Gone.

93 (1990) 125-128
Elsevier 125

GENE 03659

Short Communications

Use of uracil D N A glycosylase to control carry-over contamination in polymerase chain reactions

(Recombinant DNA; dUTP; primers; oligodeoxyribonucleotides; papillomavirus)

Mary C. Longo, Mark S. Beminger and James L. Hartley

Life Technologies, Inc., Gaithersburg. MD 20877 (U.S.A.)

Received by A.J. Podhajska: 31 January 1990

Revised: 22 April 1990
Accepted: 15 May 1990

SUMMARY

Polymerase chain reactions (PCRs) synthesize abundant amplification products. Contamination of new PCRs with trace
amounts ofthese products, called carry-over contamination, yields false positive results. Carry-over contamination from some
previous PCR can be a significant problem, due both to the abundance of PCR products, and to the ideal structure of the
contaminant material for re-amplification. We report that carry-over contamination can be controlled by the following two
steps: (i) incorporating dUTP in all PCR products (by substituting dUTP for dTTP, or by incorporating uracil during
synthesis of the oligodeoxyribonucleotide primers; and (ii) treating all subsequent fully preassembled starting reactions with
uracil DNA glycosylase (UDG), followed by thermal inactivation of UDG. UDG cleaves the uracil base from the
phosphodiester backbone of uracil-containing DNA, but has no effect on natural (i.e., thymine-containing) DNA. The
resulting apyrimidinic sites block replication by DNA polymerases, and are very labile to acid/base hydrolysis. Because UDG
does not react with dUTP, and is also inactivated by heat denaturation prior to the actual PCR, carry-over contamination
of PCRs can be controlled effectively if the contaminants contain uracils in place of thymines.

INTRODUCTION products and primers from previous PCRs. Of these

The polymerase chain reaction (PCR) is a widely used
and powerful analytical and preparative technique (Saiki
etal., 1988). However, the amplification that makes PCR
so useful also makes it intrinsically susceptible to contami-
nation problems. Three sources of contaminating DNA
must be considered: (/)cross-contamination between
samples, resulting in transfer of target DNA from one
sample to another; (ii)plasmid contamination from the
laboratory environment, i.e., recombinant plasmids
containing cloned target sequences; and (iii)carry-over
contamination of amplified target DNA, i.e., amplification
Correspondence to: Dr. J.L. Hartley, Life Technologies, Inc., P.O. Box
6009, Gaithersburg, MD 20877 (U.S.A.) Tel. (301)670-8380;
Fax (301)948-8977.
Abbreviations: bp, base pair(s ); dT, (deoxy)thymidine;dU, deoxyuridine;
EtdBr, ethidium bromide; HPV, human papiilomavirus; kb, kilobase(s)
or 1000 bp; nt, nucleotide(s); oligo, oligodeoxyribonucleotide; PCR, poly-
merase chain reaction; UDG, uracil DNA glycosylase.
sources, carry-over contamination is the major source of
concern (Gibbs and Chamberlain, 1989; Kwok and
Higuchi, 1989) due to its relative abundance and its ideal
structure for re-amplification. Carry-over contamination of
reagents, pipetting devices, laboratory surfaces, or even the
skin ofworkers (Kitchin et al., 1990),can yield false positive
results. Determining the source of such contamination can
be difficult and time-consuming, and extraordinary steps
are sometimes taken to keep it from occurring in the first
place.
Since the researcher ordinarily has little or no control
over either the composition of the target DNA in the speci-
mens to be amplified, or the plasmids containing target
sequences, only stringent physical isolation procedures can
control contamination from these sources. However, it is
possible to arrange a discrimination against DNA products
of PCRs, so as to control carry-over contamination. We
describe two related procedures that achieve this discrimi-
nation. Both utilize the enzyme UDG.

0378-1! 19/90/$03.50 © 1990ElsevierSciencePublishersB.V. (BiomedicalDivision)

126

U D G has the in vivo function of removing uracil bases
from DNA, since slow chemical deamination of cytosine
produces uracil in vivo, and consequent C --, T transitions,
The mutagenic effect of the C --, T conversion accounts for
the universal distribution of U D G (Duncan, 198 l), which
quickly eliminates uracil from DlqA by cleaving the N-gly-
cosidic bond between the base and the sugar-phosphate
backbone. The resulting apyrimidinic sites block replication
by D N A polymerases and are recognized and repaired in
vivo by other D N A repair enzymes (Friedberg et al., 1981).
U D G does not remove uracil from R N A or free nt, but only
from D N A (single- or double-stranded) longer than about
4 nt (Duncan, 1981),
EXPERIMENTAL A N D DISCUSSION
(a) Priueiples of the method
We hypothesized that if PCR products contained dU
instead of dT, incubation of such D N A with U D G would
render it unamplifiable in new PCRs. In contrast, target
DNA, i.e., sample D N A whose amplification is desired,
would not contain dU and would be unaffected by the
U D G and therefore would amplify normally. We have
modified PCR to produce D N A that contains dU. Incu-
bation of subsequent PCRs with U D G results in removal
of uracil from any contaminating carry-over P C R products,
blocking reamplification, while leaving natural D N A un-
affected.
(1) In one procedure, referred to as the d U T P protocol,
d U T P is substituted for dTTP during all PCRs, and the
products of these reactions contain many dUs throughout
their length (except the 5' ends, which are comprised of the
primers). U D G treatment of the fully pre-assembled start-
ing reactions prevents amplification of any contaminating
P C R products. U D G is inactivated at the heat-denatu-
ration steps in the normal temperature-cycling profde, so
that new PCR products, which contain many dUs, are
unaffected.
(2) In a second procedure, called the dU primer proto-
col, the PCR is performed with primers that contain dU,
which replaces dT during automated chemical synthesis.
Uracils in the P C R products are found only in the 5' ends
of the strands, but U D G treatment of such D N A is
sufficient to block amplification because the primers must
be copied by the Taq polymerase to allow re-priming at
successive cycles. In the dU primer protocol, reactions
must be reopened after the U D G has been heat-inactivated
so that the primers can be added, since the primers them-
selves are substrates for the enzyme. The d U T P protocol
controls carry-over contamination from all sources, in-

Fig,I,
A Fig, 2,

1 2 3 4 M 1 2 3 4 56 7 M 1 2 3 M

bp bp t)p
4369 4369 ~1369
4246 4 246 4 246

4123 ~1123 ~1123

Fig. I. Replacement of dTTP w/th dUTP in PCR reactions. PCRs contained I #M oligo primers ($'-GGTCGATGTATGTCTTGTTGand
5'-GTCTACGTGTGTGCTTrGTAC;Synthecell,Inc.), 200 ~M dNTPs (dATP, dCTP, dGTP, and either dTTP or dUTP), and Z$ units of TaqDNA
polymerase (Cetus/Perkin-Elmer) in 50 pl of25 mM Tris. HCI pH 8.3/5 mM M~21a/$0mM NaCI/0.01% gelatin under mineral oil. After 5 min at 94°C
and 30 cycles of I rain at 94°C, 2 rain at 55°C, and 3 rain at 72°C in a Cetus/Perkin-Elmer Thermal Cycler, aliquots of reactions were electrophoresed
on 2% agarose/2pg per ml EtdBr/40mM Tris. acetate pH 7.9/I mM EDTA gels. (Panel A)Target was I0 P8 of linearized plasmid HPVI6pTTI3 (a
full.length clone of the HPV type 16 genome,propagated in u~ + £. co//). Lanes I and 2, duplicate reactions containing dTTP; lanes $ and 4, duplicate
reactions containing dUTP.M, 123-bpladder size standards (Bethesda Research Laboratories). Bands migrating at 284 bp are amplification products.
Bands migratingbelow 123 bp are primer and primer dimer. (Panel B) Titration of input PCR product DNA (made with dUTP) into new PCRs containing
dUTP substituted for dTTP. Lane I, no target. Lanes 2 through 7 contained tenfold serial dilutions from I0- " - I 0 - te g ofdU-contalning 284-bp DNA.

Fig. Z Effect of UDG upon 'natural' target DNA, i.e., DNA not containing uracil. PCRs containing I0 pg of linearized plasmid HPVI6pTTI3 were
assembled, with dUTP replacing dTTP, and in the absence of UDG (lane I), the presence of UDG (5 ng; lane 3), or the presence of heat-inactivated
UDG (100°C, 30 min; lane 2). Followingincubation at 37°C for 15 min, thermal cyclingwas beam with a standard 5 rain 94°C heating step to inactivate
the UDG and denature the DNA. UDG was the generous gilt of Dr. J. van de Sande (Varshney et al., 1988). This UDG preparation (5 ng) released
5 pmol of uracil from DNA in 30 min at 37°C in 70 mM Hepes pH 7.8/I mM EDTA. For other details see legend to Fig. I.

cluding primers, enzyme, and nt, because the fully pre-
assembled amplification reactions are not opened after the
incubation with UDG.
(b) Substitution of dUTP for d T r P
The substitution of dUTP for dTrP in PCR was de-
monstrated for amplification of a 284-bp region of HPV
type 16. In one experiment, 10-pg quantities oflinear HPV
plasmid DlqA were amplified with either dUTP or dT~:P
(Fig. 1A). The extent of amplification was nearly identical.
In a second experiment, DNA products of a PCR reaction
were serially diluted and added to new PCR reactions to
mimic contamination; the new reactions contained dUTP
substituted for dTTP. Amplification of 10-16g of input
target to an estimated 5 × 10- s g (Fig. 1B, lane 7) after 30
cycles indicated that amplification near the theoretical
maximum (i.e., twofold per cycle) had been achieved. This
would not have been possible if incorporation of dUTP by
Taqpolymerase into the product DNA had been inefficient.
However, amplification of longer targets (1-2 kb) was less
efficient with dUTP than with dTTP.
(c) Effect of UDG on the PCR amplification
Incubation with UDG had minimal effect on the PCR
amplification of normal DNA (i.e., DNA that lacked
uracil). This was demonstrated by preincubating linearized
plasmid (10 pg) with UDG (Varshney et al., 1988) prior to
normal PCR and comparing the resulting amplification
with a reaction done without UDG treatment, or after heat
inactivation of the UDG prior to the addition of the DNA
(Fig. 2). A slight reduction of the intensity of the amplifica-
tion products was seen (lane 3 vs. lanes I and 2). However,
this experiment represented an extreme test of both the
purity of the UDG from nucleases and the inactivity of the
UDG on 'natural' DNA, since the target was the only DNA
in the reaction, and the weight of the UDG used (5 ng) was
Mabcde f g h i j kl mno
bp
869))
2461)
123))
Fig. 3. Control of contamination of PCRs with the dUTP protocol. PCR
amplification products (t0fg-10ng, estimated from agarose gels)
containing dU instead ofdT were added to new PCR to mimic accidental
contamination. Mixtures were incubated for 15 min at 37°C in the
presence or absence of 5 ng of UDG and taken through the normal PCR
temperature cycling (with dUTP). Lanes a through g and h through n
contained 10 ng, I ng, 100p g, 10p g, I pg, 100 fg, or 10 fg of dU-con-
taining PCR product, respectively. Lanes a through g, with UDG; lanes
h through n, no UDG; lane o, no DNA. For other details see legend to
Fig. 1.
127
M 1 2 3 4 5 6
bp
369m,
2461)
1231)
Fig. 4. Control ofmock carry-over contamination with the dUTP proto-
col in the presence of large amounts ofheterolngous DNA. PCR product
DNA containing dU was added to new PCR that contained $ #g of HeLa
DNA and incubated in the absence 0enes I-3) or presence 0anes 4-6)
of $ ng UDG. The amounts of dU-containing PCR product added were
2 ng (lanes I and 4), 20 pg 0anes 2 and S), or 200 fg (lanes $ and 6)~
Amplifications shown were done with dUTP. For other details see legend
to Fig.t.
5000 times that of the DNA. Normal PCR amplification of
408 bp of human globin sequence from placental DNA was
obtained -after incubation with UDG (5 ng) and substi-
tution of dUTP for dTrP (data not shown).
The efficiency of the dUTP protocol in preventing carry-
over contamination of PCR is illustrated in Fig. 3. The
284-bp product (10 fg) contaminating a new PCR resulted
in a band visible in an EtdBr-stained gel (lane n). However,
when the PCR products contained dU replacing dT, a
15-rain preincubation of each PCR with 5 ng of UDG
abolished amplification of 10 ng of contaminant (lane a).
This result showed that very few molecules of the added
dU-containing DNA survived the UDG treatment, since
0.1 fg of the 284-bp PCR product could be reamplified to
give a visible band (Fig. IB). UDG prevented re-amplifica-
tion of mock contaminants in the presence of large amounts
of heterologous DNA (Fig. 4).
(d) Uracil-containing primers
An alternative approach to control of contamination of
PCR is to incorporate uracil into th~ oligo primers. Because
each molecule of PCR product contains the synthetic
primer in the 5' end of each strand, and because the primer
sequence must be copied by the DNA polymerase to
achieve amplification, removal ofuracil from the 5' ends of
contaminating DNA should abolish amplification of the
contaminant. This is demonstrated in Fig. 5. A slight
amount of residual amplification occurred despite the
UDG treatment (lane 4). It is reasonable that inhibition of
re-amplification by UDG might be less efficient with uracils
only in the primers, since the number of target sites for the
UDG is much smaller than in the dUTP protocol. Uracil
probably must be present near the 3' end of each primer for
maximum UDG inhibition of re-amplification.
 128
1 2 3 4 5 6 M dU. Contamination with DNA containing dT cannot be
bp
4369
4246
4123
Fig. 5. Demonstration of the dU primer protocol. Lanes I and 2 show
amplification of the 284-bp HPVI6 sequence as above, using normal
primers (lane I), or primers in which the dTs had been replaced by dUs
(dU primers, Syntheceli, Inc.; lane 2). These PCR products (10 fg ofeach)
were added to new PCR in the presence or absence of UDG, but without
primers. After 15 min incubation at 37 °C, all reactions were heated at
94°C for 5 rain to inactivate the UDG, cooled, and the corresponding
primers were added for reamplification (i.e., dT primers were used to
reamplify lane I products, and dU primers were used to reamplify lane
2 products). Lane 3, dU primer products from lane 2, no UDG. Lane 4,
dU primer products from lane 2, with UDG. Lane S, dT primer products
from lane I, no UDG. Lane 6 dT primer products from lane I, with UDG.
For other details see legend to Fig. 1.
The dU primer protocol controls contamination from all
sources except the primers themselves (i.e., a contaminant
in the primers cannot be eliminated with UDG, since the
primers themselves contain uracils and would be degraded).
As a consequence, amplification reactions must be opened
for the addition of the dU primers after the UDG has been
incubated and inactivated, It should be noted that we have
found oligos containing uracil to be valuable reagents,
because the reactions they mediate (restriction cutting,
priming, etc.) can be specifically stopped with UDG treat-
ment, while leaving all other DNA and RNA in the
reactions unaffected.
(e) Conclusions
The dUTP protocol has the advantages of(/) eliminating
contamination from all sources (including contamination of
primers); (//) allowing the normal PCR temperature profle
to be followed without any re-opening of tubes, preceded
only by a short incubation to allow UDG to act on potential
contaminants; and (///) being relatively robust, in the sense
that tens or hundreds of uracils (sites for UDG action) will
be present in any contaminant molecule. The dUTP proto-
col thus requires only two changes in standard PCR: the
substitution of dUTP for dTYP in all reactions, and the
incubation of all PCR reactions with UDG prior to tem-
perature cycling. It should be emphasized that this
approach to controlling carry-over contamination of PCR
can be effective only if all amplification products contain
controlled with UDG treatment.
The presence of uracil should not affect most subsequent
uses of PCR products, including sequencing (dideoxy) and
hybridizations. However, we are not aware of any investi-
gation of the fidelity with which Taq polymerase incor-
porates dUTP in place of dTI'P during PCR. Until these
data are available, PCR product DNA containing dU
should be used cautiously. For cloning of PCR products
containing dU, the amplification products must be intro-
duced into an ung- (i.e., UDG-deficient) E. coil host
(Duncan et al., 1981)to avoid destruction of the amplified
DNA.
Any alteration in PCR products that allows discrimi-
nation against them prior to subsequent amplifcafions
could be used to control contamination problems. The use
of UDG as described here provides two approaches to
achieve this discrimination. The dUTP protocol described
here appears especially attractive because the enzyme
attacks many sites within any contaminating PCR product,
while at the same time it is inactive on both the reaction
components (nt and primers), and virtually any naturally
occurring DNA whose amplification is desired.
ACKNOWLEDGEMENTS
We thank J. van de Sande for his generous gift of UDG,
and S. Starr for expert assistance.
REFERENCES
Duncan, B.K.: DNA glycosylases. In Boyer, P. (Ed.), The Enzymes, 3rd
ed., part A. Academic Press, New York, 1981, pp. 565-586.
Friedberg, E.C., Bonura, T., Radany, E.H. and Love, J.D.: Enzymes that
incise damaged DNA. In Boyer, P. (Ed.), The Enzymes, 3rd ed., part
A. Academic Press, New York, pp. 251-2"/9.
Gibbs, R.A. and Chamberlain, J.S.: The polymerase chain reaction: a
meeting report. Genes Develop. 3 (1989) 1095-1098.
Kitchin, P.A., Szotyori, Z., Fromholc, C. and Almond, N.: Avoidance of
false positives. Nature 344 (1990) 201.
Kwok, S. and Higuchi, R.: Avoiding false positives with PCR. Nature 339
0989) 237-238,
Saiki, R.K., Oeifand, D.H., Stoffel, S., Scharf, SJ., Higuchi, R., Horn,
G.T., MulUs, K.B. and Erlich, H.A.: Primer-directed enzymatic
amplificationof DNA with a thermostable DNA polymerase. Science
239 (1988) 487-491.
Varshney, U., Hutcheon, T. and Van de Sande, J.H.: Sequence analysis,
expression, and conservation of Escherichia coli uracil DNA glyco-
sylase and its 8ene (ung). J. Biol. Chem. 263 (1988) 7776-7784.