A new strategy for full-length Ebola virus glycoprotein expression in E.coli

Ebola virus (EBOV) belongs to the Filovirdae family and it is a filamentous, enveloped, non-segmented, negative-sense RNA virus that causes severe viral hemorrhagic fever with a mortality rate of 50%-90% (Feldmann et al., 2003; Ascenzi et al., 2008). EBOV was first identified in 1976 in outbreaks in West-Central Africa, which led to hundreds of deaths (Sun et al., 2009). Subsequently, there were intermittent outbreaks of EBOV in the region and, in 2014, there was an Ebola epidemic that caused more than 11, 300 deaths according to the reports of the World Health Organization as of March 2016 (http://apps.who.int/ebola/current-situation/ebola-situation-report-30-march-2016). The duration, infectivity and mortality rate associated with this outbreak greatly exceeded previous outbreaks and caused much concern about the influence of EBOV on global health.

The genome of EBOV is approximately 19 kb in size and it encodes seven proteins comprising nucleoprotein (NP), matrix viral proteins 35, 40, 30 and 24 (VP35, VP40, VP30 and VP24), glycoprotein (GP) and an RNA-dependent RNA polymerase (RdRp). Among these proteins, VP40 is the most abundant protein, and it plays an essential role in virus assembly and budding (Jasenosky and Kawaoka, 2004; Licata et al., 2004; Hartlieb and Weissenhorn, 2006). The expression of VP40 alone enhanced the release of EBOV virus-like particles (VLPs) in mammalian and insect cells (Jasenosky et al., 2001; Noda et al., 2002; Ye et al., 2006).

The gp gene of EBOV encodes the secreted glycoprotein (sGP) and the surface glycoprotein (GP). GP is the only envelope protein that forms the spikes on the surface of the virions (Elliott et al., 1993; Sanchez et al., 1993; Volchkov et al., 1998). GP is co-and post-translationally processed into GP1 and GP2 by cellular proteases. Disulfide bonds can link GP1 and GP2 (Dolnik et al., 2004; Falzarano et al., 2006), and GP1 and GP2 play an essential role in virus attachment, entry and cytotoxicity (Ascenzi et al., 2008), as well as stimulating the host protective immune responses (Jones et al., 2005; Swenson et al., 2005; Sullivan et al., 2006; Bukreyev et al., 2007; Sun et al., 2009). Therefore, it is important to analyze the expression and structure of GP to elucidate the mechanism of EBOV attachment and fusion and to develop an antibody treatment.

EBOV GP and its fragments have been expressed in multiple expression systems including E.coli, cell-free, baculovirus and mammalian cell systems (Table 1). Full-length GP and subfragments of GP along with VP40 and NP can be produced in baculovirus and mammalian cell expression systems, which are often used in research on vaccines or fundamental studies of viruses. However, there are no reports on the successful expression of full-length GP or subfragments of GP (residues 17-501, 221-501 or 17-653) in E. coli. This may be due to their toxicity to cells, as only one subfragment of GP (residues 158-368) has been successfully expressed in inclusion (Das et al., 2007). Sullivan et al.proved that GP selectively decreased the expression of mammalian cell surface molecules that are essential for cell adhesion and immune function, leading to cell detachment and death (Sullivan et al., 2005). It has also been suggested that the mucin-like domain of GP plays a critical role in cell cytotoxicity, which might be involved in the extracellular signal-regulated kinases (ERK)/ mitogen-activated protein kinase (MAPK) pathway and the dynamin-dependent protein-trafficking pathway (Sullivan et al., 2005; Zampieri et al., 2007). In accordance with these studies, the GP subfragments A and B (residues 369-632 and 450-550, respectively) were chosen to be cloned into a prokaryotic vector, and we investigated whether small regions of the mucin-like domain have an adverse influence on the expression of the subfragments. In addition, there are many monoclonal antibody epitopes in the domain, and they could be used for developing an antibody treatment if they could be expressed in E. coli (Lee and Saphire, 2009b; Audet et al., 2014).

The current information on the crystal structure of GP is based on fragments of GP that do not contain the mucin-like domain and several other domains (Lee et al., 2008; Lee and Saphire, 2009a), which may mean that the effect of some domains has been overlooked in the analysis of the structure. To obtain full-length GP in E. coli cells, we used a new strategy. The gp gene was fused to the vp40 gene and then cloned into a prokaryotic expression vector and the full-length GP was successfully expressed in E. coli. The results are of importance for further analyzing the relationship between the structure and function of GP and for developing an antibody treatment.

We used E. coli strain DH5α and BL21, the bacterial expression vector pET28a(+), the baculovirus pFastBac^TM Dual vectors and the mammalian expression vector pcDNA3.1(+), which were stored in our laboratory. HEK-293T cells (which were originally derived from human embryonic kidney cells) were maintained in DMEM supplemented with 10% fetal bovine serum. Sf9 cells (which are insect cells that are commonly used for recombinant protein production using baculovirus) were cultured in Grace's Insect Cell Culture Medium supplemented with 10% fetal bovine serum. Monoclonal antibodies against Ebola GP were kindly provided by Dr. Ayato Takada (Hokkaido University Research Center for Zoonosis Control, Japan). Monoclonal antibodies against Ebola VP40 were kindly provided by Dr. Songtao Yang (Institute of Military Veterinary, Academy of Military Medical Sciences, Changchun, China) and polyhistidine (His₆)-tag monoclonal antibodies were purchased from Proteintech Group, Inc. (Chicago, Illinois, USA).

The vp40 and gp genes of EBOV (Zaire strain, Mayinga isolate) were synthesized using artificial chemical gene synthesis. The vp40 gene and subfragments of the gp gene sequence that corresponded to residues 369-632 and 450-550 were cloned into pET28a(+) vectors, which were labeled pET-V, pET-A and pET-B, respectively (Figure 1). In addition, a subfragment of gp (residues 369-632) was fused to the vp40 gene via a Tobacco Etch Virus (TEV) protease cleavage motif (i.e., ENLYFQS), both upstream and downstream of the vp40 gene. They were then cloned into pET28a(+) vectors, which were labeled pET-VA and pET-AV, respectively. In the same way, the full-length vp40 gene was also fused to the gp gene and cloned into a vector, which was labeled pET-VG. Moreover, the gp gene was fused to the vp40 gene via a furin cleavage motif (i.e., RRTRR) and cloned into a vector, which was labeled pET-GV (Figure 1).

Figure 1.  Schematic representation of recombinant plasmids and baculoviruses. The upper panel is a schematic of EBOV GP and the positions of subfragments A and B. The lower panels show that Ebola vp40 and gp subfragments A (residues 369-632) and B (residues 450-550) were cloned into pET28a(+) vectors, which were labeled pET-V, pET-A and pET-B, respectively. In addition, the vp40-A, A-vp40, and vp40-gp fusion genes, which were fused via a Tobacco Etch Virus (TEV) cleavage site, and the gp-vp40 fusion gene, which was fused via a furin cleavage site, were also cloned into pET28a(+) vectors, which were labeled pET-VA, pET-AV, pET-VG and pET-GV, respectively. The vp40, gp and gp-vp40 fusion genes were also cloned into pcDNA3.1(+) and pFastBac^TM Dual vectors, which were labeled pc-V, pc-G, pc-GV, rBV-V, rBV-G and rBV-GV, respectively.

The vp40 and gp genes were also cloned into pcDNA3.1(+) and pFastBac^TM Dual vectors, which were labeled pc-V, pc-G, rBV-V and rBV-G, respectively. In addition, gp was fused to vp40 via a furin cleavage motif (i.e., RRTRR), both upstream of the vp40 gene, and cloned into vectors, which were labeled pc-GV and rBV-GV, respectively (Figure 1).

The colonies of E. coli BL21 strains containing the recombinant plasmids were inoculated in 5 mL lysogeny broth (LB) containing 50 μg/mL kanamycin and they were allowed to grow overnight at 37 ℃ in a shaker. Subsequently, the overnight culture was diluted to 1/100 with 200 mL LB containing 50 μg/mL kanamycin. The suspension was shaken at 37 ℃ until an optical density at 600 nm (OD₆₀₀) of 0.5-0.6 was obtained and isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to medium to induce expression of the target proteins at 37 ℃ for 5 h, or at 16 ℃ overnight. The suspension was centrifugated at 5000 rpm/min for 10 min to harvest the cells, which were resuspended in 20 mL Buffer A (50 mmol/L NaH₂PO₄ at pH 8.0 and 300 mmol/L NaCl). The cells were lysed by sonication and then centrifugated to separate the cell debris and supernatant. The cell debris were resuspended in the same amount of Buffer A as that in supernatant.

The separated supernatant and E. coli cell debris were separately mixed with 5× SDS PAGE buffer and boiled for 10 min. Proteins in the supernatant and cell debris were separated using 10% SDS-PAGE and then stained with Coomassie Brilliant Blue to investigate the expression of the target proteins. In addition, the samples were transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, Massachusetts, USA), incubated with 5% skim milk in Tris-buffered saline (TBS) to block nonspecific antibody binding sites and then probed with the primary antibodies (anti-GP, anti-VP40 and anti-His₆). The blots were then treated with horseradish peroxidase-conjugated goat anti-mouse immunoglobulin (Merck Millipore, Darmstadt, Germany) and, lastly, visualized with FluorChem^® HD2 (Alpha Innotech, San Leandro, California USA).

HEK-293T cells were grown in a 12-well plate. When the confluence of the HEK-293T cells was around 85%, they were transfected with 3.2 μg plasmids pc-GV or pc-VP along with pc-G, respectively and 4 μL Lipofectamine^TM 2000 (Invitrogen, Carlsbad, CA, USA). The cells were then collected at 48 hours post transfection (hpt).

Sf9 cells were also grown in a 12-well plate. When the confluence of the Sf9 cells was around 85%, they were infected with recombinant baculoviruses rBV-GV (100 μL) or rBV-V (130 μL) along with rBV-G (180 μL) (multiplicity of infection (MOI) = 2). The cells were then collected at 72 hours post infection (hpi).

The cell debris were pelleted in 1.5 mL Eppendorf PCR tubes by centrifugation at 12000 rpm/min for 2 min and resuspended in 1 mL phosphate-buffered saline (PBS). The cells were mixed with 5× SDS PAGE buffer and boiled at 95℃ for 10 min, and then the samples were used in the WB.

On the following day, the HEK-293T and Sf9 cells were plated in 12-well plates. When the confluence was around 85%, they were transfected with 3.2 μg constructed plasmids, pc-GV or pc-V, and pc-G, with 4 μL Lipofectamine^TM 2000 (Invitrogen), or they were incubated with Ebola recombinant baculoviruses rBV-GV (100μL) or rBV-V (130 μL) along with rBV-G (180 μL) (MOI=2). At 48 or 72 hpi, the cells were fixed with 4% paraformaldehyde for 30 min, then washed with PBS and permeabilized using PBS containing 0.3% Triton X-100 for 30 min at room temperature. Prior to incubation with primaryantibodies, the cells were blocked with 1% bovine serum albumin diluted in PBS for 1 h at room temperature. The cells were then incubated with Ebola anti-GP or anti-VP40 monoclonal antibodies overnight at 4℃. The antibodies were all diluted in PBS containing 1% BSA. The cells were washed with PBS and then incubated with goat anti-mouse immunoglobulin conjugated with cyanine dye 3 (CY3) (Wuhan Boster Biological Technology, Wuhan, China). The cells were then detected and photographed using an OLYMPUS inverted microscope (Tokyo, Japan).

The recombinants pET-A (32 kDa) and pET-B (14 kDa) were constructed to attempt to produce the subfragments of GP (residues 369-632 and 450-550, respectively) in E. coli cells. The results of the SDS-PAGE showed that no target proteins were detected at either 37 ℃ or 16 ℃ (Figure 2A), even though pET-B only contained 100 amino acid residues. However, the experiments involving the recombinant pET-V containing the full-length vp40 gene led to the expression of VP40 at 37℃ and 16 ℃, and VP40 could be detected in both the supernatant and cell debris (Figure 2B, see arrowheads). In addition, a higher level of VP40 was produced in the supernatant at 16 ℃ compared to at 37 ℃ (Figure 2B).

Figure 2.  The expression of target proteins associated with pET-A, pET-B, pET-V, pET-VA, pET-AV, pET-VG and pET-GV was detected using 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDSPAGE). The samples of pET28a(+) alone were used as negative controls. S and C represent the supernatant and cell debris of the samples, respectively. The arrows indicate the target proteins.

The two fusion genes containing vp40 and gp subfragment A, pET-VA (68 kDa) and pET-AV (68 kDa), could not be expressed in E. coli, so no target proteins were detected in either the supernatant or the cell debris at 37 ℃ or 16 ℃ using SDS-PAGE (Figure 2B) or WB with anti-GP, anti-VP40 or anti-His₆ antibodies (Figure 3).

Figure 3.  The expression of target proteins associated with pET-V, pET-VA, pET-AV, pET-VG and pET-GV were detected using Western blotting (WB). The samples of pET28a(+) alone were used as negative controls. S and C represent the supernatant and cell debris of the samples. The arrows indicate the bands of the target proteins.

Unexpectedly, the other two fusion genes containing vp40 and full-length gp, pET-VG and pET-GV (114 kDa), were expressed in E. coli at 16 ℃ and the target proteins were detected in the supernatant and cell debris using SDS-PAGE (Figure 2C, see arrowheads) and WB with anti-GP, anti-VP40 and anti-His₆ antibodies (Figure 3). However, the target proteins of the two fusion genes were not expressed at 37 ℃ (Figure 2C). In addition, it was clearly shown that the quantity of the target protein produced from pET-VG was higher than for pET-GV (Figure 2C and Figure 3), which indicates the position within VP40 that affects the production of the fusion protein.

In previous studies, EBOV GP and VP40 could be expressed in baculovirus and mammalian expression systems by co-transfection of two plasmids or co-infection of two recombinant baculoviruses (Noda et al., 2002; Ye et al., 2006). We generated and successfully expressed a gp-vp40 fusion gene construct in E. coli and, here, we further tested this fusion gene plasmid expression strategy in eukaryotic systems (HEK-293T and Sf9 cells). The WB results revealed that the GP-VP40 fusion protein could be produced from pc-GV and, compared to the mobility of GP from pc-G with pc-V, it demonstrated that approximately half of the fusion protein could be cleaved into GP and VP40 by a furin protease (Figure 4A, left panel). Therefore, VP40 was detected in the HEK-293T cells. However, the mobility of the GP-VP40 fusion protein produced from rBV-GV was apparently lower than that of GP from rBV-G with rBV-V, and no VP40 was detected in the WB associated with rBV-GV (Figure 4A, right panel). This indicated that the GP-VP40 fusion protein was not cleaved in the Sf9 cells. The IFA results indicated that the target proteins could be detected in the HEK-293T and Sf9 cells using anti-GP or anti-VP40 antibodies, and it further proved that VP40, GP and the GP-VP40 fusion protein can be produced in mammalian and insect cells (Figure 4B).

Figure 4.  The expression of EBOV VP40, GP and the GP-VP40 fusion protein in HEK-293T and Sf9 cells. HEK-293T cells were transiently transfected with 3.2 μg plasmids pc-GV or pc-VP along with pc-G. Sf9 cells were infected with recombinant baculoviruses rBV-GV (100 μL) or rBV-V (130 μL) along with rBV-G (180 μL) [(multiplicity of infection (MOI) = 2)]. At 48 hours (HEK-293T) or 72 hours (Sf9 cells) post transfection/infection, cells were collected for Western blotting (WB) or fixed for immunofluorescence assay (IFA). (A) The expression of VP40, GP and the GP-VP40 fusion protein were detected with Ebola anti-GP or anti-VP40 monoclonal antibodies. (B) Cells were fixed and probed using Ebola anti-GP or anti-VP40 monoclonal antibodies, followed by staining with an anti-c-myc antibody goat anti-mouse immunoglobulin conjugated with cyanine dye 3 (CY3) (Wuhan Boster Biological Technology, Wuhan, China). The cells were then detected and photographed using an OLYMPUS inverted microscope (Tokyo, Japan). Original magification 200x.

The crystal structure of EBOV GP reveals a pre-fusion conformation (Lee et al., 2008). In such a structure, GP1 is tethered to GP2 by a disulfide bond, and the structure indicates that the mucin-like domain is located between the cap domain and the N-terminus of G2 (Figure 5B). In the context of full-length GP, the signal peptide (SP) should be closer to the mucin-like domain than the transmembrane region (TM), which means that the structure/function of the mucin-like domain is more vulnerable when VP40 is fused to the N-terminus than the C-terminus of GP (Figure 5C and 5D). Taking the expression experiment into consideration, we cautiously speculated that the interaction between VP40 and GP influences the productivity and solubility of fusion proteins in a space-dependent manner, and then make it possible to express the full-length GP in the GP-VP40 fusion protein.

Figure 5.  Illustration of the VP40-GP and GP-VP40 fusion proteins. (A) Crystal structure of EBOV GP1+GP2 (pdb:3csy). The crystal structure of GP1+GP2 is shown as a diagram with a translucent surface. The base, head, and cap domain of GP1 are in green, blue and cyan, respectively. GP2 is in orange. In the crystal structure, the black sphere marked "N" indicates the N-terminal residue, R31, and that marked "C" indicates the C-terminal residue, W597. (B) Model of full-length GP. The mucin-like domain is located between the cap domain (T309) and GP2 (A503). The mucin-like domain, signal peptide (SP), and C-terminal membrane-proximal external region (MPER) and transmembrane (TM) region are indicated by boxes with dashed lines. (C) Structural arrangements of the VP40-GP construct from our model. (D) Structural arrangements of the GP-VP40 construct from our model. VP40 is shown as a gray circle with black border. N-VP40 and C-VP40 indicate the fusion of VP40 to the N-terminus and C-terminus of GP, respectively. (E) Schematic diagrams of VP40-GP (top) and GP-VP40 (bottom).

The three-dimensional structure ofGP provides crucial knowledge for understanding the mechanism of EBOV attachment and entry into host cells. Most of the information on the crystal structure of EBOV GP has been based on a truncated protein (Lee et al., 2008). Regions such as the mucin-like domain, which has recently been demonstrated to be involved in enhancement of EBOV infection (Favier et al., 2016), were missing from the truncated protein. Furthermore, there may be differences between the crystal structures of truncated and full-length GP. Therefore, it is necessary to produce full-length GP in E. coli to obtain accurate information on its crystal structure. Until now, full-length GP and even some subfragments of GP have not been expressed in E. coli (Das et al., 2007). In this study, we demonstrated that two subfragments of GP (residues 369-632 and 450-550) could not be expressed in E. coli, indicating that some regions of GP, such as the mucin-like domain, may be toxic to cells (Sullivan et al., 2005; Das et al., 2007; Zampieri et al., 2007). To successfully express full-length GP, we employed a new strategy in this study. The VP40 protein was fused to full-length GP via a protease cleavage motif. The results demonstrated that the VP40-GP and GP-VP40 fusion proteins could both be expressed in E. coli. Furthermore, as the GP-VP40 fusion protein was also detected in HEK-293T and Sf9 cells, the strategy of fusion protein expression can be used in not only in E. coli but also in mammalian and baculovirus expression systems.

According to the purported crystal structure of EBOV GP (Figure 5B), the mucin-like domain is located on surface of GP. As a result, we speculated that fusing VP40 to GP would influence the structure of GP (including its mucin-like domain), and then reduce its toxicity to E. coli (Figure 5C and 5D) and lead to the expression of VP40-GP and GP-VP40 fusion proteins in E. coli. According to the predicted structural arrangements of EBOV fusion proteins (Figure 5C and 5D), VP40 was closer to GP when it was fused to the N-terminus of GP. This potentially explains why the VP40-GP fusion protein was expressed more easily than the GP-VP40 fusion protein.

We also demonstrated that the VP40-GP fusion protein was expressed in both the supernatant and cell debris. As the molecular weight of VP40 is around 40 kDa, which is much bigger than the small fusion tags (His₆and glutathione S-transferase (GST)), its size affects the folding of GP. This means that the conformation of GP in the VP40-GP fusion protein is different to that of GP alone. VP40 must be digested from the fusion protein using a TEV protease in order to obtain GP alone when VP40-GP is purified. The conformation of GP alone in the supernatant is much closer to its native conformation than the conformation of GP in the VP40-GP fusion protein and it can therefore be used to analyze the crystal structure of GP and to study its function.

In our study, we noticed that there were someother protein bands besides the target bands, indicating the abortive expression or degradation of fusion proteins in the E. coli cells. Unfortunately, the purification of the VP40-GP fusion protein was unsuccessful because the target protein was not absorbed by the Ni²⁺ resin column. There is only one His₆-tag in the N terminus of the VP40-GP fusion protein (114 kDa) and the lack of a strong correct interaction between the His₆-tag and Ni²⁺ resin might be the main reason for the failure to purify the VP40-GP fusion protein. Furthermore, the small His₆-tag might be hidden by the large VP40-GP fusion protein, which may prevent the absorption in the column. Further investigation is needed into the effect of VP40 on the folding of the fusion protein and to establish a procedure for the purification and digestion of the fusion protein to obtain the soluble fully intact GP protein. In addition, it was puzzling that the fusion protein involving VP40 fused to GP subfragment A was not expressed in E. coli. As the full-length GP could be produced via fusion with VP40, this outcome may indicate that the GP subfragment A is particularly toxic to E. coli.

In conclusion, full-length EBOV GP was expressed in E. coli after fusing it with VP40. The quantity of the fusion protein produced was greatly influenced by the fused sequence of VP40 and GP.

This project was jointly supported by the National Science and Technology Major Project (2012ZX10004219 and 2012ZX10004403), the Presidential Fund of the Chinese Academy of Sciences and the Wuhan Key Laboratory on Emerging Infectious Diseases and Biosafety. We thank Dr. Guoliang Lu (Structural Biology of Viral Genome Replication, Wuhan Institute of Virology) for helpful advice.

JJZ and ZMY conceived the experiments. JJZ and YHY performed the experiments. JJZ, HX, BZ and ZMY analyzed the results, and JJZ wrote the first version of the manuscript. HX, BZ and ZMY checked and finalized the manuscript.