Proc. Natl. Acad. Sci. USA
`Vol. 92, pp. 10859-10863, November 1995
`Biochemistry
`
`Catalytic editing properties of DNA polymerases
`
`(pucleotide analogues/termination/active site/mechanism)
`BRUNO CANARD*, BRUNO CARDONA*, AND ROBERT S. SARFATIt
`*Faculte de MWdecine, Unite de Recherche Associe-Centre National de la Recherche Scientifique 1462, 06107 Nice Cedex 2, France; and tlnstitut Pasteur,
`Unite d& Chimie Organique, 28, Rue du Dr. Roux, 75724 Paris Cedex 15, France
`
`Communicated by Charles C. Richardson, Harvard Medical School, Boston, MA, July 7, 1995 (received for review November 23, 1994)
`
`incorporation
`2',3'-
`of
`Enzymatic
`ABSTRACT
`dide-ixynucleotides into DNA results in chain termination. We
`report that 3'-esterified 2'-deoxynucleoside 5'-triphosphates
`(dNTPs) are false chain-terminator substrates since DNA
`polymerases, including human immunodeficiency virus re-
`verse transcriptase, can incorporate them into DNA and,
`subsequently, use this new 3' end to insert the next correctly
`paired dNTP. Likewise, a DNA substrate with a primer
`chemically esterified at the 3' position can be extended
`efficiently upon incubation with dNTPs and T7 DNA poly-
`merase lacking 3'-to-5' exonuclease activity. This enzyme is
`also able to use dTTP-bearing reporter groups in the 3'
`position conjugated through amide or thiourea bonds and
`cleave them to restore a DNA chain terminated by an amino
`group at the 3' end. Hence, a number of DNA polymerases
`exhibit wide catalytic versatility at the 3' end of the nascent
`DNA strand. As part of the polymerization mechanism, these
`capabilities extend the number of enzymatic activities asso-
`ciated with these enzymes and also the study of interactions
`between DNA polymerases and nucleotide analogues.
`
`DNA polymerases are multifunctional enzymes involved in the
`replication and repair of DNA (1). Understanding the poly-
`merase reaction mechanism might lead to major developments
`in modern molecular biology, medicine, and drug design. DNA
`polymerases, such as Escherichia coli DNA polymerase I, carry
`on.-the same polypeptide chain various enzymatic activities
`separated into different domains (2). Standard protein chem-
`istry aimed at elucidating the enzymatic mechanism has not
`been as rewarding for the DNA polymerases as for other
`enzyme families such as the protease (3). Photolabeling of E.
`coli DNA polymerase I has been reported with an azido-DNA
`(4), dTTP (5), and dNTP analogues (6, 7) and has identified
`several amino acid residues in the dNTP binding site. Crys-
`tallographic studies and site-directed mutagenesis have pro-
`vided insight into the structure-function relationship of poly-
`merases, such as the Klenow fragment of DNA polymerase I
`(8), human immunodeficiency virus (HIV) reverse tran-
`scriptase (RT) (9), T7 RNA polymerase (10), and rat DNA
`polymerase 3(11, 12). Crystal structure analysis of the 3'-to-5'
`exonuclease domain of the Klenow fragment has also sug-
`gested a related polymerization mechanism involving two
`metal ions (13).
`Thus, with sequence alignments (14), two crucial Asp res-
`idues have been pinpointed in the polymerase active site, but,
`to our knowledge, no direct evidence of their mode of action
`has been obtained to date. From the rat DNA polymerase 13
`structure, Davies et al. (11) suggested that Asp-190 could
`activate the 3'-hydroxyl end of DNA to produce a nucleophilic
`attack on the a-phosphate of the bound nucleotide, while
`Pelletier et al. (12) suggested Asp-256 for this role and
`proposed a detailed polymerization mechanism. Thus, despite
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement" in
`accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`refined structural knowledge of these enzymes, precise assign-
`ments of catalytic roles of amino acid side chains remain to be
`made. With these results in mind, we felt that a modification
`of the 3'-hydroxyl group that would retain an oxygen atom
`might drastically alter the catalytic behavior of the enzyme.
`However, the effect of a substitution at this 3' oxygen on the
`polymerization of nucleotides is not well documented, prob-
`ably because it was anticipated that incorporation of these
`analogues would not occur or would lead to mere chain
`termination and would, thus, be of limited value for the
`understanding of the mechanism. During the course of syn-
`thesizing bonified 3'-modified dNTPs (15-17), we found var-
`ious analogues unable to inhibit DNA synthesis. In the present
`report, we present evidence that DNA polymerases, including
`HIV-RT, exhibit catalytic versatility at the nascent 3' end of
`DNA by removing 3'-oxygen and 3'-nitrogen substituents. We
`propose that this capability might be part of the polymerization
`mechanism.
`
`MATERIAL AND METHODS
`Reagents. DNA oligonucleotides were synthesized at the
`Institut Pasteur (Paris) or Eurogentec (Brussels) and purified
`by two successive polyacrylamide gel electrophoreses. Modi-
`fied T7 DNA polymerase (Sequenase 2.0) was from United
`States Biochemical; Taq DNA polymerase was from Eurobio
`(Paris) or Boehringer Mannheim; HIV-RT was from Boehr-
`inger Mannheim. No significant 3'-to-5' exonuclease activity
`was detected after prolonged incubation with a 5'-32P-labeled
`primer/template.
`Chemical Synthesis of Nucleotide and Oligonucleotide An-
`alogues. The 2'-deoxy-3'-anthranyloyl-dNTPs (3'-ant-dNTPs)
`were synthesized as described (15). The 3'-amino-3'-deoxy-
`dTTP (3'-NH2-dTTP) and 3'-{N3-[3-carboxylato-4-(3-oxido-
`6-oxo-6H-xanthen-9-yl)phenyl]thioureido}-3 '-deoxythymi-
`dine 5'-triphosphate (3'-fluothioureido-dTTP) were synthe-
`sized as described (18). Details of the syntheses of 3'-deoxy-
`3'-(N-methylanthranyloylamino)thymidine 5'-triphosphate
`(3'-amd-dTTP) from 3'-azido-3'-deoxythymidine (AZT), 3'-
`O-[N6(N-methylanthranyl)amidohexanoyl]-dGTP (3 -chain-
`dGTP), and 3'-O-[N6(anthranyl)amidohexanoyl]-dATP (3'-
`chain-dATP) will be described elsewhere. The synthesis and
`use of 3'-O-[N6(N-methylanthranyl)amidohexanoyl]-dCTP
`(3'-chain-dCTP) are described (17). In all cases, successive
`HPLC purification procedures as described (17) gave homog-
`enous products (>99.99% pure). Chemical synthesis of 3'-ant-
`21-mer will be described elsewhere. Briefly, 5'-DMT-
`ATACTTTAAGGATATGTATCC-3' (where DMT is dimeth-
`
`Abbreviations: HIV, human immunodeficiency virus; RT, reverse
`transcriptase; FITC, fluorescein isothiocyanate; AZT, 3'-azido-3'-
`deoxythymidine; 3'-ant-, 3'-anthranyloyl-; 3'-fluothioureido-, 3'-{N3-
`[3-carboxylato-4-(3-oxido-S-oxo-6H-xanthen-9-yl)phenyl]thioure-
`ido}-; 3'-amd-, 3'-(N-methylanthranyloylamino)-; 3'-chain-dGTP and
`-dCTP, 3'-O-[N6-(N-methylanthranyl)amidohexanoyl]-dGTP and
`-dCTP, respectively; 3'-chain-dATP, 3'-O-[N6-(anthranyl)aminohex-
`anoyl]-dATP; dd-, 2',3'-dideoxy-.
`
`10859
`
`Illumina Ex. 1050
`IPR Petition - USP 10,435,742
`
`

`

`Proc. Natl. Acad. Sci. USA 92 (1995)
`
`such as a 3'-deoxy group (20). We have reported (15) that high
`concentrations of one 3'-modified analogue (3'-ant-dNTP)
`were required to get good incorporation yields. This was
`expected to preclude their use in conjunction with classical
`dNTPs because of the large discrimination ratio against them.
`Moreover, an unnoticed contamination of the 3'-ester ana-
`logue with its corresponding dNTP might lead to the favored
`incorporation of the latter if the former is not a substrate for
`the DNA polymerase. Consequently, we positioned a 6-carbon
`tether between the sugar ring and the fluorescent reporter
`group (Fig. la) (17). This allowed us to greatly decrease the
`analogue concentration to get the same incorporation yield
`with Sequenase. As reported (17), DNA chains terminated
`with a 3'-chain-dCMP showed two products on gel autoradio-
`grams. The upper product was more fluorescent than the lower
`after gel purification, but cross-contamination of the latter by
`the former was hard to assess (data not shown). Because the
`lower band comigrated with authentic 3'-dCMP-terminated
`DNA chain, it prompted us to test the possibility that free
`3'-OH groups had been generated during the incorporation
`reaction. This was done with a standard Sanger sequencing
`reaction in which dCTP had been replaced with 3'-chain-
`dCTP. We found that this analogue was unable to stop DNA
`synthesis at several dCMP insertion sites but produced a
`specific DNA chain-termination at others. This incorporation
`pattern was rather unexpected since dCITP and 3'-chain-dCTP
`are widely separated on HPLC chromatograms that yield
`analogue with no detectable accompanying dCTP (<0.01%).
`Moreover, putative contaminating dCTP would likely be used
`first by the polymerase, producing gradually more stops at
`dCMP insertion sites because of gradual dCTP pool exhaus-
`tion. This was clearly not the case (data not shown). Thus, it
`raised the possibility that, at least at specific sites in a cDNA
`sequence, esters at the nascent 3' end of the DNA were cleaved
`off and did not prevent chain extension.
`The Catalytic Editing of 3'-Esters by DNA Polymerases.
`When a mixture of all four 3'-ant-dNTPs was used, one
`product corresponding to multiple additions was observed
`with several DNA polymerases, such as Sequenase (Fig. 2) or
`HIV-RT. However, the very same analogue mixture gave rise
`to only one addition product of the expected size with Taq
`DNA polymerase. This clearly demonstrates that read-through
`was not due to a minute concentration of 3'-unprotected
`deoxynucleotides because one would expect a distribution of
`products larger than the primer rather than a single adduct.
`Mn2+-containing buffers (20) gave the high incorporation
`levels and thus were used for subsequent editing experiments
`
`T
`
`d
`
`H2N
`
`T
`
`0 -
`
`o
`NH-CC-
`RH~
`
`09P3-0
`
`b
`
`Structure of 3'-modified dNTP substrates. (a) 3'-chain-
`FIG. 1.
`dATP for R = H or 3'-chain-dGTP and 3'-chain-dCTP for R = CH3.
`(b) 3'-amd-dTTP (R = CH3). (c) 3'-fluothioureido-dTTP. (d) 3'-
`NH2-dTTP.
`
`10860
`
`Biochemistry: Canard et al.
`oxytrityl) was treated with N-methylisatoic anhydride under
`slightly alkaline conditions and the mixture was desalted. UV
`(254 nm)-absorbing fractions were treated with glacial acetic acid,
`lyophilized, and purified by reverse-phase HPLC. Unreacted
`21-mer, 3'-ant-21-mer, and 5'-DMT-21-mer were eluted at 9.9,
`11.72, and 14.97 min, respectively. The homogenous peak corre-
`sponding to 3'-ant-21-mer was recovered by lyophilization.
`Deprotection Assay of the 3'-Ant-21-mer. The 3'-ant-21-mer
`was annealed to its complementary template bearing (dG)5 or
`(dT)s protruding from the 5' termini under mild conditions (10
`min at 37°C and pH 7.3). Subsequent incubation (1 min at
`37°C) was at equimolar concentrations of primer/template and
`enzyme (from 50 to 500 nM) and [a-32P]dCTP or [a-33P]dATP
`at various concentrations. The reaction was stopped with 100
`mM EDTA, aliquots were spotted onto DE81 filter papers and
`washed three times in 2x SSC, and radioactivity was measured
`by liquid scintillation counting. Unprotected 21-mer was
`treated similarly and taken as 100% incorporation control.
`Aliquots were also run through a 15% denaturing polyacryl-
`amide gel, which was autoradiographed.
`Incorporation of Nucleotide Analogues into DNA by Using
`an End-Labeled Primer-Extension and Gel Assay. The stand-
`ing start reaction was performed as described (15, 19). A
`synthetic 5'-32P-labeled DNA primer/template (4 AM) made
`of 5'-ATACTTTAAGGATATGTATCC-3' annealed to 5'-
`TTTTTTTTTCGGATACATATCCTTAAAGTAT-3' (tem-
`plate C) was incubated with 3'-esterified dNTPs and 5 units of
`Sequenase or 0.5 unit of HIV-RT in 20 Al. The same primer/
`template was also used with 5 units of Taq DNA polymerase
`and 3'-esterified dNTPs at 45°C in Mn2+/citrate buffer (20).
`The nature of the base adjacent to the primer 3' end was varied
`depending on the assay (template G, A, T, or C) and is
`underlined above. Samples were electrophoresed through a de-
`naturing 15% polyacrylamide gel, which was autoradiographed.
`3'-Ester Displacement Monitored with an End-Labeled
`Primer-Extension and Gel Assay. The same primer/template
`C was incubated with 3'-ant-dGTP (2 mM) and 5 units of
`Sequenase for 3 min, 5 units of Taq DNA polymerase for 15
`min in Mn2+/citrate buffer, or 3'-chain-dGTP (1 mM) and 0.5
`unit of HIV-RT, as indicated. When needed, the reaction
`mixture was heat-treated for 15 min at 68°C, and the oligo-
`nucleotides were reannealed for 1 h in a waterbath from 68°C
`to room temperature. Chase experiments used dATP (200
`,uM), 3'-ant-dATP (2 mM), 3'-chain-dATP (1 mM), or 2',3'-
`dideoxyadenosine 5'-triphosphate (ddATP) (200 ,LM) and
`were incubated for 15 min at 37°C for Sequenase or HIV-RT
`or for 15 min at 45°C for Taq polymerase.
`Incorporation of 3'-amd-dTTP, 3'-NH2-dTTP, and 3'-
`fluothioureido-dTTP. The same 5'-P32-labeled primer was
`annealed to template A and incubated with 1 mM 3'-amd-
`dTTP, 1 mM 3'-fluothioureido-dTTP, or 100 ,uM 3'-NH2-
`dTTP and 5 units of Sequenase at 37°C or 5 units of Taq DNA
`polymerase at 45°C. When needed, the incorporation mixture
`was heat-treated for 15 min at 68°C, the oligonucleotides were
`reannealed as above, and 5 units of fresh Taq DNA polymerase
`was added with 200 ,uM dATP. When indicated, Taq DNA
`polymerase was added in the presence of 200 ,tM dATP.
`Labeling of the Incorporation Products with Fluorescein
`Isothiocyanate (FITC). The incorporation products were eth-
`anol-precipitated, dried under vacuum, and treated with FITC
`in NaHCO3 buffer exactly as described (18).
`
`RESULTS
`Incorporation of 3'-Modified Nucleotides into DNA. The
`sugar ring of nucleotide analogues can reportedly tolerate
`considerable structural modifications and still be a substrate
`for DNA polymerases. However, when present with classical
`dNTPs, nucleotide analogues can be discriminated several
`hundredfold when carrying modest chemical modifications
`
`

`

`Biochemistry: Canard et al.
`
`Sequenase
`3
`2
`4
`
`0 1
`
`5
`
`0
`
`6
`
`Taq
`7
`
`8
`
`_
`
`__p+l
`_-
`_.'.....- p
`End-labeled primer extension and gel assay using 3'-ant-
`FIG. 2.
`dNTPs and DNA polymerases. 3'-ant-dNTPs were as described (15).
`Lanes: 1-5, 3'-ant-dNTPs (400 ,uM) incubated with primer/template
`C and Sequenase for 1, 2, 3, 4, and 5 min, respectively; 6-8, the same
`3'-ant-nucleotide mixture (all four nucleotides, each at 2 mM) and Taq
`DNA polymerase for 5, 10, 15 min, respectively; 0, no enzyme. p,
`Primer (21-mer).
`
`(see below), but same results were obtained with 3'-chain-
`dNTPs at lower concentrations (2 ,uM, around the sensitivity
`limit of this assay, see below). When 3'-esterified nucleotides
`were preincubated with Sequenase in the absence of DNA,
`heated to inactivate the Sequenase, and then mixed with a
`DNA template/primer and Taq DNA polymerase and incu-
`bated, a single addition product was observed as before,
`indicating that the 3' esters were not hydrolyzed prior to
`incorporation and that the ternary enzyme-DNA-analogue
`complex was a prerequisite for the editing reaction.
`Catalytic Editing Is Dependent on the Next Correctly Paired
`Nucleotide. Thus, we synthesized chemically a 3'-esterified
`primer that mimicked an incorporation product as well as a
`putative substrate for 3' editing. This 3'-blocked primer (Fig.
`3a) was hybridized to its cognate template bearing a (dG)5 or
`(dT)5 protruding from the 5' terminus and incubated with
`Sequenase and increasing [a-32P]dCTP or [a-33P]dATP, re-
`spectively. Fig. 3b shows that this template/primer was able to
`be extended and that [a-32P]dCTP titrated the 3'-ant-primer/
`template, as expected for the reaction depicted in Fig. 3a.
`Incorporation of radioactive label onto the 3'-blocked primer
`a
`
`NHCH3
`
`0
`
`dClP*
`+
`5' - ATAC''r'AA(;(;ATA'T;TA'rTcc
`31' -I'AT(;AAAI-rTCCl'A l ACA 1 A(; (;(;(;1 (;(;t;
`
`t
`
`Proc. Natl. Acad. Sci. USA 92 (1995)
`
`10861
`
`relative to an unblocked primer was assessed by using a
`filter-binding assay and showed that the amount of label in Fig.
`3b, lane 9, accounted for at least 20% of the primer under these
`conditions. Comparatively, terminal deoxynucleotidyltrans-
`ferase did not incorporate any label into 3'-ant-21-mer.
`We then evaluated the ability of several 2'-deoxynucleotides
`and analogues to enhance the 3'-esterase-like activity intrinsic
`to the DNA polymerase once the DNA primer had been
`terminated by a 3'-esterified nucleotide. By using Sequenase or
`HIV-RT, we found that dATP, ddATP, 3'-chain-dATP, or
`3'-ant-dATP was able to displace the ester present at the 3' end
`of the dGMP incorporated previously (Fig. 4), whereas only
`dATP was able to do so when Taq DNA polymerase was used.
`These results indicated that correct positioning of the next
`nucleotide in the nucleotide binding site is important for full
`expression of this 3'-esterase-like activity and that Taq poly-
`merase did not accommodate ddNTPs well enough in its dNTP
`binding site to promote efficient removal of the ester. We
`conclude that Sequenase exhibited a strong 3'-esterase-like
`activity on the 3' end of DNA that allowed the use of
`3'-esterified DNA primer and the DNA polymerization to
`proceed in the presence of 3'-esterified dNTPs. Furthermore,
`the displacement of the 3' ester during the addition of the next
`correct nucleotide suggests that the 3' esterase and the poly-
`merase active sites may be the same or at least be near one
`another and work closely together.
`The 3'-Ester-Related Bonds Are Subjected to DNA Poly-
`merase-Mediated Hydrolysis. The 3' substituents with related
`3' bonds would be instrumental in gaining insight into this last
`point. Amides and thioureas are related to ester bonds but
`harder to hydrolyze. We reasoned that if the ester was sub-
`jected to a polymerization-mediated hydrolysis, positioning
`another carbonyl or thiocarbonyl group in a spatially similar
`position might lead to the same elimination should the ana-
`logue be incorporated. Therefore, we synthesized 3'-amd-
`dTTP (Fig. lb) from AZT and used it with Sequenase and Taq
`DNA polymerases in a similar fashion (Fig. 5). This analogue
`alone was not a substrate for Taq DNA polymerase (Fig. 5, lane
`1, and data not shown) unless the next correct nucleotide
`(dATP) was present in the reaction mixture (Fig. 5, lane 2).
`With Sequenase, a unique addition product with a slightly
`smaller Rf was observed (Fig. 5, lanes 3, 5, and 9), indicating
`that the chemical nature of the addition products was probably
`
`NHCH3
`-
`
`CO
`
`H
`
`0
`
`Seqsuelt
`
`5' - AIAL
`
`GI-I A1I(I 'AI c(;(.;(
`
`+PPi
`
`b
`
`+ a .32p-dCTp
`
`+
`
`mer
`4U_4IM -26
`.k * .
`-2 5
`-24
`-23
`-22
`-21
`
`C 1
`9
`8
`6
`3
`2
`7
`5
`4
`(a) Structure and deprotection reaction of 3'-ant-21-mer with Sequenase and [a-32P]dCTP. (b) 3'-ant-21-mer extension with Sequenase
`FIG. 3.
`and [a-32P]dCTP at various concentrations, as in a. Primer/(dG)5 template concentration was 50 nM. Lanes: C, kinase-treated control; 1-9, dCTP
`at 2.5, 5, 10, 25, 50, 100, 250, 500, and 1000 nM, respectively.
`
`

`

`10862
`
`Biochemistry: Canard et al.
`
`Proc. Natl. Acad. Sci. USA 92 (1995)
`
`Pol.
`T
`T
`S
`S
`S
`S
`T
`T
`S
`S
`S
`S
`H H H H H
`HT+Taq -- +
`+
`+
`+
`-
`d A dd - dAdd- d A dd- -
`Chase
`d dd C A
`-
`Lane
`0 1
`2
`3
`4
`5
`6
`7
`8
`0 13 14 15 16 17
`9 10 11 12
`
`-
`
`wa
`
`U
`
`q.
`
`. ~ ~
`
`~
`
`~
`
`~
`
`~
`
`_
`
`'
`
`~
`
`~
`
`3
`
`_* $rjA-P+2
`-
`
`_
`
`Displacement of 3' ester monitored with an end-labeled primer-extension and gel assay. Primer/template was incubated with
`FIG. 4.
`3'-ant-dGTP and Sequenase for 3 min (lanes 1-4 and 9-12), Taq DNA polymerase for 15 min in Mn2+/citrate buffer (lanes 5-8), or HIV-RT and
`3'-chain-dGTP for 15 min (lanes 13-17). For lanes 9-12, the reaction mixture was heat-treated and the oligonucleotides were reannealed. Reactions
`were chased with dATP (lanes 2, 6, 10, and 14), 3'-ant-dATP (lanes 3, 7, 11, and 17), 3'-chain-dATP (lane 16), or ddATP (lanes 4, 8, 12, and 15).
`After heat treatment (HT), a further 5 units of fresh Taq polymerase was added for lanes 10-12. P, primer (21-mer); A, 3'-ant-dATP; C,
`3'-chain-dATP; d, dATP; dd, ddATP; Pol., polymerase; S, Sequenase; T, Taq polymerase; H, HIV RT.
`
`different (Fig. 5, lane 4). Consequently, we treated part of the
`incorporation mixture with FITC, a fluorescent labeling re-
`agent for primary amines that does not react with DNA (21).
`This resulted in a large shift in mobility for the Sequenase
`product only (Fig. 5, lanes 6 and 11). Furthermore, the
`3'-amd-dTTP Sequenase addition product comigrated with
`the 3'-NH2-dTTP (Fig. ld) addition product (Fig. 5, lanes 9
`and 10), and both were converted to the same major product
`upon FITC treatment (Fig. 5, lanes 11 and 12). Fluorescence
`spectroscopy analysis also indicated that an anthranylic deriv-
`ative and a fluorescein derivative were present in the Taq and
`in the FITC-treated Sequenase addition products, respectively.
`With the same objective in mind, we tested 3'-fluothioureido-
`dTTP (18), an analogue on which a fluorescein molecule is
`attached to the 3' position through a thiourea bond (Fig. lc).
`Complete incorporation and removal of the 3' reporter group
`were observed with Sequenase (Fig. 6) as well as with HIV-RT
`(data not shown).
`Thus, Sequenase is able to incorporate 3'-amido and 3'-
`thioureido analogues and hydrolyze them leaving a 3'-amino-
`terminated DNA chain, whereas Taq DNA polymerase in the
`
`Pol
`dATP
`FITC
`
`T
`
`-
`-
`
`T
`+
`
`--
`
`S
`
`S,T
`+
`+
`
`S
`
`S
`
`.
`
`T
`+
`-
`
`T
`+.
`
`N
`N
`S S S S
`
`+
`
`+
`
`am.
`
`Ai
`
`X
`
`,
`
`_
`
`2-....
`
`....9
`
`.Am 4w,
`":".
`'im
`
`A.-40'
`
`presence of the next correct nucleotide utilizes 3'-amd-dTTP
`without hydrolyzing the 3'-amido group once incorporated.
`We conclude that Sequenase is also able to cleave amido and
`thioureido bonds at the nascent 3' end of DNA.
`
`DISCUSSION
`We have shown that nucleotide analogues with a bulky 3'-0-
`substituent are incorporated into DNA and block further
`incorporation of nucleotide analogues by Taq DNA poly-
`merase, and the fluorescent reporter group can be observed at
`the nascent 3' end of DNA. This is consistent with the fact that
`the 3' block can be released upon alkali treatment (15). The
`incorporation products observed here cannot be due to con-
`taminating deprotected dNTPs that would have been incor-
`porated in place of the analogue (i) because a single incorpo-
`ration product is obtained with Taq polymerase and a high
`concentration (2 mM) of the 3'-ant-dNTP mixture, whereas a
`distribution of products larger than P+1 (22-mer) would be
`expected with contaminating dNTPs (Fig. 2), (ii) because Taq
`polymerase would have used this new 3'-OH primer end upon
`secondary incorporation of putative contaminating dNTP, 3'
`ester, or ddNTP (Fig. 4), and (iii) because the "editing"
`efficiency is sequence-specific. More interestingly, we report
`that 3' esters do not absolutely block the ability of several
`
`Nucl.
`
`-
`
`N
`
`FITC:
`
`-
`-
`0 1
`
`N
`
`+
`2
`
`F
`
`-
`3
`
`F
`
`F
`
`-
`+
`4 5
`
`P
`
`-
`
`_
`
`--£9i..:'
`
`..
`
`woo
`
`a
`1
`3
`5
`4
`2
`0
`8
`7
`6
`9
`12
`11
`FIG. 5.
`Incorporation of 3'-amd-dTTP and 3'-NH2-dTTP. The
`5'-32P-labeled primer was annealed to template A and incubated with
`3'-amd-dTTP (lanes 0-9 and 11) and Sequenase (lanes 3-6, 9-12) or
`Taq DNA polymerase (lanes 1, 2, 7, and 8). For lane 4, the incorpo-
`ration mixture was heat-treated, the oligonucleotides were reannealed,
`and 5 units of Taq DNA polymerase was added with 200 ,uM dATP.
`For lanes 2, 3, 7, and 8, incorporation was performed in the presence
`of 200 ,uM dATP. For lanes 6, 8, 11, and 12, the incorporation products
`were treated with FITC. Arrows and arrowheads indicate the position
`of migration of the two incorporation products. P, primer (21-mer); S,
`Sequenase; T, Taq DNA polymerase; N, when 3'-amd-dTTP was
`replaced by 3'-NH2-dTTP in the incorporation reaction.
`
`10
`
`-P
`
`Incorporation of 3'-fluothioureido-dTTP and 3'-NH2-
`FIG. 6.
`dTTP with Sequenase and FITC treatment. Lanes: 0, no enzyme; 1
`and 2, 3'-NH2-dTTP for 5 min; 3-5, 3'-fluothioureido-dTTP for 1, 5,
`and 5 min, respectively. Postincorporative derivatization with FITC is
`in lanes 2 and 5. P, primer (21-mer); N, 3'-NH2-dTTP; F, 3'-
`fluothioureido-dTTP.
`
`

`

`Biochemistry: Canard et aL
`polymerases to synthesize DNA. Indeed, once a primer has
`been terminated with a 3' ester, addition of the next correctly
`paired dNTP leads to further extension by several DNA
`polymerases lacking 3'-to-5' exonucleolytic activity. Also, a
`chemically attached 3' ester can be removed upon incubation
`with dNTP, indicating that this "editing" activity is distinct'
`from pyrophosphorolysis and inherent with the DNA poly-
`merization mechanism. The incubation times (1 min at 37°C)
`used to deprotect the 3'-blocked primer make it unlikely that
`an unwanted exonuclease (not detected) would first act to
`remove the last 3'-esterified nucleotide before addition of
`labeled dCTP. Furthermore, identical results were obtained by
`using either labeled dATP or dCTP as chasing agents. Exo-
`nucleolytic cleavage of 3'-ant-dCMP, if any, would yield a
`20-mer primer that would be extended by dCTP and not dATP.
`We conclude that only the 3' ester has been removed to allow
`chain extension. The 3'-esterified 21-mer substrate is invalu-
`able in establishing that this 3' editing occurs at the DNA level
`but does not rule out that editing could also occur before or
`simultaneously with the phosphodiester bond formation, as
`shown with 3'-N-substituted analogues. Indeed, 3'-amido-
`dTTP and 3'-thioureido-dTTP are incorporated and edited
`quantitatively into 3'-amino-dTMP-terminated DNA chain by
`Sequenase in the absence of the next correct nucleotide
`(dATP). Interestingly, 3'-amido-dTTP is incorporated by Taq
`polymerase without being edited into 3'-amino only when the
`next correct nucleotide (dATP) is present, consistent with
`incorporation without edition of 3' esters by this enzyme.
`DNA polymerases are able to use modified nucleotides
`relative to their classical deoxynucleotide (dNTP) counter-
`parts (1, 15, 16, 20, 22, 23). The 3'-deoxy analogues, such as
`AZT (22), have been investigated mainly as inhibitors but have
`been of moderate value as active-site mapping probes. Cata-
`lano and Benkovic (24) reexamined the work of Abboud et al.
`(25) by using adenosine 2',3'-epoxide triphosphate and con-
`cluded that the strong inhibition observed was due to tight
`binding of the 2',3'-epoxide-terminated DNA chain without
`detectable covalent modification of the polymerase. The re-
`sults reported here may indicate that the polymerase-bound 3'
`end of DNA lies in the vicinity of a powerful nucleophilic
`group in the active site involved either in polymerization or in
`3'-substituent removal, a finding in accordance with crystal-
`lographic data for several DNA polymerases (8-12). "Read-
`through" (16), "quenching" (18), and results reported by
`others (23, 26, 27) may be explained by a quantitative removal
`of the fluorescent reporter upon enzymatic chain extension.
`These catalytic properties (that we call catalytic editing)
`raise several important questions. Indeed, these activities may
`be a fortuitous side reaction or have a biological meaning, i.e.,
`to cope with illegitimate 3' modifications that may occur in
`vivo. The 3'-hydroxyl group of a 2'-deoxynucleotide is its most
`chemically reactive functional group. Thus, it makes sense that
`DNA polymerases are able to cope with potential illegitimate
`3'-coupling that can stop DNA synthesis. Interestingly, 3'-
`phosphates are potent blocks to DNA polymerization (1). This
`may reflect the fact that a phosphate cannot receive a nucleo-
`philic attack unless it is precisely ligated to a metal ion and that
`enzymes exist in vivo that can remove phosphates at 3' termini
`(1). Several retroviral RTs, including that of HIV, also utilize
`tRNA primers (28) that may be acylated with amino acids
`whose structures resemble the anthranyloyl substituents used
`here. The HIV-RT capability that we demonstrated above may
`be relevant to the initiation of reverse transcription during
`early retroviral life cycle. Other substituents at the 3' end of
`DNA at such a position and distance of the sugar ring may also
`receive a nucleophilic attack. Strikingly, the central nitrogen of
`the 3'-azido group of AZT bears a partial positive charge (29)
`in a position similar to the carbonyl group of 3'-esterified
`nucleotides, and the 3'-azido group of AZT is metabolized into
`a 3'-amino group in vivo (30). Finally, this also identifies a
`
`Proc. Natl. Acad. Sci. USA 92 (1995)
`
`10863
`
`position to be modified for DNA polymerase substrate ana-
`logues that does not prevent incorporation into DNA.
`
`We thank S. Wain-Hobson, I. Varlet, P. Glaser, P. Marliere, and F.
`Hecht for helpful discussion. Excellent technical expertise of C.
`Guerreiro, L. Schmitt, A. Namane, and J. Uguetto-Monfrin is grate-
`fully acknowledged. This work was supported by the Groupement
`d'ttude et de Recherche sur les Genomes, the Institut Pasteur, and
`Boehringer Mannheim.
`
`1.
`
`2.
`
`3.
`
`4.
`
`5.
`
`6.
`
`7.
`
`8.
`
`9.
`
`10.
`
`11.
`
`12.
`
`13.
`14.
`
`15.
`16.
`
`17.
`
`18.
`
`19.
`
`20.
`
`21.
`
`22.
`
`23.
`
`24.
`
`25.
`
`26.
`
`27.
`
`28.
`
`29.
`
`30.
`
`Kornberg, A. & Baker, T. A. (1992) in DNA Replication, eds.
`Kornberg, A. & Baker, T. A. (Freeman, San Francisco), pp. 408,
`446-449.
`Klenow, H. & Henningsen, I. (1970) Proc. Natl. Acad. Sci. USA
`65, 168-175.
`Fersht, A. (1985) in Enzyme Structure and Mechanism, ed. Fersht,
`A. (Freeman, New York), pp. 248-260.
`Gibson, K. J. & Benkovic, S. J. (1987) Nucleic Acids Res. 15,
`6455-6467.
`Pandey, V. N., Williams, K. R., Stone, K. L. & Modak, M. J.
`(1987) Biochemistry 26, 7744-7748.
`Pandey, V. N. & Modak, M. J. (1988) J. Biol. Chem. 263, 6068-
`6073.
`Joyce, C. M., Ollis, D. L., Rush, J., Steitz, T. A., Konigsberg,
`W. H. & Grindley, N. D. F. (1986) UCLA Symp. Mol. Cell. Biol.
`32, 197-205.
`Ollis, D. L., Brick, P., Hamlin, R., Xuong, N. G. & Steitz, T. A.
`(1985) Nature (London) 313, 762-766.
`Kohlstaedt, L. A., Wang, J., Friedman, J. M., Rice, P. A. & Steitz,
`T. A. (1992) Science 256, 1783-1790.
`Sousa, R., Chung, Y. J., Rose, J. P. & Wang, B. C. (1993) Science
`364, 593-599.
`Davies, J. F., Almassy, R. J., Hostomska, Z., Ferre, R. A. &
`Hostomsky, Z. (1994) Cell 76, 1123-1133.
`Pelletier, H., Sawaya, M. R., Kumar, A., Wilson, S. H. & Kraut,
`J. (1994) Science 264, 1891-1903.
`Beese, L. S. & Steitz, T. A. (1991) EMBO J. 10, 25-33.
`Delarue, M., Poch, O., Tordo, N., Moras, D. & Argos, P. (1990)
`Protein Engin. 3, 461-467.
`Canard, B. & Sarfati, S. (1994) Gene 148, 1-6.
`Metzker, M. L., Raghavashari, R., Richards, S., Jacutin, S. E.,
`Civitello, A., Burgess, K & Gibbs, R. A. (1994) NucleicAcids Res.
`22, 4259-4267.
`Sarfati, R. S., Berthod, T., Guerreiro, C. & Canard, B. (1995) J.
`Chem. Soc. Perkin Trans., in press.
`Herrlein, M. K., Konrad, R. E., Engels, J. W., Hollertz, T. &
`Cech, D. (1994) Helv. Chim. Acta 77, 586-596.
`Boosalis, M. S., Petruska, J. & Goodman, M. F. (1987) J. Biol.
`Chem. 262, 14689-14696.
`Tabor, S. & Richardson, C. C. (1989) Proc. Natl. Acad. Sci. USA
`86, 4076-4080.
`Smith, L., Fung, S., Hunkapiller, M. W., Hunkapiller, T. J. &
`Hood, L. E. (1985) Nucleic Acids Res. 13, 2399-2412.
`Mitsuya, H., Weinhold, K. J., Furman, P. A., Clair, M. H. S.,
`Lehrman, S. N., Gallo, R. C., Bolognesi, D., Barry, D. W. &
`Broder, S. (1985) Proc. Natl. Acad. Sci. USA 82, 7096-7100.
`Kutateladze, T. V., Kritzyn, A. M., Florentjev, V. L., Kavsan,
`V. M., Chidgeavadze, Z. G. & Beabealashvilli, R. S. (1986) FEBS
`Lett. 207, 205-212.
`Catalano, C. E. & Benkovic, S. J. (1989) Biochemistry 28, 4374-
`4382.
`Abboud, M. M., Sim, W. J., Loeb, L. A. & Mildvan, A. S. (1978)
`J. Biol. Chem. 253, 3415-3421.
`Chidgeavadze, Z. G., Beabealashvilli, R. S., Atrazhev, A. M.,
`Kukhanova, M. K., Azhayev, A. V. & Krayevsky, A. A. (1984)
`Nucleic Acids Res. 12, 1671-1686.
`Chidgeavadze, Z. G., Beabealashvilli, R. S., Krayevsky, A. A. &
`Kukhanova, M. K. (1986) Biochim. Biophys. Acta 868, 145-152.
`Weiss, R., Teich, N., Varmus, H. & Coffin, J. (1982) Molecular
`Biology of Tumor Viruses: RNA Tumor Viruses, Part 1 (Cold
`Spring Harbor Lab. Press, Plainview, NY), 2nd Ed.
`Treinin, A. (1971) in The Chemistry oftheAzido Group, ed. Patai,
`S. (Wiley-Interscience, New York), pp. 1-55.
`Cretton, E. M., Xie, M.Y., Bevan, R.J., Goudgaon, N. M.,
`Schinazi, R. F. & Sommadossi, J. P. (1991) Mol. Pharmacol. 39,
`258-266.
`
`