High-level expression of a novel liver-targeting fusion interferon with preferred Escherichia coli codon preference and its anti-hepatitis B virus activity in vivo

Background In our previous study, a novel liver-targeting fusion interferon (IFN-CSP) combining IFN α2b with plasmodium region I peptide was successfully constructed. IFN-CSP has significant inhibition effects on HBV-DNA replication in HepG2.2.15 cells. The aim of the present investigation was focused on how to produce high levels of recombinant IFN-CSP and its in vivo anti-hepatitis B virus (HBV) activity. Methods A modified DNA fragment encoding IFN-CSP was synthesized according to Escherichia coli (E. coli) preferred codon usage and transformed into E. coli BL21 (DE3) for protein expression. The induction conditions were systematically examined by combining one-factor experiments with an orthogonal test (L(9)(3)(4)). The antigenicity of the purified protein was characterized by western blot analysis. The in vivo tissue distribution were assayed and compared with native IFN α2b. HBV-transgenic mice were used as in vivo model to evaluate the anti-HBV effect of the recombinant IFN-CSP. Results The results showed that the E. coli expression system was very efficient to produce target protein. Conclusion Our current research demonstrates for the first time that IFN-CSP gene can be expressed at high levels in E. coli through codon and expression conditions optimization. The purified recombinant IFN-CSP showed liver-targeting potentiality and anti-HBV activity in vivo. The present study further supported the application of IFN-CSP in liver-targeting anti-HBV medicines.


HBV transgenic mice
Background Over the past decade, a number of functional proteins have been produced successfully by recombinant DNA methods [1,2]. The most commonly used host cell is Escherichia coli (E. coli) because of high yield and low cost [3]. However, high-level expression of recombinant proteins using E. coli as the host cell has been affected by some factors, such as the different preferential codons in different expression systems and various induction conditions of target protein [4,5].
The interferons (IFNs) are a family of protein with the ability to induce antiproliferative, immunomodulatory, and antiviral activities [6][7][8]. For more than a decade, interferon therapy is the gold standard in treatment for certain forms of viral hepatitis and carcinogenesis [9]. However, therapeutic efficacy has been limited because interferon does not have organ-specific affinity and its in vivo half-life is short [10]. Incorporation of plasmodium region I peptide was demonstrated to be a promising strategy for the development of liver-targeting drug [11,12]. In our previous study, a novel liver-targeting fusion interferon (IFN-CSP) combining IFN α2b with plasmodium region I peptide was successfully designed [13]. The results of in vitro anti-hepatitis B virus (HBV) activity of IFN-CSP showed that IFN-CSP has significant inhibition effects on HBV-DNA replication in HepG2.2.15 cells [14]. In vitro liver tissue binding analysis revealed that IFN-CSP specific targeting to liver tissue [13]. IFN-CSP may be an excellent candidate as a liver-targeting anti-HBV agent. However, the in vivo tissue distribution and anti-HBV activity of IFN-CSP requires further investigate. Moreover, it is desirable to produce IFN-CSP in a large scale for therapeutic application.
The main purpose of the present investigation was focused on how to produce high levels of recombinant IFN-CSP in a cost-effective way. We also investigated the in vivo tissue distribution and anti-HBV activity of IFN-CSP. Therefore, a modified DNA fragment encoding the IFN-CSP was synthesized by overlapping extension-PCR method according to E. coli preferred codon usage. The IFN-CSP gene was cloned into the bacterial expression vector pET-21b and transferred into the expression strain E. coli BL21. The suitable induction conditions were systematically optimized by combining one-factor experiments with an orthogonal experiment (L(9)(3)(4)). The antigenicity of the purified protein was characterized by western blot analysis. The in vivo tissue distribution were assayed and compared with native IFN α2b. HBV-transgenic mice were used as in vivo model to evaluate the anti-HBV effect of the recombinant IFN-CSP.

Methods
Pasmids, strains and culture media pMD20-T (Takara, Japan) was used for gene cloning. pET-21b (Novagen, USA) was employed to construct expression vector. E. coli strain DH5α (Novagen, USA) was applied as the host for gene manipulation. E. coli strain BL21 (DE3; Novagen, USA) served as expression host for fusion protein. Luria-Bertani (LB) medium was used for bacterial growth and protein expression.
Construction of the liver-targeting fusion interferon gene using E. coli preferred codon To improve the expression level of IFN-CSP in E. coli, a new IFN-CSP coding sequence was designed. The optimized codons usage pattern of E. coli genes was employed according to the codon usage pattern of E. coli (http:// www.kazusa.or.jp/codon/). Based on the method of polymerase chain reaction (PCR)-based gene synthesis and gene splicing by overlap extension, a modified three-step method [15], called "splicing by overlapping extension-PCR (SOE-PCR)", was adopted to construct the synthetic IFN-CSP gene. We designed 16 oligonucleotides encoding for the IFN-CSP protein. The sense and antisense oligonucleotides with 19 bp complementary overlapping sequence are presented in Table 1.

Construction of expression plasmids IFN-CSP/pET-21b
The synthetic IFN-CSP gene fragments were cloned into the pMD20-T vector (Takara) and transformed into E. coli DH5a according to the procedures described by the manufacturer. The generated recombinant plasmids IFN-CSP/pMD20-T were digested and the inserts were cloned into Nde I/Xho I restriction sites of the expression vector pET-21b (Fig. 1a). The resulting expression plasmid IFN-CSP/pET-21b was finally transformed into E. coli BL21 (DE3) for IFN-CSP expression.

Optimization of IFN-CSP expression
To improve the expression level of IFN-CSP, the induction conditions [16] like cultivation temperature, induction timing, inducer concentrations, induction time were systematically examined by combining one-factor experiments with an orthogonal test (L(9)(3)(4) Purification, antigenicity and biological activity analysis Cells were collected by centrifugation. After lysis by ultrasonication in an ice bath, an improved seven-step process was conducted to obtain purified IFN-CSP based on previously described methods [13,15,17]. Briefly, cellular pellet was separated from the cell lysate by centrifugation at 6000 g and 4°C for 10 min. IB material was washed with 1 % Triton X-100 and 2 M urea (containing 2 % deoxycholate), and then dissolved in 6 M guanidine hydrochloride (GuHCl) (containing 2.5 mM DTT and 50 mM Tris-HCL buffer, pH 8.0). The solubilized IFN-CSP was refolded by dialysis with the 50 mM Tris-HCl buffer systems (containing 0.2 mM glutathione oxidized and 2 mM glutathione reduced), followed by changes of the buffer containing decreasing GuHCl concentrations (4, 2 and 1 M) and finally with Tris-HCl buffer (pH 7.4). The refolded IFN-CSP was purified by heparin affinity chromatography according to the instruction of HiTrap affinity column (GE healthcare, USA). The IFN-CSP was eluted with a linear gradient of 0.1-2 M NaCl. After dialysis against the Tris-HCl buffer, the sample was further purified to remove endotoxin by polymyxin B column (Bio-Rad, USA). After detecting the lipopolysaccharides (LPS) content with the chromogenic limulus amoebocyte lysate assay (Associates of Cape Cod, USA), the purified IFN-CSP was lyophilized and stored at -80°C. The antigenicity of the purified protein was characterized by western blot analysis. We performed western blot with primary goat polyclonal anti-IFN α antibody (1:200; Santa Cruz Biotechnology, USA) and peroxidaseconjugated rabbit anti-goat IgG (1:2500; Santa Cruz Biotechnology, USA) as previously described [13]. The purity of the purified protein was assessed by reverse phase high-performance liquid chromatography (RP-HPLC) on a C18 column (250 × 4.6 mm, 5 μm, and 300 Å, Agilent, USA) in an analytical Alliance HPLC System (Waters, . The molecular weight of purified IFN-CSP was characterized by mass spectrometry on a matrix-assisted laser desorption/ionization (MALDI) mass spectrometer (Applied Biosystems, USA).
Biological activity of IFN-CSP was determined in a standard cytopathic effect inhibition assay using vesicular stomatitis virus (VSV)/human amniotic cells (WISH) measure system according to China Biologicals Requirements.

In vivo tissue distribution experiment
All animal experimental protocols were approved by the Guangdong Pharmaceutical University Animal Care and Use Committee. Normal Balb/c mice (18-22 g) were purchased from the Center for Experimental Animals of Guangdong Province (Guangzhou, China). Balb/c mice (n = 90) were injected via the tail vein with 9.01 μg/kg (0.46 nmol/kg) of control IFN α2b or 10.09 μg/kg (0.46

In vivo anti-HBV experiment
Balb/c-HBV transgenic mice (18-22 g) were purchased from Infectious Disease Center of No. 458 Hospital (Guangzhou, China) and kept under the pathogen free condition in the facility of the Guangdong Pharmaceutical University. The Balb/c-HBV transgenic mice were randomly assigned to either treatment or control groups (n = 6). Six non-transgenic Balb/c mice were used as normal control mice. Each mouse was intramuscular injected with IFN-CSP (at dose of 10 1 U/g body weight, 10 2 U/g body weight, 10 3 U/g body weight, respectively) once a day for 28 days. Sterile physiological saline was used as control. IFN α2b (10 3 U/g body weight) was used as drug control. Mice were sacrificed at day 28. Serum samples and livers were collected.

Measurement of serum HBsAg and HBV-DNA
Serum HBsAg was measured by enzyme-linked immunosorbent assay (ELISA) using a HBV diagnostic kit (Shanghai Kehua Biotech Co. Ltd., China) following the manufacturer's recommendations. Serum HBV-DNA was measured using a commercially available real-time fluorescent quantification PCR (FQ-PCR) diagnostic kit (Da-An Gene Co., Guangzhou, China) according to the manufacturer's instruction.

Immunofluorescence and western blot analysis for HBV core protein
Tissues were fixed in 4 % paraformaldehyde for 1 h, then in 30 % sucrose solution overnight for cryoprotection. 10-12 μm thick sections were cut using a Leica cryostat. Immunohistological detection of HBcAg was performed on frozen sections using primary rabbit polyclonal anti-HBcAg antibody (1:200; Abcam, England) and Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:200; Jackson ImmunoResearch, USA). Finally, the sections were stained with 4′,6-diamidino-2-phenylindole (DAPI) for nuclear indication. Immunofluorescence analyses were performed with a fluorescence microscope (Leica, German). HBV core protein was also analyzed by a standard western blot procedure using primary polyclonal anti-HBcAg antibody (1:200; Abcam, England), anti-actin monoclonal antibody (1:1000; Santa Cruz Biotechnology, USA) and peroxidase-conjugated secondary antibodies. The image was digitized using a scanner and signal was quantified using of Quantity One software (Bio-Rad).

Statistical analysis
Each measurement was performed at least in triplicate. All data were expressed as the mean ± standard error of the mean (SEM). All statistical analyses were performed by SPSS (version 13.0 for Windows) statistical software. Differences between mean values were analyzed using One-Way Analysis of Variance (ANOVA). Statistical significance was defined by a P value of less than 0.05.

Results
Construction of the liver-targeting fusion interferon gene using E. coli preferred codon To create a modified IFN-CSP gene with enhanced translation efficiency compared to the native gene, we designed a codon-optimized version of the IFN-CSP gene (GenBank Accession No. KP027473), which the rare codons of E. coli were changed to the preferential codons (Fig. 2). The new modified IFN-CSP gene has the identical amino acid sequence as the native gene. To increase the translation termination efficiency, the preferred stop codon TAAT was used. Finally, restriction enzyme sites for Nde I and Xho I were introduced at the 5′-and 3′-ends of the IFN-CSP gene to provide convenient restriction sites for cloning into the bacterial expression vector. Using SOE-PCR method, the synthetic IFN-CSP gene with the optimized codons was obtained (Fig. 3) and identified by DNA sequencing.

Plasmids construction and protein expression
The above synthetic IFN-CSP gene fragment was cloned into the bacterial expression vector pET-21b to construct the expression plasmid IFN-CSP/pET-21b (Fig. 1a). The correct synthetic sequence in the recombinant plasmid IFN-CSP/ pET-21b was identified by restriction enzymes with Nde I/Xho I, showing the expected about 600 bp DNA fragment (data not shown). The plasmid was then transformed into the bacterial expression host E.coli BL21 (DE3) to obtain recombinant strain. The target protein, which about 21.5 kD, was successfully expressed (Fig. 1b) in the form of inclusion bodies (IB).

The effects of different expression conditions
Several one-factor experiments and an orthogonal test (L(9)(3)(4)) were employed to optimize the induction conditions.

Cultivation temperature
It is well-known that temperature has a great impact on the recombinant protein expression. Effects of temperature on IFN-CSP expression were investigated at 17, 22, 27, 32, 37 and 42°C, respectively. The results (Fig. 4) revealed that IFN-CSP achieved a high percentage in total proteins and a high concentration in the temperature range of 32-37°C.

Induction timing
IPTG induction initiates the translation of heterologous protein. The effect of induction timing was evaluated by adding IPTG at different stages of growth phase. The results (Fig. 5) revealed that the percentage of target protein to total proteins varied from 20.12 to 61.62% at different induction timing. The maximum percentage was observed when induced at the midexponential phase, corresponding to the OD 600 value of 0.4-0.8.

IPTG concentration
Effects of final IPTG concentration on IFN-CSP expression were investigated at 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, 1.2 and 1.5 mM, respectively. The results (Fig. 6) revealed that the percentage of target protein in the total proteins and the concentration of IFN-CSP achieved the highest level in the IPTG range of 0.4-0.8 mM.

Post-induction time
The results of optimal post-induction time (Fig. 7) revealed that the percentage of IFN-CSP to total proteins varied from 32.55 % to 60.89 % at different post-induction time. The maximum percentage was observed the product reached the highest concentration after a 6-10 h induction.

Orthogonal test
Based on the results of one-factor experiments, four parameters were further optimized by an orthogonal test (L(9)(3)(4)). The result of orthogonal test was showed that  the optimal condition of maximum expression quantity was inducing the culture at OD 600 = 0.6 with 0.8 mM IPTG for 6 h at 37°C. The target protein was up to 2.17 g per L culture (Fig. 8). Main influence factor was induction temperature (data not shown).

Purification, antigenicity and biological activity analysis
The total improved isolation scheme was showed in Fig. 9. The results of SDS-PAGE and RP-HPLC revealed that the target protein was purified to over 95 % homogeneity with no degradation (Fig. 9a and c). The molecular weight 21,481.76 Da determined by MALDI-MS (Fig. 9d) was consistent with its calculated value of 21,481.7 Da.
Step yields were calculated and showed in Table 2. The improved isolation scheme were highly efficient in producing pure IFN-CSP, approximately 690 mg of the pure recombinant IFN-CSP was obtained from 1 L of E. coli culture.
The results of western blot analysis revealed that purified IFN-CSP was strongly and specifically reacted with the IFN α antibody (Fig. 9b). The results of chromogenic  limulus amoebocyte lysate assay showed that the LPS content in the purified sample was less than 0.5 Eu/mg proteins. Biological activity analysis using the WISH/ VSV system showed that the above purified sample had ability to inhibit the cytopathic effect caused by VSV on WISH cells (3.78 × 10 8 U/mg).

In vivo tissue distribution
To determine whether IFN-CSP was able to accumulate in liver for potential treatment of hepatitis B, the tissue distribution of IFN-CSP were evaluated and compared to native IFN α2b. The ELISA method for IFN α2b detection is feasible and accurate (data not shown). The IFN-  The core protein was also quantified by western blot analysis. c: Optical densities of the core protein were analyzed using Quantity One software. The data are the mean ± SEM from six mice. *P < 0.05 vs physiological saline controls, **P < 0.01 vs physiological saline controls, # # P < 0.01 vs IFN α2b controls CSP increased the liver IFN α2b levels more significantly, up to 5.48-to 3.22-fold higher than those of native IFN α2b (p < 0.01) at 30, and 60 min interval postadministration (Fig. 10).

IFN-CSP reduces serum HBsAg and HBV-DNA in HBV-transgenic mice
To evaluate the in vivo anti-HBV activity of the recombinant IFN-CSP, Balb/c-HBV transgenic mice were treated with different concentrations of IFN-CSP for 28 days. As shown in Fig. 11, IFN-CSP reduces serum HBsAg and HBV-DNA in HBV-transgenic mice in a dose-dependent manner. Compared with the native IFN α2b group, the administration of IFN-CSP at comparable concentrations significantly reduced serum HBsAg and HBV-DNA.

IFN-CSP reduces HBV core protein in the liver of HBV-transgenic mice
Livers from HBV-transgenic mice were detected with anti-HBcAg antibody by immunofluorescent staining and western blot analysis. The results (Fig. 12) showed that the signal of HBV core protein is obviously decreased in the liver of HBV-transgenic mice treated with IFN-CSP in a dose-dependent manner. IFN α2b control also suppressed the expression of HBV core protein, but the degree of suppression less than the same dose of IFN-CSP.

Discussion
The expression of heterologous proteins in E. coli is an attractive strategy to obtain the active form of protein in a large scale for therapeutic application. In our previous work, the strategy proved to be efficient to achieve highlevel expression [15,17]. While, DNA sequence analysis of genes encoding either high abundance or low abundance proteins in E. coli has revealed a pattern of favored codon usage [18]. Lowly expressed genes show the greatest degree of conformity to the rare codons, which correspond to the rare tRNAs in the E. coli cell. So, it is necessary to analyze the DNA sequence of IFN-CSP before recombination expression using E. coli expression system. The IFN-CSP contains 184 amino acids, and 15 amino acids are encoded by the rare codons of E. coli. Particularly, R12, R13, R22, R23, R33, R120, R125, R144, R149, and R162 were encoded by the least used codons, AGG and AGA, which were discovered to dramatically reduce the maximum level of protein synthesis. Some examples show that codon optimization can significantly improve the expression level [5,19]. Srivastava et al. reported that a codon optimized IFN α2b gene expressed as IB only about 40 % of the total protein [20]. Valente et al. using an artificial IFN α2b gene that was codon optimized for its rare tandem arginine codons and its gene product was produced 16.6 mg/L E. coli culture [21]. Another investigation used E. coli codon optimized artificial gene produced His-IFN α2b in the form of IB -210 mg/L E. coli culture [22]. Based on the above knowledge, we applied a codon optimization strategy to over-express IFN-CSP fusion protein in E. coli. Moreover, stop codon bias and context also correlates with the efficiency of translation [23], so the preferred sequence of translation termination in E. coli, 'TAAT' was used.
The induction conditions will affect the expression level of target proteins as reported numerous times elsewhere [16,24]. For this reason, the induction conditions like cultivation temperature, induction timing, inducer concentrations, and induction time were also examined. In our previous study, we found that the fusion IFN-CSP without any optimization expressed as IB only accounting for 39.2 % of the total insoluble protein [13]. In the present, the highest expression level was achieved when the recombinant E. coli at the mid-exponential phase was effectively induced at 37°C with 0.8 mM IPTG concentration for 6 h. Although most of the IFN-CSP protein in the current work was produced in the form of IB (Fig. 1b), this is the first report to show that the novel liver-targeting fusion interferon (IFN-CSP) was expressed in such high level, about 61.62 % relative to total host proteins. Using the improved purification scheme (Fig. 13), the final yield of purified IFN-CSP was up to 690 mg per L culture, which is 2.56-fold higher than the previous investigation [13]. The purity and the molecular weight of purified IFN-CSP were demonstrated by SDS-PAGE, RP-HPLC and MALDI-MS (Fig. 9). The result of western blot analysis characterized the antigenicity of IFN-CSP (Fig. 9b). Biological activity analysis according to China Biologicals Requirements showed that IFN-CSP had ability to inhibit the cytopathic effect caused by VSV on WISH cells (3.78 × 10 8 U/mg).
In line with previous in vitro experiments, we have further evaluated the in vivo tissue distribution and anti-HBV activities of IFN-CSP. In vivo tissue distribution assay showed that IFN-CSP was able to accumulate in mouse liver compared to native IFN α2b (Fig. 10). Small animal models amenable to infection with human hepatotropic viruses are in great need for studying new anti-HBV drugs. The HBV-Tg mouse lineage was initially produced on a BALB/c background. The transgene in these mice consists of 1.3 copies of the HBV ayw complete genome. The HBV-Tg mice express high level of HBsAg in their serum and have detectable HBV-DNA in their serum, which allows HBV-DNA replication, is an appropriate tool to evaluate anti-HBV drugs [25][26][27][28]. In the present study, we found that the recombinant IFN-CSP and native IFN α2b in vivo were able to reduce serum HBsAg (Fig. 11a), HBV-DNA (Fig. 11b) and liver HBV core protein (Fig. 12) in HBV-transgenic mice. However, IFN-CSP showed a significant improvement in anti-HBV activity when compared with native IFN α2b at similar doses. These results indicate that incorporation of plasmodium region I peptide can specifically target to liver, to achieve higher local concentrations and to improve their anti-HBV efficacy in vivo.

Conclusion
In this report, synthetic codon optimized IFN-CSP gene was overexpressed at high expression level and the purified protein was confirmed to IFN-CSP based on the results of antigenicity and biological activity analysis. The purified IFN-CSP was able to accumulate in mouse liver compared to native IFN α2b and reduce serum HBsAg, HBV-DNA and liver HBV core protein in HBV-transgenic mice. This current work reports for the first time that codon and expression conditions optimized approach is a strategy to produce IFN-CSP at large scale, and would also be very helpful to produce other recombinant protein for further medical research. The present study further supported the application of IFN-CSP in liver-targeting anti-HBV medicines.