Selection of appropriate signal peptide for antibody expression
Signal peptides play an important role in the secretion of recombinant proteins expressed in mammalian cells (Trosse et al. 2007; Young and Rance 2008). Many studies have shown that optimization of signal peptides could improve the production of recombinant proteins (Haryadi et al. 2015; Kober et al. 2013). Based on these observations, four promising signal peptides (listed in Table 2) previously reported to increase the antibody production (Kober et al. 2013; Luo et al. 2007), were tested in this study.
Suspension culture CHO-S cells were co-transfected with pTT5 expression plasmid containing genes of the heavy chain and the light chain of B45 hAb fused with the signal peptides in Table 1. Samples were taken from cultures every day for productivity testing and cell counting. Maximum productivity for each condition is shown in Fig. 1a. We found that co-expression of 8Hsp-H and 8Ksp-K led to the lowest expression, 30 mg/L. A comparable yield, 40 mg/L was obtained by co-expression of B-H mix with B-K(B-H + B-K) or E-H with E-K(E-H + E-K). However, there were no significant differences in the cell viability and density (Fig. 1b, c).
The cleavage of signal peptide accompanied by the protein secretion is even more important for industrial production than productivity. To predict the cleavage of the signal peptide, the DNA sequences of the kappa chains or heavy chains with signal peptide at N-terminal were submitted to the Signal P 4.1 Server. As seen in Fig. S-2, all signal peptides were predicted to result in a correct cleavage, and all signal peptides could be cut off completely.
In order to verify the cleavage site, the products were purified using protein A and analyzed by reduced CE-SDS separation. For all products (shown in the left column of the Fig. 2), the area of each K-chain was about 34%, and the H-chain was about 65%. As a result, the K/H for each product was 1:1 (mol/mol). The retention time of the signal peak was almost coincident. Subsequently, the light chain and heavy chain were assayed by MS. As shown in the right column of Fig. 2, the K-chain with E signal peptide (Fig. 2a), H-chain with B signal peptide (Fig. 2b), and H-chain with 8Hsp (Fig. 2c) resulted in correct N terminals. For other fusions, there were unwanted amino acid residues from the signal peptide on the N-terminal. For safety, additional amino acid residues are undesirable. Therefore, K-chain with E signal peptide (E-K), H-chain with B signal peptide (B-H), and H-chain with 8Hsp (8Hsp-H) were selected as candidates.
To explore antibody productivity with optimal signal peptide, the E-K were mixed with B-H or 8Hsp-H and were co-expressed in CHOs separately. The 8Hsp-H and E-K combination led to a higher titer, 45 mg/L, than the combination of B-H and E-K (P = 0.005, Fig. 3a). However, there was no significant difference in the viable cell density (Fig. 3b) and viability (Fig. 3c).
To investigate the splicing site of the signal peptide, the products were analyzed by MS. The peptide spectrum from MS confirmed that all chains possessed the correct N-terminal peptide without any additional or missed amino acids (data not shown). Taking into account the productivity and signal peptide cleavage, the combination of K-chain fused with E and H-chain fused with 8Hsp were selected for antibody expression.
Enhance the expression with codon optimization and UTR
The constant region of human IgG 1 was from the human cDNA library, and the codons are biased to human. The variable regions of B45 hAb were from phage library (Luo et al. 2016), and there was no obvious codon bias in the library. A web server, Codon OptimWiz (Genewiz, China), was used to optimize the codon of the variable region and constant region to bias to CHO cell. All sequences are listed in Fig. S-3. The information of sequence-optimized genes is summarized in Table S-1. Compared to the sequence before codon optimization, the codon adaptation index (CAI)s were reached more than 0.95, and the GC content was adjusted to about 60%.
To investigate the productivity enhancement after codon optimization, different vector combinations containing light and heavy chains with default signal peptide (8Ksp and 8Hsp) and optimal codon were co-transfected into CHOs. As expected, the yield was increased by 100% with codon optimization, and the 45KOC made a greater contribution to the yield improvement because co-transfected 45KOC with H led to an equivalent yield to co-transfected 45KOC and 45HOC (Fig. 4a). There were no significant differences in cell density and viability for all tests (data no shown).
Kozak sequence, a typical 5′ UTR, is reported to play a role in efficient initiation of the translation process. The 3′ UTR of murine Ig kappa chain v-j2-c (GeneBank: M35669, 30 bp) was used to stabilize mRNA in numerous studies (Li and Chen 2011). In order to improve the productivity further, a 5′ UTR was added to the 5′ end of the sequence with optimal signal peptide and codon, and a 3′ UTR was added to the 3′ end. A 50% increase was observed from transient expression of the vector with UTR (Fig. 4b), and a correct N-terminal sequence was detected from the product examined by MS (data not shown). So far, an optimal vector with an appropriate signal peptide, codon, and UTR has been constructed and could express antibody with the correct N-terminal sequence and high yield.
The SEC-HPLC and binding activity assay for products from TGE
The TGE culture with vector containing optimized signal peptide, codon, and UTR was purified using protein A and subjected to SEC-HPLC assay. The result is shown in Fig. 5a. There was only one sharp peak with retention time at 15.8 min. After analysis by Empower 3, it was found that the percentage of HMW was 1.2%, the LMW was 0%, and the main peak was 98.8%. This result indicates that the antibody expressed with the optimized TGE method was of a good homogeneity.
To explore the biological activity of the product with optimized TGE, the purified product and standard reference material were tested by CLA for binding activity with HBsAg. The reaction curves of the product with optimized TGE and standard material are coincident (Fig. 5b). Subsequently, CLA data were entered into Prism for calculation of EC50. EC50 of standard material was 0.002108, and EC50 of the product with optimized TGE was 0.001981. These results demonstrate that the optimized TGE method does not affect the biological activity of the product.
Optimization of expression for other antibodies
In order to investigate how application of the TGE optimized strategy could be applied generally, the TGE method was applied for expression of two other antibodies: Herceptin, a humanized antibody against Her2, and 37 hAb, a humanized antibody against H5N1.
The VK and VH sequences of Herceptin were taken from NCBI (Hoerr et al. 2016). The cDNAs of 37 hAb were taken from a phage display library (Luo et al. 2008). The signal peptides in Table 2 were fused to the N-terminal of the variable region for TGE. We found that the highest TGE was achieved by B signal peptide for Herceptin and antibody 37 hAb. The highest yield of Herceptin was 83 mg/L (P < 0.05) and the highest yield of 37 hAb was 64 mg/L (P < 0.05) (Fig. 6a, b). The products were purified and assayed by MS. Fortunately, we found all signal peptides had the correct N terminals (Fig. 6c, d). With the highest yield, B signal peptide was chosen for further study.
Next, the codons of the variable regions of 37 hAb and Herceptin were optimized by Codon OptimWiz. Codon optimization is shown in Fig. S-4. The parameters of optimized codons are summarized in Table S-2. After codon optimization, the codon adaptation index (CAI) increased to more than 0.96, and the GC content was adjusted to about 61%.
Finally, a UTR sequence was added to the 5′ and 3′ of the antibody ORF, and the cDNAs of 37 hAb and Herceptin with UTR were separately inserted into Ptt5 at multiple cloning sites to generate expression vectors. The plasmids of light chain and heavy chain were co-transfected into CHO cells. The yield of TGE for Herceptin post optimization was 2.2-fold higher than that of control, and the yield for 37 hAb was 2.3-fold higher (Fig. 7). The yield of Herceptin was increased from 51 to 109.9 mg/L, while the productivity of 37 hAb was raised from 30 to 69 mg/L after optimization (Fig. 7). There was no significant change in the binding activity of the products (data not shown). These results indicate that TGE with optimized signal peptide, codon, and UTR can improve the expression of antibody, and this strategy could be replicated in expression of other antibodies.
Optimization the expression of stable cell pool
Transient expression may be affected by a variety of factors, and the yield may be varied from lot to lot. To verify if this optimization strategy is applicable in stable gene expression, the genes of the antibody were cloned into the GS expression vector pGS2 (Fig. S-1) to generate stable expression vector. The CHO-K1 were transfected with the stable expression vectors and screened for stable mini pools with MSX. The top 30 mini pool in titer of each transfection were subjected to batch culture. It is exhibited in Fig. 8a that a higher frequency and higher yields were gained post optimization with codon and UTR. The maximum yield of control group (8 Hsp-H + 8 Ksp-K) is 73 mg/L, and the maximum yield of the group with signal peptide optimization (8 Hsp-H + E-K) is 104 mg/L. The highest yield of the mini pools with codon optimization (8 Hsp-HOC + E-KOC) is achieved to 203 mg/L. Furtherly, a yield of 277 mg/L was reached after adding UTR. Figure 8b shows the expression level of top 10 mini pools with fed-batch culture, the mean titer of the control group (8 Hsp-H + E-K) is 557 mg/L, and the average yield of the group with signal peptide, codon, and UTR optimization (UTR + 8Hsp-HOC + E-KOC) is up to 1770 mg/L. The mean values in Fig. 8b is consistent with the values in Fig. 8a. It proves that the optimization of signal peptide, codon, and UTR can also effectively improve the level of stable expression.