Gette. 99 (1991) 105-108

105

Elsevier

GENE

03957

Deletion mutagenesis during polymerase chain reaction: dependence on DNA polymerase (Recombinant

DNA;

hairpin;

ribosomal

RNA)

Neal F. Cariello”*b, William C. Thilly”, James A. Swenberg”,’ and Thomas R. Skopek” ” Pathology Department, U~live~sit~v ofNorth Carolina at Chapel Hf1’1,Chapel Hill, NC 27599 (U.S.A.); h Integrated Toxicology Program, Duke ~ni~~er~~t~, ~urharn* NC 27710 (U.S.A.): and ’ Center &orE~iviro~ment~l Health Sciences, Massachusetts Institute of Teehno~o~, El&666, ~an~bridge. h4A 02139 (U.S.A.) Tel. (61 T/253-6220 Received by D.T. Denhardt: 7 September Revised: 3 October 1990 Accepted: 12 November 1990

1990

SUMMARY

Polymerase chain reaction (PCR) was performed with two polymerases, 7”ermus aquaticus DNA polymerase (Taq), and modified T7 DNA polymerase (Sequenase’“‘). Both polymerases were used to amplify the same portion of the human 18s rRNA gene. We report a PCR artifact, namely a deletion of 54 bp, when Taq polymerase was used to amplify a portion of the human IKS rRrlA gene. PCR performed with Sequenasetm did not produce this artifact. The deletion eliminated a potenti~ly stable hairpin loop. Our data are consistent with the following model for generation of the deletion: (i) the formation of an intrastrand hairpin, and (ii) polymerization across the base of the hairpin, thus deleting the nucleotide sequence in the hairpin. Furthermore, we show that the deletion occurs mainly during synthesis of the ( - )DNA strand. Our observations suggest that similar artifacts may occur in other sequences containing stable secondary structures.

INTRODUCTION

Polymerase chain reaction (PCR) (Saiki et al., 1985; Mullis and Faloona, 1987) has found numerous applications in molecular biology. Different DNA polymerases have been used (Saiki et al., 1985; Keohavong and Thilly, 1989), but typically the thermostable polymerase from Thermus aquaticus (Tag) (Saiki et al., 1988) is employed. Taq polymerase has a temperature optimum of about 75”C, and the elevated temperature used with the enzyme has reduced (i) the unwanted amplification of nontarget C~r~e~~onde~~e to: Dr.

N.F.

Cariello,

No. 7095, Glaxo

Research

Carolina,

Hill, NC 27599 (U.S.A.)

Chapel

Abbreviations:

Laboratory,

bp, base pair(s);

Pathology

dd, dideoxy;

or 1000 bp; nt, nucleotide(s);

reaction;

ribosomal

RNA-encoding

EtdBr, ethidium PCR,

0

IYYt

Elsevier

Science

bromide;

polymerase

chain

gene; Taq, Thermus aquaticus;

u, unit(s).

037X-i 1 tY,‘Yl:$03.50

CB

of North

EXPERIMENTAL

AND DISCUSSION

Tel. (919)966-6143.

kb, kilobase &VA,

Department,

Rm. 142, University

nt sequences in PCR (Saiki et al., 1988) and (ii) the problem of sequencing through G + C-rich tracts (Innis et al., 1988). Although G + C-rich tracts may have the potential to form stable hairpin structures, these hairpins should be less likely to form at the elevated temperatures used with Taq. We wished to amplify a segment of the human 185 rRNA-encoding gene (185’ rRNA) with Taq polymerase. However, Taq polymerase deleted a section of the 183 rRNA molecule. We present a series of experiments designed to investigate this PCR artifact, and a method to amplify the 18s rRNA sequence without deletion.

Publishers

R.V.

(a) PCR conditions Plasmid pB, containing most of the human 185 rRNA sequence, was a gift from Dr. Iris Gonzalez (Hahnemann University, Philadelphia, PA). PCR amplification of line-

106 arized plasmid pB with Sequenasetm (United States Biochemical, Cleveland, OH) using primers 18S-UP and 18S-DOWN (Fig. 1) produced a single product of the expected size. After 15 cycles, a single band of the expected size, 389 bp, was visible on an EtdBr-stained gel (10 ~1 loaded on gel). The remainder of the reaction mixture was loaded onto a polyacrylamide gel, the 389-bp band was excised from the gel without exposure to EtdBr or ultraviolet light, and the DNA was recovered by electroelution. The gel-purified 389-bp fragment produced with Sequenasetm was used as the DNA template for all amplifications with Tag polymerase. This ensured that no COampli~cation from non-~8s r&VA sections of the plasmid occurred. Subsequent sequence analysis of the template produced with Sequenasefm confirmed that it was fulllength and contained the 18s rRNA sequence. Conditions for the first set of experiments with Tuq polymerase were: 50 ~1 total volume/l PM primer I8S-DOWN and 18S-UP/SO or 250kiM dNTPs (Sigma, St. Louis, MO)~Perkin-Elmer Cetus Taq buffer (1 x = 10 mM Tris. HCl pH 8.3i.50 mM KClj1.5 mM Mg~l~/O.OOl “/b gelatin; Perkin-Elmer Cetus; Norwalk, CT)/varying MgCl, concentrations of 1.5, 3.0, 6.0 or 10 mM/5 x lo8 copies of fulllength template/2.5 u of Taq DNA polymerase (PerkinElmer Cetus). One cycle consisted of 1 min at 95’ C, 2 min at 53 o C, and 2 min at 70°C. After 17 cycles, 10 ~1 was analyzed on a poIyacrylanlide gel, and two bands were visible with EtdBr staining. One band was of the expected size and the other

PRIMER 18S-UP AAGCTCGTAGTTGGATCTTG GTT-GCTCGTAGTTGGA PRIMER 18S-UP-V4 5’ 661 3’

701

CTGCA~TT~G~TCGTAGTTGGATCTTgggaqcgggC --_-_____+_________+-__--_-_-+_____-___+ GACGTCAATTTTTCGAGCATCAACCTAGAAccctcgcccg gggcggtccgccgcgaggcgagccaccgcccgtccccgcc _________+_------__+_________+_____-___+ cocgccaggcggcgctccgctoggtggcaggggcgg

741

c~CCTCTCGGCGCCCCCT~GATGCTCTTAG~TGAGTG...~~ -------__+-________+_-___--__-t_-_-______+ ggaacGGAGAGCCGCGGGGGAGCTACGAGAATCGACTCAC...//

1021

ATTAATCAAGAACGAAAGTCGGAGGTTCGAAGACGATCAG -________+_--------+-________+____-----+

(+)

TAATTAGTTCTTGCTTTCAGCCTCCAAGCTTCTGCTAGTC

(-)

AGCCTCCAAGCTTCTGCTAG PRIMER 18S-DOWN Fig.

1. Sequence

ofthe human I8SrfuvA

used for PCR. Primer sequences

are given above and below the sequence. The lower-case the sequence

deleted by Taq polymerase.

4 nt identical numbering

to the 3’ 4 nt of primer

are from locus

have been omitted

HUMRRN18S

to save space.

letters represent

The underlined letters represent 18%UP.

The nt sequence

in GenBank;

and

bp 781-1020

was about 50 bp smaller (data not shown).

In all cases, the

relative intensities of the bands were similar. Varying the MgCl, and the dNTP concentrations did not reduce the amount of the shorter fragment. Raising the annealing temperature to 60 or 64°C did not eliminate the production of the shorter fragment. The template for Taq polymerase was the full-length 389-bp product, thus, the smaller-ill, product observed was not due to co-amplified sequences from another portion of the plasmid. (b) Sequence of truncated PCR product produced by Tug polymerase The smaller fragment was excised from a polyacrylamide gel and sequenced using fluorescent primers and an Applied Biosystems 370A DNA sequencer. Asymmetric PCR was performed with unequal primer concentrations to produce a single-strand template that contained a sequence complementary to the fluorescent primers. dNTPs were removed using a Centricon 30 microcon~entrator (Amicon, Danvers, MA). Fluorescent primers were hybridized to the singlestrand template and extended with Sequenase’“’ in the presence of the appropriate ddNTP. Only one strand was sequenced. The truncated fragment contained a 54-bp deletion (Fig. 1). One explanation for the deletion was that primer 18S-UP was simply hybridizing at an undesired site on the template. The four nt at the 3’ end of primer lXS-UP are complementary to four bp at the 3’ end of the deletion (Fig. l), and primer hybridization and DNA polymerization at this site could cause the observed deletion. (t) Truncated PCR product was also produced using a different primer To test the possibility that primer 18.SUP was hybridizing to an unwanted site, a second primer, 18S-UP-V4, which does not have homology to nt sequences at the 3’ end of the deletion was used (Fig. 1). This was the second set of experiments with Tug polymerase; the conditions for amplification are given in the legend to Fig. 2. However, the shorter PCR product still appeared and, in some cases, it proved to be the only product (Fig. 2, lanes 4, 6, and 8). Sequenasetm produced only the full-length product (Fig. 2, lane 1). Under certain conditions, Taq polymerase produced multiple bands, some of which appeared to be larger than the expected size (Fig. 2, lane 3). The larger bands (i) may contain nt sequences added by Taq polymerase, or (ii) may look larger because of a salt effect since the Sequenase’“’ and Taq buffers have different concentrations of salt. The truncated DNA fragment in lane 8 was excised from the gel and sequenced as described above. The fragment contained the same 54-bp deletion that was produced with

107

cg

a

9

c-g c--Q

g--c c--Q c a t l g g--c o9--c 9 g--c g---c c---Q g--c g---c g--c c--Q a9 l 9 9 l 9 l

I23456789 Fig. 2. EtdBr-stained Tuq polymerase

polyacrylamide

gel showing

PCR performed

with

and Sequenase’“.

Primers

l&S-UP-V4 and ISS-DOWN

were used;

10% of the reaction

mixture

was loaded

acrylamide

gel and

Tris

run

for 3 h at

I borate12 mM EDTA). The Sequenase’”

100 ~1 and contained: copies

boiling

water,

addition

of

conditions

of plasmid

45 s at room

concentrations

(Pharmacia,

(Perkin-Elmer 1, PCR

of

of

1min in

water

18S-DOWN~vary~ng

PM dNTPs/l.5

of

MgCI,.

2-8,

mM MgCI,;

PM dNTPs/3.0

9, pBR322/MspI

l

GCCTCT

dNTP

of full-length

Tuq polymerase

FM dNTPsj4.5

mM MgCI,:

One

Fig. 1). Polymerization

the nt deleted

by the arrow)

for the observed

by Taq polymerase

deletion. during

Lower-case

across the base of this hairpin structure

could produce

the observed

letters

PCR (also shown

in

(indicated

deletion.

FM mM

6, 0.25 PM

7, 1.0 pM primers/200

8, 0.25 pM primers/200

indicate

2, 0.2 PM

3, 1.0 FM primers/200

FM dNTPs’3.0 mM MgCl,;

Fig. 3. Possible mechanism

(2 u)

and 2 min at 72°C.

Taq-polymerase;

4, 0.25 PM primers]200

5, 1.0 PM primers/200 mM MgCI,;

ATCTT

cccc t ccccg t t

100 mM solution pH 7.5)/l x Perkin-Elmer

Sequenase’“;

mM MgCi,;

bath,

at 37” C. The

were: 100 ~1 volume/varying

I min at 94°C. 2 min at 53°C

using

dNTPs/4.5

MgCI,;

and

given in section a)/lO’ copies

concentrations

primers/50

dNTPsjl.5

15 s in a 37°C

mM of

NJ)/1 pM

Cetus) was added and 35 PCR cycles were performed.

cycle consisted

MgCI,;

amplification

18S-UP-V4

buffer (composition

primers/200

One cycle consisted

temperature,

for Tuq polymerase of primers

Lanes:

DNA.

Piscataway,

1u of enzyme, vortexing, and 2 min incubation

template/varying

(89 mM

volume was

10 mM Tris . HCI pH 8.0/S mM MgC1,/2.25

concentrations Cetus

buffer

amplification

each dNTP (100 mM solution pH 7.5, Pharmacia, primers/IO’

on an 8% poly-

150 V with TBE

a

FM dNTPsjl.5

fiM mM

marker.

the initial primer, 18S-UP. Thus, the deletion produced by Tag polymerase was not due to unwanted primer hybridization at a second site since the same deletion occurred with two different primers. (d) The deleted sequence is contained in a potential hairpin The human 18 rRNA sequence contains a great deal of potential secondary structures, which are thought to be important for the functioning of the 18s rRNA in the ribosome. The portion of the 18s rRNA sequence deleted by Taq polymerase (Fig. 3) is contained in a hairpin structure as proposed by Gonzalez and Schmickel (1986). Our results with Tuq polymerase may be explained by the formation of an intrastrand hairpin and the polymerase incorrectly reading across the base of the hairpin, thus deleting the hairpin. This model has been suggested by Glickman and Ripley (1984) based on analysis of E. cc& deletion mutants. (e) Deletion occurs primarily during synthesis of one strand The deletion could be occurring during synthesis of one or both strands. To test these possibilities, full-length 18s

DNA was used as the PCR template in the presence of radiolab~led nt and a single radiolabeled primer. When only the upstream primer was used in the PCR, most of the radioactivity appeared in a single band in the autoradiogram (Fig. 4), indicating that mostly full-length product was produced. However, when the downstream primer was used, in addition to the full-length product, two intense discreet smaller-ll/i, bands were visible, representing DNA with deleted sequence. These results suggest that deletion occurred primarily when the ( + )strand was used as the DNA template. The deleted sequence is located at the 5’ end of the ( + )strand, distal to the primer hybridization site. Since Taq polymerase synthesizes DNA at a rate of about 60 bp/s (Innis et al., 1988), it should take 5-6 s for the polymerase to reach the deleted sequence. During this time, an intrastrand hairpin could form and Taq could misread across the base of the hairpin. Deletion did not occur when the ( - )strand was used as the DNA template. In this case, primer hybridization occurred immediately adjacent to the potential hairpin, and Taq may have synthesized through this portion of the gene before the hairpin has formed. It should also be noted that the base-pairing potential of the hairpin in the ( - )strand is less than the base-pairing potential of the hairpin in the ( + )strand. In the ( + )strand, G : T hydrogen bonding can occur (Fig. 3); however, in the corresponding ( - )strand hairpin, G : T mismatches will be replaced by C: A mis-

108 used only the standard Perkin-Elmer Cetus buffer; it may be possible that the deletion would not occur in other buffer systems. Researchers attempting to use PCR to amplify sequences with a great deal of potential secondary structure may

experience similar difficulties with Tuq polymerase. Deletions occurring (i) using ‘inverse PCR’ (Triglia et al., 1988; Ochman et al. 1988) where DNA of unknown sequence is amplified, or (ii) using standard PCR to amplify across an unknown internal sequence could prove particularly troublesome.

Fig. 4. Autoradiogram

showing

from ten PCR cycles DOWN:

primer

18S-DOWN

The single strong full-length DOWN

DNA primer

band

(nt sequence

products

UP: primer

of primers

seen with the UP primer

was produced. indicates

Primers were end-labeled

the amplification

with a single primer.

The multiple

is given in Fig. I). shows

bands

that DNA with deleted

ACKNOWLEDGEMENTS

produced 18S-UP-V4; that

with [y-3ZP]ATP (New England

3000

reaction

gM

was

IOOfil and

copies of full-length

contained:

template/50

800 Ci/mmol)/l

x

1 PM

England

Nuclear,

given in section a)/2 u Taq polymerase

Ten cycles were performed porated

formamide acrylamide/7

loading

(Amicon,

Perkin-Elmer

Cetus

buffer

(Perkin-Elmer

(comCetus).

as given in the legend to Fig. 2. Unincor-

nt and primers were removed

30 Microcentrator

primer/200

PCi of [G(-32P]dATP (New

position

Danvers,

by centrifugation

REFERENCES Glickman, B.W. and Ripley, L.S.: Structural intermediates of deletion mutagenesis: a role for palindromic DNA. Proc. Nat]. Acad. Sci. USA 81 (1984) 512-516. Gonzalez,

I.R. and Schmickel,

gene: evolution

in

sequencing

onto an 89~ poly-

sequencing

M urea gel. The gel was dried and exposed to Kodak XAR5

x-ray lilm

Acad.

ofpolymerase

D.H. and Brow, M.A.D.: polymerase

RNA

chain reaction-amplified

DNA

and direct

DNA. Proc. Nat].

P. and Thilly, W.G.: Fidelity of DNA polymerases Proc. Nat]. Acad.

H., Gerber,

and Arnheim, sequences anemia.

and restriction Science

Mullis,

amplification

Genetics

applications

of an

120 (1988) 621-623.

F., Mullis, K.B., Horn, G.T., Erlich, H.A.

N.: Enzymatic

Saiki, R.K., Gelfand,

D.L.: Genetic

chain reaction.

Saiki, R.K., Scharf, S., Faloona,

in DNA

Sci. USA 86 (1989) 9253-9257.

A.S. and Hart],

inverse polymerase

G.T.,

18s ribosomal

Sci. USA 85 (1988) 9436-9440.

Keohavong Ochman,

K.B., Gelfand,

Thermus aquaticus DNA

with

amplification.

matches, which have little or no hydrogen-bonding potential. Reduced hydrogen-bonding potential in the hairpin, as well as proximity to the primer, may explain why deletion did not occur when the ( - )strand was used as the DNA template. We found it surprising that Tuq polymerase produced the deletion, since the 72” C polymerization temperature would be expected to destabilize potential secondary structures. Both Sequenasetm and Taq polymerase are reported to be highly processive, synthesizing thousands of bases before disassociating from the DNA (Tabor and Richardson, 1987; Innis et al., 1988) so differences in polymerase processivity are unlikely to explain our results. We have

R.D.: The human

and stability. Am. J. Hum. Genet. 38 (1986) 419-427.

Innis, M.A., Myambo,

using a Centricon

MA). DNA was resuspended

buffer, boiled 3 min, and loaded

ES-070-31-13.

was made.

Nuclear,

The

volume

in part by NRSA

visible with the

sequence

Ci/mmol) and T4 kinase to a specific activity of l-3 x 10” cpm/pmol. dNTPs/lO’

NFC was supported

mainly

amplification

site analysis

of beta-globin

for diagnosis

genomic

of sickle cell

230 (1985) 1350-1354. D.H., Stoffel, S., Scharf,

K.B.

and

Erlich,

H.A.:

S.J., Higuchi,

R., Horn,

Primer-directed

of DNA with a thermostable

enzymatic

DNA polymerase.

Science

239 (1988) 487-491. Tabor,

S. and Richardson,

bacteriophage

CC.: DNA sequence

T7 DNA polymerase.

analysis with a modified

Proc. Nat]. Acad. Sci. USA 84

(1987) 4767-4771. Triglia,

T., Peterson,

amplification known

M.G. and Kemp,

of DNA segments

sequences.

D.J.: A procedure

that lie outside

for in vitro

the boundaries

Nucleic Acids Res. 16 (1988) 8186.

of