Posted - 03/11/2005 : 08:07:38
Protein engineering through directed evolution.
One of the few advantages of
reading through months of articles stored due to laziness keeping abreast of literature is that on occasion you find something really interesting.
Every once in a while you run into a creationist who claims evolution is "useless" because it makes no predictions and has yielded no new insights or products. While it is obvious to anyone with a rudimentary understanding of the ToE itself and the experimental evidence supporting it that these claims are blatantly false, it never hurts to have some examples ready to make your point.
The concept of directed evolution is such an example.
Although the method has been known for some time this article does a nice job in explaining the most common methods used in this approach and gives a number of recent examples.
The article starts with the observation that directed evolution has rapidly replaced rational design in protein engineering. The reason for this is that our understanding of protein structures and functions is still rather limited. Directed evolution in contrast does not rely on a detailed understanding of the relationship between structure and funtion, but instead uses the darwinian principles of mutation and selection. As long as one can generate sufficient diversity and has a method for screening and selecting the desired variants this approach can be applied succesfully, even without a complete understanding of why they work.
Several methods are available for generating diversity and these can be divided into two main groups: non-recombinative and recombinative.
The non-recombinative methods create diversity through point mutations, insertions or deletions, either in a single location, in a small region or across a whole gene. The simplest way of introducing mutations in a gene is the error-prone polymerase chain reaction (EP-PCR), which unreliably amplifies genes with a modifiable amount of mutations per generation. Typically low mutation speeds (1 or 2 per generation) are used although people have reported succesful optimisation with close to 30 errors per generation. Using this method a highly active TEM-1 beta-lactamase mutant could be created.
The second method is recombinative, in which similar genes from different sources are recombined. The most popular of these method is "DNA shuffling".
Step 1: mix related genes and digest under controled conditions. Step 2: Denature and mix. Step 3: Recombine into random selections. Step 4: Amplify products
In this approach the genes of interest are fragmented under controlled conditions followed by a reassambly step. This then results in a random combination of two or more genes. Recent succesfull examples include the E.coli beta-glucuronidase gene, from which a mutant was generated with inverted specificity for another compound and a three order of magnitude increase in activity.
The most critical step in directed evolution experiments remains the screening. Functional genetic selections, where the host organism can only survive if the desired activity is present are the most usefull. However, for a number of properties this is unworkable, and for these it remains necessary to assay each mutant individually. A number of clever methods have been generated for individual cases, but these don't seem to lend themselves for generalisation.
The authors go on to describe a number of systems that have been succesfull and also discuss some that may benefit in the near future from directed evolution. The most interesting one, IMO, is the "genetic circuit" described in another paper(abstract). By the looks of it, ID creationism arrived on the scene just in time to get nuked by the advances in this field.