Most IgG molecules have a symmetrical structure consisting of four polypeptide chains including two heavy chains (HC) and two light chains (LC) (Figure 3). In the endoplasmic reticulum, the antibody binding protein (BiP) briefly binds to the heavy-chain polypeptide of the antibody, and the light chain is secreted out of the cell as polymerized dimers. Before assembling into a tetramer, BiP maintains a binding state with the heavy-chain polypeptide. Once the light-chain polypeptide is lacking, the heavy chain will not be secreted outside the cell, causing the proteasome in the cell to be unable to degrade the heavy-chain polypeptides secreted to extracellularly; thus, the unit productivity of antibody is significantly reduced (Vanhove et al., 2001; Ha et al., 2019). Therefore, excess light-chain polypeptides should be ensured to improve the expression of antibodies. In general, in the traditional process of antibody preparation, researchers often construct light- and heavy-chain genes of antibodies into two vectors to further express the antibody molecules. The disadvantage of this method is that the ratio of light and heavy chains is not controllable. In addition, the ratio of light and heavy chains affects the quality of monoclonal antibodies, such as the formation of aggregates and glycosylation modification. Chung et al. confirmed that excessive heavy-chain polypeptides may be the main cause of polymerization (Chung et al., 2018). In transient expression, the ratio of polypeptides can be controlled by adjusting the relative amount of each vector. However, this method is difficult to operate because the integration of plasmids is often random, and the number of gene copies and integration sites cannot be controlled artificially (Schlatter et al., 2005). Therefore, the industrial production of antibody drugs requires stable transfection of cell lines and proper control of the ratio of light and heavy chains, which are particularly important to achieve high expression, low polymerization, and consistent N-glycosylation profile of antibodies.

The heavy- and light-chain genes of antibodies use different vectors and then enter the host cells by co-transfection. This method has very low requirements on the vector, but the transfection efficiency is high, which is a common strategy for recombinant antibody expression (Carrara et al., 2021). However, the main defect of this strategy lies in the random integration of light- and heavy-chain genes into chromosomes, and the insertion positions and numbers are different, resulting in uncontrollable expression levels of light and heavy chains (Ahmadi et al., 2016). The second arrangement is to construct an independent expression unit of light and heavy chains on the same vector; the light and heavy chain genes use separate promoters and polyadenylic acid (Poly A) tail sequences of different lengths. This strategy can theoretically achieve equal mole expression levels of light- and heavy-chain genes. However, such adjacent reading frames may cause transcriptional interference, resulting in the imbalanced transcription of light and heavy chain and ultimately inhibiting the expression of recombinant antibodies (Davies et al., 2011). Based on the advantages and disadvantages of the two gene arrangements, researchers adopted a third arrangement, in which an internal ribosome enter site (IRES) or 2A peptide in the same open reading frame to connect light-chain genes, namely, triscistronic expression vector. This strategy can effectively avoid the occurrence of transcription interference and control the proportion of light and heavy chain expression. When expressing monoclonal antibodies using IRES-mediated tricistronic vector, the IRES connects heavy-chain genes, light-chain genes, and resistance genes for expression on the same vector. This strategy can maintain over 70% productivity of positive clones and 2-fold increase in the yield of recombinant antibodies (Li et al., 2007; Yeo et al., 2018). Ho et al. designed four IRES-mediated tricistronic vectors to control the ratio of light and heavy chains, depending on the different positions of the light and heavy chains and selection marker genes on the vector; they also compared the expression level and quality of antibodies, including aggregate formation, N-glycosylation, and conformational stability. The results show that excess light chain is essential for high expression of antibody and also reduces the occurrence of polymerization. In addition, the main cause of polymerization is that the excessive heavy chains cannot effectively fold with the limited light chains and can change the N-glycosylation and reduce the conformational stability (Ho et al., 2013b). Controlling the ratio of light and heavy chains is crucial for the expression and quality of monoclonal antibodies under stable transfection conditions (Ho et al., 2012). In particular, excessive light chains cannot only achieve high antibody expression but also reduce polymerization and maintain low fragmentation levels, which are very beneficial for obtaining high-yield antibodies (Ho et al., 2013b). However, the fragment length of IRES itself is relatively large, which occupies more space, and the expression of IRES upstream and downstream antibody molecules is seriously imbalanced (Bayat et al., 2018). In addition, the activity of IRES is not easy to adjust, which will affect the biological activities of other expressed proteins. In recent years, a prominent approach is based on self-cleavage 2A peptide-mediated multi-gene construction method (Chng et al., 2015; Van der Weken et al., 2019). This strategy effectively avoids the disadvantage of low expression of downstream genes caused by the simultaneous expression of multiple genes and is commonly used in multi-gene expression (Luke and Ryan, 2018; Li et al., 2020). Ho et al. compared the effects of F2A and IRES on the production of a monoclonal antibody in CHO-DG44 cells. The expression level of the monoclonal antibody produced by F2A-mediated tricistronic vector was significantly higher than that of IRES-mediated vector under transient or stable transfection conditions; the expression level of the recombinant antibody was affected by the position of light- and heavy-chain cistrons (Ho et al., 2013a).

Antibody molecules usually have the molecular structure of tetramer glycoprotein. Light- and heavy-chain genes in antibody molecules need to have coordinated expression to biosynthesize tetramer IgG. However, when expressing recombinant antibodies, the special secondary structure in the gene sequence and other factors often lead to low expression of antibody, which can be remedied by codon optimization strategy. Codon optimization involves gene synthesis, gene transcription, mRNA translation, etc., with the ultimate goal of efficiently expressing recombinant antibodies (Ayyar et al., 2017; Mauro and Chappell, 2018; You et al., 2018). Before expressing a gene, the rare codon in the gene should be searched first. If there are too many rare codons, then the translation rate of antibody will be affected. The expression of the target gene in transgenic hosts can be increased by selecting the codon preferred by the receptor without changing the amino acid sequence (Mauro, 2018). When the codon of the variable region of an antibody is replaced with the preferred codon of the natural human antibody gene, the antibody expression levels in mammalian cells are significantly increased by two- to 3-fold (Carton et al., 2007).

Signal peptide is a key factor for the secretion of recombinant antibodies, and the high-efficiency expression of these antibodies is closely related to the signal peptide (Haryadi et al., 2015; Wang et al., 2016). When expressing recombinant antibodies, in addition to the use of their own signal peptides, the following strategies are often applied: ① replace the signal peptide sequence of the efficiently expressed secreted proteins (Attallah et al., 2017); ② modifying the primary structure of the signal peptide sequence in the original antibody; ③ replace the protein signal peptide sequence in some viruses; however, viral vectors are generally not recommended when considering the safety of the expressed recombinant antibodies; and ④ select the preferred host codon. When selecting a suitable signal peptide, the preferred codons of the expression host should be fully considered, and the signal peptide should be further optimized. Ramezani et al. showed a 2-fold increase of the pertuzumab production in CHO cells by optimizing codons and selecting appropriate signal peptide strategies (Ramezani et al., 2017).

The integration site status of target gene on the mammalian cell chromosome plays a decisive role in its expression level and its stability in host cells. Only clones formed by cells whose integration sites are in transcription active region of chromosomes can express the target gene at a high level. However, transgene silencing often occurs due to the random integration of the target gene after transfer into cells (Hilliard and Lee, 2021). During the expansion of cell culture, the promoter methylation will lead to the attenuation of expression. Therefore, selecting an appropriate promoter and optimizing the combination of promoters and different regulatory elements can improve the expression of recombinant antibodies and increase the stability of expression. Some chromatin-modifying elements can prevent transgene silencing, including matrix attachment regions (MAR) (Buceta et al., 2011; Mohammadian et al., 2019; Zhang et al., 2020), locus control regions (LCR) (Sharma et al., 2019; Morgan et al., 2020), ubiquitous chromatin opening elements (UCOEs) (Pfaff et al., 2013; Harraghy et al., 2015; Nematpour et al., 2017; Rocha-Pizaña et al., 2017), stabilizing anti-repressor elements (STAR) (Kwaks et al., 2003; Otte et al., 2007; Van Blokland et al., 2007), and insulators (Benabdellah et al., 2014; Chetverina et al., 2014; Naderi et al., 2018; Pérez-González and Caro, 2019). Moreover, artificial chromosome expression (ACE) and targeted integration technology, including Flp-In and recombinase-mediated cassette exchange system (RMCE), can overcome the shortcomings of random integration (Kennard et al., 2009; Soler et al., 2018; Reinhart et al., 2019; Ng et al., 2021). The introduction of these functional elements into the vector construction can greatly increase the proportion of high-expression clones and shorten the construction cycle of engineered host cells (Table 2).

Transcription is the initial step of gene expression, among which promoters, enhancers, and transcription termination signals have very important impacts on transcription efficiency and mRNA stability. CMV promoters are considered to be one of the strongest viral promoters, and hEF-1α promoters have the stronger transcription initiation efficiency and are more suitable for large-scale production of recombinant antibodies (Ebadat et al., 2017). In recent years, the construction of synthetic or heterozygous promoters will become an effective strategy to improve the transcription efficiency (Patel et al., 2021).

Enhancers can improve the transcription efficiency and can function over long distances. Constructing heterozygous enhancers is a good strategy to improve transcription efficiency and obtain high transcriptional activity. Xu et al. used the CMV promoter and CA hybrid promoter to insert the SV40 enhancers downstream of SV40 Poly A; compared with the CMV promoter, cells driven by the CA hybrid promoter can increase the production of target proteins by two times (Xu et al., 2001).

Introns can increase the expression of foreign genes. Xu et al. compared the effects of five different introns on transgene expression in CHO cells. Under transient and stable transfection conditions, the SV40 intron can obtain the highest transgene expression level among five introns, which can also obtain high-level of recombinant protein production in CHO cells (Xu et al., 2018).

The regulation of translation level and processing efficiency of translation products will also significantly affect the expression of target gene. Eisenhut et al. developed the 5′-untranslated region (UTR) RNA-structures to impact translation efficiency, further systematically tune protein expression levels in mammalian cells and eventually help to optimize recombinant protein expression (Eisenhut et al., 2020). Vivirius et al. found a universal translation enhancer element in the 5′-end non-translation area of Hsp70 mRNA, which can enhance the translation efficiency of the cap-dependent structure (Vivinus et al., 2001).

At present, the two common gene amplification screening systems are dihydrofolate reductase (DHFR) and glutamine synthetase (GS) (Budge et al., 2021; Huhn et al., 2021). The CHO-DHFR amplification system mainly uses the CHO cell line of DHFR defect to obtain high yields of recombinant antibodies in the presence of methotrexate (MTX). Akbarzade-sharbaf et al. used DHFR system to express a therapeutic antibody (Trastuzumab); after seven rounds of MTX, the total antibody valence can be as high as 50–60 mg/L/day (Akbarzadeh-Sharbaf et al., 2013). As another mature and widely used system, the GS amplification system is an explicit gene amplification selection marker, which usually requires pressure and stable selection of two rounds of methionine sulphoximine (MSX) to obtain high-expression cell lines and reduce the screening time. However, the two systems have some defects, such as the decline in the antibody production after long-term screening; therefore, these systems need to be further optimized. The strategy of using low activity of GS as the screening marker could obtain high-yield clones after MSX removal, further efficiently increase the antibody production (Lin et al., 2019).

Two strategies can be employed to weaken the screening marker. One is to mutate the screening marker and reduce its activity. Neomycin-phosphotransferase (NPT) is one of the commonly used screening markers in eukaryotic expression systems. The second strategy is to reduce the expression of the screening marker. Noguchi et al. started the transcription of DHFR through the late promoter of weak promoter SV40 and used the method of IR/MAR-dhfr fusion to easily separate the cell line of high-expression recombinant protein, which further improved the production of the recombinant proteins (Noguchi et al., 2012).

The enzymes responsible for glycosylation can be modified in non-humanized cells, so the expressed recombinant antibody is more similar to the natural humanized protein in terms of glycosylation level and type, thereby improving the biological effect of recombinant antibody. Glycosylation mainly includes genetic and non-genetic modification strategies. The glycosylation of protein can affect the pharmacological activity and pharmacodynamics of the antibody, so researchers can optimize the recombinant antibodies through genetic engineering to achieve the glycosylation modification of cell lines (Yang Y et al., 2021). For example, the therapeutic activity of recombinant IgG3 antibodies was significantly improved after transfecting murine 2, 6-sialyltransferase into CHO cells (Jassal et al., 2001).

Anti-apoptosis is a hot spot in the current host cell transformation strategy, and the overexpression of anti-apoptotic genes is one of the common strategies. Effective anti-apoptotic genes include Bcl-2, Bcl-xL, and so on. Lee et al. achieved the overexpression of Bcl-2 and Beclin-1 in CHO-DG44 cells; the results showed that the time of cell culture became longer, the cell survival rate was significantly improved, and the occurrence of apoptosis was suppressed (Lee et al., 2013). Kim et al. overexpressed Bcl-xL in recombinant CHO cells, which can improve cell survival rate, prolong culture time, and inhibit apoptosis and autophagy by inhibiting the activation of caspase-3and caspase-7 (Kim et al., 2009).

In mammalian cell culture, the metabolic engineering strategy changes the cell metabolism pathway, which can effectively promote cell growth and product synthesis, discover more novel metabolite additives, and reduce the accumulation of metabolic inhibitors (Toussaint et al., 2016; Gupta et al., 2017; Fouladiha et al., 2020; Yao et al., 2021). In cells with low lactate/glucose, the expression levels of lactate dehydrogenase and the accumulation of lactate are reduced, resulting in an increase in the synthetic products. Zhou et al. used siRNA technology to reduce the expression levels of lactate dehydrogenase A (LDHa) and pyruvate dehydrogenase kinase (PDHK) genes, reducing the lactate levels by 90% and increasing the production of therapeutic monoclonal antibodies (Zhou et al., 2011). In addition, the production of recombinant antibodies can be multiplied by several times by effectively blocking engineered cells in the G₁ phase.

In the large-scale cell culture process, with the continuous exploration of the regulatory mechanism of cell cycle, researchers have applied cell cycle regulation genes to cell proliferation control. Fussenegger et al. found that p21, p27, and p53 are cell cycle G_l/S suppressor proteins; after CHO cells express these proteins, they can prevent cells from entering the S phase, keep the growth in a static state, and increase the production of secreted alkaline phosphatase (SEAP) to 10–15 times (Fussenegger et al., 1998).

According to different connection modes, antibody glycosylation is usually divided into N-linked glycosylation and O-linked glycosylation. The Fc fragments of two heavy chains in the monoclonal antibody molecule contain an N-glycosylation modification site at the 297th aspartic acid. The N-glycosylation modification of monoclonal antibody drugs usually presents biantennary sugar chains, and sometimes fucosylation and sialylation may occur (Szabo et al., 2022). In addition, some monoclonal antibody drugs undergo O-glycosylation modification. For example, a humanized monoclonal antibody drug produced by CHO cells has O-glycosylation modified by a single glucose molecule (Tanaka et al., 2013). At present, most of the antibody drugs on the market are produced by CHO cells. CHO cells can produce antibodies that are close to the glycotype of human serum antibodies; however, the glycosylation of antibodies produced by most engineered cell lines is different from that of human serum antibodies. For example, antibodies produced from mouse-derived animal cells have a high proportion of fucose modifications and a low proportion of galactosylation modifications (An, 2009). In contrast to human cells, CHO cells lack the expression of α-2,6-sialyltransferase, while only express α-2,3-sialyltransferase (Jenkins et al., 1996; Lin et al., 2015; Chung et al., 2020). Consequently, CHO cells inherently cannot produce glycoproteins with similar terminal sialic acid content as compared to human cells (Yin et al., 2017). Furthermore, CHO cells lack N-acetylglucosamine transferase (GnT-III) expressed by human cells, resulting in the differences in the modification of N-acetylglucosamine glycosylation from human cells (Butler, 2005). These findings indicate that different cell types produce different types of antibody glycosylation (Attallah et al., 2020; Dyukova et al., 2021; Zhao et al., 2021). In summary, the type of glycosylation modification of monoclonal antibody drugs is closely related to the production system, selected cell line, and incubation process.

Although the mass of sugar chains only accounts for 2% of the total molecular weight of antibody, the sugar chains on the Fc and Fab fragments play a very important role in the affinity, structural maintenance, metabolism, and immunogenicity of the antibody molecule (Krapp et al., 2003; Chung et al., 2008; Alessandri et al., 2012). The physiological activity of therapeutic antibodies is mainly mediated by two mechanisms: one is mediated by the affinity between the variable region of the antibody and antigen, causing the neutralization or apoptosis of the target antigen, which is mainly by means of the Fab fragment of antibody to recognize and bind antigenic substances; the other is the immune effect mediated by the Fc fragment of antibody, including antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) (Salinas-Jazmín et al., 2022). Metabolic analysis and mathematical model can be used to analyze the production process to control the type and degree of glycosylation modification of recombinant antibodies to reduce the immunogenicity of antibody drugs and optimize their effector function (Sha et al., 2016). The types of glycosylation modification of recombinant antibodies are shown in Figure 4.

Sialylation modification: Sialylatedglycans are the key components of glycoproteins. According to relevant studies, monoclonal antibody sialylation can inhibit inflammatory response and reduce cell toxicity by means of different receptors in the Fc fragment (Scallon et al., 2007). The anti-inflammatory activity of the over-sialylated monoclonal antibody obtained by affinity chromatography purification will be significantly enhanced. Moreover, the higher degree of monoclonal antibody sialylation is closely related to a decrease in ADCC activity (Zhang et al., 2019; Nguyen et al., 2020).

Core fucosylation modification: Related studies have shown that core fucose is an important glycosylation structure that affects the ADCC activity of recombinant antibodies (Matsumiya et al., 2007). Most recombinant IgG produced by CHO cells have core fucose in their Fc sugar chains (Zimmermann et al., 2021). To improve the binding activity of the low-affinity receptor IIIa (FcγR3a) of the antibody to the Fc fragment as well as ADCC activity, researchers have adopted a variety of strategies to reduce the level of fucosylation of IgG, including the use of α-1,6-fucose cell lines in which α-1,6-fucosetransferase gene is knocked out (Davies et al., 2001; Wang et al., 2018).

Galactosylation modification: Approximately 95% of recombinant IgG produced by CHO cells contains galactose as the terminal sugar. Terminal galactosylation plays an important role in the conformation of the Fc fragment, and the core fucosylation modification alone has slight effect on the conformation of the Fc fragment (Kotidis et al., 2019). The existing results indicate that galactosylation at the end of the Fc fragment will seriously affect the CDC activity of IgG, and the reduction of its glycosylation level will also weaken the CDC activity (Wei et al., 2021).

Mannosylation modification: The content of a high-mannose structure should be minimized as much as possible during the production of recombinant antibodies due to the high immunogenicity of high-mannose structures. In addition, the content of mannose in the Fc fragment of the antibody varies greatly between different cultured cells and different batches (Brantley et al., 2021). Recombinant antibodies have two clearance pathways in organisms; one is the asialoglycoprotein receptor in the liver that binds and mediates the clearance, and the other is to bind to mannose receptors on the surface of macrophages in the liver. The high-mannose structure antibody molecules in the Fc fragment can be quickly eliminated from the plasma, further reducing the efficacy of recombinant antibodies (Liu, 2015).

Expression of GnT-III: Studies have shown that the bisected glycosylated epidermal growth factor receptor (EGFR) monoclonal antibody is prepared by introducing the GnT-III gene and highly expresses bisecting acetylglucosamine residues; this strategy can increase the ADCC activity by 3 times and increase the anti-proliferative activity by 1.36 times, and almost no α-Gal was detected. The bisecting EGFR monoclonal antibody prepared by glycosylation engineering contains only a small amount of α-Gal, which greatly improves the biological activity in vitro. At present, this study has not been further verified in vivo (Jenkins and Curling, 1994; Yi et al., 2014).