Designer enzymes: Learning to rival nature's wizardry

• 16 July 2012 by James Mitchell Crow
• Magazine issue 2873. Subscribe and save (New Scientist)

Biologically inspired catalysts could usher in a revolution in chemistry – if only we can figure out the best way to create them

RICHARD HECK, Ei-ichi Negishi and Akira Suzuki may have won the Nobel prize for chemistry in 2010, but they are actually pretty average at what they do.

Not that you can fault their work finding new ways to make complex organic molecules. It is just that nature is so much better at that sort of thing. You, me, the birds and the bees, even educated fleas, trees and the humblest bacteria: we are all constantly churning out the sort of carbon-containing stuff that the Nobel trio spent their careers learning to synthesise. As carbon-based life forms, we just can't help it.

But what wouldn't we do to go one better. If we could harness our own chemical powers, that could lead us to new drugs, greener materials and more efficient ways to generate energy. To do that, though, would mean unlocking nature's catalysts - something that has been eluding us for decades.

From the humblest lab to the largest petrochemical refinery, catalysts lie at the heart of much successful chemistry. Unchanged in reactions, they latch on to the reacting stuffs and assist the flows of electrons that accompany the breaking and making of chemical bonds.

Artificial catalysts tend to be relatively rudimentary affairs. Often they consist of little more than a single reactive metal atom that acts as a temporary sink and source for electrons, plus a few organic-based side-chains that guide the approach of reacting molecules. Heck, Negishi and Suzuki's prizewinning carbon manipulations, for example, used palladium-based catalysts as chemical matchmakers to create new organic compounds that have commercial uses ranging from anti-inflammatory and asthma drugs to fungicides. Ruthenium-based catalysts underlie what is arguably the most important artificially catalysed reaction, the Haber-Bosch process that makes ammonia for the chemical fertilisers sustaining about 40 per cent of the world's population.

Nature's own catalysts - enzymes - are a little different. Their hearts are often metallic too, generally made of naturally abundant elements such as iron, copper and nickel. But rather than being surrounded by a few side-chains, this metal core is embedded in a protein chain scrunched up into a complex three-dimensional structure, rather like a dense tangle into which the cord of your headphones might self-assemble.

This tangle is far from random. An enzyme's shape is determined by the patterns of electrostatic attraction and repulsion between the amino acids that link together to make protein chains. They form a cavity around the metallic core perfectly shaped to accept the particular molecules that the enzyme has evolved to manipulate. Rather than using chemical hooks to get a limp grip on their prey and haul it in as artificial catalysts do, enzymes envelop molecules like a glove fitting over a hand.

Within this firm grasp, molecules are stitched and unstitched with a proficiency that makes the best of chemists green with envy. "Enzymes are the product of four billion years of evolution, and they're pretty darned good at what they do," says Frances Arnold, a chemical engineer at the California Institute of Technology in Pasadena. They spit out rearranged molecules hundreds or even thousands of times a second, compared with the measly few molecules a second the average industrial catalyst can manage.

The hand-in-glove fit also means that enzymes work only on specific molecules and create only specific products, with none of the unwanted by-products that often gunge up conventionally catalysed reactions. Add to this selectivity the generally widely available elements that enzymes rely on, and you can see why we'd love to get to know them better.

But how? Most enzymes are fragile and take badly to life outside biological cells. Their complexity is difficult to engineer, and nature has evolved enzymes only for the reactions it wants to do. When it comes to tweaking enzymes to assist different reactions, that selectivity, so vital in the busy confines of a biological cell, becomes annoying exclusivity.

But enzymes also give us a way under their skins. "Proteins adapt readily," says Arnold. The first attempts to re-engineer enzymes to do different things, in the 1980s, followed the path of "rational design". That involved taking X-ray snapshots of crystallised enzymes and using 3D modelling to work out how amino acids shaped an enzyme's structure. The enzyme's genetic code was then tweaked to include different amino acids, the altered blueprint inserted into a tame bacterium to churn out the modified enzyme, and its catalytic effect on a particular reaction tested.

That was the theory, anyway. Unfortunately, enzymes are dynamic structures in constant motion, and could only be partially captured in X-ray snapshots. Even then, the resulting structures were so complex that it was impossible to predict with confidence how enzymes' structure influenced what they did. "Details matter, and we don't understand the details," says Arnold.

In the mid-1990s, she and others began to tinker with an alternative approach, "directed evolution". This used automated processes to shuffle genetic codes and generate huge numbers of mutant enzymes. The mutant with the best catalytic performance for a particular reaction was used as the starting point for a new generation of shuffling, thus gradually evolving an optimised enzyme.

That meant you didn't necessarily need to know the details of an enzyme's structure to find one that worked, and the ability to use genetic codes to generate and optimise structures at will became another reason to favour enzymes over conventional catalysts. "There's no counterpart in chemistry to evolution; you can't evolve chemical catalysts," says Arnold.

There were distinct successes. In the late 1990s, Manfred Reetz, a chemist then at the Max Planck Institute for Coal Research in Mülheim an der Ruhr in Germany, used directed evolution to make enzymes that catalysed chirally selective reactions. The living world is built from structures that can exist in two mirror-image, or chiral, forms, but only one of the two exists naturally. This asymmetry is a fundamental existential puzzle, but in practical terms it means that any molecule intended to interact with nature, such as a drug molecule, must also have the correct chiral structure. In a landmark paper in 2001, Reetz and his team used directed evolution to morph an enzyme that churned out equal amounts of the two forms of an organic ester molecule into an enzyme that produced 50 times more of one than the other (Angewandte Chemie International Edition, vol 40, p 3589).

But this was patience-sapping work. Reetz's efforts involved generating 50,000 enzymes and only scratched the surface of possible mutants. Even a small enzyme 100 amino acids long would have 20¹⁰⁰ possible mutants if each of the 20 amino acids that exist in nature were tried at each position along the chain - more than the number of atoms in the universe. Generating large mutant libraries isn't so hard to do, but testing the performance of every one is a monumental task. "Screening is the bottleneck of directed evolution," says Reetz.

Rational design

And so Reetz and others set about combining some of the strengths of rational design and directed evolution. Rational-design models might indicate, for instance, that a particular amino-acid sequence was the key to forming the glove structure around the metallic core. By focusing the directed-evolution process on this sequence, the numbers of mutants to be made and screened could be made more manageable. In 2005, Romas Kazlauskas at the University of Minnesota, Twin Cities, and his colleagues used this "rational directed evolution" approach to create an enzyme five times more efficient at generating a basic chemical building block known as methyl 3-bromo-2-methylpropionate (Chemistry & Biology, vol 12, p 45). When Reetz and his team revisited their chiral ester enzyme in 2010, they needed just 4000 mutants to find an enzyme almost 600 times as likely to spew out one mirror form of the molecule than the other (Journal of the American Chemical Society, vol 132, p 9144). "It was like day and night, it was a fantastic result," says Reetz.

Such breakthroughs mean we are finally in a position, in some instances, to start to equal nature's chemistry - and perhaps better it. "We can make enzymes that are just as good on their new substrates as their ancestors were on their natural substrates," says Arnold.

She and her team have been using rational directed evolution to make fuels touted as a greener alternative to conventional gas and oil. They have modified an enzyme that normally oxidises large molecules called fatty acids to make it oxidise propane, a molecule so svelte it is a gas at room temperature, into the promising liquid biofuel propanol.

The real deal, though, is to string together many reactions and take simple starting materials straight through to complex products, as nature does in cells. Getting teams of enzymes to work together to build a molecule piece by piece is a tough task, but Arnold's team is using this approach to convert plant stuffs into isobutanol, a biofuel that behaves like conventional petrol. Others such as Jay Keasling at the University of California, Berkeley, and his team are developing teams of enzymes to produce biodegradable polymers, clean up environmental contaminants and generate the antimalarial drug artemesinin.

For all such advances, however, the techniques we have developed so far are frustratingly limited. Rational directed evolution relies on having at least some clues from modelling as to what parts of an enzyme do what, and often we don't even have that, says Arnold. That takes us back to the numbers game of straight directed evolution.

An elegant workaround might come from David Baker of the University of Washington in Seattle and his colleagues. Their approach is a bit like rational design, only backwards. Rather than attempting to model an existing enzyme and then tweak it, they start by making computer models of the molecules they wish an enzyme to work on: the hand, rather than the glove. They then build a virtual glove around it, and search a library of protein structures to find a close fit. The genetic instructions are then passed to a bacterium that spews out the enzyme for testing.

Baker and his team's most celebrated product to date came in 2010, with an enzyme that can perform the Diels-Alder reaction to make cyclohexane structures (Science, vol 329, p 309). Although these six-membered rings are a common molecular motif throughout nature, some quirk of evolution means that enzymes have never evolved to make cyclohexanes in a Diels-Alder-style single transformation, preferring a more roundabout approach. The result might seem one up for human chemistry, but again the procedure does not tick all the boxes.

"On the positive end, we've been able to create enzyme catalysts for a number of reactions for which there weren't previous enzyme catalysts," says Baker. "But on the not-so-positive side, the ones we've designed so far are far less active than naturally occurring ones."

Fortunately, his is not the only promising approach to teaching old enzymes new tricks. Another is to develop hybrid enzymes, in which structures from the most effective artificial catalysts, such as Heck, Negishi and Suzuki's palladium cores, are inserted into the more effective structure of a natural enzyme. "It is really combining the best of both worlds," says Thomas Ward of the University of Basel, Switzerland.

Last year, Ward and his team engineered a hybrid enzyme that could get a metathesis reaction going (Chemical Communications, vol 47, p 12065). Like Heck and his partners' reactions, this class of reaction, which garnered Yves Chauvin, Robert Grubbs and Richard Schrock the 2005 Nobel prize in chemistry, involves a chemical dance in which participating molecules switch partners - A-B and C-D become A-C and B-D, for example - to form new combinations of organic molecules used for everything from pharmaceuticals to polymers for more durable baseball bats. "Metathesis is a very fashionable reaction," says Ward, "and enzymes don't know how to do it."

So far, the process is no faster and no more selective than that using the raw artificial catalyst. But the next step will be to exploit the glove-like properties of the protein structure to control how the metathesis partners come together. Reetz has also worked on hybrid enzyme catalysts, and has high hopes of such work. "On paper, it's a concept that could solve any problem," he says. "But there is a lot of basic research still to be done."

One big hurdle is to get directed-evolution techniques to work with hybrid enzymes. At the moment, cellular junk left over from the bacteria used to churn out the enzymes sticks to and inactivates the reactive metal centre. To avoid this, thousands of enzymes all have to be carefully purified before use. Ward thinks he has finally cracked that problem, but won't give details. "We're filing the patent right now," he says.

Like Arnold, Ward's immediate aim is to use his hybrid enzymes for generating biofuels. But the sheer versatility of carbon chemistry puts few limits on what is up for grabs, if only we can learn to combine our ingenuity with nature's highly evolved systems. Arnold's ultimate ambition is to bypass nature's inefficient photosynthetic machinery and optimise artificial enzymes to directly make useful fuels and other chemicals from carbon dioxide, powered by sunlight - a dream of chemists for decades (New Scientist, 14 April, p 28). It's a dream that supplies all the motivation we need for efforts to crack enzymes. "Nature does it," says Arnold. "So some day we should be able to do it too."

James Mitchell Crow is a freelance writer based in Melbourne, Australia