Synthetic biologists believe it's possible to design organisms out of kits of parts, but evolution has a few nasty surprises for them, E&T reveals.
"I think I may have been one of the first people to have their picture taken with fecal matter," jokes Professor Andrew Ellington of the University of Texas at Austin. Ellington was not the model for a piece of concept art or an undergraduate prank but the willing model for an experiment that set out to demonstrate how living organisms could be made to behave like photographic film.
They developed the image on a glass plate full of bacteria, reprogrammed by adding extra DNA to them.
Ellington is a little embarrassed to have his own image recorded by a landmark experiment in synthetic biology and later points to one of Charles Darwin they made using the same technique. "In the US, we actually celebrate Darwin Day because we have to. Our picture of Chuck is a homage," Ellington explains, referring to the running battle between creationists and devotees of intelligent design against the "evilutionists".
The work at Ellington's lab among others may pay Darwin greater homage than a copy of a 150-year-old image. It is showing that biological design is a lot harder to achieve than simply letting the processes of evolution run their course.
A few thousand kilometres from Ellington's base in Austin, scientists from the Massachu-setts Institute of Technology (MIT) decided that the time had come to bring engineering discipline to biology. The view of Professor Tom Knight and former students such as Drew Endy, who has gone on to head up his own lab at Stanford, and Ron Weiss of Princeton University, is that biology can be engineered. For them, genetic engineering is a misnomer: little more than tinkering with pre-existing genomes and hoping that something useful pops out. They see synthetic biology as the way to design in specific traits simply by piecing DNA strands together in a molecular jigsaw puzzle.
There are many aspects of biology that remain a mystery but Knight argues: "There is a whole set of people who have the outlook that biology is inherently complicated that every system will have to be handcrafted. I just reject it. Yes, we are in a very primitive state right now. But I think in another five years we will have a very robust and contained set of components that will let us put components together and it will change the world."
Synthetic biology is where life sciences meet engineering head-on, and is arguably the most cross-disciplinary part of current biology. Many of its practitioners are not life scientists by training. Knight started as an electronics engineer and crossed over into synthetic biology when he realised that conventional methods for making circuits were running out of steam.
"The problem in microelectronics is not that you can't make the transistor, but the device, once you make it, is not going to be under control," claims Knight. "You can try to fix it but the problem is that the devices are being fabricated with a statistical process. You stand back with a shotgun and blast dopants at it and hope for the best. That is not going to work in this century. You need to put atoms where you want them. That is chemistry and not physics and the best kind of chemistry is biochemistry."
Ellington has a conventional biochemistry background but his main collaborator is a computer programmer, Zack Booth Simpson, who moved from games developer Electronic Arts to designing life.
"Zack likes to talk about how, when building his systems, he did it in a top-down fashion," says Ellington. But biology requires a different approach. "What we do with molecules, we are building from the bottom up. For that, you have to have lots and lots of self-organisation."
Beyond a certain level of complexity, Ellington explains, top-down design fails to work. But nature has proven able to construct complex systems using comparatively simple 'programs', if you regard the DNA that encodes them as a kind of instruction set.
Professor John Mattick, director of the centre for molecular biology and biotechnology at the University of Queensland points to how little DNA it takes to provide all the information required to make a human being. "After a few glasses of wine, I like to say that biology can teach IT a thing or two," he says, pointing to the fact that the human genome requires only about 6MB of storage. "That is not as much code as there is in Microsoft Windows, but I can walk and talk."
DNA information storage
The way that DNA stores information intrigues Professor Ron Weiss of Princeton University. A computer scientist by training, he told scientists at a Royal Society of Chemistry seminar on synthetic biology last autumn: "Twelve years ago, I became fascinated by the notion that we might be able to program cells with the ease with which we program computers."
"One thing that happens in living systems is this idea of emergent systems. You have elements that engage in localised decision making. But out of that, interesting global behaviour emerges," says Weiss.
Some of the experiments conducted by Weiss provide hints as to the way in which animals form spots and stripes on their coats through chemical signals exchanged by cells, and ultimately to the way that organs form. He started by trying to coax cells into producing coloured rings on the surface of a petri dish. "It's not too difficult to engineer on paper. It took about 20 minutes
to put the slide together in Powerpoint. But it took three years to do the implementation.
"One of the reasons we wanted to do this was to work out whether we could program spatial patterns by design and, based on these patterns, have the cells differentiate into bone, muscle, cartilage or organs - engineer tissues in ways that are not possible unless you reprogram the cells."
Weiss's team developed a mathematical model based on an idea put forward by Alan Turing, and from that came up with a set of genes, one of which provides the recipe for a protein that makes a fluorescent dye, that would interfere with each other. "We did a simulation on a lawn of virtual cells in the computer. We found you can form a variety of patterns: dots, lines and so on."
A movie of the experimental systems compresses 40 hours of relative inaction into a few seconds that show how patterns such as stripes form on an otherwise undifferentiated lawn of cells. "For ten hours, nothing happens. Then the cells start to make decisions about the kinds of cell they should be. We are building a model to understand the differences between the domain sizes and the original predictions but I think there is good correlation between the model and the experiment.
"While I am not claiming that we can build the coat of a cheetah or a giraffe, there is some correlation between the lawn of bacteria and what we see in nature," Weiss claims, pointing to the eventual aim. "Can you take an initial cell and form structures that are useful in a medical setting? We are trying to program tissue generation. We would like to put cells in the appropriate places and have them make decisions based on communication between them so they know what to do."
A cure for diabetes?
One of Weiss's initial targets is Type I diabetes, a condition where the immune system turns on insulin-producing cells in the pancreas. "We would like to engineer a system where we start with stem cells and maintain in a patient the level of these beta cells," he explains.
The immune response will still kill off beta cells. Weiss's plan is to engineer into stem cells a control loop that monitors how many of the cells are present and, if the level drops below a threshold, have them divide and differentiate to make more insulin-producing cells. In effect, the team would create a genetic program to control the stem cells.
The process is trickier than it sounds. Early experiments found that the cells produced an all-or-nothing response. The cells synchronised to each other, making the decision to convert to beta cells all at once. The population of cells simply collapsed.
Weiss tweaked the program and added an oscillator with the sole function of breaking symmetry in the system. "With an oscillator, some of the cells go out of synchronisation. So not all of the cells make the decision to commit to become beta cells at the same time," he explains.
The progress so far has been promising. "We have implemented about 80 per cent of the elements that we need for the system," claims Weiss, who reckons within five years, the approach could be trialled.
One potentially big advantage of using synthetic biology to make beta-cell factories is that it makes it possible to use the patient's own cells. Not only can the stem cells have the beta-production logic inserted into their genome, the stem cells themselves can be derived from normal cells using the same kind of genetic reprogramming. The portion of the DNA that leads to stem cells differentiating into other cell types could be deleted, once the parts of the genome responsible for those processes have been identified.
Once altered in a test tube - Weiss says he is wary of trying to do such complex gene therapy in the body - they would be injected back into the patient. A further change might be to have the cells commit suicide should they wind up outside the pancreas. "We would have part of the design there to get the cells to recognise where they were," he explains, using the absence of chemicals peculiar to the pancreas to provide the kill signal.
Evolution in control
Given the progress of these efforts, it is easy to believe that writing software in DNA is not all that much different from writing embedded code for a microcontroller. Recent research has shown how much control theory can inform biology. The Circadian rhythms that control how plants grow and give us jetlag are readily modelled as control loops, although scientists differ on how complex those loops are.
Unfortunately, not everything goes to plan. Living organisms have the nasty habit of working around the modifications made by scientists. Or, as Ellington sums up: "Cells suck".
Lingchong You says the idealised picture of synthetic biology, one that makes the technology look tractable is to picture a wall between the logic circuit that scientists want to implement and the rest of the cell. A word that often crops up in these discussions is 'chassis': the cell into which you insert the genetic logic simply accepts the changes and provides all the necessary support machinery.
"The chassis is an attractive metaphor but not a good one," argues Professor Victor de Lorenzo of the National Centre of Biotechnology, Madrid. "We can't take the chassis, ignore it and put something on top."
"There is no such thing as modularity," claims Alexander Ninfa of the University of Michigan, although he has worked with colleagues on a technique that could make genetic subsystems interfere with each other less than they do today.
"The reality," You explains, "is that there are hidden interactions between the circuit and chassis that result in unexpected behaviour. When we write our mathematical models they are not taken into account."
Very often, the circuit simply fails to operate. More surprisingly, the experiment can seem to work but for all the wrong reasons. In one case, You wanted to build a genetic circuit that acted as a bistable switch. But, after performing some other experiments to confirm the results, You discovered that the cell was showing the right behaviour but not because of the action of the new genetic circuit. The new DNA had caused the cells to grow in a way that resembled the behaviour that You wanted.
In many cases, a genetic circuit's effect on growth is what causes it to fail. This is where evolution makes its presence felt, particularly in fast-evolving organisms such as bacteria.
"I find cells annoying because they are evolving entities," says Ellington, recalling how one experiment was stymied by the adaptiveness of living cells. A protein that the cells were supposed to make was so toxic that the ones in which the circuit worked grew very slowly. The cells that grew and replicated more efficiently were those where the circuit did not work properly. The cell simply evolved the genetic circuit out of the way.
"The cell has a resistance to implantation of proteins that are not part of its normal make-up," says de Lorenzo.
Evolution is a major headache for synthetic biologists who hope to bolt genetic circuits together using readymade parts and subsystems. The BioBricks Foundation, established by scientists from Harvard University, MIT and the University of California at San Francisco, aims to collect DNA with known functions in a public registry and is defining standards for how those components should be made. In principle, bioengineers will be able to pick and mix parts from the library to put together more complex functions. But the tendency of cells to reject the engineering if it turns out to damage their growth remains a problem.
De Lorenzo contrasts the approach taken by those brought up on the MIT school of synthetic biology and one that takes evolution into account, methods that are more common in European labs. He sees evolution as something to be harnessed: it can improve the efficiency of a synthetic genetic circuit and get around the problems that plague proponents of a purely rational-design approach.
"You can make the new system isolated," says de Lorenzo, pointing to the use of subcellular partitions as a way of keeping toxic proteins out of the way of the main part of the cell. Nature uses this very technique to stop cells poisoning themselves. "Or you can let the system evolve by itself the very behaviour that you want it to have."
De Lorenzo wants a protein that changes colour in the presence of 2-4-dinitrotoluene, a common component of landmines. A bacterial colony that produces this protein sprayed on a field would show the location of buried landmines to observers in a helicopter flying overhead using the chemical vapour rising through the soil to trigger the reaction. De Lorenzo's problem was that no such protein exists in nature.
In principle, it should be possible to design enzymes on a computer to perform specific tasks, such as convert a foodstuff into fuel. Conceptually, you take the shape of a target chemical and design a protein that wraps neatly around its major features. A change in shape as the protein wraps around might be enough to cause the colour shift or expose a fluorescent marker.
Unfortunately, not enough is yet known about the mechanics of these reactions. Protein folding remains an area of active research that has millions of computers crunching through quantum molecular simulations every night in the hope it will reveal how it happens in nature.
Professor Alfonso Jaramillo of the Ecole Polytechnic near Paris has come up with simpler models of protein folding to try to make it easier to put together new proteins. His team has built a database of known protein folds that is used by software to calculate how likely it is an enzyme will form a certain shape. But Jaramillo says the software can only get you so far: to get the full function you need to use directed evolution, a technique already used in the second-generation biofuel industry.
In 2003, researchers in Homme Hellinga's lab claimed to have designed proteins that could bind trinitrotoluene, another component of bombs and landmines, and the neurochemical serotonin. Other researchers tried to repeat the experiment but found the designed proteins did not work as expected. Almost five years after publishing the work, Hellinga retracted the papers.
To make his dinitrotoulene-sensing protein, de Lorenzo used a technique called directed evolution. The term is a bit of a misnomer as it's not possible to direct evolution as such - the technique works by having selection favour particular outcomes.
He started with a strain of the bacterium pseudomonas putida that grows in toluene, a near relation to the dinitrotoluene target, and mutated its toluene-sensing protein to try to make it specific to the landmine chemical. Many bacteria that convert a chemical such as toluene have associated proteins that act as sensors - they are key to regulating the production of enzymes that perform the conversion and stop the cell from wasting energy on making too many of them.
"We go to the protein and look at the variants," explains de Lorenzo. In doing so, the team found one reason why rational protein is so hard.
"We found that amino acids not in the pocket could change the specificity of the pocket. So it may be a good idea to not concentrate on the design of the pocket," says de Lorenzo. In the case of the toluene sensor, the pocket hardly changed at all.
The initial mutations made the protein less fussy about the types of toluene it would detect. This happened quite quickly. A much slower process then adapted the protein so that it would only attach to dinitrotoluene. "I think this tells us something about how proteins evolve in nature to acquire new specificities," says de Lorenzo.
"You can't avoid evolution. Darwinian selection exists at every level. But it is something that synthetic biologists who come from engineering don't want to hear," concludes de Lorenzo.
Breaking the code: making shapes
People tend to look at DNA as being just a code in which the only important aspect of the double helix is the order in which the four bases appear. And so, it's an easy leap from the concept of code to that of a computer. There is a growing body of evidence that the shapes that DNA adopts play a major role in the polymer's behaviour. This should not be a surprise: shape plays a vital role in reactions mediated by enzymes.
In chromosomes, DNA sequences that are rich in pairs of the two bases adenine (A) and thymine (T) are often the sites of 'promoters' - lengths of DNA that unzip more readily than others. This gives the enzymes that transcribe DNA a toehold from which they can crawl along the DNA that makes up each gene (see box 'The dogma of DNA').
AT-rich sequences do not just unzip easily. They are bent in a way that makes it easier for the transcription enzyme to stick in some cases but prevent this process in others. Sometimes, the DNA is bent to prevent transcription taking place and it is only when other enzymes attach and alter the shape of the DNA that transcription can get under way. The way that transcription factors bind to DNA plays a big role in the control loops that permeate cell biology. You can put the transcription of one gene, and therefore the production of its corresponding protein, under the control of another simply by altering the promoter sequence to favour the binding of the protein produced by that other gene.
One of the most studied control loops in molecular biology, which mediates how lactose is converted to energy for the cell, uses the cross-interference between genes that transcription factors can produce.
It is even possible to produce basic logic gates from promoter sequences. Working at the Weizmann Institute of Science, Shai Kaplan, Anat Bren and others worked out a way of mapping how a set of 20 genes in Escherichia coli cells react to changes in sugar levels. E coli is able to use several different sugars as sources of carbon:
In principle, the sugar genes are under the control of logic AND gates that are embedded in the promoter. In practice, the logic functions that involve more than one transcription factor are more complex, often resulting in mixtures of AND and OR functions. In artificial promoter design, researchers have succeeded in creating a variety of logic gates.
As with much in synthetic biology, it is easy to believe that promoters provide the key to transcription. But the situation is more complex. A single gene can give rise to more than one transcribed product, a phenomenon known as pleiotropy. The DNA can even be read in reverse, a property that some think could be used to prevent cells evolving synthetic circuit out of the picture.
Writing in the journal BioEssays, Anne Buchanan and colleagues from Pennsylvania State University and Yale University School of Medicine, claimed: "Most genes are pleiotropic, and for few if any do we know all their functions."
Despite all the coverage of scientists having found genes for everything from diabetes to religious belief - a tendency lampooned by scientists such as Denis Noble - genes often work together and their products recruited to help in many different jobs.
Even the idea that a gene codes for a single protein has been successfully challenged in recent years. The Encode project has found that DNA is promiscuous, any part of a given strand can wind up providing the template for a protein or a piece of RNA that then acts as an enzyme or interferes with some cellular process.
Tom Gingeras at the Cold Spring Harbor Laboratory says: "The knee-jerk response when someone says the word 'gene' is to think of a protein-coding element."
Promoters do not so much switch transcription on and off, they are more like amplifiers. The RNA polymerase enzyme seems to be able to attach to other parts of the DNA strand and just start pumping out RNA, much of which may be nonsense. However, it may have a purpose.
Buchanan writes: "Pleiotropy may be widespread because new traits evolve by reuse of existing genetic pathways by relatively easy change of expression context that can be brought about by mutation in a gene's flanking regulatory sequences."
Genes rarely act alone and some of this unexpected transcription can lead to new links forming in the network of control paths that link them together, possibly with a beneficial effect for the cell.