Production of Biopharmaceuticals in E. coli: Current Scenario and Future Perspectives

Abstract

Escherichia coli is the most preferred microorganism to express heterologous proteins for therapeutic use, as around 30% of the approved therapeutic proteins are currently being produced using it as a host. Owing to its rapid growth, high yield of the product, costeffectiveness, and easy scale-up process, E. coli is an expression host of choice in the biotechnology industry for large-scale production of proteins, particularly non-glycosylated proteins, for therapeutic use. The availability of various E. coli expression vectors and strains, relatively easy protein folding mechanisms, and bioprocess technologies, makes it very attractive for industrial applications. However, the codon usage in E. coli and the absence of post-translational modifications, such as glycosylation, phosphorylation, and proteolytic processing, limit its use for the production of slightly complex recombinant biopharmaceuticals. Several new technological advancements in the E. coli expression system to meet the biotechnology industry requirements have been made, such as novel engineered strains, genetically modifying E. coli to possess capability to glycosylate heterologous proteins and express complex proteins, including full-length glycosylated antibodies. This review summarizes the recent advancements that may further expand the use of the E. coli expression system to produce more complex and also glycosylated proteins for therapeutic use in the future.

Introduction

The drug discovery and development process entails the expression of large numbers of recombinant proteins in copious amounts and properly folded as 3D structures. These therapeutic recombinant proteins are generally used for the treatment of various diseases, or are used as a target protein for screening of new and novel drugs. E. coli is one of the most desirable host for the expression of several recombinant proteins owing to its rapid growth rate, easier genetic manipulations, and high level of recombinant protein synthesis rates [63, 65]. In a few instances, very high levels of expression are achieved; that is, up to 30% of total cellular protein. Escherichia coli was the first expression host that was used for manufacturing a biopharmaceutical, which resulted in the regulatory approval of human insulin in 1982 for the treatment of diabetes. The approval of bovine growth hormone (bGH) in 1994 set a new standard for manufacturing heterologous proteins from E. coli for therapeutic use (Table 1). It is even more impressive that both insulin and bovine growth hormone require oxidative protein folding, and, in addition, insulin is a heterodimer. Success achieved with recombinant insulin and growth hormone highlights the versatility and cost-effectiveness of E. coli-based production.

Table 1.All data obtained from corporate websites and http://www.fda.gov/. rh, recombinant human; G-CSF, Granulocyte-colony stimulating factor; EU, European union; US, United States of America.

However, in E. coli, there is a probability of translational errors due to the presence of a large number of rare codons in the heterologous gene(s). In the case of therapeutic proteins, these errors, even at low levels, can cause adverse immunogenic responses in humans. These cellular errors during protein translation may impact the tertiary structure and thus affect the biological activity of the recombinant protein. In this review, we will be focusing on the “desired modulation” of key biochemical parameters required to obtain properly folded and overexpressed biologically active therapeutic proteins in E. coli.

 

Codon Usage in E. coli

In E. coli, codon usage is manifested by the level of cognate aminoacylated tRNAs in the cytosol. Major codons are generally present in genes that are expressed at a high level; on the other hand, rare codons are commonly encountered in genes that are expressed at low levels. Codons that are rare in E. coli are found to be abundant in eukaryotic genes [31]. Expression of heterologous genes harboring rare codons can result in translational errors, due to ribosomal stalling at positions where amino acids coupled to rare codon tRNAs have to be incorporated [43]. In addition, translational errors due to the presence of rare codons in heterologous genes might include amino acid substitutions, frame-shift mutations, or premature termination of translation [31, 69]. Kane et al. [30] had reported inframe, two amino acids “hops” at a rare AGA codon. It had also been reported that protein quality is affected to a great extent by codon bias, due to the incorporation of lysine for arginine at the AGA codon [4, 67]. Hence, expression of recombinant proteins, even at very high levels, is of no use if the quality of the protein is compromised as a result of translational errors. The most problematic rare codons in E. coli are AGA, AGG, CGG, CGA, CGG (arginine), AUA (isoleucine), GGA (glycine), CUA (leucine), CCC (proline), and AAG (lysine) [85]. It has been documented that rare arginine codons, AGG and AGA, occur at frequencies of 0.14% and 0.21%, respectively, in E. coli [31].

Several strategies are employed to circumvent the issue of codon bias in E. coli. One approach is to synthesize the whole gene based on codon usage, which is currently a preferred method to improve the expression of heterologous proteins in E. coli [59, 61]. However, a major drawback is the high cost associated with total gene synthesis. Another approach involves the site-directed mutagenesis of the heterologous gene sequence to generate codons, which reflects the tRNA pool of E. coli. However, this process may become very tedious and expensive, whenever several nucleotides need to be modified. Another strategy requires the co-transformation of E. coli strains with plasmid harboring a gene encoding the tRNA cognate to the rare codons [12]. By enhancing the copy number of the limiting tRNAs, E. coli can be manipulated to match the codon usage frequency in heterologous genes. Several commercial plasmids, such as pRARE, are now available for rare tRNA coexpression in E. coli. Moreover, these plasmids contain the p15A replication origin, which enables their maintenance in the presence of the ColE1 replication origin in the E. coli expression vectors. There are also several commercial E. coli strains available, as listed in Table 2, that harbor plasmids containing gene sequences encoding tRNA for rare codons, such as BL21(DE3) CodonPlus-RIL, BL21(DE3) CodonPlus-RP (Stratagene, USA), and Rosetta (DE3). Several studies have employed these excellent strategies to enhance the expression of heterologous proteins in E. coli. [13, 33, 45, 46, 71]. This approach has been successfully used to enhance the industrial production of human recombinant interferon. Co-transformation of E. coli BL21 (DE3) with the plasmid harboring human interferon-α2a gene and the argU gene that encodes for rare tRNAs for Arg (AGG/AGA) resulted in much higher levels of expression of IFN-α2a, and the recombinant protein constituted about 25% of the total proteins [26]. In another study, it was observed that production of human interferon alpha 2b protein was increased 9.5-11.5-fold by replacing the rare arginine codons with more frequently used codons [78]. The E. coli Rosetta (DE3) strain containing plasmid pRARE was used to express different human proteins, and it was shown that the yields of the recombinant protein were increased dramatically for about 35 of the 68 proteins tested [17, 75].

Table 2.Source: http://wolfson.huji.ac.il/expression/bac-strains-prot-exp.html http://www.emdmillipore.com/life-science-research/novagen http://www.invitrogen.com/1/3/stratagene-products https://www.neb.com https://www.lucigen.com https://www.genlantis.com

 

Protein Translation

Efficient translation initiation in E. coli requires a ribosomal binding site that includes the Shine-Dalgarno (SD) sequence and a translation initiation codon [69]. The Shine-Dalgarno sequence is generally located 7-9 nucleotides upstream from the initiation codon AUG [62]. It has been observed that translation initiation is more efficient from mRNAs containing the consensus SD sequence AAGGAGG. The secondary structure of ribosomal binding site is very critical for translation initiation, and the translation efficiency is further increased by the presence of a large number of thymine and adenine [35]. The efficiency of translation initiation is also affected by the nucleotide that follows the initiation codon, and it has been observed that adenine is very common in highly expressed genes [74]. Taken together, translation initiation is affected by various factors, including a consensus Shine-Dalgarno sequence, nucleotides upstream of the initiation codon, and the secondary structure of the ribosomal binding site [73]. It had also been reported that mRNA secondary structures that prevent ribosome binding might affect the protein expression to a great extent [11, 14]. It was observed that a single base change compromised the stability of the secondary structure near the SD region and resulted in a 500-fold change in the expression levels of the coat protein of RNA bacteriophage MS2. Park et al. [55, 68] have developed a very efficient and simple method for designing 5’-untranslated region (5’UTR) variants for tunable expression in E. coli. Since 5’UTR containing the SD sequence and the AU-rich sequence play a significant role in protein translation, the expression levels of recombinant proteins can be manipulated by incorporating simple variations in the 5’UTR. It has been found that secondary structures in mRNA could be disrupted by RNA helicases such as the DEAD protein of E. coli. It was shown that expression of the DEAD protein enhanced the expression of β-galactosidase by 30-fold from the T7 promoter, but there was no significant increase in the expression levels from the lac promoter [22]. However, in the absence of DEAD protein, β-galactosidase synthesis from a T7 promoter was found to be 10-fold less as compared with the expression from the lac promoter, even though the transcription rate was 10-fold higher. It was proposed that the DEAD-box protein plays an important role in stabilizing the mRNA [6, 23, 39]. The DEAD-box protein thus can be exploited to additionally improve the expression of genes with suspected problematic mRNA secondary structures.

In the bacterial genome, UAA is the most frequently used termination codon, followed by UGA and UAG. During translation, an error in reading the termination codon leads to extended protein synthesis until another termination codon is encountered in the mRNA. This read-through results in the synthesis of a larger peptide, with several additional C-terminal amino acids. It had been shown that replacing UGA with the UAA terminator in human IFN-α2b resulted in a 2-fold increase in protein expression level [66]. It had been shown that transcription terminators stabilize the mRNA by creating a stem loop structure at the 3’ end of the mRNA [49]. In E. coli, stop codon UAA is used more commonly for translation termination. The efficiency of translation termination can be improved by adding consecutive stop codons or by using a prolonged UAAU stop codon [57]. Translation errors during protein synthesis can cause frame-shift mutations, premature truncation, lower expression, and misincorporation of amino acid, and thus adversely affect the quality of recombinant protein production in E. coli [3, 56, 67]. High expression levels and correct apparent molecular mass does not always ensure the translation integrity of the recombinant protein [4, 18, 52].

 

Molecular Chaperones to Optimize Protein Folding

One major issue in the production of biopharmaceuticals using E. coli host is the accumulation of heterologous proteins, mainly as insoluble aggregates in the form of inclusion bodies. Extraction of recombinant protein from these inclusion bodies is a tedious and cumbersome process that requires denaturation and several renaturing steps to obtain a soluble and properly folded recombinant protein. One strategy to improve protein solubility in E. coli is the use of molecular chaperones. Inclusion bodies in E. coli comprise misfolded aggregates of heterologous proteins. During protein synthesis, molecular chaperones interact with nascent polypeptide chains to prevent aggregation during the folding process. Some chaperones are shown to prevent protein aggregation, while other chaperones promote refolding and solubilization of misfolded proteins [5]. The most widely used and important cytoplasmic chaperones in E. coli are DnaK, DnaJ, GrpE, GroEL, GroES, and Trigger factor (Table 3). These chaperones have been used either singly or in combination to improve the protein solubility in E. coli [1, 9, 10, 38, 64]. The most efficient chaperone combinations that are widely used to improve protein refolding are GroEL-GroES and DnaK-DnaJ-GrpE [27, 50, 84, 80]. In addition, it was observed that DnaKDnaJ-GrpE assists in the release of unfolded proteins and GroEL-GroES chaperones prevent the degradation of peptides [72]. Trigger factor has been shown to interact with GroEL and improves GroEL-substrate binding to facilitate protein folding [19]. Other chaperones such as ClpB (Hsp100), in association with DnaK, solubilizes protein and prevents aggregation [2, 47]. Other heat shock proteins, such as IpbA and IpbB, prevent aggregation of heat denatured proteins [20]. For improving the production of recombinant heterologous proteins in E. coli, different combinations of molecular chaperones should be analyzed to identify the most efficient one. It has been shown that coexpression of Skp and FkpA chaperones increased the solubility of antibody fragments in E. coli [53]. It was revealed that coexpression of GroEL–GroES resulted in production of about 65% of anti-B-type natriuretic peptide single chain antibody (scFv) in soluble form, which was 2.4-fold higher than in the absence of chaperones [40].

Table 3.ALDH3A1, aldehyde dehydrogenase 3 family, member A1; xynB, xylanase B; IL-6, Interleukin-6; Fab, fragment antigen-binding; HSP, heat shock protein.

Another strategy involves the lowering of the growth temperature to enhance the solubility of recombinant proteins [32]. Growth at a lower temperature decreases the rate of protein synthesis and thus prevents the accumulation of folding intermediates in the cytosol. Moreover, it also reduces the aggregation of protein by preventing inter- and intramolecular hydrophobic interactions, thus minimizing the formation of inclusion bodies [15]. This approach has been very efficient in enhancing the solubility of various therapeutic proteins such as interferon-α-2, human growth hormones, and Fab fragments [24, 79]. In addition, to facilitate the production of soluble recombinant proteins in E. coli, it is crucial to optimize several other parameters, such as effect of medium composition, choice of expression vectors, choice of promoters, choice of expression hosts, application of various fusion tags, rate of protein synthesis, inducer concentration, and duration of induction [16, 36, 54].

It has been suggested that the secretion of heterologous proteins into the periplasm of E. coli may provide an opportunity to produce complex therapeutic proteins. The Dsb protein family in the periplasm assists in disulfide bond formation, and the oxidizing environment in the periplasm facilitates the proper folding of heterologous proteins. Moreover, the periplasm contains very few host protein and low proteolytic activity, which facilitate the downstream processing to recover high yields of therapeutic proteins [41]. Various therapeutic proteins have been successfully manufactured by periplasmic secretion, such as commercialized Fab fragments (i.e., Leucentis and Cimza) and some full-length aglycosylated antibodies and scFvs [29, 48]. Various secretion signals have been used to target heterologous proteins to the periplasm of E. coli, such as Omp A, LamB, PhoA, STII, PelB, and endoxylanase [7]. Yim et al. [86] exploited the endoxylanase signal peptide to obtain very high-levels of expression of granulocyte colonystimulating factor (G-CSF) at 4.2 g/l in the periplasm of E. coli. Another study reported high-level periplasmic expression of insulin-like growth factor I (IGF-1), with a yield of 4.3 g/l, by using the LamB signal peptide [28]. It must be noted that besides signal peptides, the efficiency of periplasmic secretion also depends on the E. coli strain, promoter strength, and growth temperature. In some circumstances, coexpression of periplasmic chaperones, including disulfide bond oxidase (DsbA), isomerase (DsbC), and peptidylprolyl isomerase, could enhance the production of heterologous proteins in E. coli [34]. Reilly and Yansura [58] reported that overexpression of DsbA and DsbC improved the efficiency of the assembly of the light chain and the heavy chain of a full-length antibody in the periplasm, and increased the production from 0.1 to 1.05 g/l [58]. Another study reported that the yield of anti-CD20 scFv was increased, along with improved antigen binding affinity, when it was coexpressed with the periplasmic chaperone Skp [42]. Lee et al. [37] reported the development of an efficient expression system for full-length IgG in E. coli. Their study demonstrated that modification of the 5’UTR sequence and coexpression of DsbC foldase resulted in very high expression levels of heavy and light chains, and assembly of IgG was also dramatically improved in the periplasm. Under highly optimized condition, fully assembled and functionally active IgG was produced, as high as 362 mg/ml [37]. These studies clearly suggested that through proper engineering of the E. coli host, it would be possible to produce therapeutic monoclonal antibodies in E. coli in a very cost-effective and timely manner in the near future.

 

Post-Translational Modifications

E. coli is a favorite microorganism of biotechnologists for the large-scale production of therapeutic proteins [60]. However, the absence of post-translational modification processes in E. coli limits its use for the production of recombinant biopharmaceuticals. Various post-translational modifications, including glycosylation and phosphorylation, which are critical for functional activity, do not take place in E. coli due to its lack of such cellular machinery [25, 82]. N-Linked glycosylation of proteins is one of the most important post-translational modifications in eukaryotes. Wacker et al. [81] identified a novel N-linked glycosylation pathway in the bacterium Campylobacter jejuni and also showed the successful transfer of a functionally active Nglycosylation pathway into E. coli [81]. Campylobacter jejuni harbors pgl gene clusters, which are involved in the synthesis of various glycoproteins. By successfully transferring the pgl pathway into E. coli, various glycosylated proteins were produced in E. coli. Although the bacterial N-glycan structure is not similar to that seen in eukaryotes, the molecular engineering of the glycosylation pathway of C. jejuni into E. coli has paved the way for expressing glycosylated proteins in E. coli [77, 51]. Valderrama et al. [77] successfully engineered the eukaryotic glycosylation pathway in E. coli. Four eukaryotic glycosyltransferases, which include yeast uridine diphosphate-N-acetylglucosamine transferases Alg13 and Alg14, mannosyl-transferases Alg1 and Alg2, and bacterial oligosaccharyltransferase PglB from C. jejuni, were coexpressed in E. coli for synthesizing glycans, which were then successfully transferred to asparagine residues in the target eukaryotic protein [77]. This approach can also be utilized to develop glycoconjugate vaccines against several bacterial pathogens, which could be a more cost-effective and convenient alternative method to presently employed chemical-based methods of vaccine production. Currently, a glycoconjugate vaccine against Shigella dysenteriae O1, developed using this technology, has successfully cleared the phase I clinical trials. Initial efficacy and safety studies demonstrated that the glycoconjugate vaccine was safe and also elicited a strong immune response. This novel approach for glycoconjugate vaccine production using the engineered N-linked glycosylation system of Campylobacter jejuni can be exploited to produce vaccines against both gram-positive and gram-negative pathogens [8, 18, 21, 44, 83].

 

Future Perspectives

Although there are several different expression systems such as yeast, mammalian cell lines, transgenic animals, and plants that are currently being used for the production of recombinant proteins, new technological advancements are continuously being made to improve the E. coli expression system. Production of heterologous proteins in E. coli is always preferred because of the ease of genetic manipulations, well-characterized genome, accessibility of versatile plasmid vector, availability of different kinds of host strains, cost-effectiveness, and very high expression levels as compared with other expression systems. However, there are certain limitations that might hinder the efficient use of the E. coli system to overexpress heterologous genes, such as biased codon usage, protein solubility, mRNA stability, and secondary structure. The translational errors due to the presence of rare codons in heterologous proteins may result in amino acid substitutions or frame-shift mutations, which ultimately leads to undesired products. Hence, the codon usage of the recombinant protein plays a critical role in defining its expression levels. Expression of therapeutic proteins in E. coli can be enhanced by replacing rare codons in the genes with more favorable major codons. Similarly, the coexpression of genes coding for tRNAs for rare codons could increase the expression levels of therapeutic proteins in E. coli. In addition, periplasmic secretion of heterologous proteins offers several advantages, including proper folding, solubility, ease in purification, and higher yield of proteins. Antibody fragments that have been approved for therapeutic use were produced in the periplasm of E. coli, suggesting that this approach is commercially viable. Various recent studies have shown that E. coli strains can be modified specifically for each therapeutic protein to achieve high product yields as well as high-quality products.