Towards Enhancing Therapeutic Glycoprotein Bioproduction: Interventions in the PI3K/AKT/mTOR Pathway
2019 Volume 44 Issue 2 Pages 75-83
Details
2019 Volume 44 Issue 2 Pages 75-83
Recombinant glycoproteins produced in mammalian cells are clinically indispensable drugs used to treat a broad spectrum of diseases. Their bio-manufacturing process is laborious, time consuming, and expensive. Investment in expediting the process and reducing its cost is the subject of continued research. The PI3K/Akt/mTOR signaling pathway is a key regulator of diverse physiological functions such as proliferation, global protein, and lipid synthesis as well as many metabolic pathways interacting to increase secretory capabilities. In this review we detail various strategies previously employed to increase glycoprotein production yields via either genetic or pharmacological over-activation of the PI3K/Akt/mTOR pathway, and we discuss their potential and limitations.
Key words: mTORC1, CRISPR, specific productivity, translation
Although not of human origin, immortalized Chinese hamster ovary (CHO) cells are the most frequently utilized host cells for the commercial production of glycoproteins-based drugs (Walsh, 2014). Compared to other producer cells, over 70% of the approved therapeutic recombinant proteins (r-proteins) involved in manufacturing systems are produced in CHO cells (Kim et al., 2012). From the original CHO cell line established in 1957, diverse genetically modified CHO-derived cell-lines have been developed (e.g., CHO-K1, CHO-S, and DG44). These cell lines meet various production needs (Kuo et al., 2018). In 1986, the FDA approved the first therapeutic protein, the human plasminogen activator (tPA), produced in CHO cells (Mohan et al., 2008). Thus, production in CHO cells has been applied for human use for over three decades.
It is not accidental that CHO cells have been and still are the most preferable host-cells employed for r-proteins production. CHO cells have distinct characteristics that best meet manufacturing requirements; CHO cell-based bio-production is a safe process, simplifying regulatory approval (Kim et al., 2012). CHO cells are more resistant than other production systems to a variety of viruses, since they lack key receptors found in viral infections. Thus, viruses are unable to efficiently utilize the molecular machineries of CHO cells for propagation (Berting et al., 2010; Xu et al., 2011). Besides safety issues, the most important advantage of CHO cells is their ability to grow to high densities in serum-free, chemically defined media in large-scale industrial stirred-tank bioreactors. This decreases production expenses while maintaining quality (Sinacore et al., 2000). CHO cells introduce human-like glycosylations (Butler and Spearman, 2014; Kim et al., 2012). This is significant, since post translational modifications (PTMs) in general and glycosylation in particular can have profound effects not only on the spatial structure of the molecules, but also on their pharmacological activity and/or stability in humans (Jefferis, 2016; Walsh and Jefferis, 2006). PTMs are responsible for shaping the proper three-dimensional topology of polypeptides, enabling interaction with their cellular targets. It has become clear in recent years that glycosylations also have a substantial effect on the pharmacokinetics of many glycoproteins, specifically the elimination half-life from the blood circulation. For instance, desialylated glycoproteins have shorter half-lives in the circulation compared to the native form of the same glycoprotein, since they are cleared faster by specific hepatic receptors (Ashwell and Harford, 1982). Thus, supporting non-immunogenic human-permissive PTMs is imperative and represents a crucial factor in the clinical development of biological drugs.
CHO cell-mediated production of glycoproteins has many disadvantages despite their wide use and industrial popularity. This has encouraged the design of other enhanced cell-lines/systems. Disadvantages of CHO cells include lack of enzymatic ability needed to produce essential human modifications (such as α-2,6-sialylation and α-1,3/4-fucosylation), and deficient production of γ-glutamyl-carboxylation (required for several therapeutic proteins such as coagulation factors) (Kumar, 2015; Patnaik and Stanley, 2006). In addition, some glycoproteins that require a specific proteolytic cleavage for maturation are not properly processed for their biological activity in CHO cells (Fischer et al., 1995). Structural analyses show that CHO cell-produced glycoproteins contain several PTMs not present in humans. Although found at low levels, galactose-α1,3-galactose (α-gal) and N-glycolylneuraminic acid (Neu5Gc) were observed (Bosques et al., 2010; Ghaderi et al., 2010). If these human-foreign modifications exceed a certain threshold, they are often not compatible with clinical use and may trigger adverse immune responses against administered glycoproteins (Galili et al., 1984; Noguchi et al., 1995). Murine cells in general exhibit much higher levels of these glycans. Alternate systems have, therefore, been developed to overcome these problems in product quality. Widely used human cell-lines include the suspension adapted HEK293 (isolated from human embryonic kidney) and fibrosarcoma derived HT1080 cells (Swiech et al., 2012). Other non-human mammalian cells that have obtained regulatory approval for glycoprotein production include baby hamster kidney (BHK) (Durocher and Butler, 2009), NS0, and Sp2/0 murine myeloma cells (Barnes et al., 2000). The productivity of these cell lines is typically in one order of magnitude lower than CHO cells (Lalonde and Durocher, 2017).
Some proteins, including insulin, are manufactured in non-mammalian expression platforms, such as bacteria, yeast, and plants. Although relatively inexpensive and producing high titers, the main disadvantage of bacterial systems is their inability to introduce PTMs, such as N-linked glycosylation. Bacterial expression systems, for example, can produce high yields at relatively low cost. The produced proteins, however, tend to aggregate, as they lack the molecular chaperones needed for their proper folding (Graumann and Premstaller, 2006). Hence, this mode of production requires in vitro folding following protein isolation.
Yeast expression systems have been developed, but they produce high levels of mannose glycan, that may be immunogenic in humans (Dean, 1999; Gerngross, 2004). Plant and insect expression systems are used for glycoprotein manufacturing, however their endogenous glycosylation pattern is different from mammalians, and sophisticated engineering is needed to support human-like PTMs. In recent years, transgenic animals have emerged as a promising host for glycoprotein production. Genetically engineered transgenic goats (Kling, 2009), rabbits (Walsh, 2018), and chicken eggs (Harvey and Ivarie, 2003) have been utilized. These technologies are adequate only for a subset of therapeutic proteins, while CHO cells will likely remain the main platform for bio-therapeutic production.
As the need for therapeutic glycoproteins in the clinic is rising, many are investing in researching methods to maximize cell performance during production. While major improvements have been made in the last decade in cultivating mammalian cells for industrial purposes, these cellular platforms still suffer from insufficient growth, low specific productivity for hard-to-express proteins, and time-consuming cultures. This means an expensive production process especially for manufacturing difficult-to-express glycoproteins (Fischer et al., 2015). Therefore, genetic engineering and/or pharmacological intervention by targeting key effectors of biosynthetic/survival pathways have generated new strategies for controlling product quality and increasing harvested titers (Schmidt, 2004).
Increase in the produced titers is achieved either by increasing viable cell density (VCD) over time and/or enhancing cell-specific productivity (qp) (Josse et al., 2016). This can be achieved by media optimization, bioreactor engineering, genetic modification of the producing cells, and addition of pharmacological modifiers to the growth media (Fig. 1).
Currently employed approaches for increasing productivity of therapeutic glycoproteins.
Early efforts in cell engineering for the purposes of bio-production have focused primarily on cell survival and proliferation. Initially, anti-apoptotic genes were stably over-expressed, which improved biomass and titers (Fischer et al., 2015). These manipulations, however, have not provided a clear advantage over wild-type (wt) hosts. This lack of benefit is mainly attributed to improvements in cell media and advancements in bioreactor engineering, which enable commercial cells to proliferate to high densities with minimal evidence of post-production cell death. Nonetheless, developments in gene editing technologies have opened new possibilities to easily incorporate or delete multiple genes in mammalian cells (Adli, 2018; Bussow, 2015). Together, an understanding of cell functioning under normal and stress conditions, these novel possibilities facilitate cell engineering for the enhancement of growth and survival with metabolic advantages. In this respect, the PI3K/Akt/mTOR pathway is a central metabolic regulator. This review focuses on past and current strategies that have targeted the PI3K/Akt/mTOR signaling pathway to increase overall productivity.
The phosphatidylinositol 3-kinase/protein kinase-B/mammalian target of rapamycin (PI3K/AKT/mTOR) are interconnected signaling pathways, involved in essential physiological functions such as cellular growth, survival, metabolism, protein and lipid synthesis (Porta et al., 2014). Several activating mutations in the PI3K/AKT/mTOR signaling cascade have been identified, and they are intrinsically linked to the development of various types of cancers (Polivka and Janku, 2014). This signaling pathway has become an attractive target for genetic or pharmacological manipulations towards designing high-producing cell-lines given the diverse activities that positively influence cellular proliferation and protein synthesis.
PI3Ks are a lipid kinase family involved in the signal transduction initiation cascade (Porta et al., 2014). The canonical activation process of PI3Ks occurs primarily through stimulated receptor tyrosine kinases (RTKs) (Fig. 2). The activated PI3Ks facilitate the conversion of membrane-bound phosphatidylinositol-(4,5)- bisphosphate [PtdIns(4,5)P2; PIP2] to phosphatidylinositol-(3,4,5)- trisphosphate [PtdIns(3,4,5)P3; PIP3] (Yu and Cui, 2016). Once generated, the secondary messenger PIP3 recruits and activates various pleckstrin homology (PH) domains containing proteins such as PI3K-dependent kinase-1 (PDK1) and AKT. A key factor that regulates cellular PIP3 levels is the phosphatase and tensin homolog (PTEN) protein. PTEN is a phosphatase that cleaves the PIP3 activating phosphate group, converting it to its inactive form, PIP2. AKT is a serine/threonine kinase containing a PH domain. This structural domain directs AKT to the plasma membrane, where it is activated at different sites through phosphorylations by both PDK1 and mTORC2. The activated AKT is a potent signal transducer that phosphorylates substrates such as glycogen synthase kinase 3 (GSK3), tuberous sclerosis 2 (TSC2), and PRAS40 (AKT1S1).
The molecular regulation of PI3K/Akt/mTOR signaling pathway. Upon growth factor binding, the stimulated RTK undergoes autophosphorylation that facilitates PI3K activation by multiple mechanisms. The membrane-bound PIP2 is converted into PIP3 through PI3K mediated phosphorylation. Conversely, the secondary messenger PIP3 can be converted back into its inactive form PIP2 by PTEN. PIP3 recruits PH containing proteins and facilitates AKT dual phosphorylation by PDK1 and mTORC2. The activated Akt phosphorylates various substrates such as TSC2/TSC1 complex, FOXO, and GSK3 (not shown). As the TSC2/TSC1 complex is phosphorylated and blocked, RHEB can then lead to the activation of mTORC1. The active form of mTORC1 phosphorylates and activates S6K1. In parallel, mTORC1 also phosphorylates 4E-BP1, which leads to its detachment from eIF4E, enabling cap-dependent mRNA translation.
The main target of AKT is the activation of mTOR, a serine/threonine protein kinase that belongs to the phosphoinositide 3-kinase (PI3K)-related kinase family (Laplante and Sabatini, 2009). This signaling pathway senses diverse cellular and extracellular signal inputs and plays a vital role in maintaining cellular homeostasis such as cellular growth, metabolism, and survival. Two structurally distinct complexes of mTOR called mTORC1 and mTORC2 have been identified in mammalian cells. Each complex interacts with a specific set of substrates (Liu et al., 2015). Since their downstream effectors are distinct, mTORC1 and mTORC2 are responsible for different physiological functions. The activities of mTORC1 have generated extensive research. While more than thirty different substrates of mTORC1 have been documented (Laplante and Sabatini, 2009), only AKT in a positive loop manner has been conclusively shown to be a substrate of mTORC2 (Sabatini, 2017). Even so, the exact physiological relevance of mTORC2 to proliferation and metabolism is not fully understood.
mTORC1 is a key regulator of cellular growth, protein translation, autophagy, and metabolism (Laplante and Sabatini, 2009). By sensing nutrient availability and other upstream stimulatory inputs, active mTORC1 can elicit phosphorylation-based signals to its downstream effectors. Mechanistically, the AKT-mediated phosphorylation of TSC1/2 complex prevents the binding to RAS homolog enriched in brain (RHEB), which consequently activates mTORC1. Once activated, mTORC1 directly phosphorylates its downstream substrates. Two canonical substrates have been identified, eIF4E binding proteins (4EBP) proteins and S6K1. The eIF4E binding protein-1 (4EBP1), a translation inhibitory protein that binds to and sequesters the eIF4E initiation factor, disrupts the translation initiation process. Upon its phosphorylation, 4EBP1 releases the initiation factor eIF4E to interact with eIF4G and allow cap-dependent mRNA translation. This activates protein synthesis (Ma and Blenis, 2009; Richter and Sonenberg, 2005). Besides 4EBP-1, mTORC1 also leads to the phosphorylation of S6 kinases (S6Ks) and S6 proteins that participate positively in cellular growth/survival, polypeptide translation and ribosomal biogenesis processes (Ma and Blenis, 2009; Nawroth et al., 2011). It is, therefore, not surprising that mTORC1 activation is associated with polysome formation (Gentilella et al., 2015).
In addition to proteins, intracellular lipids and nucleotide pools are essential for cellular growth and homeostasis maintenance (DeBerardinis et al., 2008). Lipids serve as the building blocks of biological membranes, a source of energy, and as precursors of diverse signaling molecules (Saxton and Sabatini, 2017). Several studies demonstrate that active mTORC1 promotes de novo lipid biosynthesis by activation of sterol responsive element-binding protein (SREBP) transcription factors that regulate key metabolic genes involved in cholesterol and fatty acid biosynthetic processes (Porstmann et al., 2008). Nucleotide availability plays a vital role during cellular proliferation, as nucleotides are required for DNA replication, RNA formation, and ribosome biogenesis (Saxton and Sabatini, 2017). Various studies show that de novo nucleotide biosynthesis is positively regulated by mTORC1. It has been suggested that activated S6K1 phosphorylates activate carbamoyl-phosphate synthetase (CAD), a key component that participates in the pyrimidine biosynthetic process (Ben-Sahra et al., 2013). Other studies show that mTORC1 activates MTHFD2, which is an important participant in the mitochondrial tetrahydrofolate cycle responsible for purine biosynthesis (Ben-Sahra et al., 2016). Thus, mTORC1 activation promotes the biogenesis of all major cellular components for proliferation.
Mounting biochemical evidence indicates a correlation between mTORC1 activity and cell line productivity. Comparisons between low and high productivity cells suggest a correlation between productivity and polysome formation, a hallmark of mTORC1 activity (Godfrey et al., 2017). The eIF4E/4E-BP1 stoichiometry positively correlates with cell productivity (Josse et al., 2016). These data indicate that manipulations of the PI3K/AKT/mTORC1 signaling may enhance r-proteins yields.
One of the typical methods to constitutively activate PI3K/AKT/mTOR pathway was achieved through mTOR over-expression (Dreesen and Fussenegger, 2011). Using CHO-K1 cells, ectopic over-expression of human mTOR has been investigated for increased production of r-proteins. Expression of mTOR increases cellular viability, proliferation, and overall titers (Dreesen and Fussenegger, 2011). When transiently expressed, mTOR leads to an approximately three-fold increase in the human placental SEAP and secreted amylase (SAMY) glycoproteins, while stable mTOR-expressing cells showed a more significant increase estimated as five or seven-fold in SEAP and SAMY production titers.
The over-expression of AKT in its active form is another strategy (Hwang and Lee, 2009) utilized for over-activation of the PI3K/AKT/mTOR pathway. AKT is a serine/threonine kinase that regulates multiple looping and dynamic downstream signaling molecules such as GSK3, FOXO, and mTORC1 (Manning and Toker, 2017). AKT is also involved in various cellular functions such as survival, proliferation and protein synthesis. While exploring the role of AKT signaling in cancer, a constitutively active version of AKT was generated by engineering a myristoylation site in its N-terminus (Kohn et al., 1996). When introduced into CHO cells and cultivated in a nutrient-limited medium, this ligand-independent active variant did not alter proliferation behavior, but conferred a 30% increase in overall titer in this study of a monoclonal antibody against a human platelets receptor (Hwang and Lee, 2009). Although the authors did not directly investigate the effect of AKT over-expression on mTORC1 activity and protein translation capability, it is reasonable that this increase in productivity is related to these effectors.
In addition to the overexpression of positive mTORC1 regulators, inhibitors of mTORC1 can be removed to increase its output. Accordingly, compromise of the main inhibitors of mTORC1, the tuberous sclerosis complex comprised of TSC1 and TSC2, results in an increase mTORC1 activity in multiple cell types. Removal of TSC2 by CRISPR/Cas9 in CHOZN (GS KO CHO cells) cells resulted in an increase in cell size and elevated the translation capacity, which led to an increase approximately two-fold of the monoclonal antibody under fed-batch conditions (McVey et al., 2016). TSC2 deletion, however, also impeded the buildup of biomass, which compromised its utility. This suggests that exaggerated mTORC1 activity is not desired during the cell growth phase. In this particular study PTEN deletion did not show a significant increase in mTORC1 output and in the produced titers. Since the effect of PTEN deletion was examined only in CHOZN cells, it is worthwhile to investigate its role in other expression systems such as HEK293 or HT1080 cells, where PTEN perturbation may play a more prominent role.
While these strategies specifically targeted the PI3K/AKT/mTORC1 pathway, applying simple genetic modifications that simultaneously activate multiple pathways may result in better performance, provided that these pathways synergize. Receptor tyrosine kinases (RTKs) are cell-surface receptors involved in diverse cellular processes such as growth and survival. In addition, various activation mutations or up-regulation of these receptors are involved in many types of cancer (Regad, 2015). Stimulation of RTKs activates multiple signaling pathways such as MAPK, Src, and PI3K/AKT/mTOR (Lemmon and Schlessinger, 2010). For therapeutic purposes, the producing cells are cultured in serum-free and chemically defined medium that generally does not contain sufficient RTK ligands or other necessary growth factors, a limitation that may adversely affect cell performance and may lead to apoptosis (del Castillo et al., 2008). Introduction of ligand-independent and constitutively active RTKs into producer cells has been explored as a genetic tool to compensate for the cell`s under-activated signaling pathways. KIT, also known as CD117, is a type-III tyrosine kinase that may possess the D816V mutation found in many cancer types (Longley et al., 2001). Since this oncogenic version of KIT is constitutively active in a ligand-independent manner, it has been suggested that over-expression of this mutant in CHO producing cells may serve as a simple genetic strategy to boost their production capacity (Mahameed and Tirosh, 2017). A significant increase in proliferation rates was observed, when D816V-KIT was stably expressed in producing CHO cells, and the cultured cells were resilient to cultivation-encountered stressors (such ER stress, starvation and hypoxia). It has been shown that the D816V mutant resulted in an almost two-fold increase in harvested titers during a batch assay. This oncogenic version of KIT promotes the overall protein translation efficiency through elevation in S6 and 4EBP1 phosphorylation levels, presumably through mTORC1 activation (Mahameed and Tirosh, 2017).
Rapamycin is a high-affinity allosteric inhibitor of mTORC1 used to genetically clone the complex initiated by Hall and coworkers (Benjamin et al., 2011). Initial studies that addressed the role of mTORC1 in bio-production used rapamycin. These analyses yielded contradicting observations. In some systems the addition of rapamycin enhanced productivity primarily by improving cell survival. In this particular study inclusion of rapamycin elevated mAB titers almost twofold (Dadehbeigi and Dickson, 2015). When 4E-BP1 and S6K1 phosphorylation status was investigated during a 10-day fed-batch as mTORC1 activation read-outs, it was shown that rapamycin treatment initially decreased phosphorylated levels of these downstream targets, but later their phosphorylation status was significantly increased by an unexplained mechanism. In other studies rapamycin strongly reduced productivity by affecting the specific productivity (Courtes et al., 2014). The underlying reasons for these discrepancies are complex and probably related to the fact that mTORC1 suppression improves the survival of cells under stress conditions and allows the cells to preserve energy and recycle nutrients by autophagy. When growth conditions are better, and the producing line is supplemented with optimal media, reduction in mTORC1 activity results in polysome disassembly and reduced productivity.
Since mTORC1 is a prominent target for cancer and other pathologies, various inhibitors have been developed. There is however, no specific compound that activates the complex, even though some (such as MHY1485 and vanadate derivatives that inhibit phosphatases including PTEN) are reported to have these properties (Choi et al., 2012; Spinelli et al., 2015). None of these molecules has been tested for enhancing bio-production. It should be emphasized that a molecule must be inexpensive and exhibit a significant benefit for a large array of molecules to be incorporated into a medium as an instigator of productivity. From a process development perspective, this approach is attractive as it obviates the need to alter the host.
Cadmium chloride is another chemical additive that increases mammalian cell production in a variety of expression systems (Mahameed et al., 2019). Few sub-micromolar concentrations of this heavy metal elevate the phosphorylation status of 4E-BP1 and hence, improve the overall translational capacity in a dose-dependent manner. When cultured in the presence of 0.5 μM cadmium chloride, the production of a secretable fluorescent protein was doubled compared to untreated cells. It has been suggested that this bivalent-cation is a potent inhibitor of PPM1, a phosphatase that cleaves the phosphate group from 4E-BP1, which may explain its effect on protein synthesis.
Elevated intracellular adenosine levels also enhance productivity of recombinant proteins (Chong et al., 2009). An increase of 1.4-fold in the titers and 2.5-fold in the average specific productivity in interferon-gamma (IFNα) has been shown, when 1 mM of adenosine is added to the culturing medium. The exact mechanism of how adenosine affects protein production has not been fully delineated, but it has been suggested that high adenosine levels increase the intracellular ATP concentration and cause an increase in the phosphorylation level of 4E-BP1 through unknown mechanisms.
In addition to small molecules, r-proteins that are prepared at low costs in bacteria are used to improve CHO performance. Typically, the addition of exogenous growth factors such as insulin, insulin-like growth factor (IGF)-1, and transferrin have been employed in medium design processes (Miki and Takagi, 2015; Yun et al., 2003). It has been suggested that the addition of these growth factors increases the production titers and inhibits cellular apoptosis during the culturing period. As described above, these components are normally required for RTK activation that lead, among other effects, to the activation of PI3K/Akt/mTOR signaling pathway.
One of the challenges of upstream cell line development is the identification of clones appropriate for large scale production. These clones must possess key features such as stability, fast doubling time, and high specific productivity. This is a laborious process that requires the evaluation of a large number of single cell clones, and it is dependent on a strong and efficient selection process. Thus, host cells that resist apoptosis may grow better and secrete more. These cells, however, may impede the selection process, resulting in a large number of clones that will grow well but secrete low amounts of the r-protein.
In contrast to inhibiting the apoptosis process, the PI3K/AKT/mTOR pathway is an attractive process for manipulation owing to its subtle effects on cell survival. In this review, we lay out the current strategies for increasing this signalling output for improved bioproduction (Fig. 3). Our TSC2 deletion experiments clearly show that this manipulation should be done in a bi-phasic manner. A strong mTORC1 activation in the logarithmic phase prior to reaching high VCD may impede growth and dramatically change the nutritional requirement of the cells. At the stationary phase, when cells cease to proliferate and increase their secretory properties, instigating the PI3K/AKT/mTOR pathway may have a significant effect on productivity. Cell engineering and media additives that allow this bi-phasic manipulation should offer a significant advantage for bio-production.
Genetic and/or pharmacological interventions that lead to PI3K/Akt/mTOR over-activation. 1. Growth factor addition; 2. RTKs overexpression; 3. PTEN KO/inhibition; 4. AKT overexpression; 5. TSC1/2 KO 6. mTOR overexpression; 7. PPM1 inhibition.
Mohamed Mahameed is the recipient of the Zvi Yanai doctoral fellowship of the Israeli Ministry of Science and Technology. Boaz Tirosh holds the David Eisenberg Chair in Pharmacy of the Hebrew University and is affiliated to the David R. Bloom Center for Pharmacy and the Dr. Adolph and Klara Brettler Center for Research in Pharmacology.
