Bruce Alberts wrote in the standard biology textbook ‘The Cell’ in 1998 that: “Nearly every major process in a cell is carried out by assemblies of 10 or more interlocking protein molecules.” This is documentary evidence that the importance of protein complexes has been known for some time. Nevertheless there are relatively few descriptions of the structure of such protein machines. The reason is that the production of large amounts of complexes of this kind is extremely laborious. However that could soon change, because ETH Zurich biologists led by Imre Berger and Timothy Richmond at the Institute for Molecular Biology and Biophysics have succeeded (1) in developing a method for production of entire protein complexes in larger quantities relatively quickly. A paper about it was published in the scientific journal “Structure” in March 2007 (2).

Up to 6000 litres of cytoplasm for a single structure

The research that led to this breakthrough in methodology, however, goes back a long time. About five years ago Imre Berger already saw a threat to his research into the structure of the human protein complex TFIID. The latter is involved in translating DNA into RNA. At that time the researcher said this of the situation: “Solving the structure of Polymerase II, a relatively common complex in the cell, needs 6000 litres of cytoplasm. That means years of harvesting work.” Since this effort was not feasible for him, Berger had the idea of preparing his complex by a recombinant technique, i.e. producing the proteins by genetic engineering in a host organism. Although they were working in a structure laboratory, his manager Timothy Richmond gave him the green light to proceed with this methodology research.

Berger made a search and hit on the baculovirus as a possible engine to produce the proteins of higher, so-called eukaryotic organisms. This virus can infect insect cells, where it expresses its genes as proteins. At first Berger thought it would probably suffice to allow the insect cells to be infected by several viruses each with one gene, each of which coded for one protein within an entire complex. However, although it appeared plausible on paper, this super-infection did not work properly. The researcher explains that “For TFIID we would have had 14 sub-units, which means 14 genes and thus 14 viruses. The 14 viruses would have had to be titrated out in a very precise ratio relative to one another for the protein complex to form. That is scarcely achievable in practice.” He says the individual production of proteins belonging to complexes often also fails to work because they are frequently insoluble in isolation.

In the newly developed MultiBac technology, host cells live longer after the virus infection and can produce more heterologous proteins in better quality. The heterologous proteins produced in insect cells infected with the MultiBac virus can be seen on the left, coloured red. For comparison, infection by a commercial virus with which the cells die sooner is on the right. (Photo: I. Berger) large

In this unsatisfactory situation Imre Berger talked to the baculovirus specialist Polly Roy from London. She advised him simply to incorporate all the genes for the proteins of a complex into a tailor-made baculovirus. Lo and behold, it worked. In 2004 it was already possible to produce several proteins of a complex in one virus. Admittedly, according to Berger, this version was still not very user-friendly. “It needed cut and paste enzymes, known in the jargon as restriction enzymes and ligases, which did not always work 100% reliably.” However the ETH Zurich researchers had not yet fully developed the method either. The virus genome contained a protease that split the resulting proteins non-specifically, thus making them unusable. The scientists removed the corresponding gene, eliminating one other gene at the same time. This was a stroke of luck because as a result the host cells lived longer and were able to produce larger amounts of proteins.

Worldwide interest

The international research community did not fail to notice the ETH Zurich researchers’ success. After the publication in Nature Biotechnology (2004), invitations to conferences and presentations arrived on a weekly basis (4). Numerous research groups wanted to order the tailor-made virus and the other Zurich System reagents – the addressees were from Harvard to Novosibirsk, including the crème de la crème in structural biology. Many laboratories with a gene therapy orientation also showed interest. That was because baculoviruses can also infect human cells, although they do not destroy them. The problem was obviously virulent, so “Nature Methods” invited the ETH Zurich scientists to publish a detailed protocol of their method in this scientific journal (5). It was published in 2006 and obviously contained the latest version, which was no longer based on restriction enzymes and ligases but rather used cleverly arranged recombination processes instead. This meant that in principle the method was amenable to automation.

This development is also imminent. Imre Berger and his colleagues published the article mentioned above in “Structure” in March 2007. The scientist says “For this we ‘streamlined’ our method so much further that in principle it can be implemented in a robot script.” This implementation is to take place in collaboration with the Paul Scherrer Institute. As a structural biologist, Berger hopes he will then obtain even more and bigger complexes. Thanks to his method, he has already been able to prepare part of the protein machine of his TFIID with 10 sub-units and a size of 700 kilodaltons. Looking to the future, Berger thinks: “We have still not reached the system’s limit. If good separation methods are also available, we may possibly even be able to prepare several complexes at the same time with one virus.” Obviously the structural biologist still seems to enjoy working on methodology.