growth in live mouse cells [Nature, 3 7 2 , 642 (1994) ]. This was the first report of a DNA-binding protein engineered de novo to inhibit gene expression. The group of Carl 0. Pabo, professor of biophysics and structural biology and a Howard Hughes Medical Institute inves tigator at Massachusetts Institute of Technology, has developed phage-display methods to identify zinc finger pro teins that recognize specific sequences. Pabo and coworkers have also fused zinc finger motifs to other DNA-binding ele ments to create hybrid transcription fac tors with novel sequence specificities—a technology to which Ariad Gene Thera peutics has obtained an exclusive license. One such hybrid has been used success fully by Ariad researchers to bind to and activate synthetic therapeutic genes in troduced into animals. Ariad also is developing a technique in which a DNA-binding domain and an acti vation domain come together, in the pres ence of an orally available drug, to form a transcription factor. The idea is to eventu ally be able to use such transcription fac tors to turn synthetic genes on and off in patients for gene therapy applications. And Dervan and coworkers have de veloped cell-permeable polyamides that fold into hairpin loops and undergo se quence-specific binding to base pairs in the minor groove of DNA, where they can play a regulatory role. Binding of these polyamides modulates gene ex pression by blocking other proteins, such as transcription factors, that would other wise be able to bind to those sites. The researchers demonstrated three years ago that polyamides targeted at key se quences in the promoter region of a human immunodeficiency virus gene strongly inhibited viral replication [Proc. Nat Acad. Sci USA, 9 5 , 12890 (1998)]. In unpublished work, Dervan's group and that of professor Mark S. Ptashne, head of the Gene Regulation Laboratory at Sloan-Kettering Institute, New York City, have created artificial transcription factors in which a small peptide activation domain is combined with a DNA-binding polyamide. The col laborative team finds that these synthet ic transcription factors are capable of ac tivating gene transcription in vitro. Der van says he is particularly excited about these findings since previous polyamide constructs have been able to repress gene expression but not to activate it. Dervan's synthetic polyamides can zero in on most any DNA sequence. The Barbas group's designed zinc finger pro
teins can't do that yet They bind up to 18unit chunks of repeating GNN triplet se quences, where G is guanine and Ν is any of the other three types of DNA bases. Barbas and coworkers have reported the development of 16 distinct types of zinc fin ger domains, but 64 such domains would be required to bind any sequence at will. Nevertheless, Barbas says, "it turns out that stretches of sequence that re peat the GNN motif six times are actual ly quite common in genes, and in every gene we've looked at we can find at least one of those sites. And we don't have to place the protein at very specific sites within the gene to regulate it. So we should be able to control most genes with our 16 zinc finger domains, and we believe that many of the 64 codonrecognition domains will become acces sible in the next year or two." Barbas notes that in unpublished studies his group has prepared many of the addi tional domains required for the recogni tion of any given sequence. Choo, Klug, and coworkers have al ready developed a new zinc finger dis play strategy that makes it possible to target more or less any 18-base-pair DNA sequence—not just repeating
GNN triplets. A scientific paper on the new strategy "is already written up and ready to go off," Choo notes, but some of the information has already appeared in patent applications filed by Gendaq Ltd., London, a company founded last year by Choo and Hug. At Gendaq, "we aim to apply customized transcription factors for gene regulation in functional genomics, agricultural biotechnology, and human therapy," Choo says. With regard to potential applications of designed transcription factors, Barbas points out that gene therapy trials have been carried out in which patients with heart ailments have been injected with vi ruses containing VEGF genes. An alter native approach, he says, "would be sim ply to activate the endogenous VEGF gene found in every cell with a designed transcription factor." Furthermore, "you could imagine introducing a zinc finger protein that activates insulin and is under the control of a nontoxic, orally available pill," such as an aspirin-like derivative. "We have encoded in our own genes the solutions to many diseases," Barbas says. "What awaits is just a way to switch on those critical genes." Stu Borman
Enzyme's Activity )esigned To Order
R
esearchers in England have modi fied a natural enzyme into a de signed enzyme with an entirely new catalytic function. They accomplished this by cutting two sections out of the gene for the natural enzyme, inserting replace ments containing mutations, shuffling the mutations, and screening expressed li braries of modified proteins for the de sired activity—a process called directed evolution [Nature, 4 0 3 , 617 (2000)]. Alan R. Fersht and coworkers at Cam bridge Centre for Protein Engineering and Cambridge University Chemical Lab oratory created the designed enzyme by substituting modified catalytic or binding units into an enzyme with an α/β-barrel structure—a common type of protein structural foundation or "scaffold." In the study, Fersht and coworkers aimed to convert the activity of indole-3glycerol phosphate synthase (IGPS) into that of phosphoribosylanthranilate isomerase (PRAI). The deck was loaded somewhat in that the two enzymes cata lyze sequential in vivo reactions in the tryptophan biosynthetic pathway and PRAI's catalytic product is IGPS's sub-
strate. Nevertheless, the activities of the two enzymes are quite different, so it was no easy feat to get IGPS to mimic PRAI's catalytic function. To do so, Fersht and coworkers snipped substrate-binding and catalytic loops out of the IGPS gene, inserted mod ified sequences in the gaps, shuffled the mutations, and screened the expressed enzyme variants for PRAI-like activity by testing them in PRAI-deficient bacteria that need tryptophan to grow. In the end, the researchers identified a modified IGPS with an activity and catalytic effi ciency strikingly similar to those of PRAI. The work has potential applicability to the creation of novel biocatalysts, and it advances scientists' understanding of the type of natural evolutionary mechanisms that enzymes might use to develop new functions and adapt to changing environ ments. The findings also suggest that the α/β-barrel scaffold—which may be found in as many as 10% of all soluble enzymes—could be generally useful for creating biocatalysts with a range of novel activities. "Alan was trying to demonstrate that FEBRUARY 21, 2000 C&EN
35
science/technology
SHi^
N-terminus β2α2
β1α1
β6α6
β8α8
β7α7
Fersht and coworkers "evolved" IGPS (left) into an enzyme with an activity similar to that of PRAI (right). IGPS struc
this α/β barrel is, in fact, highly evolvable, and that you could conceivably, with a minimal amount of work, graft and evolve new catalytic activities into that ba sic framework, just as nature has done" in the course of α/β-barrel enzyme evolu tion, says chemical engineering and bio chemistry professor Frances H. Arnold of California Institute of Technology. "His group has shown that to be the case. I'm truly impressed by what they did because I know how hard it is." Arnold notes that "it's been a Holy Grail-type problem to create new pro tein functions at will. That's why people were so excited about catalytic antibod ies. The idea that you could create a cat alyst from a noncatalytic molecule just by imprinting the transition site was so attractive. The enzymologists have been green with envy over this, because we know that enzymes are much better cat alysts, but we've had no good mecha nism for creating new ones." It is clear "that evolution is a very pow erful design algorithm—much more powerful, in my opinion, than any kind of rational design approach," but directed evolution by itself hasn't been very suc cessful in solving this kind of problem, she says. 'What Fersht has done is taken the strongest features from both rational design and directed evolution" to create a new catalytic function. The work is reminiscent of some pre vious studies. Arnold points to two nota ble examples: Organic chemistry pro fessor Donald Hilvert (now at the Swiss Federal Institute of Technology, Zur ich) and coworkers at Scripps Research Institute, La Jolla, Calif., used structure36
FEBRUARY 21,2000 C&EN
tures modified in the transformation process are labeled and color-highlighted at center.
based design and directed evolution to change a dimeric enzyme into a mono melic protein with similar activity [Sci ence, 2 7 9 , 1958 (1998)]; and Eric Quéméneur and coworkers at the French Atomic Energy Commission, Saclay, engineered a new catalytic activity into the enzyme cyclophilin by mutating three amino acids to form an active site similar to that of a serine protease [Nature, 391,301 (1998)]. But the Fersht group's study "is the first time, as far as I know, where somebody has taken as good a catalytic design as rational design will give you—which is usually pretty poor—and coupled that with directed evolution to create a really good enzyme," Arnold says. Assistant professor of biopharmaceutical sciences and pharmaceutical chemistry Patricia C. Babbitt of the University of California, San Francisco, calls the findings of Fersht and coworkers "an absolutely stunning result." Most of the enzyme engineering studies carried out heretofore have resulted in "fairly minor changes in enzyme properties—changes in substrate specificity, tweaking the efficiency of a protein, and that sort of thing," Babbitt says, whereas Fersht and coworkers have achieved a wholesale change in the type of reaction catalyzed by an enzyme. For some time, a number of research groups—including Babbitt's—have been looking at the way nature "evolves" enzymes to carry out new functions. 'There have been some hypotheses out there for a long time that the α/β barrel architec ture is set up to change its chemistry in a pretty simple and discrete way" because
substrate binding and catalysis occur in that system at sites sufficiently separable to be dealt with individually, Babbitt says. 'That gives nature an architectural strat egy it can use to evolve new functions. Fersht's group took advantage of that to answer a question that nature never asked, as far as we can tell. It's stunning evidence that the hypothesis about the α/β scaffold is on the right track." For researchers like Babbitt who are trying to understand "how nature does biology," she says, "this gives us some new clues about what's possible for this scaffold. It tells us that it can do difficult new chemistries, and it gives us a feel for how straightforward a set of changes nature can make to get those new chem istries. Biology has to do that—when an organism moves into a new ecological niche, for example." To recruit a scaf fold that's already there to do some thing new "is very important from a bio technology standpoint—in terms of the ability to develop new enzymes—but also from the basic science standpoint of understanding how enzymes evolve," Babbitt says. Fersht notes that "from a personal point of view, I was in the first wave of protein engineers and ended my first re view of the subject in 1984, in Angewandte Chemie, with this sentence: The ulti mate goal is to design a tailor-made en zyme for every reaction.' Some of my colleagues thought I was rather overoptimistic. So this result is of great emotional significance for me since, after waiting 15 years, I can see light at the end of the tunnel." Stu Borman
