CA2911712C - Pyrophosphorolysis activated polymerization (pap) - Google Patents

Abstract

A novel method of pyrophosphorolysis activated polymerization (PAP) has been developed. In PAP, pyrophosphorolysis and polymerization by DNA polymerase are coupled serially for each amplification by using an activatable oligonucleotide P* that has a non- extendibIe 3'-deoxynucleotide at its 3' terminus. PAP can be applied for exponential amplification or for linear amplification. PAP can be applied to amplification of a rare allele in admixture with one or more wild-type alleles by using an activatable oligonucleotide P* that is an exact match at its 3' end for the rare allele but has a mismatch at or near its 3' terminus for the wild-type allele. PAP is inhibited by a mismatch in the 3' specific sequence as far as 16 nucleotides away from the 3' terminus. PAP can greatly increase the specificity of detection of an extremely rare mutant allele in the presence of the wild-type allele. Specificity results from both pyrophosphorolysis and polymerization since significant nonspecific amplification requires the combination of mismatch pyrophosphorolysis and misincorporation by the DNA polymerase, an extremely rare event. Using genetically engineered DNA polymerases greatly improves the efficiency of PAP.

Description

F..\=

TITLE OF .THE INVENTION
.PYROPHOSPHOROLYSIS ACTIVATED POLYMERIZATION (PAP) =
BACKGROUND OF THE INVENTION
[0001] This invention relates to nucleic acid polymerization and amplification. In particular, it relates .to a novel and general method for nucleic acid amplification, in which pyrophosphorolySis and polymerization are serially-coupled. The method has been adapted for allele-specific amplification and can greatly increase the specificity to detect an extremely rare allele in the presence of wild-type alleles. We refer to the method as pyrophosphorolysis ió activated polymerization (PAP).
[00021 The publications and other materials used herein to illnminate the background of the invention or provide additional details respecting the practice are for convenience respectively grouped in the appended Bibliography.
[0003] Multiple methods for detecting mutations present in less than 10% of cells (i.e. rare is alleles) have been developed, including PCR amplification of specific. alleles. (PASA), peptide nucleic acid (['NA) clamping blocker PCR,.. allele-specific competitive bIocker PCR, mismatch amplification mutation assay (MAMA), restriction fragment-length-polymoiphism (RFLP)/PCR
(Parsons and Heflich, 1997) and QE-PCR. (Ronai and Min. amoto, 1997). These methods: i) amplifY the rare allele. selectively, ii) destroy the abundant wild-type allele, or spatially 20 separate the rare allele frOm the wild-type allele. The specificity achievable under typical research/clinical conditions is 10-3 (Parsons and Heflich, 1997), although a few publications = reported higher specificity: of detection (Pourzand and Cemtd, 1993;
Knoll et al., 1996). These methods either do not generally achieve the higher specificity or are not suitable for routine analysis.
zs [00041 A robust method of detecting one mutant allele in. 104409 wild-type alleles would be advantageous for many applications including detecting -minimal residual disease (recurrence = after remission or rare remaining cancer cells in lymph nodes and other neighboring tissues) and = measurement of mutation load (the frequency and pattern of somatic mutations present in normal tissues). Individuals with. a high mutation load may be at increased risk for cancer due to 30 either environmental exposure or endogenous defects in any of hundreds of genes necessary to maintain the integrity of the genome. For those individuals found to have a high. mutation load, clues to etiology can be obtained, by defining the mutation pattern.
[00051 There are many DNA sequencing methods and their variants, such as the Sanger . _sequencing using dideoxy termination and denaturing gel electrophoresis (Sanger et al., 1977), =

2 Maxam-Gilbert sequencing using chemical cleavage and denatiring gel electrophoresis (Maxam and Gilbert, 1977), pyro-sequencing detecting pyrophosphate (PP) released during the DNA
polym.erase reaction (Ronaghi et al., 1998), and sequencing by hybridi7ation (SBH) using oligonucleotides (Lysov et at, 1988; Bains and Smith, 1988; Dmianac et al., 1989; Khrapko et s al., 1989; Pevzner et al., 1989; Southern et al., 1992).
[00061 There are multiple gel-based methods for scanning for unknown mutations including single stranded conformation polymorphism (SSCP) and the SSCP-hybrid methods of dideoxy fingerprinting (ddF), restriction endonuclease -6ngetprinting (REF), and Detection Of Virtually All Mutations-SSCP (DOV AM-S), denaturing gradient gel electrophoresis (DGGE), denaturing' io HPLC (dTTPLC) chemical or enzymatic cleavage (Sarkar et al., 1992; Liu and.
Sommer, 1995;
" Liu et at., 1999; Myers et al., 1985; Cotton et al., 1988; Liu et al., 1999;
Buzin et at, 2000;
Spiegelman et al., 2000). DOVAM-S and chemical cleavage reactions have been shown in blinded analyses to identify essentially all mutations (Bnzin et at, 2000).
dHPLC, which is based on. reverse phase chromatography, also may identify essentially all mutations under is appropriate conditions (O'Donovan et al., 1998; Oefner and Underhill, 1998; Spiegelman et al., 2000). Efforts are under way to develop general scanning methods with higher throughput.
[00071 Sequencing by hybridization (SBH) is being adapted to scanning or resequencing for unknown mutations on microarrays (Southern, 1996). This continues to be a promising area Of intense study. However it is not possible as yet to detect most microinsertions and deletions with 20 this approach and the signal to noise ratio for single base changes precludes detection of 5-10% -of single nucleotide changes (Hacia, 1999). Alternative _approaches warrant exploration.
[00081 It is becoming increasingly apparent that in vivo chromatin structure is crucial for mammalian gene regulation and development Stable changes in chromatin structure often involve changes in methylaiion and/or changes in histone acetylation.
Somatically heritable 25 changes in chromatin structure are commonly called epigenetic changes (Russo and Riggs, 1996) and it is now clear that epigenetic "mistakee' or epimutations are frequently an important contributing factor to the development of cancer (Jones and Laird, 1999).
[00091 One of the few methods for assaying in vivo chromatin structure, and the onlr. method with resolution at the single nucleotide level, is ligation-mediated PCR (LM-PCR.) (Mueller and 30 Wold, 1989; Pfeifer et at, 1989) and its variant of terminal transferase-mediated PCR (fl)-PCR) (Komura and Riggs, 1998). Many aspects of chromatin struCture can be determined by LM-PCR, such as the location of methylated cytosine residues, bound transcription factors, or positioned nucleosomes. It is readily apparent that LM-PCR works better with some primer sets

3 than with others. Thus, it is desired to develop a more robust method of measuring chromatin structure.
=
[0010] Thus, it is an object of the present invention to develop alternative methods for amplification of DNA, fot sequencing DNA and for analysis of chromatin structure. This object is accomplished by the use of-the novel pyrophosphorolysis activated polymerization (PAP) as described herein. PAP has the potential to enhance dramatically the specificity of the amplification of specific alleles, for resequencing DNA and for chromatin structure analysis.
SUMIVIARY OF THE INVENTION
o [0011] The invention is a pyrophosphorolysis activated polymerization (PAP) method of synthesizing a desired nucleic acid strand on a nucleic acid template strand.
The method comprises the following steps carried out serially.
[0012] (a) Annealing to the template strand a complementary activatable oligonucleotide P*.
This activatable oligonucleotide has a non-extendible 3' terminus that is activatable by is pyrophosphorolysis (hereinafter referred to as a non-extendible 3' terminus or a 3' non-extendible end or a non-extendible 3' end). The non-extendible 3' terminus (or end) is a nucleotide or nucleotide analog which has the capacity to form a Watson-Crick base bair with a complementary nucleotide and which lacks a 3' OH capable of being extended by a nucleic acid =
polymerase. In one embodiment, the non-extendible 3' terminus May be a non-extendible ' 20 3 'deoxynucleotide, such as a dideol.rynucleotide. In a second embodiment, the non.-estendible 3' terminus may be a chemically modified. nucleotide lacking the 3' hydroxyl group, such as an acyclonucleotide. Acyclonucleotides- substitute a 2-hydroxyethoxyhaethyl group for the 2'-deoxyribofuranosyl sugar normally present in dNMPs. In other embodiments, the non-extendible 3' terminus may be other blockers as described herein. In one embodiment, the 25 activatable oligonucleotide P* has no nucleotides at or near its 3' terminus that mismatch the corresponding nucleotide's on the template strand. In. a second embodiment, the activatable, oligonucleotide P* has a mismatch at or within 16 nucleotides of its 3' terminus with respect to a corresponding nucleotide on- the template strand. The tentiftal 3'-deoxyntteleotide is hybridiZed to the template strand when the oligonucleotide P* is a-nnealed.
30 [0013] (b) Pyrophosphorolyzing the annealed activatable oligonucleotide P* with pyrophosphate and an enzyme that has pyrophosphorolysis activity. ThiS
activates the oligonucleotide P* by removal of the hybridized non-extendible 3' terminus. =

4 =
[00141 (c) Polymerizing by extending the activated oligonucleotide P* on the template strand in presence of four nucleoside triphosphates of their analogs and a nucleic acid polymerase to synthesize the desired nucleic acid. strand.
[00151 The PAP method can be applied to amplify a desired nucleic acid strand by the following additional steps.
[00161 (d) Separating the desired nucleic acid strand of step (c) from the template strand, and [0017] (e). Repeating steps (a)-(d) until a desired level of amplification of the desired nucleic acid strand is achieved.
[0018] In a preferred aspect, the PAP method as described above is applied to allele-specific amplification (PAP-A). In this application, the nucleic acid template strand is a sense or antisense strand of one allele and is present in admixture With the corresponding (sense or antisense) nucleic acid strand of the second allele (the allelelic strand).
The activatable oligonucleotide P* has at least one nucleotide or analog at or near its 3' termirmS, e.g., within .16 nucleotides of the 3' terminus, that mismatches the corresponding nucleotide of the allelic strand.
is Because of the mismatch, in step (a) of the PAP. method the non-extendible 3' tenninus of oligonucleotide P* is not substantially hybridized to the allelelic strand. In step (b) the pyrophosphorolysis does not substantially remove the non-hybridized non-extendible 3' terminus from the activatable oligonucleotide P* annealed to the allelic strand. In step (c) the oligonucleotide P* is not substantially extended by polymerization on the allelic strand. As a 1:-) result, the desired nucleic acid strand synthesized on the template strand is amplified preferentially over any nucleic acid, strand synthesized on the allelelic strand.
[00191 In a second preferred aspect, the PAP-A method described above can be.
performed -bidirectionolly (Bi-PAP-A). Bidirectional-PAP (Bi-PAP) is a novel design that preferably uses two opposing pyrophosphorolysis activatable oligonucleotides (P*) with one nucleotide overlap at their 3' termini. Thus, in Bi-PAP, PAP-A is performed with a pair of opposing activatable oligonucleotide P*s. Both the downstream and upstream P*s are specific for the nucleotide of interest at the 3' termini (e.g., an A:T base pair). In the initial round of amplification from genoraic DNA, segments of undefined size are generated. In subsequent rounds, a seginent equal to the combined lengths of the oligonucleotides minus one is amplified exponentially.
30 Nonspecific amplification occurs at lower frequencies because- this design elimitatog misincorporation error from an unblocked upstream. The P's may be 3060 nucleotides for most efficient amplification.

[90201 The PAP method can be used to amplify either RNA or DNA. When used to amplify DNA, the activatable oligonucleotide P* may be a 2'-deoxyoligonucleotide, the non-extendible = 3' terminus may be, e.g., a 2',3'-dideoxynucleotide or an acyclonucleotide or other blOckers as described herein, the four nucleoside triphoSphates are 2'-deoxynticleoside triphosphates or their s analogs, and the nucleic acid polymerase is a DNA polymerase. The DNA
polyinerase used in = step (c) can also be the enzyme having pyrophosphorolysis activity used in step (b). Preferred DNA polymerases having pyrophosphorolysis.activity are thermostable Tfl, Mg, and genetically engineered DNA polymerases, such as ArapliTaqFs and ThermoSequenaseTm. These genetically engineered DNA polymerases have the mutation F667Y or an equivalent mutation in their active sites. The use of genetically engineered DNA polymerases, such as AmpliTaqFs and ThennoSequenaseTm, greatly improves the efficiency of PAP. These Family I
DNA
polymerases can be used when the activatable oligonucleotide P* is a 3' dideoxynucleotide or an acyclonucleotide. When the activatable oligonucleotide P* is an acyclonucleotide, Family archaeon DNA polyinerases can also be used. Examples of such polymerases include, but are not limited to, Vent (exo-) and Pfu (exo-). These polymerases efficiently =
amplify 3 'acyclonucleotide blocked P*. Two or more polymerases can also be used in one reaction. If the template is RNA, the nucleic acid polymerase may be RNA polymerase, reverse transcriptase, or their variants. The activatable oligonucleotide P* may be a ribornicleotide or a T-d.eoxynucleotide. The non-extendible 3' terminus may be a 3' deoxyribo-nucleotide or an acyclonucleotide. The four nucleoside triphosphates may be ribonucleoside triphosphates., 2' deoxynucleoside triphosphates or their analogs. For convenience, the description that follows uses DNA as the template. However, RNA is also included, such as described for the Present aspect.
= [00211 Amplification by the PAP method can be linear or expOnential.
Linear amplification is obtained when the .activatable oligonucleotide P* is the only complementary oligonucleotide . used, Exponential amplification is obtained when a second opposing oligonucleotide, which may be a P*, is present that is complementary to the desired nucleic acid strand.
The activatable oligonucleotide P* and the second oligonnclebtide flank the region that is targeted for amplification: In step (a) the second oligonucleotide anneals to the separated desired nucleic acid strand product of step (d). In step (c) polymerization extends the second oligonucleotide on the desired nucleic acid strand to synthesize a copy of the nucleic acid template strand. In. step = (d) the synthesized nucleic acid template strand is separated from the desired nucleic acid strand.

Steps (a) through (d) are repeated until the desired level exponential amplification has been = achieved.
[0022] In the PAP method, a mismatch between the activatable oligonucleotide P* and the template strand results in no substantial amplification, if the mismatch occurs in the 3' specific s subsequence of P* at the .3' terminus of P* or within 16 nucleotides of the 3' terminus of P*.
This lack of amplification for such mismatches in the 3' specific subsequence of P* provides four billion different and specific oligonucleotides with one base substitution resolution.
[00231 In a preferred aspect, the PAP. method is used for exponential amplification of a rare, - mutant allele in a mixture containing one or more wild-type alleles, Strands of the alleles are io separated to provide single-stranded nucleic acid, and then the following steps are carried out serially.
[00241 (a) Annealing to the sense or antisense strands of each allele a complementary activatable =T-deoxyoligonucleotide P* that has a non-extendible 3' terminus.
The non-extendible 3' terminus may be, e.g., a non-extendible 2',3'-dideoxynucleotide or an is acyclonucleotide. P* has no 2'-deoxymicleotides at or near its 3' terminus that mismatch the corresponding 2'-deoxynncleotides on the mutant strand, but has at least one 2'-deoxynucleotide at or near its 3' terminus that mismatches the corresponding 2'-deoxynucleotide on the wild-type stand. Consequently, the non-extendible 3' terminus is hybridized to the mutant strand but not to the wild-type strand when the oligonucleotide P* is annealed. Simultaneously, a second 20 2'-deoxyoligonucleotide that is complementary to the anti-parallel strands of each allele is annealed to the anti-parallel strands. The activatable T-deoxyoligonucleotide P* and the second 2'-deoxyoligonucleotide fin.* the region of the gene to be amplified.
[00251 (b) Pyrophosphorolyzing the activatable 2'-deoxyoligonucleotide P* that is annealed to a mutant stiand with pyrophosphate and an enzyme that has pyrophosphorolysis activity. This 25 activates the 2'-deoxyoligonucleotide P* that is annealed to the mutant strand by removal of the hybrie1i7ed non-extendible 3' terminus. = It does not substantially activate the 2'.-deoxyoligonucleotide P-* when it is armealed to the mutant strand because the non-hybridized non-extendible 3' terminus is not substantially removed by the pyrophosphorolysis.
[0026] (c) Polymerizing by extending the activated oligonucleotide P* on the mutant strand in 30 presence of four nucleoside triphosphates or their analogs and a DNA
polymerase and extending the second 2'-deoxyoligonucleotide on both mutant and wild-type anti-parallel strands.
[00271 (d) Separating the extension products of step (c);

[00281 (e) Repeating steps (a)-(d) until the desired level of amplification of the mutant allele has been achieved.
[0029] The activatable 2'-deoxyoligonucleotide P* is annealed to the antisense strands of the alleles and the second 2`-deoxyoligonucleotide is annealed to the sense strands, or vice versa.
[00301 Steps (a) to (c) of PAP can be conducted sequentially as two or more temperature stages on a thermocycler, or they can be conducted as one temperature stage on a thermocycler.
[00311 Nucleoside triphosphates and 2'-deoxynucleoside triphosphates or their chemically ' modified versions may be used as substrates for multiple-nucleotide extension by PAP, i.e., when one nucleotide is incorporated the extending strand can be further extended, i.o 2',3'-dideoxynucleoside triphosphates, their chemically modified versions, acyclonucleotides or other blocked nucleotides which are terminators for further extension may be used for single-nucleotide extension. 2`,3'-dideoxynucleoside triphosphates may be labeled with radioactivity or dye for differentiation from the 3' terminal dideoxynucleotide, if present, of oligonucleotide P*.
Mixtures of nucleoside triphosphates or 2'-deoxynucleotide triphosphates or their analogs, and 2',3'-dideoxynucleoside triphosphates or their analogs may also be used.
[00321 PAP can be used in a novel method of DNA sequence determination. In PAP,.
pyrophosph.orolysis and polymerization by DNA polymerase are coupled serially by using P*, an oligonucleotide containing a non-extendible 3' terminus. The non-extendible 3' terminus may be, e.g., a non-extendible 3' -deoxynucleotide or an acyclonucleotide. This principle is based on the specificity of PAP and in, turn on the base pairing specificity of the 3' specific subsequence.
This property of the 3' specific subsequence can be applied to scan or resequence for unknown sequence variants, to determine de novo DNA sequence, to compare two DNA
sequences, and to monitor gene expression profiling in large scale. A P* array is possible in these methods. That is, each of the P*s can he immobilized. at an individual dot or a solid support, thus allowing all the PAP reactions to be processed in parallel.
[00331 Thus in one aspect, the PAP method is used for scanning or resequencing unknown sequence variants within a predetermined sequence by carrying out the following steps serially.
[00341 (a) Mixing under hybridization conditions a template strand of the nucleic acid with.
multiple sets of four activatable oligonucleotides P* which are sufficiently complementary to the template strand to hybridize therewith. Within each set the oligonucleotides P* differ, from each other in having a different non-extendible 3' terminus; so that the non-extendible 3' terminus is hybridized to the template. strand if the template strand is complementary to the non-extendible 3' terminus. The number of sets corresponds to the number of nucleotides in the sequence. The non-extendible 3' terminus may be, e.g., a non-extendible 3 '-deoxynucleotide or an acyclonucleotide.
[00351 (b) Treating the resulting duplexed P*s with pyrophosphate and an enzyme that has pyrophosphorolysiS activity to activate by pyrophosphorolysis only those oligonucleotides P*
which have a non-extendible 3' terminus that is hybridized to the template strand.
[0036] (c) Polymering by extending the activated oligOrticleotides P* on the template strand in presence of four nucleoside hiphosphateS or their analogs and a nucleic acid polymerase.
[00371 (d) Separating the nucleic acid strands synthesized in step (c) from the template strand.
[00381 (e) Repeating steps (a)-(d) until a desired level of amplification is achieved, and ic [00391 (f) Arranging the nucleic acid sequence in order by analyzing overlaps of oligonuclotides P* that produced amplifications.
10040] In a second aspect, the PAP method is used fer determining de novo the sequence of a nucleic acid by carrying out the following steps serially.
[00411 (a) Mixing under hybridization conditions a template strand of the nucleic add with is multiple activatable oligonucleotides P*. All of the oligonucleotides P*
have the same number n of nucleotides as the template and constitute collectively all possible sequence's having n nucleotides. All of the oligonucleotides P* have a non-extendible 3' terminus.
The non-extendible 3' terminus may be, e.g., a non-extendible3,-deoxynucleotide or an acyclonucleotide.
Any oligonucleotides P* that are sufficiently complementary will hybridize to the template 20 strand. The non-extendible 3' terminus -will hybridize to the template strand only if the template strand is complementary at the position corresponding tO the 3' terminus'.
[00421 (b) Treating the resulting duplexed P*s. with pyrophosphate and an enzyme that has pyrOphosphorolysis activity to activate only those hybridized oligonucleotides P* which have a non-extendible 3' terminus that is hybridized to the template strand, by pyrophosphorolysis of 25 those hybridized non-extendible 3' termini.
[00431 (c) Polymerizing by extending the activated oligonucleotides P* on the template strand in presence of four nucleoside triphosphates or their analogs and a nucleic acid polymerase.
[00441 (d) Separating the nucleic acid strands synthesized in step (c) from the template stand.
[00451 (e) Repeating steps (a)-(d) until a desired level of amplification has been achieved, and 30 [00461 (f) Detennining the sequence of oligonucleotides P* that produced amplifications, then arranging the nucleic acid sequence in order by analyzing overlaps of these oligonucleotides.
[00471 PAP can also be used to study chromatii structure analogously to ligation-mediated PCR
(LM-PCR) by carrying out the following steps serially. LM-PAP has been used for the determination of primary nucleotide sequence, cytosine methylation patterns, DNA lesion formation and repair and in vivo protein-DNA footprints (Dai et al., 2000;
Mueller and Wold, 1989; Pfeifer et al, 1989; Pfeifer et al.; 1999; Becker and Grossman, 1993).
Ligation-mediated PAP (LM-PAP) involves cleavage, primer extension, linker ligation and PAP that can be applied s for analysis of in vivo chromatin structure, such as, methylated state of chromesomes, and for other nucleic acid. analysis as for LM-PCR:
[00481 The nature of LM-PAP is that the template is synthesized before PAP, such as by ligation reaction or by extension using terminal transferase. PAP may be any type of PAP: with only one P*, with two opposing oligonucleotides where at least one is P*, Bi-PAP, matched io PAP, mismatched PAP, and. so on. Thus, at its simplest, LM-PAP is the application of PAP to a =
presynthesized template. LM-PAP may be performed by steps (i), (iv) and (v), by steps (i), (iii) and (Vi); by steps (ii), (iii), (iv) and (v) or by steps (ii), (iii) and (vi), where the -steps are as follows.
[0049] (i) The cleavage occurs chemically, enzymatically or naturally to "breakdown" nucleic . 15 acid strands: The nucleic acid usually is genomic DNA that may have lesions or nicks produced in vivo.
100501 (ii) The oligonucleotide P1 is gene-specific and its extension includes: 1) annealing to . the template strand a substantially complementary oligonucleotide; 2) extending the oligonucleotide on the template strand in the presence of nucleoside triphosphates or their zo analogs and a nucleic acid polymerase, the extension "runs off' at the cleavage site on the template strand. Steps 1) and 2) may be repeated:
[00511 The primer extension may be replaced by a P* extension, in which the above PAP is performed with only one activatable oligonucleotide P*.
[00521 (iii) The linker ligation step includes " ligation of a linker to the 3' terminus of the synthesized nucleic acid strand. The linker ligation step may be replaced by a terminal transferase extension that is non-template dependent polymerization and an extra nucleic acid sequence is added to the 3' terminus .of the synthesized nucleic acid strand.
[00531 (iv) PCR. is performed with a second gene-specific oligenucleotide (P2) together with an.
oligonucleotide specifie for the linker or the added Sequence by terminal transferase.
3 [00541 (v) A third gene-specific P* (P3) is used to detect the PCR
generated fragments. PAP
method is applied with only one activatable oligonucleotide P*. The extension of the activated oligonucleotide P* "runs off" at the end of the template strand generated in IV. The PAP
method may be applied in an allele-specific manlier. The activatable oligonucleotide P* may =
contain one or more nucleotides that are not complementary to the template strand. The uncomplimentary nucleotide(s) of P* may locate at the 3' terminus of P*.
[00551 (vi) Instead of steps (iv) and (v), PAP method can be applied with two opposing oligonucleotides of which at least one is the activatable oligonucleotide P*.
The activatable

5 oligonucleotide P*(P3) is gene-specific. The second oligonucleotide is specific for the linker or the added sequence by terminal transferase. The second oligonucleotide may be another activatable oligonucleotide P* or a regular oligonucleotide. The PAP method may be applied in an allele-specific manner. The activatable oligonucleotide P* (P3) may contain one or more nucleotides that are not complementary to the template strand. The uncomplimentary o nucleotide(s) of P* may locate at the 3' terminus of P* (P3).
[00561 The third gene-specific oligonucleotide (P3) is then usually used to label and allow visualintion of the PCR generated fragments. P3 is labeled at the 5' terminus with 32P or, more recently, with near infrared fluorochromes such as [RD 700 or [RD 800 (Li-Cor Inc.) (Dai et al., 2000).
is [0057) PAP can be used to detect a target nucleic acid. In one embodiment this method involves the following steps:
[00581 (a) -adding to a nucleic acid containing sample an oligonucleotide P*, wherein the oligonucleotide P* has a non-extendible 3' terminus, wherein the 3' terminal residue of oligonucleotide P* is removable by pyrophosphorolysis, and wherein the oligonucleotide P*
anneals to a substantially complementary strand of the target nucleic acid present in the sample;
[0059] (b) removing the 3' non-extendible terminus of the oligonucleotide P*
annealed to the substantially complementary strand of the target nucleic acid by pyrophosphorolysis to unblock the oligonucleotide P* to produce an unblocked oligonucleotide; and [0060] (c) detecting the presence of the target nucleic acid, wherein the sequence of the target nucleic acid is substantially complementary to the sequence of the oligonucleotide P*.
[00611 The method of the first embodiment may further include before the detection step the step: (b1) extending the unblocked oligonucleotide using a nucleic acid polymerase to produce an extended oligonucleotide. The method may also include the addition of a second oligonucleotide which may or may not have a 3' non-extendible terminus. The second oligon.ucleotide may anneal to the substantially complementary strand of the target nucleic acid or it may anneal to the complement of the substantially complementarty strand of the target nucleci acid.

= 11 [0062] In a Second embodiment for detecting a nucleic acid. the method involves the following steps:
[0063J (a) adding to a nucleic acid containing sample two oligonucleotide P*s, wherein each oligonucleotide P* has a non-extendable 3' terminus, wherein the 3' terminal residue of each s oligonucleotide P* is removable by pyrophosphorolysis, wherein one oligonucleotide -P*
overlaps with the other oligonucleotide P* by at least one nucleotide at their respective 3' ends, and wherein one oligonucleotide P* anneals to a substantially complementary strand of the rargot acid-present in. -the- sample- and-- the -other-oligonucieotide--P*- -anneals-ta-. a -complement of the substantially complementary strand of the target nucleic acid;
[0064] (b) removing the 3' non-extendable terminus of the oligonucleotide P*s annealed to the target nucleic acid by pyrophosphorolysis to unblock the' oligonucleotide P*s to produce unblocked oligonucleotides; and [0065] (c) detecting the presence of the target nucleic acid, wherein the sequence of the target nucleic acid is substantially complementary to the sequence of the oligonucleotide P*s.
= 15 [0066] The method of the second embodiment may thither include before the detection step the step: (bl) extending the unblocked oligonucleotide using a nucleic acid polymerase to produce an 'extended oligonucleotide. =
- [0067] In one embodiment, the detection of the nucleic acid in step (c) is performed by detecting the tinblOcking of oligonucleotide P*. In one aspect, the unblocking is detected by loss of a label contained in the 3' terminal residue of oligonucleotide P*. In a second aspect, the unblocking is detected by detecting the presence of a 3' OH on the 3' terminal residue that is capable of extension or ligation. In this aspect, the detection is determined by extending the =
unblocked oligonucleotide or by ligating the unblocked oligonucleotide to an oligonucleotide.
In a second embodiment, the detection of the nucleic acid in step (0) is.
performed by detecting the extended oligonucleotide. In one aspect, the extended oligonucleotide is detected by the presence of a label in the extended oligonteleotide. The label is part of a nucleotide or nucleotide analog used in the extension step. In a second aspect, the extended oligonucleetide is detected by gel electrophoresis: In a third aspect, the extended oligonucleotide is detected by the binding or incorporation of A dye or spectral' material.
[00681 The P* oligonncleotides are selected to be "substantiallyi!
complementary" to the = different strands of each specific sequence to be smpiiiied, Therefore, the P* oligonucleotide sequence need. not reflect the exact sequence of the template. For example, a non-complementary nucleotide segment may be attached to the 5'-end of the P*
oligonucleotide, with the remainder of the P* oligonucleotide sequence being complementary to the strand.
Alternatively, non-complementary bases or longer sequences can be interspersed into the P*
oligonucleotide, provided that the P* oligonucleotide sequence has sufficient complementarity with the sequence of the strand to be amplified to hybridize therewith and form a template for s synthesis of the extension product of the other P* oligonucleotide.
The ability to detect nucleic acid sequences which are substantially complementary to oligonucleotide P* is particularly useful for the detection of multiple mutations; such as seen in high viral load, where the detection of the presence of the virus is important and not necessarily the exact nucleic acid sequence of the virus. This method is also capable of detecting nucleic acids that are completely o complementary.
[0069] The present invention also includes other modifications of PAP.
[00701. The activatable oligonucleotide P* may contain blocked nucleotides at other positions in addition to the 3' terminus.
[00711 = The introduction of internal blocking nucleotides results in. an interface between 15 amplification and PAP which would permit PAP to amplify in a non-exponential manner (e.g., quadratic or geometric) with higher fidelity, i.e., errors made by the polymerase would not be prop agatable.
[0072] = The activatable oligonucleotide P* may contain modified nucleotides that are extendible as well as the 3' blocked nucleotide. Thus, anywhere 5' to the 3' terminus, there may 20 be either blocking or non-blocicing modified nucleotides.
[0073] = A polymerase that pyrophosphorolyzes the mismatched primer rather than the matched primer could be used to detect rare mutations in which the II* that mismatched at the 3' terminus is activated and extended.
[0074] = The detection of a rare mutation is based on no mismatch anywhere along the length of 25 the oligonuoleotide because a Mismatch inhibits the activation of P*s.
[00751 = Activation may occur by another mechsnism, such as a 3' exonuelease.
The 3' exonuclease may have specificity for the matched primer or the mismatched primer so that it discriminates as to whether there is a mismatch at the 3' end. The 3' exon.uclease can be used either way. If it prefers a mismatch, it can be used as described above, but its ability to detect 30 uncommon mutations would depend on some specificity for activation, although that specificity may come partly from internal mismatches.

[00761 = The extension reaction can be performed by a DNA polymerase, an RNA
polymerase or a reverse transcripta.se, the template may be a DNA or an RNA, and the oligonucleotide P*
may be a DNA, an RNA, or a DNA/RNA heteromer.
[0077] = Pyrophosphorolysis and the extension can be performed by different polyrnerases. For s example, the P* may include a penultimate modified oligonucleotide that could not be extended by pyrophosphorolyzing polymerase but could be extended by another polymerase.
One example is a 3' dideoxy that could be pyrophosphorolyzed by a DNA polymerase, but the presence of a ribonucleotide in the penultimate position would require extension by an RNA
polymerase. =
[0078] = PAP can. be generalized as an inactive oligonucleotide that is activated by a nucleic acid metabolizing enzyme, such as helicases, top oisomerases, telomerases, RNase H or restriction enzymes.
[0079] = Methylases would detect the presence or absence of a methyl group in genomic DNA.
Methylase,s could be coupled with truncating amplification which forces the polymerase back to the template.
[0080] = A P* in which the 3' end is a dideoxy and penultimate few nucleotides are ribos can be used as a tool for differentially making a protein product derived from a specific mutation that was desired, or for making a protein product whose expression is linked, to the presence of a particular sequence. Pyrophosphorolysis would activate the P* if there was a precise match to the mutation at the 3' end. The activated oligonucleotide is then a substrate for the generation of RNA by an RNA polyinerase. The RNA could then be translated in vitro to produce the protein product [0081] = PAP (PAP, Bi-PAP, matched or mismatched PAP, simplex PAP, multiplex PAP and.
others) can be used for quantification. The yield of the amplification products is quantitatively associated with the amount of input template. The association may be proportional or otherwise.
[00821 = In PAP, product may accumulate linearly, exponentially or otherwise.
BRIEF DESCRIPTION OF THE FIGURES
[0083] Figure 1 shows a schematic of the detection of a rare mutation by allele-specific PAP
(PAP-A).
= [00841 Figure 2 shows a schematic of bidirectional PAP-A (Bi-PAP-A).
[00851 Figure 3 shows a schematic of PAP-based resequeucing (PAP-k) performed on a microarray with programmable photochemical oligonucleotide.

[0086] Figure 4 shows a schematic of microarray-based .resequencing to detect a G to A
mutation.
=
[0087] Figure 5 shows a schematic of ligation-mediated PR (LM-PCR).
[0088] Figures 6A and 613 are a schematic illustrating use of PAP to detect the G allele at =
s nucleotide 229 of the D1. dopamine receptor gene. The procedure is described in detail in Example 1 below..
[0089] Figure 6C is an autoradiogram of PAP from the GIG, A/A and G/A
genotypes of the huthan dopamine receptor gene.
[0090] Figures 7A and 7B are diagrams illustrating enhanced specificity of PAP
relative to o PASA.
[0091] Figures 8A and 813 are autoradiograms showing the results of electrophoresis of samples obtained in Example 1 below.
[0092] Figure 9 is an autoradiogram showing the results of electrophoresis of samples obtained in Example 1 be/ow.
is [0093] Figure 10 is an autoradiogram showing the results of electrophoresis of samples obtained in Example 1 below.
[0094] Figure 11A is a schematic illustrating enhancement of PAP efficiency.
[0095] Figure 1113 is an autoradiogram of PAP from the GIG, .AJA and G/A
genotypes of the human dopamine receptor gene.
20 [0096] Figures 12A-12E are autoradiogra-ms showing the results of electrophoresis of samples obtained in Example 2 below.
[0097] Figure 13 is an. autoracliograrn showing the results of electrophoresis of samples obtained in Example 2 below..
[0098] Figure 14 is an autoradiograni showing the results of eleetrophoresis of-samples obtained =
25 in. Example 2 below.
[0099] Figure 15 is an autoradiograrn showing the results of electrophoresis of samples obtained in. Example 3 below.
[0100] Figures 16A-16B show UV footprinting by LM-PAP. Fig. 16A shows allele-specific LM-PAP versus allele-specific LM-PCR for the dopamine D1 receptor gene. Fig.
16B shows 30 LM-PAP for the pgic gene.
[01011 Figures 17A-17B show PAP amplification directly from human .(Fig. 17A) and mouse (Fig. 17B) genonaic DNA using PAP and Bi-PAP, respectively.

=
[01021 Figures 18A-18E show PAP amplification using 3' terminal acyclonucleotide blocked P*. Fig 18A: Model: A duplexed DNA template of the lad gene is shown. The mutated template contains a G at the nucleotide position 369, while the wild-type template contains a T
at the nucleotide position 369 of the /ad" gene. P*
pyrophosphorolysis activatable oligonucleotide. The P* has an acycloN1VIT or a ddNIAP at the 3' terminus. The P* is specific to the mutated template but mismatches to the wild-type template at the 3' terminus (Table 6). 0 =
oligodeoxynucleotide. PAP was performed with P*1 and 01, P*2 and 02, or P*1 and P*2, respectively. Fig. 18B: PAP with 30 mer P*s: The P*s are specific for the mutated template but mismatch the wild-type template at their 3' terminus. In lanes 1-8 are 3' terminal io acyclonucleotide blocked P*s. In lanes 9-16 are 3' terminal dideoxyn_ucleotide blocked P*s for comparison. In lanes 1-4 and 9-12, the mutated template is used. In lanes 5-8 and 13-16, the wild-type template is used. The PAP product and P* are indicated with their sizes. Lane M is 12Ong of 4a174-PUC19/HaeLTI DNA marker. Fig. 18C: PAP with 35-mer P*s: The experiment is the same as in Fig. 18B except with 35-mer P*s that are 3' co-terminal with the 30-mer P*s is and five nucleotides longer at their 5' termini. Fig. 18D: PAP with Vent (exo-) polymerase. The experiment is the same as in Fig. I8B except that Vent (exo-) was used. Fig.
18E: PAP with Pfu (exo-) polymerase. The experiment is the same as in Fig. I9B except that Pfu (exo-) was used.
[0103] Figure 19 shows that PAP has high selectivity to detect rare mutations in the abundance of the wild-type template. In the example of nucleotide position 190, the mutation-specific P*
matches the mutated A template but mismatches the wild-type T template at the 3' terminus.
Specific and efficient amplification is indicated, by thick arrows. When hybridized to the mutated A template, the P* cannot extend directly from the 3' terminal dideoxynucleotide, the 3' terminal ddTIVf2 must be removed by pyrophosphorolysis and the activated oligonucleotide is then extended efficiently. Two types of nonspecific amplification ikom the T
template are zs indicated as Types I and 11 The nonspecific amplification -occurs rarely when mismatch pyrophosphorolysis occurs to generate a/ wild-type product that will not support efficient amplification as template for subsequent cycles (Type I) (the error is indicated by thin arrow and estimated frequency of as low as 10-5). When both mismatch pyrophosphorolysis and misincorporation occur extremely rarely to generate a mutated product (Type 11) (the errors are 3o indicated by thin arrowe and estimated. coupled frequency of 3.3x 101). Once the errors occur, the mutated product can be amplified exponentially in subsequent cycles and so it determines the selectivity.

[0104j Figures 20A-20B show Bi-PAP amplification. Fig. 20A: Schematic of Bi-PAP to detection a rare mutation: The mutation-specific assay with two mutated P* for nucleotide 190 is shown. The downstream and upstream P*s contain a dideoxy T and a dideoxy A at their 3' termini, respectively. They are specific for the T:A allele at nucleotide 190 (on the right), but are mismatehed to the A:T wild-type allele at their 3' termini (on the left). The P*s are 40 nucleotides long and overlap at their 3' termini by one nucleotide. On the left, no substantial product is generated from the wild-type template because of the mismatch. On the right, the mutated product is generated efficiently frorri the mutated template. Fig.
2013: Bi-PAP
amplification directly from X DNA. Each of the wild-type and mutation-specific Bi-PAP assays =
o at nucleotide 190 was used to amplify a 79-bp segment of the lad gene from X
DNAs. For the wild-type assay in. lanes 1-3, the two wild-type P*s have 3' terminal ddA and ddT, respectively.
For the mutation-specific assay in lanes 4-6 and lanes 7-9, the two mutated P*s are with ddT and ddA at their 3' termini, respectively. In lanes, 1, 4 and 7, 2000 copies of the wild-type template were added to each reaction. In lanes 2, 5 and 8, 2000 copies of the mutated template were added to each reaction. In lanes 3, 6 and 9; no template was added. In lanes 7-9, 200 ng of human genomic DNA was added as carrier. The product and P* are indicated Lane M is 12Ong of +X174-PUC19/Haelli DNA marker.
[01051 Figures 21A-21C show titration of template for sensitivity and selectivity of Bi-PAP.
With the mutated P*s, the wild-type template was amplified to generate tfie ifintataproduct in Experiment I. The mutated template was = amplified to generate the mutated product in Experiments II, III and W. Fig. 21A: The mutation-specific Bi-PAP assay for A190T. In Experiment I, the copies of the wild-type A, DNA are indicated in lanes 1-5.
Lane 6 is a negative contror 'With no DNA. In Experiment II, the copies of the mutated 2t, DNA are indicated in lane 7-11: Lane 11(0.2 copy) is a negative control to support the dilution accuracy of copy number.
Lane 12 is a negative control with no DNA. In Experiment Ili, the copies of the mutated X DNA
in the presence of 2x109 copies of the wild-type X DNA are indicated in lane 13-17. Lane 18 is a negative control with only the wild-type X DNA. In Experiment IV, the copies of the mutated X
DNA. in the presence of 100 rig of human genomic DNA are indicated in lanes 19-23.-Lane 24 is negative control only with the human genomic DNA. Lane "C WT" is the wild-type product 30.
control in which the wild-type P*s were used to amplify 2000 copies of the .wild-type A, DNA.
Lane "C Muf' is the mutated product control in which the mutated P*s were used to alnplify 2000 copies of the mutated I DNA. The wild-type and. mutated products with unique mobilities =

= =

are indiCated. Fig. 2113: The mutation-specific Bi-PAP assay for T369G. Fig.
21C: The mutation-specific Bi-PAP assay for T369C.
[0106] Figure 22 shows a design of P* rnicroarray for Bi-PAP resequencing. Bi-PAP can be used. for resequencing to detect unknown mutations in a known region on a microarray. The P*s s are designed according to the wild-type template. The two opposing P*s for each Bi-PAP are anchored in a micro array spot. Each pair of arrows represents four Bi-PAPs for one nucleotide position. A mutation is indicated On the template, and it spans six overlapped P*s. On the micro array, many Bi-PAPs can be processed in a parallel way.
[01071 Figures- 23A-23B show a schematic of 13i-PAP resequencing. Fig. 23A:
Detection of the . 10 wild-type sequence:. This is a close look at the microanay. The P*s are designed according to the wild-type sequence. On the position of nucleotide A, four Bi-PAPs are synthesized with four pairs of P*s. The four downstream P*s have identical sequence, except that at the 3' terminus either ddAMP; ddTIVIP, ddGMP or ddCMP, corresponds to the wild-type sequence and the three possible single base substitutions. The four corresponding upstream P*s have identical sequence, except that at the 3' terminus either ddTMP, ddAMP, ddCMP or ddGMP. Each pair of P*s have one nucleotide overlap at their. 3' termini. On the next nucleotide C, another four pairs of P*s are synthesized (not shown). If the wild-type sample is added, only the. wild-type Bi-PAPs generates the- specific product that is labeled by fluorescence. In this way, to scan a I kb region, you need 89.00 P*s. Fig. 23B: Detection of an A to T mutation. On the mutated nucleotide T, the mutation-specific Bi-PAP generates the mutated product. On the next nucleotide G; no produet . of Bi-PA_P is generated because each pair of P* contains one or two mismatches (not shown).
= [0108] Figures 24A-24B -show Bi-PAP resequencing naicroarray. Fig. 24A:
Detection of the wild-type sequence. Four pairs of P*s are designed for each nucleotide position according to the = wild-type sequence. Each pair of P*s are downstream and upstream directed, and have one 25 overlap and complementary nucleotide at their 3' teniaini. The wild-type P* pair are specifically õ
amplified on each nucleotide position. If all of the wild-type P* pairs specifically amplified, the wild-type sequence can be determined. Fig. 2413: Detection of the A to T
mutation. With the mutated template; the mutation-specific Bi-PAP is P7cip1i19 ed. There is a window of no Bi-PAP
signals centered by the mutation-specific Bi-PAP and three successive nucleotides on each side.
30 The paired specific subsequoace is supposed to be seven nucleotides long. Any unknown single-base substitution peal be determined, even if if is a heterozygous mutation.
Also, small deletions and insertions can be detected and localized.

[01 09j Figure 25 shows PAP de novo sequencing on microarray. PAP can also be used for de novo DNA sequencing of an unknown region. The paired specific subsequence is supposed to fifteen nucleotides long. P* pairs of a complete set of the paired specific subsequence are on a microarray with known addresses. After the unknown DNA sample is added, Bi-PAP
is = s performed. All the amplified B-PAP products are collected and. then the paired specific subsequences of the amplified P* pairs are assembled by one-nucleotide overlapping. Thus, the unknown complementary sequence is reconstructed.
[01101 Figures 26A-26C show the detection of somatic mutations. Fig. 26A:
Eighteen genomic DNA samples of the laar transgenic mice were chosen. 2 pg genomic DNA of each sattple was amplified with the assay B to detect the T369G Mutation two times. Samples 1-10 are from livers of 25-month old mice. Samples 11-14 are from hearts (samples 11;13 and 14) and adipose (sample 12) of 6-month *old mice. Samples 15-18 are from brains of 10-day old mice. P =
positive control that amplified the mutated X DNA, N = negative control with no DNA, + =
" amplified product, - = no product Fig. 26B: The assay B was performed. In lanes 11-12, 13-16 as and 17-20, 2 lig, 0.5 pg and 0.125 pg of the lad- mouse genomic DNA
of sample 12 were used in each reaction, respectively. Lanes 1-10 are controls; the copy number of the mutated X DNA
per reaction was reconstructed by two-fold serial dilutions. In lanes 1-10 and 13-20, each reaction also contained 1 pg of the lad- mouse genomic DNA carrier. ss =
single-stranded, ds =
double-stranded. Fig, 26C: The assay B was performed. In. lanes 11-14, 2 lig of the Tact mouse genomic DNA of sample 3 was used in each reaction. In lanes 15-18, 2 pg of the lad+
mouse genomic DNA of sample 9 was used in each reaction. Lanes 1-10 are controls; the copy number of the mutated X. DNA per reaction is indicated. Each control reaction also contained 1 pg of the lad r mouse genomic DNA carrier.
,DETAMED DESCRIPTION OF TBE INVENTION
[01111 The following temainology is used herein.
[0112] Pyrophosphorolysis: zornoval of the 3' nucleotide from a nucleotide strand chain by to generate the nucleotide triphosphate.
DNA polymerase in the presence of pyroplolVilait, (PP) This is the reverse of the polymerization reaction.
[01131 PAP: Pyrophosphorolysi activated polymerization. PAP c.tet.ise one P*
or can use two opposing oligonucleotides in which at least one is P.
[01141 P*: an oligonudeotide with a non-extendible 3' terminus (or end) that is actii-,-,Asible by-pyrophosphorolysis.
=

[0115] PAP-A: PAP-based allele-specific amplification that can be used for detection of rare = mutations (Fig. 1).
- [01161)31-PAP-A: PAP-A perfomied with a pair of opposing P*, i.e., bidirectional (Fig. 2) with atleast one nucleotide overlap at their T-termini.
s [01171 PAP-R: PAP-based rese,quencing for detection of unknown mutations within a known sequence (Figs. 3 and 4).
= [01181 LM-PAP: ligation-mediated PAP. The nature of LM-PAP is that the template is synthesized before PAP, such as by ligation reaction or by extension using terminal transferase.
[0119] LM-PCR: ligation-mediated PCR (Fig, 5).
[01201 G'w or As' alleles: alleles of the common polymorphism of the dopamine DI receptor gene that was used as a model system herein (also referred to herein as G or A
alleles).
[0121] Linear PAP: PAP with only one P* for linear product accumulation.
[0122] Exponential PAP: PAP with two opposing oligonucleotides for exponential product accumulation, and at least one is P*.
[0123] Noise rate (%): the relative yield of the mismatched product to the matched product. A
specific signal for PAP is defined as a noise rate of less than 10%.
[01241 PASA: PCR amplification of specific. alleles (also known as allele-specific PCR or AR1V1S).
[0125] Resequeneing: scanning for unknown mutations and determining the precise sequence o changes within aknown sequence. Resequencing is distinguished from de novo sequencing.
[0126] Mutation toad: the frequency and pattern of somatic Mutations within a tissue.
[0127] Minimal residual disease: e.g., rare remaining cancer cells in lymph nodes and other neighboring tissues or early recurrence after remission.
[0128] Non-extendible 3' terminus (or end): a nucleotide or 'nucleotide analog at the 3' terminus (or end) of oligonucleotide P* that is non-extendible but that is actiVatable by pyrophosphorolysis. Examples of a non-extendible 3' termini (or ends) include, but are not limited to, a 2'3'-dideoxynucleotide, an acycionucleotide, 31-deoxyadenosine (cordycepin), 3'-azido-3r-deoxythymidine (AZT); 2',3'41ideoxyitioSine = (ddl), 2',3'-dideoxy-3'-thiacytidine (3TC) and 2',3'-didehydra-2',3'-dideoxythymidine (d4T).
[0129] Simplex PAP: one PAP (PAP, Bi-PAP, matched or mismatched PAP, and others) in one = reaction tube or on. a solid support.
[0130] Multiplex PAP: more than oue PAP (PAP, Bi-PAP, matched or mismatched PAP, and others) in one reaction tube or on a solid support; e.g., rnicroarray.

' [0131] Matched PAP: PAP having a match between P* and its template.
[0132] Mismatched PAP: PAP having a mismatch between P* and its template.
[0133] Nested PAP: PAP using two or more pairs of P* in which one pair is located inside a second pair on a template nucleic acid.
s [0134] Hotstart PAP: PAP in which an essential reaction component is withheld until denaturation temperatures are approached (Charo et al, 1992; Kellogg et al., 1994; Mullis, 1991; D'Aquila et al., 1991). Essential reaction components can be 'withheld by, e.g., a neutralizing antibody bound. to the polymerase, sequestering a component such as the polymers or MgC12 in wax, chemically modifying the polymerase to prevent activation until high 10 temperature incubation or separating components by wax.
[0135] Truncated Amplification: an amplification method which amplifies non-exponentially, e.g., in a quadratic or geometric manner, with over two chimeric oligonucleotides and produces truncated terminal products that are no more than three rounds of replication from the original template. (Liu et al., 2002).
15 [0136] Reactive '3 011: is a 3' OH that is capable of being extended by a nucleic acid polymerase or ligated to an oligonucleotide.
[0137] DNA polymerases, which are critical to nucleic acid amplificadon, catalyze some or all of the following reactions: i) polymerization of deoxynucleotide triphosphates or their analogs;
ii) pyrophosphorolysis of duplexed DNA in the presence of pyrophosphate (PP), [dNME]I
20 x[PPij + x[dNTP]; 3'-51exonuclease activity (which does not require PPi), and iv) 5'-3' exonuclease activity (Duetcher and Komberg, 1969; Komberg and Baker, 1992). For Tag and Tj/ DNA polymerases, polymerization and 5'-3' exonuclease activity have been reported . (Chien et al., 1976; Kaledin et al., 1981; Longley et al., 1990). For T7 Sequenasem and Thermo.
Sequenaseml DNA polymerases, pyrophosphorolysis can lead to the degradation of specific dideoxynucleotide- terminated segments in Sanger sequencing reaction (Tabor and Richardson, 1990; Vander Hom et al., 1997).
[0138] Pyrophosphorolysis is generally of very minor significance because PP i is degraded by pyrophosphatase under normal physiological conditions. However, in the presence of high in vitro concentrations of PPI, pyrophosphorolysis can be substantial. For oligonucleotides with a 3' terminal dideoxy nucleotide, only pyrophosphorolysis is possible. Once the dideoxy nucleotide is removed, the activated oligonudeotide can be extended by polymerization.
[0139] Pyropb.osphorolysis activated polymerization (PAP) offers a novel approach for retrieving a diversity of information from nucleic acids. The exceptional specificity of PAP

derives from the serial coupling of two reactions. PAP involves the activation by pyrophosphorolysis of a 3' terminal blocked oligonucleotide (P*) followed by extension of the activated oligonucleotide by DNA polymerization. Operationally, PAP involves the use of an activatable oligonucleotide (P*) in place of a normal oligonucleotide that can be directly s extended. Examples of P* include an inactive dideoxy terminated oligonucleotide P* or an inactive chemically modified nucleotide lacking a 3' hydroxyl group, such as an acyclonucleotide, or having a non-extendible nucleotide terminated oligonucleotide P*.
Acycloclonucleotides (acycloNTPs) in which the sugar ring ig absent are known to act as chain terminators in DNA sequencing (Sanger et al., 1977; Trainor, 1996; Gardner and Jack, 2002).
The activation of P* is inhibited by mismatches throughout the length of the oligonucleotide.
Mismatches even two nnoleotides from the 5' terminus inhibit PAP
amplification.
[0140] Activation of a P* by pyrophosphorolysis offers extraordinary specificity throughout the length of P*. The enhanced specificity can be used to detect rare known mutations, to elucidate unknown mutations by resequencing, to determine unknown sequence by de nova sequencing, to Is measure gene expressiori levels, to compare two sequences, and to increase the specificity of in vivo analysis of chromatin structure. Microarray-based programmable photochemical oligonucleotide synthesis and PAP are synergistic technologies. Thus, the enhanced specificity . can be used for rapid, microarray-based resequencing, de nova sequencing, gene expression profiling and SNP detection.
101411 A number of methods for enzymatic nucleic acid amplification in vitro have been developed and can be adapted to detect known sequence variants. These include polymerase chain reaction (PCR) (Saild et al., 1985; Saiki et al., 1988), ligase chain reaction (LCR) (Landegen, 1998; Barany, 1991) and rolling Girdle amplification (RCA) (Baner et al., 1998;
Lizardi et al., 1998). Herein, we describe pyrophosphorolysis activated polymerization (PAP), an approach that has the potential to enhance dramatically the specificity of PCR allele-specific smp1i-Hcation (Sommer et al., 1989). PAP differs from corrections with PCR in multiple ways: i) the P* oligonucleotide is= blocked at the 3' terminus and must be activated by pyrophosphorolysis, pyrophosphorolysis and polymerization are serially coupled for 'each apiplification, iii) PAP may be performed with one P* for linear amplification or with two oligonucieotides for exponential amplification, iv) PP; is necessary for the amplification, v) significant nonspecific amplification would require the serial coupling of errors of both mismatch pyrophosphorolysis and misincorporation.

[0142] The enhanced specificity of PAP relative to PASA is provided by serially coupling pyrophosphorolysis and polymerization. Significant nonspecific amplification requires mismatch pyrophosphorolysis and misincorporation by DNA polymerase, an extremely rare event For example as described herein, DNA polynaerase Was utilized to detect the G allele at nucleotide 229 of the DI dopamine receptor gene. P* was synthesized either With ddA, ddT, ddG
or ddC at the 3' terminus: The 3' terminal dideoxynucleotide inhibits direct extension by polymerization, but can be removed by pyrophosphorolysis in the presence of pyrophosphate (PP) when the P* is specifically hybridized with the complementary strand of the G allele, The activated oligonucleoticte can be extended by polymerization in the 5'-3' direction. =
a.o [0143] Evidence is presented that pyrophosphorolysis followed by polymerization can be used to increase the specificity of RASA. Significant nonspecific amplification with PAP requires the serial coupling of the two types of errors, i.e., mismatched pyrophosphorolysis and misincorporation (Fig. 1). The rate of mismatched pyrophosphorolysis is expressed as the relative rates of removal of a 3' miSmatch deoxynucleotide relative to the correct 3'.
as deoxynucleotide. The rate of mismatch pyrophosphorolysis is less than 104 for T7 DNA
polymerase (Komberg and Baker, 1992; Wong et at, 1991). The misincorporation rate to create a substitution mutation by polymerization, expressed- as the incmporation rate of an incorrect versus a correct dNMP, was reported to be 10-5 for T7 DNA polymerase and to be 10-4 for E.
coli DNA polymerase I (Komberg and Baker, 1992; Wong et al., 1991; Bebenek et al., 1990).
20 Similar results were reported for Mq DNA polymerase, 3'-5' exonuclease-deficient mutants of Ti DNA polyrnerase and E. coli DNA polymerase I (Komberg and Baker, 1992; Wong et at, 1991; Bebenek et al., 1990; Eckert and Kunkel, 1990).
[01441 PAP is a method of synthesizing a desired nucleic acid strand on a nucleOtide acid template strand. In PAP, pyrophosphorolysis and polymerization are "serially coupled for nucleic 25 acid- amplification using pyrophosphorolysis activatable oligonncleotides (P*). P* is an = oligonucleotide that is composed of N nucleotides or their analogs and has a non-extendible nucleotide or its analog at the 3' terminus, such as 3%5' dideoxynacleotide.
When substantially hybridized on its template strand, P* could not be extended directly from the 3' terminal nucleotide or its analog by DNA polymerase, the 3' terminal nucleotide or its analog of the P*
30 can be removed by pyrophosphorolysis and then the activated oligonucleotide N) can be extended on. the template.
[0145] The method comprises the following steps canied out serially.
=

= =

[01461 Annealing to the template strand a substantially complementary activatable oligonucleotide P*'. This activatable oligonucleotide P* has a non-extendible nucleotide or its analog at the 3' tenninuS.
[01471 (b) - Pyrophosphorolyzing the annealed activatable oligenucleotide P*
with =
pyrophosphate and an. enzyme that has pyrophospliorolysis activity. This activates oligonucleotide P* by removal of the 3' terminal non- extendible nucleotide or its analog.
[01481 (c) PolymeriZina by extending the activated oligonucleotide P* .on the template strand in = the presence of nucleoside tiph.osphates or their analogs and a nucleic acid polymerase to synthesize the desired nucleic acid strand.
o [01491 The PAP method can be applied to amplify a desired nucleic acid strand by the following additional steps.
[01501 (d) Separating the desired nucleic acid strand of step (C) from the template Strand, and [01511 (e) Repeating steps (A)-(D) until a desired level of amplification of the desired nucleic . acid strand is achieved. =
= is [01521 The above PAP method can be applied to allele-:specific Amplificatiori. The activatable oligonucleotide P* has one or more nucleotides that are not complenientary to the template strand. The uncomplimentary nucleotide(s) of P* may locate at the 3' terminus of P*. The above =- step of (A), (B) or (C) could not occur substantially. As the result, the desired micleic acid strand is synthesized substantially less.
' 20 [01531 The above PAP method can be applied with only one activatable oligonucleotide P. (e) 'Repeating steps (a)-(d); a desired levet of amplificationof the desired nucleic acid strand may be achieved li-hearly. The targeted nucleic acid region outside the annealing region may be of different sizes or of different sequence contexts, so the synthesized nucleic acid strands are of different sizes or of different sequence context.
25 [0154] The above PAP method can be applied with two opposing oligonucleotides of which at least one is the activatable oligonucleotide P. The activatable oligonucleotide P* and the second oligonucleotide are targeted for amplification of a nucleic acid region. Steps (a),-(c) occur to = the activatable oligonucleotide P*. The second oligottucleclide ia SubStantially = coMplementary to the other template'strand. If the second oligonucleotide is another activatable 30 oligonucleotide 154i, steps (a)-(c) occur. If the seCond oligonucleotide is. a regular extendible oligonucleotide, steps (a) and (c) occur: (modified a) annealing to its template strand, followed by (Modified c) polymerizing by extending the oligonucleotide on its teniplate Strand in the preadride of nucleoside triphosphates or their analogs and a nucleic acid pobijneta.se to synthesize the desired nucleic acid strand. (e) Repeating steps (a)-(d), or steps (a), (c) and (d), a desired level of amplification of the desired nucleic acid strand may be achieved, e.gõ
exponentially. The targeted nucleic acid region between the two annealing regions of the two opposing oligonucleotides may be of different sizes or of different sequence contexts, so the s synthesized nucleic acid strands are of different. sizes or of different sequence contexts:
[01551 LM-PAP involves cleavage, primer extension, linker ligation and PAP
that can. be applied for analysis of in vivo chromatin structure, such as, methylated state of chromosomes.
[015.61 LM-PAP may be performed by steps (1), (ii), (iii), (iv) and (v), by steps (i), (ii), (iii) and (vi), by steps (ii), (iv) and (v) or by steps (ii), (iii) and (vi), where the steps are as follows.
io [01571 The cleavage occurs chemically, enzymatically or naturally to "breakdown" nucleic acid strands. The nucleic acid usually is genoraic DNA that may have lesions or nicks produced in vivo.
101581 (ii) The primer of P1 is gene-specific and its extension includes: 1) annealing to the template strand a substantially complementary primer; 2) extending the primer on the template 15 strand in the presence of nucleoside triphosphates or their analogs and a nucleic acid polymerase, the extension "runs off' at the cleavage site on the template strand. Steps 1) and 2) may be repeated.
[01591 The primer extension may be replaced by a P* extension (The above PAP
with only one activatable oligonucleotide P*).
20 [01601 (iii) The linker ligation step includes ligation of a. linker to the 3' terminus of the synthesized nucleic acid strand. The linker ligation step may be replaced by a terminal transferase extension that is non-template dependent polymerization and. an extra nucleic acid -sequence is added to the 3' terminus of the synthesized nucleic acid strand.
[0161] (iv) PCR is performed with a second gene-specific primer (P2) together with a primer 25 specific for the linker or the added sequence by. terminal transferase.
[01621 (v) A third gene-specific P* (P3) is used to detect the PCR generated fragments. PAP
. method is applied with only one activatable oligonucleotide P*. The extension of the activated oligonucleotide P* "runs off" at the end of the template strand generated in step (iv). The PAP
method may be applied in allele-specific manners. The activatable oligonucleotide P* may 30 contain one or more nucleotides that are not complementary to the template strand. The uncomplimentary nucleotide(s) of P* may locate at the 3' terminus of P*.
[01631 (vi) Instead of steps (iv) and (v), PAP method can be applied with two opposing oligon.ucleotides a which at least one is the activatable oligonucleotide P*.
The activatable =

oligonucleotide P*(P3) is gene-specific. The second oligonucleotide is specific for the linker or the added sequence by terminal transferase. The second oligonucleotide may be another activatable oligonucleotide P* or a regular primer. The PAP Method may be applied in allele-specific manners. The activatable oligonueleotide P* (P3) may contain one or more nucleotides s that are not complementary to the template strand. The uncomplimentary nucleotide(s) of P*
may locate at the 3' terminus of P* (P3).
[0164] Fig, 1 shows detection of a rare mutation by allele-specific PAP (PAP-A). PAP-A can detect a rare allele with extremely high specificity because an allele-specific oligonucleotide with a 3' dideoxy terminus (P*) permits the serial coupling of pyrophosphorolysis and =
o polymerization. For example, if an allele-specific oligonucleotide has a 3' dideoxy terminus (P*) that matches a rare "T" allele, activation occurs by pyrophosphorolytic removal of the dideoxy nucleotide and is followed by polymerization (Situation A). Activation by pyrophosphorolysis does not normally occur with a mismatch at the. Tterminus as with the wild-type "C" allele (Situation B). Rarely, pyrophosphorolysis does occur at a mismatch (estimated 15 frequency 10-5), but the activated oligonucleOtide is extended to produce wild-type sequence (Situation C). A product that supports efficient amplification is generated when mismatch pyrophosphorolysis occ-urs; a polymerase error that inserts A opposite C in template DNA
(Situation. 1)). The frequency of mismatch pyrophosphorolysis coupled with the polymerase mutation is estimated at 105x3x10-6= 3x10111.
20 [0165] PAP has a potential specificity of 3x10711. Approaching this potential requires a design that eliminates confounding sources of error. For example, extension errors from non-blocked upstream oligonucleotides can generate a product with the mutation of interest. If the misincorporation rate for TaqFS is about 10-5 per nucleotide and. only one of the three misincorporations generates the mutation of interest, the error rate is about 3.3x1(16. POIYMeraSeS

that contain a proofreading function might have an error rate per specific mutation of 3x10-7.
Polymerases or polymerase complexes with lower error rates would improve specificity further.
[0166] One approach utilizes linear PAP. Linear PAP-A may be performed for 40 cycles with only P* in the presence of a fluorescent or radiolabeled ddNTP. A labeled terminated product of defined size will be generated when P* is activated. Linear PAP-A has the advantage of utilizing only the original genoraic DNA and eliminating error due to misincorporation from extension of an unblocked upstreamprimer. However, the sensitivity of detection is limited because the level of amplification is not greater than the number of cycles: For a simple genorne like lambda pbage, a detection specificity of 10-6 is possible; The specificity of linear PAP-A depends =

=

critically on the absence of unblocked, extendible oligonucleotides. To achieve a robust specificity of 10-6, unblocked extendible oligonucleotides should be present at le. This may be achieved by treating gel purified P* (about 99.99% pure with our present protocol) with a 3' to 5' exonuclease to degrade unblocked molecules followed by repurification by gel electrophoresis.
[0167] A secOnd approach is bidirectional PAP-A (Bi-PAP-A; Fig. 2). In Bi-PAP-A, both the downstream and upstream oligonucleotides are P*s that are specific for the nucleotide of interest. The P*s overlap at their 3' temnai by one nucleotide. This design eliminates extension error from a non-blocked upstream oligonucleotide. This design should not he limited by small amounts of active contaminating oligonucleotide to which the dideoxy terminus has not been io added (about 0.01% with our Current protocol) because the product generated will be that of a control and will not be a substrate for efficient amplification iti subsequent cycles, [0168] Bi-PAP-A generates a product that is the size of a primer dimer.
However,, it is not a primer dimer in the conventional sense, in that template DNA with a mutation of interest is an intermediate required to generate a product that is an efficient substrate for amplification in Is subsequent cycles. Bidirectional PAP-A eliminates important bottlenecks to specificity and has the potential to reach a specificity of [0169] As shown in Fig. 2, both the downStream and the upstream P*s are specific for the nuclecitide of interest at the 3' termini: (an A:T base pair in this example).
In the initial round of = amplification from genomic DNA, segments of undefined size will be generated. In subsequent 20 rounds, a segment equal to the combined lengths of the oligonucleotide mints one will be amplified exponentially. Nonspecific amplification occurs at lower frequencies because this design eliminates misincorporation error fauni an unblocked upstream oligonucleotide that can = generate the A:T template from a G:C wild-type template with an error tate of 3x104. The P*s may be 30-60 nucleotides for most efficient attplification. Situation A shows -that a template 25 with a rare A:T allele Will be amplified efficiently. Both the upstrenm and the downstream P*s are amplified efficiently. Situation 33 shows that if the DNA template contains the wild-type G:C sequence, neither the downstream nor the upstream P* will be activated substantially.
[0170] Rapid reseqtericing Will facilitate elncidation of genes that predispose to cancer and other complex diseases. The specificity of PAP lends itself to resequencing;
154's may be =
30 photochemically synthesized on rnicroarrayS using fiekible digital micromirror arrays.
[0171] Microarrays of ithrnohili7ed DNA Or oligonncleotides can be fabricated eitter by in situ light-directed cOmbinational synthesis or by conventional synthesis (reviewed by Ramsay, 1998;
Marshall and Hodgson, 1998). Massively parallel analysis can be perfOmaed.
Photochemical synthesis of oligonucleotides iS a powerful Means fOr combinatorial parallel synthesis of = addressable oligonucleotide microarrays (Singh-Gasson et at, 1999;
LeProust et at, 2000). This = flexible alternative to a large number of photolithographic Masks for each chip utilizes a masIdess array synthesizer with virtual masks generated on a computer. These virtual masks are s relayed to a digital micromirror array. An Ultraviolet imageof the virtual mask is produced on the active surface of the glass substrate by a 1:1 refiectiVe imaging system.
The glass substrate is mounted in a flow cell reaction chamber connected to a DNA synthesizer. Cycles of programmed chemical coupling occur after light exposure. By repeating the procedure with additional virtual masks, it is possible to synthesize oligonucleotide microarrays with any o desired sequence. The prototype developed by Singh-Gasson, et al.
synthesized oligonucleotide microarrays containing more than 76,000 features measuring 16 square microns.
= [01721 By combining programmable photochemical .oligonucleotide synthesis with digital minors and oligonucleotide extension of P*, a high throughput and automated method of resequencing is possible. PAP-R may detect virtually 100% of single base substitutions and ..15 other small sequence variants because of its high redundancy; the mismatch spanned by the several overlapping P* oligonucleOtides Will prevent activation of :a cluster of overlapping P*s.
One strategy for resequencing is shown in Figs. 3 and 4. Fig. 3 shows a schematic of PAP-R
performed on a microan-ay with programmable photochemical oligonucleotide: PAP
can be used for.resequencing to detect unknown mutations. On this roicroarray, the wild-type template is 20 indicated. The P.*s are designed according to the wild-type template. The P*s that overlap -with the mutation generate little or no signal indicated as "Low" PAP signal.
= [01731 Fig. 4 shows an example of solid support-based, e.g., microarray-based, resequencing to detect a G to A mutation. Linear PAP is performed with four different dye-labeled ddNTPS as substrates for single-base extensions. P*s have a specific region of 16 nucleotides within the 3' 25 region of the oligonucleotide. Homozygous or hemizygous DNA template is utilized in the example. Sets of four P*s, with identical sequence except for the four ddNIVIE's at the 3' terminus, are synthesized for each nucleotide position on the sense strand of the wild-type sequence. The P* with a ddA at the 3' tennintS generates a PAP signal at the site of the G-A
mutation. The mutation also creates a 15 base "gap" of no PAP signal for the subsequent 30 overlapping 15 sets of Psi For heterozygous mutation, the P*s with ddA and ddG provide PAP
signals. The heterozygous mutation also generates the 15-base "gap" of 50%
signal intensity (which is flanked by signals of 100% intensity). For added redundancy with heterozygotes samples, antisense P*s can be utilized (not shown). An unknown single-base substitution can be determined by combination of the two sets of P*s. Small deletions and insertions can be detected and localized.
[0174] With 100,000 oligonucleotides per microarray, about 12 kb can be resequenced from downstream. and upstream directions. The detection of virtually all mutations requires supplementation of the standard Geniome instrument software. For wild-type sequence, the signal intensities may vary. Certain oligonucleatides will generate a weaker signal due to secondary structure and other factors. The pattern of signal from wild-type samples should be ' distinguished reliably from the pattern generated by a given sequence change.
The preliminary data suggest that almost all mismatches will inhibit aetivation dramatically.
Because of the redundancy, mutations may be reliably distinguished from the wild-type even if a significant minority of single base mismatches does not inhibit activation substantially.
[0175] It is becoming increasingly apparent that in vivo chromatin structure is crucial for mammalian gene regulation and development. Stable changes in chromatin structure often involve . changes in methylation and/or changes in histone acetylation.
Somatically heritable changes in chromatin structure are commonly called epigenetic changes (Russo and Riggs, 1996) and it is now clear that epigenetic "mistakes" or epimutations are frequently an important contributing factor to the development of cancer (Tones and Laird, 1999).
[0176] One of the few methods for assaying in vivo chromatin structure, and the only method with resolution at the single nucleotide level, is ligation-mediated PCR (LM-PCR) (Mueller and Wold, 1989; Pfeifer et al., 1989). LM-PCR has been used to assess chromatin strueture, .
inethylation and damaged DNA. Fig. 5 shows a schematic of LM-PCR in which a DNA lesion in the starting DNA is indicated by a small diamond. LM-PCR involves cleavage, primer = extension, linker ligation and PCR amplification. LM-PAP is similar to LM-PCR except that activatable oligonucleotide P*s are used.
[0177] LM-PCR has proven to be an important technique, now having been used in over 100 published studies (Pfeifer et al., 1999). Many aspects of chromatin stnicture can be determined by LM-PCR, such as the location of methylated cytosine residues; bound transcription factors, or positioned nucleosomes. Importantly, the- structure is determined in cells that are intact and have been minimally perturbed: UV photo-footprinting, for example, is performed by LTV
irradiating tissue cultuTe cells in a Petri dish, immediately extracting the DNA, and performing .
LM-PCR to determine the location of thymidine dimers, the formation of which is affected by bound transcription. factors.

-[0178] Allele-specific LM-PAP can be applied to quantitatively detenrdne the level of in vivo methylation. The background of LM-PCR currently limits reliable estimation of the level of methylation. It is generally considered that 0 %, 50 % and 100% methylation can. be determined, but distinguishing finer gradations is not reliable. With a marked reduction.
in background in LM-PAP, 0%, 20%, 40%, 60%, 80%, and 100% Methylation standards may be distinguished reliably. It will be of particular interest to utilize allele-specific LM-PAP
to examine the level of methylation in imprinted regions, or in active verses inactive X-chromosomal genes in females.
It is anticipated that LM-PAP will decrease the skill and. experience needed to examine chromatin structure, thereby facilitating analysis of chromatin structure by more laboratories.
[0179] LM-PAP has a diversity of applications. It will be of particular-interest to utilize allele-specific PAP to examine differential methylation and chromatin structure in imprinted genes or in active versus inactive X chromosomal genes in females. In addition, the relationship between mutagens, DNA damage, and mutagenesis can be examined.
[0180] In PAP, as described above and illustrated herein, pyrophosphorolysis and is polymerization by DNA polymerase are coupled serially by using pyrophosphorolysis activatable oligonucleotide: In PAP sequencing, the principle is based on the specificity of PAP
and in turn on the base pairing specificity of the 3' specific subsequence.
This property of the 3' specific subsequence can be applied to scan for unknown sequence variants, to determine de novo DNA sequence, to compare two DNA sequences and to monitor gene expression profiling.
[0181] PAP is highly sensitive to mismatches along the length of P* in PAP
with one P* and one opposing unclocked oligonucleotide. The specificity of PAP is also affected by P* length and mismatch. If the allele-specific nucleotide of P* is at the 3' terminus, only the specific allele is amplified and the specificity is not associated with P* length. If the allele-specific nucleotide.
is not at the 3' terminus of P*, the specificity is associated with P* length.
26 mer P* has a 3' specific subsequence of three-nucleotides = within this region any mismatch inhibit the amplification. 18-mer has a 3' specific subsequence of 16 nucleotides.
[0182] Bi-PAP is a form. of PAP. In Bi-PAP with two opposing P*s, each P* has its own 3' subsequence, i.e., within this region any mismatch inhibit the amplification of BiTAP. For example, when the allele-specific nucleotide of the P* pair is at their 3' termini, only the specific allele was amplified, no matter what the lengths of the P*s are 40, 35 or 30 nucleotides. The length of the paired specific subsequence is addition of the P* pair minus one.
[0183] The length: of the paired specific subsequence may be affected by the sequence context and size of each the type of the 3' terminal non-extendible nucleotide; the template sequence, =
the DNA polymerase, other components like ions, and cycling conditions. When the template contains repeated sequences or homogenous polymer runs longer than the length of the P* pair, P* may lose specificity for anchoring.
[0184] Resequencing is the sequencing of a known region to detect unktio-wn mutations. The s property of the paired specific subsequence of Bi-PAP can be applied to scanning for unknown sequence variants or re-sequencing of predetermined sequences in a parallel way.
. [0185] A Bi-PAP resequencing is shown in Figs. 22, 23A, 23B, 24A and 23B.
Briefly, the wild-type sequence can be determined, and any single base substitution can be determined with the type and position. An unknown small deletion and insertion can be detected and localized. = In io order to identify a specific type of deletion or insertion, it is possible to add corresponding Bi-PAPs. For fingerprinting, which can provide infomiation regarding mutation position, fewer numbers of Bi-PAPs can be used.
[0186] The concept of Bi-PAP de novo DNA sequencing makes use of the complete set of paired specific subsequence of the P* pair to identify the presence of the paired specific 15 subsequence in the de novo sequence.
[0187] Bi-PAP de nova DNA sequencing on microarray is shown in Fig: 25.
Briefly, the procedure first collects all the Bi.,PAP amplifications with their P* pairs and then reconstructs the unknown DNA sequence from this collection by ordering the paired specific subsequences.
[0188] For comparison of two DNA sequences to see if they are the same or different, there is a 20 simple way to reduce the number of P* pairs by using an incomplete set of the specific subsequences of the P* pair; By arranging theni in a particular order, it is possible to identify the chromosomal locations as well as the sequences.
[0189] To monitor gene expression profiling, where up to 6x104 to 105 transcripts are expressed and details of the precise sequence are unnecessary, Bi-PAP can be applied. A
set of P* pairs 25 which can specifically amplify unique motifs in genes can be designed for Bi-PAP.
[01901 This property of the base pairing specificity of Bi-PAP can be applied to seat for unknown sequence. variants, to: determine de novo DNA sequence, to compare two DNA
sequences' and to monitor gene expression profiling, A Bi-PAP array is possible. Each pair of two opposing P*s can be immobilized at an individual spot on a solid support, e.g., mieroarray, 30 thus allowing all the Si-PAP reactions to be processed in parallel.
[0191] For PAP, the activatable oligonucleotide has a non-extendible 3' terminus that is activatable by pyrophosphorolysis (hereinafter referred to as a non-extendible 3' tettiainns). Any 3' terminal non-extendible oligonucleotide can be used, if it can hybridite with the template =
=

strand., the 3' terminal. nucleotide can be removed by pyrophosphorolysis, and the activated oligonucleotide can be extended. Examples. of non-extendible 3' terminus include, but are not limited to, a non-extendible 3' deoxyriucleotide, such. as a dideoxynucleotide, or a chemically modified nucleotide lacking the 3' hydroxyl group, such as. art acycionucleofide.
Acyclonucleotides substitute a 2-hydroxyethoxymethyl group for the 2'41.eoxyribofuratosyl sugar normally present in. dN1v1Ps:
[0192] Alternative blocking agents may increase the selectivity of pyrophosphoroloysis for a complete match, thereby further enhancing the selectivity of PAP for detecting rare mutations.
Finally, alternative blocking agents may be less .expensive or more readily autoinatable, thereby 3...o improving the cost-effectiveness of PAP and facilitating PAP
microarray-based resequencing.
[0193] In addition, P*s not blocked with dideoxynucleotides extends. the selection. of DNA
polymerases which can be used for PAP. As demonstrated herein., Family I
polyinerases maybe used for PAP when the 3' terminal non-extendible oligonucleotide contains a dideoxymicleotide = or an acyclonucleotide. Family II polyineraSes may he used for PAP when the 3' terminal non-is extendible oligonucleotide contains an acyclonucleotide.
EXAMPLES =
=
[0194] The invention can be understood from the following Examples, which illustrate that PAP
can be used, to identify a known mutation in a polymorphic site within the human Di dOp-triirie 20. receptor gene.
The effects of the dideoxyoligonucleotide sequences, acyclonucleotide .sequences, DNA polym.erases, PP i concentrations, allele-specific templates, pH, and dNTP
concentrations- were examined: The experiment reported in the Examples Were conducted for proof of principle. The following examples are offered by way of illtistration and are not intended to limit, the invention in any manner. Stan-lard techniques well known in the. art or the 25 techniques specif,cally described therein were utilized. , EXAMPLE I
Preparation of template by PCR
[0195] A 640-bp region of the hi-1mm Di dopamine receptor gene was amplified by PCR with 30 two primers = 5' GAC CTG CAG CAA GGG---AGT CAG AAG 3' (SEQ rD NO:1) wad U = 51.
TCA TAC CGG- AAA GGG CTG GAG ATA 3' (SEQ ID NO:2)) (Fig. 6A). The TU:UT
duplexed product spans AuCleotides 33 to 672 in GenBank X55760 arid the G+C
content is 55.3% A common A to G polymorphism. is located at nucleotide 229, resulting in three =
=

= genotypes of GIG, A/A and G/A (Liu et al., 1995): The PCR mixture contains a volume of 50 pd.: 50 mM KC1, lOnalVI Tris/HCI, pH 8.3, 1.5 DIM MgC12, 200 pM each of the four dNTPs (Boehringer Mamahefin), 0.1 p.M of each primer, 2% DMSO, 1 U of Taq DNA
polymerase = (Boehringer Mannheim) and 250 ng of genomic DNA from GIG homozygete, A/A
homozygote s or G/A heterozygotes. Cycling conditions included: denaturation at 95 C for 15 seconds, amiealing at 55 C for 30 seconds, and elongation at 72 C for one minute, for a total of 35 cycles (Perkin-Elmer GeneAmp PCR system 9600). The PCR product was purified from primers and other small molecules by approximately 10,000-fold by three times of retentiOn on a Centricon 100 microconcentrator (Araicon). The amount of recovered PCR product was determined by . o IN absorbance at 260 run.
Synthesis of P* by adding a 3'-dideoxynucleotide.
01961 .The deoxynucleotide oligonucIeotide w.as synthesized by Perseptive Biosystems 8909 Synthesizer. (Framinsham).and purified by oligopure cartridges (Hamilton) in the City of Hope DNA/RNA Chemistry Laboratory. The 3' terminal dideoxynucleotide was added by terminal 15 transferase. The mixture contained a total volume of 40 p.l: 200 mM
potassium cacodylate, 25 mM Tris/HCI (pH 6.6 at 25 C), 2.5 mM CoC12, 0.25 mg/ml of BSA, 4000 pM of the oligonucleotide, 2.5mM 2'3'-ddNTP (the molar ratio of the 3'-OH terminus to ddNTP was 1:25) Boehringer Mannheim); 125 U of terminal transferase (Boehringer Mannheim). The reaction was incubated at 37 C for 1 hour and then stopped by adding EDTA at 5 mM final 20 concentration. After desalting by using butanol, the dideoxyoligonucIeotide was purified by preparative 7M urea/20% polyacrylarnide gel electrophoresis in TBE buffer (90 mM Tris/borate, 1mM EDTA, pH 8.3) (Maniatis et al., 1982). The amount of the recovered P* was determined by UV absorbance at 260.-nm.
[01971 Since small amounts of =terminated oligonucleotide would result in non-specificity of 25 pyrophosphorolysis, each dideoxyoligonucleotide was 32P-labeled at the 5 terminus by T4 polynucleotide kinase and then. was electrophoresed through a 7M urea/20%
polyacrylarnide gel.
Only P* products were visible even when the gel was overexposed (data not shown). It is estimated that more than. 99.99% of P* contained a dideokynticleotide at the 3' terminus.
Pyrophosphorolysis activated polymerization 30 [0198] A 469-bp region Within. the TIP.I.TT duplexed template was amplified by PAP with oligonncleotides P* and U, or with only one P* (Table I and Fig. 6A). The PU:UP duplexed product corresponds -to nucleotides 204 to 672 in GenBank X55760 and the G+C
content is 55.6%. Unless stated, the PAP reaction mixture contained a total voluble of 25 1 for Tfl DNA

polymerase: 75 mIVI KC1, 20 mM Tiis/HC1 (pH 74), 1.5 mM MgC12, 40 uM each of the four DNTPs (dATP, dTTP, dGTP and dCTP), 0.2 tiM P*, 0.05 iM U oligonucleotide, 300 Na4PPi (the 20 KM stock solution was adjusted by Ha to pH 8.0), 1 Ci of [a-32P1-dCTP
(3000Ciinmole, Arriersham), 1 U of Tfl DNA polymerase (Promega) and 2 ng of TU:UT. For s Tag DNA polymerase, the reaction mixture was the same except for 50 rriM
KC1, 10 mM
Tris/HC1 (pH 7.4); 2.0 mM MgC12 and 1 U of Tag DNA polyinerase (Boehringer Marmheim).
The mixtures of PCR and other controls Were the same except for the primers added. Cycling conditions included: 94 C for 15 seconds,µ 55 C for one minute, ramping to 72 C for one minute and 72 C for two minutes, for a total of 15 cycles. .

= Oligonucleotides used in PAP
= Tem-plate 5 L . .AATCTGACTGACCCCTATTCCCTGCTT GGAAC . . 3 ' (SEQ ID NO: 3 ) = A
Name Oligonucleotide Sequence 51-3' (SEQ ID NO:) . Purpose ACTGACCCCTATTCCCTGCTTb (4) Control _ ACTGACCCCTATTCCCTGCTTG*b ( 5 ) 3' ddG and G allele specificity co-localized DG' ACTGACCCCTATTCCCTGCTTGG-k ( 6) G allele specificity 5' to ddG
= D3G7 ACTGACCCCTATTCCCTGCTTGGG*
(7) G allele specificity 5' to ddG
= D.4G' ACTGACCCCTATTCCCTGCTTGGGG* (8)= 3' ddG
mismatches template Da TCTGACTGACCCCTATTCCCTGCTTG* ( 9 ) DO, With 51 extended bases "
= D6K TGACTGACCCCTATTCCCTGCT
TA* (10) 3' ddA. and A allele-specificity co-localized -;TCATACCGGAAAGGGCTGGAGATA (11). Upstream oligonucleotide =
Name 3' terminal Allele-specific Size Tõ/ Amplification' nucleotide' nucleotide (base) ( C)e Type Match Type From 3' =G allele A allele terminus (bp) DI dT Yes - +1 21 64 Yes Yes . .
ddG Yes G 0 22 68 No No =
_ ddG Yes G -1 23 72 No No D3G7 ddG Yes G -2 24 76 Yes No ddG No G -3 25 80 No No =
D5A. ddG Yes G 0 26 80 Yes No = ________________________ -TK
ddA Yes A 0 24 72 No No =

U. dA Yes - 24 72 Yes Yes a DIG. was produced by adding a G dideOxynucleotide to the 3' tenninus of the D1,4 a dideoxynucleotide at the 3' terminus.
b The T means the 3' terminus is T deoxynucIeotide and G* means the 3' terminus is G
dideoxynucleotide. The bold capital G and A are the G and A bases corresponding to G and A
alleles, respectively. The first base at the 5' terminus corresponds to nucleotide 208 in GenBank X55760.
e The 3' terminal base is a deoxynucleotide or dideoxynucleotide, and creates a match (Yes) or a mismatch (No) with the corresponding base on the cornplementary strand of the template.
d The allele-specific nucleotide is G or A and its distance to the 3' temnnus is assigned: 0= at the 3' terminus +1 = one base downstream from the 3' terminus, -1 = one base upstream from the 3' terminus, -2 = two bases upstream from the 3' terminus, and -3 three bases upstream from the 3' terminus.
e The Tõ for oligonucleotides was estimatecf to be 4 C X (G + C) + 2 C X (T +
A) at 1 M NaC1 (Miyada. and Wallace, 1987).
= The amplification with TT and one P* or with only one P*. =
[0199] The reaction was electrophoresed through a standard 2% agarose gel. The gel was stained with ethidium bromide for UV photography by a CCD camera (Bio-Rad Gel Doc 1000), dried and subjected to Kodak X-OMATTm AR film for autoradiography.
" 20 Restriction digestion [0200] Each of the three restriction endonucleases of Acil (5'CYCGC3'/3'GGCAG5') Rad (51PyTGGCCPu.31/ 31PuCC.GGAPy5t) and Eco0109I (51PuGYGNCCPy3r/ 3'PyCCNGAGPu5') has a restriction site within the PU:UP duplex. The G/G alleles were amplified by PAP with D5G*
and U; PCR amplification with DI and U Was used as the control. 40 p.1 of the PAP reaction and 2 ul of the PCR reaction were purified and concentrated with a Cennicon0 100 microconcentrator, and the products digested by the restriction endonucIease:
2.5 U of AciI in =
lx NE buffer 5; or 3 T.; of EaeI in.1X NE buffer 1; or 30 U of Eco0109I in NE
buffer 4 with .
BSA (all of the above enzymes and buffers from New England BioLabs). 10 p1 of the reaction was incubated at 37 C for 2 hours. The digestion reaction was electrophoresed through a standard 2% agarose gel as described above.
Principle of PAP
[02011 Tfl and Tag DNA polymerases were shown to contain pyrophosphorolysis activity. 71/
DNA polymerase was utilized to detect the G allele at nucleotide 229 of the DL
dopamine receptor gene (I-Au et al., 1995) (Fig. 6A). P* was synthesized with either ddG or ddA. at the =
Tterminus (see Table 1). The nernain.al dideoxynucleotide inhibits direct extension by . polymerization, but can be removed by pyrophosphorolysis in the presence of pyrophosphate (pp) when the P* is specifically hybridized with the complementary strand of the G allele. The degraded oligonucleotide can be extended by polymerization in 5'-3'direction (Figs. 6B and 6C).

[0202] The enhanced specificity of PAP relative to PASA is provided by serially coupling pyrophosphorolysis and= polymerization.
Significant nonspecific amplification requires mismatch pyrophosphorolysis and misincorporation by DNA poly-merase, an extremely rare event (Fig. 7).
5 Specific amplification with D5G* and D3G*
[0203] PAP was perforined With two oligonucleotides (13* and U), 1)7 DNA
polymerase and DNA template of the GIG and A/A alleles. Multiple P* were tested (Table 1).
D5G* (the allele-specific nucleotide and dideoxynucleotide are co-localized to the 3' terminus and D3G* (the allele-specific nucleotide is two bases from the 3' terminus) specifically amplified the G allele in o the presence of PP; (Fig. 8A). Without added PP, no specific product .was observed with D5G*, indicating that added PP i was an essential component for PAP (Fig. 8B, lanes

6 and 15). Faint products with D3G* in lane 4 and with D4G* in lane 5 were observed (Fig. 833) (see below).
Effects of pH, [PP and {dNTP} and enzyme [0204] Each of the above parameters was examined. PAP was most efficient at pH
between 7.4 is and. 7.7, at {PP] between 200 gM and 400 M, and at [dNTPs] between 25 gM and 50 Al (Table 2). Taq DNA polymerase can substitute for 7:fl with similar efficiencies (Table 2).

Parameters affecting PAP
Parameter PAP efficiencYb D5G*-U D3G*-U
8.1 pHa= . 7.9' -

7.7 ----7.5 -H 111 -7.4 -H-7.15 +

ppia GEM) 600 400 ++ lIM

All d_NTPs 100 changed' = 36 50 ++ +-H-25 -H- ++++

dGTP

chanaeec dATP 50 changed' G allele and PPf Taq DNA A allele arid PPi polymerase G allele and no PPi ___________________________________________________ =
Tfl DNA polymerase was used to amplify the GIG alleles under the conditions in Materials and Methods, except for the factors indicated b The PAP efficiency is indicated as: -, no specific product(s); + very weak specific product(s);. +, weak specific product(s); +4-, moderate specific product(s);
strong specific product(s); I ____ , very strong specific product(s).
The indicated concentration was changed but the others were kept at 200 M.
Identity of specific products [02051 In order to confirm the identity of the specific products, restriction endon_uclease digestion was performed (Fig. 9). Each of the three restriction endonucleases of Acil, Ead and Eco01.09 has a restriction site with the PU:UP duplex. The expected restriction fragments were found. Similar results were observed with D36* and U.
[02061 The specific products of PAP with D5G* and. U revealed two specific bands on the = agarose gel, i.e., PU:UP and UP; because U was more efficient than D5G*, under our amplification conditions. In order to confinn this, the G/G alleles were amplified by PAP using T77 DNA polymerase with D5G* and U as previously. The products were denatured and electrophoresed through a denaturing polyacryla-mide gel. Only one specific band in single-stranded form was observed, indicating that the specific PAP products contain the duplexed and single stranded segments. The same result was observed with D3G* and U.
Linear PAP
[0207] PAP was performed for linear amplification with only one P* from the GIG and A/A
alleles in the presence of PPi. The specific products of PAP were obtai-ned With D3G* and with D5G*, but not with the other P* (Fig. 10, lanes 4 and 6). The efficiency of P*
was affected by the oligonucleotide size, the 3'-terminal dideoxynucleotide and the position of the allele-specific nucleotide.

[02081 Figs. 6A-60 show schematic of PAP. Pig. 6A. A duplexed DNA template TU:UT is amplified with two oligon.ucleotides P* and U, Tfl DNA polymerase, dNTPs, pyrophosphate and {06-3211}-dCTP. P* = pyrophosphorolysis activatable oligonucleotide. hi this example P* is D5G* and TU:UT is a 640-bp sapient of the dopamine D1 receptor gene. Fig. 6B.
D5G* has a s G
dideoxynucleotide at the 3' terminus, and it is specific to the complementary strand of the G
allele, but mismatches the A allele at the 3! terminus (Table 1). Removal of the dideoxy G by pyrophosphorolysis is followed by polymerization for each amplification. Fig.
6C.
Autoradiogram of PAP from the GIG, A/A and G/A genotypes. When the G allele is present, the radioactively labeled specific products of 469 bases (duplex PU:UP and.
excess antisense o strand UP) are produced, since the low rate of pyrophosphorolysis by Tfl polymerase implies that Oligonucleotide U has a much higher efficiency than oligonucleotide P.
Electrophoresis for a longer period separates PU:UP from UP. Other products of UT and UT:TU
are indicated.
Note that TU:UT derives from annealing of excess radioactively labeled UT with non-radioactively labeled TU original template. PAP was also perfonned with D3G*
and U from the is GIG, .AJA and G/A genotypes, and similar results were obtained.
[02091 Figs. 7A.,713 show enhanced specificity of PAP with D5G*. The specificity of PAP is compared with that of PASA to exponentially amplify a template pool of G and A
alleles. Fig.
7A. The 'specific amplification of PASA derives from the high efficiency of primer extension when the primer matches the G allele. The nonspecific amplification results from mismatch 20 extension from the A allele. When this occurs, it results in an efficiency substrate for further Pmplification. The thickness and position of the arrow represent the amplification efficiency in each cycle. Fig. 7B. The specific aniplification of PAP from the G allele occurs at high efficiency. Two types of nonspecific amplifications originate from the A
allele: (i) nonspecific Amplification can occur at low efficiency by mismatch pyrophosphorolysis resulting in a A:T
25 hoi33o-duplex PU:UP product, which is not an efficient template for subsequent amplification;
(ii) nonspecific aMplification can occur at extiemely low efficiency by both mismatch pyrophosphorolysis and misi-ncorporation to produce a G:T hetero-duplex PU:UP
product, but once i occurs, it provides an efficiency template for subsequent amplification. A similar tendency of nonspecific amplifications is suggested for iii-ear amplification by PAP with only 30 D5G*. It should be noted that allele-specific nucleotide of P*, such as D3G*, may be near but not at the 3' terminus. In that ease nonspecific amplification of PAP requires both mismatch pyrophosphorolysis and mismatch extension: While both variations of PAP should have higher specificity than PASA, the highest specificity is. predicted when the 3' terminal dideoxy nucleotide is also the allele-specific nucleotide.
[02101 Figs. 8A-8B show specific amplification with D5G* and D3G*. PAP was performed in the presence (Fig, 8A) or absence (Fig. 8B) of added PP i with two oligonucleotides for s exponential amplification. The oligonucleatides are listed in Table 1.
Extension controls with only U identify the positions of TU:UT and UT. Extension. controls with Di identify the position of PU. PCR controls of Di and U identify the positions of PU:UP and PU:UT. Only 20% of the extension reaction with Di and the PCR reaction were loaded relative to other lanes.
[02111 Fig. 9 shows restriction endonuclease digestion. TO show specificity of PAP, samples 1.o from the experiment shown in Fig. 8 were digested with Acil, Eael and.
Eco01091 restriction endonucleases. Each enzyme has a restriction site within PU:UP. PAP amplified the GIG
alleles with D5G-* and U, and 5% of PCR reaction with Di and U were taken as control. Acil = produces a 236 bp and a 233 bp fragments from PU:UP and a 407 bp and a 233 bp fragments from TU:UT. EaeI produces a 289 bp and a 180 bp fragments from PU:UP and a 460 bp and a 1.5 180 bp fragments from TU:UT. Eco01.09I produces a 348 bp and a 121 bp fragments from PU:UP and a 107 bp, a 412 bp and a-121 bp fragments from TU:UT. The arrows indicate the digestion products expected from PU:UP.
[02121 Fig. 10 .shows linear PAP. PAP was performed with only one P* in the presence of added PP. 20% of the reaction with Di was loaded relative to other lanes (lanes 1 and 10). No 20 = no oligonucleotide added.
Enhanced specificity of PAP with D5G*
[02131 Example 1 provides evidence that pyrophosphorolysis followed by polymerization may be used to increase the specificity of PASA. Significant nonspecific amplification requires the serial coupling of the two types of errors (Fig. 7). The mismatch pyrophosphorolysis rate to 25 remove a mismatch deoxynucleotide at the 3' terminus, expressed as the removal rate of an incorrect versus a correct cINMP, was reported at less- than 10-5 for' T7 DNA
polymerase (Komberg and Baker, 1992; Wong et al.., 1991). The misincorporation rate to create a substitution mutation by polymerization, expressed as the incorporation rate of an incorrect versus a correct dNMP, Was reported as to be 10-5 for 17 DNA polymerase and to be 10-4 for 30 E.coli DNA polymerase I (Kornberg and Baker, 1992; Wong et al., 1991; Bebenek et al., 1990).
Sitailar results were reported for Taq DNA polymerase and for 3'-5' exotablease-deficient mutants of 17 DNA polymerase and .E. colt DNA polymerase I (Kornberg and Baker, 1992;
Wong et al., 1991; Eckert and Kunkel, 1990). The specificity due to the (i) nonspecific =

amplification in. PAP with D5G* is estimated to be 10-5 per cycle, if the mismatch pyroph.osphorolysis rate of a ddNMP is the same as dna. The specificity due to the (ii) nonspecific amplification is estimated to be 3.3x1041, if the mismatch pyrophosphordysis and the misincorporation are serially coupled.
s Essential components of PAP
[0214] Each P* was tested by utilizing 7:11 or Taq DNA polymerases to amplify the GIG and A/A alleles. The specific amplification requires the presence of PP; and allele-specific template.
In. addition, the amplification efficiency is affected by the oligonucleotide size, the 3' terminal dideoxynucleotide, the position of the allele-specific nucleotide relative to the 3' terminus of P*.
o [0215] It is not clear why DIG* and D2G* did not generate the specilic signals, but it may be related to a threshold stability of duplex between P* and the template. D6A*, which contains A
dideoxynucleotide at the 3' terminus, did not generate the specific signal, which may be associated with different incorporation efficiencies of ddNTPs by polymerization. Klenow fragment of E. coil DNA polymerase I, Taq DNA poiyirterase and Araq DNA
polymerase as incorPorate ddGTP more efficiently than other ddNTPs (Sanger et at, 1977; Tabor and Richardson, 1995; Vander Horn et al., 1997). The rate of ddNTP incorporation also varies depending on the template sequence and can. be 10-fold higher at some bases relative to others (Sanger et at, 1977). Another possibility is that D6A* is shorter in size with a lower - [0216] In. PAP without added PP, very faint false signals were generated with D3G* and with 20 D4G* (Fig. 8B). One possibility is that oligonucleotide dimers can form and trigger nonspecific pyrophosphorolYsis of P* in later cycles after "endo-" PP; is released from the by-polymerization to generate UT. 3Ttennin.al degraded D3G* and D4G* can be hybridized and extended as false signal. Oligonucleotide climers were observed with D3G* and D4G*. Another possibility with D3G* is that the specific pyrOphesphorolysis can occur in later cycles after "endo.," PP; is 25 released. A third possibility is that D30* and D4G* were contami-nated by minimal D3 and D4 which were not fully added by G dideoxynucleotide at 3' termitit . Coniparison with other technologies [02171 A number of methods for enzymatic nucleic acid amplification in vitro have been developed and can be adapted"to detect known sequence variants. These include polymerase 30 chain reaction (PCR) (Saud et al., 1985; Saila et at, 1988), ligase chain reaction (LCR) (Landegren et al.; 1988; Barmy, 1991) and rolling circle amplification (RCA) (Lizardi et al., 1998; Ban& et al., 1998). PAP is different in many ways: i) pyrophosphOrolysiS
and polyrnerization are serially cOupled for each amplification, ii) there is at least one =

dideoxyoligonucleotide for PAP. Other chemically modified nucleotides lacking the 3"-hydroxyl group at the 3' terminus, such as acyclonucleotides can serve the same function (see Example 12 below), iii) one format is for linear amplification and the other is for exponential amplification, iv) PP' is necessary for the amplification; v) significant nonspecific amplification requires both mismatch pyrophosphorolysis and misincorporation, vi) PAP can detect known point mutations and greatly increase the specificity to detect an extremely rare mutant allele from the wild-type allele.
[0218] The mechardstic basis is that two or more reactions are serially coupled for amplification with increased specificity. The key component of PAP is a pyrophosphorolysis activatable 10 oligonucleotide. The blocked 3' terminus in these experiments is a dideoxy nucleotide, but any non-extendible nucleotide susceptible to pyrophosphorolysis could in principle be substituted.
Indeed, any enzyme that cleaves an oligonucleotide 5'= to a mismatch could serve the same function as pyrophosphorolysis activation. For example, a blocked oligonucleotide including the methylated recognition sequence (such as GmATC) is annealed to its target with. the 15 unmethylated recognition sequence, then restriction endonuclease (such as Dpnl) can only cleave the methylated site and so activate the oligonucleotide for extension.
If a mismatch is located 5' to the cleavage site, significant nonspecific amplification requires the serial coupling of mismatch cleavage and a misincorporation, which is a rare event.
Activatable oligonucleotides may also be combined with "minisequencingu primer extension.
This may 20 provide a more specific assay for detection of single base changes that might be particularly amenable to chip technology in which specificity can be a problem (Syva-nen, 1999).
. Demonstration that PAP can occur in the linear format (Fig. 10) supports the feasibility of this approach.
[02191 Nucleoside triphosphates and 2'-deoxynucleoside triphosphates or their chemically 25 modified versions may be used as substrates for multiple-nucleotide extension by PAP, i.e., when one nucleotide is incorporated the extending strand can be further extended. 2',3'-dideoxynucleoside triphOsphates or their chemically modified versions that are terminators for further extension may be used for single-nucleotide extension. 2',3'-dideoxynucleoside triphosphates may be labeled with radioactivity or fluorescence dye for differentiation from the 30 3' terminal dideoxynucleotide of oligonucleotide P. Mixtures of nucleoside triphosphates or 2'-deoxynucleotide triphosphates and 2`,3'-dideoxyaucleoside triphosphates may also be used, 102201 In PAP, specific nucleic acid sequence is produced by using the nucleic acid containing that sequence as a template. If the nucleic acid contains- two strands, it is necessary to separate the strands of the nucleic acid before it can be used as the template, either as a Separate step or . .simultaneously. The strand separation can also be accomplished by any Other suitable method including physical, chemical or enzymatic means: =
[0221] When. it is desired to produce more than one specific product.dom the original nucleic acid Or mixture of nucleic acids, the appropriate number of different oligonucleotides are utilized. For example, if two different. specific products are to be produced exponentially, four oligonucleotides. are utilized. Two of the oligonucleotides ( are specific for one of the specific nucleic acid sequel-ides and the other two oligonucleotides ( P*:L1) are specific for the .
second specific nucleic acid sequence. In this manner, each of the two different specific o sequences can be produced exponentially by the present process.
[02221 The DNA or RNA may be single- or double-stranded, may be a relatively pure species or a component of a mixture of nucleic acids, and may be linear or cirtular. The nucleic acid or acid S may be obtained from. any source, for example, from plasmid,. from cloned. DNA or RNA, . , .
or from natural DNA or RNA from any source, including bacteria, yeast, viruses, and higher organisMs such as plants or animals. DNA Or RNA may be extracted from blood, tissue material such as chorionic villi or amniotic cells by a -Variety Of techniques such as that described by Maniatis et al. (1982).
[0223] The P* oligonucleotides are selected to be. "substantially"
complementary" to the different strands of each specific sequence to be amplified. Therefore, the P*
oligOnucleotide 20 seqUence need not reflect the exact sequence of the template. For example, a non-complementary nucleotide segment may be attached to the 51-end of the P*
oligdnucleotide, with the retnailidet of the P* oligon.utleotide sequende= being complementary to the strand.
AlternatiVeiy, non-compleinentary bases or longer sequences Can be interspersed into the P*
oligorinclecitide, provided that the P* oligonticleotide sequence has sufficient coMpleMentarity with the sequence of the strand to be amplified to hybricii76 therewith and form a template for 'synthesis of the extension product of the other P* Oligonucleotide. As used in the claims, the = term. "complementary" should be understood to mean "substantially complementary," as.
discussed herein.
[02241 Any specifid nucleic acid' sequence can be produced by the present prOceSs. It is only necessary that a sufficient number of bases at both ends of the sequence be known in Sufficient =
detail so that.tWO oligonucleotides can hybri1i7e to different strands of the desired sequence at relatiVe pOsitIonS along the sequence. The greater the knowledge about the bases at both ends of the Seqtiende, the greater can be the specificity of the oligliticleotides for the target nucleic. acid =
=

= 42 -sequence, and thus the greater the efficiency of the process. It will be understood that the word oligonucleotide as used hereinafter may refer to more than one oligonucleotide, particularly in the case where there is some ambiguity in the information regarding the terminal sequence(s) of the segment to be amplified.
[0225] The present invention can be performed in a step-wise fashion where alter each step neW
= reagents are added, or simultaneously, where all reagents are added at the initial step, or partially = = step-wise and partially simultaneOus, where fresh reagent is added after a given nuMber of steps.
The simultaneous method may be utilized when an enzymatic means is used. for the strand separation. step: In the simultaneous procedure, the reaction mixture may contain the io strand-separating enzyme (e.g., helicase), an appropriate energy source for the strand-separating enzyme, such as ATP. Additional materials may be added as necessary.
[0226] The nucleic acid polymerase may be any compound or system that will function to accomplish the amplification. Suitable enzymes for this purpose include, for example, 1)7 DNA
polymerase, Taq DNA polymerase, E. coil DNA polymerase I, Klenow fragment of E. coil DNA
is polymerase I,. T4 DNA polymerase, T7 DNA polymerase, other available DNA polymerases, RNA polymerases or their variants, reverse transcriptase or its variants, and other genetic engineered versions. It is predicted on the basis of the relationship between reverse and forward reactions that. a DNA polymerase will have high and even pyrophosphorolysis activity for the P*
activatable oligonucleotide, if it incorporates dciNTPs efficiently (compared with dNTPs) and 20 evenly (compared among the four ddNTPs). Of all the DNA polymerases, the genetic engineered version may be the best in the future, such as ThermoSequenase (Vander Horn et al., = 1997). Generally, the synthesis will be initiated at the 3' end of each oligonucleotide and proceed in the 5' direction on the template strand. However, inducing agents which initiate synthesis at the 5' end and proceed in the other direction can also be used in the PAP
method as described 2s above.

Preparation of template by PCR
[02271 A 640-hp region of the human DI dopamine receptor gene was amplified by PCR with o two primers (T = 5' GAC CTG CAG CAA GGG AGT CAG AAG 3' (SEQ ID NO:1) and U= 5' TCA TAC CGG AAA GGG CTG GAG ATA. 3' (SEQ ID NO:2)). The MITT duplexed product spans nucleotides 33 to 672 in GenBank X55760 and the G-FC content of the product is 55%.. A
common A to G polymorphism is located at nucleotide 229, resulting in three genotypes of G/G, = A/A and G/A. The PCR volume is 50 Ill: 50 mM KCI, 10 inM Tris/HC1, pH

8.3, 1.5 mM MgC12, 200 gM each of the four dNTPs, 0.1 1.1M of each primer, 2% DMSO, 1 U of Taq DNA
polymerase (Boehrhiger Mannheim) and 250 ng of genomic DNA from GIG
homozygote, A/A
.homozygote or G/A heterozygotes. Cycling conditions included: denaturatiOn at 94 C for 15 s sec., annealing at 55 C for 30 sec., and elongation at 72 C for one min., for a total of 35 cycles with a GeneAmpTm PCR: System 9600 (Perkin-Elmer Applied Biosystems). The PCR
product was =
purified from primers and other small molecules by approximately 10,000-fold by three times of retention on a Centricons 100 microconcentrator (Amicon). The amount of recovered PCR
product was determined by UV absorbance at 260 rim:
io Synthesis of P* by adding a 3'. dideoxynucle,otide.
102281 The deoxynucleotide oligonucleotide was synthesized -by Perseptive BiOsystems 8909 Synthesizer (Framin.sham) and purified by oligopurecartridges (Hamilton) in the City of Hope DNA/RNA. Chemistry Laboratory. The 3' terminal dideoxynucleotide was, added by terminal transferase. The mixture. contained a total volume of 30 ill: 100 mM potassium cacodylate (pH
. is 7.2), 2.0 mM CoC12, 0.2 mM DTT, 2500 pM of the oligonucleotide, 2 inM
2`,. 3'-ddNTP (the . =
molar ratio of the 3'-OH terminus to ddNTP was 1:24)(Boehringer Mannheim), 100 U of terminal transferase (GIBCO BRL). The reaction was incubated at 37 C for 4 hr and then stopped by adding EDTA at 5 ni.M final concentration. After desalting using a Centri-spinrm 7ooltirrin .(Princeton Separations), P* was purified by preparative 7 M
urea/20% polyacrylamide 20 gel electrophoresis in TBE buffer (90 mM Tris/borate, 1 mM EPTA, pH
8.3) (Mattiatis..et al., 1982). The amount of the recovered P* was determined by UV abSorbance at 260 tun.
= [02291 Since small amounts of unterminated oligonucleotide would result in rionspecificity of pyrophosphorolysis, each P* ra 32P-labeled at the 5' terminus by T4 polyttucleotide kinase and then was electrophoreSed through a 7 M urea/20% polyacrylamide gel. Only P*
products were 25 visible even when the. gel =was' overexposed. It is estimated that More than 99.95% of P*
contained a dideoxynucleotide at the 3! terminus: The purity of P* wat supported by the abseuce of PCR. product or PAP product at pH 8.3.
Pyrophosphorolysis activated polymerization . [0230]. Regions from 445 to 469 bp within the TU:UT duplexed template were amplified by 30 RAF with oligonucleotides P* and U, or with only P*. The PU:UP
duplexed product corresponds to nucleotides 204-228 to 672 in GenBank X55760 and its (3i-C
content is 56%.
The PAP reaction mixture contained a total volume of 25 RI: SO mM =1, 10 mM
Tris/HCI (pH
7.6), 1.5 mM MgC12, 100 uM each of the four dNTPs (dATP, dTTP, dalP and dCTP), 0.1 ILM

P*, 6.1 1.1M U oligonucleotide (TCATACCGGAAAGGGCTGGAGATA (SEQlD NO:2)), 300 1.1M Na4PPi, 2% DMSO, 1 CI of {a-32P] dCTP (3000CM:1=01e, Amersham), 1 U of = AmpliTaqFS DNA polymerase (PE Applied Biosystems) or 0.5 U of each of ArapliTaqFS and 'Tag DNA polymerases, and 10 ng of TU:UT. ThermoSequenase (Amershani Pharmacia) was s also tested under the same conditions except for 8U ThermoSeqnenase or 4U
ThermoSequenase plus 0.5U Taq and 2.5mM MgC12. The cycling conditions included: denaturation at 94 C for 10 sec., annealing at 60 C for 1 min. (at 55 C for TheftuoSequenase), and elongation at 72 C for 2 .
min., for a total of 15 cycles.
[02311 The product was. electrophoresed through a standard 2% agarose gel. The gel was stained lo with ethidiura bromide for UV photography by a CCD camera (Bio-Rad Gel Doc 1000) and Multi-Analyst software, dried and subjected to Kodak X-OMATrm AR film for autoradiography. The PAP. yield was quantitated with a PhosphorImager with lmageQuantim software (Molecular Dynamics) as the total number of pixels in the PCR band mints the = background, indicated as a random units 15 Enhanced PAP efficiency [0232] In Example 1, only the P* with ddG at the 3! terminus was amplified using native V/ or Taq. DNA polymerase. AmpliTaqFS and ThoimoSequenase DNA polyinerases were found to achieve much higher PAP efficieney with much less discrimination against any kind of dideoxynncleotide (ddAMP, dd(MP, ddGMP or ddCMP) at the 3' terminus of P*. For example, 20 P*(212)18G and P*(212)18A ,. which are 18-mers of the dopamine DI
receptor gene but have ddGMP and ddAIVIP at the 3' termini (Table 3), specifically amplified the G
and A alleles, respectively. Their yield ratio was 1.4 (compare lanes 9 with 11 in Fig. 11B), and scis P*(212)18G is estimated to be 4% more efficient per cycle than: P*(212)18A .
Another =
=
P*(228)26A-24 =5' TAGGAACTTGGGGOGTOTCAGAGCCC*. Y (SEQ ID NO:12), which is as a 26-tner With. ddCMP at the 3' terminus, was amplified. as efficiently as a primer without ddCMP at the 3' terminus, and the yield was estimated to be increased 1,0.00 fold compared with that by using Tfl or Taq. Moreover, PAP amplified segments directly .from taiman genoraic DNA.
. .
=
= .

=
. . ,.
' = .
-. TABLE 3 PAP specificity affected by P* length and mismatch , _______________________________________________________________________________ ________ Name Sequence (SEQ ID NO:) Mismatch T.
Noise base ( C)d ratio - Type Distance' -Template Strand G _ 51...AATCTGACTGACCCCTATTCCCTGCTT GG1AC...3' (3) A
_______________________________________________________________________________ _________ _ -11*(204)26G" 51tCtgac H
tgACCCCTATTCCCTGCTTG*b (13) G 0 - 80 0.0 _ o --P*(208)22G 51actgACCCCTATTCCCTGCTTG* (14) G
0 68 0.5 N.) P*(210)20GY 5'tgACCCCTATTCCCTGC1'TG* (15) G -0 62 - 0.1 ko 1-, P*(212) 1869 . 5 'ACCCCTATTCCCTGCTTG*
(16) G 0 56 0.3 - .4 1-, _ (A_ -P*(216)26G' 5'ctattcccTGCTTGGGAACTTGAGGG* (17) G -12 80 107.1 N.) _______________________________________________________________________________ _________ _ P*(220)22G-12- 5'tcccTGCTTGGGAACTTGAGGG* (18) G
-12 70 95.5 (xi 1-, 1-, P*(222)20G41 5'ccTGCTTGGGAACTTGAGGG* (19) _ G . -12 64 75.8 1 1-, P*(224)18G"" 5 ' TGCTTGGGAACTTGAGGG* (20) G
42 56 7.0 , P*(206)26A-2- 5'tgactgacCCCTATTCCCTGCTTAGG* (21) A -2 80 - 30.4 .
P*(210)2.2A-2--- 5'tgacCCCTATTCCCTGCTTAGG* (22) A
-2 68 3.3 .
P*(212)20A-4 5 ' acCCCTAT TCCCTGCTTAGG k ( 2 3 ) A -2 62 2.0 P*(214)18A-2 5 ' CCCTATTCCCTGCTTAGG* ( 2 4 ) A -2 56 0.0 'P*(206)26G-9 5'tgactgacCCCTATTCGCTGCTTAGG* (25) __ C-+G -9 80 95.0 -P*(210)220-9 = 5 1 tga cCCCTATTCGCTGCTTAGG* ( 2 6 ) C--->G -9 ^ 68 88.1 ' _ P*(212)20G-9 5'acCCCTATTCGCTGCTTAGG* (27) C---)-G- -9 62 49.5 -=
'P*(214)18G-9 5' CCCTATTCGCTGCTTAGG* (28) C-+G 56 4.7 P*(206)26Y1 51tgactgacCCTTATTCCCTGCTTAGG* (29) C-->T -15 78 89.0 P*(210)22T-15- 51t gacCCTTATTCCCTGCTTAGG* (30) C--->T -15 66 47.8 13*(212)20T-15 5 ' acCCTTATTCCCTGCTTAGG* (31) C-+T -15 60 - 3.4 P*(214) 18T-15 5 I
CCTTATTCCCTGCT,TAGG* (32) C-->-T -15 54 0.0 - aP*(204)26Gu is a P* with a G dideoxynucleotide at the 31 terminus. "0"
means the allele-specific base is at the 3' terminus:, The -first base at 5' terminus corresponds to nucleotide 204 in GenBank X55760. Its length is 26 bases.
b The bold G or A are the G or A allele specific base and the underlined base is designed mismatch.
The distance from the 3' terminus to the allele-specific base: "0 "----,- at the 3' terminus, -3 = three bases from the 3' terminus.
d The Tin for oligonuckotide was estimated to be 4 C X (G + C) + 2 C X
(T + A) under condition of 1M NaCI . The length of 0 each P* is'18 bases.
The noise rate of PAP (%) is defined as the relative yield of non-specific allele product to specific allele product by the same 13", or as the relative yield of the designated. mutated P* to its native form by using the same template. A specific signal is denoted as <10% noise rate.

=
=
=
=
. = =

[02331 AmpliTaeS has two mutations compared with native Taq. One mutation in the 5' nuclease domain eliminates 5'-3' exonuclease activity and the second mutation F667Y in the active site (Innis and Gelfand, 1999). ThernaoSequenase has the same mutation F667Y in the active site but a deletion of the 5'-3' exonuclease domain (Tabor and Richardson, 1995; Van der s Horn et al, 1997). They do not distinguish between dNTP and ddNTP for incorporation. The pyrophosphorolysis of ddNIVIPs, which is the reverse reaction, is supposed to be much higher and less discriminated by these enzymes. Although either AmpliTtiqFS or ThermoSequenase DNA. polyno.erases used was formulated to contain a thermostable pyrophosphatase (manufacturers' instructions) that can hydrolyze PP i in the reaction so as to decrease PAP
o efficiency, PAP was still amplified under our conditions. AmpliTaeS and ThermoSequenase DNA polymerases will work better in their pure form without the contaminated pyrophosphatase:
The 3' specific subsequence of P*
[02341 Various Ps Were examined with different lengthS and mismatches using AinplintqFS
is (Table 3). The effect of length and mismatch on PAP efficiency is expressed as the relative yield (%) between two P*3 of different lengths from the same template (Fig. 12), which varied from 0.0% to 201.5% with each two to four less bases in length. The specificity of PAP is also affected by P47 length and mismatch (Table 3). 'The noise rate N is defined as the relative yield of the mismatch product to the match product, and a specific signal is scored with <10% noise 20 rate. If the allele-specific base of P* was at the 3' terminus, only the specific allele was amplified and the specificity was not associated with P* length (Fig. 12A). If the allele-specific base was not at the 3' terminus of P*, the specificity was associated with P*
length. Any non-3'-terminal mismatch in the 18-mer P*, which was up to 15 bases from the 3' terminus, caused no amplification (Figs. 12-12E), but even two such mismatches in the 26-mer 25 caused non-specific amplification.
= [02351 The 18-mers were further examined using "stacked" P*s, which span the allele-specific base at different positions (Fig. 13 and. Table 4). The noise rate eio varied from 9.0% to 7.1%.
. The length of the 3' specific subsequence was 13 bases.

PAP specificity with differently positioned. P*s Nunn Se4ne (SEQ ID NO:) Template = 8.1 GACTGACCCCTATTCCCTGCTT-GGAACTTGAGGGGTGTC . . . 3' (33) A
P*(21.2) 1.8dr¨ 5 'ACCCCTATTCCCTGCTTG* (16) P*(212)18Ag 5 'ACCCCTATTCCCTGCTTA* (34) P*(214)18A"` 5 'CCCTATTCCCTGCTTAGG* (24) P*(218) 18G" 5' TTCCCTGCTTGGGAACT* (35) P*(221) 18G" 5 CCCTGCTTGGGAACTTGA* (36) P*(224) 18if" 5' TGCTTGGGAACTTGAGGG* (37) Name 3' terminal Allele-specific Tm ( C) Noise rate (%)a dideoxy base =
Type Dist Exponential Linear PAP template ance PAP
P*(212)1Gu ddG- G 0 56 2.7 0.0 P*212)18.Au ddA A 0 54 3.8 1.1 - =
P*(214)18A ddG A -2 56 4.7 0.0 P*(218)18G-6- ddT G -6 54 0.0 0.0 . = P*(221)18G" ddA. G -9 56 1.7 1.7 P*(224)18G-" ddG G -12 56 7.1 0.6 a The amplificationfrom the G and A templates by PAP with two oligonucleotides or linear PAP
. with one P*. The nOise rate of PAP (%) is the relative yield of the non-specific allele product to = the specific allele product [02361 Similar results were obtained by using 1:4s whfch match and mismatch the G allele at different positions (Table 5). The noise rate with one mismatch was various from 0.8% to 5.6%.
The length of the 3' specific subsequence Was 16 bases. The noise rate with two mismatches was 0% (compare lane 2 with lanes 10-15 in Fig. 14).
=

.= . , , . .
. 1 .
. PAP specificity with differently niinaatched P*s .
.
_ Name The 3' terminal iVrisinatcha Noise rate (%)''' Sequence (SEQ ID NO;) dideoxy . TB, (.c) *
Exponential Linear Type Distance = Reg' PAP
__ 14(212)18G 51ACCCCTATICCCTOCTrG* (16) = Dda-- 56 -1 1.0 0.0 _ P*(212)18A,' -51.ACCCCTATTCCCTG4TTG** (38) DdG C-A. -3 54 1.3 0.0 -._ P*(212)18G' = 51ACCCCTATTCCGTGCTTG* (39) DdG ' C-G -6 56 = 0.8 0.6 o P*(212)18C'r VACCCCTA,TCCCCTGCTTG* (40) . DdG ' T-C -9 = 58 . 1,8 0.4 0 .
1..) ko _ P*(212)18G'" -51ACCCCGATTCCCTGCITG* (41) DdG T-G -12 58 .5.6 :1.7 1-, -.3 -p- 1-, P*(212)18Tb YACTCCIATTCCCTG e la G* (42) DdG C-T- -15 54 3.3 12 o "
tv o . a match or mismatch with the G allele.
-b noise rate (%) is the relative yield between a mismatched P''' and P*(212)18G with the G allele-specific template. 01 1-, 1-, 1-, . . . .
=
=

[0237] Linear PAP was examined using only 18 met P*s and higher specificity was observed - with lower noise rate (Tables 4 and 5). Linear PAP takes a different mechanistic pathway in Which every non-specific product is generated from the starting template which requires mismatched pyrophosphorolysis with the 3' terminal mismatched P*, or both mismatched pyrophosphorolysis and mismatched extension with the non-3' terminal mismatched P.
[0238] PASA was performed with 17-mer primers without adding a ddNMP at the 3' terminus (see Tables 4 and 5). A mismatched 17-mer primer strongly amplified a nonspecific product with 30% noise rate when the mismatch was as near as 6 bases to 3' terminus, showing a much.
shorter 3' specific subsequence. Similar results were reported elsewhere previously (Sarkar et al., 10 1990)..
[0239] In "summary, P* (1-length) has two subsequences: a 3' specific subsequence (n = the number of bases of the 3' specific subsequence determines the specificity, i.e.,within this region any mismatch to its complementary strand of the template results in no substantial amplification; and a 5' enhancer subsequence (in = the number of bases of 5' enhancer 15 subsequence 0) enhances the amplification efficiency. PAP specificity is co-determined by the base pairing specificity of the 3' specific subsequence, the pyrophosphorolysis specificity and the polymerization specificity. Thus, the base pairing specificity of the 3' specific subasequence is a mirrinrum requirement of the PAP specificity.
[0240] The length of the 3' specific subsequence of P* may be affected by the sequence context 20 and size of the P*, the :type of the 3' terminal dideoxynucleotide, the template sequence, the DNA polymerase, other components like ion, and cycling conditions. When. the template contains repeated sequences > 1 or homogeneous polymer runs > 1, P* loses specificity for anchoring. The length of the 3' specific subsequence of P* may be affected by the sequence context and size of the p*, the type of the 3' terminal dideoxynudeotide, the template sequence, 25 the DNA polymerase, other components like ion, and cycling conditions.
When the template contains repeated sequences > 1 or homogeneous polymer runs > L P* loses specificity for anchoring.
Scanning or Resequen.cing for Unknown. Sequence Variants [0241] The property of the 3' specific subsequence of P* can be applied to scanning for o unknown sequence variants or re-sequencing of predetermined sequences in a parallel way. Each nucleotide on. the complementary strand of the predetermined sequence is queried by four downstream P*s, such as 18-mers (Fig. 11), which have identical sequence except that at the 3' . terminus, either ddAlVfP, ddTMP, ddGIVIP or ddCMP corresponds to the wild-type sequence and the three possible single. base substitutions. The number of P*s scanning the complementary strand of X bases is multiplication of 4 and X, which is suitable for either exponential or linear PAP. The four downstream P*s can even be hinnobilized on a single dot when.
ddAMP, ddTMP, ddGMP and ddC1VIP at the 3' termini are labeled differently for differontiation, such as by four S fluorescence dyes. The amplification signal can thus be represented by intensity decrease of each dye when ddINIMP is removed from P* by pyrophosphorolysis. One advantage of linear PAP is that the four ddNTPs can be used as substrates for single base extensions, with are labeled with different dyes for differentiation:
[0242] Briefly, if only all the P*s corresponding the wild-type sequence are specifically io amplified, the wild-type sequence can be arranged in order by analyzing overlaps. A 13* with a single base substitution at the 3' terminus is amplified at the position of hemi- or homo-point mutations. The mutation also creates a "gap" of no PAP signal, which spans a region of several successive nucleotides. For single base substitution, the gap size (bases) + 1 = the length of the 3' specific subsequence.
= . is [0243] Furthermore, we can also scan the sense strand by designing a second set of upstream P*s. An unknown single .base substitution can be determined by combination of the two sets of .
P*s, even in lieterozygotes. An unknown. small deletion and insertIon can be detected and localized. In order to identify a specific type of deletion or insertion, it is possible to add = corresponding P*s. For fingerprinting, which can provide information of mutation position, 20 there iS a simple stacking way that the stacked region of each two successive P*s < the 3' specific subsequence on the array to reduce the number of P*s by up to n.
fold.
Determination of de novo DNA sequence [0244] The concept of de novo DNA. sequencing by PAP makes use of all the possible 3' specific subsequences el)* to identify the presence of the 3' specific subsequence in de novo 25 sequence. A complete set of the 3' specific subsequences of P* is e.
Each of the 3' specific subsequence has a complete subset of the 5' enhancer subsequence of 4311. For example, a complete set of 16-mer as the 3' specific subsequence and 2-mer as the 5' enhancer subsequence can be indicated as (A, T; G, C)(A, T, G, N16 = 418.
[02451 Briefly, the procedure first determines the list of all the specific PAP amplifications and 30 then reconstructs the unknown. DNA complementary sequence from this list by ordering the 3' specific subsequences with the given length by using the Watson-Crick pairing rules.
[0246] The assembly process ls intenupted wherever a given 3' specific subsequence of P* is encountered two or more times. One of the factors influencing the maximum sequencing length = 52 is the length of the 3' specific subsequence. The length of a random sequence that can be reconstructed unambiguously by a complete set of the 3' specific subsequence with the given length is approximately the square root of the number of the 3' specific sequence in the. complete set with .:5.0% possibility that any given 3' specific subsequence is not encountered -two or more times. Octatners of the 3' specific subsequence, of which there are 65;536, may be useful in the range up to 200 bases. Decanucleotides, of which there are more than a million, may analyze up to a kilObase denovo sequence. 18 mer P*s containing 16 mer as the 3' specific subsequence, = which Complete set is 418 of P*s, may sequence maximum 77,332 bases.
= [02471 When there is neighbored known. sequence to deSign an opposite oligonucleotide for io PAP With two oligonucleotides. The maximum sequencing length is mainly limited to the opposite oligonucleotide, but not to the length of the 3' specific subsequence of P*, termed = Conditional de novo DNA sequencing.
Other Applications for PAP
[0248] For fingerprinting which compares two DNA sequences to see if they are the same or is different, there is a simple way to reduce the number of P*s by using an incomplete set of the 3' specific subsequences, By arranging them in. a particular Order, it is possible to identify the Chromosomal locations as well as sequences. Considering the 3 x 1.0? bp DNA in human geteme, PAP with two oligonacleotides is preferred over PAP with only one P*
to increase the specificity.

20 [0249.1 To monitor gene expression profiling, where up to 6 x 10 to 105 transcripts are expressed and details of the precise sequence are nnnecessary, PAP with only one. P* can be applied and a set of P* which identify nnique motifs in genes can be designed with a total length of up to 22- mer. Between each two Ps, there is at least a sequence difference at the 3'. terminus or > 2 sequence differences at the non-3' terminus.
25 Comparison with. Sequence by Hybridization [0250] In SBH by using oligbitatleotide; the DNA sequence is deteitained- by the hybridization and assembly of positively hybridizing probes through overlapping portions. It has been known = for a long time that a single oligonucleotide hybridiatioii on. a immobilized sample can be very specific in optimal hyblidization ahd washing conditions (Wallace et al., 1979), -thus it is. =
30 possible to discriminate perfect hybrids- from Cones containing a single internal mismatch. The oligonucleotideS- in array ate 11-20 nucleotides in length and have 7-9 bases specific region in-the middle, the non-specifie signal is generated by mismatched hybridization.
Under standard hybridization and washing -coriditiotA the duplex stability between match and rniSinatch is also - =

=

affected by the terminal. mismatch and the flanking sequence (Drmana.c et al., 1989; Khrapke et al.; 1989; Ginot, 1997).
[0251] SHB. can be modified with enzymes in several ways (Miyada and Wallace, 1987;
Southern, 1996): Primer extension by DNA polymerase incorporates bases one at a time only if s they match the complement strand. Ligase has similar requirements: twO
oligonacleotides can be .
joined enzymatically provided they both are complementary to the template at the position of joining.
[02521 Figs. 11A-11B show the enhancement of PAP effidericy. Fig. 11A. PAP is amplified with two oligonucleoticleS P*. and U from duplex TU:UT template. Each of the four P*s has a ddA, dd.T, dd0 and ddC at the 3' terminus. The 3', terminal base is either specific to the complementary strand of the G or A alleles, or not matched: Fig. 11B.
Autoradiogram of PAP
from the GIG, .A/A and G/A genotypes of the human dopamine receptor gene. Tho radioactively = labeled specific products of 461 bases (dupieX PU:UP and excess antisense strand UP) are produced. Other side products UT and UT:TU are indicated. Note that TU:UT
derives from annealing of excess radioactiVoly labeled UT with non-radioactively labeled.
TU original template.
[02531 Figs. .12A-12E show the effect of* length and mismatch on PAP
effiCiency. PAP was amplified with P* and U oligonucleotide (see Table 3): In each of Figs. 12A-12E, P*s have the .
sample 3' termini but are different in. length Fig. 12A. = In lanes 1-4, the P*s niatched. and amplified the G allele. In lanes 5-8, the P*s mismatched at the 3' termini but amplified the A
allele. Fig. 12B. In lanes 942, the P*s matched and amplified the G allele: In lanes 13-16, the P*s, mismatched. at 42 bases to the 3' termini but arciplified the A allele. =
Fig. 12C.. In lanes 1.7-20, the P*s matched and amplified the A allele. In lanes 21-24, the P*s mismatched at -2, .
bases to the 3' tennini but amplified the G allele.. Fig: 12D.- In lanes-25-28, the P*s mismatched . 25 at -9 bases to the. 3' termith. but amplified the A allele. Fig. 12E. In lanes 29-32, the P*s = mismatched at -15 bases to the termini but amplified the A allele. The length effect is indicated as..the yield ratio in, one lane (4) tO the previous lane (Lõ..1)..
The length effect waS-not = , she-Wain lanes 5-43 bocanto the. signa1:11re at or close to. the background, [02541 Fig. 13 shows PAP specificity with differently positioned P*s. PAP was amplified. with a pi= and U oligonucleotide (see Table 4). The P* matched to and amplified the G allele in lanes 2-7, but mismatched to and amplified the A allele in lanes 9-15. Lanes 1 and 9 were PCR control with D1(212)17 mer and U. Lanes 8 and 16 were extension control with only U.
=

[0255] Fig. 14 shows PAP specificity with differently mismatched 134's. PAP
was amplified with a P* and U oligonudeotide (see Table 5). In lanes 2-7, the P* amplified the G allele with match or one mismatch: In lanes 9-15, the P* amplified the A with one or two Mismatches.
Lanes 1 and 9 were PCR control with D1(212)17 mei: and U. Lanes 8 and 16 \Veit extension control with only U.

PAP Amplification From Genomic DNA
[0256] This example illustrates PAP amplification directly from genoniic DNA.
The oligonucleatides used in this example are listed below. Lane numbers refer to lanes in Fig. IS.
[0257] The downstream oligonucleotides in 0.1 AM concentration are:
_ Lane 1: Di(204)25D 5' TCTGACTGACCCCTATTCCCTGCTT 3' (SEQ NO:43) =
Lane 2: P*(206)24A 5' TGACTGACCCCTATTCCCTGCTTA* 3' (A allele specific; =
SEQ NO:44) Lane 3: P*(204)26G9 5' TCTGACTGACCCCTATTCCCTGCTTG* 3' (G allele specific; SEQ NO:45).
Lane 4: P*(206)24G-2 = 5' ACTGACCCCTATTCCCTGCTTGGG* 3' (G allele specific;
= SEQ ID NO:46) Lane 5: P*(228)26A-24 5' TAGGAACTTGGGGGGTGTCAGAGCCC*. 3' (A allele specific; SEQ ID NO:47) = [0258] The opposite upstream oligonucleotide in 0.1 jaM concentration is:
D1(420)24U
5' ACGGCAGCACAGACCAGCGTGTTC 3' (SEQ ID NO:48), which was paired with each - downstream oligonucleotide. See Footnotes of Table 3 for detail.
= [02591 The other components were the same as in Example 2, except for the following: 0.5 U of each of AmpliTaqFS and Tag DNA polymerases, and 100 ng of heterozygous GIA.
allelic genamic DNA were used per 25 ul reaction by using 30 cycles.
[0260] The PAP product size range from 193bp to 218 bp. One double stranded and one single stranded product was observed on the gel, indicating the exhaust of PP i hydrolyzed by the contaminated thermostable pyrophosphatase.

=
=

=
Comparison of Specificity of LM-PCR and LM-PAP
[0261] The LM-PCR protocol includes primer extension, linker ligation, PCR
amplification, and directed labeling in the human dopamine DI receptor gene model system: (Fig.
16). LM-PCR
5 was performed with the addition by terminal deoxynucleotidyl transferas8 (TdT) (this protOcolis =
',Mown as TD-PCR) .on UV-treated genoinic DNA samples essentially as described (Pfeifer et al., 1999), except that VentR (exo-) DNA poIymerase was used in the first 10 cycles of Primer eXtension (P1 primer 5' TTGCCACTCAAGCGGTCCTCTCAT 3 (SEQ ID NO:49)).
Temperature cycles were 1 Min at 95 C, 3 Min. at 63 C, and 3 min at 72 C.
To enhanoe the .io signal, terminal transfetase was added to the protocol, and this variation of LM-PCR is called TD-PCR. Dynabeads were used to enrich target DNA molecules before terminal deoxyriucleotidyl transferase (TdT) tailing. PCR was performed using Expand Long Template PCR System 3 (Bm-,B) aS described by the manufactiter (P2 primer 5' GAAGCAATCTGGCT
GTGCAA_AGTC 3' (SEQ ID NO:50)), The PCR products were purified using Q1Aquick PCR
is Purification Kit (Q1AGEN) before performing the direct labeling. A
portion of the cleaned PCR
product was used for direct labeling with Amplinq DNA Polyinerase (Perkin-Elmer) with 32P-labeled plinierst P3A: (5' TCTGACTGACCCCTATTCCCTGCTTA 3' (SEQ
NO:51; the 3' terminal deoxynucleotide is A allele specific) and =
20 P3G: (5' TCTGACTGACCCCTA.TTCCCTGCTTG 3' (SEQ ID NO:52; the 3' terminal deoxynucleotide i G allele specific). =
[02621 LM-PAP was performed as allele-specific PCR except for the direct labeling step by PAP (Fig. 16A), The purified PCR product was used for direct labeling with 3211 labeled primers:
P3A: 5' TCTGACTGACCCCTATTCCCTGCTTA* 3' (SEQ ID NO:53; the 3' terminal 25 deoxynucleotide is A allele specific) and P3G: 5' TCTGACTGACCCCTATTCCCTGCTTG* 3' (SEQ ID NO:54; the 3' terminal deoxynucleotide is G allele specific) using PAP reaction conditions in. a 10 ill Volume (50 niM KC1, 10 niM Tris/HC1 (pH 7.6),1.5 niM MgC12, 100 11M of each dNTP, 0.1 tiM P*, 300 i.t.M NA4PPi, 2% DMSO, 0.25U.dath of 30 AinialircieS and AMpliTaq DNA Polynierases (Perkin-Elmer). The cycling conditions were 940 C, 10 sec.; 60 C, I. mitt. and 72 C, 2 min. Mt a total of 8 or 16 cycles. LM-PAP was dt aniatically more specific than LM-PCR. The initial data with the &par-line D1 gene shows a lower background with LM-PAP than with the identical unblocked oligonucleotide with LM-=

PCR. Also, LM-PAP can be performed with the PGK gene, a gene with a very high GC rich region (70%) (Fig. 16B).
10263j Fig, 16A shows a UV footprinting of the dopaniine D1 receptor 'gene with a eon parison of allele-specific LM-PA,P and allele-specific LM-PCR. A direct comparison of LM-PAP with a s P* and LM-PCR with an. unblocked primer of identical sequence shows that two alleles can be distinguished with LM-PAP, but not with LM-PCR. Both methods were performed on DNA that was untreated (C), in vitro treated (T) or in vivo treated (V) with UV. The direct labeling reaction using PAP conditions (lanes 7-18) with 32P labeled primers P3A* (lanes 7-9 and 13-15) and P3G* (lanes 10-12 and 16-18) was done with AmpliTaqFS and AmpliTaq for 8 and 16 cycles. For LM-PCR the direct labeling reaction was done with AmpliTaq (lanes 1:6) and 32P-labeled primers P3A (lanes 1-3) and P30 (lanes 4-6) for 8 cycles:
Allelic primers P*s, P3A* and P3G* for LM-PAP clearly distinguish the two alleles, while unblocked allelic primers of identical sequence, P3A and P3G; were unable to distinguish the alleles by LM-PCR.
[0264] Fig: 16B shows a UV footprinting of the pgK gene. The LM-PAP procedure for PGK
was essentially the same as for the dopamine Di receptor except that Pfu Turbo DNA
polYmerase was used in. the primer extension, as well as. 7-deaza-dGTP/dGTP in a 3:1 ratio.
Temperature cycles were 950 1 min, 60 2 min., and 76 3 min The PCR step was performed.
using Vent (exo-) DNA Polymerase at 97 1 min., 60 2 nun.,. 76 3 min. also with deaza dGTP.
The purified PCR products were used for direct labeling with the 32P P3G* and P3C* primers lasing PAP reaction conditions in. a 25 pi volume (50 rolVIKCL, 20 mM Hepes, pH 6.95, 10 mM
(NH4)2SO4, 1.5 mM MgC12, 40 pM dNTP, 150 uM Na4PPi., 4% DMSO, and 1 unit of AinpliTaq FS DNA.Polymerase. The conditions for cycling were 94 15 sec., 60 30 sec., and 72 1 min:
for 10 cycles.

Optimization of PAP-A to Detect a Mutation in 1 of 104-105 Templates [0265] One ug of lambda phage DNA contains 2 x 101 copies of template. The specificity of PAP is determined by mixing one part mutant lac templates with 104 to 105 parts control DNA
templates, e.g., wild-type lad. The specificity of PAP-A is a function of the error rate of the polymerase, the purity of P* (<2 x le by current purification protocol) and the potential for damage of the DNA template in the extraction process. The yield and specificity of PAP is -optiroind by testing enzyme type and concentration and the concentrations of other =
=

=
components, such as dNTP, PP, Mgf+ or Mn. Hotstart PAP using antibody-activated enzyme, such as DNA polymerase,, at room temperature can be used to eliminate spurious amplifications.
[0266] Wild-type and mutant lambda phage DNA, which are used. in the laboratory as a model system to study Spontaneous mutation in mammals', are prepared from infected E. colt SCS-8 . 5 cells (Nishino et al., 1996): The lambda phage is graft under high fidelity cenditions and DNA
is isolated: with. care under conditions with low rateS of DNA damage (Stratagene manual) (Nishind et al., 1996; Hill et al., 1999).
[0267] The mutants include one example of each of the two types of transitions, the four types of transversions and a= one-base nucleotide deletion. P*s specific for each, of the mutations is =
synthesized: These DNA. templates. are used for reconstruction experiments in which mutated DNA is serially diluted into wild-type DNA. The spiked samples are used to optimize PAP-A.
The most robust polyrnerases are chosen_ = based on yield, and specificity using TaqFS, ThermoSequenase, and SequiThenn. Excel II (Epicentre). Other components of the. reaction are optimized systematically; including thermocycling = parameters 5 ofigonucleotide. lengthy and is reagent concentrations of PPi, dNTP and Mg-3-4- or Mn. Quantitative detection of the yield of PAP product is achieved with antoradiography or fluorescence on a SSCP gel.
Theft data aids in.
= the optimization of PAP-R and LI\CPAP (below). The optimization of these various parameters result in a specificity of 1 part in 104-105.
= [0268] The Optimized conditions are also tested for detecting mutations in the human factor IX
gone by mixing human mutant genomic DNA templates with up= to 104 wild-type templates. As = with the lambda experiment, exponential PAP is performed with appropriately designed = oligonucleotides (using Oligo5 softWare) for .40 cycles and. strong signal is achieved by antoracliography or by fluorescence detection.
=
. EXAMPLE 6 . Optimization of PAP-R
[02691 In a model System, mismatches along the length. of p.* intribitiq activation, even when the II *Smatch is two IiiideotiddS. from. the 5' end (Fig,14), An additional set of 18. meta of P*s, -whose 5 termini were displaced 2, 6, 9, and 12 nucleotides downstream, also showed inhibition of activation (Fig. 13). In addition, 20 and 22 mers also show inhibition with single nucleotide mismatches (Fig. 12). To extend these mdings and to lay the foundation for a robust method of teg6qtencing,.thd relationship between the location of single base mismatches and activation of P4's 15 analyzed further.. . =

[0270] The factor IX gene is used as a model system. because more than 1,000 DNA samples from, hemophilia patients and family members have been ascertained from previous work on the niolecular epidemiology of gennline mutations in humans (Soinrner, 1995;
Ketterling et al.;
1999). TWO 20-nttclecitide regions of exon B and exon H in the human factor IX
gene are used as model sygten18. The region of exon B is designed from nucleotides 6460 to 6479 (5' CGAGAAGTTTTTGAAAACAC 3'.(SEQ JD NO:55;Yoshitake et at, 1985), within which eight different single base mutations are available. The region of exon H is from nucleotides 30845 to 30864 (5' GAACATACAGAGCAAAAGCG 3' (SEQ ID NO:56), within which seven mutations at different positions are available: P*s identical to wild-type regions B and H will be o synthesized. Identical P*s are synthesized, with the exception of a single nucleotide mismatch:
= [02711 The wild-type factor IX sequence is used in the initial studies. A
few P*s that match the wild-type sequence or that miSinatch at selected:sites within the 5' third of the oligonucleotide sequence are helpful in performing pilot experiments to assess the optimal length of the = oligonucleotide. The effects of polymerases and. reaction conditions can be assessed.
[02721 FrOm preliminary data, it appears that 18 mers or larger may be an optimal size. It is = also possible that 25 mers. or even 30 DION may be optimal. For. the present example, it is assmiled that 20 mers are an optimal size. Wild type P* and twenty P*S with one of the possible single base mismatches at each nucleotide of the position region of exon B are synthesized.
= Eight of these P* are a perfect match to a Mutation in a patient with hemophilia B. As positive =
0011trOlg, it is shown that these P*s activate efficiently when the appropriate mutated DNA
sample is used. Exponential PAP and linear PAP are performed and the noise rate is determined.
The noise tate for linear PAP is generally lower and is used.
[0273] To confinn preliminary data in another sequence context; a similar experiment is performed in exon H. The seven mutations in that region of exon H are analyzed in a blinded *manner to determine if the precise match is detected. The effects of the position of the mismatch or the type of mismatch. on P* activation is determined. The effects of different polymerases, reaction temperattre, and other reaction conditions can also be determined:
Another set of 20 . P*s provides additional data from misniatcheS 12.20 nucleotides froth.
the 3' terminus:

Optimization of LM-PAP
[02741 The lairnan dopamine Di receptor gene and the mouse Peci gene are used as model = systems to compare the analysis of chromatin structure when LM-PAP or LM-PCR is utilized.

The dopamine DI receptor gene has. been described above: X chromosome inactivation occurs at an early embryonic stage. Since the two alleles in female cells maintain a different expression status, this is an advantageous system for studies of gene regulation. Pgld is an X-linked housekeeping gene encoding phosphoglycerate kin ase (PGK). PGK is an important enzyme in glycolysis and the gene is expected . to be active all the time except. in the inactive X
chromosome (Xi) of female somatic cells and in male germ cells.
[02751 The preliminary data shows a dramatic enhancement of specificity.with LM-PAP relative to LM-PCR in the dopamine DI receptor gene, a gene not previously analyzed for chromatin structure (Fig. 16A). In this example, LM-PAP and LM-PCR are performed. Three sets of oligonucleotides that generated LM-PCR profiles and seven sets of primers that generated LM-PCR profiles with unacceptable background in the Pgld (and other X-chromosomal genes) are used = to compare LM-PAP with LM-PCR. Deoxy-terminating and dideoxy-tenninating oligonucleotides of identical sequence are utilized to perfonn LM-PAP and LM-PCR, respectively. The level of signal relative to background is also quantitated by a Phospholmager.
The average signal-to-noise ratio. is determined. Optimization data derived from analyses with PAP-A and PAP-R are also useful. in the LM-PAP protocol: LM-PAP is optimized for the two regions to determine if the signal-to-noise ratio can be reduced further.
3.

Optimization of Allele-Specific LM-PAP
[02761 Polymorphic sites of pgIclo .and lb gene in both coding and non-coding regions have been reported (Boer et al., 1990)õ These are used to design the allele-specific P. One allele-specific oligonucleotide is chosen prospectively from the Pea gene and one is chosen prospectively from. the d.opam-hie D1 receptor gene. Blocked and unblocked oligonucleotides of .
identical sequence are synthesized and allele-specific LM-PAP and LM-PCR are performed;
respectively. The signal to noise ratio is quantitated and compared, PAP-R on. a Microarray [02771 The initial experiment will focus on the two 20 nucleotide regions of exons B and H. as described above. The experimental design of PAP-R is similar to the experiments described above, except for digital light-direct synthesis of P* oligonucleotides on a microarray, e.g., with the Geniom instrument. A total of 160 oligonucleotides are synthesized complementary to = 60 =
wild-type and to all the single base mismatches for 20 bp regions of eX011S B
and H of the factor IX gene. As a positive control; 160 oligouncleotides, each out of registered by one nucleotide, are synthesized to match exactly an adjacent 160 bp region Of the faCtor DC
gene. anomie DNA
frorn wild-type and 'mutant .saMples is amplified; annealed to the oligonncleotides and primer eXtension will be perfornied with a fluorescent dideoxy terminator. The protocol is optimized for the solid support. Adjustment of primer length, enzyme utilized and reaction conditions is performed such that most, if not all, of the oligonucleotides that mismatch the two 20 bp nucleotide regions of factor IX generate little if any signal, while most of the 160 control oligonucleotides generate a strong signal.
io [0278] One strategy for resequencing is shown. in Figs. 3 and 4. Each nucleotide in the complementary strand of the predetermined sequence is queried by four downstream P*s, such 'as 20 mers, which have identical sequence eXcept for the 3' terminus, which is either ddA, ddT, ddG or ddC. For a 1 kb segment, 4,000 P*S are needed in the dow:astream direction.. In the second set of experiments, exons B and.H of the factor DC gene are resequenced. Samples from:
more than 200 patients With different mutations in these regions are available for. analysis. False == positives and false negatives are assessed by blinded. analysis.
Heterozygous female samples are = available for many of the mutations. For the remaining male patient samples one to one mixing experiments with, wild-type or a second mutated sample generates the equivalent of heterozygotes or compound heterozygotes, respectively. Subsequently, all the regions ,of likely functional significance (the putative promoter region, the coding regions, and the splice kinction. ) are resequenced (2.2' kb). Since more than 600 independent mutations are available, it = is possible to determine whether mere than 99% of all Sequence changes are identified (the sequence changes ill these samples have been determhied by direct sequencing Or the course of a decade).
[02101 A F*.with a single base substitution at the 3' terminus generates a signal at the position of hemizygous or homozygous point mutations: The mutation also createS a "gap"
of no PAP
signal, which spans a region of several successive nucleotides. When a single base substitution occurs, the gap, size (nucleotides) + 1 = the length of the 3' specific subsequence (Figs. 3 and 4).
[0280] To analyze samples with higher G+C content (55%), mutations in the lac/
gene are utilized : These' Mutations from the Big Blue Transgenie Molise Mutation Detection System, have. the pcitential to facilitate the definition of a strategy that detects more that 99.9% of nitrtatiotts, since more than 6,000 mutations are available in this system.
The relestantre are.
analyzed with the help of robotic devices. In addition, hundreds of mutations o 'polyMorphisms =
=

are available for aralysis in other genes with (3+C contents of 30-75%. The dystrophin gene is particularly amenable to testing perforniance under conditions in which naegabases of sequence require scanning. In this gene in which 90 segments are amplified by a robotic device, virtually all sequence variants have been defined by DOVAM-S followed by DNA sequencing.
This is s advantageous because many molecular epidemiological and molecular diagnostic applications benefit from resequencing that detects virtually 100% of the mutations.
= EXAMPLE 1.0 PAP Amplification Directly from Haman and Mouse anomie DNAs io [02811 PAP was performed with each of two P*s, Pl* (SEQ ID NO:45, G allele specific)) or P2* (SEQ ED NO:47, A allele specific) and an upstream unblocked primer (U;
(SEQ ID NO:48) to amplify 180-bp segments of the D1 dopamine gene. The P* are 26-mers with ddC and ddG at the 3' temaini. 100 ng of lun-na3i genomic DNA was amplified for 35 cycles followed by 2% gel , electrophoresis. The PM reaction mixture contained a total volume of 25 ill:
50 mM KC', 20 InM HEPES/NaOH (p11 6.9 at 25 C), 10 tn.M (NH4)2804, 1.5 m114 MgC12, 40 uM
each of the = four dNTPs (dATP, dTTP,,dGTP, dCTP),. 0.1 tM U, 150 uM NaRpi, 2% DMSO, 0.5.,U of 2.0 AmpliTagPS Polymerase (PE Applied Biosystems),.. 0.5 U Tag polymerase and 100 ng of human 4 .2 genoinic, DNA. The cycling conditions were 94 C for 15 sec, 65 C for 30 sec and 72 C for 1 min.- Fig. 17A shows the results for PAP amplification of the DI 41opainine.
gene. In lanes 2 and 20 5, P1* is specific for the A allele template at 24 nucleotides from the 3' terminus, so there is little or no discritnination between the G/G and A/A genotypes.. In lanes 3 and 6, P2* is specific for the A allele template at 2 nucleotides from the 3' terminus, so there is specific amplification of the A/A genotype: Lanes 1 and 5 are PCR controls. Lanes 4 and 8 are negative controls without P. Lane M is 120 ng cpx DNA/HAM-Sr marker.

[0282] Three Bi-PAP assays were tested directly from. mouse genomic DNA. Bi-PAP was performed with two.P*s containing a clideoxynucleotide blocker at the 3' terminus to amplify an 80-bp segment of the lad gene: The P*S are specific to the Wild-type template and are 40-42 nucleotides long. In each of the three Bi-PAP assays, two opposite P* with one nucleotide .
overlap at their 3' termini were used to amplify 400 copies of the lad gene using 35 cycles..
30 The sequences of the P*s are as fellows:
5' GAAGCGGCGTCGAAGCCTG-TAAAGCGGCGGTGCACAATCT* 3' (SEQ ID
NO:67) and 5' GCGGATAGTTAATGATCAGCCCACTGACGCGTTGCGCGAGAA.* 3' (SEQ ID NO:68) in lanes 1. and 2;
=

=
. .
5' GATGGCGGAGCTGAATTACATTCCC.AACCGCGTGGCACAA* 3' (SEQ ID
NO:69 and 5.' GGCAA.CGCCAATCAGCAACGACTGTTTGCCCGCCAGTTGT* (SEQ ID
NO:70) in lanes a and 4; and 5' TACATTCCCAACCGCGTGGCACAACAACTGGCGGGCAAAC*- 3' (SEQ ID
NO:t71) and 5' GGGCCAG-ACTGGAGGTGGCAACGCCAATCAGCAACGACTG* (SEQ
ED NO:72) in_ lanes 5 and 6:
[02831 The PAP reaction mixture contained a total volume of 25 p.1: 50 niM
KC1, 20 mM
'HEPES/NaOH (pH 6.9 at 25 C), 10 mM (NH4)2SO4, 1.5 rriM MgC12, 40 IrM each of the four dNTPs (dA11), dTTP, dGTP, dCTP), 0.1 !IM U, 150 p.M NaiPpi, 4% DMSO, 1.0 U of lo AmpliTaqFS polynierase (PE Applied Biosystems) and 400 copies of mouse genomic DNA, The cycling conditions were 94 C for 15 sec, '65 C for 30 sec and 72 C for 1. min. The - unincorporated P*s were separated well fromthe product on 2% agarose gel. Na diner was seen. Fig. 17B shows the results for these Bi-PAP assays. In lanes 1, 3 and 5, the wild-typo templates are amplified. Lanes 2, 4 and 6 are negative controls without mouse genomic DNA.
15 [0284] Three PAP assays directly amplified .18043p. segment . of the D1 receptor gene from human genomic DNA with strong signals of PAP products:, The allele-specificity of26-mer P*
remains when the mismatch_ is at 2 nucleotides from the terminus, but the allele-specificity is lost when the mismatch is at 24 nucleotides from the 3' terminus: Three Bi-PAP
assays directly amplified as low as four hundred copies of the lac/ gene from mouse genomic DNA. The P*
20 oligorrticleotides have different deoxynucleotides blocked at the 3' termini's. and all can be efficiently activated. Addition of extra human. DNA did not affect the amplification of the lad =
gene in "mouse genornic DNA. The product of Bi-PAP was easily distinguished from "imincorporated P*s. P* does not form dimmers because P* needs long and.
perfectly matched regions at the 3' terminus for activation.

=

PAP with Acyciomicleotides and Various Polymerases X Phage DNA Template [0285] The Wild-type X phage-DNA template that Contains an inserted wild-type Lc/ gene of E.

coil (Kohler et at., 1991) was purchased from Stratagene. The mutant X phage DNA template was prepared from 7c phage plaques transforined into SCS-8 E. coil della according to Maniatis, et at. (im). It contained a T to G mutation at nucleotide 369 in the lad' gene. The slnount of X
phage DNA was determined by UV absorbance at 260 urn. =
=

63 =
Synthesis of P* by Adding Acyclonucleotide or a Dideoxynucleotide at the 3' Terminus [0286] The 3' terminal acyclonucleotide or 3 terminal dideoxynucleotide was added to a deoxynucleotide oligonucleotide by terminal transferase. The mixture contained a total volume of 25 100 I/1M potassium cacodylate (pH 7.2), 2.0 mM CoCl2, 0.2 mM DTT, 2 n_M of the oligonucleotide, 2.4 mIVI acycloNTP (the molar ratio of the 3'-OH terminus to acycloNTP was 1:30) (New England BioLabs), or 2.4 mM 2',31-d.dNTP (the molar ratio of the 3'-01-1 terminus to ddNTP was 1:30)(Roche), 100 U of terminal transferase (Invitrogen). The reaction was incubated at 37 C for 6 hr and then stopped by adding EDTA to a 5 mM final concentration.
After desalting using a Centri-spin-2 column (Princeton Separations), P* was purified by preparative 7 M urea/18% polyacrylam i de gel electrophoresis with 30 mM
triethanolamine/tricine buffer (pH 7.9 at 25 C) (Maniatis, et al., 1982; Liu, et al., 1999b). The amount of recovered P* was determined by UV absorbance at 260 inn.
[02871 Since small amounts of unterminated oligonucleotide would result in unexpected PCR
, amplification, the purity of P* was tested by the absence of PCR product at pH
8.3 in which Is pyrophosphorolysis is inhibited. It is estimated that more than 99.99% of P*
contained an acyclonucleotide or a dideoxynucleotide at the 3' terminus.
r:.. PAP Amplification [02881 PAP was examined with P*1 and 01, with P*2 and. 02, and with P*1 and P2*
= respectively (Fig. I 8A and Table 6). The P*s were 30 or 35 nucleotide long and contained an acyclonucleotide or a dideoxynucleotide at the 3' terminus.
=

=

List of Oligonucleotides PAP Amplification (allele) Desig. Name' - Sequence (ID NO:) 3' Terminal G: C T:A
P*1 P*(340)30D CGAAGCCTGTAAAGCGGCGGTGCACAATCG* (57) acycloGIVIP .
Yes No or ddGMP
01 -0(502)25U ACTGTTGATGGGTGTCTGGTCAGAG (58) dGMP
P*2 P *(398)3 OU TGATCAGCCCACTGACGCGTTGCGCGAGAC* (59) acycloCMP
Yes No or dddMP 0 02 0(190)21D A.CAACTGGCGGGCAAACAGTC (60) dalIP
1.) a The position of the first nucleotide of the transcript in the lad- gene of E. coli is assigned the nucleotide position 1 1.) (Farabaugh, 1978). As an example for P*1, P* = pyrophosphorolysis activatable bligonucleotide, it may be a 3 terminal 1.) acydonucleotide blocked P* or a 3' terminal dideoxynucleotide blocked P*.
(340)30D = 5' end of the P'' begins at 340, the length is 30 nucleotides and the direction is downstream (i.e., in the direction of transcription). The precise sizes and Locations of the amplified fragment can be obtained from the informative names. The 30-mer P*s are indicated above.
The 35-mer Pss are 3' co,terminal with the 30-xner P*s and 5 nucleotides longer at their 5' termini. 0 , =
, .

[0289] The PAP reaction mixture with AmpliTaqFS DNA polyinerase contained a total volume of 25 pi: 50 inN1 KCI, 20 m_M BEPES/NaOH (PH 6.9 at 25 C), 10 inM (NH4)2SO4, 1.5 niM
= MgC12, 50 jiM each of the four dl\ITPs (dATP, dTTP, dGTP and dCTP), 0.1 plW of each.
oligonucleotide, 150 1.1M Na4PPi, 4% DMSO, 1 U of AmpHMO'S' DNA polyraerase (PE-s Applied Biosysterns), 0.1 ng phage DNA teinplate. The eyeling conditions were 92 C
- =
. fOr 10 see, 65 C for 30 sec, and 72 C for I Min for a total of 30 cycles. A
denaturing step of 92 = C for 1 min was added before the first cyele.
.
[0290] The PAP reaction mixture with Vent (exo..) or Pfu (exo-) contained a total volmte of 10 mM KC1, 20 .n1M EEPES/NaOH (pH 7.19 at 25 C), 10 m.M (NH4)2S03, 1.2 inM
o MgCl, 50 pM each of the four dNTPS (dATP, dTTP, dGTP and d.CTP), 0.1 pM= of each oligonucleotide, 150 1.1A4 Na4PP1, 4% DMSO, 1 U of Vent (exo-) DNA polymerase (New England BioLabs) or Pfu (exo-) DNA polyinerase (Stratagene), 0.1 ng = of the.2 phage DNA
template. The cycling conditions were 94 C for 15 sec, 60 C for 30 sec, and 72 C for 1 min for a total of 30 cycles. A.denaturing step of 94 C for 1 min was added before the fttst cycle.

[02911 The product was electrophoresed through a standard 2% agarose gel. The gel was stained with ethidium bromide for UV: photography by a CCD camera (Bio-Rad Gel Doc 1000).
[0292] As shown above, TagFS, a genetically engineered DNA polyinerase (Innis and Gelfand, 1999), greatly improved the efficiency of PAP. 3' terminal dideoxyMicleotide blocked P*s can.
be activated by pyrophosphorolysis to remove the 3' terminal dideoXynteleotide in the presence 20 = of pyrophosphate (IT) and the complementary strand of the allelic template. Then the actiVated p* can be extended by DNA polymerization.
[0293] PAP was perfolined= with 3' acyclonucleotide blocked P*s by nail-1g X
phage DNA
containing the lad- gene as model system. P*1 and P*2 are downstream and npstreant blocked oligonucleotides, respectively, for the same mutation (Fig. 18A and Table 6):
The P*1 and P*2 25 have an acycloGM2 and acycloCIAll at their 3' termini, respectively.
Amplification products were absent without pyrophosphate added at pH 8.3 where pyrophosphorolysis is inhibited, showing that P*1 and P*2 were not directly extendible.
[0294]. P*1 and P*2 are specific to the mutated template but mismatch to the wild-type template at their 3' termini. The mutated template was amplified efficiently by PAP
with one 30 acyclonucleotide blocked P* and one opposing unblocked oligonucleotide and by PAP with tWO
opposing 3' terminal acyclonucleotide blocked P*s (lanes I. and 2 in Fig.
18B), with two opposing acyclonucleotide blocked P*s (a special form of PAP where the tWO
Opposing P*s are .
overlapped at their 3' termini by one nucleotide),(Land 3 in Fig. 1813):
However, no product was =

generated from the wild-type template because of the misniatch at the 3' terminus, showing the = specificity (lanes 5-7 in. Fig. I 813). PAP with the 3' dideoxynucleotide blocked P* showed similar results (lanes 9-16 in Fig. 18B). Direct sequencing analysis confirmed the correct sequence of the amplified product. The effect of P* length Was also tested.
Similar results were s obtained with 35-m.er P*s that are co-terminal with the 30-mer P*8 and five nucleotides longer at "
their 5' termini (Fig: 18C), Other P*s specific for the wild-type sequence at the 3' terminus (with acycloTMP and ddTMP) were also tested with similar results.
[02951 Family It DNA polymerases Vent (exo-) and PA (exo-) were tested using the above model system. With the acycIonucleotide blocker and. perfect match at the 3' terminus, the o mutated template was amplified efficiently by PAP with one P* (lanes 1 and 2 in Figs. 18D and . 18E) and one opposing unblocked oligonucleotide and by PAP with two opposing P*s of P*1 and P*2 (a special form of PAP where the two opposing P*s are overlapped at their 3' temiini by one. nucleotide) (lane 3 in Figs. 18D and 18E). However, no product was generated from the wild-type template because P*1 and P*2 mismatch the wild-type template at their 3' termini, is showing the specificity (lanes 5-7 in Figs. 1813 and 18E). Vent (exo-) and Pfu (exo-) polymerases could not amplify with the 3' dideoxynucleatide blocked P* (lanes

9-16 in Figs.
18D and 18E). Direct sequencing analysis confirmed the con-ect sequence of the P*1/01 and P*2/0 2 products. Similar results were obtained With AcycloPol (Perkin-Elmer), a genetically engineered Family II archeon DNA polymerase. It is not clear why PAP with Vent (exo-) and 20. Pfu (exo-) DNA polymerases discriminates against 3' dideoxyribonucleotide blockers.
=
Other Blockers [0296] These results demonstrate that two terminators used in Sanger sequencing can be used as blockers in PAP. Terminators have also been described as therapies of viral illnesses, such as AIDS, and for cancer therapy, such as, 3'-deoxyadenosine (cordycepin), 3'-azido-3'-deoxythyraidine (AZT), 2',3'-dideoxyinosine (ddl), 2',3'-dideoxy-3'-thiacytidine (3TC) and 2',3'-didehydro-2',3'-dideoxythymidirie (d4T). DNA polymerase can incorporate their triphosphate form into the synthesizing strand, and the incorporation cause termination of the extension (Gardner and Jack, 1999; Cheng et al., 1987; St. Clair et al., 1987; Ueno and Mitsuya, 1997).
The raonophosphate nucleotides of 3'-azido-3'-deoxythymidine (AZT), T,3'-dideoxy-3'-thiacytidine (3TC) and 2',3'-didehydro-2',3'-dideoxythymidirie (d4T); when located at the 3' termini of oligonucleotides, can be removed by pyrophosphorolysis by MY
reverse transcriptase or its variants (Anon et al., 1998; Gate et al., 2000; Meyer et al., 2000;
Urban et al., 2001).
=

6.7 These results indicate the application of PAP for various types of blocicers and for RNA.
templates.
[02971 In summary, PAP amplification occurred efficiently and specifically with 3' acyclonucleotide and 3' dideOxynucleotide .blockers using TaqFS. DNA
polymerase, and only s with acyclonucleotide blockers using Vent (exo-) and Pfu (exo-) DNA
polymerases. Other 3' terminal nonextendible oligonucleotides and other DNA polyinerases can. be used, if the 3' terminal nucleotide can be. removed by pyrophosphorolysis, and the activated oligonucleotide can be extended.
o EXAMPLE 12 Detection of Extremely Rare Alleles by Bi-PAP
Phage DNA Template [02981 The wild-type X phage DNA template that contains an inserted wild-type lad gene of E.
coil (Kohler et- al., 1991) was Purchased from Stratagene, Three mutated X.
phage DNA.
=
IS templatds were prepared from 2t, phage plaques transformed into SCS.-8 E. coil cells according to Mani atis etal. (1982) . They contain an A to T Mutation at nucleotide position 190, a T to G
=:
mutation at. nucleotide 369 and a T to C mutation at nucleotide 369 in the lad gene, z ¨
respectively. The amount of X phage DNA was determined by UV abSorbance at 260 iitrt.
'Synthesis of P* by Adding a 3' Dideoxyiaucleotide .20 [02991 The 3' ternainal ditleOxyniicleoticle was added- to an oligodeoxynucleOtide by terminal transferase. The mixtUre contained a total volume of 25 1.11: 100 mM potassium pacodylate(pH
2.0 raM= CoCii, 02 mM DTT, 2 nM of the oligonuclootide, 2:4 3'-dd.NTP (the molar ratio of the 3'-0H terminus. to .ddl\ITP was 1:30)(Roche), 100 U of terminal transferase (Invitrogen). The reaction was incubated at 37 C for 6 hr and then stopped bY
adding EDTA to 25 a 5 mM Thial cdneentration.. After desalting using a Centri-spitia column i (Princeton Separations), P* was purified by preparative 7 M urea/16% polyacrylamide gel electrophoresis with 30 mM Triethanolarnirie/Tricine buffer (pH 7.9 at 25 C) (Maniatis et al .
, 1982, Lin et al., I999b). The atiount of recoVeredP* was determined by UV absorbance at 260 um.
[0300] Sirico small amitaintS of untentinafed oligonncleotide would result in unexpected PCR
30 amplification, P* was =32P-labeled at the 5' terminus by T4 polyn.ucleotide kinase and then was electrophoresed through a 7 M uted20% polyaciylarnide gel. Only P* prod-acts Were visible even when the gel waS=overeXposed. It is estiniated that More than 99.99% of P.* contained a =

=
dideoXynucleotide at the 3' terminus The parity of P* WAS supported by the absence of PCk product at pH 8.3 in which pyrophosphorolysis is inhibited. .
PAP Amplification [0301] Bi-PAP assays for nucleotide 190 and lincleotide 369 of the lad gene Were examined.
s The P's were 40 nucleotides long except that the upstream P*s = for 'position 369 are 42 nucleotides. Each P* contained the. sequence-specific nucleotide = at the 3 term-in-LIS. The PAP
reaction mixture contained a total volume of 25 ul: 50 inM Kel, 20 in.M
HEPES/Na011 (pH 6.=9 at 25 C), 10 naM (NH4)2504, 1.5 naM MgC12, 40 ILM each of the four dNTPs (dATP, dTTP, dGTP and dCTP), 0.1 WA each P*, 150 uM Na4PPi, 4% DMSO, 1 u.Ci of [a-3211-4:1CTP
(3000Ciinamole, Amersh.ara), 1 U of AmpliTu0S- DNA polymerase (PE Applied Biosystems), 2,000 copies of the phage DNA template or stated elsewhere. The cycling conditions were 92 C for 6 sec, 68 C for 20 sec, and 72 C for 20 sec for a total of 15 cycles. A denaturing step of 92 C for 1 min was, added before the first cycle.
[03.021 The product was electrophoresed through a standard 2.5% agarose gel, and the gel was . stained with ethidium bromide for UV photography by a CCD camera (Bia.Rad Gel Doc 1000):
[0303] In order to differentiate the mutated product from. the wild-type product of the same size, non-denaturing SSCP gel electrophoresis was perfonted (Orita et al., 1989):
The reaction was.
=
mixed with two-fold volume of loading buffer (7M urea. and 50% formarnide), bOiled= and rapidly cooled on ice. The product in 10 41 of the mixed reaction was electrophoresed through an 8% .non-denaturing PAGE-PLUS (A3nresco) gel with 30 In.M
Ethanolarnine/Capsco buffer (pH 9.6) (Liu et al., I999b) at 4 C. The gel was dried and exposed to Kodak X-OMATTm AR
film for autoradiography. Three or four bands from each amplified product were seen on a gel.
=
The upper one or two bands were double strained DNA due to hybridization of de-natured single-stranded segments during the electrophoresis as a result of the substantial amountS of amplified product present. Increasing the concentration of the amplified product further increase the intensity of the upper bands: =
Highly Efficient PAP Amplification [0304] TaqFS, a genetically engineered DNA polymerase greatly improved, the efficiency of PAP. The conditions of PAP were further optimized. for dramatically higher efficienCieS
allOwing PAP to amplify directly from a few copies of % phage DNA or human genomic DNA
template. The reaction components and the thennocycling regime were optimized, including: 1) decreased concentrations of PPi in that keeping the PPi to dNTP ratio essentially constant, use of low pH HEPES buffer (pH 6.9 at 25 C), iii) addition of (NH4)2803, iv) increased amount of Ta0S, and v) higher annealing temperature.
Bi-PAP
[0305] PAP has a potential selectivity of 3.3x1011:1 (Fig. 19). Approaching this potential s requires a design that eliminates confounding sources of error. The A190T
mutation of the lac/
gene of A. DNA is used as a model system. In PAP with one downstream P* and one upstream unblocked eligonucleotide,' extension errors fitin the non-blocked upstream oligonucleotide can produce the rare mutation of interest, thus reducing the selectivity. If the nfisincorporation rate of TaqF8 is 104 per incorporated nucleotide and one of the three possible misincorporations 2.. o generates the A--->T mutation on the newly synthesized upstream strand, the selectivity decreases to 3.3x10-5 due to the side effect. In. order to remove this limitation, Bi-PAP was developed (Fig. 20A). In Bi-PAP, both the downstream and upstream oligonucleotides are P*s that are specific for the nucleotide of interest at "their 3' termini. The P's overlap at their 3' termini by = one nucleotide.
-15 [0306] Bi-PAP amplified efficiently and specifically at nucleotide position 190 using X phage = DNA containing the lad l gene as template (Fig. 20B). Addition of human genomic DNA did not = affect the amplification. The 79-bp product of Bi-PAP was easily distinguished from nnincorporated P*s. P* did not form dimers because P* needs a perfectly matched region at the 3' tenninus" for activation. Similar results were observed at nucleotide position 369. Direct 20 sequencing analysis confirmed the correct sequence of the amplified product.
Sensitivity and Selectivity of Bi-PAP
[0307] In order to demonstrate the extremely high selectivity of Bi-PAP, more than 1010 copies of DNA template was Used for a Bi-PAP reaction. X DNA containifig the in'fq"
gene of E.coli was chosen as the model system because I ig of X DNA contains 2x1010 vector genomes, while 25 lug of human genomic DNA only contains 3.3X.105 genomes. In order to avoid potential contamination of the wild-type X DNA in this laboratory, mutation-specific Bi-PAP assays with mutated P*8 were chosen to amplify the wild,type X DNA. The relative frequency of a spontaneons mutation of the lac/ gene in the wild-type X DNA is estimated to be less than le by examining X, phage plaques infecting E. coll.
= 30 [0308] The sensitivity and selectivity of Bi-PAP were examined using three mutation-specific 131-PAP assays with their corresponding mutated X DNA (see Table 7 footnotes for definitions).
Four titration experiments were performed for each mutation-specific B1-PAP
assay (Figs. 21A-.

2IC). Experiment I tested how much the mutated pi, can "tolerate" the wild-type DNA template (i.e., the maximum copies of the wild-type template without a detectable Mutated product). The wild-type 2k. DNA was titrated fi-ana 2x1010 copies to 2x106 copies. The maximum tolerances were 2x109 to 2x101 , 2x107 to 2x 108, and 2x107 to 2x 108, respectively, for the three mutation = s sPecifie Bi-PAP as8A3is, respectively (Figs. 21A-21C). Experiment 11 tested the sensitivity of Bi-PAP. The mutated X DNA was titrated from 2x103 to 0 copies: The ratio of the maximum tolerance (Experiment I) to the sensitiVity is the selectivity. Experiment II
was repeated in the presence of large amount of wild-type template (Experiment III) or large amounts of human genomic DNA (Experiment IV) without effects (Fig. 21A; data not shown for T3690 and

10 T369C). A dose response with template copy number was observed.

Summary of the three mutation-specific Bi-PAP assay?
Assay Position" Type' SensitiVityc Selectivitityd A 190 2 109:1 to 101":1 =
369 2 10':1 to 108:1 369 2 101:1 to 108:1 15 'In each of the three mutation-specific Bi-PAP assays, two opposite P*s with one nucleotide overlap at their 3' termini were used: The P*s are 40-42 nucleotides long. They are 5' GATGGCGGAGCTGAATTACATTCCCAACCGCGTGGCACAT* (SEQ ID NO:61) and 5' GGCAACGCCAATCAGCAACGACTGrUGCCCGCCAGTIGA* (SEQ NO:62) in Assay A;
5' GAAGCGGCGTCGAAGCCTGTAAAGCGGCGGTGCACAATCW (SEQ
NO:63) and 20 5' GCGGATAGTTAATGATCAGCCCAC TGACGCG1TGCGCGAGAC* (SEQ ID NO:64) in' Assay B; 5' GAAGCGGCGTCGAA.GCCTGTAAAGCGGCGGTGCACAATCC* (SEQ ID NO:65) and 5' GCGGATAGTTAATGATCAGCCCACTGACGCGTTGCGCGAG AG* (SEQ ID NO:66) in Assay C.
." The position of the first n.ueleotide of transcript in the /ad gene is assigned the nucleotide position 1 25 (Farabaugh, 1978), The 3' nucleotide of the P* is located at the indicated position and is complementary to the corresponding mutation.
The sensitivity is defined as the minimum copies of the Mutated template from which a detectable mutated product is generated when a mutation-specific Bi-PAP assay is used. It was determined by Experiment II (Figs. 21A-21C).
30 d The selectivity is the ratio of the Makin:ant copies of the wild-type template with undetectable product the minimum copies of the mutated template with detectable product to, when a mutation-specific Bi-. PAP assay is used.
[03091 The approximately 100-fold difference in selectivity betWeert the nucleotide positions 35 190 and 369 my deriVe frOM: i) the preSence of spontaneous mutations at the position .360 at a 71 =
frequency of 10-7 to le in. the wild-type X DNA, ii) iniptirity of P*
oligonucleotides, iii) specificity of pyrophosphorolysis for a perfect match at. the 3' terminus and fidelity of DNA
polymerate to incorpota.te a correct nucleotide may be assothated With sequence context such that the Type II non-specific amplification occurs at a frequency of 104 to le. In the latter = 5 case, a 100-4bld difference in selectiVity could = arise = from a 10-fold , difference in pyrOphOSphOrolVis specificity aiid a .10-fold difference in DNA polymerase fidelity with - sequence context [05101 The rate of a .spontaneous niutatiOn. of X phage M. E. coil varlet.
from locua to 1od1.18; bii the average from 10-9 to 104' per incorporated nucleotide. The amplified signal Seen in.
10. Expotiment I Might be caused by rate spontaneous mutations; =
[0811] There iS a possible side reaction due to the impurity of P*
contamination of unblocked.
= oligonteleotide where the didebxy tenni:ring has not been added, although no unblocked = oligonucleotide was detected in the P*. However, this selectiVity may net be limited severely by r small 'amounts of unblocked oligonucleotide because, the product generated Would be much =
= TT- more likely to he the wild-type rather than the specific mutation (3.3x105:1).
¨.T. [05121 Bi-PAP has = extremely high sensitivity and selectivity.- Bi-PAP Can selectively detect two copies of rare mutatedallele with a single base substitution from Up to.
2X109 copies of the Wild-type allele. Bi-PAP is a simple, rapid, attomatable method for detecting any tare allele of interest. =
20 =
EXAMPLE 13 =
- Meaanrement Of Mutation Loadirt Mouse Tisanes by BiTAP
Materials and Methods [0313] Liver, heart, adipose tisane; cerebrum and cerebellum from 10-day to 25-Month old mice 25 were snap froten and stated undet liquid nitrogen until used. DNA Was extracted according. to the Big Blueprotocol (Stratagene Matructien. Manual). In brief, tisanes "Were homogenized and digested with Proteinage K. The genomic DNA Was extratted with phenol/chloroforin and precipitated with ethanbl. The DNA Was diStelVecl in TE bti#er (10 MM
Tris/IICI, 1 tnIVI
E15TA"., pH 8.0) aftd stilted at 4 C. The amount of the Iii0i1S6 genoMic DNA
Wag deterrnined by 30 U\i" abSofbatee at 2.60 rim.
[08141 =The nintatiOnApeCific Bi-PAP assay for -T3690 (ASSay 8: the two opposite P4's are cadeoxyfiddeotiitte blockedwith one nucleotide OVerlap at their 3' terinir'd.
arci =

5cGAAGCGGCGTCGAAGCCTGTAAAGCGGCGGTGCACAATCG*31 (SEQ ID
NO:63) and 5' GCGGATAGTTAATGATCAGCCCACTGACGCGTTGCGCGAGAC*31 (SEQ ID
NO:64) s of the /Cie/ gene was performed as above except that i) the reaction contained .2 pg of the mouse genomic DNA (-20 kb it size), unless otherwise stated; ii) mouse DNA in 20 ul (1.25x HEP-ES
buffer, 5% DMSO without MgCl2) was heated at 100 C for 2 min and quickly cooled on ice;
. before the other components added; iii) a denaturing step at 95 C for 1 min was added before the first cycle; iv) the denaturing step was 95 C for 10 sec.
o [0315] 10 p_l of the 25 p_l reaction was mixed with .10 pl of the denaturing loading buffer, boiled and rapidly cooled on ice. The product was electrophoresed through a 8% 7M
urea /PAGE gel with 90 in)\11 T13E buffer at room temperature. The gel was dried and exposed to Kodak X-OMATTm AR film for autoradiography.
= - Results and Discussion 15 [0316] Transgenic mouse mutation detection systems permit determination of the frequency and pattern of spontaneous or induced mutations in vivo. The Big Blue! system uses transgenic mice.
harboring chromosornally-integrated A phage. DNA containing the K. coil lad gene as the mutational target (Grpssen and Vijg, 1993; Gossen et al.; 1989; Kohler et al., 1990. The lad gene is integrated within each mouse diploid genome in 40 tandemly repeated X
DNAs.
20 [0317] The Big Blues mutation detection system assay is performed by isolating genomic DNA.
fi-om transgenic mouse tissues and mixing it with X packaging extracts. The packaged X phage can infect E. coll. In the presence of X-gal substrate, lad mutants give rise to blue plaques on a background of colorless wild-type plaques. Observed mutants derive overwhelmingly from the mouse (Hill et al., 1999): The mutant frequency is detennined by dividing the number of circular 25 blue plaques by. the total mrrnber of plaques. Of 5000 sequenced mutant plaques; 31. T369G
mutants have been found in a total of 149x106 plaques screened from various ages, genders and treatments in this laboratory (frequency = 2,1x1e). .
[03181 TO assess the utility of Hi-PAP for measuring ultra-rare mutations in mammalian cells;
the T369G mutation was analyzed in genomic DNA from the Big Blue mice. Two pg of mouse 30 genomic DNA was amplified in 25 Pl reaction containing a total of 1.2X107 copies of the lad gene. The mutation-specific Bi-PAP assay for T369G (Assay B) was performed for 18 samples in duplicate (Fig. 26A). Three categories of results were denned; each with similar number of =

samples: 1) six samples were positive two times (5, 11-15), 2) seven samples were positive one = time (1, 3, 6, 9, 16-18), and 3) five samples were negative two times (2, 4, 7, 8, 10).
[0319] Two samples in each category were studied further (Figs. 261B-26C, Table 8). In category 1, for the two samples 5 and 12 with the strongest amplified signals (Fig.
26A), a four-fold s dilution to 0.5 lag and 16-fold dilution to 0.125 pg of mouse genomic DNA were performed for further quantitation (Fig. 26B). The T369G mutant frequency for each sample was estimated and varies 370-fold among the six samples (Table 8). The average T369G mutant frequency of 2.9x1(17 was within. 50% of the average T369G mutant frequency of 2.1 x10-7 measured from 4x107 plaques using the Big Blue mutation detection system and confirmed by direct sequencing.

Somatic mutant frequency measured by Bi-PAP
Mouse genomic DNA Frequency of positive amplification" Estimated Sample' mutant frequency"
- Tissue Age 2 pg of 0.5 lag of 0.125 pg of DNA DNA DNA
1 12 Adipose 6 months 8/8 8/8 4/8 (0.69) 9.25x10-7 2 5 Liver 25 months (0.98) 1.31x10-6 3 3 Liver 25 months 8/24 (0.41) 3.38x10-8 4 9 Liver 25 months 13/24 (0.78) 6.50x104 .
5 7 Liver 25 months' 2/24 (0.09) 7.25x10-9 6 10 - Liver 25 months 1/24 (0.04) 3.52x10-9 Average 2.91x10-' a see Fig. 26A.
"the ratio of the number of positive signals for the T369G mutation relative to the total number of reactions.
the average number of T369G mutants per reaction is estimated using a formula (the frequency of zero mutants per reaction = e , x is the average number of mutants per reaction) suppose that the mutant distributes in the reaction according to a Poisson distribution and that if one or more mutants are in the reaction, the amplification is positive, and if zero mutant is in the reaction, it is negative.
the frequency of the T369G mutant of the lad gene in mouse genome per reaction is estimated assuming that the mutant distributes in mouse genomic DNA according to a Poisson distribution and that one or more mutants are positive in the detection. For each of samples 12 and 5, a total of ¨6.0x106, copies of the lad gene are used for the estimate, and for each of samples 3, 9, 7 and 10, ¨2.9x108 copies are used assuming that 2 lig of the lad- mouse genomic DNA contains -4 .2l copies of the lad gene.

=
[0320] The 370-fold variation in mutant frequency was observed in livers of five mice at 25 months of age. This large variation could be due to difficulties in amplifying one copy of the template. To address this issue, each of the analyses was repeated at Mast two titres with similar results: For example, in sample 9, seven of 14 reactions with 2 -p.g Of DNA
Were positivehi one experiment,: three of fbur such reactions were positive in another eXperithent, and two of four such reactions were positive hi a third experiment For sample 7, there wag one positive in eight and one positive in 14 reactions./ The product was sequenced to donfitni the T369G mutation after re-amplification from the positive reaction. hi addition, positive controls (2 p:g of the lad+
mouse DNA with ¨10 copies of T369G) and negative controls (mouse genorniC DNA
without io .the lad- target, i.e., the lad mouse DNA) were performed. As additional positive controls, reconstruction experiments were performed in that the copy number of the mutated X DNA per reaction was serially diluted by two-fold in the presence of the lad" genomic DNA carrier.
Reproducible amplifications from as low as one copy of template were demonstrated. (Figs. 26B, 26C).
as [0321] In retrospect, the 370-fold variation in the frequency of T369G mutant observed among the six mice may not be surprising because the T369G mutant frequency ainong mice is over dispersed, implying a hyper-Poisson distribution (Nishino et al., 1996;
Piegorsch et al., 1994).
Among six mice the inter-animal variation in the overall mutant frequency assayed by the Big Blues mutation detection system might be 3 to 4 fold, with significant founder effects in one or 20 a few of the mice. The variation might be in the range= of 2x1(15 to 8xle which is the sum of more than 1,000 different mutations_ Here, only the T369G mutation is assayed.
It is anticipated.
that the great majority of the signal derives from duplex mutated templates (EEll et al., 1999), but it should be noted that unresolved mismatch intermediates derived primarily from DNA
replication or DNA repair would also generate a signal. Thus, the physical limit of sensitivity is 25 actually one half of a duplex DNA molecule per reaction.
[03221 In conclusion, we demonstrate that Bi-PAP can analyze ultra-rare mutations at frequencies as low as 10-7 to 109, depending on the assay. It is shown that Bi-PAP can detect single copies of the somatic mutation directly front mammalian genomic DNA.
The inter-assay variation may reflect locus-specific variability hi the assay sensitivity or in the frequency of the z assayed mutants among the samples. More work is necessary to distinguish between these possibilities. In mammalian DNA, the number of copies of template is limited by the enormous genome size. Two lig of genbraic DNA contains only 600,000 mouse haploid genomes, yet the reaction is viscous. Our analysis of the Big Blue mouse genomic DNA was facilitated, by the 20 =

copies of the /c/c/ gene per haploid gamine. To measure mutation load in humans, genomic DNA in one reaction could be increased ,at least three fold by reducing the viscosity (e.g., shearing the DNA into small segments by ultrasonic treatment) and another four fold by expanding the reaction_ volume to 100 pl. Mutation load in human genomic DNA
might be 5 facilitated .by analyzing- segments of virtually identical sequence, e.g.. there are three 9.6 kb - segments. with 99+% sequence identity on human X chromosome involved in a common in-version, mutation in hemophilia A (Lakich et aL, 1993). Less complex genomes including C-elegans, Drosophila, and human mitochondria gen.ome or chronic viral infections (e.g., hepatitis B) also should be analyzable with this protocol.
[0323] While the invention has been disclosed in this patent application by reference to the details of preferred embodiments of the invention, it is to be understood that the disclosure is intended in an illustrative rather than in a limiting sense, as it is contemplated that modifications.
will readily occur to those skilled in the art BIBLIOGRAPHY =
Anon, D. et al., Biochemistly 37, 1590845917 (1998).
Bains W. and Smith G.C., J27wor Biol 135, 303-307(1988).
Bailer, 3. et al., Nucleic Acids Res 26, 5073-5078 (1998).
Barony, F., ProC Natl Acad Eci USA 88, 189-193 (1991). =
Bebenek, K. et at., IBiol Chem 265, 13878-13887 (1990).
Becker, M.M. and Grossmann, a, in Footprinting of nucleic acid-protein complexes (ed.
Revzin, A.), pp. 129-159 (Academic Press, New York, 1993).
Boer, P.H. et al., Biochem Genet 28, 299-308 (1990).
Buzin, C.H. et al., BioTechniques 28,746-753 (2000).
cieng, Y.C. et al., J Bid Chem 262, 2187-2189 (1987).
Chien, A. et al., Bacterial 127, 1550-1557 (1976).
Chou, Q. et al.,Nucl Acids Res. 20,1717-1723 (1992).
Cotton, R.G. et al., Proc. Nati Acad Set USA 85, 4397-4401 (1988).
Dal, S.-M. et at., Nature Biotechnology 18:1108-1111 (2000).
R.J. et al., Nucl Acids Ras 19,3749 (1991).
Drmanac, R. et at., Genomics 4, 114-128 (1989).
Duetoher, M.P. and Kornberg, A., J Biol Chem 244, 3019-3028 (1969).

= =
Eckert, K.A. and Kunkel, T.A., Nucleic Acids Res. 18, 3739-3744 (1990).
Gardner, A.F. and Jack, W.E, Nucleic Acids Res 30:605-613.(2002).
Gardner, A.F. and Jack, W.E., Nucleic Acids Res 27:2545-2553(1999).
Ginot, F., Hum Mutat 10, 1-10 (1997).
Gossen, J. and Vijg, J., Trends Genet 9, 27-31 (1993).
Gossen, J.A. et al., Proc Nad Acad Sci USA 86, 7971-7975 (1989).
Gotte, M. et at, J Viral 74, 3579-3585 (2000).
Hacia, J. et al., Nat Genet 21, 42-47 (1999), , Hill, K.A. et al., Mutat Res Mini Reviews 436, 11-19 (1999).
3.o Innis, M.A. and Gelfand, D.Hõ in PCR APPLICATIONS Protocols for Functional Geizornics (eds. Innis, MA., Gelfand, D.H. & Sninsky, II), pp. 3-22 ( Academic Press, 1999).
Jones, P.A. and Laird, P.W., Nat Genet 21, 163-167 (1999). =
Kaledin, A.S. et al., Biokl2india 46, 15764584 (1981).
Kellogg, D.E. et at, Biotechniques 16, 1134-1137 (1994).
Ketterling, R.P. et al., Hum Genet 105, 629-640 (1999).
Khrapko, K.R. et al., ,FEBS Letts 256. 118-122 (1989).
Knoll, A. et al., Hum Genet 98, 539-545 (1996).
= Kohler, S.W. et al., Strategies in Mal Bid l 3, 19-21 (1990).
Kohler, S.W. et al., Proc Nati Acad Sci USA 88, 7958-7962 (1991).
20 Komura, I. and Riggs, A.D., Nucleic Acids Res 26, 1807-1811 (1998).
Komberg, A. and Baker, T.A., DNA Replication, (eds., Second Edition), pp. 113-226 (W.H.
Freeman and Co., New York 1992).
Lakich, D. et at, Nature Genet 5, 236-241 (1993).
Landegren, U. et al., Science 241, 1077-1080 (1988).
=
25 LeProust, E. et al., J Comb Chem 2, 349-354 (2000).
Liu, Q. and Somn-ter, S., BioTechniques 29, 10724083 (2000).
Liu, Q. and Sommer, S., BioTechniques 18,470-477 (1995).
Liu, Q. et al., Am J Med Genet (Neuropsych Genet) 60, 165-171 (1995).
Liu, Q. et al., BioTechniques 26, 932-942 (1999).
3 o Liu, Q. et al., BioTechniques 33, 129-138 (2002).
Liu, Q. et al., Anal Biochem 270, 112-122 (1999b).
Lizardi, P.M. et al., Nature Genetics 19, 225-232 (1998).
Longley, M.J. et al., Nucleic Acids Res 18, 7317-7322 (1990).
Lysov, I. et al., Dokl Akad Nauk SSSR 303, 1508-1511 (1988).

Mathatis, T. et al., Molecular Cloning.. a Laboratory Manual, Cold Spring Harbor, New York:
Cold Spring Harbor Laboratory, 1982. .
Marshall, A. and Hodgson, I.., Nat Biotechnol 16; 27-31(1998) Maxana, A.M. and Gilbert, W., ProcNatl Acad Sci USA 74, 560-564 (1977).
s Meyer, P et al., EMBO J19, 3520-3529 (2000).
.Miyada, C.G. and Wallace, R.B., Methods in. Enzymology 154, 94-107 (1987).
Mueller, P.R. and Wold, B., Science 246, 780-786 (1989). [published erratum, appears in Science 248, 802 (1990).]
Mullis, K.B., PCR Methods Appl 1, 11-4 (1991).
io Myers, R.M. et at., Science 230, 1242-1246 1985..
Nishino, H. et al.,.Environ Mal Mutagen 28, 299-312 (1996).
= Nishino, H. et al., Environ Mal MUtagen 28, 414-417 (1996).
= O'Donovan, M.C. et al., Genomics 51, 44-49 (1998).
Oefner, P. and Underhill, P., Current Protocols in Human Genetics Supplement 19:7.10.1-7 15 .10.12 (1998).
. Orita, M. et al., Proc Natl Acad Sci USA 86:2766-2770 (1989).
Parsons, B.L. and Heflich, R.H., Mutat Res 387, 97-121 (1997).
. Pevzuer P.A., JBionzo/ Struct Dyn 7, 63-73 (1989).
Pfeifer, G.F. et al., Science 246, 810-813 (1989).
20 Pfeifer, G.P. et at., Methods Enzymol 304, 548-571 (1999).
Piegorsch, W.W. et al., Environ Mol Mutagen 23, 17-31(1994).
Pourzqnd, C. and Cerutti, P., Mutat Res 288, 113-121 (1993).
Ramsay, G., Nat Biotechnol16, 4044 (1998). .
Ronaghi, M. et al., Science 281, 363, 365 (1998).
2s Ronai, Z. and Minamoto, T., Hum Mutat 10, 322-325 (1997).
= Russo, E. and Riggs, A.D., Epigenetics Mechanics of gene regulation.
Anonymous 1996 Saiki, R.K. et al., Science 230, 1350-1354(1985).
R.K. et al., Science 239, 487-491 (1988).
Sanger, F. et at, Proc Nail Acad Sci USA 74, 5463-5467 (1977).
30 Sarkar, G. et at., Anal Biochem 186, 64-68 (1990).
Sarkar, G. et al., Nucleic Acids Res 20, 871-878 (1992).
Singh-G-asson, S. et al., Nat Biotechnol17, 974-978 (1999).
Sommer, S.S., Trends Genet 11, 141-147 (1995).
= Sommer, S.S. et at, Mayo Clinic Prac. 64, 1361-1372 (1989).
35 Southern, E.M. et al., Genomics 13, 1008-1017 (1992), /

Southern, E.M. Trends Genet 12, 110-115 (1996).
Spiegebnan, LI. et al., Biorechniques 29, 10844092(2000).
St. Clair, M.H., Antimicrob Agents Cheinother 31, 1972-1977 (1987).
Syvanen, A.C., Hum Mutat 13, 140 (1999).
s Tabor, S. and Richardson, C.C., J Biol Chem 265, 8322-828 (1990).
Tabor, S. and Richardson, C.C., ProcNatl Acad Sci USA 92, 6339-6343 (1995).
Trainor, GL., US Patent No. 5,558,991 (1996).
Ueno, T. and IVIitsuya, H., Biochemistry 36, 1092-1099 (1997).
Urban, S. et al., Proc Nall Acad Scie USA 98, 4984-4989 (2001).
lo Vander Horn, P.B. et al., BioTechniques 22, 758-762 (1997).
Wallace, R.B.et al., Nucleic Acids Res 6, 3543-3557 (1979).
Wong, I. et al., Biochensishy 30, 526-537 (1991).
Yoshitake, S. et al., Biochemist?), 24, 3736-3750 (1985).
=
=
=
=
=

Claims (5)

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A process which comprises serial coupling of two reactions, the first reaction being activation of an inactive oligonucleotide which, if not activated, would prevent the oligonucleotide from being extended on a nucleic acid template and the second reaction being an extension of the activated oligonucleotide on the nucleic acid template, wherein the inactive oligonucleotide has a 3' match with the nucleic acid template wherein the inactive oligonucleotide is activated by a nucleic acid metabolizing enzyme, and wherein the nucleic acid metabolizing enzyme is a topoisomerase, a helicase, RNase H, or a telomerase.

2. The process of claim 1, wherein the inactive oligonucleotide has a 3' end block.

3. The process of claim 1, wherein the extension is performed in the presence of four nucleoside triphosphates and a nucleic acid polymerase.

4. The process of claim 1, wherein the oligonucleotide is at least partially hybridized to the template before and during the first reaction.

5. The process of claim 1, wherein the extension is part of an amplification of the nucleic acid template.