al., 1993), and secondary rabbit anti-mouse polyclonal antibody conjugated with horseradish peroxidase (Dako).
Expression of VP1 in Drosophila cells
A Bam HI/Bgl II DNA fragment containing the wild-type VP1 open reading frame (ORF) was excised from the baculovirus transfer vector pVL1393VP1 (Forstová et al., 1993) by restriction endonuclease digestion and ligated to Bam H1 cleaved phosphatase treated Drosophila expression vector, pRmHa3 (Bunch et al., 1988). Clones containing full-length insertions of the VP1 sequence in the correct orientation were identified by restriction enzyme analysis. For cloning into the secretion vector pMT/BiP/V5 (Invitrogen), the gene was amplified by the polymerase chain reaction with Bgl II and Eco RI sequences engineered into the primers (amino terminal and carboxy terminal to the VP1 gene, respectively) and ligated into the Bgl II/Eco RI cleaved expression vector. The PCR amplified VP1 sequence was verified to be wild type and in frame with the secretion signal by sequencing from the final vector. A gene for the C114S mutant of VP1 (a gift from R.L. Garcea) was cloned for expression in the Drosophila system as described for the wild-type gene, above.
Drosophila melanogaster Schneider 2 cells (Sc2; Schneider, 1972) were co-transfected by calcium phosphate precipitation (as previously described) (Forstová et al., 1995) with wild type or secretion vector plasmids mixed at a ratio of 20:1 (w/w) with pCoHYGRO (Invitrogen) encoding the hygromycin resistance gene, and cells selected for 3 weeks in the presence of 300 µg/ml hygromycin. Stably selected lines were grown in Schneider’s Drosophila Medium (Invitrogen) supplemented with 10% foetal calf serum (FCS) at 23°C in closed flasks. To facilitate purification of secreted protein, lines were adapted to grow free of serum in Ultimate Insect Medium (Invitrogen). Recombinant VP1 expression was induced with CuSO4 (1 mM) for 24–48 h, unless otherwise stated, in logarithmically growing cells, prior to harvesting.
Analysis of VP1 by western blotting
Cell pellets washed in phosphate buffered saline (PBS) and lysed by sonication into Laemmli sample buffer (LSB) (Laemmli, 1970), or supernatant or lysates mixed 1:1 in LSB, were fractionated by SDS–polyacrylamide gel electrophoresis (SDS–PAGE) and analysed by western blotting as previously described (Forstová et al., 1993). Briefly, cell lysates or culture supernatants were fractionated by SDS–PAGE, transferred by electroblotting to PVDF membrane (Millipore Inc) and VP1 species detected with a monoclonal antibody followed by horseradish peroxidase conjugated secondary antibody. Immunocomplexes were visualised by chemiluminescence using ECL (Amersham Pharmacia Biotech, UK) and exposure to X-ray film for 10 s unless otherwise stated.
VLP purification and gene transfer
Confluent VP1 wt or VP1sec cells, typically from six 162 cm2 flasks, were induced with 1 mM CuSO4. Two days later, intracellular VLPs were prepared according to the protocol used for baculovirus VLPs, as previously described (Krauzewicz et al., 2000). For secreted VP1sec protein, the culture supernatant was separated from cells by low speed centrifugation (1000 x g at 4°C for 5 min) and then further clarified by centrifugation at 8000 rpm. The supernatant was overlayed onto a sucrose cushion (20% sucrose w/v in 10 mM Tris pH 7.6) and centrifuged at 150 000 xg for 1 h at 20°C. The supernatant was harvested from above the sucrose phase. Pellets containing high molecular weight assemblies were resuspended in 10 mM HEPES buffer, pH7.5 (600 µl) and analysed directly or further purified on CsCl gradients as for recombinant baculovirus VLP purification.
For gene transfer VLPs were produced by mixing VP1 nanospheres and plasmid DNA (pEGFP, Clontech) at a ratio of 30:1 (w/w), and incubated with cos7 cells for 90 min in DMEM and then for 48 h in DMEM supplemented with 10% FCS, as previously described (Krauzewicz et al., 2000). EGFP expression was scored by counting green fluorescent cells by epifluorescence microscopy using an FITC filter set.
Two millilitre of high molecular weight VP1 were mixed with 9 ml of PBS, adsorbed to carbon coated, formar coated, copper grids and stained with phosphotungstic acid (pH 6.8). Photographs were taken on a transmission electron microscope (Philips, Netherlands) operating at 100 kV.
Pure VP1 capsids (3 mg/ml>95% pure VP1) isolated by differential centrifugation from VP1 recombinant baculovirus infected cells, as previously described (Krauzewicz et al., 2000), were fractionated by SDS–PAGE, the gel stained with Coomassie blue and the 86 kDa band excised. The gel slice was destained and subjected to trypsin digestion on the MassPREPTM digestion robot (Micromass). Resulting peptides were analysed on a MALDI-TOF instrument (M@LDI from Micromass). The peptide mass fingerprint generated was used to search against a peptide fragment database using Mascot from Matrix Science (www.matrixscience.com). The protein was identified as polyoma virus VP1. The identity of the protein was confirmed using nanospray MS/MS on an ion trap (LCQ Deca from ThermoQuest). Prior to analysis, the digest was desalted using a standard C18 zip tip (Millipore) and eluted into a nanospray needle (New Objective). MS/MS spectra were obtained from a number of peptides and the data were combined. The database was queried using the Mascot MS/MS search program. Eleven peptides were analysed all of which gave a significant match with the VP1 coat protein.
Proteins samples were denatured in buffer (0.5% SDS, 1% β-mercaptoethanol), at 100°C for 10 min. Buffer (50 mM sodium phosphate, pH7.5), supplemented with 1% NP-40 was then added and the samples incubated with 0.5 unit of PNGaseF at 37°C for 1 h.
Tunicamycin treatment of cells
Cells were induced with CuSO4 (1 mM) and incubated with tunicamycin (10 µg/ml) for the last 24 h prior to harvesting.
Purified VP1 (0.2 µg) or VP1sec (0.15 µg) VLPs in buffer (50 mM NaCl, 0.01 mM CaCl2, 20 mM Tris pH 7.6) were adjusted to 0.33 µg/µl total protein with bovine serum albumin. Half of each sample was treated with DTT (0.1M) and EGTA (50 mM) and incubated for 15 min at 37°C. Trypsin (0, 0.00003, 0.0003, 0.003, 0.03 and 0.3 µg) was then added to 20 µl aliquots of both treated and non-treated batches. The samples were then left to incubate at 37°C for 15 min before being analysed by SDS–PAGE and western blotting for VP1.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Characterisation of VP1 from Sc2 cell lines expressing wild type or secreted protein
The ORF for the mouse polyoma virus major coat protein, VP1, was cloned into Drosophila expression plasmids, either alone, or as a fusion product with the signal sequence of the chaperone protein, BiP. In both cases, the ORF was under the transcriptional control of the inducible metallothionin promoter. Drosophila Sc2 cell lines carrying stable copies of the VP1 genes were isolated by hygromycin selection, as described in Materials and Methods.
Cell lines carrying the wild-type VP1 (VP1wt) gene, or VP1 fused to the BiP signal sequence (VP1sec), were grown to logarithmic phase and, following induction of the gene with Cu2+, cells and culture supernatants were analysed for VP1 by western blotting, as shown in Fig. 1A. VP1 immuno-reactive species were observed in cell lysates (lanes 1–3) and culture medium (lanes 4–6) from both VP1wt and VP1sec cells. As expected, more VP1sec protein was found in the culture medium relative to intracellular lysate than VP1wt protein. However, both cellular and secreted VP1sec protein migrated through the gel more slowly than the wild-type species and the 86 kDa VP1 related species [indicated by an arrowhead (Friedmann, 1974; Hewick et al., 1975)] observed in wild-type VP1 preparations was absent from the VP1sec cell line. Time courses of expression (Fig. 1B) demonstrated that whereas VP1sec protein was easily detectable in culture supernatant after day 1 (lane 2) and increased with time (lanes 4 and 6), VP1 in the supernatant of wild-type expressing cells was only detected with prolonged incubation of the cells in the presence of Cu2+ (compare lanes 1 and 3 with 5). This observation suggested that the latter was the product of material released from lysed cells. Varying amounts of lysis were experienced with VP1wt cells in the following experiments, which may be due to the toxic nature of VP1 when expressed intracellularly in isolation from other polyoma viral proteins (our unpublished observations). It was estimated by antibody detection of dot blotted samples against known concentrations of purified VP1, that VP1sec protein was produced at 2 mg/l culture supernatant of secreted protein, with 0.04 mg/l associating with the cells, whereas the culture supernatant of VP1wt cells contained 0.02 mg/l VP1 with the majority of protein (4 mg/l) being associated with the cells.
|
Targeting through the secretory pathway results in glycosylation of VP1
The above analysis, suggested that VP1 was successfully targeted through the secretory pathway by the signal sequence from BiP, culminating in its release into the culture medium. However, VP1sec protein migrated on SDS–PAGE with an apparent molecular weight (Mr) 1–2 kDa higher than VP1wt protein. In some experiments, in VP1sec cell lysate but not supernatant, a second band migrating with wild-type VP1 was observed (Fig. 2, lane 6) suggesting partial processing of the protein. This increase in Mr could be due to lack of cleavage of the secretion tag, or VP1sec protein may be glycosylated as it passes through the secretory pathway. Wild-type VP1 is transported to the nucleus following synthesis (Chang et al., 1992), rather than entering the ER/Golgi lumen and is not glycosylated. However, examination of the primary amino acid sequence revealed two potential glycosylation sites (conforming to the consensus NXS/T/CX, where X is anything but P), NNTL and NYTG [amino acids 92–95 and 243–246 respectively; numbering according to (Soeda et al., 1980)]. To test whether VP1sec protein is glycosylated, cells were grown in the presence of tunicamycin, a fungal antibiotic that inhibits N-acetylglucosamine transferase resulting in loss of N-glycosylation (Tkacz and Lampen, 1975). Cell lysates and supernatants were then analysed for VP1 (Fig. 2A). In the experiment shown, the level of expression of VP1wt protein was high compared to VP1sec protein (lanes 1 and 5 vs. 2 and 6) and significant VP1wt protein was found in the culture supernatant. This high level expression may have contributed to increased cell death in the wild-type expressing cells and consequent release of VP1wt protein into the extracellular medium (lane 1). This interpretation is consistent with the observation that treatment with tunicamycin did not affect the level or nature of VP1wt protein in the supernatant (lane 3). By contrast, VP1sec protein was present in the culture medium in control cells but was absent in supernatant from cells treated with tunicamycin (compare lanes 2 and 4). Further, whereas intracellular VP1sec protein migrated as a doublet, likely representing processed and unprocessed forms (lane 6), in drug treated cells this was resolved into a single band migrating with an Mr similar to the wild-type species (lane 8).
|
These data suggest that VP1sec protein may be glycosylated. To confirm this supposition and to determine whether VP1sec protein was glycosylated via asparagine residues, protein from VP1wt or VP1sec cells was incubated with the N-glycosidase, PNGase F. The data (Fig. 2B) demonstrate that, whereas VP1wt protein was unaffected by treatment with PNGase F (lanes 1 and 2), VP1sec protein migrated faster following treatment, with an Mr similar to wild-type VP1 (lane 4). These data confirmed that VP1sec protein was modified by aberrant N-linked glycosylation.
Assembly of VP1sec into capsid-like structures
To determine whether VP1sec protein was able to assemble into VLPs, VP1sec culture supernatants were clarified by centrifugation at 8000 xg to remove cell debris and then high molecular weight assemblies were separated from low molecular weight species by centrifugation through a 20% sucrose cushion (Fig. 3A). The majority of VP1sec related material was collected in the sucrose and supernatant overlay (lane 2), indicating that much of the protein was present as low molecular weight species. However, a small but significant fraction was pelleted through the cushion (lane 3), comprising approximately 0.3% of the total VP1sec protein, and representing approximately a yield of 3 µg of VP1sec high molecular weight assemblies, as assessed on western blots by comparison with pure VP1 of known concentration. It is noteworthy that in these higher molecular weight assemblies, VP1sec protein was glycosylated (as indicated by the higher Mr) and the 86 kDa VP1 dimer was absent (Fig. 3A).
|
Analysis of this material by negative staining electron microscopy, revealed a proportion of the VP1sec protein assembled into
50 nm diameter nanospheres (Fig. 3B, left panel, large black arrows) that were similar in appearance to wild-type nanospheres isolated from recombinant baculovirus infected cells (right panel). Further, other structures in the VP1sec preparation had an appearance consistent with deformed nanospheres (small black arrows). The concentration of VP1sec VLPs (0.005 mg/ml) adsorbed onto the grid was 1/30 of that used for the recombinant baculovirus VLPs (0.15 mg/ml). In our best views of the grid, VP1sec VLPs averaged 10 nanospheres compared to 150 for the recombinant baculovirus VLPs; therefore, we estimate that a similar proportion of the high molecular weight VP1 was assembled into nanospheres or nanosphere-like structures in each case. Analysis of the stability of VP1sec VLPs by exposure to trypsin
A number of reports have indicated that disulphide bond formation is essential for the stability of polyoma (Walter and Deppert, 1975; Stehle et al., 1994; Stehle and Harrison, 1996; Schmidt et al., 2000) as well as papilloma (Li et al., 1998; Sapp et al., 1998; Chen et al., 2001) viruses and VLPs. Further, it has been suggested that the 86 kDa species represents a dimer of VP1 linked via an unusually stable disulphide bond during the capsid assembly process (Friedmann, 1974; Hewick et al., 1975). Since the morphology of VP1sec VLPs appeared less well formed compared to the recombinant baculovirus particles and the VP1sec protein fails to form the 86 kDa dimer, this suggested that the VP1sec VLPs may not have wild-type disulphide bonding. Disulphide bonding plays an important role in capsid assembly through stabilisation of calcium ion chelation in human polyomavirus particles and this can be observed by exposing particles to calcium chelators under reducing conditions, causing the VP1 to become less resistant to trypsin (Chen et al., 2001). Therefore, to assess the stability of the VP1sec VLPs, we tested their resistance to trypsin degradation under control and reducing (dithiothreitol and EGTA) conditions compared to wild-type VLPs. As shown in Fig. 4, VP1sec VLPs were degraded at lower concentrations of trypsin than recombinant baculovirus derived particles (lanes 20–24 vs. 8–12). Further, their sensitivity to trypsin did not increase under reducing conditions, whereas recombinant baculovirus derived particles were three orders of magnitude more sensitive (compare lanes 14–18 and 20–24 with 2–6 and 8–12). These results suggest that disulphide bonding and/or Ca2+ chelation in the VP1sec VLPs differs from wild type and although not conclusive are consistent with the VP1sec derived VLPs being less stable due to a lack of the correct disulphide bridges.
|
The relationship between VP1sec VLP disulphide bonding, the 86 kDa dimer and VP1sec protein glycosylation
Although the precise disulphide-bonding pattern in the mouse polyomavirus VLP structure is not documented, there are two conserved cysteines (C114 and C274, murine strain A2 and 3 sequence numbering) that occur throughout polyoma VP1 sequences from different species. The former is proposed in the murine strains to form a stabilising linkage in pentamers in the final capsid structure (Stehle et al., 1994; Stehle and Harrison, 1996) and the second is located near the putative Ca2+ binding pocket (Haynes et al., 1993). Coincidentally, the two putative glycosylation sites identified in the present study are positioned 18 and 27 amino acids, respectively, N-terminal to the conserved cysteines. To address which cysteine might be destabilised in the VP1sec VLPs, leading to the loss of the 86 kDa dimer and the less stable capsids, we further investigated the structure of the 86 kDa species.
First we characterised this species, isolated from wild-type VLPs, by mass spectrometry of peptides derived from tryptic digestion, to confirm that it was a homodimer of VP1 and not a heterodimer with another cellular polypeptide, as suggested by earlier reports (Friedmann, 1974). This was found to be the case as accurate mass measurements were obtained for 11 peptides, all of which corresponded to predicted peptides in VP1, indicating that the protein was composed entirely of VP1 (probability based Mowse score of >200). As can be seen from the data, summarised in Fig. 5A, no peptides were identified from the N-terminal 60 amino acids, or in the region of C114, consistent with these sequences having altered molecular masses.
|
To determine directly whether C114 is involved in forming the 86kDa dimer, we expressed a mutant of VP1 in the Drosophila cells, in which C114 has been replaced by serine. Analysis of the products by SDS-PAGE and western blotting (Fig. 5B) demonstrated a VP1-related species at 86 kDa, showing that the dimer formation does not involve this late forming disulphide bridge. In agreement with this we could also detect the dimer in VP1 isolated from the cytoplasm of VP1wt cells (data not shown). This observation suggests the disulphide bond represents an intrapentameric bond formed early in assembly, i.e. in pentamer assembly in the cytoplasm, prior to transport to the nucleus and that the nuclear environment is not necessary for its formation. We also noted from the mass spectrometry data that no peptides were identified from the region spanning the second putative glycosylation site (amino acids 243–246), and containing the putative Ca2+ binding site (amino acids 266–277) and the second conserved cysteine, C274. We, therefore, propose that if glycosylation is directly affecting the disulphide bond formation in VP1sec proteins, for instance by sterically hindering its formation, it is more likely to be via this second site, affecting either C274 or the nearby cysteine at 283.
Gene transfer mediated by VP1sec VLPs
We next asked whether VP1sec VLPs were still able to mediate gene expression despite the fact that they appeared less well formed, lack the 86 kDa dimer and were more sensitive to enzymatic degradation. VP1wt and VP1sec cells and VP1sec culture supernatants were partially purified using the scheme developed for isolating VLPs from recombinant baculovirus infected cells (Krauzewicz et al., 2000; See Materials and methods). The resulting material was separated on CsCl density gradients and collected fractions analysed for VP1 by western blotting, as shown in Fig. 6A. In each case, high molecular weight complexes were isolated with buoyant densities similar to empty recombinant baculovirus expressed VLPs (refractive indices of 1.359–1.363, corresponding to CsCl fractions 7–9). These data supported the finding by electron microscopy that VP1sec can assemble into VLP-like structures. In the VP1sec culture supernatant, additional VP1 related species were observed with higher molecular weights (Fraction 4, bottom panel), but the nature of these complexes was not further investigated, since they were unable to mediate gene transfer (data not given).
|
Material from fractions 7–9 of VP1wt and VP1sec VLP purifications were each pooled and the final concentration of VP1 estimated by comparison with purified VP1, as described above. A comparison of the pooled fractions by SDS–PAGE demonstrated that VP1sec from the high molecular weight assemblies migrated with a higher Mr than for VP1, showing that they were composed of glycosylated and therefore secreted protein (Fig. 5B). This fact was further confirmed by treating a sample of the material with PNGase F, resulting in a shift in VP1sec to the wild-type size (data not shown). The products were complexed to pDNA carrying the EGFP gene and incubated with Cos7 tissue culture cells, as previously described for the baculovirus derived VLPs (Krauzewicz et al., 2000
). In the small-scale trial attempted here, insufficient VLPs were produced to carry out our standard transfection procedure of 7.5–15 µg VLPs per 4 x 105 cells, particularly from the cell lysates. Therefore, material from pooled fractions 7–9 of each gradient were used in a single experiment and the numbers of transfected cells obtained were compared with dilutions of complexes formed from recombinant baculovirus derived VLPs. The results, shown in Table I, give the final yields of VP1 proteins, the amounts of pDNA complexed to them and the resulting transfected cell scores. The data show that VP1sec supernatants gave the best yield of VLPs (2.5 µg/1 x 108 cells) and that despite being glycosylated, were able to enhance gene transfer rates compared to DNA alone and at a level similar to recombinant baculovirus derived VLPs.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The purpose of this study was to investigate an alternative source for producing VLPs that would be suitable for scaling up for in vivo studies. Advantages of the Drosophila expression system were considered to be that: first, the recombinant protein is expressed from continuously growing cell lines, avoiding the need for recombinant baculovirus production; second, the system is eukaryotic, increasing the chances of faithfully reproducing the multiple post translational modifications of VP1. Most importantly, third, if the protein was successfully secreted, this would prevent contamination of the final preparation with VLPs containing random fragments of cellular or baculoviral DNA that constitute up to 50% of the total yield of VLPs from the baculovirus system and simplify the purification procedure. The data obtained demonstrated the potential of the system to fulfil the study objectives, but revealed several limitations that would need to be overcome to exploit it fully.
Wild-type VP1 was expressed from the recombinant Drosophila cell lines at a concentration of 4 mg/l cell culture. This level of recombinant protein production is significantly less than the yield from the baculovirus system, where VP1 constitutes up to 25% of the total cellular protein in some infections. However, since the baculovirus system relies on producing large quantities of high titre virus, the Drosophila system which uses a continuously growing cell line to produce VP1 provides significant advantages in terms of ease of scaling-up production and reliability. Expression of the wild-type protein was relatively toxic as judged by the rate of cell lysis, even though the VP1 gene was expressed under the control of an inducible promoter allowing the culture of large numbers of cells prior to gene induction. This finding is in agreement with previous observations that over-expression of VP1 in mammalian cells appears to be toxic (unpublished results).
Cell lines expressing VP1 fused to the signal peptide from the chaperone BiP (VP1sec) produced a similar level of protein to that from the VP1wt cells (approximately 2 mg/l cell culture). However, the VP1sec protocol shows more promise for scale up as the majority of the protein (98%) was exported to the extracellular space, providing VP1 relatively free of cell components. We have also successfully adapted the cell lines to grow in serum-free medium, with little loss in yield of VP1 species, which would allow the VP1 to be harvested in a background of soluble, relatively well-defined proteins (data not given). Analysis of the secreted protein showed that it migrated at a higher Mr than the wild-type protein and that this was due to glycosylation. We suggest that exposure to the ER/Golgi lumen revealed cryptic glycosylation sites on VP1 [NNTL and NYTG; amino acids 92–95 and 243–246 respectively; numbering according to (Soeda et al., 1980)], which with the wild-type gene are not glycosylated as the protein is transported to the nucleus following synthesis.
VP1sec protein was able to self assemble into VLPs that were active in the gene transfer assay. Since we have shown that intracellular VP1sec species are not glycosylated, as the assembled particles were composed of glycosylated VP1 species they, therefore, must derive from secreted extracellular material. This fact represents a significant advantage over the current baculoviral protocol, as these VLPs, whether self assembled or further purified and then assembled in vitro will be free of random fragments of cellular and baculovirus DNA which contaminate the baculoviral VLP preparations. Only a small percentage of glycosylated VP1sec protein formed VLPs, but this demonstrates that the exported protein is able to assemble into T = 7 icosahedra, i.e. the 50 nm nanospheres needed for gene transfer. The composition of the extracellular medium is not favourable to conditions determined for assembly of T = 7 VP1 icosahedra (Salunke et al., 1989), nor is the more dilute concentration of the protein resulting from export into the medium. These limitations are reflected in the heterogeneity of high molecular weight species, indicative of aberrant assembly, revealed in sucrose gradient analysis (Fig. 5). Consequently, experiments are underway to investigate methods for purifying VP1sec protein from this source by standard purification techniques. In this regard, a cell line that produced similar levels of VP1 sec protein, but that grows in serum free medium has been isolated and preliminary experiments show that the bulk of this VP1 can be precipitated with 25% ammonium sulphate. Further purification would then yield a protein source that could be used for in vitro reassembly and packaging studies. This provides a second important advance in that one of the limitations of the baculoviral VLP prototype is that the DNA packaging protocol results in only partial packaging of the pDNA (tokrová et al., 1999; Bishop et al., 2006). Unpackaged DNA is subject to degradation and accumulating evidence shows that such exposed DNA can trigger cellular defence mechanisms leading to loss or silencing of the transgenes and consequent inefficiency of the vector (Bishop et al., 2006).
The influence of the chemical nature of the extracellular medium compared to the cell nucleus on VLP assembly and stability is well documented. Early studies with baculovirus expressed VP1 demonstrated that differences in the cytoplasmic environment vs. the nucleus, governed whether VP1 was assembled into pentamers or VLPs (Montross et al., 1991). Both calcium binding and disulphide bonding are known to be necessary for capsid stability and are influenced by environmental conditions. VP1 chelates calcium ions allowing folding of the polypeptide such that capsids can be formed (Leavitt et al., 1985; Salunke et al., 1986; Salunke et al., 1989) and the final particle is thought to be stabilised by the formation of specific intrapentamer disulphide bonds (Stehle et al., 1994; Stehle and Harrison, 1996). In the case of the stability and structure of the VP1sec VLPs, the evidence presented here suggests a role for disulphide bonding, as VP1sec VLPs lack the 86 kDa VP1 dimer, which is thought to result from linkage via an unusually stable disulphide bond (Friedmann, 1974; Hewick et al., 1975). With the related JC virus, disulphide bonding protects the calcium ion binding and, thus, prevents the particles from disassembling (Chen et al., 2001). Here, we found that VP1sec VLPs were more sensitive to trypsin than their wild-type counterparts and their sensitivity did not increase in the presence of reducing/chelating agents consistent with the possibility that they lack the appropriate disulphide bonding for full stability. Thus, the low level assembly of VP1sec into VLPs and their relative instability may be due to the fact that cellular components necessary for disulphide bond formation are lacking in the extracellular medium. In this regard, this deficiency could be addressed in future expression/purification protocol by modifying the chemical environment, or, as it has been shown to be possible to create structural disulphide bonds in vitro for VLPs derived from JC polyomavirus VP1 (Chen et al., 2001), a similar approach could be used to stabilise the VP1sec VLPs. The data presented here indicate, however, that the bond formation lacking in VP1sec is not entirely due to chemical environment, as we have detected the 86 kDa dimer in cytoplasmic preparations of wild-type VP1 (data not shown), showing its formation is not dependent on the nuclear environment. An alternative explanation is that the covalent attachment of the glycosyl moiety impeded assembly of the VP1sec VLPs, or affected their stability. Positioning of the glycosylation sites close to cysteines involved in potential disulphide bridging is consistent with the lower capsid stability and lack of the 86 kDa species, both of which are indicative of deficiencies in disulphide bond formation. By analyses of the 86 kDa dimer (Fig. 5), we have ruled out the involvement of the cysteine located at amino acid position 114 and our data are consistent with the disruption of disulphide bonding by the second glycosylation site. Nonetheless, the small fraction of VP1sec VLPs obtained were able to mediate gene transfer to a level comparable to the wild-type VLPs, demonstrating that glycosylation does not prevent the particles interacting with or transferring pDNA to cells.
The putative glycosylation sites identified on VP1 would be located on the outer side of the capsid structure [as determined from the X-ray crystallography structure of the virus (Stehle et al., 1994)]. As it is possible to attach bulky ligands to the VLP surface without compromising biological activity (Gleiter and Lilie, 2001; Schmidt et al., 2001; Gleiter and Lilie, 2003), it is conceivable that the glycosylation has not affected the ability of the VP1sec VLPs to interact with cells. The effects on capsid stability, however, would need to be addressed to make a successful stable gene therapy vector. Ways of overcoming this problem could include enzymatic removal of the modification during purification, or mutating the consensus site to prevent glycosylation (although this latter approach may prevent export of the protein in a manner similar to treating cells with tunicamycin). Furthermore, if increased stability and assembly can be achieved, it may even be feasible to use the glycosylation as a platform for covalently introducing novel targeting ligands.
In summary, secretion of polyoma virus VP1 to the extracellular space may provide readily purified VP1 in a scaleable process to generate VLPs for gene therapy vector production, but to fully maximise the system, modifications need to be made to improve stability and assembly of VLPs from this source.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
This work was supported by the RPMS Hammersmith Hospital Special Trustees, the Medical Research Council and the European Community (grant No. BIO4-CT97-147).
|
Footnotes |
|---|
1 Contributed equally.
2 Present address: Department of Cancer Genetics, Division of Geneticsand Molecular Medicine, Guy’s Hospital, London SE1 9RT, UK.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
J. Ng and O. Koechlin contributed equally to this work. We thank C. F. Higgins for helpful discussions, R. L. Garcea for VP1 mutants and V. Emons for assistance with electron microscopy.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Bishop C.L., Ramalho M., Nadkarni N., May Kong W., Higgins C.F., Krauzewicz N. Mol. Cell Biol. (2006) 26:2583–2594.
Bolen J.B., Anders D.G., Trempy J., Consigli R.A. J. Virol. (1981) 37:80–91.
Bunch T.A., Grinblat Y., Goldstein L.S. Nucleic Acids Res. (1988) 16:1043–1061.
Burton K.S., Consigli R.A. Virus Res. (1996) 40:141–147.[CrossRef][Web of Science][Medline]
Chang D., Haynes J.I. 2nd, Brady J.N., Consigli R.A. Virology (1992) 189:821–827.[CrossRef][Web of Science][Medline]
Chen P.L., Wang M., Ou W.C., Lii C.K., Chen L.S., Chang D. FEBS Lett. (2001) 500:109–113.[CrossRef][Web of Science][Medline]
Forstová J., Krauzewicz N., Sandig V., Elliott J., Palková Z., Strauss M., Griffin B.E. Hum. Gene Ther. (1995) 6:297–306.[Web of Science][Medline]
Forstová J., Krauzewicz N., Wallace S., Street A.J., Dilworth S.M., Beard S., Griffin B.E. J. Virol. (1993) 67:1405–1413.
Friedmann T. Proc. Natl Acad. Sci. USA (1974) 71:257–259.
Garcea R.L., Gissmann L. Curr. Opin. Biotechnol. (2004) 15:513–517.[CrossRef][Web of Science][Medline]
Gleiter S., Lilie H. Protein Sci. (2001) 10:434–444.[CrossRef][Web of Science][Medline]
Gleiter S., Lilie H. Biol. Chem. (2003) 384:247–255.[CrossRef][Web of Science][Medline]
Haynes J.I., Chang D., Consigli R.A. J. Virol. (1993) 67:2486–2495.
Hewick R.M., Fried M., Waterfield M.D. Virology (1975) 66:408–419.[CrossRef][Web of Science][Medline]
Johanson K., et al. J. Biol. Chem. (1995) 270:9459–9471.
Krauzewicz N., tokrová J., Jenkins C., Elliott M., Higgins C.F., Griffin B.E. Gene Ther. (2000) 7:2122–2231.[CrossRef][Web of Science][Medline]
Laemmli U.K. Nature (1970) 227:680–685.[CrossRef][Medline]
Leavitt A.D., Roberts T.M., Garcea R.L. J. Biol. Chem. (1985) 260:12803–12809.
Li M., Garcea R.L. J. Virol. (1994) 68:320–327.
Li M., Beard P., Estes P.A., Lyon M.K., Garcea R.L. J. Virol. (1998) 72:2160–2167.
Ludlow J.W., Consigli R.A. J. Virol. (1987) 61:1708–1711.
Ludlow J.W., Consigli R.A. J. Virol. (1989) 63:2881–2884.
May T., Gleiter S., Lilie H. J. Virol. Methods (2002) 105:147–157.[CrossRef][Web of Science][Medline]
Montross L., Watkins S., Moreland R.B., Mamon H., Caspar D.L., Garcea R.L. J. Virol. (1991) 65:4991–4998.
Salunke D.M., Caspar D.L., Garcea R.L. Cell (1986) 46:895–904.[CrossRef][Web of Science][Medline]
Salunke D.M., Caspar D.L., Garcea R.L. Biophys. J. (1989) 56:887–900.[Web of Science][Medline]
Sapp M., Fligge C., Petzak I., Harris J.R., Streeck R.E. J. Virol. (1998) 72:6186–6189.
Schmidt U., Rudolph R., Bohm G. J. Virol. (2000) 74:1658–1662.
Schmidt U., Rudolph R., Bohm G. Protein Eng. (2001) 14:769–774.
Schneider I. J. Embryol. Exp. Morphol. (1972) 27:353–365.[Web of Science][Medline]
Soeda E., Arrand J.R., Smolar N., Walsh J.E., Griffin B.E. Nature (1980) 283:445–453.[CrossRef][Medline]
Stehle T., Harrison S.C. Structure (1996) 4:183–194.[Medline]
Stehle T., Yan Y., Benjamin T.L., Harrison S.C. Nature (1994) 369:160–163.[CrossRef][Medline]
tokrová J., Palková Z., Fischer L., Richterová Z., Korb J., Griffin B.E., Forstová J. FEBS Lett. (1999) 445:119–125.[CrossRef][Web of Science][Medline]
Tkacz J.S., Lampen O. Biochem. Biophys. Res. Commun. (1975) 65:248–257.[CrossRef][Web of Science][Medline]
Walter G., Deppert W. Cold Spring Harb. Symp. Quant. Biol. (1975) 39:255–257.[Web of Science][Medline]
Received October 31, 2006; revised September 17, 2007; accepted October 11, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||