Getting hold of cells for research is not difficult. You can grow colonies of them in a petri dish, clone them in a test tube, or lift some right out of your own skin, for that matter. But microbiologist Jack Szostak has opted for a less conventional route: he is building his own cell from scratch. ìIím trying,î he says casually, ìto create life.î
Without question, a cell passes any test for being alive. A single-celled organism like a bacterium can eat, grow, adjust to the environment, reproduce, evolve. So if Szostak succeeds, he will indeed have created a living organism of sorts, albeit one not quite like any other known. His creation will have a bubblelike membrane akin to the outside of a natural cell, a noteworthy development in itself. But Szostakís real triumph will be the tiny package within the membrane, a package consisting of several strands of custom-built RNA. In natural cells RNA often serves as a messenger molecule. It carries blueprints from the cellís library of instructions--its DNA--to factories where proteins are made according to those instructions. But Szostakís RNA will do something that seemed impossible until just a few years ago: it will not only carry instructions but also execute them. And those instructions will tell each strand of RNA to build copies of itself--a function widely considered to be the very essence of life.
To make his scheme work, Szostak has to simplify the cellular machinery that exists in nature. Instead of a bagful of specialized molecules--DNA for storing blueprints, RNA for transferring instructions, specialized proteins for a variety of cellular construction jobs--he needs one RNA molecule that can do everything. It has to carry its own blueprint, transfer the blueprint to copies of itself, and use parts of itself as the copying tools. Puttering away inside its membrane, reproducing, and possibly even evolving, Szostakís RNA, it seems safe to say, will be the most lifelike thing ever to emerge from a test tube.
Yet oddly enough, Szostak may not be the first to create this strange entity. Nature beat him to it by several billion years, according to a theory now widely accepted among biologists. A self-replicating strand of RNA, many believe, was a precursor to all the more complex, DNA-based forms of life that exist on Earth today. ìItís generally believed that there was a time roughly four billion years ago when RNA was running the show,î says Gerald Joyce, a leading RNA researcher at the Scripps Research Institute in La Jolla, California. ìWe just donít know how an RNA life-form came about.î However it may have happened, that original RNA has long since disappeared. The all-purpose molecule gave way to specialized offspring that were far more efficient, when acting in concert, at maintaining and reproducing themselves.
Just how nature managed to come up with the portentous first strand of self-replicating RNA has been a deep and long-standing mystery. One might think that viruses could provide a clue, because many viruses carry all the self-reproduction information they need on single strands of RNA. But those strands canít act on their own: they insert their instructions into the complex machinery of host cells theyíve invaded.
Today every known type of replicating RNA needs the help of enzymes. These flexible molecules enfold the single strand of RNA and, using this original strand as a template, string together building-block molecules called nucleotides into a complementary copy of it. If you think of the original strand as being made up of four differently colored types of nucleotide--say, blue, orange, purple, and yellow--then each nucleotide of the new strand will be complementary to its original counterpart. Where the original has a blue, the complement will have an orange; where it has a purple, the complement will have a yellow. The complementary strand becomes like half a zipper, interlocking with the original RNA all along its length. Eventually the two strands come apart and another enzyme repeats the process on the complementary strand, stringing together nucleotides that are now a precise copy of the original RNA. The complementís orange is now matched by a nucleotide that reproduces the originalís blue; its yellow by one that reproduces the originalís purple.
The enzymes that do all this work are a type of protein, which means that they must be assembled according to instructions embedded in RNA. In other words, RNA must direct the assembly of the enzymes that put together more RNA. Itís a nice system--but how could it ever get started? Without enzymes, something like todayís RNA couldnít have copied itself and evolved; but without highly evolved RNA, you couldnít have any enzymes. It is biologyís ultimate chicken-and-egg question.
Nearly 30 years ago researchers posited an elegant way out: a primitive RNA molecule that could also act as an enzyme for its own replication. Two of these identical molecules acting in concert--one as enzyme, the other as template--might have been able to churn out a third, complementary molecule unaided. The complementary molecule could then have turned out a replica of the original RNA, which would then crank out another complementary molecule, and so on. Eventually, after many mistakes that didnít survive and a few lucky improvements that did, the RNA could have evolved into more complex versions capable of synthesizing separate enzymes to carry out the job more efficiently. And the rest, as Darwin might have said, is natural history.
The only problem was that no one had ever found a form of RNA that could do double duty as an enzyme, and so this particular origin-of- life scenario came to be regarded as no more than an interesting speculation. But in 1981 biochemist Tom Cech found a strand of RNA capable of performing some simple enzymatic functions, a strand he had isolated from a paramecium-like pond creature called Tetrahymena. That discovery earned Cech a Nobel Prize three years ago and seemed to give the self- replicating-RNA theory the boost it needed.
But the sense of triumph was soon muted. ìWhen RNA enzymes were first discovered,î says Joyce, ìthere was a flurry of papers and commentary saying, ëWell, that solves it, we can see there was an RNA life-form.í But the difference between what these molecules could do and self-replication is actually quite significant.î Specifically, it had quickly become clear that Cechís RNA, along with 80 or so related enzymatic RNAs that were discovered in the next few years among a variety of microorganisms, plants, and fungi, couldnít manage much more than hacking apart its own strands in a few specific places and perhaps sticking a few nucleotides onto one end-- an enzymeís handiwork, to be sure, but a far cry from replication. And no one saw a way to prove that some as yet undiscovered version of RNA could do better. Thatís when Jack Szostak decided to get involved.
Szostak doesnít fit the mold of his specialty. Many origin-of- life biochemists are flamboyant, outspoken characters. The soft-spoken 39- year-old Szostak, in contrast, exhibits little affinity for the limelight and politics of high-powered research. He works in the mazelike interior of the new building in Boston that serves as Massachusetts General Hospitalís research digs, and it would be easy to mistake him, in his casual attire, for one of the many thousands of graduate students slaving away in the bowels of Greater Bostonís bevy of universities.
Seven years ago Szostak began the hunt for an RNA enzyme that could be coaxed into self-replication. He hadnít picked the worldís easiest chore; there was a biochemical chasm between a molecule that could cut itself apart and one that could deftly weave nucleotides into a precise copy of itself. ìThe general feeling was that I was trying to do the impossible,î says Szostak. ìBut I was optimistic. I really did believe that the origin of life happened this way and that if it did, I should be able to do it in the lab.î
Szostakís overall strategy was dazzlingly unexciting: Think small. ìIf you looked at this goal in terms of a big jump, it really did seem impossible,î he says. ìBut if you looked at it as the sum of a lot of little, manageable steps, it didnít seem as hard. So we decided to simplify the problem by breaking it down into several smaller ones. Of course, we knew that any one of these steps could have been a roadblock that would have stopped us cold. We just tried not to think about that.î Joined by graduate student Jennifer Doudna in 1986, Szostak got to work.
Step one was to take Cechís Tetrahymena RNA enzyme, with its ability to snip itself apart, and get it to perform its chopping trick on a separate molecule, since an RNA strand serving as a replication enzyme would have to act not on itself but on another strand that would be acting as a template. To that end, Szostak and Doudna divided the Tetrahymena RNA into two pieces: a large chunk of about 300 nucleotides to perform the splitting action, and the remaining chunk of some 40 nucleotides that was the target site for the splitting.
After synthesizing the two strings separately in a test tube, they confirmed that the larger string on its own could indeed slice apart the smaller, unattached target string. Even better, the researchers were able to make the reaction run in the other direction--that is, they showed that the larger string of RNA, like other splicing enzymes, could not only cut but join, or ìligate,î the errant halves of the target back together. Thus the team had come up with an RNA enzyme that could convert two small strings, called oligos, of about 20 nucleotides each into a single longer strand. ìIt was a very important, very exciting first step,î says Szostak.
Next Szostak and Doudna had to find a way to get their enzyme to do more than just put back together the two severed halves of the target RNA. What they were looking for, after all, was replication, and for this the enzyme would have to join free-floating oligos--oligos that would be synthesized by the researchers--into a sequence that would be complementary to a target strand. (That is, when the assembled oligos were laid alongside the target, everywhere the target had a sequence of nucleotides that ran, say, blue-yellow-orange, the complementary oligo would have a string of nucleotides running orange-purple-blue.) The enzyme couldnít join together just any two oligos; it had to use the target as a template. Otherwise the oligos would be strung together randomly instead of in a precise sequence-- hardly a mechanism for accurate copying.
If the enzyme was to follow a template faithfully, it would have to weld together only oligos that were complementary to the template at every point along their length, and to refrain from joining those sequences that didnít quite match. There are four types of nucleotide in RNA (corresponding to our four color-coded versions), and they have preferred pairings: adenine (A) prefers uracil (U) as its complement, while guanine (G) prefers cytosine (C). These two happy couples, A-U and G-C, zip right up; they are called Watson-Crick pairs, while other, less felicitous combinations are called wobble pairs. When an enzyme comes by, all it has to do is make sure that each nucleotide on the oligos is lined up next to its Watson-Crick pal on the template, so that there are no wobble pairs, and then weld together the ends of the oligos. When the complementary strand in turn becomes a template, it will reproduce the original strand simply by following its natural preferences.
Unfortunately, the Tetrahymena RNA enzyme seemed determined to join oligos that incorporated wobble pairs, a habit that would introduce mistakes later on in replication. This was a serious problem: it would prevent the enzyme from accurately copying long chains of RNA.
To remedy the situation, the two researchers looked for a way to modify the enzyme, the template, or the oligos just enough to get matches that were entirely Watson-Crickish. But given the complexity of enzymatic reactions, this sort of fine-tuning is always a hit-or-miss proposition; Szostak and Doudna therefore decided simply to throw chemicals at the enzyme and hope one of them did the trick. One of the first substances they tried was spermidine, a small, electrically charged molecule that had been shown in other labsí experiments to twist the molecular structure of RNA.
It was a shot in the dark, but luck was on their side. When the researchers added spermidine to the RNA mixture, the enzyme readily ligated oligos perfectly matched to the templates, without wobble pairs. The researchers are still not sure what happened. Whether the spermidine altered the shape of the enzyme, the template, or the oligos, or all three for that matter, they couldnít say. ìThe bottom line is that it worked,î Szostak says with a shrug.
Pressing their good fortune, Szostak and Doudna then tried to get the enzyme to ligate not just oligos complementary to the small target chunk of Tetrahymena RNA but also a variety of oligos complementary to longer, synthetic RNA strands cooked up in their laboratory. Again the enzyme worked perfectly, joining as many as five different oligos into a single strand in a sequence that made the whole string complementary to the synthetic RNA. The chances of those oligos bumping into each other in that order by themselves, without a template, and at the same time joining end to end were almost zero. Clearly, the reaction was working just the way Szostak wanted it to: oligos were bumping into the synthetic RNA and sticking to a complementary part of it; the enzyme was then jumping in and joining those oligos end to end.
If the enzyme could perform this trick on synthetic templates, why shouldnít it be able to make copies of itself? By 1988 Szostak and Doudna were sure they were hot on the trail of self-copying RNA. ìThis was the first thing that really looked like replication,î says Szostak.
In fact, what they had come up with fell short of self- replication: their enzyme could string together only a few relatively long oligos, not individual nucleotides, according to a template. Thatís a critical difference because itís not likely that enough of these oligos would have been floating around in the primordial ooze. ìThe system was really only giving us back a longer version of whatever we put in,î says Szostak. ìWe wanted to get it down to handling two- or three-nucleotide oligos because then we could put in all the random combinations and let it pick the right ones.î
But the enzyme didnít have anywhere near the precision or speed necessary to match and ligate the roughly 115 three-nucleotide oligos it would have needed to construct a copy of itself. Speed is important because oligos donít bond very strongly to the template; the enzyme has to do its work before the oligos can get away.
Szostak and Doudna had two options: they could reengineer the enzyme to handle a large number of oligos quickly, or they could find a shorter RNA enzyme that could be constructed from a smaller number of oligos. Szostak decided to take the second route.
ìWe had started with the Tetrahymena RNA because it was so well studied,î he explains. ìBut we had known from the beginning it might not work out, and we were prepared to switch to a different system.î Looking at 50 or so related RNA enzymes that had been identified since they had started their research, Szostak and Doudna settled on a strand of RNA called sun Y, found in a bacterial virus known as T4. Sun Y had one overriding grace: at about 200 nucleotides, it was the shortest in this class of RNA enzymes.
But though it was a bit over half the length of the Tetrahymena RNA and even better at ligating oligos, sun Y was still too long. It would never be able to string together reliably the nearly 70 three-nucleotide oligos it would need to replicate itself; that would be like presenting a child with a jigsaw puzzle of too many pieces and too little time to assemble them. To improve the situation, Szostak and Doudna relentlessly hacked away nucleotides from the enzyme to see how far they could shrink it--in effect, they were simplifying the jigsaw puzzle. But it was a delicate game of trade-offs: while a shorter sun Y would make an easier-to- replicate template, the same molecule when acting as an enzyme still needed all the loops and folds critical for embracing matched-up oligos and performing its ligating stunt. If the researchers hacked off too much, sun Y would become a simpler template but would no longer be an effective enzyme.
The two researchers eventually got down to a 160-nucleotide version of sun Y. At first it proved too weak as an enzyme, but they soon discovered they could restore its enzymatic strength by shuffling around a few of the nucleotides. Still, after all this inspired tweaking, the resulting enzyme remained much too large for its alternate role as a template. And this time the researchers were stumped.
ìEvery time we had come to a roadblock, Jennifer and I had sat and talked about it and come up with some strategy to get around it,î says Szostak. ìIt seemed amazing that we had gotten this far. But this was a severe problem, and we were very discouraged. We were forced to sit back and think about whether there was some completely different way to approach it.î
They needed a breakthrough--and it wasnít long in coming. ìI was sitting in my office staring at sun Y molecular-structure diagrams,î recalls Szostak, ìwhen all of a sudden I had this idea. What if we could split the enzyme into pieces that self-assembled?î
Szostak had little trouble finding places where he could break the molecule into three pieces of similar length. This novel strategy seemed to meet the teamís conflicting needs. In a solution, at any given time, some of the three sun Y subunits would be linked together, and some would be floating unattached. The RNA that was completely assembled would act as an enzyme; the unattached pieces would be short enough to be copied. In theory, the fully assembled molecules would join oligos into complementary strands alongside the shorter chunks acting as templates. Then the complementary strands would act as templates themselves, churning out replicas of the original templates with the help of the original enzyme. Eventually, a lot of brand-new sun Y subunits would be floating around, and some would self-assemble to become new enzymes.
ìI ran out of the office and told Jennifer,î says Szostak, ìand she immediately got excited about the idea and went to work on it.î Just two weeks later Szostak and Doudna were watching their new strategy work on a lab bench. ìIt was one of the most exciting moments of the project,î he says.
In a very rough sense, their three-piece RNA enzyme is indeed self-replicating. However, the researchers still need to feed it prefab oligos of about eight nucleotides each. Until the system makes accurate copies out of randomly clumped two- and three-nucleotide chunks, it canít be considered a plausible model for what nature did 4 billion years ago. Right now the enzyme is too slow and sloppy to stitch enough tiny oligos together: because it deviates from the template about one out of every three times it tacks on an oligo, the chances of its being able to get 20 right are minute. ìWe really need to focus on this question right now,î says Szostak. ìWe have to go from seventy percent accuracy to ninety-nine percent accuracy.î
Ideally, says Szostak, heíd like nature to do the necessary design work for him. ìIf the process was a little more efficient and accurate,î he explains, ìwe could let the enzyme replicate itself, errors and all. Some of the errors would lead to mutants that would replicate even better, and these would eventually dominate and lead to other mutants that did even better, and pretty soon evolution would take over.î
For now, though, he has turned to a hand-operated version of evolution. When Szostak synthesizes fresh batches of his sun Y RNA in a test tube, he purposely shuffles some of its nucleotides in the hope that a handful of the possible trillions of new combinations will turn out to be more efficient enzymes. To isolate these talented few, he has the entire batch ligate various molecules and then strains the mixture to pull out the longest strings. He picks out the mutants that produced those strings, synthesizes a fresh batch of the winners, and repeats the entire process. ìGetting to real evolution is still a distant goal for us,î he says, ìbut Iím pretty optimistic.î
In addition to his work on RNA, Szostak is experimenting with phospholipids, soapy substances that ball up in water to form microscopic globules. When he adds a phospholipid to a test tube full of his RNA molecules in solution, the emerging globules trap tiny drops of liquid containing RNA molecules. Although that doesnít even begin to approximate the complex way in which modern DNA directs the construction of its own cell membrane, Szostak thinks his globules can perform many of the functions of a real membrane. He has already found ways of getting them to ìgrowî by combining with other globules, as well as divide by squeezing through porous materials--both of which could have happened 4 billion years ago. ìA lot of people neglect the idea of compartmentalizing the RNA, but separating it from the rest of the world is crucial for evolution,î he says. ìYou want the RNA replicating its own interesting mistakes instead of wandering off in the solution.î
Some scientists arenít convinced that Szostak has found the recipe for creating life. ìHeíll stand right up there and tell you heís trying to build a cell, and I applaud his chutzpah,î says Norman Pace, a biochemist at Indiana University. ìBut unless he has some new gimmick Iím not familiar with, I donít think heís anywhere close to getting a self- replicating molecule. And even if he were, thereís no reason to believe it has anything to do with the origin of life. RNA is much too fragile to have survived the early conditions on the planet.î
But Gerald Joyce of Scripps is one of the many who weigh in on Szostakís side. ìJack has continued to amaze over the last several years,î he says. ìHeís not trying to recapitulate in detail something that happened on the primitive Earth; heís trying to capture the fundamental principles of how an RNA-based system could evolve. I tend to be hard-nosed about this sort of thing, but Iím optimistic about Jackís efforts.î Tom Cech, who kicked off the whole field of RNA enzymes, agrees: ìThat three-component system convinced me this whole thing is going to work. If they keep going at this rate, theyíll have RNA self-replication within two or three years.î
Szostak continues to plug away. Like many scientists these days, heís finding that his biggest challenge may be to keep his funding: Hoechst, the German chemical and pharmaceutical company that has given tens of millions of dollars to Mass Generalís biochemical research program, has cut back its support. ìIím going the traditional route now, which means writing lots of grant proposals,î he shrugs. If heís concerned, he doesnít show it. Maybe heís just thinking that if it came down to it, he could always scrounge up the money in small chunks, one modest grant at a time. After all, itís an approach thatís worked pretty well for him so far.