KEYSTONE MILLENNIUM MEETING: CONFERENCE REPORT September 2000 © 2000 Elsevier Science Ltd. All rights reserved26

Genes and disease

“The science is

young, the

hurdles are

many, and the

expectations are

too high.”

Inder Verma

Cells infected with a gene-carrying retrovirus maychurn out lots of the therapeutic protein in cell culture, butprotein levels seem to drop once they are returned to thebody. “That has been a bit of a handicap,” says Verma.

Verma’s group has also tried adenovirus vectors. Theyengineered a virus to express factor IX, a protein essentialfor blood clotting, and injected it directly into the musclesof mice and dogs. Although the animals briefly had highlevels of factor IX, immune cells started attacking both theviral and transgenic products.

The body’s immune system poses a serious problemfor gene therapy. “For me, the biggest question is ‘What arethe immunological challenges?’” says Verma.The Food andDrug Administration halted one gene therapy trial in cysticfibrosis patients several years ago when an adenovirusprovoked too much inflammation. Gelsinger’s death alsoseems to have involved a severe immune reaction.

So one thing gene therapy desperately needs now,Verma says, is a better vehicle for putting genes into cells.An ideal vector would have little or no immune problems,would infect both dividing and non-dividing cells, andwould be easy to generate in large quantities in the lab.One potential answer, which Verma first proposed in 1996,shocked many of his colleagues. “Everyone said ‘Oh myGod,AIDS virus, are you nuts?’” remembers Verma. But hemaintains that a modified, pared-down form of HIV couldfulfill all three of his criteria.

Already,Verma has shown that genetically modified HIVcan deliver a therapeutic gene to haematopoietic stem cells,which generate the blood and immune system.These cellsare normally hard to infect because they divide so infre-quently. But Verma’s HIV vector had no trouble transfer-ring the gene for green fluorescent protein into culturedhuman haematopoietic stem cells. For more than 24 weeks,5–15% of the cells continued to generate the foreign protein(Science, vol. 283, p. 682).

Another promising new way to get genes into cells isthe adeno-associated virus, or AAV. It causes no knownhuman disease, and genes transferred to cultured cells thisway are expressed for a long time. After successfully halt-ing bleeding disorders in mice with an AAV carrying thefactor IX gene, researchers have begun clinical trials. So farthe three patients with haemophilia B who received thevirus have shown no ill effects, and two have been able toreduce their injections of factor IX (Nature Genetics, vol. 24,p. 257). The only apparent drawback to AAV is that it isdifficult to produce in large quantities.

Verma thinks that the first gene therapy procedures toemerge successfully from clinical trials to become acceptedpractice will involve inherited, single-gene disorders likehaemophilia. In recent years, compelling accounts of genetherapy for acquired diseases have appeared in the press. Forexample, Jeffrey Isner at the St Elizabeth’s Medical Center inBoston has used gene therapy to grow new blood vessels inpeople with angina who were not eligible for heart bypasssurgery. But such stories are not clear-cut, Verma says. “I

consider those anecdotal because the clinical end points aren’treally defined,” he says. “There are to me more measurableend points than saying the patients felt better.”With a diseaselike haemophilia, researchers can measure precise quantitiesof gene products to see how well the gene is working.

Ultimately, people will be able to control the genes thatare added to their bodies,Verma says, although that technol-ogy is a long way off. Currently researchers can’t controlthe expression of a gene once they get it into cells, so theyare experimenting with ways to make the gene respond todrugs like tetracycline. In theory, a person could have genetherapy to treat something like diabetes and then take apill to turn the insulin gene on only when necessary.“There is substantial interest in that,”Verma says.

Verma recognizes that one of the first steps in that direc-tion must be to restore public confidence in the technology,in part through education. He says the public knows verylittle about the progress that’s been made because theresults from human studies have been so unimpressive. “Wewent too fast into the clinical trials.The enthusiasm hasn’tbeen borne out,” he says. “The science is young, the hur-dles are many, and the expectations are too high.” But heremains confident in its future. “Rarely comes a technol-ogy which can have such a vast impact on public health.”

Nell Boyce

Once upon a timeHow did life begin? One way to find out is to make it happenagain in the laboratory. Scientists have yet to generate lifefrom scratch. The best they have managed is to simulatethe Earth’s primordial soup with educated guesses at whatchemicals must have been present four billion years ago;they have then coaxed from it some basic biological build-ing blocks such as amino acids. But this is just child’s play.In the coming century, Harvard geneticist Jack Szostak

predicts that researchers will reproduce in the lab likelycandidates for the Earth’s original self-replicating molecules.They may even recreate the first cells.

Anything that doesn’t duplicate itself can hardly be consid-ered alive, so the very first life forms must have been able toreproduce.The system by which modern cells grow and repli-cate is far too complex to have emerged directly from theprimordial soup. It requires at a minimum the coordinatedaction of three types of molecule.The first is DNA, whichholds the entire genetic blueprint of an organism in every cell.Then RNA carries the information as it is needed from thenucleus to the cytoplasm.There it directs the manufacture ofthe third type of molecule: proteins, the workhorses of thecell. This three-stage process provides the framework inwhich mutation and natural selection take place. The firstlife forms must have been simpler, and Szostak believes wemay soon discover how life actually emerged.

So how might early life forms have transferred infor-mation from one generation to the next? RNA seems a likely

KEYSTONE MILLENNIUM MEETING: CONFERENCE REPORT September 2000 © 2000 Elsevier Science Ltd. All rights reserved 27

Genes and disease

“The only way to

really see what

RNA can do is to

make those

molecules

themselves.”

Jack Szostak

candidate for this originalgenetic material. Not only canit be copied into new RNA,just as DNA can be copied intoDNA, but it may also at onetime have been able to copyitself. A number of specializedforms of RNA, called ribo-zymes, have been discoveredin the past decade that actuallybreak chemical bonds and cata-lyze reactions. RNA, it seems,can do the work of both DNAand proteins. If this is so, andRNA was the original moleculeof life, the ribozymes that sur-vive today may be relics of thisancient RNA world.

Scientists are, however, di-vided over whether RNA couldever have catalyzed all the bio-chemical reactions necessaryfor life.That’s because RNA has many fewer possibilities asan enzyme than proteins have. RNA has only four buildingblocks, whereas proteins have 20, so the number of dif-ferent possible RNA molecules of any given length ismuch smaller. Also, the building blocks themselves have amuch smaller variety of chemically active groups thanproteins, limiting the types of reactions RNA could catalyze.“The only way to really see what RNA can do is to makethose molecules themselves,” says Szostak.

So Szostak’s team has made huge pools of random syn-thetic RNA sequences – around 1015 different combinations– and exposed them to Darwinian selection pressures, in thehope of creating new ribozymes. “We started off evolvingfairly simple functions, evolving RNA to recognize and bindto other targets,” he says.To drive this evolution, the research-ers extracted from a random pool of RNAs only thosecapable of binding to a specific target.They then amplifiedthe number of copies of the extracted RNAs in a way thatallowed occasional sequence mistakes to be made, so someof the new copies were mutated. Then the selection wasapplied again until an RNA emerged that could efficientlybind to its chemical target. “Surprisingly, there’s often some-thing in the collection that actually works,” Szostak says.

Once Szostak was satisfied he could select RNA that couldbind to targets, the researchers began trying to evolve RNAsthat could catalyze reactions.The first ribozymes they gener-ated this way catalyzed the sort of phosphate chemistry usedby the modern enzymes that copy DNA. Now Szostak’s groupand other laboratories are trying to make RNAs that accelerateother reactions, such as those involved in protein synthesis.

“Our work shows that RNA can catalyze a wide rangeof reactions,” Szostak says. And if it can happen in a lab, itcould almost certainly have happened somewhere on Earthfour billion years ago, he says. So it is at least plausible that

ribozymes could have beenresponsible for the replicationfunctions and metabolic re-actions of the first cells.Assum-ing this is what happened,somewhere along the line, pro-teins must have joined in.Whatwere the first proteins like?Did ribozymes manufacturethem? And if so, what types ofprotein were the earliest RNAscapable of producing? To tacklethe problem, Szostak has devel-oped a way to drive proteinevolution in the test tube.

One way of evolving pro-teins would be simply to trans-late a random pool of thou-sands of RNAs into proteins,and select the ‘best’ protein fromthe mix – the one that propelleda chemical reaction the fastest,

say.The problem is that proteins, unlike RNA, do not retainthe genetic information that gave rise to them. Furthermore,determining the order of amino acids in a protein is tech-nically much more cumbersome than reading the order ofbases in an RNA. So once the optimum protein was in hand,it would be very hard to identify the gene that encoded it.

Szostak and his colleague Richard Roberts jumped thishurdle three years ago by engineering a way for each proteinmade in a test tube to carry its own coding sequence aroundwith it.Normally RNA and protein go their separate ways afterthe ribosome has finished translating the RNA code into afinished protein. Szostak and Roberts put a special linker onthe end of their RNAs that got pulled into the ribosome andattached to the tail end of the new protein.The proteins madethis way still functioned normally, and could be tested fortheir performance. But after the most efficient was selected,its genetic code could be read directly off its RNA tag.

Szostak’s team and others are now thinking about the nextstep in the laboratory simulation of primordial life: artificialcells. In the next few years, Szostak expects to see the devel-opment of very simple cells in test tubes.These might carryan RNA genome replicated by an RNA ribozyme, and shouldbe capable of performing simple metabolic functions relatedto building membrane material and producing the ribo-nucleotide bases of which RNA is made.“Basically we’re talk-ing about simple systems with the bare minimum of com-ponent parts,” Szostak says.“Presumably if we can do this, wecan gradually add more pieces to make the cell work better.”

Will such experiments really be able to recreate thefirst living things on Earth? “They can’t really tell us whatactually happened,” Szostak says, “but at least they can tellus what is possible.”

Matt Walker

Further reading

Landerweb.mit.edu/biology/www/Ar/99lander.html

Vermawww.salk.edu/faculty/verma.html

Szostakxanadu.mgh.harvard.edu/szostakweb/web2

Jack Szostak Trying to recreate the first life forms