Improvements in protein production in mammalian cells from targeted metabolic engineering

Abstract

Bioprocess optimization has yielded powerful clones for biotherapeutic production. However, new genomic technologies allow more targeted approaches to cell line development. Here we review efforts to enhance protein production in mammalian cells through metabolic engineering. Most efforts aimed to reduce toxic byproducts accumulation to enhance protein productivity. However, recent work highlights the possibility of regulating other desirable traits (e.g., apoptosis and glycosylation) by targeting central metabolism since these processes are interconnected. Therefore, as we further detail the pathways underlying cell growth and protein production and deploy diverse algorithms for their analysis, opportunities will arise to move beyond simple cell line designs and facilitate cell engineering strategies with complex combinations of genes that together underlie a phenotype of interest.

Graphical Abstract

Introduction

Over the past few decades, protein-based products have emerged as important biopharmaceuticals that treat complex human diseases (e.g., inflammatory disorders, cancer and infectious diseases). These drugs are predominantly synthesized in mammalian cells lines since the cells often produce high quantities of therapeutic proteins with appropriate critical quality attributes (CQAs) that impact potency and immunogenicity (e.g., glycosylation) [1]. The market for mammalian-produced therapeutic proteins has gradually grown and is projected to further increase to ~20% of the pharmaceutical market in 2017 [2].

High volumetric productivity and product titer are important to obtain more affordable protein therapeutics [3]. A cell line may achieve these goals from a combination of changes that collectively make the host system a protein “superproducer” [4]. Thus, these attributes are selected by manufacturers and include high translation efficiency, secretory capacity, growth capacity, duration of viability at maximum cell density, and human-like post-translational modifications [4, 5].

To date, most improvements in protein production have been achieved by media and bioprocess optimization (e.g., feeding strategies and process parameter control) [6]. Random mutagenesis is also been used to find cell factories with desired phenotypes. However, the availability of high-throughput omics data (genomic, transcriptomic, proteomic and metabolomic) [7] and the emergence of genome editing tools [3] provide novel opportunities for targeted genome engineering of the host cell. Indeed, they have enabled the overexpression or down-regulation of specific gene candidates to increase yield during culture and control product quality [3, 4, 8–10]. Several prominent strategies have targeted cell metabolism, cell cycle regulatory machinery, the protein secretion pathway, apoptosis, and protein glycosylation [9, 11].

Core metabolic engineering for an effective host cell alteration

The engineering approaches developed to control protein production attributes have resulted in mixed levels of success. Most successful studies targeted energy metabolism to reduce the accumulation of toxic by-products (i.e., lactate and ammonia) and/or increase metabolic efficiency (Table 1).

Table 1.

Metabolic engineering strategies used to reduce accumulation of lactate and ammonia in the culture medium

Target gene	Effect	References
glucose transporter (GLUT1)	Decreased the transporter affinity to reduce flux through glycolysis and lactate production.	[12]
fructose transporter (GLUT5)	Transfected GLUT5 to allows cell to grow on fructose as sole carbon source which results in a reduction of sugar consumption and lactate production.	[13, 14]
lactate dehydrogenase (LDH)	Reduced the conversion of lactate from pyruvate and the regeneration of NAD+.	[15, 16]
pyruvate carboxylase (PC)	Increased the conversion of pyruvate to oxaloacetate and its entry into the TCA cycle. The overexpression of PC is often associated with a reduction in specific glucose, glutamine consumption rates and lactate to glucose yield.	[17–20]
Co-overexpression of alanine aminotransferase (ALT1) and taurine transporter (TAUT)	Increased transamination between 2-oxoglutarate and alanine, which accumulates early in the culture period due to the TAUT introduction. Pyruvate and glutamate were formed, thus increasing the flux through TCA cycle and reducing lactate formation.	[22]
carbamoyl phosphate synthetase I (CPS I) and ornithine transcarbamoylase (OTC).	Improved the first and the second steps of urea cycle, leading to decreased ammonia secretion.	[23]
glutamine synthetase (GS)	Improved the synthesis of glutamine from glutamate and eliminated the need of exogenous supplied glutamine and reduced ammonia accumulation	[24–26]
malate dehydrogenase (MDH)	Improved the conversion of oxaloacetate to malate, forced flow into TCA cycle, increased ATP and NADH intracellular levels, and improved growth.	[27]

Several studies have successfully decreased glucose uptake by up to 50%, leading to reduced lactate production. This has been accomplished by knocking down the glucose transporter GLUT1 [12] or by transfecting the fructose transporter GLUT5 [13, 14]. An 80% reduction in lactate production was also achieved by knocking down lactate dehydrogenase [15, 16] leading to an improved product titer from 2–3-fold. Other strategies have overexpressed pyruvate carboxylase to improve the connection between glycolysis and the TCA cycle [17–20]. Interestingly, the overexpression of pyruvate carboxylase decreased glutamine consumption and extended cell viability. Zhou et al. [21] attenuated the expression of both lactate dehydrogenase and pyruvate dehydrogenase kinase using siRNAs, thus forcing the conversion of pyruvate to acetyl-coA instead of lactate. This strategy successfully reduced lactate production by almost 90% and increased specific and volumetric productivity by 75% and 68%, respectively, without decreasing cell growth. Another double target strategy to reduce lactate production was performed by [22] by overexpressing alanine aminotransferase and the taurine transporter. The overexpression of the taurine transporter in CHO cells leads to an accumulation of alanine in the early period of the culture, but an overexpression of alanine aminotransferase converts the alanine to pyruvate, which is subsequently metabolized in the TCA cycle.

In addition to lactate, efforts have been made to control other toxic byproducts that impact cell viability and product quality. For example, ammonia production has been reduced by overexpressing the first two steps of the urea cycle by up to 35% [23] and by overexpressing glutamine synthetase. Such efforts have improved the synthesis of glutamine from glutamate [24–26] and thereby improved cell phenotypes.

Beyond the aim to eliminate toxic byproducts, Chong et al. [27] engineered metabolism to improve integral cell number. They had observed that malate accumulated in the medium of CHO cultures. This suggested that a bottleneck existed at the step of the malate dehydrogenase reaction in the TCA cycle. To overcome this, they overexpressed malate dehydrogenase II to reduce NAD⁺ concentration and increase NADH, which was further used by oxidative phosphorylation to produce ATP, thus increasing cell viability by 1.9-fold.

Thus, numerous efforts have been made to engineer the central metabolic pathways in mammalian cell culture to reduce toxic byproducts and increase cell viability, usually by targeting just one or two genes. Further efforts will continue, especially since recent work has identified many more endogenous metabolites that inhibit growth [28] and so metabolic engineering efforts will aim to control their concentration in mammalian cell culture.

Controlling other metabolic pathways while engineering central metabolism

Most of the metabolic engineering efforts in the aforementioned studies aimed to reduce toxic by-product accumulation by targeting single genes related to carbohydrate metabolism (Table 1). However, studies are now tracking the influence of central carbon metabolism on other pathways, including the biosynthesis and metabolism of amino acids, nucleic acids, lipids, and ultimately protein synthesis and protein quality. The knowledge of these pathway connections can be used, for example, to track the cellular switch occurring between exponential growth and protein production, since this event changes the balance of different central metabolic pathways (e.g., pentose phosphate pathway flux and oxidative TCA cycle) [29]. For example, the ratio of flux between the pentose phosphate pathway (PPP) and glycolysis flux has been suspected to significantly impact protein production. Indeed, the PPP activity in a human cell line was much lower compared to CHO K1 cell line that produced higher amounts of protein [30, 31]. Furthermore, Mulukutla et al. [32] recently investigated the regulation of glucose metabolism, including the connections between glycolysis, the pentose phosphate pathway, nucleotide synthesis, glycerol-3 phosphate metabolism and serine/glycine/threonine biosynthesis. The interconnection between these pathways reiterates the central role of carbon metabolism of regulating the global metabolic state of a cell and other cellular processes (e.g., apoptosis, glycosylation) influencing protein quality attributes (Figure 1).

Figure 1. Central role of carbon metabolism to cell physiology, product quality, and bioprocess optimization.

The engineering of central carbon metabolism enables the regulation of key cellular processes (i.e., glycosylation, secretory pathways, cell cycle regulation, apoptosis, intracellular trafficking, and endocytosis), involved in the acquisition of protein quality attributes (i.e., protein composition, aggregation, stability, specificity, complexity, and folding). Moreover, it helps in efforts to control the global metabolic state of a cell to ensure the achievement of optimal culture process goals (i.e., cell density, viability, productivity, product titer, quality, and reproducibility).

With carbon metabolism connected to other cell processes, it is possible to control desired cellular attributes by only manipulating genes of central metabolism. For example, Majors et al. [33] inhibited apoptosis by modifying GLUT1 and/or hexokinase expression, since a high glycolytic activity can slow the onset of apoptosis by improving energy efficiency [34]. Moreover, recent studies have altered glycosylation patterns by maintaining efficient glucose metabolism and avoiding accumulation of toxic byproducts, since central metabolism controls the supply of nucleotide sugar precursors [35, 36]. Finally, Mulukutla et al. [32] highlighted several potential engineering targets within central metabolism (e.g. PFK, F26BP, PKM2, PGAM, 3PG, 3PGDH and 6PGD) that globally impact cell growth and protein production. Thus, knowledge of how different pathways are connected will guide future cell engineering efforts and enable the use of metabolic engineering to regulate the many pathways related to protein production.

Conclusion - A more holistic picture of mammalian cell physiology can guide more complex engineering strategies

The ideal CHO cell line for protein production does not have any exact, defined phenotypic traits in particular. However, it typically includes attributes of high cell viability, cell density, and titer. It also would exhibit robust growth, high stability of expression and control over desired post-translational modifications. To achieve these attributes, multiple modifications are needed in CHO cells, and such changes would target diverse cell pathways and physiological functions.

To begin identifying genes whose expression correlate with desirable attributes of protein synthesis and secretion, several studies have use diverse experimental data types to compare mammalian factories that produce high or low titers of recombinant protein. These have included data types such as transcriptomics [5, 37, 38], metabolomics [39], proteomic profiling [40, 41], and miRNA expression patterns [42]. These studies highlight the importance of intracellular trafficking, endocytosis, cytoskeletal elements, lipid metabolism, the unfolded protein response, protein-processing constituents in the Golgi, and cell cycle related functions (e.g., apoptosis resistance and proliferation) [3, 5]. However, we are still missing fundamental knowledge about how these cellular mechanisms are organized and link together. Furthermore, it is often not clear how they are connected to process conditions, and how these factors all impact protein production [43]. Thus, efforts to further enhance drug production will be facilitated as the molecular basis of these processes are studied and linked to protein production. Pathway maps and interaction networks [44–46] are starting points that link the processes, and can help identify new process conditions and cell engineering strategies that control product quantity and quality.

To completely map out the cellular pathways, it is first critical to know the genetic basis of the cells. Recent efforts to sequence the Chinese hamster genome [47–50] have enabled this for CHO cells. Furthermore, variations in the cell lines, such as mutations and epigenetic changes in individual cell lines [51], can be catalogued and analyzed to help develop engineering strategies for improved protein factories, focusing on a specific cell line.

With the genetic basis established, a holistic understanding of the cellular basis for high productivity could be achieved within the systems biology context. This will be accomplished by the development of detailed metabolic pathway and interaction maps of the major cell processes, and identifying the genes associated with the pathways [45, 52]. These metabolic networks can be converted into mathematical models that can guide engineering efforts by quantifying the connection of cellular processes to desired phenotypes and protein production using metabolic flux analysis [44, 45, 53]. Furthermore, these models will allow the analysis and integration of the avalanche of high-throughput data available at the genomic, transcriptomic, proteomic, and metabolomic levels thanks to innovations in these fields [54]. The analyses of these data in the context of cellular pathways will be particularly informative when investigated along with phenotypic differences of different cell lines, such as variations in growth media, feeding strategies, process conditions, and the type and amount of produced protein.

Finally, with the genetic basis and pathways mapped out, we can move toward the design and implementation of more complex genetic changes, using multiplex genome edits in CHO cells [55–57]. Similarly, we can harness the multiplex targeting of miRNA [58, 59]. Thus, with the guidance of predictive mathematical models of CHO cell metabolism, and emerging concepts of dynamic [60, 61] and combinatorial [8], more complex and targeted approaches can be explored to improve mammalian factories. These tools will be invaluable in the future of engineering, wherein multiplex metabolic engineering strategies account for details of cell line, culture environment and product, in the pursuit of a perfect therapeutic production factory.

Highlights.

• Mammalian cell metabolism impacts therapeutic protein production

• Recent sequencing efforts allow more targeted engineering approaches in CHO cells

• Metabolic engineering of CHO metabolism has improved production processes

• Systems biology tools can guide future engineering of metabolic phenotypes