The Unusual Origin of the Polymerase Chain Reaction

Home , Chain reaction, Dideoxynucleotide, Kary Mullis

The Unusual Origin of the Polymerase Chain Reaction A surprisingly simple method for making unlimited copies of DNA fragments was conceived under unlikely circumstances-during a moonlit drive through the mountains of California

by Kary B. Mullis

ometimes a good idea comes tal tissue specimen, from a single hu- it at random points along its length. to you when you are not looking man hair, from a drop of dried blood Consequently, if the DNAis removed S for it. Through an improbable at the scene of a crime, from the tis- from 1,000 identical cells, there will combination of coincidences, naivete sues of a mummified brain or from a be 1,000 copies of any given gene, but and lucky mistakes, such a revelation 40,000-year-old woolly mammoth fro- each copy will be on a DNAfragment came to me one Friday night in April, zen in a glacier. of differing length. 1983, as I gripped the steering wheel In the seven years since that night, For years this problem made it diffi- of my car and snaked along a moonlit applications for the PCRhave spread cult to study genes. Then in the 1970's mountain road into northern Califor- throughout the biological sciences: enzymes known as restriction endo- nia's redwood country. That was how I more than 1,000 reports of its use nucleases were discovered: these en- stumbled across a process that could have been published. Giventhe impact zymes snipped strands of DNAat spe- make unlimited numbers of copies of of the PCRon biological research and cific points. The endonucleases made genes, a process now known as the its conceptual simplicity, the fact that it possible to cut DNA into smaller, polymerase chain reaction (PCR). it lay unrecognized for more than 15 sturdier, more identifiable pieces and Beginningwith a single molecule of years after all the elements for its thereby made it easier to isolate the the genetic material DNA,the PCRcan implementation were available strikes pieces containing a gene of interest. generate 100 billion similar molecules many observers as uncanny. By the late 1970's, therefore, mo- in an afternoon. The reaction is easy to lecular biologists were busily study- execute: it requires no more than a he polymerase chain reaction ing DNAwith endonucleases and with test tube, a few simple reagents and a makes life much easier for mo- other molecules called oligonucleotide source of heat. The DNAsample that Tlecular biologists: it gives them probes. An oligonucleotide is a short one wishes to copy can be pure, or it as much of a particular DNAas they chain of specifically ordered nucle- can be a minute part of an extremely want. Casual discussions of DNAmol- otide bases. Under the right condi- complex mixture of biological materi- ecules sometimes make them sound tions, an oligonucleotide willbind spe- als. The DNAmay come from a hospi- like easily obtained objects. The truth cifically with a complementary -se- is that in practice it is difficult to get a quence of nucleotides in single-strand well-defined molecule of natural DNA DNA.Therefore, radioactively labeled, KARYB.MUlliSdescribes himself as from any organism except extremely man-made oligonucleotides can serve "a generalist with a chemicalprejudice." simple viruses. as probes for determining whether a In addition to the polymerase chain reaction, he is also known for having in- The difficulty resides in the nature sample of DNAcontains a specific nu- vented a plastic that changes color rap- of the molecule. DNA is a delicate cleotide sequence or gene. In 1979 the idly when exposed to ultraviolet light. chain made of four deoxynucleotides: Cetus Corporation in Emeryville,Calif., Whileworking as a biochemistry grad- deoxyadenylate (A),deoxythymidylate hired me to synthesize oligonucleo- uate student at the University of Cali- (T), deoxyguanylate (G) and deoxycy- tide probes. fornia, Berkeley,he published a paper tidylate (C);the sequence of these bas- By 1983 the charm of synthesizing in Nature entitled "The Cosmological es encodes the genetic information. oligonucleotides for a living had en- Significance of Time Reversal." Mullis tered a decline-a decline that most received his Ph.D. in biochemistry in Rarelydoes one find a single strand of r 1972. After working as a postdoctoral DNA;. usually pairs of strands with of us so employed were happy to wit- fellowat the Universityof Kansas Medi- complementary sequences form dou- ness. The laborious but very quaint cal Schooland the Universityof Califor- ble helixes in which the A's in one chemical art form for making oligonu- I nia, San Francisco,Mullisjoined the Ce- strand bind with the T's in.the other, cleotides manually, to which we had ~ tus Corporation, where he discovered and the G's bind with the C's [see grown comfortably numb, had given , I the polymerase chain reaction. In 1986 illustration on opposite page]. Inside way to a much less charming but reli- he became the director of molecular a cell this DNA helix is surrounded able automated technique. It was an biology at Xytronyx, Inc. Today Mullis I and further coiled by various proteins. immense improvement. works in Lajolla, Calif.,as a private con- In the aftermath of this minor sultant on polymerase-chain-reaction When biologists try to isolate a naked technology and nucleic acid chemistry. DNAchain, the DNAis so long and thin industrial revolution, we nucleotide that even mild shearing forces break chemists found ourselves success-

56 SCIENTIFICAMERICAN April 1990 fully underemployed. Laboratory ma- facts about DNA.A strand of the mole- if the template nucleotide is a G, the Q chines, which we loaded and watched, cule has one end that is known, by enzyme attaches a C.Byrepeating this were making almost more oligonucle- chemical convention, as three-prime process, the polymerase can extend otides than we had room for in the and one end that is five-prime. In a the primer's three-prime end all the freezer and certainly more than the double helix of DNA,the complemen- way to the template's five-prime ter- molecular biologists-who seemed to tary strands are said to be antiparallel, minus [see illustration on page 59]. In be working even more slowlyand tedi- because the three-prime end of one a double helix of DNA, each strand ously than we had previously suspect- strand pairs with the five-prime of the serves as a template for the other II ed-could use in their experiments. other strand, and vice versa. during replication and repair.

Consequently, in my laboratory at Ce- In 1955Arthur Kornberg of Stanford Now for dideoxy sequencing, which I tus, there was a fair amount of time University and his associates discov- is also commonly called the Sanger i available to think and to putter. ered a cellular enzyme called a DNA technique after one of its inventors, I found myself puttering around polymerase. DNA polymerases serve Frederick Sanger of the British Medi- II Ii, with oligonucleotides. several natural functions, including cal Research CouncilLaboratory of Mo- :1 the repair and replication of DNA. lecular Biology.This technique uses knew that a technique for easily These enzymes can lengthen a short a DNApolymerase, template strands, determining the identity of the nu- oligonucleotide "primer" by attaching primers, nucleotide triphosphates and Icleotide at a given position in a an additional nucleotide to its three- special dideoxynucleotide triphos- DNAmolecule would be useful, espe- prime end, but only if the primer is phates (ddNTP's) to determine DNA cially if it would work when the com- hybridized, or bound, to a comple- sequences. like ordinary nucleotides, plexity of the DNAwas high (as it is in mentary strand. called the template. ddNTP's can be attached to growing human DNA)and when the available The surrounding solution must also primers by polymerases; however, a quantity of the DNAwas small. I did contain nucleotide triphosphate mol- ddNTPwill "cap" the three-prime end not see why one could not use the ecules as building blocks. of a primer and prevent the addition enzyme DNApolymerase and a varia- The nucleotide that the polymerase of any more bases. The Sanger tech- tion of a technique called dideoxy se- attaches willbe complementary to the nique produces primers that have quencing, and so I designed a simple- base in the corresponding position on been lengthened to varying extents minded experiment to test the idea. the template strand. For example, if and then capped by a ddNTP.By ar- To understand the approach I had the adjacent template nucleotide is an ranging these fragments according to in mind, it is worth reviewing certain A, the polymerase attaches a T base; length and by knowing which ddNTP's --~-- 'II'

~ -"".. 110 II

DNAconsists of two strands of linked nucleotides: deoxyaden- are always opposite T's, and the G's are opposite C's-and this ylates (A's), deoxythymidylates (T's), deoxyguanylates (G's) complementarity binds the strands together. Each strand has a and deoxycytidylates (C's). The sequence ofnucleotides in one three-prime and a five-prime end Because their orientations strand is complementary to that in the other strand-the A's oppose one another, the strands are said to be antiparallel.

SCIENTIFICAMERICAN April 1990 57

... have been added, an investigator can the presence of a C in the template. would therefore terminate immediate- determine the sequence of bases in In the modified version of this ly after the addition of one base from the template strand. For example, if technique that I was contemplating, a ddNTP to the chain. If I knew which a dideoxyadenine (ddA) base were I would use" only polymerases, tem- ddNTP had been added to the primers, added at a given position, the corre- plates, ddNTP's and primer mole- I would also know the identity of the sponding complementary base in the cules-that is, I would omit the ordi- corresponding base in the template template would be a T; the addition nary nucleotide triphosphates from strand. In this way, I could deduce the of a dideoxyguanine (ddG) implies the mixture. Extension of the primers identity of a base in the template

00 32 COPIES 64 COPIES----7 16 COPIES-- 8 COPIES / -- HOPIES -- -/ -- -\ / -- -- 2COPlES_/-, -- / -- -- ./ -- "\ / -- 1 COpy I -- -. -- / -- ./ -- -- -\ / -- -Ie -- -, -- / -- -- ./ -- -\ / -- --

POLYMERASE CHAIN REACTION is a simple technique for copy- agents. Because the number of copies increases exponential- ing a piece of DNA in the laboratory with readily available re- ly, more than 100 billion can be made in only a few hours.

58 SCIENTIFICAMERICAN April 1990 ~ ~"""

strand adjacent to the site where the 5' 3' primer binds. hat I did not realize at the time was that there were W many good reasons why 3' 5' my sequencing idea could not work. ~ The problem was that oligonucleotides sometimes hybridize with DNA POLYMERASE sequences other than those intend- ed; these unavoidable pairings would 0 have made my results ambiguous. Even in the hands of those skilled in the art of careful hybridization, it was impossible to bind oligonucleotides 5' T 3' to whole human DNAwith sufficient specificity to get anything even ap- proaching a meaningful result. It was because of this limitation that researchers had resorted to more dif- ~ 3' 5' ficult procedures for looking at human DNA.For instance, restriction enzymes could be employed to cleave POLYMERASE the DNA sample into various fragments that could be separated from each other by electrophoresis; in this cb way,the sample could be "purified," to some extent; of all DNA except the target fragment before the hybridiza- 5"" 3' tion of oligonucleotide probes. This approach reduced erroneous hybridi- zations sufficiently to provide meaningful data, but just barely. Moreover, this procedure was lengthy and would ~ 3' 5' not work on degraded or denatured samples of DNA. Another technique that was much POLYMERASE too lengthy for routine DNAanalysis db involved cloning. A human DNA sequence of interest could be cloned, or copied, into a small ring of DNAcalled a plasmid. Copies of this plasmid and the targeted sequence could then be 5' . 3' produced in bacteria, and sequence information could be obtained by oligonucleotide hybridization and dideoxy sequencing. In the early 1980's dideoxy sequencing of cloned DNA 3' . 5' was the method by which most human ~ DNAsequence information had been DNAPOLYMERASE,an enzyme, can lengthen a short strand of DNA, called an oli- obtained. gonucleotide primer, if the strand is bound to a longer "template" strand of DNA. In proposing my simple-minded ex- The polymerase does this by adding the appropriate complementary nucleotide to periment, I was implicitly assuming the three-prime end of the bound primer. If a dideoxynucleotide triphosphate that no such cloning or other step (ddNTP) such as dideoxyadenine (ddA) is added, however, no further extension is would be necessary to detect specif- possible, because the three-prime end of the ddA will not link to other nucleotides. ic human DNAsequences by a single oligonucleotide hybridization. In to- ken defense of my misguided putter- were somewhere near the cutting edge free. On that particular night I was ing, I can point out that a group down of DNAtechnology. thinking about my proposed DNA-se- the hall led by Henry A. Erlich, one of quencing experiment. Cetus's senior scientists, was trying ne Friday evening late in the Myplans were straightforward. First another method based on the hybridi- spring I was driving to Men- I would separate a DNA target into zation of a single oligonucleotide to a Odocino County with a chemist single strands by heating it. Then I human DNAtarget. No one laughed friend. She was asleep. U.S. 101 was would hybridize an oligonucleotide to out loud at Henry, and we were all undemanding. I liked night driving; a complementary sequence on one of being paid regularly. In fact, we were every weekend I went north to my the strands. I would place portions of being paid enough to lead some of us cabin and sat still for three hours in this DNAmixture into four different to assume, perhaps brashly, that we the car, my hands occupied, my mind tubes. Each tube would contain all

SCIENTIFICAMERICAN April 1990 59 four types of ddNTP's, but in each and winds upward through the coast- Although I did not realize it at that tube a different type of ddNTPwould al range, I decided the determination moment, with the two oligonucleo- be radioactively labeled. Next I would would be more definitive if, instead of tides poised in my mind, their three- add DNApolymerase, which would ex- just one oligonucleotide, I used two. prime ends pointing at each other on tend the hybridized oligonucleotides The two primers would bracket the opposite strands of the gene target, I in each tube by a single ddNTP. By targeted base pair I hoped to identify. was on the edge of discovering the electrophoresis I could separate the Bymaking the oligonucleotides of dif- polymerase chain reaction. Yet what extended oligonucleotides from the ferent sizes, I would be able to distin- I most felt on the edge of was the residual ddNTP's;by identifying which guish them from each other. Bydirect- mountain road. radioactively labeled ddNTPhad been ing one oligonucleotide to each strand incorporated into the oligonucleotide, of the sample DNAtarget, I could get hat night the air was saturated I could determine the corresponding complementary sequencing informa- with moisture and the scent of complementary base in the target tion about both strands. The experi- Tflowering buckeye. The reckless strand. Simple. ment would thereby contain an inter- white stalks poked from the roadside Around Cloverdale,where California nal control at no extra inconvenience into the glare of my headlights. I was 128 branches northwest from US.IOl [seeillustration below]. thinking about the new ponds that

TARGETED BASE PAIR

5' ~~ 3'

3'~ 5'

I MELT APART STRANDS,

t ATTACH PRIMERS 5' 3'

3' 5' 5'~ 3'

3'~ 5'

I ADD POLYMERASE,

t LABELEDddNTP's 5' 3'

5' 5' l

5' IDENTITYOFADDEDddNTP's REVEALS BASESIN TARGETEDPAIR

TO DETERMINETHE IDENTITYof a targeted base pair in a which would allow each of the primers to be extended by only piece of DNA,the author hoped to apply a variation on a tech- one base. The identity of the added ddNTP bases would re- nique called dideoxy sequencing. First two primers would be veal what the complementary targeted bases were. The tech- bound to the opposing strands in the DNAat sites flanking nique could work with only one primer, but the use of two the targeted pair. DNApolymerase and dideoxynucleotide tri- would provide a control for checking the results. Planning this phosphates (ddNTP's) would then be added to the mixture, experiment led the author to the polymerase chain reaction.

60 SCIENTIFICAMERICAN April 1990 I was digging on my property, while tended oligonucleotides from the DNA I realized something else about the also hypothesizing about things that targets. True, the extended oligonucle- products of the reaction. After a few might go wrong with my base-se- otides would still be in the sample; but rounds of extending the primers, dis- quencing experiment. because there would be far more un- sociating the extension products, rehy- From my postdoctoral days in Wolf- extended primers than extended ones bridizing new primers and extending gang Sadee's laboratory at the Uni- in the mixture, the DNAtargets would them, the length of the exponential- versity of California at San Francis- probably hybridize with unextended ly accumulating DNA strands would co, where John Maybaum was devis- primers when the mixture cooled. I be fixed because their ends would be ing clinical assays for nucleotides, I could then add ddNTP's and more sharply defined by the five-prime ends remembered that my DNA samples polymerase to perform my sequenc- of the oligonucleotide primers. I could might contain stray traces of nucleo- ing experiment. replicate larger fragments of the orig- tide triphosphates. It would compli- Yet some questions still nagged at inal DNAsample by designing prim- cate the interpretation of the gel, I me. Would the oligonucleotides ex- ers that hybridized farther apart on it. figured, if stray nucleotides intro- tended by the mock reaction inter- The fragments would always be dis- duced with the sample added them- fere with the subsequent reactions? crete entities of a specified length. selves to the three-prime end of the What if they had been extended by I stopped the car again and started primers before the planned addition many bases, instead of just one or drawing lines of DNAmolecules hy- of the labeled ddNTP's. two? What if they had been extended bridizing and extending, the products One thought I had was to destroy enough to create a sequence that in- of one cycle becoming the templates any loose nucleotide triphosphates cluded a binding site for the other for the next in a chain reaction.... in the sample with alkaline phospha- primer molecule? Surely that would Jennifer protested again from the edge tase, a bacterial enzyme. This enzyme cause trouble. . . . of sleep. "You're not going to believe would chew the reactive phosphate No,far from it! Iwas suddenly jolted this," I crowed. "It's incredible." groups off any nucleotide triphos- by a realization: the strands of DNAin She refused to wake up. I proceed- phates, thereby rendering them inert the target and the extended oligonu- ed to the cabin without further stops: to a polymerase reaction. Yet I would cleotides would have the same base The deep end of Anderson Valley is then somehow have to eliminate the sequences. In effect,the mock reaction where the redwoods start and where phosphatase from the sample, or else would have doubled the number of the "ne'er-do-wells" have always lived. it would also destroy the ddNTP's DNAtargets in the sample! Mydiscovery made me feel as though . when I added them. Normally one Suddenly, for me, the fragrance of I was about to break out of that old can deactivate unwanted enzymes by the flowering buckeye dropped off valley tradition. It was difficult for me heating them and altering their essen- exponentially. to sleep that night with deoxyribonu- tial shape; Ibelieved, however,bacteri- clear bombs exploding in my brain. al alkaline phosphatase could refold nder other circumstances, I itself into its original form. I therefore might not have recognized the et in the morning I was too rejected alkaline phosphatase as an Uimportance of this duplication tired not to believe that some- answer to the problem. so quickly. Indeed, the idea of re- Yone, somewhere, must have I was, in fact, mistaken. Much later I peating a procedure over and over tried this idea already. Thousands of learned that alkaline phosphatase can again might have seemed unaccept- investigators had, for various reasons, be irreversibly denatured by heating ably dreary. I had been spending a lot extended single oligonucleotides with if no zinc is present in the solution. of time writing computer programs, polymerases; surely someone would As it turned out, my mistake was ex- however, and had become familiar have noticed the possibility of a poly- traordinarily fortunate: had I known with reiterative loops-procedures in merase chain reaction. But if it had better, Iwould have stopped searching which a mathematical operation is re- worked, I was sure I would have heard for alternatives. peatedly applied to the products of about it: people would have been us- Every mile or so another potential earlier iterations. That experience had ing it all the time to amplify, or multi- solution arose but fell short. Then, as I taught me how powerful reiterative ply, DNA fragments. began the descent into Anderson Val- exponential growth processes are. The Back at Cetus on Monday I asked one ley, I hit on an idea that appealed to DNAreplication procedure Ihad imag- of the librarians, George McGregor, to my sense of aesthetics and economy: I ined would be just such a process. run a literature search on DNA poly- would apply the same enzyme, DNA Excited,I started running powers of merase. Nothing relevant to amplifica- polymerase, twice-first to eliminate two in my head: two, four, eight, 16, tion turned up. For the next few weeks the extraneous nucleotide triphos- 32 I remembered vaguely that two I described the idea to anyone who phates from the sample, then to incor- to the tenth power was about 1,000 would listen. No one had heard of its porate the labeled ddNTP's. and that therefore two to the twenti- ever being tried; no one saw any good I reasoned that jf there were enough eth was around a million. I stopped reason why it would not work; and yet ., nucleotides in the sample to inter- the car at a turnout overlooking An- no one was particularly enthusiastic fere with the experiment, there would derson Valley. From the glove com- about it. In the past, people had gener- also be enough for the DNApolymer- partment I pulled a pencil and pa- ally thought my ideas about DNAwere I, ase to act on. Byrunning the sample per-I needed to check my calcula- off the wall, and sometimes after a few , through a kind of preliminary mock tions. Jennifer, my sleepy passenger, days I had agreed with them. But this reaction with oligonucleotide primers objected groggily to the delay and the time I knew I was on to something. and polymerase but without ddNTP's, light, but I exclaimed that Ihad discov- Years ago, before biotechnology- I could easily deplete any nucleotides ered something fantastic. Nonplussed, when being a genetic engineer meant in the mixture by incorporating them she went back to sleep. I confirmed that you, your dad and his dad all into the extending oligonucleotides. that two to the twentieth power really drove trains-our building at Cetus Then, by raising the temperature of was over a million and drove on. had been owned by the Shell Develop- I the sample, I could separate the ex- About a mile farther down the road ment Company. Our laboratory space, I i SCIENTIFICAMERICAN April 1990 61

I ,j tr

I~ I

whose rear windows looked grandly solutions to use, what the relative and When everything was ready, I ran my out on the Berkeley hills, had given absolute concentrations of the reac- favorite kind of experiment: one in- birth to the "No-Pest Strip." It did not tants should be, how much to heat and volving a single test tube and produc- escape my notice that the PCRmight cool the mixtures, how long the mix- ing a yes or no answer. Would the PCR someday travel as far as its sibling tures should run and so on. Some of amplify the DNA sequence I had se- invention, that distinctively scented Kornberg's early papers on DNApoly- lected? The answer was yes. piece of yellow plastic. merase helped. To run the experiment, Walking out of the lab fairly late in Months passed as I prepared for my I selected a 25-base-pair target frag- the evening, I noticed that Albert Hal- first experiment to verify whether the ment of a plasmid and two oligonu- luin, the patent attorney for Cetus, PCRwould work. I had to make many cleotide primers that were 11 and 13 was still in his office. I told him that educated guesses about what buffer baseslong,respectivel~ I had invented something and de-

SEPARATE TARGETSTRANDS EXTEND PRIMERSTO MAKE AND ATTACH PRIMERS COPIES OF TARGETS .I\. .J\.. "\ r "\ ("

t ~ CYCLE1 J 1 j t

-=nmm:o 0 ~ --mIJIIIIb j t

0 mmIIIb - CYCLE3 ' cmm:o:o 0

j t AD INFINITUM

POLYMERASECHAIN REACTION is a cyclic process; with each to allow primers to bind to them. Next DNApolymerases ex- cycle, the number of DNA targets doubles. The strands in each tend the primers by adding nucleotides to them. In this way, targeted DNA duplex are separated by heating and then cooled duplicates of the original DNA-strand targets are produced.

64 SCIENTIFICAMERICAN April 1990 I

y scribed the PCR.AI was the first per- II 1- son, out of maybe a hundred to whom i I had explained it, who agreed that it was significant. He wanted to see the autoradiogram showing the experi- mental data right away;it was still wet. Some people are not impressed by one-tube experiments, but AIwas not noticeably skeptical. Patent attorneys, after all, have a vested interest in in- ventions. Hehad followed my explana- tion of the process in his office and agreed that it made sense. Nowin the lab he was even a little excited and suggested that I get to work on the experiment and write a patent disclo- sure. As he left, he congratulated me.

or the next few months I contin- ued to study and refine the PCR Fwith the help of Fred A. Faloona, a young mathematics wizard whom I had met through my daughter. Fred MACHINE that performs the polymerase chain reaction is shown being loaded with had helped me with the first PCRex- samples of DNA. Such devices are rapidly becoming common fixtures in laboratories. periment by cycling the DNA mixture-in fact, that had been his very first biochemistry experiment, and he precedented to be accepted readily. But he looked back at my poster and I celebrated on the night of its No one was prepared for a process with an expression that I have almost success with a few beers. that provided all the DNA one could come to expect. I think that Josh, after IIII In the following months we con- want. The reaction seemed self-evi- seeing the utter simplicity of the PCR, firmed that the PCRwould work on dent to Fred and to me because it was was perhaps the first person to feel :1 larger and larger fragments of plasmid our toy. For most people, it took some what is now an almost universal first II DNA. Eventually we obtained some getting used to. response to it among molecular biol- human DNAfrom Henry Erlich's lab- ogists and other DNAworkers: "Why Ii oratory and produced evidence for n the spring of 1984,while working didn't I think of that?" And nobody the amplification of a fragment from a on the patent, I presented a poster really knows why; surely I don't. Ijust single-copy gene. Idescribing the PCRat the annual ran into it one night. Today many of the initial hitches Cetus Scientific Meeting. These meet- or inefficiencies of the PCRhave been ings were always fun, because Cetus worked out. Several slightly different had some first-rate scientific advisers, RJR THER READING protocols are now in use. I usually and I was looking forward to talking SPECIFIC ENzYMATIC AMPLICATION OF recommend that the DNAsamples be with them about my invention. DNA IN VI1RO: THE POLYMERASECHAIN cycledbetween temperatures of about REACTION. Kary Mullis, Fred Faloona, Yet nobody seemed to be interested Stephen Scharf, Randall Saiki, Glenn 98 degrees Celsius, just below boil- in my poster, and I felt increasingly Horn and Henry Erlich in Cold Spring ing, and about 60 degrees C. These anxious. People would glance at it and Harbor Symposia on Quantitative Biolo- cycles can be as short as one or two keep walking. Finally, I noticed Josh- gy, Vol. 51, No.1, pages 263-273; 1986. minutes; during each cycle the num- ua Lederberg, president of the Rocke- SPECIFIC SYNTHESIS OF DNA IN VITRO VIA ber of DNAtarget molecules doubles. feller University, nearby, and I snared A POLYMERASE-CATALYZEDCHAIN REAc- The primers are usually from 20 to 30 him into looking at my results. Josh TION. Kary B. Mullis and Fred A. FaIoona in Methods in Enzymology, Vol. 155, bases long. One of the most important looked the poster over carefully and Part F, pages 335-350; 1987. improvements in the process is the then turned his enormous head, the AMPLIFICATION OF HUMAN MINISATEL- use of a particular DNApolymerase Nobel-laureated head, the head that LITES BY THE POLYMERASECHAIN REAc- originally extracted from the bacteri- had deduced in 1946 that bacteria TION: TOWARDS DNA FiNGERPRINTING um Thermus aquaticus, which lives in could have sexual intercourse. "Does OF SINGLE CELLS.Alec J. Jeffreys, Victo- hot springs. The polymerase we had it work?" He seemed amused. ria Wilson, Rita Neumann and John originally used was easily destroyed Pleased, I confirmed that it did, and Keyte in Nucleic Acids Research, Vol. 16, by heat; consequently, more had to we talked for a long time. At one point pages 10953-10971; 1988. DNA SEQUENCING WITH THERMUS AQUAT- be added during each cycle of the re- he mentioned that about 20 years ICUS DNA POLYMERASE AND DIRECT SE- action. The DNApolymerase of Ther- previously, after Kornberg had discov- QUENCING OF POLYMERASE CHAIN REAc- mus aquaticus, however, is stable and ered DNApolymerase, the two of them TION-AMpLIFIED DNA. M. A. Innis, K. B. active at high temperatures, which had considered the notion that the Myambo, D. H. Gelfand and Mary Ann means that it only needs to be added enzyme could somehow be harnessed D. Brow in Proceedings of the National at the beginning of the reaction. This to make large quantities of DNA.They Academy of Sciences, Vol. 85, No. 24, high-temperature polymerase is now had not figured out exactly how to do pages 9436-9440; December, 1988. THE POLYMERASECHAINREACTION.T. ]. produced conveniently by genetically it, however. I reminded hiin that oligo- White, Norman Arnbeim and H. A. Er- engineered bacteria. nucleotides were not readily available lich in Trends in Genetics, Vol. 5, No.6, The virtually unlimited amplifica- at that time and that there was hardly pages 185-188; June, 1989. tion of DNAby the PCRwas too un- any DNAsequence information either.

SCIENTIFIC AMERICAN April 1990 65