Pharmaceutical Biotechnology: Drug Discovery and Clinical Applications

Part One

Concepts and Methods for Recombinant Drug Production

1

Pharmaceutical Biotechnology and Industrial Applications – Learning Lessons from Molecular Biology

Oliver Kayser and Heribert Warzecha

1.1 Introduction

To date, biotechnology has produced more than 200 new therapies and vaccines, including products to treat cancer, diabetes, HIV/AIDS, and autoimmune disorders. There are more than 400 biotech drug products and vaccines currently in clinical trials, targeting more than 200 diseases, including various cancers, Alzheimer’s disease, heart disease, diabetes, multiple sclerosis, AIDS, and arthritis. These few figures demonstrate the importance of biotechnological methods and techniques, which are increasingly dominating the process of drug research and development [1].

An average approval of 10–15 products a year indicates that pharmaceutical biotechnology is a highly active sector. Amongst these, the number of genuinely new biopharmaceuticals is around 40%, indicating the high innovative character of research; some of these products are likely to be future blockbusters (Table 1.1). Examples are monoclonal antibody-based products such as Rituximab (Rituxan®/MabThera®) for the treatment of cancer with $18 billion in sales in 2009, insulin and insulin analogues ($13.3 billion/2009), and finally erythropoietin-based products ($9.5 billion/2009). The global market is growing by 7% per year for protein-based therapeutics and among all blockbuster drugs only one is a classical low molecular drug, the other four top selling drugs (Table 1.2) are derived from the biotechnology sector [3]. In addition to new drug entities (NDE), biosimilars or follow-up-biologicals will continue to increase in market value; this is the focus of Chapter 13. This trend is supported by new or adapted approved routes from the regulatory bodies such as the EMA (European Medicines Agency) and the FDA (Food and Drug Administration) (see Chapter 11).

Table 1.1 Classification of recombinant proteins for human use

(according to [1]).

a) No longer available.

Table 1.2 The ten top selling recombinant proteins for human use in 2010

(source: LaMerie Business Intelligence, Barcelona [2]).

Established molecular biology techniques for protein engineering, such as phage display, construction of fusion proteins or synthetic gene design, have matured to the level where they can be transferred to industrial applications in recombinant protein design. Traditional engineering has focused on the protein backbone, while modern approaches take the complete molecule into account. We want to discuss recent advances in molecular engineering strategies that are now paying off with respect to engineered proteins with improved pharmacokinetic and pharmacodynamic profiles, as reviewed in Chapter 14. In designing muteins, glycoengineering and post-translational modification with non-natural polymers such as polyethylenglycol (PEG) have affected around 80% of approved protein therapeutics [1].

1.2 Research Developments

1.2.1 Protein Engineering

The term protein engineering refers to the controlled and site specific alteration of a gene sequence encoding the transcription to a polypeptide to a mutated protein with introduced changes in the amino acid sequence. In principle, deletions and insertions of one or more triplet codes and amino acids are possible, but mostly alteration of a protein sequence is limited to exchange of amino acids at calculated sites. Since the first experiments in molecular biology to obtain insights into diseases, protein engineering has been introduced successfully into drug development of recombinant proteins to improve pharmacodynamics and pharmacokinetic profiles [4]. At the biotechnology level, tailoring of proteins has been documented for commercially relevant proteins such as insulin, erythropoietin, growth hormones, and various antibodies. Today the important objectives for protein engineering are:

improving the pharmcodynamic profile to obtain a drug that acts faster or slower;

alteration of the pharmacological half-life and development of controlled release kinetics;

alteration of receptor binding specificity;

reducing the immunogenicity of the protein;

increasing physical and chemical protein shelf half-life.

From the 25 genuine new biological entities (NBEs) approved in Europe and the USA until 2009, 17 proteins have already been engineered. The dominant group are antibodies (11), and of these six are fully human, and one is bispecific (Revomab®); out of 25 drugs 17, or in other words around 70%, are modified from a total number of 25 NBEs, and four are humanized antibodies. Among the 25 products, two are fusion proteins (rilonacept, Arcalyst and romiplostim, Nplate). Romiplostin is a so-called peptibody consisting of the Fc fragment of the human antibody IgG1 and the ligand-binding domains of the extracellular portions of the human interleukin-1 receptor component (IL-1RI). It is used for the treatment of Familial Cold Auto-inflammatory Syndrome (FCAS) or Muckle-Wells Syndrome (MWS). Interestingly the functional domain consists of peptide fragments designed by protein modeling to bind highly specifically on the thrombopoetin receptor.

1.2.2 Muteins

Based on the genetic code, a significant number of proteins, which have been approved for clinical use, are subjected to directed change and amino acid substitution to improve the pharmacokinetic and pharmacodynamic activity, and also to develop antagonist functionality. These derived proteins with site directed mutations are called muteins and show interesting pharmacological features, which is why a bright future is in prospect. As in classical recombinant biotechnology, insulin was the first candidate with site directed mutations. Insulin lispro was approved in May 1996 as the first mutein, and only a few months later, in November 1996, Reteplase was also approved as a tissue plasminogen activation factor. The number of muteins has since increased significantly and is now dominated by recombinant antibodies. Briefly we want to discuss the potential of muteins for analogs of insulin, tissue plasminogen activator (tPA), and humanized antibodies.

Native insulin associates from dimers up to hexamers at high local concentrations are what are usually found at the site of injection, leading to retarded dissolution and activity in the body. As a result of structure elucidation, proline and lysine at positions 28 and 29, respectively, in the B chain were identified to play a crucial role and were therefore subjected to site directed mutagenesis. Switching B28 and B29 of proline and lysine reduced the association affinity 300-fold, resulting in faster uptake and action, as well as shorter half-life [5]. In contrast, to increase the time of action towards a retarded drug delivery profile, the same concept of site-directed mutation was also applied. Insulin glargin (Lantus®) is a mutein where in the A chain A21 glycine is introduced instead of asparagine, and in the B chain two more ariginines are added at the C-terminal end [6]. As a physicochemical consequence, the isoelectric point is shifted towards the physiological pH at 7.4, resulting in precipitation and slow dissolution into the blood stream.

Tissue plasminogen activators (tPA) play an important role in the breakdown of blood clots. As with insulin, tPA is converted from plasminogen into plasmin, the active enzyme responsible for clot breakdown. tPA is manufactured by recombinant biotechnology, and is used extensively in clinics, but a disadvantage is fast elimination from the body. To overcome this problem a deletion mutant was constructed to reduce binding of the protein at hepatocytes via the EGF-domain (epidermal growth factor) encoded by an amino acid sequence starting from position 4 to 175. The remaining 357 of the 527 amino acids in Reteplase (Retavase®, Rapilysin®) showed increased half-lives of 13–16 min and, interestingly, increased fivefold activity [5, 7]. The historic development with a brief outline of the near future, for example, non-invasive delivery systems, has been described well by Heller et al. [8].

The beauty of antibodies can be addressed through the ability of binding to highly specific surface structures and a fairly uniform structure. Apart from vaccinations, antibodies were introduced early on in the therapy of neoplastic diseases and for the prevention of acute tissue rejection in patients with organ transplants. Muromonab CD3, with the tradename Orthoclone OKT3®, is an immunosuppressant monoclonal antibody that targets the CD3 receptor on the surface of T cells. It is approved to prevent acute rejection of renal transplants. As an adverse reaction, anti-mouse antibodies can be formed leading to reduced efficacy after repeated injection. To improve tolerance, chimera between mouse and humans were designed. From the protein sequence of the established murine antibodies, the genetic code was deciphered and substituted in the conserved Fc region by the respective human genetic code. These antibodies are called chimeric, in contrast to humanized antibodies where the framework regions are also substituted. Examples are Daclizumab, Zenapax (humanized) [9], Abciximab in ReoPro® (chimeric) [10], and Rituximab in Mabthera® (chimeric) [11] as antineoplastic antibodies for non-Hodgkin lymphoma.

1.2.3 Post-translational Engineering

Several approved recombinant therapeutic products are engineered post-biosynthesis. From the molecular biology background, post-translational engineering is associated with glycosylation or lipidation post-biosynthesis. Post-translational biosynthesis today is the covalent attachment of a chemical group, not a mandatory glycosylation, but attaching fatty acids or PEG-chains alteration of a pre-existing post-translational modification, and has been reviewed best by Walsh [9]. Novo Nordisk′s Victoza® (liraglutid) is an example of a non-insulin once-daily medication that may help improve blood sugar levels in adults with type II diabetes. It contains the glucagons-like peptide 1 (GLP-1) analog with 97% sequence homology and with an attached C16 fatty acid (N-ε-(γ-Glu[N-α-hexadecanoyl]) at Lys26 [10].

Glycosylation is the most complex and widespread form of post-translational modification. Glycoengineering therefore becomes of greater interest, and by directed and targeted alteration of the glycosylation pattern at the protein backbone, significant changes tof the pharmacokinetic profile can be enforced. Approximately 40% of the approved proteins are glycosylated and the use of mammalian cell lines is dominating the manufacturing process (e.g., Chinese hamster ovary (CHO) cells, baby hamster kidney (BHK) cells). A recent trend in engineering the glycocomponent is to also use plant systems (such as carrot cells for β-glucocerebrosidase) or Saccharomyces cerevisiae and Pichia pastoris. The production of the glucocerebrosidase analog imiglucerase (Cerezyme®) for the treatment of Morbus Gaucher has been carried out in CHO cells. Alternatively, recent interesting advances by the company Protalix showed that glucocerebrosidase for oral administration can be produced in the carrot cells (Daucus carota, Apiaceae) (Figure 1.1) [11]. Oral glucocerebrosidase is a plant cell-expressed form that is naturally encapsulated within carrot cells, which are genetically engineered to express the enzyme. Plant cells have the unique attribute of a cellulose cell wall that makes them resistant to enzyme degradation when passing through the digestive tract, which is the main idea behind the oral administration concept. In plant systems, one of the most notable biotransformation reactions has been developed for genetically modified mosses. The Heilbronn, Germany-based company Greenovation has constructed a moss (Physcomitrella patens) lacking in xylose and fucose transferase activity. Biolex Therapeutics, Pittsboro, NC, USA, has an alternative system in engineered duckweed (Lemna minor), where fucosyl and xylose transferase activity is inhibited by RNA interference (RNAi) mechanisms [12].

Figure 1.1 Production of glucocerebrosidase with Daucus carota plant cell suspension cultures at Protalix.

In contrast to the endogenous human glucocerebrosidase, imiglucerase (Cerezyme®), which is naturally glycosylated and produced in a CHO cell line and downstream processing, includes an enzyme-based processing step using an exoglycosidase. Imiglucerase must be biochemically modified by cutting off capping oligosaccharide chains (sialinic acid, galactase, and N-acetyl-glucosamine) down to mannose by exoglycosidases (neuraminidase, β-galactosidase, N-acetyl-glucosaminidase). Exposing the remaining mannose residues facilitates specific uptake by macrophages via macrophage cell surface mannose receptors. In this way the enzyme is taken up by macrophages in a very efficient way. Unmodified glucocerebrosidase, if administered, is quickly removed from the bloodstream in the liver.

A non-natural post-translational modification is PEGylation of the protein backbone. Polyethylene glycol (PEG) is a more frequently used technique to alter the physical, chemical, and biological profile of the desired protein [13]. PEGylated proteins and peptides have found promising applications in pharmaceutical biotechnology and related biomedical areas:

to improve solubility,

to improve thermal and mechanical stability,

to reduce immunogenicity,

to reduce renal excretion and clearance,

to protect proteins from degradation such as proteolysis, and

to optimize pharmacokinetic properties such as increased blood circulation and extend plasma half-lives.

It started initially with PEGylated interferons (Pegasys®, Viraferon®), but now four more recombinant proteins (Somavert®, Neulasta®, Oncaspar®, Mircera®) have been approved. Pegvisomant (Somavert®) is a PEGylated analogon of the human growth hormone (hGH) that is produced in Escherichia coli. Four to five PEG chains are attached to the protein backbone to form a hydrodynamic shell surrounding the protein, giving an improved solubility and longer half-life. Neulasta® and Oncaspar® are two more growth hormones, but Mircera is a PEGylated Erythropoietin analogon. Consequently, PEGylation has gradually become a platform technology in pharmaceutical technology. A detailed outline of the new formulation strategies is presented in Chapter 10. The chemistry of protein and peptide PEGylation has attracted more and more attention as further PEG-conjugates have reached late phase clinical trials. The discovery and development of upcoming recombinant proteins with undesirable biopharmaceutical hurdles makes PEGylation an attractive approach to drug formulation. New routes to site specific PEGylation and new reversible PEGylation are likely concepts in the near future, as discussed in Chapter 11.

1.2.4 Synthetic Biology

Synthetic Biology is a new emerging field in gene technology and system biology. In the continuation of metabolic engineering research, this new discipline tries to integrate engineering, nanobiotechnology, genetics, and bioinformatics [14]. Key enabling technologies have their roots in molecular biology and genetics. The concept behind synthetic biology is to abstract the hierarchical order and to allow standardization of biological devices, also called biobricks, such as promotors, transcription factors for use in complex biological systems. Synthetic biologists rely on massive DNA sequencing, protein engineering, and final assembly of designed biobricks to fabricate a production host of interest. A high and constantly increasing number of genomes have been sequenced, but valuable information regarding plant and microorganism genes encoding, for example, the diverse secondary natural product metabolism, are still limited. Despite the fact that synthetic biology is in its infancy, today’s achievements are impressive. In 2000, researchers at Washington University, USA, reported synthesis of the 9.6 kbp Hepatitis C virus genome from chemically synthesized 60- to 80-mers. In 2002, researchers at SUNY, Stony Brook, USA, succeeded in synthesizing the poliovirus genome from its published sequence and producing the second synthetic genome. The first bacterial genome was assembled in 2006 by scientists at the J. Craig Venter Institute. Mycoplasma laboratorium is derived from M. genitalium and contains a minimal synthetic genome, allowing complete functionality in a living cell as host [15].

Synthetic biology is receiving more and more interest for pharmaceutical applications. In 2004, artemisinic acid, as a precursor towards the biosynthesis of artemisinin, an important antimalarial drug, was successfully transferred into E. coli and later in 2006 into S. cerevisiae. In 2007 the plant derived kaempferol and quercetin were heterologously synthesized in E. coli and on average two secondary natural products per year can be added to the list of combinatorial biosynthetic compounds. The role of secondary natural products is clearly highlighted in Chapter 2. Synthetic biology is still in its infancy and has not been exposed to wide use in the pharmaceutical laboratories [16]. However, the challenges and opportunities are clear and range from host design to producing non-natural chimeric recombinant proteins and also low molecular drugs (see Chapter 19), adapting the host for improved process design in pharmaceutical and chemical engineering, and to enforcing personalized medicine (see Chapter 21). Today it cannot be predicted that nucleic acids as drugs and somatic gene therapy will also benefit, and that synthetic biology may shape the road to the design of new safe gene delivery systems.

1.3 Production Hosts and Upstream/Downstream Processing

A close look at the production organisms of the approved biopharmaceuticals over the last five years reveals that out of 58 products, 32 are produced in cells derived from mammalian organisms (Chinese hamster ovarium, CHO) (see Chapter 3), 17 in E. coli, four in Saccharomyces cerevisiae, and two in transgenic animals (see Chapter 5). New production hosts have entered the stage, such as Pichia pastoris (Ecallanttide, Kalbitor), a baculovirus-insect cell based system (Cervarix®), and Daucus carota for the production of a human glucocerebrosidase (imiglucerase alpha) (see Chapters 4 and 19), and antithrombins (Atryn®, Macugen®) in goats.

To improve product efficiency of the host systems used in these times of increased competition with upcoming biosimilars, reduced budgets costs, and market-price setting by public healthcare reforms, expression levels must be improved. For recombinant proteins in mammalian production systems, a yield of 5 g/l is considered to be a standard level. In the future, yields far above the typical levels of today could be achieved by construction of high producer cell lines, where we will definitively see the impact of synthetic biology and smart metabolic engineering. System biotechnology will allow rational process design to identify and overcome metabolic bottlenecks and media optimization. From the total costs involved in a drug manufacturing process, up to 80% can be considered to be the downstream processing (see Chapter 8). This is due to extraction and purification of single compounds from a complex metabolic broth and the increasing biosafety aspects in the highly regulated GMP (Good Manufacturing Practice) Pharma environment. Process-scale columns can cost more than US$1 million, depth-filter sets up to US$30 000, and centrifuges for biomass separation more than half a million US$.

In recent years, disposable bioreactors or single-use bioreactors have been introduced into pharmaceutical manufacturing. Disposable bags consisting of biocompatible ethylene vinyl acetate–polyethylene copolymers, are γ-radiated for sterilization and are available from 10 up to 2000 l. Apart from GE Healthcare, Xcellerex, Millipore, and Thermo Scientific, only Satorius Stedim Biotech offers a satisfying solution for a complete production line. Disposable biobags have to be certified by drug authorities and validated, as we know from experience with the traditional steel and glass bioreactors. Besides the minor ecological aspects, the eco-efficiency balance is not negative, and disposable biobags fulfill all requirements for GMP production, but it is doubtful if they will become accepted in companies who have invested heavily in running a steel-based infrastructure.

Synthetic biology has already arrived in the Pharma industry and is improving biosynthetic processes. Two examples may illustrate the potential of metabolic engineering and synthetic biology to influence bioprocesses in the future. DSM, a Dutch biotech company engaged in natural product, food additives, and antibiotic production, has improved the existing process for the commercial production of cephalexin. By cutting out 13 chemical steps and replacing them by biotransformation, a new innovative process with significant energy and cost savings has been established. The main metabolic engineering involved the introduction and heterologous expression of acyl transferase and expandase for the direct fermentation of dipoyl-7-aminodesacetoxycephalosporanic acid [15]. Sitagliptin, a dipetidyl peptidase-4 inhibitor, is a synthetic compound for the treatment of type II diabetes. Codexis, a company headquartered in the USA, has developed a biocatalytic process using transaminase for producing this compound with a higher degree of stereoselectivity than the existing organic synthesis processes, which use a metal catalyst. Codexis won the US Presidential Green Chemistry Challenge Award from the US Environmental Protection Agency (EPA), and showed how synthetic compounds can benefit from metabolic engineering and re-engineering strategies, and tailoring biocatalysts for non-natural substrates [17].

1.4 Future Outlook

The development and production of therapeutic proteins represents the first truly industrial application of recombinant DNA technology. At the beginning of the biotech revolution, the main goal was the expression and efficient production of recombinant proteins as known from humans. The success story of insulin documents this process in detail. Based on the urgent need to accommodate the demand for insulin in the world, highly sophisticated manufacturing units have been built over the last two decades. Today, sufficient amounts are available and the production is virtually free from contamination risks, from pathogens or prions for example. Advances in protein sciences, genetics, and molecular biology have provided new opportunities to the production of tailored recombinant proteins to meet the demands of better disease management and more specific active drugs.

In the future, the design of engineered proteins will become more complex and will be specified by synthetic biology techniques, allowing de novo protein design in silico, and also better designed integrated manufacturing processes. Innovation for the pharmaceutical industry is based on innovative and safe drugs, but also cost effectiveness and performance, even if the product can be sold at a higher price in comparison with that from other industries. Embedding synthetic biology and integrating biotechnology and genomic sciences in the whole drug development process allow companies to save up to US$300 million per drug – about one third of the costs today – and the prospect of bringing the drug onto the market one or two years earlier. Each day lost before market entry leads to a loss of approximately US$1.5 million per day, indicating the value of efficient and optimized research and operational strategies.

Synthetic biology arose from combined activities between (bio)engineers, biophysicists, and computer scientists, but today the integration of clinicians is essential to allow successful transfer into clinical applications. Furthermore, microorganisms were the playground for synthetic biological experiments, but the move towards mammalian cells is necessary to prove developed circuits and constructs for the patient [18]. In Chapters 6 and 7 actual trends and drugs in the approval pipelines are highlighted and discussed extensively.

References

1 Walsh, G. (2010) Biopharmaceutical benchmarks 2010. Nat. Biotech., 28, 917.

2 La Merie Business Intelligence (2010) R&D Pipeline News. Special edition, March 2010. http://www.pipelinereview.com (accessed 10 November 2011).

3 Goodman, M. (2009) Sales of biologicals to show robust growth to 2013. Nat. Rev. Drug Discov., 8, 837.

4 Singh, R. (2011) Facts, growth and opportunities in industrial biotechnology. Org. Process Res. Develop., 15, 175.

5 Pancholia, A.K., Sambi, R.S., and Krishna, C.K. (2009) Current problems with thrombolytic agents in the management of acute myocardial infarction. Indian Heart J., 61 (5), 476–479.

6 Berti, L., et al. (1998) The long acting human insulin analog HOE901: characteristics of insulin signaling in comparison to Asp(B10) and regular insulin. Hormon. Metab. Res., 30, 123.

7 Simpson, D., Siddiqui, M.A., Scott, L.J., and Hilleman, D.E. (2006) Reteplase: a review of its use in the management of thrombotic occlusive disorders. Am. J. Cardiovasc. Drugs, 6, 265.

8 Heller, S., Kozlovski, P., and Kurtzhals, P. (2007) Insulin’s 85th anniversary – an enduring medical miracle. Diabetes Res. Clin. Practice, 78, 149.

9 Walsh, G. (2010) Post-translational modifications of protein biopharmaceuticals. Drug Dis. Today, 15, 773.

10 Rebel, U., et al. (2004) NN2211: a long-acting glucagon-like peptide-1 derivative with anti-diabetic effects in glucose-intolerant pigs. Eur. J. Pharmacol., 451, 217.

11 Shaaltiel, Y., Bartfeld, D., Hashmueli, S., Baum, G., Brill-Almon, E., Galili, G., Dym, O., Boldin-Adamsky, S.A., Silman, I., Sussman, J.L., Futerman, A.H., and Aviezer, D. (2007) Production of glucocerebrosidase with terminal mannose glycans for enzyme replacement therapy of Gaucher’s disease using a plant cell system. Plant Biotechnol. J., 5, 579.

12 Decker, E.L. and Reski, R. (2007) Moss bioreactors producing improved biopharmaceuticals. Curr. Opin. Biotechnol., 18, 393.

13 Jevsevar, S., Kunstelj, M., and Porekar, V.G. (2010) PEGylation of therapeutic proteins. Biotechnol. J., 5, 113.

14 Tyo, K.E., Alper, H.S., and Stephanopoulos, G.N. (2007) Expanding the metabolic engineering toolbox: more options to engineer cells. Trends Biotechnol., 25, 132.

15 Gibson, D.G., et al. (2008) Complete chemical synthesis, assembly, and cloning of a Mycoplasma genitalium genome. Science, 319, 1215.

16 Aldrich, S.C., Newcomb, J., and Carlson, R. (2007) Genome Synthesis and Design Futures: Implications for the U.S. Economy (Bio-Economic Research Associates [bio-era], LLC, Stockbridge, VT, 2007). http://www.bio-era.net/reports/genome.html (accessed 17 January 2012).

17 Lutz, S. (2010) Biochemistry. Reengineering enzymes. Science, 329, 285–287.

18 Ruder, W.C., Lu, T., and Collins, J.J. (2011) Synthetic biology moving into clinics. Science, 333, 1248.

Weblinks

Biotechnology Industry Organization: Washington, DC Everyday products, synthetic biology report, Current Uses of Synthetic Biology for Chemicals and Pharmaceuticals http://bio.org/ind/syntheticbiology/Synthetic_Biology_Everyday_Products.pdf (accessed 17 January 2012)

Landau, R. and Arora, A. (October 1999) The chemical industry: from the 1850s until today. Retrieved from Business Services Industry: http://findarticles.com/p/articles/mi_m1094/is_4_34/ai_56973853/?tag=content;col1 (accessed 17 January 2012)

US Presidential Green Chemistry Challenge Award for Sitagliptin synthesis. http://www.pharmamanufacturing.com/industrynews/2010/130.html (accessed 17 January 2012)

Atryn, the first approved antothrombin from transgenic goats. http://news.bbc.co.uk/2/hi/science/nature/4746736.stm (accessed 17 January 2012)

2

Prokaryotic Cells in Biotech Production

Andriy Luzhetskyy, Gabriele Weitnauer, and Andreas Bechthold

2.1 Introduction

The origins of producing natural products from microorganisms extend back to individual impressive scientists and individual companies (e.g., Sandoz, Ciba, Hoechst, Bayer, Eli-Lilly). In 2010 natural products and enzymes and proteins produced by prokaryotes made up a large proportion of the pharmaceuticals production. They are most prominent in the antibiotics sector but examples can be given from nearly all disease areas. The most important producers of natural products are actinomycetes [1], myxobacteria [2], cyanobacteria [3], and marine sponges [4], and biopharmaceuticals are mostly produced by Escherichia coli strains.

This chapter, which is an updated and expanded version of one that was published in this book series in 2003 [5], discusses the role of microorganisms as sources of pure natural products or derivatives of natural products, and as leads for novel synthetic compounds. The impact of the increase in available genomic information and manipulation and the impact of bioinformatics are discussed in one section, followed by a section on the potential of microbes as producers of biopharmaceuticals, such as proteins, enzymes, and vaccines.

2.2 Production of Natural Products by Microorganisms

Natural products include thousands of compounds, and a high proportion of these compounds are produced by microorganisms. Among these microorganisms, actinomycetes are especially well known for their high production capacity. Many compounds produced by microorganisms are active antibiotically and are the major source for anti-infective drugs. At present about 7000 antibiotics are known and about 100 of these are produced commercially by microbial fermentation processes. A list of commercially important secondary metabolites originating from actinomycetes are given in [5].

2.2.1 Production of Libraries of Natural Products

Genetic engineering of microbial strains either by introducing deletions into one or several genes, by heterologous expression of genes, and by combining both have become important routes for drug design [5, 6]. These technologies have been used successfully to generate libraries of many natural compounds (Table 2.1, Figure 2.1).

Table 2.1 Natural products that have been used as lead structures for the generation of natural compound libraries.

Figure 2.1 Structure of selected compounds that have been used as lead structures for the generation of novel natural compounds.

A recent successful story is about lipopetides. The story started with the cloning and sequencing of clinically useful antibiotics such as daptomycin produced by Streptomyces roseosporus [31], and friulimicin produced by Actinoplanes friuliensis [32] and by cloning the calcium dependent antibiotic CDA [33]. Genetic engineering of the producer strains, heterologous expression of complete gene clusters, and expression of sets of genes either from one or from different producer strains resulted in nearly 50 novel lipopetides, some of them with fundamental changes in properties [11, 12].

Based on its know-how in generating product libraries, a biotech company, Biotica, recently formed successful partnerships with leading pharmaceutical companies indicating the importance of the new technologies for the pharmaceutical industry. One of the key projects was the development of rapamycin analogs through genetic engineering technologies.

2.2.2 Production of Natural Products by Cloning and Expression of Biosynthetic Gene Clusters

Biosynthetic genes are clustered in actinomycetes. As original producers often do not produce more than 1–5 mg/l of a given compound scientists started to express gene clusters in others strains to improve production rate. Examples are listed in Table 2.2. It is worth mentioning that in a few cases the production rate could be increased significantly when the cluster was heterologously expressed. The highest amount obtained was 500 mg nikkomycin produced by S. lividans TK23 containing the nikkomycin gene cluster.

Table 2.2 Natural products produced after heterologous expression of biosynthetic gene clusters.

In the last few years, microbial natural product research has been revolutionized by genomic technologies. The complete sequence of microbial genomes revealed a remarkable number of gene clusters encoding enzymes involved in the production of undetected and unknown secondary metabolites [55–58]. This finding led to the rapid development of novel technologies to wake up these silent clusters in the original producer strain, and also to novel genomic-driven approaches for cloning and expressing these gene clusters in other strains.

2.2.3 Culture Manipulation to Wake Up Silent Gene Clusters

In 2002 the OSMAC theory (one strain many compounds) was published by Zeeck and coworkers [59]. These workers described microbial strains that are metabolically responsive to a broad range of different media, cultivation vessels, solid versus liquid fermentation techniques, and enzyme inhibitors. Following this approach, several new compounds have been detected in bacteria and other organisms. Recently the microbial coculture technique was rediscovered by several scientists. Based on the premise that bacteria live in complex communities, workers observed that cocultivating microorganisms elicitated natural product biosynthesis. One example is about bacterial–fungal interaction. When Aspergillus nidulans was cocultivated with actinomycetes the fungal secondary metabolism was specifically activated [60]. One major factor limiting an effective use of culture manipulation is that we do not know why natural products are produced under different conditions.

2.2.4 Genomic Driven Approaches to Wake Up Silent Gene Clusters

It was Ecopia (now Thallion Pharmaceuticals, Inc.), a small biotech company, which was the first company to develop a screening platform for searching secondary-metabolite encoding biosynthetic gene clusters in bacteria genomes. Ecopia showed that it is possible to predict structures of natural products based on genome information [61, 62].

In recent years scientists have developed heterologous expression techniques allowing the production of novel compounds in suitable host strains (see also Table 2.2). Although these techniques are still not applicable to high-throughput investigations, many novel compounds have been produced in this way. Today scientists are working on overcoming several significant problems in connection with this approach.

2.2.5 E. coli, an Interesting Host Also for Natural Product Synthesis

Although actinomycetes are used in industry as convenient producer strains, the advantages of using E. coli motivated scientists to develop this bacterium as a producer of natural products.

2.2.5.1 Production of Polyketides in E. coli

Engineering of E. coli to produce polyketides mainly focused on 6-methyl-salicyclic acid (6-MSA), 6-deoxyerythronolide B (6dEB), erythromycin, and epitholone [63–65] As E. coli does not naturally produce polyketides scientists faced major problems that had to be solved.

1) Expression of polyketide synthases was only possible at low temperature (22–30 °C).

2) The biosynthesis of polyketides requires phosphopantetheinylation, which can be achieved by introducing a P-pant transferase into E. coli.

3) Adequate levels of precursors (e.g., malonyl-CoA, propionly-CoA, NADPH) are needed for high production. Depending on the precursor genes, encoding enzymes responsible for the biosynthesis of these precursors have to be introduced into E. coli.

The final amount of 6-MSA was 75 mg/l, of 6-dEB 1 g/l, of 0.4 mg/l erythromycin, and of epothilone 0.001 mg/l. The amount of erythromycin produced by E. coli could then be drastically increased by synthesizing artificial polyketide synthase genes with codons appropriate to the host’s t-RNA and GC biases.

2.2.5.2 Metabolic Engineering of E. coli for Isoprenoid Biosynthesis

Many isoprenoids are found in nature, but mainly only in low abundance. Therefore, in isoprenoid research the development of renewable production processes has been much more in the focus of scientists than the generation of libraries. Beside Saccharomyces cerevisiae, E. coli has become the most utilized microorganism for the synthesis of isoprenoids [66].

Isoprenoid metabolic engineering in E. coli started with carotenoids, such as lycopene and ß-carotene. Genes of the carotenoid pathway isolated from Erwinia uredovora were expressed in E. coli resulting in the production of zeaxanthin, ß-carotene, and lycopene [67]. Recent research has focused on the diterpene taxadiene, an intermediate of paclitaxel biosynthesis [68], on monoterpenes such as (−)-carvone and (−)-limonene [69], and on amorpha-4,11-diene, an intermediate of artemisinic acid [70].

The most challenging issue for isoprenoid biosynthesis in bacteria is the limitation in the supply of the universal precursors IPP (isopentenyl diphosphate) and DMAPP (dimethylallyl diphosphate). Early attempts focused on overexpression of single or several genes of the non-mevalonate pathway, either by providing extra copies of genes [71] or by replacing native promoters [72]. Very recently, the group of Prof. Dr. J.D. Keasling engineered the mevalonate pathway from Saccharomyces cerevisiae in E. coli, resulting in a high producer of amorphadien, a precursor of the anti-malaria drug artemisinin [73, 74]. Production was even higher when yeast genes for HMG-CoA synthase and HMG-CoA reductase (the second and third enzymes in the pathway) were replaced with equivalent genes from Staphylococcus aureus, more than doubling production. Amorpha-4,11-diene titers were further increased by optimizing the culture conditions resulting in the production of 27.4 g/l. The conversion of amorpha-4,11-diene into artemisinin is possible by chemical synthesis with a yield of about 30%. Thus the conversion of amorpha-4,11-diene via genes and enzymes is economically desirable. A plant P450 enzyme (8-cadinene hydroxylase) was found to catalyze this reaction, but attempts to express the corresponding gene in E. coli were very challenging. So far the artemisinic acid production in E. coli is around 105 mg/l. Useful approaches for increasing production rates, which were also applied, are optimizing codon usage, enhancing production of rate-limiting enzymes, and eliminating the accumulation of toxic intermediates or by-products [75]. In other approaches various methods (transposon mutagenesis, computational based genome-wide stoichiometric flux balance analysis (see also below)) were used to identify non-biosynthetic genes that influence the flux of carbons from the central metabolism of the bacteria cell towards the non-mevalonate pathway. E. coli strains were generated containing deleted genes and genes that were overexpressed resulting in higher production rates of lycopene [76–78].

Key enzymes in the biosynthesis of isoprenoids are synthases (e.g., taxadiene synthase, amorphadiene synthase, limonene synthase) and P450 dependent oxygenases. Engineering of these enzymes is an effective way to create novel molecules. Recently random and site-specific mutagenesis was performed to convert a (+)-δ-cadinene synthase into a germacrene D-4-ol synthase. In addition, based on the crystal structure of a P450 oxygenase involved in the conversion of (+)-campher into 5-exo-hydroxycamphor, a mutant was designed with altered substrate specificity. This mutant is able to convert (+)-α-pinene into (+)-cis-verbenol [79].

2.2.6 Global-Scale Strategies for Strains Improvement

The major objectives of pharmaceutical biotechnology include the increase in products yield, the introduction of pathways leading to new products, the deletion or reduction of byproduct formation, and the enhancement of strain tolerance. The microbial strains can be engineered for the desired phenotypes based on the global information obtained from the results of genomic (studies on whole sets of genes), transcriptomic (studies on mRNA expression levels), proteomic (studies on interactions and functional dynamics of whole sets of proteins), metabolomic (studies on the concentration of metabolites), and fluxomic (studies on the complete flux of metabolites in the metabolic network). These data are then used for computational modeling and simulation of metabolism followed by the wet lab experiments [80].

2.2.6.1 System Biology, System Biotechnology, and Omic Approaches

Scientists working in the area of system biology are trying to understand the function of a cell rather than the function of single components of a cell. The nature of the links that join different components are used in mathematical models to perform quantitative analysis of biological system. Data are then used for cell design, which is part of system biotechnology. The successful design requires collection, analysis, and integration of genome data and the prediction of the behavior of the biological system in response to exogenous triggers. The aim is to predict which genetic changes will lead to a desired phenotype. Recent advances in whole-genome sequencing techniques and bio-informatic analyzes have propelled reconstruction of genome-scale biochemical reaction networks in microorganisms. Examples are E. coli K-12 MG1655 [81], B. subtilis [82], Methanosarcina barkeri [83], Saccharomyces cerevisiae [84], and S. coelicolor [85]. The metabolism reconstruction is based on annotated genes, physiological and biochemical information, and linear programming. The reaction sets are composed in a stoichiometric matrix with two dimensions representing the number of metabolites and the number of reactions. This is the starting point for various mathematical analyzes. Additional information about the enzyme complex formation, protein localization, and regulation can also be associated with the matrix. To predict the effects of gene inactivation on whole cell metabolic flux distribution, the flux value of the corresponding reaction should be constrained to zero during the simulation [86].

This simulation method has been proven to be a powerful tool for predicting targets for metabolic engineering. For instance, Nielsen and coworkers have successfully engineered an S. coelicolor strain to produce actinorhodin and prodigiosin at a five-times higher yield than the wild type, by combining an in silico gene knockout simulation and rational metabolic engineering [85]. A genome-scale metabolic model of S. coelicolor predicted that the deletion of the phosphofructokinase gene will lead to an increased flux through the pentose phosphate pathway, and thus accumulation of NADPH. As NADPH (nicotinamide adenine dinucleotide phosphate) is a cofactor of polyketidesynthase (a central enzyme involved in actinorhodin biosynthesis), its elevated level led to the overproduction of actinorhodin. This theoretical calculation was supported experimentally.

There are many other successful examples of in silico simulation and its use to predict gene targets to be knocked or overexpressed. Using these approaches, the mutant strains of E. coli overproducing L-valin and L-threonine have been obtained [87, 88]. Gene targets for metabolic engineering can also be predicted by comparison of the transcriptomes of the wild types and classically improved overproducers. For instance, novel genes in the C. glutamicum genome, NCgl0855 (putatively encoding a methyltransferase) and the amtA-ocd-soxA operon, which could improve the production of lysine were identified using a DNA microarray. After their overexpression, total lysine production was increased by about 40% [89]. Comparative transcriptomes profiling were performed for three different actinomycetes strain, Saccarpolyspora erythrea, S. fradiae, and S. avermitilis [90, 91]. Wild type strains were compared with overproducing strains. Genes that were overproduced in the superior strains were detected. These results will pave the way for manipulating strains in the future for obtaining overproducers.

2.2.6.2 Synthetic Biology Tools

Synthetic biology is the discipline that aims at reconstructing, designing, and building novel biological systems in the same way as engineers design electronic and mechanic systems. This discipline provides the ability to manipulate (separate and combine) modular biological components such as promoters, genes, RNA translational control devices, and the whole metabolic pathways for the production of valuable pharmaceuticals. Synthetic biology has still not achieved satisfactory levels of precision, robustness, and reliability to become a form of engineering. Factors preventing this are incompatibility of biological parts and modules, functional overlaps between them, and the dependence of the functionality of the parts on the context. Deeper characterization of the biological components and their standardization will help to solve these problems. Despite all these difficulties, synthetic biology approaches have already been achieved in many important pharmaceuticals [92]. One notable example of a synthetic biology application is the production of artemisinin, which was discussed in Section 2.2.5.2.

Another example is the construction of synthetic operons leading to the production of activated deoxysugars, which are an important part of many commercial antibiotics and antitumor drugs. Using standard parts (promoters) and devices (genes from different biosynthetic pathways) Salas and coworkers constructed a number of modules (synthetic operons) that are responsible for the deoxysugar production. These modules have a high level of compatibility as their combination with other modules (responsible for polyketide production) and devices (different glycosyltransferases) have yielded a library of antibiotics and antitumor compounds [93].

Synthetic biology uses well-characterized, modular parts that can be put together to create new functionality in a directed, predictable manner. Therefore, one of the most important goals is to create the libraries of such modular and well characterized parts. In the following sections two examples are presented describing the construction of modular libraries.

2.2.6.2.1 Synthetic Promoter Library

Gene expression in bacteria is mainly controlled at the transcriptional level, and therefore the promoter is the most tunable element. A wide range of promoter strengths is necessary to express a certain gene at the desired level. Several strategies have been developed to construct synthetic promoter libraries for fine-tuning gene expression. Usually, these methods rely on random modification of existing promoters in their spacer regions between the consensus sequences, as they affect a promoter’s strength most strongly. The promoter activity is tested using different reporter genes such as gfp, beta-galactosidase, and others [94]. The synthetic promoter library should represent as many variations in promoter strength as possible. For example, if only a low-range of expression is required because of gene toxicity, then a set of promoters driving low levels of expression will be used. If one needs to maximize a valuable protein, then expression from the strong promoters is necessary. For instance, a synthetic promoter library was used to assess the impact of deoxy-xylulose-P synthase levels on lycopene production, and the optimal expression level of this gene was identified for maximal desired phenotype [94].

2.2.6.2.2 Synthetic Tunable Intergenic Regions Library

In many cases, more than one gene needs to be expressed to produce different pharmaceuticals (e.g., antibiotics). In order to produce these molecules the genes must be expressed at appropriately balanced levels, which help to avoid the accumulation of toxic intermediates resulting in growth inhibition or suboptimal yields. Therefore, it is important to combine multiple genes into a synthetic operon with a single promoter, and fine-tune expression of each gene at the translational level. The tunable control elements are mRNA secondary structure, RNase cleavage sites, and ribosome binding sites at intergenic regions. Libraries of tunable intergenic regions (TIGRs) were generated and screened to tune expression of several genes in an operon [95]. TIGRs can vary the relative expression of two reporter genes over a 100-fold range. The sevenfold increase in the mevalonate production was achieved using different TIGRs for the coordination of three genes expression in an operon [95]. Thus, using TIGRs is expected to be valuable for engineering of complex biosynthetic pathways in order to improve the phenotype.

2.2.6.3 Whole Genome Engineering Approaches

Rational methods for strain improvement, mainly based on a single-gene inactivation or on heterologous expression, have been developed in recent decades. However, owing to the complex nature of the biosynthetic machinery used for the production of many pharmaceutically relevant molecules, and due to the fact that not all regulatory genes involved in the biosynthesis of a compound are known, these methods have been successfully applied for the development of just a few industrial microorganisms. Many recent efforts focused on the development of global engineering approaches such as artificial transcription factor engineering [96], ribosome engineering [97], global transcription machinery engineering [98], and genome shuffling [99] for obtaining strains with desirable characteristics.

2.2.6.3.1 Protoplast Fusion and Genome Shuffling

The whole genome engineering approach is based on protoplasts fusion (Figure 2.2) and genome shuffling enable construction of many bacteria in a more global fashion. Usually protoplast fusion-based techniques are used for the improvement of phenotypes of microorganisms. Protoplasts fusion has been established since the late 1970s. There are a few reports demonstrating that protoplasts fusion allows recombination events to take place throughout the genome and at a much longer range relative to other methods. A high frequency of recombination is achieved by fusing complete genomes. Genome shuffling is a technology of recursive protoplast fusion between multi-parent strains to obtain a desired phenotype. Through this method, overproducers of tylosin, teicoplanin, natamycin, epothilone, riboflavin, and pristinamycin produced by S. fradiae, Actinoplanes teichomyceticus, S. gilvosporeus, Sorangium cellulosum, Bacillus subtilis, and S. pristinaespiralis, respectively, have been obtained [100–103]. The success of the genome shuffling strategy strongly depends on the availability of a high-throughput screening method for a desired trait. This method became popular after a publication by Stemmer and coworkers in 2002, where they described the development of a tylosin overproducer [100]. It is worth mentioning the publication of Fedorenko and coworkers from 1993. In this paper, the process of recursive protoplast fusion of multi-parental populations of Saccharopolyspora erythrea is described. The selected mutant strains, overproducing erythromycin, were later used for the industrial production of erythromycin in the Ukraine [104]. There are a few reports on the interspecies protoplast fusion using streptomyces strains. When S. fradiae 261-27E (mycaminose producer) and Streptomyces sp. AM 4900 N3-4, (pikronolide producer) were fused, an unstable prototrophic fusant produced a macrolide antibiotic, which was not found in parent strains [105]. A novel antibiotic was also generated by interspecies protoplast fusion treatment using S. griseus and S. tenjimariensis [106]. Taking into consideration the efficiency and time for the strain development, genome shuffling will play an important role in engineering of microorganisms in the future.

Figure 2.2 Electron microphotographs demonstrating the protoplast fusion of Streptomyces kanamyceticus.

(The photographs were recorded by O.R. Kulachkovskyy from the Lviv National University, Ukraine).

2.2.6.3.2 Engineering of Translational and Transcriptional Machineries

Global changes in bacterial gene expression have been observed after the alteration of the ribosomal proteins and RNA polymerase subunits. These changes led to the overproduction of antibiotics and enzymes by actinomycetes and bacilli and enhanced the tolerance of Pseudomonas to different chemicals. The convenient and simple method of modulating ribosomes and RNA polymerases is to treat strains with low concentrations of different antibiotics (e.g., streptomycin, gentamycin, neomycin, tetracycline, paromomycin, thiostrepton, and rifampicin) [107]. This leads to resistant strains accumulating mutations, which are located in genes encoding ribosomal proteins and the RNA polymerase subunits. A dramatic activation of natural product synthesis has been observed in many actinomycetes containing a mutation in rpsL, encoding the ribosomal protein S12, which confers resistance to streptomycin [108]. The beneficial effect on antibiotic production of mutations in the rpsL gene have been described by the isolation of mutants in S. lividans, S. coelicolor, S. chattanoogensis, S. antibioticus, S. lavendulae, and S. albus producers of actinorhodin, undecylprodigiosin, fredericamycin, actinomycin, formycin, and salinomycin, respectively [109–112]. Mutations in rsmG, which encodes a 16S RNA methyltransferase, also increased actinorhodin and actinomycin production in S. coelicolor and S. parvulus, respectively [113]. Combining the rpsL and rsmG mutations resulted in a further increase in antibiotic production [114]. Introduction of streptomycin resistance mutations also increased antibiotic production in Bacillus and Pseudomonas.

The simultaneous introduction of several resistant mutations to streptomycin, gentamicin, and rifampin has a synergistic effect on antibiotic production, leading to improvements in actinorhodin, salinomycin, and thiazolylpeptide GE2270 production in S. coelicolor, S. albus, and Planobispora rosea strains, respectively [115]. Mutations lead to an increase of the hyperphosphorylated guanosine nucleotide (ppGpp) level in the cells of actinomycetes [116]. ppGpp can affect gene expression by altering the selectivity of the RNA polymerase [117]. Similarly, mutations in rpoB, which encodes a β-subunit of the RNA polymerase, conferred resistance to rifampicin and led to overproduction of the antibiotic [118].

Another example describes the biosynthesis of actinorhodin, undecyprodigiosin, and the calcium dependent antibiotic. All three pathways were activated after introducing mutations into rpoB. It was shown that the mutated RNA polymerase has a better affinity to the gene promoters of the gene clusters responsible for the formation of the three compounds [119]. The introduction of antibiotic resistant mutations also improves the production of extracellular enzymes in Bacillus subtilis [120]. More recently, it was shown that drug-resistant bacteria produce novel natural products in contrast to their wild type strains. Streptomyces species that did not produce antibacterials started to produce antibacterials after selection on streptomycin and/or rifampicin. Again mutations were detected in rpoB and/or rpsL genes [121]. In one case a novel class of antibiotics, named piperidamycins, has been discovered.

2.3 Prokaryotes as Producers of Recombinant Therapeutic Proteins

Biopharmaceuticals (recombinant proteins and antibodies) are becoming more and more important as they help to overcome sicknesses that are not treatable without them. Recombinant proteins intended for the drug market are required in large amounts. Thus, the aim is to express high levels of stable, soluble, and functional proteins. Once a protein can be produced in small amounts, scaling up the production rate is an important and difficult task. For instance, bacterial cultures of high cell density, as are frequently used in industry, show different characteristics to cultures on a small scale. Drawbacks are, among others, the limited availability of dissolved oxygen, insufficient mixing efficiency of the fermentors, carbon dioxide (CO2) levels that can decrease growth rates and stimulate cell toxic acetate formation, or the fairly expensive use of antibiotics for selection [122]. However, several technological developments and feeding schemes can now improve the fermentation process. In this section, different strategies for successful intracellular expression of recombinant proteins, in particular mammalian proteins, in prokaryotes will be reviewed.

2.3.1 Prokaryotic Expression Systems

Today there are various expression systems available. As hosts, bacteria, yeasts, moulds, mammalian cells, plants, insects, or even transgenic animals are used. Which system is chosen mainly depends on the protein structure, production speed, and yield [123]. Non-glycosylated proteins are usually produced in Escherichia coli (E. coli) or yeasts, N-glycosylated proteins in mammalian cells, for example, Chinese hamster ovary (CHO) cells, which mimic human glycosylation. However, even these glycoproteins are not exactly the human type, and possibly need to be modified [123].

Bacterial hosts, with E. coli being the most important, are still widely used for heterologous protein expression. This is due to several advantages of E. coli expression systems in particular, including ease of handling, inexpensive culture media, very well-known genetics, availability of improved genetic tools, fast growth of the cells, and high yields of recombinant protein achieving up to 20–30% of total cellular protein [123–125]. Still much effort is being given to improve and develop prokaryotic expression systems in order to overcome deficiencies. In the following, microbial expression systems, which are employed in pharmaceutical industry, will be described.

2.3.1.1 Host Strains

2.3.1.1.1 E. coli

E. coli systems are most commonly used for industrial and pharmaceutical protein production and large-scale production systems are established [126]. Besides the advantages mentioned above, there are also some limitations such as acetate formation in high-density cultures resulting in cell toxicity, lack of post-translational glycosylation ability, and accumulation of endotoxins, which must be removed during the purification process. Moreover, proteins are often built as inactive inclusion bodies and require refolding and proteins containing disulfide bonds are fairly difficult to express [123].

However, many advances have been achieved to improve the E. coli production system. Successful measures that have been taken are, for example, the use of different promoters to regulate expression, controlling the oxygen level, optimization of the specific growth rate, coexpression of chaperones, secretion of proteins into the periplasmic space or into the culture medium, or addition of a fusion partner [123]. Different mutant E. coli strains have been developed, which show specific features to meet the special demands of particular proteins (Table 2.3) [126, 127]. E. coli BL21 and its derivatives are the most widely used ones [126, 127]. The difference between the codon usage of different species or organisms can also cause problems, particularly for the overexpression of eukaryotic genes in E. coli. Insufficient tRNA pools can lead to translational stalling, formation of shortened polypeptide chains, translation frameshift, and amino acid misincorporation [126]. To enhance production of heterologous proteins that contain codons rarely used in E. coli, special strains were created which supply additional tRNAs under control of their native promoters (Table 2.4).

Table 2.3 Recombinant proteins that are used as human therapeutics and are approved by the EMA (status July 2010) (http://www.vfa.de/gentech; www.ema.europa.eu)

Table 2.4 E. coli strains frequently used as hosts for heterologous protein production [126, 128, 131].

Alternatively, the codon usage can be optimized for the respective host by replacing the natural genes that are to be expressed with synthetic genes. For some proteins, for example, insulin, formation of inclusion bodies is an advantage rather than a disadvantage because of an easier isolation and purification procedure. However, others loose their functionality when falling out as inclusion bodies due to misfolding. To support the folding process certain plasmids for coexpression are available on which genes of bacterial chaperones are encoded. Different origins of replication and selection markers prevent compatibility problems with other expression vectors. The best characterized chaperone systems in the cytoplasm of E. coli are the DnaK-DnaJ-GrpE and GroEL-GroES systems. It is important to mention that success is by no means guaranteed and highly dependent on the nature of the overexpressed recombinant protein [126, 128, 129].

It is often desirable that heterologous expressed proteins are secreted either into the periplasm or into the culture medium. Firstly, the isolation and purification process will require less effort. Secondly, during the translocation process usually a so-called leader peptide will be cleaved at the N-terminus resulting in proteins without methionine at their N-terminus. This may be essential to maintaining the bioactivity of eukaryotic proteins. Another advantage is that correct formation of disulfide bonds can be facilitated, because in contrast to the cytoplasm in the periplasmic space a more oxidative environment predominates [126, 130–132]. Alternatively, proteins with disulfide bonds can be overexpressed in the cytosol of thioreductase-deficient (trxB) and glutathione reductase-deficient (gor) mutant strains (Table 2.4) [126, 131, 132]. The most prominent protein secretion systems used in E. coli strains are the Sec system and the twin-arginine translocation (Tat) system [132]. To secrete recombinant proteins into the periplasmic space, fusion with a signal sequence at the N-terminus is necessary. Examples of commonly used signal peptides are MalE, OmpA, OmpT, PhoA, PhoE, and Tat signal peptides [126, 127, 130, 132]. Most problems with secretion of heterologous produced proteins are caused by incomplete translocation across the inner membrane, proteolytic degradation, and a limited number of available gates. For optimized results expression, rate and transport capacity have to be balanced carefully [127, 131].

2.3.1.1.2 Bacillus

Other useful host systems are those of the Gram-positive bacilli. Bacillus species are often preferred for expression of enzymes such as proteases (for detergents) and amylases (for starch and baking) [123]. They are currently not used for the production of pharmaceutical therapeutics approved by the European Medicines Agency (EMA) (Table 2.3). Some advantages of Bacillus as the expression host are its high capability for secretion of the desired proteins into the fermentation medium, the absence of a lipopolysaccharide containing outer membrane and its metabolic robustness [126]. In addition, bacilli are also genetically well characterized, and highly developed transformation and gene manipulation technologies are available [127]. The species most frequently used are Bacillus (B.) megaterium, B. subtilis, B. licheniformis, and B. brevis [123].

2.3.1.1.3 Other Bacteria

Besides E. coli and Bacillus some other bacteria have been tried for overexpression of proteins. An improved Gram-negative host has been developed using Ralstonia eutropha. This system displays a reduced inclination for inclusion body formation and seems to be amenable to high-cell-density fermentations due to its more efficient carbohydrate metabolism compared with E. coli systems [123, 133]. Good yields were reported for the production of a mammalian protein by Staphylococcus carnosus [126, 133]. Streptomycetes are also considered to be suitable for the development of efficient expression and secretion systems. However, the results reported so far are still not satisfactory and below cost-efficient ranges [133]. Further research is necessary to make Streptomyces systems competitive.

As prokaryotes show only a limited ability for post-translational modification their employability remains restricted to the production of naturally non-glycosylated proteins, for example, insulin or human somatotropin (STH), or to natively glycosylated proteins that are pharmacologically also active without glycosylation, such as interleukins, interferons or human tissue plasminogen activator (htPA). For the production of proteins that are only active when they are prost-translational modified, eukaryotic expression systems are the best choice.

2.3.1.2 Expression Vectors

As E. coli is the most frequently used host for the production of recombinant proteins and the only prokaryotic system employed for the production of drugs approved by the EMA, this section focuses on vector systems that are suitable for transformation of E. coli strains.

A variety of expression vectors are commercially available. Successful production of foreign proteins in E. coli depends not only on the host strain itself and on fermentation conditions, but also to a great extent on the type of expression vector. Generally, a typical expression vector contains several genetic elements that affect both transcriptional and translational steps of protein biosynthesis. Essential elements are a promoter, ribosome binding site (RBS), start codon, multiple cloning site (MCS), transcription terminator, origin of replication, selection marker, and, additionally, it may contain a fusion tag sequence.

2.3.1.2.1 Replication Rate

The origin of replication located on a plasmid determines the plasmid copy number in the E. coli cell. The plasmid copy number, on the other hand, determines the gene dosage accessible for expression. In many cases it was found that the higher the plasmid copy number the higher the yield of protein [122, 132, 134, 135]. However, the relation between plasmid copy number and amount of gene product is not proportional at all. In some cases it might be more successful to use medium or low copy number plasmids. Lower gene dosages lead to lower translation rates, so that proteins have an increased chance of correct folding [125]. For protein secretion slower and sustained protein synthesis is also preferred as mentioned earlier in this chapter, and high copy number may not be necessary or even desirable [136]. Depending on the target protein, low copy plasmids may have some more advantages, such as tight control of gene expression, the ability to replicate large pieces of DNA precisely, and a low metabolic burden on the host strain [135]. However, it should be noted that the final rate of protein production is the result of an interplay of many factors. It additionally depends on promoter strength, mRNA stability, and efficiency of translation initiation, which will be discussed in the following sections [136].

2.3.1.2.2 Promoter Choice

A variety of promoters are available by now. They differ in strength (strong/weak) and inducibility (inducable/constitutive). Several criteria have an impact on the selection of an appropriate promoter. If inclusion body formation is intended, a strong promoter may be helpful, and if the protein of interest is toxic to cell growth a tight regulation of the promoter is essential. Moreover, inducibility of the promoter to varying degrees should be possible in a simple and cost-effective manner [122, 137]. Constitutive promoters will be applicable only in a few cases. Frequently used promoters are summarized in Table 2.5. It is common to express the gene of interest from an inducible promoter controlled either by a repressor or by an activator [137]. Thereby negatively regulated expression systems are predominating. The most widely applied promoter system for research purposes is the T7 RNA polymerase system. However, the use of IPTG (isopropyl β-D-1-thiogalactopyranoside) for induction in large-scale production of recombinant therapeutic proteins is undesirable due to its toxicity and high cost.

Table 2.5 Examples of E. coli promoter systems used for heterologous protein production.

The most widely used promoters for large-scale production use thermal (e.g., thermosensitive variants of the LacI repressor protein) or nutritional (e.g., trp) induction [122]. Often promoters are modified in order to optimize expression for single target genes. Prokaryotic promoters contain the so-called -35 and -10 sequences consisting of six nucleotides each and a 15–19 bp spacer in between. Among