Introduction

Protein synthesis is an energy demanding process: Calculating the energy required for (1) the synthesis of a codon, which is the nucleic acid information unit for an amino acid, (2) the charging reaction of a tRNA by its synthetase with the correct (cognate) amino acid, and (3) the subsequent incorporation of this amino acid into the nascent peptide chain, a total of 10 energy-rich bonds need to be sacrificed. In this context an energy rich bond means an acidic-anhydride bond of adjacent phosphate residues of ATP or GTP, each with an energy content of about ΔG 0′ = −6 kcal/mol. The enormous energy requirement explains why a cell has developed intricate systems for controlling energy consumption. This is precisely the Achilles’ heel of modern in vitro translation systems, where protein synthesis is severely restricted by a huge squandering of energy: Usually no more than 5% of the energy is used for actual synthesis of proteins, with the rest wasted by uncontrolled and useless energy drains. In striking contrast, a bacterial cell can channel up to 70% of its energy consumption into protein synthesis. In the following sections we will outline the major points one has to follow in order to avoid energy losses and to direct most of the energy – almost like the living cell – into useful work, in this case, protein synthesis.

At least three major reasons for energy losses in vitro systems can be identified: (1) the metabolic energy drain, (2) idle ribosomes and an excess of the elongation factor EF-G, and (3) a metabolization of amino acids. In this order we consider these points.

Preventing Metabolic Energy Drain

The standard way to regenerate energy (ATP, GTP) used in protein synthesis is the addition of millimolar concentrations of phosphoenolpyruvate (PEP) and the enzyme pyruvate kinase that transfers the phosphate group from PEP to AMP and ADP yielding ADP and ATP, respectively (see Fig. 1a). Likewise, GMP and GDP are also substrates for this enzyme. PEP contains the energy richest bond of all biological samples, the phosphoester bond has an energy content of ΔG 0′ = − 12 kcal/mol. This system allows only the synthesis of limited amounts of proteins due to the inhibitory effect resulting from the generation of orthophosphate. Orthophosphate byproducts, not only reduce the pH, but also bind Mg2+ ions, thus reducing the free Mg2+ concentration and severely effecting the structure of all ribonucleoproteins in the system, the most prominent one being the ribosome. The two positive valences of Mg2+ shield the negatively charged phosphate groups of the RNA and maintaining the high density of the ribosome, thus reducing the free concentration of Mg2+ sets the repelling phosphate groups free and expands the ribosome, impairing its efficiency.

Fig. 1
figure 1

Various ATP regeneration systems for in vitro protein expression. a Conventional scheme for energy regeneration using phosphoenol puryvate (PEP) and pyruvate kinase. The generated orthophosphate (Pi) inhibits protein synthesis. b Improved energy regeneration using pyruvate, pyruvate oxidase and endogenous acetate kinase (according to Kim and Swartz 2000). c Near in vivo system based on phosphotransacetylase (Pta) to produce acetylphosphate. The boxes in b and c show a common set of reactions for regeneration of ATP. The PEP synthase (Pps) is blocked by the presence of oxalate. Ack acetate kinase, Ldh lactate dehydrogenase, Pdh pyruvate dehydrogenase, Pyk pyruvate kinase. According to Jewett and Swartz (2004), modified

Full size image

One way to get rid of this unwanted effect is to recycle the liberated phosphate by addition of pyruvate oxidase, together with some prosthetic groups, such as TPP and FAD (thiamin pyrophosphate and flavin dinucleotide, respectively). This system generates acetylphosphate, a standard component of the regeneration system of a cell, by using molecular oxygen and H2O (Fig. 1b), and this improvement in conditions leads to a definite prolongation in protein synthesis in vivo (Kim and Swartz 2000).

This system, however, is still far from an ideal situation. The need of molecular oxygen is the drawback, since the uptake of oxygen is limited and becomes worse when larger reaction volumes are used because of the decreasing relative surface area. Jewett and Swartz (2004) recently reported a convincing solution of this problem: by combining the classical PEP/pyruvate kinase system with the synthesis of acetylphosphate via acetyl-CoA, they efficiently recycle orthophosphate (Pi) during the synthesis of acetylphosphate (Fig. 1c). This solution represents a breakthrough: for the first time several mg of proteins can be synthesized per one ml, making in vitro systems a serious alternative for synthesis of proteins that can be used for structural and functional analyses. Table 1 compiles the precise conditions that allow the synthesis of mg of a protein in a coupled transcription–translation system, where the gene of interest is introduced on a plasmid behind a T7 promoter (for details see Iskakova et al. 2006).

Table 1 Final concentrations of components in an in vitro system for protein synthesis
Full size table

Prevent Idle Ribosomes in the Presence of an Excess of EF-G

GTP-binding (G) proteins are an important class of regulatory proteins. They bind GDP or GTP, and undergo a conformational change depending on the presence of which nucleotide is bound. In the GTP conformer (the “on”-state) they bind to their target and promote a distinct reaction of the target, with the energy being paid by the binding energy generated during the binding step of the G protein. The target then signals that the reaction was successfully accomplished and triggers the hydrolysis of “GTP → GDP + Pi” in the corresponding enzymatic center of the G protein. Upon release of the inorganic phosphate, the G protein switches into the GDP conformer (“off”-state) and looses it’s affinity for the target and thus dissociates from it. Four G proteins participate in translation, two of which are components of the central activity of the ribosome, the elongation cycle, where in a cycle of reactions the nascent chain is prolonged by one amino acid (aa). One factor is EF-Tu, which brings the aa in the form of aa-tRNA to the decoding center of the ribosome, the other is EF-G, which translocates the tRNAs on the ribosome by one codon length (reviewed by Wilson and Nierhaus 2003).

Usually a G protein recognizes a specific configuration in the target before promoting or even triggering a specific reaction. This is also true for EF-G; it recognizes specifically the ribosome state before translocation and dissociates from the ribosome following GTP cleavage and Pi release. However, EF-G is the only G protein involved in translation that also interacts effectively with empty or idle ribosomes, probably because empty ribosomes seem to have “conformational memory” of the two main states before and after translocation (Mesters et al. 1994). Thus empty ribosomes can trigger EF-G dependent GTP hydrolysis with a high turnover (Fig. 2), which is not seen with EF-Tu. It follows that in a minimal system with poor energy resources both ribosomes and EF-G must be present in limiting amounts. The latter is established, if the molar ratio of EF-G: ribosomes is about 0.2 to 0.3, i.e. 2–3 EF-G molecules per 10 ribosomes.

Fig. 2
figure 2

Kinetics of EF-G dependent hydrolysis of GTP on 70S ribosome. Taken from Qin et al. 2006, where the experimental details can also be found

Full size image

Considerations on the Metabolization of Particular Amino Acids (Arg, Cys, Ser and Trp)

When we follow synthesis of the Green Fluorescent Protein (GFP) kinetically in an efficient coupled transcription–translation system, synthesis occurs almost linearly for about seven hours, but then starts to plateau (Fig. 3a). One reason for this plateau could be consumption of the energy-rich compounds (e.g. ATP). However, when we simultaneously follow the synthesis of mRNA in the same reaction, we observe that the total mRNA (synthesis and decay) displays a minimum at 7 h, but as GFP synthesis ceases, the concentration of mRNA starts to increase again. This clearly indicates that there is still enough energy in the system. Another possibility therefore could be the metabolic degradation of some amino acids, which would block protein synthesis completely. Indeed, some amino acids have been identified that are significantly metabolized in vitro systems, in particular arginine, cysteine, serine and tryptophan (Michel-Reydellet et al. 2004); for review see (Swartz 2006). Consistent with this idea, refreshing the amino acid pool after a 7-h incubation period significantly improved the overall yield (Fig. 3b).

Fig. 3
figure 3

Synthesis of GFP in a near in vitro protein system. a A comparison of the synthesis of GFP and that of GFP-mRNA. Continuous line, total synthesis of GFP determined by SDS PAGE; dashed line, amount of GFP-mRNA as determined by Northern blotting. b Total synthesis of GFP; (−)aa, 2 mM amino acids present from the beginning; (+)aa, a second addition of amino acids after 7 h incubation leading to a total amino acid concentration of 4 mM. Taken from Iskakova et al. (2006), see also for experimental details

Full size image

Conclusions

Together with our article in chapter 6.3, we have tried to estimate the minimal set of genes needed for a minimal cell. A brave estimate of the minimal components for the translational apparatus today comprises no more than 200 genes, of which more than 120 are associated with the translational apparatus, encoding about 40 genes for ribosomal proteins, two rRNAs (omitting the 5S rRNA), 21 tRNAs, 20 synthetases, six factors and at least 20 tRNA modifying enzymes. In addition, a minimum of 30 genes are needed for both the household energy as outlined in this article and the synthesis of at least some of the amino acids (note that some of the amino acids were formed in the Stanley Miller experiments mimicking the atmosphere and the physical environment more than 3 Bya (Miller 1953) and therefore need not be synthesized by the cell).

Even if we consider a “limping” life form, we are ending up with a total of at least 150 genes. This is in a reasonable agreement with the most restricted life form known today, namely the bacteria Carsonella rudii with 189 genes (Nakabachi et al. 2006). This bacteria is an obligatory endosymbiont living in the “jumping plant lice” (psyllids), namely in specialized cells (bacteriocytes) made by the insects. The bacteria belong to the γ-proteobacteria as Escherichia coli, are transmitted by vertical transmission through the host generations (no exogenous infections!) and contain a full translation system as well as prepare some amino acids for the host. We think therefore that it is not possible at the moment to construct a primitive cell “getting-by” with a number of genes significantly below the 150 genes mentioned above.