Biotechnology Letters

, Volume 35, Issue 3, pp 413–421

Codon optimization of the human papillomavirus type 58 L1 gene enhances the expression of soluble L1 protein in Saccharomyces cerevisiae

Authors

  • Hyoung Jin Kim
    • College of Pharmacy, Chung-Ang University
  • Hye-Lim Kwag
    • College of Pharmacy, Chung-Ang University
    • College of Pharmacy, Chung-Ang University
Original Research Paper

DOI: 10.1007/s10529-012-1097-y

Cite this article as:
Kim, H.J., Kwag, H. & Kim, H. Biotechnol Lett (2013) 35: 413. doi:10.1007/s10529-012-1097-y

Abstract

The effect of codon optimization of L1 gene on the production of the L1 protein of human papillomavirus (HPV) was investigated in a yeast expression system. Saccharomyces cerevisiae was transformed with a plasmid containing either the wild type (WS)-HPV type 58 L1 (HPV58 L1) gene or a codon-optimized (MO)-HPV58 L1 gene. The proportion of soluble L1 protein expressed from MO-HPV58 L1 was significantly higher than that expressed from WS-HPV58 L1. Moreover, the amount of purified MO-HPV58 L1 protein recovered was 2.5-fold higher than the amount of WS-HPV58 L1 protein. Codon optimization of HPV58 L1 gene thus increases the proportion of soluble L1 protein and the amount of purified product that can be used as antigen to generate vaccines.

Keywords

Codon optimizationHuman papillomavirusL1 proteinSaccharomyces cerevisiae

Introduction

L1 protein is the major capsid protein of human papillomavirus (HPV), which can self-assemble into a virus-like particle (VLP) with a structure similar to that of naturally occurring HPV virions (Conway and Meyers 2009). Therefore, recombinant L1 proteins expressed in insect cells (Sf9) and Saccharomyces cerevisiae (S. cerevisiae) have been used as antigens in prophylactic vaccines (Garland and Smith 2010). The S. cerevisiae expression system that generates VLPs has a low production cost and short culture period (Park et al. 2008). However, productivity is sub-maximal. Improved fermentation technology and promoter systems are considered important for enhancing productivity (Kim et al. 2011; Mattanovich et al. 2004), but there has been little investigation into the possibility of codon optimization, which has been studied as a means of enhancing transgene expression for two decades, mostly in E. coli expression systems.

There are two strategies for codon optimization: one is to alter codons in a way that reduces mRNA secondary structure, and the other is to use the preferred codons of the relevant species (Angov 2011). Reducing mRNA secondary structure increases protein expression by increasing translation initiation (Kozak 2005), while changing to high frequency usage codons increases the binding of isoacceptor tRNA molecules to mRNA (Bulmer 1987; Ikemura 1985). There is disagreement over which strategy is more useful in enhancing transgene expression. The results of a recent study strongly suggested that reducing mRNA secondary structure was more effective in enhancing transgene expression than codon changes per se (Kudla et al. 2009).

The L1 ORFs of high-risk HPVs contain two putative start codons (ATGs) (Webb et al. 2005). The second in-frame ATG is located 78 nucleotides downstream of the first. The short L1 gene, which uses the second in-frame ATG as start codon, has been used to produce HPV L1 VLP (Kirnbauer et al. 1993; Zhou et al. 1991). The long L1 gene, which contains both ATGs, does not produce L1 VLPs efficiently (Webb et al. 2005), perhaps because of some effect of the nucleotide sequence between the first and second ATG on L1 production.

In this study we used three types of L1 gene; the long wild type (WL)-HPV58 L1 gene, the short wild type (WS)-HPV58 L1 gene, and a codon-optimized short (MO)-HPV58 L1 gene (Supplementary Fig. 1). The sequence of the codon-optimized gene was designed to reduce mRNA secondary structure without altering the amino acid sequence. Most previous studies have focused on the effect of codon optimization on protein expression levels. In contrast, in this study we focused on the effect of codon optimization on the proportion of soluble L1 protein made, and the amount of L1 protein recovered after purification.
https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig1_HTML.gif
Fig. 1

Comparison of L1 protein expression. Nine clones (for each gene) of cells transformed with WL-, WS- or MO-HPV58 L1 genes were cultured in 5 ml of SD-ura broth for 1 day and mixed. The cell mixtures were cultured in 150 ml YPDG medium at 30 °C for 48, 96 or 144 h with shaking at 230 rpm. To determine the levels of total and soluble L1 protein, western blotting was performed using cell lysates before and after removal of cell debris, respectively

Materials and methods

Construction of plasmids containing the WL-, WS-, and MO-HPV58 L1 genes

To construct the plasmid containing the WL-HPV58 L1 gene, the DNA sequence of the gene (containing both the first and second ATGs) was amplified from HPV58 DNA. (Matsukura and Sugase 1990). The 1,575 bp PCR product was ligated into pGEM-T-easy vector (Promega, USA) and subsequently into vector YEGα-MCS, using the EcoRI and SalI sites, to create YEGα-MCS-WL-HPV58 L1 (Supplementary Fig. 1). To construct the plasmid containing the WS-HPV58 L1 gene, the DNA sequence of the gene (containing only the second ATG) was amplified from the HPV58 DNA above. The 1,497 bp PCR product was ligated into pGEM-T-easy vector and subsequently into YEGα-MCS, using HindIII and SalI sites, to create YEGα-MCS-WS-HPV58 L1 (Supplementary material I). We confirmed that the WL- and WS-HPV58 L1 sequences were identical to those of the relevant sections of HQ357755.1 in GenBank. The MO-HPV58 L1 gene (1,497 bp) was synthesized by Blue Heron Biotechnology Inc. (USA) and ligated into YEGα-MCS, using HindIII and SalI sites, to create YEGα-MCS-MO-HPV58 L1 (Supplementary material I). HPV58 DNA was kindly provided by Dr. Jong Sup Park (College of Medicine, Catholic University of Korea, Seoul, South Korea), and the YEGα-MCS vector was kindly provided by Prof. Hyun Ah Kang (Chung-Ang University, Seoul, South Korea).

Calculation of folding free energies and codon adaptation indices

The folding free energies and codon adaptation indices (CAIs) of L1 mRNAs transcribed from WL-HPV58 L1, WS-HPV58 L1 and MO-HPV58 L1 were calculated using the software provided by the Vienna RNA WebServer (http://rna.tbi.univie.ac.at/) and the E-CAI server program (http://genomes.urv.es/CAIcal/E-CAI/), respectively.

Yeast transformation and cell culture

Saccharomyces cerevisiae Y2805 was transformed with YEGα-MCS-WL-HPV58L1, YEGα-MCS-WS-HPV58 L1 and YEGα-MCS-MO-HPV58 L1 as described previously (Soni et al. 1993), and transformants were selected on SD-ura medium (a synthetic complete medium without uracil). Fifteen colonies on SD-ura plate were selected for each gene and culture in 5 ml SD-ura broth for 2 days. Thereafter nine clones with high growth rates among 15 clones were selected for each gene. The nine clones were further cultured in 5 ml SD-ura broth for 1 day and mixed with the other clones transformed with the same gene. The mixtures were then inoculated into 150 ml YPDG medium containing 1 % (w/v) yeast extract, 2 % (w/v) peptone, 8 % (w/v) glucose and 1 % (w/v) galactose (Kim et al. 2011) and grown in flasks at 30 °C with shaking at 230 rpm for 48, 96, or 144 h. Cell densities were calculated from the OD600 values.

Preparation of cell lysates

For measurement of total L1 protein level, cell lysates were prepared without removing cell debris, while for measurement of soluble L1 protein level cell lysates were prepared with cell debris removed. For both preparations cells were harvested from 1 ml of cultures by centrifugation at 10,000×g for 10 min, mixed with disruption buffer (10 mM NaH2PO4, 150 mM NaCl, 1.7 mM EDTA, 0.01 % Tween 80, pH 7.2) and disrupted by vortexing with glass beads (BioSpec Products, Inc., USA). Cell lysates for measurement of total L1 protein level were then diluted 50-fold with disruption buffer. Cell lysates for measurement of soluble L1 protein level were centrifuged at 12,000×g for 10 min to remove cell debris, then diluted 20-fold with disruption buffer.

SDS-PAGE and western blotting

Diluted cell lysates were mixed with Laemmli sample buffer (Laemmli 1970) and heated for 12 min at 77 °C to perform SDS-PAGE (Cook et al. 1999). (Heating at 77 °C increases the resolution of L1 protein band.) Proteins were fractionated on 12 % acrylamide gel. The fractionated proteins were visualized with silver staining. Western blotting was performed as described previously (Kim et al. 2011). The band corresponding to L1 protein was detected using rabbit anti-HPV16 L1 polyclonal antibody (Pab) followed by HRP-conjugated anti-rabbit IgG (Bethyl laboratories, USA). L1 band intensities were determined with NIH open source software Image J (http://rsbweb.nih.gov/ij/), and calculations were performed as described previously (Kim et al. 2011). To determine intracellular BiP levels, 50 μg protein per well was loaded onto gels for western blotting. Tubulin was used as an internal control. Anti-KAR2 antibody (Santa Cruz Biotechnology, USA) and anti-tubulin antibody (Abcam, UK) were used for detection of BiP and tubulin, respectively.

Purification of L1 proteins

L1 proteins were purified from cultures after 48, 96, and 144 h as described previously (Kim et al. 2010). Briefly, protein in the cell lysates was precipitated with 40 % saturated (NH4)2SO4. The precipitated samples were clarified by removing precipitated contaminants and the L1 proteins were isolated by cation-exchange chromatography using P-11 resin (Whatman, UK). The purified proteins were analyzed by SDS-PAGE (Laemmli 1970), and western blotting. For SDS-PAGE and western blot, proteins were fractionated on 12 % acrylamide gel.

Determination of protein concentration

Protein concentrations were determined with a Bradford protein assay kit (Bio-Rad Laboratories, USA) with bovine serum albumin as standard.

Transmission electron microscopy (TEM)

Purified HPV58 L1 was dialyzed against PBS containing 0.33 M NaCl and 0.01 % Tween 80, and negatively stained with 2 % phosphotungstic acid (Kim et al. 2011). Electron microscopy was performed on a LIBRA 120 energy-filtering transmission electron microscope (Carl Zeiss, Germany) at a final magnification of 50,000×.

Evaluation of immunogenicity of purified HPV58 L1 proteins

Six-week-old female BALB/c mice were divided into three groups (PBS, WS-HPV58 L1 and MO-HPV58 L1), each consisting of 7–14 mice and immunized subcutaneously with PBS or purified HPV58 L1 proteins four times at 2-week intervals. Those in the PBS group received 100 μl PBS containing 200 μg Al(OH)3 per dose while those in the WS-HPV58 L1 and MO-HPV58 L1 groups received 10 ng purified WS-HPV58 L1 or MO-HPV58 L1 and 200 μg Al(OH)3 per dose.

Determination of anti-HPV58 L1 IgG titer

After four immunizations sera were obtained, anti-HPV58 L1 IgG titers were determined as described previously (Kim et al. 2012). 96-well ELISA plate was coated with purified WS-HPV58 L1 VLP (100 ng/well) and blocked with 3 % BSA in PBS-T (PBS containing 0.5 % Tween 20). Thereafter, serial dilution of the mice sera were incubated with the coated L1 VLPs for 1 h at 37 °C. The anti-HPV58 L1 antibody bound with coated HPV58 L1 VLPs was detected using HRP-conjugated anti-mouse IgG polyclonal antibody (Bethyl, USA), and the reactions were developed using o-phenylenediamine (Sigma. USA). The anti-HPV58 L1 IgG titers were determined by end-point titrations. To determine anti-HPV58 neutralizing antibody titers of mice sera HPV pseudovirus (PsV)-based assays were performed as described previously (Pastrana et al. 2004).

Results

Comparison of the WL-, WS- and MO-HPV58 L1 genes

WL-HPV58 L1 contains two ATGs, and its DNA sequence from the second ATG to the stop codon is identical to that of the WS-HPV58 L1 gene (Supplementary Fig. 1). To design MO-HPV58 L1 gene, the rare codons were removed, and the codons were substituted into preferred codons of S. cerevisiae. Then the codons of the resulting DNA were substituted for reducing secondary structure of mRNA. 60 % of the codons in the MO-HPV58 L1 gene differ from those in WS-HPV58 L1 (Supplementary Fig. 2). The folding free energies of WL-HPV58 L1, WS-HPV58 L1 and MO-HPV58 L1 calculated with the RNA fold program were −400, −377 and −265 kcal/mol, respectively. The frequencies of preferred S. cerevisiae codons for Ala, Arg, Gln, Glu, Gly, Lys and Pro were higher in MO-HPV58 L1 than in WS-HPV58 L1, while those for Asn, His, Ile, Leu, Phe, Thr, Tyr and Val were lower in MO-HPV58 L1 than in WS-HPV58 L1 (Supplementary Table 1). The CAIs of the MO-HPV58 L1 and WS-HPV58 L1 genes for S. cerevisiae were 0.80 and 0.83, respectively. The similarity of these values indicates that the alteration of codons in the MO-HPV58 L1 gene has little effect in terms of codon usage optimization; the gene is instead optimized for reducing mRNA secondary structure.
https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig2_HTML.gif
Fig. 2

Comparison of L1 protein finally recovered from purification. Panela shows the amounts of L1 proteins in soluble fractions of cell lysates. To analyze the amount of soluble L1 proteins, supernatants were obtained after removing cell debris and were diluted 1:25 with cell disruption buffers, and 5, 2.5 and 1.2 μl the supernatants were loaded for western blotting. The L1 proteins were purified as described in “Materials and methods” section, and the L1 protein contents were determined by Bradford protein assay (b). Data are mean ± SD of three independent purifications

Comparison of L1 protein expression and amounts of L1 protein recovered after purification

Cells transformed with WS-HPV58 L1 yielded the highest level of total L1 protein and those transformed with WL-HPV58 L1 produced some L1 protein but most of the L1 protein produced in either case was lost after removing cell debris (Fig. 1). In contrast, the L1 protein produced in cells transformed with MO-HPV58 L1 remained after removal of cell debris (Fig. 1). The absolute level of soluble L1 protein was thus highest in cells transformed with MO-HPV58 L1 (Fig. 2a). No L1 protein was purified from cells transformed with WL-HPV58 L1 (Fig. 2b), and the amount of L1 protein purified from cells transformed with MO-HPV58 L1 was 2.5-fold higher than the amount purified from cells transformed with WS-HPV58 L1 (Fig. 2b). The amount of L1 protein purified from 150 ml culture of cells transformed with MO-HPV58 L1 gene for 144 h was 0.48 mg (Fig. 2b). This amount of L1 protein corresponds to 24 human doses for vaccination.

Comparison of soluble L1 protein levels in different transformed clones

We measured the amounts of soluble L1 protein in 12 different clones transformed with WS-HPV58 L1 and 12 transformed with MO-HPV58 L1. For WS-HPV58 L1 clones 7 and 8 produced high levels of soluble L1 protein while for MO-HPV58 L1 the highest levels were produced by clones 5, 7, 8, 9, 11 and 12 (Fig. 3a). There is significant difference between different clones in the L1 protein production. Therefore, screening to select the clone with higher production level of L1 protein is thought to be important. Again, the mean L1 band intensity on Western blots was higher for cells transformed with MO-HPV58 L1 than those transformed with WS-HPV58 L1 (Fig. 3b). This result indicates that transformation with the MO-HPV58 L1 gene provides a greater opportunity to select clones with high levels of L1 protein expression than transformation with WS-HPV58 L1 gene.
https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig3_HTML.gif
Fig. 3

Comparison of soluble L1 protein levels in cells transformed with WS- and MO-HPV58 genes. 12 clones with a high growth rate were selected from transformed cells. Cells were cultured at 30 °C in 150 ml YPDG medium for 144 h. 1 ml of each culture was harvested, and the cell debris removed by centrifugation at 12,000×g for 10 min. The supernatant protein was loaded onto gels for western blots (a). To enable analysis of L1 band intensities, 300 and 150 ng of purified WS-HPV58 L1 proteins were loaded as standards. The band intensity of the 300 ng standard was set at 100 %, and the band intensities of other L1 protein samples are given as percentages of that band intensity. The band intensities are presented as mean ± SEM (b)

Conformation and immunogenicity of purified WS- and MO-HPV58 L1 proteins

Highly purified HPV58 L1 proteins were obtained from cells transformed with the WS- and MO-HPV58 L1 genes (Fig. 4a). TEM analysis confirmed that both WS- and MO-HPV58 L1 proteins self-assembled into virus-like particles (VLPs) (Fig. 4b). There was no difference between WS- and MO-HPV58 L1 proteins with respect to their immunogenicity: both efficiently stimulated the production of anti-HPV58 L1 IgG (Fig. 5a) and neutralizing antibody (Fig. 5b) in mice.
https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig4_HTML.gif
Fig. 4

SDS-PAGE, western blotting and TEM analysis of purified L1 proteins. HPV58 L1 proteins were purified from cells transformed with WS- and MO-HPV58 L1 genes. The purified L1 proteins were analyzed by SDS-PAGE and western blotting (a). To analyze the L1 protein content, 200 ng of purified L1 protein samples was loaded onto gels. TEM images of purified WS-HPV58 L1 and MO-HPV58 L1 proteins were also obtained (b). The results for each VLP preparation were obtained from three sections on carbon coated grids Magnification: bars = 100 nm

https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig5_HTML.gif
Fig. 5

Immunogenicity of purified WS- and MO-HPV58 L1 proteins. Mice were immunized subcutaneously four times with purified WS- or MO-HPV58 L1 proteins (10 ng of L1 protein per dose, in combination with Al(OH)3). Anti-HPV58 L1 IgG (a) and neutralizing antibody titers (b) were determined. Horizontal bars indicate median values of titers (PBS, n = 7; WS-HPV58 L1, n = 14; MO-HPV58 L1, n = 14)

Codon optimization of HPV58 L1 gene reduces the intracellular BiP level

We measured the intracellular BiP level between cells transformed with WL-, WS- and MO-HPV58 L1 genes to investigate the stress response of the cells. As shown in Fig. 6, the BiP level in cells transformed with MO-HPV58 L1 gene was lower than those transformed with WL- and WS-HPV58 L1 genes. These results indicate that MO-HPV58 L1 gene produces more abundant soluble form of L1 protein with lower stress response.
https://static-content.springer.com/image/art%3A10.1007%2Fs10529-012-1097-y/MediaObjects/10529_2012_1097_Fig6_HTML.gif
Fig. 6

Comparison of intracellular BiP levels in cells transformed with WL-, WS- and MO-HPV58 L1 genes. Cells were cultured for 24, 48, 96 or 130 h, harvested and disrupted as described in “Materials and methods” section. To determine intracellular BiP levels, western blotting was performed as described in “Materials and methods” section. Tubulin was used as an internal control. Panela shows the results of western blots detecting BiP and tubulin. The BiP band intensities were determined relative to the intensity in 24 h-cultured cells transformed with the WL-HPV58 L1 gene, which was set at 100 % (b)

Discussion

We have compared levels of L1 proteins derived from the WL-HPV58 L1, WS-HPV58 L1 and MO-HPV58 L1 genes in S. cerevisiae. It has previously been unclear whether L1 protein is produced from the long L1 gene in yeast. We found that L1 protein was produced from WL-HPV58 L1 at a detectable level but none of that protein was soluble (Fig. 1). Overall expression of L1 protein was highest from the WS-HPV58 L1 gene but the proportion of soluble L1 protein was significantly lower than for the MO-HPV58 L1 gene (Fig. 1). Expression of soluble L1 protein was highest from the MO-HPV58 L1 gene, despite the low level of total L1 protein produced. These results suggest that the nucleotide sequence between the first and second ATGs hinders the expression of L1 protein, and that the codon composition of the wild type L1 gene of HPV58 also reduces soluble L1 protein production.

High levels of non-functional protein are associated with increased cellular stress responses (Mattanovich et al. 2004). In the endoplasmic reticulum (ER), the quality of newly synthesized protein is controlled by post-translational modifications such as glycosylation and disulfide bond formation (Feige and Hendershot 2011; Hulsmeier et al. 2011). Failure of quality control for newly synthesized protein results in accumulation of misfolded proteins, which in excess causes endoplasmic reticulum (ER) stress (Ron and Walter 2007). The BiP protein derived from the KAR2 gene is a well-known marker of ER stress in yeast (Lajoie et al. 2011; Lindl et al. 2007; Mattanovich et al. 2004). We found that the BiP level in cells transformed with MO-HPV58 L1 genes was significantly lower than in cells transformed with WL- or WS-HPV58 L1 genes (Fig. 6). Based on these findings, it is likely that the codon optimization of the MO-HPV58 L1 gene reduces cell stress by decreasing the amount of misfolded protein produced, and the reduced cell stress enhances the performance of the cellular machinery.

Codon usage frequency and secondary structure of mRNA have been thought as critical factors to affect the expression levels of heterogeneous genes. CAI value of WS-HPV58 L1 gene for yeast is higher than those for human and mammals, indicating that the codon composition of L1 gene is relatively favorable for expression in yeast: CAI of WS-HPV58 L1 gene for yeast is 0.83, while those for human and rat are 0.73 and 0.67, respectively. Therefore, it was thought that there is little margin for increasing L1 protein expression in codon usage optimization for yeast. In this study, we used modified L1 DNA designed to reduce the mRNA secondary structure and confirmed that the strategy is useful to obtain the soluble L1 protein.

There have been numerous studies aimed at enhancing protein expression using codon optimized genes. However, there has been little research on codon optimization in S. cerevisiae. Our results suggest that the proportion of soluble protein and amount of final product should be taken into account in codon optimization for S. cerevisiae. We found that the codon optimized MO-HPV58 L1 gene enhanced the expression of soluble L1 protein in S. cerevisiae, and that this protein could be purified and has potential for use as a prophylactic vaccine.

Supplementary material

10529_2012_1097_MOESM1_ESM.doc (134 kb)
Supplementary material 1 (DOC 134 kb)

Copyright information

© Springer Science+Business Media Dordrecht 2012