Expression of mammalian proteins for diagnostics and therapeutics: a review

Mammalian and non-mammalian cell lines

Growing numbers of revolutionary biological products and biosimilars have thus intensified the demand for high productivity host cell lines as the choice of host cells is the most notable and influential method to increase volumetric cell yield while minimising production cost [71]. However, while achieving high product yield at low cost is much coveted, it is important to consider the structural quality of the produced molecules as the selection of expression system can heavily influence the type and degree of the product’s structure, particularly glycosylation [29]. Some main traits of a good host for therapeutic production are rapid development, known organelles and functions for protein processing and secretion, as well as possessing strain engineering toolkits to integrate recombinant protein into host genome and engineer post-translational modifications [72]. Additionally, intrinsic qualities such as proven specific productivities, lack of co-secreted proteases, proteome’s amount of host cell proteins, process flexibility, and absence of mammalian viral infectivity, in addition to other factors including usage of cheaper, less-refined feeds, along with robustness to shear stress, pH, and temperature can direct the selection of a prospective host as well [71]. Recombinant proteins such as antibodies can be expressed using various platforms and there have been multiple expression systems identified.

The most common strategy for producing recombinant proteins is using bacteria, typically Escherichia coli but, the foreign protein being toxic to the host, besides, protein folding machinery differences, and the absence of post-translational modifications in bacteria makes them inapt hosts for expression of functional eukaryotic proteins [8]. Yeast, insect, and plant cell-based expression systems are applicable as well but unfortunately, they form glycan structures, unlike a human’s which carries the risk of undesired immunogenicity for therapeutics [29]. Additionally, the development of the aforementioned systems are fragmented, making it challenging to compare them to mammalian cell lines for potential benefits [71].

As a result, the primary framework for generation for therapeutic recombinant proteins is utilising mammalian cell lines as for ectopically expressed mammalian genes, they are the nearest likeness to the native environment and in this way have an inherent capability to create functional mammalian proteins [8]. Therein, mammalian cells are the largest contributors to recombinant proteins on the market at about 70% [73]. They can perform the proper and necessary folding, post-translational modifications and glycosylation in patterns similar to those in humans which reduces the risk of rejection besides being able adapt to large-scale production systems and produce large volumes of therapeutic antibodies [74, 75]. Glycosylation in antibodies is an important factor to note as it has incredible diversity and complexity of which antibody structure and function is crucially dependent on. Glycosylation patterns can determine structure and effector functions of antibodies including ADCC, CDC and its pharmacokinetic/pharmacodynamics properties [4]. Despite that, recombinant proteins expressed in mammalian cell lines can encounter complications in expression yields, time required to produce over-expressing lines, along with increased cost [8].

The most popular cell line used in industrial therapeutic production is CHO cell lines. CHO cell lines can expand rapidly to large densities in bioreactors, generate high quantities of recombinant protein, increased resistance to human virus transmissions, and generate human compatible post-translational modifications, besides possessing a genome with considerable plasticity which increases its flexibility to being genetically modified and manufacturing process scales [7]. Despite these many advantages, their characteristics of being genomically unstable and heterogeneous results in regular genotypic variation and phenotypic instability, which brings about cell line instability and therefore, concerns about the process reproducibility and consistency [6]. This issue generates genomic instability over time, causing a drop in protein production and thus fabricates titer and product quality instability [7]. Expression of mAbs from mouse and hamster-derived cell lines has its downsides as nonhuman cell lines express proteins with glycan structures which differ substantially from humans, and they are N-glycolylneuraminic acid and galactose-alpha 1,3-galactose group which elicits immunogenic responses in human cells [76]. Furthermore, the α-1,3-galactose linkage induce anaphylaxis reactions at specific levels [77]. These cell lines secrete fucosylated IgG-G0 glycoforms which is able to activate complement through MBL binding when clustered [78]. It has been well identified that nonhuman glycan structures are produced by rodent cell lines [79]. Sialic acid production differs as well. In human cells, the formation of Neu5Gc is inhibited, but it is identified in murine derived proteins which triggers antibody production to contribute to serum sickness in humans and many other pathophysiological implications [80, 81]. Additionally, murine derived proteins may consist of xenoantigenic gal-a-gal linkages which can cause hypersensitivity reaction and as a result increases risk of murine derived antibody-based therapeutics [82]. Consequently, product quality, chiefly glycan structures of murine derived biologics must be screened but this result in acceptable proteins being discarded from these concerns.

To circumvent immunogenicity, human cell lines can be exploited. Upsides of employing human cell lines are that they form immunogenic glycans, but the downsides are that they are vulnerable to human viral infection. Thus, viral inactivation on human cell lines is essential and HEK293 cell line which stands for Human Embryonic Kidney 293 is a viral DNA transfected human cell line widely used for protein biologics production [29]. It is the leading human cell line platform for protein expression with diverse variants although its efficiency is restricted by low transgene expression levels in these cells [76].

Both CHO and HEK cell lines are extremely stable, amenable to genetic manipulations, and have rapid growth and proliferation rates, indicating a high protein synthesis ability [10].

Cell culture medium

In another note, although choosing high expressing host cell clone is vital in bioproduction, developing cell culture medium is critical for cost-effectively promoting cell growth and productivity, as well as for reducing batch-to-batch variance during production [12]. Furthermore, to improve cell lines and cell culture processes with high efficiency and quality, it's crucial to use the right media [11]. Variations in cell culture media can influence therapeutic behaviour of the produced antibodies through glycosylation or otherwise [4]. The importance of cell culture medium stems from the fact that it is able to promote cell viability and proliferation along with cellular function which indicates that medium nature has a direct impact on research findings, biopharmaceutical output rates, and assisted reproductive technology treatment outcome [13]. As a result, the selection of an appropriate medium is critical.

Synthetic media are currently divided into various categories depending on the form of ingredients used, such as serum-containing media, serum free media (SFM), protein-free media, and chemically defined media. Serum-containing media consist of various serum-derived compounds, thus obscuring the medium makeup. Thereby, causing proportions to vary from batch-to-batch and because of this, the culture findings are less predictable and there is a threat of contamination whether in form of bacteria, viruses or prions which also impedes further optimisation and standardisation of systems [83]. This unfortunately limits its use for therapeutic protein production as serum content increases the risk of contamination such as pathogen transmission and additionally, its high protein content can interfere with product purification during the downstream process [11, 15]. On the bright side, this media is customizable to increase cell proliferation for a broad variety of mammalian cell types since serum contains a large number of active substances required for animal cell survival and development [84].

Serum, the supernatant of clotted blood, is an integral part of mammalian cell culture as it is a source of numerous compounds, reduces shear stress, improves the medium’s pH-buffering capacity and modify the culture substratum to allow proliferation of adherent cells [13, 15]. There are various serums to date, with some examples such as fetal bovine serum (FBS), calf serum (CF), and horse serum. FBS is the go-to for most applications as it is abundant in growth factors and lacking in γ‐globulins that stimulate cell proliferation but comparatively, CS has a reduced ability to promote cell growth and there is a high degree of homogeneity between lots in horse serums [13]. FBS owns its popularity to its ability to promote cell proliferation, metabolism, and differentiation, support cellular functions, supply necessary molecules, buffer pH, and inhibit protease of which their combined role is to promote cell development [85]. However, in terms of biosafety, FBS, like most other serums has unknown biochemical makeup, varying between lots, hence hindering culture medium standardisation. An alternative is knockout serum replacement, with a specified composition of organic components, trace elements, insulin, transferrin, and albumin [86].

Serum-free media (SFM), on the other hand, have a standardised formulation, leading to high reliability of results and the ability to evaluate the cultivation process. The development of conventional SFM to protein-free and chemically defined medium stems from increasingly defined components and nutrients necessary to improve cell proliferation [85, 87]. Subgroups of SFM, protein-free media comprises of zero protein whatsoever and chemically defined media which contains only known ingredients. These factors improve stability and reproducibility, easing cellular secretion detection while reducing microbial contamination. The general makeup of this media are amino acids, vitamins, glucose, inorganic salts, anti-shear protectants and nuclei acids with additions of supplements such as chemical components and growth factors [87].

Component in cell culture media is vital for its unique functions [87–89]. Energy sources, i.e. glucose plays a vital role as glycosylation precursors and thus can affect glycosylation profiles of the produced mAbs. Amino acids serve as important energy and nitrogen sources. Vitamins are essential components of SFM because they help maintain cell growth and regulate and monitor cell metabolism. Lipids are essential components of cell membrane structure that plays a role in cell signalling, serves as energy storage in cells as well as increase protein expression and glycosylation. Trace elements regulate cell metabolism and enzyme activity. Lastly, inorganic salts modulate osmotic pressure during cell culture phase besides regulating the cells’ metabolism.

The design of culture media and supplements influences mAb production as it can improve bioprocesses where certain pathways are favoured and lipid limitations are eradicated [90]. However, the downsides to serum free media are weaker growth performance when compared to their counterparts besides the difficulty of design as to date, there are very limited cell lines blessed with serum free media [84].

Transient and stable expression

The procedure of producing a mammalian recombinant protein expressing system starts with transfection of recombinant gene vector into cells and followed by selection using selection markers to select cells with successful plasmid integration to generate clones [29]. Antibodies can be expressed in mammalian cell lines transiently or stably by integrating expression constructs into the host’s genome. Choice of expression systems can be done solely based on quantity of protein required and convenience as this does not affect their biochemical characteristics as reported in these studies where there is no structural discrepancies between proteins produced via transient and stable expression systems [54, 55, 91, 92]. This is vital as biochemical properties influences the affinity and thus stoichiometry, stability, and expressibility of antibodies, making them are essential parameters for antibodies in research and medical tools [93].This study reports no biochemical distinction from stable and transient expressed antibodies with the only difference being expression levels which was reasoned to be from dissimilar promoters employed [92]. Based on this report, antibodies from both systems exhibit similar properties from antibody folding, assembly, and affinity [54]. A challenge faced by bispecific antibody expression is misassembly but as reported in this study, both systems generate similar amounts of misassembled bispecific antibodies although most expressed proteins were correctly assembled [91]. Furthermore, this study reports no biochemical discrepancies in proteins expressed from transient and stable systems but antibodies produced from stable has a slightly high level of proper assembly and expression, the authors noted co-expression in stable cells remains the predominant approach for IgG-like bispecific antibody production [55]. Additionally, the review from Wang reports on numerous bispecific antibodies expressed from various hosts and expression systems where both transient and stable systems were utilized, indicating no particular desirable mode of expression with both reporting high stable formulations [55]. All in all, proteins from stable cell lines and transient cells are similar biochemically with no reported differences on its stoichiometry.

Transient gene expression has a short time frame for protein harvesting but a short production time frame. Transiently transfected cells generate a lower yield. Transient expression is based on transgene-carrying plasmids transfected into the cell nuclei are transcribed as non-integrated plasmids rather than being permanently incorporated into the genome and thus the foreign gene is not replicated and gets diluted over time as it is lost through cell division, degradation or other factors [15]. Transient gene expression is most frequently and successfully employed technique for functional studies such as biochemical and preclinical assessments as well as initial screening for new expression strategy, especially when examining the design of the vector construct and molecular candidates [16]. It is unfortunately unapt for most structural studies and large-scaled productions, and thus as a result, transiently transfected cell lines are only used in short-term experiments [16]. However, as of recent years, through engineering of new expression vectors, transfection optimizations, and development of new culture processes, transient expression systems have advanced to the point where milligram to gram quantities of recombinant proteins can now be generated and although the yield is still comparatively lower than yield from stable gene expression systems, this is an incredible progress [15]. HEK 293 cells are the preferred mammalian cell-based transient protein expression systems for their flexible conversion to suspension cultivation in SFM, good transfection efficiency, high productivity, ability to grow to high cell densities and the inclusion of an episomal replication mechanism [16, 94].

Furthermore, transiently transfected cells may express inconsistently throughout the population. A dated notion believes that all in vitro generated antibodies will possess similar profiles to those produced in vivo but time has proven otherwise and minor variation in culture conditions, cell quality, medium content as well as transfection conditions cause major differences in glycan content regardless of being produced in one and the same production laboratory [17]. Using stable cell lines, the disparity and variation in expression levels associated with repeated transient transfection can be eradicated. Another downside to transient gene expression is that in transient transfection assays, several gene regulatory regions or inducible promoters work incorrectly which may be as many long-range enhancer and silencer components do not operate unless incorporated into the genome or a lack of sufficient chromatin formation, therefore transient transfection assay is not a suitable platform for determining the full impact of regulatory regions on a gene’s expression [19]. Stable expression is especially important and thus the standard practise for continued genetic research or industrial protein manufacturing.

Meanwhile stable clones may take several months to be generated and but it is able to express large amount of consistent, stable quality protein of interest with regulatory familiarity. Stable cell lines can continually express the transgene as the expression construct is integrated stably into the host cell’s genome and thus stable cell lines are able to carry and pass on the same modified genetic characteristics to its future generations [18, 73]. Although it can be noted that as stable cell lines consist of a single gene copy number of the integrated DNA, it results in a lower level of protein expression as compared to their transient counterparts with a higher copy number [95]. In contemporary era, the overflowing market for antibody medicines induces a need for their much-improved counterparts and the patent expiration of many “blockbuster” antibody drugs that drove the production of several biosimilar antibodies has all in all increases the demand for time effective, quality and quantity of antibodies [41]. Stable cell lines are able to metaphorically tick all these boxes as for long term production. These cell lines are cost-effective and manufactures stable antibodies in high yield to cater for the demand.

Principally, stable cell lines are generated using strong viral or cellular promoters and enhancers to drive expression by genomic integrating the GOI through an expression vector encoding the recombinant gene [18]. Based on integrant quantity and their sites, there may be variation in protein synthesis levels in the same cell batch [8]. A selective advantage to the transgenic cells such as antibiotic resistance in the genome can be introduced for selection of cell lines. Gene transfer to stable mammalian cells lines can be done virally and non-virally. Nonetheless, the expression levels are dependent on transgene integration position, quantity of integrations per cell, and individual cell variations amongst many others. Consequently, screening for high producer cell lines is necessary to obtain a cell line of fast growth and high productivity.

Position effect is a phenomenon from randomly integrated transgene, interfering with expression monitoring as it is dependent on the differential influences of its genomic surroundings [96]. This is due to clonal variation which brings about integration site concerns as the current human cell line generation processes are based on random integration of GOI into the genome which limits predictive value, process streamline, and cost-effective therapeutic glycoprotein production which also makes it impossible to directly compare between wild-type and mutant reporters [19, 20]. Besides, an integrated gene’s copy number tend to be vague and differs between cells, making the cell population heterogenous. To preserve the transgene, continuous selection pressure is necessary, as cells prefer to eliminate randomly inserted alien genetic elements, which may alter genome integrity [18]. The encoded resistance markers are not able to completely neutralize such selective agents i.e., antibiotics, which will then lead to cumulative mutations in the host cells, and thus must be removed from the proteins before usage [97]. Robust exogenous promoters such as pCMV elevate and express the recombinant gene non-physiologically [98]. Recombinant protein over-expression from multiple integrants can impose a significant strain on host cells, hence negatively impacting proper protein production [18].

Thus, there are strategies to eliminate clonal variance while generating stable high expressing cell lines expressing protein of interest in a controllable, quantifiable, and efficient fashion. Site-specific targeting based on recombinases manipulate genomic DNA with high efficiency and fine spatiotemporal control by inserting the GOI into the chromatin in a transcriptionally active region, or genomic hot spot [99, 100]. These systems do not possess preference for specific cell types but require genetically engineered parental cell lines with “landing pads” which are suitably arranged recombinase recognition sites as chromosomal targets [101]. Site-specific recombination is a precise genome editing tool to produce more controllable and predictable stable cell clones with high productivity and is likely to dramatically reduce the time needed for cell line screening, thus making them indispensable for biomedical research [79]. Due to their exceptional ability to excise, insert, inverse, and translocate genomic DNA in living organisms, site-specific recombinases are vital genome engineering techniques [100]. Anyhow, there are some technical issues that comes with the use of site-specific recombinase. The systematics of temporal and spatially regulated recombination must be improved, the expression of a recombination reporter does not necessarily signify target gene deletion, traditional lineage tracing for cell fate mapping is solely dependent on recognized cell type markers, and hence to perform regional and prompt gene expression and deletion, novel strategies are required [100]. Some examples of the genetic manipulation toolbox are the Cre/Loxp and Flp/Frt, which depend on a reporter gene’s random integration to identify genomic hot spots [79].

The most widely used site-specific recombinase, Cre (causes/cyclisation recombinase) has been used to understand gene function and modify many forms of DNA, in any cellular setting and in several species in vivo [100]. In site-specific recombination, the two necessary components are the transacting function, which in this case is Cre and a site which is loxp of which the Cre recombinase catalyses the recombination between the loxp recognition sites [100]. The expression and activity of the Cre recombinase, which is encoded by the bacteriophage P1 gene Cre is regulated by tissue-specific promoters and ligand tamoxifen to allow conditional genetic modifications [102]. However, there are some setbacks in the usage of this system as they possess an inconsistent nature and the epigenetic silencing of the promoters used can create highly variable expression patterns and alien phenotypes, additionally, as the event is invisible, there may be variable recombination or gene deletion efficiency which is why novel technologies such as iSuRe-Cre can be used [102].

A similar, straightforward and recurrent technology is the Flp-FRT recombination system derived from Saccharomyces cerevisiae where the flippase (Flp) recombinase catalyses DNA rearrangements through engineered genomic Flp recognition target (FRT) sites to achieve precise gene integration [21, 103]. This technology is used to modify genomes both within the germline and somatic tissue where gene expression is manipulated by utilising previously integrated FRT sites in the genome or extrachromosomal arrays to modify DNA elements at specific sites [104]. However, there is a range of possible shortcomings to this current system. The existence of bacterial sequences that are CpG-rich and thus vulnerable to methylation in mammalian cells, which could result in the reporter being silenced [19]. Additionally, the Flp-FRT systems integrate entire vector sequences, including vector backbone, into the FRT site [21].

Homologous recombination is another strategy to be adopted for eradicating clonal variation where targeted transgene insertions are site-specific using designer nucleases which allows considerable flexibility in target site selection in comparison to site-specific recombination and is known as biological scissors for genome editing [101]. An example is the genome editing RNA guided system, CRISPR/Cas9. This system is made up of two elements, Cas9 enzyme and a guide RNA (gRNA), where Cas9, a RNA guided endonuclease is used to form targeted double-strand breaks (DSBs) in mammalian cell genome which will be patched by non-homologous end-joining (NHEJ) or, to a lower degree, homology-directed repair (HDR) [100, 105]. Utilising clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein (Cas9)-mediated targeted integration combined with recombinase-mediated cassette exchange (RMCE), this system is a platform for producing isogenic cell lines [20].

As of recent years, new systems consisting of two or more of the aforementioned techniques are designed to offer the best of all worlds where the limitations of each system can be countered with another. An example is a new integration system where large DNA constructs can be inserted into specific genome sites by using a combination of recombinases, Flp and Cre as regular methods as the application of CRISPR technology is laborious and time consuming [104]. Another novel technology is “Cas-Pi”, which stands for cascaded precise integration, where large, complex, exogenous, and repetitive DNA sequences can be precisely integrated into genomic loci in cultured cells [21]. This is especially unique as these DNA sequences are difficult to integrate with contemporary genome targeting technology and this protocol came to be with a combination of Cre-loxp and Flp-FRT recombination events as well as CRISPR technology [21]. In genomic engineering, there is also the use of different hybrid enzymes such as the development of Flp-TAL recombinases, which comprises of Flp, a site-specific recombinase and DNA-binding site of transcription activator-like effector (TAL) to integrate and delete genes [22].

Nonetheless, the design of these systems and concurrent necessity expertise as well as the cost impedes their deployment.

Other strategies in increasing antibody yield

Low protein yield is a major drawback of producing recombinant therapeutic proteins from mammalian cell lines such as HEK293 cells. Optimizing the design of the expression vector components is one of the many efficient strategies for increasing transgene expression levels [106]. Use of promoters such as EF1α, SV40, TK, and CMV promoters can regulate mammalian cell line expression [20]. The most popular promotor is pCMV for its high expression but yield of transgenes powered by the CMV promotor diminish with culture time due to complex transcriptional silencing and DNA methylation issues from its many potential methylation sites [76]. Therefore, artificial promoters, such as the fusion of CMV with chicken beta-actin promoter (CAG) are preferred instead [76]. Besides, while the transgene expression from pCMV promoter is high in initial stages after transduction or transfection, it decreases soon after during prolonged culturing of the transduced or transfected cells and instead the transgene expression was noticeably more stable and increased under the control of the pCNA promoter which was noted through this study as a potential universal promoter for gene therapeutic constructs [23]. CMV may be the choice promotor for short-term expression while for long-term expression, CMV-CAG hybrid or CNA may be chosen instead. Thus, the choice of promoters is especially important as the promoter is a crucial component of an expression vector, and choosing the right one will boost the expression and stability of a GOI.

Another strategy is to reduce cell culture temperature to a slightly hypothermic temperature to increase transgene expression. This will empirically extend the active life of the culture by slowing apoptosis and nutrient degradation, and, for certain proteins, this technique will also improve the yield of correctly folded protein [10]. Besides improving the assembly and secretion rates of the expressed protein, lower temperatures reduce product aggregation due to improved endoplasmic reticulum machinery [107]. Hypothermia causes a rise in transgene expression, as well as an accumulation of cells in the G1 process and a decrease in cellular metabolism [15]. However, while reducing culture temperature does not impact protein function or activity, it decreases cell growth rate and impedes cell division [108].

Furthermore, specific recombinant protein productivity is correlated with culture osmolality as it influences cell size which in turn influences biosynthetic capacity and nutrient exchange [24]. Hyperosmolar conditions which are customarily induced from repeated feed addition and metabolic by-product build-up or salt supplementation with sodium chloride or potassium chloride brings on increased nucleus mitochondria, endoplasmic reticulum, as well as protein and gene expression [24]. However, the quality and biopotency of mAb production are guided by application time, hyperosmolality level and culture mode [88]. It is important to note that hyperosmolar conditions can suppress cell growth and apoptosis [24, 88].

Moreover, increase specific productivity can be obtained by cell cycle arrest when the cells are at the largest size. Harvesting of antibodies can be optimised to be done with the highest productivity during or prolonging the stationary phase and late-stage culture by starving the cells or adding molecule regulators as they are the key development phases in recombinant protein production where there is a rise in cell percentage in the G1 process and an increase in cell diameter [24]. Without increase cell density, this strategy is makes it possible to obtain and increase specific volumetric productivity as cell size is correlated to extracellular nutrient conditions which enables mammalian cell size to be manipulated through media and feeds to improve cell growth and specific productivity [25]. Correlating with osmolarity, there is an accumulation of extracellular metabolites which can be further intensifies by overfeeding during the late-stage culture contributing to an increase in osmolarity [24].

Downstream processing

The recovery and purification of biosynthetic is critical in assuring its quality. After the proteins are produced and harvested, the next step is purification. The principal concern for purification is purity, followed by speed, overall yield, cost, and process throughput [52]. However, although purification will remove most unwanted molecules, heterogeneities will still occur [39]. Various modifications such as incomplete disulphide bond formation, glycosylation, N-terminal pyroglutamine cyclisation, C-terminal lysine processing, deamidation, isomerization, and oxidation, over the protein lifespan amass to heterogeneity [109]. Furthermore, heterogeneity may be caused by subsequent exposure to culture media, stress from pH and temperature as well as post-translational modifications, degradation, and other chemical modifications during preparation and storing [39].

A purification system is chosen on the basis of its robustness, reliability, and scalability. Affinity purification are often the choice method for antibodies although its effectiveness is reliant on the antibody’s ability to bind to an affinity adsorbent [110]. For process scale purification, Protein A affinity chromatography derived by the cell wall of Staphylococcus aureus is the benchmark as it utilises specific ligand interactions between mAb Fc region and immobilised Protein A to achieve high selectivity at over 95% [52]. For pharmaceutically accepted purity levels, Protein A affinity chromatography is followed up by cation-exchange chromatography and then anion-exchange chromatography [52]. It is then proceeded with virus inactivation and ultrafiltration/diafiltration. Other purification systems include histidine-ligand-affinity chromatography, metal-affinity chromatography, thiophilic interaction, lectin-affinity chromatography, and purification of recombinant antibodies using affinity tags [110].