Saturday, August 21, 2010


Interesting progress is afoot in designing enzymes from scratch.

While the vain, egomaniacal self-sequencing Craig Ventner has grabbed headlines for "creating life", more interesting and useful projects are afoot. While his employees have inserted a replica of a natural genome into an existing cell and gotten it to propagate, that was never an interesting problem. We have altered the DNA of organisms for decades through recombinant DNA technology. Whether replacing large amounts or small amounts, the true intellectual difficulty is not whether copied DNA can work in a host cell, it is how to design such DNA to do new things- making novel genes and gene networks that inform us about their biology, synthesize new chemicals, cure diseases, create truly new life forms, etc.

So I was far more impressed by a recent pair of papers in science about new successes in designing enzymes, in one case more or less from scratch and by computational means (accompanied by a review). It is the protein that is the main actor in cells and in life. Anything that needs doing in biology, from digesting food to lifting weights, is done by proteins. The only thing they don't do is store information for their own synthesis, and all the informational and catalytic tasks done by RNA.

Proteins are also dauntingly complex- little dynamic chemical packages that fold into intricate shapes and sometimes harness advanced quantum mechanics, (advanced for us, that is), like in the photosynthetic reaction center), to do magical feats of chemical transmutation. No wonder that it has been extremely difficult for scientists to consider themselves capable of designing new proteins- evolution has set the bar very high. Even understanding how existing proteins work has been hugely challenging. X-ray crystallography has allowed scientists to gain detailed pictures of protein structures, but these are static images. Much of the magic of protein action lies in their dynamics, where complex electronic surfaces and structural rearragements take place on a routine basis.

A classic example is myosin, which uses ATP to bend its head against actin to power our muscles, in a power cycle that has only recently been elucidated. But changes in shape are common, occuring also, for example, in hemoglobin as it picks up oxygen and then dumps it out in the peripheral tissues. These dynamic aspects, along with electronic and even quantum mechanical elements, have made proteins quite difficult to understand, and thus also to model and design for our own purposes.

Nevertheless, Siegel et al. decided to design a novel enzyme for a reaction that, as far as known, is not carried out by any biological enzyme- the joining of two separate molecules into one ring structure in a Diels-Alder reaction. Step one was to imagine the ideal transition state for the reaction- the structure of the reactants just as they cross the energetic divide between being unjoined and being joined. In organic chemistry, this state is typically promoted by using high pressure and exotic catalysts like niobium pentachloride. Enzymes do utilize a wide variety of metals, (like iron to bind oxygen in hemoglobin), but these workers wanted to start simple and begin with just protein-based building blocks. This involved using quantum mechanics and organic chemistry to model that state in terms of both its shape, the electronic fields that would stablize it, and some key hydrogen bond acceptors and donors that could help the reaction along.

Image of one designed catalytic active site (left) with the substrates in color.
At right, the scaffold's structure in green (right), and the designed changes shown in red.

Step two was then to attempt to design a protein which would supply a "pocket" that could fit that transition shape while allowing the reactants access and giving them extra assistance with properly shaped electric fields and hydrogen bond donors, etc. This was done using protein design software based on the relevant chemical principles. The hard part was then next to translate this small "pocket" design into a protein that could provide the backbone and folding to bring such a pocket together.

For this, they consulted a library of known small protein shapes, (which they called scaffolds), and chose 84 candidates for synthesis into proteins. This means that they read out the protein scaffold sequence, then superimposed their designed active site onto that sequence, substituting amino acids as needed in key places, (involving 13 mutations in one case), and lastly back-translated it all (conceptually) into the DNA of a gene they could synthesize on a machine. This DNA was then linked to a promoter that would drive its synthesis and inserted into the genome of a bacterium. They then grew up a bunch of these cells, popped them open and purified out the translated (actual) protein, and tested it for its ability to carry out the new Diels-Alder reaction.

What did they find? Only two designs had detectable, though paltry, enzymatic activity, with turnover of about 4 reactions per minute. Then they went back in, looked under the hood with their modelling programs, and made a few additional mutations on one of these candidates that then increased this catalytic rate a hundred fold. Not too shabby!

Schematic of the reaction and its transition state which was designed around.

What these experimenters do not go on to do is harness the power of evolution to optimize their design. That is doubtless because they do not have a selective test for their reaction- i.e. a way for cells to depend on successfully carrying out this reaction which is not biologically significant. That could doubtless be arranged, but to show something of the sort, another paper in the same issue takes over the story.

In this case, industrial biochemists with a small biotech (Codexis) and partner Merck wanted a better way to perform a pharmaceutical synthesis previously catalyzed by rhodium under harsh conditions, by doing it with an enzyme. They were able to find an existing enzyme that performs a similar reaction (transamination) and used a bit of design, and a lot of mutagenesis and quasi-evolutionary selection to optimize it to industrially useful levels. A bonus of such a switch is that the new process is more stereo-specific, (as enzymes typically are), creating precisely the correct product, suffers from no rhodium contamination of the products, and can be done under mild conditions, rather than the 250 psi previously used. This is a good example of "green" chemistry.

In brief, these workers slightly opened the binding pocket of their candidate enzyme (from a soil bacterium) using molecular design software to allow it to have some activity on their chosen (unnatural) substrates. They then mutated the active site with abandon to find improved variants, and then finally unleashed the enzyme in more general mutagenesis + selection system to accommodate several other parameters needed for large-scale synthesis. This was done by creating libraries of mutated variants, (35,000 in all), and using industrial-scale screening to test each for its activity individually. They also optimized for performance in their chosen conditions of high substrate concentration and a high level of organic solvents- conditions that a natural enzyme would never see.

The result? An enzyme that is industrially usable, tens of thousands-fold faster than the original semi-designed natural enzyme, giving 100% correct product in 50% DMSO solvent at 40ÂșC with no heavy metal contamination and less waste and cost all around. While these researchers didn't use a true evolutionary system to optimize their enzyme, (it was not done in cells using an endogenous replication system), there are numerous ways to do this kind of thing, and the principle is the same- make lots of mutations, test them, then take the best candidates and repeat the process.

So, if you are after truly novel functions rather than vanity projects, focus on proteins- the workhorses of life- to do new and useful things. The ability to design truly novel molecular functions opens enormous vistas in both biotechnology and in the more heady project of tinkering with life itself.

  • Krugman keeps getting it.
  • A song for our times- Pieces of a man, by Gil Scott-Heron.
  • How much does consciousness owe to language?
  • The "ground zero" mosque debate: cynically divisive and unnecessary, except for Neanderthals.
  • Bill Mitchell quote of the week, regarding public surpluses and supposed public saving schemes like Social Security:
"Thus the concept of a fiat-issuing Government saving in its own currency has no meaning. Governments may use their net spending to purchase stored assets (spending the surpluses for instance on gold or in sovereign funds) but that is not the same as saying when governments run surpluses (taxes in excess of spending) the funds are stored and can be spent in the future. This concept is erroneous. Please read my blog – The Futures Fund scandal – for more discussion on this point."

No comments:

Post a Comment