Primer-Directed Enzymatic Amplification of DNA with a Thermostable DNA Polymerase

Author(s): Randall K. Saiki, David H. Gelfand, Susanne Stoffel, Stephen J. Scharf, Russell
Higuchi, Glenn T. Horn, Kary B. Mullis and Henry A. Erlich
Source: Science, New Series, Vol. 239, No. 4839 (Jan. 29, 1988), pp. 487-491
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/1700278
Accessed: 19-09-2016 04:27 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1700278?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to digitize,
preserve and extend access to Science

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms
18. I thank many colleagues at Scripps for comments on the manuscript. This work was supported in part by
If the K-T boundary isotopic spike is
the ideas expressed in this report, in particular G. grants from the National Science Foundation and
indeed the result of impact-related acid rain, Arrhenius, S. Galer, J. Gieskes, M. Kastner, D. Lal,the National Aeronautics and Space Administration.
the oceanic strontium isotope record may G. Lugmair, and H.-G. Stosch. Comments from
two anonymous reviewers also improved the origi-
reveal other large impacts. The seawater 28 September 1987; accepted 7 December 1987
nal manuscript. I thank P. Hey for preparation of
strontium curve of Burke et al. (9), which
spans the past 500 million years, shows at
least two other prominent high spikes in the
87Sr/86Sr ratio, one in the mid-Cretaceous, at
-100 million years, and the other in the
Primer-Directed Enzymatic Amplification of DNA
Pennsylvanian, at -290 million years. The
first appears to precede by a few million with a Thermostable DNA Polymerase
years the mass extinction event at the Ceno-
manian-Turonian boundary. There is also a
RANDALL K. SAIKI, DAVID H. GELFAND, SUSANNE STOFFEL,
large increase in 87Sr/86Sr across the Permi-
STEPHEN J. SCHARF, RUSSELL HIGUCHI, GLENN T. HORN,
an-Triassic boundary (9), the time of the
KARY B. MULLIS,* HENRY A. ERLICH
most extreme mass extinction in the Phaner-
ozoic record (17). However, the increase
appears to be rather gradual, extending over A thermostable DNA polymerase was used in an in vitro DNA amplification
20 million to 25 million years, and is thus procedure, the polymerase chain reaction. The enzyme, isolated from Thermus aquati-
quite different in character from the K-T cus, greatly simplifies the procedure and, by enabling the amplification reaction to be
spike. Nevertheless, data are sparse for this performed at higher temperatures, significantly improves the specificity, yield, sensitiv-
interval, and more work will be required to ity, and length of products that can be amplified. Single-copy genomic sequences were
determine the exact nature of the increase. amplified by a factor of more than 10 million with very high specificity, and DNA
The occurrence of a spike toward higher segments up to 2000 base pairs were readily amplified. In addition, the method was
values in the seawater 87Sr/86Sr record at the used to amplify and detect a target DNA molecule present only once in a sample of 105
K-T boundary is tantalizing evidence for cells.
enhanced continental weathering, possibly
due to impact-related acid rain. Detailed T HE ANALYSIS OF SPECIFIC NUCLEO- fragment, approximately 2", where n is the
strontium isotopic studies through this and tide sequences, like many analytic number of cycles.
other intervals where such spikes appear are procedures, is often hampered by the One of the drawbacks of the method,
required to determine precisely the nature of presence of extraneous material or by the however, is the thermolability of the
the isotopic variations with respect to stra- extremely small amounts available for exami-Klenow fragment of Escherichia coli DNA
tigraphy, and particularly with respect to nation. We have recently described a meth- polymerase I used to catalyze the extension
mass extinctions. od, the polymerase chain reaction (PCR), of the annealed primers. Because of the heat
denaturation step required to separate the
that overcomes these limitations (1, 2). This
REFERENCES AND NOTES technique is capable of producing a selective newly synthesized strands of DNA, fresh
1. J. Hess, M. L. Bender, J.-G. Schilling, Science 231, enrichment of a specific DNA sequence by enzyme
a must be added during each cycle-a
979 (1986). factor of 106, greatly facilitating a variety tedious
of and error-prone process if several
2. L. W. Alvarez, W. Alvarez, F. Asaro, H. V. Michel,
subsequent analytical manipulations. PCR samples are amplified simultaneously. We
ibid. 208, 1095 (1980).
3. R. G. Prinn and B. Fegley, Jr., Earth Planet. Sci. has been used in the examination of nucleo- now describe the replacement of the E. coli
Lett. 83, 1 (1987). tide sequence variations (3-5) and chromo- DNA polymerase with a thermostable DNA
4. J. Lewis, H. Watkins, H. Hartman, R. Prinn, Geol.
Soc. Am. Spec. Pap. 190 (1982), p. 215. somal rearrangements (6), for high-efficien- polymerase purified from the thermophilic
5. M. R. Palmer and H. Elderfield, Nature (London) cy cloning of genomic sequences (7), for bacterium, Thermus aquaticus (Taq), that
314, 526 (1985). can survive extended incubation at 95?C
direct sequencing of mitochondrial (8) and
6. D. J. DePaolo and B. L. Ingram, Science 227, 938
(1985). genomic DNAs (9, 10), and for the detec- (12). Since this heat-resistant polymerase is
7. F. Surlyk and M. B. Johansen, ibid. 223, 1174 tion of viral pathogens (11). relatively unaffected by the denaturation
(1984).
PCR amplification involves two oligonu- step, it does not need to be replenished at
8. Since strontium is removed from the oceans mainly
by biogenic carbonate precipitation, removal rates cleotide primers that flank the DNA seg- each cycle. This modification not only sim-
may have been much less for some period afterment the to be amplified and repeated cycles of plifies the procedure, making it amenable to
K-T boundary event, leading to higher strontium
concentrations in seawater and a temporary increase heat denaturation of the DNA, annealing of
automation, it also substantially improves
in "residence time." the primers to their complementary se- the overall performance of the reaction by
9. W. H. Burke et al., Geology 10, 516 (1982).
quences, and extension of the annealed increasing the specificity, yield, sensitivity,
10. M. A. Wadleigh, J. Veizer, C. Brooks, Geochim.
Cosmochim. Acta 49, 1727 (1985). primers with DNA polymerase. These prim- and length of targets that can be amplified.
11. D. W. Mittlefehldt and G. W. Wetherill, ibid. 43, ers hybridize to opposite strands of the Samples of human genomic DNA were
201 (1979).
12. D. W. Graham, M. L. Bender, D. F. Williams, L. D.
target sequence and are oriented so DNA subjected to 20 to 35 cycles of PCR amplifi-
Keigwin, Jr., ibid. 46, 1281 (1982). synthesis by the polymerase proceeds across
13. A. A. Afifi, 0. P. Bricker, J. C. Chemerys, Chem. the region between the primers, effectively
Geol. 49, 87 (1985). R. K. Saiki, S. J. Scharf, R. Higuchi, G. T. Horn, K. B.
14. R. M. Garrels and F. T. Mackenzie, Evolution of doubling the amount of that DNA segment.
Mullis, H. A. Erlich, Cetus Corporation, Department of
Sedimentary Rocks (Norton, New York, 1971). Moreover, since the extension products are Human Genetics, 1400 Fifty-Third Street, Emeryville,
15. G. W. Brass, Geochim. Cosmochim. Acta 40, 721 CA 94608.
also complementary to and capable of bind-
(1976). This paper provides data for both strontium D. H. Gelfand and S. Stoffel, Cetus Corporation, De-
isotopic ratios and strontium-to-calcium ratios for ing primers, each successive cycle essentially partment of Microbial Genetics, 1400 Fifty-Third Street,
limestone. doubles the amount of DNA synthesized in Emeryville, CA 94608.
16. C. Officer, C. Drake, J. Devine, Eos 66, 813 (1985).
17. J. J. Sepkoski, Jr., Geol. Soc. Am. Spec. Pap. 190
the previous cycle. This results in the expo- *Present address: Xytronyx, 6555 Nancy Ridge Drive,
(1982), p. 283. nential accumulation of the specific target Suite 200, San Diego, CA 92121.

29 JANUARY I988 REPORTS 487

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms
cation with optimal amounts of either Nevertheless, Southern blot analysis with a mum of the enzyme. (12). During the tran-
Klenow or Taq DNA polymerase and ana- 3-globin hybridization probe reveals the 1-sition from 40? to 70'C, poorly matched
lyzed by agarose gel electrophoresis (Fig. globin amplification fragment in all samples primer-template hybrids (which formed at
1A) and Southern hybridization (Fig. 1B). in which the 3-globin target sequence was 40'C) dissociate, and only highly comple-
The PCR primers direct the synthesis of a present. mentary substrate remains as the reaction
167-bp segment of the human 3-globin A substantially different electrophoretic approaches the temperature at which cataly-
gene (13). Electrophoretic examination of pattern is seen in the amplifications per- sis occurs. Furthermore, because of in-
the reactions catalyzed by the Klenow poly- formed with Taq DNA polymerase, where creased specificity, there are fewer nonspe-
merase reveals a broad molecular size distri- the single predominant band is the 167-bp cific extension products to compete for the
bution of amplification products (7) that is target segment. This specificity is evidently polymerase, and the yields of the specific
presumably the result of nonspecific anneal- due to the temperature at which the primers target fragment are higher.
ing and extension of primers to unrelated are extended. Although the annealing step is As the values for extent of amplification
genomic sequences under what are essential- performed at 40'C, the temperature of Taq and overall efficiency indicate (Fig. 1), the
ly nonstringent hybridization conditions: polymerase-catalyzed reactions was raised exponential accumulation of PCR amplifica-
Klenow polymerase reaction buffer at 370C.
to about 70'C, near the temperature opti- tion products is not an unlimited process.
Eventually, a level of amplification is
reached where more primer-template sub-
A B strate has accumulated than the amount of
1 2 3 4 5 6 7 8 9 10 1112 1 2 3 4 5 6 7 8 9 10 11 12
enzyme present is capable of completely
.1.35 extending in the allotted time. When this
1.08
`0.872 occurs, the efficiency of the reaction de-
I 0.603 clines, and the amount of PCR product
- 0.310 accumulates in a linear rather than an expo-
=0.281,0.271
- 0.234 nential manner. Under the conditions de-
- 0.194 scribed (Fig. 1), the Klenow polymerase
reaction begins to "plateau" around the 20th
_ |l 9 - 0 . 1 1~~~~~~~~.18
cycle, after a 3 x 105-fold amplification of
- 0.072 the 3-globin target sequence. The higher
specificity of the Taq polymerase-catalyzed
reaction, however, permits it to proceed

Fig. 1. Comparison of Klenow and Taq DNA polymerase-catalyzed PCR amplification products of the
efficiently for an additional five cycles, to an
human 3-globin gene. (A) Electrophoretic analysis of the PCR products obtained with Kienow amplification level of 4 x 106 before the
polymerase (lanes 1 to 6) and Taq polymerase (lanes 7 to 12) after 0 cycles (lanes 1 and 7), 20 cycles
activity of the polymerase becomes limiting
(lanes 2 and 8), 25 cycles (lanes 3 and 9), 30 cycles (lanes 4 and 10), and 35 cycles (lanes 5, 6, 11, and
(14).
12) of amplification. The DNA samples that were amplified were prepared from the human cell lines
The specificity of the Taq DNA polymer-
MOLT4 (lanes 1 to 5 and 7 to 11) and GM2064 (lanes 6 and 12). MOLT4 is homozygous for the
normal 3-globin gene (2), GM2064 possesses a homozygous deletion of the entire 3-globin gene ase-mediated amplifications can be affected
complex (21). The molecular size marker is 250 ng of Hae III-digested 4X174 replicative form (RF) by the time allowed for the primer extension
DNA (New England Biolabs). (B) Southern analysis of the gel with a 32P-labeled oligonucleotide step and by the quantity of enzyme used in
probe. Compared against standards, the intensities of the bands in the Klenow amplifications were
the reaction. The electrophoretic patterns of
estimated to be equivalent to increases of 2.8 x 105, 1.1 x 106, 2.2 x 106, and 2.2 x 106 (lanes 2 to 5)
with corresponding overall reaction efficiencies of 87, 74, 63, and 52% [calculated according to (2)1. PCR products obtained with different ex-
The values for the intensities of the bands in the corresponding Taq reactions were 2.8 x 10,tension times and units of Taq DNA poly-
4.5 x 106, 8.9 x 106, and 1.7 x 10' (lanes 8 to 11) with overall efficiencies of 87, 85, 70, and 61%. merase on otherwise identical samples indi-
Amplification of genomic targets by PCR with Klenow polymerase was performed as described (7). cate that the amount of nonspecific DNA
Briefly, 100-1?l reactions containing 1 ,ug of genomic DNA in 50 mM NaCI, 10 mM tris (pH 7.6), 10
increases as the extension times become
mM MgCI2, 10% dimethyl sulfoxide, 1 pM each primer (PC03 and KM38), and 1.5 mM each of the
deoxyribonucleotide triphosphates (dNTP: dATP, dCTP, TTP, dGTP). The reactions were performed longer or the number of Taq polymerase
units increases (Fig. 2A).
by 20 to 35 repetitions (cycles) as follows. The samples were heated from 37? to 950C over a 2.5-minute
period (to denature the DNA) and cooled to 37C (3 minutes) (to anneal the primers); 1 unit of A significant improvement in specificity is
Klenow polymerase (USB) was added to each sample and then incubated at 37C for 2 minutes (to
obtained when the temperature of the prim-
extend the bound primers). Amplifications with Taq polymerase took place in 100-1?l reaction mixtures
er annealing step is raised from 40? to 55GC.
containing 1 ,ug of genomic DNA in 50 mM KC1, 10 mM tris (pH 8.4), 2.5 mM MgCI2, each primer
This effect is demonstrated by the amplifica-
(PCO3 and KM38) at 1 pM, each dNTP (dATP, dCrP, TITP, dGTP) at 200 pM, gelatin at 200 1?g/ml,
and 2 units of polymerase. The samples were overlaid with several drops (- 100 ,ul) of mineral oil totion of the 13-globin gene in a set of dilutions
prevent condensation and subjected to 20 to 35 cycles of amplification as follows. The samples were
of normal genomic DNA into the DNA of a
heated from 70? to 95?C over a 1-minute period (to denature the DNA), cooled to 40?C over 2 minutes
mutant cell line with a homozygous deletion
(to anneal the primers), heated to 70?C in 1 minute (to "activate" the polymerase), and incubated at that
temperature for 0.5 minute (to extend the annealed primers). Additional Taq DNA polymerase wasof notthe 13-globin gene (Fig. 2B). These sam-
added to the samples during amplification. One unit of enzyme is the amount that will incorporate 10 ples, each containing 2 ptg of DNA, repre-
nmol of total deoxyribonucleotide triphosphates into acid-precipitable material in 30 minutes at 74?Csent 13-globin gene frequencies that range
with activated salmon sperm DNA as template (12). The enzyme was prepared from T. aquaticus, strain
from one copy per genome (two copies per
YT1, by a modification of published procedures (22). Thermal cycling was performed in a programma-
ble heat block (Perkin Elmer-Cetus Instruments). After the last cycle, all samples were incubated for diploid
an cell) in the undiluted normal DNA
additional 5 to 10 minutes at 370 or 70?C to ensure that the final extension step was complete. After sample, to as little as one copy per 106
precipitation with ethanol and resuspension in 100 ,ul TE buffer (20), each sample (8 ,ul) was resolved genomes (one copy per 500,000 cells) in the
on a composite gel of 3% NuSieve and 1% SeaKem agaroses (FMC) in tris-borate buffer and stained 10-6 dilution. After 40 cycles of PGR with
with ethidium bromide (20). Southern transfers were performed essentially as described (23) onto
primer annealing at 400G, the specific ampli-
Genetrans-45 nylon membranes (Plasco). A 19-base oligonucleotide probe specific for the amplified 1-
globin fragment, 19A, was 5' end-labeled with 32P and hybridized to the filter (3). The autoradiogram fication fragment can be seen in the 10-2
was exposed for 3 hours with a single intensification screen. dilution by electrophoretic examination and

488 SCIENCE, VOL. 239

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms
in the lo- dilution by Southern analysis of the 13-globin gene. If only one target readily effect the synthesis of segments long-
(15). However, in parallel samples amplified molecule was present initially, the intensities er than 400 bp. With other primers and
with 550C annealing, much less nonspecific of the PCR products in the 10-6 samples are longer extension times, this enzyme can
DNA is present and the product band is equivalent to amplifications of about 109 to amplify genomic target sequences of up to
visible in the original gel in the i0' dilu-
1010. This experiment supports the conclu- 2.0 kb (Fig. 4B); longer fragments can be
tion and, by Southern analysis, in the 10-6sion that Taq polymerase-mediated PCR synthesized but with reduced efficiency and
dilution (15). Although annealing at 550C is
performed with primer annealing at 550C is
yield.
usually of limited value when amplifying able to successfully amplify a single target In an application exploiting this capability
genomic targets present at one or two copies molecule in a DNA background of 105 cells. of Taq polymerase, complementary DNA
per cell (Fig. 2B, lane 1 compared to lane 9), Subsequent investigations have demonstrat- (cDNA) inserts in the phage Xgtl 1 cloning
the reduction in the amount of nonspecific ed that the reaction can detect a single copy vector were amplified from crude phage
primer extension products improves the lim- of target DNA in 10 ptg of DNA, the suspensions with primers that flank the Eco
it of sensitivity by two orders of magnitude. equivalent of 1.5 million diploid cells (15). RI insertion site of the vector. We chose 16
Since 2 ptg of the 10-6 dilution would The variation in the electrophoretic pat- plaques at random-15 with inserts and 1
contain, on average, only 0.6 copy of the 13-
tern of PCR products (Fig. 3A) probably without-and subjected them to 25 cycles of
globin gene, these data suggest that PCR reflects the stochastic nature of nonspecific amplification. A single band corresponding
priming events. The stringency of the prim-
with Taq polymerase may be able to amplify to the amplified insert DNA was observed
a single target molecule in 105 to 106 cells.
er annealing step is sufficiently high that the for each phage isolate; these fragments were
This result was confirmed by demonstrat- priming of a nontarget sequence is rare and 400 to 2000 bp in length (Fig. 5). The
ing a Poisson distribution of successful am- does not always occur in every sample. Only phage that did not contain an insert pro-
plifications on limiting amounts of the 13- when such a priming event occurs during duced an 87-bp fragment, the distance be-
globin template. Fifteen identical 1-ptg sam- the first few cycles of amplification is there tween the primers on the uninterrupted
ples of the same 10-6 dilution were ampli- an opportunity for an electrophoretically vector. Between 0.5 and 1.0 ptg of each
fied for 60 cycles with annealing at 550C and visible band to accumulate. Similarly, the cDNA insert fragment was synthesized,
analyzed for the presence of a 13-globin which, on the basis of 106 phage per plaque,
variability of the intensities of the specific IB-
amplification product by Southern hybrid- globin signal (Fig. 3B) is presumably the represents an amplification of approximately
ization. Each of these samples of genomic result of the failure of the polymerase to 106. The extent and purity of these amplifi-
DNA (equivalent to that in 150,000 cells) locate and extend all template strands during cations suggest that PCR with Taq DNA
should have, on average, 0.3 copy of the the 13-first few cycles of the reaction. Because polymerase may be an effective and efficient
globin gene. The fraction of those samples there is initially only one target molecule, way to isolate the inserts of plasmid or
that contain at least one copy is expressed as any inefficiency of this sort would strongly
A
effect the final yield of amplification prod- 1 2 3 4 5 6 7 8 9 10 1112 13 14 15
1 - e-0 = 0.26
uct.
or 4 out of 15; the remaining 11 should Three pairs of primers which, in various
contain no 13-globin templates. The ob- combinations, define segments of 110 to
served frequency was 9 of 15 (Fig. 3). 400 bp were used to examine the relative
Although this is twice as high as the expect-
abilities of the Klenow and Taq polymerases
ed value, it does not significantly affecttothe
synthetize longer PCR products. South-
interpretation of the results. The distribu- ern analysis of the reaction products after 25
tion of successful amplifications is consistentcycles of amplification reveals that the
with a concentration of approximately one Klenow polymerase does not sustain the
P-globin gene per 500,000 cells which indi-exponential accumulation of DNA se-
B
cates that of the nine positive samples, most, quences much greater than 250 bp (Fig. 1 2 3 4 5 6 7 8 910 11 12 1314 15

if not all, originally contained a single copy 4A). The Taq polymerase, in contrast, can

Fig. 2. Factors that influ- A B

ence the specificity of Taq 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
polymerase-catalyzed am-
plifications. (A) Electropho-
retic analysis of specificity as Fig. 3. Poisson
ogoueoide distribution
prbe RS8, speifi fo th of single
affected by increasing exten- quences in samples of i0' cells. (A) E
sion times: 0, 0.5, 2, and 8 retic analysis of PCR products in a
_ - - 11111REPORTI 489
minutes (lanes 1 to 4, re- samples (lanes 1 to 15). The arrow in
spectively); and by increas- position of the 150-bp amplification pr
ing amounts of enzyme: is visible in some samples (lanes 3, 4, and 6).
0.25, 0.5, 1, 2, 4, and 8 Molecular markers as described (Fig. 1). (B3)
units (lanes 5 to 10, respec- Southern analysis of the gel with a P-globin-
tively). (B) Electrophoretic specific oligonucleotide probe. Fifteen I-I~g sam-
ples of
examination of the effect of annealing temperature on specificity at 400 (lanes 1 to 8) and 550C (lanes 9 the 10-6 dilution of MOLT4 DNA in
to 16) with serial dilutions of MOLT4 DNA in GM2064 DNA: undiluted MOLT4 (lanes 1 and 9), GM2064 DNA prepared previously (Fig. 2) were
10-' (lanes 2 and 10), 10-2 (lanes 3 and 11), 10-3 (lanes 4 and 12), 1O' (lanes 5 and 13), 10-5 amplified for 60 cycles with 550C annealing, and
(lanes 6 and 14), 10-6 (lanes 7 and 15), and undiluted GM2064 (lanes 8 and 16). Molecular size the primers RS79 and RS8O. The amplified sam-
marker as described (Fig. 1). The amplifications were generally conducted as described (Fig. 1). The
reactions for (A) contained 1 Rg of DNA and were subjected to 30 cycles with the primers PCO3 and
PCO4 (1 10-bp product) with the indicated times of extension and units of enzyme. The samples for (B)
each contained 2 pug of DNA and were amplified for 40 cycles with either 400 or 550C annealing with
the primers RS79 and RS80 (150-bp product). A portion of the reaction, 5 RI1, was subjected to
electrophoresis on a NuSieve-SeaKem agarose gel.

29 JANUARY I988

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms
phage recombinants, eliminating the need Although this value is somewhat greater made by the Klenow enzyme were com-
for plasmid or phage growth, vector purifi- than the 8 x 1O-5 misincorporation rate posed of a mosaic of the different alleles
cation, and insert isolation. observed with Klenow polymerase-cata- being amplified. These "shuffled" clones ap-
The fidelity of the thermostable Taq poly- lyzed PCR (7, 18), the errors made by Taq parently arise from incomplete extension of
merase in the amplification reaction was DNA polymerase should not be a problem. the annealed primer during one cycle. In
assessed by cloning and sequencing individ- Analytic procedures that use a significant later cycles, these incomplete products may
ual amplification products (7) with primers portion of the reaction product, such as hybridize to other allelic templates and be
that define a region of the HLA-DPP gene direct sequencing (see below) or filter hy- extended, thus producing the mosaic (18).
(16). The products of 30-cycle PCR amplifi- bridization with allele-specific oligonucleo- Few shuffled clones have been observed
cations were cloned into an M13 vector, tide probes (3), are not affected by the small with the Taq DNA polymerase, which may
multiple isolates of the same allele obtained, fraction of misincorporated bases. Cloning be the result of higher processivity.
and their sequences compared. In 28 sepa- and sequencing individual amplification The higher specificity of Taq polymerase-
rate clones, each with 239 bp of amplified mediated amplifications can facilitate the
fragments may include these errors, but they
DPP genomic DNA, 17 misincorporatedare readily identified by analyzing several direct sequencing of single-copy human
bases were identified representing an overallisolates and establishing a consensus. A diffi- genes, particularly those that may be mem-
error frequency of 0.25% (17). These misin- culty could arise if PCR were attempted on a bers of a gene family (10). A 11O-bp frag-
corporations occurred throughout the am- sample initially containing only a few copies ment of the 13-globin gene was amplified
plified product and no deletions or inser- of the target template. In that situation, a
tions were detected. Because each misincor- misincorporation during the early stages of A B
poration event is retained and propagated the reaction would represent a substantial G A T C G A T C

through succeeding cycles of amplification, fraction of the molecules present and could -..;.SR _ C A _ A
the frequency of errors observed in the complicate the analysis of the amplification
cloned products is cumulative and a func- product. However, the amplification and
tion of the number of doublings; the actual comparison of several samples would reveal
rate of misincorporation is lower. If con- and resolve any inconsistencies.
stant over the 30 PCR cycles, the misincor- Even though the Klenow DNA polymer-
poration rate per nucleotide per cycle for ase has better fidelity in PCR amplification,
Taq polymerase is estimated at 2 x 10-4it is more likely to produce a different type 3-globin &-globin f3-globin
(17). of sequence artifact. Some PCR products Kienow TaqT

Fig 6. Direct sequencing of a PGR-amplified

Fig. 4. Examination of the A B
a b c d e I g h i 1 2 3 4 5
human genomic target. Autoradiograms of se-
ability of Kienow and Taq
quencing gels derived from (A) Kienow polymer-
DNA polymerases to ampli-
-1.35,1.08 ase-catalyzed and (B) Taq polymerase-catalyzed
fy longer target segments. 0.872
-0.603 -3.68 amplification reactions. Nucleotide sequences de-
(A) Autoradiogram of
0 -~~~~~0.310
.1 = * ~ o23,2.32,1.93 termined from the films are listed beside the
Southern filter comparing S 0 0 ~~~~~~~~0.281,0.271 P -1.35 autoradiograms, including the ambiguous bases
relative efficiencies of
_S :0.194 -0.872 ascribed to 8obin 8-. A 11o-bp region of f3-globin
Klenow and Taq polymerase -0.118 -0.603 was amplified in samples of MOLT4 DNA with
at amplifying fragments of
-0.072 the primers PGO3 and PCO4 and either Kienow
increasing size. The lanes are or Taq polymerase (Fig. 1). The sequencing of
grouped in pairs, Klenow
the PGR products was done as described (8) with
polymerase amplifications on the left and Taq polymerase amplifications on the right, according to the
modifications to permit the substitution of re-
primers used: 110 bp, primers PCO3 and PC04; 167 bp, PC03 and KM38; 250 bp, PC03 and GH21; verse transcriptase for the Klenow fragment of
205 bp, KM29 and PC04; 262 bp, KM29 and KM38; 345 bp, KM29 and GH21; 268 bp, GH20 and DNA polymerase I (24). Approximately 0.1 to
PC04; 325 bp, GH20 and KM38; and 408 bp, GH20 and GH21 (lane pairs a to i, respectively). 1.0 pmol of microconcentrator-purified PCR
Molecular markers as described (Fig. 1). (B) Autoradiogram of Southern blot for examining capacity of and 3 to 5 pmol of 32P-labeled sequenc-
product
Taq polymerase to synthesize products greater than 1 kb: 989 bp, RS40 and RS80; 1390 bp, GH20
ing primer, RHf9, were combined in 10 spl of 50
and RS80; 1988 bp, GH20 and RS114; 2489 bp, GH20 and RS115; and 3015 bp, GH20 and RS116 mM KGC, 50 mM tris (pH 8.0), 5 mM Mgei2,
(lanes 1 to 5, respectively). Molecular marker is a combination of BstE I1-digested X DNA and HaemM dithiothreitol. The reaction was
and 10
III-digestcd XX174 RF DNA. Amplifications were performed for 25 (A) or 30 cycles (B) with the heated at 95scr for 10 minutes then quenched on
indicated primers and 2 minutes (A) or 10 minutes (B) primer extension steps. Electrophoretic ice, and 2.2 m o was added to each of four 2PCl
rcsolution and Southern analysis were performed on 10 RI of each sample. solutions: "dddG" is 20 FM dGTP, 10 zM
dATP, 100 an M TTP, 100 p M dCTP, and 6 s zM
Fig. 5. Amplification of insertsa in
b 1 2 3 4 5 6 7 8 9 1011
a phage X 1213141516 b a ddGTP; "dddA" is 100 czM dGTP, 10 oM dATP,
cDNA library. Lanes 1 to 15, phages containing 100 KCM TTP, 100 zM dCTPH, 2.5 MM ddATP;
inserts; lane 16, phage without a detectable insert. "dddT" is 100 1M dGTP, 10 1M dATP, 20 p1M
Molecular markers are either 500 ng of BstE II- ITP, 100 mM dCTP, 15 hrM ddTTP; "dddrC is
digested X DNA (lanes a) or 250 ng of Hae III- 100 s iM dGTP, 10 2M dATP, 100 M TTP, 20
digested OX174 RF DNA (lanes b). A Xgtll M dCTP, 5 IAM ddCTP [each in 50 maM KCI,
human fibroblast cDNA library (Clontech) was 50 mM tris (pH 8.0), 5rM MgCT2, and 10 mMA
plated on X-gal plates at high dilution by standard dithiothreitol]. Five units of AMV reverse tan-
techniques (20). Well-isolated plaques were se- scriptase (Life Sciences) was added to each reac-
lected at random, 15 clear (with insert) and 1 blue tion and incubated at 3TC for 15 minutes. The
(without insert), and excised with the tip of a reactions were held for an additional 15 minutes
Pasteur pipette. The agarose plugs were eluted in after the addition of 1 ,ul of a solution containing
0.2 ml of deionized water for 30 minutes, and 50 1 mM each dNTP. Five microliters of formamide-
pI of the eluates was subjected to 25 cycles of dye stop mLic was added to each reaction and
amplification as described (Fig. 1). The primers heated for 5 minutes at 95?C before loading 2.5
used were two 24-base sequencing primers, 1218 ,lon an 8% polyacrylamide sequencing gel (25).
and 1222 (New England Biolabs) that flank the Electrophoresis
amplified samples (10 iLl) was resolved on a and autoradiography were by
1.4%
Eco RI insertion site of the vector. Each of the SeaKem agarose gel. standard techniques (20).

490 SCIENCE, VOL. 239

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms
with either Klenow or Taq polymerases. A derived from a single progenitor sequence, number of cycles or degree of amplification. Rather,
it is the concentration of total PCR product, the
third primer, complementary to a region of PCR based on Taq DNA polymerase repre- concentration of the enzyme, and the length of
the DNA between the two PCR primers, sents a form of "cell-free molecular cloning" extension time at 70C that defines the conditions
was end-labeled with 32P and used in the that can accomplish in an automated 3- to 4- under which the activity of the enzyme becomes
limiting. Sufficient molar excesses of deoxyribonu-
chain-termination sequencing reaction (19). hour in vitro reaction what might otherwise cleotide triphosphates and primers were present in
The sequence of the Klenow polymerase- take days or weeks of biological growth and the reactions so the consumption of these reagents
was not a factor.
catalyzed amplification product displays biochemical purification.
15. R. K. Saiki, unpublished observations.
base pair ambiguities at several positions 16. T. L. Bugawan et al., manuscript in preparation.
(Fig. 6). The origin of these extra bands is REFERENCES AND NOTES 17. Thirteen of the errors involved transitional changes
(where one purine nucleotide replaces the other),
attributed to the presence of 8-globin gene 1. K. B. Mullis and F. A. Faloona, Methods Enzymol.
ten of them resulting in a G-C pair. Of the four
155, 335 (1987).
sequences. The 8-globin gene is closely relat- transversional misincorporations (where a purine
2. R. K. Saiki et al., Science 230, 1350 (1985).
ed to 3-globin, and both of the PCR prim- nucleotide replaces a pyrimidine), two were A-T to
3. R. K. Saiki, T. L. Bugawan, G. T. Horn, K. B.
T-A and two were G-C to T-A. The formula used to
ers match 8-globin at 18 out of 20 positions Mullis, H. A. Erlich, Nature (London) 324, 163
calculate the misincorporation rate is m = 2(fld),
(1986).
(2). Because of the relative nonspecificity of wherefis the observed error frequency in the PCR
4. S. H. Embury et al., N. Engi. J. Med. 316, 656
product and d is the number of doublings. [W.
the Klenow-mediated amplifications, 8-glo- (1987).
Hayes, The Genetics of Bacteria and Their Viruses
bin is coamplified to at least 10% of the level 5. J. L. Bos et al., Nature (London) 327, 293 (1987).
(Wiley, New York, 1965).]
6. M.-S. Lee et al., Science 237, 175 (1987).
of P-globin (7). However, the higher speci- 18. G. T. Horn, unpublished observations.
7. S. J. Scharf, G. T. Horn, H. A. Erlich, ibid. 233,
19. F. Sanger, S. Nicklen, A. R. Coulson, Proc. Natl.
ficity of Taq polymerase reactions per- 1076 (1986).
Acad. Sci. U.S.A. 74, 5463 (1977).
8. L. A. Wrischnik et al., Nucleic Acids Res. 15, 529
formed with 550C annealing does not per- 20. T. Maniatis, E. F. Fritsch, J. Sambrook, Molecular
(1987).
mit the primers to anneal to 8-globin and Cloning: A Laboratory Manual (Cold Spring Harbor
9. G. McMahon, E. Davis, G. N. Wogan, Proc. Natl.
Laboratory, Cold Spring Harbor, NY, 1982).
only the 3-globin segment is amplified (Fig.Acad. Sci. U.SA. 84, 4974 (1987).
21. D. Tuan, E. Feingold, M. Newman, S. M. Weiss-
10. C. Wong et al., Nature (London) 330, 384 (1987).
6). man, B. G. Forget, Proc. NatI. Acad. Sci. U.SA. 80,
11. S. Kwok et al., J. Virol. 61, 1690 (1987).
6937 (1983).
The amplification of RNA transcripts can 12. D. H. Gelfand et al., manuscript in preparation.
22. A. Chien, D. B. Edgar, J. M. TrelaJ. Bacteriol. 127,
also be performed with Taq polymerase 13. Names and sequences of the synthetic oligonucleo-
1550 (1976).
tides used in this report. GH20: GAAGAGC-
PCR. After conversion of the messenger CAAGGACAGGTAC, GH21: GGAAAATAGAC-
23. K. C. Reed and D. A. Mann, NucleicAcids Res. 13,
7207 (1985).
RNA (mRNA) to first-strand cDNA with CAATAGGCAG, KM29: GGTTGGCCAATCTAC-
24. G. M. Air, Virology 97, 468 (1979).
TCCCAGG, KM38: TGGTCTCCITAAACCTGT-
oligo(dT) primers and reverse transcriptase 25. A. M. Maxam and W. Gilbert,MethodsEnzymol. 65,
CEI7G, PC03: ACACAACTGTGITCACTAGC,
by standard methods (20), the resulting 499 (1980).
PC04: CAACITCATCCACGITCACC, RH09: A-
26. We thank T. L. Bugawan and C. Long for assistance
single-stranded cDNA can be directly ampli- ACCTCAAACAGACACC, RS40: A7FJTMCCCAC-
with cloning and sequencing of PCR products; L.
CC(TAGGCTG, RS79: CATGCCTC'lTGCACC-
fied by PCR. With the HLA-DQjx PCR AITC, RS80: TGGTAGCTGGAITGTAGCTG,
Goda, D. Spasic, and C. Levenson for synthesis of
primers reported previously (7), mRNA oligonucleotides; L. Johnson, R. Leath, D. Jones,
RS81: CTGGGITAAGGCAATAGCA, RS114: CC-
and J. Widunas for engineering and instrument
transcripts present at about 0.01% in 100 TCCAAATCAAGCCTCTAC, RS115: ATCCTGA-
support; D. Carlin for statistical interpretation; S.
GGAAGAATGGGAC, RS116: GTTTGATGTAG-
ng of cDNA prepared from lymphoblastoid CCTCAC(TC, 19A: CTCCTGAGGAGAAGTCT-
Nilson and E. Ladner for graphic services; A. Wil-
son for the phrase "cell-free cloning"; and J. Sninsky
polyadenylated [poly(A)+] RNA could be GC, 1218: GGTGGCGACGACTCCTGGAGCCC-
and T. White for advice and encouragement.
G, 1222: ITGACACCAGACCAACTGGTAATG.
easily amplified to generate approximately 1
14. The "plateau" effect is not directly determined by9 the
October 1987; accepted 17 December 1987
Vug of the specific 242-bp amplification frag-
ment.
Our data demonstrate the highly specific
nature of Taq polymerase-mediated PCR
and its effect on the efficiency and sensitivity
Genomic Amplification with Transcript Sequencing
of the reaction. The amplification of both
DNA and RNA targets was readily accom-
plished by means of this thermostable en- E. S. STOFLET, D. D. KOEBERL, G. SARKAR, S. S. SOMMER*
zyme, often with yields and purities compa-
rable to fragments prepared from clonally A sequencing method called genomic amplification with transcript sequencing
isolated recombinants. This facilitates rapid (GAWTS) is described that is based on amplification with the polymerase chain
sequence analysis of mutants and variants at reaction (PCR). GAWTS bypasses cloning and increases the rate of sequence acquisi-
a known locus by allowing the PCR product tion by at least fivefold. The method involves the attachment of a phage promoter onto
to be sequenced directly. Similarly, the anal- at least one of the PCR primers. The segments amplified by PCR are transcribed to
ysis of unknown sequences could be expedit- further increase the signal and to provide an abundance of single-stranded template for
ed by PCR amplification of the cloned reverse transcriptase-mediated dideoxy sequencing. An end-labeled reverse transcrip-
segments with vector-specific primers that tase primer complementary to the desired sequence generates the additional specificity
flank the insertion site. The ability to ampli- required to generate unambiguous sequence data. GAWTS can be performed on as
fy and manipulate a target sequence present little as a nanogram of genomic DNA. The rate of GAWTS can be increased by
only once in a sample of 105 to 106 cells coamplification and cotranscription of multiple regions as illustrated by two regions of
should prove valuable in many areas of the factor IX gene. Since GAWTfS lends itself well to automation, further increases in
molecular biology. Clinical applications in- the rate of sequence acquisition can be expected.
clude the diagnosis of infectious diseases and
of rare pathologic events such as chromo- JN CONTRAST TO AUTOSOMAL RECES- represents an independent mutation. From
somal translocations. Moreover, the sensi- sive mutations, deleterious X-linked the perspective of efforts to understand the
tivity of the procedure should enable the mutations are eliminated within a few
analysis of gene expression or rearrangement generations because the affected males re- Department of Biochemistry and Molecular Biology,
in single cells. By virtue of the exponential Mayo Clinic/Foundation, Rochester, MN 55905.
produce sparingly if at all. Thus, each family
accumulation of literally billions of copies in an X-linked disease such as hemophilia B *To whom correspondence should be addressed.

29 JANUARY 1988
REPORTS 491

This content downloaded from 136.159.235.223 on Mon, 19 Sep 2016 04:27:55 UTC
All use subject to http://about.jstor.org/terms